Characterization of collagens synthesized by cultured bovine corneal ...

4 downloads 0 Views 641KB Size Report
Scheffer C. G. Tsengs, Naphtali Saviong, Denis Gospodarowiczg, and. Robert Sterns. From the +Department of Pathology and @Cancer Research Institute and ...
THEJOURNALOF BIOLOGICAL CHEMISTRY

Vol. 256, No. 7,Issue of April 10, pp. 3361-3365,1981 printed in U.S. A.

Characterization of Collagens Synthesizedby Cultured Bovine Corneal Endothelial Cells* (Received for publication, May 12, 1980)

Scheffer C. G. Tsengs, Naphtali Saviong, Denis Gospodarowiczg, and Robert Sterns From the+Department of Pathology a n d @Cancer ResearchInstitute a n d the Department of Medicine, University of California, Medical Center, Sun Francisco, California 94143

lagens synthesized by cultured corneal endothelial cells, the Collagen is themajormatrixproteinproducedby corneal endothelial cell cultures and represented 1%of major protein of the ECM, as a first step in understanding the total protein synthesis. The different types of collagenmechanism whereby this matrix can replace the requirement were determined by a combination of DEAE- andCM- for growth factors. cellulose column chromatography usingknown internal markers. Type 111 collagen was the major compoEXPERIMENTALPROCEDURES nentbothdepositedintheextracellularmatrixand Materials-Fibroblast growth factor waspurifiedfrombovine secreted into the media. The basement membrane collagens, types IV and V, were also found in each com- brains as previously described (6). Bovine calf serum, DMEM, type partment, although the latter was associated prefer- H-16, and glutamine were purchased fromGibco. Tissue culture were from Falcon Plastics. ~-[2,3,4,5-’H]proline(80 to 110 Ci/ entially with the cell matrix.The ratios of types 1:III: dishes was obtained from Amersham. L-Ascorbic acid wasfrom IV + V collagens synthesized by corneal endothelial mmol) Calbiochem. BAPN, PMSF, and PCMB were the products of the cells were3:16:1. Sigma Co., and benzamidine of the Eastman Kodak Co. DEAEcellulose (DE52) and CM-cellulose (CM52) werefrom Whatman. Pepsin A was from Worthington and Aquasol from New England Nuclear. Collagen standards were prepared as follows: unlabeled type The corneal endothelium is the innermost lining of the I collagen was extracted from lathyritic rat skin by neutral NaCl cornea. It forms an orderly cell monolayer whose basal surface fractionation (7); type 111 collagen was extracted from human fetal is in contact with a thick basement membrane termed Des- skin following pepsin digestion (8); removal of the major proportion cemet’s membrane. This separates thecornea proper from the of type I collagen was achieved by NaCl fractionation and subsequent anterior chamber. The function of the corneal endothelium in molecular sieve column chromatography; type IV collagen from bovine anterior lens capsule and type V (placenta A, B, and C chains) uzuo isdependent on the integrity of all the cells which from whole human placenta, were extracted by pepsin digestion and comprise this layer and their attachment to Descemet’s mem- NaCl fractionation, as described by Dehm and Kefalides (9) and by brane. By acting as a transport pump it maintains the corneal Sage and Bornstein (IO), respectively. The purity of each standard stroma in a permanentstate of deturgescence and ensures the was confmed by SDS-polyacrylamide gel electrophoretic analysis (data not shown) and in each case yielded the expected band profiles. transparency of the corneal stroma. Tissue Culture of Bovine Corneal Endothelial Cells-Bovine corAddition of FGF’ to the culture medium facilitates the establishment in long term cultureof the corneal endothelium neal endothelial cell cultures were established from steer corneas (1). Stock cultureswere maintained in DMEM (H-16)supplemented with (1). When exposed to this mitogen, sparse cultures have a 10% fetal calf serumand 5% calf serum. Endothelial cellswere high mitotic index and upon reaching confluence adopt the passaged weekly at a 1:64 split ratio and FGF (100 ng/ml) was added morphological organization characteristic of corneal endothe- every other day until the cultures became subconfluent. Metabolic Labezing-Subconfluent cultures of bovine corneal enlium in vivo (2), a well organized monolayer of contact-inhibited flattened cells with a polarity such that the highly dothelial cells were preincubated for 24 h in glutamine-free DMEM thrombogenic ECM is elaborated exclusively on the under- (H-16) supplemented with 10% bovine calf serum and ascorbic acid (25 pg/ml). The cultures were then exposed (24 h, 37 “C) toascorbic surface of cells. acid (25 pg/mI) and ~-[2,3,4,5-~H]proline (40 pCi/ml). BAPN (80 pg/ The ECM produced by corneal endothelium is a major cell m l ) was added to prevent collagen cross-linking. After 24 h, the product at confluence. This ECM is capable of supporting medium containing the secreted procollagen was removed and the proliferation of the corneal endothelium itself, in addition to cultures were washed three times with cold (4 “C) PBS. The plates a variety of other cultured cells (3-5), and obviates the re- were then stored at -70 “C. Culture medium and the first wash were quirement for any exogenous growth factors (3-5). These then combined. To inhibit further proteolysis, Tris-HC1 (pH 8.0), EDTA, PMSF, PCMB, and benzamidine were added a t a final conobservations stimulated us to examine the type-specific col- centration of 50, 20, 0.1, 1, and 0.1 mM respectively. The media were * This work was supported by Grant EY01930 from the National then centrifuged (800 X g, 10 min) to remove cell debris and the Institutes of Health (to D. G.).Portions were supported by grants supernatants were stored at -70 “C. from the Academic Senate and the Research Evaluation Allocation Determination of Radiolabezed Proline, 4-Hydroxyproline, a n d 3Committee, University of California a t San Francisco, and by Grant Hydroxyproline-To quantitatethe radiolabeled proline and two CA 25179 awarded by the National Cancer Institute (to R. S.). The isomers of hydroxyproline (3-hydroxyproline and 4-hydroxyproline), costs of publication of this article were defrayed in part by the the samples of media and the cell layer were processed as follows. payment of page charges. This article must therefore be hereby The media were dialyzed stepwise against 1 M NaCI, 0.2 M NaCl, and marked “aduertzsement” in accordance with 18 U.S.C. Section 1734 distilled water to remove the free radiolabeled proline. The cell layers solely to indicate this fact. which contained the cells and theECM were scraped into ahypotonic ’ The abbreviations used are: FGF, fibroblast growth factor; ECM, solution containing 10 mM NaCl, 3 mM MgCb, 0.58 Triton X-100 (pH extracellular matrix; BAPN, P-aminopropionitrile fumarate; PMSF, 7.5), homogenized, and dialyzed in the same fashion as described for phenylmethylsulfonyl fluoride; PCMB, p-hydroxymercuribenzoate; the media. After dialysis, both samples were lyophilized and hydroPBS, phosphate-bufferedsaline, pH 7.4; DMEM, Dulbecco’s-modified lyzed with 6 N HCl under vacuum for 24 h. The hydrolysates were Eagle’s medium; BCE, bovine corneal endothelium. then evaporated to dryness with air and reconstituted with distilled

3361

Corneal Endothelial Collagen Synthesis Cells in

3362

water from which aliquots were taken for the determination of total incorporation. The remainder was subjected to Dowex 50-X8cationic exchange chromatography (11) to quantitateradiolabeled proline and hydroxyproline. To separate and quantitate the two isomers of hydroxyproline, samples were applied to high voltage paper electrophoresis, as described by Tseng et al. (12). The ratios of both [3H]hydr~xyproline/[~H]proline and [3H]3-hydroxyproline/[YH]4-hydroxyproline + [3H]3-hydroxyproline were presented assuch without corrections for the loss of radioactivity which occurs when proline residues become hydroxylated. DEAE-cellulose Chromatography-To separateand quantitate procollagens which are secreted into the media, DEAE-cellulose chromatography was performed as described (13, 14). After labeling for 24 h with [3H]proline in the presence of ascorbic acid and BAPN as described above, media were dialyzed a t 4 "C against 0.15 M NaC1, 50 mM Tris-HCI, pH 7.5, containing 20 m EDTA, 1 mM PMSF, 1 mM PCMB, and 1 mM benzamidine, to remove the traces of free radiolabeled amino acid. Saturated ammonium sulfate (100%w/v) was added, to make the final concentration 30% (v/v). The resulting suspension was then stirred gently overnight at 4 "C in the presence of 3 mg of unlabeled carrier lathyritic rat skin collagen. Precipitates were collected by centrifugation (7000 X g, 30 min, 4 "C), dissolved in the initial buffer, 2 M urea, 50 m Tris-HC1, pH 7.5, and dialyzed overnight (4 "C) against the same solution. Prior to chromatography, any insoluble material was removed by filtering through cotton. Samples were applied to a column of DEAE-cellulose (1.5 X 4 cm) and washed with 25 ml of the initial buffer. Bound proteins were eluted using a linear gradient (0 to 0.2 M ) of NaCl (total volume of 160 ml, flow rate 6.2 ml/h). Fractions of2.45ml were collected, of which a 0.3-ml aliquot was removed for counting. Three ml of Aquasol was added and samples were counted in aBeckman Counter LS 8000, with a counting efficiency of 17.6% for the tritiated material. In one experiment, the remainder of the fractions were pooled from each peak region for the determination of hydroxyproline and proline. Pepsin Digestion and Salt Fractionation-The culture medium and the cell layer were treated with pepsin separately. Ammonium sulfate (30% v/v) was added to the media. Following centrifugation, the pellets were dissolved and dialyzed (24 h, 4 "C) against 0.5 N acetic acid and then lyophilized.The lyophilized samples were further dissolved in 0.5 N acetic acid containing pepsin A (100 pg/ml). The cell layers, together with the ECM, were suspended in 0.5 N acetic acid and homogenized. Pepsin A (100 pg/ml) was then added. Both samples (media and cell layer) were digested separately for 24 h a t 4 "C and the reaction was then stopped by neutralization with NaOH to pH 8.5. Both solutions were dialyzed against 1 M NaC1, 50 mM Tris-HC1 (pH 7.5), and collagens precipitated stepwise by the slow addition of NaCl to a final concentration of 1.7, 2.6, and 4.5 M, respectively (15). The developing precipitates were stirred overnight (4 "C) and collected by centrifugation (7000 X g, 4 "C, 30 min). Each fractionated precipitate was then dissolved in 0.5 N acetic acid, dialyzed against 0.1 N acetic acid, and lyophilized. CM-cellulose Chromatography-CM-cellulose column chromatography was pe ormed as described previously (16). Radioactive samples from eac salt fractionation were dissolved in 40 mM sodium acetate, 2 M urea, pH 4.8, the initial buffer. Lathyritic rat skin collagen, containing primarily type I, was added (2 to 5 mg) as an internal standard and carrier. For certain samples, type I11 collagen prepared from fetal skin, or type V (A, B, C chains) from placenta, was added. Prior to chromatography, samples were denatured by heat treatment (42 "C, 30 min) and the insoluble material was removed by centrifugation (1000 X g, 10 min) at room temperature. Aliquots were taken for determination of the total radioactivity in each salt fraction. Samples were then applied to a jacketed CM-cellulose column (0.7 X 10 cm), pre-equilibrated with the deareated intial buffer a t 42 "C, and washed with 25 ml of the initial buffer. Bound proteins were eluted using a linear gradient (0 to 0.1 M) of NaCl (total volume, 150 ml, flow rate 37 ml/h). Fractions of 2.5 ml were cpllecteci. Absorbance at 230 nm was detected by a Gilford spectrophotometer, model 2521, for the unlabeled carrier collagen. Aliquots of labeled material were counted by a Beckman LS 8000 scintillation counter with a counting efficiency of 28.8% for tritiated material. Conductivities were measured at room temperature. Cell Counting-The cell number was obtained by trypsinizing triplicate cultures and counting cells in a Coulter counter.

TABLE I / 3 H J P r incorporation ~ and collagen accumulation by subconfluent bovine corneal endothelialcells

I."..

cells

Medium Cell layerd Total

27.3' 48.5 75.8

line

tion

6.45

cells 17.6

1.41 3.20

8.0 6.8 7.1 24.4

2.17 0.47 1.08

6.7

a C/P indicates the ratio of collagen to total protein. It is calculated assuming that thehydroxyproline residue content of collagen is 12.2% and that theaverage proline content in total cellular protein is 4.1% (17). Two isomers of hydroxyproline, 3-hydroxyproline and 4-hydroxyproline, were separated andmeasured by high voltage paper electrophoresis as described by Tseng et al. (12). Ratios of 3-hydroxyproline to thesum of 3- and 4-hydrox~1~roline are reported. The results presented here are the mean of four different measurements. The cell layer denotes the cells and their ECM.

SUBCONFLUENT BOVINE CORNEAL ENDOTHELIALCELLS

f

RESULTS

Collagen Synthesis in Corneal Endothelial Cells-Bovine corneal endothelial subconfluent cell cultures (BCE) were

FRACTION NUMBER FIG. 1. DEAE-cellulosechromatograms of C3H]proline-labeled medium proteins of ( A ) BCE cells and (B) human skin fibroblasts. Samples were processed and chromatographed as described under "Experimental Procedures." Arrows indicate start of NaCl gradient. Conductivities were measured at room temperature. Peak 1 was unabsorbed fractions containing collagens and procollagen type IV. Peaks 2 and 3 corresponded to procollagen types I and 111 respectively. Peaks 4 and 5 were tentatively identified as precursors for type V collagen.

labeled with C3H]proline for 24 h. As shown in Table I, the amount of C3H]proline incorporated into proteins present in the cell layer was nearly twice that secreted into themedium. The cell layer refers to both cells and their underlying ECM. However, the radioactivity recovered in the form of hydrox-

Collagen Synthesis in Corneal Endothelial Cells

3363

yproline in the medium was 2.5 times (1.76 X lo5 cpm per lo6 cells) that recovered in the cell layer (6.8 X lo4 cpm per lo6 cells) (Table I). By calculation (17), 2.17%and 0.47%of total protein synthesis was devoted to collagen in the medium and in the cell layer, respectively (Table I). The media and cell layers were further analyzed for the distribution of the hydroxyproline isomers, 3- and 4-hydroxyproline, which are more abundant for basement membrane collagen and interstitialcollagen, respectively (18).3-Hydroxyproline was found to be accumulated in greater amounts in the cell layer, reflecting the preferential association of basement membrane collagen with the cell layer (8.0%in the cell layer and 6.7%in the medium). DEAE-cellulose Chromatography of Medium ProteinsThe types of procollagens secreted by cells into the culture media were analyzed. Media proteins precipitated by30% ammonium sulfate were chromatographed on a DEAE-cellulose column. Chromatography of the BCE medium gave an elution profie (Fig. 1A) composed of two major peaks (denoted l and 3) and three minor peaks or regions (denoted 2, 4, and 5). The fist major peak, which waseluted in the initial wash, represented an unabsorbed fraction. The other major peak was eluted during the course of NaCl gradient between fractions 42 to 50 (peak 3). The ratio of hydroxyproline/ proline in these 5 peaks or regions, 23%for peak 1,9%for peak 2,16% for peak 3, 6% for region 4, and 3% for region 5,

..

_--

””””

”””

_””

-

1.-

FRACTIONNUMBER FIG. 3. CM-cellulose chromatograms of pepsin-digested

[3H]proline-labeledproteins in themedia (A) and the cell layers (B) of BCE cells which were precipitated at 2.6 M NaCl under neutral conditions. Both samples were procesed and chro-

“”-*-

I

b

matographed as described under “Experimental Procedures.” Lathyritic rat skin collagen, mainly type I, was co-chromatographed in both cases. A distinct peak co-eluted with the al(1) standard and was readily observed in the profiie of the cell layer ( B ) . However, the media ( A ) contained primarily type I11 collagen (al(III))3. Arrows indicate the start of the gradient. Conductivities were measured a t room temperature.

t

5, 10 0

I

I I 0 Y

FRACTIONNUMBER FIG. 2. CM-cellulose chromatograms of pepsin-digested [31-Ilproline-labeled proteins in themedia ( A ) and the cell layers (B) of BCE cells which were precipitated at 1.7 M NaCl under neutral conditions. The samples were processed and chromatographed as described under “Experimental Procedures.” For the media ( A ) ,1.7 M NaCl precipitate of human fetal skin, which contained 60%type I11 collagen, was co-chromatographed as an internal standard and carrier. For the cell layers ( B ) , lathyritic rat skin collagen, mostly type I, was employed. Both media and cell layers exhibited the same elution profiies with a prominent peak of type I11 (~yl(II1))~ collagen. Arrows indicated the start of the NaCl gradient. Conductivities were measured at room temperature.

indicated that all contained collagenous materials. Further identification of each peak or region was made using human skin fibroblast procollagens as markers (Fig. 1B). By comparing elution profiies (Fig. 1, A and B ) , peak 2 was identified as procollagen type I and peak 3 as procollagen type 111. Procollagen type IV is reported to elute from the DEAE-cellulose column in the initial wash (19, 20)) corresponding to peak 1. This suggested that peak 1 contained procollagen type IV. Regions 4 and 5 were tentatively identified as the elution positions for the precursors of type V collagen? The ratio between procollagen type I11 and procollagen type I was nearly 501. Thus, type I11 procollagenwas the principal collagen secreted and accumulated into the media by BCE cells. CM-celluloseChromatography of pepsin-digested Proteins from the Mediaand the Cell Layers-To quantitate thetypes of collagens produced by BCE cultures, samples from media and cell layers were analyzed after pepsin digestion under acidic conditions, followedby salt fractionation. NaCl was added stepwise to a final concentration of 1.7, 2.6, and 4.5 M under neutral conditions. Each salt concentration precipitates preferentially certain type-specific collagens(10,16,21).Each salt fractionate was analyzed by CM-cellulose chromatography. Acidic noncollagenous moieties are eluted in the initial S. C.G. Tseng, N. Savion, D. Gospodarowicz, and R. Stern, unpublished experiments.

3364

Collagen Synthesis in Corneal Endothelial Cells

wash, while different collagen chains are resolved by NaCl gradient elution under denaturing conditions. When the 1.7 M NaCl fractions of both the media and the cell layers were analyzed, the elution profiies werethe same (Fig. 2, A and B ). Only type I11 collagen ( ( ~ l ( I 1 1 )was ) ~ observed. A prominent peak was observed which eluted prior to theelution position of a2 chain (Fig. 2B). This peak co-eluted with type I11 collagen (Fig.2A). This carrier, prepared from the 1.7 M NaCl precipitate of fetal skin collagen, contained 60% type 111 collagen as documented by both sodium dodecyl sulfate-polyacrylamide gel and molecular sieve column chromatography (results notshown). Minor peaks co-eluting with the 21 position were also detected. When 2.6 M NaCl fractions were analyzed from both the media and the cell layers, al(1) and a2 of type I collagen were detected and were observed to coelute with the type I standard carrier. However, negligible amounts of al(1) and a2 were observed in the media as compared to the cell layer (Fig. 3, A and B ) . This reflected increased rates of processing of type I collagen, as described previously by Goldberg (22). The 4.5 M NaCl fractions of both samples were also analyzed, fractions containing primarily type IV and type V collagens (10, 21). Both cell layers and media yielded an

1. U 0 I(rr

3

X

TABLEI1 Ratios of differentcollagen types synthesized by subconfluent bovine corneal endothelial cells IIV I11 v Total radioactivity (cpm Ratio'

X

1

29.8 2.6

177.3 15.7

3.5

3.9

Total radioactivity foreach type of collagen is calculated fromthe results of CM-cellulosechromatograms,byintegratingdesignated peaks fromeach salt fractionate and combining results of both media and cell layer. Type I collagen is calculated from the al(1) peak by a correctionfactor of 3/2, basedon its heteropolymerstructure as ((aI)na2).Data from the two cell cultures are presented after adjusting for DNA levels. Ratio is measured by combining the values for collagen types IV and V.

'

elution profiie with two major peaks which emerged between the elution positions of al(1) and a2 chain of type I collagen (Fig. 4, A and B ) ) .This elution profie was similar to that of type V collagen, as shown when purified A, B, and C chains from a placental extract were co-chromatographed with media sample (Fig. 4A). We have identified the first peak as the A chain and the second peak as B, and possibly C chains, which co-elute and cannot beresolvedfrom each other by this method (10).The shoulder of the A chain peak and the minor peak preceding the al(1) elution position were tentatively identified as al(IV), type IV collagen, when compared to the elution profiie of a type IV collagen standard, which had major peaks with elution positions preceding that of al(1) (9). When the 4.5 M NaCl supernatant fraction was analyzed, no peaks could be detected, demonstrating that all collagenous materials had been precipitated by 4.5 M NaCl (results not shown). By integrating the peaks for the different types of collagen and combining the results of medium and cell layer, it was found that BCE cellssynthesized collagen types I and 111 and basement membrane collagens at a ratio of 3:161 (Table 11). DISCUSSION

In the eye, primary mesenchyme originating from the primary streak gives rise to corneal endothelium as well as to the vascular endothelium lying between the retina andsclera (2326). Thus these two types of endothelial cells have a common embryologicalorigin. Secondary mesenchyme whichoriginates from the neural crest (27, 28) gives rise to the stroma proper of the cornea, composed entirely of fibroblasts. Thus, corneal and vascular endothelial cells, although of mesenchymal origin, are derived from an embryological source entirely different from that of fibroblasts. Wewished to examine whether cells of a common embryological source maintain a similarity in the phenotypic expression of type-specific collagen synthesis. Collagen is the major protein of ECM. Currently, at least five different types of collagen have been observed, each a FRACTIONNUMBER distinct gene product. The present study on corneal endotheFIG. 4. CM-cellulose chromatograms of pepsin-digested lium, together with previous experiments on vascular endo['Hlproline-labeled proteins in the media (A) and the cell lay- thelium from this laboratory3 and others (30, 31), establishes ers (B) of BCE cells which were precipitated at 4.5 M NaCl that type I11 collagen is the major collagen synthesized by all under neutral conditions. The samples were processed and chromatographedas described under "Experimental Prodedures." the For cultured endothelial cells. This clearly distinguishes endothemedia (A), placenta extracts containing A, B, C, chains were co- lial cells from fibroblasts, which synthesize type I and type I11 chromatographed as an internal standard and carrier. For the cell at a ratio of approximately 3:l (32). A preponderance of type layers ( B ) ,however, lathyritic rat skin collagen, containing primarilyI11 collagen appears to be a characteristic of all endothelial type I, was co-chromatographed.Both the media and the cell layers cells, regardless of origin. However, small amounts of other exhibited the sameprofdes with two majorpeakswhich eluted collagensare also found among different kinds of endothelium. the A and Bchains respectively. between al(1) and a2 and represented A minor peakeluting prior to al(1) was identified as al(1V). Arrows indicate the start of the NaCl gradient. Conductivities were measured 3S. C. G. Tseng, N. Savion, R. Stern, and D. Gospodarowicz, unpublished experiments. at room temperature.

Collagen Synthesis

in Corneal Endothelial Cells

Type I collagen is made by corneal endothelium, while vascular endothelium does not produce it, even when both cell types are grown and maintained under optimal culture cond i t i o n ~Thus, ~ . some degree of heterogeneity in the phenotypic expression of collagen synthesis occurs among similar types of cells, although the major collagen types remain the same. Increases in production of type I collagen are usually associated with suboptimal growth conditions in cultured cells: vascular endothelial cells repeatedly passaged at lower density in the absence of FGF3, skin fibroblasts asa function of increasing age (33), or chondrocytes with progressive passage number (34). A portion of the observed heterogeneity in the phenotypic expression of collagen synthesis may also be a reflection of differences in status of the cellorin culture conditions. The collagenous composition of the intact bovine cornea has been analyzed by several investigators (35-38). However, such work has focused on the cornea as a whole without preliminary separation of the component layers. The presence of type I11 collagen cited in these studies has been an inconsistent observation and remains controversial. An additional unresolved problem is whether type I11 collagen is present in vivo in the adult bovine corneal endothelium. To clarify this point, characterizations of collagen composition in adult and fetal corneal endothelia are now in progress. One of the major characteristics of corneal endothelium i n vivo is that it is formed by a honeycomb-like monolayer of flattened, contact-inhibited, and highly polarized cells, which deposit an ECM exclusively underneath or on the basal cell surface. This ECM forms the distinct structure, Descemet’s membrane. The aforementioned characteristics are retained when cells are grown in long term culture in the presence of FGF. These cells therefore offer a model for the study of the synthesis and asymmetrical deposition of ECM. The present study has established that, in addition to interstitial collagen types I11 and I, basement membrane collagens types IV and V are also elaborated by corneal endothelial cells in vitro into their ECM. This is compatible with previous results demonstrating that Descemet’s membrane in vivo contains type IV and V (A and B chains) collagens (37,39). Corneal endothelial ECM is capable of supporting proliferation of a variety of cultured cells,in addition to corneal endothelial cells themselves (3-5). The matrix completely obviates the requirement of these cells for exogenous FGF. Characterization of the collagens synthesized by BCE cells in culture as described herein may provide a mechanism which will explain why a preformed matrix can substitute for an exogenous peptide growth factor. Acknowledgments-We wish to acknowledge the invaluable help of Mr. Harvey Scodel in the preparation of this manuscript. REFERENCES 1. Gospodarowicz, D., Mescher, A. L. & Birdwell, C. R. (1977) Exp. Eye Res. 25, 75-89.

3365

2. Gospodarowicz, D., Vlodavsky, I. & Savion, N. (1980) Vision Res. 21,87-I03 3. Gospodarowicz, D., Delgado, D. & Vlodavsky, I. (1980) Proc. Natl. Acad Sci. U. S. A. 77,4094-4098 4. Gospodarowicz, D. & Ill, C. R. (1980) Exp. Eye Res. 31, 181-199 5. Gospodarowicz, D. & Ill, C. R. (1980) J. Clin. Znuest. 65, 13511364 6. Gospodarowicz, D., Bialecki, H. & Greenburg, G. (1978) J. Biol. Chem. 253,3736-3743 7. Bornstein, P. & Piez, K. A. (1966) Biochemistry 5, 3460-3473 8. Byers, P. H., McKenney, K. H., Lichtenstein, J. R. & Martin, G . R. (1974) Biochemistry 13,5243-5248 9. Dehm, P. & Kefalides, N. A. (1978) J. Biol. Chem. 253,6680-6686 10. Sage, H. & Bornstein, P. (1979) Biochemistry 18, 3815-3822 11. Cutroneo, K. R., Guzman, N. A. & Liebelt, A. G. (1972) Cancer Res. 32, 2828-2833 12. Tseng, S. C. G., Stern, R. & Nitecki, D. (1980) Anal. Biochem. 102,291-299 13. Smith, B. D., Byers, P. H. & Martin, G. R. (1972) Proc. Natl. Acad. Sci. U. S. A. 69,3260-3262 14. Burke, J. M., Balian, G., Ross, R. & Bornstein, P. (1977) Biochemistry 16, 3243-3249 15. Chung, E. & Miller, E. J. (1974) Science 183, 1200-1201 16. Epstein, E. H., Jr., Scott, R. D., Miller, E. J. & Piez, K. A. (1971) J. Biol. Chem. 246, 1718-1724 17. Green, H. & Goldberg, B. (1963) Nature 250, 1097-1098 18. Man, M. & Adams, E. (1975) Biochem. Biophys. Res. Commun. 66,9-I6 19. Crouch, E. & Bornstein, P. (1978) Biochemistry 17,5499-5509 20. Liotta, L. A., Wicha, M. S., Foidart, J . M., Rennard, S. I., Garbisa, S. & Kidwell, W. R. (1979) Lab. Invest. 41, 511-518 21. Kresina, T. F. & Miller, E. J . (1979) Biochemistry 18,3089-3097 22. Goldberg, B. (1977) Proc. Natl. Acad.Sci. U. S. A . 74,3322-3325 23. Lopashov, G. V. & Stroeva, 0.G. (1961) Adu. Morphog. 1, 331337 24. Sabin, F.R. (1920) Contrib. Embryol. Carnegie Znst. 9, 213-262 25. Trelstad, R. L., Hay, E. D. & Revel, J. P. (1967) Deu. Biol. 16,78106 26. Hay,E. D. & Revel, J. P. (1968) in Epithelial-Mesenchymal Interactions, pp. 31-55, Williams & Wilkins Co., Baltimore 27. Horstadius, S. (1950) The Neural Crest, Oxford University Press, London 28. Le Douarin, N. M., Teillet, M. A. & Le Lievre, C. (1977) in Cell and Tissue Interactions (Lash, J. W. & Burger, M. M., eds) pp. 11-26, Raven Press, New York 29. Deleted in proof 30. Barnes, M. J., Morton, L. F. & Levene, C. I. (1978) Biochem. Biophys. Res. Commun. 84,646-653 31. Sage, H., Crouch, E. & Bornstein, P. (1979) Biochemistry 18, 5433-5442 32. Smith, B. D., Byers, P. H. & Martin, G. R. (1972) Proc. Nutl. Acad. Sci. U. S. A. 69,3260-3262 33. Digelmann, R. F. & Peterkofsky, B. (1972) Deu. Biol. 28,443-453 3 4 . Benja, P. D., Padilia, S. R. & Nimni, M. E. (1978) Cell 15, 13131321 35. Schmut, 0. (1977) Exp. Eye Res. 25,505-509 36. Freeman, I. L. (1978) Znuest. Ophthalmol. Visual Sci. 17, 171177 37. Davison, P. F., Hong, B. S. & Cannon, D. J. (1979) Exp. Eye Res. 29,97-107 38. Praus, R., Brettschneider, I. & Adam, M. (1979) Exp. Eye Res. 29,469-477 39. Kefalides, N. A., Cameron, J. D., Tomichek, E. A. & Yanoff, M. (1976) J . Biol. Chem. 251, 730-733