The Expression of novH in Adrenocortical Cells Is Down-regulated by TGFβ1 through c-Jun in a Smad-independent Manner*

The human NOV secreted glycoprotein (NOVH) is abundant in the fetal and adult adrenal cortex. The amount of NOVH increases in benign adrenocortical tumors and decreases in malignant adrenocortical tumors, suggesting that NOVH plays a role in tumorigenesis in the adrenal cortex. Transforming growth factor β1 (TGFβ1), fibroblast growth factor 2 (FGF2), and insulin growth factors (IGFs) play crucial roles in the physiology of the adrenal cortex. We investigated the effects of these factors on the expression of novH in the NCI H295R adrenocortical cell line. The amounts of NOVH protein and novH transcripts were down-regulated by TGFβ1 and up-regulated by FGF2, whereas IGFs had no effect. Furthermore, the TGFβ1-dependent inhibition ofnovH promoter activity was completely abrogated following site-directed mutation of two activating protein (AP-1) sequences (positions −473 and −447), whereas the stimulatory effect of FGF2 was not affected. Co-transfection with dominant negative forms of c-Jun and MEKK1 also abrogated novH-targeted regulation by TGFβ1, whereas the overproduction of Smad proteins or dominant negative forms of Smad had no effect. Taken together, these results suggest that c-Jun and MEKK1 signaling but not Smad signaling are involved in the TGFβ1-dependent decrease in NOVH in NCI H295R cells. In conclusion, our data provide evidence that novH is a new target of TGFβ1; unlike other members of the CCN ( c yr61, c tgf, n ov) family, however, its expression is repressed rather than induced.

From ‡INSERM U515 and §INSERM U482, Hôpital Saint-Antoine, 75571 Paris Cedex 12,France The human NOV secreted glycoprotein (NOVH) is abundant in the fetal and adult adrenal cortex. The amount of NOVH increases in benign adrenocortical tumors and decreases in malignant adrenocortical tumors, suggesting that NOVH plays a role in tumorigenesis in the adrenal cortex. Transforming growth factor ␤1 (TGF␤1), fibroblast growth factor 2 (FGF2), and insulin growth factors (IGFs) play crucial roles in the physiology of the adrenal cortex. We investigated the effects of these factors on the expression of novH in the NCI H295R adrenocortical cell line. The amounts of NOVH protein and novH transcripts were down-regulated by TGF␤1 and up-regulated by FGF2, whereas IGFs had no effect. Furthermore, the TGF␤1-dependent inhibition of novH promoter activity was completely abrogated following site-directed mutation of two activating protein (AP-1) sequences (positions ؊473 and ؊447), whereas the stimulatory effect of FGF2 was not affected. Co-transfection with dominant negative forms of c-Jun and MEKK1 also abrogated novH-targeted regulation by TGF␤1, whereas the overproduction of Smad proteins or dominant negative forms of Smad had no effect. Taken together, these results suggest that c-Jun and MEKK1 signaling but not Smad signaling are involved in the TGF␤1dependent decrease in NOVH in NCI H295R cells. In conclusion, our data provide evidence that novH is a new target of TGF␤1; unlike other members of the CCN (cyr61, ctgf, nov) family, however, its expression is repressed rather than induced.
Relatively little is known about the functions of NOV. However, several reports have suggested that it up-or down-regulates cell proliferation, depending on the cell type, (3,12,13). There are also several lines of evidence indicating that NOV is involved in cell adhesion. Indeed, the multidomain structure of NOV and the other CCN proteins suggests that they bind to components of the extracellular matrix, including heparin-like oligomers (5). The finding that fibulin 1C, an extracellular matrix-associated protein (14,15), interacts with the human NOV protein (NOVH) (16) suggests that NOVH has a role in signaling pathways involving the extracellular matrix. It was recently shown that recombinant NOV can promote the adhesion of vascular smooth muscle cells in vitro and that changes in nov expression occur following injury to the arterial walls (13).
In normal tissues, the expression of nov is tightly regulated during the development of the central nervous system (17,18) and skeletal and visceral muscles (18) and during chondrogenesis. 2 novH is highly expressed in the adrenal cortex during embryogenesis, and in adults novH is more strongly expressed in the adrenal cortex than in other endocrine tissues (19). Thus, novH may play an autocrine/paracrine role in the development and/or differentiation of these tissues.
Interestingly, the expression of novH is altered in several human tumors, including Wilms' tumors (4, 10) and adrenocortical tumors (19). In adrenocortical tumors, which have a very poor prognosis (20,21), quantitative and qualitative changes in novH expression are correlated with the acquisition of the tumoral phenotype by adrenocortical tissue (19). Significant differences have been detected in the concentrations of NOVH and novH mRNA in benign and malignant tumors. Furthermore, the NOVH protein profiles are different in the two types of tumor, suggesting that novH plays a role in the early stages of tumorigenesis. The enhanced expression of novH in benign tumors may contribute to the benign phenotype by increasing cell adhesion, whereas the lower expression of novH in malignant tumors could be involved in cell invasiveness (19). Alternatively, the down-regulation of novH in malignant adrenocortical tumors suggests that novH could act as a tumor suppressor. This hypothesis is supported by the inverse corre-lation between tumorigenicity and novH expression in glioma cells (22) and by the fact that the ectopic expression of novH in glioma cells reduces their tumorigenicity in xenografts (23). 3 Therefore, we decided to investigate the molecular mechanisms responsible for the alterations in novH expression in tumoral adrenocortical cells. For this purpose, in the present study we used the human NCI H295R cell line, which is derived from a human adrenocortical carcinoma that produces steroids (24,25). A number of growth factors and cytokines such as epidermal growth factor (EGF), transforming growth factor-␣ (TGF␣), tumor necrosis growth factor (TNF␣), interleukins (26 -28), insulin-like growth factors (IGFs) (for a review see Refs. 26 and 29), fibroblast growth factor 2 (FGF2) (26), and transforming growth factor ␤1 (TGF␤1) (for a review see Refs. 26 and 29) regulate adrenal growth and functions in normal and fetal adrenal glands. The concentration of IGF-II, which also plays a role in adrenocortical tumorigenesis (24,30,31), is inversely correlated to novH expression in several malignant adrenocortical tumors, suggesting that IGF-II regulates the expression of novH or vice versa (19). Moreover, FGF2 and TGF␤1 can induce the production of other members of the CCN family such as CTGF and CYR61 in fibroblasts and in some epithelial cells (32)(33)(34)(35)(36).
Thus, we examined whether novH expression is affected by IGFs, FGF2, and TGF␤1 in NCI H295R cells. We showed that the expression of novH is not modulated by IGFs but is upregulated by FGF2 and down-regulated by TGF␤1. These regulations occur at the transcriptional level. Further studies indicated that two AP-1 consensus binding sites (Ϫ473 and Ϫ447) within the novH promoter play a crucial role in TGF␤1 regulation but not in the stimulatory effect of FGF2. Finally, we provide evidence that c-Jun and MEKK1, but not Smad, can mediate the TGF␤1-dependent decrease of novH expression.
Site-directed Mutagenesis-Either one or both of the AP-1 consensus sites present at positions Ϫ473 (GGTGACAAACT) and Ϫ447 (CATGAC-TAAC) (Fig. 7A) of p625NH-Luc were changed to TGctgAA using a two-step PCR strategy. Both strands of all the constructs were fully sequenced (Genome Express, Grenoble, France) to confirm the mutations before use.

Cells
Cell Culture-NCI H295R cells (ATCC) were maintained in Dulbecco's modified Eagle's/F12 medium supplemented with 2% Ultroser G (Invitrogen), 5 g/ml insulin, 5 g/ml transferrin, and 5 ng/ml sodium selenite (Sigma), 200 units/ml penicillin, 200 g/ml streptomycin, and 2.5 mM L-glutamine. Primary cultures of mouse astrocytes were obtained as follows. Brains from 3-4-day-old mice were dissected and crushed in minimum Eagle's medium. Cells were recovered by filtration through sterile filters (70-m pores), resuspended in minimum Eagle's medium containing 25 mM Hepes, nonessential amino acids, and 10% fetal calf serum and plated out. Fibroblasts were allowed to adhere for 4 h, and any non-adhering astrocytes were replated on complete medium.
To eliminate the influence of serum, NCI H295R cells were transferred into Dulbecco's modified Eagle's/F12 medium supplemented with transferrin (5 g/ml), selenium (5 ng/ml), 200 units/ml penicillin, 200 g/ml streptomycin, and 2.5 mM L-glutamine. All experiments were performed with cell lines obtained from passages 2-8 following thawing or with primary cultures from passage 2.
Cytokine Treatment of Cell Cultures-Adrenocortical cell lines (NCI H295R) plated out at a density of 5 ϫ 10 6 cells per 100-mm dish were incubated in a serum-free medium for 24 h and then treated with TGF␤ 1 (Sigma) or FGF2 (R&D Systems) or left untreated as indicated in the text. In some experiments, cycloheximide (10 g/ml, Sigma) and actinomycin D (5 g/ml, Sigma) were added to the medium 1 h before the addition of the growth factor.
RNA Extraction and Northern Blotting-Total RNA was isolated from cultured cells by use of the acid-guanidium-thiocyanate-phenolchloroform extraction kit according to the manufacturer's instructions (Tri-Reagent, Sigma). Total RNA samples (10 g) were loaded onto a 1% agarose-2.2 mol/liter formaldehyde gel, subjected to electrophoresis, and transferred onto nylon membranes. The membranes were hybridized as previously described with the 1.9-kb EcoRI novH probe or the 2.3-kb EcoRI-XhoI mouse nov (novM) probes (4, 10, 43) labeled by random hexamer priming (Amersham Biosciences) in the presence of [ 32 P]dCTP. The signal for novH or novM was normalized according to the intensity of the gapdh signal (Clontech).
Luciferase Reporter Assays-NCI H295R cells plated in 6-well plates (5 ϫ 10 5 cells per well) were transfected using LipofectAMINE Plus (Invitrogen) as described in the user's manual. For reporter assays, the reporter constructs (0.5 g) were co-transfected with 0.1 g of pCMV-␤Ϫgalactosidase as an internal transfection control (Clontech). For assays in which the role of the transacting proteins was to be tested, 1.5 g of the empty pcDNA3 (Invitrogen) vector or pcDNA3 encoding the protein of interest was used. In dose response experiments, the total amount of the expression vectors was kept constant by use of the empty vector. In each assay, cell cultures were serum starved prior to treatment with TGF␤1 (4 ng/ml) or FGF2 (10 ng/ml). Luciferase and ␤-galactosidase activities were assayed by use of kits from Promega and PerkinElmer Life Sciences (Galacto-Star system), respectively. The data are presented as means Ϯ S.E. of representative experiments performed in triplicate on at least two separate occasions.
Protein samples from the lysates (10 g) or from the conditioned medium were subjected to electrophoresis in 12% reducing SDS-polyacrylamide gels before being transferred to polyvinylidene difluoride membranes (Hybond P, Amersham Biosciences) for immunological detection. The membrane was incubated with the K19M anti-NOVH (1:500 dilution) polyclonal antibody (10) for 1 h at 37°C. Immunoreactive proteins were detected by ECL (Amersham Biosciences) according to the manufacturer's instructions. For the detection of proteins encoded by transfected expression vectors, protein samples (40 g) derived from the same cell lysates used for luciferase were subjected to immunoblotting. The anti-Myc (9E10; Santa Cruz Biotechnology), monoclonal antibody was used to detect c-Myc-tagged Smad2, 3, and 4. Immunoreactive proteins were visualized by ECL.
Densitometry-Western blots were scanned with a GS700 imaging densitometer and processed with the Molecular Analyst data system (Bio-Rad). Northern blots were analyzed with a Storm PhosphorImager (Amersham Biosciences).

RESULTS
Expression of novH in NCI H295R Is Down-regulated by TGF␤1 and Up-regulated by FGF2-NCI H295R cells are con-sidered to be a good cellular model for adrenocortical tumors (24). To determine whether the expression of novH is influenced by environmental conditions, we used Western blotting to examine the amount of NOVH present in a conditioned medium when these cells were plated out at different densities. In serum-free medium, the amount of NOVH detected, corresponding to the same number of NCI H295R cells (10 5 ) tested, increased with cell density (Fig. 1A). The concentration of secreted NOVH also increased in the presence of serum (Fig. 1B). Thus, the production of NOVH in NCI H295R cells may be regulated by cell-cell contact and by growth factors present in serum.
We next tried to identify growth factors that affect the production of NOVH in these cells. We focused on TGF␤1, FGF2, and IGFs because they play important roles in adrenocortical development and physiology (see Ref. 26 for a review). As shown in Fig. 2, the amount of NOVH did not change following treatment with IGF-II or IGF-I; however, the amount of NOVH (in cell lysates or in medium) was decreased (ϳ 80%) by TGF␤1 (4 ng/ml) and increased (ϳ3-fold) by FGF2 (10 ng/ml). The amount of NOVH increased only slightly in the presence of both factors, suggesting that TGF␤1 inhibits both basal and FGF2induced expression of NOVH.
Next, we examined whether protein synthesis was required for the regulation of novH by TGF␤1 or FGF2 in NCI H295R. Pretreatment with the translation inhibitor cycloheximide 1 h before the addition of TGF␤1 or FGF2 did not block the effects of TGF␤1 but completely abolished the effects of FGF2 (Fig.  4A). Thus, TGF␤1 directly regulates the expression of novH, and FGF2 requires de novo protein synthesis, which could be rapidly induced.
To gain further insight into the molecular mechanisms by which TGF␤1 and FGF2 regulate the amount of novH mRNA, we investigated the effects of these factors on the steady-state levels of transcripts. From Fig. 4B it can be seen that when transcription was blocked with actinomycin D for 9 h, the basal level of novH mRNA decreased (ϳ3-fold), indicating that the half-life of novH transcripts is less than 9 h in these cells. Under these conditions, TGF␤1 and FGF2 did not significantly modulate the inhibitory effect of actinomycin D. This is consistent with the hypothesis that these factors regulate the expression of novH at the transcriptional level.
TGF␤1 and FGF2 Regulate novH Promoter Activity-To better understand the mechanism by which TGF␤1 and FGF2 regulate the transcriptional activity of novH, we analyzed their effects on the novH promoter fused to the luciferase reporter gene in transient transfections in NCI H295R cells. We assessed the regulation of three different promoter constructs, p2540NH-Luc (Ϫ2540 to ϩ87), p625NH-Luc (Ϫ625 to ϩ87), and p492NH-Luc (Ϫ492 to ϩ87) by TGF␤1 and FGF2 (Fig. 5A). Treatment of all three constructs with TGF␤1 resulted in ϳ50% inhibition, and treatment with FGF2 resulted in ϳ150% stimulation (Fig. 5, B and C). Thus, the promoter region between Ϫ2540 and Ϫ492 is not involved in the regulation of novH expression by these two factors. Furthermore, the stimulatory effect of FGF2 on p625NH-Luc promoter activity was reduced following the addition of TGF␤1 (Fig. 5D), which is consistent with our results for the endogenous NOVH. Therefore, we carefully examined the promoter region beyond position Ϫ492 to try to identify any specific cis-acting elements that could be involved in the regulation of novH by TGF␤1 and FGF2.
We also investigated the ability of TGF␤1 and FGF2 to modulate the expression of nov in cells from other species. Using primary cultures of mouse astrocytes, we observed that FGF2 had no effect, whereas TGF␤1 also decreased the amount of nov RNA in these mouse cells (Fig. 6). Therefore, we compared the sequences of the human (novH) and murine (novM) FIG. 4. Regulation of the expression of novH by TGF␤1 and FGF2 following treatment with cycloheximide (CHX) or actinomycin D (AD). Northern blot analysis of novH in NCI H295R (10 g of total RNA). A, cells were pretreated for 1 h with CHX (10 g/ml) before the addition of TGF␤1 and FGF2 for 24 h. B, cells were pretreated for 1 h with AD (5 g/ml) before the addition of TGF␤1 and FGF2 for 9 h. As the expression of gapdh was affected by AD in these conditions, the amount of novH mRNA was normalized relative to the 18 S ribosomal RNA. Densitometric analyses of the normalized novH RNA concentrations are presented in the lower panels. Two independent experiments gave the same results. C, control; AU, arbitrary units.
promoter sequences (Fig. 7A). We found two consensus sequences corresponding to AP-1 binding sites in the novH promoter region (at positions Ϫ473 and Ϫ447). Interestingly, these sequences were also present in the novM promoter region.
The AP-1 family of transcription factors is implicated in various regulatory activities of TGF␤1 (44 -47) and FGF2 (48,49). To determine the role of AP-1 in the regulation of nov expression by TGF␤1 and FGF2, we used site-directed mutagenesis to alter the AP-1 sites (Fig. 7B). Each of the point mutations resulted in a substantial decrease in basal promoter activity (ϳ4 -5-fold) when these constructs were used to transfect NCI H295R. No further effect was observed when both AP-1 sites were mutated simultaneously (Fig. 7, B and C).
As shown in Fig. 7C, none of the AP-1 mutations prevented FGF2 from stimulating novH promoter activity (ϳ100%), indicating that this process does not involve the binding of AP-1 to these sites. However, each of these mutations completely abrogated the effects of TGF␤1 on novH promoter activity in these cells (Fig. 7D) even following stimulation by FGF2 (data not shown). These data show that these AP-1 sites mediate the inhibitory effects of TGF␤1 and also suggest that the mechanism by which TGF␤1 inhibits FGF2 stimulation of novH expression is not a direct competition between transcription factors for binding on AP-1 sites.

FIG. 5. Effects of TGF␤1 and FGF2 on novH promoter activity.
A, schematic structures of novH-Luc reporter constructs. The potential consensus sequences of the transcription binding sites are indicated (37). B and C, NCI H295R cells were transfected with p2540NH-Luc, p625NH-Luc, or p492NH-Luc (0.5 g). Cells were or were not treated with TGF␤1 or FGF2 24 h prior to lysis and subjected to a luciferase assay. The results are expressed as the mean Ϯ S.E. of a representative experiment performed in triplicate. D, NCI H295R cells were transfected with p625NH-Luc. 4 h later the cells were or were not treated with FGF2. After 20 h they were treated with TGF␤1 in the presence or absence of FGF2 (10 ng/ml). Cells were subjected to the luciferase assay 24 h later. The increase in p625NH-Luc promoter activity (ϳ370%) due to FGF2 in these experiments compared with control was probably due to longer treatment times of cells with the growth factor. The results are expressed as the mean of a representative experiment, performed in triplicate, Ϯ S.E. AU, arbitrary units.

C-Jun and MEKK1
Are Involved in the Effects of TGF␤1 on novH Expression-Next, we were interested in determining which signaling pathways mediate the inhibitory effect of TGF␤1 on novH expression. TGF␤1 signaling is mediated by two types of serine-threonine kinase receptors (50,51). The highly conserved Smad proteins act as downstream signal transducers (51,52). Smad2 and Smad3 are restricted to the TGF␤/activin pathway. After phosphorylation by TGF␤1-activated type I receptors, pathway-restricted Smads form hetero-meric complexes with Smad4 and then translocate to the nucleus where they control the expression of a number of genes (53). TGF␤ also initiates other pathways such as the SAPK/ JNK pathway (42). This intracellular signal leads to the phosphorylation of c-Jun by JNK, which increases its transcriptional potential (54 -56). c-Jun is a member of the AP-1 family of transcription factors, which can bind to and activate transcription from AP-1 or 12-O-tetradecanoylphorbol-13-acetate (TPA)-responsive element sites (57). Several lines of evidence have indicated that c-Jun is a downstream target of TGF␤ signaling (42). However, only a few examples of a down-regulation of gene regulation by TGF␤1 involving AP-1 and c-Jun have been reported (41,58).
To investigate whether the down-regulation of novH expression by TGF␤1 involves c-Jun, we examined whether a dominant negative form of c-Jun (TAM67) lacking the region between amino acids 3 and 122 and encompassing the transactivation domain and the SAPK/JNK binding site could abrogate the effect of TGF␤1 on the p625NH-Luc reporter construct (42). As shown in of TGF␤1 on p625NH-Luc promoter activity. Similar results were also obtained ( Fig. 8A) with another dominant negative form of c-Jun in which the JNK phosphorylation sites (Ser-63 and Ser-73) were replaced by alanine (41). These data therefore strongly suggest that c-Jun through the JNK pathway plays a crucial role in the down-regulation of novH expression by TGF␤1.
We carried out further experiments to determine whether the activation of JNK contributes to the down-regulation of novH expression by TGF␤1. NCI H295R cells were treated for various periods of time with TGF␤1, and endogenous JNK activity was examined by an immune complex kinase assay using GST-Jun (1-79) as a substrate. Under our experimental conditions, the basal phosphorylation level of GST-Jun observed was relatively high (Fig 8B), and it remained elevated without any significant increase for all the time periods studied (up to 24 h). A weak but not reproducible increase was detected at 6 and 24 h in this representative experiment. We also checked whether JNK, was not transiently activated within the first 15 min as has been reported in some cells (59). Immunoblotting analysis of total cell lysates from NCI H295R with the anti-JNK antibody demonstrated that approximately equivalent amounts of the JNK, protein were present (Fig 8B). Thus, although c-Jun must be phosphorylated by JNK if novH is to be down-regulated by TGF␤1, because TGF␤1 did not significantly activate JNK, our results suggest that the basal level of JNK activity detected in these cells is sufficient for this inhibition to occur. Consistent with this, the production of increasing concentrations of a constitutively active MKK7 protein, a specific activator of JNK (60), neither decreased the basal level nor increased the TGF␤1-induced down-regulation of the novH promoter activity, which was still inhibited by ϳ50% (data not shown).
MEKK1 is an upstream activator of the JNK pathways that is also able to mediate the effects of TGF␤1 activation on AP-1-responsive promoters (59). Transient transfection of NCI H295R cells with a dominant negative interfering MEKK1 mutant (K432A) significantly blocked the TGF␤1-induced down-regulation of both p625NH-Luc and AP1-Luc promoter activities (Fig. 8C). These results suggest that additional components besides those activated by the JNK pathway are involved in the TGF␤1-mediated inhibition of these two reporter constructs.  1.2 g). The total amount of transfecting DNA was kept constant (1.6 g) by adding an empty pCDNA3 vector. Transfected cells were treated as in A. Results are presented as in A. AU, arbitrary units.

The Smad Pathway Is Not Required for the Down-regulation of novH Expression by TGF␤1-No
(CAGA) have been found in the p625NH-Luc promoter sequence (61,62); however, the Smad and JNK pathways may converge at the transcriptional levels (58). In particular, c-Jun physically interacts with Smad2, Smad3, and Smad4 (63), resulting in a synergy of activation on AP-1 site-mediated transcription (63,64). In contrast, c-Jun was shown to repress a TGF␤1-inducible promoter containing the Smad3/4 binding element CAGA (58,61). MEKK1 was also shown to modulate Smad2-mediated transcriptional activation selectively (65).
To determine whether the Smad pathway was functional in NCI H295R cells, we used the CAGA reporter containing nine copies of the Smad-binding site derived from the PAI-1 promoter (61). Treatment of the transfected NCI H295R cells with TGF␤1 led to a ϳ100-fold increase in the CAGA reporter activity (Fig. 9A), indicating that TGF␤1 can induce the Smad pathway in these cells.
We therefore co-transfected NCI H295R cells with p625NH-Luc and increasing concentrations of either the Smad2, Smad3, or Smad4 expression vector. As presented in Fig. 9B, we observed that the overexpression of Smad2, Smad3, or Smad4 in the absence of TGF␤1 did not significantly affect the basal novH promoter activity. This is consistent with previous studies of CAGA-mediated transcription (61,66). More importantly, TGF␤1 still down-regulated novH promoter activity as efficiently as it does in the absence of co-transfected Smad proteins. In all of these experiments, the expression of the c-Myc tagged-Smad proteins was checked by immunoblotting using an anti-c-Myc monoclonal antibody (Fig. 9B). A similar conclusion could be drawn when Smad2 or Smad3 were transfected together with Smad4 (data not shown). We also investigated the effects on this process of the overexpression of a dominant negative interfering form of Smad4 (DN Smad4) and Smad7, a natural inhibitor of the Smad pathway (67,68). The expression of both DN Smad4 and Smad7 in NCI H295R significantly decreased the TGF␤1-induced CAGA promoter activity but did not affect the TGF␤1-dependent down-regulation of novH promoter activity (Figs. 9, C and D). Thus, these results suggest that Smad signaling does not participate in the TGF␤1-dependent down-regulation of novH expression.

DISCUSSION
Because IGF-I, IGF-II, FGF2, and TGF␤1 are involved in the physiological functions of adrenocortical cells, (26,29), we analyzed their effects on the expression of novH in the NCI H295R cell line, which is derived from a human adrenocortical carcinoma (25). This cell line allowed us to show for the first time that novH expression is up-regulated by FGF2 and downregulated by TGF␤1, whereas IGF-I and II have no influence. These data suggested that FGF2 and TGF␤1, which are involved in the development of various tumors (69,70) and are also produced by adrenocortical cells (71,72), could be considered as potential candidates involved in the modulation of novH expression in adrenocortical tumors (19). Our results also show that TGF␤1 reduces the up-regulation of novH expression induced by FGF2, suggesting that the basal level of novH expression results from a balance between the actions of these two growth factors. This balance may vary during tumorigenesis, and FGF2 detected in adrenocortical tumors (72) may play a major role in the overexpression of novH during the earlier stages of tumorigenesis. We cannot exclude the possibility that other factors also influence the levels of novH expression during adrenocortical tumorigenesis.
FGF2 has been shown to up-regulate the expression of novH in NCI H295R cells; however, we showed that down-regulation of the expression of novH by TGF␤1 is not restricted to tumoral adrenocortical cells, because it was also observed in primary astrocytes and could be detected in human as well as murine cells. The regulation of novH by TGF␤1 might, however, present some specificity, because it has not been reported in human prostatic cells (36).
novH is the first member of the CCN family that has been shown to be down-regulated by TGF␤1. The other members of this family such as ctgf and cyr61 are induced by TGF␤1 in different cell systems (33,36). It is noteworthy that nov is also regulated oppositely from ctgf and cyr61 in chicken embryo fibroblasts. In these cells, ctgf and cyr61 behave as immediateearly genes induced by serum and oncogenes (73,74), whereas the expression of nov is down-regulated by these factors and associated with quiescence (75). Primary cultures of mouse astrocytes are another cell type in which the expression of nov and ctgf is inversely regulated by TGF␤1. 4 These observations suggest that NOV and CTGF and CYR61 have antagonistic functions in certain cell systems. However, the expression of nov and ctgf has been reported to be down-regulated by Wilms' tumor suppressor gene 1 (WT1) in renal cells (37,76), indicating that these two molecules may also cooperate in some cells.
Under our experimental conditions, the expression of ctgf and cyr61 was barely detectable in NCI H295R cells, and the expression of ctgf was only slightly stimulated by TGF␤1. 4 In fibroblasts, CTGF can function as a downstream mediator of TGF␤1 activity. For example, it can stimulate cell proliferation and extracellular matrix protein synthesis (77,78). In vivo, CTGF plays a role in TGF␤1-mediated formation of granulation tissue and cooperates with TGF␤1 to induce persistent fibrosis (79). Whether NOVH mediates a function of TGF␤1 in adrenocortical cells remains to be determined. However, no correlation could be found between the levels of novH expression in NCI H295R cells treated with TGF␤1 or FGF2 and their proliferation state as assessed by [ 3 H]thymidine incorporation. 4 TGF␤1 has been reported to be a strong inhibitor of steroidogenesis in adrenocortical cells (26); therefore, novH may antagonize the effect of TGF␤1 in this function. This hypothesis is currently under investigation. TGF␤1 can potentiate tumorigenesis by down-regulating the genes involved in cell-cell adhesion and by up-regulating the expression of genes involved in cell-extracellular matrix association, ultimately improving the migration and invasiveness of the cell (69). These properties are more consistent with NOVH having a role as an adhesive protein (13) that is able to regulate the expression of genes involved in extracellular matrix remodeling. 3 Because the expression of novH is directly regulated by TGF␤1 and different elements of the TGF␤1 signaling pathway can also be altered in cancer (69), we analyzed the signaling pathway involved in the TGF␤1-mediated down-regulation of novH. Comparison of the human nov promoter region, which is targeted by TGF␤1, with the corresponding mouse sequences revealed a high degree of sequence homology (69%). These conserved regions included several consensus sequences involved in the binding of transcription factors (such as USF, NFB, NFY, and AP-1), suggesting that novH and novM could be subjected to common regulations. Our data provide an example of one of those, as the expression of novH and novM can be down-regulated by TGF␤1. We further demonstrated that down-regulation of novH expression is mediated by AP-1 sites, which are found in the same region of the two promoters. Our results suggest that the TGF␤1 signaling pathway targets AP-1 sites in the novH promoter to inhibit the expression of novH. However, we consistently observed that the mutation of AP-1 sites also results in a decrease in the basal activity of the novH promoter. The molecular mechanisms involved in maintaining the expression of novH in unstimulated NCI H295R cells are currently unknown. The identification of environmental cues that regulate the expression of novH will help to clarify this point.
Whereas both ctgf and novH can be regulated by TGF␤1, quite different promoter sequences are involved in this regulation, because the Smad pathway is responsible for the upregulation of ctgf expression by TGF␤1 in fibroblasts (80). The induction of gene expression by TGF␤1 involving Smad or AP-1 binding sequences and c-Jun has been well documented (81,82), but there are only a few reports of the down-regulation of gene expression by TGF␤1 involving c-Jun (41,58). For example, TGF␤1 down-regulates the expression of the gene that encodes the metalloproteinase MMP12 (83) through AP-1 sites, but this inhibitory effect is dependent on signaling through Smad3. Our results showed that the mechanism by which novH is negatively regulated by TGF␤1 in NCI H295R cells is different. We demonstrated that although the Smad pathway in these cells was induced by TGF␤1, which is in agreement with a previous report (84), this pathway is not involved in the TGF␤1-mediated inhibition of novH expression. Our data concerning the novH promoter mutated in the AP-1 sites and the dominant forms of c-Jun mutated in the JNK binding domain or in JNK-specific phosphorylation indicate that the SAPK/ JNK pathway is required in this regulation. Two other studies (59,85) showed that the activation of JNK independently of Smads leads to the regulation of fibronectin and insulin-like growth factor binding protein 5 (IGFBP5) by TGF␤1. In contrast, our data suggest that the down-regulation of novH expression by TGF␤1 requires a basal level of JNK activity to phosphorylate c-Jun and an additional TGF␤1-dependent mechanism. We provide evidence that MEKK1 could play a crucial role in this regulation in a manner that does not involve the activation of JNK. It has recently been shown that MEKK1 is able to directly activate, independently of JNK, other proteins such as p300/cAMP-response element-binding proteinbinding protein (86). It has also been reported that by reinforcing the association between c-Jun and TGIF, TGF␤1 leads to the repression of AP1-mediated transcriptional activity (41). It is therefore tempting to speculate that when NCI H295R are treated with TGF␤1, a factor specifically regulated by MEKK1 could participate in an interaction between c-Jun and TGIF, resulting in the down-regulation of novH expression. However, the molecular mechanism by which MEKK1 could participate in the TGF␤1-negative regulation of novH expression awaits further investigation. The study of novH regulation in NCI H295R cells, in which the Smad pathway is functional, may therefore represent a good model for a better understanding of the molecular mechanisms involved in the TGF␤1-mediated inhibition of gene expression.
In summary, the data presented here demonstrate that novH is a new target for TGF␤1. Further studies aimed at determining which of the functions of TGF␤1 are mediated by novH might be useful to the development of therapeutic agents for the treatment of diseases involving also other members of the CCN family such as fibrosis or cancer.