Regular paper on-line at: www.actabp.pl The DnaK chaperones from the archaeon Methanosarcina mazei and

Hsp70 (DnaK) is a highly conserved molecular chaperone present in bacteria, eukaryotes, and some archaea. In a previous work we demonstrated that DnaK from the archaeon Methanosarcina mazei (DnaK(Mm)) and the DnaK from the bacterium Escherichia coli (DnaK(Ec)) were functionally similar when assayed in vitro but DnaK(Mm) failed to substitute for DnaK(Ec) in vivo. Searching for the molecular basis of the observed DnaK species specificity we compared substrate binding by DnaK(Mm) and DnaK(Ec). DnaK(Mm) showed a lower affinity for the model peptide (a-CALLQSRLLS) compared to DnaK(Ec). Furthermore, it was unable to negatively regulate the E. coli sigma32 transcription factor level under heat shock conditions and poorly bound purified sigma32, which is a native substrate of DnaK(Ec). These observations taken together indicate differences in substrate specificity of archaeal and bacterial DnaKs. Structural modeling of DnaK(Mm) showed some structural differences in the substrate-binding domains of DnaK(Mm) and DnaK(Ec), which may be responsible, at least partially, for the differences in peptide binding. Size-exclusion chromatography and native gel electrophoresis revealed that DnaK(Mm) was found preferably in high molecular mass oligomeric forms, contrary to DnaK(Ec). Oligomers of DnaK(Mm) could be dissociated in the presence of ATP and a substrate (peptide) but not ADP, which may suggest that monomer is the active form of DnaK(Mm).


INTRODUCTION
The Hsp70 proteins form a family of evolutionarily conserved heat shock proteins which act as molecular chaperones, assisting in the folding of nascent polypeptides, in the refolding of partially denatured proteins and in many other cellular pro-cesses.They have a weak ATPase activity and function by reversibly binding substrates at the expense of ATP hydrolysis.The Hsp70s of prokaryotic organisms are called the DnaK proteins, after the model protein of Escherichia coli, and function as a triad, together with the co-chaperones Hsp40 (DnaJ) and GrpE (Bukau et al., 2000;2006;Truscott et al., 2003; 2007 M.A. Żmijewski and others Deuerling & Bukau 2004;Young et al., 2004;Kultz 2005).
DnaK of E. coli is composed of two domains: an approximately 44 kDa amino-terminal ATPase domain and a carboxy-terminal substrate binding domain (SBD) of about 27 kDa.The crystal structures of the ATPase domain (Harrison et al., 1997) and that of the SBD (Zhu et al., 1996) have been solved.The ATPase domain contains a groove for ATP binding; it also binds a dimeric form of the nucleotide-exchange factor GrpE.In the structure of SBD two subdomains exist: the first one, composed mostly of β-sheets, containing a cavity which binds the substrate polypeptide (displaying short stretches of hydrophobic residues on its surface), and the second one, composed mostly of α-helices, which functions as a lid covering the peptide-binding cavity (Zhu et al., 1996).
According to the current model, the α-helical subdomain forming a latch over the substrate-binding cavity is in an open conformation in the ATPbound state, allowing release of the folded polypeptide and subsequent binding of another molecule of substrate to initiate a second folding cycle.The capture and retention of the polypeptide for folding involves closing of the latch, a conformational change that requires energy from the hydrolysis of the bound ATP, which is converted to ADP.Since the ADP-bound DnaK has a higher affinity for the peptide compared to the ATP-bound form, in order to release the polypeptide, ADP has to be exchanged for ATP.Both the ATPase activity of DnaK and nucleotide exchange are regulated by the co-chaperones DnaJ and GrpE.The amino-terminal DnaJ domain binds to DnaK and stimulates ATP hydrolysis, which promotes substrate binding.Subsequently, GrpE promotes ADP dissociation thus enabling DnaK to bind ATP, which in turn promotes the DnaK-polypeptide complex dissociation and release of the folded polypeptide (Bukau et al., 2000;2006;Mayer et al., 2000;Deuerling & Bukau, 2004;Erbse et al., 2004;Young et al., 2004).
In E. coli, the dnaK, dnaJ and grpE genes belong to a heat shock regulon positively controlled by the rpoH gene product, a promoter-specific σ 32 subunit of RNA polymerase.DnaK negatively regulates the level of σ 32 by binding and promoting its proteolysis (Yura & Nakahigashi, 1999).The binding of σ 32 by DnaK is an example showing that not only unfolded polypeptides but also a native, non-denatured protein may be a substrate for this chaperone (Liberek et al., 1992).
The Hsp70/DnaK proteins are highly conserved in sequence and distribution in Bacteria and Eukarya, but in the Archaea DnaK is present only in some species, one of which is a mesophilic archaeon, Methanosarcina mazei (Macario et al., 1991).It is hypothesized that the archaeal DnaK system has been obtained by lateral transfer of the dnaK-dnaJ-grpE genes from bacteria.This hypothesis is well documented, mainly on the basis of sequence similarity (Gribaldo et al., 1999;Macario & Conway De Macario, 2001;Macario et al., 2004) and, recently, by showing a functional and structural similarity of purified M. mazei and E. coli DnaK proteins (Żmijewski et al., 2004; 2007).These similarities notwithstanding, there is a significant functional difference, revealed by the fact that the dnaK gene of M. mazei is unable to complement dnaK mutations in E. coli.E. coli dnaK mutant bacteria are thermosensitive and this phenotype can not be rescued by DnaK of M. mazei (Żmijewski et al., 2004), which indicates DnaK species-specificity.
Generally, chaperones are considered to be promiscuous, since in vitro they interact with a variety of unfolded polypeptides and the presence of exposed hydrophobic groups seems to be the only requirement for Hsp70 binding (Erbse et al., 2004).However, DnaK from one species introduced into another is frequently unable to fully substitute for the DnaK of the host cell (Sussman & Setlow, 1987;Mehlert & Young, 1989;Tilly et al., 1993;Minder et al., 1996;Mogk et al., 1999).Since the species-specificity of DnaK proteins is an unsolved problem and because very little is known so far about function of the archaeal DnaKs, we decided to investigate molecular basis of the observed species-specificity of the M. mazei DnaK.
In this work we compared the binding of a typical peptide substrate, representing an unfolded protein, and of a native protein, sigma 32 transcription factor, by DnaK Mm and DnaK Ec .
Searching for structural differences we analyzed the oligomerization status of the purified DnaKs.Our results showed a significantly lower affinity of the DnaK Mm for both types of substrates, and also differences in the quaternary structures of the purified DnaKs.

MATERIALS AND METHODS
Bacterial strains, plasmids, media.The Escherichia coli strains and plasmids used in this study are listed in Table 1.LB medium (Luria-Bertani broth) and LA (Luria agar) were prepared according to Sambrook and coauthors (1989), and were supplemented with appropriate antibiotics (when necessary): 100 µg ml −1 ampicillin, 68 µg ml −1 chloramphenicol, and 30 µg ml −1 kanamycin.
Chemicals.Deuterium oxide (99.9% 2 H 2 O), 2 HCl, and NaO 2 H were purchased from Aldrich (Sigma-Aldrich S.r.l., Milan, Italy).All other chemicals were commercial products of the purest quality purchased from Sigma (Poznań, Poland), or were obtained as indicated in the text.
Protein modeling.Modeling of M. mazei chaperone proteins was done by use of the Swiss Model server and Deep View/Swiss-PdbViewer (Peitsch 1996;Guex & Peitsch, 1997).Model of the ATPase domain of the DnaK Mm protein was done with the E. coli structure (1DKGD) used as a template.After modeling, the structures of the M. mazei proteins were minimized with the GROMOS96 force field implementation of Swiss-PdbViewer.
Fourier-transform infrared spectroscopy.Preparation of protein samples and Fourier-transform infrared spectroscopy was carried out as described previously (Żmijewski et al., 2007) Size exclusion chromatography.Size-exclusion chromatography was performed using an Agilent HPLC 1100 system equipped with a diode array detector and a Zorbax GF-250 column (Agilent Biotechnologies).In all experiments Zorbax GF-250 was equilibrated with 50 mM Tris, pH 7.5, 100 mM KCl buffer.DnaK proteins (100 µg/100 µl, or as stated in figure legends) were incubated in the presence (3 mM ATP or 3 mM ADP+Pi) or in the absence of nucleotide, or in the presence of 200 nM peptide NR (NRLLLTG) in 50 mM Tris, pH 7.5, 50 mM KCl, and 10 mM MgCl 2 , for at least 30 min; they were then injected onto an HPLC column (a 100 µl loop was used), and resolved with a 1-ml per minute flow rate.In the case of DnaK incubated with ATP, the nucleotide was included in the elution buffer (as described by Wawrzynów and Żylicz, 1995) or was omitted, as explained in the text.For each set of experiments, the column was calibrated (in the presence or absence of ATP) with the following Sigma molecular-mass standards: bovine thyroglobulin (669 kDa), apoferritin from horse spleen (443 kDa), yeast alcohol dehydrogenase (150 kDa), and ovalbumin (45 kDa).Spectra were collected at 230, 260, and 280 nm.For the final presentation of data the 230-nm spectra were chosen, because DnaK Mm contains no tryptophan, and therefore shows low absorbance at 280 nm.Furthermore, the absorbance is poorly defined in the presence of ATP, which shows a maximum absorbance at 296 nm.Determination of fluorescence spectra.A Perkin-Elmer LS55 spectrofluorimeter, equipped with a thermoregulated cuvette holder with stirrer and a Julabo thermostat, was used to record fluorescence emission spectra.All experiments were performed in 0.5 × 0.5 cm quartz cuvettes at 25°C, in a buffer containing 25 mM Hepes pH 7.0, 100 mM KCl, and 10 mM MgCl 2 .Kinetics of complex formation of acrylodan-labeled peptide a-p4 with DnaK was measured with the excitation wavelength set at 370 nm (bandpass 4.6 nm, with excitation slits set at 4.6 nm), and the spectra were recorded at 510 nm (bandpass 18.6 nm, with emission slits set at 18.6 nm) for 900 s at 25 o C. The reaction was started by addition of DnaK at final concentrations ranging from 0.1 to 5 µM.The concentration of the acrylodan-labeled peptide was held constant in all experiments (200 nM), and the final volume of the reaction mixture was 400 µl.
Determination of k obs , kinetic constants (k +1 , k -1 and K d ) and evaluation of kinetic measurements for the DnaK-a-p4 peptide binding.To determine the rate constants k +1 and k -1 for complex formation between DnaK Ec or DnaK Mm and acrylodan-labelled peptide a-p4, the apparent rate constants k obs were calculated for 200 nM peptide a-p4 and increasing concentrations (0.1 to 5 µM) of DnaK.Collected data were fitted to a single exponential equation (1): where k on represents first-order rate constant, ΔF, amplitude and F 0 , initial fluorescence.For all tested concentrations of DnaK the error of fitting for k obs was lower than 2% and fitting to a double-exponential equation resulted in an increase of fitting error.The k obs values were plotted against concentration of DnaK protein and data were fitted to the linear function.The association (k +1 ) and dissociation (k -1 ) rate constants were calculated according to equation (2): (2).
The dissociation equilibrium constants (K d ) of DnaK Mm -a-p4 and DnaK Ec -a-p4 complexes were calculated according to equation (3): All calculations were performed with Origin 7.5 software (similar results for DnaK Ec were obtained with the Dynafit program).
Assay of σ 32 levels.E. coli strains were transformed with appropriate plasmids according to (Sambrook et al., 1989).The transformed bacteria were grown in LB medium with aeration at 30°C to an OD 595 of 0.6, and then were heat-shocked at 45°C for 1 h.The heat shock was followed by a 1-h recovery period at 30°C.Equivalent amounts of each culture were harvested before and after heat shock, and after the recovery period, and were analyzed by 12.5% SDS/PAGE followed by Western blotting with anti-σ 32 antibodies.
Surface plasmon resonance spectroscopy.Studies on interaction of DnaK proteins with σ 32 were carried out by use of the BIAcore 2000 biosensor (Pharmacia Biosensor AB, Uppsala, Sweden) and determination of SPR changes.Research-grade CM5 chips were coated with σ 32 to about 2000 resonance units (RU) using the EDC/NHS coupling kit (Nethyl-N'-(dimethylaminopropyl)carbodiimide, N-hydroxysuccinimide; Pharmacia Biosense AB).The cell temperature was 25 o C, and the running buffer was 50 mM Hepes (pH 7.5), 100 mM KCl, 10 mM MgCl 2 , 0.005% (v/v) Tween 20.The DnaK Ec or DnaK Mm protein was passed over the σ 32 -coated chip and an empty control chip with a flow rate of 10 µM/min at 1 µM concentrations.As a control, BSA was used with the same concentration and conditions.ATP (12.5 µM) was added at least 5 min before injection.The signal obtained from the non-coated chip was subtracted from the signal obtained from the chip with coupled σ 32 .The σ 32 -coupled chip was regenerated by the addition of 20 µl of 1 M urea at a flow Substrate binding by archaeal DnaK rate of 10 µl/min.Spectra were evaluated with the software supplied with the BIAcore instrument.

Binding of substrates by DnaK Mm and DnaK Ec
In our previous work we demonstrated that DnaK Mm has a chaperoning capacity comparable to that of DnaK Ec ; however, the substrate (polypeptide)binding properties of the archaeal protein were not investigated (Żmijewski et al., 2004).In this work, substrate binding was assessed to determine whether or not DnaK Mm and DnaK Ec differ in specificity.For this purpose, we compared the abilities of the two proteins to bind the acrylodan-labeled peptide a-p4 (a-CALLQSRLLS), a known DnaK Ec substrate successfully used for kinetic measurements (Gisler et al., 1998).We measured by spectrofluorometry the kinetics of the binding of the a-p4 peptide by the two DnaKs, and then used the data to calculate the kinetic constants, as described under Materials and Methods.The results are shown in Fig. 1 (k obs values plotted against DnaK concentrations); Table 2 shows the calculated kinetic constants.The k +1 values were significantly different for the two proteins, indicating that the chaperone-a-p4 peptide complex formation was considerably slower (by approx.8-fold, in the experiment shown) for DnaK Mm than for DnaK Ec .The dissociation equilibrium constants (K d ) differed between the two chaperones to a smaller extent, showing nonetheless that binding of the peptide to DnaK Mm was at least 2-fold weaker (e.g., 0.91 µM for DnaK Ec vs. 2.57 µM for DnaK Mm ).
The differences in the substrate-binding properties between the archaeal and bacterial chaperones must be due to fine structural differences, most importantly those occurring in the substrate-binding domain (SBD).Our modeling of DnaK Mm showed a high degree of similarity between the SBDs of the M. mazei and E. coli proteins (Żmijewski et al., 2007).However, close examination of the substrate-binding cavities of the two DnaKs revealed that the highly conserved methionine 404 in DnaK Ec , forming an arch over the cavity, is replaced in DnaK Mm by leucine 378.Also, in the latch region, DnaK Ec histidine 544 is replaced in DnaK Mm by asparagine 518, and DnaK Ec aspartic acid 540 is replaced in DnaK Mm by glutamic acid 514 (Fig. 2A, B).

DnaK Mm is unable to down-regulate the E. coli heat-shock transcription factor σ 32 in vivo
In E. coli, DnaK Ec participates in its own regulation, by interacting with the transcription factor σ 32 and steering it toward degradation, which results in down-regulation of the dnaK gene when the cellular need for this chaperone diminishes.To find out whether DnaK Mm is able to interact with σ 32 in the cell, we investigated whether DnaK Mm can downregulate σ 32 in E. coli, as DnaK Ec does.We used E. coli strains lacking DnaK (E. coli ∆dnaK) and strains that produce a mutant DnaK protein that lacks the self-regulatory ability (E. coli dnaK756); both of these strains are unable to shut off the heat-shock response, in contrast to the wild-type.These strains were transformed with plasmids bearing the gene coding for DnaK Mm or the genes encoding DnaK Mm and DnaJ Mm .We carried out the experiment with both the dnaK Mm and dnaJ Mm genes to rule out the possibility that a lack of DnaK Mm function was caused by a poor cooperation of DnaK Mm with DnaJ Ec .The level of σ 32 protein in cell lysates from the transformants was assessed by Western blotting (Fig. 3A).We also confirmed the presence of DnaK Mm in the transformants by Western blotting (not shown).Neither the presence of the dnaK Mm gene alone nor the presence of both genes from M. mazei, dnaK Mm and dnaJ Mm , in the transformed mutant E. coli cells, caused a decrease in the levels of σ 32 under heat shock conditions (when σ 32 is elevated).In mutant E. coli cells transformed with a plasmid carrying dnaK Ec alone, or transformed with plasmids carrying both E. coli genes, dnaK Ec and dnaJ Ec , the σ 32 protein was barely detectable or undetectable, as expected.Likewise, in the control, wild-type bacterium with its own dnaK gene (E. coli B178), the σ 32 protein was undetectable.These results taken together indicate that DnaK Mm did not negatively regulate the σ 32 transcription factor in E. coli, presumably because the chaperone was unable to bind the transcription factor, contrary to what happens with DnaK Ec .

DnaK Mm poorly interacts with E. coli heat shock transcription factor σ 32 in vitro
In order to check whether DnaK Mm is unable to physically interact with σ 32 as efficiently as DnaK Ec does, we compared the binding capacities of DnaK Mm and DnaK Ec to E. coli σ 32 , by applying an enzyme-linked immunosorbent assay (ELISA) and surface plasmon resonance (SPR) spectroscopy.In ELISA, we determined the amount of DnaK protein bound to σ 32 immobilized on the plate wells, using anti-DnaK antibodies.The amount of DnaK Mm bound to σ 32 was significantly lower than that of DnaK Ec (Fig. 3B).Reverse experiments were also performed.In these, the DnaK proteins were immobilized onto the ELISA-plate wells first, and then they were incubated with serial dilutions of σ 32 ; this was followed by detection of bound transcription factor with antiσ 32 antibodies.The results, not shown, were essentially the same as those obtained with immobilized σ 32 , displayed in Fig. 3B.
The kinetics of interaction of either DnaK with immobilized σ 32 protein was measured by SPR spectroscopy.DnaK Mm interacted with σ 32 poorly as compared with DnaK Ec (Fig. 3C).In conclusion, the ELISA and SPR spectroscopy results showed that the DnaK Mm interaction with E. coli σ 32 in vitro was considerably weaker than the interaction of DnaK Ec with the same factor.This finding is in agreement with the observed inabil-  (Moro et al., 2004).(C) The substrate binding domain (SBD) of DnaK, with the latch region marked by a box.ity of DnaK Mm to negatively regulate the σ 32 levels in vivo.

Purified DnaK Mm exists predominantly in oligomeric forms
It is known that oligomerization status modulates chaperoning ability, e.g. in small heat-shock proteins (Narberhaus, 2002;Franzmann et al., 2005), in eukaryotic Hsp70 (Blond-Elguindi et al., 1993), and in some bacterial DnaKs (Motohashi et al., 1994;Schonfeld et al., 1995).Searching for differences in quaternary structure which might contribute to the differences in substrate binding by DnaK Mm and DnaK Ec we performed experiments to elucidate whether the archaeal chaperone oligomerizes under conditions resembling those in vivo.
Size-exclusion chromatography showed that DnaK Mm formed oligomers over a broad range of concentrations (0.1-5 mg/ml).The high-molecular mass oligomeric forms of DnaK Mm were predominant under all tested conditions, even at the lowest (A) Western blotting results.E. coli ΔdnaK (I), or E. coli dnaK756 (II) bacteria were transformed with the plasmid pAM1-dnaK Mm or with the plasmid pAM3-dnaKJ Mm .As positive controls, the E. coli B178 (dnaK + ) strain and the mutant strains transformed with the plasmids pMOB45-dnaK Ec or pJZ-589-dnaK Ec dnaJ Ec , were used.The E. coli ΔdnaK strain transformed with pAK1-rpoH plasmid (ΔDnaK + σ 32 ) and purified σ 32 protein (σ 32 ) were included to show the position of σ 32 in the gels.The various combinations tested are indicated above the corresponding lanes.E. coli was grown at 30°C, then heatshocked at 45°C and the heat shock was followed by a recovery period at 30°C.Equivalent amounts of each culture were harvested before (1) and (2) after heat shock, and after the recovery period (3), and were analyzed by 12.5% SDS/ PAGE followed by immunodetection with anti-σ 32 antibodies.(B) Enzyme-linked immunosorbent assay (ELISA).Shown are binding curves produced by a series of concentrations of DnaK Ec and DnaK Mm and a constant amount of immobilized σ 32 (closed triangles and circles, respectively), and bovine serum albumin (open triangles and circles).Mean values of at least three independent assays are presented; error bars represent standard deviations.(C) Surface plasmon resonance (SPR).In the example shown, 1 µM of DnaK Mm or DnaK Ec was passed over immobilized σ 32 in the presence of 12.5 µM ATP, as described under Materials and Methods.The data were normalized, and the DnaK Ec -σ 32 maximum reading was set as 100 relative units.The same results were obtained in two other independent runs: the three sets of data were collinear, with variations not exceeding 5%.
protein concentration, as shown by high-pressure liquid chromatography (HPLC) (Fig. 4A).DnaK Mm showed the coexistence of three major forms, eluting as proteins with molecular masses of 440, 300 and 135 kDa; the ratio of these forms changed very little with concentration (these molecular masses represent mean values calculated from the results of five independent experiments; molecular-mass standards were run in each experiment).DnaK Ec subjected to HPLC under similar conditions showed substantial aggregation only at high (non-physiological) concentrations (5-10 mg/ml), and the lowest-molecular-mass form (eluting as a 140-150 kDa protein) was predominant within the range of 0.1-5 mg/ml (Fig. 4B).SDS/PAGE of the fractions containing a mixture of all the various DnaK forms did not resolve these forms, but showed the bands for DnaK-Mm and DnaK Ec monomers (not shown).
The smallest DnaK Ec and DnaK Mm forms eluted (by HPLC) as species larger than 100 kDa.We assume that the 130-140 kDa form of DnaK Mm is the monomer, since it eluted like the smallest DnaK Ec form, and did not change its position upon ATP addition, when high-molecular-mass forms dissociated (Fig. 5A).Other investigators have also noticed that DnaK Ec does not behave like a 70 kDa globular protein should (Zylicz & Georgopoulos, 1984;A. Żylicz, personal communication).The presence of at least three forms with different molecular masses and the predominance of the high molecular-mass forms in the DnaK Mm preparations were confirmed by native gel electrophoresis (Fig. 5C, lanes 3, 8).In contrast, under the same experimental conditions, the low molecular-mass forms predominated in the DnaK Ec preparations (Fig. 5C, lanes 1, 7).
In order to investigate whether the DnaK Mm oligomerization has a functional significance with regard to the ATP cycle and substrate binding, we preincubated DnaK Mm with ATP, ADP+Pi, or the peptide NRLLLTG (NR), and then carried out sizeexclusion chromatography.In these experiments we used the NR peptide as a model substrate, since it has previously been shown to efficiently bind DnaK (Gragerov et al., 1994;Buczynski et al., 2001).In the presence of ATP or NR, the high-molecular-mass oligomers dissociated, and the majority of DnaK Mm eluted as about 140 kDa protein (Fig. 5A).In a similar experiment with DnaK Ec , the major peak eluted as an about 130 kDa protein.Also, the small shoulder on the ascending portion of the curve, representing oligomers, decreased in magnitude in the presence of ATP (Fig. 5B), as expected from previous knowledge on dissociation of DnaK Ec (Palleros et al., 1993).The elution profile of DnaK Mm preincubated with ATP was identical in the absence (Fig. 5A) and in the presence (not shown) of ATP in the elution buffer, which reflects the stability of the DnaK Mm -ATP complex.Dissociation of DnaK Mm in the presence of ATP or NR was confirmed by native electrophoresis (Fig. 5C).The observed dissociation in the presence of ATP and a substrate suggests that the monomeric form of DnaK Mm may be the active form, participating in peptide binding.ADP, in contrast to ATP and NR, did not cause dissociation of the DnaK Mm oligomers (Fig. 5A and 5C, lane 10), which resembled the situation described for DnaK Ec (Palleros et al., 1993).
Protein oligomerization may theoretically lead to a restricted accessibility of certain region(s) of a molecule to a solvent, which could result in a decreased access of the substrate to the protein.Our previous analysis of DnaK Mm by Fourier-transform infrared spectroscopy showed that the 1 H/ 2 H exchange of the amide hydrogens of the polypeptide chain was less complete than in the case of DnaK Ec (Żmijewski et al., 2007).The extent of the 1 H/ 2 H exchange can be measured by monitoring the intensity of the residual amide II band (encompassing the 1600-1500 cm −1 interval) absorption, i.e., the absorption of the amide II band after the 1 H/ 2 H exchange.In 1 H 2 O medium, the amide II band intensity was about 2/3 of the intensity of the amide I band (not shown).In 2 H 2 O medium, the intensity of the amide II band decreases as a consequence of the ex- change of amide hydrogens with deuterium (Osborne & Nabedryk-Viala, 1982;D'Auria et al., 2004).The larger the decrease in intensity of the amide II band, the larger the 1 H/ 2 H exchange and, in turn, the greater the accessibility of the solvent ( 2 H 2 O) to the protein.Fig. 6A, B shows the temperaturedependent changes of the DnaK Mm and DnaK Ec infrared absorbance spectra in the 1600-1500 cm −1 interval.At 20°C the absorbance of DnaK Mm close to 1550 cm -1 is higher than in the DnaK Ec spectrum, indicating a lower accessibility of 2 H 2 O to DnaK Mm .A temperature increase causes a decrease of absorption close to 1550 cm −1 (residual amide II band) due to a further 1 H/ 2 H exchange, a consequence of tem-perature-dependent molecular dynamics and/or protein denaturation.A more detailed comparison was possible when the 1 H/ 2 H exchange as a function of temperature was calculated, Fig. 6C, as done previously (D'Auria et al., 2004).Full 1 H/ 2 H exchange was reached by DnaK Mm at higher temperatures compared to DnaK Ec and the temperature at which a full exchange occurred can be estimated to be 55-60°C for DnaK Ec and about 65°C for DnaK Mm (Fig. 6C), the temperatures corresponding approximately to the melting points (Tms) previously calculated for DnaK Ec and DnaK Mm , respectively (Żmijewski et al., 2007).On the other hand, the maximum rate of exchange (midpoint of the 1 H/ 2 H exchange curves) was

DISCUSSION
It has been shown previously that there is a high functional similarity between the DnaK of M. mazei and DnaK of E. coli (Żmijewski et al., 2004), which is based on the overall structural similarity of these proteins (Żmijewski et al., 2007).In spite of these similarities, there is a significant functional difference, indicated by the fact that the dnaK gene of M. mazei is unable to complement dnaK mutations in E. coli (Żmijewski et al., 2004).To better understand the molecular basis of the observed species-specificity, we compared substrate binding by the M. mazei and E. coli DnaK proteins.We used two types of substrates: (1) a peptide, which represents DnaK general substrates -unfolded proteins, and (2) E. coli sigma 32 transcription factor, a protein which is bound by DnaK Ec in its native (folded) state.
DnaK Mm differed from DnaK Ec in its affinity for the substrate, peptide ap-4 (acrylodan-labeled CALLQSRLLS) (Fig. 1 and Table 2).SBD modeling of DnaK Mm showed a high similarity of this domain to SBD of DnaK Ec (Żmijewski et al., 2007) but also predicted small differences in the latch region, which could influence the closing and opening rates of the DnaK Mm .For example, the highly conserved methionine 404, forming an arch over the substrate-binding cavity in DnaK Ec , is absent in DnaK Mm ; DnaK Ec histidine 544 is replaced in DnaK Mm by asparagine 518, and DnaK Ec aspartic acid 540 is replaced in DnaK Mm by glutamic acid 514 (Fig. 2A, B).According to (Rudiger et al., 2000), even small changes in this region of DnaK Ec , in particular replacement of methionine 404 by another amino acid, will modify substratespecificity. Recently, it has been demonstrated that interactions forming the latch are critical in maintaining DnaK functional cycle, because their disruption renders the protein unable to refold protein substrates (Fernandez-Saiz et al., 2006).Thus, it is likely that the putative structural differences between the DnaK Mm SBD and that of DnaK Ec revealed by the model, do operate in reality, and that they contribute to the observed lower affinity of DnaK Mm for the substrate (Fig. 1, and Table 2).It has been shown previously that Hsp70s from different organisms recognize peptide substrates with differing affinities (Fourie et al., 1994), a fact which could be attributed to amino-acid substitutions within the DnaK substrate-binding cavity and in the lid region (Mayer et al., 2000).It is possible that different affinity of DnaK Mm for unfolded protein substrates could be one of the reasons why it is unable to substitute for DnaK Ec in a cell.
DnaK Mm interacted poorly, if at all, with E. coli σ 32 , a natural ligand for DnaK Ec , and it did not negatively regulate the σ 32 level in E. coli cells (Fig. 3).The observed lack of σ 32 down-regulation is not likely to be the reason why DnaK Mm does not suppress the thermosensitivity of the E. coli dnaK mutants.High levels of σ 32 promote transcription of heat shock genes which should help bacterium survive at high temperatures (Yura & Nakahigashi, 1999).However, there may be other native proteins, as yet unidentified, which are substrates for DnaK Ec , and whose binding is essential for bacterial survival at elevated temperatures.The failure to recognize such proteins by DnaK-Mm would result in its inability to suppress the thermosensitivity of the E. coli bacteria without a functional DnaK Ec .Mogk and coworkers (1999) showed that DnaK from the Gram-positive bacterium Bacillus subtilis, a protein that is very close to DnaK Mm in phylogenetic trees and shows a high percentage of identical amino acids, did not have the ability to cause degradation of σ 32 .They found that the degradation-promoting activity depends on the integrity of a carboxy-terminal DnaK Ec region (about 10 kDa, encompassing amino acids 543-637).While the primary sequence of this small region is the least-conserved portion of the DnaK proteins (about 35% identity), the putative three-dimensional structures of this region do not differ appreciably between DnaK Mm and DnaK Ec (Żmijewski et al., 2007).It remains to be determined which amino acids in the carboxy-terminal region of DnaK Ec are directly involved in σ 32 binding and unfolding, the processes that precede degradation of the transcription factor.
Searching for differences in the quaternary structure which might contribute to the differences in substrate binding, we performed size-exclusion chromatography and native gel electrophoresis, and found that DnaK Mm occurred in solution mainly as high-molecular-mass oligomeric forms (Fig. 4), and that these forms dissociated upon ATP or peptide binding (Fig. 5).Under the same conditions, DnaK Ec formed smaller oligomers, in smaller quantities.It is possible that dissociation of the DnaK Mm oligomers is necessary for full activation of the chaperone, as has been postulated for DnaK from the bacterium Thermus thermophilus (Watanabe & Yoshida, 2004).Moreover, it has been shown that the monomeric form of DnaK Ec is fully active M.A. Żmijewski and others (Palleros et al., 1993).Furthermore, the ATP-driven monomerization has been shown to occur for the eukaryotic DnaK homologs Hsp70 and Hsc70, and it has been suggested that the non-aggregated forms exert the chaperone activity (Angelidis et al., 1999).It is possible that formation of highly oligomeric forms by DnaK Mm decreases the efficiency of substrate binding/releasing and thus may be one of the reasons why DnaK Mm functioning in E. coli cell is impaired.The FTIR analysis of the 1 H/ 2 H exchange of the amide hydrogens of the polypeptide chain indicated that DnaK Mm is less accessible to the solvent than is DnaK Ec (Fig. 6).This might be caused by DnaK Mm oligomerization and could theoretically lead to a decreased access of the substrate to the protein.It can also be speculated that oligomerization of DnaK Mm may be part of an additional regulatory mechanism of DnaK function in M. mazei, and lack of this regulation in E. coli may contribute to the inefficient functioning of DnaK Mm in E. coli cells.
It is worth pointing out that this is the first study showing substrate binding by a purified archaeal DnaK and the first attempt to solve the species-specificity of an archaeal Hsp70.Furthermore, this work shows a new aspect of the archaeal DnaK protein, namely its highly oligomeric structure.

Figure 1 .
Figure 1.Kinetics of peptide binding by DnaK proteins.To follow the binding of the test polypeptide to the chaperone, we added DnaK Mm or DnaK Ec to the assay buffer containing acrylodan-labeled a-p4 (a-CALLQSRLLS) and measured the increase of acrylodan fluorescence with time, as described under Materials and Methods.Final concentrations were: a-p4, 0.2 µM; DnaK, 0.01-5 µM.The binding curves obtained were fitted to a single exponential function, and the k obs values were derived for DnaK Ec (circles) and DnaK Mm (squares).The k obs values were plotted as a function of DnaK concentration; the plots were obtained from least-squares fitting of the data to an equation for a straight line.

Figure 2 .
Figure 2. Putative differences between the substrate-binding domains of the archaeal and bacterial DnaKs.Shown are the latch regions (see the boxed structure in C) in the DnaK Ec SBD structure (A) and in the DnaK Mm SBD model (B), with some amino acids displayed in ball-and-stick format.Methionine (M404) in DnaK Ec , forming an arch over peptide-binding cavity, is replaced by L378 in DnaK Mm .Other appreciable differences at critical positions are H544 and D540 in DnaK Ec , which are replaced in DnaK Mm by N518 and E514, respectively.Also shown are the salt bridges (orange) between K548 and D431 in DnaK Ec and K522 and D405 in DnaK Mm , and hydrogen bonds (green) between amino acids of the α-helix and the outer loops L 4,5 and L 5,6 ; these salt bridges and hydrogen bonds are believed to stabilize the latch in DnaK Ec(Moro et al., 2004).(C) The substrate binding domain (SBD) of DnaK, with the latch region marked by a box.

Figure 3 .
Figure 3. Archaeal DnaK does not down-regulate bacterial σ 32 transcription factor.(A)Western blotting results.E. coli ΔdnaK (I), or E. coli dnaK756 (II) bacteria were transformed with the plasmid pAM1-dnaK Mm or with the plasmid pAM3-dnaKJ Mm .As positive controls, the E. coli B178 (dnaK + ) strain and the mutant strains transformed with the plasmids pMOB45-dnaK Ec or pJZ-589-dnaK Ec dnaJ Ec , were used.The E. coli ΔdnaK strain transformed with pAK1-rpoH plasmid (ΔDnaK + σ 32 ) and purified σ 32 protein (σ 32 ) were included to show the position of σ 32 in the gels.The various combinations tested are indicated above the corresponding lanes.E. coli was grown at 30°C, then heatshocked at 45°C and the heat shock was followed by a recovery period at 30°C.Equivalent amounts of each culture were harvested before (1) and (2) after heat shock, and after the recovery period (3), and were analyzed by 12.5% SDS/ PAGE followed by immunodetection with anti-σ 32 antibodies.(B) Enzyme-linked immunosorbent assay (ELISA).Shown are binding curves produced by a series of concentrations of DnaK Ec and DnaK Mm and a constant amount of immobilized σ 32 (closed triangles and circles, respectively), and bovine serum albumin (open triangles and circles).Mean values of at least three independent assays are presented; error bars represent standard deviations.(C) Surface plasmon resonance (SPR).In the example shown, 1 µM of DnaK Mm or DnaK Ec was passed over immobilized σ 32 in the presence of 12.5 µM ATP, as described under Materials and Methods.The data were normalized, and the DnaK Ec -σ 32 maximum reading was set as 100 relative units.The same results were obtained in two other independent runs: the three sets of data were collinear, with variations not exceeding 5%.

Figure 4 .
Figure 4. DnaK Mm and DnaK Ec differ in oligomerization state.DnaK Mm (A) and DnaK Ec (B) were examined, in the absence of ATP, at various concentrations (10-1000 µg in 100-µl volumes) by size exclusion chromatography.Shown are UV spectra of the effluent at 230 nm.Above the panels are the respective electrophoregrams obtained with 5 µg of DnaK, using 10% PAGE under non-denaturing conditions.The mobilities of the molecular-mass standards, as resolved by HPLC under the same conditions, are indicated by vertical lines, with the corresponding molecular masses in kDa shown above each line.

Figure 5 .
Figure 5. Effects of ATP, ADP, and substrate on DnaK oligomerization state.(A, B) size-exclusion chromatography.DnaK Mm (A) or DnaK Ec (B) was incubated in the presence or absence of ATP, or ADP+Pi , or in the presence or absence of NR peptide (NRLLLTG) , as indicated.The incubated mixtures were resolved by size-exclusion chromatography and the UV spectra at 230 nm of the effluents are shown.For clarity, the spectra for DnaK Ec are displayed at different heights; otherwise, they would be indistinguishable, due to their great similarity.In contrast, the spectra for the various mixtures including DnaK Mm differ considerably from one another and are displayed at the same level.(C) PAGE.DnaK Mm or DnaK Ec was incubated for 30 min in the presence or absence (the latter indicated by a minus sign above the lane) of ATP, ADP+Pi or peptide NR (NRLLLTG), as shown.The incubated mixtures were analyzed by 10% PAGE under non-denaturing conditions.Lanes 1-4 and 5-10 are from two separate gels in which the controls DnaK Mm and DnaK Ec without additives were repeated (lanes 1, 3, 7, and 8).

Table 1 . Escherichia coli strains and plasmids used in this study.
a BDUG, Biochemistry Department, University of Gdańsk M.A.Żmijewski and others

Table 2 . Rates of formation, and dissociation equilibrium constants, of DnaK Mm and DnaK Ec complexes with a-p4 (a- CALLQSRLLS) peptide
For 200 nM peptide a-p4 and 1 µM DnaK protein.bTherateconstantsk+1 and k -1 were determined by titration of acrylodan-labeled peptide a-p4 with increasing concentrations of DnaK (see Fig.1).cDissociation equilibrium constant K d was calculated from ratio k +1 /k -1 .For details on the determination of k obs and the kinetic constants, see Materials and Methods. to occur at 46.3 for DnaK Ec and at 54.8°Cfor DnaK Mm .The analysis of 1 H/ 2 H exchange thus indicates that DnaK Mm is less accessible to the solvent than is DnaK Ec and that the proteins undergo maximum rate of 1 H/ 2 H exchange at temperatures much lower than the corresponding Tms, temperatures at which the exchange is almost complete.
a Mm , and at 46.3 °C for DnaK Ec .All determinations and the calculation of arbitrary units (a.u.) were done as described in Materials and Methods.calculated