MINIREVIEW / MINISYNTHÈSE

Mechanism of substrate recognition by thechaperonin GroEL

Walid A. Houry

Abstract: The bacterial chaperonin GroEL functions with its cofactor GroES in assisting the folding of a wide rangeof proteins in an ATP-dependent manner. GroEL–GroES constitute one of the main chaperone systems in theEsche-richia coli cytoplasm. The chaperonin facilitates protein folding by enclosing substrate proteins in a cage defined bythe GroEL cylinder and the GroES cap where folding can take place in a protected environment. The in vivo role ofGroEL has recently been elucidated. GroEL is found to interact with 10–15% of newly synthesized proteins, with astrong preference for proteins in the molecular weight range of 20–60 kDa. A large number of GroEL substrates havebeen identified and were found to preferentially contain proteins with multipleαβ domains that haveα-helices andβ-sheets with extensive hydrophobic surfaces. Based on the preferential binding of GroEL to these proteins and structuraland biochemical data, a model of substrate recognition by GroEL is proposed. According to this model, binding takesplace preferentially between the hydrophobic residues in the apical domains of GroEL and the hydrophobic faces ex-posed by theβ-sheets orα-helices in theαβ domains of protein substrates.

Key words: chaperone, folding, binding, hydrophobic interaction, structure.

HouryRésumé: La chaperonine bactérienne GroEL en association avec son cofacteur GroES aide au repliement d’une grandevariété de protéines par un mécanisme ATP-dépendant. Le système GroEL–GroES est un des principaux systèmeschaperons dans le cytoplasme deEscherichia coli. La chaperonine facilite le repliement des protéines en enfermant lessubstrats protéiques dans une cage, formée du cylindre GroEL et du couvercle GroES, à l’intérieure de laquelle lerepliement peut se faire dans un environnement protégé. Le rôle de GroEL in vivo a récemment été élucidé. GroELinteragit avec 10–15% des protéines nouvellement synthétisées et elle a une forte préférence pour les protéines demasse moléculaire entre 20 et 60 kDa. Un grand nombre de substrats de GroEL ont été identifiés; ce sont des protéinesayant plusieurs domainesαβ contenant des hélicesα et des feuilletsβ à grande surface hydrophobe. En se basant surla liaison préférentielle de GroEL à ces protéines et sur des données structurales et biochimiques, un modèle de recon-naissance du substrat par GroEL est proposé. Selon ce modèle, la liaison s’établit préférentiellement entre les résidushydrophobes des domaines apicaux de GroEL et les faces hydrophobes exposées des feuilletsβ ou des hélicesα setrouvant dans les domainesαβ des substrats protéiques.

Mots clés: chaperon, repliement, liaison, interaction hydrophobe, structure.

[Traduit par la Rédaction] 577

Introduction

The successful folding of newly translated proteins is es-sential to the viability of the cell. Since the environment inthe cell is very crowded (Fulton 1982; Goodsell 1991) andviscous (Swaminathan et al. 1997; Partikian et al. 1998;Elowitz et al. 1999), there is a high probability that newlytranslated proteins will misfold and aggregate. Consequently,the cell employs quality-control mechanisms to ensure that

the folding process of any newly synthesized polypeptidechain results in the formation of properly folded protein andthat the folded protein is maintained in an active conforma-tion throughout its functional lifetime (Netzer and Hartl1998; Horwich et al. 1999; Wickner et al. 1999). Molecularchaperones are key components of this quality-control ma-chinery. They are a fundamental group of proteins whichhave been identified only relatively recently (Ellis andHemmingsen 1989). They have been shown to play essentialroles in cell viability under both normal and stress condi-tions by assisting in the folding of newly translated proteinsand by maintaining the conformational integrity of preexist-ing proteins (Hartl 1996).

Molecular chaperones are known to be very promiscuousin vitro (Bukau et al. 1996; Coyle et al. 1997). In a test tube,they have the ability to bind to a wide range of proteins inseveral conformational states. In vivo, however, there is

Biochem. Cell Biol.79: 569–577 (2001) © 2001 NRC Canada

569

DOI: 10.1139/bcb-79-5-569

Received March 26, 2001. Revised May 29, 2001. AcceptedMay 31, 2001. Published on the NRC Research Press Website at http://bcb.nrc.ca on October 3, 2001.

W.A. Houry. Department of Biochemistry, Medical SciencesBuilding, University of Toronto, 1 King’s College Circle,Toronto, ON M5S 1A8, Canada(e-mail: walid.houry@utoronto.ca).

mounting evidence that chaperones have a high affinity for adefined set of proteins and therefore preferentially assist inthe folding and conformational maintenance of only this setof proteins. Key advances have recently been made in char-acterizing the mechanism of function of the bacterialchaperonin GroEL in vitro and in vivo.

GroEL function in vitro

The chaperonin GroEL with its cofactor GroES are thetypical representatives of the Hsp 60 and Hsp 10 families ofmolecular chaperones, respectively (Hsp refers to proteinsinduced by heat shock). GroEL–GroES is the only essentialchaperone system inEscherichia colicytoplasm under allgrowth conditions (Fayet et al. 1989). The structures ofGroEL, GroES, and the GroEL–GroES complex have beenextensively investigated and have been visualized at atomicresolution by X-ray crystallography (Braig et al. 1994; Huntet al. 1996; Xu et al. 1997) (Fig. 1). GroEL is a homo-oligomer of fourteen 57-kDa subunits. The subunits arearranged into two heptamericrings forming a cylindricalstructure with two large cavities. GroES is a heptamer of 10-kDa subunits, and it forms a cap over the cavities in theGroEL cylinder. The GroEL subunit is divided into apical(residues 191–376), intermediate (residues 134–190 and377–408), and equatorial domains (residues 1–133 and 409–547). Protein binding takes place at the outermost apical do-mains. Residues in the apical domain which are importantfor substrate binding are predominantly hydrophobic(Fenton et al. 1994). These residues form a hydrophobicpatch at the rim of each of the GroEL rings. The intermedi-ate domain forms a short flexible linker between the apicaland equatorial domains, and the equatorial domains form thecontact region between the two GroEL rings. ATP bindingand hydrolysis take place at the equatorial domains. The api-cal and intermediate domains of GroEL undergo largeconformational changes as a result of nucleotide and GroESbinding (Fig. 1).

The mechanism of GroEL function is well established invitro (Fenton and Horwich 1997; Martin and Hartl 1997;Sigler et al. 1998; Rye et al. 1999) (Fig. 2). The substrate-acceptor state of GroEL is thought to be an ADP bullet inwhich GroES and seven ADPs are bound to the same ring ofGroEL. This ring might also contain a previously encapsu-lated polypeptide. Another substrate protein then binds tothe apical domains of the opposite free ring of GroEL in anon-native compact conformation mainly through hydropho-bic interactions to hydrophobic residues in helices H (resi-dues 234–243) and I (residues 256–268) and a loop region(residues 196–211) (Fenton et al. 1994). ATP and thenGroES subsequently bind to the newly occupied ring ofGroEL, resulting in displacement of the protein into an en-closed cavity defined by the GroEL cylinder and the GroEScap. This also results in the release, from the opposite ring,of ADP and GroES and any previously encapsulatedpolypeptide. The folding process of the new substrate thentakes place in the enclosed cage in which aggregation andother possible unproductive interactions are prevented. Afterabout 10–20 s of GroES binding, ATP on the polypeptide-containing ring is hydrolyzed to ADP. This results in the for-mation of a new ADP bullet which can act as the acceptor

state for another substrate molecule. ATP and GroES thenbind to the opposite ring, resulting in the release of ADP,GroES, and substrate protein from the distal ring. The cyclethen repeats as described above. Some proteins require mul-tiple rounds of binding and release to reach their nativestate.

There is some controversy regarding the role that GroELplays in the actual folding process of substrate proteins.GroEL is usually observed to retard the folding of proteinswhich can fold rapidly in vitro (Itzhaki et al. 1995;Katsumata et al. 1996; Tsurupa et al. 1998). It has also beenobserved to accelerate the refolding rate of those proteinswhich fold slowly and are prone to aggregation (Martin et al.1991; Ben-Zvi et al. 1998; Beissinger et al. 1999). Further-more, it has been proposed that GroEL can actively unfoldsubstrate proteins during its functional cycle (Walter et al.1996; Shtilerman et al. 1999). In any case, there seems to bea general consensus that, even though GroEL might affectthe folding rate of proteins, it does not change the foldingmechanism of these proteins (Coyle et al. 1999). It only pro-vides a folding cage without contributing any steric informa-tion to the folding process itself.

GroEL function in vivo

Although the in vitro mechanism of GroEL function iswell studied, the role of GroEL in folding newly translatedproteins or in the maintenance of preexisting proteins invivo is poorly understood, until recently. My colleaguesand I wanted to quantitatively analyze the contribution ofGroEL to protein folding in theE. coli cytoplasm undernormal and heat-shock conditions and identify thoseE. coliproteins which utilize GroEL for folding and (or) mainte-nance. The strategyemployed was to visualize those pro-teins which directly interactedwith GroEL in vivo. Tothis end, affinity-purified anti-GroEL antibodies wereused to immunoprecipitate GroEL-substrate complexesfrom E. coli cells in mid-log phase (Ewalt et al. 1997;Houry et al. 1999). Cells were converted to spheroplasts bytreatment with lysozyme and then lysed gently in a hypo-osmotic buffer in the presence of EDTA, which stopped theATP-dependent polypeptide release from GroEL. Severalcontrol experiments were carried out to ensure thatimmunoprecipitations were specific. Antibodies used wereshown to be specific in recognizing only the GroEL bandby treating the cell lysate with 1% SDS, diluting to 0.1%SDS, and then immunoprecipitating using anti-GroEL anti-bodies. Only GroEL was immunoprecipitated after this pro-cedure and no other protein. Moreover, protein bandswhich coimmunoprecipitated with GroEL were shown to betrue substrates of GroEL, since upon addition of GroESand ATP regenerating system to the cell lysate, the majorityof these proteins were released from GroEL. Also, the ad-dition of noncycling GroEL mutant, which can bind sub-strates but does not release them (Weissman et al. 1994),during cell lysis did not change the amount or pattern ofGroEL-bound polypeptides, indicating that the binding toGroEL occurred in the intact cell and not during or afterlysis (Ewalt et al. 1997). It should be noted that no GroESwas detected in our anti-GroEL immunoprecipitations inthe presence of EDTA, in agreement with in vitro observa-

© 2001 NRC Canada

570 Biochem. Cell Biol. Vol. 79, 2001

tions made using purified proteins which demonstrated thatthe binding of GroES to GroEL requires the presence ofnucleotides (Martin et al. 1993; Todd et al. 1994).

To investigate how newly translated proteins utilizeGroEL for folding, the kinetics of passage of these proteinsthrough GroEL (flux kinetics) was followed by carrying outpulse-chase experiments onE. coli cells in mid-log phaseunder normal and heat-shock conditions. Cells were pulsedwith radiolabelled methionine for a short time of about 15 sand then chased with nonradiolabelled methionine. At differ-ent time points after the chase, cells were lysed rapidly andGroEL–substrate complexes were immunoprecipitated as in-dicated previously. Complexes were then separated on SDS-PAGE (one-dimensional) or on two-dimensional gels andvisualized by phosphorimage analysis. One-dimensionalgels enabled us to follow the flux kinetics of the total set ofproteins, and two-dimensional gels enabled us to follow theflux kinetics of individual proteins. Similar experimentswere carried out on steady-state labelled cells to determinethe role of GroEL in the conformational maintenance of pre-existing proteins. Also, model proteins were overexpressedin vivo or in E. coli translation extracts to verify observa-tions made with bulk proteins. The techniques employedproved to be very powerful in elucidating GroEL function invivo. The following major conclusions were reached:(1) Three classes of proteins are present in the cell cyto-

plasm: proteins with a chaperonin-independent foldingpathway, proteins with an intermediate chaperonin de-pendence for which normally only a small fraction tran-sits GroEL, and highly chaperonin dependent proteins

which require sequestration of aggregation-sensitiveintermediates within the GroEL cavity for successfulfolding.

(2) About 300 proteins, representing 10–15% of total cyto-plasmicE. coli proteins, utilize GroEL for de novo fold-ing under normal growth conditions and about twice asmuch under heat-stress conditions.

(3) GroEL preferentially interacts with proteins of molecu-lar weight between 20 and 60 kDa, as compared to cyto-plasmic E. coli proteins. About two-thirds of theseproteins flux through GroEL with time constants be-tween 20 s and 2 min, reflecting the requirement of oneto several rounds of GroEL binding and release to reachtheir native state. For the other substrates, a fraction ofthe population of a particular protein remains associatedwith GroEL after the chase. This later set of proteinsrepresents preexisting proteins that require repeated cy-cles of GroEL binding and release for conformationalmaintenance during their lifetime in the cell. This set ofproteins was found to be relatively unstable in vivo,since, under heat shock, the same set interacted moreextensively with the chaperonin. Hence, GroEL is im-portant for folding newly translated proteins and forconformational maintenance of preexisting proteins.

Substrate recognition by GroEL

Although purified GroEL is able to bind in vitro to-50% ofsolubleE. coli proteins in their unfolded or partially folded states(Viitanen et al. 1992), the finding that only -300 proteins

© 2001 NRC Canada

Houry 571

Fig. 1. Structure of the chaperonin machinery. GroEL, GroES, and the GroEL–GroES complex are displayed using the ribbon repre-sentation. Also shown is the conformational change that a GroEL subunit undergoes upon ATP and GroES binding. Structures weredrawn using MolMol (Koradi et al. 1996) and the PDB files 1DER (Boisvert et al. 1996) and 1AON (Xu et al. 1997).

require the chaperonin for folding in vivo indicates thatGroEL, in the cell, preferentially recognizes features presentmainly in this small fraction of proteins. It was necessary toidentify these proteins to determine the features that distin-guish GroEL substrates from the rest ofE. coli proteins. Tothis end, large-scale immunoprecipitation of GroEL–sub-strate complexes fromE. coli cells at mid-logphase wascarried out. These complexes were separated on two-dimensional gels and protein spots were identified bypeptide-mass fingerprint analysis using MALDI-TOF (Houry et al. 1999).Fifty-two proteins were identified unequivocally using this pro-cedure. Substrates included essential components of the tran-scription translation machinery and enzymes involved in severalmetabolic pathways.

Sequence analysis failed to reveal statistically significantconsensus sequences among the 52 identified substrates, inagreement with in vitro studies which also did not find aconsensus sequence among model peptides that bound toGroEL (Coyle et al. 1997). However, 24 out of the 52GroEL substrates were amenable to structural classificationusing domain classification databases such as SCOP(Hubbard et al. 1999) or CATH (Orengo et al. 1999). Thisclassification demonstrated that GroEL substrates preferen-tially contained two or moreαβ domains compared tosolubleE. coli proteins (Fig. 3). The preference was statisti-cally significant at the 95% confidence limit. The most com-mon domain architectures in GroEL substrates were those of

the three-layerαβα and the two-layerαβ sandwiches. Thesedomains typically haveα-helices andβ-sheets with extensivehydrophobic surfaces. These surfaces are ideally suited forrecognition by the hydrophobic residues in the apical bindingdomains of GroEL.

For αβ proteins, the formation of theβ-sheet is expectedto be a difficult step in the folding process since, unlike theformation of anα-helix, the assembly of the sheet requiresthe formation of proper long-range contacts in the properorientation. This is further complicated by the fact that thehydrophobic surface of the assembled sheet has to be buriedaway from the surrounding solvent and hence has to be sta-bilized by the packing of helices, or some other secondarystructure elements, against the sheet surface through hydro-phobic interactions. Misfolding can occur during the foldingprocess either through the improper packing of helicesagainst the sheet within one domain or the packing of heli-ces of one domain against a sheet in another domain. Alter-natively, helices or sheets with exposed hydrophobicsurfaces in one protein molecule might pack against a sheetin another proteinmolecule, leading to aggregation.

Since the identified GroEL substrates have preferentiallymultiple αβ domains compared toE. coli proteins, they areexpected to fold slowly (Plaxco et al. 1998) and to be struc-turally labile. Mogk et al. (1999) recently identified 57thermolabileE. coli proteins. Significantly, 11 of these pro-teins are also GroEL substrates:CLPX_ECOLI, G3P1_ECOLI,GATY_ECOLI, GLF_ECOLI, METE_ECOLI, METF_ECOLI,MIND_ECOLI, NUSG_ECOLI, RPOA_ECOLI, RPOB_ECOLI,and SYT_ECOLI.Furthermore, 24 of the thermolabile pro-teins were amenable to structural classification using theSCOP database (Hubbard et al. 1999). It was striking to findthat their set of thermolabile proteins was structurally verysimilar to the set of 24 GroEL substrates (Fig. 3). Both setswere enriched with proteins containing multipleαβ domainscompared toE. coli proteins. It should be noted, however,that the average protein size of the 24 thermolabileE. coliproteins and the 24 GroEL substrates is 425 and 328 aminoacids, respectively. This might indicate that GroEL’s prefer-ence for binding substrates with multipleαβ domains is evenstronger than that dictated by random selection of therm-olabile proteins. The significance of this finding will be veri-fied once the structures of moreE. coli proteins are known.

The apical domains of GroEL can assist in the folding ofsuch multidomainαβ proteins in several ways. Residues im-plicated in polypeptide binding in the apical domains ofGroEL could provide a scaffold on which the hydrophobicβ-sheet(s) or amphipathic helices of the substrate protein canassemble; subsequent folding of the protein could then pro-ceed in the enclosed GroEL–GroES cavity. Alternatively, if awrong or improper packing occurred between helices andβ-sheet(s) in a protein, the apical domains of GroEL could in-terfere by displacing the helices andβ-sheet(s) and thenbinding to the exposed hydrophobic surfaces. Upon GroESbinding to GroEL, the apical domains would then be re-leased from these hydrophobic faces, and the protein wouldattempt to refold to achieve the right packing interactions.

In this regard, it has been demonstrated biochemically thatthe binding of unfolded rhodanese, malate dehydrogenase,and rubisco to the chaperonin requires recognition by multi-ple GroEL apical domains (Farr et al. 2000). All three

© 2001 NRC Canada

572 Biochem. Cell Biol. Vol. 79, 2001

Fig. 2. The GroEL functional cycle. The cycle starts in theupper-left corner and proceeds counterclockwise. The ATP-dependent GroEL-assisted folding of a compact substrate labelledI is shown. I′ refers to another conformation of I, and N refers tothe native state of I.

proteins are stringent substrates of GroEL, requiring GroESand ATP for efficient folding, and all three consist of twoαβ domains. In support of the biochemical data, the struc-tural data of Wang et al. (2000) indicate that the large sizeof theαβ domains in the identified substrates would indeedrequire more than three GroEL apical domains for binding.

The scheme shown in Fig. 4 provides our working modelfor the binding between the apical domains of GroEL andthe hydrophobic face of the sheet(s) or helices inαβ proteindomains. At this level, the model does not distinguish be-tween the binding of GroEL to the hydrophobic surface as awhole or to individual strands and helices which make upthat surface.

Evidence for the GroEL binding model

Recent structural data have demonstrated that only thosepeptides with a high tendency to present a hydrophobic sur-face exhibit high affinity for GroEL binding, irrespective ofwhether the peptide adopts a helical or a strand conforma-tion (Wang et al. 1999; Chatellier et al. 1999). In the X-raystructure of the apical domain of GroEL (residues 191–376),Buckle et al. (1997) observed that seven residues in the N-terminal tag of one molecule bind through hydrophobic in-teractions to helices H and I of a second molecule in amainly extended conformation. Chen and Sigler (1999)solved the crystal structure of the apical domain of GroEL incomplex with a model peptide. They found that the peptidebound to helices H and I in aβ hairpin conformation. Al-though no helical substrates have yet been seen bound toGroEL at atomic resolution, transferred nuclear Overhauserenhancements experiments have demonstrated that some

peptides adopt a helical conformation when bound to GroEL(Landry et al. 1992; Kobayashi et al. 1999). These differentstructural studies reveal the ability of the apical domain ofGroEL of promiscuous binding to a wide range of nonnativesubstrates provided a hydrophobic surface is available forbinding.

GroEL has been observed to be able to bind proteins intheir native conformation. Upon incubation of native pre-β-lactamase (Laminet et al. 1990) or native human dihydro-folate reductase (DHFR) (Viitanen et al. 1991) with GroEL,both proteins become inactivated and are bound by GroEL.Both enzymes areαβ proteins which containαβ (pre-β-lactamase) orαβα (DHFR) sandwiches. The observed bind-ing of the native proteins to the chaperonin could be the resultof the binding of helices H and I in the apical domain ofGroEL to the buried sheet in these proteins after displace-ment of the endogenous helices. This would lead to thedestabilization and inactivation of the proteins. Interestingly,Frieden and coworkers (Clark et al. 1996; Clark and Frieden1997) have observed thatE. coli DHFR interacts only tran-siently with GroEL upon refolding, whereas murine DHFRexhibits a strong interaction with the chaperonin. The twoproteins are very homologous and are expected to have simi-lar structures consisting ofαβα sandwiches. The major dif-ference between the two structures is the presence of longloop regions in the murine DHFR which are absent in theE. coli protein. Upon insertion of these loops intoE. coliDHFR, Clark et al. (1996) observed strong binding toGroEL. Significantly, one of the inserted loops replaces atight turn region inE. coli DHFR which links the buriedsheet to the opposite helix. Hence, based on our model, thelonger loop would allow for easier access of the apical

© 2001 NRC Canada

Houry 573

Fig. 3. Structural analysis of GroEL substrates. The SCOP domain classification database was used to analyze the structures of GroELsubstrates of Houry et al. (1999), thermolabileE. coli proteins of Mogk et al. (1999), and totalE. coli proteins. Protein domains wereclassified through sequence homology to domains in the SCOP database. Proteins containing a stretch of more than 100 contiguousamino acids not covered by sequence homology to domains in SCOP were excluded from the analysis. The numbers of proteins ame-nable for structural classification in the set of GroEL substrates, thermolabileE. coli proteins, and totalE. coli proteins were 24, 24,and 492, respectively (refer to Houry et al. 1999 for details of the method used).

domain of GroEL to the buried sheet and would explain theincreased binding of the mutatedE. coli DHFR.

In addition, there have been many investigations wheremodel proteins bound to GroEL were subjected to limitedproteolysis to determine what protein fragments were boundby GroEL. In one such study, Hlodan et al. (1995) identifiedtwo fragments of rhodanese which remain bound to GroELafter limited proteolysis of the rhodanese–GroEL complex.In the native structure, these fragments contain hydrophobicand amphiphilicα-helices and extensiveβ-sheet regions. Inanother study, Gervasoni et al. (1998) proposed a model forthe β-lactamase–GroEL complex in which a carboxy-terminal helix and other secondary structure elements inβ-lactamase are melted out upon binding to GroEL, causing apartial uncovering of the large buriedβ-sheet. Their result isconsistent with our proposed model for the binding betweenGroEL and its substrates.

Hence, most of the observations in the literature supportthe model shown in Fig. 4. It should be emphasized, how-ever, that the model represents only what is proposed to bethe predominant mode of binding between GroEL and its

substrates, especially that most GroEL substrates are ex-pected to haveαβ domains. Other possible modes of bindingby GroEL are not excluded because the chaperonin can alsorecognize and help in the folding of all alpha (e.g., citratesynthase; Mendoza and Campo 1996) and all beta proteins(e.g., Fab fragment of a monoclonal antibody; Schmidt andBuchner 1992) in vitro and has also been shown to interactwith an all alpha protein in vivo (ferritin; Houry et al. 1999).

GroEL versus DnaK

The other major chaperone system in theE. coli cyto-plasm, in addition to the Hsp 60 system, is the Hsp 70 sys-tem. This system consists of DnaK (Hsp 70), its cofactorDnaJ (Hsp 40), and a nucleotide exchange factor GrpE.DnaK has been studied extensively in vitro and in vivo. LikeGroEL, DnaK has an ATPase activity and undergoes ATP-dependent conformational changes during its functional cy-cle, which is regulated by DnaJ and GrpE (Szabo et al.1994; Pierpaoli et al. 1997; Laufen et al. 1999; Russell et al.1999). The binding of DnaK to target polypeptides occurs

© 2001 NRC Canada

574 Biochem. Cell Biol. Vol. 79, 2001

Fig. 4. Proposed interaction surfaces between GroEL and its substrates. The GroEL ring is shown viewed from the top. Residues im-plicated in polypeptide binding in the apical domain of GroEL and hydrophobic residues in theαβ domain of substrate protein on theburied surface of the sheet or the helices are drawn as CPK models with different shades of blue. The domain shown is anαβ two-layer sandwich of RL7_ECOLI (PDB code 1CTF), which is one of the identified GroEL substrates. Structures were drawn usingSwissPdbViewer (Guex and Peitsch 1997) and MolMol (Koradi et al. 1996).

through a channel defined mainly by loop regions in the C-terminal domain of DnaK (Zhu et al. 1996). The preferredpeptide-binding motif of DnaK is proposed to consist of alinear stretch of five hydrophobic residues that are flankedon both sides by regions enriched in basic residues (Rudigeret al. 1997).

The role of DnaK in vivo has been the subject of somecontroversy. Under normal growth conditions, deletion ofthe dnaKJ gene in a wild-type strain results in lethality(Teter et al. 1999), whereas in other strain backgrounds orunder certain conditions the deletion results only in the rapidaccumulation of suppressor mutations (Bukau and Walker1990; Kang and Craig 1990; Teter et al. 1999). Under heat-stress conditions, however, DnaK seems to be essential forviability under all genetic backgrounds. Recently, DnaK hasbeen shown to assist in vivo in the de novo folding of 5–10% of total solubleE. coli proteins at 30°C, including ribo-some-bound nascent chains (Deuerling et al. 1999; Teter etal. 1999). These proteins are preferentially in the size rangeof 30–75 kDa. However, the identities of these substrates re-main largely unknown. Only four proteins are currentlyknown to directly interact with DnaK under normal growthconditions at 30°C (Mogk et al. 1999).

Since DnaK recognizes polypeptide segments in extendedconformation, whereas GroEL recognizes compact interme-diates, and since DnaK has the ability to bind to ribosome-associated nascent chains, whereas no such binding has beenobserved for GroEL, it is reasonable to assume that DnaKacts in the cell at an earlier stage in the folding of target pro-teins than does GroEL. The possibility that the set of DnaKsubstrates might overlap with that of GroEL substrates comesfrom the observation that both DnaK and GroEL preferen-tially bind to proteins in a similar molecular weight range:30–75 kDa for DnaK and 20–60 kDa for GroEL. Whetherthere is a direct transfer of substrates from DnaK to GroELrequires further investigation. Such a direct transfer has beendemonstrated in vitro (Langer et al. 1992) and has been sug-gested to occur in vivo (Gaitanaris et al. 1994; Teter et al.1999). It is most likely that a large fraction of newly synthe-sized polypeptides would require both the Hsp 70 and Hsp60 systems for proper folding and, hence, the two chaperonesystems have to cooperate in folding these proteins.

Future directions

Although much is now known about GroEL structure andfunction, the identification of the natural substrates of thechaperonin raises several major issues which remain to beresolved. The extent of interaction of the identified proteinswith the chaperonin has to be determined, both in vitro andin vivo. The mode of binding of these substrates to GroELalso has to be further investigated by using a battery of bio-physical and imaging techniques. Lastly, the role that GroELplays as part of a multichaperone network has to be investi-gated by the identification and characterization of substratesof other chaperone systems. Work is currently in progress onall of these tracks. The goal, ultimately, is to be able to elu-cidate at the molecular level the function of differentchaperone systems in the in vivo folding process ofE. coliproteins.

Acknowledgements

I would like to thank F.U. Hartl, D. Frishman, C.Eckerskorn, F. Lottspeich, and K. Anderson for their supportduring the course of this work.

References

Beissinger, M., Rutkat, K., and Buchner, J. 1999. Catalysis, com-mitment and encapsulation during GroE-mediated folding. J.Mol. Biol. 289: 1075–1092.

Ben-Zvi, A.P., Chatellier, J., Fersht, A.R., and Goloubinoff, P.1998. Minimal and optimal mechanisms for GroE-mediatedprotein folding [published erratum appears in Proc. Natl. Acad.Sci. U.S.A. 96(10): 5890]. Proc. Natl. Acad. Sci. U.S.A.95:15 275 – 15 280.

Boisvert, D.C., Wang, J.M., Otwinowski, Z., Horwich, A.L., andSigler, P.B. 1996. The 2.4 angstrom crystal structure of the bac-terial chaperonin GroEL complexed with ATP-gamma-S. Nat.Struct. Biol. 3: 170–177.

Braig, K., Otwinowski, Z., Hegde, R., Boisvert, D.C., Joachimiak,A., Horwich, A.L., and Sigler, P.B. 1994. The crystal structureof the bacterial chaperonin GroEL at 2.8 A. Nature (London),371: 578–586.

Buckle, A.M., Zahn, R., and Fersht, A.R. 1997. A structural modelfor GroEL-polypeptide recognition. Proc. Natl. Acad. Sci.U.S.A. 94: 3571–3575.

Bukau, B., and Walker, G.C. 1990. Mutations altering heat shockspecific subunit of RNA polymerase suppress major cellular de-fects ofE. coli mutants lacking the DnaK chaperone. EMBO J.9: 4027–4036.

Bukau, B., Hesterkamp, T., and Luirink, J. 1996. Growing up in adangerous environment—a network of multiple targeting andfolding pathways for nascent polypeptides in the cytosol. TrendsCell Biol. 6: 480–486.

Chatellier, J., Buckle, A.M., and Fersht, A.R. 1999. GroEL recog-nises sequential and non-sequential linear structural motifs com-patible with extendedβ-strands andα-helices. J. Mol. Biol.292:163–172.

Chen, L.L., and Sigler, P.B. 1999. The crystal structure of aGroEL/peptide complex: plasticity as a basis for substrate diver-sity. Cell, 99: 757–768.

Clark, A.C., and Frieden, C. 1997. GroEL-mediated folding ofstructurally homologous dihydrofolate reductases. J. Mol. Biol.268: 512–525.

Clark, A.C., Hugo, E., and Frieden, C. 1996. Determination of re-gions in the dihydrofolate reductase structure that interact withthe molecular chaperonin GroEL. Biochemistry,35: 5893–5901.

Coyle, J.E., Jaeger, J., Gross, M., Robinson, C.V., and Radford,S.E. 1997. Structural and mechanistic consequences ofpolypeptide binding by GroEL. Folding & Design,2: R93–R104.

Coyle, J.E., Texter, F.L., Ashcroft, A.E., Masselos, D., Robinson,C.V., and Radford, S.E. 1999. GroEL accelerates the refoldingof hen lysozyme without changing its folding mechanism. Nat.Struct. Biol. 6: 683–690.

Deuerling, E., Schulze-Specking, A., Tomoyasu, T., Mogk, A., andBukau, B. 1999. Trigger factor and DnaK cooperate in foldingof newly synthesized proteins. Nature (London),400: 693–696.

Ellis, R.J., and Hemmingsen, S.M. 1989. Molecular chaperones:proteins essential for the biogenesis of some macromolecularstructures. Trends Biochem. Sci.14: 339–342.

© 2001 NRC Canada

Houry 575

Elowitz, M.B., Surette, M.G., Wolf, P.E., Stock, J.B., and Leibler,S. 1999. Protein mobility in the cytoplasm ofEscherichia coli.J. Bacteriol.181: 197–203.

Ewalt, K.L., Hendrick, J.P., Houry, W.A., and Hartl, F.U. 1997. Invivo observation of polypeptide flux through the bacterialchaperonin system. Cell,90: 491–500.

Farr, G.W., Furtak, K., Rowland, M.B., Ranson, N.A., Saibil, H.R.,Kirchhausen, T., and Horwich, A.L. 2000. Multivalent bindingof nonnative substrate proteins by the chaperonin GroEL. Cell,100: 561–573.

Fayet, O., Ziegelhoffer, T., and Georgopoulos, C. 1989. The GroESand GroEL heat shock gene products ofEscherichia coliare es-sential for bacterial growth at all temperatures. J. Bacteriol.171:1379–1385.

Fenton, W.A., and Horwich, A.L. 1997. GroEL-mediated proteinfolding. Protein Sci.6: 743–760.

Fenton, W.A., Kashi, Y., Furtak, K., and Horwich, A.L. 1994. Res-idues in chaperonin GroEL required for polypeptide binding andrelease. Nature (London),371: 614–619.

Fulton, A.B. 1982. How crowded is the cytoplasm? Cell,30: 345–347.

Gaitanaris, G.A., Vysokanov, A., Hung, S.C., Gottesman, M.E.,and Gragerov, A. 1994. Successive action ofEscherichia colichaperones in vivo. Mol. Microbiol.14: 861–869.

Gervasoni, P., Staudenmann, W., James, P., and Pluckthun, A.1998. Identification of the binding surface on beta-lactamase forGroEL by limited proteolysis and MALDI-mass spectrometry.Biochemistry,37: 11 660 – 11 669.

Goodsell, D.S. 1991. Inside a living cell. Trends Biochem. Sci.16:203–206.

Guex, N., and Peitsch, M.C. 1997. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modeling.Electrophoresis,18: 2714–2723.

Hartl, F.U. 1996. Molecular chaperones in cellular protein folding.Nature (London),381: 571–579.

Hlodan, R., Tempst, P., and Hartl, F.U. 1995. Binding of definedregions of a polypeptide to GroEL and its implications forchaperonin-mediated protein folding. Nat. Struct. Biol.2: 587–595.

Horwich, A.L., Weber-Ban, E.U., and Finley, D. 1999. Chaperonerings in protein folding and degradation. Proc. Natl. Acad. Sci.U.S.A. 96: 11 033 – 11 040.

Houry, W.A., Frishman, D., Eckerskorn, C., Lottspeich, F., andHartl, F.U. 1999. Identification of in vivo substrates of thechaperonin GroEL. Nature (London),402: 147–154.

Hubbard, T.J.P., Ailey, B., Brenner, S.E., Murzin, A.G., andChothia, C. 1999. SCOP: a structural classification of proteinsdatabase. Nucleic Acids Res.27: 254–256.

Hunt, J.F., Weaver, A.J., Landry, S.J., Gierasch, L., andDeisenhofer, J. 1996. The crystal structure of the GroES co-chaperonin at 2.8C resolution. Nature (London),379: 37–45.

Itzhaki, L.S., Otzen, D.E., and Fersht, A.R. 1995. Nature and con-sequences of GroEL-protein interactions. Biochemistry,34:14 581 – 14 587.

Kang, P.J., and Craig, E.A. 1990. Identification and characteriza-tion of a newEscherichia coligene that is a dosage-dependentsuppressor of a dnaK deletion mutation. J. Bacteriol.172: 2055–2064.

Katsumata, K., Okazaki, A., and Kuwajima, K. 1996. Effect ofGroEL on the re-folding kinetics of alpha-lactalbumin. J. Mol.Biol. 258: 827–838.

Kobayashi, N., Freund, S.M., Chatellier, J., Zahn, R., and Fersht,A.R. 1999. NMR analysis of the binding of a rhodanese peptideto a minichaperone in solution. J. Mol. Biol.292: 181–190.

Koradi, R., Billeter, M., and Wuthrich, K. 1996. MolMol—a pro-gram for display and analysis of macromolecular structures. J.Mol. Graphics,14: 51.

Laminet, A.A., Ziegelhoffer, T., Georgopoulos, C., and Pluckthun,A. 1990. TheEscherichia coliheat shock proteins GroEL andGroES modulate the folding of the beta-lactamase precursor.EMBO J. 9: 2315–2319.

Landry, S.J., Jordan, R., McMacken, R., and Gierasch, L.M. 1992.Different conformations for the same polypeptide bound tochaperones DnaK and GroEL. Nature (London),355: 455–457.

Langer, T., Lu, C., Echols, H., Flanagan, J., Hayer, M.K., andHartl, F.U. 1992. Successive action of DnaK, DnaJ and GroELalong the pathway of chaperone-mediated protein folding. Na-ture (London),356: 683–689.

Laufen, T., Mayer, M.P., Beisel, C., Klostermeier, D., Mogk, A.,Reinstein, J., and Bukau, B. 1999. Mechanism of regulation ofhsp70 chaperones by DnaJ cochaperones. Proc. Natl. Acad. Sci.U.S.A. 96: 5452–5457.

Martin, J., and Hartl, F.U. 1997. Chaperone-assisted protein fold-ing. Curr. Opin. Struct. Biol.7: 41–52.

Martin, J., Langer, T., Boteva, R., Schramel, A., Horwich, A.L.,and Hartl, F.U. 1991. Chaperonin-mediated protein folding atthe surface of GroEL through a ‘molten globule’-like intermedi-ate. Nature (London),352: 36–42.

Martin, J., Mayhew, M., Langer, T., and Hartl, F.U. 1993. The re-action cycle of GroEL and GroES in chaperonin-assisted proteinfolding. Nature (London),366: 228–233.

Mendoza, J.A., and Campo, G.D. 1996. Ligand-inducedconformational changes of GroEL are dependent on the boundsubstrate polypeptide. J. Biol. Chem.271: 16 344 – 16 349.

Mogk, A., Tomoyasu, T., Goloubinoff, P., Rudiger, S., Roder, D.,Langen, H., and Bukau, B. 1999. Identification of thermolabileEscherichia coliproteins: prevention and reversion of aggrega-tion by DnaK and ClpB. EMBO J.18: 6934–6949.

Netzer, W.J., and Hartl, F.U. 1998. Protein folding in the cytosol:chaperonin-dependent and independent mechanisms. TrendsBiochem. Sci.23: 68–73.

Orengo, C.A., Pearl, F.M.G., Bray, J.E., Todd, A.E., Martin, A.C.,Lo Conte, L., and Thornton, J.M. 1999. The CATH databaseprovides insights into protein structure/function relationships.Nucleic Acids Res.27: 275–279.

Partikian, A., Olveczky, B., Swaminathan, R., Li, Y.X., andVerkman, A.S. 1998. Rapid diffusion of Green Fluorescent Pro-tein in the mitochondrial matrix. J. Cell Biol.140: 821–829.

Pierpaoli, E.V., Sandmeier, E., Baici, A., Schonfeld, H.J., Gisler,S., and Christen, P. 1997. The power stroke of theDnaK/DnaJ/GrpE molecular chaperone system. J. Mol. Biol.269: 757–768.

Plaxco, K.W., Simons, K.T., and Baker, D. 1998. Contact order,transition state placement and the refolding rates of single do-main proteins. J. Mol. Biol.277: 985–994.

Rudiger, S., Germeroth, L., Schneidermergener, J., and Bukau, B.1997. Substrate specificity of the Dnak chaperone determined byscreening cellulose-bound peptide libraries. EMBO J.16: 1501–1507.

Russell, R., Wali Karzai, A., Mehl, A.F., and McMacken, R. 1999.DnaJ dramatically stimulates ATP hydrolysis by DnaK: insightinto targeting of Hsp70 proteins to polypeptide substrates. Bio-chemistry,38: 4165–4176.

Rye, H.S., Roseman, A.M., Chen, S.X., Furtak, K., Fenton, W.A.,Saibil, H.R., and Horwich, A.L. 1999. GroEL-GroES cycling:ATP and nonnative polypeptide direct alternation of folding-active rings. Cell,97: 325–338.

© 2001 NRC Canada

576 Biochem. Cell Biol. Vol. 79, 2001

© 2001 NRC Canada

Houry 577

Schmidt, M., and Buchner, J. 1992. Interaction of GroE with anall-beta-protein. J. Biol. Chem.267: 16 829 – 16 833.

Shtilerman, M., Lorimer, G.H., and Englander, S.W. 1999.Chaperonin function: folding by forced unfolding. Science(Washington, D.C.),284: 822–825.

Sigler, P.B., Xu, Z.H., Rye, H.S., Burston, S.G., Fenton, W.A., andHorwich, A.L. 1998. Structure and function in GroEL-mediatedprotein folding. Annu. Rev. Biochem.67: 581–608.

Swaminathan, R., Hoang, C.P., and Verkman, A.S. 1997.Photobleaching recovery and anisotropy decay of Green Fluo-rescent Protein GFP-S65T in solution and cells—cytoplasmicviscosity probed by Green Fluorescent Protein translational androtational diffusion. Biophys. J.72: 1900–1907.

Szabo, A., Langer, T., Schroder, H., Flanagan, J., Bukau, B., andHartl, F.U. 1994. The ATP hydrolysis-dependent reaction cycleof the Escherichia coliHsp70 system DnaK, DnaJ, and GrpE.Proc. Natl. Acad. Sci. U.S.A.91: 10 345 – 10 349.

Teter, S.A., Houry, W.A., Ang, D., Tradler, T., Rockabrand, D.,Fischer, G., Blum, P., Georgopoulos, C., and Hartl, F.U. 1999.Polypeptide flux through bacterial Hsp70: DnaK cooperateswith trigger factor in chaperoning nascent chains. Cell,97: 755–765.

Todd, M.J., Viitanen, P.V., and Lorimer, G.H. 1994. Dynamics ofthe chaperonin ATPase cycle: implications for facilitated proteinfolding. Science (Washington, D.C.),265: 659–666.

Tsurupa, G.P., Ikura, T., Makio, T., and Kuwajima, K. 1998.Refolding kinetics of staphylococcal nuclease and its mutants inthe presence of the chaperonin GroEL. J. Mol. Biol.277: 733–745.

Viitanen, P.V., Donaldson, G.K., Lorimer, G.H., Lubben, T.H., and

Gatenby, A.A. 1991. Complex interactions between thechaperonin 60 molecular chaperone and dihydrofolate reductase.Biochemistry,30: 9716–9723.

Viitanen, P.V., Gatenby, A.A., and Lorimer, G.H. 1992. Purifiedchaperonin 60 (GroEL) interacts with the nonnative states of amultitude ofEscherichia coliproteins. Protein Sci.1: 363–369.

Walter, S., Lorimer, G.H., and Schmid, F.X. 1996. A thermody-namic coupling mechanism for GroEL-mediated unfolding.Proc. Natl. Acad. Sci. U.S.A.93: 9425–9430.

Wang, Q., Buckle, A.M., and Fersht, A.R. 2000. Fromminichaperone to GroEL 1: information on GroEL–polypeptideinteractions from crystal packing of minichaperones. J. Mol.Biol. 304: 873–881.

Wang, Z.L., Feng, H.P., Landry, S.J., Maxwell, J., andGierasch, L.M. 1999. Basis of substrate binding by thechaperonin GroEL. Biochemistry,38: 12 537 – 12 546.

Weissman, J.S., Kashi, Y., Fenton, W.A., and Horwich, A.L. 1994.GroEL-mediated protein folding proceeds by multiple rounds ofbinding and release of nonnative forms. Cell,78: 693–702.

Wickner, S., Maurizi, M.R., and Gottesman, S. 1999.Posttranslational quality control: folding, refolding, and degrad-ing proteins. Science (Washington, D.C.),286: 1888–1893.

Xu, Z.H., Horwich, A.L., and Sigler, P.B. 1997. The crystal struc-ture of the asymmetric GroEL-GroES-(ADP)(7) chaperonincomplex. Nature (London),388: 741–750.

Zhu, X.T., Zhao, X., Burkholder, W.F., Gragerov, A., Ogata, C.M.,Gottesman, M.E., and Hendrickson, W.A. 1996. Structural anal-ysis of substrate binding by the molecular chaperone Dnak. Sci-ence (Washington, D.C.),272: 1606–1614.