Molecular mechanisms of DAX1 action

Transcription

1 Molecular Genetics and Metabolism 83 (2004) Minireview Molecular mechanisms of DAX1 action Anita K. Iyer a, Edward R.B. McCabe a,b,c,d,e, a Department of Human Genetics, David GeVen School of Medicine at UCLA, Los Angeles, CA, USA b Department of Pediatrics, David GeVen School of Medicine at UCLA, Los Angeles, CA, USA c UCLA Molecular Biology Institute, Los Angeles, CA, USA d Mattel Children s Hospital at UCLA, Los Angeles, CA, USA e UCLA Center for Society, The Individual and Genetics, Los Angeles, CA, USA Received 25 May 2004; received in revised form 12 July 2004; accepted 13 July Abstract DAX1 (dosage sensitive sex reversal (DSS), adrenal hypoplasia congenita (AHC) critical region on the X chromosome, gene 1) encoded by the gene NR0B1, is an unusual orphan nuclear receptor that when mutated causes AHC with associated hypogonadotropic hypogonadism (HH), and when duplicated causes DSS. DAX1 expression has been shown in all regions of the hypothalamic pituitary adrenal gonadal (HPAG) axis during development and in adult tissues, suggesting a critical role for DAX1 in the normal development and function of this axis. Steroidogenic factor 1 (SF1, NR5A1) knockout mice show similar developmental defects as AHC and HH patients, but paradoxically, DAX1 is a negative coregulator of SF1 transactivation. The function of DAX1 as an antagonist of SF1 in gonadal development is consistent with the fact that in humans, duplication of the region of the X chromosome containing DAX1 causes a similar phenotype as mutations in SF1. However, how disruption of DAX1 leads to adrenal, hypothalamic, and pituitary developmental defects similar to SF1 disruption remains to be clariwed. The exact mechanism of DAX1 action in each of these tissues during adulthood and critical stages of development are not fully understood. Recent evidence suggests a broader functional role for DAX1 as a negative coregulator of estrogen receptor (ER, NR3A1-2), liver receptor homologue-1 (LRH- 1, NR5A2), androgen receptor (AR, NR3C4), and progesterone receptor (PR, NR3C3), each by distinct repression mechanisms. DAX1 may have pleiotropic roles in addition to its function as a negative regulator of steroidogenesis during the development and adult function of the HPAG axis Elsevier Inc. All rights reserved. Keywords: DAX1; NR0B1; Adrenal hypoplasia congenita; SF1; AR; ER; PR; LRH-1; Hypothalamic pituitary adrenal gonadal axis; Nuclear receptor Introduction Adrenal hypoplasia congenita (AHC) is an inherited disorder characterized by underdevelopment of the adrenal cortex [1]. It has an estimated frequency of 1:12,500 live births and presents in two histological forms: the miniature adult and the cytomegalic forms. The adrenal glands of patients with the miniature adult form have a * Corresponding author. Fax: address: emccabe@mednet.ucla.edu (E.R.B. McCabe). permanent zone that has normal structural zonation but is smaller than normal, with a minimal or absent fetal cortex. This form is generally associated with abnormal central nervous system and pituitary development and function, and is either sporadic or inherited in an autosomal recessive manner. In the cytomegalic form of AHC, associated with NR0B1 mutations, the permanent zone of the cortex is absent or nearly absent, and the residual adrenal cortical tissue is structurally disorganized with large vacuolated cells that most closely resemble those in the fetal adrenal cortex, resulting in an adrenal cortex that lacks normal postnatal zonation and is dysfunctional /$ - see front matter 2004 Elsevier Inc. All rights reserved. doi: /j.ymgme

2 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) This form primarily avects males as it is inherited in an X-linked manner. Patients with X-linked cytomegalic AHC present with adrenal insuyciency early in childhood, and exhibit saltwasting, hypotension, hyperpigmentation, hyponatremia, hyperkalemia, hypoglycemia, decreased glucocorticoid and aldosterone production, and increased levels of adrenocorticotropic hormone (ACTH) [1]. This disease is lethal without glucocorticoid and mineralocorticoid replacement therapy. Some patients with X-linked AHC who survive beyond childhood develop hypogonadotropic hypogonadism (HH), in which a mixed hypothalamic and pituitary defect in the secretion of gonadotropins prevents normal puberty and requires treatment with testosterone for sexual maturity [1,2]. X-linked AHC was originally mapped to Xp21 and the NR0B1 gene was subsequently identiwed by positional cloning as the causative agent of AHC, with mutations or deletions in the NR0B1 gene identiwed in AHC patients [3 5]. AHC can also present as part of an Xp21 contiguous gene syndrome along with phenotypes of glycerol kinase (GK) dewciency, Duchenne muscular dystrophy (DMD), and mental retardation due to a large deletion encompassing NR0B1 and the neighboring GK and DMD loci [1]. AHC patients with defects in the NR0B1 gene also develop HH, thus identifying the mutant DAX1 as the causative agent of both disorders [3 5]. DAX1 has also been shown to be involved in sex determination and gonadal development [6,7]. Male to female sex reversal in XY individuals with an intact SRY gene was mapped to a 160 kb region of Xp21, which includes NR0B1 [6]. Duplication of NR0B1, the gene that encodes DAX1, in these sex reversed patients makes NR0B1 a very strong candidate for the dosage sensitive sex reversal gene (DSS). NROBI genomic organization and DAX1 protein domain structure NR0B1 has a very simple genomic structure with two exons separated by a single intron [3,5]. Exon 1 is 1168 base pairs (bp) in length, Exon 2 is 245 bp, and the intron is 3385 bp [8]. The 1413 nucleotide cdna encodes a 470 amino acid protein. Hossain et al. [9] have recently identiwed an alternatively spliced isoform designated as DAX1α that speciwes a protein of 401 amino acids encoded by Exon 1 and a previously unidentiwed Exon 2α. This isoform was shown to be expressed in a broad range of tissues, but elucidation of the signiwcance and functional role of DAX1α will require further investigation. DAX1 has been classiwed as an orphan member of the nuclear receptor superfamily [10 12]. Superfamily members have a characteristic domain structure consisting of subregions A E (Fig. 1A). The A/B region is the Fig. 1. Comparison of functional domain structure of members of the nuclear receptor superfamily (A) with DAX1 (B). most evolutionarily divergent and varies in size among family members. It is considered a modulator domain and may contain a hormone independent transactivation domain (Activation Function 1 or AF-1). The C region is typically the most highly conserved and represents a DNA-binding domain (DBD) containing two zinc Wngers that allow the receptor to recognize and bind hormone response elements in the promoters of target genes. The C region also contains receptor dimerization interfaces. The D region serves as a hinge between the DBD and the ligand-binding domain (LBD) and has been shown to serve as a docking site for corepressors. The E region is the second most highly conserved LBD, and mediates ligand binding, dimerization, and nuclear localization. It consists of 12 helices with an AF-2 hormone-dependent transactivation domain in helix 12 that undergoes allosteric conformational changes in response to ligand binding. The DAX1 domain structure is rather unusual (Fig. 1B) [11,12]. The carboxy-terminal domain (CTD) is homologous to the LBD of other nuclear receptors and also contains an AF-2 transactivation domain, but DAX1 lacks the conventional DBD (Region C), modulator domain (Region A/B), and hinge region (Region D). Instead, the DAX1 amino-terminal domain (NTD) has a novel structure consisting of 3.5 alanine/glycine rich repeats of a amino acid motif that has no known homology to any other proteins, with the exception of the related nuclear receptor superfamily member, small heterodimer partner (SHP), encoded by NR0B2 [1]. The repeats show 33 70% identity to each other, and also contain cysteine residues in conserved positions that could potentially form zinc Wngers [1]. The C-terminal domain of DAX1 has strongest amino acid similarity to the LBD of the testis receptor, COUP-TF, and retinoid X receptor (RXR) [12,13]. However, the similarities with other receptors are unable to provide information regarding a putative ligand, and to date, no ligand has been identiwed for DAX1. DAX1 is structurally most similar to SHP, in the sense that SHP also lacks the typical nuclear receptor DBD, but has an N-terminal domain similar, but shorter than DAX1 that contains one amino acid repeat, and contains a C-terminal region homologous to nuclear receptor LBDs [14].

3 62 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) Molecular mechanisms of DAX1 action The complex endocrine phenotype caused by DAX1 defects is consistent with its pattern of expression throughout the hypothalamic pituitary adrenal gonadal (HPAG) axis. DAX1 expression has been shown in the developing adrenal cortex, gonad, anterior pituitary, and hypothalamus, and also in adult adrenal cortex, Sertoli and Leydig cells in the testis, theca, granulosa, and interstitial cells in the ovary, anterior pituitary gonadotropes, and the ventromedial nucleus of the hypothalamus [15 18]. Such a pattern of expression suggests that DAX1 is involved in the development and function of the HPAG axis. The DAX1/SF1 functional paradox Interestingly, the tissue expression prowle of DAX1 in development and in the adult organism is similar to that of SF1/NR5A1, an orphan nuclear receptor that is an essential regulator of HPAG axis diverentiation and steroid hormone biosynthesis [19]. The phenotype of SF1-disrupted mice is similar to the phenotype of AHC patients. These mice have aplasia of the gonads and adrenals, and also have pituitary and hypothalamic defects [19]. This observation led to the idea that these two receptors could be involved in a common developmental pathway, with the concerted evort of SF1 and DAX1 necessary for normal development of the HPAG axis. DAX1 and SF1 were shown to have a colocalized tissue prowle during embryonic development and also in adult tissues [16,17], suggesting a functional interaction between DAX1 and SF1, perhaps in a transcriptional cascade or through cooperative transcriptional regulation. Classifying DAX1 as a member of the nuclear receptor superfamily suggested that it may have a bona Wde receptor function to activate target genes involved in HPAG axis development and function [12,20]. Paradoxically, DAX1 has been shown to function primarily as a transcriptional repressor. It has been proposed that DAX1 inhibits expression of steroidogenic acute regulatory protein (StAR) by binding to DNA hairpin structures in the StAR promoter [21]. Most notably, however, DAX1 was shown to act as an inhibitor of SF1-mediated transcriptional transactivation [20,22]. SF1 functions as a transcriptional activator of many genes involved in steroid hormone biosynthesis in the HPAG axis, and DAX1 appears to act by complexing with and inhibiting the activator function of SF1 [23]. The function of DAX1 as an antagonist of SF1 in gonadal development is consistent with the fact that duplication of the NR0B1 gene causes a similar XY sex reversal phenotype as mutations in SF1, with DAX1 repressing, or SF1 unable to activate, genes involved in sex determination. The requirement of SF1 for proper development of the hypothalamus, pituitary, and adrenal glands is consistent with its role as an activator, in which the loss of SF1 function leads to decreased transcription of critical target genes. On the other hand, how disruption of DAX1, the antagonist of SF1 function, can lead to adrenal, hypothalamic, and pituitary developmental defects remains a functional conundrum. DAX1 expression colocalizes with that of SF1 [16,17]; however, the colocalization, as shown by double immunoxuorescence studies, is not absolute [17]. DAX1 and SF1 show distinct expression patterns during gonadal development. DAX1 positive, SF1 negative cells were found in the embryonic mouse testis, postnatal ovary, and most notably in the developing pituitary and hypothalamus, which suggests a function for DAX1 independent of SF1. The expression of DAX1α in an even broader range of tissues [9] suggests far greater pleiotropy. Immunohistochemical studies in developing and adult tissues have shown DAX1 to have primarily a nuclear localization [17,24]. However, studies of DAX1 in cultured cell lines and in ES cells show that DAX1 is present in both the nucleus and the cytoplasm [25 29]. Kawajiri et al. [29] also showed nucleocytoplasmic localization in vivo with immunohistochemical analyses of DAX1 within the developing pituitary in Rathke s pouch, a region where SF1 is not expressed. In addition, Lalli et al. [26] showed an association of DAX1 with polysomes and RNA in the cytoplasm, suggesting a regulatory role at the posttranscriptional level, though the physiological consequence of such an association is not clear. This group also demonstrated shuttling of DAX1 between the nucleus and cytoplasm. These observations suggest that DAX1 may function both in the nucleus and the cytoplasm independent of SF1 and possibly independent of transcriptional silencing. DAX1 may have pleiotropic roles with diverent molecular functions in the context of developing compared with adult tissues, or may have distinct mechanisms of action in speciwc cell types and tissues during development. How disruption of DAX1 leads to AHC and HH appears to be more complex than previously considered, and likely involves molecular functions beyond its role as a repressor of SF1 action. DAX1 contains an AF-2 transactivation domain that is characteristic of nuclear receptors, and therefore it is possible that certain physiological conditions and/or cellular developmental environments could convert this repressor into an activator. Recent studies provide evidence for a wider functional role for DAX1 as a transcriptional repressor of other nuclear receptors expressed in the HPAG axis. DAX1 can also repress the action of the androgen (AR; NR3C4), estrogen (ER; NR3A1-2) and progesterone receptors (PR; NR3C3), and also liver receptor homologue-1 (LRH-1; NR5A2) [22,28,30,31]. What is known about molecular mechanisms of DAX1-mediated

4 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) repression of SF1 and its involvement with these other nuclear receptors are described below. DAX1-mediated repression of SF1 action DAX1 is a negative regulator of SF1-mediated transactivation of many genes in the steroid biosynthetic pathway [23]. In addition, DAX1 inhibits the transcriptional synergy of SF1 with other associated coregulators throughout the HPAG axis. DAX1 antagonizes the cooperation of SF1 with Wilms Tumor 1 (WT1) and with GATA-4 in regulation of Müllerian inhibiting substance (MIS) in the gonad [32,33], SF1 with GATA-6 in transcriptional activation of genes in androgen biosynthesis in the adrenal [34], SF1 with Egr-1 in the transcription of LHβ in pituitary gonadotropes [35], and SF1 with CREB in the transcription of gonadotropin-inducible ovarian transcription factor 1 (GIOT1) in ovarian granulosa cells [36]. DAX1 also negatively regulates SF1 and SREBP-1a-mediated transcription of the HDL-R gene [37]. Mechanisms of repression Transcriptional silencing of SF1 is thought to involve direct protein protein interactions between DAX1 and DNA-bound SF1 via the DAX1 N-terminal domain, with the subsequent recruitment of corepressors to the promoters of target genes via a DAX1 C-terminal transcriptional silencing domain [20,38] (Fig. 2A). It has been suggested that DAX1 inhibition of SF1 requires multiple SF1-binding sites in the promoters of target genes [39]. Transcriptional coactivators typically bind the AF-2 domain of activated nuclear receptors through a Fig. 2. Mechanisms of DAX1-mediated repression of SF1, ER, and LRH-1. (A) DAX1 binds the AF-2 domain of the nuclear receptors via its LXXLL motifs and recruits corepressor proteins to target gene promoters. (B) EVects of intracellular levels of DAX1 and SF1 on transcriptional repression. Increased SF1 levels relative to DAX1 favor transcriptional activation (left), and increased levels of DAX1 relative to SF1 favor transcriptional repression (right). conserved LXXLL motif called the NR box. DAX1 contains three LXXLL motifs, and has been shown to bind the AF-2 domain of SF1 via these motifs, making DAX1 an LXXLL containing-corepressor [22,30]. Mutations or deletions of the DAX1 LXXLL motifs impair its repressor activity against SF1 [22]. SF1 has been shown to interact with a number of transcriptional coactivators, including CBP/p300, GRIP1, and SRC-1 [40]. SRC-1 typically interacts through its LXXLL motifs and has been shown to interact with the AF-2 domain of SF-1 [41]. It is possible that DAX1 and the transcriptional coactivators compete for a similar binding site on SF1, with DAX1 displacing the coactivators from SF1 in certain conditions to promote transcriptional repression. This coactivator competition mechanism of repression has been shown for SHP, the orphan nuclear receptor that has strongest structural similarity to DAX1 and is also an LXXLL-containing corepressor [42]. SHP has been shown to interact with the AF-2 domains of its target receptors, including HNF-4α, RXR, and ER, and has been shown to compete with coactivators for binding to the AF-2 [42,43], which makes this a likely mechanism for DAX1-mediated repression. The transcriptional silencing domain of DAX1 has been mapped to a bipartite region in the C-terminal LBD-like (LBD-L) domain which includes two groups of residues, one at each end of the LBD-L [20,38]. Interestingly, all documented AHC missense mutations also map to this region [44], and it has been shown that AHC mutations abolish transcriptional silencing activity, suggesting that a lack of silencing could be involved in the pathogenesis of AHC. The mechanism of DAX1-mediated transcriptional repression involving corepressor recruitment was prompted by a squelching experiment where cotransfection of excess DAX1 was able to relieve DAX1-mediated repression, suggesting that this relief was due to the titration of cellular corepressors by the excess DAX1 [38]. DAX1 was subsequently shown to interact with the corepressors NCo-R and Alien, but not SMRT [45,46] through portions of the C-terminal domain that corresponded with transcriptional silencing. However, the interaction between DAX1 and NCo-R appears to be very weak [27,45], which indicates that the interaction could involve protein domains diverent from those used in these studies, or could be stabilized by unknown proteins or conditions in speciwc cell types where DAX1 action could be taking place. A recent study by Eckey et al. [47] demonstrates a direct interaction between Alien and mixed lineage kinase 2 (MLK2) and also shows an enhancement of DAX1-mediated repression in the presence of MLK2, suggesting that the interaction of Alien and MLK2 possibly stabilizes a corepressor complex or recruits additional complexes for eycient repression. In addition, DAX1 could not fully relieve its own transcriptional silencing [38], suggesting corepressor-independent mechanisms of

5 64 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) transcriptional repression. It is also possible that DAX1 could interact with other novel unidentiwed corepressors similar to those that interact with other nuclear receptors that function as corepressors, such as testis receptor and COUP-TF. Though DAX1 and SHP share structural similarities, mechanisms of SHP-mediated transcriptional repression appear to be diverent from DAX1. Similar to DAX1, transcriptional silencing by SHP not only involves coactivator competition mediated by the LXXLL motifs, but also a C-terminal repression domain. The precise mechanism of silencing via this domain is unclear but is thought to involve the recruitment of unidentiwed corepressors, as SHP has been shown not to interact with NCo-R [48], but has been shown to interact with the cofactor EID-1, which antagonizes the function of CBP/ p300 coactivators [49]. DAX1 and SHP both contain a 25 and 12 amino acid insertion, respectively, between helix 6 and helix 7 of the LBD-L/transcriptional silencing domain that is not present in other members of the superfamily [13,38,50]. Recent studies by Park et al. [50] elucidated the role of this insertion in DAX1 and SHPmediated transcriptional repression. This insertion appears to have an important role in repression by SHP, but not DAX1, as deletion of this insertion impaired the ability of SHP to repress its target orphan receptors, but did not impair the ability of DAX1 to repress SF1. The insertion was necessary for the interaction of SHP with EID-1, as it is thought to provide structural spacing for eycient interaction. Interestingly, DAX1 did not interact with EID-1, providing further evidence for a distinct repression mechanism for SHP, and also lending credence to the concept of DAX1 recruiting corepressors involved in histone deacetylase function. Though the insertion did not impair the ability of DAX1 to repress SF1, DAX1 could repress other nuclear receptors through additional mechanisms that may rely on this insertion. Gene dosage and subcellular localization Transcriptional regulation of SF1 target genes, including those in steroidogenesis both during development and in the adult organism appear to be, in part, regulated by intracellular levels of SF1 and DAX1, with the ratio of these two factors determining whether the target genes are activated or repressed (Fig. 2B). If more SF1 is present in a cell than DAX1, SF1 molecules will outnumber the SF1 DAX1 complexes and the target genes will be activated. Conversely, if more DAX1 is present, the SF1 DAX1 complexes will outnumber the SF1 molecules and the target genes will not be activated and therefore will be repressed. This mechanism also alters the balance of coactivators and corepressors that are recruited to the promoter. This concept applies to target genes that require the synergy of SF1 with an associated cofactor, such as WT-1 and GATA-4 in the transcription of MIS [32,33]. DAX1 does not inhibit transcription of or interact with WT-1 and GATA-4 alone, but does inhibit synergy with SF1. It is not clear whether DAX1 forms a complex with SF1 and the cofactors, or whether DAX1 competes with the cofactors for binding to SF1, though evidence from Nachtigal et al. [32] seem to suggest the former. This type of regulation emphasizes the critical nature of gene dosage in development. For example, reduction of WT1 in Denys Drash syndrome, or increased DAX1 levels in DSS seem to explain the observed antagonized male development. Both of these situations shift the balance of cofactors to favor the DAX1 SF1 interaction and thus transcriptional repression. An increase in DAX1 expression and a downregulation of SF1 expression via the MAPK pathway has been suggested to explain the decreased steroidogenesis in response to prolonged stimulation of ovarian granulosa cells with gonadotropins [51]. In the zona glomerulosa in the adrenal gland, aldosterone biosynthesis is promoted by angiotensin II in part by downregulating DAX1 expression, and by stimulation of the camp signal to mimic an ACTH signal that results in downregulation of DAX1 expression and upregulation of SF1 expression [52]. The subcellular localizations of DAX1 in cultured cell lines and in ES cells have been shown by many groups to be nuclear and cytoplasmic [25 29], with evidence that DAX1 can shuttle between the compartments [26]. Coexpression of DAX1 and SF1 in cultured cells results in DAX1 shifting completely to the nucleus [29]. A similar phenomenon has been observed for SHP and HNF4α [53]. This nuclear localization with SF1 appears to require direct interaction with SF1 as the LXXLL motifs in isolation were able to promote nuclear localization in the presence of SF1, and mutation of the these motifs impaired this process. This phenomenon also appears to require the AF-2 domain of DAX1, as mutation of these residues also lowered the frequency of nuclear localization in the presence of SF1, implicating the AF-2 in nuclear localization processes. This observation appears to indicate that in order for DAX1 to carry out its function as a repressor of SF1, it needs to localize to the nucleus. The mechanism of nuclear import is not clear. DAX1 may shuttle between the cytoplasm and nucleus as previously described [26], and may be retained in the nucleus by SF1. Alternatively, DAX1 may bind SF1 in the cytoplasm followed by translocation into the nucleus. Once in the nucleus, DAX1 can repress SF1 as described above. A recent confocal imaging study in KGN granulosa cells describes how the protein kinase A (PKA) pathway stimulates SF1 transactivation and involves DAX1 [54]. In the absence of stimulation with forskolin, the nuclear distribution of SF1 is very divuse, but in the presence of stimulation, SF1 is rearranged into distinct foci, which are indicative of sites of transcrip-

6 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) tionally active nuclear receptors. In the presence of stimulation, these foci also colocalize with transcriptional coactivators GCN5 and TRRAP suggestive of a functional interaction of SF1 and the recruitment of a coactivator complex. DAX1 and SF1, when coexpressed in the absence of stimulation, have a speckled/dotted distribution with no divuse background, and colocalize completely in the nucleus. This is consistent with the results of Kawajiri et al. [29]. When stimulated, the speckled/ dotted distribution remains, but DAX1 and SF1 appear to separate from each other. The activation of the PKA pathway may weaken the DAX1 SF1 interaction. Fluorescence recovery after photobleaching (FRAP) studies show that DAX1 immobilizes SF1 and anchors it to a given site in the nucleus. These results suggest that DAX1 may be capable of interacting with the nuclear matrix. This immobilization is relieved in the presence of a PKA signal. This phenomenon has been observed for ER in the presence of antagonist [55]. The simultaneous distribution between DAX1, SF1, and the GCN5/ TRRAP complex was not studied. These results are likely not an artifact of overexpression as transiently transfected cells that expressed amounts of protein close to physiological levels were selected for imaging. These observations suggest that coexpression of DAX1 and SF1 causes DAX1 to localize fully to the nucleus where it can repress SF1 transactivation through direct protein interactions and the recruitment of corepressors and also by immobilizing SF1. Activation of pathways that stimulate SF1 action cause DAX1 and SF1 to dissociate, promoting SF1 interactions with coactivators, and activation of SF1 target genes. DAX1-mediated repression of ER and LRH-1 action DAX1 has been shown to repress transcriptional transactivation of ER and LRH-1 [22,30,56]. This repression is thought to occur through a similar mechanism as the repression of SF1 action (Fig. 2A). The LXXLL motifs of DAX1 have been shown to interact with the AF-2 domain of ER [22,30], and these motifs appear to be required for interaction with LRH-1 [22]. DAX1 interacts with both ERα and ERβ. The interaction with ERα is ligand dependent, whereas the interaction with ERβ appears to be ligand independent. DAX1 does not interfere with dimerization or DNA binding of ER, but forms a ternary complex on ER response elements. DAX1 is expressed in many estrogen target tissues, with a suggested coexpression in testis and ovary during development, making the ER a likely physiological target in the developing and adult reproductive system [30]. It is possible that DAX1 is involved in the control of the physiological response to estrogen in the adult organism, or may repress ER target genes during critical times and cell types in development. SpeciWc target genes have not yet been identiwed, and further studies of colocalization of DAX1 and ER expression during development and adult reproductive processes need to be conducted in order to understand fully the physiological role of this interaction. LRH-1, in addition to expression in pancreas, liver, and intestine and its function in bile acid metabolism, is also expressed in the ovary, testis, and adrenal. LRH-1 is a monomeric orphan nuclear receptor that can recognize the same response elements as SF1 and can substitute for SF1 in the activation of steroidogenic enzymes, suggesting a function in the regulation of steroidogenesis [57]. LRH-1 has been shown to upregulate expression of aromatase and 3β-hydroxysteroid dehydrogenase type II (3βHSD2) in the ovary [56,58]. SF1 and LRH-1 are expressed in diverent sites in the ovary at diverent times during the menstrual cycle and during pregnancy [58]. SpeciWcally, during the conversion of a mature ovarian follicle to the corpus luteum following the LH surge, there is a shift to progesterone biosynthesis. LRH-1 is thought to be involved in this regulation of steroidogenesis as SF1 expression is downregulated in the corpus luteum [56,58]. DAX1 and LRH-1 both appear to be expressed in granulosa cells [58,59] and DAX1 was shown to inhibit LRH-1-mediated transcription of 3βHSD2 in granulosa cells [56]. It is possible that DAX1 can repress other LRH-1 target genes in the ovary. The DAX1 interaction with LRH-1 appears to be noticeably stronger than the interaction with SF1 as seen in an in vitro binding assay and also a mammalian two-hybrid assay [22]. These results may be indicative of Wne tuning of a regulatory mechanism to allow for a critical amount of repression with regard to SF1 to prevent gain of function and loss of function phenotypes as observed in AHC and DSS. Perhaps the SF1/DAX1 interaction is transient in the context of physiological and developmental function, occurring for very speciwc lengths of time that could alter an overall physiological response should the interaction be stronger and more stable. SF1 may be in a conformation to discourage a strong interaction compared to LRH-1. It is also possible that there are cell-speciwc coregulators in cell types within which SF1 and DAX1 would exert their actions to stabilize the interaction. Since DAX1 appears to interact with ER and LRH-1 through the LXXLL motifs and repress ER and LRH-1 by similar mechanisms, it can be expected that coexpression of DAX1 with either ER or LRH-1 will cause DAX1 to localize to the nucleus. This has been shown for ERα [29] and thus seems highly likely for LRH-1. It is also possible that DAX1 could immobilize these receptors in the nucleus, but this remains to be determined. Moore et al. [60] have recently shown an interaction between an LXXLL motif of DAX1 and the thyroid receptor β (TRβ) LBD through an in vitro binding assay. The functional implications of this physical interaction require further investigation, but it is possible

7 66 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) that DAX1 could repress TRβ, and could do so via a repression mechanism similar to that of SF1, ER, and LRH-1. DAX1-mediated repression of AR and PR action DAX1 has been shown to interact with and function as a negative coregulator of AR [28,31,61] and PR [31], suggesting a role for DAX1 in the normal development and adult function of the reproductive system, but the mechanism of DAX1-mediated repression of these receptors appears to be diverent from that of SF1, ER, and LRH-1 (Figs. 3A and B). Unliganded AR is present in the cytoplasm, but is transported to the nucleus upon ligand binding in order to carry out its function. Holter et al. [28] showed that DAX1 inhibits AR transactivation by interfering with the dimerization and interdomain communication of the AR, and further showed through transient transfections and intracellular localization studies that cytoplasmic DAX1 tethered AR in the cytoplasm in the presence of ligand, preventing its translocation into the nucleus. This tethering occurs independently of ligand and is between the AR LBD and the N-terminus of DAX1. Nuclear DAX1 colocalized with AR in the nucleus in the presence of ligand, suggesting not only that the liganded-ar can enter the nucleus properly, but that DAX1 may also possess a mechanism for repressing AR in the nucleus. Agoulnik et al. [31] further investigated DAX1 repression of AR and provided evidence of a nuclear mechanism of transcriptional repression. DAX1 was shown to repress AR activity in the presence of an agonist and an antagonist. An in vivo interaction of endogenous DAX1 with AR was also conwrmed in an agonist-treated prostate cancer cell line, suggesting a physiologically relevant Fig. 3. Other mechanisms of DAX1-mediated repression. (A) Nuclear DAX1 inhibits AR- and PR-mediated transactivation through interfering with functional dimerization of the nuclear receptors. (B) Cytoplasmic DAX1 inhibits AR transactivation by preventing its translocation into the nucleus. (C) DAX1 may inhibit transcription by binding to hairpin elements in the promoters of target genes. interaction. The repression of AR by DAX1 does not appear to involve histone deacetylases, as there was no relief of repression in the presence of a histone deacetylase inhibitor. DAX1-mediated silencing was reversed by the addition of excess coactivator. DAX1 does not interfere with AR interaction with SRC-1, rather it interferes with the dimerization and interdomain communication of AR, similar to the results of Holter et al. [28]. DAX1 appears to interfere with the formation of functional coactivator complexes by binding to sites diverent than coactivators like SRC-1 while also preventing the formation of transcriptionally active AR dimers. Repression of AR by DAX1 involves both the N- and C-terminal domains of DAX1, consistent with the previous observation that the DAX1 AR interaction involves both halves of DAX1 [28]. Whereas Holter et al. [28] showed that cytoplasmic tethering involves the AR LBD, Agoulnik et al. [31] show that DAX1 is capable of repressing a truncated AR lacking the LBD. Perhaps this provides evidence for two distinct mechanisms by which DAX1 can repress AR depending on its localization. Despite the lack of cytoplasmic tethering in the truncated AR lacking the LBD, DAX1 can still repress AR via the AR N-terminus in the nucleus [31]. DAX1 does not interfere with binding of AR to DNA, which seems to conxict with the concept of DAX1 tethering AR in the cytoplasm and thus preventing its binding to DNA. However, the localization of DAX1 varies depending on culture conditions, and the number of cells in a given population with cytoplasmic versus nuclear DAX1 varies. DAX1 is present in both the nucleus and the cytoplasm regardless of the presence of AR, so it is feasible for both cytoplasmic and nuclear mechanisms to be valid. While these results seem to suggest that DAX1 may possess two mechanisms of repressing AR, depending on whether DAX1 is present in the cytoplasm or in the nucleus, it is also possible that the cytoplasmic tethering mechanism is an artifact of overexpression. It has been suggested that DAX1 may interact with AR through the LXXLL motifs for cytoplasmic retention [28]; however this logic conxicts with the observation that an interaction through the LXXLL motifs with SF1 and ER causes translocation of DAX1 and the nuclear receptor into the nucleus. The speciwc residues involved in the DAX1 AR interaction have not been studied. These results indicate a repression mechanism distinct from that of SF1 and ER. Agoulnik et al. [31] also studied DAX1 repression of the PR. DAX1 repressed agonist-, but not antagonistbound PR activity. DAX1 interacted with PR-A and PR-B in vivo in an agonist-treated breast cancer cell line. DAX1 did not interfere with the interaction of PR and SRC-1, indicating that DAX1 binds to diverent sites on PR than SRC-1. DAX1 did, however, interfere with PR homodimerization. A functional interaction takes place between the PR LBD and the N-terminus of DAX1,

8 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) though the C-terminus may also be involved as full length DAX1 was a more potent repressor of PR. It is possible that this interaction may involve the LXXLL motifs of DAX1 and the AF-2 of PR, but this seems unlikely in light of the lack of competition for binding of PR to SRC-1. DAX1 did not interfere with the binding of PR to DNA. These results suggest a mechanism diverent from that of SF1 and ER, but also slightly diverent from AR with regard to protein domains involved. DAX1 may interfere with the speciwc mechanisms of transcriptional activation of a given receptor. DiVerences in mechanisms of DAX1-mediated repression DAX1 has been shown to repress transactivation of other nuclear receptors by both nuclear and cytoplasmic mechanisms. For example, DAX1 functions as a transcriptional repressor of SF1 by heterodimerization with SF1 in the nucleus, and thereby interfering with the protein protein interactions between SF1 and the heterodimeric partners with which SF1 cooperates in transcriptional activation [32 37] (Fig. 2). As another example, however, DAX1-mediated repression of AR action appears to involve cytoplasmic and nuclear mechanisms (Figs. 3A and B). DAX1 has been observed to tether AR in the cytoplasm, even in the presence of ligand, and to prevent nuclear translocation of AR [28]. DAX1 also appears to repress AR in the nucleus [28,31]. Transcriptional repression by DAX1 displays speci- Wcity for certain nuclear receptors. DAX1 does not repress the action of HNF4α, RORα, VDR, and p53 [22,31]. Thus, molecular mechanisms of DAX1 action appear to be restricted in terms of target receptor speciwcity, but DAX1 repression of ER, LRH-1, AR, PR, and possibly TRβ, suggests that DAX1 may have broader functional roles in HPAG development and adult function than previously considered (Fig. 4). It is informative to speculate on the origin and nature of DAX1 repression. DAX1 serves as a transcriptional silencer for a broad range of nuclear receptors and yet DAX1 is restricted in its target speciwcity. DAX1 evidences nuclear or cytoplasmic mechanisms for repression of its target receptors transactivation functions, and for some targets DAX1 may function, perhaps by distinct and independent mechanisms, in both cellular compartments. Therefore, the breadth of receptors with which DAX1 interacts could be a consequence of diverse molecular mechanisms acquired in the course of its evolution [62]. Alternatively, these mechanistic diverences could rexect variations in the experimental designs employed by various investigators. Supra-physiological levels of DAX1 expression, for example in transient transfection assays [31], could lead to spurious interactions with nuclear receptors that would not be targeted under normal conditions or would interact through a diverent mechanism in the presence of physiological concentrations of DAX1 protein. Role of DAX1 in HPAG axis development and adult function Though DAX1 expression has been shown in all regions of the HPAG axis during development and in adult tissues, and coregulator and posttranscriptional functions have been shown in vitro, the exact mechanisms Fig. 4. Role of DAX1 and mechanisms of action. DAX1 may have pleiotropic roles in early embryonic development, bone cell development, and HPAG axis development and adult function, and may repress the action of various nuclear receptors depending on cellular and physiological context. Solid lines show dewnitive stimulatory and inhibitory relationships. Dotted lines indicate relationships where tissue coexpression has been shown, but functional activity remains to be determined. See text for details.

9 68 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) of DAX1 action in each of these tissues, and how disruptions of these mechanisms cause AHC and HH are not fully understood. The following sections outline possible roles for DAX1 in HPAG axis development and function. Combined evidence from in vitro and in vivo models suggest that DAX1 may have pleiotropic roles, with complex and distinct functions in development and adult function throughout the HPAG axis (Fig. 4). DAX1 as an inhibitor of steroidogenesis The observation of DAX1 functioning as a negative coregulator of SF1 transactivation suggested that DAX1 may have a role as an inhibitor of steroidogenesis in the adrenal gland and the gonads [20]. DAX1 blocks overall steroid production in forskolin-stimulated Y-1 adrenocortical cells at multiple levels of the steroid biosynthetic pathway without an evect on the camp signaling pathway [63]. DAX1 has been shown to inhibit SF1-mediated transactivation of many steroidogenic enzymes [23]. Gonads The transcriptional inhibition of speciwc steroidogenic enzymes by DAX1 seems to depend on the sex of the organism, the organ, and physiological context. DAX1 was shown to inhibit transcription of LRH-1- mediated transcription of 3βHSD2 in a granulosa cell line, implicating DAX1 in the regulation of ovarian steroidogenesis during the menstrual cycle [56]. In the Leydig cells of DAX1-deWcient mice, aromatase is speciwcally upregulated at the mrna and protein level, while StAR, P450 side chain cleavage enzyme (P450scc), 17α-hydroxylase (Cyp17), 3βHSD2, and 17βHSD3 were not [64]. These results suggested that increased estrogen production could explain the infertility in these mice. In addition, aromatase expression was not upregulated in the ovaries of female DAX1-deWcient mice [64]. The observation of the lack of DAX1 repression for other steroidogenic enzymes appears to conxict with observations in cultured cell lines that DAX1 inhibits StAR and 3βHSD expression [21,56,63]. These diverences could represent bona Wde physiological tissue speciwcity, or they could also be artifacts of overexpression of DAX1 in transfected cultured cell lines or due to the possible presence of a hypomorphic allele in the DAX1-deWcient mice [65]. Adrenal gland DAX1 has been shown to repress SF1-mediated transcription of genes involved in adrenal androgen biosynthesis (StAR, P450scc, and Cyp17) in cultured cells [34]. DAX1 has also been shown to repress aldosterone biosynthesis in bovine adrenal zona glomerulosa cells that can be relieved by angiotensin II [52]. DAX1-deWcient mice appear to have increased levels of 21-hydroxylase (Cyp21) and the ACTH receptor in the adrenal glands in response to stress compared with wildtype [66]. DAX1 represses SF1-mediated transcription of Cyp17 in a human adrenocortical cell line [39]. These results portray DAX1 as a negative regulator of steroidogenesis in the adrenal gland. Disruption of normal DAX1 function in the adrenal gland and pathogenesis of AHC Patients with the cytomegalic form of AHC present with adrenal insuyciency due to low levels of steroid production [1]. Although there is a great deal of evidence for DAX1 functioning as a negative regulator of steroidogenesis, this appears to be inconsistent with the patient phenotype. These AHC patients have small, nonfunctioning adrenal glands, and thus the adrenal insuyciency is due to a developmental defect rather than a defect in postnatal function. These observations indicate a critical role for DAX1 in the normal development of the adrenal cortex. DAX1 is expressed throughout adrenal cortical development, but its function in adrenal development is poorly understood. The phenotype of DAX1-deWcient mice [65] was quite diverent from that of AHC patients. These mice had fully developed adrenal glands with functional zonation and normal serum corticosterone levels, but the X zone (roughly equivalent to the human fetal zone) failed to regress at the normal time of puberty. These observations suggest that DAX1 is involved in fetal adrenal degeneration but is not necessary for formation of the dewnitive zone or for steroidogenesis. The diverences between AHC patients and the mice may be due to species diverences or the disrupted DAX1 gene acting as a hypomorphic allele. DAX1 may be involved in adrenocortical growth and may interact with SF1 in an antagonistic manner during development. This is supported by the fact that SF1 haploinsuycient mice display adrenal glands that are smaller than wildtype, while compound DAX1-deWcient, SF1 haploinsuycient mice do not display the growth defect [67]. DAX1-deWcient mice contain a unique steroidogenically active cell layer contiguous with the X zone that is rescued by SF1 haploinsuyciency, suggesting that DAX1 is involved in the regression of this particular cell layer [68]. The signiwcance of steroid production during development is unclear [23], but it is possible that DAX1 could inhibit steroid production at critical periods in development. The possibility exists that DAX1 may also play an unknown function in the context of development, possibly as a transcriptional activator, or through modulating the actions of other nuclear receptors, that could explain the AHC phenotype. The function of DAX1 as a negative regulator of steroidogenesis may have a more signiwcant role in the adult adrenal. Adrenal steroid production is tightly regulated, as steroids are secreted on demand in response to

10 A.K. Iyer, E.R.B. McCabe / Molecular Genetics and Metabolism 83 (2004) physiological stimuli with secretion regulated at the level of biosynthesis. DAX1 is likely involved in the control of adrenal steroid production, possibly repressing steroidogenic genes when steroid synthesis is not required, keeping steroid production in control when synthesis is required, and also regulating the transcription of speciwc biosynthetic genes depending on the steroid being secreted. DAX1 may be involved in negatively regulating the physiological response to stress. DAX1-deWcient mice have increased corticosterone production and also increased adrenal responsiveness following restraint stress and also after treatment with exogenous ACTH [66]. DAX1 mutations found in AHC patients result in reduced transcriptional silencing activity which could contribute to AHC pathogenesis [20,38,45]. It was previously thought that these mutations caused a direct transcriptional evect with a reduced capacity for DAX1 mutants to interact with its target nuclear receptors and corepressor proteins [45,46]. However, it has been shown that these DAX1 mutants localize predominantly to the cytoplasm presumably due to protein misfolding [27,69]. This observation explains the impaired repressor activity of these mutants, since mutant DAX1 is thus unable to enter the nucleus to interact with nuclear receptors and recruit corepressors for full repressor function. Disruption of normal DAX1 function and pathogenesis of sex determination/diverentiation abnormalities Sex determination Since patients with a duplication of a 160 kb region of Xp21 that contains DAX1 are XY sex reversed females, DAX1 has been thought to function as an anti-testis or pro-ovarian factor during gonadal development [6]. Overexpression of DAX1 in mice in the presence of a weakened SRY allele caused XY sex reversal, suggesting that DAX1 is acting as an anti-testis gene that antagonizes SRY and male development [70]. DAX1 has been shown to repress transcription of MIS by antagonizing transcriptional synergy of SF1 with WT1 or GATA-4 [32,33]. MIS repression prevents regression of the Müllerian ducts. These observations are consistent with the repressor function of DAX1 and also its putative role in DSS. Testicular development and function Studies using the DAX1-deWcient mouse model provide evidence that DAX1 is necessary for proper testicular development and function, suggesting a role beyond that of simply an anti-testis factor. The DAX1-deWcient mice are hypogonadal and infertile, and have testicular and spermatogenic defects with loss of germ cells and degeneration of the seminiferous epithelium, contrary to the hypothalamic or pituitary defect seen in HH patients [65]. Meeks et al. [71] recently showed that DAX1-deWcient XY mice in the presence of the weakened SRY allele develop as phenotypic females, suggesting that DAX1 is required for testis determination, though humans with mutations in DAX1 develop as males. Subsequent studies have shown that DAX1 is involved in testis cord organization during development [72]. Sertoli and Leydig cell-speciwc expression of DAX1 in the DAX1-deWcient mice appear to rescue fertility and sperm production, suggesting a role of DAX1 in spermatogenesis. However, expression of DAX1 in each of these cell types individually was not suycient to rescue the testicular pathology, suggesting that DAX1 function in both Sertoli and Leydig cells, in addition to other somatic cell types is necessary for proper testicular development [73,74]. It has been proposed that embryonic Leydig cell development involves cooperative actions of DAX1 and SF1, as DAX1-deWcient/SF1 haploinsuycient mice show no improvement in testicular pathology [75]. DAX1 is thought to repress aromatase expression, and thus estrogen production in the testis, as estrogen levels were close to normal in the transgenic rescue mice [73,74], making aromatase a physiological target for DAX1. Increased estrogen production in the infertile DAX1-deWcient mice suggests increased ER action, indicating that DAX1 repression of ER action may have functional signiwcance in the testis and adult reproductive function. LRH-1 expression has also been detected in Leydig cells, and aromatase expression was stimulated by LRH-1 in a Leydig cell line [76]. It is possible that DAX1 could repress LRH-1 action in the testis. MIS expression is also present in adult testes [77], and DAX1 may be involved in regulating its expression during puberty and adulthood. The demonstration that expression of DAX1 in rat Sertoli cells is hormonally regulated by FSH and also during postnatal development further suggest a role for DAX1 in spermatogenic cell development [78]. The testicular role of DAX1 in spermatogenesis is consistent with the observation that the infertility in some AHC/HH patients is due to spermatogenic defects in addition defects in gonadotropin secretion. These patients show azoospermia with resistance to gonadotropin treatment [79]. Ovarian development and function The function of DAX1 in ovarian development is not clear. DAX1 expression is detected in the developing ovary at various stages of embryonic development [17], but homozygous DAX1-deWcient female mice, except for a slight ovarian follicular defect, were otherwise normal and fertile, implying that DAX1 is not required for ovarian development [65]. DAX1 may have a role in the adult ovary as a regulator of steroidogenesis during the menstrual cycle. DAX1 expression varies among follicles [17,24], and therefore DAX1 likely plays a role in regulating gene expression in