Está en la página 1de 28

The electrochemical basis of corneal hydration, swelling

and transparency
Peter M. Pinsky Xi Cheng
April 20, 2017

Author for correspondence:


Peter M. Pinsky
e-mail: pinsky@stanford.edu

Abstract
Fluid pressure, water content and charge concentration are intimately coupled in the corneal
stroma. The fluid pressure is composed of a hydrostatic pressure component associated
with the intraocular pressure and an osmotic pressure component associated with the excess
concentration of ions in the stroma compared to the aqueous humor. While the value of
the intraocular pressure is fixed, in the living cornea, ion pumps located in the endothelium
actively modulate the osmotic pressure. The osmotic pressure in turn causes water exchange
between the stroma and aqueous humor. The resulting hydration of the stroma will have
a direct effect on the collagen fibril lattice and therefore on the transparency of the tissue.
The entire system is driven by the energy of various charges, as well as the metabolic energy
used in ion transport. In this chapter, we attempt to provide a comprehensive model for
corneal hydration, swelling and transparency. Many parts of this system have been both
experimentally investigated and modeled in previous groundbreaking studies. In this work,
we describe a macroscopic model for in vivo corneal hydration based on a novel energy
approach that can be used for predicting stromal hydration under various conditions, for
example, under variations in metabolic state, and for improving the description of corneal
biomechanics.

Keywords: cornea, stroma, corneal hydration, collagen-swelling interaction, collagen, pro-


teoglycan, glycosaminoglycan, osmotic pressure.

1 Introduction
The fluid pressure within the corneal stroma is composed of hydrostatic and osmotic compo-
nents. The hydrostatic pressure results from the intraocular pressure of the anterior chamber,

1
whereas the osmotic pressure results from the interplay of fixed charges and mobile ions. In
fact, from an electrochemical perspective, the corneal stroma is a highly-hydrated polyelec-
trolyte gel consisting of an interacting mixture of fluid, solid and ionic phases. Stromal
water saturates the collagen and proteoglycan solid phase, solvates the ionic phase, and ac-
counts for about 78% of the cornea by weight [1]. A small portion of the water is cellular
or bound to the stromal collagen [2, 3], but most of the water is free to flow within the
stroma in response to gradients in fluid pressure. The solid phase is comprised of a flexible
collagen network, organized as fibrils within lamellae, and associated proteoglycans (PGs).
Stromal PGs1 have sulfated linear sidechains of negatively charged disaccharide units called
glycosaminoglycans (GAGs)2 that are covalently bound at one end to the PG core protein
[4, 5]. At normal pH, the stromal fixed charge is almost entirely due to GAG ionization [6].
The ionic phase includes dissolved salts, primarily Na+ and Cl− , and metabolites such as
C3 H5 O− −
3 (lactate ion) and HCO3 (bicarbonate ion).
Stromal mobile ions interact electrostatically with the GAG fixed charges and form cloud-
like distributions, giving rise to osmotic pressure. The aqueous humor, filling the anterior
chamber, acts as an ionic bath for the stroma and provides a reservoir of water that is
available for exchange with the stroma. Water transport across the permeable endothelial
layer3 , which partitions the stroma and aqueous humor, is driven by the osmotic pressure
difference between the two phases. When ionic concentration in the stroma exceeds that in
the aqueous humor, a positive osmotic pressure difference is created and water will tend to
flow from the aqueous chamber into the stroma, and vice-versa when the osmotic pressure
difference is negative. In the metabolically functioning (in vivo) cornea, ion pumps, located
in the endothelial layer, actively transport ions from the stroma into the aqueous humor.
This lowers stromal osmotic pressure and modulates the exchange of water with the aqueous
humor and is referred to as the pump-leak mechanism of stromal hydration [7, 8, 9].
The regulatory system and details of the molecular mechanisms responsible for active ion
transport, which requires Na+ , K+ , ATPase and carbonic anhydrase activity to transport
HCO− − +
3 , Cl and possibly Na , are not yet fully understood [9]. It may be noted that the
pump-leak mechanism is not unique to the cornea. A number of tissues employ epithelial
and/or endothelial layers as barriers to separate phases and to actively mediate the exchange
of solutes and solvent between those phases. Other examples include the lining of blood
vessels, the kidney, organs of the gastrointestinal tract, and the choroid plexus in the brain.
Charge can produce powerful forces, either directly by electrostatics or indirectly by
osmotic effects, and the cornea exploits charge in a variety of remarkable ways. One of
those, we believe, is the way in which charge is used for transparency. The transparency
of the cornea requires individual collagen fibrils to be maintained in a quasi-regular lattice
with short-range order. The origin and nature of forces that must necessarily act on fibrils to
maintain the stability of the lattice have long been the subject of speculation. In this chapter,
we apply the proposed thermodynamic theory of stromal osmotic pressure to describe how
restoring forces can arise from GAG-based osmotic and electrostatic considerations in a
1
PGs found in adult stroma include decorin, lumican, keratocan, and mimecan.
2
Common stromal GAGs include keratan sulfate (KS), dermatan sulfate (DS) and chondroitin sulfate
(CS)
3
The anterior stroma is sealed by the tight boundaries of epithelial cellular layer, making it nearly im-
permeable to water (although permeable to O2 and CO2 ) [10].

2
manner that maintains the lattice even when strong random variations in GAG distributions
are considered.
A number of models for corneal swelling have previously been presented. Notable is the
non-equilibrium model of Klyce and Russell [11], based on the phenomenological membrane
transport theory of Kedem-Katchalsky [12], and the steady-state model of Bryant [13], based
on the triphasic theory of Lai et al. [14]. These works and their extensions, for example
[15, 16, 17], have modeled the stroma as one-dimensional coupled flow of solvent (water) and
ionic solutes across the corneal thickness.
In this chapter, a different modeling approach, based on characterizing the free energy
of the stromal polyelectrolyte gel, will be introduced. The approach is intrinsically three-
dimensional, provides an explicit expression for the stromal osmotic pressure, accounts for
the nanoscale spatial distribution of GAG-based fixed charge, and is well-suited to implemen-
tation in finite element codes. It is shown that steady state active endothelial ion transport
reduces stromal ionic concentrations, which then occur in a modified Boltzmann distribution.
This leads to a modification of the osmotic pressure and stromal hydration, manifesting the
macroscopic effects of the pump-leak mechanism.
In Sect. 2, the stromal fixed charge distribution is modeled. In Sect. 3, the stromal
free energies are identified and characterized. In Sect. 4, the ex vivo cornea (no metabolic
activity) is analyzed and the model is applied to stromal swelling pressure and an investi-
gation of the stability of the collagen fibril lattice underlying transparency. In Sect. 5, the
theory is extended to the in vivo cornea including active endothelial ion transport and the
pump-leak mechanism of corneal hydration. In Sect. 6 the model is used to investigate the
biomechanics of Fuch’s dystrophy in which endothelial cells have compromised ion pumping
capacity and to examine the effects of LASIK on fluid pressure in the stroma.

2 Glycosaminoglycan-based charge in the stroma


2.1 Corneal hydration measures
Before characterizing the nature of charge distributions in the stroma, it is first necessary to
establish a suitable definition for the level of tissue hydration. A common definition, denoted
Hw , is water weight per unit dry (collagen) weight. However, this definition is not convenient
for our purpose since we will employ notions from continuum mechanics. An alternative
measure is the tissue volume dilation J which is defined simply as the swollen volume of
the tissue divided by the normo-hydrated volume. Because we will employ the dilation J
throughout this work to signify hydration level, it useful to relate the two definitions.
By assuming that keratocytes have the mass density of water, the hydration Hw of a
sample of stromal tissue can be related to its volume dilation J by

(J − φr φcol )ρw
Hw (J) = , (2.1)
φr φcol ρcol

where ρw = 1 g/cm3 and ρcol = 1.36 g/cm3 are the mass density of the water and dry
collagen fibrils [18], respectively, and φcol = 0.249 is the collagen fibril volume fraction at
normal hydration [3]. The volume fraction of collagen molecules within the fibrils (which

3
excludes intrafibrillar bound water) is denoted φr . [6, 19]. We take normal hydration to
be set at Hw = 3.2 [6] and J = 1. Using these conditions in (2.1), we infer that φr = 0.75.
Therefore, equation (2.1) may be simplified to
Hw (J) = 3.94J − 0.74 (2.2)
The linearity is fully consistent with measurements [20, 21]. A validation of this relationship
is provided by the fact that the inferred value of φr is in close agreement with the value of
0.77 reported by Goodfellow [19].
In the following subsections, we aim to characterize the magnitude and nanoscale distri-
bution of the GAG ionization charge in human stroma. This is an essential step in describing
the stromal polyelectrolyte gel.

2.2 Evidence for a glycosaminoglycan-based fibril coating


X-ray scattering studies under varying tissue hydration [2] suggest that some portion of
stromal PGs form a charge-rich and water-binding PG-coating surrounding each collagen
fibril. The radius of the coating has been measured [2] to be rc = 18.25 nm and this radius
is insensitive to hydration over a wide range. The existence of such a surface ultrastructure
on the collagen fibrils is corroborated by image studies from [22] and [23]. In addition, a
theoretical study on transparency by [24] proposed that collagen fibrils must be centered
in a transparent coating and the coated fibrils occupy approximately 60% of the matrix
volume, giving rc = 21.56 nm. Recent three-dimensional electron microscopy reconstructions
of corneal collagen and GAG chains [4] also suggests that some GAG chains are in close
association with the fibrils while the remaining GAG chains have random orientation in the
interfibrillar fluid.
A theoretical study by Cheng and Pinsky [25], based on comparing predicted stromal
swelling pressure to measurements, indicated that approximately 35% of stromal fixed charge
was in the non-swelling fibril coating region, and that 65% occupied the interstitial region
between coatings (Fig. 1). The study concluded that this charge arrangement was capable
of producing osmotic restoring forces that maintain the stability of the fibril lattice, and that
the coating region provided an effective interfibrillar electrostatic repulsion to prevent fibril
clustering.

2.3 Model for glycosaminoglycan charge concentration


The molecular structure of GAG chains is shown in Fig. 2. Each sulfate group carries one
negative charge per group. The KS disaccharides are sulphated at the 6−carbon position of
both the Gal and GlcNAc residues. The CS disaccharides are sulfated at either the 4− or
6−carbon position of the GalNAc residue and are designated CS4 or CS6, respectively (Fig.
2). However, a detailed FACE (fluorophore-assisted carbohydrate electrophoresis) analysis
[25, 26] shows that on average only 72.6% of all available GAG monosaccharides are actually
sulfated and contributing charge. If this ionization fraction is denoted fion , the stromal
average charge concentration in mM is given by,
eNg Lc fion
Cstroma = (2.3)
Fb
4
where e is the unit (negative) charge, Ng is the average number density (per unit volume) of
GAG chains with average contour length Lc , F is the Faraday constant, and b is the length
of a monosaccharide unit. The value of Cstroma has been measured indirectly [6, 27], and
the value used in this study, Cstroma = 38.6 mM, has been taken from [25].
The corneal stroma is composed of parallel arrays of collagen fibrils aligned with the
lamella direction and arranged with psuedo-hexagonal packing [4]. Based on the above noted
evidence, it is assumed that there exists a GAG-dense coating around each collagen fibril
surface. The simplest three-dimensional representative unit cell within a stromal lamella is
that of an equilateral triangular prism with vertices at the center of three fibrils (Fig. 1c).
The unit cell is divided into three regions: fibril, fibril coating and interstitial regions (Fig.
1d). Assuming that collagen fibrils are non-swelling and that keratocyte cells, which reside
between adjacent lamellae, dilate with the tissue dilation [3, 25], the normalized fixed charge
densities in the fibril coating and interstitial regions may be shown to be given by [25]
 
λ 1 − λ Cstroma
Ccoat (J) = + , (2.4)
Jφ0k − φf φc 2C0
 
λ Cstroma
Cinter (J) = 0
. (2.5)
Jφk − φf 2C0
In these expressions, J is the tissue macroscopic dilation, λ = 0.65 is the interstitial charge
partition parameter, φk , φf and φc denote the volume fractions for keratocytes, collagen
fibrils and coatings, respectively, and φ0k = 1 − φk . The normalization concentration is 2C0 ,
where C0 is the ionic concentration of the aqueous humor. For convenience, we combine the
above charge concentrations into a single expression as follows

Ccoat in the coating region
Ccell = (2.6)
Cinter in the interstitial region
The expressions for coating and interstitial charge concentrations take into account the fact
that GAG fixed charge cannot occupy the fibril or keratocyte volumes. Thus, when the
unit cell volume decreases, it does so only by reduction in interstitial fluid which causes
a rapid increase in the interstitial fixed charge concentration and osmotic pressure. This
volume-exclusion effect built into the charge model (2.4) and (2.5) is crucial for explaining
the strongly nonlinear variation of osmotic pressure with dilation in the cornea.
For future reference, it is noted that the volumes of the coating and interstitial regions
of the unit cell are given by
vcoat = φc v0 , (2.7)
vinter (J) = (Jφ0k − φf − φc ) v0 , (2.8)
where v0 is the volume of the normo-hydrated unit cell.

3 Basic polyelectrolyte theory for the stroma


3.1 Gibb’s free energy
From the point of view of electrochemistry, the stroma is a polyelectrolyte (polymeric elec-
trolyte) gel. It is in contact, via the endothelium, with the aqueous humor that therefore

5
acts as an ionic bath for the stromal electrolyte. In the living cornea, the pump-leak mech-
anism, alluded to in Sect. 1, has a strong influence on the tissue osmotic pressure. In this
section, however, we develop basic theory for isolated stroma without metabolic activity,
that is for in vitro tissue. Extension to the in vivo case is undertaken in Sect. 5
The stroma consists of a mixture of solid, fluid, and ionic phases, immersed in an ionic
bath (aqueous humor) at constant electrostatic potential φbath , hydrostatic pressure Pbath ,
and ionic concentration C0+ = C0− = C0 . Since no fixed charge exists in the bath, it is
reasonable to take φbath = 0. We now wish to describe the deformation of the unit cell,
introduced in Sect. 2, as the tissue undergoes a change in hydration. Let the reference
(normo-hydrated) and current (swollen) configurations of the unit cell be denoted Ω0 and Ω,
respectively, with volumes v0 and v, respectively. Using standard results from continuum me-
chanics, the motion is denoted ϕ (X) : Ω0 → Ω, the deformation gradient is F = ∂ϕ (X) /∂X,
and the displacement field is u (X) = ϕ (X) −√X. The Cauchy-Green deformation tensor is
defined by C = FT F and volume dilation J = det C (which is related to stromal hydration
through (2.2)).
It is next assumed that the Gibbs free energy of the unit cell electrolyte can be additively
decomposed [28] as follows,

W (φ, C) = Welectrolyte (φ, J) + Wbath (J) + Welastic (C)


Z
= [Welectrolyte (φ, J) + Wbath (J) + Welastic (C)] dΩ (3.1)
Ω0

where Welectrolyte , Wbath , and Welastic denote the free energy density (per unit reference
volume) of the electrolyte, bath solution and the elastic fibers of the gel, respectively, and φ
is the electrostatic potential.
A suitable form for Welectrolyte is given by the mean-field approximation [29] and is ex-
pressed as
     
F w 2
Welectrolyte (φ, J) = J zcell F Ccell (J) φ−2RT C0 cosh φ −1 − |∇φ| , (3.2)
RT 2

where zcell = −1 is the GAG-based fixed charge valence value, and R, T , F and w are the
gas constant, temperature, Faraday constant, and dielectric permittivity of the electrolyte
solvent, respectively. The first term on the right hand side measures the free energy of the
fixed charge, the second term measures the excess mean concentration of the mobile ions at
any point in the electrolyte compared to the bath and, after multiplication by RT, may be
interpreted as the osmotic work of introducing excess ions into the neighborhood of the fixed
charges Ccell , and the last term measures the dielectric free energy.
The bath free energy density, which measures the potential of the bath hydrostatic pres-
sure Pbath to do work associated with configuration volume change, by

Wbath (J) = −Pbath (J − 1) . (3.3)

For reasons of brevity, the elastic free energy Welastic appearing in (3.1), which corre-
sponds to the elasticity of the collagen fibrils and their three-dimensional organization, is
not discussed in this chapter but full details may be found in [30].

6
3.2 Thermodynamic equilibrium
In thermodynamic equilibrium, electrostatic equilibrium requires

∂W (φ, C)
=0 (3.4)
∂φ
C=constant

and mechanical equilibrium requires



∂W (φ, C)
=0 (3.5)
∂C
φ=constant

where the Gibb’s free energy W is given by (3.1). Condition (3.4) implies [31] that
  
2 F F
∇ φ= Ccell (J) + 2C0 sinh φ (3.6)
w RT
holds over Ω. This result can also be displayed in the form of the Poisson-Boltzmann equation
!
F X
− ∇2 φ = −Ccell (J) + zi Ci (φ) , (3.7)
w i=+,−

wherein the mobile ion concentrations Ci (moles per liter) satisfy the Boltzmann distribution
given by  
F
Ci (φ) = C0 exp −zi φ , i = +, − (3.8)
RT
and in which the binary ion valence numbers are z+ = 1 and z− = −1.
Condition (3.5) implies [31] that

div σ elastic − grad (Posmotic + Pbath ) = 0 (3.9)

holds over Ω. In this expression, σelastic is the effective Cauchy stress, Posmotic is the osmotic
pressure defined by
∂Welectrolyte
Posmotic = − , (3.10)
∂J
and Pbath is the bath (aqueous humor) hydrostatic pressure defined by
∂Wbath
Pbath = − . (3.11)
∂J
Finally, from equation (3.9), we may identify the stromal fluid pressure Pfluid as

Pfluid = Posmotic + Pbath (3.12)

This is a useful formulation because it provides an explicit constitutive equation for the
osmotic pressure (3.10). However, the Poisson-Boltzmann equation (3.7) must be solved over
the unit cell for the electrostatic potential φ that will be rapidly varying at the nanoscale.
To address this challenge and to arrive at a tractable theory, we modify the above approach
by employing the concept of energy-based homogenization.

7
3.3 Volume-averaged free energy
The idea is to employ a homogenization of the electrolyte free energy density based on its
volume average over the unit cell. This is given by
Z
1
Welectrolyte (φ, J) = Welectrolyte (φ, J) dΩ, (3.13)
v0 Ω0

where v0 is the volume of the normo-hydrated (reference) unit cell Ω0 . Now the homogenized
osmotic pressure is defined by

∂Welectrolyte
P osmotic = − . (3.14)
∂J
As noted above, the electrostatic potential φ must be solved from the Poisson-Boltzmann
equation (3.7). However, a good approximation for φ is given by the so-called Donnan
potential φ.
e The Donnan potential φe is defined to be that potential which satisfies the
Poisson-Boltzmann equation (3.7) when the fixed charge concentration is uniform (constant).
Since Ccell is piecewise constant over the coating region and interstitial region, the Donnan
potential will likewise be piecewise constant over the unit cell and is given by
RT
φecell (J) = − sinh−1 Ccell (3.15)
F
with (
φecoat in the coating region
φecell = (3.16)
φeinter in the interstitial region
The accuracy of this approximation has been confirmed by comparing the Donnan and true
electrostatic potentials for the unit cell [30]. The electrolyte free energy density is now found
by replacing φ with φecell in (3.2) resulting in
    
F e
Welectrolyte (J) = J −F Ccell φcell −2RT C0 cosh
e φcell −1 (3.17)
RT

Notice that Welectrolyte now depends only on J since Ccell (J) and φecell (J) . This result may
be introduced into (3.13) to arrive at the homogenized electrolyte free energy density
1
Welectrolyte (J) = [Wcoat vcoat + Winter vinter ] , (3.18)
v0
where vcoat and vinter are the reference volumes of the coating and interstitial regions and
given by (2.7) and (2.8), respectively. The homogenized osmotic pressure may then be
determined from (3.14) which yields
q  
0 2 φc −1 −1

P osmotic = 2φk RT C0 Cinter + 1 − 1 + Cinter sinh Ccoat − sinh Cinter .
Jφ0k − φf
(3.19)

8
Table 1: Parameters of the electrolyte model
Parameter Value
φf , volume fraction of the collagen fibrils (%) 24.9a
φk , volume fraction of the keratocytes (%) 11.2a
φc , volume fraction of the PG-dense coating (%) 13.3a
Cstroma , average charge density of the corneal stroma (mM) 38.6b
λ, the charge partition parameter 0.65c
Q, dimensionless parameter by the ionic pumping effect 0.965d , 1.0e
P0 , the hydrostatic pressure in bath solution (mmHg) 0e , 15.0d
C0 , bath concentration (mM) 150
T , temperature (K) 298
a
Values derived from [3, 25].
b
Calibrated value by [25].
c
Value estimated by [25].
d
Values for the in vivo corneas [30].
e
Value for in vitro corneas.

The first term in this expression describes a Donnan-like osmotic pressure from the GAGs
in the interstitial region and the second term describes the contribution from the GAGs
in the coating region. It is important to note that the model has no free parameters; the
osmotic pressure is expressed only in terms of experimentally determined quantities such as
collagen fibril and keratocyte volume fractions and stromal average charge concentration. In
the next section, we explore the application of the presented theory to in vitro swelling and
the self-organization of the fibril lattice.

4 The stroma in vitro


4.1 Swelling behavior
To obtain a validation of the model, the above osmotic theory was incorporated into a finite
element model of the cornea, including collagen fibrils aligned according to X-ray scattering
measurements [30]. The model was used to compare predicted osmotic pressure (given by
(3.19)) with experimental measurements of swelling pressure [20, 32]. Stromal samples, in the
form of 7 mm diameter disks, were immersed in an ionic bath of physiological concentration
C0 = 150 mM, and loaded by a porous piston across their thickness (Fig. 3a). The piston
load was systematically varied and the equilibrium sample thickness recorded against piston
load. The experiments were performed in confined [32] and unconfined [20] experimental
conditions. A finite element model was created to simulate the experimental setup, and the
model parameters are summarized in Table 1. The bath hydrostatic pressure Pbath is zero
in this case (and the fluid pressure in the stromal sample is purely osmotic pressure).
The stroma sample was meshed with 2,500, 27-node hexahedral elements. In order to
model confined and unconfined experimental conditions, edge boundary conditions were

9
taken as zero radial displacement or free, respectively (Fig. 3a). Sample thickness versus
osmotic pressure is shown in Fig. 3b. The predictions show good agreement to measurements
for the full range of sample hydration (Hw = 0.05 − 5.2). At the lowest hydration, the
osmotic pressure exceeds 2, 000 mmHg. Swelling pressure predictions for the confined and
unconfined cases are nearly identical. This is expected because lateral expansion of the
sample is restricted by the collagen fibrils and, consequently, stromal volume change in these
tests is due to changes in sample thickness [6]. This study provides confirmation of the
proposed model for the in vitro case.
The free swelling of the described cylindrical sample of in vitro stroma is modeled to
investigate the effect of the bath ionic concentration. Results for edge-confined free swelling
of the stroma sample are shown in Fig. 4, which depicts the ratio of swollen thickness
to original thickness (swelling ratio) versus bath ionic concentration C0 . For dilute bath
solutions, the model predicts that the stroma sample will swell to approximately four times
its original thickness, which may be compared to experimental measurements in deionized
water [33, 34] in which human corneas were observed to swell to approximately three times
their original thickness. For concentrated bath solutions, the abundance of ions results in
ionic shielding of the fixed charges, reduction in osmotic pressure and minimal swelling. Both
limiting states are captured by the swelling predictions.

4.2 Self-organization the collagen lattice and transparency


The stroma, which comprises about 90% of the corneal thickness, is organized into lamellae
(i.e. fibers) which contain collagen types I and V in the form of 25-30 nm diameter fibrils.
The collagen fibrils within a lamella appear in parallel arrays following the direction of the
lamella. In a lamella cross-section, the fibrils assume a quasi-regular packing arrangement
with center-to-center interfibrillar distance of 40 − 60 nm; this arrangement is referred to as
the fibril lattice. The short-range order of this lattice produces optical interference effects
that render the tissue transparent [35, 36, 37, 38].
GAGs have an important role in the maintenance of the fibril lattice. Keratan sulfate,
the predominant stromal GAG component, has been shown to be involved in modulating
the fibril lattice by the knockout of Chst5 in the mouse [39]. Scott [40] proposed that two
or more GAG chains, originating at different core proteins on neighboring fibrils, may form
an antiparallel duplexed association which appears as a bridge-like structure spanning the
interfibrillar distance in electron microscopy after staining, and that such structures may be
capable of controlling the interfibrillar spacing. The open question of how negatively charged
and mutually repulsive GAG chains might form durable antiparallel associations has been
discussed by Knupp et al. [41]. The idea of an entropic elastic interconnected fibril-GAG
network was employed in the theoretical study on lattice forces and spatial order by Farrell
and Hart [42] who proposed a model based on a six-fold symmetric elastic network. A six-fold
symmetric structural model was also proposed by Muller et al. [23] in which GAG bridges
were proposed to link next-nearest-neighboring fibrils. However, recent three-dimensional
electron tomography imaging [4, 41] suggests that GAG chains do not appear to exhibit any
approximation to a regular six-fold symmetric bridge network or any kind of regularity in
spatial organization. Cheng and Pinsky [25] modeled the GAGs as a random network of
worm-like chains and concluded that the entropic elasticity of these polymers falls far short

10
of what is needed to maintain the lattice. All of this suggests that fibrillar bridges cannot
mediate the lattice through a mechanical mechanism [4].
Maurice [1, 38] speculated that repulsive forces must exist between adjacent fibrils to keep
them separated and that the origin of these forces was related to the swelling force within
the extracellular matrix. Here we propose a mechanism for collagen fibril self-organization
based on osmotic pressure gradients that is robust with respect to the random organization
of GAG chains [25]. Consider an individual fibril in the lattice that is perturbed into to a
new position in the lattice while all other fibrils maintain their positions. A restoring force
must exist if the lattice is stable. The linear GAG chains are associated with each fibril
and attach to that fibril at one end only via the PG core protein. It is reasonable that the
attached GAG chains will displace with their supporting fibril. Because the longer chains
reach into the interfibrillar fluid, a local change in position of GAG charge concentration
will be created such that the concentration will increase in advance of the displacement and
reduce behind it (Fig. 5). The shorter fibril coating GAGs will not play any role in the
restoring force until the coatings become close and at which point electrostatic repulsion will
add an additional component.
The external work required to move the fibril from its lattice position r0 to perturbed
position r1 , while fixing the position of other fibrils in the lattice, is given by,
Z r1
Wext = ffibril (r) · dr. (4.1)
r0

In thermodynamic equilibrium, the force ffibril is given by


∂Welectrolyte
ffibril = − , (4.2)
∂r
which measures the energy cost of moving the fibril through the electrolyte. In order to
estimate this force, the fibril lattice is assumed to be perfectly hexagonal. The region between
the central (perturbed) fibril and its six neighbors is divided into six subcells with initial
volume Vk . The central fibril is then displaced from r0 to a new position r, causing a change
∆Vk in the volume of each subcell (Fig. 5). The GAG fixed charges within subcells are
assumed to be conserved during fibril displacement, which is consistent with the GAG-fibril
attachement. As the fibril moves into its peterbured position, the charge concentration in
each of the six lattice subcells can be estimated as a function of r and the force acting on
the fibril is then given by the chain rule as
6 6
X ∂Wkelectrolyte X
k ∂J k
ffibril = − = Posmotic (4.3)
k=1
∂r k=1
∂r

where the k−subcell osmotic pressure is given by

k
∂Wkelectrolyte
Posmotic =− (4.4)
∂J k
This result indicates that the local osmotic pressure variation between the subcells serves
as the source of an interfibrillar restoring force. It may be easily shown that the vector

11
∂Jk /∂r is directed toward the center of the lattice for all subcells [25]. Then equation (4.3)
describes a set of centrally oriented restoring forces. When a fibril is perturbed away from
its natural lattice position, the osmotic pressure gradient that results always produces a
restoring force oriented towards the natural lattice position. When the fibril occupies that
natural lattice position, the osmotic pressure is uniform and the restoring force vanishes. A
graphical interpretation of the origin of the lattice restoring force is provided in Fig. 5d.
Since the stromal free energy Welectrolyte has been modeled, it possible to develop esti-
mates for the fibril restoring force ffibril . Consider an ideal hexagonal cell with fibrils at each
vertex (Fig. 5a). The stroma is assumed to be normally hydrated and GAG fixed charge
concentrations are uniform and equal in each subcell. In this case, the fibril restoring force
magnitude may be computed from equation (4.3); the result is depicted in Fig. 6 for two
assumed fibril coating radius values of 18.25 nm and 21.56 nm. The force-fibril perturba-
tion relation exhibits two quasi-linear regimes with a transition where the coatings begin to
interact. For small perturbations, the osmotic-based restoring force is dominant, but as the
perturbation increases to the point where the PG-coatings begin to overlap, there is a marked
increase in repulsion force between the fibrils. This provides a mechanism preventing fibril
aggregation.
The lattice restoring force and its role in lattice self-organization was further studied
by using a transient dynamic model [25]. The model included randomness in the lattice
so that the fibrils are never arranged in perfect hexagonal symmetry. The goal was to
investigate whether the proposed osmotic restoring forces maintained the lattice in a dynamic
and random environment. The results were interpreted by computing the predicted radial
distribution function (RDF) of the fibrils, which is a measure of the order of a lattice [43].
RDF results from the dynamic simulation were shown to closely match measurements based
on electron micrographs [43] and indicated short-range order in the lattice with correlations
between any two fibrils becoming nearly random after a distance of 200 nm (Fig. 7). The
average interfibrillar distance, which corresponds to the location of the first peak in the RDF,
was maintained at 53 nm in all random cases studied. These results support the proposed
osmotic theory for lattice self-organization.

5 The stroma in vivo


5.1 Active ion transport and hydration control
The corneal endothelium is a 4 µm thin cellular monolayer with a very high density of
mitochondria, indicating high metabolic activity, and is located on the posterior surface of
the cornea. It is permeable to water, metabolic species, including glucose and lactate ion,
and other salt ions. The corneal endothelium is responsible for maintaining the hydration
of the cornea through a pump-leak mechanism [8]. The pump function is provided by active
ion transport across the endothelium [9] that regulates the osmotic pressure of the stroma.
The leak function refers to the passive exchange of water across the endothelium through
water channels in response to the stromal osmotic pressure.
In this section we introduce a macroscopic model for the pump-leak mechanism and use
the model to quantitatively illustrate how stromal hydration is modulated. The model is

12
formulated in three steps. First, steady state (but non-equilibrium) conditions are used to
establish the jump in electrochemical potential across the endothelial layer as a function of
the steady active ion fluxes. Then thermodynamic equilibrium within the stroma is used to
establish a modified Boltzmann distribution that governs ionic concentrations throughout
the stroma. Finally, a modified free energy provides the stromal osmotic pressure and
describes its dependence on the endothelial active ion transport.
To assess the electrochemical potential jump across the endothelial layer, it is treated as
an ideal membrane in which the action of the ion pumps is represented by an independently
specified active steady anion flux Ja− and cation flux Ja+ , transporting ions out of the stroma
and into the aqueous humor [13]. Because the endothelium is thin compared to the stroma,
it is assumed that transport across the endothelium is one-dimensional. A coordinate z
is introduced with z = z0 at the endothelium-aqueous humor interface and z = z ∗ at the
endothelium-stroma interface. At steady-state, net ionic fluxes resulting from both active
and passive transport will vanish so that
dµi
Ji = −Li − Jai = 0, i = +, − (5.1)
dz
must hold across the endothelial layer thickness z ∈ [z0 , z ∗ ] . In (5.1), Li is the membrane
permeability for ionic species i, µi is the electrochemical potential, and Jai > 0 is the steady
active ionic flux across the endothelium. Integrating (5.1) across the endothelium results in
the jump condition for the electrochemical potential
Jai
µ∗i − µ0i = −h , (5.2)
Li
where h = z ∗ − z0 is the thickness of the endothelial layer, and µ∗i and µ0i are the ionic
electrochemical potentials at z = z ∗ and z = z0 , respectively. Since no active transport is
taking place within the stroma, thermodynamic equilibrium requires µi = µ∗i where µi is the
electrochemical potential at any point in the stroma. Noting that
RT Fφ
µi = µref
i + ln Ci + zi , i = +, − (5.3)
Mi Mi
where M+ and M− are the atomic weights for anions and cations, respectively, results in
RT Ci Fφ
µi − µ0i = ln + zi . (5.4)
Mi C0 Mi
Substituting (5.4) in (5.2) and rearranging gives
   
h Mi Jai Fφ
Ci = C0 exp − exp −zi . (5.5)
RT Li RT

Observing that the first expression in parenthesis on the right hand side of (5.5) is dimen-
sionless, we define the membrane active transport factor Qi
 
h Mi Jai
Qi = exp − . i = +, − (5.6)
RT Li

13
Finally, stromal ionic concentration (5.5) may be written as
 

Ci = Qi C0 exp −zi , (5.7)
RT

which generalizes the Boltzmann distribution (3.8). Observe that 0 ≤ Qi ≤ 1 and the value
of Qi reduces with increasing active flux Jai . When no endothelial ionic pumping occurs,
Jai = 0 and Qi = 1. Equation (5.7) expresses the idea that when active ionic pumping
occurs, Qi < 1 and stromal ionic concentrations are reduced compared to when there is
no pumping and Qi = 1. Reduced stromal ionic concentrations produce reduced osmotic
(swelling) pressure, and vice versa. Simply stated, active ionic pumping directly reduces
the swelling tendency of the stroma. In this macroscopic description, the pump effect is
embedded in the membrane active transport factor Qi .
Based on (5.7), a modified Poisson-Boltzmann equation (see (3.7)) for the electrostatic
potential in a stromal unit cell is introduced as
 !
F X F φ
− ∇2 φ = −Ccell + zi Qi C0 exp −zi , (5.8)
w i=+,−
RT

The Donnan potential over the coating and interstitial regions are solved from (5.8) by
setting the right hand side to zero and are given as
s !
2
RT C cell C cell Q+
φecell (J) = ln − + + , (5.9)
F Q− Q2− Q−

where the normalized fixed charge concentration Ccell is given by (2.4-2.6). The electrolyte
free energy density is then
"   !#
 
e J = J −F Ccell φecell −RT C0
X F
Wcell φ, Qj exp −zj φecell −2 . (5.10)
j=+,−
RT

Introducing the Donnan potentials (5.9) into (5.10) and the result into (3.18) provides
the homogenized unit cell free energy Welectrolyte . Now employing (3.14) yields the modified
unit cell osmotic pressure
q  
0 2 φc
 −1 −1
P osmotic = 2φk RT C0 Cinter +Q−1 + g Ccoat − sinh
Cf2 sinh g Cinter ,
Jφ0k − φf
(5.11)
in which a modified inverse hyperbolic sine function was introduced
−1
 p 
sinh (x) = ln x + x + Q ,
g 2 (5.12)

and where Q = Q+ Q− is the membrane active transport parameter that describes the
combined effect of both the cation and anion active transport.
Eq. (5.11) shows that the osmotic pressure is dictated by the combined membrane active
transport parameter Q. Estimation of the combined membrane active transport parameter

14
Q is challenging. Based on imbibition pressure measurements in rabbit cornea, a value of
Qphysio = 0.965 has been established [30]. Stroma in vitro has no active ionic transport and
Q = 1. In this case, the above model reduces to the in vitro model described in Sect. 3.
In the subsections below we apply the in vivo osmotic theory to examine the effects of
reduced endothelial ion pumping (Fuch’s dystrophy) and the fluid pressure effects of LASIK.

5.2 Effects of reduced endothelial active ion transport


Fuch’s dystrophy is usually characterized by morphological changes in endothelial cells or by
an accelerated loss of endothelial cells [10, 44]. In this situation, the cornea will swell due
to increasing endothelial ionic permeability, decreasing active ion flux, or both mechanisms
simultaneously. This situation can be modeled by modifying the membrane active transport
parameter using (5.6), such that
 
RJ
ln Qpatho = ln Qphysio (5.13)
RL
where Qpatho is the pathological transport parameter, Qphysio = 0.965, RJ is the patholog-
ical/physiological fraction of ionic flux and RL is the pathological/physiological fraction of
ionic permeability [31].
A finite element model embedding the above in vivo osmotic theory and realistic collagen
organization was created [30] and used to analyze of a cornea with central corneal thickness
(CCT) of 520 µm, and anterior and posterior radii of 7.87 and 6.7 mm, respectively. Fig.
8a shows the predicted swollen CCT for 1 ≤ RJ ≤ 4.5 and RJ = 1 and Fig. 8b shows the
predicted swollen CCT for 0.2 ≤ RJ ≤ 1 and RL = 1. In both cases, the model predicts
corneal swelling, with predicted maximum swollen CCT of approximately 690 µm. This lies
in the range of clinical observations for Fuch’s dystrophy [45].
The model also predicts that swelling is concentrated in the posterior region, a prediction
that agrees with clinical observations [45] that the anterior surface of the cornea is nearly
normal among patients with Fuch’s dystrophy, whereas the posterior surface shows significant
change. Fig. 9 shows deformations due to swelling when RJ = 0.22. The results suggest
the stability of the anterior surface with respect to swelling resulting from the presence of
inclined lamellae.

5.3 Osmotic effects in LASIK


Laser in-Situ Keratomileusis (LASIK) is commonly used to correct refractive errors including
myopia and hyperopia. An anterior circular hinged flap of about 150 µm thickness is first
created using a femtosecond laser. The flap is then folded back to expose the stromal bed
which is reshaped by excimer laser ablation. The procedure is completed by returning the
flap to conform with the ablated profile. Although laser ablation creates a new central corneal
curvature with the goal of achieving emmetropia, the biomechanical response of the residual
(significantly thinned) stromal tissue is an important issue [46]. In a study [31], normal active
endothelium ionic pumping was assumed (Q = Qphysio ) and the osmotic model employed to
investigate the stromal fluid pressure and swelling resulting from a typical myopic correction.

15
The ablation depth profile was determined using Munnerlyn’s theory [47] for an intended
correction of 5 Diopter. The CCT was 520 µm, the flap thickness was 150 µm, and the
maximum ablation depth was 87 µm, resulting in a residual central stromal thickness of 283
µm. The flap is returned after ablation but it is not mechanically integrated with the stroma
[46].
The finite element model consisted of 7,008 27-node hexahedral elements. The solution
was obtained in two steps: (i) a pre-surgical step in which the normal cornea was loaded by
an IOP of 15 mmHg and the collagen prestress obtained, and (ii) a post-surgical step in which
the flap is separated and ablated tissue removed. Predicted curvature change, postsurgical
corneal radius and spherical equivalent agreed well with clinical measurements by Ortiz et
al. [48] on 85 eyes with myopia or myopic astigmatism [30].
As noted in Sect. 5.1, the active transport parameter Qphysio was calibrated based on
direct experimental measurement of imbibition pressure in in vivo rabbit cornea [49]. Those
measurements imply that in the normal in vivo (rabbit) cornea, the osmotic pressure is
approximately zero and the fluid pressure is approximately IOP [30]. This normal physio-
logical condition is modeled with the active ion transport parameter Qphysio = 0.965 [30]. In
this case, the normal pre-surgical fluid pressure is approximately Pfluid = IOP throughout
the stroma.
Recalling that the stromal fluid pressure is given by Pfluid = Pbath + Posmotic (see equation
(3.12)), where Pbath = IOP, the post-surgical fluid pressure is determined from Pfluid =
IOP +Posmotic , and where Posmotic could be positive or negative. The predicted fluid pressure
Pfluid for the LASIK model is shown in Fig. 10a. It was observed that the osmotic pressure
becomes negative and the stromal fluid pressure reduces to approximately half of its normal
value in the ablation region. The change in stromal fluid pressure resulting from the surgical
procedure may be inferred from Fig. 10a. Since the pre-surgical stromal fluid pressure is
Pfluid = IOP , the change in stromal fluid pressure can be observed by simply subtracting
IOP = 15 mmHg from the pressure scale in Fig. 10a. For example, at the center of the
ablation zone, the pre-surgical fluid pressure is 15 mmHg and the post-surgical fluid pressure
is 4.1 mmHg, which represents a change of −10.9 mmHg.
The predicted volume dilation J for the LASIK model is shown in Fig. 10b. Modest
swelling is indicated in Fig. 10b and is concentrated in the anterior region of the residual
stroma, particularly where the ablation is deepest and lamella inclination is lost. However,
if IOP increases, then the swelling increases significantly [30].

6 Concluding remarks
This chapter has presented an energy-based theory for stromal swelling and fluid pressure
and an application of the theory to demonstrate that GAG-based osmotic forces are capable
of stabilizing the collagen lattice, necessary for the transparency of the cornea. The model,
which has been validated against stromal swelling experiments, has no ”free parameters” and
the physical constants which do appear have been found through independent measurements.
The precise quantification of the fluid and osmotic pressures for the in vivo human cornea
remains an open question, as direct measurements are not currently feasible. Likewise,
measurement of the active transport parameter Qphysio , (see expression (5.6)), which depends

16
on the endothelial ionic permeability and active ionic flux, cannot be directly measured.
We have employed experimental measurements of imbibition pressure in rabbit to assign a
value to Qphysio [30], but this is not entirely satisfactory. Our investigations suggest that
stromal fluid pressure is close to IOP and osmotic pressure is maintained close to zero by the
endothelial ion pumps. But a final accounting requires more experimental data; the general
theory presented should, however, remain relevant.
The role of stromal collagen in swelling must also be considered for a full understanding of
how the cornea responds to swelling forces. This important topic has not been included in the
scope of this chapter, however, we remark about one interesting aspect. The anterior stroma
terminates at Bowman’s layer, which supports the epithelium. In and below Bowman’s, the
stromal fluid is pressurized whereas above Bowman’s the fluid pressure is zero (atmospheric
pressure can be cancelled out). How is the pressure difference equilibrated? This must be
through lamella inclination, as illustrated in Fig. 11. In the posterior stroma, the anterior
chamber IOP and stromal fluid pressure must balance and no lamella inclination is needed
(or exists). It may be concluded that anterior lamella inclination stabilizes the cornea’s
refractive surface. A detailed study of this problem has been conducted by Cheng et al. [30].
Another aspect of the model not discussed is the convenience with which it may be
implemented in finite element codes. The finite element method is an ”energy” method.
The proposed model is also an energy method and the electrolyte free energy can be readily
exploited in a finite element code. This is a significant advantage compared to all previous
models which are based on balance principles for fluxes of solvents and solutes. A detailed
discussion of the finite element implementation has been given by Pinsky and Cheng [31].

References
[1] Maurice DM. 1962 Clinical physiology of the cornea. Int. Ophthalmol. Clin. 2, 561-572.

[2] Fratzl P, Daxer A. 1993 Structural transformation of collagen fibrils in corneal stroma
during drying. Biophys. J. 64, 1210-1214.

[3] Meek KM. 2008 The cornea and sclera. In Collagen: structure and mechanics. (ed. P.
Fratzl), pp. 359-396. Springer.

[4] Lewis PN, Pinali C, Young RD, Meek KM, Quantock AJ, Knupp C. 2010 Structural
interactions between collagen and proteoglycans are elucidated by three-dimensional
electron tomography of bovine cornea. Structure 18, 239-245.

[5] Scott JE. 1988 Proteoglycan-fibrillar collagen interactions. Biochem. J. 252, 313-323.

[6] Elliott GF, Hodson SA. 1998 Cornea, and the swelling of polyelectrolyte gels of biological
interest. Rep. Prog. Phys. 61, 1325-1365.

[7] Maurice DM. 1972 The location of the fluid pump in the cornea. J. Physiol. 221, 43-54.

[8] Fischbarg J, Maurice DM. 2004 An update on corneal hydration control. Exp. Eye. Res.
78, 537-541

17
[9] Bonanno JA. 2012 Molecular mechanisms underlying the corneal endothelial pump.
Exp. Eye. Res. 95, 2-7.

[10] Dawson DG, Ubels JL, Edelhauser HF. 2011 Cornea and Sclera. In Levin LA et al. (ed)
Adler’s physiology OF the eye, 11th edn. Elsevier, pp 71-130

[11] Klyce SD, Russell SR. 1979 Numerical solution of coupled transport equations applied
to corneal hydration dynamics. J. Physiol. 292, 107-34.

[12] Kedem, O, Katchalsky, A, 1958 Thermodynamic analysis of the permeability of biolog-


ical membranes to non-electrolytes. Biochim. Biophys. Acta. 27, 229-246.

[13] Bryant MR, McDonnell PJ. 1998 A triphasic analysis of corneal swelling and hydration
control. Trans. ASME 120, 370-381.

[14] Lai WM, Hou JS, Mow VC. 1991 A triphasic theory for swelling and deformation
behaviors of articular cartilage. J. Biomech. Eng. 113, 245-258.

[15] Ruberti, JW, Klyce, SD. 2003 NaCl osmotic perturbation can modulate hydration con-
trol in rabbit cornea. Exp. Eye. Res. 76, 349-359.

[16] Li LY, Tighe B. 2006 Numerical simulation of corneal transport processes. J. R. Soc.
Interface 3, 303-310

[17] Leung BK, Bonanno JA, Radke CJ. 2011 Oxygen-deficient metabolism and corneal
edema following epithelial hypoxia in the rabbit. Prog. Retin. Eye. Res. 30, 471-492.

[18] Meek KM, Fullwood NJ, Cooke PH, Elliott GF, Maurice DM, Quantock AJ, Wall
RS, Worthington, CR. 1991 Synchrotron x-ray diffraction studies of the cornea, with
implications for stromal hydration. Biophys. J. 60, 467-474.

[19] Goodfellow JM, Elliot GF, Woolgar AE. 1978 X-ray diffraction studies of the corneal
stroma. J. Mol. Biol. 119, 237-252.

[20] Olsen T, Sperling S. 1987 The swelling pressure of the human corneal stroma as deter-
mined by a new method. Exp. Eye. Res. 44, 481-490.

[21] Hedbys BO, Mishima S. 1966 The thickness-hydration relationship of cornea. Exp. Eye.
Res. 5, 221-228

[22] Miyagawa A, Kobayashi M, Fujita Y, Hamdy O, Hirano K, Nakamura M, Miyake Y.


2001 Surface ultrastructure of collagen fibrils and their association with proteoglycans in
human cornea and sclera by atomic force microscopy and energy-filtering transmission
electron microscopy. Cornea 20, 651-656.

[23] Muller LJ, Pels E, Schurmans LR, Vrensen GF. 2004 A new three-dimensional model
of the organization of proteoglycans and collagen fibrils in the human corneal stroma.
Exp. Eye. Res. 78, 493-501.

18
[24] Twersky V. 1975 Transparency of pair-correlated, random distributions of small scat-
terers, with applications to the cornea. J. Opt. Soc. Am. 65, 524-530.
[25] Cheng X, Pinsky PM. 2013 Mechanisms of self-organization for the collagen fibril lattice
in the human cornea. J. R. Soc. Interface 10, 20130512.
[26] Plaas AH, West LA, Thonar EJ, Karcioglu ZA, Smith CJ, Klintworth GK, Hascall
VC. 2001 Altered fine structures of corneal and skeletal keratan sulfate and chon-
droitin/dermatan sulfate in macular corneal dystrophy. J. Biol. Chem. 276, 39788-
39796.
[27] Hodson SA. 1971 Why the cornea swells. J. Theor. Biol. 33, 419-427.
[28] Katchalsky A, Michaeli I. 1955 Polyelectrolyte gels in salt solutions. J. Polym. Sci. 15,
69-86.
[29] Sharp KA, Honig B. 1990 Calculating total electrostatic energies with the nonlinear
Poisson-Boltzmann equation. J. Phys. Chem. 94, 7684-7692.
[30] Cheng X, Petsche SJ, Pinsky PM. 2015 A structural model for the in vivo human cornea
including collagen-swelling interaction. J. R. Soc. Interface 12, 20150241.
[31] Pinsky PM and Cheng X. 2017 A constitutive model for swelling pressure and volumetric
behavior of highly-hydrated connective tissue. J. Elasticity, DOI: 10.1007/s10659-016-
9616-z
[32] Hedbys BO, Dohlman CH. 1963 A new method for the determination of the swelling
pressure of the corneal stroma in vitro. Exp. Eye. Res. 2, 122-129.
[33] Meek KM, Leonard DW, Connon CJ, Dennis S, Khan S. 2003 Transparency, swelling
and scarring in the corneal stroma. Eye 17, 927-936.
[34] Muller LJ, Pels E, Vrensen GF. 2001 The specific architecture of the anterior stroma
accounts for maintenance of corneal curvature. Br. J. Ophthalmol. 85, 437-443.
[35] Benedek GB. 1971 Theory of transparency of the eye. Appl. Opt. 10, 459-473.
[36] Cox JL, Farrell RA, Hart RW, Langham ME. 1970 The transparency of the mammalian
cornea. J. Physiol. 210, 601-616.
[37] Kostyuk O, Nalovina O, Mubard TM, Regini JW, Meek KM, Quantock AJ, Elliott GF,
Hodson SA. 2002 Transparency of the bovine corneal stroma at physiological hydration
and its dependence on concentration of the ambient anion. J. Physiol. 543, 633-642.
[38] Maurice DM. 1957 The structure and transparency of the cornea. J. Physiol. 136, 263-
286.
[39] Hayashida Y, Akama TO, Beecher N, Lewis P, Young RD, Meek KM, Kerr B, Hughes
CE, Caterson B, Tanigami A, Nakayama J, Fukada MN, Tano Y, Nishida K, Quantock
AJ. 2006 Matrix morphogenesis in cornea is mediated by the modification of keratan
sulfate by GlcNAc 6-O-sulfotransferase. Proc. Natl. Acad. Sci. USA 103, 13333-13338.

19
[40] Scott JE (1992) Morphometry of cupromeronic blue-stained proteoglycan molecules
in animal corneas, versus that of purified proteoglycans stained in vitro, implies that
tertiary structures contribute to corneal ultrastructure. J. Anat. 180, 155-164.

[41] Knupp C, Pinali C, Lewis PN, Parfitt GJ, Young RD, Meek KM, Quantock AJ. 2009
The architecture of the cornea and structural basis of its transparency. Adv. Protein.
Chem. Struct. Biol. 78, 25-49.

[42] Farrell RA, Hart RW. 1969 On the theory of the spatial organization of macromolecules
in connective tissue. Bull. Math. Biophys. 31, 727-760.

[43] Freund DE, McCally RL, Farrell RA, Cristol SM, L’Hernault NL, Edelhauser HF. 1995
Ultrastructure in anterior and posterior stroma of perfused human and rabbit corneas.
Relation to transparency Invest. Ophthalmol. Vis. Sci. 36, 1508-1523.

[44] Elhalis H, Azizi B, Jurkunas U. 2010 Fuchs endothelial corneal dystrophy. The Ocular
Surface 8, 173-184.

[45] Brunette I, Sherknies D, Terry MA, Chagnon M, Bourges JL, Meunier. 2009 3-D char-
acterization of the corneal shape in fuchs dystrophy and pseudophakic keratopathy.
Invest. Ophthalmol. Vis. Sci. 52, 206-214.

[46] Dupps WJ, Wilson SE. 2009 Biomechanics and wound healing in the cornea. Exp. Eye.
Res. 83, 709-720

[47] Munnerlyn CR, Koons SJ, Marshall J. 1988 Photorefractive keratectomy: a technique
for laser refractive surgery. J. Cataract. Refract. Surg. 14, 46-52

[48] Ortiz D, Alio JL, Pinero D. 2008 Measurement of corneal curvature change after me-
chanical laser in situ keratomileusis flap creation and femtosecond laser flap creation.
J. Cataract. Refract. Surg. 34, 238-242

[49] Hedbys BO, Mishima S, Maurice DM. 1963 The imbibition pressure of the corneal
stroma. Exp. Eye Res. 2, 99-111.

20
(a)
Epithelium

Stroma

IOP = 15 mmHg

Endothelium

(b) (c)

Figure 1: (a) The cornea is a polyelectrolyte gel loaded by the IOP. Stromal lamellae follow
the corneal curvature except in the anterior region where lamellae exhibit inclination and
insert into Bowman’s layer under the epithelium. (b) An illustration of the organization of
the corneal stroma showing several lamellae and a keratocyte cell. Collagen fibrils within
lamellae form a lattice with short range order. (c) Cross-section of a unit cell representing
the hexagonal collagen fibril lattice. The GAG-based coating region around fibrils and the
interstitial GAG chains are illustrated. (d) The unit cell is a triangular prism (shown in
cross-section); the coating and interstitial regions are indicated.

21
Figure 2: Disaccharide repeat units of KS, DS, CS4 and CS6.

Figure 3: (a) Illustration of confined and unconfined swelling pressure experiments [20, 32].
In both experiments, a porous piston loads the cornea in the transverse direction. (b)
Comparison of predicted and experimentally measured [20, 32] values of swelling pressure Ps
versus corneal thickness t.

22
5

3
t m /t 0

0
10 -3 10 -1 10 1 10 3 10 5
C (mM)
0

Figure 4: Predicted free swelling of stroma in an ionic bath ionic vs. bath concentration C0 ;
the swelling ratio is swollen thickness tm divided by original thickness t0 .

23
Figure 5: The GAG-based osmotic restoring force mechanism acts on fibrils to stabilize the
lattice. (a) Shows a set of fibrils in a regular lattice arrangement with their associated GAG
chains forming fibril coatings and interstitial projections. The distribution of mobile ions
is also illustrated. (b) Depicts the osmotic pressure in each subcell around the undisturbed
fibril. Because the osmotic pressure is uniform, the net restoring force is zero (indicated by
the red dot). (c) Shows a single fibril perturbed from its lattice position. The GAG chains
attached to the fibril move with the fibril and the GAG fixed charge concentration increases
in advance of the fibril motion and reduces behind it. (d) Indicates the resulting osmotic
pressure in each of the subcells, which is higher where GAG fixed charge concentration has
increased and lower where it has reduced. The forces exerted by each of the subcells on the
fibril act to restore the fibril towards its regular lattice position (indicated by the red arrow).
This mechanism operates regardless of the direction of the perturbation.

24
(a) (b)
-3 r c = 18.25 nm r c = 21.56 nm
6
×10 ×10 -3
2.5
λ = 0.3 λ = 0.3
λ = 0.65 λ = 0.65
5 λ = 1.0 λ = 1.0
2

4
f fibril (N/m)

(N/m)
1.5
3

fibril
1

f
2

1 0.5

0 0
0 5 10 15 20 0 5 10 15
r (nm) r (nm)

Figure 6: Electrostatic restoring force magnitude on a horizontally displaced fibril in the


ideal hexagonal cell for two coating radius values: (a) rc = 18.25 nm, and (b) 21.56 nm.

2.5 Prediction
Freund et al (1995)

1.5
g(r)

0.5

0
0 50 100 150 200
r (nm)

Figure 7: Comparison of the predicted RDF, denoted g(r), from the fibril dynamics model
with the measured RDF from an electron micrograph of human cornea [43].

25
(a)
700 700
680 680
660 660
640 640
CCT (µ m)

CCT (µ m)
620 620
600 600
580 580
560 560
540 540
520 520
1 1.5 2 2.5 3 3.5 4 4.5 0.2 0.4 0.6 0.8 1
RL R
J

Figure 8: The predicted swollen CCT in Fuch’s dystrophy when (a) the endothelial ionic
permeability is increased such that 1 ≤ RL ≤ 4.5 and with normal active ionic flux RJ = 1,
and (b) the active ionic flux rate is reduced such that 0.2 ≤ RJ ≤ 1 and with normal anionic
permeability RL = 1.

Figure 9: Fringe plot of vertical displacements of a cornea with simulated Fuch’s dystrophy.
Ion pumping is reduced to 22% of normal (RJ = 0.22). Most swelling occurs in the posterior
stroma.

26
Figure 10: Fringe plots of (a) the fluid pressure and (b) volume dilation J of a postsurgical
cornea after LASIK designed for an intended correction of 5 Diopter.

27
Figure 11: At the anterior surface, inclined lamellae insert into Bowman’s layer and act
as anchors resisting the stromal fluid pressure. In the posterior stroma, the stromal and
anterior chamber fluid pressures are balanced and the collagen does not need to be inclined.
This arrangement stabilizes the refractive surface.

28

También podría gustarte