Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Structure and Composition of Rat Precorneal Tear Film A Study by an In Vivo Cryofixation Hai-Bo Chen, Shigeki Yamabayashi, Bo Ou, Yuko Tanaka, Shinichi Ohno,* and Shigeo Tsukahara Purpose. To visualize the in vivo structure and to investigate the composition of rat precorneal tear film. Methods. An in vivo cryofixation with freeze substitution method of electron microscopy was used for the study. For light and transmission electron microscopy, a small amount of aluminum powder was used as a tracer spread on the corneal surface. The eyeballs were immediately and quickly frozen by pouring an isopentane-propane mixture cooled by liquid nitrogen directly over the eyes. For scanning electron microscopy, the corneal surface was freezefractured after the cryofixation. The specimens were then freeze-substituted and prepared conventionally for microscopic observation. Results. The tear film appeared as a layer of homogeneous and fine network-like structures varying from 2 to 6 //m in thickness on the corneal surface, with a membrane-like layer covering its surface. The aluminum powder was located on the surface of the tear film. The tear film could be removed completely by applying 10% or 20% acetylcysteine, but not by phosphate buffer. Conclusions. The in vivo structure of the rat tear film is composed primarily of mucus, with a lipid layer covering its surface but without a free aqueous layer. The "three layers theory" of tear film structure requires revisions. Invest Ophthalmol Vis Sci. 1997; 38:381-387. 1. he optical integrity and normal function of the eye can not be maintained without the presence of the precorneal tear film, a very thin fluid film on the exposed part of the ocular surface. The precorneal tear film serves an optical function by maintaining an optically uniform corneal surface, a mechanical function by flushing cellular debris and foreign matter from the cornea and conjunctival sac and by lubricating the surface, a corneal nutritional function, and an antibacterial function. 12 In spite of these important functions, however, little information is available about the morphology of the tear film. It is generally thought that it consists of three layers: the superficial lipid layer (less than 0.1 fxm in thickness) from Meibomian secretion; the middle aqueous layer (approxiFrom the Department of Ophthalmology and the * Department of Anatomy, Yamanashi Medical University, Tamaho, Yamanashi, 409-38 Japan. Supported by a Scientific Research Grant ofJapan Monbusyou (No. 07407050). Submitted for publication June 17, 1996; revised September 3, 1996; accepted September 16, 1996. Proprietary interest categmy.N Reprint requests: Shigeo Tsukahara, Department of Ophthalmology, Yamanashi Medical University, Tamaho, Yamanashi, 409-38 Japan. mate 7 /xm), which is the main component, secreted by the lacrimal gland and the accessory glands of Krause and Wolfring; and the deepest mucus layer (0.02 to 0.05 fim) elaborated by goblet cells of the conjunctiva.1'2 This "three layers theory" of the structure and composition of tear film was proposed initially by Wolff,3 based on histologic studies of the eye surface and on observation of the corneal surface with a slitlamp. Many investigators have studied and measured the tear film with various methods.4"8 No study has so far provided enough evidence to confirm the three layers theory or demonstrated the entire structure of the precorneal tear film morphologically, because the tear film could not be preserved by conventional chemical fixation.910 Accordingly, the three layers theory of tear film structure remains unconfirmed. Recent studies of the tear film have reported results that refute the three layers theory and suggest that this theory has to be revised.11"17 We have reported that a layer of homogeneous material considered to be tear film could be preserved on the rat corneal surface by an in vivo cryofixation with a freeze Investigative Ophthalmology & Visual Science, February 1997, Vol. 38, No. 2 Copyright © Association for Research in Vision and Ophthalmology Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933422/ on 06/18/2017 381 382 Investigative Ophthalmology & Visual Science, February 1997, Vol. 38, No. 2 substitution (VC-FS) method of electron microscopy,9'10 but whether that was its entire structure remains unresolved. In the current study, we again used the VC-FS method to confirm the entire structure of the tear film with a tracer, to investigate the composition of tear film by a rnucolytic agent, and to estimate the thickness of the tear film. • / METHODS Animal experiments conformed with the recommendations of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Nineteen Wistar rats (19 eyes) at the age of approximately 15 weeks were used for the study. Sodium pentobarbital was injected intraperitoneally (4 mg/100 g) to anesthetize the rats for the experiments. Specimens Prepared by the VC-FS Method The eyelids were artificially opened not beyond a physiologic width after anesthesia. A small amount of aluminum powder, used as a tracer, was spread over the surface of five eyes (five rats) to locate the surface of the tear film. Then 10 eyeballs of 10 rats (including the above five eyes) were quickly frozen and freezesubstituted by the VC-FS method described previously.9'10 Then, for light microscopy (LM) and transmission electron microscopy (TEM), the five eyes that had the aluminum powder spread on them were embedded in epoxy resin and semithin or ultrathin sections were cut on a microtome (MT2-B; Sorvall). After staining the semithin sections with toluidine blue and the ultrathin sections with uranyl acetate and lead citrate, the specimens were observed under LM and TEM (H-500; Hitachi). For scanning electron microscopy (SEM), the corneal surface of another five eyes was freeze-fractured with a scalpel in liquid nitrogen after in vivo freezing. After freeze-substitution, the specimens were coated with platinum in a vacuum evaporator (E-1030; Hitachi) after being dried in a critical point-drying apparatus (CP-2; Hitachi) and then observed under a SEM (S-4500; Hitachi). Specimens Prepared by the VC-FS Method After Mucolytic Agent Treatment To investigate the composition of tear film, 10% and 20% solutions of acetylcysteine, a type of mucolytic agent,18 in 0.1 M phosphate buffer (PB) were applied according to the method described by Prydal et al.15'16 The solutions were applied for 5 minutes to each of three eyes, after which the eyes were irrigated for 30 seconds with PB. Three control eyes were treated only with PB in the same way. After these procedures, the eyes were in vivo cryofixed, and the specimens for LM, TEM, and SEM were prepared as described above. FIGURE l. Photomicrograph of transmission electron microscopic image. Tear film (asteiish) appears to be a layer of a homogeneous material on the corneal surface. Glycocalices (arrows) are denser and mainly distributed on the tip of the microvilli. (inset) Photomicrograph of LM image. The tracer of aluminum powder, appearing as black masses (arrows), are confirmed to be on the surface of the tearfilm(ast&isks). E = corneal epithelial cell.. RESULTS Specimens Prepared Directly by the VC-FS Method Under LM, the tear film appeared as a layer of homogeneous material on the corneal surface. The aluminum powder, which appeared as black masses, was present on the surface of the tear film (Fig. 1, inset). Under TEM, xhe tear film appeared to be a layer of a homogeneous and fine network-like structure of 2 to 6 jum thickness on the corneal surface (Fig. 1). A dense line measuring about 12 nm in thickness covered the surface of the network-like structure (Figs. 1,2). Under high magnification, the dense line consisted of two parts differing in electron density. The inner part, in contact with the network-like structure and almost as thick as half of the phospholipid bilayer of the epithelial cell membrane, was denser than the outer one that faced the atmosphere (Fig, 2). The glycocalices of the surface epithelial cells, which were denser and mainly distributed on the tip of the microvilli, overlapped the network-like structure of the tear film; no interface could be identified between them (Figs. 1,2). Under SEM, contrary to the findings with the conventional fixation,1019'20 the outlines and types of surface epithelial cells (light, dark, and medium cells) could not be seen because of the presence of the tear film (Fig. 3). When the edge of the freezefractured region was observed, however, the tear film showed network-like structures as under TEM three- Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933422/ on 06/18/2017 383 In Vivo Structure of Rat Precorneal Tear Film ture of the tear film was almost completely removed after application of 10% acetylcysteine solution, but the glycocalices still remained between the microvilli and no definite damage was found (Fig. 7). DISCUSSION Previous studies19"21 as well as our reports910 have shown that preparation of corneal specimens by conventional chemical fixation removes the tear film on the corneal surface. In the current study, we investigated the structure and composition of the tear film using the VC-FS method. We have found that the tear film appears as a layer of homogeneous and fine network-like structure of 2 to 6 fim in thickness with a membrane-like layer covering its surface. Aluminum powder, which was used as a tracer, was confirmed to be located on the surface of the tear film. The structure of the tear film could be removed completely by applying 10% or 20% acetylcysteine, but not by PB. Nichols et al" have demonstrated a similar structure on the corneal surface of guinea pig eyes by cryofixation with a metal contact method. Their conclusion that this structure was only the mucus layer of the tear film was probably based on the unconfirmed three layers theory. No other evidence supporting the conclusion was provided. Moreover, some loss of, or FIGURE 2. High magnification of transmission electron miinterference with, the tear film probably occurred croscopic image, showing that the tear film appears as a when the eyes were enucleated and had the cooled homogeneous and fine network-like structure (asterisk) with metal applied. These problems, however, were a dense line covering its surface (arroiuhead). The dense line avoided in the current study by using the VC-FS consists of two parts differing in density. The inner part, method, a better way to preserve the ultrastructure of which is in contact with the network-like structure and almost as thick as half of the lipid bilayer of the epithelial cell membrane, is denser than the outer one that faces the atmosphere. Glycocalices [arrows) overlap the network-like structure of the tear film. E = corneal epithelial cell. dimensionally (Fig, 4). The dense line on the surface of the network-like structure was confirmed to be a membrane-like layer (Figs. 2,4). Specimens After Mucolytic Agent Treatment In contrast to control eyes washed with PB, in which the structure of the tear film was still observed on the corneal surface and the outlines or microvilli of the surface cells could not be seen (Fig. 5), neither tear film nor glycocalices of the surface cells could be found on the corneal surface after treatment with 20% acetylcysteine solution. The superficial epithelial cells seemed to be damaged by this concentration, so that cytolysis occurred and the cytoplasm looked transparent under TEM (Fig. 6a). The cell membrane showed spotty breakage under SEM (Fig. 6b). Outlines and microvilli of the surface cells could be seen as in the conventional method (Fig. 6b).10ll9>2° Also, the struc- FIGURE 3. Photomicrograph of scanning electron microscopic image, showing the freeze-fractured region (F) and unfractured region (UF) of the corneal surface. Arrows indicate the freeze-fractured edge. Outlines or the microvilli of the surface epithelial cells cannot be seen when the unfractured region is studied. Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933422/ on 06/18/2017 384 Investigative Ophthalmology & Visual Science, February 1997, Vol. 38, No. 2 of these molecules through disulfide bridges produces polymers of high molecular weight that bind similar polymers to form a gel by weak interactions of their carbohydrate chains. The mucus has been demonstrated to be a fine network-like structure. 2e ~ 28 The homogeneous and fine network-like structure, the main part of the tear film found in the current study, was similar to that of mucus. It overlapped with the glycocalices of the epithelial cells, which also consist primarily of glycoproteins,29'30 without any morphologic difference between them except for electron density. Furthermore, this structure could be completely removed by a mucolytic agent but not by PB. Accordingly, we conclude that it is a layer of mucus. Mucus in dilute solution is known to be highly hydrated. 24 Tear fluid from the lacrimal gland provides such a diluent in the precorneal tear film. Therefore, the mucus of the tear film might be dilute. This proposal also derives support from unpublished data of a histochemical study for polysaccharides: the networklike structure was weakly stained by the periodic acid-thiocarbohydrazide-silver protein (PA-TCHSP) method in TEM. A tear film in this state might be more suitable for its moistening and protective functions, 12 according to the properties and functions of the mucus.24'25 FIGURE 4. Part of Figure 3 under high magnification. At the freeze-fractured edge, the tear film is also composed of the network-like structure (asterisks), as seem under transmission electron microscopic, and the dense line that is three-dimensionally confirmed to be a membrane-like layer (M). The network-like structure shown here on scanning electron microscopic is looser than on transmission electron microscopic, probably because of loss of soluble proteins in the tear film during freeze-substitution. Arrows indicate the microvilli of the surface epithelial cell, arrowheads indicate the underlying epithelial cells. E = corneal epithelial cells. tissue in its natural state. 9 The fact that the aluminum powder was observed on the surface of the tear film confirms that all the constituents on the corneal surface were preserved. In other words, the structure of the tear film shown in the current study was complete. It seems unlikely that any additional layers, such as an aqueous layer, could have been present and not found by our techniques. Although there may be other sources, such as that from conjuncdval nongoblet epithelial cells or corneal epithelial cells, 21314 the mucus of the tear film has been concluded to be mainly elaborated by conjunctival goblet cells. 1 " 31122 Mucus from different sources has a similar structure. 1822 " 25 Mucus is composed of glycoproteins containing many carbohydrate groups. These glycoproteins are large, elongated molecules with a protein backbone to which oligosaccharides are attached in a botde-brush configuration. Crosslinking With slitlamp microscopy, the interference patterns of a lipid layer can be observed on the corneal surface1'3'5'12 and this lipid layer has been physically proven to be present on the anterior surface of the tear film.31 This lipid layer plays an important role in reducing the evaporation rate of the tear film, thereby stabilizing it.1"3'6 In the current study, the dense line or the membrane-like layer was located on the outermost surface of the tear film. Its composition has been shown to be different from that of the network-like structure by means of the PA-TCH-SP method, with which it failed to stain, whereas the network-like structure did stain (unpublished data). We conclude, therefore, that it is most likely to be the lipid layer of the tear film. The tipid layer of the tear film is secreted by the Meibomian glands1'2'3 and is composed both of nonpolar lipids, such as wax esters, and polar lipids, such as phospholipids. 2 ' 32 ' 33 The phospholipids are capable of forming a monolayer molecular membrane on a water/air interface and of forming the bilayer structure of the living biological membrane. 3435 The cell membrane, a phospholipid bilayer membrane, is visible as two parallel railroad tracks on electron micrographs when their polar hydrophilic heads are reduced by osmium tetroxide. 29 ' 35 In the current study, the inner part of the dense line covering the surface of the network-like structure (Fig. 2) was almost as thick as half of the phospholipid bilayer of corneal epithelial cell membranes (like a single railroad track) and, therefore, is probably a monolayer membrane Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933422/ on 06/18/2017 385 In Vivo Structure of Rat Precorneal Tear Film FIGURES. Specimens after PB washing, (a) Photomicrograph of transmission electron microscopic image. The tear film is still present on the corneal surface and shows almost the same findings as in Figure 1 or Figure 2. The loose network-like structure (asterisks) results from poor cryofixation, which might be due to an increase in the aqueous component when washing. Arrow indicates the dense line, (b) Photomicrograph of scanning electron microscopic image. The corneal surface also shows the same findings as in the unfractured region in Figure 3; surface epithelial cells cannot be seen. E = corneal epithelial cells. formed by the polar lipids, whereas the outer one might be the nonpolar lipids, such as wax esters. Recently Greiner et al33 have suggested that the phospholipids secreted by the Meibomian glands form a planar noncellular membrane separating hydrophobic lipids from an aqueous environment. Our findings seem to provide morphologic evidence supporting this suggestion. The lipid layer has been measured to be from 0.04 to 0.5 /xm in the human 1 ' 2 5 8 ; this has been determined 6. Specimens after 20% acetylcysteine solution treatment, (a) Photomicrograph of transmission electron microscopic image. Neither tear film nor glycocalices of the surface cells can be found on the corneal surface (arrow). The superficial epithelial cells (E) seem to be damaged by this concentration, so that cytolysis has occurred and the cytoplasm looks transparent, (b) Photomicrograph of scanning electron microscopic image. The tear film has been completely removed and the structures of the corneal surface were the same as that found by the conventional method, in which the oudines or the three types of surface cells can be seen clearly. Breakage of the cell membrane (arrows) resulting from damage is mainly found in light (L) or medium (M) cells. D = dark cell. FIGURE Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933422/ on 06/18/2017 386 Investigative Ophthalmology & Visual Science, February 1997, Vol. 38, No. 2 FIGURE 7. Specimens after 10% acetylcysteine solution treatment, (a) Photomicrograph of transmission electron microscopic image, (b) Photomicrograph of scanning electron microscopic image, (c) Part of b was examined under high magnification. Tear film is completely removed, but glycocalices still remain between the microvilli (mrows). No definite damage is found. E = corneal epithelial cell, L = light cell; M = medium cell; D = dark cell. mainly by means of measuring the reflection of the tear film and depends on various theoretical foundations.5'8 It is possible that the different thickness of the lipid layer seen in the rat tear film is a species difference. Collectively, our data indicate that, at least in rat, the structure of die tear film is mainly composed of dilute mucus with a lipid layer covering its surface, but without a free aqueous layer. Holly 6 once suggested that the aqueous layer of the tear film consisted of two parts of dilute semigel near the epithelium and a highly dilute mucin solution toward the surface of the tear film. Measuring the thickness of the tear film after the application of a mucolytic agent, Prydal et al15'16 have proposed that the tear film was largely composed of mucus without a free aqueous layer. Observing the lipid layer of the tear film widi noncontact specular microscopy, Danjo et al12 have shown that the lipid layer must be in contact widi mucin for interaction. Recendy Tiffany et al17 also suggested that the theories of tear film structure need revision based on their study of the soluble mucins in tears. Our findings support these authors' suggestions. The thickness of the tear film has been measured by various methods. 4 ' 7 '" Prydal et al 1516 have criticized these methods and estimated the tiiickness of tear film in various animal species; the tear film thickness was reported to be 34 to 45 fira in human and 12 fim in rat. These results are larger dian earlier reports of 7 ^ m i,2,4.7 Moreover, their results showed that there was no variation in thickness at different sites on die cornea surface. Our results, in which the tear film varied from 2 to 6 //m in thickness, do not agree with those of Prydal et al.15 The variation in tear film thickness might be partially due to the process of exfoliation of aged cells, resulting in their elevation so that the tear film on them becomes thinner. 9 It is still uncertain why the thickness of the tear film in the current study differs so much from that of Prydal et al.15 In our results, the thickness of the tear film seemed to be affected by the width of the palpebral fissure. The tear film fissure was reported to measure 0.8 to 1 ^m when the eyes were cryofixed in the protruded posidon for easy enucleation after being frozen.9 It might be that protruding die eyes resulted in extremely wide opening of the palpebral fissure so that die tear film became thinner, whereas in the current study the eyes were cryofixed when the eyelids were not held open beyond a physiologic widdi. The dependence of the thickness of the tear film on the width of the palpebral fissure suggests that die tear film is so dilute diat it flows easily. Some difference may be explained by the measuring methods used by Prydal et all;j'16; for example, difficulty in focusing the laser beam on the illdefined (even with an interferometer and slitlamp microscope) surface of the tear film and the effect of oil on the tear film when the confocal microscope was applied. Key Words in vivo cryofixation, microscopy, mucus, precorneal tear film, three-layered structure References 1. Milder B. The lacrimal apparatus. In: Moses RA, Hart WMJ, eds. ADLER'S Physiofogyof the Eye, Clinical Applica- tion. Washington: The C. V. Mosby Company; 1987: 15-35. 2. Holly FJ and Lemp MA. Tear physiology and dry eyes. Surv Ophthalmol 1977; 22:69-87. 3. Wolff E. The mucocutaneous junction of the lid margin and the distribution of the tear fluid. Trans Ophthalmol Soc UK. 1946;66:291-308. Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933422/ on 06/18/2017 387 In Vivo Structure of Rat Precorneal Tear Film 4. Mishima S. Some physiological aspects of the precorneal tear film. Arch Ophthalmol. 1965; 73:233-241. 5. McDonald JE. Surface phenomena of the tear film. Am J Ophthalmol 1969; 67:56-64. 6. Holly FJ. Formation and rupture of the tear film. Exp Eye Res. 1973; 15:515-525. 7. Benedetto DA, Shah DO, Kaufman HE. The instilled fluid dynamics and surface chemistry of polymers in the preocular tear film. Invest Ophthalmol. 1975; 14:887-902. 8. Olsen T. Reflectometry of the precorneal film. Ada Ophthalmol. 1985; 63:432-438. 9. Chen HB, Yamabayashi S, Ou B, Ohno S, Tsukahara S. Ultrastructural studies on the corneal superficial epithelium of rats by in vivo cryofixation with freeze substitution. Ophthalmic Res. 1995;27:286-295. 10. Chen HB, Ou B, Yamabayashi S, Ohno S, Tsukahara S. Ultrastructural study on rat precorneal tear film by the quick freezing freeze-substitution method. J Jpn Ophthalmol Soc. 1996; 100:453-457. 11. Nichols BA, Chiappino ML, Dawson CR. Demonstration of the mucus layer of the tear film by electron microscopy. Invest Ophthalmol Vis Sci. 1985; 26:464473. 12. Danjo Y, Hamano T. Observation of precorneal tear film in patients with Sjogren's syndrome. Ada Ophthalmol Scand. 1995; 73:501-505. 13. Gipson IK, Yankauckas M, Spurr-Michaud SJ, Tisdale AS, Rinehart W. Characteristics of a glycoprotein in the ocular surface glycocalyx. Invest Ophthalmol Vis Sci. 1992;33:218-227. 14. Watanabe H, Fabricant M, Tisdale AS, Spurr-Michaud SJ, Lindberg K, Gipson IK. Human corneal and conjunctival epithelia produce a mucin-like glycoprotein for the apical surface. Invest Ophthalmol Vis Sci. 1995; 36:337-344. 15. PrydalJI, Campbell FW. Study of precorneal tear film thickness and structure by interferometry and confocal microscopy. Invest Ophthalmol Vis Sci. 1992; 33: 1996-2005. 16. Prydal JI, Artal P, Woon H, Campbell FW. Study of human precorneal tear film thickness and structure using laser interferometry. Invest Ophthalmol Vis Sci. 1992;33:2006-2011. 17. Tiffany JM, Pandit JC and Bron AJ. Soluble mucins 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. and the physical properties of tears. ARVQ Abstracts. Invest Ophthalmol Vis Sci. 1996;37:S845. Creeth JM. Constituents of mucus and their separation. BrMedBull. 1978;34:17-24. Hoffmann F. The surface of epithelial cells of the cornea under the scanning electron microscope. Ophthalmic Res. 1972;3:207-214. Hazlett LD. Corneal and ocular surface histochemistry (Review). ProgHistochem Cytochem. 1993;25:1-60. Hogan MJ, Alvarado JA, Weddell JE. Histology of the Human Eye. Philadelphia: WB Saunders Company; 1971:55-109. Moore JC, Tiffany JM. Human ocular mucus. Origins and preliminary characterization. Exp Eye Res. 1979; 29:291-301. Allen A, Pain RH, Robson TR. Model for the structure of the gastric mucous gel. Nature. 1976;264:88-89. Allen A. Mucus-a protective secretion of complexity. Trends Biochem Sci. 1983;8:169-173. Meyer FA. Comparison of structural glycoproteins from mucus of different sources. Biochim Biophys Ada. 1977;493:272-282. Specian R, Neutra MR. Mechanism of rapid mucus secretion in goblet cells stimulated by acetylcholine. JCellBiol. 1980; 85:626-640. Sandborn EB. Cells and Tissues by Light and Electron Microscopy. Volume 2. Tokyo: Academic Press, Inc; 1970:25-89. Schumacher GFB. Biochemistry of cervical mucus. FertilSteril. 1970;21:697-705. Darnell J, Lodish H, Baltimore D. Molecular Cell Biology. 2nd ed. New York: WH Freeman; 1990:491-530. Lehninger AL. Biochemistry. 2nd ed. New York: Worth Publishers, Inc; 1979:249-277. Brauninger GE, Shah DO and Kaufman HE. Direct physical demonstration of oily layer on tear film surface. Am J Ophthalmol. 1972; 73:132-134. Tiffany JM. The meibomian lipids of the rabbit. I. Overall composition. Exp Eye Res. 1979;29:195-202. Greiner JV, Glonek T, Korb DR, Booth R and Leahy CD. Phospholipids in meibomian gland secretion. Ophthalmic Res. 1996;28:44-49, Rawn JD. Biochemistry. Burlington: Neil Patterson Publishers; 1989:209-232. Lehninger AL. Biochemistry. 2nd ed. New York: Worth Publishers; 1979:279-306. Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933422/ on 06/18/2017