13
Detection of specific collagen types in normal and keratoconus corneas David A. Newsome, Jean-Michel Foidart, John R. Hassell, Jay H. Krachmer, Merlyn M. Rodrigues, and Stephen I. Katz Keratoconus is a corneal disease of unknown cause that involves a progressive thinning and. scarring of the corneal connective tissue. We examined normal human and keratoconus cor- neas, including one healed penetrating keratoplasty specimen. Organ cell cultures of normal and keratoconus corneal specimens were labeled with radioactive proline and analyzed, by CM-cellulose chromatography and slab gel electrophoresis to determine collagen biosynthesis. Collagen types I and HI were synthesized in similar amounts by normal and keratoconus stromacytes in culture. Specifically purified antibodies were used to determine the distribution of collagen types in tissue sections by immunofluorescence. The distribution of collagen types I, III, and IV in keratoconus was also similar to that in normal corneas, except that scarred regions in keratoconus and at the host-graft: juncture were largely type HI. Immunofluorescent reaction of the antitype IV collagen antibodies with Bowman's layer, in particular, and Descemet's membrane in keratoconus specimens indicated extensive destruction. Basement membrane destruction may play an important role in this disease. Key words: keratoconus, corneal collagens, collagen types, immunofluorescence, normal and abnormal corneal extracellular matrix Ke. ;ratoconus is a slowly progressive, seri- ous corneal disorder of uncertain heritability and unknown cause. The clinical hallmark of the disease is the irregular, cone-shaped pro- trusion of the central cornea which results from weakening and thinning of the stromal elements. Although the disease has been re- ported to exhibit familial patterns and an au- From the Section on Retinal and Ocular Connective Tis- sue Diseases (D. N., J. H.) and Ocular Pathology (M. R.), Clinical Branch, National Eye Institute, the Laboratory of Developmental Biology and Anomalies, National Institute of Dental Research (J-M. F.), Der- matology Branch, National Cancer Institute (S. K.), National Institutes of Health, Department of Health, Education and Welfare, Bethesda, Md., and the De- partment of Ophthalmology, University of Iowa (J. K.). J-M. F. is a Fellow of the FNRS in Belgium. Submitted for publication July 5, 1979. Reprint requests: Dr. David Newsome, Building 10, Room 10 D-17, National Institutes of Health, Be- thesda, Md. 20205. tosomal recessive mode of inheritance has been postulated, 1 most cases appear to be sporadic. In other clinical reports, kera- toconus has been linked to a heritable disor- der of connective tissue, Ehlers-Danlos syn- drome. 2 " 4 Extensive biomicroscopic and pathological studies of keratoconus have so far failed to demonstrate conclusively whether the pri- mary site of the disease is the corneal stroma or the epithelium and its basement mem- brane. 5 Ectasia of the stroma likely involves either primary or secondary alterations in its major structural macromolecules or in their interactions with other matrix elements. Col- lagen, the most abundant stromal element (about 71% of dry weight) has naturally received much attention. Histopathological studies have documented stromal scarring and degradative changes in the stromal lamel- lae but have been of little help in elucidating pathogenetic mechanisms. 738 Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Detection of specific collagen types in normal and keratoconus

Embed Size (px)

Citation preview

Page 1: Detection of specific collagen types in normal and keratoconus

Detection of specific collagen types innormal and keratoconus corneas

David A. Newsome, Jean-Michel Foidart, John R. Hassell, Jay H. Krachmer,Merlyn M. Rodrigues, and Stephen I. Katz

Keratoconus is a corneal disease of unknown cause that involves a progressive thinning and.scarring of the corneal connective tissue. We examined normal human and keratoconus cor-neas, including one healed penetrating keratoplasty specimen. Organ cell cultures of normaland keratoconus corneal specimens were labeled with radioactive proline and analyzed, byCM-cellulose chromatography and slab gel electrophoresis to determine collagen biosynthesis.Collagen types I and HI were synthesized in similar amounts by normal and keratoconusstromacytes in culture. Specifically purified antibodies were used to determine the distributionof collagen types in tissue sections by immunofluorescence. The distribution of collagen types I,III, and IV in keratoconus was also similar to that in normal corneas, except that scarredregions in keratoconus and at the host-graft: juncture were largely type HI. Immunofluorescentreaction of the anti—type IV collagen antibodies with Bowman's layer, in particular, andDescemet's membrane in keratoconus specimens indicated extensive destruction. Basementmembrane destruction may play an important role in this disease.

Key words: keratoconus, corneal collagens, collagen types, immunofluorescence,normal and abnormal corneal extracellular matrix

Ke.;ratoconus is a slowly progressive, seri-ous corneal disorder of uncertain heritabilityand unknown cause. The clinical hallmark ofthe disease is the irregular, cone-shaped pro-trusion of the central cornea which resultsfrom weakening and thinning of the stromalelements. Although the disease has been re-ported to exhibit familial patterns and an au-

From the Section on Retinal and Ocular Connective Tis-sue Diseases (D. N., J. H.) and Ocular Pathology(M. R.), Clinical Branch, National Eye Institute, theLaboratory of Developmental Biology and Anomalies,National Institute of Dental Research (J-M. F.), Der-matology Branch, National Cancer Institute (S. K.),National Institutes of Health, Department of Health,Education and Welfare, Bethesda, Md., and the De-partment of Ophthalmology, University of Iowa(J. K.). J-M. F. is a Fellow of the FNRS in Belgium.

Submitted for publication July 5, 1979.Reprint requests: Dr. David Newsome, Building 10,

Room 10 D-17, National Institutes of Health, Be-thesda, Md. 20205.

tosomal recessive mode of inheritance hasbeen postulated,1 most cases appear to besporadic. In other clinical reports, kera-toconus has been linked to a heritable disor-der of connective tissue, Ehlers-Danlos syn-drome.2"4

Extensive biomicroscopic and pathologicalstudies of keratoconus have so far failed todemonstrate conclusively whether the pri-mary site of the disease is the corneal stromaor the epithelium and its basement mem-brane.5 Ectasia of the stroma likely involveseither primary or secondary alterations in itsmajor structural macromolecules or in theirinteractions with other matrix elements. Col-lagen, the most abundant stromal element(about 71% of dry weight) has naturallyreceived much attention. Histopathologicalstudies have documented stromal scarringand degradative changes in the stromal lamel-lae but have been of little help in elucidatingpathogenetic mechanisms.

738

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 2: Detection of specific collagen types in normal and keratoconus

Volume 20Number 6 Collagen types in normal and keratoconus corneas 739

Fig. 1. Fluorescence micrograph of a frozen section of normal cornea reacted with anti-typeIV collagen antibodies demonstrates a wide, discrete area of fluorescence in the epithelialbasement membrane area. (x250.)

In the present study we have used spe-cifically purified antibodies to characterizethe distribution of collagen types in normalhuman corneal and anterior scleral tissue aswell as in keratoplasty specimens from pa-tients with well-documented keratoconus,including one postmortem specimen from apatient who had successful perforating kera-toplasty for keratoconus. We also charac-terized biochemically the collagens synthe-sized by normal and keratoconus stromalcells and normal scleral cells. Our data dem-onstrate the cornea-specific distribution ofcollagen types as well as some alterations incollagenous components in keratoconus speci-mens.Materials and methods

Corneal specimens. Five normal and five kera-toconus corneas were studied. Normal tissue was

obtained at autopsy from individuals aged 7, 22, 28,35, and 56 years. Corneoscleral specimens wereembedded in OCT embedding medium (Ames) andthen frozen in liquid nitrogen. Frozen sections (4jum thick) were transferred to albumin-coated glassslides to serve as substrate for a modified directimmunofluorescent staining reaction. Keratoconusspecimens from patients in the age range 28 to 49years, all with typical clinical criteria for the dis-ease as documented by one of us (J. K.), wereplaced into cold M-K corneal storage medium atthe time of surgery or autopsy and embedded,frozen, and sectioned within 20 hr by the methodsdescribed above. The keratoconus specimens hadvariable amounts of obvious scarring; all had irregu-lar thinning and Fleisher pigmented rings.

Corneal stromal cell cultures were established aspreviously described6 from explants of pure stromaprepared under a dissecting stereomicroscope.Cell cultures were similarly started from cleanlydissected explants of sclera and conjunctival

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 3: Detection of specific collagen types in normal and keratoconus

740 Newsome et al. Invest. Ophthalmal. Vis. Sci.June 1981

Fig. 2. Fluorescence micrograph of a frozen section of keratoconus cornea reacted with anti-type IV collagen antibodies demonstrates a marked reduction in and discontinuity of theamount ol fluorescence in the epithelial basement membrane area as compared with that seenin the normal specimen in Fig. 1. (X500.)

stroma. Primary outgrowths of cells as well as cellsin the third serial passage forming confluent layerswere used. Cells were maintained in polystyreneculture dishes (Falcon) in Eagle's minimal essen-tial medium with 5% fetal calf serum, 100 u/m]penicillin, and 50 /ng/ml streptomycin, and ex-posed to 37°, 100% humidity, and a 95% air-5%CO2 atmosphere. Confluent cultures were ra-diolabeled by 24 hr incubation in serum-freeDulbecco-Vogt medium (NIH) with 100 ju.Ci/ml3H-proline (NET-323; New England Nuclear, Bos-ton), 100 jti.g/ml vitamin C, and 50 yu-g/ml/3-aininoproprionitrile fumarate as a lathyrogen.Medium and cells together with carrier collagenfrom skins of lathyritic rats were extracted in 0.5Macetic acid containing 1.0 mg/ml pepsin. The colla-gen was purified by NaCl precipitations (25%; twoprecipitations) in the presence of carrier collagenfrom skins of lathyritic rats and eluted from a car-boxymethyl (CM)-cellulose chromatography col-

umn (1.6 by 5 cm) as described elsewhere, whichpermitted an average (n = 4) recovery of 74% ofapplied counts.7"9 Peak fractions in the effluentwere pooled, lyophilized, and electrophoresed inthe presence of 0.05M urea on sodium dodecylsulfate (SDS)-polyacrylamide slab gels.10 Purifiedtypes I and III collagen standards were includedwith each electrophoresis. Reduced samples wereobtained by dividing the original 250 /xl samplesinto two and adding 5 ju.1 of/3-mercaptoethanol toone set of the samples. After electrophoresis thegels were stained with Coomassie blue and wereimpregnated with 2,5-diphenyloxazole (PPO) indimethylsulfoxide. Autofluororadiograms on Ko-dak X-Omat film were prepared by exposing thefilm at —74° for periods up to 7 days.

Preparation of antigens. Types I and 111 colla-gens were isolated from fetal calf skin,8 type IIfrom a rat chondrosarcoma," and type IV from amurine sarcoma.12 Type I procollagen was ex-

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 4: Detection of specific collagen types in normal and keratoconus

Volume 20Number 6 Collagen types in normal and keratoconus corneas 741

tracted from the skin of dermatosporactic calvesand was a gift from Dr. C. M. Lapiere. Type IIIprocollagen was purified from fetal calfskin.8 Theidentity and purity of the collagens was deter-mined by amino acid analysis after hydrolysis in6N HC1 at 110° for 18 hr; by CM-cellulose anddiethylaminoethyl-cellulose elution profiles, andby SDS-polyacrylamide slab gel electrophoresiswith and without /3-mercaptoethanol reductionand with and without cyanogen bromide diges-tion. Human fibronectin was purified from freshhuman plasma by affinity chromatography on agelatin column. The material eluted with 1M po-tassium bromide was electrophoresed after reduc-tion of the disulfide bonds with 20 mM dithio-treitol. The 220,000 dalton chain of fibronectinwas eluted from slices of the acrylamide gels andused to immunize rabbits.

Preparation of antibodies to collagens. Rabbitsor guinea pigs were immunized by repeated sub-cutaneous or foot pad injections of 0.5 to 5.0 mg ofthe antigen in Freund's complete adjuvant.Antibodies were isolated and purified by cross-immunoabsorption and checked for cross-reac-tivity to collagen and proteoglycans by indirectimmunofluorescence, micropassive hemaggluti-nation assay,13 and radioimmunoassay. Antibodiesto collagens were purified from the antisera byaffinity chromatography and cross-immunoab-sorption.H Serum from immunized animals waspassed sequentially over columns of types II andIII collagen and types IV and V collagens whichhad been covalently bound to Sepharose 4B. Theunbound material was then passed over a type Icollagen column which was extensively rinsedwith phosphate-buffered saline (PBS). Antibodieswere eluted from the column with 0.5N aceticacid-0.5M NaCl. This eluate was neutralized with3M Tris, dialyzed against PBS, and concentratedby filtration on a UM20 Amicon filter. Antibodiesto the other antigens were purified according to asimilar protocol.

Testing of antibodies to collagens. The spe-cificity of the antibodies was determined by radio-immunoassay (RIA), by immunofluorescence la-beling of target tissues, and by enzyme-linkedimmunoabsorbent assay (ELISA) as reported pre-viously.15 For RIA, types I and III collagens werelabeled with iodine-125 by the chloramine-Tmethod16 (sp. act. 1 to 2 fxCi/ixg). The RIA wasperformed according to the method of Rohde etal.17 In quantitative inhibition tests preincubationof the purified antibody with increasing amountsof the autologous collagen inhibited the binding ofthe antibody with the labeled antigen, whereas no

Table I. Distribution of ocular connectivetissue proteins in normal andkeratoconus specimens as detectedby immunofluorescence

Site Proteins detected

Cornea] epithelium

Cornea! epithelial basementmembrane

Cornea! stroma*

Descemet's membrane

Corneal scars

Conjunctival blood vesselsSclera

No reaction forcollagens/fibronectinType IV collagen

Type 1 collagen;fibronectinType IV collagen;fibronectinType III collagen;some type IType IV collagenType III collagen;some type I

*Type III collagen has been extracted from corneal stroma andcharacterized biochemically but was not reliably detected byimmunofluorescence in our studies.

significant inhibition was observed when the anti-body was preincubated with the heterologous col-lagen. Indirect immunofluorescence studies wereperformed on fresh-frozen sections of human skin,liver, tendon, and spleen. They established thatthe purified antibodies reacted only with thosestructures known to contain these molecules. Fi-nally, indirect immunofluorescence blocking andcompetition studies confirmed the specificity ofthe anti-type I and anticollagen antibodies, sincea preincubation with the specific antigen blockedthe binding of antibodies to known high-affinitysubstrates. For example, a preincubation withtype I collagen blocked the binding of anti-type Icollagen antibodies to human skin but did not in-terfere with the binding of anti-type III collagenantibody.

Preparation of antibodies to fibronectin.Specific antibody to fibronectin was purified byaffinity chromatography.H Serum from im-munized rabbits was passed over a column offibronectin which had been covalently bound toSepharose 4B. The unbound material was thenwashed away with PBS. Antibody to fibronectinwas then eluted from the column with 0.5M aceticacid, pH 2.9, neutralized with 3M Tris, dialyzedagainst PBS, and concentrated on a UM20 Amiconfilter.

Testing of antibody to fibronectin. The spec-ificity of the antibody was determined by RIA, byELISA,15 by radial immunodiflusion according toOuchterlony, and by immunoelectrophoresis. ForRIA, labeled fibronectin was purified from the cul-ture medium of human fibroblasts grown with

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 5: Detection of specific collagen types in normal and keratoconus

742 Newsome et al. Invest. Ophtlialmol. Vis. Sci.June 1981

Fig, 3. Inset of the superficial area of a scarred keratoconus cornea shows distruption (asterisk)of Bowman's layer (arrows) and superficial stromal fibrosis. E, epithelium; S, stroma. (To-luidine blue; X350.) The transmission electron micrograph demonstrates irregular subepithe-lial connective tissue excrescences in the basal epithelium (E). Bowman's layer is replaced byirregular connective tissue. F, fibroblast. (x27,000.)

[U-3H]leucine.l8 Fibronectin was also labeledwith l2oI by the chloraniine-T method.16 Preincu-bation of the antibody with other antigens, includ-ing types I and III collagens, laminin, and type IVcollagen, did not inhibit the binding of the anti-bodv to labeled fibronectin. Bv radial immu-

nodiffusion and immunoelectrophoresis a singleline of precipitation was obtained with an an-tifibronectin antibody and purified fibronectin ornormal human serum.

Immunofluorescence reaction. Modified directimmunofluorescent staining was performed on

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 6: Detection of specific collagen types in normal and keratoconus

Volume 20Number 6 Collagen types in normal and keratoconus corneas 743

Fig. 4. These sections of a healed perforating keratoplasty specimen from a patient withkeratoconus were reacted with anti-type I (A) and anti-type III (B) antibodies. Note thestrong fluorescence of the scar with anti-type III and the similarity of staining of the graft andhost stromas with both antibodies. (X300.)

sections of cornea and corneoscleral tissue withantibodies and control sera in two dilutions: 10 and20 /Ag/ml PBS. After a 30 min incubation the sec-tions were washed in PBS and then exposed tofluorescein-conjugated goat-anti-rabbit immuno-globulin for 30 min, washed again in PBS, andmounted with glycerine. Specimens were exam-ined with a Leitz ultraviolet microscope equippedwith a camera attachment and a BG 12 filter. Ex-posures were made with an automatic exposuremeter on Kodak Ektachrome 200 film.

Results

With the use of the techniques describedherein it was not generally possible to assessimmunofluorescent reactivity quantitatively.We were able to localize the heterogeneouspopulation of human corneal collagens tospecific sites, This distribution is summarizedin Table I.

Corneal epithelium. The epithelial layer it-self did not react with any antibodies. Thisfinding was similar in normal and keratoco-nus corneas. The zone just deep to the epi-thelium showed a strong reaction with an-tibodies to type IV collagen {Fig. 1). In thekeratoconus specimens with more advancedstromal thinning, the,reaction with anti-typeIV in the epithelial basement membrane

area was strikingly decreased and irregular(Fig- 2).

Bowman's layer. Bowman's layer reactedsimilarly with antibodies to types I and IIIcollagen as did the anterior stroma. How-ever, the positivity for anti-type I wasnoticeably reduced as compared with moreposterior layers of the stroma. Bowman'slayer reacted more strongly with antibody tofibronectin than did the stroma. Except forthe presence of type Ill-containing scars thatincluded part of the Bowman's area in somekeratoconus specimens, there was no differ-ence between the types of detectable colla-gens in the normal and keratoconus material.The histological appearance of a typical scarcan be seen in Fig. 3.

Corneal stroma. In both normal and kera-toconus tissue the stroma was strongly posi-tive for type I. Procollagen type I reactionwas consistently weaker in all specimens ofcornea. In some of the heavily scarredkeratoconus specimens there was a patchyincrease in detectable procollagen type I inthe subepithelial anterior stroma.

Type III collagen was not clearly detectedin the normal corneal stroma by fluorescencemethods. Consistent reactivity in normal tis-

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 7: Detection of specific collagen types in normal and keratoconus

744 Newsome et al.Invest. Ophthalmol. Vis. Sci.

June 1981

Fig. 5. Fluorescence micrographs (A, X450; B, X300) of a section of normal cornea reactedwith antifibronectin antibody exhibits fluorescent anterior and posterior portions of Des-cemet's membrane identical to that pattern seen with anti-type IV antibodies. Right panel,Transmission electron micrograph of normal tissue. The fluorescence pattern was seen in bothnormal and undisrupted keratoconus Descemet's membrane and may correspond to the nor-mal 100 nm wide-spaced collagen arrow zone (immiinoHuorescent dark interleaf) and anteriorand posterior granular basement membrane-like zones of material. S, stroma; DM, posteriorDescemet's membrane; E, endothelial cell. (x30,000,) Fibronectin is also detectable in stromaitself in the fluorescence micrographs.

sues was detected only on the edges ofstromal lamellae. No differences were visiblebetween the reaction for type III and procol-lagen type III in normal and keratoconus un-scarred stroma.

In the specimen of a successful graft ofnormal donor into a keratoconus corneal bed,both areas showed the same reaction withantibodies to types I and III. In the grafted

specimen the healed junction between donorkeratoconus recipient reacted strongly withantibodies to type III and procollagen typeIII. The zone of scarring extended into boththe host and donor tissues (Fig. 4). Type IIIcollagen could be clearly detected also in themore thinned and scarred keratoconus spec-imens as patches corresponding to stromal

scars.

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 8: Detection of specific collagen types in normal and keratoconus

Volume 20Number 6 Collagen types in normal and keratoconus corneas 745

Fig. 6. Fluorescence micrograph of a normal sclerocorneal section reacted with anti—type IIIantibodies reveals strong fluorescence in Tenon's and the sclera at the liinbal transition zone.Note the relative lack of reaction in the clear cornea (C). This strong reaction was evident in allareas of sclera studied in normals and keratoconus. (X450.)

Both normal and keratoconus corneal stro-mas exhibited high levels of detectablefibronectin. There was a particularly concen-trated positive reaction at the level of thestroma adjacent to Descemet's membrane andin the epithelial and Descemet's basementmembrane areas themselves (Fig. 5).

Descemet's membrane. Descemet's mem-brane reacted strongly with antibodies totype IV collagen, exhibiting a striking lami-nar pattern with bright fluorescence on thestromal and endothelial faces and a dark, ap-parently unstained interleaf (Fig. 5). Whenmeasured on a fluorescence micrograph, thebright leaves each were about 1 /am wide at4x, and the dark interleaf 2 /xm. These rela-tive widths were comparable to the threeanatomical zones discernible ultrastructural-ly: the anterior zone of granular basementmembrane-like material at the stromal inter-faces, the middle zone of wide-spaced colla-gen, and the posterior zone of basementmembrane-like material (Fig. 5). In kerato-conus specimens, the staining pattern of

Descemet's membrane was often irregular,and the laminar pattern less defined. Thisfinding appeared to be associated with thefrequent folds in Descemet's membrane nearthe thinned stroma.

Normal Descemet's membrane had de-finitely detectable fibronectin by the immu-nofluorescent stain. The distinctive laminarpattern was present, with more intense stain-ing on the stromal as compared with the endo-thelial surface. The interruption of the patternof fibronectin reaction in most of thekeratoconus specimens was in contrast to thepicture in the normal tissues (Fig. 5).

Sclera. Normal human sclera had a pre-ponderance of detectable type III collagenover type I collagen. This pattern, oppositeto that seen in the corneal stroma, was mostapparent at the scleral spur region, where thetwo tissues are juxtaposed (Fig. 6). Procolla-gen type I reaction was consistently weaker inall scleral specimens than that of type I itself.

Conjunctiva. Normal human conjunctivalstroma reacted strongly with antibodies to

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 9: Detection of specific collagen types in normal and keratoconus

746 Newsome et at.Invest. Ophthalmol. Vis. Sci.

June 1981

40,000

30,000

20,000

10,000 -

B50,000 r

40,000

30,000 •

20,000 •

10,000 •

12,500

10.000

7500

2500 -

20 30 40

TUBE NUMBER

Fig. 7. CM-cellulose chromatograms demonstratethe elution profiles of the radioactively labeled col-lagens synthesized by cultured corneal stroma-cytes from normal (A) and keratoconus (B) corneasand sclera (C). Arrows indicate the elution posi-tions of the carrier rat skin type I collagen peaks.Note the large peak containing the type III colla-gen in C and the similarities of the profiles in Aand B.

type I collagen. Walls of blood vessels in theconjunctiva had clearly detectable type IVcollagen and fibronectin. The conjunctivalepithelium itself did not react with any an-tibodies.

Identification of collagen synthesized invitro. Normal and keratoconus cornealstromal cells synthesized both types I and IIIcollagen in vitro. The presence of these twocollagens was confirmed by the elution posi-tion from CM-cellulose columns (Fig. 7) andSDS slab gel electrophoresis patterns. SDSgel profiles of the migration patterns ofpooled fractions from the three collagenouspeaks seen on CM-cellulose chromatographyshowed co-migration of peak 1 collagenchains with purified alpha 1 (I) collagen andpeak 3 with authentic alpha 2 collagen chains(Fig. 8). Peak 2 materials migrated largelywith the standard type III and exhibiteddisappearance of the trimeric form whenelectrophoresed under reducing conditions(Fig. 8).

The ratio of type I to type III collagen asdetermined from the CM-cellulose profileswas 10:1 for normal corneal tissue and 9:1for keratoconus tissues (average of five spec-imens). The preponderance of type III colla-gen in the sclera was demonstrated by a ratioof 2.1:1, type I to type III (average of fivespecimens). An example of this large amountof type III collagen synthesized by scleralfibroblasts in culture can be seen in Fig. 7.The material from peak 2 of the chromato-gram of the scleral collagens again showedreduction with beta-mercaptoethanol whensubjected to SDS gel electrophoresis.

Discussion

Corneal transparency depends upon theorderly deposition of collagen and glycocon-jugates in the stroma.19' 20 Maintenance oftransparency must depend, at least in part,on the controlled turnover of this extracel-lular matrix. Collagens, the most plentifulgroup of structural macromolecules in thecorneal stroma, are present largely as fibrillarproteins that interact with the stroma-specificproteoglycans to produce a normal opticallyuseful stroma. Thus modifications of the

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 10: Detection of specific collagen types in normal and keratoconus

Volume 20Number 6 Collagen types in normal and keratoconus corneas

Type Id

ai(l)a2

1 1R 2R 3R

Fig. 8. Fluororadioautograph of an SDS-polyacrylamide slab gel of the collagen chains elutedin the three peaks of the CM-cellulose chromatogram demonstrates the presence of the types Iand III collagen synthesized by cultured keratoconus stromaeytes. 1, Pooled fractions fromthe a-1 peak region; 2, pooled from the interpeak; 3, pooled from the a-2 peak. Note thedisappearance of the type III trimer after reduction with b- mercaptoethanol (2R). The posi-tion of the radiolabeled type III collagen was identical to that of coelectrophoresed authentictype III, as was that of the labeled type I. These patterns were similar between normal andkeratoconus cell synthetic products; the scleral samples had more prominent type III bands onthe gels.

types of collagens synthesized or defects inposttranslational processing of the moleculecould lead to corneal disease.

Recent advances in collagen characteriza-tion have revealed that there are at least fourgenetically distinct types which are now re-ferred to as types I, II, III, and IV.21 Thetissue-specific distribution of collagen typesappears to play an important role in mor-phogenesis, particularly in the cornea,22'23< 27

and in association with the sensory retina.24'25

Corneal stromal matrix stability may de-pend in part upon the proportion of the typesof collagen present. We have shown thatthere are two major collagen types synthe-sized in vitro by normal and keratoconus cor-nea cells, with a similar proportion of types Iand III in both. This mixture of types I andIII is similar to the collagens that accumulatein vivo and have been extracted and iden-tified.26 It is possible that an alteration in theproportion of these major collagen compo-nents could destabilize the connective tissue

matrix, as in Ehlers-Danlos type IV, wherethe synthesis of type III collagen is low toabsent.27 Maumenee28 reported that twokeratoconus corneal specimens contained in-creased type III collagen as detected by im-munofluoesence. In our specimens type IIIcollagen was not reliably detected in unin-volved corneal stroma. We did demonstratecollections of type III collagen in keratoconuscorneal scars and at the site of host-graftwound healing in a perforating keratoplastyspecimen. Type III collagen has previouslybeen reported in scars in nonocular tissues,even those such as tendon which do not nor-mally have type III.29

The presence of type III collagen in normalcornea, including human, has been contro-versial,30' 3l and type V collagen preliminarilyidentified.32 Type III had been identified inbovine material,33 and type V in lapine.32

Newsome et al.26 have recently submittedevidence from collagenase sensitivity, cyano-gen bromide peptides, trypsin sensitivity,

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 11: Detection of specific collagen types in normal and keratoconus

748 Newsome et al.Invest. Ophthahnol. Vis. Sci.

June 1981

and slab gel electrophoresis with and withoutreduction identifying types III and V in nor-mal human cornea. These data support theobservations of others noted above. Differ-ences with other reports30 may be due at leastpartially to technique, since type III is heav-ily crosslinked and difficult to extract fromnative tissues. It is more difficult to explainthe conflict of the present data from cul-tures,31 since extraction is not a problem invitro. We used different media with a shortlabeling period and early passage cells, incontrast to Stoesser et al.34 In the presentstudy preliminary results with anti-type Vantibodies revealed a positive reaction pat-tern identical to that for type IV in both nor-mal and keratoconus corneas. We have notemphasized these results, since our anti-bodies were not, at the time of this work,rigorously proved to be completely specific.Subsequent purification has, however, con-firmed the type V specificity (J-M. F., un-published data).

Our results emphasize some similaritiesand indicate certain differences between thedeveloping chick cornea and human tissue.Type I collagen was the major detectable col-lagen in our study of the human cornealstroma, a finding corresponding to what hasbeen reported for chick.35 Type III collagenis a small component of the human corneabut is reportedly absent in the chick. Thesynthesis of both types I and III by a uniformcell population has not been previously re-ported for corneal stromacytes but is knownto occur with smooth muscle cells36 andfibroblasts.29 We could not reliably detecttype III collagen in all normal stromas,perhaps due to the limits of the immu-nofluorescent reaction. Factors such as mask-ing of type III by other matrix componentsmay also have influenced its detectability.

Type II collagen is in the chick a promi-nent component of Bowman's layer, the sub-epithelial stroma, and Descemet's mem-brane.35 The presence of type II-bearingprimary corneal stroma, thought to be criticalto the morphogenesis of the adult stroma inthe chick, has not been confirmed in humantissue (for review, see ref. 37). We could notdemonstrate type II in the Bowman's or sub-

epithelial areas of either the normal orkeratoconus specimens.

The laminar staining pattern of Descemet'smembrane may indicate a regional distribu-tion of type IV collagen and of the noncol-lagenous fibronectin within the membrane.Such a specific distribution could explain, atleast in part, the distinct ultrastructure of thismembrane. Jakus38 and later Hay and Revel39

described the approximately 100 nm period-icity of the anterior third of Descemet'smembrane, presumably reflecting the pack-ing of membrane subunits in this area. Suchorderly packing could be in response to thepresence here of type IV fibrils or aggragates.The interface between the stroma andDescemet's membrane is also distinguishedultrastructurally. Patchy, basement mem-brane-like material has been reported hereby transmission electron microscopy. Thestrong reaction in this region with an-tifibronectin suggests that this material con-sists, at least in part, of fibronectin. The pre-cise localization of fibronectin within thebasement membrane-like material at thestromal-Descemet's interface must await ul-trastructural studies.

Type IV collagen was also prominently de-tected at the interface of the endothelium withDescemet's membrane. This observationoffers further evidence that this cell layer, soimportant to the maintenance of proper cor-neal hydration, synthesizes basement mem-brane.40' 41 The presence of fibronectin here inposterior Descemet's membrane is consistentwith its detection in a wide variety of humanbasement membranes.42 This protein is alsocommonly associated with and synthesized byfibroblasts (for review, see ref. 43). It is possi-ble that the stromacytes synthesize thefibronectin that accumulates in Descemet'smembrane. However, the corneal endothe-lium is embryologically derived from theneural crest, as are the stromacytes proper,44

and thus may share common attachment pro-teins. This observation contrasts with the ap-pearance of the epithelial basement mem-brane area which had no detectable fibronec-tin. And, of course, the epithelium is not aneural crest derivative.

The striking amount of type III collagen in

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 12: Detection of specific collagen types in normal and keratoconus

Volume 20Number 6 Collagen types in normal and keratoconus corneas 749

the sclera may in part explain its anatomicaldifferences from the type I-rich cornealstroma. The proportion of collagen types syn-thesized may affect fibril aggregation and uni-formity. In the sclera the matrix fibrils arerandomly arrayed in a feltwork with no signsof the orderly lamellae so characteristic of thecorneal stroma. Scleral fibril diameters aremuch more heterogenous, and the fibrilsmuch larger than those of the cornea. Lapieret al.45 have shown that type III collageninfluences the size and type of bundlesformed by type I collagen in vitro. Suchan interaction may condition the collagenbundle organization in various connectivetissues.

It is significant that the collagen profilesobtained from cultured corneal, scleral,and conjunctival stromacytes differed amongthemselves, and that these differences corre-sponded to differences in immunofluorescentreactivity. For example, the sclera reactedstrongly with anti-type III and synthesizedthe largest proportion of type III collagen invitro. The persistence of these differences inpatterns of synthesis of collagens, each typeof which is a distinct gene product, is in con-trast to the rapid decay of keratan sulfate pro-teoglycan synthesis by in vitro cornealstroma, even in organ culture.46

The present work has extended our knowl-edge of the site-specific distribution of thegenetically distinct collagen types in normalhuman and keratoconus corneas and normalsclera. It appears that a normal complementof corneal stromal collagens is present innormal proportions in keratoconus. Thus, ifkeratoconus is a true collagen disease (i.e.,has a defect in collagen on the molecularlevel), the defect probably occurs as a resultof an error in assembly or in the control ofturnover, for example, by the overproductionof collagenolytic enzymes. Further investi-gations of these possibilities are in progress.

REFERENCES1. Hammerstein W: Zur Genetik des Keratokonus. Al-

brecht Von Graefes Arch Klin Exp Ophthalmol190:293, 1974.

2. McKusick VA: Heritable Disorders of ConnectiveTissue, ed. 4. St. Louis, 1972, The C. V. Mosby Co.

3. Robertson I: Keratoconus and the Ehlers-Danlos

syndrome. A new aspect of keratoconus. Med J Aust1:571, 1975.

4. Kuming BS and Joffe L: Ehlers-Danlos syndromeassociated with keratoconus. A case report. S AfrMed J 52:403, 1977.

5. Pollack FM: Contributions of electronmicroscopy tothe study of corneal pathology. Surv Ophthalmol20:375, 1976.

6. Newsome DA, Takasugi M, Kenyon KR, Stark WF,and Opelz G: Human corneal cells in vitro: mor-phology and histocompatibility (HL-A) antigens ofpure cell populations. INVEST OPHTHALMOL 13:23,1974.

7. Chung E and Miller EJ: Collagen polymorphism:characterization of molecules with the composition[a l (III)]3 human tissues. Science 183:1200, 1974.

8. Epstein EH Jr: [a l (III)]3 human skin collagen: re-lease by pepsin digestion and preponderance in fetallife. J Biol Chem 249:3225, 1974.

9. Trelstad RL: Human aorta collagens: evidence forthree distinct species. Biochem Biophys Res Com-mun 57:717, 1974.

10. Lammli UK: Cleavage of structural protein duringthe assembly of the head of bacteriophage T4. Na-ture 227:680, 1970.

11. Smith BD, Martin EJ, Miller EJ, Dorfman A, andSwarm R: Nature of the collagen synthesized by atransplanted chondrosarcoma. Arch Biochem Bio-phys 166:181, 1975.

12. Orkin RW, Gehron P, McGoodwin EB, Martin GR,Valentine T, and Swarm R: A murine tumor produc-ing a matrix of basement membrane. J Exp Med145:204, 1977.

13. Andreopoulos NA, Mestecky J, Wright GP, and Mil-ler EJ: Characterization of antibodies to the nativehuman collagens and to their component a chains inthe sera and joint fluids of patients with rheumatoidarthritis. Immunochemistry 13:709, 1976.

14. Nowack H, Gay S, Wick G, Becker U, and Timpl R:Preparation and use in immunohistology of an-tibodies specific for types I and III collagen and pro-collagen. J Immunol Methods 12:117, 1977.

15. Rennard SI, Berg R, Martin GR, Foidart J-M, andGehron Robey P: Enzyme-linked immunoassay(ELISA) for connective tissue components. AnalBiochem (in press).

16. McConahey PJ and Dixon FJ: A method of traceiodination of proteins for immunological studies. IntArch Allergy 29:185, 1966.

17. Rohde H, Nowack H, Becker U, and Timpl R: Ra-dioimmunoassay for the aminoterminopeptide ofprocollagen p a l (I) chain. J Immunol Methods11:135, 1976.

18. Engvall E and Rouslahti E: Binding of a soluble formof fibroblast surface protein fibronectin to collagen.Int J Cancer 20:1, 1977.

19. Trelstad RL and Coulombre AJ: Morphogenesis ofthe collagenous stroma in the chick cornea. J CellBiol 50:840, 1971.

20. Maurice DM: The structure and transparency of thecornea. J Physiol 136:263, 1957.

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018

Page 13: Detection of specific collagen types in normal and keratoconus

750 Newsome et al.Invest. Ophthalmol. Vis. Sci.

June 1981

21. Martin GR, Beyers PH, and Piez KA: Procollagen.Adv Enzymol 42:167, 1975.

22. Dodson JW and Hay ED: Secretion of collagenousstroma by isolated epithelium grown in vitro. ExpCell Res 65:215, 1971.

23. Hay ED: Origin and role of collagen in the embryo.Am Zoologist 13:1085, 1973.

24. Newsome DA, Linsenmeyer RF, and Trelstad RL:Vitreous body collagen. Evidence for dual originfrom the neural retina and hyalocytes. J Cell Biol71:59, 1976.

25. Smith GN, Linsenmeyer TF, and Newsome DA:Synthesis of type II collagen in vitro by embryonicchick neural retina tissue. Proc Natl Acad Sci USA73:4420, 1976.

26. Newsome DA, Gross J, and Hassell JR: Humancorneal stroma contains three distinct collagens.(Manuscript submitted.)

27. Pope FM, Martin GR, Lichtenstein JR, PenttinenR, Gerson G, Rowe DR, and McKusick BA: Patientswith Ehlers-Danlos syndrome type IV lack type IIIcollagen. Proc Natl Acad Sci USA 72:1314, 1975.

28. Maumenee IH: The cornea in connective tissue dis-ease. Ophthalmology 85:1014, 1978.

29. Epstein EH Jr and Munderloh NH: Isolation andcharacterization of CNBr peptides of human [a l (III)]3

collagen and tissue distribution of[a l ( I ) ] 2a2and[al(III)]3 collagens. J Biol Chem 250:9304, 1975.

30. Freeman IL: Collagen polymorphism in maturerabbit cornea. INVEST OPHTHALMOL VIS SCI 17:171,1978.

31. Church RL: Procollagen and collagen produced bynormal bovine corneal stroma fibroblasts in cell cul-ture. INVEST OPHTHALMOL VIS SCI 19:192, 1980.

32. Yue B, Baum JL, and Smith BD: Collagen synthesisby cultures of stroma] cells from normal human andkeratoconus corneas. Biochem Biophys Res Com-mun 86:465, 1979.

33. Schmut O: The identification of type III collagen incalf and bovine cornea and sclera. Exp Eye Res25:505, 1977.

34. Stoesser TR, Church RL, and Brown SI: Partialcharacterization of human collagen and procollagen

secreted by human corneal stromal fibroblasts in cellculture. INVEST OPHTHALMOL Vis SCI 17:264, 1978.

35. von der Mark K, von der Mark H, Timpl R, andTrelstad RL: Immunofluorescent localization of col-lagen types I, II, and III in the embryonic chick eye.Dev Biol 59:75, 1977.

36. Layman DL, Epstein EH Jr, Dodson RF, and TitusJL: Biosynthesis of type I and III collagens by cul-tured smooth muscle cells human aorta. Proc NatlAcad Sci USA 74:671, 1977.

37. Hay ED: Development of the vertebrate cornea. IntRev Cytol (in press).

38. Jakus MA: Studies on the cornea. II. The fine struc-ture of Descemet's membrane. J Biophys BiochemCytol 2(Suppl):243, 1956.

39. Hay ED and Revel JP: Fine Structure of the Devel-oping Avian Cornea. New York, 1969, S. Karger.

40. Perlman M, Baum JL, and Kay GI: Fine structureand collagen synthesis activity of monolayer culturesof rabbit corneal endothelium. J Cell Biol 63:306,1974.

41. Kefalides NW, Cameron JD, Tomichek EA, andYanoff M: Biosynthesis of basement membrane col-lagen by rabbit corneal endothelium in vitro. J BiolChem 251:730, 1976.

42. Stenman S and Vahen A: Distribution of a majorconnective tissue protein, fibronectin, in normalhuman tissues. J Exp Med 147:1054, 1977.

43. Yamada KM and Oldern K: Fibronectin-adhesiveglycoproteins of cell surface and blood. Nature275:179, 1978.

44. Johnston MC, Nodin DM, Hazleton RD, Cou-lombre JL, and Coulombre AJ: Origins of avian ocu-lar and periocular tissues. Exp Eye Res (in press).

45. Lapier CM, Nusgens B, and Pierard GE: Interac-tion between collagen type I and type III in condi-tioning bundles organization. Connect Tissue Res5:21, 1977.

46. Klintworth GK and Smith CF: A comparative studyof extracellular glycosaminoglycans synthesized byrabbit corneal fibroblasts in organ and confluent cul-tures. Lab Invest 35:258, 1976.

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933327/ on 04/13/2018