Biosynthesis and Regulation of Type V Collagen in Diploid Human Fibroblasts*

The biosynthesis of type V collagen and its regulation were studied using diploid human gingival fibroblasts. Cells were metabolically labeled with radioactive amino acids and labeled proteins were subjected to limited pepsin digestion, DEAE-cellulose chromatography at 15 “C, and polyacrylamide gel electrophoresis. Proteins eluted from DEAE-cellulose columns by 0.25 M NaCl contained a collagen species which was resistant to mammalian collagenase and had a chains with hydroxylysine/lysine ratios and CNBr peptide patterns similar to al(V) and a2(V). Procollagen(V) fractions obtained by DEAE-cellulose chromatography and immunoprecipitates of type V collagen antibody contained polypeptides with M, = 239,000, 219,000, 198,000,174,000,157,000, and 132,000. By compar- ing the CNBr peptide maps of these proteins with those of standard al(V) and a2(V) chains, the first three polypeptides were shown to be related to al(V) and the others to a2(V). It was concluded that the gingi- val fibroblasts synthesize type V collagen, that the proal(V) and the proa2(V) chains have M, = 239,000 and 174,000, respectively, and that the al(V) and a2(V) chains laid in the form of fibrils have M, = 198,000 and 132,000, respectively. A detectable amount of type V collagen was synthesized

V synthesized were increased when the cells were labeled in the presence of serum, and the increase was accompanied by a decrease in type 111. This effect was dependent on serum concentration. Serum obtained from platelet-poor plasma failed to elicit this effect, and it was restored by the addition of platelet-derived growth factor. Platelet-derived growth factor was effective in medium with and without platelet-poor serum. Thus, it appears that platelet-derived growth factor may be an important regulatory factor in the synthesis of types V and I11 collagens.
The collagens represent a group of at least five closely related but genetically distinct structural proteins designated as types I, 11,111,IV,and V (1). Types I and 111 are found in many connective tissues, whereas the presence of types I1 and IV is restricted to cartilage and basement membranes, respectively. Type V collagen was first isolated from fetal membranes (2); however, later studies revealed that it is present * This work was supported by National Institutes of Health Grants DE-02600 and.DE-03301. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. in most connective tissues along with types I and 111 or I1 (3), and that it has a unique ultrastructural localization in pericellular spaces and near basement membranes (4-6). The type V collagen appears to be involved in many important biological processes such as platelet aggregation, epithelial cell migration, substrate attachment, and binding of other interstitial collagen fibrils (7-12).
Many aspects of the structure and biosynthesis of type V collagen are not fully known. Pepsin digests of tissues contain all and a 2 chains as the major type V components; however, their ratio varies from tissue to tissue and with the developmental status of the animal (2,3, 13). These and the thermal denaturation studies of Rhodes and Miller (3) indicate that type V collagen exists in two major molecular forms with chain composition (011)~ and (~y l )~a 2 .
However, the pepsin extracts also contain an additional a3 chain (14-16), and it is not clear whether it is a constituent of type V molecules or not. Biosynthetic studies using cell and organ cultures have indicated that the type V collagen molecules laid down as fibrils may retain a significant portion of nonhelical propeptide segments (17)(18)(19). These studies and the electron microscopic observations of Bachinger et al. (20) also indicate that the type V procollagen has larger globular extensions than procollagen type I, but both have similar sized helical regions. The proal(V) and proa2(V) chains elaborated by a human rhabdomyosarcoma cell line have M, = 220,000 and 150,000 (21), and whether these values hold true for normal cells as well is not known. Also nothing is known about mechanisms which regulate synthesis of type V collagen. An understanding of the regulatory mechanisms is important because abnormal amounts of type V collagen are present in many diseases such as atherosclerosis, chronic inflammation, and carcinoma (22)(23)(24)(25).
We report here studies on the synthesis of type V collagen by diploid fibroblasts, which are the predominant cell type responsible for synthesis of collagens in the gingiva (26). Normal human gingiva contains al, a2, and additional as yet unidentified type V chains, and the nature of the chain composition has not been resolved (27). The content and proportion of type V collagen is elevated several-fold in chronically inflamed gingiva (24). The studies described here were undertaken to determine the form and chain composition of type V collagen in gingiva and to investigate mechanisms which may regulate type V collagen biosynthesis and accumulation.

MATERIALS AND METHODS
cell Culture and Labeling-Fibroblasts were obtained from explants of human gingiva from an individual with clinically and Biosynthesis and Regulation of Type V Collugen radiographically healthy periodontal tissues. Cells were maintained and labeled as described previously (26). Briefly, confluent cultures in 75 cmz flasks were preincubated for 1 h in serum-free Dulbecco-Vogt medium containing 50 pg/ml each of ascorbic acid and j3aminopropionitrile, and then labeled for 24 h in 5.0 ml of the same medium containing 10 pCi/ml each of ~-[2,3-~H]proline (specific activity 20-40 Ci/mmol) and [2-3H]glycine (specific activity  Ci/mmol) (New England Nuclear). Separation of Native Collagens-After the labeling, medium was separated and the cells were taken in 0.5 M NH3. Both fractions were pooled, and then subjected to limited pepsin digestion at 15 "C for 6 h (26), and collagens were precipitated by 2.0 M NaCI. After centrifugation, they were dissolved in and dialyzed versus 50 mM Tris buffer, pH 8.6, containing 20 mM NaCl and 2 M urea. The material was loaded onto a DEAE-cellulose column maintained at 15 "C and eluted stepwise first by the buffer, then with 0.11 M NaCl, and finally by 0.25 M NaCl in the buffer. Fractions were monitored for radioactivity and peak fractions were pooled, desalted by dialysis versus 0.1 M CH&OOH, and then lyophilized.
DEAE-cellulose Chromatography of Procollagens-Cells were taken in 0.5 M NH3 containing 10 mM of N-ethylmaleimide, 1 mM phenylmethylsulfonyl fluoride, and 25 mM EDTA. Procollagens were precipitated at 50% saturated (NH4),S0, at 4 "C and recovered by centrifugation. The precipitated material was dissolved in 50 mM Tris buffer, pH 7.5, containing 2 M urea, dialyzed versus the same buffer, and loaded onto a DEAE-cellulose column at 4 "C (28). The column was washed with the buffer and then eluted stepwise with 0.13 M NaCl and 1.0 M NaCl in the same buffer. Fractions containing radioactive peaks were pooled, desalted by dialysis, and lyophilized.
Sodium Dodecyl Sulfate-Polyacrylamide Slab Gel Electrophoresis and Fluorescence Autoradiography-Proa and a chains of collagens and their CNBr-digested peptides were separated on 5 and 12.5% gel slabs, respectively, as described elsewhere (27). Samples were run without reduction except when specified. Gels were fixed in 10% CC13COOH, 10% CH3COOH, 30% CHIOH, treated with Enhance (New England Nuclear) and exposed to Kodak X-Omat AR film.
Qwntitation of Collagens-Exposed films were scanned in a Helena autoscan densitometer at 610 nm. Relative proportions of radioactive collagen chains were quantitated from respective peak areas. From this value and from total radioactivity of DEAE-cellulose fractions containing collagens, amounts and proportions of different collagen types were calculated.
CNBr Digestion-Collagen and procollagen chains were first separated on 1.3-mm thick 5% slab gels. Bands were located either by Coomassie blue staining or by fluorography of representative runs. Corresponding bands from untreated gels were cut, suspended in 70% HCOOH, and digested with 50 mg of CNBr. They were washed in 100 mM Tris-HCI buffer, pH 6.8, containing 20% glycerol, layered on a 12.5% gel slab (1.6-mm thickness), and subjected to electrophoresis (29).
Digestion with Mammalian Collagenase-Radioactive proteins were digested with 50 pg of purified human skin fibroblast collagenase (obtained as a gift from Dr. Eugene A. Bauer, Washington University, St. Louis, MO), in 50 mM Tris buffer, pH 7.5, containing 1 mM phenylmethylsulfonyl fluoride, 2 mM N-ethylmaleimide, and 5 mM CaCIZ. After incubation a t 25 "C for 24 h, the materials were lyophilized and separated on 8% polyacrylamide gels (30).
Hyl/Lys Ratios-For measurement of Hyl/Lys ratios, cells were labeled with ~-[U-"C]lysine (2.5 pCi/ml, >300 mCi/mmol, New England Nuclear), and collagens were obtained by DEAE-cellulose chromatography as described above. a chains were separated on 5% gels, along with 10 pg of carrier types I and V collagens, and bands were located by staining with Coomassie blue. Bands containing individual a chains were cut from the gel slabs, extracted by vigorously shaking four times with 25% isopropanol, followed by thrice with 10% methanol, and then lyophilized (21). They were suspended in 50 mM (NH4)*CO3 buffer, pH 8.8, and then digested with 50 pg/ml of chromatographically purified porcine pancreatic elastase (Sigma E-0127, specific activity 120 units/mg) for 24 h at 42 "C. This procedure solubilized >94% of the radioactive proteins. These were lyophilized and hydrolyzed with 6 M HCl at 108 ' C for 24 h. Amino acid analysis was carried out on a Beckman 12OC analyzer as described elsewhere Immunoprecipitation-Rat antibody to type V collagen was prepared by immunizing rata with chromatographically purified gingival type V collagen (31). The antibody showed no cross-reactivity with type I, 111, and IV collagens when examined by the enzyme-linked immunosorbent assay technique (data not shown). For precipitation (27). of radioactive proteins, labeled cell and medium proteins containing proteinase inhibitors were dialyzed versus 100 mM NaH2P04, 150 mM NaCl buffer, pH 7.2, and then treated with 100 pl of a 1:lO diluted antiserum for 1 h at mom temperature. Then 100 pl of 5 mg/ml goat anti-rat I& (Cappel Laboratories, Inc.. Cochranville, PA) was added and incubation was continued at mom temperature for 4 h and then overnight at 4 "C. Precipitates were recovered by centrifugation in a Microfuge and washed with phosphate buffer containing 10 mg/ml of bovine serum albumin and 0.1% Tween 20.

RESULTS
In preliminary experiments we observed that fibroblasts which are not confluent and those which have only recently become confluent did not synthesize detectable amounts of type V collagen; significant amounts were made only after the cells had remained confluent for 7 days or more. Thereafter, synthesis increased with increasing time in culture for up to 15 days then remained the same until 20 days (data not shown). Whereas the majority of type I and I11 collagens was secreted into the medium (62-79'31 and 84-9796, respectively), the presence of type V was restricted to the cell layer, and very little was present in the culture medium (data not shown). The radioactivity in al(V) and a2(V) chains represented 0 . 2 4 5 % of the total present in cells plus medium; this amount was greater by several-fold when labeling was carried out in the presence of 10% fetal calf serum (see below). Therefore, for most of the studies described here, cells were labeled after 15 days in culture in the presence of 10% serum, and both medium-and cell-associated collagens were combined and processed together.
Characterization of Type V CoUagen Chains-For this purpose, labeled proteins were subjected to limited pepsin digestion, and collagens were fractionated on a DEAE-cellulose column at 15 "C. Initially separation was attempted with a 20-300 mM NaCl gradient; however, due to inefficient separation, a stepwise elution procedure was developed (Fig. l); 65 * 12% (n = 21) of the loaded radioactivity was recovered in fractions I, 11, and 111, and no additional radioactive material was recovered by further washing with l M NaCl and 0.4 M NaOH. Electrophoresis on 5% sodium dodecyl sulfatepolyacrylamide gels showed that fraction I (breakthrough peak), which contained approximately 1-7% of the loaded radioactivity, did not contain collagen chains ( Fig. 1, imet, I).
Fraction 11, which was eluted by 0.11 M NaCl, contained FIG. 1. DEAE-cellulose chromatography of native collagen molecules obtained by limited pepsin digestion of labeled cell and medium proteins. The column was maintained at 15 "C and 9.2 X 106 cpm were loaded. Arrows indicate the start of 0.11 M (left) and 0.25 M (right) NaCl elution. Fractions I, 11, and I11 contained 0.2, 6.4, and 1.2 X l@ cprn (2, 69, and 13% of the loaded radioactivity, respectively). Inset, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (5%) patterns of fractions I, 11, and I11 under nonreducing conditions. The proteins in fraction 111 (bands a, b, c, and d ) were used for further characterization. The arrowheads from top to bottom indicate the migration of standard disulfide-linked trimer al(III)a, al(I), and a2(Ij chains, respectively.

Biosynthesis and
Regulation of Type V Collagen proteins migrating similar to al(I), a2(I), and the disulfide trimer (~l ( I 1 1 )~ ( Fig. 1, inset, ZI), demonstrating the presence of types I and I11 collagens. Characterization of these collagens has been described previously (26), and therefore is not elaborated upon here. Fraction 111, the 0.25 M eluate, contained four bands (a-d, Fig. 1, inset, HZ). These proteins had electrophoretic migration similar to the a chains of types I and V collagens. No other a chains were detected, and the ratio of a/b and c/d was 2.0. They were characterized as follows. In the first experiment, cells were labeled with [I4C]lysine and the extent of lysine hydroxylation in each of the four bands was determined as described under "Materials and Methods." From Table I it is seen that values for bands a and b are greater than those for c and d, and that values for a and b and c and d resembled a1 and a2 chains of type V and I collagens, respectively. In the second experiment, fraction I11 was subjected to digestion with human fibroblast collagenase, and degradation products were separated on 8% polyacrylamide slab gels. Fraction I1 was treated similarly for a positive control, and it caused the removal of al(I), a2(I), and al(II1) chains, and the appearance of new bands at locations expected for TCA and TCR peptides corresponding to M, = 75,000 and 25,000, respectively (Fig. 2, ZZ). In fraction 111, bands c and d behaved in a similar fashion; however, bands a and b remained unaffected, indicating that collagen composed of these a   Fig. 1, fraction 111. chains is resistant to mammalian collagenase (Fig. 2, ZZZ). In the third experiment, fraction I11 was first separated on 5% gels, and the four bands obtained were digested with CNBr; peptides formed were then separated on 12.5% gels. The resulting fluorogram was compared to a similar run for the nonradioactive a chains of type I and V collagens, but stained with Coomassie blue. From Fig. 3 it is seen that peptide patterns for bands a, b, c, and d are identical, respectively, to those of a1 and a2 chains of types V and I collagens.
Characterization of Procollagen(V) Chains-For these studies labeled proteins were taken in buffer containing enzyme inhibitors, precipitated with 50% (NH)4)ZS04r and then separated by DEAE-cellulose chromatography at 4 "C. Separation was initially carried out by a 0-200 mM NaCl gradient. However, satisfactory separation of procollagen(V) could not be achieved from the relatively high amounts of procollagen(1). Therefore, a stepwise elution procedure was developed (see "Materials and Methods"). The column was loaded and then eluted in sequence with buffer (fraction I), 0.13 M NaCl (fraction II), and 1.0 M NaCl (fraction 111) (figure not shown). Further elution of the column with 0.5 M HCl, 6 M urea recovered <0.7% of radioactivity and the eluted material after pepsin digestion yielded al(1) and a2(I) (data not shown). Fraction I contained processed a chains and noncollagenous material (data not shown; Ref. 26), whereas fraction I1 contained precursor a chains of type I and 111 collagens, and the latter yielded type I and I11 a chains after pepsin treatment (Fig. 4, a-d). Characterization of these proteins has been described previously (26); therefore, they are not discussed further. Fraction I11 contained 6-9% of loaded radioactivity, and electrophoresis under nonreducing conditions revealed the presence of two major bands near the origin and at least four other faster migrating bands (Fig. 4f). Prior reduction with MSH removed a major band near the origin, intensified 'k," and gave rise to a minor band between u and v; other bands were not affected (Fig. 4e). All these proteins were removed by digestion with bacterial collagenase (Fig. 4h). After pepsin digestion, al(V) and a2(V) chains were the predominant species (al/a2 ratio 2.0), and no type I a chains were present (Fig. 4g). Thus, the collagen components present in fraction I11 appear to be type V. In order to confirm this  (a-d), and the standard collagens by Coomassie blue staining (e-h). The bands were cut, digested with CNBr, layered over a 12.5% gel, and electrophoresed as described under "Materials and Methods." a, 6, c, and d represent patterns for the respective bands in the Fig. 1 inset. e, al(V); f, a2(V); g, al(1); h, a2(I). Peptides derived from type I collagen chains are identified. 37 'C, i, antibody precipitate. d and h were exposed for 2 weeks, and i was exposed for 34 days. All other patterns were obtained after 47 h of exposure. conclusion and to characterize the individual bands, they were subjected to CNBr digestion followed by electrophoresis on 12.5% gels. From Fig. 5 it is seen that the peptide patterns fall into two groups; p, q, r, and s are similar to each other, as are t, u, and v. When these patterns are compared with those obtained for al(V) and a2(V) chains in Fig. 3, e and f, it is observed that peptide maps of p, q, r, and s resemble al(V), and t, u, and v are similar to a2(V). From five separate experiments, the molecular weights of bands p, q, r, s, t, u, and v were calculated, based on collagen standards, to be 370,000 f 10,000, 239,000 f 6,000, 219,000 f 15,000, 198,000 f 10,000, 174,000 & 12,000, 157,000 f 10,000, and 132,000 f 12,000, respectively. Immune precipitates obtained using type V antibody on electrophoresis revealed the presence of three protein bands, one of M , = 373,000 remaining near the origin, and two others of M , = 282,000 and 195,000 (Fig. 4i). However, the pattern obtained under reducing conditions was similar to Fig. 4e, and after limited pepsin digestion, only two proteins migrating with al(V) and a2(V) chains were observed (not shown). Thus, these data indicate that the procollagen preparations contain two groups of proteins which are related to al(V) and a2(V), respectively.
Modulation of Type V and 1 1 1 Synthesis-We have reported previously that labeling fibroblasts in the presence of 10% fetal calf serum causes a reduction in the proportion of type I11 collagen synthesis (32). In the present study it was observed that in the presence of serum type V collagen synthesis was increased, and the increase was 3-to 6-fold in different experiments ( Fig. 6 and Table 11). In order to investigate whether the changes in the synthesis of type I11 and V    * Prepared by coagulating plasma after removal of platelets, followed by complement inactivation by heating at 55 "C for 30 min (34). collagens occur simultaneously, the quantities of these collagens produced at various serum concentrations were measured. As shown in Fig. 7, changes in synthesis of types V and I11 are dependent upon serum concentration, and their quantities vary in a reciprocal manner.
Even though serum contains a mixture of several growth factors, PDGF' is a major mitogenic component for fibroblasts (33). Therefore, we investigated whether PDGF influences synthesis of type I11 and V collagens by fibroblasts. In the first experiment normal and platelet-poor human sera were prepared as described by Rutherford and Ross (34), and the proportion and quantities of collagens synthesized in their presence were measured. From Table 111 it is seen that type V collagen synthesis by cells labeled in medium containing platelet-poor serum was less than half, and type I11 collagen more than double that seen in cultures containing plateletrich serum. In the second experiment, we compared collagens after labeling the cells in the presence of 5.6 and 94.0 ng/ml (concentration of approximately IX and 17X of 10% serum; Ref. 33) of purified human PDGF (gift of Drs. Elaine Raines and Russell Ross, University of Washington, Seattle, WA) added to platelet-poor serum. As expected, type V collagen The abbreviation used is: PDGF, platelet-derived growth factor (human). synthesis increased and type I11 decreased in the presence of the growth factor, and the effect was more pronounced at 94.0 ng/ml than at the lower concentration (Table IV, experiment 1). Essentially the same results were obtained in a different experiment when the effect of PDGF was studied in medium containing no serum (Table IV, experiment 2). DISCUSSION We have shown that human gingival fibroblasts synthesize a collagenous protein comprised of a chains, the electrophoretic mobilities of which are similar to those of a l ( V ) and a2(V) chains. This protein was shown to be type V by its resistance to digestion by mammalian collagenase, by the higher degree of lysine hydroxylation of the a chains, and by their CNBr peptide maps (1, 3, 14, 30).
Pepsin extracts of placenta, gingiva, and other tissues contain a l , a2, and a3 chains (14)(15)(16)27), and it is not clear whether all these proteins are constituents of type V collagen. However, gingival fibroblast type V collagen fractions obtained by ion exchange chromatography and pepsin digests of procollagen (V) preparations and antibody precipitates contained only d ( V ) and a2(V) chains with a ratio of 2:l. Even though a radioactive protein with electrophoretic migration similar to the a 3 chain was present in 0.7-1.2 M NaCl fractions of pepsin digests, and occasionally in fraction I1 of the DEAEcellulose column eluates, it was not detected in the type V collagen or procollagen fractions, or in antibody precipitates. Therefore, we conclude that type V collagen of gingiva has the chain composition ( (~1 )~a 2 , and that the "a3" chain described previously in the gingival tissue (27) is a contaminant and not a constituent of type V. Our studies do not rule out the possibility of a separate (al)(a2)(a3) molecule, even though such a molecule would have chromatographic properties similar to the (al)'a2 species.
A portion of type I collagen always co-eluted with type V in fraction I11 from the DEAE-cellulose columns. Even extensive washing of the columns failed to remove this type I from V (Fig. 1). That the co-eluent was indeed type I was confirmed by electrophoretic migration, a l / a 2 ratio, conversion to TCA and TCB by mammalian collagenase, and finally by the CNBr peptide pattern. However, the lysines of the a1 and a2 chains in fraction I11 were hydroxylated to a lesser extent than the respective chains of fraction I1 (23% and 29% uersus 35% and 50%, respectively). Thus, one reason for the difference in the elution behavior between the two type I species may be differences in their lysine hydroxylation (35) and possibly in glycosylation. The reason for the presence of two type I species which differ in their lysine hydroxylation is not clear, although it may be caused by one or more of the nutritional and growth factors present in serum.
The DEAE-cellulose purified procollagen fraction contained at least seven proteins ranging in molecular size from 132,000 to 370,000; all of these are apparently related to type V collagen, since after digestion by pepsin only a1 and a2 chains of type V collagen were present (Fig. 4g). Proteins q, r, and s migrated faster under nonreducing than reducing conditions, indicating the presence of interchain disulfide loops (18,19,21). Reduction did not induce major changes in pattern except for the possible conversion of the major band near the origin to "r," and the appearance of a minor band between u and v (Fig. 4, e and f). The latter's CNBr peptide pattern was similar to those oft, u, and v (not shown). From this observation and from the CNBr peptide maps, it can be concluded that a portion of the al(V) and a2(V) chains are linked by disulfide bonds (19), and that p, q, r, and s, and t, u, and v represent al(V) and a2(V) chains, respectively, at various processing stages. Band "p" stayed near the origin even after reduction; it had a M , = 370,000 and a CNBr peptide pattern identical to those of q, r, and s, and similar to al(V), and there was no indication of the presence of a2(V) in it. Thus, it is likely that band p represents the nascent unprocessed proal(V) chains, or alternatively, a dimer of 198,000 species, although the nature of their association is not clear. Our results show that the al(V) and a2(V) chains have at least three proteins each at various stages of processing, and the total number of intermediates detected is greater than those reported for chick embryo crop, tendon cells, and A204 rhabdomyosarcoma cells (18,19,21). The sizes of unprocessed proal(V) and proa2(V) have been reported to be 220,000 and 150,000, respectively (21). These proteins were observed in our experiments, but larger molecules with M , = 239,000 and 174,000 were also present, indicating that sizes of prod(V) and proa2(V) elaborated by gingival cells are greater than those reported in other cultures. The M , of prod(V) may be still larger because the reduced procollagen fraction contained a minor protein band migrating with M, = 290,000 ( Fig. 4e) which had a CNBr pattern similar to al(V) (data not shown). Pulse-chase experiments to confirm these conclusions were not possible because of low counts during short pulse periods. Bands s and v (MI = 198,000 and 132,000) appear to be the final products of the a1 and a2 chain processing since smaller molecules were not observed. The value for a1 agrees with 190,000 reported for rhabdomyosarcoma cells, but the gingival a2(V) appears to be smaller (132,000 uersus 150,000;Ref. 21).
Normal gingiva contains about 1% type V collagen (24), and this value parallels the proportion of type V made by cultures of gingival fibroblasts. Previous publications have shown that cell density, serum concentration, and epidermal growth factor affect the amount of type I11 collagen made by fibroblasts (32,36,37). Our present data shows that cell density and PDGF regulate type V production and that there is a reciprocal relationship between synthesis of types I11 and V. PDGF is effective even in the absence of serum, indicating that it may be the major serum factor responsible for modulating the synthesis of these collagens. The mechanism by which collagen synthesis is regulated remains obscure.
In chronically inflamed gingiva, and in certain other diseases such as atherosclerotic aorta wall, the proportion of type V collagen is enhanced greatly (22)(23)(24)(25). One feature these seemingly disparate diseases have in common is the exposure of resident tissue fibroblasts and muscle cells to plasma cornponents such as PDGF which they normally do not encounter (33). Our experiments indicate that factors such as PDGF may play a significant role in modifying collagen composition of tissues under pathological conditions through interaction with cells responsible for collagen synthesis.