Shaoyao Gancao Tang (SG-Tang), a formulated Chinese medicine, reduces aggregation and exerts neuroprotection in spinocerebellar ataxia type 17 (SCA17) cell and mouse models

Spinocerebellar ataxia (SCA) type 17 is an autosomal dominant ataxia caused by expanded polyglutamine (polyQ) tract in the TATA-box binding protein (TBP). Substantial studies have shown involvement of compromised mitochondria biogenesis regulator peroxisome proliferator-activated receptor gamma-coactivator 1 alpha (PGC-1α), nuclear factor erythroid 2-related factor 2 (NRF2), nuclear factor-Y subunit A (NFYA), and their downstream target genes in the pathogenesis of polyQ-expansion diseases. The extracts of Paeonia lactiflora (P. lactiflora) and Glycyrrhiza uralensis (G. uralensis) have long been used as a Chinese herbal medicine (CHM). Shaoyao Gancao Tang (SG-Tang) is a formulated CHM made of P. lactiflora and G. uralensis at a 1:1 ratio. In the present study, we demonstrated the aggregate-inhibitory and anti-oxidative effect of SG-Tang in 293 TBP/Q79 cells. We then showed that SG-Tang reduced the aggregates and ameliorated the neurite outgrowth deficits in TBP/Q79 SH-SY5Y cells. SG-Tang upregulated expression levels of NFYA, PGC-1α, NRF2, and their downstream target genes in TBP/Q79 SH-SY5Y cells. Knock down of NFYA, PGC-1α, and NRF2 attenuated the neurite outgrowth promoting effect of SG-Tang on TBP/Q79 SH-SY5Y cells. Furthermore, SG-Tang inhibited aggregation and rescued motor-deficits in SCA17 mouse model. The study results suggest the potential of SG-Tang in treating SCA17 and probable other polyQ diseases.


INTRODUCTION
A group of inherited neurodegenerative diseases including Huntington's disease (HD), spinobulbar muscular atrophy (SBMA), hereditary spinocerebellar ataxia (SCA) types 1, 2, 3, 6, 7 and 17, and dentatorubral-pallidoluysian atrophy (DRPLA) are caused by an expansion of unstable trinucleotide (CAG) repeats encoding expanded polyglutamine (polyQ) tracts [1]. Among them, SCAs are characterized by cerebellar dysfunction alone or in combination with other neurological abnormalities. SCA17 is an autosomal dominant ataxia caused by an allele containing expanded repeats longer than 43 in the TATA-box binding protein (TBP) gene, a transcription initiation factor [2]. In addition to progressive ataxia, the features of this rare disease also include seizure, cognitive dysfunctions, psychiatric symptoms, and Shaoyao Gancao Tang (SG-Tang), a formulated Chinese medicine, reduces aggregation and exerts neuroprotection in spinocerebellar ataxia type 17 (SCA17) cell and mouse models AGING pyramidal and extrapyramidal signs such as spasticity, dystonia, chorea, and parkinsonism. Identification of disease-causative genes has led to the development of model systems for exploration of disease mechanisms and discovery of drug therapy. A prominent pathological feature of the polyQ-mediated diseases is the intranuclear and cytoplasmic accumulation of aggregated polyQ proteins in neurons. Substantial evidence has shown that polyQ-mediated diseases are the result of a toxic gain of function that occurs at the protein level. The presence of expanded polyQ proteins leads to transcriptional dysregulation, mitochondrial damage, oxidative stress, defect in axonal transport, chaperone-proteasome impairment, autophagolysosome dysfunction, and unfolded protein response (UPR) in endoplasmic reticulum (ER) [3,4]. Among them, transcriptional dysregulation is one of the main pathogenic mechanisms of SCA17 [5]. We and Huang et al. have shown that TBP-containing expanded polyQ interacted aberrantly with nuclear factor-Y (NFY) subunit A (NFYA), which would result in reduced heat shock 70 kDa protein 5 (HSPA5) expression [6,7], a major ER chaperone and master regulator of UPR [8].
We have also previously shown downregulation of HSPA5 in lymphoblastoid cells of SCA17 patients, suggesting that decreased ER chaperones may contribute the cell dysfunction of SCA17 [9]. Moreover, elimination of HSPA5 in Purkinje cells leads to accelerated cerebellar degeneration in a mouse model, suggesting an important role of HSPA5 in maintaining neuronal survival [10]. Thus NFY-HSPA5 may serve as a potential target for development of therapeutics for SCA17.
A polyQ mutation can induce reactive oxygen species (ROS) that directly contribute to cell death in vitro [11,12] and in vivo [13,14]. Antioxidants have been shown to attenuate aggregation and cell death in SCA1, SCA3, and HD models [15][16][17][18]. The nuclear factor erythroid 2related factor 2 (NRF2) and the antioxidant response elements (AREs) signaling pathway is regarded as the major response in the cell to protect against oxidative stress [19]. NRF2 binds to AREs and recruits the general transcriptional machinery for ARE-dependent gene expression when cells respond to oxidative stress. The target genes upregulated by NRF2 including heme oxygenase (decycling) 1 (HMOX1), NAD(P)H dehydrogenase, quinone 1 (NQO1), glutamate-cysteine ligase catalytic subunit (GCLC), and glutathione Stransferase pi 1 (GSTP1) are belonging to the endogenous phase II antioxidative enzymes. Mutant huntingtin and ataxin 3 impaired NRF2 activation and decreased the ARE binding activity, which contributed to mitochondrial dysfunction and enhanced susceptibility to oxidative stress in HD and SCA3 cell models [18,20].
Peroxisome proliferator-activated receptor gamma, coactivator 1 alpha (PGC-1α) is a known regulator of mitochondrial biogenesis and antioxidative response genes including superoxide dismutase 2, mitochondrial (SOD2) and cytochrome c, somatic (CYCS). PGC-1α null mice developed spongiform neurodegeneration in selective brain areas, which suggests the direct role of PGC-1α in neuronal survival [21]. PGC-1α was recently found also to upregulate the NRF2 transcription [22]. Transcriptional repression of PGC-1α by mutant huntingtin resulting in mitochondrial abnormality and neurodegeneration has also been shown in a HD mouse model, suggesting that agents enhancing the transcriptional activity of PGC-1α may be potential therapeutics for HD [23,24]. Indeed, previously we have shown that Glycyrrhiza inflata herb extract and its constituents, licochalcone A and ammonium glycyrrhizinate, activated PGC-1α activity and NRF2-ARE signaling to increase mitochondrial biogenesis, decrease oxidative stress, and reduce aggregate formation in SCA3 cellular models [18]. Therefore, we propose that PGC-1α and NRF2 pathways may be also compromised in SCA17 and compounds that enhance PGC-1α and/or NRF2 expression may have potential to treat SCA17.

SG-Tang ameliorated behavioral deficits and reduced aggregation in SCA17 TBP/Q109 TG mice
Lastly, we examined the neuroprotective effect of SG-Tang on an established SCA17 TBP/Q109 TG mouse model [33]. The experimental timeline is shown in Figure 5A. SG-Tang (0.4%) was supplied in the drinking water of SCA17 mice from 10 to 21 weeks old. To identify the effect of SG-Tang on motor function, rotarod, locomotor, and footprint tasks were performed. On the rotarod task to test motor coordination of mice, the SCA17 TG mice showed significantly decreased latency to fall at 10-21 weeks of age when compared with the WT mice (111 s-140 s versus 284 s-337 s, P < 0.001). However, the SG-Tang-treated TG mice showed significantly increased latency compared with the vehicle-treated TG mice at 17 weeks (159 s versus 123 s, P = 0.030) and 21 weeks (201 s versus 121 s, P < 0.001) of age ( Figure 5B). On the locomotor task to test motor activity of mice, the SCA17 TG mice showed significantly increased total distance traveled in an open field (hyperactivity) at 10 weeks (3814 cm versus 2627 cm, P < 0.001) and 19 weeks (4003 cm versus 2830 cm, P < 0.001) of age when compared with the WT mice. However, the hyperactivity of TG mice was ameliorated when they were 19 weeks old after SG-Tang treatment (3576 cm versus 4003 cm, P = 0.015) ( Figure 5C). On the footprint task to test gait coordination of mice, the SCA17 TG mice showed significantly increased print position at 10 weeks (1.7 cm versus 1.5 cm, P = 0.022), 19 weeks (1.8 cm versus 1.1 cm, P < 0.001), and 22 weeks (1.9 cm versus 1.2 cm, P < 0.001) of age when compared with the WT mice. However, SG-Tang treatment significantly improved print position in the TG mice compared with vehicle treatment at 22 weeks of age (1.6 cm versus 1.9 cm, P = 0.048) ( Figure 5D). It is noticed that the rotarod performance of SG-Tang treated TG mice at 19 weeks of age showed a trend toward prolonged latency on rod compared with the vehicle-treated TG mice, but the improvement did not reach significance. This result may be due to the relatively small sample size with large variations in motor performance of tested TG mice. These results suggest that SG-Tang treatment ameliorated behavioral deficits in SCA17 TBP/Q 109 TG mice.

AGING
In humans, the cerebellum plays an important role in motor control and is most adversely affected by SCAs. Calbindin in cerebellar Purkinje cells is a crucial determinant in motor functions [38]. We thus examined the cerebellar Purkinje cells in vehicle-or SG-Tangtreated SCA17 TG mice at 22 weeks of age using calbindin staining. As shown in Figure 5D, the disrupted alignment and structure of Purkinje cells in SCA17 TBP/Q109 TG mice was improved with the SG-Tang treatment, accompanied with reduced TBP aggregation (from 90% to 74%, P = 0.020).

DISCUSSION
SCA17 is one of the neurodegenerative diseases caused by accumulation of expanded polyQ, the effective treatments for which are lacking. The elucidation of pathogenic mechanisms is important for development of therapeutics. Studies have shown PGC-1α/SOD2/CYCS and NRF2/GCLC/NQO1 pathways are involved in pathogenesis of polyQ-mediated diseases including HD and SCA3 [18,20,23,24], and impaired NFYA/HSPA5 expression has been shown in SCA17, including cellular and animal models and lymphoblasts from patients [6,7,9]. PGC-1α/SOD2/CYCS, NRF2/GCLC/NQO1, and NFYA/HSPA5 pathways may therefore serve as the therapeutic targets for SCAs. SG-Tang is composed of P. lactiflora and G. uralensis, both of which have been used as a traditional medicine to treat certain illnesses for their antioxidative, anti-inflammatory and cytoprotective properties. P. lactiflora or its constituent paeoniflorin have been shown to exert the protective effects in different rodent models of Alzheimer's disease, the 6-OHDA-lesioned rodent model of Parkinson's disease, and the SCA3 cell model [28,[39][40][41][42]. G. uralensis and its bioactive compounds have displayed anti-inflammatory and antioxidative activities in vitro and in vivo [43][44][45]. SG-Tang has been used to inhibit inflammatory chemokine production in HaCaT cells [46]. Due to the multiple constituents in P. lactiflora and G. uralensis and their effects on different pathways, we set out to investigate if SG-Tang is beneficial to SCA17, the pathogenic pathways of which are also multiple. We also explored how SG-Tang acts on the underlying pathogenic pathways.
It is evident that the SG-Tang has a low cytotoxicity on uninduced and induced TBP/Q79-GFP 293 cells (as indicated by IC50 > 1 mg/ml in Figure 1) with the effective aggregation-inhibitory concentrations ranged from 0.001 to 100 μg/ml. Since the expression of TBP was not changed by SG-Tang, the aggregationinhibitory effect may be related to decreased ROS rather than reduced TBP expression. Similarly, a very low cytotoxicity of SG-Tang on uninduced TBP/Q79-GFP SH-SY5Y cells at doses 0.1−100 μg/ml was shown ( Figure 2A). The aggregation-inhibitory and neurite outgrowth promoting effects of SG-Tang at dose 100 μg/ml were further demonstrated in TBP/Q79-GFPexpressing SH-SY5Y cells ( Figure 2D). Our in vitro result of low cytotoxicity of SG-Tang on TBP/Q79-GFPexpressing 293 and SH-SY5Y cells is consistent with previous studies suggesting its low cytotoxicity on ΔK280 tauRD-expressing 293 and SH-SY5Y cells [31] or HaCaT human keratinocytes [46], but some adverse effects of the main constituents have been shown. For example, paeoniflorin and albiflorin, two constituents of P. lactiflora, markedly suppressed the CYP3A4 and CYP2D6 activity [47]. In addition, paeoniflorin caused some allergic skin reactions [48]. G. uralensis was reported to significantly alter the metabolic rates of anticoagulants or antiplatelet drugs [49]. Glycyrrhizin, a main constituent of G. uralensis, at 1200-2600 mg/kg through oral intake displayed hypertension, polydipsia, (A) Experimental flowchart. TBP/Q79-GFP SH-SY5Y cells were plated on dishes with retinoic acid (RA, 10 µM) added at day 1. Next day, cells were transfected with siRNA (100 nM; PGC-1α, NRF2, NFYA, or scrambled control). At 24 h post-transfection, SG-Tang (100 µM) was added to the cells for 8 h followed by inducing TBP/Q79-GFP expression (+ Dox, 5 µg/ml) for 6 days. Then, the cells were collected for PGC-1α, NRF2, and NFYA protein analysis (by Western blot, GAPDH as a loading control), or stained with Hoechst 33342 for aggregation, neurite outgrowth, process, and branch analyses (by HCA). (B) Western blot analysis of PGC-1α, NRF2, and NFYA protein levels in SG-Tang-treated cells transfected with PGC-1α, NRF2, NFYA, or scrambled control siRNA. To normalize, the relative PGC-1α, NRF2, or NFYA level of uninduced cells was set as 100%. P values: induced versus uninduced cells ( # : P < 0.05 and ## : P < 0.01); SG-Tang-treated versus untreated cells (*: P < 0.05 and **: P < 0.01); or PGC-1α, NRF2, or NFYA siRNA versus scrambled control siRNAtransfected cells ( & : P < 0.05) (n = 3). (C) Representative microscopic images of TBP/Q79-GFP-expressing SH-SY5Y cells transfected with scrambled control siRNA and SG-Tang-treated cells transfected with scrambled control, PGC-1α, NRF2, or NFYA siRNA. Nuclei were counterstained with Hoechst 33342 (blue, left) and aggregates were marked in red (middle). Shown left are images of the neurites and cell bodies outlined by red color for neurite outgrowth, process, and branch quantification. (D) Aggregation and neuronal outgrowth, process, and branch assays of SG-Tang-treated TBP/Q79-GFP-expressing SH-SY5Y cells transfected with scrambled control, PGC-1α, NRF2, or NFYA siRNA. To normalize, the relative aggregation, outgrowth, process, or branch of scrambled control siRNAtransfected cells without SG-Tang treatment was set as 100%. AGING bracycardia, and behavioral changes in rats [50]. These potential side effects should be further investigated by in vivo studies.
NRF2, a transcription factor that regulates expression of many cytoprotective genes, mediates redox adaptations to exercise as well as defendant responses to oxidative stress [19,54]. Since oxidative stress may lead to neurodegeneration, NRF2 activators have been suggested to be a therapeutic target for the treatment of neurodegenerative diseases [55]. NRF2 enhancers have been applied to treat different models of polyQ diseases [18,[56][57][58]. Our study results further support the potential of NRF2 activators for treating SCA17 and other polyQ diseases.
The trimeric NFY is formed by three subunits, NFYA, NFYB, and NFYC. The sequence specific interactions of the complex are made by the NFYA subunit, suggesting its role as the regulatory subunit [59]. Recent studies have shown that mutant huntingtin aggregates sequester NFYA and NFYC leading to the reduction of HSPA1A gene expression in the brain of a HD mouse model, indicating NFY components as modulators of the HD pathological process [60]. Similarly, mutant TBP with expanded polyQ interacts aberrantly with NFYA, which impairs the transcriptional function [6,7,61]. Because NFYA is a key regulator of HSPA5 transcription [62] and downregulated HSPA5 has been shown in SCA17 lymphoblasts [9], treatments through enhancing NFYA expression may be beneficial to SCA17. Our results demonstrate that SG-Tang rescues downregulated NFYA and HSPA5 in SCA17 cell model.
According to the antiaggregation effect on SCA17 SH-SY5Y cells (Fig. 2), the effective dosage for SG-Tang was 100 μg/ml. To assume the bioactivity of SG-Tang as 10% and protein binding rate as 90%, the dosage for treating animals can be roughly estimated to be 10 mg/25 g/day or 0.4 g/kg/day per mouse. The SCA17 transgenic mice performed poorly on an accelerating rotarod around 2 and 4 months of age and showed continued declines in latency on the rod [33]. Thus 10 weeks of age was selected to receive 0.4% SG-Tang in drinking water for 12 weeks. Average daily consumption of water for these mice was 3 ml. The dosage of 3 ml 0.4% SG-Tang administered to SCA17 mouse in the present study would be equivalent to 0.48 g/kg/day per mouse. The dosage of SG-Tang for a 50 kg human adult to alleviate pain recommended by the Sun Ten Pharmaceutical Co. Ltd. is 4 g 3 times daily, or roughly 0.24 g/kg/day. The dosage of SG-Tang used to treat SCA17 mice was about twice amount used to relieve pain in human. Further experiments will need to be carefully designed to examine the possible adverse effects of SG-Tang in SCA17 mice and optimal dosage for future clinical application.
There are several limitations in our study. Since there are multiple constituents in SG-Tang, the main components or compounds responsible for the protective effects in our cellular models are not known. However, paeoniflorin, the bioactive compound of P. lactiflora demonstrating protective effects through activating the NRF2/ARE pathway in other cell models of oxidative injury has been shown [63,64]. The bioactive compounds of G. uralensis such as licochalcone A, glycyrrhetinic acid, liquiritigenin, isoliquiritigenin and liquiritin were found to be all potent NRF2 or NQO1 inducers [18,65,66]. Paeoniflorin also protected Aβ25-35-expressed SH-SY5Y cells from cytotoxicity by preventing mitochondrial dysfunction [67]. Previously we also showed that licochalcone A and ammonium glycyrrhizinate could activate PGC-1α activity in a SCA3 cell model [18]. We therefore propose that the effects of PGC-1α, NRF2, and NFYA enhancement may be mediated by paeoniflorin, licochalcone A, ammonium glycyrrhizinate, glycyrrhetinic acid, liquiritigenin, isoliquiritigenin, or liquiritin in SG-Tang. Future studies examining how individual bioactive compounds of SG-Tang exert the specific effects on the PGC-1α/SOD2/CYCS, NRF2/ GCLC/NQO1, and NFYA/HSPA5 pathways are necessary to consolidate our results and clarify the therapeutic mechanism. Besides, there are protective effects of SG-Tang through other pathways not AGING examined in the present study, which requires further investigation in the future. Since the main pathology of SCA17 is in brain, the most important prerequisite for a good therapeutic agent for treating neurodegenerative diseases is its high blood brain barrier (BBB) permeability. Previous studies have shown a good BBB permeability of isoliquiritigenin and 18β-glycyrrhetinic acid, a major metabolite of glycyrrhizin in G. uralensis [68][69][70]. Our studies also have another limitation that the in vitro models did not show the BBB permeability of SG-Tang. It will facilitate proceeding of clinical trials of SG-Tang in SCA17 and other polyQ diseases if efficient BBB permeability of SG-Tang and its main bioactive compounds can be shown in the future animal studies. Finally, although we have shown the improved motor performance and decreased cerebellar aggregates in SCA17 mice treated with SG-Tang, we did not investigate how SG-Tang exerts its neuroprotection effect on mice. In the future, examining the PGC-1α/SOD2/CYCS, NRF2/GCLC/NQO1, and NFYA/ HSPA5 pathways in SCA17 mice treated with specific constituents of SG-Tang is warranted to uncover the underlying mechanisms.
In conclusion, we have provided evidence that PGC-1α/SOD2/CYCS, NRF2/GCLC/NQO1, and NFYA/ HSPA5 pathways are downregulated in SCA17 cell models. We further demonstrated that SG-Tang may serve as a protective agent for SCA17 via upregulating the PGC-1α/SOD2/CYCS, NRF2/GCLC/NQO1, and NFYA/HSPA5 pathways ( Figure 6). Based on these results, future work is warranted to uncover the main constituents in this Chinese medicine formula and its adverse effects on human health.

Formulated SG-Tang and its ingredients
SG-Tang (Code: 0703H), provided by Sun Ten Pharmaceutical Co. Ltd. (New Taipei City, Taiwan), is a formula in the traditional Chinese medical Shang Han Lun. SG-Tang consisting of two Chinese herbs Paeoniae Radix Alba and Glycyrrhizae Radix et Rhizoma (use Honey baked) at 1:1 (w/w) ratio. The botanical origins of each ingredient are Paeonia lactiflora Pall. and Glycyrrhiza uralensis Fisch., harvested in An Hui (Paeonia lactiflora Pall.) and Inner Mongolia (Glycyrrhiza uralensis Fisch.), China, respectively. These materials had passed the strict quality tests with methods and acceptance criteria based upon the requirements of Chinese pharmacopoeia. Quality control factors include macroscopic identification, microscopic identification, thin layer chromatographic identification, loss on drying, total ash, acid-insoluble ash, water soluble extract, dilute ethanol soluble extract, high performance liquid chromatography fingerprint, marker substance assay, residual pesticides, heavy metals, sulfur dioxide, aflatoxins, and microbiological contaminants. The SG-Tang is manufactured by decocting (boiled with one hour), concentrating, fluid-bed granulating, and sieving. Decocted intermediate is granulated with corn starch and cellulose as diluents and magnesium stearate as a glidant. The certificate of analysis for finished product includes physical description, thin layer chromatographic identification, loss on drying, total ash, acidinsoluble ash, water soluble extract, dilute ethanol soluble extract, particle size examination, weight AGING conformity, heavy metals (arsenic, cadmium, mercury, and lead) contaminants, and microbiological (total viable aerobic count, Escherichia coli, and Salmonella) contaminants with suitable specifications accordingly. The certificate of analysis is available on request.

TBP/Q 79 aggregation assay
GFP fluorescence was evaluated to reflect TBP aggregation in TBP/Q79-GFP-expressing 293 cells. Briefly, cells were plated on 96-well (2 × 10 4 /well) dishes, grown for 24 h and treated with different concentrations of SG-Tang (0.001−1000 μg/ml), or a positive control suberoylanilide hydroxamic acid (SAHA, 0.1 µM) (Cayman Chemical, Ann Arbor, MI, USA). After 8 h, doxycycline (10 µg/ml) and oxaliplatin (5 µM) (Sigma-Aldrich, St Louis, MO, USA) were added to the cells for 6 days to induce TBP/Q79-GFP expression and inhibit cell cycle progression [71], respectively. The cells were kept in the medium containing doxycycline, oxaliplatin and SG-Tang for 6 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) (482 nm excitation and 536 nm emission for enhanced GFP; 377 nm excitation and 447 nm emission for Hoechst 33342). Aggregation was determined by Transfluor technology [72] based on GFP fluorescence intensity. To quantify aggregation, the relative aggregation level in untreated cells was set as 100%. In addition, IC 50 value (half-maximal inhibitory concentration) for SG-Tang was evaluated based on the survived cell number.

SH-SY5Y TBP/Q79-GFP aggregation and neurite outgrowth assays
2 × 10 4 of TBP/Q79-GFP SH-SY5Y cells were seeded on a 24-well plate, and retinoic acid (10 µM; Sigma-Aldrich) was added to initiate neuronal differentiation. On the second day, cells were treated with SG-Tang (100 µg/ml) for 8 h before TBP/Q79-GFP expression induction by adding doxycycline (5 µg/ml). The cells were kept in the medium containing retinoic acid, doxycycline and SG-Tang for 6 days. On day 8, cells were stained with Hoechst 33342 (0.1 µg/ml) and the aggregation percentage was assessed by HCA as described. In addition, the morphologic differentiation of TBP/Q79-GFP-expressing cells was assessed by using Metamorph microscopy automation and image analysis software (neurite outgrowth application module, Molecular Devices). To quantify neurite outgrowth, the relative outgrowth in untreated cells was set as 100%.

RNA interference
To knock down the expression of specific genes in TBP/Q79-GFP SH-SY5Y cells, small interfering RNA (siRNA) targeting PGC-1α (sc-38884), NRF2 (sc-37030), NFYA (sc-29947) and a scrambled control (sc-37007) (Santa Cruz Biotechnology) were used. Cells were plated at a density of 8 × 10 5 /well on 6-well plates in the presence of retinoic acid on day 1 as described. On day 2, the cells were transfected with siRNA (100 nM) using T-Pro NTR II reagent (3 µl) (T-Pro Biotechnology, New Taipei City, Taiwan). Twenty-four hours post-transfection, the culture medium was changed and the cells were pretreated with SG-Tang (100 µg/ml) for 8 h followed by inducing TBP/Q79-GFP expression with doxycycline (5 µg/ml) for 6 days. Cells were then collected for PGC-1α, NRF2, and NFYA protein analysis or stained with Hoechst 33342 and analyzed for aggregation and neurite outgrowth as described.

SCA17 mice and SG-Tang treatment
Previously, SCA17 transgenic (TG) mice were established with Purkinje cell-specific Pcp2 promoter driving the expression of human TBP/Q109 [33]. SCA17 TG mice and their wild-type littermates (WT) were kept in individually ventilated cages (Lasco, Taipei, Taiwan) under controlled temperature (25 ± 2°C), humidity (50%), and normal light/dark (12h/12h) cycle. Mice were maintained ad libitum on food and water at the Animal House Facility of National Taiwan Normal University (NTNU). At age of 3 weeks old, both WT and TG mice were randomly divided into vehicle and SG-Tang groups (n = 18 in each group). From 10-21 weeks of age, mice in SG-Tang group received 0.4% SG-Tang in drinking water for 12 weeks. Behavioral analyses were performed during (rotarod task) and at the beginning and near the end (locomotor and footprint tasks) of the period to evaluate the treatment effect. All procedures were conducted in compliance with the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines and approved by the Institutional Animal Care and Use Committee of NTNU (Permit Number: 103002).

Rotarod, locomotor, and footprint tasks
The motor coordination of WT and TG mice was analyzed by rotarod (Ugo Basile, Comerio, VA, Italy) from 10 to 21 weeks of age. Mice were tested starting at 4 rpm and accelerating to 30 rpm over a period of 5 min and then maintaining a constant speed of 30 rpm for 1 min. The time (latency) it takes the mouse to fall off the rod was recorded. The latencies of four trials given on 2 days were averaged as the performance of each mouse.
For locomotor activity monitoring, 10-and 19-week-old mice were placed in the center of an open field apparatus (30 × 30 × 30 cm) and monitored under low lighting (2 lux) for 10 min. An EthoVision system (Noldus, Wageningen, Netherlands) was used to record and analyze the distance traveled.
Mouse footprints were monitored when the mice were 10, 19, and 22 weeks old using the CatWalk XT system (Noldus). Each mouse was allowed to walk three times on the glass plate, with paw prints recorded and analyzed using the CatWalk XT 9.1 software (Noldus).

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