PHYSIOLOGICAL REVIEWS   Vol. 78 No. 1 January 1998, pp. 1-33
Copyright ©1998 by the American Physiological Society

Regulation of Sexual Dimorphism in Mammals

CHRISTOPHER M. HAQQ AND PATRICIA K. DONAHOE

Pediatric Surgical Research Laboratories, Massachusetts General Hospital, and Department of Genetics and Surgery, Harvard Medical School, Boston, Massachusetts

I. INTRODUCTION
II. SEX-DETERMINING ROLE OF HUMAN Y CHROMOSOME
III. IDENTIFICATION OF THE SEX-DETERMINING GENE SRY
IV. MUTATIONS IN SRY EXPLAIN A SUBSET OF PATIENTS WITH 46, XY PURE GONADAL DYSGENESIS
V. TRANSLOCATION OF SRY TO THE X CHROMOSOME OR AUTOSOMES EXPLAINS A SUBSET OF PATIENTS WITH 46, XX TRUE HERMAPHRODITISM
VI. BIOCHEMICAL ACTIVITIES OF SRY: HOW DOES SRY DETERMINE SEX?
    A. SOX Genes and Functions
    B. Factors Involved in MIS Regulation
    C. Mutated SRY From Sex-Reversed Patients Fails to Activate MIS
    D. Superactivator SRY Isoleucine 68 Phenylalanine
    E. Functional Mapping of the MIS Promoter Response to SRY
    F. Additional Factors Involved in Regulation of the MIS Promoter
VII. SRY TRANSCRIPTION UNIT: WHAT REGULATES SRY EXPRESSION?
VIII. AUTOSOMAL AND X-LINKED GENES INVOLVED IN SEX DETERMINATION AND DIFFERENTIATION
    A. MIS and Sex Differentiation
    B. Mutation of the MIS Type II Receptor and MIS Receptor Signal Transduction
    C. Role of Androgens and Sex Steroids in Sex Differentiation
    D. Defects in Steroid Biosynthesis Lead to an Intersex Phenotype and Congenital Adrenal Hyperplasia
    E. Defects in the Androgen Receptor Lead to Testicular Feminization
    F. Defects in Steroid 5-Reductase Result in an Intersex Phenotype
    G. Additional Autosomal Genes Involved in Sex Differentiation
IX. TISSUE-SPECIFIC MARKERS OF TESTICULAR DIFFERENTIATION
X. CONCLUSIONS
REFERENCES

	ABSTRACT

Top References

Haqq, Christopher M., and Patricia K. Donahoe. Regulation of Sexual Dimorphism in Mammals. Physiol. Rev. 78: 1-33, 1998.  Sexual dimorphism in humans has been the subject of wonder for centuries. In 355 BC, Aristotle postulated that sexual dimorphism arose from differences in the heat of semen at the time of copulation. In his scheme, hot semen generated males, whereas cold semen made females (Jacquart, D., and C. Thomasset. Sexuality and Medicine in the Middle Ages, 1988). In medieval times, there was great controversy about the existence of a female pope, who may have in fact had an intersex phenotype (New, M. I., and E. S. Kitzinger. J. Clin. Endocrinol. Metab. 76: 3-13, 1993.). Recent years have seen a resurgence of interest in mechanisms controlling sexual differentiation in mammals. Sex differentiation relies on establishment of chromosomal sex at fertilization, followed by the differentiation of gonads, and ultimately the establishment of phenotypic sex in its final form at puberty. Each event in sex determination depends on the preceding event, and normally, chromosomal, gonadal, and somatic sex all agree. There are, however, instances where chromosomal, gonadal, or somatic sex do not agree, and sexual differentiation is ambiguous, with male and female characteristics combined in a single individual. In humans, well-characterized patients are 46, XY women who have the syndrome of pure gonadal dysgenesis, and a subset of true hermaphrodites are phenotypic men with a 46, XX karyotype. Analysis of such individuals has permitted identification of some of the molecules involved in sex determination, including SRY (sex-determining region Y gene), which is a Y chromosomal gene fulfilling the genetic and conceptual requirements of a testis-determining factor. The purpose of this review is to summarize the molecular basis for syndromes of sexual ambiguity seen in human patients and to identify areas where further research is needed. Understanding how sex-specific gene activity is orchestrated may provide insight into the molecular basis of other cell fate decisions during development which, in turn, may lead to an understanding of aberrant cell fate decisions made in patients with birth defects and during neoplastic change.

	I. INTRODUCTION

Top Next References

Fetal sexual differentiation relies on translation of chromosomal sex established at fertilization into gonadal sex and somatic sex as development proceeds. In cases where chromosomal, gonadal, and somatic sex are incongruent in human infants and children, rapid establishment of the diagnosis and implementation of medical and surgical management is of paramount importance (reviewed by Refs. 97 and 115), since gender identity is so important to psychological well-being throughout life. In this review, we briefly discuss the history of research on sex determination, since this has been the topic of recent reviews (e.g., Refs. 47, 148, 159), and then focus on more recent developments in the field. Rather than being comprehensive, we have highlighted work that may define molecules important in the processes of sex determination and sex differentiation.

Chromsomal determination of sex was discovered and influenced by work in Drosophila melanogaster at the turn of the century. T. H. Morgan showed that XX fruit flies develop as females, whereas XY flies develop as males. These studies showed that the critical signal for sex determination in flies was the ratio of X chromosomes to autosomes, since XX flies are female and XO flies are sterile males. This was adopted as a model for sex determination in mammals in the 1920s when T. S. Painter showed that female mammals have XX constitution whereas males have X and Y chromosomes (319).

This early model was overturned in the late 1950s when it was found that the human Y chromosome specifies development of the testis (reviewed in Ref. 279). With the development of specific staining techniques for chromosomes, Ford et al. (121) and Jacobs and Strong (207) were able to demonstrate that the Y chromosome was associated with testicular differentiation in human intersex gonads. These workers determined that XO humans, unlike fruit flies, are female. Also, no matter what the dosage of the X chromosome, the presence of a Y chromosome was sufficient to determine maleness since 47 XXY, 48 XXYY, 48 XXXY, and 49 XXXXY individuals underwent testicular differentiation.

It is now recognized that differences in Drosophila and human sex determination were first glimpses at a wide array of sex determination mechanisms that have evolved independently in different species (for reviews, see Refs. 14, 77, 151, 213). For example, many animals use environmental cues such as crowding, pH, and, as proposed by Aristotle, temperature. Birds use a chromosomal sex determination system where steroids play an important role and can reverse chromosomal sex determination. Unlike humans and flies, birds are heterogametic for the female sex rather than the male (ZW female vs. ZZ male), whereas fish such as the Japanese Medaka have chromosomal sex determination where the male-determining Y chromosome is morphologically indistinguishable from its pairing partner; it is thought that the inability to distinguish X and Y chromosomes in Medaka reflects recent evolution of a chromosomal sex-determining mechanism from an ancestral pair of autosomes (3, 392). One fish, the Atlantic Silverside, actually uses chromosomes to determine sex in part of its geographic range, while using temperature to determine its sex in another part (68).

	II. SEX-DETERMINING ROLE OF HUMAN Y CHROMOSOME

Top Previous Next References

In mammals, the requirement for the Y chromosome to achieve testicular differentiation suggested the existence of a Y chromosomal gene responsible for inducing the state of maleness, termed the testis-determining factor (TDF). Subsequently, the search for the TDF was narrowed to the short arm of the Y by cytogenetic analysis of abnormal Y chromosomes. Loss of the Y long arm or formation of an isochromosome consisting of duplicated short arms without any long arm material was compatible with male development, whereas in contrast, isochromosomes consisting exclusively of long arm material showed female phenotype (206). Although male-specific histocompatibility H-Y antigen looked promising as the TDF (429), both men and mice (278) were found who possessed testes but lacked H-Y antigen, and thus, the search for TDF continued.

Gene mapping studies on the Y chromosome presented unique problems due to the distinct properties of three types of genetic material present on the Y. The first, the pseudoautosomal regions (PARS), located at the tips of the short and long arms of the Y chromosome (109, 129, 146, 316), contain DNA shared between both the X and Y sex chromosomes (see Fig. 1). Pseudoautosomal regions are needed to guide correct pairing and segregation of the sex chromosomes during male meiosis (134, 290). Unique among chromosome partners, the sex chromosomes form a synaptonemal complex at their tips rather than at their centromeres, and while paired, a single X-Y recombination almost always occurs within the Y short-arm PAR (41, 351, 377). The long-arm PAR is smaller than the short-arm PAR, <500 kb in length. XY recombination or gene conversion in the long-arm PAR is rare, present in only 4% of male meioses (129). Recombination maps of the Y chromosome were developed (439) based on recombination frequencies in the short-arm PAR region.

View larger version (24K): [in this window] [in a new window]   FIG. 1.   Schema of the X and Y chromosomes is shown during meiosis, featuring a crossover between X and Y DNA in the short-arm pseudoautosomal pairing region. Also shown is region of synaptonemal complex formation. Inset: location of SRY testis determining gene within unique Y chromosome DNA on Yp short arm.

The second type of genetic material on the Y chromosome consists of repetitive satellite DNA visible by special staining. Because the satellite DNA of one species will not cross-hybridize with the satellite DNA of another, these regions are not thought to have any function (374). In fact, up to 70% of human, mouse, and Drosophila Y DNA consists of repeated DNA (70), and sex determination proceeds normally when the distal long arm of the human Y is deleted (262), showing that this region harbors no information necessary for sex determination.

The third type of Y genetic material is unique Y chromosomal DNA that does not undergo recombination with the X chromosome. It is in this unique Y DNA that the TDF is located. Unique Y region DNA cannot be analyzed by traditional genetic methods that depend on recombination frequency to determine the distance between loci; instead, the unique Y region can be analyzed by correlating presence of Y-specific DNA markers to Y chromosomes with cytogenetic abnormalities. Maps of the Y chromosome based on DNA hybridization (147, 449) and maps based on deletions occurring in human patients (427) were developed to assist not only in the search for the testis-determining gene, but also to facilitate study of Y genes involved in spermatogenesis (11, 16, 260, 458, 460) and to facilitate understanding of chromosome structure and function.

	III. IDENTIFICATION OF THE SEX-DETERMINING GENE SRY

Top Previous Next References

By analysis of chromosomal translocations where Y DNA was transferred to XX phenotypic males and where Y DNA was lost by deletion in XY phenotypic females, Page et al. (318) were able to define a critical region for testis determination that was just proximal to the pseudoautosomal X-Y pairing region on the short arm of the Y, and focused their attention on a 140-kb segment that was present in one XX male (LGL203) and absent in another XY female patient (WHT1013). A highly conserved gene in this portion of the Y (called zinc-finger Y, or ZFY) encodes a zinc-finger containing DNA-binding protein that was initially thought to be the TDF (318). Evidence consistent with the assignment of ZFY as TDF included finding ZFY-related sequences on Y chromosomes of all mammals tested and finding that a ZFY-related gene, Zfy-1, was present in the mouse sex determining region, a region termed Sxr'. However, additional study revealed that ZFY could not be the TDF: a close homolog of ZFY was found on the X chromosome of eutherian mammals (361). Also, ZFY expression was found to be in germ cells, which are not required for testicular differentiation, and testes differentiation was unaffected in W^e/W^e mice that lack germ cells and lack detectable expression of Zfy-1 and Zfy-2 (234). Finally, Y-derived sequences lacking ZFY were found in four XX men (322), and the patient whose translocation led to the cloning of ZFY was found to have a more complicated DNA rearrangement than first appreciated (317), so the search for the primary sex-determining gene continued.

Additional studies of XX men with Y chromosomal DNA but not ZFY permitted precise localization of the TDF. One marker probe recognized an 8.5-kb band in normal males, but hybridized to smaller DNA fragments in XX males who harbored translocated fragments of the Y chromosome. This result implied that the probe was at or near the breakpoint of the translocations and that the testes-determining gene had to lie within the region delimited by the pseudoautosomal boundary and this probe, a distance of 35 kb. A single-copy gene that encodes a protein with homology to high-mobility-group (HMG) protein nuclear transcription factors was identified within this region (156, 372) and termed SRY, for sex-determining region Y gene. At present, an overlapping set of DNA clones spanning the entire euchromatic portion of the Y chromosome has been developed (120), and the DNA sequence of the sex-determining region and pseudoautosomal boundary has been determined (444).

The HMG proteins were first characterized as chromatin components with a high electrophoretic mobility in polyacrylamide gels, and sequence analysis of various HMG proteins shows that they can be grouped into three families, the HMG-14/-17 group, the HMG-1/-2 group, and the HMG-I/-Y group (44). The SRY gene belongs to the HMG-1/-2 family, and when the term HMG box is used in this review, it refers to the homologous DNA binding domain of proteins in the HMG-1/-2 family.

The HMG-1/-2 DNA binding domain defines an evolutionarily ancient superfamily whose members appear in animals, plants, and yeast. The family can be subdivided into two main classes: canonical members of the family bind to non-B-DNA conformations such as B-Z junctions, stem loops, cruciforms, four-way junctions, and cisplatin-modified DNA (18, 44); however, several members of the family, including LEF-1, IRE-ABP, TCF-1, Rox1, and SRY itself, bind to closely related variants of the A/T A/T C A A A/T G DNA motif with sequence specificity independent of any known requirement for DNA conformation (80, 141, 167, 171, 300, 413).

Several lines of evidence support the role of SRY as the TDF. In a line of XY sex-reversed female mice, the SRY gene is deleted from the defective Y chromosome (156). Also, the phenotype of XX mice transgenic for a 14-kb EcoR I fragment containing the mouse Sry locus recapitulates maleness. About 30% of these XXSry mice develop testes and secondary somatic sex characteristics and exhibit male mating behavior (235). The fact that not all XXSry transgenic mice exhibit sex reversal implies that dosage and timing of Sry expression are critical for its effects, and these parameters may be subject to variation depending on the insertion site of the transgene. Indeed, human and mouse Y chromosome deletions that leave the Sry ORF intact but remove sequence including the members of the multicopy ribosomal binding motif (Rbm) gene family (260) between Sry and the centromere result in sex reversal as a result of decreased urogenital ridge expression of Sry (50, 244, 275). The decrease in Sry expression is hypothesized to be the result of an increase in proximity of the Sry gene to transcription-repressing effects of a centromere-associated chromatin domain.

SRY is appropriately present in the gonadal primordia at the outset of sexual differentiation. The most precise ontogeny studies have been performed in mice, where Sry can be detected by reverse transcription (RT)-polymerase chain reaction (PCR) and ribonuclease (RNase) protection between embryonic days 10.5 and 12.5 (161, 236, 245, 350). Expression of Sry was not dependent on the presence of primordial germ cells (236). It is interesting that the initiation of SRY transcription occurs subsequent to the condensation of cells that form the urogenital ridge by embryonic day 10. Presumably, an additional regulatory system is in place to orchestrate the initial epithelial-mesenchymal transition that begins the formation of the indeterminate gonad (161). That precise control of Sry expression is needed is attested to by the presence of three ATGs in the 5'-untranslated region of the SRY mRNA and four copies of an AUUUA sequence reminiscent of the Shaw and Kamen destabilizer (367) in the 3'-untranslated region (161). These features may serve to limit the expression of the rare SRY mRNA by regulating translation and mRNA stability, respectively.

After birth, Sry is again transcribed in round spermatids of the adult testis, where Sry is expressed as a circular mRNA. The circular form is thought to be favored, since surrounding genomic sequences 5' and 3' to the open reading frame are repetitive elements that share homology and could mediate base-pairing to form an RNA-RNA duplex. Circular transcripts are not translated, since no Sry protein can be detected in round spermatids (51). Expression of Sry in round spermatids is not essential for their development, since male mice lacking Sry were fertile (255). This is true for other species as well, for example, the creeping vole, which eliminates the entire Y chromosome in the germ line (309).

Expression of SRY has also been detected by RT-PCR but not by RNase protection at an additional site in mice, the preimplantation blastocyst (34, 46, 135, 161, 465). Although it was originally suggested that Sry expression might account for the faster growth rate (greater numbers of somites and greater weights at equivalent days post coitum) seen for male compared with female embryonic mice, these early sex dimorphisms are unperturbed by removal of Sry from the mouse strain 129 Y chromosome. In addition, XO fetuses have a growth advantage over XX fetuses, suggesting a contribution of an X-linked locus that acts to repress size to early embryonic sexual dimorphism before onset of gonadal differentiation (43). Further work is needed to examine various X- and/or Y-linked candidate genes to define at the molecular level the locus or loci responsible for this phenomenon.

In nonrodent species, SRY exhibits much less tissue specificity. All fetal human tissues examined expressed SRY, and in adult males, heart, liver, kidney, and testis showed SRY expression by RT-PCR analysis (63). SRY is also expressed in HepG2 and NTERA-2 cl.D1 cell lines, derived from adult tissues that express SRY (63). The significance of widespread SRY expression is unclear. Indeed, in the tammar wallabee, Sry expression begins at the time the gonadal primordium is present, 4 days before an increase in cytoplasmic-to-nuclear ratio heralds the onset of Sertoli cell differentiation (173). The long time delay between expression of SRY and differentiation of Sertoli cells suggests the presence of auxiliary autosomal or Y-linked factors in marsupials necessary to cooperate with SRY to regulate gonadal sex. Like human SRY, wallabee Sry is expressed in many fetal and several adult tissues including brain; these results make possible a widespread role for Sry in development of mammals. Certain Old World rodents have multiple copies of the Sry gene (298); it is unknown whether all these copies are transcribed or if some have been recruited for additional roles in development. Two rat Sry transcripts of distinct size and quantity are present in urogenital ridge cell lines (Haqq et al., unpublished data).

SRY cannot be the only genetic switch molecule involved in mammalian sex development; rather, a cascade of molecules seems to be operating. For example, the wood lemming has three sex chromosomes, an X, a Y, and a modified X*, which functions to suppress the action of the Y (127). Even more striking, Sry is absent from the voles Ellobius lutescens and Ellobius tancrei, which have no Y chromosome (222). In these species, the X remains unpaired in male meiosis, making their mechanism of sex determination an interesting topic for future investigation.

Furthermore, in marsupial mammals where the Y chromosome determines gonadal sex, many sexual determinants such as scrotum versus pouch formation occur before differentiation of the gonads (307), under control of a separate system based on the ratio of X chromosomes to autosomes (366). Further investigations in these model systems promise to lead to identification of additional sex-determining molecules, and it will be interesting to see how Y-determining and X to autosome-determining systems can coexist.

	IV. MUTATIONS IN SRY EXPLAIN A SUBSET OF PATIENTS WITH 46, XY PURE GONADAL DYSGENESIS

Top Previous Next References

The differential diagnosis of patients who are 46, XY but have female characteristics includes deficiency of testosterone or its 5-reduced metabolite, defects in the androgen receptor, chromosomal mosaicism, or gene mutation resulting in failure of gonadal differentiation. Defects in the steroid hormone biosynthetic and signal transduction pathway are discussed in a section VIIIF. XY sterile females with bilateral streak gonads and sexual infantilism are said to have the pure gonadal dysgenesis syndrome, which is distinguished from the more common syndrome of mixed gonadal dysgenesis where mosaicism for 45, XO/46, XY, commonly leads to a streak gonad on one side and a testis on the other (283).

In pure gonadal dysgenesis, both gonads are frequently streaks, and the height of patients is normal, in contrast to the 45, XO Turner syndrome where patients have short stature, webbed neck, cardiac defects, and dorsal pedal edema. The short stature of Turner patients is likely due to a Y long-arm gene mapped between Y intervals 4B and 5G that influences height (355); this gene may or may not be identical to a Y long-arm gene that plays a role in tooth size (5a). In addition, 30% of pure gonadal dysgenesis patients develop gonadoblastoma (420), a rare neoplasm containing aggregates of germ cells intermixed with a stromal component resembling immature Sertoli and granulosa cells. Because 45, XO Turner syndrome patients who also have dysgenetic gonads never develop gonadoblastoma, it has long been suspected that a Y-linked locus is responsible for gonadoblastoma, and the critical genetic region has now been mapped (404).

Mutation of SRY is the first molecular explanation for the syndrome of 46, XY pure gonadal dysgenesis. As of March 1995, nearly 20 mutations in human SRY have been described (see Fig. 2 for an overview of 46 XY female patient mutations and mutations involved in SRY DNA binding-F67A and I68F; Refs. 27, 36, 176, 177, 182, 209, 274, 295, 337, 385, 424, 425). Most of these mutations lie within the region of homology to the conserved HMG box DNA binding domain, with the exception of one COOH-terminal mutation resulting in premature termination of the SRY open reading frame 41 amino acids short of the full-length protein (385). The recent finding that the COOH-termimal region binds a novel PDZ domain-containing protein, SRY-interacting protein-1 (SIP-1) (336), adds to the significance of this mutation. The fact that only 15% of 46, XY pure gonadal dysgenesis patients have an SRY mutation suggests that SRY is only one of several genetic loci whose mutation can lead to an identical phenotype. The etiology of the remaining patients must be due to mutations in genes that affect the expression of SRY itself or that play independent roles in a cascade of regulatory events specifying gonadal development.

View larger version (26K): [in this window] [in a new window]   FIG. 2.   Published mutations associated with female phenotype in 46, XY individuals are displayed above a schema of sex-determining SRY gene. In addition, 2 mutations not derived from sex-reversed patients, F67A and I68F, were introduced into SRY to facilitate structure-function correlation. Note that all but 1 of the mutations associated with sex reversal affect structure of SRY DNA binding domain, the high mobility group (HMG) box.

	V. TRANSLOCATION OF SRY TO THE X CHROMOSOME OR AUTOSOMES EXPLAINS A SUBSET OF PATIENTS WITH 46, XX TRUE HERMAPHRODITISM

Top Previous Next References

In addition to the group of 46, XY sex-reversed females, there exists the converse type of sex reversal where 46, XX individuals display a masculinized phenotype. The bulk of such patients now come to medical attention in infancy, suffering the effects of steroid biosynthetic defects leading to congenital adrenal hyperplasia, which is discussed in section VIIID. This is distinguished from true hermaphrodites, 90% of whom are 46, XX with a mixture of ovarian and testicular tissue either on different sides of the body or within a single gonad, or rarely bilateral testes (81, 113, 163, 251). True hermaphrodites come to attention during infancy due to ambiguous genitalia (97), or later in life due to infertility. A high incidence of true hermaphroditism occurs in African Bantu tribal nations (416). Gonadoblastoma is not a frequent finding (2.6%; Ref. 416) in true hermaphrodites, even those who have inherited Y DNA, presumably because the gonadoblastoma susceptibility region is not always coinherited with the SRY locus (163). Likewise, reports of rare paternity may correlate with translocation of sufficient Y material to include spermatogenesis loci along with the testis-determining locus.

SRY translocation to the X chromosome or an autosome accounts for ~30% of true hermaphrodites (163, 251). Genotype-phenotype correlation shows that individuals with SRY translocations have normal male external genitalia, whereas those without SRY present genital ambiguity (113, 306). Although mosaicism with a Y-bearing cell line could explain remaining cases of true hermaphroditism, there is a dearth of supporting evidence, and it is likely that the bulk of cases arise from mutations of autosomal loci important to sex differentiation.

	VI. BIOCHEMICAL ACTIVITIES OF SRY: HOW DOES SRY DETERMINE SEX?

Top Previous Next References

SRY must act to control the process of gonad formation. Normally, differentiation of the gonadal primordium into male Sertoli and Leydig cells depends on the presence of SRY and the Y chromosome (41; reviewed in Ref. 48). However, the requirement is not absolute, and there are certain situations in which male differentiation can occur in the absence of the Y. For example, XX gonadal primordia transplanted under the kidney capsule of adult male hosts develop into ovotestes (391). In these experiments, XX Sertoli cell development is delayed by 7 days relative to XY Sertoli cells. Sertoli development requires contact with the host tissue (390), and Müllerian inhibiting substance (MIS) is produced concomitant with loss of germ cells in the resulting ovotestes (389).

The embryological origin of Sertoli and Leydig cell precursors is not well understood. In light and electron microscopic studies, both contributions from coelomic epithelium and mesonephros are implicated (357). In vivo, many connections of the convoluted mesonephric tubules continuous with the gonadal primordium are seen in both male and female mouse embryos at the earliest time studied, day 11.5 post coitum, and are extinguished by day 12.5 (225). Mesonephros-gonad connections could contribute cells that would give rise to the Sertoli cell lineage; however, RT-PCR for the linear urogenital ridge SRY transcript in microdissected day 11.5 post coitum mesonephri and gonads found expression only in the gonad proper, so if pre-Sertoli cells come from the mesonephros, they are not yet competent to express Sry (211). Experiments in organ culture using testes older than 11.5 days post coitum show cell recruitment from the mesonephros contributing to the myoid, endothelial and possible Leydig cell populations (225); in fact, placement of a semipermeable membrane between mesonephros and gonad led to the failure of further gonadal development (40, 285). Another source of mesenchyme, the hindlimb bud, is able to substitute for mesonephros in organ culture experiments, but the sex of the donor limb tissue had a marked effect on subsequent testosterone output by the putative Leydig cells (292).

In our work, we have made immortalized rat gonadal cells using retrovirus to introduce the v-myc oncogene at day 14.5 post coitum (168; and Haqq et al., unpublished data). At this time, primordial Sertoli cells are condensing into cords, and myoid, Leydig, and germ cells are all present. An analysis of gene expression by RT-PCR shows that cloned cell lines from the mixture of immortal cells express urogenital ridge markers (Fig. 3, A and B). Each cell type has a distinct morphology in phase-contrast light microscopy. In experiments where two cell types were mixed together, we were intrigued to find that CH-17 cells have the ability to organize other cell types into aggregates that bear an at least superficial resemblance to early testicular cords (Fig. 3C). Much work lies ahead to develop molecular markers capable of distinguishing the various early gonadal cell types. Comparing patterns of gene expression in the urogenital cell lines described here may be a reasonable way to discover markers specific to particular early cell types present in the gonad.

View larger version (70K): [in this window] [in a new window]   FIG. 3.   A: reverse transcription (RT)-polymerase chain reaction (PCR) analysis of urogenital ridge cell line RNA for WT-1 gene expression using primers indicated. B: table summarizing RT-PCR experiments to characterize expression of several urogenital ridge markers in immortalized urogenital ridge cell lines. C: ability of CH-17 cells to organize CH-23 cells but not control CH-15 cells into structures reminiscent of seminiferous tubules. CH-17 cells were able to organize other urogenital ridge cell lines including CH-4 and CH-34 into similar structures.

SRY biochemistry provides some inroads to begin addressing the question of how this transcription factor operates to activate the events necessary to form the gonad. The 80-amino acid HMG box DNA binding domain is highly diverse, ranging in particular HMG box proteins from 72 to 83 amino acids in length and containing three -helical regions. SRY's HMG box is 77 amino acids in length, of average length compared with other HMG boxes. In addition, there are no strictly conserved amino acid sequences between all members of the superfamily: only three amino acids are conserved in >80% of HMG boxes: P at position 8, W at position 45, and K at position 53 (243).

Some members of each of the HMG groups are present in large quantities, bind DNA with little sequence specificity, and bind novel DNA structures including four-way junctions that are thought to be intermediates involved in DNA recombination. In contrast, the group of HMG-1/-2 box proteins to which SRY belongs possesses both sequence-specific and DNA structure-specific binding. Because mutations in SRY found in cases of XY sex reversal are located in the HMG DNA binding domain, sequence-specific DNA binding is thought to play a critical role in its sex-determining function. A similar role for structure-specific DNA recognition of four-way junction elements by SRY is unlikely, since mutations that cause sex reversal do not affect the ability of SRY to interact with four-way junction DNA (328). SRY is closely related to a family of HMG box proteins termed the SOX (Sry bOX) family, and because many SOX genes can bind the same consensus DNA target sites that SRY binds, and many are coexpressed in the same tissues as SRY, understanding the mechanism of SRY action must take into account the presence of these proteins.

Homeodomain proteins, long thought to function solely by binding to DNA, have been shown to bind RNA, exerting control of translation to establish a molecular gradient during early Drosophila development (104, 347). These observations provide a precedent for families of proteins operating to control both transcription and translation. It is intriguing that one member of the HMG protein family, HMG2, can interact with the POU domain of transcription factors Oct1 and Oct2 via its HMG domain (464), causing a marked increase in the DNA binding activity of the Oct proteins. The inability of the SRY HMG domain to interact with Oct2 in parallel experiments (464) suggests that this type of protein-protein interaction is specific, and it is tempting to speculate that the SRY HMG domain may itself be a protein-protein interaction domain.

The structures of SRY coding regions in various species have been evolving at a rapid rate (243, 406, 445), and little homology is present in these molecules outside of the DNA binding HMG box region. The lack of homology outside of the HMG box led to the proposal that the only important part of the SRY molecule was its HMG DNA binding domain. This hypothesis has been reconsidered since the COOH-terminal non-HMG region of the SRY molecule is changing faster than can be explained by accumulation of neutral mutations having no effect on SRY function. Instead, it seems that changes in the SRY COOH terminal must give SRY a species-specific advantage (406, 445). In support of there being a role for non-HMG box regions of SRY, a mutation in an XY female that results in deletion of the 41 most COOH-terminal amino acids has been described (385; see Fig. 2), and more recently, this region was found to bind a novel protein termed SIP-1 (336). This PDZ domain containing protein may favor protein-protein interactions in a higher order transcription complex.

SOX genes are present in several species as closely related groups of molecules, as in the Drosophila melanogaster HMG-D cluster, the mouse Sox 1 cluster, and the reptile Sox 12 cluster. Because whole families of HMG box proteins are present in single species like Drosophila and close homologs are not present in any other animal group (243), SOX genes may have rapidly adapted during evolution to subserve species-specific roles. Very little is known about the putative species-specific or tissue-specific function of the majority of the SOX genes.

Expression patterns for several Sox genes have been published within the last years: Sox 1 is expressed in developing brain and testicular somatic cells (66); Sox 2 is expressed in developing brain, gut endoderm, and testicular germ cells (66, 410); and Sox 3 is expressed in developing brain, spinal cord, adrenals, liver, thymus, spleen, pancreas, and testicular somatic cells (66, 410). Because Sox 3 maps to a part of the X chromosome associated with mental retardation, Sox 3 is a candidate gene for X-linked mental retardation syndromes (383). Sox 4 is implicated as a transcription activator in lymphocytes containing a transcription activation domain that can function when fused to the DNA binding domain of GAL4 in yeast (112, 414); Sox 5 is expressed in postmeiotic germ cells (83); Sox 6 is expressed in developing brain and adult testis (71); Sox 11 is expressed in mesoderm and neuroepithelium, with highest levels concomitant with postmitotic neural differentiation (410); Sox 18 is expressed in adult heart, lung, and skeletal muscle (188); and zebrafish Sox 19 is expressed in late gastrulation stage embryos (428).

Knowing that campomelic dysplasia was associated with primary gonadal dysgenesis (403a), and finding the sequence of the SRY-related gene SOX 9 (125) at the breakpoint of a patient with campomelic dysplasia, led to the realization that mutations in SOX 9 were responsible for both syndromes (401, 430). SOX 9 is expressed at the right time in the developing urogenital ridge to cooperate with SRY itself in sex determination. Ontongeny analysis of SOX 9 indicates that it is expressed at higher levels in the somatic cells of the male gonad immediately after SRY expression (291). Because these two gene products are coexpressed in the soma of the gonad in a male specific pattern, analysis of how the two interact may shed light on the mechanism of Sry action and sex determination. Alternately, it is possible that SOX 9 may be expressed in a cell type that interacts with the Sertoli cells expressing SRY, to specify competence to respond to a Sertoli cell SRY-induced signal (proposed in Ref. 47). Because SOX 9 plays an important role in the development of mesoderm-derived skeletal elements, the SOX 9 positive cell type might be the mesonephric cells that migrate into the gonad as seminiferous tubule myoid architecture is defined (40). Sox 1, Sox 3 (66), and the Sox 4-related gene IRE-ABP (M. Alexander-Bridges, personal communication) are all coexpressed with Sry and Sox 9 in somatic cells of the developing mouse gonad.

One of the best clues to Sox gene function comes from homology comparison outside of the HMG box: Sox 1,2,3 and zebrafish Sox 19 share three regions of homology outside the HMG box (410, 428; see Fig. 4). A fourth non-HMG box region of homology is evident in comparisons of Sox 4 and Sox 11 (Fig. 4). We speculate that these regions of conservation define protein-nucleic acid or protein-protein interaction domains, and it will be interesting to see if they specify homotypic or heterotypic interactions.

View larger version (48K): [in this window] [in a new window]   FIG. 4.   Comparison of Sox gene sequences and identification of 3 amino acid domains of cross-species evolutionary conservation. We speculate that these domains mediate protein-protein or protein-DNA interactions. [Data from Uwanogho et al. (410) and Vriz and Lovell-Badge (428).]

Researchers in the SRY field can gain insight in studying SOX genes that bind to identical or very closely related binding sites in the same cell type by considering investigations of HMG box proteins in lymphocytes. For example, LEF-1 (also known as TCF-1) and TCF-1 are two HMG proteins present in lymphocytes that can bind the same ATTGTT consensus. LEF-1 was first characterized as a pre-B and T lymphocyte specific DNA binding protein that bound to the T-cell receptor -chain enhancer (402, 435). Subsequent studies showed LEF-1 to help assemble a multiprotein complex on the TCR- enhancer (141-143) and to be widely expressed in embryonic tissues (310). TCF-1, identified in T cells (412), was also subsequently shown to be widely expressed during embryogenesis (310) and later during development (461). In addition, the LEF-1 knock-out had no detectable immune system phenotype and rather led to striking effects on tooth, hair follicle, and mammary gland development (415), whereas the TCF-1 knock-out led to a specific phenotype in maturation of CD8+ to CD4+/CD8+ during T-cell maturation (418).

Thus studies aiming to understand SRY's action will need to be similarly inclusive, since more than one SOX protein can interact with its target element, and SRY protein is present at levels orders of magnitude lower than other members of this family. The question of how such factors can achieve specificity in their actions is of major importance. It has been proposed that SRY might cooperate with other transcription factors to achieve specificity (254). Some investigators have suggested that SRY may cooperate with AP-1-type transcription factors (64). In support of such models, Sox 2 is required to cooperate with Oct-3 at a fibroblast growth factor-4 enhancer that is active in embryonal carcinoma cell lines (457). Specificity of this interaction was apparent, since Sox 5 and Oct-2 were unable to substitute in cotransfection assays. These assays suggest that SRY can only activate in a specific context of combined binding sites.

Posttranslational modification of SRY might modify the activity of SOX proteins. TCF-1/LEF-1 can activate transcription through a single copy of its response element in nonlymphoblastoid cell lines, whereas it only mediates activity of the -enhancer with intact binding sites for CREB and TCF-2 nearby in T cells (435). Similarly, GATA-1 can activate minimal promoters bearing an upstream GATA site in nonerythroid but not erythroid cells (311). This demonstrates that posttranslational modifications, or a requirement for cooperation with tissue-specific factors, restrict LEF-1 and GATA-1 activity to particular cell types. This may serve as a model for tissue-specific modulation of SOX protein activity. -Catenin was recently shown to interact with TCF and LEF transcription factors. The resulting constitutive gene activation can be modulated by the adenomatous polyposis coli (APC) tumor suppressor gene in APC-deficient colon carcinoma cells (293) or melanoma cells (352).

The biology of the MIS is reviewed in detail in section VIIIA. Here, we consider the interaction of SRY and MIS, a logical putative downstream target of Sry activity, since MIS transcription begins within 20 h of Sry expression in mouse urogenital ridge (161) and is male specific until MIS transcription begins after birth (reviewed in Ref. 159). We determined that the bacterially expressed HMG domain of human SRY could protect the human MIS promoter in deoxyribonuclease (DNase) footprint assays with a nanomolar dissociation constant (167), and we and others (167, 170) determined that SRY binding to human MIS promoter nucleotides 75 to 45 was diminished by nucleotide point mutations. These analyses also showed that the MIS binding site was not as high affinity a target site as SRY sites identified by PCR selection starting from random sequence (5'-ATTGTT; Refs. 168, 172). The weak nature of the SRY binding sites suggested they might be relevant only in the context of occupancy of neighboring binding sites for additional transcription factors. Thus we made cell lines from the urogenital ridge at the precise time that differentiation of the testis was in progress (168), in the hope that such lines would contain all factors necessary to transcribe the MIS promoter.

Human SRY does not have a transcription activation domain detectable by fusing SRY to the DNA-binding domain of GAL4; in contrast, mouse Sry possesses a COOH-terminal domain that has activity in transcription activation when fused to the DNA binding domain of GAL4 and transfected with a reporter containing GAL4 promoter sites (103). In a male urogenital ridge cell line, CH-34, cotransfection of an expression vector directing synthesis of human SRY with an MIS reporter plasmid resulted in a dramatic induction of transcription (168). This activation appeared to be specific, since no activation was seen for the Rous sarcoma virus LTR promoter or the mouse actin promoter (Fig. 5A).

View larger version (28K): [in this window] [in a new window]   FIG. 5.   A: cotransfection of indicated chloramphenicol acetyltransferase (CAT) reporter vectors (5 µg) with control empty expression vector pCMV (7) or expression vector directing expression of full-length human SRY (2 µg) into urogenital ridge cell line CH-34. Human Müllerian inhibiting substance (MIS) reporter vector was generated by cloning the fragment 299 to +8 generated by PCR using primers CMH-106 5'-ggg Agg CTTggCCgTCACTCCCAgCCTg and CMH-107 5'-CggAAgCTTggg TgCTgCCAggggCTg, into Hind III site of SVO-CAT (149a). Forty hours after transfection by the calcium phosphate method, cells were harvested and assayed for CAT activity by thin-layer chromatography (232). B: comparison of fold induction of 114 bp human MIS-luciferase reporter (2 µg) by SRY, LEF-1, Sox 2, IRE-ABP, or control empty expression vectors (1 µg). The 114 to +8 fragment of human MIS promoter was generated by PCR using primers CMH-128 5'-ggTACCgggCACTgTCCCCCAAgg and CMH-107. Resulting fragment was sequenced for confirmation and subcloned between Kpn I and Hind III sites of luciferase reporter pA3 (451). Although SRY was a strong activator in this assay, Sox 2 was a strong repressor, whereas IRE-ABP was a weak activator and LEF-1 had no consistent effect. Luciferase data are an average of 3 independent transfections ± SE. All transfections included 0.5 µg pXGH5, which directs expression of growth hormone under the control of a ubiquitously active promoter (364). Growth hormone determinations were used as an internal standard to control for variations in transfection efficiency.

In addition, the response of the transfection system to SRY is highly specific at the protein level. We find that SRY is the only HMG protein able to greatly activate the human MIS promoter reporter constructs; HMG-2 and LEF-1 are incapable of upregulating in our assay when equal amounts of expression plasmid are transfected. IRE-ABP gives upregulation of the reporter to levels only about one-tenth those achieved with SRY, and Sox 2 transfected in our assay leads to a strong repression of MIS transcription (Fig. 5B). Mouse SRY (data not shown) was also unable to activate the human SRY reporter in this assay. Ongoing experiments should address the question of whether SRY and other HMG proteins known to be expressed in the urogenital ridge might exhibit synergism or antagonism, but specificity with the Sox gene family.

Mutated SRY protein I68T found in an XY sex-reversed phenotypic female had diminished DNA binding activity in bandshift assays and failed to activate transcription of the reporter plasmid (168). Biophysical analysis of the interaction of this mutant protein I68T led to the identification of a hydrophobic wedge that intercalates between adjacent A-A base pairs in SRY binding sites (168, 230, 253, 441, 442; Ukiyama et al., unpublished data; Fig. 6A). An analogous mechanism of nucleic acid recognition is used by the TATA-binding protein (228, 229), the Hae III methyltransferase (342), and PurR (363), DNA-binding proteins belonging to diverse gene families, suggesting that this mechanism is widely utilized.

View larger version (48K): [in this window] [in a new window]   FIG. 6.   A: structure of SRY with the 4 amino acids M64, F67, I68, and W98 constituting the hydrophobic wedge (see text) are shown in bold. B: sequences of SRY HMG DNA binding domain and additional HMG boxes were obtained from GenBank and aligned by use of University of Wisconsin Genetics Computer Group software. Positions of amino acids whose side chains constitute hydrophobic wedge are boxed and are present within helix 1 and helix 2 of HMG structure predicted by analogy to known structures of SRY and LEF-1 (253, 441). C: transfection analysis performed as in Fig. 5B, using expression vectors encoding mutant SRY strains indicated. Luciferase data are presented as a mean of 3 independent transfections ± SE after normalization to an internal growth hormone control. All 4 mutations at hydrophobic wedge positions decreased transcription response seen in transfection assay. M64I and I68T from sex-reversed patients, as well as M64A and F67A, predicted to have defective function based on SRY structure, all show decreased transcription of 114 bp human MIS-luciferase reporter vector. Note that M64I has partial activity in assay (see text). D: transfection of a variable amount of SRY mutant I68F or wild-type expression vector (0-5 µg) with a fixed amount of 114 bp human MIS-luciferase (2 µg). Empty expression vector pCMV was used to normalize total amount of DNA present in each assay. All points on graph are means ± SE of 3 replicates, corrected for transfection efficiency using an internal growth hormone standard. SRY I68F shows a markedly enhanced ability to transactivate the 114 bp human MIS-luciferase reporter vector.

Nuclear magnetic resonance structure determination has defined the four residues M64, F67, I68, and W98 that constitute the hydrophobic wedge, which is conserved in many HMG box sequences (442; Fig. 6B), and we have now extended our analysis of defective SRY proteins to include hydrophobic wedge mutants M64I, M64A, F67A, I68T, and W98R. Three of these mutants, M64I, I68T, and W98R, are clinically associated with sex reversal.

Mutants M64A, F67A, and I68T are well folded after synthesis in Escherichia coli, and all show decreased affinity in DNA-binding gel-shift assays (E. Ukiyama, C. Haqq, Y. Morikawa, M. Weiss, and P. Donahoe, unpublished data). W98R fails to fold properly and was therefore excluded from further analysis. Figure 6C presents an analysis of transcriptional activation by mutant SRY proteins M64A, M64I, F67A, and I68T assayed by transient cotransfection of mutant SRY molecules with MIS-luciferase reporter plasmids in urogenital ridge cell line CH-34. All mutations of hydrophobic wedge positions resulted in substantial loss of transcription response. Mutation of the isoleucine-68 to threonine results in 50-fold decreased DNA binding and lack of transcription response (168; Fig. 6C). Because isoleucine-68 is the position that plays an important structural role in DNA binding by intercalation into a widened minor groove, the lack of transcription seen when the mutant protein is transfected into cells proves that DNA binding activity is necessary for SRY to show upregulation in the assay, arguing against artifactual transcription due to nonspecific protein-protein interactions.

Although mutation of the first hydrophobic wedge position, methionine-64 to alanine, results in abolition of the transcription response, mutant protein M64I, which is associated with sex reversal clinically, had partial activity in the transfection assay. Analysis at this position again demonstrates a tight correlation of SRY activity in the transfection assay with biochemical analysis. Previous investigation of the M64I mutation showed that the DNA binding activity was largely intact (<2-fold decrement in DNA binding affinity; Refs. 171, 334; Ukiyama et al., unpublished data), but that the angle induced upon DNA binding was greatly altered (334). For this reason, it was proposed that a defect in DNA bending was the explanation for sex reversal with the M64I mutation. However, recent results prompt reevaluation of this hypothesis. Our group has determined that the decrement in DNA bending by the M64I mutation is only 30° (Ukiyama et al., unpublished data). We believe the difference stems from lengths of the bacterially expressed SRY HMG domains used by the two groups, the former using 77 amino acids that did not include the basic domain just COOH terminal to the classic HMG box, wherease the latter used an 85-amino acid protein that did include this region. The TCF-1 and LEF-1 basic region contributes to DNA binding and DNA bending (253, 341). Experiments are now under way to compare protein-induced DNA bend angles in the identical experimental assay with the two lengths of SRY peptides. Changes in SRY protein-induced DNA bend angle are unlikely to be responsible for the diminished biological activity of the M64I molecule, unless the cotransfection assay is sensitive to changes in bend angle of only 30°.

We reasoned that the hydrophobic wedge of SRY might not be optimal for intercalation into the minor groove and interruption of base stacking, since the isoleucine at position 68 is considerably less hydrophobic than the phenylalanine present at the analogous position of other HMG proteins, including chimpanzee Sry. We predicted that if we replaced the isoleucine-68 position of SRY in the hydrophobic wedge with phenylalanine, we could make a superactive SRY molecule. In fact, cotransfected SRY I68F yields more activation than wild-type SRY when equal concentrations of expression vectors are cotransfected (Fig. 6D). High concentrations of SRY I68F yield a maximal activation that is at least four times higher than wild type, and 16 times higher than M64I (data not shown), suggesting a qualitative gain of function in this mutant. Future studies using this superactivating I68F mutant molecule may be able to shed light on the mechanism of transcription stimulation by SRY.

Analysis of deletion constructs of the human MIS promoter showed that coexpressed SRY was able to stimulate even very small fragments of the MIS promoter as long as they contained the start site for transcription. However, the quantitative amount of SRY-induced transcription was greatly reduced by each deletion. The fact that the smallest fragments that do not contain a binding site for SRY still upregulated suggested that the cotransfection system might be biased because of SRY binding to sites within the vector itself (Morikawa et al., unpublished data). In support of this, empty reporter vector without an MIS promoter insert could be stimulated by transfection of large amounts of SRY. Sox 18's transactivation properties include the same activation of the empty reporter vector, with further transcription resulting from ATTGTT sites placed upstream of a minimal promoter (188).

To our knowledge, all available chloramphenical acetyltransferase, growth hormone, and luciferase reporter vectors used for these experiments contain the simian virus 40 (SV40) polyadenylation signal in multiple tandem copies just 5' to the promoter cloning site, and each SV40 polyadenylation signal contains a perfect ATTGTT match to the consensus site for binding of SRY and SOX proteins; however, reporter vectors that use polyadenylation signals without these ATTGTT sites continued to upregulate in this assay (T. Clarke and P. Donahoe, unpublished data). Until the reporter vectors are purged of ATTGTT sites, it will not be clear whether SRY acts directly at SRY as suggested by DNase footprint analysis, or whether SRY induces expression of another factor which in turn acts to upregulate MIS transcription. It will also be interesting for future experiments to address the role of SRY in transgenic mice, since the detailed chromatin configuration of the MIS promoter in host chromosomes compared with the plasmids used in transient transfection analysis might be different enough to show distinct effects on MIS transcription.

Analysis of MIS initiator region (MIS-InR), 22 to +8, indicates that it is capable of mediating SRY-induced transcription. Mutation of the initiator by randomization of the 32-bp nucleotide sequence, or replacing the 3' 16 nucleotides of the MIS-InR, which contains the transcription start site, by a heterologous sequence from phage  reduced transcription to control levels seen with empty reporter vector. In contrast, replacing the 5' 16 nucleotides of the MIS-InR with a heterologous sequence from phage  had little effect. We have not been able to detect direct DNA binding of SRY to the MIS-InR region; instead, SRY may undergo protein-protein interactions with basal transcription factors bound to the InR in vivo, or might induce synthesis of basal transcription factors to high levels resulting in a secondary increase in MIS transcription levels.

We previously demonstrated that SRY could bind to and protect sites in the rat aromatase P-450 promoter (167). Although aromatase is important to female sexual differentiation, aromatase transcription is seen in mouse fetal ovaries and testes without sex specificity (245); on the other hand, no aromatase activity is detected in rat ovaries during fetal life (440) and expression of aromatase in the brain appears to be highly sex specific during fetal life (193). Although human SRY upregulated a 534 to +100 bp rat aromatase-chloramphenicol acetyltransferase promoter containing the SRY footprinted region (Fig. 5A), and cannot regulate a 227 to +100 bp aromatase-luciferase promoter lacking this region (Haqq et al., unpublished data), further work is required to see if precise mutation of the SRY footprinted region is the site of promoter regulation. Control of aromatase activity at the protein level is also possible, since it has been known for many years that MIS can downregulate aromatase enzymatic activity (423). Contributions of transcription, translation, and stability of mRNA and protein to this phenomena have not yet been distinguished.

This assay developed for human SRY and the MIS promoter (168) along with assays developed for SRY and the fra-1 promoter (64) and for mouse Sry and reporters containing the consensus binding element (103) are the first assays available to test function of the SRY molecules. We have used our assay to demonstrate that SRY mutations associated with sex reversal in humans demonstrate defects in transcription activity of the MIS promoter (see Figs. 2 and 6).

Recent cloning of the mouse (296), rat (166), and chicken (111) MIS genes, along with promoter sequencing upstream of the human and bovine genes (54, 55, 157), now permits precise studies of the regulation of MIS expression in efforts to define transcription factors whose activity may work with SRY or downstream of it in the pathway leading to male sexual differentiation. The 5'-region upstream of MIS is notable for its close proximity to the transcription unit for a spliceosome-associated protein (SAP62), which is transcribed with the same orientation as the MIS gene (100). In mouse, SAP62 transcripts use a polyadenylation signal at 424 relative to the MIS ATG start codon; in humans, SAP62 terminates at 789. Use of an approximately 300 mouse MIS promoter fragment to drive lacZ expression in transgenic mice is sufficient to correctly initiate MIS expression in males at the same time the endogenous gene becomes active, embryonic day 12 in mice (162). However, these MIS promoter-lacZ transgenes are turned off by embryonic day 15, whereas the endogenous gene continues to be expressed in males at high levels until at least embryonic day 18. This suggests that all the information for initiation of MIS expression is contained in a very compact minimal promoter region, but that additional factors needed for maintenance of MIS expression interact with sites upstream of or within the SAP62 gene, or within MIS introns, exons, or 3'-sequence.

A MIS promoter element termed MIS-RE-1 (368), or M2 (human MIS nucleotides 114 to 93 bp; Refs. 159, 169), matches the consensus sequence for an SF-1 binding site 5'-CCAAGGTCG. SF-1 rendered constitutively active by deletion of its hormone binding domain can activate the MIS promoter in Sertoli cells (368); it was proposed that a tissue-specific SF-1 ligand is present in Sertoli cells, since SF-1 responses were stronger in Sertoli cells than in extragonadal HeLa cells. Intact SF-1 in a male urogenital ridge cell line transcriptionally repressed the MIS promoter through this site (168).

Although the above results suggest that SF-1 may play a role in MIS transcription, SF-1 is present in the ovaries from embryonic days 13.5 to 16.5 (199), a period when ovaries never express MIS (296). Thus, if SF-1 is responsible for regulation of the MIS promoter in vivo, additional tissue and sex-specific factors, possibly including the SF-1 hormone binding domain ligand, must preclude SF-1 from acting at this time. Homozygous disruption of the SF-1 gene in mice leads to a phenotype where gonad and adrenal development fails before the stage of MIS transcription (257), so this knock-out cannot help determine whether SF-1 is involved in MIS transcription in vivo. An alternative hypothesis to explain these findings is that additional factors can bind at the MIS-RE-1/M2 site to cause MIS transcription.

The M2/MIS-RE-1 sequence containing the binding site for SF-1 (6 bp) is part of a larger region of evolutionary conservation across species (12 bp), and when the entire 12-base element is used for gel-shift experiments using testicular or ovarian nuclear extracts, the pattern of specific DNA-binding activities detected is complex. In addition to three male specific bands of distinct migration, female-specific ovarian gel shifts are present concomitant with ovarian MIS expression (Fig. 7A). We believe these gel shifts are due to the granulosa cell compartment, since they were detected in nuclear extract from a clinical granulosa cell tumor specimen (Haqq et al., unpublished data). Our laboratory is in the process of purifying an ~20 kDa M2-DNA binding activity (Fig. 7B) distinct in molecular weight compared with 51-kDa SF-1 from newborn bovine testes. This M2 binding protein might modify SF-1's transcription activity, or might exert its transcription influence independent from SF-1. It is equally important, however, to rule out that this NH₂-terminally blocked band (Haqq et al., unpublished data) is not merely an enzymatically induced fragment of SF-1.

View larger version (61K): [in this window] [in a new window]   FIG. 7.   A: electrophoretic mobility shift assay of nuclear extract prepared by microdissection of tissues labeled above gel lanes at indicated times during rat embryogenesis, performed by standard protocol (128) in 5% 38:2 acrylamide-bis-acrylamide low cross-linking gels. Nuclear protein extract (10 µg), quantitated by the Bradford-Lowry assay, was used in each lane; probe was rat M2 5'-CAggCACTgTCCCCAAggTCACC annealed to its 32P-labeled complement. Control gel shifts were performed using a probe for ubiquitous protein H2TF1 (14a) and showed a robust H2TF1 gel shift in all cases, confirming integrity of all extracts. B: purification of a protein of ~22 kDa binding to M2 MIS promoter DNA element. Nuclear extracts were prepared from newborn bovine testes by modification of an established procedure (89) making nuclei in a Waring blender. Crude nuclear extract was fractionated on heparin-agarose and subsequently salt eluted, dialyzed, and subjected to 2 rounds of purification on a DNA affinity column (223) made of a support matrix bound to multimerized M2 DNA sites. Coomassie-stained 15% SDS-PAGE of affinity-purified M2 DNA binding protein (5, 10, and 25 ml) were compared with known quantities (10, 20, and 100 pmol) of the 85-amino acid SRY HMG box expressed in Escherichia coli.

	VII. SRY TRANSCRIPTION UNIT: WHAT REGULATES SRY EXPRESSION?

Top Previous Next References

Full-length copies of SRY have been obtained from mouse, human (372), rabbit (445), bovine (79), and rat (Haqq et al., unpublished data) species. In addition, partial clones of SRY from many other species have been obtained by PCR amplification of their HMG DNA binding domains (72, 124). Several notable species-specific differences exist in the organization of the SRY locus, the structure of the gene itself, and the function of the expressed protein in transgenic mice and in biochemical assays. The putative promoter region of the SRY gene contains multiple potential SP1 binding sites, as well as possible autoregulatory binding sites for SRY itself. However, the human SRY promoter lacks a TATA box, which is present in all other species. Transcription start sites have been mapped for human SRY by two groups. One group expressed human SRY in a mouse cell line finding two different start sites for SRY transcription at 137 and 78 nucleotides upstream of the ATG (384). In contrast, mapping human SRY start sites using mRNA in tissues in which SRY is normally expressed reveals start sites at 91 and 501 (63). The discrepancy between start sites found in transfection experiments compared with endogenous RNA from tissue sources suggests that tissue- or species-specific factors may be involved in transcription at the SRY locus.

	VIII. AUTOSOMAL AND X-LINKED GENES INVOLVED IN SEX DETERMINATION AND DIFFERENTIATION

Top Previous Next References

Finding that Sry as a transgene can make XX female mice develop as males indicates that most if not all downstream genes in the sex determination pathway are autosomal. Two autosomal gene products, testosterone and MIS, are well-characterized modulators of the sexual phenotype and are discussed in separate sections below.

DAX-1, an orphan member of the steroid hormone transcription factor family related to steroidogenic factor 1, is responsible for adrenal hypoplasia congenita (297, 459). In patients with adrenal hypoplasia congenita, hypogonadotrophic hypogonadism results from the defect in pituitary-gonadal-adrenal endocrine axis. DAX-1 maps to the same region of the X chromosome that is associated with male to female sex reversal when it is duplicated (26, 308, 360, 381); other duplications of Xp that do not give rise to sex reversal were used to limit the critical interval to Xp21.3-22.11 (16, 241, 299). The sex reversal function is named dosage-sensitive sex reversal (DSS) (297), and it is possible that DSS and DAX-1 are identical. However, other genes are present in the critical region of Xp21, so further experiments are needed to address the timing and tissue specificity of DAX-1 expression, whether the gene is subject to X-inactivation, and whether duplicated DAX-1 in transgenic animals gives the same phenotype as DSS. It will clarify matters greatly if molecular target genes downstream of DAX-1 are identified. Masafumi and Jameson (264) recently discovered a direct interaction between DAX-1 and SF-1 which resulted in inhibition of expression of SF-1 heterologous and homologous reporter constructs.

Every normal person, XX or XY, has the potential to develop both male and female reproductive structures. During fetal life either the Wolffian or the Müllerian duct must develop while the other must regress in response to differentiation of bipotential gonads into either testes or ovaries. Professor Alfred Jost, in the late 1940s, first discovered that the influence of the testis on male development required two hormones (219). Jost surgically removed the gonads of fetal rabbits during various stages of sexual differentiation. Removal of either testes or ovaries before, or at the outset of gonadal differentiation, resulted in female somatic development, the Wolffian ducts regressed instead of going on to form the vas deferens and seminiferous tubules, while the Müllerian ducts persisted and developed into the oviducts, uterus, and upper vagina. When Jost transplanted a testis into a female fetus, Müllerian ducts were inhibited and Wolffian ducts persisted. The converse transplant of ovary to male had no effect. Finally, when a crystal of synthetic androgen, instead of an entire testis, was placed in the abdomen of a castrated male, Wolffian development was supported but the Müllerian system failed to regress. Therefore, Jost (219, 220) postulated the existence of two hormones, one an androgen that would support male Wolffian duct development and a second called l'hormone inhibitrice or the Müllerian inhibitor, responsible for regression of developing female Müllerian structures.

L'hormone inhibitrice, later called MIS (217) or anti-Müllerian hormone (AMH), was subsequently found to be a Sertoli cell protein (32). It was purified to homogeneity (38, 39, 331), and both bovine and human MIS genes were cloned (54, 332). Müllerian inhibiting substance is one protein of a large superfamily whose members play key roles in development, wound healing, suppression of tumor growth, and feedback control of the pituitary-gonadal hormone axis. Related proteins include transforming growth factor- (84), nodal (461a), activins (190, 250, 277, 411), inhibin (265), decapentaplegic and related molecules (90, 315), Vg-1 and related molecules (214, 359, 437), osteogenic proteins (313, 314), bone morphogenesis factors (452), cartilage-derived morphogenetic proteins (59), and growth/differentiation factors (189, 247, 277, 281). Expression studies of MIS revealed synthesis of MIS not only in fetal and neonatal males, but also in females, where MIS is first synthesized in postnatal ovarian granulosa cells that surround and nourish developing oocytes, and then produced in dynamic fashion with rising levels throughout the follicular phase, peaking just before ovulation, throughout female reproductive life (28, 296, 386, 409). Müllerian inhibiting substance is an inhibitor for several processes, including oocyte meiosis (387, 408), the production of aromatase in the fetal ovary (86, 423), and levels of aromatase and lutenizing hormone receptor in the postnatal ovary (87), and MIS blocks epidermal growth factor-induced synthesis of progesterone in postnatal granulosa cells (227), suggesting that it plays a role in the control of granulosa cell steroidogenesis.

Cloning of MIS with subsequent investigations has confirmed Jost's foresight and enabled novel observations of MIS activities that suggest additional roles. Overexpression of MIS in transgenic female mice resulted in a phenotype characterized by regression of Müllerian ducts as predicted by Jost's experiments (20). However, the ovaries of these female mice exposed to supraphysiological levels of MIS became depleted of germ cells and formed structures reminiscent of seminiferous tubules. Mutation of the MIS gene is associated with the persistent Müllerian duct syndrome (52, 200, 233). Surprisingly, many of these mutations are located in the NH₂ terminal of MIS, whose role in bioactivity is not fully understood (447). Enzyme-linked immunosorbent assay detection of mutant protein levels are lower than wild type, perhaps indicating that these mutations affect translatability or stability of the protein; however, the same result would hold if the mutations interfered with recognition by the antisera, so further work will be required to show the precise molecular mechanism for these mutations leading to the persistent Müllerian duct syndrome.

One-fourth of male mice exposed to supraphysiological levels of human MIS showed feminized testes and external genitalia (20). Thus MIS may play a role in the formation of gonadal and external genital structure, perhaps by affecting expression or function of the androgen receptor in target tissues. Catlin et al. (56, 57) have shown that MIS decreases levels of disaturated phosphatidlycholine, the major component of surfactant in fetal lungs, which led to the hypothesis that MIS may contribute to the male preponderance in newborn infants of respiratory distress syndrome. Finally, the role of MIS as a negative growth regulator in fetal life suggested that MIS might have antitumor activity against neoplastic cells, especially those derived from Müllerian duct structures. These effects were apparent for ovarian and uterine carcinoma cell lines (61, 98) and primary tumors (133). In whole animal experiments conducted in nude mice, recombinant human MIS had potent inhibitory activity against the growth and reduced the number of metastases of an ocular melanoma (323).

Homozygous deletion of the MIS gene resulted in male mice with retained Müllerian ducts. Female MIS / mice were fully fertile, whereas most males were infertile due to anatomic blockage of sperm egress; in vitro fertilization with their sperm resulted in production of normal offspring (20). Newborn males showed no respiratory distress or developmental lung defects (