Hormone-Dependent Expression of a Steroidogenic Acute Regulatory Protein Natural Antisense Transcript in MA-10 Mouse Tumor Leydig Cells

Cholesterol transport is essential for many physiological processes, including steroidogenesis. In steroidogenic cells hormone-induced cholesterol transport is controlled by a protein complex that includes steroidogenic acute regulatory protein (StAR). Star is expressed as 3.5-, 2.8-, and 1.6-kb transcripts that differ only in their 3′-untranslated regions. Because these transcripts share the same promoter, mRNA stability may be involved in their differential regulation and expression. Recently, the identification of natural antisense transcripts (NATs) has added another level of regulation to eukaryotic gene expression. Here we identified a new NAT that is complementary to the spliced Star mRNA sequence. Using 5′ and 3′ RACE, strand-specific RT-PCR, and ribonuclease protection assays, we demonstrated that Star NAT is expressed in MA-10 Leydig cells and steroidogenic murine tissues. Furthermore, we established that human chorionic gonadotropin stimulates Star NAT expression via cAMP. Our results show that sense-antisense Star RNAs may be coordinately regulated since they are co-expressed in MA-10 cells. Overexpression of Star NAT had a differential effect on the expression of the different Star sense transcripts following cAMP stimulation. Meanwhile, the levels of StAR protein and progesterone production were downregulated in the presence of Star NAT. Our data identify antisense transcription as an additional mechanism involved in the regulation of steroid biosynthesis.


Introduction
Steroid hormones are essential for maintaining normal homeostasis and reproductive capability. Biosynthesis of all steroid hormones starts in the mitochondrion with conversion of cholesterol into pregnenolone by the cholesterol side-chain cleavage enzyme cytochrome P450 (P450scc; CYP11A1) [1]. Transport of cholesterol from the outer to inner mitochondrial membrane, where the conversion to pregnenolone occurs, constitutes the rate-limiting step of steroidogenesis [2,3]. This trophic hormone-regulated step involves the formation of a macromolecular signaling complex that includes the outer mitochondrial membrane-localized translocator protein (TSPO, 18 kDa), TSPO-associated protein PAP7 (ACBD3), the regulatory a subunit of cAMP-dependent protein kinase (PRKARIa), steroidogenic acute regulatory protein (StAR; STARD1), the voltage-dependent anion channel (VDAC), and extracellular signal-regulated kinases (ERK 1/2 or MAPK3) and their upstream activator (MEK1/2) [4,5].
In the adrenal and gonads, regulation of steroidogenesis is mediated partially by mechanisms that enhance the transcription, translation, and/or activity of StAR [6,7]. Studies have demonstrated that regulation of StAR expression is complex, involving interaction between a diversity of hormones/factors and multiple signaling pathways [8,9]. Synthesized as a 37-kDa precursor molecule, StAR is imported into mitochondria, where it is cleaved to generate a 30-kDa mature form [10][11][12][13]. To render this protein fully active in its capacity to support cholesterol transfer, StAR phosphorylation by cAMP-protein kinase A (PKA) [14] and ERK1/2 [5,15] is required. Moreover, numerous transcription factors have been identified to bind the Star promoter and mediate transcription of this gene [16]. Another important regulatory mechanism of Star transcription involves acetylation and methylation of histones bound to the Star promoter [17,18].
In addition, post-transcriptional mechanisms, such as polyadenylation, also regulate Star mRNA. Rodent steroidogenic cells express two main transcripts (1.6-and 3.5-kb), as well as a minor 2.8-kb form [6,19,20], due to differential polyadenylation in exon 7. These transcripts share the same 59-untranslated region  and open reading frame, differing only in their 39-UTR. Thus a single protein is synthesized from all of them [20]. Mouse,

In silico prediction of NATs for the Star gene
To identify potential NATs specific to the Star mRNA, we performed a computational analysis by aligning the murine Star gene (Gene ID: 20845) with mouse expressed sequence tags (ESTs) using BLAST and the UCSC Genome Browser [45]. Homologous sequences were compared against the cDNA of the longest murine Star mRNA (NCBI Reference Sequence: NM_011485) and properties such as sequence orientation and localization of poly(A) signals and tails were analyzed. According to these criteria, several EST clones demonstrated transcription from the opposite direction (Fig. S1) and are potential NATs for Star. 59 Rapid Amplification of cDNA Ends (59 RACE) screening in MA-10 cells Given that 10 to 20% of EST sequences in UniGene were annotated in the wrong direction [46], reliable in silico screenings must be based on stringent parameters which could underestimate or exclude some genuine antisense transcripts. Nevertheless, we proceeded to experimentally validate our bioinformatics results by performing 59 RACE experiments on total RNA isolated from the MA-10 Leydig tumor cell line. Three different groups consisting of three sequence-specific primers were used for the RT, PCR, and nested PCR analyses respectively ( Fig. 1A and Table 1). Before any reaction, RNA was treated with RNase-free deoxyribonuclease I (DNase I) to eliminate any contaminating genomic DNA. Moreover, the identity of the amplified products was confirmed by sequencing in both directions. This analysis verified the presence of antisense sequences of different sizes that were completely complementary to Star mRNA (Fig. 1B). Many overlapped with each other, suggesting that they could be fragments of a longer, unique antisense sequence.

Star NAT expression in MA-10 cells
To evaluate the presence of a long Star antisense transcript, sequence-specific RT-PCR was performed using DNase I-treated total RNA from MA-10 cells. The relative orientation of transcription was assessed by restricting which primer was present during single-strand cDNA synthesis. For this purpose, the first primer of the G3 group designed for the 59 RACE assay was used. For the PCR and nested PCR amplifications, the remaining two primers of the G3 group were used as reverse primers. The two forward primers were designed based on the RACE results ( Fig. 2A).
Cloning and sequencing the PCR product confirmed the presence of a 2757-bp transcript that was perfectly complementary to the Star mRNA sequence, including parts of the coding region (i.e., exons 4, 5, 6, and part of exon 7) and the 39-UTR of the longest Star sense mRNA (Fig. 2B). Thus, hereafter we will refer to this transcript as the Star NAT.
To verify this result, several negative controls were included. A ribonuclease (RNase)-treated RNA control was analyzed to identify any RNA contamination in the reaction. In addition, RT-PCR of a DNase I-treated RNA sample without reverse transcriptase was performed. Two controls were conducted in order to evaluate the strand specificity of the reverse transcription: first, an RT reaction conducted in the absence of primer and a second RT reaction done in the presence of a non-specific primer. PCR product was not detected from these negative control reactions. Moreover, RT reactions were carried out at high temperature (i.e., 50uC) in order to avoid primer-independent cDNA synthesis [47] and to improve strand specificity in RT-PCR.

Star NAT characterization
Since total RNA contains both polyadenylated [poly(A) + ] and non-polyadenylated [poly(A) -] RNA molecules. To determine the class in which Star NAT should be categorized, MA-10 cell total RNA was divided into two fractions and analyzed in parallel by sequence-specific RT, PCR, and nested PCR following DNase I treatment. Fig. 2C shows that Star NAT was amplified from poly(A) + but not poly(A) 2 RNA, indicating that this transcript belongs to the class of polyadenylated RNAs. However, neither a classical polyadenylation consensus region nor a poly(A) tail was detected in the Star NAT sequence.
To evaluate Star NAT missing 39end sequence, 39 RACE experiments on DNase I treated-total RNA isolated from MA-10 cells were performed (Fig. 3A). Analysis of the amplified products showed an extension on Star NAT sequence of 389-bp reaching its poly(A) tail. We next performed a sequence-specific RT-PCR to confirm that this new fragment is part of a full length Star NAT transcript. Three primers were designed based on the results of the 39 RACE experiments, one for the RT reaction and the remaining two as reverse primers for PCR and nested PCR amplifications (Table 1). Fig. 3B shows a PCR product that confirmed the presence of a 3146-bp transcript that was perfectly complementary to the Star mRNA sequence, including the 59-UTR, the coding region and part of the 39-UTR of the longest Star sense mRNA. Interestingly, the last 75-bp of Star NAT 39end is complementary to the Star genomic sequence, beyond the 59 end of Star mRNA.
To further characterize this antisense transcript, the presence of open reading frame (ORF) coding proteins was analyzed. Then, probable amino acid sequences derived from this transcript were compared against murine protein databases using BLAST. Star NAT does not seem to encode any known protein unlike many other antisense transcripts described in the literature [48]. We also explored the possibility of a promoter region that regulates Star NAT transcription; however, none was identified within the Star gene or its flanking genomic sequences.

Expression of Star NAT in mouse tissues
To further validate our observations, we studied the expression of Star endogenous antisense transcripts in mouse tissues. 59 RACE experiments were performed with total DNase I-treated RNA isolated from mouse testis, ovary, adrenal gland, prostate, liver, kidney, and brain. Fig. 4A demonstrates that steroidogenic tissues, namely the ovary, adrenal gland, and testis, display a similar outcome to MA-10 cells, suggesting the presence of overlapping antisense sequences that were completely complementary to Star mRNA. Also, other partially complementary antisense sequences for Star were found in the liver, ovary, and brain. To assess the presence of a specific long Star NAT in mouse tissues, sequence-specific RT, PCR, and nested PCR were conducted as described previously for MA-10 cells. A single band of the expected size was detected and the identity of this amplification product was confirmed by sequencing. Fig. 4B shows that the Star NAT found in the MA-10 cell line is also expressed in classical steroidogenic tissues, such as adrenal gland, ovary, testis, and brain.

Hormonal regulation of Star NAT expression
To elucidate a possible functional role for Star NAT expression, hormonal regulation of this transcript was assessed. Semiquantitative RT-PCR was performed on MA-10 cells treated with human chorionic gonadotropin (hCG) for varying times (i.e., 0 to 6 h). Fig. 5A demonstrates that hCG-induced Star NAT expression was time-dependent. Maximum levels were observed after 2 to 3 h of hormone stimulation. Star NAT expression also increased in a time-dependent manner when cells were stimulated with cAMP (Fig. 5B), with maximum levels observed 3 h after stimulation.
To support these data, an RNase protection assay (RPA) was conducted to confirm the presence of the antisense transcript. In this experiment, MA-10 cells were incubated with 8Br-cAMP (1 mM) for varying times. Total RNA was extracted, treated with DNase, and then co-precipitated with single chain riboprobes specific to the Star mRNA sense strand and NAT (Fig. 6A). Hybridization and RNase A digestion were subsequently performed. Moreover, yeast tRNA with and without RNase treatment were used as controls. Fig. 6B shows that, consistent with the RT-PCR data, cAMP treatment increased Star NAT expression levels in a time-dependent manner. This experiment also demonstrated coordinated regulation of both sense and antisense Star transcripts after hormone stimulation. Since the Star sense probe targeted the coding region, joint expression of the three sense mRNAs (1.6-, 2.8-and 3.5-kb) was assessed in this approach. The data show that the antisense transcript exists simultaneously with its sense counterpart. For further confirmation, this experiment was repeated using a Star NAT probe that was complementary to the antisense transcript in a different region, its 39-end. Identical results were obtained with this approach (Fig. 6C), thereby validating our methodology as well as the presence and hormonal regulation of Star NAT via cAMP. These findings suggest that Star NAT may play a role in hormonal regulation of StAR protein expression and therefore steroid synthesis.

Effect of Star NAT overexpression on Star sense transcripts, StAR protein levels, and steroid production
To investigate the functional role of Star NAT expression in steroid synthesis, MA-10 cells were transiently transfected with a pcDNA3.1(+) vector expressing Star NAT. At 24 h post-transfection, cells were stimulated with 8Br-cAMP for varying times. Total RNA, mitochondria, and culture media were harvested and analyzed. First, the expression levels of Star sense transcripts were determined by semi-quantitative RT-PCR. Since the three Star mRNAs differ only within their 39-UTR, discriminating between them was difficult. Thus, three primer pairs targeting different regions of mouse Star mRNA were used ( Fig. 7A and Table 1). This analysis revealed that Star NAT overexpression had a differential effect on Star sense transcripts expression ( Fig. 7B and C). While no significant variation was found in the level of the 3.5-kb transcript, a dramatic increase was observed when the 2.8and 3.5-kb forms were determined together, indicating that Star NAT overexpression increases 8Br-cAMP-induced expression of the 2.8-kb mRNA. When the levels of all three Star sense transcripts were examined jointly, this increase was also detected. Nevertheless, this experiment did not allow us to evaluate the contribution, if any, of the 1.6-kb mRNA in the observed effect.
Thus, to address this and establish the selectivity of Star NAT, the expression levels of Star sense transcripts in transfected MA-10 cells was determined by Northern blot analysis. Although relatively less sensitive, this assay enabled us to evaluate each mRNA individually. The data confirmed that Star NAT overexpression increased 8Br-cAMP-promoted 2.8-kb and 1.6-kb Star mRNAs expression after 4 and 6 h of stimulation respectively, while expression of the 3.5-kb transcript was unaffected ( Fig. 8A and B).  StAR protein expression was assessed by western blot analysis in mitochondria from transfected cells. Star NAT overexpression caused a significant decrease in StAR protein after 4 to 6 h of stimulation ( Fig. 9A and B). This result was supported by a concurrent reduction in progesterone production, as determined by radioimmunoanalysis of the cell culture medium (Fig. 9C). Compared to mock-transfected cells, progesterone levels in Star NAT-transfected cells decreased 10 to 15% following incubation with 8Br-cAMP for 3 to 4 h. A 30% decrease was observed after 6 h of cAMP stimulation. Altogether these results indicate that, after 4 h of 8Br-cAMP stimulation, Star NAT elicits a differential effect on the three Star sense transcripts, which ultimately results in reduced StAR protein levels and steroid production. From these results, we conclude that Star NAT hormone-regulated expression may be involved in modulating StAR protein expression, a complex process in which post-transcriptional regulation of mRNAs is required.

Discussion
NATs have been implicated in numerous mechanisms that affect, directly or indirectly, virtually all levels of transcriptional control of eukaryotic gene expression [38,44].
Here we have identified a new endogenous antisense transcript that we named Star NAT because it complements the Star mRNA sequence, thus playing a role in the regulation and function of StAR. We have demonstrated its expression in MA-10 Leydig cells and steroidogenic murine tissues. Furthermore, we have established that hCG increases Star NAT expression via cAMP. This study constitutes one of a few experimentally proven examples of hormonal regulation of antisense transcripts.
Star NAT is 3146-bp long and has full sequence complementarity to the spliced StAR sense 3.5-kb transcript. Similar overlapping and length characteristics have been described for other antisense transcripts [49]. Based on a 200-nucleotide cut-off according to RNA purification protocols, antisense transcripts are classified as short RNAs and long non-protein-coding RNAs [48]. The latter are often several hundred (to thousands) of nucleotides in length and display strict homology to their corresponding sense sequence [38,50]. While NATs may contain potential ORFs, most are non-coding [35]. We found that Star NAT does not translate  59 RACE was performed as described in Fig. 1. A representative image of the RT-nested PCR products generated using the G1 and G3 primers is shown. MW, DNA molecular weight ladder. B. Star NAT sequencespecific reverse transcription was performed using DNase I-treated total RNA from different mouse tissues, followed by PCR/nested PCR as described in Fig. 2. L19 mRNA, a housekeeping ribosomal protein, was used as a loading control. A representative image of the RT-nested PCR product is shown. The identity of the amplification products was confirmed by sequencing. doi:10.1371/journal.pone.0022822.g004 into any known protein, suggesting that this RNA is most likely a non-protein-coding RNA (ncRNA) [48].
Star NAT is polyadenylated since it was efficiently amplified from poly(A) + , but not poly(A) 2 , RNA fractions. Moreover, its 39end was amplified in a 39 RACE experiment using an adapter primer which initiates the first strand synthesis at the poly(A) tail of mRNA. Sequencing confirmed the presence of a poly(A) tail in Star NAT full sequence. Although many of the NATs found in mice represent atypical transcripts, they tend to be localized in the nucleus and non-polyadenylated [51]. Some NATs are mRNAlike since they possess poly(A) tails and are expressed in the cytoplasm [49,50] where they may potentially interact with overlapping sense RNAs. It has been suggested that the poly(A) tail is localized to the 39-end of antisense transcripts, similar to sense transcripts [52]. The presence of a poly(A) tail would confer increased stability to the molecule and thus indicates its localization to the cytoplasm [53].
Three different mechanisms have been proposed to describe the origin of antisense transcripts: 1) antisense synthesis could occur either by transcription of the opposite strand of the corresponding gene (cis-NATs) [54]; 2) transcription of a pseudogene (trans-NATs) [55]; or 3) transcription of the sense mRNA by an RNAdependent RNA polymerase in the cytoplasm [52,56]. The mechanism by which Star NAT is generated is an interesting question. For instance, this NAT could be produced by transcription from the opposite strand of the Star gene. This is consistent with the complementarity of Star NAT last 75 bp to Star genomic sequence. However, the remaining sequence of Star NAT is the complement of the spliced Star mRNA. The use of atypical donor and acceptor sites for splicing of the antisense would be required to obtain an exact match between sense and antisense transcripts. Such sites have already been reported and bidirectional transcription has also been proposed [52,57]. In this case, the precise removal of six introns, as well as a promoter region that regulates Star antisense transcription, would be required. However, the current evidence does not support the existence of an origin derived from bidirectional transcription at the gene locus, as has been observed for other NATs [38,58,59].
Transcription of the Star NAT from a putative spliced pseudogene sequence can be excluded because antisense RNA arising from a pseudogene should exhibit several mutations and therefore display partial complementarity with the sense mRNA [55]. We found only a few point mutations at different positions within the sequence of different clones, which we attributed to classical PCR artifacts produced by the non-proofreading DNA polymerase. Screening the mouse genome database also showed no evidence of a Star pseudogene.
Nevertheless, Star NAT could be produced by transcription of processed Star mRNA in the cytoplasm [60,61]. This possibility is consistent with the precise and full complementarity of these sense/antisense RNAs. Therefore, an enzyme homologous to RNA-dependent RNA polymerase must be identified [62]. However, direct evidence for such an enzyme in mammalian cells is still lacking [63].
The function of most NATs remains undetermined. Several NATs may constitute transcriptional noise; however, there are clear indications that some have a gene regulatory impact [38,64]. Furthermore, investigation of the entire antisense transcriptome has identified common structural characteristics of some NATs and point towards conserved themes in gene regulation by antisense transcripts [39,42,51].
Expression of NATs in specific tissues provides clues about their physiological role [64]. Analysis of the tissue distribution of Star NAT revealed that it is expressed in steroidogenic tissues such as testis, adrenal gland, ovary, and brain, suggesting a role in regulating Star sense RNA expression. Comparative expression analysis [65] has demonstrated that, as in our case, NATs tend to be correlated with the expression of the corresponding sense transcript. However, these studies did not specify whether sense and antisense transcripts are expressed within the same cell [66,67], and frequently this linked expression is not found [68]. Moreover, a hormone-dependent increase in Star antisense RNA expression was evidenced by RT-PCR and RPA. Our data show that Star NAT expression reached a maximum level at 2 to 3 h after stimulation and indicate that this effect was mediated by cAMP. Our results are among the few recently described examples of effective regulation of NAT expression by hormones [69][70][71][72]. Sense-antisense Star RNA co-expression in MA-10 cells was demonstrated directly by RPA, suggesting that these transcripts could be coordinately regulated. Mammalian RNAs that form sense-antisense pairs have been reported to exhibit reciprocal expression patterns [73]. However, other studies have concluded that NATs display a tendency to be positively correlated with the expression of their sense counterparts [35,74].
There are several mechanisms by which NATs can influence the expression of their complementary transcripts. These mechanisms have been categorized into four main groups: those related to transcription, RNA-DNA interactions, RNA-RNA interactions in the nucleus, and RNA-RNA interactions in the cytoplasm [34]. Given the results presented here, Star NAT could form RNA duplexes with its sense counterpart, which occurs frequently when transcripts are long and completely overlapping [40,75]. However, this duplex formation has only been detected in a few cases [49,55,76] since such hybridization is a complex, transitory, and fragile. Since Star NAT appears to be polyadenylated, formation of a cytoplasmic RNA duplex could be postulated as part of its mechanism in regulating StAR expression. Cytoplasmic senseantisense duplex formation can alter sense mRNA stability and/or translation efficiency, mask protein-binding sites, or generate endogenous small interfering RNA [34].
Star is expressed in steroidogenic cells as 3.5-, 2.8-, and 1.6-kb transcripts that differ only in their 39-UTR, which is derived from alternative polyadenylation sites in exon 7 through the 39-UTR [20]. In the mouse MA-10 testis and Y-1 adrenal lines, 8Br-cAMP stimulates the Star 3.5-kb mRNA preferentially. This level of selectivity has also been observed with adrenocorticotropic hormone stimulation in primary bovine adrenocortical cells. In MA-10 cells, expression of the 3.5-kb mRNA peaks at 3 h but then declines rapidly, whereas the 1.6-kb mRNA is maintained at a steady-state level after 6 h of stimulation [24,29,77]. The 3.5-kb Star mRNA is intrinsically much less stable when expressed from vectors but is similarly translated [19]. This differential timing in expression of the 3.5-and 1.6-kb Star transcripts is most likely due to differences in post-transcriptional regulation. The time course of Star NAT expression after hormone stimulation nearly paralleled that of the Star 3.5-kb sense transcript. After achieving peak levels 2 to 3 h after stimulation, Star antisense RNA levels declined. In addition, we did not observe any changes in the expression of longer sense transcript levels after Star NAT overexpression, suggesting that the antisense transcript may play a role in stabilizing the 3.5-kb Star mRNA in the cytoplasm. The overlapping region may affect mRNA stability by altering its secondary or tertiary structure or by interacting with mRNA decay regulatory proteins, such as AURE-binding proteins. Interestingly, the 3.5-kb Star mRNA possesses an AURE within the extended 39-UTR. The zinc finger protein ZFP36L1/TIS11b binds to UAUUUAUU repeats in the extended 39-UTR, enhancing Star mRNA turnover. Surprisingly, TIS11b expression is induced concurrently with Star 3.5-kb mRNA. This co-regulation, despite an adverse affect on mRNA stability, may be explained by the presence of Star NAT [29]. Star antisense RNA could mask the AURE motifs required for TIS11b-mediated destabilization. Indeed, the antisense transcript overlaps the region that contains the first two UAUUUAUU repeats shown to be necessary for enhanced turnover. Therefore, cAMP could increase Star 3.5-kb mRNA stability by concomitantly increasing Star NAT expression, thus allowing for rapid StAR protein synthesis and cholesterol transport. Later, Star antisense transcript levels could decline, enabling destabilizing factors (e.g., AURE-dependent or -independent) to promote mRNA decay. An increase in mRNA stability provides a means to quickly respond to stimulation by cytokines, growth factors, and protooncogenes [78,79].
Alternatively, longer 39-UTRs frequently possess target sites for miRNAs, which mediate mRNA degradation [31]. Many NATs may have the ability to mask miRNA-binding sites following cytosolic RNA duplex formation [34]. Although the impact of miRNAs on Star expression has yet to be examined, one prospective miRNA (mmu-miR-706) site has already been identified within the 39-UTR of rodent Star mRNA [19,32]. The Star NAT sequence also overlaps this region, implying that a miRNA-related mechanism may be involved.
Overexpression of Star NAT resulted in an increase in 2.8-and 1.6-kb Star mRNAs with a concomitant decrease in StAR protein.
Although much of the mechanism remains to be elucidated, we can postulate that it does not involve an AURE-dependent increase in stability because Star 2.8-and 1.6-kb transcripts lack this sequence within their 39-UTR. Since low levels of StAR protein were observed despite high levels of the mid-and shortlength mRNAs, these transcripts may not be actively translated, but instead could be playing a role in regulating expression of the 3.5-kb mRNA.
Differences in how Star NAT and sense transcripts interact may be due to variations in their secondary and tertiary structures, as well as on the complexity of forming RNA duplexes [80]. The exact mechanisms underlying the possible functional role of Star NAT in the cAMP signal transduction pathway remain to be investigated. Nevertheless, our data indicate that regulation of StAR protein levels may depend on the presence of a natural antisense transcript. Hormonal regulation of this transcript and its role in balancing the expression of a crucial protein involved in cholesterol transport emphasize the importance of these results. Moreover, the involvement of NATs provides an additional mechanism for rapid hormonal regulation of steroidogenesis.
Regulation of StAR expression is a complex process involving transcriptional and post-transcriptional mechanisms. Our work establishes the concept that StAR protein levels are modulated by a fine balance between sense and antisense transcripts that enables a quick response to hormonal stimulation. Participation of the Star NAT in this process may explain the preferred formation of a longer Star transcript, which facilitates the rapid synthesis of StAR protein and cholesterol transport after hormone stimulation. In conclusion, the present results demonstrate, for the first time, that antisense transcription adds another level of control to gene expression in steroid synthesis regulation.

Computational analysis
The murine Star gene (Gene ID: 20845) was aligned with a mouse EST database using BLAST [81]. The results were visualized using the UCSC Genome Browser [45]. Homologous sequences were compared against the murine cDNA derived from the longest Star mRNA (NM_011485) and analyzed using the Unigene database (NCBI), Blast2sequence (NCBI), and the Vector NTITM Suite 8 (InforMax, Inc.) software. The latter program was also used to identify open reading frames (ORFs) and the identity of all sequenced fragments. A potential promoter region for Star NAT transcription was evaluated using PROSCAN version 1.7 (bimas.dcrt.nih.gov/molbio/proscan).

Mammalian cell culture and animals
The MA-10 cell line is a clonal strain of mouse Leydig tumor cells that produces progesterone rather than testosterone as the main steroid. MA-10 cells were generously provided by Mario Ascoli from the University of Iowa, College of Medicine (Iowa City, IA) and were handled as described previously [82]. The growth medium consisted of Waymouth MB752/1 containing 1.1 g/l NaHCO 3 , 20 mM HEPES, 50 g/ml gentamicin, and 15% heat-inactivated horse serum. Flasks and multiwell plates were maintained at 36uC in a humidified atmosphere containing 5% CO 2 .
Human chorionic gonadotropin (purified hCG, batch CR-125 of biological potency 11900 IU/mg; gift from NIDDK, NIH) was used to treat the MA-10 cells (20 ng/ml) for the times indicated. 8Br-cAMP (Sigma-Aldrich, St. Louis, MO) a permeable analog of cAMP, was used to treat the MA-10 cells (1 mM) for the times indicated.
According to approved protocols of animal care and use, tissues were obtained from 60-day-old male and female Balb/c mice (purchased from the School of Pharmacy and Biochemistry, University of Buenos Aires, Buenos Aires, Argentina).

Total and poly(A) + RNA isolations
Total RNA from MA-10 cells and Balb/c tissues was isolated with Tri-Reagent (Molecular Research Center, Inc. Cincinnati, OH) according to the manufacturer's instructions. Poly(A) + and poly(A) 2 RNA from MA-10 cells were fractionated by oligo(dT) cellulose chromatography as published [83]. RNA concentration was quantitated in triplicate. Any residual genomic DNA was removed by treating RNA with DNase I (15 min at room temperature), which was subsequently inactivated by incubation with 2.5 mM EDTA for 10 min at 65uC.

RACE
59 RACE was performed using the SMARTTM RACE cDNA Amplification kit (Clontech Laboratories, Inc., Mountain View, CA) according to the manufacturer's instructions. Three different groups of three sequence-specific primers were used for the RT, PCR, and nested PCR analysis. A list of the oligonucleotides used is shown in Table 1. Total DNase I-treated RNA was reversetranscribed using the first sequence-specific primer of each group. PCR and nested PCR amplifications were conducted with the two remaining oligonucleotides of each group. PCR conditions were: one cycle at 94uC for 5 min; 5 cycles of 94uC for 30 s, 70uC for 30 s, 72uC for 3 min; 5 cycles of 94uC for 30 s, 68uC for 30 s, 72uC for 3 min; 25 cycles of 94uC for 30 s, 66uC for 30 s, 72uC for 3 min; and one cycle at 72uC for 10 min. PCR products were resolved on 1.5% (wt/vol) agarose gels containing 0.5 mg/ml of ethidium bromide to determine the molecular size of the amplicons. Bands were visualized by UV transillumination and images digitally recorded. Bands were excised, eluted, and ligated into the pGEMH-T Easy vector, and then sequenced using T7/ SP6 primers and the ABI 3700 sequencer (Applied Biosystems-Life Technologies, Carlsbad, CA). Results were analyzed using BLAST and Vector NTITM Suite 8 software.

RACE
39 RACE was performed using the Invitrogen kit (Carlsbad, CA) according to the manufacturer's instructions. Total DNase Itreated RNA was reverse-transcribed at 50uC using the adapter primer (AP) which initiates the first strand synthesis at the poly(A) tail of mRNA. PCR and nested PCR amplifications were conducted with two sequence-specific primers and the two universal amplification primers provided with the kit. A list of the oligonucleotides used is shown in Table 1. PCR conditions were: 1 cycle at 94uC for 5 min; 5 cycles at 94uC for 30 s, 64uC for 30 s, 72uC for 3 min; 5 cycles at 94uC for 30 s, 62uC for 30 s, 72uC for 3 min; 30 cycles at 94uC for 30 s, 60uC for 30 s, 72uC for 3 min; and 1 cycle at 72uC for 10 min. PCR products were resolved on 1.5% (wt/vol) ethidium bromide-stained agarose gels to determine the molecular size of the amplicons. Bands were excised, eluted, and sequenced. Results were analyzed using BLAST and Vector NTITM Suite 8 software.

RT-PCR and nested PCR amplifications
RNA was reverse-transcribed with M-MLV reverse transcriptase and G3.1 or Rv.RT Star NAT-specific primers. RT reactions were carried out at high temperature (50uC) to avoid primerindependent cDNA synthesis [47] and to improve the strand specificity of RT-PCR detection.
PCR amplification was performed using GoTaqH DNA polymerase with the pair of primers listed in the corresponding section of Table 1. Amplification conditions were the same as those employed for 59 or 39 RACE assays. PCR products were resolved on 1.0% (wt/vol) ethidium bromide-stained agarose gels, excised, eluted, and finally cloned into a pGEMH-T Easy vector for sequencing. As mentioned previously, RNA was treated with RNase-free DNase I to eliminate genomic DNA contamination. In addition, several negative controls were added. First, an RNasetreated RNA control was included to determine whether any nucleic acid contamination was present in the reaction. Second, DNase-treated RNA was subjected to RT-PCR without reverse transcriptase to ensure the lack of contaminants and effectiveness of DNase I treatment. Finally, two controls were conducted in order to asses the strand specificity of reverse transcription: RT reactions were carried out in either the absence of primers, or in the presence of a non-specific primer that is not related to the target sequence. The latter targets the murine acyl-CoA synthetase long-chain family member 4 mRNA. As expected, no PCR product was detected from these negative control reactions.

Semi-quantitative RT-PCR
For semi-quantitative RT-PCR of Star NAT (amplicon size 460 bp), sequence-specific RT was performed using the G3.1 primer, followed by PCR amplification using the primers in Table 1. L19 ribosomal protein mRNA (amplicon size 405 bp) was used as internal standard to compare amplified amounts of Star NAT from different RNA samples. In this case, the RT reaction was conducted using random primers. PCR conditions were one cycle at 94uC for 5 min, followed by 27 (for Star NAT) or 23 cycles (for L19) of 94uC for 30 sec, 50uC for 30 sec, 72uC for 1 min, and one cycle at 72uC for 10 min. The number of cycles used was optimized for each transcript to fall within the linear range of PCR amplification.
For semi-quantitative RT-PCR of Star sense transcripts, sequence-specific reactions were also performed. Since the three StAR mRNAs differ only in their 39-UTR, it was difficult to discriminate between them. Thus, three primer pairs targeting different regions within mouse Star mRNA were used (Table 1). One pair (Sense 1, amplicon size 435 bp, 22 cycles) was designed to target only the extended 39-UTR of the longest 3.5-kb mRNA. A second pair (Sense 2, amplicon size 719 bp, 23 cycles) targeted a region that is shared by both 2.8-and 3.5-kb Star mRNAs. A third pair (Sense 3, amplicon size 566 bp, 21 cycles) was specific to the coding region and was thus common to all three Star sense transcripts (1.6-, 2.8-and 3.5-kb). PCR conditions were the same as described above. PCR products were resolved on 1.5% (wt/vol) ethidium bromide-stained agarose gels to determine the molecular size of the amplicons. Bands were visualized and transcript levels quantitated using the computer-assisted image analyzer Gel-Pro (IPS, North Reading, MA). Bands were excised, eluted, and cloned into the pGEMH-T Easy vector for sequencing.

In vitro transcription of RNA probes
Single-chain Star sense and Star NAT [ 32 P]-labeled riboprobes were synthesized by in vitro transcription. The Star NAT RNA probe (NAT probe 1, 779 bp) was complementary to 719 bp of Star NAT sequence. The template for its synthesis was a pGEMH-T Easy vector containing a Star sense fragment amplified by semiquantitative RT-PCR using the Sense 2 primer pair and cloned in the sense position relative to the T7 RNA polymerase promoter. The Star sense RNA probe (sense probe, 626 bp) was complementary to 566 bp of the coding region of sense transcripts. The template for its synthesis was a pGEMH-T Easy vector containing a Star sense fragment amplified by semi-quantitative RT-PCR using the Sense 3 primer pair and cloned in the antisense position relative to the T7 RNA polymerase promoter. Both plasmids were linearized with Sal I and gel-purified. Another Star NAT probe (NAT probe 2, 680 bp) was synthesized to be complementary to 590 bp of the Star NAT sequence in a region different from the previous NAT probe used. The template for its synthesis was the Star cDNA sequence cloned into a pGEMH-T Easy vector in the sense position relative to the T7 RNA polymerase promoter as previously generated in our laboratory [84]. This plasmid was linearized with Xho I and gelpurified. The choice of probes size (779-, 626-and 680-bp) was conducted according to the recommendations for RPA methodology [85]. Riboprobe synthesis reactions contained 0.5 -1 mg of linearized DNA plasmid; ATP, GTP, CTP (0.5 mM each); 12 mM UTP; 100 mCi [a 32 P] UTP (3000 Ci/mmol specific activity from New England Nuclear, Boston, MA); 5 mM dithiothreitol; 20 U RNAsinH inhibitor; 20 U T7 RNA polymerase; and 1X transcription buffer. After 2 h at 37uC, DNA templates were digested with RNase-free DNase RQ1. Transcription products were resolved on 5% denaturing polyacrylamide gels (8 M urea). Gel slices containing full-length transcripts were excised and incubated overnight at 37uC in gel elution buffer (0.5 M ammonium acetate, 1 mM EDTA, 0.1% SDS). The concentration of RNA probes was 3-6610 4 cpm/ml, with a theoretical specific activity of 3610 8 cpm/mg.

RNase Protection Assay (RPA)
Probes (sense and NAT 1 probes or NAT 2 probe alone, 72610 4 cpm each) and DNase I-treated total MA-10 cells RNA (80 mg) were co-precipitated and hybridized overnight at 42uC in hybridization buffer (40 mM PIPES pH 6.4; 0.5 M NaCl, 1 mM EDTA, 75% formamide). Then, RNase A (40 mg/ml) digestion was performed for 1 h at 32uC. The reaction was stopped by adding 0.5% SDS, followed by digestion with 12.5 mg/ml proteinase K at 37uC for 15 min. Next, a phenol/chloroform extraction was performed by adding 10 mg yeast tRNA to each sample as carrier. After ethanol precipitation, protected doublechain RNA fragments were subjected to a 5% denaturing polyacrylamide gel (8 M urea) chromatography and then visualized by autoradiography. The negative control (tRNA(+)R-Nase) consisted of yeast tRNA instead of MA-10 RNA. No bands are expected in this sample following digestion. The tRNA(2)R-Nase control consisted of yeast tRNA without ribonuclease treatment and was used to visualize the full-length probes.

Plasmid preparation and Star NAT overexpression by transient transfection
Star NAT sequence (2757-bp length, obtained from RT-PCR amplification based on 59RACE data), was subcloned from a pGEMH-T Easy vector into the eukaryotic expression vector pcDNA3.1(+) using Not I restriction sites.
For transient transfection, MA-10 cells were seeded the day before and grown up to 80% confluence. Transfection was performed with a pcDNA3.1(+) plasmid expressing Star NAT or an empty vector in Opti-MEM medium and LipofectamineTM 2000 reagent according to the manufacturer's instructions. Transfection efficiency was estimated at approximately 30% by counting fluorescent cells transfected with the pRc/CMVi plasmid expressing enhanced green fluorescent protein (EGFP) [84].

Protein quantification and western blotting
Mitochondria were isolated as described previously [86]. Mitochondrial proteins were determined by the Bradford assay [87] using BSA as a standard. Proteins were separated by SDS-PAGE (12% acrylamide gels) and transferred to PVDF membranes (Bio-Rad Laboratories Inc., Hercules, CA) as described previously [84]. Anti-StAR antibody was generously provided by Dr. Douglas Stocco (Texas Tech University Health Sciences Center, Lubbock, TX). Membranes were sequentially blotted with anti-StAR and anti-OxPhos complex III core 2 subunit (Invitrogen, Carlsbad, CA) antibody, and immunoreactive bands were detected using enhanced chemiluminescence (GE Healthcare, Buckinghamshire, UK).

Radioimmunoanalysis (RIA) and statistics
Progesterone production in cell culture media was measured by RIA as described previously [88]. Data are represented in ng/ml. Statistical significance was determined using the Student's t test or analysis of variance (ANOVA) followed by the Student-Newman-Kuels test. Figure S1 Genomic view of Star gene related RNA and EST sequences. The potential antisense transcripts (NATs predicted in silico) are highlighted with a pink circle in front of the sequence name. (PPT)