The Role of the DIF Motif of the DnaJ (Hsp40) Co-chaperone in the Regulation of the DnaK (Hsp70) Chaperone Cycle*

To perform effectively as a molecular chaperone, DnaK (Hsp70) necessitates the assistance of its DnaJ (Hsp40) co-chaperone partner, which efficiently stimulates its intrinsically weak ATPase activity and facilitates its interaction with polypeptide substrates. In this study, we address the function of the conserved glycine- and phenyalanine-rich (G/F-rich) region of the Escherichia coli DnaJ in the DnaK chaperone cycle. We show that the G/F-rich region is critical for DnaJ co-chaperone functions in vivo and that despite a significant degree of sequence conservation among the G/F-rich regions of Hsp40 homologs from bacteria, yeast, or humans, functional complementation in the context of the E. coli DnaJ is limited. Furthermore, we found that the deletion of the whole G/F-rich region is mirrored by mutations in the conserved Asp-Ile/Val-Phe (DIF) motif contained in this region. Further genetic and biochemical analyses revealed that this amino acid triplet plays a critical role in regulation of the DnaK chaperone cycle, possibly by modulating a crucial step subsequent to DnaK-mediated ATP hydrolysis.

Molecular chaperones assist the folding of proteins, mainly by protecting them from intracellular aggregation (1). Of these, the highly conserved Hsp70 chaperone machine plays a central role in a plethora of cellular processes related to protein folding. Hsp70 assists the folding of newly synthesized polypeptides, the assembly and disassembly of protein complexes, as well as protein translocation, disaggregation, and targeting for degradation (1,2). All of these Hsp70 functions depend on the ability of Hsp70 to interact in an ATP-dependent manner with extended hydrophobic stretches of polypeptide substrates (2). To actively perform as a bona fide chaperone, Hsp70 necessitates the assistance of its Hsp40 co-chaperone partners, which efficiently stimulate both its intrinsically weak ATPase activity and facilitate its interaction with polypeptide substrates (2)(3)(4). Some Hsp70 family members additionally require other co-chaperone partners, such as GrpE, Bag-1, or HspBP1, to regulate substrate release and nucleotide exchange (5)(6)(7).
All of the members of the Hsp40 family are characterized by a highly conserved sequence of ϳ70 amino acids, called the J-domain. This domain is absolutely necessary for the stimulation of Hsp70 ATPase activity (3, 8 -11). Active J-domains contain a canonical His-Pro-Asp (HPD) triplet (in some cases Asp is replaced by Glu) that is essential for their activity. The members of the Hsp40 proteins family can be grouped in three classes with respect to their additional conserved domains (9,10). Type I members, such as the well studied DnaJ protein from the bacterium Escherichia coli, possess adjacent to their N-terminal J-domain (i) a glycine-and phenyalanine-rich (G/Frich) 3 sequence of unknown function, (ii) a central cysteine-rich domain containing two zinc-binding sites with apparently distinct functions, one related to a DnaK-independent chaperone function and the second most likely involved in a mechanism associated with the stabilization of DnaK-substrate complex (12)(13)(14)(15), and (iii) a less conserved C-terminal domain involved in substrate binding and homodimerization (16 -20). Although the type II Hsp40 family members lack the central zinc-binding domain, the type III members only share the J-domain signature sequence with their related type I and II members (9,10).
The role of the conserved G/F-rich region of both major type I and II family members has remained enigmatic. This region was originally proposed to be a flexible link between the J-domain and the central zinc-binding domain, but later evidence suggested different functions. For example, in E. coli, the N-terminal J-domain together with the flanking G/F-rich region of the type I Hsp40 member DnaJ were necessary and sufficient to activate DnaK (3,21). A DnaJ construct where the G/F-rich region was replaced by an unrelated sequence (HMGSHM hexapeptide) could efficiently stimulate DnaK ATPase activity and bind the 32 substrate but could not efficiently activate DnaK itself to bind 32 (22). These results suggested that, in contrast to the N-terminal J-domain, the G/F-rich region does not directly trigger the stimulation of DnaK ATPase activity per se but may be rather involved in a mechanism that modulates substrate binding by DnaK. In the yeast Saccharomyces cerevisiae, this region seems to confer a certain degree of specificity. Indeed the G/F-rich region of the S. cerevisiae type II Hsp40 protein Sis1 could not be replaced by the G/F-rich region of the S. cerevisiae type I Ydj1 Hsp40 homolog, when complementation for the lethal phenotype of a ⌬sis1 strain was monitored (23). An additional central aspect of the Sis1 G/F-rich region is that, although not essential for yeast growth, this region was specifically required for the maintenance of the yeast prion [RNQ ϩ ] (24,25). Interestingly, prion maintenance was specifically dependent upon the presence of a short stretch of amino acid residues in the Sis1 G/F-rich region not present in the Ydj1 homolog (24).
The NMR solution structure of the J-domain and the flanking G/F sequence of E. coli DnaJ revealed that in contrast to the compact structure of the J-domain, the G/F-rich region adopts an extended disordered conformation (26). However, a short stretch of amino acids in the G/Frich region presents reduced local flexibility when compared with the rest of the region and undergoes rapid exchange between favored conformations (26). This short segment corresponds to conserved Asp-Ile/ Val-Phe triplets of amino acids in the DnaJ G/F-rich region, referred to as the DIF motif (see Fig. 1A) (22). In this study, we have used in vivo and in vitro approaches to examine the role of the G/F-rich region and its DIF motif in the biology of DnaJ.
Alanine substitution mutants in the DIF motifs of DnaJ and DnaJ12 were constructed by PCR-based, site-directed mutagenesis using the appropriated primers and plasmids pWKG90 or pWKG100. The entire deletion of the G/F-rich region (⌬72-108) of DnaJ was constructed by two-step PCR using pWKG90 and the primers: ⌬G/F-for, 5Ј-gtatggtcatcaacgtgcggcgcgcggtgct-3Ј, and ⌬G/F-rev, 5Ј-cgcacgttgatgaccatactgatcgtatgc-3Ј.
To construct the chimerae consisting of the J-domain of DnaJ and the G/F-rich region of Ydj1 (DnaJ12-Ydj1 G/F), the G/F-rich region of Ydj1 was first PCR-amplified using primers Ydj1G/F-for (5Ј-ggggtaccgatggtctaagtggtgct-3Ј) and Ydj1G/F-rev (5Ј-gcggatccttatcttttgtgcgccaccagc-3Ј) and S. cerevisiae DNA (a gift of Dr. Martine Collart, University of Geneva). The resulting 124-bp fragment was digested with KpnI and BamHI and ligated into pWKG90KPN digested with the same enzymes. To construct the DnaJ12-Mdj1 G/F chimera, the G/F-rich region of Mdj1 was PCR-amplified using primers Mdj1G/F-for (5Ј-ggggtaccgctgccttcggcggcggc-3Ј) and Mdj1G/F-rev (5Ј-gcggatccttatgcgccaccgctaccgcg-3Ј) and S. cerevisiae DNA. The resulting 202-bp fragment was digested with KpnI-BamHI and ligated into pWKG90KPN digested with the same enzymes. The chimera comprising the DnaJ J-domain flanked by the G/F-rich region of Hdj2 (DnaJ12-Hdj2 G/F) was constructed by PCR using primers Hdj2G/F-for (5Ј-cgtggtacccaggcaattaaagagggt-3Ј) and Hdj2G/F-rev (5Ј-cgtagatctttatacaacatttttacctct-3Ј) and HeLa cells cDNA (a gift of Dr. Samuel Deutsch, University of Geneva) as template. The resulting 138-bp fragment was digested with KpnI and BglII and ligated into pWKG90KPN digested with the same enzymes.
DjlA⌬TM with the G/F-rich region of DnaJ attached at its C terminus was constructed as follows. A KpnI site was first introduced before the djlA stop codon by PCR using primers pdTMN-term-for (5Ј-ccggaattcatgcaggataaagcccgtagccgt-3Ј) and pdjlAC-termKpnI-rev (5Ј-cagcggtacctttaaaccctttctgctgctt-3Ј) and pWKG52. The resulting PCR fragment was digested with EcoRI and KpnI and ligated into pWKG100KPN digested with the same enzymes.
To construct plasmid pGPPK, the dnaK gene plus its native promoter sequences was first amplified from plasmid pDM38 (dnaKdnaJ) using the primers 5Ј-gtgcagcacagcatcaggcta-3Ј and 5Ј-ggaagcttggatccttattttttgtctttgacttcttcaaattc-3Ј. The resulting 2596-bp fragment was digested with HindIII and ligated into the EcoRV/HindIII-digested p29SEN. To construct DnaK with a C-terminal His 6 tag, the dnaK gene was amplified with primers 5Ј-gcccatgggtaaaataattggtatcgacctg-3Ј and 5Ј-gcaagcttcaatgatgatgatgatgatgttttttgtctttgacttcttc-3Ј and cloned as an NcoI-HindIII fragment into pSE380 digested with the same enzymes. Control experiments both in vivo and in vitro indicated that the C-terminal addition of His 6 to DnaK did not noticeably alter DnaK function. To construct the pGPE plasmid, the grpE gene was amplified by PCR using appropriated primers and cloned as an EcoRI/BglII fragment into the pSE380⌬NcoI vector digested with the same enzymes.
Plasmids pGPJ-His and pGPJ3DIF-His, encoding respectively for the DnaJ wild type and the DnaJ-DIF proteins with a C-terminal His 6 tag, were obtained as follows. The dnaJ wild type and dnaJ (3DIF) mutant sequences were amplified by PCR, using the primers dnaJNterm (5Ј-gggaattcaccatggctaagcaagattattac-3Ј) and dnaJC-term His6X (5Ј-gcagatctcaatgatgatgatgatgatggcgggtcaggtcgtcaaaaaac-3Ј). The 1167-bp fragments (containing dnaJ linked in frame with six additional Cterminal histidines) were cleaved with EcoRI-BglII and cloned into pSE380⌬NcoI digested with the same enzymes. The dnaJ (H33Q) mutant was constructed using the same strategy, except that the EcoRI-BglII fragment was inserted in pBAD22. Control experiments indicated that the C-terminal addition of His 6 to DnaJ did not noticeably alter DnaJ function.
The pBAD33 DnaJ and DnaJ-3DIF derivatives were obtained by subcloning the EcoRV-SphI-digested fragment of pWKG90 containing dnaJ wild type or dnaJ (3DIF) into the EcoRV-SphI-digested pBAD33 vector. All of the constructs obtained by PCR were sequenced-verified using the appropriate primers.
In Vivo DnaJ Activity Assays-Bacterial motility and bacteriophage plating assays were performed at 30°C as described previously (27). To assess bacterial viability, fresh plasmid transformants were grown overnight at 30°C, serially diluted, and spotted on LB agar plates supplemented when necessary with the appropriate antibiotics (100 g/l ampicillin, 20 g/l chloramphenicol) and various L-arabinose inducer concentrations. The plates were incubated at 30 and 40°C for 20 h and at 14°C for 6 days.
Immunoblot Analysis-For steady state protein expression, whole cell extracts were prepared as described previously (32). DnaJ proteins were resolved by 12% SDS-PAGE, transferred to nitrocellulose membranes (Schleicher & Schuell), probed with anti-DnaJ rabbit polyclonal antibodies, and developed by enhanced chemiluminescence using goat anti-rabbit horseradish peroxidase-conjugated IgG as secondary antibodies (Amersham Biosciences).
Protein Purifications-To avoid a toxic effect of protein overexpression in wild type E. coli, DnaJ and DnaJ-3DIF were purified from the ⌬ 4 strain. Fresh overnight cultures were diluted 1:100 in 500 ml of LB broth supplemented with 100 g/l ampicillin and grown with vigorous shaking at 30°C. At an A 600 of 0.5, 1 mM IPTG was added, and the cultures were incubated for an additional 3 h. For DnaJ-H33Q-His 6 overexpression, IPTG was replaced by 0.5% of L-arabinose. The cells were harvested at 5000 rpm for 30 min in a Sorvall GSA rotor and resuspended in 10 ml of J1 buffer (50 mM Tris, pH 8.0, 10% (w/v) sucrose, 10 mM DTT, 1 mM EDTA, 1 mg/ml lysozyme, and 0.6% (w/v) Brij 58). The cells were incubated on ice for 45 min and then sonicated twice for 10 s to reduce viscosity. After centrifugation at 28,000 rpm for 1 h in a Beckman 35 Ti rotor, the pellets were resuspended in 10 ml of J2 buffer (1 M NaCl, 50 mM Tris, pH 7.3, 2 mM ␤-mercaptoethanol, 10% sucrose, 0.5% (v/v) Triton X-100, 500 l of protease inhibitor mix; Sigma) and then gently tumbled at 4°C for 1 h. Insoluble material was removed by centrifugation at 28,000 rpm for 90 min in a Beckman 35 Ti rotor. Supernatants (5 ml) were applied to 2 ml of nickel-nitrilotriacetic acid columns preequilibrated with 5 volumes of J2 buffer (without protease inhibitors). The following steps were performed as described in the procedure from Qiagen for purification of His 6 -tagged proteins from E. coli using nickelnitrilotriacetic acid superflow under native conditions, using buffer J2 supplemented with 20 mM imidazole as washing buffer and with 250 mM imidazole as elution buffer. The proteins were stored at Ϫ80°C in buffer containing 25 mM HEPES, pH 7.6, 0.4 M KCl, 1 mM DTT, 10% DnaK with a C-terminal His 6 tag was purified from strain ⌬ 4 . The cells were grown at 30°C to mid-exponential phase, protein expression was induced with 1 mM IPTG for 3 h, and purification was performed as described in the Qiagen protocol. The proteins were stored at Ϫ80°C in TEG buffer (20 mM Tris, pH 7.5, 100 mM NaCl, 0.1 mM EDTA, 1 mM DTT, 20% (v/v) glycerol).
Purified GrpE protein was a kind gift of Dr. Krzysztof Liberek (University of Gdansk, Gdansk, Poland). Protein concentrations were routinely measured by a Bradford assay using bovine serum albumin as standard and estimated by SDS-PAGE analysis.
Luciferase Aggregation and Refolding Assays-Luciferase aggregation was performed as follows. 25 M luciferase (Promega) was denatured in 6 M guanidinium hydrochloride, 30 mM Tris-HCl, pH 7.6, 5 mM DTT for 2 h at 22°C. The reactions were initiated by a 100-fold dilution of denatured luciferase into a premix containing the indicated concentrations of DnaJ proteins in 30 mM HEPES, pH 7.5, 40 mM KCl, 50 mM NaCl, 7 mM magnesium acetate, 1 mM DTT. Aggregation kinetics were followed at 22°C by measuring light scattering at 320 nm. Luciferase refolding was essentially assessed as described (35), with minor modifications. Specifically, luciferase 25 M (Sigma) was denatured as described above and diluted 100-fold into a premix containing 10 mM MOPS, pH 7.2, 50 mM KCl, 5 mM MgCl 2 , 0.015% (w/v) bovine serum albumin, 0.1 mg/ml creatine kinase, 20 mM creatine phosphate, 5 mM ATP, 2 M DnaK, and 0.5 M DnaJ. After 10 min, the indicated concentration of GrpE was added. The luciferase activity was monitored at 22°C by using the luciferase assay system from Promega (E1500) and a Turner luminometer (TD-20/20).
Co-immunoprecipitation-The co-immunoprecipitation experiments were‫گ‬carried out as described previously (36), except that Pansorbin was replaced by IgGsorb (The Enzyme Center, Malden).
Steady State and Single-turnover ATPase Assays-The steady state ATPase assay was performed essentially as described, with minor modifications (4). The standard reaction conditions were 30 mM HEPES, pH 7.6, 40 mM KCl, 50 mM NaCl, 7 mM magnesium acetate, 2 mM DTT, 0.29 mg/ml bovine serum albumin, 100 M ATP, 1 Ci/ml [␥-32 P]ATP (3000 Ci/mmol), 1 M DnaK, 1 M GrpE, and DnaJ or DnaJ mutants in the range 0 -0.8 M. When necessary, 250 nM of luciferase (Promega) denatured with 6 M guanidinium hydrochloride was added in the reaction mix. Aliquots were removed from the reaction at the specified time points and spotted onto polyethyleneimine thin layer chromatography plates (Merck). The plates were resolved with a 1:1 solvent mixture 1 M formic acid, 1 M LiCl. The amount of liberated ␥-phosphate was quantified using phosphorimaging.
Single-turnover ATPase assay was performed as described by Mayer et al. (37) with minor modifications. Briefly, 15 M DnaK was incubated with 800 M ATP and 12 Ci [␣-32 P]ATP (3000 Ci/mmol) in buffer A (25 mM HEPES, 50 mM KCl, 10 mM MgCl 2 ) for 2 min on ice. DnaK-[␣-32 P]ATP complexes were then rapidly separated from free nucleotides with a Sephadex G50 Nick-column (Amersham Biosciences). The aliquots were flash frozen in liquid nitrogen and stored at Ϫ80°C. The reactions were initiated by adding DnaK-[␣-32 P]ATP complex to a premix containing 0.4 M DnaJ in buffer A maintained at 25°C. Aliquots (2 l) were spotted onto polyethyleneimine thin layer chromatography plates (Merck) and separated with 150 mM formic acid, 0.150 mM LiCl, and the ADP was quantified using phosphorimaging.

Functional Conservation of the G/F-rich Region among Hsp40 Family
Members-The G/F-rich region of Hsp40 family members is characterized by a relatively high frequency of glycine and phenylalanine residues, with one universally conserved sequence motif DIF (Fig. 1A). To gain insight into the function of this region, we first asked whether functional conservation could be observed among type I Hsp40 G/F-rich regions from several organisms. To this end, we constructed a series of chimerae where the G/F-rich region of E. coli DnaJ was replaced by the corresponding region from either yeast (cytosolic Ydj1 and mitochondrial Mdj1) or human (cytosolic Hdj2) type I Hsp40s. To minimize complexity, these chimerae were constructed in the stringent context of DnaJ12, which comprises the N-terminal amino acids 1-108 containing the J-domain and the G/F-rich region and is the smallest known part of E. coli DnaJ sufficient and necessary to partially support its DnaK cochaperone function (21). These constructs were then tested in vivo for their ability to complement the various phenotypes of the E. coli strain ⌬ 3 , which lacks all three Hsp40 co-chaperones of DnaK, namely DnaJ, CbpA, and DjlA (29). We observed that none of the yeast and human G/F-rich regions tested are able to fully functionally replace the E. coli G/F-rich region (Fig. 1B). Specifically, although all the chimerae efficiently support bacterial growth at high temperature and propagation in the ⌬ 3 strain, no complementation for bacterial growth at low temperature was observed for any of the chimerae (Fig. 1B; compare DnaJ12 to DnaJ12-Ydj1 G/F, DnaJ12-Mdj1 G/F, and DnaJ12-Hdj2 G/F). Steady state accumulation levels of the chimerae at permissive temperature (30°C) were similar to those of DnaJ12 (data not shown). Among the chimerae tested, the DnaJ12-Hdj2 G/F construct exhibited the weakest complementation activity and, in contrast to the yeast chimerae, even exerted a negative effect on E. coli growth and dnaJ ϩ transducing phage propagation at a permissive temperature (30°C).
We next asked whether addition of the G/F-rich region of DnaJ could convert a non-type I Hsp40 homolog into a functional DnaJ-like cochaperone in vivo. To answer this, we grafted the G/F-rich region of DnaJ to the C-terminal part of the E. coli type III Hsp40 DjlA protein, a bona fide DnaK co-chaperone that does not efficiently complement DnaJ function in vivo for either bacterial growth at high and low temperatures or propagation (29,35). Interestingly, the chimera (DjlA⌬TM G/F) was able to rescue all of the phenotypes of the E. coli ⌬ 3 mutant strain tested, including bacterial growth at low temperature (Fig. 1B). Taken together, these results indicate that the G/F-rich region of DnaJ retains a substantial degree of specificity and is critical for DnaJ function in E. coli.
The DIF Motif of the G/F-rich Region Is Critical for DnaJ Function in Vivo-The G/F-rich region of the Hsp40 proteins includes the conserved Asp-Ile/Val-Phe triplet of amino acids, referred to as DIF (Fig. 1A). The E. coli DnaJ G/F-rich region possesses three such repeats. To investigate whether this DIF motif is important for DnaJ function, we first replaced all of the nine amino acid residues comprising the DIF motif by alanines in the stringent context of DnaJ12 (DnaJ (1-108)) and tested the resulting mutant for complementation in our in vivo assays. As judged by the inability of the DnaJ12-3DIF mutant to support propagation of bacteriophage and bacterial growth at high and low temperatures, the DIF motif appears essential for the in vivo co-chaperone functions of DnaJ12 (Figs. 1B and 2A). In addition, we observed that the DIF amino acid substitutions exert a toxic effect on E. coli growth at permissive temperatures (see below) and even partially block the propagation of dnaJ ϩ transducing phage.
To further assess the critical role of the DIF motif, we asked whether its substitution in the context of the full-length DnaJ protein exerts an effect similar to that observed in the DnaJ12 context. To do so, we generated both a full-length DnaJ with the three DIF repeats substituted by alanines (3DIF) and a DnaJ lacking the whole G/F-rich region (⌬(72-108)) and tested for their activity in vivo (Fig. 2). As found with DnaJ12, both the 3DIF and the ⌬G/F mutations completely abolish complementation of bacterial growth and motility (Fig. 2, B and C; data not shown). The steady state expression levels of both DnaJ wild type and DnaJ-3DIF are comparable at permissive temperature (30°C) in the ⌬ 3 mutant strain (Fig. 2B). The steady state expression levels of these proteins at the nonpermissive temperatures for ⌬ 3 (16°and 40°C) are also indistinguishable when the mutant protein was expressed in the dnaJ single mutant able to grow at such temperatures (data not shown). Furthermore, the same inhibitory effect was observed when the DIF motif was replaced by glycines instead of alanines (data not shown).
In contrast to bacterial growth and motility, propagation is partially supported by both the 3DIF and ⌬G/F mutants ( Fig. 2D; data not shown). This functional difference may be due to the specific role of the DnaK/DnaJ/GrpE chaperone machine in the destabilization of an oth-erwise stable preprimosomal protein-DNA complex at the origin of replication (38). This specific task of DnaJ contrasts with its generic protein folding activity with nascent polypeptide chains and may explain the differential behavior of the DnaJ-DIF mutants in E. coli and physiology.
To test the importance of individual DIF repeats and each residue of the DIF motif, we next generated a series of single, double, and triple alanine substitutions and tested them in our in vivo assays (Fig. 2). When the corresponding residues of each of the three repeats are simultaneously substituted by alanine, the resulting mutants behave similarly to the 3DIF mutant (see constructs 3D, 3I/V, and 3F). Similar observations were made for each DIF motif, although the DIF3 substitution seems to partially complement at high temperature and for cellular motility (Fig.  2, B and C; see constructs DIF1, DIF2, and DIF3). Furthermore, a single alanine substitution of any residue of the first DIF motif exerts a major inhibitory effect on DnaJ function at low temperature, but not at high temperature nor for cellular motility (Fig. 2, B and C; see constructs D1, I1 and F1). As with the 3DIF mutant, these constructs exert only a minor effect on propagation (data not shown). As shown in the bottom panel of Fig. 2B, all of the mutant constructs are well expressed at the permissive temperature of 30°C, and their steady state level is comparable with that of wild type DnaJ. Taken together, our results indicate that the DIF motif plays a subtle, yet crucial role in the ability of DnaJ protein to function as a bona fide DnaK co-chaperone.
The DIF Mutants Exhibit a DnaK-dependent Toxic Effect-Besides their inability to functionally substitute for the wild type DnaJ protein, we observed that the DIF substitutions, in both the DnaJ12 and fulllength DnaJ contexts, exert a toxic effect on E. coli growth. As seen in Figs. 1B and 3A, expression of the mutant proteins completely blocked the ability to form colonies at the otherwise permissive temperature of 30°C. In contrast, the overexpression of either wild type DnaJ or DnaJ-H33Q (substitution His 33 to Gln in the J-domain), known to abolish all DnaJ co-chaperone functions, exerts no such toxic effect. A level of DnaJ-3DIF ϳ3-fold higher than endogenous DnaJ at 30°C was sufficient to severely impair bacterial growth (data not shown). A similar effect on E. coli growth was also observed in the dnaJ single mutant or the dnaJ cbpA or the dnaJ djlA double mutants. However, the simultaneous co-expression of the wild type DnaJ significantly reduces DnaJ-3DIF toxicity (data not shown).
To investigate whether toxicity of the mutant is due to an impaired interaction with its chaperone partner DnaK, we tested the effect of DnaJ-3DIF in a strain lacking DnaK plus its three Hsp40 co-chaperone partners, namely DnaJ, DjlA, and CbpA (strain called ⌬ 4 ). Consistent with a possible impaired interaction with DnaK, the toxicity of the DnaJ-3DIF mutant is totally abolished in the absence of DnaK (Fig. 3A). We next tested whether the absence of toxicity is directly linked to the absence of dnaK or if the effect is indirectly due to the accumulation of extragenic suppressors, known to readily accumulate following the deletion of the dnaK gene (39) in the ⌬ 4 strain. To do so, we transformed the ⌬ 4 strain expressing the toxic DnaJ-3DIF construct with the dnaK gene under the control of its native promoter on a compatible low copy number plasmid and tested for toxicity. Clearly, the toxicity of the DnaJ-3DIF mutants reappears in the presence of plasmid-encoded DnaK in the ⌬ 4 strain (Fig. 3A). These results demonstrate that the toxic effect of the 3DIF mutation in DnaJ is strictly DnaK-dependent and therefore directly related to its co-chaperone function.
It is known that the J-domain of DnaJ efficiently stimulates the weak ATPase activity of DnaK, a necessary step for DnaK-substrate stable complex formation (40). To assess whether DnaJ-mediated stimulation of DnaK ATPase activity occurs prior to the DIF motif signal, we generated a double mutant that combines the 3DIF substitution together with the well characterized, disabling H33Q substitution in the J-domain and tested the construct for toxicity in a ⌬ 3 strain. As seen in Fig.  3 (see construct H33Q-3DIF), the 3DIF-mediated toxicity is totally abolished when the DIF substitution is combined with the J-domain H33Q disabling mutation. These results suggest that under wild type conditions, the DIF motif plays a role at a step subsequent to ATP hydrolysis by DnaK.

FIGURE 2. The DIF motif of the DnaJ is critical for full-length DnaJ in vivo.
A, summary of the alanine substitutions in the G/F-rich region of the full-length DnaJ protein analyzed in this study. Only the amino acid sequence of the DnaJ G/F-rich region is presented. B, complementation for bacterial growth at high and low temperatures in strain ⌬ 3 (dnaJ cbpA djlA triple mutant) and Western blot analysis showing the steady state levels of the various DnaJ constructs in strain ⌬ 3 at 30°C. A typical set of results using 0.02% of L-arabinose inducer is presented. C, complementation for bacterial motility in strain ⌬ 2 (dnaJ cbpA double mutant) using the L-arabinose-inducible pBAD22-based DnaJ constructs depicted in A. A typical set of results using 0.02% of L-arabinose inducer is presented. D, complementation for bacteriophage plaque forming ability in strain ⌬ 2 using the L-arabinose-inducible pBAD22-based DnaJ and DnaJ-3DIF constructs. A typical set of results using 0.05% of L-arabinose inducer is presented.

Increased Level of GrpE Inhibits the Poisonous Effect of DIF on the
DnaK Chaperone Cycle-It was previously shown that the deletion of the grpE gene encoding for GrpE, the nucleotide exchange factor of DnaK, can only be generated in the absence of wild type DnaK in the cells (41). This DnaK-dependent lethality of GrpE depletion closely resembles the observed DnaK-dependent toxicity of the DnaJ-DIF mutants on E. coli growth. Thus, both situations may reflect the accumulation of poisonous DnaK-substrate complexes.
To assess the validity of this hypothesis we asked whether GrpE overexpression could rescue the toxicity of the DnaJ-DIF mutants. Indeed, as shown in Fig. 3B, the toxicity of DnaJ-3DIF is significantly relieved following GrpE co-overexpression. An ϳ5-10-fold increased level of GrpE is sufficient to suppress the observed toxicity of the DnaJ-3DIF mutant (data not shown). Although GrpE overexpression was sufficient to rescue toxicity of the DnaJ-3DIF mutant at the otherwise permissive temperature, it does not restore colony formation at either high or low temperature in the presence of the DnaJ-3DIF mutant (data not shown). These results suggest that once stimulation of DnaK-substrate complex formation by the DnaJ-3DIF mutant has occurred, the putative toxic DnaK-substrate intermediates formed can be destabilized by the GrpEmediated acceleration of substrate release (6).
Substitution in the DIF Motif Does Not Affect DnaJ Chaperone Functions in Vitro-To gain further insight into the involvement of the DIF motif, we carried out several in vitro studies. We first tested the ability of the purified DnaJ-3DIF protein to bind and protect denatured substrates from aggregation in a DnaK-independent manner. Using the model substrates firefly luciferase and bovine rhodanese, we observed that the DnaJ-3DIF mutant protein protects both proteins from aggregation, in a manner indistinguishable from wild type DnaJ (Fig. 4; data not shown). These results suggest that the DIF motif of DnaJ is not involved in substrate binding and are in agreement with previous published work (22,42). Indeed, replacement of the whole G/F-rich region of DnaJ by the random HMGSHM amino acid segment had no effect on its affinity for 32 (22). Likewise, in yeast, the deletion of the whole G/F-rich region of Sis1 had no effect on Sis1 binding to luciferase or to Rnq1 (42). In addition, other experimental evidence suggests that the polypeptide-binding sites of the type I and type II Hsp40 homologs Ydj1 and Sis1 are localized at the C-terminal domain (16 -18).
The DIF Motif Does Not Directly Participate in the Stimulation of DnaK ATPase Activity-The in vivo data suggest that the DIF motif of DnaJ regulate the DnaK chaperone cycle at a post-ATP hydrolysis step, such as the proper locking in of substrate or the timely release of DnaJ from DnaK. To investigate this possibility, we tested the effect of the DIF substitutions on the stimulation of DnaK-mediated ATP hydrolysis, both under single-turnover and steady state conditions. First, we tested the ability of purified DnaJ-3DIF to stimulate DnaKmediated ATP hydrolysis under single-turnover conditions, where the kinetics of only a single round of DnaK-mediated ATP hydrolysis is measured (37). Under such conditions, the DIF substitutions in DnaJ do not exert a noticeable effect on the stimulation of DnaK ATPase activity, compared with wild type DnaJ (Fig. 5A). This result confirms that DIF involvement is at a step subsequent to that of the stimulation of DnaK ATP hydrolysis.
Considering that the DIF motif does not influence the stimulation of DnaK ATP hydrolysis but is absolutely needed for DnaJ co-chaperone functions and that the DnaJ-3DIF mutant exhibits a DnaK-dependent toxicity in vivo, one may expect to see an effect on DnaK-mediated ATP hydrolysis following multiple rounds of DnaK cycling modulated by its co-chaperones DnaJ and GrpE. To address this possibility, we performed DnaK ATP hydrolysis experiments under steady state conditions, in the presence of DnaJ and GrpE. We found that indeed, steady state ATP hydrolysis rates by DnaK are considerably lower (20 -30%) in the presence of DnaJ-3DIF than in the presence of wild type DnaJ (Fig.  5B). As a negative control, we show that as expected, the DnaJ-H33Q protein does not stimulate DnaK under these conditions (Fig. 5B). We conclude that the DIF motif of the DnaJ co-chaperone is not involved directly in the stimulation of DnaK ATPase activity but instead is necessary for a subsequent, yet unknown, critical step in the DnaK chaperone cycle.  The DIF Motif Regulates the DnaK Chaperone Cycle-To determine the involvement of the DIF motif in the regulation of the DnaK chaperone cycle, we tested whether DnaJ-3DIF could function as a DnaK co-chaperone in vitro. To do so, we tested the ability of the purified DnaJ-3DIF protein to assist DnaK in the refolding of chemically denatured firefly luciferase (Fig. 6A). We observed that in contrast to wild type DnaJ, DnaJ-3DIF does not activate DnaK for the refolding of denatured luciferase. Increased concentrations of DnaJ-3DIF (up to 5 M) does not allow luciferase reactivation (data not shown). These results indicate that the DIF motif is critical for the activation of the DnaK chaperone function in vitro.
To our knowledge, the DIF mutations in DnaJ are the only DnaJ mutant proteins known to exhibit such a DnaK-dependent toxic effect in vivo. To test whether the DIF mutants can still efficiently transfer substrates to DnaK, we performed co-immunoprecipitation of DnaK and its luciferase substrate in the presence of either DnaJ or DnaJ-3DIF. As control, we performed the same experiment with DnaJ-H33Q. This well characterized mutation in the J-domain does not affect DnaJ chaperone functions (like DIF mutations) but totally abolishes all known DnaK-dependent co-chaperone functions, including the stimulation of DnaK ATPase activity and the efficient substrate transfer to DnaK (10). Furthermore, in contrast to DIF, this mutation does not exhibit any DnaK-dependent toxic effect (27) (Fig.  3A). As expected, the DnaJ-H33Q protein does not activate the formation of the DnaK-luciferase complex, as judged by the co-immunoprecipitation data (Fig. 6B). In contrast to DnaJ-H33Q, the DnaJ-3DIF mutant efficiently transfers substrate to DnaK (Fig. 6B). These results are in agreement with the in vivo toxicity and the reduced DnaK ATPase cycle observed under steady state conditions. For both DnaJ and DnaJ-3DIF, the addition of GrpE in the co-immunoprecipitation reaction identically reduces the yield of DnaK-bound luciferase (data not shown).
Because GrpE overexpression alleviates the DnaK-dependent toxicity of the DnaJ-3DIF mutant in vivo, we asked whether increased GrpE concentrations could partially restore luciferase refolding in the presence of DnaJ-3DIF. Although a wide range of GrpE concentrations was tested, no detectable improvement in luciferase refolding was observed (Fig. 6C). In a similar manner, increased GrpE concentrations do not  improve the steady state ATPase activity of DnaK in the presence of DnaJ-3DIF (data not shown). These results are in agreement with the in vivo observation that GrpE overexpression does not support bacterial growth at a nonpermissive temperature in the presence of DnaJ-DIF. In turn, this suggests that accelerating substrate release from DnaK or inhibiting its interaction with DnaK prevents the formation of toxic forms of DnaK-substrate complexes in the presence of DnaJ-3DIF.

DISCUSSION
In this study, we have investigated the role of the DnaJ G/F-rich region and the DIF motif in the regulation of the DnaK chaperone cycle in E. coli. We first show that the G/F-rich region retains an exquisite specificity that is required for full E. coli DnaJ activity in vivo. Specifically, our chimerae study revealed that although functional complementation by the G/F-rich region from either human or yeast type I Hsp40 homologs is observed during bacterial growth at high temperatures and for bacteriophage propagation, none of these chimerae can functionally replace the DnaJ G/F-rich region for bacterial growth at low temperatures. Such specificity has been observed among G/F-rich regions of type I and type II Hsp40 homologs in yeast (23, 24,42). Indeed, the G/F-rich region of the Sis1 type II Hsp40, which is critical for [RNQ ϩ ] prion maintenance and cell survival in a ⌬sis1 mutant, could not be replaced by the G/F-rich region of the Ydj1 type I Hsp40 (24). The specificity of action of the G/F-rich region is further emphasized by our results obtained with the E. coli type III Hsp40 homolog DjlA. Indeed, DjlA, which possesses no G/F-rich region and by itself does not complement for DnaJ functions in vivo (29,35), could efficiently replace DnaJ when the G/F-rich region of DnaJ was grafted to its C-terminal region.
In addition, we showed that deletion of the whole G/F-rich region of DnaJ (DnaJ⌬(72-108)) exerts a noncomplementing phenotype in vivo and that this phenotype is also seen with single amino acid alterations in a stretch of amino acids that we previously named the DIF motif (22). An alteration of any of the repeats of the DIF motif mimics the deletion of the whole G/F-rich region, both in the context of the full-length DnaJ and of the smallest active DnaJ construct (DnaJ(1-108)), which contains only the J-domain and the flanking G/F-rich region. These results suggest that these motifs may, at least in part, be responsible for the specificity exhibited by the G/F-rich region.
In this respect, it is interesting to note that the NMR solution structure of the J-domain and the G/F-rich region of DnaJ shows that although the whole G/F-rich region is overall highly unstructured and does not form a globular core, amino acids 90 -103 exhibit reduced local flexibility and thus tend to adopt a preferential conformation (26). Remarkably, this short segment contains all the 11 amino acids of the critical DIF motif.
The in vitro analysis of the DnaJ-3DIF mutant shows that the DIF motif neither participates in substrate binding (Refs. 14, 16 -18, 22, and 42 and this work) nor directly stimulates DnaK ATPase activity. However, substitutions in the DIF motif significantly slow down the steady state ATPase cycle of the DnaK/DnaJ/GrpE chaperone machine and inhibit its ability to refold luciferase substrate in vitro. In addition, DIF mutations exert a remarkable DnaK-dependent poisonous effect on E. coli growth at otherwise permissive temperatures but do not block the activation of the formation of DnaK-substrate complexes in vitro. This suggests that the poisonous effect may be due to the formation of kinetically trapped DnaK-substrate or DnaJ-DnaK-substrate complexes. Some of the trapped substrates may be essential E. coli proteins, and the failure to properly fold or release these proteins in a timely fashion may inhibit bacterial growth. This situation may be analogous to the previously observed lethality of a grpE deletion seen only in the presence of functional DnaK (41). Such a possible scenario is further supported by the fact that either overexpression of GrpE or a mutation in the J-domain of DnaJ, known to block the stimulation of DnaK ATPase activity, alleviates DIF toxicity.
While our study was in progress, Aron et al. (42) observed a relatively similar situation in yeast. The deletion of the G/F-rich region of the type II Hsp40 homolog Sis1 also exerted a poisonous effect on wild type yeast growth. Additional experiments indicated that the absence of a 13-amino acid segment of the G/F-rich region was the cause of the toxicity. The same segment was also shown to be necessary for Sis1-mediated [RNQ ϩ ] prion maintenance (24,42). Interestingly, the toxicity of Sis1 ⌬G/F was suppressed by point mutations in the Sis1 C-terminal domain, which weakened the overall interaction of Sis1 with its Hsp70 partner Ssa1 (42). However, the specific motif of the G/F-rich region of the yeast type II Hsp40 Sis1 involved is also necessary for [RNQ ϩ ] prion maintenance and is not present in the type I Hsp40 homologs, Ydj1 in yeast and DnaJ in E. coli.
Based upon these observations, a model for the role of the DIF motif of DnaJ in the function of the DnaK/DnaJ/GrpE chaperone machine is proposed in Fig. 7. Both DnaJ wild type (Fig. 7A) and DnaJ-DIF mutant (Fig. 7B) bind similarly to substrate and equally stimulate DnaK ATPase activity. Upon ATP hydrolysis by DnaK and substrate transfer from DnaJ to DnaK, DnaJ wild type is rapidly released from DnaK, leaving the DnaK-substrate complex accessible to interact with the GrpE dimer, thus accelerating nucleotide exchange and the subsequent release of the substrate from DnaK. In the case of the DnaJ-DIF mutant, we propose that more stable DnaK-substrate or DnaJ-DnaK-substrate complexes are formed, perhaps trapped in an erroneous conformation. The formation of such a kinetically trapped complex could either delay or block the folding or disaggregation of some essential substrate(s). In the absence of DnaK, this putative essential substrate(s) may be eligible for folding with the help of the other E. coli chaperone machines. Clearly, more experiments are warranted to validate such a model.