|Home | About | Journals | Submit | Contact Us | Français|
Protein disaggregation in Escherichia coli is carried out by ClpB, an AAA+ (ATPases associated with various cellular activities) molecular chaperone, together with the DnaK chaperone system. Conformational changes in ClpB driven by ATP binding and hydrolysis promote substrate binding, unfolding, and translocation. Conserved pore tyrosines in both nucleotide-binding domain-1 (NBD-1) and -2 (NBD-2), which reside in flexible loops extending into the central pore of the ClpB hexamer, bind substrates. When the NBD-1 pore loop tyrosine is substituted with alanine (Y251A), ClpB can collaborate with the DnaK system in disaggregation, although activity is reduced. The N-domain has also been implicated in substrate binding, and like the NBD-1 pore loop tyrosine, it is not essential for disaggregation activity. To further probe the function and interplay of the ClpB N-domain and the NBD-1 pore loop, we made a double mutant with an N-domain deletion and a Y251A substitution. This ClpB double mutant is inactive in substrate disaggregation with the DnaK system, although each single mutant alone can function with DnaK. Our data suggest that this loss in activity is primarily due to a decrease in substrate engagement by ClpB prior to substrate unfolding and translocation and indicate an overlapping function for the N-domain and NBD-1 pore tyrosine. Furthermore, the functional overlap seen in the presence of the DnaK system is not observed in the absence of DnaK. For innate ClpB unfolding activity, the NBD-1 pore tyrosine is required, and the presence of the N-domain is insufficient to overcome the defect of the ClpB Y251A mutant.
Escherichia coli ClpB is a molecular machine that utilizes the energy of ATP hydrolysis to dissolve and reactivate aggregated proteins that accumulate in cells as a result of extreme stress conditions (1–3). ClpB is a member of the AAA+ (ATPases associated with various cellular activities) superfamily of proteins as well as a molecular chaperone in the Clp/Hsp100 family (1). To function in protein disaggregation and reactivation in the cell, ClpB requires an additional ATP-dependent chaperone, DnaK (3, 4). DnaK is a ubiquitous molecular chaperone that is involved in protein remodeling in conjunction with two cochaperones, DnaJ and GrpE (5). In vitro, ClpB with DnaK, DnaJ, and GrpE is able to disaggregate large insoluble aggregates (4, 6). Despite much work, the role of each chaperone in protein disaggregation remains unclear.
Like other Clp/Hsp100 proteins, ClpB forms a hexameric ring with a central pore (Fig. 1A) (7, 8). Each protomer of ClpB is composed of two AAA+ domains, nucleotide-binding domain-1 and 2 (NBD-1 and NBD-2), an N-terminal domain (N-domain), and a unique coiled-coil middle domain (M-domain) (1, 8). NBD-1 and -2 contain motifs characteristic of nucleotide-binding domains, including Walker A and B, sensor-1 and -2 motifs, and an arginine finger (9–11). Additionally, both nucleotide-binding domains contain a highly conserved tyrosine (9, 11, 12) that extends into the central pore of the ClpB hexamer, although they are not visible in the crystal structure due to the flexibility of the loops (Fig. 1, A–C) (8). These pore tyrosines interact directly with substrates (12, 13). When the NBD-1 pore loop tyrosine is mutated, protein disaggregation by ClpB in conjunction with the DnaK system is reduced, and when the tyrosine in the NBD-2 pore loop is mutated, protein disaggregation is abolished (12, 13).
The current models for ClpB-mediated disaggregation propose that the DnaK system interacts with the aggregate and perhaps initiates some aggregate remodeling. DnaK then recruits ClpB through interactions with the M-domain of ClpB (14, 15), and the pair of chaperones disassembles the aggregate by binding and threading polypeptides through the central pore of the ClpB hexamer (13, 16). Unfolded polypeptides are released and either refold spontaneously or with the help of additional chaperones. In the absence of the DnaK system, ClpB reactivates some soluble protein aggregates (17–19). Mixtures of ATP and ATPγS,2 a poorly hydrolysable ATP analog, elicit innate unfolding activity by ClpB. Mutations that allow ATP binding but not hydrolysis also evoke ClpB activity in the absence of the DnaK system (18, 19). These conditions allow for relatively stable substrate interactions with ClpB, an activity that requires ATP binding, and at the same time substrate unfolding, an activity that requires ATP hydrolysis (18). However, ATPγS cannot replace the DnaK system for all activities (17), indicating there are additional roles for DnaK during substrate disaggregation. One role for DnaK is to modulate ATP hydrolysis by ClpB (17, 19). Moreover, protein remodeling is enhanced when both ClpB and the DnaK systems are present (17). In this study, we examined the effects of substitutions to the NBD-1 and -2 pore tyrosines on protein remodeling in both the presence and absence of the DnaK system.
ClpB(ΔN) and ClpB-Trap(ΔN) plasmids were a generous gift from M. Zolkiewski (Kansas State University). GFP-15 (20), 15-GFP (20), GFP15-1 (20), 1–70GFP (21), GFP-X7-His6 (22), GFP-38 (19), full-length ClpB, ClpB(ΔN) and ClpB mutants (23), DnaK (24), DnaJ (24), GrpE (24), and GroEL(Trap) (25) were prepared as described. 1–24βgal-GFP contains the first 24 amino acids of β-galactosidase fused to the N terminus of GFP; this is the product from pGFPuv (Clontech) and was purified like GFP-15 (20). ClpB mutant plasmids were created by QuikChange (Stratagene) mutagenesis. FITC-casein was from Sigma. MDH (Roche Applied Science) was labeled with 3H as described (26). Protein concentrations given are for monomeric DnaK, GFP fusion proteins, and FITC-casein, dimeric DnaJ, GrpE, and MDH, hexameric ClpB, and tetradecameric GroEL(Trap).
Reaction mixtures (100 μl) contained buffer A (20 mm Tris-HCl, pH 7.5, 100 mm KCl, 5 mm DTT, 0.1 mm EDTA, and 10% glycerol (v/v)), 0.005% Triton X-100 (v/v), 0.2 mg/ml BSA, 4 mm ATP or 2 mm ATP, and 2 mm ATPγS (Roche Applied Science) as indicated, an ATP-regenerating system (20 mm creatine phosphate and 6 μg of creatine kinase), 10 mm MgCl2, 0.4 μm GFP fusion protein, 3.0 μm GroEL(Trap), and 1.0 μm total ClpB unless indicated otherwise. A mutant form of GroEL, GroEL(Trap), that binds but does not release unfolded proteins was included in the reaction mixtures to capture unfolded polypeptides released from ClpB (25). Unfolding was initiated by the addition of nucleotide and MgCl2 and monitored over time at 23 °C using an LS55 fluorescence spectrophotometer (PerkinElmer Life Sciences) with a well plate reader. Excitation and emission wavelengths were 395 and 510 nm, respectively. Unfolding rates were determined from the initial linear decrease in fluorescence intensities of the GFP fusion proteins.
Reactivation assays (100 μl) contained buffer A, 4 mm ATP, an ATP-regenerating system as above, 10 mm MgCl2, 5 μl of heat-aggregated GFP-38 (heated 15 min at 80 °C at 14 μm in buffer A, frozen on dry ice, thawed, and used immediately), 0.7 μm DnaK, 0.15 μm DnaJ, 0.08 μm GrpE, and 0.3 μm total ClpB, either ClpB(WT), ClpB mutant, or a mixture containing both in various ratios as indicated. Reactions were initiated by the addition of ATP and MgCl2, and reactivation was monitored over time at 23 °C as described above for GFP fusion protein unfolding. GFP-38 reactivation rates were determined from the initial linear increase in fluorescence intensities.
Reaction mixtures (100 μl) contained buffer A, 0.005% Triton X-100 (v/v), 50 μg/ml BSA, 10 mm MgCl2, 4 mm ATP, 0.4 μm GFP-15, and 2 μm ClpB(Trap) variant as indicated. Reactions were incubated at 23 °C for 5 min and loaded onto a prewashed (0.1% Triton X-100 (v/v)) Microcon YM100 filter, and complexed GFP-15 was separated from free GFP-15 by spinning for 12 min at 5400 rpm at 23 °C in a microcentrifuge. Fluorescence intensity of both the retained and filtered fractions was counted in an LS55 fluorescence spectrophotometer (PerkinElmer Life Sciences) with a well plate reader. Excitation and emission wavelengths were 395 and 510 nm, respectively.
Reaction mixtures (60 μl) contained buffer A, 5 mm ATP, an ATP-regenerating system as above, 0.005% Triton X-100 (v/v), 17 mm MgCl2, 25 μl of heat-aggregated [3H]MDH (prepared by heating 2.0 μm [3H]MDH in 50 mm Tris-HCl, pH 7.5, 150 mm KCl, 2 mm DTT, and 0.1 mm EDTA for 33 min at 47 °C), 0.9 μm DnaK, 0.18 μm DnaJ, 0.09 μm GrpE, and 0.25 μm total ClpB, either ClpB(WT), ClpB mutant, or a mixture containing both in various ratios as indicated. Reactions were initiated by the addition of ATP and MgCl2. Following a 60-min incubation at 30 °C, reactions were stopped with 75 mm EDTA and 75 mm (NH4)2SO4. The soluble fraction of [3H]MDH was separated by centrifugation, and radioactivity was measured.
Reaction mixtures (50 μl) contained buffer A, 0.005% Triton X-100 (v/v), 4 mm ATP, 0.1 μCi of [γ-33P]ATP (>3000 Ci/mm; GE Healthcare), 20 mm MgCl2, and 0.4 μm of either ClpB(WT) or ClpB mutants as indicated. Reactions were initiated by the addition of ATP and MgCl2, incubated 60 min at 25 °C, and analyzed as described (27).
ClpB (0.2 μm) was incubated for 5 min at 25 °C with 4 mm ATP and 10 mm MgCl2 in buffer A supplemented with 0.005% Triton X-100. FITC-casein (0.014 μm) was added and incubated an additional 10 min. Fluorescence anisotropy measurements were performed on an LS55 spectrofluorimeter (PerkinElmer Life Sciences) with excitation and emission wavelengths set at 490 and 525 nm, respectively, and with polarizers in place. Temperature was maintained at 25 °C with a water bath.
Each ClpB protomer contains two loops, one in NBD-1 and the other in NBD-2, which extend into the central pore of hexameric ClpB (Fig. 1). Each pore loop contains a highly conserved tyrosine (9, 11, 12). Previous reports investigating functional regions of ClpB have shown that mutation of the tyrosine residue in the NBD-2 pore loop abolishes protein reactivation activity by ClpB in collaboration with the DnaK system (13). In contrast, mutation of the conserved tyrosine in the pore loop of NBD-1 reduces, but does not eliminate, ClpB disaggregation activity with the DnaK system (12, 13).
To further investigate the role of the NBD-1 pore-exposed tyrosine in DnaK-dependent protein reactivation, we constructed a ClpB double mutant protein. This mutant, referred to as ClpB(ΔN-Y1), contained both a deletion of the N-domain and a Y251A substitution in the NBD-1 pore loop. We monitored reactivation of heat-aggregated nonfluorescent GFP-38, a GFP fusion protein that contained a 38-amino acid C-terminal peptide, by incubating it with wild type or mutant ClpB, ATP, and the DnaK system and measuring the increase in fluorescence over time. When we tested ClpB(ΔN-Y1), we observed that the initial linear rate of GFP-38 reactivation was 4.6 ± 3.0% of ClpB(WT) (Fig. 2A). In experiments testing each individual mutation, ClpB(ΔN) reactivated GFP-38 similarly to ClpB(WT), whereas ClpB(Y1) reactivated GFP-38 at 41 ± 1.4% the initial rate of wild type (Fig. 2A). These results indicate that combining the N-domain deletion with the Tyr substitution in pore loop-1 of ClpB reduces the DnaK-dependent protein reactivation activity of ClpB(WT) by more than 90%.
To exclude the possibility that ClpB(ΔN-Y1) was unable to disaggregate heat-inactivated GFP-38 because of impaired ATP hydrolysis, we measured rates of ATP hydrolysis by wild type and mutant ClpB proteins. The rate of ClpB(ΔN-Y1) ATPase activity was slightly stimulated compared with that of ClpB(WT) (Fig. 2B). These results show that either the N-domain or the tyrosine in the NBD-1 pore loop is required for ClpB collaboration with the DnaK system in protein remodeling. Moreover, the data suggest that the N-domain and the NBD-1 pore loop tyrosine may have an overlapping function, possibly in initial substrate interaction.
To test if the functional overlap for the NBD-1 pore tyrosine and the N-domain is in substrate engagement, we introduced ClpB(ΔN) and ClpB(Y1) mutations into the ClpBTrap mutant. ClpBTrap contains glutamine to alanine changes in each of the Walker B motifs (E279A and E678A) and binds but does not release substrates (28). We used fluorescence anisotropy to monitor the binding of a model substrate, FITC-casein, to the ClpB mutant proteins. Under the conditions tested, ClpBTrap(ΔN-Y1) bound ~78% less FITC-casein compared with ClpBTrap(WT) (Fig. 3A). The decrease in binding observed for ClpBTrap(ΔN) was ~40% and ClpBTrap(Y1) was ~65% compared with ClpBTrap(WT), consistent with previous observations using other substrates (12, 29). These data show that stable substrate binding by ClpB is facilitated by the NBD-1 pore tyrosine and the N-domain.
In another substrate binding assay, complex formation between GFP-15, a GFP fusion protein containing a C-terminal 15 amino acid peptide, and ClpBTrap proteins was monitored by ultrafiltration. We observed that GFP-15 binding by ClpBTrap(ΔN-Y1) was reduced about 70% compared with ClpBTrap(WT) (Fig. 3B). GFP-15 binding by ClpB(ΔN) was similar to ClpBTrap(WT), and binding by ClpBTrap(Y1) was reduced ~50% (Fig. 3B). These observations are consistent with previous studies showing that the N-domain is dispensable for the binding of some substrates (28–30). Thus stable substrate binding by ClpB is facilitated by the NBD-1 pore tyrosine and participation of the N-domain depends on the substrate. In contrast to ClpBTrap(Y1), ClpBTrap(Y2) bound GFP-15 similarly to ClpBTrap(WT) (Fig. 3B), suggesting that the NBD-2 pore tyrosine does not participate in initial substrate engagement prior to translocation.
In the ClpB hexamer, the pore loops are arranged in two rings of six, referred to here as pore ring-1 and pore ring-2 (Fig. 1, A–C) (3, 7). To investigate if all six tyrosines in pore ring-1 of ClpB function cooperatively in protein disaggregation, we monitored disaggregation by heterohexamers composed of ClpB(WT) subunits and defective ClpB(Y1) subunits. Mixing of wild type and mutant ClpB subunits has been previously demonstrated, and the ClpB hexamer has been shown to dissociate and reassociate with a half-life of a minute (19, 31, 32).
When ClpB(WT) and ClpB(Y1) were mixed in various ratios with the total ClpB concentration constant, we observed an initial stimulation in substrate reactivation as the percentage of ClpB(Y1) was increased, which was followed by a decrease in activity to the level of ClpB(Y1) alone (Fig. 4A). The stimulation was maximal at a ratio of 5:1 ClpB(WT) to ClpB(Y1), whereas heterohexamers at a ratio of 1:1 ClpB(WT) to ClpB(Y1) had activity similar to 100% ClpB(WT) homohexamers (Fig. 4A). Thus, incorporation of a few defective ClpB subunits stimulates disaggregation activity. One possible explanation for the stimulation is that six pore ring-1 tyrosines per hexamer cause stabilization of substrate engagement, which results in a decrease in the rate of unfolding and translocation. Interestingly, incorporation of several ClpB(ΔN) subunits into heterohexamers with ClpB(WT) also stimulates GFP-38 disaggregation activity in the presence of the DnaK system (Fig. 4B). Nagy et al. (33) previously observed a similar stimulation of disaggregation when they monitored reactivation of heat-aggregated glucose-6-phosphate dehydrogenase by mixtures of ClpB(ΔN) and ClpB(WT) in vitro and heat-inactivated luciferase in vivo. They suggested the possibility that the presence of six N-domains per ClpB hexamer may be thermodynamically unfavorable leading to slower rates of disaggregation. Additionally, Nagy et al. (33) proposed that heterohexamers of ClpB(ΔN) and ClpB(WT) might optimize ClpB binding to extractable structural elements in the aggregate. The results with heterohexamers of ClpB(Y1) and ClpB(WT) (Fig. 4A) and those with heterohexamers of ClpB(ΔN) and ClpB(WT) (33) (Fig. 4B) are consistent with our above suggestion that that the N-domain and the pore tyrosine in NBD-1 have an overlapping function in protein disaggregation, likely in initial substrate engagement.
When we monitored disaggregation of another aggregated substrate, MDH, in conjunction with the DnaK system, we observed that a 1:1 mixture of ClpB(Y1) and ClpB(WT) at a final heterohexamer concentration of 0.26 μm exhibited similar activity to ClpB(WT) homohexamers at a concentration of 0.13 μm (Fig. 4C). These results suggest that the cooperation between ClpB pore ring-1 tyrosines during disaggregation in conjunction with the DnaK system depends on the substrate.
We also examined the effect of incorporating ClpB(Y2) subunits into heterohexamers with ClpB(WT). ClpB(Y2) was unable to reactivate GFP-38 with the DnaK system, consistent with previous results using other aggregates (Fig. 4D) (13). When ClpB(WT) and ClpB(Y2) were mixed in various ratios, we observed a near linear decrease in GFP-38 reactivation as the percentage of ClpB(Y2) increased (Fig. 4D). When MDH was used as the substrate, similar results were obtained (Fig. 4C). These results suggest that the six pore ring-2 tyrosines in the ClpB hexamer do not function cooperatively during DnaK-dependent substrate disaggregation, but they likely utilize a random or probabilistic mechanism.
In the absence of the DnaK system, ClpB has innate protein remodeling activity that is elicited by using mixtures of ATP and ATPγS (18). We wanted to know if mutating the pore loop tyrosines in NBD-1 and NBD-2 of ClpB have similar effects on innate protein remodeling activity as observed for DnaK-dependent protein reactivation. We monitored protein unfolding by ClpB wild type and mutants using as substrates GFP fusion proteins containing an N- or C-terminal 15-amino acid peptide, 15-GFP and GFP-15. In contrast to protein reactivation in the presence of the DnaK system (Fig. 2A), we were unable to detect protein unfolding of either 15-GFP or GFP-15 by ClpB(Y1) (Fig. 5, A and C). These results show that the tyrosine in the NBD-1 pore loop is essential for innate protein remodeling by ClpB alone.
We also investigated protein unfolding by ClpB(ΔN) and by ClpB(ΔN-Y1) in the absence of the DnaK system. ClpB(ΔN) had innate protein unfolding activity similar to ClpB(WT), whereas ClpB(ΔN-Y1), like ClpB(Y1), was unable to unfold GFP-15 (Fig. 5, C and D). These data show that the N-domain is not essential for the innate ClpB remodeling activity. Taken together, the results indicate that although the NBD-1 pore loop tyrosine and the N-domain have an overlapping role in substrate engagement by ClpB, the action of Tyr-251 is absolutely required for innate ClpB activity.
To determine whether cooperative activity of all six tyrosines in pore ring-1 is required for protein unfolding, we monitored substrate unfolding by heterohexamers composed of ClpB(WT) subunits and defective ClpB(Y1) subunits. As the percentage of ClpB(Y1) in heterohexamers increased, the rate of 15-GFP unfolding decreased in a cooperative manner, such that when the heterohexamer contained ~50% ClpB(Y1), ~25% of the unfolding activity remained (Fig. 5B). When we repeated the experiments using GFP-15, we observed that there was modest stimulation of activity when one to four ClpB(Y1) protomers were incorporated per hexamer followed by inhibition when ~5 were incorporated (Fig. 5C). For example, when the heterohexamer contained ~50% ClpB(Y1), ~75% of the GFP-15 unfolding activity remained (Fig. 5C). Together these results show that the six tyrosines in pore ring-1 act cooperatively in protein unfolding, and the degree of cooperativity varies depending on the substrate being remodeled (Fig. 5, B and C).
We next examined the pore ring-2 ClpB mutant, ClpB(Y2), for innate remodeling activity (Fig. 1E). This mutant lacks protein disaggregation activity in combination with the DnaK system in vitro (Fig. 4D) (13). When we monitored fluorescence of 15-GFP or GFP-15 in unfolding assays with ClpB(Y2), there was no detectable decrease in fluorescence signal with either substrate (Fig. 5, E and G). Our results indicate that the tyrosine in the pore loop of NBD-2 is essential for ClpB protein remodeling activity in the absence of the DnaK system.
When we tested for cooperative activity of the pore ring-2 tyrosines using subunit mixing experiments, we observed that as the percentage of ClpB(Y2) in heterohexamers with ClpB(WT) was increased, the rate of 15-GFP unfolding decreased in a cooperative manner (Fig. 5F). When the hexamer contained ~50% ClpB(Y2), only ~9% of the activity remained. A cooperative decrease in GFP-15 unfolding was also seen as the percentage of ClpB(Y2) in the hexamer increased, although the decrease in activity was less than for 15-GFP (Fig. 5G). These results suggest that two to three subunits with a tyrosine in the NBD-2 pore loop are required for unfolding activity, depending upon the substrate. Moreover, they indicate that the tyrosines in pore ring-2 function cooperatively when ClpB acts without the DnaK system.
Taken together, our studies show that the tyrosines in both the NBD-1 and NBD-2 pore loops are essential for substrate unfolding in the absence of the DnaK system. The data also suggest that the tyrosines in pore ring-1 and those in pore ring-2 function cooperatively, and the cooperativity varies depending on the substrate. This conclusion was further substantiated by results using four additional tagged GFP substrates for unfolding by heterohexamers of ClpB(WT) and ClpB(Y1) or ClpB(WT) and ClpB(Y2) (Table 1).
We have shown that ClpB homohexamers with six tyrosines in one pore ring and six tyrosine to alanine substitutions in the other pore ring are defective in protein unfolding (Fig. 5). However, heterohexamers are partially functional with at least two to three tyrosines in one of the pore rings, when the other ring contains six tyrosines. We next wanted to determine whether ClpB heterohexamers are able to function with only six pore tyrosines if the tyrosines are distributed between the NBD-1 and the NBD-2 pore loops. To test this, we mixed ClpB(Y1) with ClpB(Y2) generating heterohexamers and used GFP-15 as substrate. The highest activity was observed at a 1 ClpB(Y1) to 1 ClpB(Y2) ratio, whereas ClpB(Y1) and ClpB(Y2) alone were inactive (Fig. 6A). However, there was barely detectable unfolding activity (~5% compared to WT) at a 1:1 ratio of the two mutants when the total ClpB concentration was 1 μm (Fig. 6A). When we increased the total ClpB concentration used in the mixtures to 10 μm, the maximal activity observed was ~25% of ClpB(WT) activity at 1 μm. These data show that when each protomer has a single pore tyrosine and there are six pore tyrosines per hexamer, ClpB can support protein unfolding but with significantly reduced activity. Moreover, these results may suggest that ClpB preferentially transfers substrate from pore ring-1 to pore ring-2 by an intraprotomer mechanism.
To address whether cooperation between pore tyrosines within the same protomer is important for ClpB in unfolding reactions, we mixed ClpB(WT) with a ClpB mutant containing a tyrosine to alanine substitution in both pore loops (Y251A and Y653A), referred to as ClpB(Y1,Y2). This mixture yielded heterohexamers with tyrosines in both pore rings, although each protomer had either a tyrosine in both pore loops or an alanine substitution in both. Using GFP-15 as a substrate for unfolding, we observed a near linear decrease in activity as the percentage of ClpB(Y1,Y2) in the heterohexamer with ClpB(WT) was increased (Fig. 6B). For example, when there were ~3 pore tyrosines in pore ring-1 and ~3 in pore ring-2, protein unfolding activity was ~50% of ClpB(WT) (Fig. 6B). These data indicate that ClpB heterohexamers promote efficient protein unfolding when there are six pore tyrosines per hexamer if each protomer contains two pore tyrosines or two alanine substitutions. Moreover, ClpB still retains unfolding activity when only ~4 pore tyrosines are present on two protomers within the hexamer. Taken together, these results suggest that the ClpB tyrosines in the pore loops of NBD-1 and NBD-2 in an individual protomer act cooperatively in protein remodeling.
We next asked whether the NBD-1 pore tyrosine and the NBD-2 pore tyrosine need to reside in a ClpB protomer that is hydrolytically active for ClpB to function in protein unfolding. To address this question, we used a previously described assay where ClpB unfolding activity is elicited by heterohexamers of ClpB(WT) and ClpBTrap in the presence of ATP, a condition under which neither protein is active alone (Fig. 7A) (19). One explanation for the unfolding activity of these heterohexamers is that some protomers bind but do not hydrolyze ATP, a state that stabilizes substrate interactions with ClpB, although other subunits are active for ATP hydrolysis, which is necessary for driving substrate unfolding and translocation (19).
When we monitored GFP-15 unfolding by mixtures of ClpBTrap and ClpB(Y1), we found that heterohexamers were as active as heterohexamers of ClpBTrap and ClpB(WT) (Fig. 7A). Thus, full activity of the heterohexamer is elicited when the subunit that is able to bind the substrate stably, ClpBTrap, has a wild type NBD-1 pore loop, and the ATP hydrolysis-competent subunit, ClpB(Y1), has a mutant NBD-1 pore loop. This implies that intersubunit substrate transfer can occur along the unfolding pathway consistent with what was observed in Fig. 6A. No significant differences in ClpB(WT) and ClpB(Y1) ATP hydrolysis rates were observed (Fig. 2B). Heterohexamers of ClpB(WT) and ClpBTrap(Y1) exhibited ~40% of the activity of the ClpBTrap and ClpB(WT) mixture at a 1:1 ratio (Fig. 7A). Thus, unfolding activity is reduced when the ClpBTrap subunit also contains a tyrosine to alanine substitution in the NBD-1 pore loop. Taken together with the substrate binding data (Fig. 3), these results indicate that the primary role for pore tyrosines in ring-1 is initial substrate engagement and suggest that substrate binding at pore ring-1 prior to translocation is rate-limiting for the process of protein unfolding, translocation, and release. These data further indicate that substrate transfer between subunits can occur along the pathway of substrate unfolding.
We next tested protein unfolding by heterohexamers of ClpB(Y2) and ClpBTrap and by heterohexamers of ClpBTrap(Y2) and ClpB(WT) (Fig. 7B). There was observable protein unfolding by both mixtures, but the maximal rate of unfolding activity was ~12% that of 1:1 mixtures of ClpBTrap and ClpB(WT) (Fig. 7, A and B). In combination with the observation that pore ring-2 tyrosines do not participate in initial substrate binding (Fig. 3B), the results from ClpB(Y2) and ClpBTrap heterohexamers indicate that the pore ring-2 tyrosines are likely involved in translocation activity. Additionally, the loss of unfolding activity of ClpBTrap(Y2) and ClpB(WT) heterohexamers (Fig. 7B) compared with ClpBTrap and ClpB(WT) heterohexamers (Fig. 7A) implies that stable substrate interaction with the pore ring-2 tyrosines is a prerequisite for translocation by NBD-2.
Our current model for protein disaggregation by ClpB and the DnaK system is based on the observations presented here and previously (Fig. 8). The DnaK chaperone system interacts with an aggregate, likely performing some loosening and partial remodeling of the aggregate. DnaK then recruits ClpB through direct interactions with the ClpB M-domain. In an engagement step that requires ATP binding, the ClpB N-domain and pore loop-1 tyrosine bind polypeptides exposed on the aggregate surface. However, the ClpB N-domain or the pore loop-1 tyrosine can each act independently in polypeptide binding depending upon the substrate. Rounds of ATP binding and hydrolysis trigger conformational changes in ClpB leading to substrate binding at the pore loop-2 tyrosine and translocation through the ClpB channel. The DnaK system likely participates in this step of the reaction by coupling ATP hydrolysis by ClpB to substrate translocation. Following translocation, the polypeptide is released in an unfolded conformation and either spontaneously refolds or is refolded with the aid of cellular chaperones. Our observed engagement step, also referred to as a commitment or priming step, is similar to an observed or proposed step for both Hsp104 and ClpA (33–36).
Here, we show that ClpB requires either the N-domain or the NBD-1 pore loop tyrosine for protein disaggregation in collaboration with the DnaK system. A ClpB mutant with an N-domain deletion and a Tyr to Ala substitution in the NBD-1 pore loop is unable to perform substrate remodeling alone or with the DnaK system (Fig. 2A). For many proteins and small aggregates, the N-domain of ClpB is not required for binding (Fig. 3B) (12, 37). However, Zolkiewski and co-workers (29, 33) observed N-domain-dependent substrate binding for glucose-6-phosphate dehydrogenase and MDH, and they showed that ClpB binding to glucose-6-phosphate dehydrogenase depends on the size of the aggregate. Similarly, a Thermus thermophilus ClpB N-terminal deletion mutant is defective in binding some but not all substrates (38). Thus, the N-domain, like the NBD-1 pore loop, appears to take part in substrate engagement prior to translocation. This functional overlap of the N-domain and the NBD-1 pore loop could make it possible for ClpB and other AAA+ proteins to regulate substrate recognition and binding by providing more regions for interaction with substrates.
The N-domain and NBD-1 pore loop may also regulate substrate interaction by limiting or controlling access to the central channel of ClpB. We present data showing a stimulation in disaggregation activity for mixtures of ClpB(WT) and ClpB(Y1) compared with wild type (Fig. 4A). Nagy et al. (33) previously observed a similar stimulation in disaggregation activity by heterohexamers of ClpB(WT) and ClpB(ΔN) compared with ClpB(WT) homohexamers (Fig. 4B). Together, these data suggest that six N-domains and six pore ring-1 tyrosines per hexamer slow ClpB remodeling activity, and the substitution of mutant subunits stimulates activity of the ClpB hexamer. The mechanisms behind these effects are not known. One possibility is that substrates are tightly bound by the six N-domains or the six pore tyrosines in NBD-1 causing unfolding and entry of the substrate into the channel to slow. Another possibility is that a full six N-domains or six NBD-1 pore loop tyrosines may physically occlude the central pore and slow substrate entry. Recently, it has been proposed that conformational flexibility of the N-domain is important for ClpB disaggregation activity (33, 39).
In vivo, E. coli ClpB appears to have a mechanism for avoiding the inhibition caused by six N-domains per hexamer. In addition to synthesizing full-length ClpB, E. coli makes a truncated ClpB protein, which lacks the N-domain, as the result of a translation initiation site within the clpB transcript. Cells expressing both forms of ClpB are more resistant to thermal killing than cells engineered to express only full-length ClpB (40, 41). Unlike ClpB, the yeast homolog Hsp104 does not have an internal start site and is therefore a hexamer composed of six full-length monomers, which may lead to differences in substrate recognition and remodeling. However, a Δhsp104 Saccharomyces cerevisiae strain that expresses Hsp104ΔN from a plasmid exhibits no defects in thermotolerance or prion propagation, although they are defective in prion curing (42, 43).
Our results additionally show that a mutant ClpB protein with a Tyr to Ala substitution in the NBD-1 pore loop is unable to catalyze protein unfolding in the absence of the DnaK system (Fig. 5, A–C), although the mutant is active in the presence of the DnaK system (Fig. 2A) (12, 13). Because the pore ring-1 tyrosines are involved in substrate engagement prior to translocation (Fig. 3) (12, 13), this observation indicates that the DnaK system plays a role in stabilizing substrate interactions with ClpB. This interpretation is consistent with previous observations and proposals suggesting that the DnaK system facilitates aggregate recognition by ClpB (13, 16, 44, 45). It is also consistent with our recent observation that DnaK interacts with helices 2 and 3 of the ClpB M-domain (14) and thus is in a position to stabilize a handoff of substrates to ClpB. Additionally, the DnaK system may have other roles in protein disaggregation. For example, we previously found that in the presence of aggregated substrate, the rate of ATP hydrolysis by ClpB and the DnaK system is ~2-fold higher than the sum of the rates of each chaperone alone with aggregated substrate. This suggests that DnaK modulates the activity of ClpB and that modulation results in the synergistic action of the pair of chaperones (17, 19). Despite a better understanding of the roles of the ClpB pore loops and the N-domain in substrate recognition, binding, unfolding, and translocation, many questions remain to be answered about the mechanism of ClpB and the interplay between ClpB and the DnaK system.
We thank Jodi Camberg for critical reading of the manuscript and many helpful discussions. We thank Michal Zolkiewski (Kansas State University) for the plasmid pET20b-ΔN and pET20b-ΔNtrap.
*This work was supported, in whole or in part, by National Institutes of Health Intramural Research Program, NCI, Center for Cancer Research.
2The abbreviations used are: