Structural basis for regulation of bifunctional roles in replication initiator protein
- Akira Nakamura*,
- Chieko Wada†, and
- Kunio Miki*,‡,§
- *Department of Chemistry, Graduate School of Science, Kyoto University, Sakyo-ku, Kyoto 606-8502, Japan;
- †Graduate School of Biostudies, Kyoto University, Sakyo-ku, Kyoto 606-8315, Japan;
- ‡RIKEN SPring-8 Center at Harima Institute, Koto 1-1-1, Sayo, Hyogo 679-5148, Japan
-
Edited by Johann Deisenhofer, University of Texas Southwestern Medical Center, Dallas, TX, and approved September 28, 2007 (received for review June 15, 2007)
Abstract
DNA replication initiator protein RepE stringently regulates F plasmid replication by its two distinct molecular association states. A predominant dimer functions as an autogenous repressor, whereas monomers act as replication initiators, and the dimer requires actions of the DnaK molecular chaperone system for monomerization. The structure of the monomeric form is known, whereas the dimeric structure and structural details of the dimer-to-monomer conversion have been unclear. Here we present the crystal structure of the RepE dimer in complex with the repE operator DNA. The dimerization interface is mainly formed by intermolecular β-sheets with several key interactions of charged residues. The conformations of the internal N- and C-terminal domains are conserved between the dimer and monomer, whereas the relative domain orientations are strikingly different, allowing for an efficient oligomeric transition of dual-functional RepE. This domain relocation accompanies secondary structural changes in the linker connecting the two domains, and the linker is included in plausible DnaK/DnaJ-binding regions. These findings suggest an activation mechanism for F plasmid replication by RepE monomerization, which is induced and mediated by the DnaK system.
Initiation of DNA replication in prokaryotic and eukaryotic cells requires a cis-element (origin) and a trans-factor (initiator protein). Strictly controlled plasmid replication in Gram-negative bacteria is a simple model to investigate initiation of DNA replication and has been studied for several plasmids (1). A plasmid replication origin and a plasmid-encoded initiator protein are common elements, and plasmid replication is modulated by the initiator, called Rep. The F plasmid Rep protein, RepE (M r 29,000), plays an essential role in stringent regulation of the plasmid copy number in an Escherichia coli cell, so that the F plasmid is stably maintained at one to two copies per host chromosome (2). RepE exhibits two functions correlated with its oligomeric states, monomeric and dimeric (3), and the DnaK molecular chaperone system is essential for F plasmid replication (4, 5), mediating the conversion of the prevailing RepE dimers to monomers (6 and C.W., A.N., G. Kobayashi, F. Matsunaga, M. Ueta, and K.M., unpublished data) (Fig. 1 A). The RepE monomer works as an initiator that binds to an iteron in the replication origin, ori2. By the effects of RepE binding to four tandem repeats of the iteron, a significant bending on the DNA chain is induced with the assistance of a histone-like protein HU, and then localized DNA melting occurs in the 13-mer and the AT-rich regions, followed by the formation of a prepriming complex for replication initiation (7). In contrast, the RepE dimer functions as an autogenous transcriptional repressor that binds to the repE promoter/operator region located just upstream of the repE gene. The repE operator and the iteron share an 8-bp sequence (underlined in Fig. 1 A), and two common 8-bp sequences repeat inversely in the operator (Fig. 1 B). Negative regulation of replication is also given by an interaction between the ori2 and incC iterons mediated by RepE proteins (8), in which a RepE-bridged DNA loop is formed (6). Both initiator and repressor activities, including the interaction between RepE and DNA, have been discussed using several mutants (3, 9–14), and the first crystal structure of the prokaryotic replication initiator RepE54 [a stable monomeric mutant R118P of RepE isolated from a dnaJ-defective host (10)] in complex with an iteron DNA (RepE54–iteron) has been revealed (Fig. 1 D) (15). RepE consists of two domains, each containing a winged helix–turn–helix motif (16). The N-terminal domain [also termed WH1 (17)] is involved in the dimerization of RepE (18) and an auxiliary binding to DNA, whereas the C-terminal domain (WH2) is related to a sequence-specific binding to the common 8 bp of the iteron. Further, the WH1 domain structure of the pPS10 RepA protein is already known (19), and the similarity between Rep proteins and initiators of archaea (Cdc6p) and eukaryote (Orc4p) has been proposed because they have partially similar structures (19, 20) and essentially common activation machinery (17, 20), in which Hsp70 (DnaK) chaperone activates the initiators by facilitating dissociation of the replication-inert oligomers and/or conformational changes in monomeric initiators. Thus, Rep activation can be a fundamental process for the regulation of DNA replication. However, structural details of the chaperone-mediated monomerization remain unclear because of the lack of a whole dimeric structure of Rep. In the hypothetical dimeric model based on the RepE54–iteron structure, the two N-terminal domains collide with each other in one central major groove of the operator DNA (15). This model suggests that a drastic structural change is necessary for dimerization. To provide direct evidence for the structural alteration, we have determined the crystal structure of the RepE dimer in complex with the repE operator DNA (RepE–operator). Together with the RepE54–iteron structure, we here present the structural basis for regulation of the dual-functional roles in RepE. Its conformational changes allow us to suggest an activation mechanism of the initiator induced by the molecular chaperones DnaK, DnaJ, and GrpE.
Structures and functions of RepE. (A) Schematic representation of functions of RepE initiator protein of mini-F plasmid. (B) DNA sequence used in crystallographic analysis. Arrows indicate the common 8-bp sequences between the iteron and repE operator. (C and D) Two orthogonal views of RepE–operator (C) and RepE–iteron (D; PDB ID code 1REP) complexes. Each dimer protomer is colored green (molecule A) or yellow (molecule B). The DNA models are omitted in the lower panels. The secondary structural elements of the RepE dimer are designated according to those of the RepE54 structure (15).
Results and Discussion
Overall Structure of RepE–Operator.
The full-length replication initiator proteins such as RepE have never been subjects of structural study because of their aggregating natures. However, we have attempted to prepare intact RepE protein for x-ray crystallographic analysis, and we have successfully purified wild-type RepE under high-salt conditions, thus obtaining the crystals of its complex with the repE operator DNA (21). The crystal structure of the dimeric RepE bound to the 33-bp DNA containing the repE operator sequence was solved at 3.14 Å resolution (Fig. 1). In the crystal asymmetric unit, there is one protein–DNA complex with dimensions of ≈50 Å × 60 Å × 100 Å (Fig. 1 C). A RepE protomer, like RepE54, is divided into an N-terminal domain (NTD; residues 1–132) with a C-terminal domain (CTD; residues 160–251) and a linker region tethering the two domains (Fig. 1 C and D and Fig. 2). The secondary structural elements (α1–α4′ and β1–β4′) of RepE54 are well maintained in RepE, except for the α1 helix and the linker [Fig. 2 and supporting information (SI) Fig. 7]. The N-terminal 20 residues and the loop regions between α2 and β2, which are also disordered in RepE54, are not included in the present model because only discontinuous electron density maps were observed in these regions. The α4 helix of each NTD is almost linearly lying over the central major groove of the operator without any contacts between protein and DNA (Fig. 1 C), which is one of the key factors involved in solving predicted steric hindrance (15). Although the RepE dimer contacts the operator DNA only by CTD residues, the winged helix–turn–helix (wHTH) motif, a canonical DNA-binding motif, is found in the NTD as well as in the CTD. The DNA presents in a global B-form with a slight bend of ≈20° in the direction opposite that of the RepE54–iteron complex.
Sequence alignment of Rep proteins. Amino acid sequences of Rep proteins, RepE of F plasmid (E. coli) and RepA of pPS10 plasmid (Pseudomonas syringae), are aligned. Identical and homologous residues are highlighted in red backgrounds and in red letters, respectively. The secondary structural elements of RepE (in this work; blue), RepE54 (15), and RepA (19) are also represented based on their crystal structures. The solid and broken lines indicate random coils and disordered regions in the crystal structures, respectively. The R118P mutation site of RepE54 is indicated by a red star. The sequence alignment was carried out by using ClustalW (22), and this figure was generated by ESPript (23) with manual modifications.
The structures of two RepE protomers (designated molecules A and B) in the dimer are nearly identical and are related to each other by a noncrystallographic dyad. The superimposition of these protomers, however, shows a subtle structural plasticity in the wHTH region (α2–β2–α3–α4, β3–β4) of the NTD. Although the protomers are well superimposed by the CTD with a root mean square deviation (rmsd) of 0.56 Å at 87 Cα positions calculated by using LSQKAB (24), the two NTD wHTH regions are placed in slightly different locations at a maximum distance of 10 Å between the two Thr-74 residues (SI Fig. 8B). In addition, the plot of the differences in dihedral angles (ΔΦ + ΔΨ) shows sharp peaks at Leu-31 and Gly-91, excluding the peaks in the loop regions (SI Fig. 8A). Further, Phe-71 (α3) and Arg-37 (α2) form a π-cation interaction to fix these helices in meaningful relative positions. These results indicate that the wHTH area of the NTD has the potential for a rigid-body motion.
RepE–Operator Interaction.
As described above, only the CTD of the RepE dimer is responsible for the DNA binding (Fig. 1 C). Even though several corresponding residues that interact with the iteron DNA in the NTD of the RepE54–iteron structure point toward the major groove located between the two common 8-bp sequences of the operator, the NTD is >5 Å away from the DNA (Fig. 3 A). The RepE dimer recognizes the common 8-bp sequence of the operator in the same manner as the RepE54–iteron (SI Fig. 9). The specific charged residues in RepE (Arg-200, Asp-203, Arg-206, and Arg-207) directly bind to the bases on the DNA, and several residues interact with the phosphate backbones (Fig. 3 B and SI Fig. 9). Because the CTD of one protomer (molecule B) has a higher average B-factor (53.0 Å2) than that of the other (34.2 Å2; molecule A), a poor electron density map is observed between the DNA and the CTD of the protomer B.
RepE–operator interactions. (A) Central region of the complex. Corresponding residues interacting with DNA in the RepE54–iteron complex are shown as yellow and green sticks. The final σA-weighted 2F o–F c electron density map (contoured at 1.2σ) is superimposed on the DNA. (B) Details of the DNA recognition by the CTD of RepE (molecule A). Amino acid residues interacting with the operator DNA are shown as yellow sticks. Hydrogen bonds are shown as dotted lines.
The overview of RepE–operator interaction, especially the relation between the NTD and DNA, is quite different from the pPS10 RepA–operator model based on the partial structure of the dimeric NTD of RepA (19). No interaction is found between the NTD of RepE and DNA, and the two α4 helices are located over the major groove, whereas the NTD of RepA would form interactions with the minor groove by using the corresponding α-helices. The differences are probably attributable to ambiguity of the relative domain orientation of RepA because of the lack of its CTD in the crystal structure and/or to the different properties of the two operators. The repE operator has a 9-bp interval between the two common 8-bp sequences repeated inversely (Fig. 1 B), whereas the repA operator of pPS10 contains an 11-bp interval (25). This difference might lead to divergent interaction forms of Rep proteins bound to the operator of its gene.
RepE Dimerization.
Two RepE protomers assemble into a dimer with a common surface of 2,600 Å2, which is buried by the association of two monomers. The widely distributed hydrogen bonds in the two antiparallel β-sheets composed of the β-strands β2–β4–β3/β2b–β2a from both protomers are main contributors to dimer formation (Fig. 1 C) as observed in the NTD dimer of pPS10 RepA (19). There are 10 intermain-chain hydrogen bonds between β2b of one protomer and β3 of the other protomer along the intermolecular contacts for a distance of 45 Å from Glu-101 to Ser-123 (Fig. 4). Further, the side chains of His-130 (β4) and Tyr-134 (α5) form hydrogen bonds between the two protomers (data not shown in Fig. 4). Additionally, several van der Waals contacts and two polar interaction sites are found in the interface, which enhances the interaction of two monomers. The two Gly-91 and the two Leu-116 residues are located at the center of the dimerization contacts, and they form van der Waals interactions around the noncrystallographic twofold axis. Arg-118, which is the mutation site of RepE54 (10), forms polar interactions with Glu-93 and Ser-111 of the other protomer (Fig. 4 A). On the opposite side of the β-sheet, Glu-110 interacts with Lys-117 of the counterpart as well as Arg-98 of the same protomer (Fig. 4 B). These residues correspond to the chaperone-independent copy-up (9, 10) or to the dimerization-defective (14) mutation sites, except for Arg-98 and Lys-117. Two mutations in which the negatively charged residues are replaced by the positively charged ones, E93K (RepE10) and E110K (RepE26) (9), could result in electrostatic repulsion and destabilization of the dimer interface. In two other representative mutants, S111P (14) and R118P (RepE54), which have markedly increased initiator activities and are stable as monomers, the disruption of the hydrogen bond network would be caused by a conformational change around the introduced proline residue. Therefore, these RepE mutants could not form stable dimers.
RepE dimerization. (A and B) Two opposite side views of the dimerization β-sheet. The coloring scheme is the same as that in Fig. 1 C. Amino acid residues involved in dimerization are represented by sticks. Hydrogen bonds by main-chain atoms are shown as dotted lines. Polar interactions between side-chain atoms are shown as dashed red lines. For clarity, the DNA model was omitted in both A and B, and the α4 helices are not shown in A. (B) The position of the noncrystallographic twofold axis is indicated by a black ellipse.
The specific leucine residues (Leu-24, Leu-31, and Leu-39 of RepE; Figs. 2 and 5 A) that are conserved in plasmid replication initiators were once thought to be involved in dimerization (14) and predicted to be exposed to the dimer interface (15, 26). However, they were located in the internal hydrophobic core of the RepA NTD dimer of pPS10 plasmid as found in RepE54 (Fig. 5 B), and it has been suggested that the rigid V-shaped conformation of the α1–α2 helices containing these leucines helps the correct placement of the α5 helix to stabilize the dimeric form (19). The leucine residues of the RepE dimer adopt configurations similar to those of pPS10 RepA and form strong hydrophobic interactions with residues of the α5 linker connecting the NTD and the CTD (Fig. 5 A). In the ΔN42 mutant lacking N-terminal 42 residues of RepE (14), the defect in the linker stabilization could affect the decrease in dimerization efficiency. Thus, it is also suggested that the NTD leucine residues conserved among plasmid initiator proteins are not implicated in the architecture of the dimer interface but in a scaffold of the linker helix for stabilizing the dimeric structure.
Structural comparison of linker regions. (A and B) Stereo views of hydrophobic interactions around the linker regions of the dimeric form (A) and the monomeric form (B) of RepE. Three helices (α1, α2, and α5) are shown in a ribbon representation, and the residues involved in the hydrophobic interactions are shown as sticks. The conserved leucine residues (Leu-24, Leu-31, and Leu-39) are yellow. This V-shaped helix bundle is anchored by the hydrogen bond (dotted line) between the Lys-36 side chain (blue stick) and Tyr-29 carbonyl oxygen atom (red) only in the dimeric form.
Conformational Changes in Two Forms of RepE.
Comparing the overall structure of RepE (the dimer) with that of RepE54 (the monomer), it is found that the monomer adopts a more compact conformation than the dimer (Fig. 1 C and D). Therefore, the two structures cannot be superimposed. However, once each whole structure is divided into the two domains, the discrete superimpositions of the domain structures between RepE (molecule A) and RepE54 show that the two NTD moieties are similar (Fig. 6 B; rmsd 2.41 Å for 87 Cα atoms) as are the two CTD (Fig. 6 A; rmsd 0.54 Å for 87 Cα atoms). Both association forms have the same internal hydrophobic core constructed by the conserved residues of initiators not only in the CTD but also in the NTD (SI Fig. 10). In addition, their side-chain conformations are remarkably similar. In contrast, with respect to the CTD superposition, the NTD rotates around the α2′ helical axis by ≈40° (Fig. 6 A). These findings demonstrate that the relative orientations of the two domains are different between the dimer and monomer, retaining the conformations of each domain. This domain-based conformational change correlates with secondary structural alterations in the linker region (SI Fig. 11 A and B). The linker consists of the long α-helix (α5) in the dimeric structure, whereas the shorter α5 helix, a loop structure, and a β-strand (β1′) construct it in the monomeric RepE (Figs. 2 and 5). Therefore, the interdomain β-sheet of RepE54, comprising β1 and β1′, is no longer observed in the dimer. Moreover, in contrast to the completely folded α1 helix of the dimer, that of the monomer is partly unfolded. This unwinding is associated with disruption of the V-shaped conformation of α1–α2 anchored by the conserved Lys-36 interacting with the carbonyl oxygen of Tyr-29 (Fig. 5).
Conformational changes between the dimer and monomer. (A and B) Superpositions of the crystal structures of RepE protomer (molecule A; green) and RepE54 (gray). Structural alignments were performed by using LSQKAB, based on the CTD (A) and the NTD (B). The CTD model is not present in B. The kinked residue Ile-140 is indicated by an arrow. (C and D) Interdomain interactions with structural rearrangement. Intramolecular interactions of the dimeric and monomeric structures are focused in C and D, respectively. Residues involved in the interdomain interactions are shown as yellow sticks. Polar interactions are shown as dotted red lines.
These two forms are structurally stabilized by several interactions. The interdomain β-sheet (β1–β1′) is formed in the monomer (Figs. 1 D and 6 B and D). In the dimeric forms, its characteristic long α5 helix is stabilized by hydrophobic interactions with α1–α2 (Fig. 5) and anchored by the interactions at both ends (Tyr-134–His-130′ and Arg-150–Asp-244) (Fig. 6 C). The relative B-factor values of the linker region in the dimeric structure are lower than those of the monomeric structure, which implies that the linker is stable in the dimeric form (SI Fig. 11C). In addition, one of the conserved polar residues Glu-26 (α1; NTD) is related to the interdomain interaction with conserved Arg-164 (α2′; CTD) (Fig. 6 C). This Glu-26 alternatively contacts conserved Gln-171 (α2′) in the monomeric form (Fig. 6 D). Conserved Arg-37 (α2) and Phe-71 (α3) residues form a π-cation interaction in the wHTH of the dimeric NTD (Fig. 6 C and SI Fig. 8), whereas Arg-37 forms an interdomain interaction with the carbonyl oxygen atom of Lys-155 (α1′) in the monomer (Fig. 6 D). Therefore, it is apparent that the polar interaction pairs of RepE are exchanged depending on its oligomeric state associated with domain relocation. These results imply that the conserved residues in plasmid initiator proteins are responsible for the stabilization of both monomeric and dimeric structures.
Implication of the Chaperone Interaction in RepE Dimer-to-Monomer Conversion.
Structural comparisons of the two states in RepE show structural details of the dimer-to-monomer conversion of RepE. The domain-based structural transition and the structural transformation of the linker region are responsible for this conversion. With regard to the conformational switch, it has been shown that the DnaK molecular chaperone system (DnaK, DnaJ, and GrpE) is necessary for replication of the mini-F plasmid (4, 5) and in particular for the conversion from RepE dimers to monomers in vitro, allowing RepE to bind to the iteron (6, 11, and C.W., A.N., G. Kobayashi, F. Matsunaga, M. Ueta, and K.M., unpublished data). Therefore, we expected interaction sites of these chaperone proteins on the RepE structure, and we proposed an initiator activation mechanism of RepE triggered and mediated by the DnaK system. DnaK is known to bind to a hydrophobic region of a substrate, especially to the DnaK-binding motif consisting of a 4- to 5-residue hydrophobic core flanked by 4-residue basic regions (27). DnaJ also interacts with a hydrophobic core of a substrate, including an 8-residue aromatic/hydrophobic region (28). The deletion, peptide competition, and mutant analyses of a homolog protein of RepE, RepA of P1 plasmid, show that DnaJ recognizes residues 180–200 of P1 RepA (29). According to the amino acid sequence alignment between P1 RepA and F RepE (30), the corresponding DnaJ-binding region of RepE is assigned to the linker helix α5 and α1′ (residues 133–155). Moreover, this area contains continuous hydrophobic residues, i.e., the DnaJ-binding motif (28) (135LIPFFIGL142). On the other hand, several possible DnaK recognition sites (27) are found in the proximity of α5, in which the center of the α2 helix has the highest probability of DnaK binding (SI Fig. 12). DnaJ has been known to stimulate DnaK ATPase activity, accompanied by the alteration of DnaK properties (31, 32). Moreover, it has been proposed that DnaJ associates with a polypeptide substrate before ATP-DnaK that exhibits low affinity for substrates and then ATP-DnaK is converted to ADP-DnaK with high affinity for substrates by the cooperation of DnaJ on the same substrate (28, 33). Therefore, it is likely that DnaJ first interacts with α5, followed by DnaK binding to the RepE dimer (SI Fig. 13). Considering the flexibility of the N terminus and of the α2–β2 loop of RepE, DnaK could widely bind around α2 and perturb the hydrophobic network distributed among the helix bundle (α1, α2, and α5). It has been reported that the helical contents of DnaK-bound substrates are reduced (34). Thus, DnaK binding probably causes unfolding at α5 and α1–α2, losing the Lys-36–Tyr-29 anchor contact (Fig. 5). This helical unwinding of α1 as well as of α5 could further weaken the interactions among the helix bundle. The partially unfolded α5 linker, which might be covered with DnaK to prevent a hydrophobic surface from exposure to the solvent during the conversion, may protrude outward from the core (kinked at Ile-140; Figs. 5 B and 6 B) because of the insufficient interaction among the helix bundle. A structural perturbation of α1–α2 by DnaK can be propagated to the whole wHTH region of the NTD, containing α2 and showing the rigid structure indicated in SI Fig. 8. If the wHTH forming the dimer interface moves inward to take on compact conformations, the hydrogen bond network between the protomers is disrupted. Therefore, this conformational change induces dissociation of the RepE dimer. Both relaxation of the linker and dissociation of the dimer decrease the driving force to fix interdomain orientation, leading to the conformational changes of RepE into a form appropriate for binding to the iteron. At the last stage of monomerization, by an action of another component of the DnaK system, GrpE, which functions as a nucleotide exchange factor of DnaK (converting its ADP state to ATP state), DnaK dissociates from RepE and the partly exposed hydrophobic surface of the NTD is simultaneously overlaid by the N-terminal β1 strand of RepE as a lid.
Before and after this conversion of the RepE association states, the core structure of each domain is unchanged, whereas the relative domain orientation undergoes a significant change, and alternative interdomain interactions are formed to stabilize both the oligomeric conformations. This dynamic structural change independent of intradomain rearrangement is an important structural basis for the dual-functional RepE to work efficiently in the two distinct oligomeric states. Further, the linker possesses two roles in the conversion process: one is as a determinant of the relative domain orientation in the dimeric form, and the other is as a structurally flexible hinge involved in the domain relocation. Because these inconsistent properties are accomplished by the structural change at the linker, interactions of the molecular chaperones are indispensable for the oligomeric conversion. The structural information of both the RepE–operator and RepE–iteron will provide valuable insights into how the DnaK system acts on a native protein substrate. The proposed activation mechanism of the RepE initiator might be a prototypic model for the regulation of DNA replication.
Materials and Methods
Crystallization and X-Ray Data Collection.
The RepE–DNA complex was prepared, and the crystal and its diffraction data were obtained as described previously (21). Gold derivative crystals were prepared by soaking crystals in mother liquor containing 0.1 mM KAu(CN)2 for 3–5 days. Iodine derivative crystals were obtained by cocrystallization with the operator DNA substituted with 5′-iodouracil for a thymine base at the T13 and T55 positions (Fig. 1 B). X-ray diffraction data on heavy-atom derivatives were collected at either 100 K or 90 K on beamline BL26B1 or BL44B2 at SPring-8. All diffraction data were processed by using HKL2000 (35) and are summarized in SI Table 1.
Structure Determination.
The molecular replacement was carried out with MOLREP (36) using the RepE54–iteron structure [Protein Data Bank (PDB) ID code 1REP], especially the C-terminal moiety of the complex (161–246 residues of the protein and 1–9 and 35–43 residues of the DNA) as a search model. Two clear solutions were obtained, and solvent flattening and histogram matching with the molecular replacement phases were then performed with the program DM (37) of the CCP4 suite (38). An obvious electron density map of continuous nucleotide pairs resulted as if the two start models were connected, which allowed us to build almost all of the nucleotide chain by using the graphic program O (39). The NTDs, which had not been included in the model, were roughly introduced with the aid of MOLREP by using 32–139 residues of RepE54 as a search model. The incorporated model of RepE–operator was then improved by iterative cycles of manual rebuilding using O and refinement with CNS (40) at 3.14 Å resolution. The sites of gold and iodine atoms were determined by means of isomorphous difference Fourier maps using phase information from a partially refined structure (R work = 35.2%, R free = 42.2%). Subsequent refinement was carried out by using the MLHL target function with the combined phase probability distribution calculated from the three experimental SIR phases and a refinement-ongoing model. Isotropizing based on the anisotropic B-factor correction values of the CNS output and B-factor sharpening (41, 42) (overall B sharp = −60 Å2) were applied to the observed anisotropic amplitudes (21) followed by density modification only for map calculations, whereas unmodified amplitudes from the diffraction data were used in refinement. After application of bulk solvent correction and atomic B-factor refinement using the MLF target function, the final model was given with a crystallographic working R-factor of 26.6% and a free R-factor of 31.3% (SI Table 1). In the Ramachandran plot, 75.3%, 24.0%, and 0.8% of residues are in the most favored, in the additional allowed, and in the generously allowed regions, respectively. All molecular graphics figures were generated with PyMOL (http://pymol.sourceforge.net).
Acknowledgments
We thank the staffs at beamlines BL26B1, BL41XU, and BL44B2, SPring-8, Japan, and beamline NW12, Photon Factory, Japan, for kind assistance with x-ray diffraction data collection. This work was supported by grants-in-aid for scientific research (to C.W. and K.M.) and by a grant from the National Project on Protein Structural and Function Analysis from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (to K.M.).
Footnotes
- §To whom correspondence should be addressed. E-mail: miki{at}kuchem.kyoto-u.ac.jp
-
Author contributions: A.N., C.W., and K.M. designed research; A.N. performed research; A.N., C.W., and K.M. analyzed data; and A.N., C.W., and K.M. wrote the paper.
-
The authors declare no conflict of interest.
-
This article is a PNAS Direct Submission.
-
Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID code 2Z9O).
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0705623104/DC1.
- © 2007 by The National Academy of Sciences of the USA