Impaired conversion of procollagen to collagen by fibroblasts and bone treated with tunicamycin, an inhibitor of protein glycosylation.

Tunicamycin, an inhibitor of lipid carrier-dependent protein glycosylation, was used in studies of procollagen synthesis, secretion, and proteolytic modification by chick cranial bones in organ culture and by chick tendon fibroblasts in tissue culture. Tunicamycin inhibited the incorporation of D-[2-3H]mannose into procollagen by greater than 90% whereas general protein synthesis and collagen synthesis were decreased by only 10 to 20%. The procollagen synthesized in the presence of tunicamycin was secreted normally and its immunological characteristics, as detected by an antiserum to the intact protein, were unchanged. However, tunicamycin caused an accumulation of biosynthetic intermediates containing disulfide-bonded COOH-terminal extensions in both cell and bone culture. Cleavage of NH2-terminal extensions was not detectably impaired. These findings provide additional support for the involvement of more than one enzyme in the limited proteolytic conversion of procollagen to collagen.

Tunicamycin, an inhibitor of lipid carrier-dependent protein glycosylation, was used in studies of procollagen synthesis, secretion, and proteolytic modification by chick cranial bones in organ culture and by chick tendon fibroblasts in tissue culture. Tunicamycin inhibited the incorporation of o-[2-"Hlmannose into procollagen by greater than 90% whereas general protein synthesis and collagen synthesis were decreased by only 10 to 20%. The procollagen synthesized in the presence of tunicamycin was secreted normally and its immunological characteristics, as detected by an antiserum to the intact protein, were unchanged. However, tunicamycin caused an accumulation of biosynthetic intermediates containing disulfide-bonded COOH-terminal extensions in both cell and bone culture. Cleavage of NH,terminal extensions was not detectably impaired. These findings provide additional support for the involvement of more than one enzyme in the limited proteolytic conversion of procollagen to collagen.
The identification of a biosynthetic precursor of collagen, procollagen (see Refs. 2 and 3 for reviews), has been followed by studies directed toward an understanding of the molecular structure of the precursor, its function, and the process by which it is converted to collagen. Intact type I procollagen contains nontriple helical domains at both the NHB-and COOHterminal ends of the molecule while interchain disulfide bonds, characteristic of the precursor, are confined to the larger COOH-terminal domain (4)(5)(6).
In addition, procollagen contains carbohydrate moieties, at least some of which have been shown to be located in the COOH-terminal nontriple helical domain (7-9). The biological functions of the nontriple helical domains are not well understood. Recent data support the postulate (10) that the COOH-terminal end serves to initiate chain association and subsequent triple helix formation (11). Previously, studies of the rate of polymerization and cross-link formation using dermatosparactic calf skin collagen had indicated that these processes were retarded unless NH,-terminal extensions were excised (12).
Since peptide sequences from both the NH,-and COOH-terminal regions must be removed during conversion of procollagen to collagen, it seems likely that more than cne enzymatic activity is involved. When labeling was done with a radioactive amino acid, that amino acid was omitted from the labeling medium; when a radioactive sugar was used, the glucose concentration in the labeling medium was reduced to 50 mglliter and the medium was supplemented with 1.1 g/liter of sodium pyruvate as an energy source (231. At the end of the labeling period protein synthesis was stopped by rapidly cooling the plates or bottles to 0". All subsequent treatments were performed at 0", and the medium was treated separately from the tissue or cell layer. The media were decanted and cell monolayers and bones were washed three times with phosphate-buffered saline. The cells were homogenized using a glass Dual1 tissue grinder (Kontes Glass Co.), and the bones were homogenized using a Polytron tissue grinder model SDT (Tekman Co.). Total protein was determined by a modification (24) of the Lowry method using bovine serum albumin as a standard.
Proteins of both the medium and the tissue were precipitated by addition of 10% trichloroacetic acid and collected by centrifugation at 38,000 x P at 0" for 15 min: the nellets were washed three times with 5% trichloroacetic acid and aliquots were assayed for radioactivity in a Beckman LS-230 liquid spectrometer using Aquasol (New England Nuclear) as a scintillant to determine total protein synthesis.
Collagen in the medium and cell layer of cultured tendon libroblasts was measured by the method of Peterkofsky and Diegelmann (25) with minor modifications described by Kruse and Bornstein (23 and procollagen-derived intermediates was performed according to a method described previously (23,25). A l-ml sample of 13H1proline-labeled cell culture medium, prepared as described above, was dialyzed against 0.15 M NaCl, 0.05 M TrislHCl (pH 7.5), 5 mM CaCl,, and made 10 mM in N-ethylmaleimide. The dialyzed sample was incubated for 2 h at 37" with 50 wg of purified collagenase (25). The reaction was terminated by addition of glacial acetic acid and heating to 65" for 10 min; aliquots were then removed for sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
Analytical Sodium Dodecyl Sulfate-Polyacrylamide Gel Electro-Dhoresis -Gel electronhoresis was carried out on 5% acrvlamide gels 713 x 0.6 cm) for procollagen and procollagen-derived intermediates (13) and 7.5% acrylamide gels (13 x 0.6 cm) for the procollagenderived COOH-terminal peptides from culture medium of bones-(7). The gel electrophoresis system was used as described by Goldberg et al. (27) except that an equimolar concentration of N,N'-diallyltartardiamide was substituted for methylenebisacrylamide as a crosslinking agent (28). Dansvlated collagen al chains, R,, components, and al(I)-CB7 were used as internal standards. Following electrophoresis the gels were frozen and sliced into l-mm slices on a Mickle gel slicer. Slices were dissolved in 0.5 ml of 2% periodic acid and the radioactivity measured with 10 ml of Triton X-114 scintillation mixture in a Beckman LS-230 scintillation counter. for 2 h, as described above.

RESULTS
Effect of Tunicamycin on Protein Synthesis-The effect of tunicamycin was tested in chick embryo cranial bones and chick embryo tendon fibroblasts. In both systems the incorporation of amino acids into macromolecules was decreased by only 10 to 20% in the presence of tunicamycin ( Table I) was inhibited by about 80% in tendon fibroblasts. Incorporation of D-[2-3H]mannose, which loses the label during enzymatic conversion to fructose and to all other sugars except fucose, was inhibited almost completely (>95%) in the presence of tunicamycin (Table I). Thus, in agreement with previous work (221, tunicamycin inhibits the incorporation of sugars into glycoproteins without markedly inhibiting protein synthesis. Effect of Tunicamycin on Collagen Synthesis and Secretion -The effect of tunicamycin on the synthesis and secretion of collagenous proteins by chick embryo leg tendon fibroblasts was tested as follows. Following a preincubation period of 6 h in the presence or absence of 0.5 to 5.0 yg of tunicamycin/ml of labeling medium, cells were further incubated for 24 h with ["Hlproline in medium containing these different concentrations of tunicamycin, and the medium and cell layer were analyzed for collagenous protein at the end of this period. Tunicamycin inhibited [3Hlproline incorporation into total protein and into collagenous proteins by about 20%, both in the medium and the cell fraction, but the distribution between medium and cell layer was unchanged (Table II). The drug therefore does not specifically inhibit secretion of procollagen. Long Term Double Labeling OfFibroblasts-Although synthesis and secretion of procollagen were not significantly al-  in the presence or absence of 2 pglml of tunicamycin. Proteins from the medium were isolated and separated by polyacrylamide gel electrophoresis.
When the media of control (Fig. 1, A and   B) and tunicamycin-treated fibroblasts (Fig. 1, C and D) were examined a marked inhibition of incorporation of radioactive mannose into proteins was observed. The only fraction in which such inhibition was not observed was the highest molecular weight component that appeared near the top of the gel. This component probably contained protein-polysaccharides. Procollagen, disulfide-bonded intermediates which migrated 15 to 25 mm from the top of unreduced gels, and (Y chains which migrated between Dns-p,, and Dns-~ul markers were collagenase-sensitive. After reduction, procollagen and its intermediates also migrated between the above Dns markers (Fig. 1, B and D).
In medium from tunicamycin-treated cells (Fig. 1, C and D) the distribution of the ['4C]proline label differed markedly from that of control cells. The ratio, calculated from 6 experi-ments, of 14C radioactivity in procollagen plus intermediates to 14C radioactivity in uz chains was 0.93 +-0.2 for control cultures and 2.52 + 0.5 for tunicamycin-treated cultures. The accumulation of procollagen and intermediates in the presence of tunicamycin indicated an inhibition of conversion of procollagen to collagen. It was interesting to note that a collagenaseresistant proline-and marmose-labeled peak, corresponding in position of migration to the elastic fiber microfibrillar protein isolated by Muir et at. (35) was absent from the tunicamycintreated medium. This glycoprotein, in the reduced form, is indicated by the black bar in Fig. lB.
When the collagenous proteins in the cell layer were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis most of the [14C]proline label appeared in (~1 and cu2 chains. There was no accumulation of intracellular procollagen in either control or tunicamycin-treated cultures (data not shown).
To further characterize the procollagen produced in the presence of tunicamycin, medium from flbroblasts labeled with ['4C]proline and ["Hlmannose was precipitated with antibodies directed against determinants in COOH-and NH,terminal extensions of procollagen. Approximately 38% of the total [14C]proline in the control medium and 34% of the total [14C]proline in the tunicamycin-treated medium was precipitated by the antiserum (Fig. 2A). The fraction precipitated from the medium of bone organ cultures labeled with [3H]tryptophan was also the same for control and tunicamytin-treated cultures (Fig. 2B). These findings indicate that the antigenicity of procollagen lacking carbohydrate was unchanged, as detected by the antisera used, and suggest that carbohydrate side chains may not be the major antigenic determinants in procollagen. It should be noted that immune precipitation of collagenous proteins does not distinguish between procollagen and the several intermediates in conversion of procollagen to collagen since all of these proteins contain COOH-terminal or NH,-terminal extensions, or both. After denaturation and electrophoresis, a different spectrum of intermediates may be revealed by the different relative proportions of LY chains and disulfide-bonded forms, as seen in Fig. 1  effect in conversion of procollagen to collagen observed with tunicamycin, a pulse-chase experiment was performed with fibroblasts in culture. A pulse period of 45 min with [3H]proline was followed by an 18-h chase with medium containing an excess of unlabeled proline, The collagenous proteins were separated by slab gel sodium dodecyl sulfate-polyacrylamide electrophoresis under reducing conditions and the 3H-labeled protein bands visualized by fluorescent autoradiography (Fig. 3). At the end of the pulse, proa! chains were present in both tunicamycin-treated and control medium. However, the course of conversion was considerably slower in tunicamycin-treated cultures. At the end of an 18-h chase, only procu and pcol chains2 were present in tunicamycintreated cultures whereas LY chains made up a prominent portion of the collagenous proteins in control cultures both in the medium (Fig. 3) and in the cell layer (data not shown). These findings indicate that the proteolysis of the COOH-terminal extension was markedly inhibited in the tunicamycin-treated cultures.

MEDIUM
Long Term Labeling of Bones-The course of procollagen to collagen conversion has been well studied in chick cranial bones in culture. Conversion can occur within a physiological setting (in contrast to cells in culture) and intermediates in the process have been identified (6, 13). Chick embryo cranial bones were incubated as described under "Experimental Procedures" and the proteins in the culture medium analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Fig. 4). The medium from control cultures (Fig. 4, A and B) contained two major reducible peaks that were labeled with cystine and mannose. A high molecular weight, collagenasesensitive peak was identified as procollagen; a second peak migrated ahead of Dns-cul in the nonreduced gel and was identified as COOH-terminal procollagen-derived peptide (7). This peptide migrated more rapidly than the Dns-cwl-CB7 marker in the reduced gel and contained approximately 10% of the [Vlcystine radioactivity.
In addition to these two reducible peaks there were other nonreducible peaks which were not identified.
In the medium from tunicamycin-treated bones (Fig. 4, C and D) procollagen and the COOH-terminal procollagen-derived peptide lacked the mannose label almost completely. The medium peptide peak appeared smaller (presumably due 2 In accord with a recently proposed nomenclature (3), chains intermediate in length between proa and (Y chains are termed pa chains. Pa chains containing a COOH-terminal extension are designated pc a and pcu chains containing an NH,-terminal extension are designated pea. Intermediates containing COOH-or NH&erminal extensions are termed pc collagen or pN collagen, respectively. to lack of carbohydrate) and migrated further into the gel. In three experiments the total [YS]cystine radioactivity found in the medium peptide peak of tunicamycin-treated bones was 50% of that found in the medium peptide of control bones, i.e. approximately 5% of the total radioactivity. When the media were precipitated with antiprocollagen serum the ratio of [3S]cystine:[3H]mannose was 0.3:1 in control medium and 8.7:l in tunicamycin-treated medium. To further characterize the glycosylated and nonglycosylated COOH-terminal peptides from procollagen, bones were incubated with [3H]tryptophan for 2 h in the presence or absence of tunicamycin, the media were collected, and proteins precipitated by trichloroacetic acid. The [3H]tryptophanlabeled COOH-terminal peptide was further purified by chromatography on DEAE-cellulose (7). Peptides from control and tunicamycin-treated bones were then analyzed by electrophoresis in sodium dodecyl sulfate-polyacrylamide gels at acrylamide concentrations of 7.5, 10, 12.5, or 15% to determine an asymptotic value for migration distance relative to the observed migration of Dns-al-CB7, a nonglycosylated collagenous peptide of approximately 25,000 daltons ( Fig. 51. The reduced procollagen-derived medium peptide synthesized by bones in the presence of tunicamycin showed the same migration as Dns-al-CB7 at all gel concentrations, whereas the reduced peptide, synthesized under control conditions, showed a slower migration typical of proteins with high sugar content (36). The asymptotic value for the molecular weight of the reduced glycosylated peptide relative to three noncollagen protein standards was 39,500 while that for the reduced nonglycosylated peptide was 37,500.
Pulse-Chase Experiments in Bones -Pulse-chase experiments were performed in bones to follow the conversion process and to further clarify the nature of-the tunicamycin effect. For these experiments cranial bones were preincubated for 6 h and then incubated with [3H]proline for 18 min in the presence or absence of tunicamycin, transferred to medium with an excess of unlabeled proline, and incubated further for 30 or 90 min. The collagenous proteins were then extracted and the samples analyzed by electrophoresis in sodium dodecyl sulfatepolyacrylamide gels. After an Wmin pulse, virtually all the radioactivity in nonreduced control and tunicamycin-treated samples was found in a single, high molecular weight peak with a molecular weight in excess of 400,000 (Fig. 6, A andB).
After reduction, two bands with apparent molecular weights of approximately 150,000 were obtained (Fig. 6, A and B) and identified as pro& and procu2 (131. After a 90-min chase (Fig.  6, C and D), there was a decrease in the relative amount of intact procollagen accompanied by a corresponding increase of two lower molecular weight, nonreducible components whose mobilities coincided with those of (~1 and (~2 chains. There was a marked difference between control bones and tunicamycintreated bones. In the control bones (Fig. 6C), almost all the procollagen was converted to collagen and only 10% of the  20 40 60 80 100 radioactivity was found in high molecular weight intermediates. In contrast, 55% of the radioactivity remained in high molecular weight intermediates in bones treated with tunicamycin (Fig. 6D). These results are in agreement with those previously obtained with iibroblasts in culture (Fig. 1).
To demonstrate the step at which inhibition of conversion occurred, a shorter chase period of 30 min was used. After an 18-min pulse, both control and tunicamycin-treated bones contained procul and procu2 chains (Fig. 7). In control samples, after a 30-min chase (Fig. 7C), all the potential intermediates in the conversion pathway, pr (Y and p,a chains as well as procu and (Y chains, were present (6, 13, 15). The predominant peaks were the (~1 and (r2 chains as expected in the normal conversion process. In the tunicamycin-treated bones (Fig.  701, there were striking differences. Smaller amounts of (~1 and a2 chains and an accumulation of pc (~1 chains were seen, indicating an impairment in the activity of the COOH-terminal cleavage enzyme in the presence of tunicamycin.
The accumulation of pc (~1 chains appeared more pronounced than that of ko2. No corresponding increase in p~c~l was observed suggesting that cleavage of NH,-terminal extensions was not impaired by tunicamycin. Consistent results were obtained by examination of samples prior to reduction. Thus, after an 18-min pulse and 30-min chase, pccollagen was found in similar amounts in control and tunicamycin-treated bones (data not shown). This finding again suggests that the activity of the NH,-terminal procollagen protease was not markedly affected by tunicamycin.

DISCUSSION
The data presented in this paper clearly indicate that in the presence of tunicamycin the conversion of procollagen to collagen is inhibited. The limited proteolysis of procollagen is a complex process that involves the cleavage of at least 6 peptide 3 B DNS-B,, DNS-a, DNS-a,-CB, Dye bonds, during the course of which several intermediates are formed (6,13,15). One proteolytic enzyme which cleaves the NH,-terminal peptides has been partially purified and characterized (37); one or more additional enzymes cleave the COOHterminal domain. The accumulation of R collagen in bones incubated with tunicamycin suggests that the COOH-terminal peptide is cleaved by a different enzyme than that which cleaves the NH,-terminal peptides. The selective accumulation of peal could indicate the presence of more than one COOH-terminal protease but this matter will require further study.
Although complete inhibition of procollagen conversion by tunicamycin was obtained in fibroblast cultures (Fig. 3), the inhibition caused by tunicamycin in bone organ culture was incomplete as demonstrated by the appearance of (~1 and (~2 chains in bones (Fig. 6) and of the COOH-terminal procollagen fragment in culture medium (Fig. 4). This may be due to the presence of a pre-existing active enzyme which remained tightly bound to extracellular matrix in the bone system. Several studies (38,39) have indicated that glycoprotein precursors of viral structural proteins failed to be converted properly when glycosylation of the protein was inhibited by deoxyglucose or glucosamine; in the case of influenza virus hemagglutinin, the presence of heterogenous cleavage products was attributed to lack of substrate specificity caused by the lack of sugar side chains (381. In preliminary studies with enzyme extracted from fibroblast medium (151, we found that tunicamycin was not a direct inhibitor of the COOH-terminal cleaving enzyme, that nonglycosylated procollagen was converted by active enzymatic activity, and that medium from cells treated with tunicamycin did not contain the enzymatic activity.R Therefore, a possible explanation for the inhibition of conversion by tunicamycin is that the COOH-terminal procollagen peptidase is a glycoprotein that is synthesized as an inactive zymogen and normally activated extracellularly.
In the presence of tunicamycin, glycosylation of this protease is inhibited and this may result in impaired secretion of the enzyme. Alternatively, lack of glycosylation of the protease may directly affect its activity or may directly inhibit its activation.
The mechanism by which tunicamycin inhibits glycoprotein synthesis is not entirely understood; however, recent experiments with calf liver microsomes (161, chick embryo microsomes (17), and yeast protoplasts (18) have shown that tunicamycin specificially inhibits the formation of polyisoprenyl Nacetylglucosaminyl pyrophosphate from UDP-N-acetylglucos-' D. Duksin and P. Bornstein, manuscript in preparation. amine. Therefore, it is likely that tunicamycin will inhibit primarily the synthesis of those sugar side chains in glycoproteins which contain N-acetylglucosamine.
The inhibition of incorporation of mannose by tunicamycin strongly suggests that this saccharide follows N-acetylglucosamine on the oligosaccharide side chain of procollagen. Indeed, in most glycoproteins in which carbohydrate is attached to asparagine, Nacetylglucosamine linked N-glycosidically, is followed by mannose (40,41). Clark and Kefalides (91 have recently reported the presence of glucosamine, galactosamine, and mannose in type I procollagen isolated from chick tendon fibroblasts Evidence against linkage of oligosaccharides to seryl or threonyl residues was provided, suggesting, by exclusion, linkage to asparaginyl side chains. The experiments described in this study also demonstrate that mannose is a constituent of procollagens synthesized by chick tendon flbroblasts and cranial bones. Additional experiments have demonstrated that proa and P~(Y, but not pNo chains are rapidly labeled with [2-3H]mannose suggesting that the bulk of this sugar is contained in the COOH-terminal extension." Estimates based on molecular weight determinations of control and nonglycosylated procollagen-derived fragments suggest that approximately 10 saccharides, accounting for a molecular weight difference of about 2000, exist on each COOH-terminal extension of the triple-stranded molecule. The hypothesis that sugar moieties are necessary substituents of proteins designed for export from the cell (421 has been questioned (43). In this study, procollagen synthesized in the presence of tunicamycin was secreted normally by fibroblasts in culture and probably by cranial bone. Such procollagen can be assumed to lack oligosaccharides linked by asparagine-l\racetylglucosamine bonds. Tunicamycin probably does not inhibit the addition of galactosyl and glucosyl moieties to hydroxylysyl residues in the triple helical region of the protein. The latter saccharides are probably not required for secretion of procollagen since secretion appears to be relatively normal in individuals who are deficient in lysyl hydroxylase activity (44) and who therefore synthesize procollagen with very low levels of hydroxylysyl glycosides. Nevertheless, the requirement for some glycosylation of procollagen prior to secretion must be considered unsettled.
There is evidence that proteins destined for transmembrane transport are synthesized with short NH,-terminal peptide extensions that facilitate the binding of ribosomes to the membrane of the endoplasmic reticulum and the subsequent transmembrane movement of the protein into the cisternal space (45). However, it is likely that other events modulate the progress of a protein in this secretory pathway. Preliminary Procollagen Conversion Impaired by Tunicamycin findings in this work suggest that the secretion of another major protein synthesized by fibroblasts is impaired by tunicamycin. This protein resembles the elastic fiber microtibrillar protein recently described by Muir et al. (35) and is thought to be related to cold insoluble globulin and fibroblast surface antigen4 It is possible that some proteins, such as the elastic fiber microfibril protein, IgG (461, and perhaps the COOHterminal procollagen protease are secreted through a glycosylation-dependent pathway whereas others can be secreted independently of glycosylation.
Many biologically active proteins contain oligosaccharide chain(s) (41). Some glycoproteins, like interferon (471, can undergo partial removal of the carbohydrate chain(s) by the action of glycosidases and remain active while others, such as human chorionic gonadotropin (48) and y-globulin (491, lose normal activity when deglycosylated. The use of tunicamycin could provide us with a better understanding of the biological function of oligosaccharide side chains in glycoproteins.

Acknowledgments -The skillful technical assistance of Ms. Kathleen
Williams-Geiger is gratefully acknowledged. We thank Dr. G. Tamura for his generous gift of tunicamycin.
We also thank Drs. Peter Byers and Jeffrey Davidson for their helpful suggestions.