| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Prince Henry’s Institute of Medical Research, Clayton 3168, Victoria, Australia; and Medical Research Council Molecular Haematology Unit, Institute of Molecular Medicine, John Radcliffe Hospital, Headington, Oxford OX3 9DS, United Kingdom
Correspondence: Address all correspondence and requests for reprints to: Dr. Vincent R. Harley, Prince Henry’s Institute of Medical Research, Monash Medical Center, P.O. Box 5152, Clayton 3168, Australia. E-mail: Vincent.harley{at}med.monash.edu.au
 |
Abstract |
|---|
 Top Abstract I. Introduction and Historical...II. SRY, the Testis-Determining...III. SOX9, the Campomelic...D. SOX9 function in...IV. Regulation of SRY...V. Regulation of SOX9...VI. Initiating Sex...VII. Conclusions and Future...References |
|---|
|
I. Introduction and Historical Review |
|---|
|
TopAbstractI. Introduction and Historical...II. SRY, the Testis-Determining...III. SOX9, the Campomelic...D. SOX9 function in...IV. Regulation of SRY...V. Regulation of SOX9...VI. Initiating Sex...VII. Conclusions and Future...References |
|---|
Among humans, rare individuals arise who carry two X chromosomes but are phenotypically male (XX male) or carry a Y chromosome but are phenotypically female (XY females). Molecular analysis of the genomes of these so-called sex-reversed patients led to the isolation of the sex-determining region on Y gene, SRY (Sry in the mouse). Before this discovery, candidate Y-located genes had been isolated that subsequently failed to meet established criteria (5). SRY met the criteria, as discussed in Section I.A, and its discovery 12 yr ago was envisaged as the entry point through which the molecular genetic basis of mammalian sex determination would be quickly unraveled. However, progress has been slow, hampered by the seemingly intractable nature of the SRY protein. On a positive note, advances have been made on other fronts in the last decade: reverse genetic approaches on human sex-reversal syndromes and mouse gene knockout studies have led to the identification of sex-determining and gonad-formation genes such as those encoding the transcription factors SOX9 [SRY-type high-mobility group (HMG) box 9], DMRT1, GATA4, DAX1, SF1, WT1, LHX9, and cell-signaling molecules AMH, WNT4, FGF9, and DHH. Together with the output of a large number of genes from recent microarray and mouse mutagenesis screens, the genetic and biochemical interactions between these genes and proteins are beginning to reveal the nature of sex determination.
This review focuses on the regulation, function, and molecular interactions of two key mammalian testis-determining factors: SRY and the SRY-related protein, SOX9, allowing us to propose a model of how sex determination is initiated in mammals. Excellent reviews cover other genes in the pathway in more depth (6, 7) or compare pathways between species (8). A recent book arising from a Novartis symposium broadly covers the genetics and biology of sex determination (9).
|
II. SRY, the Testis-Determining Factor |
|---|
|
TopAbstractI. Introduction and Historical...II. SRY, the Testis-Determining...III. SOX9, the Campomelic...D. SOX9 function in...IV. Regulation of SRY...V. Regulation of SOX9...VI. Initiating Sex...VII. Conclusions and Future...References |
|---|
B. Gonadal expression of SRY
Analysis of chimeric mouse gonads constructed by aggregation of male and female embryos showed that almost all Sertoli cell precursors bear a Y chromosome (17), thus suggesting that these cells are the site of Sry expression (18). In the mouse, the adreno-genital primordium arises from the celomic epithelial cells and mesenchymal cells in the mesonephros. The cells of the adreno-primordium are identified as a group of SF1-expressing cells (19). Germ cells migrate to the primordium and cluster with somatic cells to form the genital ridge, first visible as a thickening of the mesonephros around 10 d post coitum (dpc). The timing of Sry expression in the mouse is entirely consistent with a role in sex determination. Sry is detected in the genital ridge at 10.5 dpc, and at 11.5 dpc, but by 12.5 dpc, Sry is detectable only at low levels (14). Sex-specific differences are apparent at about 11.5 dpc when the male gonad takes on a striped appearance, probably due to cells of the supporting cell lineage differentiating into Sertoli cells and aligning into testis cords, the presumptive seminiferous tubules. The ovary forms 2 or 3 d later in female embryos, behaving as a default pathway, initiated in the absence of specification of the male pathway forming granulosa or theca cells. The Sertoli cells and granulosa cells surround the germ cells in their respective organs. Germ cells are not required for testis determination because mice homozygous for the We (white spotting) mutation lack germ cells, yet form somatically normal testes (20). If germ cells migrate to an ectopic tissue, they go into meiotic arrest, which is the fate of germ cells in the ovary (20).
Sex determination can be defined as the proliferation, migration, and differentiation of supporting cells to become Sertoli cells, committing the fate of the gonad to the testis pathway. Sry expression correlates with proliferation of bipotential supporting cells, some of which are the precursors of Sertoli cells in the XY gonad (21). Sry also induces the migration of cells into the XY gonad from the adjacent mesonephros (22). Sry expression progresses from the anterior to the posterior end of the testis over 2 d, with any one cell expressing Sry for no more than a few hours (23). The transient expression of Sry suggests that in mice, whereas it is responsible for determining Sertoli cell fate, Sry is not involved in maintenance of the Sertoli cell phenotype (6). Sertoli cells act as organizers, causing surrounding cells to differentiate into either Leydig or myoid cells. Additionally, Sertoli cells induce mitotic arrest in germ cells, conferring a spermatogenic fate upon them. Few effector molecules involved in these processes are known. A recent advance, however, is the observation that XY mice lacking the secreted fibroblast growth factor Fgf9 develop as phenotypic females (24); these mice show defects in testis cord formation and organization as a consequence of reduced and abnormal Sertoli cells and/or reduced migration and proliferation of interstitial cells.
In humans, as in mice, the onset of SRY mRNA expression defines testis determination. The human gonadal ridge forms around 33 d gestation, and SRY is detected at 41 d in XY embryos. SRY levels peak at 44 d, when the testis cords are first visible (25). Unlike in the mouse, human SRY is not switched off in the gonads and continues into adulthood. By 52 d gestation the germ cells are surrounded by Sertoli cells, which continue to express SRY at a low level (25), suggesting other roles for SRY, possibly in spermatogenesis. SRY protein has been detected in the nuclei of Sertoli cells in the embryonic testis, consistent with a role as a transcription factor. RNA expression data suggest SRY is widely expressed in adult tissues and cell lines as well as in germ cells and preimplantation blastocysts (26, 27). However, these RT-PCR experiments have not been confirmed by a quantitative technique. In summary, human and mouse SRY are expressed just before the differentiation of the bipotential gonad into a testis.
C. The SRY protein
The open reading frame (ORF) of human SRY is contained within a single exon and encodes a 204-amino-acid protein. The protein can be divided into three regions (Fig. 1
). The central 79 amino acids encode the HMG domain, which functions as a DNA-binding and DNA-bending domain and also contains two nuclear localization signals. The human SRY C-terminal domain has no obvious or conserved structure, except for the final seven amino acids, which interact in vitro with a PDZ domain protein (28). The N-terminal region of the protein also has no obvious structure, but phosphorylation of a sequence within this domain enhances DNA binding activity (29). Comparison of the amino acid sequence of the SRY HMG domains from human, mouse, rabbit, wallaby, marsupial mouse, and sheep reveal 70% identity. In contrast, there is no sequence conservation outside the HMG domain. Mouse Sry contains a glutamine-rich region absent from other mammalian species (and some mouse subspecies). The SRY gene has evolved rapidly (30) and the lack of homology outside the HMG domain may reflect the evolution of additional functions for SRY outside of sex determination, e.g., in spermatogenesis or in the brain.
|
Because the in vivo DNA targets of many SOX proteins are known, it is likely that SRY also binds to DNA in vivo to regulate gene expression. It is intriguing that an RNA binding activity for SOX6 and SRY has been reported (34).
The nuclear magnetic resonance structure of the HMG domain of SRY in complex with DNA has been determined (35) (Fig. 2). It consists of three 
-helices forming an L-shape. Helix 1 and 2 run antiparallel and form the short arm. The longer helix 3 runs antiparallel to the amino-terminal strand and forms the long arm at a right angle to the short arm. Conserved aromatic amino acids from each helix pack together to form the hydrophobic core of the protein, stabilizing the structure. The DNA is severely unwound with a widened shallow minor groove, and the bound SRY causes the DNA to bend about 80 degrees, which is in agreement with biochemical studies. A number of amino acids make DNA contacts to specific bases (Asn65, IIe68, Ser88, lle90, Gln117, Lys129, Tyr129), the sugar-phosphate backbone (Arg59, Arg62, Met64, Arg72, Arg75, Lys 92, Trp98, Arg121) or both (Phe67).
|
D. Clinical mutations in SRY
Mutations in SRY are associated with human male-to-female sex reversal in XY females. DNA sequencing has demonstrated that SRY is found in most XX males, and that SRY point mutations or deletions are found in only about 15% of XY females (41). This supports SRY being TDF and also suggests the existence of mutations in sex-determining genes other than SRY. Histologically, the gonads of XY females can be classified into two groups: 1) pure or complete gonadal dysgenesis, in which patients have dysgenic or "streak" ovaries, identical with those in individuals with Turner syndrome (45,X females); and 2) partial gonadal dysgenesis, in which some testicular material is present. Significantly, almost all XY female patients with SRY mutations show complete gonadal dysgenesis, consistent with a critical role for SRY early in testis formation (41). These patients develop as normal females with female internal and external genitalia and, due to the complete gonadal dysgenesis and lack of ovarian function, these XY females present at clinics with primary amenorrhea. In about 50% of cases gonadal tumors (gonadoblastoma or dysgerminoma) are associated with the gonadal dysgenesis, with the result that gonads are routinely removed by surgery.
The positions of SRY point mutations causing XY sex reversal are shown in Fig. 3. Although nonsense mutations are scattered throughout the SRY gene, missense mutations tend to cluster in the central region of the gene, which encodes the HMG domain. This strongly suggests that the DNA binding motif is essential in vivo. However, not all SRY mutations cause complete sex reversal. There are cases of familial SRY mutation in which the fertile father of an XY sex-reversed individual also carries the SRY mutation. In such cases, it is thought that genetic background can compensate for the mutation and that this is not occurring in the affected individual.
|
It is difficult to draw correlations between the activities of the mutated SRY proteins in vitro and the clinical severity of the mutation in the patient. A spectrum of effects are observed in vitro, from a small reduction in DNA binding or bending activity, to a total loss of DNA binding activity. Most have the same outcome in vivo: complete gonadal dysgenesis. The fact that relatively small reductions in DNA binding activity are sufficient supports the idea that SRY is expressed just above a threshold activity level.
Although most mutations are de novo, first arising in the affected individual and causing complete gonadal dysgenesis, there are a few mutations with a familial inheritance (I90M, F109S) that show near-wild-type DNA binding activities. Surprisingly, a number of XY females carry mutations, which do not affect their DNA binding or DNA bending activities in vitro; this suggests that other essential activities of SRY must exist. Also, the results confirm our earlier suggestion that SRY functions at a biochemical threshold and that familial mutations are close to this threshold level and manifest in certain genetic backgrounds (13). We have recently shown that one of these mutants, R133W, is defective in its interaction with importin ß, a protein that SRY requires for nuclear import (46A ). It is quite likely that other interacting proteins remain to be discovered and that some of these SRY mutants will be abnormal in their interaction with them. Alternatively, the in vivo DNA site bound by SRY could vary substantially from the consensus SRY/SOX site used in in vitro binding assays.
Two individuals carry missense mutations outside the HMG domain in the N-terminal and C-terminal domains; both have unusual phenotypes. The 4-yr-old boy who was affected by the N-terminal domain mutation S18N had partial gonadal dysgenesis, had retained both Wolffian and Mullerian ducts, and had ambiguous external genitalia. However, both his father and adult brother carry the mutation and appear to be normal males (47). A patient carrying the K43 stop mutation failed to develop a uterus (48). In vitro DNA binding and DNA bending activity of the S18N mutant was normal (46). In each case, SRY function is affected in a unusual manner, suggesting that the domains flanking the HMG domain influence SRY function in vivo.
F. SRY: activator or repressor?
Evidence that human SRY is a transcriptional activator or repressor is equivocal with reporter gene studies in transfected cell cultures showing activation or repression. The main limitations of these studies has been the lack of a relevant DNA target and/or appropriate cell lines. One study suggested the AMH gene (also known as MIS) as a potential target of SRY (49). SRY was transfected into a rat gonadal ridge cell line with a reporter construct containing a fragment of the AMH promoter. Activation of the reporter construct was approximately 15-fold higher than basal levels, and the response was reduced when a variant of SRY with a 50-fold reduced DNA binding affinity was used. However, these effects were not due to a direct interaction between SRY and the AMH promoter because the removal of the single SRY binding site in the AMH promoter did not affect the levels of SRY-mediated activation of AMH. In a second study, SRY was shown to activate the fra-1 promoter (fos-related antigen) in Chinese hamster ovary cells about 20-fold higher than basal levels (50), suggesting that SRY has the capacity to activate gene expression in cells, although a SOX protein other than SRY is likely to be involved in fra-1 regulation because fra-1 is not expressed in the genital ridge (51).
On the other hand, it has been suggested that SRY represses the activity of a gene that would otherwise repress genes required for the development of the male sex (52). This model could explain why XX males with SRY rarely have ambiguous genitalia, whereas XX males lacking SRY commonly have ambiguous genitalia or are true hermaphrodites. It is argued that there would be varying degrees of repression of male sexual development genes in the absence of SRY, depending on which of these unidentified genes contained mutations. In support of this theory, there has been a reporter gene study in transfected cells demonstrating that SRY can repress transcription. A construct containing multiple SRY binding sites upstream of a reporter gene (thymidine kinase promoter) was transfected into monkey epithelial (COS7) cells with an SRY expression vector (29). The 2-fold level of repression was increased with phosphorylation, which enhances DNA binding activity, and no repression was detected when the phosphorylated residue of SRY was mutated. However, in a similar study in the same cell line but using a different promoter (adenovirus E1b promoter) a 3-fold activation was observed (P. Tang and V. R. Harley, upublished observation). Thus, it is becoming increasingly apparent that until the in vivo target of SRY has been identified, we will not know whether SRY is an activator or repressor—or indeed both!
In contrast to studies of human SRY, studies on mouse Sry show that it strongly activates transcription through its C-terminal glutamine-rich region (53), which is completely absent in SRY from other mammals and certain mouse subspecies (Fig. 1
). It is hard to envisage a different mechanism operating through the poorly conserved non-HMG domain regions in different mammalian species. More likely, the SRY HMG domain itself carries the necessary information for transcription through structure- and sequence-specific DNA recognition together with coactivator proteins to establish the correct architecture in chromatin, analogous to the HMG domain protein LEF-1 (54). Non-HMG domain regions may subserve the HMG domain function.
G. The SRY HMG box: all you need?
Arguing against the suggestion that the HMG box is all you need, the C-terminal domain of mouse Sry can act as a transcription activator in cultured cells (53). A translation stop engineered just before the glutamine region prevented the Sry transgene from causing female-to-male sex reversal in XX mice (55), but a negative result in transgenic experiments must be viewed with caution. In humans, the SRY K43 stop mutation in a human XY female might represent a similar situation, where the C-terminal domain had been deleted, resulting in loss of function.
In favor of the suggestion that the HMG box is all you need, when the human and mouse ORFs are compared, there is no homology outside the HMG box (Fig. 1). Of the 32 known SRY missense mutations, only two are located outside the HMG box; more would be expected in these domains, which represent 120 amino acids of the 204-amino-acid protein. Recent experiments indicate that the human SRY ORF can cause sex reversal of XX transgenic mice (56). This suggests that there is no requirement for the C-terminal glutamine-rich domain of mouse Sry for sex determination, although these regions may play a role in processes other than sex determination, e.g., spermatogenesis or regulating male behavior. These studies suggest that "a" C-terminal domain is required for Sry/SRY function, if only for protein stability. Supporting this suggestion, the HMG domain is never located at the C-terminal end of an SRY protein among mammalian species.
Clearly, the HMG domain is essential, but can any HMG domain do the job? In vitro studies suggested that SRY/SOX proteins differ in their intrinsic DNA sequence specificity (33), probably reflecting a difference in DNA binding specificity in vivo. Although SRY and SOX9 are capable of binding the same target sequences in vitro, their affinities are different (33). In vivo support for this observation came from work by Eicher and colleagues, which showed that only when overexpressed could SOX9 or SOX3 HMG domains substitute for that of SRY to cause sex reversal in XX transgenic mice (57). It was noted that greater amounts of transgene were required, presumably because the protein dose needed to be greater to override the weaker affinities of the SOX3 and SOX9 HMG domains for the target of SRY, currently unknown. However, swapping does not always work. When the HMG domain of SOX9 was swapped with that of SOX1, it greatly affected the transactivation potential of SOX9 (58), confirming that SOX9 HMG domain appears to possess some DNA target specificity not present in other SOX proteins.
H. In vivo targets of SRY
The essential nature of the SRY HMG box and the numerous examples of SOX proteins binding in vivo to DNA targets suggest a likely role for SRY as a DNA binding protein to a promoter/enhancer/silencer element within a downstream gene in Sertoli cell precursors in vivo. The early report of SRY binding to the AMH promoter is unlikely given that their expression profiles in the genital ridge are nonoverlapping (59). However, a number of genes are expressed soon after the expression of SRY, e.g., SOX9, FGF9, DHH, and VNN1. Furthermore, a number of recent microarray screens hold much promise in revealing more candidate genes (60, 61, 62). Clearly, a direct demonstration of SRY binding to a candidate gene DNA in vivo will rely upon difficult procedures such as chromatin immunoprecipitation. Only then will the issue of whether SRY acts as a repressor or activator be more easily resolved.
It is likely that whatever SRY directly acts upon, the ultimate function of SRY is the up-regulation of SOX9. Shortly after the induction of Sry, Sox9 becomes activated, moving from the cytoplasm to the nucleus, and up-regulated in the male gonad (63). To address the precise function of SOX9 in the gonad, transgenic mice ectopically expressing Sox9 driven by the Wtl promoter were produced (64). XX transgenic mice developed testes with apparently normal Sertoli cells and Leydig cells. This suggests that Sox9 can replace Sry and implies that Sry’s only function is to up-regulate Sox9. Up-regulation of Sox9 is a key phenomena in all vertebrates, regardless of the switch mechanism controlling sex determination, i.e., SRY in mammals [except for the mole vole (65)], ZW chromosome gene/s in birds (66), and temperature sensitivity of egg incubation in turtles and crocodiles (67, 68).
Support for SRY being a repressor of a repressor comes from the creation of the Odsex (ocular degeneration with sex reversal) mouse. Sex reversal is caused by the insertion site of a transgene 1–2 megabases (Mb) upstream of Sox9, causing a 150-kb deletion. XX mice transgenic for Odsex develop as males with levels of Sox9 transcripts identical with those found in wild-type XY males at 11.5 and 14.5 dpc. The insertion/deletion did not affect the expression of Sox9 in cartilage or bone, which suggests that the transgene insertion deleted a regulatory element that is only required for the repression of Sox9 in the gonads. It was proposed that SRY normally disrupts the binding or function of this repressor (69). A candidate for the role of repressor of SOX9 is DAX1.
I. SRY interacting proteins
Two proteins implicated in nuclear import function have been identified as interacting with SRY. Nuclear localization signals (NLSs) are responsible for targeting proteins for recognition by cytoplasmic proteins, which facilitate translocation of proteins generally larger than 45 kDa through the nuclear pore complex. SRY protein appears to be localized in the nucleus of pre-Sertoli cells consistent with a transcriptional function (25). SRY, although only 27 kDa, is exclusively nuclear and therefore actively transported. SRY contains a bipartite NLS (61 KRpmnafivwsRdqRRK) at the N-terminal end and an simian virus 40-type NLS (126KyRpRRKaK) at the C-terminal end of the HMG box (70, 71) (Fig. 4). The N-terminal NLS forms part of a calmodulin (CaM)-binding domain, and a role for CaM in nuclear import of other proteins (72) suggested that nuclear import by this NLS is mediated by CaM. Recent studies using SOX9 and CaM antagonists demonstrate this to be the case (72A ). Binding studies of NLSs with importins reveal that the C-terminal NLS of SRY interacts with importin ß. The SRY mutations in the C-terminal NLS, R133W, showed reduced nuclear import due to reduced binding to importin ß (46A ) (Fig. 5). This provides a molecular explanation for sex reversal in this variant, which also showed wild-type DNA binding and DNA bending activity.
|
|
|
III. SOX9, the Campomelic Dysplasia/Autosomal Sex-Reversal Gene |
|---|
|
TopAbstractI. Introduction and Historical...II. SRY, the Testis-Determining...III. SOX9, the Campomelic...D. SOX9 function in...IV. Regulation of SRY...V. Regulation of SOX9...VI. Initiating Sex...VII. Conclusions and Future...References |
|---|
B. Clinical mutations in SOX9
SOX9 mutations occur throughout the ORF, unlike those in SRY that generally cluster within the HMG box. Mutations in the SOX9 ORF detected outside the HMG box are nonsense and frameshift mutations that disrupt the C-terminal domain of the protein and alter the ability of SOX9 to efficiently activate transcription of target genes (77, 78, 79). Mutations in SOX9 include splice acceptor/donor changes and missense, nonsense, translocation, and frameshift mutations (Figs. 6 and 7) (75, 76, 79, 80, 81, 82). There is no correlation between severity of the disease or associated sex reversal and mutation type with many of the same mutations causing varying degrees of gonadal dysgenesis and bone malformations (81). For example, a mutation that causes a frameshift and subsequent premature stop at codon 254 was identified in three siblings. One sibling was a 46,XY true hermaphrodite with ovotestes and ambiguous genitalia; another was 46,XY with bilateral ovarian tissue and normal female genitalia; and the third sibling was a 46,XX with normal ovaries and female genitalia (83). In another case, two unrelated individuals carried a mutation that resulted in a premature stop (Y44OX) (Fig. 7). One of these individuals exhibited testicular dysgenesis with sex reversal, whereas the other had unambiguous male genitalia (81, 82). In addition, a CD patient with a nonsense mutation that removes 80% of the protein was alive at 12 yr of age, whereas other nonsense and frameshift mutations that leave more of the SOX9 protein intact were found in CD patients who died in the neonatal period (81) (Fig. 7). Hence, both sex reversal and disease severity are due to the degree of expressivity of the SOX9 mutation, as well as differences in genetic background, rather than the specific type of mutation. Although most mutations occur in one allele of SOX9, compound heterozygosity has also been detected in one patient with CD with a different mutation in each allele of SOX9 (76). Recently, an SRY-negative female-to-male sex reversal patient with a duplication of chromosome band 17q23–24 including the SOX9 gene was isolated (84). Thus, an extra dose of SOX9 may be sufficient to initiate testis differentiation in the absence of SRY. No cases of sex reversal without CD, due to mutations in SOX9, have been detected (85).
|
|
C. SOX9 protein structure and activity
The human SOX9 gene encodes a protein of 509 amino acids consisting of the HMG box, which shares 70% amino acid homology to the HMG box of SRY. In addition, the SOX9 protein contains additional protein domains, including two transcriptional activation domains, downstream of the HMG box. The SOX9 protein, unlike SRY, is very highly conserved through vertebrate evolution (Fig. 8).
|
-crystallin minimal enhancer, DC5, with SOX9, which activates the type II collagen minimal enhancer, Col2a1. SOX proteins have been classified according to the protein sequence homology of their HMG domains (90). The replacement of the HMG box of SOX1 with that of SOX9 did not alter transcriptional activity of the DC5 enhancer in lens cells (58). However, the reciprocal experiment resulted in a reduction in the ability of SOX9 protein (containing the SOX1 HMG box) to activate the type II collagen minimal enhancer, Col2a1 (58). Hence, the SOX9 HMG box confers some degree of specificity that may relate to interactions with specific partner proteins. The presence of promoter and/or enhancer sites with varying SOX9 DNA binding affinities may also contribute to differential regulation of target genes. All known missense mutations in SOX9 occur in the HMG box (75, 76, 79, 80, 81, 82, 83). The majority of SOX9 missense mutations show altered DNA binding compared with wild type, although there are a few exceptions (79, 81). The A119V mutation showed near-wild-type DNA binding and bending; and another mutation, P17OR, while showing near-wild-type DNA binding and bending, had altered DNA binding specificity (79). This suggests that other essential biochemical activities, in addition to DNA binding and bending, are impaired in these SOX9 mutations. In addition, no missense mutations identified have shown normal binding with altered DNA bending.
2. SOX9 transcriptional activity.
Several studies have shown that the SOX9 protein acts as a potent transcriptional activator both in vitro and in vivo. The SOX9 protein is known to activate transcription of the type II collagen gene and anti-Mullerian hormone (91). The removal of the last 108 amino acids from the C terminus of SOX9, which is rich in proline, glutamine, and serine residues, abolished transcriptional activity (77, 78, 79). The region adjacent to the proline-glutamine-serine (PQS) domain, which consists entirely of prolines, glutamines, and alanine residues (PQA), was also found to be required for maximal transactivation; unlike the PQS domain, the PQA domain varies in length between different species and is completely absent in rainbow trout Sox9 (92) (Fig. 8). Species that lack the PQA domain would be predicted to have a weaker transactivation potential. Interestingly, this domain is only present in vertebrates and may relate in some way to organisms with an SRY sex-determining mechanism.
Some SOX9 mutations that encode proteins with C-terminal truncations showed decreased transactivation capabilities compared with wild-type SOX9. Two truncation mutations, Y440X and the 507fs frameshift mutation, while leaving the majority of the transactivation domain intact, reduced transactivation activity (81). The 507fs mutation encodes a mutant SOX9 protein that has an extended ORF creating an unstable mRNA and/or protein product (81). The transactivation potential of other SOX9 mutations that truncate the C-terminal domain or those that cause an alteration in DNA binding affinity have not been investigated but are predicted to have reduced transcriptional activity compared with wild-type SOX9. Hence, the biochemical defect in these mutants is likely to be an inability to adequately activate transcription.
Apart from sharing a high level of homology between orthologs, the C-terminal domain of SOX9 also shares homology with other Group E SOX proteins, in particular, the transactivation domain of SOX10 (Fig. 9). The C-terminal 23 amino acids are highly conserved in Sox8, SOX9, and SOX10 although the PQA domain is not present in either Sox8 or SOX10 (93, 94) (Fig. 9). It is not clear how the C-terminal domain of SOX9 may confer transcriptional activity; it is likely that both the PQA and PQS domains interact with as-yet-uncharacterized partner proteins to activate transcription. These proteins may include tissue-specific transcription factors and/or components of the basal transcription machinery.
|
). These phosphorylation sites, as in SRY, may be involved in regulating SOX9 DNA binding, nuclear import, and transcriptional activity.
SOX9 was shown to interact specifically with the catalytic subunit of PKA (95). This interaction between SOX9 and PKA suggests that phosphorylation directly regulates SOX9 activity. SOX9 contains two putative PKA sites, flanking the HMG box, which are completely conserved in SOX9 (Fig. 4). PKA phosphorylation of SOX9 increased DNA binding activity and enhanced the activity of SOX9-dependent Col2a1 enhancer activity in RCS cells (95). In addition, PKA-phosphorylated Sox9 was detected in the prehypertrophic zone of the growth plate in chondrocytes (95). This suggests that phosphorylation of SOX9 by PKA in prehypertrophic chondrocytes is regulated by parathyroid hormone-related peptide signaling, which is known to increase cAMP levels. Bone morphogenetic protein-2 (BMP-2), which plays an important role in connective tissue morphogenesis, was also previously shown to induce chondrocyte differentiation and activate PKA in cultured chondrocytes (96). BMP-2 increased the expression levels of Sox9 and a number of other chondrocyte markers, including aggregan and cartilage oligomeric matrix protein, in C3H10T1/2, a mesenchymal progenitor cell line. Hence, both parathyroid hormone-related peptide and BMP-2 signaling may cause phosphorylation of SOX9 by PKA in chondrocyte cells (97). Fibroblast growth factors (FGFs) enhance Sox9 expression in C3H10T1/2 cells (98). FGF increased the activity of Sox9-dependent Col2a1 activity, consistent with the increase in Sox9 expression in primary chondrocyte cells (98). This increase is mediated by the MAPK [MAPK kinase (MEK)-MAPK] pathway because the addition of specific inhibitors of MEK-MAPK to primary chondrocytes inhibited FGF-induced expression of Sox9 (98).
The regulation of SOX9 activity in chondrocytes is complex, with phosphorylation playing a major role. A number of signaling pathways act during chondrogenesis to regulate the expression, DNA binding, and transactivation activities of Sox9. Although these studies demonstrate that cAMP-mediated pathways regulate the activity of Sox9 in chondrocyte cells, the identification of several casein kinase II and protein kinase C sites in SOX9 suggest the existence of other phosphorylation pathways that also regulate the activity of SOX9 in vivo. In particular, the role of phosphorylation in the Sertoli cell lineage, including the regulation of anti-Mullerian hormone (AMH) expression, has not been investigated.
4. Nuclear transport of SOX9.
In order for transcription factors to act, they must be transported to the nuclei of cells to bind to and activate (or repress) transcription of target genes. In the developing mouse gonad, in pre-Sertoli cells, Sox9 transcripts are very low and perinuclear. Upon expression of Sry at 10 dpc, the level of Sox9 is unregulated and the protein is localized in the nuclei of cells (63). The temporal and spatial expression of SOX9 and SRY is critical for proper gonadal development.
The NLSs in the C and N termini of the HMG box of SRY are highly conserved throughout SOX proteins, and thus the mechanism of nuclear import is likely to be conserved (Fig. 4). The SOX9 HMG box, similar to SRY, was found to interact directly with the transport receptor importin ß and, although not tested individually, importin ß binding is likely to occur through the carboxy-terminal nuclear localization signal (C-NLS) (99). Unlike SRY, no clinical mutations have been identified in the NLSs of the SOX9 HMG box; however, one mutant (A158T), which lies close to the C-NLS, was found to have altered nuclear accumulation and may disrupt the function of the C-NLS (99). The fact that A158T bound with wild-type affinity to importin ß suggests that, whereas this recognition step is normal, other components of the importin ß-mediated pathway could be affected. The demonstration that a mutation that lies outside the NLS regions affects nuclear localization raises the possibility that a large number of clinical mutations in SOX9 and SRY could affect nuclear import, in addition to or distinct from DNA binding and bending.
Other mechanisms of nuclear import have been described, including a well characterized pathway involving CaM (72). CaM, an intracellular calcium receptor, plays roles in numerous processes in the cell including DNA replication, mitosis, DNA repair, and other nuclear functions such as the regulation of the condensation/relaxation of chromatin (100, 101) Like the SRY protein, SOX9 also interacts with CaM, in a calcium-dependent manner, through a basic region that acts as an NLS and is highly conserved among SOX proteins (102) (Fig. 4). A reduction in SOX9 in vitro CaM binding was observed on native gels upon addition of CaM antagonists (72A ). In addition, significant reductions in transcriptional activity and nuclear accumulation of SOX9 were observed upon treatment with specific CaM antagonists (72A ). This suggests that CaM is involved in the nuclear transport of SOX9 (and all SOX proteins) in a process likely to involve a direct interaction (72A ). It is likely that a highly specific mechanism acts to regulate SOX9 nuclear transport because the subcellular localization of SOX9 changes from cytoplasmic to nuclear upon onset of SRY expression in the developing mouse XY gonad (63). Components of both the importin/Ran-GTP and CaM-dependent pathways may act to regulate nuclear accumulation of SOX9 in Sertoli cells to allow a sufficient amount of SOX9 protein to activate the AMH gene (Fig. 5).
Although a number of groups have investigated nuclear import of SOX proteins, little is known about SOX nuclear export. Apart from the NLSs with the HMG box, SOX9 (as well as a number of other SOX proteins) also contains leucine-rich motifs that are putative nuclear export signals (103). In relation to nuclear import, clinical mutations located with the HMG box, in particular those located within the NLS and/or CaM binding domain, may reduce the efficiency of SOX9 protein being transported to the nucleus, which is critical for the proper regulation of the gonadal and chondrocyte pathways.
|
D. SOX9 function in gonad and bone formation |
|---|
|
TopAbstractI. Introduction and Historical...II. SRY, the Testis-Determining...III. SOX9, the Campomelic...D. SOX9 function in...IV. Regulation of SRY...V. Regulation of SOX9...VI. Initiating Sex...VII. Conclusions and Future...References |
|---|
One gene in the Sertoli cell lineage, the expression of which SOX9 was shown to directly regulate, is AMH. AMH is an essential component of the male sexual differentiation pathway, secreted by Sertoli cells, causing the regression of the female (Mullerian) reproductive tract (107, 108). Deoxyribonuclease 1 footprint analysis showed that SOX9 binds to a SOX-like site in the AMH proximal promoter, adjacent to the MISRE1 site and, together with SF1 (steroidogenic factor 1), WT1 (Wilms tumor gene 1), and GATA4, directly activates expression of AMH (108, 109, 110). Heterozygote mutations in the SF1 gene have been associated with complete XY sex reversal and adrenal failure in humans, suggesting that SF1 regulates the regression of the Mullerian ducts (111, 112). Several key experiments have shown that regulation of the AMH gene requires cooperative interaction between SOX9 and SF1. SF1 binds to the MISRE1 site, and reporter gene assay experiments have shown a cooperative increase in transcriptional activity with the cotransfection of SF1 and SOX9 compared with SF1 alone (108). A comprehensive study of Amh promoter mutations in transgenic mice (109) showed that mice with targeted mutations in the SOX site within the Amh promoter resulted in complete retention of Mullerian duct-derived organs with a complete absence of Amh transcript (109). This indicates that Sox9 is essential for Amh transcription. The complete absence of Amh transcripts in XY mice homozygous for the mutant SOX binding sites also suggests that Sox9 is required to initiate Amh transcription (109). In the case of SF1, mutations in the MISRE1 site caused partial decrease in Amh levels and partial regression of the Mullerian ducts, indicating that SF1 is required for the up-regulation of Amh transcription (109). GATA4, which is abundantly expressed in Sertoli cells, interacts through its zinc finger domain with SF1 to synergistically activate AMH transcription (110). Similarly WT1, also required for male gonadal development, interacts with SF1 to up-regulate expression of AMH transcription (113).
Recently, SOX9 has been shown to interact with HSP70 (heat shock protein 70) in testicular and chondrocyte cell lines (114). The interaction involves the C-terminal domain of HSP70 with a 100-amino-acid region of SOX9 between the HMG box and the PQA domain, hitherto of unknown function. HSP70 plays numerous roles in transcriptional regulation and interacts strongly with WT1 in vivo (115). Although binding sites to both SOX9 and SF1 are conserved within the AMH promoter, WT1 has no conserved binding site and it interacts with SF1 only weakly. Considering that WT1 strongly interacts with HSP70 in vivo (115), it is possible to speculate that WT1 binding at the AMH promoter is stabilized by the formation of a SOX9-HSP70-WT1 protein complex.
Clearly, numerous protein-protein interactions exist at the Amh promoter. The ability of the SOX9 HMG box to bend DNA may bring SF1 and GATA4 in closer proximity to each other and along with WT1 and HSP70 form a tightly associated protein complex that activates transcription of the AMH gene (Fig. 10).
|
At present, it is not known what factor(s) initiate the Sox9 promoter in Sertoli cells. The up-regulation of Sox9 in Sertoli cells upon expression of Sry is consistent, however, with a role for Sry in the activation of Sox9 expression (63, 106). A minimal interval located between -193 and -73 bp from the transcription start site of Sox9 was sufficient to direct maximal promoter activity in male and female gonadal somatic cells (116). This region contains putative SOX-like binding sites, which are conserved in human, mouse, and chicken (116). Luciferase activity from this minimal region was significantly higher in testicular cell lines than in other cell lines tested with the difference abolished when this region was deleted (116). This suggests that the -193 to -73-bp region contributes, but is not sufficient, to direct testis-specific Sox9 expression. To further delineate the regulatory elements in the SOX9 promoter, mice transgenic for large fragments of human SOX9 in yeast artificial chromosomes were generated (117). Several regulatory elements required for SOX9 expression in chondrocyte development were identified in a 350-kb region upstream of SOX9 (117). However, testis-specific SOX9 expression was not detected in any of the transgenic mice tested, indicating that additional testis-specific regulatory elements are located outside this region (117). A translocation breakpoint identified in a CD patient, which maps more than 950 kb from SOX9, supports this observation (87). A 150-kb gonad-specific regulatory element, 1 Mb upstream of Sox9, which mediates the repression of Sox9 expression in XX fetal gonads, was mapped (69). Transgenic mice with an insertional mutation, odsex (ods), which lack this 150-kb region developed as sterile XX males lacking Sry and showed no skeletal defects (69). Hence, regulation of SOX9 expression in the testis is likely to be complex and involve proteins bound at distant regulatory elements at the SOX9 promoter and/or enhancer.
2. Role of SOX9 in chondrogenesis.
Sox9 is expressed at high levels at all sites where cartilage is being laid down with expression most abundant in mesenchymal condensation just before overt chondrocyte differentiation (78, 89). Sox9 is expressed in the first and second branchial arches, in the sclerotomes, and the lateral plate mesoderm that gives rise to the appendicular skeleton (78, 89). The expression of Sox9 and type II collagen (Col2a1), an early marker of chondrocyte differentiation, coincide and peak at 11.5–14.5 dpc (78, 89). The severe skeletal malformations seen in patients with CD, which include bowing of the femora and tibiae, hypoplastic scapulae, pelvic malformations, and bilateral clubfeet, and the expression pattern of Sox9 in cartilage are consistent with a role of Sox9 in chondrogenic development (75, 76).
In an effort to identify transcription factors that control chondrocyte differentiation, a minimal sequence was identified in intron 1 of the Col2a1 gene, which contains SOX-like binding sites and was able to direct chondrocyte-specific expression in transgenic mice (118). Sox9 binds to this enhancer sequence and activates transcription of a reporter construct in COS7 cells (78, 89). Other SOX proteins tested, however, failed to activate transcription, suggesting that type II collagen is a direct target for Sox9 (78, 89). In addition, mutation of these sequences abolished chondrocyte-specific expression of a Col2a1-driven reporter gene in transgenic mice (88).
A number of in vivo experiments performed illustrate the importance of SOX9 in chrondrogenesis. In transgenic mice, ectopic expression of Sox9 resulted in ectopic expression of endogenous Col2a1 in a number of tissues (88). However, in the genital ridge of transgenic mice, Col2a1 is not expressed, although there is a high level of endogenous Sox9 present in this tissue. Therefore, Sox9 alone is not sufficient to direct Col2a1 expression in vivo and suggests that Sox9 interacts with other chondrocyte-specific enhancer binding proteins. SOX9 is known to form a complex with Sox6 and L-Sox5, which also binds to the Col2a1 enhancer (119). The lack of expression of Col2a1 in nonchondrogenic tissues in transgenic mice that express Sox9 could be due to the lack of or insufficient expression of Sox6, L-Sox5, or additional chondrocyte-specific enhancer-binding proteins. Chimeric mice were produced that are mosaic for a targeted deletion of Sox9, replaced with the lacz gene (120). At 15.5 dpc when chondrogenesis is occurring, Sox9-/- cells express Col2al, Col9a2, Col11a2, and aggrecan. These Sox9-/- cells are interspersed with wild-type cells (120). Sox9-/- embryonic stem cells did not differentiate into chondrocytes as per wild-type cells. In contrast, Sox9-/- cells were not localized to the cartilage and lacked both morphological characteristics of chondrocytes and expression of early chondrocyte markers (120). Earlier in development, at 9.5 dpc when mesenchymal condensation is occurring, neither were Sox9-/- cells interspersed with wild-type cells nor did the cells express Col2a1. This indicates that Sox9 is an early marker of chondrocyte development and is required for the activation of specific collagen genes.
Mice heterozygous null for Col2a1 show only mild skeletal defects, suggesting that the more severe skeletal defects observed in CD are likely to be due to inadequate activation by Sox9 of additional genes (other than Col2a1) that are required for chondrogenesis such as Col11a2, Col9a2, and aggrecan. This is supported by the observation that Sox9 can bind to the promoters and activate transcription of the Col11a2, cartilage-derived retinoic acid-sensitive (CD-RAP), and aggrecan (agg) genes and is therefore likely to be involved in regulating the expression of these genes (121, 122).
|
IV. Regulation of SRY Expression in the Developing Gonad |
|---|
|
TopAbstractI. Introduction and Historical...II. SRY, the Testis-Determining...III. SOX9, the Campomelic...D. SOX9 function in...IV. Regulation of SRY...V. Regulation of SOX9...VI. Initiating Sex...VII. Conclusions and Future...References |
|---|
When it was first demonstrated that an Sry transgene was capable of inducing male gonad development in XX mice, a 14-kb genomic fragment containing the Sry coding region and 8 kb of 5'-sequence was used (15). Expression of Sry from the original 14-kb construct is more widespread and persists for longer than the endogenous Sry gene (19, 57, 125), indicating that the transgene is subject to position effects or lacks cis-acting repressors that temporally and spatially limit its expression. Either way, it would seem that tight positional regulation of Sry expression is not necessary for sex determination in the mouse. Surprisingly, constructs with progressive deletions of 5'-sequences to within 57 bp of the transcription start site were still capable of inducing testis development (125).
Conventionally, promoter elements are referred to in terms of their location relative to the site of transcript initiation; however, the initiation codon has been used as a reference point for discussing the location of SRY elements because multiple initiation sites have been detected in humans and mice. In characterizing the human transcript, one study identified two SRY transcript start sites at -78 bp and -136 bp (126), and other studies have identified a major initiation site at -91 (26, 127) and a minor initiation site at least 410 bp further upstream (26) (Fig. 11). In the mouse, three major and one minor transcription initiation site have been identified between -269 bp and -256 bp (59) (Fig. 11).
|
