Complementation of an Escherichia coli DnaK Defect by Hsc70-DnaK Chimeric Proteins

Abstract

Escherichia coli DnaK and rat Hsc70 are members of the highly conserved 70-kDa heat shock protein (Hsp70) family that show strong sequence and structure similarities and comparable functional properties in terms of interactions with peptides and unfolded proteins and cooperation with cochaperones. We show here that, while the DnaK protein is, as expected, able to complement an E. coli dnaK mutant strain for growth at high temperatures and λ phage propagation, Hsc70 protein is not. However, an Hsc70 in which the peptide-binding domain has been replaced by that of DnaK is able to complement this strain for both phenotypes, suggesting that the peptide-binding domain of DnaK is essential to fulfill the specific functions of this protein necessary for growth at high temperatures and for λ phage replication. The implications of these findings on the functional specificities of the Hsp70s and the role of protein-protein interactions in the DnaK chaperone system are discussed.

The heat shock proteins of 70 kDa (Hsp70s) are among the most conserved proteins in nature and are found in most prokaryotic cells and in most compartments of all eukaryotic cells (1, 22). They are known to protect cells against damage by high temperatures and to assist protein folding and assembly by ATP-dependent cycles of substrate binding and release. They cooperate in these functions with various cofactors, such as the ubiquitous members of the DnaJ chaperone and GrpE families (6, 11, 14).

Escherichia coli DnaK and rat Hsc70 are two prominent members of this family that have been extensively studied. While bacterial DnaK is a bona fide heat shock protein, for it is strongly inducible by heat shock (1) and is able to efficiently protect cells at high temperatures, eukaryotic Hsc70 is not and is in fact a constitutive protein expressed at normal temperatures (10, 16) that plays little or no role in heat stress protection. DnaK is involved in negative regulation of the heat shock response, in host and bacteriophage replication, in the prevention of protein denaturation and aggregation during stress, and in the refolding of heat-denatured proteins (18), while Hsc70 interacts with a wide range of specific and well-folded cellular proteins and possesses specialized functions, such as clathrin uncoating from coated vesicles (8). Moreover, these proteins differ in their abilities to interact with a defined set of cochaperones. For instance, while DnaK and Hsc70 chaperones are both slow ATPases that have similar hydrophobic peptide-binding specificities, they cooperate with different cochaperones to accomplish their functional cycles of substrate binding and release through nucleotide hydrolysis and exchange. E. coli DnaK uses the ATPase-activating factor DnaJ and the nucleotide exchange factor GrpE (13, 26, 31), whereas Hsc70 does not bind to GrpE, although it still interacts with Hsp40, a DnaJ homolog, and uses Hip and Bag-1, a set of cochaperones with no counterpart in E. coli (15, 29, 32). In fact, it was proposed that the interaction of GrpE with DnaK, but not Hsc70, is at the basis of the diversification and functional specificity of Hsp70 chaperone systems (4).

Nevertheless, these two relatives have very similar three-dimensional structures, as indicated by the X-ray and nuclear magnetic resonance structures available (12, 13, 21, 23, 34), and are both made of three domains: an N-terminal ATPase domain, a peptide-binding domain composed essentially of a β sandwich with a shallow peptide-binding pocket followed by an α-helical segment supposed to form a lid controlling the accessibility to the peptide-binding pocket, and a C-terminal α-helical domain (7, 9, 12, 23) (Fig. 1).

FIG. 1.

Three-dimensional structure of DnaK/Hsc70 showing the three domains: the N-terminal ATPase domain, N (1 to 384), the substrate-binding domain, P (389 to 557), which contains the β sandwich (β) and helices α1 and α2, and the C-terminal helical domain, C (557 to 607), which is composed of α3, α4, and α5 helices. Residues 386 and 557 (circles) constitute the junction points for the construction of the chimeras. The primary structures of DnaK and Hsc70 in the N (ATPase) and P domains (particularly in the β sandwich) are very similar, but they differ slightly in helices α1, α2, and α3 to α5 of the C domain (see text for explanations).

Thus, in spite of a high sequence and structure similarity, these proteins appear to have different functional properties. To gain insight into the structural origin of these differences, a series of chimeric proteins, made by swapping respective domains having similar structures but different functions, have been generated and analyzed in vivo for the complementation of two E. coli phenotypes, growth at high temperatures and propagation of λ phage. The results of this in vivo study are discussed with respect to the available in vitro structural and functional information for these two proteins.

MATERIALS AND METHODS

Plasmids, strains, and media.

The various chimeric proteins used in this work have been constructed with the pDnaK and pUHE21-2FdΔ12 plasmids, a kind gift from Bernd Bukau (University of Heidelberg, Heidelberg, Germany). All strains and plasmids are listed in Table 1.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Genotype and/or description	Source or reference(s)
MC4100	F− araD139 Δ(argF-lac)U169 rpsL150 relA1 deoC1 ptsF25 rbsR fbB301	20
BB1553	MC4100 ΔdnaK52::cm sidB1	20
BB2393	C600 dnaK103(Am) thr::Tn10	19, 30
pUHE21-2fdΔ12	pBR322 derivative containing a multiple cloning site downstream of the lac promoter	5
pDMI-1	pUC18 derivative encoding the LacI repressor under the control of the lac promoter	20
pDnaK	pUHE21-2fdΔ12 derivative encoding DnaK (NPC) under the control of the lac promoter	20
pHsc70	pUHE21-2fdΔ12 derivative encoding Hsc70 (N′P′C′) under the control of the lac promoter	This study
pJP1	pUHE21-2fdΔ12 derivative encoding the chimera N′PC	This study
pJP2	pUHE21-2fdΔ12 derivative encoding the chimera NP′C′	This study
pJP3	pUHE21-2fdΔ12 derivative encoding the chimera NP′C	This study
pJP4	pUHE21-2fdΔ12 derivative encoding the chimera N′PC′	This study
pJP5	pUHE21-2fdΔ12 derivative encoding the chimera NPC′	This study
pJP6	pUHE21-2fdΔ12 derivative encoding the chimera N′P′C	This study

Ultracompetent cells from E. coli strain XL2-Blue were used for the various constructions and were from Stratagene. The E. coli strain used for complementation studies, BB2393 [C600 dnaK103(Am) thr::Tn10], is from Bernd Bukau.

Luria-Bertani (LB) medium was used for bacterial growth. Tryptone, yeast extract, and agar were obtained from Difco Laboratories, while ampicillin and kanamycin were from Sigma.

Construction of Hsc70/DnaK chimeric proteins.

To obtain a plasmid coding for rat Hsc70, the hsc70 coding sequence of pFB7 (2) was inserted between the BamHI and HindIII restriction sites of pUHE21-2fdΔ12. This was performed after modifying the internal HindIII site of the hsc70 coding sequence, with the QuikChange kit (Stratagene), and introducing the 5′ and 3′ restriction sites by PCR. The resulting plasmid was used to transform XL2-Blue ultracompetent cells. Single colonies were picked for overnight culture at 30°C, and the plasmids were purified by the MidiPreps kit (Bio 101).

For the construction of chimeras, restriction sites were introduced in the coding sequence of dnaK and hsc70 by site-directed mutagenesis using the QuikChange kit. Since the restriction sites had to be unique sites and identical in both plasmids in order to perform the domain swap, only the possibilities that resulted in minimal changes in the amino acid sequence have been retained. Thus, and based on structural alignment of the two proteins (34), AflII sites were introduced into the coding sequences for the interdomain region separating the ATPase domain (N) and the peptide-binding domain (P) (positions 386 to 387 in DnaK and 389 to 390 in Hsc70), and SpeI sites were introduced into the coding sequences for the loop separating the peptide-binding domain (P) and the C-terminal domain (C), between helix α2 and helix α3 (positions 557 to 558 in DnaK and 563 to 564 in Hsc70) (Fig. 1). The creation of AflII and SpeI sites in DnaK coding sequence led to the replacement of valine 386 by a leucine and the insertion of a valine in position 558, whereas, in Hsc70, valine 389 was replaced by a leucine, glutamine 390 was replaced by a lysine, isoleucine 563 was replaced by a leucine, and asparagine 564 was replaced by a valine. Complementation properties with these plasmids were indistinguishable from those with the parental, unmodified plasmids.

After gel electrophoresis in 2% agarose, products of digestion were purified with the Geneclean kit provided by Bio 101, and the desired restriction fragments were mixed in order to obtain a given chimera. The DNA sequence corresponding to all chimeric proteins was verified by automatic sequencing (MWG-Biotec, Ebersberg, Germany).

High-temperature growth studies.

To ensure a strong repression of the lac promoter under the control of which DnaK, Hsc70, and their chimeras are expressed, strain BB2393 was transformed with the pDMI.1 plasmid encoding the LacI repressor. The resulting strain was then transformed by the various constructions. Transformant cells were plated on LB media containing ampicillin (100 μg/ml) and kanamycin (25 μ/ml).

For each construction, a single colony was picked and inoculated into 2 ml of LB medium containing ampicillin and kanamycin for an overnight culture at 30°C. Aliquots of 10 μl of this sample and successive 10-fold dilutions of it were spotted on an LB agar plate containing ampicillin and kanamycin with or without IPTG (isopropyl-β-d-thiogalactopyranoside; 100 μM). For each construction, test plates were incubated at 30 and 43°C for 24 h. After the test, to control the results, each plasmid was purified and used to transform again competent BB2393 cells carrying the pDMI.1 plasmid. Each test was performed three times.

λ phage growth studies.

Tests measuring the levels of λ phage resistance or sensitivity of the different E. coli strains were performed after growth overnight at 30°C in kanamycin- and ampicillin-containing LB medium supplemented with 10 mM MgSO₄ and 0.2% maltose, with or without IPTG (100 μM). The cells were then spread with 0.8% top agar on agar plates containing the same components. Serial dilutions of a λvir phage stock (5 × 10⁹ PFU/ml) were spotted on the top agar, and plates were incubated overnight at 30°C, resulting in lysis or no lysis of each bacterial strain.

SDS-PAGE, immunoblots, and quantification.

Exponentially growing 30°C cultures of MC4100, BB1553, BB2393, and BB2393 carrying the different constructions and the pDMI.1 plasmid were induced by using 100 μM IPTG for 5 h to allow expression of wild-type or chimeric proteins. Two milliliters of each culture was subjected to sonication and then centrifugation. To partially purify the wild-type and chimeric Hsp70s from the extracts, 300 μl of the soluble protein fraction was incubated for 5 min with 100 μl of ATP agarose beads in buffer A (20 mM Tris-HCl [pH 7.5], 3 mM MgCl₂, 1 mM β-mercaptoethanol, 1 mM EDTA) containing 20 mM KCl. After three washes with buffer B (buffer A containing 250 mM KCl), Hsp70s were released from the beads with 100 μl of buffer E (buffer A containing 20 mM KCl and 3 mM ATP). A degree of purification of about 80% could be achieved by this procedure. Cell extracts as well as partially purified proteins were subjected to sodium dodecyl sulfate-12% polyacrylamide gel electrophoresis (SDS-12% PAGE), stained with Coomassie blue or transferred to nitrocellulose paper (Hybond-C; Amersham), and then immunoblotted with anti-DnaK polyclonal rabbit antibodies (provided by Bernd Bukau). Detection was performed with the ECL detection system (Amersham) as described by the manufacturer.

To determine the cellular levels of relevant proteins, 2 ml of each exponentially growing culture at 30°C, induced with 100 μM IPTG for 5 h, was subjected to sonication and then centrifugation. The pellets were resuspended in 1 ml of buffer A, and protein concentration was determined by Lowry assay. Loading of the SDS-12% PAGE gel for each sample was adjusted based on the protein concentration data. To obtain a linear range of detection for immunoblot quantification, increasing amounts of purified DnaK ranging from 0 to 20 ng were treated in the same manner. The contents of the gels were then transferred to nitrocellulose membranes (Hybond-C; Amersham) and immunoblotted with rabbit anti-DnaK polyclonal antibodies (DAKO), followed by incubation with I¹²⁵-protein A. Detection was performed with a PhosphorImager, and quantification was obtained with ImageQuant software.

RESULTS

Rat Hsc70 is unable to complement an E. coli DnaK-deficient strain for growth at high temperatures and λ phage propagation.

The BB2393 strain used for the complementation studies reported here carries an amber mutation on the dnaK gene, dnaK103(Am) and is devoid of a functional DnaK protein (19) (Table 1). The absence of the DnaK protein was verified on immunoblots of cell extracts with polyclonal anti-DnaK antibodies. As shown in Fig. 2, whereas DnaK is present in cell extracts of the wild-type strain (Fig. 2A, lane 2), it is absent in those of the BB2393 strain, just as it is absent in those of BB1553 (ΔdnaK52), a strain in which the dnaK gene has been deleted (Fig. 2A, lanes 3 and 4). The same results were observed after partial purification of DnaK from these strains (Fig. 2B). The BB2393 dnaK103(Am) strain was chosen in this study over the BB1553 (ΔdnaK52) deletion strain since it has about normal levels of functional DnaJ cochaperone by contrast to the deletion strain, in which the essential DnaJ cochaperone level is reduced by more than 95% (19, 28). Normal amounts of DnaJ have been shown to be of great importance for studies of the complementation of E. coli DnaK defects by Bacillus subtilis DnaK (20).

FIG. 2.

Immunoblots of BB2393 [dnaK103(Am)] and BB1553 (ΔdnaK52) cell extracts before (A) and after (B) purification of DnaK (see Materials and Methods). Lane 1, purified DnaK; lane 2, MC4100 wild-type strain; lane 3, BB2393[dnaK103(Am)]; lane 4, BB1553 (ΔdnaK52) (lane 4).

As shown in Fig. 3B, line 1, the BB2393 dnaK103(Am) strain does not grow at 43°C, although it grows normally at 30°C, and is unable to support the growth of λ phage, by contrast to the MC4100 wild-type strain, which grows normally at 30 and 43°C and which supports the growth of λ phage (not shown). As expected, the IPTG-induced expression of the wild-type DnaK protein (NPC [Fig. 1]) in this strain complemented these two phenotypes (Fig. 3B, line 2). However, expression of rat Hsc70 (N′P′C′, where N′, P′, and C′ are the domains of Hsc70 that correspond to DnaK N, P, and C, respectively) was not able to do so, and neither thermoresistance at 43°C nor growth of λ phage was observed (Fig. 3B, line 3) even though the protein was present (Fig. 3C, line 3) at an intracellular level comparable to that of DnaK (Fig. 3D, lines 2 and 3). Thus, there seems to be no correlation between the protein expression level and complementation properties. Note, however, that DnaK and Hsc70 are overexpressed in these strains at levels about 20 times those for the wild-type strain, which has about 5 ng of DnaK/μg of total soluble proteins.

FIG. 3.

Complementation of the E. coli BB2393 dnaK103 strain by DnaK and Hsc70 proteins. (A) Schematic structures of the E. coli DnaK and rat Hsc70 used for complementation of the E. coli BB2393 dnaK103 strain showing the three domains: the N-terminal ATPase domain (N in DnaK and N′ in Hsc70), the substrate-binding domain (P in DnaK and P′ in Hsc70), and the C-terminal domain (C in DnaK and C′ in Hsc70). (B) Cell growth at high temperatures. Overnight cultures (30°C) of BB2393 dnaK103:pDMI.1:pUHE21 (no insert), BB2393 dnaK103:pDMI.1:pdnaK (DnaK), and BB2393 dnaK103:pDMI.1:phsc70 (Hsc70) were prepared as described in Materials and Methods. Serial dilutions (top of the panel) of these saturated overnight cultures were then spread on kanamycin- and ampicillin-containing LB plates, in the absence (not shown) or presence of IPTG (100 μM) and incubated at 30 and 43°C for 24 h. Undiluted aliquots of overnight cultures were used to harvest bacteria for measuring the lytic growth of λ phage, as described in Materials and Methods. (C) Immunoblots. See Materials and Methods. (D) Cellular levels of relevant proteins, obtained as described in Materials and Methods, are expressed as nanograms of relevant protein per microgram of total soluble proteins in extracts. DnaK and Hsc70 are overexpressed in these strains at levels about 20-fold higher than that for the wild-type strain (not shown), which has about 5 ng of DnaK/μg of soluble proteins.

Based on this result, it was therefore of interest to determine the structural elements of DnaK required to ensure growth at high temperatures and propagation of λ phage.

Rationale for the design of the Hsc70-DnaK chimeric proteins by domain swapping.

The rationale for the design Hsc70-DnaK chimeric proteins was that of whole-domain exchange between Hsc70 and DnaK, taking into account the modular structure of these proteins. Indeed, the fact that the three domains composing the Hsp70s can be expressed separately in and purified from E. coli or obtained by limited proteolysis indicates that these domains behave as true independent folding and structural units. Moreover, the structure of the three isolated domains has been established by X-ray crystallography and nuclear magnetic resonance, and their associated functional properties have been studied (7, 9, 12, 23). Thus, the respective domains of the different members of the Hsp70 family can be swapped with confidence since the structural integrity and overall stability of the parent proteins should be maintained in the resulting chimeric proteins.

Therefore, the eight possible combinations among the three respective domains, N′, P′, and C′ of Hsc70 and N, P, and C of DnaK, were constructed and analyzed for their ability to complement the temperature sensitivity phenotype of the BB2393 dnaK103 strain and growth of λ phage. Two splice junctions corresponding to the domain boundaries defined by structural and functional studies were introduced in solvent-accessible connecting loops (Fig. 1): a first junction point at residue 386 between the N and P domains of DnaK, corresponding to residue 389 in Hsc70, and a second junction point at residue 557 between the P and C domains (at the end of helix α2, which forms the putative lid), which corresponds to residue 563 of Hsc70. The introduction of these junction points entailed some substitutions and insertions in the protein sequences (see Material and Methods). Nevertheless, even though these modifications, located at solvent-exposed loops connecting the domains, were not expected to change the functional properties of the proteins, it was verified that the complementation properties of DnaK and Hsc70 were not affected by these changes and were indistinguishable from those of the wild-type proteins reported in Fig. 3 (results not shown).

The peptide-binding domain of DnaK is essential for growth of E. coli cells at high temperatures and for λ phage replication.

As shown in Fig. 4, cells bearing the NP′C′ chimera, having the N-terminal domain of DnaK and the peptide-binding and C-terminal domains of Hsc70, do not grow at 43°C and do not support λ phage growth (Fig. 4B, line 2), even though the protein is expressed at levels comparable to those of other chimeras (Fig. 4C and D, line 2). This indicates that the presence of the N-terminal domain of DnaK in the chimera is not sufficient to restore growth. However, its counterpart, chimera N′PC, having the peptide and C-terminal domains of DnaK and the N-terminal domain of Hsc70, is able to restore growth (Fig. 4B, line 1), indicating that the presence of the peptide-binding and C-terminal domains of DnaK in the hybrid protein is necessary for complementation of both phenotypes. This is due to the sole presence of the peptide-binding domain of DnaK in the chimeric protein, since a strain carrying Hsc70 in which only the peptide-binding domain is replaced by that of DnaK (N′PC′) is able to grow at 43°C and to support λ phage growth (Fig. 4B, line 4). As shown in Fig. 4B and D, the difference in complementation properties between the various chimeras is not related to differences in intracellular amounts of the relevant protein, since all chimeras are expressed at comparable levels, but rather reflects intrinsic functional differences. Thus, whether the N-terminal and C-terminal domains come from Dnak or Hsc70 (N or N′ and C or C′, respectively) in the hybrid protein, only the peptide-binding domain of DnaK (P) appears to be the determinant for complementation of the dnaK strain for growth at high temperatures and for propagation of λ phage (compare lines 1, 4, and 5 of Fig. 4B).

FIG. 4.

Complementation of the E. coli BB2393 dnaK103 strain by the different chimeric Hsp70 proteins. The nomenclature of the different domains (dark boxes, Dnak; white boxes, Hsc70) of the proteins and the procedures are as described in the legend to Fig. 3 (for corresponding plasmids, see Table 1).

DISCUSSION

From these studies, it appears that rat Hsc70, which has more than 50% sequence identity in the N-terminal domain and peptide-binding domain with E. coli DnaK (3, 34), is unable to ensure growth of the BB2393 dnaK103(Am) strain at high temperatures or to support the growth of λ phage. Nevertheless, an Hsc70 in which the peptide-binding domain is replaced by that of DnaK (chimera N′PC′) can restore these two phenotypes, indicating that the P domain of DnaK is the determining factor for growth at high temperatures and λ phage propagation. Most importantly, the P domain seems also to have a species specificity since an E. coli DnaK in which only the P domain is replaced by that of rat Hsc70 (chimera NP′C) is inefficient and unable to ensure thermoresistance and phage growth. These findings, which suggest that functional specificity is related to peptide binding specificity, are in contrast with those reported for Saccharomyces cerevisiae Hsp70 Ssa-Ssb chimeric proteins (17). However, the phenotypes analyzed in such studies, cold sensitivity and hygromycin B sensitivity, are distinct from thermoresistance and phage growth, addressed in this work,. Moreover, the chimeras used by James et al. (17) were made using Ssa, a yeast “classical” Hsp70 that is functionally related to DnaK and the Hsc70 family, and Ssb, an “unconventional” Hsp70 that has divergent functional properties (24).

Functional specificity of the peptide-binding domain of DnaK for growth at high temperatures and λ phage multiplication should depend on the peptide-binding site itself and/or on the dynamics of the helical lid. In this respect, the peptide-binding domain of DnaK (P) and that of Hsc70 (P′) are both composed of two regions (Fig. 1): a β sandwich subdomain, holding the peptide-binding site, and an α-helical region, which forms a lid controlling the accessibility to the peptide-binding pocket (9, 23, 34). As far as the substrate-binding site is concerned, it is exceptionally well conserved in Hsp70s in general and in DnaK and Hsc70 in particular, and most residues involved in peptide binding are identical in both proteins. Moreover, substrate specificities in vitro for Hsc70 and DnaK are comparable; both proteins bind short peptides of 5 to 7 residues, mostly hydrophobic (12a, 32, 33, 34), and Hsc70 can substitute for DnaK in protein renaturation in vitro (35). Thus, it is unlikely that functional specificity of the peptide-binding domain of DnaK is due exclusively to the peptide-binding site, unless the latter has a more stringent peptide binding specificity in vivo than in vitro. However, functional specificity could be related to the helical region forming the lid over the binding site, which regulates access to it by a latch-like mechanism (20a, 34). Indeed, even though P and P′ are very similar in the substrate-binding site, there is a strong sequence variation between them in the helical region that forms the lid. In fact, it has been proposed that changes in amino acid composition (25) and orientation (21) of this latch in Hsc70 relative to DnaK are the determinant of DnaK chaperone activity (19). Hence, dynamics in the latch opening and closing may be involved in discriminating substrates in vivo and ultimately in conferring a specific target protein-binding capacity to P but not to P′. Finally, the ability of P, but not of P′, to complement may also be due to specific interactions in vivo with the cochaperones DnaJ and GrpE, or yet-unknown interactions with critical components of the cell machinery

It is well established that thermoresistance and λ phage propagation in E. coli are based on the ability of the DnaK-DnaJ-GrpE chaperone system to prevent heat-induced damage and to interact with the phage replication protein complex (20, 27). DnaJ is known to bind to the N-terminal ATPase domain of DnaK and other Hsp70s, including Hsc70. However, GrpE is known to bind to DnaK but not to Hsc70 since the latter lacks the primary binding sites in the N-terminal ATPase domain (4). Thus, all the proteins studied here that are able to complement can in principle bind to DnaJ through their N-terminal domains be it N′ of Hsc70 or N of DnaK. However, only proteins having the ATPase domain of DnaK (N) bind GrpE. In spite of this, two chimeras having the ATPase domain of Hsc70 (N′PC and N′PC′) are still able to complement, even though they may not be able to bind GrpE. It is then possible that these chimeras do not need GrpE binding to activate nucleotide exchange, since nucleotide exchange is already fast, and that they do not have to be stimulated, as has been shown for Hsc70 (15). Alternatively, GrpE may interact with the P domains of these chimeras, as has been proposed on the basis of crystallographic and mutagenesis data that additional GrpE binding sites in the C-terminal domain of DnaK do exist (13). This is corroborated by the fact that even the chimeras in which the N-terminal domain of DnaK is present and where an interaction with GrpE is expected to be effective can complement the loss of DnaK only if the peptide-binding domain of DnaK is present.

Altogether, complementation results presented here indicate that the peptide-binding domain of DnaK is essential for the protection of E. coli cells at high temperatures and for phage growth.