Inhibition of growth by transforming growth factor-beta following fusion ...

THEJOURNAL

OF

BIOLOGICAL CHEMISTRI

Vol. 267, No. 4, Issue of February 5, pp. 2588-2593, 1992 Printed in U.S.A.

Inhibition of Growth by Transforming Growth Factor-@ following Fusion of Two Nonresponsive Human Carcinoma CellLines IMPLICATION OF THE TYPE

I1 RECEPTORINGROWTHINHIBITORYRESPONSES* (Received for publication, July 12, 1991)

Andrew G. GeiserS, James K. Burmester, Renee Webbink, Anita B. Roberts, and Michael B. Sporn From the Laboratoryof Chemopreuention, National Cancer Institute, National Institutes of Health, Bethesda, Maryland20892

Loss of growth regulation by transforming growth factor+ (TGF-b) may be an important step in carcinogenesis. We have used a cell fusion system toshow that inhibition of growth by TGF-/3 can be restored to carcinoma cell lines that are unresponsive to the inhibitory effects of TGF-8.In a previous study, theEJ bladder carcinoma line was fused to the SW480 colon adenocarcinoma line and found to produce nontumorigenic hybrid cells along withone hybrid cell clone of low tumorigenicity. Here we show that the capacity of the nontumorigenic hybrid cells to respond to either TGF-81 or TGF-82 has been restored, while theparental or tumorigenic hybrid cells show little or no inhibition of growth following TGF-8 treatment. Crosslinking analyses with labeled TGF-81 demonstrated much higher levels of thetype I1 (85 kDa) receptor in the hybrid cells compared with the parental tumor lines. Both the parental and tumorigenic hybrid cell lines werecapable of responding to TGF-0 as evidenced by increased levels of mRNA for fibronectin, type IV collagenase, and plasminogen activator inhibitor after treatment with TGF-Bl. These results suggestthat the type I1 receptor is necessary for mediating the effects of TGF-@on inhibition of growth but not on gene activation of the hybrid cells.

The process of tumorigenesis may in part be due to such an imbalance involving TGF-P (10). Many cultured malignant or transformed epithelial cells have been shown to have lost sensitivity to regulation of growth by TGF-P. This may be indicative of escape from normal regulation of growth during the carcinogenic process. Such an escape could bedue to either theloss of specific receptors for TGF-P, or to alterations in the postreceptor signaling pathway(s) of TGF-/3 that lead to inhibition of growth. Both ways of rendering cells unresponsive to TGF-/3 have been documented (3, 11-13). Whole cell fusions betweentumorigenic and nontumorigenic cells and even fusions between pairs of cells that are both tumorigenic often lead to the creation of nontumorigenic hybrid cells (14, 15). This suppressionof tumorigenicity demonstrates the dominant nature of normal growth regulatory mechanisms. It isthoughtthatthe fusions resultinthe complementation of functional growth regulatory (tumor suppressor) genes that were lost or compromised duringthe process of carcinogenesis. These types of studies have been instrumental in identifying chromosomes on which specific tumor suppressor genes reside. However, little has been reported about how the hybrid cells respond to molecules like TGF-/3 that regulate growth. It would thus be of interest to see if regulation of growth by TGF-P can be restored as the nontumorigenic phenotype is restored to the hybrid cells. In this study,we have used cell hybrids as an approach to Members of the transforming growth factor /3 family of understanding therole of TGF-(3 in thecarcinogenic process. Here we describe the complementationof the TGF-P responproteins havebeenshown tohave diverseeffects onthe sive phenotype in whole cell hybrids formed from the fusion growth of many different cell types (for review see Ref. 1). Mammalian cells have the potential to express three different of two human carcinomacell lines shown to be nonresponsive isoforms of TGF-@’ (TGF-01,$2, and $3), all of which have to TGF-P. The reappearance of responsiveness to TGF-/3 in broadly similar effects on cells affected by TGF-8. TGF-/3 the hybrid cells correlated with a nontumorigenic phenotype I1 (85 kDa) cross-linked affects the growth either positively or negatively of many and with an increase in the type TGF-P receptor. different cell types both i n vitro and i n vivo (2-9). TGF-/3 is thus seen as an important central regulator which helps to MATERIALS AND METHODS coordinate the normal processes of growth and differentiation, Cell Culture Conditions and DNA Synthesis Assays-The generaas well as responses due to tissue injury and repairprocesses tion of the hybrid cells was previously described (16). The EJ cell (1).Should an imbalance in either production of, or reponse line was originally established from a human bladder carcinoma (see to, TGF-@ occur, normal regulation of growth may become Marshall et al. (17)) while the SW480 cell line (ATCC CCL 228) was perturbed. established from a human colon adenocarcinoma (18).All cells were routinely passaged in DMEM medium (GIBCO) supplemented with sodium pyruvate, penicillin + streptomycin, and 10% fetal bovine the payment of page charges. This article must therefore be hereby serum (GIBCO).For the DNA synthesis assays the cells were counted marked “aduertisement” in accordance with 18 U.S.C. Section 1734 and seeded in 24-well plates ata density of 2.5 X lo4 cells/well in 0.5ml DMEM supplemented with 1.0% fetal bovine serum. One hour solely to indicate this fact. after cells were seeded, TGF-P1 or TGF-P2 (R&D Systems) was added $ To whom correspondence should be addressed Lab. of Chemoprevention,NationalCancer Institute,NIH, Bldg. 41, Rm. C629, in growth medium plus 0.1% BSA to duplicate wells for each concentration of TGF-P used. After the desired incubation period, 0.25 rCi Bethesda, MD 20892. Tel.: 301-496-5391; Fax: 301-496-8395. ’ The abbreviations used are: TGF-P, transforming growth factor- of 5-[’”SI]iodo-2’-deoxyuridine (Amersham Corp.) was added to each well and incubated for 2 h at 37 “C. Cells were then fixed by direct @;BSA, bovine serum albumin; SDS, sodium dodecyl sulfate; PAI-1, plasminogen activator inhibitor-1; kb, kilobase(s); DMEM, Dulbec- addition of 1.0 ml of 3:l methanokacetic acid for 2h at room CO’Smodified Eagle’s medium; HEPES, 4-(2-hydroxyethyl)-l-piper- temperature. The cells were washed twice with 80% methanol and then solubilized with 1.0 ml of 1.0 M NaOH for 30 min. The solubilized azineethanesulfonic acid.

* The costs of publication of this article were defrayed in part by

2588

Complementation of TGF-6 Growth Inhibitory Responses extract (0.8 ml) was then counted in a y counter for relative DNA incorporation of 5-["'1]-iodo-2'-deoxyuridine. Receptor Cross-linking-Cells were plated at a density of 0.5 X lo6 cells/35-mm dish in growth medium 18 h prior to cross-linking. The monolayers of cells were washed once with binding buffer (DMEM, 25 mM HEPES, pH 7.4,l.O mg/ml BSA) and then incubated for 3 h at 4 "C with 100 PM 1""ITGF-/31, or 100 pM '*'I-TGF-/31 plus 40 nM unlabeled TGF-01in binding buffer where indicated. Monolayers were then washed 3 times with cold BSA-free binding buffer and cross-linked for 1h at 4 "C in the presence of BSA-free binding buffer containing 300 p~ disuccinimidyl suberate (Pierce). The cells were washed 3 times with cold buffer containing 250 mM sucrose, 10 mM Tris, pH 7.4, and 1 mM EDTA and then were scraped from the plate into buffer containing 1%Triton X-100, 10 mM Tris, pH 7.4, 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 1 +M pepstatin, and 1 p M leupeptin. Following centrifugation at 16,000 X g for 5 min to remove insoluble material, 100 pg of protein was loaded onto a 10% acrylamide gel for electrophoretic separation; the gel was then dried and exposed to film for 1-2 weeks. RNA and NorthernBlot Analysis-Cells were incubated in DMEM supplemented with 1.0% fetal bovine serum alone (control) or with 5 ng/ml TGF-pl for a period of 24 h. Total RNA was then isolated by the guanidinium hydrochloride method (19). RNA (15 pg/sample) was loaded onto 1.2% agarose, 2.2 M formaldehyde gels for electrophoretic separation, then transfered to Nytran filters (Northern blot). Hybridizations were done using the conditions described by Church et al. (20). Hybridization probes were labeled with [32P]dCTPusing a random primer labeling kit (BRL) and 3 X lo6 cpm of the probe per milliliter of hybridization buffer was used. Gelatin Zyrnography-Gelatinolytic activity was measured on minipolyacrylamide gels as by Brown et al. (21). Aliquots of conditioned medium (16 pl) were loaded onto 9% acrylamide gels containing 1 mg/ml gelatin (ICN Biochemicals). After removal of SDS, gels were incubated overnight at 37 "C in 50 mM Tris-HCI, pH 7.6, containing 0.2 M NaC1, 5 mM CaCI2,and 0.02% (w/v) Brij-35. After staining, gelatinolytic activity was seen as a clear band against a blue gelatin background.

RESULTS

The cells used in these studies have been described previously (16). Briefly, the human cell lines EJ and SW480 were fused and the hybrid cell colonies were isolated and characterized to confirm their hybrid status. Subsequent injections into nude (athymic) mice indicated that five of the six hybrid cell clones tested were nontumorigenic (HYB 1-5), while both of the parental cell lines were consistently tumorigenic. The tumorigenicity of the one hybrid clone which formed tumors was reduced compared to the parental lines and designated here as LT-HYB for Low Tumorigenic =rid. The studies described below were done on the parentalcell lines (EJ and SW480), LT-HYB, and HYB 1,2, and3 cell clones. Growth Response to TGF-p-We chose this particular hybrid system because both of the parentallines (and therefore the hybrid cells) were of epithelial cell derivation. While the growth of most normal epithelial cells is inhibited by TGF-P, many carcinoma-derived cells have been found to be refractory to inhibition of growth by TGF-fi (see Introduction). We therefore tested the ability of TGF-P1 toinhibit DNA replicationin theparentalandinthe hybrid cell lines after exposure of the cells to TGF-(31 for different time periods. While replication of theparental cells was not affected, growth of the hybrid cell clones was inhibited after treatment for 48 h and maximally inhibited in a dose-dependentmanner after a 72-h treatment with TGF-P1 (Fig. 1). Maximal suppression of DNA synthesis of approximately 80% was achieved with a concentration of -40 PM TGF-Pl. It should be noted thatthe one hybrid clone found to be slightly tumorigenic (LT-HYB) was only slightly inhibited by TGFPl, to a maximum of 30% at higher concentrations of TGF,Bl.Thus, thecell fusions resulted not only in the complementation of the tumor-suppressed phenotype from two tumorigenic parental cells, but also the complementation of the

2589

TGF-I31 (ng/ml)

FIG. 1. Inhibition of DNA synthesis by TGF-Dl. DNA synthesis was measured as described under "Materials and Methods" and percent inhibition was calculated as the percent difference between the counts of "'I (cpm) measured in cells treated with TGF01 for 72 h compared with nontreated cells. The cells were treated with a maximum of 3.0 ng/ml and 1:3 serial dilutions were made to generate the doses indicated. Each value represents the average of duplicate samples measured for each TGF-P1 concentration used. 100 "a-

-20 01

Lr-m

,

1

"I

10

TGF-131 (ng/ml)

FIG. 2. Inhibition of DNA synthesis by TGF-82. DNA synthesis and percent inhibition were plotted as in Fig. 1. Serial dilutions of 1:2 were made from a maximal concentration of4.0 ng/ml TGF-

02. TGF-p growth-responsive phenotype from two nonresponsive cells. In order to determine if this was a general response to TGF-0 isoforms, we also used TGF-02 and obtained similar results interms of extent of DNA suppression and concentration of TGF-02 needed for the suppressive effect (Fig. 2). The LT-HYB cell clone was again slightly inhibited, but only with the highest doses of TGF-p2. TGF-/3 Receptor Types-Complementation of the TGF-/3 growth-responsive phenotype could have occurred atthe TGF-0 receptor level or at thesignal transduction level upon stimulation of the receptor. To examine the possibility of receptor changes, radiolabeled TGF-01 was cross-linked to cell surfaces of the parental andhybrid cells and differences in the quantity or gel mobility of the three receptor forms (22-24) were noted. We found significant differences in the relative quantities of the receptor forms when comparing the parental with the hybrid cells from three separate crosslinking experiments (Fig. 3 and Table I). Both the EJ and SW480 cell lines displayed high levels of the 200-kDa (betaglycan-type 111)form and low levels of the 60-kDa (type I) and 85-kDa (type 11)forms. In contrast, the nontumorigenic hybrids displayed reduced levels of the type I11 receptor and a large increase in the level of the type I1 form (seen as a doublet, see also Ref. 25). A small increase in the level of the type I receptor was also noted in the nontumorigenic hybrids, along with an increase in ahigh molecular weight band which may correspond to thebetaglycan core (26). Furthermore,the LT-HYB showed reduced type I11 receptor but no enhancement of the type I1 receptor as seen in the nontumorigenic hybrids. Densitometric analysis of the autoradiograms was

+

Complementation of TGF-/3Growth Inhibitory

2590

200-

] Type Ill

97-

] Type I1

68-

-Type I

43-

29-

18-

FIG. 3. Receptor cross-linking to ‘251-TGF-f11.The indicated cells were cross-linked in duplicate and extracts were loaded onto adjacent lanes. Molecular weight markers are indicated on the left. The receptor types are indicated on the right showing the type I receptor, a doublet for the type I1 receptor, and two bands which correspond to the type 111 (betaglycan) receptor in which the lower band is thoughtto be the unglycosylatedcore protein. Free monomeric “‘I-TGF-@l can be seen at the bottom of the gel.

TABLE I Relative expression of TGF-@receptor types Densitometric measurements were performed on three independent receptor cross-linking gels. The first set of data was measured from the gel shown in Fig. 2. The values allow comparison of the intensities of the individual receptor types between cells and are normalized to theintensitv of the SW480 receDtors. SW480 E3 LT-HYB HYB-1 HYB-2 HYB-3 Type I 1.00 0.56 0.98 1.60 3.83 3.62 Type I1 1.00 1.06 0.13 21.52 23.66 15.89 Type 111 1.00 0.49 0.05 0.23 0.19 0.26 Type I 1.00 0.47 0.83 2.01 3.18 2.93 Type I1 1.00 4.16 47.44 3.31 42.49 28.16 0.22 0.26 0.23 Type 111 1.00 0.27 0.23 Type I 1.00 0.92 0.97 0.11 6.35 1.51 Type I1 1.00 0.20 0.78 9.38 8.54 20.84 Type 111 1.00 0.21 0.16 0.32 0.48 0.44 TABLE I1 Gene activation by TGF-@1 Values represent the ratios of mRNA expression of the indicated genes in TGF-@1-treatedcompared with untreated cells. Densitometric analyses of the Northern blots shown in Fig. 4 were used to determine the relative values. The type IV collagenase bands for EJ and SW480 seen on long exposures were belowthe intensity required for densitometry. GAPDH, glyceraldehydephosphate dehydrogenase. Fibronectin PAI-1 Type IV Col. GAPDH 43.9 EJ 5.6 0.9 4.5 1.7

SW480 2.8 LT-HYB 5.5 HYB-2 HYB-3 5.2

1.6 2.2 2.7 6.4

1.3 16.3

2.7

1.3 0.5

used to quantify the relative shift in receptor distribution from predominantly type I11 receptor in the parental cells to type I and type I1 receptors in the nontumorigenic hybrids (Table I). While differencesin the relative intensities of crosslinked bands were apparent between different experiments, the above observations held true in each experiment. Thus, the differential responses of these cells may lie the in receptor complement that they express. A large increase in the expression of the type I1 receptor (10-20-fold) and a smaller increase

Responses

in the type I receptor correlated with the hybrid cells that were strongly inhibited for DNA synthesis by TGF-P. These findings are compatible with those of others who have shown involvement of type I and type I1 receptors in growth responses to TGF-/3 (25, 27). Gene Expression in Response to TGF-PI-TGF-P affects the expression of many genes, several of which are involved in the accumulation and maintenance of the extracellular matrix (28-30). We wanted to determine if regulation of gene expression by TGF-P could occur in the absence of growth regulation of the two parental tumor lines and to see how regulation might be modified in the hybrid cells. Each cell line and hybrid cell clone was treated for 24 h with a high concentration (200 PM) of TGF-Dl. RNA was then extracted from both treated and untreated cells and used in Northern blot analysis. As seen in Fig. 4 and Table 11, the parental as well as the hybrid cells expressed significantly higher levels of fibronectin message after treatment with TGF-Dl. The amount of fibronectin mRNA was foundto be 50-fold higher in the hybrid cells compared with the tumorigenic parental cells and produced even higher levels upon stimulation by TGF-6. While SW480 cells showed onlya slight increase (of questionable significance) in fibronectin message levels, the EJ cell line was stimulated by TGF-81 to the same degree as the hybrid cells. Other genes known to be regulated by TGF-/3 in certain cells were also examined for changes in expression by TGF0. Plasminogen activator inhibitor-1 (PAI-1) (30,31)gene expression was strongly induced by TGF-P1 in both the parental and hybrid cells. Again, the SW480 cells responded only slightly, while the EJ line and each hybrid produced much higher PAI-1 message levels in response to TGF-Pl. Although type I collagenase expression has generally been reported to be inhibited by TGF-P, expressionof the type IV collagenase (72-kDa progelatinase) gene has been shown to increase in response to TGF-P in both fibroblasts and melanoma cells (21, 32). Expression of type IV collagenase was both much higherand inducible in allof the hybrids compared EJ rGF-01 I -

SW480 LT-HYB HYB 2 HYB 3 ‘I-

+ ‘I -

+‘I-

+ ‘I- +‘

8kb-

k

32kb-

I

2.3kb-

3.1kb1.4kb -

FIG.4. Enhanced gene expression in response to TGF-Bl. Probes used for hybridization and thesize of the transcripts are (from the top to thebottom panel);rat fibronectin (8.0-kb), human plasminogen activator inhibitor type 1 (PAI-1) (3.2 and 2.3 kb), human type IV (72 kDa) collagenase (3.1 kb), and rat glyceraldehyde phosphate dehydrogenase (1.4 kb). The lanes marked + were RNA from cells treated with 5.0 ng/ml TGF-@lfor 24 h, and the - lanes were from untreated cells. Glyceraldehyde phosphate dehydrogenase was used as a control for the amount of RNA loaded onto the gel. For the fibronectin hybridization, EJ and SW480 lanes were exposed for 24 h, while the hybrids were exposed for only 25 min. Exposure times for PAI-1, type IV collagenase, and glyceraldehyde phosphate dehydrogenase were 2,3, and 6 h, respectively, at room temperature. The fibronectin and collagenase probes were hybridized to the same Nytran filter, while the PAI-1 and glyceraldehyde phosphate dehydrogenase probes were hybridized to a second Nytran filter. Both agarose gels were loaded with equal quantities (12 pg) of RNA.

Complementation of TGF-p Growth Inhibitory Responses

2591

with the parental cells. Only after longer exposures to film (data not shown)could any expression of type IV collagenase be seen in the SW480cells, although no inductionby TGF-P was seen. Even such long exposures did not detectexpression of type IV collagenase in untreated EJ cells. However, a low EJ cells treated with level expression was seen in mRNA from TGF-0. As wewere initially surprised to find the IV type collagenase gene activated to such a high degree in the hybrid cells, we also examined expression at the protein level by measuring type IV collagenase activity. Collagenase activity was measured by zymographic analysis (33) from conditioned medium taken from the parental and hybrid cells either untreated or treated with 200 PM TGF-P2. As shown in Fig. 5A, both of the hybrids produced high levels of type IV collagenase activity, whereas the parentalcells producedno detectableactivity. Treatment with TGF-P2 increased the gelatinolytic(collagenase) activity in the hybrids (Fig. 5 B ) . Interestingly, a band representing the92-kDa form of type IV collagenase (34) was not detectedfrom untreated SW480cell conditioned medium but was clearly detectable after treatment with TGF-P (Fig. 5 B ) . Regulation of 92-kDa type IV collagenase activity by TGF-P indicates the ability of SW480 to respond to TGF-P. Additionalgenes that have been shown by others to be transcriptionally stimulated by TGF-P were also tested (data not shown). Induction by TGF-/3 of its own message (35,36) was seen in thehybrid cells and in EJ but the induction was only 2-fold. Other genes were examined but did not show alterations in transcriptlevels in response to TGF-Dl. These include the TIMP I and TIMP I1 genes (inhibitors of type I and type IV collagenase activity, respectively) (32, 37) and the c-myc and TGF-P2genes, both of which have been shown t o be down-regulated by TGF-P (38-40). Most importantly, these datashow increased expressionof selected genes in both the hybrid and parental cells by TGFP, demonstrating that defects in the pathway(s) leading to regulation of growth by TGF-P need not interfere with pathways leading to gene regulation by TGF-8.

type I1 receptor, however, were not required for activation of gene expression by TGF-6. These studies lend support to the theory carcinogenesis that is a stepwiseprocess that involves thesequential loss of normal regulatory mechanisms, along with the activation of separate growth-promoting mechanisms (14). Using cell fusion, we have demonstrated the restoration of one growth regulatory pathway involving TGF-P. Certainly, the complementation of other genes involved in growth regulation may have contributed to the nontumorigenic phenotype of the hybrid cells. Indeed, these same hybrids respond to retinoic acidwhile the parental tumor cells do not.' However, the possibility that TGF-/3 participates in control of growth and that loss of such regulation may be a contributing factor to carcinogenesis is supported by our findings. Several studies have demonstrated the loss of sensitivity to TGF-8 in transformed or tumorigenic epithelial cells. For example, rat liver epithelial cells are normally inhibited by TGF-P, but transformation by carcinogens (2) or oncogenes (41, 42) can render the cells unresponsive to TGF-8. Transformation of the rat liver cells somehow altered the growth inhibitory signaling of TGF-p. There is also evidence that TGF-P can still inhibit the growth of colon carcinoma cells if they are from well differentiated tumors. However, as the cells develop into poorly differentiated, aggressive cells, they lose their sensitivity to inhibition by TGF-P (39, 43). Thus, the progression from a normal to a malignant phenotype can often beassociated with the loss of sensitivity to TGF-P, althoughotherfindingsindicatethatthisisfar from an absolute correlation (10,44). Our data suggest that a decrease in expression of the type a lesser degree, the typeI receptor) I1 TGF-P receptor (and to may have been the mechanism by which the tumor cell lines losttheirabilityto respond to TGF-/3 growthinhibitory signals. Fusion of the tumorcell lines resulted in greater type I1 receptor expression indicating complementationof regulatory mechanisms that allowed increased expression of functional type I1 receptors. Similar receptorcomplementation results were found by Laiho et al. (45) who found that fusion DISCUSSION of various type I and typeI1 receptor mutantsof CCL64 cells of recepIn this study, we report the reestablishment of regulation restored normal functionalexpression for both types of growth by TGF-P. of growth by TGF-P in hybrid cells following the fusion of tor and restored the inhibition The cross-linking studies presented here indicate that a two carcinoma cell lines that were not inhibited by TGF-B. The results suggest that thetwo cell lines have complementing critical threshold of functional type I1 receptors may be neclesions in the growth inhibitory response to TGF-P. Further, essary for signaling inhibitionof growth. Should thereceptor the type I1 (85 kDa) receptor maybe involved in the restored number fall below this threshold, other responses to TGF-P at much such as transcriptionalregulation could still occur but growth regulation of growth, as this receptor was expressed higher levels in thehybrids compared with the nonrespondingresponses would be lost. However, the tumorigenichybrid tumorigenic parental cells. The higher expression levels of (LT-HYB) was stillcapable of responding weakly to growthinhibitory signals from TGF-/3 even though type I1 receptor levels were low. This might indicate some post-receptor signaling advantage of the hybrid cells which allows the same low level of type I and type I1 receptors as expressed in the parental tumor lines to more effectively tranduce the weak receptor signal. An alternative possibility is that thereduced type I11 receptor level in the LT-HYBallows more TGF-/3 to bind to the signal-transducing type I and type I1 receptors. As mutational studieshave indicatedthat the type I11 receptor is not involved in TGF-P signal transduction (see below), it is possible that the typeI11 receptor plays a negative role by FIG.5. Zymographic analysis of 72-kDa type IV collagen- sequestering available TGF-/3 away from the signaling recepase activity. Aliquots of conditioned medium were electrophoresed tors. on gelatin-containing polyacrylamide gels with SDS. Theconditioned Increasing evidence suggests that the typeI and/or type I1 medium was removed from cells following a 48-h serum-free incubareceptors are responsible for transducing the growth regulation in the presence (panel B)or absence (panel A ) of 200 pM TGF82. The major gelatinolytic activity was from the 72-kDa type IV collagenase form, while a minor activity can be seenat 92 kDa, corresponding to a different form of type IV collagenase.

* K. J. Busam, A. G . Geiser, A. B. Roberts, and M. B. Sporn, manuscript in preparation.

2592

Complementation of TGF-/3 Growth Inhibitory Responses

tory signals of TGF-P. Cross-linking studies based on the highly TGF-@-responsive mink lung epithelial cell line, MvlLu (CCL64) showed that stable mutantsresistantto inhibition of growth by TGF-P were deficient in the type I or type I1 receptors but unchanged in the type I11 receptors (25, 46). These mutants were grouped by phenotype according to the reduction or complete absence of the type I receptor alone, or the reduction/absence of type I together with the type I1 receptors (25). In no cases, however, did they find alterations in type I1 receptors independent of alterations in typeI receptors. Furthermore, they found that these receptor mutants were defective in the ability to respond to TGF-p with elevated fibronectin or plasminogen activatorinhibitor-1 expression or by promoting cell flattening (45, 46). Thus, it was concluded that the type I and type I1 receptors are coordinately expressed and that both are necessary for all of the responses to TGF-P. In contrast, our findings with the hybrid cells described here indicate that expression of the type I1 receptor was independent of type I expression and that inhibition of growth and gene activation by TGF-p can be independentprocesses. Other findings have shown that the type I1 receptor is important in mediating suppression of growth by TGF-8. One study investigated the mechanism by which basic fibroblast growth factor decreased the ability of TGF-P to inhibit the growth of bovine aortic endothelial cells. It was found that basic fibroblast growth factor selectively decreased type I1 receptor expression (47). In addition, a recent comparison of growth response and receptor profiles of TGF-P was made with murine and human embryonic palate mesenchymal cells (48). While the human cells were slightly stimulated by TGFP, the murine cells were significantly inhibited. TGF-p receptor profile analysis showed the presence of an 86-kDa (type 11) receptor in the mouse cells but not the human cells. Finally, bovine endothelial cells were chemically mutagenized from a TGF-@-inhibited to TGF-@-resistant a phenotype (49). In this case, a glycosylation defect was found that altered the electrophoretic mobility of the type I1 but not the types I or I11 receptors. Although the growth inhibition response to TGF-P was completely lost, the activation of fibronectin and plasminogen activator inhibitor-1by TGF-/3still occurred, similar to our findings. The distinction between the inhibition of growth and regulation of certain responsive genes by TGF-P could be made by comparing responses of the tumor cells with those of the hybrid cells. The EJ tumor cell line was unaffected by TGFP in terms of DNA synthesis, yet demonstratedlarge increases in fibronectin andPAI-1 mRNA production. The SW480 tumor line showed only marginal increases in transcription of sensitive genes when treated with TGF-p. However, SW480 cells expressed a 92-kDa form of the type IV collagenase upon treatment with TGF-P. Interestingly, none of the other cells expressed higher levels of 92-kDa collagenase after TGF-P treatment but the hybrid cells did increase expression of the 72-kDa form of type IV collagenase. It is possible that increased type IV collagenase expression in response to TGF-P could have a role in matrix remodeling under conditions that involve wound healing or repair. The nontumorigenic hybrids were responsive to TGF-/3 in all ways tested including inhibition of DNA synthesis, while the tumorigenic hybrid (LTHYB) was only marginally affected in terms of inhibition of DNA synthesis yet was very responsive in gene activation. Thus, TGF-@signaling can involve different pathways which allow some responses (gene activation)in the absence of others (inhibition of growth) and may depend upon the number of functional receptors available to initiate thesignals.

Acknowledgments-We thank William Stetler-Stevenson, David Kliner, and Peter Brown for the zymographic analysis and type IV collagenase cDNA, Peter Andreasen for the PAI-1 cDNA, and Michael O’Reilly and Klaus Busam for helpful suggestions. REFERENCES 1. Roberts, A. B., and Sporn, M. B. (1990) Peptide Growth Factors and their Receptors. Handbook of Experimental Pharmacology (Sporn, M. B., and Roberts, A. B., eds) Vol. 95, pp. 419-472, Springer-Verlag, Heidelberg 2. McMahon, J. B., Richards, W. L., del Campo, A. A., Song, M.-K., and Thorgeirsson, S. S. (1986) Cancer Res. 46, 46654671 3. Shipley, G. D., Pittelkow, M. R., Wille, J. J., Scott, R. E., and Moses, H. L. (1986) Cancer Res. 46, 2068-2071 4. Silberstein, G . B., and Daniel, C.W. (1987) Science 237, 291293 5. Frater-Schroder, M., Muller, G., Birchmeier, W., and Bohlen, P. (1986) Biophys. Res. Commun. 137,295-302 6. Kehrl, J. H., Roberts, A. B., Wakefield, L. M., Jakowlew, S. B., Sporn, M. B., and Fauci, A. S. (1986) J. Immunol. 137, 38553860 7. Wahl, S. M., Hunt, D.A., Wong, H. L., Dougherty, S., McCartney-Francis, N., Wahl, L. M., Ellingsworth, L., Schmidt, J. A., Hall, G., Roberts, A. B., and Sporn, M. B. (1988) J. Immunol. 140,3026-3032 8. Shipley, G. D., Tucker, R. F., and Moses, H. L. (1985) Proc. Natl. Acad. Sci. U. S. A . 82,4147-4151 9. Robey, P. G., Young, M. F., Flanders, K. C., Roche, N. S., Kondaiah, P., Reddi, A. H., Termine, J. D., Sporn, M. B., and Roberts, A. B. (1987) J. Cell Bwl. 105,457-463 10. Wakefield, L. M., and Sporn, M. B. (1990) in Tumor Suppressor Genes (Klein, G., ed) pp. 217-243, Marcel Dekker, Inc., New York 11. Arteaga, C. L., Tandon, A. K., Von Hoff, D. D., and Osborne, C. K. (1988) Cancer Res. 48,3898 12. Kimchi, A., Wang, X.-F., Weinberg, R. A., Cheifetz, S., and Massaguk, J. (1988) Science 240, 196-240 13. Massagu6, J., Boyd, F. T., Andres, J. L., and Cheifetz, S. (1990) Ann. N. Y. Acad. Sci. 593, 59-72 14. Stanbridge, E. J. (1990) Annu. Rev. Genet. 24,615-657 15. Geiser, A. G., and Stanbridge, E. J. (1989) CRC Crit. Reo. Oncogenesis 1, 261-276 16. Geiser, A. G., Anderson, M. J., and Stanbridge, E. J. (1989) Cancer Res. 49,1572-1577 17. Marshall, C. J., Franks, L. M., and Carbonell, A. W. (1977) J. Natl. Cancer Inst. 58,1743-1747 18. Leibovitz, A. (1976) Cancer Res. 36,4562-4569 19. Maniatis, T. E., Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual,Cold Spring HarborLaboratory, Cold Spring Harbor, NY 20. Church, G. M., and Gilbert, W. (1984) Proc. Natl.Acad.Sci. U. S. A . 81, 1991-1995 21. Brown, P. D., Levy, A. T., Margulies, I. M. K., Liotta, L. A., and Stetler-Stevenson, W. G. (1990) Cancer Res. 50, 6184-6191 22. Massaguk, J., and Like, B. (1985) J. Biol. Chem. 260,2636-2645 23. Cheifetz, S., Like, B., and Massague, J. (1986) J. Biol. Chem. 261,9972-9978 24. Cheifetz, S., Weatherbee, J. A., Tsang, M. L.-S., Anderson, J. K., Mole, J. E., Lucas, R., and Massague, J. (1987) Cell 48, 409415 25. Laiho, M., Weis, F. M. B., and Massaguk, J. (1990)J. Biol. Chem. 265,18518-18524 26. Cheifetz, S., Andres, J. L., and MassaguC.,J. (1988) J. Bid. Chem. 263,16984-16991 27. Segarini, P. R., Rosen, D. M., and Seyedin, S. M. (1989) Mol. Endocrinol. 3, 261-272 28. Ignotz, R. A., and Massague, J. (1986) J. Biol. Chem. 261,43374345 29. Rossi, P., Karsenty, G., Roberts, A. B., Roche, N. S., Sporn, M. B., and de Crombrugghe, B. (1988) Cell 52,405-414 30. Laiho, M., Saksela, O., Andreasen, P. A., and Keski-Oja, J. (1986) J . Cell B i d . 1 0 3 , 2403-2410 31. Lund, L. R., Riccio, A,, Andreasen, P. A., Nielson, L. S., Kristensen, P., Laiho, M., Saksela, O., Blasi, F., and Dano, K. (1987) EMBO J. 6, 1281-1286

Complementation of TGF-0 Growth Inhibitory Responses 32. Overall, C.M., Wrana, J. L., and Sodek, J. (1989) J. Biol. Chem. 264,1860-1869 33. Heussen, C., and Dowdle, E. B. (1980) Anal. Biochem. 102,196202 34. Wilhelm, S. M., Collier, I. E., Marmer, B. L., Eisen, A. Z., Grant, G. A., and Goldberg, G . I. (1989) J . Biol. Chem. 2 6 4 , 1721317221 35. Van Obberghen-Schilling, E., Roche, N. S., Flanders, K. C., Sporn, M. B., and Roberts, A. B. (1988) J. Biol. Chem. 263, 7741-7746 36. Kim, S.-J., Angel, P., Lafyatis, R., Hattori, K., Kim, K. Y., Sporn, M. B., Karin, M., and Roberts, A. B. (1990) Mol. Cell. Biol. 10, 1492-1497 37. Stetler-Stevenson, W. G., Brown, P. D., Onisto, M., Levy, A. T., and Liotta, L. A. (1990) J . Biol. Chem. 2 6 5 , 13933-13938 38. Coffey,R. J., Bascom, C. C., Sipes, N. J., Graves-Deal, R., Weissman, B. E., and Moses, H. L. (1988) Mol. Cell. Biol. 8, 3088-3093 39. Mulder, K. M.,Zhong,Q., Choi, H. G., Humphrey, L. E., and Brattain, M. G. (1990) Cancer Res. 5 0 , 7581-7586 40. Bascom, C. C.,Wolfshohl, J. R., Coffey, R.J., Madisen, L., Webb,

41. 42. 43. 44. 45. 46. 47. 48. 49.

2593

N. R., Purchio, A. R., Derynck, R., and Moses,H. L. (1989) Mol. Cell. Biol. 9 , 550&5515 Houck, K. A., Michalopoulos, G. K., and Strom, S. C. (1989) Oncogene 4, 19-25 Huggett, A. C., Hampton, L. L., Ford, C. P., Wirth, P. J., and Thorgeirsson, S. S. (1990) Cancer Res. 5 0 , 7468-7475 Hoosein, N. M., Levine, A. E., Mulder, K. M., Childress, K. E., Brattain, D. E., and Brattain,M. G. (1989) Exp. Cell Res. 181, 442-453 Wollenberg, G. K., Semple, E., Quinn, A., B. and Hayes, M. A. (1987) Cancer Res. 47,6595-6599 Laiho, M., Weis, F. M.B.,Boyd, F. T., Ignotz, R. A., and Massague, J . (1991) J. Biol. Chem. 266,9108-9112 Boyd, F. T.,and Massagub, J. (1989) J. Biol. Chem. 2 6 4 , 22722278 Fafeur, V., Terman, B. I., Blum, J., and Bohlen, P. (1990) Growth Factors 3,237-245 Linask, K. K., D’Angelo, M., Gehris, A. L., and Greene, R. M. (1991) Exp. Cell Res. 1 9 2 , 1-9 Fafeur, V., O’Hara, B., and Bohlen, P. (1991) J. Cell. Biochem. 15B. 172