Small heat shock proteins, ClpB and the DnaK system form a functional triade in reversing protein aggregation

(1)

Molecular Microbiology (2003) 50(2), 585–595 doi:10.1046/j.1365-2958.2003.03710.x

Blackwell Science, LtdOxford, UKMMIMolecular Microbiology 1365-2958Blackwell Publishing Ltd, 2003502585595Original ArticlesHsps, ClpB and DnaK form a functional triadeA. Mogk et al.

Accepted 7 July, 2003. *For correspondence. E-mail a.mogk@zmbh.uni-heidelberg.de; Tel. (+49) 6221 546863; Fax (+49) 6221 545894.

Small heat shock proteins, ClpB and the DnaK system form a functional triade in reversing protein aggregation

Axel Mogk,^1,2* Elke Deuerling,¹

Sonja Vorderwülbecke,¹ Elizabeth Vierling³ and Bernd Bukau¹*

1ZMBH, Universität Heidelberg, Im Neuenheimer Feld 282, Heidelberg D-69120, Germany.

2Institut für Biochemie und Molekularbiologie, Hermann-Herder-Str. 7, D-79104 Freiburg, Germany.

3Department of Biochemistry and Molecular Biophysics, University of Arizona, 1007 E. Lowell St., Tucson AZ 85721, USA.

Summary

Small heat shock proteins (sHsps) can efficiently pre- vent the aggregation of unfolded proteins in vitro. However, how this in vitro activity translates to func- tion in vivo is poorly understood. We demonstrate that sHsps of Escherichia coli, IbpA and IbpB, co-operate with ClpB and the DnaK system in vitro and in vivo, forming a functional triade of chaperones. IbpA/IbpB and ClpB support independently and co-operatively the DnaK system in reversing protein aggregation.

A

D D D DibpAB

DDD

DclpB double mutant exhibits strongly increased protein aggregation at 42

∞∞∞∞C compared with the single mutants. sHsp and ClpB function become essential for cell viability at 37

∞∞∞∞C if DnaK levels are reduced. The DnaK requirement for growth is increas- ingly higher for

D D D DibpAB,

D D D

DclpB, and the double

DDD Dib- pAB

D D D

DclpB mutant cells, establishing the positions of sHsps and ClpB in this chaperone triade.

Introduction

The ensemble of molecular chaperones and proteases constitutes the cellular system for de novo folding and quality control of proteins (Hartl, 1996; Gottesman et al., 1997; Wickner et al., 1999; Dougan et al., 2002). This system is operative under regular growth conditions but becomes particularly important under stress conditions, such as heat shock. The functions of the major cytosolic chaperone and protease systems during heat stress are best understood in Escherichia coli. The GroE and the

DnaK chaperone systems prevent the aggregation of thermolabile proteins and support their refolding to the native state (Langer et al., 1992; Viitanen et al., 1992; Horwich et al., 1993; Schröder et al., 1993). The DnaK system in addition disaggregates small aggregates of heat- denatured proteins (Schröder et al., 1993; Skowyra et al., 1990; Mogk et al., 1999; Diamant et al., 2000). The AAA+ chaperone ClpB allows the DnaK system to efficiently disaggregate even large aggregates of misfolded proteins (Goloubinoff et al., 1999; Mogk et al., 1999; Zolk- iewski, 1999), similar to the homologous Hsp104 and Hsp70 chaperone systems of Saccharomyces cerevisiae (Glover and Lindquist, 1998). Non-native proteins may alternatively be degraded by proteases. Lon, ClpAP, ClpXP and HslUV represent the major ATP-dependent proteolytic systems of the E. coli cytosol and act synergistically in vivo in the degradation of abnormal proteins (Kanemori et al., 1997). Chaperones and proteases also exhibit synergistic and overlapping functions in the protein quality control system. Importantly, ClpXP and Lon become essential for viability at high temperatures under conditions of limiting amounts of the DnaK system (Tomo- yasu et al., 2001). Depletion of the DnaK system (KJE) in clpXP and lon mutant cells caused strong protein aggregation, indicating overlapping functions of proteases and KJE in preventing protein aggregation during heat stress.

The role of small heat shock proteins (sHsps) in the protein quality control network is poorly understood. The lack of strong phenotypes of mutants lacking sHsp activity has prevented a clear assignment of sHsp function in the chaperone network. While overproduction of sHsps in different cells including E. coli can increase stress tolerance, sHsp function is not essential for thermotolerance in S.

cerevisiae or E. coli (Petko and Lindquist, 1986; Thomas and Baneyx, 1998; Kitagawa et al., 2000). Temperature- sensitive growth phenotypes, caused by missing sHsp function, have only been reported for Neurospora crassa (Plesofsky-Vig and Brambl, 1995) and Synechocystis sp.

PCC6803 (Giese and Vierling, 2002). The behaviour of sHsps in vivo also appears to contrast with their reported in vitro chaperone activity. In vitro sHsps can efficiently prevent the aggregation of heat denatured proteins (Hor- witz, 1992; Jakob et al., 1993; Lee et al., 1997). In vivo, however, sHsps frequently localize to insoluble protein fractions of heat stressed cells. Even at normal growth First publ. in: Molecular Microbiology 50 (2003), 2, pp. 585-595

Konstanzer Online-Publikations-System (KOPS) URL: http://www.ub.uni-konstanz.de/kops/volltexte/2007/3719/

URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-37190

(2)

586 A. Mogk et al.

temperatures in the presence of high levels of unfolded proteins sHsps will become insoluble. Indeed they were first defined in E. coli as ‘inclusion body binding proteins’

(IbpA and IbpB), accumulating in association with over- produced heterologous proteins in inclusion bodies (Allen et al., 1992; Laskowska et al., 1996). Because solubilization of stress-induced aggregated proteins is still possible in DibpAB mutant cells, the functional importance of the proposed chaperone activity of sHsps in E. coli has remained in question (Mogk et al., 1999; Kuczynska- Wisnik et al., 2002).

Here we link the function of sHsps to the chaperone network of E. coli both in vitro and in vivo. We demonstrate that sHsps co-operate with ClpB and the DnaK system in reversing protein aggregation. In vitro refolding of substrates bound to IbpB is mediated most efficiently by ClpB and KJE. Presence of IbpA/IbpB in insoluble protein aggregates accelerated the ClpB/KJE mediated disaggregation reaction in vivo. Functions of sHsps and ClpB become essential for cell viability and protein resolubilization at elevated temperatures under conditions of limiting DnaK levels. These data provide direct evidence for a functional triade of sHsp, ClpB and DnaK in reversing protein aggregation in vivo.

Results

ClpB stimulates the KJE-dependent refolding of substrates bound to IbpB

sHsps are found associated with aggregated proteins in heat stressed E. coli cells (Allen et al., 1992; Laskowska et al., 1996; Mogk et al., 1999). We therefore tested whether the chaperone ClpB, which acts to dissociate protein aggregates, can affect in vitro the refolding of substrates from sHsp/substrate complexes using thermolabile malate dehydrogenase (MDH) and firefly luciferase as model substrates (Fig. 1). In these studies, we did not include IbpA because this protein could not be overpro- duced and purified in a soluble state, as observed earlier (Shearstone and Baneyx, 1999). MDH (2 mM) was pro- tected from heat-induced aggregation at 47∞C by IbpB (16 mM). Reactivation of MDH from soluble MDH/IbpB complexes was only observed in the presence of the DnaK system (KJE), similar to published results (Veinger et al., 1998). Denaturation of MDH in the presence of sHsps was a prerequisite for KJE activity, as aggregated MDH was no longer refolded by KJE (data not shown).

The GroE system (ELS) and ClpB, alone or in combina- tion, did not exhibit refolding activities. However, addition of either ELS or ClpB accelerated the KJE-dependent refolding of MDH, increasing the refolding rate by 1.7-fold to 3.9 nM MDH min^-¹ (Fig. 1A). Importantly, together ELS and ClpB co-operatively stimulated the KJE-mediated

reactivation of MDH bound to IbpB. The refolding rate increased to 14.5 nM MDH min^-¹ and complete refolding of sHsp-bound MDH was observed within 60 min. These results indicate that ESL and ClpB act during different stages of the MDH reactivation process.

Similar data were obtained when luciferase was used as an alternative substrate (Fig. 1B). Heat denaturation of luciferase (0.2 mM) in the presence of IbpB (1.6 mM) allowed substrate refolding only by KJE but not by ELS or ClpB alone. Protection of luciferase from aggregation by IbpB was again essential for the subsequent KJE-dependent refolding as KJE did not support luciferase reactivation from aggregates (data not shown). In the case of Fig. 1. ClpB stimulates the KJE-dependent refolding of substrates bound to IbpB.

A. 2 mM malate dehydrogenase (MDH) was heat denatured at 47∞C in presence of 16 mM IbpB resulting in the formation of soluble IbpB/

MDH complexes. MDH refolding was started by diluting the IbpB/

MDH complexes 1:1 with the indicated cytosolic Escherichia coli chaperones at 30∞C (KJE: 1 mM DnaK, 0.2 mM DnaJ, 0.1 mM GrpE;

ELS: 4 mM GroEL, 4 mM GroES; ClpB: 1.5 mM ClpB) and MDH activities were determined at the indicated time points. The enzymatic activity of native MDH was set at 100%.

B. 0.2 mM luciferase was heat-denatured at 43∞C in the presence of 1.6 mM IbpB, resulting in the formation of soluble IbpB/luciferase complexes. Luciferase refolding was initiated by diluting the IbpB/

luciferase complexes 1:1 with the indicated chaperones (KJE: 0.5 mM DnaK, 0.1 mM DnaJ, 0.05 mM GrpE; ELS: 2 mM GroEL, 2 mM GroES;

ClpB: 1 mM ClpB) and luciferase activities were determined at the indicated time points. The enzymatic activity of native luciferase was set at 100%.

(3)

sHsps, ClpB and DnaK form a functional triade 587 IbpB/luciferase complexes ELS did not influence the

KJE-dependent refolding process, indicating that the beneficial effects of ELS on MDH reactivation are specific for this substrate. In contrast ClpB accelerated the KJE- dependent refolding of luciferase increasing the refolding rates by 2.9-fold (0.59 nM luciferase min^-¹). These data demonstrate that, in general, ClpB co-operates with KJE in the reactivation of sHsp-bound substrates in vitro.

DibpAB mutant cells exhibit delayed resolubilization of protein aggregates

Our in vitro data suggest that sHsp/substrate complexes are substrates for the ClpB/KJE bi-chaperone system. To obtain evidence that sHsps also facilitate the action of ClpB/KJE in vivo, we followed the kinetics of protein disaggregation after severe heat shock treatment (45∞C for 30 min) in E. coli wild-type and DibpAB mutant cells. While the total level of heat-aggregated proteins was not affected in the DibpAB mutant, the subsequent solubilization of protein aggregates was reproducibly delayed compared with wild-type cells (Fig. 2). The defect of DibpAB mutant cells was mild compared with DclpB and DdnaK52 mutant cells which showed almost no protein disaggregation after heat-stress to 45∞C (see Fig. 3; Mogk et al., 1999). We conclude that presence of sHsps in protein aggregates is not essential for protein disaggregation in vivo but can accelerate the ClpB/KJE-mediated disaggre- gation process.

Fate of sHsp/substrate complexes in vivo

The precise distribution of sHsps in soluble and insoluble protein fractions of heat-treated cells, which contained or

lacked the activities of KJE and ClpB, and the processes which lead to dissociation of sHsps from aggregated proteins in vivo are unknown. We therefore followed the fate of sHsps upon heat treatment in E. coli wild-type and DclpB and DdnaK52 null mutant cells by quantitative immunoblotting. After heat shock to 45∞C, more than 50%

of IbpA/IbpB was present in the insoluble protein fraction of wild-type cells (Fig. 3A–C). This value was increased to over 80% of total IbpA/IbpB in DclpB and DdnaK52 mutant cells, which are affected in preventing and reversing protein aggregation (Mogk et al., 1999). Interestingly, increased amounts of IbpA/IbpB, observed in DclpB and DdnaK52 mutant cells after heat stress, could not prevent heat-induced protein aggregation.

Disappearance of IbpA/IbpB from the insoluble cell fraction during the recovery phase at 30∞C occurred in wild- type cells with the same kinetics as the solubilization of protein aggregates (data not shown). Such solubilization of sHsps from protein aggregates was slowed down by threefold in DclpB and nearly abolished in DdnaK52 mutant cells (Fig. 3D). Identical results were obtained when the fate of [³⁵S]-methionine pulse-labelled IbpA/IbpB was followed in wild-type and chaperone mutant cells (data not shown). These in vivo findings parallel those obtained in vitro in that dissociation of sHsp/substrate complexes was strictly dependent on KJE and accelerated by the additional presence of ClpB.

Analysis of DibpAB DclpB double mutants

Despite the noticed co-operation of IbpA/IbpB and ClpB in KJE-mediated protein disaggregation, DibpAB and DclpB mutants are viable at 42∞C (Thomas and Baneyx, 1998). To determine whether both chaperone systems act synergistically in vivo, we constructed a DibpAB DclpB double mutant. Double mutants were characterized with

Fig. 2. Escherichia coli cells lacking sHsp func- tion exhibit delayed resolubilization of protein aggregates. Cultures of E. coli wild type and DibpAB mutant strains were grown in LB medium at 30∞C to logarithmic phase, then shifted to 45∞C for 30 min, followed by incubation at 30∞C for 60 min. Aggregated proteins were isolated at the indicated time points and analysed by SDS–PAGE followed by staining with Coomassie brilliant blue. The amount of aggregated proteins was quantified by Bradford assay and calculated in relation to the total protein content (set at 100%). Asterisk indicates the position of IbpA/IbpB in the wild-type proteins samples.

(4)

588 A. Mogk et al.

respect to growth at high temperatures, formation of protein aggregates after heat shock to 42∞C and the development of thermotolerance. Growth of DibpAB DclpB double mutant cells was still unaffected at 42∞C (data not shown).

However, the absence of IbpA/IbpB and ClpB function caused increased protein aggregation at 42∞C, which was not observed for the single mutant strains (Fig. 4). Levels of aggregated proteins in DibpAB DclpB mutant cells were still low compared with those found in DdnaK52 mutants.

To test for the ability to develop thermotolerance, we determined the survival rates of E. coli wild-type, DibpAB, DclpB and DibpAB DclpB mutant cells either directly upon heat shock from 30∞C to 50∞C, or after applying a 15 min pre- shock to 42∞C for cell adaptation (Fig. 5A–B). In both

cases, DibpAB mutants reproducibly exhibited a 1.5-fold reduced survival rate, indicating a contribution of IbpA/

IbpB in the development of thermotolerance. The defects of DibpAB mutant cells were mild compared with DclpB mutant cells, which revealed a five- to 10-fold reduced survival after heat-shock to 50∞C. Viability of DclpB mutant cells was also largely unaffected by the 42∞C pre-shock in contrast to wild-type and DibpAB mutants. Survival of DibpAB DclpB double mutant cells at 50∞C was slightly reduced compared with DclpB mutants if cells were directly heat-shocked to 50∞C. Together these findings indicate an activity of sHsps in preventing and/or reversing protein aggregation. The dominating effect in thermotolerance development, however, is provided by ClpB.

Fig. 3. Fate of E. coli sHsps (IbpA/IbpB) during heat stress.

A. Cultures of wild-type, DclpB and DdnaK mutant strains were grown in LB medium at 30∞C to logarithmic phase, then shifted to 45∞C for 30 min, followed by incubation at 30∞C for 2 h. Aggregated proteins were isolated at the indicated time points and analysed by SDS–

PAGE followed by staining with Coomassie brilliant blue. The amount of aggregated proteins was quantified by Bradford assay and calculated in relation to the total protein content (set at 100%).

B–C. The distribution of IbpA/IbpB in soluble (white bars) and insoluble (black bars) cell fractions of E. coli wild-type, DclpB and DdnaK mutant strains was determined by quantitative immunoblot analysis using IbpB-specific antibodies.

D. IbpA/IbpB levels of pellet fractions were calculated in relation to the sHsp amount, determined directly after heat-shock to 45∞C (set at 100% for each strain).

(5)

sHsps, ClpB and DnaK form a functional triade 589

Lack of ClpB and sHsp function influences the heat shock response

The mildness of phenotypes of DibpAB, DclpB and DibpAB DclpB mutant cells might result from a regulatory adjustment in the mutant cells involving overproduction of other chaperones to compensate for the lack of IbpA/IbpB and ClpB activity. In particular, increased levels of the KJE system would reduce the levels of misfolded proteins and the formation of protein aggregates and thereby decrease the necessity for chaperones acting at the level of protein aggregates. To test this possibility, we analysed the levels of DnaK at 37∞C in E. coli wild-type and chaperone mutant cells (Fig. 6A). Immunoblot analysis revealed that DnaK levels were indeed 1.5- to twofold elevated at 37∞C in DclpB and DibpAB DclpB mutant cells compared with wild- type cells, whereas the levels of Trigger factor, which is not regulated as part of the heat shock response, remained constant (Fig. 6A). These increased levels of DnaK, and conceivably also of other heat shock proteins, are expected to mask potential phenotypes of DibpAB and DclpB mutant cells.

It is also possible that the kinetics of the heat shock response is affected in cells lacking sHsp and/or ClpB function. To test this possibility E. coli wild-type, DibpAB, DclpB and DibpAB DclpB mutant cells were grown at 30∞C in M9 minimal medium and heat-shocked to 42∞C (Fig. 6B–C). The heat shock response was transient for E. coli wild-type and DibpAB mutant cells with a peak in GroEL synthesis at 10 min after upshift. In contrast, DclpB and DclpB DibpAB mutant cells exhibited a signifi-

cant delay in the shut-off phase of the heat shock response (Fig. 6B and C). Downregulation of Hsp synthesis is mediated by the DnaK chaperone system (Straus et al., 1990). A delayed downregulation suggests that especially the lack of ClpB function leads to a requirement for increased DnaK levels to cope with heat stress at 42∞C, which in turn might compensate for lacking ClpB activity.

IbpA/IbpB and ClpB become essential for viability in cells with reduced DnaK levels

Based on the above findings, we tested whether increased levels of KJE can compensate for IbpA/IbpB and ClpB. We took advantage of an E. coli strain in which the dnaKJ operon is placed under lacI control (KJ- Fig. 4. Escherichia coli cells lacking sHsp and ClpB function exhibit

protein aggregation at 42∞C. E. coli wild-type, DibpAB, DclpB, DibpAB DclpB and DdnaK mutant cells were grown in LB medium at 30∞C to mid-exponential phase and shifted to 42∞C for 30 min. Aggregated proteins were isolated and analysed by SDS–PAGE followed by staining with Coomassie brilliant blue. The amount of aggregated proteins was quantified by Bradford assay and calculated in relation to the total protein content (set at 100%). The position of IbpA/B is indicated.

Fig. 5. Escherichia coli cells missing sHsp or ClpB function exhibit reduced thermotolerance. E. coli wild-type, DibpAB, DclpB and Dib- pAB DclpB mutant strains were grown in LB medium at 30∞C to mid- exponential phase. 10^-2 dilutions of each strain were heat-shocked to 50∞C and incubated for the indicated time (A). Alternatively aliquots of all strains were first shifted to 42∞C for 15 min to allow induction of the heat shock response and transferred to 50∞C afterwards (B).

Various dilutions of the stressed cells were spotted on LB plates and incubated at 30∞C. After 24 h, colony numbers were counted, and survival rates were calculated in relation to unstressed cells. Mean values of three independent experiments are shown. The average variation between the experiments was less than 20%.

(6)

regulatable strain) and thus is no longer subject to heat shock regulation (Tomoyasu et al., 1998). This strain allows adjustment of the cellular levels of DnaK and DnaJ according to the IPTG concentration in the growth medium. We introduced the DibpAB, DclpB or DibpAB DclpB mutations into this strain background and determined whether the levels of DnaK and DnaJ affect cell viability on LB agar plates at various growth temperatures (Fig. 7). At 30∞C, all three KJ-regulatable mutant strains (DibpAB, DclpB, DibpAB DclpB) remained viable even in the absence of IPTG (i.e. with greatly reduced KJ levels). At 37∞C, none of the mutants was able to form colonies in the absence of IPTG, while the KJ-regulatable wild-type control strain showed no growth defects.

The lack of IbpA/IbpB and/or ClpB function thus causes synthetic lethality in cells with limiting KJ levels (Fig. 7).

Our experiment further revealed that different KJ levels were required to compensate for missing IbpA/IbpB or ClpB function. Growth of DibpAB mutants was restored at low IPTG concentrations (40 mM), which allows for KJ

production to levels found in wild-type cells at 30∞C.

Growth of DclpB mutants required 70 mM IPTG, allowing for production of DnaK and DnaJ to increased levels and even higher IPTG concentrations (80 mM) were neces- sary for growth of DibpAB DclpB double knockout cells.

At 42∞C, the KJ-regulatable wild-type strain also required IPTG for growth, as expected from the known temperature sensitive growth phenotype of dnaK mutant cells.

The different mutant cells showed a similar relative requirement for KJ as compared with 37∞C, except that the IPTG threshold concentration was generally higher.

Taking these data together, the DibpAB, DclpB or DibpAB DclpB mutant cells require KJ function for growth at

≥37∞C. This requirement follows a hierarchy, with Dib- pAB, DclpB and DibpAB DclpB mutant cells showing increasing requirements in that order, indicating that the IbpA/IbpB, ClpB and KJE chaperone systems form a co- operative functional network that is essential for viability at regular growth temperature and under heat shock conditions.

Fig. 6. Loss of ClpB and IbpA/IbpB function alters the heat shock response.

A. Levels of DnaK and Trigger factor were determined by immunoblotting of total lysates (5 mg) prepared from E. coli wild-type (WT), DibpAB, DclpB and DibpAB DclpB mutant cells grown in LB medium at 37∞C.

B. E. coli wild-type (WT), DibpAB, DclpB and DibpAB DclpB mutant cells were grown at 30∞C in M9 minimal medium without ^L-methionine to mid- exponential growth phase and shifted to 42∞C. At the indicated time points, aliquots were pulse-labelled with [³⁵S]-L-methionine (7.5 mCi) for 1 min and subsequently precipitated by TCA (10% (v/v)). Samples were subjected to SDS–PAGE (12%) followed by development with a phosphoimager and synthesis rates of GroEL were determined (C). Values were normalized to the maximal value of E. coli WT (set as 100%). The positions of GroEL and DnaK are indicated.

(7)

sHsps, ClpB and DnaK form a functional triade 591

IbpA/IbpB, ClpB and the DnaK system act co-operatively to reverse protein aggregation

To elucidate the functional interplay between IbpA/IbpB, KJE and ClpB in the protein quality control network more precisely, we determined the degree of protein aggregation in DibpAB, DclpB and DibpAB DclpB mutants that have KJ adjusted to various levels (Fig. 8). Because ClpB and IbpA/IbpB do not prevent protein aggregation in vivo (Mogk et al., 1999), increased amounts of aggregated proteins in the respective mutants would indicate a less efficient protein disaggregation. At 30∞C no protein aggregation was detectable for all tested mutant strains, even in cells with greatly reduced KJ levels. After a 30 min incubation at 42∞C, 5% of cellular proteins aggregated in all mutant cells, provided that IPTG was omitted from the growth medium (Fig. 8). Increasing KJ levels (by addition of IPTG) reduced the amount of aggregated proteins in each strain, but to different degrees dependent on the mutant background. While 50 mM IPTG in the growth medium was sufficient to eliminate aggregates in DibpAB cells, 2% and 5% of total proteins still aggregated in DclpB and DibpAB DclpB mutant cells respectively. Even in presence of DnaK/DnaJ levels corresponding to heat shock conditions (100 mM IPTG), 2% of cellular proteins remained aggregated in DibpAB DclpB mutant cells.

These findings are in complete agreement with the hierarchal complementation of growth defects of these mutant cells at high temperatures and demonstrate the co-

operative action of IbpA/IbpB and ClpB in the KJE- mediated removal of protein aggregates in vivo.

Discussion

In this study, we analysed the functions of sHsps in the protein quality control system of E. coli. We provide in vitro and in vivo evidence that sHsps co-operate with the ClpB/

KJE bi-chaperone system in reversing protein aggregation. This functional triade of chaperones is operative at regular growth temperatures and during heat stress and becomes essential if KJE levels are limiting.

Using MDH and luciferase as thermolabile model substrates, we could demonstrate that in vitro refolding of substrates bound to IbpB is exclusively mediated by KJE or, more efficiently, by ClpB/KJE. The dependency of substrate refolding on KJE and ClpB/KJE can likely be explained by the high stability of sHsp/substrate complexes (Ehrnsperger et al., 1999). Indeed, substrates bound to sHsps are not released spontaneously to signif- icant amounts (Mogk et al., 2003). We therefore suggest that sHsp/substrate complexes represent protein aggregates and, consequently, substrate refolding relies on the active extraction of substrates by KJE or ClpB/KJE.

While efficient in forming soluble complexes with unfolded proteins in vitro, sHsps extensively co-localize with insoluble protein aggregates in heat treated E. coli cells. In view of our findings, this suggests that their pri- Fig. 7. sHsps and ClpB become essential at elevated temperatures in E. coli cells with reduced DnaK/DnaJ levels. BB7352 (P_A1/lacO-1 dnaK, dnaJ, lacI^q), BB6406 (PA1/lacO-1 dnaK, dnaJ, lacI^q, DibpAB), BB6410 (PA1lacO-1

-1 dnaK, dnaJ, lacI^q, DclpB) and BB6411 (PA1/lacO-1 dnaK, dnaJ, lacI^q, DibpAB, DclpB) cells were grown at 30∞C in LB medium supplemented with 1 mM IPTG until late log phase. Various dilutions (10^-3-10^-6) were spotted on LB plates containing the indicated IPTG concentrations and incubated at 30∞C, 37∞C or 42∞C for 24 h.

(8)

mary function is to facilitate the solubilization process of aggregated proteins. Although it is difficult to determine the nature of the in vivo formed sHsp/substrate complexes, a large heterogenity of species can be assumed.

In E. coli wild-type cells, approximately 50% of IbpA/IbpB was found associated with aggregated proteins after heat stress, while the remaining sHsps stayed soluble, poten- tially representing smaller sHsp/substrate complexes.

Based on our in vitro findings it is likely that the size of such aggregates determines the chaperone requirement for solubilization of the aggregated proteins. In vitro KJE alone can dissociate small, soluble sHsp/substrate complexes, whereas ClpB is additionally needed for efficient solubilization of insoluble sHsp/substrate complexes. Fur- thermore, the presence of sHsps in protein aggregates aggregates facilitates the solubilization process by ClpB/

KJE (Mogk et al., 2003). This latter finding parallels the delayed protein disaggregation in DibpAB mutant cells (Fig. 2).

The analysis of dissociation of IbpA/IbpB from aggregated proteins in DclpB mutant cells revealed a slow disaggregation reaction, which was no longer detectable in DdnaK mutant cells (Fig. 3). This indicates that dissocia-

tion of sHsp/substrate complexes by DnaK alone is possible in vivo. However, solubilization of such complexes was much faster in the presence of ClpB and represented only a subfraction of stress-induced protein aggregates.

Thus ClpB and sHsps make different contributions to the KJE-mediated protein disaggregation. While ClpB plays a direct and dominant role in this process, sHsps exert their function indirectly by facilitating the extraction of unfolded proteins from aggregates. This hierarchal function of IbpA/

IbpB and ClpB in the chaperone triade is also indicated by increased requirement for KJE if one member of the network is missing. Importantly, by influencing different aspects of the disaggregation reaction sHsps and ClpB can work co-operatively and, consistent with this sugges- tion, DibpAB DclpB double mutants needed the highest KJE levels for growth at 37∞C or 42∞C and exhibited strongly increased protein aggregation compared with the single knockout strains (Figs 7 and 8). Recent findings indicate that the functional interplay between sHsps, ClpB and KJE is not restricted to E. coli, but may represent a conserved principle in bacteria to guarantee efficient protein disaggregation. Co-operation of sHsps and ClpB has been suggested in the development of thermotolerance in Fig. 8. Escherichia coli cells missing sHsp and ClpB function accumulate aggregated proteins at high temperatures. BB7352 (PA1/lacO-1 dnaK, dnaJ, lacI^q), BB6406 (P_A1/lacO-1 dnaK, dnaJ, lacI^q, DibpAB), BB6410 (PA1/lacO-1 dnaK, dnaJ, lacI^q, DclpB) and BB6411 (PA1/lacO-1 dnaK, dnaJ, lacI^q, DibpAB, DclpB) cells were grown at 30∞C overnight in LB medium supplemented with 1 mM IPTG. Pre-cultures were washed twice with LB medium and inoculated into LB medium containing the indicated IPTG concentrations. Cells were cultured further to mid-exponential growth phase and shifted to 42∞C for 30 min or left at 30∞C. Aggregated proteins were isolated and analysed by SDS–PAGE followed by staining with Coomassie brilliant blue. The amount of aggregated proteins was quantified by Bradford assay and calculated in relation to the total protein content (set at 100%). Two prominent bands visible at 30∞C were identified as membrane proteins OmpA and OmpF and do not represent protein aggregates.

(9)

sHsps, ClpB and DnaK form a functional triade 593 Synechocystis sp. PCC6803 (Giese and Vierling, 2002).

The existence of the sHsp/ClpB/KJE triade in Mycobacte- rium tuberculosis is also suggested by their specific HspR-dependent co-regulation (Stewart et al., 2002).

The functions of sHsps and ClpB in KJE-mediated protein disaggregation are most important during severe stress; however, this network is already operative at 37∞C.

The absence of phenotypes in DibpAB and DclpB mutant cells can be explained by a redundancy of the cellular protection machinery. The high abundance of DnaK in the E. coli cytosol guarantees a high potential to prevent pro- tein aggregation. However, we could show that sHsps or ClpB become essential for viability of E. coli cells at regular and heat shock temperatures under conditions of low DnaK levels. Small changes in the DnaK/J levels had dramatic effects on the viability of the chaperone mutant cells as compared with wild-type control cells. Depletion of DnaK increases protein aggregation and at the same time reduces the disaggregation potential of the cells.

Such conditions may exist in cells subjected to prolonged or repeated stress treatment. Importantly, increased levels of other chaperone systems or proteases, caused by heat shock induction in presence of reduced DnaK concentrations, cannot replace the missing sHsp or ClpB function.

The co-aggregation of IbpA/IbpB with misfolded proteins during heat stress is likely explained by low sHsp levels in unstressed E. coli cells. sHsp levels become abundant only after induction of the heat shock response (Mogk et al., 1999; Richmond et al., 1999). We speculate, that most bacteria tend to keep the sHsp concentration low, in order to direct the flux of unfolded proteins to chaperones with folding capacity such as ELS and KJE.

Interestingly, high sHsp levels can also interfere with protein disaggregation, as overproduction of IbpA/B retarded the resolubilization of protein aggregates in vivo (data not shown; Kuczynska-Wisnik et al., 2002). Furthermore, sHsps assemble into oligomeric structures that may con- stitute inactive resting states. Activation appears to involve dissociation of the oligomers into dimers; for some sHsps heat shock temperatures shifts the oligomer-dimer equi- librium to the active dimer form (Haslbeck et al., 1999; van Montfort et al., 2001). Therefore increased levels and activation of sHsps seem to be restricted to stress conditions.

We also noticed a third strategy to ensure low sHsp levels even after strong heat shock induction: only a fraction of IbpA/IbpB, after the dissociation from protein aggregates, was regained in the soluble cell fraction, indicating degradation of sHsps after dissociation from the aggregates (Fig. 3B and C). Proteolysis of IbpA/IbpB would guarantee that high sHsp levels are restricted to heat stress conditions, but are actively reduced during the recovery phase of cells.

Together these findings demonstrate how the quality control system of cells regulates and uses the chaperone

activity of sHsps. Different mechanisms guarantee low levels of active sHsps under non-stress conditions to pro- tect cells from their extraordinary binding capacity. Induc- tion of sHsps at stress conditions allows association with unfolded proteins destined for aggregation, if other chaperone systems are overburdened. This co-aggregation represents a second defence line and has important contributions to survival of cells during severe stress by accel- erating the ClpB/KJE-dependent protein disaggregation.

Experimental procedures Strains and culture conditions

E. coli strains used were derivatives of MC4100 [araD139 D(argF-lac)U169 rpsL150 relA1 flbB5301 deoC1 ptsF25 rbsR]. Mutant alleles of chaperone genes were introduced by P1 transduction into MC4100 background to generate strains BB4561 (DclpB::kan), BB4563 (DibpAB::kan), BB4566 (DibpAB::kan, zic4901::Tn10, DclpB::kan), BB2395 (lon146::Tn10), BB7999 clpP::kan, BB7352 (PA1/lacO-1 dnaK, dnaJ, lacI^q), BB6406 (PA1/lacO-1 dnaK, dnaJ, lacI^q, DibpAB), BB6410 (PA1/lacO-1 dnaK, dnaJ, lacI^q, DclpB) and BB6411 (PA1/lacO-1 dnaK, dnaJ, lacI^q, DibpAB, DclpB). Bacterial strains were cultured at 30∞C in Luria broth (LB) or M9 minimal medium supplemented with glucose (0.2%) and all L-amino acids (30 mg ml^-1) except L-methionine. Chloramphenicol, kanamycin and tetracycline were used at final concentrations of 10, 20, and 10 mg ml^-1 respectively. Temperature-shift experiments were performed in orbital shaking water baths.

For pulse-labelling with [³⁵S]-L-methionine, cells were grown in M9 minimal medium (see above). Labelling was done by adding [³⁵S]-L-methionine to heat shocked cells (Amersham SJ1515; 15 mCi ml^-1, 1000 Ci mmol^-1) to 30 mCi ml^-1 cell culture for 5 min, followed by addition of unlabeled L-methionine to 200 mg ml^-1.

Proteins

Purifications of DnaK, DnaJ, GrpE, ClpB, GroEL, GroES and IbpB were performed as described previously (Goloubinoff et al., 1999; Mogk et al., 1999). Pyruvate kinase was pur- chased from Sigma, pig heart muscle MDH and firefly Luciferase from Roche. Protein concentrations were determined with the Bio-Rad Bradford assay using BSA as standard. Protein concentrations refer to the protomer.

Protein denaturation and chaperone activity assays MDH (2 mM) was denatured at 47∞C for 30 min in buffer A (50 mM Tris, pH 7.5, 150 mM KCl, 20 mM MgCl2, 2 mM DTT) in the presence of IbpB (16 mM). Luciferase (0.2 mM) was denatured at 43∞C for 15 min in buffer A in the presence of 1.2 mM IbpB. Protein refolding was started by diluting IbpB/

MDH or IbpB/luciferase complexes 1:1 into the indicated final concentrations of chaperones in buffer A at 30∞C. All assays were performed in presence of an ATP-regenerating system (3 mM phosphoenol pyruvate; 20 mg ml^-1 pyruvate kinase;

2 mM ATP). Determination of enzymatic activities followed

(10)

published protocols (Schröder et al., 1993; Goloubinoff et al., 1999). Refolding rates were calculated from the linear increase of substrate activities.

Determination of protein synthesis rates

Cells were grown at 30∞C to mid-exponential phase in M9 minimal medium (see above). Cells were heat-shocked to 42∞C and labelled with 150 mCi ml^-1 [35S]-L-methionine (Amersham SJ1515, 15 mCi ml^-1) at the indicated time points for 1 min. The pulse-labelled cells were immediately precipitated with trichloracetic acid (TCA; 10% final concentration), the protein pellet was washed with acetone, dissolved in sample buffer and further prepared for SDS–PAGE. Dried gels were scanned and the incorporated radiolabel was quantified by MacBAS software (Fuji Film).

Plating efficiency, spot test and isolation of aggregated proteins

Survival of E. coli cells after exposure to lethal temperature was determined by calculating the plating efficiency. Cells were grown in LB medium to mid-exponential growth phase at 30∞C. Aliquots were diluted to 10^-2 in LB medium and incubated at 50∞C for the indicated time. After further serial dilution from 10^-3 to 10^-7 in LB medium, 10 ml cells were spotted onto LB plates and incubated for 24 h at 30∞C and colony numbers were determined afterwards. Synthetic lethality of missing DnaK and sHsp or ClpB function was followed by spot tests. Spot tests were performed using cells cultured in LB medium containing 1 mM IPTG. Aliquots of the cultures were diluted from 10^-3 to 10^-6 in LB medium and aliquots (10 ml) were spotted onto LB plates containing different IPTG concentrations as indicated. Plates were incubated at 30∞C, 37∞C or 42∞C for 24 h, and colony numbers were determined. Defects in protein disaggregation in chaperone mutant strains were followed by isolation of aggregated proteins as described (Tomoyasu et al., 2001).

SDS–PAGE, immunoblotting and protein quantifications Gel electrophoresis was carried out according to standard protocols (Laemmli, 1970) using 15% SDS-polyacrylamide gels and stained with Coomassie brilliant blue. Immunoblot- ting was performed according to standard procedures, using rabbit anti-sera specific for DnaK, Trigger factor or IbpB (exhibiting strong cross-reactivity to IbpA) as primary anti- body, and developed with a Vistra ECF fluorescence immunoblotting kit (Amersham) using alkaline phosphatase/

conjugated anti-rabbit IgG as secondary antibodies (Vector Laboratories). Developed immunoblots were scanned using a fluoroimager (FLA-2000) and quantified using MacBAS software (Fuji Film).

Acknowledgements

We thank D. Dougan and K. Turgay for discussions and critical reading of the manuscript; A. Schulze-Specking for technical assistance. This work was supported by grants from

the DFG (Bu617/14-1) and the Fond der Chemischen Indus- trie to B.B. and A.M.

References

Allen, S.P., Polazzi, J.O., Gierse, J.K., and Easton, A.M.

(1992) Two novel heat shock genes encoding proteins produced in response to heterologous protein expression in Escherichia coli. J Bacteriol 174: 6938–6947.

Diamant, S., Ben-Zvi, A.P., Bukau, B., and Goloubinoff, P.

(2000) Size-dependent disaggregation of stable protein aggregates by the DnaK chaperone machinery. J Biol Chem 275: 21107–21113.

Dougan, D.A., Mogk, A., and Bukau, B. (2002) Protein folding and degradation in bacteria: to degrade or not to degrade?

That is the question. Cell Mol Life Sci 59: 1607–1616.

Ehrnsperger, M., Lilie, H., Gaestel, M., and Buchner, J.

(1999) The dynamics of Hsp25 quaternary structure.

Structure and function of different oligomeric species. J Biol Chem 274: 14867–14874.

Giese, K.C., and Vierling, E. (2002) Changes in oligomeriza- tion are essential for the chaperone activity of a small heat shock protein in vivo and in vitro. J Biol Chem 23: 23.

Glover, J.R., and Lindquist, S. (1998) Hsp104, Hsp70, and Hsp40: a novel chaperone system that rescues previously aggregated proteins. Cell 94: 73–82.

Goloubinoff, P., Mogk, A., Peres Ben Zvi, A., Tomoyasu, T., and Bukau, B. (1999) Sequential mechanism of solubilization and refolding of stable protein aggregates by a bichap- erone network. Proc Natl Acad Sci USA 96: 13732–13737.

Gottesman, S., Wickner, S., and Maurizi, M.R. (1997) Protein quality control: triage by chaperones and proteases.

Genes Dev 11: 815–823.

Hartl, F.U. (1996) Molecular chaperones in cellular protein folding. Nature 381: 571–580.

Haslbeck, M., Walke, S., Stromer, T., Ehrnsperger, M., White, H.E., Chen, S., et al. (1999) Hsp26: a temperature- regulated chaperone. EMBO J 18: 6744–6751.

Horwich, A.L., Brooks Low, K., Fenton, W.A., Hirshfield, I.N., and Furtak, K. (1993) Folding in vivo of bacterial cytoplas- mic proteins: role of GroEL. Cell 74: 909–917.

Horwitz, J. (1992) Alpha-crystallin can function as a molecular chaperone. Proc Natl Acad Sci USA 89: 10449–10453.

Jakob, U., Gaestel, M., Engel, K., and Buchner, J. (1993) Small heat shock proteins are molecular chaperones. J Biol Chem 268: 1517–1520.

Kanemori, M., Nishihara, K., Yanagi, H., and Yura, T. (1997) Synergistic roles of HslVU and other ATP-dependent proteases in controlling in vivo turnover of s³² and abnormal proteins in Escherichia coli. J Bacteriol 179: 7219–

7225.

Kitagawa, M., Matsumura, Y., and Tsuchido, T. (2000) Small heat shock proteins, IbpA and IbpB, are involved in resis- tances to heat and superoxide stresses in Escherichia coli.

FEMS Microbiol Lett 184: 165–171.

Kuczynska-Wisnik, D., Kcdzierska, S., Matuszewska, E., Lund, P., Taylor, A., Lipinska, B., and Laskowska, E. (2002) The Escherichia coli small heat-shock proteins IbpA and IbpB prevent the aggregation of endogenous proteins denatured in vivo during extreme heat shock. Microbiology 148: 1757–1765.

(11)

sHsps, ClpB and DnaK form a functional triade 595 Laemmli, U.K. (1970) Cleavage of structural proteins during

the assembly of the head of bacteriophage T4. Nature (London) 227: 680–685.

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

Laskowska, E., Wawrzynow, A., and Taylor, A. (1996) IbpA and IbpB, the new heat-shock proteins, bind to endogenous Escherichia coli proteins aggregated intracellularly by heat shock. Biochimie 78: 117–122.

Lee, G.J., Roseman, A.M., Saibil, H.R., and Vierling, E.

(1997) A small heat shock protein stably binds heat- denatured model substrates and can maintain a substrate in a folding-competent state. EMBO J 16: 659–671.

Mogk, A., Tomoyasu, T., Goloubinoff, P., Rüdiger, S., Röder, D., Langen, H., and Bukau, B. (1999) Identification of thermolabile E. coli proteins: prevention and reversion of aggregation by DnaK and ClpB. EMBO J 18: 6934–6949.

Mogk, A., Schlieker, C., Friedrich, K.L., Schönfeld, H.-J., Vier- ling, E., and Bukau, B. (2003) Refolding of substrates bound to small Hsps relies on a disaggregation reaction mediated most efficiently by ClpB/DnaK. J Biol Chem 278:

31033–31042.

van Montfort, R.L., Basha, E., Friedrich, K.L., Slingsby, C., and Vierling, E. (2001) Crystal structure and assembly of a eukaryotic small heat shock protein. Nat Struct Biol 8:

1025–1030.

Petko, L., and Lindquist, S. (1986) Hsp26 is not required for growth at high temperatures, nor for thermotolerance, spore development, or germination. Cell 45: 885–894.

Plesofsky-Vig, N., and Brambl, R. (1995) Disruption of the gene for hsp30, an alpha-crystallin-related heat shock protein of Neurospora crassa, causes defects in thermotolerance. Proc Natl Acad Sci USA 92: 5032–5036.

Richmond, C.S., Glasner, J.D., Mau, R., Jin, H., and Blattner, F.R. (1999) Genome-wide expression profiling in Escheri- chia coli K-12. Nucleic Acids Res 27: 3821–3835.

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

Shearstone, J.R., and Baneyx, F. (1999) Biochemical char-

acterization of the small heat shock protein IbpB from Escherichia coli. J Biol Chem 274: 9937–9945.

Skowyra, D., Georgopoulos, C., and Zylicz, M. (1990) The E.

coli dnaK gene product, the Hsp70 homolog, can reactivate heat-inactivated RNA polymerase in an ATP hydrolysis- dependent manner. Cell 62: 939–944.

Stewart, G.R., Wernisch, L., Stabler, R., Mangan, J.A., Hinds, J., Laing, K.G., et al. (2002) Dissection of the heat-shock response in Mycobacterium tuberculosis using mutants and microarrays. Microbiology 148: 3129–3138.

Straus, D., Walter, W., and Gross, C.A. (1990) DnaK, DnaJ, and GrpE heat shock proteins negatively regulate heat shock gene expression by controlling the synthesis and stability of s³². Genes Dev 4: 2202–2209.

Thomas, J.G., and Baneyx, F. (1998) Roles of the Escheri- chia coli small heat shock proteins IbpA and IbpB in ther- mal stress management: comparison with ClpA, ClpB, and HtpG in vivo. J Bacteriol 180: 5165–5172.

Tomoyasu, T., Ogura, T., Tatsuta, T., and Bukau, B. (1998) Levels of DnaK and DnaJ provide tight control of heat shock gene expression and protein repair in E. coli. Mol Microbiol 30: 567–581.

Tomoyasu, T., Mogk, A., Langen, H., Goloubinoff, P., and Bukau, B. (2001) Genetic dissection of the roles of chaperones and proteases in protein folding and degradation in the Escherichia coli cytosol. Mol Microbiol 40: 397–

413.

Veinger, L., Diamant, S., Buchner, J., and Goloubinoff, P.

(1998) The small heat shock protein IbpB from Escherichia coli stabilizes stress-denatured proteins for subsequent refolding by a multichaperone network. J Biol Chem 273:

11032–11037.

Viitanen, P.V., Gatenby, A.A., and Lorimer, G.H. (1992) Puri- fied chaperonin 60 (groEL) interacts with the nonnative states of a multitude of Escherichia coli proteins. Protein Sci 1: 363–369.

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

Zolkiewski, M. (1999) ClpB cooperates with DnaK, DnaJ, and GrpE in suppressing protein aggregation. A novel multichaperone system from Escherichia coli. J Biol Chem 274:

28083–28086.