Heat shock promotes inclusion body formation of mutant huntingtin (mHtt) and alleviates mHtt-induced transcription factor dysfunction

PolyQ-expanded huntingtin (mHtt) variants form aggregates, termed inclusion bodies (IBs), in individuals with and models of Huntington's disease (HD). The role of IB versus diffusible mHtt in neurotoxicity remains unclear. Using a ponasterone (PA)-inducible cell model of HD, here we evaluated the effects of heat shock on the appearance and functional outcome of Htt103QExon1–EGFP expression. Quantitative image analysis indicated that 80–90% of this mHtt protein initially appears as “diffuse” signals in the cytosol, with IBs forming at high mHtt expression. A 2-h heat shock during the PA induction reduced the diffuse signal, but greatly increased mHtt IB formation in both cytosol and nucleus. Dose- and time-dependent mHtt expression suggested that nucleated polymerization drives IB formation. RNA-mediated knockdown of heat shock protein 70 (HSP70) and heat shock cognate 70 protein (HSC70) provided evidence for their involvement in promoting diffuse mHtt to form IBs. Reporter gene assays assessing the impacts of diffuse versus IB mHtt showed concordance of diffuse mHtt expression with the repression of heat shock factor 1, cAMP-responsive element-binding protein (CREB), and NF-κB activity. CREB repression was reversed by heat shock coinciding with mHtt IB formation. In an embryonic striatal neuron–derived HD model, the chemical chaperone sorbitol similarly promoted the structuring of diffuse mHtt into IBs and supported cell survival under stress. Our results provide evidence that mHtt IB formation is a chaperone-supported cellular coping mechanism that depletes diffusible mHtt conformers, alleviates transcription factor dysfunction, and promotes neuron survival.

Protein misfolding and aggregation are hallmarks and contributing pathogenic mechanisms of many neurodegenerative diseases (ND) 6 (1)(2)(3)(4). Causative mutations identified in familial ND result in the aggregation of disease-related proteins: polyQhuntingtin (mHtt) in Huntington's disease, ␣-synuclein in Parkinson's disease, ␤-amyloid in Alzheimer's disease, and prion in Creutzfeldt-Jakob disease, as examples. Although we do not know how protein aggregation may lead to neurodegeneration, the well-known association has spawned therapeutic strategies and extensive research efforts aimed at preventing and clearing such aggregated protein deposits (5).
Age is the primary risk factor in the development of ND, heritable familial cases included. That an individual harboring a mutation appears normal for decades before succumbing in mid-/later life to the pathogenic effects of the disease is confounding. In this context, our observation of an age-dependent and neuron-specific dysfunction of HSF1 for the induction of HSP chaperones is significant and relevant (6 -10). HSF1 is a master transcription regulator that links stress biology to the induction of a network of HSP chaperones that collectively function as a protein quality control (PQC) mechanism to promote protein homeostasis (11,12). The importance of PQC is underscored by the fact that mutations in HSP70 co-chaperones have been identified as causing inherited neurodegenerative and cardiac disorders, directly linking the HSP70 chaperone system to human disease (13).
Accordingly, we reason that conditions that enhance this PQC would be beneficial to counter pathologies caused by dysregulation of protein folding and aggregation in aging and ND (3, 14 -19). Using an inducible cell model of HD (the 14A2.6 line of PC12 cells) (20), we evaluated the effects of heat shock and the role of HSP70 and HSC70 chaperone on the structuring and functional readout of Htt103Q Exon1 -EGFP (mHtt) chimera protein expression. A transient and modest heat shock (HS) at 42°C in the course of mHtt expression had a robust and reproducible effect in promoting the formation of large cytosolic inclusion bodies (IBs) and small nuclear IBs at the expense of diffusible assemblies of mHtt. Functionally, accumulation of diffuse mHtt repressed proportionately the activities of HSF1, CREB, and NF-B as demonstrated by reporter gene analysis. The repressed CRE-reporter was reversed by HS and coincided with mHtt IB formation. Together, these results suggest that IB formation likely represents a cellular coping mechanism to rid itself of toxic mHtt conformers. This study adds to our understanding of the role of HSPs in protein homeostasis that, in addition to promoting protein folding, disaggregation, solubilization, and degradation, HSP chaperones can drive the structuring and aggregation of the intrinsically disordered mHtt protein into forming IBs for beneficial outcomes.

Effect of heat shock on mHtt expression profile
We used the 14A2.6 line of PC12 cells with stably integrated ecdysone receptor-based inducible expression of the Htt103Q Exon1 -EGFP chimera (mHtt) as our HD cell model (20,21). PC12 cells harboring the normal Htt construct Htt25Q-EGFP (cHtt) served as control (20). We note that transgenic mice expressing the human mHtt Exon1 exhibit a progressive neurological phenotype that exhibits many of the motor and nonmotor features of HD (22,23). Fig. 1 presents images of the control and mutant Htt cells at 48 h of induction with 5 M of ponasterone (PA), an analogue of ecdysone. To test for the effects of heat shock, cells were shifted to 42°C for 2 h at the 24-h time point, returned to 37°C and harvested and fixed at 48 h. Expression of HSP70 was probed with a rabbit polyclonal antibody against HSP70 followed by Alexa Fluor 594 goat antirabbit IgG, and nuclei were stained with Hoechst 33342. The three sets of mHtt images (Fig. 1, C and D, E and F, and G and H) were from three separate experiments and are representative of many experiments done at different times in the course of this study. Results show a robust expression of both the cHtt-and mHtt-EGFP chimera protein at the 48-h time point of PA induction, with diffusible assemblies of cytosolic Htt protein (Fig. 1, A-H, black arrows with white outline) predominant in both. In the mHtt cells, we could see occasional large cytoplasmic IBs in cells maintained at 37°C (Fig. 1, C, E, and G; white arrows). HS decreased the diffuse cytosolic mHtt signal and conversely increased the number of IBs in both the cytosol and Briefly, heat shock (HS) was at the 24-h time point by placing designated wells/strips of cells in a 42°C incubator for 2 h and then returned to 37°C incubator. Cells were harvested and fixed at 48 h. To track the expression of HSP70 as a response to heat shock, we used a rabbit polyclonal antibody against HSP70 followed by Alexa Fluor 594 -labeled goat anti-rabbit IgG secondary antibody. Nuclei were stained with Hoechst 33342. Cells were imaged with EVOS ® FL microscope with a 10ϫ objective; all images of a given experiment were captured at the identical settings of exposure time and light intensity/contrast. A-H, cropped and contrasted areas of representative images of the 25QHtt and 103QHtt cells under control and heat shock conditions as indicated. Each of the image frames represents an area of ϳ200 ϫ 266 m. Black arrows point to diffuse Htt-EGFP signal; white arrows point to IB; yellow arrows point to punctate nuclear IB. No IB was detected in the Htt25Q PC12 cells. I, mHtt quantitation. Averaged intensities per unit cell of total, diffuse, and IB Htt-EGFP of images from five independent experiments were done over a 2-year period with 25 independent cell samples per condition. Results shown here are relative to that of the control total of 100%. J, scatter plot of the HSP70 (red) versus Htt103Q-EGFP (green) signal of control versus HS cells. Individual cells of representative images of control (black symbol) and heat shock cells (red symbol; the two red lines represent results of two separate images) were marked as region of interest (ROI). The red (HSP70) and green (mHtt) signals of each ROI were determined by ImageJ and displayed as scatter plots. Probability of difference p Ͼ0.05 is defined as not significant, between 0.01 and 0.05 is significant (*), Ͻ0.01 is very significant (**), and Ͻ0.001 is extremely significant (***). A-H, full-frame original images are included in Fig. S1, A-H. Error bars represent S.D.
Heat shock for mHtt IB formation and TF de-repression nucleus ( Fig. 1, D, F, and H) as compared with their control counterparts ( Fig. 1, C, E, and G). The cytosolic IBs were substantial in size, with some approaching the size of the 10 -12 m cell nuclei. HS also had a striking effect in increasing the number of small nuclear IBs. It should be noted that IB formation depends on polyQ expansion of the mHtt protein as this was not observed with the 25Q cHtt protein. The full-size original images presented in Fig. 1 are included as Fig. S1, A-H.
To quantify these changes in the expression and distribution of mHtt, we created "Macro" programs to measure the diffuse and IB Htt-EGFP intensity and of HSP70, and score the number and size of nuclei and IBs in ImageJ Fiji. Notably, the term diffuse is operationally defined as the microscopic diffuse appearance of the Htt protein; such diffusible assemblies can include disordered and meta-stable monomers and oligomers and low order fibrillary forms of Htt. For all IBs of varying sizes, a scatter plot of size versus green fluorescence intensity yielded a straight line with RSQ (R 2 ) value of 0.9991 suggesting the same mHtt packing density in IBs, size notwithstanding. Diffuse and IB mHtt had significantly different mean fluorescent intensities, a result consistent with previous observations (24). This difference enabled us to analyze and score the diffuse mHtt signal separate from the IB mHtt signal. Macro programming greatly facilitated analysis of images of cells plated in 96-well plates. The Macro programs used for scoring the diffuse and IB Htt-EGFP signals and for profiling the number and size of IBs are included in Fig. S2.
The bar graph in of Fig. 1I presents our results on Htt signal quantitation. For mHtt cells at the 48-h time point of induction with 5 M PA, the diffuse and IB forms of mHtt constituted on average 80 and 20% of the total signal intensity, respectively. HS of cells reduced the total mHtt load by about 20% and changed the relative abundance of diffuse versus IB forms of mHtt to constitute, respectively, 60 and 40% of the total. In contrast, the cHtt (25Q) protein never formed IB regardless of the level of cHtt expression, and HS reduced the total cHtt intensity by about 20%, similar to that of the mHtt cells. In sum, heat shock had two specific and notable effects on mHtt: 1) it depleted the diffuse cytosolic mHtt to form large cytosolic IBs, and 2) it greatly increased the appearance of small nuclear IBs. A scatter plot scoring individual cells for diffuse mHtt versus HSP70 (Green versus Red) in Fig. 1J shows that HS increased the slope of this scattered plot over the control. To affirm that the small, punctate-appearing IB in nuclei are indeed of nuclear localization, we imaged cells by confocal microscopy. Confocal images in Fig. S3 show that nuclear IB and the Hoechst 33342-stained nucleus are in the same focal plane.
To better define conditions necessary for IB formation, we assessed their appearance vis à vis the dose of PA as well as the time to induce mHtt expression. Under control conditions and with a lower PA concentration (e.g. Ͻ1.25 M) ( Fig . Quantitation of IB number and size of the control and HS cells showed that HS increased the number of IB per cell but reduced slightly the averaged IB size because of the increase in small nuclear IBs (Fig. 2D). These results are consistent with a nucleated polymerization process in IB formation as proposed (23).
To quantitate the effects of HS on IB formation, we scored the numbers as well as size distribution of IBs of control and HS cells. For this, we selected six representative images of control and HS cells (42°C for 2 h; 1ϫ at 24 h; 2ϫ at 24 and 30 h) at the 48-h time point of induction with PA. The results in Fig. 3A of IB size distribution showed that IBs ranged in size from Ͻ1 m to Ͼ10 m. Heat shock of cells resulted in a significant increase in the number and integrated intensity of IBs. Profiling of IB in  Fig. 3B showed that there were two populations of IBs with averaged size of ϳ2 and ϳ6 m in diameter, principally representing nuclear and cytosolic IBs, respectively. In the visual images, the punctate-appearing nuclear IB was a notable feature of the HS cells. Fig. 3C summarizes the quantitative results of IB in number per cell and averaged size. HS of cells resulted in statistically significant ϳ2 to 3ϫ increases in the total as well as nuclear IB number per unit cell. The averaged IB size of the HS cells was statistically smaller than the controls in large part because of the marked increase in small nuclear IB population. Because of their substantially smaller size, the nuclear IB constituted a small percentage of the total IB integrated intensity.

Effects of HSP70 and HSC70 knockdown on the restructuring of mHtt
Heat shock elicits the activation of HSF1 for the induction of HSP chaperones such as HSP90, 70, 40, and others in various subcellular compartments. Collectively, the chaperones promote the folding/refolding, disaggregation, and disposition of cellular proteins for recovery after stress (19). HSP90 binds many meta-stable client proteins including the normal and the polyQ-expanded Htt (25,26), whereas HSP70 primarily functions to chaperone exposed hydrophobic patches of nascent and nonnative proteins (27). Several of the HSP chaperones work in tandem and in cooperation for the folding versus degradation of substrate proteins (28). To probe the contribution of individual HSPs in the HS-induced dynamic change of mHtt, we asked if knocking down of HSP70 or HSC70 affected the restructuring of mHtt under either control or heat shock condition. For HSP70 KD, cells were first transfected with control and HSP70 DsiRNA (Dicer-substrate siRNA, a mixture of three Hspa1a siRNAs, Integrated DNA Technologies) and then processed for mHtt induction by PA, heat shock, and image analysis. Representative images of control and HSP70 DsiRNA-transfected cells kept under a basal versus HS condition in Fig. 4A clearly showed that HSP70 DsiRNA blunted the heat shock induction of HSP70 (Fig. 4A, red). Quantitation of the total, diffuse, and IB mHtt intensity in the control versus HSP70 DsiRNA-transfected cells showed that knocking down HSP70 reduced mHtt IB formation under both control and HS conditions, with the reduction being of greater magnitude and statistical significance in the HS cells over the controls (Fig. 4C).
The heat-inducible HSP70 (HSPA1A) is a member of the HSP70 multigene family that includes the constitutively expressed HSC70 protein (HSPA8) involved in processes such as folding and trafficking of nascent polypeptide under normal conditions (29 -31). To evaluate specificity of effects among members of the HSP70 protein family in the remodeling of mHtt, we determined the effects of knocking down HSC70. Representative images of control versus HSC70 KD cells under control and heat shock conditions in Fig. 4B show that HSC70 is constitutively expressed at 37°C, and heat shock at 42°C for 2 h gave a modest increase (Ͻ2 times) in this expression, a result consistent with previous observations (30,31). Quantitation across multiple images of the total, diffuse, and IB mHtt intensity per unit cell in Fig. 4D shows that HSC70 KD selectively affected the restructuring of diffuse mHtt into IB under the basal 37°C condition such that the IB signal decreased from ϳ20% of the total in the control to ϳ7% in the HSC70 KD cells. In HS cells, IB represented ϳ46 and ϳ32% of the total in control versus HSC70 KD cells, respectively. Together, the results in Fig. 4 show that KD of HSP70 selectively blunted the HS-induced restructuring of diffuse mHtt into forming IBs, whereas KD of HSC70 selectively suppressed IB formation at 37°C.

Heat shock for mHtt IB formation and TF de-repression mHtt expression and transcription factor dysregulation
Transcription dysregulation is an important pathogenic mechanism in HD, and major changes in gene expression have been detected in postmortem HD brains and experimental model systems (32)(33)(34). Given the well-documented role of HSF1 as a driver of the HSP-dependent PQC system and its dysfunction in many model systems of aging (6 -8, 11, 14, 17, 35), we asked if expression of mHtt may in turn affect the regulation and function of HSF1. For this, cells were first transfected with HSE-firefly luciferase DNA along with the internal control humanized Renilla luciferase DNA. Transfected cells were split into two halves, with and without 5 M PA to induce mHtt expression. HSE-reporter gene activity was tested 24 h after PA induction, a time when Ͼ80% of the mHtt appeared as diffusible assemblies, by heat shock of cells at 42°C for 2 h followed by recovery at 37°C for 4 h. Results in Fig. 5A show that expression of mHtt significantly repressed the heat-induced HSE-reporter gene activity. Fig. 5B shows the dose-response effect of PA on HSE-firefly luciferase reporter gene activity under control and heat shock-, MG132-, and arseniteinduced conditions. HS, MG132, and arsenite were all effective in inducing the HSE-firefly luciferase reporter, with HS being the most efficacious. For each of the inducers, there was a PA dose-dependent reduction in the HSE-reporter gene activity in proportion with the increased expression of diffuse mHtt. This observation is consistent with previous demonstrations of pro-

Heat shock for mHtt IB formation and TF de-repression
gressive disruption of the PQC machinery in models of polyglutamine diseases (36).
We then asked if other TF driven reporter genes may be similarly affected. In particular, we were interested in TFs implicated in memory formation and in stress response (37)(38)(39)(40)(41). Fig. 6A presents the result of PA dose-dependent effect on the basal and induced HSE-, CRE-, NF-B-, and SRE-firefly luciferase reporter gene activity. Results showed that in each case, PA gave a dose-dependent repression of the induced HSE-, CRE-, and NF-B-firefly luciferase reporter activity, whereas the serum-induced SRE-reporter gene activity was unaffected. This suppression of reporter gene activities correlates with expression of the diffuse mHtt protein as shown in fluorescent/phase images of cells from the same experiment in Fig. 6B. The basal reporter gene activity was low and it was difficult to gain statistically meaningful results of the effects of mHtt expression on the basal (uninduced) reporter gene activity. Parallel studies done with the 25Q PC12 cells did not show a PA dose-dependent effect on reporter gene activity, indicating a polyQ dependence and specificity in reporter gene repression. Expression of mHtt had little or no effect on the SREreporter gene under the same experimental condition. Our results are consistent with previous observation of the efficacy of polyQ and amyloid-like aggregates to sequester many metastable regulatory proteins, CREB, HSF1, and NF-B included (33,42). Together, our results in Figs. 5 and 6 show that increased expression of the diffusible assemblies of mHtt suppressed the transcriptional activity of HSF1, CREB, and NF-B, whereas SRE-promoter was unaffected.
To further define the functional readout of diffuse versus IB forms of mHtt in this paradigm of TF repression, we used a 2-h HS protocol to drive diffuse mHtt into forming IBs and tested the consequence of such manipulation on the CRE-promoter activity. The time line of the experiment is as shown in Fig. 7E. Representative images of cells at the 24-h and at the 48-h time point of PA induction, without and with pre-HS, are in Fig. 7, B-D, respectively. Results from the CRE-reporter gene assay (Fig. 7A) showed that PA-induced expression of mHtt repressed CRE-reporter gene activity at both the 24-and 48-h time points as compared with controls. Most importantly we showed that a 2-h HS of the PA-induced cells, but not of the control cells (without PA), at 24 h rescued the repressed CREreporter gene activity when measured at the 48-h time point (Fig. 7A). Collectively, these results demonstrate that heat shock, a condition that induces HSP chaperones to drive the structuring of diffuse mHtt into forming IBs, has beneficial consequences in the regulation and function of CREB. This also supports the notion that IB formation is an HSP-supported coping mechanism in the management and disposition of pathogenic proteins to alleviate TF dysfunction. Qualitatively similar result was observed with the NF-B-reporter gene. Because HS affects both the remodeling of mHtt and the HSE-reporter gene, it was difficult to tease out the cause and effect relationship, and the HSE-reporter gene was not included in this analysis.

Chemical chaperone promoted compaction of mHtt and survival of striatal neuron-derived cells
Striatal neurons are noted for their vulnerability in HD (43). To ascertain disease relevance of chaperone-dependent IB formation and the concomitant restoration of cell function, we used the ST14A line of cells derived from E14 rat striatum primordia by retroviral transduction of a temperature-sensitive SV40 large T antigen to analyze the processing and functional readout of mHtt expression (44,45). The temperature-sensitive growth phenotype of these cells (33°C permissive, 39°C restrictive) negated the use of HS to induce HSP chaperones for the structuring and compaction of mHtt. Instead, we used sorbitol as a chemical chaperone to promote protein structuring (46 -48). Cells were transfected with plasmid expression vectors of HttQ74 -EGFP and the effects of sorbitol on the structuring and compaction of mHtt and cell viability under stress

Heat shock for mHtt IB formation and TF de-repression
were determined. Representative images of control (Fig. 8, A  and B) and sorbitol-treated (150 mM) (Fig. 8, C and D) ST14A cells show that whereas HttQ74 -EGFP mostly had a diffuse appearance in control cells maintained at 33°C, in the sorbitoltreated cells HttQ74 -EGFP appeared primarily in the form of IBs. Shifting the ST14A cells from 33°C to 39°C under a serumdeprived condition resulted in a ϳ60% decrease in cell viability at 48 h after the shift (44), whereas for the sorbitol-treated cells the decrease was ϳ45%. Sorbitol had little effect on survival of the untransfected ST14A cells or cells transfected with the HttQ23-EGFP DNA. Together, the results in Fig. 8 show that sorbitol promotes the structuring and compaction of mHtt into forming IB and confers a statistically significant survival advantage to the ST14A striatal neurons under stress.

Discussion
The protein deposits found in ND, including Huntington's, Alzheimer's, Parkinson's, and prion diseases, are formed of completely unrelated proteins. The presence of such protein aggregates is often the most distinct postmortem histological signs of ND, and it was hypothesized that protein aggregation is toxic to neurons and that therapeutic strategies aimed at clearing such aggregated protein deposits would be beneficial. Over time, it has become clear that the correlation of the location of protein deposits and brain regions of neuron death is weak at best (49). This and other considerations raise important questions regarding the identity of pathogenic protein conformer(s) and the cause of neuron death in ND (49 -52).
Our observation that HS has a timely and reproducible effect in promoting mHtt IB formation is in general agreement with that of previous studies (53,54), although the tenor, emphasis, and conclusion of the studies are different. In this research, we looked at changes of both diffuse and IB forms of mHtt and arrived at the conclusion that HS through induction of HSP chaperones promotes the structuring and compaction of diffuse mHtt into forming IBs with beneficial functional outcomes. Although our standard experimental protocol was to HS cells at the 24-h time point in the course of PA induction, the effective window for HS to drive the depletion of diffusible assemblies of mHtt to form IBs extended from 6 to 30 h of PA induction. Earlier than 6 h and later than 30 h, the results became less clear-cut. The significance of the size difference between cytoplasmic and nuclear IBs is not entirely clear. One possibility is a difference in mHtt concentration in the subcellular compartments. Alternatively, the presence of different HSPs and their stoichiometry in the nuclear versus cytoplasmic compartment may impart a different remodeling outcome. mHtt and other ND-associated proteins belongs to a family of intrinsically disordered proteins (IDPs), proteins without a defined stably folded structure and with abundant interacting surfaces in their native states (55)(56)(57). Their aggregation begins with the meta-stable, intrinsically disordered conformers and ends up as IBs in Huntington's disease, ␤-amyloid deposits in Alzheimer's disease, and Lewy bodies in Parkinson's disease, as examples. There is a growing body of evidence that conditions that promote protein structuring guide the assembly and aggregation of IDPs, including ␣-synuclein, prion, and tau (48,58,59). Consistent with this concept, we hereby show that induction of HSP chaperones by HS greatly enhances the assembly of diffuse mHtt into forming IBs and KD of HSP70 expression blunted this process. Our observation that KD of HSC70 selectively reduced the basal rate of IB formation is consistent with the acknowledged role of HSC70 in protein folding and trafficking under normal growth conditions (29) and lends additional support to the role of protein structuring in mHtt IB formation.
The assembly of the disease protein monomer and oligomers into visually distinctive fibers and IBs is a multistep process

Heat shock for mHtt IB formation and TF de-repression
with many ill-defined intermediates including unstructured andquasi-structuredmonomerandoligomers,higherorderoligomers, protofibrils, and fibril bundles that may dynamically interconvert (60,61). It is unclear which one or combinations of these may be the toxic entity that predisposes neurons to their eventual demise. An emerging concept, as exemplified by the ␤-amyloid protein of Alzheimer's disease, is that small intermediates including soluble oligomers adversely affect synaptic structure and plasticity (62). Studies in other ND suggest that a broadly similar process of neuronal dysfunction is caused by diffusible monomers and oligomers of disease-causing proteins (54,(62)(63)(64)(65)(66). In the case of HD, longitudinal tracking of individual neuron fate showed that the presence of diffuse mHtt is a prognostic indicator of neuronal death whereas IB formation is correlated with neuron survival (24,67,69,70). Further, studies using an epitope-specific mAb 3B5H10 suggest that monomer or small oligomer of mHtt as opposed to higher order aggregate can best predict neuron death (71). Consistent with these observations, we showed using the ST14A embryonic striatal neuron-derived cells that an osmolyte sorbitol depleted diffuse mHtt and promoted their structuring and compaction into forming IB to support neuron survival under stress.
Identification of mHtt conformers that are directly or indirectly neurotoxic is a challenging proteomics problem with significant implications in framing therapeutics development for ND. To relate protein aggregation to disease pathogenesis, it is necessary to take into consideration the cellular components affecting protein aggregation, such as chaperones, proteases, as well as molecular crowding (72). In vitro analysis may not mimic scenarios in cells; reasons include the following: 1) intermediate conformers are likely to be meta-stable, 2) cell lysis and fractionation affect protein concentration which is a primary determinant in aggregation (73), and 3) salts and/or detergents used in the in vitro analysis would change the distribution and solubility of protein conformers (74). In this research, we relied principally on image analysis of live as well as fixed cells to track the expression and progression of diffuse versus IB forms of mHtt.
An important cause of the neurotoxicity of disease proteins implicated in ND is the anomalous interactions with many metastable regulatory proteins causing multifactorial toxicity and collapse of essential cellular functions (33,75). In HD, there is widespread transcription dysfunction (32)(33)(34). Herein, we show that although HSF1-mediated induction of HSPs guides diffuse mHtt to form IBs and improves cell fitness, there is a negative feedback loop, namely, the expression of diffuse mHtt limits and constrains the activity of HSF1. Our observation is in agreement with previous observations which show that the expression of model ␤-sheet proteins leads to deficiency of the normal cytosolic stress response mediated by HSF1 and NF-B (33), and of an impaired heat shock response in mHtt expressing cells (38,40,76). Likewise, a deficiency in striatal neuron proteostasis machinery likely contributes to the select demise of these neurons in HD (77).
The repression of TFs by diffusible assemblies of mHtt as a pathogenic mechanism is not limited to HSF1. We showed that the CREB-and NF-Bdriven reporter genes are similarly repressed. CREB is a TF involved in a number of processes required for memory formation (38,40). Given that ND often involves memory compromises, it is perhaps not surprising that disruption of CREB function leads to neurodegeneration (39), that inhibition of CREB reduces neuronal excitability and plasticity associated with ND (37), and that CREB is down-regulated in Alzheimer's disease brains (79). Likewise, NF-B functions as a crucial TF for glia and neurons, and expression of mHtt down-regulates NF-B signaling (41,80).
Our observation that a transient heat shock can reverse the effect of mHtt in secluding TFs has three implications: 1) IB formation is protective and represents a chaperone-supported coping mechanism; 2) part of the cyto-protective activity of heat shock is because of the reversal of TF seclusion such as HSF1, CREB, and NF-B; and 3) for PQC purposes, the HSP chaperone network can help either by promoting the proper folding/refolding and degradation of nonnative proteins or by structuring and aggregating toxic protein conformers into IBs. Our observation is consistent with previous result on the yeast prion protein [RNQ ϩ ] in which the HSP40 chaperone protein Sis1 stimulated amyloid assembly of the otherwise toxic protein in yeast (63). These and other considerations (81, 82) suggest that chaperone-supported structuring and compaction of meta-stable IDPs can enhance cellular fitness. We do not know whether it is the remodeling of mHtt in the cytosol or the nucleus to form IB that account for the improved TF readout. HSF1 and NF-B are located in the cytosol under normal and unstressed condition and become activated and shuttled into the nucleus upon stress (83). Teleologically, an increase in the presence of diffusible assemblies of mHtt in the cytosol can entrap HSF1 and NF-B to constrain their translocation into the cell nucleus. For CREB, a constitutive nuclear TF, it was shown that ␤-amyloid blocked the nuclear translocation of activated CREB (84). Indeed, aggregation-prone artificial ␤-sheet proteins and authentic disease proteins have been shown to cause a pronounced impairment of nucleocytoplasmic transport and a redistribution of nuclear shuttle factors to the cytosol (85,86).
We further note that independent of the mHtt-dependent seclusion of key TFs involved in cell fitness and survival, there is also age-dependent attenuation of various prosurvival and adaptive responses, a decreased activation of HSF1 in various aging models being a prime example (8,14). Together, these two changes likely would have a snowballing effect to progressively limit and constrain the HSF1-dependent PQC machinery leading eventually to PQC collapse. In the context of a therapeutics framework for HD, this would mean that an earlier intervention to enhance the chaperone network in the course of disease progression would have a significantly greater efficacy, whereas treatment effectiveness would greatly diminish as the disease progresses. Heat shock for mHtt IB formation and TF de-repression gene construct was described previously (13). The other three reporter gene constructs were cloned in the pGL4 vector (Promega): pCRE-firefly luciferase (E8471), pNF-B-firefly luciferase (E8491), and pSRE-firefly luciferase (E1340), humanized Renilla luciferase DNA (phRLSV40), the Dual-Glo Luciferase Assay Reagent (E2920), and the CellTiter-Glo reagent (G7572) were all obtained from Promega. Plasmid expression vectors for the Q23 (plasmid no. 40261) and Q74mHtt-EGFP (plasmid no.40262) were from Addgene. The ST14A rat E14 striatal neuron-derived cell line was from Coriell Institute (CH00066; CHDI-90000066). Lipofectamine 2000 reagent used for DNA transfection was from Invitrogen. Rabbit polyclonal antibody against HSP70 -RTG76 antibody that we generated against a recombinant histidine-tagged human HSP70 protein (9) or from Enzo Life Sciences, ADI-SPA-812, was used at 1:200 dilution for probing of the heat-inducible HSP70. The SPA-816 rabbit polyclonal antibody (Enzo Life Sciences, ADI-SPA-816) was used at 1:40 dilution for probing of HSC70. All other biochemical and chemical reagents were of molecular biology or reagent grade.

Cell culture
This study uses PC12-derived cell lines with stably integrated DNA of the normal (25Q) and polyQ-expanded (103Q; 14A2.6 line) Htt Exon1 sequences under the control of the inducible VgRxR promoter. Cells were cultured and maintained in DMEM supplemented with 5% fetal bovine serum, 5% horse serum, 50 g/ml streptomycin, 50 units/ml of penicillin, 100 g/ml G418, and 200 g/ml zeocin essentially as described previously (20). For imaging experiments, cells were plated in 96 Stripwell plate at a density of ϳ1-2 ϫ 10 4 cells/well. For reporter gene assay, cells were plated at ϳ5-6 ϫ 10 4 cells/well. Ponasterone (Invitrogen) was added to induce the expression of Htt-EGFP and the induction was monitored by live-cell imaging using an EVOS ® FL microscope. Unless indicated otherwise, heat shock was for 2 h at 42°at 24 h after PA addition followed by recovery at 37°C for the remainder of the experiment. Culturing of the ST14A rat E14 striatal neuron-derived cell line was done according to instructions provided by the HD Community Biorepository via Coriell Institute for Medical Research in DMEM with 10% fetal bovine serum.

Image acquisition and quantification
Cells were fixed with 4% paraformaldehyde in PBS for 15 min followed by blocking with 10% fetal bovine serum in PBS and 0.1% Triton X-100. All subsequent washes and incubation were in PBS with 0.1% Triton X-100. Fixed cells were incubated overnight at 4°C with ␣-HSP70 antibody, followed by washing three times for10 min each. Cells were then incubated for 2 h at room temperature with Alexa Fluor 594 goat anti-rabbit IgG (Invitrogen, A11037) and Hoechst 33342 for nuclear stain (300 -500 ng/ml). Cells were imaged with an EVOS ® fluorescent microscope at an exposure setting to optimize the difference between the diffuse and IB mHtt signals. In a given experiment, all images were captured under identical conditions.
For quantification of Htt intensities in the diffuse versus IB format, we wrote Macro programs to quantitate Htt intensity and scoring of individual cells for diffuse mHtt signal versus HSP70 intensity. These Macro programs are included in Fig. S2.

DsiRNA knockdown of HSP70 expression and antisense knockdown of HSC70 expression
A rat HSP70 RNAi kit, TriFECTaRNAi, was obtained from Integrated DNA Technologies. For our experiment, we used the three Dicer-substrate siRNAs of rat HSP70 (rn. Ri.Hspa1a13.1, 13.2, and 13.3), a negative control DsiRNA, and a Tye563 transfection control DsiRNA, all at 20 pmol siRNA/transfection of a 35-mm plate of cells. All siRNA was prepared in Opti-MEM and mixed with Lipofectamine 2000 according to protocols provided by Invitrogen. The hsc70 antisense vector was generated by inserting hsc70 cDNA (1042-2187) in an inverted orientation into pcDNAZeo (Ϫ) (78). Freshly plated 14A2.6 cells were transfected with the designated DNA-or oligoRNA-lipid mixture. After transfection, cells were plated in 96-well plate and mHtt was induced by the addition of 5 M PA. HS was done at 24 h after plating at 42°C for 2 h followed by recovery at 37°C. The time line of the experiment is as indicated in Fig. 4.

Reporter gene and cell viability assays
The four reporter genes used were HSE-, CRE-, NF-B-, and SRE-firefly luciferase. Reporter gene activity was assayed using standard transient transfection protocol as reported previously (68). Briefly, freshly plated 80 -90% confluent cells were used, the amount of each DNA was 0.5 g/35 mm plate or 1.5 g/60 mm plate. The amount of Lipofectamine 2000 (in microliters) was three times that of the total amount of DNA (in micrograms). 6 h after DNA transfection, cells were trypsinized and plated into individual wells of a 96 Stripwell TM plate at a density of ϳ5-6 ϫ 10 4 cells/well (Corning/Costar 9102). The identically transfected cells provided a very stable baseline to assess the effects of mHtt expression and of treatment conditions (e.g. heat shock or dibutyryl cAMP) on reporter gene activity. When specified, the reporter DNA was transfected along with the internal control of phRLSV40 (synthetic humanized Renilla luciferase DNA) to normalize for possible nonspecific effects of treatment conditions on transcription and translation processes. Conditions to induce the promoters of reporter genes were heat shock (2 h at 42°C plus 4 h of recovery at 37°C) for HSE, dibutyryl cAMP (1 mM, 6 h) for CRE, TNF␣ (20 ng/ml, 6 h) for NF-B, and fetal bovine serum for SRE (20%, 6 h). Results are the average Ϯ S.D. of four separate determinations of each sample/condition. Within each experiment, the reporter gene signal was tight, with a Ͻ10% sample to sample variation.
The Dual-Glo Luciferase Assay Reagent from Promega (E2920) was used to assay for first the firefly then the Renilla luciferase activity according to manufacturer's instructions. Luciferase activity was measured using the PerkinElmer VICTOR 2 Multiplate Reader equipped with dual injectors. Where indicated, result of the Hsp70 -firefly luciferase activity (in relative luminescence units, RLU) was normalized against that of the Renilla luciferase (RLU), and to facilitate comparison across experiments for statistical analysis across experiments this ratio was set at 1 for the control (without heat shock) Heat shock for mHtt IB formation and TF de-repression (see Fig. 5A). By normalizing the Hsp70 -firefly luciferase activity against that of the Renilla luciferase internal control, we effectively negated experimental variables such as differences in cell number, as well as nonselective effects of the treatment conditions/reagents on gene expression. Cell viability assay was done using the CellTiter-Glo regent from Promega (G7572) as described previously (9,10).

Data and statistical analysis
ImageJ Fiji and Macro programs (Fig. S2) were used for image analysis and quantitation. Statistical analyses were done using GraphPad InStat or GraphPad Prism 6. Data shown are the mean Ϯ S.D. The significance of difference between groups of data were determined using ANOVA followed by post hoc Tukey-Kramer multiple comparisons test. Probability of difference p Ͼ0.05 is defined as not significant, between 0.01 and 0.05 is significant (*), Ͻ0.01 is very significant (**), and Ͻ0.001 is extremely significant (***).