correlation of corneal endothelial pump site density, barrier function

September 1985 Vol. 26/9

Investigative Ophthalmology& Visual Science

A Journal of Dosic and Clinical Research

Articles

Correlation of Corneol Endotheliol Pump Site Density,Barrier Function, and Morphology in Wound Repair

Richard W. Yee, Dayle H. Geroski, Mamoru Morsudo,* Eugene J. Champeau,Lorry A Meyer, and Henry F. Edelhauser

After transcorneal freezing, physiologic function (pump and barrier) of the regenerating rabbit cornealendothelium was evaluated and compared with the morphologic differentiation that occurs duringwound healing. Endothelial pump function was investigated utilizing the specific binding of tritiatedouabain to endothelial Na+/K+ ATPase (pump sites); the permeabilities of isolated de-epithelializedcorneas to labeled inulin and dextran were measured to determine endothelial barrier function.Endothelial recovery after transcorneal freezing can be described as a three-stage process. Stage oneis characterized by the establishment of an initial coverage of the wound by pleomorphic spindle-shaped cells which form a functional but incomplete barrier and minimal pump site density. In stagetwo, the cells assume a flattened configuration consisting of irregular polygons and establish nearlynormal pump capacity. In stage three, a significant remodeling of the monolayer continues despitethe layer's early physiologic recovery. Invest Ophthalmol Vis Sci 26:1191-1201, 1985

The regenerative capacity of the rabbit endotheliumafter various forms of injury is well known.1"9 Duringunstressed conditions, no mitosis occurs in the rabbitendothelial monolayer.1'3 However, when the endo-thelium is damaged, cells at the wound peripheryeffect healing by migration and mitosis. His-tologic48'1011 and ultrastructural studies6-912"16 havedocumented the morphologic appearance of the en-dothelium during the regeneration process.

Although endothelial morphology during woundhealing is well documented, little is known about theability of the corneal endothelium to regain its normalphysiologic function after wounding. Khodadoust andGreen5 and Van Horn et al,6 using corneal thicknessas an index of endothelial function, followed the time

From the Departments of Physiology and Ophthalmology, MedicalCollege of Wisconsin, Milwaukee, Wisconsin.

Supported in part by grants RO1-EY-OO933, R01-EY-05609,T32-EY-07016, and P3O-EY-O1931 from the National Institutes ofHealth. Dr. Edelhauser is a 1982 Olga K. Wiess Research Scholarof Research to Prevent Blindness, Inc.

* Dr. Mamoru Matsuda is a Visiting Scientist from the Depart-ment of Ophthalmology, Osaka University Medical School, Osaka,Japan.

Submitted for publication: December 12, 1984.Reprint requests: Henry F. Edelhauser, PhD, Department of

Physiology, Medical College of Wisconsin, 8701 Watertown PlankRoad, Milwaukee, WI 53226.

course of corneal swelling and deturgescense aftertranscorneal freezing of rabbit corneal endothelium.They reported that the functional capacity of theregenerated endothelium lags behind the apparentmorphologic repair of the injured area by severaldays. Minkowski et al,9 in addition to observingcorneal thickness, measured endothelial barrierfunction using fluorophotometry after transcornealcryoinjury. They suggested endothelial pump functionrecovered after endothelial permeability had returnedto normal.

The corneal endothelium functions both as a pump,which requires metabolic energy to remove fluid fromthe stroma, and as a physical barrier to retard aqueousflow back into the stroma. While corneal barrier andpump functions are reflected by corneal thickness,17

alterations in either or both of these functions canaffect thickness. Previous studies have demonstratedthat endothelial permeability can be measured invitro using radioactive labeled substances of 3H-inulinand 14C-dextran.18"20 More recently, measurement ofendothelial Na+/K+ ATPase "pump sites" has pro-vided a new parameter for the in vitro measurementof endothelial transport capacity.21 Using these meth-ods it is possible to independently quantitate bothbarrier and pump functions. The functional devel-opment of the regenerating endothelium can thus beevaluated and compared with the morphologic differ-

1191

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933125/ on 02/12/2018

1192 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / September 1985 Vol. 26

entiation that occurs during wound healing. Thepresent study was undertaken to correlate the reestab-lishment of endothelial function (pump site densityand permeability) and morphology during the woundhealing process in the rabbit cornea.

Materials and Methods

Corneal Freezing

Two- and six-mm diameter brass probes, machinedto approximate the curvature of the cornea, wereused to wound the corneas of 2-2.5 kg New ZealandWhite rabbits. The use of rabbits in this study adheresto the ARVO Resolution on the Use of Animals inResearch. After an intramuscular injection of ketam-ine HC1 (10-15 mg/kg) and xylazine (6 mg/kg) tosedate the rabbit, two drops of proparacaine wereapplied topically to the cornea. The brass probes werecooled in liquid nitrogen and placed on the centralcorneal surface for 5 sec. The 5-sec application wasselected because it provided consistent endothelialinjury with the least amount of morbidity (inflam-mation, retrocorneal membrane, and stromal scar-ring). After freezing, the eyes were left proptosed untilthe ice ball was completely thawed. The total endo-thelial area damaged by the freezing procedure wasapproximately 7 mm2 and 39 mm2, respectively, forthe 2- and 6-mm probes.

Measurement of Corneal Thickness

Central corneal thickness was measured using amodified Haag-Streit pachymeter.22 Prior to freezingand at regular intervals from 1-30 days after trans-corneal freezing, corneal thickness was taken as themean of three readings.

Measurement of Pump Site Density

At various times after freezing, rabbits were killed,their eyes were enucleated, and the corneas wereexcised. Matched pairs of de-epithelialized corneaswere incubated in tritiated ouabain for the determi-nation of Na+/K+ ATPase pump site densities, usingmethods previously described.21 Briefly, corneas wereincubated in 10~7 M carrier-free 3H-ouabain (18 jzCi/mmol), with one cornea of each matched pair receiv-ing additionally 10~4 M of unlabeled glycoside. Pre-vious studies21 have shown that, at this concentrationof glycoside, Na+/K+ ATPase sites are 80% saturated,and nonspecific binding accounts for less than 2% ofendothelial ouabain uptake. After 3 hr of incubation,the central (wounded) area of each cornea was re-moved with an 8-mm diameter trephine, and traceruptake was measured in the endothelium isolatedfrom each of these corneal buttons. Ouabain binding

was calculated as the difference in tracer uptake inthe absence and presence of 10~4 M unlabeled gly-coside.

Measurement of Barrier Function

At various times after a 6-mm transcorneal freeze,rabbits were killed and their eyes were enucleated.Corneas were excised and mounted as matched pairsin Lucite-block perfusion chambers as previouslydescribed.23"25 Glutathione bicarbonated Ringer's(GBR) was added to the reservoirs bathing either sideof the cornea. Constant mixing and pH (7.3) of theGBR were maintained by an airlift siphon using 95%air and 5% carbon dioxide. The temperature of theperfusion chambers was maintained at 37°C by acirculating water bath. The exposed corneal area was1.1 cm2 in all experiments. One cornea of a matchedpair served as the untreated control while its matereceived the freeze injury as previously described. Inadditional paired experiments designed to measuremaximal permeability, the endothelium of one corneawas completely removed with a cotton-tipped appli-cator while its mate served as the untreated control.

GBR solution dual-labeled with 1.0 /iCi/ml dextran(carboxyl-14C, MW 50,000-70,000, New EnglandNuclear; Boston, MA) and 1.05 A*Ci/ml inulin (3H,MW 5,000, New England Nuclear) was added to theendothelial surface of the corneas. After addition ofthe isotopes, tissues were allowed to equilibrate for 1hr. At the end of the equilibration the chamber facingthe de-epithelialized surface or "cold" chamber (1.8-ml volume) was rinsed with 2.0 ml of GBR. A second2.0-ml rinse was collected immediately in a taredvial, weighed, and counted to provide a backgroundor residual count. At 30-min intervals for 3 hr, the"cold" chamber was evacuated, collected, and ana-lyzed in a similar manner. After the final collectionof the "cold" chamber, the "hot" chamber wasevacuated, collected, and assayed. All samples werecounted by liquid scintillation, and the counts werecorrected for the 10.5% 14C energy overlap into the3H channel and the 0.5% 3H overlap into the I4Cchannel. The permeability coefficient, Ktrans, for eachtracer was calculated according to Maffly et al26 asfollows:

K-transincrease in counts on the unlabeled sideconcentration of counts on labeled side

X area of membrane X time

Morphologic Evaluation

Rabbits were killed at various intervals after trans-corneal freezing. The corneas were excised and stained


No. 9 PUMP AND BARRIER FUNCTION FOLLOWING ENDOTHELIAL INJURY / Yee er ol. 1193

6mm ENDOTHELIAL WOUND

Fig. 1. Changes in cornealthickness and pump sitedensity with time after trans-corneal freezing of centralcornea with a 6-mm brassprobe. (Each point representsthe mean ± SEM; pump sitedensity was measured on 5-6 isolated endothelial sheetsat each point.)

I I.2Hin

1.0-

i -S0-\

< .60-LJ

.40-|

Freeze ^ ^ Complete coverage

0 1 2 3 4 5 6 7 101 / / ' 7

14 21 30DAYS

with 1% alizarin red and 0.25% trypan blue.27 Thecentral wounded area was photographed. An addi-tional group of frozen corneas was fixed at similarintervals in 2.7% phosphate-buffered glutaralde-hyde and processed for scanning electron micros-copy (SEM).28

•9.0

8.0

•7.0

•6.0

•5.0

4.0

-3.0

PU

MP

<

D31

IS

3"3ro

2.0

Specular Microscopy and Computer-Assisted Morphometry

At various times after freezing, wide-field specularphotomicrographs of the central corneal wound weretaken with a Keeler-Konan specular microscope

Fig. 2. Top, Morpho-logic appearance of theendothelium by specularmicroscopy immediatelyafter 6-mm wounding(• denotes wound; bar= \00 fim). Bottom, Mor-phologic appearance ofthe central endothelialwound at 2, 3, and 5days after wounding,'us-ing alizarin red and try-pan blue stain (* denoteswound).

IMMEDIATELY AFTER WOUNDING

t2 days 3 days 5 days


1194 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Seprember 1985 Vol. 26

Fig. 3. Scanning electron microscopic appearance of the central endothelium after 6-mm transcorneal freezing: A, I day; B, 2 days; C, 3days; D, 4 days; E, 7 days; F, 30 days (bar = 50 ^m, A-D; bar = 100 //m, E and F).

(Keeler Instruments Inc.; Broomall, PA) using Tri-Xpan film (ASA 400, Eastman Kodak Company;Rochester, NY). The black and white film was devel-oped in Acufine (Acufine, Inc.; Chicago, IL) for 5min at 75°F. Based on cell boundary clarity, the best

frame of each eye at various intervals after freezingwas selected and enlarged (X400) on grade 4 Koda-bromide RC paper (Eastman Kodak Company;Rochester NY). Prior to printing, a calibration nega-tive of a micrometer scale (0.01 mm) was placed in



the enlarger to assure accurate cornea-to-print mag-nification.

One hundred individual cells in a cluster perspecular micrograph were outlined and numberedconsecutively to minimize sampling and analyzingerrors.29 Individual cells were digitized by touchingcell apices with a Hewlett Packard 9111A graphicstablet pen (Hewlett-Packard Company; Brookfield,WI). The coordinates entered onto the digitizer tabletwere analyzed by a Hewlett Packard 85-B computer(Hewlett-Packard Company; Brookfield, WI) usingendothelial analysis software as previously described.30

Computer-assisted analysis of each individual celldetermined cell size parameters (perimeter, mean cellarea, standard deviation of the cell area, cell density,and coefficient of variation in cell area) and cell shapeparameters (the relative frequency of each cell shape)at various intervals after transcorneal freezing. Thecell shape can be described by the number of sidesof each cell, and the frequency of each cell shape wascounted automatically by numbering each apex thatpermitted quantitation of variation in cell shape(pleomorphism). Cell density was calculated by divid-ing 106 nn\2/mm2 by the mean cell area Ozm2).31 Thecoefficient of variation (CV) was calculated by dividingthe standard deviation of the cell area by the meancell area. The CV is a dimensionless index indepen-dent of cell area which provides a quantitative mea-surement of cell area variation (polymegethism).32

Results

6-mm Endothelial Wound

After transcorneal freezing, corneal thickness (mean± SEM) rapidly increased from 0.40 ± 0.004 mm to1.2 ±0.005 mm at day 2 (Fig. 1). This markedincrease in thickness was due to loss of the endothelialbarrier function as the cells sloughed from Descemet'smembrane. Associated with the endothelial cell loss,the pump site density also decreased from a prefreezevalue of 8.2 ± 0.6 X 109 sites/mm2 to 2.3 ± 0.3X 109 sites/mm2 at day 2 (Fig. 1). It is important tonote that, after transcorneal freezing, the time pointof minimum pump site density is also the point atwhich corneal thickness was greatest.

Endothelial morphology early in the regenerativeperiod was evaluated by preparing corneal flat prep-arations stained with 1% alizarin red and 0.25%trypan blue,27 and by SEM since specular microscopyduring this period was not possible because of markedcorneal edema. Immediately after transcorneal freez-ing, all endothelial cells over the frozen area werelost except for some areas covered with degeneratingendothelial cells and debris (Fig. 2, top). The cellsoutside the wounded area appeared not to be damagedby the freeze, nor was any damage observed in

s<

6.50

6.00

5.50

5.00

4.50

4.00

3.50

•SCRAPED

2 4TIME (DAYS)

B0.550

0.500

0.450

0.400

0.350

0.300

0.250

SCRAPED

0 1 2 4 7TIME (DAYS)

Fig. 4. Changes in permeability coefficient (K,rans) for inulin (A)and dextran (B) after transcorneal freezing of central endotheliumwith 6-mm brass probe (each point represents the mean ± SEM of4-8 isolated corneas; • denotes significance from prefreezing values,P < 0.05).

Descemet's membrane (Figs. 2, 3A). The diameter ofthe wounded area was approximately 1 mm largerthan the diameter of the 2- and 6-mm probes.

At both 1 and 2 days after freezing, both vitalstaining and SEM demonstrated that the denudedarea was smaller than immediately after freezing. Thecells at the wound margin were elongated andstretched toward the center of the wound (Figs. 2,3A). Between 2 and 3 days, the 6-mm wound wascompletely covered with pleomorphic spindle-shapedcells having extensive cytoplasmic processes whichoverlapped neighboring cells (Figs. 2, 3B, 3C).

Within the 3- to 7-day time period (in the 6-mmwounded cornea), the random pleomorphic spindle-shaped cells rapidly flattened and retracted as amonolayer of cells forming a confluent polygonalmosaic (Figs. 2, 3C-E). These morphologic transfor-


1196 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / September 1985 Vol. 26

Fig. 5. Specular microscopic appearance of the central endothelial monolayer after 6-mm transcorneal freezing. A, prefreeze; B, day 7; C,

day 14; D, day 30 (bar = 50 urn).

mations could not be visualized in vivo with specularmicroscopy because of stromal edema, but they couldbe observed with vital staining (Fig. 2) and SEM(Figs. 3A-D).

Once Descemet's membrane was covered by thepleomorphic spindle-shaped cells on days 2 to 3,barrier function was reestablished as measured by theendothelial permeability to inulin and dextran (Figs.4A, 4B). Prior to wounding, the endothelial perme-ability, Ktrans (mean ± SEM), for inulin and dextranwere 3.57 ± 0.163 X 106 cm/sec and 0.270 ±0.012X 106 cm/sec, respectively (Figs. 4A, 4B). For com-parative purposes, complete removal of the endothe-

Table 1. Cell size analysis (x ± SEM)

Woundsize

6 mm

2mm

Day afterfreezing

pre * '• 7 - 1 '

• . 1 4 -

, 30

pre457

1430

Celt density(cells/mm2)

3646 ± 9 5 ••2237 ±.326*246J''±202*'''3042. ± 188

3591 ± 4 31883 ± 83*2252 ± 294*2835 ± 65*3227 ± 2313500 ± 4 1

Mean cellarea (tun2)

274.6 ± 7.0"465-.8±65*• 409.1 '±'33.6*

331.5 ±-.21.7 •

278.6 ± 3.3532.2 ± 23.4*459.1 ± 57.7*353.1 ±8.2*315.1 ±24.4285.8 ± 3,4

Coefficient ofvariation

. • 0.19 ±0:014;0.36'±-0.03*

. ' 0.33 ±-0.047*"• > '0.16 ±0.007

0.17 j0.33 j0.37 d0.34 d0.29 d0.20 H

t 0.009t 0.023*t 0.021*t 0.010*t 0.036t 0.037

* Statistical significances refer to comparison with normal values, P < 0,05.

Hum by scraping increased the Ktrans to 6.40 ± 0.33X 10"6 cm/sec for inulin and 0.538 ± 0.051 X 10"6

cm/sec for dextran. After 6-mm transcorneal freezing,the Ktrans at day 1 increased to 5.3 ± 0.20 X 10~6

cm/sec for inulin (Fig. 4A) and 0.444 ± 0.03 X 10"6

cm/sec for dextran (Fig. 4B). By day 2, as thewounded area was covered with migrating and divid-ing cells, endothelial barrier function was partiallyrestored as manifest by a decrease in the endo-thelial permeability to both inulin and dextran(Figs. 4A, 4B).

Although barrier function was partially restored byday 2, the pump site density remained low (Fig. 1),Pump function did not begin to increase until com-plete endothelial coverage was achieved at day 2.Pump site density then rapidly increased as thespindle-shaped endothelial cells reorganized into po-lygonal cells (Figs. 2, 3D). During this period (days3-7) pump site .density increased from 3.0 ± 0 . 3

;X\107mm2 to 6.6 ± 0.7 X.109/mm2.By day 7, the endothelial cells had reorganized

into a complete polygonal array (Figs. 3E, 5B) andendothelial barrier function was not significantly dif-ferent from prefreezing values (Figs. 4A, 4B). Directlycorrelated to the polygonal endothelial morphologywas the return of the endothelial pump site density(Fig. 1). With both the barrier and pump functions



Table 2. Cell shape analysis: percentage (x ± SEM) of cells having 4-9 sides at various times after freezing

Woundsize

6 mm

2 mm

Day afterfreezing

pre7

1430

pre457

1430

4-sided

02.0 ± 00.5 ± 0.5

0

05.0 ± 2.04.7 ± 2.73.3 ± 0.9*1.3 ±0.60.7 ± 0.7

Percentage

5-sided

15.3 ±0.732.3 ± 2.3*27.5 ± 3.5*15.3 ±0.3

12.7 ±0.327.5 ± 2.5*27.0 ± 3.5*25.0 ± 1.0*25.8 ± 1.6*16.3 ± 0.7

• of cells having designated number of sides

6-sided

72.0 ± 1.537.3 ± 4.7*48.0 ± 8.0*70.3 ± 0.7

75.3 ± 0.336.0 ± 0*36.0 ± 1.2*45.0 ± 1.7*48.8 ± 2.4*67.0 ± 1.5

7-sided

12.3 ±2.324.3 ± 2.4*19.5 ± 0.5*14.3 ± 0.9

11.3 ±0.327.5 ± 1.5*27.0 ± 2.5*22.0 ± 0*19.5 ± 1.214.3 ± 0.7

8-sided

0.33 ± 0.32.7 ± 1.24.0 ± 3.0

0

04.0 ± 1.0*4.0 ± 1.5*3.3 ± 1.3*3.0 ± 0.9*1.3 ±0.7

9-sided

01.3 ±0.3*0.5 ± 0.5

0

00

1.0 ±0.61.3 ±0.3*1.8 ± 1.1

0.33 ± 0.3

Statistical significances refer to comparison with normal values, P < 0.05.

of the endothelium restored to control values, therewas a concomitant decrease in corneal thickness.

The return of corneal thickness to 0.7 mm enabledendothelial specular microscopy to be performed (Fig.5B). Computer morphometric analysis of the centralendothelium showed that significant remodeling ofthe individual cells occurred after a confluent mono-layer was established (Tables 1, 2; Fig. 6). It was notpossible to perform quantitative computer analysis ofthe regenerating cells prior to the establishment ofthe monolayer because of the morphological natureof the early regenerating cells (Figs. 2, 3A-D).

Although the endothelial monolayer had been es-tablished at day 7, endothelial cell density and thefrequency of hexagons were still decreased while thecoefficient of variation in cell area (CV) and relativefrequency of non-six-sided cells were increased (Tables1, 2). Analysis of endothelial specular micrographs atday 14 (Fig. 5C) showed that endothelial cell densitywas still decreased below the prefreeze values, the CVwas high, and the original hexagonal array was sig-nificantly compromised (Tables 1, 2; Fig. 6). However,by day 30 complete endothelial remodeling had oc-curred. Cell density, CV, and percent hexagonal cells(70%) had all returned to their respective prefreezevalues (Tables 1, 2; Fig. 6).

2-mm Endothelial Wound

After transcorneal freezing, there was a twofoldincrease in corneal thickness on day 1 and a significantdecrease in the number of measured pump sites,which corresponded to the area of lost endothelialcells (Fig. 7). By day 2, corneal thickness decreased.This decrease was associated with the reestablishmentof the endothelial barrier when the endothelial cellsmigrated from the wound edge (Fig. 8A). Completecoverage of Descemet's membrane occurred earlierin the 2-mm wound (between 1 and 2 days) than in

4000

3 0 0 °2000

IOOO

B

^T l . 2

0.80 ^

•J-0.40 [>!

to7 14TIME (days)

30

4 0

.30

.20

.10

$

-H.2

0.80

Q£

7 14TIME (days)

30

to5:Q

t>o

7 0

50

30

10

s

0

T ^S

7 14TIME (days)

m 8jr ^

<"' j 12 >

0.80 >j

0.40 ^

^>to

30

Fig. 6. Changes in the cell density, coefficient of variation in cellarea (polymegethism) and hexagonality (pleomorphism) after trans-corneal freezing of central cornea with 6-mm brass probe (dottedcurve is the corneal thickness data from Figure 1; each pointrepresents the mean ± SEM of 6-8 determinations).


1198 INVESTIGATIVE OPHTHALMOLOGY G VISUAL SCIENCE / September 1985 Vol. 26

2mm ENDOTHELIAL WOUND

J. 1.0-CO

£.90ox .80-

< .60-

I.40H

Mean±SE

_i Pump site density

Comeal thickness

Freeze ••Complete coverage

0 I 2 3 4 5 6 7 10DAYS

14

-II-4

21 30

•9.0

•8-0

7.0 C

"0

6-0 wHm

5.0 g

4.0

3.0

2.0

Fig. 7. Changes in comealthickness and pump site densitywith time after transcornealfreezing of central cornea witha 2-mm brass probe (each pointrepresents the mean ± SEM;pump site density was measuredon 5-6 isolated endothelialsheets at each point).

the 6-mm wound. Although the decrease in thicknessoccurs as the barrier is reestablished, the pump sitedensity remained decreased until after the reestablish-ment of the monolayer, as in the 6-mm wound (Fig.7). By day 4, the endothelium was covered by large,variable-shaped polygonal cells. Though cell densitywas low (Table 1), corneal thickness decreased to 0.5mm. The large pleomorphic polygonal cells are illus-trated in Figure 8B. Quantitation of these cells bycomputer-assisted morphometry showed that only36% of the cells are hexagonal compared with 75%for the prefreeze values (Table 2).

As endothelial cells became polygonal (day 4),there was a marked increase in the number of pumpsites. Pump site density returned to normal by day5. During the period from day 7 to 30, the cornealendothelium remodeled to the extent that 67% of theendothelial cells returned to their hexagonal shape,and cell density returned to its normal prefreeze value(Tables 1, 2). There was no further change in endo-thelial pump site density during this time. It isinteresting to note that endothelial morphology aftera 2-mm wound at day 2 is similar to the morphologyafter a 6-mm wound at day 4.

DiscussionResults of this study indicate that, after transcorneal

freezing, the normalization of central corneal thick-ness, endothelial pump site density, and endothelialpermeability precede the establishment of a normalendothelial mosaic. The functional characteristics of

the regenerating endothelium correlate with its mor-phologic appearance (Fig. 9).

Though a 2-mm wound heals more quickly thana 6-mm wound, the general scheme of wound healingis applicable to both small and large wounds. Basedon endothelial morphology, the wound healing processafter a 2-mm and 6-mm freezing can be divided intothree stages: I, the denuded area is covered by dividingand migrating endothelial cells; II, the spindle-shapedcells form a flattened monolayer and assume a pleo-morphic polygonal pattern; and III, the cells graduallyremodel to a normal hexagonal configuration (Fig.9). Corresponding to these three morphologic stagesof wound healing are changes in corneal thickness,Na+/K+ ATPase pump site density, and barrier func-tion.

Stage I begins immediately after freezing withcomplete destruction and loss of the endothelial cellswithout rupture of Descemet's membrane. As longas a wound is present, endothelial barrier function iscompromised as is evidenced by a rapid rise incorneal thickness and endothelial permeability. Thepassage of inulin and dextran occurs freely throughthe endothelial defects if these defects are no smallerthan the effective molecular radius of inulin (14 A)and dextran (38 A).19 Pump site density declines toa minimum at 1-2 days after freezing, and it is alsoat this time that corneal thickness is greatest. Thereduction in endothelial transport capacity is propor-tional to the damaged area produced by either the2- or 6-mm freezing.


No. 9 PUMP AND DARRIER FUNCTION FOLLOWING ENDOTHELIAL INJURY / Yee er ol. 1199

Fig. 8. Scanning electron microscopicappearance of the central endothelium after2-mm transcorneal freezing. A, 2 days; B,4 days (bar = 50 pm).

Coverage of the central wounded area occurs bycell migration and mitosis in the rabbit cornea.1"9

Studying endothelial incorporation of tritiated thy-midine, previous authors have demonstrated endo-thelial DNA synthesis within 24 hr after injury.Endothelial maximal incorporation is seen at 24-48n r 6,33 once a confluency of cells is formed, DNAsynthesis and cell migration stop and contact inhibi-tion of cell growth or density-dependent growth in-hibition occurs. The migrating and proliferating fi-broblastic-like cells appear to function solely to coverthe denuded area and to reestablish a barrier. Hence,no increase in pump site density occurs during stageI of healing. In contrast, corneal thickness and en-

dothelial permeability begin to decrease as the woundbecomes smaller, and they approach prefreezing valueswhen Descemet's membrane is completely coveredwith cells. The regenerating cells form primary con-tacts with their neighboring cells by lateral expansion.15

Although large intercellular spaces exist, apical cyto-plasmic extensions bridge over these spaces formingsecondary contacts.15 Together the primary and sec-ondary contacts form a functional but incompleteendothelial barrier. Permeability studies using asmaller molecule such as fluorescein (4A effectiveradius) have been performed by Minkowski andassociates and presumably would reflect the endothe-lial barrier function more accurately. Unfortunately,


1200 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / September 1985 Vol. 26

SUMMARYCORNEAL MORPH-

STA6E DAY STATE OF HEALING THICKNESS PUMP BARRIER OLOGY

0 - 1 d a y w o u n d p r e s e n t + +

. o , cel ls s l i d i n g , m i t o s i s . . . .I -2days wound | jn size ++++

wound3 days completely covered +++

4 - 6 d a y s establishment7 days of monolayer NL NL

Fig. 9. Summary of thestages of endothelial woundhealing and correlation be-tween corneal thickness,pump site density, perme-ability, and morphology af-ter transcorneal freezing(changes relative to prefreezevalues are expressed as fol-lows: 1 and - = decrease; += increase; NL = normal, ie,no change).

remodeling ofI E 8-30 days monolayer NL NL NL

the authors were not able to evaluate the endothelialbarrier function during the first 7 days postcryoinjurydue to stromal edema which precluded fluoropho-tometry.9

Stage II of the regenerative process does not beginuntil the wound is repopulated with cells and presum-ably after contact inhibition has occurred. Once anendothelial layer is reformed, the cells become lessspindle-shaped, and over the next several days anintact endothelial mosaic comprised of pleomorphicpolygonal cells is formed. During this time period, asignificant cellular differentiation occurs as intercel-lular spaces narrow and apical junction and lateralintercellular contacts are reestablished.1415 Associatedwith these morphologic changes are also changes atthe membrane level as evidenced by the increase inpump site density to prefreeze values.

Although endothelial permeability and pump sitedensity are near normal once the endothelial mosaicis present, corneal thickness remains elevated. Thismay be due to the stromal damage with the loss ofstromal proteoglycans and glycoproteins that occursduring corneal edema.34"36 Similarly, pump site den-sity may be normal, but total pumping capacity(turnover) may still be decreased at this time.

Stage III of endothelial healing begins as a flattenedbut pleomorphic endothelium is established. This isaccompanied by the gradual return of pump sitedensity, permeability, and corneal thickness to controlvalues. While these parameters of endothelial functionare approaching normal, the regenerated mosaic pat-tern is still morphologically quite abnormal. Duringthis healing period, the cells undergo significant re-

modeling. As cell density increases, the cells becomemore uniform in size (reduced CV) and assume anormal hexagonal configuration as shown by theincrease in the frequency of hexagonal cells. By 30days, all the morphologic parameters have returnedto prefreeze values.

These observations demonstrate that rapid func-tional recovery of the regenerated endothelium occursonce a monolayer is reestablished over the injuredarea. Although the newly regenerated monolayer hasnormal endothelial function (permeability and pumpsite density) early in the course of wound healing(stage II), normal structural appearance of the mono-layer is not seen until later (stage III).

Key words: corneal endothelium, Na/K pump site density,barrier function, transcorneal freezing, morphology, woundhealing, sodium potassium adenosine triphosphatase

Acknowledgments

The authors thank Mr. Tony Kuzma for technical assis-tance with the specular photographic processing and Mrs.Barbara Olson for excellent secretarial assistance.

References

1. Binder RF and Binder HF: Regenerative processes in theendothelium of the cornea. Arch Ophthalmol 57:11, 1957.

2. Morton PL, Ormsby HL, and Basu PK: Healing of endotheliumand Descemet's membrane of rabbit cornea. Am J Ophthalmol46:62, 1959.

3. Chi HH, Teng CC, and Katzin HM: Healing process in themechanical denudation of the corneal endothelium. Am JOphthalmol 49:693, 1960.

4. Capella JA: Regeneration of endothelium in diseased andinjured corneas. Am J Ophthalmol 74:810, 1972.


No. 9 PUMP AND BARRIER FUNCTION FOLLOWING ENDOTHEUAL INJURY / Yee er ol. 1201

5. Khodadoust AA and Green K: Physiological function of regen-erating endothelium. Invest Ophthalmol 15:96, 1976.

6. Van Horn DL, Sendele DD, Seideman S, and Buco PJ:Regenerative capacity of the corneal endothelium in rabbit andcat. Invest Ophthalmol Vis Sci 16:597, 1977.

7. Fujikawa LS, Wickham MG, and Binder PS: Wound healingin cultured corneal endothelial cells. Invest Ophthalmol VisSci 19:793, 1980.

8. Yano M and Tanishima T: Wound healing in rabbit cornealendothelium. Jpn J Ophthalmol 24:297, 1980.

9. Minkowski JS, Bartels SP, Delori FC, Lee SR, Kenyon KR,and Neufeld AH: Corneal endothelial function and structurefollowing cryo-injury in the rabbit. Invest Ophthalmol Vis Sci12:1416, 1984.

10. von Sallmann L, Caravaggio LL, and Grimes P: Studies onthe corneal endothelium of the rabbit. I. Cell division andgrowth. Am J Ophthalmol 51:955, 1961.

11. Chi HH and Kelman CD: Effects of freezing on ocular tissues.I. Clinical and histologic study of corneal endothelium. Am JOphthalmol 61:630, 1966.

12. Faure JP, Kim YZ, and Graf B: Formation of giant cells inthe corneal endothelium during its regeneration after destructionby freezing. Exp Eye Res 12:6, 1971.

13. Kreutziger GO: Lateral membrane morphology and gap junctionstructure in rabbit corneal endothelium. Exp Eye Res 23:285,1976.

14. Renard G, Hirsch M, Faure JP, Usui M, and Pouliquen Y:Ultrastructural study on the regeneration of the corneal endo-thelial cells. Jpn J Ophthalmol 20:243, 1976.

15. Hirsch M, Renard G, Faure JP, and Pouliquen Y: Formationof intercellular spaces and junctions in regenerating rabbitcorneal endothelium. Exp Eye Res 23:385, 1976.

16. Hirsch M, Renard G, Faure JP, and Pouliquen Y: Study ofthe ultrastructure of the rabbit corneal endothelium by thefreeze-fracture technique: apical and lateral junctions. Exp EyeRes 25:277, 1977.

17. Mishima S: Corneal thickness. Surv Ophthalmol 13:57, 1968.18. Mishima S and Trenberth SM: Permeability of the corneal

endothelium to nonelectrolytes. Invest Ophthalmol 7:34, 1968.19. Kim JH, Green K, Martinez M, and Paton D: Solute perme-

ability of the corneal endothelium and Descemet's membrane.Exp Eye Res 12:231, 1971.

20. Hull DS, Farkas S, Green K, Laughter L, Elijah RD, andBowman K: Radial keratotomy effect on cornea and aqueoushumor physiology in the rabbit. Arch Ophthalmol 101:479,1983.

21. Geroski DH and Edelhauser HF: Quantitation of Na/K ATPase

pump sites in the rabbit corneal endothelium. Invest OphthalmolVis Sci 25:1056, 1984.

22. Mishima S and Hedbys BO: Measurement of corneal thicknesswith the Haag-Streit pachymeter. Arch Ophthalmol 80:710,1968.

23. Edelhauser HF, Hoffert JR, and Fromm PO: In vitro ion andwater movement in corneas of rainbow trout. Invest Ophthalmol4:290, 1965.

24. Hull DS, Hine JE, Edelhauser HF, and Hyndiuk RA: Perme-ability of the isolated rabbit cornea to corticosteroids. InvestOphthalmol 13:457, 1974.

25. Dikstein S and Maurice DM: The metabolic basis to the fluidpump in the cornea. J Physiol 221:29, 1972.

26. Maffly RH, Hays RM, Lamdin E, and Leof A: The effect ofneurohypophysical hormones on the permeability of the toadbladder to urea. J Clin Invest 39:630, 1960.

27. Taylor MJ and Hunt CJ: Dual staining of corneal endotheliumwith trypan blue and alizarin red S: importance of pH for thedye-lake reaction. Br J Ophthalmol 65:815, 1981.

28. Cleveland PH and Schneider CW: A simple method of pre-serving tissue for scanning electron microscopy. Vision Res 9:1401, 1969.

29. Tanimura K: A quantitative analysis of corneal endothelialcells. Folia Ophthalmol Jpn 32:1835, 1981.

30. Schultz RO, Matsuda M, Yee RW, Edelhauser HF, and SchultzKJ: Corneal endothelial changes in type I and type II diabetesmellitus. Am J Ophthalmol 98:401, 1984.

31. Waring III G, Krohn M, Ford G, Harris R, and Rosenblatt L:Four methods of measuring human corneal endothelial cellsfrom specular micrographs. Arch Ophthalmol 98:848, 1980.

32. Rao G, Waldron W, and Aquavella J: Morphology of graftendothelium and donor age. Br J Ophthalmol 64:523, 1980.

33. Mills NL and Donn A: Incorporation of tritium-labeled thy-midine by rabbit corneal endothelium. Arch Ophthalmol 64:159, 1960.

34. Anseth A: Slides on corneal polysaccharides. V. Change incorneal glycosaminoglycans in transient stromal edema. ExpEye Res 8:297, 1969.

35. Anseth A and Fransson LA: Studies on corneal polysaccharides.VI. Isolation of dermatan sulfate from corneal scar tissue. ExpEye Res 8:302, 1969.

36. Kaye GI, Edelhauser HF, Stern ME, Cassai ND, Weber P, andDische Z: Further studies of the effect of perfusion with aCa++-free medium on the rabbit cornea: extraction of stromalcomponents. In The Structure of the Eye, Hollyfield JG, editor.New York, Elsevier North Holland, 1982, pp. 271-278.


correlation of corneal endothelial pump site density, barrier function

Documents