Inhibition of the Assembly and Secretion of Procollagen by Incorporation of a Threonine Analogue, Hydroxynorvaline*

Hydroxynorvaline (DL-a-amino-&hydroxyvaleric acid) was shown to competitively inhibit the activation of threonine and valine when tested with tRNA and synthetases prepared from whole chick embryos. However, the hydroxynorvaline was transferred only to threonyl-tRNA and not valyl-tRNA. The hydroxynorvaline had no effect when tested with other amino acids. The K, for threonine was 25 fiM and the K, for hydroxynorvaline was 181 pM. When fibroblasts from embryonic chick tendons were incubated with [3H]threonine and increasing concentrations of hydroxynorvaline, there was a progressive decrease in the incorporation of [3H]threonine so that at 1 mM hydroxynorvaline the incorporation into nondialyzable protein was 26% of the control value. A much smaller decrease in the incorporation of other radioactive amino acids was observed. When the cells were incubated with [“Clproline and 1 mM hydroxynorvaline, the labeled procollagen containing hydroxynorvaline accumulated intracellularly and very little was secreted. Control experiments demonstrated that free hydroxynorvaline did not inhibit the secretion of unsubstituted procollagen. Although the individual


Hydroxynorvaline
(DL-a-amino-&hydroxyvaleric acid) was shown to competitively inhibit the activation of threonine and valine when tested with tRNA and synthetases prepared from whole chick embryos. However, the hydroxynorvaline was transferred only to threonyl-tRNA and not valyl-tRNA. The hydroxynorvaline had no effect when tested with other amino acids. The K, for threonine was 25 fiM and the K, for hydroxynorvaline was 181 pM. When fibroblasts from embryonic chick tendons were incubated with [3H]threonine and increasing concentrations of hydroxynorvaline, there was a progressive decrease in the incorporation of [3H]threonine so that at 1 mM hydroxynorvaline the incorporation into nondialyzable protein was 26% of the control value. A much smaller decrease in the incorporation of other radioactive amino acids was observed. When the cells were incubated with ["Clproline and 1 mM hydroxynorvaline, the labeled procollagen containing hydroxynorvaline accumulated intracellularly and very little was secreted. Control experiments demonstrated that free hydroxynorvaline did not inhibit the secretion of unsubstituted procollagen. Although the individual pro (Y chains containing hydroxynorvaline were of normal molecular weight (125,000) and hydroxyproline content, only about 50% of this intracellularly retained procollagen was triple helical within the cell at 37" as measured by sensitivity to pepsin digestion. Also only approximately 50% of the pro cy chains were disulfide-linked to form triple stranded molecules as compared to greater than 85% linkage in unsubstituted procollagen. We postulate that incorporation of hydroxynorvaline alters the conformation of the propeptide extension sufficiently so that: (a) normal assembly of disulfide-linked, triple helical molecules is reduced and (b) assembled triple helical molecules are not properly recognized by the secretory mechanism.
It is now well established that collagen is first synthesized as a precursor molecule called procollagen, in which the individual polypeptide chains are larger than the cy chains of interstitial collagen because of an extension of the NH, terminus of each chain (l-4). These extensions are each approximately 25,000 daltons and differ from the cy chain portion in their amino acid composition and conformation. The extensions are relatively poor in imino acids, relatively rich in acidic and hydroxyamino acids (5,6), and are susceptible to a number of proteolytic enzymes, including pepsin (l-3). In contrast, the rest of the molecule is in a triple helical conformation, which is resistant to proteolysis by these.enzymes and becomes susceptible to proteolysis only if the triple helical conformation is disrupted by heating or other means (7). The extensions present in the procollagen molecule are held together by disulfide bonds (8)(9)(10). The role of the extensions has not been clearly defined, but it has been postulated that they may serve *This investigation was supported by the National Institutes of Health Research Grants AM-14439 and AM-14526 and the National Institute of Dental Research Grant DE-02623 from the United States Public Health Service.
$ To whom reprint requests should be sent.
to initiate triple helix formation by serving as registration peptides, they may facilitate secretion of collagen into the extracellular matrix, or they may control fibrillogenesis (11). The procollagen molecule undergoes a number of modifications within the cell before secretion. These include hydroxylation of some of the prolyl and lysyl residues, and glycosylation of some of the hydroxylysyl residues. When the hydroxylation of proline and lysine was inhibited in embryonic tibiae (12) and fibroblasts (13,14), unhydroxylated procollagen was secreted at a markedly reduced rate and accumulated intracellularly. It now appears that hydroxyproline stabilizes the collagen triple helix (15)(16)(17)(18)(19), and that unhydroxylated collagen is found within cells at 37" in a random coil conformation.
Since triple helix formation normally occurs intracellularly before collagen is secreted (20), these observations suggested that triple helix formation may be required for normal secretion. This hypothesis is supported by the finding that unhydroxylated procollagen molecules can be secreted by cells incubated at lower temperatures at which the molecules are largely triple helical (21).
When tibiae (22)(23)(24)(25) or fibroblasts (26)(27)(28)(29) were incubated with analogues of proline or lysine, the analogues were incorporated into collagen and the analogue-containing collagens were 7624 secreted more slowly than normal collagen. The collagens containing proline analogues were largely nontriple helical at 37" within cells, but some of the intracellularly retained procollagen molecules containing the lysine analogue appeared to be triple helical. However, by and large, these experiments with the analogues supported the concept that the triple helical conformation was essential for normal secretion. In these experiments no attempt was made to examine possible alterations in the conformation of the propeptide extension by incorporation of the analogues. In the present study we show that cu-amino-p-hydroxyvaleric acid' specifically replaces threonine in procollagen synthesized by chick fibroblasts.
The procollagen chains containing the analogue were of normal molecular weight and hydroxyproline content but formed interchain disulfide linkages less efficiently and were secreted much more slowly than normally.
We postulate that incorporation of hydroxynorvaline alters the conformation of the propeptide portion sufficiently so that: (a) normal assembly of triple stranded chains is reduced, and (b) assembled triple helical molecules are not properly recognized by the secretory mechanism. relatively crude preparation of activating enzyme (Fig. 1). The K, for threonine was 25 pM and the K, for hydroxynorvaline was 181 pM. Hydroxynorvaline was not competitive with any other amino acid tested except valine (Table I). In order to determine whether hydroxynorvaline was transferred to threonyl-and/or valyl-specific tRNA, the following experiment was performed. Aliquots of tRNA were incubated under conditions of maximal acylation with either ["Clhydroxynorvaline, valine, threonine, or no amino acid. The tRNA was then dialyzed and aliquots were reacylated with either [3H]valine or [3H]threonine.
If the tRNA were maximally acylated and remained acylated during the dialysis period, then no radioactivity would be incorporated.
The tRNA sample which was incubated without any "C-amino-acid served as a control for maximal acylation with the radioactive amino acids. The results of this experiment demonstrated that hydroxynorvaline was transferred only to threonyl-specific tRNA since the analogue had no effect on the incorporation of [3H]valine and was as effective as threonine in inhibiting the incorporation of [3H]threonine (Table II). In the case of threonine, apparently a significant fraction of the tRNA which was charged with nonradioactive amino acid deacylated during the dialysis and recovery of the tRNA.
Deacylation has been observed at neutral pH and low temperature and varies with the amino acid and specific conditions (34).
Inhibition of Incorporation of Radioactive Amino Acids by Hydroxynorualine-In order to confirm the specific replacement of threonine by hydroxynorvaline and to study the effects of this replacement in procollagen, chick fibroblasts were incubated with radioactive amino acids in the presence or absence of hydroxynorvaline for 2 hours. Increasing concentrations of the analogue progressively diminished the incorporation of [3H]threonine, so that at 1 mM hydroxynorvaline the incorporation of [Wlthreonine into nondialyzable protein was only 26% of the control value and at 4 mM it was 6% (Fig. 2). A much smaller effect on the incorporation of other radioactive amino acids was observed. After 1 hour of incubation in t,ha  presence of 1 mM hydroxynorvaline, the incorporation ranged from 59 to 73% of the control values for the other amino acids studied, compared to 35% for threonine (Table III). The inhibition of incorporation of other amino acids besides threonine was independent of the concentration of hydroxynorvaline above 1 mM in contrast to the results with threonine in which there was a progressive inhibition of incorporation. These findings, in addition to demonstrating the relatively specific effect on threonine incorporation, suggested that protein synthesis was being inhibited and that this inhibition increased with incubation time. This effect was studied in more detail using ["Clproline (see below). the system and this inhibition increased with incubation time (Fig. 3). In the control after 90 min of incubation, 49% of the incorporated radioactivity was found secreted into the medium. Approximately 90% of this secreted label has been shown previously to be in the form of procollagen (31). In the presence of the analogue, only 13% of the incorporated radioactivity was in the medium. After 30 min of incubation, total incorporation in the presence of the analogue was 75% of the control while after 90 min it declined to 49%.
In order to demonstrate that hydroxynorvaline had to be incorporated into procollagen in order to inhibit secretion, an experiment was designed so that hydroxynorvaline was present either during the synthesis of the procollagen but absent during the period of secretion or it was absent during synthesis but present during the period of secretion. The results of this experiment, presented in Table IV, clearly demonstrate that hydroxynorvaline exerts its effect only when it is present during the synthesis of procollagen and that free hydroxynorvaline has no effect on the secretion of unsubstituted procollagen.
Molecular Weight, Conformation, and Hydroxyproline Content of Procollagen Containing Hydroxynorvaline-In order to estimate the molecular weight of the individual procollagen chains synthesized in the presence of hydroxynorvaline and to determine whether they were in a triple helical conformation, aliquots of cells containing procollagen labeled with ["Clproline and synthesized in the presence of 1 mM hydroxynorvaline were incubated with or without pepsin at 15". The digest mixtures were neutralized and then chromatographed on agarose A-5 columns in sodium dodecyl sulfate (Fig. 4). When pepsin was omitted from the digestion, the analogue-containing procollagen was recovered in a position identical with that of control procollagen with an approximate molecular weight of 125,000 (control not shown). The procollagen peaks were recovered and analyzed for their ['Clhydroxyproline content. The degree of hydroxylation of the procollagen peak containing the hydroxynorvaline was 37.6% compared to a value of 35.6% for control procollagen obtained from cells incubated with colcemide to cause intracellular retention. When the analoguecontaining proteins were digested at 15" with pepsin before chromatography, the quantity of radioactivity recovered as (Y chains was only about 53% of the amount originally found as pro (Y chains. In cells incubated under control conditions, greater than 85% of the intracellular procollagen was resistant to pepsin and recovered as (Y chains (21). These results suggested that a significant fraction of the analogue-containing procollagen was not triple helical within the cell.
In order to examine the stability of the procollagen in more detail, ["Clproline-labeled procollagen was extracted from cells incubated with 1 mM hydroxynorvaline and aliquots of this procollagen were incubated with pepsin at different temperatures up to 37". As demonstrated in Table V, the fraction of procollagen resistant to pepsin remained constant at approximately 56% up to 37". These results indicated that a majority of the intracellular hydroxynorvaline-containing procollagen chains were found in triple helical molecules which were equal in termal stability to unsubstituted procollagen. The finding of two classes of procollagen chains, pepsin-sensitive and pepsin-resistant, suggested that incorporation of hydroxynorvaline might inhibit disulfide bond formation and hence decrease the efficiency of assembly of procollagen chains into triple helical molecules. To test this possibility, intracellular ['Clproline-labeled procollagen synthesized in the presence of hydroxynorvaline was subjected to electrophoresis on sodium dodecyl sulfate polyacrylamide gels in the presence or absence of a reductant.
In the presence of a reductant, procollagen chains migrate with an approximate molecular weight of 125,000 while in the absence of a reductant the linked chains migrate much more slowly. By comparing the quantity of radioactivity recovered as individual pro LY chains in the presence and absence of a reductant, we have found that only one-half of the analogue-containing procollagen chains were disulfide-linked, in contrast to unsubstituted procollagen in Chick fibroblasts were incubated with or without 1 mM hydroxynorvaline for 20 min and then ["C jproline was added to a final concentration of 2 &i/ml. Aliquots, 1 ml, of the cell suspension were taken at the times indicated and centrifuged. The cells were resuspended in 0.01 M phosphate buffer, pH 7.4, and the cells and media were then treated and counted as described in Fig. 2. A-A, total incorporation; O-O, intracellular; 0---0, secreted. which greater than 80% of the chains were linked (Fig. 5).
The small amount of procollagen synthesized in the presence of 1 mM hydroxynorvaline which was secreted into the medium was characterized with respect to pepsin resistance and disulfide linkage (Fig. 6). In contrast to the intracellular molecules, approximately 90% of the extracellular procollagen was disulfide-linked and pepsin-resistant and by these two criteria was identical with control procollagen.

Hydroxynorvaline
was shown by Buston et al. (35) to specifically antagonize the effects of threonine in stimulating the growth of S. fuecalis but did not antagonize the effect of valine. Since labeled hydroxynorvaline was not available and since the chemical amounts of protein synthesized by the tendon cells during the incubations was extremely small, it was not possible for us to demonstrate directly the incorporation of hydroxynorvaline into protein. However, there is little doubt that it specifically replaced threonine in being transferred to tRNA (Table II), and that it markedly inhibited the incorporation of [3H]threonine into protein synthesized by the tendon cells (Fig. 2). Once amino acid analogues have been transferred to tRNA there is no known further physiological process which will prevent their incorporation into protein. Although we can not rigorously exclude other unknown effects of the analogue, it is likely that hydroxynorvaline effectively replaced threonine in the synthesized procollagen.
Although the mechanism of secretion of collagen has been studied for a number of years in different types of cells, many aspects remain unresolved both from the morphological and biochemical points of view. The intracellular pathway has been investigated mainly through the use of autoradiography at the electron microscopic level. Ross and Benditt (36) concluded that collagen passed directly from the endoplasmic reticulum to the cell exterior in fibroblasts.
In cartilage, Cooper and Prockop (37) and Salpeter (38) proposed possible passage through the cytoplasm while Revel and Hay (39) believed that collagen passed through the Golgi apparatus prior to secretion. Weinstock and Leblond (40) using rat odontoblasts, which are polarized and secrete collagen at one surface, have demonstrated that collagen appears to accumulate within the endoplasmic reticulum, enters the Golgi, and is then secreted via exocytosis of granules. The biochemical mechanisms whereby secreting cells distinguish which proteins are to be secreted and which are to be retained intracellularly have not been elucidated. Two general hypotheses can be formulated: (a) mRNA coding for secreted proteins contains a tag which allows it to selectively bind to membrane-bound ribosomes. All proteins  The cell suspension  was rapidly  chilled  and  centrifuged for 2 min at 1,500 x g. The cell pellet was resuspended in 5 ml of 0.5 M acetic acid and the suspension shaken for 24 hours at 4" to extract the procollagen. The suspension was then centrifuged at 12,006 x g for 15 min. and the supernatant containing the extracted procollagen was dialyzed against 0.5 M acetic acid. Aliquots of the procollagen were incubated with or without pepsin at the listed temperatures and the digested proteins chromatographed on agarose A-5 columns in sodium dodecyl sulfate as described in Fig. 4. The degree of pepsin resistance was obtained by comparing the counts found in the derived a chains to the counts in the original procollagen. synthesized on membrane-bound ribosomes are vectorially discharged through the membrane into the cisternae. (b) Endoplasmic reticular membranes selectively recognize and bind proteins which are to be secreted and they are transferred into the cisternae. In both mechanisms, proteins transported into the cisternae are then secreted, possibly passing through other cell organelles or compartments.
In the present work we have found that incorporation of an analogue specific for threonine interferes with the secretion of procollagen.
This inhibition of secretion was not caused by interference with the hydroxylation of proline resulting in destabilization of the triple helical portion of the molecule since the molecules containing the analogue were hydroxylated to the same extent as the control. However, approximately 50% of the substituted procollagen chains found intracellularly were not disulfide-linked, a much higher fraction than that found in the control synthesized in the presence of colcemide. It is likely, but not conclusively established, that it is this fraction which was not perfectly triple helical and hence sensitive to pepsin at all temperatures in the range 15-37" ( Table V). The remaining 50% of the substituted molecules appeared to be triple helical and were as stable by this test as normal procollagen.
It is unlikely, therefore, that incorporation of hydroxynorvaline leads to destabilization of the triple helix per se but rather interferes with formation of the helix by inhibiting disulfide bond formation, possibly by modifying the conformation of the propeptides. This impairment of triple helix formatiom may explain the failure of these molecules to be secreted. However, if the remaining set of intracellular molecules which is pepsin-resistant is identical with that which is disulfide-linked, then this set of molecules is normal by the criteria presently available. In order to explain their intracellular retention we postulate that there may be an alteration in the conformation of the propeptide of this set and that the correct conformation of the propeptide is essential for efficient secretion. Alternatively, the pepsin-sensitive molecules and the