Detection and Partial Characterization of Collagen Synthesis Stimulating Activities in Rat Mammary Adenocarcinomas*

Rat mammary epithelium and adenocarcinomas de- rived from it synthesize type IV collagen, a structural protein of basement membranes. In cultures of cells, net production of collagen is stimulated 2-fold more than total cell protein by epidermal growth factor. Mammary adenocarcinoma cells also respond to epidermal growth factor but to a much reduced extent. This difference in growth factor responsiveness appears to be due to the production of collagen synthesis stimulating factors by the mammary tumor cells. Such factors have been partially purified and shown to differentially stimulate the incorporation of proline into collagenase-sensitive protein by 2.5-10-fold in normal rat mammary epithelium, normal rat kidney, and mouse 3T3 cells. The tumor factors do not stimulate net collagen production in cultures of tumor cells from which the factors are derived, suggesting that tumor cells produce sufficient stimulatory factors for optimal synthesis of collagen. Pulse-chase studies indicate that the tumor factors stimulate collagen synthesis rather than block collagen turnover. The activities in the ex- tract have been partially purified by gel filtration, ion exchange column chromatography, and isoelectric fo- cusing. The major species has a molecular weight of about 68,000 and a PI of 5.9. A smaller peak of activity with a molecular weight of 6,000 is also present. Since collagen synthesis appears to be necessary for the growth of mammary adenocarcinomas in vivo, production of these collagen synthesis stimulating factors may be important for tumor growth. The 7.5% SDS-polyacrylamide gels and stained according to Laemmli (10) to determine the nature of the proteins in the active fractions. Effects on Collagen Production and Cell Growth-Either soluble tumor extracts or column fractions prepared from them as described were added to the various cell cultures. The effects on cell growth were determined three days after factor addition during which the cells were in log phase. Control cultures received an equivalent amount of phosphate-buffered saline. Effects on collagen production were assessed either by determining the amount of labeled amino acid ([3H]lysine or ['4C]proline, 2 pCi/ml) incorporation into collagenase sensitive protein according to methods described by Peterkofsky and Diegelmann (11) or by determining the amount of hydroxyproline and hydroxylysine mass relative to cell number or to total lysine in the cell protein fraction. In the former a protease-free collagenase (Form Biofactures, Inc., Lynnbrook, NY) was utilized as (3). In the latter the cell layers were recovered from the culture dishes and these were solubi- lized in 0.1 M NaOH. After digestion for 30 min at 37 "C, the sample precipitated with trichloroacetic acid, 1% tannic acid (11). The precipitate was washed two times with the same and then extracted to remove the trichloroacetic acid. The sample was hydrolyzed in 6 N HCl for 24 h at 110 "C in sealed tubes in a nitrogen atmosphere. After decoloring with activated charcoal and removal of the the samples were dissolved in 0.2 M citric acid, pH 2, and amino acid composition determined on a Beckman 121 MB amino acid analyzer.

Rat mammary epithelium and adenocarcinomas derived from it synthesize type IV collagen, a structural protein of basement membranes. In cultures of cells, net production of collagen is stimulated 2-fold more than total cell protein by epidermal growth factor. Mammary adenocarcinoma cells also respond to epidermal growth factor but to a much reduced extent. This difference in growth factor responsiveness appears to be due to the production of collagen synthesis stimulating factors by the mammary tumor cells. Such factors have been partially purified and shown to differentially stimulate the incorporation of proline into collagenase-sensitive protein by 2.5-10-fold in normal rat mammary epithelium, normal rat kidney, and mouse 3T3 cells. The tumor factors do not stimulate net collagen production in cultures of tumor cells from which the factors are derived, suggesting that tumor cells produce sufficient stimulatory factors for optimal synthesis of collagen. Pulse-chase studies indicate that the tumor factors stimulate collagen synthesis rather than block collagen turnover. The activities in the extract have been partially purified by gel filtration, ion exchange column chromatography, and isoelectric focusing. The major species has a molecular weight of about 68,000 and a PI of 5.9. A smaller peak of activity with a molecular weight of 6,000 is also present. Since collagen synthesis appears to be necessary for the growth of mammary adenocarcinomas in vivo, production of these collagen synthesis stimulating factors may be important for tumor growth.
The epithelium of the normal rat mammary gland synthesizes type IV collagen, a constituent of the basement membrane upon which these cells rest in vivo (1). Formation of this protein is also characteristic of differentiated mammary tumors such as those induced by DMBA' or NMU (2, 3).
Synthesis of type IV collagen by these tumors is of considerable interest because their growth can be arrested by several * 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.
$ To whom all correspondence and requests for reprints should be addressed.
The abbreviations used are: DMBA, 7,12-dimethylbenz(a)anthracene; NRK, normal rat kidney; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; EGF, epidermal growth factor; NMU, N-methylnitrosourea: T-DMBA, transplantable 7,12dimethylbenz(a)anthracene-induced rat mammary tumor: T-NMU, transplantable N-methylnitrosourea-induced rat mammary tumor; CHP, cis-hydroxyproline; BSA, bovine serum albumin, MTWS, a highly differentiated rat mammary tumor; MTWSa, a less-well differentiated tumor derived from the MTWS tumor. proline analogs which are selective inhibitors of collagen production. One such inhibitor is CHP. It blocks collagen synthesis 10 times more efficiently than total protein synthesis in cultured mammary epithelium ( 2 ) and blocks the growth of both DMBAand NMU-induced mammary tumors in vivo (2, 3). CHP effects on tumor growth are related to the ability of mammary tumors to synthesize type IV collagen since certain transplantable tumors no longer make this protein and their growth in vivo is unaffected by CHP (3, 4).
The growth of the normal mammary rat epithelium also appears to require the production of type IV collagen. CHP blocks the growth of the epithelial cells both in vivo and in vitro, but the inhibitory effect of CHP can be reduced by plating the normal mammary epithelium on type IV collagen coated culture dishes (5, 6), suggesting that the type IV collagen-cellular interaction is important for cell growth and/ or survival.
These observations have prompted us to study the control mechanisms for collagen synthesis and deposition in both normal and neoplastic mammary epithelium. We have shown that production of type IV collagen is controlled both at the level of alterations in biosynthetic and in degradative rates (7). In the present report, we show that normal and tumor cells differ in that collagen production is stimulated by EGF in cultures of normal cells but not tumor cells. In attempting to explain this difference, we found that the tumor cells contained factors that stimulate net collagen production. This suggests that these factors might be important for tumor cell growth. We also report here on the partial purification and characterization of the tumor factors.

EXPERIMENTAL PROCEDURES
Cell Cultures-The epithelium from normal rat mammary glands or from DMBA-induced rat mammary tumors was isolated according to Wicha et al. (8) and cultured in a serum-free growth medium as described by Salomon et al. (7). NRK (clone 49-F) cells were obtained from Dr. Joseph DeLarco, National Cancer Institute, and Balb/c 3T3 fibroblasts were supplied by Dr. Janice Chu, National Institute of Child Health and Human Development. Both cell types were maintained in Dulbecco's modified Eagle's medium containing 10% fetal calf serum.
Tumor Extraction-Aciditied ethanol extracts of tumor tissue were prepared according to Roberts et al. (9). Briefly, the material solubilized in acidified ethanol was adjusted to pH 5.3 and precipitated by the addition of 2 volumes of ethanol and 4 volumes of ether. The precipitate was collected, dialyzed against 0.17 M acetic acid, and lyophilized to dryness. This residue was dissolved in phosphatebuffered saline (pH 7.4) giving a final concentration of 0.5-1 mg/ml after removal of insoluble residue by low speed centrifugation. The soluble tumor extract was sterilized by filtration through 0.2 p Millipore filters.
Bio-Gel PlOO Column Chromatography-Approximately 20 mg of soluble protein from the extract was applied to a column (1.5 X 90 cm) of Bio-Gel PlOO equilibrated with phosphate-buffered saline. The column was eluted with the same buffer at a flow rate of 12 ml/h. Column fractions were collected into tubes to which 0.1 ml of 1% BSA had been added. This step was necessary to prevent denaturation or absorption of the active fractions since little or no activity was detected otherwise. Column fractions were sterilized by fitration and aliquots added directly to cell cultures for assessment of their effects on cell growth and collagen production.
CM-cellulose Column Chromatography-55 mg of crude tumor extract was suspended in 10 ml of 0.01 M ammonium formate, pH 5.2, and the suspension stirred for 30 min. The solubilized material was applied to a column (1.5 X 25 cm) of CM-cellulose-52 resin. The column was washed with 16 ml of the same buffer and then elution was performed with a linear gradient of ammonium formate (100 ml of 0.01 M ammonium formate and 100 ml of 1.0 M ammonium formate, pH 5.2). Approximately 5-ml fractions were collected into tubes containing 1% BSA. The fractions were tested for biological activity after lyophilization.
Isoelectric Focusing-200 mg of tumor extract was suspended in 10 ml of focusing buffer (10 m Buffalyte, Pierce Chemical Co., pH 3-9) and the soluble fraction recovered after centrifugation. This fraction was adjusted to a total volume of 48 ml with the same buffer, then mixed with 111 g of Pevicon C-870 and electrofocused for 4 h at IO "C, 1230 V, 10 watts, and 50 mA in an LKB model 2117 Multiphor apparatus. Afterwards, the Pevicon gel was sectioned into 30 consecutive fractions. To each fraction, 2 ml of distilled water was added and pH measurements made. Then 0.2 ml of 2 M ammonium formate, pH 5.2, was added to elute protein from the Pevicon matrix. The solubilized fractions were dialyzed against I O liters of 0.2 M ammonium formate buffer and lyophilized to dryness. The residual material was resuspended in 2 ml of phosphate-buffered saline and aliquots tested for activity in cell cultures. In the fractions containing collagen synthesis stimulating activity, aliquots were electrophoresed on 7.5% SDS-polyacrylamide gels and stained according to Laemmli (10) to determine the nature of the proteins in the active fractions. Effects on Collagen Production and Cell Growth-Either soluble tumor extracts or column fractions prepared from them as described were added to the various cell cultures. The effects on cell growth were determined three days after factor addition during which the cells were in log phase. Control cultures received an equivalent amount of phosphate-buffered saline. Effects on collagen production were assessed either by determining the amount of labeled amino acid ([3H]lysine or ['4C]proline, 2 pCi/ml) incorporation into collagenase sensitive protein according to methods described by Peterkofsky and Diegelmann (11) or by determining the amount of hydroxyproline and hydroxylysine mass relative to cell number or to total lysine in the cell protein fraction. In the former method, a protease-free collagenase (Form 111, Advanced Biofactures, Inc., Lynnbrook, NY) was utilized as described previously (3). In the latter method, the cell layers were recovered from the culture dishes and these were solubilized in 0.1 M NaOH. After digestion for 30 min at 37 "C, the sample was precipitated with trichloroacetic acid, 1% tannic acid (11). The precipitate was washed two times with the same solution and then ether extracted to remove the trichloroacetic acid. The sample was hydrolyzed in 6 N HCl for 24 h at 110 "C in sealed tubes in a nitrogen atmosphere. After decoloring with activated charcoal and removal of the solvent, the samples were dissolved in 0.2 M citric acid, pH 2, and amino acid composition determined on a Beckman 121 MB amino acid analyzer.

Differential Effects of EGF on Normal and Tumor Cells of the Mammary Gland
The net incorporation of proline into cell-associated collagen in cultured normal mammary epithelium has previously been shown to be differentially stimulated by EGF (12). Mammary tumor cells such as those derived from primary mammary tumors induced by NMU or DMBA show a reduced response to EGF as is demonstrated in Table I. EGF produces about a doubling in the net percentage of incorporation of precursor amino acids into collagenase sensitive protein in the cell layer in cultures of normal cells while it produces only an 8% differential labeling of collagen in the tumor cell layer. This difference in the EGF responsiveness of the tumor and normal cells is also apparent when the effects of the growth factor are assessed on the basis of the mass of cell-associated hydroxyproline in cultures propagated with and without EGF the cell layers were analyzed for the amount of collagen either by quantitating the amount of collagenase sensitive labeled cell-associated protein, or by measuring the mass of hydroxyproline present following acid hydrolysis of the cell layer. The values depict the amount of collagen produced normalized against total labeled cell protein or normalized against the total cellular lysine, respectively. These normalized values were then compared to normalized values for control cultures with EGF omitted. A value of 0% means that collagen production is not stimulated more than total protein synthesis is stimulated. The tumor cells were prepared from primary NMUinduced tumors.  (Table I) as shown by amino acid analysis. In our attempts to explain this difference in EGF responsiveness, we found that there were acid-ethanol extractable factors in tumors which differentidy promoted the amount of labeled collagen in the cell layer of cultures of normal mammary epithelium. This factor(s) had relatively little effect on the relative amount of labeled collagen found in association with tumor cells, however. For example, cell-associated collagen was labeled 300% more efficiently than total cell protein in cultures of normal cells containing the tumor factors than in cultures without tumor factors (Table 11). With cultured tumor cells, however, there is only a 25% differential increase in cell-associated collagen label produced by adding the same amount of the tumor extract (Table 11); z.e. the normal cells were 10 times as responsive to the tumor extracts than were the tumor cells.
At the concentration of tumor extract utilized (-100 pg of protein/ml), there was only a 40 and 70% stimulation of total protein synthesis for normal and tumor cells, respectively. As will be subsequently demonstrated, the differentially enhanced labeling of cell-associated collagen was not effected by altering collagen turnover rates or by altering the relative distribution of the newly synthesized collagen between the cell Iayer and culture medium.

Effect of the Tumor Extract on Collagen Turnover
In order to characterize the activities in the tumor extract, an abundant supply of extract-responsive cells was desirable. Consequently, several cell types were screened as potential candidates. As shown either by the increased production of labeled collagenase sensitive protein formed or by the mass of hydroxyproline formed, normal rat kidney cells were quite responsive. That is, the relative amount of cell-associated collagen was increased by the extract (Table 111). 3T3 cells were also responsive but an EGF receptorless cell line (NR6) (13), derived from 3T3, was not. Because the NRK cells were from the same species as the tumors (rat), we opted to use these cells for determining the biological properties of the tumor extract and for routine screening during purification of the activities in the extract.
Previously, type IV collagen accumulation was shown to be favored in mammary cell cultures supported with glucocorticoid hormones. The hormones acted by blocking the turnover of newly synthesized collagen (7). This was deduced by pulse labeling the collagen in the presence of the hormone followed Effect of primary NMU tumor extract on collagen production in different cell types Assays were performed as described in the legend of Table 11. In the case of the NRK, 3T3, and NR6 cells, growth medium contained 10% fetal calf serum, whereas the mammary epithelium was cultured in serum-free medium. The effects of the tumor extract on hydroxyproline production in the NRK cells was also determined. After three days of incubation, the control cultures had only ?A as much 4hydroxyproline/unit of cell protein as did the cultures grown in the presence of the tumor extract.  (14), a 15-min pulse with ['4C]proline followed by a 6 h chase without label was also performed to assess effects of tumor extract on collagen turnover. As shown in Table IV about % of the cell-associated collagen labeled in 15 min was still present in the cell layer a t the end of the chase period whether the cultures were grown in the presence of tumor extract, EGF, or control buffer. Thus, we can conclude that for NRK cell cultures the effects of the tumor extract or EGF in differentially enhancing the accumulation of newly synthesized collagen in the cell layer is not via inhibiting the turnover of cell layer collagen. Likewise, neither factor affected the relative amount of collagen secreted into the growth medium. Analyses of acid precipitable, collagenase sensitive labeled protein in the cell layer and growth medium after the 6-h chase showed that 90, 83, and 87% of the acid precipitable labeled collagen was cell-associated in the control, in tumor extract, or EGF supplemented cell cultures, respectively. An analysis was also made of the total associated hydroxyproline in the control cultures. For the cultures with tumor extract added, the values were 20,100 and 28,930 cpm, respectively. Thus, 53% of the total labeled hydroxyproline was in the cell layer for the controls and 59% for the tumor extract-treated cultures. The above results strongly indicate that the activities in the primary mammary tumor extract differentially promote collagen synthesis.

Sources of the Collagen Synthesis Stimulating Activity
Before attempting to purify and characterize the collagen synthesis stimulating activity, a number of tissues were screened as potential sources of starting material. Using identical fractionation procedures, extracts were prepared from proliferating rat mammary tissue, adult rat liver, primary rat mammary tumors induced by NMU or DMBA, and several transplantable tumors including MTW9, MTWSA, T-NMU, and T-DMBA. None of the extracts of normal tissues were active in differentially stimulating collagen synthesis in cultures of NRK cells or in normal rat mammary epithelium cultures. Activity was highest in extracts of primary mammary tumors induced by NMU or DMBA. Activity was moderate in MTW9 and MTW9A tumors, low but detectable in T-NMU tumors, and absent in T-DMBA tumors. In general, the activity in the tumors correlated with the degree of differentiation of the tumor tissue as estimated from histological examination. Because the primary NMU tumor extracts possessed the most activity, extracts of this tissue were utilized for further characterization and purification of the collagen synthesis stimulating activity.

Characterization of the Active Fractions in the Tumor Extract
Gel Filtration-When the soluble fraction prepared from the tumor extract was chromatographed on Bio-Gel PlOO columns and the fractions screened for effects on collagen production by NRK cells, two major peaks of activity were seen. These had elution positions approximately equal to bovine serum albumin and porcine insulin markers and, therefore, had apparent molecular weights of 68,000 (the major species) and 6,000 (the minor species), respectively, under neutral conditions. The fractions were tested in cultures grown in medium containing 5 or 10% fetal calf serum. Total protein synthesis stimulating activities were detected in the column fractions, but the peaks of activity were not coincident with the collagen synthesis stimulating activity peaks. In the most active fraction, collagen synthesis was stimulated 7.5-fold,  1 day in medium with the buffer, tumor extract (100 bg of protein/ml), or EGF (10 ng/ml) in medium containing 10% serum. [3H]lysine (50 pCi/ml) was added for 15 min, then quickly aspirated, and the cell layers washed 3 times with medium without [3H]lysine. Dishes were returned to the incubator for 6 h or immediately frozen on dry ice. The cpm collagenase sensitive protein in the cell layer was determined for the pulse or pulse-chase samples. In this experiment, labeled collagen production was differentially stimulated 4-fold by the tumor extract and 2-fold by EGF when normalized against total cell protein counts. while total protein synthesis was increased by 3.5-fold (Fig.  2).
The nonidentity of the collagen synthesis stimulating activities and total protein synthesis stimulating activities deduced from analysis of the high molecular weight fractions from the Bio-Gel PlOO column was also evident since the two activities were differentially heat sensitive. Thus, a 2-min heating of the tumor extract at 90 "C reduced the collagen synthesis stimulating activity by 95%, while only reducing the total cell protein synthesis stimulating activity by 53% (Table V).
CM-cellulose Column Chromatography-Ion exchange chromatography also resulted in the separation of the collagen synthesis stimulating activity into two major peaks, the larger eluting from the column at 0.28 M ammonium formate and the less abundant species at 0.56 M salt. The major collagen synthesis stimulating activity was partially resolved from the total protein synthesis stimulating activity but the two activities fractionated coincidently in the case of the lesser abundant, higher salt eluting species (Fig. 3). As was the case with the Bio-Gel PlOO column fractionation, both the collagen synthesis and total protein synthesis stimulating activities were only partially separated from the bulk protein of the tumor extract, especially the higher molecular weight species. At the peak of the collagen synthesis stimulating activities there was a 3.1-and 2.0-fold differential stimulation of collagen labeling for the major and minor activity peaks, respectively.
Isoelectric Focusing-Upon isoelectric focusing of the crude tumor extract from the primary NMU tumor, one major peak of collagen synthesis stimulating activity was detected Fig. 4, fraction 15). The PI of this fraction was 5.9. Total protein synthesis stimulating activity was also present as a broad peak focusing between a PI of 5.6 and 6.3. Several minor collagen synthesis stimulating peaks were also present but most of the activity was found in fraction 15. Gel electrophoresis was performed on fractions 13-1 7 for further characterization of the proteins present in the active fractions. The profiles are depicted in Fig. 5. The major stainable protein bands were broad and diffuse, with apparent molecular weights ranging from 25,000 to 45,000. Fraction 15, having the most prominent collagen synthesis stimulating activity, contained several minor bands not present in the other fractions. One band had a molecular weight of 68,000, which is the molecular weight of the major collagen synthesis stimulating activity as detected by Bio-Gel PlOO column chromatography. Other protein bands were also present only in this fraction. These are estimated to have molecular weights of 16,000, 11,000, and smaller. Whether more than one band possesses collagen synthesis stimulating activity is not certain, but the   68,000 molecular weight species is active. The gel slices corresponding to the 68,000 molecular weight species were extracted and the extracts tested for activity with NRK cells (Table VI). Compared to control gels with no extract applied, the 68,000 molecular weight species stimulated collagen synthesis 2.6-fold, while stimulating total protein synthesis by In addition to the direct demonstration that the 68,000 molecular weight protein recovered from the SDS-gels was biologically active (Table VI), further evidence that this protein was the entity responsible for stimulating collagen synthesis was obtained as follows. Equal amounts of extracts (100 mg of protein) of tumors possessing collagen synthesis stimulating activity or tumors not having this activity were isoelectrically focused and the PI-5.9 fraction recovered and subsequently electrophoresed on SDS-gels as described in Fig. 5. As shown in Fig. 6, a stainable protein band coincident with the bovine serum albumin marker was present in the extracts of primary NMU tumors (lane 2), primary DMBA tumors

(lane 3), and MTW9 tumors (lane 4). No stainable bands
were seen in this region of the gel in the case of the transplantable NMU tumor (lane 5) or the transplantable DMBA tumor (lane 6). Thus, the relative abundance of the 68,000 molecular weight protein with PI of 5.9 correlates roughly with the amount of collagen synthesis stimulating activity detected in the extracts of the various tumor types.
A rough estimate is that the 68,000 molecular weight collagen synthesis stimulating species represents about 0.0001% of the protein in the tumor. From the relative staining intensity of the 68,000 molecular weight band in fraction 15 from the isoelectric focusing column compared to the albumin standard staining intensity, we can estimate that there was approximately 100 ng of this material added in the assay for effects of fraction 15 on collagen synthesis in the NRK cell culture. This produced a 300% differential stimulation of collagen synthesis over and above the stimulation of total protein synthesis (Fig. 4). A maximal stimulation of collagen synthesis of about 100% is observed with 10 ng of EGF, which was the optimal amount of this factor for the NRK cells in so f a r as stimulation of collagen synthesis is concerned (not only 1.l-fold.  (Fig. 4) were run on Laemmli gels (7.58) after reduction with mercaptoethanol. Molecular mass: myosin, 2OOK, &galactosidase, 116K, phosphorylase 6, 92K, bovine serum albumin, 68% and ovalbumin, 45K. Note that a band of protein with the same molecular weight as the albumin marker is present in fraction 15 only. K, kilodaltons.

Effect of the 68,000 molecular weight protein on collagen synthesis
Bovine serum albumin was dansylated and run on SDS-slab gels in slots adjacent to the primary NMU tumor extracts (50 pg of tumor protein/slot). Control gels to which no tumor extract was applied were also developed. After electrophoresis, the fluorescent bovine serum albumin was localized with a UV light and corresponding gel areas recovered, minced in 0.5 ml of phoshate-buffered saline containing 50 pg of bovine serum albumin carrier protein. After solubilizing overnight at 4 " C , the sample was centrifuged a t 2000 rpm for 15 min and the supernatant fluid recovered, sterilized by filtration, and added to cultures of NRK cells. ['4C]Proline was added and the cultures were incubated for two days and then analyzed for the amount of labeled collagen and total protein label present in the cell layer using the collagenase assay. Three control slots (control gel slice extract) and 3 tumor gel slots were Dooled for the assavs.

Addition
Total protein Collagen shown). Thus, the 68,000 molecular weight tumor collagen synthesis stimulating factor is a more effective stimulator of collagen synthesis than an equal molar concentration of EGF. It is clear that the tumor derived activity is not rat EGF 68.000 molecular weight protein with a PI of 5.9. One hundred milligrams of extracts of various tumor types were isoelectrically focused as described in Fig. 4. The material with a PI of 5.9 was recovered and analyzed for the size distribution of proteins in this fraction on SDS 7.5% slab gels. Twenty-five micrograms of protein from the PI 5.9 fraction of each tumor type was added to the gels which, following electrophoresis, were stained with silver using a staining regimen and reagents supplied by Bio-Rad. because the latter species has a PI of 4.6 (15), because EGF does not bind to CM-cellulose at pH 5.2 (15), and because the molecular weight of EGF is %o that of the tumor factor. In addition to the collagen synthesis stimulating activities, the tumors also contain activities that stimulate the growth of NRK cells in soft agar. In preliminary gel filtration studies, these activities, the soft agar growth promoting and collagen synthesis stimulating activities, co-fractionated but they were not coincident in the isoelectrically focused fractions, the collagen synthesis stimulating activity being localized in fraction 15 while the soft agar growth promoting activities were present in fractions 13, 14, and 17 but absent from fraction 15. The characteristics of the soft agar growth promoting activity will be described more fully in a separate publication.

DISCUSSION
The cellular controls for modulating the amount and types of collagen synthesis are poorly understood. With normal mammary epithelium, collagen production is alterable depending on the substratum upon which the cells are plated. For example, these cells produce about twice as much type IV collagen when plated on tissue culture plastic surfaces or on type I collagen-coated dishes as the cells produce on type IV collagen-coated dishes (16). Collagen production by the normal rat mammary cells is also influenced by modulating turnover and, apparently, by enhancing the biosynthetic rate. Glucocorticoids have been shown to inhibit the production of type IV collagenase while EGF promotes collagen production in these cells by another mechanism (7). Substratum effects on collagen production by other cells has also been observed. Thus, Enders et al. (17) have found that limiting the degree of flattening of cells by varying the degree of coating of culture dishes with polyhemin selectively affects collagen production. Another example of a surface effect is in the change from the production of interstitial type to a basement membrane type of collagen by kidney mesenchyme as it interacts with embryonic notochord (18) Thus, cells can respond to physical, hormonal, or growth factors and selectively modify the amount and/or types of collagen they produce. Also, a change in collagen species or amount may or may not accompany transformation (19-21), depending on the cell type.
There is evidence suggesting that collagen production can be regulated at the level of translation of the collagen mRNA as well as by altering the amount of mRNA produced. In in uitro translation systems, for example, the NHn-terminal peptides from procollagen selectively inhibit collagen mRNA translation (22-25). Quantitation of collagen mRNA levels during development has revealed that the level of collagen production may be proportional to the level of the collagen messenger RNA but in some cases it is not (26).
Regulation of collagen production may be effected by modulation of the degradative rate of collagen within the cell. Berg et al. (27) have proposed that such a mechanism of control could be provided by changes in the ability of newly formed collagen to form a mature triple helix depending on the degree of hydroxylation of proline. These authors demonstrated that about ' h of the collagen labeled in a short pulse was degraded, a value identical to our results with the NRK cells. In our present studies, the amount of turnover of collagen was the same for control cultures or cultures with the tumor factors (or EGF) present. Thus, we can conclude that the effects of the tumor collagen synthesis stimulating activity (or EGF) in enhancing net collagen production is not via inhibiting intracellular turnover. Based on the lack of turnover of labeled collagen formed during long labeling times during a subsequent chase, it is also unlikely that extracellular turnover is affected by the tumor factors or EGF. The probability is, then, that these two agents act by similar mechanisms, i.e. by enhancing collagen mRNA production or translation, possibilities that are currently being investigated.
Because of the complexities in the regulation of collagen production and the possibility that such changes might not be of physiological significance, some caution in interpreting the observed effects of factors on collagen production in culture is necessary. In a search for such factors that might be physiologically important, some criteria for their identification can be formulated. One would anticipate that a cell that produced such a factor would not be responsive to the exogenous addition of that factor. Additionally, a producer cell might be expected to be unresponsive to other growth factors that stimulated collagen production by a similar mechanism. One would also anticipate that such a factor would be active at low concentrations and it would probably be produced in small amounts.
Based on our present knowledge of the collagen synthesis stimulating activities we have detected in some mammary tumors, it would appear that most of these criteria are met and that such factors are present in these tumors. Thus, collagen synthesis in cultures of primary NMU tumor cells it only marginally affected by exogenously added collagen synthesis stimulating activity derived from the primary NMU tumor. The same cells are also unresponsive to EGF, a growth factor that, like the collagen synthesis stimulating activity derived from primary NMU tumors, modulates collagen pro-duction independently of affecting intracellular or extracellular turnover of collagen. That is, the primary tumor-derived factor is EGF-like since neither the tumor-derived factor nor EGF stimulates collagen synthesis in cultures of primary NMU tumor cells. As demonstrated, the 68,000 molecular weight factor is also present in very low concentrations. Based on the intensity of the staining of the 68,000 molecular weight species with a PI of 5.9, we can roughly estimate that this species represents about 0.0001% of the protein in the tumor. This is a rough estimate because recoveries may be poor. The 6000 molecular weight species was not recovered at all following isoelectric focusing, for example. Using the same rough estimate of the abundance of the 68,000 molecular weight protein, we can calculate that it is more active than EGF on NRK and normal mammary cells.
What would be the significance of the production by the mammary adenocarcinomas of a collagen synthesis stimulating activity? In these tumors, as in the normal mammary epithelium, collagen synthesis appears to be restricted to a minority population, the basal or myoepithelial cell (4). One might speculate that such an activity is produced by one cell type in the gland (possibly the epithelial cell) and it acts on the basal calls to enhance collagen production. This condition might favor tissue growth since it has been demonstrated that a variety of proline analogs such as cis-hydroxyproline, thioproline, and azetidine carboxylate selectively block collagen deposition and also block primary mammary adenocarcinoma tumor cell growth (2,3). Collagen synthesis stimulating activity might also be produced by normal mammary epithelium but its levels of production might be too low to be detected in the present assays. Further studies will be required to evaluate these possibilities.