“Molecular Mechanisms Underlying the Corneal Endothelial Pump”

Abstract

The corneal endothelium is responsible for maintaining the hydration of the cornea. This is through a “Pump-Leak” mechanism where the active transport properties of the endothelium represent the “Pump” and the stromal swelling pressure represents the “Leak”. For the “Pump”, Na⁺,K⁺ ATPase activity and the presence of HCO₃⁻, Cl⁻,and carbonic anhydrase activity are required. Several basolateral (stromal side) anion transporters, apical (facing the aqueous humor) ion channels and water channels have been identified that could support a model for ion secretion as the basis for the endothelial pump, however evidence of sustained anion fluxes, osmotic gradients or the need for water channels is lacking. This has prompted consideration of other models, such as Electro-osmosis, and consideration of metabolite flux as components of the endothelial pump. Although the conditions under which the “Pump” is supported are known, a complete model of the endothelial “Pump” has yet to emerge.

1. Introduction

The topic of molecular mechanisms underlying the corneal endothelial pump was extensively reviewed in 2003 (Bonanno 2003). The purpose here is to summarize the important features of the endothelial pump and provide an update on recent significant findings. Development of new techniques, e.g. siRNA knockdown and in vivo viral transfection of shRNA, has produced more specific testing of pump features. Similarly, the use of knockout models (e.g., AQP1) has prompted significant re-thinking of the nature of solute-fluid coupling. These molecular approaches together with new studies using customary pharmacological approaches for studying corneal endothelial fluid transport have led to re-evaluations of the corneal endothelial pump.

2. Background

The cornea is the major refractive element of the eye. This requires optical transparency and smooth curved surfaces. The cornea has five layers: the outer epithelium, Bowman’s (basement) membrane, the stroma, Descemet’s (basement) membrane, and the inner surface endothelium. The smooth regular surface, tight packing of cells, relative paucity of organelles (especially mitochondria), and lack of blood vessels in the stratified squamous corneal epithelium (~50 μm) reduce light scatter thereby contributing to transparency. The corneal stroma reduces light scatter by its regular spacingamong collagen fibers and uniform diameter of fibers. This contrasts with the sclera where spacing and fiber diameters have a very broad distribution. The regular spacing of the stromal fibers is controlled by the hydrophilic glycosaminoglycan (GAGs) ground substance. When the hydration of the stroma is ~3.5 mg H₂O/mg dry tissue or less, the stroma is relatively transparent. A slit-lamp view of the stroma however, indicates that it scatters more light than the epithelium. This is due to the thin flat keratocytes that lie between stromal lamellae. Keratocytes are responsible for producing collagen and GAGs (among many other extracellular substances) and will be activated during trauma to repair the stroma. Repair alters the expression profile of keratocytes, which can lead to increased light scatter. In contrast, the corneal endothelium is a thin (4 μm) confluent monolayer that has a very high density of mitochondria, but because of its extremely short optical pathlength, scatters very little light.

The epithelium provides the barrier to the outside world, the stroma provides the refractive shape and the endothelium maintains the nutrition of the corneal cells and the hydration of the stroma.

Except for oxygen, all the nutrients for the cornea come from the aqueous humor and through the endothelium. For example, glucose transporters are present on both the apical (aqueous humor side) and basolateral (stromal side) endothelial cell membranes to allow transcellular glucose flux (Kumagai, Glasgow et al. 1994). Moreover, 85% of the glucose consumed by the cornea is converted to lactate, which diffuses posteriorly across the endothelium (Riley 1969). The corneal endothelium also maintains the hydration of the stroma through active transport mechanisms. Because of the swelling pressure (~60 mmHg) exerted by molecular repulsion from the highly negatively chargedstromal GAGs, bare stroma can swell to many times its normal thickness (see (Kwok and Klyce 1990) for further details) and as a consequence the spacing between fibers becomes non-uniform, light scatter increases, and corneal transparency is lost. This tendency to swell is counteracted by the endothelial pump through the membrane active transport mechanisms. Steady-state hydration occurs when the endothelial pump rate equals the GAG driven leak. Maurice (Maurice 1981) called this the “Pump-Leak” mechanism for maintenance of corneal hydration and transparency. Because of the presence of the continuous leak, loss of endothelial ion transport activity leads to corneal edema, loss of transparency, and impaired vision.

3. Corneal Endothelial Pump Description

The endothelial pump function is best demonstrated using rabbit corneas mounted in modified Ussing chambers in vitro. If carefully mounted and perfused with appropriate media, the cornea will maintain its thickness for several hours. Another approach is to remove the epithelium, expose the anterior stroma for a short period to Ringer’s solution allowing the stroma to swell, remove the anterior solution and replace it with silicone oil. The corneal thickness will then slowly decrease exponentially in a process called deturgescence. From these experiments it was determined that the cornea will swell, or not deturgesce, if the endothelium is exposed to the cardiac glycoside ouabain, an inhibitor of the Na⁺,K⁺ ATPase, indicating that the endothelial pump is dependent on primary active transport. Removal of HCO₃⁻ from the endothelial perfusing solution had a similar inhibitory effect on the pump and addition of carbonic anhydrase inhibitors slowed the pump by ~30%,indicating a major role for HCO₃⁻(Dikstein and Maurice 1972; Fischbarg and Lim 1974; Hodson and Miller 1976; Riley, Winkler et al. 1995). Because Cl⁻ removal did not cause initial swelling, the early pump model was described as an exclusiveHCO₃⁻ secretory mechanism. Subsequent studies indicated that Cl⁻ was important for maintaining the pump (Winkler, Riley et al. 1992), suggesting that an anion exchanger, (Cl⁻/HCO₃⁻) played a role. Consistent with an anion transport mechanism, anion transport blockers, e.g. DIDS (4,4′-Diisothiocyano-2,2′-stilbenedisulfonic acid), significantly slow the pump(Kuang, Li et al. 2004; Riley, Winkler et al. 1996). There must also be a role for secondary Na⁺ fluxes because amiloride, which blocks the Na⁺/H⁺ exchanger and at high concentrations the Na⁺ channel ENaC (Epithelial Sodium Channel), also slows fluid transport(Liebovitch and Fischbarg 1982). Assembling these disparate observations into a clear model for endothelial pump function has been very challenging and is described below.

4. The Anion (Bicarbonate& Chloride) Secretion Model

Fluid secretion that is coupled to ion fluxes is dependent on active transport mechanisms to produce local osmotic gradients that move water across the cellular layer. This is best described for secretory glands and kidney reabsorption processes. More recently, evidence for direct coupling of water to ion fluxes in cotransporters has also been presented (Hamann, Kiilgaard et al. 2003; Meinild, Loo et al. 2000) and could have a significant role in fluid absorption across the intestinal mucosa (Loo, Zeuthen et al. 1996).The best tissue models for bicarbonate secretion are the gall bladder and pancreas. In both tissues transporters and anion channels have been identified in both basolateral and apical membranes that could provide the net transcellular bicarbonate flux that can unequivocally be measured. In contrast, several groups have measured HCO₃⁻ and Cl⁻ fluxes across the rabbit corneal endothelium. However, there is no consensus that a net anion flux can be generated (Bonanno 2003). Moreover, the basolateral membrane HCO₃⁻ permeability of the corneal endothelium is significantly higher than the apical permeability (Bonanno, Guan et al. 1999) and with the exception of apical anion channels; no apical HCO₃⁻ transporter has been identified.

Figure 1 shows a possible bicarbonate transport model for the corneal endothelium. The basolateral Na⁺,K⁺ ATPasecreates a low intracellular [Na⁺] and high intracellular [K⁺], and in conjunction with K⁺ channels a negative membrane potential of about −55 mV(Watsky and Rae 1991). On the basolateral side there are two bicarbonate transporters, a 1Na⁺:2HCO₃⁻ cotransporter (SLC4A4,NBCe1; Sodium Bicarbonate Cotransporter electrogenic) and the Cl⁻/HCO₃⁻ exchanger (SLC4A2, AE2; Anion Exchanger),as well as a Na⁺/H⁺exchanger (SLC9A6, NHE1; Sodium Hydrogen Exchanger). NBCe1 can directly load HCO₃⁻ into the cell. The Na⁺/H⁺exchanger can indirectly load HCO₃⁻ into the cell because removal of protons favors formation of HCO₃⁻ from CO₂, catalyzed by carbonic anhydrase II. Conversely, the Cl⁻/HCO₃⁻ exchanger is poised to remove HCO₃⁻ from the cell and add Cl⁻. Moreover, a basolateral Na⁺:K⁺:2Cl⁻ cotransporter helps load intracellular chloride to ~40 mM, above electrochemical equilibrium (~12 mM). On the apical side, at least two anion channels have been identified: CFTR (Cystic Fibrosis Transmembrane conductance Regulator) and CaCC (Calcium activated Chloride Channel). Anion permeability of both channels favors Cl⁻ vs HCO₃⁻by ~ 4:1.There is no evidence for apical Cl⁻/HCO₃⁻exchange. As such the model (see Figure 1) predicts that HCO₃⁻ is taken up on the basolateral side and efflux of HCO₃⁻ across the apical side is through anion channels driven by the negative membrane potential. Figure 1shows an additional potential indirect route for apical HCO₃⁻efflux. Because the influx mechanisms for HCO₃⁻ exceed direct efflux mechanisms and the differential must be converted to CO₂(Bonanno and Giasson 1992), the CO₂ can diffuse in any direction and some will diffuse across the apical membrane where carbonic anhydrase IV could catalyze the conversion back to HCO₃⁻ and thereby contribute to a local buildup of osmotic pressure at the apical membrane. The following sections discuss recent studies of several corneal endothelial transport mechanisms and their potential contribution to an anion secretion model.

Figure 1.

Bicarbonate Secretion Model for Endothelial Pump. Fluid coupled anion secretion requires transendothelial net flux of Cl⁻ and/or HCO₃⁻. The movement of net negative charge creates a small potential difference (0.5 mV, apical side negative) that attracts Na⁺through the paracellular pathway& across the tight junction (TJ). The net flux of NaHCO₃⁻and/or NaCl constitutes the osmotic driving force for water movement. HCO₃⁻ uptake on the basolateral membrane is through the actions of the 1Na⁺:2HCO₃⁻cotransporter (NBCe1) and the Na⁺/H⁺exchanger (NHE1). Cl⁻ uptake is primarily via the Na⁺:K⁺:2Cl⁻cotransporter (NKCC1) and the Cl⁻/HCO₃⁻exchanger (AE2). The high intracellular [Cl⁻]and[HCO₃⁻], together with the negative membrane potential can then drive anions across the apical membrane through anion selective channels. An additional route for net HCO₃⁻ flux is for the high intracellular [HCO₃⁻] to be converted to CO₂, facilitated by carbonic anhydrase II (CAII), apical CO₂ diffusion and conversion back to HCO₃⁻, facilitated by carbonic anhydrase IV (CAIV) at the apical surface. This pathway is less attractive because it does not contribute to the transendothelial potential.

4a. NKCC1

The role for Cl⁻ is unclear. Intracellular [Cl⁻] is about 40 mM in bovine endothelium (Srinivas, Bonanno et al. 1998), which is above electrochemical equilibrium. This is due to the presence of a bumetanide sensitive basolateral Na⁺:K⁺:2Cl⁻cotransporter (NKCC1) (Diecke, Zhu et al. 1998; Diecke, Wen et al. 2005; Jelamskii, Sun et al. 2000). Together with the apical Cl⁻ channels, the potential for Cl⁻ secretion exists. One study could not show any effect of the Na⁺:K⁺:2Cl⁻blocker bumetanide on rabbit corneal deturgescence (Riley, Winkler et al. 1997), however small, but significant effects were noted in a later study (Diecke, Wen et al. 2005). Certainly NKCC1 is involved in regulation of cell volume, however given the lack of clear evidence for a net transendothelial Cl⁻ flux,the role NKCC1 could have in the endothelial pump is uncertain. One suggestion is that Cl⁻ efflux through anion channels offsets the membrane hyperpolarizing effect of the electrogenic1Na⁺:2HCO₃⁻ cotransporter. By keeping the membrane potential relatively depolarized through Cl⁻efflux, the driving force for continual 1Na⁺:2HCO₃⁻ influx is maintained.

4 b. NBCe1

The dependence of endothelial fluid transport on active transport and the presence of bicarbonate suggest an important role for the 1Na⁺:2HCO₃⁻cotransporter (NBCe1). NBCe1 has two splice variants, pNBCe1 and kNBCe1, for pancreas and kidney type, respectively. The kidney proximal tubule form of NBC (kNBC) has a 1:3 stoichiometry, which favors cellular efflux of HCO₃⁻, and the pancreas form of NBC (pNBC) has a 1:2 stoichiometry, which favors HCO₃⁻influx. However, more recent studies have shown that the stoichiometry of either kNBC or pNBC can change depending on the cell type in which it is expressed(Gross, Hawkins et al. 2001). Previous reports indicated the pNBC variant was expressed in bovine(Sun, Bonanno et al. 2000), human(Sun and Bonanno 2003)and rat corneal endothelium(Bok, Schibler et al. 2001), and one report suggested both variants were expressed in human endothelium (Usui, Seki et al. 1999).Immunohistochemistry studies in cultured and fresh bovine, rat, and human endothelium indicate that NBCe1 exclusively locates to the basolateral membrane; however, one report (Diecke, Wen et al. 2004) suggests apical expression as well. NBCe1specific siRNA knockdown reduced NBCe1 expression by 80-90% four days post transfection in cultured bovine corneal endothelial cells (Li, Sun et al. 2005). This treatment significantly reduced basolateral HCO₃⁻permeability and also reduced the basolateral to apical HCO₃⁻flux induced by basolateral to apical [HCO₃⁻] gradient. Importantly, apical HCO₃⁻ permeability and apical to basolateral HCO₃ flux was unaffected by siRNA knockdown. This study shows that functional NBCe1 is only present on the basolateral membrane and that it is necessary for a large portion of basolateral HCO₃⁻ influx.

4c. Calcium Activated Chloride Channels

In bovine corneal endothelial cells, apical HCO₃⁻permeability can be enhanced by increasing [Ca²⁺]_i via either activation of purinergic receptors (e.g., by ATP) or inhibition of the sarcoendoplasmic reticulum Ca²⁺ -ATPase (SERCA) (Xie, Zhang et al. 2002). In either case there is a release of Ca²⁺ from intracellular stores with concomitant Ca²⁺ entry from the extracellular bath through store-operated channels in a process called capacitive calcium entry (CCE). CCE agonists enhanced apical HCO₃⁻ permeability in Ca²⁺-rich Ringer solution, but not in the absence of extracellular Ca²⁺, which was blocked by the CCE inhibitor2-aminoethoxydiphenyl borate (2-APB)(Zhang, Li et al. 2006; Zhang, Xie et al. 2002). ATP release from endothelial cells does occur and can be stimulated by injury or cell volume increase (Srinivas, Mutharasan et al. 2001) indicating that this pathway could be important during cellular stress.

A likely candidate for the apical Ca²⁺sensitive channel is CLCA (chloride channel, calcium activated)(Cunningham, Awayda et al. 1995).CLCA1 antiserum detected the 90kD band in corneal endothelium. Immunofluoresence staining showed an apical membrane location(Zhang, Li et al. 2006). In endothelial cells transfected with CLCA1 specific siRNA, CLCA1 expression was reduced by 80%. Activating CCE increased apical HCO₃⁻ permeability in siControl transfected cells, while having no effect in CLCA1 specific siRNA transfected cells. Baseline HCO₃⁻ permeability however, was not different between control and siRNA treated cells. These studies indicate that the calcium-activated chloride channel (CLCA1) is expressed in corneal endothelial cells and can contribute to Ca²⁺ dependent apical HCO₃⁻ permeability, but not resting permeability across the corneal endothelium.Since release of ATP activates the purinergic pathway, this channel may have a role during stress, e.g. inflammation, to enhance “Pump” activity.

4d. CFTR

Increasing intracellular [cAMP] by activating membrane adenylyl cyclases with forskolin stimulates an NPPB (5-nitro-2-(3-phenylpropylamin benzoic acid) and glibenclamide-inhibitable apical Cl⁻ and HCO₃⁻ permeabilitythat has been identified as CFTR (Cystic Fibrosis Transmembrane conductance Regulator)in bovine corneal endothelium (Sun and Bonanno 2002). Knocking down CFTR expression using siRNA eliminated forskolin induced increases in apical Cl⁻ and HCO₃⁻ permeability(Li, Allen et al. 2008). However, the siRNA knockdown had no effect on baseline unstimulated anion permeability. Stimulation of CFTR can also induce a steady-state apical bath alkalinization. In CFTR siRNA treated cells, the baseline □□□□ pH was similar to control, butstimulation did not produce an apical alkalinization.These findings indicatethat cAMP induced increases in apical HCO₃⁻ permeability in bovine corneal endothelium are due to CFTR. However, CFTR does not have a major role in determining baseline apical chloride or HCO₃⁻ permeability. This is consistent with reports indicating that there are no corneal abnormalities in humans with cystic fibrosis and studies showing that a specific CFTR blockerhad no significant effect on baseline corneal thickness (Diecke, Wen et al. 2004). CFTR may, like CLCA, have a role only during stimulated stressful conditions.

4e. CA IV

As described in figure 1, there is the potential for apical CO₂ flux contributing to net basolateral to apical HCO₃⁻ flux. This requires an apical carbonic anhydrase. An apical CAIV is presentin bovine corneal endothelium (Sun, Li et al. 2008). The relatively membrane impermeant carbonic anhydrase inhibitor benzolamide or CAIV siRNA knockdown reduced apical CO₂ fluxes due to an imposed [CO₂] gradient by ~20%, however there was no effect on HCO₃⁻ permeability or HCO₃⁻ flux(Sun, Li et al. 2008). If a significant net flux of CO₂ occurred across the apical membrane with conversion to HCO₃⁻ + H⁺ (figure 1), then blocking the apical CAIV would cause an alkalinization of the apical bath.However, apically applied benzolamide and siRNA treated cells showed an acidification of the apical side relative to basolateral, which is inconsistent with a net cell to apical CO₂ flux. Because CAIV does not facilitate steady-state cell to apical CO₂ flux, apical HCO₃⁻ permeability, or basolateral to apical HCO₃flux, but apparently maintains apical side pH, CAIV may have a role in buffering the apical surface.

4f. Role of NBCe1: In Vivo Studies

A mouse knockout for NBCe1 exists, however NBCe1 knockouts have significant developmental and systemic abnormalities and die shortly after birth, limiting their usefulness in analyzing the role of NBCe1 in the cornea. Therefore, to follow up on the previous in vitro studies, lentiviral vectors encoding NBCe1 specific shRNA were constructed and injected into the anterior chamber of live rabbits to determine if the bicarbonate fluxes produced by NBCe1 were needed to support corneal endothelial fluid transport in vivo (Liu, Cheng et al. 2010). This study showed that significant (up to 90%) endothelial NBCe1 knockdown could occur in vivo, however there was large variability in knockdown efficiency. If transendothelial HCO₃⁻transport were an integral part of the endothelial pump, then knocking down NBCe1 should induce significant corneal swelling. Surprisingly, there was no effect on corneal thickness. However, relative to control eyes that were injected with a Lentivirus encoding a scrambled sequence shRNA, the eyes receiving the NBCe1 specific shRNA were more sensitiveto the topical carbonic anhydrase inhibitor brinzolamide, i.e. there was more corneal swelling. These results were interpreted to suggest that the corneal endothelium has a large functional reserve (e.g., NBCe1 & CA activity), however they could also be interpreted to suggest that transendothelial HCO₃⁻flux is not an integral part of the “Pump”.In summary, the recent studies presented above do not directly support the anion secretion model.

5. The Electro-osmosis Model

The conventional view of epithelial cell secretion and absorption of water is that epithelial cell layers create local osmotic differences in the lateral spaces between the cells and/or on the apical surfaces of cells within an unstirred layer. The osmotic gradients generated serve as the driving forces for water movement across the epithelium.However, in many epithelial cells, including the corneal endothelium, there is no evidence that these gradients exist. This has led to the consideration of other mechanisms, such as Electro-Osmosis(Sanchez, Li et al. 2002). In this process, cells generate a transepithelial potential (0.5 mV apical side negative in corneal endothelium),through the concerted action of apical anion channels (HCO₃⁻ and Cl⁻),that draws counter-ions, i.e. Na⁺, through a paracellular pathway (the lateral space between cells)lined with fixed negative charges that make the paracellular pathway Na⁺specific. This produces electro-osmotic fluid coupling across the tight junction.Evidence for electro-osmosis and against other modes of fluid transport are recently reviewed (Fischbarg 2010). For electro-osmosis, a “leaky” junction with low specific resistance (20 ohm cm²) must have a relatively intense current density. This high Na⁺current generates fluid movement by electro-osmotic coupling. The Na⁺current loop is complete due to the presence of apical Na⁺channels (ENaC) (Mirshahi, Nicolas et al. 1999). Of interest, apical membrane density of ENaC can be enhanced by SGK1 activity (serum glucocorticoid kinase), which has also been shown to increase Na/K-ATPase activity in corneal endothelium (Hatou, Yamada et al. 2009; Rauz, Walker et al. 2003). This could be interpreted in favor of electro-osmosis or the increased ENaC density may simply be needed to offset the hyperpolarizing effect of increased Na/K-ATPase activity in order to maintain a stable membrane potential. Moreover, glucocorticoids thin rabbit(Hara 1970), but not human corneas (Baum and Levene 1968). A major problem with electro-osmosis in corneal endothelium is that a significant reduction in endothelial cell density, which is a normal process of ageing, does not result in reductions in endothelial function or production of corneal edema. The integrated effect of the high current density would be dissipated since the total paracellular area is significantly decreasing with age.

6. Aquaporins

The role of aquaporins in fluid transport has been controversial. Knockout models of AQP1, the most ubiquitous aquaporin, showed essentially no phenotype unless the animal was stressed by withholding water (Agre 1998). Subsequent studies indicated that knockout of aquaporins had effects on secretory tissues if the rate of secretion was relatively high(Ma, Fukuda et al. 2000; Ma, Song et al. 1999). Corneal endothelium has a high density of AQP1 located on both apical and basolateral membranes(Kuang, Yiming et al. 2004). In the knockout mouse, corneal thickness is thinner than wild type rather than the expected thicker more hydrated cornea if AQP1 was an important part of the “Pump”. The endothelial osmotic permeability is significantly reduced, as expected, and the rate of corneal thickness change in response to an osmotic challenge is slowed. These results suggest that AQP1 is present as a passive water pathway that speeds stromal hydration changes due to osmotic pressure fluctuations. In normal physiological situations this would include eye closure, which swells the cornea by ~4%, or exposure to hypo-osmotic conditions, e.g. swimming. This is consistent with the findings in other tissues that aquaporins do not have a role in secretion if the rate is relatively low.

7. Fluid Transport Agonists

The most highly studied fluid transport agonists in corneal endothelium are those that result in increased [cAMP]. Adenosine was the first agonist discovered (Dikstein and Maurice 1972) and was subsequently shown that it increased [cAMP] in endothelial cells (Riley, Winkler et al. 1996)through activation of A2_b receptors(Tan-Allen, Sun et al. 2005). Other agonists that increase [cAMP], e.g. forskolin or the phosphodiester inhibitor rolipram, also cause corneal thinning(Wigham, Turner et al. 2000). It is important to note that this could occur in de-epithelialized preparations so it was not due to the well-characterized corneal epithelial cAMP dependent Cl⁻ secretion (Klyce 1975). Besides being protective for corneal endothelial cells (Li, Allen et al. 2011), increased cAMP will phosphorylate and activate the apical CFTR channel leading to increased Cl⁻ and HCO₃⁻ flux(Sun, Zhai et al. 2003). It also relaxes endothelial cells leading to better cell-cell apposition and increased transendothelial resistance (Ramachandran and Srinivas 2010). ATP is also a potential fluid transport agonist. Purinergic receptor activation increased apical anion channel permeability as described above. Moreover, ATP released by the endothelial cells, can be broken down to adenosine by the presence of ecoATPases(Soltau, Zhou et al. 1993). Again, the presence of this physiologic response is probably in place to stimulate endothelial function during inflammatory, oxidative, or traumatic stress to the cell layer, but this will need to be tested.

8. Facilitated Lactate Transport

Often overlooked is the fact that in the in vivo cornea there are substantial bulk gradients of ions, e.g. glucose, HCO₃⁻ and lactate⁻(Chhabra, Prausnitz et al. 2009). [HCO₃⁻] is higher in the anterior chamber than in the cornea, especially when the eyelid is open. Thus HCO₃⁻will diffuse into the cornea passively in the opposite direction of the putative bicarbonate secretory mechanism. The cornea gets all of its glucose from the anterior chamber. Glucose in the cornea is consumed thereby maintaining an inward gradient for glucose diffusion. The high concentration of mitochondria in the endothelial cells would minimize glucose consumption allowing for supply to the very glycolytic keratocytes and epithelial cells. Overall, Riley (Riley 1969) has shown that 85% of glucose consumed in the cornea is converted to lactic acid. Therefore, for every 100 glucose molecules entering the cornea, 170 lactate molecules are produced. Klyce (Klyce 1981) demonstrated that the epithelium is impermeable to lactateand that the [lactate] in the cornea was about twice that in the aqueous humor, so there is a large gradient for lactate flux from cornea (~13 mM) to anterior chamber (~7 mM) across the endothelium. Lactate:H⁺cotransport was demonstrated in the endothelium (Giasson and Bonanno 1994) and the monocarboxylate transporters (MCTs) responsible are being studied in our laboratory.

Corneal hypoxia increases [lactate] and causes corneal swelling (Klyce 1981) indicating that lactate is an important osmolyte. It follows that any process that facilitates lactate efflux from the cornea will help maintain dehydration. Conversely, any interference with lactate efflux will cause corneal swelling. Figure 2 illustrates a model for lactate efflux across the endothelium that is facilitated by HCO₃⁻ buffering, carbonic anhydrase activity, 1Na⁺:2HCO₃⁻ cotransport, and Na⁺/H⁺exchange.An important role for buffering acid is supported by studies indicating that deturgescence can be maintained in the absence of HCO₃⁻ if a high concentration of other buffers are present (Doughty and Maurice 1988). Studies in our laboratory are currently examining the extent to which this process may contribute to the endothelial pump.

Figure 2.

Facilitated Lactate Transport Model. The bulk movement of lactic acid from cornea to anterior chamber via transcellular Monocarboxylate Cotransporters (MCTs) could contribute to net fluid movement out of the cornea. Intracellular buffering by the high intracellular [HCO₃⁻], provided by the 1Na⁺:2HCO₃⁻(NBCe1), Na⁺/H⁺exchanger (NHE1), and carbonic anhydrase II work to facilitate basolateral lactate:H⁺entry. Apical efflux through an MCT is facilitated by robust apical surface buffering, which is supplied by aqueous humor HCO₃⁻ and apical surface carbonic anhydrase IV.

9. Conclusion

The corneal endothelial pump is an active transport dependent process that requires both Cl⁻ and HCO₃⁻, but is highly dependent on HCO₃⁻ and facilitated by carbonic anhydrase activity. While a HCO₃⁻ secretory model for the pump has been preferred, there are several issues that argue against this interpretation. These include: the lack of apical anion transporters, equivocal measures of net anion fluxes from stroma to anterior chamber (probably due to the extremely leaky nature of the endothelium), no data indicating the presence of local osmotic gradients, a bulk fluid HCO₃⁻gradient that is against the putative secretory mechanism, and the greater potential for a Cl⁻ secretory mechanism. Alternatives such as electro-osmosis are being considered, but this suffers from the fact that significant reductions in electro-osmotic potential, i.e. the paracellular space, do not have direct effects on fluid pumping. The lactate efflux mechanism is attractive, but requires significant experimental testing. Increasing endothelial cAMP enhances pump function by a combination of anion channel activation and greater barrier function. Further studies are needed to determine if this mechanism could be used as a therapy for corneal edema.

Research Highlights.

• Review of corneal endothelial pump function

• Evidence supporting an anion transport based pump is limited

• siRNA and shRNA approaches provide new insights on endothelial function

• Evidence for alternate endothelial pump models including electro-osmosis & lactate flux