Suramin Inhibits Hsp104 ATPase and Disaggregase Activity

Hsp104 is a hexameric AAA+ protein that utilizes energy from ATP hydrolysis to dissolve disordered protein aggregates as well as amyloid fibers. Interestingly, Hsp104 orthologues are found in all kingdoms of life except animals. Thus, Hsp104 could represent an interesting drug target. Specific inhibition of Hsp104 activity might antagonize non-metazoan parasites that depend on a potent heat shock response, while producing little or no side effects to the host. However, no small molecule inhibitors of Hsp104 are known except guanidinium chloride. Here, we screen over 16,000 small molecules and identify 16 novel inhibitors of Hsp104 ATPase activity. Excluding compounds that inhibited Hsp104 activity by non-specific colloidal effects, we defined Suramin as an inhibitor of Hsp104 ATPase activity. Suramin is a polysulphonated naphthylurea and is used as an antiprotozoal drug for African Trypanosomiasis. Suramin also interfered with Hsp104 disaggregase, unfoldase, and translocase activities, and the inhibitory effect of Suramin was not rescued by Hsp70 and Hsp40. Suramin does not disrupt Hsp104 hexamers and does not effectively inhibit ClpB, the E. coli homolog of Hsp104, establishing yet another key difference between Hsp104 and ClpB behavior. Intriguingly, a potentiated Hsp104 variant, Hsp104A503V, is more sensitive to Suramin than wild-type Hsp104. By contrast, Hsp104 variants bearing inactivating sensor-1 mutations in nucleotide-binding domain (NBD) 1 or 2 are more resistant to Suramin. Thus, Suramin depends upon ATPase events at both NBDs to exert its maximal effect. Suramin could develop into an important mechanistic probe to study Hsp104 structure and function.


Introduction
For proteins to perform their biological function, folding into the appropriate three-dimensional shape is of paramount importance [1]. Protein misfolding can result in cellular toxicity and lead to catastrophic diseases, such as Parkinson disease, Huntington disease and amyotrophic lateral sclerosis [1][2][3]. Thus, cells have evolved sophisticated chaperone systems to promote successful protein folding and preserve proteostasis [4,5]. While most chaperones act by preventing protein misfolding [5], Hsp104 is capable of reversing protein aggregation [3,[6][7][8].
Hsp104 is a member of the AAA+ family of ATPases and utilizes energy from ATP hydrolysis to dissolve disordered protein aggregates as well as amyloid fibers [3,6,8,9]. It assembles into a homohexameric ring structure with a central channel [7]. Hsp104, and its bacterial homolog ClpB, drive protein disaggregation by directly translocating substrates through this channel [10][11][12][13][14][15]. Each Hsp104 monomer contains an N-terminal domain, two AAA+ nucleotide-binding domains (NBD1 and NBD2), a coiledcoil middle domain, and a C-terminal region required for hexamerization [16]. Both NBDs contain Walker A and Walker B motifs that are critical for nucleotide binding and hydrolysis, respectively [17]. Most ATP hydrolysis happens at NBD1, whereas NBD2 has a primarily nucleotide-dependent oligomerization function [18,19].
Hsp104 is highly conserved in eubacteria and eukaryotes [23,24]. Indeed, Hsp104 is essential for cell viability in challenging conditions when proteins tend to aggregate more readily [26,27]. Animal cells do not have an Hsp104 homolog [23,24]. Thus, Hsp104 is a promising drug target against a myriad of microorganisms. For instance, Hsp101, the Hsp104 homolog in the malaria parasite Plasmodium falciparum is essential for parasite survival and has become a prime drug target [28,29]. Indeed, a small molecule Hsp104 inhibitor could potentially treat a great variety of infections. Moreover, such a small molecule could greatly aid in the study of the structural and mechanistic basis of Hsp104 activity. Not only would a small-molecule inhibitor provide a way to rapidly silence Hsp104, but it might also hold the key to stabilizing Hsp104 hexamer structure to achieve a crystal structure that has remained so elusive. However, only one small-molecule inhibitor of Hsp104 activity is known to date: guanidinium hydrochloride (GdmCl), which is effective at millimolar concentrations [30,31]. High-throughput screening has led to small molecule inhibitors for other molecular chaperones such as Hsp70 and Hsp90, as well as other AAA+ proteins, including p97 and even ClpB [32][33][34][35][36]. Here, we employ a highthroughput screen of over 16,000 compounds and identify 16 novel inhibitors of Hsp104 ATPase activity. We then excluded small molecules that inhibit Hsp104 by non-specific colloidal mechanisms. Thus, we isolated Suramin as a robust inhibitor of Hsp104 ATPase and disaggregase activities. Suramin also interfered with the unfolding and translocation activities of Hsp104. Hsp104 inhibition by Suramin was not rescued by Hsp70 and Hsp40. Interestingly, Suramin cannot inhibit ClpB to the same extent as Hsp104, thus highlighting the functional differences between these two related proteins [9,16,37]. Suramin does not act by disrupting Hsp104 hexamers, but depends upon ATPase activity at NBD1 and NBD2 to exert its maximal effect on Hsp104.

Materials
All chemicals were purchased from Sigma Aldrich (St. Louis, MO) unless otherwise specified. Creatine kinase was purchased from Roche. Firefly luciferase and FITC-casein were purchased from Sigma Aldrich. Quantilum recombinant luciferase was purchased from Promega (Madison, WI). Hsp70 and Hsp40 were purchased from Enzo Life Sciences (Farmingdale, NY).

High-Throughput Screening Assay Reagents
ATPase activity was assessed by the release of inorganic phosphate determined using a malachite green phosphate detection kit (Innova Biosciences, Cambridge, United Kingdom). Microtiter plates (384-well) were purchased from PerkinElmer (Waltham, MA). Three libraries were screened: The NIH Clinical Collection (450 small molecules), the Spectrum Collection (2,000 compounds, MicroSource, Gaylordsville, CT) and the HitFinder Collection (14,400 compounds, Maybridge).

High-Throughput Hsp104 Plate Assay
We adapted the inorganic phosphate colorimetric assay to fit a 384-well plate format. A Z-factor between 0.8 and 1 was calculated for the assay [42]. A 2X Hsp104 stock solution was dispensed into each well using a Biotek microflo (Biotek Instruments, Winooski, VT). Compounds were then transferred using a Janus 96/384 Modular Dispensing Tool (PerkinElmer). Finally, a 2X ATP solution was dispensed using the Biotek microflo. Final concentrations of Hsp104 (monomer) and ATP were 0.25 mM and 1 mM respectively. Seeking to avoid the discovery of competitive inhibitors, we designed our highthroughput screen to use a 100-fold excess of ATP with respect to the test compounds (10 mM each). Final buffer conditions were 20 mM HEPES-KOH pH 7.4, 20 mM NaCl and 10 mM MgCl 2 . The ATPase reaction was allowed to continue for 60 min, at which time the reaction was terminated by addition of the colorimetric reagent. Absorbance was measured at 635 nm on an EnVision Xcite Multilabel plate reader (PerkinElmer). Hits were tested in duplicate. Sixteen compounds were found to lower ATPase activity of Hsp104. Ten of these were subjected to a 6point titration (40, 13.3  The amount of P i produced was measured by absorbance at 635 nm. Raw (non-normalized) absorbance values were transformed to a color scale, where the average absorbance produced by Hsp104 WT is represented as black. The color red represents absorbance values five-fold higher than average absorbance for Hsp104 WT . Green represents absorbance values five-fold lower than the average absorbance obtained for Hsp104 WT . Hsp104 DWB was used as a negative control (2). Hsp104 WT in the absence of inhibitors was used as a positive control (+).

Manual ATPase Assay
Wild-type or mutant Hsp104 (0.25 mM monomer) in ATPase buffer (20 mM HEPES-KOH pH 7.4, 20 mM NaCl and 10 mM MgCl 2 ) was incubated for 60 or 10 min at 25uC in the presence of ATP (1 mM) as noted. ATPase activity was assessed by the release of inorganic phosphate determined by using a malachite green phosphate detection kit (Innova). To ensure Suramin did not interfere with the colorimetric detection of free phosphate, Hsp104 WT (0.25 mM monomer) was incubated with ATP (1 mM) for 60 min and then different concentrations of Suramin were added to the reaction followed by the addition of the colorimetric reagents.

Gel Filtration
Hsp104 was diluted into ATPase buffer with or without Suramin (100 mM) in the presence or absence of ATP (1 mM). The Hsp104 concentration was adjusted to 120 mM monomer and 20 mL of sample were fractionated by a Superdex 200 10/300 column (GE Healthcare Life Sciences, Piscataway, NJ) at room temperature and at a flow rate of 0.5 mL per minute in line with an Optilab T-rEX Refractive Interferometer (Wyatt Technologies, Santa Barbara, CA). Molecular weight markers were purchased from Bio-Rad Laboratories (Hercules, CA).

Luciferase Activity Determination
Quantilum recombinant luciferase (native) was diluted to 5000, 500, 50, 5 and 0.5 nM in the presence or absence of Suramin. Dilutions were mixed with luciferase assay system (Promega). Luciferase activity was assessed by luminescence measured on a Safire 2 microplate reader.

Results and Discussion
High-Throughput Screening for Small-Molecule Inhibitors of Hsp104 ATPase Activity We aimed to identify small molecules capable of inhibiting Hsp104. To do so, we employed high-throughput screening using Hsp104 ATPase activity as a proxy for function. To detect changes in Hsp104 ATPase activity, we used a colorimetric assay detecting free inorganic phosphate (P i ) based on the formation of a colored complex between P i and a dye molecule under acidic conditions [43,44]. P i is produced from the Hsp104-driven hydrolysis of ATP. The colored complex concentration, a surrogate for P i concentration and ATPase rate, is determined by measuring the absorbance of a sample at 635 nm. We used recombinant Hsp104 purified from E. coli. An Hsp104 mutant with mutations in both NBD1 and NBD2 Walker B sequence motifs (Hsp104 DWB , carrying ATPase ablating mutations E285Q: E687Q) [45] was used as a negative control for ATPase activity [9]. Hsp104 WT activity results in a high A 635 (high ATPase), while Hsp104 DWB activity results in a low A 635 (low ATPase). We optimized the P i colorimetric assay to fit a 384-well plate format. We selected the Hsp104 concentration (0.25 mM) on the basis of previous experience in our laboratory [9]. The ATP concentration was selected to be in a 100-fold excess of ATP with respect to the test compounds. Furthermore, we found that starting the reaction by the addition of ATP (as opposed to Hsp104) yielded the most reliable results; similarly, a reaction time of 60 minutes (compared to 10 minutes, 30 minutes and overnight) maximized the response for the assay (data not shown). We screened 16,850 compounds from the NIH Clinical Collection as well as the Spectrum and Maybridge compound libraries. Sixteen compounds were found to inhibit the ATPase activity of Hsp104 (Table 1). For instance, hexachlorophene and tannic acid inhibit Hsp104 ATPase activity by ,80%, while Merbromin and Sennoside a inhibit by ,20% ( Table 1).
Several of the small molecules uncovered by our screen contained catechol groups, which are frequently a feature of promiscuous compounds in biochemical high throughput screens [46]. To exclude inhibition by non-specific colloidal effects [47,48], we performed a dilution series for 10 out of the 16 inhibitors in the presence of: (1) Triton X-100, (2) BSA, and (3) both BSA and Triton X-100 ( Figure S1). A common mechanism behind false-positive inhibition is the formation of promiscuous aggregates by self-assembly of small organic molecules in aqueous solution [47,48]. These aggregates bind proteins and nonspecifically inhibit their activity [47,48]. By counter-screening in the presence of detergent, we can exclude small molecules that are likely to be inhibiting via the formation of aggregates as these are sensitive to detergent [49]. To further exclude non-specific interactions, we chose to use BSA, which has been proposed as an alternative to ionic detergents in systems where detergents are not well tolerated [49]. The presence of 0.1% Triton X-100 relieved inhibition of ATP hydrolysis by many of the small molecules ( Figure 1A, second column from the left). BSA, at a concentration of 0.1 mg/mL, also appeared to lessen the effect of the inhibitors, though to a smaller extent than Triton X-100 ( Figure 1A, third column from the left). Out of our initial ten inhibitors, we found Suramin and Cisplatin to most significantly inhibit Hsp104 ATPase activity in the presence of both Triton X-100 and BSA ( Figure 1A, right column).
Suramin, an FDA-approved drug, is a symmetrical polysulfonated naphthylamine urea derivative ( Figure 1B) [50]. This drug was developed more than one hundred years ago as a treatment for human African sleeping sickness [51]. More recently, Suramin has been shown to be an effective anticancer agent [52] and to reverse several autism-related features in a mouse model of the disease [53,54]. Interestingly, Suramin and its derivatives have been shown to reduce the levels of prion protein in infected cells [55] by inducing its aggregation at the cell surface and thus inhibiting prion replication [56]. Suramin has a strong affinity for a variety of proteins and enzymes [57]; however, despite being a promiscuous inhibitor, it does not inhibit its targets by colloidal effects [58]. Suramin is also known to inhibit pyruvate kinases; its phenyl sulfonate groups bind their active sites in place of ATP [59]. Lastly, Suramin is known to inhibit the ATPase activity of RecA, a DNA-dependent ATPase involved in DNA repair [60].
Cisplatin ( Figure 1C, cisplatinum or cis-diamminedichloroplatinum (II)), also an FDA-approved drug, is a highly effective chemotherapeutic agent used to treat several types of cancer including ovarian, testicular, penile, cervical, lung and bladder cancers [61]. It interacts with DNA to form DNA adducts which activate several signal transduction pathways and culminate in the activation of apoptosis [62]. Cisplatin can also form protein adducts by covalent modification of cysteine and methionine residues, which can even induce protein crosslinking and aggregation [63]. Curiously, we were unable to replicate the inhibitory effect of Cisplatin outside of the 384-well format (Figure 2A, yellow bars). Hence, we did not pursue Cisplatin any further.

Suramin Inhibits Hsp104 ATPase Activity More Effectively Than GdmCl
We first set out to validate the effect of different concentrations of Suramin on the amount of ATP hydrolysis by Hsp104. In the high-throughput assay, we measured the amount of P i produced by Hsp104-mediated ATP hydrolysis after 60 minutes. We scaled up the volume of our reactions for manual validation, but kept all other conditions the same. Under manual assay conditions, GdmCl a well documented inhibitor of Hsp104 ATPase activity [30,31,64,65], only very mildly inhibited Hsp104 ATPase activity at the highest concentration tested (30 mM; Figure 2A, green bars). This finding is consistent with previous studies that found a large fraction of Hsp104 ATPase activity remained even at GdmCl concentrations of 100 mM [30]. By contrast, Suramin greatly diminished the amount of P i produced in a concentrationdependent manner and was effective at micromolar concentrations ( Figure 2A, orange bars). We confirmed Suramin did not interfere with the colorimetric detection of free phosphate (data not shown). To assess the effect of Suramin on the initial rate of ATP hydrolysis by Hsp104, we allowed the ATPase reaction to proceed for only 10 minutes in the presence of varying concentrations of Suramin ( Figure 2B). As before, Suramin greatly diminished the amount of P i produced in a concentrationdependent manner and was effective at micromolar concentrations ( Figure 2B). It is important to note, however, that the ATPase activity of Hsp104 was not completely ablated for any of the Suramin concentrations tested ( Figure 2B). Based on these data, we calculated the half maximum inhibitory concentration (IC 50 ) of Suramin to be ,3.39 mM.

Suramin Inhibits Hsp104 Refolding, Unfolding, and Translocation Activities
To assess Hsp104 function, the capacity of Hsp104 to refold amorphous protein aggregates was determined by measuring the amount of urea-denatured luciferase that could be reactivated by Hsp104 alone or in the presence of inhibitors. The activity of refolded luciferase is quantified by luminescence and then used as a proxy for Hsp104 refolding activity. To bypass the requirement for other chaperones, we used a 1:1 ratio of ATP to ATPcS to activate Hsp104 [25]. Despite only having a modest effect on Hsp104 ATPase activity (Figure 2A), all concentrations of GdmCl tested very strongly inhibited the luciferase reactivation activity of Hsp104 ( Figure 3A). Importantly, these GdmCl concentrations do not inhibit firefly luciferase refolding by other molecular chaperones [66], indicating a very strong inhibition of Hsp104 disaggregase activity as noted previously [8,67]. Thus, GdmCl grossly perturbs Hsp104 disaggregase activity without having an equivalent effect on Hsp104 ATPase activity (e.g. at 1 mM GdmCl). This disproportionate inhibitory effect of GdmCl on Hsp104 disaggregase activity compared to ATPase activity  suggests that the major effect of GdmCl is to uncouple Hsp104catalyzed ATP hydrolysis from protein disaggregation. Suramin, at a concentration of 100 mM, completely ablated the ability of Hsp104 to refold denatured luciferase, whereas 5 mM GdmCl was needed to elicit the same level of inhibition ( Figure 3A, B). This effect was gradually reduced with decreasing concentrations of Suramin ( Figure 3B). We calculated Suramin to have an IC 50 of ,10.1 mM ( Figure 3C). Unlike GdmCl (3 mM), Suramin (100 mM) impaired neither thermotolerance nor [PSI + ] propagation in Dpdr5 yeast (which lack Pdr5, an ABC transporter that expels small molecules from the cell [68]), two activities that absolutely require Hsp104 (data not shown) [26,27,65,69]. This lack of activity in vivo is likely due to poor uptake by yeast cells or titration by other Suramin-binding proteins. Indeed, Suramin is known to interact strongly with many proteins [57,70].
Despite the lack of in vivo activity, Suramin could still serve as a useful mechanistic probe to study Hsp104 function in vitro. Intriguingly, inhibition by Suramin was not rescued by inclusion of Hsp70 and Hsp40 in the disaggregation reaction ( Figure 3D, red bars). As a control, we verified that Suramin did not inhibit luciferase activity itself. At the luciferase concentration used in our refolding assay (50 nM), Suramin (100 mM) did not inhibit luciferase activity (data not shown). These data suggest that the observed reduction in luciferase refolding is most likely due to a reduction in Hsp104 activity.
Next, we established that Suramin hampers substrate translocation by Hsp104 using an engineered Hsp104 variant, HAP, which anchors to the bacterial peptidase ClpP to form a novel proteolytic system [13]. Thus, HAP translocates fluorescein isothiocyanate (FITC)-casein for degradation by ClpP. Fluorescence increases as FITC is released from degraded casein and can be used as a surrogate for substrate translocation [40]. Suramin, at a concentration of 100 mM, completely abolished substrate translocation ( Figure 4B, purple line). In contrast, treatment with 25 mM inhibitor permits translocation activity equivalent to that by untreated HAP in conjunction with ClpP ( Figure 4B, orange and green lines). Note that both the unfolding and translocation activities appear less sensitive to Suramin than disaggregase activity ( Figures 3B, 4A and 4B). Thus, disaggregation of protein aggregates makes more stringent demands on the Hsp104 hexamer than the unfolding or translocation of soluble substrates. Altogether, we have established that Suramin greatly decreases Hsp104 ATPase activity, drastically impairs its capacity to refold luciferase and hinders its substrate unfolding and translocation activities.

ClpB Displays Resistance to Suramin
We also assessed the effect of Suramin on ClpB, the E. coli homolog of Hsp104. Curiously, ClpB ATPase activity is not greatly hindered by Suramin even at high concentrations ( Figure 5A). For instance, 100 mM Suramin inhibits over 60% of Hsp104 ATPase activity, while the same concentration only inhibits 24% of ClpB ATPase activity ( Figure 2B and 5A). Next, we assessed ClpB-mediated refolding of chemically-denatured luciferase aggregates in the presence of Suramin. As before, we bypassed the use of co-chaperones by using a 1:1 ratio of ATP to ATPcS to stimulate ClpB, in the absence of DnaK, DnaJ and GrpE (Hsp70, Hsp40 bacterial homologs and a nucleotide exchange factor, respectively) [25]. Only high concentrations of Suramin (100 mM) drastically inhibited the ClpB-mediated refolding of luciferase ( Figure 5B). Nevertheless, 16% of ClpB disaggregase activity remains even at this concentration. Indeed, ClpB is much less sensitive to Suramin than Hsp104; for instance, at a Suramin concentration of 50 mM, ClpB retains 100% of its refolding power ( Figure 5B), while Hsp104 is practically inactive ( Figure 3B). Importantly, this result confirms that Suramin is not inhibiting the refolding of luciferase itself. Strikingly, low concentrations of Suramin increased the amount of luciferase ClpB was able to reactivate ( Figure 5B). The divergent behavior of Hsp104 and ClpB in response to Suramin further supports the notion that these two machines function by different mechanisms despite the similarities in their sequence and architecture [9]. Indeed, functional differences between ClpB and Hsp104 abound. For instance, NBD1 is primarily responsible for ATPase activity in Hsp104 [18,39,72,73], whereas both NBDs contribute in ClpB [74]. Furthermore, nucleotide binding to NBD1 is essential for ClpB hexamerization [74]; in contrast, nucleotide binding to NBD2 is needed for Hsp104 to hexamerize [18,72,75]. Hsp104 and ClpB even diverge in their mechanisms of collaboration with Hsp70 [37]. Most interestingly, Hsp104 is able to rapidly process amyloid substrates while ClpB cannot [6,9,76]. In addition to these functional disparities, recent data draws attention to potential structural differences between these proteins. Cryoelectron microscopy structures of the BAP variant of ClpB (which binds ClpP analogously to HAP) are not compatible with some of the disulfide bonds engineered in yeast Hsp104 [37,77]. Our results with Suramin further support the existence of operational differences between ClpB and Hsp104.

Suramin Does Not Disrupt Hsp104 Hexamers
Since Suramin is known to interfere with the oligomerization of proteins [78], we asked whether it might disrupt Hsp104 hexamers. Hsp104 must exist as a hexamer to hydrolyze ATP and perform its functions [72,79], and thus a small molecule hindering its oligomerization would inhibit activity. Figure 6 shows gel filtration elution profiles of Hsp104 with or without ATP in the absence or presence of Suramin (100 mM). Without nucleotide or inhibitor, Hsp104 eluted as a broad peak with an apparent size of ,600 kDa ( Figure 6, dark red line). Interestingly, the presence of Suramin did not significantly change the elution profile, demonstrating that the inhibitor does not interfere with hexamerization ( Figure 6, red line). When 1 mM ATP is present in the sample and running buffer, the majority of Hsp104 still eluted as a hexamer; however, a smaller peak corresponding to monomeric Hsp104 appears ( Figure 6, light blue line). The fraction of Hsp104 eluting as monomers in the presence of ATP could be due to ATP hydrolysis by Hsp104 on the column. Consistent with this idea, when Suramin is added and ATP hydrolysis is inhibited, only the ,600 kDa peak is observed ( Figure 6, dark blue line). From these data, we conclude that Suramin does not disrupt Hsp104 hexamers, and thus must inhibit the activity of hexameric Hsp104.

A Potentiated Hsp104 Variant Is Particularly Sensitive to Suramin
To learn more about how Suramin interacts with Hsp104, we examined its effect on a hyperactive variant of Hsp104, Hsp104 A503V [40,80,81]. This mutation is located in the middle domain and is thought to weaken autoinhibitory interactions that diminish Hsp104 activity or to induce conformational changes that mimic an allosteric activation of the protein [40]. The A503V mutation circumvents the need for Hsp70 and Hsp40 in remodeling amorphous aggregates (even in the absence of ATPcS) and displays increased ATPase activity, substrate translocation speed, unfoldase activity, and amyloid disaggregase activity [40]. Resembling our results for Hsp104 WT , Suramin reduced Hsp104 A503V ATPase activity in a concentration-dependent manner ( Figure 7A). Both Hsp104 and Hsp104 A503V ATPase are maximally inhibited at an inhibitor concentration of 25 mM. However, Hsp104 A503V is slightly more sensitive to inhibition. For example, at 25 mM Suramin, Hsp104 WT has an ATPase activity of 30% (relative to uninhibited Hsp104) while Hsp104 A503V has an activity of 18% ( Figure 7A).
Intriguingly, we found Hsp104 A503V luciferase reactivation activity to be considerably more sensitive to Suramin than that of Hsp104 WT ( Figure 7B). While Hsp104 WT has a 100% refolding activity in the presence of 1 mM Suramin, Hsp104 A503V has less than 50% activity at the same inhibitor concentration ( Figure 7B). Based on refolding data, the IC 50 for Hsp104 A503V is ,0.61 mM, which is much lower than that for Hsp104 WT (IC 50 ,10.1 mM). These observations further suggest that Hsp104 A503V hexamers are regulated differently than Hsp104 WT [40]. Suramin appears to be exploiting this difference to preferentially inhibit the potentiated variant.

Suramin Preferentially Inhibits Disaggregase Activity Catalyzed by ATP Hydrolysis at NBD2
To evaluate the effect of Suramin on individual NBDs, we took advantage of the AAA+ sensor-1 mutants T317A and N728A [39]. Hsp104 T317A and Hsp104 N728A can bind ATP but are unable to hydrolyze it at NBD1 and NBD2 respectively [39]. If, for instance, Suramin acts preferentially on NBD1, we would expect Hsp104 T317A to be resistant to Suramin. On the other hand, if Suramin acts on NBD2 then we would expect Hsp104 N728A to be unaffected.
Suramin inhibited Hsp104 T317A and Hsp104 N728A ATPase activity in a dose-dependent manner ( Figure 7A). However, the extent of inhibition was much lower for both Hsp104 T317A and Hsp104 N728A compared to Hsp104 WT and Hsp104 A503V . For instance, at 100 mM Suramin, Hsp104 WT ATPase activity is ,35% of that of the uninhibited protein, whereas Hsp104 T317A and Hsp104 N728A exhibit rates of ,65% and ,75% respectively ( Figure 7A). As these two mutants are only able to hydrolyze ATP at one NBD, they do not ''cycle'' through hydrolysis events at both NBDs [39], unlike Hsp104 WT and Hsp104 A503V . These findings suggest that maximal inhibition by Suramin depends on both NBD1 and NBD2 being able to hydrolyze ATP. Indeed, Hsp104 A503V cycles through more hydrolysis events at both NBDs than Hsp104 WT [40,82,83] and is even more sensitive to Suramin ( Figure 7A). Nonetheless, Suramin inhibits the ATPase activities of both Hsp104 T317A and Hsp104 N728A to a similar extent, indicating that it can inhibit the global ATPase activity of Hsp104, including ATPase reactions occurring at both NBDs. It is important to note that Suramin does not preferentially inhibit NBD1 or NBD2 ATPase activity. This finding differentiates Suramin from GdmCl, which inhibits the ATPase activity of NBD1 but not NBD2 of Hsp104 [84].
We found both Hsp104 T317A and Hsp104 N728A luciferase refolding activities to be inhibited by Suramin in a concentration-dependent manner ( Figure 7B). Akin to our results for ATPase activity, the extent of the inhibition was much lower for both Hsp104 T317A and Hsp104 N728A compared to Hsp104 WT . Thus, we find both sensor-1 mutants are more refractory to Suramin than Hsp104 WT ( Figure 7B). This result further reinforces that Suramin is not inhibiting the refolding of luciferase after it is released from Hsp104. Luciferase reactivation by Hsp104 N728A is more resistant to Suramin, revealing an inhibitor preference for protein disaggregation catalyzed by ATP hydrolysis at NBD2. For instance, at 50 mM Suramin, Hsp104 N728A retains ,36% refolding activity, while Hsp104 T317A is approximately 8% active ( Figure 7B). Likewise, Hsp104 N728A retains ,100% refolding activity in the presence of 1 mM Suramin, while Hsp104 T317A has approximately 75% activity at the same inhibitor concentration. These findings suggest that Suramin uncouples ATP hydrolysis from protein disaggregation more effectively at NBD2 than NBD1. Hence, we conclude that Suramin preferentially inhibits disaggregase activity catalyzed by ATP hydrolysis at NBD2. Based on refolding data, the IC 50 values for Hsp104 T317A and Hsp104 N728A are ,20.5 mM and ,92.4 mM respectively, which are much higher than that for Hsp104 WT (,10.1 mM) and Hsp104 A503V (,0.61 mM). These results suggest that Suramin more effectively inhibits Hsp104 when it cycles between ATP hydrolysis events at NBD1 and NBD2, as the sensor-1 mutants primarily hydrolyze ATP at one NBD and are less susceptible to the molecule than Hsp104 WT .
Hsp104 NBD1 and NBD2 have been proposed to adopt two distinct conformations (relaxed, R, and tense, T) that are reciprocally regulated by multiple allosteric pathways [17]. The tense conformation has low activity, while the relaxed conformation is highly active. It is proposed that upon ATP binding to NBD1, Hsp104 switches from a less active NBD1

Conclusions
We performed the first high throughput screen for Hsp104 ATPase inhibitors, encompassing over 16,000 small-molecule compounds. We found 16 molecules that inhibit Hsp104 ATPase activity in vitro. Of these, Suramin hinders the rate of ATPase activity in a specific, non-colloidal manner. Suramin also inhibits Hsp104 disaggregase, unfoldase, and translocase activities. Suramin inhibits Hsp104 without disrupting the oligomerization state of the disaggregase. Suramin-mediated inhibition of Hsp104 is not rescued by Hsp70 and Hsp40. Intriguingly, ClpB is much less sensitive to Suramin than Hsp104, which supports prior observations that these homologs function by different mechanisms. A potentiated variant of Hsp104 proved to be more sensitive to Suramin than the wild-type protein. Variants defective in ATP hydrolysis revealed a preference for Suramin to inhibit disaggregase activity catalyzed by NBD2 over NDB1. Overall, our data suggests Suramin takes advantage of Hsp104 ''cycling'' between ATP hydrolysis events at NBD1 and NBD2 to exert its maximal inhibitory effects. Future experiments will delineate the precise mechanism by which Suramin engages Hsp104 and exerts its inhibitory effects. We hope that Suramin will greatly aid in the study of the molecular mechanisms underlying Hsp104 function. Figure S1 Small Molecules that Inhibit Hsp104 ATPase Activity. Chemical structures and common names are shown for nine molecules found to inhibit Hsp104 ATPase activity. Gossypol-acetic acid complex was omitted for its similarity to Gossypol. (TIF)