Transforming Growth Factor β1 Decreases Cholesterol Supply to Mitochondria via Repression of Steroidogenic Acute Regulatory Protein Expression*

Transforming growth factor-βs (TGF-βs) constitute a family of dimeric proteins that affect growth and differentiation of many cell types. TGF-β1 has also been proposed to be an autocrine regulator of adrenocortical steroidogenesis, acting mainly by decreasing the expression of cytochrome P450c17. Here, we demonstrate that TGF-β1 has a second target in bovine adrenocortical cells, namely the steroidogenic acute regulatory protein (StAR). Indeed, supplying cells with steroid precursors revealed that TGF-β1 inhibited two steps in the steroid synthesis pathway, one prior to pregnenolone production and another corresponding to P450c17. More specifically, TGF-β1 inhibited pregnenolone production but neither the conversion of 25-hydroxycholesterol to pregnenolone nor P450scc activity. Thus, TGF-β1 must decrease the cholesterol supply to P450scc. We therefore examined the effect of TGF-β1 on the expression of StAR, a mitochondrial protein implicated in intramitochondrial cholesterol transport. TGF-β1 decreased the steady state level of StAR mRNA in a time- and concentration-dependent manner. This inhibition occurs at the level of StAR transcription and depends on RNA and protein synthesis. It is likely that the TGF-β1-induced decrease of StAR expression that we report here may be expanded to other steroidogenic cells in which a decrease of cholesterol accessibility to P450scc by TGF-β1 has been hypothesized.

Although initially identified by its ability to induce reversible phenotypic transformation of non-neoplastic cells by stimulating anchorage-independent cell growth, TGF-␤ 1 1 is now recognized as a physiological mediator of growth and differen-tiation of various cell types (1). Its major effects are on the cell cycle in epithelial cells and on the synthesis of extracellular matrix proteins in most cell types. TGF-␤ 1 has also been identified as a potential physiological autocrine/paracrine regulator of bovine adrenocortical steroidogenic functions (2,3). Our laboratory has shown that bovine adrenal cortical (BAC) cells from the fasciculata zone express specific high affinity TGF-␤ 1 receptors (4) and produce and secrete TGF-␤ 1 under a latent form (5,6). Immunolocalization of TGF-␤ 1 in the bovine or rat adrenal gland revealed its presence in the glomerulosa and fasciculata zones (5,7).
The mechanism of TGF-␤ 1 action on adrenal steroidogenesis is not fully understood. In TGF-␤ 1 -treated BAC cells, low density lipoprotein uptake and metabolism are inhibited (8), and angiotensin II binding is reduced (9). TGF-␤ 1 also decreases the expression of several steroidogenic enzymes. The major target has been shown to be cytochrome P450c17 which is strongly inhibited by TGF-␤ 1 in bovine (10), ovine (11), and human adrenal cells (12). In addition, TGF-␤ 1 has been shown to inhibit 3␤-hydroxysteroid dehydrogenase isomerase (3␤-HSD) (13) and cytochrome P450 side chain cleavage (P450scc) (14) enzyme expressions, although regulation of these enzymes seems to differ from one species to another. The negative effect of TGF-␤ 1 on steroidogenesis has also been observed in other cell types such as testicular Leydig cells (15,16) and ovarian thecal cells (17).
Cholesterol is an obligatory precursor of steroid hormones in steroidogenic cells. Endogenous cholesterol is either synthesized from acetate or absorbed from blood, principally through LDLs in the bovine adrenal cortex, and is stored mainly as cholesterol esters in lipid droplets in the cytoplasm. The biosynthesis of steroid hormones upon ACTH stimulation starts with the transport of the substrate cholesterol from extramitochondrial stores to the inner mitochondrial membrane where the first enzyme in the pathway, P450scc, cleaves cholesterol into pregnenolone. The intramitochondrial transport of cholesterol is the rate-limiting step in steroidogenesis and the main site for regulation by physiological stimuli during acute stimulation of steroid production. Activation of this step is known to require de novo protein synthesis (18). Recently, the steroidogenic acute regulatory protein (StAR) has been implicated as the regulator of intramitochondrial cholesterol translocation (for a review see Ref. 19). StAR is synthesized as a 37-kDa labile cytoplasmic precursor that is imported into the mitochondrion, where the cleavage of the mitochondrial targeting sequence occurs, yielding a 30-kDa protein. Transfection of MA-10 Leydig tumor cell line with the mouse StAR cDNA (20) or that of COS-1 cells with the human StAR cDNA in combi-nation with the cholesterol side chain cleavage enzyme system (21) results in enhanced steroid production. Mutations in the StAR gene were subsequently found to be responsible for lipoid congenital adrenal hyperplasia, an autosomal recessive disease in which the synthesis of all gonadal and adrenal steroids is severely impaired (22). Fifteen different mutations in the StAR gene were found, and all rendered the StAR protein inactive in functional assays (22,23).
Closer examination of P450c17 inhibition by TGF-␤ 1 in bovine adrenocortical cells suggested that TGF-␤ 1 might have an earlier target in the biosynthesis of steroids. Testing this hypothesis, we found that TGF-␤ 1 regulated two distinct steps in steroid biosynthesis as follows: 1) the conversion of pregnenolone to 17␣-hydroxypregnenolone due to the inhibition of P450c17 expression, and 2) the supply of cholesterol to P450scc, suggesting inhibition of cholesterol transport to the inner mitochondrial membrane. This prompted us to study the effect of TGF-␤ 1 on StAR expression. We report here that TGF-␤ 1 decreases the steady state level of StAR mRNA in a concentrationand time-dependent manner and that this inhibition requires transcription and translation of an intermediate protein.
Bovine Adrenal Fasciculata Cell Preparation-Bovine adrenal glands were obtained from a local slaughterhouse. Fasciculata-reticularis cells were obtained by successive tryptic digestions, seeded in 10-cm plates or in 12-well plates, and grown in Ham's F12 medium supplemented with 10% horse serum (Eurobio, Les Ulis, France) and 2.5% fetal calf serum (Life Technologies, Inc., Cergy-Pontoise, France) as described previously (24). The cells were used at day 2 or 3 of primary culture.
Steroid Production-BAC cells were incubated for 15 h with or without TGF-␤ 1 (2 ng/ml). The medium was then removed and replaced for 2 h with fresh medium for cortisol production or fresh medium containing 10 M trilostane (3␤-HSD inhibitor) and 40 M SU10603 (P450c17 inhibitor) to measure pregnenolone production. To study steroid production from exogenous precursors, either 25-hydroxycholesterol (50 M), pregnenolone (20 M), or 17␣-hydroxypregnenolone (20 M) were added to the fresh medium during the 2-h incubation.
P450scc Activity-BAC cells were incubated for 15 h with or without TGF-␤ 1 (2 ng/ml). The cells were scraped and then homogenized with a Potter-Kontes homogenizer (1200 rpm, 35 strokes) in 5 mM Tris-HCl, pH 7.4, 275 mM sucrose. The homogenate was centrifuged at 500 ϫ g for 15 min to remove large debris and nuclei. Mitochondria were collected by centrifugation at 10,000 ϫ g for 10 min and washed once with the same buffer, and protein concentrations were determined using the Bradford method (25). 50 g of isolated mitochondria were equilibrated at 37°C in 10 mM Tris-HCl, pH 7.4, 250 mM sucrose, 10 mM KCl, 10 mM potassium phosphate, 5 mM MgCl 2 , 5 M metyrapone (11␤-hydroxylase inhibitor), 10 M trilostane, and 200 M 25-hydroxycholesterol in a final volume of 500 l. A mix of malate/isocitrate (1:1, v/v, final concentrations 10 mM) was added to start the reaction. The reaction was carried out for 10 min and was stopped by the addition of 5 ml of dichloromethane (26). Pregnenolone was extracted and assayed by RIA as described above.
RNA Preparation and Northern Blot Analysis-Total RNA was isolated from cells using the RNAgents® kit (Promega, Charbonnières, France). 25 g of total RNA were separated by electrophoresis through a 1% agarose gel containing 1.9% formaldehyde. RNA was then trans-ferred to a Hybond-N membrane (Amersham, Les Ulis, France). Blots were sequentially probed with a full-length murine StAR cDNA and an 18 S rRNA probe that were labeled by random priming with [␣-32 P]dCTP (111 TBq/mmol, ICN Pharmaceuticals, Orsay, France) using the Radprime DNA labeling kit (Life Technologies, Inc., Cergy-Pontoise, France). Prehybridization and hybridization at 65°C were performed in Rapid-Hyb buffer (Amersham, Les Ulis, France). The blots were washed twice at room temperature in 2ϫ SSC (1ϫ SSC ϭ 0.15 M NaCl, 15 mM sodium citrate) and 0.1% sodium dodecyl sulfate, once in 1ϫ SSC and 0.1% sodium dodecyl sulfate at 65°C, and once in 0.1ϫ SSC and 0.1% sodium dodecyl sulfate at 65°C. Hybridizing bands were visualized on a ␤-imager (PhosphorImager, Molecular Dynamics, Sunnyvale, CA) and quantified using the ImageQuant TM program (Molecular Dynamics, Sunnyvale, CA). Values for StAR mRNA were normalized to values for the 18 S rRNA.
Statistics-Statistical analysis was performed with Student's t test for comparison of two groups. Differences were considered significant when p Ͻ 0.05.

TGF-␤ 1 Alters at Least Two Steps in the Cortisol Biosynthesis
Pathway-Recently, using primary cultures of bovine adrenocortical (fasciculata-reticularis) cells, we re-examined the effects of TGF-␤ 1 on several steps of the cortisol biosynthesis pathway. Cortisol synthesis begins with the transport of the substrate cholesterol to the first enzyme in the pathway, P450scc, which cleaves cholesterol into pregnenolone which will then be hydroxylated by P450c17 into 17␣-hydroxypregnenolone. The subsequent steps involve 3␤-HSD, P450c21, and P450c11␤ enzymes. We used a cell-permeant analog of cholesterol and some of the intermediate substrates to determine whether inhibition by TGF-␤ 1 could be overcome by skipping any of these steps. BAC cells were pretreated for 15 h with or without TGF-␤ 1 (2 ng/ml); the medium was then replaced with fresh medium containing the steroid precursors, and cortisol production was measured after 2 h. The results shown in Table  I confirm that TGF-␤ 1 is a very potent inhibitor of basal cortisol production (81% inhibition). When the cells are supplied with either 25-hydroxycholesterol (a membrane-permeant analog of cholesterol) or pregnenolone, TGF-␤ 1 -induced inhibition of cortisol synthesis is reduced to only 39 and 33%, respectively. TGF-␤ 1 then must act prior to the production of pregnenolone, apparently at the level of cholesterol supply to P450scc. When 17␣-hydroxypregnenolone is supplied as substrate, TGF-␤ 1 no longer significantly inhibits cortisol synthesis (3%), confirming that TGF-␤ 1 reduces the conversion of pregnenolone into 17␣hydroxypregnenolone. This block corresponds to the inhibition of P450c17 expression that we have previously demonstrated (10,27).

TGF-␤ 1 Inhibits Pregnenolone Production and Decreases
Cholesterol Accessibility to P450scc-To confirm that TGF-␤ 1 inhibits an early steroidogenic step, we measured its effect on pregnenolone accumulation using inhibitors of 3␤-HSD (trilostane) and P450c17 (SU 10603) to block pregnenolone metabolism. As shown in Fig. 1, pregnenolone production is inhibited by TGF-␤ 1 in a time-dependent manner. This inhibition occurred as early as 6 h after treatment and reached a maximum at around 12 h in both unstimulated and ACTHstimulated cells (Fig. 1, A and B, respectively). We then measured the effect of TGF-␤ 1 on the conversion of 25-hydroxycholesterol into pregnenolone. Table II shows that when this permeant cholesterol analog was supplied as a steroid precursor, TGF-␤ 1 no longer inhibited pregnenolone production (6% inhibition in the presence of 25-hydroxycholesterol versus 58% in its absence). These results suggest that TGF-␤ 1 decreases cholesterol accessibility to the inner mitochondrial membrane rather than directly affects P450scc. This was confirmed by measuring the level and activity of P450scc protein in mitochondria isolated from cells treated or not treated with TGF-␤ 1 . We observed no difference in either P450scc protein level (data not shown) or activity in the absence of TGF-␤ 1 (228 Ϯ 21 ng of pregnenolone/mg protein/h) or in its presence (241 Ϯ 57 ng of pregnenolone/mg protein/h).

TGF-␤ 1 Decreases the Steady State Level of StAR mRNA-
The recent cloning of the StAR cDNA has allowed us to demonstrate that the corresponding protein plays a fundamental role in the mitochondrial import of cholesterol (20). We therefore decided to investigate the effect of TGF-␤ 1 on the steady state level of StAR mRNA. Northern blot analysis of BAC cell total RNA using either a full-length mouse cDNA ( Fig. 2A) or a partial bovine cDNA probe (data not shown) revealed two major StAR transcripts (3 and 1.7 kb) that have been previously described (28) and a minor transcript (1.3 kb). The 3-kb and the 1.7-kb transcripts account for 60 and 30% of StAR mRNA expression, respectively, whereas the 1.3-kb transcript accounts for 10%. All three transcripts showed coordinate induction, and therefore, the most abundant transcript of 3 kb was quantified as a representative of the three transcripts. Continuous treatment of BAC cells with TGF-␤ 1 (2 ng/ml) caused a time-dependent decrease in the steady state level of StAR mRNA in both unstimulated and ACTH-stimulated cells. This decrease was significant after 4 h of exposure to TGF-␤ 1 , and the effect was nearly maximal after 12 h and sustained for up to 24 h (85% inhibition) (Fig. 2B). Fig. 2C shows the kinetics of StAR mRNA induction in ACTH-stimulated cells versus control cells. We found that StAR mRNA levels in BAC cells were markedly increased within 1 h and peaked at 2 h after ACTH treatment, returning to a basal level after 12 h of stimulation.
The inhibitory effect of TGF-␤ 1 was concentration-dependent, and a 12-h exposure of BAC cells to TGF-␤ 1 concentrations ranging from 2 pg/ml to 2 ng/ml decreased StAR transcript levels by 15-75% (Fig. 3). The EC 50 for StAR inhibition was found to be 20 pg/ml TGF-␤ 1 .
TGF-␤ 1 Does Not Modify StAR mRNA Stability-To determine whether the effects of TGF-␤ 1 on StAR mRNA levels were due to changes in transcript stability, BAC cells were incubated in the presence or absence of TGF-␤ 1 (2 ng/ml) for 12 h and subsequently treated with actinomycin D (2.5 g/ml) for 6 -24 h. The half-life of StAR mRNA in transcriptionally arrested BAC cells was determined to be 12 h in both control and TGF-␤ 1 -treated cultures (Fig. 4). This suggests that TGF-␤ 1 does not modify the degradation of StAR mRNA but rather  affects its transcription rate.
The TGF-␤ 1 Effect Depends on RNA and Protein Synthesis-To determine whether the effect of TGF-␤ 1 on StAR mRNA levels was dependent on transcription and/or translation, BAC cells were treated with TGF-␤ 1 in the presence or absence of actinomycin D (2.5 g/ml) or cycloheximide (10 g/ml). We found that, in the presence of either of these two inhibitors, TGF-␤ 1 no longer decreases StAR transcripts levels (Fig. 5). Thus, inhibition by TGF-␤ 1 seems to depend on the synthesis of other proteins. DISCUSSION TGF-␤ 1 treatment of adrenocortical steroidogenic cells has been shown to result in the inhibition of steroid hormone biosynthesis, mainly attributable to a decrease in P450c17 expression (10,11,27). In the present study, we show that TGF-␤ 1 inhibits not only P450c17 expression but also that of StAR, a recently described protein implicated in intramitochondrial cholesterol translocation. Furthermore, we could demonstrate that this inhibition occurs at the level of StAR transcription and requires the transcription and translation of new proteins.
A number of studies performed on bovine and ovine adrenocortical fasciculata and glomerulosa cells (29 -31), in porcine testicular Leydig cells (16), and in porcine ovarian thecal cells (17) have led to the hypothesis that TGF-␤ 1 modifies cholesterol accessibility to P450scc. In most of these works, TGF-␤ 1 has been shown to decrease pregnenolone synthesis, which could be restored by the addition of 22(R)-or 25-hydroxycholesterol. In the present study, we used precursor steroids, 25-hydroxycholesterol, pregnenolone, and 17␣-hydroxypregnenolone, to demonstrate that TGF-␤ 1 has at least two targets in BAC cells as follows: one that lies between pregnenolone and 17␣-hydroxypregnenolone, previously identified as P450c17 (10,27), and another that seems to affect cholesterol accessibility to P450scc. Indeed, addition of 25-hydroxycholesterol fully restored pregnenolone synthesis in TGF-␤ 1 -treated cells. Pregnenolone production in isolated mitochondria from TGF-␤ 1 -treated cells in which cholesterol was allowed to accumulate was also inhibited (data not shown). Finally, we found that TGF-␤ 1 inhibits neither the level nor the activity of P450scc. Altogether, these results clearly demonstrate that TGF-␤ 1 decreases cholesterol accessibility to the inner mitochondrial membrane, in accordance with the studies mentioned above. The only hypothesis in the literature to explain this effect was formulated by Hotta and Baird (8) who showed that TGF-␤ 1 induces a decrease in the number of LDL receptors, which are in part responsible for cholesterol entry into the cell. However, we found that TGF-␤ 1 still inhibits pregnenolone production in the absence of LDLs in the medium (30 Ϯ 4% inhibition). In view of our results, we cannot completely exclude that TGF-␤ 1 might decrease the number of LDL receptors; however, it is clear that TGF-␤ 1 hits another target. We examined the effect of TGF-␤ 1 on the expression of StAR, a mitochondrial protein required for steroid hormone biosynthesis, which regulates cholesterol transfer to the inner mitochondrial membrane. Our results demonstrate that TGF-␤ 1 strongly inhibits StAR mRNA expression in a time-and dose-dependent manner. Interestingly, TGF-␤ 1 -induced decrease of StAR mRNA expression occurs with similar kinetics and EC 50 as TGF-␤ 1 inhibition of pregnenolone synthesis. To definitively establish a direct link between these two inhibitions, we are currently testing whether overexpression of StAR relieves the TGF-␤ 1 inhibition of pregnenolone synthesis.
It is long known that a newly synthesized protein(s) is absolutely required for the acute regulation of steroidogenesis (32,33). Several candidate proteins have been proposed, including the sterol carrier protein 2, the steroidogenesis activator protein, the peripheral benzodiazepine receptor with the diazepam binding inhibitor, and a family of proteins of approximately 30 kDa that are rapidly synthesized and phosphorylated in response to corticotropic hormone (for review, see Ref. 34). More recently, Clark et al. (20) cloned the cDNA encoding the 30-kDa phosphoprotein that was called StAR. A model has been proposed that dissociates the import to the mitochondrion and processing of StAR from its steroidogenic activity. It was shown both in vivo (23) and in vitro (35) that N-terminal truncated StAR proteins can increase pregnenolone synthesis without entering the mitochondria, probably by acting on the outer mitochondrial membrane. This suggests that StAR acts through its C-terminal domain to induce cholesterol import into the inner mitochondrial membrane.
It is clear that StAR plays a fundamental role in steroidogenesis. This implies that the expression of such an important protein must be tightly regulated. StAR is rapidly synthesized in response to cAMP analogs in the Leydig tumor cell line MA-10 (36) and in ovary cells (37) and in response to angiotensin II through a Ca 2ϩ -dependent pathway in bovine glomerulosa cells (38). StAR is regulated both at the mRNA and the protein level. In particular, it has been found that the 37-kDa StAR protein has a very short half-life of 4 -6 min (39). We have performed Northern blot analysis to study the expression of StAR mRNA in BAC cells. We found two major transcripts (3 and 1.7 kb) that have been previously described (28,40) and a minor transcript (1.3 kb). This smaller transcript has not been described before in bovine cells, but this could be due to low sensitivity of Northern blots in the earlier studies, as three transcripts have also been shown in both human (21) and mouse (36). All three transcripts are long enough to encode a full-length protein. The functional significance of the different transcripts is not yet known; however, all three are coordinately induced by ACTH in BAC cells. StAR mRNA levels in BAC cells were markedly increased within 1 h of ACTH treatment, peaked at 2 h, and returned to a basal level after 12 h. This is a little faster than in mouse cells stimulated with a cAMP analog (starting after 2 h and peaking between 4 -6 h) (36) and much faster than in ovary cells (maximum at 24 h) (37). We found that TGF-␤ 1 significantly decreased StAR mRNA levels as early as 4 h, reaching a maximum at 12 h in both unstimulated and ACTH-stimulated cells. Our results show that ACTH induction of StAR mRNA expression is faster than its decrease by TGF-␤ 1 . The EC 50 for this inhibition was around 20 pg/ml, which corresponds to the EC 50 described for the inhibition of epithelial cell proliferation (41). The inhibitory effect of TGF-␤ 1 on the steady state level of StAR mRNA was not due to an increase in its degradation, as we found a similar half-life (12 h) for StAR mRNA in both control and TGF-␤ 1treated cells. It can be surmised that TGF-␤ 1 inhibition is due to a decrease in StAR gene transcription. This is the first report of a transcriptional inhibition of StAR. Reduction of StAR mRNA expression by prostaglandin F2␣ in the rat ovary has been reported (42), but the mechanism was not elucidated.
Importantly, we found that the inhibition of StAR in BAC cells is dependent upon ongoing transcription and protein synthesis, as addition of actinomycin D or cycloheximide completely abolishes the effect of TGF-␤ 1 . Thus the inhibition of StAR expression by TGF-␤ 1 requires the production of additional intermediate regulators. TGF-␤ 1 signals through type I and type II transmembrane serine/threonine kinase receptors (for a review, see Ref. 43). Activation of the receptor complex occurs when the type II receptor kinase transphosphorylates the type I receptor in response to ligand binding. This activates the type I kinase, which transiently associates with and phosphorylates Smad proteins. Phosphorylated Smads translocate to the nucleus and regulate transcription of TGF-␤-responsive genes. It will be interesting to see whether Smads are involved in TGF-␤ 1 signal transduction in BAC cells, as it has not yet been demonstrated that Smads are implicated in all biological responses to TGF-␤. Possible candidates for the intermediary protein include the transcription factors c-Jun and/or c-Fos that bind to AP-1 sites. Indeed, the expression of these two immediate-early genes is induced by TGF-␤ 1 in many different cell types (44,45). Furthermore, phorbol myristate acetate prevents StAR mRNA induction by a cAMP analog in the human ovary and in the ovine corpora lutea (37,42), by activating protein kinase C, a kinase that is known to modulate responses through AP-1-binding sites.
The StAR promoter, like that of the cytochrome P450 hydroxylases, lacks typical cAMP-responsive elements; instead, alternate cAMP-responsive sequences mediate the hormonal induction of these genes (46). Furthermore, the steroid hydroxylases and StAR are all apparently transcriptionally regulated by the orphan nuclear receptor, SF-1 (47). It is tempting to speculate that TGF-␤ 1 -induced inhibition of StAR and P450c17 expression may occur via the same mechanism involving a common regulator like SF-1. DAX-1, which also belongs to the orphan nuclear receptor family and has been found to inhibit SF-1 transactivation (48), is another potential target of TGF-␤ 1 action.
Altogether, the data reported here demonstrate that TGF-␤ 1 is a major regulator of differentiated functions of steroidogenic cells, modifying the two rate-limiting steps in cortisol biosynthesis, StAR and P450c17 expression. We are currently investigating the possibility of a common repressive mechanism of TGF-␤ 1 on StAR and CYP17 promoters. The present observations support the potential role of TGF-␤ 1 as a negative regulator of adrenocortical cell-differentiated functions, in balance with the major positive agent ACTH, to control the overall homeostasis of this endocrine tissue.