Multiple Promoter Elements Are Required for the Stimulatory Effect of Insulin on Human Collagenase-1 Gene Transcription

Several of the complications seen in patients with both type I and type II diabetes mellitus are associated with alterations in the expression of matrix metalloproteinases. To identify the cis-acting elements that mediate the stimulatory effect of insulin on collagenase-1 (matrix metalloproteinase-1) gene transcription a series of collagenase-chloramphenicol acetyltransferase (CAT) fusion genes were transiently transfected into HeLa cells. Multiple promoter elements, including an Ets and activator protein-1 (AP-1) motif, were required for the effect of insulin. The AP-1 motif appears to be a target for insulin signaling because it is sufficient to mediate an effect of insulin on the expression of a heterologous fusion gene, whereas the data suggest that the Ets motif acts to enhance the effect of insulin mediated through the AP-1 motif. Multiple promoter elements were also required for the stimulatory effect of phorbol esters on collagenase-CAT gene transcription, and the AP-1 motif was also a target for phorbol ester signaling. However, thecis-acting elements required for the effects of insulin and phorbol esters were not identical. Moreover, phorbol esters were a much more potent inducer of collagenase-CAT gene transcription than insulin, a difference that may be explained by selective effects of insulin and phorbol esters on AP-1 expression.

The maintenance of the extracellular matrix is accomplished by a balance of synthesis and degradation, the latter being determined by the relative activities of a family of extracellular matrix proteinases, the matrix metalloproteinases (MMPs), 1 and the tissue inhibitors of MMPs (1)(2)(3). Tissue inhibitors of MMPs block MMP action by binding covalently to MMPs and preventing both activation of the MMP by enzymatic modification and the ability of MMPs to bind substrate (1)(2)(3).
The regulation of MMP gene expression has received considerable attention primarily because of the involvement of increased MMP activity in tumor progression, specifically the development of malignant carcinomas (4). However, several of the complications associated with both type I, insulin-dependent diabetes mellitus and type II, non-insulin-dependent diabetes mellitus, including glomerulosclerosis of the kidney (5,6), retinopathy (7) periodontal disease (8), and some forms of cardiac disease (9), are also characterized, in part, by alterations in the amount and composition of extracellular matrix protein.
In the streptozotocin rat model of type I diabetes, collagenase-1 (MMP-1; referred to henceforth as collagenase) gene expression in the glomerulus is decreased (10), and this would be predicted to contribute to an increase in the mesangial matrix and/or the glomerula basement membrane, one of the characteristics of glomerulosclerosis (5,6). This decrease in collagenase gene expression is reversed by insulin treatment (10), but it is unclear whether this represents a direct effect of insulin as opposed to an indirect effect mediated through changes in glucose concentration. The effect of insulin on collagenase expression in glomerula-derived cell lines has not been studied; however, in NIH 3T3 (11), Chinese hamster ovary (12), and HeLa cells (13) insulin stimulates the expression of reporter genes when ligated to the collagenase promoter. Interestingly, if insulin does directly regulate collagenase gene expression in the glomerulus, then collagenase expression would actually be predicted to be increased in hyperinsulinemic individuals because the insulin resistance associated with muscle, adipose tissue, and liver may not be manifest in the kidney (14).
cis-Acting elements that mediate the action of insulin on gene transcription are referred to as insulin response sequences (IRSs) or elements (15). Several IRSs have been identified, but it is apparent that a single consensus IRS does not exist (15). Instead, it is predicted that multiple classes of consensus IRSs will be found, only three of which have currently been identified (15). One of these has the consensus sequence T(G/A)TTT(T/G)(T/G) and mediates the insulin-dependent transcriptional inhibition of several hepatic genes such as those encoding phosphoenolpyruvate carboxykinase (PEPCK), insulin-like growth factor-binding protein-1, tyrosine aminotransferase, apolipoprotein CIII, and glucose-6-phosphatase (16 -20). The other consensus IRSs are the Ets motif (21) and the serum response element (15,22), which mediate stimulatory effects of insulin on several genes. However, unlike the PEPCK-type IRS that confers a selective effect of insulin (15), the Ets motif and serum response element can mediate the effects of multiple other hormones on gene transcription (23,24).
The activator protein-1 (AP-1) motif, which binds members of the Fos and Jun transcription factor families (25) and, like the Ets motif and serum response element, mediates transcriptional changes in response to multiple ligands (26), may represent a fourth class of consensus IRS. Thus, Rutter et al. (12) have shown that in Chinese hamster ovary cells, mutation of the AP-1 motif in the collagenase promoter abolishes the stimulatory effect of insulin on the expression of a collagenaseluciferase fusion gene. Similarly, in rat H4IIE hepatoma cells, an AP-1 motif is required for the stimulatory effect of insulin on the expression of a malic enzyme (ME)-chloramphenicol acetyltransferase (CAT) fusion gene (13). However, paradoxically, in HeLa cells insulin has almost no effect on ME gene transcription, whereas it markedly stimulates collagenase gene transcription (13). This observation raises the question as to why the AP-1 motif appears to only mediate an insulin-dependent activation of gene transcription in some cell contexts. To indirectly address this question, we have analyzed the promoter elements required for the stimulatory effect of insulin on collagenase gene transcription in HeLa cells. The results demonstrate that multiple promoter elements are required for the effect of insulin on collagenase gene transcription in addition to the AP-1 motif. However, the collagenase AP-1 motif appears to be a target of insulin signaling because it is sufficient to mediate an effect of insulin on the expression of a heterologous fusion gene. The additional collagenase promoter elements that are required for the full stimulatory effect of insulin may bind accessory factors that act to enhance the effect of insulin mediated through the AP-1 motif. The data also suggest that the mechanism of insulin and phorbol ester signaling through the collagenase-1 AP-1 motif are distinct.

Plasmid Construction
DNA manipulations were accomplished by standard techniques (27), and DNA sequencing was performed using the USB Sequenase kit. All plasmid constructs were purified by centrifugation twice through cesium chloride gradients (27).
A BglII-HindIII fragment of the human collagenase-1 promoter, spanning the sequence from Ϫ518 to ϩ64, was isolated from the plasmid pCol⅐Luc (12) and ligated, in the same orientation as that of the endogenous gene, into the BglII-digested polylinker of the pCAT(An) expression vector, a generous gift from Dr. Howard Towle (28). The noncompatible 3Ј HindIII-BglII junction was filled in using the Klenow fragment of Escherichia coli DNA polymerase I prior to blunt-end ligation. The pCAT(An) vector has polyadenylation signals located 5Ј of the polylinker to prevent read-through transcription (28). Control experiments demonstrated that there was no basal CAT expression and no effect of insulin or PMA when the pCAT(An) vector, minus the collagenase promoter, was transiently transfected into HeLa cells (data not shown). A series of truncated collagenase-CAT fusion genes was then generated, with the 5Ј end points shown in Fig. 4, using the Ϫ518/ϩ64 pCAT(An) construct as a template, by either restriction enzyme digestion or polymerase chain reaction, using standard techniques (27). The 3Ј polymerase chain reaction primer was designed to conserve the junction between the collagenase promoter and CAT reporter gene to be the same as that in all other collagenase-CAT fusion gene constructs. All promoter fragments generated by polymerase chain reaction were completely sequenced to ensure the absence of polymerase errors, whereas promoter fragments generated by restriction enzyme digestion were only sequenced to confirm the 5Ј end points. Site-directed mutants of the collagenase Ets and AP-1 motifs in the context of the Ϫ97 to ϩ64 promoter fragment, as shown in Fig. 5, were generated by polymerase chain reaction in conjunction with 5Ј primers that incorporated the mutated sequence (Table I).
A plasmid designated X(An) was generated by ligating a minimal Xenopus 68-kDa albumin promoter (29) with the following sequence (5Ј-AAGCTTGATCTCTCTGAGCAATAGTATAAAACTCGAG-3Ј; Hin-dIII and XhoI restriction enzyme sites underlined; TATA box in italics) into HindIII-XhoI cleaved pCAT(An). The plasmid XMB was then generated by removing a BamHI site located just 5Ј of the HindIII site by the use of mung bean nuclease. Double-stranded complementary oligonucleotides, representing various regions of the collagenase promoter (Table I), were synthesized with HindIII compatible ends and ligated into HindIII-cleaved XMB in multiple copies. The number of inserts was determined by restriction enzyme analysis and confirmed by DNA sequencing.

Cell Culture and Transient Transfection
Human HeLa cervical carcinoma cells were grown to 90% confluence in T150 flasks in DMEM containing 10% (v/v) calf serum and were replated the day before use into 55-cm 2 culture dishes (1 flask to 26 dishes). Attached cells were transfected by addition of 0.5 ml of a calcium phosphate-DNA co-precipitate (30), containing the reporter gene construct (15 g), an expression vector for ␤-galactosidase (2.5 g), and an expression vector encoding the insulin receptor (5 g), courtesy of Dr. Jonathan Whittaker, to the 10 ml of culture medium. After an overnight incubation the medium was removed and the cells incubated in 10 ml PBS for 10 min at room temperature. The cells were then incubated for a further 8 -24 h in 10 ml serum-free DMEM supplemented with or without various concentrations of PMA or insulin, as indicated in the Figure legends, prior to harvesting.
Hamster insulinoma tumor cells were grown to 70% confluence in T150 flasks in DMEM containing 2.5% (v/v) fetal bovine serum and 15% (v/v) horse serum and were replated the day before use into 55-cm 2 culture dishes (1 flask to 14 dishes). Attached cells were then cotransfected as described previously (31) by addition of 0.5 ml of a calcium phosphate-DNA co-precipitate containing 15 g of reporter plasmid DNA and expression vectors encoding ␤-galactosidase (2.5 g) and the insulin receptor (5 g) to the 10 ml of culture medium. After incubation for between 4 and 6 h, the cells were treated for 2 min with 20% glycerol in serum-free DMEM (5 ml/dish). The cells were then rinsed for 5 min with serum-free DMEM (5 ml/dish) prior to incubation for 24 h in serum-free DMEM (10 ml/dish) supplemented with or without various concentrations of insulin, as indicated in the Figure legends.
Rat hepatoma H4IIE cells were grown in 12-well plates in DMEM containing 10% (v/v) fetal bovine serum. Attached cells were transfected at ϳ60% confluence with the plasmid pCol⅐Luc (2 g; Ref. 12) using 4 l of the Tfx-50 transfection reagent (Promega) in a final volume of 400 l serum-free DMEM, as described previously (32). After incubation for 2 h, the cells were then overlaid with 2 ml of serum-containing DMEM. After an overnight incubation, the cells were washed with serum-free DMEM for 2 h prior to incubation for 24 h in serum-free DMEM supplemented with or without various concentrations of insulin, as indicated in the Figure legends. Under these growth and transfection conditions 100 nM insulin was not toxic to these cells (see Fig. 2), in contrast to previous experiments in which the cells were grown in DMEM containing 2.5% (v/v) fetal calf serum and 2.5% (v/v) newborn calf serum and were transiently transfected in solution using calcium phosphate-DNA co-precipitation (13).

CAT, ␤-Galactosidase, and Luciferase Assays
Cells were either harvested by trypsin digestion and sonicated in 300 l of 250 mM Tris (pH 7.8) containing 2 mM phenylmethylsulfonyl fluoride, or for luciferase assays, cells were extracted by scraping into passive lysis buffer (Promega). CAT, ␤-galactosidase, and luciferase assays were performed exactly as described previously (20,30,32). To compare the relative basal CAT expression obtained with the various reporter gene constructs described, CAT activity in samples from control cells was corrected for the ␤-galactosidase activity in the same sample. Because phorbol esters and insulin affect Rous sarcoma virus-␤ galactosidase expression in HeLa cells (data not shown), CAT activity in these samples was corrected for the protein concentration in the cell lysate, as measured by the Pierce BCA assay, and each plasmid construct was analyzed in duplicate in multiple transfections, as specified in the Figure legends.

Gel Retardation Assay
Labeled Probes-Oligonucleotides representing the sense and antisense strands of the collagenase AP-1 and Ets motifs (Table I) were synthesized with BamHI or HindIII compatible ends, respectively, gel purified, annealed, and then labeled with [␣-32 P]dATP using the Klenow fragment of Escherichia coli DNA polymerase I to a specific activity of approximately 2.5 Ci/pmol.
Nuclear Extract Preparation-HeLa nuclear extracts were prepared exactly as described previously (13).
AP-1 Binding Assay-Labeled AP-1 oligonucleotide (ϳ7.5 fmol, ϳ30,000 cpm) was incubated with HeLa (3 g) nuclear extract in a final reaction volume of 20 l containing 20 mM HEPES, pH 7.8, 100 mM NaCl, 0.38 mM spermidine, 0.08 mM spermine, 0.1 mM EDTA, 1 mM EGTA, 2 mM dithiothreitol, 12.5% glycerol (v/v), and 1 g of poly(dI-dC)⅐poly(dI-dC). After incubation for 10 min at room temperature, the reactants were loaded onto a 6% polyacrylamide gel and electrophoresed at room temperature for 90 min at 150 V in a buffer containing 25 mM Tris⅐HCl at pH 7.8, 190 mM glycine, and 1 mM EDTA. Following electrophoresis, the gels were dried and exposed to Kodak XAR5 film, and binding was analyzed by autoradiography.
Ets Binding Assay-When the Ets oligonucleotide was used as the labeled probe, the binding conditions were identical to those described for AP-1 except that the NaCl concentration was decreased to 50 mM and poly(dI-dC)⅐poly(dI-dC) was reduced to 0.5 g. In addition, visualization of specific Ets binding (see Fig. 8B) required preincubation of HeLa nuclear extract with 10 ng of chymotrypsin for 2 min at room temperature prior to addition of the labeled probe and binding buffer and a further 10 min of incubation at room temperature. Binding was then analyzed by acrylamide gel electrophoresis as described above.
Competition Experiments-For competition experiments (see Fig. 8), the indicated unlabeled double-stranded oligonucleotides (100-fold molar excess) were mixed with the labeled oligomer prior to addition of nuclear extract. Binding was then analyzed by acrylamide gel electrophoresis as described above.
Gel Supershift-Gel supershift assays (see Fig. 9) were carried out by incubating HeLa nuclear extract (3 g) with the indicated antisera for 10 min at room temperature, prior to the addition of the labeled AP-1 oligonucleotide probe and binding buffer and incubation for an additional 10 min.

Insulin Stimulates Collagenase-CAT Fusion Gene Transcription in HeLa
Cells-To begin to study the regulation of collagenase gene transcription by insulin, a collagenase-CAT fusion gene construct, containing collagenase promoter sequence from Ϫ518 to ϩ64, relative to the transcription start site at ϩ1 (33), was transiently transfected into HeLa cells (Fig. 1). An effect of insulin on reporter gene expression was only detected when the collagenase-CAT fusion gene was co-transfected with an expression vector encoding the insulin receptor (Fig. 1A). In the absence of insulin, co-transfection with the insulin receptor alone was insufficient to activate collagenase-CAT fusion gene expression, suggesting a low level of signaling through the basal receptor (Fig. 1B). Stanley (34) has previously shown that co-transfection with an insulin receptor expression vector is also required to observe an effect of insulin on the expression of a transiently transfected prolactin-CAT fusion gene in rat pituitary tumor GH4 cells. Interestingly, insulin stimulates the expression of the endogenous prolactin gene in these cells in the absence of receptor co-transfection (34). The maximal effect of insulin on collagenase-CAT gene expression was seen at 100 nM ( Fig. 2A). This concentration of insulin is 10-fold higher than that required to see a maximal repression of glucocorticoid-stimulated PEPCK-CAT fusion gene expression in rat H4IIE hepatoma cells (35,36). This requirement for 100 nM insulin to manifest the maximal effect of the hormone is not explained by the degradation of insulin by HeLa cells because a similar EC 50 for this effect is obtained following an 8 or 24 h incubation with insulin ( Fig. 2B). In addition, the maximal effect of insulin on collagenase-CAT gene expression was also seen at 100 nM in hamster insulinoma tumor cells (Fig. 2C), H4IIE cells (Fig. 2D), and Chinese hamster ovary-T cells (data not shown). Although cell line-dependent variations in insulin sensitivity could arise due to differences in the expression of molecules in the insulin signaling pathway, our results suggest that different signaling pathways are used by insulin to regulate PEPCK and collagenase gene transcription.

Multiple, Distinct cis-Acting Elements Are Required for the Actions of Insulin and Phorbol Esters on Collagenase Gene
Transcription-The result shown in Fig. 1 suggests that an IRS is present in the collagenase promoter between Ϫ518 and ϩ64. Because the cis-acting elements that mediate the stimulatory effect of phorbol esters on collagenase gene transcription have been studied in detail (1,2,33,37), and given the long running controversy concerning the potential role of protein kinase C in insulin action (38), it was of interest to determine whether insulin and phorbol esters both mediate their stimulatory action on collagenase gene transcription through the same elements. The phorbol ester PMA has a biphasic effect on collagenase-CAT gene transcription, with a maximal stimulation of ϳ150-fold seen at 10 nM (Fig. 3). The reduction in PMA-stimulated CAT expression at higher PMA concentrations probably reflects the down-regulation of protein kinase C (39). By contrast, the maximal stimulation obtained with insulin was ϳ30fold ( Fig. 2A), and the stimulation of collagenase-CAT fusion gene expression by insulin and PMA was not additive (data not shown).
To delineate the collagenase IRS, a series of 5Ј deletion mutations of the collagenase promoter was constructed in the FIG. 1. Insulin stimulates collagenase-CAT fusion gene transcription in HeLa cells. HeLa cells were transiently co-transfected, as described under "Experimental Procedures," with a collagenase-CAT fusion gene (15 g) containing promoter sequence from Ϫ518 to ϩ64, an expression vector encoding ␤-galactosidase (2.5 g), and either with (ϩIREV) or without (ϪIREV) an expression vector encoding the insulin receptor (5 g). Following transfection, cells were incubated for 24 h in serum-free medium in the presence or absence of 10 nM insulin. The cells were then harvested, and both CAT and ␤-galactosidase activity were assayed as described previously (20,30). In A, results are presented as the ratio of CAT activity, corrected for protein concentration in the cell lysate, in insulin-treated versus control cells and are expressed as fold induction. In B, results are presented as the ratio of CAT to ␤-galactosidase activity in control cells and are expressed as arbitrary units. Results represent the mean of Ϯ S.E. of three experiments, in which each condition was assayed in duplicate.
CAT reporter plasmid pCAT(An) (28), and the effect of insulin on CAT expression directed by these constructs was analyzed following transient transfection into HeLa cells. Fig. 4A shows that insulin-stimulated collagenase-CAT gene transcription was unchanged when the region of the promoter between Ϫ518 and Ϫ159 was deleted. By contrast, the effect of PMA was reduced more than 50% (Fig. 4B), indicating that these two agents do not have an identical mechanism of action on collagenase gene transcription. Further deletion of the collagenase promoter sequence between Ϫ158 and Ϫ124 resulted in a slight decrease in insulin-stimulated collagenase-CAT gene transcription (Fig. 4A) and a slight increase in phorbol ester-stimulated collagenase-CAT gene transcription (Fig. 4B), as previously reported in a study by Auble and Brinckerhoff (40) on the rabbit collagenase-1 promoter in fibroblasts. Two additional collagenase-CAT fusion genes, in which the collagenase promoter sequence from Ϫ123 to Ϫ98 and then from Ϫ97 to Ϫ80 were deleted, both directed decreased phorbol ester-and insulin-stimulated CAT expression (Fig. 4, A and B).
The progressive loss in insulin-stimulated collagenase-CAT gene expression following deletion of the promoter sequence from Ϫ158 to Ϫ80 is indicative of the presence of either multiple insulin response sequences in this region or multiple binding sites for accessory factors that enhance the action of a more proximal IRS. These same deletions result in decreased basal collagenase-CAT expression (Fig. 4C). Whether the transcription factors that confer high basal collagenase gene expression through the Ϫ158 to Ϫ98 promoter region are the same as those that mediate/enhance the insulin effect remains to be determined. Instead, we decided to focus this study on the more proximal portion of the promoter, between Ϫ97 and ϩ64, because this region contains the Ets and AP-1 motifs that have previously been shown to be important in mediating the induction of collagenase gene transcription by phorbol esters (1,2,24,37,41). Moreover, this proximal region of the promoter confers a response to both insulin and phorbol esters that is 50% of that seen with the full-length promoter (Fig. 4, A and B).
The Collagenase Ets and AP-1 Motifs Are Both Required for the Stimulatory Effects of Insulin and Phorbol Esters on Gene Transcription-To directly assess the relative importance of the collagenase Ets and AP-1 motifs in mediating the induction of gene transcription by insulin and phorbol esters, these cisacting elements were mutated within the context of the Ϫ97 to ϩ64 promoter fragment (Fig. 5). Mutation of the Ets motif reduced the stimulatory effects of both insulin and phorbol esters on collagenase-CAT fusion gene transcription, whereas mutation of the AP-1 motif abolished the response to both agents (Fig. 5, A and B). Thus, an intact AP-1 motif is essential for the induction of collagenase-CAT gene expression by insulin and phorbol esters, whereas the Ets motif is not.
The collagenase Ets and AP-1 motifs, particularly the latter, are also important for basal collagenase gene transcription (Fig. 5C). Thus, although mutation of the Ets motif reduces basal collagenase-CAT gene expression, mutation of the AP-1 motif reduces basal gene expression to barely detectable levels, despite the presence of an intact Ets motif (Fig. 5C). In addition, mutation of the AP-1 motif within the context of the Ϫ79 to ϩ64 promoter fragment reduces basal fusion gene expression to the same level as that seen when the AP-1 motif is mutated in the context of the Ϫ97 to ϩ64 promoter fragment, despite the presence of the intact Ets motif (Fig. 5C).
Mutation of the Ets motif within the context of the Ϫ97 to ϩ64 promoter fragment has the same quantitative effect, on both basal fusion gene expression and the response to insulin, as deletion of the Ϫ97 to Ϫ80 sequence (Fig. 5, A and C). By contrast, deletion of the Ϫ97 to Ϫ80 sequence is more deleterious to the phorbol ester response than a mutation of the Ets motif in the context of the Ϫ97 to ϩ64 promoter fragment (Fig.  5B). This suggests that the transcription factor(s) binding the HeLa cells were transiently co-transfected, as described under "Experimental Procedures," with a collagenase-CAT fusion gene (15 g), containing promoter sequence from Ϫ518 to ϩ64, and expression vectors encoding ␤-galactosidase (2.5 g) and the insulin receptor (5 g). The insulin receptor expression vector was included to be consistent with all other transfection experiments. Following transfection, cells were incubated for 24 h in serum-free medium in the presence or absence of various concentrations of PMA. The cells were then harvested, and CAT activity was assayed as described previously (20,30). Results are presented as the ratio of CAT activity, corrected for protein concentration in the cell lysate, in PMA-treated versus control cells and are expressed as fold induction. Results represent the mean of Ϯ S.E. of four experiments, in which each condition was assayed in duplicate.

FIG. 2. Concentration dependence of insulin-stimulated collagenase-CAT fusion gene transcription.
HeLa and hamster insulinoma tumor (HIT) cells were transiently co-transfected using calcium phosphate co-precipitation, as described under "Experimental Procedures," with a collagenase-CAT fusion gene (15 g), containing promoter sequence from Ϫ518 to ϩ64, and expression vectors encoding ␤-galactosidase (2.5 g) and the insulin receptor (5 g). H4IIE cells were transiently transfected using the Tfx-50 transfection reagent, as described under "Experimental Procedures," with a collagenase-luciferase fusion gene (2 g), containing promoter sequence from Ϫ518 to ϩ64. Following transfection, cells were incubated for either 24 h (A, C, and D) or 8 h (B) in serum-free medium in the presence or absence of various concentrations of insulin. The cells were then harvested and CAT or luciferase activity was assayed as described previously (20,30,32). Results are presented as the ratio of CAT or luciferase activity (corrected for protein concentration in the cell lysate in A-C although not in D) in insulin-treated versus control cells and are expressed as fold induction. Results represent the mean of Ϯ S.E. of three to eight experiments, in which each condition was assayed in duplicate.
Ets motif mediates both basal gene expression and the response to insulin conferred through this region but that an additional element is important for the action of phorbol esters. Finally, mutation of the AP-1 motif in the context of the Ϫ79 to ϩ64 promoter fragment almost completely abolishes the small effect of insulin and phorbol esters on the expression of this truncated fusion gene, just as it does in the context of the Ϫ97 to ϩ64 promoter fragment (Fig. 5, A and B).
The Collagenase AP-1 and Ets Motifs Confer a Stimulatory Effect of Insulin and Phorbol Esters on the Expression of a Heterologous Promoter-The preceding experiments demonstrate that the AP-1 motif is required for the action of both insulin and phorbol esters on collagenase gene transcription and that the Ets motif enhances the action of both agents, although it is inactive in the absence of the AP-1 motif. However, these experiments do not prove that the transcription factors binding the AP-1 and/or Ets motifs are the targets of insulin and phorbol ester signaling, rather than acting as accessory factors to enhance the effects of these agents mediated through an unidentified response sequence. To determine whether the AP-1 and/or Ets motifs are the targets of insulin and phorbol ester signaling, multiple copies of an oligonucleotide representing the collagenase promoter sequence between FIG. 4. Progressive deletion of the collagenase promoter sequence between ؊518 and ؊80 reduces both basal collagenase-CAT fusion gene transcription and the stimulatory effects of insulin and PMA. HeLa cells were transiently co-transfected, as described under "Experimental Procedures," with a series of collagenase-CAT fusion genes, with 5Ј deletion end points as shown on the abscissa, and expression vectors encoding ␤-galactosidase and the insulin receptor. Following transfection, cells were incubated for 24 h in serum-free medium in the presence or absence of 10 nM insulin or 100 nM PMA. The cells were then harvested, and both CAT and ␤-galactosidase activity were assayed as described previously (20,30). Results are presented as the ratio of CAT activity, corrected for protein concentration in the cell lysate, in insulin-treated (A) or PMA-treated (B) versus control cells, and are expressed as fold induction. In C, results are presented as the ratio of CAT to ␤-galactosidase activity in control cells and are expressed as arbitrary units. Results represent the mean of Ϯ S.E. of 4 -13 experiments, in which each construct was assayed in duplicate.

FIG. 5. Site-directed mutation of the Ets and AP-1 motifs in the collagenase promoter reduces both basal collagenase-CAT fusion gene transcription and the stimulatory effects of insulin and PMA.
HeLa cells were transiently co-transfected, as described under "Experimental Procedures," with a series of collagenase-CAT fusion genes, with 5Ј deletion end points and mutations as shown on the abscissa, and expression vectors encoding ␤-galactosidase and the insulin receptor. Following transfection, cells were incubated for 24 h in serum-free medium in the presence or absence of 10 nM insulin or 100 nM PMA. The cells were then harvested, and both CAT and ␤-galactosidase activity were assayed as described previously (20,30). Results are presented as the ratio of CAT activity, corrected for protein concentration in the cell lysate, in insulin-treated (A) or PMA-treated (B) versus control cells and are expressed as fold induction. In C, results are presented as the ratio of CAT to ␤-galactosidase activity in control cells and are expressed as arbitrary units. Results represent the mean of Ϯ S.E. of 4 -13 (insulin) or 3 (PMA) experiments, in which each construct was assayed in duplicate.
Ϫ97 and Ϫ64, which contains both the Ets and AP-1 motifs (Table I), were ligated into the polylinker of a heterologous vector, designated XMB, which contains a minimal Xenopus 68-kDa albumin promoter (29) ligated to the CAT reporter gene, and the resulting construct was transiently transfected into HeLa cells (Fig. 6). Neither basal CAT expression nor insulin/PMA-induced CAT expression were detected with the basic XMB vector (Fig. 6) or with a fusion gene construct in which a single copy of the Ϫ97/Ϫ64 oligonucleotide had been ligated into the polylinker of the XMB vector (data not shown). By contrast, the multimerized Ϫ97/Ϫ64 oligonucleotide was able to mediate a stimulatory effect of both insulin and phorbol esters on fusion gene expression, although the effect of insulin was small relative to that of phorbol esters (Fig. 6).
We next investigated the ability of multimerized oligonucleotides (Table I), containing mutations in either the Ets or AP-1 motifs, to mediate an insulin or phorbol ester response in the context of the XMB vector (Fig. 6). Mutation of the Ets motif reduced both basal fusion gene expression (Fig. 6C) and the stimulatory effect of phorbol esters (Fig. 6B), whereas mutation of the AP-1 motif abolished the stimulatory effect of phorbol esters (Fig. 6B) and greatly reduced basal gene expression, despite the presence of an intact Ets motif (Fig. 6C). The effects of these mutations on both basal fusion gene expression and the response to phorbol esters mimic the effects of these same mutations when made in the context of the endogenous collagenase promoter (Fig. 5, B and C). By contrast, mutation of the Ets motif in the heterologous context of the XMB vector had only a minor effect on the ability of insulin to stimulate expression of the fusion gene, whereas the same mutation in the context of the collagenase promoter markedly reduced the insulin response (Fig. 5A). The explanation for this discrepancy is unclear but may indicate that in this context, the presence of multiple AP-1 motifs reduces the requirement for the Ets motif with respect to insulin, although not phorbol ester signaling. This is consistent with the observation that the mechanisms of insulin and phorbol esters signaling through the AP-1 motif are distinct (see below). Mutation of the AP-1 motif in the heterologous context of the XMB vector abolished the effect of insulin (Fig. 6A), as did the same mutation in the context of the collagenase promoter (Fig. 5A).
Both Insulin and Phorbol Esters Signal through the Collagenase AP-1 Motif-The preceding experiments demonstrate that the AP-1 motif is required for the action of both insulin and phorbol esters on collagenase gene transcription (Fig. 5) as well as for mediating an effect of these agents on the expression of a heterologous XMB fusion gene through the Ϫ97/Ϫ64 collagenase sequence (Fig. 6). In the collagenase promoter the Ets motif enhances the action of both agents, although it is inactive TABLE I Sequence of oligonucleotides used in these studies All nucleotide positions are relative to the collagenase gene transcription start site at ϩ1 (33). The collagenase AP-1 and core Ets motifs are boxed. Wild-type (WT) and mutated (MUT) sequences are shown in uppercase and lowercase letters, respectively. The oligonucleotide representing the collagenase sequence between Ϫ97 and Ϫ64 was synthesized with either BamHI (GATC)-or HindIII (AGCT)-compatible ends. Both the BamHI-and HindIII-compatible ends contribute to the collagenase sequence such that the oligonucleotide with BamHI ends actually represents the collagenase sequence between Ϫ98 and Ϫ64, whereas the oligonucleotide with HindIII ends actually represents the collagenase sequence between Ϫ97 and Ϫ61.
FIG. 6. The collagenase promoter sequence between ؊97 and ؊64 can confer a stimulatory effect of both phorbol esters and insulin on the expression of a heterologous Xenopus-CAT fusion gene. HeLa cells were transiently co-transfected, as described under "Experimental Procedures," with expression vectors encoding ␤-galactosidase and the insulin receptor and the basic XMB vector or constructs in which oligonucleotides representing the wild-type (WT) or individually mutated (MUT) Ets (E) and AP-1 (A) motifs in the collagenase promoter sequence from Ϫ97 to Ϫ64 had been ligated into the HindIII site of the Xenopus promoter in multiple (4 -5) copies. Following transfection, cells were incubated for 24 h in serum-free medium in the presence or absence of 10 nM insulin or 100 nM PMA. The cells were then harvested, and both CAT and ␤-galactosidase activity were assayed as described previously (20,30). Results are presented as the ratio of CAT activity, corrected for protein concentration in the cell lysate, in insulin-treated (A) or PMA-treated (B) versus control cells, and are expressed as fold induction. In C, results are presented as the ratio of CAT to ␤-galactosidase activity in control cells and are expressed as arbitrary units. Results represent the mean of Ϯ S.E. of three experiments, in which each construct was assayed in duplicate.
in the absence of the AP-1 motif (Fig. 5). However, it is unclear from these experiments whether the Ets motif-binding protein can also directly mediate an insulin/phorbol ester response or whether it just acts as an accessory factor to enhance the effects of these agents mediated through the AP-1 motif. To address this question, oligonucleotides representing just the individual collagenase Ets or AP-1 motifs (Table I) were ligated into the polylinker of the XMB vector (Fig. 7). Neither basal CAT expression nor insulin/PMA-induced CAT expression were detected with fusion gene constructs containing single copies of the Ets and AP-1 oligonucleotides (data not shown). However, the multimerized AP-1 motif was able to mediate an induction of fusion gene expression by both insulin and phorbol esters, whereas the multimerized Ets motif was unable to confer a response to either agent (Fig. 7). Mutation of the AP-1 motif in this context abolished the induction of XMB fusion gene expression (Fig. 7), just as it did in the context of the collagenase-CAT fusion gene (Fig. 5).
Protein Binding to the Collagenase Ets and AP-1 Motifs-The specific binding of nuclear extract proteins to the collagenase AP-1 motif was analyzed using the gel retardation assay (Fig. 8). Two broad protein-DNA interactions were detected when a double-stranded oligonucleotide representing collagenase promoter sequence from Ϫ78 to Ϫ63 (Table I), which con-tains the AP-1 motif, was used as the labeled probe (Fig. 8A). A 100-fold molar excess of the unlabeled Ϫ78/Ϫ63 oligonucleotide competed effectively against the labeled probe for binding of the upper complex, indicating that this protein-DNA complex represents a specific interaction (Fig. 8A). By contrast, an oligonucleotide representing the same Ϫ78 to Ϫ63 collagenase sequence, but with a mutation in the AP-1 core sequence ( Table  I) that abolishes insulin and phorbol ester signaling through this element (Fig. 7), failed to compete for protein binding with the labeled probe, even in a 100-fold molar excess (Fig. 8A). Thus, the formation of this specific protein-DNA complex correlates with the effect of insulin and phorbol ester mediated through this sequence in the context of the collagenase promoter and the heterologous XMB vector. This specific protein-DNA complex detected in the gel retardation assay contains members of the AP-1 transcription factor family (Ref. 13 and see below).
We also analyzed protein binding to an oligonucleotide representing the wild-type collagenase Ets motif (Table I) but failed to detect a protein-DNA interaction, the formation of which was selectively competed for by an excess of the unlabeled wild-type oligonucleotide and not by an oligonucleotide containing the same mutated Ets sequence that abolished the effects of insulin and phorbol esters mediated through this element in the context of the collagenase promoter (Fig. 8B, left  panel). This result contrasts with those of White et al. (42), who were able to detect specific binding to the collagenase Ets motif. We would speculate that this discrepancy may reflect a difference in the amount or nature of the Ets family transcription factors expressed in HeLa cell nuclear extracts, as used here, and fibroblast nuclear extracts, as used by White et al. (42).
A number of investigators have previously shown that DNA binding by Ets proteins is repressed by an intramolecular mechanism (43-45) that can be relieved through partial proteolysis (45). To determine whether a similar phenomenon was preventing the detection of specific protein binding to the wildtype Ets oligonucleotide, HeLa nuclear extract was preincubated with 10 ng of chymotrypsin prior to analysis of protein-DNA binding using the gel retardation assay. Using this strategy, a protein-DNA interaction was now detected, the formation of which was selectively competed for by an excess of the unlabeled wild-type oligonucleotide and not by an oligonucleotide containing the same mutated Ets sequence that abolished the effects of insulin and phorbol esters mediated through this element in the context of the collagenase promoter (Fig. 8B, right panel, arrow). Additional gel retardation experiments with an oligonucleotide containing both the collagenase AP-1 and Ets motifs, using HeLa nuclear extract without chymotrypsin treatment, failed to reveal the presence of a protein-DNA interaction that might be indicative of the formation of a ternary complex, although specific binding to the AP-1 motif was detected (data not shown). Thus, the presence of AP-1 did not appear to be sufficient to relieve the repression of Ets protein binding.
The Mechanisms of Insulin and Phorbol Ester Signaling through the Collagenase AP-1 Motif Are Distinct-The magnitude of insulin-and phorbol ester-stimulated collagenase-CAT fusion gene transcription are markedly different, with phorbol ester being the more potent activator (Figs. [2][3][4][5]. This difference in potency was maintained when the effects of insulin and phorbol esters were compared on the expression of heterologous constructs containing multimerized oligonucleotides representing either the collagenase promoter sequence between Ϫ97 and Ϫ64, which contains both the Ets and AP-1 motifs (Fig. 6), or collagenase promoter sequence between Ϫ78 to Ϫ63, which contains just the AP-1 motif (Fig. 7). This suggests that the FIG. 7. The collagenase AP-1 motif, in contrast to the Ets motif, can confer a stimulatory effect of both phorbol esters and insulin on the expression of a heterologous Xenopus-CAT fusion gene. HeLa cells were transiently co-transfected, as described under "Experimental Procedures," with expression vectors encoding ␤-galactosidase and the insulin receptor and XMB vector constructs in which oligonucleotides representing the wild-type collagenase Ets motif (E(WT)) or the wild-type or mutated collagenase AP-1 motif (A(WT) and A(MUT), respectively) had been ligated into the HindIII site of the Xenopus promoter in multiple (3)(4)(5) copies. Following transfection, cells were incubated for 24 h in serum-free medium in the presence or absence of 10 nM insulin or 100 nM PMA. The cells were then harvested and CAT activity was assayed as described previously (20,30). Results are presented as the ratio of CAT activity, corrected for protein concentration in the cell lysate, in insulin-treated (A) or PMA-treated (B) versus control cells, and are expressed as fold induction. Results represent the mean of Ϯ S.E. of three or four experiments, in which each construct was assayed in duplicate. difference in potency of insulin and phorbol esters on collagenase gene transcription can be at least partly ascribed to a difference in signaling through the AP-1 motif.
Because AP-1 is composed of homodimers and heterodimers of various members of the Fos and Jun transcription factor families (25), we hypothesized that the difference in potency of insulin and phorbol ester signaling might be due to differences in the ability of these factors to selectively induce the binding of specific members of the Fos/Jun families. To test this hypothesis, nuclear extracts were first prepared from HeLa cells incubated for 5 h in the presence or absence of insulin and phorbol esters, and then polyclonal antisera to specific Fos/Jun family members were assayed for their ability to supershift the specific protein-DNA complex detected using the collagenase AP-1 motif in gel retardation assays (Fig. 9). These experiments showed that although both insulin and phorbol esters induced AP-1 binding activity, several differences were apparent in the composition of the induced complex. Most notably, phorbol esters stimulated a marked increase in c-Fos binding and a slight increase in Fra-1 binding (Fig. 9B), whereas insulin was without effect on these proteins (Fig. 9A). By contrast, insulin induced the expression of Fra-2 and Jun D (Fig. 9A), whereas phorbol esters only had a minor effect on the binding of these factors (Fig. 9B). Because phosphatase inhibitors were not present during the preparation of nuclear extracts, it is likely that these changes in AP-1 binding reflect changes in AP-1 expression rather than phosphorylation. Although insulin induces c-Fos mRNA in multiple cell types, this effect is transient (13), which may thus explain the lack of insulin-stimulated c-Fos binding (Fig. 9A). DISCUSSION This analysis of the cis-acting elements that mediate the stimulatory effects of insulin and phorbol esters on collagenase gene transcription reveals that multiple elements are required for the action of both agents. The results obtained with respect to phorbol ester signaling are similar to those reported by Brinckerhoff and colleagues (reviews in Refs. 1 and 37) in their studies on the regulation of the rabbit collagenase Ϫ1 gene in fibroblasts. Although several of these elements are required for the effects of both insulin and phorbol esters, it is apparent that there are also unique features to their mechanisms of action. Thus, phorbol ester-stimulated collagenase-CAT gene transcription was reduced when the region of the promoter between Ϫ518 to Ϫ159 was deleted, whereas the effect of insulin was unchanged (Fig. 4). In addition, deletion of the Ϫ97 to Ϫ80 sequence was more deleterious to the phorbol ester response than a site-directed mutation of the Ets motif in the context of the Ϫ97 to ϩ64 promoter fragment, whereas the insulin response was equally reduced by both manipulations (Fig. 5). Perhaps the most striking difference, though, was the observation that the mechanisms of insulin and phorbol ester signaling through the AP-1 motif appear to be distinct (Fig. 9). Thus, these agents induce selective increases in the binding of specific members of the Fos/Jun transcription factor families; most notably, phorbol esters stimulated a marked increase in c-Fos binding (Fig. 9B), whereas insulin was without effect (Fig. 9A). We hypothesize that these selective effects on AP-1 binding may explain the greater potency of phorbol esters in stimulating collagenase gene transcription, although the existence of additional differences in insulin-and phorbol ester-stimulated changes in AP-1 phosphorylation cannot be excluded.
We first became interested in the regulation of collagenase gene transcription by insulin as a result of earlier studies on the regulation of ME gene transcription (13). Those experiments demonstrated that whereas insulin clearly stimulated ME-CAT gene transcription in H4IIE cells, it had little effect on ME-CAT gene transcription in HeLa cells (13). The action of insulin on ME gene transcription in H4IIE cells was shown to be mediated, in part, through an AP-1 motif in the ME promoter (13). Because insulin stimulates collagenase-CAT gene transcription in HeLa cells (13) (Fig. 1) and because the stimulatory effect of insulin on collagenase gene transcription requires an intact AP-1 motif in the collagenase promoter (12) two questions arose, namely (i) why does insulin stimulate collagenase-CAT expression, but not ME-CAT expression, in HeLa cells given that both promoters contain AP-1 motifs and (ii) why does insulin stimulate ME-CAT gene transcription in H4IIE cells but not in HeLa cells? We postulated that the latter may reflect an inherent difference in insulin-signaling through AP-1 motifs in the H4IIE and HeLa cell lines, whereas the answer to the first question could reflect some inherent difference between the AP-1 motifs in the ME and collagenase promoter (13). Alternatively, we proposed that both observations might be explained by a difference in the promoter context in which the AP-1 motif was located; in other words, the nature of the transcription factors associated with the promoter other FIG. 8. Specific protein-DNA complexes are detected in gel retardation assays using the collagenase AP-1 and Ets motifs as the labeled probes. Protein binding to labeled oligonucleotide probes, representing either the collagenase AP-1 motif sequence between Ϫ78 and Ϫ63 (A) or the Ets motif sequence between Ϫ97 and Ϫ76 (B), was analyzed using HeLa cell nuclear extracts and the gel retardation assay, as described under "Experimental Procedures." The labeled probes were incubated in the absence (Ϫ) or presence of a 100-fold molar excess of unlabeled oligonucleotides representing either the wild-type (WT) or mutant (MUT) collagenase Ϫ78/Ϫ63 AP-1 sequence (A) or Ϫ97/Ϫ76 Ets sequence (B) prior to the addition of nuclear extract. In B, nuclear extract was pretreated with or without 10 ng of chymotrypsin. Representative autoradiographs are shown. In A, the AP-1 complex and a nonspecific (NS) protein-DNA interaction are indicated by the arrows, whereas in B, the specific protein-DNA interaction revealed by chymotrypsin treatment is indicated by the arrow. FP, free probe. than AP-1 (13). The data presented in this paper demonstrates that the latter explanation is at least part of the answer. Thus, whereas the AP-1 motif is required, the marked stimulation of collagenase-CAT gene transcription by insulin in HeLa cells is mediated through the interaction of multiple collagenase promoter elements and their associated factors. By analogy, the presence of an AP-1 motif in the ME promoter would be predicted to be insufficient by itself for the induction of ME-CAT gene transcription by insulin in HeLa cells. The question of why insulin stimulates ME-CAT gene transcription in H4IIE cells but not in HeLa cells remains to be answered. We have previously shown that phorbol esters also fail to induce ME-CAT fusion gene expression in HeLa cells and that there is an inherent difference in the ability of the collagenase and ME AP-1 motifs to mediate a phorbol ester response in a heterologous context (13). However, it is apparent that the marked stimulation of collagenase-CAT gene expression by phorbol esters is also dependent on multiple elements in addition to the AP-1 motif (Figs. 4 and 5).
Based on the data presented in Figs. 5 and 6, we propose that the transcription factor binding the collagenase Ets motif acts as accessory factor to enhance the effect of insulin mediated through the AP-1 motif. Such an arrangement would be similar to that found in the glucose-6-phosphatase promoter in which HNF-1␣ acts as an accessory factor to enhance the effect of insulin mediated through a more proximal IRS (46). Interestingly, Ets and AP-1 motifs are found juxtaposed in several promoters, including those encoding collagenase (41,47), matrilysin (48), and tissue inhibitor of MMP-1 (49) genes and the polyoma virus enhancer (50). In these genes, the Ets and AP-1 motifs mediate a synergistic activation of transcription in response to various agonists (for review, see Ref. 2). In addition, it has been shown that Ets proteins are capable of directly interacting with components of the AP-1 complex in vitro (49,51,52) and in vivo (51,52), thus providing a molecular model to explain the ability of Ets motif-binding proteins to enhance signaling through AP-1 motifs.
Although the multimerized collagenase Ets motif fails to mediate an insulin response when ligated to the heterologous XMB vector (Fig. 7), the possibility that insulin directly signals through the Ets binding factor as well as through AP-1 cannot be entirely ruled out. Thus, Blackshear and colleagues (22) have demonstrated that at least in some cell lines, the stimulatory effect of insulin on c-fos gene transcription is mediated by MAP kinase through the phosphorylation of the Ets motifbinding protein, p62 TCF . p62 TCF only binds the c-fos serum response element as a ternary complex with p67 SRF (23). By analogy, the possibility therefore exists that insulin does directly signal through the collagenase Ets motif-binding protein, but that this cannot be demonstrated in a heterologous context because AP-1 is required to stabilize binding of the Ets factor (22). MAP kinase also phosphorylates the Ets protein GABP␣, which mediates the stimulatory effect of insulin on prolactin gene transcription (53), although the significance of this phosphorylation is unclear because recent data has shown that PD98059 does not block this effect of insulin (54).
In summary, the experiments described in this report demonstrate that multiple promoter elements are required for the stimulatory effect of insulin on collagenase gene transcription in HeLa cells. The AP-1 motif in the collagenase promoter appears to be the target of both insulin and phorbol ester signaling; however, the mechanisms of insulin and phorbol ester action are distinct. Future studies will examine whether insulin also regulates collagenase gene expression in glomerula-derived cell lines. Although insulin receptor levels are low in the mesangial cells of the glomerulus (55), they increase in the diabetic state (56). Furthermore, it is interesting to note that in the streptozotocin rat model of type I diabetes, collagenase gene expression in the glomerulus decreases (10). This may indicate that the regulation of collagenase gene expression in this tissue by phorbol esters differs from that reported in HeLa cells because the hyperglycemia in these animals has been postulated to lead to activation of protein kinase C (57), which would be anticipated to lead to an increase, rather than a decrease, in collagenase gene expression. decreased collagenase-1 gene expression in the pathophysiology of glomerulosclerosis. HeLa and hamster insulinoma tumor cells were kindly provided by Roland Stein and Eva Henderson. Data analysis was performed in part through the use of the Vanderbilt University Medical Center Cell Imaging Resource (CA68485 and DK20593).