Isolation of a new procollagen V chain from chick embryo tendon.

Whole tendons of chick embryos synthesize procollagens V which consist of three pro-alpha chains: pro-alpha 1'(V), pro-alpha 1(V) and pro-alpha 2(V). This report shows that while the pro-alpha 1'(V) chain is similar to the pro-alpha 1(V) chain in many respects, such as similar but not identical peptide maps, it also distinctly differs from it in size and in other ways. The new chain is denoted as pro-alpha 1' to indicate the relationship. We have failed to see conversion of one chain into the other and they are regarded as variants, although we do not know whether they are different transcripts of one gene or products of two closely related genes. The pro-alpha(V) chains are assembled into the disulfide-linked homotrimer (pro-alpha 1'(V))3 and the heterotrimer [(pro-alpha 1'(V)S-S-pro-alpha 2(V))pro-alpha 1(V)] and a smaller amount of [(pro-alpha 1(V)2pro-alpha 2(V)]. The pro-alpha 1'(V) chains are processed similarly to the pro-alpha 1(V) by the initial removal of the presumed carboxyl propeptide yielding p-alpha 1'(V) and then by reduction in the size of the noncollagenous, presumed amino propeptide to yield alpha 1'(V). A size difference between the alpha 1'(V) and alpha 1(V) series of molecules is demonstrated by velocity sedimentation and by electrophoretic mobility of the reduced molecules. This difference is ascribed to a difference in the size of the propeptides because after pepsin digestion the products of both series of molecules are the same size and electrophorese like alpha 1(V)(pepsin). The carboxyl propeptides of pro-alpha 1(V) and pro-alpha 1'(V) are the same size, but the amino propeptide of pro-alpha 1'(V) is smaller than that of pro-alpha 1(V). The amino propeptide of pro-alpha 1'(V) and p-alpha 1'(V) also lacks asparagine-linked complex carbohydrate, which is linked to propeptides of the p-alpha 1(V) and p-alpha 2(V) chains. Differences between the alpha 1(V) and alpha 1'(V) series of molecules remain in material synthesized in the presence of tunicamycin. Primary cultures of tendon cells synthesize procollagen V consisting of the above three chains, but the procollagen V made by cultured tendon sheath synovial cells predominantly contains [(pro-alpha 1(V))2pro-alpha 2(V)].

Liselotte I. Fessler, Nancy Shigaki, and John  The two cartilage collagen chains l a and 2a are probably also closely related to them (1). The a3(V) chain has not been described in chick tissues and peptide mapping clearly shows that it is not a variant of the al(V) chain. In tendon we have now found a procollagen V chain, which we denote as proal'(V) (2). This new chain is similar to pro-al(V) but differs from it at least in size, attachment of complex carbohydrate, and interchain disulfide formation.' Furthermore, we were unable to yield interconversion of pro-al'(V) and pro-al(V) chains and conclude that their differences are not solely due to postribosomal modifications. We suggest that these two polypeptides are either encoded in two closely related genes or are different peptide translation products of a single gene. Fig. 1 summarizes our limited understanding of individual pro-a(V) chains and the p-a(V) and a(V) chains into which they are physiologically converted (2)(3)(4)(5)(6). Electron microscopy of procollagen shows a 300-nm long thread, corresponding to the 3 X 100,000-dalton collagen helix, with a knob at each end (7). The knobs correspond to large, collagenase-resistant peptide regions. Although the amino and carboxyl ends of procollagens V have not been established, comparison with procollagens I, 11, and I11 suggests that the smaller noncollagenous peptides of pro-al(V) and pro-al'(V) are their carboxyl ends: they are internally disulfide-linked glycopeptides of similar size and the physiological cleavage of all of them is inhibited by 50 mM arginine (3-6). These are cut off when pcollagen is formed, and at the same time any disulfide links between component pro-a(V) chains are lost. However, intrachain disulfide links exist in the remaining noncollagenous regions, and correspondingly the electrophoretic mobilities of the denatured, individual p-a(V) chains are slowed on reduction.
To help to understand properties of the p-collagen V chains that were utilized in this study, we describe the pro-a2(V) and p-a2(V) chains, and some noncollagenous "P peptides" which are disulfide-linked to the p-a2(V) chains. The pro-a2(V) chain has only one large noncollagenous peptide region, and this may be at the carboxyl or amino end. Its location is under current investigation. This large noncollagenous region is retained in the p-a2(V) chain, but is cut during conversion to a2(V), with a portion remaining attached to the collagen helical sequence of a2 (V). Surprisingly, all p-a2(V) chains have more than one copy of an approximately 30,000-dalton noncollagenous peptide, named peptide P, attached by disulfide links (4, 5). The nature and source of the P peptides are unknown; they could be derived from the smaller noncollagenous regions of the pro-al(V) and pro-al'(V) chains. The covalent, denatured p-a2(V). P, complex sediments significantly faster than reduced, denatured p-a2(V) chains and also In the preliminary publication (2) we referred to the new procollagen V chain of tendon as pro-a4(V) and its products as p-a4(V) and a4(V). The names have been changed to pro-al'(V), p-al'(V), and al'(V) in the present publication. has a correspondingly slower electrophoretic migration (5).
The P peptides are lost during physiological, proteolytic conversion to a(V) collagens. All the chains of the final, a(V) collagens retain significant noncollagenous peptide regions. Throughout these processing steps the new pro-oll'(V) chain and its derivatives maintain their identities and are similar to, but different from, pro-d(V) and its derivatives. Pepsin digestion, as used by others to solubilize collagens V from tissues, removes the noncollagenous peptide regions present in a(V) collagen chains to give correspondingly smaller a(V)(pepsin) chains. The electrophoretic mobilities of al'(V)(pepsin) and al(V)(pepsin) are identical, and the heterogeneity of chains which we describe here would not be apparent in materials solubilized by pepsin treatment.
A peculiar feature of procollagens V is that, in contrast to procollagens I, 11, 111, and IV, only a small proportion of the molecules have all three component chains disulfide-linked to each other. Most chains are linked pair-wise, or not at all (4, 6), and the homotrimer ( p r o -~d ( V ) )~ of hamster lung cells lacks any interchain disulfide links (6). The present work originated in a re-examination of this problem of interchain disulfide linkage. We had found earlier that the degree of disulfide linkage could not be changed significantly by prolonging the residence time of chains within cells (4). A relatively higher proportion of disulfide-linked procollagen V molecules in chick embryo tendon was now found to be due to the new pro-al'(V) chain. We now studied procollagen V synthesis by whole tendon and in two sets of cell cultures separately derived from it: one mostly consisting of the major, tendon matrix fibroblasts and the other of synovial cells which form the tendon covering and sheath (16,17). Chick embryo crop and blood vessels, which essentially fail to make prod ' ( V ) chains, served as controls.

MATERIALS AND METHODS
Unless stated otherwise, the details of methods were the same as those which we described before (3-6).
Isolation of Type V Molecules from Tendons-Tendons from 18day chick embryos were excised and preincubated in Dulbecco's modified Eagle's medium (DMEM'), from which those amino acids were omitted which were later added in radioactive form, and supplemented with 100 pg/ml of ascorbic acid, 64 pg/ml of B-aminopropionitrile, and 0.02 M Hepes, pH 7.4. The tissue was labeled either with 50 pCi/ml of [5'-3H]proline and [4,5'-3H]le~cine (ICN Corp.)

or [35S]
methionine (Amersham Corp.). In a pulse-chase experiment the labeled tendons were rinsed and incubated in DMEM supplemented as above plus 250 pg/ml of cycloheximide (Sigma) and with 0.1 mg/ml of proline and leucine and 1% fetal calf serum. The tissues were frozen in liquid Nz and extracted at 0 "C with 1 M NaCl buffer containing protease inhibitors, and the extract was chromatographed on a DEAE-cellulose column as described (3,5). The native proteins in the concentrated "low salt" or "high salt" fractions were further purified by velocity sedimentation in a Beckman SW 60 rotor at 4 "C on a 5-20% sucrose gradient containing 2 M urea, 0.5 M NaCl, 0.05 M Tris-HC1, pH 7.5, and 0.1% Triton X-100 (3). The native procollagen V fractions obtained by sedimentation or the proteins of the high salt fraction were denatured in 6 M urea at 40 "C for 30 min and sedimented on a buffered 5-20% sucrose gradient containing 6 M urea at 22 "C in a Beckman SW 60 rotor. Fractions were collected and electrophoresedeither nonreduced or reduced on SDS-polyacrylamide gels (18) and fluorographed (19). Densitometric measurements were made on an Optronics 1000 densitometer coupled to a VAX 11/780 computer. Sedimentation coefficients were calculated as described (20).
The type V molecules were identified by comparison with known molecules isolated from chick blood vessels (5). The samples in native form were digested with 100 pg/ml of pepsin at 4 "C for 18 h in 0.2 M acetic acid and electrophoresed on SDS-4.5% polyacrylamide gels. Specific bands identified on SDS-4% polyacrylamide gels by fluorography were cut out and also digested with V8 protease (Sigma) or clostripain (Millipore), and the resulting peptides were electrophoresed on SDS-10-15% polyacrylamide gels as described (3).
Native type V molecules were immunoprecipitated with antibodies prepared in rabbits against type V collagen extracted by pepsin digestion of decapitated 12-day-old chick embryos and purified by salt precipitation. The immune complex was formed at 4 "C for 18 h and precipitated with Pansorbin (Sigma). The precipitate was sedimented through 1 M sucrose, 150 mM NaCI, 50 mM Tris-HC1, pH 7.5, 20 mM EDTA, 0.1% Triton X-100, and the precipitate was washed with this buffer minus sucrose. The proteins were solubilized at 100 'C in 1% SDS and 2 M urea. Aliquots of the proteins in the immune precipitate and supernatant were reduced and electrophoresed on a SDS-4.5% polyacrylamide gel and fluorographed.
Separation of Nonco1lugenou.s Peptides-Type V molecules separated electrophoretically on SDS-4.5% polyacrylamide gels were identified by fluorography and the individual bands were cut out, swelled with buffer (150 mM NaCl, 50 mM Tris-HC1, pH 7.6, 7 mM CaClZ, 0.1% Triton X-100) and bacterial collagenase (Worthington CLSPA) which had been further purified (21,22), and incubated at 37 "C for 3 h. Then 1% SDS, 10 mM dithiothreitol, and 10% glycerol were added and the gel pieces were heated at 100 "C for 3 min and electrophoresed on a SDS-10% polyacrylamide gel (4). The noncollagenous P peptides attached to p-a2(V). P, were identified electrophoretically after reduction of denatured p-a2(V). P, isolated by velocity sedimentation. This fraction was also digested with bacterial collagenase and the propeptide. P, fragment was identified electrophoretically.
Tendon Cell and Synovial Cell Cultures-The cells were isolated as ' The abbreviations used are: DMEM, Dulbecco's modified Eagle's medium; Hepes, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; SDS, sodium dodecyl sulfate. described (16) from 17-day chick embryo tendons. These were incubated in phosphate-buffered saline containing 0.1% bacterial collagenase (Worthington CLS) and 0.25% trypsin (Difco) and 0.1% glucose for 20 min at 37 "C to release synovial cells. The residual tendons were rinsed with buffered saline and incubated as above for 2 h a t 37 "C to release tendon cells. The cell suspensions were filtered through Nytex gauze, and the cells were washed 3 times in DMEM containing 0.1% fetal calf serum, 25 pg/ml of ascorbic acid, and 10 mM Hepes buffer, pH 7.4. The tendon cells were allowed to attach to tissue culture plates in this medium for 2 h and the nonattached cells were removed. The cells were then kept for 1 h either with fresh DMEM as above or in DMEM containing 0.22 pg/ml of tunicamycin AI, kindly donated by Dr. D. Duskin (23). The cultures were then labeled for 12 h with [5'-3H]proline and [4,5'-3H]leucine (each 25 pCi/ml). Some tendon cells were cultured for 3 days until the cells were confluent and labeled as above. The synovial cells were suspended in DMEM containing 5% fetal calf serum. The cells were allowed to attach for 12 h, fresh medium was added, and the cultures were grown to near confluency in 3 days. On the third day the cultures received 25 pg/ml of ascorbate and on the fourth day the cells were labeled for 18 h in medium containing [5'-3H]proline, [4,5'-3H]leucine (25 pCi/ml each), and 25 pg/ml of ascorbate, 64 pg/ml of @-aminopropionitrile, 0.1% fetal calf serum, and 5 mM Hepes buffer, pH 7.4.
The type V molecules were isolated from the medium and cell layer as described (6). The DEAE-cellulose chromatography was by stepwise elution of the effluent fraction and the 0.15 M NaCl and 1 M NaCl eluates. The products were examined electrophoretically.

RESULTS
Characterization of Tendon Procollagen V-Chick embryo tendons were incubated in DMEM supplemented with 100 pg/ml of ascorbic acid and 64 pg/ml of P-aminopropionitrile in the presence of [3H]leucine and [3H]proline. After a 4-h continuous labeling period the tissue was extracted and the extract was chromatographed on DEAE-cellulose. The effluent fraction contained collagen I and its precursors, the procollagens V eluted predominantly in the low salt fractions and the processed intermediates, p-collagens V, together with the final collagen V were eluted with a high salt buffer (Fig.  2).
The tendon procollagens V were further partially purified in native form by velocity sedimentation. For comparison, procollagen V from blood vessels, consisting of pro-al(V) and pro-a2(V) chains, was sedimented at the same time. The sedimentation coefficients of tendon and blood vessel procollagens V were the same. Electrophoretic analyses of the sedimented, reduced samples of tendon procollagens V showed that the three component chains of this procollagen are present in the same proportion in each fraction (Fig. 3). These chains are pro-al(V) and pro-a2(V), and a chain with slightly different electrophoretic mobility, and we refer to this as proal'(V). Electrophoretic separation of nonreduced aliquots of the same fractions indicated that these procollagens V were present as disulfide-linked trimers, dimers, and pro-a chains (not shown). These were separated after denaturation of the procollagens V by velocity sedimentation, as is shown in Fig.  4. The disulfide-linked trimer fractions consisted almost exclusively of pro-al'(V) chains, the dimers were composed of equal amounts of pro-al'(V) and pro-aB(V), and the monomers which were not disulfide-linked were predominantly proa l ( V ) chains and a small amount of pro-a2(V) chains. From this we conclude that pro-al'(V) exists as a homotrimer [(proal'(V))-s-s-pro-al'(V)-S-S-pro-al'(V)] and as a heterotrimer consisting of 3 different chains [(pro-al'(V)-S-S-pro-aB(V))pro-al(V)]. In addition, smaller amounts of [(procul(V)),pro-a2(V)] may also be present.
Processing of Tendon Procollagen V-When tendons were labeled for 30 min the only radioactive type V chains were the three pro-a(V) chains, and they were labeled in about the same proportions as shown in Fig. 3. However, assembly into Leu. The tissue extract was chromatographed on DEAE-cellulose, and aliquots of successive fractions of the concentrated "low salt fractions" and "high salt fractions" were reduced and electrophoresedon a SDS-5% polyacrylamide gel. The fluorogram is shown. Marker type I and V molecules were isolated from blood vessels and electrophoresed on the same gel. The mobilities of the blood vessel type V molecules were the same as those of their tendon counterparts. The positions of pro-al(1) and pro-a2 (1)  Leu for 2 h. The tendon extract was chromatographed on DEAE-cellulose and the concentrated low salt fraction containing procollagen V was sedimented on a buffered sucrose gradient containing 2 M urea at 4 "C at 56,000 rpm for 22 h. Aliquots of the reduced fractions were electrophoresed on a SDS-4.5% polyacrylamide gel and the fluorogram is shown. Pro-al(V) and pro-a2(V) had the same electrophoretic mobility as marker molecules isolated from chick blood vessels (5). The direction of sedimentation is from right to left. disulfide-linked trimers and dimers was incomplete and about half of the chains were present as non-disulfide-linked proal(V), pro-al'(V), and pro-a2(V). It was not investigated whether these non-disulfide-linked pro-a chains existed as single pro-a chains or whether they were present as triple helically folded molecules. In a pulse-chase experiment, tendons were labeled for 60 min and then transferred to medium containing cycloheximide, and the incubation was continued. Fig. 5 shows that only procollagen V was present at 30 min, but by 60 min of incubation some conversion was seen and thereafter tendon procollagen V was converted first to pcollagen V and then to collagen V. The rates of processing of pro-al'(V) and pro-al(V) chains were similar.  Fig. 3, was denatured in 6 M urea a t 40 "C for 30 min and sedimented on a buffered sucrose gradient containing 6 M urea a t 56,000 rpm a t 27 'C for 19 h. Aliquots were counted and aliquots of fractions 6 + 7, 10 + 11, 13 + 14, and 15 were electrophoresed, either nonreduced or reduced, on a SDS-4.5% polyacrylamide gel, and the fluorogram is shown. As a control, blood vessel procollagen V was denatured and sedimented at the same time and the same fractions were electrophoresed (not shown). These trimers and dimers consisted of pro-al(V) and pro-a2(V), and monomeric pro-al(V) and pro-a2(V) were present, in agreement with our previous publication (5). a1 '(V)-The native p-collagens V sedimented slightly faster than the native collagens V on a sucrose gradient, and in Fig.  6A the electrophoretic analyses of the sedimented, reduced fractions are shown. The p-collagens V consisted of p-al(V), p-al'(V), and p-a2(V), and the collagens V consisted of al(V), al'(V), and aB(V). The electrophoretic mobilities of reduced p-a2(V) and al'(V) are closely similar, but the nonreduced al'(V) chain is clearly distinguishable from the p-aS-P, molecules, which have a much slower electrophoretic mobility like p-al(V) (not shown) (4,5). When the components of this high salt fraction were denatured in 6 M urea at 40 "C for 30 min and sedimented, distinct differences in the sedimentation coefficients of these chains could be seen in Fig. 6B and Table   I. Thus, differences in size between the p-al(V) and p-al' (V) and al(V) and al'(V) molecules are indicated both by electrophoretic mobilities and sedimentation coefficients. The P peptides, which are disulfide-linked to the p-aP(V) chain, were also present in tendon p-collagen V and the denatured p-a2(V). P, sedimented ahead of all the other chains.-Following pepsin digestion, separated tendon procollagen V and p-collagen V plus collagen V yielded only two electrophoretic bands which were the same as the al(V)pepsin and a2(V)pepsin isolated from blood vessels or crop. We conclude that the helical portions of the al'(V) series of molecules are identical or very similar to the al(V)pepsin chains.
Antibodies prepared against purified collagen V isolated from pepsin digests of decapitated chick embryos precipitated all of the tendon p-collagen V and collagen V, including all of the al'(V) series of chains (Fig. 7). The results of Fig. 4 show that a significant portion of the total al'(V) chains in this material exist as homotrimers. Therefore these homotrimers were precipitated by the antibodies, as well as a l ' ( V ) chains in heterotrimeric combination with a l ( V ) and a2(V) chains. Therefore, this new collagen chain is included in the type V group by immunological reaction with antibodies that were made against antigens which met previously set criteria for collagen type V. Comparisons of the peptide maps obtained by protease digestion of [3H]Pro-and [3H]Leu-labeled pro-al(V) and proal'(V) and of p-al(V) and p-al'(V) show marked similarities, but some distinct differences can be seen in Fig. 8. The peptide pattern of the a2(V) series is clearly different.
These findings may be interpreted that pro-al(V) and proal'(V) and their products differ only in their noncollagenous segments. Therefore, we separated the [35S]methionine-labeled or [3H]Pro-and [3H]Leu-labeled, reduced chains electrophoretically and digested the molecules with bacterial collagenase. The remaining noncollagenous, reduced peptides were then electrophoresed on a SDS-10% polyacrylamide gel together with globular molecular weight standards. Nonre- TABLE I Sedimentation coefficients of denatured p-collagen V and collagen V molecules The denatured components of the high salt fraction, obtained by DEAE-cellulose chromatography, were sedimented at 56,000 rpm for 28 h at 27 "C, 62 fractions were collected, electrophoresed, and fluorographed, the peak positions for each molecule were determined, and the sedimentation coefficients were calculated as described (20) Fig. 6B). This fraction was then either reduced to release and identify the P peptides, or the material was digested first with bacterial collagenase and then the products were analyzed electrophoretically before and after reduction on SDS-10% polyacrylamide gels. Table I1 summarizes the findings and the diagram in Fig. 1 depicts the differences between the three series of molecules. Since these noncollagenous peptides are glycosylated and sulfated (24), the relative molecular sizes shown are only approximate. The noncollagenous peptides, presumed to be the carboxyl propeptides, cleaved from pro-al(V) and pro-al'(V) during the first step of processing, have the same electrophoretic mobilities. The peptides presumed to be the amino propeptides of proa l ( V ) and p-al(V) are larger than the corresponding peptides of pro-al'(V) and p-al'(V). In the next processing step these are cleaved and the remaining noncollagenous peptide of the cul(V) chain is larger than that of the al'(V) chain.
Although the data given above suggest distinct differences in size of the complete chains, post-translational addition of complex carbohydrate may affect the electrophoretic mobilities of the noncollagenous peptides markedly. The inhibition of this modification of procollagen I with tunicamycin has been analyzed in detail by Duskin and co-workers (23, 25). Procollagen I synthesis was not affected markedly in the cell cultures examined, but processing of PC-collagen I to collagen I was decreased, presumably due to a deficiency of the carboxyl-procollagen peptidase (25). Using freshly isolated tendon cells in culture, which were labeled with [3H]Pro and [3H] Leu for 12 h, we found that procollagen I was almost completely converted to an equal amount of PC-collagen I and collagen I in the culture medium in control cultures. In the presence of tunicamycin AI (0.22 pglml), procollagen I processing was reduced and PC-collagen I accumulated in the culture medium while only a small amount of collagen I was present. The electrophoretic mobilities of pro-al(1) and pro-a2(I) and PC-d(1) and pC-a2(I) were increased when complex carbohydrate addition was prevented by tunicamycin. The synthesis of type V procollagens and the processing to pcollagen V was nearly the same in the presence or absence of tunicamycin. Under the specific culture conditions no processing to final collagen V occurred. The p-al(V) and p-a2(V) chains showed altered electrophoretic mobilities in cultures treated with tunicamycin, but no change was seen in the mobilities of the p-al'(V). chains as is shown in Fig. 9. However, the mobility of the pro-al'(V) chains was altered. Comparisons of the electrophoretic mobilities of the noncollagenous peptides of these molecules from control and tunicamycin-treated cultures are given in Table 111 mobilities when addition of complex carbohydrate was prevented. The "carboxyl propeptide" of pro-d'(V), on the other hand, contains complex carbohydrate as evidenced-by increased mobility of this peptide derived from pro-al'(V) of a tunicamycin-treated culture.
Site of Synthesis of the New Type V Collagens-Tendon cells consist primarily of tendon fibroblasts and of an outer sheath of tendon synovial cells. These two cell types can be separated by differential attachment to tissue culture plates (16). The tendon cells attach within 2 h in the presence of 0.1% fetal calf serum, and in the presence of 25 pg/ml of  ascorbic acid these cells synthesize collagen at the same rate as in the whole tissue (26). During the first 24 h of culture gene expression does not appear to be altered. These tendon fibroblasts make types I and V collagen, but no type I11 collagen, while the sheath cells make type I11 collagen as well (16,17). Tendon fibroblasts maintained in tissue culture for a few days initiate type I11 procollagen synthesis (27). Under the conditions used in our experiments the tendon fibroblasts synthesized types I and V procollagens, but no type I11 collagen was detectable when the medium proteins were digested with pepsin and the nonreduced molecules were electrophoresed. These tendon fibroblasts released most of the type V The reduced type V molecules were electrophoresed as in Fig. 9, the identified bands were cut out and treated with bacterial collagenase, and the products were electrophoresed on a SDS-12.5% polyacrylamide gel. The mobilities in cm are given. The p-a2 noncollagenous peptide also showed increased mobility when complex carbohydrate addition was inhibited with tunicamycin (control = 3.15 cm; + tunicamvcin = 3.5 cm).

Large (amino) peptide
Small ( molecules into the medium and processing to p-collagen V occurred (Fig. 9). Tendon cell cultures grown for 3 days and then labeled also synthesized pro-al(V), pro-a2(V), and proal'(V) chains.
Tendon synovial cells are readily released from whole tendons with bacterial collagenase and trypsin, and these cells attach to tissue culture plates only in medium containing 5% fetal calf serum. These cells were cultured for 3 days to obtain sufficient cells for labeling. These cells also synthesized type V molecules. Most of the type V molecules were deposited in the cell layer, some p-al(V) and p-a2(V) and mostly al(V) and a2(V) were present, and only a trace of p-al'(V) and al'(V) were seen. The latter could have been made by contaminating tendon cells. The synovial cells also made type I11 procollagen (as confirmed by pepsin digestion of the medium proteins). This could be a specific cell type marker expressed by these cells, as described (16,17), but the possibility is not ruled out that during 3 days of cell culture this gene is turned on in response to the in vitro culture conditions. We conclude that the pro-al' (V) containing molecules are synthesized predominantly by the tendon matrix fibroblasts and probably to a lesser extent or not at all by the surrounding tendon sheath synovial fibroblasts.

DISCUSSION
There are at least three differences between the pro-al(V) and the pro-al'(V) chains: 1) the mass of the noncollagenous region near the presumed amino end of the collagen helical portion, 2) attachment of complex carbohydrate to the presumed amino noncollagenous region, and 3) the propensity to form disulfide links with other members of the same triplechained molecule.
The masses of the pro-al'(V), p-al'(V), and al'(V) chains are each smaller than those of, respectively, the pro-al(V), pal(V), and a l ( V ) chains. The different electrophoretic behavior of collagenous and noncollagenous peptides and the influence of glycosylation only allow nominal estimation of molecular masses (Fig. 1). However, the same sequence of relative molecular sizes is indicated by the electrophoretic mobilities and the sedimentation velocities of the denatured chains. This same sequence of relative electrophoretic mobilities is maintained when attachment of complex carbohydrate is prevented by tunicamycin, even though some individual electrophoretic mobilities are changed. Therefore, there are underlying differences which are independent of attachment of complex carbohydrate. After digestion with bacterial collagenase, the equivalent differences of electrophoretic mobilities are shown by the separated, presumed amino noncollagenous peptides. In contrast, the carboxyl propeptides of pro-al(V) and proal'(V) have the same mobility. As al(V)(pepsin) and al'(V) (pepsin) also have identical electrophoretic mobilities, we conclude that one key difference between the pro-d(V) and pro-al'(V) chains is in the masses of those parts of the presumed amino noncollagenous peptide regions which remain attached in the al(V) and al'(V) chains, i.e. the masses of their telopeptides. Differences in the peptide maps produced by protease digestion are consistent with this view. In addition, our unpublished experiments3 show that some tyrosine residues of these noncollagenous peptides of p-al(V), al(V), p-al'(V), and al'(V) are sulfated. When [35S]sulfatelabeled molecules were digested with Staphylococcus aureus V8 protease, the patterns of radioactive peptides obtained from the p-al(V) and al(V) series were very different from those of the p-al'(V) and al'(V) chains. This provides further evidence that the noncollagenous segments of al(V) and al'(V) are different.
Purified tunicamycin caused closely similar modulations of procollagen I synthesis and processing by tendon fibroblasts to those described by Duskin and co-workers (23, 25), who verified the action of the drug by mannose incorporation measurements. As these cells simultaneously made procollagens V, the effect of tunicamycin on the electrophoretic mobilities of the procollagen V chains can be interpreted as affecting attachment of complex carbohydrate. Tunicamycin changed the electrophoretic mobility of pro-al'(V) but not of p-al'(V) chains (Fig. 9). We conclude both that complex carbohydrate is attached to the carboxyl propeptide and that, necessarily, the drug had equal access to the biosynthetic paths of pro-al(V) and pro-al'(V). Interestingly, complex carbohydrate is attached to the noncollagenous peptides of both p-al(V) and p-a2(V) chains (Table 111). At present we do not know whether complex carbohydrate is attached to the noncollagenous regions of al(V) and a2(V).
Our conclusion that most of the pro-al'(V) chains are disulfide-linked to neighbors within the three-chained molecules, in contrast to pro-al(V) chains which are mostly unlinked, is based on a range of experiments of which only one is illustrated in Fig. 4. The sedimentation velocity of these materials under native conditions indicated that they consisted of individual three-chained molecules, and, therefore, the disulfide-linked dimers and trimers represent their constituents. Fig. 4 strongly indicates the existence of heterotrimeric molecules which contain all three chains. Denaturation yielded mostly individual pro-al(V) chains in the necessary amount to provide the third chain required by the disulfidelinked pairs of pro-al'(V) and pro-a2(V) chains. When electrophoretograms of denatured procollagen V are interpreted, it is essential to be aware of the small decreases in electrophoretic mobility of individual chains that accompany reduction of the disulfide-links which hold folds within one chain together. As a consequence, reduced pro-al'(V) chains have almost the same mobility as nonreduced pro-al(V) chains. Complete reduction before analysis avoids this problem.
We have not succeeded in separating the central, collagenous regions of the al(V) and al'(V) chains after pepsin digestion of the mixture of native, helical molecules. Cyanogen bromide digest maps of unfractionated, labeled collagen V(pepsin) from tendon and from blood vessels, which do not contain the al'(V) chain, were very similar. Because of the possibilities of incomplete digestion and other effects, we excluded these experiments on mixtures of collagen V(pepsin) chains from the present considerations. However, together with the protease peptide maps, the evidence clearly indicates that the collagenous portion of the pro-al'(V) chain and its L. I. Fessler, S. Brash, S. Chapin, and J. H. Fessler, unpublished experiments. derivatives is similar to that of the pro-al(V) and unlike the a3(V)(pepsin) derived from human placenta. The a3(V) chain has not been described in chick tissues. The cyanogen bromide peptide patterns of human al(V)(pepsin) and a3(V)pepsin are quite different. The new chain we describe in chick could not be chick a3(V) without inordinately extensive changes in the occurrence of methionine residues between human and chick.
The similarities of the pro-al(V) and pro-al'(V) chains and their derivatives raise the possibility that they are different, secondary modifications of a single polypeptide. We looked for evidence towards this, failed to find it, and conclude that they represent different products of ribosomal peptide synthesis. The pulse-chase experiments gave no indication of interconversion and showed similar and simultaneous processing of pro-al(V) and pro-al'(V) chains. Our unpublished experiments3 on the sulfation of tyrosine residues in prod ( V ) and pro-al'(V) indicate that the relative timing of sulfation and cleavage to p-a(V) chains differs in the two series of chains. The tendon matrix cells are representative of the major tendon fibroblasts, and under these conditions of culture their collagen biosynthesis closely follows that of intact tendon (26). Similarly to intact tendon, these cultures synthesize both pro-al(V) and pro-al'(V). This makes it unlikely that differences between the chains are due to any effect of the local tissue environment on secondary modifications of a single precursor polypeptide. The existence of the heterotrimer molecules [(pro-al'(V).S-S.pro-a2(V))(proal(V))] necessitates that both the pro-al'(V) and the prod ( V ) chains are synthesized simultaneously within the same individual cells and follow a common intracytoplasmic path of synthesis and transport.
The hypothesis of a single polypeptide precursor for both chains would require at least the following secondary modifications to some chains, but not to others: 1) attachment of complex carbohydrate to the "amino" noncollagenous region of the pro-al(V) chain, but not to the pro-al'(V) chain (even though complex carbohydrate is attached to the carboxyl propeptides of both chains); 2) additional increase of mass of this amino noncollagenous region of pro-al(V), separately from attachment of complex carbohydrate, by some unspecified process (the complementary alternative is removing mass from within the common precursor at a location at least 40,000 daltons from one end to produce the pro-al'(V) chain.); or 3) by some unspecified process assuring that interchain disulfide links are made by the pro-al'(V) chains at a location at least 100,000 daltons away, beyond the other end of the central collagen peptide sequence, from the first two modifications. It seems improbable that three such effects would all be exercised on one copy of a polypeptide and not on another identical one passing along the same intracytoplasmic path. We suggest that there must be differences in the precursor polypeptides themselves. These differences could be due to different genes, to different transcripts of one gene, or to different splicing combinations of one set of transcripts into two different mRNAs.
The pro-al'(V) chains may be a specialization of the tendon matrix cells. They are not made by the smooth muscle cells of crop and blood vessels and probably also not by synovial tendon cells. The role of the substantial noncollagenous regions which seem to remain in the a(V) chains after physiological processing is unknown, and it would be interesting if there were tissue-specific modification of these regions.