HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 48, 46743-46752, November 29, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/48/46743    most recent M207796200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Guo, F. Articles by Xia, D. Search for Related Content PubMed PubMed Citation Articles by Guo, F. Articles by Xia, D. Social Bookmarking                      What's this?

Crystal Structure of ClpA, an Hsp100 Chaperone and Regulator of ClpAP Protease^*

Fusheng Guo, Michael R. Maurizi, Lothar Esser, and Di Xia

From the Laboratory of Cell Biology, Center for Cancer Research, NCI, National Institutes of Health, Bethesda, Maryland 20892

Received for publication, August 1, 2002, and in revised form, August 20, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Escherichia coli ClpA, an Hsp100/Clp chaperone and an integral component of the ATP-dependent ClpAP protease, participates in regulatory protein degradation and the dissolution and degradation of protein aggregates. The crystal structure of the ClpA subunit reveals an N-terminal domain with pseudo-twofold symmetry and two AAA⁺ modules (D1 and D2) each consisting of a large and a small sub-domain with ADP bound in the sub-domain junction. The N-terminal domain interacts with the D1 domain in a manner similar to adaptor-binding domains of other AAA⁺ proteins. D1 and D2 are connected head-to-tail consistent with a cooperative and vectorial translocation of protein substrates. In a planar hexamer model of ClpA, built by assembling ClpA D1 and D2 into homohexameric rings of known structures of AAA⁺ modules, the differences in D1-D1 and D2-D2 interfaces correlate with their respective contributions to hexamer stability and ATPase activity.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Energy-dependent intracellular protein degradation is needed for cell growth, mediation of stress responses, and supervision of protein quality control (1). ATP-dependent proteases, such as the Clp proteases and proteasomes, are composed of a proteolytic core particle, in which the active sites are compartmentalized, and an ATP-dependent regulatory particle with ATPase and chaperone activity (2). In Escherichia coli ClpAP and ClpXP, the functional units are ClpP, a protease core of 14 identical subunits in two stacked heptameric rings (3) and either ClpA or ClpX, ATP-dependent chaperones consisting of six identical subunits arranged in hexameric rings (2). The ATPase subunits are responsible for substrate selection, protein unfolding and translocation to the proteolytic core, and allosteric modulation of the ClpP activity; they also have stand-alone chaperone activity, catalyzing limited structural remodeling of proteins and disassembly of stable protein complexes (4) in the absence of the proteolytic component. ATP binding and hydrolysis have distinct roles and are essential in various activities of ClpA or ClpX (5-7).

ClpA belongs to the AAA⁺ superfamily of ATPases associated with various cellular activities, a ubiquitous family of ATP-dependent molecular machines with one or two 230-residue AAA⁺ modules containing well conserved sequence and structural motifs (8). Atomic structural information is available for several single-AAA⁺ modules (9-17), all of which consist of a large and a small sub-domain, with the nucleotide bound at the sub-domain interface. The AAA⁺ modules assemble into homologous or heterologous rings, with hexameric rings being most common. Structural analysis of AAA⁺ proteins in several nucleotide states has led to models of possible conformational changes that accompany ATP hydrolysis and exchange of nucleotides and that may have a role in chaperone activity and promotion of protein degradation (18).

ClpA subunits have two non-identical AAA⁺ modules in tandem, similar to the autonomous molecular chaperone Hsp104 (ClpB) (19) and the membrane-fusion proteins N-ethylmaleimide sensitive factor (NSF)¹ (20) and p97 (21). In contrast, ClpX, like ClpY (HslU), the regulatory particle of the proteasome, and diverse others, possess a single AAA⁺ module. The two ATPase domains have non-equivalent roles in different AAA⁺ proteins; the domain most involved in hexamer assembly (D1 in ClpA) or possessing higher ATPase activity (D2 in ClpA) differs between ClpA and Hsp104 or NSF (5, 22, 23). Electron microscopy (EM) studies (2) of ClpA show a two-tiered structure with topologically separate rings formed by the two AAA⁺ modules. In proteolytic complexes, only one ring surface of ClpA interacts with ClpP, whereas the other binds protein substrates (24). Contact with ClpP is mediated in part by an exposed loop in D2 called the ClpP loop (25, 26). In addition to the AAA⁺ modules, ClpA subunit has a unique N-terminal domain (N-domain) of 150-160 residues, which has not been definitively visualized by EM studies. Here, we report the crystal structure of the full length ClpA, the first atomic resolution structure of a complete AAA⁺ protein with two ATPase domains, and discuss its functional implications.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Protein Purification and Crystallization of the Full Length and N-Domain of ClpA-- The expression plasmids for the ClpA-M169L mutant protein and for the N-domain were derived from pSK39 (27). A stop codon was introduced after the residue R143 to generate the N-domain expression vector. E. coli strains SG22176 clpA::kan (28) containing corresponding plasmids were grown in LB medium at 30 °C to an A₆₀₀ of 0.7, 1 mM isopropyl -D-thiogalactoside was added, and cells were incubated a further 2.5 h. The cells were harvested and broken in a French press at 20,000 pounds/square inch. The crude extracts were treated with 0.05% polyethyleneimine, and the precipitates were removed by centrifugation. For full length ClpA, the supernatant after the polyethyleneimine treatment was separated on a SP Sepharose column (Amersham Biosciences) as described (27), and then applied to a hydroxyapatite (Bio-Rad) column equilibrated with 20 mM potassium phosphate, pH 7.2, and eluted with a gradient of phosphate from 20 to 200 mM. After concentration, the protein was run over a Bio-Sil TSK250 gel filtration column (Bio-Rad) in 50 mM Tris/HCl, pH 7.5, 0.2 M KCl, and 10% glycerol. ClpA was precipitated with 50% saturated ammonium sulfate, dissolved to a concentration of 12 mg/ml on ice, dialyzed against three changes of 1 liter of buffer H (0.1 M HEPES, pH 7.5, 10% glycerol) for 24 h, and stored at 80 °C. For N-domain, the supernatant after polyethyleneimine treatment was applied to a Q Sepharose column (Pharmacia Biosciences) and a Bio-Sil TSK250 gel filtration column (Bio-Rad) in sequence. The purified fractions were precipitated with 60% ammonium sulfate, dialyzed against 20 mM HEPES buffer, pH7.5, and stored at 80 °C.

For crystallization, frozen aliquots of the full length ClpA were thawed on ice, and 8 mM ATPS or ADP and 40 mM MgCl₂ were added. Optimal condition for crystallization is achieved by mixing 10 µl of protein solution with an equal amount of well solution containing HEPES (0.1 M, pH 7.5), KCl (0.5 M), glycerol (5%), isopropanol (8%), and sodium azide (0.01%). ClpA crystals up to 400 × 400 × 400 µm in size were routinely obtained after incubation at 21 °C for 1 month. ClpA crystals were soaked in a stepwise manner in cryo-protection solution containing HEPES (0.1 M, pH 7.5), KCl (0.3 M), PEG 4000 (30%), and glycerol (35%). Crystals were frozen in liquid propane. Heavy atom derivative crystals were prepared following the same protocol as the cryo-soaking, except for the last soaking cycle, at which point crystals were treated with cryoprotection solution containing 1-10 mM heavy metal compounds. In a like manner, frozen aliquots of N-domain protein were thawed on ice, and a 2 × 2 µl hanging drop was set up with the reservoir buffer containing PIPES (0.1 M, pH 7.0), PEG 8000 (35%), and sodium citrate (0.32 M). Crystals of N-domain were produced in 3 months at 21 °C.

Data Collection and Structure Determination-- Native and derivative data sets were collected mainly at beamline X9B of NSLS (National Synchrotron Light Source), Brookhaven National Laboratory, at BioCARS and Dow-Northwestern-Dupont beamline of APS (Advanced Photon Source), Argonne National Laboratory, on ADSC 2 × 2 CCD and MAR CCD detectors. Cryo-frozen full length ClpA crystals diffracted x-rays to a maximum of 2.4-Å resolution. They have the symmetry of space group P6₅ and average cell dimensions of a = b = 123.7 Å, c = 97.8 Å. Diffraction images were processed with the programs in the HKL package (29). Native and heavy atom derivative data sets were scaled together with the programs TRUNCATE, CAD, and SCALEIT of the CCP4 package (30). Difference Patterson map between the MeHgAc derivative and native2 (Table I) was calculated with the FFT program and three heavy atom sites were identified. For all other derivatives, heavy metal sites were determined by the difference Fourier technique with single isomorphous replacement phases from the single Hg derivative. Heavy atom parameters for all derivatives were refined and multiple isomorphous replacement phases were calculated with the MLPHARE program (31). Phasing statistics for the four derivatives are given in Table I. Density modification using a solvent content of 45% with the program DM (32) yielded an interpretable electron density map, into which 713 residues of 758 were built with the program O (33). The model was refined with the program REFMAC (34) against a higher resolution native data set (Native2, Table I) with iterative manual model rebuilding. Crystals of ClpA N-domain diffracted x-rays to 1.8-Å resolution at room temperature in space group P2₁2₁2₁ and have average cell dimensions of a = 32.0 Å, b = 52.0 Å, c = 65.1 Å. The structure was determined using a crude polyalanine tracing model from the N-domain of the full length ClpA in a molecular replacement rotational and translational search, followed by refinement of several search solutions using the CNS-solve program (35). One solution was refined successfully, and the final statistics for the refinement are given in Table I. The 1.8-Å resolution N-domain structure was subsequently used in the refinement of the full length ClpA structure.

Structure Alignments, Analysis, and Modeling of Hexameric ClpA-- Least-squares structure alignments of ClpA D1 to D2, and to other known AAA⁺ modules were performed in the program O. Buried surface area was calculated with the program AREAIMOL (36) in CCP4. The hexameric structures of NSF-D2 (PDB entry 1D2N) and HslU (PDB entry 1DO0) were generated by supplying appropriate crystallographic symmetry operators. To build a hexameric model of ClpA, the hexameric N-D1 and D2 models were built using the NSF hexamer and HslU hexamer as templates, respectively. The resulting N-D1 and D2 rings were energy-minimized separately with imposed 6-fold symmetry and joined together co-axially in an orientation and separation that rendered minimum van der Waals energy and a distance of 4 Å between the C-terminal residue of N-D1 domain and that of the N terminus of the D2 domain.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Overall Structure of the Monomeric ClpA-- We solved and refined the structures for both the full length ClpA in the space group P6₅ at 2.6-Å resolution and the N-domain of ClpA (residue 1-142) in the space group P2₁2₁2₁ at 1.8-Å resolution, respectively. Details of structure solutions are given under "Experimental Procedures," and the statistics for crystallographic data collection, phasing, and refinement are provided in Table I. Stepwise soaking of glycerol for cryoprotection made ClpA crystals mosaic, with mosaicities in the range between 0.8° and 1.5°. The best native crystal diffracted x-rays to 2.4-Å resolution at synchrotron, but diffraction intensities fell off rapidly beyond 3.5 Å. Diffraction spots higher than 2.6 Å were not included in refinement because they failed to follow the Wilson statistics (37). More than 30 heavy atom derivative data sets were collected, and four were sufficiently isomorphic to the native (Table I) to contribute to the MIR phasing. There are several disordered regions in the ClpA structure; all of them are loops connecting secondary structure elements or between domains (Fig. 1b). There appeared to be some variations in crystal packing of ClpA domains because the TLS refinement in the Refmac (34) significantly reduced R_free (35). The diffraction data set for the N-domain crystal was collected at room temperature at synchrotron from a single crystal (Table I) that diffracted x-rays to 1.8-Å resolution. The data set was incomplete especially at high resolution. The N-domain structure was nevertheless refined successfully with all data included, and the final refined electron density showed characteristics of a high resolution structure.

View larger version (40K): [in this window] [in a new window]   Fig. 1.   The structure of the full length ClpA. a, in ribbon presentation, the five structural domains of ClpA monomer are color coded; the N-domain, red; the large and small sub-domain in D1, dark and light green, respectively; the large and small sub-domain in D2, yellow and purple, respectively. Disordered loops are shown as dashed coils. Two ADP molecules located at the interface between sub-domains are rendered as ball-and-stick models. The 65 crystallographic axis is shown as the red rod, and each secondary structure element is as labeled. The illustration is produced with the programs Bobscript (50), Molscript (51), and POVRAY (www.povray.org) interfaced with GL-Render (L. Esser, unpublished work). b, assignment of secondary structure elements and sequence motifs to the primary sequence of ClpA. The sequence is color-coded according to the structural domains in Fig. 1a. Below the sequence, -helices are shown as sinuous curves, -strands as solid arrows, and loops as straight lines. Secondary structure elements are named to reflect their locations in sub-domains and their topological equivalence between D1 and D2; N-domain elements are numbered; large domain elements are indicated by Roman letters; and small domain elements are indicated by Greek letters. In the N-domain, the first four helices are N-H1 through N-H4, and the symmetry-related helices in the second half are N-H1' through N-H4'. The five strands of the -sheet in the D1 large domain are D1-SA through D1-SE, and those in the D2 large domain are D2-SA through D2-SE. Helices preceding a -strand are given the same letter designation as the -strand; for example, D1-HA precedes D1-SA, and D1 HE1 and D1-HE2 both precede D1-SE. The first helices in D1 and D2 are not universally found in AAA+ molecules and are designated as transition helices, D1-HT and D2-HT, respectively; two strands of a -hairpin found only in D2 are designated D2-S1 and D2-S2, respectively. In D1 and D2 small domains, topologically equivalent elements have the same designation regardless of secondary structure. D1-H1 and D1-H2 combined are equivalent to D2-H, D1-S to D2-S, and so on. D1 ends with helix D1-H whereas D2 terminates in strand D2-S. Also highlighted in the sequence are conserved residues in the ClpA family (8) along with labels describing their functions. c, ribbon diagram of N-domain. -Helices in the first half (red) are marked sequentially (H1, H2, H3, and H4), and pseudo-symmetry related helices in the second half (green) are marked H1', H2', H3', and H4'. The pseudo-twofold symmetry axis is perpendicular to the plane of the paper. Conserved hydrophobic residues are shown as ball-and-stick models and point into the hydrophobic core; conserved hydrophilic residues are as labeled and located at the ends of H1 and H1' and flank the hydrophobic core.

The full length ClpA subunit consists of five tandemly linked structural domains (Fig. 1a) corresponding to three functional groups: an N-domain and two AAA⁺ ATPase modules. The N-domain has 142 residues; the first AAA⁺ ATPase module (D1) has 270 residues and the second ATPase module (D2) contains 319 residues, each consisting of a large and a small sub-domain (Fig. 1b). Both domains are larger than the conventionally defined 230 residue AAA⁺ module (8). The increased size comes from a set of insertions or extensions, such as the transition helices (D1-HT and D2-HT) at the start of D1 and D2 by which they are connected to the preceding domains; D1-H, which forms the anchor in D1 by which it is connected to D2; the ClpP loop, a function-specific insertion within D2; and the C-terminal -strands, S and S. Despite these additions, the ClpA ATPase modules are markedly analogous to other AAA⁺ modules whose structures are known. In the ClpA crystal, the two AAA⁺ modules are connected head-to-tail with the longest dimension inclined with respect to the 6-fold screw axis, placing a neighboring D2 domain directly underneath the D1 domain. This arrangement differs from the model of another two-AAA⁺ domain protein, p97, derived from separate crystal structures for p97 N-D1 and NSF D2 fit into a cryo-EM density envelope, which indicates a tail-to-tail connection of the two AAA⁺ modules along the 6-fold molecular axis (14). The different arrangements may reflect fundamental differences in the activities of these proteins. The head-to-tail order may be present in ClpA because the vectorial transfer of unfolded proteins into the degradation chamber of ClpP requires the sequential activity of the two domains.

Two symmetry-related N-domains have contacts with the D1-D2 domains in the crystal. The electron density for the 25 residues connecting the N-domain and D1 was not visible, but we were able to identify the cognate N-domain based on three criteria: the distance between the C-terminal residues of the N-domains and the first determined residue of D1, the surface area buried by each N-domain-D1D2 contact, and the location of contact interface. The distances for both candidates were similar (59 Å and 55 Å), but there was a significant difference in the corresponding buried surface areas, 1371 Å² versus 611 Å². Furthermore, the N-domain with the larger buried surface area was exclusively in contact with the D1, similar to the interaction seen in the p97 N-D1 structure (14), whereas the other N-domain had most of its contact with D2. The N-domain thus assigned best fit all three criteria. Other regions for which density was not well defined included residues 252-255, residues 295-300, and notably residues 611-625 (Fig. 1b). All three appear to be functionally important loops subject to conformational variability. In contrast, the electron density for the D1-D2 connector was visible without ambiguity. The overall structure is consistent with results of limited proteolysis (25) and provides precise domain boundary information.

The Structure of the N-Domain Reveals a Novel Fold-- The N-domain has a tightly packed helical structure (Fig. 1c). The secondary structure consists of eight -helices corresponding to 61% helical content (Fig. 1b). Helices H1-H4 can be superimposed onto helices H1'-H4' with a root mean square deviation of 1.56 Å for 58 pairs of C atoms. The first four helices are related to the second four by a rotation of 178°, so that the entire N-domain has pseudo-twofold symmetry. Structure based sequence alignment of the two halves of the ClpA N-domain and its homologs from different species (data not shown) reveals a weak sequence conservation to the consensus sequence for the two halves producing roughly 19 and 15% identity, respectively. In each half, all but four conserved residues are hydrophobic and are concentrated in helices, especially in H2 and H2', which form the hydrophobic core of the N-domain (Fig. 1c). Despite weak sequence identity, the tertiary structure conservation between the two halves is very significant, suggesting that the origin of the repeating motif is an early evolutionary gene duplication event.

Searches of the protein structure data base (38) with either the whole N-domain or with a single repeating four-helix motif did not produce a single match of significance; therefore, the N-domain of ClpA may represent a new fold. The first half of the N-domain has an apparent Zn²⁺-binding site consisting of residues His-20, His-22, and Glu-63 in a spatial arrangement similar to that of metalloproteases. In an accompanying paper we report the structure of the N-domain with Zn bound at this site.² These residues are conserved in only a subset of ClpA proteins, suggesting that the site may be required for a ClpA-specific interaction with substrates or adaptors. We have found that the isolated N-domain of ClpA forms a tight heterodimeric complex with ClpS (data not shown), a small protein that associates with ClpA in many organisms and modulates ClpA activity (39). The non-homologous N-domains of other AAA⁺ proteins, such as p97 or NSF, act in an analogous manner by binding unique adaptors that direct their activities toward specific intracellular pathways (40-42).

ClpA N-domain makes contacts with the D1-small domain mainly through helix H4 with >74% of the total buried surface area, and less significantly, with the D1-large domain through helix H3'. The association at the three domain interface with 1371 Å² of buried surface area appears weak, consistent with in vitro experiments showing that isolated N-domains do not bind tightly to ClpA153 (data not shown). These data together with the poor visibility of N-domains in EM reconstruction (25) suggest that the N-domain of ClpA, like those of p97 and NSF (14, 43), are highly mobile. Coupling of conformational changes in the D1 small domain to the transitions in the N-domain would provide a mechanism by which nucleotide hydrolysis can affect the binding or the position of bound substrates or adaptor proteins.

Nucleotide Binding Environments Reflect Functional Differences in D1 and D2-- Both the D1 and D2 large sub-domain have an analogous structure of a core / folding motif with a five-stranded parallel -sheet flanked by three -helices on either side. The D1 small sub-domain has five tightly packed -helices plus a small two-stranded parallel -sheet contributed by the residues prior to the H1 and H helices (Fig. 1, a and b). Despite an overall similarity in topology with the D1 small sub-domain, the D2 small sub-domain ends in the -strand S instead of a helix with a prominent surface-exposed three-stranded mixed -sheet emerging as a unique feature (Fig. 1a). In both D1 and D2, the two sub-domains unite to form a hinged quarter-moon structure with a concave and a convex surface on either side (Fig. 2). Upon assembly of the oligomer, the concave surfaces provide snug docking sites for the convex surfaces of the respective domains from an adjacent subunit. The nucleotide-binding cavity penetrates the concave surface, and the ADP molecule, required for growth of the full length ClpA crystal, lies at the interface between the two sub-domains (Fig. 1a). The rim of the ADP-binding cavity of the D1 is highly negatively charged, contributing an interface that is complementary in both charge and shape to the surface of a neighboring D1, which donates eight positively charged residues (Figs. 2, a and b, and 3a). Consequently, a number of ADP-interacting residues come from the neighboring symmetry-related D1. The concave side of the D2 surface, however, is steeper, whereas the convex side is rather smoother as compared with the surface of D1 (Figs. 2, c and d, and 3b). As a result, the nucleotide-binding site of D2, unlike that of D1, has few electrostatic interactions and does not receive contributions of residues from neighboring subunits. These differences could also underlie the greater contribution of D1 to hexamer assembly or stability.

View larger version (59K): [in this window] [in a new window]   Fig. 2.   Electrostatic potential surfaces and surface topology of the AAA+ modules of ClpA. The surfaces are rendered in GRASP (52) and superimposed with embedded ribbon presentation of structures of AAA+ modules of ClpA. a, the surface of D1-domain is shown with the concave side facing the reader. The domain is orientated with the D1-small domain on top and large domain at the bottom. The subunit interface and ADP-binding cavity are outlined black in an oval. Blue represents positive potential and red, negative. The bound ADP is shown as the ball-and-stick model with phosphorus atoms in red, oxygen in gray, carbon in yellow and nitrogen in blue. b, 180° rotated surface from a along a vertical axis showing the convex side of the D1 domain. The subunit interface is also outlined. c, surface of the D2 domain is shown with an orientation similar to that of D1 domain in a and atoms are depicted as in a. Subunit interface and the ADP binding pocket are as outlined. d, surface of D2 with a180° rotation from c.

Attempts to elucidate the mechanism of ATP-dependent protein unfolding by AAA⁺ chaperones have been complicated by the dual functions of nucleotide in promoting assembly of functional oligomers and catalyzing conformational changes that are conferred to bound protein substrates. In ClpA, ATP binds to the D2 site in unassembled subunits, but this interaction is not sufficient to allow hexamer assembly (44). ATP hydrolysis at the D2 site requires assembly promoted by nucleotide binding at D1. The mutation, K220Q, in the D1 Walker A motif causes a moderate loss of ATPase activity, primarily by impairing assembly of hexamers. In contrast, mutations in K501 of the Walker A motif in D2 impair 90% of the ATPase activity of ClpA yet allow hexamer formation (5). The structure of ClpA suggests a basis for these effects. The protein surfaces surrounding the nucleotide-binding sites in ClpA serve as subunit interfaces for hexamerization in both D1 and D2; interacting residues are contributed by large and small sub-domains and additionally in D1 by the large sub-domain of the adjacent subunit. The nucleotide-binding site in D1 is formed in part by contributions from D1 residues of a neighboring symmetry-related subunit (Sym-D1), whereas the D2 site appears self-contained (Fig. 3). Arg-339 of the Sym-D1 is salt-bridged to the -phosphate (3.2Å) and -phosphate (3.2 Å) of ADP and is in a position to act as an arginine-finger residue involved in inter-subunit catalysis of ATP hydrolysis (45); Arg-206 of the Sym-D1 is H-bonded to 2'-OH (3.0 Å) and 3'-OH (2.6Å) of the ribose moiety of ADP. An additional five positively charged residues from the same neighbor contribute mostly electrostatic interactions: Arg-204 forms two salt-bridges with Glu-404 (2.8Å) and Asp-400 (2.9Å), Arg-205 is 2.7Å away from Asp-186, Arg-206 is 3.0Å from Asp-186, and Lys-335 has a distance of 2.7Å from Glu-286. In contrast, within the D2 nucleotide-binding site, the closest residue from a neighboring subunit is more than 9 Å away, creating the middle zone filled by many structural water molecules (Fig. 3b). An estimate of the coulombic energy between charged residues of two neighboring subunits within a radius of 5 Å is 1,400 kJ/mol from 35 ion pairs for the D1-D1 interface, whereas the D2-D2 interface has none. Interestingly, many of the D1-D1 contact residues in the nucleotide pocket are conserved between ClpA and ClpB or Hsp104, suggesting that a deficiency in ClpA D2-D2 interactions, rather than differences in the stability of the respective D1 interfaces, contributes to the apparently greater relative contribution of D2-D2 interactions to assembly of Hsp104 (22). The increased intrinsic stability of Hsp104/ClpB D2 interactions may be required because these chaperones do not interact with a component such as ClpP that helps stabilize D2 interactions in ClpA.

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Subunit interface and protein environment of the D1 and D2 nucleotide-binding sites. ADP molecules are drawn as ball-and-stick models caged within the 2Fo-Fc electron density (forest green) at 1.2  level. For the ADP molecule, carbon atoms are black, nitrogen blue, phosphorus yellow, and oxygen purple. Residues interacting with the nucleotide are shown as stick models. Secondary structures are shown in purple for the ADP-binding subunit and in green for those of the neighboring symmetry-related D1; residues are labeled in red and green, respectively. a, stereo pair of the ADP site in D1. Eight positively charged and conserved residues from the neighboring D1 subunit are seen to interact with the ADP and its surrounding residues. Salt bridges and H bonds that are less than 3 Å are shown with dashed lines in gray. b, stereo pair of the ADP site in D2. No residues from the adjacent subunit make contact with the ADP, which is instead bounded by water molecules (W).

The ClpA structure also provides a clue to the dependence of hexameric assembly on ATP binding. Thr-323 and Asn-606 are found at the end of strand D of their respective domains in a position to interact with the -phosphate in the ATP state (Fig. 3); these hydrophilic residues each constitute part of Sensor 1 motif (8), which is suggested to contribute directly to nucleotide hydrolysis in some AAA⁺ proteins (46) and in others to distinguish between ATP and ADP in the binding pocket (47, 48). Thr-323 in D1 is connected by a short helix (D1-HE1) to a so-called SRH (second region of homology) motif that includes the "arginine-finger" residue, Arg-339, and other residues at the D1-D1 interface involved in detecting nucleotide states and in cooperative hexamerization with a neighboring subunit, suggesting an intra-subunit signal-transduction pathway. In D2, however, the intervening helix in a topologically equivalent position is the much longer ClpP-loop, and assembly of D2 ring does not directly provide an "arginine finger" from the adjacent subunit. Although the Arg-643 is in a topologically similar position as the putative D1 arginine-finger residue, Arg-339, it is >12 Å from the nucleotide-binding site and is unlikely to serve a catalytic or signaling function without a significant conformational switch. In this regard, D2 resembles HslU, in which the equivalent residue, Arg-325, is between 7 and 12 Å away from the -phosphate of the bound nucleotide and is not in a position to act catalytically (9). Our data is more consistent with D1-mediated hexamerization leading to stabilization of the active conformation of the D2 nucleotide-binding site.

Spatial Locations and Possible Functions of Conserved Residues-- Sequence alignment of ClpA family proteins reveals that the N-domains are the least conserved part of the protein, whereas the structural conservation as a pseudo dimer implies a preserved function for this region. The large sub-domains of the AAA⁺ modules are the most conserved regions, whereas the small sub-domains are moderately conserved. Many of the conserved residues within the ClpA family have not yet been assigned functional or structural roles, but their locations in the structure now provide clues as to possible functions (Fig. 1b). Residues 186-191 in D1 and 460-464 in D2 precede helix A and supply residues that interact with the amine group in position N6 of the bound nucleotide. Residues 201-210, including three positively charged residues, form part of the contact surface between the D1 domains; residues at equivalent positions of the D2 domain are less well conserved, suggesting that subunit interactions may be controlled separately or have different functional roles in D1 and D2. Two equivalent sequence regions located between the Walker A and B motifs, residues 250-266 of D1 and 520-542 of D2, form mobile loops that line the narrow passages in the modeled ClpA hexamer (see below, Fig. 5a). The passages resemble camera diaphragms and may open or close in response to changes in nucleotide state or the presence of substrates. The D1 and D2 diaphragms differ, implying that each makes a unique contribution to the control of protein translocation. In D2, the diaphragm contains the sequence GYVG, one of the most conserved motifs in Clp/Hsp100 proteins. This motif was identified as a possible pore forming motif in HslU (ClpY) (11). Residues Ile-617, Gly-618, and Leu-619 of D2 are conserved within the ClpA (and ClpX) family only in members that interact with a ClpP and are involved in activation of ClpP (25, 26). These residues are located on the D2 distal ring surface and are mobile in the ClpA structure. The apparent mobility of the ClpP-loop may reflect the need for flexibility in this interacting loop to allow more than one loop from hexameric ClpA to interact with its docking site on heptameric ClpP.

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Structural superposition of AAA+ modules. a, superposition of ClpA-D1 to ClpA-D2 domain where the ClpA-D1 is in blue and the ClpA-D2 is in red. b, alignment of ClpA-D1 to NSF-D2. The ClpA-D1 is in blue and the NSF-D2 is in yellow. c, the ClpA-D2 is superimposed to HslU (without the I-domain). ClpA-D2 is in red and HslU is in green.

AAA⁺ Modules of ClpA Subunit in Crystal Represent ADP-binding Conformation-- Structure superposition of D1 and D2 gives an overall root mean square deviation of 1.5 Å for 116 C atoms. The superposition of large domains between D1 and D2 required a 90° rotation of D1 relative to its orientation in the crystal (Fig. 4a); the ADP molecules in both domains are also superimposed well, indicating that the structural configuration of catalytic and nucleotide-binding residues are essentially identical in each domain. Although the loop connecting the large and small domains as well as helices D1-H1 and D2-H of the small domains are topologically similar, only 35 C atoms within the first portion of the small domains are superimposable with a root mean square deviation of 1.5 Å, and the small domains differ considerably in their respective C-terminal regions. The differences cause the two small domains to be oriented differently (6°) relative to their respective large domains and also affect the manner in which the small domains interact with the adjacent large domain in the hexamer. Structure alignments with known AAA⁺ modules demonstrate that ClpA-D1 is most similar to the NSF-D2 and p97-D1, whereas ClpA-D2 is strikingly similar to HslU with bound nucleotide (Fig. 4). These alignments provide a structural basis for constructing a ClpA hexameric model (below).

In full length ClpA, both D1 and D2 domain of ClpA demonstrate typical AAA⁺ folding, both have bound ADP molecules, and both are superimposable nicely to known nucleotide-bound AAA⁺ structures as well as to each other. We conclude that each AAA⁺ module of ClpA represents the ADP bound conformation. Such a conformation should occur as part of the catalytic cycle of ClpA and probably other AAA⁺ proteins. Hexamer stability precludes this conformation being present in all six subunits simultaneously, but such a condition would arise under energy-deprived conditions or when ADP and ATP exchange becomes a rate-limiting step in ATP hydrolysis, as suggested in kinetics studies of Hsp104 (46).

Construction of a Hexameric Ring Model of ClpA-- The D1 and D2 domains of ClpA are connected by a short six-residue hinge (Ile-436-Val-441) between helices D1-H and D2-HT1, which are present in most two AAA⁺ domain proteins analyzed. Two residues (Ile-436 and Pro-437) in the hinge are highly conserved in ClpA/Hsp104 family members. This short peptide segment is well ordered in the crystal and is expected to be rigid because of the proline and large side chains of the residues in the sequence. ClpA subunits form a hexameric spiral in the crystal; starting from the spiral conformation, a 15° rotation between D1 and D2 around this hinge would allow the subunits to be assembled into a planar hexagonal model (see below). In the crystal D1 and a sym-D1, and D2 and a sym-D2 have buried surface area of 2800 Å² and 2700 Å², respectively. In contrast interaction between D1 and D2 domains amounts to only 800 Å², supporting the notion that D1 and D2 are allowed to move with respect to each other.

To model the full length ClpA subunit into a hexagonal ring, we used NSF-D2 and HslU hexamers as templates for constructing the respective N-D1 and D2 hexameric rings on the basis of structural alignments. No changes were made to the structure of either domain or to the geometric relationship between large and small sub-domains. The resulting full length model satisfies the following additional criteria. First, the C terminus of N-D1 and the N terminus of D2 are in close proximity. Second, the model is consistent with limited proteolysis experiments (25). Third, the relative orientations of the large and small domains for D1 and D2 and the bound nucleotide were not changed from their crystal states. Finally, the maximum diameter of the hexamer is 170 Å and the height is 87 Å. Without the N-domain, the diameter is 130 Å, consistent with the 124 Å measured by EM, in which the N-domain is not visible.

Viewed down the 6-fold axis, ClpA is hexagonal with crenated edges produced by the N-domains and the D2 small domains (Fig. 5, a and b). A lateral view shows a two-layered structure similar to EM images, the larger layer being the N-D1 ring and the smaller layer being the D2 ring. D1 and D2 from the same subunit are inclined with respect to the 6-fold axis (Fig. 5c). The six subunits of ClpA hexamer enclose a vase-shaped central cavity with two spacious compartments connected by a narrow passage and an opening on each end (Fig. 5d). The constriction between the two compartments has its narrowest point midway through the D2 domain, so that the compartments are of unequal size. Compartment I, which includes D1 and a portion of D2 before the constriction, has an estimated volume of 110,000 Å,³ a value similar to that observed by EM (49). In D1, a narrow opening is located at the surface and should serve as an entrance pore for protein substrates. Compartment II is smaller, with an estimated volume of 40,000 Å³, and is somewhat funnel-shaped with a surface opening 60 Å in diameter. Compartment II opens to the ring surface that interacts with ClpP and represents part of the "vestibule" where stalled protein substrates have been observed in the ClpAP complex (24). The "ClpP loop," containing the sequence IGL, which has been implicated in activation of ClpP activity by ClpA (25, 26), was not visible in the crystal. However, residues 611 and 624, which flank the loop, are located on the distal surface of D2 at about 30-35 Å from the center (Fig. 5, b and d). A distinctive feature of the central cavity is the three highly negative potential belts separated by hydrophobic regions, which might contribute to the disruption of hydrogen bonding network of substrate proteins.

View larger version (60K): [in this window] [in a new window]   Fig. 5.   Hexameric model of ClpA. Electrostatic potential surface of the modeled planar ClpA hexagon as rendered in GRASP, with negative potential in red, positive in blue, and neutral in white. a, the hexagonal ring is viewed along the 6-fold axis with the D1 domains facing out. The hexagon has a crenated edge and maximum diameter of 170 Å. The larger crenations are made by the six N-domains, which are attached to the outer edge of the D1 domains; the smaller crenations are formed by extensions of the D2-small domains. b, the D2 side is facing out, showing the wide opening of the central cavity (red) and residues forming part of the ClpP loop (yellow). c, side view of the modeled ClpA hexagonal ring. The height is about 87 Å. The six subunits are shown in different colors. D1 and D2 from the same ClpA subunit are tilted with respect to the ring axis and make little contact with each other. Each domain makes extensive contacts with both D1 and D2 of a neighboring subunit. d, cross section through the center and parallel to the 6-fold axis of the modeled ClpA hexagonal ring. The surface of the central cavity is colored to show the three negatively charged belts (red) and the hydrophobic surfaces surrounding the channels (gray). The borders of the cavity are outlined in black. The two constrictions and the two compartments are as labeled. The positions for the three remaining ClpP loop are indicated in yellow.

A model for the conformational transitions in ClpA in solution and in crystal can be derived from differences in D1-D2 orientation, buried surface area, and electrostatic interaction between ClpA in the spiral and in the hexamer (Fig. 6). In solution, ClpA readily forms a hexamer in the presence of ATP but not in the presence of ADP. In fact, ADP causes hexamers to dissociate. Although overwhelming evidence supports the functional form of ClpA being a hexamer, the spiral form as observed in our crystal may be derived from packing of twisted (closed conformation, Fig. 6) ClpA monomers or dimers in solution with ADP present and partially reinforced into a perfect spiral by crystal-lattice contacts. In a functional scenario, when ATP is converted to ADP, a ClpA subunit could be somewhere between a closed and an open conformation by varying the orientation between D1 and D2. When ATP is sufficient, nucleotide exchange would prevent a complete transition from occurring, but slight rotation of D1 and D2 away from each other (toward the position in the spiral) could exert forces on bound substrates that either help unfold them or, if coupled to changes in channel and chamber residues, provide directional translocation. Although speculative, the model of a ClpA hexamer provides a framework for future studies of substrate recognition, unfolding, and translocation to ClpP.

View larger version (14K): [in this window] [in a new window]   Fig. 6.   A hypothetical model describing transitions of ClpA subunits in solution to form a spiral in crystal in the presence of ADP and to assemble into a planar hexamer in solution in the presence of ATP. ClpA subunits are postulated to undergo an open and a closed conformation by rotating D2 with respect to D1 via the hinge between two domains.

	ACKNOWLEDGEMENTS

We thank Dr. Z. Dauter for assistance in data collection at X9B of NSLS, Dr. F. Dyda for helping with the use the x-ray facility at National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health for crystal diffraction tests and heavy metal screening, Drs. R. Henning and Z. Ren for help at BioCars beamline at Advanced Photon Source, Drs. Z. Wawrzak and S. Weigand at Dow-Northwestern-Dupont beamline of APS for data collection, Dr. X. Gao for help in data collection at Synchrotron x-ray sources, and Drs. S. Gottesman and M. Gottesman for reading the manuscript and providing helpful comments.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The coordinates for the full length ClpA and N-domain (access codes 1KSF and 1K6K, respectively) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).

To whom correspondence should be addressed: 37 Convent Dr., Bldg. 37, Rm. 1B22, Bethesda, MD 20892. Tel.: 301-435-6315; Fax: 301-435-8188; E-mail: dixia@helix.hih.gov.

Published, JBC Papers in Press, September 28, 2002, DOI 10.1074/jbc.M207796200

² Guo, F., Esser, L., Singh, S. K., Maurizi, M. R., and Xia, D. (2002) J. Biol. Chem. 277, 46753-46762

³ L. Esser, unpublished work.

	ABBREVIATIONS

The abbreviations used are: NSF, N-ethylmaleimide sensitive factor; N-domain, N-terminal domain; ATPS, adenosine 5'-O-(thiotriphosphate); Hsp, heat shock protein.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				L. Zhu, J. O. Wrabl, A. P. Hayashi, L. S. Rose, and P. J. Thomas The Torsin-family AAA+ Protein OOC-5 Contains a Critical Disulfide Adjacent to Sensor-II That Couples Redox State to Nucleotide Binding Mol. Biol. Cell, August 1, 2008; 19(8): 3599 - 3612. [Abstract] [Full Text] [PDF]

				S. M. Doyle, J. R. Hoskins, and S. Wickner Inaugural Article: Collaboration between the ClpB AAA+ remodeling protein and the DnaK chaperone system PNAS, July 3, 2007; 104(27): 11138 - 11144. [Abstract] [Full Text] [PDF]

A. Shamaei-Tousi, J. P. Halcox, and B. Henderson Stressing the obvious? Cell stress and cell stress proteins in cardiovascular disease Cardiovasc Res,