HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 45, 43463-43473, November 8, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/45/43463    most recent M204616200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Narendra, U. Articles by Weiss, M. A. Search for Related Content PubMed PubMed Citation Articles by Narendra, U. Articles by Weiss, M. A. Social Bookmarking                      What's this?

Sex-specific Gene Regulation

THE DOUBLESEX DM MOTIF IS A BIPARTITE DNA-BINDING DOMAIN^*

Uma Narendra^§, Lingyang Zhu^§¶, Biaoru Li, Jill Wilken, and Michael A. Weiss^**

From the  Department of Biochemistry, Case Western Reserve University School of Medicine, Cleveland, Ohio 44106 and  Gryphon Sciences, Inc., South San Francisco, California 94080

Received for publication, May 10, 2002, and in revised form, August 22, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Sex-specific gene expression in Drosophila melanogaster is regulated in part by the Doublesex (DSX) transcription factor. Male- and female-specific splicing isoforms share a novel DNA-binding domain, designated the DM motif. This domain is conserved among a newly recognized family of vertebrate transcription factors involved in developmental patterning and sex determination. The DM motif consists of an N-terminal zinc module and a disordered C-terminal tail, hypothesized to fold on specific DNA binding as a recognition -helix. Truncation of the tail does not perturb the structure of the zinc module but impairs DNA binding and DNA-dependent dimerization. Chemical protein synthesis and alanine scanning mutagenesis are employed to test the contributions of 13 side chains to specific DNA binding. Selected arginine or lysine residues in the zinc module were substituted by norleucine, an isostere that maintains the aliphatic portion of the side chain but lacks a positive charge. Arginine or glutamine residues in the tail were substituted by alanine. Evidence is obtained that both the zinc module and C-terminal tail contribute to a bipartite DNA-binding surface. Conserved arginine and glutamine residues in the tail are required for high affinity DNA recognition, consistent with its proposed role as a nascent recognition -helix.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Sex determination in metazoans is regulated by diverse pathways (1). Formation of the mammalian testis and subsequent male development are initiated by Sry (2-5), a single-copy gene on the Y chromosome that encodes a high mobility group (HMG)¹ box (6, 7). Distinct genetic mechanisms operate in Drosophila melanogaster (8-14) and Caenorhabditis elegans (15, 16) wherein sex is determined by the X:autosome ratio, a process linked to X-dosage compensation (1, 17). Although the pathways of the fly and worm (Fig. 1A) seem otherwise unrelated, a cysteine-rich DNA-binding domain (the DM motif; Fig. 1B) is conserved within downstream transcription factors Doublesex (DSX) and MAB-3, respectively (18). This domain exhibits a distinctive pattern of cysteines and histidines (boxed in Fig. 2A) (18-20). These residues participate in two intertwined Zn²⁺-binding sites (Fig. 1B) (21). The DM motif defines a newly recognized family of metazoan transcription factors (22).

View larger version (31K): [in this window] [in a new window]   Fig. 1.   Overview of DM motif. A, sex determination pathways of the fruit fly (a) and the nematode C. elegans (b). For clarity, other target genes of tra factors in respective branched pathways are not shown. For review see Ref. 1. B, ribbon model (stereo pair) of DSX zinc module (residues 40-80) based on NMR studies (21). Helical segments consist of residues 46-50 and 71-79. Cys and His side chains are shown; two zinc ions are shown as spheres. Dashed line at the C terminus indicates the beginning of disordered tail. Coordinates have been deposited in the Protein Data Bank (accession code 1LPV).

View larger version (48K): [in this window] [in a new window]   Fig. 2.   DM sequences and DNA binding studies. A, alignment of metazoan DM sequence motifs. Cysteines and histidines that coordinate Zn2+ are aligned as two intertwined binding sites (boxes), site I and site II (see Fig. 1B). Sites of substitutions employed in the present study are shown in red (impaired DNA binding), green (unimpaired DNA binding), or gold (Arg-46; slightly decreased DNA binding with confounding structural perturbation). Stable -helical elements are highlighted by magenta ribbons above the DSX sequence. Nascent C-terminal -helix is indicated by a black dashed extension. Arrowheads without parentheses indicate sites of point mutations in dsx or mab-3 associated with intersex development; substitutions in parentheses indicate variants characterized only by biochemical assays (19). DSX, DM domain in D. melanogaster (GenBankTM accession number M25292). MAB-3a and MAB-3b, The first and the second DM domains in C. elegans protein (Z99278). Other C. elegans DM sequences: F10C1.5, cosmid F10C1 (U49831); C34D1.1, cosmid C34D1 (Z78060); K08B12, cosmid K08B12 (U97001); F13G11, cosmid F13G11 (Z83317); T22H9.4, cosmid T22H9 (AAC69225); CAA93739 (Z69883); and CAA21612.1 (AL032637.1). Vertebrate DM sequences: hDMRT1 and hDMRT2, human homologs on short arm of chromosome 9 (DMRT1h AF130728; DMRT2h AF130729). Other homologs: mDmrt1 (murine, AF202778), pDmrt1 (porcine, AF216651), and cDmrt1 (chicken, AF123456). TERRA, DM domain of zebrafish terra (AF080622). Sequences shown include genes (such as TERRA; see Ref. 70) not known to be involved in sex determination. B, deletion of C-terminal tail impairs DNA binding and cooperativity. Lane 1, intact DM domain at 16 nM concentration in low ionic strength assay buffer. Lane 2, free fbe probe (29 bp). Lanes 3-12, successive DM concentrations 14, 28, 42, 56, 92, 130, 180, 280, 370, and 960 nM. C, alanine scanning mutagenesis of Arg and Gln residues in C-terminal tail of DSX DM domain. Wild-type control is provided in lane b. Alanine substitutions at residues 79, 80, 81, 90, 91, and 93 impair specific DNA binding to dsxA. By contrast, alanine substitutions at residues 74, 95, 98, and 99 do not impair specific DNA binding. Composite GMSA gel showing specific fbe gel shift at DSX domain concentration 140 nM. Percentages of DNA probe shifted to the C1 and C2 forms were as follows: wild type (wt) (C1 4% and C2 38%); R74A (3 and 27%); R79A (non-detectable and 10%); Q80A (2 and 3%); R81A (2 and 11%); R90A (< 1 and 4%); R91A (1 and 8%); Q93A (1 and 5%); Q95A (7 and 60%); Q98A (2 and 55%); and R99A (6 and 30%). Apparent binding in C is weaker than in prior reports (11, 19, 21) due to higher [KCl] and poly [d(I-C)] concentrations.

DM genes are conserved among vertebrates (23). A subset (designated Dmrt1, -2, -3, etc.; see Ref. 23) is expressed in the differentiating gonad and is associated with human sex reversal (46, XY gonadal dysgenesis; see Ref. 24). Additional DM homologs of unknown function are encoded in the C. elegans genome (middle group in Fig. 2A) (25). Although the biochemical properties of these DM-related proteins have not been characterized, the DM domains of DSX and MAB-3 exhibit specific zinc-dependent DNA binding (19, 20). Mutations in these domains are associated with intersex development (arrowheads in Fig. 2A) (26-28). Phenotypes correlate with loss of DNA binding activity (18, 19). DSX mutations in the invariant cysteines or histidines impair both zinc coordination and DNA recognition (19). Point mutations in the DM domains of human DMRT1 or DMRT2 (linked genes on chromosome 9p) are apparently rare in patients with unexplained sex reversal or intersex phenotypes, presumably due to functional redundancy (24). Deletions spanning DMRT1 and DMRT2 are nonetheless associated with 46, XY gonadal dysgenesis, giving rise to intersex phenotypes with high risk of gonadoblastoma (the 9p syndrome; see Refs. 29-36). Selective deletion of the murine Dmrt1 gene by homologous recombination causes XY gonadal dysgenesis and azospermia without sex reversal (37). There is no apparent phenotype in the female mouse.

The DSX DM domain contains an ordered moiety and disordered tail (21). In ¹H NMR studies the ordered moiety (spanning DSX residues 40-80, including invariant cysteines and histidines; Fig. 1B) exhibits marked dispersion of chemical shifts, whereas the tail (residues 81-105) is manifest by an envelope of poorly resolved spin systems at near-random-coil chemical shifts. The zinc module consists of tetrahedral CCHC and HCCC metal-binding sites; the two sites are contiguous within a single hydrophobic core. The tail is hypothesized to function as a nascent recognition -helix (21). Mutations in either zinc-binding site or tail (R91Q; see Ref. 19) can lead to an intersex phenotype. The DSX DM domain, itself monomeric, binds as a dimer to a specific target site, designated dsx^A, within the fat body enhancer (38, 39), a well characterized genetic response element with sex-specific regulation (40, 41). Critical base pairs in a consensus target site (as defined by random binding site selection) are palindromic about a central AT bp (20, 28). Studies of DNA analogs suggest that the motif binds in the DNA minor groove. Despite such groove targeting, DSX-induced DNA bending is negligible as inferred from permutation gel electrophoresis (21). Absence of sharp DNA bending stands in contrast to the marked electrophoretic anomalies induced by binding of HMG boxes (including SRY; see Refs. 7 and 42). The DNA-binding properties of DSX are proposed to underlie its function in combinatorial gene regulation (21, 38, 39). To our knowledge, the structure of a DM-DNA complex has not been determined.

This study investigates structure-function relationships in the DSX DM domain. Deletion of the tail is shown not to alter the folding of the zinc module but precludes formation of a specific dimeric complex. DNA binding assays employ the dsx^A control element and corresponding DNA half-site. Selected residues important for specific DNA binding are identified by site-directed mutagenesis. Two mutagenesis strategies are employed. The first approach focuses on the disordered C-terminal tail wherein charged or polar side chains (arginine or glutamine) are substituted by alanine; site-directed mutagenesis is effected in Escherichia coli. The second approach employs chemical protein synthesis by native ligation (43); arginine or lysine residues in the zinc module are substituted by norleucine. This strategy is meant to distinguish between the possible structural role of the aliphatic side chain (retained in norleucine) and the functional role of the positive charge (-amino group of lysine or guanidinium group of arginine; absent in norleucine). Of the 13 substitutions tested, 8 significantly impair specific DNA binding and 5 do not. Together, evidence is obtained that protein-DNA contacts are made by side chains in both the zinc module and tail to define a bipartite DNA-binding motif. Multiple conserved arginine and glutamine residues in the tail are required for specific DNA binding, consistent with its proposed role as a nascent DNA-recognition -helix.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Chemical Protein Synthesis-- Wild-type and norleucine analogs were prepared by solid-phase peptide synthesis and native fragment ligation (Fig. 3A) (43). The N-terminal peptide consisted of DSX residues 35-67; the C-terminal peptide consisted of residues 68-105 (DM domain) or 68-86 (DM fragment; outlined in schematic form in Fig. 3A). Ligation products were purified by reverse phase-HPLC (Fig. 3, B and C). Peptide fragment synthesis employed an Applied Biosystems 430A synthesizer. t-Boc amino acids were used with the following protection: Arg(tosyl), Asn(xanthyl), Asp(OcHxl), Cys(4MeBzl), Glu(OcHxl), His(DNP), Lys(2ClZ), Ser(Bzl), Thr(Bzl) and Tyr(BrZ). N-terminal peptides (DSX residues 35-67 and variants) were synthesized on a C-terminal thioester-generating resin. C-terminal peptides (DSX residues 68-105 and 68-86) were respectively synthesized on a Boc-Glu(OcHxl)-OCH₂-Pam resin and Boc-Gln-OCH₂-Pam resin. Full-length polypeptides were prepared by native chemical ligation (43). Electrospray mass spectra of domains with and without addition of 2 eq of Zn²⁺ (Fig. 4) yielded values differing by 125-126 daltons, consistent with twice the atomic mass of zinc (130.76 Da) minus six, the number of cysteines from which sulfhydryl protons are removed on metal binding. The following four DM analogs were prepared: R46Z, K57Z, K60Z (using variant N-terminal fragments; where Z designates norleucine), and R91Q (using a variant C-terminal fragment). A 19-residue C-terminal tail peptide (DSX residues 87-105; sequence TALRRAQAQDEQRALHMHE) was also synthesized. The peptide contained an N-terminal acetyl group; its observed molecular mass (2303 ± 0.6 Da) is in accord with its calculated mass (2303.6 Da). The residue numbers refer to native DSX sequence; DSX residue 35 thus corresponds to residue 1 of DM consensus sequence (Fig. 2A). Use of native DSX numbering facilitates correspondence with prior genetic analysis (19).

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Chemical protein synthesis by native ligation. A, schematic overview of ligation scheme whereby an N-terminal peptide (33 residues) is joined to native or truncated C-terminal peptides (38 or 19 residues, respectively) to yield DM or DM polypeptides. The cysteine participating in the ligation reaction is highlighted (filled oval); the remaining five cysteines are indicated by open ovals. B, rp-HPLC chromatograms showing purity and elution positions of DSX DM domain (a) or DM fragment (b). C, example of ligation reaction in which purified N-terminal peptide thioester (a) is coupled to purified C-terminal 19-mer (b) to yield ligated product (c) (labeled P, asterisk) and a mixture of free C- and N-terminal peptides (labeled C and N in c). The ligation product is readily purified by HPLC (asterisk in d).

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Electrospray mass spectrometric analysis of synthetic ligation products: A, intact DM domain; B, DM fragment. Spectra were obtained in the presence (red) or absence (black) of Zn2+ ions at a stoichiometry of 2 zinc ions per polypeptide. Mass differences (125 and 126 Da) are in accord with atomic mass of two zinc ions (130.72 Da) minus six sulfhydryl protons lost on metal binding. Upper panels contain raw data; lower panels show inferred molecular masses.

Bacterial Expression-- The DSX cDNA coding region (female isoform) was kindly provided by P. Wensink (Brandeis University). A DNA segment encoding the wild-type DSX DM domain (residues 35-105) was recloned by polymerase chain reaction into an overexpression plasmid in which a tac promoter (inducible by isopropyl--thio-galactopyranoside) drives expression of an N-terminal His₆-tagged fusion protein. The fusion protein contains an N-terminal staphylococcal nuclease domain followed by a thrombin-sensitive linker and C-terminal DSX domain (44). The purified DSX domain contains two N-terminal non-native N-terminal amino acids (Gly-Ser) derived from the thrombin site. This expression system is a modification of a plasmid originally designed by Markley and co-workers (45). Alanine scanning mutations were introduced into the DSX coding region in phage M13mp19RF by oligonucleotide-directed mutagenesis by polymerase chain reaction and recloned into the above expression plasmid. Fidelity of mutagenesis was verified in each case by DNA sequencing. None of the mutations altered the efficiency of overexpression in E. coli.

Protein Purification-- Recombinant proteins were purified from E. coli lysates by Co²⁺ affinity chromatography (Clonetech, Inc.) using an imidazole elution gradient. Following thrombin digestion using bead-immobilized enzyme, cleaved DM domains were purified by reverse phase-HPLC (Vydac C4 column, 1 × 25 cm). The yield in each case was ~2 mg/liter of fermentation. Molecular masses of purified proteins were verified by mass spectrometry to exclude proteolytic degradation or inadvertent mutation. None of the alanine substitutions significantly altered the solubility or chromatographic properties of the variant proteins.

DNA Binding Assays-- The sequence of the DNA site (29 bp) was 5'-GTGCACAACTACAATGTTGCAATCAGCGG-3' and complement (3'-CACGTGTTGATGTTACAACGTTAGTCGCC-5'; dsx^A site in boldface). The sequence of a consensus half-site (10 bp) was 5'-AGCTACATTG-3' and complement (critical bases in bold) (19, 20). A nonspecific DNA control site (17 bp) was provided by  phage operator site O_L1 (5'-TACCACTGGCGGTGATA-3' and complement). Immediately prior to DNA binding studies, the proteins were reduced in 50 mM dithiothreitol in 50 mM Tris-HCl (pH 8.0), repurified by rp-HPLC as above, lyophilized, and reconstituted with a 20% excess of ZnCl₂ in the assay buffer (below). Specific DNA binding was monitored by gel mobility shift assay (GMSA; 10% acrylamide with 29:1 bisacrylamide) run in 45 mM Tris borate (pH 8.0) without EDTA (omitted to avoid competitive chelation of zinc ions) at 200 V and 4 °C. In studies of alanine scanning mutants (Fig. 2C) and norleucine analogs binding (Fig. 8) reactions were conducted in 20 mM Tris-HCl (pH 7.4), 150 mM KCl, 5 mM MgCl₂, 0.1 mM ZnCl₂, 5% glycerol, 33 µg/ml bovine serum albumin, and either 0.08 (Fig. 8) or 0.10 µg/µl (Fig. 2C) poly(dI-dC) competitor DNA. In studies of DM fragment (Fig. 2B) and half-site binding (Fig. 6), binding reactions were conducted in a lower ionic strength buffer consisting of 10 mM Tris-HCl (pH 7.4), 50 mM KCl, 0.1 mM ZnCl₂, 5% glycerol, 33 µg/ml bovine serum albumin, and 0.06 µg/µl poly(dI-dC) DNA (unlabeled). The concentration of ³³P-labeled DNA was less than 1.5 nM. GMSA studies of biosynthetic alanine variants included control lanes containing the wild-type biosynthetic domain; GMSA studies of synthetic variants included control lanes containing the wild-type synthetic domain. Quantification of GMSA bands was obtained using the PhosphorImager software package as described by the vendor (Amersham Biosciences).

DNA Binding Cooperativity-- A formalism is obtained from the method of Senear and Brenowitz (46) based on classical thermodynamic models (47, 48). In brief, the monomeric DM domain binds to the pseudo-palindromic half-sites of the dsx^A DNA target with intrinsic association constants k₁ and k₂, respectively, and cooperativity parameter k₁₂. The concentration of labeled DNA probe is much less than the range of protein concentrations tested. At a given protein concentration [P] the fraction of DNA sites that are unbound (F₀), singly occupied (F₁; labeled C1 in the present GMSA studies), or doubly occupied (F₂; labeled C2) is given by Equations 1-3 (46),

(Eq. 1)

(Eq. 2)

(Eq. 3)

The formalism may be simplified by the assumption that k₁ = k₂ = k, i.e. the two half-sites have similar intrinsic affinities. Under these conditions, the cooperativity parameter is given by Equations 4 and 5 (46),

(Eq. 4)

(Eq. 5)

where F_1,max is the maximal value of the singly-occupied complex observed in the GMSA titration. The low values of F_1,max observed in the present studies (see "Results") indicate a high degree of positive cooperativity unaffected by the norleucine substitutions. In this limiting case, the fold change in intrinsic dimer-specific association constants (K_a = k₁k₂k₁₂ = k²k₁₂) effected by the substitution is essentially equal to the ratio of protein concentrations where F₂ = 0.50 or in the case of the K60Z variant (for which 50% occupancy was not achieved) where F₂ = 0.25. Decrements in the apparent protein-DNA association constant due to mutation can be ascribed primarily to perturbation of DNA contacts only if the mutation does not proportionately perturb the cooperativity factor.

Circular Dichroism-- Zinc-dependent protein folding of DM fragment and norleucine domains was evaluated by circular dichroism (CD) and proton nuclear magnetic resonance (NMR) at 600 MHz as described (21). CD spectra were obtained using an Aviv spectropolarimeter equipped with temperature control. In CD studies of half-site binding (Fig. 6), the DSX DM and DNA concentrations were 10 µM in 50 mM Tris-HCl (pH 7.4) and 50 mM KCl. The light path length for the CD cell was 1 mm. Protein concentrations were determined by UV absorbance and verified by quantitative amino acid analysis. In studies of protein-DNA interactions CD difference spectra² (designated 1 and 2) were calculated to prove induced helical structures; mean residue ellipticities are relative to the protein component.

Nuclear Magnetic Resonance Spectroscopy-- For NMR spectroscopy, solutions were purged with N₂ and contained 2 mM deuterated dithiothreitol (Cambridge Isotopes, Inc., Woburn, MA). Spectra were recorded in 50 mM deuterated Tris-HCl (pH 6.5, pD 6.1, or pD 8.0; see Fig. 7 legend) and 5 mM deuterated dithiothreitol at a Zn²⁺:protein ratio of 2.5:1 in 90% H₂O and 10% D₂O at 25 °C.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The DSX DM domain contains an ordered moiety (the zinc module) and disordered C-terminal tail, proposed to fold on specific DNA binding as a DNA recognition -helix (21). Binding of the DSX DM domain to dsx^A leads to discrete 1:1 and 2:1 complexes (labeled C1 and C2 in lane 1 of Fig. 2B). The predominance of the 2:1 complex under conditions containing appreciable free DNA demonstrates its cooperative assembly (cooperativity factor k₁₂ >1; see "Experimental Procedures"). Previous mutagenesis studies have focused on the importance of the conserved cysteine and histidine residues involved in zinc coordination (19). Deletion of the tail (yielding fragment DM; residues 35-86) leads to >100-fold reduction in specific DNA binding; only a weak 1:1 band is observed (labeled C1* in lanes 3-12 in Fig. 2B). Retention of a low affinity 1:1 complex indicates that DM retains a portion of the DNA-binding surface. Addition of an isolated C-terminal peptide (residues 87-105; 19 residues) at concentrations up to 100 µM does not restore formation of a specific dimeric complex. No DM complex is observed in control GMSA studies of the  operator site O_L1 (not shown).

The ¹H NMR spectrum of DM is similar to that of the intact domain (Fig. 5) (21), demonstrating native folding of the zinc module in the fragment. Spectra of the intact domain and DM fragment exhibit identical chemical shift dispersion but differ by the presence (DM) or absence (DM) of broad amide resonances near random-coil frequencies³ (asterisk in a and d of Fig. 5). These are assigned by elimination to the C-terminal tail. The pattern of nuclear Overhauser enhancements (NOEs) between side chains in the metal-binding sites (including contacts between histidine and cysteine side chains; b and e of Fig. 5) and between amide protons in -helices (d_NN contacts; c and f) is essentially identical. An upfield-shifted methyl resonance is observed in each case, assigned to the Ile-54-CH₃. This core side chain packs against the imidazole ring of His-59 (NOEs labeled in b and e). Differences in chemical shifts between DM and DM spectra are small (< 0.1 ppm) and localized near the site of truncation.

View larger version (34K): [in this window] [in a new window]   Fig. 5.   NMR studies of DM domain (A) and DM fragment (B). Truncation of C-terminal tail does not affect folding of zinc module but removes envelope of near-random-coil resonances from NMR spectrum. a and d, one-dimensional spectra of exchangeable NH and aromatic resonances at pH 6.5. Spectra highlight the presence (intact DM domain; a) or absence (DM; d) of a broad envelope of amide resonances near 8.1 ppm (asterisks); these unresolved resonances exhibit prominent water exchange cross-peaks in NOESY spectra (not shown) and are assigned to nascent helix in C-terminal tail. Amide resonances in both domains are notable for resolved downfield shifts of cysteate residues Cys-70, Cys-73, and Cys-47. b and e, two-dimensional NOESY spectra in D2O (pD 6.1) of contacts in metal-binding sites between side chains of histidine and cysteine. Comparison of amide region of NOESY spectra in H2O is remarkable for similarity of chemical shifts among side chains in (Cys-44, His-50, His-59, Cys-63, and Cys-67) or adjoining (Ile-54) the zinc-binding sites. The  methyl resonance of Ile-54 is shifted to high field (0.63 ppm) presumably due to the aromatic ring current of His-59; prominent NOEs are observed between these side chains. c and f, dNN contacts in helical segments (DSX residues 46-80 (red) and 71-79 (green) at pH 6.5; DM consensus residues 12-16 and 37-45) are outlined. The NOESY mixing times were in each case 175 ms.

Single amino acid substitutions extrinsic to the core of the metal-binding sites enable the importance of individual side chains in DNA recognition to be tested. Because of the different structural characteristics of the zinc module and disordered tail, we employed two different approaches. The first is alanine scanning mutagenesis of the tail. Such an approach has been widely applied to nascent DNA recognition elements (49-51). Because alanine contains a small side chain of high intrinsic helical propensity (52), such substitutions are unlikely to perturb the bound structure of the tail or introduce steric obstacles to induced fit. The second approach employed chemical protein synthesis to introduce norleucine at sites of conserved lysine or arginine side chains in the zinc module. Use of a non-standard amino acid was motivated by analysis of the NMR structure of the zinc module; the long methylene chains of some basic side chains pack near the metal-binding sites while their terminal positive charges are exposed to solvent. Norleucine (unlike alanine) in principle retains such hydrophobic packing, enabling the specific role of the positive charge to be evaluated.⁴ Because the DM polypeptide is too long for conventional peptide synthesis, the variant proteins were prepared by native peptide ligation (43) as illustrated in Fig. 3 (see "Experimental Procedures").

Alanine Scanning Mutagenesis of Tail-- Systematic alanine substitution of 10 arginine or glutamine residues in the tail was effected by site-directed mutagenesis (residues Arg-74, Arg-79, Gln-80, Arg-81, Arg-90, Arg-91, Gln-93, Gln-95, Gln-98, and Arg-99; highlighted in color in Fig. 2A). Recombinant proteins were purified to homogeneity, and GMSA studies of each were undertaken as a function of protein concentration. The results are summarized in a composite gel in which representative lanes containing the same protein concentration, extracted from individual gels, are compared (Fig. 2C). Percent shifts of the single bound complex (designated C1) and doubly bound complex (C2) are given in the legend. A striking correlation is observed between critical sites and degree of sequence conservation. Conserved basic or carboxamide side chains in two patches contribute to DNA recognition (positions 79-81 and positions 90, 91, and 93; highlighted in red in Fig. 2A), whereas substitution of non-conserved side chains does not impair specific DNA binding (positions 74, 95, 98, and 99; green in Fig. 2A). Position 91 is a site of intersex substitution R91Q (19); GMSA studies of this variant indicate a decrement in binding more marked than that of the R91A substitution (not shown). Although small variations occur in the relative intensity of the intermediate 1:1 band (C1 in Fig. 2C), deleterious substitutions do not affect this ratio (to the extent that the 1:1 band can be seen), implying that decreased specific DNA binding is not secondary to decreased cooperativity.⁵ That multiple alanine substitutions in the tail impair specific DNA binding strongly supports the hypothesis that the tail functions as a DNA recognition element.

Folding of the Tail on a DNA Half-site-- CD spectra of the DM polypeptide in the presence and absence of ZnCl₂ (at a stoichiometry of two zinc ions per polypeptide) demonstrate the presence of a metal-dependent folding transition (Fig. 6A). Binding of Zn²⁺ accentuates the partial -helical propensity of the apodomain in accord with the solution structure of the motif (Fig. 1B) (21). The dimeric protein-DNA complex formed on binding of the DSX DM domain to dsxA has been investigated previously by CD. Spectra are remarkable for additional specific DNA-dependent -helical difference features and thermal stabilization of the C-terminal tail as an induced -helix (21). Such induced fit can in principle occur at the protein-DNA interface or secondary to DNA-dependent dimerization. The DSX domain forms a low affinity 1:1 complex on binding to a dsx^A half-site (5'-AGTACATTG-3' and complement; Fig. 6B). Spectra of the free domain, free dsx^A half-site and 1:1 complex at 37 °C are shown in Fig. 6C. An -helical CD difference feature (designated 1 in Fig. 6D) is obtained by subtracting the CD spectrum of the free DNA from that of the complex. The relative magnitude of this feature (but not its detailed shape) is similar to that of the 2:1 dsx^A complex as shown by comparison of difference spectra 2 (open squares in Fig. 6E). A DNA-related difference feature is observed in the near-UV region (250-300 nm), presumably due to widening of the minor groove. This feature is similar at 4 and 37 °C (dashed line versus open squares in Fig. 6E), unlike the far-UV difference feature, which is attenuated at low temperature due to base-line folding of the free tail. Formation of a 1:1 half-site complex is sufficient to confer thermal stability between 4 and 40 °C as indicated by the slope of []₂₂₂ versus temperature (Fig. 6F); the anomalously steep slope of the free domain (dashed line in Fig. 6F) reflects non-cooperative thermal fraying of nascent helical structure in the tail. DNA-dependent folding and stabilization of the tail can thus occur in the absence of DNA-dependent dimerization. Together with the results of alanine scanning mutagenesis (above), these data strongly support the hypothesis that the tail functions as an induced recognition -helix.

View larger version (35K): [in this window] [in a new window]   Fig. 6.   DSX DM tail folds on binding to half-site DNA. A, far-UV CD spectrum of zinc-DM complex (open circles) and apodomain (solid circles). B, upper panel, GMSA of DSX DM domain binding to half-site of dsxA. Bands C1 and f designate 1:1 protein-DNA complex and free DNA, respectively. Lanes a-h correspond to low affinity 1:1 complex with protein concentrations of 0, 8, 32, 64, 128, 256, 512, and 1000 nM. Concentration of 33P-labeled DNA site was 0.8 nM. Lower panel, sequence of DSX half-site DNA (10 bp). The arrow points to position corresponding to central bp in dsxA. C and D, helix formation in half-site 1:1 complex is inferred from comparison of CD spectra of DM domain (dotted line) and complex (solid line) at 37 °C. The free DNA is shown as dash-dot-dot-dot-dashed line. D, far-UV CD difference spectrum 1 (dash-dot-dash) represents difference between the CD spectrum of the protein-DNA complex and the CD spectrum of the free DNA. Induced helical feature at 222 nm is labeled. Curves for DM domain (solid line) and free DNA (dash-dot-dot-dot-dashed line) are provided for comparison. E, difference spectra 2 of half-site 1:1 complexes at 4 °C (dash-dotted line) and 37 °C (open squares) and native 2:1 complex at 37 °C (solid line) (reprinted from Ref. 21). Difference spectra 2 is defined as the difference between the spectrum of the complex and the sum of the spectra of the free DNA and free protein. Marked changes in structure of complexes are reflected in difference spectra. The large amplitude of far-UV feature (200-250 nm) in 1:1 complex at 37 °C suggests that helical content is stabilized by the monomer-DNA interaction. DNA-specific difference features are observed in the near-UV (250-300 nm). For 2:1 dsxA complex the protein and DNA concentrations were 10 and 5 µM, respectively. For 1:1 half-site DNA complex the protein and DNA concentrations were each 20 µM. F, CD melting curves of free DM domain (dashed line) and complex (solid line) as monitored at 222 nm from 4 to 95 °C. Curves demonstrate normalization of anomalous initial slope in DM curve on half-site DNA binding (arrows).

Non-standard Mutagenesis-- Like the tail, the DSX zinc module contains multiple basic residues: seven arginines and five lysines. The positions of these side chains in the NMR ensemble are shown in Fig. 7. The abundance of basic side chains suggests that the zinc module may also contact DNA, consistent with the residual specific DNA binding affinity of the DM monomer (above). To test this hypothesis, norleucine substitutions (designated Z) were introduced at positions Arg-46, Lys-57, and Lys-60. Among DM sequences residue 46 (DM consensus position 12) is conserved as lysine or arginine (Fig. 2A) and is a site of intersex mutation (Arg  Trp) in MAB-3 (18). This side chain is well defined in the solution structure of the free domain (side chain root mean square deviation 0.94 Å) and of limited solvent accessibility (24% relative to an extended peptide) (21). Residue 57 (DM consensus position 23) is also conserved as lysine or arginine, whereas residue 60 (consensus position 26) is usually lysine but is not conserved within two divergent C. elegans sequences in which alanine or serine is observed (Fig. 2A). The side chains of Lys-57 and Lys-60 are less well ordered in the solution structure (respective side chain root mean square deviations 1.77 and 2.17 Å) (21). Lys-57 is largely exposed (relative solvent accessibility 82%), whereas Lys-60 projects into solution and is disordered.

View larger version (26K): [in this window] [in a new window]   Fig. 7.   Positions of basic side chains in DM zinc module and NMR screening of analogs. A, stereo pair showing solution structure of DSX domain with selected side chains; cysteines and histidines in metal-binding sites, Ile-54 (which exhibits an upfield shifted  methyl resonance; see Fig. 5), and basic side chains on protein surface. Basic side chains in disordered N-terminal residues (Arg-39) and C-terminal tail are not shown. B, TOCSY spectrum of wild-type domain showing aromatic spin systems at pD 6.1 (assignments as labeled, see Ref. 21). The mixing time was 55 ms. C, TOCSY spectrum of DM; conditions as in B. D, one-dimensional NMR spectra of wild-type and K60Z and R46Z variants. In the R46Z spectrum arrow indicates ortho resonance of Phe-65, shifted slightly downfield in R46Z analog, and asterisk indicates C2H resonance of His-50. Differences in histidine resonances near 8.1-8.2 ppm arise from small differences in pD near pKa values of unliganded histidines in tail. E, aromatic spectra of analogs at pD 8.0. Histidine spin systems are as labeled, and asterisk indicates upfield perturbation in  resonance of His-50 in R46Z analog. Arrow and asterisk are as in D. Wild-type one-dimensional spectrum is essentially identical to that of R91Q tail analog.

Native zinc-dependent folding of the norleucine analogs was verified by CD and ¹H NMR spectroscopy. Evidence of a native-like fold is provided by aromatic and aliphatic NMR spectra, which in each case exhibit a similar pattern of chemical shifts. Resolved markers are provided by aromatic spin systems as outlined in the wild-type TOCSY spectrum (Fig. 7B); these resonances are not perturbed by truncation of the C-terminal tail (Fig. 7C). Small non-local perturbations are observed in the spectrum of the R46Z analog (Fig. 7, D and E, arrow and asterisk) but not other analogs. Results of GMSA assays and plot of percent shifts are shown in Fig. 8. The norleucine substitutions do not significantly affect the relative intensities of the 2:1 and 1:1 shifted bands, indicating that positive charges at these sites are not required for DNA-dependent dimerization. In each case the maximal fractional occupancy of the singly bound complex (C1) is small; values of F_1,max (see "Experimental Procedures") are 5 (wild type), 8 (R46Z), 4 (K57Z), and 7% (K60Z). Due to errors in integration arising from non-uniform background and trailing edges of bands, we estimate uncertainties of ±2% in these values. The cooperativity factor k₁₂ is in each case greater than 100. 

View larger version (35K): [in this window] [in a new window]   Fig. 8.   Specific DNA-binding GMSA studies of norleucine analogs: A, wild-type and K57Z analog; B, K60Z; and C, R46Z. The analogs retain cooperative binding to form 1:1 and 2:1 complexes (bands C1 and C2, respectively). Whereas the K57Z and R46Z substitutions impair DNA binding by only 1.5- and 2-fold, respectively, the affinity of the K60Z analog is reduced by 10-fold. For comparison, lanes 6 and 11 in A each contain 48 nM protein and exhibit similar percent shifts. Similarly, lanes 7 and 12 in A, lane 7 in B, and lane 14 in C each contain 96 nM protein. Protein concentrations are as follows: A, lanes 2-7, 4, 8, 12, 24, 48, and 96 nM; lanes 8-14, 6, 12, 24, 48, 96, 144, and 192 nM; B, lanes 1-9, 0, 24, 36, 48, 60, 72, 96, 144, and 240 nM; C, lanes 10-15, 12, 24, 48, 72, 96, and 144 nM. Apparent affinity of wild-type domain is higher than in Fig. 2C due to lower concentration of poly(dI-dC) in assay buffer. D, plot of percent of DNA probe shifted to C1 (circles) and C2 (rectangles) forms with respect to protein concentration. The wild type (wt) is shown in black; mutants K57Z, R46Z, and K60Z are shown in red, green, and blue, respectively. The amount of DNA probe shifted to the dimeric complex in the K60Z mutant is 10-fold less than that of wild-type domain, whereas mutants K57Z and R46Z exhibit decrements of only 1.5- and 2-fold, respectively.

Under these conditions relative dimer-specific association constants may be estimated by comparison of the protein concentrations at which 50% of the labeled DNA is shifted to the dimeric complex. Inspection of Fig. 8E thus indicates that substitutions K57Z and R46Z impair specific DNA binding only to a small extent (decrements of 1.5- and 2-fold, respectively). Substitution K60Z more significantly impairs specific DNA binding; although 50% occupancy of the dimeric complex was not achieved in the protein concentration range tested, 10-fold higher concentrations were required to obtain 25% occupancy. The similar F_1,max values of the wild-type and K60Z domains indicate that this perturbation is not primarily due to impaired cooperativity. In light of the native-like NMR spectrum of the K60Z analog, we speculate that the positive charge of Lys-60 participates in a salt bridge to the DNA backbone. The small decrements in binding of the R46Z and K57Z analogs may reflect either similar loss of local contacts, structural perturbations in the zinc module, or (in the case of R46Z) a subtle effect on cooperativity. In any case markedly impaired binding of the K60Z norleucine analog, taken together with the results of tail deletion and scanning mutagenesis, suggests that both parts of the DM motif contact DNA.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

DSX contains an N-terminal DM domain and a C-terminal dimerization domain (11, 28). This modular organization is reminiscent of phage  repressor (53); the N-terminal domain in each case mediates DNA recognition, whereas the C-terminal domain in each case enhances dimerization and enables cooperative binding to pairs of DNA sites (12, 54). Mutations in either DSX domain are associated in vivo with intersex phenotypes (19, 28). The DM domain (but not other regions of the protein) is conserved among a newly recognized family of transcription factors. Prototype members of this family are provided by orthologs doublesex (dsx; see Ref. 26) in D. melanogaster and mab-3 in C. elegans (55). These sex-determining genes function downstream of inequivalent "master" regulatory factors to define one branch of a ramifying pathway (Fig. 1A). Remarkably, dsx and mab-3 in part regulate analogous dimorphic tissues (40, 56, 57). The present study has investigated by mutagenesis the role of 13 side chains on the surface of the DSX zinc module and in its C-terminal tail. The results suggest that both structural elements contact DNA and substantiate the hypothesis that the tail functions as a recognition -helix.

The DSX DM domain contains a distinctive zinc module with intertwined CCHC and HCCC zinc-binding sites (21). CD studies suggest that a C-terminal tail, disordered in the free domain, is stabilized on DNA binding. The present results demonstrate that such induced fit occurs in a monomeric half-site complex and is thus independent of DNA-dependent dimerization. It is not known whether the tail forms a single contiguous -helix or two or more helical segments separated by turns or bends. Unlike classical zinc fingers and zinc modules (58, 59), the DM domain binds in the minor groove of DNA (21). Unlike SRY and other architectural motifs targeted to the minor groove (42, 60), the DM motif does not induce sharp DNA bending. Our results demonstrate that two patches of conserved arginine and glutamine side chains in the tail are necessary for high affinity DNA recognition. The first patch (RQR; DSX residues 79-81 at DM consensus positions 45-47) is precisely conserved among mammalian Dmrt genes (Fig. 2A). The second patch (RRAQ; DSX residues 90, 91, and 93 at DM consensus positions 56, 57, and 59) is conserved as RRQQ among mammalian Dmrt genes. Such conservation suggests that structure-function relationships in DSX generalize to the human proteins. It is noteworthy that the center of the first patch (residue 80) is separated from the Arg-90 by 10 residues, which would correspond to three turns of a single contiguous -helix.

Tail sequences vary among invertebrate DM sequences. Whereas the first patch is not conserved in some C. elegans DM sequences (Fig. 2A), the second patch is not conserved in one of the two DM domains in MAB-3.⁶ Such divergence suggests that DM domains can exhibit inequivalent sequence specificities or atomic mechanisms of recognition. In fact, MAB-3 and DSX exhibit distinct (but related) DNA binding specificities as defined by random-binding site selection (20). The biological target sites of MAB-3 are not well characterized. Interestingly, the male-specific isoform of DSX functions in a mab-3 XO worm (ordinarily a chromosomal male with an intersex phenotype; see Ref. 55) to rescue male features (18). Such complementation is consistent with overlapping DNA binding specificities and demonstrates that downstream mechanisms of gene regulation are in part conserved (18). It is not known whether the divergent tails of MAB-3 domain a or other DM domains contact DNA and, if so, whether their mode of binding differs from that of a consensus DSX-like domain.

The present study differs in part from that of a previous analysis of the DSX DNA-binding domain (10, 12). Wensink and co-workers (10) expressed a His₆-tagged fragment of DSX (residues 39-104; 72 residues including N-terminal tag) in a baculoviral system and characterized the oligomerization- and DNA-binding properties of a partially purified preparation. Whereas the present domain (residues 35-105) was found to be monomeric by equilibrium ultracentrifugation and NMR spectroscopy in the 50 µM to 2 mM concentration range (21), the His₆-tagged fragment was observed to oligomerize at lower concentrations as inferred from glutaraldehyde cross-linking experiments (12). GMSA studies with a dsx^A probe, conducted in 25 mM HEPES buffer (pH 7.6) containing 100 mM NaCl, revealed formation only of a 2:1 complex; a 1:1 complex corresponding to band C1 in the present studies was not observed. The shape of the DNA-binding isotherm was sigmoidal, indicating positive cooperativity. Analysis of the thermodynamic coupling between dimerization of the free domain and specific DNA binding was performed based on the model shown in Fig. 9A (12). Despite the difference in appearance between this model and that of cooperative DNA binding (Fig. 9B), the thermodynamic implications are similar. In particular, the ratio of the dimerization constant (K₁; 430 nM) to the dimer-specific dissociation constant (K₂; 0.48 nM) is formally equivalent to the cooperativity factor k₁₂ in the Senear-Brenowitz formalism (46). We may therefore apply Equation 5, see "Experimental Procedures," to estimate the corresponding F_1,max in a cooperative model (Equation 6),

(Eq. 6)

This estimate is in good agreement with the maximal fractional occupancy of the singly-bound complex (C1) observed in the present GMSA studies⁷ (5 ± 2%).

View larger version (18K): [in this window] [in a new window]   Fig. 9.   Thermodynamic models of specific protein-DNA binding. A, pre-assembly mechanism. Dimerization of the free DM domain (black ovals) precedes specific binding of the protein dimer (dimerization constant K1) to the pseudo-palindromic dsxA site (box, protein-DNA dissociation constant K2). The monomer does not bind DNA. B, DNA-dependent assembly mechanism. Although the protein is strictly monomeric, cooperative DNA binding leads to DNA-dependent assembly of the 2:1 protein-DNA complex (C2). The dimeric interface of the protein is induced on DNA binding. Intrinsic association constants k1 and k2 represent binding constants to the individual DNA half-sites; k12 is the cooperativity parameter.

The DM domain provides an unusual example of a minor groove DNA-binding motif that employs a flexible basic tail. The use of flexible basic regions to recognize specific DNA sequences is well characterized among major groove DNA-binding motifs. Such basic regions fold on DNA to form -helices (61). Examples include the leucine zipper/bZIP and helix-loop-helix motifs (62-65). The following similarities and differences are noteworthy.

(i) The basic arms of bZIP and the basic arm/helix-loop-helix motif domains are positioned by a structured dimerization element that does not itself contact DNA, either a parallel-coiled coil (the leucine zipper) or parallel four-helix bundle (62-65). In contrast, the present studies suggest that the DSX DM zinc module itself contacts DNA.

(ii) Among major groove DNA-binding proteins, conserved arginine and asparagine side chains play critical roles in DNA recognition, often making specific hydrogen bonds at the edge of base pairs or contacting the DNA backbone within an extended network of hydrogen bonds (58). The minor groove of DNA is by contrast more hydrophobic than the major groove and lacks as distinctive a pattern of base-specific functional groups (66).

(iii) We propose that the aliphatic portions of lysine and arginine side chains in DSX may make van der Waals interactions within a widened minor groove, whereas their basic groups can interact with the DNA backbone. Such interactions are seen in the minor groove T domain-DNA complex (67) and in one segment of the resolvase structure⁸ (68). Conserved glutamine residues in the DSX tail may contact either DNA bases or backbone. Base-glutamine contacts, although common among major groove protein-DNA complexes, have seldom been observed in minor groove complexes.⁹

(iv) CD difference features in the near-UV (250-320 nm), ascribed to changes in DNA structure, are observed in both major groove and minor groove protein-DNA complexes but differ in detail. Structural interpretation of such differences will require a more extensive data base of spectra and high resolution crystal structures.

We imagine that the DM-DNA complex, like other minor groove complexes, is stabilized by electrostatic interactions and hydrogen bonds within an overall framework of non-polar contacts. It is intriguing that the DSX tail contains a series of non-polar side chains (VMAL; DM consensus positions 48-51) between the two basic patches identified herein. It is not known whether the tail contributes to an induced dimer contact as well as to the protein-DNA interface. Although truncation of the tail blocks DNA-dependent dimerization, none of the residues tested herein are critical to the stability of the induced dimer interface. It is possible that impaired cooperativity of DM is secondary to its very weak DNA affinity.

The present study illustrates the complementary utility of biosynthetic expression and total protein synthesis by native ligation of peptide fragments (43). The latter synthetic methods may prove useful in structural genomics (69). This technology permits introduction of non-standard amino acids and may enable preparation of proteins or analogs refractory to biosynthetic expression. Cysteine-rich motifs are in particular attractive targets as the cysteines providing convenient sites of peptide ligation. Application to the DM motif allowed rapid preparation of the DSX analogs containing norleucine side chains as isosteric replacements for the aliphatic portions of lysine or arginine. The K60Z substitution impairs specific DNA binding, whereas K57Z is well tolerated. The R46Z analog exhibits structural perturbations, confounding local interpretation of its small change in DNA binding affinity. In the native NMR structure Arg-46 packs against His-50 in metal-binding site II and contacts the conserved aromatic side chain of Phe-65 to seal one edge of the hydrophobic core (21). In the future it will be of interest to investigate whether the aliphatic portion of this invariant arginine is integral to the stability of the metal-binding sites. An integrated understanding of structure-function relationships in the DM domain will require structural dissection of determinants of zinc-dependent protein folding, DNA recognition, and cooperativity.

	ACKNOWLEDGEMENTS

We thank S. B. Kent for advice regarding peptide synthesis and native ligation chemistry; N. B. Phillips for advice regarding protein purification; L. Han for assistance with DNA binding assays; Q.-X. Hua and W. Jia for assistance with CD and NMR spectroscopy; H. T. Keutmann for amino acid analysis; G. Reddy for mass spectrometry; R. Singh for gel quantification and assistance with the revised manuscript; E. Collins for preparation of the manuscript; and members of the Weiss laboratory for advice and discussion.

	FOOTNOTES

^* This work is a contribution from the Cleveland Center for Structural Biology and was supported in part by a grant from the National Institutes of Health (to M. A. W.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The atomic coordinates and the structure factors (code 1LPV) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).

^§ Both authors contributed equally to this work.

^¶ Present address: Array BioPharma, 3200 Walnut St., Boulder, CO 80301.

^** To whom correspondence should be addressed. Tel.: 216-368-5991; Fax: 216-368-3419; E-mail: weiss@biochemistry.cwru.edu.

Published, JBC Papers in Press, August 26, 2002, DOI 10.1074/jbc.M204616200

² Two different types of difference spectra are shown in Fig. 6. The first, designated 1 (D), is defined as the difference between the CD spectrum of the protein-DNA complex and the CD spectrum of the free DNA. The second, designated 2 (E), is defined as the difference between the spectrum of the complex and the sum of the spectra of the free DNA and free protein.

³ The NMR spectrum of DM contains unusually sharp resonances assigned to its truncated tail (residues 80-86). Motional narrowing of the foreshortened tail contrasts with conformational broadening of the intact tail due to intermediate exchange among nascent helical structures (21).

⁴ Use of a non-standard amino acid was motivated by our recent experience in studies of insulin wherein alanine scanning mutagenesis (71) was confounded at a key site by non-local structural perturbations (72). Such limitations were overcome by non-standard mutagenesis (73). Although it is not known whether alanine substitutions in the zinc module would likewise perturb its folding, the side chain of Arg-46 appears integral to the structure, and R46A would be predicted to create a crevice.

⁵ Attenuation of the 1:1 band in GMSA studies of alanine variants could be due either to enhancement of cooperativity or to kinetic instability of the variant 1:1 complexes in the course of electrophoresis. The latter could be an indirect consequence of a mutation at the protein-DNA interface.

⁶ Whereas DSX contains a single DM domain and binds to its DNA target site as a dimer (10), MAB-3 contains two DM domains and binds as a monomer (20). The MAB-3 DM domains are each required in vivo and are proposed to function coordinately in DNA binding as an "internal dimer" (21). It is not known whether each domain contacts DNA.

⁷ Although Wensink and co-workers (10) observed no discrete C1 band between the free probe and the dimeric complex, they noted that the presence of DNA radiolabel between these bands of integrated intensity was less than 5%. It is possible that these counts in part represent background due to rapid dissociation of a singly bound species.

⁸ The crystal structures of the T domain and resolvase each exhibit docking of an -helix within a widened minor groove (67, 68). Like the DM domain, the T domain does not induce sharp DNA bending. Neither complex employs glutamine at a minor groove interface.

⁹ HMG boxes contain glutamines that contact DNA phosphates or deoxyribose moieties (74-77). Sequence-specific HMG boxes contain a conserved asparagine (consensus position 10) whose carboxamide function contacts edges of base pairs in the minor groove. This interaction contributes to sequence specificity (7).

	ABBREVIATIONS

The abbreviations used are: HMG, high mobility group; bZIP, basic arm/leucine zipper motif; GMSA, gel mobility-shift assay; NOE, nuclear Overhauser enhancement; NOESY, NOE spectroscopy; rp-HPLC, reverse phase-high performance liquid chromatography; SRY, sex-determining region of the Y chromosome; DSX, Doublesex.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES