| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 278, Issue 45, 44457-44466, November 7, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
From the
Department of Molecular, Cellular, and Developmental Biology, University of Michigan, Ann Arbor, Michigan 48109-1048
Received for publication, July 11, 2003 , and in revised form, July 31, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The functions of the "foldase" DnaK and the nucleotide exchange factor GrpE are quite well understood. In contrast, however, the roles of DnaJ and its homologues are still somewhat enigmatic (2). DnaJ consists of four domains that are conserved to various degrees among the homologues. The N-terminal 76-amino acid J-domain defines proteins as members of the DnaJ/Hsp40 family and is, thus, present in all DnaJ/Hsp40 family members (3). It contains the conserved tripeptide His-Pro-Asp, which is essential for DnaJ-DnaK interactions and for the stimulation of the ATPase activity of DnaK (4–6). Adjacent to the J-domain resides the glycine- and phenylalanine-rich G/F domain (residues 77–107). Together with the J-domain, the G/F domain is essential for maximal stimulation of the ATPase activity of DnaK (7). The central domain of DnaJ is the zinc-binding cysteine-rich CR-domain (residues 144–200), which is followed by a poorly conserved C-terminal CTD-domain (8).
Hsp40 homologues that contain all four domains are called class I DnaJ homologues. They exert autonomous, DnaK-independent chaperone activity as well as a DnaK-dependent co-chaperone activity (9). The autonomous chaperone activity is characterized by the ability of excess amounts of DnaJ to bind to unfolded substrate proteins and to prevent their irreversible aggregation (10, 11). The co-chaperone activity of DnaJ, on the other hand, is based on the synergistic interaction of DnaJ and DnaK with substrate proteins and results in the efficient suppression of aggregation at much lower DnaJ concentrations than DnaJ alone is capable of. DnaJ works as the "holdase" by binding to and delivering unfolded substrate proteins to the foldase DnaK, which folds the protein. Interaction of the DnaJ-substrate protein complex with the DnaK-ATP-substrate protein complex maximally stimulates ATP hydrolysis. This is thought to cause conformational changes in DnaK that increase DnaK affinity for unfolded substrate proteins and that causes the lock-in of substrate proteins to the substrate binding site of DnaK (7, 12–15).
Class II DnaJ/Hsp40 homologues lack the central cysteinerich zinc binding domain but have two C-terminal domains, CTD1 and CTD2, instead (16). They have not been found to exert any autonomous, Hsp70-independent ability to interact with unfolded substrate proteins. However, in concert with Hsp70 homologues, they will support the refolding of unfolded proteins such as luciferase in vitro, albeit with about 4-fold less efficiency than class I DnaJ homologues (9). Class III DnaJ representatives only have the J-domain in common. They appear unable to work as molecular chaperones in vitro. In vivo, they have been found to be involved in protein secretion (17).
The cysteine-rich zinc binding region of class I DnaJ homologues is thought to be involved in the DnaK-dependent chaperone activity based on domain truncation studies and cysteine mutagenesis (9, 18, 19). However, conflicting results have been reported concerning the role of the entire zinc binding region in the autonomous DnaK-independent chaperone activity (8, 9, 19). Furthermore, the role of the individual zinc centers in the DnaK-dependent and -independent chaperone function of DnaJ has not been investigated. The cysteine-rich CR-domain contains four repeated CXXCXGXG motifs, which are invariable to all class I DnaJ proteins (20, 21). It has long been known that the two sulfur atoms of each consensus motif are involved in zinc coordination, resulting in the tetrahedral coordination of two zinc ions per DnaJ monomer (18, 19). However, the identity of the specific residues involved in each zinc coordination remained unclear until the NMR structure of the isolated cysteine-rich domain was solved in 2000 (Fig. 1) (21). Prior to the solution of the structure, it was assumed that the four immediately adjacent cysteine residues were involved in each zinc site, with the four N-terminal cysteine residues forming zinc center I and the four C-terminal cysteines forming zinc center II (19). The solution structure clarified the arrangement of the individual zinc centers and revealed that zinc center I is actually formed by the two non-adjacent cysteine motifs 1 (Cys-144, Cys-147) and 4 (Cys-197, Cys-200), which are furthest apart in primary sequence. Zinc center II is coordinated by the conserved cysteines of the two adjacent cysteine motifs 2 (Cys-161, Cys-164) and 3 (Cys-183, Cys-186) (21) (Fig. 1). This somewhat surprising arrangement meant that a number of experiments that had been conducted to investigate the function of the zinc centers in type I DnaJ homologues needed to be re-interpreted. For instance, experiments that were designed to disrupt the zinc centers individually probably only succeeded in altering one zinc center (19). This also meant that no experiments had been conducted as yet, that separately investigated the specific role of these highly conserved zinc centers.
|
To analyze the exact functional role of the two individual zinc centers in type I DnaJ homologues in vitro and in vivo, we constructed two DnaJ mutants, based on the NMR structure, which had either zinc center I or zinc center II disrupted. Our investigations revealed that zinc center I plays an important role in the autonomous, DnaK-independent chaperone activity of DnaJ and that the absence of this activity only slightly impairs the in vivo function of DnaJ. Zinc center II, on the other hand, is crucial for the in vivo function of DnaJ. Zinc center II mutants appear to lack a necessary additional interaction site between DnaJ-substrate complexes and DnaK-ADP substrate protein complexes, which seems to be essential for the refolding activity of the DnaK/DnaJ/GrpE foldase machinery.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
ZnI and C161S, C164S, C183S, C186S to generate DnaJZnII were constructed with the QuickChange site-directed mutagenesis kit (Stratagene) using pUH21–12fd12 (gift from Dr. Bernd Bukau) as the parental plasmid. The resulting plasmids were sequenced and transformed into the dnaJ deletion strain PK11 (MG1655 dnaJ Thr::Tn10) (gift from Dr. Elizabeth Craig) for phenotypical studies or into wild-type Escherichia coli WM3110 containing the plasmid pDMI (22) for protein purification.
Protein Purification—Wild-type DnaJ as well as the DnaJZnI and DnaJZnII mutant proteins were purified as previously described (22) with modifications to remove all residual detergents. After elution of the proteins from S-Sepharose Hitrap-column (Amersham Biosciences), the protein solutions were applied to a hydroxyapatite column, equilibrated in 50 mM Tris-HCl, pH 7.5, 2 M urea, 2 mM DTT,1 50 mM KCl, 0.1% Brij58. The column was then washed in 50 mM Tris-HCl, pH 7.5, 2 M urea, 2 mM DTT, 1 M KCl followed by a wash with 50 mM Tris-HCl, pH 7.5, 2 M urea, 2 mM DTT, 50 mM KCl. DnaJ, and the mutant proteins were eluted using a gradient from 0 to 400 mM KH2PO4. The pooled fractions were diluted into 50 mM Tris-HCl, pH 9.0, 2 M urea, 2 mM DTT, 100 mM KCl and applied onto a Q-Sepharose Hitrap column (Amersham Biosciences). The flow-through contained >99% pure DnaJ or the mutant proteins, respectively. The DnaJ variants were dialyzed against 30 mM HEPES-KOH, pH 8.2, 120 mM KCl, 15 mM NaCl, 4% glycerol, and the protein concentration was determined using the extinction coefficient 
280 = 13.400 M–1 cm–1. Wild-type DnaK, overexpressed from the plasmid pMOB45, and GrpE overexpressed from the plasmid pWKG20, were purified as described (22).
Zinc Measurements—The zinc content of DnaJ was determined using a combined PAR/PMPS assay (23). In this assay, free zinc readily interacts with 4-(2-pyridylazo) resorcinol (PAR) (Sigma) to form a Zn-(PAR)2 complex, which strongly absorbs at 500 nm. To determine the amount of free zinc in the protein solution, 1 µM DnaJ in 30 mM HEPES-KOH, pH 8.2, 120 mM KCl, 15 mM NaCl, 4% glycerol was mixed with 100 µM PAR and the A500 nm was monitored. To determine the amount of zinc that is cysteine-coordinated in wild-type DnaJ and mutant proteins, 2-µl aliquots of a 1 mM p-hydroxymercuriphenylsulfonic acid (PMPS) (Sigma) titration solution (in 100 µM PAR, 40 mM HEPES-KOH, pH 7.5) were added, and changes in A500 nm were monitored after each addition, until no further changes were detected (24). PMPS forms stoichiometric mercaptide bonds with thiols. This leads to the release of zinc into the solution, where it is immediately complexed by PAR, thereby turning the solution red. A zinc standard was used to determine the extinction coefficient of the Zn(PAR)2 complex at 500 nm in 30 mM HEPES-KOH, pH 8.2, 120 mM KCl, 15 mM NaCl, 4% glycerol.
CD Measurements—All DnaJ proteins were dialyzed against 30 mM phosphate buffer pH 8.2, 120 mM KCl, 15 mM NaCl, 4% glycerol. Then, the protein concentrations were adjusted to 10 µM. Cuvettes with pathlengths of 1 or 10 mm were used for far-UV-CD or near-UV-CD, respectively. CD spectra were recorded using a Jasco J810 spectropolarimeter.
Luciferase Aggregation and Reactivation Assay—8 µM firefly luciferase (Roche Applied Science) was denatured in 5 M guanidinium-HCl in 30 mM Tris acetate, pH 7.5, for at least 2 h at room temperature. Aggregation and reactivation was initiated by a 1:100 dilution into 40 mM MOPS, pH 7.5, 50 mM KCl, 2 mM MgATP at 30 °C. The buffers were either not supplemented with any chaperones or supplemented with the indicated concentrations of DnaK, DnaJ, and GrpE. The extent of luciferase aggregation was determined 600 s after the dilution using light scattering measurements (Hitachi Fluorimeter F4500, 
ex/em = 320 nm, slit widths of 5.0 nm). To determine luciferase reactivation, aliquots were withdrawn and diluted into assay buffer (100 mM potassium phosphate pH 7.8, 25 mM glycylglycine, 70 µM luciferin, 0.2 mM EDTA, 0.5 mg/ml bovine serum albumin). Luciferase activity was analyzed by monitoring chemiluminescence 1 min after the start of the reaction using a Hitachi Fluorometer F4500 (900 V, em = 557 nm, 20-nm slit). The chemiluminescence was followed for 45 s at room temperature, and the initial activity was obtained by linear extrapolation (25).
Determination of the DnaK ATPase Activity—The influence of DnaJ on the ATP hydrolysis of DnaK was determined using single turnover experiments (26). Briefly, 10 µM DnaK was incubated with 40 µM ATP, 0.1–0.2 µCi/ml [
-32P]ATP in 40 mM MOPS pH 7.5, 50 mM KCl, 10 mM MgCl2 for 2 min on ice to allow the DnaK-[-32P]ATP complex to form. Free ATP was separated from the complex using a micro Bio-Spin chromatography column (BioRad). DnaK-[-32P]ATP aliquots were kept frozen until further use. DnaK concentration was determined using a Bradford assay. To determine the rate of ATP hydrolysis, 30-µl reactions containing 0.16 µM wild-type DnaJ or mutant proteins in 40 mM MOPS, pH 7.5, 50 mM KCl were prepared and incubated at 25 °C. Immediately after addition of chemically denatured luciferase (final conc. 0.08 µM), the reaction was started with the addition of 0.8 µM DnaK-[-32P]ATP complex. At the indicated time points, 2 µl of the reaction were withdrawn and spotted directly onto a PEI-cellulose TLC-plate (Merck). The plates were developed in 2 M formic acid/0.5 M LiCl. The amount of ADP formed was determined using a phosphorimager.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
ZnI) or of zinc center II (DnaJZnII) replaced by serine residues. To test the in vivo function of the two mutant proteins, plasmids expressing either the mutated dnaJ genes or wild-type dnaJ were transformed into the dnaJ deletion strain PK11. dnaJ deletion strains such as PK11 are unable to propagate -growth (27), do not grow at heat shock temperatures (
43 °C) (28), and are unable to swarm (29). These three phenotypes appear to depend on slightly different functions of the DnaJ protein. We, therefore, decided that analyzing the extent to which the two mutant proteins are able to rescue those diverse phenotypes should provide a valuable first tool to functionally evaluate our mutant proteins. To exclude the possibility that any observed phenotype is due to the under- or overexpression of the mutant proteins, quantitative Western blot analysis was performed of PK11 strains expressing the two mutant proteins. We found that both mutant proteins migrated slightly different on SDS-PAGE than wild-type DnaJ, but were expressed to levels that were very similar to DnaJ expressed from its own promoter off the chromosome at 37 °C (data not shown).
In vivo analysis of the two mutant proteins revealed dramatic functional differences. PK11 strains expressing the zinc center II mutant protein failed to grow at 43 °C and formed smaller colonies at 37 °C than PK11 strains expressing wild-type DnaJ (Fig. 2, A and B). Mutants lacking zinc center I, on the other hand, showed no significant growth defect at 37 °C and were able to form small colonies at 43 °C (Fig. 2, A and B). Moreover, zinc center II mutants were non-motile at 37 °C, whereas zinc center I mutants were fully motile (Fig. 2C). These functional differences between zinc center I and II mutants were not due to solubility and/or stability problems of the mutant DnaJ proteins. This became evident when -growth was tested in these strains. -Propagation only requires the presence of a functional N-terminal domain of DnaJ (30) and was found to be nearly indistinguishable in both mutant and wild-type strains (Table I).
|
|
These in vivo results provided the first evidence that the two highly conserved zinc centers might not form a single functional entity in DnaJ but exert rather distinct functional properties. The absence of zinc center I showed only a slight influence on the DnaJ function at heat shock temperatures and did not appear to affect growth or flagellum synthesis at 37 °C. Lack of zinc center II, on the other hand, dramatically influenced the function of DnaJ under both heat shock and nonstress temperature conditions.
Zinc Center I and II of DnaJ Form Independently—In order to analyze the in vitro function of the two zinc center mutant proteins, we investigated the following properties: (i) zinc content, (ii) protein conformation, and (iii) in vitro chaperone activity of the mutant proteins alone and in concert with the DnaK/GrpE machinery. The DnaJZnI and DnaJZnII mutant proteins were overexpressed in E. coli and purified to homogeneity. The purification of DnaJZnI and DnaJZnII mutant protein was performed according to the purification protocol for wild-type DnaJ (see "Experimental Procedures"). The behavior of the two mutant proteins and DnaJ wild type was indistinguishable on the chromatography columns, suggesting that the deletion of zinc center I or zinc center II did not dramatically alter the physical-chemical properties of DnaJ.
To determine whether the zinc centers can form independently of each other, the zinc content of the two mutant proteins and wild-type DnaJ was analyzed using the PAR/PMPS titration assay (23, 24). In this assay, the total amount of zinc, which is specifically coordinated via thiol groups in a protein, can be determined. The assay is based on the presence of the chelator PAR, which quickly associates with free zinc and forms a bright red Zn(PAR)2 complex, whose absorption can be determined at 500 nm. When the zinc binding constant of the thiol-containing zinc center exceeds that of the Zn(PAR)2 complex (Ka = 2 x 1012 M–2 in 40 mM HEPES-KOH, pH 7.0); however, addition of PAR is not sufficient to extract zinc from proteins (23). In this case, thiol-modifying agents such as PMPS have to be used, which form stoichiometric thiol-mercaptide bonds and cause the release of zinc. We found that both mutant proteins coordinated close to one molecule of zinc per molecule of protein (0.8 + 0.1) via the four remaining cysteines of each zinc center (Table I). The strict requirement for PMPS to release the zinc from the remaining zinc center of the mutant proteins (1 µM) into a solution containing 100 µM PAR, indicated that the zinc binding constant Ka for the individual zinc centers must be still higher than that of PAR under these conditions. These results revealed that each of the two mutant proteins harbor one intact zinc center and suggested that DnaJ mutants, which lack zinc coordination in one zinc center, still retain sufficient protein structure to bind and coordinate zinc in the second zinc center. We, therefore, concluded that the formation of the two individual zinc centers in DnaJ can occur independently.
To determine to what extent the lack of zinc center I or II causes conformational changes in the mutant proteins that might explain the inability of DnaJZnII to rescue the observed phenotypes, far and near UV-CD spectra were recorded. These spectra showed that neither the secondary structure (Fig. 3) nor the tertiary structure (Fig. 3, inset) of DnaJ was significantly disturbed by the lack of the individual zinc centers, suggesting that the individual zinc centers do not play a major role in the overall secondary or tertiary structure of DnaJ. This agreed well with the NMR structure of the isolated zinc binding domain, which showed that this domain forms an autonomous folding domain (21) (Fig. 1).
|
Zinc Center II, Required for Optimal Refolding Activity of the DnaJ/DnaK/GrpE System—The DnaK/DnaJ/GrpE chaperone machinery supports the refolding of denatured proteins both in vivo and in vitro (for recent review see Ref. 31). In vitro analysis of the reactivation of chemically or thermally denatured luciferase in the presence of DnaK/DnaJ/GrpE is a widely used way to measure the activity of the individual protein components (11). This assay has been used, for instance, to show that the presence of full-length DnaJ is required to result in high yields of reactivated luciferase, while DnaJ variants that were missing both the cysteine-rich zinc binding domain and the C-terminal domain showed no refolding activity (19). DnaJ mutants that were lacking zinc center II because of mutations in Cys-183 and Cys-186 showed 10-fold reduced activity, while mutants that were lacking zinc center II because of mutations in Cys-161 and Cys-164 had no activity (19). This suggested that zinc center II is important and may even be essential for the DnaK-dependent chaperone activity of class I DnaJ homologues. There was also evidence suggesting that zinc center I might be involved in this activity, because mutants in the yeast homologue Ydj1p that were lacking one of the highly conserved zinc center I cysteines, also revealed a significantly reduced ability to cooperate with yeast Hsp70 in the refolding of luciferase (8).
To clearly define the role of the DnaJ individual zinc centers in cooperation with DnaK and GrpE, we utilized this well established in vitro system to test our mutant proteins. Refolding of luciferase in the absence of any chaperones resulted in less than 10% reactivated luciferase molecules (Fig. 4A, squares). In the presence of a 2:10:10 molar ratio of wild-type DnaJ:DnaK:GrpE over luciferase, however, more than 80% of chemically denatured firefly luciferase reactivated within the timeframe of the experiment (Fig. 4A, closed circle). These results agreed well with the published data (15).
|
The DnaJZnI mutant protein, which showed slightly reduced activity under heat shock conditions in vivo, displayed near wild-type activity in vitro. Presence of a 2:10:10 ratio of DnaJZnI mutant protein to DnaK and GrpE supported luciferase refolding to nearly the same levels as wild-type DnaJ (Fig. 4A, open circle), indicating that the absence of the highly conserved zinc center I does not influence the co-chaperone activity of DnaJ. The presence of the DnaJZnII mutant protein, on the other hand, resulted in reactivation yields that were only slightly higher than the reactivation yields obtained in the absence of any chaperones (Fig. 4A, triangle). This finding agreed with previous reports (19) and suggested that the observed lack of in vivo function of the DnaJZnII mutant protein might be due to a loss of its chaperone activity.
To further analyze the precise extent by which zinc center I and II mutations influence the co-chaperone activity of DnaJ, luciferase refolding assays were performed in the presence of varying concentrations of DnaJ and constant concentrations of DnaK and GrpE. Wild-type DnaJ supported luciferase reactivation to the highest extent when present in a 2:10:10:1 molar ratio of DnaJ:DnaK:GrpE:luciferase (Fig. 4B, closed circles) (12, 19). DnaJ concentrations higher than this optimum decreased the refolding yield of luciferase, presumably because the autonomous, DnaK-independent chaperone activity of DnaJ competes with DnaK for substrate binding (Fig. 4B, closed circles). Refolding reactions that contained lower ratios of DnaJ to DnaK:GrpE:luciferase also showed decreased refolding yields. This might be caused by denatured luciferase that fails to interact with DnaJ and forms irreversible aggregates.
The zinc center I mutant protein DnaJZnI was only slightly less active than wild-type DnaJ at lower concentrations (Fig. 4B, open circles). Interestingly, however, the luciferase reactivation yields did not decrease with increasing DnaJZnI concentrations. The yields of reactivated luciferase remained at the same high level even when a 100:10:10:1 molar ratio of DnaJZnI:DnaK:GrpE:luciferase was used. This was unexpected and actually suggested that the autonomous, DnaK-independent substrate binding affinity of DnaJ might be impaired when zinc center I is absent from the protein, because at high wild-type DnaJ concentrations, this autonomous chaperone activity has been shown to interfere with the refolding of luciferase. The co-chaperone activity of DnaJ, which is responsible for binding and delivering substrate proteins to DnaK, on the other hand, appeared not to be significantly influenced by the lack of zinc center I.
Substitution of wild-type DnaJ with the DnaJZnII mutant protein dramatically reduced the activity of the DnaK/DnaJ/GrpE folding machinery. With a 10:10:10:1 molar ratio of DnaJZnII to DnaK:GrpE:luciferase, only 60% luciferase was reactivated (Fig. 4B, triangles), while lower concentrations of DnaJZnII resulted in refolding yields that were only slightly higher than the yields obtained in the absence of any chaperones. Interestingly, in the presence of excess DnaJZnII mutant protein, on the other hand, the same reduction in luciferase refolding yields was observed as with wild-type DnaJ. A 50:1 molar ratio of either DnaJZnII mutant or wild-type protein to luciferase completely blocked the refolding of luciferase, indicating that the ability to interact with unfolded substrate proteins might actually not be affected by the deletion of zinc center II. This was in contrast to previous studies, which concluded that zinc center II mutant proteins are less capable of interacting with unfolded substrate proteins (19). Our results suggested, however, that zinc center II might not play a major role in the autonomous, DnaK-independent chaperone activity of DnaJ but might be specifically involved in the interaction with DnaK. In contrast, zinc center I appeared to be involved in the autonomous, DnaK-independent chaperone activity of DnaJ and dispensable for the DnaK-dependent chaperone activity of DnaJ. These results provided the first evidence that the chaperone activity that DnaJ displays when acting in concert with the DnaK/GrpE machinery might be distinct and can be disconnected from the chaperone activity that DnaJ displays on its own.
Zinc Center II, Involved in a Rate-limiting Step of the DnaK/DnaJ/GrpE-mediated Luciferase Refolding—In the absence of any chaperones, the refolding rate of luciferase is rather slow. In contrast, however, the presence of the optimal 2:10:10:1 molar ratio of DnaJ:DnaK:GrpE:luciferase accelerates the rate of luciferase refolding by about 4-fold (Fig. 4C, closed circles). Under these conditions, no luciferase aggregation is observed and the ATPase activity of DnaK is thought to be optimally stimulated by the DnaJ-substrate protein complex. Lower concentrations of DnaJ cause a slowing down of the refolding reaction. This could be either due to the presence of luciferase micro-aggregates, whose dissociation by the DnaK/DnaJ/GrpE chaperone machinery might become a rate-limiting step in the refolding of luciferase, or caused by a suboptimal stimulation of the ATPase activity of DnaK. Higher concentrations of DnaJ also slow down the reactivation rate presumably because of the competitive binding of DnaJ to the refolding substrate proteins.
A quite similar but right-shifted dependence of the reactivation rate on the DnaJ concentration was found when the DnaJZnI mutant protein was used instead of wild-type DnaJ (Fig. 4C, open circles). Here, the same high reactivation rates were reached, albeit in the presence of higher molar ratios of DnaJZnI mutant protein to DnaK:GrpE and luciferase. This was consistent with the slightly lower activity of the DnaJZnI mutant protein under heat shock conditions in vivo. The results also showed clearly that optimal luciferase refolding yields do not necessarily require optimal refolding rates and indicated that absence of zinc center I might be involved in some rate-limiting steps in the refolding reaction of luciferase that can be compensated by higher concentrations of the DnaJZnI mutant protein. In contrast, however, the DnaJZnII mutant protein did not accelerate the refolding reaction of luciferase at any concentration used (Fig. 4C, triangles). Even at a 10:10: 10:1 molar ratio of DnaJZnII to DnaK:GrpE:luciferase, where 60% of luciferase molecules refolded to the native state (Fig. 4B), the refolding rate was still similar to the slow spontaneous refolding rate of luciferase. This suggested that the DnaJZnII/DnaK/GrpE complex was no longer catalytically active. Zinc center II in DnaJ appears to mediate a rate-limiting step in the DnaK/DnaJ/GrpE foldase machinery, which allows DnaJ to accelerate the folding of luciferase and, which, in its absence, cannot be compensated by higher concentrations of the DnaJZnII mutant protein.
Zinc Center II, Essential for Locking in Substrate onto DnaK?—The inability of the DnaJZnII mutant protein to function as a co-chaperone in the DnaK/DnaJ/GrpE machinery might be due to a specific inability of this mutant protein to interact with and transfer substrate to DnaK, and not, as suggested by Szabo et al. (19), due to a general lack in substrate binding affinity. To specifically assess the interaction of our two mutant proteins with substrate proteins and DnaK, the influence of our DnaJ variants on the aggregation of chemically denatured luciferase was investigated in the absence and presence of DnaK. The buffer and reaction conditions that were used for these experiments were identical to the conditions employed for analyzing luciferase reactivation.
As shown in Fig. 5A, in the absence of any molecular chaperones, chemically denatured luciferase quickly aggregated after dilution into refolding buffer. This is well known and explains the very low yield of refolded luciferase that is obtained under these conditions (Fig. 4A). In the presence of a 0.5:1 molar ratio of wild-type DnaJ to luciferase, the aggregation of luciferase was suppressed by about 50% (Fig. 5A). Increasing concentrations of DnaJ over luciferase prevented the aggregation even more and a 2:1 molar ratio of DnaJ to luciferase was sufficient to almost completely suppress luciferase aggregation (Fig. 5C). This reflects the autonomous, DnaK-independent chaperone activity of DnaJ. DnaK by itself, on the other hand, does not influence the aggregation of luciferase even when present in a 10:1 molar ratio to luciferase (Fig. 5B). This is due to the negligible affinity of DnaK-ATP complexes to substrate proteins (32, 33). The presence of both DnaJ and DnaK in a 0.5:10:1 molar ratio of DnaJ:DnaK:luciferase, however, completely suppressed luciferase aggregation (compare Fig. 5A with Fig. 5B). This synergistic effect is caused by the cooperative chaperone action of DnaJ and DnaK (8). Here, the simultaneous interaction of DnaK with the substrate protein and DnaJ allows DnaJ to stimulate the ATP hydrolysis of DnaK. This is thought to cause the locking in of substrate proteins to the substrate binding site of DnaK, which in turn leads to the effective prevention of luciferase aggregation (8, 12). Additional presence of GrpE does not influence the aggregation behavior (data not shown) but leads to the efficient refolding of nearly 60% luciferase under these concentration conditions (Fig. 4B).
|
Analysis of the DnaJZnII mutant protein in this assay showed that absence of zinc center II does not significantly influence the ability of DnaJ to interact with unfolded substrate proteins (Fig. 5, A and C) but renders the mutant protein unable to productively cooperate with DnaK (Fig. 5B). In the absence of DnaK, the DnaJZnII mutant protein was nearly as efficient as wild-type DnaJ in interacting with refolding luciferase intermediates and preventing their irreversible aggregation (Fig. 5, A and C). In contrast to the experiments with wild-type DnaJ, however, the simultaneous presence of DnaJZnII and DnaK did not cause a further decrease of the light scattering signal (Fig. 5B). Zinc center II appears, therefore, not to be directly involved in the DnaJ-substrate interaction. This excluded the possibility that the failure of the DnaJZnII mutant protein to promote and accelerate the refolding reaction of luciferase (Fig. 4) is due to the inability to interact with substrate proteins.
The observed lack of synergistic interaction between the DnaJZnII mutant protein and DnaK is also not substrate specific because very similar results were obtained when chemically unfolded rhodanese was used (data not shown). As in the case with luciferase, the DnaJZnII mutant protein was fully able to bind to folding intermediates of rhodanese and to prevent their nonspecific aggregation, but additional presence of DnaK did not increase the protective action of the DnaJZnII mutant protein.
Zinc Center II, Providing a Second Necessary Interaction Site with DnaK—The observed lack of synergistic DnaJ-DnaK interaction in the absence of zinc center II could be attributed to several causes: (i) DnaK might be unable to interact with substrate proteins in the presence of the DnaJZnII mutant protein. (ii) DnaJZnII mutant protein might be incapable of effectively stimulating ATP hydrolysis of the DnaK-ATP-substrate protein complexes, or (iii) DnaJZnII mutant protein might fail to provide a necessary second DnaK interaction site, whose existence has been proposed previously and, which might be necessary for a catalytically active DnaK/DnaJ/GrpE complex (34, 35).
One assay that allows us to distinguish between those possibilities is the stimulation of the DnaK ATPase activity in the simultaneous presence of DnaJ and substrate proteins. Maximal stimulation of the DnaK ATPase activity has been shown to be achieved when the J-domain of DuaJ interacts with the N-terminal ATPase domain of DnaK while DnaK and DnaJ are binding simultaneously to the same unfolded polypeptide chain (12, 13). This situation is thought to cause the substrate to become locked into the high affinity DnaK-ADP complex (12). We, therefore, decided to utilize this assay to investigate the extent to which absence of zinc center II influences this DnaK-unfolded substrate protein-DnaJ interaction. Single turnover experiments were performed under conditions that were identical to the refolding and aggregation assays described above.
In agreement with the literature, we found that in the absence of unfolded substrate proteins, a 2:10 molar ratio of wild-type DnaJ to DnaK stimulated the DnaK ATPase activity 
50-fold (Fig. 6, inset). In the simultaneous presence of unfolded luciferase in a 2:10:1 molar ratio of DnaJ:DnaK:luciferase, a more than 600-fold stimulation of the DnaK ATPase activity was observed (Fig. 6, closed circles, Table I). When we tested our DnaJZnII mutant protein it became evident that this high stimulation of the ATPase activity of DnaK is not sufficient for optimal foldase activity of the DnaK/DnaJ/GrpE chaperone machine. The DnaJZnII mutant protein behaved very similarly to wild-type DnaJ and stimulated the ATPase activity 70-fold in the absence of substrate (Fig. 6, inset) and about 600-fold in the presence of substrate proteins (Fig. 6, triangles and Table I). These results indicated that the absence of zinc center II in DnaJ neither influences the interaction between DnaK and substrate protein nor the interaction between the N-terminal J-domain of DnaJ and the ATPase domain of DnaK. These findings led us to propose, however, that to be catalytically active, the DnaK/DnaJ/GrpE foldase machinery requires additional interactions between DnaJ and DnaK, which appear to be mediated by zinc center II. These additional interactions between DnaJ and DnaK might cause conformational changes in DnaK that are essential for converting DnaK-ADP into its high affinity binding state.
|
Zinc Center I, Important for DnaK-independent Chaperone Activity of DnaJ—DnaJZnI mutant protein showed an at least 10-fold reduced ability to prevent the aggregation of chemically denatured luciferase (Fig. 5C, open circles) and rhodanese (data not shown) as compared with wild-type DnaJ, indicating that the absence of zinc center I causes a strong reduction in the ability of DnaJ to interact with unfolded substrate proteins. Importantly, the significantly reduced ability of the DnaJZnI mutant protein to prevent the aggregation of folding intermediates such as luciferase or rhodanese did not appear to influence the DnaK-dependent co-chaperone activity of DnaJ (Fig. 5, A and B). In the presence of a 0.5:1 molar ratio of DnaJZnI mutant protein to luciferase, no substantial influence in the aggregation behavior of luciferase was observed (Fig. 5A). In contrast, the simultaneous presence of DnaJZnI and DnaK led to a very similar suppression of luciferase aggregation that was observed when wild-type DnaJ and DnaK was used (Fig. 5B). These findings indicated that absence of zinc center I did not significantly impair the interaction between DnaJ and DnaK and explained how the DnaJZnI mutant protein can be as effective as wild-type protein in the DnaK/DnaJ/GrpE-mediated refolding of chemically unfolded luciferase (Fig. 4, A and B). These results, furthermore, suggested that a rather transient interaction between DnaJ and substrate proteins must be sufficient for a successful cooperative interaction with DnaK-substrate protein complexes.
Maximal stimulation of the DnaK ATPase activity requires the simultaneous presence of DnaJ and DnaK bound to the same polypeptide (13). Therefore, more transient interactions between DnaJ and substrate protein should lower the effective concentrations of substrate-bound DnaJ, which in turn should decrease the extent of ATPase stimulation. That this was indeed the case is shown in Fig. 6 and summarized in Table I. While the DnaJZnI mutant protein was nearly as active as wild-type DnaJ in stimulating the ATPase activity of DnaK in the absence of substrate proteins, a lower additional stimulation of the ATPase activity of DnaK was observed when unfolded luciferase was present (Fig. 6, open circles). This might explain why higher concentrations of the DnaJZnI mutant protein are required during the luciferase refolding reactions to achieve maximal luciferase refolding rates (Fig. 4C).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
We constructed the two zinc center mutants by individually replacing the four cysteines of each zinc center with serine residues. Both DnaJZnI and DnaJZnII mutant proteins were found to be stable proteins in vivo and circular dichroism measurements of the purified proteins did not reveal any major structural changes when compared with wild-type DnaJ. This agreed well with the modular character of DnaJ proteins and suggested that the 79 amino acid cysteine-rich domain is also an autonomous folding unit in the full-length protein. The results were also in good agreement with studies using metal-free full-length DnaJ, which demonstrated that even in the absence of any zinc coordination, DnaJ undergoes probably only local structural changes (36). These local changes most likely include the partial unfolding of the zinc binding region, because in the isolated cysteine-rich domain, absence of zinc has been shown to cause the unfolding of the 
-hairpin structure (21). When zinc was titrated back into the metal-free cysteine-rich domain, zinc center II formed first and independently of zinc center I, while zinc center I only formed when the Zn2+ to protein ratio was increased above 1. This led the authors to conclude that folding of the two symmetric zinc sites is zinc-dependent and that one site forms before the other (21). Our studies now showed that both mutants of DnaJ are able to correctly assemble the one remaining zinc center. This suggests that the unfolding of one zinc center might not affect the second zinc site. The differences between our results and those obtained previously could be based on the DnaJ constructs used. While we performed our analysis with full-length protein, Martinez-Yamout et al. (21) used the isolated zinc binding domain. The two zinc centers I and II, which fold into symmetric zinc binding modules, form the "wings" of a V-shaped structure (Fig. 1). Two antiparallel -strands form the outside of each zinc center. In zinc center II, the -strands are connected by a -hairpin loop, while in zinc center I the N-and C-terminal strands need to approach each other to form the -strands (Fig. 1B). It is, therefore, conceivable that in the isolated cysteinerich domain, the absence of zinc coordination in zinc center I causes the separation of the N and C termini and, therefore, the spatial separation of the two cysteine motifs of zinc center I. This would make re-coordination of the metal in zinc center I significantly more difficult than in zinc center II, where the four cysteines are relatively close in the primary sequence.
Zinc Center I, High Affinity Binding Site for Unfolded Substrate Proteins—Functional analysis of our mutant proteins revealed that zinc center I is involved in the high affinity binding of unfolded substrate proteins. This activity, which allows DnaJ to bind to non-native proteins and prevent their nonspecific aggregation processes in vitro, has been characterized as the autonomous DnaK-independent chaperone activity of DnaJ (10, 11). Zinc center I, which is built by two interlinked turns, is connected with zinc center II via a long loop that forms the central base of the "V-apex" (Fig. 1B) (21). This central part of the cysteine-rich domain harbors a groove of conserved and hydrophobic residues, which has been suggested to be a potential substrate binding site of DnaJ (Fig. 1C) (21). It is conceivable that unfolding of zinc center I extends into this adjacent loop region. This could change the conformation of this putative substrate binding site and therefore, reduce its ability to interact with unfolded substrate proteins.
Until now, the autonomous DnaK-independent chaperone function of DnaJ has always been linked to the DnaK-dependent chaperone activity and was thought to be a necessary functional requirement for DnaJ to present unfolded substrate proteins to DnaK. We have now discovered that this high affinity substrate binding activity of DnaJ appears to be largely dispensable for the synergistic interaction of DnaJ with DnaK. Zinc center I mutant proteins, which show at least a 10-fold reduction in their ability to bind unfolded substrate proteins, are not significantly impaired in their cooperative interaction with DnaK to suppress protein aggregation or in their ability to refold unfolded proteins when acting as part of the DnaK/DnaJ/GrpE chaperone machine. This suggested that a much more transient interaction between DnaJ and unfolded substrate proteins must be sufficient for the cooperative interaction with DnaK-substrate complexes. These findings agree with functional studies of class II DnaJ homologues such as Sis1p, which appear to lack the autonomous chaperone activity, but are able to interact with Hsp70 to suppress protein aggregation and to refold non-native proteins (9).
Where is the low affinity substrate binding site located in DnaJ, which appears to be still present in mutant proteins lacking zinc center I and, which seems to be sufficient to bind and present unfolded substrate proteins to DnaK? Earlier studies showed that DnaJ deletion mutants that lack the complete zinc binding region are still capable of interacting with substrate proteins such as 
32 (18). This suggested that the low affinity binding site might be located in the C terminus of DnaJ. This agreed with in vivo and in vitro studies that have shown that residues in the first 68 amino acids of the C terminus of class I DnaJ homologues are involved in the DnaK-dependent chaperone activity of DnaJ (37, 38). Absence of the highly conserved C-terminal G242DLYV246 motif has been found to dramatically influence the synergistic interaction with DnaK, but shows little effect on the ability to suppress the aggregation of unfolded proteins such as luciferase (38). This is presumably because the autonomous, DnaK-independent high affinity binding site of zinc center I is still functional in these mutants.
Zinc Center II, a New Interaction Site with DnaK—Previous studies that were conducted to identify the role of zinc coordination in the function of DnaJ led to the conclusion that zinc center II plays a significant role in the DnaJ/DnaK/GrpE-mediated refolding of unfolded proteins (19). The authors explained their observed effects with the significantly reduced ability of these mutant proteins to bind to unfolded substrate proteins and to prevent their aggregation (19). These findings did not agree with our mutant protein studies that showed that lack of the complete zinc center II did not significantly change the ability of DnaJ to interact with unfolded proteins such as luciferase or rhodanese. We do not know the reason for the observed discrepancy but by performing all of our aggregation, refolding, and ATPase studies in the same buffer supplemented with identical amounts of ATP and co-chaperones, we were able to exclude the possibility that even small differences in the conformational stabilities of our mutant proteins might influence the outcome of our experiments.
We found that absence of zinc center II specifically affects the ability of DnaJ to convert DnaK into its high affinity binding state. This is not due to the inability of DnaJ to interact with unfolded proteins because the DnaJZnII mutant protein exerts a similar high affinity for unfolded proteins compared with wild-type DnaJ. We could also exclude that this is due to unusually stable complexes between the DnaJZnII mutant protein and unfolded substrate proteins that would prevent substrate transfer between DnaJ and DnaK. The stability of protein-protein complexes is known to increase with increasing protein concentrations. Therefore, luciferase refolding would have been expected to decrease with increasing DnaJZnII mutant protein concentrations. Quite in contrast, however, we found that luciferase refolding peaked in the presence of a 10-fold molar excess of the DnaJZnII mutant protein.
We showed that the DnaJZnII mutant protein behaved identically to wild-type DnaJ and stimulated the ATPase activity of DnaK maximally when present together with unfolded substrate proteins. Since it had been demonstrated that only the simultaneous binding of DnaJ and DnaK to the same polypeptide chain facilitates the high stimulatory effect of DnaJ on the DnaK ATPase (13), we must, therefore, assume that our DnaJZnII mutant protein is able to interact with substrate proteins and DnaK like wild-type DnaJ. These results provided excellent evidence that the maximal stimulation of the ATPase activity of DnaK is not, as previously suggested, sufficient for the locking-in of substrate proteins into the substrate binding site of DnaK. It appears, however, that additional interactions between DnaK and DnaJ are required to promote the conversion of DnaK into its high affinity binding state, and that these additional interactions are mediated by a functional zinc center II. In the presence of these additional interactions, the DnaK/DnaJ/GrpE foldase machine becomes catalytically active and accelerates the refolding of luciferase. These findings are in good agreement with studies conducted by Banecki et al. (18), who showed that a DnaJ truncation mutant, which lacks the complete zinc binding region of DnaJ, is able to bind native substrate proteins, and stimulates the ATPase activity of DnaK. This mutant protein was, however, like our zinc center II mutant unable to efficiently convert DnaK conformation into its high affinity binding state.
Based on earlier models and our studies, we propose a model in which the conversion of the DnaK-substrate protein complex into a high affinity binding state requires at least two distinct steps: ATP hydrolysis in DnaK and additional interactions between DnaK and the zinc center II of DnaJ. These interactions might cause DnaJ to release its substrate proteins to DnaK, and allow DnaK to lock in the substrate protein (Fig. 7). The requirement for such additional interactions between DnaJ and DnaK-substrate protein complexes have been suggested before in DnaK suppressor studies that showed that DnaJ binds to at least two sites in DnaK (35). The first interaction, which has been recognized for many years, seems to occur between the N-terminal J-domain of DnaJ and the N-terminal ATPase domain of DnaK. The second interaction site was proposed to be between an unidentified site in DnaJ and the C-terminal substrate binding site in DnaK (35). It appears that we have succeeded in identifying that this required additional interaction site in class I DnaJ homologues is mediated by the highly conserved zinc center II.
|
|
FOOTNOTES |
|---|
To whom correspondence should be addressed. Tel.: 734-615-1286; Fax: 734-647-0884; E-mail: ujakob{at}umich.edu.
1 The abbreviations used are: DTT, dithiothreitol; PAR, 4-(2-pyridylazo) resorcinol; PMPS, p-hydroxymercuriphenylsulfonic acid; MOPS, 4-morpholinepropanesulfonic acid.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
) or in the presence of 0.16 µM wild-type DnaJ (•), DnaJ
), or DnaJ
). The activity of luciferase was determined as described. The activity of native luciferase was set to 100%. B and C, chemically denatured luciferase (0.08 µM) was reactivated in the presence of 0.8 µM DnaK and 0.8 µM GrpE and various concentrations of wild-type DnaJ (•), DnaJ
