Intracellular Collagen and Protocollagen from Embryonic Tendon Cells*

Abstract When matrix-free tendon cells from chick embryos were incubated with [14C]proline and then extracted with sodium dodecyl sulfate and mercaptoethanol, a major part of the newly synthesized 14C-protein was found to consist of collagen polypeptides which were complete in that their content of [14C]hydroxyproline and their size was the same as that of polypeptides of the precursor form of collagen secreted into the medium. When the cells were incubated with 0.3 mm α,α'-dipyridyl so as to inhibit protocollagen proline hydroxylase and protocollagen lysine hydroxylase, secretion of 14C-protein was inhibited and [14C]protocollagen accumulated within the cells. [14C]Protocollagen extracted from the cells was comprised of polypeptides which were of the same size as polypeptides of the intracellular collagen or about 125,000 as estimated by gel filtration in sodium dodecyl sulfate. In contrast, the small amount of peptide-bound 14C still secreted in the presence of α,α'-dipyridyl was shown to consist of small peptides which were in part derived from the intracellular degradation of [14C]protocollagen. The results demonstrated therefore that in freshly isolated tendon cells protocollagen itself is not secreted. Since the rate of protein synthesis in the presence of α,α'-dipyridyl was the same as under control conditions for about 90 min, the data suggested that the intracellular [14C]protocollagen accumulated in some postribosomal compartment. Extraction of control cells with acetic acid solubilized a large fraction of the intracellular collagen, and amino acid analyses indicated that 40 to 50% of the protein in the extracts was collagen. The protein extracted from cells incubated with α,α'-dipyridyl was similar in amino acid composition, but it contained essentially no hydroxylysine and hydroxyproline and was correspondingly rich in lysine and proline. In addition, 40% of the proline in the protein extract was converted to hydroxyproline after incubation with pure protocollagen proline hydroxylase. Experiments involving limited pepsin digestion provided the first demonstration that intracellular collagen and proto-collagen are largely in a native, triple-helical conformation. After acetic acid extracts from control cells and from cells incubated with α,α'-dipyridyl were dialyzed against ATP, segment long spacing aggregates were obtained. The aggregates were similar to those formed by extracellular, fibrillar collagen except that they had a 130-A NH2-terminal extension which was indistinguishable from that seen in aggregates of the precursor form of collagen secreted by the same cells.


SUMMARY
When matrix-free tendon cells from chick embryos were incubated with [l%]proline and then extracted with sodium dodecyl sulfate and mercaptoethanol, a major part of the newly synthesized 14C-protein was found to consist of collagen polypeptides which were complete in that their content of [l%]hydroxyproline and their size was the same as that of polypeptides of the precursor form of collagen secreted into the medium. When the cells were incubated with 0.3 mrd a,&-dipyridyl so as to inhibit protocollagen proline hydroxylase and protocollagen lysine hydroxylase, secretion of 14C-protein was inhibited and [14C]protocollagen accumulated within the cells.
[14C]Protocollagen extracted from the cells was comprised of polypeptides which were of the same size as polypeptides of the intracellular collagen or about 125,000 as estimated by gel filtration in sodium dodecyl sulfate. In contrast, the small amount of peptide-bound l4C still secreted in the presence of cr,cr'-dipyridyl was shown to consist of small peptides which were in part derived from the intracellular degradation of [14C]protocollagen. The results demonstrated therefore that in freshly isolated tendon cells protocollagen itself is not secreted. Since the rate of protein synthesis in the presence of a,a'-dipyridyl was the same as under control conditions for about 90 min, the data suggested that the intracellular [14C]protocollagen accumulated in some postribosomal compartment.
Extraction of control cells with acetic acid solubilized a large fraction of the intracellular collagen, and amino acid analyses indicated that 40 to 50% of the protein in the extracts was collagen. The protein extracted from cells incubated with ar,cr'-dipyridyl was similar in amino acid composition, but it contained essentially no hydroxylysine and hydroxyproline and was correspondingly rich in lysine and proline. In addition, 40% of the proline in the protein extract was converted to hydroxyproline after incubation with pure protocollagen proline hydroxylase.
Experiments involving limited pepsin digestion provided the first demonstration that intracellular collagen and proto-* This investigation was supported in part by National Institutes of HeaIth Grants FR-107 and AM-14,526 from the United States Public Health Service.
$ Postdoctoral Fellow of the Arthritis Foundation. 5 To whom correspondence should be addressed at the Department of Biochemistry, The Rutgers Medical School, New Brunswick. New Jersey 08903. collagen are largely in a native, triple-helical conformation. After acetic acid extracts from control cells and from cells incubated with cr,cr'-dipyridyl were dialyzed against ATP, segment long spacing aggregates were obtained. The aggregates were similar to those formed by extracellular, fibrillar collagen except that they had a 130-A NH&erminal extension which was indistinguishable from that seen in aggregates of the precursor form of collagen secreted by the same cells.
Collagen is assembled by a series of sequential steps consisting of assembly of proline-rich and lysine-rich polypeptides, hydroxylation of some of the prolyl and lysyl residues either while the nascent chains are still being assembled or after the completed polypeptides are released from ribosomes, and glycosylation of some of the hydroxylysyl residues before the molecule is secreted into the extracellular matrix (for recent review, see Reference 1). Recent studies have demonstrated that collagen is first synthesized as a precursor form which is larger than the collagen found in extracellular fibers (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13) because of an extension of about 130 A at the NHz-terminal end (9-11). This precursor form has been known either as "procollagen" (5,9,10,12,13) or "transport form" (3, 8, II), and it apparently cannot self-assemble into fibers until the NHz-terminal extension is cleaved off (9, 11). In the case of embryonic tendon it has been shown that the extension is not removed until after the molecule is secreted into the extracellular matrix (8).
The two enzymes which synthesize the hydroxyproline and hydroxylysine in collagen require as co-factors or co-substrates 02, Fez+, Lu-ketoglutarate, and perhaps ascorbate (for review, see Reference 1). When cartilage from chick embryos was incubated under conditions where the two hydroxylases were inhibited either by excluding 02 from the system (14,15) or by adding an iron chelator such as 01, oc '-dipyridyl (14-16), the tissue was found to accumulate W-labeled proline-rich and lysine-rich polypeptides called "protocollagen" which served as minced feet (18) and calvaria of chick embryos (19). However, attempts to characterize the protein more fully were handicapped by the limited amounts available, its contamination with matrix collagen, and the tendency of the protocollagen to form large insoluble aggregates (17)(18)(19).
The studies with embryonic cartilage in vitro strongly suggested that protocollagen was not secreted from cells and that it accumulated intracellularly (15). After the hydroxylases were inhibited for several hours and then the inhibition was reversed by exposing the tissue to 02 or iron, the protocollagen which had accumulated during the period of inhibition was hydroxylated to the same extent as collagen in control tissues. Studies with a metal chelator which does not enter cells (20) and experiments with puromycin (21) demonstrated that protocollagen proline hydroxylase is an intracellular enzyme. Accordingly, the fact that protooollagen was hydroxylated in the tissue up to 2 hours after it was synthesized (15) indicated that it remained intracellular. Autoradiographs of the cartilage incubated with [3H]proline supported the conclusion that protocollagen was not secreted from the cells (15,22).
Subsequent studies with fibroblasts in monolayer cultures raised the possibility that under certain conditions some protocollagen was secreted from cells (23)(24)(25). However, most of the reports were not cIear as to how much of the total protocollagen synthesized by the cells was secreted, or as to the size of the protein recovered in the culture medium. A recent paper (7) in fact suggested that most of the protocollagen-like material recovered in the medium from fibroblast cultures consisted of small peptides which were probably degradative products.
We have recently obtained cells by enzymic digestion of chick embryo tendon which are free of extracellular matrix and which under the conditions we have employed remain in suspension and synthesize collagen at a rapid rate for several hours (6,26). We here report the use of this biological system to demonstrate directly the intracellular accumulation of protocollagen in the presence of a,@'-dipyridyl and to prepare sufficient quantities of both intracellular collagen1 and protocollagen for the partial characterization of these proteins. proline was followed, the medium and cell fractions were dialyzed against running tap water for 24 hours, and the retentates were hydrolyzed in sealed tubes in 6 N HCl at 120" overnight. The hydrolysates were evaporated on a steam bath and they were r We here use the term "intracellular collagen" to denote collagen which was extracted from ceIIs and which apparently is hydroxylated to about the same degree as the collagen secreted by the same cells (see "Discussion").
As indicated by the data presented here, both the intracellular collagen and the protocollagen extracted from the cells contain the NH*-terminal extension of about 130 A.
2 The abbreviations used are: SDS, sodium dodecyl sulfate; SLS, segment long spacing. dissolved in 4 ml of distilled water. The ['%]hydroxyproline content was assayed by a specific chemical procedure and a separate aliquot of each sample was taken for assay of total WY.3 (27).
Amino acid analyses on protein fractions were carried out after hydrolysis in 6 N HCl at 120" overnight. The samples were evaporated in a rotary evaporator and they were analyzed on a Spinco model 116 amino acid analyzer. The amino acid analyzer was sensitive enough to carry out either a long column or a short column analysis on about 15 pug of collagen.
Specific assays for hydroxyproline were carried out with a chemical procedure using half of the volumes previously recommended (28). With this modification the method was sensitive to about 0.5 pg of hydroxyproline.
Isolation and Incubation of Matti-free Cells from Tendon-Cells were isolated from leg tendons of 17-day-old chick embryos by digestion with trypsin and purified bacterial collagenase as described previously (6,26). After the digestion the cells were filtered through lens paper and isolated by centrifugation. The cells were washed three times with modified Krebs medium (6) containing 10% fetal calf serum and were then resuspended in Krebs medium for the incubation.
In the experiments in which the synthesis of [14C]collagen and [14C]protocolIagen was followed, 5 x lo7 cells isolated from the tendons of four to six embryos were incubated in 20 ml of modified Krebs medium. The samples were preincubated with or without 0.3 mM cr,c~'-dipyridyl for 30 min and then 20 $Zi of [14C]proline were added to both the control samples and the samples containing a! ,cu'-dipyridyl. The incubation was carried out in a siliconized Erlenmeyer flask with shaking at 37". At each time point an aliquot of 2.0 ml was removed with a plastic pipette and immediately pipetted into 0.2 ml of medium containing 220 pg of cycloheximide and 2.2 pmoles of ac , a'-dipyridyl in an ice bath in order to stop any further incorporation of 14C or hydroxylation of [14C]protocollagen. The samples were centrifuged at 1600 x g for 12 min at room temperature in order to separate the medium from the cells. The pellet containing the cells was washed with modified Krebs medium containing 1 InM a! ,a'-dipyridyl and 100 pg per ml of cycloheximide and recentrifuged. The wash solution was discarded.
Gel Filtration in SDS-Agarose-Gel filtration was carried out on SDS-agarose as described by Jimenez et al. (8). To prevent degradation of the intracellular protein by hydrolytic enzymes, the previous procedure was modified slightly so that the cell fraction was suspended in 3.0 ml of water at the end of the incubation period and immediately heated in boiling water for 3 min. The medium from the cells was also immediately heated in boiling water for 3 min. The samples were then stored frozen until denatured and reduced with SDS-mercaptoethanol. The latter procedure was carried out by adjusting both the cell fraction and the medium to 1 y0 SDS and 1% mercaptoethanol in 0.02 M sodium phosphate, pH 7.4, in a final volume of 4 ml, and incubating them at 37" for 3 hours. The samples were dialyzed at room temperature for 4 hours and then 15 hours against 100 volumes of 0.1% SDS and 0.1 M sodium phosphate, pH 7.4, and chromatographed at room temperature on a column (1.5 by 90 cm) of 6% agarose (Bio-Gel A-5m, 200 to 400 mesh, from Bio-Rad) equilibrated and eluted with 0.1% SDS and 0.1 M sodium phosphate, pH 7.4. Fractions of 2 ml were collected and aliquots of 0.2 ml were assayed in a liquid scintillation counter with an efficiency of 76 to 780/ The recovery of %-protein from the SDS-agarose column was over 90%. All of the chromatograms were obtained with the same SDS-agarose column and the same values for V,/Bt were obtained with standard proteins over a period of 4 months.
However, t,he sharpness of the elution peaks decreased somewhat with time. The Tit varied from 114 to 120 mi and the chromatograms were adjusted to a Vt of 116 ml to facilitate comparison of the elution patterns obtained in different experiments. g for 20 min and resuspended in 0.5 ml of 0.1 M Tris-HCl buffer, pH 7.8, at 25". The protein was again precipitated with acetone and dissolved in the 0.1 M Tris-HCI buffer.

Treatment of Collagen Precursors with Pepsin-To determine
Extraction of Collagen and Protocollagen from Cells-To obtain maximal amounts of intracellular collagen or protocollagen from the matrix-free cells, from 0.4 to 1.1 X log cells from the leg tendons of about 60 embryos were used for a single experiment. The cells in a concentration of 5 x lo6 per ml were incubated either under control conditions or with 0.3 m&I cr,oc'-dipyridyl as described above except that 10% fetal calf serum was added to the medium of cells incubated with OL , cY'-dipyridyl to obtain slightly better yields of protocollagen.
Although the rate of collagen synthesis was not consistently linear for more than 2 hours, the deviation from linearity was generally not great and the total amount of collagen in the system continued to increase for up to 8 hours (see below).
For this reason incubation periods of 4 or 6 hours were used in many of the experiments for the preparation of quantitative amounts of collagen or protocollagen. The cell pellet was homogenized in about 30 ml of 0.1 M acetic acid and dialyzed exhaustively at 4" against 0.1 32 acetic acid for 48 to 72 hours. The suspension was centrifuged at 20,000 x g for 30 min and the supernate was dialyzed against 0.4 M NaCl and 0.1 M Tris-HCl buffer, pH 7.5. After dialysis for 24 to 48 hours, t,he protein in the sample was precipitated by adding 176 mg per ml of solid ammonium sulfate (Uaker Chemical Co.). The precipitate was sedimented by centrifuging the sample at 20,000 x g for 30 min and the pellet was extracted into 1.5 ml of 0.1 M acetic acid overnight.
After centrifugation at 20,000 x g for 30 min, the clear supernate was dialyzed against 0.1 M acetic acid and then taken directly either for amino acid analysis or for other studies.
All these procedures were carried out at 4". Hydroxylation of Isolated Protocollagen with Protocollagen whether the intracellular collagen and the protocollagen were resistant to pepsin, the cells were extracted with acetic acid and precipitated with ammonium sulfate as described above. The pellets were dissolved in 0.1 M acetic acid, dialyzed against 0.1 M acetic acid, and then stored at 4". For the enzymic treatment (3,11,31), an aliquot of 0.2 ml was adjusted to 0.5 M acetic acid and incubated with 100 pg per ml of pepsin (Sigma Chemical Co.). The amounts of protein were 3 to 15 pg, and the final volume for the incubation was 2.0 ml. The enzymic digestion was carried out at 15" for 6 hours and then the reaction was stopped by adding 0.35 ml of 3 N NaOH.
The samples were adjusted to a final volume of 3.0 ml containing 1% SDS, 1% mercaptoethanol, and 0.02 M sodium phosphate, pH 7.4, by adding concentrated stock solutions. The samples were incubated at 37" for 3 hours and then were dialyzed at room temperature against 0.1 y. SDS and 0.1 M sodium phosphate buffer, pH 7.4, before chromatography on the SDS-agarose column. Preparation of segment Long Spacing Aggregates-To prepare SLS aggregates. 0.8 to 1.1 X 10" cells were incubated under control conditions or with 0.3 mM c~,a'-dipyridyl for 4 hours, and the cells were separated from the medium by centrifugation. The cells were extracted with 0.1 M acetic acid, and the extracts were precipitated with ammonium sulfate as described above. The ammonium sulfate pellet was dissolved in 1.5 ml of 0.1 RI acetic acid and dialyzed against 0.01 M acetic acid. After cen-Proline Hydroxylase-To observe the synthesis of chemical amounts of hydroxyproline, a total of 90 pg of protein extracted from cells incubated with a!, oc'-dipyridyl was divided into two equal samples. The test sample was incubated with 100 units of pure protocollagen proline hydroxylase (29) with a specific activity of 1500 units per mg, 2 mM ascorbic acid (Fisher Scientific), 0.05 mM FeS04, 1 mg per ml of bovine serum albumin, 0.1 InM dithiothreitol (Eastman Organic Chemicals), 0.5 InM a-ketoglutarate (Calbiochem), and 50 mM Tris-HCl buffer, adjusted to pH 7.8 at 25", in a final volume of 1 ml (30). The control sample was incubated under the same conditions except that no enzyme was added. After incubation for 2 hours at 37", the reaction was stopped by adding an equal volume of concentrated HCl. The samples were then hydrolyzed and assayed for hydroxyproline with the specific chemical assay (see above). trifugation at 20,000 x g for 30 min, the supernatant was dialyzed for 24 hours against 0.01 M acetic acid containing 0.2yc disodium ITP (Sigma Chemical Co.). One drop of the contents of the bag was placed on a carbon-coated grid, and excess fluid was removed with a filter paper. The sample was stained and examined in a Hitachi model HU-11F electron microscope operated at 100 kV. Slide plates (Kodak Contrast Projector) were exposed to the image at a magnification of 60,000. The magnification of the microscope was calibrated using the 86-h spacing of catalase (32).

Synthesis of Intracellular
[W]Collagen and [14C]Protocollagen-Matrix-free cells from tendon were incubated with [%]proline under control conditions or in the presence of 0.3 m&r LY ,Q'dipyridyl for up to 3 hours. Under control conditions 17.5 to 25.9 y0 of the '"C in intracellular proteins was [%]hydroxyproline (Table I). From 37.5 to 41.4% of the 1% in the protein secreted into the medium was [Y!]hydroxyproline.
As discussed elsewhere (6), the ratio of [IYJhydroxyproline to total 14C should be about 43yo for pure tendon collagen in which only the prolyl and hydroxyprolyl residues are radioactively labeled.
In the a!, c&dipyridyl-treated samples hydroxylation of proline was markedly inhibited so that less than 0.37, of the protein-bound 1% in either the cells or the medium was accounted for by 114Clhvdroxvuroline.
In experiments in which [14C]protocollagen was used as a substrate and the synthesis of [lJC]hydroxyproline was measured, the hydroxylation was carried out under the same conditions except that the amount of enzyme used was 40 to 60 units, catalase (Sigma Chemical Corp.) was added to a final concentration of 0.2 mg per ml, and the final volume was 4.0 ml. The amount of [YJprotocollagen added was 5,000 to 60,000 dpm which on the basis of amino acid analysis was equivalent to 0.5 to 6 pg of protein.
When ['G]protocollagen from the SDSagarose column was used as substrate, the phosphate was removed by dialysis against distilled water and the SDS was removed by precipitating the protein with 10 volumes of cold acetone. The protein was isolated by centrifuging at 15,000 X In the control samples the incorporation of [14C]proline was linear for about 180 min from the time the isotope was added (Fig. IA).
In the samples incubated with c~,oc'-dipyridyl the incorporation of l4C was the same as in the control for 60 min after the isotope was added. Thereafter the apparent rate of incorporation decreased.
When the X-protein in the cells and medium was assayed separately, it was apparent that a major effect of the o( , ol'-dipyridyl was a decrease in the secrebion of l4C-protein ( Fig. 1, B and c).  1B). In contrast, the amount of "C-protein secreted into the medium was only about 15% of the control (Fig. 1C).
Size of W-Peptides Secreted in Presence of (Y ,a-DipyridybTo determine the size of the W-peptides secreted under control conditions and in the presence of (~,a'-dipyridyl, the medium from the samples was removed, treated with SDS-mercapto- In contrast, the smaller amount of nondialyzable 14C recovered from the medium of cells incubated with cw ,ar'-dipyridyl eluted much later from the SDS-agarose column (Fig. 2B). The peak eluted at about the same position as cytochrome c with a molecular weight of 13,500, and therefore it consisted of IJC-peptides just large enough to be retained during the dialysis steps. The amount of peptide-bound l4C which was recovered as small peptides (fractions of 38 to 52 in Fig. 2, A and B) was significantly greater in the medium of cells incubated with o( , ar'-dipyridyl than in the control sample, even though the number of cells, the amount of [iG]proline, and other conditions were the same. As indicated above (Table I), the l*C-peptides in the medium contained essentially no [l%]hydroxyproline.
When the i4Cpeptides recovered from the column (Fig. 23) were incubated with protocollagen proline hydroxylase in two separate experiments, 7.1 and 8.3y0 of the l*C was converted to peptide-bound [i%]hydroxyproline, a result which indicated that these peptides were at least in part derived from collagen precursors.
Size of Intracellular ]%']Collagen and ['4C]Protocollagen-The size and amount of 'Gprotein in the cell fractions were also examined on the SDS-agarose column (Fig. 3). Preliminary experiments demonstrated that when cells which had been incubated with [Wlproline were extracted with 1% SDS and 1% mercaptoethanol at 37" for 3 hours, over 95% of the nondialyzable i4C was solubilized.
The elution pattern obtained from the SDS-agarose column was not significantly different if the extraction at 37' in 1% SDS-mercaptoethanol was carried out for 12 hours instead of 3 hours. In the control sample a large fraction of the nondialyzable W-protein was recovered in a peak with an elution position about the same as the major peak observed with IJC-protein from the medium (compare Fig. 3 with Fig. 2A).

The ratio of [W]hydroxyproline
to total 14C in the major peak was 36%. As indicated in Fig. 3, the total amount of nondialyzable 14C was increased in cells incubated with cr ,a'-dipyridyl and most of the additional i4C-protein was recovered in a peak with an apparent molecular weight of about 125,000. Both the control and the test sample contained 1.5 X 107-&ells and they were incubated under the same conditions for 3 hours in 6.0 ml of medium with 5 &i of Wlproline.
The total cell fraction was treated with SDS-mercaptoethanol as described under "Experimental Procedure" and the chromatographic conditions were the same as in Fig. 2A. The aliquot placed on the column corresponded to one-eighth of the total cell fraction.
The [i*C]collagen from control cells and the [14C]protocollagen from the cells incubated with 01, ol'-dipyridyl was partially purified and fractionated by extraction with acetic acid and precipitation with ammonium sulfate (see '%xperimental Procedure"). From 50 to 70% of the nondialyzable 14C was recovered with these procedures.
When the partially purified 14C-protein was denatured and reduced with SDS-mercaptoethanol and then chromatographed on the SDS-agarose column, most of the 14C eluted in a sharp peak with an apparent molecular weight of about 125,000 (Fig. 4, A and B). With the control sample the ratio of ["Clhydroxyproline to total i*C in the peak was 46.6%, indicating that the proline in the intracellular collagen was hydroxylated as fully as the collagen in the medium (6,26).
The '*C-protein extracted from cells incubated with (Y ,a'-dipyridyl contained no [14C]hydroxyproline, but, when the SDS in the fractions was removed and the 'V-protein was incubated with protocollagen proline hydroxylase, there was a synthesis of ['*C]hydroxyproline.
The final value for the ratio of [14C]hydroxyproline to total r*C after incubation with the enzyme was 29.6%. Cells, 6 X 108, were incubated in 56 ml of medium for 6 hours with 8 pCi of [WJproline. The cells were treated as described under "Experimental Procedure" and an aliquot corresponding to one-fifteenth of the total extract was placed on the column. The chromatographic conditions were the same as in Fig Conditions were similar to those in the experiment shown in A but the total amount of W-protein is not directly comparable because the number of cells was 2.1 X 108, the cells were incubated in 40 ml of modified Krebs medium containing 10% fetal calf serum, 10 pCi of [r4C]proline were used, and the incubation time was 4 hours. l -* , elution of W; 0---0, [W]hydroxyproline content of fractions before incubation with protocollagen proline hydroxylase ; A-* -A, ['*C]hydroxyproline content after incubation with protocollagen proline hydroxylase.  Amino Acid Analysis of Acetic Acid Extracts of Control Tendon Cells and Tendon Cells Incubated with a, ol'-Dipyridyl--Xs reported previously (ll), a net synthesis of co1lagc~11 could hc ob served with the tendon cells (Table II). When 30 x 10" ~~11s were incubated for 4 hours, thcrc was a Ilet, incrcasc> iu proteinbound hydroxyproline in the mc>dium of 3.77 pg. When tjhc: same number of cells was incubated for 8 hours, there was a net increase of 6.24 c(g of hydrosyproline.
Xt the samt: time there was a small decrease in the protein-bound hydroxyproline in the cell fraction.
There was no significant difference in the net synthesis of collagen hydroxyproline when the cells were incubated in Eagle's Minimum EssenCal medium instead of the Krebs medium, indicating &at during the incubation period the amount of collagen synthesized was not affected by the presence of exogenous essential amino acids and vitamins.
From 0.4 to 1.1 x log cells which had been incubated under control conditions or with cz,a'-dipyridyl for 4 hours were extracted with acetic acid and the extract was precipitated with ammonium sulfate. The samples were then hydrolyzed and examined in an amino acid analyzer.
Since collagen is the only protein in tissues such as tendon which contains any significant amount of hydroxylysine or hydroxyproline, and since collagen has an unusually high content of glycine, the contents of these three amino acids were used to estimate the collagen content of the cell extracts.
Collagen from chicken tendons has been reported to contain 9.6 residues of hydroxylysine, 99 residues of hydroxyproline, and 331 residues of glycine/lOOO amino acid residues (33). An acetic acid extract of the tendons from which the inatrix-free cells were prepared was found to contain 10 residues/l000 of hydroxylysine, 80 of hydroxyproline, and 283 of glycine (Table  III), suggesting that 80 to 90% of the protein in the extract was collagen.
The amino acid analysis of the extract from the control tendon cells indicated that the proteiu fraction contained 8.3 residues of hydroxylysine, 37 residues of hydroxyproline, and 168 residues of glycine/lOOO. On this basis it was concluded that 40 to 50% of the protein extracted from the control cells was collagen.
Amino acid analyses of the protein extracted from the tendon cells which had been incubated with ac,a'-dipyridyl for 4 hours were similar to the amino acid analyses of the protein from control cells, but there were several major differences (Table  III).
The prot,ein from the cells incubated with cr,ru'-dipyridyl tendon cells were f 4 residues of the mean. b In two samples the height of the hydroxylysine peak was less than 0.0005 absorbance unit). Assuming an elution time of 2 min for the peak, the hydroxylysine content was less than 0.4 residue/ 1000. c In five of six samples the height of hydroxyproline peak was less than 0.0005 absorbance unit and therefore was too small to measure.
Assuming an ellltion time of 2 min for the peak, the hydroxyproline content in 2 samples was less than 3 residuesjlOO0 and in three samples was less than 2 residues/lOOO. In the sixth sample, the height of t.he peak was 0.0008 absorbance unit and calculated on the basis the hydroxyproline content was 1.5 residues/1000. d Amino acids not, assayed in t,he samples indicated.
contained less than 0.4 residue of hydroxylysine and less than 2 residues of hydroxyproline. The protein cont.ained 43 more residues of proline so that its total content of imino acid was within 6% of the value for the protein extracted from the control tendon cells. The lysine content was higher by 15 residues, and therefore the cont.ent of hydroxylysine plus lysine was within 10 y0 of the value for the protein extracted from the control cells.   (Table  III). As indicated (Table  IV), the ratio of hydroxyproline to proline in the protein increased over acid extracts from cells which had been incubated with ~(,a!dipyridyl for 4 hours. Before treatment with pepsin, 67% of the 14C eluted in the same fractions as the precursor polypeptides from the medium (Fig. 5C). After treatment with pepsin, 57% of the initial 14C elut,ed in the same fractions as (Y chains (Fig.   50).
However, if the '%-protein was denatured by heating at 100" for 10 rnin before treatment with pepsin, essentially all of the 14C-protein was digested to smaller peptides ( Fig. 50 and Table IV).

Long Spacing Aggregates jrom Intracellular
Collagen-Control cells and cells which had been incubated with (Y ,a'-dipyridyl for 4 hours were extracted with 0.1 M acetic acid, and the 14C-protein in the extracts was precipitated with ammonium sulfate.
The samples were redissolved in acetic acid and dialyzed against disodium ATP in acetic acid. A flocculent precipitate appeared in the bag and 70 to 80% of the 14C-protein in the samples was sedimented when the sample was allowed to stand at 4' overnight.
Electron microscopy indicat.ed the presence of regular aggregates which were indistinguishable from SLS aggregates of collagen except, that they contained an extension at the NH&erminal or A-end (Fig. 6). The NHrterminal ext.ension was defined most clearly when the aggregates were stained with ammonium molybdate, and measurement of the extension in 40 aggregates indicated a length of 134.0 A f 1.5 (S.E.M.). I:nder the conditions employed the length of the SLS aggregates was 2836 A & 13 (S.E.M.), and therefore the NHz-terminal extension accounted for 4.7 y. of the total length.
As discussed elsewhere (II), there is as yet no clear explanation as to why the XHz-terminal extension increases the length of the SLS aggregates by about 5%, whereas the polypeptide chains appear to be about 20 to 30% larger when their size is examined by gel filtration (8, 12) or polyacrylamide gel electrophoresis (13). The control sample was prepared by incubating 9 X 108 cells for 6 hours in 120 ml of medium with 12 pCi of ['4C]proline. The a+'-dipyridyl sample was prepared by incubating 6 X lo8 cells for 4 hours in 80 ml of medium with 6 &i of ['%]proline. The cells were extracted with acetic acid, treated with pepsin at 15", and then treated with SDS-mercaptoethanol as described under "Experimental Procedure." With the control samples an aliquot corresponding to oneeighth of the tot,al extract was placed on the column. With the a,~'-dipyridyl sample an aliquot corresponding to one-sixteenth of the total sample was placed on the column. The chromatographic conditions were the same as in Fig. 2A There was no difference in the appearance of the SLS aggregates prepared from control cells and those prepared from cells incubated with tu , cw'-dipyridyl for 4 hours. Also the apparent yield of SLS aggregates was about the same as judged by the number of aggregates seen on the grids and by the relative amount of W-protein which sedimented when the samples were allowed to stand at 4" overnight.
Although it was previously reported (11)  nal extension of the collagen secreted by tendon cells did not stain positively with either phosphotungstic acid or uranyl acetate, further studies demonstrated that the NHz-terminal extension did stain positively (Fig. 7) when the SLS aggregates were exposed to uranyl acetate for 5 min instead of 1 to 2 min. After staining with uranyl acetat.e for 5 min a broad, positively stained band was seen at, the NHP-terminal extension. The same pattern was seen in SLS aggregates from control cells, from cells incubated with o( , a'-dipyridyl, and from the medium of control cells (11).

DISCUSSION
Pulse-label and chase experiments (6) and quantitative assays of hydroxyproline (Table II) demonstrated that most of the collagen secreted by the matrix-free cells from tendon is recovered in the incubation medium. The system therefore makes it possible to isolate intracellular collagen which is not contaminated by extracellular fibrillar collagen. Extraction of the cells with SDS-mercaptoethanol under the conditions used here solubilized all of the W-protein, and gel filtration on SDS-agarose indicated that a major part of the newly synthesized intracellular W-protein consisted of polypeptides of about 125,000 daltons, or of about the same size as the polypeptides of the precursor form of collagen secreted into the medium by the matrix-free cells (8,11 traction and ammonium sulfate precipitation, the ratio of [l"C]hydroxyproline to total 14C was equal to the value observed with collagen polypeptides from the medium.
The results suggested therefore that a significant part of the intracellular '%-protein consisted of collagen polypeptides which were complete in the sense that they were the same length and they contained the same amount of hydroxyproline as the collagen secreted by the cells. Studies are now in progress3 to establish whether the collagen polypeptides are also complete in the sense of containing the same amount of hydroxylysine and glycosylated hydroxylysine as the secreted collagen.
If these values are the same as for the secreted collagen, the 14C-protein extracted from the cells should probably be regarded as an intracellular collagen' and should be distinguished from partially hydroxylated protocollagen, even though it can probably be hydroxylated further by incubation with large amounts of either protocollagen proline hydroxylase (34) or protocollagen lysine hydroxylase.4 It was of interest that under the conditions used here, or when the cells were pulse labeled for only 2 to 15 min,3 no significant amounts of 14C-polypeptides were found with molecular weights of more than 125,000 as estimated by chromatography on SDSagarose. Accordingly, there was no evidence that the matrixfree cells from tendon synthesized a collagen precursor with a molecular weight of 500,000 to 600,000 as has been recently described in cultures of fibroblasts (35).
Incubation of the matrix-free cells from embryonic tendon with OL , ar'-dipyridyl prevented the conversion of peptide-bound [%]proline to [14C]hydroxyproline, inhibited the secretion oi '4C-protein by the cells, and produced an intracellular accumulation of 14C-protein.
The 14C-protein which was accumulated in the cells was [14C]protocollagen as defined by its size, the 3 J. Uitto, P. Dehm, and D. J. Prockop, in preparation. 4 We have recently observed that the hydroxylysine content of collagen rat tendon can be increased by incubating it with relatively large amormts of protocollagen lysine hydroxylase (K. I. Kivirikko and D. J. Prockop, in preparation).
absence of [l%]hydroxyproline, and its ability to serve as a substrat,e for the synthesis of [14C]hydroxyproline by pure prot'ocollagen proline hydroxylase.
The small amount of pcptidcbound 1% still secreted into the medium in the presence of (Y ,oc'-dipyridyl was shown to consist of small 14(:-peptides. The results directly demonstrated therefore that in freshly isolated tendon cells protocollagen itself is not secreted. The data also demonstrated that inhibition of the hydroxylation of protocollagen does not immediately affect protein synthesis, since the incorporation of [14C]proline was the same as in the control for up to 90 min from the time at which a!,(~'dipyridyl was added. The synthesis time for a polypeptide chain of collagen has been estimated to be about 6 min (12). The results obtained here therefore support earlier indications (15,22,36,37) that when the hydroxylation is inhibited complete chains of protocollagen are released from ribosomes and protocollagen accumulates in some post-ribosomal compartment .
The apparent decrease in 14C incorporation after 90 min of exposure t,o (Y ,a!-dipyridyl and m-ith 60 min of continuous labeling with [14C]proline may reflect a delayed feedback inhibition of protein synthesis, but it may also be explained by achievement of a new steady state in which t.he synthesis of [14C]protocollagen is balanced by an increased rate of intracellular degradation of t,he [14C]protocollagen t'o both nondialyzable and dialyzable 14C-peptides.
In the presence of 0.3 mM cy , ol'-dipyridyl the total amount of peptide-bound l4C in the medium was markedly decreased, but after incubation of the cells for 180 min, the amount of small 14C-peptides in the medium was greater than in the control.
Since the small 14Cpeptide fractions served as a substrate for the synthesis of [14C]hydroxyproline by protocollagen proline hydroxylase, the results suggested that after [14C]l)rotocollagen had accumulated in the cells for some time, some of the intracellular [14C]protocollagen was degraded to small 14C-peptides which were then secreted. A similar situation apparently occurs in fibroblast cultures incubated with (I(, ol'-dipyridyl (7). The precursor form of collagen known as "procollagen" or "transport form" had different solubility properties from the collagen which can be extracted from collagen fibers (2, 3, 5-8, 10-12).
Because of this, and because of the small amounts of protein which were available, the intracellular collagen and protocollagen could not be purified extensively. However, half or more of the intracellular collagen and protocollagen was solubilized by extracting the cells with acetic acid under conditions known to solubilize many collagens in native form. After the extracts from control cells were fractionated with ammonium sulfate, 40 to 50y0 of the protein was collagen on the basis of its content of hydroxyproline, hydroxylysine, and glycine. Comparable extracts from cells incubated with cy , ac'-dipyridyl had about the same amino acid composition except that they contained essentially no hydroxyproline or hydroxylysine and they were correspondingly rich in proline and lysine. These results are compatible with the conclusion that the cells incubated with o(, a!-dipyridyl contained protocollagen and little, if any, hydroxylated collagen. The conclusion was supported by the fact that 40% of the proline in the protein extracts was converted to hydroxyproline by pure protocollagen proline hydroxylase. Both the intracellular collagen and protocollagen extracted from the cells were largely in a native conformation. These observations provide the first experimental evidence to indicate that collagen becomes triple helical before it is secreted into the extracellular matrix. It was previously demonstrated in cartilage (13,22,36) that [14C]protocollagen could be hydroxylated to a normal extent several hours after it accumulated in cells, and similar experiments have recently been carried out with tendon cells3 It is apparent therefore that a triplehelical conformation does not prevent the intracellular hydroxylation of peptidyl proline by protocollagen proline hydroxylase. A similar conclusion has recently been reached on the basis of enzymic studies with a synthetic substrate which forms triplehelical structures in solution (38).
The helical conformation of the intracellular collagen was further demonstrated by the preparation of SLS aggregates, since previous work (39,40) had demonstrated that only triplehelical collagen will form such regular structures.
The major distinguishing feature of the aggregates from the intracellular collagen was the presence of a 130-A NHz-terminal extension which was indist,inguishable from that seen in SLS aggregates of the precursor form of collagen secreted into the medium by the same cells (11). The NHz-terminal extension was also indistinguishable morphologically from the extension seen in SLS aggregates prepared from the skin of cows with dermatosparaxis (9, 10). With the extracts from tendon cells no SLS aggregates were seen which did not contain the NHz-terminal extension, an observation which was consistent with the fact that essentially all of the [14C]collagen polypeptides in the acetic acid extracts had an apparent molecular weight of 125,000 by gel filtration in SDS-agarose. The observation was also consistent with the earlier demonstration (8) that in tendon the NHz-terminal extension is not cleaved until after the collagen is secreted int,o the extracellular milieu.
The SLS aggregat#es from cells incubated with oc,a'-dipyridyl were essentially the same as those from control cells and they contained the same NHt-terminal extension. The apparent yield of SLS aggregates from cells incubated with CX,CX'-dipyridyl was about the same as from controls, and, on the basis of the amino acid analyses, no more than 2% of the protein in the samples was fully hydroxylated collagen.
Because it is not possible to quant!itate the yields of SLS aggregates, it may be that some of the aggregates which were seen consisted of intracellular collagen which contaminated the preparation of protocollagen.
However, the results suggested that protocollagen formed SLS aggregates similar to both intracellular collagen and the collagen secreted in the medium by the matrix-free cells.
Positive staining of the SLS aggregates for a longer time than was used previously (11) demonstrated that the NHz-terminal extension stained more readily with uranyl acetate than with phosphotungstic acid, a result which suggested that the extension contained an excess of negatively charged groups. This observation is consistent with reports indicating that the precursor forms of collagen from the skin of cows with dermatosparaxis (4, 9, lo), from membranous bone of chick embryos (13), and from the medium of matrix-free tendon cells (41) contain an excess of acidic amino acids when compared to fibrillar collagen.