Control of Collagen Production by Human Diploid Lung Fibroblasts*

The fibroblast is a differentiated mesenchymal cell which produces and exports collagen, a macromolecule that plays a critical structural role in the function of most organs. To evaluate the control soft tissue fibroblasts have over collagen production, HFL-1, a diploid human lung cell strain, was studied during periods of rapid cell growth and relatively slow growth over 25 population doublings. To minimize environmental influences, the extracellular milieu of the cells was kept constant throughout the study period. Rates of collagen production per cell per unit time were quantitated by labeling HF’L-1 with [14C]proline and measuring the production of [“C]hydroxyproline after taking into consideration the specific activity of [14C]proline within the free intracellular proline pool and the per cent hydroxylation of proline residues in newly synthesized collagen. Although the specific activity of intracellular free proline and the per cent hydroxylation of proline in collagen varied considerably depending on the growth rates of the cells, collagen production by HFL-1 was constant, even during periods of rapid cell growth. Thus, under conditions of a stable environ- ment, populations of soft tissue fibroblasts rigidly control their collagen production. In cultures that main- tained a constant doubling time, this stability was maintained over at least 25 population doublings, sug-gesting that on the average, collagen production ap- pears to be tightly controlled and dissociated from the events and sequelae of cell division. The fibroblast is a differentiated mesenchymal

The fibroblast is a differentiated mesenchymal cell with characteristic morphologic and functional features. One major role of this cell is to produce collagen, an extracellular macromolecule that significantly influences organ structure and function. In soft tissue such as skin, lung, and gingiva, the fibroblast comprises at least 40 to 6 0 % of the cell population (1-3) and it is thought to be responsible for the majority of collagen present (4). Fibroblasts cultured from such organs manifest a characteristic functional state in vitro. Most importantly, they produce collagens I and I11 (5-7), quantitatively the major collagen types found in these tissues. In addition, since 2 to 10% of total protein production of soft tissues is devoted to collagen production (8, 9), and since cultures of fibroblasts derived from such organs also commit the same proportion of protein production to collagen (10,l l), * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. it is generally considered that such cells function similarly in vitro and in vivo.
The quantity of collagen in soft tissues has a significant impact on organ structure and function. Since under normal, steady state conditions, soft tissue fibroblasts are constantly undergoing renewal (12), yet the organ as a whole is producing unchanging amounts of collagen (9), it would be predicted that under conditions of a constant extracellular milieu, replicating populations of soft tissue fibroblasts continue to control rigidly the amount of collagen they produce.
To evaluate how well a replicating population of fibroblasts regulates the production of intact collagen molecules, the present study was designed to quantitate collagen production in cultures of soft tissue fibroblasts during periods of rapid and slow growth over many population doublings. To accomplish this, a diploid fibroblast strain derived from human lung, HFL-1, was maintained under conditions of a relatively constant extracellular milieu and was evaluated several times during more than 25 population doublings to determine the average number of collagen molecules each population of fibroblasts produced.
HFL-1 Cells-The cell strain used throughout these studies, designated HFL-1 (American Type Culture Collection CCL 153), is a fibroblast derived from the lungs of a male human fetus of 16-weeks gestation. HFL-1 has a normal karyotype, exhibits contact inhibition, and contains no mycoplasma or viral contaminants. Two vials of HFL-1 frozen in liquid nitrogen at 3 X lo6 cells/vial at Subcultivation 2 were used for all of the studies to be described, and are referred to as A and B, respectively. Vial A cells were studied at Subcultivations cultured in growth medium (DME medium containing 10% fetal calf serum, 100 units/ml of penicillin, 100 pg/ml of streptomycin, 5 pg/ml of Fungizone, and 0.06% glutamine) in a 37°C incubator with 90% air, 10% Con. To obtain cells at Subcultivation 10, 15, and 25, cells were seeded in 100-mm plastic Petri dishes at 4 X lo6 cells/plate in 10 ml of growth medium and the medium was changed 4 days later. On Day 7, when there were approximately 12 X lo6 cells/plate, the cells were subcultivated in a 1:3 split using 0.25% trypsin (3 &/plate, 10 min, Cell Number, DNA Synthesis, and Mitotic Index-To determine the cell number in each plate, the cells were removed from one plate with trypsin, diluted to a fixed volume with DME medium, and an aliquot was mixed with an equal volume of 0.25% trypan blue solution. The number of cells was counted in a hemacytometer; in all cases greater than 95% of the cells were viable. DNA synthesis was quantitated by rinsing duplicate plates twice with 5 ml of phosphate-buffered saline and incubating for 4 h with ['Hlthymidine-labeling medium (growth medium without fetal calf serum but containing 50 pg/ml of ascorbic acid and 1 pCi/ml of ["Hlthymidine). At the end of the incubation, the cells were rinsed with phosphate-buffered saline, removed from the plate with trypsin, precipitated with 5% trichloroacetic acid, and counted in 10 ml of Bray's scintillator. DNA was quantitated on parallel plates by a modification of the method of Burton (14,15) and DNA synthesis was expressed as disintegrations per min of ['Hlthymidine incorporated in 4 h/pg of DNA.
To determine the mitotic index of HFL-1 at different subcultivations, cells were grown in 150-mm plates containing glass microscope slides. At 4 and 10 days after subcultivation, the slides were removed, placed in a mixture of DME medium:H20 (1:3 v/v) for 5 to 10 min. The cells were then fixed (methano1:acetic acid, 3:l v/v) for 5 min, air-dried, and stained with Giemsa. The number of mitoses in 500 cells was counted and the mitotic index expressed as the percentage of cells in mitosis.
Labeling of HFL-1 Cells-To evaluate collagen production in HFL-1 cells at different subcultivations and at different times within each subcultivation, the following general design was used. At Subcultivations 9, 14, and 24, confluent cells were removed from the plates and seeded at 5 X lo5 cells/plate in 100-mm plastic Petri dishes with 10 ml of growth medium. The growth medium was changed every other day. On Days 4,6,8, 10, 12, and 14 (Subcultivation lOA), and Days 4 and 12 (Subcultivations lOB, 15A, 15B, and 25A), the growth medium was removed from the cultures, the plates rinsed twice with phosphate-buffered saline, and the cells reincubated for 4 h at 37°C with 4 ml of labeling medium (DME medium containing 5 pCi/ml of ["Clproline, 50 pg/ml of ascorbic acid, 100 units/ml of penicillin, 100 pg/ml of streptomycin, 5 pg/ml of Fungizone, and 0.06% glutamine).
Measurement of the Specific Activity of ('4C/Proline in the Free Intracellular Proline Pool-Following the labeling period, the medium of one to three plates was discarded, the cell layers were placed on ice, rinsed eight times with phosphate-buffered saline at 4"C, and harvested together by scraping with 5 ml of 1% picric acid. After allowing the proteins to precipitate overnight at 4°C. the precipitate was pelleted (1200 X g, 10 min) and the supercntant containing the free intracellular proline pool was passed through disposable prefded 100 to 200 mesh AG 1-X8 columns to remove the picric acid. The columns were washed twice with 5 ml of Hz0 and the combined eluates lyophilized. This material was solubilized in 800 p1 of 0.2 N citrate buffer, pH 2.2, and aliquots were fractionated on a Beckman model 119C amino acid analyzer to quantitate the disintegrations per min and nanomoles of proline in the intracellular free proline pool. To measure the disintegrations per min of ["Clproline, 1.4-ml fractions were collected in the region of the proline peak and counted in 10 ml of Aquasol with corrections made for the disintegrations per min lost because of the reaction of the proline carbon atoms with ninhydrin (11%) and the quenching by citrate (4 to 6%). Specific activity of the free proline pool was expressed as disintegrations per min ['4C]proline/mol of proline. For each day at each subcultivation, it was measured in triplicate.
Quantitation of Collagen Production/Cell-Six plates were labeled as described above and paired into three sets.  (14,16). The per cent hydroxylation of proline residues in newly synthesized collagen was determined using parallel aliquots of the dialyzed sample. Each aliquot was placed in a 12-ml conical tube; 50 pl of clostridial collagenase was added to one tube; the other tube served as a blank. Both samples were incubated at 37°C for 16 h. At the end of the incubation period, 2 mg of bovine serum albumin (40 mg/ml) was added to each tube followed by 0.9 ml of trichloroacetic acid/sodium tungstate (17%/0.42%). The samples were blended on a Vortex mixer, kept at 4°C for 18 h, and the precipitated proteins were pelleted (1200 X g, 10 min). An aliquot of the resulting supernatant was used to measure [14C]hydroxyproline using the Juva-Prockop method and additional aliquots were used to measure total counts of DNA was measured in 2-ml aliquots (14,15). Values for DNA/cell (9.2 f 0.88 pg/cell) were similar for all subcultivations and at different times within each subcultivation ( p > 0.6 all comparisons). Thus, cell number and DNA were used interchangeably throughout the study. In general, most estimates of cell number were made by quantitating DNA and then converting to cell number using the value 9.2 pg of DNA/cell. There were two reasons for this: 1) DNA could be quantitated from the same cell homogenate as the other parameters used to measure collagen production, whereas quantitation of cell number required parallel plates; and 2) in our experience, the variation of DNA measurement was less than that for cell number.
Using these measurements, the rate of collagen production/cell was calculated as follows: rate of collagen production/cell = ((total dpm ['4C]hydroxyproline in the culture)/(% hydroxylation of proline residues in the collagenous regions of newly synthesized collagen))/ [((pg DNA)/(9.2)) ((specific activity of ["Clproline in the free intracellular proline pool in dpm/mo1)/(6.02 X loz3)) (labeling time in hours) (241)l. The rationale underlying derivation of this formula is detailed in the appendix. "Labeling time" was 4 h in all experiments unless otherwise noted. The units of rate of collagen production/cell were number of procollagen chains produced/h/cell. Per Cent Collagen Production-Per cent collagen production was quantitated using aliquots of the same dialyzed samples used for quantitation of the rate of collagen production/cell. The disintegrations per min of [14C]proline incorporated into all proteins was quantitated from 100-p.l aliquots of the dialyzed sample using trichloroacetic acid precipitation onto Millipore filters (17). The disintegrations per min of newly synthesized proteins sensitive to clostridial collagenase was taken from the term "((total dpm in the supernatant of the collagenase-treated sample) -(total dpm in the supernatant of the blank))" used in the quantitation of per cent hydroxylation of proline residues in newly synthesized collagen (see above). The efficiency of collagenase in solubilizing the collagenase regions of the newly synthesized procollagen was calculated as follows: efficiency = ((dpm [ "C]hydroxyproline solubilized by collagenase) -(blank))/((dpm ['4C]hydroxyproline in the untreated homogenate) -(blank)).
Evaluation of HFL-1 for Active Mammalian Collagenuse-Using the 14C-labeled collagen fibril assay, no collagenase was detected in the media of rapidly growing or confluent HFL-1 (data not shown). In addition, when a known quantity of 14C-labeled type I or type III collagen was incubated with HFL-1 cultures for 4 h, all of the labeled procollagen remained intact (data not shown). Thus, under the conditions used in this study, HFL-1 did not produce an active collagenase.
Relationship  moles of free proline within the cell (Fig. 3A), it doubled the total disintegrations per min of free [14C]proline within the cell (Fig. 3B). Thus, when HFL-1 cells were incubated in the presence of 10 pC!i/ml of ["Clproline in the culture media, the specific activity of [Wlproline in the intracellular free proline pool was twice as high as when the cells were incubated in the presence of 5 &i/ml of ['%]proline in the media (Fig. 30. This apparent linear relationship between the specific activity of the intracellular free proline pool and the concentration of in the culture media held over a wide range of culture media tracer concentrations. It was not possible to alter this relationship by increasing the quantity of unlabeled proline in the culture medium (i.e. decreasing the specific activity of the isotope in the culture medium).
Thus, the specific activity of the intracellular free proline pool in HFL-1 could not be "saturated" with either labeled or unlabeled proline such that it became independent of the concentration or specific activity of the ['4C]proline in the culture medium. If proline incorporation was not expressed in relation to the intracellular [%]proline specific activity, the apparent rate of protein production in HFL-1 cultures labeled with 10 &i/ml of [%]proline was approximately 2-fold greater than HFL-1 cultured in 5 @i/ml of [14C]proline (Fig. 4A). In comparison, ifproline incorporation was expressed in relation to the intracellular [14C]proline specific activity, the apparent rate of protein production by HFL-1 cells was independent of the concentration of ['"Clproline in the culture medium (Fig. 4B).
It might be argued that even though the measured rate of incorporation of [14C]proline into protein is dependent on the concentration of the tracer in the culture medium, if the same concentrat.ion of tracer was used in all studies, the problem would be alleviated (i.e. even though the measured rates of protein production were dependent upon the amounts of tracer used, comparisons between measured rates would still be valid). This, however, was not the case, since with identical concentrations of [%]proline in the culture media, HFL-1 cells had markedly different specific activities of ['4C]proline in the free intracellular proline pool at different times after subcultivation (Fig. 5). For example, at Subcultivation lOA, Day 4, the specific activity of the intracellular proline pool was 29 X lo3 dpm/nmol, whereas at Subcultivation lOA, Day 14, the specific activity of the intracellular proline pool was 13 X lo3 dpm/nmol. This 2-to 3-fold decrease in the specific activity of the free intracellular proline was consistent from subcultivation to subcultivation and from cells derived originally from the two different frozen vials of HFL-1.
The two parameters which govern the specific activity of free intracellular proline (the size of the free intracellular proline pool and the amount of uptake of ['4C]proline from the media into the cell) both contributed to the 2-to 3-fold decrease in intracellular proline specific activity observed in HFL-1 with time after subcultivation (data not shown). However, since the uptake of the intracellular free proline pool at different times after subcultivation was remarkably consistent from subcultivation to subcultivation (Fig. 5). For example, even though there was more than a 2-fold decrease in specific activity from Day 4 to Day 12 in all subcultivations evaluated, the coefficient of variation of the specific activities at different subcultivations was 8.8% at Day 4 and 22% at Day 12.
The validity of using specific activity of the free intracellular proline pool was further confirmed by independent measurement of the specific activity of proline in the newly synthesized collagen. Increasing the specifk activity of the free pool 3.8fold resulted in a 3.6-fold increase in specific activity of proline in the collagen product. Thus, the specific activity of the free intracellular proline pool is likely a very close approximation to the specific activity of ['4C]proline in prolyl-tRNA.
When HFL-1 cells were fully confluent (Day 12 after subcultivation), the majority of newly synthesized collagen chains contained proline residues that were fully hydroxylated (Fig.  6). In comparison, the proline residues in newly synthesized collagen were underhydroxylated when cells were in the log phase of their growth curve, independent of the fact that significant quantities of ascorbic acid were used in the labeling medium. This trend for low levels of hydroxylation in the log phase of growth and more complete hydroxylation in stationary phase was consistent from subcultivation to subcultivation; in general, hydroxylation at Day 4 was approximately 60% that of Day 12, independent of subcultivation number (Fig. 6 ) . Per cent hydroxylation in collagen produced by HFL-1 cells was unaffected by: 1) the addition of more ascorbic acid to the labeling medium; 2) "preincubating" the cells in ascorbic acid prior to the addition of labeling medium; 3) having ascorbic acid in the growth medium as well as the labeling medium; or 4) using ascorbic acid from different suppliers. Quantitation of the levels of active prolyl hydroxylase per cell in HFL-1 at various times after subcultivation revealed a level at Day 12, 2-fold greater than that found at Day 4. At Day 12, there was sufficient active prolyl hydroxylase to hydroxylate 47.2 pg of proline/lOs cells in 30 min, whereas at Day 4, the active prolyl hydroxylase in 10' cells only hydroxylated 20.5 pg of proline in 30 min.
Because underhydroxylation of proline residues has been recognized in other cell strains used to study collagen production in vitro, many investigators have turned to the use of clostridial collagenase to quantitate collagen production. However, at least for HFL-1 cells labeled for 4 h, collagenase usually did not solubilize consistently 100% of the hydroxyproline-containing regions of the newly synthesized collagen (Fig. 7). It is clear, therefore, that dependence on the collagenase method alone to quantitate collagen production in represents 100% solubilization of newly synthesized collagen. The relative efficiency of collagenase was unaffected by increasing the amount of enzyme or the time of incubation with the enzyme. In comparison to the variability of this enzyme preparation in completely cleaving newly synthesized HFL-1 collagen in the presence of the entire culture contents, the same enzyme cleaved >98% of ['*C]proline-labeled type I or type I11 collagen purified from the media of HFL-1 cells.

5255
HFL-1 cells could result in underestimation of the rate of collagen production by as much as 20%. This observation was i f , 12 not dependent on subcultivation number or the time after subcultivation. It did, however, seem to be partially dependent  Because of the problems inherent in relying on either the hydroxyproline orthe collagenase methods to quantitate collagen production in HFL-1, all subsequent measurements utilized a combination of both methods (see "Experimental Procedures" and "Appendix" for details).
Control of Collagen Production by HFL-1-From Subcultivation 10, Day 0, to Subcultivation 25, Day 12, HFL-1 cells underwent approximately 25 population doublings. During that period, HFL-1 produced collagen at an average of 6.3 X lo5 f 0.41 procollagen chains/h/cell (Fig. 8). Comparison of the rate of collagen production per cell at different days after subcultivation and at different subcultivations demonstrated that collagen production by HFL-1 was independent of both parameters. Thus, under conditions of a defined, relatively constant extracellular milieu, HFL-1 rigidly controlled collagen production such that it was invariant over at least 25 population doublings. This was true even when different populations of HFL-1 were used to start the experiment since the invariance of collagen production was observed for cells derived from different vials of HFL-1 that had been frozen at Subcultivation 2.
Evaluation of the proportion of total protein production represented by collagen over the same period demonstrated that per cent of collagen production remained relatively constant at 3.72 & 0.21%. However, although per cent of collagen production did not vary from subcultivation to subcultivation ( p > O.l), there was a small, but sigmkant, difference between per cent of collagen production at Day 4 after subcultivation and Day 12 ( p < 0.05) (Fig. 9). This difference was only seen when all data were averaged, as there was no difference for Subcultivations lOB, 15A, and 25A between Days 4 and 12 ( p > 0.1, all comparisons). The small difference between Days 4 and 12 that was seen when all data were averaged resulted from a slight decrease in non-collagen protein production at  Day 12, since there was no change in collagen production (Fig.   8).

DISCUSSION
The present study demonstrates that under conditions of a relatively constant extracellular milieu, not only does a population of replicating diploid fibroblasts derived from a human soft tissue continue to produce collagen, but the cells do so at the same level as when they are not replicating. In addition, in cultures that maintain a constant doubling time, this level of collagen production remains stable over at least 25 population doublings.

Quantitation of Collagen Production
Evaluation of the Specific Activity of the Labeled Amino Acid Precursor-When isotopic tracers are used to quantitate rates of protein synthesis in biologic systems, true assessment of the average rate of protein production per cell necessitates knowledge of the specific activity of the precursor amino acid just prior to its incorporation into protein. There are various general approaches that can be used to estimate this value, but for several reasons, we accounted for variations in the specific activity of the precursor amino acid by quantitating the specific activity of [14C]proline in the free intracellular amino acid pool (20,21).
First, the specific activity of ['4C]proline in the intracellular free proline pool is proportional to the specific activity of proline actually incorporated into protein by HFL-1 cells. Under steady state conditions, when the specific activity of the intracellular proline pool is doubled, the disintegrations per min of newly synthesized protein is doubled, but the actual quantity of newly synthesized protein is unchanged (Figs. 4 and 5). Thus, even though there is controversy concerning the actual amino acid pool utilized as the source of aminoacyl-tRNA (20,(22)(23)(24)(25)(26), quantitation of the specific activity of the free intracellular proline pool is a valid estimate of the specific activity of proline residues being incorporated into newly synthesized protein.
Second, measurements of the specific activity of [I4C]proline in the free intracellular proline pool pennits evaluation of the specific activity of ["C]proline residues incorporated into protein without perturbing the culture conditions. In contrast, attempts to expand the intracellular amino acid pool (so that it will not change during variable experimental conditions) may significantly alter normal cellular function including rates of protein production (22,(27)(28)(29).
Third, the use of the free intracellular amino acid pool to evaluate the specific activity of the tracer amino acids can be carried out with reasonable amounts of biologic material.
Approximately lo7 cells are needed to make three reproducible measurements of the specific activity of ['4C]proline in the intracellular free proline pool, less than that needed to quantitate the specific activity of [I4C]proline in prolyl-tRNA.
Fourth, there is little change in the total amount of free intracellular proline during the labeling period. In 4 h, cultured fibroblasts have little turnover in their free intracellular proline (30). Proline is a nonessential amino acid; fetal lung fibroblasts synthesize it from glutamine, arginine, and ornithine (30). Proline from degraded proteins can be recycled into new collagen (31), but during a 4-h labeling period, this pathway contributes negligible amounts of free proline to the general pool (30). Thus, during a 4-h labeling period, the specific activity of the intracellular proline pool is dependent on only two parameters: the size of the free proline pool and the rate of transport of the labeled proline from the media into the cell. Since the latter process is rapid (32-34), the specific activity of the intracellular proline pool rapidly approaches a constant level which will reflect the specific activity of proline incorporated into protein synthesized during the labeling period.
Evaluation of the specific activity of the free intracellular proline pool in HFL-1 cells during different conditions demonstrates that it is significantly higher during the log phase of growth than in stationary phase. Furthermore, this more than 2-fold variation in specific activity during cell growth is a constant finding for at least 25 population doublings (Fig. 5). These findings are consistent with the observations in other cell lines that: 1) intracellular proline pool sizes are smaller in log compared with stationary phase (30, 33); 2) " A ' system amino acids (e.g. proline) are transported into the cell more rapidly in log than in stationary phase (33, 34); and 3) the growth phase variations in intracellular free proline pool sizes and "A" system transport rates are consistent over many population doublings (35).
These observations are important in evaluating the rates of protein production in cultured cells during periods of cell growth. Since there are marked differences in the specific activity of intracellular free proline in the log and stationary phases of fibroblast cultures, and since the specific activity of intracellular free proline is proportional to the specific activity of the proline residues incorporated into newly synthesized protein, the approach used circumvents possible misinterpretations of experimental data inherent in using labeled tracers to quantitate protein production.
Evaluation of the Amount of Tracer Incorporated into Collagen-When it was recognized that hydroxyproline was a relatively specific marker for the newly synthesized collagen molecule, quantitation of the incorporation of ['*C]proline into ['4C]hydroxyproline became the standard approach to evaluate rates of collagen production in cultured cells (36). It is now generally recognized, however, that the processes of collagen synthesis and prolyl hydroxylation are not necessarily coupled (37). The hydroxylation process is highly sensitive to environmental factors (e.g. cell crowding, lactic acid, oxygen tension, ascorbate availability) which do not inevitably have a coordinate effect on collagen production (38-41). A consistent finding in studies of collagen production by cultured fibroblasts is that in periods of rapid growth, the relative hydroxylation of newly synthesized collagen molecules is significantly less than that which occurs when the cells are confluent (37). This is also true for HFL-1, in which the average per cent of hydroxylation of prolyl residues is 27 -+ 3% for collagen synthesized during the log phase of growth while it is 42 * 2% when the cells reach confluency (Fig. 6).
The reason for the relatively lower hydroxylation during rapid growth is not completely defined, but it is associated with the relative quantities of active prolyl hydroxylase present. Independent of the mechanisms for this phenomenon, it causes a major problem in using labeled hydroxyproline as an estimate of collagen production, since the relative deficiency of prolyl hydroxylation in periods of rapid growth results in an underestimation of the quantity of collagen produced.
The collagenase method for collagen quantitation was developed to obviate the problem concerning the lack of coordination of collagen production with hydroxylation of prolyl residues (9). The collagenase method is an excellent one, but for it to be used successfully in quantitating collagen production by cultured cells, two conditions must be met. First, the enzyme used must specifically cleave only collagen and not non-collagen proteins, and second, the enzyme must cleave all of the newly synthesized (i.e. labeled) collagen in the culture. In usual practice, collagenase is considered adequate for use if it completely degrades a purified ['4C]proline-labeled collagen substrate but does not degrade a [14C]tryptophan-labeled non-collagen substrate (9). While such testing fulfas the first condition, it does not f u l f d the second, as complete degradation of purified collagen in solution is different than completely degrading labeled procollagen in the presence of the cellular debris found in homogenates of cell culture. For HFL-1 cells, purified clostridial collagenase was often less than 100% efficient in digesting all of the collagen synthesized by the cultures (Fig. 7). This variability could not be minimized by altering the amount of enzyme, the source of enzyme, time of incubation, or other incubation conditions (e.g. Ca", Nethylmaleimide, or NaCl concentrations) and it persisted even though the enzyme consistently cleaved >98% of pwifed ['4C]proline-labeled type I or type I11 collagen isolated from the media of HFL-1 cells. There are two possible reasons for this variability: 1) clostridial collagenase may not be able to attack the a! chain-like hydroxyproline containing region of the N-propeptide portion of procollagen (see "Appendix"); and 2) other cell products (e.g. fibronectin) may bind to regions of the collagen molecule and protect them from collagenase attack. However, independent of the mechanisms limiting the efficiency of collagenase, the observed variability points out the problems in relying only on collagenase-released material for quantitatively assessing collagen production by these cells.
To circumvent the possibility of underestimating collagen production using the hydroxyproline or the collagenase methods, we combined the two methods and capitalized on the fact that although clostridial collagenase does not always digest 100% of the newly synthesized collagen, the purified enzyme specifically degrades only collagen, and hence, the material released by the enzyme can be used to measure accurately the per cent of hydroxylation of proline residues in newly synthesized collagen. These data, together with the value for the total ['4C]hydroxyproline present in the culture, can be used to determine the total amount of ["C]proline incorporated into collagen, independent of the proportion of incorporated prolyl residues that are actually hydroxylated (see "Appendix"). Together with measurements of DNA and the specific activity of the free intracellular [I4C]proline, this approach permits an accurate assessment of the number of procollagen chains produced per cell per unit time.

Fibroblast Collagen Production 5257
Collagen "Production " versus Collagen "Synthesis " The production of intact, functional collagen is a complex process involving a number of post-translational modifications of the newly synthesized molecule prior to its eventual incorporation into a mature collagen fibril. Recent evidence suggests that mechanisms also exist within the cell to degrade completely a significant proportion of the newly synthesized collagen molecules before they leave the cell (42,43). While the mechanisms and function of intracellular collagen degradation are still to be defined, it is clear that there is a significant difference between the number of collagen molecules "synthesized" by fibroblasts and the amount of intact collagen chains actually "produced" by these cells. Although intracellular degradation may play an important role in controlling the quantity and fidelity of collagen produced per cell, from the viewpoint of evaluating the stability of a population of cells with respect to collagen, the important parameter is not how much collagen is synthesized, but rather the actual net production of collagen. In the present study, the experimental design was specifically constructed to evaluate only net collagen production, i.e. all homogenates of cultures were dialyzed to remove degraded collagen fragments prior to quantitation of the amount of labeled collagen present (42,43).

Stability of Collagen during Cell Growth
The present study demonstrates that not only does a population of fibroblasts produce collagen during periods of cell proliferation, but the population of cells does so at the same average rate per cell as when the cells are quiescent. It is dX1cult to compare these results to studies with other fibroblast types, because early investigators did not account for the decrease in per cent prolyl hydroxylation in the log phase of growth (36), and no other studies have accounted for the intracellular specific activity of the tracer used to quantitate collagen production.
The fact that a population of fibroblasts produces collagen at the same average rate per cell during periods of rapid and slow growth does not prove that each fibroblast produces collagen at a constant rate throughout its own cell cycle. Although studies with synchronous cultures have suggested that fibroblasts produce collagen during both the S and G2 phases of the cell cycle (44, 45), it is possible that levels of collagen production during S and G2 are different from that during G1, or even that the level of collagen production varies through all phases of the cell cycle. However, the fact remains that such possible variations do not seem to influence the average per cell level of collagen production by the entire population.
In addition to the fact that the average rate of collagen production per cell is independent of the growth state of the fibroblast population, these studies also suggest that, except for small differences in some subcultivations, collagen production by HFL-1 cultures is generally coordinate with overall protein production during the same periods. Similar coordination between collagen production and overall protein production has been found in replicating fibroblasts from other species, or organs, or both, including chick embryo fibroblasts (ll), L929 cells (46), 3T6 cells (44), human foreskin (FSlO) fibroblasts (45), chick tendon fibroblasts (47), and rat PR105 cells (44).

Stability of Collagen Production over Many Population Doublings
Under usual culture conditions, serially cultured human diploid fibroblasts derived from soft tissues demonstrate a finite in vitro life span. Three phases of growth have been described. Phase I, the time in which the culture is initiated; Phase 11, the major life span of the cell; and Phase 111, in which cell senescence becomes manifest. For most human fetal fibroblasts, Phase I11 begins at 40 to 45 population doublings. While the mechanisms responsible for initiation of Phase 111 are unclear, this phenomenon is generally regarded as a model of the in vitro aging process at the cellular level (48).
The present study was designed specifically to examine collagen production in human fetal fibroblasts during Phase 11, when the cells are likely representative of the majority of fibroblasts in a healthy, normally functioning soft tissue organ. The finding that HFL-1 produces collagen at a constant rate per cell from Subcultivations 10 through 25, a period representing approximately 25 population doublings, is consistent with the observation that cultures of human fibroblasts from a variety of donors and soft tissue organs accumulate a similar amount of collagen per week for at least 25 to 40 subcultivations (49,50).
The invariance of the average rate of collagen production by HFL-1 over at least 25 population doublings does not prove that all cells comprising the culture are producing collagen at the same level. Although there is evidence that all fibroblasts comprising in vitro cultures are actively producing collagen (5,51,52), current technology does not permit accurate assessment of the actual net production of collagen per cell on the single cell level. However, it is of interest that, even though cells comprising fibroblast cultures have different interdivision times (53) and manifest differences in clone size within 10 days (54,55), populations of soft tissue fibroblasts maintained under a constant environment produce collagen at a constant average rate per cell.

"Soft Tissue" versus "Firm Tissue" Fibroblasts
HFL-1 fibroblasts were derived from the lung parenchyma of a 16-week-old human male fetus. Like fibroblasts derived from other soft tissues such as skin and gingiva, HFL-1 has specific characteristics which distinguish it from f i i tissue fibroblasts such as those from tendon or bone. HFL-1 produces both collagens types I and I11 (data not shown) while fm tissue fibroblasts produce only type I (56). Like most other soft tissue fibroblasts, HFL-1 devotes less than 10% of its total protein production to collagen while tendon and bone fibroblasts dedicate as much as 50% of their total protein production to collagen (57).
The present study also points out another important difference between soft and fm tissue fibroblasts: recent evidence suggests that f i i tissue fibroblasts do not have sufficient internal controls over collagen production to maintain normal in vivo levels of collagen production in vitro. For example, even though 50% of the total protein product of tendons is type I collagen, fibroblasts derived from tendon, after a few days in culture, produce significantly less collagen (58,59). Interestingly, these cells can be reverted to their normal in vivo levels of collagen production by specific environmental conditions (58). In contrast, populations of soft tissue fibroblasts maintained in vitro do not show a drastic decline in per cent of collagen production and, in addition, do not respond to environmental influences such as ascorbate or serum concentration by increasing collagen production.
It should be kept in mind, however, that it is possible that primary cultures of pure fibroblasts might have a far higher rate of collagen production and proportion of total synthesized protein than does a fibroblast strain passaged 10 or more times. In addition, the present study was conducted with human fibroblasts derived from a 16-week-old fetus, while studies with tendon fibroblasts have been performed with 5258 Fibroblast Collagen Production cells derived from chick embryos. These differences in the species and developmental age may be as important as, or more important than, the organ from which the cells are derived.
during dialysis, at this time it is best to omit this region in the calculations.
If the leader sequence were retained by the dialysis step, and if it had the same average number of amino acids as the N-propeptide (both of which are unlikely), omitting this region from the final calculations would result in underestimates of collagen production of less than 5%.

Inherent versus Enuironmental Control of Collagen Production
The present study was designed to minimize possible influence of the extracellular milieu on collagen production by HFL-1. Under such conditions, it is clear that the production of collagen is stable and independent of cell growth and population doublings. However, these findings do not imply that soft tissue fibroblasts are incapable of altering collagen production.
Several studies have demonstrated environmental control of collagen production including mediators of inflammation (7), pharmacologic agents (60), proteases (61), hormones (62), lipids (63), and nutrients (64 When the hydroxyproline and collagenase methods were first applied to quantitate collagen production in cultured cells, the formulas used were based on the concept that the collagen a chain was the primary translation product of collagen mRNA (36,46). However, it is now known that collagen is initially made in a precursor form with large NH2-and COOH-terminal extensions to the a chain. Since the amino acid composition of these extensions is markedly different than that of the a chain, accurate assessment of the rate and percentage of collagen production by cultured cells requires consideration of the structure of the so-called "(pre)pro a chain" (Fig. 10).
Region a-A leader sequence (also called the "pre piece" or "signal peptide") has been described at the NH2 terminus of the primary translation product of chick al(I) mRNA (65,66). It is at least 19 amino acids long; estimates from sodium dodecyl sulfate gel electrophoresis suggest it may contain 50 to 100 amino acids. No data are available on leader sequences of other collagen chains. Since this region is probably very transient during biosynthesis, may be rapidly degraded intracellularly, and likely is removed from the cell homogenate Region &The NH*-terminal globular region (also called the "Co1 1" region) of the N-propeptide of the pro-a chain (for types I and III). This region is 53 to 100 residues in length (67); it contains 8 to 14 imino acids, all of which are proline (67).
Region c-The collagenous region (also called the "Co1 3" region) of the N-propeptide (68) is 26 to 41 residues long and it contains hydroxyproline (5 to 9 residues) as well as proline (2 to 8 residues) (67). It has been suggested that in the intact procollagen form, Region c may not be susceptible to clostridial collagenase (65). This may partially account for the variability in the efficiency of collagenase in cleaving HFL-1 collagen (Fig. 7); it is not known to what extent Region c of HFL-1 pro I and pro III chains is digested under the conditions used in this study. However, independent of this variability, the hydroxyprolme residues in Region c will always be accounted for in the measurement of ['4C]hydroxyproline.

Region d-This
region is a globular segment (also called the "Co1 2" region) on the COOH-terminal portion of the Npropeptide (68). Little is known about Region d except that it is short (5 residues in the pro al (I) chain, 28 in the pro al (III) chain), and contains few imino acids (67, 68). The NHa-terminal procollagen peptidase cleaves between Regions d and e.
Region e-This is the NHz-terminal portion of the final a chain and is often called the "teleopeptide." For al(I) and a2 chains, Region e is 19 residues in length and contains one to two imino acids (67).
Region f-The collagenous a chain. Region f contains 81% of the imino acids in pro al(I), 98% in pro a2, and 85% in pro al(II1).
Region g-The COOH-terminal teleopeptide of the final a chain. Like the NH*-terminal teleopeptide, it is short, containing, at most, 25 amino acids of which 3 are imino acids.
Region h-The C-propeptides of the procollagen chains contain between 300 and 350 residues (69). For procollagens I and III, less than 4% of these residues are proline, and none are hydroxyproline.

Quantitation of the Rate of Collagen Production
The rate of collagen production per cell, y, can be quantitated using the formula: y = (x/p)/(r/t)(q/u)(s)(u) where:  for most of the experiments with HFL-1, s = 4 h, u = the average number of imino acids in the colIagenous regions (Regions c + f) of pro-a chains. The term "u" converts the units of the rate of collagen production from "molecules of proline incorporated into the collagenous regions of collagen per cell per h' to "molecules of collagen produced per cell per h." For pro-al (I) there are 253 imino acids in Regions c + f; for a2 there are 217, and for pro al(II1) there are 245. Since HFL-1 make approximately 90% type I, 10% type III, and type I contains 2 pro al(I) chains/pro a2 chain, there is an average of 241 imino acids in Regions c + f of an average pro-a chain produced by HFL-1. However, varying the ratio of type I to type III actually introduces little error; if HFL-1 produced only type III, failure to change the term "a" would result in less than a 3% increase in the calculation of the rate of collagen production. The units of y are: number of procollagen molecules produced per h per cell. The assumptions in the above formulation are as follows: 1. It is assumed that the specific activity of free intracellular ['4C]proline represents the true specific activity of ['4C]proline just prior to its incorporation into protein; that DNA and cell number are directly proportional; and that incorporation of ['4C]proline into protein by HFL-1 is linear with time for more than 4 h. The validity of each of these assumptions is discussed in the text.
2. The formulation is independent of whether the pro-a chains are in the triple helical or denatured form or whether they are intact pro-a chains or processed (by procollagen Nand C-proteases) to an intermediate or final a chain. It is assumed, however, that if processed, the intact N-propeptide (Regions b + c + d) is retained by dialysis, and thus the hydroxyproline residues in Region c are quantitated when measuring the total ['4C]hydroxyproline in the post-dialysis culture material.
3. It is assumed that the measured per cent of hydroxylation represents the average value for Regions c + f. Whereas, it is known that Region f is completely solubilized by collagenase, Region c may not be (depending on the form of the pro-a chain) (65). Thus, the value for "p" may be derived only from Region f and not from the average of both regions. However, the number of residues in Region c is so few compared to Region f, this assumption would have little consequence if Region c were not solubilized.
In addition, it is also assumed that the noncollagenous Regions d + e are not solubilized by collagenase. However, if collagenase completely solubilizes Regions c and f, Region d + e might be solubilized (it is not known whether under the conditions used, trichloracetic acid will precipitate a peptide the size of Region d + e; for procollagens I and III, Region d + e is less than 25 residues). If Region d + e were solubilized, this would reduce slightly the value for per cent of hydroxylation since the 2 or 3 proline residues in this region would be added to the denominator in the determination of per cent hydroxylation of the collagenous regions.
4. For the value "u," it has been assumed that pro-a chain composition of human, sheep, calf, and chick collagens are comparable; when possible, composition data were used from the species closest to human. 5. It is assumed that hydroxyproline is specific for collagen. For HFL-I, this assumption is valid. Human fibroblasta do make CIq, a hydroxyprolme containing protein, but the hydroxyprohne content of Clq is far less than procollagen and fibroblasts synthesize less than 10m3 Clq molecules/procolh+ gen molecule (70).

Quantitution of Per Cent Collagen Production
The commonly used formula for per cent of collagen pro- However, as pointed out by Rowe et al. (59), this formulation ignores the fact that collagen chains are produced as pro-a chains, and thus, a significant proportion of the N-and Cpeptides will be falsely counted as non-collagen (i.e. not solubilized by collagenase) and will not be counted as collagen. Thus, Regions b, c, d, and h must be accounted for in this formula. If they are degraded to dialyzable peptides by the tissue during the labeling period they can be ignored, but available evidence (71) suggests that in human fibroblast cultures, there is minimal cleavage of the precursor collagen peptides.
From these considerations, per cent of procollagen production was quantitated Regions c + f). From these data, G is equal to: (J/K) + 0.1 (J/K) which is equivalent to 1.1 (J/K). The term "0.1 (J/ K)" is the total disintegrations per min of imino acids in Regions b + d + e + g + h; the term "1.1&Z/K)" is the total disintegrations per min of imino acids in the entire pro-a chain. Z = 3.71, the relative number of imino acid residues in procollagen compared to non-collagen. The number for procollagen is averaged from the available data for collagens type I and III (66-70); the number for non-collagen is taken from direct measurement of human fetal lung noncollagen (9). H = total disintegrations per mm of imino acid incorporated into non-collagen during the labeling period. His quantitated from: 3) Regions d + e are not solubilized by collagenase; 4) the composition data for human pro-a chains are similar to the species used; and 5) hydroxyproline is specific for collagen. Each of these assumptions is discussed above.