Characterization of the rat Star gene that encodes the predominant 3.5-kilobase pair mRNA. ACTH stimulation of adrenal steroids in vivo precedes elevation of Star mRNA and protein.

The steroidogenic acute regulatory protein (STAR) participates in steroidogenesis through the mitochondrial transfer of cholesterol to cytochrome P450scc. The rat adrenal Star gene is transcribed as a 3. 5-kilobase pair (kb) and 1.6-kb mRNA with the larger mRNA predominating ( approximately 85% of total) in vivo. Hypophysectomy (HPX) produced a 3-5-fold decrease in Star mRNA along with a loss of adrenal steroids, whereas P450scc mRNA decreased by less than 2-fold. Adrenocorticotropic hormone (ACTH) treatment of HPX rats maximally stimulated steroidogenesis rates within 5 min with over 10-fold elevation of steady state blood levels occurring within 10 min. For intact rats there was a 5-10-fold larger increase, paralleling previously observed elevations of cholesterol-cytochrome P450scc association and metabolism in subsequently isolated adrenal mitochondria. ACTH did not increase either total STAR protein or a group of modified forms until at least 30 min after completion of acute stimulation, indicating that elevated translation of STAR protein cannot alone mediate this acute stimulation. Parallel slow changes in STAR protein and corticosterone formation after ACTH treatment are consistent with participation of STAR forms as co-regulators of these hormonal responses. ACTH stimulation of HPX rats increased Star mRNA by 2.5-fold within 20 min and by 4.5-fold after 1 h, thus preceding the rise in the STAR protein. A 3.5-kb Star cDNA clone isolated from a rat adrenal cDNA library exhibited a 0.9-kb open reading frame and a 2.5-kb 3'-untranslated region (3'-UTR). The open reading frame sequence differed at only 12 amino acids from that of the mouse Star. The rat Star gene seven exons with exon 7 encoding the entire 2.5 kb of 3'-UTR of the 3.5-kb mRNA. The 3'-UTR sequence suggests that 1.6- and 3.5-kb mRNA are formed by an alternative usage of different polyadenylation signals. Multiple UUAUUUA(U/A)(U/A) motifs also suggest additional regulation through this extended 3'-UTR. Although elevation of STAR protein by ACTH does not cause the acute increase in adrenal cholesterol metabolism, changes in the turnover or distribution of an active STAR subfraction cannot be excluded.

The conversion of cholesterol to pregnenolone is the first step of steroid synthesis and is catalyzed by cytochrome P450 side chain cleavage enzyme (P450scc) 1 localized on the matrix side of the inner mitochondrial membrane (1,2). This conversion is the rate-limiting step in steroidogenesis and is hormonally activated (3). The availability of cholesterol to P450scc which limits this conversion (4,5) requires hormonal activation of cholesterol mobilization to the mitochondria and then trophic hormone-dependent transport of cholesterol from mitochondrial outer membrane to inner membrane (6,7). This latter process is blocked within 10 min by protein synthesis inhibitors, such as cycloheximide, which also correlate with the loss of steroid synthesis after removal of ACTH (8,9). A series of phosphoproteins (30)(31)(32)(33)(34)(35)(36)(37) have been identified in cultured adrenal cells that localize to the mitochondria and increase in response to hormone stimulation (10,11). The larger precursor forms are rapidly removed by cycloheximide concomitant with mitochondrial proteolysis (10,11). It has been proposed by Epstein and Orme-Johnson (12) that the effects of these proteins account for the cycloheximide sensitivity and hormonal stimulation in intermembrane cholesterol transport. Subsequent studies from several laboratories have shown that these changes in 30-kDa proteins occur in adrenal, testis, and ovarian cells of several species (13)(14)(15)(16)21). The peripheral benzodiazepine receptor that is localized in the adrenal mitochondrial outer membrane is also essential for this step, suggesting the involvement of a complex multistep process (17,18).
Recently, a cDNA encoding the 30-kDa protein has been cloned and renamed the Steroidogenic acute regulatory protein (STAR) (19). Several sites for phosphorylation by cAMPdependent protein kinase have been recognized in the sequence which otherwise shows no similarity to other proteins (20). Star expression in COS-1 cells that have been previously transfected with expression vectors encoding P450scc and adrenodoxin further enhances cholesterol metabolism (21). Recent work shows that this STAR activity in COS-1 cells is independent of the mitochondrial uptake that is seen in steroidogenic cells (22). Mutated STAR has been found in humans born with the steroid deficiency disease, congenital lipoid adrenal hyperplasia (23)(24)(25). In these cases C-terminal truncation of the protein removes steroidogenic activity. Disruption of the Star gene in mice produces a similar phenotype (58). Star has been detected in all steroidogenic tissues except for brain and placenta (26).
Two species of Star mRNA (1.6 and 3.4 kb) are stimulated by cAMP in cultured mouse Leydig tumor cells (26). The regulation of the relative proportions of these Star mRNA species may be important in determining the post-transcriptional activation that has been implicated in the rapid hormone stimulation of steroidogenesis (19 -21). Here we have investigated acute effects of ACTH on levels of Star mRNA and protein in the rat adrenal gland in vivo under conditions that exhibit ACTHenhanced transfer of cholesterol to P450scc within adrenal mitochondria (4 -6). We have particularly addressed the kinetics of the activation of steroidogenesis in relation to changes in Star mRNA and protein, in particular whether changes in STAR are sufficiently fast to account for the acute response.
Although the gene encoding STAR has been cloned from both mouse (19) and human cells (21), we know little about the mechanism of transcriptional and translational regulation. The present structures of the mouse and human STAR gene only account for the expression of a 1.6-kb Star mRNA even though steroidogenic cells show extensive expression of a 3.5-kb mRNA. Here we show that this longer mRNA is the predominant species in rat adrenals in vivo, thus raising questions about this extended sequence. Several elements in the 3Ј-UTRs are targets for RNA-binding proteins which regulate their processing, translational efficiency, and stability (27,28). Here we describe an extended structure for the Star gene encoding the complete rat 3.5-kb mRNA that is a key step toward characterizing the acute control of the activation of this protein. We show that the extended sequence indeed contains motifs that may affect regulation of this Star mRNA.

MATERIALS AND METHODS
Animal Treatment-Intact and hypophysectomized female Sprague-Dawley rats (175-200 g) were obtained from Harlan Sprague-Dawley (Madison, WI). All animal treatments were performed at 10:00 a.m. Animals were kept in stress minimizing environment. Each rat was then injected with ACTH. Intact rats were treated intraperitoneally with 4 units/rat of Cortrosyn (Organon, West Orange, NJ) up to 20 min prior to sacrifice (20-min group), or subcutaneously with 4 units/rat of Acthar gel (Rhone-Pouleuc Rorer Pharmaceuticals Inc., Collegeville, PA) 24 h prior to sacrifice (24-h group), or subcutaneously (4 units of Acthar gel/rat/day) for 3 consecutive days (3-day group). Hypophysectomized rats were allowed to recover after surgery for 3 days when they were maintained on Purina Lab Chow diets supplemented with 5% sucrose solution. A group of rats was given 4 units of Cortrosyn intraperitoneally for stimulations of up to 20 min prior to sacrifice. For 1 h stimulation, a second injection was given after 30 min to compensate for in vivo degradation of ACTH. For prolonged administrations ACTH was injected subcutaneously as Acthar gel (4 units) 24 h prior to sacrifice. In 24-h groups a second subcutaneous injection of 4 units of Acthar gel was given 3 h prior to sacrifice. All animals were sacrificed by decapitation, and blood was collected into vacutainers containing 15% EDTA, and adrenals were surgically removed, defatted, and rapidly homogenized in TRIzol reagent (Life Technologies, Inc.) Adrenal protein and RNA fractions from this treatment were analyzed for STAR protein and mRNA (see below).
Determination of Plasma Corticosterone-Steroids were extracted from plasma with ethyl acetate/acetone (2:1). Cortisone was used during extraction as an internal standard. The organic phase was then evaporated under nitrogen with heating at 37°C. The residue was dissolved in 100 ml of the methanol and subjected to high pressure liquid chromatography analysis using a reverse-phase C18 column (Beckman, 5 m, 4.6 mm ϫ 25 cm) with a linear gradient of methanol (50 -100%). Detection was carried out at 254 nm by a Beckman System Gold programmable Detector Module 166, and the data were analyzed by System Gold Software via a Beckman System Gold Analog Interface Module 406.
Analysis of STAR Protein-Two-dimensional gel electrophoresis was performed as described previously (29) using the Millipore Investigator Two-dimensional Electrophoresis System. In brief, tube gels (200 ϫ 1 mm) were prepared with isoelectrofocusing gel solution (ESA, Inc., Chelmsford, MA), broad range ampholytes (Millipore, pH 3-10), and 10% ammonium persulfate. The protein fraction from rat adrenal tissue homogenized in TRIzol reagent (50 -100 mg of tissue/1 ml of reagent; Life Technologies, Inc.) was obtained according to the protocol from Life Technologies, Inc. The protein pellet was washed with ethanol and then solubilized with 1% SDS solution. Protein concentration was determined by BCA method, according to the manufacturer's instructions (Pierce) using bovine serum albumin as a standard. Protein samples were mixed with an equal volume of a buffer containing 10 mM Tris, 120 mM dithiothreitol, 0.06% SDS, 3.2% (v/v) Nonidet P-40, 1.8% ampholytes, and 8 M urea (all reagents from ESA, Inc.). 80 g of protein was loaded on the tube gel and subjected to electrofocusing at 1000 V for 18 h. Gels were extruded from the tubes and equilibrated for 2 min in a 373 mM Tris (pH 8.6), 50 mM dithiothreitol, 3% SDS, and 0.02% bromphenol blue buffer. Tube gels were placed on top of slab gels, consisting of 375 mM Tris (pH 8.8), 0.1% SDS, and 12.5% Duracryl. Molecular weight standards, prepared in 1% agarose, were applied on the top of the slab gel at either end of the tube gel. Gels were run at a constant power of 16 watts/gel for 5 h. After electrophoresis, the proteins from the gel were transferred to the nitrocellulose membrane. The position of the molecular weight markers was determined by staining the membrane with Ponceau S. Western immunoblot analysis (30) was completed, using a rabbit antibody (1:6000) raised against a peptide sequence of mouse Star, kindly provided by Drs. N. Boujrad and V. Papadopoulos (Georgetown University, Washington, D. C.), followed by the secondary horseradish peroxidase-conjugated goat anti-rabbit antibody (Promega, Madison, WI). Proteins were visualized by exposing the nitrocellulose membranes following treatment with the ECL detection reagent (Amersham Pharmacia Biotech), to the ECL x-ray film (Amersham Pharmacia).
Isolation of Rat Star cDNA-Construction of rat cDNA library in ZAPII was performed as described previously (31). The coding region of rat Star cDNA was cloned by PCR using pfu DNA polymerase (Stratagene, La Jolla, CA) with primers designed against mouse Star cDNA sequences 15-34 and 973-992 (19). The 0.9-kb amplified product was then subcloned into the SmaI site of the pBluescript SK(ϩ) plasmid (Stratagene, La Jolla, CA). The nucleotide sequence was analyzed by the dideoxy method using Sequenase Version II kit (U. S. Biochemical Corp.), and sequence comparison with reported mammalian Star genes was completed using the GCG program (Genetics Computer Group, Inc.). 5 ϫ 10 5 plaques of the cDNA library were screened using the 0.9-kb rat Star cDNA as a probe. The probe was labeled by the random primer method (Prime-It kit II, Stratagene) using [␣-32 P]dCTP (3000 Ci/mmol) (NEN Life Science Products) and nitrocellulose (Schleicher & Schuell) filter replicas for hybridization. From more than 100 positive clones, 18 were randomly selected and excised into the pBluescript vector and then characterized. cDNA clones were isolated and purified using the Qiagen plasmid kit (Qiagen Inc.). The dye terminator cycle sequencing reaction was carried out for both strands using GeneAmp PCR System 9600 (Perkin-Elmer).
Rat Star Genomic DNA Cloning-5 ϫ 10 5 plaques of the WiStar Furth rat genomic DNA library were screened using a 0.9-kb rat Star cDNA as described above. From 10 positives, 5 clones were selected and characterized. Five clones were amplified by the standard plate lysate and liquid culture method, prepared by the polyethylene glycol method (32), and then purified by Qiagen plasmid kit. The gene structure was verified by PCR using a set of specific primers derived from the 0.9-kb cDNA sequence (Table I). For determination of genomic sequence and, specifically, the sequences for introns and exon-intron junctions, primers were designed against sequences from the ends of each exon. We assumed a distribution of ORF sequences between exons equivalent to the mouse sequence (see Table I and Fig. 6). PCR products were purified using a QIAEX-gel extraction kit and then directly sequenced using GeneAmp PCR System 9600. The gene sequences from within each exon were compared with those of rat Star cDNA.
Subcloning of a 5Ј-Flanking Region of the Rat Star Gene-Restriction digestion of the five genomic clones was carried out with a set of restriction enzymes, followed by Southern blotting using a probe synthesized against a partial 5Ј-flanking region (ϳ150 bp) of the rat Star gene. A 4.4-kb DNA fragment produced by SacI digestion of one genomic clone was recognized by the 5Ј-flanking region cDNA probe, was subcloned, and was partially sequenced. A dye terminator cycle sequencing reaction was carried out for both strands of the construct using the GeneAmp PCR System 9600.
Primer Extension Study-Rats were injected with long acting ACTH 24 h prior to sacrifice. Total RNA (20 g) was then prepared from the adrenal glands with TRIzol reagent. A reverse primer [5Ј-CCC AGC ACA CAG CTT GAA TGT AGC TAG TAA-3] was synthesized against bases 4 -33 downstream to the translation initiation codon, ATG. The extended product was extracted with phenol/chloroform and precipitated with 100% ethanol. The pellet was washed with 70% ethanol and briefly dried for a few minutes and then dissolved in 5 ml of TE buffer (pH 7.4) with 5 ml of formamide loading buffer and kept at Ϫ20°C until use. The sample was heated with boiling water for 3 min just before loading and analyzed on a 6% sequencing gel together with the samples for dideoxynucleotide sequencing using 32 P-end-labeled primer. 5Ј-End labeling of the primer with [␥-32 P]ATP was conducted with T4 polynucleotide kinase, and the labeled primer (5 ϫ 10 4 cpm) was mixed with RNA. Hybridization was performed at 30°C overnight, and primer extension reaction was performed at 42°C for 90 min with 100 units of reverse transcriptase. The same labeled primer (10 ng) was used for the dideoxynucleotide sequencing reaction using a construct containing rat Star 5Ј-flanking region as a template. Primer extension product was then loaded on a denaturing sequencing gel in parallel to the DNA sequencing ladder.
Analysis of Star mRNA-Northern hybridization was conducted with either the 32 P-labeled 0.9-kb Star ORF cDNA after agarose-formaldehyde gel (1%) electrophoresis. Quantitation of Northern hybridization was performed by laser densitometry using the NIH image program.
Rat Star mRNA species containing poly(A) ϩ tails were identified by using a nested RT-PCR according to the method reported previously (33). Briefly, cDNA was synthesized using Superscript reverse transcriptase (Life Technologies, Inc.), 2 g of total RNA isolated from untreated rat adrenal gland, and an adapter primer, 5Ј-TAC GCC AAG CTC GAA ATT AAC CCT CAC TAA AGG G(T) 16 -3Ј. Long distance PCR was then performed using Taq polymerase under previously published conditions (34,35). We used a set of 4 primers as follows: the Star exon 1-specific primer 5Ј-CTG CAG CAC TAC CAC AGA AAG CAT-3Ј and adapter-specific primer 5Ј-TAC GCC AAG CTC GAA ATT AAC CCT C-3Ј, followed by a second nested PCR with Star exon 1-specific primer 5Ј-CTA CAT TCA AGC TGT GTG CTG GGA-3Ј and adapter-specific primer 5Ј-TCG AAA TTA ACC CTC ACT AAA GGG-3Ј in the presence of RNasin. After incubation at 37°C for 1 h, reverse transcriptase was inactivated by boiling for 5 min followed by vigorous shaking to inactivate RNasin. Single strand cDNA was then obtained by digestion of template mRNA with RNase A. The product was dissolved in 50 ml of TE buffer (pH 7.4). The first PCR reaction was conducted using onetenth of the product as a template. PCR was performed according to the automatic program designed as follows: denaturation, 94°C for 3.5 min; cycle denaturation, 98°C for 7 s; annealing, 65°C for 1 min; autosegment extension, 72°C, 4 min ϩ 10 s/cycle; 30 cycles. At the end of the last cycle, the tubes were held at 70°C for 10 min for elongation reaction and then cooled down to 4°C. The second nested PCR reaction was performed with a 100 ϫ diluted sample of the first PCR product as a template, using the conditions and primers described above.

Expression of STAR Protein in Relation to Steroidogenesis
following ACTH Stimulation-We have characterized the expression of immunodetectable STAR protein in relation to the time course of steroid synthesis following ACTH stimulation of rats in vivo. Fig. 1 shows the time course of the increase in corticosterone synthesis in hypophysectomized (HPX) and intact rats. Similar to previous studies, we show that removal of ACTH through HPX lowers circulating adrenal steroid levels by over 10-fold. Administration of ACTH to the HPX rats produced a biphasic steroidogenic response, including an acute phase which is complete in 20 min and a chronic phase which occurs between 1 and 24 h post-treatment. In the acute phase there is a short lag (3-5 min) followed by a rapid increase to a steady state that is reached in 10 min and is maintained at least until 1 h after stimulation. The chronic phase corresponds to a further 3.5-fold increase in serum glucocorticoids, which occurs between 1 and 24 h after stimulation. In intact rats the acute phase kinetics are retained but with much higher amplitude of the response. The change to a new steady state at 20 min corresponds to a 3-fold increase in the rate of secretion of glucocorticoids relative to the basal secretion rate.
We analyzed the effect of ACTH treatment on STAR protein by carrying out immunoblotting from two-dimensional electrophoresis gels of total adrenal protein. This approach has been used previously to resolve modified forms of rat STAR, notably phosphorylated forms, which locate at a more acidic pH on these gels (52). Prior to ACTH treatment of HPX and intact rats, we detected two proteins whose pI (pI 6.5 and 6.6) and size (30 kDa) are fully compatible with previous determinations for rat STAR in unstimulated cultured adrenal cells (Fig. 2) (13,52). In the same animals that were used for these steroid measurements, the STAR protein levels were 2-4 times lower in HPX rats but showed absolutely no response to ACTH within the 20-min period of the acute phase response (Fig. 2B). Increases in total STAR were only apparent 60 min after stimulation (data not shown) and rose 3-4-fold by 24 h corresponding to the period of chronic changes in steroidogenesis. During this period, initial pair of STAR protein increased while an additional pair of low pI STAR forms (pI 6.2-6.3) become apparent to an extent equal with elevated levels of the initial forms. The position of these proteins is consistent with phosphorylated forms of STAR identified in other studies by 32 P labeling (12,41). These data establish that increased levels of STAR do not begin until substantially after the acute phase response. However, the slow changes in STAR levels in the chronic phase 1-24 h after ACTH stimulation of HPX rats were paralleled by a comparable increase in the serum corticosterone level. This methodology did not detect the labile 37-and 32-kDa precursor fors that have been detected previously by 35

S labeling.
Regulation of Two Rat Star mRNA Species by cAMP-To characterize the responsiveness of Star mRNA in vivo to hormonal stimulation, we generated a rat Star cDNA by PCR from a rat adrenal cDNA library. This provided a sequence that was highly similar to Star sequences from mouse (26), human (21), and bovine (16) (Fig. 3). This 0.9-kb rat Star cDNA was used as a probe for Northern hybridization of rat adrenal mRNA (Fig.  4A). Two mRNA species were detected with sizes of 1.6 and 3.5 kb that were similar to those previously reported for MA-10 cells (26). The amount of the larger mRNA was consistently much higher than the 1.6-kb mRNA (ratio 5.1 Ϯ 0.6 for three separate isolations), paralleling previous findings for in vivo measurement of Star mRNA for bovine (16) and ovine (15) luteal tissue. HPX (3 days) diminished the level of both Star messages (3-5-fold in several experiments) compared with intact rats, whereas the ratio of the two mRNA species was not affected (Fig. 4A).
Acute treatment of HPX rats markedly induced both Star mRNAs within 20 min (2.5-fold) and by 4.5-fold after 60 min (Fig. 4B). The level of Star mRNA in intact rats was approximately comparable to this acute stimulation of HPX rats. A 20-min treatment of intact rats with ACTH did not elevate Star mRNA (Fig. 4A). Intact rats treated with ACTH for 3 consecutive days also showed selective expression of the 1.6-kb species relative to 3.5-kb species (Fig. 4A). Similar preferential expression of the 1.6-kb mRNA with long term stimulation with cAMP has been reported in MA-10 cells (26) and in cul-tured rat adrenal cells. 2 Characterization of a 3.5-kb Star cDNA-To identify a cDNA that could encode the predominant 3.5-kb Star mRNA, we screened a rat adrenal cDNA library with the 0.9-kb Star PCR product. We identified two clones each with a 3.5-kb insert in addition to several clones with shorter inserts. The sequences demonstrated that each clone covered the entire coding region and additionally shared a ϳ2.5-kb 3Ј-UTR. There was a total of four polyadenylation signals (AATAAA) in the 3Ј-UTR, the last one being responsible for the formation of the 3.5-kb transcript.

FIG. 2. Immunoblots of STAR protein in adrenal glands from intact (A)
and HPX (B) rats. Total adrenal protein was obtained from the same ACTHtreated rats in which the levels of corticosterone were analyzed (Fig. 1). Protein was separated by two-dimensional gel electrophoresis and immunoblotted with STAR antibodies as described under "Materials and Methods." Protein loading was verified by immunoblotting the same membrane with glyceraldehyde 3-phosphate dehydrogenase (GAPDH).
The sequence of the ORF was identical to those obtained from the 0.9-kb PCR product (Fig. 3). Fig. 5 shows the sequence of the 3Ј-UTR in comparison with the shorter 3Ј-UTR reported for the mouse Star cDNA (19). Northern hybridization of rat adrenal mRNA with a cDNA probe targeted to the end of the 3Ј-UTR selectively recognized 3.5-kb Star mRNA but not the 1.6-kb species (data not shown). This indicates that this longer mRNA indeed differs by the additional presence of an extended 3Ј-UTR.
The involvement of polyadenylation signal for the formation of 1.6-kb mRNA was established through nested RT-PCR. Two Star-specific forward primers from the 5Ј-end of the ORF and a poly(dT) reverse primer were used to detect the poly(A) ϩ terminal sequences as depicted in Fig. 6. Two major polyadenylated species were detected in the size range ϳ1.5 to 1.6 kb corresponding to the second and third polyadenylation sites in the rat Star 3Ј-UTR (Fig. 6A). A minor shorter PCR product was also detected with a size of approximately 1.2-kb corresponds to the first polyadenylation site in the Star 3Ј-UTR. This 1.2-kb mRNA has been observed as a minor band in rat Leydig cells (37). Thus, 1.2-kb mRNA was not evident in rat adrenals. The longer 3.5-and 2.6-kb poly(A) ϩ mRNA species were only weakly detected by this RT-PCR approach due to the inefficiency of amplifying longer species.
Rat Star Gene Structure-To study the structure of the rat Star gene, a rat genomic library was screened with the 0.9-kb cDNA probe. Positive clones were characterized by long distance PCR analysis using primers designed against the se-quence of the mouse Star 5Ј-flanking region and the rat Star exon 7, as described previously (Table I, PCR7). Four of the five clones produced a 4.3-kb amplification product indicative of the presence of the complete Star gene (PCR7 in Fig. 6). The rat Star introns were characterized by applying PCR with exonspecific primers to rat genomic clones that span the full 3.5-kb mRNA sequence (Table I, PCR1 to PCR6, Fig. 6B). These primers were selected from the rat Star ORF sequence based on mouse and human STAR gene exon distributions (26,38). These PCR products were sequenced to characterize the rat Star introns. Sequences of the exon-intron junctions and the estimated size of the introns from the rat Star gene are summarized in Table II. PCR with primers corresponding to the 3Ј-UTR generated products of the same size (ϳ2.2-kb) using both the genomic and cDNA clones as a template (forward primer 5F, reverse primer 7R (Table I) and Fig. 6C). This clearly shows that the 3Ј-UTR of the 3.5-kb cDNA is coded by exon 7. Sequences of these genomic and cDNA segments confirm their identity.
The sequence upstream of the ORF has been determined and used to identify the transcription Start site by primer extension (Fig. 7A). The Start site is 5 bases on the 3Ј side of the reported Start site for mouse Star mRNA transcription in MA-10 cells. This establishes that exon 1 contains 87 bp of 5Ј-untranslated sequence. A TATA-like element (TTTAA) which exists in both human and mouse Star gene was located 26 -30 bp upstream from the transcription Start site. The orphan receptor SF-1/ Ad4BP binding element that was reported in the 5Ј-flanking region of the mouse Star gene was conserved at a position 134 bp upstream of the Start site (Fig. 7B), whereas several additional SF-1 binding sequences were located within 2.0-kb upstream of the transcription initiation site. We have shown that a rat Star-luciferase reporter construct containing 1028 bp of this upstream sequence is both basally expressed in Y-1 mouse adrenal cells and is stimulated 4-fold by cAMP. 3

DISCUSSION
In this study we have examined the relationship between the stimulation of corticosterone synthesis in rat adrenals in relation to changes in the level and multiplicity of STAR protein and mRNA. We have focused on changes occurring immediately after ACTH stimulation of hypophysectomized (HPX) and quiescent intact female rats. These conditions for stimulation of corticosterone secretion in HPX and intact rats correspond almost exactly to conditions used in previous studies that demonstrate an increased pool of reactive mitochondrial cholesterol (4,5). Thus, administration of ACTH or ether stress 10 -20 min prior to isolation of adrenal mitochondria increases the pool of cholesterol that is rapidly metabolized in these mitochondria, the proportion of cytochrome P450scc bound by cholesterol and the rate of metabolism of exogenous cholesterol (5,39). In this protocol, administration of ACTH in combination with aminoglutethimide, an inhibitor of cytochrome P450scc, further stimulates the level of reactive cholesterol and P450scc-cholesterol complexes, whereas cholesterol accumulates in the inner mitochondrial membrane (40). This change is inhibited by cycloheximide concomitant with accumulations of outer membrane cholesterol (5), and this has been linked by Orme-Johnson et al. (11) and Stocco and Chen (41) to loss of STAR activity.
Here we have shown that ACTH stimulation of both HPX and intact rats elevates secretion of corticosterone from the adrenal to peak rates after 5-10 min, whereas no change in the detectable STAR proteins can be seen even 20 min after stimulation. The absence of any increase in total STAR protein argues strongly against an ACTH-mediated activation process 3 D. Zhao, unpublished observations. that simply involves binding of STAR to an outer mitochondrial membrane receptor. This is inconsistent with activation of steroidogenesis without processing of STAR that has been indicated by a study with recombinant Star in COS-1 cells (22).
At about 60 min after stimulation of HPX rats, adrenal levels of STAR protein begin a 3-4-fold increase. This work, therefore, raises important questions about whether there are other functions for the large increases in STAR that occur subsequent to attainment of a maximum acute stimulation of steroidogenesis. This secondary slow increase in STAR protein certainly parallels an increase in corticosterone secretion. Thus, when intact rats are challenged with ACTH the acute response is 5-10-fold greater consistent with the 2-4-fold higher level of STAR protein compared with HPX rats. Therefore, although other adrenal enzymes will also increase during this period, these changes are consistent with a role for STAR as a regulator of ACTH-induced cholesterol metabolism and the magnitude of the acute response. Nevertheless, some as yet undefined component including an unresolved STAR form mediates this acute cAMP effect.
Under unstimulated conditions STAR is predominantly present as a pair of proteins. Following stimulations by ACTH, a set of low pI forms appears that reaches 50% of the total STAR, similar to 35 S-labeled proteins seen in cultured adrenal and testis cells (11)(12)(13)(14)20). The gel location of low pI STAR forms almost exactly matches the positions of phosphorylated STAR forms identified in these previous studies (12,20). This increase in modified STAR parallels the increase in total STAR protein. This is consistent with the proposal of Krueger and Orme-Johnson (10) that increased protein synthesis is neces- sary to generate co-translationally phosphorylated STAR forms that accumulate inside the mitochondria. However, we have failed to detect either 37-or 32-kDa STAR precursors that are readily seen in pulse 35 S-labeling of adrenal or MA-10 cells (10,12,20). These phosphorylated precursors are selectively processed in cultured adrenal cells to a single phosphorylated STAR form (12). The brief [ 35 S]methionine exposures used in this previous study label rapidly synthesized, short lived proteins disproportionately relative to proteins that turn over slowly. This consequently overrepresents these labile proteins relative to the immunoblots used in our experiments. However, although we may be unable to detect a key STAR component, this work establishes that such a component must be regulated entirely separately from the predominant pool of STAR forms, including the activation of their p37 and p32 precursors.
ACTH stimulation of Star mRNA transcription is a limiting factor in this secondary ACTH stimulation of Star protein levels. Removal of ACTH stimulation by HPX lowers total Star FIG. 6. PCR analysis of multiple Star mRNA species. A, a poly(A) ϩ Star mRNA species transcribed in rat adrenal glands. Below the gel picture is a schematic representation of the 3Ј-rapid amplification of cDNA ends RT-PCR method used to isolate Star mRNA species containing a poly(A) ϩ tail. The second PCR products were analyzed by Southern hybridization with 0.9-kb rat Star probe. In lane MIX, total PCR products were loaded, and the other lanes were loaded with PCR products that were gel-purified from the mixture. The arrow indicates the rat Star 3.5-kb cDNA clone which was applied as a positive control for hybridization. B, PCR products used to characterize each intron in the rat Star gene. Exon-specific primers were synthesized according to the reported conserved sequences of the mouse and human STAR exons. (Table I). The generated PCR products shown in the gel picture were sequenced and summarized in Table II. C, comparison of the 3.5-kb cDNA and the genomic clone using primer pair 8 (PCR8). Lanes M1 and M2 are DNA size markers of FX174-HaeIII digest and -HindIII digest, respectively. PCR products, PCR1-8 are shown below the gel picture in relation to each exon in the rat Star gene.

TABLE I Primers used for the generation of PCR products
Location of primers is shown in Fig. 6.  mRNA (3-5-fold). Stimulation by injection of ACTH remarkably reverses more than half of this loss within 20 min, and 1 h of stimulation is sufficient for near complete reversal. This rise in mRNA substantially precedes any increase in STAR protein, thus indicating that the level of Star mRNA for these HPX rats is a major determinant of the protein expression. In intact rats ACTH did not increase Star mRNA during an equivalent period. Basal ACTH levels may therefore be sufficient to maintain the expression of Star mRNA at the level seen after the acute response in HPX animals. Prolonged (24 h) ACTH stimulation in HPX and intact rats leads to further 2-3-fold rises in Star mRNA and protein suggesting a secondary slower ACTHsensitive control over Star expression. The delay in HPX rats between the rise in mRNA and the increase in protein (ϳ30 min) implies a slow translation step. Since the ratio of STAR protein to Star mRNA remains approximately constant during this process, ACTH stimulation of translation has no significant impact. The rise in mRNA in these HPX rats could, therefore, directly mediate formation a pool of labile, cycloheximidesensitive p30/STAR that has been observed in rat adrenal cells (12). In intact rats, however, ACTH increased protein levels between 1 and 3 h after stimulation without a comparable Star mRNA rise, 4 thus indicating ACTH stimulation of translation.
The rat Star gene is transcribed to two mRNA (1.6 and 3.5 kb) corresponding to short and long 3Ј-UTR, respectively. This is also seen in all other sources of Star expression (1.6 and 3.4 kb (mouse MA-10 cells) (26), 1.8 and 3.0 kb (bovine luteal cells) (16), 1.6 and 2.8 kb (ovine luteal cells) (15), and 1.6 and 4.4 kb (human) (21)). However, unlike MA-10 cells, the rat adrenal in vivo expresses the 3.5-kb mRNA as the predominant form (42). We have shown that this 3.5-kb mRNA arises from a 1.9-kb extension of the 1.6-kb species in the 3Ј-UTR region which is encoded by a continuation of exon 7 of the Star gene. The open reading frame (284 amino acids) for rat Star differs from the mouse sequence by 12 amino acids (Fig. 3). Surprisingly, 7 of these amino acids were identical to those in the bovine and human STAR sequences. Of 36 amino acid differences between the rat and human sequences, 17 are highly conserved substitutions, whereas 6 differences occur in a sequence of 8 amino acids close to the putative N-terminal processing site. Comparison of the four sequences reveals a remarkably conserved sequence including three potential phosphorylation sites as noted previously (20). This sequence differs substantially from a recently reported rat Star sequence that indicates an additional 86 amino acids at the N terminus (53).
The initial response of HPX rats to ACTH elevates both 1.6-kb and 3.5-kb Star mRNA equally. However, more prolonged treatment causes the 1.6-kb form to progressively increase relative to the longer form, and 3-day ACTH stimulation of intact rats results in nearly equal expression of each form (Fig. 4A). This treatment causes a 3-5-fold increase in STAR protein, which parallels the increase in 1.6-kb mRNA. This suggests that cAMP not only stimulates transcription of Star but also stimulates a secondary process that preferentially increases formation of the 1.6-kb mRNA which may then be preferentially translated. Insight into the mechanism of this process has been obtained from analysis of the extended 2.5 kb in the 3Ј-UTR of the 3.5-kb Star cDNA. In addition to the poly(A) ϩ sequence this 3Ј-UTR sequence has three polyadenylation signals that correspond to each of the shorter Star mRNA species (1.5 and 1.6 kb) and to the still shorter form (1.2 kb). RT-PCR using a poly(dT) 3Ј-adapter primer shows that poly(A) ϩ mRNA products are formed corresponding to each of these polyadenylation sites. The 1.5-and 1.6-kb Star mRNA are not resolved on Northern blots but appear to be present in equal amounts based on this RT-PCR analysis. The 1.2-kb poly(A) ϩ mRNA resulting from the first polyadenylation signal is a minor Star species in the rat adrenal gland but has been detected in rat testis (37).
The function of the 3Ј-UTR of the mammalian mRNA has not been well established; however, it may play a role at least at the level of mRNA stability. AU-rich elements within 3Ј-UTR appear to be crucial for their function as determinants of mRNA instability in mammalian cells. The AUUUA motif has been implicated in binding factors that affect both degradation and stabilization (43,44). The binding and the resulting mRNA stabilization is caused by activation of an adenosine-uridine binding factor via phosphorylation by protein kinase C (45). Since previous studies have shown that Star expression may be also regulated by Ca 2ϩ -signaling pathway (46), protein kinase C-dependent Star mRNA stabilization may provide a potential regulation process. The Elav-like RNA-binding proteins (HuD and HuR) are shown to be trans-acting factors involved in selective mRNA degradation by binding with high affinity to an AU n A element (n ϭ 3-5) (47,48). Recent studies have shown that a consensus sequence, UUAUUUAU, can function as a destabilizing element, suggesting a complex 3Ј-UTR regulatory mechanism (49). Significantly, we observe three such elements within 330 bp from the end of the 3Ј-UTR of rat Star and an AU 3 AAU 4 A element immediately prior to the terminal polyadenylation signal. These terminal elements are particularly noteworthy since they will be deleted through processing of the 3.5-kb mRNA to the 1.6-kb Star mRNA. 4 I. Artemenko, unpublished data. The putative binding site for steroidogenic factor 1 (SF-1) which has been reported in the mouse Star gene is also represented by bold letters with underlines. The putative consensus sequence for the Sp1 binding site is underlined. Transcription initiation site is indicated by bold and italic letters with arrows.
Exons 1-6 and the corresponding introns are closely matched to those seen for mouse and human STAR genes (26,38). We show here that this extended 3Ј-UTR arises from an additional 2.2 kb at the 3Ј-end of exon 7. Presumably, the Star genes from other mammalian species also have this much larger exon 7. The predominance of the full-length sequence in rat adrenals in vivo suggests that the intermediate polyadenylation sites are typically not recognized by the polyadenylation processing proteins. However, under conditions of prolonged hormonal stimulation (probably cAMP) a higher proportion of intermediate polyadenylation occurs at the 1.5/1.6-kb sites, possibly through selective binding of a cAMP-induced protein to 3Ј-UTR motifs. This analysis of the regulation of Star in rat adrenal gland raises many questions about the physiological significance of Star. Star transcription is evidently rapidly mediated by cAMP, and the 5Ј-flanking sequence of the gene contains several SF-1 sites that have been implicated in this stimulation (50). However, this is so sensitive to cAMP that levels of ACTH in unstressed animals, even at the low point of diurnal fluctuation used for our in vivo experiment (8 -10 a.m.), maintain sufficient Star mRNA for effective stimulation of steroidogenesis. This points to a mechanism where the activation of intramitochondrial cholesterol transfer is fully primed for stimulation. We can also question the function of this evidently conserved capacity to switch between short and long mRNA species. We suspect that the extended 3Ј-UTR of the 3.5-kb mRNA causes differences in stability and/or translation efficiency. Thus, in mouse Y-1 adrenal tumor cells, removal of cAMP stimulation causes a selective rapid loss of the 3.5-kb mRNA before changes in the 1.6-kb mRNA. 5 Regulatory signals in the 3Ј-UTR may provide a mechanism for other signaling processes, notably cytokines or chemical stress activation, that may override the ACTH stimulation of cholesterol metabolism. Such a mechanism has recently been reported for Star mRNA in the testis (51).
Levels of STAR protein expression in HPX, quiescent intact, and 24-h ACTH-stimulated HPX rats parallel the steroidogenesis rates. This is consistent with a rate-limiting function for STAR while requiring an additional rapid response factor for acute stimulation which is maximally activated within 5 min. It remains possible that a labile pool of STAR could provide this acute mediation as previously proposed by Epstein and Orme-Johnson (12) or that redistribution of STAR rather than net synthesis is critical. Contact sites provide a likely site for cholesterol transfer (52), and rapid intramitochondrial transfer and turnover of STAR precursor could facilitate formation of these sites (12,20,52). In either case a labile STAR mediator could be below our detection limits. In addition, the rapid changes seen in the peripheral benzodiazepine receptor (18) and the strong evidence for the involvement of this protein (36) suggest an important contribution to the acute ACTH response.
This leaves questions concerning possible functions for these more modified forms of STAR which became apparent only with prolonged stimulation. The recent identification of other forms of the STAR family in breast tumors and Caenorhabditis elegans (54,55) suggests that rat Star may have multiple functions. Significantly, the breast carcinoma protein is not located in the mitochondria but nevertheless shares conserved motifs in the C terminus where the dysfunctional mutations occur in human lipoid adrenal hyperplasia (23,24,56,57). This breast protein when expressed in COS-1 cells even exhibits a small stimulation of cholesterol metabolism (55).