The Corneal Endothelium in the Blowfish (Torquigener Pleurogramma)

The Corneal Endothelium in the Blowfish (Torquigener pleurogramma)

Shaun P. Collin, M.Sc. (Melb), Ph.D.(Qld), and H. Barry Collin, A.M., Ph.D. (Melb), D.Sc. (NSW), F.R.C.Path. (Lond)

Purpose. In vertebrates, a corneal endothelium is essential for humans (1), monkeys (2,3), and various experimental the maintenance of corneal transparency in a variety of envi- animals, including rabbits (4), guinea pigs (5,6), and rats ronments. Knowledge of the surface structure of the corneal (7). However, there are few reports on the surface struc- endothelium may assist our understanding of this unique tissue and its evolutionary development. Although there have been ture and density of corneal endothelial cells in teleosts. many studies of the corneal endothelium of humans and some In mammals, the appropriate hydration levels of the mammals, there have been few in other vertebrates. Methods. cornea are maintained by the endothelium, which ac- The field emission scanning electron microscope was used to tively pumps bicarbonate ions and water out of the cor- study the surface structure of the corneal endothelium in the neal stroma (8). The interdigitating endothelial cells are blowfish, Torquigener pleurogramma (Tetraodontidae, Teleos- tei), and to examine cell density. Cell areas were measured by joined by junctional complexes, which help reduce fluid using image-analysis software. Results. The endothelium is flow between cells (9) and are covered by a viscous composed of a sheet of interdigitating hexagonal and pentago- coating, which reduces lipid membrane surface tension nal cells with a mean area of 154 ␮m2 and a density of 6,486 2 to promote wetability (10,11). The role of the endothe- cells/mm . Two types of surface features are identified; pri- lium in maintaining corneal transparency is vital. mary cilia and microvilli. The cilia are cylindrical, protrude from either a pore or circular indentation in the cell center, and The presence of a primary cilium protruding from the possess a knob-like ending. The microvilli are button-like pro- endothelial cell surface in only a small range of mam- trusions with a density of ∼3.5×105 microvilli/mm2 or 54 mals including humans (1,2,12), other primates (2,3), microvilli/cell in central cornea. Conclusion: The results show and rabbits (4) has led to the conclusion that these struc- that the surface structure of teleost endothelial cells is similar to tures are not widely found in other vertebrate classes. those described for other vertebrates and indicate that cell density varies across classes, with the presence of cilia a more However, with the recent finding of cilia in birds (13), widespread occurrence than previously believed. endothelial cilia may be a ubiquitous structure through- Key Words: Corneal endothelium—Primary cilia—SEM— out the vertebrate kingdom. Cell density—Fish. The aim of this field emission scanning electron microscopy study was to examine the surface morphology and cell density of the corneal endothelial cells of a In vertebrates, a transparent cornea is essential for representative teleost, the blowfish, Torquigener pleuro- clear vision, which in turn is essential for survival. How- gramma (Tetraodontidae, Teleostei). The data will in- ever, vertebrates live in a great variety of environments, crease our understanding of the environmental con- which require aerial, terrestrial, and aquatic vision and straints placed on the nonmammalian cornea and add to which place different demands on the cornea. The struc- our knowledge of the evolutionary development of this ture of the endothelium has been extensively reported for unique tissue.

Submitted December 17, 1998. Revision received April 29, 1999. MATERIALS AND METHODS Accepted May 3, 1999. From the Department of Zoology, The University of Western Aus- Three specimens of the blowfish, Torquigener pleu- tralia, Nedlands, Western Australia (S.P.C.), and Department of Op- tometry and Visual Sciences, The University of Melbourne, Parkville, rogramma (Tetraodontidae, Teleostei), were collected Victoria (H.B.C.), Australia. from Cockburn Sound, Western Australia, and subse- Address correspondence and reprint requests to Dr. S.P. Collin, De- quently anesthetized and killed in methanesulfonate salt partment of Zoology, the University of Western Australia, Nedlands 6907, Western Australia, Australia. E-mail: [email protected]. (MS 222; 1:2,000) in accordance with the ethical guide- edu.au lines of the National Health and Medical Research Coun-

231 232 S.P. COLLIN AND H.B. COLLIN

The central corneal surfaces of six corneas were examined by using a Joel FESEM (field emission scanning electron microscope) with an accelerating voltage of 3 kV. Results were recorded both on 35-mm film and dig- itally. The areas of Յ20 individual endothelial cells were obtained by digital analysis of the computer images by using Image-analysis software (Optimas, Adept Elec- tronic Solutions, Australia). By using this procedure, the mean endothelial cell density and standard deviation were calculated, providing the extent of variation in cell area and a measure of polymegethism. Measurements of cilia and microvilli were made on photographic prints by using a magnifier and graticule.

FIG. 1. Low power scanning electron micrograph showing the corneal endothelium of the blowfish, Torquigener RESULTS pleurogramma, comprising a mixture of elongated hexagonal and pentagonal cells covered by numerous micro- The corneal endothelium of the blowfish, T. pleuro- villi. gramma, is composed of a sheet of interdigitating cells with a mean area of 154 ± 28 ␮m2 and a density of 6,486 cil of Australia. After enucleation, the cornea was dis- ± 1,194 cells/mm2. The cell borders are difficult to dif- sected free and fixed in Karnovsky’s fixative (2% para- ferentiate but are often distinguished by the different formaldehyde, 2.5% glutaraldehyde, 0.1 M sodium level of the neighboring cell(s), which forms an inter- cacodylate buffer, 2% sucrose, and 0.1% calcium chlo- rupted ridge (Յ2.5 ␮m long and 91 nm in width). The ride at pH 7.2) and rinsed in 0.1 M sodium cacodylate cells are irregular in shape but are predominantly pen- buffer. tagonal and hexagonal (Fig. 1.). Postfixation of all specimens was in 0.1% osmium The smooth surface of the cells is interrupted by two tetroxide in 1.0 M cacodylate buffer, followed by dehy- types of features, which project into the anterior cham- dration in a graded series of alcohols. Specimens were ber. The predominant feature comprises small (Յ460 nm critical-point dried in a Polaron critical-point dryer and in width), circular microvilli, which are scattered mounted on 10-mm aluminum stubs with double-sided throughout the cell with a density of 350,700 microvilli/ tape. Each specimen was oriented and/or hemisected so mm2. The microvilli are distributed randomly over the that one half of the cornea was inverted and both sides surface of each cell (54.4 ± 17.3 microvilli/cell), al- were displayed. This allowed direct comparison and dif- though a higher density of microvilli lies along the cell ferentiation between epithelial and endothelial surfaces. borders. The mounted specimens were coated with 12–15 nm of In the central region of approximately one third of the gold–palladium in a Polaron sputter coater and placed in endothelial cells, a single cilium, 1.16 ± 0.14 ␮min an oven at 40°C overnight before being examined. length and 105 ± 16 nm in diameter, also projects into the

FIG. 2. a: Scanning electron micrograph of the corneal endothelium of the blowfish, showing its smooth surface interrupted by button-like microvilli. Note also the central cilium protruding from a pore in the cell surface. b: High-power micrograph showing the interdigitations and raised ridges along the junction of three endothelial cell borders.

Cornea, Vol. 19, No. 2, 2000 CORNEAL ENDOTHELIUM IN THE BLOWFISH 233 anterior chamber (Figs. 2a and 3). On rare occasions, two brates is typically a mixture of hexagonal and pentagonal cilia may project from the surface of a single cell. Typi- cells, in which the cell borders are irregular and inter- cally, each cilium is cylindrical, projecting from either a digitating (26). This is confirmed in our study, in which pore or a circular indentation of the endothelial cell sur- all of the endothelial cells in the blowfish were either face. The base of each cilium is ∼30% thicker than the hexagonal or pentagonal and markedly irregular. These central shaft, with the end of each cilium expanded into findings are also similar to the regular array of either a knob (147±9nmindiameter). hexagonal or pentagonal cells bordered by a ruffle of interdigitations for the Florida garfish, Lepisosteus platy- DISCUSSION rhincus (15), and an irregular endothelial cell array in the trout, Salmo gairdneri (7). However, in the goldfish, The presence of a corneal endothelium in the blowfish, Carassius auratus, endothelial cells have been described T. pleurogramma, supports other studies on the teleost as nonpolygonal with rounded borders like those of a cornea (7,13–17). A complete endothelial cell layer has jigsaw pattern (7). also been confirmed in agnathans, although the endothe- The endothelial cell density for the blowfish (6,486 ± lium is covered by a basement membrane (18–20), am- 1,934) appears to be similar to that of the Perth herring, phibians, reptiles, birds (7), and mammals (5,6), includ- Nematalosa vlaminghi (13), but is appreciably higher ing primates (21,22). Therefore, it appears that some spe- than those reported for two species of teleosts by Yee et cies of elasmobranchs are the only group of vertebrates al. (7). However, different techniques were used in the that do not possess an endothelial cell layer, because no two studies. Yee et al. (7) used specular microscopy on endothelial cells were found in 12 species of cartilagi- anesthetized goldfish (C. auratus, with 431 cells/mm2) nous fishes (23–25). It is claimed that an endothelium and trout (S. gairdneri, with 578 cells/mm2), whereas appears unnecessary, as these corneas swell very little Collin and Collin (13) and this study used FESEM after and retain their transparency when immersed in distilled fixation, alcohol dehydration, and critical-point drying, water (24,25). In contrast, in the ray, Torpedo ocellata, which would result in some tissue shrinkage. Doughty endothelial cells were found (25). (27) reported tissue shrinkage of 35.8 ± 1.2% as a result The shape of the corneal endothelial cells of verte- of fixation and critical-point drying for SEM. Hence the

FIG. 3. a–d: High-power micrographs illustrating the structure of a variety of endothelial cilia. Note the base of each cilia is usually associated with a pore or indentation of the surface membrane and a knob-like ending.

Cornea, Vol. 19, No. 2, 2000 234 S.P. COLLIN AND H.B. COLLIN true cell densities will be considerably less, perhaps of as in the rabbit, the cilium is a normal component of the order of 30%, than those reported here. every endothelial cell, is capable of regeneration after Observation of the corneal endothelium through the being broken off, and simply may not have been ob- anterior corneal tissue could also modulate the size of the served previously because of their fragility during pro- image in comparison to direct observation of the endo- cessing. The failure to find cilia in the endothelial cells in thelial surface as in our study. Sperling (28) found that five other species of teleosts by Collin and Collin (13) estimates of endothelial cell densities in intact eyes were may be explained in this way or may be a species- 9.7% less than the estimates in excised eyes. Although it specific difference. However, our finding in teleosts sug- appears not to have been studied in most teleosts, the gests that the endothelial cilium may not be rare or an effect of the corneal optics on the image of the endothe- organelle found only sporadically throughout the verte- lium obtained in specular microscopy should be minimal. brate kingdom, as previously believed. An exception may be the cornea of the sandlance, Lim- The average length of the endothelial cilia in the blow- nichthyes fasciatus (13,14,29), in which the cornea con- fish is 1.16 ± 0.14 ␮m. This compares with 3.9 ± 0.5 ␮m tributes ∼200 diopters to the refractive power of the eye in the rabbit (4), 1.2 ␮m in emu, Dromaius novaehol- and has a thickness of 0.14 mm (14), and hence would landiae (13), 0.5 ␮m in length in the barred owl, Bubo have a magnifying effect on the appearance of the cor- strix (13), and 0.13 ␮m in length in the galah, Eolophus neal endothelium as viewed from the anterior surface. roseicapillus (13). In the rabbit, the cilia appear to pro- The large range of endothelial cell densities found in ject at a preferred angle, which may be dependent on the teleosts (1,900 ± 197 cells/mm2 in the black bream, physiologic activity of the endothelial layer (4). Under Acanthopagrus butcheri, to 11,133 ± 523 cells/mm2 in physiologic stress, these cilia appear to retract within the the seahorse, Hippocampus angustus) by Collin and Col- cell and may not be visible at the cell surface (4). Thus lin (13) is also reported for the small sample of other the smaller length of the cilia described for T. pleuro- representative (nonmammalian) vertebrate classes stud- gramma (Յ∼1.3 ␮m) and the three species of birds may ied (6,7,13). There are, in fact, no systematic differences be due to species-specific differences, tissue shrinkage, among any of the cell densities of the various classes and and/or the result of physiologic stress associated with species examined, in spite of the range of corneal envi- tissue preparation, resulting in the incomplete protrusion ronments represented, for example, aerial, terrestrial, and of the cilia through the cell membrane. aquatic (13). The ciliary shaft in the rabbit endothelium is com- The size and regularity of the shape of endothelial posed of nine peripheral doublets without central fibrils cells are influenced by their ability to regenerate. In (9+0 axoneme), which converts to an 8+1 when a dou- many vertebrates, the corneal endothelial cells mitose blet, which was located peripherally near the base, shifts and regenerate, and in some mammals, there is a mitotic toward the center closer to the tip (4). At the base of the turnover of endothelial cells and a mitotic response to cilium is a centriole pair (4). The structure of the cilium endothelial injury (1). However, in both human and non- in T. pleurogramma was not investigated. human primates, although mitosis occurs in the young, The function of the cilium throughout the vertebrates the endothelial cell population does not regenerate in any is unclear (2,4). It does not appear to be motile and is significant fashion in the adult (1,30). This gives rise to structurally associated with the cell centriole pair (35). a decreasing cell density with age (21,22,31–34). Unfor- Klyce and his colleagues (35) suggested that in the hu- tunately, no systematic studies of endothelial cell– man, the cilium may be related to the inability of the cells density changes with age exist for nonhuman vertebrates. to undergo mitosis in the adult, because the corneal en- However, the high density of endothelial cells (11,133 dothelium of humans is not a self-renewing cell layer cells/mm2) in the larval seahorse, H. angustus (13), when (36). Other postulated functions for the cilium include compared with the mean of 3,142 ± 1,052 cells/mm2 for chemoreception, osmoregulation, and/or pressure detec- the remainder of nonlarval teleosts (but of indeterminate tion (4). age) studied by SEM (13,15, and this study) suggest that The small microvilli that project from all the endothe- a relationship may exist, but a thorough developmental lial cells in T. pleurogramma have been described in study in one species is required. many vertebrate groups. The density of microvilli in the This is the first study identifying endothelial cilia in blowfish (54 per cell) is higher than that in humans (35), the teleost cornea, although the presence of cilia was rabbits (26), the emu, D. novaehollandiae (13), and the reported in other vertebrate classes. A single primary barred owl, B. strix (13), which all possess between 20 cilium, which protrudes into the anterior chamber, is oc- and 30 microvilli per cell. Similarly, the microvilli in the casionally found in the endothelial cells of humans (1,2, blowfish (Յ0.46 ␮m in width) are thicker than those 12), the Japanese cynomolgus (Macaca irus), and stump- described in humans [0.1–0.2 ␮m wide (35)], rabbits tail (Macaca speciosa) monkeys (2,3), rabbits (4), and [0.1–0.2 ␮m wide (26)], the barred owl, B. strix [0.15 three species of birds (13). Gallagher (4) considered that, ␮m in width (13)], and the freshwater garfish, Lepisos-

Cornea, Vol. 19, No. 2, 2000 CORNEAL ENDOTHELIUM IN THE BLOWFISH 235 teus platyrhincus [0.2 ␮m in diameter (15)]. The in- cornea, lens and iris in the pipefish, Corythoichthyes paxtoni creased surface area provided by the microvilli may aid (Syngnathidae, Teleostei). Histol Histopathol 1995;10:313–23. 17. Collin HB, Collin SP. The fine structure of the cornea of the in ion transport associated with the endothelial stromal salamanderfish, Lepidogalaxias salamandroides (Lepidogalaxi- dehydrating mechanism, in which both bicarbonate and idae, Teleostei). Cornea 1996;15:414–26. sodium ions are transported from the stroma to the aque- 18. Van Horn DL, Edelhauser HF, Schultz RO. Ultrastructure of the ous humor, creating an osmotic gradient that balances the primary spectacle and cornea of the sea lamprey. J Ultrastruct Res 1969;26:454–64. swelling tendency of the corneal stroma (37). The mi- 19. Pederson HJ, Van Horn DL, Edelhauser HF, Ultrastructural crovilli may also stabilize the endothelial coating changes associated with loss of transparency in the primary spec- (10,11). However, the number and size of microvilli may tacle and cornea of spawning sea lamprey. Exp Eye Res 1971;12: 147–50. be markedly changed after application of toxic sub- 20. Dickson DH, Graves DA, Moyles MR. Corneal splitting in the stances (38–40), suggesting that a more detailed study developing lamprey Petromyzon marinus L. eye. Am J Anat 1982; may be necessary to indicate interspecific differences in 165:83–98. function with respect to the environment. 21. Laule A, Cable MK, Hoffman CE, Hanna C. Endothelial cell population changes of human cornea during life. Arch Ophthalmol 1978;96:2031–5. Acknowledgment: We thank Michael Archer of the Depart- 22. Yee RW, Matsuda M, Schultz RO, Edelhauser HF. Changes in the ment of Zoology, The University of Western Australia, and normal corneal endothelial cellular pattern as a function of age. Brendon Griffin of the Centre for Microscopy and Microanal- Curr Eye Res 1985;4:671–8. ysis, The University of Western Australia, for technical assis- 23. Faure JP. Le de´velopement embryonnaire de la corne´echezun tance. Dr. S.P. Collin is currently an ARC QE II Research Se´lacien, la Rousette (Scyliorhinus canicula L.). Arch Ophthalmol Fellow. This work was funded by Australian Research Council (Paris) 1970;30:883–906. and National Health and Medical Research Council grants to 24. Keller N, Pouliquen Y. Ultrastructural study of the posterior cornea Dr. S.P. Collin. of the dogfish “Scyliorhinus canicula L.” Cornea 1985;4:108–17. 25. Keller N, Pouliquen Y. Ultrastructural study of posterior cornea in cartilaginous fishes. In: Cavanagh HD, ed. The cornea: transac- REFERENCES tions of the world congress of the cornea III. New York: Raven Press, 1988:253–8. 1. Beuerman RW, Pedrosa L. Ultrastructure of the human cornea. 26. Blu¨mcke S, Morgenroth K Jr. The stereo ultrastructure of the ex- Microsc Res Tech 1996;33:320–35. ternal and internal surface of the cornea. J Ultrastruct Res 1967; 18:502–18. 2. Svedbergh B, Bill A. Scanning electron microscopic studies of the corneal endothelium in man and monkeys. Acta Ophthalmol 1972; 27. Doughty MJ. Analysis of areas and shapes of cells on the corneal 50:321–36. surface of the albino rabbit by scanning electron microscopy. Curr 3. Sugita A, Nakano H. Surface ultrastructure of the transition zone Eye Res 1990;9:295–305. between the cornea and the trabecular meshwork in the Japanese 28. Sperling S. Estimates of endothelial cell densities in whole human monkey. Nippon Ganka Gakkai Zasshi 1980;84:1759–64. eyes and excised corneas. Cornea 1986;5:15–8. 4. Gallagher BC. Primary cilia of the corneal endothelium. AmJAnat 29. Pettigrew JD, Collin SP. Terrestrial optics in an aquatic eye: the 1980;159:475–84. sandlance, Limnichthyes fasciatus (Creediidae, Teleostei). J Comp 5. Collin HB, Grabsch BE. The effect of ophthalmic preservatives on Physiol A 1995;177:397–408. the shape of corneal endothelial cells. Acta Ophthalmol 1982;60: 30. Hogan MJ, Alvarado JA, Weddell JE. Histology of the human eye: 93–105. an atlas and textbook. Philadelphia: WB Saunders, 1971. 6. Collin HB, Grabsch BE, Johnston AW. An accurate in vitro as- 31. Binkhorst CD, Loones LH, Nygaard P. The clinical specular mi- sessment of the corneal endothelium. Am J Optom Physiol Opt croscope. Doc Ophthalmol 1977;44:57–75. 1982;59:5–12. 32. Hirst LW, Snip RC, Stark WJ, Maumenee AE. Quantitative cor- 7. Yee RW, Edelhauser HF, Stern ME. Specular microscopy of ver- neal endothelial evaluation in intraocular lens implantation and tebrate corneal endothelium: a comparative study. Exp Eye Res cataract surgery. Am J Ophthalmol 1977;84:775–80. 1987;44:703–14. 33. Bourne WM, O’Fallon WM. Endothelial cell loss during penetrat- 8. Hodson S, Miller F, Riley M. The electrogenic pump of rabbit ing keratoplasty. Am J Ophthalmol 1978;85:760–6. corneal endothelium. Exp Eye Res 1977;24:249–53. 34. Seta¨la¨ K. Corneal endothelial cell density in iridocyclitis. Acta 9. Kreutziger GO. Lateral membrane morphology and gap junction Ophthalmol 1979;57:277–86. structure in rabbit corneal endothelium. Exp Eye Res 1976;23:285– 93. 35. Klyce SD, Beuerman RW. Structure and function of the cornea. In: Kaufman HE, Barron BA, McDonald MB, Waltman SR, eds. The 10. Wolf J. The secretory activity and the cuticle of the corneal endo- cornea. Melbourne: Churchill Livingstone, 1988:3–54. thelium. Doc Ophthalmol 1968;25:150–94. 11. Sperling S, Jacobsen SR. The surface coat on human corneal en- 36. Gipson IK. The anatomy of the conjunctiva, cornea and limbus. In: dothelium. Acta Ophthalmol (Copen) 1980;58:96–102. Smolin G, Thoft RA, eds. The cornea: scientific foundations and clinical practice. 3rd ed. Boston: Little, Brown, 1994:3–24. 12. Andrews PM, Jester JV, Petroll WM, Barry PA, Ichijima H, Cavanagh HD. In vivo osmotic perturbation of intercellular fluid 37. Hodson S, Miller F. The bicarbonate ion pump in the endothelium channels in the rabbit corneal endothelium. Cornea 1994;13: which regulates the hydration of the rabbit cornea. J Physiol 1976; 253–8. 263:563–77. 13. Collin SP, Collin HB. A comparative study of the corneal endo- 38. Collin HB, Grabsch BE, Carroll N, Pain PR. In vivo corneal en- thelium in vertebrates. Clin Exp Optom 1998;81:245–54. dothelial changes due to chlorbutanol. Int Eyecare 1985;1:435–9. 14. Collin HB, Collin SP. The cornea of the sandlance, Limnichthyes 39. Collin HB, Grabsch BE, Carroll N, Pain PR. In vivo effects of fasciatus (Creediidae). Cornea 1988;7:190–203. Na2EDTA on rabbit corneal endothelium. Int Eyecare 1986;2:157– 15. Collin SP, Collin HB. The visual system of the Florida garfish, 62. Lepisosteus platyrhincus (Ginglymodi): II. Cornea and lens. Brain 40. Collin HB, Carroll N. Ultrastructural changes to the corneal endo- Behav Evol 1993;42:98–115. thelium due to benzalkonium chloride. Acta Ophthalmol 1986;64: 16. Collin HB, Collin SP. Ultrastructure and organisation of the 226–31.

Cornea, Vol. 19, No. 2, 2000