Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Datos Académicos:
Programa de Doctorado cursado: Biología Funcional y Molecular
Departamento responsable: BIOLOGIA FUNCIONAL
Departamento en que presenta la tesis doctoral: BIOQUIMICA Y BIOLOGIA MOLECULAR
Título definitivo de la Tesis: RELEVANCIA DE LOS DOMINIOS CITOPLASMÁTICOS DEL CANAL DE POTASIO
H-ERG EN SUS PROPIEDADES CINÉTICAS Y EN SU REGULACIÓN HORMONAL
Resolución
El Departamento BIOQUIMICA Y BIOLOGIA MOLECULAR en su reunión de fecha 10 de Octubre de 2008, acordó dar su conformidad
para la presentación de la tesis doctoral a la Comisión de Doctorado, en cumplimiento de lo establecido 35.2 de la “Modificación del
Reglamento del tercer ciclo de estudios universitarios, la obtención y expedición del título de doctor y otros cursos de postgrado”, aprobada
por el Consejo de Gobierno, en su sesión del día 23 de octubre de 2008 (BOPA del 19 de diciembre de 2008).
Asimismo el director/directores de la tesis doctoral, cumplen con el requisito establecido en el artículo 2 de la “Modificación del
Reglamento del tercer ciclo de estudios universitarios, la obtención y expedición del título de doctor y otros cursos de postgrado”,
aprobada por el Consejo de Gobierno, en su sesión del día 23 de octubre de 2008 (BOPA del 19 de diciembre de 2008).
Curso: 2008/2009
Datos personales:
Apellidos: ALONSO RON Nombre: CARLOS
D.N.I. 76942119N Dirección: Illas, 3, 2-1
C.P. 33012 Localidad : OVIEDO Teléfono: 620887800
Datos Académicos:
Programa de Doctorado cursado: Biología Funcional y Molecular
Departamento responsable: BIOLOGIA FUNCIONAL
Departamento en el que presenta la tesis doctoral: BIOQUIMICA Y BIOLOGIA MOLECULAR
FOR-OFE-VCE-021
Fdo: FRANCISCO BARROS DE LA ROZA Fdo: MARIA DEL PILAR DE L PEÑA CORTINES
Este trabajo ha permitido identificar a los primeros 16 residuos del extremo amino
del canal y a la región N-terminal del lazo intracelular S4-S5 como protagonistas de la
interacción entre ambos dominios, que constituye un factor determinante de las
peculiares características cinéticas de h-ERG, así como para su capacidad de ser
regulado hormonalmente.
“En igualdad de condiciones la solución
más sencilla es probablemente la más correcta”.
(Guillermo de Ockham, 1280/1288-1349).
Abreviaturas i
Introducción 1
I. Características generales del canal h-ERG. 1
I.1. El canal h-ERG (Kv 11.1): un canal iónico
dependiente de voltaje. 1
I.2. Características e structurales de los canales ERG. 3
I.3. Características biofísica s del canal h-ERG. 6
I.4. Importancia fisiológica de lo s canales ERG. 8
I.4.1. Papel de lo s canales ERG en la repolarización
del potencial de acción cardiaco . 9
I.4.2. Papel de lo s canales ERG en el sistema nervio s o
y en las células neuroendocrinas. 11
I.4.3. Papel de lo s canales ERG en las células tumorales. 12
II. Relacione s e structura-función en h-ERG
y otros canales Kv , CNG y HCN. 13
II.1. El proce s o de inactivación en h-ERG y otros canales Kv . 14
II.2. El proce s o de apertura y cierre en h-ERG
y otros canales Kv , CNG y HCN. 15
II.2.1. El sens or de voltaje de lo s canales Kv , CNG y HCN. 15
II.2.2. El sens or de v oltaje de h-ERG. 17
II.2.3. El papel del extrem o a mino de h-ERG
en la apertura y el cierre. 17
II.2.4. Papel de otras regione s de h-ERG
en la apertura y el cierre. 20
II.3. Estudio de la propiedade s cinéticas y term odinámica s
de h-ERG bajo condicione s de auténtico e stado e stacionario. 21
III. Regulación horm onal de h-ERG. 22
III.1. Regulación de lo s canales ERG. 22
III.2. Regulación horm onal de los canales ERG por TRH. 24
III.3. El papel del do minio proximal en la regulación horm onal
de h-ERG por la TRH. 26
IV. Los o o cito s de Xenopu s laevis
c o m o siste ma de expresión heteróloga. 27
Objetivos 29
Materiales y Métodos 31
I. Medio s , disolucione s y reactivo s . 31
I.1. Medio s para el cultivo de Escherichia coli. 31
I.2. Medio s utilizado s para la obtención, mantenimiento
y registro electrofisiológico de lo s oo cito s de Xenopu s laevis. 31
II. Técnicas de Biología Molecular. 32
II.1. Transformación de células co mpetentes
de Escherichia coli. 32
II.2. Purificación y se cuenciación de DNA. 32
II.3. Reac ción en Cadena de la Polimerasa (PCR). 33
II.4. Mutagéne sis dirigida mediante PCR. 33
III. Plás mido s y variantes de h-ERG. 36
III.1. Plás mido s rec o m binantes . 36
III.2. Generación de variantes de h-ERG. 36
III.2.1 Variantes c on delecione s . 36
III.2.2. Mutantes puntuales y co m binado s . 38
III.2.3. Variantes c on sub stitucione s
en el do minio proximal. 40
IV. Síntesis in vitro de cRNA. 42
V. Expresión en oo cito s de Xenopu s laevis. 43
VI. Registro electrofisiológico de lo s oo cito s. 44
VII. Análisis mate mático. 45
VII.1. Ajuste s mate mático s . 45
VII.2. Análisis e stadístico. 46
VIII. Modelo s de ho m ología e structural. 47
Resultados 49
I. Relevancia del extrem o amino y del lazo S4-S5 de h-ERG
en las propiedade s cinéticas y termodinámica s del canal. 49
I.1. Efecto de la modificación e structural del extrem o a mino
de h-ERG sobre la dependencia de v oltaje de activación
del canal en estado e stacionario. 50
I.2. Efecto de la modificación e structural del extrem o a mino
de h-ERG sobre la velocidad de activación. 55
I.3. Efecto de las m odificacione s a minoterminales de h-ERG
en la cinética de cierre. 57
I.4. Influencia de las mutacione s en el laz o S4-S5 de h-ERG
s o bre las propiedade s de activación del canal
en e stado e stacionario. 59
I.5. Influencia de las mutacione s en el lazo S4-S5 de h-ERG
s o bre la velocidad de activación. 63
I.6. Influencia de las mutacione s en el lazo S4-S5 de h-ERG
s o bre la cinética de cierre. 65
I.7. Alteracione s en el proce s o de apertura y cierre de h-ERG
inducidas por mutacione s en el laz o S4-S5 en canales
carentes del extrem o amino. 67
II. Regulación horm onal de h-ERG por TRH:
relevancia de los do minios citoplas mático s del canal. 73
II.1. Efecto de la eliminación de regione s del extrem o a mino
s o bre la regulación horm onal de h-ERG. 73
II.2. Efecto de la modificación de la secuencia 326-332
(RYRTISK) so bre la regulación hormonal de h-ERG. 78
II.3. Mantenimiento de la regulación horm o nal
en canales h-ERG carentes del seg mento 362-366
de aminoácido s bá sico s ( se cuencia KIKER). 82
II.4. Reduc ción de lo s efecto s horm onales inducido s por la TRH
en canales h-ERG carentes del seg mento 155-209. 87
II.5. Supresión de lo s efecto s horm onales inducido s por la TRH
en canales h-ERG con mutacione s puntuales en el lazo S4-S5. 89
III. Estudio de las interaccione s intramoleculares en el canal h-ERG. 94
III.1. Análisis de po sibles interaccione s intram oleculares
m ediante el estudio de la modificación del cierre de h-ERG
por el pH extracelular. 94
III.1.1. Efecto de la eliminación de regione s N-terminales
en la modulación del cierre de h-ERG
por la acidificación extracelular. 94
III.1.2. Efecto de las mutacione s del lazo S4-S5
en la modulación del cierre de h-ERG
por la acidificación extracelular. 98
III.2. Análisis de las interaccione s intramolec ulares en h-ERG
m ediante la óxido-reducción de cisteínas . 102
III.2.1. Introducción de cisteínas al inicio del extrem o a mino
y en el lazo S4-S5 de h-ERG para el análisis
de la proximidad entre amba s regione s. 103
III.2.2. Efecto de la oxidación de cisteínas introducidas
en el inicio del extrem o amino y el lazo S4-S5
s o bre las corrientes de h-ERG. 104
III.2.3. Reversión del efecto de la oxidación de cisteínas
introducidas en el inicio del extrem o amino
y el lazo S4-S5 de h-ERG. 106
III.2.4. Efecto de la óxido-reducción de cisteínas introducidas
en el extrem o a mino y el lazo S4-S5 de canale s h-ERG
carentes de do minios citoplas mático s . 107
III.2.5. Cinética y dependencia de e stado de la oxidación
de cisteínas introducidas en el extrem o amino
y el lazo S4-S5 de h-ERG. 110
III.2.6. Especificidad de lo s efecto s de la oxidación
para el par de cisteínas 3 y 542 introducidas en h-ERG. 111
Discusión 115
I. Relevancia del extrem o amino y del lazo S4-S5 de h-ERG
en las propiedade s cinéticas y termodinámica s del canal. 115
II. Regulación horm onal de h-ERG por TRH:
relevancia de los do minios citoplas mático s del canal. 121
III. Estudio de las interaccione s intramoleculares en el canal h-ERG. 124
Conclusiones 135
Publicaciones
Abreviaturas
Abreviaturas - i
Todos los canales “de lazo en el poro” parecen haber evolucionado de un antepasado
común con una estructura equivalente a la de los actuales Kir (“ir” de “inward rectifier”)
constituidos por un tetrámero de subunidades, cada una con sólo dos segmentos
transmembranales entre los que se sitúa el lazo del poro. A este canal ancestral único se
agregó un módulo proteico formado por las hélices transmembranales S1-S4 que conforma
el llamado dominio sensor de voltaje (VSD, “Voltage Sensor Domain”), dando lugar en
conjunto a la típica estructura de seis hélices transmembranales (S1-S6, con S1-S4 del VSD
y S5-S6 del dominio poro) de cada subunidad . Una interesante excepción a esta
organización con 6 hélices se presenta en los componentes de la familia slo [e.g. KCa1.1
(BK) y KCa5.1] en los que la presencia de un segmento hidrofóbico adicional aminoterminal
da lugar a una séptima hélice transmembranal (denominada “S0”), que sitúa
excepcionalmente al extremo amino del polipéptido en la cara extracelular de la membrana,
y no en la cara interna en la que se localizan tanto el extremo amino como el carboxilo del
resto de los canales “de lazo en el poro”.
A las subunidades que constituyen el cuerpo central del canal iónico pueden
asociarse otros polipéptidos que, por contraposición y de modo general, suelen denominarse
2 - Introducción
Figura 1. Superfamilia de canales iónicos “de lazo en el poro” o “relacionados con los operados por
voltaje” (VGL).
Se muestra la representación gráfica de las relaciones entre los “canales relacionados con los operados por
voltaje” (VGL) adaptada de Yu et al., 2005. Se destaca en color rojo la posición de los canales ERG (Kv11). La
topología estructural característica de los canales prototípicos de las diferentes familias se esquematizan en el
exterior.
Es interesante indicar que, dentro del dominio sensor de voltaje (VSD) constituido por
las hélices S1-S4, el segmento transmembranal S4 presenta una secuencia aminoacídica
aparentemente poco frecuente en una hélice transmembranal, en la que cada dos residuos
hidrofóbicos se alterna uno con carga positiva (lisina o arginina). Este segmento S4
constituye precisamente un motivo central para la detección del campo eléctrico, cuyas
Introducción - 3
Dentro de las varias familias englobadas en los canales “de lazo en el poro” operados
por voltaje se encuentra la familia denominada eag (“ether-à-go-go”), en la que se incluye
h-ERG. El prototipo de esta familia de canales es el codificado por el locus eag de
Drosophila y cuya denominación proviene del peculiar fenotipo que exhiben las moscas
mutantes en el mismo, con una agitación característica de las patas bajo anestesia con éter.
En la nomenclatura actual para los canales dependientes de voltaje selectivos para iones K+,
esta familia comprende los canales Kv10-12 y agrupa tres subfamilias: Kv10,
correspondiente a eag; Kv11, correspondiente a los erg (“eag-related gene”); y, Kv12 ó elk
(“eag-like”). Hace algunos años se identificaron 3 tipos del locus erg en rata, erg1, erg2 y
erg3 (Shi et al., 1997) y, posteriormente, sus homólogos humanos h-erg1, h-erg2 y h-erg3
(Bauer et al., 2003). Se ha venido utilizando la nomenclatura genérica de “h-erg” para
referirse a la isoforma génica h-erg1, aunque en los últimos años se está utilizando también
el término KCNH2. En esta Memoria utilizaremos “h-ERG” para referirnos al canal codificado
por el gen h-erg1 ó KCNH2 y correspondiente a Kv11.1 en la actual nomenclatura.
Como hemos comentado, h-ERG, al igual que muchos otros canales “de lazo en el
poro”, presenta la estructura cuaternaria básica de cuatro subunidades , cada una de las
4 - Introducción
cuales con seis hélices transmembranales S1-S6 (Figura 2). Es interesante hacer notar
que desde la última hélice del VSD (S4) a la primera del dominio poro (S5), se extiende el
lazo conector intracelular S4-S5. Debido a su situación, este lazo es considerado como el
elemento que transfiere las reorganizaciones del VSD en respuesta a variaciones en el
potencial de membrana hasta el dominio poro, para que la compuerta del canal cambie de
conformación y permita el acceso de los iones al conducto de permeación. De hecho, este
papel del lazo S4-S5 como acoplador entre el sensor de voltaje y la compuerta ha sido
propuesta para varios tipos de canales como los Kv, CNG y HCN (Chen et al., 2001; Lu et
al., 2002; Latorre et al., 2003; Decher et al., 2004; Soler-Llavina et al., 2006) así como en el
propio h-ERG (Sanguinetti & Xu, 1999; Tristani-Firouzi et al., 2002; Piper et al., 2005; Ferrer
et al., 2006). También es interesante indicar que dentro del polipéptido, la parte
correspondiente al cuerpo central hidrofóbico es la más conservada filogenéticamente entre
los canales Kv, CNG y HCN. En el caso de h-ERG, la secuencia en esta región central
mantiene un 49% de identidad con el canal EAG de Drosophila (Warmke & Ganetzky, 1994)
que alcanza un 96% respecto a su homólogo de rata r-ERG (Bauer et al., 1998).
El grupo de canales que engloba a los EAG, CNG y HCN muestra una gran homología
en su región C-terminal, presentando todos un Dominio de Unión a Nucleótidos cíclicos
(cNBD, “Cyclic Nucleotide Binding Domain”). Se ha sugerido que el cNBD estaría implicado
en la regulación de h-ERG por cAMP, bien mediante una unión directa del nucleótido al
mismo o mediante fosforilación del dominio vía Proteín Quinasa A (Cui et al. 2000, 2001). En
lo que respecta a la región final del extremo carboxilo, ésta se ha relacionado con la
expresión del canal en la membrana, tanto con su tránsito hacia ella (Kupershmidt et al.
1998, 2002) como con su ensamblaje en forma de tetrámero y su inserción en la membrana
(Jenke et al., 2003). La obtención de la estructura tridimensional de parte del extremo
carboxilo del canal HCN2 por difracción de rayos X (Zagotta et al., 2003) ha resultado de
gran relevancia debido a su homología entre este grupo de canales. Así, es posible que en
Introducción - 5
el extremo carboxilo de este grupo de proteínas haya tres regiones que puedan participar en
la tetramerización de las cuatro subunidades del canal: la región que une el final de la hélice
S6 con el cNBD denominada “C-linker”, el propio cNBD y la región final del extremo
carboxilo.
que los primeros 135 aminoácidos suelen denominarse dominio eag/PAS. El dominio PAS,
con estructuras del tipo “hélice-lazo-hélice”, que está presente en algunas proteínas
procarióticas y eucarióticas, está implicado en interacciones proteína-proteína para factores
que funcionan como sensores (e.g. fotorreceptores) o transductores de señales. Hace
tiempo que se ha resuelto la estructura tridimensional del dominio PAS de h-ERG, no
habiéndose podido determinar la disposición de los primeros 16 residuos, al estar
desorganizados en la estructura cristalina (Morais-Cabral et al., 1998). A este dominio
eag/PAS del extremo amino de h-ERG se le ha atribuido un papel esencial en el control del
proceso de cierre del canal (Morais-Cabral et al., 1998; Wang et al., 1998, 2000).
Al igual que la mayoría de los canales Kv, h-ERG presenta al menos tres estados
básicos: cerrado, abierto e inactivo, de los cuales sólo el estado abierto es conductor de
iones K+ (Figura 3A). En la población de canales de la célula existe un equilibrio entre las
tres conformaciones que es desplazado hacia un lado u otro por el potencial de membrana,
dado que las transiciones entre los estados son dependientes de voltaje. Así, un potencial
de membrana más positivo en el interior que en el exterior tiende a desplazar el equilibrio
hacia la(s) conformación(es) de abierto e inactivo, mientras que un potencial más negativo
en el interior tiende a desplazar el equilibrio en sentido opuesto hacia la conformación de
cerrado.
Según este equilibrio conformacional, para un potencial de membrana negativo (en
reposo), la población de canales se encuentra en su mayoría en estado cerrado. Si por
cualquier circunstancia (e.g. un voltaje aplicado), la membrana plasmática es despolarizada
hacia un potencial menos negativo o positivo, los canales pasan en su mayoría al estado
abierto. Este proceso se conoce como “activación” o “apertura” y ocurre en casi todos los
canales Kv de forma muy rápida en cuestión de milisegundos. Como consecuencia, la
conductancia global de la membrana a K+ aumenta originándose corrientes macroscópicas
de potasio. Cuando los canales Kv se activan por despolarización se dice que presentan una
“rectificación normal”. En canales como h-ERG, si la despolarización se mantiene en el
Introducción - 7
Desde el punto de vista estructural, h-ERG se clasifica dentro de los canales Kv que se
activan por despolarización y que presentan por ello una rectificación normal. Sin embargo,
debido a su particular cinética, h-ERG se caracteriza por tener una “rectificación anómala” al
ser las corrientes mayores para potenciales negativos que positivos (Figura 3B).
8 - Introducción
I.4.1. Papel de los canales ERG en la repolarización del potencial de acción cardíaco.
h-ERG (KCNH2 ó Kv11.1), asociado con otra subunidad reguladora que podría ser MiRP1
(KCNE2; Abbott et al., 1999; Weerapura et al., 2002) y/o minK (MacDonald et al., 1997). A
pesar de la rápida aparición de IKr, su contribución al mantenimiento de la meseta se basa
en la limitación de esta corriente durante la despolarización debido a las características
cinéticas de h-ERG. De hecho, el comportamiento de h-ERG como rectificador anómalo
hace que la corriente IKr sea mayor al comienzo de la repolarización del potencial de acción.
Por ello, para la finalización del potencial de acción cardíaco es fundamental la corriente IKr,
es decir, el correcto funcionamiento de h-ERG (revisado por Tristani-Firouzi et al., 2001),
junto con la lenta aparición de IKs y la activación de otra corriente tipo rectificador anómalo
denominada IKl.
Es importante notar asimismo que aparte de h-ERG, en los distintos tipos de células
cardíacas se expresan otros varios canales Kv. Además, aparte de su expresión bien
conocida en el ventrículo, h-ERG también es particularmente abundante en otras zonas del
corazón como el nódulo sinoatrial o la aurícula (Schram et al., 2002). Aunque no se conoce
la contribución exacta in vivo de cada corriente a la conductancia total al K+ durante el
potencial de acción cardíaco, es lógico pensar que pequeñas variaciones en el patrón de
expresión de IKr, así como en su cinética y/o magnitud producidas por cualquier proceso
regulador, pueden alterar significativamente las características del potencial de acción de los
cardiomiocitos (Jurkiewicz & Sanguinetti, 1993).
I.4.2. Papel de los canales ERG en el sistema nervioso y las células neuroendocrinas.
potenciales de acción de estas células (Chiesa et al., 1997), sugiriendo que la expresión de
erg podría contribuir a la excitabilidad de algunas células neuronales de mamíferos.
Por otro lado, en el sistema nervioso, los canales ERG se han implicado no sólo en la
excitabilidad y actividad eléctrica, sino también en la neuritogénesis y la diferenciación
neuronal (Arcangeli et al., 1993, 1995; Faravelli et al., 1996). Este papel se debe a la
contribución de ERG al mantenimiento del potencial de membrana basal. Así, los canales
rectificadores anómalos harían que, en células como las neuronas o las musculares, el
potencial basal se mantenga entre los -60 y -80 mV. Sin embargo, en algunos tipos
celulares, al no expresarse los rectificadores anómalos “clásicos” y hacerlo en su lugar
canales ERG, el potencial basal se mantiene entorno a -50 mV. De esta manera, en ciertos
tejidos del sistema nervioso se produce una expresión transitoria de erg durante los estadios
más tempranos de diferenciación neuronal, mientras que en los más tardíos se expresan
otros rectificadores anómalos clásicos (IRK1). La expresión secuencial de las corrientes
codificas por erg y por IRK1 provoca dos niveles de potencial basal de membrana: un
potencial más positivo durante las primeras etapas, que se desplaza hacia valores más
negativos a medida que se progresa en la diferenciación (Arcangeli et al., 1995, 1997).
Además de haberse involucrado la expresión de erg en el desarrollo del sistema nervioso,
también se le ha implicado en el del tejido muscular (Crociani et al., 2000). Por último, se ha
descrito la expresión de erg en células neuronales periféricas, detectándose una corriente
con las características de la codificada por este locus en células intersticiales de Cajal de
intestino delgado, proponiéndose su participación en el control del potencial de membrana
basal así como en la actividad marcapasos de las mismas (Zhu et al., 2003).
El papel de los canales ERG en células endocrinas es más conocido que en las
neuronas, contribuyendo tanto a la exicitabilidad como al mantenimiento del potencial de
membrana basal. Así, son varios los trabajos en los que se ha descrito el relevante papel de
ERG en la excitabilidad necesaria para que las células adenohipofisarias, -pancreáticas o
cromafines desempeñen su función fisiológica de secreción de neurotransmisores y
hormonas (Corrette et al., 1995; Bauer et al., 1999; Rosati et al., 2000; Lecchi et al., 2002;
Gullo et al., 2003). De hecho, fue precisamente en nuestro laboratorio donde por primera
vez se describió que la corriente inhibida por la TRH para incrementar la frecuencia de
producción de potenciales de acción (y con ello la concentración de Ca2+ intracelular y los
niveles de secreción) era precisamente ERG (Barros et al., 1997). Ello ha enfatizado
asimismo la importancia fisiológica de las corrientes codificadas por erg en las células
adenohipofisarias en la excitabilidad y mantenimiento del potencial de membrana basal
(Barros et al., 1992, 1994; Bauer et al., 1999; Schäfer et al., 1999).
ciclo celular (Wonderlin & Strobl, 1996), comprobándose que una gran variedad de
bloqueantes de canales de K+ son capaces de detener dicho ciclo en diferentes puntos
(Pardo, 2004; Felipe et al., 2006).
Desde los primeros trabajos de Hodgkin y Huxley (Hodgkin & Huxley, 1952) hasta
nuestros días, el avance en las técnicas de Biología Molecular y Celular, la Electrofisiología
y el perfeccionamiento de técnicas de determinación estructural como la Cristalografía de
14 - Introducción
El canal h-ERG muestra una inactivación tipo C, pero con algunas peculiaridades. En
primer lugar, es un proceso extremadamente rápido que, a diferencia de la inactivación tipo
N ó C clásicas, es dependiente de voltaje en el rango en el que también lo es la activación.
En segundo lugar, la recuperación de la inactivación en h-ERG es un proceso sumamente
rápido y puede ser ralentizada al aumentar la concentración de iones K+ externos, al
Introducción - 15
contrario de lo que ocurre con la inactivación tipo C “clásica” (Rasmusson et al., 1998).
Finalmente, las altas concentraciones extracelulares de Cs+ también ralentizan la
recuperación de la inactivación, al contrario de lo que ocurre en canales como Shaker
(López-Barneo et al., 1993; Zhang et al., 2003).
Los determinantes moleculares de la particular inactivación tipo C de h-ERG aún están
por esclarecer. En algunos trabajos se ha demostrado que se pueden transferir las
características de inactivación de h-ERG al canal codificado por m-eag (homólogo de ratón
de eag) al introducirle a éste la región del poro y parte de la hélice S6 del primero (Herzberg
et al., 1998), indicándose asimismo que el lazo entre el segmento S5 y el poro, el cual
presenta una secuencia muy diferente en h-ERG respecto a otros canales Kv, desempeña
una función importante en la inactivación del canal (Liu et al., 2002; Torres et al., 2003). Por
otro lado, se ha sugerido que existe un sensor de voltaje independiente del que activa al
canal (el VSD) para la inactivación, haciendo este proceso también dependiente de voltaje
(Wang et al., 1996; Johnson et al., 1999). También se ha postulado que la hélice S4 actúa
como sensor para la activación y la inactivación, moviéndose lentamente en el primer
proceso y rápidamente en el segundo (Smith & Yellen, 2002; Piper et al., 2003, 2005).
Como ya se ha indicado anteriormente, la inactivación tipo C rápida y dependiente de
voltaje de h-ERG, es un determinante esencial para su comportamiento como rectificador
anómalo, debido al solapamiento de ésta con la lenta activación, que limita enormemente las
corrientes a voltajes positivos (Smith et al., 1996). De hecho, se podrían obtener niveles
notables de corriente al despolarizar si la activación fuese, al igual que la inactivación, un
proceso rápido, como ocurre en los canales Shaker de Drosophila. Por último, recordar que
este comportamiento como rectificador anómalo es fundamental para el papel de h-ERG en
los procesos fisiológicos en los que interviene.
Desde hace un tiempo se considera que no es sólo el segmento S4, sino todo el VSD
en conjunto (S1-S4), el que sufre cambios conformacionales en respuesta a variaciones en
el potencial de membrana (Durell et al., 1998) y que serán transmitidos subsiguientemente al
módulo S5-S6 que configura el poro (dominio poro). El movimiento de la hélice S4 es hoy en
día objeto de gran controversia. Algunos autores apoyan el modelo de “paletas”, que se
basa en considerar a la segunda mitad de la hélice S3 y la S4 como una unidad funcional,
constituyendo un motivo “hélice-giro-hélice” que actúa como una “paleta” que abre el canal.
Así, en el estado cerrado la “paleta” se situaría en la hemicapa lipídica interior y, cuando el
canal se abre, ésta se trasladaría a la hemicapa exterior, quedando parte de ella expuesta al
medio extracelular. De esta forma, el segmento S4 pasaría de una posición horizontal
paralela a la membrana a una posición vertical, experimentando un largo movimiento de
traslación a través de la bicapa lipídica que tiraría del lazo S4-S5, produciendo un
desplazamiento del segmento S5 que culminaría con una apertura de la “compuerta” del
canal a nivel de la hélice S6. Además, el movimiento de las cargas positivas (residuos
básicos) del segmento S4 sería asistido por cargas negativas (residuos ácidos) en el
segmento S2 (Jiang et al., 2003a, 2003b; Long et al., 2007).
Sin embargo, una mayoría de autores apoyan un modelo tipo “transportador”, basado
en el mecanismo por el cual algunas de estas proteínas (e.g. bomba Na+/K+) responden a
cambios en el potencial de membrana. Así, en la región que rodea a la hélice S4 existirían
profundas hendiduras por las cuales el medio acuoso penetraría en la bicapa lipídica,
determinando alrededor de dicha hélice un campo eléctrico muy diferente al del resto de la
membrana, confiriendo a esta región la máxima sensibilidad a cambios de voltaje. Por ello,
la hélice S4 no efectuaría movimientos de traslación tan largos, sino que experimentaría un
movimiento longitudinal corto, acompañado de una rotación sobre su eje y un cambio en su
inclinación (traslación-rotación-torsión), de forma que las hélices S4 y S5 se mantendrían
próximas entre sí, produciéndose contactos entre ellas (Cha et al., 1999; Glauner et al.,
1999; Broomand et al., 2003: Ghandhi et al., 2003; Lee et al., 2003; Neale et al., 2003; Laine
et al., 2003; Posson et al., 2005; Soler-Llavina et al., 2006; revisado por Bezanilla, 2008).
Por último, en un trabajo relativamente reciente, utilizando un modelado molecular de Kv1.2
y KvAP basado en el método computacional “Rosetta”, se ha propuesto que el movimiento
de la hélice S4 y de todo el VSD puede tener diferentes amplitudes según el tipo de canal
que se trate (Yarov-Yarovoy et al., 2006).
Introducción - 17
voltaje dado. En este trabajo también se demostró que al eliminar tan sólo 19 aminoácidos
del dominio proximal (región 355-373) se produce cerca de la mitad de estos efectos
observados sobre la apertura (Viloria et al., 2000). En conjunto, nuestro trabajo ha permitido
demostrar que la presencia del dominio proximal exclusivo de h-ERG es un determinante
esencial de su lenta apertura, lo que junto con su rápida inactivación contribuye a su
operación como un rectificador anómalo. En trabajos posteriores hemos propuesto un
modelo cinético según el cual el dominio proximal actuaría como un freno molecular que
ralentiza el proceso de apertura (Gómez-Varela et al., 2002).
Puesto que la modulación del cierre de h-ERG por la región inicial del extremo amino
ha de implicar alguna interacción directa o indirecta con el cuerpo central del canal, desde
hace años y por analogías con el mecanismo de “bola y cadena” implicado en la inactivación
tipo N, aunque sin ningún dato directo que lo sustente, se ha postulado una interacción del
extremo amino de h-ERG con el lazo S4-S5 (Morais-Cabral et al., 1998; Wang et al., 1998;
Chen et al., 1999; Sanguinetti & Xu, 1999; Gómez-Varela et al., 2002). De hecho, tras la
resolución de la estructura del dominio eag/PAS de h-ERG y al comprobarse que éste
contenía un dominio PAS eucariótico, relacionado en otros sistemas con interacciones
proteína-proteína, se postuló que dicho dominio podría interaccionar con el cuerpo central
del canal para regular el cierre, probablemente a nivel del lazo S4-S5 (Morais-Cabral et al.,
1998). No obstante, es importante recordar que dentro del dominio eag/PAS no es el PAS el
que ha sido relacionado con la maquinaria de apertura y cierre, sino los primeros 16
residuos sin estructura definida (Wang et al., 1998, 2000). Por otra parte, se ha comprobado
que la mutación a cisteína de ciertos residuos del lazo S4-S5 ocasiona una aceleración del
cierre semejante a la que tiene lugar al eliminar los 16 primeros aminoácidos N-terminales,
proponiendo a dicho lazo S4-S5 como el sitio de interacción con el extremo amino (Wang et
al., 1998, 2000; Sanguinetti & Xu, 1999). Asimismo, se ha sugerido que la aceleración del
cierre inducida por la acidificación extracelular podría deberse precisamente a una
interrupción de las interacciones entre el domino eag/PAS y el cuerpo central del canal,
probablemente a nivel del lazo S4-S5 (Liu et al., 2003).
Igualmente, en nuestro trabajo demostrando que la eliminación del dominio proximal
facilita la apertura de h-ERG, ya postulamos que ello era debido a que se favorecía la
interacción del dominio eag/PAS con el cuerpo central del canal (Viloria et al., 2000). En
estudios posteriores de nuestro laboratorio centrados en la secuencia DRY del lazo S4-S5
(muy conservada en otras proteínas de membrana y con dos de los tres residuos cargados
del lazo S4-S5), se comprobó que mutaciones en dicha secuencia eran capaces de
compensar el efecto producido por la eliminación del dominio proximal (Cristina Gutiérrez
Viloria, 2000, Tesis Doctoral). A raíz de estos resultados, en el modelo cinético propuesto
por nuestro laboratorio, sugerimos que la función del dominio proximal como freno molecular
de la activación de h-ERG se debía al entorpecimiento de la interacción entre el dominio
eag/PAS y el cuerpo central del canal, probablemente a nivel del lazo S4-S5 (Gómez-Varela
et al., 2002). Esta hipótesis ha resurgido hace poco tiempo, tras un trabajo en el que se
comprobó que los efectos del dominio proximal se deben fundamentalmente a una
secuencia con alta densidad de carga positiva situada en la región terminal próxima al
segmento S1 de dicho dominio, lo que de nuevo ha llevado a postular que estos residuos
20 - Introducción
positivos del dominio proximal podrían interaccionar con el lazo S4-S5, interrumpiendo la
normal interacción de dicho lazo con el dominio eag/PAS (Saenen et al., 2006).
Desde hace algunos años se ha propuesto que, en los canales “de lazo en el poro” ó
VGL, la segunda mitad de las hélices S6 de las cuatro subunidades del tetrámero estarían
dirigidas hacia la hemicapa lipídica inferior, empaquetadas y entrelazadas hacia la derecha,
constituyendo una “compuerta” que da paso a una cavidad hidrofílica previa al filtro de
selectividad del canal (revisado por Yellen, 2002). En un trabajo reciente se ha estudiado el
papel de dos residuos de glicina de la hélice S6 en la apertura de h-ERG, ya que para otros
canales estos residuos en posiciones equivalentes se han propuesto como “bisagras” para
la apertura de la “compuerta”. Tras el análisis por mutagénesis dirigida de estas glicinas, se
concluyó que las hélices S6 en h-ERG tienen la suficiente flexibilidad como para no
necesitar dichos residuos para la activación del canal, lo que explicaría en parte la tendencia
natural del mismo a mantenerse abierto (Hardman et al., 2007).
En el apartado anterior se han enumerado una serie de estudios en los que se propone
al lazo S4-S5 como el sitio de interacción con el extremo amino de h-ERG, y concretamente
con el dominio eag/PAS. Debido a la localización de dicho lazo en los canales Kv, CNG y
HCN, siempre conectando la parte de la proteína que siente el voltaje (región S1-S4, VSD)
con la que constituye el poro y la “compuerta” (región S5-S6, dominio poro), este lazo S4-S5
se ha considerado también como el elemento que acopla los cambios en el potencial de
membrana a la apertura de los canales (Lu et al., 2002; Soler-Llavina et al., 2006; Latorre et
al., 2003; Chen et al., 2001; Decher et al., 2004; Sanguinetti & Xu, 1999; Tristani-Firouzi et
al., 2002; Piper et al., 2005; Ferrer et al., 2006). En un primer momento se describió que las
mutaciones en el lazo S4-S5 no sólo afectan al cierre (Wang et al., 1998) sino también a la
activación de h-ERG, proponiendo a dicho lazo como fundamental en el mecanismo de
apertura y cierre, al interaccionar probablemente con la “compuerta” del canal a nivel de la
hélice S6 (Sanguinetti & Xu, 1999). Posteriormente, se ha indicado que la interacción entre
el lazo S4-S5 y la hélice S6 ocurre entre los residuos 540 y 665, ya que al introducir cargas
positivas en ambas posiciones el canal es capaz de “reabrirse” a potenciales muy negativos
(Tristani-Firouzi et al., 2002). Finalmente, se ha propuesto que en el estado cerrado de
h-ERG existe una proximidad física entre el aspartato 540 y la arginina 665 y/o leucina 666,
ya que se ha podido inducir la formación de puentes disulfuro en dicha conformación entre
cisteínas introducidas en esas posiciones mediante mutagénesis (Ferrer et al., 2006).
Aparte del reconocido papel del extremo amino de h-ERG en la modulación del
proceso de apertura y cierre, se han publicado algunos trabajos acerca del papel del
extremo carboxilo en las características cinéticas del canal. Así, hace algunos años se
encontró una variante de h-ERG, originada por “splicing alternativo” y carente de
aproximadamente la mitad del extremo carboxilo, con una expresión el doble de abundante
que el transcrito normal y denominada “h-ERGUSO”. Esta variante no produce canales
funcionales, pero tras su co-expresión con el canal nativo genera corrientes, aunque con
Introducción - 21
una activación acelerada y una dependencia de voltaje desplazada hacia valores negativos
(Kupershmidt et al., 1998). También se ha estudiado la mutación S818L, presente en algún
caso de síndrome QT largo II, comprobándose que dicha mutación hace que los canales
mutantes tampoco sean funcionales, pero al co-expresarlos con el canal nativo se producen
corrientes h-ERG con un desplazamiento de la dependencia de voltaje de activación hacia
potenciales negativos y con un cierre acelerado (Nakajima et al., 2000). Por último, se ha
indicado que las influencias del extremo carboxilo sobre los parámetros de apertura y cierre
precisan de la interacción entre éste y el extremo amino (Aydar & Palmer, 2001).
Al ser h-ERG un canal con una relevante función fisiológica en distintos tipos celulares,
para interpretar correctamente dicha función es importante conocer los parámetros cinéticos
y termodinámicos que gobiernan su apertura y cierre. La única forma de obtener dichos
parámetros de forma precisa e independiente de las condiciones experimentales es
mediante su estudio bajo condiciones de equilibrio o estado estacionario. Además, los
parámetros así obtenidos permiten comparar adecuadamente el impacto que produce la
mutación de un determinado residuo o dominio proteico en la funcionalidad del canal. De
hecho, los valores publicados de voltaje de activación semimáxima (V1/2) para los mismos
mutantes de h-ERG varían incluso decenas de milivoltios al usar condiciones consideradas
erróneamente de estado estacionario (Wang et al., 1997a; Kupershmidt et al., 1998; Chen et
al., 1999; Aydar & Palmer, 2001; Paulussen et al., 2002, 2005; Liu et al., 2003; Subbiah et
al., 2004, 2005; Zhang et al., 2005; Saenen et al., 2006).
Puesto que una de las características distintivas de h-ERG es su lenta cinética de
apertura y cierre, para voltajes cercanos al V1/2, las tasas de apertura y cierre son
característicamente lentas (Schönherr et al., 1999; Viloria et al., 2000; Miranda et al., 2005),
haciendo necesarios pulsos de muy larga duración para alcanzar las condiciones de
equilibrio. Es importante recordar que la generación de curvas I/V en condiciones de estado
estacionario permite la obtención no sólo de valores de V1/2 reales, sino de una serie de
parámetros termodinámicos que nos permiten conocer el equilibrio conformacional de la
apertura y el cierre del canal. Así, podemos obtener el valor de “energía libre” (G0) que
dirige el proceso de activación y que nos permite conocer el potencial químico que gobierna
los cambios conformacionales del canal, en ausencia de un potencial eléctrico que los dirija
(i.e. a 0 mV). Por otro lado, se puede conocer el número de cargas traslocadas a través del
campo eléctrico durante la activación (zg) y con él el potencial electrostático (-zgEF) que
dirige la apertura del canal, definiendo la contribución del potencial eléctrico en dicho
proceso. De esta manera, cuando un canal posea una modificación estructural que cambie
sus propiedades de apertura, mostrará valores particulares de G0 y zg de manera que, para
un mismo valor de potencial eléctrico, el potencial electroquímico total [-(G0-z gEF)] que
dirige la activación será distinto. Por eso, es de suma importancia comparar las tasas de
apertura de un canal no sólo en función del voltaje aplicado a la membrana, sino también en
función del potencial químico y eléctrico total que dirige la activación. Dado que el
conocimiento de la velocidad de apertura aporta una idea de la barrera energética del
22 - Introducción
estado de transición (G‡) entre la conformación abierta y cerrada del canal, cuando esta
barrera energética es reducida, la transición entre ambas conformaciones procederá
rápidamente y, por tanto, la tasa de activación será rápida. Sin embargo, cuando la barrera
entre los dos estados es alta, la transición procederá lentamente y la tasa de activación será
por ello lenta.
Durante los últimos años se han descrito distintos procesos de regulación de los
canales tipo ERG, estudiándolos tanto en líneas celulares que expresan el canal de forma
nativa como en sistemas de expresión heteróloga. Los procesos de regulación se han
analizado preferentemente mediante la utilización de agentes farmacológicos, o tras la
sobre-expresión de los componentes reguladores. Así, se han implicado en la regulación
hormonal de los canales ERG diferentes componentes de varias cascadas de señalización
intracelular, como proteín quinasas, proteín fosfatasas, proteínas adaptadoras de
señalización como las de la familia 14-3-3 e, incluso, segundos mensajeros que se unen
directamente al canal como el cAMP o el fosfatidil inositol 4,5-bisfosfato (PIP2). A
continuación se comentan algunos de los múltiples trabajos publicados sobre la regulación
de h-ERG que pudieran ser relevantes para los resultados que se describirán en esta
Memoria.
Hace unos años se describió en nuestro laboratorio que el éster de forbol -forbol-12-
miristato-13-acetato ó PMA inhibía la función de h-ERG expresado en oocitos de Xenopus,
anulándose este efecto por el compuesto “GF109203X”. Teniendo en cuenta que el PMA y
el GF109203X son un activador y un inhibidor, respectivamente, de la Proteín Quinasa C
(PKC), concluimos que h-ERG era regulado por esta proteín quinasa (Barros et al., 1998).
Sin embargo, tras comprobar en oocitos de Xenopus que el efecto del PMA era mimetizado
y anulado, respectivamente, por activadores e inhibidores específicos de la Proteín Quinasa
A (PKA), se propuso a este enzima como el regulador de h-ERG en dicho sistema,
argumentándose un efecto inespecífico del PMA en la activación de proteín quinasas (Kiehn
et al., 1998). Posteriormente, se reafirmó el papel de la Proteín Quinasa A tras comprobar
incrementos y decrementos inducidos por el cAMP en la corriente, que eran similares a los
observados con activadores e inhibidores de esta quinasa, y tras identificar en h-ERG cuatro
residuos potencialmente fosforilables por la misma (Thomas et al., 1999; Kiehn, 2000).
Finalmente se ha reconocido que los efectos del PMA eran debidos a la activación tanto de
la PKA como de la PKC, aunque se ha propuesto que esta última no fosforilaba
directamente h-ERG, sino algunos mediadores de la señalización (Thomas et al., 2003).
En relación con la posible regulación de h-ERG por la PKA y el cAMP, también se han
hecho estudios en células CHO y HEK-293 transfectadas, en las que se ha propuesto un
mecanismo dual de regulación por cAMP, según el cual el nucleótido, por un lado, se uniría
Introducción - 23
directamente al cNBD del canal y, por otro, activaría la PKA, la cual fosforilaría a h-ERG en
sus cuatro residuos fosforilables (Thomas et al., 1999; Kiehn, 2000). La actividad de h-ERG
sería entonces modificada en direcciones opuestas por la unión del cAMP y la fosforilación
por la PKA (Cui et al., 2000, 2001; Kiehn, 2000). Por último, utilizando estudios de doble
híbrido, se ha sugerido un mecanismo alternativo para la regulación de h-ERG por el cAMP
y la PKA, por el cual el nucleótido activaría al enzima y éste fosforilaría a h-ERG en las
serinas 283 (del dominio proximal) y 1137 (del final del extremo carboxilo), haciendo que al
canal se puedan unir dímeros de proteínas de la familia 14-3-3, produciendo un aumento de
la corriente (Kagan et al., 2002). Mas recientemente también se ha propuesto que existen
proteínas adaptadoras interaccionando con h-ERG que dirigen las subunidades reguladoras
de la PKA hacia el canal. Estos complejos moleculares regularían los canales ERG vía
cAMP/PKA durante la respuesta adrenérgica cardíaca (Li et al., 2008).
En cardiomicitos ventriculares de cobaya se ha comprobado que la estimulación de
receptores -adrenérgicos activaba tanto a la PKA como a la PKC, regulando IKr (Heath &
Terrar, 2000). De esta forma se describió que la estimulación de estos receptores
-adrenérgicos, así como de su cascada de transducción, producía un aumento en las
corrientes ERG mediado por la PKA y la PKC, al modificar la inactivación de los canales
cambiando su rectificación. Sin embargo, se postuló que la activación de la PKC se debía a
una activación previa de la vía de la PKA. En un trabajo mas reciente en células HEK-293
transfectadas, se ha descrito que la estimulación de receptores M3-muscarínicos produce
una ralentización de la apertura de h-ERG y un desplazamiento de su dependencia de
voltaje de activación, proponiéndose que estos efectos son debidos a la PKC. En este caso,
esta alteración cinética se debería a la fosforilación de h-ERG por parte de dos tipos de
PKC, la activada por Ca2+ y la activada por 1,2-diacilglicerol (DAG), para la cual sería
necesaria la presencia del extremo amino del canal (Cockerill et al., 2007).
Tanto la PKA como la PKC y la Proteín Quinasa B (PKB) son serín/treonín quinasas
que se consideran relacionadas entre sí y que presentan una alta homología de secuencia
(Jones et al., 1991). Al igual que se ha descrito la regulación de canales ERG por las Proteín
Quinasas A y C, también se ha estudiado su modulación por la PKB (Zhang et al., 2003). Se
pudo comprobar así que, para un normal funcionamiento de h-ERG en células HEK-293, se
necesita una actividad basal de la PKB por vías dependientes e independientes de la
Fosfoinosítido 3-Quinasa, proponiéndose como sitios de fosforilación por esta quinasa las
serinas 331 (localizada en el dominio proximal) y 890 (del extremo carboxilo).
Por último, también se han publicado trabajos sobre la regulación de h-ERG por
segundos mensajeros sin actividad enzimática. Ya se han comentado anteriormente los
trabajos acerca de la unión directa del cAMP al cNBD de h-ERG (Cui et al., 2000, 2001), así
como de proteínas adaptadoras de la familia 14-3-3 cuando el canal ha sido fosforilado por
la PKA (Kagan et al., 2002). Hace unos años se describió que el aumento del estrés
oxidante por incrementos en especies reactivas de oxígeno alteraba la dependencia de
voltaje de inactivación de h-ERG, expresado en oocitos de Xenopus, haciendo que el canal
perdiese sus propiedades de rectificación anómala. Posteriormente, se demostró que el
óxido nítrico (NO) ejercía el efecto contrario, proponiéndose que esto era debido a su
capacidad para neutralizar dicho estrés oxidante. Esta posible regulación de h-ERG se
relacionó con los cambios en la excitabilidad celular en situaciones fisiológicas y patológicas
de isquemia, acompañadas de estrés oxidante, postulándose que el NO producido en estas
situaciones podría contrarrestar dicho efecto (Taglialatela et al., 1997, 1999).
Por otra parte, en ciertos trabajos también se ha descrito que, en células CHO
transfectadas con h-ERG, las elevaciones intracelulares de PIP2 producen un aumento de
la corriente, una aceleración de la activación junto con un desplazamiento de su
dependencia de voltaje hacia potenciales más negativos, así como una ralentización de la
inactivación. Se ha propuesto una unión del fosfolípido al canal a través de interacciones
entre la carga negativa del grupo fosfato del fosfolípido y las cadenas laterales de residuos
positivos de h-ERG. Se han señalado dos regiones C-terminales, con densidad de carga
positiva, como las implicadas en la unión del PIP2 : los aminoácidos 883 a 894, responsables
de los efectos sobre la activación, y los aminoácidos 672 a 697, responsables de los efectos
sobre la inactivación. Finalmente, en estos trabajos también se demostró que la estimulación
de receptores 1A-adrenérgicos produce una disminución de la densidad de la corriente
mediada por canales ERG, tanto en células HEK-293 transfectadas como en un sistema
nativo de cardiomiocitos ventriculares de ratón. Esta regulación se ha propuesto que es
debida precisamente al descenso del PIP2 causado por la activación de la PLC (Bian et al.,
2001, 2004; Bian & McDonald, 2007).
Entre los modelos experimentales más utilizados para el estudio de los mecanismos de
acción de la TRH, se encuentra la línea celular lactotrofa/somatrofa GH3 derivada del cultivo
primario de un tumor de hipófisis de rata (Tashjian et al., 1968). En estas células que, al
igual que los lactotrofos normales, generan espontáneamente potenciales de acción, la TRH
produce una modificación bifásica de la actividad eléctrica, de los niveles intracelulares de
Ca2+ y de la secreción de PRL y hormona de crecimiento. La unión de la TRH a su receptor
Introducción - 25
Además de los descritos hasta ahora, la TRH produce diversos efectos intracelulares
que podrían estar implicados en la secreción a corto o largo plazo, como son la activación
de la MAP Quinasa ERK (Palomero et al., 1998), la producción de NO (Tsumori et al., 1999)
o la liberación de ácido araquidónico (Judd et al., 1986). Si bien ya se ha mencionado la
existencia de un mecanismo de regulación de h-ERG por NO (Taglialatela et al., 1999), así
como de otros canales Kv por la ERK (Adams et al., 2000), ninguna de estas vías parece
participar en la cascada de señalización que acopla la activación del receptor de TRH a la
regulación de ERG en células adenohipofisarias (Schledermann et al., 2001).
Se ha propuesto una vía alternativa para la regulación de r-ERG por TRH en células
adenohipofisarias de rata GH4/C1, en la que la cascada de transducción acoplada al receptor
de TRH participan la proteína G13 y la proteína G pequeña RhoA (Storey et al., 2002). A
pesar de ello, y aunque se ha sugerido que en esta vía podría estar implicada la Rho
Quinasa, en nuestro laboratorio se demostró, mediante el uso del inhibidor específico de
esta quinasa “Y-27632”, que este enzima no está implicado en la inhibición por TRH de la
corriente r-ERG en las células GH3 (Gómez-Varela et al., 2003b).
Por último, en un trabajo mas reciente de nuestro laboratorio, estudiando la
especificidad de acoplamiento del receptor de TRH para la regulación de los canales ERG,
mediante el bloqueo funcional de proteínas G con cDNAs codificantes de mutantes
dominantes negativos de subunidades , se ha demostrado que, si bien el receptor se
acopla a proteínas Gq/11 en la transducción de la señal de Ca2+ durante la fase I de
respuesta hormonal, esta proteína G no está involucrada en la inhibición de las corrientes
r-ERG por la TRH en células GH3. Además, se identificó a las proteínas Gs y G13, así como
las proteínas G pequeñas de la familia Rho, como piezas clave en la regulación hormonal de
ERG por la TRH, sugiriéndose además que los dímeros liberados de estas proteínas
podrían participar en la modulación de los canales expresados tanto en células GH3 como
en células HEK-293 transfectadas (Miranda et al., 2005).
Hace años se identificó y clonó en nuestro laboratorio el cDNA codificante del receptor
de TRH de las células GH3 (de la Peña et al., 1992a, 1992b). Mediante la co-expresión en
oocitos de Xenopus de los cRNAs codificantes del receptor de TRH y del canal h-ERG, se
Introducción - 27
demostró de forma directa que la activación del receptor por la hormona ocasionaba la
regulación del canal, modificando sus parámetros cinéticos (Barros et al., 1998). Al estudiar
los parámetros cinéticos específicos que eran regulados por la TRH, se pudo comprobar que
la adición de la hormona a los oocitos producía una ralentización de la apertura de h-ERG y
un desplazamiento de su dependencia de voltaje de activación hacia potenciales más
positivos, así como una aceleración del cierre, mostrándose sin embargo inalteradas la
inactivación y la recuperación de la inactivación del canal.
Teniendo en cuenta que la eliminación del dominio proximal, exclusivo de h-ERG,
produce alteraciones de los parámetros cinéticos de activación del canal contrarias a las
inducidas por la regulación hormonal por TRH, se estudiaron en este sistema los efectos
hormonales en canales h-ERG carentes del dominio proximal. Al analizar la regulación
hormonal de canales con diferentes deleciones seriadas en el citado dominio proximal, se
comprobó que los efectos de la TRH sobre la activación de h-ERG eran anulados en
canales carentes de todo el dominio proximal (variante 138-373), así como al eliminar las
regiones 223-373, 284-373 y 326-373 del dominio, pero que los efectos reguladores eran
completamente normales en un canal carente de la región 346-373. A la vista de estos
datos, se concluyó que la regulación hormonal del canal parecía depender de la presencia
de la región 326-346 del dominio proximal (Gómez-Varela et al., 2003a). Por otro lado,
también se comprobó que la regulación hormonal del proceso de cierre de h-ERG dependía
de una región diferente de la implicada en la activación, ya que los efectos hormonales
sobre el cierre del canal se reducen al ~20% eliminando todo el dominio proximal, pero se
recuperan hasta el ~50% en canales carentes de la región 223-373 y hasta el ~90% en los
carentes de la región 284-373 (Gómez-Varela et al., 2003a).
observadas entre las correspondientes entidades nativas y mutadas (Rudy & Iverson, 1992;
de la Peña & Barros, 1999).
Para ello, en este trabajo se han abordado los siguientes objetivos concretos:
1. Conocer el impacto real de la alteración estructural del extremo amino y del lazo
intracelular S4-S5 de h-ERG sobre la funcionalidad del canal. Para ello, se estudiarán las
características cinéticas y las propiedades termodinámicas de variantes de h-ERG
modificadas en ambos dominios bajo condiciones de auténtico estado estacionario.
mM K+, en el que se sustituyó parte del NaCl con DTT (“Fluka”), a fin de mantener la
osmolalidad. Para ello se añadió DTT sólido hasta una concentración final de 20 mM,
reduciendo concomitantemente la concentración final de NaCl hasta 24.5 mM.
En una visión general, para llevar a cabo este tipo de mutagénesis se requieren cuatro
oligonucleótidos. Dos de estos oligonucleótidos (sentido y antisentido) contienen la mutación
a introducir y han de hibridar con las dos hebras de DNA molde en la posición donde se
desea efectuar dicha mutación. Los otros dos cebadores (sentido y antisentido) han de
hibridar en posiciones alejadas y más o menos equidistantes a la mutación. Además, estos
oligonucleótidos “externos” han de hibridar en zonas cercanas a sitios de restricción de
interés en el DNA molde.
pSP64A+ h-ERG. Plásmido que contiene la secuencia codificante del canal h-ERG,
cedido por el Dr. Enzo Wanke (Universidad de Milán). Fue obtenido por la subclonación del
cDNA de h-ERG entre los sitios HindIII y BamHI del vector plasmídico comercial pSP64A+
(“Promega”). Dicho vector contiene el gen de resistencia a ampicilina (AmpR), un sitio de
reconocimiento para la RNA Polimerasa del virus bacteriófago SP6 y una cola de poli A+, lo
que permite la transcripción in vitro del inserto, produciendo un transcrito poliadenilado. Las
variantes del canal h-ERG utilizadas en este trabajo (y que aparecen recogidas en la Tabla
2) se generaron a partir de este plásmido recombinante.
deleción 326-341. Finalmente, el fragmento digerido con HindIII y XhoI de esta última
construcción se substituyó por el correspondiente fragmento nativo del plásmido pSP64A+
h-ERG.
pSP64A+ h-ERG D540C, R541H, R541A, Y542C, S543C, E544C, Y545C, G546C.
Los canales con las mutaciones puntuales en el lazo intracelular S4-S5 D540C,
R541H, R541A, Y542C, S543C, E544C, Y545C y G546C se generaron siguiendo la
estrategia basada en el método descrito por Ho et al. (ver apartado II.5). Los
Materiales y Métodos - 39
oligonucleótidos empleados que contienen las mutaciones puntuales, al igual que el resto de
oligonucleótidos que introducen variaciones en h-ERG, se recogen en la Tabla 3.
En los mutantes R541H, R541A, Y542C y G546C, el cebador externo 5’3’ empleado
cubre la secuencia codificante para los residuos 368-376 de h-ERG, en la que se solapa el
sitio único BstEII con las secuencias de los residuos 374-375. Por otro lado, el cebador
externo antisentido empleado en estos mutantes R541H, Y542C y G546C contiene la
secuencia complementaria a la codificante para los residuos 694-701, en la que se solapa,
en las secuencias de los residuos 697-698, el sitio XhoI, único en la secuencia de h-ERG.
En ocasiones, también se utilizó como oligonucleótido externo 3’5’ uno que contiene la
secuencia complementaria de los residuos 666-673 de h-ERG.
Para generar los mutantes D540C, S543C, E544C e Y545C, se emplearon como
oligonucleótidos externos 5’3’ una serie de cebadores semejantes entre sí, con una
longitud de 25-33 pb y que cubren la secuencia codificante de los residuos 320 al 334. Estos
cebadores, empleados anteriormente, portan mutaciones puntuales, para mutagénesis, que
son eliminadas al sustituir los fragmentos resultantes en el plásmido nativo. Además,
también se utilizó como oligonucleótido externo en la tercera PCR el cebador sentido que
contiene el sitio BstEII, citado anteriormente.
En el caso del cebador externo 3’5’ de la primera PCR, se utilizó un oligonucleótido
que hibrida con las secuencias codificantes de los residuos 756-764 y que, además,
contiene una diana para el enzima SalI. Dicha diana, inexistente en h-ERG, la cual se
empleó en otros trabajos, fue eliminada en la construcción final según lo descrito
anteriormente. A su vez, en la tercera PCR se cambió este cebador externo antisentido por
el mismo que ya ha sido descrito y que cubre la secuencia del sitio XhoI.
Finalmente, los fragmentos amplificados tras la última fase del método de mutagénesis
dirigida fueron purificados y, posteriormente, digeridos con los enzimas BstEII y BglII. Tras
ello, los fragmentos fueron nuevamente purificados para reemplazar al correspondiente
fragmento nativo en el plásmido pSP64A+ h-ERG.
La síntesis in vitro del cRNA de las distintas construcciones del canal h-ERG se realizó
utilizando la RNA Polimerasa SP6, la cual utiliza el promotor para esta polimerasa viral,
existente en la secuencia del vector pSP64A+. Por su parte, la síntesis del cRNA del
receptor de TRH se realizó empleando RNA Polimerasa T7, que utiliza uno de los
promotores presentes en el vector pcDNA3B y bajo el cual se encuentra la secuencia del
receptor en la orientación correcta.
El análisis de los datos de corrientes, los ajustes matemáticos para la obtención de los
parámetros cinéticos y la representación gráfica de los mismos, se realizaron mediante los
programas “PulseFit®” (“HEKA Electronik”, Lambrecht, Alemania) e “Igor Pro®” (“Wavemetric
Inc.”, Lake Oswego, OR, USA).
Por su parte, las tasas de cierre se determinaron ajustando la fase de caída de las
corrientes de cola obtenidas al repolarizar a diversos voltajes, después de despolarizar a un
voltaje fijo. Dichos ajustes se efectuaron utilizando una ecuación biexponencial del tipo:
y = Af exp(1/f · x) + As exp (1/ s · x) + C
donde f y s son las constantes de tiempo del componente rápido y lento, Af y As las
amplitudes relativas de estos dos componentes, y C una constante. Para hacer estos
ajustes se situó el primer cursor por detrás del máximo inicial de las colas, una vez
completada la recuperación de la inactivación.
Para determinar la evolución temporal de las corrientes durante la perfusión con los
medios oxidante y reductor, las corrientes de cola fueron normalizadas respecto a la
corriente máxima según la expresión Inorm = (I/I0)100, en la que I0 indica el valor de la
corriente de cola en el momento anterior a la entrada del medio con el agente oxidante
tbHO2 e I el valor de la corriente de cola en cada pulso.
La cinética de inhibición de las corrientes de cola tras la adición del agente oxidante,
se estimó ajustando el curso temporal de la disminución de la corriente normalizada a
ecuaciones biexponenciales del tipo de la mostrada anteriormente [y = Af exp(1/f · x) + As
exp (1/ s · x) + C] para determinar las tasas de cierre de la corriente de cola. En algunos
casos en los que el ajuste biexponencial no rindió resultados satisfactorios, se utilizaron
ecuaciones monoexponenciales del tipo y = A exp(1/ · x) + C.
de oocitos y “p” el nivel de significación obtenido mediante los diferentes test comparativos
utilizados.
La importancia del dominio eag/PAS del extremo amino de h-ERG como determinante
de la lenta cinética de cierre del canal es conocida desde hace tiempo (Morais-Cabral et al.,
1998; Wang et al., 1998, 2000; Chen et al., 1999). Sin embargo, el posible papel que este
dominio pudiera tener en el proceso de activación de h-ERG no ha sido estudiado aún. Es
por ello que en este trabajo se ha abordado la caracterización termodinámica y cinética del
proceso de apertura en canales h-ERG modificados en la región inicial de la proteína
[residuos 1-135, en la cual se encuentra el dominio eag/PAS y cuya estructura tridimensional
ha sido caracterizada hace tiempo (Morais-Cabral et al., 1998)]. En primer lugar, se generó
un canal en el que se eliminó la totalidad de dicha región inicial (2-135). Además, se
generó otra variante del canal carente de los primeros 16 residuos (2-16), cuya estructura
tridimensional no es conocida, al aparecer desordenados en la estructura cristalina de la
región 1-135. Las características de ambos canales con deleciones se compararon con las
de la variante 2-370, carente de la práctica totalidad del extremo amino, así como con las
de la construcción 138-373, que carece específicamente de la región correspondiente al
dominio proximal.
V1/2, zg y G0 a -42.6 ± 0.72 mV, 4.16 ± 0.09 y -4.08 ± 0.05 kcal/mol. Por otro lado, cuando a
canales h-ERG nativos se les aplican los pulsos de 10 s desde potenciales basales de -80 y
0 mV (para los cuales se espera que todos los canales estén en su conformación cerrada o
abierta, respectivamente), las curvas I/V obtenidas aún aparecen separadas unos 30 mV.
Esto demuestra que, para canales nativos, despolarizaciones de 10 s no son aún lo
suficientemente largas como para que se alcance un verdadero equilibrio, lo que significa
que, incluso con estas despolarizaciones tan largas, sólo se obtienen parámetros en un
estado “cuasiestacionario” o isocrónico, pero no los correspondientes a un auténtico estado
estacionario. Por ello, mientras para canales con cinéticas suficientemente rápidas, los
parámetros de activación se pueden obtener mediante ecuaciones de Boltzmann ajustadas
a curvas I/V coincidentes y desde los dos potenciales basales extremos (i.e. -100/-80 y
0/+40 mV) y/o con despolarizaciones largas de 10 s, para canales con cinéticas más lentas,
dichos parámetros han de obtenerse a partir de curvas promedio generadas utilizando las
curvas I/V aún no coincidentes, incluso tras despolarizaciones largas y desde los dos
potenciales basales a los que se esperan canales basalmente cerrados o abiertos,
respectivamente (ver Figs. 5 y 6). Esto garantiza que las curvas I/V sean exclusivamente
una función dependiente de los pulsos despolarizantes y no del estado previo que presenten
los canales (abierto o cerrado). Según esto, las curvas tienden a converger al aumentar la
duración del pulso independientemente de que los canales provengan del estado abierto o
cerrado y, por ello, el valor medio extrapolado marca el final de la tendencia de cada curva y
da una estimación muy razonable de los parámetros en condiciones de verdadero estado
estacionario (Schönherr et al., 1999; Viloria et al., 2000; Miranda et al., 2005). Nótese que el
promediado de las curvas ha de hacerse en horizontal y no utilizando los valores de
corriente a cada voltaje individual, para evitar que dicho promediado altere la pendiente de
las mismas. Por último, y como un claro ejemplo del impacto producido por las desviaciones
del estado estacionario, el valor de G0 de -1.65 kcal/mol obtenido mediante pulsos de 1 s
desde un potencial basal de -80 mV en canales h-ERG nativos, se convierte en -4.08
kcal/mol con pulsos de 10 s, un valor que se desvía aún claramente de las -5.70 kcal/mol
obtenidas en condiciones de auténtico equilibrio.
asimismo de tan sólo 2.3 ± 1.6% (n = 6) tras pulsos de 1 ó 10 s. Por ello, se puede afirmar
que las variaciones en los parámetros debidas a las diferentes situaciones cinéticas con
pulsos cortos y largos, afectan a la misma población de canales. Además, éste es un
comportamiento cinético inherente a h-ERG y que no está relacionado con la elección de los
oocitos de Xenopus como sistema de expresión y/o con las posibles interferencias de
corrientes de cloro endógenas de dichas células, ya que se han observado desviaciones
similares de las curvas I/V en células de mamífero HEK-293 en las que se han expresado
canales h-ERG nativos (Miranda et al., 2005).
Figura 5. Efecto de las modificaciones en el inicio del extremo amino de h-ERG sobre las corrientes
registradas mediante protocolos diseñados para obtener la dependencia de voltaje de activación en
condiciones de estado estacionario.
(A) Corrientes representativas obtenidas de un oocito que expresa canales nativos h-ERG, en respuesta a
despolarizaciones de 10 s a diferentes voltajes según se muestra en el esquema de la parte superior. Al
potencial basal de -80 mV (izqda.) la población de canales se mantiene inicialmente en estado cerrado y a 0
mV en estado abierto (dcha.). En los insertos se muestran de forma ampliada las corrientes al final de los
pulsos despolarizantes y las corrientes de cola obtenidas tras repolarizar la membrana a -100 mV.
(B) Corrientes de un oocito que expresa canales h-ERG carentes de los primeros 16 residuos (2-16). Las
corrientes de cola se obtuvieron al repolarizar a -80 mV tras pulsos de 10 s entre -90 y +50 mV en incrementos
de 10 mV. Para abreviar, solamente se representa el final de los pulsos de despolarización y las corrientes de
cola. Los potenciales basales (“P.B.”) se indican en la gráfica. (C) Corrientes de un oocito que expresa canales
sin la región inicial del N-terminal de h-ERG, incluyendo el domino eag/PAS. Las corrientes de cola se
obtuvieron a -100 mV tras pulsos entre -90 y +60 mV.
Resultados - 53
Figura 9. Efecto de las mutaciones puntuales en el lazo S4-S5 sobre las corrientes de h-ERG
obtenidas a diferentes voltajes de despolarización.
Los protocolos utilizados para la obtención de las familias de corrientes se muestran en la parte superior de los
registros. Se aplicaron a los oocitos pulsos despolarizantes de 1 s a voltajes entre de -80 y +60 mV en
incrementos de 10 mV desde un potencial basal de -80 mV. Nótese cómo los mutantes presentan diferentes
cinéticas de cierre en las corrientes de cola obtenidas durante las repolarizaciones a -100 mV.
60 - Resultados
Figura 10. Efecto de las mutaciones en el lazo S4-S5 sobre la dependencia de voltaje de activación
de h-ERG en estado estacionario.
Las curvas I/V mostradas se obtuvieron mediante los protocolos y familias de corrientes equivalentes a los
indicados en la Fig. 9. Los círculos negros y blancos corresponden a los datos obtenidos mediante pulsos
despolarizantes de 10 s desde potenciales basales extremos hiperpolarizados y despolarizados,
respectivamente, para los cuales los canales se mantendrían en la conformación cerrada y abierta. Debido a
que el mutante D540C exhibe una rapidísima cinética tanto de apertura como de cierre, las curvas I/V
obtenidas mediante despolarizaciones de 1 s desde ambos potenciales basales se presentan superpuestas. Las
líneas continuas se corresponden con las funciones de Boltzmann que mejor se ajustan a los datos. Las líneas
punteadas corresponden a curvas de Boltzmann cuyos parámetros son los deducidos para el estado
estacionario a partir de las curvas obtenidas para los dos potenciales basales. Las flechas indican la posición del
V1/2 en estado estacionario. Las aspas corresponden a los datos obtenidos con prepulsos de 1 s desde un
potencial basal hiperpolarizado y se muestran para una mejor comparación. En el panel inferior se comparan
las curvas de estado estacionario de todos los mutantes. En la parte inferior se resumen los datos cinéticos y
termodinámicos en condiciones de estado estacionario. *p<0.05 frente al canal nativo.
Resultados - 61
El lazo citoplasmático S4-S5, situado entre la última hélice del VSD y la primera del
dominio poro (ver Introducción), se ha propuesto como el elemento que acopla las
reorganizaciones del sensor en respuesta a cambios de voltaje y la apertura de la
“compuerta” que da acceso a los iones al poro. Esta posible función se ha descrito para
diferentes canales de K+ como Shaker (Lu et al., 2002; Soler-Llavina et al., 2006) y el mismo
h-ERG (Sanguinetti & Xu,, 1999; Tristani-Firouzi et al., 2002; Piper et al., 2005; Ferrer et al.,
2006). Así, es posible que las alteraciones en la activación de h-ERG detectadas al mutar el
lazo S4-S5 se deban, al menos en parte, a la perturbación del acoplamiento entre el sensor
de voltaje y la maquinaria de apertura del canal. Según esto, y con el fin de evaluar el
impacto que la mutación de cada uno de los aminoácidos del lazo produce en la activación
de h-ERG, se calculó el incremento de energía libre respecto al control (G0, ver Métodos).
En la Figura 11, se representan los valores de incremento de energía libre de
activación respecto al canal nativo (tanto positivos como negativos) de todos los mutantes
puntuales del lazo S4-S5. Considerando arbitrariamente como relevantes cambios en la
energía libre mayores de 1 kcal/mol, se puede clasificar el impacto producido por cada
mutación según se desestabilice el estado de abierto frente al de cerrado (G0 > 1
kcal/mol), o bien el estado de cerrado frente al de abierto (G0 < -1 kcal/mol). Como se
puede observar, al sustituir la arginina 541 por una histidina, el equilibrio conformacional
permanece prácticamente inalterado. A su vez, si este residuo se reemplaza por una
62 - Resultados
D540C
R541H
R541A
Y542C
S543C
E544C
Y545C
G546C
Figura 12. Efecto de la modificación del lazo S4-S5 de h-ERG sobre la velocidad de activación.
(A) Familias de corrientes obtenidas mediante el protocolo ilustrado en el inserto superior utilizando oocitos
que expresan canales con diferentes mutaciones puntuales en el lazo S4-S5. (B) Variación de las tasas de
activación frente al voltaje de los prepulsos (izqda.) o frente a la energía total de activación (dcha.) en los
diferentes mutantes. Para una mejor comparación, se incluyen los datos correspondientes al canal nativo y a la
variante 138-373, caracterizada por una fuerte aceleración de la velocidad de apertura (Viloria et al., 2000;
Saenen et al., 2006). (C) Histogramas que resumen las tasas de activación a 0 mV (izqda.) y a 4 kcal/mol
(dcha.). Las tasas de activación se representan como el inverso del tiempo de activación semimáxima. Los
valores de t1/2 a 4 kcal/mol se determinaron a partir de los puntos de cruce entre la línea discontinua situada a
4 kcal/mol y las gráficas de variación de la tasa de activación, ilustrada en el panel derecho de B. *p<0.05
frente al canal nativo.
Resultados - 65
Figura 13. Efecto de la mutación del lazo intracelular S4-S5 en la cinética de cierre de h-ERG.
(A) Familias de corrientes representativas obtenidas de oocitos que expresan canales con mutaciones
puntuales en el lazo S4-S5 mediante el protocolo que se muestra en el inserto. (B) Variación de las tasas de
cierre frente al potencial de repolarización (izqda.) o frente al potencial electroquímico (dcha.).
(C) Histogramas que resumen los efectos sobre la velocidad de cierre a -100 mV (izqda.) y a -4 kcal/mol de
energía total impulsora de la desactivación (dcha.). Las tasas de cierre se representan como el inverso de las
constantes de tiempo del componente rápido. Los valores de a -4 kcal/mol se determinaron a partir de los
puntos de cruce entre la línea discontinua y las gráficas de la variación de la tasa de cierre frente al potencial
electroquímico mostradas en el panel derecho de B. *p<0.05 frente al canal nativo.
Resultados - 67
una cisteína en posición 546 del canal nativo, se transforma en una reducción mínima y en
sentido contrario (-0.7 kcal/mol) al introducir la misma mutación en el canal 2-370.
Estas diferencias en el impacto que las mutaciones del lazo intracelular S4-S5
producen en canales con o sin extremo amino, también se extienden a la velocidad de
activación. Según se ilustra en la Figura 15, el t1/2 a 0 mV se reduce de forma significativa
desde los 99 ± 8.6 ms del canal 2-370 hasta los 73 ± 4 ms que muestra el canal
R541A/2-370. Las diferencias son más pronunciadas tras compensar las tasas con sus
respectivos valores de energía total de activación. Así, la tasa de activación más lenta del
mutante R541A respecto al control a 4 kcal/mol, se acelera hasta ser prácticamente cuatro
veces más rápida en la construcción R541A/2-370 respecto a h-ERG 2-370.
También se observan diferencias claras si se hace un análisis similar en los mutantes
Y542C y G546C. Así, las tasas de activación a 0 mV de estos mutantes, que son
significativamente más lentas que las del canal nativo, pasan a ser similares, o permanecen
lentas respecto a 2-370, al combinar estas mutaciones con la deleción del extremo amino
[t1/2 de 111 ± 9.5 ms y 216 ± 11.0 ms para Y542C/2-370 y G546C/2-370,
respectivamente]. Además, al comparar estos dos mutantes para un potencial
electroquímico equivalente de 4 kcal/mol, la velocidad de activación, que era 4.6 veces más
rápida en Y542C respecto al canal nativo, se reduce a 1.7 veces en Y542/2-370 respecto a
2-370. Por último, para el mutante G546C no se observan diferencias en la tasa de
activación a 4 kcal/mol, independientemente de que dicha mutación se introduzca en el
canal nativo o en el canal carente de extremo amino.
Finalmente, también se compararon los efectos que las mutaciones en el lazo S4-S5
inducen en la cinética de cierre de canales h-ERG, con o sin extremo amino. Los resultados
de este análisis se recogen en la Figura 16. Sorprendentemente, se encontraron efectos
aditivos en la aceleración del cierre al combinar la deleción del extremo amino con las
mutaciones de los residuos 541 y 542, tanto a -100 mV como a un potencial electroquímico
equivalente de -4 kcal/mol. Por otro lado, no se observó este efecto aditivo para el caso del
mutante G546C, ya que la aceleración del cierre que se produce al eliminar el extremo
amino, se reduce al introducir una cisteína en posición 546. Así, en el canal 2-370, las
constantes de tiempo del componente rápido de cierre son 9.0 y 6.1 veces más rápidas a
-100 mV y -4 kcal/mol, respectivamente, que en el canal nativo. Sin embargo, son tan sólo
3.3 y 4.4 veces más rápidas en el canal G546C/2-370 respecto 2-370.
Como se discutirá posteriormente, todos estos resultados sugieren que ninguna de las
mutaciones en el lazo intracelular S4-S5 es plenamente neutral respecto a la cinética de
cierre. Además, indican que, probablemente, los fenotipos similares mostrados por los
canales con mutaciones en el lazo S4-S5 y los canales modificados en la región inicial del
extremo amino, no son exclusivamente debidas al entorpecimiento de la interacción entre
ambas regiones, ya que de ser así no se esperarían efectos aditivos entre las mutaciones en
el lazo S4-S5 y la deleción del extremo amino.
Resultados - 69
Figura 14. Dependencia de voltaje de activación en estado estacionario de canales h-ERG sin
extremo amino y con mutaciones puntuales en el lazo S4-S5.
(A) Familias de corrientes representativas en oocitos que expresan canales h-ERG con la doble modificación
tras despolarizaciones de 10 s a los potenciales indicados en la parte superior. Solamente se muestran las
corrientes registradas al final de los pulsos despolarizantes y las corrientes de cola durante la repolarización,
como se indica en los recuadros. Nótese el potencial de repolarización más negativo para el canal h-ERG
G546C/2-370. (B) Dependencia de voltaje de activación en estado estacionario de canales con la modificación
doble. Los círculos blancos y negros corresponden a los datos obtenidos con prepulsos de 10 s desde
potenciales basales hiperpolarizados y despolarizados, respectivamente. Las líneas continuas representan
ajustes a ecuaciones de Boltzmann a los datos obtenidos. Las líneas punteadas representan las curvas teóricas
en condiciones de estado estacionario, las cuales se obtienen al promediar las curvas de los dos potenciales
basales extremos. Nótese la práctica superposición de las curvas independientemente del potencial basal. Las
flechas indican la posición de los valores de V1/2. (C) Curvas de activación en estado estacionario de diferentes
variantes. Para una mejor comparación, se incluyen las curvas tanto del canal nativo como de los canales con
cada una de las modificaciones simples. (D) Recopilación de los parámetros cinéticos y termodinámicos de
todas las variantes. * p<0.05 frente al canal nativo. # p<0.05 frente a los respectivos mutantes simples o
frente a la variante 2-370.
70 - Resultados
Figura 15. Efecto de mutaciones puntuales en el lazo intracelular S4-S5 sobre la velocidad de
activación de canales h-ERG carentes de extremo amino.
(A) Variación del tiempo de activación semimáxima de los diferentes canales frente al potencial despolarizante
aplicado al oocito (izqda.) o frente a la energía total de activación (dcha.). (B) Resumen de los efectos sobre la
velocidad de activación. Se muestran los datos del canal nativo y los canales con cada modificación simple,
para una mejor comparación. *p<0.05 frente al canal nativo. #p<0.05 frente a los respectivos mutantes
simples o frente a la variante 2-370.
Resultados - 71
Así pues, en conjunto, el estudio de las propiedades de los mutantes del lazo S4-S5
combinados con la deleción 2-370, permite concluir que, en al menos tres de las
mutaciones estudiadas (R541A, Y542C y G546C), el cambio estructural introducido produce
una alteración en la maquinaria de apertura y cierre por sí mismo e independiente de la
presencia o no del extremo amino, ya que las mutaciones producen un efecto diferente en el
canal nativo o en la variante 2-370. Esto no contradice la posible existencia de una
interacción física entre la región inicial del extremo amino y el lazo S4-S5, y pone claramente
de manifiesto de nuevo el papel fundamental que dicho lazo desempeña en la maquinaria de
apertura y cierre (Sanguinetti & Xu, 1999; Tristani-Firouzi et al., 2002; Piper et al., 2005;
Ferrer et al., 2006).
Por otra parte, en trabajos previos de nuestro laboratorio (Viloria et al., 2000), se ha
postulado que la facilitación de la apertura de h-ERG observada al eliminar el dominio
proximal, podía ser debida a que se favorece la interacción entre la región inicial de extremo
amino y el cuerpo central del canal. Si esta interacción se diese a nivel del lazo S4-S5, sería
posible que al alterar esta región se recuperasen los efectos producidos al eliminar el
dominio proximal. Por ello, también estudiamos los efectos de la modificación del lazo S4-S5
en canales carentes del dominio proximal, combinando los mutantes R541A, Y542C y
G546C con la deleción 138-373 (datos no mostrados). Los resultados indicaron que, si
bien en algunos parámetros y algunos mutantes parecía recuperarse en parte el fenotipo
silvestre (e.g. la dependencia de voltaje de activación en Y542C/138-373, la velocidad de
apertura G546C/138-373 o la velocidad de cierre en R541A/138-373), en la mayoría de
los casos, existía un efecto dominante de la deleción 138-373 sobre la mutación del lazo
S4-S5. Esto demostraría que la eliminación del dominio proximal produce una alteración de
la activación y el cierre de h-ERG que es independiente de la existencia de un lazo S4-S5
intacto.
Todos estos resultados parecen indicar que el lazo S4-S5 desempeña un papel
fundamental en la apertura y cierre de h-ERG, pero que dicho papel podría ser modulado
por las regiones del extremo amino. Para saber si la contribución de las regiones
N-terminales a la función del lazo S4-S5 es debida a una interacción física directa entre
ambos, son necesarios otros abordajes experimentales complementarios que apoyen estos
estudios cinéticos y termodinámicos. Algunos de los resultados obtenidos al intentar
confirmar de una forma directa la existencia de interacciones intramoleculares entre el
extremo amino y el cuerpo central del canal, se contemplan en los apartados siguientes de
esta Memoria.
72 - Resultados
Figura 16. Efecto de la mutación puntual del lazo intracelular S4-S5 en la velocidad de cierre de
canales h-ERG carentes de amino terminal.
(A) Variación de la constante de tiempo del componente rápido de cierre en función del potencial repolarizante
aplicado al oocito (izqda.) y del potencial electroquímico (dcha.). (B) Histogramas que recogen las tasas de
cierre a -100 mV (izqda.) y a -4 kcal/mol de potencial electroquímico (dcha.). Las tasas de cierre se
representan como el inverso de las constantes de tiempo del componente rápido. Para una mejor comparación,
se incluyen los datos tanto del canal nativo como de los canales con cada una de las modificaciones simples.
*p<0.05 frente al canal nativo. #p<0.05 frente a los respectivos mutantes simples o frente a la variante
2-370.
Resultados - 73
Hace algunos años, se demostraron por primera vez en nuestro laboratorio los efectos
reguladores de la Hormona Liberadora de Tirotropina (TRH) sobre el canal h-ERG, tras
co-expresar funcionalmente en oocitos de Xenopus el receptor para la hormona y el canal
(Barros et al., 1998). Nuestros estudios posteriores han permitido constatar que ciertas
regiones citoplasmáticas N-terminales de la proteína de h-ERG, son esenciales para que
tengan lugar los efectos de la TRH sobre los parámetros cinéticos del canal. Así, mientras la
TRH produce una ralentización de la apertura del canal y un desplazamiento de su
dependencia de voltaje de activación hacia valores positivos (Barros et al., 1998), estos
efectos hormonales se anulan completamente cuando se elimina una gran parte del dominio
proximal del extremo amino de h-ERG (como en los canales 138-373, 223-373 y 284-
373) y, también, desaparecen cuando se elimina una secuencia de 48 residuos del citado
dominio, como en el canal 326-373. Por el contrario, la regulación hormonal por TRH
permanece inalterada cuando se eliminan los residuos 345-373. Así pues, el hecho de que
al incrementar en 19 aminoácidos el tamaño de la deleción del dominio proximal, se pierda
la regulación hormonal, llevó a concluir que la secuencia 326-345 era esencial para la
regulación hormonal por TRH del canal h-ERG (Gómez-Varela et al., 2003a; David Gómez
Varela, 2004, Tesis Doctoral).
Por otra parte, a diferencia de lo observado con la apertura, también se comprobó que
la aceleración del cierre inducida por la TRH se mantiene al eliminar aproximadamente la
mitad del dominio proximal (284-373), mientras que dicho efecto sólo se anula totalmente
al eliminar el dominio proximal completo (138-373). Así, la región comprendida entre los
residuos 138 y 284, es la que parece jugar un papel crucial en el control hormonal de la
desactivación de h-ERG. Por lo tanto, parecen ser diferentes las regiones implicadas en la
regulación hormonal de la activación y del cierre del canal (Gómez-Varela et al., 2003a;
David Gómez Varela, 2004, Tesis Doctoral).
En este trabajo, hemos intentado ampliar estos estudios tratando de identificar qué
secuencias concretas del dominio proximal de h-ERG son necesarias para la regulación
hormonal por TRH de los parámetros de apertura y cierre. Para este fin, inicialmente se
construyó un canal carente de los residuos 333-373 del dominio proximal y se estudiaron de
nuevo los efectos de la TRH sobre el proceso de apertura de h-ERG. Además, también se
analizó la modulación por la hormona de la activación del canal 2-135, que carece del
dominio eag/PAS.
Figura 17. Efecto de la TRH sobre la dependencia de voltaje de activación de canales h-ERG que
presentan diferentes deleciones en el extremo amino.
(A) Esquema del extremo amino del canal h-ERG en las diferentes construcciones. La región correspondiente al
dominio eag/PAS (residuos 1-135) se representa como una barra gris mientras que la región correspondiente al
domino proximal (residuos 135-397) se representa como una barra blanca. Las deleciones internas se ilustran
como líneas discontinuas. (B) Familias de corrientes representativas obtenidas en oocitos que co-expresan el
receptor de TRH y el canal indicado, antes (“Control”, panel superior) o después de la adición de 1 μM TRH
(“TRH”, panel inferior), al aplicar los protocolos mostrados en el inserto. Los pulsos fueron aplicados cada 20-30
s. Los oocitos fueron perfundidos de forma continua con medio 50 mM KCl. Nótese la extremadamente rápida
cinética de desactivación característica del canal 2-135. (C) Efecto de la eliminación de regiones N-terminales
de h-ERG sobre el desplazamiento inducido por la TRH en las curvas I/V. Los datos promediados se obtuvieron
a partir del número de células que se indica en el panel D. Los círculos vacíos y los rellenos se corresponden a
los datos obtenidos en ausencia y en presencia de la hormona, respectivamente. Las líneas continuas
representan funciones de Boltzmann que mejor se ajustan a los datos obtenidos. Los valores de V1/2 antes y
después de la adición de TRH para el canal 333-373 son -11 y +10 mV, respectivamente. Para las
construcciones 2-135, 326-373 y 345-373 los valores de V1/2 con y sin hormona son -20.1/-6.9, -60/-58.5
y -44/-24 mV, respectivamente. Las funciones de Boltzmann en ausencia de TRH para el canal nativo y para el
canal carente del dominio proximal (138-373) se incluyen como referencia. (D) Histograma en el que se
compara los desplazamientos en las curvas I/V inducidos por la TRH en los diferentes canales. También se
incluyen los datos correspondientes al canal nativo y al canal sin dominio proximal 138-373. ** p<0.001, n.s.
no significativo frente al canal nativo.
76 - Resultados
Por último, una vez estudiados los efectos de la TRH sobre la activación, se estudió la
modulación hormonal del proceso de cierre. Como una confirmación de que los efectos de la
TRH sobre la desactivación de h-ERG parecen depender, en gran medida, de la región
comprendida entre los residuos 222 y 285 del extremo amino, se comprobó que la hormona
no aceleraba el cierre de ninguno de los canales con las deleciones del extremo amino
descritas en los párrafos anteriores. Así, como se observa en la Figura 19, que ilustra el
decremento en la constante de tiempo del componente rápido de cierre a -100 mV tras el
tratamiento con la hormona, los efectos de la hormona sobre el cierre de los canales
326-373, 333-373 y 345-373 son similares a los del canal nativo, mientras que en la
variante a la que le falta el dominio proximal completo (138-373) los efectos hormonales
son destacadamente reducidos. Dichos efectos de la hormona sobre el cierre parecen
también relativamente disminuidos en el canal carente del dominio eag/PAS (2-135), pero
dado que esta construcción presenta una desactivación marcadamente acelerada de forma
basal (ver Figura 8), la interpretación de su regulación del cierre por la TRH resulta muy
complicada.
Resultados - 77
Figura 18. Efecto de la TRH sobre la velocidad de activación de canales h-ERG que presentan
diferentes deleciones N-terminales.
(A) Corrientes representativas obtenidas en oocitos que co-expresan las construcciones indicadas y el receptor
de TRH, al estudiar la modulación por la hormona de la velocidad de activación. Se muestran los registros
generados a partir del protocolo indirecto mostrado en el inserto, antes (“Control”, panel superior) y después
(“TRH”, panel inferior) de la adición de 1 μM TRH. Los pulsos de despolarización se realizaron a 0 mV para
todos los canales a excepción de la construcción 326-373, en la cual se realizaron a -40 mV para compensar
el desplazamiento en su dependencia de voltaje. En todos los casos, las repolarizaciones se realizaron a -100
mV. (B) Variación de la corriente de cola en el pulso de repolarización en función de la duración del pulso
condicionante de despolarización antes (“Control”) y después (“TRH”) del tratamiento con la hormona. Los
datos promediados se muestran como normalizados al máximo de corriente obtenida. En los paneles se
incluyen los valores de t1/2 obtenidos en el punto de cruce entre la línea punteada y el gráfico. (C) Histograma
en el que se comparan los efectos de la TRH sobre la velocidad de activación de canales carentes de distintas
regiones N-terminales. Un valor del 100% indica que el tiempo de activación semimáxima se ha incrementado
el doble tras el tratamiento hormonal. ** p<0.001, n.s. no significativo frente al canal nativo.
78 - Resultados
Figura 19. Efecto de la TRH sobre el componente rápido de cierre a -100 mV de canales h-ERG con
diferentes deleciones N-terminales.
(A) Constantes de tiempo () del componente rápido de cierre a -100 mV del canal nativo y los distintos
canales carentes de diferentes regiones N-terminales. (B) Histograma en el que se comparan los efectos sobre
la cinética de cierre de las distintas construcciones de h-ERG tras su tratamiento con 1 μM TRH. Los
decrementos en la constante de tiempo del componente rápido de cierre se comparan con el del canal nativo,
correspondiente a un 22.6 ± 3.7 % (n=24). Una reducción del 50% en la constante de tiempo () se
corresponde con una velocidad de cierre el doble de rápida. ** p<0.001, n.s. no significativo frente al canal
nativo.
alberga otros consensos de unión para segundos mensajeros. De este modo, la presencia
de tres residuos con cargas positivas hacen del segmento 326-332 una región hipotética de
unión para el fosfatidil inositol 4,5-bisfosfato (PIP2), que también ha sido propuesto como un
posible modulador de la actividad de h-ERG (Bian et al., 2001, 2004; Bian & McDonald,
2007). Por último, se ha descrito la regulación de h-ERG mediante proteínas de la familia
14-3-3 y se ha estudiado la unión de algunos miembros de esta familia al canal (Kagan et
al., 2002), para las que la secuencia RYRTISK sería también una variante de su motivo
consenso de unión [consenso RXXS(P)/T(P), donde (P) representa el residuo fosforilado].
Figura 20. Comparación de los efectos de la TRH sobre canales h-ERG con mutaciones en la
secuencia RYRTISK (región 326-332).
(A) Histograma en el que resumen los desplazamientos en la dependencia de voltaje de activación, tras el
tratamiento con 1 μM TRH, de los oocitos que co-expresan el receptor de TRH y canales h-ERG con mutaciones
en los residuos 326, 327, 328, y 331. La construcción YTS-A corresponde al triple mutante en el que los
aminoácidos 327, 329 y 331 fueron mutados simultáneamente a alanina. La dependencia de voltaje de
activación se estudió según se indica en la Fig. 17. Los datos se presentan normalizados respecto al
desplazamiento inducido por la hormona en el canal nativo. (B) Histograma en el que se compara la
ralentización de la activación inducida por la TRH en los diferentes canales. Los valores de t1/2 se estudiaron a 0
mV según se indica en el pie de la Fig. 18. Los incrementos en el valor de t1/2 se presentan como relativos al
incremento inducido por la hormona en el canal nativo. (C) Histograma en el que se compara la aceleración del
cierre tras el tratamiento de las diferentes construcciones con TRH. La constante de tiempo () del componente
rápido de cierre se obtuvo según se ha descrito anteriormente. Los decrementos en se presentan
normalizados al valor obtenido en el canal nativo. n.s. no significativo frente al canal nativo.
Resultados - 81
Figura 21. Efecto de la TRH sobre la activación de canales h-ERG con la secuencia RYRTISK
eliminada o substituida por otras no relacionadas.
(A) Esquema del extremo amino del canal h-ERG nativo y de las diferentes construcciones. Las deleciones
internas se ilustran con líneas punteadas mientras que las substituciones de la región 326-341 se representan
como barras en dos tonalidades de azul. La secuencias aminoacídicas que substituyen a la nativa en la región
326-341 se especifican en Métodos. (B) Efecto de la TRH sobre la dependencia de voltaje (izqda.) y la
velocidad (dcha.) de activación del canal 326-341. Las curvas I/V y el desarrollo temporal de la activación,
antes y después de la adición de 1 μM TRH, se generaron según lo detallado en los pies de las Figs. 1 y 2. Se
incluyen en los gráficos los valores de V1/2 y t1/2. Para una mejor comparación se incluye la curva I/V
correspondiente a la dependencia de voltaje del canal nativo en ausencia de TRH. (C) Histograma en el que se
compara el desplazamiento inducido por la TRH en las curvas I/V de activación de los diferentes canales.
También se incluyen, para una mejor comparación, los datos correspondientes al canal nativo y a la variante
326-373. (D) Histograma de barras en el que se compara la ralentización de la activación inducida por la TRH
en los diferentes canales. Los incrementos del 100% en el valor de t1/2 indican que dicho valor es el doble que
en el control. ** p<0.001, n.s. no significativo frente al canal nativo.
82 - Resultados
Figura 22. Efecto de la TRH sobre el componente rápido de cierre a -100 mV de canales h-ERG que
presentan eliminada o substituida la secuencia RYRTISK.
(A) Constantes de tiempo () del componente rápido de cierre a -100 mV del canal nativo y los distintos
canales con la región 326-341 (secuencia RYRTISK y sus alrededores) eliminada o substituida por otras no
relacionadas con h-ERG. (B) Histograma en el que se comparan los efectos sobre la cinética de cierre de las
distintas construcciones de h-ERG tras su tratamiento con 1 μM TRH. Los decrementos en la constante de
tiempo del componente rápido de cierre se comparan con el del canal nativo según se describe en el pie de la
Fig. 19. n.s. no significativo frente al canal nativo.
completa mediante una deleción interna (canal 363-373). Por otro lado, con el fin de
estudiar la importancia relativa de las secuencias KIKER y RYRTISK, y de analizar la posible
necesidad de la presencia de ambas para la regulación hormonal del canal, las
modificaciones en la secuencia KIKER se combinaron con la mutación triple YTS-A y con la
deleción en la secuencia RYRTISK, generándose así el triple mutante doble YTS-A/KKR-A y
la variante YTS-A/363-373. También mediante recombinación, se construyó el canal
326-342/363-373 que presenta delecionados ambos tramos de secuencia KIKER y
RYRTISK.
Las Figuras 23 y 24, ilustran los resultados obtenidos al estudiar el impacto de estas
mutaciones combinadas en la regulación del proceso de activación de h-ERG por la TRH.
Como se puede observar, y en concordancia con los estudios previos (Saenen et al., 2006),
al anular las cargas positivas de la secuencia KIKER mediante la triple mutación KKR-A, se
produce un desplazamiento en la dependencia de voltaje de activación del canal hacia
potenciales más negativos (V1/2 = -39.3 ± 2.3 mV). Este desplazamiento hacia valores
hiperpolarizantes es aún mayor en el canal delecionado 363-373 (V1/2 = -51.5 ± 0.4 mV).
Teniendo en cuenta que la modificación en la carga local de ese segmento del dominio
proximal es equivalente para las dos construcciones (ver panel A de Fig. 23), estos
resultados indican que, en el caso de la deleción, existe otro tipo de alteración adicional a la
neutralización de las cargas positivas, que potencia el efecto sobre la activación del canal.
Es interesante destacar que estas diferencias en la dependencia de voltaje de
activación observadas en condiciones basales en ausencia de TRH, también se manifiestan
tras la adición de la hormona. Así, el desplazamiento de la curva I/V producido tras el
tratamiento hormonal en el canal KKR-A (22.3 ± 3.0 mV) es similar al que ocurre en el canal
nativo (17.3 ± 1.05 mV), mientras que este desplazamiento es ligeramente reducido,
respecto al del canal nativo, en el caso de 363-373 (11.9 ± 0.6 mV, Fig. 23D). Por último,
cuando las modificaciones en torno a la secuencia RYRTISK (YTS-A y 326-341), se
combinan con las modificaciones en la secuencia KIKER (KKR-A y 363-373) y se analizan
los efectos hormonales, se obtienen resultados similares, con un efecto ligeramente
aumentado en presencia de la modificación KKR-A y ligeramente reducido tras la deleción
de la región 363-373 (Fig. 23D).
Así pues, estos resultados indican que, aunque el grupo de residuos básicos
constituido por la secuencia KIKER tenga un papel crucial en las propiedades basales de
activación de h-ERG, esta secuencia no es demasiado relevante para los efectos
reguladores de la TRH sobre dicho proceso. Para corroborar esta idea, se estudiaron
también los efectos de la TRH sobre la velocidad de activación de estos canales. Como se
puede observar en la Figura 24, en el canal KKR-A se produce una ralentización de la
apertura tras la adición de la TRH, similar a la ocurrida en el canal nativo. Por su parte, el
efecto hormonal es reducido a un 63% en la variante 363-373. Además, cuando la triple
mutación KKR-A se combina con las modificaciones en RYRTISK (YTS-A ó 326-341), el
efecto hormonal es algo más pronunciado que en el canal nativo. Asimismo, tras la deleción
en la secuencia KIKER (363-373) combinada con el triple mutante YTS-A o con la deleción
326-341 en la secuencia RYRTISK, la apertura del canal es ligeramente ralentizada por la
TRH, de forma similar a como ocurría en el canal 363-373 (Figs. 24 B y D).
84 - Resultados
Figura 23. Efecto de la TRH sobre la dependencia de voltaje de activación de canales h-ERG con
modificaciones en las regiones 326-332 (secuencia RYRTISK) y/o 362-366 (secuencia KIKER).
(A) Esquema del extremo amino de h-ERG en el que se muestra de forma ampliada la región 326-376 para
ilustrar las modificaciones que presentan cada uno de los diferentes mutantes. Las secuencias RYRTISK y
KIKER, así como los aminoácidos mutados, están resaltados en cursiva y en negrita. Nótese que la última lisina
puede considerarse como el residuo 362 ó 373, ya que ambas posiciones son ocupadas por este aminoácido en
la secuencia normal del canal. (B) Corrientes representativas generadas antes (“Control”, panel superior) y
después (“TRH”, panel inferior) de la adición de 1 μM TRH, obtenidas mediante el protocolo mostrado en el
inserto. (C) Efecto de la TRH en la dependencia de voltaje de activación de los diferentes canales. Los datos
promediados para generar las diferentes curvas I/V se obtuvieron del número de oocitos indicado en el panel D.
En los gráficos se indica el valor de V1/2 antes del tratamiento hormonal. (D) Histograma en el que se comparan
los desplazamientos en las curvas I/V inducidos por la TRH en canales con modificaciones en la secuencia
KIKER sola o combinada con modificaciones en la secuencia RYRTISK. * p<0.05, ** p<0.01, n.s. no
significativo frente al canal nativo.
Resultados - 85
Figura 24. Efecto de la TRH en la velocidad de activación de canales h-ERG con modificaciones en la
secuencia RYRTISK y/o en la secuencia KIKER.
(A) Corrientes representativas obtenidas mediante el protocolo mostrado en el inserto, antes (“Control”, panel
superior) y después (“TRH”, panel inferior) de la adición de 1 μM TRH. Para compensar los desplazamientos en
la dependencia de voltaje de activación, los potenciales de membrana utilizados fueron de -20 mV para KKR-A
y 326-341, y de -40 mV para 326-341 y 326-341/ 363-373. En todos los casos, las corrientes de cola se
obtuvieron repolarizando los oocitos a -100 mV. (B) Efecto de la TRH en el desarrollo temporal de la corriente
de cola de las diferentes construcciones. Los datos promediados se muestran como normalizados a la corriente
máxima y se corresponden con el número de oocitos mostrados en el panel D. Se muestran los valores de t1/2 ,
que se obtienen en el punto de cruce entre la línea punteada y los gráficos, antes (círculos blancos) y después
(círculos negros) de la adición de hormona. (C) Histograma en el que se compara la ralentización de la
activación de h-ERG inducida por la TRH en los diferentes canales con modificaciones en la secuencia KIKER
sola o combinada con modificaciones en la secuencia RYRTISK. Los datos se presentan como relativos al canal
nativo donde un incremento del 100% significa que el valor de t1/2 es el doble. * p<0.05, ** p<0.01, n.s. no
significativo frente al canal nativo.
86 - Resultados
Por último, también se estudió el efecto que las modificaciones de las secuencias
KIKER y RYRTISK ejercen sobre las características y la regulación hormonal del cierre de
h-ERG. Así, en la Figura 25, se puede observar que la TRH induce, en todas las
construcciones con modificaciones en la región KIKER, una aceleración del componente
rápido de cierre a -100 mV, análoga a la producida en el canal nativo. Nótese asimismo que
los efectos de la TRH sobre la desactivación, son similares a los inducidos en el canal
nativo, independientemente de que las modificaciones en la secuencia KIKER se presenten
solas o combinadas con las modificaciones en la secuencia RYRTISK.
De nuevo, estos resultados indican que las secuencias KIKER y RYRTISK no están
relacionadas con el control hormonal del proceso de cierre de h-ERG y son coherentes con
los datos previos que indican que, en los efectos hormonales sobre el cierre, parecen ser
determinantes otras regiones como la comprendida entre los residuos 223 y 284 del dominio
proximal del canal (Gómez-Varela et al., 2003a).
Figura 25. Efecto de la TRH sobre el componente rápido de cierre a -100 mV de canales h-ERG con
modificaciones en la secuencia RYRTISK y/o en la secuencia KIKER.
(A) Se muestran las constantes de tiempo () del componente rápido de cierre a -100 mV del canal nativo y de
las distintas variantes con modificaciones en la secuencia RYRTISK y en su entorno, y/o en la secuencia KIKER.
(B) Histograma en el que se comparan los efectos sobre la cinética de cierre de las construcciones de h-ERG
tras su tratamiento con 1 μM TRH. Los decrementos en la constante de tiempo del componente rápido de cierre
se comparan con el del canal nativo según lo descrito en el pie de la Fig. 19. n.s. no significativo frente al canal
nativo.
Resultados - 87
Figura 26. Efecto de la TRH sobre canales 155-209 carentes de 55 residuos en la región N-terminal
del domino proximal.
(A) Esquema del extremo amino de los canales h-ERG nativo, 326-373 y 155-209. (B) Corrientes
representativas de oocitos que co-expresan el receptor de TRH y la construcción 155-209, antes (“Control”,
panel superior) y después (“TRH”, panel inferior) de la adicción de 1 μM TRH. Los registros se obtuvieron para
el estudio de la dependencia de voltaje (izqda.) y la velocidad (dcha.) de activación en respuesta a los
protocolos mostrados en los insertos. (C) Efecto de la TRH sobre la dependencia de voltaje (izqda.) y la
velocidad de activación a 0 mV (dcha.) de canales 155-209. Para una mejor comparación, también se
muestra la dependencia de voltaje de activación en ausencia de TRH con el canal nativo (línea discontinua).
(D) Histograma en el que se comparan los efectos de la TRH en la dependencia de voltaje y la velocidad de
activación, así como en el componente rápido de cierre de los canales nativo, 326-373 y 155-209. * p<0.05,
** p<0.001, n.s. no significativo frente al canal nativo.
Resultados - 89
Los resultados expuestos hasta ahora, indican que es necesaria una estructura global
relativamente inalterada en el dominio proximal de h-ERG para que tengan lugar los efectos
reguladores de la TRH sobre la actividad del canal. Sería posible pues, que al eliminar
regiones del dominio proximal se alterase, además de su estructura, su posicionamiento, o
el de otros dominios N-terminales como el dominio eag/PAS, respecto al cuerpo central del
canal, donde se encuentra la maquinaria de apertura y cierre.
Como ya se ha indicado anteriormente, el lazo intracelular que une las hélices
transmembranales S4 y S5 de h-ERG, ha sido propuesto como el posible sitio de interacción
entre la región inicial N-terminal y el cuerpo central del canal y, además, como el elemento
transmisor de las reorganizaciones estructurales en respuesta al voltaje para la apertura de
la “compuerta” del mismo (Wang et al., 1998; Chen et al., 1999; Sanguinetti & Xu, 1999;
Viloria et al., 2000; Tristani-Firouzi et al., 2002; Piper et al., 2005; Ferrer et al., 2006; Saenen
et al., 2006). Por otro lado, de existir interacciones entre las estructuras N-terminales y el
lazo S4-S5 (residuos 540 a 546), es muy probable que dichas interacciones proteína-
proteína se vean alteradas al eliminar el dominio proximal de h-ERG (Viloria et al., 2000;
Gómez-Varela et al., 2002). Así pues, partiendo de esta idea, y en base a nuestros
resultados (ver apartado I), nos planteamos la posibilidad de que la regulación hormonal de
h-ERG, que es significativamente alterada por cambios en las estructuras N-terminales del
canal, se vea también severamente afectada al modificar el lazo intracelular S4-S5.
La Figura 27, recoge los resultados obtenidos al estudiar el impacto que las
mutaciones puntuales en el lazo intracelular S4-S5 ejercen sobre la regulación por la TRH
de la dependencia de voltaje de apertura de h-ERG. Como se puede observar, los
desplazamientos en la curva I/V son marcadamente reducidos, o completamente anulados,
en los canales con las mutaciones puntuales D540C, R541H e Y542C. A su vez, cuando la
arginina 541 se substituye por una alanina en vez de por una histidina (R541A), también se
observa una reducción significativa en el desplazamiento de la dependencia de voltaje
inducido por la hormona. Sin embargo, esta supresión de la acción hormonal al mutar la
región N-terminal del lazo S4-S5, no se observa cuando las mutaciones se introducen en la
sección C-terminal del lazo, ya que como se ilustra en la Figura, los canales S543C, E544C,
Y545C y G546C muestran una regulación hormonal de la dependencia de voltaje normal e
incluso ,en algunos casos, algo mayor que la del canal nativo.
En la Figura 28, se puede observar que también se obtienen resultados similares
cuando se analiza el efecto hormonal sobre el desarrollo temporal de la corriente. Así, la
ralentización de la apertura del canal inducida por la TRH, es mínima en los canales D540C,
R541H, R541A e Y542C, mientras que el efecto hormonal es normal, o incluso algo más
acentuado, en los canales S543C, E544C, Y545C y G546C. Es importante resaltar aquí que,
como se señaló en el caso del canal con la deleción 155-209, también en esta ocasión, la
presencia o ausencia de efecto hormonal no se correlaciona con la modificación que sobre
el proceso de activación del canal pueda causar la mutación en sí. Así, se puede comprobar
que el efecto de la TRH es reducido tanto en los mutantes D540C e Y542C (los cuales
muestran un marcado desplazamiento de su dependencia de voltaje hacia valores más
90 - Resultados
positivos y un decremento en la pendiente de la curva I/V; ver Fig. 10), como en los
mutantes R541H e R541A, con unas propiedades de activación no muy diferentes a las del
canal nativo. Esto de nuevo indica que la reducción o supresión de los efectos hormonales
sobre la apertura, son debidos a la alteración de un mecanismo de regulación, y no a la
alteración previa del proceso de apertura causada por las mutaciones. Por otro lado, estos
resultados señalan, además, a la sección N-terminal del lazo intracelular S4-S5 como un
determinante esencial en la transmisión del efecto regulador de la TRH desde las regiones
citoplasmáticas hasta la maquinaria de apertura-cierre, situada en el cuerpo central del
canal, embebido en la membrana plasmática.
Por último, también se estudió el efecto de las mutaciones del lazo S4-S5 sobre la
aceleración del componente rápido de cierre a -100 mV inducida por la TRH. Los resultados
obtenidos se muestran en la Figura 29. Se puede comprobar que la hormona no produce
ningún efecto sobre el cierre de los canales D540C, R541H, R541A e Y542C, mutados en el
sección N-terminal del lazo. Sin embargo, el proceso de cierre sí parece acelerarse tras la
adición de la TRH a los canales S543C, E544C, Y545C y G546C, mutados en la sección
C-terminal del lazo, aunque la aceleración del cierre es algo menor en los mutantes E544C e
Y545C. No obstante, es importante señalar en este punto que el análisis de los efectos
hormonales sobre el cierre de h-ERG, se complica enormemente en aquellos canales que
(como la variante 2-135) presentan ya una tasa de cierre marcadamente más rápida que la
del canal nativo, en condiciones basales en ausencia de hormona. Este es precisamente el
caso de los canales D540C, R541H, R541A, Y542C, E544C e Y545C (ver Fig. 13), por lo
que puede ser posible que la falta de efecto hormonal sobre el cierre de alguno de estos
canales, sea debido, en parte, a la aceleración basal del cierre causada por la propia
mutación.
Así pues y a modo de resumen global de este apartado de Resultados, podemos decir
que el trabajo previo de nuestro laboratorio, utilizando canales con deleciones seriadas,
sugirió que los residuos 326-345 del dominio proximal eran necesarios para la regulación
hormonal de la activación de h-ERG por la TRH (Gómez-Varela et al., 2003a) y que, aún
cuando al estudiar nuevas construcciones pareció delimitarse todavía más esta región a los
residuos 326-332, el hecho de que los mutantes puntuales en la misma sean regulados
normalmente por la hormona (David Gómez Varela, 2004, Tesis Doctoral) indica que el
mantenimiento de la secuencia no es importante para la regulación hormonal del canal. De
hecho, esta interpretación es definitivamente confirmada al comprobar que los efectos
inducidos por la TRH, se mantienen también en canales en los que la agrupación de
aminoácidos básicos 362-366 es eliminada. Por último, hemos podido constatar que al
mutar la región N-terminal del lazo S4-S5 de h-ERG, el canal pierde su regulación por la
hormona. Estos resultados sugieren que es necesaria una correcta organización estructural
de las regiones N-terminales para la regulación de h-ERG por la TRH y que una
modificación de las mismas podría alterar una posible interacción entre el dominio eag/PAS
y el lazo S4-S5, la cual sería fundamental para la transmisión del mecanismo regulador
desde los dominios citoplasmáticos al cuerpo central del canal.
Resultados - 91
Figura 27. Efecto de la TRH en la dependencia de voltaje de activación de canales h-ERG con
mutaciones puntuales en el lazo intracelular S4-S5.
(A) Corrientes representativas de oocitos que co-expresan el receptor de TRH y los diferentes canales
mutados, antes (“Control”, panel superior) y después (“TRH”, panel inferior) de la adición de 1 μM TRH. Los
registros se obtuvieron tras aplicar el protocolo mostrado en el inserto. Nótese que la escala temporal se
presenta ampliada en los mutantes D540C, Y542C y E544C para resaltar la rápida desactivación de las
corrientes de cola mostrada por estos canales (ver apartado I.6 de Resultados). (B) Curvas I/V promedio
obtenidas a partir del número de oocitos mostrado en C. La dependencia de voltaje del canal nativo en ausencia
de hormona también se representa en los gráficos (línea discontinua). (C) Histograma en el que se compara el
desplazamiento inducido por la TRH en los distintos mutantes con el del canal nativo. * p<0.05, ** p<0.001,
n.s. no significativo frente al canal nativo.
92 - Resultados
Figura 28. Efecto de la TRH en la velocidad de activación de canales h-ERG con mutaciones
puntuales en el lazo intracelular S4-S5.
(A) Corrientes representativas de oocitos que co-expresan el receptor de TRH y los diferentes canales
mutados, antes (“Control”, panel superior) y después (“TRH”, panel inferior) de la adición de 1 μM TRH. En
todos los casos, el voltaje del pulso despolarizante fue de 0 mV a excepción del mutante G546C, en el cual fue
de +20 mV, debido a su velocidad de activación ligeramente más lenta en comparación con la del canal nativo.
Para todas las construcciones las corrientes de cola se obtuvieron repolarizando las membranas de los oocitos a
-100 mV. (B) Desarrollo temporal de las corrientes de los diferentes canales. Se muestran los valores de t1/2,
que se obtienen en el punto de cruce entre la línea punteada y los gráficos, antes (círculos blancos) y después
(círculos negros) de la adición de hormona. La escala temporal representada en el eje de abscisas se reduce a 3
s para una mejor visualización del retraso inducido por la TRH. (C) Histograma en el que se compara el efecto
de la TRH sobre la velocidad de activación de los canales mutados en el lazo intracelular S4-S5 con el del canal
nativo. Un incremento del 100% indica que t1/2 alcanza un valor doble en respuesta a la TRH. * p<0.05,
** p<0.001, n.s. no significativo frente al canal nativo.
Resultados - 93
Figura 29. Efecto de la TRH sobre el componente rápido de cierre a -100 mV de canales h-ERG con
mutaciones puntuales en el lazo intracelular S4-S5.
(A) Constantes de tiempo () del componente rápido de cierre a -100 mV del canal nativo y de los distintos
mutantes en el lazo S4-S5. (B) Histograma en el que se comparan los efectos hormonales sobre la cinética de
cierre de los distintos mutantes de h-ERG, tras su tratamiento con 1 μM TRH. Los decrementos en la constante
de tiempo del componente rápido de cierre se comparan con el del canal nativo según se describe en la Fig. 19.
* p<0.05, ** p<0.001, n.s. no significativo frente al canal nativo.
94 - Resultados
En las Figuras 30 y 31, se muestran los resultados obtenidos al analizar los efectos de
la acidificación extracelular sobre la cinética de desactivación del canal nativo. De acuerdo
con trabajos anteriores (Anumonwo et al., 1999; Jiang et al., 1999; Liu et al., 2003), al
disminuir el pH externo desde 7.5 a 6.5, se produce una ligera reducción de la magnitud de
las corrientes de cola obtenidas a diferentes potenciales negativos, así como una clara
aceleración del proceso de cierre. Se puede observar que la acidificación reduce las
Resultados - 95
constantes de tiempo tanto del componente lento (lenta) como, predominantemente, del
componente rápido (rápida) de cierre, el cual es el mayoritario en las corrientes de cola
registradas a potenciales muy negativos. Esta aceleración del cierre es similar tanto en 50
mM como en 2 mM KCl, sin bien, para potenciales negativos próximos o por debajo del
potencial de reversión del catión, el efecto de la acidificación fue algo más reducido al utilizar
el medio con baja concentración de potasio (datos no mostrados). Por todo ello y para
maximizar la magnitud de las corrientes de cola, las comparaciones cinéticas fueron
realizadas en un medio extracelular conteniendo 50 mM KCl.
Estos resultados podrían indicar que los efectos de acidificación sobre el cierre del
canal h-ERG, precisarían de la presencia de los primeros 16 residuos aminoacídicos de la
proteína. Sin embargo, la ausencia de aceleración del cierre inducida por la acidificación,
también se observa cuando se elimina el dominio proximal (138-373). Es interesante
destacar también que en otra variante de h-ERG, carente de un segmento del carboxilo
terminal (864-1010), se observa una aceleración del cierre por la acidificación extracelular
que es análoga a la del canal nativo (datos no mostrados), indicando que la anulación o
reducción de los efectos de la acidificación extracelular sobre el cierre, es dependiente
específicamente del extremo amino del canal. Nótese asimismo que, si bien los canales con
diferentes deleciones en el comienzo del extremo amino (2-16, 2-135 y 2-370) ya
presentan una cinética basal de cierre más rápida que la del canal nativo, la ausencia de
aceleración del cierre también ocurre en la construcción 138-373, cuya cinética basal de
cierre es similar a la del canal nativo. Parece claro pues que la ausencia de modulación por
el pH extracelular, no se debe exclusivamente a la aceleración basal del cierre causada por
algunas de las modificaciones N-terminales.
96 - Resultados
Por último, la insensibilidad del cierre del canal 138-373, carente de dominio proximal
a la acidificación extracelular, sugiere que los efectos producidos al eliminar el citado
dominio, podrían ser colaterales a la modificación del posicionamiento de la región inicial
N-terminal respecto al cuerpo central del canal. Así, es posible que al eliminar el dominio
proximal se propicie o estabilice una interacción constante entre la región inicial N-terminal y
la maquinaria de apertura y cierre, que funcionaría como el “interruptor molecular” del cierre
de h-ERG, y que, por ello, se minimice el efecto de la acidificación. Alternativamente, la
ausencia de aceleración del cierre inducida por la acidificación en el canal 138-373, podría
deberse a una interrupción permanente de esa interacción intramolecular, haciendo al canal
incapaz de “sentir” los cambios en el pH extracelular.
Resultados - 97
Figura 31. Efecto de la eliminación de regiones N-terminales sobre la aceleración del cierre de
h-ERG inducida por la acidificación extracelular.
(A) Efecto de la acidificación extracelular en la variación de las constantes de cierre frente al voltaje aplicado a
la membrana (panel superior) o frente al potencial electroquímico [panel inferior; -( G0 -EzgF)], en las distintas
variantes de h-ERG. Para estimar la cinética de desactivación se ajustaron las fases de caída de las corrientes
de cola durante los distintos pulsos repolarizantes a ecuaciones biexponenciales, según lo descrito en los
Métodos. Se muestra, en coordenadas semilogarítimicas, la variación de la constante de tiempo tanto del
componente lento (lenta, triángulos) como del componente rápido (rápida, círculos), obtenidas a pH extracelular
de 7.5 (símbolos blancos) y 6.5 (símbolos negros). (B) Histograma en el que se compara la aceleración del
componente rápido de cierre a -100 mV, inducida por la acidificación extracelular. Solamente se muestran los
datos correspondientes al componente rápido, mayoritario a potenciales muy negativos. Un incremento del
100% significa que la rápida se ha reducido a la mitad. (C) Histograma en el que se compara la aceleración
inducida por la acidificación extracelular para el mismo valor de potencial electroquímico de -4 kcal/mol. Los
parámetros termodinámicos de activación en estado estacionario a pH 7.5 se obtuvieron en la forma indicada
en el apartado I de Resultados. * p<0.05 frente al canal nativo.
98 - Resultados
III.1.2. Efecto de las mutaciones del lazo S4-S5 en la modulación del cierre de
h-ERG por la acidificación extracelular.
Figura 32. Efecto de la acidificación extracelular sobre el cierre de canales h-ERG mutados en el lazo
intracelular S4-S5.
Se muestran familias de corrientes representativas obtenidas de oocitos que expresan los mutantes de h-ERG
indicados en la parte superior de los registros, a un pH extracelular de 7.5 y de 6.5. Para generar las corrientes
de cola a los distintos voltajes de repolarización, se utilizó el protocolo mostrado en la parte superior. Solo se
representan los trazos de corriente obtenidos cada 20 mV y los primeros 3 s de las corrientes de cola, para una
mayor claridad y mejor comparación de la fase de caída de las mismas.
100- Resultados
Figura 33. Efecto de la mutaciones en el lazo S4-S5 sobre la aceleración del cierre de h-ERG inducida
por la acidificación extracelular.
Efecto del acidificación extracelular en la variación de las constantes de tiempo de cierre en función del voltaje
aplicado a la membrana o del potencial electroquímico [-(G0-zgEF)]. Para estimar la cinética de desactivación,
se utilizó el método ya descrito y comentado previamente en la Figura 31. Se muestra, en escala
semilogarítmica, la variación de las constante de cierre tanto del componente lento (lenta, triángulos) como del
componente rápido (rápida, círculos), obtenidas a pH extracelular de 7.5 (símbolos blancos) y 6.5 (símbolos
negros).
Resultados -101
Figura 34. Efecto de la mutaciones en el lazo S4-S5 de h-ERG sobre la aceleración de la constante de
tiempo del componente rápido de cierre ( rápida) inducida por la acidificación extracelular.
(A) Histograma en el que se compara la aceleración del componente rápido de cierre a -100 mV inducida por la
acidificación extracelular. (B) Histograma en el que se compara la aceleración inducida por la acidificación para
el mismo valor de potencial electroquímico de -4 kcal/mol. El número de células utilizado en los promedios se
muestra en el panel A. Los parámetros termodinámicos de activación en estado estacionario a pH 7.5 se
obtuvieron en los estudios ya expuestos en el apartado I de los Resultados. *p<0.05 frente al canal nativo.
Trabajos previos de varios laboratorios, incluido el nuestro, han constatado que las
alteraciones estructurales en la región inicial N-terminal de h-ERG y las mutaciones del lazo
S4-S5, modifican de modo equivalente las características cinéticas del canal (Morais-Cabral
et al., 1998; Wang et al., 1998; Chen et al., 1999; Sanguinetti & Xu, 1999; Gómez-Varela et
al., 2002). Esta correspondencia entre las características de los canales modificados en
ambas regiones intracelulares, ha llevado a proponer la existencia de una interacción física
entre las mismas, la cual explicaría porqué diferentes alteraciones estructurales en estos
dominios alejados en la secuencia proteica, ocasionan el mismo impacto sobre la
funcionalidad.
Una indicación adicional que apoya la existencia de dichas interacciones
intramoleculares, es aportada por los datos recogidos en el apartado I de Resultados, en el
que tras el análisis de las características cinéticas y termodinámicas de variantes de h-ERG
modificadas en ambas regiones, de nuevo se ha demostrado el paralelismo existente entre
los efectos producidos sobre los parámetros cinéticos. De igual forma, hemos observado
una reducción equivalente de la aceleración del cierre inducida por la acidificación
extracelular, tanto en canales carentes de regiones N-terminales como en los mutados en el
lazo S4-S5, sugiriendo de nuevo la posible existencia de un contacto físico entre ambos
dominios proteicos.
Asimismo, y aunque también de modo indirecto, esta hipótesis es apoyada por un
estudio reciente de nuestro laboratorio (Miranda & Manso et al., 2008) en el que, tras un
análisis mediante FRET (“Fluorescence Resonance Energy Transfer”) de canales h-ERG
etiquetados con proteínas fluorescentes derivadas de GFP (“Green Fluorescence Protein”),
se comprobó que la región inicial N-terminal se sitúa más próxima a la membrana que la
región final del extremo carboxilo, proponiéndose que tanto el dominio eag/PAS como el
dominio proximal se localizan cercanos al cuerpo central del canal. Por último, en un estudio
previo de nuestro laboratorio (Viloria et al., 2000), también se ha postulado que la facilitación
de la apertura del canal que provoca la eliminación del dominio proximal de h-ERG, es
debida a una mejor interacción entre el dominio eag/PAS y la maquinaria de apertura-cierre,
situada en el parte central del canal embebida en la membrana plasmática.
V3C -17.9 ± 1.05 (n = 12)* 272.3 ± 15.9 (n = 15)* 46.1 ± 0.01 (n = 9)*
Y542C -5.9 ± 0.55 (n = 20)* 274.5 ± 11.2 (n = 8)* 21.3 ± 2.5 (n = 9)*
Tabla 4. Características cinéticas de los canales h-ERG mutantes V3C e Y542C, y del mutante doble
V3C/Y542C.
Se muestran los parámetros cinéticos de dependencia de voltaje (V1/2) y tiempo (t1/2) de activación
semimáxima, así como la constante de tiempo del componente rápido de cierre (rápida), de los canales h-ERG
con las mutaciones V3C e Y542, y V3C/Y542C, así como del canal nativo. La dependencia de voltaje de
activación se obtuvo despolarizando los oocitos a diferentes voltajes mediante pulsos de 1 s desde un potencial
basal de -80 mV (ver apartado I de Resultados) y después de ajustar los datos a ecuaciones de Boltzmann
según lo descrito en Métodos. Para obtener los valores de t1/2 se utilizó el protocolo indirecto, descrito en el
apartado I de Resultados, con despolarizaciones a 0 mV. Los valores de rápida se obtuvieron ajustando la fase
de caída de las corrientes de cola a -100 mV a ecuaciones biexponenciales (ver Métodos), mediante el protocolo
descrito en el apartado I de Resultados. *p<0.001 respecto al canal nativo.
Se puede observar que, tanto en los mutantes simples como en el mutante doble, se
produce un cierto desplazamiento de las curvas I/V de activación hacia potenciales más
positivos, si bien dicho desplazamiento de la dependencia de voltaje de activación es
mínimo en el canal V3C, y algo mayor en los mutantes Y542C y V3C/Y542C, que muestran
una dependencia de voltaje totalmente análoga. También observamos una cinética de
apertura a 0 mV significativamente más lenta en los mutantes V3C e Y542C, pero similar a
la del canal nativo en el mutante combinado V3C/Y542C.
Por último, y de modo totalmente esperable, dada la conocida influencia del extremo
amino y el lazo S4-S5 en las propiedades de desactivacón de h-ERG, se puede observar
que la cinética de cierre a -100 mV está acelerada en todos los canales mutados. Estos
resultados indican que los canales mutantes y, particularmente, el mutante doble, no
presentan graves distorsiones funcionales que los diferencie ni del canal nativo ni de los
mutantes simples correspondientes.
En la Figura 35, se muestra el efecto de la adición del agente oxidante tbHO2 sobre las
corrientes de cola de h-ERG, en oocitos que expresan tanto los mutantes simples V3C e
Y542C, como el doble mutante V3C/Y542C o el canal nativo. En todos los casos, los
canales se mantuvieron en estado basal cerrado a -80 mV, abriéndolos periódicamente para
su análisis mediante despolarizaciones de 1 s a +40 mV, seguidas de una repolarización
posterior a -100 mV y con una frecuencia de 15 s. La perfusión del medio con el oxidante
tbHO2 2 mM, se hizo de forma continua mediante un flujo lo suficientemente rápido como
para asegurar un recambio total de la solución en la cámara de registro en sólo unos pocos
segundos. Se puede observar en los oocitos que expresan el canal V3C/Y542C, que la
corriente de cola es rápidamente reducida en tan sólo 1 min tras la adición del oxidante. Sin
embargo, a ese tiempo, la corriente fue muy poco reducida en el resto de canales,
manteniéndose en un 94.7% en el canal nativo, y en un 88.2 y 81% en los mutantes Y542C
y V3C, respectivamente. Tras 5 min de perfusión con el oxidante, la corriente se redujo
hasta el 23.5% en oocitos que expresan el doble mutante V3C/Y542C, pero sólo disminuyó
hasta el 80.3, 57.5 y 51.3% en los que expresan los canales nativo, Y542C y V3C,
respectivamente. Estos datos demuestran que, aunque en los mutantes V3C e Y542C se
produce una inhibición de la corriente por el tbHO2 que es algo mayor que la que sufre el
canal nativo, dicha inhibición es mucho mayor e, inicialmente, presenta una cinética mucho
más rápida en el mutante doble V3C/Y542C.
Resultados -105
Figura 35. Efecto de la oxidación de cisteínas introducidas en las posiciones 3 del extremo amino y
542 del lazo S4-S5 de h-ERG.
(A) Corrientes representativas de oocitos que expresan los mutantes simples h-ERG V3C e Y542C, así como el
doble mutante V3C/Y542C y el canal nativo. Los registros corresponden al momento anterior a la adición del
agente oxidante tbHO2 2 mM y a los obtenidos durante los cinco primeros minutos de tratamiento con el
oxidante, a intervalos de 1min. En el inserto de la parte superior de la figura, se muestra el protocolo utilizado
para generar las corrientes. Las corrientes de cola aparecen de forma ampliada en los recuadros. (B) Curso
temporal del efecto del tbHO2 sobre la magnitud de las corrientes de cola. La magnitud de la corriente fue
cuantificada como la diferencia entre el pico inicial y la registrada al final de la repolarización a -100 mV. Los
datos se presentan normalizados respecto al valor de la corriente justo antes de la adición de tbHO2 (ver
Métodos). Nótese la escala semilogarítmica del panel central y los mismos datos representados en escala lineal
en el inserto.
106- Resultados
En conjunto, estos resultados sugieren que las cisteínas en posición 3 del extremo
amino y 542 del lazo S4-S5, se encuentran lo suficientemente próximas como para que sea
posible la formación de un puente disulfuro entre ambas. De hecho, la rapidez y la magnitud
del efecto oxidante inicial del tbHO2 sobre el mutante doble V3C/Y542C parecen superiores
a las esperables si se suman los efectos observados en los dos mutantes simples, V3C e
Y542C. No obstante, la presencia de un cierto efecto del agente oxidante en cada uno de los
mutantes simples por separado, tiende a complicar la interpretación de los resultados. Por
ello, resultó conveniente llevar a cabo una serie de controles adicionales para analizar la
posibilidad de que los resultados con el doble mutante V3C/Y542C no fueran debidos a la
formación de un puente disulfuro entre las últimas, sino a la mera oxidación individual de
ambas, o bien a que dicho puente disulfuro no se estableciera directamente entre ellas, sino
con otras cisteínas ya presentes en el canal y distintas del par considerado.
Como una primera confirmación de que los efectos del tbHO2 son debidos a la
formación de puentes disulfuro entre cisteínas, se comprobó la posible reversión de los
mismos con el agente reductor ditiotreitol (DTT). Este compuesto, tiene un poder reductor lo
suficientemente alto como para poder revertir la formación de puentes disulfuro,
caracterizados por su elevado grado de oxidación (Zahler & Cleland, 1968), aportando una
demostración directa e inequívoca de la formación de dichos enlaces covalentes.
En la Figura 36, se muestran los efectos del DTT sobre la corriente de los canales
V3C, Y542C y V3C/Y542C, así como sobre la del canal nativo, tras su tratamiento con el
oxidante tbHO2. Como se puede observar, el agente reductor causa un rápida reversión del
efecto del oxidante sobre el canal mutante V3C, situando a la corriente de cola a un nivel
similar al observado con el canal nativo tras los mismos tratamientos. Sin embargo, el
agente reductor revierte de forma mínima el nivel de inhibición causado por el oxidante en el
mutante Y542C, que del 39.6% de la corriente presente al final del tratamiento con tbHO2,
se incrementa en un 8% (hasta un 47.6%) en presencia del DTT. Por su parte, el agente
reductor revierte parcialmente la inhibición de la corriente causada por la oxidación en el
doble mutante V3C/Y542C, precisamente hasta el nivel exhibido por el mutante sencillo
Y542C, cuya corriente no es prácticamente revertida en presencia del DTT como acabamos
de indicar.
Estos datos son totalmente coherentes con la posibilidad de que entre la cisteína 3 y la
cisteína 542, ambas introducidas en h-ERG, se establezca un puente disulfuro, ya que el
efecto inhibitorio rápido del oxidante, es revertido totalmente en el doble mutante
V3C/Y542C, hasta el nivel irreversible alcanzado en el mutante sencillo Y542C. Por su
parte, la irreversibilidad observada en el canal Y542C sugiere que, salvo que dicha cisteína
forme un puente disulfuro con otra cisteína nativa de h-ERG situada en un estado no
accesible al DTT, el efecto del oxidante probablemente se deba a la mera modificación de la
cisteína 542 hacia derivados oxidados sulfónicos (-SO3-) y sulfénicos (-SO-). Estos posibles
derivados de la cisteína 542, serían totalmente incapaces per se de formar puentes
Resultados -107
disulfuro, pero su formación probablemente compita con la reacción que da lugar a los
puentes disulfuro entre las cisteínas 3 y 542 del doble mutante. Este hecho sería asimismo
completamente coherente con la lenta tasa de modificación exhibida por el mutante sencillo
Y542C, ya que la velocidad de reacción de formación de derivados no disulfidos es mucho
menor que la de formación de puentes disulfuro (revisado por Bass et al., 2007). Este es
también el caso del otro mutante sencillo V3C, que también muestra una manifiesta lentitud
en el efecto del oxidante sobre la corriente. Sin embargo, la clara reversibilidad del efecto,
en este caso, complica la interpretación de los datos. Por ello, se planteó el estudio de
algunas construcciones adicionales en las que una parte de las cisteínas nativas de h-ERG
hubieran sido eliminadas de determinadas regiones.
Figura 36. Reversión del efecto del tbHO2 en canales con cisteínas introducidas en la posición 3 del
extremo amino y 542 del lazo S4-S5 de h-ERG.
Se muestra, en escala semilogarítmica, la evolución temporal de la magnitud de la corriente de cola
normalizada durante la perfusión con medio oxidante (“tbHO2 2 mM”) y tras la adición de medio reductor con
20 mM DTT. Las corrientes de cola fueron obtenidas aplicando a los oocitos el protocolo de estimulación
indicado en la Figura 35.
Figura 36. Efecto de la óxido-reducción sobre canales h-ERG con cisteínas en el extremo amino y el
lazo S4-S5, y con deleciones en dominios citoplasmáticos para la eliminación de cisteínas nativas.
(A) Evolución temporal de la corriente de cola normalizada durante la perfusión con el medio oxidante (“tbHO2
2 mM”) y reductor (“DTT 20 mM”) de los canales V3C y V3C/864-1010. Para una mejor comparación, también
se incluye la evolución temporal correspondiente a los canales nativo y doble mutante V3C/Y542C.
(B) Evolución temporal de la corriente de cola durante la perfusión con medio oxidante y reductor de los
canales Y542C, Y542C/864-1010 e Y542C/2-370.
110- Resultados
Como se puede apreciar en la Figura 38, la rápida cinética que muestra la inhibición
de la corriente debida a la oxidación en h-ERG V3C/Y542C, cuando los canales
permanecen basalmente cerrados, se ajusta a una ecuación biexponencial con un
componente rápido muy prominente y mayoritario. Sin embargo, la cinética de inhibición de
las corrientes de cola es notablemente más lenta, cuando el tratamiento con el oxidante se
realiza manteniendo los canales abiertos (e inactivos) a un potencial basal de +40 mV
(desde el que las corrientes de cola son obtenidas mediante repolarizaciones sucesivas a
-100 mV a intervalos de 15 s). Esto supone que las constantes de tiempo de los
componentes rápido y lento durante la reducción biexponencial de la magnitud de las
corrientes de cola, son incrementadas desde los 34 ± 0.2 y 327 ± 36 s (n = 13),
correspondientes a las medidas desde un potencial basal de -80 mV (canales cerrados),
hasta los 94.5 ± 24 y 406 ± 62 s, correspondientes a un potencial basal de +40 mV (canales
abiertos/inactivos). Además, al situar los canales abiertos/inactivos a +40 mV, el porcentaje
del componente con desaparición rápida a -80 mV de potencial basal, correspondiente a un
93.7 ± 1% del total, es drásticamente disminuido hasta un 61.2 ± 6%. Por último, dado que,
en estos experimentos, la obtención de las corrientes de cola se llevó a cabo cada 15 s y
que su cuantificación conlleva la repolarización de la membrana a -100 mV con el
consiguiente cierre periódico de los canales, se repitió el análisis de los efectos del oxidante
a +40 mV de potencial basal, pero con menos incursiones a potenciales negativos (una vez
cada dos minutos) para la generación de las corrientes de cola. Esta estrategia provocó una
ralentización aún mayor de la cinética de reducción de las corrientes por el oxidante, hasta
valores muy similares a los obtenidos con cualquiera de los mutantes simples V3C o Y542C.
De hecho, debido al reducido número de valores experimentales generados en estas
condiciones y a la ralentización de la cinética de reducción de las corrientes, en este caso,
Resultados -111
sólo fue posible realizar un ajuste monoexponencial a la cinética, que supuso una caída de
la magnitud de las corrientes mucho más lenta y con una única constante de tiempo de
250 s.
Estos datos claramente demuestran que la reducción de las corrientes del doble
mutante V3C/Y542C, provocada por la oxidación, es mucho más rápida cuando se
mantienen los canales en estado cerrado durante el máximo tiempo posible. Como se
discutirá posteriormente, estos resultados sugieren que las cisteínas introducidas en las
posiciones 3 y 542, se encuentran mucho más próximas o en una configuración mucho más
favorable para la formación de un puente disulfuro, cuando el canal se encuentra en el
estado cerrado que cuando h-ERG se sitúa en estado abierto.
Como hemos indicado en los apartados anteriores, los resultados obtenidos hasta el
momento no permiten excluir completamente la posibilidad de que los efectos del agente
oxidante tbHO2 sobre el doble mutante V3C/Y542C no sean debidos, en parte, a la simple
oxidación de las cisteínas en ambas posiciones, o a la formación de puentes disulfuro entre
cada una de ellas y alguna otra cisteína nativa presente en la secuencia de h-ERG. Para
aportar una prueba adicional de la especificidad de los efectos de la oxidación para el par
V3C/Y542C, se planteó la posibilidad de buscar una pareja alternativa para la cisteína 3 en
el lazo S4-S5, introduciendo la mutación V3C en el mutante G546C, el cual tiene una
cisteína reemplazando a una glicina en la zona C-terminal del lazo S4-S5, generándose así
el doble mutante V3C/G546C.
En la Figura 39, se muestra el efecto de la adición del agente oxidante tbHO2, sobre la
magnitud de las corrientes de cola del mutante doble V3C/G546C. Se puede observar que la
caída de la corriente de este doble mutante es prácticamente idéntica que la del mutante
sencillo V3C. Además, por otro lado, la reducción de la corriente del mutante simple G546C
por efecto del oxidante es mínima e igual a la que muestra el canal nativo. Por tanto, parece
que la cinética de oxidación exhibida por el mutante V3C/G546C es debida sólo y
exclusivamente a la presencia de la mutación sencilla V3C. Estos resultados claramente
demuestran que, cuando se introducen dos cisteínas en posiciones 3 y 542 de h-ERG, se
obtiene una inhibición por el oxidante tbHO2 amplia y particularmente rápida, que no se
obtiene cuando las cisteínas se introducen en posiciones 3 y 546. Así pues, existe una
especificidad de oxidación cuando la cisteína 3 se combina con la 542, lo cual afianza la
idea de la proximidad entre la región inicial del extremo amino y la zona N-terminal del lazo
intracelular S4-S5.
En resumen, este grupo de resultados indican que existe una proximidad física entre el
inicio del extremo amino y el lazo intracelular S4-S5, situado en el cuerpo central del canal.
Suponen, además, un inicio muy prometedor para explorar con más profundidad las
interacciones intramoleculares en el canal h-ERG, que son cruciales tanto para su
funcionalidad como para su regulación hormonal.
112- Resultados
Figura 38. Cinética y dependencia de estado de los efectos de la oxidación sobre h-ERG con cisteínas
introducidas en el extremo amino y en el lazo S4-S5.
(A) Evolución temporal de la magnitud de las corrientes de cola, tras la adición del agente oxidante tbHO2 a
canales V3C/Y542C. La evolución temporal, a un potencial basal de -80 mV, de las corrientes de cola de los
mutantes V3C e Y542C, y del canal nativo, se muestran mediante líneas punteadas, para una mejor
comparación. En el inserto, se muestran los mismos datos, en representación semilogarítmica. Las curvas de
inhibición de la corriente se ajustaron a ecuaciones exponenciales. (B) Resumen de los parámetros cinéticos de
las curvas de inhibición de la corriente. Se muestra el potencial basal (“P.B.”) al que se mantuvieron los
oocitos, la frecuencia (“f”) de las repolarizaciones a -100 mV que generan las corrientes de cola, las constantes
de tiempo de los componentes rápido y lento de caída de las ecuaciones biexponenciales y de cada una de las
monoexponenciales (rápida, lenta), y el porcentaje de componente rápido de la cinética inhibición biexponencial
(“% de comp. rápido”).
Resultados -113
Figura 39. Efecto de la oxidación de cisteínas introducidas en las posiciones 3 del extremo amino y
546 del lazo S4-S5 de h-ERG.
Se muestra el curso temporal de la magnitud de las corrientes de cola durante la perfusión con medio oxidante
(“tbHO2 2 mM”) del doble mutante de V3C/G546, así como de cada uno de los mutantes simples V3C y G546C.
También se incluyen las curvas correspondientes al canal nativo y el doble mutante V3C/Y542C, para una mejor
comparación. Los datos se muestran normalizados a la corriente registrada justo antes de la adición del
oxidante y en escala semilogarítmica.
114- Resultados
Discusión
Discusión -115
también en otros que muestran una marcada aceleración del cierre (e.g. 2-16, 2-135,
2-370, R541H, R541A y Y542C).
En estudios previos sobre variantes de h-ERG con alteraciones en el dominio eag/PAS
del extremo amino, se ha demostrado el papel fundamental de esta región en la cinética de
cierre (Morais-Cabral et al., 1998; Wang et al., 1998, 2000; Chen et al., 1999). Por otro lado,
en otros trabajos, incluidos los de nuestro grupo, se ha demostrado el papel que en la
cinética de activación de h-ERG juega el dominio proximal, situado entre el eag/PAS y la
hélice S1 (Viloria et al., 2000; Gómez-Varela et al., 2002; Saenen et al., 2006). Sin embargo,
hasta este trabajo, no se había reconocido que la porción inicial aminoterminal de h-ERG
jugara ningún papel en la activación del canal. Nuestros resultados demuestran la
importancia de este segmento inicial de la proteína en la apertura de h-ERG. Así, al
modificar esta región, se produce un desplazamiento de la dependencia de voltaje de
activación hacia valores más positivos y una reducción del valor de z g en la curva I/V de
estado estacionario, haciendo que el valor de G0 sea menos negativo, es decir,
desplazando el equilibrio conformacional hacia los estados cerrados. Hemos obtenido las
características termodinámicas del estado estacionario para comparar las velocidades de
apertura de los canales en las mismas condiciones energéticas, observando que todos los
canales modificados en la región inicial del extremo amino presentan tasas de activación al
menos dos veces más rápidas que las del canal nativo, si se comparan a potenciales
electroquímicos equivalentes. Teniendo en cuenta esta apertura más rápida y que, además,
todos estos canales presentan acelerado el cierre, podemos concluir que las modificaciones
del extremo amino estabilizan el estado de transición entre los estados de abierto y de
cerrado, disminuyendo la barrera energética entre ambos (G‡), acelerando así la apertura y
el cierre del canal. Es importante destacar que nuestro análisis no permite diferenciar cuales
de las transiciones individuales durante la activación de h-ERG son afectadas por las
modificaciones estructurales, tan sólo permite identificar regiones proteicas implicadas en el
proceso global de apertura y cierre. Es importante asimismo considerar que se está
estudiando un proceso dependiente de voltaje y, por ello, la velocidad del mismo estará
directamente relacionada con la cantidad de energía (potencial electroquímico) que lo dirige.
Por tanto, hay que asegurarse que las diferencias en las tasas de apertura y cierre no son
sólo una consecuencia de los desplazamientos en la dependencia de voltaje,
comparándolas a una energía equivalente.
En lo referente al dominio proximal del extremo amino de h-ERG, nuestro grupo había
demostrado su papel modulador sobre el proceso de apertura, no considerando relevante en
ese momento su participación en el cierre del canal (Viloria et al., 2000). Sin embargo, en
este trabajo de Tesis, hemos observado una clara aceleración del cierre en canales carentes
de dominio proximal, cuando se compensan las tasas de cierre por la diferencia en el
potencial electroquímico. Es importante hacer el tratamiento de los datos con precaución, ya
que al compensar la cinética de cierre por la energía que dirige la activación, se asume que
el proceso de cierre es una reversión del de apertura con idénticas transiciones en ambos
sentidos, suponiendo entonces que el cierre está acoplado al movimiento del sensor de
voltaje, lo que es una cuestión aún controvertida. Así pues, las constantes de tiempo de
cierre se deberían de corregir por los valores de z g y V1/2 de activación, solamente si el cierre
está acoplado a los cambios conformaciones del VSD. Teniendo en cuenta que las
alteraciones detectadas en la activación y en el cierre se producen al modificar el dominio
Discusión -117
proximal y la región inicial N-terminal, respectivamente, esto parece indicar que ambos
procesos no son una simple reversión uno del otro, ya que están determinados por dominios
proteicos diferentes. De cualquier forma, y dado que se producen alteraciones comunes
sobre la apertura y sobre el cierre al modificar ambos dominios del extremo amino, se podría
especular que tanto la apertura como el cierre son procesos relacionados, acoplados a
reorganizaciones estructurales en el sensor de voltaje, a pesar de que pueden estar
ampliamente influenciados por otras regiones de la proteína.
Nuestro análisis cinético y termodinámico bajo condiciones de equilibrio, aporta una
información muy importante para identificar las regiones del canal que participan en la
maquinaria de apertura/cierre, permitiendo conocer la estabilidad relativa de los estados
conformacionales del canal bajo diferentes condiciones. Se observa, por ejemplo, que las
variantes de h-ERG con modificaciones en la región inicial del extremo amino, presentan un
desplazamiento de las curvas I/V de estado estacionario hacia potenciales positivos. Como
consecuencia, estos canales se caracterizan por un valor de G0 menos negativo, indicando
un desplazamiento del equilibrio conformacional hacia los estados de cerrado. Por el
contrario, la variante carente de dominio proximal presenta valores de V1/2 y de G0 mucho
más negativos, indicando que el equilibrio está desplazado en sentido contrario, esto es,
hacia los estados de abierto. De cualquier forma, cuando se tienen en cuenta en ambos
tipos de canales, las diferencias en el potencial electroquímico [i.e. -(G0-z gEF) o las
diferencias energéticas entre los estados abierto y cerrado], se observa en ambos casos una
clara aceleración de la velocidad de apertura. Esto sugiere que ambas regiones del extremo
amino contribuyen a una serie de interacciones químicas específicas implicadas en la
elevación de la barrera energética de activación (G‡), es decir, de la energía necesaria
para alcanzar el estado de transición entre las conformaciones de abierto y cerrado.
Se ha descrito que los efectos moduladores del dominio proximal sobre la apertura de
h-ERG, se deben a la secuencia de aminoácidos básicos 362-366 (Saenen et al., 2006), y a
su influencia electrostática sobre la maquinaria de apertura/cierre del canal. A pesar de ello,
permanece por esclarecer si estas interacciones electrostáticas se dan de forma directa con
las estructuras que dirigen la apertura y el cierre (e.g. el VSD, el lazo S4-S5 o la “compuerta”
de la hélice S6), o de una forma indirecta mediante alguna estructura intermediaria (e.g. el
dominio eag/PAS). De acuerdo con nuestros resultados, tanto en los canales sin dominio
proximal como en los carentes de todo el extremo amino, se produce una facilitación similar
de la apertura para valores equivalentes de potencial electroquímico. Teniendo en cuenta
que los efectos que dichas alteraciones ejercen sobre las tasas de activación y cierre son
similares y no aditivos (ver características de 2-370; Figuras 7C y 8C), se podría especular
con la necesidad de una correcta ubicación del dominio eag/PAS (especialmente de los
primeros 16 residuos) respecto al cuerpo central del canal, no sólo para un cierre normal,
sino también para la activación. Como consecuencia, la organización estructural del extremo
amino propiciaría la adecuada orientación del dominio eag/PAS hacia el cuerpo central del
canal. De esta forma, si se altera dicha organización, se afectaría la correcta colocación de
la región inicial y, con ello, se modificarían las características de activación del canal. Esto
podría explicar el efecto dominante de la deleción de los 16 primeros residuos del extremo
amino frente a la eliminación del dominio proximal. Así, el desplazamiento de la
dependencia de voltaje hacia valores negativos del canal 138-373, es revertido cuando se
118- Discusión
eliminan ambas regiones, el dominio proximal y la porción inicial del extremo amino. De ahí
que se observe una dependencia de voltaje de activación similar en las variantes 2-16,
2-135 y 2-370. Es decir, al eliminar el dominio proximal se produce un desplazamiento
negativo de la dependencia de voltaje sólo si se conservan los 16 primeros aminoácidos de
la proteína. De esta forma, al eliminar el extremo amino al completo (i.e. variante 2-370) y,
por ello, el dominio proximal y los 16 primeros aminoácidos, se produce un desplazamiento
de la dependencia de voltaje hacia potenciales positivos, que es similar al producido cuando
se eliminan tan sólo estos primeros 16 residuos o todo el dominio eag/PAS.
Como ya se ha indicado, hace tiempo que se propone que la lenta cinética de cierre de
h-ERG se debe a una interacción de la región inicial del extremo amino con el lazo S4-S5
(Morais-Cabral et al., 1998; Wang et al., 1998; Chen et al., 1999; Sanguinetti & Xu, 1999;
Gómez-Varela et al., 2002). Dicho lazo también se ha propuesto como el elemento de
acoplamiento mecánico entre el movimiento del VSD en respuesta a cambios en el voltaje y
la maquinaria de apertura/cierre en el dominio poro (Sanguinetti & Xu,, 1999; Tristani-Firouzi
et al., 2002; Piper et al., 2005; Ferrer et al., 2006). Asumiendo que exista una interacción
física entre el inicio del extremo amino y el lazo S4-S5, que pueda ser importante para
modular el proceso de activación y cierre, se podría esperar que la alteración de cualquiera
de ambas regiones produjese efectos análogos sobre las propiedades del canal. Además,
también se esperaría que dichos efectos no fuesen aditivos. Aunque algunas de las
propiedades en estado estacionario de los mutantes del lazo S4-S5 dependen de la posición
que ocupe el residuo mutado, se observan claros desplazamientos positivos en la
dependencia de voltaje del estado estacionario en los mutantes D540C, Y542C, E544C e
Y545C. Como consecuencia, se obtiene un valor de G0 menos negativo respecto al canal
nativo, indicando que el equilibrio conformacional está desplazado hacia los estados de
cerrado. También es común a estos cuatro mutantes la notable aceleración que muestran
tanto en la cinética de activación como en la de cierre. Dado que estas características
también aparecen en las variantes con modificaciones en la región inicial del extremo amino,
parece que realmente existe una correspondencia entre el efecto de las modificaciones
aminoterminales y el de las mutaciones en el lazo S4-S5.
Adicionalmente, también se estudió el efecto de combinar mutaciones en el lazo S4-S5
con deleciones del extremo amino completo. Sorprendentemente, en todos los casos
estudiados (R541A/2-370, Y542C/2-370 y G546C/2-370), el comportamiento de estos
canales doblemente modificados, no se corresponde con el de sus relativos mutantes
puntuales. En principio, si se asume que los efectos de las mutaciones en el lazo S4-S5 se
deben a un entorpecimiento del contacto físico entre éste y la región inicial aminoterminal,
no se esperaría que el efecto de las mutaciones puntuales fuese distinto con o sin extremo
amino. Estos datos no contribuyen a afianzar la idea de la existencia de interacciones entre
ambos dominios, sin embargo no contradicen necesariamente que dicha interacción pueda
existir, ya que es posible que las mutaciones del lazo S4-S5 no sean neutras, sino que
afecten por sí solas a la maquinaria de apertura/cierre, dado el supuesto papel del citado
lazo como elemento de acoplamiento entre el VSD y el dominio poro. Es necesario pues, ser
muy cauto en la interpretación de las características de estos canales respecto a la
existencia de interacciones entre ambos dominios citoplasmáticos.
En base a lo anteriormente mencionado, sería razonable pensar que, para mantener
las propiedades de activación de h-ERG, es importante preservar una organización
Discusión -119
adecuada de todo el extremo amino, que permita una correcta orientación de su porción
inicial respecto al cuerpo central del canal. Así, podríamos pensar que al eliminar el dominio
proximal se produce una reorganización del todo el extremo amino, que reposiciona su
región inicial hacia el cuerpo central del canal, facilitando su apertura; una hipótesis que ya
se ha propuesto en anteriores trabajos (Viloria et al., 2000; Gómez-Varela et al., 2002).
Asumiendo pues que la región inicial N-terminal ha de tener un posicionamiento adecuado
para su interacción con el lazo S4-S5, es posible que la alteración del lazo entorpezca la
interacción entre ambos dominios. Esto podría explicar de nuevo porqué el desplazamiento
de la dependencia de voltaje de activación del canal 138-373, carente de dominio proximal,
va en sentido contrario al que muestran los canales 2-16, 2-135 y 3-370. Para validar
esta posibilidad, se estudiaron también las características de las mutaciones del lazo S4-S5
pero, en este caso, en canales carentes del dominio proximal (datos no mostrados). De
nuevo, y a pesar de que en ciertos mutantes y en determinados parámetros, parece haber
un efecto dominante de la deleción frente a la mutación, en otros casos, se da el fenómeno
contrario o el efecto es nulo. Por tanto, al igual que en el caso de los mutantes carentes del
extremo amino completo, las características de los canales doblemente modificados en el
dominio proximal y el lazo S4-S5 son distintas a las de los canales modificados, bien por
mutación o por deleción. Esto confirma la idea de que las mutaciones en el lazo S4-S5
alteran la maquinaria de apertura/cierre de h-ERG por sí mismas, si bien la facilitación de la
apertura producida al eliminar el dominio proximal domina, en algunos casos, sobre las
mutaciones en el lazo S4-S5.
Un caso particularmente interesante es el comportamiento de los mutantes D540C e
Y542C. En ambos, aparte de los cambios en su cinética de apertura, también se observa
una notable reducción del valor de zg, obteniéndose un perfil energético casi plano en la
activación de estos canales mutantes. Este perfil se caracteriza no sólo por una escasa
tendencia termodinámica a producirse la transición de cerrado a abierto, en ausencia de un
campo eléctrico aplicado (i.e., G0 ~ 0), sino también por una baja energía de activación
(i.e. un reducido valor de G‡ del estado de transición entre los estados cerrado y abierto).
Además, el canal mutante D540C presenta una dependencia de voltaje muy reducida en sus
tasas de activación. Esto concuerda con un trabajo anterior en el que se mostraba cómo al
neutralizar la carga negativa en la posición 540 de h-ERG, aparece un gran componente de
“corrientes de compuerta” a potenciales negativos a los cuales el canal aún no conduce
corriente (Piper et al., 2005), es decir, se produce un desplazamiento de las curvas de
activación hacia potenciales negativos en lo que respecta a la relación Q/V (carga movida
frente a voltaje del pulso despolarizante) y hacia potenciales positivos en la relación G/V
(conductancia frente a voltaje, equivalente a las curvas I/V). Junto con nuestros resultados
respecto a las variaciones en los parámetros de activación y cierre, todo ello sugiere que la
mutación D540C hace que el movimiento del VSD se desacople de la apertura del canal
iónico en el dominio poro. Este fenómeno de desacoplamiento también se ha descrito en
varios trabajos con el canal Shaker, mutado en diferentes regiones de la proteína: la hélice
S4 (Ledwell & Aldrich, 1999; Pathak et al., 2005), el lazo S4-S5 (Schoppa & Sigworth, 1998)
y el extremo de la hélice S6 (Ding & Horn, 2003). En este último caso, parece que el
movimiento del sensor de voltaje es incapaz de acoplarse a la apertura de la “compuerta”.
La explicación de ello supone que al mutar el final de la hélice S6 que constituye la
“compuerta”, ésta se aleja del lazo S4-S5, permitiendo un movimiento más fácil de la hélice
120- Discusión
Con el fin de delimitar aún más la región del dominio proximal esencial para que tenga
lugar la regulación por TRH de la activación de h-ERG, hemos generado una nueva
construcción con una deleción intermedia entre las de los canales 326-373 y 345-373, en
la que se ha eliminado la región 333-373. Al estudiar el comportamiento de esta variante
frente al tratamiento hormonal, se comprobó que presentaba el mismo comportamiento que
el canal nativo. Teniendo en cuenta que al aumentar el tamaño de la deleción en otros 7
residuos en la variante 326-373 (respecto al canal 333-373) se pierden los efectos
reguladores de la TRH, se consideró que la secuencia 326-332 era imprescindible para la
regulación hormonal de la apertura del canal. La secuencia 326-332 (RYRTISK) contiene
varios consensos para modificación covalente y/o unión de segundos mensajeros,
albergando varios sitios de fosforilación para diferentes proteín quinasas y para la unión de
proteínas de familia 14-3-3 o de fosfatidil inositol-4,5-bisfosfato (PIP2), todos los cuales ya
habían sido propuesto como agentes moduladores de las corrientes ERG (Kiehn et al.,
1998; Heath & Terrar, 2000; Kiehn 2000; Thomas et al., 1999, 2003; Cui et al., 2000, 2001;
Bian et al., 2001, 2004; Cayabyab et al., 2002; Cayabyab & Schlichter, 2002; Kagan et al.,
2002; Zhang et al., 2003; Bian & McDonald, 2007; Cockerill et al., 2007; Li et al., 2008;
Zhang et al., 2008). Sin embargo, sorprendentemente, al eliminar en el segmento 326-332
mediante mutagénesis dirigida (de forma individual o combinada) los sitios susceptibles de
ser fosforilados, no se produjo ninguna modificación de los efectos hormonales de la TRH
sobre la actividad del canal. Además, al substituir las argininas 326 y 328, cuyas cargas
positivas podrían ser determinantes para la unión de PIP2, además de fundamentales para
la unión de proteínas de la familia 14-3-3, tampoco se observaron diferencias en la
regulación hormonal respecto al canal nativo. Estos resultados demuestran que la pérdida
de los efectos hormonales observada al incrementar el tamaño de la deleción (retirando
desde el residuo 333 hasta el 326), no se debe a la eliminación de la secuencia diana para
algún agente regulador.
122- Discusión
No obstante, los datos parecen indicar que la supresión de los efectos hormonales se
relaciona con el tamaño de la deleción del dominio proximal, que ocasiona alguna
modificación estructural en el extremo amino de h-ERG. Esta indicación está claramente
respaldada por el hecho de que las variantes de h-ERG en las que se elimina la secuencia
RYRTISK y sus alrededores (canal 326-341), o en las que se substituye por otra secuencia
no relacionada con la nativa (canales 326-341/SubA y 326-341/SubB), presentan una
regulación hormonal totalmente normal.
Una vez demostrada la irrelevancia de la secuencia 326-332 (RYRTISK) de h-ERG
para su regulación hormonal por TRH, nos planteamos la posibilidad de que los canales no
regulados fuesen aquellos en los que las alteraciones estructurales modifican per se los
parámetros cinéticos del canal. Este podría ser el caso de las variantes en las que se ha
eliminado desde el segmento 326-373 hasta la totalidad del dominio proximal, las cuales
presentan una alteración de las propiedades de apertura que podría explicar su incapacidad
para ser reguladas por la hormona. Así, se podría explicar la ausencia de regulación
hormonal de la apertura en las variantes 326-373, 284-373, 223-373 y 138-373, las
cuales presentan una activación marcadamente acelerada y una dependencia de voltaje
claramente desplazada hacia potenciales negativos (Viloria et al., 2000; Gómez-Varela et
al., 2003a). Esta hipótesis también sería coherente con el hecho de que el control hormonal
de la activación en el canal 333-373 sea normal, teniendo en cuenta que en este canal la
apertura está sólo ligeramente ralentizada y con un pequeño desplazamiento de la curva I/V
hacia potenciales positivos. También se podría explicar así la ausencia de regulación sobre
el componente rápido de cierre en el canal 2-135, en el cual el cierre ya está basal
acelerado.
Sin embargo, este razonamiento no es en absoluto válido para los canales 345-373 y
355-373, que también muestran una marcada facilitación de la apertura, acelerada y con
un prominente desplazamiento de la dependencia de voltaje hacia potenciales negativos,
pero que son regulados normalmente por la TRH (Gómez-Varela et al., 2003a). Tampoco
resulta válido en el caso del canal 155-209, que presentando unos parámetros de
activación similares a los del canal nativo, muestra unos efectos hormonales de la TRH muy
reducidos. No parece existir pues una relación entre la modificación cinética inducida por
una alteración estructural, y el hecho de que del canal no pueda ser regulado. Para
demostrar que la citada relación no existe, está el caso de los mutantes en el lazo
intracelular S4-S5. En primer lugar, ninguno de estos canales mutantes presenta una
modificación de los parámetros de activación semejante a la encontrada en las deleciones
del dominio proximal (i.e. facilitación de la apertura) y, sin embargo, los canales con
mutaciones en la porción inicial del lazo S4-S5 presentan una regulación hormonal de la
activación nula o muy reducida. En segundo lugar, a pesar de que el mutante Y542C y,
sobre todo, el D540C, presentan una notable alteración de la activación (desplazamiento de
la curva I/V hacia potenciales positivos así como un decremento en su pendiente), apenas
son modulados por la TRH. Tampoco lo son los mutantes R541H y R541A, los cuales
presentan una apertura muy similar a la del canal nativo. Una aparente excepción a la
independencia entre la alteración basal y ausencia de regulación hormonal se observa en la
modulación del componente rápido de cierre, ya que todos los mutantes que lo presentan
basalmente acelerado, se caracterizan por una regulación hormonal nula, o marcadamente
Discusión -123
reducida (D540C, Y542C, R541A/H, E544C é Y545C). Nótese sin embargo que, al contrario
de lo que ocurre con la aceleración basal de la apertura, que puede compensarse midiendo
las tasas a potenciales para los cuales la velocidad es comparable a la del canal nativo, para
analizar la velocidad de cierre tras la adición de TRH, es necesario estimarla a potenciales
negativos para los cuales la contaminación residual de las corrientes de cloro activadas por
la hormona es nula. Como consecuencia, las velocidades de cierre se comparan para el
mismo voltaje sin compensar las diferencias en la aceleración basal que presenten los
canales mutantes respecto al canal nativo. Esto complica el poder distinguir claramente si la
ausencia de regulación sobre el cierre de estos canales se debe a la aceleración basal de
este proceso, o a la alteración del control hormonal.
Una interpretación alternativa a las expuestas hasta ahora para nuestros resultados,
sería que la ausencia de control hormonal fuese debida a que los canales presentan
alteraciones estructurales que provocan la desestabilización de algún dominio citoplasmático
y/o su orientación respecto a la maquinaria de apertura y cierre del canal. Los resultados
obtenidos con el canal 155-209 respaldarían esta hipótesis, ya que dicho canal, con una
deleción de un tamaño semejante a la del canal 326-373 (55 aminoácidos para la primera y
48 para la segunda) y situada en el extremo opuesto del dominio proximal, muestra unos
efectos de la TRH sobre la activación también muy reducidos.
Por su parte, los efectos hormonales sobre la apertura del canal 2-135, que carece
del dominio eag/PAS, son normales respecto a los observados en el canal nativo. Estos
resultados apuntan a que sólo las modificaciones en el dominio proximal, y no en el dominio
eag/PAS, perturban la regulación hormonal de la apertura de h-ERG. Como se expuso
anteriormente, debido a la aceleración del cierre que presenta esta variante carente de
extremo amino, es difícil comprobar si este proceso sigue siendo regulado por la TRH.
Por otro lado, algunos estudios recientes (Saenen et al., 2006) han enfatizado el papel
de la secuencia positivamente cargada 362-366 (KIKER) del dominio proximal en la cinética
de apertura de h-ERG. Al analizar si la secuencia KIKER era relevante para el control
hormonal de la apertura de h-ERG, comprobamos que, tras anular las cargas positivas de
esta secuencia mediante mutagénesis (KKR-A) o mediante la deleción 363-373, los
efectos hormonales de la TRH son similares a los del canal nativo.
Por último, al estudiar la regulación hormonal de la activación en canales con
mutaciones en el lazo S4-S5, se pudo comprobar que los efectos hormonales eran nulos o
marcadamente reducidos en aquellos mutantes en los que el residuo modificado se
localizaba en la porción N-terminal del citado lazo. Esta reducción de los efectos hormonales
se produce no sólo en mutantes que presentan alterado el proceso de apertura (D540C é
Y542C), sino también en los canales en los que los parámetros de activación son similares a
los del canal nativo (R541H y R541A).
En resumen, el conjunto de nuestros resultados sugiere que, para que tenga lugar la
regulación hormonal de la activación de h-ERG, es necesario tanto una correcta
organización de la estructura de los dominios del extremo amino, como un posicionamiento
adecuado de éstos respecto al cuerpo central del canal. Además, nuestros datos han
permitido identificar la región N-terminal del lazo S4-S5 como un determinante esencial para
la transmisión de los efectos reguladores de los receptores acoplados a la Fosfolipasa C,
desde los dominios citoplasmáticos hasta el cuerpo central del canal h-ERG.
124- Discusión
Como se ha venido indicando a lo largo de esta sección, los resultados de este trabajo
nos llevan a proponer la existencia en h-ERG de interacciones entre la región inicial del
extremo amino y el lazo intracelular S4-S5, que son fundamentales para el normal
funcionamiento del mecanismo de apertura y cierre. Además, el mantenimiento de la
organización de los dominios aminoterminales parece conllevar una orientación y/o
posicionamiento del extremo amino hacia el cuerpo central del canal, a nivel del lazo S4-S5,
que es necesaria también para la regulación hormonal por la TRH. A pesar de ello, no se
han aportado pruebas directas de la existencia de tales interacciones entre el inicio del
extremo amino y el lazo S4-S5.
Como se ha indicado anteriormente, se ha postulado que la aceleración del cierre de
h-ERG al disminuir el pH extracelular, se debe a la protonación de residuos cargados
negativamente de la región S1-S3, los cuales establecerían interacciones electrostáticas con
residuos de arginina de la hélice S4. El mantenimiento de dichos puentes salinos se vería
favorecido por una interacción proteína-proteína entre el dominio eag/PAS del extremo
amino y el lazo intracelular S4-S5, presumiblemente constituyendo un “interruptor molecular”
que controla la cinética de cierre del canal (Jiang et al., 1999; Liu et al., 2003). En base a
este postulado, en este trabajo, para estudiar de una forma indirecta dichas interacciones,
nos planteamos analizar las contribuciones relativas del extremo amino (particularmente del
dominio eag/PAS y del dominio proximal) y del lazo S4-S5 de h-ERG, a la aceleración del
proceso de cierre que induce la acidificación extracelular.
aspartato 411 y una carga negativa cercana, que desestabiliza la(s) conformación(es) de
cerrado. Como resultado, al disminuir el pH, se favorecería la activación del canal, al
producirse un desplazamiento de la dependencia de voltaje hacia potenciales negativos y,
quizá, también al acelerarse la apertura (Liu et al., 2003). A este respecto, nuestros
resultados sugieren que el efecto de la acidificación sobre la apertura es únicamente
resultado de la modificación de la carga superficial. Así, se obtienen desplazamientos de las
curvas I/V similares en todos los canales, tanto en el nativo como en aquellos con
modificaciones aminoterminales o en el lazo S4-S5. Además, el desplazamiento en el V1/2 al
disminuir el pH externo, es equivalente en el canal nativo y en la variante sin dominio
proximal, la cual muestra una fuerte desestabilización de los estados de cerrado, con un
marcado desplazamiento de su dependencia de voltaje de activación hacia potenciales
negativos, y una notable aceleración basal de la velocidad de apertura (ver apartado I de
Resultados; Viloria et al., 2000; Gómez-Varela et al., 2002).
central del canal y muy cercano tanto al lazo S4-S5 como a la “compuerta” limitada por la
parte final de la hélice S6 (Miranda & Manso et al., 2008).
Desafortunadamente, aunque muy sugerentes, todos estos datos no han aportado una
prueba directa de la existencia de una interacción física entre el extremo amino y el lazo
S4-S5. En un trabajo relativamente reciente (Ferrer et al., 2006), se ha indicado que entre el
residuo 540 del lazo S4-S5 y los residuos 665/666 del extremo de la hélice S6, existe una
proximidad suficiente como para que entre ellos se formen puentes disulfuro, tras introducir
en estas posiciones cisteínas y añadir un agente oxidante. Basándonos en este tipo de
abordaje experimental, nos planteamos el estudio de la proximidad espacial entre los
primeros residuos del extremo amino y en la región N-terminal del lazo S4-S5 de h-ERG,
mediante la introducción en esas regiones de cisteínas y la formación de puentes disulfuro
por oxidación con tert-butilhidroperóxido (tbHO2). Es previsible que, en el caso de que estos
enlaces covalentes se formen entre dos puntos cuya interacción dinámica sea esencial para
la actividad del canal, se produzca alguna alteración funcional medible, como, por ejemplo,
la eliminación de las corrientes mediadas por el canal (Ferrer et al., 2006).
Nuestros resultados demuestran que al introducir por mutagénesis dos cisteínas en la
posición 3 (V3) del extremo amino y en el residuo 542 (Y542) del S4-S5, la adición del
agente oxidante tbHO2 a oocitos que expresan el doble mutante V3C/Y542C, provoca una
rápida disminución de las corrientes h-ERG. Esta eliminación de la corriente es
prácticamente nula en el canal h-ERG nativo. Además, el curso temporal de la disminución
de la corriente sigue una cinética biexponencial con un marcado y predominante
componente rápido inicial. Esta rápida cinética es totalmente coherente con una reacción de
formación de puentes disulfuro, la cual se produce a una tasa extremadamente rápida
cuando dos cisteínas se encuentran a la distancia y orientación adecuadas, y son sometidas
a condiciones oxidantes, debido al aumento de la probabilidad de colisión entre los grupos
sulfhidrilo y a la reactividad de los mismos (revisado por Bass et al., 2007). Así pues, dadas
las restrictivas condiciones necesarias para favorecer la formación de puentes disulfuro, los
resultados obtenidos con el doble mutante V3C/Y542C indican que entre estas dos cisteínas
existe una gran proximidad, y que sus cadenas laterales están orientadas en un ángulo
correcto como para que se formen dichos enlaces covalentes.
Es importante indicar que cuando se introducen cisteínas aisladamente en las
posiciones 3 ó 542 de h-ERG, también se produce una reducción de las corrientes por
efecto del oxidante tbHO2. Sin embargo, dicha inhibición es mucho menos marcada y no
ocurre de forma tan rápida como en el doble mutante, ajustándose básicamente a una
ecuación monoexponencial con una lenta constante de tiempo. Mientras la rápida cinética
del doble mutante es una buena indicación de la formación de un puente disulfuro, la lenta
tasa exhibida por los mutantes sencillos resulta más difícil de interpretar. Así, cuando el
grupo sulfhidrilo (-SH) de la cadena lateral de una cisteína se somete a condiciones
oxidantes, da lugar a un derivado reactivo del tipo tiolato (-S-). Si dos de estos grupos se
encuentran a una distancia de menos de 4.6 Å entre sus respectivos carbonos ,
presentando además la orientación angular adecuada, podrían formar puentes disulfuro
entre ambos (revisado por Bass et al., 2007). No obstante, si se mantienen las condiciones
oxidantes en el tiempo, y debido a las fluctuaciones térmicas que puede experimentar el
esqueleto hidrocarbonato de la proteína, dos cisteínas separadas una mayor distancia (de
incluso 15 Å) pueden llegar a colisionar en su forma reactiva y establecer puentes disulfuro,
Discusión -127
aunque de forma mucho más lenta (revisado por Bass et al., 2007). Así, el efecto del agente
oxidante sobre los mutantes individuales V3C e Y542C, se podrían explicar por la formación
de derivados reactivos tipo tiolato en sus cadenas laterales que no son capaces de formar
puentes disulfuro rápidamente con otros próximos, pero que, con el tiempo y debido a las
fluctuaciones de la proteína, podrían acabar formándolos, produciendo la inhibición de la
corriente de cola. Esto explicaría la lenta cinética monoexponencial de caída de las
corrientes de los canales V3C e Y542C, así como la presencia del componente lento,
minoritario, en la caída biexponencial del mutante doble V3C/Y542C. De hecho, es posible
que en este mutante doble V3C/Y542C, la adición del tbHO2 podría ocasionar que en la gran
mayoría de canales se formasen puentes disulfuro entre las posiciones 3 y 542, mientras
que en una pequeña población, se podrían formar puentes disulfuro alternativos de cada
cisteína con otras próximas. Así, podría existir cierto grado de competencia entre la rápida
formación de puentes disulfuro entre el par 3 y 542, favorecida por su proximidad, y la lenta
reacción de formación de puentes disulfuro alternativos, dando lugar al componente rápido y
lento de caída de la corriente, respectivamente.
ya que una escasa accesibilidad de este puente disulfuro secundario al DTT es complicada
de conciliar con la clara accesibilidad mostrada al agente oxidante.
De las 23 cisteínas endógenos presentes en el canal h-ERG nativo, 5 se encuentran
repartidas entre el lazo extracelular S1-S2 y las hélices S5 y S6, insertadas en la membrana
plasmática, mientras que existen 18 cisteínas situadas en dominios intracelulares y que son
posibles candidatas para formar puentes disulfuro con las introducidas en las posiciones 3 y
542. Tras combinar la mutación V3C con la deleción 864-1010 y comprobar que los efectos
del agente oxidante y reductor eran análogos a los observados en el mutante sencillo V3C,
parece claro que la pareja de la cisteína 3 para formar un posible puente disulfuro
secundario no es ninguna de las 2 presentes en la región del extremo carboxilo eliminada.
Teniendo en cuenta que la inhibición de la corriente conlleva, en último término, un bloqueo
de la maquinaria de permeación en el cuerpo central del canal, podríamos pensar que el
efecto del oxidante sobre el canal V3C se debe a la lenta formación de un puente disulfuro
entre la cisteína 3 y alguna otra situada en la proximidad de la maquinaria de permeación,
para influir sobre ella. De esta forma, de las 16 restantes posibles cisteínas que se podrían
emparejar con la 3, las máximas candidatas podrían ser alguna de las 4 presentes en la
región que une el final de la hélice S6 con el dominio cNBD (“C-linker”). Aunque
indirectamente, esta idea estaría apoyada por el modelo, propuesto por nuestro laboratorio,
para la arquitectura de los dominios citoplasmáticos de h-ERG, que sitúa al dominio
eag/PAS rodeando al conjunto de hélices perpendiculares a la membrana que constituye el
“C-linker” del canal, de modo análogo a lo propuesto para otros canales (Miranda & Manso
et al., 2008).
También hemos comprobado que el efecto del oxidante y el reductor sobre la corriente
de los canales en los que la mutación Y542C se combina con la deleción del extremo
carboxilo 864-1010 o con la deleción del amino 2-370, es similar al producido sobre el
mutante sencillo Y542C. De esta manera, hemos podido descartar 12 cisteínas endógenas
de h-ERG como posibles parejas para la lenta formación de puentes disulfuro con la 542: las
2 eliminadas con la deleción 864-1010 y otras 10 eliminadas con la 2-370. Por tanto, de
las 18 posibles cisteínas endógenas del canal que pudieran ser candidatas para formar
puentes disulfuro, sólo quedarían 6 después de éste análisis. Dichas cisteínas se
encontrarían en el “C-linker” (4) y la región final del extremo carboxilo (2). Es interesante que
la irreversibilidad del efecto del oxidante por el DTT observada con el mutante sencillo
Y542C, también se observa en los mutantes combinados con las deleciones, indicando que
la falta de accesibilidad del agente reductor, o el alto grado de oxidación alcanzado por la
cisteína 542, es insensible a la eliminación de las citadas regiones amino y
caboxiloterminales. La escasa reversibilidad de la oxidación, junto con la lenta cinética de
caída de la corriente, apoyan la hipótesis de que en la cisteína 542, el grupo sulfhidrilo de la
cadena lateral, es oxidado a radicales sulfónicos o sulfénicos, con un estado de oxidación
que no puede ser revertido por el DTT.
se formen derivados sulfónicos y/o sulfénicos de la cisteína 542, que, por su alto grado de
oxidación, no son reducidos por el DTT. En ambos casos, los procesos siguen una cinética
que se ajusta a ecuaciones monoexponenciales con una lenta constante de tiempo. Por el
contrario, en el mutante doble V3C/Y542C, debido a la proximidad y orientación correcta de
las cisteínas 3 y 542, se establecen puentes disulfuro entre ambos residuos en un proceso
muy rápido que originaría la rápida caída de las corrientes de cola. Muy probablemente, las
reacciones lentas de formación de puentes disulfuro secundarios entre la cisteína 3 y otra(s)
endógena(s) del canal, así como de formación de derivados altamente oxidados de la
cisteína 542, competirían en alguna medida con la rápida reacción de formación de puentes
disulfuro entre las cisteínas 3 y 542, dando lugar al componente lento de caída de la
corriente del mutante doble V3C/Y542C.
Una prueba adicional que confirma el establecimiento de puentes disulfuro entre las
cisteínas 3 y 542, es la dependencia de estado de la inhibición de la corriente en el mutante
doble V3C/Y542C. Así, la cinética de caída de la corriente, cuando los canales se mantienen
basalmente abiertos, se ajusta a una ecuación biexponencial, con unas constantes de
tiempo mucho mayores que las obtenidas cuando los canales se mantienen basalmente
cerrados. Estos datos indican que la cisteína 3 y la 542 se encuentran mucho más próximas
y/o mejor orientadas para la formación de puentes disulfuro, cuando el canal se presenta
preferentemente en su conformación cerrada.
En un principio, la mayor eficiencia en la formación de los puentes disulfuro en el
estado cerrado, parece contradecir la hipótesis mantenida hasta el momento, que propone
que la interacción entre el extremo amino y el lazo S4-S5 contribuye a ralentizar el cierre de
h-ERG, o que la facilitación de la apertura al eliminar el dominio proximal, se debe a que se
favorece la interacción entre el dominio eag/PAS y el cuerpo central del canal (Viloria et al.,
2000). Sin embargo, la disposición exacta de los residuos implicados en la formación de los
puentes disulfuro, puede variar en los canales mutantes y/o con modificaciones estructurales
o cinéticas. Además, el hecho de que la máxima proximidad entre el inicio del extremo
amino y el lazo S4-S5 se produzca predominantemente en la conformación cerrada del
mutante V3C/Y542C, no excluye la posibilidad de que en el canal nativo dicha proximidad se
alcance en la conformación abierta. Teniendo en cuenta que la cinética de cierre del
mutante V3C/Y542C es muchísimo más rápida que la del canal nativo, parece que, en esta
variante, la conformación cerrada está estabilizada frente a la de abierta. Por tanto, el
mutante V3C/Y542C tiene una mayor tendencia a mantenerse cerrado que el canal nativo y,
quizá por ello, la región aminoterminal pueda mantenerse más próxima al lazo S4-S5 en
esta conformación.
En cualquier caso, la existencia de diferencias detectables en la eficiencia de
formación de puentes disulfuro entre las cisteínas 3 y 542, entre los canales en estado
predominantemente abierto o en estado cerrado, reflejan las diferencias en la conformación
del canal en ambos estados, por lo que pueden constituir, además, una forma de analizar
los cambios conformacionales que sufre el canal durante el proceso de apertura y cierre.
el cuerpo del canal, y una línea de residuos polares en la cara opuesta. Sin embargo, en
h-ERG, el lazo S4-S5 no tiene tal naturaleza anfipática, ya que, en él, los residuos polares
se sitúan en la primera mitad del lazo, mientras que los no polares lo hacen en la segunda
mitad de su secuencia (Figura 40A, B). Por otro lado, la introducción de la arginina 541 de
h-ERG en la posición correspondiente a una glicina en Kv1.2, hace que en el modelo de
homología se produzca una colisión entre las hélices, ya que la cadena lateral voluminosa
de la arginina no encaja en el hueco correspondiente a la glicina de Kv1.2. Asimismo, se ha
descrito una interacción directa entre el aminoácido D540 y los aminoácidos R665/L666,
localizados en el extremo C-terminal de la hélice S6, en la conformación cerrada de h-ERG
(Ferrer et al., 2006). Sin embargo, en el modelo de homología de h-ERG, tal interacción es
imposible, dada la enorme distancia entre estos residuos, debida a la ausencia del motivo
PVP en la hélice S6 y, por tanto, a la curvatura que en Kv1.2 tiende a aproximar el extremo
carboxilo de la hélice S6 a la superficie interior de la membrana.
Por último, los efectos relativamente similares en la funcionalidad de h-ERG,
observados al mutar la secuencia DRY de la región N-terminal del lazo S4-S5, son difíciles
de correlacionar con el dispar posicionamiento espacial de estos residuos en los modelos.
Sería pues necesario clarificar si esto es debido a una orientación de las cadenas laterales
de estos aminoácidos hacia la misma superficie de la proteína (a diferencia de lo esperado
en base a los modelos), o bien si la modificación de cualquiera de estos residuos produce
una alteración funcional análoga, aún cuando se encuentren dirigidos hacia diferentes partes
de la proteína, como predicen los modelos de homología. Finalmente, es importante hacer
notar que ambos modelos aportan escasa información acerca de la localización espacial y la
organización estructural de los extremos amino y carboxilo de h-ERG, algo que es
sumamente relevante dada la importancia funcional de estas regiones y de sus posibles
interacciones con las estructuras del cuerpo central transmembranal, puestas de manifiesto
a lo largo de este trabajo.
Como resumen final, podríamos concluir que los extremos citoplasmáticos del canal
h-ERG, particularmente el extremo amino, participan de modo crucial en el control de las
propiedades de activación y desactivación del canal, y constituyen un factor determinante
para la transmisión de los efectos reguladores sobre la apertura y el cierre en respuesta a la
activación hormonal de la Fosfolipasa C. En ambos casos, nuestros resultados indican que
una interacción física entre la región inicial de extremo amino de la proteína y la porción final
de la hélice S4 o la inicial del lazo S4-S5, parece constituir la base molecular de la
regulación funcional de la maquinaria de apertura/cierre, localizada en el cuerpo central
transmembranal del canal.
132- Discusión
A
S4 S5 S6
Kv1.2 KLSRHSKGLQILGQTLKASMRELGLLIFFLFIG....GYGDMVPTTIGGKIVGSLCAIAGVLYIALPVPVIVSNFNYFYHR
306 419
MlotiK KPLRDSTFFPVLGRVLANEARNLIGVTTLFGVV....GYGDTIPQSFAGRVLAGAVMMSGIGIFGLWAGILATGFYQEVRR
107 220
h-ERG RVARKLDRYSEYGAAVLFLLMCTFALIAHWLAC....GFGNVSPNTNSEKIFSICVMLIGSLMYASIFGNVSAIIQRLYSG
534 669
Discusión -133
Figura 40. Modelos de homología de h-ERG basados en las estructuras cristalinas de Kv1.2 y MlotiK.
(A) Alineamiento de las secuencias, en código aminoacídico de una letra, de la región correspondiente al lazo
intracelular S4-S5 y al segmento transmembranal S6 de los canales Kv1.2, MlotiK y h-ERG. Se resaltan en
negrita los residuos con carga negativa situados en la región C-terminal del segmento transmembranal S4, el
motivo GYGD, o el equivalente en h-ERG, del poro de permeación, las glicinas de la hélice S6, que le confieren
cierta flexibilidad, y el motivo PVP, que produce la curvatura característica de la S6 en Kv1.2. Los aminoácidos
del lazo S4-S5 se muestran en color rojo. Para una mejor visión se omiten los residuos que abarcan desde la
segunda mitad de la hélice S5 hasta el GYGD del poro de permeación (motivo H5). Los residuos del lazo S4-S5
que interaccionan con sus “parejas” de las hélices S5 y S6 de la misma subunidad o de la adyacente en la
estructura de Kv1.2, se representan en color azul y verde, respectivamente, en la secuencia de Kv 1.2. El
aminoácido D540 del lazo S4-S5 y los residuos R665 y/o L666 del segmento S6 de h-ERG, entre los cuales se
ha sugerido que pueda existir una interacción directa (Ferrer et al., 2006), se muestran como subrayados.
(B) Estructura tridimensional de la región correspondiente al lazo S4-S5 de una de las subunidades de K v1.2
(izquierda, ID en el PDB: 2a79) y modelo de homología obtenido al intercambiar los residuos del citado lazo de
Kv1.2 por sus homólogos de h-ERG (derecha). Las cadenas laterales de los aminoácidos de las hélices S4 y S5
de Kv1.2 se representan con bastones y aquellas cadenas de estos residuos que tienen carga positiva se
representan, además, en rojo. Los aminoácidos del lazo S4-S5 de ambas estructuras se representan con
esferas. (C) Representación con cintas del tetrámero del modelo de homología, en el que se ha substituido la
región comprendida desde el final del segmento S4 hasta el inicio S6 (i.e. el lazo S4-S5 y sus alrededores) de
h-ERG en la estructura de Kv1.2 (izquierda) o de MlotiK (derecha; ID en el PDB: 3co2). El lazo S4-S5 con las
cintas de la hélice correspondiente al mismo, se remarca en color rojo, a excepción de una de las subunidades
en la que se representa la superficie total de la estructura proteica correspondiente a dicho lazo. Nótese que el
motivo PVP, presente en el segmento S6 de Kv 1.2, produce una curvatura que sitúa dicho segmento justo por
debajo del lazo S4-S5. Ese motivo está ausente en MlotiK (y en h-ERG) haciendo que dicho segmento se
proyecte hacia el citosol, de forma casi perpendicular a la membrana. (D) Visión ortogonal ampliada de la
región correspondiente al lazo S4-S5 (cintas rojas) de una de las subunidades mostradas en C, en la cual se
representan los contactos con el segmento S6 de la misma subunidad (rosa) y el S5 de la subunidad adyacente
(verde). Las cadenas laterales de los residuos del lazo S4-S5 y de los segmentos S5 y S6 que interaccionan con
dicho lazo, se representan como bolas y bastones. Se muestra superpuesta y transparente la superficie global
del lazo S4-S5. Nótese que en el modelo basado en Kv1.2 (izquierda) se produce una colisión entre la arginina
541 de h-ERG y la hélice S5 de la subunidad adyacente (no mostrado), que no existe en Kv1.2, en el que el
residuo correspondiente a la arginina 541 de h-ERG es una glicina. Nótese también que ni en el modelo basado
en Kv1.2, con el motivo PVP presente, ni en el basado en MlotiK (derecha), sin PVP, se predice la interacción
propuesta entre el residuo D540 del lazo S4-S5 y los residuos R665/L666 del final del segmento S6 (Ferrer et
al., 2006). Las figuras se obtuvieron utilizando el programa “USF Cimera ” y los residuos de K v1.2 y MlotiK se
substituyeron por sus homólogos de h-ERG utilizando el programa “Swiss-Pdb Viewer”.
134- Discusión
Conclusiones
Conclusiones -135
3. Los residuos aspartato 540 y tirosina 542, situados en la región N-terminal del
lazo intracelular S4-S5 de h-ERG, están involucrados en el acoplamiento funcional
entre el movimiento del sensor de voltaje y la maquinaria de apertura y cierre del
canal.
10. El residuo 3 del extremo amino y el 542 de la región N-terminal del lazo
S4-S5 de h-ERG, se encuentran lo suficientemente próximos como para que se
formen entre ellos puentes disulfuro, de forma dependiente de estado, cuando ambos
aminoácidos son substituidos por cisteínas y la proteína es sometida a condiciones
oxidantes.
11. La existencia de interacciones entre los primeros residuos del extremo amino
y la región N-terminal del lazo S4-S5, parece esencial para el mantenimiento de la
particulares propiedades de apertura y cierre del canal h-ERG y para su regulación
hormonal por receptores acoplados a proteínas G.
Referencias
bibliográficas
Referencias bibliográficas -137
A n um onw o J .M., H o rta J., D el mar M., T a ffe t S .M., J a lif e J . (1999). “Proton and
zinc effects on HERG currents”. Biophys. J. 7 7: 282-298.
Ba ker O .S., Larss on H .P ., M an n u zzu L.M., Isac off E .Y. (1998). “Three
transmembrane conformations and sequence-dependent displacement of the S4 domain
in Shaker K+ channel gating”. Neuron. 2 0: 1283-1294.
Bar ros F., G óm ez- Var el a D., Vi to ria C.G., Pal o me ro T ., Gir áld e z T., d e l a
Peña P. (1998). “Modulation of human erg K+ channel gating by activation of a G
protein-coupled receptor and protein kinase C”. J. Physiol. 5 1 1: 333-346.
B eza nill a F. (2008). “How membrane proteins sense voltage”. Nat. Rev. Mol. Cell. Biol.
9: 323-332.
Bia n J.S ., Ka ga n A., Mc D on ald T.V . (2004). “Molecular analysis of PIP2 regulation
of HERG an IKr”. Am. J. Physiol. Heart Circ. Physiol. 2 87: H2154-H2163.
Bia nc h i L ., Wib le B., Ar ca ng eli A., Tag l iala tela M., M orr a F., Cas ta ld o P .,
Cr oci an i O., R osa ti B., Far av ell i L., Ol ivo tto M ., Wa n ke E . (1998). “HERG
encodes a K+ current highly conserved in tumors of different histogenesis: a selective
advantage for cancer cells?”. Cancer Res. 5 8: 815-822.
Bi ng g eli R., W ei ns tei n R .C. (1986). “Membrane potentials and sodium channels:
hypotheses for growth regulation and cancer formation based on changes in sodium
channels and gap junctions”. J. Theor. Biol. 1 2 3 : 377-401.
Ca yab yab F.S ., Sc hli ch ter L .C. (2002). “Regulation of an ERG K+ current by Src
tyrosine kinase”. J. Biol. Chem. 2 77: 13673-13681.
Ca yab yab F.S., Ts ui F.W ., Sc hlic h ter L .C . (2002). “Modulation of the ERG K+
current by Tyrosine Phosphatase, SHP-1”. J. Biol. Chem. 2 27: 48130-48138.
C ha A., S ny der G.E ., Selvi n P .R., Be zan i lla F . (1999). “Atomic scale movement of
the voltage-sensing region in a potassium channel measured via spectroscopy”. Nature.
402: 809-813.
C hi esa N., R osa ti B ., Arc an g eli A ., O liv o tto M ., Wa n ke E. (1997). “A novel role
for HERG K+ channels: spike-frequency adaptation”. J. Physiol. 5 01: 313-318.
Cla y ton G .M., Al ti eri S ., H egi nb o tha m L., Ung er V .M ., Mo rais- C abra l J .H .
(2008). “Structure of the transmembrane regions of a bacterial cyclic nucleotide-
regulated channel”. Proc. Natl. Acad. Sci. USA. 1 05: 1511-1515.
140- Referencias bibliográficas
C oc kerill S .L ., T obi n A .B ., T orr eci lla I., Will ars G .B., Sta nde n N.B .,
Mi tc h es on J.S . (2007). “Modulation of hERG potassium currents in HEK-293 cells by
protein kinase C. Evidence for direct phosphorilation of the pore forming subunits”. J.
Physiol. 5 8 1: 479-493.
C oetzee W .A., A mar ill o Y., C hi u J., C how A ., La u D., Mc C or mac k T.,
M or eno H ., Nadal M .S., O za ita A., P o u ntn ey D., Sag a nic h M., Ve ga- Sa e n z
de M ie ra E ., Rud y B . (1999). “Molecular diversity of K+ channels”. Ann. NY Acad.
Sci. 8 68: 233-285.
C ui J., M el ma n Y.F ., Pal ma E., F ish ma n G.I., Mc Do n ald T.V . (2000). “Cyclic
AMP regulates the HERG K+ channel by dual pathways”. Curr. Biol. 1 0: 671-674.
Em mi A ., W en zel H.J., Sc hwa rtzkr oi n P.A., T agl ial atel a M., Cas ta ldo P.,
Bia nc h i L ., Nerb on ne J ., Ro berts o n G.A., Ja ni gr o D . (2000). “Do glia have
heart? Expresión and functional role for ether-a-go-go currents in hippocampal
astrocytes”. J. Neurosci. 2 0: 3915-3925.
Ferr er T ., R upp J ., Pi per D.R ., Tr istan i-Fir o u zi M. (2006). “The S4-S5 linker
directly couples voltage sensor movement to the activation gate in the human ether-à-go-
go-related gene (hERG) K+ channel”. J. Biol. Chem. 2 81: 12858-12864.
Fon ta na L., D’ A mic o M ., Cr oci an i O., B iagi o tti T ., So la zzo M ., Ros ati B .,
Arc an g eli A., W a n ke E., O liv o tto M. (2001). “Long-term modulation of HERG
channel gating in hypoxia”. Biochem. Biophys. Res. Commun. 2 8 6: 857-862.
Ga nd hi C .S., C lar k E., L o ots E., Pr all e A., Is aco ff E .Y. (2003). “The orientation
and molecular movement of a K(+) channel voltaje-sensing domain”. Neuron. 4 0: 515-
525.
Góm ez- V ar ela D ., de la P e ña P., Gar cí a J ., Gir áld ez T., Ba rros F . (2002).
“Influence of amino-terminal structures on kinetic transitions between several closed
and open status in human erg K+ channels”. J. Membr. Biol. 1 87: 117-133.
142- Referencias bibliográficas
H ea th B.M ., T erra r D .A. (2000). “Protein kinase C enhaces the rapidly activating
delayed rectifier potassium current, IKr, through a reduction in C-type inactivation in
guinea-pig ventricular myocytes”. J. Physiol. 5 2 2: 391-402.
H ol mgr en M ., Ju rma n M.E ., Yel le n G . (1996). “N-type inactivation and the S4-S5
region of the Shaker potassium channel”. J. Gen. Physiol. 1 08: 195-206.
H os hi T., Za go tta W .N., A ldric h R .W. (1991). “Two types of inactivation in Shaker
K+ channels: effects of alterations in the carboxy-terminal region”. Neuron. 7 : 547-556.
Isac off E.Y., J a n Y.N., J a n L.Y. (1991). “Putative receptor for the cytoplasmic
inactivation gate in the Shaker K+ channel”. Nature. 3 53: 86-96.
Kemp B .E., P earson R.B . (1990). “Protein kinase recognition sequence motifs”.
Trends Biochem. Sci. 1 5: 342-346.
Kieh n J., Lac erd a A .E., Br ow n A.M . (1999). “Pathways of HERG inactivation”.
Am. J. Physiol. 2 77: H199-H210.
Lai n é M., Li n M .C., Ba n nis ter J.P., Sil ver ma n W .R., M oc k A .F ., Ro u x B.,
Papa zi an D .M. (2003). “Atomic proximity between S4 segment and pore domain in
Shaker potassium channels”. Neuron. 3 9: 467-481.
144- Referencias bibliográficas
Lam ber ts S .W., M acL e od R .M. (1990). “Regulation of prolactin secretion at the level
of the lactotroph”. Physiol. Rev. 7 0: 279-318.
Lecc h i M., Re da elli E ., Ros ati B ., G urr o la G ., Fl ori o T ., C roc ia ni O., Cu ria
G., Cass uli ni R .R., Ma si A., Arc an g eli A ., Ol iv otto M ., Sc h etti ni G., P oss an i
L.D ., Wa n ke E. (2002). “Isolation of a long-lasting eag-related gene-type K+ current in
MMQ lactotrophs and its accommodating role during slow firing and prolactin release”.
J. Neurosci. 2 2: 3414-3425.
Led well J.L ., Aldr ic h R.W . (1999). “Mutations in the S4 region isolate the final
voltage dependent cooperative step in potassium channel activation”. J. Gen. Physiol.
113: 389-414.
Lu Z., K lem A.M ., R am u Y. (2002). “Coupling between voltage sensor and activation
gate in voltage-gated K+ channels”. J. Gen. Physiol. 1 20: 663-676.
Ma n n u zzu L.M ., M oron n e M.M ., Is ac off E.Y. (1996). “Direct physical measure of
conformational rearrangement underlying potassium channel gating”. Science. 2 71: 213-
216.
Ma rti n T.F., Hsi e h K- P., P or ter B .W . (1990). “The sustained second phase of
hormone-stimulated diacylglycerol accumulation does not activate protein kinase C in
GH3 cells”. J. Biol. Chem. 2 95: 7623-7631.
Mi ra nda P., M ans o D .G ., B arr os F., Car re ter o L., H u gh es T.E ., Al o nso- R o n
C., Domí ng u e z P ., d e la P e ña P. (2008). “FRET with multiply labelled HERG K+
channels as a reporter of the in vivo coarse architecture of the cytoplasmic domains”.
Biochim. Biophys. Acta-Mol. Cell. Res. 178 3: 1681-1699.
146- Referencias bibliográficas
Pa ul ussen A.D., R aes A., Jo n gbl o ed R.J., Gil isse n R.A ., Wild e A .A.,
S nyd ers D.J., Sm eets H.J ., Aerssens J. (2005). “HERG mutation predicts short
QT bases on channel kinetics but causes long QT by heterotetrameric trafficking
deficiency”. Cardiovasc. Res. 6 7: 467-475.
Referencias bibliográficas -147
Papa M., B osc ia F., C a nita no A., C asta ldo P ., S el li tti S., A n n u n zia to L .,
Tagl ial atel a M. (2003). “Expression pattern of the ether-a-go-go-related (ERG) K+
channel-encoding genes ERG1, ERG2 and ERG3 in the adult rat central nervous
system”. J. Comp. Neurol. 4 6 6: 119-135.
Pill ozzi S ., Bri zzi M .F ., Bal zi M ., Crocia ni O., C h er ubi ni A., G uas ti L. ,
Bar tol o zzi B., B ecc h etti A ., Wa n ke E., B er nab ei P.A ., O liv o tto M ., P e go rar o
L., Arca ng el i A. (2002). “HERG potassium channels are constitutively expressed in
primary human acute myeloid leukemias and regulate cell proliferation of normal and
leukemic hemopoietic progenitors”. Leukemia. 1 6: 1791-1798.
Poss on D.J., G e P ., Mill er C., B e za nil la F., Se lvi n P.R . (2005). “Small vertical
movement of a K+ channel voltage sensor measured with luminescence energy transfer”.
Nature. 4 36: 848-851.
R eic hli n S. (1985). “Regulation of prolactin secretion at the level of the lactotroph”.
Wilson J.D., & Foster D.W. (eds). William textbook of endocrinology. Saunders,
Philadelphia. pp. 492-548.
R osa ti B ., Marc h etti P ., C roc ia ni O., Lecc h i M., L upi R ., A rca ng el i A.,
Oliv otto M., W a n ke E. (2000). “Glucose- and arginine-induced insulin secretion by
human pancreatic beta-cells: the role of HERG K(+) channels in firing and release”.
FASEB J. 1 4: 2601-2610.
R udy B., Iv ers o n L.E . (eds., 1992). “Ion channels”. Methods Enzymol. 2 07.
Sa enen J .B., Lab ro A.J., R aes A ., S n yd ers D .J. (2006). “Modulation of HERG
gating by a charge cluster in the N-terminal proximal domain”. Biophys. J. 9 1: 4381-
4391.
Sa ng ui n e tti M .C., Xu Q.P. (1999). “Mutations of the S4-S5 linker alter activation
properties of HERG potassium channels expressed in Xenopus oocites”. J. Physiol. 5 1 4:
667-675.
Sc hön h err R., H ei n em a nn S.H . (1996). “Molecular determinants for activation and
inactivation of HERG, a human inward rectifier potassium channel”. J. Physiol. 4 9 3:
635-642.
Sc hön h err R ., R os ati B., H eh l S., Ra o V .G ., A rca ng eli A., O liv o tto M .,
H ei ni em a nn S.H ., Wa n ke E . (1999). “Functional role of the slow activation property
of ERG K+ channels”. Eur. J. Neurosci. 1 1: 753-760.
Sm ith P.L., Ba u kr ow itz T., Yell e n G. (1996). “The inward rectification mechanism
of the HERG cardiac potassium channel”. Nature. 3 79: 833-836.
Sm ith P .L ., Yell en G. (2002). “Fast and slow voltage sensor movements in HERG
potassium channels”. J. Gen. Physiol. 1 19: 275-293.
Sp ector P.S., Cu rra n M.E ., K eati ng M.T., Sa n gu i ne tti M.C . (1996a). “Class III
antiarrhythmic drugs block HERG, a human cardiac delayed rectifier K+ channel. Open-
channel block by methanesulfonanilides”. Circ. Res. 7 8: 499-503.
Stor ey N.M., O’ Br ya n J.P., Ar ms tro n g D.L . (2002). “Rac and Rho mediate
opposing hormonal regulation of the ether-a-go-go-related potassium channel”. Curr.
Biol. 1 2: 27-33.
Tagl ial atel a M., C as tal do P., I oss a S ., P an nac ci on e A ., F r esi A., F ic ker E.,
A n nu n zi ato L. (1997). “Regulation of the human ether-a-go-go related gene (HERG) K
channels by reactive oxygen species”. Proc. Natl. Sci. USA. 9 4: 11698-11703.
Tagl ial atel a M., P a n na cci one A ., I oss a S ., C as tald o P., A n n u n zia to L . (1999).
“Modulation of the K+ channels encoded by human ether-a-go-go-related gene-1 (hERG1)
by nitric oxide”. Mol. Pharmacol. 5 6: 1298-1308.
Tas hji an A.H . Jr ., Yasa m ura Y., L ev in e L., S ato G .H., Par ker M.L. (1968).
“Establishment of clonal strains of rat pituitary tumor cells that secrete growth
hormone”. Endocrinology. 8 2: 342-352.
Titus S.A ., War m ke J.W., G an e tzky B. (1997). “The Drosophila erg K+ channel
polypeptide is encoded by the seizure locus”. J. Neurosci. 1 7: 875-881.
Tru deau M .C., W arm ke J .W ., Ga n etzky B., Ro be rts o n G .A. (1995). “HERG, a
human inward rectifier in the voltage-gated potassium channel family”. Science. 2 69 :
92-95.
Vil lal ob os C., G arcí a-S an ch o J . (1995). “Capacitive Ca2+ entry contributes to the
Ca2+ influx induced by thyrotropin-releasing hormone (TRH) in GH3 pituitary cells”.
Pflügers Arch. 4 30: 923-935.
W ond er li n W.F., Str obl J .S. (1996). “Potassium channels, proliferation and G1
progression”. J. Membr. Biol. 154: 91-107.
Ya rov- Yar ovo y V., Ba ker D ., Ca tte rall W.A . (2006). “Voltage sensor
conformations in the open and closed states in ROSETTA structural models of K+
channels”. Proc. Natl. Acad. Sci. USA. 1 03: 7292-7297.
Yell en G . (2002). “The voltage-gated potassium channels and their relatives”. Nature.
419: 35-42.
Yu F .H., Ya rov- Yar o voy V., G u tm an G.A ., Ca tterall W .A. (2005). “Overview of
molecular relationships in the voltage-gated ion channel superfamily”. Pharmacol. Rev.
57: 387-395.
Z ei P.C ., Ald ric h R .W. (1998). “Voltage-dependent gating of single wild-type and S4
mutant KAT1 inward rectifier potassium channels”. J. Gen. Physiol. 1 12: 679-713.
Z hou W., Cay ab yab F .S., P e nn ef ath er P .S., S ch lic h ter L.C ., De C o urs ey T .E .
(1998). “HERG-like K+ channels in microglia”. J. Gen. Physiol. 1 11: 781-794.
Carlos Alonso-Ron, Pilar de la Peña, Pablo Miranda, Pedro Domı́nguez, and Francisco Barros
Departamento de Bioquı́mica y Biologı́a Molecular, Edificio Santiago Gascón, Campus del Cristo, Universidad de Oviedo,
E-33006 Oviedo, Asturias, Spain
ABSTRACT Gating kinetics and underlying thermodynamic properties of human ether-a-go-go-related gene (HERG) K1
channels expressed in Xenopus oocytes were studied using protocols able to yield true steady-state kinetic parameters. Channel
mutants lacking the initial 16 residues of the amino terminus before the conserved eag/PAS region showed significant positive
shifts in activation voltage dependence associated with a reduction of zg values and a less negative DGo, indicating a deletion-
induced displacement of the equilibrium toward the closed state. Conversely, a negative shift and an increased DGo, indicative of
closed-state destabilization, were observed in channels lacking the amino-terminal proximal domain. Furthermore, accelerated
activation and deactivation kinetics were observed in these constructs when differences in driving force were considered,
suggesting that the presence of distal and proximal amino-terminal segments contributes in wild-type channels to specific
chemical interactions that raise the energy barrier for activation. Steady-state characteristics of some single point mutants in the
intracellular loop linking S4 and S5 helices revealed a striking parallelism between the effects of these mutations and those of the
amino-terminal modifications. Our data indicate that in addition to the recognized influence of the initial amino-terminus region on
HERG deactivation, this cytoplasmic region also affects activation behavior. The data also suggest that not only a slow movement
of the voltage sensor itself but also delaying its functional coupling to the activation gate by some cytoplasmic structures possibly
acting on the S4-S5 loop may contribute to the atypically slow gating of HERG.
INTRODUCTION
The human ether-a-go-go-related gene (HERG) encodes a librium and/or steady-state conditions to obtain voltage de-
potassium channel that mediates the cardiac repolarizing cur- pendencies and for derivation of thermodynamic parameters
rent IKr (1,2). ERG potassium channels play a key role setting is of primary interest for at least two reasons: 1), it provides a
the electrical behavior of a variety of cell types (3–10). Mal- model- and researcher-independent way to quantify the ki-
function of HERG currents is responsible for the type 2 long netic characteristics, and 2), it allows for an accurate com-
QT syndrome (1,11–16), a clinical disorder characterized by parison of these characteristics among different channel
prolongation of the QT interval on the electrocardiogram as- constructs that could help to understand the impact of a given
sociated with an increased risk of life-threatening arrhythmia residue and/or protein domain on channel functionality. How-
(11,17,18). The critical determinant of HERG physiological ever, disparate V½ values for activation of different HERG
roles is its atypical kinetic behavior, characterized by slow mutants, sometimes differing by several tens of millivolts,
activation kinetics and a very fast voltage-dependent inacti- have been reported using so-called fully activated or steady-
vation process on depolarization. In addition, a fast recovery state activation curves (2,23–33). As previously emphasized
from inactivation followed by a much slower deactivation (22,34,35), the slow rates of HERG activation and deacti-
process take place on repolarization, helping to maintain the vation at voltages around the V½ values of the activation
channels open before closing at negative potentials. This curves make the use of very long pulses necessary to reach a
makes HERG operate as an inward rectifier because little complete steady state. On the other hand, such slowness will
outward current passes through it during depolarization, al- shift the voltage dependence of activation, making impossi-
though an increased current is induced during repolarization, ble an accurate comparison of two channel mutants showing
even though the driving force for potassium decreases at different activation time courses, unless the steady-state
negative voltages under physiological conditions (19–23). condition is fulfilled.
It is important to understand the activation and deactiva- As for other voltage-dependent potassium channels, the
tion gating kinetics to properly interpret the contribution of main determinant of HERG voltage sensitivity is the occur-
HERG channels to cell function. In this case, use of equi- rence of conformational changes in the voltage-sensing do-
main contributed by the S1-S4 transmembrane segments
(29–33,36–38). However, it has been also recognized that
Submitted July 5, 2007, and accepted for publication December 28, 2007. other protein regions can contribute to the gating machinery
Address reprint requests to Dr. Francisco Barros, Departamento de and influence the activation and deactivation transitions.
Bioquı́mica y Biologı́a Molecular, Edificio Santiago Gascón, Campus del Thus, an interaction of the N-terminus with the S4-S5 linker
Cristo, Universidad de Oviedo, E-33006 Oviedo, Asturias, Spain. Tel.: has been proposed as a determinant of HERG slow deacti-
34-985103565; Fax: 34-985103157; E-mail: fbarros@uniovi.es. vation (19,23,39), although direct proof of such a physical
Editor: Francisco Bezanilla.
Ó 2008 by the Biophysical Society
0006-3495/08/05/3893/19 $2.00 doi: 10.1529/biophysj.107.116731
3894 Alonso-Ron et al.
coupling between these structures is still lacking. It has also ments were digested with BstEII/BglII and ligated into BstEII/BglII-digested
been proposed that an interaction of the initial eag domain wild-type HERG in the psP64A1 vector. To combine S4-S5 loop mutations
and the D2-370 amino-terminal deletion, BstEII/BglII restriction fragments
with the channel core could be involved in facilitation of from each of the S4-S5 channel mutants were used to replace the corre-
opening after elimination of the proximal domain in the sponding segments in HERG deleted construct D2-370. All constructs were
amino terminus of HERG (34). Electrostatic interactions sequenced to confirm the mutations and to ensure the absence of introduced
between a small section of the proximal domain (close to the errors.
S1 helix) and the gating machinery (possibly at the level of
the S4-S5 loop), have been proposed as a determinant of the
modulatory effects of this domain on gating (33). Finally, a Expression of HERG channel variants and
role of the S4-S5 linker coupling the voltage sensor move- current recording conditions in Xenopus
ment to the activation gate of HERG has also been recently laevis oocytes
recognized (40). Procedures for frog anesthesia and surgery, obtaining oocytes, and micro-
In this study we characterized the thermodinamic and ki- injection have been detailed elsewhere (34,41–43,45). Oocytes were main-
netic gating properties of HERG channel mutants either tained in OR-2 medium (in mM: NaCl 82.5, KCl 2, CaCl2 2, MgCl2 2,
lacking specific domains of the amino terminus or carrying Na2HPO4 1, HEPES 10, at pH 7.5). Cytoplasmic microinjections were
performed with 30–50 nl of in vitro synthesized cRNA per oocyte. HERG
several point mutations in the S4-S5 loop, using protocols currents were studied in manually defolliculated oocytes. Recordings were
designed to yield kinetic parameters under true steady-state systematically obtained in high-K1 OR-2 medium in which 50 mM KCl
conditions. The results were also compared with those ob- replaced an equivalent amount of NaCl to maximize tail current magnitude at
tained with channels combining structural alterations in both negative repolarizing potentials. A continuous perfusion of the cells at 1–5
regions. Our findings emphasize the key role of these cyto- ml/min was used to minimize the possible influence of K1 variations during
long test pulses. The amount of injected cRNA was calibrated to maintain
plasmic structures modulating HERG gating properties and inward current levels in the 1–6 mA range at repolarizing voltages around
indicate that not only the movement of the voltage sensor 100 mV to ensure proper voltage control. Functional expression was
itself but also the S4-S5 linker and the N-terminal regions act typically assessed 2–3 days after microinjection. Recordings were made at
as important determinants of activation and deactivation room temperature using the two-electrode voltage-clamp method as de-
voltage-dependent properties. scribed previously (34,42,43,45). Membrane potential was typically clamped
at 80 mV except for constructs showing a left shift in voltage dependence
of activation, in which a 100/110 mV basal voltage was used to com-
pensate for this shift. When indicated, a holding potential of 0/120 mV was
METHODS also used to maintain the channels open to obtain true steady-state activation
data. Ionic currents sampled at 1 KHz were elicited using the voltage pro-
Plasmids and preparation of cRNA tocols indicated in the graphs.
Kinetic parameters of activation and deactivation were obtained as pre-
The original plasmid containing the cDNA for the HERG channel viously described (34,42). The voltage dependence of activation was as-
(psP64A1-HERG) was a generous gift of Dr. E. Wanke (University of sessed by standard tail current analysis using depolarization pulses of
Milan, Italy). For in vitro cRNA synthesis, the HERG constructs cloned in variable amplitude. For slowly deactivating channels, the size of the peak
the psP64A1 vector were linearized, and capped cRNA was synthesized inward tail currents gives a reasonable estimate for the fraction of channels
from the linear cDNA templates by standard methods using SP6 RNA poly- activated (and inactivated) at the end of the depolarization. Alternatively, and
merase as described previously (41,42). for very rapidly deactivating constructs, fitting the relaxation of the tail
currents and extrapolating the magnitude of the decaying current to the
moment the depolarizing pulse ended were used to determine the amount of
Generation of HERG channel mutants current passing through channels opened on depolarization without influence
Procedures for generation of the D138-373 and D2-370 mutants have been of rapid inactivation (2,42). To take into account the possible influence of the
detailed elsewhere (34,43). To construct the mutant D2-16, a forward poly- inactivation recovery process on peak tail current or extrapolated current
merase chain reaction (PCR) primer was synthesized to introduce a HindIII magnitudes, the integral of the completely deactivated tail currents was also
restriction site and an ATG codon, plus 27 bp of HERG coding sequence used to estimate the voltage dependence of these channels. In all cases, very
corresponding to amino acids 17–26 (59-GGAAGCTTTCAGGATGACC- small variations of the half-maximum activation voltage (V½) were observed
ATCATCCGCAAGTTTGAGGGCCAGAG). The reverse primer was de- by any of the three alternative procedures, never reaching differences above
signed to cover the coding sequence of HERG corresponding to amino acids 63 mV. Tail current magnitudes normalized to maximum were fitted with a
400–407 (59-CCTTGAAGGGGCTGTAATGCAGGAT). The resulting PCR Boltzmann function to estimate the V½ and equivalent gating charge (zg):
product was digested with HindIII and BstEII, gel purified, and ligated into
HindIII/BstEII-digested wild-type HERG in the psP64A1 expression vector. Itail =Imax ¼ 1=½ð1 1 expððV½ VÞzg F=RTÞ;
For the D2-135 mutant, a forward primer containing the HindIII site, an ATG, where V is the test potential, and F, R, and T are Faraday constant, gas con-
and the coding sequence corresponding to amino acids 136–145 (59-GGAAG- stant, and absolute temperature, respectively. The activation data were also
CTTTCAGGATGGACATGGTGGGGTCCCCGGCTCATGACACC) was fitted with a second Boltzmann function of the form:
used in PCR with the same reverse primer as above covering amino acids 400–
407 of the HERG sequence. The PCR product generated was digested with Itail =Imax ¼ 1=½ð1 1 expððDGo Vzg FÞ=RTÞ;
HindIII/BstEII, and the purified fragment was used to replace the corre-
sponding HindIII/BstEII wild-type fragment in the psP64A1-HERG construct. where DGo is the work done at 0 mV. Although both equations are equiv-
The point mutants D540C, R541H, R541A, Y542C, Y545C, and G546C alent, from the last expression, the effect of every mutation on changes in the
were created by site-directed mutagenesis using the PCR-based overlap chemical potential (DGo) and electrostatic potential (VzgF) that drives
extension method as previously described (34,43,44). The final PCR frag- activation can be obtained. The time course of voltage-dependent activation
was studied using an indirect envelope-of-tail-currents protocols, varying the exhibited by HERG, the ionic currents during depolarizing
duration of depolarizing prepulses, and following the magnitude of the tail pulses used to activate the channels are rather small, and their
currents on repolarization. The time necessary to reach a half-maximal tail
current magnitude was used to compare the speed of activation of the
kinetics are obscured by superposition of activation and in-
different channels. activation transitions over a wide range (2,22,24,29,30,34).
The rates of deactivation were determined from negative-amplitude Big deactivating tail currents can be measured after the return
biexponential fits to the decaying phase of tail currents using a function of the to negative voltages following the depolarizing steps that
form: trigger the activation (and inactivation) of the channels. Such
y ¼ Af expðinvt f xÞ 1 As expðinvt s xÞ 1 C; currents, inward at the negative potentials and ionic condi-
tions used here, result from fast inactivation removal, fol-
where tf and ts are the time constants of fast and slow components, Af and As
lowed by a decay caused by a relatively slow channel closing.
are the relative amplitudes of these components, and C is a constant. In this
case, the first cursor of the fitting window was advanced to the end of the Analysis of the magnitude of these tail currents as a function
initial hook because of the recovery of inactivation. of previous depolarization voltage can be used to generate
I versus V curves from which the voltage dependence of
channel activation can be obtained (Fig. 2). For wild-type
Statistics
channels, Boltzmann fits (see Methods) to the normalized tail
Data values given in the text and in figures with error bars represent the current data obtained in response to depolarization pulses of
mean 6 SE for the number of indicated cells. Comparisons between data 1 s duration from a holding potential of 80 mV yielded V½,
groups were at first performed by parametric Student’s unpaired t-test (two- zg, and DGo values of 21.9 6 0.55 mV, 3.23 6 0.06, and
tailed) or ANOVA. Where nonhomogeneous variances (as evidenced after
Bartlett’s test) were obtained, alternate Welch’s test assuming Gaussian
1.65 6 0.05 kcal/mol, respectively (n ¼ 25). However,
populations with unequal SD and nonparametric Wilcoxon or Mann-Whit- for such channels as wild-type HERG characterized by very
ney tests, which do not make any assumption about the scatter of the data, slow activation/deactivation kinetics, no steady state will be
were also used to evaluate significance of mean differences between cell achieved unless much longer pulses are applied. Thus, at a
populations. In all cases, p-values ,0.05 were considered indicative of
holding potential of 80 mV the position of the Boltzmann
statistical significance.
curve along the voltage axis was strongly shifted toward
more negative values by increasing the duration of the de-
RESULTS polarizing step from 1 to 10 s. This modified the aforemen-
tioned kinetic parameters to 42.6 6 0.72 mV, 4.16 6 0.09,
Influence of specific amino-terminal deletions on
and 4.08 6 0.05 kcal/mol (n ¼ 5). On the other hand, and
steady-state activation voltage dependence of
perhaps more important, when 10 s preconditioning test
HERG channels
pulses were applied to wild-type channels from two holding
Exhaustive mutagenesis and kinetic studies using HERG voltages (80 and 0 mV), at which closed and open channels
channels modified in the initial protein segment located be- are expected, respectively, the activation curves were still
tween residues 1 and 135, which includes the conserved eag/ separated by nearly 30 mV. This demonstrates that for wild-
PAS domain, have been used to emphasize the crucial role of type channels test depolarizations of up to 10 s are not long
this region in setting the slow closing kinetics of the channel enough to reach a true equilibrium because if a steady-state
(19–21). However, the possible impact of these structural condition is attained during the depolarizing test pulse, the
alterations on activation gating has not been similarly as- current magnitudes and the position of the derived I/V curves
sessed. must be independent of the previous channel state (holding
To investigate the possible influence of the initial amino- voltage) or the test pulse duration. Therefore, only fractional
terminal domains on HERG activation properties, we gen- or isochronal but not real steady-state parameters can be
erated two channel constructs lacking either the first 135 derived from these data. Subsequently, V½ values under
amino acids including the previously characterized PAS steady-state conditions and the charge and free-energy
structure located between residues 16 and 135 (mutant D2- parameters for all channel constructs were obtained from
135) or only the initial 16 amino acids (mutant D2-16) that Boltzmann fits to activation curves either coming from
remained disordered in the crystalline structure (21). The coincident curves at two extreme holding voltages and long
characteristics of these channels were compared with those of 10-s depolarizations or from the extrapolated mean of such
mutant D2-370 in which not only the initial portion but most still noncoincident curves (Fig. 2). This guaranteed that I/V
of the amino terminus has been deleted, and with D138-373 curves were exclusively a function of test pulse character-
channels lacking only the protein segment corresponding to istics, regardless of the previous (open or closed) state of
the proximal domain. In all cases, these constructs were ex- the channels. As previously shown, this extrapolated value
pressed in Xenopus oocytes, and currents were measured marks the end of the I/V curve trend to converge as a result
using a two-electrode voltage clamp. To elucidate the voltage of increased test pulse duration from any holding voltage,
dependence of activation, depolarization pulses to different yielding a reasonable estimation of the true steady-state V½
potentials were applied, and tail currents were recorded on (22,34,35). As an example of the impact caused by steady-
repolarization (Fig. 1). Because of the rapid inactivation state deviations on the derived kinetic parameters, the 1.65
kcal/mol DGo value obtained for wild-type HERG using 1-s pared with those obtained in the same oocytes using 1-s
depolarization pulses from a holding potential of 80 mV is depolarizing steps. Analogously, maximal current levels
increased to 4.08 kcal/mol with 10-s conditioning pulses, a amounting to 4.0 6 1.3% (n ¼ 7) and 2.3 6 1.6% (n ¼ 6) of
value still deviated from the 5.70 kcal/mol under steady- those recorded in response to 1-s depolarizing steps were
state conditions. obtained when 10-s depolarizations were applied to wild-type
It is important to note that the observed variations in ki- and G546C channels, respectively. Therefore, the short- and
netic behavior after short (1-s) and long (10-s) conditioning long-term derived parameters originate from the different ki-
pulses are not a consequence of a secondary process (e.g., netic situations affecting the same channel population. On the
recruitment of silent channels or long-term open probability other hand, this kinetic behavior is inherent to HERG itself and
variations as a result of a slow and uncontrolled cellular not related to the oocyte expression system or to contamina-
process) taking place during long depolarizations because 1), tion of recordings with endogenous chloride currents during
for the same conditioning pulse, opposite I/V curve shifts are long voltage steps because analogous deviations are observed
obtained as a function of the holding potential, and 2), the in HEK293 cells expressing wild-type HERG channels (35).
same amount of current was obtained at saturating positive The steady-state values of V½, zg, and DGo of the different
voltages for the 1- and 10-s-derived I/V curves. Thus, with constructs are summarized at the bottom of Fig. 2. It can be
those constructs showing relatively small displacements of observed that a negative shift in V½ from 53.5 6 1.1 mV for
the 10-s I/V curve and almost coincident 10-s-derived curves wild-type (n ¼ 5) to 78.3 6 1.2 mV for D138-373 (n ¼ 5) is
at two holding potentials (i.e., D2-16, D2-135, D2-370, R541A induced by deletion of the proximal domain, but the voltage
and Y545C channels), only nonsignificant variations (,5%) dependence of activation is shifted in the opposite direction
in the maximal amount of current were observed when I/V in channels lacking the initial amino acids of the protein.
curves obtained using 10-s conditioning pulses were com- Thus, mutants D2-16, D2-135, and D2-370 showed V½
values of 20.0 6 1.6 (n ¼ 6), 34.1 6 1.6 (n ¼ 6), and presence of the proximal domain tends to stabilize the closed
34.3 6 0.5 mV (n ¼ 7), respectively. Interestingly, a V½ state(s) of HERG, whereas the initial segment of the protein
value equivalent to that obtained here under steady-state tends to stabilize the group of open states relative to the group
conditions has been previously reported for wild-type HERG of closed ones.
using 30-s depolarization steps (23). The data in Fig. 2 also
indicate a significant reduction of zg for D2-16 and D138-373
Effect of amino-terminal deletions on HERG
compared with wild-type channels. On the other hand, whereas
activation rates
a larger chemical potential drives the opening of the proximal-
domain-deleted D138-373 channels (DGo ¼ 7.2 6 0.12 As indicated above for the activation voltage dependencies,
kcal/mol as compared with 5.7 6 0.09 kcal/mol for wild- direct measurement of HERG activation rates during depo-
type), a less negative DGo is observed for those constructs larization steps is not possible because of overlapping of
deleted in the initial region of the amino terminus, regardless rapid inactivation and channel opening. Therefore, we stud-
of the presence or absence of the proximal domain (1.7 6 ied the time course of voltage-dependent activation using an
0.22, 3.5 6 0.23, and 3.7 6 0.13 kcal/mol for D2-16, envelope-of-tails protocol in which the duration of a depo-
D2-135, and D2-370 mutants). These data suggest that the larizing prepulse is varied at each potential, and the magni-
tude of the tail current peak at a constant repolarizing voltage 8.6 ms (n ¼ 6) necessary to reach the half-maximal tail am-
is measured at the end of every conditioning prepulse. Be- plitude of the wild-type channels at this voltage were either
cause of the presence of several preopen closed states, the slightly increased to 211 6 21 ms (n ¼ 6) and 213 6 11 ms
HERG activation time course shows a sigmoidal shape (n ¼ 7), or slightly decreased to 100 6 9 ms (n ¼ 6) for D2-
(24,30,33,45). Therefore, we estimated the rate of activation 16, D2-135, and D2-370 channels, respectively. However,
at a given voltage from the time necessary to attain half- consistent with previous results (34), a marked acceleration
maximum tail current magnitude at every depolarization in activation kinetics of around fivefold (t½ ¼ 28 6 1 ms (n ¼
potential (34,35). The data obtained with the different con- 6)) was obtained with D138-373 channels selectively deleted
structs carrying deletions in the amino terminus are compared in the proximal domain.
with those from wild-type channels in Fig. 3. Plots of acti- The analysis of the mutation effects on activation rates is
vation rates as a function of depolarization membrane po- complicated by the shifts in voltage dependence of steady-state
tential showed a similar slope for all channel types through activation (V½) and by the alterations in the effective number
the tested range. With a value of 0 mV taken as a reference, of gating charges moving across the membrane electric field
relatively similar activation kinetics was observed for chan- (zg). Because this would change the total activation driving
nels carrying deletions in the first 135 amino acids, as com- force (chemical plus electrostatic potential) for every channel
pared with those of the wild-type channels. Thus, the 143 6 mutant at a given voltage, we also plotted the activation rates
for the different constructs versus the total energy driving Influence of S4-S5 loop mutations on
activation (i.e., (DGo zgEF)) (30,31). These plots showed steady-state activation voltage dependence of
similar slopes along the tested range. Interestingly, when HERG channels
compensated for the differences in driving force, the activation
The interaction of the initial HERG region with the S4-S5
kinetics of all mutants appeared clearly accelerated as com-
linker has been proposed as a determinant of slow deactiva-
pared with the wild-type channel. If a level of 4 kcal/mol is
tion (19,23,39). The aforementioned modifications of HERG
taken as a reference, this acceleration corresponded to almost
activation properties, observed when the initial segment of
2-fold for the D2-135 mutant, 3.5-fold for D138-373 and
the amino terminus is eliminated, prompted us to check if
D2-370 channels, and 4.3-fold for the D2-16 construct.
mutations in the S4-S5 loop could similarly affect activation
These results indicate that although the mutations cause
parameters. For this purpose, we generated several single-
shifts in the steady-state voltage dependence of activation
point mutations between residues 540 and 546 of this linker
that could affect opening rates at submaximal activation
that were characterized using procedures analogous to those
potentials, the changes observed in the rates are not sec-
described above for the amino-terminal constructs. Addition-
ondary to those shifts. They also suggest that 1), not only the
ally, some of these mutations were combined with amino-
N-terminal region corresponding to the proximal domain
terminal deletions to check for possible additive effects on
(33,34,45) but also the initial part of the amino terminus may
gating (see below).
contribute to set the activation characteristics of HERG, and
Data regarding the activation voltage dependence of
2), the amino-terminal structures exert rate-limiting control
D540C, R541A, Y542C, Y545C, and G546C mutants are
of gating transitions during the activation process. Interest-
summarized in Figs. 5 and 6. Results of channels in which
ingly, our corrected data show that a facilitation of activation
arginine 541 was replaced by histidine (R541H) are also
(faster activation rate) is induced not only by deletion of
shown. This change introduced a histidine that is highly
the proximal domain (D138-373) that causes a shift of the
conserved in the same position of several Kv channels, in-
equilibrium toward the open state (more negative DGo) but
cluding some members of the eag family, and whose muta-
also in the mutants lacking the initial amino-terminal region
tion to arginine has been reported to compensate for the
(D2-16, D2-135, and D2-370) in which DGo is shifted by the
alterations in activation gating caused by N-terminal dele-
deletions to less negative values. This demonstrates the utility
tions of the rat eag channel (46). The position of the steady-
of this approach to separate alterations in the transition en-
state Boltzmann curves along the x-axis remained the same in
ergy of the gating process from changes in energy difference
cells expressing R541A and G546C channels as compared
between, or relative stability of, the closed and open states.
with wild-type, whereas a small rightward shift of around 10
mV was observed with R541H and Y545C mutants (Fig. 6).
Effect of amino-terminal deletions on HERG However, strong displacements to more depolarized poten-
deactivation kinetics tials were obtained after introducing cysteines in residues
540 and 542. In this case the V½ values were modified from
Fitting the decaying phase of tail currents to a biexponential
the 53.5 mV of the wild-type to 21.6 6 1.4 (n ¼ 6) and
function at different potentials after a fixed depolarization
19.5 6 0.7 mV (n ¼ 10) in D540C and Y542C channels,
pulse allows the estimation of the voltage dependence of
respectively. These huge modifications in steady-state volt-
deactivation kinetics. As illustrated in Fig. 4, the time con-
age dependencies were also accompanied by a substantial
stant of the fast deactivation component that accounts for
reduction in the slope of the I/V curves that lowered the zg
most of the current amplitude at negative voltages (21,30,34)
value from 4.5 in the wild-type to 1.3 6 0.04 and 2.0 6 0.08
became strongly reduced in all channel mutants lacking the
in D540C and Y542C channels. As a result of these altera-
initial 16-amino-acid segment. Thus, values of 14.5 6 1.3,
tions, the chemical potential for activation DGo was raised
14.4 6 2.0, and 16.0 6 1.3 ms (n ¼ 8–14) were obtained for
from the control 5.7 kcal/mol to 0.8 6 0.02 and 0.9 6
D2-16, D2-135, and D2-370 compared with 145 6 14 (n ¼ 9)
0.05 kcal/mol, respectively. These data indicate that without
for wild-type channels at 100 mV, a potential well below
an applied membrane field (e.g., at 0 mV) the tendency of the
the threshold for activation of all constructs. However, a
channels to open is nearly abolished (Y542C) or slightly
value closer to that of wild-type was observed with the
reversed (D540C) by the mutations.
proximal domain-deleted D138-373 mutant (113 6 4 ms, n ¼
6). These results emphasized the key role of the amino-ter-
minus initial segments on regulation of HERG deactivation
Effect of S4-S5 loop mutations on HERG
properties (34). Interestingly, although this reduction in de-
activation rates
activation time constants remained almost unaltered for all
the initial segment-deleted constructs, once differences in To characterize the time dependence of the activation pro-
total potential energy are considered, a 3.75-fold acceleration cess, an envelope-of-tails protocol (see above) was applied
of closing was also observed for D138-373 channels at to mutants in the S4-S5 linker. As shown in Fig. 7, plots of
equivalent driving forces. half-activation-time variation as a function of depolariza-
tion potential showed similar slopes for the S4-S5 mutants mutation of residue 540 became more pronounced when
as compared with wild-type and proximal-domain-deleted activation rates were compensated for differences in driv-
channels. However, there was a notable exception to this ing force. Nevertheless, the flat energetic profile of the
trend in the case of the D540C mutant, which showed only a D540C mutant complicated the performance of a rigorous
small variation of t½ values at different depolarizing volt- comparison at the more positive levels of energy driving
ages. Comparison of activation rates at 0 mV indicated very activation. With the 4 kcal/mol level taken as a reference,
little change in the kinetics of opening after modification we also observed a slightly slowed activation rate with
of residues 541 and 545, but almost twofold slower rates R541A, R541H, and G546C channels. Finally, the Y542C
were observed with Y542C and G546C mutants. Interest- and Y545C mutants exhibited 4.6- and 2.6-fold faster ac-
ingly, considerably faster rates were obtained with D540C tivation rates than wild-type at equivalent driving force.
channels at 0 mV, although this difference tended to dis- Altogether, these results indicate that not only deactiva-
appear at positive voltages because of the reduced voltage tion gating (19) but also voltage-dependent activation prop-
dependence of the activation rates exhibited by this mutant. erties can be affected by structural alterations in the S4-S5
Indeed, the acceleration of channel opening induced by linker (39).
Effects of S4-S5 loop mutations on HERG force of 4 kcal/mol also showed that although very small
deactivation kinetics accelerations of deactivation are obtained with R541H and
G546C mutants, the time constant of deactivation is notably
The effects of S4-S5 loop mutations on the time constant
smaller in mutants D540C, R541A, Y542C, and Y545C than
of the fast deactivation component are summarized in Fig. 8.
in wild-type, meaning deactivation accelerations of about
No significant differences in the rate of deactivation as a
seven-, three-, nine-, and fivefold, respectively. Interestingly,
function of voltage were observed for G546C as compared
whereas the accelerations of deactivation at 100 mV in
with wild-type. However, considerably faster rates were ob-
the S4-S5 loop mutants were always smaller than that of the
tained for the rest of the S4-S5 mutants. This acceleration of
D2-370 construct lacking most of the amino terminus, equiv-
deactivation ranged from about sevenfold faster for Y542C
alent or greater accelerations than that of the D2-370 mutant
than wild-type at 100 mV to almost threefold as observed
were observed at the same driving force with D540C, Y542C,
with R541H. It is important to note the smaller inclination of
and Y545C channels.
the D540C and Y542C t-versus-voltage plots as compared
with those of the rest of mutants. This tended to equalize the
tail-current kinetics at very negative voltages, suggesting a
Gating modifications induced by S4-S5 loop
reduction of the voltage dependence of closing from intro-
alterations in channels lacking the
duction of a cysteine in these positions. When the deactiva-
amino terminus
tion rates were plotted as a function of total potential energy
((DGo zgEF)), a clear tendency of the slopes obtained The aforementioned results demonstrate that in addition to
with the different mutants to be equal was observed, except the previously reported effects of amino-terminal deletions
for D540C channels, for which a steeper trace was apparent. and S4-S5 loop mutations on HERG deactivation properties
Comparison of deactivation rates at an equivalent driving (19,21,34,47,48), clear modifications on activation gating
can also be induced by structural alteration of these domains. the amino terminus. Thus, the small changes in V½ and zg
Therefore, the effects of S4-S5 loop mutations were also values (0.9 mV and 1.1 equivalent gating charges) obtained
assessed in the background of the D2-370 variant, which is with the R541A single mutant, largely differ from the 13 mV
lacking almost the whole amino terminus. As an initial ap- depolarizing shift and the 2.3 reduction in zg induced by the
proach, three S4-S5 mutants were used, corresponding to mutation in the D2-370 channel background. Subsequently,
those showing no effect (G546C) or intermediate (R541A) the 1.1 kcal/mol of DGo increase between R541A and
and strong (Y542C) accelerations of closing kinetics. wild-type channels is changed to 2.6 kcal/mol between the
Fig. 9 summarizes the effects of the double mutations on R541A 1 D2-370 and D2-370 variants. Therefore, the slight
activation voltage dependence. Fig. 9 A shows current traces reduction of free energy needed for channel activation (de-
in response to the voltage-clamp protocols depicted on top, stabilization of closed states relative to activated states)
and Fig. 9 B illustrates steady-state activation curves con- caused by mutating residue 541 is changed to a clear net in-
structed from the relation between depolarization membrane crease in free energy (destabilization of activated states rel-
voltage and the amplitude of tail currents, using 10-s depo- ative to closed ones) by introducing the R541A mutation in
larizing steps and two holding potentials. Comparison of the D2-370 background. Performance of similar comparisons
Boltzmann curves for the double mutants and those of wild- with the Y542C mutation reveals that, whereas the single
type and single mutants in Fig. 9 C indicates that the impact Y542C mutant exhibits a positive shift of 34 mV and a zg
of the mutation is clearly different in channels with or without reduction of 2.0, a similar reduction in the slope of the I/V
curve but a shift of only 0.8 mV is induced by mutating the time necessary to reach half-maximal activation at 0 mV was
same residue in the D2-370 construct. Concomitantly, the significantly reduced from 99 6 8.6 ms (n ¼ 6) in the D2-370
4.8 kcal/mol increase in DGo obtained with the Y542C mu- mutant to 73 6 4 ms (n ¼ 6) in the R541A 1 D2-370 con-
tation is reduced to 1.8 kcal/mol when the mutation is com- struct. This contrasts with the almost identical activation rate
bined with the deletion of the amino terminus. However, a observed at that voltage when wild-type and R541A channels
quite different pattern is observed with the G546C mutation. are compared. These differences were still more pronounced
In this case, the limited 0.6 mV rightward shift obtained with after compensation of activation rates for differences in
the single mutant is changed to an 8.5 mV leftward shift driving force because the slightly slower rate of activation
between the G546C 1 D2-370 and the D2-370 constructs, exhibited by the R541A mutant at 4 kcal/mol (see also above)
and the 0.9 reduction in zg is diminished to 0.5. This means a was changed to an almost fourfold faster rate in the R541A 1
small change in DGo increase from 1.1 kcal/mol after intro- D2-370 construct as compared with the D2-370 channel.
ducing the G546C mutation in wild-type channels, to –0.7 Clear differences were also observed after performance of a
kcal/mol after introduction of the same mutation in D2-370. similar analysis with the Y542C and G546C mutations. Thus,
The differences in the impact of the S4-S5 loop mutations compared with wild-type activation rates, the significantly
in channels with or without the amino terminus can be ex- slower rates exhibited by the Y542C and G546C mutants at
tended to the activation time course. As shown in Fig. 10, the 0 mV were either abolished by introducing a cysteine in
position 542 of the D2-370 channel (Y542C 1 D2-370 ac- or 542, both at 100 mV and at an equivalent driving force of
tivation t½ 111 6 9.5 ms, n ¼ 7) or remained slow in the –4 kcal/mol. On the other hand, the acceleration of closing
G546C 1 D2-370 mutant (216 6 11.0 ms, n ¼ 8). Further- induced by removal of the amino terminus was significantly
more, the 4.6-fold faster than wild-type activation shown by reduced by the G546C mutation. Thus, the 9.0- and 6.1-fold
the Y542C mutant at 4 kcal/mol was reduced to a 1.7-fold faster than wild-type deactivation rates of D2-370 channels at
faster rate of the Y542C 1 D2-370 construct as compared 100 mV and 4 kcal/mol, respectively, became 3.3- and
with the D2-370 channel at the same driving force. Finally, 4.4-fold faster in the G546C 1 D2-370 construct than in D2-
no changes in activation rates were observed at 4 kcal/mol 370. As discussed below, this suggests that neither of the
regardless of the background in which the G546C mutation structural alterations in the S4-S5 linker is entirely neutral in
was introduced. respect to deactivation characteristics. In addition, the similar
As a final verification, we also compared the effects of phenotypes exhibited by the S4-S5 linker mutants and the
mutating S4-S5 loop residues on closing kinetics with and channel variants lacking amino-terminal segments are prob-
without the amino terminus. The results of this analysis are ably not exclusively caused by disruption of an interaction
summarized in Fig. 11. Surprisingly, clearly additive accel- between the amino terminus and the S4-S5 loop because in
erations of closing were obtained after combining the elim- this case no further effect of the mutations should be expected
ination of the amino terminus with mutations of residues 541 in channels lacking the whole amino terminus.
vation (nN) curves. Thus, very small shifts in nN curves are strong alterations of activation properties (33,34,45). How-
obtained, not only with D138-373 channels showing a se- ever, a possible role of the initial amino-terminal segment on
lective acceleration of activation kinetics but also with sev- activation properties has not been previously recognized. Our
eral constructs (e.g., D2-16, D2-135, D2-370, R541H, current results indicate that the relevance of this protein
R541A, or Y542C) exhibiting activation kinetics equivalent segment in normal activation gating is greater that previously
to those of wild-type but a marked acceleration of closing. depicted. Thus, significant positive shifts in steady-state ac-
Previous work with channel constructs modified in the tivation voltage dependence associated with a reduction of zg
initial region of HERG amino terminus including the eag/ values and a less negative DGo are caused by selective de-
PAS domain indicated the crucial role of this segment in letion of this domain. These effects map to the initial region
setting deactivation characteristics and suggested that it can of the protein because they are also observed in channel var-
interact with the gating machinery, likely at the S4-S5 loop iants lacking only the first 16 residues of the amino terminus.
level (19–21,23). A specific role on HERG activation gating The differences in activation kinetics also extend to opening
has also been proposed for the proximal domain located in rates once differences in driving force are taken into account
the amino terminus between the eag/PAS and the first because the activation of the deleted constructs is at least
transmembrane helix, because deleting this region or modi- twofold faster at equivalent electrochemical potentials. Be-
fying a short cluster of basic residues near S1 helix caused cause these deleted channels also show a marked acceleration
of deactivation, it is conceivable that the alterations in the deactivation rates are measured at potential ranges at which
amino terminus cause stabilization of a transition state subsecond time constants are present but very long voltage
leading to a smaller energetic barrier between the group of steps are necessary to derive the equilibrium parameters,
closed and open states and hence faster activation and de- different gating transitions are being isolated in both cases.
activation kinetics. Note, however, that although activation and deactivation
It is important to emphasize that our analysis does not al- rates cannot be individually measured at potentials around
low us to identify the individual steps in the gating pathway the activation V½ (e.g., 53 mV for wild-type channels) be-
that are influenced by the mutations but only to recognize cause of superpositioning of the two processes, extrapolation
positions or structures altering the whole gating process. It is of their values up to these voltages indicates that, for exam-
also evident that for a voltage-dependent event, the speed of ple, for wild-type channels, activation half-time and fast
the process will be related to the amount of force (chemical deactivation rate are longer than 1 s. This suggests that the
plus electrostatic potential) driving it. Therefore, the distinct measured rates and the equilibrium parameters both qualify
deviations from electrochemical equilibrium must be taken the global gating process. It also validates the use of the
into account in comparing mutants to ensure that differences steady-state V½ and zg values to correct and compare the
in gating rates are not simply secondary to shifts in voltage gating rates of the different mutants. Interestingly, except for
dependence. It could be argued that, because activation and the D540C mutant, which shows a reduced voltage depen-
dence of activation rates, the similar slope of the plots relating to act through an electrostatic influence on the gating ma-
the gating parameters of the different constructs either with chinery (33). Whether this influence is exerted by a direct
voltage or with driving force over a wide range indicates that interaction of this sequence with the gating structures them-
similar conclusions about the effect of the mutations and the selves (e.g., voltage sensor, S4-S5 linker, or channel gate at
role of different regions in gating would be reached if acti- the bottom of S6) or indirectly through some intermediate
vation (and deactivation) rates extrapolated to potential structure (e.g., distal amino or eag/PAS domains) remains an
values near V½ are used. open question. Our results indicate that a similar facilitation
Deletion of the proximal domain located after the eag/PAS of activation is apparent at the same driving force in channels
segment in the amino terminus causes a strong alteration in lacking the initial portion of the amino terminus, the proximal
activation parameters, but an irrelevant participation of this domain, or both. Because the effects of such deletions on
domain on deactivation has been proposed (34). However, activation (and deactivation) rates are similar and nonaddi-
a significant acceleration of closing is observed in these tive (as illustrated by D2-370 channels; Figs. 3 C and 4 C),
proximal-domain-deleted channels once differences in total it is possible to speculate that an adequate positioning of
potential energy driving deactivation are considered. As pre- the eag/PAS-containing region (most probably the first 16
viously pointed out (33), care should be taken when com- N-terminal residues) toward the channel core is important not
pensating deactivation rates for driving force because the only for normal deactivation as previously proposed but also
extent to which the gate remains coupled to the voltage sen- for activation gating. Consequently, it is possible that main-
sor movement during closing is still an open question. Thus, tenance of a normal organization in the amino terminus is
the time constants of deactivation have to be corrected for the essential for proper orientation of this region and that dis-
corresponding variations in zg and voltage dependence of rupting it could contribute to modifications of activation
activation only if channel closure remains directly coupled properties. This would also be consistent with the dominant
to voltage sensor movements. Until now, the selective al- effect of its deletion on activation voltage dependence. Thus,
terations of activation and deactivation gating by specific the negative shifts caused by removal of the proximal domain
modification of the proximal and the initial amino-terminus in the D138-373 variant are readily reversed when both
domains, respectively, have been taken as an indication that proximal and the initial amino-terminus domains are elimi-
both processes were not simple reversions of each other. nated, as demonstrated by the similar voltage dependencies
However, it is tempting to speculate that the common alter- exhibited by the D2-16, D2-135, and D2-370 constructs.
ations in response to both structural alterations demonstrated It has been proposed that an interaction of the initial part of
here are produced because activation and deactivation gating the amino terminus with the S4-S5 linker acts as a determi-
are related and similarly coupled to sensor reorganizations, nant of HERG slow deactivation (19,23,39). It has also been
although they can also be strongly influenced by other do- shown that some mutations in this linker alter HERG acti-
mains or structures within the protein. vation properties and that it constitutes a crucial component
Although the molecular mechanism by which some struc- of the activation process by acting as a coupler between
tural alterations modify the distribution of HERG closed and voltage sensor movements and the gate (39,40). We reasoned
open states is still unclear, a combination of steady-state and that if physical interaction(s) between the amino terminus and
kinetic data can provide important information to identify the S4-S5 loop also participate in modulation of activation
gating-sensitive regions in the channel and to show the rel- gating, mutations introduced in the contact point(s) at the
ative stability of those states under different conditions. We level of the S4-S5 loop could phenocopy the effect of the
observed that all mutants lacking the initial portions of the amino-terminal deletions. Furthermore, it could also be ex-
amino terminus showed a positively shifted V½ of steady- pected that the effects of both structural alterations would not
state activation and a less negative DGo compared with wild- be additive. Steady-state properties of mutants in residues
type, which indicates that such deletions shift the equilibrium 540 to 546 show some differences as a function of the mu-
toward the closed state. Conversely, a negative shift and an tated residue. Clear displacements of the V½ to more positive
increased DGo indicating a destabilization of the closed state voltages and a notably less negative value of DGo were ob-
were observed for proximal domain-deleted channels. Inter- tained after introducing a cysteine at residues D540, Y542,
estingly, in both cases an accelerated activation is obtained and Y545. On the other hand, a remarkable acceleration of
once differences in electrochemical potential energy driving activation (and deactivation) kinetics was observed in these
activation (i.e., (DGo zgEF) or the differences in total three mutants, indicating an agreement between the behavior
energy between the open and closed states) are taken into of the channels lacking the initial segment of the amino ter-
account. This suggests that the presence of any amino-terminal minus and that of the S4-S5 loop-altered constructs. As an
segment contributes in wild-type channels to specific chemical additional control to strengthen the possibility of an amino
interactions that raise the energy barrier for activation. terminus/S4-S5 loop interaction, we characterized the effect
Recent work has localized the modulatory effects of the of mutations in the S4-S5 linker using an amino-terminal-
proximal domain on gating activation to a short positively deleted channel as background. Surprisingly, in all cases the
charged sequence between residues 362 and 366 that seems behavior of the double mutants did not correspond to any of
the single mutants. It is difficult to conceive a mechanism by terminus can influence normal function of the HERG gating
which the S4-S5 single-point mutations could cause addi- machinery. Some of these effects cannot be exerted only on
tional effects on channels lacking virtually the whole amino the deactivation process as previously recognized (19,23,39)
terminus if such an effect strictly relies on a physical inter- but also affect the activation behavior, probably by modu-
action between the two structures. Although these data are lating the transduction of charge movement into channel
not contrary to the possibility that such an interaction exists, opening and closing. Previous studies measuring time con-
they demonstrate that in no case are the structural and/or stants without correcting for driving force or for the shift in
functional alterations induced in the gating machinery by the V½ determined at true steady state may have misinterpreted
S4-S5 mutations neutral by themselves. Therefore, caution observations about how amino-terminal regions or the S4-S5
must be exercised when using them as an argument in favor linker modulate relative stability of closed and open states
of or against the existence of the interaction. and the energy of gating transitions between them. Our data
The changes in activation kinetics of the D540C and also suggest that in addition to the slow movement of the
Y542C channels were accompanied by a strong zg reduction. voltage sensor itself (30,36,37), delaying voltage sensor and
Thus, our data indicate the existence of an almost flat ener- activation gate functional coupling by other channel struc-
getic profile for these mutants, characterized not only by a tures can contribute to the atypically slow activation of
small thermodynamic tendency of the closed-to-open tran- HERG. Although further work is needed to elucidate the
sition to take place at 0 mV (i.e., near zero DGo) but also by a exact interactions involved, it is possible that physical cou-
low activation energy (i.e., small DGz of the transition state pling of the initial amino-terminal segment to the bottom of
between the closed and open states). Remarkably, we de- helix S4 or the S4-S5 linker could contribute to these mod-
tected a very small voltage dependence of the D540C channel ulatory effects.
activation rates. It has been shown that neutralization of the
This work was supported by grants SAF2003-00329 from Ministerio de
negative charge in position 540 makes a prominent compo-
Ciencia y Tecnologı́a, BFU2006-10936 from Ministerio de Educación y
nent of gating charge movement appear at negative voltages, Ciencia of Spain (both partially cofinanced by FEDER European Funds)
where channels do not open (49). Altogether, the negative and IB05-002 from Principado de Asturias (Spain). C.A.R. is a predoctoral
and positive shifts in the Q-V and the G-V relations, respec- fellow from the Spanish Ministerio de Ciencia y Tecnologı́a (refs. BES-
tively, and the effects on activation and deactivation kinetics 2004-3872). P.M. holds a predoctoral fellowship from FICYT of Asturias
suggest that the mutation uncouples the movements of the (ref. BP03-108).
voltage sensor and the activation gate. Albeit with slight
variations, similar alterations have been thoroughly charac- REFERENCES
terized in Shaker channels in which several mutations in the
S4 helix (50,51), the S4-S5 linker (52), and the bottom of S6 1. Sanguinetti, M. C., C. Jiang, M. E. Curran, and M. T. Keating. 1995. A
mechanistic link between an inherited and an acquired cardiac arrhyth-
(53) have been shown to disrupt the normally tight coupling mia: HERG channel encodes the IKr potassium channel. Cell. 81:299–
between voltage sensor and gate. In that case, movement of 307.
the voltage sensor appeared to be hindered by coupling to the 2. Trudeau, M. C., J. W. Warmke, B. Ganetzky, and G. A. Robertson.
activation gate, thus helping to explain the effects of carboxy- 1995. HERG, a human inward rectifier in the voltage-gated potassium
channel family. Science. 269:92–95.
terminal S6 mutations in which unloading this region from
3. Barros, F., C. Villalobos, J. Garcı́a-Sancho, D. del Camino, and P. de la
the S4-S5 linker allows the S4 segment to more easily carry Peña. 1994. The role of the inwardly rectifying K1 current in resting
its gating charge without effectively opening the permeation potential and thyrotropin-releasing hormone-induced changes in cell
pathway (53). Previous work on HERG strengthened the excitability of GH3 rat anterior pituitary cells. Pflugers Arch. 426:221–
230.
evidence of an interaction between the bottom of S6 and the
4. Barros, F., D. del Camino, L. A. Pardo, T. Palomero, T. Giráldez, and
first part of the S4-S5 linker, specifically at residue 540 P. de la Peña. 1997. Demonstration of an inwardly rectifying K1
(40,54). Interestingly, sensor and pore opening uncoupling in current component modulated by thyrotropin-releasing hormone and
Shaker leads to slowed activation kinetics (e.g., ILT and V2 caffeine in GH3 rat anterior pituitary cells. Pflugers Arch. 435:119–
129.
mutants; 52,55), but the opposite happens in the HERG
5. Bauer, C. K., R. Schäfer, D. Schiemann, G. Reid, I. Hanganu, and J. R.
D540C mutant. It is possible that whereas in Shaker such Schwarz. 1999. A functional role of the erg-like inward-rectifying K1
uncoupling favors a more stable voltage sensor-independent current in prolactin secretion from rat lactotrophs. Mol. Cell. Endo-
closed state (56), an unstable HERG closed channel is gen- crinol. 148:37–45.
erated after mutation of residue 540. This would also be 6. Schäfer, R., I. Wulfsen, S. Behrens, F. Weinsberg, C. K. Bauer, and
J. R. Schwarz. 1999. The erg-like current in rat lactotrophs. J. Physiol.
consistent with the recently observed stabilization of the 518:401–416.
activation gate closed conformation by interactions between 7. Emmi, A., H. J. Wenzel, P. A. Schwartzkroin, M. Taglialatela, P.
the initial portion of the S4-S5 linker (around D540) and the Castaldo, L. Bianchi, J. Nerbonne, G. A. Robertson, and D. Janigro.
C-terminal S6 residues around R665 or L666 (43). 2000. Do glia have heart? Expression and functional role for Ether-a-
go-go currents in hippocampal astrocytes. J. Neurosci. 20:3915–3925.
In summary, our results indicate that besides the gate and
8. Cherubini, A., G. L. Taddei, O. Crociani, M. Paglierani, A. M.
the voltage sensor located in the central channel core, other Buccoliero, L. Fontana, I. Noci, P. Borri, E. Borrani, M. Giachi, A.
cytoplasmic regions such as the initial part of the amino Becchetti, B. Rosati, E. Wanke, M. Olivotto, and A. Arcangeli. 2000.
HERG potassium channels are more frequently expressed in human 27. Paulussen, A., A. Raes, G. Matthijs, D. J. Snyders, N. Cohen, and J.
endometrial cancer as compared to non-cancerous endometrium. Br. J. Aerssens. 2002. A novel mutation (T65P) in the PAS domain of the
Cancer. 83:1722–1729. human potassium channel HERG results in the long QT syndrome by
9. Overholt, J. L., E. Ficker, T. Yang, H. Shams, G. R. Bright, and N. R. trafficking deficiency. J. Biol. Chem. 277:48610–48616.
Prabhakar. 2000. HERG-like potassium current regulates the resting 28. Paulussen, A. D. C., A. Raes, R. J. Jongbloed, R. A. H. J. Gilissen,
membrane potential in glomus cells of the rabbit carotic body. J. Neuro- A. A. M. Wilde, D. J. Snyders, H. J. M. Smeets, and J. Aerssens. 2005.
physiol. 83:1150–1157. HERG mutation predicts short QT based on channel kinetics but causes
10. Rosati, B., P. Marchetti, O. Crociani, M. Lecchi, R. Lupi, A. Arcangeli, long QT by heterotetrameric trafficking deficiency. Cardiovasc. Res.
M. Olivotto, and E. Wanke. 2000. Glucose- and arginine-induced in- 67:467–475.
sulin secretion by human pancreatic b-cells: the role of HERG K1 29. Liu, J., M. Zhang, M. Jiang, and G.-N. Tseng. 2003. Negative charges
channels in firing and release. FASEB J. 14:2601–2610. in the transmembrane domains of the HERG K channel are involved in
11. Viskin, S. 1999. Long QT syndromes and torsade de pointes. Lancet. the activation and deactivation gating processes. J. Gen. Physiol. 121:
354:1625–1633. 599–614.
12. Chiang, C., and D. M. Roden. 2000. The long QT syndromes: Genetic 30. Subbiah, R. N., C. E. Clarke, D. J. Smith, J. Zhao, T. J. Campbell, and
basis and clinical implications. J. Am. Coll. Cardiol. 36:1–12. J. I. Vandenberg. 2004. Molecular basis of slow activation of the
human ether-à-go-go related gene potassium channel. J. Physiol. 558:
13. Keating, M. T., and M. C. Sanguinetti. 2001. Molecular and cellular 417–431.
mechanisms of cardiac arrhythmias. Cell. 104:569–580.
31. Subbiah, R. N., M. Kondo, T. J. Campbell, and J. I. Vandenberg. 2005.
14. Redfern, W. S., L. Carlsson, A. S. Davis, W. G. Lynch, I. MacKenzie, Trytophan scanning mutagenesis of the HERG K1 channel: The S4
S. Palethorpe, P. K. S. Siegl, I. Strang, A. T. Sullivan, R. Wallis, domain is loosely packed and likely to be lipid exposed. J. Physiol.
A. J. Camm, and T. G. Hammond. 2002. Relationship between
569:367–379.
preclinical cardiac electrophysiology, clinical QT interval prolonga-
tion and torsade de pointes for a broad range of drugs: evidence for a 32. Zhang, M., J. Liu, M. Jiang, D.-M. Wu, K. Sonawane, H. R. Guy, and
provisional safety margin in drug development. Cardiovasc. Res. 58: G.-N. Tseng. 2005. Interactions between charged residues in the
32–45. transmembrane segments of the voltage-sensing domain in the hERG
channel. J. Membr. Biol. 207:169–181.
15. Thomas, D., W. Zhang, K. Wu, A.-B. Wimmer, B. Gut, G. Wendt-
Nordahl, S. Kathöfer, V. A. W. Kreye, H. A. Katus, W. Schoels, J. 33. Saenen, J. B., A. J. Labro, A. Raes, and D. J. Snyders. 2006.
Kiehn, and C. A. Karle. 2003. Regulation of potassium channel Modulation of HERG gating by a charge cluster in the N-terminal
activation by protein kinase C independent of direct phosphorylation proximal domain. Biophys. J. 91:4381–4391.
of the channel protein. Cardiovasc. Res. 59:14–26. 34. Viloria, C. G., F. Barros, T. Giráldez, D. Gómez-Varela, and P. de la
16. Finlayson, K., H. J. Witchel, J. McCulloch, and J. Sharkey. 2004. Peña. 2000. Differential effects of amino-terminal distal and proximal
Adquired QT interval prologation and HERG: Implications for drug domains in the regulation of human erg K1 channel gating. Biophys. J.
discovery and development. Eur. J. Pharmacol. 500:129–142. 79:231–246.
17. Roden, D. M., R. Lazzara, M. Rosen, P. J. Schwartz, J. Towbin, and 35. Miranda, P., T. Giráldez, P. de la Peña, D. G. Manso, C. Alonso-
G. M. Vincent. 1996. Multiple mechanisms in the long-QT syndrome. Ron, D. Gómez-Varela, P. Domı́nguez, and F. Barros. 2005. Spec-
Current knowledge, gaps, and future directions. Circulation. 94:1996– ificity of TRH receptor coupling to G-proteins for regulation of
2012. ERG K1 channels in GH3 rat anterior pituitary cells. J. Physiol. 566:
717–736.
18. Schwartz, P. J. 2005. Management of long QT syndrome. Nat. Clin.
Pract. Cardiovasc. Med. 2:346–351. 36. Smith, P. L., and G. Yellen. 2002. Fast and slow voltage sensor
movements in HERG potassium channels. J. Gen. Physiol. 119:275–
19. Wang, J., M. C. Trudeau, A. M. Zappia, and G. A. Robertson. 1998.
293.
Regulation of deactivation by an amino terminal domain in Human
ether-a-go-go-related gene potassium channels. J. Gen. Physiol. 37. Piper, D. R., A. Varghese, M. C. Sanguinetti, and M. Tristani-Firouzi.
112:637–647. 2003. Gating currents associated with intramembrane charge displace-
20. Wang, J., C. D. Myers, and G. A. Robertson. 2000. Dynamic control of ment in HERG potassium channels. Proc. Natl. Acad. Sci. USA. 100:
deactivation gating by a soluble amino-terminal domain in HERG K1 10534–10539.
channels. Biophys. J. 115:749–758. 38. Zhang, M., J. Liu, and G.-N. Tseng. 2004. Gating charges in the
21. Morais-Cabral, J. H., A. Lee, S. L. Cohen, B. T. Chait, M. Li, and R. activation and inactivation processes of the HERG channel. J. Gen.
MacKinnon. 1998. Crystal structure and functional analysis of the Physiol. 124:703–718.
HERG potassium channel N terminus: a eukariotic PAS domain. Cell. 39. Sanguinetti, M. C., and Q. P. Xu. 1999. Mutations of the S4–S5 linker
95:649–655. alter activation properties of HERG potassium channels expressed in
22. Schönherr, R., B. Rosati, S. Heh, V. G. Rao, A. Arcangeli, M. Olivotto, Xenopus oocytes. J. Physiol. 514:667–675.
S. H. Heinemann, and E. Wanke. 1999. Funtional role of the slow 40. Ferrer, T., J. Rupp, D. R. Piper, and M. Tristani-Firouzi. 2006. The S4–
activation property of ERG K1 channels. Eur. J. Neurosci. 11:753– S5 linker directly couples voltage sensor movement to the activation
760. gate in the human ether-a-go-go-related gene (hERG) K1 channel.
23. Chen, J., A. Zou, I. Splawski, M. T. Keating, and M. C. Sanguinetti. J. Biol. Chem. 281:12858–12864.
1999. Long QT syndrome-associated mutations in the Per-Arnt-sim 41. Gómez-Varela, D., F. Barros, C. G. Viloria, T. Giráldez, D. G. Manso,
(PAS) domain of HERG potassium channels accelerate deactivation. S. G. Dupuy, P. Miranda, and P. de la Peña. 2003. Relevance of the
J. Biol. Chem. 274:10113–10118. proximal domain in the amino-terminus of HERG channels for regu-
24. Wang, S., S. Liu, M. J. Morales, H. C. Strauss, and R. L. Rasmusson. lation by a phospholipase C-coupled hormone receptor. FEBS Lett.
1997. A quantitative analysis of the activation and inactivation kinetics 535:125–130.
of HERG expressed in Xenopus oocytes. J. Physiol. 502:45–60. 42. Ho, S. N., H. D. Hunt, R. M. Horton, J. K. Pullen, and L. R. Pease.
25. Kupershmidt, S., D. J. Snyders, A. Raes, and D. M. Roden. 1998. A 1989. Site-directed mutagenesis by overlap extension using the poly-
K1 channel splice variant common in human heart lacks a C-terminal merase chain reaction. Gene. 77:51–59.
domain required for expression of rapidly activating delayed rectifier 43. de la Peña, P., L. M. Delgado, D. del Camino, and F. Barros. 1992.
current. J. Biol. Chem. 273:27231–27235. Cloning and expression of the thyrotropin-releasing hormone receptor
26. Aydar, E., and C. Palmer. 2001. Functional characterization of the from GH3 rat anterior pituitary cells. Biochem. J. 284:891–899.
C-terminus of the human ether-à-go-go-related gene K1 channel 44. Barros, F., D. Gómez-Varela, C. G. Viloria, T. Palomero, T. Giráldez,
(HERG). J. Physiol. 534:1–14. and P. de la Peña. 1998. Modulation of human erg K1 channel gating
by activation of a G protein-coupled receptor and protein kinase C. 50. Ledwell, J. L., and R. W. Aldrich. 1999. Mutations in the S4 region
J. Physiol. 511:333–346. isolate the final voltage-dependent cooperative step in potasssium
45. Gómez-Varela, D., P. de la Peña, J. Garcı́a, T. Giráldez, and F. Barros. channnel activation. J. Gen. Physiol. 113:389–414.
2002. Influence of amino-terminal structures on kinetic transitions be- 51. Pathak, M., L. Kurtz, F. Tombola, and E. Isacoff. 2005. The cooper-
tween several closed and open states in human erg K1 channels. ative voltage sensor motion that gates a potassium channel. J. Gen.
J. Membr. Biol. 187:117–133. Physiol. 125:57–69.
46. Terlau, H., S. H. Heinemann, W. Stühmer, O. Pongs, and J. Ludwig. 52. Schoppa, N. E., and F. J. Sigworth. 1998. Activation of Shaker
1997. Amino terminal-dependent gating of the potassium channel rat potassium channels. II. Kinetics of the V2 mutant channel. J. Gen.
eag is compensated by a mutation in the S4 segment. J. Physiol. Physiol. 111:295–311.
502:537–543. 53. Ding, S., and R. Horn. 2003. Effect of S6 tail mutations on charge
47. Schönherr, R., and S. H. Heinemann. 1996. Molecular determinants for movement in Shaker potassium channels. Biophys. J. 84:295–305.
activation and inactivation of HERG, a human inward rectifier 54. Tristani-Firouzi, M., J. Chen, and M. C. Sanguinetti. 2002. Interactions
potassium channel. J. Physiol. 493:635–642. between S4–S5 linker and S6 transmembrane domain modulate gating
48. Spector, P. S., M. E. Curran, A. Zou, M. T. Keating, and M. C. of HERG K1 channels. J. Biol. Chem. 277:18994–19000.
Sanguinetti. 1996. Fast inactivation causes rectification of the IKr 55. Smith-Maxwell, C. J., J. L. Ledwell, and R. W. Aldrich. 1998.
channel. J. Gen. Physiol. 107:611–619. Uncharged S4 residues and cooperativity in voltage-dependent potas-
49. Piper, D. R., W. A. Hinz, C. K. Tallurri, M. C. Sanguinetti, and M. sium channel activation. J. Gen. Physiol. 111:421–439.
Tristani-Firouzi. 2005. Regional specificity of hERG channel activation 56. Yifrach, O., and R. MacKinnon. 2002. Energetics of pore opening in a
and inactivation gating. J. Biol. Chem. 280:7206–7217. voltage-gated K1 channel. Cell. 111:231–239.
Date Submitted by
n/a
the Author:
er
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pflugers Archiv-European Journal of Physiology Page 2 of 34
1 1
2
3
4
5
6
7
8
9
10
Participation of HERG channel cytoplasmic structures on regulation by the G-
11 protein-coupled TRH receptor
12
13
14
15
16
17 Carlos Alonso-Ron, Francisco Barros, Diego G. Manso, David Gómez-Varela, Pablo
18
19 Miranda, Luis Carretero, Pedro Domínguez and Pilar de la Peña
20
Fo
21
22
23
24
rP
30
31
32
rR
33
34
35
36
37
ev
38
39
40
41
iew
42
43
44
45 Running title: Cytoplasmic domains and hormonal modulation of HERG
46
47
48
49
50
51
52
53
54
55
56
57 Send correspondence to Dr. Francisco Barros at the above address. Phone: +34-985-
58
59 103565; FAX: +34-985-103157; email: fbarros@uniovi.es
60
Page 3 of 34 Pflugers Archiv-European Journal of Physiology
1 2
2
3
4
5 ABSTRACT
6
7
8 Human ether-a-go-go-related gene (HERG) channels heterologously expressed
9
10
in Xenopus oocytes are regulated by the activation of G-protein-coupled hormone
11 receptors that, like the thyrotropin-releasing hormone (TRH) receptor, activate
12
13 phospholipase C. Previous work with serially deleted HERG mutants suggested that
14 residues 326-345 located in the proximal domain of the channels amino terminus might
15
16 be required for the hormonal modulation of HERG activation. Generation of new
17 channel mutants deleted in this region further point to the amino acid sequence between
18
19 residues 326-332 as a possible determinant of the TRH effects, but individual or
20 combined single point mutations in this sequence demonstrate that maintenance of its
Fo
21
22 consensus sites for phosphorylation and/or interaction with regulatory components is
23 not important for the modulatory response(s). The TRH-induced effects also remained
24
rP
25 unaltered when a basic amino acid cluster located between residues 362 and 366 is
26
27
eliminated. Additionally, no effect of TRH was observed in channels carrying single
28 point mutations at the beginning of the intracellular loop linking transmembrane
29
ee
30 domains S4 and S5. Our results indicate that a correct structural arrangement of the
31 amino terminal domains is essential for the hormone-induced modifications of HERG
32
activation. They also suggest that the hormonal regulatory action is transmitted to the
rR
33
34 transmembrane channel core through interactions between the cytoplasmic domains and
35
36 the initial portion of the S4-S5 linker.
37
ev
38
39
40 Keywords: HERG, potassium channel, gating, hormonal regulation, cytoplasmic
41
iew
1 3
2
3
4
5
6
INTRODUCTION
7
8
9
10
Ether-a-go-go-related gene (ERG) potassium channels play a key role setting the
11 electrical behaviour of a variety of cell types [2,3,5,8,13,27,30,34]. The human-ERG
12
13 (HERG) channel mediates the cardiac repolarizing current IKr [32,40]. Reduction of
14 this current due to mutations in HERG or to channel blockade by cardiac and non-
15
16 cardiac drugs causes congenital type 2 long QT and most of the acquired long QT
17 syndromes, respectively [9,14,20,28,32,37,42]. Long QT syndrome is a clinical disorder
18
19 characterized by prolongation of the QT interval on the electrocardiogram, associated
20 with an increased risk of life-threatening arrhythmias [29,36,42]. It is known that many
Fo
21
22 cardiac events in patients with long QT syndrome including the HERG-asssociated
23 LQT2 occur during physical or emotional stress (reviewed in [27,30,37]) suggesting a
24
rP
33
34 hormone (TRH) receptors] and the activation of protein kinase C, cause HERG current
35
36 reductions and a positive shift in the HERG activation curve [4,6,18,38]. Interestingly,
37 β-blockers remain the mainstay of long-term therapy for long-QT patients [30,37,43].
ev
38
39 However, the response of type 2 HERG-associated long-QT patients to these agents is
40 less complete than that of type 1 long-QT patients [37] and in these cases a combination
41
iew
42 of β-blockade and left cardiac sympathetic denervation, that would also block α-
43
adrenergic pathways, appears to provide significant additional protection [26,37,43].
44
45 Therefore, it is of great interest to better understand the molecular reasons for the effects
46
47
of PLC-coupled receptor activation on HERG properties.
48 We have previously demonstrated that activation of PLC-coupled receptors,
49
50 such as TRH and serotonin-1c receptors co-expressed in Xenopus oocytes with HERG
51 channels, modifies HERG channel gating; slowing activation, accelerating deactivation
52
53 and shifting the activation voltage dependence in the depolarizing direction [4,16].
54 Similar effects on activation have been reported in response to α-adrenergic receptor
55
56 stimulation and to protein kinase C activation with phorbol esters [18,38]. ERG
57 channels endogenously expressed in adenohypophysial cells are also similarly
58
59 modulated by TRH [16,25,36]. In an attempt to characterize the structural requirements
60
for hormonal regulation of channel activation, we used channel variants carrying serial
deletions in the proximal domain of the HERG amino terminus extending from several
Page 5 of 34 Pflugers Archiv-European Journal of Physiology
1 4
2
3 positions in the channel sequence up to residue 373 near the first transmembrane helix
4
5 [16]. Based on the fact that normal hormonal responses are observed with channels
6
lacking the protein segment corresponding to residues 345-373, but that the TRH-
7
8 induced modifications in activation properties are virtually abolished when an
9
10
additional segment (residues 326-373) is deleted, we concluded that the sequence
11 extending from residues 326 to 345 is essential for the hormonal modulation of HERG
12
13 activation [16]. To further identify the minimal amino acid sequence required for the
14 hormonal effects we used here a new HERG variant lacking residues 333 to 373. Since
15
16 the TRH-induced effects in activation were totally normal in the ∆333-373 mutant but
17 were lacking in the ∆326-373 construct, this initially suggested a requirement of the
18
19 326-332 sequence for the hormonal effects to take place. Surprisingly, introduction into
20 this sequence of single or combined point mutations leading to disruption of consensus
Fo
21
22 sites for phosphorylation by different protein kinases or for interaction with regulatory
23 components such as 14-3-3 proteins and inositol phospholipids, did not modify the
24
rP
33
34 hormonal effects. Our results suggest that the effect of the amino terminal deletions
35
36 antagonizing the TRH-induced regulation of HERG is not mainly due to elimination of
37 specific amino acid sequences, but instead results from structural modifications in the
ev
38
39 cytoplasmic domains and/or disruption of their normal positioning towards the
40 transmembrane channel core. Furthermore, we identify the initial portion of the S4-S5
41
iew
42 HERG linker as a key point for transmission of the regulatory actions triggered by
43
activation of the PLC-coupled receptors from the cytoplasmic domains to the channel
44
45 core.
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pflugers Archiv-European Journal of Physiology Page 6 of 34
1 5
2
3
4
5 METHODS
6
7
8 Plasmids and preparation of cRNA
9
10
The original plasmid containing the cDNA for the HERG channel was a
11 generous gift of Dr. E. Wanke (University of Milan, Italy). Isolation of the TRH-R
12
13 cDNA has been described previously [10]. For in vitro cRNA synthesis HERG
14 constructs cloned in the psP64A+ vector were linearized and capped cRNA was
15
16 synthesized from the linear cDNA templates by standard methods using SP6 RNA
17 polymerase as described previously [4,10,11].
18
19
20 Generation of HERG channel mutants
Fo
21
22 Procedures for the generation of the ∆2-135, ∆138-373 and ∆326-373 variants
23 have been detailed elsewhere [1,15,42]. Variant ∆333-373 was constructed in a similar
24
rP
25 way using a forward polymerase chain reaction (PCR) primer containing a Hind-III site
26
27
and including the coding sequence for the first six residues of HERG (5'-TTG AAG
28 CTT CTC AGG ATG CCG GTG CGG AGG GGC-3') and a reverse primer containing
29
ee
33
34 generated in a similar way using a PCR reverse primer containing HERG coding
35
36 sequence corresponding to residues 338-344 followed by the 9-bp sequence containing
37 the unique BstEII site (5'-CTG GGT GAC CTT GAG GTC CAC AAA GTT GAG G-
ev
38
39 3'). Construction of variant ∆326-341 was performed by combining two independent
40 constructs. The first one containing the 342-698 HERG sequence was obtained by PCR
41
iew
42 using a forward primer containing the SalI site followed by the coding sequence 343-
43
348 (5'-CCG TGT CGA CCT CAA GGG CGA CCC CTT C-3'). The reverse primer
44
45 contained the sequence corresponding to residues 694-701 and the unique XhoI site in
46
47
the HERG sequence. The amplification product was digested with SalI and XhoI,
48 purified and subcloned into pBluescript KS+, yielding the construct pBKS+ HERG 342-
49
50 698. The second construct containing the HERG coding sequence from amino acids 1 to
51 325 and a SalI site was obtained by PCR using a forward primer containing the Hind-III
52
53 site and covering the 5' end of the HERG cDNA including the initial ATG triplet (5'-
54 TGG AAG CTT CTC AGG ATG CCG GTG CGG AGG GGC CAC-3') and a reverse
55
56 primer that contained the HERG coding sequence corresponding to residues 319-325
57 and 342-343 and also a SalI site (5'-CCT AGT CGA CGA GGT CGC AGT CCG AGG-
58
59 3'). The amplification product was digested with HindIII and SalI, purified and
60
subcloned into the pBKS+ HERG 342-698 construct yielding the pBKS+ HERG 1-698
construct with the 326-341 deletion. Finally, the HindIII/XhoI fragment of this construct
Page 7 of 34 Pflugers Archiv-European Journal of Physiology
1 6
2
was introduced into the corresponding site of the psP64A+ plasmid containing the
3
4
5 cDNA for the wild-type HERG.
6
7 For construction of the InsA/326-341 and InsB/326-341 variants in which the
8 326-341 sequence is substituted by other sequences unrelated to HERG, a double chain
9
10 oligonucleotide 5'-TGC ACC GCT ACC CGT ACG ACG TTC CGG ACT ACG CCA
11 CCC TCA ACT TTG-3' was subcloned into the SalI site of the ∆326-341 variant. This
12
13 oligonucleotide contained a BsiWI site and two SalI sites at both ends. The sense and
14 antisense oligonucleotides used to generate the double chain segment were hybridized
15
16 after phosphorylation with ATP and polynucleotide kinase. Insertion of this double
17
18
chain oligonucleotide into the SalI site of ∆326-341 was done in both sense and
19 antisense orientation yielding the InsA/326-341 and InsB/326-341 variants in which the
20
original HERG 326-341 segment is substituted by DRYPYDVPDYATLNFV and
Fo
21
22 DKVEGGVVRNVARVAV sequences, respectively.
23
24 Construction of the ∆155-209 variant was also performed by PCR using a
rP
25 forward primer containing the sequence coding for HERG residues 152-154 and 209-
26
27 215 as well as a BstXI site that corresponds to the sequence of residues 152-154 and
28 209 (5'-CCA GCT GGG TGG ACG TGG ACC TGA CG-3'). The reverse primer
29
ee
30 contained the coding sequence corresponding to residues 365-372 and a BstEII site (5'-
31 GGA CAG GAC CTG GGT GAC CTT CTC-3'). The PCR product digested with BstXI
32
and BstEII was gel-purified and substituted into the wild-type psP64A+ HERG.
rR
33
34
The point mutants D540C, R541H and Y542C were created by site-directed
35
36 mutagenesis using the PCR-based overlap extension method as previously described
37
[1,15,42]. The resulting PCR products were digested with BstEII/XhoI and ligated into
ev
38
39 BstEII/XhoI-digested wild-type psP64A+ HERG.
40
41 All constructs were sequenced to confirm the mutations and to ensure the
iew
1 7
2
3
maintain inward current levels in the 1-6 µA range at repolarizing voltages around -100
4
5 mV to ensure proper voltage control. Functional expression was typically assessed 2-3
6
7
days after microinjection. Recordings were made at room temperature using the two-
8 electrode voltage-clamp method with a Turbo TEC01C amplifier (NPI, Tamm,
9
10 Germany) as described previously [1,4,15,16,42]. Membrane potential was typically
11 clamped at -80 mV except for constructs ∆138-373, ∆326-373, ∆333-373 and ∆345-373
12
13 in which a -100/-110 mV basal voltage was used to compensate for the left-shift in
14 voltage dependence of activation caused by the deletions. This ensures that in all cases
15
16 the whole channel population remains closed at the holding potential used. TRH data
17 collection started 2 min after adding the hormone. Stimulation and data acquisition were
18
19 controlled with Pulse+PulseFit software (HEKA Elektronic, Lambrecht, Germany)
20 running on Macintosh computers. Ionic currents sampled at 1 KHz were elicited using
Fo
21
22 the voltage protocols indicated in the graphs. Data analysis and exponential fits to ionic
23
24
currents were performed with the programs PulseFit (HEKA Elektronic) and Igor-Pro
rP
25 (WaveMetrics Inc., Lake Oswego, OR, USA). A P/n method was used for leak and
26
27 capacitive current subtraction. Kinetic parameters of activation and deactivation were
28 obtained as previously described [1,16,42]. The voltage dependence of current
29
ee
30 activation was assessed using standard tail current analysis. Tail current magnitudes
31 normalized to maximums were fitted with a Boltzmann function:
32
h(V) = Imax [1/(1 + exp((V - V1/2)/k))]
rR
33
34 where V is the test potential, V1/2 is the half-activation voltage, and k is the slope factor.
35
36 The time course of voltage-dependent activation was studied using indirect envelope of
37 tail current protocols, varying the duration of depolarising prepulses and following the
ev
38
39 magnitude of the tail currents upon repolarization. The time necessary to reach a half-
40
maximal tail current magnitude was used to compare the speed of activation for the
41
iew
1 8
2
3
4
5 RESULTS
6
7
8 Effect of amino terminus ∆2-135 and ∆333-373 deletions on the TRH-induced
9
10
modifications of HERG activation
11 Our previous studies to test the relevance of different segments of the HERG
12
13 amino terminal regions on TRH-induced regulation of activation kinetics, indicated that
14 the hormone induces a clear slowing of activation and a marked shift in activation
15
16 voltage both in wild-type channels and in those lacking residues 345-373 of the
17 proximal domain in the amino terminus, but that the hormonal effect is virtually
18
19 abolished in those channel variants from which either the 326-373 sequence or a longer
20 amino acid stretch extending up to the whole proximal domain sequence (e.g. ∆284-
Fo
21
22 373, ∆223-373 and ∆138-373 mutants) is deleted. Since the 345-373 segment is not
23 necessary for the hormonal effects to take place, but the deletion of 19 additional
24
rP
25 residues (∆326-373 construct) eliminates such effects, we interpreted that the relatively
26
27
short sequence extending from residues 326 to 345 is essential for the hormonal
28 modulation of HERG activation [16]. In an attempt to further delimit the specific region
29
ee
30 involved in the hormonal regulation, here we generated two new constructs: a variant
31 lacking residues 333-373 in the proximal domain and a ∆2-135 channel lacking the
32
eag/PAS domain (see Methods). The activation voltage dependence of these mutants
rR
33
34 and their modification upon treatment with TRH in oocytes co-expressing the channel
35
36 and the TRH receptor are compared with those corresponding to the ∆326-373 and
37 ∆345-373 constructs in Fig. 1. It can be observed that, unlike the results obtained with
ev
38
39 ∆326-373 and ∆138-373 (see also [16]), a shift in activation voltage dependence
40 analogous to that observed in wild-type HERG (from –18.7 ± 1.1 mV of V1/2 without
41
iew
42 hormone to –1.4 ± 1.2 mV after TRH, n=30) is induced by the hormone in the eag/PAS-
43
deleted ∆2-135 construct (from –20.5 ± 0.4 to –6.9 ± 2.3 mV, n=11). This indicates that
44
45 the eag/PAS region is not necessary for the hormonal effect.
46
47
Performance of an activation voltage dependence analysis with the ∆333-373
48 construct in the absence of hormone demonstrates that the activation V1/2 of this variant
49
50 is slightly shifted (≈11 mV) to more depolarized voltages as compared to wild-type.
51 This strongly contrasts with all other mutants with different size proximal domain
52
53 deletions in which the activation voltage dependence is markedly shifted to more
54 hyperpolarized voltages (e.g. from –18.7 mV in the wild-type to -60 and -44 mV in
55
56 ∆326-373 and ∆345-373 channels, respectively. See also [16,42]). Consistent with
57 previous results [32], it is likely that the hyperpolarizing modifications of basal
58
59 activation voltage dependence in channels lacking the positively charged KIKER
60
sequence (between residues 363 and 367), are compensated by substitution of this
sequence by an equivalently positive stretch (RIRTYSK using the one letter amino acid
Pflugers Archiv-European Journal of Physiology Page 10 of 34
1 9
2
3
code) in the ∆333-373 construct, leading to a wild-type behaviour. Regardless of the
4
5 modifications in the I/V curves caused by the deletion itself, a prominent shift of 20.1 ±
6
7
1.0 mV can be observed (analogous to those obtained with ∆345-373 and wild-type
8 channels) triggered by the hormone in oocytes expressing ∆333-373 channels. It is
9
10 important to note that: i) the differences in the basal or hormonally-induced behaviour
11 of the constructs were not related to the different conditions used for recording
12
13 membrane currents (i.e., 50 mM KCl) in these experiments, since the position of the I/V
14 curves and their displacement in response to TRH addition were the same when ∆326-
15
16 373 and ∆333-373 currents were recorded in medium containing 2 mM KCl (not
17 shown), and ii) the position of the I/V curves along the voltage axis did not act as an
18
19 important determinant of the hormonal effect, since an analogous shift was observed in
20 ∆345-373 and ∆333-373 channels, in spite of the fact that in the absence of hormone
Fo
21
22 their voltage dependencies appeared displaced to hyperpolarizing and depolarizing
23
24
voltages, respectively, as compared with the wild-type channel. In principle, these
rP
25 results would be compatible with the interpretation that the seven amino acid RYRTISK
26
27 segment corresponding to residues 326-332 could be acting as a determinant of the TRH
28 effects since the hormone-induced response is absent from channels lacking residues
29
ee
33
34 effects on activation rates. Due to the overlap of relatively slow activation transitions
35
36 and fast inactivation rates in HERG upon depolarization, we monitored the time course
37 of transitions from closed to open states using an indirect envelope of tail currents
ev
38
39 protocol, following the increase of tail current magnitude as a result of duration
40
increases in a previous depolarization pulse at a fixed voltage. As shown in Fig. 2, the
41
iew
42 time necessary to attain a half-maximal tail current was more than two-fold longer after
43
44
addition of TRH to ∆2-135 and ∆333-373 channels, a result similar to that obtained
45 with wild-type and ∆345-373 channels. However, the activation rate remained almost
46
47 unaltered following hormone addition to channels lacking either the 326-373 segment
48 or most of the proximal domain (∆138-373). This confirmed and extended our previous
49
50 results and indicated that increasing the length of the deletion by seven additional
51 residues (from ∆333-373 to ∆326-373) causes an almost complete loss of the TRH-
52
53 induced effects on activation parameters.
54
55
56 Effect of mutations in the 326-332 (RYRTISK) HERG sequence on TRH-induced
57
modifications of activation parameters
58
59 The identification of the short RYRTISK sequence corresponding to residues
60
326-332 as a possible determinant of the hormonal effects, opened some a priori
exciting and additional possibilities. As a matter of fact, several consensus sequences
Page 11 of 34 Pflugers Archiv-European Journal of Physiology
1 10
2
3 are localized in this short region including phosphorylation sites for tyrosine kinases at
4
5 Y327, for protein kinase A (RX1-2S/TX) and Ca-Calmodulin kinase II (RXXS/TX) at
6
T329, and for protein kinase B/Akt (RXRXXS/T) at S331 [21,22,47]. Interestingly,
7
8 protein kinases A, B and tyrosine kinase(s) have been proposed as regulators of HERG
9
10
currents [7,23,47]. Furthermore, the presence of three positively charged residues in the
11 seven residue segment 326-332 constitutes a hypothetical determinant of PIP2 binding,
12
13 a putative modulator of HERG channel activity [6]. Finally, the RYRTISK sequence is
14 a variant of the RXXpS/pT motif that constitutes a consensus for interaction with 14-3-
15
16 3 proteins, also proposed as modulators of HERG [19]. Based on this information,
17 several point mutants were generated in which residues R326, Y327, R328, T329 and
18
19 S331 were changed to alanine, and a triple mutant (YTS/AAA) in which the three
20 putatively phosphorylated residues were simultaneously changed to alanine was
Fo
21
22 constructed. Surprisingly, the modulatory effects of TRH on activation parameters
23 including the slowed activation rate and the shift in activation voltage dependence were
24
rP
25 not modified by any of the mutations (Fig. 3). This suggests that the impairment of the
26
27
hormonal effects caused by extending the size of the deletions from residue 333 up to
28 amino acid 326 was not due to elimination of any of the aforementioned consensus
29
ee
30 sites, but was possibly related instead to the deletion itself and/or to a general disruption
31 of a structural motif linked to this region in the channel molecule.
32
As a last test of the real contribution of the 326-332 sequence to the hormonal
rR
33
34 modulation, the TRH effects were tested in other constructs carrying only a short
35
36 deletion that includes the 326-332 sequence and its surroundings (construct ∆326-341),
37 and in variants in which the 326-341 sequence was replaced by sequences unrelated to
ev
38
39 HERG (constructs InsA/326-341 and InsB/326-341, see Methods). As shown in Fig. 4,
40 deleting the 326-341 segment or exchanging it with other unrelated sequences did not
41
iew
1 11
2
3 the KIKER with alanines (KKR/AAA construct) or by introducing a short deletion that
4
5 included the KIKER sequence (∆363-373 construct). Additionally, these mutations were
6
combined with the YTS/AAA triple mutant (YTS/AAA+KKR/AAA and
7
8 YTS/AAA+∆363-373 constructs) or with the ∆326-341 deletion (∆326-341+KKR/AAA
9
10
and ∆326-341+∆363-373 constructs), both affecting the RIRTYSK sequence. The
11 results obtained with all these constructs are summarized in Figures 5 and 6. It can be
12
13 observed that the triple mutant KKR/AAA shows a negatively shifted voltage
14 dependence of activation (V1/2 = -39.3 ± 2.3 mV, n=6; p<0.0001 vs wild-type). This
15
16 leftward shift was significantly bigger for the ∆363-373 construct (V1/2 = -51.5 ± 0.4
17 mV, n=9; p<0.005 vs KKR/AAA). Since an equivalent local charge is expected in the
18
19 region around the 363-373 residues in both channels (Fig. 5A), this indicates that some
20 additional disruption of the normal gating behaviour is induced by the deletion.
Fo
21
22 Remarkably, the TRH-induced displacement of the I/V curve to positive potentials
23 remained the same in the KKR/AAA mutant (22.3 ± 3.0 mV shift) and was only slightly
24
rP
25 reduced (11.9 ± 0.6 mV) in the ∆363-373 construct (Fig. 5C,D). Similar results for the
26
27
hormonal effect were obtained when the YTS/AAA or the ∆326-341 mutations around
28 the RYRTISK region were combined with the KKR/AAA or the ∆363-373
29
ee
30 modifications, respectively (Fig. 5D). Altogether, these results suggest that, in spite of
31 its crucial role in setting the basal activation parameters of HERG, the positively
32
charged KIKER sequence does not act as an important determinant of the hormonal
rR
33
34 regulatory effects on HERG activation behaviour.
35
36 As an additional indication that the KIKER sequence is not essential for the
37 TRH-induced effects on channel activation, we also compared the influence of the
ev
38
39 hormonal treatment on the time course of activation using the same constructs (Fig. 6).
40 Again, the slowing of activation gating caused by TRH in the wild-type channel
41
iew
42 remained the same in the triple mutant KKR/AAA, and was slightly reduced to 63% of
43
the control in the ∆363-373 construct. However, such slowing was more pronounced
44
45 when the KKR/AAA mutant was combined with the YTS/AAA or ∆326-341
46
47
modifications in the RYRTISK region. Finally, a slowing of gating equivalent to that
48 observed in the ∆363-373 construct was obtained when this deletion was combined with
49
50 the YTS/AAA or the ∆326-341constructs (Fig. 6B-D).
51
52
53 Impairment of the TRH-induced effects in ∆155-209 channels
54 The aforementioned results indicate that, different to our initial supposition (see
55
56 above), the impairment of the hormonal effects caused by the ∆326-373 and longer
57 amino terminal deletions is not due to the absence of a specific combination of residues,
58
59 but instead to the size of the deleted region beyond a critical point. This would be
60
coherent with the normal response observed in channels lacking the 16, 41 and 19
residues corresponding to sequences 326-341, 333-373 and 345-373, respectively, and
Page 13 of 34 Pflugers Archiv-European Journal of Physiology
1 12
2
3 with the absence of hormonal regulation in the constructs from which more than 48
4
5 residues are lacking (variants ∆326-373 to ∆138-373; see also [16]). To verify this
6
possibility we deleted a 55 amino acid region (variant ∆155-209) similar in size to that
7
8 of the ∆326-373 variant, but located near the amino terminal region of the proximal
9
10
domain. Analysis of the TRH effects on activation parameters of this new construct
11 demonstrated a significant reduction of the TRH-induced shifts in activation voltage
12
13 dependence (Fig. 7). Such a reduction of the hormonal effect was even more prominent
14 regarding the slowing down of the activation at 0 mV. Interestingly, the impairment of
15
16 the hormonal regulatory effects was observed in spite of the fact that the activation
17 properties of the ∆155-209 mutant remained the same as those of the wild-type
18
19 channels. This suggests that the lack of hormonal effects in channels lacking segments
20 326-373 (or with bigger deletions) is not due to alterations in the activation properties
Fo
21
22 induced by the deletions. It therefore supports the interpretation that the effect of the
23 longer deletions is due to structural alteration(s) in the cytoplasmic domain(s) leading to
24
rP
33
34 relatively unaltered in the amino terminal domains is important for the regulatory effects
35
36 triggered by TRH. This would explain the lack of hormonal effects once the structure
37 and/or the correct positioning of one or more of these domains towards the channel
ev
38
39 and/or its gating machinery is altered. The intracellular region that links HERG
40 transmembrane domains S4 and S5 has been proposed as a docking site for the N-
41
iew
42 terminal region and also as a possible linker between S4 movement and channel
43
opening or as a part of the activation gate itself [32,42]. Also, data from our laboratory
44
45 suggested that the interaction between amino terminal structures and the S4-S5 linker
46
47
(amino acids 540 to 546) may act as a determinant of changes in activation parameters
48 caused by deletion of proximal domain segments [1]. This prompted us to check
49
50 whether the modulatory effects of TRH on activation parameters can also be altered by
51 structural modifications in the S4-S5 linker. As shown in Fig. 8, the TRH-induced shifts
52
53 in activation voltage dependence are strongly reduced in channels with the single point
54 mutations D540C, R541H and Y542C. A sizeable but significantly smaller shift was
55
56 also observed in the R541A mutant. However, the TRH-induced alteration of activation
57 voltage dependence remained the same (or was even increased) when the amino acids
58
59 located in the carboxy terminal portion of the S4-S5 linker (residues 543, 544, 545 and
60
546) were substituted with cysteines. Analogous results were obtained when the TRH
effect on activation was tested during a time course. Thus, the slowing of activation
Pflugers Archiv-European Journal of Physiology Page 14 of 34
1 13
2
3 triggered by the hormone was also minimized by the mutations D540C, R541H, R541A
4
5 and Y542C, but not by the S543C, E544C, Y545C and G546C mutants (Fig. 9). It is
6
important to emphasize that the reduction of the hormonal effects was similarly
7
8 observed both in the D540C mutant (showing an activation voltage dependence
9
10
relatively shifted to positive values and less steep I/V curves; see [1]) and in the R541H
11 and Y542C mutants showing similar activation properties to those of wild-type
12
13 channels. This indicates that impairment of the TRH-induced effects is specifically due
14 to the disabling of the regulatory mechanism and not to a previous alteration of the
15
16 activation gating process as a consequence of the mutation itself. On the other hand, it
17 also identifies the initial amino-terminal portion of the S4-S5 linker as a crucial point
18
19 for transmission of the regulatory message to the gating machinery.
20 It is important to note that the bulk of constructs used in this study to map the
Fo
21
22 influence of amino terminal structural alterations on activation parameters, have been
23 designed to cover the second half of the proximal domain, previously proposed to be
24
rP
30 normal acceleration of closing rates equivalent to that obtained with wild-type HERG
31 was observed with most of the constructs used here (supplemental Fig 1). Note also that
32
although a clear impairment of the TRH effect on closing was observed with channel
rR
33
34 variants either deleted in the initial eag/PAS domain (∆2-135) or carrying single point
35
36 mutations in the amino half of the S4-S5 loop (e.g. D540 and R541 residues), a clear
37 interpretation of these data is complicated by the strong modification of closing rates
ev
38
39 induced by the mutations themselves (supplemental Fig 1 and ref [1]).
40
41
iew
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 15 of 34 Pflugers Archiv-European Journal of Physiology
1 14
2
3
4
5 DISCUSSION
6
7
8 A distinctive structural signature of HERG channels, that differentiates them
9
10
from the rest of voltage-dependent potassium channels, is the presence of a long stretch
11 of amino acids in the amino terminus (the proximal domain) that separates the initial
12
13 eag domain (conserved in members of the eag channel subfamily) from the first
14 transmembrane helix S1 [42,46]. Our previous studies using HERG deleted variants
15
16 indicated that elimination of the proximal domain induces kinetic alterations that favour
17 channel activation [15,42]. More recent work using the aforementioned channel variants
18
19 suggested that the proximal domain, particularly the sequence corresponding to residues
20 326-345, constituted a crucial component for TRH-induced modulation of HERG
Fo
21
22 activation in oocytes co-expressing TRH receptors and HERG channels [16]. In an
23 attempt to further delimit the molecular determinants of the hormonal effects, we
24
rP
25 generated a HERG variant from which residues 333-373 were deleted. Due to the fact
26
27
that normal TRH-induced responses are obtained with this variant but that a complete
28 elimination of the hormonal effects is achieved by increasing the size of the deletion by
29
ee
30 few additional residues (variant ∆326-373), we initially interpreted that the short
31 sequence extending from residues 326 to 332 was required for the hormonal regulation
32
of HERG activation to take place. The amino acid composition of this sequence was
rR
33
34 especially appealing since it included several consensus sites for phosphorylation by
35
36 different protein kinases and for interaction with regulatory components such as 14-3-3
37 proteins and inositol phospholipids, all of them recently proposed as putative regulators
ev
38
39 of ERG currents [6,7,19,23,47]. Surprisingly, introduction of single point mutations
40 (either individually or combined) in the phosphorylation sites of the 326-332 sequence
41
iew
42 and changing the two arginine residues included in this sequence to alanines, did not
43
alter the TRH-induced effects on activation voltage dependence and activation rates.
44
45 This suggests that the impairment of the hormonal effects caused by extending the size
46
47
of the deletions from residue 333 to amino acid 326 was not due to the elimination of
48 the consensus sites, but it is probably related to the deletion itself and/or to a structural
49
50 disruption of the cytoplasmic domains. An additional demonstration that this is the case
51 is provided by channel variants carrying: i) a short deletion that includes the 326-332
52
53 sequence and its surroundings, or ii) a short stretch of amino acids replacing the
54 sequence contained between residues 326-341 by a different sequence unrelated to
55
56 HERG. In all cases, these structural modifications do not significantly modify the TRH-
57 induced modifications of HERG activation voltage dependence and activation rates.
58
59 The demonstration that the amino acid sequence corresponding to positions 326-
60
332 does not determine by itself the hormonal regulatory effects opens up the possibility
that the lack of hormonal effects in channels without the segment 326-373 (or having
Pflugers Archiv-European Journal of Physiology Page 16 of 34
1 15
2
3 longer deletions) is due to the alteration(s) in activation behaviour caused by the
4
5 deletions themselves. This would be coherent with the absence of TRH effects in
6
variants ∆284-373, ∆223-373 and ∆138-373 showing a very accelerated activation and a
7
8 strong shift in the I/V curves to the left [16] and also with the fact that a normal TRH-
9
10
induced regulation is obtained with ∆333-373 channels with a slightly slower activation
11 and a certain displacement of the I/V curve to more depolarized values. However, a
12
13 normal TRH-induced regulation is observed in ∆345-373 and ∆355-373 channels in
14 which clear accelerations of activation rates and hyperpolarizing shifts are induced by
15
16 the deletions. Additionally, ∆155-209 channels display activation properties analogous
17 to those of wild-type HERG, but show a significant impairment of the hormonal effects.
18
19 Finally, the hormonal regulation is virtually abolished in single-point mutants of the S4-
20 S5 linker that do not show the altered activation kinetics encountered in the proximal
Fo
21
22 domain-deleted mutants. All these data together mean that no correlation exists between
23 the alterations in activation induced by a given mutation and its ability to impair the
24
rP
25 TRH-induced regulation.
26
27
Alternatively, it could be possible that the structural modifications introduced in
28 the channel variants showing a reduced hormonal response could cause a destabilization
29
ee
30 of some cytoplasmic domain(s) and/or their miss alignment towards the channel gating
31 machinery. This possibility is supported by the results obtained with variant ∆155-209
32
which has a deletion equivalent in size to that of the ∆326-373 construct but located in
rR
33
34 the amino-terminal portion of the proximal domain. Even though the hormonal
35
36 regulation was not impaired to the same extent as in the ∆326-373 channels, a
37 significant reduction of the TRH-induced effects was induced by the ∆155-209 deletion.
ev
38
39 Remarkably, the hormonal effect remained unaltered in channels from which the initial
40 135 amino acids (including the eag/PAS, ref. [25]) had been eliminated. This indicates
41
iew
42 that it is the amino terminal proximal domain but not the eag/PAS region which is
43
important for the TRH effects on HERG activation. It is also worth noting that the
44
45 modulatory effects triggered by the hormone were not impaired in channels carrying an
46
47
altered (or absent) KIKER sequence between residues 362 and 367. This demonstrates
48 that in spite of the crucial role of this charged amino acid cluster in setting the basal
49
50 HERG activation phenotype [32], it does not seem to be involved in the hormonal
51 modulation of activation parameters by TRH. Further work will be necessary to identify
52
53 the cytoplasmic structure(s) involved in transmission of the TRH effects to the HERG
54 gating machinery.
55
56 Previous work with HERG and other channels indicated that the S4-S5 linker
57 could play a crucial role in the coupling of the voltage sensing machinery to channel
58
59 activation [40]. This concept has been recently strengthened by structural evidence in a
60
Shaker family K+ channel [24]. It is also known that the S4-S5 linker acts as docking
site for regulation of HERG deactivation gating by amino terminal domains
Page 17 of 34 Pflugers Archiv-European Journal of Physiology
1 16
2
3 [15,26,34,44]. Our recent results also suggest that an interaction between the amino
4
5 terminus and the S4-S5 linker may contribute to HERG gating alterations induced by
6
amino terminal deletions [1] or extracellular acidification (Alonso-Ron et al.
7
8 Manuscript in preparation). The results reported here demonstrate that the hormonal
9
10
effects are strongly impaired by single point mutations in the amino terminal portion of
11 the S4-S5 linker. Interestingly, this is observed not only with mutants showing a
12
13 relatively strong alteration of their activation parameters (e.g. D540C channels), but also
14 with channels carrying much more conservative mutations (e.g. R541H) in which
15
16 activation properties similar to those of wild-type channels were obtained. Altogether,
17 this suggests that not only is a correct structural organization of the amino terminal
18
19 domains required, but also an appropriate arrangement of these domains towards the
20 central channel body is critical for hormonal modulation of HERG activation properties.
Fo
21
22 Furthermore, it identifies the initial portion of the S4-S5 linker as a key point for
23 transmission of the regulatory actions triggered by activation of PLC-coupled receptors
24
rP
30
31
32
rR
33
34
35
36
37
ev
38
39
40
41
iew
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pflugers Archiv-European Journal of Physiology Page 18 of 34
1 17
2
3
4
5 ACKNOWLEDGEMENTS
6
7
8 This work was supported by grants SAF2003-00329 from the Ministerio de Ciencia y
9
10
Tecnología, BFU2006-10936 from the Ministerio de Educación y Ciencia of Spain
11 (both partially co-financed by FEDER European Funds) and IB05-002 from the
12
13 Principado de Asturias (Spain). C.A.R. and D.G.M. are predoctoral fellows from the
14 Spanish Ministerio de Ciencia y Tecnología (refs. BES-2004-3872 and AP2000-4363).
15
16 P.M. holds a predoctoral fellowship from FICYT of Asturias (ref. BP03-108).
17
18
19
20
Fo
21
22
23
24
rP
25
26
27
28
29
ee
30
31
32
rR
33
34
35
36
37
ev
38
39
40
41
iew
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 19 of 34 Pflugers Archiv-European Journal of Physiology
1 18
2
3
4
5 REFERENCES
6
7
8 1. Alonso-Ron C, de la Peña P, Miranda P, Domínguez P, Barros F (2008)
9
10
Thermodynamic and kinetic properties of amino-terminal and S4-S5 loop HERG
11 channel mutants under steady-state conditions. Biophys J 94: 3893-3911
12
13 2. Barros F, Villalobos C, García-Sancho J, del Camino D, de la Peña P (1994) The role
14 of the inwardly rectifying K+ current in resting potential and thyrotropin-releasing
15
16 hormone-induced changes in cell excitability of GH3 rat anterior pituitary cells. Pflügers
17 Arch 426: 221-230
18
19 3. Barros F, del Camino D, Pardo LA, Palomero T, Giráldez T, de la Peña P (1997)
20 Demonstration of an inwardly rectifying K+ current component modulated by
Fo
21
22 thyrotropin-releasing hormone and caffeine in GH3 rat anterior pituitary cells. Pflügers
23 Arch 435: 119-129
24
rP
33
34 6. Bian J, Cui J, McDonald TV (2001) HERG K+ channel activity is regulated by
35
36 changes in phosphatidyl inositol 4,5-bisphosphate. Circ Res 89: 1168-1176
37 7. Cayabyab FS, Schlichter LC (2002) Regulation of an ERG K+ current by Src tyrosine
ev
38
39 kinase. J Biol Chem 277: 13673-13681
40 8. Cherubini A, Taddei GL, Crociani O, Paglierani M, Buccoliero AM, Fontana L, Noci
41
iew
1 19
2
3 12. de la Peña P, del Camino D, Pardo LA, Domínguez P, Barros F (1995) Gs couples
4
5 thyrotropin-releasing hormone receptors expressed in Xenopus oocytes to phospholipase
6
C. J Biol Chem 270: 3554-3559
7
8 13. Emmi A, Wenzel HJ, Schwartzkroin PA, Taglialatela M, Castaldo P, Bianchi L,
9
10
Nerbonne J, Robertson GA, Janigro D (2000) Do glia have heart?. Expression and
11 functional role for Ether-a-go-go currents in hippocampal astrocytes. J Neurosci 20:
12
13 3915-3925
14 14. Finlayson K, Witchel HJ, McCulloch J, Sharkey J (2004) Adquired QT interval
15
16 prologation and HERG: Implications for drug discovery and development. Eur J
17 Pharmacol 500: 129-142
18
19 15. Gómez-Varela D, de la Peña P, García J, Giráldez T, Barros F (2002) Influence of
20 amino-terminal structures on kinetic transitions between several closed and open states
Fo
21
22 in human erg K+ channels. J Membrane Biol 187: 117-133
23 16. Gómez-Varela D, Barros F, Viloria CG, Giráldez T, Manso DG, Dupuy SG,
24
rP
33
34 (Lond.) 547: 913-929
35
36 18. Jian M, Dun W, Fan J-S, Tseng G-N (1999) Use-dependent 'agonist' effect of
37 azimilide on the HERG channel. J Pharmacol Exp Ther 291: 1324-1336
ev
38
39 19. Kagan A, Melman YF, Krumerman A, McDonald TV (2002) 14-3-3 amplifies and
40 prolongs adrenergic stimulation of HERG K+ channel activity. EMBO J 8: 1889-1898
41
iew
42 20. Keating MT, Sanguinetti MC (2001) Molecular and cellular mechanisms of cardiac
43
arrhytmias. Cell 104: 569-580
44
45 21. Kemp BE, Pearson RB (1990) Protein kinase recognition sequence motifs. Trends
46
47
Biochem Sci 15: 342-346
48 22. Kennelly PJ, Krebs EG (1991) Consensus sequences as substrate specificity
49
50 determinants for protein kinases and protein phosphatases. J Biol Chem 266: 15555-
51 15558
52
53 23. Kiehn J (2000) Regulation of the cardiac repolarizing HERG potassium channel by
54 protein kinase A. Trends Cardiovasc Med 10: 205-209
55
56 24. Long SB, Campbell EB, MacKinnon R (2005) Voltage sensor of Kv1.2: Structural
57 basis of electromechanical coupling. Science 309: 903-908
58
59 25. Miranda P, Giráldez T, de la Peña P, Manso DG, Alonso-Ron C, Gómez-Varela D,
60
Domínguez P, Barros F (2005) Specificity of TRH receptor coupling to G-proteins for
Page 21 of 34 Pflugers Archiv-European Journal of Physiology
1 20
2
regulation of ERG K+ channels in GH3 rat anterior pituitary cells. J Physiol. (Lond.)
3
4
5 566: 717-736
6
7 26. Morais-Cabral JH, Lee A, Cohen SL, Chait BT, Li M, MacKinnon R (1998) Crystal
8 structure and functional analysis of the HERG potassium channel N terminus: a
9
10 eukariotic PAS domain. Cell 95: 649-655
11
12
27. Morita H, Wu J, Zipes DP (2008) The QT syndromes: long and short. Lancet 372:
13 750-763
14
15 28. Overholt JL, Ficker E, Yang T, Shams H, Bright GR, Prabhakar NR (2000) HERG-
16 like potassium current regulates the resting membrane potential in glomus cells of the
17
18 rabbit carotic body. J Neurophysiol 83: 1150-1157
19 29. Redfern WS, Carlsson L, Davis AS, Lynch WG, MacKenzie I, Palethorpe S, Siegl
20
PKS, Strang I, Sullivan AT, Wallis R, Camm AJ, Hammond TG (2002) Relationship
Fo
21
22 between preclinical cardiac electrophysiology, clinical QT interval prolongation and
23
24 torsade de pointes for a broad range of drugs: evidence for a provisional safety margin
rP
1 21
2
3 channel activation by protein kinase C independent of direct phosphorylation of the
4
5 channel protein. Cardiovasc Res 59: 14-26
6
39. Thomas D, Kiehn J, Katus HA, Karle CA (2004) Adrenergic regulation of the rapid
7
8 component of the cardiac delayed rectifier potassium current, IKr, and the underlying
9
10
hERG ion channel. Basic Res Cardiol 99: 279-287
11 40. Tristani-Firouzi M, Chen J, Sanguinetti MC (2002) Interactions between S4-S5
linker and S6 transmembrane domain modulate gating of HERG K+ channels. J Biol
12
13
14 Chem 277: 18994-19000
15
16 41. Trudeau MC, Warmke JW, Ganetzky B, Robertson GA (1995) HERG, a human
17 inward rectifier in the voltage-gated potassium channel family. Science 269: 92-95
18
19 42. Viloria CG, Barros F, Giráldez T, Gómez-Varela D, de la Peña P (2000) Differential
20 effects of amino-terminal distal and proximal domains in the regulation of human erg
Fo
21
22 K+ channel gating. Biophys J 79: 231-246
23 43. Viskin S (1999) Long QT syndromes and torsade de pointes. Lancet 354: 1625-1633
24
rP
25 44. Wang J, Trudeau MC, Zappia AM, Robertson GA (1998) Regulation of deactivation
26
27
by an amino terminal domain in Human ether-a-go-go-related gene potassium channels.
28 J Gen Physiol 112: 637-647
29
ee
30 45. Wang J, Myers CD, Robertson GA (2000) Dynamic control of deactivation gating
31 by a soluble amino-terminal domain in HERG K+ channels. Biophys J 115: 749-758
32
46. Warmke JW, Ganetzky B (1994) A family of potassium channel genes related to eag
rR
33
34 in Drosophila and mammals. Proc Natl Acad Sci USA 91: 3438-3442
35
36 47. Zhang Y, Wang H, Wang J, Han H, Nattel S, Wang Z (2003) Normal function of
37 HERG K+ channels expressed in HEK293 cells requires basal protein kinase B activity.
ev
38
39 FEBS Lett 534: 125-132
40
41
iew
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 23 of 34 Pflugers Archiv-European Journal of Physiology
1 22
2
3
4
5 FIGURE LEGENDS
6
7
8 Fig. 1. Effect of TRH on activation voltage dependence of HERG channels carrying
9
10
deletions in the amino terminal domains. (A) Schematic diagram of HERG amino
11 terminus in wild-type and deletion mutants. The regions corresponding to the eag/PAS
12
13 (residues 16-135) and the proximal domains up to the first transmembrane helix are
14 shown as striped and white bars, respectively. The portion corresponding to amino acids
15
16 1-16 that remained disordered in the eag/PAS crystalline structure [26] is not
17 represented in the scheme. Internal deletions are marked by dashed lines. The size of
18
19 every domain and the lengths of the deletions are represented on a horizontal scale
20 proportional to the total length of the amino terminus. (B) Families of currents obtained
Fo
21
22 in the absence (Control, upper part) or the presence of 1 µM TRH (TRH, lower part)
23 using the voltage protocol shown in the inset. Test pulses were applied once every 20-
24
30 s. Oocytes bathed in high-K+ medium were used for recordings. Note the fast
rP
25
26
27
deactivation kinetics exhibited by the ∆2-135 construct. (C) Effect of proximal domain
28 deletions on TRH-induced shifts in HERG activation voltage dependence. Averaged
29
ee
30 data for the number of oocytes indicated in panel D were used to generate the I/V curves
31 in the absence (open symbols) or the presence (solid symbols) of TRH. The continuous
32
lines correspond to Boltzmann curves h(V) = Imax [1/(1 + exp((V - V1/2)/k))], which
rR
33
34 best fitted the data with V1/2 of -11 and +10 mV for control and TRH-treated ∆333-373
35
36 channels. V1/2 values of ∆2-135, ∆326-373 and ∆345-373 channels corresponded to –
37 20.5 and –6.9, -60.0 and -58.5, and –44.0 and –24.0 mV in the absence or the presence
ev
38
39 of TRH, respectively. Boltzmann curves obtained in the absence of TRH from oocytes
40 expressing wild-type and proximal domain-deleted ∆138-373 channels are also included
41
iew
1 23
2
3 shown. The time necessary to reach a half-maximal tail current magnitude (dashed
4
5 lines) is indicated in the graphs. (C) Comparison of TRH effects on activation time
6
course of channels carrying different deletions in the amino terminus. One-hundred %
7
8 increase corresponds to a doubled t0.5 value. Data for wild-type and ∆138-373 channels
9
10
[16] are also included for comparison. The deletion mutants in this figure correspond to
11 those schematized in Fig 1A.
12
13
14 Fig. 3. Comparison of TRH-induced effects on activation parameters of HERG channels
15
16 carrying single point mutations in the 326-332 RYRTISK sequence. (A) Maintenance of
17 TRH-induced shifts in activation voltage dependence after mutation of residues 326,
18
19 327, 328, 329 and 331 to alanine. YTS/AAA corresponds to a triple mutant in which
20 residues 327, 329 and 331 were simultaneously changed to alanine. Activation voltage
Fo
21
22 dependence was studied as indicated in Fig. 1. Data normalized to the shift value
23 measured in wild-type channels are presented. n.s.: not significant vs. wild-type. (B)
24
rP
30
31 Fig. 4. Effect of TRH on activation properties of channels with structural modifications
32
around the 326-332 RYRTISK sequence. (A) Schematic diagram of HERG amino
rR
33
34 terminus in wild-type and modified HERG channels. Deletions are marked by dashed
35
36 lines. Insertions of a HERG-unrelated sequence covering the region deleted in the ∆326-
37 341 construct are represented by a dotted and a vertically striped bar in the InsA/326-
ev
38
39 341 and the InsB/326-341 variants, respectively. See Methods for more details about the
40 specific sequences introduced in these constructs. (B) Effect of TRH on activation
41
iew
42 voltage dependence (left) and the activation time course at 0 mV (right) of ∆326-341
43
channels. I/V curves in the absence or the presence of 1 µM TRH were generated as
44
45 detailed in the legend of Fig. 1. V1/2 and t0.5 values with and without TRH are indicated
46
47
in the graphs. A Boltzmann curve obtained without TRH from oocytes expressing wild-
48 type channels (dashed line) is also shown for comparison. Activation time courses were
49
50 followed as detailed in the legend of Fig. 2. (C) Comparison of TRH-induced shifts in
51 activation voltage dependence. Data for wild-type and ∆326-373 channels are also
52
53 included for clarity. (D) Comparison of TRH-induced slowing of HERG activation at 0
54 mV. Increases in t0.5 values are shown with a 100% increment representing a doubled
55
56 half-maximal activation time.
57
58
59 Fig. 5. Effect of TRH on activation voltage dependence of HERG channels with
60
modifications around the RYRTISK and KIKER sequences. (A) Schematic diagram of
HERG amino terminus and sequence alignment of residues 325-376 in the different
Page 25 of 34 Pflugers Archiv-European Journal of Physiology
1 24
2
3 constructs. (B) Families of currents obtained in the absence (Control, top) or the
4
5 presence of 1 µM TRH (TRH, bottom) using the voltage protocol shown in the inset.
6
(C) Comparison of the TRH-induced shifts in HERG activation voltage dependence for
7
8 different constructs. Averaged data for the number of oocytes indicated in panel D were
9 used to generate the I/V curves. V1/2 values before adding TRH are indicated in the
10
11 graphs. (D) Comparison of TRH-induced I/V shifts on channels with different
12
13 modifications in the region corresponding to the KIKER cluster either alone or
14 combined with mutations in the RYRTISK sequence.
15
16
17 Fig. 6. Effect of TRH on the activation time course of HERG channels carrying
18
19 modifications around the RYRTISK and KIKER sequences. (A) Families of currents
20 obtained in the absence (Control, top) or the presence of 1 µM TRH (TRH, bottom)
Fo
21
22 using the voltage protocol shown on top. Depolarizing voltages of –20 mV for
23 KKR/AAA and ∆326-341+KKR/AAA, and –40 mV for ∆363-373 and ∆326-
24
rP
25 341+∆363-373 constructs were used to compensate for the shifts in activation voltage
26
27
dependence. In all cases a repolarizing potential of –100 mV was used. (B) Comparison
28 of the activation time courses of different constructs. Averaged data normalized to
29
ee
30 maximum are shown. The time necessary to reach a half-maximal tail current
31 magnitude (dashed lines) with (solid symbols) and without TRH (open symbols) is
32
indicated in the graphs. (C) Comparison of TRH effects on activation time course of
rR
33
34 channels carrying different modifications in the region corresponding to the KIKER
35
36 cluster either alone or combined with mutations in the RYRTISK sequence. One-
37 hundred % increase corresponds to a doubled t0.5 value. The constructs in this figure
ev
38
39 correspond to those schematized in Fig 5A.
40
41
iew
42 Fig. 7. Effect of TRH on activation properties of the HERG channel construct ∆155-209
43
with a 55 amino acid deletion in the proximal domain. (A) Schematic diagram of HERG
44
45 amino terminus in wild-type, ∆326-373 and ∆155-209 HERG channels. (B) Effect of
46
47
TRH on ∆155-209 channel currents in response to the indicated voltage protocols for
48 studying the activation voltage dependence (left) and the activation time course at 0 mV
49
50 (right). (C) Effect of TRH on activation voltage dependence (left) and the time course of
51 activation at 0 mV (right) of ∆155-209 channels. A Boltzmann curve obtained without
52
53 TRH from oocytes expressing wild-type channels (dashed line) is also shown for
54 comparison. (D) Comparison of TRH-induced effects on activation voltage dependence
55
56 shifts and increases in t0.5 values of ∆155-209 channels. Data for wild-type and ∆326-
57 373 channels are also included for comparison.
58
59
60
Fig. 8. Effect of TRH on activation voltage dependence of HERG channels carrying
single point mutations in the S4-S5 linker. (A) Membrane currents from oocytes
Pflugers Archiv-European Journal of Physiology Page 26 of 34
1 25
2
3 expressing the indicated channel mutants are shown. Families of currents correspond to
4
5 individual representative oocytes in the absence (top panels) or the presence of 1 µM
6
TRH (bottom panels) using the voltage protocol shown in the inset. Oocytes bathed in
7
8 high-K+ medium were used for recordings. Note the enhanced time scale in the D540C,
9
10
Y542C and E544C mutant panels for a better view of the fast deactivating tails
11 exhibited by these channels. (B) Averaged I/V curves for the number of oocytes
12
13 indicated in panel C. Boltzmann curves obtained without TRH from oocytes expressing
14 wild-type channels (dashed lines) are also shown. (C) Comparison of TRH-induced I/V
15
16 shifts on channels with different single point mutations in the S4-S5 linker
17
18
19 Fig. 9. Effect of TRH on activation time course of HERG channels with single point
20 mutations in the S4-S5 linker. (A) Membrane currents from oocytes expressing the
Fo
21
22 indicated channel mutants. Families of currents correspond to individual representative
23 oocytes in the absence (top) or the presence of 1 µM TRH (bottom) using a voltage
24
rP
30 was used. (B) Activation time courses of the different mutants. Averaged data
31 normalized to maximum are shown. The time necessary to reach a half-maximal tail
32
current magnitude (dashed lines) with (solid symbols) and without TRH (open symbols)
rR
33
34 is indicated in the graphs. The scale of the x axes has been limited to 3 s for a better
35
36 view of the differences in the activation time course after adding TRH. (C) Comparison
37 of TRH effects on activation time courses of channels mutated in the S4-S5 linker. One-
ev
38
39 hundred % increase corresponds to a doubled t0.5 value.
40
41
iew
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 27 of 34 Pflugers Archiv-European Journal of Physiology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45 217x251mm (480 x 480 DPI)
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pflugers Archiv-European Journal of Physiology Page 28 of 34
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43 220x228mm (480 x 480 DPI)
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 29 of 34 Pflugers Archiv-European Journal of Physiology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48 178x259mm (480 x 480 DPI)
49
50
51
52
53
54
55
56
57
58
59
60
Pflugers Archiv-European Journal of Physiology Page 30 of 34
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48 213x250mm (480 x 480 DPI)
49
50
51
52
53
54
55
56
57
58
59
60
Page 31 of 34 Pflugers Archiv-European Journal of Physiology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49 229x298mm (480 x 480 DPI)
50
51
52
53
54
55
56
57
58
59
60
Pflugers Archiv-European Journal of Physiology Page 32 of 34
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
221x249mm (480 x 480 DPI)
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 33 of 34 Pflugers Archiv-European Journal of Physiology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47 215x285mm (480 x 480 DPI)
48
49
50
51
52
53
54
55
56
57
58
59
60
Pflugers Archiv-European Journal of Physiology Page 34 of 34
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48 217x288mm (480 x 480 DPI)
49
50
51
52
53
54
55
56
57
58
59
60
Page 35 of 34 Pflugers Archiv-European Journal of Physiology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48 214x287mm (480 x 480 DPI)
49
50
51
52
53
54
55
56
57
58
59
60
Supplemental Fig. 1. Deactivation properties of different constructs and effect of
structural modifications in the amino terminus and the S4-S5 loop on TRH-induced
acceleration of HERG deactivation. (A) Summary of fast deactivation rates at –100 mV.
We considered for analysis only the time constants of the fast deactivation current that
corresponds to the major component of current at negative voltages and that shows
minimal contamination with residual oocyte chloride currents triggered by the hormone
treatment. The rates of deactivation were determined from negative-amplitude
biexponential fits to the decaying phase of tail currents at –100 mV as detailed in ref [1]
of the paper. (B) Comparison of TRH effects on deactivation kinetics in channels
carrying different structural alterations in the amino-terminus. Bars in the histogram
represent TRH-induced reductions in the magnitude of fast deactivation time constant at
–100 mV as compared to that observed in wild-type channels, that amounted 22.6±3.7%
(n=24) in this set of experiments. In this case, a 50% reduction corresponds to a doubled
speed of closing. (C) Comparison of TRH effects on deactivation kinetics in channels
carrying single-point mutations in the S4-S5 loop (residues 540-546) and the 326-331
sequence of the proximal domain.
J Physiol 566.3 (2005) pp 717–736 717
Regulation of ether-à-go-go related gene (ERG) K+ or auditory stress suggests a link between hormonal
channel activity by the hypothalamic neuropeptide (e.g. adrenergic) stimulation and cardiac ion channel
thyrotropin-releasing hormone (TRH) constitutes an function including HERG (Thomas et al. 2004).
essential point of hormonal control of electrical activity, Furthermore, ERG channels also seem to play a key role
and hence of intracellular Ca2+ levels ([Ca2+ ]i ) and the setting the electrical behaviour of other cell types including
secretory response in anterior pituitary cells (Barros et al. neurones and glial, chromaffin, pancreatic β and tumour
1994, 1997; Weinsberg et al. 1997; Bauer, 1998; Bauer cells (Zhou et al. 1998a; Emmi et al. 2000; Rosati et al. 2000;
et al. 1998, 1999). The human ERG (HERG) channel Gullo et al. 2003; Sacco et al. 2003; Lastraioli et al. 2004).
has been also recognized as an important determinant Nevertheless, unlike the relatively well known molecular
of action potential characteristics in heart muscle and and kinetic characteristics of ERG channels and in spite
its inhibition by inherited mutations or a panoply of of their physiological and pathological relevance, many
cardiac- and noncardiac-related prescribed drugs has been of their molecular mechanisms of regulation by different
associated with an increased risk of cardiac arrhythmia physiological agents remain unclear.
and sudden death (Chiang & Roden, 2000; Keating & In native lactotrophs and clonal GH adenohypophysial
Sanguinetti, 2001; Redfern et al. 2003; Finlayson et al. cells, endogenous ERG currents are inhibited by activation
2004). Interestingly, the observation that arrhythmogenic of the G protein-coupled TRH receptor (TRH-R; Bauer
syncopes are usually associated with physical, emotional et al. 1990, 1994; Barros et al. 1992, 1993; Schäfer et al.
C The Physiological Society 2005 DOI: 10.1113/jphysiol.2005.085803
1999; Schledermann et al. 2001). The TRH-R is coupled to signal during the initial response to the hormone, this
a G-protein of the Gq/11 family (reviewed in Gershengorn G-protein is not involved in the TRH-induced inhibition
& Osman, 1996) resulting in phospholipase C of ERG currents. Dominant-negative variants of Gα 13 and
(PLC) activation and generation of myo-inositol Rho, but not of Gα q/11 , are able to significantly reduce
1,4,5-trisphosphate (IP3 ) and diacylglycerol (DAG) the inhibitory effect of TRH on ERG. Furthermore, a
from phosphatidylinositol 4,5-bisphosphate (PIP2 ). It prominent reduction is observed upon introduction of
is also known that TRH is able to activate several PKC dominant-negative Gα s . Interestingly, a strong reduction
isozymes in GH3 cells (Kiley et al. 1991; Akita et al. 1994). of the TRH-induced inhibition is also observed in cells
However, TRH-induced inhibition of ERG current in these overexpressing transducin α subunits (Gα t ), an agent
cells does not depend on PKC or PKA activation (Bauer known to sequester free G-protein βγ dimers (Crespo
et al. 1990, 1994; Barros et al. 1992, 1993; Schäfer et al. et al. 1994; Faure et al. 1994; Palomero et al. 1998). The
1999; Schledermann et al. 2001). Whether Gq/11 protein specificity of the dominant-negative and Gα t effects is
transduction, typically linked in many cells (including demonstrated by their failure to modify the Ca2+ response
adenohypophysial cells) to Ca2+ signalling, plays a role in the same cells. Furthermore, dominant-negative Gα q/11
in ERG channel regulation remains controversial. Thus, (but not Gα t or dominant-negative Gα s , Gα 13 and
a pathway for ERG regulation by TRH involving a G13 - Rho) was able to reduce TRH-induced release of Ca2+
and Rho-mediating signalling cascade has been described from intracellular stores in HEK293 cells permanently
in whole-cell voltage-clamped GH4 C1 cells (Storey et al. expressing HERG channels and TRH-Rs. In these cells,
2002), but the relative importance of this transduction however, only the dominant-negative forms of Gα 13 and
pathway as compared with the ‘classical’ Gq/11 -mediated Rho (but not Gα t or dominant-negative Gα s ) were able
TRH-induced signalling was not assessed. Furthermore, to antagonize the modifications in activation voltage
Gq/11 but not Gi/o or G13 has been recently shown to dependence induced by TRH on HERG currents. On
mediate muscarinic inhibition of ERG currents in tsA-201 the other hand, only Gα t and dominant-negative Gα q/11
cells coexpressing rat ERG1 channels and M1 muscarinic expression reduced the TRH-induced HERG current
receptors (Hirdes et al. 2004). In adenohypophysial cells, inibition at positive voltages. Apart from emphasizing the
the Gq/11 -mediated coupling of TRH-R to PIP2 hydrolysis specificity of the different dominant-negative constructs,
leads to an initial elevation of [Ca2+ ]i via IP3 that this also suggests that the cellular background and/or
mediates a peak of secretion associated with a transient the channel isoform may influence the transduction
hyperpolarization of the cell membrane due to activation mechanism(s) involved in hormonal regulation of ERG.
of Ca2+ -dependent K+ channels (Gómez-Varela et al.
2003b). However, normal inhibition of ERG channel Methods
activity in response to TRH has been observed in
individual cells in which the initial Ca2+ response is totally Plasmids and chemicals
absent (Barros et al. 1991, 1992, 1994). Although the The original plasmid containing the cDNA for the
PKC branch of the PLC signalling cascade is necessary for HERG channel was a generous gift of Dr E. Wanke
reversal of the TRH-induced ERG inhibition in GH3 cells (University of Milan, Italy). pEGFP-N3 plasmid was
(Gómez-Varela et al. 2003b) and PIP2 depletion could obtained from Clontech. pcDNA3.1 plasmids containing
lead to HERG current reduction in HEK293 cells (Bian dominant-negative forms of Gα q (Gα q -Q209L/D277N),
et al. 2001), the TRH-induced inhibition of the GH3 cell Gα 13 (Gα 13 -Q266L/D294N), Gα s (Gα s -Q227L/D295N)
r-ERG current also takes place after blockade of PIP2 and RhoA (RhoA-T19N 3xHA-tagged-NH2) were
consumption with a PLC inhibitor (Gómez-Varela et al. obtained from Guthrie (Guthrie cDNA Resource Center;
2003b). These results and the reported modification of the currently transferred to University Missouri-Rolla cDNA
TRH effects on r-ERG in cholera toxin-treated GH3 cells Resource Center, Rolla, MO, USA). Gα t was cloned in
(Barros et al. 1994; Bauer et al. 1994) open the possibility pcDNA3 as an EcoRI/XhoI fragment transferred from
that, at least in adenohypophysial cells, a transduction pcDNAI (provided by Dr J. S. Gutkind, N. I. of Dental
cascade(s) involving either a Gs -like protein or a G13 - and Research, N.I.H., Bethesda, MD, USA). TRH and nystatin
Rho-based pathway, couples the TRH-R to endogenous were purchased from Sigma. E-4031 and anti-HERG poly-
ERG channel inhibition. clonal antibodies were from Alomone Laboratories; Fura-2
In this report we use double mutants (Gα-QL/DN) of and Fura-2/AM were from Molecular Probes.
G-protein α subunits able to act as dominant-negative
inhibitors against specific G-proteins (Yu et al. 2000) to
GH3 cell culture and transfection
explore the specificity of TRH-R coupling to G-proteins for
ERG K+ channel inhibition in GH3 rat anterior pituitary GH3 rat anterior pituitary cells (ATCC-CCL 82.1) were
cells. Our results demonstrate that whereas the TRH-R plated in 35-mm diameter tissue culture plastic dishes
certainly uses a Gq/11 protein for transducing the Ca2+ containing sterile glass coverslips coated with poly l-lysine
C The Physiological Society 2005
and grown at 37◦ C in a humidified atmosphere of 95% 100 U ml−1 penicillin, 0.1 mg ml−1 streptomycin and 10%
air and 5% CO2 . The culture medium consisted of a fetal bovine serum. HEK-H36/T1 cells were maintained
1 : 1 mixture of Dulbecco’s modified Eagle’s medium in the presence of 1 mg ml−1 geneticin sulphate and
and Ham’s F-12 nutrient mixture (Sigma) supplemented 150 µg ml−1 hygromycin B (Gibco) and plated on
with 100 U ml−1 penicillin, 1.1 mg ml−1 streptomycin and the poly-l-lysine-coated coverslips for recording.
a serum mixture of 15% horse serum and 2.5% fetal Transient transfection of HEK-H36/T1 cells was
bovine serum. The coverslips constituted the bottom performed following the procedures indicated above for
of a small recording chamber (0.2–0.3 ml) that was the GH3 cells.
continuously perfused with saline at a rate of about
1 ml min−1 . Cells trypsinized 24 h prior to transfection
lying in poly-l-lysine-coated coverslips were transiently
Electrophysiological recordings, solutions
transfected using Lipofectamine 2000 according to the
manufacturer’s instructions (Invitrogen, Carlsbad, CA, and data analysis
USA). Unless otherwise indicated, 5.0 µg of plasmids Current recordings were performed at room temperature
containing the different constructs and pEGFP-N3 with the perforated-patch variant of the patch-clamp
codifying green fluorescent protein (eGFP) as a marker technique as previously described (Barros et al. 1991,
for transfection in a 10 : 1 ratio were used. The mixture 1992, 1994, 1997; Gómez-Varela et al. 2003b). Electro-
of 5.5 µg of total DNA and Lipofectamine was incubated des were fabricated from borosilicate or kimax disposable
in serum-free medium for 20 min and added to the micropipettes (Boralex, Rochester Scientific, Rochester,
plates containing the cells in serum-containing medium NY; Fisherbrand, Fisher Scientific, Pittsburg, PA or Kimble
without antibiotics. Recordings were performed 24–48 h glass Inc., Vineland, NJ, USA). Electrode resistance
after transfection. amounted 2–5 M when filled with the pipette solution
containing (mm): 65 KCl, 30 K2 SO4 , 10 NaCl, 1 MgCl2 ,
Generation and isolation of permanently transfected 50 sucrose and 10 Hepes (pH 7.4 with KOH). The
tip of the pipette was initially filled with nystatin-free
HEK 293 cell clones
solution and the remainder of the pipette was back-filled
HERG channel cDNA was subcloned into HindIII/BamHI with the same solution also containing 250 µg ml−1
sites of the pcDNA3 vector (Invitrogen). Monolayer nystatin, added from a stock of 50 mg ml−1 nystatin
cultures (∼50% confluent) of human embryonic kidney freshly dissolved in dimethylsulphoxide. These solutions
cells (HEK293; ATCC CRL-1573) were transfected with were sonicated just before use. The course of perforation
this construct using Lipofectamine (Gibco). Three days was followed by monitoring the progress of capacitive
after transfection, the cells were trypsinized and diluted in transients under voltage-clamp mode, setting the pipette
a medium containing 1 mg ml−1 geneticin. Subsequently voltage at a value of −70 mV. Access resistance, as
they were cultured until cell colonies were visible. estimated from the capacitive compensation circuitry on
Individual colonies were picked with cloning cylinders the amplifier, reached 10–30 M within 5–20 min after
and tested for HERG currents. A clone named H36 was the seal was made. Solution junction potentials were
selected for further transfection. Cells of clone H36 nulled before seal formation. Once patch permeabilization
were cotransfected with plasmid pcDNA3.1/Hygro(+) reached the indicated levels, the extracellular solution was
(Invitrogen) containing the cDNA for the TRH-R (de changed as indicated and the cell was voltage-clamped
la Peña et al. 1992) inserted between the HindIII/XbaI at the desired holding potential. An EPC-7 patch-clamp
sites of the vector. Hygromycin B (150 µg ml−1 ) was amplifier (HEKA Elektronic, Lambrecht, Germany) was
used to select H36 clones coexpressing HERG channels used to record membrane currents. Stimulation, data
and TRH-Rs. We chose for further work a clone named acquisition and analysis were carried out using Pulse
HEK-H36/T1 showing: (a) robust HERG currents under and PulseFit software (HEKA Elektronic) running on
voltage-clamp, (b) reproducible calcium responses Macintosh computers. Current records were sampled
when perfused with TRH after loading the cells with every 1 ms and digitally filtered at 500 Hz. r-ERG current
the fluorescent Ca2+ indicator Fura-2, and (c) a pre- data are shown without correction for leakage and
dominant level of a 155 kDa band of HERG protein, capacitative transients. A P/n method was used for leak and
corresponding to the mature and more glycosylated capacitive current subtraction of the HERG recordings in
HERG likely to be located in the plasma membrane HEK-H36/T1 cells. Further data processing was performed
(Zhou et al. 1998b; Petrecca et al. 1999), immunodetected with PulseFit and Igor Pro (WaveMetrics, Lake Oswego,
in cell extracts with HERG-specific antibodies. Cells OR, USA).
were grown at 37◦ C in a humidified atmosphere of The standard extracellular saline used for perforation
95% air and 5% CO2 . HEK 293 cells were cultured and monitoring [Ca2+ ]i contained (mm): 137 NaCl, 4 KCl,
in the same medium as GH3 cells supplemented with 1.8 CaCl2 , 1 MgCl2 , 10 glucose, and 10 Hepes (pH 7.4
C The Physiological Society 2005
with NaOH). Recordings of r-ERG currents in GH3 cells poly-l-lysine-coated coverslips as indicated above.
were performed after changing the extracellular medium In this case the coverslips were transferred to wells
to high-K+ , Ca2+ -free solution once permeabilization of containing standard extracellular saline plus 5 µm
the patches had been completed. This solution contained Fura-2/AM (Molecular Probes) and loaded with the
(mm): 140 KCl, 4 MgCl2 , 10 EGTA and 10 Hepes titrated dye for about 60 min at room temperature. After
to pH 7.4 with KOH. Inward currents were studied during loading with Fura-2, cells were washed with saline to
hyperpolarization pulses to −100 mV from a holding remove non-hydrolysed Fura-2/AM and left for another
potential of −10 mV. The hyperpolarization pulses were 30 min before recording to facilitate AM hydrolysis
preceded by a 100 ms ramp from 0 to −50 mV that by cellular esterases. Fluorescence measurements were
can yield an estimation of the membrane conductance performed in a Axiovert 100 microscope equipped with
within this voltage range and would tend to potentiate the a Plan-NeoFluar 40×/0.75 objective and epifluorescence
otherwise voltage-dependent effect of TRH (Bauer et al. accessories (Carl Zeiss), attached to a fluorescence imaging
1990; Barros et al. 1992, 1994, 1997). To prevent variations system (TILL-Photonics GmbH, Martinsried, Germany).
due to differences in deactivation rates from cell to cell, Control of the monochromator (Polychrome IV) and
the magnitude of the inward currents was estimated with the 12-bit cooled CCD camera (IMAGO) was performed
the PulseFit software as the total inward charge computed using TILLvisION imaging software. Cells were excited
between cursors located at 0.5 and 100% duration of the through a dichroic mirror reflecting less than 395 nm
hyperpolarization pulses. HERG currents were recorded light. Fluorescence signals were filtered through a 410 nm
in HEK-H36/T1 cells in standard extracellular saline long-pass filter. Cycles of sample excitation were repeated
following the pulse protocols indicated on the figures. every 500 ms consisting in 10/20 ms periods of irradiation
Kinetic parameters of activation and deactivation were with 340, 360 and 380 nm light. The ratio of the emission
obtained as previously described (Barros et al. 1998; Viloria intensities (340 nm/380 nm) was used as a measure for
et al. 2000). The voltage dependence of current activation changes in intracellular Ca2+ . When eGFP-transfected
was assessed using standard tail current analysis. Tail cells were used, a correction for eGFP fluorescence due to
current magnitudes normalized to maximum were fitted residual eGFP excitation at 340–380 nm was performed.
with a Boltzmann function: Using eGFP-containing cells without Fura-2 we estimated
h(V ) = Imax [1/(1 + exp((V − V1/2 )/k))] previously that 13% of the fluorescence recovered above
510 nm (using a GFP filter set with a 500 nm dichroic
where V is the test potential, V1/2 is the half-activation and a 510 nm long-pass filter) following eGFP excitation
voltage, and k is the slope factor. Steady-state voltage at 488 nm is also present upon eGFP excitation at
dependencies of activation were obtained as previously 380 nm using the conventional Fura-2 set-up. Less than
described (Viloria et al. 2000) applying depolarization 1% was recovered when eGFP was excited at 340 nm.
pulses of variable magnitude and up to 10 s duration from Subsequently, the eGFP-derived fluorescence at 380 nm
two holding potentials: +40 mV to hold the channels fully was estimated in every individual cell loaded with Fura-2
open and −80/−100 mV to hold them fully closed. The from the (13%) fluorescence intensity at 488 nm (a
position of the Boltzmann curves under true steady-state wavelength at which Fura-2 is not excited). This amount
conditions was estimated as an extrapolated mean from was subtracted from the total fluorescence at 380 nm to
the curves obtained at both holding potentials to ensure isolate the Fura-2-specific signal. Ca2+ concentrations
that they were a function exclusively of depolarization were estimated from the 340 nm/380 nm fluorescence
pulse characteristics, regardless of the previous (open or ratio by comparison with Fura-2 standards (Barros et al.
closed) state of the channels. The time course of activation 1994).
was monitored using an indirect envelope of tail current
protocol (Viloria et al. 2000), varying the duration of the
depolarization pulse and following the variation in the Statistics
magnitude of the tail currents recorded after going back to a Data values given in the text and in figures with error
negative voltage. The rates of deactivation were determined bars represent the mean ± s.e.m. for the number of
from negative-amplitude biexponential fits to the decaying indicated cells. Comparison between data groups was at
phase of tail currents. The first cursor of the fitting window first performed by parametric Student’s unpaired t test
was advanced to the end of the initial hook due to the (2-tailed) or ANOVA. Due to dispersion of the data in
recovery of inactivation. individual cells after some treatments (e.g. Gα 13 -QL/DN
and RhoA T/N transfections in Fig. 4), non-homogeneous
variances (as evidenced after a Bartlett’s test) were
Intracellular calcium measurements
sometimes obtained. Therefore, alternate Welch’s test
Measurements of intracellular Ca2+ concentrations assuming Gaussian populations with unequal s.d.s and
([Ca2+ ]i ) were performed in cells platted in a non-parametric Wilcoxon or Mann-Whitney test that
C The Physiological Society 2005
does not make any assumption about the scatter of the data Gα q -mediated PLC activation and activation of K+
were also used to evaluate significance of mean differences currents by Ca2+ released from intracellular stores via
between cell populations. For a posteriori comparison of IP3 , because they were determined without extracellular
two specific samples a Bonferroni or a Dunn test for Ca2+ . Furthermore, the effect of Gα q -QL/DN was specific
multiple comparisons was also used following one-way since the TRH-induced currents remained unaltered in
ANOVA or Kruskal-Wallis non-parametric ANOVA tests, cells expressing the dominant-negative α subunit of G13 ,
respectively. In all cases, P-values < 0.05 were considered a G-protein predictably unrelated to the PLC–IP3 Ca2+
as indicative of statistical significance. signalling pathway.
For an additional demonstration of the effectiveness and
Results specificity of the dominant-negative approach we studied
Validation of the dominant-negative strategy the effect of the Gα-QL/DN proteins on hormone-induced
using Gαq -QL/DN and the TRH-induced Ca2 + [Ca2+ ]i increases. For this purpose, the fluorescence of
Fura-2 was monitored in the transfected cells as indicated
response of the GH3 cells
in Methods. As an internal control, TRH-induced
To examine the transduction pathway leading to ERG variations in fluorescence of the cells present in the
inhibition upon TRH stimulation, we expressed xanthine same microscope field but not expressing the transfection
nucleotide binding mutants of different G-protein α marker (eGFP) were monitored. We determined first that
subunits. These mutants carrying a double mutation transfection or expression of the eGFP biosensor itself does
(leucine and asparagine substituting for a glutamine not modify the [Ca2+ ]i increases induced by the hormone,
and an aspartate, respectively) possess a lowered affinity using GH3 cells transfected with eGFP and pcDNA3B
by guanine nucleotides and an enhanced affinity by plasmid lacking any G-protein codifying insert. Data
xanthine nucleotides that make them form stable and from a representative experiment indicate that whereas
specific complexes with cognate receptors and compete addition of 1 µm TRH raised [Ca2+ ]i from a basal averaged
with endogenous wild-type G-proteins (Yu et al. 2000). value of 105 ± 32 nm (n = 35) to an initial maximum of
We first probed the effectivity of this approach using 236 ± 32 nm in the 35 cells of a microscope field lacking
dominant-negative Gα q (Gα q -Q209L/D277N) and eGFP fluorescence, the [Ca2+ ]i level was increased from
exploring the initial (Phase 1) TRH-dependent response 80 ± 31 to 240 ± 29 nm in the 13 cells from the same field
of the GH3 cells. This phase corresponds to a transient showing a clear eGFP expression. On the other hand, in
release of stored Ca2+ into the cytosol due to production both cases most of the cells showed a significant increase
of IP3 by PLC-catalysed PIP2 hydrolysis, leading to in peak [Ca2+ ]i regardless of the presence or the absence
an initial and transient hyperpolarization of the cell of eGFP expression.
membrane by activation of Ca2+ -dependent K+ channels The aforementioned results strongly differ from
(Gómez-Varela et al. 2003b). We reasoned that blockade of those obtained with cells transfected with eGFP and
this transduction pathway with Gα q -QL/DN should mini- a plasmid coding for Gα q -QL/DN (Fig. 2A). In this
mize all the cellular responses that characterize this phase case, TRH raised the basal [Ca2+ ]i level of 73 ± 9 nm to
such as (i) the transient membrane hyperpolarization, 141 ± 14 nm in the 26 cells showing no detectable eGFP
(ii) the transient increase in potassium currents expression. In contrast, only a peak value of 61 ± 5 nm
that determine this hyperpolarization, and (iii) the [Ca2+ ]i from a basal level of 49 ± 6 nm was obtained
transient [Ca2+ ]i increase that activates the upon TRH addition in the 11 cells of the same field
Ca2+ -dependent K+ currents. As shown in Fig. 1A the showing a clear eGFP fluorescence. It is also important
transient hyperpolarization of the cell membrane induced to note that in this case only 3 of the 11 cells expressing
by TRH in perforated-patch current-clamped GH3 cells eGFP showed any significant increase in [Ca2+ ]i as
is nearly abolished in cells expressing Gα q -QL/DN (see compared with the 18 in which [Ca2+ ]i was clearly
averaged voltage traces in the insets). Interestingly, as increased from 26 cells not expressing the transfection
shown in the individual cell recordings illustrated in marker. Furthermore, whereas only a modest [Ca2+ ]i
Fig. 1A, the increase in the rate of production of action increase from 61 ± 6 to 94 ± 4 nm was observed in those
potentials (Phase 2 of hormone action) that follows the three cells expressing eGFP, a prominent initial peak
initial hyperpolarization was maintained in the presence of Ca2+ that raised [Ca2+ ]i from 73 ± 9 to 172 ± 7 nm
of the dominant-negative. As a second validation of the was obtained in the 18 cells in which eGFP (and hence
dominant-negative effect, we studied the appearance of Gα q -QL/DN) expression was not detectable. Analogous
membrane currents immediately after TRH addition in results were obtained in two additional experiments.
patch-perforated voltage-clamped GH3 cells bathed in This demonstrates that (i) a much lower percentage
high-K+ Ca2+ -free solution. Figure 1B shows that the of cells expressing the dominant-negative variant of
current increases induced by TRH were absent in cells Gα q respond to TRH, and (ii) even in the few cells
expressing Gα q -QL/DN. Thus, these increases represent expressing Gα q -QL/DN that show a detectable response
C The Physiological Society 2005
Figure 1. Blockade by dominant-negative Gαq -QL/DN of the initial phase 1 of response induced by TRH
in GH3 cells
A, effect of dominant-negative Gα q on TRH-induced Phase 1 of hyperpolarization. Representative recordings of
membrane potential are shown in two cells either expressing the dominant-negative form of Gα q (lower trace)
or not (upper trace). Application of 1 µM TRH is marked with a horizontal line on top of the traces. Note the
maintenance of Phase 2 of increased electrical activity in the Gα q -QL/DN-transfected cell. Traces averaged point
by point from several cells are shown in the insets. Continuous current traces averaged point by point and their
corresponding S.E.M. are shown. In this cases, traces were synchronized to the time of TRH addition as indicated
with an arrow. B, effect of dominant-negative Gα q and Gα 13 on Ca2+ -dependent K+ currents elicited in GH3
cells by TRH during the initial Phase 1 of response. Continuous current traces averaged point by point and their
corresponding S.E.M. are shown for cells transfected with vector pcDNA3B (Control 3B), or with plasmids codifying
the dominant-negative mutants of Gα q (Gα q -QL/DN) and Gα 13 (Gα 13 -QL/DN). Traces were synchronized to the
time of 1 µM TRH addition as indicated with an arrow. High-K+ , low-Ca2+ extracellular solution and a potential of
−30 mV were used for better detection of inward K+ currents activated by Ca2+ released from intracellular stores
via IP3 .
C The Physiological Society 2005
Figure 2. Effect of dominant-negative Gα subunits or Gαt expression on Ca2+ liberation from GH3 cell
intracellular stores in response to TRH
The time course of variations in [Ca2+ ]i levels is shown for Fura-2 loaded cells from a microscope field transfected
with Gα q -QL/DN (A), Gα s -QL/DN (B), Gα 13 -QL/DN (C), or Gα t (D). Addition of 1 µM TRH is indicated by horizontal
lines on top of the traces. Continuous traces averaged point by point and their corresponding S.E.M. are shown.
Averaged basal levels are signalled by dashed lines. Data correspond to cells showing fluorescence of the trans-
fection marker (eGFP+ , left) or not (eGFP− , right). Variations of [Ca2+ ]i levels in the cell subpopulations showing a
detectable peak Ca2+ increase after visual inspection are illustrated. In all cases the number of averaged cells with
respect to the total number of cells present in the field is boxed. Data from the whole cell population present in
the microscope field are shown in the insets of panel A. Averaged values from all data points before TRH addition
and that of the initial maximum are indicated. Analogous results were obtained in two additional experiments.
C The Physiological Society 2005
C The Physiological Society 2005
to the hormone, the magnitude of the [Ca2+ ]i increase is present in GH3 cells and to quantify its inhibition
clearly smaller than that of cells in which the transfection by TRH, high-K+ low-Ca2+ extracellular solutions and
marker (and hence the dominant-negative Gα q ) has not established voltage protocols were used. Thus, due to
been expressed. the fast inactivation of ERG channels at depolarized
potentials and the presence of several outwardly rectifying
voltage- and calcium-dependent K+ currents in GH3 cells,
Specificity of dominant-negative Gα subunits on currents were studied during hyperpolarization pulses
TRH-induced Ca2+ response to −100 mV from a holding potential of −10 mV (see
The results presented above indicate that expression of the Methods). High-K+ low-Ca2+ extracellular solutions were
dominant-negative form of Gα q constitutes an efficient also used to increase the amplitude of the inwardly
way to suppress the Gq-dependent initial Ca2+ response rectifying r-ERG currents and to reduce Ca2+ currents
to TRH of the GH3 cells. Whereas coupling of end- and activation of Ca2+ -dependent K+ currents (Bauer
ogenous and heterologously expressed TRH-Rs to Gq et al. 1990, 1999; Barros et al. 1992, 1997; Weinsberg
for activation of PLC-mediated PIP2 hydrolysis has been et al. 1997). Furthermore, since under these conditions
widely documented, it has been also reported that in GH the TRH-induced current inhibition is almost exclusively
cells the TRH-R can interact with a number of different exerted on ERG currents, 5 µm of the ERG-specific blocker
G-proteins that may include G13 , Gi and Gs or a Gs -like E-4031 (a concentration that totally blocks the r-ERG
protein (Storey et al. 2002; reviewed by Gershengorn & current) was added at the end of the experiments to
Osman, 1996). To check the possible specificity of the subtract the E-4031-insensitive currents from the initial
dominant-negatives we also studied the Ca2+ response in ones and compare the difference between the TRH- and
cells expressing the QL/DN variants of Gα s and Gα 13 . The E-4031-blocked currents in every individual cell (see
elevations of [Ca2+ ]i in response to TRH remained the Gómez-Varela et al. 2003b). Representative examples of
same in cells expressing either Gα s -QL/DN (Fig. 2B) or TRH effects on r-ERG currents in cells transfected with
Gα 13 -QL/DN (Fig. 2C) as compared with cells from the different G-protein α subunit variants are shown in Fig. 3.
same microscope field lacking eGFP expression. Similar Only successfully expressing cells identified by their eGFP
results were obtained in cells expressing transducin α fluorescence were used for recording. As shown in Figs 3A
subunits (Fig. 2D; see also below). Furthermore, as in and 4, the inhibitory effect of TRH on the E-4031-sensitive
control cells transfected with pcDNA3B (see above) or current was not modified by the transfection procedure.
in untransfected cells (not shown), most of the cells Thus, a value of 76.8 ± 2.7% (n = 20) was obtained in cells
expressing dominant-negatives of Gα s or Gα 13 showed transfected with eGFP and pcDNA3B plasmid lacking any
significant increases in peak [Ca2+ ]i . This indicates that G-protein coding insert, as compared with the 78.6 ± 5.4%
whereas coupling of TRH-R to Gq/11 is indispensable for (n = 8) inhibition obtained in cells showing no detectable
the TRH-evoked Ca2+ response, coupling to a G-protein eGFP expression. Similar inhibitions have been previously
of the Gs or G13 type is not required for this effect. reported under identical conditions using untransfected
cells (Gómez-Varela et al. 2003b). Most importantly, the
TRH-induced inhibition was the same in cells expressing
Effect of different dominant-negative Gα subunits on dominant-negative Gα q -QL/DN (74.8 ± 6.3%, n = 6). It
TRH-induced inhibition of endogenous r-ERG currents is important to note that failure to significantly modify
in GH3 cells the TRH-induced inhibition of r-ERG was not due to
lack of efficient expression of Gα q -QL/DN, since the early
To investigate the transduction cascade linking the increases in Ca2+ -dependent K+ currents induced by TRH
TRH-R to r-ERG channel inhibition we always used (a Gq-dependent and PLC/IP3 /Ca2+ -related effect, see
perforated-patch conditions to minimize cell dialysis and above) were absent in the same cells (Fig. 3B). Apart from
to preserve intact the intracellular components necessary adding further support to specificity of the Gα q -QL/DN
for the hormonal response. To isolate the r-ERG current for the Ca2+ response, this indicates that whereas the
and those corresponding to the minimum following addition of 5 µM E-4031 were considered as 0 and 100%,
respectively. Filled circles correspond to the current traces shown above. The first one or two data points following
addition of TRH, when total inward currents become transiently enhanced by activation of Ca2+ -dependent
K+ channels due to massive liberation of Ca2+ from intracellular stores, have been deleted for clarity. Perfusion
of 1 µM TRH and addition of E-4031 to the recording chamber are indicated. Values for the data in the absence
and presence of TRH are indicated by horizontal dashed lines in the time courses. A continuous membrane current
recording at a holding potential of −30 mV around the time of hormone addition is shown in B and marked with a
thick arrow. Note the total absence of transient Ca2+ -dependent K+ currents following TRH addition in the same
cell showing a clear reduction of r-ERG currents. Similar results were obtained in the five additional cells expressing
Gα q -QL/DN.
C The Physiological Society 2005
TRH receptor certainly uses a Gq/11 protein for trans- of two subpopulations. Whereas nearly half of the cells
duction of the Ca2+ signal during the initial response showed TRH-induced r-ERG inhibitions above 50%
to the hormone, this G-protein is not involved in the (equivalent to those of controls and untransfected cells),
TRH-induced inhibition of endogenous r-ERG currents a clear reduction of the hormonal effects leaving the
in GH3 cells. TRH-induced inhibition below 50% took place in the other
Recent experiments in GH4 C1 cells using a constitutively half of the cell population (see Fig. 4).
active mutant of Gα 13 and whole-cell recording showed It has been shown that the activation of G13 - (and Gq -)
a partial inhibition of r-ERG currents equivalent to dependent pathways is able to signal to different effectors
that induced by TRH under similar conditions (Storey through the small G-protein RhoA (Seasholtz et al. 1999;
et al. 2002). Using our perforated-patch conditions, the Sah et al. 2000; Dutt et al. 2002; Vogt et al. 2003). Expression
inhibition of r-ERG by TRH was attenuated when the of the dominant-negative RhoA-T19N also significantly
G13 pathway was antagonized with dominant-negative attenuated r-ERG modulation by TRH, leaving a mean
Gα 13 . Thus, an inhibition of 56.4 ± 6.9% (n = 12, P < 0.01 inhibition of 53.4 ± 6.3% (n = 11, P < 0.01 versus control,
versus control pcDNA3B-transfected cells, Welch and Welch and Mann-Whitney tests; Figs 3D and 4). Again, a
Mann-Whitney tests) in the E-4031-sensitive r-ERG considerable dispersion of the data was obtained in this
currents was induced by TRH in cells transfected with case. Nevertheless, these data confirm previous results
1.6 µg of plasmid encoding Gα 13 -QL/DN (4 : 1 ratio in GH4 C1 cells and suggest that a transduction cascade
versus EGFP plasmid; not shown). This value was involving G13 (but not Gq ) and Rho may participate in the
slightly decreased to 43.8 ± 6.4 (n = 15) when 5.0 µg inhibition of ERG channels by TRH.
of the same plasmid were used (Figs 3C and 4). Previous results in GH3 cells showed a modification of
This effect was specific, as demonstrated by the total the TRH-induced effects on r-ERG in cells treated with the
absence of Gα 13 -QL/DN influence on TRH-induced Ca2+ Gs -modifying agent cholera toxin (Barros et al. 1994; Bauer
responses (see above). Interestingly, the magnitude of et al. 1994). Although a coupling of TRH-R to Gs with sub-
the TRH-induced inhibition in individual cells showed sequent activation of adenylyl cyclase has been reported in
a huge dispersion at both DNA concentrations, which GH3 cells, evidence against (i) stimulation of the enzyme
made it appear that the 27 cell sample was composed by TRH and (ii) specific down-regulation of Gα s following
long-term expositions of GH3 cells to TRH has been
obtained also (reviewed in Gershengorn & Osman, 1996).
Furthermore, a cholera toxin-dependent degradation of
Gα s has been demonstrated in GH3 cells (Chang & Bourne,
1989), but it is also known that cholera toxin treatment can
cause a significant reduction in the number of TRH-Rs
(Yajima et al. 1988). This prompted us to check whether
antagonization of the Gs pathway with dominant-negative
Gα s could cause any effect on the TRH-induced r-ERG
current reductions. As shown in Figs 3E and 4, a prominent
reduction of the inhibitory effect of TRH on r-ERG
was observed in the presence of dominant-negative
Gα s -QL/DN. In this case, mean inhibition amounted only
to 33 ± 3.9% (n = 10, P < 0.0001 versus control, Student’s
t and Mann-Whitney tests). This value is also significantly
smaller than that obtained in the presence of RhoA-T19N
(P < 0.02, Mann-Whitney test). It is important to note
that in the presence of dominant-negative Gα s not only is
the reduction of the TRH-induced inhibition the biggest
Figure 4. Percentage inhibition of r-ERG currents induced by observed, but also the dispersion of the data is notably
TRH in GH3 cells expressing dominant-negative Gα subunits reduced (Fig. 4). This supports the conclusion that Gs plays
and Gαt a crucial role on the inhibitory effects of TRH in GH3 cells.
Current inhibitions were estimated from total inward charge as
described in Fig. 3. Data from individual cells averaged on the bars are
shown as open circles. Values from cells transfected with dominant
Effect of transducin α subunit expression on the
negative variants of Gα q , Gα 13 , RhoA and Gα s , or transducin α
subunits (Gα t ) are shown. Data from cells transfected with vector TRH-induced inhibition of r-ERG in GH3 cells
pcDNA3B (Control 3B) and from untransfected cells (NO transf.) are
also shown for comparison. Significant variations among population
It has been recognized that receptors that activate G13 also
medians as evidenced by Mann-Whitney test or ANOVA and post hoc couple to Gq and G11 . It is also well known that Gα 13
multiple comparison test are indicated. and Gα q/11 signals induce Rho activation and subsequent
C The Physiological Society 2005
cellular responses by inducing GDP–GTP exchange in here. As shown in Fig. 7, the HEK-H36/T1 cells showed
Rho guanine nucleotide exchange factors in response to voltage-activated K+ currents typical of HERG under the
extracellular stimuli (Seasholtz et al. 1999; Sah et al. 2000; perforated-patch conditions chosen to maintain as intact
Dutt et al. 2002; Vogt et al. 2003). Thus it seemed surprising as possible the hormonal responses. Apart from HERG
that dominant-negative Gs (but not Gq ), G13 and Rho currents, we frequently detected also other outward K+
share antagonism on the TRH-induced effects on r-ERG. currents during the depolarization steps. The magnitude
To check the possibility that reduction of TRH-induced and the degree of inactivation of these currents during the
inhibition involves a component common to Gs and G13 depolarization pulses were highly variable. Nevertheless,
but not available in heterotrimeric Gq , we studied the their contribution was negligible along the tail current
consequences of expressing transducin α subunits (Gα t ), time course, as evidenced by their absence after treating
a scavenger of βγ dimers after they are released from the cells with the specific HERG inhibitor E-4031 (Fig. 5A).
G-proteins by receptor stimulation (Crespo et al. 1994; Thus only HERG current characteristics would be referred
Faure et al. 1994; Palomero et al. 1998). The inhibition to by limiting the analysis to tail currents. Addition of
of r-ERG by TRH was strongly attenuated by Gα t TRH to HEK-H36/T1 cells loaded with Fura-2 triggered
(33.5 ± 4.4%; n = 14, P < 0.0001 versus control, Student’s a transient increase in cytoplasmic Ca2+ that slowly
t and Mann-Whitney tests; Figs 3F and 4) up to levels declined to basal values upon hormone washout. On
equivalent to those observed with dominant-negative Gα s . the average, the amplitude of this Ca2+ response was
This effect of Gα t was totally specific, since both the equivalent to that elicited in the same cells by carbachol
percentage of cells showing a detectable increase in peak (data not shown) probably acting through the endogenous
[Ca2+ ]i and the magnitude of the TRH-induced Ca2+ muscarinic receptors of HEK293 cells. The TRH-induced
response remained the same regardless of the presence Ca2+ responses were absent in the cell clones containing
or the absence of Gα t (Fig. 2D). As discussed below, this HERG channels but not coexpressing TRH-Rs.
suggests that free βγ subunits released from Gs (and The simultaneous presence of TRH-R and HERG
perhaps shared by G13 ) heterotrimers may be responsible channels in HEK-H36/T1 cells prompted us to check the
for the TRH-induced inhibition of endogenous r-ERG effect of the hormone on HERG currents in this putatively
currents in GH3 cells. simple and defined cellular background. TRH reduced
tail HERG currents within 1–3 min after introducing
the hormone in the recording chamber (Fig. 5A). The
Generation of a HEK293 cell clone permanently
current level was lowered by 36 ± 2% (n = 38) when
expressing HERG channels and TRH receptors and
measured at the peak of the tails that followed pulses of
characterization of the TRH response 1 s duration at +40 mV. Furthermore, a strong shift in
Unlike the strong and quite reproducible current current availability voltage dependence to more positive
inhibitions induced by TRH on the endogenous ERG voltages was also caused by TRH. Thus the V1/2 value of
current, only modest effects of the hormone have been the Boltzmann functions describing the I–V curves was
reported on other kinetic characteristics of ERG either end- shifted from −1.5 ± 1 to +25.5 ± 2 mV (n = 14) under
ogenous or heterologously expressed in GH3 cells (Barros these conditions (Fig. 5A). It is important to note that the
et al. 1992; Bauer et al. 1998; Schledermann et al. 2001). diminished tail currents remaining after treatment with
This fact and the sometimes variable effect of dominant TRH were entirely due to the operation of HERG channels,
negatives in the individual cells prevented us from reaching since they were abolished by E-4031.
any consistent conclusion about the way in which different Due to the slow activation and deactivation kinetics
G proteins could be affecting other current parameters in of HERG channels, no steady-state conditions would
these cells. be expected within 1 s depolarizations. A possible cause
HEK293 cells permanently expressing HERG constitute of the shift in activation voltage dependence could be
an interesting and widely used model system to study easily a displacement of the V 1/2 values to more depolarized
the biochemical and electrophysiological properties of the potentials due to a slower activation rate leading to a less
channel (Zhou et al. 1998b). For this reason, we initially steady-state condition (Schönherr et al. 1999; Viloria et al.
isolated several cell clones expressing HERG. Subsequently, 2000). In fact, it is known that the activation rate of HERG
they were cotransfected with a plasmid containing the channels in Xenopus oocytes is slowed by activation of
TRH-R cDNA (de la Peña et al. 1992). Single colonies coexpressed TRH-Rs (Barros et al. 1998). For this reason,
were selected and tested for both HERG current under we also tested the effects of TRH on HERG activation
voltage-clamp and TRH-induced Ca2+ responses with rates in HEK-H36/T1 cells using an indirect envelope of
the fluorescent Ca2+ indicator Fura-2 (see Methods). tail currents protocol (Barros et al. 1998; Viloria et al.
A cell line named HEK-H36/T1 showing high HERG 2000). As shown in Fig. 5B, the time required to attain a
current density and also systematic Ca2+ increases when half-maximum current magnitude was increased by TRH
exposed to TRH was used for the experiments reported near an order of magnitude between +40 and +60 mV.
C The Physiological Society 2005
Thus the difference in the inflection potentials of the was also detected in response to TRH when the I–V
I–V curves could be due, at least in part, to the different curves were generated from tail currents following long
activation rates around the V1/2 values. Nevertheless, an depolarization steps of 10 s duration (Fig. 6A). Finally,
18 mV shift from −22.5 ± 2.2 to −4.1 ± 2.8 mV (n = 6) the position of the curves under real steady-state was
C The Physiological Society 2005
extrapolated as the mean of those obtained from hyper- The deactivation properties of HERG can be obtained
polarized (−80 mV) and depolarized (+40 mV) holding from the time course of current decay during voltage steps
potentials (Schönherr et al. 1999; Viloria et al. 2000). In to negative potentials, following depolarizing pulses at a
this case a 15 mV shift in V1/2 (from −25.0 to −10.0 mV, fixed time and voltage designed to activate (and inactivate)
n = 2) was obtained (Fig. 6B). Altogether, this suggests that the channels. In this case, the closing time constants
two distinct effects on activation contribute to the shifts can be estimated by fitting a bi-exponential function to
when studied with 1 s short depolarization steps: a genuine the decaying tail current phase that follows the initial
shift of the voltage dependence of activation and a marked peak once channel inactivation has been removed by the
slowing of activation rates moving the channels further repolarization. Both the fast and the slow exponential
away from steady-state conditions. components of current deactivation were significantly
Figure 6. Effect of TRH on HERG activation voltage dependence under steady-state conditions
A, effect of TRH on activation voltage dependence studied using long depolarization steps of 10 s. Tail currents upon
repolarization to −50 mV were obtained following 10 s depolarizations to different voltages as indicated at the
top. Only the end of the depolarizing steps and the tail currents at −50 mV are shown for clarity. Traces correspond
to a cell before (Control) and 3 min after introduction of 100 nM TRH (TRH). Averaged I–V curves before and after
the hormonal treatments are shown at the bottom. The magnitude of the peak tail current upon repolarization was
determined from recordings as shown at the top and normalized to maximum of tail current magnitude without
TRH. B, effect of TRH on HERG steady-state voltage dependence of activation. Steady-state voltage dependence
of activation was studied following the protocols shown in the two upper panels, with a prepulse of varying
magnitude and a duration ranging from 1 to 10 s, followed by a pulse test to −50 mV. Holding potentials of
+40 mV to keep the channels fully open and −80 mV to hold them fully closed were used as indicated. Families
of currents during the test pulse for Control and TRH-treated cells following 10 s prepulses to different potentials
are shown. Fractional activation curves before and after TRH treatment are shown at the bottom. Open and filled
symbols correspond to data obtained from −80 and +40 mV holding voltages, respectively. Data obtained at
prepulse duration of 1 s (circles), 5 s (squares), and 10 s (triangles) are plotted. In every curve current values were
normalized to tail current magnitude in response to the more depolarized potential of the prepulse series. The
dashed lines represent the deduced position of the activation curves under steady-state conditions obtained as
a mean of those corresponding to both holding voltages and prepulses of 10 s duration. The V1/2 values for the
steady-state curves are indicated by arrows in the graphs.
C The Physiological Society 2005
C The Physiological Society 2005
with endogenous wild-type G proteins (Yu et al. 2000). The presence of a heterogeneous population of binding
Accordingly, it could be expected that all signalling sites for TRH in GH3 cells has been previously reported
pathways lying on a given receptor became inhibited by (Gautvik & Lystad, 1981) and two TRH-R isoforms
any Gα-QL/DN variant interacting with it. Our results derived from the same gene by differential splicing
demonstrate that this is not the case for the TRH-R. mechanisms have been shown to be expressed in GH3
Figure 7. Effect of dominant-negative Gα subunits and Gαt expression on Ca2+ liberation from
HEK-H36/T1 cell intracellular stores in response to TRH
Protocols and data evaluation were performed as described in Fig. 2. Only data from the cell subpopulations
showing a visually detectable peak Ca2+ increase are plotted. In all cases the number of averaged cells respect
to the total number of cells present in the field is boxed. Analogous results were obtained in two additional
experiments.
C The Physiological Society 2005
cells (de la Peña et al. 1992). However, no significant functionality and specificity of the Gα-QL/DN variants
differences in binding or signalling properties of these allowed us to explore the TRH-R coupling to defined
receptors seem to exist (de la Peña et al. 1992; Lee et al. G proteins for transducing the inhibitory signal to ERG
1995; Gershengorn & Osman, 1996). Furthermore, the channels. Our results suggest that Gα s plays a crucial role
presence of two different molecular species of TRH-R transducing the TRH signal to native r-ERG channels
could not explain the dominant-negative specificity in GH3 cells, since dominant-negative Gα s causes the
found in HEK-H36/T1 cells in which only an isoform of greatest reduction of the hormonal effect as well as the
the receptor is expressed. Clearly, further work will be smallest data dispersion in individual cells. These data
necessary to understand the reason(s) for the observed are consistent with previous results showing that the
specificity of the dominant-negative effects. TRH-induced r-ERG inhibition is enhanced by short-term
Regardless of the exact mechanism by which the treatment with cholera toxin (an agent able to specifically
negative dominance takes place, the demonstration of the modify Gα s functionality) and nearly abolished when the
Figure 8. Effect of dominant-negative Gα subunits and Gαt on TRH-induced modifications of HERG I–V
curves in HEK-H36/T1 cells
A, effect of Gα subunit expression on HERG activation voltage dependence. Voltage dependence of activation
was studied in the absence (open symbols) or the presence (filled symbols) of 1 µM TRH by varying the magnitude
of a 1 s depolarizing pulse as detailed in the legend of Fig. 5A. Data normalized to maximum of tail current
magnitude without TRH are shown. V1/2 values obtained from the Boltzmann curves that best fitted the averaged
data (continuous lines) and the number of recorded cells are indicated in the graphs. B, comparison of TRH-induced
shifts on activation voltage dependence in cells expressing different Gα subunits. Data from the same cells used
to generate the averaged I–V curves shown in panel A were used. Values of TRH-induced shifts from the different
individual cells were averaged and are shown in the histogram. C, effect of Gα subunit expression on TRH-induced
inhibition of HERG currents at positive voltages. Data from the same cells used to generate the averaged I–V curves
shown in panel A were used. Values of TRH-induced reductions in maximal currents from the different individual
cells were averaged and are shown in the histogram. Statistically significant variations versus controls in Student’s
t or Mann-Whitney tests are indicated in B and C.
C The Physiological Society 2005
treatment with the toxin is prolonged (Barros et al. 1993, Comparison of TRH and dominant-negative effects in
1994; Bauer et al. 1994). They will be also coherent with native GH3 cells and a heterologous expression system
the reported inhibition of TRH-induced PLC stimulation such as the HEK-H36/T1 cells reveals some quantitative
in Xenopus oocytes using nucleotides antisense to Gα s (de and qualitative differences. These include (i) considerably
la Peña et al. 1995; but see Gershengorn & Osman, 1996). smaller TRH-induced current inhibitions in HEK-H36/T1
The reasons why coupling of TRH-R to Gs in GH3 cells cells, (ii) a clear inability of dominant-negative Gα s
only modestly stimulates adenylyl cyclase (Paulssen et al. to alter TRH-induced modifications of both activation
1992; Gershengorn & Osman, 1996) and why prolonged voltage dependence and HERG current magnitude in
stimulation with TRH down-regulates Gq/11 but has no HEK-H36/T1 cells, (iii) a significant reduction of the
effect on the Gα s protein levels (Kim et al. 1994) remain I–V curve shifts induced by TRH in HEK-H36/T1 cells
to be established. As suggested previously (Barros et al. upon expression of Gα 13 -QL/DN and RhoA-T/N without
1993, 1994; Bauer et al. 1994), it is possible that a Gs -like any concomitant alteration of current inhibition, and
protein that is not Gs itself, but sensitive to cholera toxin (iv) the maintenance of a TRH-induced positive shift
and blocked by Gα s -QL/DN expression, is involved in GH3 of the I–V curves in Gα t -transfected HEK-H36/T1 cells
cell r-ERG inhibition by TRH. in which the presence of the βγ scavenger abolishes
Using the dominant-negative approach we also studied the HERG current inhibition induced by the hormone.
the possible implication of Gα 13 and RhoA, two entities Use of perforated-patch conditions excludes uncontrolled
recently proposed as mediators of r-ERG modulation alterations of the intracellular environment as a cause for
by TRH (Storey et al. 2002). In this case, expression these differences. Instead, our results point to differences
of dominant-negative Gα 13 -QL/DN significantly reduced in the cellular background and/or the molecular identity
the TRH-induced inhibition, although a slightly smaller of the channel protein as determinant of them. The
effect was observed than that with Gα s -QL/DN and parallel effects of TRH on ERG channels either end-
only in around 50% of the recorded cells was the ogenous or overexpressed in GH3 /B6 cells (Schledermann
hormonal effect clearly antagonized. It is unlikely that this et al. 2001) suggest that the cellular background may
situation is caused by an unspecific inhibition of Gs by influence the transduction mechanism(s) involved in
dominant-negative Gα 13 because smaller TRH-induced hormonal regulation of the channels. This is further
inhibitions and wider dispersion of the data were also supported by recent results showing that Gq/11 , but
observed with dominant-negative RhoA, a component not Gi/o or G13 , mediates hormonal inhibition of ERG
located downstream of G13 in many cellular systems and currents in tsA-201 cells coexpressing rat ERG1 channels
(supposedly) not directly related to Gs . On the other hand, and another Gq/11 -coupled receptor, the M1 muscarinic
it seems difficult to conceive a transduction mechanism receptor (Hirdes et al. 2004). Some differences between the
involving Gα s , Gα 13 and RhoA (but not Gα q ) either signal cascade mediating the TRH-induced modulation
simultaneously or indistinctly. Our data following the of HERG in Xenopus oocytes and the physiological
overexpression of Gα t , an agent known to sequester free pathway modulating the native ERG channels in GH3
G-protein βγ dimers (Crespo et al. 1994; Faure et al. cells have been also reported (Barros et al. 1998). It
1994; Palomero et al. 1998), offer a possible explanation is important to highlight that the dominant-negative
of this apparent paradox. Thus, prominent reductions of effects reported here may vary also as a function of
the TRH-induced r-ERG inhibition equivalent to those the kinetic parameter being considered. Thus whereas
obtained with dominant-negative Gα s are observed in only an attenuation of the TRH-induced I–V shifts is
Gα t -transfected GH3 cells. The specificity of the Gα t observed in HEK-H36/T1 cells in which the G13 pathway
effect was demonstrated by its failure to modify the is blocked with dominant-negative G13 or RhoA, only the
Ca2+ response in the same cells. It is tempting to hormone-induced current inhibition but not the I–V shift
speculate that a specific set of free βγ subunits released is antagonized in Gα t -transfected cells. The possibility that
from Gs heterotrimers (or a similar combination pre- more than one hormone-activated mechanism exists for
sent in G13 but not in Gq ) may be responsible for the regulation of different channel properties is reinforced
TRH-induced inhibition of endogenous r-ERG currents by recent data showing that separate protein segments
in GH3 cells. Interestingly, both Gα and Gβγ subunits in the amino terminus and/or different structural
have been implicated in activation of Rho (Niu et al. rearrangements of the channel molecule are necessary
2003). Furthermore, GH3 cells constitute perhaps the for the TRH-induced modifications of HERG activation
best characterized example in which specific hormone and deactivation gating in Xenopus oocytes (Gómez-Varela
receptors have been shown to use G protein heterotrimers et al. 2003a).
of different αβγ subunit composition to modulate In summary, the results presented in this report lead us
an ionic channel (reviewed in Robishaw & Berlot, to propose that free βγ subunits released from Gs (and
2004). perhaps shared by G13 ) heterotrimers are responsible for
C The Physiological Society 2005
TRH-induced inhibition of the endogenous r-ERG current Bauer CK, Davison I, Kubasov I, Schwarz JR & Mason WT
in GH3 cells, in which reductions of current magnitude (1994). Different G proteins are involved in the biphasic
constitute the major component of the hormonal response of clonal rat pituitary cells to thyrotropin-releasing
response. Differences in the subunit composition of the hormone. Pflugers Arch 428, 17–25.
βγ dimers associated with Gq could explain the inability Bauer CK, Engeland B, Wulfsen I, Ludwig J, Pongs O &
Schwarz JR (1998). RERG is a molecular correlate of the
of Gq heterotrimers to serve a similar function in these cells.
inward-rectifying K current in clonal rat pituitary cells.
Variations in the cellular background upon heterologous Receptors Channels 6, 19–29.
expression of the channels may influence the transduction Bauer CK, Meyerhof W & Schwarz JR (1990). An
mechanism(s) involved in hormonal regulation of ERG, inward-rectifying K+ current in clonal rat pituitary cells and
but also the kinetic characteristics modified in response its modulation by thyrotropin-releasing hormone. J Physiol
to an agonist. Thus, as suggested by the similarities in 429, 169–189.
the effects of Gα q -QL/DN and Gα t , βγ dimers released Bauer CK, Schäfer R, Schiemann D, Reid G, Hanganu I &
from a G protein different from Gs (probably Gq ) are Schwarz JR (1999). A functional role of the erg-like
probably involved in the relatively small current inhibition inward-rectifying K+ current in prolactin secretion from rat
detected in HEK-H36/T1 cells. However, it is a Gα 13 - lactotrophs. Mol Cell Endocrinol 148, 37–45.
and RhoA-dependent pathway what seems to determinate Bian J, Cui J & McDonald TV (2001). HERG K+ channel
activity is regulated by changes in phosphatidyl inositol
the prominent alterations in HERG activation voltage
4,5-bisphosphate. Circ Res 89, 1168–1176.
dependence triggered by TRH in these cells heterologously Chang F-H & Bourne HR (1989). Cholera toxin induces
expressing HERG channels and TRH-Rs. cAMP-independent degredation of Gs. J Biol Chem 264,
5352–5357.
References Chiang C-E & Roden DM (2000). The long QT syndromes:
genetic basis and clinical implications. J Am Coll Cardiol 36,
Akita Y, Ohno S, Yajima Y, Konno Y, Saido TC, Mizuno K et al. 1–12.
(1994). Overproduction of a Ca2+ -independent protein Crespo P, Xu N, Simonds WF & Gutkind JS (1994).
kinase C isozyme, nPKCε, increases the secretion of Ras-dependent activation of MAP kinase pathway mediated
prolactin from thyrotropin-releasing hormone-stimulated by G-protein βγ subunits. Nature 369, 418–420.
rat pituitary GH4C1 cells. J Biol Chem 269, 4653–4660. de la Peña P, del Camino D, Pardo LA, Dominguez P & Barros
Barros F, del Camino D, Pardo LA, Palomero T, Giráldez T & F (1995). Gs couples thyrotropin-releasing hormone
de la Peña P (1997). Demonstration of an inwardly rectifying receptors expressed in Xenopus oocytes to phospholipase C.
K+ current component modulated by thyrotropin-releasing J Biol Chem 270, 3554–3559.
hormone and caffeine in GH3 rat anterior pituitary cells. de la Peña P, Delgado LM, del Camino D & Barros F (1992).
Pflugers Arch 435, 119–129. Two isoforms of the thyrotropin-releasing hormone receptor
Barros F, Delgado LM, del Camino D & de la Peña P (1992). generated by alternative splicing have indistinguishable
Characteristics and modulation by thyrotropin-releasing functional properties. J Biol Chem 267, 25703–25708.
hormone of an inwardly rectifying K+ current in Dutt P, Kjoller L, Giel M, Hall A & Toksoz D (2002). Activated
patch-perforated GH3 anterior pituitary cells. Pflugers Arch Gαq family members induce Rho GTPase activation and
422, 31–39. Rho-dependent actin filament assembly. FEBS Lett 531,
Barros F, Delgado LM, Maciá C & de la Peña P (1991). Effects 565–569.
of hypothalamic peptides on electrical activity and Emmi A, Wenzel HJ, Schwatzkroin PA, Taglialatela M, Castaldo
membrane currents of ‘patch perforated’ clamped GH3 P, Bianchi L et al. (2000). Do glia have heart? Expression and
anterior pituitary cells. FEBS Lett 279, 33–37. functional role for Ether-a-go-go currents in hippocampal
Barros F, Gómez-Varela D, Viloria CG, Palomero T, Giráldez T astrocytes. J Neurosci 20, 3915–3925.
& de la Peña P (1998). Modulation of human erg K+ channel Faure M, Voyno-Yasenetskaya TA & Bourne HR (1994). cAMP
gating by activation of a G protein-coupled receptor and and βγ subunits of heterotrimeric G proteins stimulate the
protein kinase C. J Physiol 511, 333–346. mitogen-activated protein kinase pathway in COS-7 cells.
Barros F, Mieskes G, del Camino D & de la Peña P (1993). J Biol Chem 269, 7851–7854.
Protein phosphatase 2A reverses inhibition of inward Finlayson K, Witchel HJ, McCulloch J & Sharkey J (2004).
rectifying K+ currents by thyrotropin-releasing hormone in Adquired QT interval prolongation and HERG: implications
GH3 pituitary cells. FEBS Lett 336, 433–439. for drug discovery and development. Eur J Pharmacol 500,
Barros F, Villalobos C, Garcı́a-Sancho J, del Camino D & de la 129–142.
Peña P (1994). The role of the inwardly rectifying K+ current Gautvik KM & Lystad E (1981). Demonstration of a hetero-
in resting potential and thyrotropin-releasing geneous population of binding sites for thyroliberin in
hormone-induced changes in cell excitability of GH3 rat prolaction-producing tumour cells and their possible
anterior pituitary cells. Pflugers Arch 426, 221–230. functional significance. Eur J Biochem 116,
Bauer CK (1998). The erg inwardly rectifying K+ current and 235–242.
its modulation by thyrotropin-releasing hormone in giant Gershengorn MC & Osman R (1996). Molecular and cellular
clonal rat anterior pituitary cells. J Physiol 510, biology of thyrotropin-releasing hormone receptors. Physiol
63–70. Rev 76, 175–191.
C The Physiological Society 2005
Gómez-Varela D, Barros F, Viloria CG, Giráldez T, Manso DG, Redfern WS, Carlsson L, Davis AS, Lynch WG, MacKenzie I,
Dupuy SG et al. (2003a). Relevance of the proximal domain Palethorpe S et al. (2003). Relationships between preclinical
in the amino-terminus of HERG channels for regulation by a cardiac electrophysiology, clinical QT interval prolongation
phospholipase C-coupled hormone receptor. FEBS Lett 535, and torsade de pointes for a broad range of drugs: evidence
125–130. for a provisional safety margin in drug development.
Gómez-Varela D, Giráldez T, de la Peña P, Dupuy SG, Cardiovasc Res 58, 32–45.
Garcı́a-Manso D & Barros F (2003b). Protein kinase C is Robishaw JD & Berlot CH (2004). Translating G protein
necessary for recovery from the thyrotropin-releasing subunit diversity into functional specificity. Curr Op Cell
hormone-induced r-ERG current reduction in GH3 rat Biol 16, 206–209.
anterior pituitary cells. J Physiol 547, Rosati B, Marchetti P, Crociani O, Lecchi M, Lupi R, Arcangeli
913–929. A et al. (2000). Glucose- and arginine-induced insulin
Gullo F, Ales E, Rosati B, Lecchi M, Masi A, Guasti L et al. secretion by human pancreatic β-cells: the role of HERG K+
(2003). ERG K+ channel blockade enhances firing and channels in firing and release. FASEB J 14,
epinephrine secretion in rat chromaffin cells: the missing 2601–2610.
link to LQT2-related sudden death? FASEB J 17, Sacco T, Bruno A, Wanke E & Tempia F (2003). Functional
330–332. roles of an ERG current isolated in cerebellar Purkinje
Hirdes W, Horowitz LF & Hille B (2004). Muscarinic neurons. J Neurophysiol 90, 1817–1828.
modulation of erg potassium current. J Physiol 559, Sah VP, Seasholtz TM, Sagi SA & Brown JH (2000). The role of
67–84. Rho in G protein-coupled receptor signal transduction.
Keating MT & Sanguinetti MC (2001). Molecular and cellular Annu Rev Pharmacol Toxicol 40, 459–489.
mechanisms of cardiac arrhythmias. Cell 104, 569–580. Schäfer R, Wulfsen I, Behrens S, Weinsberg F, Bauer CK &
Kiley SC, Parker PJ, Fabbro D & Jaken S (1991). Differential Schwarz JR (1999). The erg-like current in rat lactotrophs.
regulation of protein kinase C isozymes by thyrotropin- J Physiol 518, 401–416.
releasing hormone in GH4 C1 cells. J Biol Chem 266, Schledermann W, Wulfsen I, Schwarz JR & Bauer CK (2001).
23761–23768. Modulation of rat erg1, erg2, erg3 and HERG K+ currents by
Kim G-D, Carr IC, Anderson LA, Zabavnik J, Eidne KA & thyrotropin-releasing hormone in anterior pituitary
Milligan G (1994). The long isoform of the rat cells via the native signal cascade. J Physiol 532,
thyrotropin-releasing hormone receptor down-regulates Gq 143–163.
proteins. J Biol Chem 269, 19933–19940. Schönherr R, Rosati B, Hehl S, Rao VG, Arcangeli A, Olivotto
Lastraioli E, Guasti L, Crociani O, Polvani S, Hofmann G, M & Wanke E (1999). Functional role of the slow activation
Witchel H et al. (2004). herg1 gene and HERG1 protein are property of ERG K+ channels. Eur J Neurosci 11,
overexpressed in colorectal cancers and regulate cell invasion 753–760.
of tumor cells. Cancer Res 64, 606–611. Seasholtz TM, Majumdar M & Brown JH (1999). Rho as a
Lee TW, Anderson LA, Eidne KA & Milligan G (1995). mediator of G protein-coupled receptor signaling. Mol
Comparison of the signalling properties of the long and Pharmacol 55, 949–956.
short isoforms of the rat thyrotropin-releasing hormone Storey NM, O’Brian JP & Armstrong DL (2002). Rac and Rho
receptor following expression in Rat 1 fibroblasts. Biochem J mediate opposing hormonal regulation of the
310, 291–298. ether-a-go-go-related potassium channel. Curr Biol 12,
Niu J, Profirovic J, Pan H, Vaiskunaite R & Voyno-Yasenetskaya 27–33.
T (2003). G protein βγ subunits stimulate p114RhoGEF, a Thomas D, Kiehn J, Katus HA & Karle CA (2004). Adrenergic
guanine nucleotide exchange factor for RhoA and Rac1. regulation of the rapid component of the cardiac delayed
Regulation of cell shape and reactive oxygen species rectifier potassium current, IKr , and the underlying hERG
production. Circ Res 93, 848–856. ion channel. Basic Res Cardiol 99, 279–287.
Palomero T, Barros F, del Camino D, Viloria CG & de la Peña P Viloria C, Barros F, Giráldez T, Gómez-Varela D & de la Peña P
(1998). A G protein βγ dimer-mediated pathway (2000). Differential effects of amino-terminal distal and
contributes to mitogen-activated protein kinase activation by proximal domains in the regulation of human erg K+
thyrotropin-releasing hormone receptors in transfected channel gating. Biophys J 79, 231–246.
COS-7 cells. Mol Pharmacol 53, 613–622. Vogt S, Grosse R, Schultz G & Offermanns S (2003).
Paulssen RH, Paulssen EJ, Gautvik KM & Gordeladze JO Receptor-dependent RhoA activation in G12 /G13 -deficient
(1992). The thyroliberin receptor interacts directly with a cells. Genetic evidence for an involvement of Gq /G11 . J Biol
stimulatory guanine-nucleotide-binding protein in the Chem 278, 28743–28749.
activation of adenylyl cyclase in GH3 pituitary tumour cells. Weinsberg F, Bauer CK & Schwarz JR (1997). The class III
Evidence obtained by the use of antisense RNA inhibition antiarrhythmic agent E-4031 selectively blocks the
and immunoblocking of the stimulatory inactivating inward-rectifying potassium current in rat
guanine-nucleotide-binding protein. Eur J Biochem 204, anterior pituitary tumor cells (GH3 /B6 cells). Pflugers Arch
413–418. 434, 1–10.
Petrecca K, Atanasiu R, Akhavan A & Shrier A (1999). N-linked Yajima Y, Akita Y & Saito T (1988). Effects of cholera toxin on
glycosilation sites determine HERG channel surface the coupling of thyrotropin-releasing hormone to a guanine
membrane expression. J Physiol 515, nucleotide-binding protein in cultured GH3 cells. Mol
41–48. Pharmacol 33, 592–597.
C The Physiological Society 2005
C The Physiological Society 2005
FRET with multiply labeled HERG K+ channels as a reporter of the in vivo coarse
architecture of the cytoplasmic domains
Pablo Miranda a,1, Diego G. Manso a,1, Francisco Barros a,⁎, Luis Carretero a, Thomas E. Hughes b,
Carlos Alonso-Ron a, Pedro Domínguez a, Pilar de la Peña a
a
Departamento de Bioquímica y Biología Molecular, Edificio Santiago Gascón, Campus del Cristo, Universidad de Oviedo. E-33006. Oviedo, Asturias, Spain
b
Department of Cell Biology and Neuroscience, Montana State University, Bozeman, Montana 59717, USA
a r t i c l e i n f o a b s t r a c t
Article history: The intracellular N-terminus of human ether-a-go-go-related gene (HERG) potassium channels constitutes a
Received 29 November 2007 key determinant of activation and deactivation characteristics and is necessary for hormone-induced
Received in revised form 30 May 2008 modifications of gating properties. However, the general organization of the long amino and carboxy HERG
Accepted 2 June 2008
terminals remains unknown. In this study we performed fluorescence resonance energy transfer (FRET)
Available online 19 June 2008
microscopy with a library of fluorescent HERG fusion proteins obtained combining site-directed and trans-
Keywords:
poson-based random insertion of GFP variants into multiple sites of HERG. Determinations of FRET efficiencies
HERG with functional HERG channels labeled in different combinations localize the fluorophores, introduced in the
Potassium channel amino and carboxy ends, in two quadratic planes of 7.8 and 8.6 nm lateral size, showing a vertical separation of
Cytoplasmic domain architecture nearly 8 nm without major angular torsion between the planes. Similar analysis using labels at positions 345
FRET and 905 of the amino and carboxy terminals, located them slightly above the planes delimited by the amino and
carboxy end labels, respectively. Our data also indicate an almost vertical arrangement of the fluorophores
introduced in the NH2 and COOH ends and at position 905, but a near 45° angular rotation between the planes
delimited by these labels and the 345-located fluorophores. Systematic triangulation using interfluorophore
distances coming from multiply labeled channels provides an initial constraint on the overall in vivo arrange-
ment of the HERG cytoplasmic domains, suggesting that the C-linker/CNBD region of HERG hangs centrally
below the transmembrane core, with the initial portion of the amino terminus around its top and side surfaces
directed towards the gating machinery.
© 2008 Elsevier B.V. All rights reserved.
0167-4889/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.bbamcr.2008.06.009
1682 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
ever, this protein segment only forms monomers in crystal structures purified and ligated into HindIII/BstEII-digested wild-type HERG in the
[18], and channels truncated at residue 725 of the carboxy terminus are pSP64A+ vector. The purified HindIII/BamHI fragment obtained from
not able to assemble in spite of the presence of an intact EAG/PAS this construct was subsequently cloned into the pEYFP/ECFP-C1 vectors.
domain [11]. It has also been shown that HERG N-terminal truncations A similar strategy was used to generate C-terminally labeled HERG
that delete the EAG/PAS domain still allow the formation of functional channels using the pECFP-N1 and pEYFP-N1 vectors (Clontech). In this
channels in the plasma membrane [18–22]. case it was necessary to construct first a HERG cDNA with an EcoRI
Voltage sensors and gating elements have been mapped to the recognition site instead of the stop codon at the end of the coding
transmembrane region of voltage-gated potassium channels. How- sequence. A forward primer containing the coding sequence corres-
ever, the amino- and carboxy-terminal cytoplasmic domains also play ponding to amino acids 859–867 was used in PCR reactions with a
a key role in activation and deactivation gating of these channels (for reverse primer containing the coding sequence for amino acids 1151–
reviews see [7,23,24]). This can be particularly significant for HERG, 1159 and the EcoRI recognition site (5′-GGGAATTCCTGCCCGGGTCCG-
since it operates as an inward rectifier although it has the typical AGCCGTGTCTGTG-3′). The resulting PCR product was digested with
molecular arrangement of depolarization-activated channels [25]. SacI/EcoRI, purified and ligated into the corresponding sites of wild-
Such a kinetic behaviour is critical for normal cardiac repolarization type HERG. Finally the Hind III/EcoRI cDNA fragment from this
and setting electrical parameters in a variety of cells [reviewed in modified construct was cloned into the pECFP/EYFP-N1 vectors.
26–28]. As in other voltage-dependent K+ channels, voltage sensors The HERG channels with CFP or the venus variant of YFP (VFP) inserted
and gating elements in HERG are located within the core transmem- at different internal positions of the coding sequence were obtained by
brane region [29–34]. However, it is also known that the initial part of random insertion with an in vitro transposition system using a Tn5-
the HERG N-terminus determines its slow deactivation [18,20– derived hyperactive transposon carrying the fluorescent protein coding
22,35,36], and that the presence of a structurally intact proximal do- sequence for CFP or VFP as previously described [40,41]. These two
main, also in the amino terminus, is essential for maintenance of transposons (TcPT-1 encoding CFP and TvPT-0 encoding VFP) use different
normal activation properties and hormonal modulation of activation reading frames to increase the chance of insertion yielding functional
and deactivation parameters [21,37–39]. Therefore, unravelling the proteins [40]. Given the low probability of transposition in an in vitro
general structure of those HERG cytoplasmic regions associated with reaction, the transposons also include as selective marker a kanamycin
its gating is important for understanding how this channel controls resistance gene (Kanr) flanked by SrfI restriction sites that can be used to
fundamental physiological events such as cardiac rhythm and subsequently remove it. Two independent in vitro transposition reactions,
contraction, hormone secretion, and tumour cell proliferation [26–28]. one for each transposon, were carried out containing molar equivalents of
Given the importance of knowing the relative position and the the PCR-amplified transposon and the target plasmid (carrying the cDNA
structural organization of HERG intracellular structures, in this report of wild type HERG) with EZ::TN transposase (Epicentre Technologies).
we have tried to get some insight into the basic architecture of the After the transposition reaction and transformation, the transposed
amino and carboxy HERG terminals in vivo. For this purpose we clones expressing dual antibiotic resistance to ampicillin (transformed
generated a library of HERG fluorescent fusion proteins useful for clones) and kanamycin (marker for transposition) were transiently
fluorescence resonance energy transfer (FRET) microscopy studies, expressed in HEK293 cells to identify and select correctly orientated in-
combining the directed insertion of the labels in specific positions with frame insertions by following the appearance of fluorescent fusion
a random insertion of the cyan (CFP) and yellow (YFP) variants of the proteins. The Kanr cassette was further removed by digestion with SrfI
green fluorescent protein (GFP) into multiple sites by a transposon- and religation to obtain the full-length fusion constructs. Sequencing and
based technique [40,41]. Next we performed an electrophysiological transient expression of each fusion protein was used to confirm proper
screening of the library looking for correctly assembling constructs insertion of the fluorescent protein into the HERG coding sequence and
that might provide insights about the positioning and functional correct folding of the fluorophore regardless of the insertion site. Two
implications of some specific regions in the context of the whole collections of HERG fluorescent fusion proteins containing either CFP or
channel. Also, donor de-quenching following photobleaching of the VFP randomly inserted all along the channel coding sequence were thus
acceptor fluorophore was used for a quantitative determination of generated. Additionally, some of the fluorescent fusion channel constructs
FRET efficiencies [42–46] of several library constructs. Subsequent from each collection were subsequently used as a base for the generation
estimation of interfluorophore distances and the performance of of identical constructs carrying the complementary fluorophore. For this
“multiple molecular triangulation” with fluorophores located at va- purpose, PCR fragments containing the coding sequence of CFP and VFP
rious positions provided a useful initial constraint on the coarse fluorescent proteins were generated using as template the plasmids
structure of the channel tetramer, clarifying the relative arrangement containing both transposons [40]. After digestion with AscI and SrfI these
of some key landmarks on the cytoplasmic ends of HERG. PCR products were finally used to replace in the fluorescent fusion
channel constructs the corresponding AscI/SrfI fragment containing the
2. Materials and methods complementary fluorophore.
To generate HERG channels N and C-terminally double labeled with
2.1. CFP/YFP tagging of HERG channels both CFP and V/YFP fluorescent proteins, HindIII–BstEII restriction
fragments from channel fusion constructs carrying the fluorescent label
The original plasmid construct containing the cDNA of the HERG located in the amino end were used to replace the corresponding
channel was a gift of Dr. E. Wanke (University of Milan, Italy). To gene- segment in HERG fluorescent constructs having the complementary
rate HERG channels N-terminally labeled with the enhanced yellow fluorophore in the caboxy end. In other cases, BstEII–BamHI fragments
(EYFP) or cyan (ECFP) fluorescent protein (subsequently named YFP and from fluorescent fusion channels containing the fluorophore in the C-
CFP throughout the text) a Hind III/BamHI cDNA fragment containing terminal domain were used to replace the corresponding wild type
the entire coding region from HERG was cloned into pEYFP-C1 and fragment in constructs carrying the complementary fluorophore in the
pECFP-C1 vectors (Clontech) in-frame with the fluorescent protein N-terminal region.
sequence. For this purpose a forward primer containing a HindIII
recognition site and the coding sequence corresponding to HERG amino 2.2. Cell culture and transfection
acids 1–7 (5′-GGAAGCTTTCATGCCGGTGCGGAGGGGCCAC-3′) was used
in PCR reactions with a reverse primer containing the coding sequence Human embryonic kidney (HEK293) and Chinese hamster ovary
corresponding to amino acids 370–378 and covering the unique BstEII (CHO) cells were grown at 37 °C in a humidified atmosphere of 95% air
site. The resulting PCR product was digested with HindIII/BstEII, gel and 5% CO2 and plated in 35-mm diameter tissue culture plastic dishes
P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699 1683
containing sterile glass coverslips coated with poly-L-lysine, that activation was monitored using an indirect envelope of tail current
constituted the bottom of a recording chamber for electrophysiological protocol [21,38,47], varying the duration of the depolarization pulse
and FRET measurements. The culture medium consisted of Dulbecco's and following the variation in the magnitude of the tail currents
modified Eagle's medium nutrient mixture F-12 Ham's (DME/F12 1:1 recorded after going back to a negative voltage. The time necessary to
mixture, Sigma) supplemented with 100 U/ml penicillin, 0.1 mg/ml reach a half-maximal tail current magnitude was used to compare the
streptomycin and 10% fetal bovine serum. Cells trypsinized 24 h prior speed of activation of the different channels. Deactivation parameters
to transfection seeded on poly-L-lysine-coated coverslips were were obtained from tail currents upon membrane repolarization at
transiently transfected using Lipofectamine 2000 (Invitrogen) with the indicated potentials, following 1 s depolarization pulses to +40 mV
2–5 µg of plasmid DNA containing either the individual constructs or that displaced the channels to open (and inactive) states. The time
the co-transfected CFP- and V/YFP-labeled constructs in an approxi- constants of deactivation were determined from negative-amplitude
mately 1:3 relationship to maintain a proper molar ratio of expressed biexponential fits to the decaying phase of the tail currents using a
donor and acceptor fluorophores, ensuring that most of the cells function of the form:
contained at least one acceptor-labeled subunit per donor molecule.
The mixture of DNA and Lipofectamine was incubated in serum-free
Y ¼ Af exp −T=τf þ As expð−T=τ s Þ þ C ð2Þ
medium for 20 min and added to the plates with the cells in serum-
containing medium without antibiotics. Recordings were typically
performed 24–48 h after transfection. Electrophysiological character- where τf and τs are the time constants of fast and slow components,
ization of the constructs was systematically performed in transfected Af and As are the relative amplitudes of these components, and C is a
HEK293 cells because they are easier to patch and remain more stable constant. In this case, the first cursor of the fitting window was
during recordings. As previously reported [47] additional non-HERG advanced to the end of the initial hook due to the recovery of
outward currents are frequently detected in these cells during inactivation.
depolarizing steps, in which the magnitude and the degree of
inactivation and/or rectification of the elicited currents are highly 2.4. FRET measurements
variable. However, the contribution of the endogenous currents is
negligible along the tail current time course. Therefore, by limiting the FRET measurements were performed in CHO cells bathed in
analysis to the tail currents, only HERG current characteristics would standard extracellular saline following the increase (dequenching) in
be analysed. Since analogous current parameters were obtained when donor (CFP) fluorescence signal during incremental photobleaching
CHO cells were used, data from both cell types were pooled in the of acceptor (V/YFP). Background fluorescence of both fluorophores
Tables. Due to their higher transfection efficiency providing an easier was determined from a region of the same field located outside the
co-localization of several cells in the microscopy field, transfected CHO cell and the averaged pixel intensity of this area was subtracted from
cells were typically used for FRET measurements. the mean intensities measured in the regions of interest correspond-
ing to the individual fluorescent cell areas. A Zeiss Axiovert 100
2.3. Acquisition and analysis of ionic current data microscope with a 40×/0.75 Plan-NeoFluar objective (Carl Zeiss) was
used. The fluorescence imaging system (Till-Photonics, Gräfelfing,
Ionic current recordings in HEK293 and CHO cells were performed at Germany) consisted of a monochromator Polychrome IV, a dual band
room temperature in the whole-cell configuration of the patch-clamp CFP/YFP filter set and a 12-bit CCD camera (IMAGO) combined with
technique. To improve the quality of the seals and the stability of the a DUAL-View Micro-Imager (Optical Insights, Santa Fe, USA) for dual
recordings, cells were sometimes mildly trypsinized and seeded on poly- emission image acquisition. Control of monochromator and camera
L-lysine-coated coverslips to be used for analysis within 2–6 h. No change as well as image recording and processing were performed with
in the qualitative or quantitative characteristics of the cell currents was TILLvisION software (Till-Photonics). The standard acceptor bleach
introduced by this procedure. Patch pipettes with tip resistances of 3– protocol consisted of 30 cycles every 2 s with 50–100 ms of
5 MΩ were pulled from borosilicate or Kimax disposable micropipettes exposure of donor and acceptor near their excitation maximum (at
(Boralex, Rochester Scientific, NY; Fisherbrand, Fisher Scientific, Pitts- 440 and 515 nm, respectively) to detect CFP and V/YFP fluorescence
burg, PA or Kimble glass Inc. Vineland, NJ). The standard extracellular without V/YFP-bleaching, followed by 180–240 additional cycles
saline contained (in mM): 137 NaCl, 4 KCl, 1.8 CaCl2, 1 MgCl2, 10 glucose, separated by periods in which the V/YFP remains permanently
and 10 HEPES (pH 7.4 with NaOH). The pipette solution contained (in illuminated at 515 nm to photobleach it. Linear regression was used
mM): 140 KCl, 2 MgCl2, 0.7 CaCl2, 1.1 EGTA, and 10 HEPES (pH 7.4 with to obtain the steady-state values of FCFP increase from scatter plots
KOH). An EPC-7 patch-clamp amplifier (HEKA Elektronic, Lambrecht, of CFP fluorescence increase versus FYFP decrease with incremental
Germany) was used to record membrane currents. Stimulation and data bleaching for each cell. Photoconversion of V/YFP into CFP-like
acquisition were controlled with the Pulse+ PulseFit software (HEKA species was negligible under our photobleaching conditions as
Elektronic) running on Macintosh computers. Ionic currents sampled at evidenced by the total absence of fluorescence in the CFP channel
1 kHz were elicited using the voltage protocols indicated in the figures. when cells expressing exclusively V/YFP-labeled proteins were
Data analysis and exponential fits to ionic currents were performed with submitted to the standard acceptor bleaching protocol. Individual
the programs PulseFit and Igor-Pro (WaveMetrics Inc., Lake Oswego, OR, cells showing dim fluorescence leading to very low signal-to-noise
USA). A P/n method was used for leak and capacitive current subtraction. ratios and those in which the fluorescence of V/YFP did not average
Kinetic parameters of activation and deactivation were obtained as at least 13 or 10 fold that of the lower intensity CFP probe,
previously described [21,38,47]. When tail current analysis was used for respectively (see below), were discarded to prevent low acceptor-to-
studying activation voltage dependence, tail current magnitudes donor ratio contributing to uncontrolled and reduced FRET signals.
normalized to maximum following 1 s depolarizations were fitted with This implies that almost exclusively 2:2 and/or 1:3 CFP:V/YFP-tagged
a Boltzmann function to estimate the half-activation voltage (Vhalf) and subunit stoichiometries will contribute to FRET in the HERG tetra-
equivalent gating charge (zg): meric structure. Quantitative FRET levels were expressed as FRET
efficiency (EFRET), defined as the proportion of the excited states of
Itail =Imax ¼ 1= 1 þ exp ðVhalf −V ÞZg F=RT ð1Þ the donor that become transferred to the acceptor. As previously
pointed out [43,46,48,49] the true or “maximal” FRET efficiency can
where V is the test potential and F, R and T are Faraday constant, gas be obtained from donor dequenching fluorescence data as the
constant and absolute temperature, respectively. The time course of product of the measured or “apparent” efficiency and αD, the
1684 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
fraction of total donor molecules forming donor-acceptor complexes, Pythagorean theorem, the interfluorophore distance for two diag-
according to the expressions: onally located labels in the same plane will correspond to 1.4142d in
which d represents the adjacent distance between fluorophores
EFRET ¼ EFRET ðAPPÞ α D ¼ ½1−ðFDA =FD Þα D ð3Þ arranged in a quadratic symmetrical way and 1.4142 stands for the
square root of 2. Equivalent triangulations using the Pythagorean
in which FDA and FD are the fluorescence intensities of the donor theorem were performed to calculate interfluorophore distances as a
before and after complete photobleaching of the acceptor, respec- function of the vertical separation between the planes delimited by
tively, and αD would correspond to DA/(D + DA) or the fraction of labels introduced in two different positions of the channel sequence.
donor-acceptor complexes (DA) with respect to total donor molecules
[free (D) plus complexed (DA) donors]. In our case, the contribution of 2.5. Statistics
the αD factor was minimized i) systematically converting the
fluorescence intensities into molar ratios by normalizing the intensity All data are presented as the mean ± S.E.M. for n number of cells
ratios to those of a donor-acceptor tandem imaged at the same and the number of transfections indicated by N. Statistical signi-
settings, and excluding from the analysis those cells showing a low ficance was tested with the unpaired two-tailed Student's t-test.
acceptor to donor ratio, and ii) by systematic subtraction of signals Extrapolation of EFRET deviations to distance error values was per-
coming from donor molecules co-expressed with non-interacting formed applying the propagation of errors method to the distance
acceptor pairs. Measuring donor fluorescence changes as a function of estimations using the aforementioned equations and the experi-
time during the bleaching process and extrapolation of fluorescence mentally determined EFRET. The uncertainty calculator (http://www.
intensity to zero acceptor level [45,50] also eliminated the influence of takesomeshortcuts.com/sigdig/sigdigs.shtml) was used for this
incomplete acceptor bleaching [49] in donor fluorescence increases. purpose.
To estimate distances from the intrasubunit FRET measurements
the conventional formulas 3. Results
EFRET ¼ Ro6 =ðRo6 þ d6 Þ ¼ ðRo=dÞ6 =1 þ ðRo=dÞ6 ð4Þ 3.1. Generation and functional expression of libraries of fluorescent HERG
fusion proteins with cyan and yellow variants of GFP
and
To obtain a number of constructs carrying fluorescent proteins
d ¼ RoðEFRET1 1Þ1=6 ð5Þ
inserted into multiple sites of the HERG channel protein large enough
were used in which d corresponds to the interfluorophore distance to support our experimental approach, we used a transposon-based
and Ro (the Förster distance) is the distance at which the FRET technique in which the sequence of either the cyan fluorescent
efficiency (EFRET) is 50%. Ro was taken as 5 nm for the CFP:V/YFP pair protein (CFP) or the venus variant of the yellow fluorescent protein
[48,51,52] assuming an orientation factor of 2/3 corresponding to a (VFP) contained in a modified hyperactive Tn5 transposon were
random interfluorophore orientation. Unless otherwise indicated the randomly inserted into the HERG coding sequence by in vitro
distance estimates from the FRET efficiencies were interpreted in a transposition reactions ([40,41]; see also Materials and methods for
two-dimensional way by assuming that the fluorophores located at details). When these transposons were inserted into the HERG coding
identical positions in different subunits were located in an x–y plane sequence in the correct orientation and reading frame they produced
parallel to the cell membrane due to maintenance of a fourfold fluorescent channel fusion proteins. Following in vitro transposition
symmetry of the channels around the central pore. reactions, transformation and selection of successfully transposed
Due to the several geometries adopted by the fluorophores in the clones, we initially screened 1728 clones (864 CFP and 864 VFP
tetrameric channel structures, two general situations were considered transposed recombinants) for fluorescence by transient expression in
for the EFRET expressions as a function of the donors and acceptors HEK293 cells. From the 189 constructs that produced fluorescent
present in a single oligomeric molecule [53,54]. Thus, for the case of a fusion proteins, 68 recombinants were discarded due to insertions
donor surrounded by several acceptors at distances d1 to dn, the out of the HERG sequence or being inversely oriented or out-of-
averaged EFRET is given by: frame. This yielded 121 correct insertions (7% of the transposed
clones) from which 44 additional recombinants were also discarded
EFRET ¼ Ro6 1=d61 þ 1=d62 þ :::: þ 1=d6n =1 due to duplication of the insertion position. Thus, we finally obtained
6
þ Ro 1=d61 þ 1=d62 þ N þ 1=d6n ð6Þ two collections of HERG fusion proteins with 37 unique insertions of
CFP and 40 of VFP (Fig. 1A and Table 1). In our system, taking into
also meaning that for n equivalent acceptors at the same distance, an account the size of the HERG cDNA target sequence (3500 bp) and the
operational nRo6 factor can be used. On the other hand, for several GW1-CMV plasmid vector carrying it (8650 bp), a random transposi-
donors surrounding one acceptor, the averaged EFRET is the average of tion would be expected to result in an in-frame correctly orientated
the transfer efficiency originating from each donor: insertion into the HERG coding region with a probability of 6.7%. This
percentage value compares favourably with the 7% of correct inser-
EFRET ¼ ½EFRET ðD1 Þ þ EFRET ðD2 Þ þ N þ EFRET ðDn Þ=n ð7Þ
tions obtained. Nevertheless, it is important to note that once the
or duplicated insertions are discarded, only 4.4% of the recovered fusion
proteins correspond to unique insertions. This is consistent with the
not entirely random behaviour of the Tn5 transposon, showing a
EFRET ¼ 1=n½ððRo=d1 Þ6 =1 þ ðRo=d1 Þ6 Þ þ ððRo=d2 Þ6 =1Þ
preference for some particular locations that tends to limit the
þ ðRo=d2 Þ6 þ N þ ððRo=dn Þ6 =1 þ ðRo=dn Þ6 Þ ð8Þ number of unique insertions recovered within a given target
sequence [40]. It is important to note also that the transposition
Application of these mathematical expressions to different process involves a duplication of the 9 bp flanking the transposon
donor:acceptor combinations and determination of the correspond- insertion site. This generates a three amino acid duplication that adds
ing EFRET vs. distance curves are detailed in the figures. Equivalent to the linker flanking the fluorescent protein [40]. Therefore we
expressions and procedures for the estimation of vertical distances identified the insertion site in the HERG coding region by the number
between the square planes delimited by fluorophores located in two of the amino acid residue preceding the inserted amino acid triplet.
different positions are also shown in the figures. According to the Interestingly, the 77 constructs carrying in-frame CFP or VFP protein
P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699 1685
Fig. 1. Schematic representation of a HERG channel subunit showing the position of fluorophore insertions (A) and differences in cellular localization of HERG and TRH-R fluorescent
fusion proteins in living cells (B). (A) Localization of CFP or V/YFP insertions into the different regions of the HERG subunit. Refer to Table 1 for the exact position in the sequence where
the insertions occurred and the type of fluorophore used for labeling. Different regions/domains in the channel molecule are highlighted and signaled with numbers, corresponding to
the amino terminal EAG/PAS and proximal domains (residues 1–135 and 136–397, respectively), the central channel core (residues 397 to 672) including S1 to S6 transmembrane
helices and the pore region between S5 and S6, and the CNBD (residues 764–836) in the carboxy terminus that extends up to residue 1159 [21,25]. Black circles, functional constructs;
white circles, non functional constructs; gray circles, not tested in the functional screening. The transposon-derived insertions and those introduced in the amino and carboxy ends by
conventional cloning procedures are shown. The five positions subsequently used for FRET measurements are marked with arrows. (B) HEK293 cells were transiently transfected with
expression plasmids encoding HERG channels labeled with YFP in the amino terminus and TRH receptors fused at the carboxy end to GFP. Cells were imaged by standard bright field
(top) and epifluorescence (middle) microscopy. Fluorescence intensity profiles along the depicted white line are shown at the bottom. Calibration bars: 10 µm. A.u.: arbitrary units.
labels at different positions along the HERG coding sequence yielded the HERG coding sequence (see also Fig. 1A). Five and seventeen
in all cases fluorescent fusion proteins indicating that, as previously insertions corresponded to the EAG and the proximal domains,
reported, GFP variants are able to fold properly and form a respectively, with four additional insertions located up to the
fluorophore after insertion in a variety of protein backgrounds beginning of the first transmembrane helix around residue 400
[40,41]. The distribution of the insertions shown in Table 1 (in which [25]. This means a total of 26 insertions (33.7%) covering the initial
the eight labeled variants obtained by conventional fluorescent 400 residues (34.5%) of the channel protein that correspond to the
labeling are also listed), indicates that they are spread over most of amino terminal region. Fourteen constructs (18.1%) had insertions in
1686 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
Fig. 2. Functional characteristics of HERG channels labeled with CFP and V/YFP in the amino and carboxy terminal ends and quantification of FRET signals by acceptor photobleaching.
(A) Comparison of WT and fluorescent HERG functional characteristics in transfected HEK293 cells expressing channels with or without the fluorescent protein fused to the amino
(YFP/NH2-HERG) or the carboxy (YFP/COOH-HERG) ends, either before residue 1 or following residue 1159. Representative ionic currents in response to the indicated voltage protocol
are shown on the left. Fractional activation curves obtained by standard tail current analysis in the same cells are shown on the right. I vs. V curves normalized to maximum for
currents at the end of the depolarization step generated from cells expressing WT and YFP/NH2-HERG channels are shown in the inset. (B) Determination of donor fluorescence
recovery during disruption of energy transfer by selective acceptor photobleaching. Transiently transfected CHO cells were imaged for co-expressed HERG amino terminally fused to
CFP and YFP as indicated in Methods. Shown are the relative increases in CFP intensities (in percentages) over initial levels and the remaining fluorescence intensity of YFP. The
symbols represent the calculated mean ± SEM, whereas the thin lines correspond to single cell traces. The acceptor was bleached as indicated by the box. A linear regression analysis of
donor recovery vs. fractional acceptor photobleach is depicted in the inset.
positions 580, 596, 598 and 603, all located in the linker between S5 3.2. Implementation of a FRET system to detect intersubunit interactions
and the pore loop. On the other hand, a complete lack of functional between functional HERG channels fluorescently labeled in the amino
expression was also obtained for all tested fusions carrying the labels and carboxy termini
in the segment linking the sixth transmembrane helix and the CNBD
or in the CNBD domain itself. However, prominent HERG-type The relative positions of the HERG N- and C-terminal structures and
currents were observed in 10 out of 13 tested positions carrying their relationship to the channel core are unknown. It has been
insertions in the C-terminal segment that extends from amino acid proposed that interactions between the initial part of the EAG domain
905 to the carboxy end corresponding to residue 1159. and the S4–S5 linker are involved in slowing HERG closing, and that
1688 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
this linker could act as a docking site for the N-terminal region to FRET efficiency of 13.53 ± 1.16 (see Methods). These data demonstrate
modulate activation gating [18,22,29,31,35]. According to this idea and the existence of a substantial FRET between the N-labeled subunits of
due to the short length of the HERG S4–S5 linker, we reasoned that HERG, suggesting a relatively close packing of their amino terminal
(unless the four-fold symmetrical organization is not maintained), the structures in the tetramer.
initial segments of the four amino termini should lie relatively close To demonstrate that the recorded FRET signals are due to properly
and packed below the central transmembrane pore structure. There- assembled HERG complexes and not to unordered aggregates formed in
fore, we started generating and characterizing N-terminal fusion cons- defined cell compartments or to random encounters between diffusing
tructs of HERG with CFP and YFP that could help us to implement a overexpressed channel subunits, we performed some additional con-
FRET system useful in assessing the predicted proximity of the co- trols. Although the detection of robust HERG currents unequivocally
expressed subunits. Additionally, fluorescent channels labeled at the demonstrates the presence of properly assembled fusion proteins in the
C-terminus were also constructed for comparison. plasma membrane, the diffuse fluorescence pattern consistently
Labeling N- and C-termini with CFP and YFP yielded totally func- observed in HERG-expressing cells (see above), made it necessary to
tional fusion channels as demonstrated by the appearance of typical ensure that the detected FRET signals were not mainly due to miss-
HERG-like currents in HEK293-(Fig. 2A) and CHO-(not shown) assembled channels retained in intracellular structures. As shown in Fig.
transfected cells. These currents showed similar activation voltage 3A for an illustrative construct, the averaged FRET efficiency was
dependence and inward rectification properties as those of unlabeled indistinguishable when the determination was performed using the
channels (Fig. 2A and Table 2). However, the N-terminal fluorescent total cell area, the central part of the cells, or only the region corres-
channels showed faster deactivation kinetics. Thus, the values of deac- ponding to the outer border, mostly representing the plasma membrane.
tivation time constants for the major fast deactivation current com- Statistically equivalent FRET values were observed when data derived
ponent upon repolarization at −50 and −100 mV were significantly from whole cell images were compared with those of the same
reduced in the presence of the amino terminal fluorophore, but not in construct, using the fluorescence signals detected at the tip of patch
the constructs carrying the fluorescent protein at the end of the HERG pipettes carrying membrane patches excised from the fluorescent
sequence at positions 1158 or 1159 (Table 2). The behaviour of the N- HERG-expressing cells (Fig. 3B).
terminally labeled channels is equivalent to that obtained after intro- As a further demonstration of FRET coming from specifically
duction of a short HA epitope at the beginning of the amino terminus assembled protein complexes, we always verified that the FRET effi-
[39] and mimics the accelerations of deactivation caused by impair- ciency did not significantly vary with different CFP or YFP fluorescence
ment of the EAG/PAS domain due to deletion of the initial sixteen intensities, and that it was independent of the channel density and/or
residues [22,35] or the complete domain (Δ2–135 channels, ref. [18]). the expression levels over a wide range (Fig. 3C). Only experiments and
Altogether, these results indicate that the labeled subunits reliably channel constructs showing a similar absence of correlation between
reproduce the normal assembly and function of the unlabeled HERG measured EFRET and fluorochrome expression levels were subsequently
channels. considered for analysis. Finally, as shown in Fig. 3D, the FRET efficiency
To detect and quantify FRET signals, we measured the increase in became much lower than that obtained with N-terminally labeled
donor (CFP) emission during selective photobleaching of the acceptor pairs after using two presumably non-interacting proteins, namely
fluorophore (YFP) under static conditions. Thus the recovery of donor HERG channels and the rat vanilloid receptor VR1, both labeled in the
fluorescence emission (a measure of steady-state FRET efficiency) was amino terminus [3.78 ± 0.45 (n = 76) vs. 13.53 ± 1.16 (n = 15); P b 0.0001],
monitored and expressed as the percentage of CFP emission during the or the TRH receptor and HERG also labeled at the amino end (not
acceptor bleach. As shown in Fig. 2B, direct recording of CFP and YFP shown). This indicates that FRET is only observed between labeled
fluorescence signals from N-terminally labeled HERG constructs at subunits assembled in one channel but not between merely co-loca-
successive time points after the onset of YFP photobleaching, reveals lized subunits. It also defines a lower limit of EFRET below 4 as an
an increased CFP fluorescence emission concomitant with progressive indication of FRET not coming from specific homotetrameric assembly
YFP bleaching, which is direct evidence for FRET. Plotting normalized of HERG channel subunits.
CFP fluorescence against that of YFP also indicated that the extent of Next, we assessed FRET efficiency upon the expression of C-
CFP recovery linearly correlated with decaying YFP fluorescence (inset terminally labeled HERG constructs either alone or in combination
of Fig. 2B). As an indication of FRET efficiency, the extrapolated with N-terminal fusion constructs. Currents analogous to those
intersection of the linear regression line with the y axis at FYFP = 0 mediated by native HERG channels were obtained in cells expressing
yielded an FCFP increase of 15.64 ± 1.34% (n = 15) that corresponds to a fluorescent channels labeled at the C-terminus either at position 1159
Table 2
Electrophysiological characteristics of the fluorescent HERG fusion proteins
Label position and/or transfection V0.5 (mV) ZG Activation t0.5 (ms) Deactivation τ (MS) n
Fig. 3. Verification of FRET signals specificity between assembled tetramer subunits of amino terminus-labeled HERG channels. (A) Comparison of whole-cell (top a) FRET levels and
those determined separately either in a defined region over the plasma membrane (bottom a) or in the rest of the cell body. Calibration bar: 10 µm. The fluorometric determinations
were performed on regions of interest as indicated in a, defined over the whole-cell area of the transfected CHO cells (b,c left), the outer border mostly representing the plasma
membrane (b,c middle) or only the cell center excluding the outer border (b,c right) of the same cells. The donor fluorescence recovery during selective photobleaching of the
acceptor was determined as described in Fig. 2B. Quantitative EFRET levels obtained as indicated in Methods for the indicated areas are compared in panel c. (B) Comparison of FRET
levels in the whole-cell and in membrane patches excised from cells expressing the same HERG protein construct. a, microphotographs of the imaged cells and excised membrane
patches under epifluorescence (left) or bright-field (right) illumination. Calibration bar: 10 µm. The white circle indicates the area from which fluorescence recordings were obtained
upon patch excision. b, time-course of donor and acceptor fluorescence variation in excised patches submitted to acceptor photobleaching. Averaged EFRET levels from four
independent measurements are indicated. c, comparison of EFRET in excised patches and whole-cell recordings from cells expressing the same fusion construct. (C) Independence of
EFRET on fluorescent protein expression levels. Single cell data plotted against the expression level of YFP-HERG are shown. The results of a linear regression to the data are overlaid.
(D) Comparison of FRET levels between the N-terminally-labeled CFP-HERG/YFP-HERG pair and YFP-HERG co-expressed with VR1 labeled with CFP in the amino terminus. Linear
regression analysis of donor recovery vs. fractional acceptor photobleach for the CFP-HERG/YFP-HERG and CFP-VR1/YFP-HERG pairs is shown in the central panel. The donor
fluorescence recovery during selective photobleaching of the acceptor for the pair CFP-VR1/YFP-HERG is shown in the inset. n.s.: not significant vs. whole-cell.
(see above) or after residues 1123 or 1158 (not shown). However, only a or 1159) were co-expressed with subunits labeled with CFP in the same
basal FRET signal was observed following co-expression of N-terminally positions [EFRET 7.52± 1.48 (n = 6) and 9.37 ± 0.48 (n = 31), respectively;
CFP-labeled channels with those carrying V/YFP fluorophores at Fig. 4]. Similar FRET efficiencies were measured in cells co-expressing
positions 1158 or 1159 (EFRET 3.07 ± 0.68, n = 16 and 3.06 ± 0.5, n = 19; fluorescent channels with CFP and VFP both at positions 1123 (9.9 ± 0.47,
Fig. 4). These data further support the specificity of the FRET detected n = 21) and when channel subunits labeled at residue 1123 were co-
between the amino terminals. Interestingly, FRET signals significantly expressed with those labeled at residue 1158 (8.32 ± 0.82, n = 7). These
elevated above the control HERG/VR1 pair were obtained when C- results and the functional integrity of the fluorescent constructs suggest
terminally V/YFP-labeled HERG (with the fluorophore at residues 1158 that failure to detect significant FRET signals between channel subunits
1690 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
Fig. 4. FRET efficiencies between HERG subunits labeled in the amino and carboxy termini. Different combinations of HERG channels labeled with CFP or Y/VFP at residues indicated
on the abscissa were co-expressed in CHO cells and FRET efficiencies were quantified as described in Methods. Each bar represents the mean ± SEM of 2–5 co-transfection experiments
for the indicated number of cells. FRET levels following co-expression of the putatively non-interacting VR1 and HERG proteins are shown for comparison. Combinations showing a
significantly increased (P b 0.05) FRET signal as compared with the VR1/HERG pair are marked with an asterisk. N.s.: not significant vs. VR1/HERG. A schematic representation of a
HERG channel subunit topology showing the position of the fluorescent labels (black circles) is shown in the inset. Transmembrane helices S1–S6, the position of the pore region
between S5 and S6 and the situation of the EAG and CNBD domains in the amino- and carboxy-terminals, respectively, are highlighted. No significant differences are present between
the EFRET values of 1158/1158, 1159/1159, 1123/1123 and 1123/1158 pairs.
labeled at residues 1 and 1158 or 1159 is not due to the inability of C- in the channel tetramer as compared with those separating the amino
terminally labeled subunits to assemble properly as a result of terminal ends.
disruption of a putative tetramerization domain located in the carboxy
terminus. It can be argued that even though C-terminally labeled 3.3. Estimations of intramolecular distances between fluorophores
homotetrameric channels are readily assembled, the low FRET levels introduced in the amino and carboxy ends
observed in cells co-expressing N- and C-terminally labeled subunits
are due to lack of heterotetrameric assemblies. Two lines of evidence In a tetrameric structure, both adjacent and diagonal distances
indicate that heteromers are indeed formed in these cells. First, robust between the fluorophores will contribute to the distance estimated
HERG-like currents with electrophysiological characteristics unlike from FRET measurements. As indicated in the Methods section, almost
those observed in cells expressing channels only labeled in the amino or exclusively 2:2 and/or 1:3 CFP:YFP-labeled subunit stoichiometries are
the carboxy termini are observed in the co-transfected cells. As shown expected to produce FRET under our experimental conditions. As
in Table 2, the activation voltage dependence and kinetics of these schematically illustrated in Fig. 5A, in a 2:2 stoichiometry two
currents do not correspond to those expected from a simple mixture of equivalent arrangements of donor and acceptor labels are produced
amino- and carboxy-labeled channels, strongly arguing in favor of the in which a given donor is surrounded by an adjacent- and a diagonally-
presence of additional subunit combinations. Second, properly located (2:2a combination) or two adjacent (2:2b) acceptors. In a 1:3
assembled functional channels are obtained with double-labeled stoichiometry the energy transfer of the donor will be directed towards
constructs carrying fluorophores attached to both the amino and the two adjacent and a diagonal acceptor. Assuming a Ro value of about
carboxy termini, demonstrating that up to eight fluorescent proteins 5 nm see (Methods) and since the same residues in the different
can be introduced in these positions without impairing channel subunits of the tetramer must be located in an x–y plane parallel to the
function (see below). Interestingly, the lower EFRET values in cells cell membrane maintaining a symmetrical structure with respect to
expressing carboxy-labeled constructs as compared to those obtained the central axis of the channel, the calculation of FRET efficiencies as a
upon expression of subunits labeled in the amino terminus, also function of the interfluorophore distances for adjacent CFP/YFP pairs
indicate that although the C-terminal labels of the channel subunits are arranged in a 2:2 or 1:3 tetrameric structure yields the curves shown in
separated by a distance yielding a detectable FRET signal, such a Fig. 5B. Considering our FRET measurements once the non-specific
distance is likely higher than that between the amino termini labels. FRET (3.78 of the HERG/VR1 pair) is subtracted and using the 2:2
Moreover, the detection of similar FRET values with the 1123/1123, arrangement as a standard, the specific EFRET value obtained with the
1158/1158, 1159/1159 and 1123/1158 pairs suggests that the smaller co-expressed CFP/YFP amino terminal constructs (9.75 ± 1.61) would
FRET signals between C-terminally labeled subunits are not only due to mean a separation of 7.76 ± 0.24 nm between the labels in adjacent
variations in the relative orientation of the fluorophores, but they subunits (with a ≈11 nm diagonal separation), when the fluorophores
reflect increased distances between the carboxy termini of the subunits are located in the amino termini of the four HERG channel subunits
P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699 1691
Fig. 5. Determination of intramolecular distances between labels introduced in identical positions of the four subunits assembled in a HERG channel tetramer. (A) Diagrammatic
representations of possible energy transfer pathways for labeled subunit tetramers in a fourfold symmetrical arrangement and 2:2 or 1:3 stoichiometry. Mathematical expressions for
EFRET calculation as a function of the distance “d” between adjacent fluorophores for a single pair (1:1) and 2:2 or 1:3 stoichiometries are shown. The product of d times square root of
2 (1.4142d) corresponds to the diagonal distance according to the Pythagoream theorem. Donor and acceptor fluorophores are represented by solid and open circles, respectively. (B)
Dependence of EFRET on adjacent distance between fluorophores in individual tetramer subunits. Calibration curves were obtained using the expressions indicated in A and
correspond to a 1:1 single pair (dashed line), 2:2a and 2:2b combinations (lower and upper dotted lines, respectively), the averaged EFRET in the 2:2 stoichiometry (lower solid line)
and the EFRET in the 1:3 arrangement (upper solid line). Extrapolated distance values for EFRET obtained with HERG channels labeled in the amino terminal end or following residues
345, 905 or 1159, are signalled with arrows. Net or specific EFRET values after subtraction of unspecific FRET obtained with the VR1/HERG pair were used. For more explanations see
text. (C) Schematic representations of interfluorophore distances corresponding to specific EFRET values obtained from expressed HERG channels labeled with the fluorophores as
indicated.
(Fig. 5C). Performance of an equivalent analysis using the FRET value However, since in this case the donor and acceptor labels are in
for C-terminal labeled channel subunits (specific EFRET 5.59 ± 0.93) different subunits and they can be located at different distances from
suggests that the carboxy terminal labels would be placed at a distance the membrane (z), the number of possible pathways for energy
of 8.58 ± 0.26 nm between adjacent subunits (corresponding to ≈12 nm transference is increased due to positioning of the labels in two planes
of diagonal separation). for which torsion angles are unknown. Due to the high sensitivity of
The localization of the four amino and carboxy end labels in two FRET to distance, it could be possible that even though a relatively short
square planes of ≈7.8 and ≈8.6 nm lateral sizes and the absence of vertical distance between both planes could exist, only a negligible
specific FRET between pairs of constructs labeled in the amino and the EFRET is observed due to the diagonal separation between fluorophores
carboxy termini, excludes the possibility that the N-termini are copla- in different subunits. To solve this uncertainty, we generated a double-
nar with the carboxy ends. This lack of FRET signals could be consi- labeled HERG construct carrying the donor and acceptor fluorophores
dered as an indication that a considerable distance exists between the in the amino and carboxy ends of the same subunit. Since this construct
planes limited by the amino- and the carboxy-located fluorophores. has an inherent 1:1 CFP:YFP stoichiometry, it was also used to calibrate
1692 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
Fig. 7. Functional expression and FRET measurements with HERG channels labeled at residue 345 and in different combinations. (A) Representative ionic currents from transiently
transfected cells expressing HERG channels labeled with GFP variants at residues 379, 345 or either both 1 and 345 (Double 1/345) or 345 and 1158 (Double 345/1158). 1 s
depolarization pulses up to + 60 mV in 20 mV increments from a holding voltage of −80 mV were delivered to the cells, followed by a 2 s repolarization period at −50 mV. Note the
complete absence of HERG-like currents using the 379-labeled construct and the expanded time scale and the fast deactivating tail currents for the double 1/345 construct. (B)
Determination of donor fluorescence recovery during acceptor photobleaching in cells either co-expressing HERG channels carrying CFP and VFP at residues 345/345, 1/345 and
345/1159, or transfected with double-labeled subunits with donor and acceptor fluorophores at residues 1 and 345 or 345 and 1158. Relative increases in CFP fluorescence over
initial levels and the remaining fluorescence intensity of YFP were obtained from transiently transfected CHO cells as detailed in the legend of Fig. 2. The symbols represent the
calculated mean ± SEM, whereas the thin lines correspond to single cell traces. (C) Comparison of EFRET for channels labeled as indicated in panel B. FRET levels from the HERG/VR1
pair and from co-expressed channels labeled with CFP and YFP in the amino end are also shown for comparison. Bars represent the mean ± SEM of 2–5 transfections for the indicated
number of cells. Data corresponding to co-transfections with 4 µg DNA of the double-labeled constructs plus 4 µg of the plasmid containing wild-type HERG are also shown.
Patch-clamp analysis of all these constructs indicated that they are plasma membrane, it effectively constrains the ability of this domain to
able to form functional channels in the plasma membrane, as de- slow down HERG channel closing.
monstrated by the presence of HERG-like currents in transfected The results of FRET measurements using channel subunits labeled at
HEK293 and/or CHO cells (Figs. 7A and 8A). Interestingly, whereas slow position 345 combined with those labeled in the amino and carboxy
deactivating currents with similar kinetics as those of unlabeled HERG termini in several combinations are summarized in Fig. 7B,C. It can be
were observed in cells expressing 345, 905 and double-labeled 345/ observed that a substantial amount of specific FRET is exhibited by cells
1158 constructs, very fast decaying tail currents at −50 mV were co-expressing HERG subunits labeled with CFP and VFP both at positions
observed in cells transfected with double-labeled 1/345 and 1/905 345 (12.85 ± 0.55, n = 11) or 1 and 345 (18.48± 0.87, n = 15) and also in
channels (see also Table 2). Again, this indicates that whereas blockade cells expressing the double labeled 1/345 construct (10.55 ± 0.66, n = 29).
of EAG/PAS domain functionality by introducing the protein fluoro- However, only a basal FRET efficiency is obtained in cells carrying
phore in the amino terminus does not impair the ability of the fusion double-labeled 345/1158 channels (4.36± 0.35, n = 34: not significant
proteins to assemble, tetramerize and be functionally expressed in the versus HERG/VR1 expressing cells). A small but significant EFRET was
1694 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
Fig. 8. Functional expression and FRET measurements with HERG channels labeled at residue 905 and in different combinations. (A) Representative ionic currents from cells co-
expressing HERG channels labeled with CFP and VFP at residue 905 (top) or transfected with double labeled subunits carrying fluorophores both at residues 1 and 905 (Double 1/905)
or 345 and 905 (Double 345/905). 1 s depolarization pulses up to +60 mV in 20 mV increments from a holding voltage of −80 mV were delivered to the cells, followed by a 2 s
repolarization period at −50 mV. Note the expanded time scale and the fast deactivating tail currents obtained with the double labeled 1/905 construct. (B) Determination of donor
fluorescence recovery during acceptor photobleaching in cells either co-expressing HERG channels carrying CFP and VFP at residues 905/905, 1/905, 905/1158 and 345/905, or in cells
transfected with subunits double-labeled at residues 1 and 905 or 345 and 905. Relative increases in CFP fluorescence over initial levels and the remaining fluorescence intensity of
YFP were obtained from transiently transfected CHO cells as detailed in the legend of Fig. 2. The symbols represent the calculated mean ± SEM, whereas the thin lines correspond to
single cell traces. (C) Comparison of EFRET for channels labeled as indicated in panel B. Bars represent the mean ± SEM of 2–5 transfections for the indicated number of cells. FRET levels
from the non-interacting HERG/VR1 pair and from co-expressed channels labeled with CFP and YFP in the amino end are also shown for comparison. Data corresponding to co-
transfections with 4 µg DNA of the double labeled constructs plus 4 µg of the plasmid containing wild-type HERG are also shown.
present in cells co-transfected with constructs labeled at residues 345 with double-labeled constructs 345/1158, 1/905 and 345/905 (Figs. 7C
and 1159 (6.14 ± 0.38, n = 20; P b 0.05 vs. HERG/VR1). and 8C) or 1/1158 (EFRET 8.6 ± 0.42, n = 20). In all cases, only moderate
It is important to note that EFRET remained almost the same when reductions in EFRET ranging from 7% with the double 1/345 to 27% with
4 µg of plasmid DNA coding for unlabeled HERG plus 4 µg of DNA the double 345/1158 constructs were observed. As expected, the
containing the double-labeled 1/345 construct was used for transfec- relatively small effect caused by introduction of an excess of unlabeled
tion (Fig. 7C). These total amounts and proportions of plasmid DNA HERG is not due to reduced expression of this protein, as shown by the
were chosen to maintain as much as possible the viability of the appearance of slowly deactivating HERG currents in cells co-expres-
transfected cells and also enough of a fluorescence level of the double- sing the fast deactivating 1/1158, 1/345 or 1/905 double-labeled
labeled protein to perform the FRET measurements. Similar results channels and wild-type HERG (not shown). These results add further
were obtained when unlabeled wild-type HERG was co-expressed support to the interpretation that the FRET signals are mainly due to
P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699 1695
specific assembled proteins and not to random collisions of over- and 345-located labels. This is confirmed by the FRET level obtained
expressed molecules. It should also be noted that depending on the with the 1 and 345 pair co-expressing cells. The 14.7 ± 1.32 specific
stoichiometry and torsion of the channel oligomers, some EFRET EFRET obtained in this case (Fig. 7C), the highest value of all the tested
reductions are possible even with specifically assembled tetramers in combinations, is not consistent with the EFRET value below 10 expected
which unlabeled subunits are combined with double-labeled ones. for any vertical distance separating the 1 and 345 label planes if no
Thus, as illustrated by the possible transference pathways shown in angular torsion exists between them. Note also that if this torsion does
Supplemental Fig. 1A, lower efficiencies will be obtained if only a single not occur, such distances would never be lower than 4–5 nm due to the
donor and acceptor pair carried by a given subunit is included in the dimensions of the GFP-based fluorophores. Therefore, only an effective
tetrameric channel schemes, eliminating the additional energy range of FRET efficiencies of around 2 or less can be expected under
transferences to unlabeled adjacent partners. It might also be possible these circumstances. Finally, even though the high EFRET observed with
that some subtle modification of the relative orientation of both the 1/345 pair seems to lie above that expected for the 7.7–8.1 nm
fluorophores takes place in the otherwise specifically assembled vertical separation predicted by the double-labeled 1/345 data (see
molecules, when part of the eight fluorophore tetramers produced by above), it is likely that this distance became increased in the double-
expressing the double labeled constructs alone are changed to lower labeled construct (but not so much in the co-transfected pair) by some
order fluorescent oligomers by introduction of non-fluorescent steric collision between labels. It is also possible that in this case the
subunits, leading to some variation in the observed FRET levels. physical proximity of both labels could force a more unfavorable
Equivalent data to those corresponding to 345-labeled channels, orientation of both fluorophores, someway contributing to reduce the
but obtained with different combinations of channel subunits labeled observed FRET level.
at residue 905 and other positions, are shown in Fig. 8B,C. In these The performance of an analogous analysis using the results obtain-
cases, cells co-expressing donor- and acceptor-labeled subunits at ed with different combinations of 905-labeled channels indicates that
position 905 yielded an EFRET value of 11.9 ± 0.85 (n = 31). The FRET the 8.12 ± 1.3 net EFRET of cells co-expressing subunits labeled at residue
efficiency was even higher upon expression of the double-labeled 905 means an 8.12 ± 0.24 nm side size for the quadratic plane corres-
constructs 1/905 (17.5 ± 0.8, n = 18) and 345/905 (15.03 ± 1.1, n = 16). On ponding to the labels at this position (Fig. 5B,C). The 13.72 ± 1.25 net
the other hand, a considerably smaller EFRET was recorded in cells co- EFRET obtained with the double-labeled 1/905 construct (Fig. 8B,C)
expressing HERG subunits labeled with CFP and VFP at positions 1 and predicts a vertical separation of 6.33 to 6.96 nm between the amino
905 (7.0 ± 0.75, n = 9) or at positions 905 and 1158 (6.51 ± 0.33, n = 30). and 905 fluorophore planes, both for arrangements with no angular
Finally, a FRET efficiency of 9.23 ± 0.57 (n = 22) was obtained upon co- torsion or maximal 45° rotation, respectively. According to these
expression of 345- and 905-labeled HERG channel subunits. distances, a FRET efficiency of between 5 and 19 would be expected for
Interpretation of these FRET efficiencies as previously detailed for co-expressed 1- and 905-labeled channels if torsion is present, but
the case of the amino and carboxy end pairs indicates that, for the only of around 3 with no angular torsion (Supplemental Fig. 4C).
9.07 ± 1.0 net EFRET obtained after co-expressing CFP and VFP pairs Therefore, the 3.32 ± 1.2 net EFRET obtained under these circumstances
labeled at residue 345, the square x–y plane corresponding to the (Fig. 8C) suggests that no angular torsion exists between the planes
labels at position 345 of the tetrameric structure would have a side defined by the amino terminus- and 905-located labels. This is further
size of 7.87 ± 0.16 nm (Fig. 5B,C). According to this value and given supported by the data from cells co-expressing 905- and 1158-labeled
the ≈8.6 nm lateral size of the plane delimited by the carboxy end subunits. In this case, the 2.73 ± 0.78 net EFRET (Fig. 8B,C) would also be
labels, the negligible specific FRET obtained with the double labelled consistent with the EFRET value below 5 expected for an almost vertical
345/1158 channel means that the vertical separation between the positioning of the 905 and 1158 labels, but very different from the EFRET
345 and the 1158 label planes is at least 9–10 nm regardless of the above 10 predicted with torsion between them (Supplemental Fig. 4E).
angular torsion established between them (see Supplementary Fig. 1 Finally, a rotated 345 plane with respect to non-torsioned NH2, 905 and
for a detailed description of the procedures analogously used to find COOH planes, predicts a 4–7 EFRET level after co-expression of 345- and
the corresponding vertical distance with 1/1158 double labeled 905-labeled subunits. This is indeed coherent with the 5.45 ± 1.02 net
channels). Since these data indicate a slightly longer vertical distance EFRET measured in these cells (Fig. 8B,C), a FRET level that would not be
separating the 345 and COOH label planes than the NH2 and COOH, obtained if fluorophores at the 345 and 905 positions were vertical to
the 345-located fluorophores are slightly above those introduced at each other, since in this situation almost no FRET would be expected
the amino termini. Application of an equivalent analysis to the (Supplemental Fig. 4D).
double-labeled 1/345 data predicts a vertical separation of between
7.7 (with 45° torsion) and 8.1 nm (without torsion) for the planes 4. Discussion
corresponding to the 1 and 345 labels (but see below).
Knowing the lateral size of the NH2, COOH and 345 planes and Here, we have generated a library of fluorescent HERG channel
the vertical separation between them, it should be possible to use constructs carrying CFP and V/YFP in multiple positions of the HERG
the data obtained by co-expressing pairs of subunits labeled at those coding sequence, to study by FRET the relative arrangement of
positions to further define their relative arrangement and, more different regions in the protein. The labeled channels were functionally
importantly, to validate the previously calculated distances. The expressed in HEK293 and CHO cells. Remarkably, despite the insertion
results of this analysis are presented in Supplemental Fig. 4A,B. It can of a large fluorescent domain in the channel protein, many HERG fusion
be observed that for a vertical distance of 9–10 nm separating the proteins are still fully functional. Essentially, we found that fusion of
345 and 1159 fluorophore planes, a FRET efficiency of around 2 the fluorescent proteins to the cytoplasmic regions corresponding to
would be expected, providing that a sizeable torsion is present (see the proximal domain in the amino terminus or the final segment of the
Supplemental Fig. 4B). This nicely fits with the 2.36 ± 0.83 value of carboxy terminus do not compromise channel function. However,
net EFRET exhibited by cells co-expressing 345- and 1159-labeled apart from the construct carrying the label at the first amino acid of the
subunits (Fig. 7C), but not with the total absence of FRET expected protein, no functional channels were detected upon insertion of the
for channels in which the vertices of the 345 and 1159 label planes reporter fluorophores in the initial portion of the amino terminus
are perpendicular without any angular torsion. corresponding to the EAG/PAS domain, in the central channel core, or
Interestingly, since the planes delimited by the amino and carboxy in the segment after the sixth transmembrane helix including the
end-located labels do not show any sizeable torsion between them (see linker between S6 and the CNBD and the CNBD itself.
above), an angular rotation similar to that encountered between the The localization of the permissive insertions only in specific areas
labels in the 345 and 1159 positions should also exist between the 1- of the sequence may provide some insights about the structural
1696 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
arrangement and/or the functional implications of these regions in that in our case the fluorescent labels may be mainly localized in
the context of the whole channel. Thus, it is likely that more peripheral positions, including those introduced in the amino end.
deleterious effects are expected by insertion of the labels in an We propose therefore that the C-linker/CNBD region hangs cen-
internally located domain when compared to insertions in a surface trally below the transmembrane core, in a similar fashion to that
aqueous-exposed region. The lack of functionality of the C-linker- and described for the same tetrameric region of HCN2 [10]. Unlike the
CNBD-labeled constructs would be coherent with the localization of central T1 domain position underneath the core of Kv channels [2–6],
this region as a compact tetrameric structure in a central position the amino terminus of HERG probably tracks to the periphery
immediately below the cytoplasmic pore opening as previously surrounding the C-linker/CNBD and extending to its bottom, where
proposed for HCN and TRP channels [8,10]. Indeed, the participation the proximal domain and the distal carboxy terminus might interact.
of this region in the tetrameric assembly of HERG has been also As depicted in the homology model shown in Fig. 9A, it is likely that
proposed [11]. Similarly, a compact EAG/PAS domain would also be the EAG/PAS domain is establishing extensive contacts with the top
compatible with its ability to form tetramers in vitro in the absence of and side surfaces of the C-linker/CNBD, helping to explain the
the rest of the protein, and to inhibit the functional expression of wild- functional interaction of the initial part of the N-terminus with the
type HERG [16]. However, this protein segment only forms monomers gating machinery, but also the disruptive effects of introducing
in crystal structures [18] and channels carrying N-terminal truncations voluminous labels in the EAG/PAS that could considerably perturb
that delete the EAG/PAS domain can form functional channels in the the central channel structures. Remarkably, the five positions in the
plasma membrane [18–22], even though reduced current densities are EAG/PAS domain carrying a label (residues 43, 78, 87, 88 and 104)
systematically obtained upon expression of constructs lacking either appear in quite different locations encircling the surface of the crystal
the EAG/PAS domain or most of the amino terminus (e.g., Δ2–135 of Δ EAG/PAS structure [18], making unlikely that all of them are directed
2–370 deletions; C. Alonso-Ron et al. unpublished). Interestingly, towards a central space in the channel. However, all of them appeared
relatively long distances (corresponding to around 8–9 and 11–12 nm non-functional. This opens the possibility that the EAG/PAS is still
for adjacent and diagonal separations) between labels at equivalent surrounded by other portions of the protein as recently proposed for
positions of the four subunits, are suggested from our FRET-derived the T1 domain of Kv2.1 [55]. Thus, further work would be necessary to
measurements using fully functional HERG channels. This suggests check if the proximal domain and/or the carboxy terminus extend to
Fig. 9. Molecular arrangement of the fluorescent labels introduced in the HERG cytoplasmic domains and homology model of the channel. (A) Homology models of the HERG channel
(a) and the amino terminus-labeled construct (b). The models were constructed using surface representations of the Kv1.2 pore region (PDB ID: 2A79), the C-linker/cNBD region from
HCN2 (PDB ID: 1Q5O) and the HERG EAG/PAS domain (PDB ID: 1BYW) or a ribbon representation of YFP (PDB ID: 1F0B) showing the position of the chromophore in the center of the
molecule (yellow spheres). The different components are shown at the same scale. The Kv1.2 representation presents the four subunits of the tetramer differently coloured in green,
blue, yellow and magenta. The position of the S4–S5 loop has been marked in red. Note that the fourth C-linker/cNBD, EAG/PAS and YFP structures located nearest the viewer have been
removed for a better view of the assemblies. The rotation of these structures around the central vertical axis of the complex is arbitrary. The position of the EAG/PAS has been fixed to
direct its amino terminal end (red spheres) towards the S4–S5 loop in the channel core. Both the EAG/PAS and the YFP structures have been docked against the channel complex
preventing any superposition of the molecular volumes. Note also the good correspondence of the interfluorophore distances in the construct represented in panel b (carrying the YFP
molecules in the position corresponding to the amino ends) and those calculated from FRET measurements. The unknown structures of the proximal domain and the remainder of the
C terminus possibly located below the illustrated compositions, are not shown. (B) Relative position and approximated distances between fluorophores introduced in positions 1 (NH2),
345, 905 and 1159 (COOH) of HERG. The transmembrane core of Kv1.2 is shown at the top as a reference. A ribbon representation of YFP showing the approximate dimensions of the
molecule at the same scale is shown in the upper inset. The positions of the amino and carboxy ends of the fluorescent protein and the approximate location of the insertion point in the
host (black circle) are indicated. Lower inset. Ribbon diagram of Kv1.2 viewed from the cytoplasmic face with the S4-S5 in red. Molecular representations and combinations were
generated using the UCSF Chimera from Computer Graphics Laboratory, University of California, San Francisco (supported by NIH P41 RR-01081).
P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699 1697
the bottom of the channel either running parallel to the EAG/PAS or of those cells showing a low acceptor-to-donor ratio, ensured that
wrapping around it. only cells in which most of the donor is coupled to an acceptor were
A question about the structural combinations shown in Fig. 9A is considered. This minimized the possible influence of unpaired donors
whether the volume of the EAG/PAS domain and the distances and (αD factor, see Methods) to measured EFRET ensuring that the
volumes of the attached GFP variants could fit into the general measured values closely approach the true FRET efficiency. Control
dimensions of the model. In this respect it is worth noting that the of EFRET independence from the fluorochrome expression levels over
separation between the channel core and the CNBD domains is pro- a wide range, the demonstration of equivalent FRET levels when
bably longer than the arbitrarily set distance shown in Fig. 9A. Since the measurements are done with the whole cell fluorescence or when
three-dimensional structure of the HERG channel transmembrane core they are restricted to plasma membrane regions in which properly
is unknown, we used as a reference the atomic structure recently assembled functional channels are located, and the systematic
determined by X-ray crystallography for Kv1.2 [56], a voltage-depen- subtraction of signals coming from non-interacting molecules were
dent K+ channel showing a 26% identity and 43% similarity to HERG in used as an indication that our signals came from the specific
the transmembrane helices and pore regions (data not shown). Thus, assembly of ion channel subunits and not from random encounters
whereas in the Kv1.2 core structure the C-terminal end of the S6 helix or formation of bulk aggregates as a result of construct over-
lies closely associated to the internal surface of the plasma membrane expression.
due to its angular bending at the level of the PVP sequence, in HERG Performance of a quantitative analysis of FRET signals obtained
which lacks such a motif, the S6 should emerge from the trans- with a pair of fluorophores provides a unique opportunity to discern
membrane channel core in a more linear and tubular-like way. This the distance between donor and acceptor under physiological
topology has been recognized in the three-dimensional organization conditions and in a natural environment within the intact cell, using
of other channels lacking the PVP signature such as KcsA [57] and in the Förster equation EFRET = Ro6/Ro6 + d6 [45,48,49,52,60,62], where d
the recently determined crystal structure of the cyclic nucleotide- stands for the interfluorophore distance and Ro corresponds to the
regulated channel MlotiK [58]. As previously suggested [8,59], it is distance at which EFRET is 0.5 (≈5 nm for the CFP/YFP pair assuming a
likely that at least in some channel states the C-linker region random orientation distribution corresponding to a relative dipole
following S6 also exists in an extended and more tubular-like orientation factor κ2 of 2/3; [48,51,52]). As indicated previously, our
tetrameric arrangement than the vertically collapsed crystal structure data with the 1/1123, 1/1158 and 1/1159 pairs and the equivalent EFRET
of the isolated HCN C-linker/CNBD depicted in Fig. 9A [10]. This would values between the 1123-, 1158- or 1159-located fluorophores of the
increase the separation between the membrane surface and the CNBD four subunits suggest a quite random orientation of the labels.
structures allowing for a better approximation of the EAG/PAS and/or Previous analysis of the possible errors introduced by assuming κ2 = 2/
the GFP variants to the central axis of the tetramer. Finally, assuming 3 also indicated that the uncertainty in distance introduced by the
a peripheral localization of the GFP-derived fluorophores either orientation factor is relatively small, usually less than 20% [43,60,61].
longitudinally or transversely projecting from the channel structure There is also evidence that covalently attached GFP variants are quite
(Fig. 9Ab), the ≈8 nm lateral size of the planes defined by the centre of mobile under physiological conditions and anisotropy measurements
the labels would localize their insertion points at a shorter distance indicate that they adopt random orientations [63,64]. Altogether, this
(4–6 nm at both sides) from the central axis. For the labels introduced indicates that within certain limits our FRET data can be interpreted in
in the amino termini, that seem to interact with the S4–S5 loop, this terms of distance, allowing a reasonable view of the global dimensions
distance would be consistent with the 3–6 nm range in which this of HERG. It is important to emphasize that our measurements using
loop is located relative to the central axis (lower inset of Fig. 9). different combinations of fluorescent labels located in different
Unfortunately, the use of non-functional HERG constructs to check a positions also provide an additional method to validate the measured
more compact localization of other labels towards the protein centre distances. Thus, the final estimation of the distance between two
is hampered by the possibility of strong structural alterations in these given positions lies not only in the FRET data from channel constructs
constructs as a result of fluorophore insertion. Subsequently, we carrying the fluorophore partners in these positions, but also in the
focused our efforts in those channel constructs exhibiting an intact combination of measurements done with a number of labels in order
functionality, indicative of no major distortions as a result of labeling. to confirm that the calculated distances are coherent with those
There are two possible concerns regarding the utility of our FRET obtained with other pairs.
measurements to obtain information about the molecular organiza- A limitation of inferring distances and/or structures from GFP-
tion of HERG. The first one relates to the specificity and quantification based fluorophores is that the GFP chromophores are located in the
of the obtained signals and the second refers to the real meaning of centre of a barrel-like entity ≈2.5 nm in diameter and ≈4 nm long
the calculated interfluorophore distances. Due to the dependence of (see inset of Fig. 9B). Apart from restricting the minimum distance
FRET from intra- and intermolecular distances between fluorophores, detectable by FRET with a single CFP/YFP pair to the 5–10 nm range
it could be expected that measuring the amount of energy transfer due to donor–acceptor steric collision, the calculated distances
allows the determination of donor-acceptor distance [60–62]: the would correspond to those separating the CFP and V/YFP chromo-
farther apart the fluorophores, the less energy transfer. However, it is phores, not to those between the residues to which the labels are
also possible to obtain a considerable FRET signal in spite of a attached, limiting the spatial resolution of our measurements and
relatively long intramolecular distance due to unspecific random the interpretation of FRET in terms of intramolecular distances. The
encounters produced by overexpression of labeled molecules. Con- interfluorophore distances estimated here have been incorporated
versely, a low FRET signal and/or a long estimated distance may be into the models shown in Fig. 9 that are mostly consistent with our
obtained even when a short interfluorophore separation exists if an FRET measurements. Evidently the absolute coordinates shown
inappropriate Förster distance (Ro, see below) is used or a complex cannot be considered as indicative of the real distances separating
(e.g., tetrameric) stoichiometry is present. the protein residues, but they can contribute to better define the
Regarding the first concern, our measurements of donor de-quen- relative position of some channel domains. Our data indicate that the
ching upon acceptor photobleaching provide a simple and practical amino and carboxy end-located fluorescent labels and those
approach to unequivocally identify the existence of FRET signals, introduced at position 905 of the four channel subunits delimit
since such an increase in fluorescence would be achieved only if FRET three parallel square planes without sizeable torsion between them,
is present. Conversion of the fluorescence intensities into molar ratios placing the centre of the labels introduced after residues 905 and
by normalizing the intensity ratios to those of a donor–acceptor 1158/1159 almost vertically aligned with those attached to the amino
tandem imaged at the same settings, and exclusion from the analysis terminal end. This contrasts with the near 45° rotation of the
1698 P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699
fluorophores inserted after position 345. Due to the short vertical HERG fusion proteins could constitute an excellent basis for subs-
separation predicted between the planes delimited by the 1 and 345 tituting the GFP variants with shorter peptides to which small fluo-
labels, it could be argued that the proposed rotation of the 345- rophores such as biarsenical dyes [67] can be specifically attached. The
located labels is not related to a non-vertical arrangement of the use of these smaller fluorophores combined with the procedures,
corresponding positions in the protein structure, but forced by triangulations and expressions detailed in this report could also help
competing packing problems due to the big size of the tags. to confirm the relative dimensions reported here.
However, our data also predict a similar torsion between the 345-
and the 905- or 1158-located labels that are separated from the 345 Acknowledgments
fluorophores by a long distance. Note that if the 345-attached label
rotation is simply due to an angular trajectory of the 345-attached We thank Dr. Hugo Cabedo from University Miguel Hernández in
label that rotates it (clockwise or counter clockwise), or if it is the Alicante (Spain) the gift of the N-terminal-tagged VR1 construct. We
result of a structural torsion of the channel protein that places the also thank Dr. Luisa María Sierra for her geometrical considerations.
345 position near the amino end of a non-adjacent subunit remains This work was supported by grants SAF2003–00329 from Ministerio
to be established. de Ciencia y Tecnología and BFU2006–10936 from Ministerio de
Our data also indicate that whereas a vertical distance of around Educación y Ciencia of Spain (both partially co-financed by FEDER
8–9 nm exists between the planes limited by the amino and carboxy European funds), and grant IB05–002 from Principado de Asturias
end labels, only around 1–2 nm separate the 345 and 905 planes from (Spain). Finance support from Dirección General de Universidades e
the amino and the carboxy terminal ends, respectively. Interestingly, Investigación of Asturias for acquisition of the image equipment (ref.
the accuracy of these values is basically confirmed by the coincidence EQPT02–36) is also acknowledged. P.M. holds a predoctoral fellowship
of distances reported by the 345/NH2, 345/905, 345/COOH, 905/NH2 from FICYT of Asturias (ref. BP03–108). D.G.M. and C.A.R. are pre-
and 905/COOH pairs. doctoral fellows from the Spanish Ministerio de Ciencia y Tecnología
Even though our analysis only includes the four positions for which (refs. AP2000–4363 and BES-2004–3872).
the measured distances were unequivocally assessed and validated by
“multiple molecular triangulation”, it can yield some valuable infor- Appendix A. Supplementary data
mation about the overall channel arrangement. In fact, an upside-
down organization of the cytoplasmic labels with respect to that Supplementary data associated with this article can be found, in
schematized in Fig. 9B could also be compatible with the calculated
the online version, at doi:10.1016/j.bbamcr.2008.06.009.
distances. We find the proposed location of the amino and carboxy
termini towards the central core as the most probable arrangement of
the channel protein for three reasons. First, placing the 345 position References
away from the channel core and the carboxy end close to it would
[1] F.M. Ashcroft, From molecule to malady, Nature 440 (2006) 440–447.
imply a very long distance between the limit of the first transmem-
[2] G. Yellen, The voltage-gated potassium channels and their relatives, Nature 419
brane helix and residue 345, due to the ≈10 nm vertical separation (2002) 35–42.
reported here between the 345 and the carboxy planes, plus an [3] M. Ju, L. Stevens, E. Leadbitter, D. Wray, The roles of N- and C-terminal
additional separation from the membrane surface (top vertical arrow determinants in the activation of the Kv1.2 potassium channel, J. Biol. Chem.
278 (2003) 12769–12778.
and question mark in Fig. 9B), corresponding to the cavity that sepa- [4] O. Sokolova, A. Accardi, A.D. Gutierrez, A. Lau, M. Rigney, N. Grigorieff,
rates the cytoplasmic domains from the transmembrane core in the Conformational changes in the C terminus of Shaker K+channel bound to the rat
two-layered architecture of the HERG-related Kv and CNG channels Kvβ2-subunit, Proc. Natl. Acad. Sci. U.S.A. 100 (2003) 12607–12612.
[5] L.A. Kim, J. Furst, D. Gutierrez, M.H. Butler, S. Xu, S.A.N. Goldstein, N. Grigorieff,
[4,56,65,66]. Second, residue 345 lies only a few amino acids away from Three-dimensional structure of Ito: Kv4.2-KchIP2 ion channels by electron
the short HERG sequence (residues 362–366) recently recognized as a microscopy at 21 A resolution, Neuron. 41 (2004) 513–519.
key modulator of HERG activation by electrostatic interactions with the [6] B. Callsen, D. Isbrandt, K. Sauter, L.S. Hartmann, O. Pongs, R. Bähring,
Contribution of N- and C-terminal Kv4.2 channel domains to KChIP channel
gating machinery in the channel core [21,37]. Third, the closest posi- interaction, J. Physiol. (Lond.) 568 (2005) 397–412.
tion of the amino termini towards the central channel core as shown in [7] T.P. Roosild, K-T. Lê, S. Choe, Cytoplasmic gatekeepers of K+-channel flux: a
Fig. 9 is mostly compatible with the previously proposed interaction(s) structural perspective, Trends Biochem. Sci. 29 (2004) 39–45.
[8] N. García-Sanz, A. Fernández-Carvajal, C. Morenilla-Palao, R. Planells-Cases, E.
of the initial part of the EAG/PAS domain with the transmembrane
Fajardo-Sánchez, G. Fernández-Ballester, A. Ferrer-Montiel, Identification of a
channel structure [22,29,31,35]. It is also tempting to speculate that the tetramerization domain in the C terminus of the vanilloid receptor, J. Neurosci. 24
position of the 345 label, closer to the membrane surface but sizeably (2004) 5307–5314.
[9] J.W. Taraska, W.N. Zagotta, Cyclic nucleotide-regulated ion channels: spotlight on
rotated with respect to the amino ends might be related to the
symmetry, Structure 15 (2007) 1023–1024.
presence of a pathway allowing for a physical interaction between the [10] W.N. Zagotta, N.B. Olivier, K.D. Black, E.C. Young, R. Olson, E. Gouaux, Structural
initial portion of the amino terminus and the gating machinery, likely basis for modulation and agonist specificity of HCN pacemaker channels, Nature
at the level of the S4–S5 loop [22,29,35]. It is possible that in this case, 425 (2003) 200–205.
[11] Q. Gong, D.R. Keeney, J.C. Robinson, Z. Zhou, Defective assembly and trafficking of
the position of residue 345 above the amino terminus could be due to mutant HERG channels with C-terminal truncations in long QT syndrome, J. Mol.
its location in the connectors linking the transmembrane core and the Cell. Cardiol. 37 (2004) 1225–1233.
hanging bulk of the cytoplasmic domains [4,56,65,66]. [12] S. Kupershmidt, T. Yang, S. Chanthapahychith, Z. Wang, J.A. Towbin, D.M. Roden,
Defective human ether-a-go-go-related gene trafficking linked to an endoplasmic
We believe that the qualitative features of this model certainly reticulum retention signal in the C terminus, J. Biol. Chem. 277 (2002)
provide a useful initial constraint on the coarse structure of the 27442–27448.
channel oligomer, clarifying the relative arrangement of some key [13] F. Shoeb, A.P. Malykhina, H.I. Akbarali, Cloning and functional characterization of
the smooth muscle ether-a-go-go-related gene K+ channel, J. Biol. Chem. 278
landmarks on the cytoplasmic domains of HERG. Even though our data (2003) 2503–2514.
were obtained with large fluorophores, making it possible that they [14] A. Akhavan, R. Atanasiu, T. Noguchi, W. Han, N. Holder, A. Shrier, Identification of
are forced into a limited number of vacant spaces in the channel the cyclic-nucleotide-binding domain as a conserved determinant of ion-channel
cell-surface localization, J. Cell. Sci. 118 (2005) 2803–2812.
periphery, our results delimit at least two clearly differentiated and
[15] A.D.C. Paulussen, A. Raes, R.J. Jongbloed, R.A.H.J. Gilissen, A.A.M. Wilde, D.J.
spatially separated spots for the amino and the carboxy ends of the Snyders, H.J.M. Smeets, J. Aerssens, HERG mutation predicts short QT based on
protein. Additional triangulation with other members of our fluor- channel kinetics but causes long QT by heterotetrameric trafficking deficiency,
Cardiovasc. Res. 67 (2005) 467–475.
escent HERG channel library using the general methodology depicted
[16] X. Li, J. Xu, M. Li, The human Δ1261 mutation of the HERG potassium channel
here should render very helpful data to position other parts of the results in a truncated protein that contains a subunit interaction domain and
protein. On the other hand, it can be anticipated that our collection of decreases the channel expression, J. Biol. Chem. 272 (1997) 705–708.
P. Miranda et al. / Biochimica et Biophysica Acta 1783 (2008) 1681–1699 1699
[17] P. Phartiyal, E.M.C. Jones, G.A. Robertson, Heteromeric assembly of HERG 1a/1b [42] A. Miyawaki, R.Y. Tsien, Monitoring protein conformations and interactions by
channels occurs cotranslationally via amino terminal interactions, J. Biol. Chem. fluorescence resonance energy transfer between mutants of green fluorescence
282 (2007) 9874–9882. protein, Meth. Enzymol. 327 (2000) 472–500.
[18] J.H. Morais-Cabral, A. Lee, S.L. Cohen, B.T. Chait, M. Li, R. MacKinnon, Crystal [43] M.G. Erickson, B.A. Alseikhan, B.Z. Peterson, D.T. Yue, Preassociation of calmodulin
structure and functional analysis of the HERG potassium channel N terminus: a with voltage-gated Ca2+ channels revealed by FRET in single living cells, Neuron
eukaryotic PAS domain, Cell 95 (1998) 649–655. 31 (2001) 973–985.
[19] R. Schönherr, S.H. Heinemann, Molecular determinants for activation and [44] R.N. Day, F. Schaufele, Imaging molecular interactions in living cells, Mol. Endo-
inactivation of HERG, a human inward rectifier potassium channel, J. Physiol. crinol. 19 (2005) 1675–1686.
(Lond.) 493 (1996) 635–642. [45] E.B. Van Munster, G.J. Kremers, M.J.W. Adjobo-Hermans, T.W.J. Gadella Jr,
[20] P.S. Spector, M.E. Curran, A. Zou, M.T. Keating, M.C. Sanguinetti, Fast inactivation Fluorescence resonance energy transfer (FRET) measurement by gradual acceptor
causes rectification of the IKr channel, J. Gen. Physiol. 107 (1996) 611–619. photobleaching, J. Microsc. 218 (2005) 253–262.
[21] C.G. Viloria, F. Barros, T. Giráldez, D. Gómez-Varela, P. de la Peña, Differential [46] C.L. Takanishi, E.A. Bykoiva, W. Cheng, J. Zheng, GFP-based FRET analysis in live
effects of amino-terminal distal and proximal domains in the regulation of human cells, Brain Res. 1091 (2006) 132–139.
erg K+ channel gating, Biophys. J. 79 (2000) 231–246. [47] P. Miranda, T. Giráldez, P. de la Peña, D.G. Manso, C. Alonso-Ron, D. Gómez-Varela,
[22] J. Wang, M.C. Trudeau, A.M. Zappia, G.A. Robertson, Regulation of deactivation by P. Domínguez, F. Barros, Specificity of TRH receptor coupling to G-proteins for
an amino terminal domain in Human ether-a-go-go-related gene potassium regulation of ERG K+ channels in GH3 rat anterior pituitary cells, J. Physiol. (Lond.)
channels, J. Gen. Physiol. 112 (1998) 637–647. 566 (2005) 717–736.
[23] D.L. Minor Jr, Potassium channels: life in the post-structural world, Curr.Op. Struct. [48] V. Raicu, D.B. Jansma, R.J.D. Miller, J.D. Friesen, Protein interaction quantified in
Biol. 11 (2001) 408–414. vivo by spectrally resolved fluorescence resonance energy transfer, Biochem. J.
[24] A. Varshney, M.K. Mathew, A tale of two tails: cytosolic termini and K+ channel 385 (2005) 265–277.
function, Prog. Biophys. Mol. Biol. 83 (2003) 153–170. [49] T. Zal, N.R.J. Gascoigne, Photobleaching-corrected FRET efficiency imaging of live
[25] J.W. Warmke, B. Ganetzky, A family of potassium channel genes related to EAG in cells, Biophys. J. 86 (2004) 3923–3939.
Drosophila and mammals, Proc. Natl. Acad. Sci. U.S.A. 91 (1994) 3438–3442. [50] H. Amiri, G. Schultz, M. Schaefer, FRET-based analysis of TRPC subunit
[26] M.C. Sanguinetti, M. Tristani-Firouzi, hERG potassium channels and cardiac stoichiometry, Cell Calcium 33 (2003) 463–470.
arrhythmia, Nature 440 (2006) 463–469. [51] G.H. Patterson, D.W. Piston, B.G. Barisas, Förster distances between green
[27] E. Wanke, R. Restano-Cassulini, Toxins interacting with ether-a-go-go-related fluorescent protein pairs, Anal. Biochem. 284 (2000) 438–440.
gene voltage-dependent potassium channels, Toxicon 49 (2006) 239–248. [52] E. Biener-Ramajunan, V.K. Ramajunan, B. Herman, A. Gertler, Spatio-temporal
[28] E. Raschi, V. Vasina, E. Poluzzi, F. De Ponti, he hERG K+channel: target and kinetics of growth hormone receptor signalling in single cells using FRET
antitarget strategies in drug development, Pharmacol. Res. 57 (2008) 181–195. microscopy, Growth Horm. IGF Res. 16 (2006) 247–257.
[29] M.C. Sanguinetti, Q.P. Xu, Mutations of the S4-S5 linker alter activation properties [53] B. Corry, D. Jayatilaka, P. Rigby, A flexible approach to the calculation of resonance
of HERG potassium channels expressed in Xenopus oocytes, J. Physiol. (Lond.) 514 energy transfer efficiency between multiple donors and acceptors in complex
(1999) 667–675. geometries, Biophys. J. 89 (2005) 3822–3836.
[30] P.L. Smith, G. Yellen, Fast and slow voltage sensor movements in HERG potassium [54] E.A. Jares-Erijman, T.M. Jovin, FRET imaging, Nat. Biotechnol. 21 (2003) 1387–1395.
channels, J. Gen. Physiol. 119 (2002) 275–293. [55] B. Adair, R. Nunn, S. Lewis, I. Dukes, L. Philipson, M. Yeager, Single particle image
[31] J. Liu, M. Zhang, M. Jiang, G-N. Tseng, Negative charges in the transmembrane reconstruction of the human recombinant Kv2.1 channel, Biophys. J. 94 (2008)
domains of the HERG K channel are involved in the activation- and deactivation- 2106–2114.
gating processes, J. Gen. Physiol. 121 (2003) 599–614. [56] S.B. Long, E.B. Campbell, R. MacKinnon, Crystal structure of a mammalian voltage-
[32] D.R. Piper, W.A. Hinz, C.K. Tallurri, M.C. Sanguinetti, M. Tristani-Firouzi, Regional dependent Shaker family K+ channel, Science 309 (2005) 897–903.
specificity of hERG channel activation and inactivation gating, J. Biol. Chem. 280 [57] D.A. Doyle, J. Morais-Cabral, R.A. Pfuetzner, A. Kuo, J.M. Gulbis, S.L. Cohen, B.T.
(2005) 7206–7217. Chait, R. MacKinnon, The structure of the potassium channel: Molecular basis of
[33] M. Zhang, J. Liu, G-N. Tseng, Gating charges in the activation and inactivation K+conduction and selectivity, Science 280 (1998) 69–77.
processes of the HERG channel, J. Gen. Physiol. 124 (2004) 703–718. [58] G.M. Clayton, S. Altieri, L. Heginbotham, V.M. Unger, J.H. Morais-Cabral, Structure
[34] T. Ferrer, J. Rupp, D.R. Piper, M. Tristani-Firouzi, The S4-S5 linker directly couples of the transmembrane regions of a bacterial cyclic nucleotide-regulated channel,
voltage sensor movement to the activation gate in the human ether-a-go-go- Proc. Nat. Acad. Sci. USA 105 (2008) 1511–1515.
related gene (hERG) K+ channel, J. Biol. Chem. 281 (2006) 12858–12864. [59] K.V. Craven, W.N. Zagotta, Salt bridges and gating in the COOH-terminal region of
[35] J. Wang, C.D. Myers, G.A. Robertson, Dynamic control of deactivation gating by a HCN2 and CNGA1 channels, J. Gen. Physiol. 124 (2004) 663–677.
soluble amino-terminal domain in HERG K+ channels, Biophys. J. 115 (2000) [60] L. Stryer, Fluorescence energy transfer as a spectroscopic ruler, Ann. Rev. Biochem.
749–758. 47 (1978) 819–846.
[36] J. Chen, A. Zou, I. Splawski, M.T. Keating, M.C. Sanguinetti, Long QT syndrome- [61] R.M. Clegg, Fluorescence resonance energy transfer and nucleic acids, Meth.
associated mutations in the Per-Arnt-sim (PAS) domain of HERG potassium Enzymol. 211 (1992) 353–388.
channels accelerate deactivation, J. Biol. Chem. 274 (1999) 10113–10118. [62] P.R. Selvin, Fluorescence resonance energy transfer, Meth. Enzymol. 246 (1995)
[37] J.B. Saenen, A.J. Labro, A. Raes, D.J. Snyders, Modulation of HERG gating by a charge 300–334.
cluster in the N-terminal proximal domain, Biophys. J. 91 (2006) 4381–4391. [63] I. Riven, E. Kalmanzon, L. Segev, E. Reuveny, Conformational rearrangements
[38] D. Gómez-Varela, F. Barros, C.G. Viloria, T. Giráldez, D.G. Manso, S.G. Dupuy, P. associated with the gating of the G protein-coupled potassium channel revealed
Miranda, P. de la Peña, Relevance of the proximal domain in the amino-terminus by FRET microscopy, Neuron 38 (2003) 225–235.
of HERG channels for regulation by a phospholipase C-coupled hormone receptor, [64] J. Zheng, M.D. Varnum, W.N. Zagotta, Disruption of an intersubunit interac-
FEBS Lett. 535 (2003) 125–130. tion underlies Ca2+ -calmodulin modulation of cyclic nucleotide-gated chan-
[39] D. Gómez-Varela, P. de la Peña, J. García, T. Giráldez, F. Barros, Influence of amino- nels, J. Neurosci. 23 (2003) 8167–8175.
terminal structures on kinetic transitions between several closed and open states [65] M.K. Higgins, D. Weitz, T. Warne, G.F.X. Schertler, U.B. Kaupp, Molecular
in human erg K+ channels, J. Membrane Biol. 187 (2002) 117–133. architecture of a retinal cGMP-gated channel: the arrangement of the cytoplasmic
[40] D.L. Sheridan, C.H. Berlot, A. Robert, F.M. Inglis, K.B. Jakobsdottir, J.R. Howe, T.E. domains, EMBO J. 21 (2002) 2087–2094.
Hughes, A new way to rapidly create functional, fluorescent fusion proteins: ran- [66] M. Pioletti, F. Findeisen, G.L. Hura, D.L. Minor Jr., Three-dimensional structure of
dom insertion of GFP with an in vitro transposition reaction, BMC Neurosci. 3 the KchIP-Kv4.3 T1 complex reveals a cross-shaped octamer, Nat. Struct. Mol. Biol.
(2002) 7. 13 (2006) 987–995.
[41] T. Giráldez, T.E. Hughes, F.J. Sigworth, Generation of functional fluorescent BK [67] B.R. Martin, B.N.G. Giepmans, S.R. Adams, R.Y. Tsien, Mammalian cell-based
channels by random insertion of GFP variants, J. Gen. Physiol. 126 (2005) optimization of the biarsenical-binding tetracysteine motif for improved
429–438. fluorescence and affinity, Nat. Biotechnol. 23 (2005) 1308–1314.