Allostery in Hsp70 Chaperones Is Transduced by Subdomain Rotations

doi:10.1016/j.jmb.2009.01.062 J. Mol. Biol. (2009) 388, 475–490

Available online at www.sciencedirect.com

Allostery in Hsp70 Chaperones Is Transduced bySubdomain Rotations

Akash Bhattacharya, Alexander V. Kurochkin, Grover N. B. Yip,Yongbo Zhang, Eric B. Bertelsen and Erik R. P. Zuiderweg⁎

Department of BiologicalChemistry and BiophysicsResearch Division, TheUniversity of Michigan, 930North University Avenue, AnnArbor, MI 48109, USA

Received 5 January 2009;received in revised form27 January 2009;accepted 30 January 2009Available online4 February 2009

*Corresponding author. E-mail [email protected] addresses: G. N. B. Yip, U

Cameroon; Y. Zhang, NorthwesternIL, USA; E. B. Bertelsen, Assay DesigAnn Arbor, MI, USA.Abbreviations used: NBD, nucleo

SBD, substrate-binding domain; RDcoupling; NEF, nucleotide exchangeheteronuclear single quantum cohertransverse relaxation optimized spe3D, three-dimensional; PAS, princip

0022-2836/$ - see front matter © 2009 E

Hsp70s (heat shock protein 70 kDa) are central to protein folding, refolding,and trafficking in organisms ranging from archaea to Homo sapiens underboth normal and stressed cellular conditions. Hsp70s are comprised of anucleotide-binding domain (NBD) and a substrate-binding domain (SBD).The nucleotide binding site in the NBD and the substrate binding site in theSBD are allosterically linked: ADP binding promotes substrate binding,while ATP binding promotes substrate release. Hsp70s have been linked toinhibition of apoptosis (i.e., cancer) and diseases associated with proteinmisfolding such as Alzheimer's, Parkinson's, and Huntington's.It has long been a goal to characterize the nature of allosteric coupling in

these proteins. However, earlier studies of the isolated NBD could not showany difference in overall conformation between the ATP state and the ADPstate. Hence the question: How is the state of the nucleotide communicatedbetween NBD and SBD?Here we report a solution NMR study of the 44-kDa NBD of Hsp70 from

Thermus thermophilus in the ADP and AMPPNP states. Using the solutionNMR methods of residual dipolar coupling analysis, we determine thatsignificant rotations occur for different subdomains of the NBD uponexchange of nucleotide. These rotations modulate access to the nucleotidebinding cleft in the absence of a nucleotide exchange factor. Moreover, therotations cause a change in the accessibility of a hydrophobic surface cleftremote from the nucleotide binding site, which previously has beenidentified as essential to allosteric communication between NBD and SBD.We propose that it is this change in the NBD surface cleft that constitutes theallosteric signal that can be recognized by the SBD.

© 2009 Elsevier Ltd. All rights reserved.

Edited by M. F. Summers
Keywords: dnak; structure; dynamics; allostery; nmr
ess:

SA Peace Corps,University, Evanston,ns, 5777 Hines Drive,

tide-binding domain;C, residual dipolarfactor; HSQC,ence; TROSY,ctroscopy;le axes system.

lsevier Ltd. All rights reserve

Introduction

Hsp70 (heat shock protein 70 kDa) chaperoneproteins are central to protein folding, refolding, andtrafficking in organisms ranging from archaea toHomo sapiens under both normal and stressed cellularconditions.1 Hsp70s recognize misfolded substrateproteins by their exposed hydrophobic amino acids.In a process that requiresATPand two cochaperones,2

the Hsp70s unfold the substrate proteins by cycles ofbinding and release. The unfolded proteins thenrefold on their own accord. Much of the mechanismby which this takes places remains unknown.Recently, Hsp70s have been linked to cancer3–6 anddiseases associated with protein misfolding such asAlzheimer's,7 Parkinson's,8 and Huntington's.9,10 Ithas been suggested thatmodulation of Hsp70 activity

d.

476 Allostery and Subdomain Rotation in Hsp70

with small compounds may provide an avenue forthe treatment of these diseases.11

Hsp70s are comprised of two main domains: a 44-kDa N-terminal nucleotide-binding domain (NBD)and a 25-kDa substrate-binding domain (SBD) thatharbors the substrate binding site.Hsp70s are allostericmolecules: with ADP bound to the NBD, the substratebinds tightly to the SBD (Kd=1.5×10

−7 M); ATPbinding to the NBD promotes substrate release fromthe SBD (Kd=4.6×10

−5 M).12,13

While several high-resolution crystal and solutionstructures are available for theNBD14–16 and SBD17–22

subdomains of the Hsp70s of several species, no high-resolution structure has been reported for a completewild-type Hsp70. Hence, the allosteric mechanism isnot yet known. However, several recent studies haveshown that a conserved hydrophobic and flexiblelinker [residues 388VQDLLLLDVTP (Hsc70H. sapiens)and 380VRDVVLLDVTP (DnaK Thermus thermophi-lus)] loosely tethers the NBD and the SBD in theADP-substrate-bound state.23–25 In addition, it isanticipated that NBD and SBD are docked togetherin the ATP state,24,26 but this still needs to bedemonstrated for Hsp70s with an actual structure.Currently, the following overall model for theallosteric mechanism seems credible. With substratebound in the SBD and with ADP·Pi bound in theNBD, the SBD collides repeatedly with the NBD, butdoes not dock.25 Upon ATP binding to the NBD, a

Fig. 1. Stereograph of the superposition of five X-ray crystproteins were crystallized in the following states: wild-type APand 2QWL; chain A), wild-type ADP·VO4 (PDB 2QWM; choverlaid on the secondary structure elements of subdomain IAaccentuate potential changes in subdomain IB (residues 40–11purple), and subdomain IIB (residues 229–306; cyan). The N-domain is shown in yellow. The nucleotide, PO4

3−, Mg2+, andfigure was prepared with MacPyMOL.

change takes place in the NBD such that the collisionslead to stable docking of SBD and NBD. The dockingevent propagates through the SBD and leads to therelease of substrate. A key feature of this model is thatthe NBD conformational change upon ADP→ATPchange takes place while the SBD is loosely tethered.Hence, it is expected that the isolatedNBD should alsoshow such changes between different nucleotidestates.However, this notion is seriously challenged by

X-ray diffraction studies of the isolated NBD. Theyshow no obvious difference in the overall confor-mation of the NBD of bovine (identical withhuman) Hsc70 between the ATP state and theADP state14,27 (see Fig. 1) in the crystal. In addition,small-angle X-ray scattering methods could notdetect a difference in the radii of gyration for theATP and ADP states of the isolated NBD of Hsc70in solution,28 apparently supporting the crystal-lographic studies.We report here that subtle but extensive chemical

shift changes occur in the NMR spectra of theisolated NBD of Hsc70 between the ADP·Pi stateand the ATP state in neutral aqueous solution (seeFig 2a). Clearly, a change in nucleotide causeschanges in shifts that extend far from the nucleo-tide site. This indeed suggests that the isolatedNBD changes its conformation upon nucleotidechange in the absence of the SBD. The question is:

allography structures for bovine Hsc70 NBD (4–380). TheO (PDB 2QW9; chain A), wild-type ADP·PO4 (PDB 3HSCain A), and K71M-ATP (PDB 1KAX). The proteins were(Hsc70 count residues 1–39, 116–188, and 361–381; blue) to5; green), subdomain IIA (residues 189–228 and 307–360;terminus is shown in red, whereas the C-terminus of thistwo Na+ present in PDB 3HSC are shown in space fill. This

Fig. 2. (a) 15N–1H chemical shift differences between the ATP and ADP·Pi conformations of Hsc70-NBD. Orange:significant shift; green: no shift; gray: not known. Residues R171 (R167 in DnaK E. coli) and I181 (L177 in DnaK E. coli) arerendered as sticks. (b) 15N–1H chemical shift differences (Δδ) between the AMPPNP and ADP·Pi conformations of TTh-NBD. Red:ΔδN2σΔδ; orange: σΔδbΔδb2 σΔδ; yellow: 0.5σΔδbΔδbσΔδ; green: Δδb0.5σΔδ; gray: not known, where Δδ=(ΔδH

2+ΔδN2)1/2. ADP is shown in light blue, whereas PO4

3− is shown in dark blue. Residues R164 (R167 in DnaK E. coli)and L174 (L177 in DnaK E. coli) are rendered as sticks. See also Fig. 3. This figure was prepared with MacPyMOL.

477Allostery and Subdomain Rotation in Hsp70

What is the extent of the change? In order toanswer this question, we made the NBD of theHsp70 homologue DnaK of T.thermophilus29 (TTh-NBD) the focus of this study because the collectionof NMR data at an elevated temperature (50 °C)allows better-quality spectra than for the mesophi-

lic Hsc70. TTh-NBD is highly related to humanHsc70 (49% identity and 72% homology). Asshown in Fig. 2b, quite similar chemical shiftchanges occur in the NMR spectrum of TTh-NBDin the ADP and AMPPNP states, where the latter isa nonhydrolyzable ATP analogue.


In this work, we use residual dipolar coupling(RDC) analysis30,31 and discover that subdomain IIBin this species (see Fig. 1) rotates by as much as 20°between the AMPPNP state and the ADP state. Suchconformational changes have been observed before,but were thought to be caused by the binding of anucleotide exchange factor (NEF) such as BAG16 orGrpE,32 rather than by nucleotide change itself. Wealso discover that subdomains IA and IIA rotatesignificantly with respect to each other, opening(AMPPNP) and closing (ADP) a hydrophobic sur-face cleft between these subdomains. Significantly,this cleft area has been hypothesized to be ofimportance to the communication between NBDand SBD.23,24 The current work demonstrates that achange in this cleft occurs upon nucleotideexchange. It is likely that this is the allosteric changeon the surface of the NBD that is recognized by theSBD and/or the hydrophobic linker between NBDand SBD.

Results

The NBDs of Hsp70s are comprised of foursubdomains14 called IA, IB, IIA, and IIB, as definedin Fig. 1. The nucleotide ATP or ADP·PO4, togetherwith several cations, is located deep in the centralcleft and interacts with all four subdomains.AMPPNP binds at the same location. Several crystaland some NMR structures are available for severaldomains of the Hsp70s of several species (calledDnaK in bacteria and SS1 in fungi) molecules, but node novo structure is available for any domain of theDnaK of T. thermophilus.We used the construct33 of TTh-NBD containing

residues 1–381, giving rise to excellent 15N–1Hheteronuclear single quantum coherence (HSQC)-transverse relaxation optimized spectroscopy(TROSY) NMR spectra at 50 °C (see SupplementaryMaterials). The assignments of the backbone reso-nances for this construct were obtained previously.33

The NMR spectra of TTh-NBD in the ADP·PO4 andAMPPNP states show many chemical shift differ-ences (Fig. 2b). Amide 1H and 15N NMR assign-ments were confirmed for each state by analysis ofthree-dimensional (3D) HNCA-TROSY spectra. Theshift changes are subtle but significant, as shown inFig. 3. Mapping the shift differences on a TTh-NBDhomology model shows that the shifts extend farbeyond the location of the nucleotide (see Fig. 2b).The shifts disclose subtle but widespread changes instructure and/or dynamics between the states.At 44 kDa, TTh-NBD is too large for a high-

resolution structure determination by NMR. How-ever, the reorganization of subdomains can be probedwith excellent precision using RDC NMR analysis insolution.31 Briefly, themethod is as follows. The targetprotein is aligned in neutral aqueous buffer to whichPf1 phage is added.35 Rod-like phages align in themagnetic field, providing an anisotropic environmentfor the protein occupying the buffer between phagemolecules. The target protein aligns dynamically

about 0.2% of the time, allowing themagnetic dipolarcoupling of the amide nitrogen and hydrogen nucleito be measured by solution NMR techniques.36,37 Weuse modified HSQC-TROSY experiments that shiftthe resonances in the 15Ndimension proportionally to1JNH+1DNH. The comparison of those spectra withunshifted TROSY spectra reveals the RDCs36 (seeSupplementary Materials). This information is, inturn, used to obtain the orientation of the protein'ssubdomains with respect to the magnetic field,making use of the available high-resolution (X-ray)structures and homology models of the subdomains.Subsequently, this information is used to reconstructthe relative orientation of the subdomains. Subdo-main orientation is characterized by three axescorresponding to a rectangular parallelepiped, withthe longest axis defined as Szz, with the shortest axisdefined as Sxx, and with Syy in between.On average, 60 reproducible NH RDCs per

subdomain were obtained, with an uncertainty of~4 Hz. Only RDCs corresponding to residues thatare homologous between the NBDs of DnaK of T.thermophilus, DnaK of Escherichia coli, and Hsc70,Hsp70, and mt-Hsp70 of H. sapiens were used. Thisreduced the number of RDCs available for analysisto about 50 per subdomain (see Table 1). We used ahomology model for a reference structure of thesubdomains (see Materials and Methods).As shown in Fig. 3, our data is of sufficient quality

to define with precision the orientations of thedifferent subdomains for the two nucleotide states.In these so-called Sanson–Flamsteed globe plots, themain orientational axis Szz for all subdomains islocated approximately at 40° west and 5° south. Thisaverage location, by itself, is irrelevant, as it justreports the difference between the physical Szz axisorientation of the aligned protein and the arbitraryProtein Data Bank file z-axis direction of the modelstructure used. However, the differences betweenthe globe locations of the Szz axes of the differentsubdomains are highly relevant: they indicate thatnot all subdomains are oriented as in the modelstructure. Moreover, the relative orientations of thesubdomains change between the ADP state and theAMPPNP state. The spread in the Szz directions ofthe subdomains represents uncertainty in the rawdata, as determined by the Monte Carlo routine inthe REDCAT (residual dipolar coupling analysistool)34 program used, using RDCs correspondingto homologous positions without any further edit-ing. According to the REDCAT program, for theAMPPNP state, the subdomains are, within experi-mental error, oriented similarly to those in the modelstructure, which is based on bovine Hsc70 in theADP state. Surprisingly, this is not the case for theADP state, where the orientation of subdomain IIBhas moved away by approximately 20°. In addition,the relative orientations of the other subdomainschange as well.As a next step in the analysis, orientational

changes were modeled as changes in 3D structure.As explained in detail in Materials and Methods, themodels for the ADP and AMPPNP states were

Fig. 3. Globe graphs (Sanson–Flamsteed plots) showing the orientations and experimental uncertainties of the Szzprincipal alignment axes (around 60° west and 5° south) of the different subdomains of NBD of DnaK-TTh, as derivedfrom NMR RDC measurements using REDCAT.34 (a and b) RDC input: Conserved and experimentally reproducibleresidues. (a) Left: AMPPNP state. (b) Right: ADP·Pi state. Color codes: IA (1–37, 109–180, and 357–377; blue); IB (residues38–108; green); IIA (residues 181–219 and 307–356; purple), and IIB (residues 220–304; cyan). The reference structure forthe NBD of DnaK-TTh was modeled on the NBD of DnaK E. coli (PDB 1DKG) and bovine Hsc70.ADP·Pi (PDB 3HSC).Scale: horizontal, 20° per gridline; vertical, 10° per gridline. The Szz principal alignment axes also appear at the other sideof the globe (120° east and 5° north). The Sxx axes appear as smears around 150° west and 45° north, and 30° east and 45°south.


Table 1. RCDC calculations for the NBD of DnaK of T. thermophilus in 50 mM Hepes, 10 mM KCl, 5 mM MgCl2,5 mM sodium phosphate, and 20 mg/ml Pf1 phage (pH 7.2) at 50 °C (ADP or AMPPNP concentration: 10 mM)

Input RDCa NRDCb α β γ Szz Syy Sxx RMSDc (Hz) Qd

AMPPNP stateIA commone 45 161.8 87.6 63.1 −8.8E−04 8.6E−04 2.1E−05 8.1 0.65IA allf 58 161.2 86.9 61.8 −8.8E−04 8.1E−04 6.6E−05 9.8 0.73IA selfval averageg 34 160.2 87.4 61.9 −9.2E−04 8.6E−04 5.7E−05 9.1 0.68IA selfval RMSDg 4 6.2 4.5 5.2 6.3E−05 1.0E−04 7.4E−05 1.6 0.10IB common 26 136.4 76.4 70.0 −1.1E−03 7.5E 04 3.1E−04 10.5 0.67IB all 37 143.3 81.5 66.8 −8.5E−04 7.0E−04 1.5E−04 11.4 0.76IB selfval average 23 145.3 81.4 66.2 −9.0E−04 7.4E−04 1.6E−04 11.1 0.74IB selfval RMSD 3 26.3 5.3 4.5 6.4E−05 9.3E−05 8.6E−05 0.9 0.06IIA common 24 163.0 90.6 61.1 −8.5E−04 5.8E−04 2.6E−04 7.2 0.60IIA all 45 145.6 87.5 59.7 −9.1E−04 7.2E−04 1.9E−04 7.3 0.56IIA selfval average 27 144.6 88.0 58.7 −9.3E−04 7.4E−04 1.9E−04 7.0 0.53IIA selfval RMSD 4 6.9 5.9 4.6 5.1E−05 6.4E−05 8.9E−05 0.6 0.06IIB common 34 159.8 90.3 56.9 −1.1E−03 1.0E−03 4.0E−05 7.6 0.54IIB all 45 165.1 92.2 66.1 −7.9E−04 8.0E−04 −1.7E−05 9.2 0.71IIB selfval average 28 166.7 92.9 69.8 −8.1E−04 8.1E−04 1.0E−06 9.0 0.71IIB selfval RMSD 4 7.0 3.7 10.0 6.1E−05 6.6E−05 3.1E−05 1.2 0.10

ADP·Pi stateIA common 45 150.4 85.5 67.4 −1.1E−03 8.7E−04 1.9E−04 8.0 0.59IA all 66 155.3 86.7 68.3 −8.5E−04 6.8E−04 1.7E−04 8.8 0.70IA selfval average 41 154.4 87.1 68.2 −8.8E−04 6.9E−04 1.9E−04 8.5 0.67IA selfval RMSD 3 12.3 5.3 4.8 5.7E−05 9.5E−05 7.6E−05 0.6 0.05IB common 26 115.6 82.1 70.3 −9.5E−04 7.3E−04 2.2E−04 8.4 0.58IB all 38 107.2 85.2 71.4 −9.4E−04 6.0E−04 3.5E−04 9.0 0.66IB selfval average 24 165.4 86.7 71.8 −9.7E−04 6.7E−04 3.1E−04 8.5 0.62IB selfval RMSD 4 65.9 6.8 4.2 6.6E−05 8.0E−05 7.0E−05 1.3 0.08IIA common 24 180.5 78.2 66.2 −9.3E−04 6.1E−04 3.2E−04 5.5 0.43IIA all 43 175.8 83.4 72.1 −1.0E−03 7.4E−04 2.7E−04 6.7 0.47IIA selfval average 26 177.1 83.3 71.1 −1.0E−03 7.6E−04 2.4E−04 6.2 0.45IIA selfval RMSD 4 8.7 3.6 4.8 5.4E−05 6.1E−05 6.0E−05 0.6 0.05IIB common 34 161.5 87.5 50.1 −1.1E−03 1.1E−03 9.0E−05 7.9 0.51IIB all 50 159.1 80.1 48.4 −8.7E−04 8.7E−04 −1.5E−06 8.3 0.60IIB selfval average 30 156.2 79.9 50.1 −8.6E−04 8.5E−04 5.7E−06 7.9 0.58IIB selfval RMSD 4 7.6 5.2 6.3 5.6E−05 7.7E−05 6.4E−05 1.0 0.06

The calculations were carried out using an in-house-written grid search and optimization program, which optimized a, b, and γ (Eulerrotational angles in the z–y–z convention describing the orientation of the subdomain in the aligned protein with respect to thecoordinates of the reference structure based on DnaK E. coli andHsc70H. sapiens), overall alignment (Szz), and rhombicity [(Syy−Sxx)/Szz].

a The input RDC set used.b The number of RDCs in that set.c RMSD of the best fit between experimental and calculated RDC.d Q = RMSDffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPNRDC

i = 1

RDC ið Þexpð Þ2NRDC

se Consists of experimentally reproducible RDCs for NH groups of residues that are homologous between the following Hsp70s: DnaK

of T. thermophilus, DnaK of E. coli, and Hsc70, Hsp70, mt-Msp70, and Bip of H. sapiens. This set was further reduced by taking only thoseresidues for which RDC data were available in both the ADP state and the AMPPNP state.

f Consists of experimentally reproducible RDCs for NH groups of residues that are homologous between the following Hsp70s: DnaKof T. thermophilus, DnaK of E. coli, and Hsc70, Hsp70, mt-Msp70, and Bip of H. sapiens.

g Forty percent of the RDCs of footnote f were omitted at random for an ensemble of 30 calculations per subdomain. The mean andRMSD of the fitting parameters for these ensembles are given.


obtained by superposing the rotated subdomains onthe model structure for minimal positional RMSDusing translation of coordinates only.A comparison of the best-fit structures based on

all RDCs (as defined in the legend to Table 1) isshown in Fig. 4. Major shifts are seen for the relativeorientations of all domains between the twonucleotide states.In order to assess the statistical significance of the

differences seen in Fig. 4, we performed a “jack-knife” self-validation procedure. As shown inMaterials and Methods, it was determined that thecalculation of model structures obtained from RDC

calculations in which 40% of the RDC data wasdeleted at random yields an ensemble that repre-sents a reliable measure of the precision of structuredetermination.One of the ensembles is presented in Fig. 5, in

which subdomain IA is superimposed. The figuresclearly show that the average position of subdomainIIB in the ADP state (relative to the orientation ofsubdomain IA) lies outside of the ensemble ofpossible subdomain orientations in the AMPPNPstate. Figure S6 shows that the reverse is also true.Hence, the differences in orientation between thesestates, as seen in Fig. 4, are statistically highly

Fig. 4. Model of the ADP state of the NBD of DnaK-TTh (red) superposed on domain IA of the model for the AMPPNPstate (blue) calculated using all RDCs.


significant. The average position of subdomain IB inthe ADP state lies well inside the ensemble ofpossible subdomain IB positions in the AMPPNPstate. Hence, the differences in orientation betweenthese states, as seen in Fig. 4, are statistically notsignificant. The average position of subdomain IIAin the ADP state lies at the boundary of the ensembleof possible subdomain positions in the AMPPNPstate, and vice versa. According to t-statistics, there is

Fig. 5. The average of the TTh-NBD self-validation ensembto the self-validation ensemble of TTh-NBD ADP (salmon).

just a 3% probability that the observed difference inthe orientations of subdomains IIA in the differentnucleotide states is determined by chance (seeMaterials and Methods). Hence, the statisticalsignificance of the differences between subdomainsIIA and IIB, as seen in Fig. 4, is high.In summary, our data show that the subdomains

in the AMPPNP state are, within experimental error,similarly oriented as in the model structure, which is

le for AMPPNP (blue) and TTh-NBD ADP (red) compared

Table 2. Computed and experimental radii of gyration (Å)

ADP AMPPNP ATP

Computed values forDnaK-TTh-NBD

21.55 21.63

Experimental valuesfor Hsc70-NBD(Table 1 in Wilbanks et al.38)

22.8±0.2 22.6±0.1

Experimental values 23.1±0.1


based on an Hsc70 NBD crystal structure in the ADPstate (PDB 3HSC14). In contrast, in the ADP state,one observes statistically significant deviations by asmuch as 20° for the Szz axes of subdomains IIB. Inaddition, we observe a statistically significantrotation of subdomain IIA with respect to subdo-main IA between the different states, as disclosed inFig. 4.

for Hsc70-NBD(Table 2 in Wilbanks et al.38)

The experimental radii of gyrationwere taken fromWilbanks et al.38
Discussion
The RDC data report on the orientations andchanges in orientations of the NBD subdomains. Thedata give no information on whether the subdo-mains shift in a lateral way with respect to eachother. Therefore, it is important to emphasize thatthe structures shown in the figures are just con-venient models; they show the orientationalchanges, but changes in translation have beenminimized by superposition. The models have notbeen further minimized and contain areas in whichatoms clash (see also Materials and Methods).Figure 4 shows that domain IIB rotates about 20°

clockwise going from the AMPPNP state to theADP·Pi state. As shown in Fig. 5, this change issignificant within the accuracy of the experimentaldata. This movement is the likely explanation for thechemical shift changes in residues in domains IB andIIB that line the nucleotide binding cleft (see Fig. 2).We suggest that the origin of this rotation is asfollows. In the AMPPNP state, the left and righthalves of the protein are bridged by the rigidnucleotide mimic. For ADP·Pi, however, the γ-phosphate bond is hydrolyzed, which breaks themolecule and thus the bridge between left and right,allowing a rotation of IIB. It appears that the left andright halves can move relatively independently inthe ADP·Pi state. Possibly, the ADP·Pi structure canopen dynamically even further than displayed in themodels.Available crystal structures of the isolated NBD

correspond closely to each other regardless of theirnucleotide state. This is illustrated in Fig. 1, whichis an overlay of five crystal structures of the NBDof bovine Hsc70 in APO, ADP (twice), ADP-V2O5state, and the mutant K71M complexed with ATPfrom two different laboratories. Small-angle X-rayscattering measurements38 showed no significantdifference in the radii of gyration for the ADP andATP states of the human Hsc70 NBD fragment (seeTable 2). This appears to confirm that the con-formations of these states are also identical insolution. However, there is also no difference in thecomputed radii of gyration for the AMPPNP andADP structures of TTh-NBD as derived from theRDC data, despite the rather large (visual) differ-ences (Table 2).As it turns out, the X-ray structures of the isolated

NBD all correspond much more closely to theAMPPNP state of TTh-NBD in solution then to thesolution ADP state (Fig. 6a). The fit between TTh-NBD in the ADP state and Hsc70 in the ADP state is

inferior (Fig. 6b), mostly because of the clockwiserotation of IIB. The rotation of subdomain IIB in theADP state in solution is reminiscent of the rotationof this domain upon binding of a NEF. Figure 6cshows a superposition of the crystal structures ofbovine Hsc70-NBD with and without its NEF BAG-1.16 Similar rotations were observed for E. coli DnaKNBD upon binding to its NEF GrpE39 (data notshown) and in the very recent cocrystal structures ofyeast Hsp70 NBD with Hsp110 as exchangefactor40,41 (data not shown). Hence, the rotatedstate that exists in solution for the ADP state ofDnaK TTh-NDB can also be observed in the crystal,provided it is stabilized by a NEF.However, an earlier notion indicating that it is the

interaction between Hsp70 and NEF that forces aninduced fit that actively promotes nucleotideexchange32 should likely be modified. It seemsmore appropriate to assume that the NEF selectivelycaptures and stabilizes the ADP state. The nucleo-tide exchange process in the Hsp70 chaperones(abundant ATP replacing ADP) in this model is thencatalyzed by other processes, which may include theinteraction of GrpE's tail with the SBD, as has beensuggested before.39 Selective interaction of the NEFwith the ADP conformation also makes sense from afunctional point of view: it avoids unnecessaryinterference with the ATP state and provides adriving force for NEF release when the nucleotidehas been exchanged from ADP to ATP. The processof selective capture that we suggest here to be activefor the chaperone NEF interaction has recently beenrecognized as a common interaction paradigm forseveral other allosteric proteins42 and nucleicacids.43

The chemical shift changes in Fig. 2 suggest thatmajor changes take place in domains IA and IIAwhen changing from the AMPPNP state to theADP·Pi state. The globe plots in Fig. 3 and themodels based on these data in Fig. 4 suggest thatthese changes are due to a clockwise rotation ofdomain IIA with respect to domain IA when goingfrom the AMPPNP state to the ADP state. Figure 5proves that such a rotation is significant, with a 97%probability.This is the first time that such a conformational

change has been demonstrated for nucleotidechange in a single Hsp70 species. But as is illustratedin Fig. 7, similar differences are present between the


published structures of different species in differentnucleotide states. For instance, a crystal structure fora yeast Hsp110 dimer, locked in the ATP state, wasrecently solved.26 Hsp110 has been recognized as aprotein with a strong Hsp70 homology, but itfunctions as a NEF rather than as a chaperone. Itwas suggested that the structure of this relatedprotein might provide insight into the structure ofan Hsp70 protein in the ATP state. If one associatesthe structure of Hsc70 NBD in complex with its NEF

Fig. 6. Comparison of structures. The figures were made belements of domains IA. (a) Hsc70 bovine (PDB 3HSC; green)NBD-TTh with ADP (red). (c) Hsc70 bovine and Hsc70 bovinshown). This figure was prepared with MacPyMOL.

as a true representative of the ADP state, andHsp110 as a true representative of the ADP state,ADP–ATP change becomes quite apparent. Figure7a shows a dramatic clockwise rotation of domainIIA of Hsc70 (complexed with BAG) as compared toHsp110 structure. Remarkably, these intermolecularADP–ATP changes are very similar to the intramo-lecular changes seen for TTh-NBD, as shown in Fig.4 (which is reproduced for convenience in Fig. 7b).That conformational changes are possible in the

y superposing the Cα positions of the secondary structureand NBD-TTh with AMPPNP (blue). (b) Hsc70 bovine ande complexed with BAG (PDB 1HX1; orange; BAG is not

Fig. 6 (legend on previous page)


Hsc70 NBD has been suggested by us before.44

However, our experimental data at that time were ofinsufficient quality to disclose what the changesentailed.At this point of the discussion, it is worthwhile to

take stock. Our studies have shown that significantconformational changes occur in the NBD of DnaKof T. thermophilus upon nucleotide change. This isconfirmed by and does explain the observedchemical shift changes for this domain in itsdifferent states. It is of importance that similarchemical shift changes were observed between theADP state and the ATP state of the Hsc70 NBD,confirming that similar conformational changes takeplace for the different species. Analysis of the RDCdata and subsequent modeling discloses significantand substantial clockwise rotations for domains IIAand IIB when changing from the AMPPNP state tothe ADP state (relative to IA). Because of thesefindings, we can now be confident that thedifferences seen between structures of differentspecies in different nucleotide states are most likelydue to changes in nucleotide, and not to changes inspecies. Hence, those differences may now beinterpreted in terms of conformational changesand the mechanism of allostery. The remainder ofthis discussion is a first attempt.The relative rotations of subdomains IA and IIA are

of great interest, since they affect a hydrophobicsurface cleft between them (see Fig. 8), which likely isinvolved in linker binding and substrate domainbinding.23,24 In Hsp110, this surface cleft is open andoccupied by the linker between theNBD and the SBD.The cleft is closed in the ADP state, and the linkercannot be docked. It has been shown before for DnaKof E. coli that the linker moves freely in the ADP stateand allows the SBD to move relatively freely,24,25

while the linker becomes immobilized in the ATPstate.24 Several biochemical studies also showed thatthe linker is exposed in the ADP state, and not in theATP state.23,45,46 It has been speculated24 that changesin the surface cleft between domains IA and IIA areresponsible for the docking in the ATP state, but it isonly in our current work that this is substantiatedwith an actual observation of change in that area for asingle species.On the basis of the current work, literature reports

on a confounding large number of Hsp70 species,states, complexes, and artificial truncations, studiedby different experimental techniques, can now bemerged into the following general allosteric model: inthe ATP-substrate-free state, NBD and SBD aredocked. The NBD is relatively rigid.45,47 ATP hydro-lysis leads to a loosening of the junction between the“left” (subdomains IA and IB) and the “right”(subdomains IIA and IIB) of the NBD. Overall, theNBD has become more flexible.45,47 In this process,subdomain IIA rotates clockwise and closes the IA/IIA surface cleft. Next, the linker is expelled, and theSBD dissociates from the NBD. We suggest that theATP hydrolysis energy is spent on this step tocompensate for the lost SBD–NBD binding energy.In the SBD, the dissociation from the NBD istransduced through the β-sheet21 and affects thesubstrate-binding cleft, which rigidifies. Meanwhile,subdomain IIB rotates clockwise as well and ispredisposed to binding to the NEF, which furtherstabilizes the open ADP state and promotes nucleo-tide exchange. The rebinding of ATP causes NBD andSBD to dock and expels the substrate.We suggest thatthe regained SBD–NBD docking energy compensatesfor the lost SBD-substrate binding energy.While this scenario is likely to be correct overall,

many of the details are still missing. For instance, we

Fig. 7. Comparison of structures. The figures were made by superposing the Cα positions of the secondary structureelements of domains IA. (a) Hsc70 bovine complexed with BAG (PDB 1HX1; orange; BAG not shown) and Hsp110 (PDB1QXL; NBD of one monomer is shown in brown). (b) DnaK-TTh-NBDmodel in the AMPPNP state (blue) and DnaK-TTh-NBD model in the ADP state (red). This figure was prepared with MacPyMOL.


do not really knowhow theNBD/SBD linker binds inthe ATP state for the true Hsp70s. The Hsp110model26 is useful, but there is only marginal sequencehomology: true Hsp70s have fully hydrophobiclinkers such as 388VQDLLLLDVTP (Hsc70H. sapiens),380VRDVVLLDVTP (DnaK T. thermophilus), and387TQDILLLDVAP (Candida albicans SSA-1), whileyeast Hsp110 has 389VRPFKFEDIH. Moreover,

Hsp110 is locked in the ATP state, and the divergentlinker may actually be the cause of its inability tochange conformation.Being fully aware of the limits on the precision of

current NMR structure calculations and models, wetentatively show in Fig. 8a a comparison of the IA/IIA cleft region for the ADP and AMPPNP states ofDnaK T. thermophilus. The surface residues visible on

Fig. 8. View of the IA/IIA interface of DnaK-TTh. This view is a 90° rotation around the horizontal axis, as comparedto the other figures. It shows the “bottom” as compared to these other figures. (a) In the ADP state. (b) In the AMPPNPstate. Color codes: in domain IA (left): hydrophobics, light green; positive, light blue; negative, purple; polar, gray; indomain IIA (right): hydrophobics, green; positive, blue; negative, red; polar, white. The C-terminus (residue 372 in thisconstruct) is rendered in cyan. Domain IB (top left) is shown in gray, and domain IIB (top right) is shown in black.


this side of the protein are completely conserved inover 300 sequences of archaea, bacteria, and animalschecked. The hydrophobic nature of the cleft is clear.Access to the cleft becomes hindered in the ADPstate. This is especially obvious for Leu174 (renderedin yellow), which forms the bottom of the cleft. In theTROSY spectrum of NBD-TTh, L174 shows a largeshift upon nucleotide exchange. Its position is alsoindicated in Fig. 2b. The homologous residue, I181 inHsc70, also shifts (Fig. 2a). NMR chemical shiftswerealso observed for this residue (L177) in DnaK of E.coli upon nucleotide exchange.24

Mutagenesis studies23 strongly suggest that theuniversally conserved R167 of DnaK E. coli interactswith the universally conserved D393 of the linkersequence in the ATP state. The NMR shift data inFig. 2 and the surface view in Fig. 8 show that thisresidue is unaffected by nucleotide exchange(R164 and R171 in DnaK T. thermophilus and HscBos taurus, respectively). This is not surprising sincethe constructs studied here are truncated beforethe linker sequence. However, the finding suggeststhat this arginine residue is “passive” and justserves to steer the sense of linker insertion. This


sense corresponds to the insertion of the linker inHsp110.26

Materials and Methods

Samples

DnaK-TTh 1–381 (DnaK-TTh NBD), cloned into Pet22-b(Novagen) with an N-terminal His-tag, was expressed in E.coli BL21 cells at 37 °C. Expression was induced atOD600=0.5 in a triple-labeled M9 medium containing 98%D2O, [13C]glucose, and [15N]ammonium chloride. Cellswere harvested by spinning down at 15,000g, resuspended,and subsequently lysed using a microfluidizer. The proteinwas purified in two steps using Ni-NTA agarose with animidazole gradient and FastFlo Q− ion exchange at pH 7.2with a linear KCl gradient.47 The purified protein wasextensively dialyzed against NMR buffer (see below) andconcentrated using Centricon microfilters.

NMR experiments

NMR samples contained 0.15–0.2 mM protein in 50 mMHepes (pH 7.4), 10 mM KCl, 5 mM MgCl2, and 5 mMsodium phosphate. The ADP or AMPPNP concentrationwas 10 mM. Experiments were performed at 50 °C on an800-MHz Varian Inova spectrometer, using a triple-resonance cold probe. Backbone resonance assignmentswere obtained from a single 3DHNCA-TROSYexperimentfor each nucleotide state, using a previously obtained peaklist for DnaK-TTh in theADP-AlFx state as template.33 Twohundred eighty-one and 310 assignments were obtainedfor the ADP and AMPPNP forms, respectively.For RDC measurements, Pf1 bacteriophage in the

aforementioned buffer was added to the NMR samplesto a concentration of 20 mg/ml for partial alignment.48

The 2H2O quadrupolar splitting was 8 Hz at these cir-cumstances. RDCs were extracted from a series oftwo-dimensional TROSY experiments with a [κt1/2–180(N,H)−κt1/2] sequence49 at the beginning of the 15Nchemical shift labeling period, with κ=0, 0.75, and 1.5. Thismethod is better suited for larger proteins than the IP/APanalysis50 or the HSQC-TROSY comparison,51 since thismethod retains the full TROSY narrowing in the 1Hdimension and the partial TROSY narrowing in the 15Ndimension. The two-dimensional NMR spectra were recor-ded at 50 °C. Ten, 20, and 40 h of data acquisition wereused for the κ=0, 0.75, and 1.5 experiments, respectively,for both the ADP state and the AMPPNP state.

Data analysis

The experiments yield TROSY spectra in which the 15Ncoordinate of each cross-peak is shifted by κ⁎(1JNH+

1DNH)/2 (see Fig. S2). Using a uniformvalue of 1JNH (90.5Hz), 1DNHwas extracted from the differences between κ=0 andκ=0.75spectra and between κ=0 and κ=1.5 spectra. The expecteddifferences in real 1JNH (±2 Hz) for the different residues aresignificantly less than the experimental precision in the data(±4 Hz in the 15N dimension) for this large protein.The available RDC data for each state were trimmed

down into two successive steps. First, the RDCs obtainedfrom the differences between κ=0 and κ=0.75 spectra andbetween κ=0 and κ=1.5 spectra were compared. Those forwhich the scaled differences were larger than the scaledRMS difference between these values were rejected.The scaled difference was defined as: difference=abs

[2⁎(value1−value2)/(value1+value2)]. The remainingRDCs were labeled as “experimentally sound.” Second,only the RDCs were retained for NH groups of residuesthat are homologous between the following Hsp70s: DnaKof T. thermophilus, DnaK of E. coli, and Hsc70, Hsp70, mt-Hsp70, and Bip of H. sapiens.The remaining RDC data, termed experimentally and

homology sound, were used as inputs to orient subdo-mains IA, IB, IIA, and IIB of the DnaK-TTh homologymodel. It was assumed that each subdomain was a rigidunit by itself. While most of the data shown in this workare based on these relatively unedited RDC input data,several calculations were performed with a subset of RDCdata in which only those residues for which RDC datawere available in both the ADP state and the AMPPNPstate were retained. The results of these calculations(shown in Fig. 7b) show that the differences in conforma-tions are essentially the same as for the larger RDCdataset, establishing that the differences are not due to theselection of the RDCs.We used in-house-written programs and REDCAT34 to

transform the RDCs into orientational data. REDCAT'ssolution algorithm relies on singular value decompositionand Monte Carlo error estimation to generate an ensembleof 1000 structures compatible with the input structuresand the set of RDCs provided, based on an experimentalerror range of 4 Hz. The in-house-written Fortran programis based on a grid and minimization search of all possibleEuler rotations, overall alignment, and rhombicity to findthe best fit to the experimental data. The results areidentical with the REDCAT solutions. The structural datashown in this report were all derived from the results ofthe Fortran program, since it gave easier access to thedistribution of Euler angles.The orientations of the subdomains were computed from

the structures of the subdomains. Since no coordinates areavailable for NBD-TTh, we used a homology referencemodel. It was constructed in two steps. First, we threadedthe sequence of TTh-NBD on the coordinates of a crystalstructure of the NBD of DnaK E. coli, the protein mosthomologous to DnaK T. thermophilus for which coordinatesare available32 (PDB 1DKG, with a resolution of 2.8 Å).However, this structure was obtained in the presence of aNEF, which likely rearranged the subdomain orientations.Several crystal structures14 without cochaperones areavailable for the bovine Hsc70 NBD, in which thesubdomain orientations, especially for IIB, are differentfrom those seen in PDB 1DKG. In order to obtain therequired reference model for NBD-TTh, we adjusted theorientations of the subdomains in the model to correspondwith those in PDB 3HSC, a structure of Hsc70-NBDnominally found in the ADP state at a resolution of 1.90 Å.The adjustment was based on a superposition of the Cα

atoms of the secondary structure elements in the subdo-mains. The quality of the superposition of the TTh-NBDreference model with PDB 3HSC can be seen in Fig. S6.

Self-validation

An alternative estimation of the significance of thedifferences observed in Fig. 5 can be provided by a“jackknife” self-validation procedure. In self-validation, acertain fraction of the data is omitted at random, and theremaining data are then fitted to the relevant mathema-tical model. This process is carried out many times, and adistribution in fitting parameters is obtained. In order toassess the relationship of the self-validation data retentionpercentage, the distribution in the fitting parameters, andthe real error, we carried out the following model

†http://en.wikipedia.org/wiki/Student%27s_t-test‡http://changingminds.org/explanations/research/

analysis/t-test_table.htm


calculation. To a third-order polynomial function, randomnoise was added to an amplitude that allowed the data tobe fitted back to a third polynomial with an R2 of 0.65 (seeFig. S7). This was repeated 10 times to generate 10independent rather noisy “data sets.” The distribution inthe 10 sets of fitted parameters was taken as a measure ofachievable precision due to the “real” noise in the “data”(see Table S1). Subsequently, one of the “data sets” wasanalyzed using self-validation. When 60% of this single“data set” is kept at random and fitted back to a thirdpolynomial, and doing this 10 times, one ends up with asimilar distribution in the 10 sets of fitted parameters as forthe “real” noise (see Table S1). Hence, we view self-validation retaining 60% of the data as a realistic measureof the influence of measurement error on the precision ofthe fitted parameters.Figure S8 shows the results of 100 self-validations per

domain using 60% of the RDC data selected at randomand computed with the REDCAT program. Even whenusing only 60% of the data, one obtains statisticallysignificant differences in subdomain orientations ascompared to the model structure.

Molecular models

We obtained the molecular models for each of the statesin five steps:

1. Rotate each subdomain into its Principle Axes System, (PAS)based on the average Euler angles obtained from the 60% re-tained self validation procedure (obtaining IA-PAS, IIA-PAS,IB-PAS, IIA-PAS).2. Rotate the reference structure into the PAS of domain IA(TThREF

PAS-IA).3. Optimize the superposition of the CA's in the 2nd structureof each PAS subdomain onto the corresponding subdomain ofTThREF

PAS-IA, using translation only.4. Reassemble the domains into a single structure.5. Check that this process has not altered the relativesubdomain orientations by carrying out a REDCAT computa-tion on the obtained models. Zero-degree Euler angles shouldbe and were obtained.

The obtained model structures are available in PDBformat in Supplementary Materials.The model of TTh-NBD in the AMPPNP state was used

as a template for all superpositions of the structures shown.Corresponding Cα atom positions in secondary structureelements of subdomain IAwere used for all overlays.

Ensemble of RDC structures

Using the in-house-written Fortran programs, wecalculated self-validations at the 60% level (retained) foran ensemble of 30 structures for the ADP state and theAMPPNP state (see Table 1). The RMSDs of the three Eulerangles were computed from these ensembles, as shown inTable 1. Next, for each subdomain, we calculated amolecular structure, as described above, for each of eightdifferent sets of Euler angles: ⟨α⟩±RMSDα, ⟨β⟩±RMSDβ,and ⟨γ⟩±RMSDγ, as shown in Table 1.

Statistics of the ensemble

The 30 structures in the ensembles are represented by 8structures describing the boundaries of the ensemble, like theeight vertices of a cube. These boundaries represent theaverage RMSD multiplied by

ffiffiffi3

p. While the ensemble

contains 30 structures, it should actually be seen as anensemble of just 6 independent structures, since it takes fiveRDCs to determine an orientation (the fourfold degeneracyin orientation is not an issue here). Under the assumptionthat this distribution of 6 structures can be described as aGaussian around an average, t-statistics can be used.In the present case, the average orientation of sub-

domain IIA in the ADP state lies at the edge of a similarensemble of that domain in the AMPPNP state, and viceversa (Fig. 6). The t-statistic for similar ensembles isdefined as:†

t1;2 =hX1i � hX2irX1X2

ffiffiffi2N

qwhere hX1i and hX2iare the averages of the ensembles, andN is the number of elements in each ensemble (the same),and where:

rX1X2 =

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffir2X1

+ r2X2

2

s

with rX1X2 ir2X2

� �as the squared standard deviation of

the distributions.For the present case, with hX1i � hX2i = rX1X2

ffiffiffi3

pand

N=6, one finds t1,2=3, which means that there is just a 3%probability that the observed difference in orientations forsubdomains IIA between the two states is determined bychance.‡

Acknowledgements

We thank the W. F. Keck Foundation, the NationalScience Foundation, and the National Institutes ofHealth for providing funds for the purchase of an800-MHz spectrometer and cryogenic probe. Theresearch was supported by National Institutes ofHealth grants GM63027 and NS059690-01A1.

Supplementary Data

Supplementary data associated with this articlecan be found, in the online version, at doi:10.1016/j.jmb.2009.01.062

References

1. Bukau, B. & Horwich, A. L. (1998). The Hsp70 andHsp60 chaperone machines. Cell, 92, 351–366.

2. Schroder, H., Langer, T., Hartl, F. U. & Bukau, B.(1993). DnaK, DnaJ and GrpE form a cellularchaperone machinery capable of repairing heat-induced protein damage. EMBO J. 12, 4137–4144.

3. Garrido, C., Brunet, M., Didelot, C., Zermati, Y.,Schmitt, E. & Kroemer, G. (2006). Heat shock proteins27 and 70: anti-apoptotic proteins with tumorigenicproperties. Cell Cycle, 5, 2592–2601.


4. Nylandsted, J., Brand, K. & Jaattela, M. (2000). Heatshock protein 70 is required for the survival of cancercells. Ann. N.Y. Acad. Sci. 926, 122–125.

5. Wadhwa, R., Kaul, S. C. & Mitsui, Y. (1994). Cellularmortality to immortalization: mortalin. Cell Struct.Funct. 19, 1–10.

6. Wadhwa, R., Takano, S., Kaur, K., Deocaris, C. C.,Pereira-Smith, O. M., Reddel, R. R. & Kaul, S. C.(2006). Upregulation of mortalin/mthsp70/Grp75contributes to human carcinogenesis. Int. J. Cancer,118, 2973–2980.

7. Gestwicki, J. E., Crabtree, G. R. & Graef, I. A. (2004).Harnessing chaperones to generate small-moleculeinhibitors of amyloid beta aggregation. Science, 306,865–869.

8. Chung, K. K. & Dawson, T. M. (2004). Parkin andHsp70 sacked by BAG5. Neuron, 44, 899–901.

9. Krobitsch, S. & Lindquist, S. (2000). Aggregation ofhuntingtin in yeast varies with the length of thepolyglutamine expansion and the expression ofchaperone proteins. Proc. Natl Acad. Sci. USA, 97,1589–1594.

10. Novoselova, T. V., Margulis, B. A., Novoselov, S. S.,Sapozhnikov, A. M., van der Spuy, J., Cheetham, M. E.& Guzhova, I. V. (2005). Treatment with extracellularHSP70/HSC70 protein can reduce polyglutaminetoxicity and aggregation. J. Neurochem. 94, 597–606.

11. Brodsky, J. L. & Chiosis, G. (2006). Hsp70 molecularchaperones: emerging roles in human disease andidentification of small molecule modulators. Curr. Top.Med. Chem. 6, 1215–1225.

12. Mayer, M. P., Rudiger, S. & Bukau, B. (2000).Molecular basis for interactions of the DnaK chaper-one with substrates. Biol. Chem. 381, 877–885.

13. McCarty, J. S., Buchberger, A., Reinstein, J. & Bukau,B. (1995). The role of ATP in the functional cycle of theDnaK chaperone system. J. Mol. Biol. 249, 126–137.

14. Flaherty, K. M., Deluca-Flaherty, C. & McKay, D. B.(1990). 3-Dimensional structure of the ATPase frag-ment of a 70k heat-shock cognate protein. Nature, 346,623–628.

15. Flaherty, K. M., Wilbanks, S. M., DeLuca-Flaherty, C.& McKay, D. B. (1994). Structural basis of the 70-kilodalton heat shock cognate protein ATP hydrolyticactivity: II. Structure of the active site with ADP orATP bound to wild type and mutant ATPasefragment. J. Biol. Chem. 269, 12899–12907.

16. Sondermann, H., Scheufler, C., Schneider, C., Hohfeld,J., Hartl, F. U. & Moarefi, I. (2001). Structure of a Bag/Hsc70 complex: convergent functional evolution ofHsp70 nucleotide exchange factors. Science, 291,1553–1557.

17. Zhu, X. T., Zhao, X., Burkholder, W. F., Gragerov, A.,Ogata, C. M., Gottesman, M. E. & Hendrickson, W. A.(1996). Structural analysis of substrate binding by themolecular chaperone DnaK. Science, 272, 1606–1614.

18. Morshauser, R. C., Wang, H., Flynn, G. C. &Zuiderweg, E. R. (1995). The peptide-binding domainof the chaperone protein Hsc70 has an unusualsecondary structure topology. Biochemistry, 34,6261–6266.

19. Wang, H., Pang, Y., Kurochkin, A. V., Hu, W., Flynn,G. C. & Zuiderweg, E. R. P. (1998). The solutionstructure of the 21 kDa chaperone protein DnaKsubstrate binding domain: a preview of chaperone–protein interaction. Biochemistry, 37, 7929–7940.

20. Morshauser, R. C., Hu, W., Wang, H., Pang, Y., Flynn,G. C. & Zuiderweg, E. R. (1999). High-resolutionsolution structure of the 18 kDa substrate-binding

domain of the mammalian chaperone protein Hsc70.J. Mol. Biol. 289, 1387–1403.

21. Pellecchia, M., Montgomery, D. L., Stevens, S. Y.,Vander Kooi, C. W., Feng, H. P., Gierasch, L. M. &Zuiderweg, E. R. (2000). Structural insights intosubstrate binding by the molecular chaperone DnaK.Nat. Struct. Biol. 7, 298–303.

22. Stevens, S. Y., Cai, S., Pellecchia, M. &Zuiderweg, E. R.(2003). The solution structure of the bacterial HSP70chaperone protein domain DnaK(393–507) in complexwith the peptide NRLLLTG. Protein Sci. 12, 2588–2596.

23. Vogel, M., Mayer, M. P. & Bukau, B. (2006). Allostericregulation of Hsp70 chaperones involves a conservedinterdomain linker. J. Biol. Chem. 281, 38705–38711.

24. Swain, J. F., Dinler, G., Sivendran, R., Montgomery,D. L., Stotz, M. & Gierasch, L. M. (2007). Hsp70chaperone ligands control domain association via anallosteric mechanism mediated by the interdomainlinker. Mol. Cell, 26, 27–39.

25. Bertelsen, E. B., Chang, L., Gestwicki, J. E. &Zuiderweg, E. R. (2009). Solution conformation ofwild-type E. coli Hsp70 (DnaK) chaperone complexedwith ADP and substrate. Proc. Natl Acad. Sci. USA, inpress. doi:10.1073/PNAS.0903503106.

26. Liu, Q. & Hendrickson, W. A. (2007). Insights intohsp70 chaperone activity from a crystal structure ofthe yeast Hsp110 Sse1. Cell, 131, 106–120.

27. O'Brien, M. C., Flaherty, K. M. & McKay, D. B. (1996).Lysine 71 of the chaperone protein Hsc70 is essentialfor ATP hydrolysis. J. Biol. Chem. 271, 15874–15878.

28. Wilbanks, S. M., Chen, L. L., Tsuruta, H., Hodgson,K. O. & Mckay, D. B. (1995). Solution small-angle X-ray-scattering study of the molecular chaperone Hsc70and its subfragments. Biochemistry, 34, 12095–12106.

29. Klostermeier, D., Seidel, R. & Reinstein, J. (1998).Functional properties of the molecular chaperoneDnaK from Thermus thermophilus. J. Mol. Biol. 279,841–853.

30. Tolman, J. R., Flanagan, J. M., Kennedy, M. A. &Prestegard, J. H. (1995). Nuclear magnetic dipoleinteractions in field-oriented proteins: information forstructure determination in solution. Proc. Natl Acad.Sci. USA, 92, 9279–9283.

31. Fischer, M. W., Losonczi, J. A., Weaver, J. L. &Prestegard, J. H. (1999). Domain orientation anddynamics in multidomain proteins from residualdipolar couplings. Biochemistry, 38, 9013–9022.

32. Harrison, C. J., Hayer-Hartl, M., Di Liberto, M., Hartl,F. & Kuriyan, J. (1997). Crystal structure of thenucleotide exchange factor GrpE bound to the ATPasedomain of the molecular chaperone DnaK. Science,276, 431–435.

33. Revington, M. & Zuiderweg, E. R. (2004). TROSY-driven NMR backbone assignments of the 381-residuenucleotide-binding domain of the ThermusthermophilusDnaK molecular chaperone. J. Biomol. NMR, 30,113–114.

34. Valafar, H. & Prestegard, J. H. (2004). REDCAT: aresidual dipolar coupling analysis tool. J. Magn. Reson.167, 228–241.

35. Tjandra, N., Omichinski, J. G., Gronenborn, A. M.,Clore, G. M. & Bax, A. (1997). Use of dipolar 1H–15Nand 1H–13C couplings in the structure determinationof magnetically oriented macromolecules in solution.Nat. Struct. Biol. 4, 732–738.

36. Tolman, J. R., Al-Hashimi, H. M., Kay, L. E. &Prestegard, J. H. (2001). Structural and dynamicanalysis of residual dipolar coupling data for proteins.J. Am. Chem. Soc. 123, 1416–1424.


37. Yang, D., Tolman, J. R., Goto, N. T. & Kay, L. E.(1998). An HNCO-based pulse scheme for themeasurement of 13C[agr]–1H[agr] one-bond dipolarcouplings in 15N, 13C labeled proteins. J. Biomol.NMR, 12, 325–332.

38. Wilbanks, S. M., Chen, L., Tsuruta, H., Hodgson,K. O. & McKay, D. B. (1995). Solution small-angleX-ray scattering study of the molecular chaperoneHsc70 and its subfragments. Biochemistry, 34,12095–12106.

39. Harrison, C. J., Hayer-Hartl, M., Di Liberto, M., Hartl,F. & Kuriyan, J. (1997). Crystal structure of thenucleotide exchange factor GrpE bound to the ATPasedomain of the molecular chaperone DnaK. Science,276, 431–435.

40. Polier, S., Dragovic, Z., Hartl, F. U. & Bracher, A. (2008).Structural basis for the cooperation of Hsp70 andHsp110 chaperones in protein folding. Cell, 133,1068–1079.

41. Schuermann, J. P., Jiang, J., Cuellar, J., Llorca, O.,Wang, L., Gimenez, L. E. et al. (2008). Structure of theHsp110:Hsc70 nucleotide exchange machine. Mol.Cell, 31, 232–243.

42. Kern, D. & Zuiderweg, E. R. (2003). The role ofdynamics in allosteric regulation. Curr. Opin. Struct.Biol. 13, 748–757.

43. Al-Hashimi, H. M., Gosser, Y., Gorin, A., Hu, W.,Majumdar, A. & Patel, D. J. (2002). Concerted motionsin HIV-1 TAR RNA may allow access to bound stateconformations: RNA dynamics from NMR residualdipolar couplings. J. Mol. Biol. 315, 95–102.

44. Zhang, Y. & Zuiderweg, E. R. (2004). The 70-kDa heatshock protein chaperone nucleotide-binding domainin solution unveiled as a molecular machine that can

reorient its functional subdomains. Proc. Natl Acad.Sci. USA, 101, 10272–10277.

45. Rist, W., Graf, C., Bukau, B. & Mayer, M. P. (2006).Amide hydrogen exchange reveals conformationalchanges in hsp70 chaperones important for allostericregulation. J. Biol. Chem. 281, 16493–16501.

46. Buchberger, A., Theyssen, H., Schroder, H., McCarty,J. S., Virgallita, G., Milkereit, P. et al. (1995). Nucleo-tide-induced conformational changes in the ATPaseand substrate binding domains of the DnaK chaper-one provide evidence for interdomain communica-tion. J. Biol. Chem. 270, 16903–16910.

47. Revington, M., Zhang, Y., Yip, G. N., Kurochkin, A. V.& Zuiderweg, E. R. (2005). NMR investigations ofallosteric processes in a two-domain Thermus thermo-philus Hsp70 molecular chaperone. J. Mol. Biol. 349,163–183.

48. Hansen, M. R., Mueller, L. & Pardi, A. (1998). Tunablealignment of macromolecules by filamentous phageyields dipolar coupling interactions. Nat. Struct. Biol.5, 1065–1074.

49. Yang, D., Venters, R. A., Mueller, G. A., Choy, W. Y. &Kay, L. E. (1999). TROSY-based HNCO pulsesequences for the measurement of 1HN–15N,15N–13CO, 1HN–13CO, 13CO–13C[agr] and 1HN–13C[agr] dipolar couplings in 15N, 13C, 2H-labeledproteins. J. Biomol. NMR, 14, 333–343.

50. Ottiger, M., Delaglio, F. & Bax, A. (1998). Measurementof J and dipolar couplings from simplified two-dimensionalNMR spectra. J.Magn. Reson. 131, 373–378.

51. Lukin, J. A., Kontaxis, G., Simplaceanu, V., Yuan, Y.,Bax, A. & Ho, C. (2003). Quaternary structure ofhemoglobin in solution. Proc. Natl Acad. Sci. USA, 100,517–520.

Documents

Allostery in Hsp70 Chaperones Is Transduced by Subdomain Rotations