Mechanistic Insights into Hsp104 Potentiation

Abstract

Potentiated variants of Hsp104, a protein disaggregase from yeast, can dissolve protein aggregates connected to neurodegenerative diseases such as Parkinson disease and amyotrophic lateral sclerosis. However, the mechanisms underlying Hsp104 potentiation remain incompletely defined. Here, we establish that 2–3 subunits of the Hsp104 hexamer must bear an A503V potentiating mutation to elicit enhanced disaggregase activity in the absence of Hsp70. We also define the ATPase and substrate-binding modalities needed for potentiated Hsp104^A503V activity in vitro and in vivo. Hsp104^A503V disaggregase activity is strongly inhibited by the Y257A mutation that disrupts substrate binding to the nucleotide-binding domain 1 (NBD1) pore loop and is abolished by the Y662A mutation that disrupts substrate binding to the NBD2 pore loop. Intriguingly, Hsp104^A503V disaggregase activity responds to mixtures of ATP and adenosine 5′-(γ-thio)-triphosphate (a slowly hydrolyzable ATP analogue) differently from Hsp104. Indeed, an altered pattern of ATP hydrolysis and altered allosteric signaling between NBD1 and NBD2 are likely critical for potentiation. Hsp104^A503V variants bearing inactivating Walker A or Walker B mutations in both NBDs are inoperative. Unexpectedly, however, Hsp104^A503V retains potentiated activity upon introduction of sensor-1 mutations that reduce ATP hydrolysis at NBD1 (T317A) or NBD2 (N728A). Hsp104^T317A/A503V and Hsp104^A503V/N728A rescue TDP-43 (TAR DNA-binding protein 43), FUS (fused in sarcoma), and α-synuclein toxicity in yeast. Thus, Hsp104^A503V displays a more robust activity that is unperturbed by sensor-1 mutations that greatly reduce Hsp104 activity in vivo. Indeed, ATPase activity at NBD1 or NBD2 is sufficient for Hsp104 potentiation. Our findings will empower design of ameliorated therapeutic disaggregases for various neurodegenerative diseases.

Introduction

Protein misfolding and aggregation are associated with a wide variety of diseases, ranging from type II diabetes (1, 2) to neurodegenerative diseases, such as fatal familial insomnia (3, 4), Parkinson disease, and amyotrophic lateral sclerosis (ALS) (5–7). In Parkinson disease patients, α-synuclein (α-syn)⁶ forms toxic soluble oligomers as well as amyloid structures that accumulate in Lewy bodies and contribute to the death of dopaminergic neurons (8–12). Similarly, toxic soluble oligomers and cytoplasmic inclusions of TDP-43 or FUS are associated with ALS and frontotemporal dementia (13–20). These misfolded protein conformers are recalcitrant and represent a colossal roadblock in the treatment of these diseases.

Hsp104 is a 102-kDa AAA+ ATPase (21) from Saccharomyces cerevisiae capable of dissolving disordered protein aggregates as well as dismantling amyloid fibrils and toxic soluble oligomers (22–33). It assembles into a homohexameric barrel structure with a central channel (34–39). Hsp104 processes protein aggregates by directly translocating substrates either partially or completely through this channel (35, 36, 40–46). Hsp104 encompasses an N-terminal domain, two nucleotide-binding domains (NBD1 and NBD2), a coiled-coil middle domain (MD), and a C-terminal domain important for oligomerization (Fig. 1A) (47). Both NBDs contain Walker A and Walker B motifs that are critical for nucleotide binding and hydrolysis, respectively (48). ATP hydrolysis takes place primarily at NBD1, whereas NBD2 has a nucleotide-dependent oligomerization function (29, 34, 49–52).

FIGURE 1.

Hsp104 hexamers must contain 2–3 A503V subunits for enhanced disaggregase activity in the absence of Hsp70. A, domain structure of Hsp104. The N-terminal domain (purple), NDB1 (olive green), MD (orange), NBD2 (blue), and the C-terminal domain (red) are shown. Hsp104 mutations included in this study and their functional effects are also shown. Mutations affecting substrate binding are shown in mustard, whereas mutations affecting ATP binding and hydrolysis are shown in light blue. The potentiating mutation A503V is shown in pink. B, theoretical Hsp104 heterohexamer ensembles containing 0–6 WT subunits as a function of the fraction of WT subunit present. C, Hsp104 was mixed in varying ratios with Hsp104^A503V to create heterohexamer ensembles. Luciferase disaggregation and reactivation activity was assessed in the absence of Hsp70 and Hsp40. Values represent means ± S.E. (error bars) (n = 3). Theoretical curves are shown to illustrate the predicted result when 1–6 WT subunits are needed to inactivate the Hsp104^A503V hexamer. D, the data plotted in C are replotted. Theoretical curves are shown where adjacent pairs of A503V:A503V or A503V:WT subunits confer hexamer activity, whereas adjacent WT subunits are inactive. Each adjacent A503V:A503V pair has an activity of 1/6. Adjacent A503V:WT pairs have a repressed activity (r), and the effects of various r values are depicted.

Remarkably, Hsp104 can remodel amyloid substrates alone, without the aid of any other chaperones (22, 24, 26, 31, 33, 45, 53–56). However, to disaggregate amorphous protein aggregates, Hsp104 usually needs to collaborate with the Hsp110, Hsp70, and Hsp40 chaperone system (23, 26, 30, 32, 57). Moreover, small heat shock proteins such as Hsp26 can enhance disaggregase activity further (30, 57–59). In vitro, mixtures of ATP and ATPγS (a slowly hydrolyzable ATP analogue) enable Hsp104 to dissolve amorphous aggregates in the absence of Hsp70 and Hsp40 (26, 35, 60, 61).

Wild-type (WT) Hsp104 can resolve α-syn oligomers and fibrils, but very high Hsp104 concentrations are required (26, 31, 32). Hsp104 has limited disaggregase activity against TDP-43 and FUS fibrils (62, 63). Recently, we have engineered potentiated Hsp104 variants that mitigate TDP-43, FUS, and α-syn misfolding (62–67). Missense mutations at disparate but specific positions in the MD or the small domain of NBD1 (immediately C-terminal to the MD) resulted in potentiated Hsp104 variants (62, 67). Potentiating mutations in the MD obviate any absolute requirement for Hsp70 in disaggregation of amorphous aggregates and typically (under physiological salt conditions) enhance Hsp104 ATPase activity (62, 67). Potentiated Hsp104 variants also display accelerated substrate translocation, enhanced unfoldase activity, and enhanced amyloid-remodeling activity (45, 62). They can also recognize shorter unfolded tracts in client proteins compared with Hsp104 (63). Hsp104^A503V has been previously established as a potentiated Hsp104 variant able to ameliorate the toxicity arising from the aggregation of WT or disease-linked forms of α-syn, FUS, and TDP-43 in yeast (62, 63). Hsp104^A503V hexamers also display enhanced plasticity and are more resistant to defective subunits than Hsp104 (62). However, Hsp104^A503V is more sensitive than Hsp104 to suramin, a small molecule inhibitor of Hsp104 ATPase activity (68). Interestingly, two potentiated Hsp104 variants, Hsp104^A503S and Hsp104^{Y257F/A503V/Y662F}, rescue neurodegeneration in the metazoan nervous system and hold promise as possible treatments for neurodegenerative disease (62, 66).

Despite these advances, the molecular mechanisms underlying the potentiation of Hsp104 are incompletely understood. The N-terminal domain of Hsp104 is critical for Hsp104 potentiation, as is motif 1 (helix 1 and a portion of helix 2) of the MD (35, 47, 67). Thus, deletion of these large regions precludes potentiation (35, 67). However, beyond these domain requirements, little else is known. Here, we explore how many A503V subunits are needed per hexamer to enable enhanced activity. We also determine which Hsp104 ATPase and substrate-binding modalities are important for potentiation both in vitro and in vivo. The mechanistic insights gleaned from our studies will enable further development of potentiated Hsp104 variants as therapeutics for various neurodegenerative diseases.

Experimental Procedures

Materials

All chemicals were purchased from Sigma-Aldrich unless otherwise specified. Creatine kinase was purchased from Roche Applied Science. Firefly luciferase was purchased from Sigma-Aldrich. Luciferase assay reagent was purchased from Promega (Madison, WI). Hsp70 (Hsp72) and Hsp40 Hdj1 were purchased from Enzo Life Sciences (Farmingdale, NY).

Protein Expression and Purification

Sixteen Hsp104 variants: Hsp104, Hsp104^Y257A, Hsp104^T317A, Hsp104^A503V, Hsp104^Y662A, Hsp104^N728A, Hsp104^K218T/K620T, Hsp104^E285Q/E687Q, Hsp104^Y257A/Y662A, Hsp104^Y257A/A503V, Hsp104^T317A/A503V, Hsp104^Y662A/A503V, Hsp104^N728A/A503V, Hsp104^{K218T/A503V/K620T}, Hsp104^{E285Q/A503V/E687Q}, and Hsp104^{Y257A/A503V/Y662A} were purified as reported previously (52, 61, 69). Briefly, untagged Hsp104 was transformed into BL21-DE3 RIL cells (Agilent Technologies, Santa Clara, CA). Expression was induced at an A₆₀₀ of 0.4–0.6 with 1 mm isopropyl 1-thio-β-d-galactopyranoside for 15–18 h at 15 °C. Cells were harvested via centrifugation (4,000 rpm, 4 °C, 20 min), resuspended in lysis buffer (50 mm Tris-HCl, pH 8.0, 10 mm MgCl₂, 2.5% glycerol (w/v), 2 mm β-mercaptoethanol (Bio-Rad), 5 μm pepstatin A, and 1 Mini-complete EDTA free protease tablet per 50 ml (Roche Applied Science), and lysed by sonication. Cell debris was removed via centrifugation at 16,000 rpm at 4 °C for 20 min. The supernatant was applied to Affi-Gel Blue resin (Bio-Rad). Supernatant and resin were rotated at 4 °C for 4 h. Resin was then washed with wash buffer (50 mm Tris-HCl, pH 8.0, 10 mm MgCl₂, 100 mm KCl, 2.5% glycerol (w/v), 2 mm β-mercaptoethanol). Hsp104 was eluted with high salt buffer (wash buffer plus 1 m KCl). Hsp104 was then further purified by ResourceQ anion exchange chromatography using running buffer Q (20 mm Tris-HCl, pH 8.0, 0.5 mm EDTA, 5 mm MgCl₂, 50 mm NaCl) and eluted with a linear gradient of buffer Q+ (20 mm Tris-HCl, pH 8.0, 0.5 mm EDTA, 5 mm MgCl₂, 1 m NaCl). Eluate was exchanged into storage buffer (40 mm HEPES-KOH, pH 7.4, 150 mm KCl, 20 mm MgCl₂, 10% glycerol, 1 mm DTT), snap-frozen, and stored at −80 °C. High salt storage buffer (40 mm HEPES-KOH, pH 7.4, 500 mm KCl, 20 mm MgCl₂, 10% glycerol, 1 mm DTT) was used for storage of Hsp104^A503V and all other Hsp104 variants containing this mutation.

ATPase Assay

WT or mutant Hsp104 (0.25 μm monomer) in ATPase buffer (20 mm HEPES-KOH, pH 7.4, 20 mm NaCl, and 10 mm MgCl₂) was incubated for 10 min at 25 °C in the presence of ATP (1 mm) as noted (52). ATPase activity was assessed by the release of inorganic phosphate determined by using a malachite green phosphate detection kit (Innova Biosciences, Cambridge, UK).

Luciferase Disaggregation Assays

Luciferase reactivation was performed as described (23, 61). To assemble aggregates, firefly luciferase (50 μm) in luciferase refolding buffer (25 mm HEPES-KOH, pH 7.4, 150 mm potassium acetate, 10 mm magnesium acetate, 10 mm DTT) with 6 m urea was incubated at 30 °C for 20 min. Luciferase was then rapidly diluted 100-fold into luciferase refolding buffer, divided into 100-μl aliquots, snap-frozen in liquid N₂, and stored at −80 °C. For reactivation assays, aggregated luciferase (50 nm) was incubated with Hsp104 (1 μm hexamer) plus 5 mm ATP (or the indicated ATPγS and ATP ratio amounting to the same total) and an ATP regeneration system (10 mm creatine phosphate, 0.5 μm creatine kinase (Roche Applied Science), 0.1 mm ATP) for 90 min at 25 °C. For some assays, Hsp70 (1 μm) and Hsp40 (1 μm; Enzo Life Sciences) were added. Luciferase activity was assessed by luminescence measured on a Safire² microplate reader (Tecan, Männedorf, Switzerland).

Subunit Doping Assay

Hsp104 was mixed with Hsp104^A503V in varying ratios to give a total concentration of 0.167 μm Hsp104 hexamer, and the luciferase reactivation experiments were performed as described above. Hsp70 and Hsp40 were omitted for these experiments such that WT Hsp104 was inactive. Thus, under these conditions, we are certain that if the number of WT subunits per hexamer exceeds five, then the hexamer is inactive. We employed the approach of Reinstein and colleagues (70) to simulate the distribution of Hsp104 and Hsp104^A503V subunits within a given population of Hsp104 hexamers as described (26, 61, 62, 70). Thus, we employed the binomial distribution,

where P is the probability that a hexamer (therefore, n = 6) contains x WT subunits, and p is the probability that a WT subunit is incorporated (26, 61, 62, 70). Experiments demonstrated that WT and A503V subunits have a similar probability of being incorporated into a hexamer (26, 62). Consequently, p is calculated as the molar ratio of WT and A503V protein present.

Therefore, for any specified percentage of WT subunits, the probability distribution of Hsp104^A503V hexamers containing zero, one, two, three, four, five, or six WT subunits can be derived (Fig. 1B) (26, 61, 62, 70). Activity versus p plots could then be generated, assuming each A503V subunit makes an equal contribution to the total activity (one-sixth per subunit) (Fig. 1C) (26, 61, 62, 70). Consequently, if subunits within the Hsp104^A503V hexamer operate independently, then activity should decline in a linear manner upon incorporation of WT subunits (26, 61, 62, 70). Conversely, if subunits are coupled, then a specific number of WT subunits will be sufficient to eliminate activity (26, 61, 62, 70). Thus, zero activity is assigned to hexamers that are in breach of a specific threshold number of WT subunits (26, 61, 62, 70). In this way, we can generate activity versus p plots if we assume that 1 or more, 2 or more, 3 or more, 4 or more, or 5 or more WT subunits are required to eliminate activity (26, 61, 62, 70).

To further model the observed inhibitory effect of WT Hsp104 subunits on Hsp104^A503V activity, we employed the binomial distribution as above but imposed an additional rule whereby WT subunits repress the activity of adjacent A503V subunits by a factor of r (26, 35, 62, 71). Thus, we scored each subunit-subunit interface of every possible heterohexamer in each possible configuration as follows: interfaces were scored as 1/6 if at an A503V:A503V junction, r/6 if at an A503V:WT junction, or 0 if at a WT:WT junction (26, 62). Activity was then normalized to the predicted heterohexamer population as defined by the binomial distribution above (Fig. 1B) (26, 59).

Yeast Strains, Media, and Plasmids

All yeasts were W303aΔhsp104 (MATa, can1-100, his3-11,15, leu2-3,112, trp1-1, ura3-1, ade2-1) (72). Yeasts were grown in synthetic medium lacking the appropriate components. Medium was supplemented with 2% glucose, raffinose, or galactose. Vectors encoding TDP-43, FUS, and α-syn (pAG303GAL-TDP-43, pAG303GAL-FUS, and pAG303GAL-α-syn-GFP, respectively) were from A. Gitler (73–75). pRS416GAL-Hsp104 variants have been described previously (62). Hsp104 mutations were generated using QuikChange site-directed mutagenesis (Agilent Technologies, Santa Clara, CA) and confirmed by DNA sequencing.

Yeast Transformation and Spotting Assays

Yeast were transformed according to standard protocols using polyethylene glycol and lithium acetate (76). For the spotting assays, yeasts were grown to saturation overnight in raffinose supplemented dropout medium at 30 °C. Cultures were serially diluted 5-fold and spotted in duplicate onto synthetic dropout medium containing glucose or galactose. Plates were analyzed after growth for 2–3 days at 30 °C.

Western Blotting

Yeasts were grown and induced in galactose-containing medium for 5 h (TDP-43 and FUS) or 8 h (α-syn-GFP). Cultures were normalized to an optical density of 0.6; 6 ml of cells were then harvested and treated with 0.2 m NaOH for 5 min at room temperature. The resulting cell pellets were resuspended in 100 μl of 1× SDS sample buffer and boiled. Cell lysates were separated using SDS-PAGE (4–20% gradient; Bio-Rad) and then transferred to a PVDF membrane (EMD Millipore, Billerica, MA). Membranes were blocked using LI-COR blocking buffer for 1 h at room temperature. Primary antibody incubations were performed at 4 °C overnight. Antibodies used were as follows: rabbit anti-GFP polyclonal (Sigma-Aldrich, catalog no. G1544), rabbit anti-TDP-43 polyclonal (Proteintech (Chicago, IL), catalog no. 10782-2-AP), rabbit anti-FUS polyclonal (Bethyl Laboratories (Montgomery, TX), catalog no. A300-302A), rabbit anti-Hsp104 polyclonal (Enzo Life Sciences, catalog no. ADI-SPA-1040-F), and mouse anti-3-phosphoglycerate kinase monoclonal (Novex (Frederick, MD), catalog no. 459250). Blots were processed using goat anti-mouse and anti-rabbit secondary antibodies from LI-COR Biosciences (Lincoln, NE) and imaged using an Odyssey Fc Imaging system (LI-COR Biosciences).

Results

Hsp104 Hexamers Must Contain 2–3 A503V Subunits for Enhanced Disaggregase Activity

To gain insight into the mechanism of Hsp104 potentiation, we focused on Hsp104^A503V (Fig. 1A), which is among the strongest suppressors of α-syn, FUS, and TDP-43 toxicity in yeast (62, 63, 67). First, we explored the effects of WT Hsp104 subunits on the disaggregase activity of Hsp104^A503V hexamers. To do so, we exploited the strict requirement of Hsp104 for Hsp70 and Hsp40 to reactivate luciferase trapped in disordered aggregates in the presence of ATP (23, 26). In the absence of Hsp70 and Hsp40, Hsp104 reactivation activity was abolished (Fig. 1C) (23, 26). By contrast, Hsp104^A503V displayed robust luciferase reactivation activity in the absence of Hsp70 and Hsp40 (Fig. 1C) (62). Thus, under these conditions, we are certain that if the number of WT subunits per Hsp104 hexamer exceeds five, then the hexamer is inactive. To determine how many A503V subunits per hexamer are required to elicit disaggregase activity in the absence of Hsp70 and Hsp40, we employed a mutant subunit doping strategy (26, 70). Here, subunits were mixed to generate heterohexamer ensembles according to the binomial distribution (Fig. 1B) (26, 70). We have previously demonstrated that Hsp104 and Hsp104^A503V assemble into dynamic hexamers that rapidly exchange subunits, ensuring statistical incorporation of individual subunits (Fig. 1B) (26, 62). By applying different heterohexamer ensembles composed of Hsp104 and Hsp104^A503V subunits to reactivate disordered luciferase aggregates, we can obtain a measure of how many A503V subunits per hexamer are required for disaggregase activity in the absence of Hsp70 (Fig. 1, B and C) (26, 29, 61, 62, 70).

We assembled different heterohexamer ensembles of Hsp104^A503V and Hsp104^WT subunits (Fig. 1B) and assessed their luciferase reactivation activity (Fig. 1C). Incorporation of Hsp104 subunits into Hsp104^A503V hexamers caused a shallow curvilinear decline in luciferase disaggregase activity consistent with ∼4–5 WT Hsp104 subunits being necessary to inactivate the Hsp104^A503V hexamer (Fig. 1C, compare blue markers with purple and light blue lines). If four WT subunits inactivate the Hsp104^A503V hexamer, then two A503V subunits are not enough for activity (hence, we need three), and if five WT subunits inactivate the Hsp104^A503V hexamer, then a single A503V subunit is insufficient for activity (hence, we need two). Thus, 2–3 Hsp104^A503V subunits/hexamer are required to dissolve disordered luciferase aggregates in the absence of Hsp70. Strikingly, we previously demonstrated that two Hsp104 subunits must be capable of interacting with Hsp70 to enable Hsp104-mediated disaggregation of luciferase in the presence of Hsp70 (61). Hence, it is interesting that at least two A503V subunits were needed to enable otherwise WT hexamers to disaggregate luciferase in the absence of Hsp70 (Fig. 1C). This finding indicates that A503V subunits might mimic the conformation of Hsp70-activated, WT Hsp104 subunits. Moreover, this dose-response curve (Fig. 1C) also suggests that Hsp104^A503V subunits do not stimulate the activity of adjacent WT Hsp104 subunits (26, 62). Thus, a single Hsp104^A503V subunit is unable to potentiate an otherwise WT hexamer.

Interestingly, however, none of the theoretical curves exactly match the experimental data (Fig. 1C). Rather, the data fall mostly between the curves for four or five WT subunits being required to inhibit the activity of an Hsp104^A503V hexamer (Fig. 1C, compare blue markers with purple and light blue lines). However, we could model the observed behavior more precisely if we imposed rules whereby a WT subunit represses the activity of an adjacent A503V subunit by a factor of ∼0.3 and is inactive if adjacent to another WT subunit (Fig. 1D, compare blue markers with purple line; see “Experimental Procedures”).

Hsp104^A503V Disaggregase Activity Is Diminished by Mutations That Disrupt Substrate Binding to NBD Pore Loops

We next examined how Hsp104^A503V would respond to specific defects in substrate binding. Residues Tyr-257 and Tyr-662 are located in flexible pore loops located within each NBD and face the central channel through which substrate is translocated (Fig. 1A) (37, 38, 42–44). Each pore loop tyrosine engages substrate directly, and mutating these residues to alanine results in severely reduced (Y257A) or abolished (Y662A) disaggregation activity (43, 44). We introduced missense mutations Y257A, Y662A, or both in combination with A503V, resulting in Hsp104^Y257A/A503V, Hsp104^Y662A/A503V, and Hsp104^{Y257A/A503V/Y662A}. First, we measured the ATPase activity of these Hsp104 variants using the more sensitive low salt conditions of Hattendorf and Lindquist (52), which maximally stabilize Hsp104 hexamers. Under these conditions, surprisingly, Hsp104^A503V displays only ∼18% higher ATPase activity than Hsp104 (Fig. 2A). Thus, enhanced ATPase activity of Hsp104^A503V compared with Hsp104 determined previously at physiological salt concentrations may reflect slight variations in hexamer stability (62, 77, 78). Hsp104 displayed ATPase activity similar to that of Hsp104^Y257A and Hsp104^Y662A, but Hsp104^Y257A/Y662A ATPase activity was significantly elevated (Fig. 2A). We had not observed this increase previously at physiological salt concentrations (26). Compared with Hsp104^A503V, Hsp104^Y257A/A503V and Hsp104^Y662A/A503V had very similar ATPase activity, whereas Hsp104^{Y257A/A503V/Y662A} was ∼35% lower (Fig. 2A). Thus, the double pore loop mutation had opposite effects on ATPase activity in the wild type versus A503V background (Fig. 2A). Nonetheless, all pore-loop variants displayed robust ATPase activity under these conditions.

FIGURE 2.

Hsp104^A503V disaggregase activity is severely impaired by the Y257A mutation and ablated by the Y662A mutation. A, Hsp104 mutants (0.25 μm monomer) were incubated for 10 min with ATP (1 mm). Average absorbance from a no enzyme control reaction was subtracted from raw absorbance values. The resulting absorbance values were normalized to the average absorbance yielded by Hsp104^WT. Values represent means ± S.E. (error bars) (n = 16–48). A one-way analysis of variance with the post hoc Tukey's multiple comparisons test (*, p < 0.05; **, p < 0.01) was performed. B, urea-denatured firefly luciferase aggregates were incubated for 90 min at 25 °C with Hsp104 (1 μm hexamer) plus ATP. Luciferase reactivation was then determined and normalized to Hsp104^WT disaggregase activity. Values represent means ± S.E. (n = 6–90). C, reactions were performed as in B except that 1:1 mixtures of ATP and ATPγS replaced ATP. Values represent means ±S.E. (n = 6–84). D, reactions were performed as in B except that Hsp70 (1 μm) and Hsp40 (1 μm) were added. Values represent means ± S.E. (n = 6–84).

We next assessed the reactivation of luciferase trapped in disordered aggregates in the absence of Hsp70 and Hsp40 but in the presence of ATP. Under these conditions, Hsp104^A503V was ∼33-fold more active than Hsp104 (Fig. 2B). In the A503V background, the pore loop mutations abolished this enhanced activity and reduced refolding activity to a low level similar to Hsp104 (Fig. 2B) (23). The effect of Y662A was slightly greater than Y257A, but either mutation severely diminished Hsp104^A503V activity (Fig. 2B). Thus, Hsp104^A503V disaggregase activity is likely mediated via substrate interactions with Tyr-257 and Tyr-662.

In the presence of 1:1 mixtures of ATP and ATPγS, Hsp104 can disaggregate disordered aggregates in the absence of Hsp70 (26, 60, 61, 68). Under these conditions, Hsp104^A503V displays reduced luciferase reactivation activity compared with Hsp104 (Fig. 2C). This finding suggests that Hsp104^A503V responds differently from Hsp104 to mixtures of ATP and ATPγS, which we explore further below. In these conditions, Hsp104^Y257A and Hsp104^Y662A are devoid of luciferase reactivation activity. Likewise, all pore-loop variants bearing the A503V mutation showed diminished luciferase reactivation activity (Fig. 2C).

Next, we studied Hsp104 disaggregase activity in the presence of Hsp72 (an Hsp70) and Hdj1 (an Hsp40). Here, Hsp104^A503V luciferase-refolding activity was ∼2-fold lower than that of Hsp104 (Fig. 2D). This result was surprising because Hsp104^A503V has increased luciferase reactivation activity in the presence of Hsc70 and Hdj2 as well as Sse1, Ssa1, and Ydj1 (62, 67). Thus, the identity of Hsp70 and Hsp40 chaperone partners can modulate the activity of Hsp104^A503V differently from Hsp104. For both Hsp104 and Hsp104^A503V, Y257A or Y662A greatly reduced activity (Fig. 2D). The Y257A mutation greatly impaired activity in both Hsp104 and Hsp104^A503V, whereas the Y662A or the Y257A/Y662A mutations abolish activity (Fig. 2D) (62). Thus, Tyr-257 and Tyr-662 play a crucial role in Hsp104^A503V disaggregase activity, indicating that potentiated Hsp104 variants employ a mechanism of substrate translocation similar to that of Hsp104.

To complement our in vitro studies, we utilized yeast models of cytotoxic misfolding and aggregation of proteins involved in neurodegenerative diseases, such as Parkinson disease and ALS (74, 75, 79). In Parkinson disease, α-syn assembles into toxic soluble oligomers and amyloid fibrils (80). In yeast, expression of α-syn from the galactose promoter induces cytoplasmic aggregation and toxicity (79). In these experiments, we exploited a Δhsp104 yeast strain to avoid any potential interference from endogenous Hsp104 (Fig. 3, A and B) (62). Deletion or overexpression of Hsp104 does not affect α-syn aggregation or toxicity in yeast (62, 63, 67). Thus, we can be certain that any rescue of toxicity by an Hsp104 variant in the Δhsp104 background is due to a gain of Hsp104 therapeutic function (62, 63, 67). We first established that differences in growth were not due to changes in the expression of α-syn or Hsp104 variants via immunoblotting (Fig. 3B). As expected, a vector control, Hsp104, Hsp104^Y257A, Hsp104^Y662A, or Hsp104^Y257A/Y662A did not reduce α-syn toxicity (Fig. 3A). By contrast, Hsp104^A503V strongly suppressed α-syn toxicity (Fig. 3A) (35, 62, 63, 66). This rescue was severely impaired by the Y257A mutation and ablated by the Y662A or Y257A/Y662A mutations (Fig. 3A).

FIGURE 3.

Hsp104^A503V-mediated rescue of yeast proteinopathy models is severely impaired by the Y257A mutation and ablated by the Y662A mutation. Δhsp104 yeasts integrated with genes encoding α-syn (A), FUS (C), or TDP-43 (E) were transformed with the indicated Hsp104 variants or empty vector control. Strains were serially diluted 5-fold and spotted on glucose (off) or galactose (on) medium. B, D, and F, strains from A, C, and E, respectively, were induced, lysed, and immunoblotted. 3-Phosphoglycerate kinase (PGK) was used as a loading control. Results shown are representative of at least three independent trials.

In ALS, the RNA-binding proteins with prion-like domains, TDP-43 and FUS, mislocalize from the nucleus and form cytoplasmic aggregates in degenerating motor neurons (17, 18, 80). In yeast, expression of FUS or TDP-43 from the galactose promoter induces their cytoplasmic aggregation and toxicity (73–75). Here too, we utilized a Δhsp104 yeast strain (Fig. 3, D–F), in which either FUS or TDP-43 aggregates and is highly toxic (62, 73, 81). Hsp104^A503V mitigated cytoplasmic FUS aggregation and toxicity in yeast, whereas Hsp104 had no effect (Fig. 3C) (35, 62, 63, 66, 81). This rescue was abrogated by the Y257A mutation (Fig. 3C). No other Hsp104 variant tested here reduced FUS toxicity (Fig. 3C). Consistent with previous reports (62, 63), we found that Hsp104^A503V slightly reduced FUS expression levels (Fig. 3D). However, Hsp104^A503V was also expressed at lower levels than the other Hsp104 variants (Fig. 3D).

None of the pore loop mutants could suppress TDP-43 toxicity in yeast (Fig. 3E) (62, 73, 74). Only Hsp104^A503V rescued TDP-43 toxicity (Fig. 3E). Hsp104^A503V very slightly reduced TDP-43 expression level and was itself also expressed at lower levels than Hsp104 (Fig. 3F). We conclude that the Hsp104 potentiation conferred by A503V is severely disrupted by the Y257A mutation and ablated by the Y662A mutation.

Hsp104^A503V Responds Differently from Hsp104 to Mixtures of ATP and ATPγS

Unlike Hsp104, Hsp104^A503V was not stimulated in luciferase reactivation by a 1:1 ratio of ATP to ATPγS in the absence of Hsp70 (Fig. 2C). This finding suggested that an altered pattern of ATP hydrolysis by Hsp104^A503V hexamers might contribute to potentiation. To explore this idea further, we examined the effect of various ratios of ATP and ATPγS on Hsp104 and Hsp104^A503V disaggregase activity. We kept the total adenine nucleotide concentration constant but varied the ATP/ATPγS ratio. In Fig. 4A, we present the data normalized to WT Hsp104 maximal activity to reveal the amplitude of the activity of the different Hsp104 variants tested. In Fig. 4B, we present the data normalized to the maximal activity of each individual Hsp104 variant to reveal the optimal ATP/ATPγS ratio. Hsp104 exhibited maximal luciferase reactivation activity at ∼40–50% ATPγS (Fig. 4, A and B, blue trace) (26, 35). By contrast, the dose response of Hsp104^A503V was left shifted toward lower levels of ATPγS (Fig. 4, A and B, compare blue and green traces). Thus, Hsp104^A503V exhibited maximal luciferase reactivation activity at ∼20–30% ATPγS (Fig. 4, A and B, green trace) and was sharply inhibited at ATPγS concentrations that were optimal for Hsp104 (Fig. 4, A and B). Utilization of ATPγS in combination with ATP stimulates disaggregation of disordered aggregates by Hsp104 via slowing ATP hydrolysis at subset of nucleotide-binding sites (26, 60). Thus, the heightened sensitivity of Hsp104^A503V to stimulation by lower proportions of ATPγS indicates that Hsp104^A503V requires slowing of ATP hydrolysis at fewer nucleotide-binding sites than Hsp104.

FIGURE 4.

Hsp104^A503V requires a lower proportion of ATPγS for maximal disaggregase activity in the absence of Hsp70 compared with Hsp104. A and B, urea-denatured firefly luciferase aggregates were incubated for 90 min at 25 °C with the indicated Hsp104 variant (1 μm hexamer) in the presence of the indicated amounts of ATP and ATPγS (total nucleotide concentration was kept constant). Values represent means ± S.E. (error bars) (n = 4–5). A, all data points were normalized to the maximum refolding activity of WT Hsp104. B, all data points were normalized to the maximum refolding activity for a given Hsp104 mutant.

ATP hydrolysis can be specifically slowed at NBD1 or NBD2 by mutating their conserved Walker A motif, which harbors a lysine residue that directly contacts the phosphates of ATP (21). Mutating the conserved lysine of the Walker A motif to threonine impairs ATP binding at that site (21, 49, 51). Alternatively, ATP hydrolysis can be slowed at NBD1 or NBD2 by mutating their conserved sensor-1 motif, which harbors a threonine or asparagine that interacts with the γ-phosphate of ATP (21, 52). Mutating the conserved threonine or asparagine of the sensor-1 motif to alanine does not affect ATP binding but inhibits ATP hydrolysis at that site (21, 52). Slowing ATPase activity specifically at NBD2 with K620T (Walker A) or N728A (sensor-1), but not NBD1 with K218T (Walker A) or T317A (sensor-1) (Fig. 1A), enables disaggregation of disordered aggregates by Hsp104 in the presence of ATP and absence of Hsp70 (Fig. 4, A and B) (data not shown) (60).

Next, we assessed how the disaggregase activity of the sensor-1 variants, Hsp104^T317A and Hsp104^N728A, was affected by increasing proportions of ATPγS. Hsp104^T317A showed no disaggregase activity in the presence of just ATP (Fig. 4, A and B, black trace) (60). Surprisingly, however, mixtures of ATP and ATPγS stimulated Hsp104^T317A disaggregase activity (Fig. 4, A and B, black trace). The dose response of Hsp104^T317A was similar to that of Hsp104 except slightly right-shifted toward higher ATPγS levels (Fig. 4, A and B, compare black and blue traces). Maximal Hsp104^T317A disaggregase activity was observed at ∼50% ATPγS, and at higher ATPγS concentrations, Hsp104^T317A disaggregase activity declined (Fig. 4, A and B, compare black and blue traces). These findings indicate that, surprisingly, slowing ATP hydrolysis at NBD2 even when it is the only intact NBD can stimulate Hsp104 disaggregase activity against disordered aggregates in the absence of Hsp70.

In striking contrast, Hsp104^N728A maximally reactivated luciferase in the presence of ATP but was sharply inhibited by increasing fractions of ATPγS (Fig. 4, A and B, red trace) (60). This finding suggests that slowing ATP hydrolysis at NBD1 when it is the only intact NBD strongly inhibits disaggregase activity against disordered aggregates in the absence of Hsp70.

Hsp104^T317A/A503V displayed a response to ATPγS similar to that of Hsp104^A503V except that it was slightly right-shifted toward higher ATPγS levels (Fig. 4, A and B, compare orange and green traces). Indeed, Hsp104^T317A/A503V was less inhibited than Hsp104^A503V by ATPγS at concentrations higher than 30% (Fig. 4, A and B, compare orange and green traces). Maximal Hsp104^T317A/A503V luciferase reactivation activity was observed at ∼30% ATPγS (Fig. 4, A and B, orange trace). Thus, the T317A mutation renders Hsp104 and Hsp104^A503V less sensitive to inhibition by higher ATPγS concentrations.

Remarkably, Hsp104^A503V/N728A was even more sensitive to inhibition by ATPγS than Hsp104^N728A (Fig. 4, A and B, compare purple and red traces). Hsp104^A503V/N728A maximally reactivates luciferase in the presence of ATP but is completely inhibited by as little as 10% ATPγS (Fig. 4, A and B, purple trace). Collectively, these observations indicate that altered ATP hydrolysis patterns might contribute to the enhanced activity of Hsp104^A503V. They also illustrate the startling mechanistic plasticity of the Hsp104 hexamer for disaggregating disordered aggregates (26).

Potentiated Hsp104^A503V Activity Can Tolerate Sensor-1 Mutations in NBD1 or NBD2

Hsp104^A503V exhibits increased ATPase activity at physiological salt concentrations (62, 77, 78) and is more sensitive to suramin, a small molecule inhibitor of Hsp104 ATPase activity (68). Furthermore, we show here that Hsp104^A503V responds differently to ATPγS (Fig. 4, A and B). Moreover, Hsp104^A503V ATPase activity is inhibited and not stimulated by polylysine, unlike Hsp104 (78). These results suggest that altered ATPase activity is a key element enabling the potentiated activity of Hsp104^A503V. Thus, we next delineated the requirements for ATPase activity at NBD1 and NBD2 for potentiated Hsp104^A503V activity. To do so, we introduced the AAA+ sensor-1 mutations T317A and N728A (Fig. 1A) (52). We also introduced mutations in the Walker A and Walker B motifs of both NBDs; Hsp104^K218T/K620T was unable to bind nucleotide, whereas Hsp104^E285Q/E687Q bound nucleotide but was unable to hydrolyze it (Fig. 1A) (82).

In the WT Hsp104 background, both sensor-1 mutants displayed significantly reduced ATPase activity, whereas the double Walker A and Walker B mutants showed almost no activity (Fig. 5A) (49, 51, 52, 82). Likewise, in the A503V background, the double Walker A or double Walker B mutants eliminated ATPase activity (Fig. 5A). The NBD1 sensor-1 mutant displayed reduced ATPase activity in the A503V background (Fig. 5A; compared with Hsp104^A503V). Indeed, the ATPase activity of Hsp104^T317A/A503V was reduced by ∼32% compared with Hsp104^A503V, and this reduction was statistically significant (Fig. 5A). This reduction was similar to the effect of T317A in WT Hsp104 (Fig. 5A). By contrast, the activity of Hsp104^A503V/^N728A was reduced by only ∼18% compared with Hsp104^A503V, and this reduction was not statistically significant (Fig. 5A). This reduction was not as large as the statistically significant ∼33% reduction of ATPase activity caused by N728A in the WT Hsp104 background (Fig. 5A). However, we note that the ATPase activity of Hsp104^T317A/A503V and Hsp104^A503V/N728A were not significantly different (Fig. 5A). Collectively, these findings begin to suggest that allosteric communication between NBD1 and NBD2 may be altered in Hsp104^A503V (52, 78). Indeed, NBD1 appears to make a larger contribution to the ATPase activity in Hsp104^A503V than in Hsp104 (Fig. 5A). These observations indicate that NBD1 may retain high ATPase activity when NBD2 is bound with ATP in the A503V background. By contrast, in Hsp104, NBD1 ATPase activity is low when ATP is bound by NBD2 (52).

FIGURE 5.

A potentiated Hsp104 variant can overcome deficits in ATP hydrolysis at NBD1 or NBD2 in vitro. A, Hsp104^A503V mutants (0.25 μm monomer) were incubated for 10 min with ATP (1 mm). Average absorbance from a no enzyme control reaction was subtracted from raw absorbance values. The resulting absorbance values were normalized to the average absorbance yielded by Hsp104^WT. Values represent mean ± S.E. (error bars) (n = 16–48). A one-way analysis of variance with the post hoc Tukey's multiple comparisons test (*, p < 0.05; **, p < 0.01; ***, p < 0.001) was performed. B, urea-denatured firefly luciferase aggregates were incubated for 90 min at 25 °C with Hsp104 (1 μm hexamer) plus ATP. Luciferase reactivation was then determined and normalized to Hsp104^WT disaggregase activity. Values represent mean ± S.E. (n = 6–78). C, reactions were performed as in B except that 1:1 mixtures of ATP and ATPγS replaced ATP. Values represent mean ± S.E. (n = 6–72). D, reactions were performed as in B except that Hsp70 (1 μm) and Hsp40 (1 μm) were added. Values represent mean ± S.E. (n = 6–72).

Next, we determined the luciferase disaggregase activity of these Hsp104 ATPase variants in the presence of ATP but in the absence of Hsp70. In the presence of ATP alone, the double Walker A or Walker B mutants displayed insignificant luciferase reactivation activity in both the WT and A503V backgrounds (Fig. 5B). As expected, Hsp104 and Hsp104^T317A were also inactive in this setting (Fig. 5B). Hsp104^A503V and Hsp104^T317A/A503V displayed robust disaggregase activity under these conditions (Fig. 5B). Thus, reducing ATPase activity at NBD1 did not affect the ability of Hsp104^A503V to disaggregate disordered aggregates. Hsp104^N728A reactivated luciferase trapped in urea-denatured aggregates ∼11-fold more effectively than Hsp104^A503V (Fig. 5B). Although Hsp104^N728A has previously been shown to be active under these conditions (60), we were surprised that it was considerably more active than Hsp104^A503V. Remarkably, Hsp104^A503V/N728A displayed even greater luciferase reactivation activity (Fig. 5B). Thus, slowing ATP hydrolysis at NBD2 substantially enhances the ability of Hsp104 and Hsp104^A503V to disaggregate disordered aggregates in the presence of ATP and absence of Hsp70 (60).

In the presence of a 1:1 mixture of ATP/ATPγS, the double Walker A or Walker B mutants were unable to elicit any luciferase reactivation in both the WT and A503V backgrounds (Fig. 5C). The T317A mutation had little effect on Hsp104 activity under these conditions but slightly stimulated Hsp104^A503V activity (Figs. 4 and 5C). By contrast, the N728A mutation strongly inhibited Hsp104 and Hsp104^A503V activity at 1:1 ATP/ATPγS (Figs. 4 and 5C). Thus, slowing ATP hydrolysis at NBD2 increases the sensitivity of Hsp104 and Hsp104^A503V to inhibition by ATPγS.

In the presence of Hsp70 (Hsp72) and Hsp40 (Hdj1), the double Walker A or Walker B mutants were inactive in the WT and A503V background (Fig. 5D). Surprisingly, under these conditions, Hsp104^T317A activity was very similar to that of Hsp104, whereas Hsp104^N728A exhibited elevated activity (Fig. 5D). Similar results were observed in the A503V background. Thus, Hsp104^T317A/A503V exhibited activity similar to that of Hsp104^A503V, whereas Hsp104^{A503V N728A} exhibited elevated activity and was the most active variant in this assay (Fig. 5D). Under these conditions, slowing ATP hydrolysis at NBD2 increased Hsp104 and Hsp104^A503V luciferase reactivation activity in the presence of Hsp70 and Hsp40, whereas slowing ATP hydrolysis at NBD1 had little effect on activity (Fig. 5D). A summary of the various activities of the Hsp104 variants tested here is presented in Table 1.

TABLE 1.

Summary of Hsp104 variants and their ATPase and luciferase reactivation activities under the conditions tested in this paper

Numbers for partial or increased activities are noted. Numbers in blue represent a decrease in activity with respect to Hsp104^WT, whereas numbers in red represent increased activity.

Next, we assessed whether these Hsp104 ATPase variants could rescue toxicity of α-syn, FUS, or TDP-43 in yeast (Fig. 6, A–F). In the context of WT Hsp104, none of the ATPase variants affected the expression or toxicity of α-syn (Fig. 6, A and B), FUS (Fig. 6, C and D), or TDP-43 (Fig. 6, E and F) in yeast. Thus, despite having enhanced ability to reactivate luciferase aggregates in vitro (Fig. 5, B and D), Hsp104^N728A was inactive against the neurodegenerative disease proteins in vivo (Fig. 6, A, C, and E). Previously, we had established that an important property of potentiated Hsp104 variants bearing mutations in the MD was increased disaggregase activity against disordered aggregates in the absence of Hsp70 and Hsp40 (62, 63, 65–67). However, our findings with Hsp104^N728A (Figs. 5 (B and D) and 6 (A, C, and E)) indicate that this activity is not sufficient for potentiated activity in vivo. Despite having enhanced ability to disaggregate disordered aggregates in vitro (Fig. 5, B and D), Hsp104^N728A is unable to resolve amyloid conformers (24, 33). By contrast, Hsp104^A503V has enhanced activity against both amyloid and non-amyloid aggregates in vitro (62, 63, 67). Thus, enhanced disaggregase activity against amyloid is a better predictor of potentiated activity in vivo.

FIGURE 6.

ATPase activity at NBD1 or NBD2 in Hsp104^A503V is sufficient to sustain potentiation in yeast proteinopathy models. Δhsp104 yeasts integrated with genes encoding α-syn (A), FUS (C), or TDP-43 (E) were transformed with the indicated Hsp104 mutants or control. Strains were serially diluted 5-fold and spotted on glucose (off) or galactose (on) medium. B, D, and F, strains from A, C, and E, respectively, were induced for 5 or 8 h, lysed, and immunoblotted. 3-Phosphoglycerate kinase was used as a loading control. Results shown are representative of at least three independent trials.

Hsp104^A503V strongly rescued the toxicity of α-syn (Fig. 6A), FUS (Fig. 6C), and TDP-43 (Fig. 6E) in yeast (62, 63, 67). This potentiated activity was ablated by either the double Walker A or Walker B mutation (Fig. 6, A, C, E). Remarkably, however, both Hsp104^T317A/A503V and Hsp104^A503V/N728A retained potentiated activity in vivo and rescued toxicity of α-syn (Fig. 6A), FUS (Fig. 6C), and TDP-43 (Fig. 6E). Hsp104^T317A/A503V rescued the toxicity of all three disease proteins just as well as Hsp104^A503V (Fig. 6, A, C, and E), whereas Hsp104^A503V/N728A rescued α-syn and FUS toxicity just as well as Hsp104^A503V (Fig. 6, A and C) but had reduced ability to rescue TDP-43 toxicity (Fig. 6E). Rescue of α-syn toxicity was observed without any effect on α-syn expression level (Fig. 6B). By contrast, rescue of FUS toxicity by Hsp104^A503V and Hsp104^T317A/A503V was accompanied by a reduction in FUS expression level (Fig. 6D). However, Hsp104^A503V/N728A rescued FUS toxicity without affecting the FUS expression level (Fig. 6D). Hence, the reduction of FUS expression is not required to rescue toxicity. For TDP-43, Hsp104^A503V and Hsp104^T317A/A503V also modestly reduced TDP-43 expression level, whereas Hsp104^A503V/N728A did not (Fig. 6F). Therefore, the reduction in TDP-43 expression is also not required for rescue of toxicity. These findings suggest that the potentiated activity of Hsp104^A503V in yeast is largely unaffected by reducing ATPase activity at NBD1 or NBD2.

Discussion

Several missense mutations at specific but disparate positions in the MD or the small domain of NBD1 result in potentiated Hsp104 variants, able to dissolve protein aggregates implicated in neurodegenerative diseases (35, 62–67). These mutations probably disrupt a fragilely constrained, autoinhibited state of Hsp104 (67). However, the exact molecular basis of Hsp104 potentiation remains unclear. Here, we have dissected mechanistic aspects of a potentiated Hsp104 variant, Hsp104^A503V (35, 62–67). We have revealed how many subunits within the Hsp104 hexamer must bear a potentiating mutation to yield enhanced activity. We have also determined the ATPase and substrate-binding modalities that underpin potentiation in vitro and in vivo.

Using a mutant subunit doping strategy (26, 35, 61, 70), we have established that 2–3 subunits of an otherwise WT hexamer must bear A503V mutations to elicit enhanced disaggregase activity in the absence of Hsp70. Intriguingly, at least two Hsp104 subunits must interact with Hsp70 to enable disaggregation of disordered aggregates (61). We suggest that A503V subunits mimic the conformation of WT Hsp104 subunits that have been activated by Hsp70 binding. Hsp70 and Hsp40 are often sequestered in disease-associated aggregates, which might inhibit their function (83, 84). Thus, the ability of Hsp104^A503V to operate without Hsp70 and Hsp40 probably contributes to enhanced activity against disease-associated substrates in yeast (62, 63, 66, 67). Nonetheless, the NBD2 sensor-1 variant, Hsp104^N728A, can disaggregate disordered aggregates (but not amyloid) in vitro in the absence of Hsp70 (24, 33, 60), but, in contrast to Hsp104^A503V, Hsp104^N728A is unable to rescue α-syn, FUS, and TDP-43 toxicity in yeast. It is possible that Hsp104^N728A ATPase activity is too low to process certain aggregated substrates in vivo or that there are elements inhibiting activity in vivo that we do not reconstitute in our in vitro experiments. Regardless, these findings suggest that separation from Hsp70 and Hsp40 is not the only important determinant in Hsp104 potentiation.

We show that the Hsp104 potentiation conferred by A503V is severely impaired by mutating conserved substrate-binding tyrosines to alanine. In WT Hsp104, mutating Tyr-257 and Tyr-662 to alanine results in severely reduced (Y257A) or abolished (Y662A) disaggregation activity (43, 44). This was also the case for Hsp104^A503V. Hsp104^Y257A/A503V retained slightly more activity than Hsp104^A503V/Y662A in all conditions tested in vitro. However, the Y257A pore-loop variants are unable to rescue yeast proteinopathy models. These results suggest that Hsp104^A503V recognizes and translocates substrates via direct contact with conserved Tyr-257 and Tyr-662 pore-loop residues in a manner similar to Hsp104.

In the absence of Hsp70, Hsp104^A503V disaggregase activity against disordered aggregates responded differently from Hsp104 to mixtures of ATP and ATPγS. Indeed, optimal Hsp104^A503V disaggregase activity was observed at ∼4:1 ATP/ATPγS compared with ∼1:1 for Hsp104. Mixtures of ATPγS and ATP stimulate disaggregation of disordered aggregates by Hsp104 via decelerating ATP hydrolysis at a subset of nucleotide-binding sites (26, 60). Thus, stimulation of disaggregase activity by lower fractions of ATPγS indicates that Hsp104^A503V requires decelerated ATPase activity at fewer nucleotide-binding sites. This finding suggests that an altered pattern of ATP hydrolysis underlies Hsp104 potentiation.

Analysis of various ATPase-defective mutants revealed that enhanced Hsp104^A503V activity requires ATPase activity because double Walker A or Walker B mutants were ATPase-dead and non-functional in all assays studied here (62). Interestingly, the sensor-1 mutants, T317A and N728A (52), revealed alterations in allosteric signaling between NBD1 and NBD2 in Hsp104^A503V. In the WT background, both T317A and N728A significantly reduce Hsp104 ATPase activity (52). Indeed, based on this observation, it is suggested that Hsp104 NBD1 ATPase activity is low when ATP is bound by NBD2 (52). However, in Hsp104^A503V, only the T317A mutation significantly impairs ATPase activity. Indeed, N728A has little effect on Hsp104^A503V ATPase activity. It should be noted that despite these differences, the ATPase activities of Hsp104^T317A/A503V and Hsp104^A503V/N728A were not significantly different. Nonetheless, collectively, our findings begin to suggest that NBD1 makes a larger contribution to ATPase activity in Hsp104^A503V than in Hsp104. The essentially unaltered ATPase activity of Hsp104^A503V/N728A may indicate that NBD1 retains high ATPase activity when NBD2 is bound with ATP in the A503V background. We suggest that this alteration in allosteric signaling between NBD1 and NBD2 probably plays an important role in Hsp104 potentiation.

Remarkably, both Hsp104^T317A/A503V and Hsp104^A503V/N728A retain potentiated activity in vivo and in vitro. Thus, the potentiated activity of Hsp104^A503V is largely unaffected by reducing ATPase activity at NBD1 or NBD2. Hsp104^A503V/N728A displayed slightly reduced ability to rescue TDP-43 toxicity in yeast but still provided significant rescue. By contrast, the equivalent sensor-1 mutations in WT Hsp104 greatly reduce activity in yeast with respect to thermotolerance and prion propagation (52, 85). Thus, Hsp104^A503V displays a more robust activity with an operational plasticity that is unperturbed by mutations that greatly reduce activity of Hsp104 in vivo (52, 62, 85). A summary of the key findings of our study is presented in Fig. 7.

FIGURE 7.

Summary of key mechanistic insights into Hsp104 potentiation. PD, Parkinson disease.

These insights into the molecular underpinnings of Hsp104 potentiation help to lay foundations to further develop next generation Hsp104 variants with ameliorated therapeutic utility for neurodegenerative diseases. For example, our findings hint that the specificity of potentiated Hsp104 variants might be sharpened by more subtle alterations to the NBD1 pore-loop or by more severely reducing NBD2 ATPase activity. Potentiated Hsp104 variants with enhanced substrate or conformer selectivity might exhibit reduced off-target effects and consequently display increased therapeutic efficacy and safety (66).

Author Contributions

M. P. T. conceived and coordinated the study; designed, performed, and analyzed the experiments shown in Figs. 2, 3, 4, 5, and 6; generated Figs. 1A and 7; and wrote the manuscript. E. C. designed, performed, and analyzed the experiments shown in Figs. 2, 4, and 5. M. M. N. performed and analyzed the experiments shown in Figs. 3 and 6. M. E. J. conceived, designed, performed, and analyzed the experiment shown in Fig. 1, B–D, and prepared reagents used in Figs. 2, 4, and 5. M. S. G. contributed essential unpublished data, analysis, and interpretation for Fig. 2. J. S. conceived, coordinated, and directed the study and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.