Submitochondrial distribution of three key steroidogenic proteins (steroidogenic acute regulatory protein and cytochrome p450scc and 3beta-hydroxysteroid dehydrogenase isomerase enzymes) upon stimulation by intracellular calcium in adrenal glomerulosa cells.

In adrenal glomerulosa cells, angiotensin II (Ang II) and potassium stimulate aldosterone synthesis through activation of the calcium messenger system. The rate-limiting step in steroidogenesis is the transfer of cholesterol to the inner mitochondrial membrane. This transfer is believed to depend upon the presence of the steroidogenic acute regulatory (StAR) protein. The aim of this study was 1) to examine the effect of changes in cytosolic free calcium concentration and of Ang II on intramitochondrial cholesterol and 2) to study the distribution of StAR protein in submitochondrial fractions during activation by Ca2+ and Ang II. To this end, freshly prepared bovine zona glomerulosa cells were submitted to a high cytosolic Ca2+ clamp (600 nM) or stimulated with Ang II (10 nM) for 2 h. Mitochondria were isolated and subfractionated into outer membranes, inner membranes (IM), and contact sites (CS). Stimulation of intact cells with Ca2+ or Ang II led to a marked, cycloheximide-sensitive increase in cholesterol in CS (to 143 ± 3.2 and 151.1 ± 18.1% of controls, respectively) and in IM (to 119 ± 5.1 and 124.5 ± 6.5% of controls, respectively). Western blot analysis revealed a cycloheximide-sensitive increase in StAR protein in mitochondrial extracts of Ca2+-clamped glomerulosa cells (to 159 ± 23% of controls). In submitochondrial fractions, there was a selective accumulation of StAR protein in IM following stimulation with Ca2+ (228 ± 50%). Similarly, Ang II increased StAR protein in IM, and this effect was prevented by cycloheximide. In contrast, neither Ca2+ nor Ang II had any effect on the submitochondrial distribution of cytochrome P450scc and 3β-hydroxysteroid dehydrogenase isomerase. The intramitochondrial presence of the latter enzyme was further confirmed by immunogold staining in rat adrenal fasciculata cells and by immunoblot analysis in MA-10 mouse testicular Leydig cells. These findings demonstrate that under acute stimulation with Ca2+-mobilizing agents, newly synthesized StAR protein accumulates in IM after transiting through CS. Moreover, our results suggest that the import of StAR protein into IM may be associated with cholesterol transfer, thus promoting precursor supply to the two first enzymes of the steroidogenic cascade within the mitochondria and thereby activating mineralocorticoid synthesis.

demonstrate that under acute stimulation with Ca 2؉mobilizing agents, newly synthesized StAR protein accumulates in IM after transiting through CS. Moreover, our results suggest that the import of StAR protein into IM may be associated with cholesterol transfer, thus promoting precursor supply to the two first enzymes of the steroidogenic cascade within the mitochondria and thereby activating mineralocorticoid synthesis.
The Ca 2ϩ -mobilizing agonists angiotensin II (Ang II) 1 and K ϩ act as regulators of aldosterone synthesis and secretion in adrenal glomerulosa cells. The crucial role of the Ca 2ϩ messenger in the acute regulation of aldosterone production is firmly established (1)(2)(3)(4)(5). Indeed, the steroidogenic response of isolated adrenal cells to Ang II and K ϩ is highly dependent upon extracellular Ca 2ϩ concentration (6) and can be blocked by inhibitors of Ca 2ϩ influx across the plasma membrane (4). Moreover, calmodulin antagonists have been shown to inhibit Ang IIstimulated aldosterone production in zona glomerulosa cells (7).
Traditionally, aldosterone biosynthesis is functionally divided into three consecutive phases. (i) In the early mitochondrial steps, cholesterol is transported from intracellular lipid droplets into the outer mitochondrial membrane (OM) and then to the inner mitochondrial membrane (IM). The latter step represents the rate-limiting process in all steroidogenic pathways (8) and is followed by the conversion of cholesterol to pregnenolone by the cytochrome P450 scc enzyme. (ii) The intermediate steps take place on the endoplasmic reticulum and involve the conversion of pregnenolone to progesterone by 3␤hydroxysteroid dehydrogenase isomerase and then to 11-deoxycorticosterone. (iii) The late steroidogenic steps are localized back in the mitochondria and include the formation of corticosterone and its conversion to aldosterone by cytochrome P450 11␤ .
The regulation of intramitochondrial cholesterol transfer by cAMP-dependent mechanisms has been extensively studied (9). While the transport of cholesterol from lipid droplets to the outer mitochondrial membrane was found not to be affected by inhibitors of protein synthesis in ACTH-stimulated adrenal cells, by contrast, the subsequent delivery of cholesterol to the inner mitochondrial membrane has been shown to be blocked by cycloheximide, with a concomitant inhibition of steroid synthesis (10,11). As a consequence, newly synthesized "labile" proteins, among others, have been proposed to be significant mediators of the acute steroidogenic response. In primary adrenal cell cultures (12,13) and in MA-10 mouse Leydig tumor cells (14,15), a family of 30-kDa mitochondrial phosphoproteins are synthesized in response to stimulation with both trophic hormone and cAMP analogs. These studies have demonstrated that the appearance and the amounts of these proteins are highly correlated with the rate of steroidogenesis. Recently, the steroidogenic acute regulatory (StAR) protein has been identified, and its complementary DNA cloned (16 -19). A model has been proposed according to which cholesterol transfer from the outer to the inner membrane might occur via intermembrane contact sites (CS) during the import and proteolytic processing of the 37-kDa precursor of StAR protein into mitochondria (18,19). The most striking evidence for the essential role of StAR protein in the acute regulation of steroidogenesis was provided by studies of a lethal disease, lipoid congenital adrenal hyperplasia, which manifests itself by a complete inability of the newborn infant to synthesize steroids (20). Mitochondria from affected adrenal glands and gonads fail to convert cholesterol to pregnenolone, and as a consequence, cholesterol accumulates within the cells. The defects responsible for this disease are mutations in the StAR gene that generate truncated and nonfunctional proteins (20). In addition to the first report on StAR gene mutations causing lipoid congenital adrenal hyperplasia (20), many additional examples of mutations in StAR gene resulting in this disease are being reported (21)(22)(23) and perhaps it is not as rare as previously thought. To date, mutations in the StAR gene are the only known causes of this potentially lethal disease and, in a most dramatic manner, have demonstrated the indispensable role of StAR protein in the production of steroids.
Our laboratory has recently provided a direct demonstration of the involvement of physiological rises in cytosolic free Ca 2ϩ concentration ([Ca 2ϩ ] i ) in the activation of the early steps of steroidogenesis, namely the stimulation of pregnenolone synthesis in bovine adrenal glomerulosa cells (24). Both Ang II and K ϩ elicit sustained changes in mitochondrial free calcium concentration (25). However, the precise site(s) of action of Ca 2ϩ responsible for this increased pregnenolone formation remained to be determined. We therefore undertook this work to investigate the effect of rises in [Ca 2ϩ ] i on the intramitochondrial distribution of both cholesterol and StAR protein in bovine adrenal glomerulosa cells. Our data show that physiological rises in cytosolic Ca 2ϩ , produced either with a Ca 2ϩ inonophore or with Ang II, are indeed effective in stimulating a specific StAR protein accumulation in the inner membranes as well as a concomitant cholesterol transfer from the outer membranes to the contact sites and inner membranes, where P450 scc and 3␤-HSD are present to initiate the steroidogenic response.

Bovine Adrenal Zona Glomerulosa Cell Preparation
Bovine adrenal glands were obtained from a local slaughterhouse. Zona glomerulosa cells were prepared by enzymatic dispersion with dispase and purified on Percoll density gradients as described in detail (26). Purified glomerulosa cells were resuspended at a density of 10 6 cells/ml in a modified Krebs-Ringer buffer (136 mM NaCl, 5 mM NaHCO 3 , 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 1.8 mM KCl, 1.2 mM CaCl 2 , 5.5 mM D-glucose, and 20 mM Hepes, pH 7.4) and preincubated at 37°C for 1 h before being used in the subsequent experiments.

Calcium Clamping of Bovine Adrenal Glomerulosa Cells
After having been washed in Krebs-Ringer buffer, glomerulosa cells were Ca 2ϩ -clamped as described (24), in the presence of 2 M ionomycin and 1 mM total extracellular Ca 2ϩ , in order to achieve [Ca 2ϩ ] i ϭ 600 nM (high Ca 2ϩ clamp). CHX was used at a final concentration of 1 mM. Control cells were Ca 2ϩ -clamped in Krebs-Ringer buffer without Ca 2ϩ in the presence of 0.2 mM EGTA (low Ca 2ϩ clamp; [Ca 2ϩ ] i Ͻ 100 nM). Aminoglutethimide (500 M) was included in the incubation medium to inhibit cholesterol side chain cleavage. At the end of a 2-h incubation period at 37°C, the cells were sedimented at 200 ϫ g for 15 min. All subsequent operations were conducted at 4°C in buffers containing 500 M aminoglutethimide.

Isolation of Mitochondria and Preparation of Submitochondrial Fractions
Glomerulosa cells were homogenized with a Potter-Elvejhem homogenizer (1200 rpm, 35 strokes) in 5 mM Tris-HCl buffer, pH 7.4, containing 275 mM sucrose. The homogenate was centrifuged at 200 ϫ g for 15 min to remove large debris and nuclei. Further centrifugation of the supernatant at 10,000 ϫ g for 10 min yielded the mitochondria. The mitochondrial pellet was washed twice at 8000 ϫ g with the same buffer.
We have previously shown that sucrose density gradient fractionation of osmotically shocked adrenocortical mitochondria leads to the separation of three distinct membrane populations containing specific marker enzymes (27). Submitochondrial particles were prepared and separated by sucrose density gradient (15-50%, density ϭ 1.06 -1.23) centrifugation. Subsequently, the gradients were divided into 20 fractions of 500 l that were assayed for marker enzyme activities. Pooled fractions 4 -7 (corresponding to the monoamine oxidase activity peak; density ϭ 1.08 -1.11, specific to the outer mitochondrial membranes), 9 -11 (corresponding to the nucleoside-diphosphate kinase activity peak; density ϭ 1.13-1.15, specific to contact sites), and 13-15 (corresponding to the cytochrome c oxidase activity peak; density ϭ 1.17-1.18, specific to the inner mitochondrial membranes) from the original gradient were dialyzed against 5 mM potassium phosphate buffer, pH 7.4, and stored at Ϫ20°C until use. Protein was quantified using the Bio-Rad protein microassay and bovine serum albumin as a standard.
For some experiments, mitoplasts and outer membranes were prepared as follows. The washed mitochondria were exposed to a first swelling by incubation in 20 mM sodium phosphate buffer, pH 7.4, for 20 min at 4°C. Centrifugation at 10,000 ϫ g yielded the outer membranes (supernatant) and a pellet that was subjected to a second swelling in the same buffer (15 min, 4°C). Mitoplasts were pelleted by centrifugation at 10,000 ϫ g, washed, and resuspended in 5 mM Tris-HCl and 250 mM sucrose buffer (pH 7.4).

Cholesterol Determination
The cholesterol content of submitochondrial fractions was determined by a coupled cholesterol oxidase-peroxidase assay with cholesterol as a standard (31). Aliquots of the fractions (200 l) were transferred to glass tubes. To each sample were added 20 l of 20 mM cholate and 1% Triton X-100 in 100 mM potassium phosphate buffer, pH 7.4, followed by the addition of 25 l of 95% ethanol. The reaction mixture containing potassium phosphate buffer (100 mM), pH 7.4, cholesterol oxidase, peroxidase, and p-hydroxyphenylacetic acid was then added to each fraction in a final volume of 1 ml. Assay tubes were incubated for 1 h at 37°C. Cholesterol oxidase generates H 2 O 2 , and peroxidase catalyzes the reaction of H 2 O 2 with p-hydroxyphenylacetic acid to yield a stable fluorescent product. The fluorescence was measured in a Jasco CAF-110 fluorometer (excitation, 325 nm; emission, 405 nm).

Blotting Method and Immunodetection
SDS-polyacrylamide gel electrophoresis-resolved proteins were electrophoretically transferred onto a nitrocellulose membrane (Schleicher & Schuell) according to Towbin et al. (33). Following transfer, the membrane was incubated in blocking buffer (phosphate-buffered saline containing 0.4% Tween 20 and 5% nonfat dry milk) for 1 h at room temperature and then incubated either with antibodies raised by one of us (17) against a peptide fragment (amino acids 88 -98) of the 30-kDa StAR protein or with antisera specific for the steroidogenic enzymes 3␤-HSD and cytochrome P450 scc or for cytochrome c oxidase (1 h in phosphate-buffered saline containing 0.4% Tween). The membrane was thoroughly washed with the same buffer (3 ϫ 10 min) and then incubated for 1 h with horseradish peroxidase-labeled goat anti-rabbit IgG (Bio-Rad). The nitrocellulose sheet was washed as described above, and the antigen-antibody complex was revealed by enhanced chemiluminescence using the Western blotting detection kit and Hyper-ECL film (Amersham, Zü rich, Switzerland). Quantitation of fluorograms was performed using a Molecular Dynamics computing densitometer.

Electron Microscopy
Rat adrenal glands were removed, and small fragments were fixed in 1% glutaraldehyde solution in phosphate-buffered saline for 1 h at room temperature (34). Preparation of the tissue blocks in LR white resin (London Resin Co., Bassingstoke, United Kingdom) was performed according to Newman et al. (35). Briefly, after fixation, the tissue was washed in phosphate-buffered saline and dehydrated in a graded series of up to 70% ethanol. The non-osmicated fragments were infiltrated with LR white resin and then placed in gelatin capsules for polymerization at 50°C for 24 h (34). Thin sections were cut with a diamond knife and collected on nickel grids. Prior to incubation with antiserum, nonspecific antigenic sites were blocked by incubation for 5 min at room temperature with normal goat serum (1:100 dilution) in Tris-HCl/ Tween buffer containing 0.9% NaCl, 10 mM Tris-HCl, pH 8.2, and 0.1% Tween 20. The sections were incubated overnight (4°C) with antisera (1:20 dilution in Tris-HCl/Tween), followed by a 1-h incubation with a 1:10 dilution of 12-nm gold-labeled goat anti-rabbit IgG (Jackson Im-munoResearch Laboratories, Inc., West Grove, PA) in -TrisHCl/Tween. Finally, the sections were stained with 1% aqueous uranyl acetate and lead citrate. The sections were observed and photographed using a Philips 300 electron microscope. Cytochemical controls (data not shown) included omission of the first antibody incubation step. Additional controls included incubation of preimmune serum with the adrenal sections (data not shown).

Analysis of Data
Results are expressed as means Ϯ S.E. The mean values were compared by analysis of variance using Fisher's test. A value of p Ͻ 0.05 was considered as statistically significant. Image analysis of electron microscope micrographs was performed as follows.
Acquisition-Low power electron microscope micrographs (see Fig. 8, A and B) were scanned with a Vista-S8 scanner (UMAX Systems, Taipei, Taiwan) using a Power Macintosh 6100/60 computer (Apple Computer Corp.) at a resolution of 300 dpi and 256 gray-levels. Digitized images were analyzed on a Compaq Prosignia 300 PC using the IPWIN 1.3 software package (Media Cybernetics, Silver Spring, MD).
Processing and Analysis-Regions of interest were marked manually. Analysis of gold particle density was performed within regions of interest that delineated appropriate organelles and areas of interest within the cell. Pixel values within regions of interest were modified using a ␥-correction curve of 9.7. Images were then subjected to a 5 ϫ 5 horizontal edge detection filter. This accentuated the particles while preserving a uniform background. Data loss was negligible (data not shown). Segmentation of particles was performed using particle size, roundness, and gray-level criteria. After segmentation, particles were counted, and their density within regions of interest was determined. Fig. 1 illustrates the monoamine oxidase, nucleoside-diphosphate kinase, and cytochrome c oxidase activities measured in the pooled fractions of OM, CS, and IM isolated from mitochondria of control cells (low Ca 2ϩ clamp). As expected, the OM fraction contained the highest monoamine oxidase activity; the IM fraction was characterized by the highest cytochrome c oxidase activity; and CS displayed monoamine oxidase and cytochrome c oxidase activities in addition to nucleoside-diphosphate kinase activity. IM contained most of the mitochondrial membrane proteins (57.9 Ϯ 7.6% of the total, n ϭ 3), while CS and OM contained 36.3 Ϯ 4.3 and 5.7 Ϯ 1.7% of the total membrane proteins, respectively. Similar profiles of mitochondrial membrane marker enzymes and protein content were obtained following fractionation of mitochondria from high Ca 2ϩ -clamped cells (data not shown).

Characterization of the Submitochondrial Fractions of Bovine Glomerulosa Cells-
Calcium and Ang II Induce a CHX-sensitive Intramitochondrial Cholesterol Transfer-We have previously shown in bovine adrenal zona glomerulosa cells that the calcium ionophore ionomycin can be effectively used at low concentration to clamp the cytosolic free Ca 2ϩ concentration at various physiological levels (50 -1000 nM) (24). These submicromolar [Ca 2ϩ ] i levels stimulate in a concentration-dependent manner the early mitochondrial steps of steroidogenesis as well as aldosterone synthesis (24).
To examine the effect of Ca 2ϩ on intramitochondrial cholesterol distribution, cholesterol content was determined in OM, CS, and IM from control (low Ca 2ϩ -clamped) cells and from high Ca 2ϩ -clamped cells in the presence or absence of CHX. Fig. 2A shows that the stimulation of ionomycin-treated glomerulosa cells with Ca 2ϩ for 2 h led to a marked decrease in cholesterol content in OM (to 65.2 Ϯ 0.2% of controls, n ϭ 4; p Ͻ 0.001), with a concomitant increase in CS (to 143 Ϯ 3.2% of controls; p Ͻ 0.01) and a less pronounced but significant augmentation in IM (to 119 Ϯ 5.1% of controls; p Ͻ 0.05).
When CHX was added concomitantly with Ca 2ϩ , the decrease in cholesterol content in the OM fraction was almost entirely prevented (92.5 Ϯ 5.4% of the cholesterol content in the respective control fraction, n ϭ 3). Simultaneously, CHX reduced significantly the Ca 2ϩ -induced cholesterol increase in CS and IM ( Fig. 2A). Fig. 2B illustrates the effect of Ang II on the cholesterol content of submitochondrial fractions of intact glomerulosa cells. No significant changes were detected in the outer mitochondrial membranes (93.4 Ϯ 1.0% of controls, n ϭ 4). However, the hormone induced a pronounced increase in cholesterol content in the contact site-enriched fractions (151.1 Ϯ 18.1% of controls, n ϭ 4; p Ͻ 0.05). Finally, a less pronounced increase in cholesterol content was observed in the inner membrane fractions (124.5 Ϯ 6.5% of controls, n ϭ 4; p Ͻ 0.05).
The accumulation of cholesterol in mitochondrial contact sites that occurred during Ang II stimulation was entirely prevented by cycloheximide (Fig. 2B). Furthermore, there was a tendency toward an increase in cholesterol content in the outer mitochondrial membranes (110.3 Ϯ 20.3% of controls with Ang II alone, n ϭ 3).
Ca 2ϩ Stimulates StAR Protein Expression in Bovine Glomerulosa Cells-The 30-kDa mitochondrial StAR protein has been recently shown to be required in the acute regulation of steroid synthesis (18,19). To determine whether an increase in [Ca 2ϩ ] i affected StAR protein expression in bovine glomerulosa cells, mitochondrial proteins from Ca 2ϩ -clamped cells were analyzed by immunoblotting. Fig. 3A shows the results obtained with one representative experiment. The anti-StAR protein antibodies recognized a protein doublet of ϳ30 kDa in mitochondria from both control cells and high Ca 2ϩ -clamped cells. Densitometric analysis of this particular Western blot revealed that Ca 2ϩ increased StAR protein content to 210% of controls. CHX significantly inhibited the Ca 2ϩ -induced increase in StAR protein by 62%. In Fig. 3B, the mean results from the quantitation of the StAR protein signal in mitochondria from five separate experiments are represented. Ca 2ϩ increased significantly mitochondrial StAR protein content to 159 Ϯ 23% of controls (p Ͻ 0.05). This effect of Ca 2ϩ was prevented by CHX (116 Ϯ 8% of controls).
Ca 2ϩ and Ang II Selectively Increase StAR Protein Content in Inner Mitochondrial Membranes-We next examined the submitochondrial distribution of StAR protein in adrenal glomerulosa cells. Interestingly, no StAR protein was detected in the outer membrane (data not shown). By contrast, as shown in Fig. 4A, StAR proteins were detected by the antibody in mitochondrial CS and IM from both control (low Ca 2ϩ -clamped) and high Ca 2ϩ -clamped cells. Densitometric analysis of the immunospecific bands (Fig. 4B) revealed that no significant changes were observed in StAR protein content in mitochondrial CS from high Ca 2ϩ -clamped cells as compared with controls. By contrast, Ca 2ϩ increased StAR protein content by 4-fold in IM, indicating that the increase in StAR protein induced by Ca 2ϩ was specifically targeted to the inner mitochondrial mem-

FIG. 3. Immunodetection of StAR protein in mitochondria of Ca 2؉ -clamped glomerulosa cells.
A, shown is a Western blot from one representative experiment. Mitochondria were isolated from control cells or high Ca 2ϩ -clamped cells incubated in the absence or presence of 1 mM CHX. For each sample, 10 g of mitochondrial proteins were separated by SDS-polyacrylamide gel electrophoresis as described under "Experimental Procedures." B, the immunospecific bands for the 30-kDa StAR proteins were quantitated by densitometry in five independent experiments. C, control (low Ca 2ϩ clamp); Ca 2ϩ , high calcium clamp (600 nM); Ca 2ϩ ϩ CHX, high calcium clamp in the presence of cycloheximide; IOD, integrated optical density.
Qualitatively similar results were obtained with Ang II (Fig.  5B). The hormone induced a significant increase in IM StAR protein to 150 Ϯ 23% of controls (n ϭ 3, p Ͻ 0.05), while having no effect on StAR protein content in CS.
The Ca 2ϩ -and Ang II-induced Increase in StAR Protein Content in Inner Mitochondrial Membranes Is Sensitive to CHX-When bovine adrenal glomerulosa cells were subjected to a high Ca 2ϩ clamp in the presence of CHX, immunoblot analysis of inner mitochondrial membrane and contact site proteins revealed that CHX completely prevented the Ca 2ϩinduced increase in StAR protein in IM (Fig. 5, A and B), while no significant changes occurred in contact sites (data not shown). Similarly, the increase in IM StAR protein content observed after Ang II challenge was not observed in the presence of CHX (Fig. 5B).
Cytochrome P450 scc and 3␤-HSD Are Found in Mitochondrial Contact Sites and Inner Membranes-To determine whether StAR protein induction by Ca 2ϩ and Ang II was specific, submitochondrial fractions were analyzed by immunoblotting using anti-3␤-HSD or anti-P450 scc antibodies since we had shown previously that these two enzymes are co-localized within the mitochondria of bovine fasciculata cells (27). The increase in the synthesis of these two enzymes is known to require long-term exposure to steroidogenic hormones (9). Fig.  6 shows that mitochondrial contact sites and inner membranes from bovine glomerulosa cells contain significant amounts of 3␤-HSD and P450 scc . In addition, no changes in 3␤-HSD and P450 scc content occurred in these submitochondrial fractions with Ang II stimulation. Finally, CHX did not affect the level of these proteins in CS and IM. Identical observations were made in bovine glomerulosa cells subjected to a high Ca 2ϩ clamp (data not shown).
The presence of 3␤-HSD in inner mitochondrial membranes from bovine glomerulosa cells was further confirmed by preparing mitoplasts. The determination of enzyme marker activities confirmed that mitoplasts were practically devoid of any contamination with microsomes. Indeed, as can be seen in Table I, while cytochrome c oxidase activity was 10-fold higher in mitoplasts than in outer membranes, NADPH-cytochrome c reductase activity, a marker of the endoplasmic reticulum, was undetectable in mitoplasts and high in microsomes. Upon immunoblot analysis (Fig. 7A), a strong 3␤-HSD signal was detected in mitoplasts as well as in microsomes, thus confirming the presence of the enzyme inside the mitochondria. In fact, the 3␤-HSD signal in mitoplasts was considerably higher than in outer membranes in spite of the fact that the latter displayed a higher NADPH-cytochrome c reductase activity.
To examine whether the intramitochondrial localization of 3␤-HSD is a common characteristic of steroidogenic cells, mitochondria from MA-10 mouse Leydig cells were fractionated on a sucrose gradient as described above for adrenal mitochondria. As shown in Fig. 7B, immunoblot analysis revealed the presence of the 3␤-HSD protein in CS and IM, but not in OM.
Immunocytochemical Localization of 3␤-HSD-As an additional proof of the presence of 3␤-HSD within steroidogenic mitochondria, the ultrastructural localization of 3␤-HSD was studied by the immunogold staining technique in rat adrenal fasciculata cells (17,34). Fig. 8 shows that 3␤-HSD antigenic sites are clearly abundant in the mitochondria of these cells. The immunovisualization technique also confirms that, as expected, Ͼ60% of this enzyme is anchored to the membranes of the endoplasmic reticulum ( Fig. 8C and Table II), where the apparent density of immunoreactive 3␤-HSD is 56% higher than that observed in the mitochondria (Table II). Higher enrichment of membrane-bound 3␤-HSD was observed in stacking of smooth endoplasmic reticulum shown in Fig. 8C. Quantitation of P450 scc labeling, which serves as a classical mitochondrial marker, shows a dramatically reciprocated ratio, with the density of this enzyme per mitochondrial surface being 10 times higher than that observed in the cytoplasm (Table II).
The intramitochondrial pattern of 3␤-HSD localization (Fig.  8D) was monitored using an analytical approach similar to that previously applied to P450 scc and StAR protein patterns (34,36). Fig. 9 depicts the compartmentalization patterns of the three key steroidogenic proteins at a submitochondrial resolution. Clearly, whereas P450 scc antigenic sites were exclusively anchored to the crista membranes protruding into the mitochondrial matrix (34), the labeling of 3␤-HSD sites partitioned in all five compartments of the organelle. Nevertheless, nearly 60% of 3␤-HSD labeling was associated with the crista membranes, whereas the rest of the antigenic sites were equally distributed between the outer membranes, the mitochondrial matrix, and the intermembrane spaces. To some extent, this profile of 3␤-HSD sites also differed from that obtained for StAR protein, the majority of which was confined to the intermembrane spaces of the vesicular cristae (Fig. 9).

DISCUSSION
In a previous study on the mechanisms of activation of aldosterone biosynthesis by Ca 2ϩ , we have narrowed the potential target domain for this messenger to the early steps of the steroidogenic pathway (24). Using the ionomycin-mediated Ca 2ϩ clamp and Ang II as activators of steroid synthesis in bovine glomerulosa cells, we have recently shown that physiological [Ca 2ϩ ] i changes acutely stimulate cholesterol redistribution across mitochondrial membranes (37). We now demonstrate that this cholesterol transfer from the outer mitochondrial membranes to both contact sites and inner membranes is associated with a selective increase in StAR protein content in inner mitochondrial membranes and that both these responses require protein synthesis.
The major finding of this work is the clear-cut and specific increase in the mature form(s) of the 30-kDa StAR protein in the inner mitochondrial membranes of bovine glomerulosa cells under stimulation of the calcium messenger system. To our knowledge, this is the first demonstration of a submitochondrial modulation of StAR protein by Ca 2ϩ . Indeed, most studies that have investigated the regulation of the 30-kDa proteins involved in the acute steroidogenic response have used total cellular or mitochondrial protein extracts and have focused on trophic agents activating adenylyl cyclase and cAMP production (16,38,39). Recently, Clark et al. (40) have shown that calcium-mobilizing agonists such as Ang II, K ϩ , and the dihydropyridine agonist Bay K8644 produce significant increases in the cellular levels of StAR protein in H295R human adenocarcinoma cells within 4 -6 h. Our results extend this observation by confining the StAR protein increase to a submitochondrial compartment and by showing that it can be observed after shorter stimulation times (2 h), thus confirming the acute role of StAR protein in the initiation of the steroidogenic response.
In MA-10 Leydig tumor cells and in the human H295R adrenocarcinoma cell line, StAR protein appears as a single band upon immunoblot analysis (16,40), whereas in bovine adrenal glomerulosa cells, we detected a StAR protein doublet around 30 kDa. While the nature of this protein doublet remains to be elucidated, it is worth mentioning that a similar pattern has been observed in bovine fasciculata cells (41). In MA-10 mouse Leydig tumor cells, phosphorylation of StAR  protein on a threonine residue is required for the acute induction of steroidogenesis (42). The doublet we observed may therefore represent phosphorylated and non-phosphorylated forms of StAR protein. These proteins could also be related to the 28.5-30-kDa proteins shown by Elliott et al. (43) to be induced by Ang II in bovine glomerulosa cells.
Remarkably, Ca 2ϩ and Ang II induced StAR protein accumulation exclusively in the inner mitochondrial membranes. Interestingly, although StAR protein was present in contact sites, the increase in cholesterol induced by [Ca 2ϩ ] i rises was not associated with a concomitant increase in StAR protein content, indicating that StAR protein is only transiting via those structures. Moreover, we could not detect the 37-kDa precursor of StAR protein in any of the submitochondrial fractions. These observations confirm the short half-life of the StAR protein precursor, which is known to be rapidly processed by mitochondrial proteases (17), and indicate that the inner mitochondrial membranes constitute the final destination of mature StAR protein. The present biochemical evidence is therefore in agreement with our visualization of StAR protein localization in adrenal mitochondria labeled by immunogold staining (17).
Early studies in adrenocortical cells have demonstrated that the inhibitor of protein translation, cycloheximide, prevents the increase in the rate of steroid synthesis induced by ACTH (10,11). The cycloheximide-sensitive step in steroidogenesis has been located to the ACTH-activated cholesterol transport to the inner mitochondrial membrane (10). In adrenal glomerulosa cells, Elliott and Goodfriend (44) have reported that the cycloheximide-sensitive step in Ang II-stimulated aldosterone synthesis is mitochondrial pregnenolone synthesis, the first event in steroidogenesis following cholesterol supply to cytochrome P450 scc . Activators of steroidogenesis mobilizing either the cAMP or the Ca 2ϩ messenger system have been shown to induce a net accumulation of total cellular StAR protein (19,40), and we have obtained similar data in bovine glomerulosa cells (data not shown). These findings indicate that cycloheximide is most likely to act at a translational level. Indeed, our data show that both StAR protein accumulation and intramitochondrial cholesterol transfer induced by [Ca 2ϩ ] i rises are highly sensitive to cycloheximide, strongly suggesting that the blockade of StAR protein synthesis prevents cholesterol mobilization to the inner membranes. Taken together, these results provide strong correlative evidence that the increase in StAR protein expression and its targeting to the mitochondrial membranes are linked to cholesterol redistribution from the outer to the inner membranes induced by calcium-mobilizing agents in adrenal glomerulosa cells.
Our current model of the mechanisms of calcium-induced activation of steroidogenesis would favor a dual site of action for Ca 2ϩ : in addition to an obligatory role for Ca 2ϩ influx into the mitochondria, as demonstrated previously in permeabilized glomerulosa cells (45) and in glomerulosa cells treated with a blocker of the mitochondrial Na ϩ /Ca 2ϩ exchanger (25), the present cycloheximide results imply an effect of cytosolic Ca 2ϩ on StAR protein expression.
The effect of increased [Ca 2ϩ ] i is shown to be specific for StAR protein since the levels of two additional key steroidogenic enzymes in the mitochondria, cytochrome P450 scc and 3␤-HSD, were not affected under similar experimental manipulations. It should be noted that whereas P450 scc is a typical nuclearly encoded mitochondrial protein, 3␤-HSD has been  Fig. 8 (A and B) were scanned and analyzed as described under "Experimental Procedures." The localization of the gold particles was categorized as shown, and the relative area of each cellular compartment was defined. The particle number (percent in parentheses) and the density per unit area are shown. a ER, endoplasmic reticulum; cytoplasm, defined as such when ultrastructural preservation did not allow a clear definition of the endoplasmic reticulum; fat, lipid droplets as shown in Fig. 8 (A and B); ICS, intercellular spaces.
FIG. 9. Intramitochondrial distribution of colloidal gold-labeled 3␤-HSD, P450 scc , and StAR protein. Colloidal gold particles were manually counted in 6 -10 different mitochondria of rat adrenal fasciculata cells stained with rabbit antisera to bovine 3␤-HSD, rat P450 scc (based on previously published data (34)), and mouse StAR (based on previously published data (17)) (787, 587, and 443 particles, respectively). The localization of the particles in five different intramitochondrial compartments was categorized using high power photomicrographs (magnification ϫ 50,575) as described for Fig. 8D. Data are presented as percentage (mean Ϯ S.E.) of gold particles in each of the submitochondrial compartments. IMS, intermembrane spaces. considered as an enzyme anchored in the endoplasmic reticulum (46), with a few exceptions (47). Yet, we have previously provided evidence showing that P450 scc and 3␤-HSD co-localize in the inner membranes and contact sites of bovine adrenocortical zona fasciculata mitochondria (27), and their association into a macromolecular complex retains the catalytic activities of both enzymes (48). We report here that cytochrome P450 scc and 3␤-HSD also coexist in the contact sites and inner membranes of zona glomerulosa mitochondria. In addition, immunogold staining of rat adrenal fasciculata mitochondria, as well as immunoblot analysis of MA-10 mouse Leydig submitochondrial fractions, clearly confirmed the intramitochondrial localization of 3␤-HSD. The consistency of this finding, observed with different techniques in various steroidogenic cell types from three different species, speaks in favor of a common feature of 3␤-HSD distribution in all steroid-producing cells. Although the N terminus of the 3␤-HSD protein does not appear to include a mitochondrial targeting sequence (49), various proteins of the inner mitochondrial membrane, such as the ADP/ATP carrier (50), porin (51), and BCS1 (52), appear to contain important targeting information either internally or at their C terminus, rather than at the N terminus (53). The intramitochondrial presence of 3␤-HSD could add a facilitating step of important functional significance to the Ca 2ϩ -mediated cholesterol transfer: channeling pregnenolone to progesterone should be highly favored by 3␤-HSD located in close proximity to P450 scc and may actually represent a major pathway, in line with the fact that pregnenolone is the preferential substrate of mitochondrial 3␤-HSD (27). In this regard, it is worth mentioning that the existence of multihormonal enzyme complexes ("hormonads") has already been proposed some years ago by Lieberman and Prasad (54) and Prasad et al. (55).
Collectively, our biochemical approaches, together with the ultrastructural analysis showing that Ͼ85% of P450 scc and 60% of mitochondrial 3␤-HSD are anchored to the inner crista membranes of this organelle, raise the possibility of a supramolecular organization of P450 scc and 3␤-HSD into a functional unit directly catalyzing the metabolism of cholesterol to pregnenolone and to progesterone following StAR protein-mediated cholesterol supply via contact sites.