Tet-On PC12 cells expressing either ARQ112 or ARQ10, were previously created in our laboratory (Walcott and Merry, 2002). Cells were maintained in normal growth media (Dulbecco’s modified Eagle’s medium with 10% heat-inactivated horse serum, 5% heat-inactivated fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin/streptomycin, 200 μg/ml hygromycin (Invitrogen, Carlsbad, CA, USA), 100 μg/ml G418 (Mediatech, Manassas, VA, USA) at 37°C, 10% CO₂). All experiments were performed utilizing charcoal-stripped serum. Cells were induced with 500 ng/ml doxycycline (Dox; Clontech) to express equivalent levels of AR and treated with 10 nM DHT (Sigma Aldrich). Stable REGγ transformants were selected with puromycin (2 μg/ml; Invitrogen, Inc.). Proteasome inhibitor epoxomycin (Sigma Aldrich) was used at 500 nM concentration.

FLAG-REGγ pcDNA3.1+plasmid (provided by Dr. Martin Rechsteiner, University of Utah). Site-directed mutagenesis (Quick Change II XL, Stratagene) was performed to mutate FLAG- REGγ into both REGγP245Y and REGγK188E. Mutations were confirmed by sequence analysis.

Stable transfections of Tet-On PC12 ARQ112 and ARQ10 cells were performed utilizing a calcium phosphate transfection protocol. FLAG-REGγ or FLAG-REGγP245Y plasmids were co-transfected with puromycin pBABE plasmid (kindly provided by Dr. Karen Knudsen, Thomas Jefferson University) at a 4:1 ratio, respectively. Single colonies were isolated, expanded, and screened for overexpressed REGγ using PSME3 antibody (BD Transduction Laboratories). Cells were maintained as stated above and media was supplemented with puromycin.

PC12 cells and dissociated spinal cord cultures were lysed in 0.5% Triton-0.5% DOC lysis buffer and sonicated using a Branson cup sonifier. A DC protein assay (Bio Rad, Inc.) was performed to determine protein concentration. Protein lysates were electrophoresed by SDS-PAGE and transferred to a 0.45 μm PVDF (Immobilon-P) membrane. Western hybridization was performed using the following antibodies: [(AR(H280), AR(N20), AR(441) Santa Cruz Biotechnology), GAPDH (Fitzgerald Industries International)] (1:1000), [MDM2(SMP14); γ-tubulin (1:10,000) (Sigma Aldrich); REGγ (PSME3)] (1:2000) (BD Transduction Laboratories); FLAG (1:500) (Affinity BioReagents); SC35 (1:500) (Euromedex). Detection was carried out using ECL substrate (Pierce). Filter trap analysis was carried out as previously described (Bailey et al., 2002). Western analysis was performed using a standard protocol for antibody ARH280 (Santa Cruz Biotechnology).

PC12 cells and dissociated spinal cord cultures were fixed with 4% paraformaldehyde for 20 min, washed in PBS, permeabilized with 0.3% Triton X-100 for 15 min, blocked in 2% goat serum (Jackson ImmunoResearch, West Grove, PA, USA) for 20 min and incubated with primary antibody diluted in 1.5% goat serum for 1 h. Cells were washed in PBS and incubated with secondary antibody (FITC- or Texas Red-conjugated; Jackson ImmunoResearch, West Grove, PA, USA) for 30 min, washed with PBS, incubated in Hoechst (2 μg/ml), washed in PBS and mounted in Vectashield (Vector Laboratories, Burlingame, CA, USA). Fluorescence was visualized with a Leica (Leica Microsystems GmbH, Wetzlar, Germany) microscope and pictures were taken with a Leica camera and compiled with IP Lab software (BD Biosciences, Rockville, MD, USA). Antibodies utilized included the following: [AR(N20), AR(H280)] (1:100), p16(M-156) (1:50) (Santa Cruz Biotechnology); PSME3 (1:100) (BD Transduction Laboratories); FLAG (1:100) (Affinity BioReagents); SMI32 (1:1000) (Sternberger Monoclonal, Baltimore, MD, USA).

PC12 cells were either treated with Dox (500 ng/ml) for 48 h and DHT (10 nM) in the presence or absence of epoxomycin (500 nM) for the last 12 h or with Dox (500 ng/ml) and DHT (10 nM) for 72 h with or without epoxomycin (500 nM) for the last 12 h. Controls consisted of PC12 ARQ112 cells inducibly expressing FLAG-REGγ or FLAG-REGγP245Y that were not treated with Dox or DHT, but treated with epoxomycin for 12 h. Cells were harvested in NP-40 lysis buffer (0.15 M NaCl, 50 mM NaF, 50 mM Tris-HCl (pH 7.5), 0.5% NP-40, and protease inhibitors), sonicated and protein concentration determined. Cell lysates (250 μg) were precleared with magnetic Protein G Dynabeads (Invitrogen) for 1 h at room temperature. Precleared lysates were incubated with AR(H280) antibody (0.5 μg) overnight at 4°C. Protein G Dynabeads were incubated with lysate/antibody for 30 min at room temperature, unbound lysate was removed and the beads were washed four times with NP-40 lysis buffer. Product was eluted with 2× Laemmli buffer and electrophoresed by SDS-PAGE, transferred to PVDF and Western analysis performed.

FLAG-REGγ and FLAG-REGγK188E adenoviruses (replication incompetent (deltaE1/E3) human adenoviral type 5 genome) were constructed utilizing BD Adeno-X™ Expression System 1 (BD Biosciences). In brief, the pShuttle2 plasmid (BD Biosciences) and the FLAG- REGγ plasmids were digested with NheI and NotI restriction enzymes and then ligated to each other. The pShuttle2/FLAG-REGγ plasmid and BD Adeno-X Viral DNA (provided) were then cut with I-CeuI and PI-SceI restriction enzymes and ligated to each other. The resulting BD Adeno-X Viral plasmid containing FLAG-REGγ was sequenced to ensure proper ligation and then purified and digested with PacI restriction enzyme in order to linearize the plasmid. The PacI digested plasmids were transfected into low passage HEK293T cells and then freeze/thawed to release adenovirus. Adenoviruses were purified utilizing the Virabind™ Adenovirus Miniprep kit (Cell Biolabs, Inc., San Diego, CA, USA) and then titered using the QuickTiter™ Adenovirus Immunoassay Kit (Cell Biolabs, Inc., San Diego, CA, USA) according to published protocols.

PC12 ARQ112 cells were plated at 5 × 10⁵ cells/well in a 6-well dish and then infected with the following adenoviruses at multiplicity of infection (MOI) 100: FLAG-REGγ, FLAG-REGγK188E, LacZ. Cells were treated with 50 μM DHT and 500 ng/ml Dox for 72 h and then harvested for protein or immunostained as described above. Percent cells infected by FLAG-REGγ and FLAG-REGγKE were measured by calculating the number of FLAG-positive cells/total number of cells. LacZ viral infection was determined by performing an X-gal stain. Cells were counted and percent of cells with nuclear inclusions was calculated. Statistical analysis was carried out as described below.

For REGγ knockdown studies: PC12 ARQ112 cell pellets (4 × 10⁶ cells) were resuspended in 100 μl of Nucleofector solution V (Amaxa®) (Lonza Cologne GmbH) with 50 pmol REGγsiRNA (ON-TARGETplus SMART pool rat PSME3), 50 pmol Non-Targeting siRNA (ON-TARGETplus Non-Targeting siRNA #1) (Dharmacon) or 2 μg GFP plasmid (Amaxa). Cells were electrophoresed using the Amaxa Nucleofector System (program U-029) and then replated in 6-well and 24-well dishes. At 72 h post nucleofection, cells were treated with Dox and DHT for 48 h. Cells were harvested at 120 h total and either lysed in Triton DOC lysis buffer and run on SDS-PAGE gels or immunostained. Percentage of siRNA protein knockdown was determined by densitometry. Cells were counted and percent of cells with nuclear inclusions was calculated. Statistical analysis was carried out using the Student’s t-test.

PC12 ARQ112 cells were electroporated with the Amaxa® Nucleofector® System (Lonza Cologne GmbH) as described above with the following plasmids: FLAG-REGγ, FLAG- REGγP245Y, or GFP. Twenty-four hours post-transfection, cells were treated with Dox (500 ng/ml) and DHT (10 nM) for 72 h. Cells were then either harvested for biochemical analysis (Triton-DOC lysis buffer) or immunostained as described above.

Cells were grown in 100 mm dishes and induced with Dox (500 ng/ml) for 60 h and treated with DHT (10 nM) and epoxomycin (500 nM) for the last 12 h. Cells were lysed on the dish in tandem ubiquitin binding entity (TUBE) lysis buffer (20 nM Na₂HPO₄, 20 nM NaH₂PO₄ (pH 7.2), 50 mM NaF, 5 mM tetra-sodium pyrophosphate, 10 nM β-glycerophosphate, 2 mM EDTA, 1 mM DTT, 1% NP-40) supplemented with fresh N-ethyl maleimide (NEM; 10 mM), and protease inhibitors and then 40 μl of GST-TUBE2 (Lifesensors, Inc., Malvern, PA, USA) was added. Cells were incubated on ice for 15 mins and then clarified (12,000 rpm for 10 min). A DC protein assay was performed to determine protein concentration. Glutathione (GST) affinity resin (GE Healthcare, Inc.; 100 μl) was equilibrated according to the provided protocol. GST resin was added to the cell lysate/TUBEs (2.5 mg protein) and incubated at 4°C for 4 h. GST-resin was collected by centrifugation (1000× g, 4°C) for 5 min, unbound sample was removed and resins washed three times with TBST. Product was eluted in 2× Laemmli buffer and loaded on a 10% SDS-PAGE gel for electrophoresis and transferred to 0.45 mM PVDF (Immobilon-P). Western hybridization was performed using the following antibodies: AR(441) (1:500) (Santa Cruz Biotechnology) and Ubiquitin (1:1000) (Dako).

FLAG-REGγ-expressing PC12 cells, inducible for ARQ112 or ARQ10, were treated with 500 ng/ml Dox to express equivalent AR levels and then treated in the presence or absence of 10 nM DHT for 12 days. Cells were harvested and stained with trypan blue. Two hundred cells were counted per cell type/condition in triplicate. The percentage of trypan blue-positive cells was calculated and significance determined as described below.

Adeno-associated viruses (AAV; type 1) were constructed by the following protocol: FLAG-RE REGγ pcDNA3.1 plasmids were cut with HindIII and XhoI restriction enzymes (Fisher Scientific) and the AAV-Cis plasmid (AM/CBA-pl-WPRE-bGH containing a chicken β-actin promoter and CMV enhancer) was cut with BamHI and XhoI restriction enzymes (Fisher Scientific); the two plasmids were ligated to form the FLAG-REGγ AAV-Cis plasmid. In addition, the pcDNA™6.2-GW/ EmGFP-miR plasmid, containing scrambled miRNA or miRNA sequences against SC35 (construction of these plasmids outlined below) was cut with XhoI and partially cut with DraI, while the AAV-Cis plasmid was cut with EcoRI and EcoRV. Both plasmids were treated with Klenow to blunt the ends and ligated together to form the scrambled AAV-, SC35 miRNA #24 AAV-, SC35 miRNA #27 AAV-Cis plasmids. Low passage HEK293T cells were plated and transfected with the FLAG-REGγ Cis, Trans (H21), and Adenovirus-associated Helper (δ6) plasmids using calcium phosphate. At three–5 days post-transfection, cells were harvested, lysed in PBS, freeze/thawed three times in order to obtain AAV viral stocks and then filtered. The volume of virus (HEK293T cell lysate) to achieve maximum AAV infection of motor neurons dissociated spinal cord cultures was determined by immunostaining for the protein of interest 5 days post-infection.

Dissociated spinal cord cultures were established according to published literature (Roy et al., 1998; Montie et al., 2009). Spinal cord cultures derived from transgenic ARQ112 mice were dissected on ice from 13.5-day-old embryos. Genotyping was carried out from a tail biopsy. Transgenic ARQ112 spinal cords were pooled, plated for culture and incubated for 3 weeks in conditioned media [MEM, 3% charcoal-stripped horse serum, 35 mM NaHCO₃, 0.5% dextrose, 1% N3, 10 nM 2.5 S NGF (added after conditioning)] from 13.5-day-old non-transgenic brain. Motor neurons were identified in these mixed spinal cord cultures by staining with neurofilament heavy chain (SMI32) antibody and morphology. Three weeks after culture initiation, motor neurons were infected with the following AAV-viruses: EGFP, FLAG-REGγ, FLAG-REGγKE, FLAG-REGγP245Y, scrambled miRNA, or miRNAs against SC35 (#24 and #27). Five days following infection, motor neurons were treated with EtOH or 10 μM DHT for an additional 7 days. The work described here was carried out in strict accordance with the recommendations of The Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, with approval by the Institutional Animal Care and Use Committee of Thomas Jefferson University.

Two pre-designed oligos (forward and reverse for each, Mmi519524 and Mmi519527) containing miRNA sequences against SC35, as well as scrambled, control sequences, were purchased from Invitrogen, annealed and cloned into pcDNA™6.2-GW/ EmGFP-miR (BLOCK-iT™), containing a blasticidin resistance gene, according to the manufacturer’s recommendations. Inducible PC12 cells (ARQ112) were stably transfected with these plasmids. Twenty-four hours post-transfection, cells were selected with (15 ng/mL) blasticidin. After 2 weeks in culture, 100% of the cells expressed EmGFP. Cells were treated with EtOH or 10 nM DHT for 12 days. At the end of the treatment, cells were harvested and stained for trypan blue. Two hundred cells were counted per cell type/condition in triplicate. The percentage of trypan blue-positive cells was calculated and significance was determined as described below.

Statistical analysis was carried out using either SigmaStat or Excel. For single factor analysis of variance (ANOVA), post hoc Bonferroni-corrected Student’s t-test was used to determine statistically significant differences. For two factor ANOVA, post hoc Tukey HSD was used to determine statistically significant differences. For single sample analysis with two conditions, Student’s t-test was used after determining sample variance with the F statistic.

Numerous studies have demonstrated that components of the UPS system are present within nuclear inclusions in cell and animal models of SBMA and tissue of affected SBMA patients (Li et al., 1998; Stenoien et al., 1999; Abel et al., 2001; Walcott and Merry, 2002); we therefore wished to determine whether REGγ co-localizes with hormone-induced ARQ112 nuclear inclusions in this cell model of SBMA. We observed REGγ both in a diffuse nuclear distribution and co-localized with a portion of AR nuclear inclusions (Figure 1).

The presence of REGγ in AR nuclear inclusions suggests that nuclear REGγ-capped proteasomes may play a role in the inefficient degradation of mutant AR. Moreover, it has been previously shown that REGγ is capable of activating only the T-L activity of the 20S proteasome core, thereby limiting the degradative capacity of nuclear 11S proteasomes (Realini et al., 1997). However, mutation of REGγ at lysine 188 to an acidic residue (K188E, K188D), as found in the cytoplasmic REG homologs REGα and β (Zhu et al., 2001), allows the activation of all three catalytic activities of the REGγ-capped 20S proteasome. We therefore sought to investigate whether overexpression of REGγ or REGγK188E would impact the aggregation of expanded AR. We predicted that REGγK188E might enhance the nuclear degradation of expanded full-length AR or expanded AR fragments due to its REGα/β-like activation capabilities and thereby decrease AR aggregation. The cell model that we used to test this hypothesis is a well characterized inducible PC12 cell model of SBMA, in which hormone treatment of cells expressing an AR gene with 112 CAG repeats (ARQ112) results in the formation of nuclear inclusions comprised of an amino-terminal AR fragment (Walcott and Merry, 2002; Heine et al., 2015), as observed in SBMA patient neurons (Li et al., 1998). Moreover, nuclear inclusions correlate well with a subset of biochemical AR aggregates detected by SDS-agarose gel electrophoresis (SDS-AGE; Berger et al., 2015; Heine et al., 2015). To test our hypothesis, adenoviruses were created to express FLAG-tagged REGγ, FLAG-REGγK188E or LacZ in PC12 ARQ112 cells. Overexpression of both FLAG-REGγ and FLAG-REGγK188E resulted in a substantial increase in the frequency of cells with ARQ112 nuclear inclusions, compared to the LacZ control (Figure 2A).

We next wanted to determine whether knockdown of REGγ would impact expanded AR aggregation. Knockdown of REGγ caused a significant decrease in ARQ112 inclusion bodies, but did not eliminate them (Figure 2B). Thus, not only did the overexpression of REGγ enhance AR aggregation, but the knockdown of endogenous REGγ decreased AR aggregation, suggesting a role for REGγ in the aggregation of mutant AR.

One possible explanation for these results is that REGγ heptameric rings compete with the 19S proteasomal activator, thereby decreasing AR degradation and enhancing its aggregation. Therefore, given that REGγ has functions that are both proteasome binding-dependent and proteasome binding-independent, we asked whether proteasome binding by REGγ is required for these effects. Proline 245 was previously shown to be required for the binding of REGγ to the 20S proteasome α rings (Zhang et al., 1998; Zhang and Zhang, 2008). To determine whether interaction of REGγ with the 20S proteasome core is required for the observed REGγ effects on ARQ112 aggregation, we created PC12 ARQ112 cells that stably express either REGγ or the proteasome-binding mutant REGγP245Y. We found that stable overexpression of both REGγ and REGγP245Y significantly increased ARQ112 aggregation (Figure 3A, Supplementary Figure S1), indicating that REGγ increases expanded AR aggregation in a 20S proteasome-binding-independent manner. Moreover, mutant REGγP245Y had a slightly greater effect on increasing ARQ112 aggregation, compared to wild-type REGγ, possibly due to its limited non-proteasome binding role. The same aggregation results were observed upon transient overexpression of both constructs (data not shown). Neither overexpression of REGγ or of REGγP245Y altered the steady state levels of ARQ112 (Figure 3B).

With the observation that REGγ promotes AR aggregation in a proteasome binding-independent manner, we sought to determine the mechanism for this effect. We found that full-length ARQ112 co-immunoprecipitated with both endogenous and exogenous (overexpressed) REGγ (Figures 4A,B). Moreover, treatment of cells with the proteasome inhibitor epoxomicin enhanced the interaction of REGγ with ARQ112 (Figure 4B), without changing the steady-state levels of REGγ or AR (Figure 4B, right). We also found that REGγP245Y co-immunoprecipitated with ARQ112 (Figure 7), indicating that REGγ can interact with ARQ112 independently of REGγ association with the proteasome. We did not detect an interaction of REGγ with unexpanded ARQ10 (data not shown), suggesting that REGγ has a limited, if any, role in normal AR metabolism.

Utilizing PC12-AR-expressing cell lines that we engineered to stably overexpress REGγ or REGγP245Y, we examined whether overexpression of both isoforms modulated DHT-dependent cell death. We found that ARQ112 cells stably overexpressing either REGγ or REGγP245Y were not protected from DHT-induced AR toxicity (Figures 5A,B), and neither of the REGγ isoforms caused death in cells expressing unexpanded ARQ10 (Figure 5A, data not shown). Moreover, REGγP245Y overexpression resulted in slightly increased toxicity (Figure 5B), as observed with its effect on AR aggregation.

REGγ has been shown to promote the polyubiquitination of p53 in a proteasome binding-independent manner (Zhang and Zhang, 2008). We therefore wanted to evaluate whether REGγ overexpression modified the polyubiquitination state of ARQ112. We observed that the overexpression of both REGγ and REGγP245Y significantly decreased the polyubiquitination state of ARQ112 (Figure 6). While the total amount of ubiquitin pulled down from the lysates was equivalent across cell lines (data not shown), the amount of polyubiquitinated ARQ112 was reduced in cells expressing either REGγ or REGγP245Y. Therefore, REGγ negatively impacts ARQ112 polyubiquitination. The apparent decrease in ARQ112 polyubiquitination might be explained by increased ARQ112 aggregation and consequent insolubility. However, evaluation of ARQ112 polyubiquitination was carried out under conditions that minimized AR aggregation; indeed, no AR-containing nuclear inclusions were observed in the short time of DHT treatment (data not shown). Moreover, we evaluated AR solubility using ultracentrifugation and Western blot analysis; no increase in ARQ112 sedimentation was observed under these conditions. Consistent with its lack of interaction with ARQ10, REGγ had no effect on the polyubiquitination state of ARQ10 (data not shown).

We next wanted to evaluate the mechanism by which REGγ decreases the polyubiquitination of ARQ112. A recent study showed that REGγ facilitates the interaction of the E3 ubiquitin ligase MDM2 with p53, thus promoting the polyubiquitin-dependent degradation of p53 (Zhang and Zhang, 2008). MDM2 is a known E3 ubiquitin ligase for the AR (Lin et al., 2002); therefore, we hypothesized that REGγ might prevent the interaction of ARQ112 with MDM2, resulting in reduced polyubiquitination of the receptor. We observed MDM2 co-immunoprecipitation with polyQ-expanded AR only in the presence of the proteasome inhibitor epoxomycin (Figure 4B), which substantially increased the otherwise low levels of MDM2 in these cells (Figure 4B, right). Consistent with our finding that REGγ overexpression decreased AR polyubiquitination, we found that REGγ overexpression also decreased the interaction of ARQ112 with MDM2 (Figure 7), suggesting that REGγ may decrease ARQ112 polyubiquitination by reducing the interaction of ARQ112 with the ubiquitin ligase MDM2.

Our finding that REGγ overexpression in PC12 cells significantly increased ARQ112 aggregation led us to ask whether REGγ overexpression would impact SBMA motor neuron survival. Transgenic ARQ112 motor neurons undergo significant hormone-dependent death (Montie et al., 2009, 2011; Orr et al., 2010; Heine et al., 2015; Zboray et al., 2015). We hypothesized, based on our observations in PC12 cells, that overexpression of REGγ and REGγK188E might enhance motor neuron death after exposure to hormone. To test this hypothesis, we infected dissociated spinal cord cultures with AAV that expressed FLAG-REGγ isoforms. A high percentage of motor neurons were infected (Figure 8A), and significant overexpression of REGγ isoforms was observed (Figure 8B, right). In contrast to our observations in PC12 cells, however, we found that overexpression of either REGγ or REGγK188E not only failed to enhance DHT-dependent toxicity, but rather rescued ARQ112 motor neurons from DHT-induced toxicity (Figure 8B, left). Moreover, REGγ protected ARQ112 motor neurons from DHT-dependent toxicity in a proteasome binding-dependent manner, as overexpression of the proteasome non-binding mutant REGγP245Y did not rescue motor neuron viability (Figure 8C). This suggests that the effect of REGγ on motor neuron survival depends on a REGγ proteasome binding (11S)-dependent mechanism.

We next determined whether REGγ overexpression rescued motor neurons from DHT-dependent cell death by altering levels of the mutant polyQ-expanded AR. Western analysis revealed that REGγ overexpression did not alter AR monomer levels (Figure 8C, right), suggesting that REGγ does not promote motor neuron survival by enhancing ARQ112 degradation. The effect of REGγ on AR aggregation and inclusion formation in cultured motor neurons could not be evaluated, as inclusions do not form in these neurons during the time course of the toxicity assays performed here (Heine et al., 2015).

The observation that the neuroprotection conferred by REGγ in SBMA mouse motor neurons requires its ability to bind the 20S proteasome core prompted us to evaluate potential REGγ substrates that mediate its effects. Several REGγ-activated proteasomal substrates have been identified. These include the CDK inhibitors p21^WAF/CIP1, p16^INK4a and p19^ARF (Chen et al., 2007) the transcriptional co-activator SRC-3 (Li et al., 2006) and the E3 ligase SMURF1 (Nie et al., 2010). We evaluated levels of p16^INK4a in motor neurons in the presence of overexpressed REGγ but found no change in its levels (unpublished data). Another potential substrate of REGγ-capped proteasomes is the nuclear splicing factor SC35. REGγ-capped proteasomes colocalize with SC35-containing nuclear speckles (Baldin et al., 2008). Moreover, knockdown of REGγ leads to altered SC35 distribution. To evaluate the potential role of SC35 in REGγ-induced protection of SBMA motor neurons, we first determined the effect of ARQ112 expression on SC35 levels, using the PC12 cell model described above. SC35 is transcriptionally regulated by E2F1 (Merdzhanova et al., 2008) and is consequently present at much higher (and detectable) levels in cycling PC12 cells than in primary motor neurons (unpublished data). Expression of polyQ-expanded AR resulted in increased expression of SC35, which was further increased by DHT treatment (Figure 9A). Moreover, overexpression of REGγ, but not the proteasome binding-mutant REGγ-P245Y, resulted in the reduction in SC35 levels (Figure 9A).

To determine the extent to which decreased SC35 levels contributed to REGγ neuroprotection, we knocked down SC35, in the absence of overexpressed REGγ and determined the effect on cellular viability. ARQ112-expressing PC12 cells were stably transfected with either a plasmid expressing scrambled miRNA or a plasmid expressing miRNA against SC35 and placed under blasticidin selection for 2 weeks; after 2 weeks, all cells were confirmed to express EmGFP, which is contained on the same expression plasmid. Cells were then treated with EtOH or DHT and viability assessed after 12 days. Reducing SC35 expression provided complete protection against DHT-dependent death (Figure 9B). To determine if knockdown of SC35 would also prevent DHT-dependent ARQ112 toxicity in primary motor neurons derived from SBMA transgenic mice, we infected dissociated spinal cord cultures from ARQ112-expressing mice with AAV expressing either a scrambled miRNA or a miRNA against SC35 and evaluated the effect on DHT-dependent AR toxicity. Knockdown of SC35 in primary SBMA motor neurons also substantially reduced DHT-dependent AR toxicity (Figure 9C).