UTP induces osteopontin expression through a coordinate action of NFkappaB, activator protein-1, and upstream stimulatory factor in arterial smooth muscle cells.

Osteopontin (OPN) is an important chemokinetic agent for several cell types. Our earlier studies have shown that its expression is essential for uridine triphosphate (UTP)-mediated migration of vascular smooth muscle cells. We demonstrated previously that the activation of an AP-1 binding site located 76 bp upstream of the transcription start in the rat OPN promoter is involved in the induction of OPN expression. In this work, using a luciferase promoter deletion assay, we identified a new region of the rat OPN promoter (-1837 to -1757) that is responsive to UTP. This region contains an NFkappaB site located at -1800 and an Ebox located at -1768. Supershift electrophoretic mobility shift assay and chromatin immunoprecipitation assays identified NFkappaB and USF-1/USF-2 as the DNA binding proteins induced by UTP, respectively, for these two sites. Using dominant negative mutants of IkappaB kinase and USF transcription factors, we confirmed that NFkappaB and USF-1/USF-2 are involved in the UTP-mediated expression of OPN. Using a pharmacological approach, we demonstrated that USF proteins are regulated by the extracellular signal-regulated kinase (ERK)1/2 pathway, just as the earlier discovered AP-1 complex, whereas NFkappaB is up-regulated through PKCdelta signals. Finally, our work suggests that the UTP-stimulated OPN expression involves a coordinate regulation of PKCdelta-NFkappaB, ERK1/2-USF, and ERK1/2/NAD(P)H oxidase AP-1 signaling pathways.

Its expression is induced during vessel remodeling, as well as wound healing and bone synthesis (1). Overexpression of OPN can be found in pathological settings, such as immunological disorders, neoplastic transformation, metastasis, and formation of urinary stones. Its role in cell migration has been established in vitro in various cell types, including osteoclasts (2), macrophages (3), endothelial cells (4), and SMCs (5,6). In vivo studies have established that OPN contributes to SMC recruitment into atherosclerotic plaques (7) and neointimal thickening (8), to osteoclast recruitment during bone resorption, (2) and to macrophage recruitment at inflammatory sites (atherosclerotic plaque, granulomas) (3).
In the context of vascular remodeling, soluble factors, such as angiotensin II, basic fibroblast growth factor, plateletderived growth factor, interleukin-1 (9), and the nucleotides ATP and UTP (10), all induce expression of OPN. Moreover, nucleotides also modulate contraction, proliferation, and migration of vascular SMCs (11). They act through G proteincoupled P2Y receptors (12) and activate phospholipase C, resulting in elevated levels of intracellular free Ca 2ϩ and activation of PKC. In addition, they induce ROS production through NAD(P)H oxidase (13). Several transcription factors, such as NFAT, AP-1, cAMP-responsive element-binding protein, serum-responsive factor, STAT, and NFB are induced in UTP-stimulated SMCs (14), controlling the expression of a set of genes including OPN (10). Moreover, we have demonstrated previously that nucleotide-induced SMC migration is dependent on OPN expression and on its binding to the ␣ v ␤ 3 integrin (6).
The UTP-mediated increase of OPN protein expression results from enhanced gene transcription and stabilization of its transcript (15), but the way chemotactic factors regulate OPN expression remains poorly defined. Vitamin D receptor (16), estrogen receptor ␣ (17), Smad (18), Cbfa (19), and Ets-1 (20) have been shown to regulate OPN expression in osteoclasts, whereas metastasis-associated transcription factor (21), Tcf-1, Ets-1, Ets-2, polyomavirus enhancer A-binding protein-3, c-Jun (22) and acute myeloid leukemia-1 (23) are involved in OPN expression in cancer cells. In contrast, with respect to SMCs, only a few transcription factors have been identified. It has been found that USF-1 is associated with increased OPN levels during cell differentiation (24) and that AP-1 is involved in glucose- (25) and UTP-mediated (15) OPN induction.
In this study, we undertook to identify enhancer elements, transcription factors, and signaling pathways that mediate the UTP-induced expression of OPN in SMCs. Through a systemic analysis of the OPN promoter, we show that, in addition to AP-1, NFB and USF transcription factors also contribute to UTP-induced OPN expression. We describe the involvement of two parallel signal transduction pathways, one involving ERK1/2 activation leading to Ebox and AP-1 activation and the other operating via PKC␦ and leading to the activation of NFB. Moreover, we demonstrate that AP-1 can be activated via ROS production.

MATERIALS AND METHODS
Cell Culture and Transfection-Rat aortic SMCs were prepared from the thoracic aortas of Wistar rats and cultured as described previously (26). SMCs from passages 10 to 20 were used. Quiescent SMCs were obtained after a 72-or 24-h (for gene reporter assay) incubation period in serum-free Dulbecco's modified Eagle's medium.
SMCs were transfected in 380-mm 2 wells using SuperFect reagent (Qiagen). Cells were co-transfected with 0.4 g of pGL 2 basic vectors (Promega) containing rat OPN promoter fragments and 0.4 g of pHook-LacZ vector (Invitrogen) carrying the ␤-galactosidase gene. For experiments using plasmids encoding IB or dominant negative mutants of USF, IKK (IB kinase) or PKC␦ cells were co-transfected with 0.27 g of each plasmid. Empty pcDNA3 vector (Invitrogen) was used as a control.
Luciferase and ␤-galactosidase activities were assayed as described previously (15). For each sample, luciferase activity was normalized to ␤-galactosidase activity to compensate for differences in transfection efficiency. Each construct was assayed in triplicate in each experiment, and each experiment was performed at least three times.
Plasmid Constructs-The rat OPN promoter (GenBank TM accession number AF017274) cloned into the pGL 2 basic vector (-1994luc) was kindly provided by Dr. A. Ridall (27). Luciferase reporter plasmids containing fragments of different lengths of the 5Ј-region of the rat OPN promoter (Ϫ1599luc, Ϫ1240luc, Ϫ1051luc, Ϫ551luc, Ϫ294luc, and Ϫ96luc) have been described previously (15). Ϫ1994⌬-1599luc and Ϫ1994⌬-1837luc plasmids were obtained by cloning the KpnI-NsiI and the KpnI-PstI fragments of the Ϫ1994luc plasmid into the multicloning site of the pGL 2 promoter vector, respectively. Fragments (Ϫ1757 to Ϫ1599 and Ϫ1670 to Ϫ1599) were obtained by PCR amplification using the forward primers 5Ј-CGTGCTTAAAGGGCAGAAAG-3Ј and 5Ј-ATATTCGATAGTCACAGGTG-3Ј, respectively, and the same reverse primer chosen within the pGL 2 vector, 5Ј-TTAGGTAACCCAGTA-GATCC-3Ј. Ϫ1994⌬-1599luc plasmid was used as a template. PCR products were cut by XhoI and then inserted between the SmaI and XhoI sites of the pGL 2 promoter vector.
Western Blot Analysis-Whole cell extracts and immunoblottings were carried out as described previously (6). The monoclonal antibody MPIIIB10 used for OPN detection (Developmental Studies Hybridoma Bank), anti-Erk1/2, and anti-phospho-Erk1/2 (Cell Signaling Technology) were all three used at a 1:1000 dilution. ␣-Tubulin was used to control equal sample loading.

Two Regions of the OPN Promoter Contain UTP-responsive
Elements-To identify transcription factors involved in UTPinduced OPN expression, we first identified cis-regulating elements in the OPN promoter that were activated by UTP. Fig. 1 shows that the 4.5-fold UTP induction of the Ϫ1994 OPN promoter activity was partially lost when the Ϫ1994 to Ϫ1599 region was deleted. Moreover, the residual activity was completely lost after deletion of the Ϫ96 to ϩ66 OPN promoter sequence, pointing to two regions containing UTP-responsive elements in the OPN promoter.
In a previous study, we showed that UTP activation of the Ϫ96 to ϩ66 region was mediated by the AP-1 site located at Ϫ76 (15). UTP-responsive elements contained in the newly characterized Ϫ1994 to Ϫ1599 region were located neither in the Ϫ1994 to Ϫ1837 nor in the Ϫ1757 to Ϫ1599 regions (Fig.  1B). Consequently, we hypothesized that the region Ϫ1837 to Ϫ1757 contained this element.
NFB and Ebox Are the UTP-responsive Elements of the Ϫ1837 to Ϫ1757 Region of the OPN Promoter-Computer analysis of the Ϫ1837 to Ϫ1757 region revealed several potential cis-regulated sequences, including STAT (Ϫ1805), NFB (Ϫ1800), NFAT (Ϫ1799), AP-4 (Ϫ1786), Sry (Ϫ1775), and Ebox (Ϫ1768) binding sites. The binding capacities of these sites were explored by EMSA analyses. Two probes were used. The first one (Ϫ1806 to Ϫ1789) contains STAT, NFB, and NFAT potential sites, and the second (Ϫ1788 to Ϫ1759) contains AP-4, Sry, and Ebox sites ( Fig. 2A). When nuclear extracts of UTP-stimulated cells were incubated with the 32 P-labeled probe 1, one complex was detected by EMSA (Fig. 2B). This complex disappeared when an excess of the specific cold probe 1 was added but not when an excess of a nonspecific cold probe (AP-1) was added. Moreover, the complex also disappeared FIG. 1. Localization of UTP-responsive region in the rat OPN promoter. SMCs were transfected with plasmids containing successive deletions of the OPN promoter cloned into pGL 2 basic vector (A) or fragments of the Ϫ1994 to Ϫ1599 region cloned into the pGL 2 promoter vector (B). After a 24-h incubation in serum-free medium, transfected SMCs were stimulated (or not) with 100 mol/liter UTP for 6 h. The cell lysates were assayed for luciferase activity. Relative luciferase activity was calculated taking pGL 2 vector activity as 1. The ratio of stimulated versus unstimulated activities is shown. when a cold probe containing only the NFB site (but not the STAT or NFAT sites) was added (Fig. 2B). These data demonstrate that the NFB site located at Ϫ1800 is specifically recognized by transcription factors in SMCs. Using the same technique, nuclear extracts incubated in the presence of 32 P-labeled probe 2 revealed the formation of three complexes (Fig. 2B). The addition of cold AP-4, Sry, and Ebox probes in excess in the assay showed that only the Ebox located at Ϫ1768 (Fig. 2B) is involved in the binding of these complexes.
The binding of factors to NFB and Ebox sites were altered upon UTP stimulation but with different kinetics (Fig. 3). Binding of factors to the NFB site peaked at 15 min (Fig. 2C), whereas binding of factors to Ebox was maximum after 2 h of UTP stimulation (Fig. 2C).
To confirm the role of the NFB (Ϫ1800), Ebox (Ϫ1768), and AP-1 (Ϫ76) sites, cells were transfected with the Ϫ1994luc construct mutated on these sites. UTP-stimulated promoter activity was reduced by 58, 43, or 45% when the AP-1, Ebox, or NFB sites were mutated, respectively. These results demonstrate that, in addition to the AP-1 site (Ϫ76), Ebox (Ϫ1768) and NFB sites (Ϫ1800) are also involved in UTP-induced OPN transcription.
p65 Mediates the NFB Site Activation of the OPN Promoter-Supershift assays using anti-p65 antibody were performed to identify transcription factors binding the NFB site. When anti-p65 antibody was added to nuclear extract, a supershift band was observed (Fig. 3A). In contrast, anti-c-Fos antibody did not induce a supershift. Moreover, ChIP assay demonstrated that OPN promoter DNA was co-immunoprecipitated together with p65 (Fig. 3B). Altogether, these results demonstrate that p65 NFB binds the site located at Ϫ1800 in nuclear extracts as well as in living cells.
To confirm the role of NFB in the regulation of the OPN promoter, NFB activity was inhibited by preventing IB degradation by ammonium pyrrolidinedithiocarbamate or by overexpressing either IB or a dominant negative IKK mutant. In these conditions, UTP activation of the Ϫ1994luc construct was reduced by 48, 44, or 39%, respectively (Fig. 3C). In contrast, UTP activation of the Ϫ1994⌬NFBluc construct was not sig-

FIG. 3. p65 binds the NFB site of the OPN promoter.
A, SMCs were stimulated with 10 mol/liter UTP for 15 min or 2 h to perform EMSA analysis with 32 P-labeled probes 1 or 2, respectively. The indicated antibodies were added to nuclear extracts to produce supershift. Anti-c-Fos antibody was used as a nonspecific antibody with probe 1, and anti-p65 was used with probe 2. B, ChIP assay with anti-p65, anti-USF-1, or anti-c-Myc antibodies. SMCs were stimulated with 10 mol/liter UTP during 15 min or 2 h in p65 or USF-1 ChIP assay, respectively. -Ab, control experiment done without antibody. Total cell extract was used as a positive control of the PCR. C and D, SMCs were transfected with the Ϫ1994luc, or the Ϫ1994⌬NFB, or the pGL 2 basic plasmids and co-transfected with the IB-encoding plasmid (C), or the dominant negative IKK-expressing plasmid, or with the dominant negative encoding plasmid A-USF (D). Empty pcDNA3 was used as a control. Quiescent SMCs were stimulated (or not) with 10 mol/liter UTP for 6 h. SMCs were pretreated with 100 mol/liter ammonium pyrrolidinedithiocarbamate for 30 min before adding UTP. Co-transfection with pGL 2 basic and empty cDNA3 vectors was taken for reference (n ϭ 1). nificantly affected. These results indicate that NFB is involved in UTP-induced OPN expression.
USF-1/USF-2 Mediates the Ebox Activation of the OPN Promoter-A supershift band was observed when anti-USF-1 or anti-USF-2 (but not anti-c-Myc or anti-p65) antibodies were added to nuclear extract (Fig. 3A). ChIP assays demonstrated that OPN promoter DNA was co-immunoprecipitated with anti-USF-1 antibodies but not with anti-c-Myc antibodies (Fig.  3B). The role of USF in UTP-induced OPN expression was assessed using the USF dominant negative A-USF. In the presence of A-USF, UTP-activation of the Ϫ1994luc construct was reduced by 42% but not that of the Ϫ1994⌬Eboxluc construct (Fig. 3C). Altogether, these results demonstrate that USF-1/USF-2 binds to the site located at Ϫ1768 in nuclear extracts as well as in living cells and are unambiguously involved in UTP-induced OPN expression.
PKC␦ Is Involved in UTP-induced OPN Expression-To explore the intracellular pathways involved in UTP-induced OPN expression, we first focused on PKC. The role of PKC in UTP signaling via P2Y 2 , P2Y 4 , or P2Y 6 receptors is well established (32). Fig. 4A shows that PKCs are involved in UTP-induced OPN expression, because UTP is unable to induce OPN expression in cells depleted of PKC by a 16-h phorbol 12-myristate 13-acetate treatment.
To identify which PKC isotypes were involved in UTP-induced OPN expression, we used a set of selective PKC inhibitors. Gö6976 inhibits PKC␣ and PKC␤1 (IC 50 Ͻ 10 nM) but does not affect PKC␦, -⑀, and -activities (33,34). Bisindolylmaleimide I is a highly selective inhibitor of PKC␣, -␤ (IC 50 Ͻ 20 nM), -, -␦, and -⑀ (IC 50 Ͻ 100 nM) (35). LY379196 selectively inhibits PKC␤ isoenzymes I and II (IC 50 Ͻ 0.1 mol/ liter), and rottlerin is a highly selective inhibitor for PKC␦ (IC 50 3-6 M) (36). Western blot analysis showed that Gö6976, bisindolylmaleimide I, and LY379196 had no effect (Fig. 4, B and C), whereas rottlerin totally inhibited UTP-induced OPN expression at 3 mol/liter (Fig. 4D). Moreover, SMC transfection with a plasmid encoding dominant negative PKC␦ induced a marked decrease of the reporter OPN gene activity in response to UTP (Fig. 4E). We further show that PKC␦ inhibition did not modulate UTP-induced ERK1/2 phosphorylation (Fig.  4F). All these experiments strongly suggest that PKC␦ is an important actor in UTP-mediated OPN expression and that this pathway does not involve ERK1/2 activation. (37), we investigated whether ROS produced by NAD(P)H oxidase could be involved in UTP-induced OPN expression. Fig. 5A shows that OPN protein content was strongly decreased when UTP-stimulated SMCs were treated with the NAD(P)H oxidase inhibitor DPI, whereas in the same conditions, ERK1/2 phosphorylation was not impaired. These results suggested that the ROS produced by NAD(P)H oxidase were required for UTP-induced OPN expression but not for ERK1/2 activation (Fig. 5B). Taken together, our previous (6) and present data suggest that ERK1/2, PKC␦, and NAD(P)H oxidase were all three involved in UTPinduced OPN expression.

NAD(P)H Oxidase Is Involved in UTP-induced OPN Expression-Because ROS have been shown to be involved in angiotensin II-induced OPN expression
Role of ERK1/2, PKC␦, and NAD(P)H Oxidase in UTP-induced AP-1, USF, and NFB Activation-Involvement of ERK1/2, PKC␦, and NAD(P)H oxidase in UTP-induced AP-1, USF, and NFB activation was assessed using specific inhibitors (U0126, rottlerin, and DPI, respectively). Fig. 6, A and B, shows that UTP-induced AP-1 activation was reduced in the presence of U0126 and DPI but not in the presence of rottlerin, indicating that ERK1/2 and NAD(P)H oxidase (but not PKC␦) were involved in UTP-induced AP-1 activation. EMSA analysis performed in parallel experiments showed that NFB activation was inhibited only in the presence of rottlerin (Fig. 6, A  and B), indicating that PKC␦ (but neither ERK1/2 nor NAD(P)H oxidase) was involved in UTP-induced NFB activation. These data were confirmed by the entire inhibition of UTP-induced NFB activation when the endogenous PKC␦ was inhibited by transfection of a plasmid encoding a dominant negative PKC␦ mutant (Fig. 6C). In contrast, EMSA analysis revealed that UTP-induced USF activation was inhibited by U0126 but neither by rottlerin nor by DPI (Fig. 6, A and C), demonstrating that ERK1/2 but not PKC␦ nor NAD(P)H oxidase was involved in UTP-induced USF activation. DISCUSSION In the present study, we showed that a new region of the rat OPN promoter (Ϫ1837 to Ϫ1757) is involved in UTP-induced

FIG. 5. NAD(P)H oxidase is involved in UTP-induced OPN expression.
Quiescent SMCs were pretreated with NAD(P)H oxidase inhibitor DPI (10 mol/liter) for 30 min before UTP stimulation (10 mol/liter). A, OPN expression was analyzed by Western blot after a 6-h incubation period. Equal loading was verified by ␣-tubulin (␣-tub) detection. B, ERK1/2 phosphorylation was analyzed by Western blot after a 30-min incubation period. Equal sample loading was verified by total ERK1/2 revelation. OPN expression in arterial SMCs. Until now, two large regions of the OPN promoter have been involved in OPN regulation; the Ϫ1000 to Ϫ700 region is mainly a target for hormones (16,38), whereas the Ϫ270 to ϩ1 region has been shown to be activated by various transcription factors, including AP-1 (15). AP-1 seems to be a common element in the activation of OPN expression by growth factors, because it has been involved in FGF-2-and TGF␤-induced OPN expression in osteoblasts (39,40). Moreover, cis-regulating elements (Lef-1, Ets, and c-Jun) located upstream of Ϫ1000 were involved in the basal expression of OPN in breast cancer cells (22). These data suggest that this distal region could be involved in OPN expression in transformed cells or induced by specific growth factors.
In this work, we showed that the distal region of the rat OPN promoter contains a functional NFB cis-regulator site. The sequence of this site is highly conserved in mouse and human, suggesting a key role of this region in OPN gene regulation. Because OPN has been shown to induce itself, NFB activation (41), an amplification loop leading to OPN expression, can be hypothesized.
This work also demonstrated the involvement of a new Ebox at Ϫ1768 on the rat OPN promoter. Homologous sequences were found in equivalent regions of human and mouse OPN promoters. Two other Ebox were described previously in the OPN promoter. The first one, located at Ϫ106 on the rat OPN promoter, is associated with SMC differentiation (24) and glucose signaling (42) and is conserved in the mouse and human promoter, although its function has not been demonstrated in these species. A second Ebox, located at Ϫ81 in the mouse promoter (25) and at Ϫ84 in the human promoter (43), was also shown to be involved in glucose signaling (25). According to our data, none of these sites are involved in UTP-induced OPN expression. Thus, this newly identified Ebox seems to be specifically activated by the chemotactic factor UTP (6). OPN mRNA accumulation is detected as early as 2 h after UTP stimulation and peaks after 6 h. NFB activity is maximal 15 min after UTP stimulation, whereas USF (this study) and AP-1 (15) activation are maximal after a 2-h UTP stimulation. These data suggest a coordinate mechanism for the regulation of OPN transcription. One possible scenario is the following. NFB, which does not require protein neosynthesis, could be involved in the rapid onset of OPN transcription, whereas AP-1 activation (requiring transcription) and USF induction (which needs co-activator expression) (28) could occur in a second phase of the response and take over the regulation of OPN transcription.
Downstream signaling events that followed UTP stimulation are well documented in various cell types (32). A general consensus suggests that purinergic receptor P2Y 2 activation mediates Ca 2ϩ increase, RhoA activation, and tyrosine phosphorylation of multiple proteins, including mitogen-activated protein kinase, focal adhesion kinase, Shc, and translocation of several PKC isotypes. Our data suggest that OPN expression is dependent on the PKC␦ isotype. Although several studies report that PKC␦ mediates ERK1/2 activity (44), our data suggest that ERK1/2-mediated OPN expression is independent of PKC␦ but acts through the activation of the NFB transcription factor.
Because ROS play a major role in the NFB pathway, we verified the possible involvement of NAD(P)H oxidase. We showed that this enzyme is not involved in UTP-induced NFB activation but mediates the AP-1-dependent regulation of the OPN gene. The role of NAD(P)H oxidase in the ERK1/2 transduction pathway is unclear, because studies demonstrated that NAD(P)H oxidase induced ERK1/2-dependent (45) or -inde- FIG. 6. Identification of signaling pathways involved in AP-1, NFB, and USF transcription factor activation. A, for EMSA analysis of AP-1, NFB, and USF-1/USF-2, quiescent SMCs were pretreated with 10 mol/liter U0126, 5 mol/liter rottlerin, 10 mol/liter DPI or left untreated for 30 min before UTP stimulation (10 mol/liter) (2 h for AP-1 and USF-1/USF-2 and 15 min for NFB). Binding activity was revealed using the AP-1 probe, the NFB probe 1 or the Ebox probe in the respective assays. Ϫ, without inhibitors. B, quantification of the signal density of three experiments was determined by image analysis (Scion Image). Time 0 was taken as the reference (n ϭ 1). C, SMCs were co-transfected with either NFB reporter plasmid, its control pGL 2 promoter vector, AP-1 reporter plasmid, USF reporter plasmid or their control pGL 3 promoter vector, and with either dominant negative PKC␦ (PKC␦ DN) encoding plasmid or empty pcDNA3. After a 24-h incubation period, SMCs transfected with NFB, AP-1, or NFB reporter plasmids were stimulated (or not) with 10 mol/liter UTP for 2 or 6 h, respectively. Histograms present the ratio of luciferase activities of UTPstimulated cells versus quiescent cells. Co-transfections with control pGL 2 promoter or pGL 3 promoter and empty pcDNA3 vectors were taken as the reference (n ϭ 1). pendent (46) transduction pathways in SMCs. We showed that the NAD(P)H-mediated AP-1 activation is independent of ERK1/2 activation, as described previously in angiotensin IIand endothelin-stimulated SMCs in which NAD(P)H oxidase activates mitogen-activated protein kinase, p38, and/or JNK (46,47). Altogether, this work characterizes two new regulatory elements on the OPN promoter and suggests that UTP activates two parallel pathways in a coordinated manner in SMCs (Fig. 7), one involving the PKC␦-NFB cascade controlling the early OPN transcription and the other controlling late OPN transcription through ERK1/2-AP-1/USF regulation.