• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Mar 18, 2003; 100(6): 3137–3142.
Published online Mar 5, 2003. doi:  10.1073/pnas.0530312100
PMCID: PMC152259

The 1.6-Å crystal structure of the class of chaperones represented by Escherichia coli Hsp31 reveals a putative catalytic triad


Heat shock proteins (Hsps) play essential protective roles under stress conditions by preventing the formation of protein aggregates and degrading misfolded proteins. EcHsp31, the yedU (hchA) gene product, is a representative member of a family of chaperones that alleviates protein misfolding by interacting with early unfolding intermediates. The 1.6-Å crystal structure of the EcHsp31 dimer reveals a system of hydrophobic patches, canyons, and grooves, which may stabilize partially unfolded substrate. The presence of a well conserved, yet buried, triad in each two-domain subunit suggests a still unproven hydrolytic function of the protein. A flexible extended linker between the A and P domains may play a role in conformational flexibility and substrate binding. The α-β sandwich of the EcHsp31 monomer shows structural similarity to PhPI, a protease belonging to the DJ-1 superfamily. The structure-guided sequence alignment indicates that Hsp31 homologs can be divided in three classes based on variations in the P domain that dramatically affect both oligomerization and catalytic triad formation.

Keywords: heat shock protein|Pyrococcus horikoshii protease I (PhPI)|DJ-1 family

Exposure of living cells to high temperatures or other environmental insults leads to the accumulation of misfolded proteins that must be either refolded in a biologically active conformation or degraded to their constituent amino acids to guarantee efficient cell survival. In the cytoplasm of Escherichia coli, these tasks are performed mainly by a set of stress-inducible molecular chaperones and proteases collectively known as heat shock proteins (Hsps). Hsps involved in folding quality control bind hydrophobic stretches of amino acids that become exposed to the solvent by substrate proteins during the unfolding process. Molecular chaperones may either actively mediate the refolding of client polypeptides through multiple cycles of binding and release, or disentangle preformed thermal aggregates to allow their subsequent refolding (1, 2). On the other hand, heat shock proteases hydrolyze proteins that have become irreversibly misfolded and are beyond repair by the cellular complement of molecular chaperones (3). Nevertheless, functional lines can become blurred, as exemplified by the fact that the ATPase components of the ClpAP and ClpXP proteases exhibit protein-remodeling activity (4), and the observation that the periplasmic protease DegP (HtrA) switches from chaperone to protease function on temperature upshift (5).

Transcriptome analysis has revealed that the E. coli chromosome contains a number of heat-inducible genes of unknown function (6), raising the possibility that the current inventory of bacterial molecular chaperones and proteases is incomplete. Recently, we have shown that E. coli Hsp31 (EcHsp31), the stress-inducible (6) and H-NS-regulated (7) product of the yedU gene (now referred to as hchA for heat-inducible chaperone), is an efficient homodimeric molecular chaperone whose activity is modulated by temperature and ATP binding (8). While highly conserved orthologs of unknown function are present in a number of pathogenic eubacteria and fungi (8), EcHsp31 also shares ≈19% identity with Pyrococcus horikoshii protease I (PhPI) (9) and contains a signature sequence matching that of the α-β hydrolases (10). In an effort to gain further insight on EcHsp31 function, we determined its structure at 1.6-Å resolution. Although part of the EcHsp31 fold superimposes on PhPI (11), the architecture of the putative catalytic triad and the quaternary structure are very different. The implications of these results for a dual chaperone–protease function of EcHsp31 and the structural relevance for the Hsp31 family are discussed.

Materials and Methods

Protein Crystallization.

Selenomethionine-labeled EcHsp31 crystals were grown by vapor diffusion at 14°C. Crystals grown at both higher and lower temperatures exhibited lower diffraction quality. The drops, containing 4 μl of protein solution (14 mg/ml in 50 mM Tris, pH 7.5/1 mM EDTA/4 mM DTT) and 5 μl of the reservoir solution (25% PEG 3350/300 mM MgCl2/50 mM MES, pH 6.5/4 mM DTT) yielded crystals in 5 days, which were flash-frozen in 15% ethylene glycol. The crystals belong to space group C2 and contain one molecule per asymmetric unit with a corresponding Vm of 2.01 Å3/Da (ref. 12; see Supporting Materials and Methods, which is published as supporting information on the PNAS web site, www.pnas.org).

Structure Determination.

Multiple-wavelength anomalous dispersion (MAD) data sets were collected on beamline 19-BM (Advanced Photon Source, Argonne National Laboratories, Argonne, IL) and processed with HKL2000 (ref. 13; Table Table1).1). Using three wavelengths of MAD data at 1.6-Å resolution, solve found all seven selenium sites for initial phases (14). Density modification and automated model building were carried out with RESOLVE (15), which traced side chains for 81% of the protein model. The resulting maps were excellent, and further refinement and model building was performed with alternate rounds of REFMAC5 (16) and XTALVIEW (17).

Table 1
Data collection and refinement statistics


The EcHsp31 Monomer.

The 1.6-Å crystal structure of EcHsp31 has clear electron density for 279 residues of 282 native residues with an Rcryst of 18.7% and Rfree of 24.2% (Table (Table1).1). Hence, the model provides a solid basis for the analysis of structure, function, and evolutionary relationship within the Hsp31 family. The EcHsp31 monomer has approximate dimensions of 55 × 40 × 40 Å and contains 13 β-strands and 12 α-helices, forming two α-β domains referred to as “A” and “P.” Domain A consists of 202 residues containing the following three segments: residues 46–56, 74–209, and 229–283 (Fig. (Fig.1).1). This domain is an α-β sandwich made up of a mixed β-sheet with the following strand order: β7, β6, β3, β8, β9, β13, and β12. The latter is the only antiparallel strand of the central β-sheet. Helices α2–α9, α11, and α12, and strands β10–β11 pack against this core β-sheet (Fig. (Fig.1).1). The 65-residue P domain protrudes from the “top” of the A domain and is composed of the following three subregions: the P1 segment (residues 1–31), the P2 segment (residues 57–73), and the P3 segment (residues 210–228). This domain consists of a four-stranded antiparallel β-sheet with two α-helices on one side. Interactions between the A and P domains are extensive and bury ≈2,200 Å2. A “linker” region formed by residues 32–45 connects the two domains. Although positioned in a well-defined groove of the A domain, the linker has higher B factors than most of the structure.

Figure 1
Ribbon representation of the EcHsp31 monomer. The A domain is blue, the P domain is green (P1–P3 segments are in progressively lighter shades of green), and the linker region is purple. The catalytic triad is shown as a red ball-and-stick model ...

The interface between the A and P domain creates a pocket containing a potential catalytic triad. Secondary structure elements contributing side chains to this pocket are the C-terminal side of the parallel β strands in the A domain and helices α1 and α10 of the P domain. The putative catalytic triad is composed of Cys-185 and His-186, from the A domain, and Asp-214 from α10 of the P domain. Cys-185 perches at the tip of a tight connection between strand β9 and helix α9, having an unusually strained combination of [var phi], ϕ angles with values of 65° and 116°, respectively. This “nucleophilic-elbow” motif was originally identified in α-β hydrolases and imposes a strained conformation that poises the nucleophilic residue for attack in several enzymatic processes (10).

Although Cys-185 is solvent accessible, it lies at the bottom of a deep and restricted, two-cavity, S-shaped pocket, which is ≈17 Å long, 8 Å wide, and between 8 and 11 Å deep (Fig. (Fig.22A). The opening of the pocket is bordered by a combination of charged and hydrophobic residues that restrict the opening to 6 Å. The negatively charged nature of the pocket surroundings occurs because of residues Asp-35, Asp-245, Asp-246, Glu-219, and a number of backbone carbonyls. The shape is largely defined by hydrophobic residues Pro-263, Phe-264, and Tyr-29, which restrict the middle of the pocket, whereas Pro-32, Val-33, Pro-210, Phe-209, Ala-213, Tyr-222, and Ile-247 make up the edges. Cavity 1 contains the catalytic triad at the deepest point of the pocket, ≈11 Å beneath Tyr-29 and Phe-209 (Fig. (Fig.22B). Suggestive of an oxy-anion hole, the backbone NH moieties from His-186 and Gly-154 are within 3.3 and 4.5 Å, respectively, of the Sγ of Cys-185 and within 4.7 Å of each other. Cavity 1 and Cavity 2 are connected by a narrow passageway between Pro-263 and His-74, which is only 5 Å wide. Cavity 2 widens to ≈7 Å, but is only 8 Å deep and is lined by residues Pro-75, Ile-76, Leu-79, Ala-108, Pro-110, Ala-266, and the aliphatic portion of Glu-77's side chain, thus exhibiting a more hydrophobic environment than Cavity 1 (Fig. (Fig.2).2).

Figure 2
Putative active site pocket of EcHsp31. (A) Surface representation of the active site pocket. Surfaces from the A and P domains are blue and green, respectively. Two residues of the triad, Cys-185 in yellow and His-186 in red, are visible. (B) Residues ...

The EcHsp31 Dimer.

The EcHsp31 dimer is generated by the crystallographic twofold axis and measures ≈65 × 60 × 55 Å. Dimer formation buries ≈1,900 Å2 of solvent-accessible surface area. The P domain of subunit I contacts both the A′ and P′ domains of subunit II, with the P–P′ and P–A′ interactions burying ≈550 and ≈1350 Å2, respectively. Residues Ala-14, Ala-18, Phe-20, and Tyr-60 of the P domain contact Leu-68′ of the P′ domain, and Met-101′, Tyr-106′, and Phe-120′ of the A′ domain. In addition to these hydrophobic interactions, the dimer is stabilized by salt bridges between P1 and P2′ involving Glu-15 and Arg-59′, and between P2 and A′ involving Glu-58 and Lys-103′. No A–A′ interactions are involved in dimer stabilization.

When viewed along the twofold axis, the dimer sports a deep canyon between the A domains of the two subunits that spans almost the entire 22-Å-long waist of the dimer (Figs. (Figs.33 and and4).4). The canyon winds from one side of protein to the other, but, even at constricted areas, it remains 8 Å in width. Far away from the canyon, the catalytic triads are located on opposite ends of the dimer with ≈40 Å between Sγ of the cysteines (Fig. (Fig.3).3).

Figure 3
Two views of ribbon representation of the EcHsp31 dimer. The A and P domains of subunit I are blue and green, respectively, and are pale blue and pale green for subunit II, respectively. The linker is purple and light purple for subunits I and II, respectively. ...
Figure 4
Two views of the electrostatic potential of the EcHsp31 dimer. Each view was generated with grasp (26). Electrostatic surfaces are colored between −10 kT (red) and +10 kT (blue). Canyon and bowl are labeled. The negative groove connecting ...

The electrostatic distribution of charges on the surface of the EcHsp31 dimer reveals a number of distinctive characteristics. When viewed along the twofold axis, the majority of the surface is uncharged (Fig. (Fig.4),4), with two notable exceptions. First, a number of positive charges cluster at the distal corners of the dimer provided by loops of the A domain (Fig. (Fig.4).4). Second, the canyon is dominated by negatively charged residues such as Asp-12, Glu-15, Asp-16, Glu-58, Asp-157, Asp-166, and Glu-175, which line the base and walls of the canyon. At the end of the canyon, the trail of electrostatically negative potential continues into a negative groove that leads around the dimer to the catalytic triad pocket (the black lines in Fig. Fig.4).4). Because of the symmetry of the dimer, the negative groove connects the two distant triads like the seams on a tennis ball. Finally, on the bottom side of the dimer, the P domain interface forms a large, shallow, bowl-shaped, hydrophobic surface measuring ≈20 Å in diameter. Residues Val-11, Phe-19, Tyr-24, Leu-26, Thr-30, Tyr-106, Phe-120, and Ile-220 contribute to the hydrophobic nature of this surface (Fig. (Fig.44B). The system of canyon and grooves that connects the putative triads suggests the possibility of intriguing connections between substrate–protein binding and hydrolytic activity under certain conditions.

Comparison of EcHsp31 with PhPI.

The dali algorithm (18) identified PhPI as the closest structural neighbor of EcHsp31 with a Z score of 46.5, a 2.1-Å rms deviation of >154 Cα positions, and a sequence identity of 19%. Superposition of the corresponding Cα atoms of PhPI on to the EcHsp31 structure reveals 15 equivalent secondary structure elements, all situated in the A domain of EcHsp31 (Fig. (Fig.55A). Six strands of the core β-sheet of EcHsp31 correspond to strands S1, S5, S6, S7, S10, and S13 of PhPI. The two remaining β-strands of PhPI, S8, and S9, superimpose well with β10 and β11 of EcHsp31, and some helices flanking the β-sheet of PhPI have counterparts in EcHsp31 (Fig. (Fig.55A). Despite this high degree of similarity in the overall fold, the Met-109–Asn-141 stretch of EcHsp31 is mostly helical, whereas β-strands dominate the corresponding region in PhPI (Figs. (Figs.55 and and6).6). Additionally, the 65 residues of the P domain of EcHsp31 are absent from PhPI (Fig. (Fig.55 A and B).

Figure 5
Structural comparison of EcHsp31 and PhPI. The A and P domains of EcHsp31 are blue and green, respectively, and PhPI is orange. The catalytic triad of EcHsp31 is colored by atom type: carbons are green, oxygens are red, nitrogens are blue, and sulfurs ...
Figure 6
Structure-guided sequence alignment of the Hsp31 family of proteins. Class I, class II, and class III members of the Hsp31 family are designated with blue, black, and orange roman numerals, respectively. Red boxes show 100% conservation across ...

The quaternary structures of EcHsp31 and PhPI are entirely different. The PhPI hexamer can be described as a trimer of dimers, resulting in two types of subunit interactions, A to B and A to C (11). Interactions that form higher-order oligomers in PhPI are absent from EcHsp31 because of helices α3 and α4, which prevent the A–B interaction, and the P domain, which prevents the A–C interaction. Although the PhPI enzyme forms a hexamer in the crystal structure, Halio et al. (19) have shown that PfPI, a PhPI homolog, retains proteolytic activity, even as a dimer (11, 19). The shared active site between subunits A and C of PhPI performs proteolytic cleavage through a Cys–His–Glu catalytic triad. Cys-100 and His-101 reside on the A subunit, whereas Glu-74′ is provided by the neighboring C subunit. Superposition of PhPI and EcHsp31 places Cys-100 and His-101 of PhPI on to Cys-185 and His-186 of EcHsp31 with a deviation between Cα atoms of 0.1 and 0.4 Å, respectively, which aligns the Sγ cysteines and imidazoles of the histidines (Fig. (Fig.55C). Most interestingly, where the A–C dimer interface of PhPI completes the active site between two subunits, the P domain of EcHsp31 performs the same function within a single subunit. In this way, EcHsp31 forms its active site with a carboxylate from Asp-214, instead of relying on a neighboring subunit as in PhPI. Although the Cα positions of the PhPI Glu-74′ and EcHsp31 Asp-214 differ by 2.6 Å, the Oδ of Asp-214 and the Oepsilon of Glu-74′ are within 0.8 Å of one another and both are within 2.7 Å of the nitrogen of their corresponding histidine. Finally, the backbone amides of His-186 and Gly-154, which serve as a potential oxy-anion hole in EcHsp31, correspond to the amides of His-101 and Gly-70 in PhPI and deviate in position by only 0.4 and 0.8 Å, respectively. Structurally speaking, EcHsp31 possesses the characteristic features of an active catalytic triad as observed in PhPI.

Hsp31 Family Sequence Alignment.

blast searches yielded a number of distinct EcHsp31 homologs (20). Class I orthologs share >55% sequence identity with EcHsp31 and include closely related proteins from the human pathogens enterohemorrhagic E. coli, Vibrio cholerae, Staphylococcus aureus, Pseudomonas aeruginosa, and Pseudomonas fluorescens (8). Class II orthologs have sequence identities ranging from 24% to 29%, and contain proteins from Xylella fastidiosa, Agrobacterium tumefaciens, Sinorhizobium meliloti, Bacillus anthracis, Enterococcus faecalis, Xanthomonas campestris, Coccidioides immitis, and Schizosaccharomyces pombe (8). Although most of these proteins are hypothetical, a few belong to the disparate DJ-1/PfPI family, which hosts proteins with a wide range of functions.

Sequence similarities have previously been identified between EcHsp31 and PfPI, another member of the DJ-1/PfPI family (9). Together, PfPI and PhPI make up class III orthologs with <20% sequence identity to EcHsp31. On the basis of our structure, the sequence alignment of EcHsp31 and PfPI reported by Kim et al. (9) needed correction for the P domain (Fig. (Fig.6).6). Multiple sequence alignments with various programs led to the same misalignment (data not shown). Structural alignment of EcHsp31 and PhPI (Fig. (Fig.6)6) results in 19% sequence identity, which is much lower than the 25.9% reported on the basis of sequence alone (9). Fig. Fig.66 shows that the β-strand S1 of PhPI aligns with β3 in the A domain of EcHsp31, and H7 aligns with α11 (Fig. (Fig.6).6). In fact, the P3 segment of EcHsp31 is crucial for correct orientation of the catalytic Asp-214. Therefore, the structural counterpart of EcHsp31 Asp-214 is not Glu-125, but the catalytic Glu-74′ of PhPI (Fig. (Fig.66).

Combining the sequence alignments of Hsp31 orthologs and the DJ-1 family, together with the overall architecture of EcHsp31 and PhPI, allows for the construction of a structure-guided sequence alignment that highlights similar characteristics within this family. All members possess a domain similar to the A domain of EcHsp31 as previously suggested (9), and a motif matching the nucleophilic elbow motif (10). Class I and II members also exhibit a conserved intradomain catalytic triad.

Nevertheless, there are clear differences among the three classes of Hsp31 orthologs. First, the P domain seems to be a source of variation, changing in length throughout the class. Particularly, the P1 segment is found only in class I, whereas the P2 and P3 segments are present in various degrees within class II, but are completely absent from class III. Variation in the P domains affects the oligomerization states available to the proteins of the Hsp31 family. Within class I Hsp31 homologs, the 12 dimer-forming interactions are well conserved. Within class II Hsp31 homologs, the P1 segment, which contains 5 of the 12 residues involved in dimerization, is completely absent, and the remaining 7 interface residues show no conservation, varying in both size and charge (Fig. (Fig.6).6). Finally, the hexameric class III orthologs (11) are missing all three P regions (Fig. (Fig.6).6). Interestingly, the dimer-forming residues, which are well conserved among class III family members, are not conserved within class II Hsp31 orthologs, suggesting that the latter class of proteins may adopt yet another quaternary structure.


Chaperone Activity.

EcHsp31 represents a previously unrecognized family of Hsps (8), whose function can now be discussed in light of its structure. Dynamic light-scattering and size-exclusion studies have shown that EcHsp31 forms a dimer in solution (8). The extensive buried-surface interactions between subunits and the conservation of dimer-forming residues within class I Hsp31 orthologs suggest that this structure is a biologically relevant form of the protein (Fig. (Fig.6).6). Molecular chaperones facilitate de novo protein folding and prevent stress-induced misfolding by binding to unstructured and hydrophobic regions (4). Like other chaperones, the EcHsp31 homodimer exhibits a number of solvent-exposed hydrophobic patches (Fig. (Fig.4).4). Many of these are located on the A domain, but the largest one is the hydrophobic bowl at the P domain interface (Fig. (Fig.44B). This concave surface is highly conserved within class I homologs and may help EcHsp31 bind early unfolding intermediates and release them once stress has abated (8). Another interesting structural feature of EcHsp31 is the canyon that dominates the waist of the dimer (Fig. (Fig.4).4). Although the canyon could serve as a binding site for partially unfolded polypeptides in general, the predominance of negative charges within the region argues for substrate specificity. The groove connecting the canyon and the two putative catalytic triads suggests that the canyon may be implicated in the EcHsp31 hydrolytic function under specific conditions. From a bacterial perspective, it might make sense to degrade a misfolded protein at higher temperatures, rather than to attempt to refold it. Hence, both functions could be protective in nature.

Biochemical data indicate that structural flexibility plays an important role in the chaperone activity of EcHsp31. Fluorescence studies have shown that EcHsp31 exposes more hydrophobic surfaces at high temperatures without undergoing significant change in secondary structure (8). This exposure may be accomplished by the movement of loop regions. Because of its lack of secondary structure and high B factors, the linker region (Fig. (Fig.1)1) is a prime candidate for flexibility. Repositioning the linker would not disrupt the secondary structure of the P or A domains because it packs against the outside of the α-β sandwich. Flexibility can play a key role in other chaperone functioning, as illustrated by the two conformations of PDZ domains in DegP (21). In EcHsp31, the linker region may move out of the way at high temperatures, while serving as a low-affinity substrate mimic at low temperatures, when EcHsp31 may not be proteolytically active, but instead functions as a chaperone.

Catalytic Triad.

Catalytic triads have been identified in various classes of proteins, including proteases (22), hydrolases (10), transglutaminases (23), and amidotransferases (24). Despite their adaptability to different functions, three major elements of the triad are usually maintained for hydrolysis: the triad itself, an oxy-anion hole, and the substrate-binding site. In EcHsp31, Cys-185, His-186, and Asp-214 form a putative catalytic triad, and backbone amides of His-186 and Gly-184 could fulfill the role of the oxy-anion hole, which is usually necessary for efficient hydrolysis (25). The last element necessary for hydrolytic activity is the substrate-binding site. The pocket that holds the catalytic triad in EcHsp31 narrows to 5 Å, restricting access to the triad (Fig. (Fig.2).2). Despite the conservation of Gly-153 and Gly-154, which presumably are conserved to allow room for a substrate in cavity 1, space is still limited within the active site. Therefore, although EcHsp31 has a catalytic triad and an oxy-anion hole, the restricted space for substrate suggests that some conformational change is required to turn EcHsp31 into a protease. Such conformational changes might be the result of temperature elevation, binding of a regulatory protein, or binding of a cofactor.

For the latter possibility, GMP synthetase makes an interesting case because, although it has all three elements of an active hydrolase, it is a poor glutaminase unless ATP binds to a neighboring domain. In this way, GMP synthetase couples the removal of the amine from glutamine to the next reaction step, which requires ATP (24). In the case of EcHsp31, ATP negatively regulates the chaperone function and may also influence a hydrolytic function (8). Although the ATP-binding site has not been identified in EcHsp31, ATP binding restricts the availability of hydrophobic surfaces that are normally exposed at high temperature (8). Of course, the effect of ATP does not rule out the existence of another cofactor for EcHsp31.

Finally, lack of the knowledge regarding the in vivo substrate might explain the lack of observed proteolytic activity for EcHsp31. Although DegP serves as a general chaperone, only a handful of its native proteolytic targets have been identified (5). Preliminary assays of EcHsp31 have not yet shown proteolytic activity. Hence, discovering the proteolytic targets of EcHsp31 clearly requires further investigation.

Implications of EcHsp31 Sequence Alignment.

A surprising discovery based on our structure-guided sequence alignment is that there appears to be three classes of Hsp31 homologs (Fig. (Fig.6).6). Class I is characterized by the following: (i) a complete P domain; (ii) aspartate as the third member of an intrasubunit catalytic triad; (iii) dimer formation through P1 and P2 segments; and (iv) an extended linker region that may govern access to the active site. Class II has the following characteristic features: (i) an incomplete P domain, missing the P1 segment; (ii) glutamate as the third member of catalytic triad; (iii) an oligomerization state likely to be different from either class I or class III; and (iv) the absence of P1 linker region, which could result in a more accessible active site. Class III is represented by PhPI and PfPI and has the following different features: (i) The P domain is completely absent; (ii) the glutamate of the catalytic triad is provided by a different subunit; (iii) three dimers arrange themselves in a hexamer with D3 symmetry; and (iv) the P1 linker is absent, but access to the active site is restricted by neighboring subunits (11). This alignment makes an intriguing case for divergent evolution within the Hsp31 family because the key catalytic Cys–His dyad is maintained throughout the family, whereas the less crucial member of the triad originates from various subunits depending on class. Overall, the structure-guided sequence alignment (Fig. (Fig.6)6) shows that the insertion of the P domain, which is unique to the class I and class II Hsp31 orthologs, has repercussions on both the quaternary structure and catalytic center of this family of proteins with an intriguing evolutionary relationship between members.

Supplementary Material

Supporting Text:


We thank M. A. Robein, D. R. Davies, N. Sanishvili, M. Sastry, and S. Turley for useful discussions. W.G.J.H. thanks the Murdoch Charitable Trust for a major equipment grant to the Biomolecular Structure Center. This work was supported by American Cancer Society Grant MBC-99-335-01 (to F.B.).


heat shock protein
Pyrococcus horikoshii protease I


Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.rscb.org (PDB ID code 1N57).


1. Hartl F U, Hayer-Hartl M. Science. 2002;295:1852–1858. [PubMed]
2. Ben-Zvi A P, Goloubinoff P. J Struct Biol. 2001;135:84–93. [PubMed]
3. Gottesman S, Wickner S, Maurizi M R. Genes Dev. 1997;11:815–823. [PubMed]
4. Wickner S, Maurizi M R, Gottesman S. Science. 1999;286:1888–1893. [PubMed]
5. Spiess C, Beil A, Ehrmann M. Cell. 1999;97:339–347. [PubMed]
6. Richmond C S, Glasner J D, Mau R, Jin H F, Blattner F R. Nucleic Acids Res. 1999;27:3821–3835. [PMC free article] [PubMed]
7. Yoshida T, Ueguchi C, Yamada H, Mizuno T. Mol Gen Genet. 1993;237:113–122. [PubMed]
8. Sastry M S, Korotkov K, Brodsky Y, Baneyx F. J Biol Chem. 2002;277:46026–46034. [PubMed]
9. Kim O G, Kim I K, Kim G H, Ko J S, Park C, Suh P G, Kang S O, Lee H S, Cha S S. Acta Crystallogr D. 2002;58:1217–1219. [PubMed]
10. Ollis D L, Cheah E, Cygler M, Dijkstra B, Frolow F, Franken S M, Harel M, Remington S J, Silman I, Schrag J, et al. Protein Eng. 1992;5:197–211. [PubMed]
11. Du X L, Choi I G, Kim R, Wang W R, Jancarik J, Yokota H, Kim S H. Proc Natl Acad Sci USA. 2000;97:14079–14084. [PMC free article] [PubMed]
12. Matthews B W. J Mol Biol. 1968;33:491–497. [PubMed]
13. Otwinowski Z, Minor W. Methods Enzymol. 1997;276:307–326.
14. Terwilliger T C, Berendzen J. Acta Crystallogr D. 1999;55:849–861. [PMC free article] [PubMed]
15. Terwilliger T C. Acta Crystallogr D. 2001;57:1755–1762. [PMC free article] [PubMed]
16. Murshudov G N, Vagin A A, Dodson E J. Acta Crystallogr D. 1997;53:240–255. [PubMed]
17. McRee D E. J Struct Biol. 1999;25:156–165. [PubMed]
18. Holm L, Sander C. J Mol Biol. 1993;233:123–138. [PubMed]
19. Halio S B, Bauer M W, Mukund S, Adams M W W, Kelly R M. Appl Environ Microbiol. 1997;63:289–295. [PMC free article] [PubMed]
20. Altschul S F, Madden T L, Schaffer A A, Zhang J H, Zhang Z, Miller W, Lipman D J. Nucleic Acids Res. 1997;25:3389–3402. [PMC free article] [PubMed]
21. Krojer T, Garrido-Franco M, Huber R, Ehrmann M, Clausen T. Nature. 2002;416:455–459. [PubMed]
22. Schmid M F, Herriott J R. J Mol Biol. 1976;103:175–190. [PubMed]
23. Yee V C, Pedersen L C, Letrong I, Bishop P D, Stenkamp R E, Teller D C. Proc Natl Acad Sci USA. 1994;91:7296–7300. [PMC free article] [PubMed]
24. Tesmer J G, Klem T J, Deras M L, Davisson V J, Smith J L. Nat Struct Biol. 1996;3:74–86. [PubMed]
25. Kraut J. Annu Rev Biochem. 1997;46:331–358. [PubMed]
26. Nicholls A, Sharp K, Honig B. Proteins. 1991;11:281–296. [PubMed]
27. Kabsch W, Sander C. Biopolymers. 1983;22:2577–2637. [PubMed]
28. Kraulis P J. J Appl Crystallogr. 1991;24:946–950.
29. Molecular Simulations. Insight II Molecular Modeling System User Guide. San Diego: Molecular Simulations; 1995.
30. Gilbert D, Westhead D, Viksna J, Thornton J. J Comput Chem. 2001;26:23–30. [PubMed]
31. Gouet P, Courcelle E, Stuart D I, Metoz F. Bioinformatics. 1999;15:305–308. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...