Chaperones Hsp70 and Hsp40 suppress aggregate formation and apoptosis in cultured neuronal cells expressing truncated androgen receptor protein with expanded polyglutamine tract.

Spinal and bulbar muscular atrophy (SBMA) is one of a group of human inherited neurodegenerative diseases caused by polyglutamine expansion. We have previously demonstrated that the SBMA gene product, the androgen receptor protein, is toxic and aggregates when truncated. Heat shock proteins function as molecular chaperones, which recognize and renaturate misfolded protein (aggregate). We thus assessed the effect of a variety of chaperones in a cultured neuronal cell model of SBMA. Overexpression of chaperones reduces aggregate formation and suppresses apoptosis in a cultured neuronal cell model of SBMA to differing degrees depending on the chaperones and their combinations. Combination of Hsp70 and Hsp40 was the most effective among the chaperones in reducing aggregate formation and providing cellular protection, reflecting that Hsp70 and Hsp40 act together in chaperoning mutant and disabled proteins. Although Hdj2/Hsdj chaperone has been previously reported to suppress expanded polyglutamine tract-formed aggregate, Hsdj/Hdj2 showed little effect in our system. These findings indicate that chaperones may be one of the key factors in the developing of CAG repeat disease and suggested that increasing expression level or enhancing the function of chaperones will provide an avenue for the treatment of CAG repeat disease.

Spinal and bulbar muscular atrophy (SBMA) 1 is an X-linked neurodegenerative disease caused by the expansion of a CAG repeat in the first exon of the androgen receptor (AR) gene (1). In SBMA patients, a normally polymorphic CAG repeat (10 -36 CAGs) expands to 38 -66 CAGs. The number of CAGs is inversely correlated with the age of onset of the disease (2)(3)(4). To date, seven other CAG repeat diseases have been identified, including Huntington's disease (5), dentatorubralpallidoluy-sian atrophy (6,7), and five spinocerebellar ataxias: 1, 2, 3, 6, and 7) (8 -14). These disorders likely share a common pathogenesis involving the gain of a toxic function associated with the expanded polyglutamine tract.
Processing of the polyglutamine-containing disease protein by proteases (e.g. caspase family) may liberate truncated fragments with the polyglutamine tract (15)(16)(17)(18). Truncated proteins with the expanded polyglutamine tracts cause neurodegeneration in transgenic mice as well as Drosophila and cause cell death in transfected cells (19 -24). In addition to cellular toxicity, truncated proteins with the expanded polyglutamine tracts have been shown to form aggregates, likely through hydrogen bonding or transglutaminase activity (25)(26)(27). Studies of CAG repeat disease patients and transgenic mice have revealed that nuclear inclusions formed by the disease protein are a common pathological feature of these diseases (23, 28 -31, 33, 34). In SBMA, nuclear inclusions containing AR protein have been mainly observed in the regions of SBMA central nervous system susceptible to degeneration, including the brain stem motor nuclei and spinal motor neurons (33,34). The finding that nuclear inclusions are ubiquitinated raises the possibility that alterations in the major intracellular system for degrading proteins, the ubiquitin-proteasome pathway, may be involved in the pathogenesis of CAG repeat diseases. The proteasome is a large multicatalytic protease complex that is critical for many cellular processes including cell cycle control, differentiation, antigen presentation, and cell survival (35). Perturbations in proteasome function are associated with altered expression levels of stress response or heat shock proteins (36). These proteins function as molecular chaperones, which recognize and renaturate misfolded proteins under normal and stressed conditions. In addition, chaperones may maintain proteins in an appropriate conformation (37). Recently, overexpression of Hdj-2/Hsdj has been reported to decrease aggregate formation by expanded polyglutamine tract (38,39). It has been postulated that the Hsp70 and the Hsp40 chaperone family members act together to promote cellular protein folding and renaturate misfolded protein (40 -42). We hypothesized that the ability of Hsp70 and Hsp40 chaperones to facilitate refolding or proteolysis of mutant protein may be a key factor for neuronal cells to defend themselves against the toxic properties of expanded polyglutamine tract. The studies reported here demonstrate that overexpression of chaperones, especially a combination of Hsp70 and Hsp40, reduces aggregate formation and apoptosis in cultured neuronal cells expressing truncated androgen receptor protein with an expanded polyglutamine tract.

Plasmid Constructs
Truncated AR-Human AR cDNAs containing 24 or 97 CAG repeats (18) were subcloned into pcDNA3.1 (Invitrogen, Carlsbad, CA). After the constructs were digested with AflII and blunt-ended, the polymerase chain reaction-amplified coding sequence of GFP (0.8 kilobase) was inserted into the digested constructs to create truncated AR constructs (24 CAG repeats, 215 N-terminal amino acids; 97 CAG repeats, 442 N-terminal amino acids of AR).
Chaperones-Construction of pCMV-Hsp70 was described as previously (41). For the construction of pRC-Hsp40, a 1.5-kilobase EcoRI fragment containing the entire coding region of human Hsp40 (43) was excised from the cDNA clone and subcloned in pRC/CMV (Invitrogen). For the construction of pRC-Hsdj, the polymerase chain reaction-amplified coding sequence of Hsdj (1.1 kilobase) (44) was subcloned in pRC/CMV. All the constructs used here were confirmed by DNA sequence analysis.

Immunofluorescence
Transient expression of truncated AR and/or chaperones in Neuro2a cell line (mouse neuroblastoma cell) was accomplished by transfection with Effectene (Qiagen) in a four-chamber slide (Nalge Nunc International, Naperville, IL) coated with rat tail collagen (Roche Diagnostics GmbH, Mannheim, Germany) in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum. After overnight incubation with transfection reagents, transfected cells were cultured in differentiation medium (Dulbecco's modified Eagle's me-dium supplemented with 2% fetal calf serum and 20 M retinoic acid). At each time point (0, 24, 48, and 72 h) after transfection, cells were fixed with methanol for 5 min at room temperature, followed by counter staining with propidium iodide (Molecular Probes, Eugene, OR) and mounted in Gelvatol. 2   Overlay of the two images taken by laser confocal microscopy at 48 h after transfection. a and c, cells transfected with expanded and truncated AR-GFP (tAR97-GFP) contain tAR97-GFP aggregates. Some aggregates localized to nuclear (yellow), and others localized to cytoplasm (green). b, cells transfected with truncated AR-GFP (tAR24-GFP) show GFP signal localized to cytoplasm (green) (counter-stained with propidium iodide). Hsp70, Hsp40 or Hsdj were detected with anti-Hsp70 Ab (1:100), anti-Hsp40 Ab (1:100), or anti-Hsdj Ab (1:10), respectively, incubated at 4°C for overnight, and subsequently stained with anti-rabbit IgG conjugated with Cy-3. Counter stain was done with Hoechst33258 (Molecular Probes) or TOTO-3 (Molecular Probes).

Quantitative Analysis of TUNEL Assay and Truncated AR Aggregate Formation
The cells were assayed for the presence of fragmented DNA by TUNEL assay using a DeadEnd TM Apoptosis Detection Kit (Promega, Madison, WI) according to the manufacturer's protocol with minor modification. Briefly, cells were fixed with 4% paraformaldehyde at each time point (0, 24, 48, and 72 h) after transfection. Fixed cells were incubated with terminal deoxynucleotidyl transferase to be incorporated with biotinylated deoxynucleotides, subsequently stained with Streptavidin-Texas Red conjugate (Life Technologies, Inc.) and Ho-echst33258. A laser confocal scan microscope (MRC1024, Bio-Rad) and a conventional fluorescent microscope were used for quantitative analysis. Duplicate slides were graded blindly in two independent trials. Each slide had over 200 transfected cells. Cells were categorized by localization of aggregates: total aggregate-positive, which are cells containing aggregates in cytoplasm and/or nucleus, and nuclear aggregatepositive, which are cells containing aggregates in nucleus. The frequency was calculated as the number of indicated signal-positive cells divided by that of GFP-positive cells.

Western Blots
48 h after transfection, cells were lysed in RIPA buffer (50 mM Tris, 150 mM NaCl, 1% Nonidet P-40, 0.1% sodium dodecyl sulfate, 10 g/ml aprotinin). Insoluble debris was pelleted, and the protein concentration of the supernatant was determined using a DC protein assay (Bio-Rad). A 20-g sample was electrophoresed on a standard sodium dodecyl sulfate-polyacrylamide gel and transferred to Hybond-P (Amersham Pharmacia Biotech). Blots were probed with the indicated antibodies using standard techniques and developed with enhanced chemiluminescence reagents (Amersham Pharmacia Biotech). The signal intensity was quantified by densitometry. Relative signal intensity (R.S.I.) was computed as the signal intensity of indicated lanes divided by that of nontransfected cells.

Proliferation Assay
Neuro2a cells (500,000 cells/60-mm culture dish) were cotransfected with truncated ARs (tAR24 or tAR97) and chaperones as described above. Cells were plated at a density of 5,000/well in 96-well dish just after transfection. Transfected cells were cultured in 100 l/well of differentiation medium. 20 l/well of CellTiter 96® Aqueous One Solution (Promega) was added at each time point (0, 24, 48, and 72 h) after transfection. After 4 h of incubation at 37°C, the absorbance at 490 nm was recorded using an enzyme-linked immunosorbent assay plate reader. The background absorbance of medium alone was subtracted from these data. The proliferation index was calculated as the mean absorbance of cells at the indicated time point divided by the mean absorbance of cells just after transfection.

In Vitro Aggregation Assay
GST-tAR65-HA was purified as described previously (24). Construction of His-tagged human Hsp40 was as described (40) and expressed in Escherichia coli strain BL21. Cells were grown at 37°C to around 0.8 at A 600 and induced for 2 h with 1 mM isopropyl-␤-D-thiogalactopyanoside. His-Hsp40 was purified with TALON kit (CLONTECH, Palo Alto, CA) according to a manufacture's protocol. A bacterial expression plasmid, pTE-Hsc70, which contains human heat shock cognate 70 (Hsc70) cDNA, was a kind gift of Dr. N. Imamoto (Osaka University, Osaka, Japan). Hsc-70 was purified as described previously (47). 5 pmol of GST-tAR65-HA was incubated with 100 pM of recombinant Hsc70 and/or Hsp40 in reaction buffer (100 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM dithiothreitol, 5 mM ATP, and 5 mM MgCl 2 ). The mixture was incubated at 37°C for 30 min, terminated with equal amount of 2ϫ SDS Laemmli loading buffer and subjected to Western blot as described above. The signal intensity was quantified by densi- tometry. Aggregate index was computed as the signal intensity of aggregate band divided by that of monomer band.

Statistical Analysis
Results were analyzed using analysis of variance and Student's t test if appropriate from Statview software version 5.

RESULTS
A Neuronal Cultured Cell Model of SBMA-We first examined whether truncated and expanded AR (tAR97) forms aggregate and is toxic to Neuro2a cell line. Transient expression of tAR97 induced aggregate formation in cytoplasm and/or nucleus and cellular toxicity in Neuro2a cell line (Figs. 1 and 2). 72 h after transfection, over 90% of transfected cells (92.8 Ϯ 1.1%) had aggregates. As for the localization of aggregate, the longer the cells transfected with tAR97 were cultured, the larger the proportion of nuclear aggregate-positive cells was. 72 h after transfection, more than 80% of transfected cells (86.4 Ϯ 1.6%) had nuclear aggregates (Fig. 2B). Moreover, most of the apoptotic cells had nuclear aggregates, whereas cells containing cytoplasmic aggregate alone had few apoptotic cells (Fig. 3). In contrast, the cells transfected with tAR24 had few aggregates and few apoptotic cells (Figs. 1 and 2). The neuronal cultured cell model has shown nuclear aggregates, which are pathological hallmark of CAG repeat disease, and an apoptosis, which might be mediated by cellular toxicity of expanded polyglutamine tract. These results indicate that this neuronal cultured cell model could be utilized as a cell model of SBMA. We thus employed this cell model for further studies.
Endogenous Chaperones Are Colocalized with Expanded and Truncated AR-formed Aggregates-We next assessed whether chaperones are colocalized with tAR97-formed aggregates in our neuronal cell model of SBMA. Endogenous chaperones in tAR97-transfected and nontransfected Neuro2a cells were discernible predominantly in cytoplasm with limited nuclear staining. Laser confocal microscopic analysis has demonstrated that the endogenous chaperones Hsp70, Hsp40, and Hsdj are up-regulated in the cells containing aggregates and clearly colocalized with aggregates irrespective of the localization of aggregate (Fig. 4). These data had suggested that the cells transfected with tAR97 induce the expression levels of chaperones to protect themselves against a cellular toxicity of expanded polyglutamine tract. However, the amount of induced endogenous chaperones might be insufficient to protect themselves, because the suppression of protein aggregation by chaperone requires a relatively large molar excess of chaperones (48).   Table I). In addition, overexpression of both Hsp70 and Hsdj/Hdj2 or Hsp40 alone was less effective to reduce aggregate formation and cell death than overexpression of both Hsp70 and Hsp40 or Hsp70 alone. Furthermore, there was a parallel correlation between the reduction of aggregate formation and the suppression of apoptosis in these cotransfected experiments.

Overexpression of Chaperones Reduces Aggregate Formation and Apoptosis Induced by Expanded and Truncated AR-To
We further examined whether chaperones reduce a tAR97induced cellular toxicity. Proliferation assay revealed that cells cotransfected with both Hsp70 and Hsp40 or with Hsp70 alone were more proliferative than the cells cotransfected with other chaperones (p Ͻ 0.05 at 48 h after transfection by analysis of variance; Fig. 7). In addition, chaperones and their combinations showed neither cellular toxicity nor proliferative effect when transfected with chaperones alone in this proliferative assay (data not shown).
Expression Levels of Chaperones in the Cotransfected Experiments-We next examined expression levels of chaperones in the cotransfected experiment. Although nontransfected Neuro2a cells expressed a small quantity of Hsp70, Hsp40, and Hsdj/Hdj2 (lanes 1 in Fig. 8, A, B, and C, respectively), cells transfected with tAR24 had a large quantity of chaperones (Hsp70 (2.0 R.S.I.), Hsp40 (7.0 R.S.I.), and Hsdj/Hdj2 (3.3 R.S.I.); lanes 2 in Fig. 8, A, B, and C, respectively). Moreover, cells transfected with tAR97 expressed a larger quantity of chaperones (Hsp70 (10.7 R.S.I.), Hsp40 (15.0 R.S.I.), and Hsdj/ Hdj2 (10.7 R.S.I.); lanes 3 in Fig. 8, A, B, and C, respectively). Cells cotransfected with expanded and tAR97 and Hsp70 and/or Hsp40 have shown a much larger quantity of Hsp70 or Hsp40 (Hsp70, lanes 4 (32.6  Chaperones Decrease in Vitro Aggregation of Expanded Polyglutamine Tract-We previously reported that expanded polyglutamine tract forms aggregates in vitro in a polyglutamine length-dependent manner (24). We next assessed the ability of Hsc70 and Hsp40 to reduce in vitro aggregation of expanded and truncated AR (tAR65) using our in vitro aggregation system. As expected, 20-fold molar excess of Hsc70 and Hsp40 or Hsc70 alone effectively suppressed in vitro aggregation of tAR65 (Hsc70/Hsp40 versus bovine serum albumin, p Ͻ 0.01; Hsc70 versus bovine serum albumin, p Ͻ 0.01 by Student's t test) (Fig. 9). However, Hsp40 alone has little ability to suppress aggregation. DISCUSSION The present study first demonstrated that overexpression of chaperones, especially combination of Hsp70 and Hsp40, reduce cytotoxicity and aggregate formation induced by expanded polyglutamine tract in our cultured neuronal cell model of SBMA. These findings suggest that the chaperones may be one of the key factors in the development of CAG repeat disease.
We previously reported that a truncated and expanded AR forms aggregate and is toxic to Cos-7 and MN-1 cells (24). In addition, we observed nuclear inclusions in SBMA (33,34). In other CAG repeat diseases, truncated and expanded polyglutamine fragments of disease proteins have been also shown to form aggregates and have toxicity in transfected cells, transgenic mice, and flies (19 -24). However, the role of aggregates in the pathogenesis of CAG repeat diseases is controversial (23,49). The relationship between aggregate formation, particularly in the nucleus, and induction of apoptosis still remains to be resolved. Our results reveal that combination Hsp70 and Hsp40 or Hsp70 alone has a favorable effect in a cellular protection as well as suppression of aggregate formation and that combination of Hsp70 and Hsp40, especially, has the strongest effect among them. Our results were in accordance with the reports that the chaperone function of Hsp70 is critically dependent on the cooperation with Hsp40 (40,41,42). The chaperone activity of Hsp70 family is regulated by Hsp70 ATPase activity (52), and Hsp40 stimulates the Hsp70 ATPase by increasing the rate of ATP hydrolysis (40). In other conformational diseases, the study about mutant copper/zinc superoxide dismutase (SOD-1)-associated amyotrophic lateral sclerosis also displayed that increasing the level of Hsp70 reduced formation of mutant SOD-containing aggregates in cultured primary motor neurons expressing mutant SOD-1 and prolonged their survival (51). It is therefore reasonable to consider that disease gene product-formed aggregates are directly associated with the induction of neurodegenaration in CAG repeat disease and other conformation diseases. We thus reasoned that overexpression of both Hsp70 and Hsp40 chaperones reduce cytotoxicity induced by aggregate formation with disease gene product in CAG repeat disease and other conformation diseases.
The question is the molecular mechanism for the reduction of cytotoxicity through inhibition of expanded and truncated ARformed aggregate by overexpression of chaperones. Although it has been proposed that expanded polyglutamine tract-formed aggregates participate in inappropriate protein-protein interactions that lead to cell death, the nature of such interactions and the mechanism by which cell death is induced remain unclear. Molecular chaperones could be involved in the actual formation of expanded polyglutamine tract-formed aggregates by stabilizing the unfolded protein in an intermediate conformation that has the propensity to interact with self or other proteins. To date, several proteins interacting with polyglutamine tract-containing disease gene product have been cloned, including huntingtin-associated protein (53), huntingtin-interacting protein (54), glyceraldahyde-3-phosphate dehydrogenase (55), leucine-rich acidic nuclear protein (56), and PQBP-1 (polyglutamine tract-binding protein-1) (57). These interacting proteins are candidate players in the pathogenesis of CAG repeat disease. Chaperones might reduce cytotoxicity of expanded and truncated AR through inhibiting the interaction of expanded polyglutamine-formed aggregate with these proteins.
Another possibility is that overexpression of chaperones is enhancing the function of the ubiquitin-proteasome pathway for mutant protein degradation because the function of the ubiquitin-proteasome pathway is related with the expression level of chaperones (36). Nuclear aggregates are ubiquitinated and are colocalized with chaperones and proteasome, implicating the ubiquitin-proteasome degradation pathway in the pathogenesis of CAG repeat disease (33,34,38,58). The report that the inhibition of proteasome function accelerates aggregate formation by polyglutamine tract also implies the ubiquitin-proteasome degradation pathway plays a direct role in modulating aggregation in CAG repeat disease (58). In Alzheimer's disease, one of the conformational disease, amyloid ␤-protein, which is a major component of senile plaque, could interfere with ubiquitin-dependent protein degradation pathway by inhibiting the 26 S proteasome. Consequently, it may lead to neuronal damage observed in Alzheimer's disease (59). Similarly, expanded polyglutamine tract would interfere with ubiquitindependent protein degradation pathway and lead to neuronal damage in CAG repeat disease. Thus overexpression of chaperones would enhance the function of proteasome, leading to protecting cells expressing truncated and expanded AR against a cellular toxicity of expanded polyglutamine tract.
Recently, there are reports that overexpression of Hsdj/Hdj2 in HeLa cells decreases the frequency of mutant ataxin-1 and mutant AR aggregation (38,39); however, overexpression of Hsdj/Hdj2 has little effect of reducing aggregate formation and providing cellular proliferation in our result. These differences of results may arise from the difference in cell lineage, the different origin of Hsdj/Hdj2, or the expression level of Hsdj/ Hdj2 in transfected cells. Previous studies used a non-neural cell line (HeLa cell), whereas we used a neural cell line (Neuro2a). The difference in cell lineages could influence the relations of chaperones, aggregation, and cell death. Although Hsdj we employed in this study is 99% identical to Hsdj2 employed in other reports at the level of amino acid sequence (32,44), the relationship between Hsdj and Hdj2 remains to be studied. Alternatively, ineffectiveness of Hsdj/Hdj2 in our study might be explained by relatively low overexpression of the protein in comparison with Hsp70 and/or Hsp40, as shown in Western blotting analysis.
In contrast to our results, two groups have provided evidence against a critical role of intranuclear aggregates in neuronal cell death (49,60). The results opposite to ours could arise from the difference of cell type/animal (primary neuron/animal versus cell line) or interventions (neurotrophic factors/suppression of ubiquitin-conjugating enzyme/inhibition of ataxin-1 self-association versus overexpression of chaperones). However, we cannot rule out the possibility that our observation, a parallel correlation between the reduction of aggregate formation and suppression of apoptosis in our system, might occur coincidentally. Thus, Hsps could reduce aggregate formation as a molecular chaperone and independently suppress apoptosis in our cell system through a different molecular mechanism. Al-  though Hsp70 was reported to have an anti-apoptotic effect, the mechanism of such effect remains veiled (50). Therefore, discussion of this possibility needs to await future studies. Finally, the findings described in this report highlight that chaperones play a role in the developing of CAG repeat disease. These results suggested that increasing expression level or enhancing the function of chaperones provides an avenue for the treatment of CAG repeat disease. Further studies of cellular and animal models are required to determine the precise mechanism of neurodegeneration of CAG repeat disease mediated by expanded polyglutamine tract as well as therapeutic approach.