nak-atpase pump sites in cultured bovine corneal endothelium of

NaK-ATPase Pump Sites in Cultured Bovine CornealEndothelium of Varying Cell Density at Confluence

Kenneth M. Crawford* Stephen A. Ernst,* Roger F. Meyer,,f and Donald K. MacCallum*

Purpose. The driving force for ion and water flow necessary for efficient deturgesence of thecorneal stroma resides in the ouabain-sensitive sodium (Na) pump of corneal endothelialcells. Using a cell culture model of corneal endothelial cell hypertrophy, the authors examinedthe expression of Na pumps at the cell surface to see how this central element of the endothe-lial pump changed as corneal endothelial cell density decreased to a level associated withcorneal decompensation in vivo.

Methods. 3H-ouabain binding to NaK-ATPase at saturating conditions was used to quantitatethe number of Na pump sites on cultured bovine corneal endothelial cells as the confluentdensity decreased from ~2750 cells/mm2 to ~275 cells/mm2.

Results. The mean number of Na pump sites per cell at confluence (1.92 ± 0.07 X 106) didnot change as the cell density decreased 2.7-fold from 2763 cells/mm2 to 1000 cells/mm2.However, pump site expression doubled to ~4 X 10s sites/cell as the cell density decreasedfrom 1000 cells/mm2 to 275 cells/mm2. Despite the incremental increase in Na pump siteexpression that occurred as the cells hypertrophied below a density of 1000/mm2 to achieveconfluence, this increase was insufficient to prevent a decrease in Na pump site density ofthe intact monolayer, expressed as pump sites/mm2.

Conclusion. The confluent cell density of cultured bovine corneal endothelial cells can bevaried from that found in the normal native cornea to that associated with corneal decompen-sation. In confluent cultures with cell densities ranging from 2750 cells/mm2 to 1000 cells/mm2, the number of pump sites per cell remains relatively unchanged. Below cell densitiesof 1000 cells/mm2, the number of pump sites per cell progressively increases. The increasedNa pump site abundance in markedly hypertrophied endothelial cells cannot aJ'-quatelycompensate for the progressive reduction in the number of transporting cells per unit areawithin the intact monolayer. Even when considered with the decrease in the size of theparacellular ion conductive pathway that is a consequence of progressive endothelial hypertro-phy, the overall pumping capacity of the intact endothelial monolayer declines. Invest Oph-thalmol Vis Sci. 1995;36:1317-1326.

J. he corneal endothelium is a monolayer of neuralcrest-derived cells that lines the posterior surface ofthe cornea.' When functioning properly, the endothe-lium regulates, with some contribution by the stra-tified squamous epithelium of the anterior cornea,the state of corneal stromal hydration and cornealclarity.2"4 When the endothelial monolayer is dis-

From the * Departments of Anatomy and Cell Biology and ^Ophthalmology,University of Michigan, Ann Arbor, Michigan.Supported by National Institutes of Health grant EY03573.Submitted for publication August 15, 1994; revised November 4, 1994; acceptedDecember 22, 1994.Proprietary interest category: N.lie/mnl requests: DonaM K. MacCallum, Department of Anatomy and Cell Biology,4S12 Medical Science 11, Box 0616, University of Michigan, Ann Arbor, Ml48109-0616.

turbed by either a disease process or a mechanicalinjury, the dynamic equilibrium of fluid penetrationinto the stroma and the removal of ions and waterfrom the stroma to the aqueous by an energy-drivenendothelial pump becomes unbalanced. As a conse-quence, the corneal stroma swells, the collagen lamel-lae become disorganized, impairing visual acuity, andthe corneal epithelium loses its integrity, resulting inrecurrent epithelial erosions, stromal scarring, andeventual blindness.5

Corneal transparency depends on an endothelialmembrane pump that compensates for passive stromalhydration by coupling water flow to net transendothe-lial ionic flux to the aqueous, resulting in the mainte-nance of corneal deturgesence. Although the mecha-

Investigaiive Ophthalmology & Visual ScitCopyright © Association for Research in '

e.June 1995, Vol. 36, No. 7on and Ophthalmology 1317

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933191/ on 02/12/2018

1318 Investigative Ophthalmology & Visual Science, June 1995, Vol. 36, No. 7

nistic details remain to be completely defined, it isclear that the underlying driving force for transendo-thelial ion- and solute-linked water transport ulti-mately resides in the ouabain-sensitive sodium (Na)pump situated in the basolateral corneal endothelialcell membrane.4I>~" The critical role of the Na pumpin stromal deturgesence is demonstrated by ouabain-induced stromal swelling, a phenomenon that occurswithout compromise of endothelial barrier functionas judged by unaltered permeability to nonelectrolytesand fluorescent dyes and maintenance of normal junc-tional architecture.1'^'5

When corneal endothelial cells are lost in certainspecies, including humans, the remaining cells hyper-trophy, thereby reducing the number of cells thatcomprise the endothelial monolayer.16 In this study,we investigated a cell culture model in which the num-ber of bovine corneal endothelial cells present in con-fluent cultures can be varied (that is, the cell densityvaried). Subsequently, we determined the relationshipbetween cell size and the number of Na pump sitespresent on the cell surface, demonstrated by the spe-cific binding of the glycoside ouabain to NaK-ATPase.The experiments were undertaken to learn how thiscritical element of the corneal endothelial pumpchanges as the cells exhibit compensatory hypertrophyin a manner analogous to that which occurs after cor-neal endothelial cell loss from disease or trauma inhuman and animal corneas.l6~19

MATERIALS AND METHODS

Materials

Culture Media and Salt Solutions. Earle's balancedsalt solution (EBSS), Dulbecco's modified medium(DMEM) with 4.5 g glucose/1, and Eagle's minimalessential medium (EMEM) were obtained from Gibco(Grand Island, NY). Iron-supplemented bovine calfserum (BCS) was purchased from HyClone (Logan,UT), and crystalline trypsin was purchased from CalBi-ochem (San Diego, CA). [3H]-ouabain (15 to 30 Ci/mmol) was obtained from New England Nuclear (Bos-ton, MA), and liquid scintillation cocktail (Bio-Safe II)was obtained from Research Products International(Mount Prospect, IL). Dispase (grade II) was pur-chased from Boehringer Mannheim (Indianapolis,IN), ouabain octahydrate and all other chemicals werepurchased from Sigma Chemical (St. Louis, MO). Hu-man recombinant epidermal growth factor (hrEGF)was a gift from Chiron IntraOptics (Irvine, CA).

Cell Culture

Bovine corneal endothelial cell (BCEC) cultures wereprepared with modifications as described by MacCal-lum el al.20 Briefly, bovine eyes were obtained from a

local abattoir, and the corneas were excised with anattached scleral ring and placed, endothelium side up,in a concave plastic cup. The endothelial surface wasrinsed and subsequently incubated at 37°C for 1 '/ahours in a sterile solution of 20 mM HEPES bufferedEBSS, pH 7.4, containing 0.25% crude dispase. Endo-thelial cells were dislodged from Descemet's mem-brane by gentle scraping using a spatula with a taperedsilicone surgical rubber tip. Dislodged cells were aspi-rated from the eyes with a pipette and added to 5 mlof DMEM containing 10% BCS and 50 fig/m\ genta-micin sulfate (subsequently referred to as "standardmedium"). The cells from a single eye were gentlycentrifuged (600g; 2 minutes), resuspended in 5 mlof standard medium supplemented with 10 ng/mlhrEGF, and incubated in 25-cm2 flasks at 37°C in 95%air-5% CO2. Cultures were fed three times a week andwere typically confluent in 5 to 6 days. The addition ofhrEGF is not required for BCEC growth, but it wasadded to amplify rapidly the growth of cells from asingle eye in primary culture. Confluent endothelialcultures (~2.2 - 2.5 X 106 cells/25 cm2 flask) weresubcultivated using a 5- to 8-minute incubation inCa2+, Mg2+-free EBSS, pH 7.4, containing 0.6 mM eth-ylenediaminetetraacetic acid (EDTA) and 0.05% crys-talline trypsin and subcultured.

Cells were grown in standard medium except asnoted in Results, where either 10 ng/ml hrEGF orEMEM with 5% BCS (referred to as "basal medium")were used to adjust the number of cells in a confluentmonolayer. Confluence was defined as the conditionin which the culture surface was completely coveredby BCEC, and only rare mitotic figures were observed.Generally, experiments were conducted on cultures 2days after they had reached confluence. Except whereindicated, all cultures were fed with either standardor basal media 24 hours before ligand-binding studies.The data presented in this article were obtained fromcells in the first three passages (primary culture = 0passage.)

Cultures were routinely observed with an invertedphase-contrast microscope. For photomicrography,cultures were grown on glass coverslips, fixed in 70%methanol for 15 minutes, mounted in glycerol-phos-phate-buffered saline (9:1), and photographed with aconventional phase-contrast microscope. Cell countswere made by counting the nuclei in three 1600-^m2

microscopic fields for each of the cell densities illus-trated and averaging the cell count.

Pump Site Determination

Equilibrium binding of [3H]-ouabain was determinedin confluent bovine corneal endothelial cell culturesgrown in 12 well plates (3.8 cmVwell) at 37°C. Todetermine saturation binding conditions, cultureswere rinsed twice in potassium-free HEPES-buffered


Cultured BCE Sodium Pumps 1319

— 3.0 r

•Ocuomc'rojQCDD

o 0.0

Bound (fmol / mm )

— O —

20 40 60 80 100

[Ouabain] (nM)

FIGURE 1. Saturation binding of 3H-ouabain to the Na pump in cultured bovine cornealendothelial cells (BCEC). Specific binding (closed circles) calculated after subtracting nonspe-cific binding (open circles) in the presence of 0.5 mM unlabeled ouabain is illustrated. Satura-tion binding was achieved at 20 nM ouabain (each point is the mean ± SEM for seven 3.8cm2 wells.) The inset is a Scatchard analysis of specific binding from which maximal bindingof 3.7 fmol/mm2 and a dissociation constant (K,,) of 6.7 nM for the cultured BCEC in thisstudy were derived.

Kreb's Ringer Henseleit solution (K+-free KRH),which contained in mmol/1: 149 NaCl, 1.7 Cad,,, 1MgCl2, 1 NaHPO3, 6 NaHCO3, 5 glucose, and 10HEPES, adjusted to pH 7.4 with NaOH. The cultureswere incubated for 120 minutes with concentrationsof [3H]-ouabain ranging from 1 nM to 100 nM. Prelim-inary experiments demonstrated that 75% of equilib-rium binding was achieved in 30 minutes, 94% wasachieved in 60 minutes, and equilibrium was reachedby 90 minutes. Duplicate incubations of [sH]-ouabainwere carried out in which the medium also contained0.5 mM unlabeled ouabain. At each concentration,specific binding was calculated from the differencebetween total counts and that measured in the pres-ence of 0.5 mM unlabeled ouabain. Incubation withexcess unlabeled ouabain suppressed [3H] -ouabainbinding by 93% to 98% (see Results and Fig. 1). Imme-diately after incubation, cultures were rinsed threetimes with ice-cold K+-free KRH followed by additional5-minute and 10-minute washes to remove unboundlabel. The ouabain bound to the cells after incubation(~24% of total added counts) was firmly bound to theNa pump as an additional six washes over 25 minutesreduced binding by only an additional 0.5%. Cellswere then solubilized overnight in a mixture of 1%sodium dodecyl sulfate and 1 N NaOH. The solubi-lized cells were aspirated, and the culture well wasrinsed with water to retrieve all counts. The samples

were neutralized with 1 N HC1 and counted by liquidscintillation spectroscopy.

Determination of Cell NumberThe number of cells in a culture well grown and han-dled under conditions identical to wells used in ra-dioligand-binding experiments was determined withan automated cell counter (Coulter Electronics, Hia-leah, FL). Briefly, at the end of 2-hour incubation inK+-free KRH and subsequent washes, cultures wererinsed with Ca2+, Mg2+-free EBSS + 0.6 mM EDTAfollowed by a 3- to 5-minute incubation in Ca2+, Mg2+-free EBSS containing 0.6 mM EDTA and 0.05% crystal-line trypsin. The detached BCEC were added to 0.5ml serum to quench the action of trypsin, centrifuged(600g, 2 minutes), resuspended in an isotonic diluent,and counted.

Statistical AnalysisData obtained under identical experimental condi-tions were pooled; means were calculated and are pre-sented in Results with standard error (SE). Analysis ofvariance, regression analysis, and /-tests were appliedwhere appropriate. Results of statistical tests were con-sidered significant at P < 0.05.

Construction of a Planar Paracellular PathwayModelA simple, two-dimensional geometric model was con-structed to approximate the linear extent of die para-



cellular pathway in confluent cultures of different fi-nal cell densities. (The model actually describes theentrance to the paracellular pathway21 rather than itstotal volume, which is estimated to be 10 times endo-thelial cell thickness over a wide range of cell densitiesand cell shapes3). A plot was constructed from thecalculated area and perimeter of regular hexagons, ashape representative of normal, native corneal endo-thelium, whose sides were increased in increments of

1 fim. The area of an individual cell in 1 mm2 of amonolayer at various cell densities was calculated, andits perimeter was obtained from the area-perimeterplot. For example, in a culture of 1000 cells/mm2,each cell has an area of 1000 //m2 and a perimeter of119 fj.m. Dividing the cell perimeter by 2 (the paracel-lular pathway is formed from the junction of two cells)and multiplying by the number of cells/mm12 approxi-mates the total paracellular pathway (e.g., 119 /xm -r

2 = 59.5 yum; 59.5 //m X 1000 cells/mm2 = 59,500yjm/mm2 paracellular pathway for a cell density of1000 cells/mm2) (see Discussion and Fig. 6) To verifythis relationship, the paracellular pathway was esti-mated from a model of irregular hexagons that moreaccurately reflected the actual arrangement of cul-tured corneal endothelial cells, drawn to scale at vary-ing cell densities. The perimeter of 60 cells was mea-sured, and the paracellular pathway was calculated ashalf the total perimeter. The results derived from theirregular hexagon model were virtually identical to theparacellular pathway calculations using the geometricrelationships of the regular hexagons (see Discussionand Fig. 6).

RESULTS

Binding Kinetics

Preliminary studies were carried out to characterizethe kinetics of ouabain binding to BCEC cultures atstandard confluent densities of ~1200 BCEC/mm2.The saturation binding curve For [3H]-ouabain toBCEC is illustrated in Figure 1. Saturation binding wasachieved at [ouabain] = 20 nM. To ensure saturationof binding sites, 50 nM [3H]-ouabain was used for allsubsequent binding determinations. A Scatchard plot(inset) of specific binding (r = 0.98) revealed a singlepopulation of binding sites with a maximal bindinglevel and equilibrium dissociation constant (Ka) of 3.7fmol/mm2 and 6.7 nM, respectively.

Model of Corneal Endothelial CellHypertrophy

To achieve confluent BCEC cultures of different den-sities for ouabain binding studies, subcultures wereinitiated with a variable number of cells essentially asdescribed.22 In general, the greater the number of

cells subcultivated, the greater the resultant cell den-sity at confluence. In practice, we varied both the num-ber of cells used to initiate subcultures and the growthmedium. To achieve routinely a cell density ap-proaching the cell count in native bovine corneas,-2500 cells/mm2,22 ~0.5 X 106 cells were subculturedin DMEM + 10% BCS + 10 ng/ml hrEGF for onemedium change and then in standard medium alone.To achieve all other cell densities, -275 -2000 cells/mm2, the number of cells used to initiate the subcul-ture, was progressively reduced (essentially 1:2 to 1:64subcultivation ratios) and grown in either standardmedium (DMEM + 10% BCS) or basal medium(MEM + 5% BCS). The use of basal medium (MEM+ 5% BCS) was necessary to achieve confluent cul-tures of low density (<500 cells/mm2) widi all cellsapproximately the same size; the use of standard me-dium in similar circumstances resulted in confluentcultures composed of groups of cells exhibiting dispa-rate sizes (not shown).

Representative phase-contrast photomicrographsof cultures with density similar to that of native bovinecornea (2221 cells/mm2), intermediate cell density(958 cells/mm2), and density exhibiting significant hy-pertrophy (325 cells/mm2) are shown in Figure 2. Ascells enlarged, they became less regularly shaped whencompared to cells at higher densities.

Effect of Cell Density on Na Pump Expression

Confluent BCEC cultures with density ranging from2763 cells/mm2 to 275 cells/mm2 were used for [3H]-ouabain studies. When cell proliferation was stimu-lated with 10 ng/ml hrEGF and removed at least 48hours before determination of ouabain binding, theresultant cell density at confluence was 2396 ±112cells/mm2 (n = 11), with each cell containing 1.54 ±0.03 X 106 pump sites per cell (n = 8). When subculti-vation ratios alone were adjusted, a wide range of finalcell densities at confluence resulted. For example, a1:2 split resulted in 1104 cells/mm2, whereas 1:8 and1:64 splits resulted in 811 cells/mm2 and 308 cells/mm2, respectively (Fig. 3). At these confluent cell den-sities, 3H-ouabain binding revealed 1.71 ± 0.03, 2.10± 0.14, and 2.82 ± 0.05 X 106 pump sites/cell, respec-tively (Fig. 3). Analysis of variance (P = 0.0008) re-vealed a significant effect of using progressively fewercells to initiate subcultures on the number of ouabain-binding sites per cell in the resultant stable, confluentcultures.

Figure 4 shows the number of pump sites per cellplotted as a function of cell density at confluence forthe experiments conducted in this study. As confluentcell density decreased from normal in vivo values,-2700 cells/mm2 to 1000 cells/mm2, the number ofpump sites per cell was independent of cell density (r= -0.25, P= 0.36). In this range, 2700 — 1000 cells/

H



A A

mm2, BCEC exhibited a mean number of 1.92 ± 0.07X 10(l pump sites/cell even though the cell densitydecreased 2.7-fold. However, as confluent cell densitydecreased below 1000 cells/mm2 to a minimum of275 cells/mm2, the number of pump sites per cell

FIGURE 2. Phase-contrasL micrographs of confluent bovinecorneal endothelial cell cultures at different final cell densi-ties: {lop) 2221 cells/mm", {middle) 958 cells/mm2, {bottom)325 cells/mm2. The progressive decrease in the number ofnuclei in the microscopic field is evident. The cells at lowestcell density are slightly retracted, and the outline of one isindicated (anowhecuts).

increased significantly from 1.92 X 10f> to 4 X 10°/cell ( r= -0.77, P< 0.001).

Two features of the culture system, time and theuse of epidermal growth factor (EGF), were furtherexamined to see whether either feature could biasresults. Confluence was usually achieved within 9 daysof subcultivation for cell densities >500 cells/mm2.However, to rule out any effect of time on pump sitedensity, we determined pump site density at conflu-ence (day 9) as well as 3 and 6 days after confluence(days 12 and 15) in one experiment. The differencesbetween the number of pump sites per cell or thenumber of pump sites per mm2 did not differ signifi-cantly between confluence and 3 and 6 days after con-fluence (Table 1).

To rule out an acute effect of EGF on Na pumpsite density, we- added EGF 24 hours before con-ducting ouabain-binding experiments to half a setof matched cultures. The EGF group had a celldensity of 1863 ± 11 (n = 6) and a pump site densityof 2.01 X 106 ± 0.03 sites/cell (n = 8), whereas thecontrol group had a cell density of 2208 X 10(i ±34 (n = 6) and a pump site density of 2.05 X 106

± 0.03 sites/cell (n = 6). The differences betweenthe number of pump sites per cell in the untreatedand treated groups was not significant (P = 0.32).Additionally, analysis of the data presented in Fig-ure 4 on the pump sites per cell in cultures havinga cell density of >1000 cells/mm2, a group com-prised of both untreated and EGF treated cultures,also ruled out any special effect EGF might havehad on pump site density; in this study, pump sitedensity varied only in relation to confluent cell den-sity.

When the total number of pump sites within theentire endothelial monolayer (expressed as pumpsites/mm2) was plotted, a linear decrease (r = 0.94,P < 0.001) in pump site density was observed as thecell density fell from 2700 cells/mm2 to 275 cells/mm2. The relationship between the number of oua-bain-binding sites per unit of surface area and thecell density is illustrated in Figure 5 (for 2700 cells/mm2, 4.65 X 109 binding sites/mm2; for 1000 cells/mm2, 2.04 X 10l) sites/mm2; for 275 cells/mm2, 0.93X 10<J binding sites/mm2). Thus, although hypertro-phied bovine corneal endothelial cells do exhibit



12001:64

uU

-1.0

0.0

1:64 1:8 1:2

FIGURE 3. The relationship of initiating cultures with pro-gressively fewer cells (increasing "split ratios"), final cellnumber in a confluent culture, and the change in Na pumpsites per cell is illustrated (each bar represents the mean ±SEM for eight 3.8 cm2 wells.) As bovine corneal endothelialcells hypertrophy in culture, there is a progressive increasein the number of pump sites per cell.

increased numbers of pump sites per cell in therange of 1000 —* 275 cells/mm2, this increase didnot prevent an overall decline in the number ofpump sites within the endothelial monolayer as thecell density decreased.

Cells / mm

FIGURE 4. The number of Na pump sites per cell in confluentbovine corneal endothelial cell cultures of varying cell den-sity is illustrated for 43 different experiments. The numberof pump sites per cell does not vary statistically in cultureswith a final cell density of 1000 to 2700 cells/mm2 (the lineillustrated is the regression line for cell densities of 1000 to2700 cells/mm2; r = -0.25, P = 0.36). As cell density de-clines from 1000 cells/mm2 to 275 cells/mm2, there is asignificant increase in the number of pump sites per cell (r= -0.77, P< 0.001).

TABLE l. Number of Cells, Pump Sites/Cell,and Pump Sites/mm2 at Confluence andAfter Confluence

10s

106

10y

cells/wellpumps/cellpumps/mm2

1.902.651.33

9

± 0.03± 0.06± 0.03

Days in Culture

1.762.781.29

12

± 0.03± 0.08± 0.04

1.832.831.37

15

± 0.04± 0.06± 0.03

Each point is the mean ± SEM for eight 3.8 cm2 wells. There isno statistical difference between the number of pump sites/cell(P = 0.17) and pump sites/mm2 (P = 0.27) at the three timepoints indicated.

DISCUSSION

By varying the number of cultured bovine corneal en-dothelial cells used to initiate cultures as well as theculture conditions, a wide range of cell densities inconfluent cultures can be achieved. Cell densities inthis culture system ranged from those observed in thenative cornea2'3'23 to those associated with corneal de-compensation.15 The ability to vary corneal endothe-lial density over a wide range, coupled with reports ofaltered corneal endothelial cell pump function associ-ated with decreased cell density,16'2''24'25 prompted usto study how the Na pump density changed as a conse-quence of progressive cell hypertrophy. It is firmlyestablished that the Na pump is the driving force re-quired for vectorial ion and fluid transport to the

Cells / mm

FIGURE 5. The number of pump sites for the entire endothe-lial monolayer, expressed as pump sites per mm2, is com-pared with the cell density in confluent bovine corneal endo-thelial cell (BCEC) cultures. A linear decrease (r = 0.94, P< 0.001) in the number of pump sites per mm2 is observedas the cell number in confluent cultures declines. Eventhough there is a progressive increase in the number ofpump sites per cell as BCEC continue to hypertrophy below1000 cells/mm2 (Fig. 4), that increase is not sufficient toprevent a sustained loss of pump sites in the endothelialmonolayer as a whole.



aqueous humor and is a central element in endothe-lial-mediated deturgesence of the corneal stroma.4 Ad-ditionally, the number of functional Na pump sites onthe endothelial cell surface can be easily quantifiedby the specific binding of 3H-ouabain.2b

The dissociation constant (K,,) for ouabain ob-served in this study (K,i = 6.7 nM) agrees well withthe calculated K,, for cultured bovine corneal endothe-lium in the studies by Savion and Farzame27 (K<| = 1 0nM) and Geroski and Hadley28 (K,, = 25 nM). Themean number of Na pump sites per cell in cultureshaving a density greater than 1000 cells/mm2 was~1.9 X 10'' in this study, which is somewhat greaterthan that reported by Savion and Farzame (0.8 X 106)or Geroski and Hadley (1.4 X 106) for confluent BCECcultures. In general, there is good agreement amongthese three studies of cultured bovine corneal endo-thelium given methodologic differences that exist be-tween each study.

Na pump site density has not been measured innative bovine corneal endothelium. There is a report™that total ATPase activity, as contrasted with ouabain-sensitive activity, is lower in cultured bovine cornealendothelial cells than in native cells. In contrast, how-ever, cultured rabbit corneal endothelial cells havebeen reported to exhibit somewhat higher Na pumpsite density (3.3 X 106)28 than native cells (3.0 X 10fi)so

by ouabain-binding analysis. Regardless of the differ-ences, if any, between Na pump site density in nativeversus cultured corneal endothelium, we think theability to alter corneal endothelial cell density rapidlyin a simple monolayer cell culture system provides auseful tool in which to study the relationship betweencell density and the number of Na pumps site on thecell surface, including any compensatory increase inpump sites per cell as cell density decreases.

A surprising finding in our study was the fact thatthe number of pump sites per cell (~2 X 106) didnot change significantly from 1000 cells/mm2 to 2700cells/mm2, a 2.7-fold increase in cell density. In thefirst study to quantitate the number of pump sites incorneal endothelium, Geroski and Edelhauser30

pointed out that corneal endothelial cells could theo-retically accommodate many more cell membrane Napump sites than were present on native rabbit cornealendothelium. Our study carries that observation fur-ther by suggesting that corneal endothelial cells arecapable of regulating the number of Na pump siteson their surfaces, and they do so over a surprisinglywide range of cell densities. As discussed on the nextpage, where Na pump sites, cell density, and the sizeof the paracellular pathway are compared, it appearsthat a pump site density of ~2 X 106 sites/cell is morethan adequate to compensate for the geometric conse-quence of a larger paracellular pathway for passive ion

movement that coincides with smaller cells and morecells per mm2.

Cultured bovine corneal endothelial cells do ex-hibit an increasing number of Na pump sites per cellas they continue to hypertrophy below a density of1000 cells/mm2. We originally thought that this pro-gressive increase in the number of Na pump sites percell might actually compensate for the overall reduc-tion in the number of cells within the monolayer.However, that hypothesis proved not to be true whenthe number of pump sites within the entire mono-layer, expressed as pump sites per mm2, was calcu-lated. The number of pump sites declined in a gener-ally linear fashion as the number of cells within themonolayer declined. When considered by itself, theincreased number of pump sites per cell attendingmarked cellular hypertrophy cannot be regarded as"compensatory increase" because the pump site den-sity of the monolayer as a whole (that is, the capacityof the endothelium to serve as a fluid pump) contin-ues to decline as the cell number declines. Even whenconsidered in relation to the decreased extent of theparacellular pathway in hypertrophied cells, discussedon the next page, the number of pump sites is reducedin confluent low-density cultures compared to culturesof higher cell density.

Solute accumulation on the basal or stromal sideof corneal endothelium results in increased water up-take by osmotic equilibration.31 The so-called "fluidpumping" mechanism of the endothelium maintainsa dynamic equilibrium by countering this leak by sol-ute-coupled fluid transport to the aqueous driven bythe Na pump.31 Previous studies32 indicated that thepathway for transendothelial water movement is likelyto be by a cellular route, and recent studies33 suggestthat CHIP28-type water channels, which are localizedin apical and basolateral endothelial cell mem-branes,3"1 may account for as much 90% of endothelialcell osmotic permeability. It is also reasonable to spec-ulate that solute movement from the stroma to theaqueous, ultimately driven by the sodium pump, re-sults in water transport through the same cellular wa-ter channels.33 Within physiological limits, it would beexpected that any increase in solute accumulationand, therefore, water uptake toward the stromal sidewould be compensated by increased solute-dependentfluid movement to the aqueous to maintain homeosta-sis. Corneal endothelium is likely to have relativelyhigh paracellular or junctional ionic conductance,consistent with low transendothelial resistance mea-surements35 and discontinuous occluding junctions.3

Significant increases in the linear extent of this para-cellular pathway, such as would occur as cell densityincreases, would be expected to increase passive ionicmovement toward the stroma and, therefore, osmoticwater flow, possibly through CHIP28-type water chan-



nels. In this scenario, greater demands would beplaced on the compensatory fluid pumping system.One possible mechanism for the latter could be anincrease in the number of pumps, although increasedpump activity by previously existing pumps would obvi-ously be an additional mechanism.

To understand further the relationship among Napump sites, cell density, and the "pumping capacity"of the intact endothelial cell monolayer as they relateto the area of junctional contact, we constructed asimple, two-dimensional model of corneal endothelialcell hypertrophy and measured the linear extent ofthe paracellular pathway at different cell densities (seeMaterials and Methods). What we measured corre-sponds to the "area of intercellular space facing theanterior chamber" calculated using a differentmethod by Bourne and Brubaker21 (with which ourmethod of calculation agrees closely) and not the volu-metric extent of the paracellular pathway, the depthof which is reported to be approximately 10 timesthe endothelial cell height over a wide range of celldensities.3 The relationship between the linear (or api-cal) extent of the paracellular pathway (designatedLP, "leak pathway") in a model of either regular orirregular hexagons representing the endothelial cellmonolayer at various cell densities is shown in Figure6. Both hexagonal models exhibit the same degreeof reduction in die planar extent of the paracellularpathway as cells hypertrophy.

Despite the inherent limitations of such a simple

I

i ° irregulregula

i

ir hex.hex.

1000 1500

Cells / mm

2000 2500 3000 3500

2

FIGURE 6. The relationship between cell size and the planarextent of the paracellular ("leak") pathway is plotted forregular hexagons (dosed circles) and irregular hexagons (opencircles) (see Materials and Methods). The change in size ofthe paracellular pathway as cell density increases or de-creases is essentially the same for both geometric models.Changes in pump sites per cell and pump sites per mm2

and the extent of the paracellular pathway all play a rolein determining the pumping efficiency of the endothelialmonolayer.

Cells / mm

FIGURE 7. The relationship between endothelial cell densityand the number of Na pump sites per /xm of the planarleak path depicted in Figure 6 is illustrated. Even thoughhypertrophied cells exhibit increased numbers of pump sitesper cell and the size of the paracellular pathway decreasesas cells hypertrophy, there still is a significant (P < 0.001)decrease in the number of pump sites per fim of the paracel-lular leak path as corneal endothelial cells hypertrophy.

geometric model, it is helpful in unifying the findingsof the present study. For example, even though thenumber of Na pump sites per cell remains the samewhile the apical extent of the paracellular pathway(leak path [LP], Fig. 6) for ionic movement is en-hanced as cell density increases above 1000 cell/mm2,there is, nonetheless, an increase in the number ofpump sites per mm2 and pump sites per leak path asthe cell density increases. Thus, simply increasing thenumber of cells within the monolayer without increas-ing the Na pump site density per cell increases thepotential transport capacity of the monolayer (pumpsites per mm2) and the number of pump sites associ-ated with the linear extent of the paracellular pathwayavailable for passive ion conductance (Fig. 7).

With regard to marked cell hypertrophy, the num-ber of Na pump sites in our paracellular pathwaymodel progressively decreases (Fig. 7) even thoughthe number of pump sites per cell essentially doublesas the cells enlarge to a density of 275 cells/mm2. Theonly degree of compensation that can be inferred byhypertrophied endothelial cells exhibiting progres-sively more pump sites per cell is that if they did not doso, the decrease in pumping capacity of the monolayer(pump sites per mm2, pump sites per paracellularpathway) would be even greater than that observed inthe present study. In fact, the increased number ofpump sites present on markedly hypertrophied cellsmay be a simple reflection of maintaining cellularionic homeostasis in the face of a greatly expandedplasma membrane.

Decreased pumping capacity, as well as decreasednumbers of Na pump sites per unit length of cell mem-

r



brane, have been reported in damaged or diseasedcorneas that exhibit cell hypertrophy, suggesting thatsuch endothelia might not have the ability to increasethe number of Na pump sites on their surfaces to theextent of that observed in this study.'* Such studiesprovide ample warning about generalizing the Napump site density-cell hypertrophy relationships ob-served in this study. Encouragingly, however, Napump sites per cell calculated from a recent study offeline corneal endothelial cell repair by Bourne et al19

do fit with the Na pump sites per cell-cell densityrelationship observed with the cultured bovine cor-neal endothelial system.

The changes in Na pump site density observedin this easily manipulated corneal endothelial culturesystem emphasizes the usefulness of cultured bovinecorneal endothelium in studying this important ele-ment of the corneal endothelial ion and water trans-porting system.

Key Words

NaK-ATPase, Na pump, paracellular pathway, ouabain, bo-vine corneal endothelium, cell culture

References

1. Johnson MC, Noden DM, Hazelton RD, et al. Originsof avian ocular and periocular tissue. Exp Eye Res.1979;29:27-43.

2. Waring GO, Bourne WH, Edelhauser HF, Kenyon KR.The corneal endothelium. Ophthalmology. 1982;89:531-590.

3. Klyce SD, Beurman RW. Structure and function of thecornea. In: Kaufman HE, Barron BA, McDonald MB,Waltman SR, eds. The Cornea. New York: ChurchillLivingstone; 1988:3-54.

4. Edelhauser HF, Geroski DH, UbelsJL. Physiology. In:Smolin G, Thoft RA, eds. The Cornea. 3rd ed. Boston:Little, Brown; 1994:25-46.

5. Dohlman CH. Corneal edema. In: Albert DM, Jako-biec FA, eds. Principles and Practice of Ophthalmology.

Vol. 1. Philadelphia: WB Saunders; 1994:245-256.6. Kaye GI, Tice L. Studies on the cornea: V: Electron

microscopic localization of adenosine triphosphataseactivity in the rabbit cornea in relation to transport.Invest Ophthahnol. 1966;5:22-31.

7. Leuenberger P, Novikoff A. Localization of transportadenosine triphosphatase in rat cornea. J Cell Biol.1974;60:721-731.

8. Soltau JB, Zhou LX, McLaughlin BJ. Isolation ofplasma membrane domains from bovine corneal en-dothelial cells. Exp Eye Res. 1993;56:115-120.

9. Fischbarg J, Hernandez H, Liebovitch LS, KoniarekJP. The mechanism of fluid and electrolyte transportacross corneal endothelium: Critical revision and up-date of a model. Curr Eye Res. 1985;4:351-360.

10. Jentsch TJ, Keller SK, Wiederholt M. Ion transportmechanisms in cultured bovine corneal endothelialcells. Curr Eye Res. 1985;4:361-369.

11. Guggenheim JA, Hodson SA. Localization of Na+/K+-ATPase in the bovine corneal endothelium. BiochimBiophys Ada. 1994; 1189:127-134.

12. Trenberth SM, Mishima S. The effect of ouabain onthe rabbit corneal endothelium. Invest Ophthalmol.1968; 7:44-52.

13. Kaye GI, Silbey RC, Hoefle FB. Recent studies on thenature and function of the corneal endothelial bar-rier. Exp Eye Res. 1973; 15:585-613.

14. Geroski DH, KiesJC, Edelhauser HF. The effects ofouabain on endothelial function in human and rabbitcorneas. Curr Eye Res. 1984;3:331-338.

15. Watsky MA, McCartney MD, McLaughlin BJ, Edel-hauser HF. Corneal endothelial junctions and the ef-fect of ouabain. Invest Ophthahnol Vis Sri. 1990;31:933-941.

16. Mishima S. Clinical investigations of the corneal endo-thelium. Am J Ophthalmol. 1982;93:l-29.

17. Van Horn DL, Hyndiuk RA. Endothelial wound repairin the primate cornea. Exp Eye Res. 1975;21:113-124.

18. Van Horn DL, Sendele DD, Seidman S, Buco PJ. Re-generative capacity of the corneal endothelium in therabbit and cat. Invest Ophthalmol Vis Set. 1977; 16:597-613.

19. Bourne WH, Nelson LR, Buller CR, et al. Long-termobservation of morphologic and functional featuresof cat corneal endodielium after wounding. InvestOphthalmol Vis Sci. 1994; 35:891-899.

20. MacCallum DK, Lillie JH, Scaletta LJ, et al. Bovinecorneal endothelium in vitro. Exp Cell Res. 1982;139:1-13.

21. Bourne WH, Brubaker RF. Decreased endothelial per-meability in transplanted corneas. Am f Ophthahnol.1983;96:362-367.

22. MacCallum DK, Lillie JH, Meyer RF, Lopatin DE. Cellculture of corneal endothelial cells exhibiting highand low mitotic capacity. Anal Rex. 1987;218:82. Ab-stract.

23. Bahn CF, Glassman RM, MacCallum DK, et al. Postna-tal development of corneal endothelium. Invest Oph-thalmol Vis Sci. 1986;27:44-51.

24. Geroski DH, Matsuda M, Yee RW, Edelhauser HF.Pump function of the human corneal endothelium.Ophthalmology. 1985; 92:759-763.

25. McCartney MD, Robertson DP, Wood TO, McLaug-hlin BJ. ATPase pump site density in human dysfunc-tional corneal endothelium. Invest Ophthalmol Vis Sci.1987;28:1955-1962.

26. Ernst SA, Mills, JW. Autoradiographic localization oftritiated ouabain-sensitive sodium pump sites in iontransporting epithelia. J Histochem Cytochem.1980;28:72-77.

27. Savion N, Farzame N. Characterization of the Na, KATPase pump in cultured bovine corneal endodielialcells. Exp Eye Res. 1986;43:355-363.

28. Geroski DM, Hadley A. Endothelial Na/K ATPasepump site density in cultured bovine and rabbit cor-neal endothelium. Invest Ophthalmol Vis Sri. ARVO Ab-stracts. 1992; 33:1403.

29. Whikehart DR, Montgomery B, Hafer LM. Sodiumand potassium saturation kinetics in the Na+K+ AT-



Pase in the plasma membranes from corneal cndothe-lium: Fresh tissue vs. tissue culture. Curr Eye Res.1987;6:709-717.

30. Geroski DH, Edelhauser HF. Quantitation of Na/KATPase pump sites in the rabbit corneal endothelium.Invest Ophthalmol Vis Sri. 1984;25:1056-1060.

31. Maurice DM. The location of the fluid pump in thecornea. J Physiol (Lond). 1972;221:43-54.

32. Hodson SA, Lawton DM. The apparent reflexion coef-ficient of the leaky corneal endothelium to sodiumchloride is about one in the rabbit. J Physiol (Lond).1987; 385:97-106.

33. Echevarria M, Kuang K, Iserovich P, et al. Cul-tured bovine corneal endothelial cells express CHIP28water channels. Am J Physiol. 1993;265:C1349-CI355.

34. Nielsen S, Smith BL, Christensen El, Agre P. Distribu-tion of the aquaporin CHIP in secretory and resorp-tive epithelia and capillary endothelia. Proc Nail AcndSri USA. 1993;90:7275-7279.

35. Narula P, Xu M, Kuang K, Akiyama R, Fischbarg J.Fluid transport across cultured corneal endothe-lial cell monolayers. Am J Physiol. 1992;262:C98-C103.


nak-atpase pump sites in cultured bovine corneal endothelium of

Documents