Pathomechanism characterization and potential therapeutics identification for SCA3 targeting neuroinflammation

Polyglutamine (polyQ)-mediated spinocerebellar ataxias (SCA) are caused by mutant genes with expanded CAG repeats encoding polyQ tracts. The misfolding and aggregation of polyQ proteins result in increased reactive oxygen species (ROS) and cellular toxicity. Inflammation is a common manifestation of oxidative stress and inflammatory process further reduces cellular antioxidant capacity. Increase of activated microglia in the pons of SCA type 3 (SCA3) patients suggests the involvement of neuroinflammation in the disease pathogenesis. In this study, we evaluated the anti-inflammatory potentials of indole compound NC009-1, 4-aminophenol-arachidonic acid derivative AM404, quinoline compound VB-037 and chalcone-coumarin derivative LM-031 using human HMC3 microglia and SCA3 ATXN3/Q75-GFP SH-SY5Y cells. The four tested compounds displayed anti-inflammatory activity by suppressing NO, IL-1β, TNF-α and IL-6 production and CD68 expression of IFN-γ-activated HMC3 microglia. In retinoic acid-differentiated ATXN3/Q75-GFP SH-SY5Y cells inflamed with IFN-γ-primed HMC3 conditioned medium, treatment with the tested compounds mitigated the increased caspase 1 activity and lactate dehydrogenase release, reduced polyQ aggregation and ROS and/or promoted neurite outgrowth. Examination of IL-1β- and TNF-α-mediated signaling pathways revealed that the tested compounds decreased IκBα/P65, JNK/JUN and/or P38/STAT1 signaling. The study results suggest the potential of NC009-1, AM404, VB-037 and LM-031 in treating SCA3 and probable other polyQ diseases.

Expansion of the polyQ track in ATXN3 protein likely induces a conformational change to affect its subcellular localization and propensity to aggregate [10]. In addition, transcriptional dysregulation AGING [11,12], decreased anti-oxidative capacity [13][14][15], DNA repair dysfunction [16] and impaired ubiquitin proteasome and autophagy activity [17,18] play important roles in pathogenesis of SCA3. The misfolded and aggregated ATXN3 protein results in a concomitant increase in reactive oxygen species (ROS) levels and cellular toxicity [13-15, 19, 20]. Inflammation is one of the manifestations of oxidative stress and inflammatory process may further induce oxidative stress and reduce cellular antioxidant capacity. In pontine neurons of SCA3 patients, expression of pro-inflammatory cytokines such as interleukin (IL)-1 receptor antagonist and IL-1β was increased, accompanied with increased numbers of reactive astrocytes and activated microglial cells [21,22]. A reduced immune defense was also seen in phenotypic SCA3 mice [23]. Overexpression of cystathionine γ-lyase decreases oxidative stress and dampens the immune response, which could improve SCA3-associated fly eye degeneration [24]. In addition, neuropeptide Y ameliorates neuropathology and motor deficits via upregulating brain derived neurotrophic factor (BDNF) and reducing neuroinflammation markers IL-6 and induction of brown adipocytes 1 (Iba1) in SCA3 mouse models [25].
In IL-1β signaling, the inhibitor of kappa B (IκBα) protein inactivates the NF-κB transcription factor (P65/P50 heterodimer) by masking the nuclear localization signal of NF-κB and keeping it sequestered in an inactive state in the cytoplasm [26]. Specifically, IκB kinase (IKK) phosphorylates the inhibitory IκBα protein [27], resulting in the dissociation of IκBα from NF-κB. NF-κB then migrates into the nucleus and activates the expression of pro-inflammatory cytokines and chemokines, such as tumor necrosis factor (TNF)-α, IL-6 and C-C motif chemokine ligand 2 (MCP1) [28]. In addition, c-Jun N-terminal kinase (JNK)/Jun proto-oncogene, AP-1 transcription factor subunit (JUN) and mitogenactivated protein kinase 14 (P38)/signal transducer and activator of transcription 1 (STAT1) are two other transduction pathways downstream to IL-1β and TNFα signaling, activated to up-regulate the synthesis and secretion of inflammatory factors [29,30].
Protein aggregation, oxidative stress and neuroinflammation are common themes in neurodegenerative diseases including polyQ SCAs and Alzheimer's disease (AD). Small heat shock proteins interact with misfolded protein aggregates, like Aβ aggregates in AD and polyQ aggregates in SCAs, to reduce the toxicity or increase the clearance of these protein aggregates [31]. To search for polyQ SCAs-modifying interventions targeting neuroinflammation, four in-house or outsourcing compounds activating molecular chaperone heat shock protein family B (small) member 1 (HSPB1) to reduce Aβ or Tau protein misfolding and aggregation were tested in this study: indole compound NC009-1 (C 19 [36,37]. In addition, NC009-1 could reduce SCA17 polyQ aggregation by enhancing expression of HSPB1 chaperone [38]. We examined the anti-inflammatory effects of these four compounds on human HMC3 microglia [39] and SH-SY5Y cells with inducible SCA3 ATXN3/Q 75 -GFP expression, which we have established previously [40]. We also explored if these four compounds exert their effects via targeting the IL-1βand TNF-α-mediated IκBα/P65, JNK/JUN and/or P38/STAT1 pathways.

Anti-inflammatory activity of the tested compounds on human HMC3 microglia
The anti-inflammatory responses of these compounds were examined using interferon (IFN)-γ stimulated HMC3 microglial cells [45] (Figure 2A). Exposure of HMC3 cells to IFN-γ resulted in increased expression of CD68 molecule (CD68) and major histocompatibility complex II (MHCII) ( Figure 2B). The production of nitric oxide (NO) in the cultured medium was compounds against HMC3 and SH-SY5Y cells using MTT viability assay. Cells were treated with 0.1−100 μM tested compounds and cell proliferation was measured after 28 h of treatment in HMC3 cells or 6 days of treatment in SH-SY5Y cells (n = 3). The IC 50 of each compound was shown under the columns. To normalize, the relative viability in untreated cells is set as 100%. (C) Radical-scavenging activity of these compounds (10−160 μM) on DPPH (n = 3). (D) Molecular weight (MW), hydrogen bond donor (HBD), hydrogen bond acceptor (HBA), calculated octanol-water partition coefficient (cLogP), polar surface area (PSA), and predicted blood-brain barrier (BBB) score of these compounds.

DISCUSSION
Several lines of evidence have shown that increased oxidative stress and decreased anti-oxidative response play a crucial role in the pathogenesis of SCA3 [13-15, 19, 20]. Increased pro-inflammatory cytokines, reactive astrocytes and activated microglia have been found in pons of SCA3 patients and mice [21,22,25]. Aberrant immune responses were also demonstrated in SCA3 mice and fly [23,24]. Inflammation and microglial activation have been shown to contribute to neurotoxicity in HD [50][51][52][53], another polyQ-mediated disease. Substantial evidence has also shown that oxidative stress and inflammation interplay to confer detrimental effects on neurons [54]. Therefore, agents targeting both oxidative and inflammatory pathways may serve as a good candidate for treating diseases such as polyQ diseases including SCA3, where both inflammation and increased oxidative stress play a pivotal role in pathogenesis.
In this study, we showed anti-inflammatory, antioxidative and/or neuroprotective effects of NC009-1, AM404, VB-037 and LM-031. The anti-inflammatory effects of the tested compounds were demonstrated by using IFN-γ-stimulated human HMC3 microglia, where these compounds significantly decreased release of proinflammatory cytokines, IL-1β, TNF-α and IL-6. We then applied HMC3 conditioned medium to ATXN3/Q 75 -GFP SH-SY5Y cells to provoke inflammation-induced damaging effects including reduced cell viability and neurite outgrowth, and increased aggregation, caspase1 activity and oxidative stress, as evidently shown in Figure 4. It may be better to show if the tested compounds have effect on activation of HMC3 cells expressing ATXN3. However, this study is mainly focusing on if exogenous inflammatory stimuli exaggerate the damage of expanded polyQ on neurons and if the tested compounds rescue the cytotoxicity of the cytokines released from IFN-γ-activated HMC3 cells. Our results are in accordance with that addition of TNF-α and IL-1β induced toxicity and apoptosis of primary cortical neurons from a HD mouse model and neurons derived from HD induced pluripotent stem cells [52]. The neuroprotection effects were further shown in ATXN3/Q 75 -GFP SH-SY5Y cells inflamed by addition of CM/+IFN-γ, where NC009-1, AM404, VB-037 and LM-031 significantly increased cell viability and decreased caspase 1 activity, NC009-1, AM404 and LM-031 reduced oxidative stress and rescued the deficits of neurite outgrowth, while NC009-1, VB-037 and LM-031 ameliorated aggregation. It is noted that AM404 had significant effects on neurite length/process/branch, caspase 1 activity/LDH release and ROS level, whereas it did not reduce aggregation. The similar effects have been shown in previous studies in polyQ diseases, where treatments or compounds had significant neuroprotection, but did not effectively inhibit aggregation, which suggests the protection is acting on downstream pathological processes secondary to toxic fragments of polyQ-containing protein rather than on aggregate-inhibition [55]. Furthermore, several studies have suggested that it is the soluble fragmented protein containing expanded polyQ that is toxic rather than aggregates and aggregate-reduction alone does not necessarily rescue neurotoxicity [56][57][58].
Treatments targeting inflammation for SCA3 have been rarely reported, although several studies have shown anti-inflammatory strategies are beneficial to other polyQ diseases [52,[65][66][67]. Hsiao and colleagues showed that inhibition of TNF-α improved motor function, reduced caspase activation, diminished the aggregates, increased neuronal density and decreased gliosis in the brains of R6/2 HD mice [52]. Aikawa and colleagues demonstrated that genetic ablation of myeloid differentiation factor 88 (Myd88), a major adaptor molecule essential for Toll-like receptor (TLR) signaling, ameliorated Purkinje cell loss and partially rescued motor impairments in a SCA6 mouse model [65]. Yang and colleagues also found that the NF-κB was activated in SCA17 knock-in mice and blocking NF-κB signaling in astrocytes ameliorated neurodegeneration [66]. Recently, Dubey and Tapadia have found that expanded polyQ triggered antimicrobial peptides (AMPs) expression and JNK activation, whereas Yorkie, the co-activator of the Hippo pathway, down-regulated AMPs and p-JNK to rescue apoptosis and mitigated polyQ-mediated toxicity in the eye of polyQ-expressing fly [67]. These studies suggest that enhanced inflammatory response contributes to polyQ mediated neurodegeneration and agents targeting inflammation may serve as potential therapeutics for polyQ diseases.
ROS have been reported to activate extracellular signalregulated kinases (ERKs), JNKs and P38, but the mechanisms by which ROS can activate these kinases are unclear [68]. Our study results also show that oxidative stress is increased and JNK and P38 are activated in inflamed neurons expressing expanded polyQ, supporting the proposal that ROS may also contribute to IL-1β-and TNF-α-mediated inflammation. The treatments of test compounds decrease inflammation and/or reduce oxidative stress, both of which may contribute the rescue of neurodegeneration.
Our study results demonstrated low cytotoxicity and high predicted BBB scores of all the tested compounds, suggesting their potential of serving as a treatment for neurodegenerative diseases including SCA3. Among them, NC009-1 [33,38], AM404 [69] and LM-031 [37] have been tested for in vivo usages. The known mechanisms of action of these compounds in neuroprotection are summarized below. In addition to up-regulating HSPB1 chaperone, NC009-1, VB-037 and LM-031 also displayed chemical chaperone-like activity in thioflavin T assay of Aβ aggregation [35,36,70]. NC009-1 has been shown previously to have aggregation-reducing and neuroprotection effects by activating HSPB1 to increase pro-aggregated ΔK280 Tau RD solubility and promote neurite outgrowth in tauopathy cell model [32]. Also by increasing HSPB1 expression, NC009-1 mitigated the increased BH3 interacting domain death agonist (BID), cytochrome c release, and caspase 3 activation to reduce polyQ aggregation and apoptosis in SCA17 TBP/Q 79 cells, as well as ameliorated behavioral deficits in SCA17 TBP/Q 109 transgenic mice [38]. Moreover, NC009-1 upregulated apolipoprotein E (APOE) and tropomyosin receptor kinase A (TRKA) expression to improve neurite AGING outgrowth in Aβ-GFP SH-SY5Y cell, as well as to reduce hippocampal/cortical Aβ and Tau levels and ameliorate cognitive deficits in hyperglycemic APP Swe / PS1 M146V /Tau P301L triple transgenic mice [33]. Here, we for the first time show the anti-inflammatory effect of NC009-1 to provide neuroprotection. AM404, an anandamide transport inhibitor, was previously selected through virtual screening compound databases to search for compounds which act as a glycogen synthase kinase-3β (GSK-3β) kinase inhibitor [34]. Through increasing phospho-GSK-3β (Ser9) expression to reduce Tau phosphorylation, AM404 enhanced HSPB1 and GRP78 (glucose-regulated protein, 78 kDa) expression, increased pro-aggregated Tau solubility, and promoted neurite outgrowth in ∆K280 Tau RD AD cell model [34]. Through increasing the phosphorylation of AKT (AKT serine/threonine kinase 1) and GSK-3β, AM404 at low dose ameliorated cognitive deficit and reduced Aβ, Tau hyperphosphorylation, and inflammation in hyperglycemic 3×Tg-AD mice [69]. AM404 also inhibits directly Ca 2+ flux of L-type voltage-dependent Ca 2+ channels [71]. AM404 can reduce allodynia in a neuropathic pain model via cannabinoid CB1 receptor activation [72]. Attenuation of lipopolysaccharide (LPS)induced increases in IL-1β and IL-6 by AM404, mediated through the cannabinoid CB1 receptor, has also been demonstrated [73]. Our results further support its antiinflammatory effect, but in a SCA3 model. Recently our group has shown the anti-aggregation, anti-oxidative, and neuroprotective effects of LM-031, a novel derivative of chalcone-coumarin, against Aβ or Tau toxicity through activation of the HSPB1 chaperone, NRF2 (nuclear factor, erythroid 2 like 2)/NQO1 (NAD(P)H quinone dehydrogenase 1)/GCLC (glutamate-cysteine ligase catalytic subunit) pathway, and CREB (cAMP-response element binding protein 1)/BDNF (brain derived neurotrophic factor)/BCL2 (BCL2 apoptosis regulator) pathway [36,37]. However, the anti-inflammatory effect of LM-031 is for the first time demonstrated in the present study. We have also previously shown that VB-037, a novel quinoline compound, attenuated LPS/IFN-γinduced activation of BV-2 microglia and diminished LPS/IFN-γ-induced increase of caspase 1 activity, expression of IL-1β, and phosphorylation of P38, JNK and JUN to protect Aβ-GFP-expressing SHSY5Y cells against inflammatory damage [35]. Here, we again show its protection effect via anti-inflammatory action on another neurodegenerative disease model. However, it should be addressed that future studies in SCA3 animal models are warranted to further consolidate the neuroprotection effects of the tested compounds.

Radical-scavenging assay
The DPPH radical-scavenging activity was measured in an ethanol mixture containing 200 µM DPPH (Sigma-Aldrich) radical solution and the tested compounds (10-160 µM). The solution was vortexed and incubated for 30 min at room temperature. The scavenging capacity was measured by monitoring the decrease in absorbance at 517 nm with a microplate spectrophotometer (Multiskan GO, Thermo Fisher Scientific). The half maximal effective concentrations (EC 50 ) for inhibition of the formation of DPPH radicals were calculated using the interpolation method.

Detection of inflammatory mediators
HMC3 cells were plated into 6-well (2 × 10 5 /well) dishes, grown for 20 h and treated with the tested compounds (1-10 µM) for 8 h followed by IFN-γ (100 ng/ml) stimulation. The release of NO in cell culture medium was evaluated by Griess assay according to manufacturer's protocol (Thermo Fisher Scientific). In addition, the levels of IL-1β, TNF-α and IL-6 in medium pre-treated with 10 µM compound were determined using Human Instant ELISA TM Kit following the manufacturer's protocol (Invitrogen). The optical density at 450 nm was detected using Multiskan GO spectrophotometer.

High content ATXN3 polyQ aggregation and neurite outgrowth analyses
ATXN3/Q 75 -GFP SH-SY5Y cells were plated on 24-well (1 × 10 4 /well) dishes, and retinoic acid (10 µM; Sigma-Aldrich) was added to initiate neuronal differentiation. On the second day, cells were treated with the tested compound (10 μM) for 8 h before ATXN3-GFP expression induction by adding doxycycline (5 µg/ml). For comparison, cells without inducing ATXN3/Q 75 -GFP expression were included. The cells were kept in the medium containing retinoic acid, doxycycline and test compound for 6 days. For the cells with inflammatory stimulation, HMC3 conditioned medium with (CM/+ AGING IFN-γ) or without (CM/-IFN-γ) IFN-γ stimulation was added at a 1:1 ratio in the last two days. On the eighth day, cells were stained with Hoechst 33342 (0.1 µg/ml; Sigma-Aldrich) for 30 min, and images of the cells were automatically obtained using an ImageXpressMICRO high content analysis (HCA) system (Molecular Devices, San Jose, CA, USA). Excitation/emission filters were at 482/536 and 377/447 nm for enhanced GFP and Hoechst 33342, respectively. Aggregation was determined by Transfluor technology [75] based on GFP fluorescence intensity. Neurite length, process and branch of ATXN3/Q 75 -GFP-expressing cells were analyzed by using Metamorph microscopy automation and image analysis software (neurite outgrowth application module, Molecular Devices). In addition, cells were fixed, permeated and stained with neuronal class III β-tubulin (TUBB3) antibody (1:1000; Covance, Princeton, NJ, USA), followed by anti-rabbit Alexa Fluor ® 555 antibody (1:1000; Thermo Fisher Scientific) for neurite outgrowth analysis. To quantify neurite outgrowth, microscopic images were segmented with multi-colored mask to assign each outgrowth to a cell body for quantification. In general, 10 4 cells in each biological replicate were analyzed.

Cell viability, caspase 1 activity, and LDH release assays
Cell viability was assayed by propidium iodide (PI) staining. Briefly, retinoic acid-differentiated ATXN3/ Q 75 -GFP SH-SY5Y cells (2 × 10 5 on 6-well dishes) were pretreated with the tested compounds, induced ATXN3/ Q 75 expression, and inflamed with HMC3 conditioned medium (CM/+IFN-γ or CM/-IFN-γ) as described. On the eighth day, the cells were stained with PI (0.6 μg/ml; Sigma-Aldrich) and Hoechst 33342 (0.1 μg/ml) for 30 min, and images of the cells were automatically obtained using the HCA, with 535 nm excitation and 617 nm emission filters for PI. For caspase 1 activity assay, the cells were lysed by repeated freeze-thaw and supernatants collected after centrifugation at 12,000 × g for 10 min. Caspase 1 activity was measured with the caspase 1 assay kit based on the cleavage of substrate YVAD-AFC according to the manufacturer's instructions (BioVision, Milpitas, CA, USA). The absorbance was read using FLx800 microplate reader with excitation at 400 nm and emission at 505 nm. For LDH release assay, cell culture media were collected on day 8 and the release of LDH was examined by using LDH cytotoxicity assay kit (Cayman, Ann Arbor, MI, USA). The absorbance was read at 490 nm with Multiskan GO microplate reader.

ROS analysis
Retinoic acid-differentiated ATXN3/Q 75 -GFP SH-SY5Y cells (8 × 10 3 on 96-well dishes) were pretreated with the tested compounds, induced ATXN3/Q 75 expression, and inflamed with HMC3 conditioned medium as described. On the eighth day, fluorogenic CellROX deep red reagent (5 μM; Invitrogen) and Hoechst 33342 (0.1 μg/ml) were added to the cells and incubated at 37°C for 30 min. Images of the cells were obtained and analyzed using the HCA, with 640 nm excitation and 665 nm emission filters for CellROX deep red reagent.
For filter trap assay, protein (20 µg) was diluted in 2% SDS in PBS and filtered through a cellulose acetate membrane (0.2 μm pore size; Merck, Kenilworth, NJ, USA) pre-equilibrated in 2% SDS in PBS on a dot-blot filtration unit (Bio-Rad Laboratories, Hercules, CA, USA). After washing with 2% SDS buffer and blocking in PBS containing 5% nonfat dried milk, the cellulose acetate membrane was probed with anti-GFP antibody (1:1000; Santa Cruz Biotechnology) and the immune complexes on the membrane were detected as described.

Statistical analysis
For each data set, three independent experiments were performed and data were expressed as the means ± AGING standard deviation (SD). Differences between groups were evaluated by Student's t test (comparing two groups) or one-way analysis of variance with a post hoc Tukey test where appropriate (comparing several groups). All P values were two-tailed, with values lower than 0.05 to be considered statistically significant.