Tenuous transcriptional threshold of human sex determination. II. SRY exploits water-mediated clamp at the edge of ambiguity

Y-encoded transcription factor SRY initiates male differentiation in therian mammals. This factor contains a high-mobility-group (HMG) box, which mediates sequence-specific DNA binding with sharp DNA bending. A companion article in this issue described sex-reversal mutations at box position 72 (residue 127 in human SRY), invariant as Tyr among mammalian orthologs. Although not contacting DNA, the aromatic ring seals the domain’s minor wing at a solvent-exposed junction with a basic tail. A seeming paradox was posed by the native-like biochemical properties of inherited Swyer variant Y72F: its near-native gene-regulatory activity is consistent with the father’s male development, but at odds with the daughter’s XY female somatic phenotype. Surprisingly, aromatic rings (Y72, F72 or W72) confer higher transcriptional activity than do basic or polar side chains generally observed at solvated DNA interfaces (Arg, Lys, His or Gln). Whereas biophysical studies (time-resolved fluorescence resonance energy transfer and heteronuclear NMR spectroscopy) uncovered only subtle perturbations, dissociation of the Y72F complex was markedly accelerated relative to wild-type. Studies of protein-DNA solvation by molecular-dynamics (MD) simulations of an homologous high-resolution crystal structure (SOX18) suggest that Y72 para-OH anchors a network of water molecules at the tail-DNA interface, perturbed in the variant in association with nonlocal conformational fluctuations. Loss of the Y72 anchor among SRY variants presumably “unclamps” its basic tail, leading to (a) rapid DNA dissociation despite native affinity and (b) attenuated transcriptional activity at the edge of sexual ambiguity. Conservation of Y72 suggests that this water-mediated clamp operates generally among SRY and metazoan SOX domains.


Table of Contents
Residue Numbering. Clinical mutations in human SRY are ordinarily given in relation to residue numbers in the full-length protein. For clarity, consensus positions in the HMG box are also given; e.g., Y127 in full-length SRY is residue 72 in an HMG box consensus (1).

NMR detection of bound water
In favorable cases NMR spectroscopy can provide evidence for bound water molecules at specific sites in proteins or at protein-ligand interfaces. The behavior of a water molecule can be investigated by looking at NMR properties reflecting in exchange processes at the micro-to millisecond timescale as in principle probed by four different nuclei: these are the three isotopes of hydrogen: proton ( 1 H), deuterium ( 2 H), tritium ( 3 H), and oxygen 17 O. Of these, the most widely and extensively studied nucleus is 1 H; only limited applications of 2 H and 17 O NMR studies of water have been described. Hydration waters that exchange with the bulk water on a time scale of seconds was deduced from 18 O tracer experiment (2). The dispersion of the water 1 H longitudinal relaxation rate address micro-to-millisecond residence time (3).
NMR is sensitive to exchange of water molecules in and out of a specific hydration site and can within certain ranges be used for quantitative measurements of exchange lifetimes (4). NMR experiments for the detection of intermolecular NOEs between a proton in a protein (or within DNA) with water include: (a) Water suppression using pair of spin-lock pulses, (b) Watergate and Diffusion filter; (c) Selective water excitation and Dipolar field effects also employed for NOEs between water and biomacromolecules (5). The sign of the NOE cross relaxation rate changes for water residence times in the range 0.1-1ns. Thus, NOE measurements provide a tool to distinguish "slow" and "fast" water molecules on this time scale.
In the original studies of the homeodomain-DNA complex by Wuthrich and colleagues (4, 6, 7), experimental evidence for interfacial water molecules was indirect: cavities and crevices at or near the protein-DNA recognition surfaces. Interpretation of these gaps was enabled by MD simulations. This example motivated efforts to develop and apply direct NMR methods to detect bridging water molecules. Clore et al subsequently reported that NOEs between protein protons and surface hydration water can be quenched when the effective correlation time for positional rearrangement of the water protons (relative to the protein surface)-determined either by chemical exchange or by independent rotational motions of the water molecules-is much shorter than the rotational correlation time of the protein (8).
NOEs and ROEs between bound water and protein protons attached to 13 C or 15 N can be observed by recording 12 C-filtered two-dimensional (2D) H2O-ROE/NOE-1 H-13 C or H2O-ROE/NOE-1 H-15 N heteronuclear single quantum coherence (HSQC) spectra (9). Molecular dynamics simulations were found to be critical to the interpretation of NOE data in a zinc finger-DNA complex (10). Only 6 protons failed to show intermolecular NOEs to solvent showed nearby long-resident water molecules in the MD simulations.
The effect of internal hydration on protein structure and stability were investigated by Brunne et al. (6). The single interior hydration water molecule located in the loop of BPTI replaced by the serine hydroxyl group in the mutant BPTI (variant G36S). The slightly reduced stability of mutant can be accounted by the loss of a hydrogen bond due to the fact that a hydroxyl moiety can donate only one hydrogen atom to potential acceptor atoms, whereas a water molecule can donate two hydrogen atoms. MD simulations predicted a residence time of the water molecule within BPTI in the range 19-200 picoseconds (ps), with no correlation between solvent-accessible residue and their location (11). Thus, interaction between solute-water, water-water and entropic effect is more favorable than charged or non-polar solute atom. Interior hydration water molecules exchange with the bulk water on a millisecond time scale or faster, specifically between 200 μs to 1 ns for BPTI as calculated by NMR (4). MD calculation of Antennapedia homeodomain-DNA complex suggest residence time for interior waters are on the nanosecond time scale (~600 ps) which is lower end of the range determined by NMR (100 times longer than the lifetime of a surface water molecule (12)).
Pertinent to studies of mutations at protein-DNA interfaces, the NMR-derived solution structure of a variant complex (a mutant Antennapedia homeodomain-DNA complex with Cys39 replaced by Ser) provided a foundation for MD simulations in a water bath (see below), which suggested that the mutation affects the pattern of bound water molecules at the variant interface (12).
In addition to isolated proteins and the protein-DNA systems highlighted in this study, applications have also been reported to water channels (13), water-filled silica nanopores (14), and mineral-water interface (15). The changes in water-fiber interactions, for example, could be probed by various NMR relaxation parameters, double-quantum filtered (DQF), and 1D and 2D translational-diffusion experiments (16). A restricted diffusion and anisotropy of water selfdiffusion could be studied by one-and two-dimensional pulse-field-gradient NMR (17).
MD simulation (500 ps) of human SRY-DNA predict only one water mediated hydrogen bond between SRY and DNA; i.e., Asn10-C4 and T14 along with other 5 salt bridge (18). Thus, it is hydrophobic interaction which facilitate protein-DNA binding. In the present case, we can see more than one water mediated hydrogen bond between SRY-DNA. The possible reason for this difference is duration of MD calculation.
Future direction of the human SRY-DNA complex will be to explore the idea of detecting Tyr72 bound water. Selective 13 C-labeling of Tyr72 might be useful to observe 13 C-edited NOE's to water held at the interface. In contrast, 13 C-labeled Tyr74 will lack such NOE. In a variant domain-DNA complex, a 13 C-edited NOE for Phe72 with water might not be observable or very weak as the distance would be close to 4 Å. Middle, mutations that localize to either the major wing hydrophobic core or a DNA-dependent mini-core (22). Mutations in parenthesis at position Tyr 72 are the site of investigation in this and our companion paper (23). Right, mutations that affect proper nuclear localization (24,25) or nucleo-cytoplasmic shuttling of human SRY (26). Human SRY HMG box-DNA structure PDB entry 1J46 (1).                The structure of SOX18 HMG-DNA is shown in a cartoon rendering with the protein in green and DNA in sand. Tyr72 (consensus numbering) in the SOX18 structure, along with residues 2-4 including I3 are shown in a green stick rendering and two nucleotides are shown in gray (the color refers to C atoms, O are red, nitrogen blue, and P orange). The hydrogen bonds between the bridging water molecule, the protein, and the DNA are shown as black dashed lines. (B) A closeup of the hydrogen-bonding network involving the bridging water molecule is shown for SOX18. The water molecule is shown as a blue sphere. (C) HMG boxes for SOX18 (green), SOX2 (cyan), and SOX11 (yellow) are shown superimposed in cartoon renderings with the DNA from the SOX18 structure. In these structures the Y72 equivalent residues shown in a stick renderings superimpose exactly. Close-up stick renderings for SOX2 (D) and SOX11 (E) show that the bridging water molecule interactions are identical in each of these structures. PDB entries are: (Sox18) 4Y60, (Sox2) 1GT0, and (Sox11) 6T78 as described in references (28,31,32).         Values are the average of two independent measurements a Width is full width at half-maximum (FWHM) b Global (χ 2 ) co-optimized global fitting of all FRET data to obtain best single distance distribution.
The DNA site is 5΄-TCGGTGATTGTTCAG-3΄ and complement. b Δδ and b Δδ is defined as the difference between chemical shifts in the wild-type ad variant respectively. Primes in column indicate lower strand SRY-p complex and the free DNA. d ΔΔδ is defined as the difference in chemical shifts between wild-type and mutant complexes.     Table generated from Human Genome Mutation Database (HGMD) https://digitalinsights.qiagen.com/products-overview/clinical-insights-portfolio/human-genemutation-database/ See also footnote 1. Footnotes 1 Homologous clinical mutations in SRY and SOX genes give rise to different clinical syndromes depending on the respective biological function of the specific family members. Whereas mutations in SRY, for example, yield DSD phenotypes, mutations in the HMG box of SOX10 are associated with Waardenburg syndrome type II and type IV (also designated Waardenburg-Shah syndrome), characterized by impaired hearing with changes in skin, hair, and eye coloring (97). Mutations in the HMG box of SOX9 cause campomelic dysplasia, characterized by abnormalities of cartilage and bone in variable association with XY DSD (98). Developmental abnormalities of bone are also associated with mutations in SOX4 and SOX11 (99). The diverse SOX-associated genetic syndromes are collectively designated "SOXopathies" (100).