Collagen cross-linking in corneal scar formation

Download Collagen cross-linking in corneal scar formation

Post on 25-Aug-2016




6 download


  • Biochimica et Biophysica Acta, 412 (1975) 18-25 Elsevier Scientific Publishing Company, Amsterdam- Printed in The Netherlands

    BBA 37175


    DONALD J. CANNON* and CHARLES CINTRON Department of Fine Structure, Boston Biomedical Research Institute, Department of Cornea Research, Retina Foundation, Boston, Mass. 02114 (U.S.A.)

    (Received May 27th, 1975)


    The collagen of rabbit corneal scar tissue contains a pattern of reducible radioactive cross-links marked by a high content of glycosylated dihydroxylysino- norleucine. The presence of the glycosylated aldimine cross-link may reflect differ- ences in the distribution of carbohydrate in scar collagen. These differences may be relevant to the regulation of fibril diameter which may in turn be a causative factor in the loss of corneal transparency.


    The major macromolecular constituent of corneal stroma and corneal scar tissue is the protein, collagen. The collagen of normal mammalian cornea is contained in fibrils which are arranged in layers with each layer consisting of fibrils of uniform diameter [1]. In general, fibrils are constructed from the aggregation of individual collagen molecules that are secreted by connective tissue cells as procollagen [2]. Procollagen is a biosynthetic precursor of collagen consisting of three ct chains with additional amino-terminal extensions (pro-ct chains) [3]. A number of functions have been postulated for procollagen including involvement in the homeostatic control of fiber formation [3, 4]. Davison [4] has recently reviewed possible mechanisms for the control of extracellular fiber formation.

    Cross-linking in the collagen molecule occurs extracellularly either before or during fiber formation. Lysine and hydroxylysine residues in the collagen chains are oxidized to their respective aldehydes [5]. These aldehydes either through aldol con- densation or Schiff-base formation or both are converted to cross-links between the polypeptide chains of the protein [5].

    In certain penetrating corneal wounds healing results in scar tissue formation which causes varying degrees of vision loss. Benedek [6] has suggested that the loss in transparency in the corneal scars can be due to the presence of collagen fibrils with abnormal diameters. This suggests a possible loss of regulation of collagen fibrillo- genesis in healing scar. To gain a clearer understanding of the basic processes oper-

    * To whom reprint requests should be addressed.

  • 19

    ative in the formation of corneal scars the present study examines the collagen mole- cule and its lysine-derived cross-links from experimentally induced corneal scar tissue.


    Collagen. Corneal tissue from normal and scarred eyes was obtained from adult albino rabbits. The procedure of Cintron et al. [7] for corneal scar production was used without modification. Corneas were also obtained from human fetal eyes at autopsy and from fetal and neonate rabbits. Whole corneas were minced and either used directly for reduction or suspended in 0.1 M acetic acid containing pepsin (5 70, w/w, Worthington, twice crystallized) and digested for 24 h at 20 C on a rotary shaker. The soluble collagen was harvested by centrifugation at 35 000 g and the pH adjusted to 8.0 with 2 M Tris to inactivate pepsin. After 3 h at 4 C the pH was readjusted to 3.5 with acetic acid and collagen was precipitated by the addition of NaC1 to a concentration of 8 70, harvested and redissolved in 0.05 M acetic acid. Precipitation of collagen was repeated and the protein redissolved in acetic acid and dialyzed exhaustively versus the same buffer.

    Determination of cross-links. Reduction of native corneal collagen and corneal scar tissue with NaBaH4/NaBH4 was performed as described by Davison et al. [8]. Aliquots of reduced samples were hydrolyzed in both 6 M HC1 and 2 M NaOH for 24 h at 105 C. Amino acid and radioactivity elution patterns were obtained from chromatography on columns of Aminex A5 resin (Bio-Rad Laboratories)[8]. In certain experiments the column effluent stream was split permitting both ninhydrin assay using the Technicon system (Technicon Instruments Co.) and the collection of fractions. The identification of radioactive peaks was confirmed by matching authentic labelled samples previously identified from collagen [8].

    Electrophoresis. The a,fl, and higher molecular weight polymers of soluble collagens were separated by the polyacrylamide gel method of Laemmli [9] using a 9 70 cross-linked gel in a slab apparatus. Gels were stained with Coomassie blue and after destaining were scanned in a Joyce-Loebl densitometer [10].

    Labelling. Rabbits previously scarred [7] received either [U-~4C]lysine or [3H]lysine (New England Nuclear) administered by injecting a 0.2 ml solution of saline containing 10#Ci, 14C or 20#Ci aH into the anterior chamber of the eye of anesthetized rabbits immediately after the removal of an equal volume of aqueous humor. This procedure was performed on days 3, 6, 9 and 12 post-wounding. The animals were sacrificed on day 14, and the scar tissue carefully removed.

    Chromatography. Pooled radioactive material from ion-exchange chromato- graphy was desalted on a Bio-Gel P-2 column, 3 55 cm, using 0.! M ammonium formate, pH 7.0, as eluant.


    Identification of cross-links Results of the acid hydrolysis of native collagen and scar collagen of 2- and

    24-week duration are presented in Table I. The major peak in reduced native rabbit cornea following acid hydrolysis is histidinohydroxymerodesmosine with prominent peaks at hydroxylysinonorleucine and dihydroxylysinonorleucine. Dihydroxylysino-

  • 20



    The results are expressed as the percentage of total eluted radioactivity.

    Compound Native Scar Scar 2-week 24-week

    Dihydroxylysinonorleucine 19 45 34 Hydroxylysinonorleucine 31 29 34 Histidinohydroxymerodesmosine 41 21 26

    norleucine is the major peak in 2-week-old scar tissue collagen following acid hydro- lysis and is still significant in 24-week scar tissue although decreased relative to the 2-week sample. The content of hydroxylysinonorleucine does not vary greatly between normal and scar tissue, however, the content of histidinohydroxymerodes- mosine is markedly reduced in scar samples.

    The elution pattern of radioactive compounds from alkaline hydrolysates of normal and scar corneal tissue is shown in Fig. 1 and summarized in Table II. In

    (a) (b)

    2000- ~




    - ooo.,

    40 40 60 2b


    1 ; ;

    4 ~'~ lo

    56 11


    Fig. 1. Distribution of tritium in native (a) and scar (b) corneal tissue. Hydrolysis was performed in 2 M NaOH. Elution positions are numbered as follows: l, dihydroxynorleucine; 2, hydroxynor- leucine; 3, pre-aldol component; 4, aldol condensation product; 5, 6, unknown; 7, dihydroxylysinor- leucine; 8, hydroxylysinonorleucine; 9, lysinonorleucine; 10, histidinohydroxymerodesmosine; II, unknown.

    addition to the compounds observed in acid hydrolysis (Table I) alkaline hydrolysis reveals the presence of the reduced aldol condensation product (aldol), a peak eluting just prior to aldol (pre-aldol) and a persistent double peak of radioactivity eluting after histidinohydroxymerodesmosine (Fig. l) [8, 11 ].

    Quantitative comparison of native and 2-week scar tissue revealed an approx. 6-fold increase in the level of the pre-aldol component in the scar tissue. After 24 weeks the pre-aldol component, though decreased, was still a significant peak. The

  • 21



    The results are expressed as the percentage of total eluted radioactivity.

    Compound Native Scar Scar 2-week 24-week

    Pre-aldol 4 25 15 Reduced aldol condensation product 16 11 17 Dihydroxylysinonorleucine 15 17 11 Hydroxylysinonorleucine 15 8 10 Histidinohydroxymerodesmosine 22 14 16

    changes in the remaining reducible cross-links are complex and attention was focused on the identification of the pre-aldol component.

    Fig, 2 shows the elution pattern of the desalted pre-aldol peak from rabbit corneal scar tissue and the products of mild and complete acid hydrolysis. The con- version of the pre-aldol peak to dihydroxylysinonorleucine via an intermediate com- pound, which elutes in our system following the aldol compound, is identical to the results of Bailey et al. [12] in which galactosyldihydroxylysinonorleucine is an inter- mediate in the acid cleavage of glucosylgalactosyl-dihydroxylysinonorleucine to di- hydroxylysinonorleucine. In addition, pre-aldol, isolated from native rabbit fetal cornea, is cleaved by acid in an identical manner to the pre-aldol of scar tissue col-




    , -





    Fig. 2. Elution profile of the pre-aldol component from alkaline hydrolysis of scar tissue before and after acid hydrolysis. Rechromatography of isolated desalted pre-aldol (a), after 0.2 M HC1 hydro- lysis (b), after 2 M HC1 hydrolysis (c), (I is pre-aldol; II tentatively identified as galactosyl dihy- droxylysinonorleucine and III is dihydroxylysinonorleucine).

  • 22

    lagen. We have found that following alkaline hydrolysis the major reduced tritiated component of both human and rabbit fetal corneal collagen is the pre-aldol peak. In fetal rabbit cornea pre-aldol accounts for 40-42 ~ of the total eluted radioactivity, 20 ~o in the one week neonate and 10 ~ at one month of age.

    The amino acid compositions of pepsin solubilized scar and native corneal tissue were consistent with the composition of previously published corneal collagens. Differences between scar and normal stroma were confined to the contents of iso- leucine, leucine and tyrosine.

    The biosynthesis of the pre-aldol was examined over the initial 2 weeks of wound healing by repeated injections of radioactive lysine as described in Materials and Methods. During the first 2 weeks of wound formation lysine is incorporated into scar collagen (Fig. 3). Furthermore both ~4C and 3H counts were observed in the elution positions of the reduced aldehydes, dihydroxy- and hydroxynorleucine, pre- aldol, aldol, hydroxylysine and one unidentified component.


    Z W


    Q m" fl.,~

    ~b ~ 2b 3b 4b FRACTION NO.

    - r


    Fig. 3. Elution profile of tritiated, reduced 2-week scar collagen following alkaline hydrolysis. Scar formation proceeded in the presence of [14C]lysine as described in Materials and Methods. Shaded areas represent elution of 14C counts.

    The radioactivity eluting in the position of hydroxylysine has been attributed mainly to this amino acid rather than the coincident hydroxylysinonorleucine because the hydroxylation of lysine precedes the formation of the cross-link. The presence of counts in the reduced aldehydes does not exclude the possible existence of ~4C- labelled hydroxylysinonorleucine. The presence of counts in the scar collagen com- pounds are consistent with the known post-translational modifications of lysine in collagen, hydroxylation, oxidative deamination and glycosylation.

  • 23

    The electrophoresis of pepsin-solubilized native and scar tissue collagen revealed no differences in the migration positions of the a and/3 chains (Fig. 4). The a and fl positions were identical with those from tendon collagen and distinct from cartilage collagen (not shown).

    co NEALScAR


    Fig;. 4. Densitometric tracing of polyacrylamide gel electrophoresis of pepsin-soluble native rabbit corneal collagen and pepsin-soluble corneal scar collagen.


    Collagen, synthesized following a trephination excision wound in rabbit cornea, reveals major differences in the pattern of lysine-derived reducible cross-links when compared with unwounded rabbit corneal stroma. The predominance of dihydroxy- lysinonorleucine in acid hydrolysates was a preliminary indication that a major bio- synthetic pathway of scar collagen cross-links was directed toward the formation of an aldimine derived from two hydroxylysine residues; whereas in normal stroma, lysine, hydroxylysine and histidine residues contribute to the major reducible cross- link histidinohydroxymerodesmosine. Forrest et al. [13] have observed similar changes in the cross-link pattern of normal and wounded guinea pig dermis following acid hydrolysis of insoluble collagen.

    Further evidence of the unique character of the cross-links of scar collagen was only obtained following alkaline hydrolysis. The appearance of radioactive com- pounds, which do not survive acid hydrolysis [8] and the appearance of a new peak eluting just prior to the aldol condensation product, emphasize the importance of examing both acid and alkaline hydrolysates.

    Bailey and co-workers [12, 14] have demonstrated that the cross-link glucosyl- galactosyl-dihydroxylysinonorleucine is present in alkaline hydrolysates of granuloma tissue collagen, and in the collagen of bone, cartilage, and fetal and adult rat skin. This cross-link is also present in the body wall of the sea cucumber [15].

    Present results are consistent with the existence of this compound in corneal scar collagen based on its lability to acid hydrolysis and conversion to dihydroxylysi- nonorleucine. Incorporation studies indicated that the pre-aldol was lysine derived. Furthermore this cross-link was found to be a major component in both human and rabbit fetal corneas.

    Our results are consistent with the suggestion that in scar collagen there is a

  • 24

    decrease in the level of hydroxylysine glycosides relative to normal stroma [16]. If we assume that reduction introduces one tritium atom into the pre-aldol component and knowing the specific activity of NaB3H4 and the amount of collagen analyzed there is 0.1-0.2 residue per 1000 residues of pre-aldol in 2-week-old scar tissue. This amount, however, would not account for the total decreased levels of hydroxylysine glycosides seen in scar tissue. The collagen of corneal scar tissue differs from that of normal stroma in both the quantity of hydroxylysine glycosides and in the pattern of reducible cross-links. Whether these differences reflect different gene products or differences in glycosylation and hydroxylation is not known.

    Some observations relative to this point can be made. The collagen chain composition in certain embryonic tissues has been shown to consist of both the (al(I)2a2) and al(III)3 types [17]. The similarity of the cross-linking pattern in embryonic and wound tissue suggested the possibility that different collagens were being synthesized in scar tissue. However, gel electrophoresis of scar collagen under conditions which can reveal heterogeneity in a chain composition indicated only al and a2 chains and an al/ct2 ratio of 2.3. There were differences in the amino acid content of native and scar collagen so the possibility of collagen chain heterogeneity must await the isolation of sufficient material to examine other separation methods including carboxymethylcellulose chromatography. This work is currently in progress.

    During the time-course of wound healing there were shifts in the proportions of the aldimine cross-links hydroxy- and dihydroxylysinonorleucine (Table II) and of the aldol and pre-aldol compounds (Table I). Robins et al. [18] have demonstrated similar shifts in cross-link distribution in the maturation of bovine collagen. Although these changes are in the direction of the normal cross-linking pattern of native col- lagen, after 24 weeks both gross scar tissue and a reIatively high level of the glycosyl- ated aldimine cross-link were present. This would seem to indicate a slow rate of collagen turnover and replacement and that the remodelling of corneal scar collagen is not controlled in the same manner as in embryonic collagen. The cells synthesizing collagen in these tissues may be responsible for this difference. The cellular changes seen in corneal wound healing [19] differ from the cellular events in corneal embryo- genesis [17]. Despite certain cross-linking similarities of fetal and scar corneal col- lagen [20] it is probable that corneal scar collagen is not an embryonic-type collagen in view of the marked differences in the overall metabolism and cellular history of the two tissues.

    Comparison of the cross-links in scar versus fetal tissue may be relevant to fiber formation. A pattern of increased levels of lysine hydroxylation [21], increased formation of dehydro-dihydroxylysinonorleucine and glucosylgalactosyl disachar- ride [14] in certain fetal versus mature collagens has been established and may be important in the initial fibrillogenesis of collagen matrices.

    In cornea scars, collagen fibrils have large and variable cross-sectional di- ameters [22]. Furthermore, evidence has been presented demonstrating an inverse relationship between the carbohydrate content and the cross-sectional diameter in collagen fibrils [23]. A decrease in the glycosylation of hydroxylysine in scar versus normal corneal tissue [16] is consistent with the ultrastructural changes of the collagen fibrils and the increase observed here in the reducible cross-links dihydroxylysinonor- leucine and its glucosylgalactosyl disacharride may be a reflection of the changes in the carbohydrate content. However, until the manner in which collagen is synthesized

  • 25

    and remodelled in the corneal scar is understood the relationship between the cross- links of collagen and the development of the scar matrix remains unknown. Present studies are directed toward this goal.


    This research was supported by funds from the National Institutes of Health (HD-05970, EY-01199 and EY-00208) and in part by the Charles A. King Trust, and the Massachusetts Lions Eye Research Fund, Inc.

    The authors wish to thank Ellie Carlson and Claire Kubl in for their excellent technical assistance.


    1 Trelstad, R. L. and Coulombre, A. J. (1971) J. Cell Biol. 50, 840-858 2 Bellamy, G. and Bornstein, P. (1971) Proc. Natl. Acad. Sci. U.S.A. 68, 1138-1142 3 Bornstein, P. (1974) Annu. Rev. Biochem. 43, 567-603 4 Davison, P. F. (1973) C.R.C. Crit. Rev. Biochem. 1,201-245 5 Gallop, P. M., Blumenfeld, O. Oo and Seifter, S. (1972) Annu. Rev. Biochem. 41,617-672 6 Benedek, G. B. (1971) Appl. Opt. 10, 459--473 7 Cintron, C., Schneider, H. and Kublin, C. (1973) Exp. Eye Res. 17, 251-259 8 Davison, P. F., Cannon, D. J. and Andersson, L. P. (1972) Connect. Tissue Res. 1,205-216 9 Laemmli, U. K. (1970) Nature 277, 680

    10 Davison, P. F. and Winslow, B. (1974) J. Neurobiol. 5, 119-133 11 Cannon, D. J. and Davison, P. F. (1973) Exp. Gerontol. 8, 51-62 12 Bailey, A. J., Bazin, S. and Delaunay, A. (1973) Biochim. Biophys. Acta 328, 383-390 13 Forrest, L., Shutlleworth, A., Jackson, D. S. and Mechanic, G. L. (1972) Biochem. Biophys.

    Res. Commun. 46, 1776-1781 14 Robins, S. P. and Bailey, A. J. (1974) FEBS Lett. 38, 334-336 15 Eyre, D. R. and Glimcher, M. J. (1973) Proc. Soc. Exp. Biol. Med. 144, 400-403 16 Cintron, C. (1974) Biochem. Biophys. Res. Commun. 60, 288-294 17 Trelstad, R. L. (1973) J. Histochem. Cytochem. 21, 521-528 18 Robins, S. P., Shimokomaki, M. and Bailey, A. J. (1973) Biochem. J. 131,771-780 19 Matsuda, H. and Smelser, G. K. (1973) Exp. Eye Res. 16, 427-442 20 Bailey, A. J. and Robins, S. D. (1972) FEBS Lett. 21. 330-334 21 Barnes, M. J., Constable, B. J., Morton, L. F. and Kodicek, E. (1971) Biochem. J. 125, 925-928 22 Jakus, M. (1962) Invest. Ophthalmol. 1,202-225 23 Schofield, J. D., Freeman, I. L. and Jackson, D. S. (1971) Biochem. J. 124, 467-473


View more >