Zeta PKC plays a critical role during stromelysin promoter activation by platelet-derived growth factor through a novel palindromic element.

Stromelysin is a metalloproteinase with the widest substrate specificity that plays a critical role in the induction of the metastatic phenotype in cancer cells. The mechanisms whereby growth factors and oncogenes control stromelysin expression are beginning to be characterized. We have recently demonstrated that protein kinase C isotypes down-regulatable by chronic exposure to phorbol esters are not involved in stromelysin gene expression in response to platelet-derived growth factor, ras oncogene, and phosphatidylcholine-hydrolyzing phospholipase C. We also identified a region in the stromelysin promoter, distinct from the 12-O-tetradecanoylphorbol-13-acetate-responsive element, responsible for the promoter activity in response to these stimulants. In this paper, we further characterize that promoter fragment and demonstrate that the region encompassing nucleotides -1218 to -1202, including the palindromic sequence ACTAGT, is necessary and sufficient for the control of stromelysin gene expression. The involvement of zeta-protein kinase C but not of c-raf in the stimulation of stromelysin promoter activity in response to platelet-derived growth factor is also demonstrated here. All these data suggest the existence of a bifurcation downstream of ras in the signaling mechanisms leading to stromelysin expression and DNA synthesis.

Recently, we and others have identified a novel required step in these cascades; activation by growth factors and oncogene products of a phospholipase C specific for phosphatidylcholine (PC-PLC) has been shown to be both necessary and sufficient for mitogenic activation (20)(21)(22)(23)(24)(25)(26)(27)(28). Furthermore, we recently found that PC-PLC activation, besides governing cell growth and tumor transformation, is also involved in the regulation of stromelysin and, therefore, could be implicated in the appearance of metastatic phenotypes (29). In that study it was also demonstrated, a t a gene transcriptional level, that the signaling mechanisms utilized by PDGF'lras p21PC-PLC differ from those triggered by P W K C . Thus, PDGF, ras p21, and PC-PLC activate stromelysin expression in fibroblasts lacking PMA-sensitive PKCs to an extent similar to that found in cells with normal PKC levels (23,29). Furthermore, this route does not involve the 12-0-tetradecanoylphorbol-13-acetate-responsive element existing in the stromelysin promoter but utilizes a potentially novel element located in the region encompassing nucleotides -1240 to -1145 of that promoter (29). These results would be consistent with our recent data demonstrating the critical role played by a PMA-unsensitive PKC isotype, termed (, in the mitogenic signaling pathways activated by ras p21/ In the study presented here we further characterize that promoter region and demonstrate that a fragment located between nucleotides -1218 and -1202, which includes the palindromic sequence ACTAGT, is necessary and sufficient for the control of stromelysin gene expression. Moreover, following PDGF stimulation of fibroblasts, the induction of a factor binding to this region was detected by mobility shift assays. Even more important, we demonstrate here the involvement of (PKC in the stimulation of stromelysin promoter activity in response to PDGF.

MATERIALS AND METHODS
Cell Cultures-NIH-3T3 fibroblasts were cultured and maintained as described (29) in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum, penicillin (100 unitslml), streptomycin (100 pg/ml), and 2 m~ L-glutamine in standard tissue culture flasks in a humidified air/CO, (19:l) incubator at 37 "C. Cells were made quiescent by incubation for 24 h in the presence of serum-free medium supplemented with transferrin (5 pg/ml) and Na,SeO, (1 p~) .
Gel Shift Mobility Assays-Nuclear extracts were isolated as previously described (32). Cells at 75% confluence in 100-mm diameter culture dishes were made quiescent by serum starvation for 24 h, after which they were either untreated or stimulated with PDGF (10 ng/ml) for 24 h. Afterward, cells were washed with ice-cold phosphate-buffered saline and scraped into 1 ml of the same buffer containing 0.5 mM DTT, 0.5 m~ phenylmethylsulfonyl fluoride, 2 p~ leupeptin, 5 m~ NaF, and 1 m~ Na3V0,. Cells were sedimented by centrifugation a t 500 x g for 5 min at 4 "C, resuspended with 1 ml of lysis buffer (10 m~ Tris, pH 7.9, 1.5 m~ MgCl,, 10 mM KCl, 0.5 m~ D m , 0.5 mM phenylmethylsulfonyl fluoride, 2 p~ leupeptin, 5 mM NaF, 1 m~ Na,VO,), and allowed to swell on ice for 10 min. Afterward, cells were homogenized with 10 strokes of a glass Dounce homogenizer, and the nuclei were collected by centrifugation at 100 x g for 5 min at 4 "C. The nuclei were resuspended in 0.7 ml of extraction buffer (20 mM Tris, pH 7.9,20% glycerol, 1.5 m~ MgCl,, 500 mM KC1, 0.5 m~ DTT, 0.5 m~ phenylmethylsulfonyl fluoride, 2 p~ leupeptin, 5 m~ NaF, 1 m~ Na3V0,), rotated gently at 4 "C for 45 min, and centrifuged at 15,000 x g for 15 min. The supernatant was dialyzed against dialysis buffer (20 mM Tris, pH 7.9,2% glycerol, 100 rn KCl, 0.2 m~ EDTA, 0.5 m~ DTT, 0.5 m~ phenylmethylsulfonyl fluoride, 0.5 m~ NaF, 0.5 m~ Na,VO,) at 4 "C. The dialysate was cleared by centrifugation a t 15,000 x g for 15 min, and the supernatant was aliquoted and stored a t -70 "C. Ten pg of nuclear extracts were preincubated with or without unlabeled competitors for 15 min at room temperature and incubated a t room temperature for an additional 30 min with 1 ng of the different ,'P-labeled probes (50,000 cpm) in the following gel shift binding buffer (5 x): 37.5% glycerol, 5 mM MgCl,, 0.25 m~ EDTA, 2.5 mM Dl", 175 mM NaI, 37.5 mM Hepes pH 8.0,4.0 pg of poly(dI.dC), and 3 pg of bovine serum albumin. The nuclear protein complexes were separated on a nondenaturing Tris borate/EDTA 5% polyacrylamide gel at 150 V for 3 h.
Dansfection a n d Gene Expression Assays-NIH-3T3 fibroblasts were seeded a t a density of 7 x lo5 cells/lOO-rnm dish and maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Twenty-four h later, the medium was changed and cells were transfected following a 4-h preincubation period with 10 pg of total DNA (5 pg of plasmid and 5 pg of carrier calf thymus) in calcium phosphate for 16 h (32). f i r w a r d , medium was changed, and cells were incubated for another 24 h in quiescent medium, after which different stimuli were added according to the experiments, and chloramphenicol acetyltransferase assays were performed on cell extracts as described previously (32).

RESULTS
Transient Gene Expression Assays-Previously published results demonstrated that the region encompassing nucleotides -1303 to -754 of the stromelysin promoter is necessary and sufficient for transcriptional activation in response to PDGF and PC-PLC (29). Further deletion analysis of this region revealed that the fragment from -1240 to -1145 is necessary to drive the transcription of a reporter gene linked to a minimal promoter (29). In order to further determine the element responsible for the activation of the stromelysin promoter by PDGF, the following plasmids were constructed. pA63HACAT harbors the fragment extending from -1240 to -754 of the stromelysin promoter cloned upstream of an SV40 minimal promoter (plasmid pCATprom). pA78HACAT, pA89HACAT, and pA102HACAT are further 5' deletions of pA63HACAT (see Fig. 1). Results from Fig. 2 demonstrate that the deletion of the region extending from -1214 to -1201, respective to the transcription initiation site, completely abolishes the stromelysin promoter activity induced by PDGF. There is a palindromic sequence in this region (ACTAGT) that we hypothesized could be a critical element responsible for stromelysin promoter activation. Plasmid palCAT was constructed in which a single copy of this element (nucleotides -1218 to -1202) was cloned upstream of the herpes simplex virus TK minimal promoter in a CAT reporter plasmid. Interestingly, results from Fig. 2 demonstrate that this fragment was sufficient to drive CAT synthesis in response to PDGF even more efficiently than the fragment from -1240 to -754 (plasmid pA63HACAT). This is a novel sequence with no obvious similarity to any other enhancer element so far described; from now on, this nucleotide sequence will be termed SPRE for gtromelysin PDGF-responsive element.
PDGF-stimulated Fibroblasts Contain Elevated Levels of a Specific SPRE Binding Nuclear Factor-The possible existence of DNA-binding proteins interacting with the SPRE was investigated by gel shift mobility assays. A double-stranded oligonucleotide (WT; nucleotides -1221 to -1203) containing the SPRE sequence was labeled and incubated with nuclear extracts from cells either unstimulated or treated with PDGF for 24 h. Nucleoprotein complexes were resolved on native poly- acrylamide gels, and the appearance of retarded bands was detected following autoradiography. Results from Fig. 3 demonstrate that cells treated with PDGF displayed a major band that was not seen in nuclear extracts from untreated cells (Fig.  3, compare lanes 2 and 3 ) . The formation of this DNA-binding complex was abolished by incubation of extracts with an excess of cold oligonucleotide (Fig. 3, lane 4 ) . These results suggest that PDGF induces a nuclear factor that binds to SPRE.
To further demonstrate the critical role played by the palindromic sequence in the binding of the putative nuclear factor, the following series of gel shift mobility assay experiments was carried out. Nuclear extracts from PDGF-treated cells were incubated with an increasing molar excess of either the wildtype double-stranded oligonucleotides ( W T ) or a deletion mutant (DEL) that lacks the palindromic sequence ACTAGT (see Fig. 4, upper panel) previous to the addition of the labeled WT probe. Results from Fig. 4 demonstrate that the integrity of the palindromic sequence is a requisite for the binding of the nuclear factor to take place. In addition, two mutations that disrupt the palindrome were performed, M U T l and MUT2 (see Fig. 5, upper panel ). Gel shift mobility assay experiments carried out with the corresponding 32P-labeled double-stranded oligonucleotides harboring either mutation and nuclear extracts from PDGF-treated cells demonstrated that both mutations severely impaired binding to the probe; this effect is even more dramatic in the case of MUTl. It is noteworthy that the intensity of the band given by MUT2 when incubated with 10 pg of protein nuclear extracts was similar to that produced by 2 pg of protein nuclear extract incubated with the labeled WT probe.
Involvement of ipKC in the Regulation of SPRE Enhancer Activity-Stromelysin induction is activated by serumPDGF and PC hydrolysis (29), and these stimulants utilize P K C (30-33) during mitogenic signaling. Therefore, conceivably, this kinase could be involved in the regulation of SPRE by PDGF. To address this possibility, we initially determined whether activation by overexpression of (PKC leads to the stimulation of SPRE binding activity. Results from Fig. 6 demonstrate that nuclear extracts from (PKC overexpresser cells (previously described in Refs. 32 and 33) gave dramatically increased levels of SPRE binding activity. The retarded band observed in gel shift mobility assay of nuclear extracts from (PKC overexpressers is most probably the same as that produced by stimulation with PDGF. In this regard, it is noteworthy that the affinity of the SPRE binding activity from (PKC overexpresser cells is quite similar to that from PDGF-treated cells (Fig. 7). Thus, a similar competition curve was observed when nuclear extracts from either (PKC overexpressers or PDGF-treated cells were incubated with increasing concentrations of cold probe prior to the addition of the 32P-labeled WT double-stranded oligonucleotide (Fig. 7). Likewise, binding of the nuclear factor from the (PKC overexpresser to the labeled WT probe is not competed by an excess (40-fold) of deletion mutant double-stranded oligonucleotide and does not bind either to MUTl or MUT2 probes (not shown). This behavior is identical to that of nuclear extracts  1 and 2 ) , MUT2 (lanes 3 and 41, and WT (lanes 5-7).
Essentially identical results were obtained in another three experiments. from PDGF-treated cells (see above). These results are consistent with the notion that {PKC is involved in the regulation of stromelysin expression through SPRE. To further substantiate this notion, we determined whether transfection of a plasmid expressing a kinase-defective {PKC dominant negative mutant (pRcCMVF'; described in Refs. 32 and 33) produced any effect on the SPRE-dependent promoter activity in response to PDGF. Results from Fig. 8A clearly demonstrate that expression of {PKC"' "' severely inhibited induction of CAT activity by PDGF in cell cultures transfected with palCAT (Fig. 8A). c-ruf is another kinase that has been proposed to play a key role in mitogenic signaling by growth factors (35). In order to determine whether or not c-ruf is involved in the regulation of SPREdependent promoter activity, a dominant negative mutant of this kinase was transfected along with the palCAT reporter plasmid. R A F " ' " ' is a previously described c-ruf kinase-defective dominant negative mutant (35). Expression plasmids harboring both kinase mutants (i.e. [PKCmUt, and RAF""') severely impaired cell proliferation when separately transfected into NIH-3T3 fibroblasts (not shown). Interestingly, transfection of a plasmid expressing did not significantly inhibit the PDGF-stimulated SPRE-dependent promoter activity (Fig.   a). Transfection of the corresponding empty plasmids did not have any effect on this parameter (not shown). Consistent with the notion that SPRE plays a critical role in the regulation of the stromelysin promoter is the fact that transfection of CPKCmut but not of WmUt significantly inhibits the stromelysin promoter activity when the pHBCAT reporter plasmid is used instead of palCAT in the transfection assays (Fig. 8B). DISCUSSION We have previously demonstrated that PC-PLC activation is an important event for stromelysin induction and in the control of PDGF-triggered mitogenic signaling (20)(21)(22)(23)(24)(25)(26)(27)(28)(29). Since PC hydrolysis generates diacylglycerol, a logical hypothesis should contemplate PKC as an important intermediary in the activation of stromelysin expression by PDGFPC-PLC. However, we have already presented evidence that PMA-sensitive PKC subspecies are not required for stromelysin induction in response to these stimuli (which is in very good agreement with the notion that this route is not necessary either for activation of DNA synthesis in Swiss 3T3 fibroblasts (23) or for maturation of Xenopus oocytes (30)). Transient expression experiments with plasmids harboring different deletions and mutations in the stromelysin promoter region linked to a reporter gene, demonstrated that the 12-0-tetradecanoylphorbol-13-acetate-responsive element located in that promoter is not required to transmit signals generated by PDGFPC-PLC (29). We defined a region of 750 base pairs located between -1303 and -754, respective to the transcription start site, that is necessary and sufficient for induction of stromelysin by PDGFPC-PLC.
A careful search in that region reveals the existence of consensus sequences similar or identical to known enhancer elements. Thus, at base pair -1290, there is a sequence very similar to one termed the rus-responsive element that apparently accounts for the inducibility of certain genes by rus (36). However, a deletion in the stromelysin promoter that removes the 5' 24 base pairs including that element (29) does not affect the ability of PDGFPC-PLC to activate this promoter. That sequence, therefore, is not important for stromelysin gene induction in response to PDGFPC-PLC. Another potentially interesting sequence in this zone is the one located at nucleotide -1239 (CACCTG), identical to the consensus sequence CANNTG, which has been reported to bind Myc and/or MyoD proteins (37). However, from the results shown in Fig. 1 it is clear that deletion of such an element does not affect stromelysin promoter activity, revealing the lack of importance of that sequence in the induction of stromelysin. Interestingly, a novel region that includes a palindrome sequence was identified in the present study as necessary and sufficient for the activation of the stromelysin promoter in response to PDGF. From the data shown here it is also clear that a nuclear factor is induced by PDGF, which binds to the promoter fragment located between nucleotide positions -1221 and -1203 that includes the palindrome ACTAGT, which is critical for the binding of the nuclear complex. The nature and characterization of that factor is beyond the scope of this paper but clearly indicate the existence of either a completely novel pathway for the transcriptional regulation of stromelysin or, alternatively, classical transcription factors functioning through a novel element. There are precedents for the second possibility. Thus, for example, it has recently been shown the involvement of AP-1 family members as components of the nuclear factor of activated T cells, a critical nuclear factor binding to the interleukin-2 enhancer through an element distinct from the AP-1 binding site (38).
On the other hand, in order to better understand the signaling mechanisms involved in the regulation of stromelysin ex-pression, a series of experiments aimed at identifying critical kinases in this pathway was carried out. From our previously published data (29), it is clear that a classical PMA-sensitive PKC isotype is not involved in the activation of the stromelysin promoter activity. Interestingly, the results depicted here establish the critical role played by CPKC in the regulation of SPRE-dependent promoter activity in response to PDGF. This, together with recent data demonstrating the involvement of CPKC in the regulation of nuclear factor KB (31,321, permit one to consider CPKC as a central point in the mechanisms controlling the transcriptional machinery. Another interesting aspect of the results shown here is the fact that c-ruf does not appear to be involved in stromelysin gene expression. Both kinases, c-Raf and CPKC, have been proposed to lie downstream of rus in the mitogenic cascade (30)(31)(32)(33)35). Ras is a potent activator of stromelysin expression (29). Therefore, the results presented in this study suggest the bifurcation of signaling mechanisms downstream of Ras. One mechanism is controlled by CPKC, whereas the other one is controlled by c-ruf. Both are critical for cell proliferation, but only (PKC is involved in the control of the stromelysin promoter activity. If this model is correct, one could predict that both pathways should converge at some point in the mitogenic cascade. Actually, preliminary data from our laboratory' demonstrate that CPKC, like Raf, phosphorylates and activates mitogen-activated protein kinase kinase, which permits one to propose mitogen-activated protein kinase kinase as the meeting point for both pathways.