Minireview: Steroidogenic Factor 1: Its Roles in Differentiation, Development, and Disease

Patterns of expression in adult and embryonic tissues

The localization of SF-1 in adult tissues has been investigated extensively in mice, rats, humans, and other vertebrates. Consistent with its role as a master regulator of steroidogenesis, SF-1 is expressed in the major steroidogenic tissues, i.e. the three zones of the adrenal cortex, testicular Leydig and Sertoli cells and ovarian interstitium, theca cells, granulosa cells, and, to a lesser degree, corpus luteum ( 6, 8, 9, 10, 11). It also is expressed in pituitary gonadotrophs where it regulates the expression of the gonadotropins ( 12, 13) and the GnRH receptor ( 14). Apart from these tissues, SF-1 expression is limited to neurons in the dorsomedial portion of the ventromedial hypothalamus that are distinct from those expressing GnRH, to the endothelial linings of the venous sinuses and pulp veins in the spleen ( 11, 15), and to a subset of neurons in the hippocampus that coexpress steroidogenic acute regulatory protein (StAR) and aromatase ( 16).

In developing embryos, SF-1 first appears in the urogenital ridge and represents the earliest marker of adrenal and gonadal differentiation ( 17, 18, 19, 20). As development proceeds, SF-1 is expressed in the steroid-secreting cortical zones of the adrenal gland, in testicular Leydig and Sertoli cells, and in the human ovary; however, in mice and rats, ovarian expression of SF-1 is turned off as development proceeds and does not resume until just before birth ( 17, 19, 20). SF-1 also appears in the fetal spleen ( 15) and in the prosencephalon—that region of the brain that includes the hypothalamic primordium ( 19).

Regulation of NR5A1 promoter activity

Signal transduction pathways affecting SF-1 function

Regulation of function through protein-protein and protein-RNA interactions

Although SF-1 interacts with its target DNA element as a monomer ( 71), its activity is strongly influenced by its interactions with other transcription factors and coregulators (for brief review see Ref. 72) (Table 2). Some interacting proteins are expressed in a tissue-specific manner whereas others are more widely expressed, and it is unclear how each contributes to the coordinate and cell-selective expression of SF-1-dependent genes. Analyses of transcription complex assembly on the MC2R ( 48) and CYP17 ( 73) promoters indicate that SF-1-containing complex formation is dynamic and cyclical. Thus the nature of the complexes formed is likely to be influenced by the relative abundance of the interacting partners and by their relative affinities for the promoter and for SF-1 at any given spatiotemporal interval.

Interactions between SF-1 and DAX-1 deserve particular comment (early work is reviewed in Ref. 74). Mutations in DAX-1 (NR0B1) cause X-linked adrenal hypoplasia congenita (AHC) and hypogonadotrophic hypogonadism, phenotypes similar to those seen with SF-1 mutations (see below). In early stages of mouse embryonic development, SF-1 expression in the urogenital ridge and brain either precedes or coincides with DAX-1 expression. As embryonic development proceeds, the two transcription factors exhibit similar expression patterns in the adrenal cortex, testis, ovary, hypothalamus, and anterior pituitary. The slightly earlier onset of SF-1 expression and its ability to bind specifically to a conserved sequence in the NR0B1 5′-flanking region suggest that SF-1 may activate DAX-1 expression, and indeed SF-1 stimulates expression of the NR0B1 promoter in transient expression assays in SF-1-deficient cells ( 75, 76).

Coexpression of DAX-1 and SF-1 inhibits SF-1-mediated transactivation of many target genes. DAX-1 interacts directly with SF-1 in vitro via the carboxy-terminal regions of each protein ( 77). Additionally, DAX-1 interferes with interactions between SF-1 and WT1 (the Wilm’s tumor suppressor protein) or GATA4, each of which normally associates and synergizes with SF-1 to promote AMH expression. Finally, DAX-1 acts as an adaptor that recruits other factors, such as the nuclear receptor corepressor, to SF-1 ( 78).

Expression of DAX-1 in adrenal cells is decreased by ACTH and increased by glucocorticoid treatment, the latter occurring through a glucocorticoid receptor-SF-1 complex on the NR0B1 promoter. ACTH disrupts the formation of this complex by shifting SF-1 binding away from the NR0B1 promoter and toward the MC2R and STAR promoters ( 79, 80).

Whereas all of these findings suggest that DAX-1 acts mainly to negatively regulate SF-1 actions (and thus steroidogenesis), the inhibitory domain in DAX-1 is deleted or mutated in most AHC mutants, suggesting that loss of the inhibitory property in DAX-1 is associated with the development of AHC. Given that inactivating mutations in DAX-1 and SF-1 have similar phenotypes, this seems paradoxical and possibly reflects DAX-1 functions in development that are distinct from its regulatory roles in postnatal life. Indeed, DAX-1 can function as an SF-1 coactivator in some contexts. Both SF-1 and DAX-1 bind to steroid receptor RNA activator (SRA), a noncoding RNA that functions as a coactivator. In cotransfection experiments, an SRA expression plasmid increases the ability of SF-1 (or DAX-1) plasmids to transactivate reporter constructs, whereas knockdown of SRA expression abolishes coactivation by Dax-1. Moreover, SRA is physically associated with both SF-1 and DAX-1 in Y1 cells and is expressed in both the adrenal glands and gonads. The coactivator TIF2 also associates with DAX-1 and synergistically coactivates SF-1 target gene transcription. Corepressor vs. coactivator functions of DAX-1 may be dose dependent; although transfection of DAX-1 expression constructs usually inhibits expression of steroidogenic proteins, knockdown of endogenous DAX-1 actually down-regulates the expression of CYP11A1 and STAR in both H295R adrenal and MA-10 Leydig cells ( 81).

Regulation through posttranslational modification

In addition to phosphorylation at S²⁰³ as describe above, SF-1 is subject to SUMO conjugation and acetylation at ε-amino groups of target lysine residues. SUMO-ylation represses SF-1 function ( 82, 83, 84), whereas acetylation enhances its activity ( 54, 85).

For a brief review of protein SUMO-ylation and its consequences on gene expression, see Ref. 86 . SF-1 is SUMO-ylated within its DBD at K¹¹⁹ and within its hinge region at K¹⁹⁴ ( 82, 83, 84); K¹⁹⁴ appears to be the primary site of SUMO-ylation and suppression of SF-1 activity whereas K¹¹⁹ makes a minor contribution ( 82, 83, 84). Disruption of these SUMO-ylation sites increases the promoter-regulatory activity of SF-1 at CYP19, HSD 3B2, STAR, and at a number of isolated SF-1 response elements ( 82, 84); preventing the SUMO-ylation of SF-1 by overexpressing the SUMO-isopeptidase, SENP1, also enhances SF-1 activity ( 84). SUMO-ylation does not affect the stability, cellular localization, or dynamic association of SF-1 with chromatin or its ability to bind to canonical SF-1 response elements but does inhibit its phosphorylation by CDK7 at S²⁰³ ( 87, 88) and may (84) or may not (82) affect its ability to interact with repressors such as DP10; SUMO-ylation at the minor K¹¹⁹ site interferes with the ability of SF-1 to bind to and function at noncanonical response elements such as those found in the rat INHA promoter ( 87).

Two histone acetyltransferase proteins, PCAF ( 85) and p300, ( 54) stimulate the incorporation of labeled acetate into SF-1 and increase its transcriptional activity. Disruption of putative acetylation sites prevents acetylation of SF-1 and decreases its transcriptional activity ( 54, 85), whereas inhibition of the deacetylation of SF-1 using trichostatin A increases both its half-life and activity ( 85, 89). The acetylation of SF-1 by p300, but not by PCAF, appears to increase its localization within transcriptionally active nuclear foci ( 54). Other coactivators with histone acetyltransferase activity interact with, and modulate, the activity of SF-1 as well (Table 2).

The dynamics of SUMO-ylation and acetylation are incompletely understood. Less than 10% of SF-1 appears to be SUMO-ylated in vivo ( 82, 83, 84, 88), and it is not clear how this small fraction can have a dominant effect on SF-1-dependent transcription. The same can be said for acetylated SF-1, based on the presumption that it too represents only a small fraction of the total. Perhaps the posttranslational modifications of SF-1 are influenced by the cyclical and dynamic recruitment of SF-1 into active transcription complexes because, for example, DNA-bound SF-1 is refractory to SUMO-ylation at K¹¹³ ( 87).

Phospholipid-dependent regulation

The finding that SF-1 is a member of the nuclear receptor family of transcription factors engendered a search for ligands that regulate its function. Oxysterols enhance SF-1 function in Chinese hamster CV-1 lung cells ( 90, 91) but not in other cell types ( 61, 91, 92). X-ray crystallography and mass spectrophotometry of bacterially expressed SF-1 indicate that its ligand-binding pocket is very large, very hydrophobic, and occupied by phosphatidylethanolamine or phosphatidylglycerol ( 93, 94, 95, 96). These bacterial phospholipids can be exchanged for other potential regulatory ligands such as phosphatidic acid, phosphatidylcholine, and phosphatidylinositol di- and triphosphates ( 93, 94, 95) with little consequence on the conformation of the AF2-activating domain of SF-1, but with some effect on conformation at the entrance to the ligand-binding pocket of the protein ( 95). SF-1 purified from H295R adrenal tumor cells is associated with sphingosine, which acts as an inhibitor of SF-1 function in reporter gene assays ( 53). Treatment of cells with cAMP increases sphingosine catabolism, decreases the amount of sphingosine bound to SF-1, and enhances SF-1-dependent gene expression, whereas inhibitors of sphingosine synthesis enhance SF-1 activity. Sphingosine can be displaced from SF-1 to varying degrees by the phospholipid ligands discussed above, suggesting that SF-1 function is controlled by changes in the lipid environment secondary to regulated changes in lipid metabolism. This may have a teleological explanation given the key role of SF-1 in regulating steroidogenesis, a process requiring high concentrations of cholesterol.

As regards pharmacological ligands, low-molecular weight compounds with a rigid cis-bicyclo[3.3.p]oct-2-ene core structure selectively increase SF-1 activity ( 97) whereas 4-alkyloxy-phenol derivatives act as inverse agonists, suppressing its constitutive activity ( 98). Given the importance of SF-1 in steroidogenesis and in appetite control, these compounds may lead to the development of drugs useful in the treatment of steroid hormone excess, steroid hormone-dependent tumors, obesity, and related metabolic disorders.

Gene targets of SF-1 in adrenal glands and gonads

Many SF-1 target genes (Table 1) have been identified with use of transient transfection assays using relatively limited stretches of promoter/regulatory DNA.

In adrenocortical cells, SF-1 increases expression of the corticotropin receptor (MCR2), STAR, and all of the enzymes required for cortisol or corticosterone biosynthesis. It also increases levels of scavenger receptor B1, which is required for cellular importation of high-density lipoprotein cholesterol, the main source of cholesterol for steroidogenesis, and AKR1B7 (aldose reductase-like protein), which metabolizes the isocaproaldehyde generated by cleavage of the cholesterol side chain ( 99). Although SF-1 is generally regarded as an activator of gene expression, it is a strong negative regulator of the type 4 adenylyl cyclase, thereby regulating adenylyl cyclase isoform composition ( 100) and of CYP11B2 (aldosterone synthase), which catalyzes the final step in aldosterone biosynthesis ( 101), thereby contributing to the restriction of aldosterone biosynthesis to the adrenal zona glomerulosa. In the ACTH-responsive, Y1 mouse adrenocortical tumor cell line, cells with a mutation affecting SF-1 function exhibit decreased expression of the MC2R CYP11B1 (11ß-hydroxylase) CYP11A1 (cholesterol side-chain cleavage enzyme), and STAR ( 102). In this cell line, the SF-1 defect affects the expression of the MC2R and CYP11B1 to a much greater degree than CYP11A1 or STAR, suggesting heterogeneity among these target genes in their regulation by SF-1.

SF-1 increases the expression of many testicular genes required to develop and maintain the male phenotype. Analogous to its effects in adrenocortical cells, SF-1 stimulates expression in Leydig cells of the LH receptor, STAR, and the CYP11A1 and CYP17 enzymes required for testosterone biosynthesis. Additionally, it increases expression of insulin-like polypeptide 3, which regulates testicular descent and acts in adults as a male germ cell survival factor ( 103, 104), and the AMHR2 receptor for AMH, which is required for normal male reproductive tract development and serves as a negative regulator of Leydig cell steroidogenesis ( 105).

In Sertoli cells, SF-1 is required for the expression of the testes-determining gene products, sex-determining region Y, and SOX9 AMH, the receptor for FSH, and INHA. Thus SF-1 is necessary for male sexual differentiation, a conclusion underscored by the XY sex reversal seen in either humans or mice carrying SF-1 mutations (see below).

Whereas SF-1 is required for ovarian development (as determined from studies of knockout mice, described later), interpreting the role of SF-1 in the adult ovary is complicated by the cyclic nature of follicular maturation, steroidogenesis, and corpus luteum formation. SF-1 is expressed most highly in theca and interstitium, at lower levels in granulosa cells and at even lower levels in the corpus luteum. In contrast, LRH-1 (NR5A2) expression is abundant in cells involved in estrogen biosynthesis, granulosa cells during the estrous cycle (in rodents), and in corpora luteum in both rodents ( 106) and humans (107). LRH-1 potently transactivates both CYP19 ( 106) and hydroxysteroid dehydrogenase ( 107). Thus, LHR-1 may have a more prominent role than SF-1 in regulating luteal steroidogenesis, e.g. during pregnancy.

Strong evidence for a developmental role of SF-1 comes from studies of forced SF-1 expression in several different SF-1-negative cell types in primary culture. In the first of these studies, embryonic stem cells were induced to express the cholesterol side-chain cleavage enzyme after the forced expression of SF-1 ( 108). Subsequently, mesenchymal cells from mouse and human bone marrow cells were transformed into cells that expressed most of the enzymes required for steroid biosynthesis after the forced expression of SF-1. These cells synthesized steroid hormones de novo, including progesterone, corticosterone, cortisol, dehydroepiandrosterone, testosterone, and estradiol in response to ACTH. In the human cells, SF-1 dramatically induced the expression of the receptors for both ACTH and LH ( 109, 110). Similar results were obtained using adipose tissue-derived mesenchymal cells, but in these cells, adrenal steroid biosynthesis (e.g. corticosterone) predominated, whereas gonadal steroids (e.g. testosterone) were more apparent in bone marrow-derived cells ( 111). These results suggest the potential utility of adipose tissue-derived mesenchymal cells for autologous cell regeneration therapy for patients with adrenal insufficiency.

Although it was anticipated that SF-1 knockout mice might provide in vivo evidence for the importance of SF-1 in gene expression, particularly for those genes involved in steroidogenesis, the failure of the knockout mice to develop the steroidogenic organs and the ventromedial hypothalamus (see below) precluded such analyses in these tissues. However, supporting in vivo roles of SF-1 have been provided for several genes listed in Table 1 including AMH ( 112, 113) and LHB (114).

Gene targets of SF-1 in pituitary and hypothalamus

The finding of SF-1 expression in pituitary gonadotropes and in the VMH (see above) prompted a search for SF-1-regulated genes in these tissues. A series of in vitro DNA-binding assays and reporter gene assays of promoter activity suggested that the expression of the α-subunit of glycoprotein hormones [α-GSU ( 12)], LH ß (115), the GnRH receptor ( 116), FSH ß (117) in pituitary gonadotropes were all regulated by SF-1 As discussed below, SF-1 clearly plays an important role in the expression of these genes in vivo as well.

In silico analyses of putative SF-1-binding sites among VMH-enriched transcripts, followed by promoter activity assays in vitro, suggested roles for SF-1 in the expression of the N-methyl-d-aspartate receptor, the cell adhesion molecules Amigo2, Cdh4, Sema3a, Slit3, and Netrin3, and other genes within the VMH, including Fezf1, Nptx2, Nkx2-2, and A2bp1 ( 118). It remains to be determined whether the SF-1-dependent expression of these transcripts can be confirmed in the VMH-specific SF-1 knockout mice discussed below.

Total disruption of SF-1

Consistent with the hypothesis that SF-1 is required for adrenal and gonadal steroidogenesis, SF-1 knockout mice die shortly after birth from adrenocortical insufficiency and exhibit male-to-female sex reversal of the external genitalia. However, these abnormalities are not due to poor expression of steroidogenic enzymes, but instead to complete absence of the adrenal glands and gonads ( 119, 120). The initial stages of adrenal and gonadal development occur in the absence of SF-1 but subsequently regress. Because their gonads regress before male sexual differentiation normally occurs, the internal and external urogenital tracts of SF-1 knockout mice are female, irrespective of genetic sex. Heterozygous SF-1 knockout mice have decreased adrenal volume associated with impaired corticosterone production in response to stress ( 121).

The gonadotropes of SF-1 knockout mice also have impaired expression of a number of gene products that regulate reproduction, including LH-ß, FSH-ß, αGSU, and the receptor for GnRH ( 13, 122). Moreover, knockout mice lack the VMH, a hypothalamic region linked to feeding and appetite regulation and female reproductive behavior ( 122, 123). Finally, although the functional consequences remain to be defined, the SF-1 knockout mice have defects in their splenic parenchyma ( 15).

Tissue-specific knockout of SF-1

Mutations of SF-1

In contrast to homozygous SF-1 null mice, which can be kept alive with glucocorticoid replacement, no human has been identified who is homozygous for null mutations in the NR5A1 gene encoding SF-1. The first two SF-1-defective patients identified were heterozygous for mutations that abolished the ability of SF-1 to bind DNA, thus rendering the protein transcriptionally inactive without creating a dominant-negative effect. Both affected patients had adrenal insufficiency; one was a phenotypic female XY infant, and the other was an apparently normal XX female who was not diagnosed with adrenal insufficiency until 14 months of age ( 131, 132). A third infant, who had adrenal failure and complete 46,XY sex reversal, was homozygous for a mutation that altered a highly conserved residue in a secondary DBD that causes partial loss of DNA binding and transcriptional activity ( 133). Heterozygous carriers of this mutation were normal; thus, heterozygosity for apparently null mutations or homozygosity for a mutation that only partially reduces activity yield similar phenotypes.

These data suggested that, in humans, male gonad development and adrenal development in both sexes required SF-1 expression in a dosage-sensitive manner. However, heterozygous SF-1 mutations subsequently were identified mainly among undervirilized 46,XY individuals (i.e. XY disorders of sexual development) who do not have adrenal insufficiency ( 134). Thus, the SF-1 genotype alone seems insufficient to explain why some patients have adrenal insufficiency, whereas others do not.

Additionally, SF-1 mutations impairing transcriptional activity have been identified among 46,XX women with premature ovarian failure in four kindreds with histories of both 46,XY disorders of sex development and 46,XX primary ovarian insufficiency and in two of 25 subjects with sporadic ovarian insufficiency. None of these subjects had clinically apparent adrenal insufficiency. A range of ovarian anomalies was found including 46,XX gonadal dysgenesis and primary ovarian insufficiency ( 135).

The adverse effects of heterozygous SF-1 mutations on reproductive function in both males and females could explain the failure to identify humans who are homozygous for null mutations, although it remains possible that such a condition is embryonically lethal in humans, perhaps (for example) owing to adverse effects on CNS development.

Dysregulation of SF-1

What mechanisms regulate the expression of SF-1?

Whereas specific genetic regions have been identified as necessary for SF-1 expression in the gonads, fetal adrenal gland, pituitary gonadotropes, and VMH (see above), the regions required for SF-1 expression in the adult adrenal cortex are as yet unknown, as are many of the transacting factors that presumably regulate the ordered expression of SF-1 in the various tissues.

What mechanisms regulate SF-1 function?

SF-1 clearly is a master regulator of transcription that directs cell-selective gene expression within the endocrine system; however, its activity is regulated by phospholipids, by a large array of transcription factors and coregulators, and by posttranscriptional modification. Whether phospholipids ligands serve as SF-1 activators or merely as stabilizers of SF-1 function is uncertain and although a regulatory role for sphingosine as an inhibitory ligand is somewhat more compelling, it remains to be determined whether its actions can be extended beyond a simple reporter gene assay in a specific cell line. Furthermore, the fact that SF-1 interacts with such a large number of proteins that modify its activity in vitro (Table 2) suggests a promiscuity that seems inconsistent with the tissue-specific functions of SF-1. Ultimately, these interactions must be placed in a context that explains the temporal as well as the tissue-specific and gene-specific effects of SF-1.

What contributions does SF-1 make to appetite control and anxiety?

The VMH-specific knockout of SF-1 in mice leads to appetite and anxiety disorders; however, it is not clear whether SF-1 function in this context is merely permissive and related to the development of the VMH, or whether SF-1 regulates the expression of genes in the VMH that directly contribute to the aberrant phenotypes.

How do SF-1 mutations in humans impact on phenotype?

The apparent discordance in individual phenotypes produced by SF-1 mutations that disrupt SF-1 function requires further investigation. Although the SF-1 mutations described to date have been found in patients with adrenal and/or gonadal deficiency, one wonders if new mutations will be found with phenotypes resulting from effects at other sites, e.g. the VMH or spleen.

NURSA Molecule Pages:

• Nuclear Receptors: SF-1.