Corneal Endothelial Cytoskeletal Changes in F-Actin With Aging, Diabetes, and After Cytochalasin Exposure Eung K. Kim, M.D., Dayle H. Geroski, Ph.D., Glenn P. Holley, B.S., Steven I. Urken, B.S., and Henry F. Edelhauser, Ph.D. We investigated the changes in endothelial cytoskeletal F-actin that occur with aging, diabetes, and exposure to cytochalasin D. Rabbit corneas, human donor corneas (with or without polymegethism), and corneas of diabetic individuals were studied. Endothelial F-actin was stained using nitrobenzoxadiazole-phallacidin. Results of these experiments demonstrated that F-actin of the rabbit and human corneal endothelium was arranged in linear circumferential strands that formed a hexagonal array. After in vitro perfusion of cytochalasin D to the corneal endothelium, the F-actin became randomly distributed throughout the cytoplasm, the hexagonal shape of the endothelial cell was disrupted, and endothelial permeability to carboxyfluorescein increased. Changes in F-actin were also observed in the endothelium of the human corneas with polymegethism, and in donor tissue having had previous posterior chamber intraocular lens implantation. The corneas of diabetic individuals also showed marked irregular F-actin fibers crossing the endothelial cell cytoplasm. These abnormal patterns of F-actin may contribute in part to the polymegethism observed in the corneal endothelial cells and may be the result of constant stress in cell volume regulation, particularly in the corneas of diabetic individuals.
T H E NORMAL corneal endothelium is a monolayer of cells with a regular hexagonal pattern.
Accepted for publication June 2, 1992. From the Department of Ophthalmology, Emory Uni versity School of Medicine, Atlanta, Georgia. This study was supported by E. W. Anderson Fellowship, in part by National Institutes of Health grants EY00933, EY05609, P30 EY06360 (a departmental core grant), and Research to Prevent Blindness, Inc., New York, New York. Reprint requests to Henry F. Edelhauser, Ph.D., De partment of Ophthalmology, Emory University Eye Cen ter, 1327 Clifton Rd. N.E., Atlanta, GA 30322.
This cell arrangement is the most stable thermodynamically and geometrically for two-di mensional arrays. Since the human corneal en dothelium does not proliferate, 1 its ability to compensate for cell loss is limited. If human endothelial cells are damaged, healing occurs by enlargement and migration of the remaining cells to cover the defect. This results in an increase in the cell area. 2 The functional capacity of the endothelium can be correlated with its structure. Rao and associates 3 indicated that polymegethism is closely correlated with the susceptibility to trauma, and that more postoperative edema developed after cataract extraction in those patients with preoperative polymegethism. F-actin, a major component of the cellular cytoskeleton, has a key role in maintaining cellular structure. F-actin, assembled through the polymerization of monomeric G-actin/is not a permanent, static structure. G-actin mon omers are continually added and lost; thus the cytoskeletal shape of the cell is in dynamic equilibrium and can adapt to environmental stresses. Previous studies by Gordon, Essner, and Rothstein 5 and Gordon 6 demonstrated that normal polygonal endothelial cells are charac terized by circumferential bands of F-actin lo cated at the apical portion of cell-cell borders. Fujino and Tanishima 7 studied F-actin changes in rabbit corneal endothelium during wound healing in vivo with the nitrobenzoxadiazolephallacidin method and showed many fibers throughout the cytoplasm of cells migrating to the wound area. Jumblatt, Matkin, and Neu feld8 demonstrated that epidermal growth fac tor or indomethacin, or both, can cause elonga tion and redistribution of F-actin into a diffuse cytoplasmic pattern in cell-cultured endothelial cells. All of these studies illustrate the impor tance of F-actin in the cytoskeletal structure of endothelial cells. We evaluated F-actin changes in normal rab bit and human corneal endothelial cells after cytochalasin treatment (a known inhibitor of
©AMERICAN JOURNAL OF OPHTHALMOLOGY 114:329-335, SEPTEMBER, 1992
329
330
AMERICAN JOURNAL OF OPHTHALMOLOGY
F-actin polymerization) and compared these cytoskeletal changes to those that might devel op in the corneal endothelium of diabetic indi viduals, donor corneas with extreme polymegethism, and human corneas after intraocular surgical procedures.
Material and Methods Perfusion of the corneal endothelium—Twelve male and female 3-month-old (2.0- to 2.5-kg) and six 26-month-old (5-kg) New Zealand White rabbits were anesthetized with xylazine and ketamine HC1 and killed with pentobarbital. All rabbits were used in accordance with the Association for Research in Vision and Ophthalmology Resolution on the Use of Ani mals in Research. The eyes were enucleated and the corneas were excised and mounted in the in vitro specular microscope for endothelial per fusion as previously described. 9 Eight human donor corneas (Georgia Eye Bank, Atlanta, Georgia) from nondiabetic donors older than 40 years of age and five corneas from diabetic donors were similarly mounted for in vitro perfusion. Rabbit corneas were perfused with glutathione bicarbonate Ringer's solution 10 and human corneas were perfused (37 C) with bal anced saline solution (BSS Plus, Alcon Labora tories, Fort Worth, Texas) for a 60-minute equil ibration period. The experimental rabbit cor neal endothelium of the matched pair was then perfused with cytochalasin D (2 μg/ml) in glutathione bicarbonate Ringer's solution for 40 to 60 minutes while the control corneal endothelium was perfused with glutathione bi carbonate Ringer's solution. The experimental human corneal endothelium was perfused with cytochalasin D (5 μg/ml) in balanced saline solution for 60 minutes. The anterior corneal surface was covered with silicone oil during the perfusion, to minimize evaporation at the ante rior corneal surface and to optically couple the microscope objective to the tissue. Nitrobenzoxadiazole-phallacidin staining of Factin filament—The cornea with a 2- to 3-mm scierai rim was fixed in 10% neutral formalin for 90 minutes. A corneal button 8 mm in diameter was trephined from the central cornea and the endothelium with Descemet's mem brane (including a small portion of posterior stroma) was removed from the overlying stroma with forceps under the dissecting microscope. Eight radial incisions were made in the periph
September, 1992
ery of the sheet of endothelium and Descemet's membrane to permit a flat preparation of the cornea. 11 The flat preparations of rabbit and human corneal endothelium were stained with 1.64 x 10" 4 mol/1 nitrobenzoxadiazole-phallacidin (Molecular Probes, Inc., Eugene, Oregon) for 30 minutes at 37 C for visualization of F-actin filaments. After a buffer (phosphate-buffered saline solution; 0.01 mol/1, pH 7.2) wash, tis sues were mounted on a nonfluorescent glass slide in a buffer-glycerol (1:1) mixture and were photographed using a Nikon fluorescence mi croscope with filter DM 510 (excitation, 450 to 490 nm; barrier, 520 nm).7·12 Cytochalasin D permeability studies—The method of Watsky, McDermott, and Edelhauser13 was used to measure endothelial permea bility to carboxyfluorescein. After epithelial debridement, isolated corneas were mounted in the in vitro specular microscope and the endo thelium was perfused with glutathione bicar bonate Ringer's solution for a 60-minute sta bilization period. The perfusate was then changed to cytochalasin D (2 μg/ml) in gluta thione bicarbonate Ringer's solution and the endothelium was perfused for an additional 40 minutes. The anterior surface of the cornea was covered with silicone oil. The oil was removed after ten minutes of cytochalasin D perfusion and 0.3 ml of glutathione bicarbonate Ringer's solution containing 0.26 mmol/1 carboxyfluo rescein was applied to the corneal surface. The carboxyfluorescein was removed from the cor neal surface after 30 seconds. The anterior corneal surface was swabbed and silicone oil was reapplied. The glutathione bicarbonate Ringer's solution perfusate was then collected in a preweighed test tube for 30 minutes. The mass of carboxyfluorescein in the perfusate (Mp) was determined spectrofluorometrically. At the termination of the perfusion, corneas were blotted dry and placed in balanced saline solution for 48 hours for carboxyfluorescein elution. The carboxyfluorescein concentration in the eluate was measured by spectrofluorometry and stromal dye mass (M8) was calculated. The cornea-aqueous transfer coefficient (kcca)13 and endothelial permeability (Pac)14 were calcu lated as: k cca = [ln(Mp + M.) - In M,]/t where t is the time after dye application (30 minutes). The Rca value of 1.07 is the steady-
Vol. 114, No. 3
Corneal Endothelial F-Actin
331
state distribution ratio used in this study, 13 and q is the average stromal thickness obtained from the three final readings of the perfusion. Statistical significance was determined with a paired f-test.
Results F-actin fibers of the normal endothelium from a 3-month-old rabbit were mainly located adjacent to lateral cell membranes in circumfer ential bands that formed hexagons. No dense F-actin fibers were seen in the cytoplasm (Fig. 1). The F-actin fibers at the borders between cells were composed of two major bands run ning parallel along the periphery of the cell. Numerous interconnecting strands joined these two bands. In the normal human endothelium, the F-actin fibers were similarly arranged adja cent to cell membranes, forming a hexagonal pattern (Fig. 2). When the corneal endothelium of the 3month-old rabbit was perfused with cytochalasin D (2 μg/ml) for one hour, the cells lost their hexagonal array and separated into irregularly shaped clumps or islands of cells (Fig. 3, left) with breaks in the junctions between these islands. The actin fibers located at the cell border lost their shape and condensed F-actin fibers were found around the nucleus (Fig. 3, left). Scanning electron microscopy (Fig. 3, right) showed cells that were pulled apart in
phallacidin-stained F-actin fibers of the corneal en dothelium of a 3-month-old rabbit. F-actin fibers are mainly located adjacent to cell membranes, forming hexagons (x 250; inset, x 630).
Fig. 2 (Kim and associates). Nitrobenzoxadiazolephallacidin-stained F-actin fibers of the normal cor neal endothelium from a 78-year-old human. F-actin fibers are mainly located adjacent to cell membranes, forming hexagons, although some cells are pentago nal in F-actin shape (x 250; inset, x 630). clumps, with areas of thin cytoplasmic mem brane covering other areas. The endothelium of 26-month-old rabbits showed changes with cytochalasin D (2 μ g / m l / one hour) perfusion that were comparable to those observed in the 3month-old rabbits. Carboxyfluorescein perme ability (mean ± SEM) of the endothelium of four rabbit corneas perfused with cytochalasin D increased significantly to 7.97 ± 0.69 x 10" 4 cm/min compared to 3.96 ± 0.65 x 10" 4 c m / min for the paired control corneas (significantly different at P < .01 with a paired f-test). Human corneal endothelium, perfused with a 2.5-fold greater concentration of cytochalasin D (5 μg/ml) for one hour, showed slight chang es in F-actin with some cytoplasmic staining, These changes, however, were not as marked as those seen in the cytochalasin D-perfused rab bit corneas and the cells retained their hexagonality (Fig. 4, top left). Scanning electron microscopy (Fig. 4, top right) confirmed that the endothelial monolayer remained intact and cell junctions were preserved. However, cytochala sin D caused some irregularity of the anterior endothelial cell surface (Fig. 4, bottom left, transmission electron microscopy). The endothelium of human corneas with se vere polymegethism was characterized by an increased complexity of F-actin along the cyto plasmic membrane (Fig. 5). All of the F-actin fibers, however, were associated with the cir cumferential cytoskeleton and were not dis persed throughout the cytoplasm.
332
September, 1992
AMERICAN JOURNAL OF OPHTHALMOLOGY
F-actin was located in dense focal aggregates while most cells were in their normal hexagonal array with the F-actin located circumferentially (Fig. 6). The corneal endothelium of diabetic individ uals showed severe polymegethism and, al though cellular F-actin showed a prominent circumferential pattern, many F-actin fibers randomly crossed the cytoplasm. These chang es were noted throughout the endothelium (Fig. 7). Scattered F-actin fibers were not observed in the cytoplasm of normal human corneal endothelial cells.
Discussion
Fig. 3 (Kim and associates). Rabbit corneal endo thelium after perfusion with cytochalasin D (2 μg/ ml) for one hour. Top, Nitrobenzoxadiazolephallacidin staining shows that F-actin fibers nor mally located at the cell border are absent and con densed F-actin fibers around the nucleus are seen. Areas of cells separate and form islands of cells without hexagonal pattern (x 250; inset, x 630). Bot tom, Scanning electron microscopy shows areas of cells pulled apart into islands with thin strands of cytoplasmic membrane (arrowhead) remaining at tached to Descemet's membrane at the cell's original position (asterisk) (bar indicates 10 μηι). The corneas from donor eyes on which pos terior chamber intraocular lens implantation had been performed also showed an irregular F-actin pattern. Tightly packed aggregates of F-actin fibers were observed to accumulate along the cellular borders. In some cells, the
F-actin, which is assembled through the po lymerization of G-actin, is a major component of the cellular cytoskeleton. The polymeriza tion of G-actin is inhibited by the cytochalasin, 15 although the extent of inhibition might be dif ferent between rabbit and human tissue. The results of the cytochalasin perfusion studies showed that the F-actin is important in main taining cell shape and barrier function of the corneal endothelium. Disruption of F-actin with cytochalasin D perfusion resulted in elon gated and disarranged cell shape and increased permeability of the rabbit corneal endothelium, indicating endothelial cell dysfunction. The disruption of the F-actin that occurs with cytochalasin exposure was greater and more severe in the rabbit than in the human endothe lium. This suggests that the F-actin component of the human endothelial cytoskeleton has a greater resistance to cytochalasin D than that of the rabbit. This difference may be related to the difference of F-actin turnover rate. For example, the rabbit F-actin may have a much faster turn over rate than that of human F-actin; thus, the rabbit F-actin would be more sensitive than the human F-actin. However, other differences in cytoskeletal composition may exist between rabbit and human corneal endothelium. F-actin has an important role in the endothe lial wound healing process. Fujino and Tanishima 7 studied F-actin changes in the rabbit corne al endothelium after transcorneal freezing. They showed the development of fiber-like Factin structures three days after injury. F-actin structures having a punctate pattern scattered throughout the cytoplasm 14 days after injury were described. As the endothelium healed completely, the F-actin pattern returned to nor-
Vol. 114, No. 3
Corneal Endothelial F-Actin
333
Fig. 4 (Kim and associates). Human corneal endo thelium after perfusion with cytochalasin D (5 μg/ ml) for one hour. Top left, Nitrobenzoxadiazolephallacidin staining shows that F-actin has a pre dominantly hexagonal pattern (x480). Top right, Scanning electron microscopy shows an intact endo thelial monolayer and cell junctions were preserved (bar indicates 10 μιη). Bottom left, Transmission electron microscopy shows a relatively normal mon olayer except for some irregularity of the anterior endothelial cell surface (bar indicates 2 μπ\). mal. In our study, the endothelium of a donor cornea after cataract extraction with posteri or chamber intraocular lenses implantation showed regions of a punctate pattern of F-actin similar to that described by Fujino and Tanishima. 7 Other areas of the endothelium remained normal. Because of the limited proliferative capacity of human tissue, this punctate pattern might persist and such changes might indicate a corneal endothelium in the process of focal wound healing. Watsky, McDermott, and Edelhauser 13 found that the corneal endothelial per meability to carboxyfluorescein was increased in all donor eyes on which previous intraocular surgical procedures had been performed. The observed increase in permeability may reflect the focal wound healing and cytoskeletal changes observed in these eyes. Our studies showed many F-actin fibers crossing the cytoplasm of the corneal endothe lium of diabetic individuals, compared to small changes of endothelial cellular F-actin at the
Fig. 5 (Kim and associates). Nitrobenzoxadiazolephallacidin-stained corneal endothelium with severe polymegethism from a 77-year-old woman. Though all F-actin fibers remain associated with the circum ferential cytoskeleton, increased fiber complexity along the cell membrane can be noted (x 250; inset, X630).
334
AMERICAN JOURNAL OF OPHTHALMOLOGY
Fig. 6 (Kim and associates). Corneal endothelium of an 81-year-old woman after posterior chamber intraocular lens implantation. Nitrobenzoxadiazolephallacidin staining shows focal areas of irregular F-actin and the formation of dense aggregates. Most other cells, however, are in normal hexagonal array, with the F-actin located in its normal circumferential pattern (x 250; inset, x 630).
borders between adjacent cells in corneas with endothelial polymegethism in nondiabetic pa tients. Kleinzeller and Ziyadeh 16 described marked swelling of shark rectal gland cells treated with the organic mercurials, which are known to induce a dissociation of F-actin. They also showed that hypotonicity induced by re duction of the sodium concentration in the media resulted in a loss of cellular F-actin, and massive cellular swelling. They hypothesized
September, 1992
that the resistance of the cell membrane to stretch, imparted by the intact cytoskeleton, has a major role in restricting cellular swelling caused by osmotic forces. In the corneas of diabetic individuals, in which there were changes in the F-actin, the changes may have been related to the sorbitol accumulation in the endothelial cell and osmotic gradient between the endothelial cell cytoplasm and aqueous humor. If the corneal endothelium is under continual osmotic stress, an F-actin change to the cell cytoskeleton is most likely to occur to prevent rupture of the cell membrane. 1617 Since the osmotic stress induced in the corneal endo thelium of diabetic individuals occurs over a long time period, leading to polymegethism, the endothelial cell changes appear to be a permanent alteration to the cytoskeleton. Poly megethism develops in corneas with aging, diabetes, after a surgical procedure or infec tion, and with contact-lens wear. In corneal endothelia with polymegethism in nondiabetic individuals, there were small, but definite changes in F-actin in the borders between adja cent cells. These changes may suggest that the endothelial cells are in stress, but can maintain their structure and can resist cellular swelling with minimal changes in F-actin. Rao and asso ciates, 3 however, demonstrated that after cata ract extraction, patients with an endothelium with polymegethism had significantly higher corneal thickness than patients who had a regu lar endothelial cell pattern. Schultz and associ ates 18 studied polymegethism in corneas of diabetic individuals with quantitative
Fig. 7 (Kim and associates). In donor corneas of diabetic individuals, nitrobenzoxadiazole-phallacidin staining shows many F-actin fibers randomly crossing the cytoplasm. Left, Endothelium with severe polymegethism from a 77-year-old diabetic woman (X 250; inset, x 630). Right, Endothelium with moderate polymegethism from a 65-year-old diabetic man (x 250; inset, x 630).
Vol. 114, No. 3
Corneal Endothelial F-Actin
m o r p h o m e t r y a n d s u g g e s t e d t h a t t h e c o r n e a s of diabetic i n d i v i d u a l s m a y b e at risk in any i n t r a ocular p r o c e d u r e . O u r d a t a a l s o s h o w e d t h a t a n e n d o t h e l i u m w i t h p o l y m e g e t h i s m h a s a n al tered cytoskeleton and, thus, might be more v u l n e r a b l e to t r a u m a from i n t r a o c u l a r surgical procedures.
References 1. Doughman, D. J., Van Horn, D., Rodman, W. P., Byrnes, P., and Lindstrom, R. L.: Human corneal endothelial layer repair during organ culture. Arch. Ophthalmol. 94:1791, 1976. 2. Laing, R. A., Sandstrom, M. M., Berrospi, A. R., and Leibowitz, H. M.: Changes in the corneal endo thelium as a function of age. Exp. Eye Res. 22:587, 1976. 3. Rao, G. N., Shaw, E. L., Arthur, E. J., and Aquavella, J. V.: Endothelial cell morphology and corneal deturgescence. Ann. Ophthalmol. 11:885, 1979. 4. Goddette, D. W., and Frieden, C : Actin polym erization. The mechanism of action of cytochalasin D. J. Biol. Chem. 261:15974, 1986. 5. Gordon, S. R., Essner, E., and Rothstein, H.: In situ demonstration of actin in normal and injured ocular tissues using 7-nitrobenz-2-oxa-l,3-diazole phallacidin. Cell Motil. 4:343, 1982. 6. Gordon, S. R.: The localization of actin in divid ing corneal endothelial cells demonstrated with nitrobenzoxadiazole phallacidin. Cell Tissue Res. 229:533, 1983. 7. Fujino, Y., and Tanishima, T.: Actin in woundhealing of rabbit corneal endothelium. II. Study by nitrobenzoxadiazole-phallacidin method. Jpn. J. Ophthalmol. 31:393, 1987. 8. Jumblatt, M. N., Matkin, E. D., and Neufeld, A. H.: Pharmacological regulation of morphology and mitosis in cultured rabbit corneal endothelium. Invest. Ophthalmol. Vis. Sci. 29:586, 1988.
335
9. McCarey, B. E., Edelhauser, H. F., and Van Horn, D. L.: Functional and structural changes in the corneal endothelium during in vitro perfusion. In vest. Ophthalmol. Vis. Sci. 12:410, 1973. 10. Edelhauser, H. F„ Van Horn, D. L„ Schultz, R. O., and Hyndiuk, R. A.: Comparative toxicity of intraocular irrigating solutions on the corneal endo thelium. Am. J. Ophthalmol. 81:473, 1976. 11. Fujino, Y., and Tanishima, T.: Actin in woundhealing of rabbit corneal endothelium. I. Study by immunoperoxidase method. Jpn. J. Ophthalmol. 31:384, 1987. 12. Joyce, N. C , Matkin, E. D., and Neufeld, A. H.: Corneal endothelial wound closure in vitro. Effects of EGF a n d / o r indomethacin. Invest. Oph thalmol. Vis. Sci. 30:1548, 1989. 13. Watsky, M. A., McDermott, M. L., and Edel hauser, H. F.: In vitro corneal endothelial permeabil ity in rabbit and human. The effect of age, cataract surgery and diabetes. Exp. Eye Res. 49:751, 1989. 14. Araie, M.: Carboxyfluorescein. A dye for eval uating the corneal endothelial barrier function in vivo. Exp. Eye Res. 42:141, 1986. 15. Wessells, N. K„ Spooner, B. S., Ash, J. F., Bradley, M. O., Luduena, M. A., Taylor, E. L., Wrenn, J. T., and Yamada, K. M.: Microfilaments in cellular and developmental processes. Contractile microfilament machinery of many cell types is reversibly inhibited by cytochalasin-B. Science 171:135, 1971. 16. Kleinzeller, A., and Ziyadeh, F. N.: Cell vol ume regulation in epithelia with emphasis on the role of osmolytes and the cytoskeleton. In Beyenbach, K. W. (ed.): Cell Volume Regulation, vol. 4. Basel, Kar ger, 1990, pp. 59-86. 17. Gruen, D. W. R., and Wolfe, J.: Lateral ten sions and pressures in membranes and lipid monolayers. Biochim. Biophys. Acta 688:572, 1982. 18. Schultz, R. O., Matsuda, M., Yee, R. W., Edel hauser, H. F., and Schultz, K. J.: Corneal endothelial changes in type I and type II diabetes mellitus. Am. J. Ophthalmol. 98:401, 1984.
T H E NORMAL corneal endothelium is a monolayer of cells with a regular hexagonal pattern.
Accepted for publication June 2, 1992. From the Department of Ophthalmology, Emory Uni versity School of Medicine, Atlanta, Georgia. This study was supported by E. W. Anderson Fellowship, in part by National Institutes of Health grants EY00933, EY05609, P30 EY06360 (a departmental core grant), and Research to Prevent Blindness, Inc., New York, New York. Reprint requests to Henry F. Edelhauser, Ph.D., De partment of Ophthalmology, Emory University Eye Cen ter, 1327 Clifton Rd. N.E., Atlanta, GA 30322.
This cell arrangement is the most stable thermodynamically and geometrically for two-di mensional arrays. Since the human corneal en dothelium does not proliferate, 1 its ability to compensate for cell loss is limited. If human endothelial cells are damaged, healing occurs by enlargement and migration of the remaining cells to cover the defect. This results in an increase in the cell area. 2 The functional capacity of the endothelium can be correlated with its structure. Rao and associates 3 indicated that polymegethism is closely correlated with the susceptibility to trauma, and that more postoperative edema developed after cataract extraction in those patients with preoperative polymegethism. F-actin, a major component of the cellular cytoskeleton, has a key role in maintaining cellular structure. F-actin, assembled through the polymerization of monomeric G-actin/is not a permanent, static structure. G-actin mon omers are continually added and lost; thus the cytoskeletal shape of the cell is in dynamic equilibrium and can adapt to environmental stresses. Previous studies by Gordon, Essner, and Rothstein 5 and Gordon 6 demonstrated that normal polygonal endothelial cells are charac terized by circumferential bands of F-actin lo cated at the apical portion of cell-cell borders. Fujino and Tanishima 7 studied F-actin changes in rabbit corneal endothelium during wound healing in vivo with the nitrobenzoxadiazolephallacidin method and showed many fibers throughout the cytoplasm of cells migrating to the wound area. Jumblatt, Matkin, and Neu feld8 demonstrated that epidermal growth fac tor or indomethacin, or both, can cause elonga tion and redistribution of F-actin into a diffuse cytoplasmic pattern in cell-cultured endothelial cells. All of these studies illustrate the impor tance of F-actin in the cytoskeletal structure of endothelial cells. We evaluated F-actin changes in normal rab bit and human corneal endothelial cells after cytochalasin treatment (a known inhibitor of
©AMERICAN JOURNAL OF OPHTHALMOLOGY 114:329-335, SEPTEMBER, 1992
329
330
AMERICAN JOURNAL OF OPHTHALMOLOGY
F-actin polymerization) and compared these cytoskeletal changes to those that might devel op in the corneal endothelium of diabetic indi viduals, donor corneas with extreme polymegethism, and human corneas after intraocular surgical procedures.
Material and Methods Perfusion of the corneal endothelium—Twelve male and female 3-month-old (2.0- to 2.5-kg) and six 26-month-old (5-kg) New Zealand White rabbits were anesthetized with xylazine and ketamine HC1 and killed with pentobarbital. All rabbits were used in accordance with the Association for Research in Vision and Ophthalmology Resolution on the Use of Ani mals in Research. The eyes were enucleated and the corneas were excised and mounted in the in vitro specular microscope for endothelial per fusion as previously described. 9 Eight human donor corneas (Georgia Eye Bank, Atlanta, Georgia) from nondiabetic donors older than 40 years of age and five corneas from diabetic donors were similarly mounted for in vitro perfusion. Rabbit corneas were perfused with glutathione bicarbonate Ringer's solution 10 and human corneas were perfused (37 C) with bal anced saline solution (BSS Plus, Alcon Labora tories, Fort Worth, Texas) for a 60-minute equil ibration period. The experimental rabbit cor neal endothelium of the matched pair was then perfused with cytochalasin D (2 μg/ml) in glutathione bicarbonate Ringer's solution for 40 to 60 minutes while the control corneal endothelium was perfused with glutathione bi carbonate Ringer's solution. The experimental human corneal endothelium was perfused with cytochalasin D (5 μg/ml) in balanced saline solution for 60 minutes. The anterior corneal surface was covered with silicone oil during the perfusion, to minimize evaporation at the ante rior corneal surface and to optically couple the microscope objective to the tissue. Nitrobenzoxadiazole-phallacidin staining of Factin filament—The cornea with a 2- to 3-mm scierai rim was fixed in 10% neutral formalin for 90 minutes. A corneal button 8 mm in diameter was trephined from the central cornea and the endothelium with Descemet's mem brane (including a small portion of posterior stroma) was removed from the overlying stroma with forceps under the dissecting microscope. Eight radial incisions were made in the periph
September, 1992
ery of the sheet of endothelium and Descemet's membrane to permit a flat preparation of the cornea. 11 The flat preparations of rabbit and human corneal endothelium were stained with 1.64 x 10" 4 mol/1 nitrobenzoxadiazole-phallacidin (Molecular Probes, Inc., Eugene, Oregon) for 30 minutes at 37 C for visualization of F-actin filaments. After a buffer (phosphate-buffered saline solution; 0.01 mol/1, pH 7.2) wash, tis sues were mounted on a nonfluorescent glass slide in a buffer-glycerol (1:1) mixture and were photographed using a Nikon fluorescence mi croscope with filter DM 510 (excitation, 450 to 490 nm; barrier, 520 nm).7·12 Cytochalasin D permeability studies—The method of Watsky, McDermott, and Edelhauser13 was used to measure endothelial permea bility to carboxyfluorescein. After epithelial debridement, isolated corneas were mounted in the in vitro specular microscope and the endo thelium was perfused with glutathione bicar bonate Ringer's solution for a 60-minute sta bilization period. The perfusate was then changed to cytochalasin D (2 μg/ml) in gluta thione bicarbonate Ringer's solution and the endothelium was perfused for an additional 40 minutes. The anterior surface of the cornea was covered with silicone oil. The oil was removed after ten minutes of cytochalasin D perfusion and 0.3 ml of glutathione bicarbonate Ringer's solution containing 0.26 mmol/1 carboxyfluo rescein was applied to the corneal surface. The carboxyfluorescein was removed from the cor neal surface after 30 seconds. The anterior corneal surface was swabbed and silicone oil was reapplied. The glutathione bicarbonate Ringer's solution perfusate was then collected in a preweighed test tube for 30 minutes. The mass of carboxyfluorescein in the perfusate (Mp) was determined spectrofluorometrically. At the termination of the perfusion, corneas were blotted dry and placed in balanced saline solution for 48 hours for carboxyfluorescein elution. The carboxyfluorescein concentration in the eluate was measured by spectrofluorometry and stromal dye mass (M8) was calculated. The cornea-aqueous transfer coefficient (kcca)13 and endothelial permeability (Pac)14 were calcu lated as: k cca = [ln(Mp + M.) - In M,]/t where t is the time after dye application (30 minutes). The Rca value of 1.07 is the steady-
Vol. 114, No. 3
Corneal Endothelial F-Actin
331
state distribution ratio used in this study, 13 and q is the average stromal thickness obtained from the three final readings of the perfusion. Statistical significance was determined with a paired f-test.
Results F-actin fibers of the normal endothelium from a 3-month-old rabbit were mainly located adjacent to lateral cell membranes in circumfer ential bands that formed hexagons. No dense F-actin fibers were seen in the cytoplasm (Fig. 1). The F-actin fibers at the borders between cells were composed of two major bands run ning parallel along the periphery of the cell. Numerous interconnecting strands joined these two bands. In the normal human endothelium, the F-actin fibers were similarly arranged adja cent to cell membranes, forming a hexagonal pattern (Fig. 2). When the corneal endothelium of the 3month-old rabbit was perfused with cytochalasin D (2 μg/ml) for one hour, the cells lost their hexagonal array and separated into irregularly shaped clumps or islands of cells (Fig. 3, left) with breaks in the junctions between these islands. The actin fibers located at the cell border lost their shape and condensed F-actin fibers were found around the nucleus (Fig. 3, left). Scanning electron microscopy (Fig. 3, right) showed cells that were pulled apart in
phallacidin-stained F-actin fibers of the corneal en dothelium of a 3-month-old rabbit. F-actin fibers are mainly located adjacent to cell membranes, forming hexagons (x 250; inset, x 630).
Fig. 2 (Kim and associates). Nitrobenzoxadiazolephallacidin-stained F-actin fibers of the normal cor neal endothelium from a 78-year-old human. F-actin fibers are mainly located adjacent to cell membranes, forming hexagons, although some cells are pentago nal in F-actin shape (x 250; inset, x 630). clumps, with areas of thin cytoplasmic mem brane covering other areas. The endothelium of 26-month-old rabbits showed changes with cytochalasin D (2 μ g / m l / one hour) perfusion that were comparable to those observed in the 3month-old rabbits. Carboxyfluorescein perme ability (mean ± SEM) of the endothelium of four rabbit corneas perfused with cytochalasin D increased significantly to 7.97 ± 0.69 x 10" 4 cm/min compared to 3.96 ± 0.65 x 10" 4 c m / min for the paired control corneas (significantly different at P < .01 with a paired f-test). Human corneal endothelium, perfused with a 2.5-fold greater concentration of cytochalasin D (5 μg/ml) for one hour, showed slight chang es in F-actin with some cytoplasmic staining, These changes, however, were not as marked as those seen in the cytochalasin D-perfused rab bit corneas and the cells retained their hexagonality (Fig. 4, top left). Scanning electron microscopy (Fig. 4, top right) confirmed that the endothelial monolayer remained intact and cell junctions were preserved. However, cytochala sin D caused some irregularity of the anterior endothelial cell surface (Fig. 4, bottom left, transmission electron microscopy). The endothelium of human corneas with se vere polymegethism was characterized by an increased complexity of F-actin along the cyto plasmic membrane (Fig. 5). All of the F-actin fibers, however, were associated with the cir cumferential cytoskeleton and were not dis persed throughout the cytoplasm.
332
September, 1992
AMERICAN JOURNAL OF OPHTHALMOLOGY
F-actin was located in dense focal aggregates while most cells were in their normal hexagonal array with the F-actin located circumferentially (Fig. 6). The corneal endothelium of diabetic individ uals showed severe polymegethism and, al though cellular F-actin showed a prominent circumferential pattern, many F-actin fibers randomly crossed the cytoplasm. These chang es were noted throughout the endothelium (Fig. 7). Scattered F-actin fibers were not observed in the cytoplasm of normal human corneal endothelial cells.
Discussion
Fig. 3 (Kim and associates). Rabbit corneal endo thelium after perfusion with cytochalasin D (2 μg/ ml) for one hour. Top, Nitrobenzoxadiazolephallacidin staining shows that F-actin fibers nor mally located at the cell border are absent and con densed F-actin fibers around the nucleus are seen. Areas of cells separate and form islands of cells without hexagonal pattern (x 250; inset, x 630). Bot tom, Scanning electron microscopy shows areas of cells pulled apart into islands with thin strands of cytoplasmic membrane (arrowhead) remaining at tached to Descemet's membrane at the cell's original position (asterisk) (bar indicates 10 μηι). The corneas from donor eyes on which pos terior chamber intraocular lens implantation had been performed also showed an irregular F-actin pattern. Tightly packed aggregates of F-actin fibers were observed to accumulate along the cellular borders. In some cells, the
F-actin, which is assembled through the po lymerization of G-actin, is a major component of the cellular cytoskeleton. The polymeriza tion of G-actin is inhibited by the cytochalasin, 15 although the extent of inhibition might be dif ferent between rabbit and human tissue. The results of the cytochalasin perfusion studies showed that the F-actin is important in main taining cell shape and barrier function of the corneal endothelium. Disruption of F-actin with cytochalasin D perfusion resulted in elon gated and disarranged cell shape and increased permeability of the rabbit corneal endothelium, indicating endothelial cell dysfunction. The disruption of the F-actin that occurs with cytochalasin exposure was greater and more severe in the rabbit than in the human endothe lium. This suggests that the F-actin component of the human endothelial cytoskeleton has a greater resistance to cytochalasin D than that of the rabbit. This difference may be related to the difference of F-actin turnover rate. For example, the rabbit F-actin may have a much faster turn over rate than that of human F-actin; thus, the rabbit F-actin would be more sensitive than the human F-actin. However, other differences in cytoskeletal composition may exist between rabbit and human corneal endothelium. F-actin has an important role in the endothe lial wound healing process. Fujino and Tanishima 7 studied F-actin changes in the rabbit corne al endothelium after transcorneal freezing. They showed the development of fiber-like Factin structures three days after injury. F-actin structures having a punctate pattern scattered throughout the cytoplasm 14 days after injury were described. As the endothelium healed completely, the F-actin pattern returned to nor-
Vol. 114, No. 3
Corneal Endothelial F-Actin
333
Fig. 4 (Kim and associates). Human corneal endo thelium after perfusion with cytochalasin D (5 μg/ ml) for one hour. Top left, Nitrobenzoxadiazolephallacidin staining shows that F-actin has a pre dominantly hexagonal pattern (x480). Top right, Scanning electron microscopy shows an intact endo thelial monolayer and cell junctions were preserved (bar indicates 10 μιη). Bottom left, Transmission electron microscopy shows a relatively normal mon olayer except for some irregularity of the anterior endothelial cell surface (bar indicates 2 μπ\). mal. In our study, the endothelium of a donor cornea after cataract extraction with posteri or chamber intraocular lenses implantation showed regions of a punctate pattern of F-actin similar to that described by Fujino and Tanishima. 7 Other areas of the endothelium remained normal. Because of the limited proliferative capacity of human tissue, this punctate pattern might persist and such changes might indicate a corneal endothelium in the process of focal wound healing. Watsky, McDermott, and Edelhauser 13 found that the corneal endothelial per meability to carboxyfluorescein was increased in all donor eyes on which previous intraocular surgical procedures had been performed. The observed increase in permeability may reflect the focal wound healing and cytoskeletal changes observed in these eyes. Our studies showed many F-actin fibers crossing the cytoplasm of the corneal endothe lium of diabetic individuals, compared to small changes of endothelial cellular F-actin at the
Fig. 5 (Kim and associates). Nitrobenzoxadiazolephallacidin-stained corneal endothelium with severe polymegethism from a 77-year-old woman. Though all F-actin fibers remain associated with the circum ferential cytoskeleton, increased fiber complexity along the cell membrane can be noted (x 250; inset, X630).
334
AMERICAN JOURNAL OF OPHTHALMOLOGY
Fig. 6 (Kim and associates). Corneal endothelium of an 81-year-old woman after posterior chamber intraocular lens implantation. Nitrobenzoxadiazolephallacidin staining shows focal areas of irregular F-actin and the formation of dense aggregates. Most other cells, however, are in normal hexagonal array, with the F-actin located in its normal circumferential pattern (x 250; inset, x 630).
borders between adjacent cells in corneas with endothelial polymegethism in nondiabetic pa tients. Kleinzeller and Ziyadeh 16 described marked swelling of shark rectal gland cells treated with the organic mercurials, which are known to induce a dissociation of F-actin. They also showed that hypotonicity induced by re duction of the sodium concentration in the media resulted in a loss of cellular F-actin, and massive cellular swelling. They hypothesized
September, 1992
that the resistance of the cell membrane to stretch, imparted by the intact cytoskeleton, has a major role in restricting cellular swelling caused by osmotic forces. In the corneas of diabetic individuals, in which there were changes in the F-actin, the changes may have been related to the sorbitol accumulation in the endothelial cell and osmotic gradient between the endothelial cell cytoplasm and aqueous humor. If the corneal endothelium is under continual osmotic stress, an F-actin change to the cell cytoskeleton is most likely to occur to prevent rupture of the cell membrane. 1617 Since the osmotic stress induced in the corneal endo thelium of diabetic individuals occurs over a long time period, leading to polymegethism, the endothelial cell changes appear to be a permanent alteration to the cytoskeleton. Poly megethism develops in corneas with aging, diabetes, after a surgical procedure or infec tion, and with contact-lens wear. In corneal endothelia with polymegethism in nondiabetic individuals, there were small, but definite changes in F-actin in the borders between adja cent cells. These changes may suggest that the endothelial cells are in stress, but can maintain their structure and can resist cellular swelling with minimal changes in F-actin. Rao and asso ciates, 3 however, demonstrated that after cata ract extraction, patients with an endothelium with polymegethism had significantly higher corneal thickness than patients who had a regu lar endothelial cell pattern. Schultz and associ ates 18 studied polymegethism in corneas of diabetic individuals with quantitative
Fig. 7 (Kim and associates). In donor corneas of diabetic individuals, nitrobenzoxadiazole-phallacidin staining shows many F-actin fibers randomly crossing the cytoplasm. Left, Endothelium with severe polymegethism from a 77-year-old diabetic woman (X 250; inset, x 630). Right, Endothelium with moderate polymegethism from a 65-year-old diabetic man (x 250; inset, x 630).
Vol. 114, No. 3
Corneal Endothelial F-Actin
m o r p h o m e t r y a n d s u g g e s t e d t h a t t h e c o r n e a s of diabetic i n d i v i d u a l s m a y b e at risk in any i n t r a ocular p r o c e d u r e . O u r d a t a a l s o s h o w e d t h a t a n e n d o t h e l i u m w i t h p o l y m e g e t h i s m h a s a n al tered cytoskeleton and, thus, might be more v u l n e r a b l e to t r a u m a from i n t r a o c u l a r surgical procedures.
References 1. Doughman, D. J., Van Horn, D., Rodman, W. P., Byrnes, P., and Lindstrom, R. L.: Human corneal endothelial layer repair during organ culture. Arch. Ophthalmol. 94:1791, 1976. 2. Laing, R. A., Sandstrom, M. M., Berrospi, A. R., and Leibowitz, H. M.: Changes in the corneal endo thelium as a function of age. Exp. Eye Res. 22:587, 1976. 3. Rao, G. N., Shaw, E. L., Arthur, E. J., and Aquavella, J. V.: Endothelial cell morphology and corneal deturgescence. Ann. Ophthalmol. 11:885, 1979. 4. Goddette, D. W., and Frieden, C : Actin polym erization. The mechanism of action of cytochalasin D. J. Biol. Chem. 261:15974, 1986. 5. Gordon, S. R., Essner, E., and Rothstein, H.: In situ demonstration of actin in normal and injured ocular tissues using 7-nitrobenz-2-oxa-l,3-diazole phallacidin. Cell Motil. 4:343, 1982. 6. Gordon, S. R.: The localization of actin in divid ing corneal endothelial cells demonstrated with nitrobenzoxadiazole phallacidin. Cell Tissue Res. 229:533, 1983. 7. Fujino, Y., and Tanishima, T.: Actin in woundhealing of rabbit corneal endothelium. II. Study by nitrobenzoxadiazole-phallacidin method. Jpn. J. Ophthalmol. 31:393, 1987. 8. Jumblatt, M. N., Matkin, E. D., and Neufeld, A. H.: Pharmacological regulation of morphology and mitosis in cultured rabbit corneal endothelium. Invest. Ophthalmol. Vis. Sci. 29:586, 1988.
335
9. McCarey, B. E., Edelhauser, H. F., and Van Horn, D. L.: Functional and structural changes in the corneal endothelium during in vitro perfusion. In vest. Ophthalmol. Vis. Sci. 12:410, 1973. 10. Edelhauser, H. F„ Van Horn, D. L„ Schultz, R. O., and Hyndiuk, R. A.: Comparative toxicity of intraocular irrigating solutions on the corneal endo thelium. Am. J. Ophthalmol. 81:473, 1976. 11. Fujino, Y., and Tanishima, T.: Actin in woundhealing of rabbit corneal endothelium. I. Study by immunoperoxidase method. Jpn. J. Ophthalmol. 31:384, 1987. 12. Joyce, N. C , Matkin, E. D., and Neufeld, A. H.: Corneal endothelial wound closure in vitro. Effects of EGF a n d / o r indomethacin. Invest. Oph thalmol. Vis. Sci. 30:1548, 1989. 13. Watsky, M. A., McDermott, M. L., and Edel hauser, H. F.: In vitro corneal endothelial permeabil ity in rabbit and human. The effect of age, cataract surgery and diabetes. Exp. Eye Res. 49:751, 1989. 14. Araie, M.: Carboxyfluorescein. A dye for eval uating the corneal endothelial barrier function in vivo. Exp. Eye Res. 42:141, 1986. 15. Wessells, N. K„ Spooner, B. S., Ash, J. F., Bradley, M. O., Luduena, M. A., Taylor, E. L., Wrenn, J. T., and Yamada, K. M.: Microfilaments in cellular and developmental processes. Contractile microfilament machinery of many cell types is reversibly inhibited by cytochalasin-B. Science 171:135, 1971. 16. Kleinzeller, A., and Ziyadeh, F. N.: Cell vol ume regulation in epithelia with emphasis on the role of osmolytes and the cytoskeleton. In Beyenbach, K. W. (ed.): Cell Volume Regulation, vol. 4. Basel, Kar ger, 1990, pp. 59-86. 17. Gruen, D. W. R., and Wolfe, J.: Lateral ten sions and pressures in membranes and lipid monolayers. Biochim. Biophys. Acta 688:572, 1982. 18. Schultz, R. O., Matsuda, M., Yee, R. W., Edel hauser, H. F., and Schultz, K. J.: Corneal endothelial changes in type I and type II diabetes mellitus. Am. J. Ophthalmol. 98:401, 1984.
