Net Fluorescein Flux Across Corneal Endothelium Strongly Suggests Fluid Transport is due to Electro-osmosis.

We have presented prior evidence suggesting that fluid transport results from electro-osmosis at the intercellular junctions of the corneal endothelium. Such phenomenon ought to drag other extracellular solutes. We have investigated this using fluorescein-Na2 as an extracellular marker. We measured unidirectional fluxes across layers of cultured human corneal endothelial (HCE) cells. SV-40-transformed HCE layers were grown to confluence on permeable membrane inserts. The medium was DMEM with high glucose and no phenol red. Fluorescein-labeled medium was placed either on the basolateral or the apical side of the inserts; the other side carried unlabeled medium. The inserts were held in a CO2 incubator for 1 h (at 37 °C), after which the entire volume of the unlabeled side was collected. After that, label was placed on the opposite side, and the corresponding paired sample was collected after another hour. Fluorescein counts were determined with a (Photon Technology) DeltaScan fluorometer (excitation 380 nm; emission 550 nm; 2 nm bwth). Samples were read for 60 s. The cells utilized are known to transport fluid from the basolateral to the apical side, just as they do in vivo in several species. We used 4 inserts for influx and efflux (total: 20 1-h periods). We found a net flux of fluorescein from the basolateral to the apical side. The flux ratio was 1.104 ± 0.056. That difference was statistically significant (p = 0.00006, t test, paired samples). The endothelium has a definite restriction at the junctions. Hence, an asymmetry in unidirectional fluxes cannot arise from osmosis, and can only point instead to paracellular solvent drag. We suggest, once more, that such drag is due to electro-osmotic coupling at the paracellular junctions.

[Comparative study of the effects of two anti-glaucomatous drugs on the corneal thickness in rabbits with corneal autografts]. / Estudio comparativo de los efectos de dos drogas antiglaucomatosas sobre el espesor corneal en conejos con autoinjerto de córnea.

PURPOSE: To determine the central corneal thickness after administration of the anti-glaucomatous medications latanoprost 0.005% or dorzolamide 2%, as assessed in rabbits which have had total corneal thickness autografts. METHODS: A bilateral total corneal thickness autograft was performed in ten rabbits. One rabbit was excluded from the subsequent study in which the antiglaucomatous medication was started two months post-operatively. Latanoprost 0.005% was instilled once per day into the right eye, whereas the left eyes were treated with dorzolamide 2% twice a day. The eyes were examined by biomicroscopy and ultrasound pachymetry immediately prior to commencement, and 4, 10, 17 and 27 weeks after starting the anti-glaucomatous treatment. In each instance three assessments of the central corneal thickness in each eye were made. At the end of the study, the influence of time and treatment on the corneal thickness was analyzed using a generalized linear model for repeated measurements. All penetrating keratoplasties were performed by the same surgeon (C.H.P). RESULTS: Treatment with dorzolamide resulted in corneal edema and a significant increase in central corneal thickness, whereas the treatment with latanoprost resulted in neither corneal edema nor corneal thickness changes. CONCLUSIONS: Dorzolamide, when instilled into the eyes of rabbits with corneal autografts, could have a negative effect on the graft, impairing the endothelial function through inhibition of the ionic pump. This effect could cause graft failure, which may be able to be defined with ultrasound pachimetry.

Aquaporins and fluid transport: an evolving relationship.

How epithelia transport fluid is controversial and remains undetermined. Two routes are possible: (1) via cell membranes and their aquaporins, or (2) paracellular. Our laboratory has recently developed experimental evidence and theoretical insights for fluid transport across corneal endothelium, a leaky epithelium. Aquaporin 1 (AQP1) is the only AQP present in these cells, and its deletion in AQP1 null mice significantly affects cell osmotic permeability but not fluid transport, which militates against sizable water movements across the cell. In contrast,AQP1 null mice cells have reduced regulatory volume decrease (only 60% of control), which suggests an AQP1 role in either the function or the expression of volume-sensitive membrane channels/transporters. Fluid movements can be produced by electrical currents, and the direction of the movement can be reversed by current reversal or by changing junctional electrical charges with polylysine. A mathematical model of corneal endothelium predicts experimental observations only when based on paracellular electro-osmosis. Our novel paradigm for this preparation includes: (1) paracellular fluid flow; (2) a crucial role for the junctions as a site for electro-osmosis; (3) hypotonicity of the primary secretion; (4) an AQP role in regulation and not as a significant water pathway.

Osmotic permeabilities across corneal endothelium and antidiuretic hormone-stimulated toad urinary bladder structures.

Osmotic permeabilities of several epithelial structures have been determined with novel optical procedures based on specular microscopy. The osmotic permeabilities of several tissue layers were determined by continuously monitoring the position of the apical tissue borders while an osmotic flow was imposed across those layers. The values found were (in micrometer/s; mean +/- SE): corneal epithelium, 137 +/- 30 (n = 5): antidiuretic hormone stimulated toad bladder, 429 +/- 64 (n = 6); and corneal endothelium, 711 +/- 34 (n = 7). In addition, the osmotically-induced transient change in thickness of the corneal endothelial cells was determined with the help of a computer, and the apparent osmotic permeability measured for the apical membrane was 1420 +/- 160 micrometer/s (n = 5). It is concluded that the osmotic permeability across the endothelial layer is sizably larger than had been previously detected and that osmotic flows across such layer largely traverse the cellular membranes. With osmotic permeability values (per unit of cell membrane area) as large as presently reported, isotonic fluid transport by epithelia can be explained simply on the basis of local osmotic gradients.

Determining the kinetics of membrane pores from patch clamp data without measuring the open and closed times.

We present a new and very time saving way to determine kinetic rate constants from patch clamp data by using the correlation functions utilized in analyzing experiments with photon correlations.

Standing-gradient osmotic flow. Examination of its validity using an analytical method.

The solutions to the non-linear differential equations governing solute-solvent coupling in the intercellular spaces of epithelial layers have been obtained by using an analytical method, rather than the usual numerical ones. When the present series solution includes second-order correction terms, the concentration and velocity profiles obtained by the analytical method agree very well with those coming from numerical solutions. This method has further allowed us to examine the standing-gradient hypothesis when applied to the backwards fluid transport system of the corneal endothelium. With the information presently available for the relevant parameters (osmotic permeability, rate of transport, radius and length of the spaces, and location of the pumping sites), near-isotonicity of the transported fluid would not be explained by the standing-gradient model.

Priming of the fluid pump by osmotic gradients across rabbit corneal endothelium.

The present study shows that the inclusion of 5% Dextran (average mol. wt. 40 000) in solutions to preserve in vitro rabbit corneal endothelium induces a sizable osmotic flow across the preparation which is superimposed on the existing fluid transport. Furthermore, even after fluid transport ceases due to in vitro deterioration, the Dextran-induced flow remains for some addition time. The osmotic permeability was 162 +/- 17 micrometer/s in the presence of glucose and 451 +/- 84 micrometer/s in its absence. The latter, comparatively high value suggests that such osmotic flow traverses the intracellular junctions. In addition, temporary (10--15 min) imposition of an osmotic gradient has a separate stimulatory 'priming' effect on the rate of fluid transport. Thus, the rate of fluid pumping increased by about 40% after challenge with Dextran. It was further noted that, after addition of Dextran, preparations in the absence of glucose escape gross deterioration for a time longer than those in the presence of glucose. On the other hand, mere addition of Dextran to a glucose-containing solution does not appear to prolong the estimated 'survival time' of the pumping mechanism. The sizable osmotic flows and the priming effect described here may provide a physiological context with which previously described Dextran effects on cornea preservation can now be compared.

Osmotic water permeability of rabbit corneal endothelium and its dependence on ambient concentration.

We measured the fluid flow osmotically induced by sucrose concentration differences across the isolated rabbit corneal endothelium to determine if its osmotic water permeability, Pos, depends on the concentrations on both sides as well as on the concentration difference across that tissue. We found that when the osmotically induced fluid flow went from stroma to aqueous, Pos decreased from 35 +/- 4 mum/s to 20 +/- 3 mum/s when an additional 20 mosM sucrose was added to the solutions on both sides of the endothelium. However, when the osmotically induced fluid flow was towards the stroma, Pos remained practically unchanged. (28 +/- 4 mum/s vs. 31 +/- 5 mum/s), when additional 20 mosM sucrose was present on both sides. These changes in the measured permeability are consistent with the possibility that sucrose would be swept into the intercellular channels by the osmotically induced fluid flow. We also confirmed that an osmotic gradient can 'prime' the fluid pump. After a gradient was removed, gradients which had previously induced flow from stroma to aqueous caused an increase in the basal fluid pump rate, while gradients in the opposite direction decreased that fluid pump rate.

Inhibition of transepithelial osmotic water flow by blockers of the glucose transporter.

On the basis of evidence derived mostly from human erythrocytes, it has been suggested that water traverses cell membranes through membrane-spanning proteins such as the anion channel or the glucose transporter acting as water pores. However, specific inhibitors of such permeation processes have not been found to block water transport, and hence a precise identification of the water route has not been possible so far. We have investigated this issue by characterizing the osmotic flows across a fluid-transporting epithelium, the rabbit corneal endothelium. The rate of such flows was monitored continuously as a function of time. We confirmed prior findings of an inhibition by PCMBS on osmotic water flow, and lack of inhibition by DTNB and DIDS. On the other hand, we have found for the first time that several blockers of glucose facilitated diffusion, namely, phloretin (2 mM), phloridzin (2 mM), diallyldiethylstilbestrol (0.1 mM), cytochalasin B (20 micrograms/ml), and ethylidene-D-glucose (200 mM), all clearly inhibit osmotic flow. Our evidence is consistent with the hypothesis that both water and glucose may traverse these cell membranes through the same channel-like pathway contained in the glucose transporter membrane-spanning protein.

Osmotic permeability in a molecular dynamics simulation of water transport through a single-occupancy pore.

The aim of this work is to determine plausible values for the rate constants of kinetic models representing water transport through narrow pores. We present here the results of molecular dynamics simulations of the movement of water molecules through a single-site hydrophilic pore. The system consists of a rectangular box of water molecules, some of which are positionally restrained so as to act as a membrane. This membrane separates two compartments where water molecules move freely; one of the positions in the membrane is initially vacant (the 'single-site pore'), but can be occupied by mobile molecules. To analyze the results, we represented the pore by a two-state kinetic diagram in which the vacant and occupied states are linked by transitions corresponding to the binding and release of water molecules. The mean occupancy and vacancy times directly yield the rate constants of binding and release, which in turn yield the osmotic water permeability coefficient per pore pf. We also compute the apparent activation energies delta E* for the rate constants and for pf. The pf value was (1.56 +/- 0.04).10(-11) cm3/s (at 307 K), which is much larger than those determined for CHIP28 and for gramicidin A (of about 10(-13) and 10(-14) cm3/s, respectively). These values were compared with those arising from a model of a symmetric single-file pore through which one-vacancy-mediated water transport takes place. The model yields an expression for pf as a function of the rate constants and of the number of molecular positions (n) in the file. When n = 1, this expression becomes the one corresponding to the single-site pore studied in our current simulation. Using the rate constants of binding and release derived from our simulation, the pf values are consistent with an occupancy value of 5-6 found for gramicidin A, and with occupancies of 4-7 that can be estimated for the single-file pore of a recently proposed model for CHIP28. delta E* for pf is 3.0 kcal/mol, a value similar to that determined for CHIP28. Hence, the system simulated here appears plausible and can be used to mimic some physical properties of water transport through biological pores.

Fractal model of ion-channel kinetics.

Markov models with discrete states, such as closed in equilibrium with closed in equilibrium with open have been widely used to model the kinetics of ion channels in the cell membrane. In these models the transition probabilities per unit time (the kinetic rate constants) are independent of the time scale on which they are measured. However, in many physical systems, a property, L, depends on the scale, epsilon, at which it is measured such that L(epsilon) alpha epsilon 1-D where D is the fractal dimension. Such systems are said to be 'fractal'. Based on the assumption that the kinetic rates are given by k(t) alpha t1-D we derive a fractal model of ion-channel kinetics. This fractal model has fewer adjustable parameters, is more consistent with the dynamics of protein conformations, and fits the single-channel recordings from the corneal endothelium better than the discrete-state Markov model.

Monomeric human red cell glucose transporter (Glut1) in non-ionic detergent solution and a semi-elliptical torus model for detergent binding to membrane proteins.

The self-association state of the human red cell glucose transporter (Glut1) in octaethylene glycol n-dodecyl ether (C12E8) and n-octyl beta-D-glucopyranoside (OG) solution was analyzed in the presence of reductant by gel filtration with light-scattering, refractivity and absorbance detection, and by ultracentrifugation. The C12E8-Glut1 complex was essentially monomeric, whereas OG-Glut1 also formed dimers and larger oligomers. C12E8-Glut1 retained substantial glucose transport activity even after depletion of endogenous lipids by gel filtration, as shown by reconstitution and transport measurements. Removal of endogenous lipids from OG-Glut1 abolished the activity unless phosphatidylcholine was included in the eluent. The binding of C12E8 and OG to Glut1 was determined by gel filtration with refractivity and absorbance detection or with radioactive tracer to be 1.86 +/- 0.07 and 1.84 +/- 0.09 g/g polypeptide, respectively. A structural model was proposed in which non-ionic detergent forms a semi-elliptical torus (SET) surrounding the transmembrane protein. The torus thickness was assumed to be equal to the radius (short half-axis) of a spherical (oblate ellipsoidal) free detergent micelle and the polar head groups of the detergent molecules were predicted to be situated just outside the hydrophobic surface of the protein. The experimental detergent binding values and those obtained from the SET model together confirmed that Glut1 was monomeric in C12E8 solution and provided constraints on the shape and size of the hydrophobic transmembrane region of Glut1 in alpha-helical and beta-barrel topology models.

Kinetic analysis of water transport through a single-file pore.

We apply the diagrammatic method developed by Hill (1977. Free Energy Transduction in Biology. Academic Press, New York) to analyze single-file water transport. We use this formalism to derive explicit expressions for the osmotic and diffusive permeabilities Pf and Pd of a pore. We first consider a vacancy mechanism of transport analogous to the one-vacancy pore model previously used by Kohler and Heckmann (1979. J. Theor. Biol. 79:381-401). (a) For the general one-vacancy case, we find that the permeability ratio can be expressed by Pf/Pd = (Pf/Pd)eqf(wA,wB), where the second factor is a function of the water activities in the two adjoining compartments A and B. As a consequence, the permeability ratio in general can effectively differ from its value at equilibrium. We also find that n - 1 less than or equal to (Pf/Pd)eq less than or equal to n, a result already proposed by Kohler and Heckmann (1979. J. Theor. Biol. 79:381-401). (b) When vacancy states are transient intermediates, the model can be reduced to a diagram consisting of only fully occupied states. Such a diagram resembles the one describing a no-vacancy mechanism of transport (c), but in spite of the similarity the expressions obtained for the permeability coefficients still retain the basic relationships of the original (a) nonreduced one-vacancy model. (c) We then propose a kinetic description of a no-vacancy mechanism of single-file water transport. In this case, the expressions derived for Pf and Pd are formally equivalent to those obtained by Finkelstein and Rosenberg (1979. Membrane Transport Processes. Vol. 3. C.F. Stevens and R.W. Tsien, editors, Raven Press, New York. 73-88.) A main difference with the vacancy mechanism is that here the permeability coefficients are independent of the water activities.

Role of facilitative glucose transporters in diffusional water permeability through J774 cells.

We have reported previously that in the presence of an osmotic gradient, facilitative glucose transporters (GLUTs) act as a transmembrane pathway for water flow. Here, we find evidence that they also allow water passage in the absence of an osmotic gradient. We applied the linear diffusion technique to measure the diffusional permeability (Pd) of tritiated water (3H-H2O) through plasma membranes of J774 murine macrophage-like cells. Untreated cells had a Pd of 30.9 +/- 1.8 microns/s; the inhibitors of facilitative glucose transport cytochalasin B (10 microM) and phloretin (20 microM) reduced that value to 15.3 +/- 1.8 (50%) and 11.0 +/- 0.7 (62%) microns/s, respectively. In contrast, no significant effect on Pd was observed in cells treated with dihydrocytochalasin B (Pd = 28.4 +/- 1.5 microns/s). PCMBS (3 mM) inhibited glucose uptake by greater than 95%, and 3H-H2O diffusion by approximately 30% (Pd = 22.9 +/- 1.5 microns/s). The combination of cytochalasin B plus pCMBS reduced Pd by about 87% (Pd = 3.9 +/- 0.3 microns/s). Moreover, 1 mM pCMBS did not affect the osmotic water permeability in Xenopus laevis oocytes expressing the brain/erythroid form of facilitative glucose transporters (GLUT1). These results indicate for the first time that about half of the total Pd of J774 cells may be accounted for by water passage across GLUTs. Hence, they highlight the multifunctional properties of these transporters serving as conduits for both water and glucose. Our results also suggest for the first time that pCMBS blocks glucose transport without affecting water permeation through GLUTs. Lastly, because pCMBS decreases the Pd of J774 cells, this suggests the presence in their plasma membranes of another protein(s) exhibiting water channel properties.

Expression of multiple water channel activities in Xenopus oocytes injected with mRNA from rat kidney.

To test the hypothesis that renal tissue contains multiple distinct water channels, mRNA prepared from either cortex, medulla, or papilla of rat kidney was injected into Xenopus oocytes. The osmotic water permeability (Pf) of oocytes injected with either 50 nl of water or 50 nl of renal mRNA (1 microgram/microliter) was measured 4 d after the injection. Pf was calculated from the rate of volume increase on exposure to hyposmotic medium. Injection of each renal mRNA preparation increased the oocyte Pf. This expressed water permeability was inhibited by p-chloromercuriphenylsulfonate and had a low energy of activation, consistent with the expression of water channels. The coinjection of an antisense oligonucleotide for CHIP28 protein, at an assumed > 100-fold molar excess, with either cortex, medulla, or papilla mRNA reduced the expression of the water permeability by approximately 70, 100, and 30%, respectively. Exposure of the oocyte to cAMP for 1 h resulted in a further increase in Pf only in oocytes injected with medulla mRNA. This cAMP activation was not altered by the CHIP28 antisense oligonucleotide. These results suggest that multiple distinct water channels were expressed in oocytes injected with mRNA obtained from sections of rat kidney: (a) CHIP28 water channels in cortex and medulla, (b) cAMP-activated water channels in medulla, and (c) cAMP-insensitive water channels in papilla.

Mercurial sensitivity of aquaporin 1 endofacial loop B residues.

The water channel protein aquaporin-1 (AQP1) has two asparagine-proline-alanine (NPA) repeats on loops B and E. From recent structural information, these loops are on opposite sides of the membrane and meet to form a pore. We replaced the mercury-sensitive residue cysteine 189 in AQP1 by serine to obtain a mercury-insensitive template (C189S). Subsequently, we substituted three consecutive cysteines for residues 71-73 near the first NPA repeat (76-78) in intracellular loop B, and investigated whether they were accessible to extracellular mercurials. AQP1 and its mutants were expressed in Xenopus laevis oocytes, and the osmotic permeability (P(f)) of the oocytes was determined. C189S had wild-type P(f) but was not sensitive to HgCl(2). Expression of all three C189S cysteine mutants resulted in increased P(f), and all three mutants regained mercurial sensitivity. These results, especially the inhibitions by the large mercurial p-chloromercunbenzene-sulfonic acid (pCMBS) ( approximately 6A wide), suggest that residues 71-73 at the pore are accessible to extracellular mercurials. A 30-ps molecular dynamics simulation (at 300 K) starting with crystallographic coordinates of AQP1 showed that the width of the pore bottleneck (between Connolly surfaces) can vary (w(avg) = 3.9 A, sigma = 0.75; hydrated AQP1). Thus, although the pore width would be > or = 6 A only for 0.0026 of the time, this might suffice for pCMBS to reach residues 71-73. Alternative explanations such as passage of pCMBS across the AQP1 tetramer center or other unspecified transmembrane pathways cannot be excluded.

A method for preparing a viable corneal endothelial layer, completely denuded of overlying stroma.

A method for denuding Descemet's membrane by dissection of all overlying stroma, while preserving the corneal endothelial layer, was developed and evaluated. The technique consists in treating the stroma with trypsin and surgically removing the softened stromal layers. With an automated thickness-measuring technique, the endothelium-Descemet preparation was found to range from 23 to 42 micron in thickness. Endothelial cell morphology was normal under specular and light-transmission microscopic examinations. Under electron microscopy, the endothelial cells appeared intact, except for an increase in the number of intracellular vacuoles. Occasionally, small portions of an intercellular space were found to be mildly dilated, but the over-all integrity of the junctional complex was intact. In vitro, the viability of the preparation was comparable with that of a cornea with all layers intact. Endothelial resting membrane potentials, measured with intracellular microelectrodes, were found to be within the normal (33 +/- 2 mV) range. It is thus possible to obtain a viable endothelial layer, completely stripped of stroma.

Noninvasive measurements of pyridine nucleotide fluorescence from the cornea.

The autofluorescence of reduced pyridine nucleotides (NADH and NADPH) and oxidized flavoproteins within the rabbit cornea were noninvasively measured as a function of depth. This was accomplished by combining a corneal specular microscope with a time-shared spectrofluorometer. When either 8 mM sodium pentobarbital or sodium sulfide, known inhibitors of mitochondrial respiration were applied to cornea, the autofluorescence at 440 nm (excited at 366 nm) increased and that at 540 nm (excited at 460 nm) decreased. No autofluorescence was measurable following destruction of the cellular membranes by freezing and leaching of the cellular constituents. The 440 nm autofluorescence is from reduced pyridine nucleotides, whereas the 540 nm autofluorescence is from the oxidized flavoproteins. The time course of the pyridine nucleotide autofluorescence after the application of the pentobarbital to either the endothelial or epithelial bathing solutions made it possible to measure the diffusion properties of this drug through the cornea. The method used is useful studying the diffusion and effects of metabolically active drugs upon the cornea.

Cultured bovine corneal epithelial cells express a functional aquaporin water channel.

PURPOSE: Given recent physiological and in situ hybridization evidence for the presence of a water channel in corneal epithelium, this study was conducted to investigate its expression and characteristics using cultured bovine corneal epithelial cells (CBCEPCs). METHODS: CBCEPCs were grown in DMEM containing 2 ng/ml fibroblast growth factor and 6% fetal bovine serum. To determine their osmotic permeability (Pf), cells were passaged onto rectangular glass coverslips, and anisotonically induced volume changes were monitored by light scattering. To investigate expression, poly(A+) RNA from CBCEPCs was injected into Xenopus laevis oocytes, and the Pf of the oocytes was determined. RESULTS: For CBCEPCs challenged with a 10% hypotonic solution at 37 degrees C, the kinetic constant of volume change was k=0.52+/-0.04 seconds(-1), and the calculated Pf 72+/-6 microm/sec (n=16). The Pf of oocytes injected with water was 14+/-1.8 microm/sec (n=4); injection with poly(A+) RNA from CBCEPCs increased Pf to 77+/-6 microm/sec (n=6). This increase in Pf was inhibited by 72% (reduced to 22+/-1 microm/sec) by 0.3 mM HgCl2 and was inhibited by 56% to 58% by coinjection with aquaporin (AQP)5 antisense oligonucleotide. CONCLUSIONS: The comparatively high Pf determined for CBCEPCs, the presence of mRNA encoding water channels, and sensitivity to mercurial agents are typical of the expression of functional water channels. The predominant message is for AQP5, although the evidence was consistent with the presence of additional water channels. These findings bring renewed support for the notion that the epithelium can contribute to corneal hydration homeostasis.