Protein Kinase C Isozymes Differentially Regulate Promoters Containing PEA-3/12- O -Tetradecanoylphorbol-13-acetate Response Element Motifs*

, To investigate the regulation of promoters containing classical phorbol ester response sequences (PEA-3/12- O -tetradecanoylphorbol-13-acetate response element mo- tifs) by protein kinase C (PKC) isozymes, co-transfec-tions were performed in human dermal fibroblasts with a plasmid containing either the human collagenase promoter or the porcine urokinase plasminogen activator (uPA) promoter linked to the chloramphenicol acetyltransferase gene and a plasmid expressing an individual PKC isozyme. Using this experimental design, seven PKC isozymes were analyzed for their ability to trans- activate the collagenase and uPA promoters. Our results demonstrate that only PKC (cid:100) , (cid:101) , and (cid:104) trans-activated the collagenase promoter and that binding of Ap-1 fam- ily members to the collagenase 12- O -tetradecanoylphor-bol-13-acetate response element (TRE) was not respon- sible for the isozyme-specific trans-activation. In contrast, the uPA promoter was stimulated by all of the PKC isozymes examined (PKC (cid:97) , (cid:98) II, (cid:103) , (cid:100) , (cid:101) , (cid:122) , and (cid:104) ). These results indicate that PKC isozymes differentially regulate promoters containing PEA-3/TRE motifs and suggest that individual isozymes PEA-3 and follows. thymidine kinase from herpes simplex type was isolated from pHSV-106 on a Pvu II to Bgl II fragment. This fragment blunted with the Klenow fragment of DNA polymerase I as described above and inserted at the Xba I site of pCAT-Basic (Promega Corp.) that had also been made blunt. A 71-bp oligonucleotide containing the collagenase PEA-3 and TRE (collagenase promoter sequences (cid:50) 111 to (cid:50) 40) was then inserted into the Sph I site creating pTRETK-CAT. Cell Line, Transfection, and cat Gene Analysis— Human dermal fibroblasts (HDF) purchased from Clonetics Corp. and

Protein kinase C (PKC) 1 is a serine/threonine kinase that was first characterized by its dependence on calcium, phospholipid, and diacylglycerol for in vitro activation (Nishizuka, 1992;Asaoka et al., 1992). PKC is also activated by phorbol esters and is considered the major phorbol ester receptor in the cell (Nishizuka, 1984). Once activated, PKC plays a key regulatory role in a variety of cellular functions such as stimulation or repression of growth, changes in morphology, and modulation of gene expression (reviewed in Nishizuka 1986;Nishizuka 1989). Extensive molecular cloning and biochemical studies have revealed that PKC is a family of related polypeptides. Eleven mammalian genes have been isolated and cloned that code for 12 different PKC isozymes identified by the following Greek symbols ␣, ␤I, ␤II, ␥, ␦, ⑀, , , , , , and (reviewed in Dekker and Parker, 1994).
Structurally, the PKC isozymes are heterogeneous and can be divided into four subfamilies based on their primary structure and activation requirements. The conventional isozymes (␣, ␤I, ␤II, and ␥) contain two constant domains, C1 and C2, within their regulatory element and require calcium for activation, whereas the novel isozymes (␦, ⑀, , and ) lack a C2 region in their regulatory domain and do not require calcium for activation. The atypical isozymes (, , and ) lack a C2 region and have only one cysteine-rich stretch in their C1 region. These isozymes do not require calcium, diacylglycerol, or phorbol esters for activation (Dekker and Parker, 1994;Liscovitch and Cantley, 1994). The fourth subfamily has only one member (PKC ). The unique feature of this isozyme is its long N-terminal region that may function as a transmembrane domain. Like the novel and atypical PKC isozymes, PKC lacks the C2 domain and does not require calcium for activation (Johannes et al., 1994;Johannes et al., 1995).
In addition to structural differences, the PKC isozymes also display functional heterogeneity due, in part, to differences in tissue distribution, subcellular localization, and substrate selectivity. For example, some of the isozymes are present in the majority of tissues investigated (␣, ␦, ) whereas others (␥, , and ) are expressed in a tissue-specific manner (Dekker and Parker, 1994;Hug and Sarre, 1993). Additionally, it has been reported that in resting cells PKC ␣, ␤, and are found in the cytosol; PKC is predominantly in the particulate fraction, and PKC is located in the nucleus. The subcellular location of PKC ␦ and ⑀ is cell type-specific (reviewed in Hug and Sarre, 1993). Furthermore, substrate selectivity of the PKC isozymes has been demonstrated. Goode et al. (1992) have shown that glycogen synthase kinase-3B (GSK-3B), an enzyme involved in the activation of c-Jun (a member of the Ap-1 transcription factor family), is a much better substrate for PKC ␣, ␤I, and ␥ than it is for PKC ␤II, whereas PKC ⑀ is unable to significantly phosphorylate GSK-3B. This isozyme hierarchy for substrate phosphorylation is seen with a number of PKC substrates including the epidermal growth factor receptor (Ido et al., 1987), the neuron-specific protein F1/GAP-43 (Sheu et al., 1990), and the vitamin D receptor (Hsieh et al., 1991). This high degree of structural and functional heterogeneity found among the isozymes suggests that each isozyme plays a unique role within the cell.
PKC has been shown to be associated with enhanced expression of both the collagenase and urokinase plasminogen activator (uPA) genes (Rajabi et al., 1992;Juarez et al., 1993;Sudbeck et al., 1994;Rondeau et al., 1990;Hamilton et al., 1991;Niedbala and Stein-Picarella, 1993;Dierks-Ventling et al., 1989;He et al., 1992). Due to the large heterogeneity that exists among the PKC isozymes as described above, it is pos-sible that the isozymes regulate the collagenase and uPA promoters differently. To test this idea, co-transfections were performed in human dermal fibroblasts (HDF) with a plasmid that expresses an individual PKC isozyme and a plasmid containing either the porcine uPA or human collagenase promoter linked to the chloramphenicol acetyltransferase (cat) gene. Our results indicate that these two promoters are regulated differently by the PKC isozymes and suggest a biological explanation for the heterogeneity that exists within the PKC family.

MATERIALS AND METHODS
Plasmids-The plasmid containing the human collagenase promoter linked to the cat gene, Ϫ1200/ϩ63 Cat (Angel et al., 1987a) was kindly provided by Hans Jobst Rahmsdorf. pCAT4660 (Cassady et al., 1991), a plasmid containing the porcine uPA promoter sequences linked to the cat gene, was a gift from David A. Hume. The expression vector pS-VHNX-neo was constructed from the parent plasmid pSV2HNXB that contains the ␤-lactamase gene and a prokaryotic origin of replication from pBR322 (pBR322 sequences 2064 -4361), the SV40 origin of replication and early promoter (SV40 sequences 270-5190), a 16-bp polylinker containing unique HindIII, NruI, XhoI, and BglII sites, the SV40 t antigen intervening sequence (SV40 sequences 4100 -4710), and the SV40 early region polyadenylation signal (SV40 sequences 2770 -1782). pSVHNX-neo was then constructed by the following steps. The neomycin resistance gene was isolated from pSV2-neo (Southern and Berg, 1982) on an NdeI to BamHI fragment, and the ends were made blunt with the Klenow fragment of DNA polymerase I plus dATP, dCTP, dGTP, and TTP. This fragment was then inserted in the unique BamHI site of pSV2HNXB that previously had been blunted as described above. An oligonucleotide containing an NheI site flanked by XhoI sites was then ligated into the unique XhoI site creating a unique NheI site. The individual PKC cDNAs were then ligated (either cohesive or blunt-ended) into the NheI site; isolation and cloning of these cDNAs have been described (Basta et al., 1992;Aris et al., 1993;Barbee et al., 1993;Hocevar et al., 1993;Rice et al., 1993). A plasmid containing the PEA-3 and TRE of the collagenase promoter, pTRETK-CAT, was constructed as follows. The thymidine kinase promoter from herpes simplex virus type I was isolated from pHSV-106 (Life Technologies, Inc.) on a PvuII to BglII fragment. This fragment was blunted with the Klenow fragment of DNA polymerase I as described above and inserted at the XbaI site of pCAT-Basic (Promega Corp.) that had also been made blunt. A 71-bp oligonucleotide containing the collagenase PEA-3 and TRE (collagenase promoter sequences Ϫ111 to Ϫ40) was then inserted into the SphI site creating pTRETK-CAT.
Cell Line, Transfection, and cat Gene Analysis-Human dermal fibroblasts (HDF) were purchased from Clonetics Corp. and grown in Dulbecco's modified Eagle's medium (Sigma) supplemented with 2 mM L-glutamine (Sigma) and 10% fetal bovine serum (FBS) (Hyclone Laboratories). One day prior to transfection, HDF were plated at a density of 10 6 cells/100-mm culture dish. Calcium phosphate DNA precipitates (Wigler et al., 1979) were prepared with 5 g of either Ϫ1200/ϩ63 CAT or pCAT4660 and 10 g of either a plasmid expressing an individual PKC isozyme or the control vector pSVHNX-neo. Four hours after transfection, 1.0 ml of 15% glycerol in Dulbecco's modified Eagle's medium was placed on the cells for 60 s. The cells were washed twice with phosphate-buffered saline, and then fresh culture medium with either 10 or 1% FBS was added to the dishes. Forty-eight hours after transfection, the cells were washed two times with phosphate-buffered saline and then removed from the dishes by scraping in TEN buffer (40 mM Tris, pH 7.5, 1.0 mM EDTA, 150 mM NaCl). After centrifugation, the pellet of cells was resuspended in 100 l of phosphate-buffered saline containing 1.0 mM phenylmethylsulfonyl fluoride. A cellular lysate was prepared by three consecutive cycles of freeze/thaw, and the protein concentration was determined for each lysate using the BCA protein assay reagent (Pierce). The amount of Cat activity or the concentration of Cat protein present in the lysates was then determined. Cat activity was measured according to the method of Gorman et al. (1982). Briefly, 10 g of protein was used for each assay and diluted to 145 l with 0.25 M Tris, pH 7.8, containing 4 mM acetyl-CoA (Boehringer Mannheim) and 0.5 Ci of [ 14 C]chloramphenicol (DuPont NEN). After incubating at 37°C overnight, 1.0 ml of ethyl acetate was added to stop the reaction and to extract the acetylated and nonacetylated chloramphenicol species. The chloramphenicol species were then separated by ascending thin layer chromatography in a chamber saturated with a chloroform: methanol (95:5) solvent. The chloramphenicol species were quantitated with the Betascope 603 Blot Analyzer (Betagen Corporation). Percent-age of acetylation was calculated by dividing the quantity of acetylated chloramphenicol by the sum of all chloramphenicol species. The concentration of Cat protein present in transfected cells was measured with an ELISA kit (Boehringer Mannheim). Manufacturer's recommendations were followed with a few modifications. Each sample contained 25 g of protein dissolved in sample buffer to 100 l; this solution was transferred to a microtiter plate well containing Cat antibody and incubated for 2 h at 37°C. The rest of the procedure was followed according to the manufacturer's recommendations until the final stage when samples were incubated at room temperature with substrate plus enhancer for at least 15 min; the absorbance of each sample was then measured at 405 nm with a Dynatech MR 5000 (Dynatech Laboratories) microplate reader. Cat protein concentrations were extrapolated from a standard curve prepared with Cat enzyme supplied in the kit. The degree of trans-activation of the uPA or collagenase promoter by the PKC isozymes was calculated and expressed as fold activation. Fold activation was calculated by dividing the percentage of acetylation (TLC method) or the amount of Cat protein produced (ELISA method) from each co-transfection by that value generated from the co-transfection containing the control plasmid, pSVHNX-neo.
Statistics-Statistical analyses were performed with the statistical software JMP (SAS Institute Inc.), version 3.0. To determine if a given isozyme had activity significantly greater than the control, a one sample t test was performed on the logarithm of the fold activation for each isozyme. The logarithm transformation was used because fold activation was determined on a ratio scale. One-sided p values are reported because only increased activity is of interest. A p value of Ͻ0.05 was considered significant.
Gel Mobility Retardation Assay-HDF were transfected with 10 g of pSVHNX-neo, pSPKC⑀, or pSPKC as described above. Forty-eight hours after the glycerol shock, nuclear extracts were prepared as described by Dignam et al. (1983). The probe for the gel mobility retardation assays was a 17-bp, double-stranded oligonucleotide encompassing the TRE motif from the human collagenase promoter. The sequence of this oligonucleotide is as follows, AAGCATGAGTCAGACAC. The oligonucleotide was labeled at the 5Ј ends with T4 polynucleotide kinase (Life Technologies, Inc.) and [␥-32 P]ATP (DuPont NEN). Nuclear extracts (1 g) were incubated under conditions similar to those described by Carthew et al. (1985) at 25°C for 30 min with 0.03 pmol (approximately 10,000 cpm) of probe. The 10-l reaction mixture also contained 200 ng of poly(dI-dC)-poly(dI-dC) (Pharmacia Biotech Inc.), 15% glycerol, 20 nM HEPES (pH 7.9), 100 mM KCl, 5 mM MgCl 2 , 0.2 mM EDTA, and 0.5 mM dithiothreitol. Samples were subjected to electrophoresis on either a 4 or 4-20% gradient, nondenaturing polyacrylamide gel. After the gel was dried, the protein-DNA complexes were visualized by autoradiography. The quantity of probe in each band was determined by analyzing the dried gel with a Betascope 603 Blot Analyzer (Betagen Corp.) or by analyzing the autoradiogram with an Ultrascan XL Laser Densitometer (Pharmacia).

Promoters Containing PEA-3 and TRE Motifs Are Regulated
Differently by PKC Isozymes-To investigate PKC isozyme regulation of promoters containing classical phorbol ester-responsive sequences, co-transfections were performed in HDF with a plasmid containing either the human collagenase promoter or the porcine uPA promoter linked to the cat gene (Ϫ1200/ϩ63 CAT or pCAT 4660, respectively) and either the control vector (pSVHNX-neo) or a plasmid expressing an individual PKC isozyme ("Materials and Methods," Fig. 1). As shown in Fig. 2A, only PKC ␦, ⑀, and (members of the novel PKC subfamily) trans-activated the collagenase promoter, whereas PKC ␣, ␤II, ␥, and were unable to significantly enhance transcription above that of the control vector, pSVHNX-neo. Interestingly, the uPA promoter was significantly trans-activated to varying degrees by all of the PKC isozymes tested (Fig. 2B). Taken together these results demonstrate that the collagenase promoter is regulated in an isozyme-selective manner, and at least two phorbol ester-inducible promoters, collagenase and uPA, are not regulated in a similar manner by the PKC isozymes.
Effect of Serum on Isozyme-regulated Gene Expression-Because each isozyme significantly trans-activated at least one promoter in the study discussed above, we felt that all of the exogenous PKC isozymes were in a functional state. To inves-tigate further this issue, the co-transfections were repeated with 1% FBS instead of the usual 10% in the culture medium following co-transfection. Under these conditions, none of the isozymes trans-activated the collagenase promoter (Fig. 3A), and only PKC ⑀ weakly stimulated the uPA promoter above that seen with the control vector (Fig. 3B). Therefore, these data suggest that a substance present in the FBS was responsible for activating the exogenous isozymes that resulted in the data shown in Fig. 2.
PEA-3/TRE Motifs Involved in Isozyme-selective Trans-activation-The observation that only PKC ⑀ and strongly transactivated the collagenase promoter represents an interesting and unique finding. Therefore, additional studies were performed to investigate the mechanism for selective gene regulation. The sequence responsible for phorbol ester induction of the collagenase promoter is located between Ϫ111 and Ϫ40 and contains both a PEA-3 and a TRE motif and a TTCA sequence (Angel et al., 1987b;Gutman and Wasylyk, 1990;Auble and Brinckeroff, 1991). To determine if this sequence is involved in PKC ⑀ and trans-activation, a plasmid was first constructed that contains Ϫ111 to Ϫ40 of the collagenase promoter upstream of the HSV-1 thymidine kinase promoter linked to the cat gene, pTRETK-CAT (Fig. 4A). Co-transfections were then performed with pTRETK-CAT and either the control vector (pSVHNX-neo), pSPKC⑀, pSPKC, or pSPKC. Consistent with the results described above, PKC ⑀ and significantly transactivated this region of the collagenase promoter, whereas was unable to stimulate transcription above that seen with the control vector (Fig. 4B). Although the fold induction was less when the smaller region of the promoter was used compared with the results with the entire promoter (4.0 to 4.5 versus 5.5 to 7.5, respectively), regulation of the collagenase promoter by PKC ⑀ and appears to be mediated, at least in part, through the PEA-3/TRE motifs.
Molecular Mechanism for Isozyme-selective Trans-activation-PKC isozymes have been implicated in altering the DNA binding activity of c-Jun, a member of the Ap-1 family of transcription factors (Boyle et al., 1991;Jackson, 1992;Goode et al., 1992). In addition, Ap-1 family members have been shown to regulate transcription of the collagenase promoter at the TRE site (Angel et al., 1987b;Angel et al., 1988;Chiu et al., 1988). Because PKC ⑀ enhanced transcription of the collagenase promoter at the PEA-3/TRE, a gel mobility retardation assay was performed to determine if Ap-1 family members are bound to the collagenase TRE when exogenous PKC ⑀ is present. A 17-bp oligonucleotide probe representing the native collagenase TRE, nuclear extracts prepared from cells transfected with pSPKC⑀, and varying concentrations of unlabeled competitor oligonucleotides representing either the native or a mutant

FIG. 1. Schematic representation of pSVHNX-neo and a description of the individual PKC cDNAs and the resulting PKC expression plasmids.
Construction of pSVHNX-neo and the PKC expression plasmids is described under "Materials and Methods." Briefly, pSVHNX-neo contains the ␤-lactamase gene (AmpRes) and a prokaryotic origin of replication from pBR322. The neomycin resistance gene (neo), derived from pSV2neo (Southern and Berg, 1982), is regulated by the SV40 early promoter (SV40EP) and has downstream SV40 processing sequences (Intron, SV40poly(A)). The PKC cDNAs were ligated (either cohesive or blunt-ended) into the unique NheI site located in the polylinker (PL) of pSVHNX-neo. The PKC cDNAs are regulated by SV40 sequences (SV40EP, Intron, SV40poly(A)). The resulting PKC expression plasmids each express an individual isozyme.

FIG. 2. Effect of individual PKC isozymes on the full-length collagenase (A) or uPA (B) promoter in 10% FBS.
A, co-transfections were performed with Ϫ1200/ϩ63CAT and either pSVHNX-neo (control vector, CV) or a plasmid that expresses an individual PKC isozyme. Fold activation describes the degree of trans-activation by the individual PKC isozymes. This value was calculated by dividing the percentage of acetylation (TLC) or the amount of Cat protein produced (ELISA) from each co-transfection by that value generated from the co-transfection with the control vector. The significance of the fold activation for each isozyme was determined as described under "Materials and Methods." The data represent the average from three to six different co-transfections (mean Ϯ S.E.). B, co-transfections were performed with pCAT 4660 and either pSVHNX-neo (CV) or a plasmid expressing an individual PKC isozyme. Fold activation was calculated, and the significance of these values was determined as described above. The data shown represent the average of from three to four separate co-transfections (mean Ϯ S.E.). TRE sequence were used in the gel mobility retardation assay. The mutant sequence has previously been shown to have very little TRE activity (Angel et al., 1987b). As seen in Fig. 5A, a single band was produced in all lanes suggesting that a single protein or a single protein complex is bound to the TRE. The amount of bound probe in each lane was determined as described under "Materials and Methods." As seen in Fig. 5B, the native TRE sequence competed for Ap-1 factor binding in a dose-dependent manner, whereas the mutant TRE oligonucleotide efficiently competed for binding only at the 100-fold concentration. These results demonstrate that an Ap-1 family member is specifically bound to the TRE sequence when exogenous PKC ⑀ is present in the nuclear extracts.
Although a single band was produced in the gel mobility retardation assay described above, subtle changes in the form of the protein or complex of proteins bound to the TRE may have occurred in response to exogenous PKC ⑀. We have found that better resolution of protein-DNA complexes from gel mobility retardation assays can be obtained on gradient, nondenaturing polyacrylamide gels (Birch et al., 1996). Therefore, additional gel mobility retardation assays were performed to further investigate the molecular mechanism for PKC ⑀-selective gene regulation. As seen in Fig. 6A, two distinct bands (B1 and B2) were consistently produced when binding reactions containing nuclear extracts prepared from cells transfected with the control vector, pSPKC⑀, or pSPKC and incubated with the 17-bp TRE probe were resolved on 4 -20% gradient gels. These two bands could represent either two different Ap-1 dimers or two forms, perhaps phosphorylation alterations, of the same dimer bound to the TRE. As mentioned above, PKC isozymes have been shown to alter the DNA binding activity of Ap-1 family members (Boyle et al., 1991;Jackson, 1992;Goode et al., 1992); therefore, the density of B1 and B2 in each lane was determined as described under "Materials and Methods." Only small variations in the total amount of bound probe (B1 ϩ B2) were detected (Fig. 6B); however, the relative amount of probe present in B1 compared with B2 was altered. The amount of probe bound in B1 increased while that in B2 decreased when nuclear extracts prepared from cells transfected with pSPKC⑀ or pSPKC were compared with cells transfected with the control vector, pSVHNX-neo (Fig. 6C). These results suggest that the total amount of protein bound to the TRE was altered only slightly by exogenous PKC isozymes, whereas the protein or form of protein bound was significantly altered. Surprisingly, the binding pattern and the quantity of bound probe were very similar for extracts prepared from cells transfected with pSPKC⑀ or pSPKC. These observations suggest that trans-activation of the collagenase promoter by PKC ⑀ is not due to alterations in the binding of Ap-1 family members to the TRE motif.

DISCUSSION
Due to the heterogeneity that exists among the PKC family of isozymes and reports that suggest that individual isozymes may have distinct roles in biological functions (Nishizuka, 1988;Borner et al., 1992;Dekker and Parker, 1994), we designed the study presented here to investigate trans-activation of two phorbol ester-responsive genes by individual PKC isozymes. Our finding that the collagenase promoter is regulated only by the novel PKC isozymes ␦, ⑀, and clearly dem-

FIG. 3. The effect of individual PKC isozymes on the fulllength collagenase (A) or uPA (B) promoter in the presence of 1% FBS.
A, the plasmid Ϫ1200/ϩ63CAT was transfected with either pSVHNX-neo (control vector, CV) or a plasmid that expresses an individual PKC isozyme in medium containing 1% FBS as described under "Materials and Methods." Fold activation was calculated as described previously, and the significance of these values was determined using the statistical software JMP, version 3.0. The data represent the average of three separate co-transfections (mean Ϯ S.E.). B, a plasmid containing the full-length uPA promoter, pCAT 4660, was transfected with either pSVHNX-neo (CV) or a plasmid that expresses an individual PKC isozyme in medium containing 1% FBS. Fold activation for each isozyme was calculated, and the significance of the values was determined as described previously. The data represent the average of two separate co-transfections (mean Ϯ S.E.).

FIG. 4. Schematic representation of pTRETK-CAT (A) and the effect of PKC ⑀, , and on the human collagenase PEA-3/TRE motifs (B).
A, pTRETK-CAT contains a 71-bp oligonucleotide representing sequences from Ϫ111 to Ϫ40 of the human collagenase promoter. This region of the promoter contains both a PEA-3 and a TRE motif. The 71-bp oligonucleotide was placed upstream of the herpes simplex virus type I thymidine kinase promoter. This regulatory region was then linked to the cat gene. Construction of this plasmid is described under "Materials and Methods." B, pTRETK-CAT was transfected with pSVHNX-neo (control vector, CV), pSPKC⑀, pSPKC, or pSPKC. The results represent the average of five separate co-transfections (mean Ϯ S.E.).
onstrates that some promoters are regulated by specific PKC isozymes and supports the idea that individual isozymes have unique functions within the cell. Other investigators have also presented data demonstrating isozyme-selective gene regulation. Kariya et al. (1991) have shown that the ␤-myosin heavy chain promoter is regulated better by PKC ␤ than by ␣. In addition, the primary response gene, ST2, has recently been shown to be regulated by PKC ␤II, ␥, , and but not by ␣, ␦, or ⑀ (Kieser et al., 1995), and Hata et al. (1993) have shown that a concatenated TRE cluster is trans-activated by PKC ␣, ␤II, and ⑀ but not ␥. Although the results by these investigators are interesting, our data may more accurately reflect PKC isozyme regulation as we have used full-length cellular promoters, transiently expressed PKC isozymes, and physiological stimulation.
In addition to isozyme-selective gene regulation, we have shown that two phorbol ester-responsive promoters are regulated differently by the PKC isozymes (compare Fig. 2A and   2B). The TRE in both the human collagenase promoter (Angel et al., 1987a) and the porcine uPA promoter (Rorth et al., 1990) is required for induction of gene expression by tumor-promoting agents. In addition, the PEA-3 site on these promoters acts synergistically with the TRE to achieve maximal levels of induction (Gutman and Wasylyk, 1990;Rorth et al., 1990). Although the PEA/TRE motifs found in these promoters are similar, their sequences, the spacing between the PEA-3 and TRE, and their location within the promoter are unique to each gene (Angel et al., 1987a;Cassady et al., 1991). While it remains unknown whether these differences are responsible for the diverse trans-activation profiles seen in this study, it is possible that substrates for the different PKC isozymes differentiate the two promoters at these sites.
Similar to our finding that serum concentrations greater than 1% are necessary to achieve trans-activation (Figs. 2 and 3), other investigators have shown that a substance present in serum activates PKC (Abdel-Ghany et al., 1989). In fact, Ohno et al. (1994) report that serum selectively activates the PKC isozymes. Our data with the collagenase promoter ( Fig. 2A) is consistent with the data of Ohno et al. (1994), i.e. serum activation leads to selective trans-activation by the novel PKC isozymes. However, our findings with the uPA promoter ( Fig.  2B) suggest that all of the isozymes are functional, and the inability to trans-activate the collagenase promoter by some of the isozymes (PKC ␣, ␤II, ␥, and ) in Fig. 2A is due to a FIG. 5. Identification of protein(s) bound to the collagenase TRE motif. A, gel mobility retardation assays were performed as described under "Materials and Methods" with 0.03 pmol of native TRE oligonucleotide as probe, 1 g of nuclear extract prepared from cells transfected with pSPKC⑀, and 0, 10-, 50-, and 100-fold excess unlabeled, competitor oligonucleotide (mutant or native). The native and mutant TRE oligonucleotides differ only by the two bases that are shown in bold in the mutant TRE sequence. Bound probe was separated from the free probe on a 4% nondenaturing polyacrylamide gel. B, the percent of native TRE sequence that was bound (% of probe bound) was determined from this representative gel by scanning the bands on the autoradiogram with a densitometer as described under "Materials and Methods." The percent of bound probe was then plotted versus the amount of competitor oligonucleotide that was added to the reaction.
FIG. 6. Effect of PKC ⑀ and on the binding of Ap-1 family members to the TRE motif. A, gel mobility retardation assays were performed with 0.03 pmol of labeled, native TRE oligonucleotide and nuclear extracts prepared from cells transfected with pSVHNX-neo (control vector, CV), pSPKC⑀, or pSPKC as described under "Materials and Methods." Bound probe was separated from free probe on a 4 -20% gradient, nondenaturing polyacrylamide gel. Two distinct bands (B1 and B2) were consistently produced. The gel shown is a representative of three separate experiments. B, the amount of bound probe was determined by scanning each lane of the autoradiogram with an Ultrascan XL laser densitometer. The total amount of bound probe in each lane was determined as the area under the curve for B1 plus B2. C, the percentage of bound probe present in B1 and B2 was determined for each lane (% relative area under the curve) with the Ultrascan XL laser densitometer. mechanism distal to isozyme activation.
An Ap-1 family member was shown to be bound to the collagenase TRE when nuclear extracts prepared from HDF transfected with pSPKC⑀ were used in a gel mobility retardation assay (Fig. 5, A and B). When additional binding reactions were separated on 4 -20% gradient gels, two distinct bands were clearly seen (Fig. 6A). The two bands could represent two different Ap-1 dimers (such as c-Jun/c-Jun or c-Jun/c-Fos) bound to the TRE. Different dimers have been shown previously to migrate to different positions in a gel mobility retardation assay (de Groot et al., 1991). Alternatively, the two bands could represent two different forms, perhaps phosphorylation forms, of the same dimer. Both c-Fos and c-Jun have been shown to be phosphorylated by a number of protein kinases (Abate et al., 1991;Jackson, 1992;Smeal et al., 1991;Pulverer et al., 1991), and phosphorylation is known to alter the electrophoretic migration of c-Fos (Barber and Verma, 1987).
Although PKC ⑀ trans-activates the full-length collagenase promoter ( Fig. 2A) and the Ϫ111 to Ϫ40 region from the collagenase promoter (Fig. 4B), it does not appear to alter the binding pattern (Fig. 6A) or enhance total binding (Fig. 6B) of Ap-1 family members to the TRE. The most likely explanation for this finding is that the TRE is not solely responsible for trans-activation. The PEA-3 site or a combination of the TTCA, PEA-3, and TRE may all be required for trans-activation of the collagenase promoter as suggested by Auble and Brinckeroff (1991). Interestingly, other investigators have also shown that PKC isozyme-selective trans-activation is not necessarily consistent with the binding of Ap-1 family members at the TRE. Hata et al. (1993) have shown that PKC ␥ is unable to stimulate transcription from multiple TREs linked to the cat gene, although ␥ enhances binding of proteins at the TRE site. Our findings together with those of others suggest that additional factors and/or an alteration of bound Ap-1 family members may be necessary to achieve transcriptional activation at this site. In fact, the results shown in Fig. 6C suggest that either the type of protein (Ap-1 dimer) or form of protein bound to the TRE was altered in PKC ⑀-transfected cells; however, these alterations do not appear to correlate with isozyme-selective transcriptional activity as an increase in B1 compared with B2 was also seen with nuclear extracts prepared from cells transfected with pSPKC.
In conclusion, we have shown that two promoters (human collagenase and porcine uPA) that contain classical phorbol ester-responsive elements were regulated differently by PKC isozymes. The collagenase promoter was only trans-activated significantly by PKC ␦, ⑀, and , whereas the uPA promoter was stimulated by additional isozymes. We also show that the sequences between Ϫ111 to Ϫ40 of the collagenase promoter were at least partially responsible for the isozyme-specific activity. Although the exact mechanism for this specificity remains unknown, we demonstrate that changes in binding of Ap-1 family members to the TRE motif induced by PKC ⑀ does not correlate with isozyme selective activation.