Activation of protein kinase C in human uterine smooth muscle induces connexin-43 gene transcription through an AP-1 site in the promoter sequence.

Myometrial connexin-43 gap junctions are scarce throughout gestation but appear in large numbers at term to facilitate contractions during labor. The mechanisms that regulate this process are incompletely characterized. This report investigates the effects of protein kinase C activation on the regulation of connexin-43 gene transcription in human uterine smooth muscle cells. In primary myometrial cells treated with phorbol ester, transient increases in c-Fos and c-Jun protein levels were observed at 2-4 h, followed by significant increases in connexin-43 protein levels at 6-8 h. Nuclear run-on transcription analysis showed an increase in connexin-43 transcription 3 h after phorbol ester treatment. AP-1 sites were identified in the sequence of the 5'-flanking promoter region of the human connexin-43 gene at 44 and 1000 base pairs upstream of transcription start. Transcription from a reporter plasmid containing the proximal human connexin-43 promoter was increased in transfected primary cultures treated with phorbol ester. Mutation of the proximal AP-1 site in the promoter abolished the phorbol ester-dependent transactivation. This work provides evidence that transcription of the human connexin-43 gene is induced through protein kinase C activation in uterine smooth muscle cells, and that the induction involves up-regulation and activation of c-Jun and c-Fos.

Connexins compose a family of transmembrane proteins, which form hexameric hemi-channel structures in cell membranes. When hemi-channels of adjacent cells attach, an intercellular channel is created, which allows for the passive diffusion of ionic mediators and secondary messengers of less than about 1,500 daltons (Da) and establishes direct communication between connected cells via such small molecules. A specialized region of the membrane containing connexin channels is called a gap junction (1). Gap junctions allow for coordinate function of the cells in a tissue.
Several mammalian connexins cDNAs have been sequenced (2)(3)(4), and the proteins, named according to their respective molecular weights, appear to be tissue-specific. The primary connexin in cardiac muscle, osteoblasts, granulosa cells of the ovary, and vascular and uterine smooth muscle is connexin-43 (Cx43) (43,000 Da) (3,(5)(6)(7). Additionally, during mouse fetal development, temporal and spatial patterns of connexin-43 gene (cx43) expression occur (8). There is little information concerning the mechanisms that control connexin-43 gene expression or that govern tissue-specific differences in gene expression.
In adult tissues such as the heart muscle and bone, cx43 expression is relatively stable. In myometrium, Cx43 gap junctions are not evident except just before parturition when connexin-43 (cx43) mRNA levels increase dramatically (9). Cx43 gap junctions subsequently become abundant in myometrium (10 -12) and are believed to play an essential role in successful deliveries by synchronizing contractions during labor through the passage of molecules such as [Ca 2ϩ ] (13)(14)(15).
The mechanisms that regulate the sparsity of myometrial gap junctions during pregnancy and their abundance during parturition are unknown. Prior to labor, myometrial cx43 expression is regulated in part by the steroid hormones, estrogen, and progesterone (10,13,16). Progesterone suppresses cx43 gene transcription and inhibits the trafficking of Cx43 protein through the Golgi apparatus (12,16). Estrogen can induce premature cx43 expression and gap junction formation in the myometrium of pregnant animals (14,17). However, cx43 expression only remains high in estrogen-treated myometrial cells if internal uterine pressure is maintained after delivery (10). These data indicate that estrogen acts with a factor(s) produced through uterine distension to influence myometrial cx43 gene expression. It is plausible that such a factor or factors are bio-reactive moieties derived from phospholipid breakdown upon extension of the smooth muscle membrane.
Diacylglycerol is an early product of inositol phospholipid breakdown in the cell membrane. Diacylglycerol is the major activator of protein kinase C (PKC) 1 (18). The phorbol ester 12-O-tetradecanoylphorbol-13-acetate (TPA) also activates PKC. The PKC signal transduction cascade causes activation of cell-specific transcription factors, which in turn bind cognate sequence elements in gene promoters and modulate gene expression. Several cis-acting elements in the DNA have been identified, which regulate gene expression after activation of * This work was supported in part by National Institutes of Health (NIH) Grant HD 30482 (to J. A.), a grant from the March of Dimes Birth Defects Foundation (to R. E. G.), and NIH Grant HL02391 (to G. I. F.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U64573.
The AP-1 site is the cognate binding sequence for the Jun and Fos families of transcription factors (20,21). Jun and Fos, products of "immediate-early" genes, induce the expression of the "late" genes which contain AP-1 sites. C-Jun and c-Fos are the members of the families which are primarily expressed in uterine smooth muscle cells (22)(23)(24). In general, activation of PKC results in increased expression of jun and fos genes and AP-1-mediated transcription is stimulated.
In this report, the role of c-Jun and c-Fos in cx43 gene regulation in uterine smooth muscle was investigated using primary cultures of human uterine smooth muscle cells composed of nonpregnancy myometrial cells, as well as uterine leiomyoma cells, myometrial-like benign tumors which express many estrogen-regulated genes expressed in pregnancy myometrium (see Ref. 25 for discussion). Upon culturing, nonpregnancy myometrial and leiomyoma primary cells begin to express Cx43 protein and the expression can be suppressed by progestin (16). The present study shows that TPA treatment causes the primary cells to transiently increase expression first of c-Fos and c-Jun proteins followed by increased expression of Cx43 protein. The study suggests that up-regulation of myometrial cx43 gap junction expression at the onset of labor may involve the activation of c-Fos and c-Jun through the PKC pathway.

MATERIALS AND METHODS
Cell Culture System-Approval to utilize human tissue was given by Stony Brook University Hospital's Committee on Research Involving Human Subjects (95-1224). Human myometrial and leiomyoma tissues were obtained from surplus pieces of surgical specimens after hysterectomies. Matched myometrium and leiomyoma tissues were processed and cultured as described by Zhao et al. (16) and Andersen et al. (6). The cultures were never passaged and were used within 2 weeks of the initial plating. The source of the tissues dictated and generally limited the number of culture dishes that could be prepared for analysis (generally 10 -20 culture plates for one experiment).
Twenty-four hours prior to the experiments, the culture medium was changed to serum-free, phenol red-free (26) Dulbecco's modified Eagle's medium supplemented as described (27). All experiments with the primary cultures were performed using this defined serum-free medium. TPA (Calbiochem-Novabiochem Corp., La Jolla, CA) was added to the cultures at a concentration of 100 ng/ml. Control samples were treated with reagent vehicle Me 2 SO at 0.03% final concentration.
Immunoblot Analysis-The relative levels of Cx43, c-Fos, c-Jun, AP-2, and intermediate filament desmin proteins were measured by immunoblot analysis. Preparation of cell lysates and immunoblot analysis using enhanced chemiluminescence (Amersham Life Science) were performed as described previously (6,27). The mouse monoclonal antibody used to detect Cx43 protein was purchased from Zymed Laboratories, Inc. (San Francisco, CA) and the mouse monoclonal antibody used to detect desmin intermediate filament protein (55,000 Da) was purchased from Sigma. Antibodies specific for c-Fos (SC-52), c-Jun (SC-822), and AP-2 (SC-184) proteins were obtained from Santa Cruz Biotechnologies, Inc. (Santa Cruz, CA). The levels of c-Jun and c-Fos were analyzed, since these forms of Fos and Jun predominate in myometrial tissue (22)(23)(24). Quantitation was by densitometry of resulting autoradiographs using the Bio-Rad laser densitometer and analysis by density volume using the Molecular Analysis Program. The densities of desmin were used to normalize the values of Cx43, c-Jun, and c-Fos proteins.
Sequencing the Human cx43 Promoter-Restriction fragments of pGF-10 (4) were subcloned into pBluescript (Stratagene, Inc.). The human 5Ј-flanking cx43 promoter was sequenced independently using an automated sequencer as well as manually using Sequenase DNA polymerase and the Sanger dideoxy method (28). The two sequences were compared to each other and to the published rat cx43 promoter sequence (29) using the Hibio DNAsis Program (Hitachi Software Engineering America Ltd., Brisbane, CA).
Nuclear Run-on Transcription Analysis-The medium of the cultures was switched to the serum-free medium 24 h before the experiment, and all experiments were done in serum-free medium as described by Zhao (16). Five 10-cm plates with confluent primary cultures were analyzed per condition. Cultures were treated with 100 ng/ml TPA or reagent vehicle for 3 h before harvesting. Slot blot filters were prepared and used in the nuclear run-on analyses as described (16) using plasmids containing the cDNAs for human cx43, rat fos, and rat gapd (see Table  I). Plasmids were purified on CsCl gradients (33). Analysis of Transcription Start-A 18-mer oligonucleotide (5Ј-TTT-GGTACTTTCCTCACG-3Ј) was made, which is complementary to a region of cx43 mRNA within exon 1 (positions 51-69). The oligonucleotide was purified using an Applied Biosystems Oligonucleotide Purification Cartridge following the manufacturer's directions. Sixty-five pmol of the oligonucleotide were kinased with 300 Ci of [␥-32 P]ATP using Escherichia coli polynucleotide kinase as described previously (27) and purified on a G-25 Sephadex spin column (Boehringer Mannheim). One pmol of the unlabeled oligonucleotide was used for the corresponding Sanger sequencing reactions using 3 g of pGF-10, which contains the cx43 gene (32). Total RNA was isolated from the cultured myometrial and leiomyoma cells according to the procedure of Chomczynski and Sacchi (34). The transcription start was determined by the primer extension method as described (33) using 20 g of total RNA of each type. The final sequencing gel was dried onto Whatman paper and subjected to autoradiography.
Mutation of AP-1 Site-Mutation of an AP-1 site to 5Ј-TCGGCCG-3Ј has been shown by others to silence the function of the site (35). The proximal AP-1 in the cx43 promoter was mutated using the Kunkel method (36), and the mutagenic oligonucleotide 5Ј-TCCCAGTTCGGC-CGGTGGCT-3Ј (changed bases are underlined). A 420-bp HindIII DNA fragment of the proximal cx43 promoter was subcloned into M13mp18 vector using standard protocols (33). The protocols described by Ausubel et al. (37) were used to create the mutant AP-1 site. The HindIII fragment containing the mutant AP-1 site was subcloned into the chloramphenicol acetyltransferase (CAT) vector, pCXV1-CAT 3 . The final mutant clone was confirmed by DNA sequencing.
Transient Expression Assays-Reporter plasmids containing portions of the 5Ј-proximal cx43 promoter cloned upstream of the bacterial CAT gene (pBL-CAT 3 ) were used in transient expression assays (see Table I). A cryptic AP-1 site has been reported in the backbone of such vectors (38). This site was deleted by digesting pBL-CAT 3 with EcoO109 and NdeI, filling-in the digested ends with Klenow enzyme, and religating (33).
The primary cultures were transfected with 40 g of reporter plasmid/10-cm plate using the calcium phosphate precipitation method as described previously (27). Expression vectors for c-jun or c-fos were co-transfected with the CAT reporter plasmids at 0.5 g/10-cm plate. Transfected cells were treated with either 100 ng/ml TPA or reagent vehicle for 48 h before obtaining cell lysates. Lysates were normalized for protein concentration and analyzed for CAT activity as described previously (27).
Statistical Analysis-Statistical analysis of the results was made with GraphPad's Instat computer software for small sample sets using a two-tailed Student's t test, or with GraphPad's Prism software using two way ANOVA analyses.

AP-1 Sites Are Present Upstream of the cx43 Transcription
Start-The human cx43 gene transcription start has been determined for cardiac muscle (32). To determine the human cx43 transcription start in myometrial and leiomyoma primary cells, primer extension assays were performed using total RNA purified from myometrial and leiomyoma cells. The RNAs had previously been shown to contain cx43 mRNA through Northern blot analysis (16). The size of the myometrial and leiomyoma cx43 mRNA was found to be similar to the size found in heart muscle, about 3,000 bases (16). The transcription start ( Fig. 1) was found to be 2 base pairs upstream of the start in cardiac muscle (32). The sequence of more than 3000 base pairs of the 5Ј-flanking portion of the human cx43 promoter was determined by two independent methods (see Fig. 2A). The human cx43 promoter sequence is very similar to the the rat cx43 promoter sequence (29) having a 68.6% homology. The transcription start and first exon of the gene have been marked on the sequence ( Fig. 2A). Several canonical cis-acting elements are observed in the 5Јflanking sequence including two AP-1 sites (5Ј-TGAGTCA-3Ј) 44 bp and (5Ј-TGACTCA-3Ј) 1000 bp upstream of the transcription start determined for myometrial and leiomyoma cells (see Figs. 1 and 2A). Several putative cis-acting elements, which may be important to PKC or steroid hormone regulation of the gene, have been marked in the sequence. Three AP-2 sites (at Ϫ93, Ϫ133, and Ϫ1499) and a trimer of SP-1 sites (40) (at Ϫ1459 to Ϫ1388) can be found upstream of the transcription start. The AP-1 and AP-2 sites in similar positions to transcription start have been noted in the rat cx43 promoter (29). No complete estrogen response elements (EREs) or progesterone response elements (PREs) could be found in the 5Ј-flanking region of this gene. However, the sequence contains six 5Ј-GGTCA (at Ϫ420, Ϫ455, Ϫ702, Ϫ2068, Ϫ2528, and Ϫ2793), which may be half-palindromic ERE sites, and a 5Ј-TGTTCT (at Ϫ1746) and 5Ј-AGAACA (at Ϫ1690), which may be halfpalindromic PRE sites. Half-EREs and half-PREs have also been noted in the rat cx43 promoter sequence (29).
Activation of PKC Results in Increased cx43 Expression in Myometrial Cells-Transcription regulated through AP-1 and AP-2 sites are induced through the PKC cascade. To investigate the role of PKC activation in the regulation of myometrial cx43 gene expression, primary human uterine smooth muscle cells were treated with the phorbol ester TPA, an activator of protein kinase C. The relative levels of Cx43, desmin, c-Fos and c-Jun proteins were then determined by immunoblot analysis (Fig. 3). The levels of Cx43, c-Jun, and c-Fos, normalized to the intermediate filament protein desmin, gradually increase with time in the vehicle-treated cultures after the change to serumfree medium. The increase in Cx43 protein with time in serumfree medium is consistent with previous observations (16). The reasons for these increases are presently unknown. When compared to the time control samples, the amount of Cx43 protein increased significantly in cultures treated for 6 -8 h with TPA (n ϭ 3, p Ͻ 0.0001) (Fig. 1B). The increase in Cx43 protein levels followed transient increases at 2-4 h in the levels of c-Fos and c-Jun proteins, the transcription factors that bind AP-1 sites (Fig. 3C). In contrast, the cell-specific, 49.5-kDa transcription factor AP-2, which can also be induced through TPA treatment, was not detected (data not shown). Two forms of c-Jun were observed in the myometrial cells after TPA treatment: a 39-kDa form, which is a dormant form, and a slower mobility 46-kDa form, which presumably is an active form due to phosphorylation of serines 63 and 73 (39). Both forms increased with time after TPA treatment, indicating both an increase in c-jun gene expression and activation of Jun protein.
After 18 h of TPA treatment, the levels of Cx43, c-Jun, and c-Fos proteins approached the corresponding levels in the control samples without TPA (Fig. 3, B and C). These data suggest that activation of PKC in human primary myometrial cells results in increased gene expression of both c-fos and c-jun and the increased fos/jun expression is associated temporally with increased Cx43 protein levels.
Cell lysates of nonpregnancy myometrial tissues from different stages of the menstrual cycle and of pregnancy myometrial tissue from a woman who had been in active full-term labor were also subjected to immunoblot analysis to test for the presence of Cx43, c-Fos, c-Jun, and AP-2 proteins. We were able to detect c-Fos in the nonpregnancy myometrial tissues (especially in tissues from estrogen-dominated follicular phase), but could not detect c-Jun, AP-2, or Cx43 proteins (data not shown). In contrast, we were able to detect Cx43, c-Fos and c-Jun (at 46 kDa) proteins in the pregnancy myometrium (Fig.  3A, lane marked P) but not AP-2. This suggests that c-Jun and c-Fos (and not AP-2) play a role in the induction of cx43 in pregnancy myometrium at term and that the same regulatory pathway is exhibited in the cultured myometrial cells.
To test if the TPA-induced increases in myometrial Cx43 protein levels were the result of increased gene transcription, nuclear run-on transcription analyses were performed. Nascent radiolabeled nuclear transcripts were made from fresh nuclei isolated from primary human myometrial cells either treated with TPA or reagent vehicle, and were hybridized to specific cDNAs immobilized on nitrocellulose membrane (Fig.  4). Relative to transcription from the glyceraldehyde-3-phosphate dehydrogenase (gapd) gene, a dramatic increase in de novo transcription from the cx43 and the c-fos genes was evident in myometrial cells after 3 h of TPA treatment (Fig. 2, lane  marked TPA). These data indicate that activation of PKC induces increased cx43 gene transcription in human primary myometrial cells.
Transcription of the cx43 Gene Is Induced through the AP-1 Site following Activation of PKC-Transient expression assays were employed to further investigate the role of PKC induction in cx43 gene expression. The 5Ј-proximal region of the human cx43 promoter has the most transcriptional activity in cardiac myocytes (32) and in primary uterine smooth cells (data not shown). For analysis of the most active portions of the promoter, several restriction fragments containing the 5Ј-proximal cx43 gene promoter were cloned into CAT reporter plasmid (see Fig. 2B) from which a reported cryptic AP-1 sequence had been removed (see "Materials and Methods"). A CAT reporter plasmid containing an 870-bp restriction fragment of the cx43 gene from the EcoRV site 700 bp upstream of transcription start to a created BglII (32) site 170 bp downstream of the transcription start was primarily used in this study (see pCXV1-CAT 3 in Fig.  2B). This portion of the promoter contains the natural transcription start, two of the half-palindromic EREs, two of the AP-2 sites, the proximal AP-1 site, and most of exon I. Transient expression assays using reporter construct pCXHB-CAT 3 (see Fig. 2B), which contains 580 bp of the cx43 5Ј-flanking promoter region and lacks the two half-palindromic EREs produced similar results (data not shown) to those using pCXV1-CAT 3 .
Primary myometrial and leiomyoma cells were co-transfected with the CAT reporter construct pCXV1-CAT 3 and/or expression vectors for c-Jun and c-Fos. Subsequently the cultures were either treated with TPA or reagent vehicle for 2 days, the amount of time needed for effective transient transfections of the primary cultures. CAT activity in each cell lysate was determined and interpreted as reflecting the transcriptional activity from the reporter plasmids. We did not see a difference between the responses of myometrial and leiomyoma cells (Fig. 5). The CAT activity in lysates from transfected primary cells that were not treated with TPA was the same whether or not they were co-transfected with c-jun and c-fos expression vectors. Increases in CAT activity were observed in the lysates of TPA-treated cells that were not co-transfected with expression vectors for c-jun or c-fos; however, the increases were not statistically significant (p ϭ 0.125; n ϭ 5). Significant increases in CAT activity were observed for cultures treated with TPA when the cells were co-transfected with either a c-fos expression vector (p Ͻ 0.05) or a c-jun expression vector (p Ͻ 0.003) or both (p Ͻ 0.008) (Fig. 5B). The necessity to enhance c-fos or c-jun expression to obtain statistically significant results may reflect natural negative feedback regula- FIG. 4. Analysis of transcriptional response of cx43 and c-fos genes to TPA treatment. Nuclear run-on transcription assays were performed with nuclei from primary myometrial cells treated for 3 h with reagent vehicle (0) or with 100 ng/ml TPA (TPA). Radiolabeled nascent transcripts purified from cells that underwent each treatment were hybridized to a molar excess of indicated cDNA probes immobilized onto nitrocellulose membrane strips: human cx43, human c-fos, and rat gapd cDNAs. tion of c-fos and/or c-jun expression with time after TPA treatment (39,40). Decreases in c-Fos and c-Jun levels are observed in myometrial cells treated with TPA for 18 h and are similar to the levels in untreated cells (see Fig. 3). This suggests that a negative feedback mechanism exists, which diminishes the levels of the transcription factors despite continued TPA treatment.
Mutation of the proximal AP-1 site to 5Ј-TCGGCCG-3Ј results in loss of TPA-induced transcription (Fig. 5C). No in-crease in CAT activity was detected in the lysates from cultures treated with TPA if they were co-transfected with c-fos and/or c-jun expression vectors. These results clearly establish that the AP-1 site in the most proximal region of the human cx43 promoter has a role in inducing transcription upon activation of PKC in myometrial and leiomyoma primary cells.

DISCUSSION
Cx43 gap junctions play a crucial role in the function of the myometrium during labor by propagating action potentials during muscle contractions. The steroid hormones regulate cx43 expression in myometrial tissue during pregnancy, but it is evident that other factors such as uterine distention are also involved in the regulation. Here we show that after treatment with TPA, primary uterine smooth muscle cells exhibit a dramatic increase in levels of c-Fos and c-Jun proteins, the factors that bind to AP-1 sites. The increases in c-Jun and c-Fos are followed by an increase in Cx43 protein levels. Expression of c-Jun and c-Fos is negligible in myometrial tissue from nonpregnant women (this study and Ref. 43) and is further suppressed in the myometrium of women in late pregnancy (43). We are able to detect c-Jun and c-Fos proteins in active-labor myometrial tissue, which also expressed high levels of Cx43 protein. We show too that TPA induces increased transcription from the cx43 gene. Results from transient expression assays clearly demonstrate that the proximal AP-1 site in the human cx43 promoter is functional in inducing transcription in primary cultures of uterine smooth muscle cells upon activation of PKC. This work then supports the concept that up-regulation of cx43 expression in human myometrium at parturition is in part induced by c-Jun and c-Fos binding to the AP-1 site after activation of PKC. Activation of PKC may occur through release of diacylglycerol during uterine distention.
The 5Ј-proximal promoter region of the cx43 gene also contains two AP-2 sites just upstream of transcription start. The cognate AP-2 transcription factor is expressed in a tissuespecific manner; is regulated by retinoic acid, PKC, and cAMP (19,42); and is induced upon activation of PKC in cells that express this factor (44). Its expression is developmentally regulated during embryogenesis (45,46). We found no evidence that AP-2 expression can be induced in myometrial or leiomyoma primary cells through TPA treatment. We could not detect this protein in pregnancy myometrium. Additionally, mutation of the AP-1 site abrogates TPA-induced transcription in the presence of intact AP-2 sites. The AP-2 site in the cx43 5Јproximal promoter does not appear to regulate cx43 expression in myometrial cells after activation of PKC.
Two unique regulatory elements, which bind protein factors found in the nuclear extract of several cell lines, have been identified in the 5Ј-proximal promoter of the mouse cx43 gene (47). Homologous sequences can be found in the human cx43 promoter in similar positions but were not analyzed in the present study. An activator element (5Ј-TCCTAGCCCC) was located at Ϫ72 to Ϫ62 and appears to be needed for basal transcriptional activity. A repressive element (5Ј-CCTC-CCCGCC) was located from Ϫ102 to Ϫ92, which appears to inhibit transcription in an analysis of the most 5Ј-proximal region of the mouse cx43 promoter. It is perplexing, however, that the repressive element binds to nuclear proteins from active-labor rat pregnancy myometrium, which has high cx43 expression, and not to nuclear proteins from prelabor pregnancy myometrium, which has suppressed cx43 expression (see Ref. 47). Additionally, the element was not repressive when larger regions of the promoter were analyzed. In contrast, in the human cx43 promoter region, the 5Ј-proximal region of the promoter is the most active compared to larger pieces of the promoter in cardiac myocytes, which express cx43 (32). This FIG. 5. Induction of CAT activity with TPA in primary culture cells. A, representative autoradiograph of a thin-layer chromatography plate on which monoacetylated [ 14 C]chloramphenicol products (MAP) have been separated from the unacetylated substrate (US). The CAT activities shown are from lysates of leiomyoma primary cultures transfected with pCXV1-CAT 3 and/or expression vectors for c-fos and/or c-jun, and treated with 100 ng/ml TPA or with reagent vehicle as indicated at the bottom of the autoradiograph. B, bar graph summarizing the results for five transfections each of myometrial and leiomyoma cultures as shown in A. All transfections were with pCXV1-CAT 3 and were treated with TPA, except that the one marked 0 did not receive TPA. J, F, and J/F indicate co-transfection with expression vectors for c-jun, c-fos, or both, respectively. Standard deviations are shown by the error bars. The results from co-transfection with pCXV1-CAT 3 and expression vectors for c-fos and c-jun without TPA treatment were the same as for the transfection marked 0 and have not been included in the bar graph. C, representative autoradiograph as in panel A, except cells were transfected with pV1-AP1 mt -CAT 3 , which contains the mutant AP-1 site. In the last lane on the right, the cells were co-transfected with pERE/TK/CAT and an expression vector for human ER (see Ref. 28) and were treated with 10 nM estrogen as a positive control. part of the promoter is also the most active (data not shown) in the primary uterine smooth muscle cells, which also express cx43. It is logical that cx43 promoter activity is cell typespecific. Analyses of the cx43 promoter may differ depending on whether the cells or cell lines are actively expressing the cx43 gene.
Interestingly, the repressive element identified in the mouse cx43 promoter sequence resembles an SP-1 element, although binding of its cognate protein(s) was not significantly competed for with a consensus SP-1 binding site oligonucleotide (47). SP-1 sites can be bound by repressive members of the SP-1 family and have been shown to be necessary to attenuate transcription of the fos gene (48). A trimer of putative SP-1 sites appears 1388 -1459 bp upstream of transcription start in the human cx43 promoter. It is possible that the SP-1 sites are involved in repressing cx43 promoter activity in the cardiac myocytes and in the uterine smooth muscle cells.
Animal models indicate that during pregnancy, myometrial gap junction expression is up-regulated by estrogen and downregulated by progesterone (see Ref. 13 for review). Estrogen and progesterone activate nuclear receptors, which are transcription factors themselves. The activated receptors bind as dimers to enhancer hormone response elements (HREs). HREs can function in a number of positions relative to the gene they regulate, but usually appear in the 5Ј-flanking region of the promoter. Unexpectedly, the sequence of the 5Ј-flanking cx43 promoter (this study) did not contain full EREs or PREs. Several putative ERE and PRE half-palindromic sites were identified several hundred base pairs upstream of the gene. An estrogen-responsive half-palindromic ERE (5Ј-TGGGTCA) found in the ovalbumin gene has been shown to be a target for TPA-induced c-Jun/c-Fos transactivation independent of the binding of the estrogen receptor (ER) to the ERE (49). Our results suggest that the two half-palindromic EREs closest to transcription start are not involved in the TPA transactivation of the cx43 gene, since we see the same TPA induction with or without the half-palindromic EREs. The roles of these halfsites in the estrogen regulation of cx43 expression need to be investigated.
It is known that progesterone suppresses transcription of the human cx43 gene (16); however, the mechanism by which estrogen controls human cx43 expression has remained elusive. Yu et al. (29) reported that the 5Ј-flanking promoter of the rat cx43 gene contains estrogen-responsive elements. They observed striking induction of ER-mediated transcription in HeLa cells co-transfected with an expression vector for ER; however, paradoxically they did not observe the same induction in the ER-containing rat osteosarcoma cell line, ROS 17/2.8, even when co-transfected with an ER expression vector. Perhaps the reason for the different results between the two cell lines lies in different endogenous HeLa and ROS transcription factors. Paradoxically, HeLa cells do not normally express cx43 and Ros cells do (7). Thus, the difference in results cannot be accounted for by the presence of factors which repress cx43 expression in the ROS cells. This suggests that ER needs to interact with certain positive factors to induce transcription of the cx43 gene, and these are present in HeLa cells but not ROS cells.
In steroid-responsive tissues, cross-talk has been shown to occur between steroid hormone regulation and the regulation of AP-1 activity (50 -55). For example, expression of c-jun and c-fos is regulated by sex steroids in a tissue-specific manner (22)(23)(24). The functional AP-1 site in the human cx43 promoter opens the possibility that transcriptional regulation of the cx43 gene by the steroid hormones in female reproductive tissues may involve the Fos/Jun transcription complex. The role of potential interactions of c-Jun/c-Fos with the steroid hormone receptors in the regulation of cx43 expression needs to be explored.