Portal de Periódicos da CAPES

Zinc Binding and Dimerization of Streptococcus pyogenes Pyrogenic Exotoxin C Are Not Essential for T-cell Stimulation

2003; Elsevier BV; Volume: 278; Issue: 11 Linguagem: Inglês

10.1074/jbc.m206957200

1083-351X

Wieslaw Swietnicki, Anne M. Barnie, Beverly K. Dyas, Robert G. Ulrich,

Neonatal and Maternal Infections

Streptococcal pyrogenic enterotoxin C (Spe-C) is a superantigen virulence factor produced byStreptococcus pyogenes that activates T-cells polyclonally. The biologically active form of Spe-C is thought to be a homodimer containing an essential zinc coordination site on each subunit, consisting of the residues His167, His201, and Asp203. Crystallographic data suggested that receptor specificity is dependent on contacts between the zinc coordination site of Spe-C and the β-chain of the major histocompatibility complex type II (MHCII) molecule. Our results indicate that only a minor fraction of dimer is present at T-cell stimulatory concentrations of Spe-C following mutation of the unpaired side chain of cysteine at residue 27 to serine. Mutations of amino acid residues His167, His201, or Asp203had only minor effects on protein stability but resulted in greatly diminished MHCII binding, as measured by surface plasmon resonance with isolated receptor/ligand pairs and flow cytometry with MHCII-expressing cells. However, with the exception of the mutants D203A and D203N, mutation of the zinc-binding site of Spe-C did not significantly impact T-cell activation. The mutation Y76A, located in a polar pocket conserved among most superantigens, resulted in significant loss of T-cell stimulation, although no effect was observed on the overall binding to human MHCII molecules, perhaps because of the masking of this lower affinity interaction by the dominant zinc-dependent binding. To a lesser extent, mutations of side chains found in a second conserved MHCII α-chain-binding site consisting of a hydrophobic surface loop decreased T-cell stimulation. Our results demonstrate that dimerization and zinc coordination are not essential for biological activity of Spe-C and suggest the contribution of an alternative MHCII binding mode to T-cell activation. Streptococcal pyrogenic enterotoxin C (Spe-C) is a superantigen virulence factor produced byStreptococcus pyogenes that activates T-cells polyclonally. The biologically active form of Spe-C is thought to be a homodimer containing an essential zinc coordination site on each subunit, consisting of the residues His167, His201, and Asp203. Crystallographic data suggested that receptor specificity is dependent on contacts between the zinc coordination site of Spe-C and the β-chain of the major histocompatibility complex type II (MHCII) molecule. Our results indicate that only a minor fraction of dimer is present at T-cell stimulatory concentrations of Spe-C following mutation of the unpaired side chain of cysteine at residue 27 to serine. Mutations of amino acid residues His167, His201, or Asp203had only minor effects on protein stability but resulted in greatly diminished MHCII binding, as measured by surface plasmon resonance with isolated receptor/ligand pairs and flow cytometry with MHCII-expressing cells. However, with the exception of the mutants D203A and D203N, mutation of the zinc-binding site of Spe-C did not significantly impact T-cell activation. The mutation Y76A, located in a polar pocket conserved among most superantigens, resulted in significant loss of T-cell stimulation, although no effect was observed on the overall binding to human MHCII molecules, perhaps because of the masking of this lower affinity interaction by the dominant zinc-dependent binding. To a lesser extent, mutations of side chains found in a second conserved MHCII α-chain-binding site consisting of a hydrophobic surface loop decreased T-cell stimulation. Our results demonstrate that dimerization and zinc coordination are not essential for biological activity of Spe-C and suggest the contribution of an alternative MHCII binding mode to T-cell activation. The 24.5-kDa Spe-C 1The abbreviations used are: Spe, streptococcal pyrogenic exotoxin; CDR, complementarity-determining region; FITC, fluorescein isothiocyanate; GdnHCl, guanidine hydrochloride; HLA, human leukocyte antigen; LB, Luria-Bertoni broth; LC-MS, liquid chromatography–mass spectrometry; MHCII, major histocompatibility complex type II; r wt, recombinant C27S mutant of Spe-C protein; SE, staphylococcal enterotoxin; TCR, T-cell receptor; TSST, toxic shock syndrome toxin; wt, wild type protein is an exotoxin produced by the pathogenic bacteriumStreptococcus pyogenes. Together with staphylococcal enterotoxin B (SEB), SEC, toxic shock syndrome toxin (TSST), and several others, Spe-C belongs to the superantigen protein family. Superantigens help pathogenic bacteria colonize the host by binding to major histocompatibility complex class II (MHCII) and T-cell receptors (TCR), thus bypassing the normal signal transduction pathway essential for immune recognition. In addition, the resulting nonspecific stimulation of the host's T-cells can induce pathological levels of cytokines. There is a substantial amount of structural data available for the superantigens, both free and in complex with their TCR and MHCII receptors (1Li H. Llera A. Malchiodi E.L. Mariuzza R. Annu. Rev. Immunol. 1999; 17: 435-466Google Scholar, 2Roussel A. Anderson B.F. Baker H.M. Fraser J.D. Baker E.N. Nat. Struct. Biol. 1997; 4: 635-643Google Scholar, 3Li Y. Li H. Dimasi N. McCormick J. Martin R. Schuck P. Schlievert P.M. Mariuzza R.A. Immunity. 2001; 14: 93-104Google Scholar, 4Arcus V.L. Proft T. Sigrell J.A. Baker H.M. Fraser J.D. Baker E.N. J. Mol. Biol. 2000; 299: 157-168Google Scholar, 5Sundstrom M. Abramsen L. Antonsson P. Mehindate K. Mourad W. Dohlsten M. EMBO J. 1996; 15: 6832-6840Google Scholar, 6Petersson K. Håkanson M. Nilsson H. Forsberg G. Svensson L.A. Liljas A. Walse B. EMBO J. 2001; 20: 3306-3312Google Scholar). The superantigen fold is highly conserved despite low overall sequence similarity among protein family members. Superantigen-binding sites are found on the β-chain or on the α-chain of the MHCII receptor. The β-chain-binding site requires zinc to form a complex with the superantigen through a conserved His from MHCII and three His or Asp residues from the superantigen. The absence of zinc in certain superantigens, such as SEB, presumably precludes β-chain binding. The α-chain-binding site requires a structurally conserved positioning of a hydrophobic binding loop contributed by the superantigen, and usually a second polar binding pocket is also engaged. For the case of Spe-C, the α-chain binding loop is displaced, and the potential α-chain-binding site, if the Spe-C were to bind MHCII as a monomer, may be hidden in the interface of the zinc-dependent superantigen dimer (2Roussel A. Anderson B.F. Baker H.M. Fraser J.D. Baker E.N. Nat. Struct. Biol. 1997; 4: 635-643Google Scholar), hypothetically resulting in interaction only with the MHCII β-chain. The β-chain binding residues are conserved in Spe-C and other streptococcal superantigens, and the β-chain-binding site (3Li Y. Li H. Dimasi N. McCormick J. Martin R. Schuck P. Schlievert P.M. Mariuzza R.A. Immunity. 2001; 14: 93-104Google Scholar, 4Arcus V.L. Proft T. Sigrell J.A. Baker H.M. Fraser J.D. Baker E.N. J. Mol. Biol. 2000; 299: 157-168Google Scholar) is intact in crystallographic models in which this interaction site with MHCII critically involves zinc. Zinc was implicated in streptococcal superantigen binding to MHCII on the surface of cells (7Li P.-L. Tiedemann R.E. Moffat S.L. Fraser J.D. J. Exp. Med. 1997; 186: 375-383Google Scholar, 8Proft T. Moffat S.L. Berkahn C.J. Fraser J.D. J. Exp. Med. 1999; 189: 89-101Google Scholar, 9Proft T. Arcus V.L. Handley V. Baker E.N. Fraser J.D. J. Immunol. 2001; 166: 6719Google Scholar), suggesting an essential role in both MHCII molecular recognition and TCR-mediated signal transduction. The high affinity zinc coordination complexes of SEA and Spe-C consist of three amino acid side chains contributed by the superantigen and the fourth by the MHCII receptor. As observed in structural data for the Spe-C tetrahedral zinc coordination complex (3Li Y. Li H. Dimasi N. McCormick J. Martin R. Schuck P. Schlievert P.M. Mariuzza R.A. Immunity. 2001; 14: 93-104Google Scholar, 4Arcus V.L. Proft T. Sigrell J.A. Baker H.M. Fraser J.D. Baker E.N. J. Mol. Biol. 2000; 299: 157-168Google Scholar, 5Sundstrom M. Abramsen L. Antonsson P. Mehindate K. Mourad W. Dohlsten M. EMBO J. 1996; 15: 6832-6840Google Scholar), His and/or Asp contribute the zinc ligands from the superantigen and His81 from the β-chain of MHCII molecule. The recently determined crystal structure of SEH with the MHCII HLA-DR1 indicates a potentially lower affinity zinc coordination site, present at the interface with the receptor (6Petersson K. Håkanson M. Nilsson H. Forsberg G. Svensson L.A. Liljas A. Walse B. EMBO J. 2001; 20: 3306-3312Google Scholar), where one of the ligands is a water molecule. The binding of SEH to MHCII is dependent on zinc, and superantigen potency is diminished when the zinc binding residues are replaced by alanines (10Nilsson H. Björk P. Dohlsten M. Antonsson P. J. Immunol. 1999; 163: 6683-6693Google Scholar). However, the MHCII β-chain mutation H81A, expected to abolish the zinc-mediated SEH binding to MHCII, had minimal effect (10Nilsson H. Björk P. Dohlsten M. Antonsson P. J. Immunol. 1999; 163: 6683-6693Google Scholar). In addition, the C-terminal zinc-binding site of another dimer-forming superantigen (11Bavari S. Ulrich R.G. Infect. Immun. 1995; 63: 423-429Google Scholar), SEC1, is not necessary for T-cell stimulation (12Chi Y.I. Sadler I. Jablonski L.M. Callantine S.D. Deobald C.F. Stauffacher C.V. Bohach G.A. J. Biol. Chem. 2002; 277: 22839-22846Google Scholar). In vitro data with purified recombinant Spe-C demonstrated a large proportion of dimer at 81 μm protein concentration (7Li P.-L. Tiedemann R.E. Moffat S.L. Fraser J.D. J. Exp. Med. 1997; 186: 375-383Google Scholar), yet the K d for dimer formation from equilibrium centrifugation data is only 390 μm (3Li Y. Li H. Dimasi N. McCormick J. Martin R. Schuck P. Schlievert P.M. Mariuzza R.A. Immunity. 2001; 14: 93-104Google Scholar) compared with T-cell stimulation occurring at nanomolar concentrations of superantigen. The mutant H35A, thought to affect dimer formation, had only minor effect on the K d (3Li Y. Li H. Dimasi N. McCormick J. Martin R. Schuck P. Schlievert P.M. Mariuzza R.A. Immunity. 2001; 14: 93-104Google Scholar). Finally, the Spe-C molecule forms a zinc-less dimer in the Spe-C/TCR Vβ 2.1 crystal structure (13Sundberg E.J. Li H. Llera A.S. McCormick J.K. Tormo J. Schlievert P.M. Karjalainen K. Mariuzza R.A. Structure. 2002; 10: 687-699Google Scholar), using a surface proposed to be involved in the zinc-mediated MHCII binding, a site far removed from the one anticipated to bury the conserved residues interacting with the MHCII α-chain (2Roussel A. Anderson B.F. Baker H.M. Fraser J.D. Baker E.N. Nat. Struct. Biol. 1997; 4: 635-643Google Scholar). Therefore, to further understand binding of Spe-C to MHCII and the contribution to T-cell stimulation, we have re-examined the relationship of zinc binding to stability, oligomerization, and biological function of the Spe-C superantigen. Our results suggest a diminished functional role for zinc and dimerization in T-cell stimulation. Further, an alternative zinc-independent-binding site, involving residues structurally equivalent to the MHCII α-chain-binding hydrophobic loop and polar pocket on SEB, may be biologically relevant for TCR stimulation. Binding to this site is only possible when the biologically active form of Spe-C is a monomer but not the homodimer. Genomic DNA was purified (Wizard Genomic DNA Isolation Kit; Promega) from a Spe-C+ clinical isolate of S. pyogenes grown in a culture. The cDNA corresponding to the wt Spe-C was amplified by PCR and cloned into pRSET A vector between NheI andHindIII sites, together with a linker coding for a thrombin cleavage site (Leu-Val-Pro-Arg*Gly-Ser) at the N terminus of Spe-C. The final construct coded for a fusion protein of His6 (pRSET A vector) followed by a thrombin linker attached to the N terminus of mature Spe-C (amino acids 28–235). The protein construct was designed to have a Gly-Ser N-terminal extension after thrombin cleavage. To avoid covalent dimer formation because of an intermolecular disulfide bridge, a C27S mutation was introduced by site-directed mutagenesis (QuikChange; Stratagene). The full open reading frame corresponding to the fusion protein was sequenced on a CEQ 2000XL (Beckman, Fullerton, CA) sequencer. For protein expression, the plasmid DNA was transformed into a BL21 (DE3) strain (Invitrogen). A single colony from a freshly streaked LB + Amp (100 μg/ml) plate was grown in a 3 ml of LB + Amp (100 μg/ml) medium at 37 °C in a shaker for ∼4 h until it became visibly turbid. A 0.5-ml aliquot was used to inoculate 25 ml of LB + Amp (100 μg/ml) medium and grown for an additional 4–5 h. The culture was then stored for 12 h at 4 °C. The next day, a 6-ml aliquot was added to 500 ml of LB + Amp (100 μg/ml) medium, and the culture was grown in a shaker at 37 °C until it reachedA 600 of 1.0. At that point, isopropyl-1-thio-β-d-galactopyranoside was added to a final concentration of 1 mm to induce protein expression, and the culture was grown overnight at 37 °C. Cells were harvested by centrifugation and stored at −80 °C until further processing. To purify protein, bacterial cells from a 2-liter culture were thawed at room temperature and resuspended in ∼120 ml of denaturing buffer (6 m GdnHCl, 100 mm potassium phosphate, 10 mm Tris-HCl, pH 8.0). Cells were disrupted by sonication on ice (3 × 1 min with a 1-min rest) using a Model 300 sonic dismembrator (Fisher Scientific, Hampton, NH) equipped with a ¼-in-diameter probe. Cellular debris was removed by centrifugation (2 h, 20,000 × g, 4 °C), and the supernatant was mixed with 25 ml of nickel-nitrilotriacetic acid Superflow resin (Qiagen) for 30 min to bind the protein. The resin suspension was used to pack a XK 26/20 column (Amersham Biosciences) in a batch mode, and the column was connected to a ÄKTA fast protein liquid chromatography system (Amersham Biosciences). Most cellular impurities were removed by washing with the denaturing buffer, and the protein was refolded by running a linear gradient of 100–0% of denaturing buffer and 100 mm potassium phosphate, 10 mm Tris-HCl, pH 8.0, over a 2-h time at a flow rate of 1 ml/min. Residual impurities were removed by a wash with 50 mm imidazole, 100 mm potassium phosphate, 10 mm Tris-HCl, pH 8.0, and the protein was eluted with 500 mm imidazole, 100 mm potassium phosphate, pH 5.8. The fractions containing protein, as judged by an SDS-PAGE analysis, were pooled and dialyzed against 10 mm Tris-HCl, 1 mm EDTA, pH 8.0. The His6-thrombin linker was removed by digestion (20 °C, 7 d) with 10 units of thrombin (Amersham Biosciences) per mg of protein. The residual thrombin was removed by incubation with 1 ml of Q-Sepharose (Amersham Biosciences) resin, and the supernatant was dialyzed against 10 mmpotassium phosphate, 1 mm EDTA, pH 5.8. The protein was bound to a SP-Sepharose (Amersham Biosciences) resin in a batch mode, and the suspension was used to pack a XK 26/20 column (AmershamBiosciences). The column was washed with 10 mm potassium phosphate, 1 mm EDTA, pH 5.8, buffer and developed with a linear gradient of 0–0.5 M NaCl at a 5 ml/min flow rate. Fractions containing protein, as judged by a SDS-PAGE analysis, were pooled and dialyzed against 10 mm HEPES, pH 7.0. The protein was concentrated in a Centriplus-3 (Millipore, Bedford, MA) concentrator, aliquoted, and stored at −80 °C. Protein concentration was determined by UV using molar extinction coefficients calculated with a ProtParam program on the ExPASy proteomics server at the Swiss Institute of Bioinformatics (Genève, Switzerland). Typically, 10–15 mg of purified protein at 0.5–2.5 mg/ml was obtained from 2L of bacterial cell culture. The purity of protein was verified by a combination of a SDS-PAGE and reversed-phase liquid chromatography. Proteins were characterized by LC-MS and, if needed, by an N-terminal amino acid sequencing. The final purity was greater than 95% as judged by LC-MS. The recombinant HLA-DR1 protein was prepared as described (24Sloan V.S. Cameron P. Porter G. Gammon M. Amaya M. Mellins E. Zaller D.M. Nature. 1995; 375: 802-806Google Scholar). The protein was stored in 20% glycerol at −20 °C at 5 mg/ml. A fresh aliquot was removed from the freezer and placed on ice directly before the experiments. Mutants were constructed by site-directed mutagenesis (QuikChange; Stratagene). The full open reading frames corresponding to Spe-C fusion proteins were sequenced on CEQ 2000XL DNA sequencer (Beckman), and the plasmids were transformed into an Escherichia coli BL21 (DE3) strain for protein expression. Bacterial cell growth and protein purification were identical to the r wt protein (above). All mutant proteins contained the C27S replacement. A protein sample volume corresponding to 0.5–1.0 mg of total protein was transferred to an empty sample container, weighed, and diluted to 5 ml with 2% nitric acid. The diluted sample was then analyzed by Hewlett Packard 4500 series inductively coupled plasma spectroscopy for zinc and nickel content. Detection limits for the method were 1 μg of metal per liter of solution. The analysis was performed on a Finnigan LCQ Deca LC-MS. Typically, a 15-μl aliquot of protein solution was injected on a reversed-phase Poros II R/H 8 × 100-mm column (LC Packings, San Francisco, CA) equilibrated in 0.1% formic acid. The column was developed with a linear 0–100% gradient of 0.1% formic acid/80% acetonitrile over a period of 30 min at the flow rate of 0.250 ml/min. Eluent was also monitored at A 280on Agilent 1100 series UV-visible detector (Agilent Technologies, Palo Alto, CA). The data were stored on a computer and processed with the instrument's software. A total of 250 μl of protein solution at a concentration of 0.2 mg/ml in running buffer (100 mm potassium phosphate, 50 mm NaCl, pH 7.0) was injected onto Superdex 200 HR 10/30 (Amersham Biosciences) gel filtration column connected to a ÄKTA fast protein liquid chromatography system (Amersham Biosciences). Protein separation at the flow rate of 0.5 ml/min was monitored by followingA 280 of the eluent. The column was calibrated with gel chromatography markers (Amersham Biosciences). Molecular weight calculations of mutant Spe-C proteins were determined from plots of the logarithm of molecular weight of standards versusretention volume. All calculations assumed a globular shape of proteins. All measurements were made on a J-810 spectropolarimeter (Jasco, Inc., Easton, MD) equipped with a Peltier unit to control temperature of sample holding block. Unless otherwise specified, the block temperature was maintained at 25 °C. Far-UV spectra were collected in a 1-mm path length rectangular cuvette. Typically, at 0.2 mg/ml protein solution in 50 mm potassium phosphate, pH 7.0, data from ten spectra were collected and averaged. The spectra were corrected for the buffer, smoothed with programs included with the Jasco software, and converted to mean residue ellipticity, MRE, according to Equation 1, where Θ is the measured ellipticity in millidegree,c the protein concentration in mol/liter, d is the path length in cm, and Na is the number of amino acid residues per molecule. MRE=Θ10cdNaEquation 1 Secondary structure estimates were performed on the buffer-corrected, unconverted data with a Neural Network program from the Softsec program suite (Softwood Software). Equilibrium unfolding experiments were performed with the Jasco titrator unit equipped with two 2.5-ml Hamilton syringes. Typically, about 10 ml of protein solution with concentration of 20 μg/ml in GdnHCl (5.4–5.6 M), 50 mm potassium phosphate, pH 7.0, was prepared, and about 2 ml was added in 50-μl increments to 2.7 ml of identical solution without GdnHCl in a 1-cm path length cuvette with constant mixing. The instrument and titrator parameters were as follows: titration steps, 40; mixing time after denaturant injection, 150 s; equilibration after solution withdrawal, 10 s; wavelength, 222 nm; response time, 8 s; average, two times; and bandwidth, 2 nm. Under these conditions, the signal recovery for refolding reaction was at least 90%. Data from unfolding were converted to ellipticity versus denaturant concentration by the Jasco software, transferred to Kaleidagraph program (Synergy Software, Reading, PA), and used for a non-linear fitting according to Equation 2, where Θ222 is the mean residue ellipticity at 222 nm, a is the slope, andb is the intercept of Θ versusdenaturant concentration of the native, N, and unfolded,U, states, R is the gas constant equal to 8.315 J/mol M deg K, T is the temperature in K, m is the slope of ΔG versus denaturant concentration, c, and Cm is the concentration of denaturant at which 50% of the protein is unfolded. Θ222=aN+bNc+(aU+bUc)expm(Cm−c)RT1+expm(Cm−c)RTEquation 2 The free energy of unfolding at denaturant concentration equal to zero, ΔGo, was calculated according to Equation 3. ΔGo=mCmEquation 3 All measurements were made in duplicate or triplicate. Data are reported as means ± S.D. between independent measurements. Isolated human blood mononuclear cells were cultured for three days in 96-well plates (3–5 × 105 cells/well) in medium containing 5% fetal bovine serum (Invitrogen) and pulsed with 1 μCi of [3H]thymidine (Amersham Biosciences) for a period of 9 h. SEA, SEB, and Spe-A were obtained from Toxin Technology, Inc. (Sarasota, FL). Cells were incubated in triplicate with bacterial toxins or recombinant Spe-C mutants and disrupted osmotically, and cellular debris was transferred to counting cassettes. Radioactivity was measured on a liquid scintillation counter (TopCount-NXT; Packard Instruments). Endotoxin level measurements were measured by a Limulus-lysate assay (QCL-1000; BioWhittaker, Walkersville, MD). All protein samples had levels less than 0.24 endotoxin units/liter of solution. Surface plasmon resonance measurements of ligand-receptor binding were performed on a Biacore 3000 (Biacore Inc., Piscataway, NJ). In a typical experiment, a solution of recombinant HLA-DR1 in 10 mmHEPES, pH 7.0, 150 mm NaCl was passed over a CM5 chip surface with immobilized superantigen (about 3000 refractive unit) at 20 μl/min for 3 min at 37 °C. The dissociation phase was followed for 2 min at the same flow rate, and the surface was regenerated with 10 mm EDTA and 2 m KCl. The sensogram was always corrected on a reference signal originating from a non-derivatized surface, measured in the same experiment. All data were processed for best fit using software supplied by the manufacturer and assuming a simple 1:1 Langmuire association model for the on-rate,kon, off-rate,koff, and dissociation constant,K d, calculations. A cell-based MHCII binding assay was also used. The human B-lymphoblastoid cell line LG2 (16Ulrich R.G. Bavari S. Olson M.A. Nat. Struct. Biol. 1995; 2: 554-559Google Scholar) was incubated with FITC-labeled r wt in Hanks' basic salt solution medium supplemented with 0.1% bovine serum albumin for 30 min at 37 °C. The cells were then washed with the medium, fixed with 1% paraformaldehyde in phosphate-buffered saline, and analyzed by a laser fluorescence-activated flow cytometry (BD Biosciences). Alternatively, mouse L cells expressing human DRα/DR1β (*B0101), prepared as previously described (11Bavari S. Ulrich R.G. Infect. Immun. 1995; 63: 423-429Google Scholar), were grown to 80% confluency. The cells were removed from tissue culture flasks with 25 mm EDTA in HBSS calcium- and magnesium-free medium, washed with Eagle's minimal essential medium supplemented with non-essential amino acids, and incubated with unlabeled r wt at the concentration range of 0.16-10 μm for 30 min at 37 °C. FITC-labeled r wt was then added to a final concentration of 1.25 μm, and the mixture was incubated for an additional 30 min at 37 °C. The cells were washed with Eagle's minimal essential medium, fixed with 1% paraformaldehyde in phosphate-buffered saline, and analyzed by laser fluorescence-activated flow cytometry (BD Biosciences). Protein sequences were downloaded from a public data base through the PubMed program (National Library of Medicine). All models were constructed by Swiss Model program (26Guex N. Peitsch M.C. Electrophoresis. 1997; 18: 2714-2723Google Scholar) on the ExPASy server at the Swiss Institute of Bioinformatics (Genève, Switzerland). The models optimized by the server and examined with the WHAT IF program (27Rodriguez R. Chinea G. Lopez N. Pons T. Vriend G. Comput. Appl. Sci. 1998; 14: 523-528Google Scholar) were either corrected manually or discarded if errors were too large. The crystallographic structure of SEH in complex with HLA-DR1 (6Petersson K. Håkanson M. Nilsson H. Forsberg G. Svensson L.A. Liljas A. Walse B. EMBO J. 2001; 20: 3306-3312Google Scholar) was used to validate the homology modeling method. There was a high degree of correlation between modeled and experimental structures, presenting root mean square deviation of 0.4 and 1.0 Å for the α-C backbone of the superantigen alone and the whole complex, respectively. Molecular models were built with the Flexidock program in the Biopolymer module of Sybyl 6.7 (Tripos Software, St. Louis, MO) molecular modeling program suite. Starting templates were constructed by overlaying the α-C backbones of Spe-C on SEB from the SEB/DR1 complex (19Jardetzky T.S. Brown J.H. Gorga J.C. Stern L.J. Urban R.G. Chi Y.I. Stauffacher C. Strominger J.L. Wiley D.C. Nature. 1994; 368: 711-718Google Scholar). The Flexidock program uses a variation of a genetic algorithm (27Rodriguez R. Chinea G. Lopez N. Pons T. Vriend G. Comput. Appl. Sci. 1998; 14: 523-528Google Scholar) that employs a combination of rigid body and torsional space search. Final solutions are scored by the fitness function, which includes van der Waals, electrostatic, and torsional energy terms of the default Tripos force field, with the following modifications: hydrogen van der Waals radius, 1 Å; hydrogen bond epsilon, 0.03; and van der Waals cutoff distance for centroids, 16 Å. The strategy was used successfully for docking other proteins or small ligands (see Refs. 28Willet P. Trends Biotechnol. 1995; 13: 516-521Google Scholar, 29Meegan M.J. Hughes R.B. Lloyd D.G. Williams C.D. Zisterer D.M. J. Med. Chem. 2001; 44: 1072-1084Google Scholar, 30Dhar A. Liu S. Klucik J. Berlin D. Madler M.M. Lu S. Ivey R.T. Zacheis D. Brown C. Nelson E.C. Birkbichler P.J. Benbrook D.M. J. Med. Chem. 1999; 42: 3602-3614Google Scholar, 31Bertelli M. El-Bastawissy E. Knaggs M.H. Barrett M.P. Hanau S. Gilbert H. J. Comput. Aided Mol. Des. 2001; 15: 465-475Google Scholar). Because of the random nature of the genetic algorithm's initial search, we repeated all docking calculations for Spe-C. All solutions were the same as in the first simulation. Therefore, the structures presented represent converged, calculated structures. A typical run for docking of Spe-C or other superantigens to DR1 used default parameter settings with a 10-Å search radius around the starting binding pocket to maximize the search space. Contributions from electrostatics and all hydrogens were always included in the search parameters. A default initial seed number was used, but the total number of generations of Flexidock search was always limited to 3000 to avoid bias toward the best scoring solution. The following two approaches were selected for docking of Spe-C to DR1: 1) rigid body docking without flexing side chains; 2) all side chains of ligand and receptor were considered adjustable. Although the first strategy resulted in many solutions, all but one was considered unacceptable based on steric problems in the final models. The second strategy resulted in multiple solutions, which included, for most models, the residues structurally equivalent to the amino acids known to be critical for SEB binding to MHCII. The highest scoring solution from the second strategy was chosen for further optimization. The side-chain conformations for all residues were adjusted within the Biopolymer module, and the final models were energy-minimized with 20 steps of Simplex and 100 steps of a gradient (Pullman's method), using the Tripos force field (32Clark M. Cramer III, R.D. Opdenbosch N.V. J. Comp. Chem. 1989; 10: 982-1012Google Scholar). To test the docking strategy, we built models of Spe-A/DR1, SEA/DR1, and SEC3/DR1 complexes. Analysis of the possible solutions revealed high scoring models, as judged by total energy and minimal steric problems, with potential binding residues that were structurally equivalent to SEB amino acids critical for MHCII binding. Predictions from the SEA/DR1 complex model were verified by previously published experimental data (16Ulrich R.G. Bavari S. Olson M.A. Nat. Struct. Biol. 1995; 2: 554-559Google Scholar). Based on these test results, the strategy was assumed to be valid for searching for possible binding modes of Spe-C to DR1. The Spe-C superantigen was cloned from a clinical isolate ofS. pyogenes. r wt and mutants thought to be important for binding to MHCII molecules (Fig. 1) (discussed below) were also constructed. To avoid dimerization because of covalent, intermolecular disulfide formation, all proteins had the mutation C27S introduced. The C27S mutation did not affect Spe-C biological function (discussed later) and allowed us to accurately measure the chemical stability and propensity to form a non-covalent dimer. Recombinant proteins were expressed in E. coli as inclusion bodies and refolded on-column under non-reducing conditions. The proteins were homogeneous as judged by SDS-PAGE (data not shown). Similar results were obtained when the mutants were analyzed by LC-MS, plotting the A 280 and total ion current as a function of retention time (data not shown). The summary of mas