The male phenotype in therian mammals is determined by Sry¹ (designated SRY in primates), an architectural transcription factor encoded by the pseudo-autosomal region of the Y chromosome ( Figure 1A ) (9, 10). In late embryogenesis, SRY initiates testicular differentiation of the bipotential gonadal ridge (10, 11). A downstream gene-regulatory network (GRN) has been investigated in mouse models (12–15), identifying autosomal gene Sox9 as the principal target of Sry in embryonic pre-Sertoli cells (15). Sry binds to specific regulatory DNA sites in the upstream region of Sox9 (or human SOX9), functioning as testis-specific enhancers (designated TES and Enh13) (16–19). Whereas Sry/SRY represents an evolutionary innovation in therian mammals, Sox9 is broadly conserved among vertebrate sex-determining (VSD) pathways: an example of such pathways “growing backwards” in evolution (20, 21).

Sox9 (SOX9 in primate nomenclature) is an architectural transcription factor (22–24) and a member of the general class of eukaryotic SOX factors defined by an Sry-related high-mobility-group (HMG) box (25), a conserved motif of DNA binding and bending ( Figure 1B ) (26). Following its transcriptional activation by Sry in the embryonic gonadal ridge (red box in Figure 1A ), a Sox9-dependent GRN orchestrates programs of cell-cell communication, migration, and differentiation leading to testis formation ( Figure 1A ; upper regulation pathway). Mutations at successive levels of this pathway are broadly associated with disorders of sexual development (DSD). Endocrine functions of the fetal testis, once formed, direct subsequent somatic virilization (via secretion of testosterone) with regression of the female anlägen (via glycohormone Müllerian Inhibiting Substance [MIS/AMH]) as outlined in Figure 1A (lower pathway). Genotype-phenotype relationships in DSD reflect the molecular logic of this pathway (15, 27).

Evidence for the critical role of SRY in human testis determination was provided by the discovery of diverse clinical mutations leading to 46, XY gonadal dysgenesis with female somatic phenotype (Swyer Syndrome) (15); for review, see (28)). Such mutations usually occur de novo in paternal spermatogenesis. Swyer variants exhibit a broad range of biochemical and biophysical perturbations (15, 27). Whereas de novo (sporadic) mutations in SRY are typically associated with marked impairment of specific DNA binding, a distinct class is inherited (29–31). The latter mutations are by definition compatible with alternative developmental outcomes: testicular differentiation leading to virilization (fertile 46, XY father) or nascent ovarian differentiation leading to gonadal dysgenesis with female somatic phenotype (sterile 46, XY daughter) (32–34). Inherited mutations in SRY thus provide “experiments of nature” that define mechanistic boundaries of genetic function. These boundaries may pertain to one or more biochemical activities, such as specific DNA binding, nucleocytoplasmic shuttling, post-translational modification, or proteasomal degradation (30, 31). Swyer mutations cluster in (but are not restricted to) the SRY HMG box (15, 27). Mutations in the HMG box of SOX9 can also cause XY sex reversal (22).

The sequence-specific HMG box is an L-shaped domain containing an N-terminal β-strand, three α-helices (α₁, α_2, and α₃), and C-terminal basic tail ( Figure 1C ). Packing of the β-strand against the C-terminal portion of α₃ defines the minor wing in the HMG box, whereas α₁, α_2, and the N-terminal portion of α₃ comprise the major wing. Unlike structure-specific homologs (such as the boxes of HMG-1 and HMG-D (35, 36), which are generally well ordered in the absence of DNA (35, 37, 38), SRY’s minor wing is partly disordered: N- and C-terminal strands are unfolded, and the major wing is molten (7). DNA recognition is mediated within an expanded DNA minor groove; such binding in turn stabilizes the canonical L-shaped domain structure. The specific protein-DNA interface is remarkable for “wedge insertion” of nonpolar side chains (HMG-box positions 9, 12, 13, and 43; residues 64, 67, 68, and 98 in human SRY) in addition to canonical hydrogen bonds and salt bridges (39, 40). Specific contacts in the SRY-DNA complex are conserved within the homologous SOX co-crystal structures (41, 42). The SRY/SOX-DNA interface is extended by a C-terminal basic tail (blue in Figure 1B ), which remains in the minor groove (7) to augment the kinetic stability of the bent complex (43).

The present study has focused on three Swyer mutations at a conserved site at the junction between helix α₃ and the basic tail: box position 72 and residue 127 in human SRY. This junction is conserved among Sry orthologs ( Figure 1D ) and SOX homologs ( Figure 1E ). Two of the mutations occurred de novo (Y127C and Y127H) whereas the third was inherited (Y127F). The wild-type (WT) side chain is of dual structural interest. On the one hand, one face of the aromatic side chain buttresses a DNA-binding surface ( Figure 2 ) via a cluster of aromatic residues engaging V60 (box position 5): H120, Y124, and Y127 (respective box positions 65, 69, and 72). On the other hand, the para-hydroxyl group of Y127 faces the DNA phosphodiester backbone, but is too distant for direct engagement (6.7 Å from the nearest DNA atom; a non-bridging phosphodiester oxygen atom) (7). Whereas the de novo mutations would be expected to perturb the aromatic cluster—in turn perturbing an overlying DNA-binding surface—how and to what extent the inherited Y127F variant impairs SRY function poses an intriguing question. Although Tyr and Phe are similar aromatic amino acids and often interchangeable among homologous proteins, a previous study has reported that Y172F blocks detectable specific DNA binding by the variant SRY domain (34). If valid, this finding would pose a seeming paradox: how might the phenotype of the fertile 46, XY father be rationalized?

The present study sought to resolve this issue through comparative cell-based and biochemical studies of the three Swyer mutations (Y127C, Y127H, and Y127F). Whereas the two de novo mutations were found to markedly impair SRY function, the effects of Y127F are modest: indeed, substantial gene-regulatory function was retained in cell-based models of SRY function (30, 31) in accordance with its inheritance. Only subtle changes were observed in the variant domain’s thermodynamic stability, specific DNA affinity, and induced DNA bend angle. Based on homologous SOX-DNA crystal structures (41, 42, 46), we envision that the Y127 para-OH group anchors a water molecule at the protein-DNA interface and propose that the loss of a single water-mediated hydrogen bond attenuates the male switch at the edge of ambiguity. Our companion study in this issue (47) provides biophysical evidence that removal of the Y127 para-OH group “unclamps” the sharply bent protein-DNA complex to enhance long-range conformational fluctuations and accelerate its dissociation.

Together, our studies uncovered an otherwise cryptic contribution of a conserved aromatic residue to the function of human testis-determining factor SRY. This residue’s structure-activity relationships are likely to extend to the SOX family of architectural TFs (26) and more broadly to a metazoan superfamily of sequence-specific DNA-bending domains (35). Molecular mechanisms of a rare syndrome—Swyer syndrome as a probe of human genetic variation—may thus illuminate general principles of developmental gene regulation and its evolution.

Two SRY-responsive mammalian cell lines (embryonic rat pre-Sertoli XY CH34 (48)) and human prostate-cancer-derived LNCaP (62, 63)) provided models for functional analyses of wild-type (WT) and mutant SRYs in culture (see Supplemental Discussion, Supplemental Figures S1–S3 , and Table S1 ) (30, 63). The independent origins of these cell lines were intended to mitigate possible line-specific or species-specific effects unrelated to their shared SRY responsiveness. Following expression of WT or variant SRY by transient transfection (with plasmid dilution to control expression level (30)), SRY-directed transcriptional activation of Sox9/SOX9 was evaluated in their respective native chromosomal contexts. Employing qPCR, this assay measured the time-dependent accumulation of mRNAs encoded by downstream genes in the SRY-SOX9 GRN (10). Level of expression of the transfected SRYs was set in the physiological range (10³-10⁴ mean molecules per cell) by an empty-plasmid dilution or Tet-on expression protocols (see Supplemental Discussion , Supplemental Figure S4 and Supplemental Table S2 ) (30, 64). Because mutations may in principle affect expression level at a given plasmid-dilution ratio or Tet concentration, functional studies were repeated on chemical proteasome inhibition to equalize WT or variant SRY expression (below).

Occupancies of SRY binding sites within corresponding far-upstream testis-specific enhancer elements of Sox9/SOX9 (17) were probed by ChIP (17). The protein binds within the core elements of two testis-specific enhancer elements (designated TESCO and Enh13, respectively; Figure 3A ); results are summarized in Figure 3B . The three Swyer substitutions were evaluated at SRY position 127 (an inherited mutation Y127F, de novo clinical variants Y127C and Y127H), and negative controls by expression of I68A (box position 13), in which a known clinical mutation at the “cantilever” position of the SRY HMG box markedly impairs its specific DNA binding (65). In both cell lines, the SRY variants at position 127 exhibited, to varying extents, reduced TESCO/Enh13 enhancer occupancies. Whereas inherited Y127F SRY preserved ca. 60% of occupancy relative to WT (as quantified by ChIP-qPCR; Figure 3B ), de novo clinical variant Y127C and Y127H exhibited ~20% occupancy in both cell models ( Figure 3B ). As expected, control mutation I68A blocked detectable binding of the variant SRY at either the TESCO or Enh13 enhancers.

Reducing the formation of testis-specific enhanceosomes directly impairs Sox9 gene activation ( Figure 3C ). Further, because SOX9 is the principal factor target of SRY, and SOX9 in turn activates the downstream GRN ( Figure 1A , sequential to or in synergy with SRY), specific upregulation of Sox8 was reduced twofold on transient transfection of Y127F SRY (relative to WT) and more markedly by the two de novo mutations. Similarly, sequential activation of downstream genes prostaglandin D2 synthetase (Ptgd2) and fibroblast growth factor 9 (Fgf9) was also reduced by all three mutations, whereas in control studies transient expression of SRY did not affect mRNA accumulation of non-sex-related Sox genes or unrelated housekeeping genes ( Figure 3C ).

We next analyzed subcellular localization. SRY contains two nuclear localization signals (NLS), one near the N terminus of the HMG box and the other near its C terminus (66–68). To test whether the mutations decreased the nuclear localization since position 127 is near the C-terminal NLS motif ( Figure 1B ), we tested its nuclear import by immunocytochemistry approach ( Figure 4A ) and cellular fractionation assay ( Figure 4B ). Immunostaining and subcellular fractionation studies (validated by respective nuclear and cytoplasmic markers YY1 and tubulin, respectively) demonstrated that the mutant SRYs retained predominantly nuclear localization, like WT ( Figures 4A, B ). In these assays, control Swyer mutations R62G (perturbing N-NLS) and R133W (C-NLS) impaired nuclear entry as expected. To test whether the mutations decreased testis-specific enhancer occupancies by enhancing off-target DNA binding in the same chromosomal regions, we analyzed potential flanking sites of ChIP occupancies in CH34 cells ( Figures 4C-E ). In addition to the two validated testis-specific enhancer elements in the Sox9 gene, there are five additional chromosomal segments containing consensus or near-consensus SRY-binding DNA sequences ( Figure 4C ). Whereas WT SRY binds within Enh13 and TESCO enhancers much more strongly than other potential DNA binding sites ( Figure 4D ), mutations perturbed enhancer occupancy without increasing off-target DNA binding ( Figure 4E ). These findings suggest that the mutations do not confer altered or relaxed DNA sequence specificity.

The cellular lifetime of SRY, its mean expression level, and rate of proteasomal degradation define additional biochemical properties affecting mean transcription-factor concentration and hence the extent of target gene activation (31, 69). We thus probed the proteolytic stabilities ( Figure 5A ). Cellular turnover of the expressed epitope-tagged SRY proteins was evaluated following treatment of cycloheximide, a chemical inhibitor of translation ( Figure 5A ). The Y127C variant was found to be more rapidly degraded than WT SRY or the other variants ( Figure 5A ). Such differential degradation could be circumvented through the addition of chemical proteasome-inhibitor MG132 to “rescue” expression of Y127C SRY. Such equalization of intracellular protein concentrations in large part rescued the gene-regulatory potency of Y127C SRY in the two cell lines ( Figure 5B ). No effect of MG132 was observed in the corresponding study of WT SRY or the other variants.

Plasmid-dilution ratio provides a classical approach to control the intracellular concentration of SRY on transient transfection. Systematic exploration of plasmid-dilution ratios enabled the construction of a dose-response curve relating mean intracellular SRY concentration to fold-activation of the Sox9 target gene ( Figure 5C ). This input-output relationship exhibits switch-like behavior: the toggle range below which Sox9 expression declines (black arrow in Figure 5C ) is 1-4 x 10⁴ mean protein molecules per cell (gray strip). Because WT and Y127F SRY exhibit different gene-regulatory properties on a per-molecule basis (above), their respective dose-response relationships fall on different curves (black and orange arrows in Figure 5C ). Interestingly, Y127F SRY was found to be not equipotent to WT SRY even in the plateau region (from 10⁵-10⁶ mean protein molecules per cell), and its toggle range (arrow in Figure 5E ) was shifted to the right in seeming agreement with the greater sensitivity of the mouse Sry-regulated intersexual development. No dose-response relationship was observed in control studies of inactive SRY variant I68A (blue circles in Figure 5C ). The WT input-output relationship was replicated in a complementary Tet-on inducible system (shown in schematic form in Figure 5D ; see also Supplemental Figure S4 ) as shown in Figure 5E (overlapping gray circles vs. plasmid-dilution-derived red circles). That two independent methods to control SRY expression levels in cell culture gave indistinguishable results suggests that this input-output relationship is an intrinsic feature of the SRY-SOX9 regulatory axis.

Together, the above cell-based results indicated that de novo variants Y127C and Y127H are unable to activate downstream target gene Sox9/SOX9. Inherited allele Y127F by contrast retained partial gene-regulatory activity. To characterize the specific DNA-binding and bending properties of the variant HMG boxes in vitro, we next exploited a DNA-based FRET assay ( Figure 6A ): a protein-directed reduction in end-to-end DNA distance provided a signature of specific DNA binding and bending. This assay was based on differences in emission spectra reflecting donor intensity ( Table 1 , column 5). Spectra of WT and Y127F box-DNA complexes (respective black and orange traces in Figure 6B ) essentially overlap, indicating similar FRET efficiencies. FRET efficiency in the Y127C box-DNA complex was slightly attenuated whereas that in the Y127H complex was the most attenuated. Specific DNA affinity was also determined by FRET titration (29). Remarkably, the variant boxes each retained high affinity for the consensus DNA target site ( Figure 6C and Table 1 , column 4), quantitatively similar to WT ( Figure 6C , black traces). Specific DNA affinities and DNA bend angles are thus uncorrelated.

To further characterize protein-directed DNA bending, we employed permutation gel electrophoresis (PGE) to estimate bend angles (30). A consensus SRY target site was placed at varying distance from the 5’-end of a 150-bp DNA duplex ( Figure 7A ; target site in gold). Relative electrophoretic mobilities ( Figure 7B ) were analyzed as a function of flexure displacement (i.e., the distance of the DNA site from the 5’-end of the oligomer divided by 150 bp; Figure 7C ). These estimates are qualitatively consistent with relative FRET efficiencies ( Table 1 , column 6). Although each variant retained sharp DNA bending, the FRET- and PGE-based results together thus demonstrated that the identity of residue 72 modulates the precise degree of bending. As expected (based on the unbound minor wing’s flexibility (29, 30), CD-based assays demonstrated that the free domains each retained nativelike thermal stability ( Figure 7D ) and thermodynamic stability ( Figure 7F and Table 1 , columns 7 and 8). In addition, each variant box-DNA complex was stably maintained at 37°C at a protein-DNA concentration of 25 mM ( Figure 7E ). These in vitro data do not correlate with relative rates of proteasomal degradation in a cellular milieu (above).

The interaction between the SRY HMG box and its specific DNA target site (core consensus sequence 5’-ATTGTT and complement (71)) is remarkable for bidirectional induced fit: (i) protein-directed DNA bending within a widened and underwound minor groove (72), (ii) DNA-directed folding of the minor wing of the HMG box extended by its basic tail (73). Such binding also stabilizes the major wing (31, 69). In the free box, the N-terminal segment is detached from helix α₃; the tail is disordered. The minor wing’s mini-core stabilizes—and is stabilized by—an overlying protein-DNA interface. Critical to such bidirectional induced fit is the insertion of the aliphatic side chain of V60 (box position 5) into a cluster of aromatic side chains: H120, Y124 and Y127 (respective box positions 65, 69, and 72; Figure 1C ). Paired proline residues C-terminal to α₃ orient Y129 (box position 74) and successive basic residues to further stabilize the bent DNA complex. DSD mutations in these structural elements (variously perturbing mini-core packing, tail orientation, or direct side chain/DNA contacts) result in 46, XY gonadal dysgenesis with somatic sex reversal. These elements are invariant features of eukaryotic Sox boxes (8), including a junctional aromatic residue at box position 72 (Tyr or Trp; shaded gray rectangle in Figures 1D, E ); corresponding clinical mutations have been identified in the HMG boxes of several SOX factors (e.g., SOX2, SOX4, SOX5, SOX9, SOX10, SOX11, SOX17, and SOX18 (5, 46, 74–77)) in association with diverse phenotypes.² Such conservation suggests that the present structure-activity relationships, although probed only in SRY, will generalize to this broad family of architectural transcription factors (8).

Our study has focused on three clinical mutations at SRY residue 127 (consensus box position 72). Two occurred de novo (Y127C and Y72H): these presumably impair specific DNA-dependent folding of the minor wing through two mechanisms: loss of side-chain volume leading to a destabilizing cavity within the minor wing’s mini-core together with attenuation (Y127H) or loss (Y127C) of favorable aromatic-aromatic interactions. Y127C may also cause steric clash due to the larger atomic radius of sulfur relative to carbon at the γ side-chain position ( Figure 2 ). Whereas nativelike domain-DNA affinities are maintained (column 4 in Table 1 ), the extent of DNA bending is reduced (column 6 in Table 1 ). Estimates of SRY-directed DNA bend angles, as inferred from PGE ( Figure 7 ), yield ΔΔθ values of 6( ± 0.5)° (Y127H complex) and 9( ± 0.9)° (Y127C complex); these estimates are in qualitative accord with steady-state FRET spectra. Such attenuated DNA bending must represent a transmitted effect of the mutations as the WT side chain, and presumably also the variant side chains, is not in direct contact with the bent DNA site. These DNA bend angles each fall, however, within the range of WT species variation (60), so altered DNA bending is unlikely to be responsible for either the variant proteins’ loss of transcriptional activity in cell culture or associated DSD phenotypes (44, 45).

In embryonic pre-Sertoli cell line CH34 (40), Y127C SRY also underwent accelerated proteasomal degradation. Although the origins of such cellular turnover are not clear (the variant domain exhibits nativelike thermodynamic stability; column 7 in Table 1 ), restoration of native protein expression by chemical proteasome inhibition (histogram at right in Figure 5 ) partially rescued transcriptional activity (as monitored via autosomal target gene Sox9). Such partial rescue unmasks a subtle intrinsic defect in transcriptional activity. This modest perturbation (ca. twofold) stands in contrast to the more marked intrinsic defect in the transcriptional activity of Y127H SRY (ca. fivefold). These findings are counterintuitive as a nonpolar side chain (Cys) would in general be expected to be more perturbing than a weakly polar side chain (His) near a protein-DNA interface. To gain further insight into structure-activity relationships at box position 72, it would be of future interest to survey the specific enhancer-binding properties (in cells) and specific DNA-binding/bending properties (in vitro) of variant SRYs containing a diverse collection of substitutions at this site. The recent development of a CRISPR/Cas9-based mouse model of human SRY promises an opportunity to correlate such data with developmental outcomes in a mammal (55).

Of particular interest is inherited variant Y127F. This mutation, often observed as a conservative substitution within the evolution of protein families, poses a seeming paradox in light of (a) its chemical similarity, yet (b) dual compatibility with either somatic sex reversal (in the infertile XY daughter) and male development (in the proband’s fertile XY father). A previous study reported no detectable specific DNA-binding activity (34). If valid and extrapolatable in vivo, this finding would suggest either (i) that the father exhibited WT germ-line mosaicism; (ii) that the SRY defect could be compensated or bypassed depending on genetic background, or (iii) that the variant SRY gains a non-DNA-mediated function to sustain male gene regulation. Although cryptic mosaicism cannot be excluded, one or both of the latter two suppositions seem plausible. Because the original report relied on the gel mobility-shift assay (GMSA)—susceptible to kinetic artifacts³ (40, 80, 81)—our companion study undertook biophysical studies of the WT and Y127F domain-DNA complexes (47). The present FRET-based assays of protein-DNA affinities, an equilibrium technique, have demonstrated that the dissociation constants of WT SRY and Y127F SRY are indistinguishable within experimental error ( Figure 6 and Table 1 , column 4). Similarly, only a slight difference was observed in DNA bending by PGE or FRET ( Figures 6C , 7B ; column 6 in Table 1 ). Nonetheless, the mutation attenuates by ca. twofold both occupancy of embryonic testis-specific enhancer elements in the Sox9/SOX9 gene ( Figure 5 and Table 1 , column 3) and gene-regulatory activity (columns 2 and 3 in Table 1 ). Because these cellular perturbations are broadly in accordance with previous studies of diverse inherited Swyer alleles ( Supplemental Figure S5 and Table S3 (30, 31), cryptic germ-line mosaicism need not be invoked to rationalize the phenotype of the proband’s father.

Tyr and Phe differ only by the presence or absence of a para-hydroxyl group in corresponding aromatic rings. It is remarkable that such a subtle chemical modification in a transcription factor could underlie so dramatic a phenotypic outcome as DSD. We imagine that the para-hydroxyl group of Y127 functions in some way, not apparent in the present biochemical studies, to enhance the gene-regulatory activity of SRY. In our companion article in this issue (47) we propose that Y127 serves as an anchor point for a bound water molecule bridging the protein and DNA surfaces. Although bound water molecules were not sought or identified in the solution structure of the SRY box-DNA complex (as a technical limitation of NMR) (7), a subset of homologous SOX-DNA co-crystal structures exhibits exactly such a bridging water molecule ( Figure 8 ). We envision that this bound water molecule, presumably part of a solvation network at this interface, “locks” the bent protein-DNA complex to prolong its kinetic lifetime and enhance the precision of DNA bending. As foreshadowed by studies of the prokaryotic Trp repressor more than 30 years ago (84–86), bound water molecules can be critical elements in protein-DNA recognition. Although the specific DNA-bending properties of WT SRY and Y127F SRY are similar, the more clearly attenuated DNA bending by the Y127C and Y127H domains suggests that this solvated interface contributes to the precision of SRY-directed DNA bending. Such precision is probed in our companion study (47) through time-resolved FRET and distance-distribution analysis of bent DNA complexes (29, 43, 60).

Inherited Swyer Syndrome, although rare, is of conceptual interest as an experiment of nature probing the mechanistic boundaries of SRY function. The subtlety of such mutations (shared by a fertile XY father and sterile XY daughter) honors in the breach Waddington’s Principle of developmental canalization (see Box 1) (30, 31, 87, 88). That loss of a single Tyr-anchored water molecule from a protein-DNA interface, as posited here, might render a developmental program ambiguous and would be a striking demonstration of the “hesitant beginnings” of male sex determination (97) as uncovered in intersexual mice with strain-specific B6-Y^POS incompatibility (98–100). Inherited Swyer syndrome extends such mouse models to a human genetic switch at the edge of ambiguity. It would be of future interest to extend our ensemble analysis of mean input-output relationships in the SRY-SOX9 regulatory axis to single-cell analyses of stochastic gene expression (101).

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

Y-SC co-designed, performed, and interpreted the cell-based studies of SRY function. JR designed, performed, and interpreted the biochemical assays with purified proteins. MW conceived the project, integrated the results, provided structural interpretations, and oversaw preparation of the manuscript. MW is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors contributed to the article and approved the submitted version.

This work was supported in part by the INCITE Scholars Program of the Lilly Foundation and the Distinguished Professors Fund of Indiana University (MW).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.