Advertisement

The Structure of Lombricine Kinase

IMPLICATIONS FOR PHOSPHAGEN KINASE CONFORMATIONAL CHANGES*
      Lombricine kinase is a member of the phosphagen kinase family and a homolog of creatine and arginine kinases, enzymes responsible for buffering cellular ATP levels. Structures of lombricine kinase from the marine worm Urechis caupo were determined by x-ray crystallography. One form was crystallized as a nucleotide complex, and the other was substrate-free. The two structures are similar to each other and more similar to the substrate-free forms of homologs than to the substrate-bound forms of the other phosphagen kinases. Active site specificity loop 309–317, which is disordered in substrate-free structures of homologs and is known from the NMR of arginine kinase to be inherently dynamic, is resolved in both lombricine kinase structures, providing an improved basis for understanding the loop dynamics. Phosphagen kinases undergo a segmented closing on substrate binding, but the lombricine kinase ADP complex is in the open form more typical of substrate-free homologs. Through a comparison with prior complexes of intermediate structure, a correlation was revealed between the overall enzyme conformation and the substrate interactions of His178. Comparative modeling provides a rationale for the more relaxed specificity of these kinases, of which the natural substrates are among the largest of the phosphagen substrates.

      Introduction

      Lombricine, arginine, and creatine kinases (EC 2.7.3) are homologous phosphagen kinases that catalyze the buffering of cellular ATP levels through phosphoryl transfer to/from their respective guanidino-containing substrates. The reaction is central to short-term temporal energy buffering (
      • Schlattner U.
      • Forstner M.
      • Eder M.
      • Stachowiak O.
      • Fritz-Wolf K.
      • Wallimann T.
      ,
      • Ellington W.R.
      ) as well as in spatial shuttling of energy from production to consumption sites (
      • Tombes R.M.
      • Shapiro B.M.
      ,
      • Wallimann T.
      • Wyss M.
      • Brdiczka D.
      • Nicolay K.
      • Eppenberger H.M.
      ,
      • Ellington W.R.
      • Kinsey S.T.
      ). A wide array of endergonic processes is driven by nucleotide hydrolysis, from motion in molecular motors, active transport, and synthetic metabolism to signal transduction. Thus, the maintenance of a constant ATP/ADP ratio, displaced far from thermodynamic equilibrium in the face of high and variable rates of ATP turnover, is crucial for cellular homeostasis (
      • Ellington W.R.
      ).
      Different organisms use different phosphagen substrates, usually only one and each with its own specific phosphagen kinase (Fig. 1) (
      • Ellington W.R.
      ). Lombricine kinase, as well as taurocyamine kinase and glycocyamine kinase, is found exclusively in annelids and allied groups (
      • Ellington W.R.
      ). Phylogenetic analyses and studies of the intron/exon organization of the genes of these phosphagen kinases unique to annelids have shown that they are more closely related to creatine kinases (with which they share 50–60% sequence identity) than typical monomeric arginine kinases such as that from the horseshoe crab (
      • Suzuki T.
      • Uda K.
      • Adachi M.
      • Sanada H.
      • Tanaka K.
      • Mizuta C.
      • Ishida K.
      • Ellington W.R.
      ) (40% sequence identity). Annelids are more diverse in their choice of phosphagen, and the substrate specificities of the corresponding kinases are often lower (
      • Suzuki T.
      • Uda K.
      • Adachi M.
      • Sanada H.
      • Tanaka K.
      • Mizuta C.
      • Ishida K.
      • Ellington W.R.
      ).
      Figure thumbnail gr1
      FIGURE 1Representative phosphagen substrates. A variety of guanidino compounds serve the same purpose in different organisms of storing “high energy” phosphates. The pervasiveness of arginine kinase across diverse phyla suggests that it may be the ancestral form (
      • Ellington W.R.
      ), but different organisms have evolved enzymes that catalyze an analogous phosphoryl transfer reaction specifically with one of several different phosphagens, including creatine in vertebrates. Lombricine kinase is found in various terrestrial and marine worms and is an exception in that it has lower specificity, catalyzing reactions with both lombricine and taurocyamine (
      • Ellington W.R.
      • Bush J.
      ).
      Structural work has concentrated on the presumptive ancestral arginine kinase and the vertebrate creatine kinase (
      • Fritz-Wolf K.
      • Schnyder T.
      • Wallimann T.
      • Kabsch W.
      ,
      • Zhou G.
      • Somasundaram T.
      • Blanc E.
      • Parthasarathy G.
      • Ellington W.R.
      • Chapman M.S.
      ), but sequence alignment suggests that subunit fold, if not quaternary structure, is conserved across the family (
      • Ellington W.R.
      ). Common structural hallmarks include a small N-terminal α-helical domain connected by a flexible linker to a larger C-terminal domain of β-sheet flanked by α-helices (
      • Fritz-Wolf K.
      • Schnyder T.
      • Wallimann T.
      • Kabsch W.
      ,
      • Zhou G.
      • Somasundaram T.
      • Blanc E.
      • Parthasarathy G.
      • Ellington W.R.
      • Chapman M.S.
      ). There is also a conserved “essential” cysteine (Cys271 in the horseshoe crab (Limulus polyphemus) arginine kinase (LpAK))
      The abbreviations used are: LpAK, Limulus polyphemus (Atlantic horseshoe crab) arginine kinase; AK, arginine kinase; CK, creatine kinase; GK, glycocyamine kinase; LK, lombricine kinase; GgCKmit, Gallus gallus (chicken) mitochondrial CK; HsCKBB, Homo sapiens (human) brain CK; HsCKmit, H. sapiens sarcomeric mitochondrial CK; OcCKMM, Oryctolagus cuniculus (rabbit) CK muscle dimer; SjAK, Stichopus japonicus (sea cucumber) AK; TcCK, Torpedo californica (Pacific electric ray) CK; TSA, transition state analog; TSAC, transition state analog complex; UcLK, Urechis caupo (Innkeeper worm) LK; bis-Tris, 2-[bis(2-hydroxyethyl)amino]-2-(hydroxymethyl)propane-1,3-diol; r.m.s., root mean square; r.m.s.d., root mean square deviation; AMPPNP, adenosine 5′-(β,γ-imino)triphosphate.
      that may be involved in substrate binding synergy and/or electrostatic catalysis as a thiolate (
      • Furter R.
      • Furter-Graves E.M.
      • Wallimann T.
      ,
      • Gattis J.L.
      • Ruben E.
      • Fenley M.O.
      • Ellington W.R.
      • Chapman M.S.
      ), as well as a highly conserved NEEDH motif (
      • Ellington W.R.
      • Bush J.
      ), of which the second glutamate (LpAK Glu225) is implicated as the base that catalyzes phosphagen proton abstraction (
      • Pruett P.S.
      • Azzi A.
      • Clark S.A.
      • Yousef M.S.
      • Gattis J.L.
      • Somasundaram T.
      • Ellington W.R.
      • Chapman M.S.
      ). Differences between the paralogs are due, in part, to the varying quaternary structures among isoforms (monomer, dimer, or octamer) and adaptations to different cellular targeting (cytoplasmic or mitochondrial intermembrane space).
      Lombricine kinase (LK) is a biological homodimer with relaxed substrate specificity relative to the highly specific arginine and creatine kinases (
      • Suzuki T.
      • Uda K.
      • Adachi M.
      • Sanada H.
      • Tanaka K.
      • Mizuta C.
      • Ishida K.
      • Ellington W.R.
      ). LK catalyzes reactions with either lombricine or taurocyamine substrates but not arginine (Fig. 1). Lombricine (guanidinoethyl phosphoserine) is found as the d-serine isomer in most annelids but as the l-serine form in echiuroid worms (
      • Ellington W.R.
      ), suggesting that close homologs tolerate some stereochemical variation. The apparent dissociation constant and specificity index of Eisenia LK for lombricine are Km = 5.33 mm and kcat/Km = 3.37 s−1mm−1 and are modestly weaker for taurocyamine (Km = 15.31 mm and kcat/Km = 0.48 s−1mm−1, respectively (
      • Tanaka K.
      • Suzuki T.
      )). Our decision to work with taurocyamine in substrate complexes was predicated on the scarcity of lombricine, which must be isolated from kilogram quantities of worms.
      The two variable loops have been implicated in phosphagen kinase substrate specificity (
      • Zhou G.
      • Somasundaram T.
      • Blanc E.
      • Parthasarathy G.
      • Ellington W.R.
      • Chapman M.S.
      ,
      • Suzuki T.
      • Fukuta H.
      • Nagato H.
      • Umekawa M.
      ,
      • Azzi A.
      • Clark S.A.
      • Ellington W.R.
      • Chapman M.S.
      ,
      • Novak W.R.
      • Wang P.F.
      • McLeish M.J.
      • Kenyon G.L.
      • Babbitt P.C.
      ,
      • Jourden M.J.
      • Clarke C.N.
      • Palmer A.K.
      • Barth E.J.
      • Prada R.C.
      • Hale R.N.
      • Fraga D.
      • Snider M.J.
      • Edmiston P.L.
      ). One set (LpAK loop 59–64) and Urechis caupo lombricine kinase (UcLK) loop 53–58) is located in the small N-terminal α-helical domain and facilitates binding of the phosphagen substrate through backbone hydrogen bonds to the carboxylate of the substrate. With loop length inversely correlated to substrate size, the mechanism of the small domain specificity loop has been rationalized in terms of lock-and-key steric hindrance (
      • Azzi A.
      • Clark S.A.
      • Ellington W.R.
      • Chapman M.S.
      ,
      • Jourden M.J.
      • Clarke C.N.
      • Palmer A.K.
      • Barth E.J.
      • Prada R.C.
      • Hale R.N.
      • Fraga D.
      • Snider M.J.
      • Edmiston P.L.
      ,
      • Suzuki T.
      • Kamidochi M.
      • Inoue N.
      • Kawamichi H.
      • Yazawa Y.
      • Furukohri T.
      • Ellington W.R.
      ,
      • Lim K.
      • Pullalarevu S.
      • Surabian K.T.
      • Howard A.
      • Suzuki T.
      • Moult J.
      • Herzberg O.
      ).
      The other loop is in the large domain (LpAK loop 311–319 and UcLK loop 309–317) and serves to align and position the guanidinium of the substrate for optimal nucleophilic attack on ATP (
      • Zhou G.
      • Somasundaram T.
      • Blanc E.
      • Parthasarathy G.
      • Ellington W.R.
      • Chapman M.S.
      ,
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ). In arginine kinase, interactions between Glu312 and the guanidinium are prominent in this alignment. Creatine kinases have a valine at the corresponding position, which forms a hydrophobic mini-pocket accommodating the methyl group distinctive for creatine (
      • Novak W.R.
      • Wang P.F.
      • McLeish M.J.
      • Kenyon G.L.
      • Babbitt P.C.
      ,
      • Jourden M.J.
      • Clarke C.N.
      • Palmer A.K.
      • Barth E.J.
      • Prada R.C.
      • Hale R.N.
      • Fraga D.
      • Snider M.J.
      • Edmiston P.L.
      ,
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ). However, mediation of specificity is more complex than lock-and-key. In a chimeric construct with the CK specificity determinants in an LpAK background, it is possible to regain AK activity with additional mutations (
      • Azzi A.
      • Clark S.A.
      • Ellington W.R.
      • Chapman M.S.
      ). Furthermore, a creatine-LpAK structure shows that creatine is not excluded from the active site but is imperfectly aligned with the nucleotide (
      • Azzi A.
      • Clark S.A.
      • Ellington W.R.
      • Chapman M.S.
      ). In crystal structures, the loop has been fully resolved only in the presence of substrates. Consistent with disorder in the substrate-free states, NMR-based Lipari-Szabo analysis has identified inherent nanosecond dynamics in this region (
      • Davulcu O.
      • Flynn P.F.
      • Chapman M.S.
      • Skalicky J.J.
      ). It appears that substrates either induce an ordering or select from an ensemble of loop conformations (
      • Boehr D.D.
      • Dyson H.J.
      • Wright P.E.
      ,
      • Boehr D.D.
      • Wright P.E.
      ,
      • Ma B.
      • Kumar S.
      • Tsai C.J.
      • Nussinov R.
      ) to achieve a substrate-bound conformation that is catalytically competent for the cognate phosphagen substrate. The loop was fully resolved in one of the two subunits in the substrate-free UcLK structure reported here, providing an improved basis for understanding substrate-associated conformational changes in phosphagen kinases.
      Past NMR and kinetic studies have shown that creatine and arginine kinases share the same rapid equilibrium, random order, bimolecular-bimolecular mechanism with direct, partially associative, in-line phosphoryl transfer (
      • Hansen D.E.
      • Knowles J.R.
      ,
      • Murali N.
      • Jarori G.K.
      • Landy S.B.
      • Rao B.D.
      ,
      • Murali N.
      • Jarori G.K.
      • Rao B.D.
      ). Catalytic rates of ∼135 s−1 (
      • Blethen S.L.
      ,
      • Rao B.D.
      • Buttlaire D.H.
      • Cohn M.
      ) are consistent with turnover-limiting protein conformational changes (
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ,
      • Rhee S.
      • Parris K.D.
      • Hyde C.C.
      • Ahmed S.A.
      • Miles E.W.
      • Davies D.R.
      ,
      • Gerstein M.
      • Chothia C.
      ,
      • Boyer R.
      ). Paired substrate-free and transition state analog complex (TSAC) structures are available for both arginine kinase and creatine kinase; the TSAC has bound phosphagen, ADP, and a nitrate mimicking the γ-phosphoryl in transit (
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ,
      • Yousef M.S.
      • Clark S.A.
      • Pruett P.K.
      • Somasundaram T.
      • Ellington W.R.
      • Chapman M.S.
      ). Upon substrate binding there are domain reorientations up to 18°, together with the ordering and/or reconfiguring of two flexible loops over the substrates. The reconfiguration of the active site, which likely removes solvent water from where it could participate in a wasteful side reaction, appears to be critical in the alignment of enzyme catalytic groups and in the precise alignment of reactive substrate atoms with each other (
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ).
      NMR relaxation dispersion measurements have identified intrinsic collective motions in the interface between the N- and C-terminal domains and in LpAK active site loop 182–209 (corresponding to UcLK loop 175–203) (
      • Davulcu O.
      • Flynn P.F.
      • Chapman M.S.
      • Skalicky J.J.
      ). These motions, measured in the absence of substrates, occur at turnover-commensurate rates (
      • Davulcu O.
      • Flynn P.F.
      • Chapman M.S.
      • Skalicky J.J.
      ) at sites that are also implicated in the substrate-associated conformation changes (
      • Niu X.
      • Brüschweiler-Li L.
      • Davulcu O.
      • Skalicky J.J.
      • Brüschweiler R.
      • Chapman M.S.
      ). These observations suggest that substrate-associated changes take advantage of modes of flexibility intrinsic to the enzyme and that some of the intrinsic motions may be rate-limiting on turnover. The comparison of the nucleotide-bound and -free UcLK structures reported here adds to our understanding of the dependence of domain configuration upon the presence of substrates.

      EXPERIMENTAL PROCEDURES

       Expression and Purification of Lombricine Kinase

      Cloning of the lombricine kinase has been described previously (
      • Ellington W.R.
      • Bush J.
      ). Expression was as follows. A 1.8-liter culture of Escherichia coli was inoculated and grown at 37 °C in LB medium containing 1% glucose and 50 μg/ml carbenicillin to an A600 of ∼0.5. Induction of expression with 1 mm isopropyl 1-thio-β-d-galactopyranoside followed 1 h after reducing the temperature to 15 °C. Cells were harvested 24 h post-induction, pelleted at 10,000 × g, and resuspended in 50 ml of 50 mm Tris/HCl, 100 mm NaCl, 10 mm EDTA, 15% w/v sucrose, and 14 mm β-mercaptoethanol, pH 8.2, at 298 K. Cells were lysed using a Model 110L Microfluidizer (Microfluidics Corp., Newton, MA) at 13,000 p.s.i. Cell debris was pelleted by centrifugation at 14,000 × g and filtered through glass wool before dialysis against 10 mm Tris, 1 mm EDTA, 5 mm KCl, 0.02% w/v NaN3, and 1 mm DTT, pH 8.6, at 298 K. The protein solution was clarified by centrifugation at 14,000 × g and purified by DEAE anion-exchange chromatography on an AKTA FPLC (GE Healthcare) eluting at ∼60 mm KCl on a 0–120 mm gradient. Peak fractions were confirmed by SDS-PAGE to be Mr ≅ 41,000 and were active in the phosphorylation of lombricine and taurocyamine. The DEAE peak fractions were pooled and dialyzed against 50 mm bis-Tris, 1 mm MgCl2, 5 mm NaNO3, and 1 mm DTT, pH 6.5, at 298 K and then further purified by affinity chromatography using ADP-agarose (Sigma) eluted with a 0–80 mm NaCl gradient. This separated the desired active dimer from a misfolded inactive monomer as confirmed by Superdex 200 HR analytical size-exclusion chromatography (GE Healthcare) and measured activities of 10–18 μm/min/ml for the dimer (using a taurocyamine substrate). Purified LK was dialyzed against 10 mm Tris, 5 mm KCl, and 2 mm DTT, pH = 8.1, at 298 K, concentrated to 30 mg/ml with an Amicon pressurized cell (Millipore) prior to crystallization.

       Crystallization and Data Collection

      Initial crystallization leads (needles) were obtained through subscription to a commercial sparse matrix screening service (Syrrx, Inc.) (
      • Hosfield D.
      • Palan J.
      • Hilgers M.
      • Scheibe D.
      • McRee D.E.
      • Stevens R.C.
      ). Conditions were scaled up from nanoliter to microliter vapor diffusion and optimized in-house. Two crystal forms were obtained with similar crystallization conditions: a 1:1 mixture of 30 mg/ml LK and mother liquor, equilibrated at 277 K by vapor diffusion against a mother liquor containing 15 mm bis-Tris, 0.2 m NaNO3, 1 mm DTT, and 20% w/v PEG 3350MME, pH = 6.8. The difference was that the second form was crystallized at pH 5.8 with additional TSA components in the mixed protein/precipitant drop (50 mm taurocyamine, 40 mm ADP, and 5 mm MgCl2). The first, apo-form crystals grew to 0.15–0.4 mm with a full physiological dimer in the asymmetric unit. The second “ADP form” grew to 0.15–0.35 mm with a single subunit in the asymmetric unit and was later found to contain ADP, but not taurocyamine, in the active site.
      Crystals were cryoprotected by serial transfer to 35% glycerol in ∼2 min by 5% w/v increments. ADP-form cryoprotectant transfers were always in the presence of TSA components. Apo-form diffraction data were collected at beamline A1 at the MacCHESS resource (Ithaca, NY) at 100 K using an ADSC Quantum 210 2 × 2 CCD detector (
      • Szebenyi D.M.
      • Arvai A.
      • Ealick S.
      • Laiuppa J.M.
      • Nielsen C.
      ). Data for the ADP form were collected at SER-CAT beamline 22-BM at the Advanced Photon Source (Argonne, IL). Data were processed using the HKL package (
      • Otwinowski Z.
      • Minor W.
      ).

       Structure Determination

      Even though the structure of a homolog with 57% sequence identity was available, LK did not immediately yield to molecular replacement phasing. In retrospect, this was due to modest diffraction quality (Table 1), interdomain flexibility and the unexpected open enzyme conformation for the ADP complex in contrast to the closed form of other phosphagen kinase substrate complexes (see below). Success came with the building of a crude homology model as a molecular replacement probe. This phasing model was based on the structure with highest sequence homology, human ubiquitous mitochondrial creatine kinase (Protein Data Bank ID 1QK1 (
      • Eder M.
      • Fritz-Wolf K.
      • Kabsch W.
      • Wallimann T.
      • Schlattner U.
      )), determined in the substrate-free state. A phosphagen kinase multiple sequence alignment was constructed with ClustalX (
      • Chenna R.
      • Sugawara H.
      • Koike T.
      • Lopez R.
      • Gibson T.J.
      • Higgins D.G.
      • Thompson J.D.
      ) and visualized with GeneDoc. The sequence in the human CK structure was then changed computationally to that of LK using the program SEAMAN (
      • Kleywegt G.J.
      ), ignoring gaps. The stereochemistry of the predicted model was then refined with the program MODELLER (
      • Sali A.
      • Blundell T.L.
      ,
      • Fiser A.
      • Do R.K.
      • Sali A.
      ).
      TABLE 1Crystallographic statistics
      Crystal formApo-formADP form
      Protein solutionLK (28 mg/ml), 1 mm DTT, 15 mm bis-Tris, pH 6.8LK (28 mg/ml), 50 mm taurocyamine, 40 mm ADP, 5 mm MgCl2, 1 mm DTT, 15 mm bis-Tris, pH 5.8
      Precipitant20% PEG 3350MME, 0.2 m NaNO320% PEG 3350MME, 0.2 m NaNO3
      Space groupP21C2221
      Unit cell dimensionsa = 74.3 Å, b = 59.6 Å, c = 85.9 Å, β = 105.0°a = 67.7 Å, b = 78.0 Å, c = 141.1 Å
      Contents of asymmetric unit (non-hydrogen atoms)Dimer (2 × 5,674 protein atoms + 5 × 4 nitrate atoms + 881 H2O)Single subunit (2,860 protein + 27 in ADP + 194 H2O)
      Resolution range (outer shell)28.5–1.95 Å (1.99–1.95 Å)20.0–2.50 Å (2.57–2.50 Å)
      No. of observations178,36474,192
      No. of unique reflections50,44112,706
      Completeness95% (90%)99.8% (99%)
      Rsym0.124 (0.218)0.122 (0.174)
      I/σ(I), mosaicity14.5 (6.7), 1.3°17.0 (5.2), 0.54°
      Rwork/Rfree0.212/0.284 (0.254/0.349)0.179/0.250 (0.218/0.329)
      r.m.s. deviation from ideal
      Bond length0.004 Å0.003 Å
      Bond angles0.7°0.7°
      Mean B-value (protein/substrate/solvent)26.2 Å2 (25.4/none/28.1 Å2)27.4 Å2 (27.1/67.2/27.2 Å2)
      Maximum likelihood estimate of coordinate error0.36 Å0.34 Å
      The substrate-free homology model yielded putative solutions for the rotation and translation functions of the ADP form, first using the CNS (crystallography and NMR system) program (
      • Grosse-Kunstleve R.W.
      • Adams P.D.
      ) and then confirmed with Phaser (
      • McCoy A.J.
      • Grosse-Kunstleve R.W.
      • Storoni L.C.
      • Read R.J.
      ,
      • Brünger A.T.
      • Adams P.D.
      • Clore G.M.
      • DeLano W.L.
      • Gros P.
      • Grosse-Kunstleve R.W.
      • Jiang J.S.
      • Kuszewski J.
      • Nilges M.
      • Pannu N.S.
      • Read R.J.
      • Rice L.M.
      • Simonson T.
      • Warren G.L.
      ) and AMoRe, both from the CCP4 suite (
      • Navaza J.
      • Saludjian P.
      ). Following a preliminary refinement of the ADP complex structure, it was then possible to find a molecular replacement solution for the apo-form structure that was consistent with the 2-fold noncrystallographic symmetry revealed in its self-rotation function.
      Models initially were rebuilt with the program O (
      • Jones T.A.
      ). Refinement iterated cycles of optimization with Refmac (
      • Murshudov G.N.
      • Vagin A.A.
      • Dodson E.J.
      ) or Phenix (
      • Adams P.D.
      • Grosse-Kunstleve R.W.
      • Hung L.W.
      • Ioerger T.R.
      • McCoy A.J.
      • Moriarty N.W.
      • Read R.J.
      • Sacchettini J.C.
      • Sauter N.K.
      • Terwilliger T.C.
      ) with interactive rebuilding using Coot (
      • Emsley P.
      • Cowtan K.
      ) and model fit evaluated using Map2mod (
      • Kirillova O.
      • Minor W.
      ). During the final cycles of Phenix refinement, atomic displacements were refined combining a translation-libration-screw (TLS) approximation (
      • Howlin B.
      • Moss D.S.
      • Harris G.W.
      ,
      • Winn M.D.
      • Isupov M.N.
      • Murshudov G.N.
      ) with restrained refinement of individual atomic B-factors. Five TLS rigid groups were used, regions contiguous in sequence that approximated the subdomain assignments and the large domain specificity loop (see below).

       Structure Comparisons

      Overall comparisons of subunit structures were performed with the protein structure comparison service PDBeFold SSM (secondary structure matching) at the European Bioinformatics Institute using default parameters (
      • Krissinel E.
      • Henrick K.
      ). The anatomy of conformational change was examined with reference to conformational differences between the transition state analog and substrate-free forms of LpAK, the phosphagen kinase that has the largest of otherwise similar conformational changes through the family (
      • Niu X.
      • Brüschweiler-Li L.
      • Davulcu O.
      • Skalicky J.J.
      • Brüschweiler R.
      • Chapman M.S.
      ). The conformational changes in LpAK are approximated well by rigid-group motions of five subdomains, sometimes referred to as “dynamic domains” (
      • Hayward S.
      • Berendsen H.J.
      ) because they consist of spatial clusters of residues that need not be contiguous in sequence but that share a common displacement. For the following study, subdomains for UcLK and other phosphagen kinases were defined by overlaying those from LpAK according to the secondary structure matching alignment (see above) after making a handful of manual adjustments in which the pairwise sequence alignment could clearly be improved. Subdomain displacements were compared using malign.py, a program developed here that performs multiple gap-penalized sequence-structure alignments using the superpose_pdbs routine from the Phenix package (
      • Adams P.D.
      • Afonine P.V.
      • Bunkoczi G.
      • Chen V.B.
      • Davis I.W.
      • Echols N.
      • Headd J.J.
      • Hung L.W.
      • Kapral G.J.
      • Grosse-Kunstleve R.W.
      • McCoy A.J.
      • Moriarty N.W.
      • Oeffner R.
      • Read R.J.
      • Richardson D.C.
      • Richardson J.S.
      • Terwilliger T.C.
      • Zwart P.H.
      ). In a first pass, matching Cα coordinates throughout the comparison structure were least-squares-aligned to the corresponding LK residues. In a second pass, the alignment was localized to each subdomain in turn. Rigid-group transformations were calculated by comparing the locally and globally aligned structures.

       Prediction of the UcLK TSA Structure

      The UcLK-ADP complex was adapted into a transition state form through energy minimization and molecular dynamics using MODELLER (
      • Eswar N.
      • Eramian D.
      • Webb B.
      • Shen M.
      • Sali A.
      ) and restraints to an ensemble of AK, CK, and GK transition state or Michaelis analog complex structures: LpAK, Stichopus japonicus AK (SjAK), human HsCKBB, TcCK, rabbit muscle OcCKMM, and Namalycastis sp. (marine worm) GK (NsGK) (
      • Lim K.
      • Pullalarevu S.
      • Surabian K.T.
      • Howard A.
      • Suzuki T.
      • Moult J.
      • Herzberg O.
      ,
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ,
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ,
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ,
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ,
      • Wu X.
      • Ye S.
      • Guo S.
      • Yan W.
      • Bartlam M.
      • Rao Z.
      ). For this restrained optimization, the phosphagens were simplified to their common moiety, methyl guanidine. Lombricine was added to the lowest energy model (from ∼200 optimizations) by superimposing the guanidinium group and rotating the torsion angles to maximize overlap with substrate arginine as superimposed from the LpAK TSA structure. The coordinates of the LK protein and lombricine were then optimized by simulated annealing using Gromacs (
      • Van Der Spoel D.
      • Lindahl E.
      • Hess B.
      • Groenhof G.
      • Mark A.E.
      • Berendsen H.J.
      ), an explicit water model, and restraints to dampen the motion of the ADP, nitrate, and guanidinium moiety of the phosphagen substrate. The annealing protocol involved slow cooling from 2000 K followed by energy minimization, all using the Gromos53a6 force field (
      • Oostenbrink C.
      • Villa A.
      • Mark A.E.
      • Van Gunsteren W.F.
      ).

      RESULTS AND DISCUSSION

       Structure Determination

      Structures were determined successfully for two crystal forms: the ADP form, with a single subunit in the asymmetric unit; and the ADP form, with a dimer (Table 1). Substrates were not included in the initial phasing models for the ADP form. Electron density for the nucleotide phosphates was stronger than the average protein density, whereas ribose and base were weaker. However, there was no doubt as to its identity when the active sites of LpAK TSAC and a CK nucleotide complex (
      • Zhou G.
      • Somasundaram T.
      • Blanc E.
      • Parthasarathy G.
      • Ellington W.R.
      • Chapman M.S.
      ,
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ) were superimposed (Fig. 2). No density was visible for either the taurocyamine or the nitrate, which were also present in the crystallization solutions. Thus, attempts to crystallize UcLK as TSAC yielded only the binary nucleotide complex. Analogously, crystals of multimeric CK have been reported with substrates bound to some subunits but not others (
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ,
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ,
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ). (Such asymmetry was cited as supporting negative cooperativity (
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ,
      • Wu X.
      • Ye S.
      • Guo S.
      • Yan W.
      • Bartlam M.
      • Rao Z.
      ), but this argument has been undercut by a glycocyamine kinase structure with both subunits in the closed substrate-bound configuration (
      • Lim K.
      • Pullalarevu S.
      • Surabian K.T.
      • Howard A.
      • Suzuki T.
      • Moult J.
      • Herzberg O.
      ).) Lombricine was not available, and the concentration of the alternative substrate, taurocyamine, limited by solubility to 3× Km, may not have been sufficient to obtain TSAC crystals, even though LpAK has been crystallized with a wide variety of weakly binding substrate analogs (
      • Azzi A.
      • Clark S.A.
      • Ellington W.R.
      • Chapman M.S.
      ).
      S. A. Clark, E. A. Ruben, M. S. Yousef, J. D. Bush, M. O. Fenley, J. D. Evanseck, W. R. Ellington, and M. S. Chapman, manuscript in preparation.
      Figure thumbnail gr2
      FIGURE 2Electron density for points of interest in the substrate-free and ADP-bound crystal structures. A, an overview of the substrate-free subunit of UcLK colored by subdomain (as determined in LpAK); gray indicates unassigned. The inset shows the TSA structure of LpAK and the location of substrates (stick model) in the active site (
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ). B and C, stereographic pairs showing maps calculated with coefficients of 2mFo − DFc following maximum likelihood atomic refinement (
      • Read R.J.
      ). B, the substrate-free UcLK-(308–317) flexible loop of subunit A is colored by atom type and its map contoured at 1.1 σ. In other phosphagen kinases this substrate specificity loop is highly disordered. The LK conformation is unique among substrate-free structures but is related to the nucleotide-bound structure (orange) by a quasi-rigid 7° rotation. C, the UcLK-ADP complex is colored by atom type. Density for part of subdomain 3 is contoured at 1.25 σ and at 0.7 σ for the ADP. The protein structure is similar to that of substrate-free UcLK (cyan), except for His178, which has distinctly different side chain density. Although the ADP is similar to that in the LpAK-TSA complex (dark brown) (
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ), with mostly similar interactions, subdomain 3 is much more like substrate-free LpAK (light brown) (
      • Yousef M.S.
      • Clark S.A.
      • Pruett P.K.
      • Somasundaram T.
      • Ellington W.R.
      • Chapman M.S.
      ).

       Subunit Structure and Comparison with Homologs

      Of the known phosphagen kinase structures, UcLK is structurally most similar to that of the chicken mitochondrial creatine kinase (GgCKmit; r.m.s.d. = 0.96 Å) with which it has the highest sequence identity (57%). Other substrate-free vertebrate CKs follow shortly thereafter. Like the other phosphagen kinases, LK has a 100-residue N-terminal α-helical domain followed by a 250-residue mostly β-sheet domain.
      The overall structures of the substrate-free and ADP complex UcLK are surprisingly similar (r.m.s.d. = 0.54 Å), considering the conventional wisdom that most of the >2 Å r.m.s. conformational changes (
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ,
      • Yousef M.S.
      • Clark S.A.
      • Pruett P.K.
      • Somasundaram T.
      • Ellington W.R.
      • Chapman M.S.
      ) between open and closed state phosphagen kinases are thought to be nucleotide-induced (
      • Dumas C.
      • Janin J.
      ,
      • Forstner M.
      • Kriechbaum M.
      • Laggner P.
      • Wallimann T.
      ). In fact, the overall difference (0.54 Å) is little more than between the two subunits in the dimeric crystal form (r.m.s.d. = 0.40 Å), so systematic differences are barely observable above experimental error. With such modest differences between ADP-bound and substrate-free annelid UcLK, the ADP-bound LK is much closer to substrate-free chicken GgCKmit (r.m.s.d. = 0.9 Å) than to the most similar transition state CK, human brain (HsCKBB; r.m.s.d. = 1.9 Å) (
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ). The annelid UcLK-ADP complex shows little of the substrate-bound character described for other phosphagen kinases.
      With this in mind, both LK structures were compared with a larger group of phosphagen kinases, including representatives of substrate-free, nucleotide, and transition state complexes (Table 2). Table 2 is readily ordered to show that the overall magnitude of conformational changes between the substrate-free and TSA states varies systematically: CK (2.7 Å Cα r.m.s.) > GK (2.4 Å) > CK (2.0 Å). The substrate-free and TSA states have been regarded as the open and closed forms, respectively, but the homologs actually exhibit varying degrees of closure. The majority of binary ADP complexes have hitherto been regarded as closed form, a good first approximation, but one that will be revisited (see below). One of the subunits in the rabbit muscle OcCKMM structure was seen as an exception in open form (
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ), but with the addition of the annelid UcLK structures, the binary complexes can be interpreted in a new light.
      TABLE 2Cα r.m.s. differences between superimposed subunits of lombricine kinase and representative creatine and glycocyamine kinases, following superimposition with SSM (
      • Krissinel E.
      • Henrick K.
      )
      Within the difference matrix (Table 2), the most consistent placement of both the substrate-free and nucleotide-bound structures of annelid UcLK is between the ADP complex of rabbit muscle OcCKMM and substrate-free chicken mitochondrial GgCKmit. The open form taken by one of the subunits of the rabbit OcCKMM-ADP complex has been rationalized in terms of cooperativity (
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ). Such rationalization cannot explain the even more open annelid UcLK-ADP structure (Fig. 3, C and D), which is from a crystallographically symmetric dimer and therefore does not exhibit any cooperativity that could inhibit enzyme closure. With the addition of the annelid UcLK-ADP structure, it is now clear that in addition to the majority “closed form,” there is a second group of nucleotide complexes that are more open than those first characterized. The ADP complexes of TcCK (
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ) and most subunits of human mitochondrial HsCKmit (Protein Data Bank ID 2GL6) are more comparable to the closed form TSA CKs (0.7–1.1 Å Cα r.m.s.) than they are to the open, substrate-free forms (1.4–1.9 Å). This is mirrored by the binary complexes of rabbit muscle OcCKMM and annelid UcLK that appear closer to the substrate-free forms (∼0.9 Å) than transition state CKs (1.7–2.1 Å). Furthermore, Table 2 shows rabbit OcCKMM-ADP to be slightly more closed like than annelid UcLK-ADP. Table 2 also makes apparent the differences in human mitochondrial HsCKmit (Protein Data Bank ID 2GL6) between the mostly closed D subunit (like TcCK-ADP) and the other subunits, which are more similar to the fully closed TSA states. In summary, this analysis shows that: 1) corresponding states of homologous phosphagen kinases (and even of the same enzyme) exhibit varying degrees of openness/closure; and 2) the greatest variation is in the nucleotide-bound state, for which a full spectrum of conformers from nearly fully open to fully closed is observed. The determinants of the extent of closure on ADP binding are unknown and could be specific for each homolog. Alternatively, in solution, the ADP-bound enzyme may exist as an equilibrium between multiple states of near equal energy, with minor environmental factors selecting the conformer seen in each crystal structure. The latter becomes more plausible with NMR evidence that, in solution, substrate-free LpAK contains a minor fraction in the closed form of the enzyme (
      • Niu X.
      • Brüschweiler-Li L.
      • Davulcu O.
      • Skalicky J.J.
      • Brüschweiler R.
      • Chapman M.S.
      ), likely in dynamic equilibrium with the predominant open form, suggesting that the energetic difference between open and closed forms is small.
      Figure thumbnail gr3
      FIGURE 3His178 and subdomain 3 configurations. A, ADP interactions of UcLK His178 and corresponding residues from representative homologs. Four TSA structures (LpAK, TcCK, rabbit muscle OcCKMM, and human brain HsCKBB (all colored pink)) (
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ,
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ,
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ,
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ) are tightly clustered with a histidine hydrogen bond to the ribose O2′. Eight representative substrate-free structures (chicken mitochondrial GgCKmit, bovine retinal CK, human muscle CK, chicken brain GgCKBB, human mitochondrial HsCKmit, rabbit muscle CK, Trypanosoma cruzi AK, and human brain HsCKBB (yellow)) (
      • Fritz-Wolf K.
      • Schnyder T.
      • Wallimann T.
      • Kabsch W.
      ,
      • Fernandez P.
      • Haouz A.
      • Pereira C.A.
      • Aguilar C.
      • Alzari P.M.
      ,
      • Eder M.
      • Fritz-Wolf K.
      • Kabsch W.
      • Wallimann T.
      • Schlattner U.
      ,
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ,
      • Rao J.K.
      • Bujacz G.
      • Wlodawer A.
      ,
      • Tisi D.D.
      • Bax B.B.
      • Loew A.A.
      ,
      • Shen Y.
      • Tang L.
      • Zhou H.
      • Lin Z.
      ,
      • Eder M.
      • Schlattner U.
      • Becker A.
      • Wallimann T.
      • Kabsch W.
      • Fritz-Wolf K.
      ) are more loosely clustered over a 2 Å range, about 4 Å from the transition state location. The histidine side chains have no substrate to with which to interact and have a variable orientation. In CK binary nucleotide complexes (TcCK, HsCKBB, and HsCKmit (orange;
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ,
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ) PDB ID 2GL6), the histidine often adopts a TSA-like configuration, but the D subunit of HsCKmit and the A subunit of rabbit muscle OcCKMM (also orange) are more substrate-free-like (
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ). The UcLK-ADP complex (colored by atom type) is even more substrate-free-like, except that, relative to substrate-free UcLK (blue), there is a 120° rotation of χ1 so that the side chain can hydrogen bond with the ribose O3′ (not O2′ as in TSA structures). B, a ribbon representation shows the broader impact on UcLK residues 162–213. The biggest differences (∼4 Å) are in subdomain 3 (center) where TSA (pink) and substrate-free (yellow) structures form two clusters; the ADP-CK complexes (orange) are intermediate, whereas both LK structures, substrate-free (blue) and ADP complex (green), are like substrate-free CKs. C, comparison of UcLK substrate-free (green) and ADP complex (red) with representative substrate-free CKs (translucent cyan) after least-squares superimposition of subdomain 4 (see A for subdomain definitions), the largest and least mobile of the subdomains. Both LK structures, for the most part, fall within the range of substrate-free CKs. The N-terminal subdomain 1 in LK has a slightly more open orientation than in CK. D, similar comparison of the LK structures with representative ADP complexes of CK (translucent cyan). These CK-ADP complexes have a broader range of configurations, showing varying degrees of closure. The UcLK structures are both most similar to the most open of the CK complexes.

       Dynamic Subdomains

      The conformational changes on adoption of the transition state have been examined crystallographically in detail for LpAK and compared with the intrinsic dynamics in the substrate-free state using solution NMR (
      • Niu X.
      • Brüschweiler-Li L.
      • Davulcu O.
      • Skalicky J.J.
      • Brüschweiler R.
      • Chapman M.S.
      ). Backbone motions are well approximated by the rigid-group motions of five subdomains. Four of the group motions are almost pure rotations, the lone exception being subdomain 3, which has a larger translational component. The LpAK subdomain designations provide a basis for analyzing the subtle differences between the ADP-bound and substrate-free UcLK and for understanding the nature of differences between these structures and phosphagen kinase homologs.
      With separate superimposition of the five subdomains (Fig. 2A), the difference between the ADP-bound and substrate-free forms of UcLK are reduced from r.m.s.d. = 0.5–0.3 Å, i.e. to the estimated residual error of the structures (Table 1). Thus, the rigid-group approximation is an appropriate low parameter framework for discussion of the subtle changes in UcLK upon ADP binding. The changes in subdomains 1, 2, 4, and 5 are indeed very subtle, less than 0.4 Å, and less than the differences with the closest structural homolog to LK, chicken GgCKmit (Table 3). Interestingly, the rotations of subdomains 1, 2, and 4 (Table 3, line 1) combined with the ADP-to-TSA change (line 5) approximately equal the apo/TSA comparison (line 4), indicating that the small changes in UcLK on ADP binding overall are in the direction of closure but are only ∼10% of the full magnitude. Subdomain 3 undergoes a larger translation of 0.8 Å. The cause is likely to be the interactions of His178 with the ADP-ribose, modulating the conformation of a loop known to be dynamic (
      • Davulcu O.
      • Flynn P.F.
      • Chapman M.S.
      • Skalicky J.J.
      ) (as discussed below).
      TABLE 3Subdomain configurations
      TargetSubunitComplexSubunitComplexSubdomain rotations (r.m.s. displacement)
      12345
      UcLKBUcLK (annelid)AADP1.0° (0.3 Å)0.6° (0.3 Å)3.9° (0.8 Å)0.6° (0.2 Å)1.4° (0.4 Å)
      UcLKBLpAK (arthropod)B3.7° (0.9 Å)2.7° (0.7 Å)5.9° (1.6 Å)5.0° (0.9 Å)17.8° (3.0 Å)
      UcLKBGgCKmit (chicken)B2.9° (0.3 Å)1.0° (0.2 Å)2.1° (0.7 Å)0.6° (0.2 Å)7.7° (1.3 Å)
      UcLKBOcCKMM (rabbit)BTSA10.2° (2.2 Å)7.7° (1.5 Å)31.6° (5.1 Å)3.5° (0.8 Å)5.6° (2.7 Å)
      UcLKAADPOcCKMM (rabbit)BTSA9.5° (2.3 Å)7.1° (1.5 Å)30.3° (4.8 Å)3.0° (0.6 Å)5.1° (2.8 Å)
      UcLKAADPOcCKMM (rabbit)AADP6.5° (1.1 Å)4.7° (1.1 Å)47.8° (2.7 Å)1.5° (0.5 Å)6.4° (1.9 Å)
      When substrate-free UcLK is compared with other substrate-free structures (Table 3), there are differences again in subdomain 3 and also in subdomain 5, which has various orientations in the different structures. Subdomain 5 brackets the large domain flexible loop (which is not included in the subdomain alignments). The whole subdomain has a more uniform configuration when the loop becomes ordered through substrate interactions in the transition state form. Absent these interactions, subdomain 5 appears less constrained in orientation.
      In the prior section, it was noted that although other ADP complexes were of closed, TSA-like configuration, rabbit muscle OcCMMM and annelid UcLK were progressively open and substrate-free-like. When annelid UcLK-ADP is compared with the ADP and TSA complexes of rabbit OcCKMM on a subdomain basis, we find that rabbit OcCKMM-ADP is about halfway between UcLK-ADP and the fully closed rabbit OcCKMM-TSA (Table 3). Overlaid on the rigid-group motions are changes in the flexible loops, but before describing these, it will be helpful to consider the enzyme-nucleotide interactions in the active site.

       Comparison of Open and Closed States in the Nucleotide Binding Site

      The conformation of ADP and its binding subsite in UcLK are highly isomorphous with LpAK-TSAC and the ADP binary complex in rabbit muscle OcCKMM-subunit A (
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ,
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ). Least-squares superimposition of the ADP and 7 of 9 interacting residues shows full agreement within the precision of the structures (Fig. 2C). Comparing the UcLK-ADP structure with the LpAK-TSAC, 12 of 13 hydrogen bonds and salt bridges (
      • Chapman M.S.
      ) between ADP and enzyme are conserved. Such consistency was unexpected, because UcLK-ADP and LpAK-TSAC are in the open and closed state conformations, respectively, which AK and CK solution scattering indicates is determined primarily by nucleotide binding (
      • Dumas C.
      • Janin J.
      ,
      • Forstner M.
      • Kriechbaum M.
      • Laggner P.
      • Wallimann T.
      ). Mg2+, required for ADP- or ATP-induced changes to solution scattering, was present in the crystallization solution but was not coordinated to the nucleotide in the UcLK-ADP structure. However, the SjAK-arginine-AMPPNP structure also lacks Mg2+ but is in the closed form (
      • Wu X.
      • Ye S.
      • Guo S.
      • Yan W.
      • Bartlam M.
      • Rao Z.
      ), and the nucleotide binding sites of Mg2+-free UcLK-ADP and Mg2+-bound LpAK-TSAC and rabbit muscle OcCKMM-ADP do not differ appreciably. So, Mg2+ appears not to be the critical determinant. It is intriguing then how nucleotide binding results in substantial remote conformational changes when the binding pockets of the open state (annelid UcLK-ADP and rabbit OcCKMM-ADP) and the closed state (other ADP-TSA complexes) differ so little.
      Two differences can be seen in the nucleotide-binding pocket. The main chain of UcLK Arg117 is shifted relative to LpAK Arg124. With a different rotamer, the same side chain interactions are maintained with the β-phosphate, so that differences at this arginine appear to be an adaptation to, rather than a cause of, different subdomain orientations. Of greater interest, annelid UcLK, rabbit muscle OcCKMM-subunit A, and LpAK differ in the coordination of the ribose O2′ and O3′ hydroxyls. In LpAK-TSAC, the side chain of His185 hydrogen bonds to the O2′, but in UcLK-ADP the corresponding His178 hydrogen bonds to O3′ along with the carbonyl of Ser308. Between the O2′ coordinated open form UcLK-ADP and the O3′ coordinated closed form TSA represented by LpAK, there is a 5.2 Å difference in the position of the histidine Cα. The corresponding rabbit OcCKMM His190 has no interactions with the nucleotide, and its Cα position is intermediate between the other two (Fig. 3A). There appears to be a correlation among the type of histidine-ribose interaction, the orientation of subdomain 3 in which it is contained (Table 3), and the extent of enzyme closure (Table 2).

       Nucleotide Binding and Dynamic Loops

      By comparing the two UcLK structures, we see that the nucleotide site is largely preformed prior to binding of the ADP. The backbone of interacting residues 117, 119, 279, 281, and 307 from both subunits of the substrate-free UcLK can be superimposed on the nucleotide-bound UcLK-ADP with an r.m.s.d. of 0.2 Å. Around the adenine ring, the side chains of the substrate-free form are already in the bound configuration. There are two other parts of the active site where nucleotide binding in LK and its homologs induces conformational change.

       Large Domain Specificity Loop

      One of the changes induced in UcLK by nucleotide affects a loop known in LpAK to be intrinsically dynamic in the nanosecond regime (
      • Davulcu O.
      • Flynn P.F.
      • Chapman M.S.
      • Skalicky J.J.
      ), UcLK loop 309–317. There are three interactions of the protein backbone with the nucleotide. Hydrogen bonds form between the ribose O2′ and O3′ hydroxyls and the carbonyls of UcLK Thr309 and Ser310. The largest changes to the loop backbone (1.7 Å) are near an interaction between the peptide nitrogen of Glu312 and the α-phosphate O2. Together the interactions are associated with a near-rigid 7° rotation of the loop between hinge points at Thr309 and Val317. Internally, within the hinged region, the local conformation of the nucleotide-bound annelid UcLK loop corresponds well with those of arthropod LpAK-TSAC and rabbit OcCKMM-ADP, i.e. conserved between species and binary versus transition state complexes.
      The region between the hinge points corresponds to a part that was missing or disordered in prior substrate-free structures. In different structures, the density has been reported as weak or has been modeled with high atomic B-factors (>90 Å2) or with large local differences (up to 13 Å) between otherwise symmetric subunits (
      • Fritz-Wolf K.
      • Schnyder T.
      • Wallimann T.
      • Kabsch W.
      ,
      • Wu X.
      • Ye S.
      • Guo S.
      • Yan W.
      • Bartlam M.
      • Rao Z.
      ). These are indicators that the loop is highly dynamic and/or challenging to model. In some structures, the loop points away from the active site in several configurations with active site residues up to 18 Å from their well resolved transition state locations. In one of these structures, human brain CK (HsCKBB), the loop contains an α-helix (
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ), but NMR chemical shift shows the LpAK solution structure to be nonhelical (
      • Davulcu O.
      • Flynn P.F.
      • Chapman M.S.
      • Skalicky J.J.
      ). Thus, locally the substrate-free loop structure is not conserved and/or is influenced by the crystal environment. Noting that the loop restructuring involves motions (up to 18 Å) larger than typical for the fast (ns-ps) intrinsic dynamics as characterized by NMR in LpAK (
      • Niu X.
      • Brüschweiler-Li L.
      • Davulcu O.
      • Skalicky J.J.
      • Brüschweiler R.
      • Chapman M.S.
      ), it is likely that many of the reported loop structures are off the reaction pathway.
      With the new apo-form lombricine kinase structure, the loop remains in the active site, free from external (crystal packing) interactions. The difference in configuration of this loop between the nucleotide-bound and substrate-free form is much more modest in LK. The largest backbone changes are 1.7 Å with a 7° hinged quasi-rigid rotation, as described above. This type of substrate-associated hinged motion, inferred from the LK structures, is of a magnitude that would normally be considered commensurate with the ns-ps motions observed by NMR for the substrate-free state of LpAK (
      • Davulcu O.
      • Flynn P.F.
      • Chapman M.S.
      • Skalicky J.J.
      ). All phosphagen kinases have a single or double glycine aligned with UcLK residues 310–311, and most have a single or double glycine at the other end of the loop (UcLK residues 318–319), so a highly dynamic/disordered structure is to be expected. LK, like GK and vertebrate mitochondrial CKs, lack the glycines at positions 318–319, so perhaps slightly reduced dynamics explain why the loop can be seen in the substrate-free LK structure. It is plausible that the limited induced loop movement seen in UcLK reflects the typical reaction cycle motion among the phosphagen kinases. Despite the different structures observed, the loop is highly conserved and identical in sequence between LK and CKs, except for UcLK residues 312–313. The equivalent of residue 312, valine in all of the CKs, but glutamate in LK, GK, and AK forms a specificity pocket that discriminates methyl-guanidino substrates (like creatine) from the other phosphagen substrates (
      • Azzi A.
      • Clark S.A.
      • Ellington W.R.
      • Chapman M.S.
      ,
      • Novak W.R.
      • Wang P.F.
      • McLeish M.J.
      • Kenyon G.L.
      • Babbitt P.C.
      ,
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ). Thus, the loop motions, on which UcLK sheds light, are likely critical to function.

       Ribose-binding Histidine

      Near the ribose, we see, by comparing the UcLK-ADP structure with the substrate-free structure, that there is a ∼1 Å ADP-associated backbone motion. At its center is UcLK-ADP His178, where the χ1 rotates 120° relative to all of the substrate-free states for the side chain to hydrogen bond with the ADP O3′. The changes on substrate binding are much more modest than seen in other phosphagen kinases.
      In other phosphagen kinases, the conformations of this region fall mostly into two clusters (Fig. 3, A and B) when the backbone atoms of other (non-moving) nucleotide site residues are superimposed. All of the substrate-free structures are clustered loosely with the Cα of the histidine dispersed over a 2.2 Å range. The backbone structures are very similar, as are the side chain conformers, so most of the variation is rigid-group displacement. One exception is the side chain of the histidine in which there is much variation.
      The structures of the transition state forms are more tightly clustered, likely because of a new hydrogen bond formed between the histidine side chain and the O2′ of the substrate ribose (Fig. 3A). These structures are distinct from the substrate-free ones, with a mostly rigid 4+ Å displacement of the histidine backbone from the consensus substrate-free structure, although there are also minor torsion angle changes. The same histidine location is also adopted in 10 of 12 subunits from structures in the Protein Data Bank that are binary complexes.
      Our UcLK-ADP structure becomes the third and most egregious exception, with a ∼1 Å backbone movement near His178 relative to the substrate-free position, away from the consensus transition state location. Here, it is 5.2 Å from the presumed transition state location and at the most distal end of the range of substrate-free CK structures. The other two structures with relatively modest nucleotide-induced changes are the A subunit of rabbit muscle CK (OcCKMM; Protein Data Bank ID 1U6R) and the D subunit of human mitochondrial CK (HsCKmit; Protein Data Bank ID 2GL6). Both have intermediate structures, remaining closer to the substrate-free form but, in contrast to annelid UcLK, moving significantly toward the transition state form.
      Stepping back, we see that UcLK His178 is within the 14-residue subdomain 3 that in other phosphagen kinases undergoes a quasi-rigid rotation/translation of ∼30°/4 Å between the substrate-free and transition states (
      • Niu X.
      • Brüschweiler-Li L.
      • Davulcu O.
      • Skalicky J.J.
      • Brüschweiler R.
      • Chapman M.S.
      ) (Table 3 and Fig. 3). The sequence within subdomain 3 is ∼45% identical between LK and CKs, and the secondary structure, including a short helix, is fully conserved. His178 is within a 5-residue segment that is completely conserved in sequence throughout the phosphagen kinases. It is very clear that there are two clusters of conformations for subdomain 3, corresponding to the two dominant positions of the histidine (Fig. 3, A and B). Indeed, the interaction with the ADP O2′ is only possible with the 4 Å movement of the histidine, which requires that the helix immediately preceding it is similarly translated.
      The binary nucleotide-CK complexes have orientations for subdomain 3 that are intermediate between the substrate-free and transition state forms. Two structures are shown in Fig. 3B to represent the most common structure for nucleotide complexes, with the histidine in the transition state form, hydrogen-bonded to the ribose O2′. All of the binary CK complexes have subdomain structures that are intermediate, but most are close to the transition state form. It is likely that the presence of a phosphagen substrate (Fig. 3B, bottom left) also affects subdomain 3; it is close to the helix that precedes the hydrogen-bonding histidine. We see that although subdomain 3 depends mostly upon the nucleotide, it is only in complexes with both substrates that the full motion is observed.
      The two CK binary complexes in which histidines are not hydrogen-bonded have subdomain 3 orientations that are in the “no man's land” between the transition state and substrate-free forms. (It is clear that there are two orange structures in Fig. 3B that fall between the substrate-free and transition-state-like clusters.) One of these is the A subunit from rabbit muscle OcCKMM, and the other is the D subunit of the eight-subunit HsCKmit, both of which were noted earlier as having intermediate overall conformations.
      Adding now the new annelid UcLK-ADP structure, subdomain 3 is in a fully open form, apparently stabilized by a His178-to-ribose O3′ hydrogen bond, which is thus far unique to this structure (Fig. 3B blue). One might imagine that absent this hydrogen bond, the presence of ADP destabilizes the open form, favoring a more closed conformation. The UcLK-ADP structure completes the spectrum and makes it clear that there is not a uniform nearly closed structure for nucleotide complexes, but there is a near continuum of states with a few examples of “half-closed” and fully open ADP complexes as well. Subdomain 3 orientation as well as overall enzyme closure, it now appears, is correlated to whether the histidine has an O2′ interaction, no hydrogen bond, or an O3′ interaction with the ribose. His178 and its equivalents appear to be key modulators of wider conformational changes.
      Mutational data for the CK equivalent histidine offer some support for its proposed role, although the data are not unequivocal. Two studies have characterized mutations at each of the conserved histidines (
      • Chen L.H.
      • Borders Jr., C.L.
      • Vasquez J.R.
      • Kenyon G.L.
      ,
      • Forstner M.
      • Muller A.
      • Stolz M.
      • Wallimann T.
      ). None of the histidines was essential for catalysis, but mutation of this histidine completely ablated substrate synergy. Synergy is the increased affinity for the second substrate once the first has bound, an effect long assumed to depend on coordinated conformational changes impacting both binding sites (
      • Maggio E.T.
      • Kenyon G.L.
      ). Thus, ablation of synergy would be consistent with the equivalents of His178 helping to mediate the broad conformational changes that link partial substrate binding at the two sites. Interpretation of the mutational data should be tempered by the observation that at least partial loss of synergy was observed in mutants of several conserved histidines (
      • Chen L.H.
      • Borders Jr., C.L.
      • Vasquez J.R.
      • Kenyon G.L.
      ,
      • Forstner M.
      • Muller A.
      • Stolz M.
      • Wallimann T.
      ).
      The distribution of subdomain 3 orientations suggests that nucleotide binding is a necessary but insufficient determinant of subdomain 3 configuration. The conformational change is only fully completed when both substrates are bound. Furthermore, the extent of subdomain 3 reorientation upon nucleotide binding appears finely balanced. Although the prevalence of closed form, ADP-bound crystal structures and solution scattering (
      • Dumas C.
      • Janin J.
      ,
      • Forstner M.
      • Kriechbaum M.
      • Laggner P.
      • Wallimann T.
      ) suggests that closed forms might generally be favored slightly in the presence of ADP, the three intermediate/open form exceptions suggest that the energetics are finely balanced. NMR relaxation dispersion has shown that subdomain 3 exhibits millisecond exchange dynamics in the substrate-free state of LpAK (
      • Davulcu O.
      • Flynn P.F.
      • Chapman M.S.
      • Skalicky J.J.
      ). If such dynamics extend to the nucleotide complex, then the structural variants that have been observed might be reflections of a dynamic equilibrium between multiple conformers in each enzyme. In this case it might be an oversimplification to consider the histidine to be a rigid and/or deterministic link between enzyme closure and nucleotide binding. In a statistical mechanical framework, the histidine interaction could offer a stochastic bias that alters a fine equilibrium between multiple states (
      • Boehr D.D.
      • Dyson H.J.
      • Wright P.E.
      ,
      • Boehr D.D.
      • Wright P.E.
      ,
      • Ma B.
      • Kumar S.
      • Tsai C.J.
      • Nussinov R.
      ).

       Prediction of the LK Transition State Form and Substrate Specificity

      For insights into phosphagen substrate specificity, we turn to a computer prediction of the transition state structure of UcLK adapted from the experimental structure of the UcLK-ADP binary complex using homology to other phosphagen kinase structures. Although the details of such a prediction should not be taken too literally, there are gross features that allow rationalization of the specificities of LKs, CKs, and AKs for their respective substrates and for the broader specificity of LK. In the building of the UcLK-TSA model, the protein structure was first adapted to the TSA form, and then lombricine was added prior to further simulated annealing and energy optimization.
      Sequence alignment has suggested that a primary determinant of phosphagen specificity is the N-terminal domain specificity loop, the length of which is roughly inversely related to the size of the phosphagen substrate (
      • Suzuki T.
      • Kamidochi M.
      • Inoue N.
      • Kawamichi H.
      • Yazawa Y.
      • Furukohri T.
      • Ellington W.R.
      ). Lombricine is considerably larger than the creatine, glycocyamine, and arginine substrates that have previously been visualized in the active sites of their respective phosphagen kinases (
      • Lim K.
      • Pullalarevu S.
      • Surabian K.T.
      • Howard A.
      • Suzuki T.
      • Moult J.
      • Herzberg O.
      ,
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ,
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ,
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ,
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ,
      • Wu X.
      • Ye S.
      • Guo S.
      • Yan W.
      • Bartlam M.
      • Rao Z.
      ). However, lombricine can be placed analogously in the LK active site; when energy-optimized, many of the same parts of the specificity loop interact with the substrate, although the substrate atoms involved have to differ. A low energy structure places the lombricine phosphate in what would be the carboxylate subsite in a CK or AK, coordinated similarly with hydrogen bonds from backbone amides. In CK, there is a single H-bond from TcCK Val72. In LpAK the corresponding valine and its two glycine neighbors (64GVG66) have backbone amide H-bonds to the carboxylate oxygens of the larger arginine substrate (Table 4). The corresponding three UcLK residues interact with lombricine, Cys58, and Gly60 with the carboxylate, but Ile59 coordinates a phosphate oxygen (Fig. 4). The changes in coordination come about because the phosphate is intermediate in position between the carboxylate locations in CK and AK, whereas the carboxylate and phosphate of lombricine straddle ∼1.5 Å either side of the AK carboxylate position (Fig. 4). The altered coordination is achieved with, at most, subtle changes in the backbone (the largest difference with LpAK is 0.5 Å at Gly60).
      TABLE 4Sequence alignment for the N-domain specificity loop for representative creatine, arginine, and lombricine kinases
      Figure thumbnail gr4
      FIGURE 4Lombricine binding site. The stereo pairs show the transition state UcLK structure (A) obtained through comparative modeling, adapting the UcLK-ADP structure with restraints to AK, CK, and GK TSA structures (
      • Lim K.
      • Pullalarevu S.
      • Surabian K.T.
      • Howard A.
      • Suzuki T.
      • Moult J.
      • Herzberg O.
      ,
      • Yousef M.S.
      • Fabiola F.
      • Gattis J.L.
      • Somasundaram T.
      • Chapman M.S.
      ,
      • Lahiri S.D.
      • Wang P.F.
      • Babbitt P.C.
      • McLeish M.J.
      • Kenyon G.L.
      • Allen K.N.
      ,
      • Bong S.M.
      • Moon J.H.
      • Nam K.H.
      • Lee K.S.
      • Chi Y.M.
      • Hwang K.Y.
      ,
      • Ohren J.F.
      • Kundracik M.L.
      • Borders Jr., C.L.
      • Edmiston P.
      • Viola R.E.
      ,
      • Wu X.
      • Ye S.
      • Guo S.
      • Yan W.
      • Bartlam M.
      • Rao Z.
      ), adding lombricine, and performing simulated annealing optimization. B, superimposition of lombricine on LpAK after alignment of the protein structures, showing steric overlap of the lombricine amino acid moiety (top) with the molecular surface of LpAK. By comparison, the pocket in UcLK (A) is opened. Corresponding regions of the specificity loop have backbone amide hydrogen bonds with the arginine carboxylate in AK (B) and with both carboxylate and phosphate in LK (A).
      By contrast, larger differences in the active site are needed to accommodate the bulk of the lombricine amino acid moiety, which extends further from the guanidinium than other phosphagens (Fig. 1). The Asn57 Cα is displaced 2 Å relative to LpAK Ser63 to enlarge the binding pocket and allow for a side chain H-bond between Oδ1 and the substrate nitrogen. Differences of ∼2 Å extend back to the start of the specificity loop (Asn54) and might be stabilized by an edge-to-face (herringbone) π-stacking interaction unique to LK, among Phe55, His17, and Tyr14. Asn57 is the site of the CK specificity loop insertion (Table 4), and the presence of Gly56 (Asp in AK and Gly in CK) may give the flexibility needed to adopt different configurations. Lombricine in an AK active site would also be blocked by a steric overlap between the side chain of Phe194 and the substrate amino group (Fig. 4). Overlap of the corresponding UcLK His187 is avoided with a 2 Å shift, accommodated by Gly186 allowing an otherwise disallowed backbone configuration.
      Taurocyamine is a bond shorter than lombricine, but when its complex with UcLK is similarly optimized, its sulfate nearly superimposes on the phosphate of lombricine. The backbone interaction of Ile59 is preserved, and two more are added from Gly60 and Leu61. Again, backbone interactions from the specificity loop form an anion-binding subsite.
      From such comparative structural modeling, differences in LK phosphagen specificity relative to its homologs can be rationalized. Throughout the phosphagen kinases, the tightest interactions are with the reactive guanidinium. In creatine kinases, the N-domain specificity loop extends over the substrate; but all phosphagen kinases display backbone amide interactions with the anionic groups of the phosphagens, analogous interactions even though the nature of the anionic groups differs. Interactions with additional groups on the phosphagen are minimal. Thus, providing that there is space, the presence of additional atoms in lombricine versus taurocyamine, or the d- or l-isomer forms of lombricine may not be critical. Structural differences are required near Asn54 and His187 to accommodate the additional atoms of lombricine. This appears to be achieved by the presence of strategically placed glycines allowing for changed backbone configuration. Increased backbone flexibility might lead to further loss of specificity, which might therefore be coupled to the evolutionary adaptation of phosphagen kinases to larger substrates.

      Acknowledgments

      Diffraction data were collected at the Cornell High Energy Synchrotron Source and its MacCHESS resource, supported by the National Science Foundation (DMR 0225180) and the National Institutes of Health (RR-01646) and also at the Southeast Regional Collaborative Access Team (SER-CAT) 22-BM beamline at the Advanced Photon Source, supported by the U. S. Department of Energy (W-31-109-Eng-38).

      REFERENCES

        • Schlattner U.
        • Forstner M.
        • Eder M.
        • Stachowiak O.
        • Fritz-Wolf K.
        • Wallimann T.
        Mol. Cell. Biochem. 1998; 184: 125-140
        • Ellington W.R.
        Annu. Rev. Physiol. 2001; 63: 289-325
        • Tombes R.M.
        • Shapiro B.M.
        Cell. 1985; 41: 325-334
        • Wallimann T.
        • Wyss M.
        • Brdiczka D.
        • Nicolay K.
        • Eppenberger H.M.
        Biochem. J. 1992; 281: 21-40
        • Ellington W.R.
        • Kinsey S.T.
        Biol. Bull. 1998; 195: 264-272
        • Suzuki T.
        • Uda K.
        • Adachi M.
        • Sanada H.
        • Tanaka K.
        • Mizuta C.
        • Ishida K.
        • Ellington W.R.
        Comp. Biochem. Physiol. B Biochem. Mol. Biol. 2009; 152: 60-66
        • Fritz-Wolf K.
        • Schnyder T.
        • Wallimann T.
        • Kabsch W.
        Nature. 1996; 381: 341-345
        • Zhou G.
        • Somasundaram T.
        • Blanc E.
        • Parthasarathy G.
        • Ellington W.R.
        • Chapman M.S.
        Proc. Natl. Acad. Sci. U.S.A. 1998; 95: 8449-8454
        • Furter R.
        • Furter-Graves E.M.
        • Wallimann T.
        Biochemistry. 1993; 32: 7022-7029
        • Gattis J.L.
        • Ruben E.
        • Fenley M.O.
        • Ellington W.R.
        • Chapman M.S.
        Biochemistry. 2004; 43: 8680-8689
        • Ellington W.R.
        • Bush J.
        Biochem. Biophys. Res. Commun. 2002; 291: 939-944
        • Pruett P.S.
        • Azzi A.
        • Clark S.A.
        • Yousef M.S.
        • Gattis J.L.
        • Somasundaram T.
        • Ellington W.R.
        • Chapman M.S.
        J. Biol. Chem. 2003; 29: 26952-26957
        • Tanaka K.
        • Suzuki T.
        FEBS Lett. 2004; 573: 78-82
        • Suzuki T.
        • Fukuta H.
        • Nagato H.
        • Umekawa M.
        J. Biol. Chem. 2000; 275: 23884-23890
        • Azzi A.
        • Clark S.A.
        • Ellington W.R.
        • Chapman M.S.
        Protein Sci. 2004; 13: 575-585
        • Novak W.R.
        • Wang P.F.
        • McLeish M.J.
        • Kenyon G.L.
        • Babbitt P.C.
        Biochemistry. 2004; 43: 13766-13774
        • Jourden M.J.
        • Clarke C.N.
        • Palmer A.K.
        • Barth E.J.
        • Prada R.C.
        • Hale R.N.
        • Fraga D.
        • Snider M.J.
        • Edmiston P.L.
        Biochim. Biophys. Acta. 2007; 1774: 1519-1527
        • Suzuki T.
        • Kamidochi M.
        • Inoue N.
        • Kawamichi H.
        • Yazawa Y.
        • Furukohri T.
        • Ellington W.R.
        Biochem. J. 1999; 340: 671-675
        • Lim K.
        • Pullalarevu S.
        • Surabian K.T.
        • Howard A.
        • Suzuki T.
        • Moult J.
        • Herzberg O.
        Biochemistry. 2010; 49: 2031-2041
        • Yousef M.S.
        • Fabiola F.
        • Gattis J.L.
        • Somasundaram T.
        • Chapman M.S.
        Acta Crystallogr. D Biol. Crystallogr. 2002; 58: 2009-2017
        • Lahiri S.D.
        • Wang P.F.
        • Babbitt P.C.
        • McLeish M.J.
        • Kenyon G.L.
        • Allen K.N.
        Biochemistry. 2002; 41: 13861-13867
        • Davulcu O.
        • Flynn P.F.
        • Chapman M.S.
        • Skalicky J.J.
        Structure. 2009; 17: 1356-1367
        • Boehr D.D.
        • Dyson H.J.
        • Wright P.E.
        Chem. Rev. 2006; 106: 3055-3079
        • Boehr D.D.
        • Wright P.E.
        Science. 2008; 320: 1429-1430
        • Ma B.
        • Kumar S.
        • Tsai C.J.
        • Nussinov R.
        Protein Eng. 1999; 12: 713-720
        • Hansen D.E.
        • Knowles J.R.
        J. Biol. Chem. 1981; 256: 5967-5969
        • Murali N.
        • Jarori G.K.
        • Landy S.B.
        • Rao B.D.
        Biochemistry. 1993; 21: 12941-12948
        • Murali N.
        • Jarori G.K.
        • Rao B.D.
        Biochemistry. 1994; 33: 14227-14236
        • Blethen S.L.
        Arch. Biochem. Biophys. 1972; 149: 244-251
        • Rao B.D.
        • Buttlaire D.H.
        • Cohn M.
        J. Biol. Chem. 1976; 251: 6981-6986
        • Rhee S.
        • Parris K.D.
        • Hyde C.C.
        • Ahmed S.A.
        • Miles E.W.
        • Davies D.R.
        Biochemistry. 1997; 36: 7664-7680
        • Gerstein M.
        • Chothia C.
        J. Mol. Biol. 1991; 220: 133-149
        • Boyer R.
        Concepts in Biochemistry. Brooks/Cole, Pacific Grove, CA1998
        • Fernandez P.
        • Haouz A.
        • Pereira C.A.
        • Aguilar C.
        • Alzari P.M.
        Proteins. 2007; 69: 209-212
        • Yousef M.S.
        • Clark S.A.
        • Pruett P.K.
        • Somasundaram T.
        • Ellington W.R.
        • Chapman M.S.
        Protein Sci. 2003; 12: 103-111
        • Niu X.
        • Brüschweiler-Li L.
        • Davulcu O.
        • Skalicky J.J.
        • Brüschweiler R.
        • Chapman M.S.
        J. Mol. Biol. 2011; 405: 479-496
        • Hosfield D.
        • Palan J.
        • Hilgers M.
        • Scheibe D.
        • McRee D.E.
        • Stevens R.C.
        J. Struct. Biol. 2003; 142: 207-217
        • Szebenyi D.M.
        • Arvai A.
        • Ealick S.
        • Laiuppa J.M.
        • Nielsen C.
        J. Synchrotron Radiat. 1997; 4: 128-135
        • Otwinowski Z.
        • Minor W.
        Methods Enzymol. 1997; 276: 307-326
        • Eder M.
        • Fritz-Wolf K.
        • Kabsch W.
        • Wallimann T.
        • Schlattner U.
        Proteins. 2000; 39: 216-225
        • Chenna R.
        • Sugawara H.
        • Koike T.
        • Lopez R.
        • Gibson T.J.
        • Higgins D.G.
        • Thompson J.D.
        Nucleic. Acids Res. 2003; 31: 3497-3500
        • Kleywegt G.J.
        CCP4/ESF-EACBM Newsletter on Protein Crystallography. Vol. 32. Daresbury Laboratory, Warrington, UK1996: 32-36
        • Sali A.
        • Blundell T.L.
        J. Mol. Biol. 1993; 234: 779-815
        • Fiser A.
        • Do R.K.
        • Sali A.
        Protein Sci. 2000; 9: 1753-1773
        • Grosse-Kunstleve R.W.
        • Adams P.D.
        Acta Crystallogr. D Biol. Crystallogr. 2001; 57: 1390-1396
        • McCoy A.J.
        • Grosse-Kunstleve R.W.
        • Storoni L.C.
        • Read R.J.
        Acta Crystallogr. D Biol. Crystallogr. 2005; 61: 458-464
        • Brünger A.T.
        • Adams P.D.
        • Clore G.M.
        • DeLano W.L.
        • Gros P.
        • Grosse-Kunstleve R.W.
        • Jiang J.S.
        • Kuszewski J.
        • Nilges M.
        • Pannu N.S.
        • Read R.J.
        • Rice L.M.
        • Simonson T.
        • Warren G.L.
        Acta Crystallogr. D Biol. Crystallogr. 1998; 54: 905-921
        • Navaza J.
        • Saludjian P.
        Methods Enzymol. 1997; 277: 581-594
        • Jones T.A.
        Acta Crystallogr. D Biol. Crystallogr. 2004; 60: 2115-2125
        • Murshudov G.N.
        • Vagin A.A.
        • Dodson E.J.
        Acta Crystallogr. D Biol. Crystallogr. 1997; 53: 240-255
        • Adams P.D.
        • Grosse-Kunstleve R.W.
        • Hung L.W.
        • Ioerger T.R.
        • McCoy A.J.
        • Moriarty N.W.
        • Read R.J.
        • Sacchettini J.C.
        • Sauter N.K.
        • Terwilliger T.C.
        Acta Crystallogr. D Biol. Crystallogr. 2002; 58: 1948-1954
        • Emsley P.
        • Cowtan K.
        Acta Crystallogr. D Biol. Crystallogr. 2004; 60: 2126-2132
        • Kirillova O.
        • Minor W.
        Bioinformatics. 2006; 22: 1660-1661
        • Howlin B.
        • Moss D.S.
        • Harris G.W.
        Acta Crystallogr. A. 1989; 45: 851-861
        • Winn M.D.
        • Isupov M.N.
        • Murshudov G.N.
        Acta Crystallogr. D Biol. Crystallogr. 2001; 57: 122-133
        • Krissinel E.
        • Henrick K.
        Acta Crystallogr. D Biol. Crystallogr. 2004; 60: 2256-2268
        • Hayward S.
        • Berendsen H.J.
        Proteins. 1998; 30: 144-154
        • Adams P.D.
        • Afonine P.V.
        • Bunkoczi G.
        • Chen V.B.
        • Davis I.W.
        • Echols N.
        • Headd J.J.
        • Hung L.W.
        • Kapral G.J.
        • Grosse-Kunstleve R.W.
        • McCoy A.J.
        • Moriarty N.W.
        • Oeffner R.
        • Read R.J.
        • Richardson D.C.
        • Richardson J.S.
        • Terwilliger T.C.
        • Zwart P.H.
        Acta Crystallogr. D Biol. Crystallogr. 2010; 66: 213-221
        • Eswar N.
        • Eramian D.
        • Webb B.
        • Shen M.
        • Sali A.
        Baxevanis A.D. Stein L.D. Stormo G.D. Yates III, J.R. Current Protocols in Bioinformatics. John Wiley & Sons, Inc., New York2006: 5.6.1-5.6.30
        • Bong S.M.
        • Moon J.H.
        • Nam K.H.
        • Lee K.S.
        • Chi Y.M.
        • Hwang K.Y.
        FEBS Lett. 2008; 582: 3959-3965
        • Ohren J.F.
        • Kundracik M.L.
        • Borders Jr., C.L.
        • Edmiston P.
        • Viola R.E.
        Acta Crystallogr. D Biol. Crystallogr. 2007; 63: 381-389
        • Wu X.
        • Ye S.
        • Guo S.
        • Yan W.
        • Bartlam M.
        • Rao Z.
        FASEB J. 2010; 24: 242-252
        • Van Der Spoel D.
        • Lindahl E.
        • Hess B.
        • Groenhof G.
        • Mark A.E.
        • Berendsen H.J.
        J. Comput. Chem. 2005; 26: 1701-1718
        • Oostenbrink C.
        • Villa A.
        • Mark A.E.
        • Van Gunsteren W.F.
        J. Comput. Chem. 2004; 25: 1656-1676
        • Dumas C.
        • Janin J.
        FEBS Lett. 1983; 153: 128-130
        • Forstner M.
        • Kriechbaum M.
        • Laggner P.
        • Wallimann T.
        Biophys J. 1998; 75: 1016-1023
        • Rao J.K.
        • Bujacz G.
        • Wlodawer A.
        FEBS Lett. 1998; 439: 133-137
        • Chapman M.S.
        Vial C. Creatine Kinase. NovaScience, New York2006: 69-94
        • Chen L.H.
        • Borders Jr., C.L.
        • Vasquez J.R.
        • Kenyon G.L.
        Biochemistry. 1996; 35: 7895-7902
        • Forstner M.
        • Muller A.
        • Stolz M.
        • Wallimann T.
        Protein Sci. 1997; 6: 331-339
        • Maggio E.T.
        • Kenyon G.L.
        J. Biol. Chem. 1977; 252: 1202-1207
        • Berman H.M.
        • Westbrook J.
        • Feng Z.
        • Gilliland G.
        • Bhat T.N.
        • Weissig H.
        • Shindyalov I.N.
        • Bourne P.E.
        Nucleic Acids Res. 2000; 28: 235-242
        • Read R.J.
        Acta Crystallogr. A. 1986; 42: 140-149
        • Tisi D.D.
        • Bax B.B.
        • Loew A.A.
        Acta Crystallogr. D Biol. Crystallogr. 2001; 57: 187-193
        • Shen Y.
        • Tang L.
        • Zhou H.
        • Lin Z.
        Acta Crystallogr. D Biol. Crystallogr. 2001; 57: 1196-1200
        • Eder M.
        • Schlattner U.
        • Becker A.
        • Wallimann T.
        • Kabsch W.
        • Fritz-Wolf K.
        Protein Sci. 1999; 8: 2258-2269