Isoform-selective Genetic Inhibition of Constitutive Cytosolic Hsp70 Activity Promotes Client Tau Degradation Using an Altered Co-chaperone Complement*

Background: High levels of constitutive heat shock protein 70 (Hsc70) preserve Tau levels, possibly contributing to neuropathology. Results: A dominant negative Hsc70 (DN-Hsc70) variant mimics small-molecule inhibition and facilitates Tau clearance by altering the associated interactome. Conclusion: DN-Hsc70 recruits a pro-degradation chaperome complement to the Tau-Hsc70 complex. Significance: Selective Hsc70 inhibition could have a therapeutic benefit in tauopathies including Alzheimer disease. The constitutively expressed heat shock protein 70 kDa (Hsc70) is a major chaperone protein responsible for maintaining proteostasis, yet how its structure translates into functional decisions regarding client fate is still unclear. We previously showed that Hsc70 preserved aberrant Tau, but it remained unknown if selective inhibition of the activity of this Hsp70 isoform could facilitate Tau clearance. Using single point mutations in the nucleotide binding domain, we assessed the effect of several mutations on the functions of human Hsc70. Biochemical characterization revealed that one mutation abolished both Hsc70 ATPase and refolding activities. This variant resembled the ADP-bound conformer at all times yet remained able to interact with cofactors, nucleotides, and substrates appropriately, resembling a dominant negative Hsc70 (DN-Hsc70). We then assessed the effects of this DN-Hsc70 on its client Tau. DN-Hsc70 potently facilitated Tau clearance via the proteasome in cells and brain tissue, in contrast to wild type Hsc70 that stabilized Tau. Thus, DN-Hsc70 mimics the action of small molecule pan Hsp70 inhibitors with regard to Tau metabolism. This shift in Hsc70 function by a single point mutation was the result of a change in the chaperome associated with Hsc70 such that DN-Hsc70 associated more with Hsp90 and DnaJ proteins, whereas wild type Hsc70 was more associated with other Hsp70 isoforms. Thus, isoform-selective targeting of Hsc70 could be a viable therapeutic strategy for tauopathies and possibly lead to new insights in chaperone complex biology.

Cellular protein metabolism relies on a network of molecular chaperones that ensure proper folding and sorting of a protein for its intended function. The major chaperones in the cell, the Hsp90 2 and Hsp70 families, along with their assorted co-chaperones facilitate folding of nascent peptide chains, refold incorrectly folded proteins, or if refolding fails, send the protein for degradation. Hsp70s are extremely well conserved, multidomain proteins consisting of a nucleotide binding domain (NBD) that regulates the binding of either ATP or ADP, a substratebinding domain (SBD), and a C-terminal domain, containing a tetratricopeptide repeat domain. Exchange of ATP for ADP allosterically controls the refolding function of Hsp70s and their substrate affinity; binding of ATP stabilizes the NBD, resulting in a closed nucleotide binding cleft while simultaneously destabilizing the SBD, resulting in decreased substrate affinity due to a solvent-exposed hydrophobic cleft (1). ATP hydrolysis is stimulated by Hsp40/DnaJ proteins, a class of cochaperones that is also crucial for presentation of structurally diverse clients to Hsp70s (2,3). The exchange of ADP for ATP results in a relaxation of the SBD conformation, allowing substrates to disengage and new substrates to bind; for reviews, see Refs. 4 and 5. In addition to these refolding and degradation roles, members of the Hsp70 family of chaperones have other varied functions around the cell both constitutively as well as under conditions of cellular stress. There are ϳ10 isoforms of human Hsp70 proteins, including stress-inducible (Hsp72) and organelle-specific isoforms (Grp78, mortalin) (6,7). The constitutively expressed isoform, Hsp73/Hsc70, accounts for 2% of proteins in cells (8) and has long been implicated in clathrinmediated endocytosis (9 -11). Hsc70 is involved in a large variety of other cellular processes including exocytosis (12), trafficking of nuclear hormone receptors (13), transport of proteins from the Golgi to the endoplasmic reticulum (14), antigen presentation (15), and cytoskeletal function (16,17). A possible link between Hsc70 and microtubules is that the intrinsically disordered protein (IDP) Tau is a reported client of Hsc70 (18 -20). Tau is a microtubule-associated protein that functions to stabilize the microtubule network in neurons, providing support to the cell during axonal transport and controlling essential steps in microtubule dynamics (21)(22)(23). However, Tau is best known for its aberrant hyperphosphorylation and accumulation in a number of neurodegenerative disorders termed tauopathies (24,25) where it leads to the formation of Tau aggregates that are linked to cognitive deficits (26,27).
Building on a possible role for Hsc70 in microtubule dynamics, we previously showed that Hsc70 coordinates with the microtubule-associated protein Tau to regulate microtubules (18 -20). However, this relationship between Tau and Hsc70 can have untoward consequences in the aging brain as Hsc70 appears to contribute to the pathogenic accretion of Tau (28). We previously showed that overexpression of Hsc70 slowed Tau metabolism, an effect that could be overcome by small molecule pan Hsp70 inhibitors (28 -30). But most Hsp70 small molecule inhibitors are not isoform-selective, and none is available that specifically target Hsc70, making it challenging to define the specific role of Hsc70 activity on the turnover of clients like Tau (31)(32)(33)(34)(35)(36). Because Hsp70 isoforms appear to have opposing effects on Tau proteostasis, it is difficult to determine the impact of individual activities of Hsp70 isoforms on Tau biology (28). To answer the question about the role of Hsc70 activity on Tau turnover, we generated a mutant form of human Hsc70, E175S. Compared with wild type Hsc70, this variant lacked both refolding and ATPase activities, remained permanently in an ADP-bound conformation, and had a very different interactome that facilitated Tau reductions via the proteasome. These findings provide strong rationale to support the development of isoform-selective Hsp70 inhibitors for understanding Tau biology and possibly developing therapeutics around this target or its interaction partners.

Experimental Procedures
Reagents-Unless stated otherwise, all chemical reagents were purchased from Sigma.
Plasmids and Antibodies-For mammalian expression experiments, full-length human Hsc70 genes were cloned into pCMV6 (Origene, Rockville, MD) with an N-terminal FLAG tag to distinguish between endogenous and overexpressed isoforms. Mutations were introduced via site-directed mutagenesis (Stratagene, La Jolla, CA), and all sequences were verified before use. WT 4R0N Tau was cloned into pCDNA3.1ϩ as previously described (28). For recombinant protein preparations, WT-Hsc70, all Hsc70 mutants, and DnaJs were cloned into the pET28a vector (Midwest Center for Structural Genomics, Bethesda, MD).
Expression and Purification of Recombinant Proteins-Wild type 4R0N Tau, full-length Hsc70 (WT and E175S), and Hsc70 NBD fragments, DnaJ proteins, and BAG1 proteins were purified as described previously (28,37). For expression of isotopically 15 N-labeled full-length wild type and E175S Hsc70 protein, expression plasmids were transformed into Escherichia coli OneShot BL21 Star (DE3) cells (Life Technologies, Carlsbad, CA), and starter cultures were grown overnight in Luria-Bertani medium with the appropriate antibiotic. Overnight cultures were centrifuged, and cells were resuspended in M9 minimal media supplemented with 1 g/liter [ 15 N]ammonium chloride, 100 M CaCl 2 , 100 M MgSO 4 , 0.4% (w/v) glucose, and trace metals and vitamins, then grown at 37°C to an optical density of ϳ0.7 and induced with 1 mM isopropyl ␤-D-1-thiogalactopyranoside, and proteins were expressed at 37°C for 3 h. The cells were then harvested by centrifugation at 4000 ϫ g for 15 min at 4°C and resuspended in lysis buffer (500 mM NaCl, 20 mM Tris-HCl, 5 mM imidazole, 1ϫ protease inhibitor mixture III (EMD Millipore, Billerica, MA) and 1 mM phenylmethylsulfonyl fluoride, pH 8.0). Purification of protein from this resuspension was executed as described previously (28).
ATPase Activity for Hsc70 -Intrinsic ATPase rate was measured via malachite green assay with human WT or mutant Hsc70 (0.6 M) in the absence of co-chaperones or substrate as previously described (38,39). Briefly, proteins were incubated with nucleotide exchange factors and J proteins as indicated, and reactions were performed in the presence of ATP for 1 h, developed with malachite green reagent, and quenched with sodium citrate, and the resulting absorbance at 620 nm was measured.
Luciferase-monitored Refolding Assay-The DnaJ-stimulated refolding activity was characterized as previously described (40). Briefly, guanidine HCl-denatured luciferase was diluted into a HEPES-buffered ATP regenerating system containing WT or E175S Hsc70. Various DnaJ proteins were added to stimulate the reaction, which was allowed to proceed for 1 h at 37°C before measuring luminescence using the SteadyGlo reagent (Promega).
Fluorescence Polarization Assays-Fluorescence polarization experiments using the HLA-FAM (Anaspec) and ATP-FAM (Jena Bioscience) reporters were performed as described previously (37). Briefly, recombinant WT and E175S Hsc70 was incubated with reporter and florescence polarization was measured (excitation 485 nm, emission 535 nm) using a Spectra-Max M5 plate reader.
Fluorescence Binding by Flow Cytometry-Experiments were performed as described (37). Briefly, biotinylated recombinant WT and E175S Hsc70 were bound to streptavidin-coated polystyrene beads (Spherotech) in the presence of 1 mM nucleotide. Beads were washed and labeled BAG1 was applied. BAG1 binding was measured by quantification of bead-associated fluorescence by and Accuri C6 flow cytometer.
NMR Spectroscopy-The conformations of 15 N-labeled WT-Hsc70 and Hsc70-E175S were characterized in the ADP-Tau peptide VQIVYK-bound state and the ATP state. Typical concentration of the samples was 50 M in protein. Hsc70-hydrolyzed ATP was recycled by a regeneration system consisting of 100 mM phosphocreatine and 20 units of phosphocreatine-ADP kinase. 1 H, 15 N TROSY-HSQC spectra (transverse relaxation optimized spectroscopy-heteronuclear single quantum coherence) were collected at 700 or 900 MHz using a Bruker Avance III system equipped with a triple resonance cryo-probe (Michigan Research Corridor Facility at Michigan State University, East Lansing, MI). The TROSY spectra were collected in 10 h each, processed with NMRPipe, and plotted in Sparky (T. D. Goddard and D. G. Kneller, University of California, San Francisco, CA).
One-dimensional versions of the standard 15 N R 1 (10 h) HSQC experiments were carried out. The data set was processed in NMRPipe and exported in (ASCII) format using the Pipe2txt.tcl routine. The data were imported into Microsoft Excel and plotted. For each spectrum, the ranges 9.6 -8.6 ppm (structured core residues) and 8.6 -7.8 ppm (structured core and mobile tail residues) were integrated and fitted to a single exponential using in-house-written nonlinear least square fit code with jackknife error estimation (41). 15 N R 2 relaxation rates obtained from one-dimensional versions of a 15 N chemical shift anisotropy/ 1 H, 15 N dipolar transverse cross-correlated relaxation, xy , were obtained using symmetric reconversion. Four spectra, N xy 3 N xy , N xy 3 2N xy H z , 2N xy H z 3 N xy , and 2N xy H z 3 2N xy H z , measuring the auto-and cross-correlated relaxation rates over a period T of 10.2 ms were collected in 5 h total and processed as described above. The rates xy were obtained from the equation, where R is given by, Cell Culture, Lysis, Co-immunoprecipitation, and Immunoblotting and Viability-HEK293T cells were maintained, transfected, harvested, and immunoblotted as described previously (28). Viability was measured by 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) assay according to manufacturer's protocol (Promega, Madison, WI).
Organotypic Slice Culture, Transduction, and Immunofluorescence-All procedures involving experimentation on animal subjects were done in accordance with the guidelines set forth by the Institutional Animal Care and Use Committee of the University of South Florida. For slice cultures, post-natal day 5-9 FVB-Tg(tetO-MAPT*P301L) mice (The Jackson Laboratory, Bar Harbor, ME) were rapidly decapitated, and brains were sectioned at 400-m thickness on a vibratome (Leica Microsystems, Buffalo Grove, IL) in 0.6% glucose-PBS, pH 7.4. This model produces excess mutant P301L human Tau that pathologically condenses to the perikarya as in human disease (27). Slices were immediately plated onto Millicell inserts (EMD Millipore) in 6-well plates (Corning) containing complete culture medium (DMEM/F-12, 2 mM Glutamax, penicillin/ streptomycin, and 25% (v/v) heat-inactivated horse serum (Life Technologies)). Media was changed every 2-3 days. For adenoassociated virus (AAV), DIV2, slices were incubated with 10 l of WT-Hsc70-FLAG, AAV9 E175S-Hsc70-FLAG, or AAV9-GFP at 10 10 titer for 14 days. After transduction, slices were fixed, permeabilized, and blocked as described previously (42). Slices were incubated in the appropriate primary antibodies overnight followed by anti-goat Alexa-Fluor-conjugated secondary antibodies (Life Technologies). Nuclei were visualized with 4Ј,6-diamidino-2-phenylindole (DAPI) staining, then cover-slipped using ProLong antifade reagent (Life Technologies). Intensity analyses were performed with ImageJ.
siRNA-mediated Knockdown-siRNAs for human BECN1 and controls were purchased from Qiagen (Buffalo Grove, IL). Knockdowns were performed for 72 h on HEK293T with 40 nM siRNA using silentFECT (Bio-Rad).
Statistical Analyses-Statistical analyses were performed by one-way ANOVA tests with Tukey's post-hoc analysis for multiple comparisons or Student's t tests, as indicated in the figure legends, using GraphPad Prism 5.0 software.

A Human Hsc70 Variant Lacks ATPase and Refolding
Activities-We mutated specific residues in the NBD domain of human Hsc70 that we supposed would alter the ATPase activity of Hsc70 based on literature and structure (43)(44)(45) (Fig. 1A). These variants were screened for in vitro ATP hydrolysis and luciferase refolding in the presence of the co-chaperone Hsp40/ DnaJA2, which greatly enhances detection of these activities. Point mutations S208A, D152K, C267S, and F68L all demonstrated increased ATPase activity compared with WT Hsc70 (Fig. 1B). Only E175S Hsc70 lacked any appreciable ATPase activity compared with WT Hsc70 (Fig. 1B). We then assessed how these same point mutants impacted luciferase refolding activity. Interestingly, F68L Hsc70 increased luciferase refolding compared with WT, but E175S, D152K, and S208A Hsc70 variants had no refolding activity (Fig. 1C). The S208A and C267S mutations were able to refold luciferase similarly to WT. We then co-overexpressed each Hsc70 mutant with WT 4R0N Tau (Tau without any N-terminal inserts but four microtubule binding repeats) and assessed cellular Tau levels in HEK293T cells by Western blot (Fig. 1D) to understand which Hsc70 activity, ATPase or refolding, was most critical for Tau metabolism. Compared with mock-transfected samples, overexpression of WT Hsc70 preserved Tau levels in the cell as previously shown (28). Overexpression of E175S Hsc70 significantly decreased cellular Tau levels (Fig. 1, D and E; *, p Յ 0.05) as did D125K (Fig. 1, D and E). The S208A and C267S mutants had no significant impact on Tau stability. We were intrigued by the E175S mutant, which lacked both luciferase and ATPase activities yet significantly impacted Tau levels. We next tested cell viability, which was not significantly affected by overexpression of E175S Hsc70 in Tau-expressing cells, ruling out the possibility that Tau levels decreased due to cell death (Fig. 1F). These findings demonstrate that Tau clearance is facilitated most robustly when Hsc70 lacks both primary refolding and ATPase activities. This was similar to what we had previously observed with small molecule pan Hsp70 inhibitors (30).
Further analysis revealed that E175S Hsc70 was not able to stimulate ATP hydrolysis irrespective of which DnaJ was used to stimulate the reaction ( Fig. 2A). Then, using fluorescence polarization (FP), we found WT and E175S Hsc70 bound to the ATP-FAM reporter with similar K d ranges (410 nM Ϯ 20 nM for WT Hsc70 and 430 nM Ϯ 40 nM for E175S Hsc70, Fig. 2B) eliminating any concerns that E175S Hsc70 was unable to hydrolyze ATP due to a deficiency in nucleotide binding. E175S Hsc70 had slightly reduced affinity for ADP compared with WT Hsc70 in nucleotide competition experiments but was otherwise normal (Fig. 2B). E175S Hsc70 and WT Hsc70 both bound to BAG1, the nucleotide exchange factor similarly, suggesting that the E175S mutation does not interrupt nucleotide exchange factor interactions (Fig. 2, C and D). Similar to what was observed with ATP binding, E175S Hsc70 could not refold luciferase regardless of which human DnaJ was present (Fig.  2E). Together, these data indicate that E175S Hsc70 is unable to cycle ATP or refold luciferase despite appropriately interacting with its co-factors, which suggests this variant may have dominant negative (DN) characteristics.
To further determine if E175S Hsc70 was in fact acting as a DN variant, which would almost exactly mimic isoform-selec-tive inhibition (28), we sought to confirm that this variant could still bind substrate using FP assays with the tracer HLA-FAM (46). In these experiments we used Hsp72 as a comparison for E175S because apo-Hsp72 has superior FP signal to apo-WT Hsc70 for reasons that are not clear (  ; Fig. 2F). These data indicated that E175S Hsc70 can bind substrate, but it is locked in an ADP-like conformation with high affinity for substrates. These data provide further evidence that E175S Hsc70 is a DN form of this protein, with properties that are identical to chemically inhibited Hsc70.
Structural Characterization of E175S Hsc70 -Previous work suggests that inhibiting Hsc70 with a ligand altered the allosteric communication between the SBD and NBD of this protein (33,47,48). With this in mind, we performed 900-MHz TROSY NMR experiments comparing the ATP-and ADP-bound NBD (amino acids 1-386) of WT Hsc70 (red, blue) and ATP-and ADP-bound E175S Hsc70 (orange, cyan) (Fig. 3A) to assess the structural ramifications of the E175S mutation. Although both NBDs were globally well folded, a number of resonances were missing or shifted in the E175S mutant, indicating that it had undergone some structural rearrangement compared with WT Hsc70 (Fig. 3B). This effect was likely due to millisecond conformational exchange dynamics due to increased flexibility. Furthermore, these data confirm that E175S Hsc70 does undergo conformational change upon nucleotide binding, proving the mutant is still capable of binding nucleotide (Fig.  3C), but there are no significant structural differences between the ATP and ADP state of E175S Hsc70 1-386 (Fig. 3D), in contrast to the conformational changes associated with the ATP versus ADP state of WT Hsc70 1-386 (Fig. 3, E and F). When the chemical shifts were mapped on the crystal structure of Hsc70 NBD (Fig. 3G), they clustered to subdomains IA and IIA adjacent to the mutation. In addition, some chemical shifts were also seen in subdomains IB and IIB, suggesting that the interfaces between subdomains may have been affected. However, despite these structural changes, E175S Hsc70 still functionally resembled a WT Hsc70 NBD, able to bind nucleotides and nucleotide exchange factors.
Based on these results, we sought to determine if the structural ramifications of the E175S Hsc70 variant were sufficient to prevent the allosteric communication between the NBD and SBD, perhaps explaining its functional profile. To test this idea, we turned to one-dimensional 1 H-detected HSQC experiments. Specifically, we measured 15 N spin relaxation rates for the amide-proton signals of Hsc70 (Fig. 4) in different nucleotide-bound states (49,50) in the presence of a Tau peptide. 15 N relaxation rates provide insight into overall protein tumbling, which is influenced by shape and effective molecular weight. From this experiment, three regions were identified; the region of 1 H Ͼ 8.6 ppm corresponds to structured protein, the region 8.6 Ͼ 1 H Ͼ 7.6 ppm is dominated by signals of the unstructured areas, and the region 1 H Ͻ 7.6 ppm contains residual signals from structured side chains. As expected, the R 1 relaxation of the structured area was much slower than that of the unstructured area, corresponding to a larger rotational correlation time and a larger apparent molecular weight. For our analysis, we integrated the area 9.5 Ͼ 1 H Ͼ 8.7 ppm only and fitted the decay to single exponential. Error estimation was carried out by a jackknife procedure (41). This approach revealed that there is a large increase in apparent molecular weight for WT Hsc70 in the ATP-bound state (Fig. 4). The conversion used to translate rotational correlation time into molecular weight has been validated for spherical proteins; therefore, the large value seen for the ATP state may be due to the irregular shape of Hsc70. This result suggested that the NBD and SBD are not docked in the ADP-bound state and become docked to each other in the ATP-bound state, as inferred from studies with DnaK, the bacterial Hsc70 ortholog. When we examined E175S Hsc70 using the same methodology, the hydrodynamic properties when bound to ADP were essentially identical to that of WT Hsc70 bound to ADP. However, there was a significantly smaller increase in apparent molecular weight seen in ATP-bound E175S Hsc70 compared with ATP-bound WT Hsc70 (Fig. 4), indicating that the conformation of E175S Hsc70 favors the ADP-bound form regardless of which nucleotide is present. These findings strongly show that the E175S mutation causes Hsc70 to take on DN characteristics; this variant is unable to cycle ATP or refold luciferase despite binding cofactors and substrates appropriately, but it remains permanently in an ADP-bound conformation that has a high affinity for substrate. Thus, the Glu-175 Hsc70 variant appears nearly identical to a chemically inhibited Hsc70 (30), allowing us to test the physiological consequences of Hsc70 selective inhibition in cells.
Mechanism of Selectively Inhibited Hsc70-mediated Tau Clearance-We utilized this new DN-Hsc70 tool to determine how Hsc70 activity affects Tau metabolism and function in a neuronal context. First, we confirmed that in HEK293T cells overexpressing Tau, shRNA-mediated knockdown of Hsc70 clearly reduced Tau levels, as demonstrated previously (28) (Fig. 5A). Thus, specific inhibition of Hsc70 promotes Tau clearance. Next, utilizing confocal microscopy of HEK293T cells overexpressing WT or DN-Hsc70-FLAG and RFP-Tau, we found Tau levels were increased in cells expressing WT Hsc70-FLAG and significantly reduced in cells expressing the DN Hsc70-FLAG (Fig. 5B). These data confirmed that DN-Hsc70 was not just altering the localization of Tau but was actually facilitating its clearance. Western blot analyses of cells overexpressing WT Hsc70 or DN Hsc70 and Tau confirmed that Tau levels were significantly decreased in the DN-Hsc70-FLAG   (41)) for full-length WT Hsc70 and E175S Hsc70 are shown below the spectra. cells (Fig. 5C). These effects appeared selective for Tau, as the expression levels of another microtubule-associated protein (MAP), MAP2, were not significantly changed by either WT or DN-Hsc70 overexpression (Fig. 5D). We then designed AAV (serotype 9) constructs of WT and DN-Hsc70 to test how Hsc70 activity affected Tau turnover in brain tissue. Hippocampal organotypic slices from FVB-Tg(tetO-MAPT*P301L) mice that produce pathologic mutant human Tau were transduced with GFP AAV9, WT Hsc70-FLAG AAV9, or DN-Hsc70-FLAG AAV9. Confocal microscopy analysis revealed that WT Hsc70 overexpression slightly preserved pathological perikaryal Tau levels above that of GFP in neurons, whereas DN-Hsc70 significantly reduced pathological P301L Tau levels compared with both WT Hsc70 and GFP (Fig. 6).
Because pan-Hsp70 inhibitors triage Tau to Hsp90 for proteasomal degradation (30,51), we speculated that DN-Hsc70 was similarly triaging Tau. Indeed, Tau immunoprecipitated from cells overexpressing DN-Hsc70 is associated with more Hsp90 compared with WT Hsc70 (Fig. 7A). We then confirmed that DN-Hsc70 is triaging Tau by the proteasome by treating cells overexpressing WT or DN-Hsc70 with the proteasome inhibitor epoxomicin for 6 h. Tau degradation caused by DN-Hsc70 was completely blocked by epoxomicin (Fig. 7B). Conversely, Tau degradation by DN-Hsc70 did not occur through autophagy, as siRNA targeting the autophagy mediator, Beclin-1 (Fig. 7C), did not block Tau reductions by DN-Hsc70. Thus, in the neuronal and cellular context, isoform-selective inhibition of Hsc70 causes similar effects to that of small molecule pan-Hsp70 inhibitors with regard to Tau biology (30).

DN-Hsc70
Has an Altered Co-chaperone Complement-We then sought to determine a putative cellular mechanism for this dramatic shift in Hsc70 function caused by the E175S mutation. Given that DN-Hsc70 efficiently degraded Tau in cells compared with WT Hsc70, we speculated that the chaperone interactome (chaperome) bound to DN-Hsc70 was distinct from WT Hsc70. WT or DN-Hsc70 were co-immunoprecipitated from HEK cell lysates and analyzed by HPLC-MS/MS to identify the most robust interactors of each variant. As predicted, these interacting partners were predominantly co-chaperones and other chaperones. Co-chaperones such as CHIP, HIP, and HOP bound either variant similarly. Interestingly, WT Hsc70 tended toward increased interaction with other Hsp70 isoforms, whereas DN-Hsc70 had increased association with several DnaJ proteins and Hsp90 isoforms (Fig. 8A). We confirmed  these interactions by immunoprecipitating WT or E175S Hsc70 and probing for levels of bound DnaJA1 and DnaJC7 (Fig. 8B). Thus, the DN-Hsc70 variant, which is locked into an ADP-bound conformation, prevents the dissociation of DnaJs and abrogates the interaction with other Hsp70 family members, promoting the recruitment of Hsp90 to the complex to facilitate Tau degradation.

Discussion
Here we sought to determine the impact of selective Hsc70 inhibition on the stability of the microtubule associated protein Tau. Chaperone modulation holds great promise as a potential therapeutic for tauopathies and can allow us to better understand the biology of Tau as it relates to both its normal and pathogenic functions. Hsc70 in particular seems to be a key regulator of Tau biology in the brain, warranting the pursuit of strategies aimed at disrupting this chaperone/client interface. In this study, using genetic and biochemical approaches, we determined that isoform-selective inhibition of the most abundant Hsp70 family member, Hsc70, can dramatically enhance Tau clearance, similar to what has previously been observed with small molecule pan Hsp70 inhibitors. Almost identically to these inhibitors, the E175S Hsc70 variant took on DN properties when overexpressed in cells that led to Tau degradation through the proteasome by recruiting Hsp90 proteins (52).
The E175S Hsc70 had nearly normal affinity for nucleotides and co-chaperones yet lacked essential ATPase and luciferase refolding activities, features that were previously unknown for this variant with regard to human Hsc70 (43,44,53,54). Human Hsc70 is also thought to undergo major conformational changes within its subdomains in response to co-chaperone binding and nucleotide hydrolysis in a manner similar to Hsp70 proteins from other species (48,(55)(56)(57)(58)(59)(60), but this also remained unknown until the NMR analyses in these studies. Differences in several resonances in subdomains IA and IIA caused by the E175S Hsc70 confirm that this mutation alters the allosteric communication between the SBD and NBD of Hsc70, dramatically impacting its function. This is consistent with previous work showing that the cleft between these subdomains is important for relaying initial NBD conformational changes to the SBD in bacterial Hsp70, DnaK (48,58,60,61). Interestingly, NMR of full-length E175S Hsc70 also showed that the ADPbound conformation was essentially identical to that of WT Hsc70; however, E175S Hsc70 was unresponsive to the conformational changes typically caused by ATP/ADP exchange as demonstrated by 15 N relaxation time measurements. Thus, the E175S mutation in Hsc70 locked this chaperone into an ADPbound state regardless of which nucleotide was bound within the NBD, again almost perfectly mimicking Hsc70 in the presence of a small molecule pan Hsp70 allosteric modulator.
This tool also allowed us to shed new light on how ATPase chaperone scaffolds like Hsc70 make triage decisions (62)(63)(64)(65). Our LC-MS/MS results indicate that chaperone/chaperone and chaperone/co-chaperone dynamics change based on nucleotide exchange, providing deeper insight into how client fate is decided. Hsc70 that is locked into an ADP-bound conformation prevents DnaJ proteins from cycling off, recruiting Hsp90 proteins to promote client clearance. Our MS results also unexpectedly suggested an important role for Hsc70 nucleotide exchange in microtubule dynamics as a number of tubulin proteins had higher association with DN-Hsc70 relative to WT. This not only provides even greater evidence for a functional link between the microtubule-associated protein Tau and Hsc70, as previously suggested (20), but it also suggests that a major role of Hsc70 could be regulation of microtubule dynamics. Hsc70 has been associated with functions that require microtubules, such as spindle formation (66) and regulation of axonal transport along microtubules (67, 68). Our results here show that these functions are likely critically linked to Hsc70 activity. Another interesting and unexpected result from these MS studies was the finding that WT Hsc70 interacted with other Hsp70 isoforms in the cell, but in the absence of allosteric changes caused by nucleotide exchange these interactions were prevented, as demonstrated by the results with E175S Hsc70. Perhaps nucleotide-induced conformational changes in Hsp70 proteins are essential for intracompartmental trafficking of properly folded proteins in cells. Regardless, these results suggest that intradomain allosteric changes are essential for Hsc70 to interact with other Hsp70 isoforms to coordinate proteostasis, an area that has yet to be explored in depth in the chaperone field.
With regard to Tau pathobiology, our results here prove what we previously could only speculate: that Hsc70 is an excel-FIGURE 8. Mass spec reveals a changing chaperome for DN Hsc70 compared with WT Hsc70. A, average spectral counts plotted as percentages to illustrate whether a protein was enriched in WT (cyan) or E175S Hsc70 (blue). Data are the averages of three independent mass spectroscopy experiments. Results that were Ͻ10 spectral counts in either sample were discarded to ensure robustness of the results. B, confirmation of mass spectroscopy results; representative Western blot images of Tau immunoprecipitated (IP) from cells overexpressing Hsc70 WT Hsc70 or E175S Hsc70 and probed for DnaJA1 and DnaJC7 is shown. lent therapeutic target for tauopathies with regard to efficacy and perhaps even specificity. We previously showed that Hsc70 preserved Tau levels in cells and that the aged human brain had high levels of Hsc70 (20,28), but it remained unknown if Hsc70 could be targeted to promote Tau clearance in neurons and whether such a strategy was practical given the important role of Hsc70 in a number of cellular processes. By identifying and using the E175S Hsc70 variant, we were able to address both of these issues to some extent. First, overexpression of the E175S Hsc70 variant was not fatal to cells or neurons, suggesting that some perturbation of Hsc70 may be permissible, at least acutely. Second, we prove for the first time in neurons that targeting the activity of a single Hsp70 isoform, Hsc70, is a very potent anti-Tau strategy. Inhibitors that generically target all isoforms of Hsp70, while effective at lowering Tau (29,30), will undoubtedly have more untoward consequences in neurons compared with isoform-selective inhibitors. This is because several Hsp70 isoforms are critical for basic cellular function. For example, targeting BiP/Grp78 would disrupt endoplasmic reticulum protein quality control and induce the unfolded protein response (69,70). Targeting mortalin would certainly impact mitochondrial function and lead to neuronal death (71). And although targeting Hsc70 may also be deleterious given its important role in cellular homeostasis, our results suggest that there is some level of flexibility built into the proteostasis system that could allow for a chemotherapeutic approach targeting this Hsp70 isoform to treat tauopathies and possibly other neurodegenerative diseases. Therefore, the design and development of Hsp70 isoform-selective inhibitors will likely be essential for rationally targeting these proteins for therapeutic purposes in the future, and although challenging due to structural similarities between Hsp70 isoforms, it is possible. An Hsp72selective inhibitor was recently developed and showed good efficacy for treating cancer with fewer side effects (72). Development of Hsc70 selective inhibitors could hold similar promise for tauopathies.
In conclusion, using a DN-Hsc70, we have proven that targeting this Hsp70 isoform contributes to the majority of anti-Tau activity seen with small molecule pan Hsp70 inhibitors. This DN-Hsc70 causes Tau degradation through the same Hsp90/proteasome mechanism as that engaged by such Hsp70 inhibitors (30). DN-Hsc70 lacks the conformational dynamics induced by nucleotide exchange in Hsc70 protein, locking Hsc70 into a high affinity state for substrates that promotes client degradation. Client degradation appears critically linked to fluidity in the DnaJ-Hsc70 complex, such that if DnaJ proteins are unable to decouple from Hsc70, Hsp90 is recruited to this complex to promote client turnover. Our data show that selective inhibition of Hsc70 activity could be tractable for treating tauopathies and elucidating the mechanisms of Tau metabolism as well as the metabolism of other clients of this important chaperone protein.