Induction of Cross-links in Corneal Tissue - Science Direct

Exp. Eye Res. (1998) 66, 97–103

Induction of Cross-links in Corneal Tissue E B E R H A R D S P O E R L*, M I C H A E L H U H L E    T H E O S E I L E R Department of Ophthalmology, University of Dresden, Dresden, Germany (Received Lund 5 May 1997 and accepted in revised form 15 September 1997) The aim of this study was to investigate the possibility of induction of cross-links in corneal tissue in order to increase the stiffness as a basis for a future conservative treatment of keratectasia. Collagenous biomaterials can be stabilized by chemical and physical agents. The epithelium of enucleated porcine eyes was removed. Eight test groups, 10 eyes each, were treated with UV-light (λ ¯ 254 nm), 0±5 % riboflavin, 0±5 % riboflavin and UV-light (365 nm) blue light (436 nm) and sunlight, and the chemical agents–glutaraldehyde (1 % and 0±1 %, 10 min) and Karnovsky’s solution (0±1%, 10 min). Strips of 5 mm in width and 9 mm in length were cut from each cornea and the stress-strain behaviour of the strips was measured to assess the cross-linking process. For comparison, ten untreated corneas were measured by the same method. Compared to untreated corneas treatment with riboflavin and UV-irradiation as well as weak glutaraldehyde or Karnovsky’s solutions resulted in an increased stiffness of the cornea. The biomechanical behaviour of the cornea can be altered by glutaraldehyde, Karnovsky’s solution, and with riboflavin and UV-irradiation which offers the potential of a conservative treatment of keratoconus. To optimize this effect further investigation is necessary regarding the dose-response and in-vivo application. # 1998 Academic Press Limited Key words : biomechanics ; biomaterials ; collagen ; cornea ; cross-linking ; keratoconus ; proteoglycans.

1. Introduction Biomechanical investigation of human, keratectatic corneas reveals significant differences in elasticity compared to normal corneas, indicating a decreased stiffness of the keratoconus cornea (Andreassen, Simonsen and Oxlund, 1980 ; Edmund, 1988 ; Nash, Greene and Forster, 1982). The microscopic basis for this finding is still unknown, but a reduction of collagen cross-links and a reduction of molecular bonds between neighbouring stromal proteoglycans are thought to be relevant (Wollensak, Ihme and Seiler, 1987 ; Wollensak and Buddecke, 1990). New electron microscopic and X-ray scattering measurements have demonstrated a lack of obliquely oriented anterior-to-posterior lamellae crossing or interweaving the stress bearing lamellae in the anterior stroma (Daxer and Fratzl, 1997 ; Radner et al., 1996). Currently, no conservative treatment of keratectasia is available. The purpose of this study is to assess the value of stromal cross-link induction in order to increase the corneal stiffness, as a basis for the future conservative treatment of keratoconus (Seiler et al., 1996). Induction of cross-links is a well established procedure in polymer chemistry to increase the elastic modulus of materials (Eyre, 1984 ; Mark, 1988). Cross-linking in connective tissue may occur during aging (Albon et al., 1995 ; Balisky et al., 1996 ; Reiser, 1991 ; Yamauchi et al., 1996) and as a side-effect of * Corresponding author : Eberhard Spoerl, Universita$ ts-Augenklinik, Fetscherstr. 74, 01307 Dresden, Germany.

0014-4835}98}01009707 $25.00}0}ey970410

diabetes (Hadley, Meek and Malik, 1996 ; Malik et al., 1992 ; Sady, Khosrof and Nagaraj, 1995 ; Zhao, Nagaraj and Abraham, 1997). Recently cross-linking has been used medically to increase the stability, and reduce the biodegradation of collagen-based biomaterials for bioprostheses such as porcine heart valves, blood vessel prostheses, dural substitute and meniscal allografts (Ersek and Derlerm, 1988 ; Nimni, 1988 ; Perkins and Bui, 1996 ; Pietrucha, 1991 ; Remacle, 1995 ; Wisnewski, Power and Kennedy, 1988). In ophthalmology, it was applied to the production of collagenous biomaterials for synthetic epikeratoplasty by Harner et al. (1996). Collagens cross-linking may be induced by different methods such as nonenzymatic glycation (Hadley et al., 1996 ; Malik et al., 1992 ; McNamara et al., 1996 ; Sady et al., 1995 ; Zhao et al., 1997), irradiation using ultraviolet light with and without photosensitizers (Ersek and Derlerm, 1988 ; Hikichi et al., 1996 ; Milne and Zika, 1992 ; Pietrucha, 1991), and by aldehydereactions (Nimni, 1988 ; Perkins and Bui, 1996 ; Remacle, 1995 ; Wisnewski et al., 1988). In this study, the latter two approaches were investigated. The experiments were designed to compare the stressstrain relation of treated porcine corneas with that of untreated corneas. 2. Materials and Methods Specimen Preparation One hundred and sixty pig eyes were harvested in the slaughter house within 2 hr post mortem and stored at 5°C. Eyes with intact epithelium and a # 1998 Academic Press Limited

98

E . S P O E R L E T A L.

T I Treatment of corneas for cross-link induction Group

n

Irradiation treatment

Chemical treatment

1 2 3 4 5 6 7 8 9

10 10 10 10 10 10 10 10 80

254 nm, 20 min 365 nm, 45 min 436 nm, 45 min sunlight, 120 min none none none none none

None riboflavin 0±5 %, 45 min riboflavin 0±5 %, 45 min riboflavin 0±5 %, 45 min riboflavin 0±5 %, 45 min glutaraldehyde 1 %, 10 min glutaraldehyde 0±1 %, 10 min Karnovsky sol. 0±1 %, 10 min none

central corneal thickness of less than 1 mm were selected. After careful removal of adnexae and corneal epithelium, all corneas were dehydrated in a dextran solution (20 %, Dextran T 500, Pharmacia-Biotech, Freiburg, Germany) for 60 minutes. Afterwards, half of the eyes were treated as listed in Table I. The rest of the corneas were stored in a moist chamber in a dark room (control groups). Following treatment, all corneas (treatment groups as well as control groups) were again stored in a moist chamber in a dark room for another 45 minutes. The corneas were then excised, including a small scleral rim, and vertical strips of corneal tissue were prepared by means of two parallel razor blades set 5 mm apart. The length and width of the strips were measured by means of an electronic caliper (Helios, Basel, Switzerland) and the thickness was determined with an ultrasonic pachometer (Pachette, Technomed, Baesweiler, Germany) at 3 locations. Within 2 minutes after removal from the moist chamber, the corneal strips were clamped and preloaded with a stress of 5±10$ N}m# for 10 minutes. After that, the stress-strain measurements were started. Each measurement of a cornea of the treatment group was followed by one of the control group and so on. Treatment During exposure to blue or ultraviolet light the eyes were mounted in a moist chamber covered with a quartz plate transparent for wavelengths between 250 and 1000 nm. A mercury lamp (HNU 6, EMI, Ilmenau, Germany) served as a light source at a wavelength of 254 nm delivering an intensity of 90 W}m# through the quartz plate. A Xenon lamp (XBO 50, EMI, Ilmenau, Germany) with interference filters for 365 nm and 436 nm delivered an intensity of 20 W}m# through the quartz plate. Sunlight exposure was performed on a sunny day with an estimated light intensity at the cornea of 85 W}m#. Corneas were incubated in various fluids (Table I) within an eye holder. The eyes were mounted up-side down so that the total cornea was immersed in the

incubation solution. After exposure the corneas were rinsed with dextran solution and dried with a wet sponge. The riboflavin solution was prepared by diluting vitamin B2 (B2-inject, Jenapharm, Jena, Germany) in a 20 %-dextran solution to a concentration of 0±5 %. Standard fixatives for SEM, glutaraldehyde 25 %, were also diluted in a 20 %-dextran solution to the concentrations listed in Table I. Dextran powder was added to 0±1 % Karnovsky’s solution (0±1 % glutaraldehyde, 0±1 % paraformaldehyde, 0±1 m Na-phosphate buffer pH 7±4) to obtain a 20 %-dextran solution.

Stress-strain Measurements The stress-strain relation was measured by means of a commercially available microcomputer-controlled material tester (MINIMAT, Rheometric Scientific GmbH, Bensheim, Germany) (Spoerl et al., 1996). This apparatus is capable of applying forces ranging from 0±001 N to 20 N. All measurements were carried out under mineral oil in order to avoid dehydration. Within 2 minutes after removal from the moist chamber, the corneal strips were clamped and a prestress of 5±10$ N}m# was applied for 10 minutes. Within this time interval an equilibrium was obtained in all cases. After that, the strain was increased linearly with a velocity of 1±5 mm min−" and the stress was detected until 2±10& N}m#. The stress-strain data were stored on a hard disc of the computer for further processing.

Statistical Evaluation Ten valid measurements were performed in each treatment group and control group and the stress values necessary for a strain of 4 %, 6 % and 8 % were statistically compared with those of the control groups. Statistical significance was determined by the MannWhitney-test. Differences between treatment groups and control groups were considered significant at P ! 0±05.

I N D U C T I O N O F C R O S S -L I N K S I N C O R N E A L T I S S U E

99

F. 1. Influence of UV-radiation and riboflavin on the stress-strain relation of the cornea (group 1, 2, 4 and 9). Treatment with UV light (254 nm) alone does not reach an effect and riboflavin}436 nm produce only a weak significant stiffness of the cornea. Riboflavin}sunlight and riboflavin}365 nm show a high significant effect.

F. 2. Influence of glutaraldehyde and Karnovsky’s solution on the stress-strain relation of the cornea (group 6, 7, 8 and 9). In spite of the high standard deviation the increase of stiffness is statistically significant for all treatments.

Fluorescence Microscopy The effect of glutaraldehyde penetration into the corneal tissue was qualitatively determined using the fluorescence of chemically induced cross-links (Malik et al., 1992 ; Sady et al., 1995 ; Stolwijk et al., 1992 ;

Uma, Sharma and Balasubramanian, 1996). Three corneas of the treatment groups 6 to 8 and the control group were frozen with liquid nitrogen. Serial sections, each 5–8 µm thick, were collected from the centre of the corneas. The native sections were observed under the fluorescence microscope (PH-2, Olympus,

100


Hamburg, Germany) with a stimulating wavelength range of 450 to 490 nm and an observation wavelength range between 515 and 565 nm. 3. Results All corneas remained clear after the treatments presented. The stress-strain relations are depicted in Figs 1 and 2. The curves show the typical exponential increase of a bio-viscoelastic solid in the so-called ‘ toe range ’. The greatest stiffening effect occurred after treatment with

glutaraldehyde 1 %. However, riboflavin combined with sunlight exposure also resulted in a highly significant stiffening (Table II). Statistically significant stiffening at an elongation of 4 % was obtained in all treatments except in groups 1 (ultraviolet radiation alone) and 5 (riboflavin, no radiation). The stressstrain curve of treatment group 5 (riboflavin alone) was not significantly different from that of the control group and is therefore not shown in Fig. 1. Figure 3 demonstrates the localization of the crosslinks induced by Karnovsky’s solution 0±1 %. The density of induced cross-links decreases from anterior

T II Stiffening of treated corneas as compared to untreated controls P-value at a strain of Group 1 2 3 4 5 6 7 8

Treatment UV (254 nm, 20 min) riboflavin, 365 nm, 45 min riboflavin, 436 nm, 45 min riboflavin, sunlight, 120 min riboflavin glutaraldehyde 1 % glutaraldehyde 0±1 % Karnovsky’s solution 0±1 %

4%

6%

8%

0±390 0±0001 0±0288 0±001 0±3932 0±0093 0±0302 0±0185

0±796 0±0001 0±0433 0±005 0±8534 0±0125 0±0412 0±0185

0±579 0±0003 0±0753 0±7959 0±0164 0±0461 0±0147

F. 3. Fluorescence-microscopy of the untreated (left) and with Karnovsky’s solution 10 min treated cornea (right) to demonstrate the crosslinks. Only the anterior part of the stroma is involved. (The space bar is 100 µm)


to posterior and no cross-links were detected in the posterior third of the corneas. The endothel cells are not involved in the cross-linking process. 4. Discussion Inter- and intramolecular cross-linking is essential for many biological processes. Cross-linking of fibrin in the formation of blood clots and cross-linking of collagen in tendons for increased tensile strength are examples. Collagen is shown to contain several different lysyl oxidase-related (enzymatic) and lysinderived (nonenzymatic) covalent cross-links that have extensively been investigated for two decades (Eyre, 1984 ; Reiser, 1991 ; Yamauchi et al., 1996). Nonenzymatic cross-linking of collagen, also called glycation, is age-related (Reiser, 1991) and there is strong evidence that the total cross-link content of collagen is increased in diabetes (Andreassen, Oxlund and Danielsen, 1988 ; Williamson et al., 1986). Kent, Light and Barley (1985) found that mechanical strength increased and solubility decreased in rat tail incubated in glucose and they attributed this to formation of glucose-derived cross-links. The microscopic interpretation of such cross-links includes intrafibrillar and interfibrillar covalent bonds. The concept of interfibrillar bonds may apply to tendon where collagen fibrils are interwoven, however, this may be less relevant to the corneal stroma where most of the collagen fibrils are parallel with a relatively constant distance of 20 to 40 nanometers between them. Hadley and coworkers found that sugar-induced cross-links do not appear to affect the swelling behaviour of corneal tissue, which is in agreement with the lack of interfibrillar cross-links (Hadley et al., 1996). The effect of aging of the human corneal stroma was investigated by Malik and coworkers using high and low angle X-ray diffraction (Malik et al., 1992). From birth to 90 years, the cross-sectional area of the collagen molecule increased by 7 % and the interfibrillar spacing decreased also by 7 %. Both effects add to a significant overall decrease in the extrafibrillar space which corresponds to the observation of Scott et al. of an age-related decrease in the ratio of proteoglycans to collagen (Scott, 1985). These observations indicate the presence intrafibrillar cross-links rather than interfibrillar bonds. We do not know whether alternative cross-linking techniques (UV-radiation with and without photosensitizers, aldehydes) may induce interfibrillar bonds between collagen molecules but proteoglycans may also participate in cross-linking, since the protein cores possess tryptophan residues. To our knowledge, there is no information in the literature about protein crosslinks other than between collagen molecules. Swelling experiments on cross-linked cornea may provide further insight in such processes. In this study we were able to demonstrate that the mechanical stiffness of the cornea can artificially be

101

increased either by aldehydes or by UV-radiation with photosensitizers. We failed to induce mechanical changes with UV-radiation alone, which may be explained by the small penetration depth of UV (254 nm) in the cornea. Longer wavelengths with better penetration, however, have less potential to cross-link and make it necessary to use photosensitizers. Riboflavin is such a non-toxic photosensitizer (Dollery, 1991) which is also soluble in water and penetrates easily into the corneal stroma in the absence of epithelium. Previous investigations reported that the active species of oxygen generated by a riboflavin-sensitized photoreaction causes cross-links of bovine vitreous collagen in vitro (Hikichi, 1996). Glutaraldehyde, in low concentration (0±1–0±01 %) is the standard cross-linker used to stabilize collagenous biomaterials. Because of the slow penetration of glutaraldehyde alone it is often mixed with formaldehyde (Karnovsky’s solution) which penetrates faster into tissue specimen and is, therefore, much more practical. Several investigators have also examined the biologic response to glutaraldehyde cross-linked collagen biomaterials in various animal models, as well as cell cultures (McPherson, Sawamura and Armstrong, 1986 ; Nimni, 1988 ; Petite et al., 1995 ; Remacle, 1995). These experiments on non-ocular tissue demonstrated, that at this low concentration and with a short treatment time, cytotoxic and immunological effects, if any, are relatively small. Grafts like porcine heart valve tissue or dermal collagen were repopulated with cells when crosslinked with low concentrations of glutaraldehyde. Since the side-effects of glutaraldehyde and formaldehyde may be different in cornea and the anterior eye segment compared to such grafts, further investigation of such reactions living ocular tissue seems mandatory. Preliminary observations in 4 living rabbit eyes treated topically with 0±1 % Karnovsky’s solution for 5 minutes, however, produced only a transient subepithelial opacity, which disappeared after three weeks. Application of 0±01 % glutaraldehyde does not produce any opacity in rabbit corneas. Regarding sideeffects as well as the biomechanical change a doseresponse has to be established which will have to be done in primate eyes rather than in rabbit eyes because of the increased regeneration potential in rabbit eyes. Aqueous solutions of commercial glutaraldehyde contains significant amounts of oligomer forms which are considered to be more important for induction of cross-links than the monomer (Nimni, 1988). Therefore, after application of 0±1 % glutaraldehyde to the surface of cornea interfibrillar or interlamellar crosslinks may become possible. The induction of cross-links in collagen by low concentrations of glutaraldehyde results not only in biomechanical changes but makes the collagen also more resistant to collagenases (Nimni, 1988 ; Remacle, 1995). This property could also be used for the

102

treatment of keratoconic cornea to prevent the degradation process caused by proteolytic activity (Daxer and Fratzl, 1997 ; Radner et al., 1996). We have used this effect clinically in three cases of corneal melting in human cornea, that were refractory to conventional treatment. In all three cases the melting process was halted after the application of 0±1 % Karnovsky’s solution (soaked in blotting-paper) for 5 minutes. Again, these treated human corneas remained clear. We would emphasize that the in vitro-experiments reported here are intended to the first step in the direction of a conservative treatment of keratectasia. It is not clear how long the induced cross-links, and the increase in biomechanical strength, may persist in living tissue. Also, potential toxic side-effects of such a cross-linking treatment on the anterior segment of the eye must be studied systematically.

Acknowledgements This work was supported by the Brunenbusch-SteinStiftung, Stuttgart, Germany.

References Albon, J., Karwatowski, W. S., Avery, N., Easty, D. L. and Duance, V. (1995). Changes in the collagenous matrix of the aging human lamina cribrosa. Br. J. Ophthalmol. 79, 368–75. Andreassen, T. T., Simonsen, A. H. and Oxlund, H. (1980). Biomechanical properties of keratoconus and normal corneas. Exp. Eye Res. 31, 435–41. Andreassen, T. T., Oxlund, H. and Danielsen, C. C. (1988). The influence of nonenzymatic glycosylation and formation of fluorescent reactionproducts on the mechanical properties of rat tail tendon. Connect. Tissue. Res. 17, 1–9. Balisky, L., Lee, C. H., Greenwald, D. P. and Rowsey, J. J. (1996). Tensile strength as a function of age in human corneal tissue. Invest. Ophthalmol. Vis. Sci. 37, S3201. Daxer, A. and Fratzl, P. (1997). Collagen fibril orientation in the human corneal stroma and its implication in keratoconus. Invest. Ophthalmol. Vis. Sci. 38, 121–9. Dollery, C. (1991). Therapeutic Drugs. Vol. 2, Pp. 21–23. Churchill Livingstone : Edinburgh. Edmund, C. (1988). Corneal elasticity and ocular rigidity in normal and keratoconic eyes. Acta Ophthalmologica 66, 134–40. Ersek, R. A. and Derlerm, A. G. (1988). Processed irradiated bovine cartilage for nasal reconstruction. Annals Plastic Surgery 20, 540–6. Eyre, D. R. (1984). Cross-linking in collagen and elastin. Ann. Rev. Biochem. 53, 717–48. Hadley, J. C., Meek, K. M. and Malik, N. S. (1996). The effect of glycation on charge distribution and swelling behaviour of corneal and scleral collagen. Invest. Ophthalmol. Vis. Sci. 37, S1010. Harner, C. H., McCarey, B. E., Rao, P. R. and Chow, A. A. (1996). In vivo evaluation of biomaterials for synthetic epikeratoplasty. Invest. Ophthalmol. Vis. Sci. 37, S68. Hikichi, T., Ueno, N., Chakrabarti, B., Tremp, C. L. and Yoshida, A. (1996). Evidence of cross-link formation of vitreous collagen during experimental ocular inflam-


mation. Graefe’s Arch. Clin. Exp. Ophthalmol. 234, 47–54. Kent, M. J. C., Light, N. D. and Bailey, A. J. (1985). Evidence for glucose-mediated covalent cross-linking of collagen after glycosylation in vitro. Biochem. J. 225, 745–52. Malik, N. S., Moss, S. J., Ahmed, N., Furth, A. J., Wall, R. S. and Meek, K. M. (1992). Ageing of the human corneal stroma : structural and biochemical changes. Biochim. Biophys. Acta 1138, 222–8. Mark, H. S. (1988). Cross-linking with radiation. Encyclopedia of Polymer Science and Engineering 2nd Ed ; Vol. 4, Pp. 418–49. Academic Press : London. McNamara, N. A., Brand, R. J., Bourne, W. M., Polse, K. A. and Hodge, D. O. (1996). Hyperglycemic effects on corneal function. Invest. Ophthalmol. Vis. Sci. 37, S315. McPherson, J. M., Sawamura, S., Armstrong, R. (1986). An examination of the biological response to injectable, glutaraldehyde cross-linked collagen implants. J. Biomed. Mat. Res. 20, 93–107. Milne, P. J. and Zika, R. G. (1992). Crosslinking of collagen gels : photo-chemical measurement. SPIE Ophthalmic Technologies II 1644, 115–24. Nash, I. S., Greene, P. R. and Forster, C. S. (1982). Comparison of mechanical properties of keratoconus and normal corneas. Exp. Eye Res. 35, 413–23. Nimni, M. E. (1988). The cross-linking and structure modification of the collagen matrix in the design of cardiovascular prothesis. J. Cardiac. Surg. 3, 523–33. Perkins, R., Bui, H. T. (1996). Tympanic membrane reconstruction using formaldehyde-formed autogenous temoralis fascia : twenty years’ experience. Otolaryngol. Head Neck Surg. 114, 366–79. Petite, H., Duval, J. L., Frei, V., Malak, N. A., Luizard, M. S. and Herbage, D. (1995). Cytocompatibility of calf pericardium treated by glutaraldehyde and by the acyl azide methods in an organotypic culture model. Biomaterials 16, 1003–8. Pietrucha, K. (1991). New collagen implant as dural substitute. Biomaterials 12, 320–3. Radner, W., Zehetmayer, M., Skorpik, C., Mallinger, R. (1996). Zur Anordnung der kollagenen Lamellen beim Keratokonus. Spektrum Augenheilkd. 10, 156–60. Reiser, K. M. (1991). Nonenzymatic glycation of collagen in aging and diabetes. Proc. Soc. Exp. Biol. Med. 196, 17–29. Remacle, M. (1995). Treatment of vocal fold immobility by glutaraldehyde-cross-linked collagen injection : longterm results. Ann. Otol. Rhinol. Laryngol. 104, 437–44. Sady, C., Khosrof, S. and Nagaraj, R. (1995). Advanced Maillard reaction and crosslinking of corneal collagen in diabetes. Biochem. Biophys. Res. Commun. 214, 793–97. Scott, J. E. (1985). Proteoglycan histochemistry—a valuable tool for connective tissue biochemists. Coll. Relat. Res. 5, 541–8. Seiler, T., Spoerl, E., Huhle, M. and Kamouna, A. (1996). Conservative therapy of keratoconus by enhancement of collagen cross-links. Invest. Ophthalmol. Vis. Sci. 37, S1017. Spoerl, E., Genth, U., Schmalfuß, K. and Seiler, T. (1996). Thermo-mechanisches Verhalten der Hornhaut. Klin. Monatsbl. Augenheilkd. 208, 112–16. Stolwijk, T. R., Van Best, J. A., Oosterhuis, J. A. and Swart, W. (1992). Corneal autofluorescence : an indicator of diabetic retinopathy. Invest. Ophthalmol. Vis. Sci. 32, 92–7. Uma, L., Sharma, Y. and Balasubramanian, D. (1996). Fluorescence properties of isolated intact normal human corneas. Photochem. Photobiol. 63, 213–16. Williamson, J. R., Chang, K., Kilo, C. and Lacy, P. E. (1986). Islet transplants in diabetic Lewis rats prevent and


reverse diabetes-induced increases in vascular permeability and prevent but do not reverse collagen solubility changes. Diabetologia 29, 392–6. Wisnewski, P. J., Power, D. L. and Kennedy, J. M. (1988). Glutaraldehyde-cross-linked meniscal allografts : mechanical properties. J. Invest. Surg. 1, 259–66. Wollensak, J. and Buddecke, E. (1990). Biochemical studies on human corneal proteoclycans—a comparison of normal and keratoconic eyes. Graefe’s Arch. Clin. Exp. Ophthalmol. 228, 517–23.

103

Wollensak, J., Ihme, A. and Seiler, T. (1987). Neue Befunde bei Keratokonus. Fortschr. Ophthalmol. 84, 28–32. Yamauchi, M., Chandler, G. S., Tanzawa, H. and Katz, E. P. (1996). Cross-linking and the molecular patching of corneal collagen. Biochem. Biophys. Res. Commun. 219, 311–15. Zhao, H. R., Nagaraj, R. H., Abraham, E. C. (1997). The role of α- and ε-amino groups in the glycation-mediated cross-linking of γB-crystallin. J. Biol. Chem. 272, 14465–9.