Advertisement

Solution Structure of Enzyme IIAChitobiose from the N,N′-Diacetylchitobiose Branch of the Escherichia coli Phosphotransferase System*

  • Chun Tang
    Affiliations
    Laboratories of Chemical Physics, National Institutes of Health, Bethesda, Maryland 20892-0520
    Search for articles by this author
  • David C. Williams Jr.
    Affiliations
    Laboratories of Chemical Physics, National Institutes of Health, Bethesda, Maryland 20892-0520
    Search for articles by this author
  • Rodolfo Ghirlando
    Affiliations
    Laboratories of Molecular Biology, NIDDK, National Institutes of Health, Bethesda, Maryland 20892-0520
    Search for articles by this author
  • G. Marius Clore
    Correspondence
    To whom correspondence should be addressed: Laboratory of Chemical Physics, Bldg. 5, Rm. B1–30I, NIDDK, National Institutes of Health, Bethesda, MD 20892-0520. Tel.: 301-496-0782; Fax: 301-496-0825
    Affiliations
    Laboratories of Chemical Physics, National Institutes of Health, Bethesda, Maryland 20892-0520
    Search for articles by this author
  • Author Footnotes
    * This work was supported in part by the AIDS Targeted Anti-Viral Program of the Office of the Director of the National Institutes of Health (to G. M. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Open AccessPublished:January 14, 2005DOI:https://doi.org/10.1074/jbc.M414300200
      The solution structure of trimeric Escherichia coli enzyme IIAChb (34 kDa), a component of the N,N′-diacetylchitobiose/lactose branch of the phosphotransferase signal transduction system, has been determined by NMR spectroscopy. Backbone residual dipolar couplings were used to provide long range orientational restraints, and long range (|i - j| ≥ 5 residues) nuclear Overhauser enhancement restraints were derived exclusively from samples in which at least one subunit was 15N/13C/2H/(Val-Leu-Ile)-methyl-protonated. Each subunit consists of a three-helix bundle. Hydrophobic residues lining helix 3 of each subunit are largely responsible for the formation of a parallel coiled-coil trimer. The active site histidines (His-89 from each subunit) are located in three symmetrically placed deep crevices located at the interface of two adjacent subunits (A and C, C and B, and B and A). Partially shielded from bulk solvent, structural modeling suggests that phosphorylated His-89 is stabilized by electrostatic interactions with the side chains of His-93 from the same subunit and Gln-91 from the adjacent subunit. Comparison with the x-ray structure of Lactobacillus lactis IIALac reveals some substantial structural differences, particularly in regard to helix 3, which exhibits a 40° kink in IIALacversus a 7° bend in IIAChb. This is associated with the presence of an unusually large (230-Å3) buried hydrophobic cavity at the trimer interface in IIALac that is reduced to only 45 Å3 in IIAChb.
      The bacterial phosphotransferase system (PTS)
      The abbreviations used are: PTS, phosphotransferase system; Chb, N,N′-diacetylchitobiose; HPr, histidine-containing phosphocarrier protein; IIAChb, IIBChb, and IICChb, A, B, and C domains, respectively, of the N,N′-diacetylchitobiose transporter IIChb; IIAChb*, double mutant of IIAChb comprising a 13-residue deletion at the N-terminus and an Asp to Leu mutation at position 92 (of the double mutant); NOE, nuclear Overhauser effect; HSQC, heteronuclear single quantum coherence; TROSY, transverse relaxation optimized spectroscopy; r.m.s., root mean square.
      1The abbreviations used are: PTS, phosphotransferase system; Chb, N,N′-diacetylchitobiose; HPr, histidine-containing phosphocarrier protein; IIAChb, IIBChb, and IICChb, A, B, and C domains, respectively, of the N,N′-diacetylchitobiose transporter IIChb; IIAChb*, double mutant of IIAChb comprising a 13-residue deletion at the N-terminus and an Asp to Leu mutation at position 92 (of the double mutant); NOE, nuclear Overhauser effect; HSQC, heteronuclear single quantum coherence; TROSY, transverse relaxation optimized spectroscopy; r.m.s., root mean square.
      is a prototypical signal transduction network in which translocation of sugars across the cytoplasmic membrane is coupled to phosphorylation (
      • Kundig W.
      • Ghosh S.
      • Roseman S.
      ,
      • Postma P.W.
      • Lengeler J.W.
      • Jacobson G.R.
      ,
      • Robillard G.T.
      • Broos J.
      ,
      • Siebold C.
      • Flükiger K.
      • Beutler R.
      • Erni B.
      ). There are four branches of the pathway corresponding to four classes of enzymes II, glucose, mannose, mannitol, and lactose/chitobiose, which bear no sequence similarity to one another and for the most part no structural similarity either (
      • Postma P.W.
      • Lengeler J.W.
      • Jacobson G.R.
      ,
      • Robillard G.T.
      • Broos J.
      ,
      • Siebold C.
      • Flükiger K.
      • Beutler R.
      • Erni B.
      ). The initial steps of the PTS are common to all branches of the pathway and involve the transfer of a phosphoryl group from phosphoenolpyruvate to the histidine phosphocarrier protein, HPr, via the intermediary enzyme I. Subsequently the phosphoryl group is transferred to sugar-specific enzymes II. Each enzyme II consists of two cytoplasmic domains, IIA and IIB, and a transmembrane sugar permease IIC, which may or may not be covalently linked. IIA accepts the phosphoryl group from HPr and donates it to IIB; IIC catalyzes the transport of the sugar across the membrane and its phosphorylation by IIB. Structures of many of the individual cytoplasmic components of the PTS have been solved by either NMR (
      • Wittekind M.
      • Rajgopal P.
      • Cranchini B.R.
      • Reizer J.
      • Saier M.H.
      • Klevit R.E.
      ,
      • Kalbitzer H.R.
      • Hengstenberg W.
      ,
      • van Nuland N.A.J.
      • Hangyi I.W.
      • van Shaik R.C.
      • Berendsen H.J.C.
      • van Gunsteren W.F.
      • Scheek R.M.
      • Robillard G.T.
      ,
      • Ab E.
      • Schuurman-Wolters G.K.
      • Reizer J.
      • Saier M.H.
      • Dijkstra M.
      • Scheek R.M.
      • Robillard G.T.
      ,
      • Ab E.
      • Shuurman-Walters G.K.
      • Nijlant D.
      • Dijkstra K.
      • Saier M.H.
      • Robillard G.T.
      • Scheek R.M.
      ,
      • Garrett D.S.
      • Seok Y.-J.
      • Liao D.-I.
      • Peterkofsky A.
      • Gronenborn A.M.
      • Clore G.M.
      ,
      • Legler P.M.
      • Cai M.
      • Peterkofsky A.
      • Clore G.M.
      ) or crystallography (
      • Liao D.-I.
      • Kapadia G.
      • Reddy P.
      • Saier M.H.
      • Reizer J.
      • Herzberg O.
      ,
      • Worthylake D.
      • Meadow N.D.
      • Roseman S.
      • Liao D.-I.
      • Herzberg O.
      • Remington S.J.
      ,
      • Herzberg O.
      • Reddy P.
      • Sutrina S.
      • Saier M.H.
      • Reizer J.
      • Kapadia G.
      ,
      • Jia Z.
      • Quail J.W.
      • Waygood E.B.
      • Delbaere L.T.
      ,
      • Liao D.-I.
      • Silverton E.
      • Seok Y.-J.
      • Lee B.R.
      • Peterkofsky A.
      • Davies D.R.
      ,
      • Nunn R.S.
      • Markovic-Housley Z.
      • Genovesio-Taverne G.
      • Flükiger K.
      • Rizkallah P.J.
      • Jansonius J.N.
      • Schirmer T.
      • Erni B.
      ,
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ,
      • Van Montfort R.L.
      • Pijning T.
      • Kalk K.H.
      • Reizer J.
      • Saier M.J.
      • Thunnissen M.M.
      • Robillard G.T.
      • Dijkstra B.W.
      ,
      • Van Montfort R.L.
      • Pijning T.
      • Kalk K.H.
      • Hangyi I.W.
      • Kouwijzer M.L.C.E.
      • Robillard G.T.
      • Dijkstra B.W.
      ,
      • Schauder S.
      • Nunn R.S.
      • Lanz R.
      • Erni B.
      • Schirmer T.
      ,
      • Orniss G.L.
      • Erni B.
      • Schirmer T.
      ), and more recently we have reported the solution NMR structures of a number of protein-protein complexes of the PTS (
      • Garrett D.S.
      • Seok Y.-J.
      • Peterkofsky A.
      • Gronenborn A.M.
      • Clore G.M.
      ,
      • Wang G.
      • Louis J.M.
      • Sondej M.
      • Seok Y.-J.
      • Peterkofsky A.
      • Clore G.M.
      ,
      • Cornilescu G.
      • Lee B.R.
      • Cornilescu C.
      • Wang G.
      • Peterkofsky A.
      • Clore G.M.
      ,
      • Cai M.
      • Williams D.C.
      • Wang G.
      • Lee B.R.
      • Peterkofsky A.
      • Clore G.M.
      ). As part of our continuing structural work on the PTS, we present the solution structure of Escherichia coli N,N′-diacetylchitobiose-specific enzyme IIA (IIAChb) using multidimensional NMR spectroscopy.
      The E. coli N,N′-diacetylchitobiose-specific enzymes II (IIChb) are part of the lactose/chitobiose branch of the PTS (
      • Keyhani N.O.
      • Roseman S.
      ). The A, B, and C components of IIChb are encoded by a single operon and expressed as individual proteins (
      • Keyhani N.O.
      • Roseman S.
      ). NMR (
      • Ab E.
      • Schuurman-Wolters G.K.
      • Reizer J.
      • Saier M.H.
      • Dijkstra M.
      • Scheek R.M.
      • Robillard G.T.
      ,
      • Ab E.
      • Shuurman-Walters G.K.
      • Nijlant D.
      • Dijkstra K.
      • Saier M.H.
      • Robillard G.T.
      • Scheek R.M.
      ) and crystal structures of E. coli IIBChb have been solved and bear surprising similarity to the structure of IIBMannitol (
      • Legler P.M.
      • Cai M.
      • Peterkofsky A.
      • Clore G.M.
      ), as well as to that of the low molecular weight eukaryotic protein tyrosine phosphatases (
      • Zhang M.
      • Van Etten R.L.
      • Stauffacher C.V.
      ,
      • Su X.-D.
      • Taddei N.
      • Stefani M.
      • Ramponi G.
      • Nordlund P.
      ), despite the absence of any significant sequence identity. No structure, however, has been determined as yet for E. coli IIAChb. The analogous system in Lactobacillus and Staphylococcus comprises the lactose-specific enzymes II (IILac) (
      • Hays J.B.
      • Simoni R.D.
      • Roseman S.
      ,
      • de Vos W.M.
      • Boerrigter I.
      • van Rooyen R.J.
      • Reiche B.
      • Hengstenberg W.
      ). The crystal structure of IIALac from Lactobacillus lactis has been solved and been shown to consist of a symmetric homotrimer (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ): each subunit comprises three antiparallel helices; the trimer interface consists of a parallel coiled-coil formed by the C-terminal helix from each subunit; and the trimer interface is partially stabilized by a metal ion coordinated to the side chains of three buried, symmetry-related, aspartate residues, one from each subunit. The crystal structure of IIALac, however, displays a highly unusual feature in the form of a very large (230-Å3), completely buried cavity at the trimer interface that is associated with the presence of a ∼40° kink in helix 3 and accommodates the heavy atom derivative trimethyl lead acetate (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ). E. coli IIAChb and L. lactis IIALac share 35% sequence identity and 63% amino acid similarity with no gaps or insertions (Fig. 1). In addition, binding of divalent cations to E. coli IIAChb significantly enhances its thermostability relative to the apoform, increasing in the order Mg2+, Cu2+, and Ni2+ (
      • Keyhani N.O.
      • Boudker O.
      • Roseman S.
      ). It has also been reported, on the basis of analytical ultracentrifugation data, that E. coli IIAChb is dimeric in solution (
      • Keyhani N.O.
      • Rodgers M.E.
      • Demeler B.
      • Hansen J.C.
      • Roseman S.
      ) in clear contrast to the trimeric state of IIALac (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ). Since the three symmetrically related active site histidines in IIALac are located in a crevice formed by the interface of two adjacent subunits (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ), this result would imply that the surface topology of the active sites and the spatial relationships of side chains within the active sites are very different in IIALac and IIAChb. If true, this would be highly unexpected given the close sequence and functional relationship between IIALac and IIAChb. Intrigued by this discrepancy, we initiated our own investigation of the oligomerization state and three-dimensional solution structure of E. coli IIAChb.
      Figure thumbnail gr1
      Fig. 1Sequence alignment of E. coli IIAChb and L. lactis IIALac. The residue shown at position 92 of IIAChb is that of the IIAChb-NΔ13D92L double mutant (IIAChb*) used for structural studies; both wild-type IIAChb and L. lactis IIALac have an Asp residue at this position. The N-terminal 13 residues that were deleted from wild-type IIAChb are colored in green; residue 92, the site of the engineered Asp to Leu mutation, is colored in bold red. Identical residues between IIAChb and IIALac are shaded in yellow, closely similar ones are in magenta, and remotely similar ones are in cyan.

      EXPERIMENTAL PROCEDURES

      Protein Purification and Mutagenesis—The plasmid for wild-type E. coli IIAChb was a kind gift from Dr. Saul Roseman (The Johns Hopkins University, Baltimore, MD). The plasmid was propagated in TOP10 cells, and IIAChb was expressed in BL21-Star cells (Invitrogen) at 37 °C. After induction with isopropyl β-d-thiogalactopyranoside, cells were harvested and microfluidized in 10 mm Tris·HCl, pH 7.5, buffer. The protein was purified using a DEAE-Sepharose anion-exchange column followed by a preparative Superdex-75 size exclusion column (Amersham Biosciences). IIAChb was denatured with 6 m guanidine HCl solution (pretreated with Chelex; Bio-Rad) at pH 3 and incubated at 37 °C overnight to remove residual phosphoryl groups on the protein. The pH of the denatured protein solution was then raised by the addition of 1 m Tris·HCl buffer. The protein solution was then dialyzed against either 10 mm sodium phosphate buffer (pH 6.5) or 10 mm Tris·HCl buffer (pH 8) containing 100 μm divalent cation to remove guanidine HCl and refold the protein; seven metal ions, Mg2+, Ca2+, Mn2+, Co2+, Ni2+, Cu2+, and Zn2+, were tested individually, yielding seven different samples. Excess metal ions in each metal-loaded protein sample were washed away by extensive buffer exchange in a concentrator, and the protein was further concentrated.
      The N-terminal deletion mutation (IIAChb-NΔ13) in which the first 13 residues were removed and the subsequent point mutation, IIAChb-NΔ13/D92L, were introduced using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA). The primers were designed according to the manufacturer's instructions. The sequences of the two mutants (IIAChb-NΔ13 and IIAChb-NΔ13/D92L) were confirmed by DNA sequencing (Davis Sequencing, Davis, CA). The two mutant proteins were purified using the same procedure described above for the wild-type protein. Typically 50 mg of protein were obtained from a 1-liter culture. The masses of the unlabeled proteins (wild type, NΔ13, and NΔ13/D92L) were confirmed by electrospray ionization mass spectrometry. The double mutant NΔ13/D92L is referred to hereafter as IIAChb*.
      (Ile/Leu/Val)-methyl-protonated and otherwise fully deuterated IIAChb-NΔ13/D92L (abbreviated as ILV-IIAChb*) was expressed based on the protocol described previously (
      • Goto N.K.
      • Gardner K.H.
      • Mueller G.A.
      • Willis R.C.
      • Kay L.E.
      ). Briefly cells were grown in M9 minimal medium prepared in D2 O with [2H7, 13C6]glucose and [15N]NH4Cl as the carbon and nitrogen sources, respectively. 80 mg of α-[13C5,3-2H1]ketoisovalerate and 50 mg of α-[13C4,3,3-2H2]ketobutyrate (Cambridge Isotopes, Andover, MA) were added into 1 liter of culture 45 min prior to induction at an A600 nm of ∼0.6. After induction with isopropyl β-d-thiogalactopyranoside, cells were grown under vigorous shaking for another 4 h before harvesting. Purified in H2O buffer, the protein sample carries protons only at backbone and side-chain amides, γ-methyls of Val residues, and δ-methyl(s) of Ile/Leu residues.
      Light Scattering—Static light-scattering data were collected using analytical size exclusion chromatography on a Superdex-75 column (Amersham Biosciences) in tandem with DAWN EOS light scattering and refractive index detectors (Wyatt Technology, Santa Barbara, CA). 100 μl of 75 μg of protein was applied to the pre-equilibrated Superdex-75 column at a flow rate of 0.5 ml/min at room temperature (20 °C). The running buffer consisted of either 10 mm sodium phosphate buffer, pH 6.5, or 10 mm Tris·HCl buffer, pH 7.5, containing 0.02% NaN3, 1 mm methionine, and 300 mm NaCl. The protein elution profile was monitored by the refractive index detector, and light-scattering measurements were made every 4 μl for a total elution volume of 20 ml. The data were analyzed using the manufacturer's proprietary software.
      The translational diffusion coefficient of the IIAChb-NΔ13/D92L double mutant (IIAChb*) was determined from autocorrelation analysis of quasielastically scattered light. Autocorrelation functions were collected on a BI-9000 AT autocorrelator (Brookhaven Instruments, Long Island, NY) at an angle of 90° with sampling times ranging from 0.5 μs to 10 ms. The diffusion coefficient, D20,w, was derived from autocorrelation functions of 10 independent measurements using the software provided by Brookhaven Instruments. The corresponding Stokes radius, Rs, was calculated using the equation Rs = kT/6πηD20,w where η represents the solvent viscosity, T is the absolute temperature, and k is the Boltzmann constant. The predicted translational diffusion coefficient and Stokes radius for the crystal structure of L. lactis IIALac (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ) were calculated using the program HYDROPRO (
      • Garcia de la Torre J.
      • Huertas M.L.
      • Carrasco B.
      ).
      NMR Data Collection and Analysis—All NMR samples were prepared in 10 mm sodium phosphate buffer, pH 6.5, containing 0.02% NaN3 and 1 mm methionine. NMR spectra were collected at 30 °C on Bruker DMX500, DMX600, DRX600, DMX750, and DRX800 spectrometers equipped either with x,y,z-shielded gradient triple resonance probes or z-shielded gradient triple resonance cryoprobes. Spectra were processed with the NMRPipe/nmrDraw suite (
      • Delaglio F.
      • Grzesiek S.
      • Vuister G.W.
      • Zhu G.
      • Pfiefer J.
      • Bax A.
      ) and analyzed using the PIPP/CAPP/STAPP package (
      • Garrett D.S.
      • Powers R.
      • Gronenborn A.M.
      • Clore G.M.
      ). Sequential resonance assignments were derived from analysis of transverse relaxation optimized (TROSY)-based triple resonance three-dimensional NMR experiments (
      • Clore G.M.
      • Gronenborn A.M.
      ,
      • Bax A.
      • Grzesiek S.
      ,
      • Yang D.W.
      • Kay L.E.
      ,
      • Clore G.M.
      • Gronenborn A.M.
      ), including HNCO, HN(CO)CA, HNCA, HNCB, and HN(CO)CB recorded on the ILV-IIAChb* sample. Nearly complete side-chain assignments were obtained from analysis of a three-dimensional HCCH-TOCSY experiment recorded on a 13C, 15N-labeled IIAChb* sample. A triple resonance CBCA(CO)NH was also recorded on the 13C, 15N-labeled IIAChb* sample to obtain 13Cα/13Cβ chemical shifts free of the offsets resulting from perdeuteration.
      Backbone φ/Ψ torsion angle restraints were derived from backbone 1H, 15N, and 13C chemical shifts using the program TALOS (
      • Cornilescu G.
      • Delaglio F.
      • Bax A.
      ). Side-chain torsion angle restraints were derived from 3JNCγ, 3JC′Cγ, and 3JCαCδ coupling constants measured using quantitative J correlation experiments (
      • Bax A.
      • Vuister G.W.
      • Grzesiek S.
      • Delaglio F.
      • Wang A.C.
      • Tschudin R.
      • Zhu G.
      ).
      Intersubunit nuclear Overhauser enhancements (NOEs) were obtained from a three-dimensional 13C-separated/12C-filtered NOE spectrum collected on a sample comprising a 1:1 mixture of unlabeled IIAChb* and labeled ILV-IIAChb* (
      • Clore G.M.
      • Gronenborn A.M.
      ). To ensure complete mixing of the subunits, the 1:1 mixture of the labeled and unlabeled proteins was denatured in 6 m guanidine HCl and then refolded into the NMR buffer (10 mm sodium phosphate buffer, pH 6.5, 0.02% NaN3, and 1 mm methionine). Intramolecular long range NOEs were obtained from four-dimensional 13C/15N-separated and 13C/13C-separated NOE spectra collected on the ILV-IIAChb* sample. As methyl-methyl NOE interactions in the unfiltered NOE spectra contain both intra- and intermolecular information, care was taken to ensure that assigned intramolecular NOE cross-peaks did not appear in the three-dimensional 13C-separated/12C-filtered NOE spectrum. NOEs involving backbone amide protons were obtained from a three-dimensional 15N-separated NOE spectrum.
      Backbone 1DNH, 1DNC′, and 1DCαC′ residual dipolar couplings were obtained from the difference between the 1J scalar couplings measured in dilute liquid crystalline (15 mg/ml phage pf1 (
      • Clore G.M.
      • Starich M.R.
      • Gronneborn A.M.
      ,
      • Hansen M.R.
      • Hanson P.
      • Pardi A.
      )) and isotropic (water) media. 1JNH and 1JNC′ couplings were measured from the splittings in three-dimensional HNCO-TROSY-based experiments (
      • Yang D.
      • Venters R.A.
      • Mueller G.A.
      • Choy W.Y.
      • Kay L.E.
      ), and 1JCαC′ couplings were derived from an intensity-modulated HN(CO)CATROSY experiment (
      • Jaroniec C.P.
      • Ulmer T.S.
      • Bax A.
      ).
      Structure Calculation—NOE-derived interproton distance restraints were classified into distance ranges of 1.8–2.7, 1.8–3.3, 1.8–5.0, and 1.8–6 Å, corresponding to strong, medium, weak, and very weak NOE cross-peak intensities, respectively. An additional 0.5 Å was added to the upper distance bound of distance restraints involving methyl groups (0.5 Å per methyl group), and 0.2 Å was added to the upper bounds for strong and medium NOE restraints involving amide protons. Nonstereospecifically assigned methyl protons and ambiguous intermolecular NOEs were represented by a (Σr-6)-1/6 sum. The error ranges used for the torsion angle restraints (which are represented by square-well potentials) are ±20° for the backbone φ/Ψ angles within helical regions and ±20° for χ1 and ±30° for χ2 side-chain torsion angles. χ1 restraints for aliphatic side chains that are not in the t rotamer and not experiencing rotamer averaging (3JCγN < 1.0 Hz) were set to 0 ± 80°.
      Structures were calculated using a well established protocol (
      • Nilges M.
      • Gronenborn A.M.
      • Brünger A.T.
      • Clore G.M.
      ,
      • Garrett D.S.
      • Kuszewski J.
      • Hancock T.J.
      • Lodi P.J.
      • Vuister G.W.
      • Gronenborn A.M.
      • Clore G.M.
      ,
      • Kuszewski J.
      • Qin J.
      • Gronenborn A.M.
      • Clore G.M.
      ,
      • Clore G.M.
      • Gronenborn A.M.
      • Tjandra N.
      ) from the experimental NMR restraints by simulated annealing in torsion angle space (
      • Schwieters C.D.
      • Clore G.M.
      ) using the program Xplor-NIH (
      • Schwieters C.D.
      • Kuszewski J.
      • Tjandra N.
      • Clore G.M.
      ). The coordinates of the three subunits were restrained to their average positions (after best fitting) by a non-crystallographic symmetry restraint. The non-bonded contacts in the target function were represented by a quartic van der Waals repulsion term (
      • Nilges M.
      • Gronenborn A.M.
      • Brünger A.T.
      • Clore G.M.
      ) supplemented by multidimensional torsion angle (
      • Clore G.M.
      • Kuszewski J.
      ) and backbone hydrogen bonding (
      • Grishaev A.
      • Bax A.
      ) data base potentials of mean force. A radius of gyration restraint was used to ensure optimal packing (
      • Kuszewski J.
      • Gronenborn A.M.
      • Clore G.M.
      ). Structure figures were generated with the programs VMD-XPLOR (
      • Schwieters C.D.
      • Clore G.M.
      ) and Pymol (
      • DeLano W.L.
      ). Reweighted atomic density probability maps were calculated from the structure ensemble as described previously (
      • Schwieters C.D.
      • Clore G.M.
      ). Modeling of the phosphorylated state was carried out by manually docking a phosphoryl group in the vicinity of the Nϵ-1 atom of the active site His-89 of the restrained regularized mean structure in VMD-XPLOR followed by regularization using Xplor-NIH, keeping all coordinates fixed with the exception of the side chain of His-89 and the phosphoryl group.

      RESULTS AND DISCUSSION

      Engineering a Monodisperse Trimer—1H-15N heteronuclear single quantum coherence (HSQC) correlation spectra of 15N-labeled IIAChb in both the apoform and loaded with diverse divalent cations, including Mg2+, Ca2+, Mn2+, Co2+, Ni2+, Cu2+, and Zn2+, are all broad and poorly dispersed, indicative of chemical exchange between various states. Of these various samples, Ni2+-charged IIAChb displayed the best spectrum (Fig. 2A), suggesting slightly more favorable exchange dynamics, consistent with the prior observation that Ni2+ has the largest effect in enhancing the thermostability of IIAChb (
      • Keyhani N.O.
      • Boudker O.
      • Roseman S.
      ).
      Figure thumbnail gr2
      Fig. 2NMR spectral characterization of IIAChb wild-type and mutant proteins.1H-15N HSQC spectra of 15N-labeled wild-type IIAChb loaded with Ni2+ (A), the 15N-labeled IIAChb-NΔ13 deletion mutant loaded with Ni2+ (B), and the 15N-labeled IIAChb-NΔ13/D92L double mutant (IIAChb*) (C). Outlying cross-peaks in the spectrum shown in A that match closely to those in C are tentatively assigned in A and B and are labeled in all three spectra. In addition, the cross-peak position for the C-terminal residue, Ala-116, is essentially identical for all three constructs and is located at 130.9 ppm in the 15N dimension (not shown).
      Ni2+-charged IIAChb was then subjected to light-scattering analysis. The intensity of light scattered by a molecule is directly proportional to molar mass; thus static light scattering used in tandem with analytical size exclusion chromatography can measure the average molecular weight of the eluted fractions on the fly. Light-scattering analysis reveals that wildtype Ni2+-IIAChb is a mixture of many species of different molecular masses ranging from 35 to 70 kDa, and there is no predominant species of defined molecular weight in the mixture (Fig. 3A, blue symbols). This result fully accounts for the poor quality of the 1H-15N correlation spectrum of Ni2+-IIAChb (Fig. 2A). Significantly, even at the leading edge of the gel filtration peak where the protein concentration is only ∼140 nm, the molecular mass never falls below 3 times the monomer molecular mass (12,747.7 Da), which is the molecular mass of a trimer. Further, as increasing concentration promotes higher order oligomerization, our light-scattering results are not consistent with the previous report that IIAChb behaves as a dimer even at a concentration of 30 μm or an optical density of 1 at 230 nm (
      • Keyhani N.O.
      • Rodgers M.E.
      • Demeler B.
      • Hansen J.C.
      • Roseman S.
      ).
      Figure thumbnail gr3
      Fig. 3Static and dynamic light scattering of IIAChb wild-type and mutant proteins. A and B, size exclusion column elution profiles of IIAChb proteins characterized by steady state light scattering and refractive index. The magnitude of the refractive index is proportional to the total protein concentration (represented by filled circles and lines); a value of 0.01 for the refractive index is equivalent to ∼0.012 mg/ml protein. The average molecular masses of the eluted fractions determined by static light scattering are represented by filled triangles. Unless otherwise noted, all the samples were prepared in 10 mm sodium phosphate buffer, pH 6.5, containing 300 mm NaCl, 1 mm methionine, and 0.02% NaN3. The sequence-dependent molecular mass composition of wild-type and variant IIAChb proteins is analyzed in A. The data for wild-type IIAChb loaded with Ni2+, the IIAChb-NΔ13 deletion mutant loaded with Ni2+, and the IIAChb-NΔ13/D92L double mutant (IIAChb*) are displayed in blue, green and red, respectively. The IIAChb-NΔ13/D92L double mutant (IIAChb*) behaves as a homogeneous, monodisperse trimer. Note that additional data points are plotted for the wildtype IIAChb light-scattering data before the signal is completely buried in the noise. The influence of buffer composition on the molecular weight composition of the IIAChb-NΔ13 deletion mutant is shown in B. Data for the Ni2+-loaded protein in phosphate buffer (same as in A for comparison) and in Tris buffer (10 mm Tris·HCl, pH 7.5, 300 mm NaCl, 1 mm methionine, and 0.02% NaN3) are shown in green and purple, respectively; data for the apoprotein (sample stripped of Ni2+) in phosphate buffer is shown in orange. C, dynamic light-scattering measurements performed on the IIAChb-NΔ13/D92L double mutant (IIAChb*). The average values of the normalized autocorrelation function obtained from 10 individual experiments, corresponding to a total collection time of 12 min, are shown as open circles, and the best second cumulant fit is shown as a solid line through the experimental data points.
      To obtain a homogeneous and monodisperse sample suitable for structural studies, we constructed a series of IIAChb variants using rational mutagenesis. We first noted that (a) the N-terminal 11 residues of IIAChb are not present in L. lactis IIALac (Fig. 1), (b) the N-terminal two residues of IIALac have high B-factors in the crystal structure (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ), and (c) the secondary structure of the first 13 residues of IIAChb is predicted to be unstructured (PredictProtein, cubic.bioc.columbia.edu/predictprotein/). On this basis we constructed a deletion mutant of IIAChb lacking the N-terminal 13 residues, IIAChb-NΔ13 (Fig. 1). The light-scattering profile for Ni2+-loaded IIAChb-NΔ13 is also a mixture of different molecular mass species ranging from 35 to 70 kDa (Fig. 3A, green symbols). Unlike full-length IIAChb, however, the IIAChb-NΔ13 sample contains a major species with a molecular mass of ∼35 kDa, 3 times the monomer molecular mass. Thus, IIAChb-NΔ13 is a mixture comprising a predominant trimeric species in conjunction with high molecular weight aggregates. Judging from the gel filtration profile quantified by refractive index, the trimeric species makes up over 90% of the total protein. The trimer is in equilibrium with the high molecular weight aggregates as a sample drawn from the trimer peak yields an identical elution profile when loaded again on the gel filtration column. Consistent with the light-scattering analysis, the 1H-15N HSQC spectrum of 15N-labeled IIAChb-NΔ13 in the presence of Ni2+ displays far better resolved peaks than that of the wild-type protein (Fig. 2B).
      In addition to improving the thermostability of IIAChb-NΔ13, the presence of Ni2+ paradoxically promotes protein aggregation. Ni2+-charged IIAChb-NΔ13 prepared in Tris buffer (Fig. 3B, blue) yields a similar light-scattering profile to that prepared in phosphate buffer (Fig. 3B, green) characterized by a major species of ∼35 kDa. However, the molecular mass of the high order aggregates in Tris buffer extends beyond 100 kDa, indicative of more extensive aggregation. The difference is likely due to the weak metal cation affinity of phosphate that scavenges Ni2+ from solution and reduces metal-induced nonspecific aggregation. In contrast, metal-stripped IIAChb-NΔ13 contains no predominant species of defined molecular weight (Fig. 3B, orange symbols). This observation confirms the importance of metal cations in stabilizing the trimer interface, presumably in the same fashion as that observed in the crystal structure of L. lactis IIALac where a metal ion is coordinated by three symmetry-related buried aspartate side chains, one from each subunit (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ). As such, the role of divalent metal cations is double edged: they promote trimer formation but simultaneously induce a small degree of nonspecific protein aggregation.
      To circumvent the dual behavior of metal cations, we introduced an Asp to Leu mutation in IIAChb-NΔ13 at the site (residue 92) corresponding to the buried aspartate involved in metal ion coordination in IIALac. Since Leu and Asp have similarly branched side chains, the Asp → Leu mutation should eliminate the need for a metal cation at the trimer interface by substituting hydrophobic methyl-methyl interactions in place of the metal cation. To test this hypothesis, we conducted light-scattering measurements on the IIAChb-NΔ13/D92L double mutant (IIAChb*). The light-scattering profile shows a single species at 33.5 kDa (Fig. 3A, red symbols) in excellent agreement with the calculated molecular mass of 33.7 kDa for a IIAChb* trimer (11,241.1 Da per subunit). Unlike the IIAChb-NΔ13 deletion mutant, the choice of buffer (phosphate versus Tris) does not have any impact on the molecular weight of the sample (data not shown). To further corroborate these results, dynamic light scattering on IIAChb* was also carried out, and a single exponential autocorrelation profile was obtained that fits to a single species with a diffusion coefficient D20,w of 8.1 ± 0.2 × 10-7 cm2·s-1 and a Stokes radius of 26.5 ± 0.6 Å (Fig. 3C). These measured values are in very close agreement with those calculated (
      • Garcia de la Torre J.
      • Huertas M.L.
      • Carrasco B.
      ) from the crystal structure (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ) of trimeric IIALac (8.28 × 10-7 cm2·s-1 and 25.9 Å, respectively).
      Consistent with the light-scattering results, 15N-labeled IIAChb* displays a relatively well resolved 1H-15N HSQC spectrum: the peaks have reasonable linewidths for a ∼34-kDa protein, and the number of cross-peaks is close to that expected for a symmetric trimer of 103 residues per subunit (Fig. 2C). Moreover some of the outlying cross-peaks in the 1H-15N HSQC spectrum of wild-type IIAChb (Fig. 2A) match closely to those of IIAChb-NΔ13 (Fig. 2B) and IIAChb* (Fig. 2C), demonstrating that the structures remain largely intact despite the mutations introduced. In addition, titration of unlabeled HPr results in a systematic perturbation in the chemical shifts for a subset of cross-peaks in the 1H-15N HSQC spectrum of IIAChb* (data not shown), indicative of a specific interaction between HPr and IIAChb*. Finally 1H-15N correlation spectroscopy and mass spectrometry indicate that IIAChb* is readily phosphorylated upon incubation with phosphoenolpyruvate and catalytic amounts of enzyme I and HPr (data not shown), indicating the functional relevance of IIAChb*. Hence the current experimental data lead one to a single conclusion: wild-type IIAChb is a mixture of aggregates of different molecular weight species, whereas the double mutant IIAChb*, containing both the N-terminal deletion and the Asp → Leu point mutation, yields a homogenous, monodisperse trimeric form suitable for structural studies. The discrepancy with previous work in which it was proposed that wild-type IIChb is a dimer on the basis of analytical ultracentrifugation studies (
      • Keyhani N.O.
      • Rodgers M.E.
      • Demeler B.
      • Hansen J.C.
      • Roseman S.
      ) may possibly be due to the unusually high centrifugation speeds (>35,000 rpm) used (www.beckman.com/resourcecenter/labresour-ces/sia/pdf/rotor_speed_equilibrium.pdf).
      Methyl-based NOEs for Structure Determination—On the basis of the above results, all structural NMR work was carried out on the IIAChb-NΔ13/D92L double mutant IIAChb*. Many of the NMR experiments were carried out on a 2H/15N/13C/(Ile-Leu-Val)-methyl-protonated sample, denoted as ILV-IIAChb*. Perdeuteration lowers the proton density and slows the overall transverse relaxation rates for larger proteins, while fast rotation and 3-fold degeneracy give methyl protons a further edge in relaxation properties (
      • Goto N.K.
      • Gardner K.H.
      • Mueller G.A.
      • Willis R.C.
      • Kay L.E.
      ). Moreover, as hydrophobic side chains are often involved in protein tertiary folding, retaining methyl protons still yields a large number of long range inter-residue NOEs (|i - j| > 5) in a large system. The necessity of such a labeling strategy was dictated by the fact the ∼34-kDa IIAChb* trimer is essentially all helical with the long axes of the helices approximately parallel to the principal axis of the diffusion tensor.
      Three primary sources of NMR structural restraints were used in the structure determination: NOE-derived interproton distance restraints, backbone and side-chain torsion angle restraints, and orientational restraints in the form of backbone one-bond residual dipolar couplings. All assigned long range (|i - j| > 5) NOE-derived interproton distance restraints, both intra- and intersubunit, involve methyl groups and were derived from multidimensional NOE spectra recorded on samples in which at least one subunit was 15N/13C/2H/(Val-Leu-Ile)-methyl-protonated. A summary of the structural statistics is provided in Table I, and a best fit superposition of the backbone atoms of the final ensemble of 100 simulated annealing structures is shown in Fig. 4A. Despite the fact that no attempt was made to assign the majority of non-methyl-based NOEs, the structures calculated from the experimental restraints are of high quality in terms of both coordinate precision and structure quality indicators (Table I). As such, the current approach therefore demonstrates the feasibility of obtaining high resolution NMR structures using predominantly methyl-based NOE-derived interproton distance restraints in conjunction with torsion angle and orientational dipolar coupling restraints. This approach should also be applicable to even larger systems where it becomes increasingly difficult to obtain high quality spectra on a uniformly 13C, 15N-labeled sample due to even faster proton relaxation (
      • Tugarinov V.
      • Kay L.E.
      ).
      Table IStructural statistics
      Experimental restraints〈SA〉SA¯r
      r.m.s. deviations from experimental restraints
      〈SA〉 are the final 100 simulated annealing structures. 〈SA¯〉r is the restrained regularized mean structure derived from the mean coordinates obtained by averaging the coordinates of the 100 simulated annealing structures best fitted to each other (with respect to residues 13–73 and 83–114 of all three subunits simultaneously). The number of terms (on a trimer basis) for the various restraints is given in parentheses. None of the structures exhibits interproton distance violations >0.3 Å or torsion angle violations >5°. There are 828 interproton distance restraints comprising 246 sequential (|i–j| = 1), 258 medium range (1 < |i–j| ≤ 5), and 90 long range (|i–j >| > 5) intramolecular restraints and 234 intersubunit restraints. The intermolecular restraints are treated as ambiguous (i.e. no distinction can be made a priori for a NOE from subunit A to subunit B or C) by means of a (Σr-6)-1/6 sum representation. In addition, 462 distance restraints for 231 (77 per subunit) backbone hydrogen bonds within the three helices were also used. The torsion angle restraints comprise 255 φ, 255 ψ, and 219 side-chain χ restraints
      Distances (Å) (1290)0.008 ± 0.0020.010
      Torsion angles (°) (729)0.21 ± 0.050.41
      13Cα chemical shifts (ppm) (300)1.12 ± 0.031.19
      13Cβ chemical shifts (ppm) (285)0.64 ± 0.020.69
      Residual dipolar coupling R-factors (%)
      The dipolar coupling R-factor, Rdip, which scales between 0 and 100%, is defined as the ratio of the r.m.s. deviation between observed and calculated values to the expected r.m.s. deviation if the vectors were randomly distributed, given by [2Da2(4 + 3η2)/5]1/2 where Da is the magnitude of the principal component of the alignment tensor and η is the rhombicity (71). For a trimer the alignment tensor is always axially symmetric (η = 0). The value of DaNH, derived from the distribution of normalized dipolar couplings (74), is -12.1 Hz
      1DNH (252)7.0 ± 0.17.1
      1DNC′ (255)17.4 ± 0.915.2
      2DCaC′ (249)16.1 ± 1.017.7
      Ramachandran map analysis
      Calculated with the program PROCHECK (75). The dihedral angle G-factors for the ensemble of simulated annealing structures are 0.64 ± 0.05, 0.86 ± 0.05, 0.54 ± 0.09, and 0.21 ± 0.14 for φ/ψ, χ1/χ2, χ1, and χ3/χ4, respectively. The WHATIF first and second generation packing scores (76) for the restrained regularized mean structure are 2.10 and 3.52, respectively; values larger than -0.5 and 0, respectively, are considered to represent high quality structures
      Percentage of residues in most favored region93.5 ± 1.791.2
      Percentage of residues in additionally allowed regions5.7 ± 1.66.7
      Number of bad contacts per 100 residues
      Calculated with the program PROCHECK (75). The dihedral angle G-factors for the ensemble of simulated annealing structures are 0.64 ± 0.05, 0.86 ± 0.05, 0.54 ± 0.09, and 0.21 ± 0.14 for φ/ψ, χ1/χ2, χ1, and χ3/χ4, respectively. The WHATIF first and second generation packing scores (76) for the restrained regularized mean structure are 2.10 and 3.52, respectively; values larger than -0.5 and 0, respectively, are considered to represent high quality structures
      4.5 ± 1.67.4
      Deviations from idealized covalent geometry
      Bonds (Å)0.001 ± 00.004
      Angles (°)0.39 ± 0.020.59
      Improper torsions (°)0.58 ± 0.040.75
      Coordinate precision (Å)
      The precision of the coordinates is defined as the average atomic r.m.s. difference between the individual 100 simulated annealing structures and the corresponding mean coordinates obtained by best fitting residues 17–73 and 83–114 of all three subunits simultaneously. Residues 14–16, 74–82, and 115–116 are disordered in solution
      Backbone (N, Cα, C′)0.37 ± 0.09
      All heavy atoms0.84 ± 0.14
      a 〈SA〉 are the final 100 simulated annealing structures. SA¯r is the restrained regularized mean structure derived from the mean coordinates obtained by averaging the coordinates of the 100 simulated annealing structures best fitted to each other (with respect to residues 13–73 and 83–114 of all three subunits simultaneously). The number of terms (on a trimer basis) for the various restraints is given in parentheses. None of the structures exhibits interproton distance violations >0.3 Å or torsion angle violations >5°. There are 828 interproton distance restraints comprising 246 sequential (|i–j| = 1), 258 medium range (1 < |i–j| ≤ 5), and 90 long range (|i–j >| > 5) intramolecular restraints and 234 intersubunit restraints. The intermolecular restraints are treated as ambiguous (i.e. no distinction can be made a priori for a NOE from subunit A to subunit B or C) by means of a (Σr-6)-1/6 sum representation. In addition, 462 distance restraints for 231 (77 per subunit) backbone hydrogen bonds within the three helices were also used. The torsion angle restraints comprise 255 φ, 255 ψ, and 219 side-chain χ restraints
      b The dipolar coupling R-factor, Rdip, which scales between 0 and 100%, is defined as the ratio of the r.m.s. deviation between observed and calculated values to the expected r.m.s. deviation if the vectors were randomly distributed, given by [2Da2(4 + 3η2)/5]1/2 where Da is the magnitude of the principal component of the alignment tensor and η is the rhombicity (
      • Bax A.
      • Kontaxis G.
      • Tjandra N.
      ). For a trimer the alignment tensor is always axially symmetric (η = 0). The value of DaNH, derived from the distribution of normalized dipolar couplings (
      • Clore G.M.
      • Gronenborn A.M.
      • Bax A.
      ), is -12.1 Hz
      c Calculated with the program PROCHECK (
      • Laskowski R.A.
      • MacArthur M.W.
      • Moss D.S.
      • Thornton J.M.
      ). The dihedral angle G-factors for the ensemble of simulated annealing structures are 0.64 ± 0.05, 0.86 ± 0.05, 0.54 ± 0.09, and 0.21 ± 0.14 for φ/ψ, χ12, χ1, and χ34, respectively. The WHATIF first and second generation packing scores (
      • Vriend G.
      • Sander C.
      ) for the restrained regularized mean structure are 2.10 and 3.52, respectively; values larger than -0.5 and 0, respectively, are considered to represent high quality structures
      d The precision of the coordinates is defined as the average atomic r.m.s. difference between the individual 100 simulated annealing structures and the corresponding mean coordinates obtained by best fitting residues 17–73 and 83–114 of all three subunits simultaneously. Residues 14–16, 74–82, and 115–116 are disordered in solution
      Figure thumbnail gr4
      Fig. 4Structure of IIAChb*. A, stereoview of a best fit superposition of the backbone (N, Cα, and C′) atoms of the final 100 simulated annealing structure of the IIAChb* double mutant. The three subunits A, B, and C are colored in red, blue, and green, respectively. Best fitting was carried out with respect to the ordered residues, 17–63 and 83–114, of all three subunits. B,Cα coordinate precision and steady-state 15N-{1H} NOE represented by a bar graph and red solid circles, respectively, as a function of residue number. Precision is defined as the average atomic r.m.s. difference between the 100 individual simulated annealing structures and the mean coordinate positions. The secondary structure is delineated at the bottom of the figure.
      Overall Structure of IIAChb*—The solution structure of IIAChb* reveals that the three subunits of the homotrimer are arranged in a parallel fashion with the cross-section making an equilateral triangle (Fig. 5A, right-hand panel). Consistent with the 3-fold symmetry, the residual dipolar couplings measured for the backbone N-H, N-C′, and Cα-C′ vectors of the ordered residues fit to a fully axial symmetric alignment tensor, and the principal axis of alignment tensor coincides with the 3-fold symmetry axis of the trimer (
      • Al-Hashimi H.M.
      • Bolton P.J.
      • Prestegard J.H.
      ). Each subunit of the IIAChb* trimer comprises a three-helix bundle with an up-down topology in which helix 2 is antiparallel to helices 1 and 3. The three-helix bundle displays a small degree of left-handed supercoiling. Each helix has a span of nearly 30 residues (α1, residues 17–43; α2, residues 47–73; and α3, residues 85–113). The N-terminal 3 residues and C-terminal 2 residues are unstructured in solution (Fig. 4B) and are located on opposite sites of the protein. A tight turn (residues 44–46) connects helices 1 and 2, while an 11-residue flexible loop (residues 74–84), located in close proximity to the N terminus, bridges helices 2 and 3. These flexible/disordered regions are characterized by small heteronuclear 15N-{1H} NOE values (≤0.6) and low backbone coordinate precision (Fig. 4B). The functional and/or structural role of the flexible loop is unclear, but it is possible that its interaction with additional unstructured residues at the N terminus of wild-type IIAChb may be responsible for aggregation of the wild-type protein, rendering it unsuitable for structural studies.
      Figure thumbnail gr5
      Fig. 5Topology of IIAChb* and details of the trimer interface. A, ribbon diagram representation showing two views, parallel (left panel) and orthogonal (right panel) to the long axes of the helices with the three subunits, A, B, and C, colored in red, blue, and green, respectively. Hydrophobic side chains at the trimer interface are also shown as bonds in the respective color of the subunit. The active site histidine (His-89) is shown in purple. Note that the side-chain conformation for residue Met-95 is not well defined due to insufficient experimental restraints. The three α-helices are labeled in the right-hand panel. B, stereoview showing a close-up of the central helices (helix α3 from each subunit) at the trimer interface illustrating the leucine coiled-coil structure. The conformational space sampled by the leucine side chains (residues 99, 103, 107, 110, and 114) is represented by an isosurface of the reweighted atomic probability density map calculated from the final ensemble of 100 simulated annealing structures and drawn at a threshold of 20% (green grid). The backbone of residues 98–115 and the side-chain coordinates of the restrained regularized mean structure are shown as blue tubes and red bonds, respectively.
      Helix 3, which is arranged directly around the 3-fold symmetry axis, is primarily responsible for the assembly of the IIAChb* trimer via formation of a parallel coiled-coil structure (Fig. 5). Lining the trimer interface formed by helix 3 are a series of hydrophobic residues that include Leu-85, Val-88, Leu-92 (which is an Asp in wild-type IIAChb), Met-95, Leu-99, Leu-103, Leu-107, Leu-110, and Leu-114 (Fig. 5A). Inspection of the coiled-coil helices reveals the characteristic “knobs-into-holes” packing (
      • Chothia C.
      ), in particular for the well defined Leu side chains, reminiscent of a leucine zipper structure (Fig. 5C). Unusual for a coiled-coil structure, the top view of the central helices reveals a slight right-handed twist (Fig. 5A, right-hand panel) (
      • Lupas A.
      ,
      • Harbury P.B.
      • Plecs J.J.
      • Tidor B.
      • Alber T.
      • Kim P.S.
      ). Thus, to compensate for the lack of extensive left-handed supercoiling, the interacting hydrophobic residues do not strictly follow the alternating 4-3 heptad repeat of a canonical coiled-coil structure. About 25% of the total accessible surface area, 1917 Å2 per subunit (total of 5751 Å2) is buried at the trimer interface. Thus the IIAChb* trimer is expected to be very stable, and the structure is consistent with the observation that wild-type IIAChb behaves as a trimer at low (∼140 nm) concentrations (Fig. 2A).
      The interface between the adjacent subunits, A and C, C and B, and B and A, forms a deep depression on the surface of the molecule that is partially shielded from bulk solvent. The depressed surface comprises residues from helices 1 and 3 in one subunit (subunits A, B, or C) and from helices 2 and 3 in the other (subunits C, B, or A, respectively). Referenced with respect to helix 3, the N-terminal half of the crevice is formed by hydrophobic and non-polar side chains, whereas several charged residues spot the C-terminal half of the depression (Fig. 6A). Since the buried surface between the adjacent subunits in a coiled-coil trimer is larger than that in a dimer, it has been noted that the corresponding sites are often enriched for hydrophobic residues (
      • Woolfson D.N.
      • Alber T.
      ). In fact, charged residues of opposite signs located at positions e and g of the heptad helical repeat are one of the determinants for coiled-coil dimer formation (
      • Zeng X.
      • Zhu H.
      • Lashuel H.
      • Hu J.C.
      ,
      • McClain D.L.
      • Binfet J.P.
      • Oakley M.G.
      ). Thus, with charged residues located deep in the surface crevice, a possible stabilization mechanism for the IIAChb trimer would involve balancing of charges. Indeed pairs of counterions are formed between the following pairs of side-chains residues: Arg-32A and Glu-102C, Lys-43A and Glu-106C, Lys-43A and Glu-109C, Lys-51C and Glu-112C, Arg-101C and Asp-55C, and Lys-113C and Glu-109C. In contrast to the dimeric coiled-coil, the interacting charged residues are not restricted to the coiled-coil-forming helices (i.e. helix 3). Thus the attractive coulombic interactions between the partially buried charged residues actually further stabilize the IIAChb* trimer.
      Figure thumbnail gr6
      Fig. 6Surface representation of the depression between the two adjacent subunits of the IIAChb* trimer. A, the bottom half of the depression (i.e. closest to the C terminus) features a network of electrostatic interactions. Positively and negatively charged side chains are colored in blue and red, respectively; tyrosine and histidine are colored in purple. B, structural model for phosphorylated IIAChb*. The phosphoryl group (shown in red) bonded to the Nϵ-2 of the active site histidine (His-89 of subunit A, shown in purple) was modeled in standard geometry. Neighboring residues in the crevice are also depicted: these include His-65 (colored in purple) and Gln-91 (shown in cyan) of subunit C and His-93 of subunit A (colored in purple). In B, the side chains of interest are represented by bonds, and the molecular surface etched behind the displayed side chains is colored in orange. In both panels the subunit of origin (subunit A or subunit C) is indicated in the residue labeling by a superscript.
      The Active Site—The active site histidine, His-89, is phosphorylated at the Nϵ-2 position by HPr (
      • Keyhani N.O.
      • Boudker O.
      • Roseman S.
      ). His-89 is located near the N terminus of helix 3, surrounded by predominantly hydrophobic residues. We modeled a phosphoryl group bonded to the Nϵ-2 atom of His-89 using idealized covalent geometry (Fig. 6B). It is worth noting that, due to steric hindrance, His-89 (irrespective of rotameric state about the χ1 and χ2 side-chain torsion angles) cannot be phosphorylated at the Nδ-1 position. Deep in the pocket, the negative charge of the phosphoryl group bonded to His-89A is neutralized by (a) the positive dipole associated with the N terminus of helix 3; (b) a potential electrostatic interaction with the side chain of protonated His-93A whose Nϵ-2 atom is on average ∼4 Å away from the oxygen atoms of the phosphoryl group (Fig. 6A), reminiscent of the situation in IIAGlc (
      • Pelton J.G.
      • Torchia D.A.
      • Meadow N.D.
      • Roseman S.
      ); and (c) a potential hydrogen bond between the phosphoryl group and the carboxyamide group of Gln-91C of the adjacent subunit, which is less than 3 Å away (Fig. 6B). It has been speculated in the case of L. lactis IIALac that His-65C (in the current numbering scheme) is also involved in the stabilization of the phosphorylated state by forming a histidine triad (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ). In the structure of IIAChb*, however, His-65C is more than 8 Å away from the phosphoryl group of His-89A. It is possible that an interaction between the side chains of His-65C and Gln-91C serves to optimally orient the side chain of Gln-91C for hydrogen bonding with the phosphoryl group on His-89A. In addition, sequence conservation of His-65 in L. lactis IIALac and E. coli IIAChb (Fig. 1) may suggest that His-65 is possibly involved in protein-protein interactions with HPr and/or IIBChb. Thus, we suggest that partial sequestration of the phosphoryl group from bulk solvent coupled with electrostatic and/or hydrogen bonding interactions with the positive helix dipole and neighboring residues in the deep active site pocket are involved in stabilizing the phosphohistidine of IIAChb.
      Comparison of the Structures of E. coli IIAChb* and L. lactis IIALac—Both the crystal structure of L. lactis IIALac (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ) and the NMR structure of E. coli IIAChb* are trimeric with identical overall topologies. Thus, the trimeric nature of E. coli IIAChb* fulfills its sequence and functional homology to L. lactis IIALac. Indeed the bottom half of the depressed surface between the adjacent A and C subunits of IIALac is also spotted with charged residues. Comparison of the sequences of IIALac and IIAChb (Fig. 1) reveals that the charged residues (Fig. 6A) are either conserved (Arg-32, Lys-40, and Glu-112), conservatively substituted (Arg-101 Lys, Glu-102 → Asp), or subject to simultaneous compensatory substitutions involving ion pairs (Lys-43 → Glu and Glu-106 → His pair, and Lys-51 → Asp and Glu-112 → Lys pair). Thus, the total net charge in the surface crevice is always kept close to zero, thereby eliminating repulsive and desolvation effects that could potentially destabilize the trimer.
      There are some small differences in the extent of the helices between the two structures: helix 1 has 3 additional residues at its N terminus, helix 2 has 3 additional residues at its C terminus, and helix 3 is 2 residues shorter at its C terminus in IIALac relative to IIAChb*. The shorter loop connecting helices 2 and 3 in IIALac relative to IIAChb* (8 versus 11 residues) coupled with the longer distance between the C terminus of helix 2 and the N terminus of helix 3 in IIALac relative to IIAChb (∼21 versus ∼15 Å) and the additional presence of a proline residue in the IIALac loop may account for the observation that this loop is ordered in the crystal structure of IIALac with B-factors less than 40 Å2, whereas it is highly disordered for IIAChb* in solution (see Fig. 4).
      Excluding the disordered regions (residues 14–16, 74–84, and 115–116), the backbone atomic r.m.s. difference between the crystal structure of IIALac and the NMR structure of IIAChb* is 1.94 Å for the trimer (Fig. 7A) and 1.71 ± 0.04 Å for the individual subunits (Fig. 7B). These differences are substantial and outside the coordinate errors of the two structures. A large contribution to the atomic r.m.s. difference originates from helix 3 (Fig. 7B). Thus, the backbone atomic r.m.s. difference between the two structures for residues 17–73 (helices 1 and 2 and the turn connecting them) is reduced to 1.06 ± 0.16 Å per subunit. From the perspective of the trimer interface, it is worth noting that when helices 1 and 2 of all three subunits are superimposed, the Cγ atom of Leu-92 in IIAChb* (mutated from an Asp in the wild-type protein) is less than 1 Å from the Cγ atom of the corresponding Asp of IIALac (Fig. 7A). This clearly supports the hypothesis that the buried Asp-79 in wild-type IIAChb stabilizes the trimer by coordinating a divalent cation in the same fashion as in IIAChb and further demonstrates that the Asp to Leu point mutation has minimal impact on the structure. Note that the equivalent mutation in L. lactis IIALac does not result in any significant structural perturbation in the crystal structure (Protein Data Bank code 2E2A) relative to that of the wild-type (Protein Data Bank code 1E2A; Ref.
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ) with a Cα atomic r.m.s. difference of only 0.87 Å (0.66 Å for the helices) between the two sets of coordinates.
      Figure thumbnail gr7
      Fig. 7Comparison of the NMR structure of E. coli IIAChb* and the crystal structure of L. lactis IIALac. A and B, ribbon diagrams showing best fit superpositions of the restrained regularized mean structure of IIAChb* (blue) and L. lactis IIALac (red). Two views, orthogonal and parallel to the long axes of the helices, are shown in A and B, respectively, with the trimer displayed in A and an individual subunit in B. Best fitting was carried out with respect to the backbone atoms of helices 1 (residues 17–43) and 2 (residues 47–73) of all three subunits in A and with respect to helix 1 (residues 17–43) and the N-terminal half of helix 2 (residues 47–63) in B. For clarity, only the helices are shown in A, and the disordered N- and C-terminal residues are omitted in B. The side chains of Leu-92 of IIAChb* and the equivalent Asp of IIALac are represented as bonds in A, and their Cγ atom positions are 1.02, 0.92, and 0.92 Å apart for subunits A, B, and C, respectively. The three subunits are indicated in A, and the three helices of a single subunit are labeled in B. Note that the distance between the C-terminal end of helix 2 and the N-terminal end of helix 3 is significantly longer in IIALac (∼21 Å) than in IIAChb (∼15 Å), yet the number of residues in the loop connecting helices 2 and 3 is 3 residues fewer in IIALac. This may account for the observation that this loop is ordered in IIALac but disordered in IIAChb. C, buried hydrophobic cavity (green) in IIAChb* (left-hand panel) and the crystal structure of L. lactis IIALac (right-hand panel). The backbone atoms of residues 95–111 of the three subunits are depicted as tubes (blue), and the side chains of residues 99, 100, 103, and 107 are displayed in red as bonds (numbering scheme of IIAChb*). The cavity was calculated and displayed as a surface (green) using the program GRASP (
      • Nicholls A.
      • Sharp K.A.
      • Honig B.
      ). The volume of the cavity in IIAChb* is 45 Å3 and lined by the side chains of Leu-103 and Leu-107; the volume of the cavity of IIALac is 230 Å3 and lined by the side chains of Leu-99, Leu-100, Val-103, and Leu-107. D, comparison of the intersubunit interactions between helix 1 (subunit A) and helix 3 (subunit C) at the site of the kink in helix 3 (Glu-102) observed in the NMR structure of IIAChb* and the x-ray structure of IIALac. The backbone (blue for IIAChb* and red for IIALac) is displayed as a tube, and the side chains (cyan for IIAChb* and purple for IIALac) are displayed as bonds. The coordinates for the crystal structure of L. lactis IIALac are taken from Ref.
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      (Protein Data Bank code 1E2A).
      In the crystal structure of L. lactis IIALac, helix 3 exhibits a large 40° kink centered around residues 102 and 103 (Figs. 7B and C). This results in the presence of a very large (230-Å3), completely buried cavity in the crystal structure (Fig. 7C, right-hand panel) in which the heavy atom derivative, trimethyl lead acetate, is located (
      • Sliz P.
      • Engelmann R.
      • Hengstenberg W.
      • Pai E.F.
      ). The cavity in IIALac is entirely hydrophobic and lined by the side chains of Leu-99, Leu-100, Val-103, and Leu-107 (using the current numbering scheme, Fig. 1). In contrast, the NMR structure of IIAChb* exhibits only a 7° bend at the same location, and the volume of the corresponding cavity is reduced to 45 Å3 (Fig. 7C, left-hand panel). This smaller cavity in IIAChb* is also entirely hydrophobic and lined by the side chains of Leu-103 and Leu-107. The differences related to cavity size and the kinking of helix 3 are associated with concomitant structural differences in helix 2, which exhibits a bend of 27° centered around residues 64 and 65 in IIALac compared with only 18° in IIAChb* (Fig. 7B).
      The question immediately arises as to whether these differences are real and can be readily discerned from the NMR data. NOE-derived interproton distance restraints are limited to distances less than 5–6 Å and hence cannot distinguish a bent from a straight helix. Dipolar couplings, however, provide very sensitive long range information in the form of orientation of atomic vectors relative to the alignment tensor that can be used to quantitatively ascertain structural differences between the solution and crystal states. In this case, since the alignment tensor is axially symmetric, the orientational information is restricted to the angle between atomic vectors and the principal axis of the alignment tensor, which coincides with the 3-fold symmetry axis of the trimer. The crystal structure of IIALac was solved at 2.3-Å resolution, and one would therefore expect a dipolar coupling R-factor, RdipNH (
      • Clore G.M.
      • Garrett D.S.
      ), of around 15–20% for the N-H backbone dipolar couplings (
      • Wang G.
      • Louis J.M.
      • Sondej M.
      • Seok Y.-J.
      • Peterkofsky A.
      • Clore G.M.
      ,
      • Cornilescu G.
      • Lee B.R.
      • Cornilescu C.
      • Wang G.
      • Peterkofsky A.
      • Clore G.M.
      ,
      • Bax A.
      • Kontaxis G.
      • Tjandra N.
      ,
      • Williams D.C.
      • Cai M.
      • Clore G.M.
      ). The overall RdipNH factor for all the three helices of the IIALac trimer is 38% compared with 22% for helix 1, 25% for helix 2, and 55% for helix 3 (Table II). One can therefore conclude that whereas the path and orientation of helix 1 in the IIALac and IIAChb* trimers are essentially the same, the difference in kink angle for helix 3 is real. With respect to helix 2 of IIALac, the RdipNH factor for the N-terminal half (α2N, residues 47–63) is 16% compared with 36% for the C-terminal half (α2C, residues 66–73), which clearly indicates that the difference in bend angle for helix 2 in IIAChb* and IIALac results in a different orientation of the C-terminal half of helix 2 between the two proteins. This is reflected in the much smaller backbone atomic r.m.s. difference between IIAChb* and IIALac for helices α1 plus α2N (1.17 Å for the trimer) than for helices α1 plus α2C (1.72 Å for the trimer).
      Table IIStructural comparison of the helices of the NMR structure of IIAChb* and the crystal structure of IIALac
      Helix
      Helices α1, α2, and α3 comprise residues 17–43, 47–73, and 85–113, respectively. The crystal structure of IIALac displays a bend (∼27°) and kink (∼40°) in helices 2 (centered around residues 64 and 65) and 3 (centered around residues 102 and 103), respectively: the N-terminal halves of helices 2 (α2N) and 3 (α3N) comprise residues 47–63 and 85–101, respectively; the C-terminal halves of helices 2 (α2C) and 3 (α3C) comprise residues 66–73 and 104–113, respectively. The RdipNH values for helix 2 exclude residues 66 and 67 of subunit B and residue 65 of subunit C, which yield anomalously high fitting errors to the IIALac crystal structure; if all the N-H dipolar couplings of helix 2 are fitted to the IIALac crystal structure, RdipNH is increased to 37.1%. The backbone atomic (N, Cα, and C′) r.m.s. differences relate to the restrained regularized mean NMR structure of IIAChb* and the crystal structure (Protein Data Bank code 1E2A) of IIALac
      N-H dipolar coupling R-factor, Rdip
      The definition of the dipolar coupling R-factor is given in Footnote b in Table I
      Backbone r.m.s. difference
      IIALac (x-ray)IIAChb* (NMR)
      Restrained regularized mean structure
      %
      α1 + α2 + α337.57.31.95
      α1 + α2N20.17.51.17
      α1 + α224.06.91.54
      α2 + α343.27.22.15
      α1 + α342.07.82.03
      α122.47.51.41
      α225.46.21.63
      α2N15.57.40.52
      α2C35.92.71.56
      α354.58.12.30
      α3N47.59.32.46
      α3C50.36.21.29
      a Helices α1, α2, and α3 comprise residues 17–43, 47–73, and 85–113, respectively. The crystal structure of IIALac displays a bend (∼27°) and kink (∼40°) in helices 2 (centered around residues 64 and 65) and 3 (centered around residues 102 and 103), respectively: the N-terminal halves of helices 2 (α2N) and 3 (α3N) comprise residues 47–63 and 85–101, respectively; the C-terminal halves of helices 2 (α2C) and 3 (α3C) comprise residues 66–73 and 104–113, respectively. The RdipNH values for helix 2 exclude residues 66 and 67 of subunit B and residue 65 of subunit C, which yield anomalously high fitting errors to the IIALac crystal structure; if all the N-H dipolar couplings of helix 2 are fitted to the IIALac crystal structure, RdipNH is increased to 37.1%. The backbone atomic (N, Cα, and C′) r.m.s. differences relate to the restrained regularized mean NMR structure of IIAChb* and the crystal structure (Protein Data Bank code 1E2A) of IIALac
      b The definition of the dipolar coupling R-factor is given in Footnote b in Table I
      c Restrained regularized mean structure
      A comparison of the intersubunit interactions involving residue 102 of helix 3, the site of the 40° kink in IIALac, reveals a probable source of the principal structural differences between IIAChb* and IIALac (Fig. 7D). In IIAChb*, Glu-102 of subunit C forms a potential salt bridge with Arg-32 and a hydrogen bond with the hydroxyl group of Tyr-36 located in helix 1 of subunit A. All three side chains are partially buried and located in the deep surface crevice at the intersubunit interface formed by helices 1 and 3 (Fig. 6A). Moreover the guanidino group of Arg-32 is directed toward the protein interior, and neutralization of its positive charge is probably important for structural integrity. Glu-102 and Tyr-36 are substituted in IIALac by two shorter side chains, aspartate and leucine. The location of the carboxylate group of Asp-102 of IIALac (current numbering scheme) is essentially identical to that of Glu-102 of IIAChb* so that the salt bridge with Arg-32 is preserved, and the methyl groups of Leu-36 pack against the β-methylene group of Asp-102. Because the side chains of aspartate and leucine are ∼1.5 and 3.5 Å shorter, respectively, than those of glutamate and tyrosine, a compensatory displacement of the backbone of helix 3, in the form of a kink centered at residue 102, occurs (Fig. 7D), thereby creating a large internal hydrophobic cavity at the trimer interface of IIALac formed by helix 3 from the three subunits (Fig. 7C). Presumably the interactions involving Asp-102, Arg-32, and Leu-36 in IIALac coupled with the need to neutralize the partially buried positive charge associated with the guanidino group of Arg-32 compensate for any energetic loss that may accompany the creation of the cavity.
      Concluding Remarks—We have determined the solution structure of a double mutant of IIAChb from the N,N′-diacetylchitobiose branch of the E. coli PTS signal transduction pathway. Wild-type IIAChb is composed of a mixture of aggregates ranging in size from a trimer to high order oligomers, thereby precluding its structural characterization. Using rational mutagenesis on the basis of the crystal structure of the homologous enzyme IIALac from L. lactis, we constructed a double mutant IIAChb-NΔ13/D92L (IIAChb*) that forms a well behaved homogeneous, monodisperse trimer. As the three symmetrically related active sites of IIAChb are located at the interface of two adjacent subunits (A and C, C and B, and B and A), the recognition surface for the upstream (HPr) and downstream (IIBChb) interaction partners would be significantly different from that in the previously proposed dimer of IIAChb (
      • Keyhani N.O.
      • Rodgers M.E.
      • Demeler B.
      • Hansen J.C.
      • Roseman S.
      ). Thus, although it has been noted that mutation of just a few residues can switch the oligomerization state of a coiled-coil protein between dimeric and trimeric forms (
      • Harbury P.B.
      • Zhang T.
      • Kim P.S.
      • Alber T.
      ), the quaternary structure of E. coli IIAChb is preserved despite sequence differences with respect to L. lactis IIALac (Fig. 1). In this sense, the functional role of enzyme IIAChb as a member of the lactose/chitobiose branch of the PTS, that is its interactions with other components of this signaling network, goes hand in hand with structural conservation of a trimeric oligomerization state. Finally the present NMR structure of IIAChb* sets the stage for future NMR structural studies of the 65–70-kDa IIAChb*-HPr and IIAChb*-IIBChb complexes for which the methyl-based NOE approach combined with backbone dipolar couplings used in this study is likely to find considerable utility.

      Acknowledgments

      We thank Drs. Junji Iwahara and Mengli Cai for helpful discussions, Dr. In-Ja Byeon for assistance with light-scattering measurements, Drs. Dan Garrett and Frank Delaglio for software and technical support, and Dr. Saul Roseman for the gift of the wild-type IIAChb plasmid. This study utilized the high performance computational capabilities of the Biowulf PC/Linux cluster at the National Institutes of Health, Bethesda, MD (biowulf.nih.gov).

      References

        • Kundig W.
        • Ghosh S.
        • Roseman S.
        Proc. Natl. Acad. Sci. U. S. A. 1964; 52: 1067-1074
        • Postma P.W.
        • Lengeler J.W.
        • Jacobson G.R.
        Microbiol. Rev. 1993; 57: 543-594
        • Robillard G.T.
        • Broos J.
        Biochim. Biophys. Acta. 1999; 1422: 73-104
        • Siebold C.
        • Flükiger K.
        • Beutler R.
        • Erni B.
        FEBS Lett. 2001; 504: 104-111
        • Wittekind M.
        • Rajgopal P.
        • Cranchini B.R.
        • Reizer J.
        • Saier M.H.
        • Klevit R.E.
        Protein Sci. 1992; 1: 1363-1376
        • Kalbitzer H.R.
        • Hengstenberg W.
        Eur. J. Biochem. 1993; 216: 205-214
        • van Nuland N.A.J.
        • Hangyi I.W.
        • van Shaik R.C.
        • Berendsen H.J.C.
        • van Gunsteren W.F.
        • Scheek R.M.
        • Robillard G.T.
        J. Mol. Biol. 1994; 237: 544-559
        • Ab E.
        • Schuurman-Wolters G.K.
        • Reizer J.
        • Saier M.H.
        • Dijkstra M.
        • Scheek R.M.
        • Robillard G.T.
        Protein Sci. 1997; 6: 304-314
        • Ab E.
        • Shuurman-Walters G.K.
        • Nijlant D.
        • Dijkstra K.
        • Saier M.H.
        • Robillard G.T.
        • Scheek R.M.
        J. Mol. Biol. 2001; 308: 993-1009
        • Garrett D.S.
        • Seok Y.-J.
        • Liao D.-I.
        • Peterkofsky A.
        • Gronenborn A.M.
        • Clore G.M.
        Biochemistry. 1997; 36: 2517-2530
        • Legler P.M.
        • Cai M.
        • Peterkofsky A.
        • Clore G.M.
        J. Biol. Chem. 2004; 279: 39115-39121
        • Liao D.-I.
        • Kapadia G.
        • Reddy P.
        • Saier M.H.
        • Reizer J.
        • Herzberg O.
        Biochemistry. 1991; 30: 9538-9594
        • Worthylake D.
        • Meadow N.D.
        • Roseman S.
        • Liao D.-I.
        • Herzberg O.
        • Remington S.J.
        Proc. Natl. Acad. Sci. U. S. A. 1991; 88: 10383-10386
        • Herzberg O.
        • Reddy P.
        • Sutrina S.
        • Saier M.H.
        • Reizer J.
        • Kapadia G.
        Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 2499-2503
        • Jia Z.
        • Quail J.W.
        • Waygood E.B.
        • Delbaere L.T.
        J. Biol. Chem. 1993; 268: 22490-22501
        • Liao D.-I.
        • Silverton E.
        • Seok Y.-J.
        • Lee B.R.
        • Peterkofsky A.
        • Davies D.R.
        Structure. 1996; 4: 861-872
        • Nunn R.S.
        • Markovic-Housley Z.
        • Genovesio-Taverne G.
        • Flükiger K.
        • Rizkallah P.J.
        • Jansonius J.N.
        • Schirmer T.
        • Erni B.
        J. Mol. Biol. 1996; 259: 502-511
        • Sliz P.
        • Engelmann R.
        • Hengstenberg W.
        • Pai E.F.
        Structure (Lond.). 1997; 5: 775-788
        • Van Montfort R.L.
        • Pijning T.
        • Kalk K.H.
        • Reizer J.
        • Saier M.J.
        • Thunnissen M.M.
        • Robillard G.T.
        • Dijkstra B.W.
        Structure (Lond.). 1997; 5: 217-225
        • Van Montfort R.L.
        • Pijning T.
        • Kalk K.H.
        • Hangyi I.W.
        • Kouwijzer M.L.C.E.
        • Robillard G.T.
        • Dijkstra B.W.
        Structure (Lond.). 1998; 6: 377-388
        • Schauder S.
        • Nunn R.S.
        • Lanz R.
        • Erni B.
        • Schirmer T.
        J. Mol. Biol. 1998; 276: 591-602
        • Orniss G.L.
        • Erni B.
        • Schirmer T.
        J. Mol. Biol. 2003; 327: 1111-1119
        • Garrett D.S.
        • Seok Y.-J.
        • Peterkofsky A.
        • Gronenborn A.M.
        • Clore G.M.
        Nat. Struct. Biol. 1999; 6: 166-173
        • Wang G.
        • Louis J.M.
        • Sondej M.
        • Seok Y.-J.
        • Peterkofsky A.
        • Clore G.M.
        EMBO J. 2000; 19: 5635-5649
        • Cornilescu G.
        • Lee B.R.
        • Cornilescu C.
        • Wang G.
        • Peterkofsky A.
        • Clore G.M.
        J. Biol. Chem. 2002; 277: 42289-42298
        • Cai M.
        • Williams D.C.
        • Wang G.
        • Lee B.R.
        • Peterkofsky A.
        • Clore G.M.
        J. Biol. Chem. 2003; 278: 25191-25206
        • Keyhani N.O.
        • Roseman S.
        Proc. Natl. Acad. Sci. U. S. A. 1997; 94: 14367-14371
        • Zhang M.
        • Van Etten R.L.
        • Stauffacher C.V.
        Biochemistry. 1994; 33: 11097-11105
        • Su X.-D.
        • Taddei N.
        • Stefani M.
        • Ramponi G.
        • Nordlund P.
        Nature. 1994; 370: 575-578
        • Hays J.B.
        • Simoni R.D.
        • Roseman S.
        J. Biol. Chem. 1973; 248: 941-956
        • de Vos W.M.
        • Boerrigter I.
        • van Rooyen R.J.
        • Reiche B.
        • Hengstenberg W.
        J. Biol. Chem. 1990; 265: 22554-22560
        • Nicholls A.
        • Sharp K.A.
        • Honig B.
        Proteins. 1991; 17: 297-309
        • Keyhani N.O.
        • Boudker O.
        • Roseman S.
        J. Biol. Chem. 2000; 275: 33091-33101
        • Keyhani N.O.
        • Rodgers M.E.
        • Demeler B.
        • Hansen J.C.
        • Roseman S.
        J. Biol. Chem. 2000; 275: 33110-33115
        • Goto N.K.
        • Gardner K.H.
        • Mueller G.A.
        • Willis R.C.
        • Kay L.E.
        J. Biomol. NMR. 1999; 13: 369-374
        • Garcia de la Torre J.
        • Huertas M.L.
        • Carrasco B.
        Biophys. J. 2000; 78: 719-730
        • Delaglio F.
        • Grzesiek S.
        • Vuister G.W.
        • Zhu G.
        • Pfiefer J.
        • Bax A.
        J. Biomol. NMR. 1995; 6: 277-293
        • Garrett D.S.
        • Powers R.
        • Gronenborn A.M.
        • Clore G.M.
        J. Magn. Reson. 1991; 95: 214-220
        • Clore G.M.
        • Gronenborn A.M.
        Science. 1991; 252: 1390-1399
        • Bax A.
        • Grzesiek S.
        Acc. Chem. Res. 1993; 26: 131-138
        • Yang D.W.
        • Kay L.E.
        J. Biomol. NMR. 1999; 14: 273-276
        • Clore G.M.
        • Gronenborn A.M.
        Trends Biotechnol. 1998; 16: 22-34
        • Cornilescu G.
        • Delaglio F.
        • Bax A.
        J. Biomol. NMR. 1999; 13: 289-302
        • Bax A.
        • Vuister G.W.
        • Grzesiek S.
        • Delaglio F.
        • Wang A.C.
        • Tschudin R.
        • Zhu G.
        Methods Enzymol. 1994; 239: 79-105
        • Clore G.M.
        • Starich M.R.
        • Gronneborn A.M.
        J. Am. Chem. Soc. 1998; 120: 10571-10572
        • Hansen M.R.
        • Hanson P.
        • Pardi A.
        Methods Enzymol. 2000; 317: 220-240
        • Yang D.
        • Venters R.A.
        • Mueller G.A.
        • Choy W.Y.
        • Kay L.E.
        J. Biomol. NMR. 1999; 14: 333-343
        • Jaroniec C.P.
        • Ulmer T.S.
        • Bax A.
        J. Biomol. NMR. 2004; 30: 181-194
        • Nilges M.
        • Gronenborn A.M.
        • Brünger A.T.
        • Clore G.M.
        Protein Eng. 1988; 2: 27-38
        • Garrett D.S.
        • Kuszewski J.
        • Hancock T.J.
        • Lodi P.J.
        • Vuister G.W.
        • Gronenborn A.M.
        • Clore G.M.
        J. Magn. Reson. B. 1994; 104: 99-103
        • Kuszewski J.
        • Qin J.
        • Gronenborn A.M.
        • Clore G.M.
        J. Magn. Reson. B. 1995; 106: 92-96
        • Clore G.M.
        • Gronenborn A.M.
        • Tjandra N.
        J. Magn. Reson. 1988; 131: 159-162
        • Schwieters C.D.
        • Clore G.M.
        J. Magn. Reson. 2001; 152: 288-302
        • Schwieters C.D.
        • Kuszewski J.
        • Tjandra N.
        • Clore G.M.
        J. Magn. Reson. 2003; 160: 65-73
        • Clore G.M.
        • Kuszewski J.
        J. Am. Chem. Soc. 2002; 124: 2866-2867
        • Grishaev A.
        • Bax A.
        J. Am. Chem. Soc. 2004; 126: 7281-7292
        • Kuszewski J.
        • Gronenborn A.M.
        • Clore G.M.
        J. Am. Chem. Soc. 1999; 121: 2337-2338
        • Schwieters C.D.
        • Clore G.M.
        J. Magn. Reson. 2001; 149: 239-244
        • DeLano W.L.
        The PyMOL Molecular Graphics System. DeLano Scientific, San Carlos, CA2002
        • Schwieters C.D.
        • Clore G.M.
        J. Biomol. NMR. 2002; 23: 221-225
        • Tugarinov V.
        • Kay L.E.
        J. Am. Chem. Soc. 2004; 126: 9827-9836
        • Al-Hashimi H.M.
        • Bolton P.J.
        • Prestegard J.H.
        J. Magn. Reson. 2000; 142: 153-158
        • Chothia C.
        Annu. Rev. Biochem. 1984; 53: 537-572
        • Lupas A.
        Trends Biochem. Sci. 1996; 21: 375-382
        • Harbury P.B.
        • Plecs J.J.
        • Tidor B.
        • Alber T.
        • Kim P.S.
        Science. 1998; 282: 1462-1467
        • Woolfson D.N.
        • Alber T.
        Protein Sci. 1995; 4: 1596-1607
        • Zeng X.
        • Zhu H.
        • Lashuel H.
        • Hu J.C.
        Protein Sci. 1997; 6: 2218-2226
        • McClain D.L.
        • Binfet J.P.
        • Oakley M.G.
        J. Mol. Biol. 2001; 313: 373-383
        • Pelton J.G.
        • Torchia D.A.
        • Meadow N.D.
        • Roseman S.
        Protein Sci. 1993; 2: 543-558
        • Clore G.M.
        • Garrett D.S.
        J. Am. Chem. Soc. 1999; 121: 9008-9012
        • Bax A.
        • Kontaxis G.
        • Tjandra N.
        Methods Enzymol. 2001; 339: 127-174
        • Williams D.C.
        • Cai M.
        • Clore G.M.
        J. Biol. Chem. 2004; 279: 1449-1457
        • Harbury P.B.
        • Zhang T.
        • Kim P.S.
        • Alber T.
        Science. 1993; 262: 1401-1407
        • Clore G.M.
        • Gronenborn A.M.
        • Bax A.
        J. Magn. Reson. 1998; 133: 216-221
        • Laskowski R.A.
        • MacArthur M.W.
        • Moss D.S.
        • Thornton J.M.
        J. Appl. Crystallogr. 1993; 26: 283-291
        • Vriend G.
        • Sander C.
        J. Appl. Crystallogr. 1993; 26: 47-60