Regulation of Hsp7O Function by a Eukaryotic DnaJ Homolog*

We report that a purified cytoplasmic Hsp70 homo- log from Saccharomyces cerevisiae, Hsp7OSSA', ex-hibits a weak ATPase activity, which is stimulated by a purified eukaryotic dnaJp homolog (YDJlp). Stable complex formation between Hsp70SsA' and the permanently unfolded protein carboxymethylated a-lac- talbumin (CMLA) was assayed by native gel electrophoresis. The affinity of Hsp70ssA1 for CMLA appeared to be regulated by YDJlp. Significant reduction in both CMLA-Hsp70sSA' complex formation and the re- lease of CMLA pre-bound to Hsp70ssA1 was observed only in the presence of both YDJlp and ATP. Thus, Hsp70ssA1 and YDJlp interact functionally in the ex-ecution of Hsp70ssA1 chaperone activities in the eukar- yotic cell. Molecular Intracellular folding pg/reaction in the range for formation. in percent of control on the gel and 12sI-CMLA-Hsp70"SA1 excised I2'I y Under experimental variation in the level of '2sII-CMLA-Hsp70SSAl com- plex formation mixtures formed complex of formation observed of reaction mixtures formation complex

I To whom correspondence should be addressed.
The abbreviations used are: Hsp70, heat shock protein(s) of the 70-kDa family; BIP, Hsp70 homolog found in the lumen of the endoplasmic reticulum; Hsc73, heat shock cognate protein of 73 kDa; DTT, dithiothreitol; MOPS, 4-morpholinepropanesulfonic acid CMLA, carboxymethylated a-lactalbumin; AMP-PNP, adenosine 5'-(j3,y-imino)triphosphate. catalyzed by eukaryotic Hsp7O family members (9-15). In Saccharomyces cereuisiue different dnaJ homologs are localized to the same subcellular compartments as the different Hsp70 members; YDJl is cytosolic (9, lo), SISl partitions between the cytosol and cell nucleus ( l l ) , Sec63 is found in membranes of the endoplasmic reticulum (12,13), and SCJl is found in mitochondria (14). The YDJl gene encodes the more abundant of the two cytosolic dnaJp homologs and is required for normal cell growth (9,lO). YDJlp is farnesylated at a C-terminal CaaX box (where a is an aliphatic amino acid and X is any residue), and this modification appears to mediate partitioning of YDJlp between the cytosol and different intracellular membranes (9,16). To test for interactions between the abundant eukaryotic Hsp70 family members and dnaJ homologs, YDJlp and a cytosolic Hsp7O homolog of S. cereuisiue, Hsp70SSA', were purified. The influence of YDJlp on Hsp70 ATPase activity and polypeptide substrate binding and release were then examined. The results of such experiments are reported below.

EXPERIMENTAL PROCEDURES
Purification of YDJlp-YDJlp was overexpressed in E. coli strain BL21 (DE3) as described previously (16). Cells from a 200-ml culture were isolated, resuspended in 10 volumes of ice cold buffer A (20 mM MOPS, pH 7.5, 0.5 mM EDTA, 10 mM DTT, and 0.5 mM phenylmethylsulfonyl fluoride) and then disrupted by sonication. The lysate was cleared by centrifugation at 100,000 X g for 30 min. The SlOO was loaded directly onto a DE52 column (1.0 X 5.0 cm, Whatman) equilibrated with buffer A at 4 "C. The column was washed with 10 volumes of buffer A and bound YDJlp was eluted with a 0-300 mM NaCl gradient. Peak fractions containing YDJlp were pooled and then dialyzed against buffer B (5 mM potassium phosphate, pH 7.0, 10 mm DTT). YDJlp was next loaded onto a hydroxyapatite column (1 X 5 cm, Bio-Rad) equilibrated with buffer B at 4 "C. The column was washed with 10 volumes of buffer B, and bound YDJlp was eluted with a 5-400 mM potassium phosphate gradient. Peak fractions were pooled and then dialyzed against buffer C (10 mM Hepes, 50 mM NaC1,lO mM DTT, and 10% glycerol), concentrated, snap-frozen in liquid nitrogen, and stored at -70 'C. Protein concentrations were determined using the Bio-Rad Bradford assay kit with bovine serum albumin as the standard.

20927
This is an Open Access article under the CC BY license.
Regulation of Hsp70 Function by a Eukuryotic DnaJ Homolog cubated in reaction mixtures containing 50 mM Hepes, pH 7.4, 50 mM NaCl, 10 mM DTT, 2 mM MgC12, and ATP (as indicated, [ ( U -~~P ] ATP, 7.0 x 10' to 1.0 x 10' Cpm/pM). Reaction mixtures were set up on ice and shifted to 30 "C for the specified time. Reactions were then placed on ice, and duplicate 2 4 aliquots were assayed for ADP formation by thin layer chromatography on polyethyleneimine-cellulose plates (19). Spontaneous ADP formation was also assayed and subtracted prior to calculations for rates of ATP hydrolysis. The pH and salt conditions employed were optimized for maximal stimulation of Hsp7OsSAl ATPase activity by YDJlp.
Gel Shift Assay for 12sI-CMLA Binding to H~p7@'~'~-Binding reactions were carried out at 30 "C for 20 min in 2 0 4 reaction mixtures composed of the following: Hsp70SSA1 (2.5-3.0 pM), 50 mM Hepes, pH 7.0, 50 mM NaC1, 10 mM DTT, 0.1 mM EDTA, 0.4% bovine serum albumin, and '2sII-CMLA (0.7 p~, 4.5 X lo5 cpm/pM). The concentrations of indicated reagents were: YDJlp (0-3.4 pM), 2 mM MgC12, 1 mM ATP, 1 mM AMP-PNP. After incubation, reaction mixtures were diluted 2-fold with ice-cold 2 X reaction buffer made 20% (v/v) in glycerol and 0.01% in bromophenol blue. Diluted samples were loaded directly onto a 10-15% linear gradient native gel and run on ice at 10-20 mA. After electrophoresis, gels were immediately fixed, stained with Coomassie Brilliant Blue R-250, dried, and then used to expose x-ray film. The gel mobility shift of lz5I-CMLA migration was specific for Hsp7OSSA1 as bovine serum albumin, YDJlp, and enolase, respectively, a t 5 pg/reaction mixture had no effect on CMLA migration. The Hsp70SSA' and lzsII-CMLA concentrations used are in the linear range for complex formation. To calculate binding in percent of control bands on the gel corresponding to ' "1-CMLA and 12sI-CMLA-Hsp70"SA1 complexes were excised from the dried gel and assayed for I2'I by y counting. Under control conditions, an experimental variation in the level of '2sII-CMLA-Hsp70SSAl complex formation was observed, 10-30% of 1251-CMLA added to reaction mixtures formed a complex with Hsp70SSA1. However, no variation in the level of complex formation was observed in assays of duplicate reaction mixtures on the same gel. The level of complex formation observed has been previously documented in gel filtration assays, which monitor complex formation between other Hsp70 homologs and CMLA (5, 20).

RESULTS
Purified Hsp7O and YDJlp used in this study were obtained from two different sources. An Hsp7O fraction highly enriched in SSAlp (Hsp70SSA') was purified from S. cerevisiae strain MW141, which was genetically engineered to constitutively express only SSA1, and not the other three SSA genes, which encode cytosolic Hsp7O homologs (6). YDJlp was overexpressed and purified from E. coli. YDJlp is farnesylated in S. cerevisiae (16), but since E. coli lack protein isoprenyl transferases, purified YDJlp used in this study was not farnesylated. Both protein preparations were greater than 98% pure ( Fig. 1).
To test for interactions between Hsp70SSA' and YDJlp, the influence of YDJlp on Hsp70SSA' ATPase activity was determined. Hsp70SSA' hydrolyzed ATP at a rate of 2-5 nmol/mg/ min depending on the protein preparation. These rates are typical of other Hsp7O homologs (4,18,21). YDJlp exhibited no detectable ATPase activity (not shown). However, addition of YDJlp to reaction mixtures stimulated Hsp70SSA' ATPase activity approximately 10-fold at all time points tested ( Fig.  2A). The -fold stimulation of Hsp70SSA' ATPase activity by YDJlp was constant over a range of ATP concentrations which were above and below the K,,, of Hsp70SSA' for ATP (approximately 2.5 phi). Thus, YDJlp influences the maximal velocity of the ATPase reaction and not the affinity of Hsp70SSA' for ATP (Fig. 2B). Maximal stimulation of Hsp70SSA1 ATPase activity was observed at a YDJlp:Hsp70SSA1 molar ratio near 1.0 (Fig. 2C). Preincubation of YDJlp for 10 min at 70 "C prior to assay reduced stimulation of Hsp70SSA' ATPase activity by YDJlp 90% (not shown), indicating that the native conformation of YDJlp must be recognized in order for the two proteins to interact productively.
Conformational changes in Hsp7O due to ATP hydrolysis have been correlated with release of bound polypeptides from Hsp70 homologs (1,18,21,22). To test the influence of the ATPase stimulatory molecule, YDJlp, on polypeptide binding to Hsp70SSA1, a gel shift assay was developed to monitor stable interactions between Hsp70SSA' and polypeptide substrates.
To determine if YDJlp stimulates release of pre-bound 1251-CMLA from Hsp70SSA'* binding assays were carried out in two steps. In the first step Hsp70SSA' and '251-CMLA were incubated to allow complex formation. In the second step, the reaction mixture containing the '251-CMLA-Hsp70SSA' complex was split and incubated further. Addition of ATP to the second reaction resulted in release of 20% of the CMLA bound to Hsp70SSA'p in the first reaction (Fig. 3B, lane 1 versus 2 ) , whereas addition of YDJlp had no effect. Inclusion of both YDJlp and ATP in the second incubation resulted in dissociation of 60% of the complex (Fig. 3B, lane 1 versus 4 ) . Thus, the combination of YDJlp and ATP not only prevents substrate binding to Hsp70SSA1 (Fig. 3A) but can also can stimulate substrate release from the molecule.
CMLA is an artificial substrate of Hsp7O and not capable of folding after release from the chaperone. This prompted us to test an alternative substrate for binding to Hsp70SSA'. When a peptide that is specifically recognized by the mitochondrial import apparatus (27), F1@ 1-51, was employed as a substrate, stable complex formation with Hsp7PSA1 was were incubated for 20 min a t 30 "C. For details pertaining to the composition of reaction mixtures and assay of ADP formation see "Experimental Procedures." observed.2 As with CMLA (Fig. 3A) the combination of ATP and YDJlp was required to effect significant substrate release; indicating that YDJlp regulates the interactions between Hsp70qSA1 and at least two different protein substrates.
ATP  Hsp70SSA' (2.5 p M ) and '*'I-CMLA (0.7 pM, 4.5 X lo5 cpm/ PM) were incubated with the indicated additions a t 30 "C for 20 min. Complex formation was analyzed by electrophoresis of reaction mixtures on 10-15% linear gradient native gels (see "Experimental Procedures" for details). CMLA denotes the migration of "'I-CMLA in the absence of Hsp70. Asterisk (*) marks the migration of a radiolabeled contaminant in the "'I-CMLA preparation. The intensity of the * band was constant under all assay conditions. Hsp70-CMLA denotes migration of '"I-CMLA to a position on the gel coincident to that of Hsp70 as determined by staining gels with Coomassie Blue R-250 prior to autoradiography. 100% of control binding represents 10% of the total "'I-CMLA added to reaction mixtures (lane 3). R, YDJlp stimulates polypeptide release from Hsp7PSA'. For this experiment reactions were carried out in a two-step process. First, Hsp70 (3.0 p M ) and "'I-CMLA (0.4 pM, 4.5 X lo5 Cpm/pM) were incubated in an 80-pl reaction mixture a t 30 "C for 20 min to allow '2sI-CMLA-Hsp70"A1 complex formation. Aliquots (15 pl) of the initial reaction mixture were then removed and incubated in a second 20-pl reaction mixture, with indicated additions, for 20 min at 30 "C and then analyzed for complex formation as described above. Control assays show that the level of '*'I-CMLA -Hsp70""' complex in Fig.   3 8 (lune I ) is the same as that observed when complex formation is assayed immediately after the first incubation. Therefore changes in the levels of complex observed are resultant from a shift in the equlibrium of the binding reaction in response to addition of ATP or the combination of ATP and YDJlp. 100% of control binding represents 29% of the total "'I-CMLA added to reaction mixtures For details pertaining to the composition of reaction mixtures see "Experimental Procedures." formation by YDJlp was, however, never complete. The small amount of lZ5I-CMLA that remained bound might be due to either the formation of a nonspecific complex or due to a specific association in which the Hsp70SSA1 molecule could either not respond to YDJlp or which may require an additional component for complete release. To establish the specificity of '251-CMLA-Hsp70"A' complex formation, unlabeled competitor CMLA was included in reactions to determine if the level of CMLA bound to Hsp7OSSA' could be reduced below that measured in the presence of YDJlp and ATP (Fig. 4A,  lane 2). At a 100-fold molar excess of unlabeled CMLA, the Hsp70SSA'-CMLA complex was reduced to only 22% of control valves. This was the same level of residual binding observed unlabeled CMLA competes for binding of '*'I-CMLA to Hsp7OSSA'. Hsp70SSA1 (2.5 pM), 12'II-CMLA (0.35 pM, 4.5 X lo5 cpm/pM), and unlabeled CMLA (as indicated) were incubated at 30 "C for 20 min. Assay for binding of "'1-CMLA to Hsp70SSA1 was carried out as described in the legend to Fig. 3A. 1251-CMLA and unlabeled CMLA behave identically on native gels. CMLA at 10 pglreaction mixture was 33 pM. Other reagent concentrations were: YDJlp (2.45 pM), 2 mM MgC12, and 1 mM ATP. 100% of control binding ( l a n e 1 ) represents 12% of total label added to reactions. B, YDJlp-dependent reduction of specific "'I-CMLA binding to Hsp 70SSA'. Shown is quantitation of a binding experiment carried out as described in the legend to Fig. 3A. Reaction mixtures containing Hsp70SSA' (2.75 pM), '*'I-CMLA (0.35 pM, 4.5 X lo5 cpm/pM), 2 mM MgC12, 1 mM ATP, and YDJlp (as indicated) were incubated for 20 min at 30 "C. Specific binding was determined by subtracting binding of "'I-CMLA which could not be competed by addition of 10 pg of unlabeled CMLA to reaction mixtures from total binding observed. Total binding and nonspecific binding represented 20 and 4% of "'1-CMLA, respectively.
This indicates that the Hsp70SSA'-CMLA complex, which was resistant to YDJlp and ATP, results from nonspecific interactions.
To correlate stimulation of Hsp70SSA' ATPase activity by YDJlp (Fig. 2) with decreases in specific substrate binding to Hsp70SSA', the ratio of YDJlp to Hsp70SSA1 required to maximally reduce '251-CMLA-Hsp70SSA' complex formation was determined (Fig. 4B). When the nonspecific association of lZ5I-CMLA to Hsp70SSA1 was subtracted, YDJlp and ATP reduced specific substrate binding by 90% at a YDJlp: Hsp70SSA' molar ratio similar to that observed for maximal stimulation of ATP hydrolysis by YDJlp (Fig. 4B).

DISCUSSION
Data presented here provide the first evidence for modulation of the activity of an eukaryotic Hsp7O family member by a regulatory factor. Since the ATPase activity of some Hsp7O homologs can be stimulated by interaction with polypeptide substrates (1, 21) the possibility that YDJlp acts at an allosteric site on Hsp70SSA' to regulate its function or the polypeptide binding site needed to be resolved. Data presented here argue that YDJlp is not acting as a substrate of Hsp70SSA' but as a regulator. 1) Stoichiometry data (Figs. 1C and 4B) indicate that equamolar YDJlp concentrations are required to stimulate ATP hydrolysis maximally and also to reduce polypeptide-Hsp70SSA' complex formation. 2) YDJlpdependent reduction of Hsp70-CMLA complex formation requires ATP hydrolysis (Fig. 3A). 3) YDJlp stimulates ATPdependent release of substrates pre-bound to Hsp70SSA'. In other independent studies, we observed that YDJlp does not appear to be a general substrate for Hsp7O homologs since it does not stimulate the ATPase activity of purified dnaKp or BIP at concentrations which stimulate Hsp70SSA' ATPase activity maximally.' Furthermore, interactions between YDJlp and Hsp70SSA' do not appear as stable as those between polypeptide substrates and Hsp70SSA' since no alteration in the mobility of ''51-CMLA-Hsp70SSA1 complex is observed on native gels upon addition of YDJlp to reaction mixtures (Fig. 3A, lane 3 versus 6).
The lack of stable YDJlp binding to Hsp70SSA' raises questions about the nature of Hsp7O and dnaJ homolog interactions in vivo. Do dnaJ homologs form stable complexes with Hsp7O family members to permanently stimulate ATPase activity and cycling on and off polypeptide substrates? Alternatively, do dnaJ homologs interact transiently with Hsp70 family members to stimulate ATP hydrolysis and polypeptide release at specific subcellular locations where discharge of bound polypeptide substrates is required for entry into protein folding or protein translocation pathways? The present data support the alternative. YDJlp stimulates ATP-dependent release of substrates pre-bound to Hsp70SSA' (Fig. 3B) and is preferentially localized to membranes (9, 16) where Hsp70 is required to discharge nascent proteins for transport (6-10).
Evidence for participation of YDJlp in intracellular events that require Hsp70SSA' comes from observations that temperature-sensitive mutations in YDJlp cause defects in protein transport into mitochondria and the endoplasmic reticulum3 at the non-permissive temperature. These results support observations made here that YDJlp and Hsp70SSA' interact functionally. However, in addition to stimulating the ATPase activity of dnaKp, E. coli dnaJp can bind several protein substrates independent of dnaKp (1, 5, 24-26) and may actually target dnaKp to substrates by altering their conformation (3,24,25). Therefore, the possibility that YDJlp acts as a chaperone independent of Hsp70SSA' in protein trafficking events cannot be excluded by data presented here. However, YDJlp does not form a complex with lZ5I-CMLA or other peptide substrates that is stable enough to withstand electrophoresis on native gels (Fig. 3A, lane 2). This is not surprising, since gel filtration experiments indicate that dnaJp does not bind to linear substrates such as CMLA but does bind proteins exhibiting tertiary structure such as folding intermediates of rhodanese (5). We are currently attempting to determine if YDJlp binds to proteins competent for folding.
The stability of Hsp70SSA'-CMLA complexes observed here in the presence of ATP (Fig. 3, A and B) is noteworthy since experiments with the other Hsp7O homologs dnaKp, Hsc73, and BIP have shown that inclusion of ATP alone in reaction mixtures is sufficient to release the majority of bound substrate (5,20-22). Tight binding of polypeptides in the presence of ATP may reflect a specialization of Hsp70SSA' that allows for maintenance of unassembled or nascent proteins in an assembly or translocation competent form prior to release from the chaperone upon its interaction with YDJlp. Indeed, there is precedent for specialization of Hsp7O function. In vitro assays for uncoating of clathrin vesicles (23) and lysosomal protein degradation (28) demonstrate that there are Caplan, A. J., Cyr, D. M., and Douglas, M. G. (1992) Cell, in press. large differences in the activity of different Hsp7O homologs. In yeast, there is evidence for specialization of the different cytosolic dnaJp homologs as SISl and YDJl deletion strains exhibit different phenotypes (9-11). YDJlp is farnesylated posttranslationally, whereas SISlp is not (11,16).
Results reported here demonstrate that YDJlp stimulates release of polypeptide substrates from Hsp70SSA' through stimulation of ATP hydrolysis (Fig. 3, A and B ) . This is in contrast to a recent report in which E. coli dnaJp was found to stabilize substrate binding to dnaKp (5). Langer et al. (5) propose that dnaJp, which stimulates dnaKp ATPase activity about 2-fold (4), acts to stabilize dnaKp-polypeptide complexes by driving the conversion of ATP-dnaKp complexes to ADP-dnaKp-complexes, which have higher affinity for polypeptide substrates (20). Since grpEp was found to stimulate dissociation of dnaKp-dnaJp-polypeptide complexes (5), nucleotide exchange catalyzed by grpEp was presumed to promote polypeptide release by stimulating rounds of ATP hydrolysis (4). In preliminary comparisons of polypeptide binding to Hsp70SSA1 no increase in the level of stable complex formation was observed when ADP was substituted for ATP in binding reactions. Furthermore, addition of ATP and YDJlp to reaction mixtures containing Hsp70SSA'-polypeptide complexes formed in the presence of ADP resulted in complex dissociation similar to that observed in Fig. 3B (not shown). This result indicates that ATP hydrolysis stimulated by YDJlp, not nucleotide exchange, is limiting in the dissociation of polypeptides bound to Hsp70SSA' molecules.