REVIEWS TIBS 23 – APRIL 1998

138 Copyright © 1998, Elsevier Science Ltd. All rights reserved. 0968 – 0004/98/$19.00 PII: S0968-0004(98)01193-1

has been successful in beginning to de-fine the physiology of yeast as well asthe pathophysiology of human disease.

References1 Stearmann, R. et al. (1996) Science 271,

1552–15572 Askwith, C. et al. (1994) Cell 76, 403–4103 de Silva, D., Askwith, C., Eide, D. and Kaplan, J.

(1995) J. Biol. Chem. 270, 1098–11014 Eide, D. et al. (1992) J. Biol. Chem. 267,

20774–207815 Lesuisse, E., Crichton, R. R. and Labbe, P.

(1990) Biochim. Biophys. Acta 1038, 253–2596 Dancis, A. et al. (1992) Proc. Natl. Acad. Sci.

U. S. A. 89, 3869–38737 Georgatsou, E. and Alexandraki, D. (1994) Mol.

Cell. Biol. 14, 3065–30738 Scheiber, B. and Goldenberg, H. (1993) Arch.

Biochem. Biophys. 305, 225–230 9 Jordan, I. and Kaplan, J. (1994) Biochem. J.

302, 875–87910 Lee, G. R., Nacht, S., Lukens, J. N. and

Cartwright, G. E. (1968) J. Clin. Invest. 47,2058–2069

11 Osaki, S., Johnson, D. A. and Frieden, E. (1966)J. Biol. Chem. 241, 2746–2751

12 Harris, Z. L. et al. (1995) Proc. Natl. Acad. Sci.U. S. A. 92, 2539–2543

13 Klomp, L. W. and Gitlin, J. D. (1997) Hum. Mol.Genet. 5, 1989–1996

14 Mukhopadhyay, C. K., Attieh, Z. K. and Fox, P. L.(1998) Science 279, 714–717

15 Fleming, M. D. et al. (1997) Nat. Genet. 16,383–386

16 Gunshin, H. et al. (1997) Nature 338, 482–48817 Gruenheid, S., Cellier, M., Vidal, S. and Gros, P.

(1995) Genomics 25, 514–52518 Supek, F., Supekova, L., Nelson, H. and

Nelson, N. (1996) Proc. Natl. Acad. Sci. U. S. A.93, 5105–5110

19 Pinner, K., Gruenheid, S., Raymond, M. andGros, P. (1997) J. Biol. Chem. 272,28933–28938

20 Eide, D., Broderius, M., Fett, J. and Guerinot,M. L. (1996) Proc. Natl. Acad. Sci. U. S. A. 93,5624–5628

21 Dancis, A. et al. (1994) Cell 76, 393–40222 Knight, S. et al. (1996) Genes Dev. 10,

1917–192923 Glerum, D. M., Shtanko, A. and Tzagoloff, A.

(1996) J. Biol. Chem. 271, 14504–1450924 Cizewski Culotta, V. et al. (1997) J. Biol. Chem.

272, 23469–2347225 Lin, S. and Cizewski Culotta, V. (1995) Proc.

Natl. Acad. Sci. U. S. A. 92, 3784–378826 Lin, S. et al. (1997) J. Biol. Chem. 272,

9215–922027 Yuan, D. et al. (1995) Proc. Natl. Acad. Sci.

U. S. A. 92, 2632–263628 Dancis, A., Haile, D., Yuan, D. S. and

Klausner, R. D. (1994) J. Biol. Chem. 269,25660–25667

29 Zhou, B. and Gitscher, J. (1997) Proc. Natl.Acad. Sci. U. S. A. 94, 7481–7486

30 Klomp, L. et al. (1997) J. Biol. Chem. 272,9221–9226

31 Vulpe, C. et al. (1993) Nat. Genet. 3, 7–1332 Chelly, J. et al. (1993) Nat. Genet. 3, 14–1933 Davies, K. (1993) Nat. Genet. 361, 9834 Bull, P. C. et al. (1993) Nat. Genet. 5, 327–33735 Chelly, J. and Monaco, A. P. (1993) Nat. Genet.

5, 317–31836 Tanzi, R. E. et al. (1993) Nat. Genet. 5,

344–35037 Scheinberg, I. H. and Gitlin, D. (1952) Science

116, 484–48538 Hung, I. H. et al. (1997) J. Biol. Chem. 272,

21461–2146639 Babcock, M. et al. (1997) Science 276,

1709–171240 Campuzano, V. et al. (1997) Hum. Mol. Genet.

6, 1771–178041 Harding, A. E. (1993) Adv. Neurol. 61, 1–1442 Lamarche, J. B., Cote, M. and Lemieux, B.

(1980) J. Canad. De. Scien. Neurol. 7, 389–39543 Campuzano, V. et al. (1996) Science 271,

1423–142744 Foury, F. and Cazzalini, O. (1997) FEBS Lett.

411, 373–37745 Wilson, R. B. and Roof, D. M. (1997) Nat.

Genet. 16, 352–35746 Koutnikova, H. et al. (1997) Nat. Genet. 16,

345–35147 Rotig, A. et al. (1997) Nat. Genet. 17, 215–217

SINCE THE LATE 1980s, it has becomeclear that many proteins require assis-tance for folding in vivo. The process of assisted protein folding is carried

out by chaperones, a universally con-served class of proteins. Many chaper-ones are also stress or heat-shock proteins, whose rate of synthesis accel-erates under various protein-damagingconditions.

This review will concentrate on theGroE chaperone machine, originally iden-tified by genetic studies of bacterio-phages (reviewed in Ref. 1), with spe-cial emphasis on the lessons derivedfrom recent structural work.

A new twist: GroE machine structure andfunction

Previous work had established thatthe GroE machine is composed of twomembers, GroEL and GroES, both essen-tial proteins for Escherichia coli under allconditions tested2. Extensive structuralwork revealed that both chaperonins areorganized into rings with a seven-foldrotational axis3,4. GroEL is composed of14 subunits of 58 kDa each, arranged intwo head-to-head rings in whose central,non-connected cavities various substrateproteins can be transiently sequesteredfrom the medium and allowed to maturein solitary confinement. GroES is com-posed of seven subunits of 10.5 kDaeach that form a dome capping GroEL’scentral cavity in the presence of nucleo-tides, thereby providing additional spaceand protection for the substrate. ATPbinding and hydrolysis in the equatorialdomain of GroEL plays a key role inGroES and substrate binding to, and re-lease from, GroEL (reviewed in Refs 5, 6;see below).

The latest model of substrate matu-ration by the GroE chaperone machineis shown in Fig. 1 (adapted from Refs 7, 8).It should be emphasized here that manyof the details of this folding pathwayhave not been completely ironed out.For example, the question of Americanfootballs vs. bullets, that is, whetherone or two GroES molecules bind GroEL

The ins and outs of a molecular

chaperone machine

Alexandra Richardson, Samuel J. Landryand Costa Georgopoulos

Genetic and biochemical work has highlighted the biological importance of theGroEL/GroES (Hsp60/Hsp10; cpn60/cpn10) chaperone machine in proteinfolding. GroEL’s donut-shaped structure has attracted the attention of struc-tural biologists because of its elegance as well as the secrets (substrates)it can hide. The recent determination of the GroES and GroEL/GroES struc-tures provides a glimpse of their plasticity, revealing dramatic conformationalchanges that point to an elaborate mechanism, coupling ATP hydrolysis tosubstrate release by GroEL.

A. Richardson and C. Georgopoulos are atthe Département de Biochimie Médicale,Université de Genève, 1211 Genève 4,Switzerland; and S. J. Landry is at theDepartment of Biochemistry, TulaneUniversity School of Medicine, New Orleans,LA 70112, USA.Email: Alexandra.Richardson@medicine.unige.ch

TIBS 23 – APRIL 1998

139

in a reaction cycle, remains un-resolved, in spite of the flurryof publications by various laboratories. Furthermore, al-though a substrate can achievethe native state within the fold-ing chamber in a single binding/release cycle, it is also clearthat other substrates are oftenejected in an unfolded state.Moreover, the nature of thesubstrate, when tightly boundto the GroEL apical domains,is poorly defined. Thus, it is un-clear whether a given proteinsubstrate progresses in foldingthrough multiple chaperonininteractions9,10, or whether eachinteraction with the chaperoninresults in a new start on thefolding path11,12. This centralissue impedes our understand-ing of how chaperones increasethe efficiency of protein folding.It seems likely that the GroEmachine’s great complexity andflexibility have evolved in orderto accommodate its diverse sub-strates efficiently. This mightexplain why, depending on thespecific substrates and the ex-perimental conditions used,varying data have been ob-tained, resulting in seeminglycontradictory conclusions5,6.

Coordination by the little guy:recent work on GroES structureand function

Recently, the structures offour GroES family memberswere solved by a combination ofX-ray crystallography and NMR studies:E. coli GroES4, Mycobacterium lepraeGroES13, human Hsp10 (Ref. 14 andJ. F. Hunt et al., unpublished) and Gp3115,a distant family member encoded bybacteriophage T4 (only 14% identical toGroES at the amino acid sequence level16).

Figure 2 shows representative viewsof GroES and Gp31. These structures,coupled with genetic and biochemicalwork, emphasize the following salientfeatures. Conserved in all family mem-bers is a heptameric dome-shaped struc-ture. A prominent roof structure foundin the bacterial GroES is dynamicallyflexible in human Hsp10 and completelylacking in Gp31, suggesting that it has anonessential function, at least for T4 andE. coli growth. In agreement with this,the roof loop can be deleted from GroESwithout altering its in vivo propertiesdetectably (A. Richardson, unpublished).

A flexible loop segment, which is 18amino acid residues in GroES andhuman Hsp10 and 23 residues in Gp31,extends from the bottom of each of theseven monomers. As a result of theirmobility17, the loops were discovered byNMR and are generally not observed incrystal structures unless stabilized bycrystal lattice contacts or binding toGroEL (Figs 2 and 3). NMR studies alsodemonstrate that the mobile loops areresponsible for contacting GroEL, reveal-ing that they are immobilized upon bind-ing of GroES to GroEL. Although the mobileloop possesses some characteristics ofa completely disordered chain, its fluc-tuations seem to be biased toward theGroEL-bound conformation, as definedby trNOE (transfer nuclear Overhausereffect) NMR, and several features of itsdynamics profile suggest that its motionis constrained by structural features14.

Mobile-loop conformational dynam-ics are probably important in chaper-onin function for several reasons. First,the disorder-to-order transition in themobile loop upon binding to GroEL car-ries a significant entropic cost for bind-ing. We have argued previously that thisfeature moderates the binding affinity ofGroES to GroEL, permitting the complexto dissociate in a timely fashion, despiteits seven subunit–subunit contacts18.Second, the conformational equilibriumof the loop might have to be biased toward the GroEL-bound conformer, yetnot to the extent that binding becomestoo tight. Third, the dynamics of theloop are highly temperature-sensitive,resulting in a large increase in the en-tropic cost of binding at elevated tem-peratures. Even if there are equal andopposite changes in the free energy ofbinding owing to other factors, effects

REVIEWS

GroES

7Pi 7Pi7 ADP 7 ADP

7 ATP

I II IIIa IVa Va

IIIb IVb Vb

7 ATP

ATP ATP

ATP ATP ATP ATP ADP ADP

ADP ADP ATP ATP ATP ATP

7Pi 7 ADP

7 ATPATP ATP ADP ADP

ATP ATP ATP ATP

Unfoldedsubstrate

GroEL(cis ring)

7ADP

GroEL(trans ring)

Correctlyfolded

substrate

Figure 1 Mechanism of GroEL/GroES-assisted protein folding. For simplicity, we describe the active events ofonly one substrate molecule, depicted in green, although substrate molecules bound to both faces ofGroEL are shown. We refer to the upper ring as the cis ring and the lower as the trans ring. We dividethe reactions arbitrarily into the following stages. Stage I: the unfolded substrate, shown in green,binds to the free face of the asymmetric GroEL/ATP/GroES complex. Stage II: hydrolysis of ATP in thetrans ring is followed by rapid binding of ATP and GroES (in blue) to the substrate-bound ring. This, inturn, results in the release of GroES [and substrate (yellow)] from the trans ring, and large confor-mational changes in the cis ring, causing a doubling of the cavity size under the dome, the reduction ofexposed hydrophobic surface (red), and the release of the substrate to the interior of the cavity. Stage III:once released into the interior of the GroEL cavity, the substrate might fail to fold properly (Stage IIIa),or it might fold to the native or a folding-competent state (Stage IIIb). Stage IV: ATP hydrolysis in the cisring is necessary to destabilize the high-affinity GroEL/GroES complex. GroES release is stimulated byeither ATP or ATP/GroES binding to the trans ring. Stage V: If the substrate is not folded properly, it canrebind to the same or a different GroEL molecule (Stage Va) and the cycle is repeated as many timesas required for the substrate to attain its native state. A substrate which is properly folded or in a folding-competent state does not rebind to GroEL (Stage Vb).

REVIEWS TIBS 23 – APRIL 1998

140

on the kinetics of binding and dischargeprobably alter chaperonin function.

Temperature-dependent changes in theGroEL ATPase and quaternary structuremust also have an impact on chaperoninfunction at different temperatures andduring the stress response. For example,Goloubinoff and coworkers have shownthat GroEL/GroES complexes are lessabundant at elevated temperatures, whichcould be because of a lower steady-statelevel of GroEL/ATP19. Viitanen and co-workers find that hamster Hsp60 existsboth in single and double donut formsand that the distribution of subunits be-tween monomer and donut forms is tem-perature and ATP dependent20. It will beinteresting to see how E. coli accommo-dates these aspects of structure and dy-namics for optimal chaperonin functionat different temperatures.

A giant breathing machine: the GroEL/GroESstructure

The structure of the GroEL/GroES com-plex has been solved recently by HelenSaibil and coworkers21 using cryo-electronmicroscopy and by Paul Sigler and co-workers22 using X-ray crystallography.The structure and its dynamic movementscan be observed directly in a movie ob-tainable via the Internet (http://www.cryst.bbk.ac.uk/~ubcg16z/elmovies. html).

Nucleotide binding induces dramaticrearrangements in GroEL, which are highlysynchronized owing to positive allostericcommunication among the GroEL sub-units within a ring, and negative com-munication between the rings23. Theserearrangements are summarized in sche-matic form in Fig. 4a. The consequencesof rearrangements are multiple. For ex-ample, ADP binding to GroEL results in

the clockwise twisting of the apical do-main. ATP binding accentuates the apicaldomain’s rotation upward and outward.The pivots for this complex movementare two hinges flanking the intermediatedomain. Thus, the intermediate domainis the key channel for allosteric trans-mission from the nucleotide-binding sitein the equatorial domain up to the re-sponding apical domain. GroES bindingeither stabilizes and/or promotes thefollowing conformational changes inGroEL: the apical domain points 60° up-ward and twists 90°, thus burying previ-ously exposed hydrophobic residues.Some of these hydrophobic residues arenow involved in GroES binding whileothers are employed in new GroEL inter-subunit contacts, resulting in the re-lining of the interior GroEL cavity with a polar surface (see Refs 22 and 24 for further discussion). Two of the majorconsequences of these movements arethe synchronized release of the GroEL-bound polypeptide into the centralGroES-capped cavity, and the overall enlargement of the cavity’s volume byup to 200%.

As yet, it is unclear whether such co-ordinated conformational changes wouldbe possible for GroEL alone. WithoutGroES, uncoordinated transitions byGroEL’s subunits might not produce theefficient and timely discharge of sub-strates that contact multiple GroEL sub-units7,17. GroES could promote the simul-taneous conformational changes in theGroEL subunit by mediating interactionsbetween all seven apical domains. Thecrystal structure of the GroEL/ADP/GroEScomplex reveals the approach of aprominent GroEL loop (spanning resi-dues 300–310) to within a few angstromsof the GroES mobile loop bound to theadjacent GroEL subunit (Fig. 3c).

The GroEL/ADP/GroES crystal struc-ture pointed to amino acid 398 as a keyresidue for ATP hydrolysis. Its subse-quent mutational analysis clarified sev-eral questions concerning the role of nucleotides and the importance of thedouble ring in the mechanism of theGroEL chaperone machine8 (Fig. 1). Thechange of residue 398 from an asparticacid to an alanine results in abolition ofGroEL’s ATPase activity. Although thisresidue is not in the vicinity of the ATP-binding pocket in the GroEL–ATPgSstructure25, nevertheless it comes intoclose proximity with ATP in the contextof the GroES-induced 25° tilt of the inter-mediate domain towards the equatorialdomain (see Fig. 4). This emphasizesthat GroES binding has stabilized GroEL

Figure 2 Comparison of the E. coli GroES4 and bacteriophage T4 Gp3115 crystal structures. Top and side views of GroES (a) and Gp31 (b). GroES is shown in green and Gp31 in blue. Themobile loops are highlighted in red. See text for details.

TIBS 23 – APRIL 1998

141

in a conformational rearrangement thatGroEL apparently undergoes by itself,because its ATPase is active in the ab-sence of GroES. Rearrangement of theintermediate domain also results in asteric block of the nucleotide bindingpocket. Thus, the seven ADP moleculespresent in the cis ring are sealed off inthe crystal structure, helping to explainthe high stability of the GroEL/ADP/GroEScomplex. Until triggered by an externalforce, such as ATP and/or GroES bindingto the trans ring, the GroEL/nucleotide/GroES complex is locked in position7,8,26.

Aside from rearrangements whichblock the nucleotide-binding site steri-cally, the equatorial domain does notundergo major conformational changes.However, a 4° inward tilt of the equato-rial domain of the cis ring might be suffi-cient to transmit a complementary 2°outward tilt in the equatorial domain ofthe trans ring. Xu et al. suggest thatthese small tilts cause the trans ring toadopt a conformation that is unable tobind GroES22. Although this interpre-tation remains to be verified, it helps ex-plain why ‘footballs’ (GroEL with GroESbound at each end) are difficult to ob-serve in the presence of ADP. Binding ofATP to the distal ring could result inconformational changes that counter-balance changes imposed by the cisring, thus favoring GroES binding. Thework of Rye et al.8 shows that theGroEL/ATP/GroES complex is a very sta-ble structure. Thus, at least transiently,a football structure might well form invivo, before GroES/substrate release fromGroEL’s cis ring. Perhaps GroES bindingto the distal ring is an absolute prerequi-site for the release of certain substratesby GroEL.

Evolution of Gp31 to meet the needs ofbacteriophage T4

Early genetic and biochemical workhas established that bacteriophage T4 ex-presses Gp31, a protein that is uniquelyand absolutely essential for the correctmaturation of Gp23, the major capsidprotein (reviewed in Refs 1 and 27). Thespecialized role of Gp31 is exemplifiedby the fact that certain mutant forms ofGp23 completely bypass the need for theGp31 co-chaperonin. The X-ray crystalstructure of Gp31 has recently beensolved at 2.3 Å resolution by Hunt et al.15

Surprisingly, the seven mobile loops arevisible in this structure, owing to thefact that they interdigitate loops withanother Gp31 molecule. Despite the lowsequence identity and structural differ-ences, it has been shown recently that

Gp31 can substitute for GroES in E. coliand bacteriophage l growth (F. Keppel,pers. commun.). This result emphasizesthe fact that Gp31 and GroES must be in-teracting in an analogous manner withGroEL, so that its various substrate pro-teins are properly matured. Yet somequalitative differences must exist, be-cause only Gp31 can assure the properfolding of Gp23.

It is anticipated that binding of Gp31to GroEL results in a bigger cavity. Thisis due primarily to the following factors.The mobile loop of Gp31, being muchlonger than that of GroES, will result in ataller dome structure that may allow theGroEL/Gp31 complex to more easily ac-commodate Gp23, whose 55 kDa mass is larger than the typical GroES/GroELsubstrate14,15,28. Absent from Gp31 is the

otherwise highly conserved Tyr71 resi-due, which penetrates into the centralcavity of GroES, thus, limiting the usefulspace provided by the GroES dome.Finally, the absence of the roof loop struc-ture in Gp31 may allow a GroEL/Gp31substrate to partially extend through thetop of the dome. It should be emphasizedthat there is no proof yet that any ofthese structural differences plays a role inGp23’s preferential maturation by Gp31.

Unresolved questions and future prospectsAlthough recent structural work has

provided solid answers to some of theoutstanding questions, nevertheless, adetailed understanding of the mechanismof the GroE chaperone machine awaitsanswers to a number of remaining ques-tions, including those highlighted below.

REVIEWS

(a)

(b) (c)

Figure 3 Conformational variability of the mobile loop. (a) Three different conformations of the mo-bile loop are depicted in the context of a single GroES subunit. The mobile loop from theGroEL/GroES co-crystal is shown in deep blue. Only one of seven mobile loops was in a sta-ble conformation, and hence visible, in the crystal structure of GroES; this structure isshown in black. The conformation of the mobile loop in red was determined by trNOE NMRspectroscopy in the presence of GroEL. An arrow indicates a hypothetical movement of themobile loop from the GroES-only crystal structure conformation toward the GroEL-boundconformation18. (b) A single GroES subunit is shown in cyan with the mobile loop high-lighted in deep blue in the space-filling model of the GroEL/GroES co-crystal structure. (c)A close-up view of the GroEL/GroES contact site. A single subunit of GroES is shown in cyanwith its mobile loop colored in deep blue, and the three GroES hydrophobic residues (Ile25,Val26, Leu27) are highlighted by space-filled atoms. The other space-filled residues areGroEL’s Leu234, Leu237, Val264 (red) directly bound by the deep blue GroES loop andIle305 (yellow) of the adjacent GroEL subunit.

REVIEWS TIBS 23 – APRIL 1998

142

(1) What is the exact range, rela-tive time occupancy, and biological sig-nificance of the GroEL conformationalstates? The non-hydrolyzing GroEL mu-tant (Asp398Ala) might be used to solvethe structure of the GroEL/ATP/GroEShigh-affinity state8. Perhaps in the ATPcomplex, the apical GroEL domains ro-tate even further, allowing contact of themobile loop with two adjacent GroELsubunits, thus promoting cooperativeconformational transitions.

(2) To what state of the GroE complexdoes the substrate bind? GroEL/ATP(7)/GroES or GroEL/ADP(7)/GroES? Is thedischarge of substrates from GroEL sub-units simultaneous, random, or ordered?

(3) Previous genetic analysis has high-lighted the GroEL upper-hinge region,which connects the intermediate and api-cal domains, as being pivotal for ensuring

proper interaction with GroES29. How dothese hinge mutants [such as GroEL44(Glu191Gly) and GroEL515 (Ala383Thr)]exert such profound effects on chaper-one machine action30?

(4) How many E. coli proteins dependabsolutely on the GroE chaperone ma-chine for their correct folding in vivo?Early estimates of the percentage ofE. coli proteins that utilized that GroEchaperone machine varied from 5% to50%31,32. The recent studies and calcu-lations of Hartl and colleagues28,33 sug-gest that the number of proteins thattransit through GroEL is in the range of15–30% of all of E. coli’s proteins, withthe upper limit of molecular mass being55 kDa; however, it is not clear what per-centage of the GroEL-associated pro-teins rely truly on the GroE chaperonemachine, or whether they can also fold

with the help of other chaper-one machines.

(5) Is there a biological rolefor GroES beyond being a GroELcohort? Recently, Joachimiakand Quaite-Randall obtainedevidence suggesting that GroEScontinues to ‘chaperone’ malatedehydrogenase following itsrelease from GroEL34. This ob-servation encourages specu-lation on new roles for GroES,that is, does the co-chaperoninprovide additional protectionto the substrate, independentof its cooperation with GroEL,and/or does GroES play a rolein the oligomerization of sub-strate subunits?

(6) Does GroEL undergo post-translational modifications thatenhance its activity or increaseits function? Different groupshave reported that GroEL canbe phosphoryl-ated, resultingin the expression of new or al-tered functions of GroEL, in-cluding GroES-independent sub-strate release35 and a role as anRNA chaperone36 or an effec-tor in signal transduction37.However, the roles, if any, ofGroEL phosphorylation re-main to be established.

(7) Does GroEL employ mul-tiple cohorts or does it remainfaithful in its partnership withGroES? Recent work has shownthat trigger factor (TF) is acold-inducible PPIase that inter-acts with GroEL38. In additionto increasing GroEL’s substrate-binding capacity, TF might cap-

ture ribosome-associated polypeptidesand deliver them to GroEL, as well as determine the fate of a struggling poly-peptide – whether it should remain withGroEL, move onto another chaperonemachine, or succumb to proteolysis.

(8) Fersht and colleagues have shownthat the apical domain of GroEL (residues191–345), termed a ‘minichaperone’, canassist in the maturation and renaturationof polypeptides in vitro39. The fact thatthis minichaperone cannot substitute forGroEL function in vivo (F. Keppel, pers.commun.) emphasizes the importanceof the intricate regulation of the GroEL/GroES chaperone machine in supportingE. coli growth efficiently.

(9) Finally, bacteriophage T4 Gp31 hasevolved with a specialized function thatseems to be directed towards assisting aspecific substrate, Gp23. Are there other

E

AA

I

E

I

60°90°

25°

(a)

(b) (c)Gly192Gly375

Pro137Gly410

Figure 4 GroES stabilizes/induces dramatic conformational changes in the cis ring of GroEL. (a) A schematicmodel comparing GroEL in the free state and complexed with GroES. The arrows indicate the extent ofdomain movements in the GroEL cis ring. A=apical domain (red); I=intermediate domain (green);E=equatorial domain (blue). (b) Ribbon diagram of a single GroEL subunit from the GroEL crystal struc-ture3. A single subunit, color coded as in (a), is shown in the context of the tetradecamer in the space-filling model. (c) Ribbon diagram of a single cis GroEL subunit from the GroEL–GroES co-crystal struc-ture22. The significant conformational changes in GroEL, schematized in (a), accommodate GroESbinding. Adapted, with permission, from Ref. 22.

TIBS 23 – APRIL 1998

143

FEATURE

BECAUSE BIOLOGICAL MEMBRANES arefreely permeable to water, but relativelyimpermeable to most solutes, living cellsare affected by changes in the total soluteconcentration [or, more precisely, theosmolarity (see Box 1)] of the environ-ment. In nongrowing cells, an increase inexternal osmolarity will cause loss of

water from the cell, whereas a decrease inexternal osmolarity will cause uptake ofwater. To reduce or eliminate these os-motically induced changes in the amountof cellular water, many osmoticallystressed growing cells (including E. coli)exhibit active responses in which theamounts of various cellular solutes arechanged by biosynthesis or transport.

In this article, we review recent workthat quantifies changes in the amountsof solutes and water in E. coli harvestedfrom exponential growth in conditions

of varying osmolarity, and compare themwith the passive responses of E. coli(uptake or loss of water) to changes inosmolarity under nongrowing conditions.Given that the equilibria and kinetics ofbiopolymer noncovalent interactions arestrongly dependent on concentrations (or,more precisely, thermodynamic activ-ities) of ligands, solutes and water1,2, onewould expect these changes to have largeand widespread effects on most cellularprocesses in growing cells. In Part II ofthis review (which will appear in the Mayissue of TiBS ), we summarize the effectsof the nature and concentration of soluteson noncovalent interactions in vitro, anddiscuss mechanisms that have recentlybeen put forward to explain how E. coliis able simultaneously to respond to os-motic stress, maintain or modulate bio-polymer noncovalent interactions, andmaximize its growth rate, over a widerange of osmolarities3.

Escherichia coli, a free-living, Gram-negative, enteric bacterium, is one ofthe best-characterized organisms – bothgenetically and biochemically. It does,however, present us with a fundamentalbiophysical paradox. This paradox arisesfrom the observation that E. coli utilizeslarge changes in cytoplasmic free ionconcentrations (especially the K1 con-centration) to adapt to changes in theosmolarity of the growth medium. Invitro thermodynamic studies of a widevariety of site-specific and nonspecificprotein–nucleic acid interactions, as well

Gp31-like co-chaperonins which haveevolved to deal with either specific orabundant substrates?

Clearly, one of the outright benefits of the many complexities of the GroEchaperone machine is the continuousemployment of geneticists, biochemistsand structural biologists to provide an-swers to these and additional questions,well into the 21st century.

AcknowledgementsWe thank F. Keppel, J. Hunt and

P. Viitanen for communicating unpub-lished information, Z. Xu for providingfigures, and K. Tanner for assistance inmaking figures.

References1 Georgopoulos, C. (1992) Trends Biochem. Sci.

17, 295–2992 Fayet, O., Ziegelhoffer, T. and Georgopoulos, C.

(1989) J. Bacteriol. 171, 1379–13853 Braig, K. et al. (1994) Nature 371, 578–5864 Hunt, J. F. et al. (1996) Nature 362, 121–125

5 Hartl, F. U. (1996) Nature 381, 571–5806 Fenton, W. A. and Horwich, A. L. (1997) Protein

Sci. 6, 743–7607 Todd, M. J., Viitanen, P. V. and Lorimer, G. H.

(1994) Science 265, 659–6668 Rye, H. S. et al. (1997) Nature 388, 792–7989 Martin, J. et al. (1991) Nature 352, 36–42

10 Smith, K. E. and Fisher, M. T. (1995) J. Biol.Chem. 270, 21517–21523

11 Jackson, G. S. et al. (1993) Biochemistry 32,2554–2563

12 Todd, M. J., Lorimer, G. H. and Thirumalai, D.(1996) Proc. Natl. Acad. Sci. U. S. A. 93,4030–4035

13 Mande, S. C. et al. (1996) Science 271,203–207

14 Landry, S. J., Steede, N. K. and Maskos, K.(1997) Biochemistry 36, 10975–10986

15 Hunt, J. F. et al. (1997) Cell 90, 361–37116 Koonin, E. V. and van der Vies, S. M. (1995)

Trends Biochem. Sci. 20, 1417 Landry, S. J. et al. (1993) Nature 364, 255–25818 Landry, S. J. et al. (1996) Proc. Natl. Acad. Sci.

U. S. A. 21, 11622–1162719 Goloubinoff, P. et al. (1997) FEBS Lett. 407,

215–21920 Viitanen, P. V. et al. Methods Enzymol. (in press)21 Roseman, A. M. et al. (1996) Cell 87, 241–25122 Xu, Z., Horwich, A. L. and Sigler, P. B. (1997)

Nature 388, 741–750

23 Yifrach, O. and Horovitz, A. (1995) Biochemistry34, 5303–5308

24 Lorimer, G. H. (1997) Nature 388, 720–72325 Boisvert, D. C. et al. (1996) Nat. Struct. Biol. 3,

170–17626 Hayer-Hartl, M. K., Weber, F. and Hartl, F. U.

(1996) EMBO J. 15, 6111–612127 van der Vies, S. M., Gatenby, A. A. and

Georgopoulos, C. (1994) Nature 368, 654–65628 Ewalt, K. L. et al. (1997) Cell 90, 491–50029 Fenton, W. A. et al. (1994) Nature 371,

614–61930 Zeilstra-Ryalls, J. et al. (1993) J. Bacteriol. 175,

1134–114331 Lorimer, G. H. (1996) FASEB J. 10, 5–932 Horwich, A. L. et al. (1993) Cell 74, 909–91733 Netzer, W. J. and Hartl, F. U. (1998) Trends

Biochem. Sci. 23, 68–7334 Quaite-Randall, E. and Joachimiak, A. (1997)

FASEB J. 11, 104635 Sherman, M. Y. and Goldberg, A. L. (1992)

Nature 257, 167–16936 Georgellis, D. et al. (1995) Mol. Microbiol. 16,

1259–126837 Morioka, M., et al. (1994) J. Biochem. (Tokyo) 5,

1075–108138 Kandror, O. et al. (1997) J. Biol. Chem. 272,

1730–173439 Zahn, R. et al. (1996) Proc. Natl. Acad. Sci.

U. S. A. 93, 1502–1509

Responses of E. coli to osmotic

stress: large changes in amounts of

cytoplasmic solutes and water

M. Thomas Record, Jr, Elizabeth S. Courtenay, D. Scott Cayley and Harry J. Guttman

Escherichia coli is capable of growing in environments ranging from very di-lute aqueous solutions of essential nutrients to media containing molarconcentrations of salts or nonelectrolyte solutes. Growth in environmentswith such a wide range (at least 100-fold) of osmolarities poses significantphysiological challenges for cells. To meet these challenges, E. coli ad-justs a wide range of cytoplasmic solution variables, including the cyto-plasmic amounts both of water and of charged and uncharged solutes.

M. T. Record, Jr, E. S. Courtenay,D. S. Cayley and H. J. Guttman are at theUniversity of Wisconsin-Madison, 420 Henry Mall, Madison, WI 53706, USA.

Copyright © 1998, Elsevier Science Ltd. All rights reserved. 0968 – 0004/98/$19.00 PII: S0968-0004(98)01196-7