Proc. Nati. Acad. Sci. USA Vol. 89, pp. 3120-3124, April 1992 Medical Sciences
A growth-regulated protease activity that is inhibited by the anticarcinogenic Bowman-Birk protease inhibitor PAUL C. BILLINGS
AND
JOAN M. HABRES
Department of Radiation Oncology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104
Communicated by Clarence A. Ryan, October 18, 1991
ABSTRACT The Bowman-Birk protease inhibitor (BBI) has been shown to be an effective suppressor of carcinogenesis in vivo and in vitro. To elucidate the mechanism(s) by which BBI suppresses carcinogenesis, we believe it will be necessary to identify and characterize the target enzymes that specifically interact with the BBI. We have shown previously that several cellular proteins in C3H/1OT'/2 mouse embryo fibroblast cells specifically bind to a BBI affinity resin. In the current report, we demonstrate that one of these proteins has proteolytic activity as judged by its ability to degrade gelatin. The enzyme has a mass of 45 kDa and subcellular fractionation experiments demonstrate that this enzyme is located in the cytosol. Furthermore, the proteolytic activity was inhibited by diisopropylfluorophosphate but was not affected by EDTA, indicating that this enzyme is a serine protease. Higher levels of protease activity were found in logarithmic-phase C3H/1OTI/2 cells compared with nondividing (confluent) cells, suggesting that this protease activity is growth regulated. Similar levels of this activity were present in nontransformed and in radiationtransformed C3H/1OT'/2 cells. Treatment of nontransformed C3H/1OT1/2 cells with phorbol 12-myristate 13-acetate increased the specific activity of this protease 5- to 10-fold. Our results suggest that this protease is a target enzyme of the BBI in these cells.
mation of BALB/3T3 cells induced by radiation or chemical carcinogens (benzo[a]pyrene and P-propiolactone) (6). We have shown that BBI as well as other anticarcinogenic protease inhibitors are internalized by C3H/10T1/2 cells (7, 28, 29). A major focus of our research is to elucidate the mechanism(s) by which protease inhibitors such as BBI suppress carcinogenesis. To accomplish this goal, we believe it will be necessary to identify and characterize the target enzymes that specifically interact with the BBI. In a previous study, we showed that a 45-kDa protein from mouse fibroblasts specifically binds to a BBI affinity column (23). In this report, we demonstrate that this protein is a serine protease with unusual characteristics.
MATERIALS AND METHODS Chemicals. The BBI was purified as described (17, 30). [35S]Methionine and Enlightning were obtained from New England Nuclear. Gelatin (300 Bloom) was obtained from Sigma. CNBr-activated Sepharose was obtained from Phar-
Epidemiological data suggest that nutritional factors play a major role in the etiology of cancer at many different sites (1-4). For example, high dietary levels of legumes have been associated with low cancer rates in general and are inversely correlated with the incidence of breast, colon, pancreatic, and prostate cancer (1-4). Legumes are known to contain high concentrations of protease inhibitors (5). While protease inhibitors have been shown to have strong anticarcinogenic activity in in vivo and in vitro cancer model systems (for examples, see refs. 6-21), relatively little is known about the precise mechanism(s) by which these compounds exert their suppressive effects. We have hypothesized that protease inhibitors block carcinogenesis by inhibiting cellular enzymes involved in induction and/or expression of the transformed phenotype (22-24). The anticarcinogenic activity of the soybean-derived Bowman-Birk protease inhibitor (BBI) has been extensively studied in our laboratory (6, 8, 13, 17, 19-21). The BBI is an 8-kDa protein that effectively inhibits trypsin and chymotrypsin (25) as well as several other proteases (22, 24-27). BBI has been shown to suppress dimethylbenz[a]anthraceneinduced cheek pouch carcinogenesis when topically applied (15), 3-methylcholanthrene-induced lung tumorigenesis when injected (20), and dimethylhydrazine-induced colon and liver carcinogenesis in mice when present in the diet (8, 17, 19). The BBI has also been shown to suppress radiation-induced transformation of C3H/1OT/2 mouse embryo fibroblast cells at nanomolar concentrations (21) and to inhibit the transfor-
macia. Cells. C3H/10T1/2 mouse embryo fibroblast cells and the radiation-transformed subclones F-17 and F-29 were grown as described (31, 32). Radlolabeling of Cellular Proteins. Cells were metabolically labeled with [3 S]methionine (50 ,uCi/ml; 1 Ci = 37 GBq) for 12 hr (23). After labeling, the cells were washed twice with ice-cold phosphate-buffered saline (PBS), removed from the dishes with a rubber policeman, and pelleted by centrifugation at 1000 x g for 5 min. The cell pellet was resuspended in 50 mM phosphate buffer, pH 7.0/1 mM MgCl2/0.1% Triton X-100, or 100 mM Tris-HCI, pH 7.0/0.1% Triton X-100, sonicated on ice for 30 sec and centrifuged at 10,000 x g for 5 min at 40C. Fractionation of Cellular Proteins over the BBI Affinity Column. The BBI affinity resin was prepared as described (23). Samples were applied to the affinity column in binding buffer (100 mM phosphate buffer, pH 7.0/1 mM MgCl2) at room temperature. The column was washed with 50 ml of washing buffer (50 mM phosphate buffer, pH 7.0/500 mM NaCl); bound proteins were eluted from the column with 5 M urea. Cellular proteins were analyzed on 12% polyacrylamide gels containing 0.1% SDS (33) and stained. For autoradiography, the gels were impregnated with Enlightning, dried in a gel dryer, and autoradiographed (23). Zymogram Protease Assays. Twelve percent polyacrylamide gels containing 0.1% gelatin were cast (34). Samples were applied to the gel in standard SDS gel loading buffer containing 0.1% SDS but lacking 2-mercaptoethanol and were not boiled before loading. The gels were run at a constant 70-80 V and were then soaked in 200 ml of 2% Triton X-100 in distilled water on a gyratory shaker for 1 hr
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviations: APMA, aminophenylmercuric acetate; BBI, Bowman-Birk inhibitor; DFP, diisopropylfluorophosphate; PMA, phorbol 12-myristate 13-acetate. 3120
Medical Sciences:
Billings and Habres
at 20'C. Next, the gels were soaked in 100 mM Tris HCl, pH 8.0/5 mM CaC12 for 12 hr at 370C and then stained. The zymograms presented are representative results obtained from two or more independent experiments. Subcellular Fractionation. Logarithmically growing C3H/ 10T1/2 cells were washed twice with ice-cold PBS, scraped from the dishes, and pelleted by centrifugation at 1000 X g for 10 min. The cells were resuspended in ice-cold isotonic sucrose buffer (10 mM Tris-HCI, pH 7.0/250 mM sucrose/1 mM MgCl2) and homogenized on ice in a Dounce homogenizer (100 strokes). The homogenate was centrifuged (all centrifugations were carried out at 40C) at 1000 x g for 10 min, 10,000 x g for 10 min, and 100,000 x g for 60 min (22). Each subcellular fraction was assayed for protease and marker enzyme activity. The lysosomal enzyme P-glucuronidase and the cytosolic enzyme lactate dehydrogenase were assayed as described (22). The protein present in each sample was determined by the method of Bradford (35). Trypsin Activation of Cell Homogenates. One hundred microliters of cell homogenate (50 ug of protein) was incubated with 1 Aul of trypsin (75 gg/ml) in PBS for 1 hr at 20TC.
RESULTS We have previously shown that two proteins of =45 and =60 kDa specifically bind to a BBI affinity column (23). We detected protease activity in the fractions binding to the column by using a nonspecific substrate (23). While these results indicated that functional enzyme activity was eluting from the BBI affinity column, they did not provide definitive information about which of these proteins had protease activity. To directly address this question, C3H/10T'/2 cell proteins were labeled with [35S]methionine and passed over the BBI affinity column; bound proteins were eluted from the column with 5 M urea (Fig. 1). In these studies, 0.05-0.1% of the total counts applied to the column eluted with the urea wash. A single band of protease activity eluted with the urea wash; this single band was readily observed on gelatincontaining zymograms (Fig. 1). Autoradiography of the zymogram revealed two labeled proteins (Fig. 1). Only one of these proteins had proteolytic activity, however. Excising the band containing protease activity from the zymogram and running it on a second-dimension reducing gel (the sample was incubated in 50 mM 2-mercaptoethalaol before being run in the second dimension) revealed a single protein of -45 kDa (Fig. 1). Earlier studies demonstrated that a 45-kDa protein can be eluted from the BBI affinity resin with HCl (23). We consistently observed low levels of protease activity in this size range when total cell homogenates or material passed over a Sephadex gel filtration column were run on the zymograms. Several types of proteases, such as collagenases, require activation by trypsin or organomercurial compounds [such as aminophenylmercuric acetate (APMA)] to become functionally active proteolytic enzymes (36). To determine whether the 45-kDa protease is present in C3H/ 10T1/2 cells in a cryptic or inactive form, cell homogenates were prepared and incubated with APMA or trypsin and subsequently analyzed on gelatin-containing zymograms. Greatly increased levels of protease activity were observed in trypsin-treated samples, while no increase in activity was observed in untreated or in APMA-treated samples (Fig. 2). The trypsin-activated protease comigrates with the protease eluting from the BBI affinity column. These results suggest that this protease is normally in an inactive form in these cells. The 45-kDa protease activity was present in growing cells, while significantly lower levels of this activity were present in nondividing, confluent cells; the activity present in proliferating cells was inhibited by the BBI (Fig. 3). Similar levels of protease activity were observed in confluent cells and in
Proc. Natl. Acad. Sci. USA 89 (1992)
3121
A 200 Vt)
100
0 x
C 1-1
6 c
0
4
C.)
2
5
0
10
15
20
25
Fraction No.
1 2
B
C
wsx
le
-a
67 45
31
21
FIG. 1. Elution of protease activity from the BBI affinity column. (A) Elution profile of 35S-labeled C3H/1OTY2 cell proteins from the BBI affinity resin. Bound proteins were eluted with 5 M urea (arrow). (B) Analysis of material eluting with the urea wash. Lanes: 1, material was analyzed on a gelatin-containing zymogram (note the band of protease activity); 2, autoradiograph of gel shown in lane 1. (C) Band containing protease activity was excised from the gel and run on a second-dimension reducing gel. Numbers are molecular mass markers (kDa).
cells grown in serum-free medium (Figs. 2 and 3). Therefore, expression of this proteolytic activity is correlated with the state of cellular growth; relatively high levels of activity are observed in actively growing cells, while low levels of activity are observed in stationary cells. We have observed this proteolytic activity in two radiation-transformed C3H/ 10T'/2 cell lines, F-17 (Fig. 3) and F-29; slightly elevated levels of this activity were seen in F-29 cells (data not shown). We have observed a similar activity in AG1522 human fibroblasts, in which BBI has been shown to suppress transformation (10) and in human bladder epithelial cells (data not shown). To determine the intracellular location of this proteolytic activity, subcellular fractionation experiments were performed. The majority of protease activity was present in the 100,000 x g supernatant fraction, indicating that this protease is located in the cytosolic fraction (Fig. 4, Table 1). We also determined the ability of this protease to cleave other substrates. While the 45-kDa protease efficiently cleaved gelatin, no protease activity was observed when casein was incorporated into the zymograms. We characterized the 45-kDa activity by using a series of protease inhibitors with defined inhibitory specificity. For these experiments, two procedures were used: (i) the sam-
Proc. Natl. Acad. Sci. USA 89 (1992)
Medical Sciences: Billings and Habres
3122
8
4
58
4 8.5
P.6
FIG. 4. Analysis of subcellular fractions for protease activity.
C3H/10T'/2 cells were disrupted in a Dounce homogenizer and FIG. 2. Analysis of cell extracts and conditioned medium for proteolytic activity. Samples were unactivated or activated with trypsin and then run on gelatin zymograms. Lanes: 1, unactivated cell extract; 2, cell extract activated with trypsin before being run on the zymogram (note band of protease activity); 3 and 4, untreated (lane 3) and trypsin-activated cell extracts (lane 4) were run on the zymogram (the zymogram was soaked in incubation buffer containing BBI); 5-8, conditioned medium from untreated or from phorbol 12-myristate 13-acetate (PMA)-treated C3H/10T1/2 cells. Cells were grown in serum-free medium in the absence or presence of PMA (100 ng/ml) for 48 hr. Two-milliliter aliquots of conditioned medium were concentrated by ultrafiltration and analyzed on zymograms. Lane 5, conditioned medium from untreated cells; lane 6, conditioned medium from cells treated with PMA; lane 7, trypsin-activated conditioned medium from control cells; lane 8, trypsin-activated conditioned medium from cells treated with PMA. The 45-kDa protease is not secreted by these cells. Numbers on left are molecular mass markers (kDa).
ples were preincubated with the inhibitor and then run on the zymograms, or (ii) the samples were run on gelatin zymograms and the gels were subsequently incubated in 100 mM Tris'HCI, pH 8.0/10 mM CaCl2 in the absence or presence of .
4I
Awmw ts sf
:..,.
A
2
4
b
;
4 8 .5 B
2 6. 6
-
FIG. 3. (A) Protease activity in growing and in growth-arrested C3H/10T1/2 cells. Cell extracts from the indicated cell population were prepared and activated with trypsin. Samples were analyzed on zymograms for protease activity. Lanes: 1, confluent cells; 2, logarithmic-phase cells; 3, F-17 [radiation transformed C3H/10T'/2 cells (32)]. (B) Effect of serum, DFP, and PMA on 45-kDa proteolytic activity. Cells were grown in the absence or presence of PMA (100 ng/ml) for 48 hr in complete (containing 10%o fetal calf serum) or in serum-free medium. Lanes 1-4, cells grown in complete medium; lanes 5-8, cells grown in serum-free medium for 72 hr. Lanes: 1, untreated cell extract (not activated with trypsin); 2, cell extract from cells treated with PMA; 3, trypsin-activated control cell extract; 4, trypsin-activated cell extract from PMA-treated cultures; 5, trypsinactivated extract from untreated cells; 6, trypsin-activated extract from PMA-treated cells; 7 and 8, same as 5 and 6 except that the zymogram was incubated in reaction buffer containing 10 mM DFP. Note that DFP completely inhibits 45-kDa proteolytic activity. Numbers on left are molecular mass markers (kDa).
subcellular fractions were isolated by differential centrifugation and analyzed for protease activity on a gelatin zymogram. Lanes: 1, 3, 5, and 7, untreated fractions; 2, 4, 6, and 8, trypsin-activated fractions; 1 and 2, 1000 x g pellet; 3 and 4, 10,000 x g pellet; 5 and 6, 100,000 x g pellet; 7 and 8, 100,000 x g supernatant. The bulk of the 45-kDa proteolytic activity is present in the 100,000 x g supernatant (arrow). Numbers on left are molecular mass markers (kDa).
the indicated protease inhibitor for 12 hr at 370C. As expected, protease activity was inhibited by the BBI (Fig. 4). It was also inhibited by antipain but was unaffected by EDTA, 1,10-phenanthroline, or chymostatin (Table 2). When trypsin-activated material was preincubated with diisopropylfluorophosphate (DFP) and then run on the zymogram, no inhibition of this activity was observed. However, when trypsin-activated material was run on the zymogram and then incubated in reaction buffer containing DFP, inhibition ofthis protease activity was observed (Table 1). Maximal protease activity was observed at pH 8. These results suggest that this activity is a neutral serine protease (37, 38). To determine whether the 45-kDa proteolytic activity was secreted, conditioned medium from C3H/1OT'/2 cells was analyzed for protease activity. This proteolytic activity was not secreted from these cells (Fig. 2). We also determined the effect of PMA on the levels of this enzyme. When cells were grown in complete medium containing serum, PMA increased the levels of this activity 3- to 6-fold (Fig. 3). However, when cells were grown in serum-free medium, PMA increased the levels of this activity 5- to 10-fold. Furthermore, PMA did not induce the secretion of the 45-kDa protease into the culture medium (Fig. 2). The proteolytic activity present in PMA-treated cells was inhibited by the BBI as well as DFP (Fig. 3, Table 2). Table 1. Marker enzyme and proteolytic activity in subcellular fractions Marker enzyme activity Protease LDH activity jS-Glu Subcellular fraction ND x 2.3 0.9 (1.3) (0.5) Pellet (1000 g) ND 0.9 (0.1) 3.8 (0.3) Pellet (10,000 x g) ND 3.4 (0.5) 2.7 (0.4) Pellet (100,000 x g) + Supernatant (100,000 x g) 1.2 (0.3) 15.6 (3.7) C3H/10T1/2 cells were homogenized on ice in a Dounce homogenizer. The homogenate was centrifuged at 1000, 10,000, and 100,000 x g. Each subcellular fraction was assayed for marker enzyme and protease activity. f3-Glucuronidase (fGlu) activity is expressed as A540 per 12 hr per mg of protein. Numbers in parentheses indicate total enzyme activity present expressed as specific activity times total protein present in the sample (AS0 per 12 hr). Lactate dehydrogenase (LDH) activity is expressed as A30 per min per mg of protein. Numbers in parentheses indicate total enzyme activity present expressed as specific activity times total protein present in the sample (Amo per min). For determination of 45-kDa protease activity, each subcellular fraction was assayed (see Fig. 4). ND, protease activity not detected; +, protease activity present.
Medical Sciences:
Billings and Habres
Table 2. Effect of different compounds on 45-kDa proteolytic activity in C3H/10T1/2 cells Type of Effect on protease proteolytic affected activity Compound Thiol Inhibit DTT (5 mM) Thiol None IA (5 mM) Serine Inhibit DFP (10 mM)* Metallo None EDTA (20 mM) Metallo None 1,10-Phenanthroline (0.5 mM) Serine/thiol Inhibit Antipain (10 ,.g/ml) Serine Inhibit BBI (10 /Ag/ml) Serine None Chymostatin (10 t±g/ml) Serine Inhibit SBTI (10 ,ug/ml) Cell homogenates were activated with trypsin and then incubated in the presence of dithiothreitol (DTT) or iodoacetamide (IA) for 10 min at 20'C and then run on gelatin-containing gels. For the other inhibitors, the cell homogenates were activated with trypsin and then run on gelatin-containing gels. The gels were incubated in reaction buffer containing the indicated inhibitor. SBTI, soybean trypsin inhibitor. *Proteolytic activity was inhibited when the zymogram was incubated in reaction buffer containing DFP.
DISCUSSION The BBI has been shown to be an effective suppressor of radiation and chemical carcinogen-induced transformation of C3H/1OT'/2 cells. We have hypothesized that protease inhibitors, such as BBI, block carcinogenesis by inhibiting cellular enzymes involved in the induction and/or expression of the transformed phenotype (22-24). In this report, we have identified an intracellular proteolytic activity in C3H/10T1/2 fibroblast cells that is inhibited by the anticarcinogenic BBI. To our knowledge, an intracellular protease that is growth regulated and induced by PMA has not been previously reported. These characteristics suggest that it may play an important role in carcinogenesis. The proteolytic activity described in the current study was present in nontransformed C3H/1OT'/2 cells and in two radiation-transformed subclones, F-17 and F-29 cells. We have observed this activity in other cell types as well, including human fibroblasts and human bladder epithelial cells. Greatly reduced levels of this protease were observed in confluent cells, and in nongrowing cells maintained in serum-free medium, compared to logarithmically growing cells (Fig. 3). These results suggest that this enzyme is growth related. Further evidence that the activity is growth related is its response to a mitogen; PMA treatment increased the levels of this enzyme 5- to 10-fold. There are previous studies indicating that PMA is capable of inducing proteolytic activity-specifically, plasminogen activator (39, 40). Plasminogen activator, however, is a secreted protease, while the protease activity identified and characterized in this report is clearly an intracellular (i.e., cytosolic) activity that is not secreted into the growth medium (Fig. 2). It is likely that the 45-kDa proteolytic activity described here was not observed previously by other investigators as it is displayed only under somewhat unusual conditions. While we were able to obtain active enzyme from the BBI affinity resin, no proteolytic activity was observed when cell homogenates were run directly on the zymograms. Furthermore, size fractionation of cell homogenates over a Sephadex gel filtration column or treatment with APMA also failed to activate this activity. However, treatment of cell homogenates with small amounts of trypsin did yield a preparation with a 45-kDa proteolytic activity. The protease activity purified by affinity chromatography appears to be similar to the trypsin-activatable activity present in cell extracts for the following reasons: (i) the protease eluting from the BBI
Proc. Natl. Acad. Sci. USA 89 (1992)
3123
affinity column comigrates with the trypsin-activatable protease in the zymograms, (ii) both proteases cleave gelatin, (iii) the trypsin-activatable protease is inhibited by the BBI, as is the protease eluting from the BBI affinity column. One explanation for the lack of 45-kDa proteolytic activity in untreated cell homogenates may be that the enzyme is tightly complexed with an endogenous inhibitor that greatly reduces its activity. Our results support the existence of such an inhibitor. Proteolytic activity was observed when the cell homogenates were run over the BBI affinity column; under these conditions, the immobilized BBI may compete with the endogenous inhibitor for the active site of the enzyme. Washing the column with urea could dislodge the enzyme from the BBI and, consequently, active enzyme would elute from the column. Our results with trypsin and DFP also suggest that an endogenous inhibitor of the 45-kDa protease is present in these cells. Trypsin may activate the protease by selectively digesting the inhibitor and causing it to dissociate from the enzyme on the SDS-dontaining zymograms. While this activity was inhibited when the zymograms were incubated in reaction buffer containing DFP, no inhibition was observed when trypsin-activated material was treated with DFP before being run on the zymogram. While trypsin treatment may partially digest the inhibitor of this enzyme, the enzyme inhibitor complex may still be present in solution and hence DFP would be unable to interact with the active site of the enzyme. In contrast, inhibition of the 45-kDa activity was observed when the zymograms were incubated in DFP, suggesting that the enzyme dissociates from the inhibitor after electrophoretic separation of the samples. Endogenous inhibitors of proteases are known to exist; for example, a fibroblast-derived urokinase inhibitor (UK-1) has been described that forms an SDS-stable complex with urokinase (41). Other assays were performed that have helped to further characterize this protease activity. The proteolytic activity was not affected by EDTA and 1,10-phenanthroline, compounds that inhibit metalloproteases (36). These results suggest that this activity is distinct from collagenases, which are potently inhibited by EDTA and other chelators of metal ions (36). The fact that this protease was inhibited by the BBI [which contains specific trypsin and chymotrypsin inhibitory sites (25)], but not by chymostatin [a highly specific inhibitor of chymotrypsin (42)], suggests that the active site architecture of this protease may have some characteristics similar to those of trypsin. This enzyme is similar to plasmin by virtue of its ability to cleave gelatin and its inhibition by DFP. However, it differs from plasmin in several important respects. Plasmin is secreted from cells as an inactive precursor, plasminogen, which is converted into plasmin by plasminogen activator. The protease we have identified in C3H/ 10T1/2 mouse embryo fibroblast cells is intracellular and is not secreted from cells. We have directly compared the mobility of the 45-kDa protease activity obtained from C3H/10T'/2 cells with plasmin on gelatin zymograms; plasmin yields a band of proteolytic activity of "'85 kDa, while the protease described in this paper is about half this size. Thus, the sizes of these proteases are different, as would be expected given the report that plasmin consists of a disulfide-linked dimer containing two subunits of 60 and 25 kDa (43). In addition, the functional characteristics of these proteases are different; for example, plasmin cleaves casein, while the 45-kDa protease does not cleave casein. While the mechanism(s) by which protease inhibitors suppress carcinogenesis is not understood, it is likely that they operate by interacting with a protease. Several lines of evidence indicate that the protease described in this report is a likely target of anticarcinogenic protease inhibitors, as a typical anticarcinogenic protease inhibitor, BBI, interacts with the 45-kDa protease in a manner that corresponds to the
3124
Medical Sciences: Billings and Habres
way in which it suppresses transformation. For example, the 45-kDa protease activity is inhibited in C3H/10T'/2 mouse cells and 1522 human cells by BBI, which is known to suppress transformation in these two cell systems (10, 12, 13). The critical time period for anticarcinogenic protease inhibitors to affect transformation corresponds to the time at which the 45-kDa protease is expressed. Protease inhibitors are able to suppress transformation only if they are added to proliferating cells; they have no effect on transformation when given to confluent cells (11, 16). The 45-kDa proteolytic activity is present in rapidly dividing, logarithmically growing cells, but it is present in reduced amounts in nondividing, confluent cells. These observations indicate that the critical target enzymes of anticarcinogenic protease inhibitors must be expressed in dividing cell populations, as is the 45-kDa proteolytic activity described here. Furthermore, it is known that cell growth after carcinogen exposure is a necessary prerequisite for transformation to occur in C3H/10T1/2 cells (11, 16) as well as in other model systems. The fact that the protease is induced by PMA treatment also links it to transformation in vitro and to carcinogenesis in vivo and the suppression of these phenomena by protease inhibitors. Protease inhibitors are strongly antipromotional in both in vivo and in vitro model systems (reviewed in refs. 10 and 12). In fact, protease inhibitors are thought to be mainly antipromotional agents, as the first report that they had anticarcinogenic activity was from Troll et al. (18), who showed that synthetic protease inhibitors could suppress tumor promotion in mouse skin two-stage carcinogenesis experiments. The fact that the anticarcinogenic protease inhibitor BBI can inhibit the PMAinducible 45-kDa protease activity suggests a mechanism for the antipromotional properties of protease inhibitors. From our results, we believe the 45-kDa protease activity described in this report may play an important role in the development of a cancer cell. We thank Dr. Ann R. Kennedy, Dr. Albert Owen, and Dr. Clarence Ryan for helpful discussion. This work was supported by National Institutes of Health Grant CA 45734. 1. Correa, P. (1981) Cancer Res. 41, 3685-3690. 2. Doll, R. & Peto, R. (1981) J. Nati. Cancer Inst. 66, 1193-1308. 3. Mills, P. K., Beeson, W. L., Abbey, D. E., Fraser, G. E. & Phillips, R. L. (1988) Cancer 61, 2578-2585. 4. Phillips, R. L. (1975) Cancer Res. 35, 3513-3522. 5. Birk, Y. (1976) Methods Enzymol. 45, 695-751. 6. Baturay, N. Z. & Kennedy, A. R. (1986) Cell Biol. Toxicol. 2, 21-32. 7. Billings, P. C., Morrow, A. R., Ryan, C. A. & Kennedy, A. R. (1989) Carcinogenesis 10, 687-691. 8. Billings, P. C., Newbeme, P. M. & Kennedy, A. R. (1990) Carcinogenesis 11, 1083-1086. 9. Hozumi, M., Ogawa, M., Sugimura, T., Takeuchi, T. & Umezawa, H. (1972) Cancer Res. 32, 1725-1728. 10. Kennedy, A. R. (1984) in Vitamins, Nutrition and Cancer, ed. Prasad, K. N. (Karger, Basel), pp. 166-179.
Proc. Natl. Acad. Sci. USA 89 (1992) 11. Kennedy, A. R. (1985) Carcinogenesis 6, 1441-1446. 12. Kennedy, A. R., in Protease Inhibitors as Caflcer Chemopreventive Agents, eds. Troll, W. & Kennedy, A. R. (Plenum, New York), in press. 13. Kennedy, A. R. & Billings, P. C. (1987) in Proceedings of the 2nd International Congress on Anticarcinogenesis and Radiation Protection, eds. Cerutti, P., Nygaard, 0. F. & Simic, M. G. (Plenum, New York), pp. 285-295. 14. Kuroki, T. & Drevon, C. (1979) Cancer Res. 39, 2755-2761. 15. Messadi, D. A., Billings, P. C., Shklar, G. & Kennedy, A. R. (1986) J. Natl. Cancer Inst. 76, 447-452. 16. St. Clair, W. H. (1991) Carcinogenesis 12, 935-937. 17. St. Clair, W. H., Billings, P. C., Carew, J. A., KellerMcGandy, C., Newberne, P. & Kennedy, A. R. (1990) Cancer Res. 50, 580-586. 18. Troll, W., Klassen, A. & Janoff, A. (1970) Science 169, 1211-1213. 19. Weed, H., McGandy, R. B. & Kennedy, A. R. (1985) Carcinogenesis 6, 1239-1241. 20. Witschi, H. & Kennedy, A. R. (1989) Carcinogenesis 10, 2275-2277. 21. Yavelow, J., Collins, M., Birk, Y., Troll, W. & Kennedy, A. R. (1985) Proc. Natl. Acad. Sci. USA 82, 5395-5399. 22. Billings, P. C., Carew, J. A., Keller-McGandy, C. E., Goldberg, A. L. & Kennedy, A. R. (1987) Proc. Natl. Acad. Sci. USA 84, 4801-4805. 23. Billings, P. C., St. Clair, W., Owen, A. J. & Kennedy, A. R. (1988) Cancer Res. 48, 1798-1802. 24. Billings, P. C., Habres, J. M. & Kennedy, A. R. (1990) Carcinogenesis 11, 329-332. 25. Birk, Y. (1985) Int. J. Pept. Protein Res. 25, 113-131. 26. Carew, J. A. & Kennedy, A. R. (1990) Cancer Lett. 49, 153163. 27. Yavelow, J., Caggana, M. & Beck, K. A. (1987) Cancer Res. 47, 1598-1601. 28. Yavelow, J., Scott, C. B. & Mayer, T. C. (1987) Cancer Res. 47, 1602-1607. 29. Billings, P. C., Jin, T., Ohnishi, N., Liao, D. C. & Habres, J. M. (1991) Carcinogenesis 12, 653-657. 30. Kassell, B. (1970) Methods Enzymol. 19, 853-862. 31. Reznikoff, C. A., Brankow, D. W. & Heidelberger, C. (1973) Cancer Res. 33, 3231-3238. 32. Terzaghi, M. & Little, J. B. (1976) Cancer Res. 36, 1367-1374. 33. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 34. Heussen, C. & Dowdle, E. B. (1980) Anal. Biochem. 102,
1%-202. 35. Bradford, M. (1976) Anal. Biochem. 72, 248-254. 36. Woessner, J. F. (1991) FASEB J. 5, 2145-2154. 37. Bond, J. S. & Butler, P. E. (1987) Annu. Rev. Biochem. 56, 333-364. 38. Shaw, E. (1972) Methods Enzymol. 25, 655-671. 39. Long, S. D., Quigley, J. P., Troll, W. & Kennedy, A. R. (1981) Carcinogenesis 2, 933-936. 40. Wigler, M. & Weinstein, I. B. (1976) Nature (London) 259, 232-233. 41. Masuzawa, M., Hamazaki, H., Nishioka, K., Nishiyama, S. & Ryan, T. J. (1987) Biochem. Biophys. Res. Commun. 149, 866-873. 42. Umezawa, H. (1976) Methods Enzymol. 45, 678-695. 43. Barrett, A. J. & McDonald, J. K. (1980) Mammalian Proteases (Academic, New York), Vol. 1.
A growth-regulated protease activity that is inhibited by the anticarcinogenic Bowman-Birk protease inhibitor PAUL C. BILLINGS
AND
JOAN M. HABRES
Department of Radiation Oncology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104
Communicated by Clarence A. Ryan, October 18, 1991
ABSTRACT The Bowman-Birk protease inhibitor (BBI) has been shown to be an effective suppressor of carcinogenesis in vivo and in vitro. To elucidate the mechanism(s) by which BBI suppresses carcinogenesis, we believe it will be necessary to identify and characterize the target enzymes that specifically interact with the BBI. We have shown previously that several cellular proteins in C3H/1OT'/2 mouse embryo fibroblast cells specifically bind to a BBI affinity resin. In the current report, we demonstrate that one of these proteins has proteolytic activity as judged by its ability to degrade gelatin. The enzyme has a mass of 45 kDa and subcellular fractionation experiments demonstrate that this enzyme is located in the cytosol. Furthermore, the proteolytic activity was inhibited by diisopropylfluorophosphate but was not affected by EDTA, indicating that this enzyme is a serine protease. Higher levels of protease activity were found in logarithmic-phase C3H/1OTI/2 cells compared with nondividing (confluent) cells, suggesting that this protease activity is growth regulated. Similar levels of this activity were present in nontransformed and in radiationtransformed C3H/1OT'/2 cells. Treatment of nontransformed C3H/1OT1/2 cells with phorbol 12-myristate 13-acetate increased the specific activity of this protease 5- to 10-fold. Our results suggest that this protease is a target enzyme of the BBI in these cells.
mation of BALB/3T3 cells induced by radiation or chemical carcinogens (benzo[a]pyrene and P-propiolactone) (6). We have shown that BBI as well as other anticarcinogenic protease inhibitors are internalized by C3H/10T1/2 cells (7, 28, 29). A major focus of our research is to elucidate the mechanism(s) by which protease inhibitors such as BBI suppress carcinogenesis. To accomplish this goal, we believe it will be necessary to identify and characterize the target enzymes that specifically interact with the BBI. In a previous study, we showed that a 45-kDa protein from mouse fibroblasts specifically binds to a BBI affinity column (23). In this report, we demonstrate that this protein is a serine protease with unusual characteristics.
MATERIALS AND METHODS Chemicals. The BBI was purified as described (17, 30). [35S]Methionine and Enlightning were obtained from New England Nuclear. Gelatin (300 Bloom) was obtained from Sigma. CNBr-activated Sepharose was obtained from Phar-
Epidemiological data suggest that nutritional factors play a major role in the etiology of cancer at many different sites (1-4). For example, high dietary levels of legumes have been associated with low cancer rates in general and are inversely correlated with the incidence of breast, colon, pancreatic, and prostate cancer (1-4). Legumes are known to contain high concentrations of protease inhibitors (5). While protease inhibitors have been shown to have strong anticarcinogenic activity in in vivo and in vitro cancer model systems (for examples, see refs. 6-21), relatively little is known about the precise mechanism(s) by which these compounds exert their suppressive effects. We have hypothesized that protease inhibitors block carcinogenesis by inhibiting cellular enzymes involved in induction and/or expression of the transformed phenotype (22-24). The anticarcinogenic activity of the soybean-derived Bowman-Birk protease inhibitor (BBI) has been extensively studied in our laboratory (6, 8, 13, 17, 19-21). The BBI is an 8-kDa protein that effectively inhibits trypsin and chymotrypsin (25) as well as several other proteases (22, 24-27). BBI has been shown to suppress dimethylbenz[a]anthraceneinduced cheek pouch carcinogenesis when topically applied (15), 3-methylcholanthrene-induced lung tumorigenesis when injected (20), and dimethylhydrazine-induced colon and liver carcinogenesis in mice when present in the diet (8, 17, 19). The BBI has also been shown to suppress radiation-induced transformation of C3H/1OT/2 mouse embryo fibroblast cells at nanomolar concentrations (21) and to inhibit the transfor-
macia. Cells. C3H/10T1/2 mouse embryo fibroblast cells and the radiation-transformed subclones F-17 and F-29 were grown as described (31, 32). Radlolabeling of Cellular Proteins. Cells were metabolically labeled with [3 S]methionine (50 ,uCi/ml; 1 Ci = 37 GBq) for 12 hr (23). After labeling, the cells were washed twice with ice-cold phosphate-buffered saline (PBS), removed from the dishes with a rubber policeman, and pelleted by centrifugation at 1000 x g for 5 min. The cell pellet was resuspended in 50 mM phosphate buffer, pH 7.0/1 mM MgCl2/0.1% Triton X-100, or 100 mM Tris-HCI, pH 7.0/0.1% Triton X-100, sonicated on ice for 30 sec and centrifuged at 10,000 x g for 5 min at 40C. Fractionation of Cellular Proteins over the BBI Affinity Column. The BBI affinity resin was prepared as described (23). Samples were applied to the affinity column in binding buffer (100 mM phosphate buffer, pH 7.0/1 mM MgCl2) at room temperature. The column was washed with 50 ml of washing buffer (50 mM phosphate buffer, pH 7.0/500 mM NaCl); bound proteins were eluted from the column with 5 M urea. Cellular proteins were analyzed on 12% polyacrylamide gels containing 0.1% SDS (33) and stained. For autoradiography, the gels were impregnated with Enlightning, dried in a gel dryer, and autoradiographed (23). Zymogram Protease Assays. Twelve percent polyacrylamide gels containing 0.1% gelatin were cast (34). Samples were applied to the gel in standard SDS gel loading buffer containing 0.1% SDS but lacking 2-mercaptoethanol and were not boiled before loading. The gels were run at a constant 70-80 V and were then soaked in 200 ml of 2% Triton X-100 in distilled water on a gyratory shaker for 1 hr
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Abbreviations: APMA, aminophenylmercuric acetate; BBI, Bowman-Birk inhibitor; DFP, diisopropylfluorophosphate; PMA, phorbol 12-myristate 13-acetate. 3120
Medical Sciences:
Billings and Habres
at 20'C. Next, the gels were soaked in 100 mM Tris HCl, pH 8.0/5 mM CaC12 for 12 hr at 370C and then stained. The zymograms presented are representative results obtained from two or more independent experiments. Subcellular Fractionation. Logarithmically growing C3H/ 10T1/2 cells were washed twice with ice-cold PBS, scraped from the dishes, and pelleted by centrifugation at 1000 X g for 10 min. The cells were resuspended in ice-cold isotonic sucrose buffer (10 mM Tris-HCI, pH 7.0/250 mM sucrose/1 mM MgCl2) and homogenized on ice in a Dounce homogenizer (100 strokes). The homogenate was centrifuged (all centrifugations were carried out at 40C) at 1000 x g for 10 min, 10,000 x g for 10 min, and 100,000 x g for 60 min (22). Each subcellular fraction was assayed for protease and marker enzyme activity. The lysosomal enzyme P-glucuronidase and the cytosolic enzyme lactate dehydrogenase were assayed as described (22). The protein present in each sample was determined by the method of Bradford (35). Trypsin Activation of Cell Homogenates. One hundred microliters of cell homogenate (50 ug of protein) was incubated with 1 Aul of trypsin (75 gg/ml) in PBS for 1 hr at 20TC.
RESULTS We have previously shown that two proteins of =45 and =60 kDa specifically bind to a BBI affinity column (23). We detected protease activity in the fractions binding to the column by using a nonspecific substrate (23). While these results indicated that functional enzyme activity was eluting from the BBI affinity column, they did not provide definitive information about which of these proteins had protease activity. To directly address this question, C3H/10T'/2 cell proteins were labeled with [35S]methionine and passed over the BBI affinity column; bound proteins were eluted from the column with 5 M urea (Fig. 1). In these studies, 0.05-0.1% of the total counts applied to the column eluted with the urea wash. A single band of protease activity eluted with the urea wash; this single band was readily observed on gelatincontaining zymograms (Fig. 1). Autoradiography of the zymogram revealed two labeled proteins (Fig. 1). Only one of these proteins had proteolytic activity, however. Excising the band containing protease activity from the zymogram and running it on a second-dimension reducing gel (the sample was incubated in 50 mM 2-mercaptoethalaol before being run in the second dimension) revealed a single protein of -45 kDa (Fig. 1). Earlier studies demonstrated that a 45-kDa protein can be eluted from the BBI affinity resin with HCl (23). We consistently observed low levels of protease activity in this size range when total cell homogenates or material passed over a Sephadex gel filtration column were run on the zymograms. Several types of proteases, such as collagenases, require activation by trypsin or organomercurial compounds [such as aminophenylmercuric acetate (APMA)] to become functionally active proteolytic enzymes (36). To determine whether the 45-kDa protease is present in C3H/ 10T1/2 cells in a cryptic or inactive form, cell homogenates were prepared and incubated with APMA or trypsin and subsequently analyzed on gelatin-containing zymograms. Greatly increased levels of protease activity were observed in trypsin-treated samples, while no increase in activity was observed in untreated or in APMA-treated samples (Fig. 2). The trypsin-activated protease comigrates with the protease eluting from the BBI affinity column. These results suggest that this protease is normally in an inactive form in these cells. The 45-kDa protease activity was present in growing cells, while significantly lower levels of this activity were present in nondividing, confluent cells; the activity present in proliferating cells was inhibited by the BBI (Fig. 3). Similar levels of protease activity were observed in confluent cells and in
Proc. Natl. Acad. Sci. USA 89 (1992)
3121
A 200 Vt)
100
0 x
C 1-1
6 c
0
4
C.)
2
5
0
10
15
20
25
Fraction No.
1 2
B
C
wsx
le
-a
67 45
31
21
FIG. 1. Elution of protease activity from the BBI affinity column. (A) Elution profile of 35S-labeled C3H/1OTY2 cell proteins from the BBI affinity resin. Bound proteins were eluted with 5 M urea (arrow). (B) Analysis of material eluting with the urea wash. Lanes: 1, material was analyzed on a gelatin-containing zymogram (note the band of protease activity); 2, autoradiograph of gel shown in lane 1. (C) Band containing protease activity was excised from the gel and run on a second-dimension reducing gel. Numbers are molecular mass markers (kDa).
cells grown in serum-free medium (Figs. 2 and 3). Therefore, expression of this proteolytic activity is correlated with the state of cellular growth; relatively high levels of activity are observed in actively growing cells, while low levels of activity are observed in stationary cells. We have observed this proteolytic activity in two radiation-transformed C3H/ 10T'/2 cell lines, F-17 (Fig. 3) and F-29; slightly elevated levels of this activity were seen in F-29 cells (data not shown). We have observed a similar activity in AG1522 human fibroblasts, in which BBI has been shown to suppress transformation (10) and in human bladder epithelial cells (data not shown). To determine the intracellular location of this proteolytic activity, subcellular fractionation experiments were performed. The majority of protease activity was present in the 100,000 x g supernatant fraction, indicating that this protease is located in the cytosolic fraction (Fig. 4, Table 1). We also determined the ability of this protease to cleave other substrates. While the 45-kDa protease efficiently cleaved gelatin, no protease activity was observed when casein was incorporated into the zymograms. We characterized the 45-kDa activity by using a series of protease inhibitors with defined inhibitory specificity. For these experiments, two procedures were used: (i) the sam-
Proc. Natl. Acad. Sci. USA 89 (1992)
Medical Sciences: Billings and Habres
3122
8
4
58
4 8.5
P.6
FIG. 4. Analysis of subcellular fractions for protease activity.
C3H/10T'/2 cells were disrupted in a Dounce homogenizer and FIG. 2. Analysis of cell extracts and conditioned medium for proteolytic activity. Samples were unactivated or activated with trypsin and then run on gelatin zymograms. Lanes: 1, unactivated cell extract; 2, cell extract activated with trypsin before being run on the zymogram (note band of protease activity); 3 and 4, untreated (lane 3) and trypsin-activated cell extracts (lane 4) were run on the zymogram (the zymogram was soaked in incubation buffer containing BBI); 5-8, conditioned medium from untreated or from phorbol 12-myristate 13-acetate (PMA)-treated C3H/10T1/2 cells. Cells were grown in serum-free medium in the absence or presence of PMA (100 ng/ml) for 48 hr. Two-milliliter aliquots of conditioned medium were concentrated by ultrafiltration and analyzed on zymograms. Lane 5, conditioned medium from untreated cells; lane 6, conditioned medium from cells treated with PMA; lane 7, trypsin-activated conditioned medium from control cells; lane 8, trypsin-activated conditioned medium from cells treated with PMA. The 45-kDa protease is not secreted by these cells. Numbers on left are molecular mass markers (kDa).
ples were preincubated with the inhibitor and then run on the zymograms, or (ii) the samples were run on gelatin zymograms and the gels were subsequently incubated in 100 mM Tris'HCI, pH 8.0/10 mM CaCl2 in the absence or presence of .
4I
Awmw ts sf
:..,.
A
2
4
b
;
4 8 .5 B
2 6. 6
-
FIG. 3. (A) Protease activity in growing and in growth-arrested C3H/10T1/2 cells. Cell extracts from the indicated cell population were prepared and activated with trypsin. Samples were analyzed on zymograms for protease activity. Lanes: 1, confluent cells; 2, logarithmic-phase cells; 3, F-17 [radiation transformed C3H/10T'/2 cells (32)]. (B) Effect of serum, DFP, and PMA on 45-kDa proteolytic activity. Cells were grown in the absence or presence of PMA (100 ng/ml) for 48 hr in complete (containing 10%o fetal calf serum) or in serum-free medium. Lanes 1-4, cells grown in complete medium; lanes 5-8, cells grown in serum-free medium for 72 hr. Lanes: 1, untreated cell extract (not activated with trypsin); 2, cell extract from cells treated with PMA; 3, trypsin-activated control cell extract; 4, trypsin-activated cell extract from PMA-treated cultures; 5, trypsinactivated extract from untreated cells; 6, trypsin-activated extract from PMA-treated cells; 7 and 8, same as 5 and 6 except that the zymogram was incubated in reaction buffer containing 10 mM DFP. Note that DFP completely inhibits 45-kDa proteolytic activity. Numbers on left are molecular mass markers (kDa).
subcellular fractions were isolated by differential centrifugation and analyzed for protease activity on a gelatin zymogram. Lanes: 1, 3, 5, and 7, untreated fractions; 2, 4, 6, and 8, trypsin-activated fractions; 1 and 2, 1000 x g pellet; 3 and 4, 10,000 x g pellet; 5 and 6, 100,000 x g pellet; 7 and 8, 100,000 x g supernatant. The bulk of the 45-kDa proteolytic activity is present in the 100,000 x g supernatant (arrow). Numbers on left are molecular mass markers (kDa).
the indicated protease inhibitor for 12 hr at 370C. As expected, protease activity was inhibited by the BBI (Fig. 4). It was also inhibited by antipain but was unaffected by EDTA, 1,10-phenanthroline, or chymostatin (Table 2). When trypsin-activated material was preincubated with diisopropylfluorophosphate (DFP) and then run on the zymogram, no inhibition of this activity was observed. However, when trypsin-activated material was run on the zymogram and then incubated in reaction buffer containing DFP, inhibition ofthis protease activity was observed (Table 1). Maximal protease activity was observed at pH 8. These results suggest that this activity is a neutral serine protease (37, 38). To determine whether the 45-kDa proteolytic activity was secreted, conditioned medium from C3H/1OT'/2 cells was analyzed for protease activity. This proteolytic activity was not secreted from these cells (Fig. 2). We also determined the effect of PMA on the levels of this enzyme. When cells were grown in complete medium containing serum, PMA increased the levels of this activity 3- to 6-fold (Fig. 3). However, when cells were grown in serum-free medium, PMA increased the levels of this activity 5- to 10-fold. Furthermore, PMA did not induce the secretion of the 45-kDa protease into the culture medium (Fig. 2). The proteolytic activity present in PMA-treated cells was inhibited by the BBI as well as DFP (Fig. 3, Table 2). Table 1. Marker enzyme and proteolytic activity in subcellular fractions Marker enzyme activity Protease LDH activity jS-Glu Subcellular fraction ND x 2.3 0.9 (1.3) (0.5) Pellet (1000 g) ND 0.9 (0.1) 3.8 (0.3) Pellet (10,000 x g) ND 3.4 (0.5) 2.7 (0.4) Pellet (100,000 x g) + Supernatant (100,000 x g) 1.2 (0.3) 15.6 (3.7) C3H/10T1/2 cells were homogenized on ice in a Dounce homogenizer. The homogenate was centrifuged at 1000, 10,000, and 100,000 x g. Each subcellular fraction was assayed for marker enzyme and protease activity. f3-Glucuronidase (fGlu) activity is expressed as A540 per 12 hr per mg of protein. Numbers in parentheses indicate total enzyme activity present expressed as specific activity times total protein present in the sample (AS0 per 12 hr). Lactate dehydrogenase (LDH) activity is expressed as A30 per min per mg of protein. Numbers in parentheses indicate total enzyme activity present expressed as specific activity times total protein present in the sample (Amo per min). For determination of 45-kDa protease activity, each subcellular fraction was assayed (see Fig. 4). ND, protease activity not detected; +, protease activity present.
Medical Sciences:
Billings and Habres
Table 2. Effect of different compounds on 45-kDa proteolytic activity in C3H/10T1/2 cells Type of Effect on protease proteolytic affected activity Compound Thiol Inhibit DTT (5 mM) Thiol None IA (5 mM) Serine Inhibit DFP (10 mM)* Metallo None EDTA (20 mM) Metallo None 1,10-Phenanthroline (0.5 mM) Serine/thiol Inhibit Antipain (10 ,.g/ml) Serine Inhibit BBI (10 /Ag/ml) Serine None Chymostatin (10 t±g/ml) Serine Inhibit SBTI (10 ,ug/ml) Cell homogenates were activated with trypsin and then incubated in the presence of dithiothreitol (DTT) or iodoacetamide (IA) for 10 min at 20'C and then run on gelatin-containing gels. For the other inhibitors, the cell homogenates were activated with trypsin and then run on gelatin-containing gels. The gels were incubated in reaction buffer containing the indicated inhibitor. SBTI, soybean trypsin inhibitor. *Proteolytic activity was inhibited when the zymogram was incubated in reaction buffer containing DFP.
DISCUSSION The BBI has been shown to be an effective suppressor of radiation and chemical carcinogen-induced transformation of C3H/1OT'/2 cells. We have hypothesized that protease inhibitors, such as BBI, block carcinogenesis by inhibiting cellular enzymes involved in the induction and/or expression of the transformed phenotype (22-24). In this report, we have identified an intracellular proteolytic activity in C3H/10T1/2 fibroblast cells that is inhibited by the anticarcinogenic BBI. To our knowledge, an intracellular protease that is growth regulated and induced by PMA has not been previously reported. These characteristics suggest that it may play an important role in carcinogenesis. The proteolytic activity described in the current study was present in nontransformed C3H/1OT'/2 cells and in two radiation-transformed subclones, F-17 and F-29 cells. We have observed this activity in other cell types as well, including human fibroblasts and human bladder epithelial cells. Greatly reduced levels of this protease were observed in confluent cells, and in nongrowing cells maintained in serum-free medium, compared to logarithmically growing cells (Fig. 3). These results suggest that this enzyme is growth related. Further evidence that the activity is growth related is its response to a mitogen; PMA treatment increased the levels of this enzyme 5- to 10-fold. There are previous studies indicating that PMA is capable of inducing proteolytic activity-specifically, plasminogen activator (39, 40). Plasminogen activator, however, is a secreted protease, while the protease activity identified and characterized in this report is clearly an intracellular (i.e., cytosolic) activity that is not secreted into the growth medium (Fig. 2). It is likely that the 45-kDa proteolytic activity described here was not observed previously by other investigators as it is displayed only under somewhat unusual conditions. While we were able to obtain active enzyme from the BBI affinity resin, no proteolytic activity was observed when cell homogenates were run directly on the zymograms. Furthermore, size fractionation of cell homogenates over a Sephadex gel filtration column or treatment with APMA also failed to activate this activity. However, treatment of cell homogenates with small amounts of trypsin did yield a preparation with a 45-kDa proteolytic activity. The protease activity purified by affinity chromatography appears to be similar to the trypsin-activatable activity present in cell extracts for the following reasons: (i) the protease eluting from the BBI
Proc. Natl. Acad. Sci. USA 89 (1992)
3123
affinity column comigrates with the trypsin-activatable protease in the zymograms, (ii) both proteases cleave gelatin, (iii) the trypsin-activatable protease is inhibited by the BBI, as is the protease eluting from the BBI affinity column. One explanation for the lack of 45-kDa proteolytic activity in untreated cell homogenates may be that the enzyme is tightly complexed with an endogenous inhibitor that greatly reduces its activity. Our results support the existence of such an inhibitor. Proteolytic activity was observed when the cell homogenates were run over the BBI affinity column; under these conditions, the immobilized BBI may compete with the endogenous inhibitor for the active site of the enzyme. Washing the column with urea could dislodge the enzyme from the BBI and, consequently, active enzyme would elute from the column. Our results with trypsin and DFP also suggest that an endogenous inhibitor of the 45-kDa protease is present in these cells. Trypsin may activate the protease by selectively digesting the inhibitor and causing it to dissociate from the enzyme on the SDS-dontaining zymograms. While this activity was inhibited when the zymograms were incubated in reaction buffer containing DFP, no inhibition was observed when trypsin-activated material was treated with DFP before being run on the zymogram. While trypsin treatment may partially digest the inhibitor of this enzyme, the enzyme inhibitor complex may still be present in solution and hence DFP would be unable to interact with the active site of the enzyme. In contrast, inhibition of the 45-kDa activity was observed when the zymograms were incubated in DFP, suggesting that the enzyme dissociates from the inhibitor after electrophoretic separation of the samples. Endogenous inhibitors of proteases are known to exist; for example, a fibroblast-derived urokinase inhibitor (UK-1) has been described that forms an SDS-stable complex with urokinase (41). Other assays were performed that have helped to further characterize this protease activity. The proteolytic activity was not affected by EDTA and 1,10-phenanthroline, compounds that inhibit metalloproteases (36). These results suggest that this activity is distinct from collagenases, which are potently inhibited by EDTA and other chelators of metal ions (36). The fact that this protease was inhibited by the BBI [which contains specific trypsin and chymotrypsin inhibitory sites (25)], but not by chymostatin [a highly specific inhibitor of chymotrypsin (42)], suggests that the active site architecture of this protease may have some characteristics similar to those of trypsin. This enzyme is similar to plasmin by virtue of its ability to cleave gelatin and its inhibition by DFP. However, it differs from plasmin in several important respects. Plasmin is secreted from cells as an inactive precursor, plasminogen, which is converted into plasmin by plasminogen activator. The protease we have identified in C3H/ 10T1/2 mouse embryo fibroblast cells is intracellular and is not secreted from cells. We have directly compared the mobility of the 45-kDa protease activity obtained from C3H/10T'/2 cells with plasmin on gelatin zymograms; plasmin yields a band of proteolytic activity of "'85 kDa, while the protease described in this paper is about half this size. Thus, the sizes of these proteases are different, as would be expected given the report that plasmin consists of a disulfide-linked dimer containing two subunits of 60 and 25 kDa (43). In addition, the functional characteristics of these proteases are different; for example, plasmin cleaves casein, while the 45-kDa protease does not cleave casein. While the mechanism(s) by which protease inhibitors suppress carcinogenesis is not understood, it is likely that they operate by interacting with a protease. Several lines of evidence indicate that the protease described in this report is a likely target of anticarcinogenic protease inhibitors, as a typical anticarcinogenic protease inhibitor, BBI, interacts with the 45-kDa protease in a manner that corresponds to the
3124
Medical Sciences: Billings and Habres
way in which it suppresses transformation. For example, the 45-kDa protease activity is inhibited in C3H/10T'/2 mouse cells and 1522 human cells by BBI, which is known to suppress transformation in these two cell systems (10, 12, 13). The critical time period for anticarcinogenic protease inhibitors to affect transformation corresponds to the time at which the 45-kDa protease is expressed. Protease inhibitors are able to suppress transformation only if they are added to proliferating cells; they have no effect on transformation when given to confluent cells (11, 16). The 45-kDa proteolytic activity is present in rapidly dividing, logarithmically growing cells, but it is present in reduced amounts in nondividing, confluent cells. These observations indicate that the critical target enzymes of anticarcinogenic protease inhibitors must be expressed in dividing cell populations, as is the 45-kDa proteolytic activity described here. Furthermore, it is known that cell growth after carcinogen exposure is a necessary prerequisite for transformation to occur in C3H/10T1/2 cells (11, 16) as well as in other model systems. The fact that the protease is induced by PMA treatment also links it to transformation in vitro and to carcinogenesis in vivo and the suppression of these phenomena by protease inhibitors. Protease inhibitors are strongly antipromotional in both in vivo and in vitro model systems (reviewed in refs. 10 and 12). In fact, protease inhibitors are thought to be mainly antipromotional agents, as the first report that they had anticarcinogenic activity was from Troll et al. (18), who showed that synthetic protease inhibitors could suppress tumor promotion in mouse skin two-stage carcinogenesis experiments. The fact that the anticarcinogenic protease inhibitor BBI can inhibit the PMAinducible 45-kDa protease activity suggests a mechanism for the antipromotional properties of protease inhibitors. From our results, we believe the 45-kDa protease activity described in this report may play an important role in the development of a cancer cell. We thank Dr. Ann R. Kennedy, Dr. Albert Owen, and Dr. Clarence Ryan for helpful discussion. This work was supported by National Institutes of Health Grant CA 45734. 1. Correa, P. (1981) Cancer Res. 41, 3685-3690. 2. Doll, R. & Peto, R. (1981) J. Nati. Cancer Inst. 66, 1193-1308. 3. Mills, P. K., Beeson, W. L., Abbey, D. E., Fraser, G. E. & Phillips, R. L. (1988) Cancer 61, 2578-2585. 4. Phillips, R. L. (1975) Cancer Res. 35, 3513-3522. 5. Birk, Y. (1976) Methods Enzymol. 45, 695-751. 6. Baturay, N. Z. & Kennedy, A. R. (1986) Cell Biol. Toxicol. 2, 21-32. 7. Billings, P. C., Morrow, A. R., Ryan, C. A. & Kennedy, A. R. (1989) Carcinogenesis 10, 687-691. 8. Billings, P. C., Newbeme, P. M. & Kennedy, A. R. (1990) Carcinogenesis 11, 1083-1086. 9. Hozumi, M., Ogawa, M., Sugimura, T., Takeuchi, T. & Umezawa, H. (1972) Cancer Res. 32, 1725-1728. 10. Kennedy, A. R. (1984) in Vitamins, Nutrition and Cancer, ed. Prasad, K. N. (Karger, Basel), pp. 166-179.
Proc. Natl. Acad. Sci. USA 89 (1992) 11. Kennedy, A. R. (1985) Carcinogenesis 6, 1441-1446. 12. Kennedy, A. R., in Protease Inhibitors as Caflcer Chemopreventive Agents, eds. Troll, W. & Kennedy, A. R. (Plenum, New York), in press. 13. Kennedy, A. R. & Billings, P. C. (1987) in Proceedings of the 2nd International Congress on Anticarcinogenesis and Radiation Protection, eds. Cerutti, P., Nygaard, 0. F. & Simic, M. G. (Plenum, New York), pp. 285-295. 14. Kuroki, T. & Drevon, C. (1979) Cancer Res. 39, 2755-2761. 15. Messadi, D. A., Billings, P. C., Shklar, G. & Kennedy, A. R. (1986) J. Natl. Cancer Inst. 76, 447-452. 16. St. Clair, W. H. (1991) Carcinogenesis 12, 935-937. 17. St. Clair, W. H., Billings, P. C., Carew, J. A., KellerMcGandy, C., Newberne, P. & Kennedy, A. R. (1990) Cancer Res. 50, 580-586. 18. Troll, W., Klassen, A. & Janoff, A. (1970) Science 169, 1211-1213. 19. Weed, H., McGandy, R. B. & Kennedy, A. R. (1985) Carcinogenesis 6, 1239-1241. 20. Witschi, H. & Kennedy, A. R. (1989) Carcinogenesis 10, 2275-2277. 21. Yavelow, J., Collins, M., Birk, Y., Troll, W. & Kennedy, A. R. (1985) Proc. Natl. Acad. Sci. USA 82, 5395-5399. 22. Billings, P. C., Carew, J. A., Keller-McGandy, C. E., Goldberg, A. L. & Kennedy, A. R. (1987) Proc. Natl. Acad. Sci. USA 84, 4801-4805. 23. Billings, P. C., St. Clair, W., Owen, A. J. & Kennedy, A. R. (1988) Cancer Res. 48, 1798-1802. 24. Billings, P. C., Habres, J. M. & Kennedy, A. R. (1990) Carcinogenesis 11, 329-332. 25. Birk, Y. (1985) Int. J. Pept. Protein Res. 25, 113-131. 26. Carew, J. A. & Kennedy, A. R. (1990) Cancer Lett. 49, 153163. 27. Yavelow, J., Caggana, M. & Beck, K. A. (1987) Cancer Res. 47, 1598-1601. 28. Yavelow, J., Scott, C. B. & Mayer, T. C. (1987) Cancer Res. 47, 1602-1607. 29. Billings, P. C., Jin, T., Ohnishi, N., Liao, D. C. & Habres, J. M. (1991) Carcinogenesis 12, 653-657. 30. Kassell, B. (1970) Methods Enzymol. 19, 853-862. 31. Reznikoff, C. A., Brankow, D. W. & Heidelberger, C. (1973) Cancer Res. 33, 3231-3238. 32. Terzaghi, M. & Little, J. B. (1976) Cancer Res. 36, 1367-1374. 33. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 34. Heussen, C. & Dowdle, E. B. (1980) Anal. Biochem. 102,
1%-202. 35. Bradford, M. (1976) Anal. Biochem. 72, 248-254. 36. Woessner, J. F. (1991) FASEB J. 5, 2145-2154. 37. Bond, J. S. & Butler, P. E. (1987) Annu. Rev. Biochem. 56, 333-364. 38. Shaw, E. (1972) Methods Enzymol. 25, 655-671. 39. Long, S. D., Quigley, J. P., Troll, W. & Kennedy, A. R. (1981) Carcinogenesis 2, 933-936. 40. Wigler, M. & Weinstein, I. B. (1976) Nature (London) 259, 232-233. 41. Masuzawa, M., Hamazaki, H., Nishioka, K., Nishiyama, S. & Ryan, T. J. (1987) Biochem. Biophys. Res. Commun. 149, 866-873. 42. Umezawa, H. (1976) Methods Enzymol. 45, 678-695. 43. Barrett, A. J. & McDonald, J. K. (1980) Mammalian Proteases (Academic, New York), Vol. 1.
