The Orphan Nuclear Receptor NGFIB Regulates Transcription of 3β-Hydroxysteroid Dehydrogenase

3β-Hydroxysteroid dehydrogenase type 2 (HSD3B2) is a steroid-metabolizing enzyme that is essential for adrenal production of mineralocorticoids and glucocorticoids. Thus, HSD3B2 is expressed at high levels in the glomerulosa and fasciculata, where these steroids are produced. In contrast, the production of dehydroepiandrosterone (DHEA) and DHEA sulfate in the adrenal reticularis is inversely correlated with the expression of HSD3B2. The reasons for the zonal expression of HSD3B2 are not known but represent an important aspect in the biochemical zonation of the adrenal. Using microarray, real time reverse transcriptase-PCR, immunohistochemistry, and HSD3B2 promoter analysis, we demonstrate that the NGFIB family of nuclear hormone receptors plays a critical part in the regulation of HSD3B2 transcription and may play an important role in the functional zonation of the adrenal gland. Microarray analysis of cortisol- versus DHEA sulfate-producing adrenal tissue demonstrated that NGIFB paralleled expression of HSD3B2 with expression much higher in cortisol-producing adrenal tissue; this observation was also demonstrated using real time reverse transcriptase-PCR analysis. In addition, immunohistochemistry confirmed that within adult and fetal adrenal gland NGFIB expression paralleled expression of HSD3B2. Transient transfections into H295R adrenal cells demonstrated that NGFIB family members enhanced HSD3B2 reporter activity but had no effect on a 17α-hydroxylase (CYP17) promoter construct. Deletion and mutational analyses of the 5′-flanking region of the HSD3B2 gene identified a consensus NGFIB response element that bound NGFIB in mobility shift assays. Infection of cultured human adrenal cells with adenovirus-containing NGFIB increased cortisol production by 8-fold and increased expression of HSD3B2 mRNA 26-fold over that observed in mock-infected cells. In primary cultures of adrenal cells, ACTH, an activator of HSD3B2, rapidly induced expression of NGFIB. These results suggest that NGFIB plays a crucial role in adrenal zonation by regulating HSD3B2 gene transcription.

The enzyme 3␤-hydroxysteroid dehydrogenase type 2 (HSD3B2) 1 is essential for the adrenal biosynthesis of mineralocorticoids (aldosterone) and glucocorticoids (cortisol), but its expression is inversely associated with that of the adrenal androgens (DHEA) (1). HSD3B2 catalyzes the oxidation and isomerization of 3␤-hydroxy-5-ene (⌬ 5 ) steroids into 3-keto-4ene (⌬ 4 ) steroids, thereby permitting the adrenal gland to synthesize progesterone and 17␣-hydroxyprogesterone from their pregnenolone and 17␣-hydroxypregnenolone precursors (2). HSD3B2 can compete with CYP17 for the metabolism of pregnenolone and 17␣-hydroxypregnenolone and thus influence the production of aldosterone versus cortisol or DHEA (3). High HSD3B2 expression combined with low CYP17 activity will favor aldosterone synthesis and oppose cortisol and adrenal androgen synthesis. Conversely, low HSD3B2 expression coupled with high CYP17 activity will favor adrenal androgen production.
The human adrenal produces DHEA at high levels within the fetal adrenal and in the zona reticularis of the adult adrenal. Both of these tissues express low levels of HSD3B2 protein and mRNA (4 -11). The lack of HSD3B2 expression in the fetal zone is directly correlated with the ability of that zone to secrete large amounts of the adrenal androgens DHEA and DHEA sulfate. The same correlation between relatively low levels of HSD3B2 and high levels of DHEA production is also present in the adult adrenal reticularis. Thus, a detailed understanding of the mechanisms that regulate HSD3B2 expression would help in efforts to understand adrenal physiology.
The present study was undertaken to examine the role of the NGFIB family of orphan nuclear receptors (nerve growth factor-induced clone B or NR4 subgroup) in the regulation of HSD3B2. The NGFIB family, which includes NGFIB (NR4A1), NURR1 (Nur-related factor 1 or NR4A2), and NOR1 (neuronderived orphan receptor 1 or NR4A3) (12,13), appears to play an important role in the coordinated regulation of the hypothalamic/pituitary/adrenal axis (14 -17). Herein, we demonstrate that adrenal expression of NGFIB (NR4A1) paralleled expression of HSD3B2. NGFIB also directly increased HSD3B2 gene transcription as shown using promoter constructs. Finally, NGFIB-containing adenovirus infection of adrenal cells stimulated endogenous HSD3B2 expression and cortisol production. The inverse correlation between adrenal androgen production and the expression of NGFIB and HSD3B2 appears to be a unifying link for the production of DHEA by the fetal adrenal and adult adrenal reticularis.

EXPERIMENTAL PROCEDURES
Cell Culture-NCI-H295R (H295R) human adrenocortical tumor cells were maintained as previously described (18,19). Primary human adrenal cell cultures were prepared from normal adrenal gland that was decapsulated, minced into small pieces, and incubated in Dulbecco's modified Eagle's medium/F-12 medium containing 0.1% collagenase. Digestion and mechanical dispersion were carried out for 1 h and 30 min at 37°C with a final digestion of the dispersed cells in medium containing 0.01% DNase I. Following isolation, cells were either frozen or subcultured onto 60-mm culture dishes for subsequent steroid assay or real time RT-PCR. Fetal adrenal cells were isolated essentially as described (20) and cultured in Dulbecco's modified Eagle's medium/F-12 medium, 1% ITS (BD Biosciences, Bedford, MA), 10% cosmic calf serum, and antibiotics.
Adrenal Tissues-For cell culture experiments, human adrenal glands were obtained from the Cooperative Human Tissue Network (Philadelphia, PA). Midgestation human fetal adrenal tissues were obtained from Advanced Bioscience Resources in accordance with the Donors Anatomical Gift Act of the State of Texas; protocols were approved by the Human Research Review Committee of the University of Texas Southwestern Medical Center at Dallas. For steroidogenic studies, glands were prepared as described above. Prior to real time analysis, total RNA was extracted from tissue, using the method of Chirgwin (21), followed by DNase I treatment. For immunohistochemistry, tissues were fixed in 10% formalin and embedded in paraffin wax. Institutional Review Boards of the University of Texas Southwestern Medical Center at Dallas and at Tohoku University School of Medicine approved the use of human tissue.
Steroid Analysis-Prior to treatments, primary adrenal cells were maintained overnight in low serum medium while fetal zone cells were kept in complete medium. Cells were incubated for 1, 6, or 24 h either under basal conditions, with ACTH (10 nM) or with ACTH plus the HSD3B2 inhibitor, trilostane (10 M). The steroid content of experimental medium was determined using radioimmunoassay kits (Diagnostic System Laboratories, Webster, TX). Results of steroid assays were normalized to the cellular protein content in each well and expressed as pmol/mg cell protein. Where indicated, relative steroid levels were expressed as -fold over basal.
Microarray Analysis-Pools of RNA from adult adrenal (n ϭ 3) and fetal zone (n ϭ 3) were hybridized to an Affymetrix human HG-U133A and B oligonucleotide two-microarray set containing more than 44,000 probe sets representing over 33,000 human genes. The arrays were scanned at high resolution using an Affymetrix GeneChip Scanner 3000 located at the University of Texas Southwestern Microarray Core Facility. Results were analyzed using GeneSpring version 6.1 software (Silicon Genetics, Redwood City, CA) to identify genotypic differences between adult adrenal cortex and fetal zone cells. Hierarchical clustering algorithms were used to identify genes that exhibited similar expression patterns to HSD3B2 in order to identify candidate regulatory genes.
RNA Extraction and Real Time RT-PCR-RNA extraction and real time RT-PCR were carried out as previously described (17) with modifications. Primers and probes for real time RT-PCR were designed using the Primer Express computer program (Applied Biosystems) ( Table I). HSD3B2 mRNA and 18 S rRNA were quantified using a TaqMan ribosomal RNA reagent kit (Applied Biosystems) and 10 l of primer/ probe mix. For HSD3B2, the final concentrations of primer and probe used were 0.4 and 0.2 M, respectively. For 18 S, the final concentrations of primer and probe were 0.05 and 0.1 M, respectively. Standard curve cDNA plasmids for NGFIB, NURR1, NOR1, and HSD3B2 were used to quantify transcript levels. As an internal standard, each individual sample was normalized to its 18 S ribosomal RNA content, and mRNA levels were expressed as amol/g 18 S rRNA. Alternatively, in some experiments, relative gene expression was calculated by the ⌬⌬Ct method. Briefly, the resultant mRNA was normalized to a calibrator; in each case, the calibrator chosen was the basal sample. Final results were expressed as n-fold difference in gene expression relative to 18 S rRNA and calibrator as follows: n-fold ϭ 2 Ϫ(⌬Ct sample Ϫ ⌬Ct calibrator) , where ⌬Ct values of the sample and calibrator were determined by subtracting the average Ct value of the transcript under investigation from the average Ct value of the 18 S rRNA gene for each sample.
Transfection Assays-Transfection assays were carried out as previously described except that Transfast (Promega, Madison, WI) was used as the transfection reagent according to the manufacturer's protocol (18).
Electrophoretic Mobility Shift Assay (EMSA)-H295R nuclear extracts were prepared, and EMSA was carried out as described with certain modifications (18). Binding was carried out in 50 mM Tris-HCl (pH 7.9), 12.5 mM MgCl 2 , 1 mM EDTA, 20% glycerol, 1 mM dithiothreitol, and 2 g of poly(dI-dC) as nonspecific competitor. Electrophoresis was performed in 0.50ϫ Tris borate-EDTA buffer. Rat NGFIB, mouse NURR1, and human SF-1 were prepared using an in vitro transcription/translation system (Promega). The probe for EMSA, which was designed to include the NBRE-1 site in HSD3B2 (underlined), was as follows: 5Ј-AACCTAAAGGTCACTAT-3Ј. The probe for EMSA was double-stranded.
Immunohistochemistry-Immunohistochemical analysis was performed employing the streptavidin-biotin amplification method as previously described (17). Rabbit polyclonal antibody for NGFIB (diluted 1:200) was purchased from Geneka Biotechnology Inc. (catalog no. 1600045). The polyclonal antibody for HSD3B2 (diluted 1:2,000) was described in detail (22). For negative controls, normal rabbit or mouse IgG was used instead of the primary antibodies, and no specific immunoreactivity was detected in these sections. Histological identification of three zones of the human fetal adrenal was based on published criteria (23).
Adenovirus-mediated Expression of NGFIB-A replication-defective human adenovirus vector, containing the coding region of canine NG-FIB (also called TR3), was prepared as previously described (24,25). Subconfluent fetal zone cells were infected at 3 ϫ 10 7 plaque-forming units/ml in complete medium for 3 h. Infected cells were either treated with ACTH (10 nM) or simply harvested for RNA in order to measure HSD3B2 levels by real time RT-PCR. Additionally, after infection, steroid hormone levels in the cell culture medium were measured by RIA. Controls for this type of experiment included infections with a mock adenovirus construct, which lacked the coding region of NGFIB. Detection of incorporated NGFIB adenovirus was accomplished by monitoring the production of canine NGFIB by real time RT-PCR.

Inhibition of HSD3B2 Enzyme Activity Promotes Adrenal
Androgen Biosynthesis-In order to further characterize the role of HSD3B2 in adrenal androgen production, primary cultures of human adrenal cells were treated with ACTH in the presence and absence of trilostane an inhibitor of HSD3B2 enzyme activity. Following treatment, cortisol or DHEA(S) was measured in the medium by RIA and normalized to the amount of cell protein present. As seen in Fig. 1A, ACTH stimulated cortisol production (4.8-fold over basal) after 24 h of incubation, whereas inhibition of HSD3B2 activity completely blocked this stimulation. In contrast, when HSD3B2 activity was blocked by trilostane, ACTH-stimulated DHEA(S) production rose dramatically to 120-fold over basal levels after 24 h of incubation (Fig. 1B). ACTH alone stimulated a 6-fold increase in DHEA(S) synthesis following 24 h of incubation.

The Expression Profile of the NGFIB Family of Transcription Factors Parallels
That of HSD3B2-Despite the crucial role of HSD3B2 in adrenal steroidogenesis, the mechanisms regulating transcription of the HSD3B2 gene remain poorly defined. In order to identify potential transcription factors that might regulate HSD3B2 expression, we performed hierarchical cluster analysis of gene expression patterns across fetal zone (a DHEAS-producing tissue) and adult (a cortisol-producing tissue) adrenal microarrays in order to identify genes with similar expression profiles to HSD3B2 (Fig 2A). Transcription factors that paralleled expression of HSD3B2 included the orphan nuclear receptors NGFIB, NURR1, and NOR; all four genes were inversely correlated with the ability of the tissues to produce DHEA. NGFIB exhibited 21.7-fold greater expression in adult than in fetal adrenal gland. Likewise, expression of NURR1 and NOR1 was greater in the adult adrenal (12-and 2.9-fold, respectively).
NGFIB, along with NURR1 and NOR1, is a member of the NGFIB family of transcription factors and is prominently expressed in brain but present in most peripheral tissues including the adrenal (26,27). In order to begin an investigation into a potential role for NGFIB, NURR1, and/or NOR1 in the regulation of HSD3B2 transcription and to confirm our microarray results, we quantified mRNA expression of HSD3B2 along with each NGFIB family member in human adult and fetal adrenal glands by real time RT-PCR (Fig. 2B). All values are expressed as amol of HSD3B2 or transcription factor/g of 18 S RNA. HSD3B2 expression was high in adult (51.5 Ϯ 14 amol/g) but low in fetal (0.20 Ϯ 0.06 amol/g) adrenal. Expression of the transcription factor NGFIB paralleled expression of HSD3B2 with high levels in adult adrenal (10.8 Ϯ 4 amol/g) and low levels in fetal adrenal (0.24 Ϯ 0.08 amol/g). Comparatively low levels of NURR1 and NOR1 mRNA were detected in adult adrenal (0.69 Ϯ 0.32 and 0.67 Ϯ 0.34 amol/g, respectively). Likewise, very low levels of NURR1 and NOR1 expression were observed in fetal tissue (0.004 Ϯ 0.002 and 0.13 Ϯ 0.066 amol/ g, respectively).
Localization of NGFIB in the Fetal Adrenal Gland-HSD3B2 is present in the nuclei of the definitive and transitional zones of late gestation fetal adrenal gland, a time when cortisol is being produced at high levels, but is essentially absent from the fetal zone at this time (9). In order to determine whether the presence of HSD3B2 was correlated with the expression profile of NGFIB, we examined expression of both proteins in fetal adrenal sections using immunohistochemistry (Fig. 3). The expression patterns of HSD3B2 and NGFIB were identical with marked immunoreactivity in the nuclei of the cells of the definitive and transitional zones but with weak or no staining in the cells of the fetal zone. A similar pattern of expression has been observed in adult adrenal with little or no HSD3B2 or NGFIB produced in the zona reticularis but with both proteins detected in the zonae glomerulosa and fasciculata (11,28). Thus, the observation that HSD3B2 and NGFIB are both expressed at low or undetectable levels in DHEA-secreting cells of the fetal zone and adult reticularis supports the hypothesis that NGFIB plays a role in HSD3B2 gene transcription.
Comparing the Effects of the NGFIB Family of Orphan Receptors and SF-1 on the Transcriptional Activity of HSD3B2 and CYP17-The relative amounts of two steroidogenic enzymes, HSD3B2 and CYP17, have a direct effect on the relative level of mineralocorticoid, glucocorticoid, and adrenal androgen produced by the adrenal cortex. When the ratio of HSD3B2 to CYP17 is high, production of aldosterone can occur, whereas cortisol and androgen production are decreased. Conversely, a high CYP17 to HSD3B2 ratio favors androgen synthesis and inhibits aldosterone and cortisol production. SF-1 is an important regulator of CYP17 gene transcription and may regulate HSD3B2 as well (29,30). In order to compare the effects of the NGFIB family of orphan receptors, and also SF-1, in the transcriptional regulation of HSD3B2 and CYP17, we co-transfected H295R cells with HSD3B2 (p3BϪ963) or CYP17 (pC17Ϫ1124) promoter constructs along with expression plasmids containing the coding regions of NGFIB, NURR1, NOR1, or SF-1 (Fig. 4). The HSD3B2 promoter was activated by NG-FIB (4-fold), NURR1 (5.1-fold), and NOR1 (3.2-fold). However, no significant difference over basal levels was observed when cells were co-transfected with SF-1. As expected, the CYP17 promoter was activated by SF-1 (15.4-fold). Conversely, the CYP17 reporter was not activated by NGFIB, NURR1, or NOR1, indicating that the NGFIB family members play no role in the regulation of the CYP17 gene. Thus, these branch-point enzymes in adrenal steroidogenesis, namely HSD3B2 and CYP17, are differentially regulated by two separate classes of transcription factors, the NGFIB family and SF-1, respectively.

Identification of a Single NBRE by Deletion and Mutation
Analysis-The NGFIB family of orphan nuclear receptors are rapidly induced early response genes that regulate gene transcription by binding to a consensus sequence (AAAGGTCA) called an NBRE (31,32). In order to identify potential NBREs within the HSD3B2 gene promoter, a series of deletion constructs containing progressively shorter fragments of HSD3B2 5Ј-flanking DNA (p3BϪ963, p3BϪ585, p3BϪ345, p3BϪ210, p3BϪ166, p3BϪ101, and p3BϪ52) were prepared. These constructs were transiently transfected into H295R cells along with empty pRc/RSV expression plasmid (basal) or pRc/RSV expression plasmid containing the coding sequence of NGFIB. Fig. 5A illustrates the activity of all eight HSD3B2 reporter constructs under basal conditions or after co-transfection with NGFIB. NGFIB-stimulated reporter activity of the two constructs containing 5Ј-flanking sequences from Ϫ963 and Ϫ585

FIG. 2. Expression profile of the NGFIB family of transcription factors parallels that of HSD3B2.
A, microarray analysis of human adult adrenal versus fetal zone cells. Each dot represents a unique sequence with a total of ϳ33,000 transcripts examined from these oligonucleotide arrays. Dots within the parallel lines represent mRNAs with less than 2-fold differences in expression. HSD3B2, NGFIB, NURR1, and NOR1 are indicated. B, quantification of HSD3B2, NGFIB, NURR1, and NOR1 transcript levels in human adult and fetal adrenal glands. Real time RT-PCR was used to quantify the transcripts for HSD3B2, NGFIB, NURR1, and NOR1 in adult (AA) and fetal (FA) adrenal as described under "Experimental Procedures." Data represent the mean Ϯ S.E. of seven independent RNA samples and are expressed as amol of mRNA/g of 18 S ribosomal RNA. increased by 3.7-and 4.4-fold, respectively, above basal levels. Between Ϫ585 and Ϫ166 (constructs numbered Ϫ345, Ϫ210, and Ϫ166), basal levels of each transcript increased over that seen with the Ϫ963 construct, and NGFIB-stimulated HSD3B2 activity of these constructs ranged from 2.0-, 2.5-, and 2.1-fold above basal levels, respectively. Further deletion to Ϫ101 decreased NGFIB-stimulated HSD3B2 reporter activity to levels similar to those observed with the p3B-52 and the pGL3Basic vector, indicating the presence of at least one NGFIB-regulated cis-element between bases Ϫ166 and Ϫ52.
Sequence analysis of the HSD3B2 gene promoter revealed the presence of a consensus NBRE at Ϫ131/Ϫ124 (Fig. 5B), herein named NBRE-1. In order to confirm the role of the putative NBRE-1 cis-element, it was mutated in the context of the full-length (p3BϪ963) HSD3B2 gene promoter. As shown in Fig. 5C, mutation of the NBRE-1 cis-element reduced NG-FIB-stimulated HSD3B2 activity to the level observed with the p3BϪ52 reporter construct.
Specific Binding of NGFIB, NURR1, and SF-1 to the NBRE-1 (Ϫ131/Ϫ124) cis-Element-In order to determine whether NGFIB, NURR1, and/or SF-1 interacted directly with the NBRE-1 (Ϫ131/Ϫ124) cis-element located within the HSD3B2 gene promoter, 32 P-labeled oligonucleotides containing this region were prepared and used in an EMSA. As seen in Fig. 6, the NBRE-1 probe bound in vitro prepared NGFI, NURR1, and SF-1 protein but did not bind in vitro prepared mock protein (protein prepared with empty pcDNA3.1 zeo expression vector). Additionally, the NBRE-1 element formed two specific protein-DNA complexes when incubated with two different H295R adrenal cell nuclear extract preparations (lanes 6 and 8). The complex designated C1 comigrated with the complex formed by in vitro translated NGFIB or NURR1, whereas C2 comigrated with the complex formed by SF-1. Formation of C1 and C2 were specifically inhibited by an excess of unlabeled homologous oligonucleotide (lanes 7 and 9), indicating that the formation of both complexes was due to specific binding.
ACTH Rapidly Induces Expression of NGFIB in Primary Cultures of Human Adrenal Cells-ACTH is a major regulator of HSD3B2 expression both in vivo and in vitro (7,33,34) To examine the effects of ACTH treatment on NGFIB expression, primary adrenal cells were incubated in the presence or absence of ACTH (10 nM) for 1, 6, and 24 h followed by real time RT-PCR analysis of NGFIB and HSD3B2 transcript levels. As seen in Fig. 7, NGFIB was rapidly induced (2.9-fold over basal), following 1 h of incubation with ACTH. NGFIB expression levels were high (21.3-fold over basal) after 6 h of incubation with ACTH but returned to near control levels (4-fold over basal) following 24 h of treatment. ACTH-stimulated HSD3B2 expression remained low at 1-and 6-h incubations but increased to 13-fold over basal levels after a 24-h incubation with ACTH. Our results suggest that ACTH may act through NG-FIB to regulate HSD3B2 transcription.
Adenovirus-mediated Overexpression of NGFIB in Fetal Zone Cells Greatly Enhances Expression of HSD3B2 mRNA-Whereas transient transfection analysis, using promoter constructs, provides strong evidence of a role for NGIFB in HSD3B2 gene transcription, infection of cells with viral vectors, which greatly increase transfection efficiency, allow one to study endogenous mRNA expression following infection with selected transcription factors. In order to examine the effect of NGFIB on endogenous HSD3B2 expression, we infected human fetal zone cells, which produce little or no HSD3B2 or NGFIB (Fig. 3), with a replication-defective adenovirus vector containing either the coding region of NGFIB (also called TR3) (24,25) or a mock adenovirus construct lacking the coding region of NGFIB. Mock-infected cells were either treated with ACTH or simply harvested for RNA in order to measure HSD3B2 levels by real time RT-PCR (Fig. 8A). Treatment with ACTH increased the expression of HSD3B2 in fetal zone cells by 17-fold, whereas the expression of HSD3B2 in cells infected with NGFIB increased ϳ26-fold over that observed in mock-infected cells.
Cortisol and DHEAS steroid hormone levels present in the cell culture medium were also measured, after infection, by FIG. 3. Immunohistochemical localization of NGFIB and HSD3B2 in human fetal adrenal gland. NGFIB and HSD3B2 were studied, using immunohistochemistry, with antibodies specific for either NGFIB (A) or HSD3B2 (B). The areas of the definitive (DF), transitional (TZ), and fetal (FZ) zones are indicated. HSD3B2 and CYP17. H295R adrenocortical cells were transfected with luciferase reporter constructs containing HSD3B2 (p3BϪ963) or CYP17 (pC17Ϫ1124) reporter constructs (1 g/well). Cells were co-transfected with either empty pRc/RSV expression vector or the indicated transcription factor (1 g/well). Following recovery for 24 h, cells were lysed and assayed for luciferase activity. Data are expressed as a percentage of the basal reporter activity of p3BϪ963 or pC17Ϫ1124. The results in each panel represent the mean Ϯ S.E. of data from three independent experiments, each one done in triplicate (*, p Ͻ 0.0001 compared with basal level). RIA (Fig. 8B). Infection with NGFIB caused a substantial increase in cortisol synthesis (8-fold). As expected, treatment with ACTH, with the resulting up-regulation of HSD3B2 (Fig.  8A), caused an increase in the production of cortisol (13.4-fold over untreated cells). DISCUSSION The three zones of the human adrenal cortex function as three distinct steroid-producing tissues, with the glomerulosa producing aldosterone, the fasciculata producing cortisol, and the reticularis producing DHEA/DHEAS. This represents a functional zonation of the cortex that is largely due to the zone-specific expression of key enzymes in the steroidogenic pathway. One enzyme that exhibits zonal expression is HSD3B2. Whereas the mechanisms that regulate the zonespecific expression of HSD3B2 remain unclear, its high expression in the fasciculata and low expression in the reticularis is a critical determinant of the functional differences between these two zones. We and others have proposed that changes in adrenocortical expression of HSD3B2 are a critical component in the process regulating adrenal androgen production (10, 35). Indeed, whereas HSD3B2 is not directly involved in the syn-FIG. 5. Analysis of the HSD3B2 gene promoter to identify putative NBREs. A, deletion analysis of HSD3B2 5Ј-flanking DNA. H295R cells were transiently transfected with luciferase reporter constructs containing serial deletions of HSD3B2 5Ј-flanking DNA. Transfection of reporter constructs was done with either empty pRc/RSV expression vector (1 g/well) or expression vector containing the coding sequence for NGFIB (1 g/well). Following recovery for 24 h, cells were lysed, and luciferase activity was measured. Deletion constructs are numbered relative to the transcriptional start site. B, nucleotide sequence of cis-element NBRE-1 (Ϫ131/Ϫ124), a consensus NBRE. C, mutational analysis of the NBRE-1 cis-element. H295R cells were transfected with a luciferase reporter containing p3BϪ963 (Ϫ963 WT) or construct containing the mutated NBRE-1 sequence (Ϫ963MUT) Cells were transfected with the indicated p3BϪ963 reporter construct and either pRc/RSV empty expression vector (1 g/well) or expression vector containing the coding sequence for NGFIB (1 g/well). Following recovery for 24 h, cells were lysed and assayed for luciferase activity. Data are expressed as a percentage of the basal reporter activity of the p3BϪ963 construct. Results represent the mean Ϯ S.E. of data from three or more independent experiments, each performed in triplicate (***, p Ͻ 0.0001; **, p ϭ 0.0138; *, p ϭ 0.0015 compared with basal level). thesis of DHEA(S), it competes with CYP17 for the metabolism of pregnenolone and 17␣-hydroxypregnenolone. In this way, the levels of HSD3B2 versus CYP17 will have direct effects on the amount of DHEA(S) produced in the adrenal cortex (3).

FIG. 4. Comparison of the effects of the NGFIB family of transcription factors and SF-1 upon the transcription of
In the fetal adrenal, as in the adult reticularis, production of DHEA(S) is correlated with low levels of HSD3B2 (36). One early hypothesis suggested that high circulating levels of fetal estradiol acted as an endogenous inhibitor of adrenal HSD3B2 activity (37,38). More recent studies have shown that the fetal adrenal gland synthesizes little HSD3B2 mRNA or protein (4,5). Immunohistochemical analysis of HSD3B2 during human fetal adrenal development has demonstrated the presence of HSD3B2 in the neocortex and the transitional zone of the fetal adrenal gland (8,9). However, the fetal zone, which secretes large amounts of DHEA(S), expresses little or no HSD3B2 protein throughout gestation. Like the fetal adrenal, the reticularis produces low levels of HSD3B2 protein and mRNA, suggesting that it may be alterations in gene transcription that regulate HSD3B2 levels during fetal and postnatal life.
Several lines of evidence suggest that lack of HSD3B2 in the fetal zone and adult reticularis plays a crucial role in the zonal regulation of human adrenal androgen production. For example, when isolated adrenal fasciculata cells were incubated with pregnenolone precursor and treated with cAMP analog, these cells produced cortisol and little or no DHEA(S). Conversely, under the same incubation conditions, isolated reticularis and fetal zone cells secreted primarily DHEA(S) (10). These observed differences were accompanied by much lower levels of HSD3B2 in reticularis and fetal zone cells. Furthermore, when HSD3B2 activity was blocked in isolated fasciculata cells, these cells secreted very high levels of DHEA(S), thus providing further evidence of the inverse correlation between HSD3B2 expression and adrenal androgen production. Here, using primary adrenal cells, we were able to completely inhibit ACTH-stimulated cortisol production by blocking HSD3B2 activity with trilostane. As a consequence of this block, ACTH-stimulated DHEA(S) production rose dramatically. We also used microarray analysis to compare mRNA expression profiles between adult adrenal glands and fetal zone cells and found that HSD3B2 exhibited 94-fold greater expression in adult adrenal. Using the more quantitative method of real time RT-PCR, we showed that the difference in adult mRNA expression for HSD3B2 was greater than 200-fold over that ex-FIG. 6. EMSA of NBRE-1 in HSD3B2. EMSA was performed using a 32 P-labeled oligonucleotide probe containing the NBRE consensus sequence of HSD3B2. Radiolabeled probe alone (FP; free probe) is shown in lane 1. Lanes 2-5 correspond to labeled probe incubated with in vitro translated NGFIB, NURR1, SF-1, or pcDNA3.1 zeo (mock), as indicated. Probe incubated with two different preparations of H295R nuclear extract (NE; 2 g) is shown in lanes 6 and 8. Nonradiolabeled self-competitor DNA was added to the nuclear extract reaction mixture in a 100-fold molar excess (ϩ100ϫ cold) in order to identify nonspecific protein-DNA interactions (lanes 7 and 9). Protein-DNA complexes were separated from free probe by gel electrophoresis. Two protein-DNA complexes were present in the H295R nuclear extract; the upper complex comigrated with the complex formed by in vitro translated NGFIB or NURR1, whereas the second complex comigrated with the complex formed by in vitro translated SF-1.

FIG. 7. Time-dependent effects of ACTH treatment on HSD3B2
and NGFIB transcript levels in primary cultures of human adrenal cells. Cells were incubated for the times shown either alone (basal) or with ACTH (10 nM). Real time RT-PCR was used to determine the relative expression of HSD3B2 and NGFIB, which were subsequently normalized to 18 S. mRNA expression following ACTH treatment was compared with mRNA levels under basal conditions (control ϭ 1). Data points are the values calculated by the ⌬⌬Ct method as described under "Experimental Procedures" and represent the mean of two independent RNA samples expressed as -fold over basal levels.
pressed in fetal adrenal. Thus, our results and previous work strongly support the hypothesis that the ability to secrete large amounts of DHEA(S) is correlated with low levels of the HSD3B2 transcript.
A clearer understanding of the processes that regulate transcription of the HSD3B2 gene would help to define the mechanisms that control the zonal production of DHEA(S). One report suggested that the orphan nuclear receptor SF-1 regulates transcription of the HSD3B2 gene (29). However, SF-1 is expressed throughout the adult and fetal adrenal, making it an unlikely candidate in the zone-specific regulation of HSD3B2. The absence of HSD3B2 protein and mRNA in DHEA-secreting cells could also be attributed to the possibility that the fetal zone and adult reticularis lack a transcription factor necessary for HSD3B2 gene transcription. In our microarray analysis and in real time RT-PCR, we found that the transcription factor NGFIB exhibited much greater expression in adult adrenal than in the fetal zone. Moreover, examination of the expression pattern of NGFIB and HSD3B2 in fetal adrenal gland by immunohistochemistry revealed that NGFIB and HSD3B2 are expressed at much lower levels in the DHEA(S)-secreting fetal zone than in the cells of the definitive and transitional zones. A similar pattern of expression has been observed in adult adrenal with little or no HSD3B2 or NGFIB produced in the zona reticularis but with both proteins detected in the zonae glomerulosa and fasciculata (11,28). Here, using adenovirus-mediated infection of fetal zone cells with NGFIB, we were able to dramatically increase the level of endogenous HSD3B2 transcript present in these, normally HSD3B2-depleted, cells. As a consequence of increased HSD3B2 synthesis, the level of cortisol produced in fetal zone cultures was increased 8-fold, whereas the level of DHEA(S) remained relatively constant. These results, along with the expression patterns of HSD3B2 and NGFIB in cortisol-versus DHEAS-secreting cells of the adult and fetal adrenal, respectively, support the hypothesis that NGFIB plays an important role in HSD3B2 gene transcription.
NGFIB, NURR1, and NOR1 comprise a family (NR4A) of closely related transcription factors that are expressed primarily in the brain and, to a lesser extent, in most peripheral tissues including the adrenal (26,27). Their mechanism of regulation appears to be through binding to the consensus sequence (AAAGGTCA), called an NBRE, located within the 5Ј-region of target genes (31,32). The lack of an adrenal phenotype in NGFIB knock-out mice originally called into question the importance of these factors in steroidogenesis. However, the cloning of other members of the NR4A family (NOR1 and NURR1) and the demonstration that they are also expressed in the adrenal, suggested that this family of genes may compensate for one another in vivo (39,40).
Several lines of evidence suggest that the NR4A family plays an important role in the coordinated regulation of the hypothalamic/pituitary/adrenal axis. For example, a recent study implicated NGFIB and NURR1 as mediators of CRH function in the hypothalamus (14). In the adrenal gland, NGFIB may regulate CYP21 transcription (15,16), whereas NGFIB and NURR1 are important regulators of the hCYP11B2 gene (17). hCYP11B2, which is a glomerulosa-specific enzyme, catalyzes the final step(s) in aldosterone biosynthesis; NURR1, which is preferentially localized to the glomerulosa, may be responsible for the zonal expression of hCYP11B2. It is also possible that NURR1 may regulate HSD3B2 in the glomerulosa as well. These observations support the hypothesis that the NR4A family of transcription factors may be involved in the zonal expression of steroid-metabolizing enzymes within the adrenal.
In summary, it is clear that the level of DHEA(S) produced in the adrenal gland follows a unique pattern throughout human life that is distinct from the other adrenal steroids. Whereas the underlying mechanisms that regulate DHEA(S) synthesis are still unclear, there is an inverse correlation between the amount of DHEA(S) produced and the levels of HSD3B2 present within the adrenal gland. Thus, understanding how HSD3B2 expression is regulated becomes a key requirement for understanding DHEA(S) production. Here we present evidence that the transcription factor NGFIB is a key regulator of HSD3B2 transcription and thus plays an important role in determining the capacity of the adrenal gland to produce DHEA(S).