Structure-function analysis of the zinc finger region of the DnaJ molecular chaperone.

DnaJ is a molecular chaperone, which not only binds to its various protein substrates, but can also activate the DnaK cochaperone to bind to its various protein substrates as well. DnaJ is a modular protein, which contains a putative zinc finger motif of unknown function. Quantitation of the released Zn(II) ions, upon challenge with p-hydroxymercuriphenylsulfonic acid, and by atomic absorption showed that two Zn(II) ions interact with each monomer of DnaJ. Following the release of Zn(II) ions, the free cysteine residues probably form disulfide bridge(s), which contribute to overcoming the destabilizing effect of losing Zn(II). Supporting this view, infrared and circular dichroism studies show that the DnaJ secondary structure is largely unaffected by the release of Zn(II). Moreover, infrared spectra recorded at different temperatures, as well as scanning calorimetry, show that the Zn(II) ions help to stabilize DnaJ's tertiary structure. An internal 57-amino acid deletion of the cysteine-reach region did not noticeably affect the affinity of this mutant protein, DnaJDelta144-200, to bind DnaK nor its ability to stimulate DnaK's ATPase activity. However, the DnaJDelta144-200 was unable to induce DnaK to a conformation required for the stabilization of the DnaK-substrate complex. Additionally, the DnaJDelta144-200 mutant protein alone was unimpaired in its ability to interact with its final sigma32 transcription factor substrate, but exhibited reduced affinity toward its P1 RepA and lambdaP substrates. Finally, these in vitro results correlate well with the in vivo observed partial inhibition of bacteriophage lambda growth in a DnaJDelta144-200 mutant background.

The Hsp70 family of proteins, the DnaK for Escherichia coli being the prototype, participate in a variety of cellular functions, such as protein folding, proteolysis, protein transport, the activation of various transcriptional and replication factors to bind to specific DNA sequences, as well as the protection and renaturation of some heat-labile proteins. These observations have led to their classification as molecular chaperones (for review, see Georgopoulos et al. (1994) and ). However, recent findings have made it clear that DnaK does not function alone. Two other heat shock proteins, DnaJ (the prototype of the eukaryotic Hsp40 proteins) and GrpE (the prototype of the eukaryotic Hsp24 proteins), are also known to participate in these reactions . The initial studies of bacteriophage (Alfano and McMacken, 1989;Zylicz et al., 1989;Zylicz, 1993) and P1 DNA replication (Wickner et al., 1992), followed by detailed in vitro protein folding experiments (Langer et al., 1992;, have revealed the intricate interactions of the DnaK/DnaJ/GrpE machinery, whose activity is stoichiometrically coupled to ATP hydrolysis . Following ATP hydrolysis, DnaK changes its conformation to the DnaK*-ADP form (the * indicates that this conformation cannot be reached by simply preincubating DnaK with ADP), which destabilizes its complex with protein substrates. In this form DnaK binds and releases protein substrates very fast. In order to stabilize the DnaK-substrate complex, the presence of DnaJ is required. In this case, DnaJ changes DnaK* conformation in such a way that the affinity of DnaK for both native and denatured protein substrates is now increased . In addition, DnaJ, which alone can bind to several protein substrates  and perform molecular chaperone functions (Schroder et al., 1993), can "target" DnaK to those substrates that directly interact with DnaJ (Szabo et al., 1994;. After the DnaJ (substrate-DnaK-ADP) complex formation, the GrpE protein and ATP hydrolysis are required to release and recycle DnaK from this complex . Thus, the DnaK chaperone, in a GrpE/ATP-dependent reaction, dissociates from the substrate complex and is converted back to the DnaK*-ADP conformation, which is ready to rebind (in a DnaJdependent reaction) to its protein substrates .
Both genetic and biochemical studies of various eukaryotic DnaJ-like proteins indicate that most, if not all, of the activities of E. coli DnaJ have been functionally conserved throughout evolution (Caplan et al., 1993;Silver and Way, 1993;Cyr et al., 1994). Members of the Hsp40 family are structurally diverse, containing different combinations of four domains. All Hsp40 members contain a "J-domain" (approximately 70 amino acids long), which is the most highly conserved and is responsible for stimulating the ATPase activity of the DnaK chaperone . Recently, using NMR methodology, the tertiary structure of the "J-domain" has been determined (Szyperski et al., 1994). Following the J-domain is a 35-amino acid-long region that is rich in both the Gly and Phe amino acids (the so-called G/F module). The deletion of the G/F module drastically interferes with the DnaJ-dependent stabilization of the DnaK-substrate complexes . Downstream of the G/F-rich module of some DnaJ-like proteins (for a review, see Caplan et al. (1993)) is a Cys-rich region that contains four repeats of the sequence CXXCXGXG (Bardwell et al., 1986;Ohki et al., 1986), which strongly resembles one of the zinc finger motifs (Table I). The COOH-terminal region of all DnaJlike proteins is much less conserved and could be responsible for substrate binding.

MATERIALS AND METHODS
Protein Purification-DnaJ, DnaK, 32 , and P proteins were purified according to Zylicz et al. (1985), Zylicz et al. (1987), Liberek et al. (1992) and Zylicz et al. (1989), respectively. The P1 replication protein, RepA, was the kind gift of Dr. Dhruba K. Chattoraj (National Institutes of Health). The purity of all enzymes was greater than 95%. The DnaJ concentration was estimated by performing the amino acid analysis.
The dnaJ⌬144 -200 cysteine-rich internal deletion mutant was constructed by polymerase chain reaction using the pJ10 (5Ј-CCGGATC-CCATATGCTCTTCCAGAGTCGG-3Ј and pJ11 (5Ј-CCGGATCCCATAT-GCATGGTCATGGTCGTG-3Ј) primers. Flanking primers to pJ10 and 3Ј to pJ11 were used to amplify the DNA. Both pJ10 and pJ11 were engineered with NdeI restriction sites (underlined). The polymerase chain reaction products were then digested with NdeI and ligated together. The gel-purified products were then digested with SacI and DraIII and cloned into the corresponding sites of the pDW19dnaJ plasmid , generating dnaJ⌬144 -200. The resulting DnaJ⌬144 -200 protein has a 57-amino acid deletion (residues 144 -200) and an insertion of 2 amino acids (His-Met). The accuracy of the dnaJ⌬144 -200 construct was verified by DNA sequence analysis. DnaJ⌬144 -200 was purified essentially as wild type DnaJ, except that the urea extraction step was replaced by increasing the ionic strength of the lysis buffer (1 M KCl). DnaJ Zn(Ϫ) apoprotein was prepared by titration of DnaJ protein with PMPS 1 followed by dialysis, as described by Giedroc et al., 1986. Zinc Content Measurements-Mercurial promoted Zn(II) release experiments and atomic absorption spectroscopy measurements were performed in buffer A (20 mM Tris/HCl, pH 7.4, 120 mM NaCl, 10% (v/v) glycerol). The buffer A was passed over the Chelex 100 (Bio-Rad) column to remove any residual heavy metal and divalent metal contamination. After such a treatment, the Zn(II) content in buffer A was shown to be lower than 10 Ϫ7 M by spectrophotometric measurements (addition of 10 Ϫ4 M PAR, followed by 1 mM EDTA). In the PMPS titration experiments, 1.5-l aliquots of 1 mM PMPS in buffer A were added to a 300-l cuvette containing the 4.35 M DnaJ. The reactants were mixed by inversion and the absorbance was measured at 250 or 500 nm. When the measurements were performed at 500 nm, 0.1 mM PAR was present in the cuvette (⌬⑀ ϭ 6.6 ϫ 10 4 M Ϫ1 cm Ϫ1 at 500 nm for (PAR) 2 Zn(II) complex). The atomic absorption of DnaJ was performed in Prof. D. Winge's laboratory (University of Utah).
Preparation of Samples for Infrared Measurements-The protein samples were prepared as described by Banecki et al. (1992). Purified DnaJ protein or DnaJZn(Ϫ) were concentrated on Centricon-30 (Amicon Division), and the buffer was exchanged by repeated centrifugation and dilution with buffer B (25 mM Hepes/KOH, 125 mM KCl, 25 mM NaCl) prepared in D 2 O, pD 7.2 (pD ϭ pH meter reading ϩ 0.4) followed by incubation in the same buffer (see Banecki et al. (1992) for more details). The spectra were recorded by a Perkin-Elmer Cetus instrument 1760-x Fourier transform infrared spectrometer as described in Banecki et al. (1992). The amide I contour plots of deconvoluted absorbance spectra were fitted with gaussian/lorentzian curves according to the method reported by Blume et al. (1988). The number and position of amide I components were taken from second derivative and deconvoluted absorption spectra (Banecki et al., 1992).
Stopped Flow Measurements-Real-time kinetics of DnaK conformational changes were performed using a double monochromator, stopped flow spectrofluorimeter (Applied Photophysics DX.17 MV; sequential mixing option) with excitation wavelength at 285 nm (10 nm bandpass) and emission at 340 nm (10 nm bandpass) as described by Banecki and Zylicz (1996).
Titration Calorimetry-The calorimetric titration were performed with a Microcal (Northampton, MA) Omega titration calorimeter (equipped with a nonovoltmeter) at 27°C in buffer B supplemented with 5 mM MgCl 2 as described by Sehl and Castellino (1990) and Odaka et al. (1994). The design and functional principles of the calorimeter and the derivation of the equation used for thermodynamic analysis of the binding through the observed heat changes has been described by Wiseman et al. (1989).
Differential Scanning Calorimetry-Scanning calorimetry measurements were performed using a Microcal MC-2 scanning calorimeter. Prior to the experiments, the sample solution was dialyzed against buffer B (prepared in H 2 O) and degassed under reduced pressure. Proteins at a concentration of 40 M, were scanned at 1°C/min. The data were analyzed using the software package provided by Microcal (Origin). A cubic splines interpolation was used for base-line correction.
Enzyme-linked Immunosorbent Assay (ELISA)-The ELISA technique used for detection protein-protein interaction was described previously in detail by .

RESULTS
Determination of the Zinc Content of DnaJ-Nucleotide sequence analysis of the dnaJ gene predicts the presence of two putative metal binding sites, which resemble one of the zinc finger motifs (Table I). To determine how many Zn(II) molecules may bind to DnaJ, we incubated a highly purified DnaJ protein sample (Zylicz et al., 1985), which is fully active in DNA replication system (Zylicz et al., 1989) with PMPS, which is known to release Zn(II) from zinc-binding proteins (Hunt et al., 1984). The formation of a mercaptide bond between the free Cys residues and PMPS can be monitored by absorbance at 250 nm (Fig. 1A). The addition of a high affinity metallochrome indicator (PAR), which changes color (absorbance at 500 nm) after the formation of the PAR/Zn(II) complex, allowed the monitoring of the amount of Zn(II) released from DnaJ after PMPS treatment (Fig. 1B). As described by Hunt et al. (1984) 4 molecules of PMPS release one Zn(II) ion. As shown in Fig. 1, the absorption at 250 nm (monitoring mercaptide bond formation) and at 500 nm (monitoring PAR)/Zn(II) complex formation) reached a plateau at a ratio of 8 PMPS for each DnaJ monomer, suggesting that 8 Cys residues are involved in binding two Zn(II) ions. Thus, the use of PMPS allowed the purification of the Zn(II)-free DnaJ apoprotein, which we designate as DnaJZn(Ϫ).
By following the absorbance of PAR at 500 nm as a function of increasing amounts of free Zn(II) (results not shown), we calculated that the absorption A 500 nm ϭ 0.56, which is obtained for a PMPS/DnaJ ratio higher than 8 (see Fig. 1B), corresponds to 1.95 Ϯ 0.2 ions of Zn(II)/monomer of DnaJ. Similar results, 2.01 Ϯ 0.1 Zn(II) ions/monomer of DnaJ, were obtained using atomic absorption spectroscopy (results not shown). We conclude that a monomer of DnaJ binds 2 molecules of Zn(II) ions.
The DnaJ protein possesses 10 Cys residues. To estimate if all of these Cys are involved in the binding of Zn(II) ions, we calculated the content of free sulfydryl groups in denatured and native DnaJ protein in the presence or absence of Zn(II). After extensive dialysis of the DnaJ protein against a buffer that did not contain DTT, we added an excess of DTNB and measured the absorbance at 412 nm as described previously by Qui et al. (1994). Under non-denaturing conditions, native DnaJ protein contains 2 Cys residues, which could interact with DTNB. Denaturation of DnaJ by 8 M Urea exposes all 10 Cys residues to DTNB. Surprisingly, DnaJ Zn(Ϫ) apoprotein possesses only 4 Cys that could interact with DTNB (results not shown). These  Berg (1990) for more details) DnaJ (E. coli) (start at position 144) C-X 2 -C-(X) 13 -C-X 2 -C DnaJ (start at position 183)

Structure-Function Analysis of DnaJ Zinc Finger Region
results suggests that, following the release of Zn(II) ions, some of the free Cys residues can form Cys-Cys bridges. The near-ultraviolet circular dichroism spectra of native DnaJ and the DnaJZn(Ϫ) apoprotein support this hypothesis (Fig. 2). In the range of 200 -250 nm, we could not detect any substantial changes in the DnaJ or DnaJZn(Ϫ) CD spectra ( Fig. 2A), suggesting that the release of Zn(II) did not substantially affect DnaJ's secondary structure. However, some differences could be seen in the 250 -290 nm range. Following the release of Zn(II) ions, the ellipticity at 270 nm increases (Fig.  2B). This reaction is reversible because subsequent addition of ZnCl 2 (10 M) partially reverses this effect (Fig. 2B). Interestingly, only when ZnCl 2 is used in the presence of 2 mM DTT does the reaction fully reverse, supporting our previous hypothesis that after release of Zn(II) some of the Cys residues of DnaJ could form Cys-Cys bridges, which are disrupted only in the presence of DTT. A similar effect has been described previously for the T4 gp32 metalloprotein (Qiu et al., 1994).
Infrared Spectroscopy Studies of DnaJ and DnaJZn(Ϫ)-To further characterize DnaJ, we measured its infrared absorbance spectra in H 2 O and D 2 O (Fig. 3A). The spectrum (amide I band) obtained in H 2 O buffer exhibits a maximum at 1650 cm Ϫ1 . The amide II band, due to NH bending and CH stretching vibrations, is present at 1550 cm Ϫ1 . In a D 2 O-containing buffer, the amide I band is shifted to 1641 cm Ϫ1 and the amide II band is shifted to about 1450 cm Ϫ1 . This latter effect is probably due to a H/D exchange reaction. The deconvoluted spectra of the DnaJ protein, shown in Fig. 3B, suggest that the amide I band is composed of at least seven components (Fig.  3B). The assignment of these bands to a particular protein secondary structure was done according to well established criteria (Byler and Susi, 1986), leading to the description of the secondary structure for the DnaJ and DnaJZn(Ϫ) apoproteins (Table II). After the release of Zn, the amount of random coil and ␣-helical structures slightly increase, but, at the same the time, the amount of ␤-structure proportionally decreases. As was suggested by the CD spectra results, the differences in secondary structure between DnaJ and DnaJZn(Ϫ) apoprotein are minor and probably local (Table II).
Infrared spectroscopy may also provide indirect information on the tertiary structure of a protein. This information could be obtained from H/D exchange studies and/or infrared spectra recorded at different temperatures in D 2 O (Banecki et al., 1992). The infrared spectra of DnaJ and DnaJZn(Ϫ) apoprotein taken at different temperatures, shown in Fig. 4, provide information about their thermal denaturation. The wild type DnaJ protein starts to denature at the temperature range between 50 and 55°C. At this temperature range, the intensity of 1655 cm Ϫ1 peak decreases while the intensity of 1635 cm Ϫ1 peak, which reflects the ␤-structure content, is not changed. In the temperature range of 65-70°C, the protein is extensively unwound and is found predominantly in a random coil struc-ture (peak 1644 cm Ϫ1 ). When the temperature is further increased, two new peaks are found (at 1618 and 1684 cm Ϫ1 ) (Fig.  4), which probably correspond to the aggregated state of the protein (Casal et al., 1988). In the case of DnaJZn(Ϫ), the corresponded peaks appear at a lower temperature (Fig. 4), suggesting that the release of Zn(II) could partially destabilize the tertiary structure of DnaJ.
Differential Scanning Calorimetry of DnaJ and Dna-JZn(Ϫ)-The conclusions reached with infrared spectroscopy on the stability of DnaJ and DnaJZn(Ϫ) apoprotein were supported by the use of the differential scanning calorimetry tech- FIG. 3. Infrared absorbance spectrum of DnaJ in the 1800 -1500 cm ؊1 region. Panel A, the infrared absorbance spectrum of DnaJ (5 M)in buffer containing H 2 O (solid line) or D 2 O (dashed line) was measured as described under "Materials and Methods." Panel B, the same, as described in panel A, absorption spectrum of DnaJ amide I band (in the presence of D 2 O) after band narrowing by Fourier deconvolution using a 18 cm-1 half-bandwidth and a resolution enhancement factor of 2.7. The shape of the deconvoluted amide I bands were simulated by gaussian/lorentzian functions. The best fit was obtained with a 30% gaussian proportion for each amide I component band. The symbols ␣, ␤, t, and r refer to ␣-helices, ␤-structure, turns, and random (non-ordered) structure, respectively.

FIG. 4. Deconvoluted spectra (amide I) of DnaJ (panel A) and DnaJZn(؊) apoprotein (panel B) in D 2 O at different temperatures.
The infrared spectra were recorded at 20, 40, 45, 50, 55, 60, 65, and 70°C. The protein concentration (5 M) and buffer conditions were as described under "Materials and Methods." The deconvolution parameters for amide I band were as described in Fig. 3B.

TABLE II Calculated positions and fractional areas of the amide I component bands for the DnaJ and DnaJZn(Ϫ) chaperone proteins
The symbols ␣, ␤, t, and r refer to ␣-helices, ␤-structure, turns, and random (non-ordered) structures, respectively.  (Fig. 5). The calorimetric profile of DnaJ shows three peaks centered at 53, 57.3, and 67.3°C (Fig. 5 and Table III), arguing that the denaturation process of DnaJ does not conform to a two-state mechanism, the transition most likely involving partially folded intermediates. This result suggests that DnaJ consists of two or more domains. As was already noted in the case of the Fourier transformed infrared differentiation spectra, the release of Zn(II) from the DnaJ protein decreases its stability since the transition points are centered at 49.6, 56.2, and 68.0°C (Fig. 6, Table III). Deletion of the 57-amino acid Cys-rich region, consisting of residues 144 -200 (DnaJ⌬144 -200), leads to a further decrease in DnaJ's transition points (results not shown), suggesting that the binding of Zn(II) ions, or the disulfide bridges formed following the release of Zn(II) from the DnaJ protein are involved in the stabilization of DnaJ's tertiary structure. In control experiments, using size high pressure liquid chromatography under the conditions described by , we showed that all three forms, DnaJ, DnaJZn(Ϫ), and DnaJ⌬144 -200, are dimeric (results not shown), which suggests that Zn(II) is not involved in the formation of DnaJ dimers. Interaction of DnaJ, DnaJZn(Ϫ), and DnaJ⌬144 -200 with DnaK-Recently, we showed that in the presence of ATP, DnaJ can interact with the DnaK molecular chaperone and stabilize DnaK complexed with its various denatured or native protein substrates . The conclusion from these and other studies is that DnaJ protein can induce conformational changes in the DnaK protein, which lead to the stabilization of the substrate-DnaK complex. Such conformational changes of DnaK can be monitored by the changes in the fluorescence of the single tryptophan 102 residue, located in the amino-terminal domain of DnaK, near the ATP binding site . In contrast, DnaJ does not possess any tryptophan residues. As shown in Fig. 6, both DnaJ and DnaJZn(Ϫ) apoprotein exhibit similar properties in modulating DnaK's conformation.
The DnaJ⌬144 -200 exhibits a significant reduction in its ability to change DnaK's conformation (Fig. 6). This observation can be explained by assuming that the internal 57-amino acid deletion causes a conformational change in DnaJ, which results in its inability to interact efficiently with the DnaK chaperone. To test this possibility, we used the very sensitive ELISA assay to probe for weak interactions between the DnaJ and DnaK proteins . In this assay, DnaJ or DnaJ⌬144 -200 were first bound to the ELISA plates, and following extensive washing and blocking with excess of BSA, increasing concentrations of the DnaK protein (in the presence of ATP and BSA) were added. Interestingly, using this assay, both DnaJ and DnaJ⌬144 -200 interacted with DnaK with identical affinities (Fig. 7). As a negative control for the specificity of the ELISA assay we used DnaKc94, a truncated protein lacking the 94 carboxyl-terminal amino acids. As published previously , the substitution of DnaK by DnaKc94 abolished the DnaK-DnaJ interaction (Fig. 7).
In a second approach, we monitored DnaJ-DnaK interactions using the microcalorimetry method. The raw data for the calorimetric titration of DnaK and a mixture of DnaK and DnaJ with ATP-Mg 2ϩ are shown in Fig. 8. The area of the downward peak is proportional to the heat released during three distinct processes, namely the binding of ATP to DnaK, the hydrolysis of ATP, and the conformational changes of the DnaK protein.
The presence of the DnaJ protein substantially increases the heat released during DnaK's interaction with ATP (Fig. 8B), supporting the previous findings that DnaJ, in the absence of protein substrate, accelerates ATP hydrolysis (Liberek et al., 1991;Jordan and McMacken, 1995). In control experiments we could not detect any heat release when ATP was injected to DnaJ protein (results not shown), suggesting that DnaJ does not interact with ATP. No enthalpy changes were detected when ATP was omitted, and DnaJ was injected to DnaK (re-  Table III is represented by filled squares. Three states (dotted line) are the minimum number of states that accurately describe the denaturation process. sults not shown), suggesting that the presence of ATP is required for the DnaJ-DnaK interaction. In support of these conclusions, we showed that ATP hydrolysis is important for the DnaJ-DnaK-ADP complex formation, and such a complex can only be detected when ATP was present during both the preincubation and mobile phase of size high pressure liquid chromatography . Additionally, we found that deletion of the Cys-rich region does not significantly change the DnaJ-and GrpE-dependent stimulation of DnaK's ATPase activity (results not shown). Thus, in agreement with our previous findings, the Cys-rich region of DnaJ does not play a significant role in the binding of DnaK to DnaJ . We conclude that the deletion of Cys-rich region does not change the ability of DnaJ to interact with DnaK, but blocks the DnaJ-dependent induction of a change in DnaK's conformation. The DnaJ, DnaJ⌬144 -200, or BSA (0.5 g/ well) was loaded on an ELISA plate well in phosphate-buffered saline buffer and following binding, washing, and blocking procedures, as described by , the increasing amounts of DnaK or DnaKc94 protein in 25 mM Hepes/KOH, pH 7.4, 150 mM KCl, 25 mM NaCl and 10 mM MgCl 2 , 0.1 mM EDTA, 2 mM DTT, 5% (v/v) glycerol, 0.05% (v/v) Triton X-100, 0.2% (w/v) BSA, and 1 mM ATP. The reaction was incubated for 30 min at room temperature and crosslinked with glutaraldehyde. The amount of DnaK protein bound to DnaJ or DnaJ⌬144 -200 was estimated using an anti-DnaK serum as described by . DnaJ and DnaJ⌬144 -200 Binding to Their Protein Substrates-DnaJ is a molecular chaperone and can interact with several seemingly native proteins, e.g. 32 , RepA, P, and O . Using the ELISA assay we found that DnaJ⌬144 -200 can bind to the 32 substrate, but its ability to interact with other physiologically relevant substrates, such as P or RepA, was significantly reduced (Fig. 9,  A and B). These results suggest that DnaJ⌬144 -200 is not only blocked in the induction of DnaK's active conformational state, but itself may bind less efficiently to some of its protein substrates.
DnaJ⌬144 -200 Is Affected in Its Ability to Stimulate DnaK-Substrate Complex Formation-As shown previously, the wild type DnaJ protein and the presence of ATP is required for substrate-DnaK complex formation (Langer et al., 1992;Liberek and Georgopoulos, 1993;. We found that DnaJ⌬144 -200 does not efficiently stimulate the P-DnaK and 32 -DnaK complex formation (Fig. 10). The failure of DnaJ⌬144 -200 to efficiently stabilize the P-DnaK com-plex is probably due to two effects, namely a block in the DnaJ-dependent induction of DnaK to an active conformation and a decrease in the affinity of DnaJ⌬144 -200 for its P substrate. In contrast, 32 interacts with DnaJ⌬144 -200 with the same efficiency as does wild type DnaJ, so the inhibition of the 32 -DnaK complex formation in the presence of DnaJ⌬144 -200 is probably due mostly to a block in the transition of DnaK* to the DnaK conformation.
In support of all these findings and conclusions, we find that DnaJ⌬144 -200 is only partially active in an in vitro DNA replication system (results not shown). A similar effect is also found in vivo. As shown in Fig. 11, at 42°C dnaJ ⌬144 -200 mutant bacteria do not support bacteriophage growth to the same extent as do the wild type isogenic bacteria.

DISCUSSION
Previous studies have established that zinc fingers and other metal-binding protein domains are involved in protein/DNA interactions, protein interactions with damaged DNA, protein folding, as well as protein-protein interactions (for a review, see Berg (1990)). In this paper we show that the native DnaJ molecular chaperone is indeed a metalloprotein, which binds two Zn(II) metal ions/DnaJ monomer. The zinc binding motif described in this paper closely resembles the C 4 zinc binding domain of certain DNA-binding proteins (for review, see Berg (1990)).
The characterization of Zn(II) ions in DnaJ's structurefunction is complicated, because the release of Zn(II) from DnaJ is probably followed by the in vitro formation of S-S bridges between some of the free Cys residues, which result in a stabilization of the DnaJZn(Ϫ) apoprotein structure. Nevertheless, the release of Zn(II) ions from the DnaJ protein results in minor and probably local changes in secondary structure.
Both infrared spectroscopy and scanning calorimetry show that the release of Zn(II) ions from DnaJ results in the destabilization of DnaJ's tertiary structure, as evidenced by the decrease in the transition points during the melting of DnaJ, FIG. 9. Interaction of DnaJ and DnaJ⌬144 -200 with their various protein substrates. The various protein substrates, namely 32 (panel A) or RepA, P, or BSA (panel B) were preincubated (0.5 g/well) in ELISA plate wells in phosphate-buffered saline buffer. Following the washing and blocking procedures, an increasing amount of DnaJ or DnaJ⌬144 -200 was added in the presence of 25 mM Hepes/KOH pH 7.4, 150 mM KCl, 25 mM NaCl, 10 mM MgCl 2 , 0.1 mM EDTA, 2 mM DTT, 5% (v/v) glycerol, 0.05% (v/v) Triton X-100, 0.2% (w/v) BSA. The reaction was preincubated for 30 min at room temperature, and following glutaraldehyde cross-linking, the amount of DnaJ or DnaJ⌬144 -200 bound to different protein substrates was estimated using an anti-DnaJ serum as described by .
FIG. 10. DnaJ-dependent formation of the substrate-DnaK complex. The protein substrates 32 or P were preincubated (0.5 g/ml) in ELISA plates as described by . Following the binding, blocking, and washing procedures, an increasing amount of DnaJ or DnaJ⌬144 -200 in the presence of 50 ng of DnaK (in each well the concentration of DnaK was maintained constant) was incubated in the presence of 1 mM ATP. The buffer conditions were the same as those used in Fig. 9. After 30 min at room temperature, glutaraldehyde was added and the amount of DnaK bound to its different protein substrates was estimated using an anti-DnaK serum as described by  while simultaneously increasing its aggregation state. The role of Zn(II) ions in stabilizing the tertiary structure of a protein was originally suggested by Giedroc et al. (1987). The finding that DnaJ has three melting transition points (53.0, 57.3, and 67.3°C) suggests that DnaJ has a complicated domain structure. The release of Zn(II) ions or deletion of the Cys-rich region results in the decrease of the temperature transition points, suggesting that the structure of DnaJ is rather compact and that the changes in one domain may influence the structure of other domains.
The DnaJ molecular chaperone is known to interact with several native proteins. For example, it interacts with high affinity with RepA and 32 and with low affinity with the O and P proteins (for review, see ). In this paper we have shown that the mutant DnaJ⌬144 -200 (deletion of the entire Cys-rich region) protein interacts with wild type efficiency with 32 , but interacts with reduced efficiency with its RepA and P substrates. These results suggest that the binding of DnaJ to its substrates is complex and may involve various modes or conformations. In control experiments we found that the DnaJ⌬144 -200 mutant protein exists as a dimer in solution and binds to DnaK with normal efficiency. This last result supports our previous findings, namely that the amino-terminal "J domain," which lacks the Cys-rich region, is responsible for binding to the DnaK protein and for stimulating of DnaK's ATPase activity .
Recently, it has become clear that, in addition to the role that DnaJ plays in targeting the DnaK chaperone to the DnaJsubstrate complexes, DnaJ is also required for the stabilization of the DnaK-substrate complex. Hartl and colleagues have previously suggested that the DnaK-ATP form could bind to the DnaJ-modified substrates, and following ATP hydrolysis form a stable DnaJ(substrate-DnaK-ADP) complex (Szabo et al., 1994) (see also McCarty et al. (1995)). However, the results presented in this paper show that the situation is far more complex. For example, the DnaJ⌬144 -200 mutant protein, which stimulates DnaK's ATPase activity like wild type DnaJ ϩ , cannot efficiently change DnaK's conformation, resulting in a decrease in the stability of the DnaK-P and DnaK-32 complexes. This, in turn, suggests that perhaps the major role of DnaJ in the formation of a stable substrate-DnaK complex is not the stimulation of ATP hydrolysis but rather a change in DnaK's conformation, which allows the stable substrate-DnaK complex formation. This interpretation is in agreement with our recent stopped flow data, according to which DnaK (in the absence of DnaJ) during ATP hydrolysis changes its own conformation to the so-called DnaK*-ADP form, which possesses limited affinity for its protein substrates, but binds efficiently to DnaJ. In the presence of DnaJ, the DnaK*-ADP form is converted to a different conformation, capable of forming a stable complex with DnaK's substrates . Moreover, such DnaJ-dependent activation of DnaK's structure can occur in the absence of protein substrate . FIG. 11. Growth of bacteriophage cI on dnaJ ؉ or dnaJ⌬144 -200 bacteria. Isogenic dnaJ::kan bacteria containing a plasmid carrying either dnaJ ϩ or dnaJ⌬144 -200 were monitored for their ability to support growth. Approximately 2 ϫ 10 8 bacteria were incubated along with 1 ϫ 10 7 cI bacteriophage for 15 min at room temperature. The cells were centrifuged and washed before being resuspended in 20 ml of L-broth supplemented with 100 g/ml ampicillin and 0.4% maltose. The cultures were grown at 30°C. Samples were taken at various time points, vortexed with chloroform, and diluted, and the number of plaque-forming units was determined on the B178 strain, as described by Wall et al. (1994).