Defects in the Processing of Procollagen to Collagen Are Demonstrable in Cultured Fibroblasts from Patients with the Ehlers-Danios and Osteogenesis Imperfecta Syndromes*

This is a study of the processing of procollagen to collagen in cultures of skin and tendon fibroblasts. Processing was markedly increased by growing cells for 2-4 days postconfluence and then adding ascorbate to the medium for 2 days prior to labeling with [3H] proline. With this system, more than two-thirds of the pro-a chains of type I procollagen in the culture medium, and more than 90% of those in the cell layer, were rapidly processed to PC-a, pN-a, or a chains. Purified, exogenous procollagen was also rapidly processed in cell-free culture medium. The results showed for the first time that exogenous procollagen can be processed in conditioned cell-free medium. The system was then used to compare the processing of procollagen in the medium of normal fibroblasts, cells from one bovine and four human variants of osteogenesis imperfecta, and those from eight human variants of the Ehlers-Danlos syndrome. The cells could be divided into three groups, based on their ability to process type I procollagen: 1) normal, 2) consistently slow, and 3) very slow. The cause of the decreased processing was shown to be associated with either a mutation causing a shortening manifestations of mild and EDS (11). for cell of EDS type the Type Culture by Dr. clinical of type VI1 in patient by Dr. in patient RMS-54 by and the half-siblings and RMS-75, clinical and atrophic of clinically variant with of type the differences in migration of human and bovine collagen chains is not known, but this is characteristic of species differences in the electrophoretic mobility of collagens.

variants of EDS, and mutations that change the structure of type I procollagen have been found in a few variants of 0 1 (9)(10)(11)(12)(13)(14)(15)(16)(17). In most variants of EDS or 01, however, neither a specific genetic mutation nor decreased procollagen processing has been identified. Lichtenstein et al. (18) found that they were not able to extract procollagen proteinase activity from the cell layer or culture medium of fibroblasts from three unrelated cases of EDS, while Layman and Ross (19) were able to demonstrate procollagen peptidase activity in extracts of both the cell layer and medium of cultures of normal human fibroblasts. In a subsequent study, however, Layman (20) found that the procollagen proteinase activity was quite variable in extracts of the cell layer and culture medium of confluent cultures of fetal and adult human fibroblasts. This is consistent with numerous observations that only small and variable amounts of procollagen processing are found when normal fibroblasts are grown with the usual culture conditions (18)(19)(20)(21). Also, because of the small amount of processing that occurs in most fibroblast cultures, conflicting observations have been reported as to whether type I procollagen is processed immediately before or after secretion, whether the processing occurs at a time and place that are distant from secretion, or whether the processing can occur in cell-free culture medium (17)(18)(19)(20). Sonohara et al. (22) recently showed that considerable amounts of procollagen processing occur in postconfluent cultures of embryonic guinea pig fibroblasts. We observed similar amounts of processing in the course of studies of a bovine variant with osteogenesis imperfecta. Processing was maximal when the fibroblasts were maintained at confluence for 5-10 days or if cultures that were 2-4 days postconfluent were fed for 2 days with medium containing ascorbate. Here, we used the modified culture system to compare the processing of endogenous 3H-procollagen and chromatographically purified, exogenous 14C-procollagen that was added to cell-free culture medium and to the postconfluent fibroblasts. The system was then used to compare the processing of procollagen to collagen by normal human fibroblasts and by fibroblasts from eight human variants of EDS and four human variants of 01.

MATERIALS AND METHODS
Source of Bovine Fibroblasts-Biopsies were obtained from gastrocnemius tendon of fetal and newborn Holstein calves with the Australian variant of bovine osteogenesis imperfecta (BOA-Aust). Normal Holstein cows were bred at Cornell University with semen from the Australian Holstein bull that was shown to transmit BO1 as an autosomal dominant trait (23). The biopsies were obtained in a collaboration with Drs. Joyce A. M. Wootton, Laurence Denholm, Lennart Krook, and Charles Hall. Primary cultures of fibroblasts were established from biopsies of a normal twin 180-day-old fetus and from an affected twin with BOI. The affected twin was readily identified by the presence of thin fragile tendons, bones, and teeth. Primary cultures were also established from biopsies of normal and affected half-sibling newborn calves, but no differences were detected in the function of the fetal and newborn cells. Since the fetal fibroblasts were from affected and unaffected twins, these cells were selected for the studies of procollagen processing described here.
The cells were established from a tissue mince, grown in Dulbecco's modified Eagle's medium without pyruvate, containing 20% fetal bovine serum, 1% fungizone solution (GIBCO, 600-5295), and 1% penicillin/streptomycin solution (GIBCO, 600-5140). After 2 days of culture, the tissue mince was removed, medium without fungizone was added, and the attached fibroblasts were allowed to form large confluent colonies. These colonies were digested with 0.5% trypsin-EDTA solution (GIBCO, 610-5300), and 2.5 X lo6 cells were plated in 5 ml of medium in 25 cm2 tissue culture flasks. The cultures were fed every other day and were passed at a density of lo' cells/cm2 within 2 days of reaching confluence. In later passages all antibiotics were omitted from the medium, and the concentration of fetal bovine serum was reduced to 10%. All of the studies described here were done with cells between passage 3 and 10. When these cells were to be labeled, they were grown for 2-4 days postconfluence and were fed for 2 days with the same medium containing 25 pg/ml sodium ascorbate.
Sources of Human Fibroblasts-The sources of normal and patient fibroblasts are shown in Table I. Fibroblasts from passage 3 through 15 were grown under standard conditions (9-12) in 25-cm2 tissue culture flasks in Dulbecco's modified Eagle's medium with pyruvate, containing 10% fetal bovine serum and no antibiotics. 2-4 days after reaching confluence, the cells were fed for 2 days with the same medium containing 25 pg/ml sodium ascorbate.
Continuous Labeling Experiments-Cells fed for 2 days with medium containing ascorbate were preincubated for 30 min with 1.5 ml of Dulbecco's medium containing 1% fetal bovine serum. The prein-cubation medium was replaced with the same medium containing 10-50 pCi/ml [2,3,4,E~-~H]proline (Amersham Corp.), and the flasks were incubated for 4 h at 37 "C. To block the hydroxylation of proline and lysine in selected cultures, m,d-dipyridyl was added to both the preincubation medium and labling medium to a concentration of 0.3 mM.
Puke-Chose Experiments-To study the kinetics of procollagen processing, five flasks each of unaffected and affected cells of BOI-Aust were preincubated for 30 min and pulsed for 1 h with 50 pCi/ ml [3H]proline as described above. The labeled medium was removed, the cell layers were washed with Dulbecco's medium, and the flasks were incubated with 1.5 ml of Dulbecco's medium containing 1% fetal bovine serum without label for 0,1, 2, 4, or 8 h.
Processing of Exogenous 14C-Procollagen-To study the enzymatic processing of exogenous procollagen on the cell layer and in cell-free culture medium, a substrate of chromatographically purified 14Clabeled chicken procollagen was prepared as described previously (24)(25)(26). One mg of this purified "C-procollagen, containing 5 X 10' cpm of I%, was incubated for 4 h at 37 "C in 5 ml of cell-free culture medium or in 1.5 ml of fresh culture medium that was added to the cell layer. The cell-free medium was culture medium that was removed from postconfluent cultures of bovine or human fibroblasts after 2 days and was centrifuged for 10 min at full speed in a clinical centrifuge to remove floating cells and other particulates. At the end of the incubation, cold protease inhibitor solution and ammonium sulfate were added to both the cell-free medium and medium from the cell layer, and the cell layer was homogenized and boiled in homogenizing buffer as described below.
Analysis of Labeled Medium and Cell Layer-As described previously (9-12), the medium proteins were recovered by adding a stock protease inhibitor solution to give final concentrations of 11 mM Nethylmaleimide, 1 mM p-aminobenzamidine, 1 mM phenylmethanesulfonyl fluoride, and 30 mM EDTA adjusted to pH 7.4. Ammonium sulfate was added to 30% saturation (176 mg/ml), and the samples  6,7,13). The diagnosis of lethal 01 (type 11) was made for cell line (IMR-2962) by Dr.
Thaddeus Kelly (lo), for RMS-2 by Dr. Carol E. Anderson, and for RMS-18 by Dr. Harold Chen. The clinical manifestations in the patient with moderately severe 01 (RMS-25) were described by Nicholls et al. (37). The patient with mild atypical 01 (RMS-44) and affected members of his family were noted by Dr. Sandra Kaffe to have some manifestations of both mild 01 and EDS (11). The clinical diagnosis for the six cell lines of EDS type VI1 obtained from the American Type Culture Collection were made by Dr. Beat Steinmann (13). The clinical diagnosis of EDS type VI1 was established in patient RMS-52 by Dr. Boris G . Kousseff and in patient RMS-54 by Drs. Sandra Farrell and Rosanna Weksberg. With the two half-siblings RMS-74 and RMS-75, Dr. Kousseff found that the primary clinical signs were easy bruisability and thin atrophic scars, with very little hyperextensibility of skin or joints. Therefore, he concluded that this was a clinically unclassified EDS variant with some characteristics of EDS type IV.

Procollagen
Processing Defects in OI and EDS were stirred overnight at 4 "C and centrifuged at 20,000 X g, at 4 "C, for 1 h. The cell layer was immediately washed three times with cold (4 "C) phosphate-buffered saline and homogenized on ice in 500 ~1 of homogenizing buffer consisting of 270 mM NaCl, 22 mM N-ethylmaleimide (Sigma), 2 mM p-aminobenzamidine (Sigma), 2 mM phenylmethanesulfonyl fluoride (Sigma), 0.1% Triton X-100, and 80 mM EDTA in 270 mM Tris-HCl buffer adjusted to pH 7.4 at room temperature.
The homogenate was immediately boiled 3 min with a l/10 volume of 20% SDS and again for 3 min after adding 2mercaptoethanol to a final concentration of l-5%. The precipitate of the medium proteins was dissolved in 300 ~1 of the same homogenizing buffer and boiled 3 min with 2% SDS and again for 3 min after adding 2-mercaptoethanol to l-5%. For electrophoresis, a l/5 volume of 5 x electrophoresis sample buffer was added, and the samples were dialyzed for 24 h at room temperature against 2 changes of 1 X sample buffer consisting of 2% SDS, 10% glycerol, and 0.001% bromphenol blue in 0.01 M Tris-HCl buffer adjusted to pH 6.8 at room temperature.
Slab gels of 6% acrylamide or a 4-8% acrylamide gradient in SDS were used to separate the labeled peptides (27). In most gels urea was added to a concentration of 0.5 M, and in selected cases the same samples were rerun with either no urea or 2.0 M urea in the gel system.
These comparisons showed that increasing amounts of urea caused the band of type III procollagen to be more focused and those of type V procollagen to be less focused, but urea did not change the density of the bands of type I procollagen.
The differences in concen-A UAUA tration of the peptides in the gels changed the apparent ratio of the different types of procollagen chains, but there was no difference in the separation of the pro-nl(I) and pro-al(III) chains in the gels with different concentrations of urea. The gels were dehydrated with dimethyl sulfoxide, infiltrated with 2$diphenyloxazole, hydrated to precipitate the 2,5-diphenyloxazole, dried, and fluorograms were prepared on Kodak XAR film (28). The films were exposed for periods from 2 to 14 days, and multiple exposures of each gel were scanned at 600 nm with a Gilford Response gel scanner to quantitate the ratios of absorbance in each band containing peptides derived from type I procollagen.
Since this instrument did not reliably integrate separate peaks, the traces of the peaks from the fluorograms were cut out and weighed to confirm the quantitation of ratios of the areas of the peaks representing procollagen, partially processed procollagen, and collagen chains. In panel B, 3rd and 4th lanes, "C-labeled chicken procollagen (50 pg, 5 x lo" dpm) was added to 1.5 ml of fresh medium on the ascorbate-stimulated bovine tendon fibroblast cell layer for a 4-h incubation. After incubation, the '"C-collagens and procollagens were precipitated, reduced, and separated on a gel of 6% acrylamide. As shown t.here was no difference in the processing of either endogenous or exogenous procollagens by the cultures of unaffected and affected bovine fibroblasts.

Processing of
half of the newly synthesized type I procollagen was processed to PC-collagen, pN-collagen, or collagen in the medium of fibroblasts that were labeled for 4 h with [3H]proline. For example, densitometry of the 2nd lane showed that the distribution of 'H in pro-al(1) or processed pro-al(1) chains was 34% in pro-al(I), 30% in PC-al(I), about 8% in pN-al(I), and 26% in a(1) chains. For this calculation, the proportion of pN-al(I), which co-migrates with pro-a2(1), was estimated as twice the value of the pN-a2(1) chain. Therefore, about 65% of the newly synthesized type I procollagen was partially or completely processed. For convenience, the ratio of pCal(1) to the sum of pro-al(1) + PC-al(1) chains was used as a minimal estimate of processing. By this measure about 47% of the procollagen in the 2nd lane of Fig. 1 was processed to PC-collagen. This value is a minimal estimate, since PC-a(1) chains are also being processed to al(1) chains. Similar degrees of procollagen processing were seen in the medium of confluent cultures that were grown for 7-10 days beyond confluence in medium without ascorbate (Figs. lA,  3rd and 4th lanes). The results also showed that during 4 h of labeling there was no difference in the extent of processing of 3H-procollagen in cultures of affected and unaffected fibroblasts from BOI-Aust (Fig. 1).
In the same series of experiments, exogenous 14C-procollagen was chromatographically purified and was used to determine 1) whether the processing of procollagen occurred in cell-free culture medium and 2) whether the extent of processing of exogenous procollagen was comparable to the processing of endogenous procollagen that occurred when the enzymes and substrates were synthesized and secreted by the same cells. As shown in Fig. 1, there was little or no difference in the extent of processing of endogenous 3H-procollagen that was synthesized by the bovine fibroblasts and exogenous I4C- and chased for 0, 1, 2, 4, and 8 h with fresh medium. The labeled collagens and procollagens in the medium were precipitated with 30% saturated ammonium sulfate, reduced, and separated on a 6% gel. Densitometric scans of fluorograms of the gels showed that more than half of the chains of 3H-procollagen secreted into the medium during the chase were partially processed to pN-and PC-collagen (altered procollagen) chains or completely processed to collagen chains. with medium containing ascorbate, labeled for 1 h with [3H]proline, and chased with fresh medium for the times indicated. More than 90% of the pro-a chains were processed to collagen chains, with little evidence of an accumulation of intermediates of the processing. The cause of the differences in migration of human and bovine collagen chains is not known, but this is characteristic of species differences in the electrophoretic mobility of collagens.

Control 01
Hours of Chase  0 1 (RMS-44). Prior to labeling, the cells were fed for 2 days with medium containing ascorbate (25 pglml). The cells were labeled for 1 h with [3H]proline (50 pCi/ml) and chased with fresh medium for the times indicated. Labeled proteins in the medium were precipitated with 30% saturated ammonium sulfate, boiled, reduced with mercaptoethanol, and separated on a 6% gel. These fluorograms are representative of those used for the densitometric scans for Fig.  6. 5-10% of the pro-nl(1) chains were processed to pC-aI (1)

FIG. 5. SDS-polyacrylamide gel electrophoresis showing greatly reduced processing of 3H-procollagen in a pulse-chase experiment with postconfluent cultures of skin fibroblasts from a new patient with EDS type VI1 (RMS-54) and a new patient with an atypical form of EDS (RMS-74).
The controls were postconfluent cultures of normal human skin fibroblasts (IMR-1106 and IMR-3348). The procedures were the same as those used in the experiment described in Fig. 3. Densitometric scans of the fluorograms showed that in the cultures of both patients fibroblasts less than 10% of the 3H-procollagen chains were processed to altered procollagen or collagen chains. Panel A , IMR-54 and RMS-54; panel B,  labeled chicken procollagen that was added to fresh culture medium on postconfluent bovine fibroblasts. Essentially the same degree of processing of the exogenous 14C-procollagen occurred in cell-free medium that was removed from postconfluent bovine fibroblasts after 2 days of culture and was centrifuged to remove floating cells and other particulates (Fig. lB, 1st and 2nd lanes). Densitometry of the fluorograms shown in Fig. 1B (3rd lane) indicated that 40% of the 14C was in pro-al(I), 40% in PC-al(I), 10% in pN-al(I), and 10% in al(1) chains. These values were essentially the same as values obtained for processing of endogenous procollagen in the medium in 4 h (Fig. lA). Most (84%) of the exogenous procollagen remained in the medium, and only 16% of the recovered 14C was in the cell layer. As also indicated in Fig.  lB, there was no difference in the processing of the exogenous procollagen in the culture medium of unaffected and affected fibroblasts of BOI-Aust.
In further experiments, bovine fibroblasts were pulse-labeled for 1 h with [3H]proline, and the label was chased for 0, 1, 2, 4, or 8 h by replacing the labeled medium with fresh medium without label (Fig. 2). Densitometry of the fluorograms indicated that more than 40% of the pro-al(1) chains in the culture medium were processed during the first h of chase and more than 80% were processed by 8 h. Again, there was no difference in the rate of procollagen processing in the affected and unaffected bovine fibroblasts (Fig. 2).
Processing of Type I Procollagen in the Cell Layer-Bovine fibroblasts were labeled with [3H]proline for 4 h, and the labeled proteins in the cell layer and medium were compared.
After 4 h the cell layer contained about two-thirds of the nondialyzable 3H, but only 15% of this 3H was in labeled procollagen or collagen chains. The procollagen in the cell layer, however, was more extensively and rapidly processed.
About 80% of the labeled collagen in the cell layer was recovered as al(1) and a2(I) chains. Electron microscopy showed that thin (-20 nm) fibrils accumulated between the layers of lamellipodia of the postconfluent fibroblasts (not shown). This rapid and complete processing of procollagen to collagen in the cell layer was even more evident in fluorograms from a pulse-chase experiment. As indicated in Fig. 3A, most of the procollagen in the cell layer was processed to a chains after 2 h of chase. Similar amounts of processing occurred in the cell layer in a pulse-chase experiment with postconfluent human fibroblasts (Fig. 3 B ) .
Processing of Type I Procollagen in the Culture Medium of Human Fibroblasts-The same culture conditions used to increase the processing of procollagen by bovine fibroblasts were used to study human fibroblasts. Processing of type I procollagen was readily demonstrated in the culture medium of human fibroblasts (Figs. 4 and 5). The ratio of PC-al(1) to the sum of pro-al(1) + PC-al(1) plateaued a t about 35% during the second h of chase (Fig. 6A), while the amounts of al(1) and &(I) chains continued to increase throughout the chase in the cultures of normal human fibroblasts (Figs. 4  and 5). In contrast, in the culture medium of fibroblasts from a patient with a mild variant of 01 (RMS-44) the ratio remained less than 10% (Figs. 4 and 6B). Previous studies showed that fibroblasts from this patient synthesized equal amounts of normal and shorter pro-a2(1) chains, and procollagen molecules containing these shorter chains were resistant to cleavage in vitro by partially purified procollagen N-proteinase (10). Therefore, fibroblasts from a series of variants of 01 and EDS were examined for evidence of defects in the processing of type I procollagen.
As indicated in Fig. 6A (-) or 0 1 (-----). The per cent conversion was calculated as 100 X the ratio of the area of the PC-a(1) peak to the sum of the areas of the pro-al(1) and PC-al(1) peaks. The amounts of the chains were estimated by densitometric scans of fluorograms of SDS-polyacrylamide gel electrophoresis gels of labeled proteins secreted into the medium of postconfluent cultures that were labeled for 1 h with [3H]proline (50 pCi/ml) and chased for periods of 0, 1 , 2 , 4 , or 8 h with media without label. The shaded area in B is the range of the control values shown in A.

FIG. 6. Graph showing the per cent conversion of pro-al(1) to PC-al(1) chains in the medium of postconfluent cultures of fibroblasts from normal human skin ( A ) and fibroblasts from patients ( B ) with EDS
pro-al(1) chains. As shown in Fig. 6B and in Table 11, patient fibroblasts could be assigned to three groups on the basis of procollagen processing: 1) normal, 2) consistently slow, and 3) very slow. The slow group was defined as variants whose ratio was more than 2 S.D. below the mean of the control values at four out of five time points, and the very slow group were those with values less than 10%.
The variants showing normal procollagen processing included two variants of EDS type VI1 (CRL-1150 and CRL-11931, two variants of severe 0 1 type I1 (RMS-2 and RMS-18), and one variant each of moderately severe 01 type I11 (RMS-25) and EDS type I (RMS-52) ( Fig. 6B and Table I). The variants of EDS type VI1 (CRL-1150 andCRL-1193) were previously reported to have a defect in the procollagen processing (13), but the procollagen in the medium of these fibroblasts consistently had a normal 30-50% ratio of pCal(1) to pro-al(1) + PC-al(1) in the studies carried out here.
The variants with very slow processing of procollagen to collagen included one mild variant of 0 1 (RMS-44), two of EDS type VI1 (CRL-1274 and RMS-54), and one atypical EDS (RMS-74) ( Fig. 6B and Table I). As noted above, RMS-44 was previously shown to have a structural defect in the Nterminal region of the a2(I) chain that made the type I procollagen resistant to cleavage by partially purified procollagen N-proteinase (11). One variant of EDS (CRL-1274) was also previously shown to have decreased procollagen processing (13). However, defects in processing had not been identified previously in either new variant of EDS (RMS-54 and

Identification of a Structural Defect in Type I Procollagen and a Deficiency of Procollagen
N-Proteinase-Previous studies indicated that decreased processing of procollagen can be caused either by mutations that change the structure of a proa chain or mutations that decrease the activity of procollagen N-proteinase. Therefore, the cell lines from the three new EDS variants (RMS-54, RMS-74, and RMS-75) showing decreased processing were examined further. As shown in Fig.  7, the pro-aS(1) chains synthesized by cells from RMS-54 migrated slightly further than normal pro-a2(1) chains in polyacrylamide gels. The increased migration was observed when the procollagen chains were hydroxylated, as well as when the cells were labeled in the presence of the iron chelator and hydroxylation inhibitor a,a'-dipyridyl. The results suggest, therefore, that the slow processing in this variant was explained by a mutation that led to the synthesis of a shortened pro-a2(1) chain. Of special interest was the observation that no normal length pro-a2(1) chains were seen in these cultures (Fig. 7). In contrast, the pro-a chains from the variants RMS-74 and RMS-75 had a normal migration on polyacrylamide gels (data not shown).
Additional experiments were done to determine whether there was a decreased activity of procollagen N-proteinase in the cell lines from RMS-74 and RMS-75 who were halfsiblings. Chromatographically purified 14C-procollagen was added to cell-free medium that was removed from postconfluent cultures of these cells and control fibroblasts. The medium was centrifuged and chromatographically purified 14C-procollagen was added to the supernatant for a 4-h incubation at 37 "C (Fig. 8). Densitometry of the fluorograms indicated that the extent of processing of exogenous pro-al(1) chains to PC-al(1) chains was 23% in the cell-free medium of control fibroblasts, 16% in the medium from RMS-75, and 11% in the medium from RMS-74. Therefore, both of these variants appear to have mutations that decrease the activity of procollagen N-proteinase.

DISCUSSION
The conditions used here to increase procollagen processing were the relatively simple manipulations of allowing the cells Values for the ratio of PC-al(1) to PC-al(1) + pro-al(1) chains were used to define the three processing groups as follows: Normal group, values between 30 and 50%; Consistently slow group, values more than 2 S.D. below the normal mean but more than 10% at four out of five time points; Very slow group, values less than 10%.
* Clinical diagnoses were established as indicated in Table I.
to grow past confluence and supplementing the medium with sodium ascorbate for 2 days prior to labeling. The reason why these manipulations increased processing is not clear, but it may be related to the multilayering of fibroblasts and to the previously reported effects of ascorbate-stimulating procollagen synthesis and secretion (29)(30)(31).
The results confirm previous observations that procollagen N-proteinase activity is present in culture medium (18)(19)(20) and showed for the first time that exogenous procollagen can be processed in cell-free medium from cultured fibroblasts. The exogenous procollagen was processed a t about the same rate and to about the same extent as endogenously synthesized procollagen. Therefore, it is clear that processing can occur after secretion of procollagen. Also, it is clear that processing does not require the fibroblast cell surface or the surface of growing collagen fibrils. These surfaces or processes may play a role in procollagen processing in uiuo, but neither the fibroblast cell surface nor growing collagen fibrils are essential requirements for procollagen processing.
There was more extensive and more rapid processing of procollagen in the cell layer than in the culture medium. Unique pools of tissue fluids may be created between the layers of discoid cells as they multilayer, and these pools may represent an environment that favors procollagen processing and collagen fibril formation. Electron microscopy shows that thin collagen fibrils accumulate between the layers of cell processes in postconfluent cultures, and recent studies indicate that the processing of procollagen and assembly of collagen into fibrils may occur in specialized compartments formed by the plasma membranes of fibroblasts (32,33). The simplest explanation, however, for the more rapid and extensive processing of procollagen in the cell layer is that the substrate and enzyme are in high concentration here, whereas they are too dilute for rapid and complete processing after secretion into the relatively large volume of medium used to culture fibroblasts. Also, it is possible that there is more inhibition of processing of the C-terminal propeptides by the serum in the culture medium than in the cell layer (34).
Because of the extensive processing of procollagen to pCcollagen in the postconfluent cultures, the system described here makes it possible to identify readily two general categories of mutations: those that decrease the activity of procollagen N-proteinase, and those that alter the structure of type I procollagen so as to make it resistant to procollagen Nproteinase. Procollagen N-proteinase is one of a small class of proteinases that require a substrate in a native conforma-FIc,.

I234 gel electrophoresis
show ing decreased processing of exogenous '%-procollagen in cellfree medium of cultures of skin fibroblasts from two patients with a clinically unclassified, atypical EDS. Lane 1, "C-procollagen in fresh Dulbecco's medium containing 10% fetal bovine serum; lane 2, "C-procollagen in cell-free medium from control fibroblasts; lane 3, "C-procollagen in cell-free medium from fibroblasts of an atypical EDS variant (RMS-74); lane 4, '"C-procollagen in cell-free medium from fibroblasts from a half-sibling (RMS-75) of the atypical EDS variant.
As noted in the text, the differences in processing in the control cell line and the two patients cells were confirmed with densitometric scans of these gels.
tion (X-36). Partially purified enzyme did not cleave denatured type I procollagen (35), and there was a reduction in cleavage of a genetically altered type I procollagen from the 01 variant RMS-44 that has a deletion of 18-30 amino acids from (u2(1) CB-4, about 90 amino acids removed from the cleavage site for procollagen N-proteinase (11). Here, there was very slow processing in the medium of RMS-44, as well as in the medium of another 01 variant (IMR-2962) that was previously shown to have a deletion of about 18 amino acids from about the middle of the n2(1) chain (10). Therefore, a slowing of procollagen processing may result, from structural defects that are quite far removed from the cleavage site. This is presumably due to 1) a shift in the register of the triple helix of (Y chains N-terminal to a structural defect and 2) the strict conformational requirements of procollagen N-proteinase. As also demonstrated here, the system can be used to detect mutations that decrease the activity of procollagen Nproteinase.
Previous reports suggested that there is a consistent correlation between the clinical syndrome of EDS type VII and a decrease in the processing of the N-propeptide of type I procollagen (13). However, two variants that were previously classified clinically as EDS type VII (CRL-1193 andCRL-1150, Ref. 13) were found here to show normal processing. In addition, the results demonstrated that processing can be deficient in patients who do not meet the clinical criteria of EDS type VII. One variant with consistently slow processing (IMR-2962) died in utero with the typical manifestations of the lethal, broad-boned form of 01. Two half-siblings with decreased procollagen processing (RMS-74 and RMS-75) had easy bruisability with subcutaneous bleeding, thin atrophic scars, and only slight hyperextensibility of skin or joints. These manifestations are inconsistent with EDS type VII. Therefore, it is apparent that mutations decreasing the processing of type I procollagen can produce more than one clinical syndrome and that the clinical features of EDS type VII do not necessarily indicate that there is a defect in procollagen processing.
One consequence of these observations is that postconfluent cultures of fibroblasts may be useful in the characterization of a broad spectrum of heritable diseases of connective tissues.