Analysis of paired ATXN3-sgRNA/Cas9n activity in HEK293T cells
Paired sgRNAs can enhance Cas9n-mediated DSBs to generate highly specific genome-editing, which could reduce OTs and enhance HR modification . Considering for the distance and cleavage length of sgRNAs at the target gene, we designed a pair of sgRNAs (ATXN3_sgRNA1 and ATXN3_sgRNA2) for the upstream and downstream PAM regions of CAG repeat tract in exon 10 of ATXN3 (Fig. 1a and
Additional file 8: Table S1). Cells were transfected with plasmids expressing both wtCas9 protein and sgRNA (Fig. 1b). SgRNAs targeting CAG repeats of ATXN3 have been successfully constructed by Sanger sequencing (Fig. 1c). The first screening of sgRNA activity was performed in HEK293T cells. The transfection efficiency of HEK293T cells was 30~50% by immunofluorescence and flow cytometry after 24~48h transfection (Fig. 1d). The cellular genome was isolated after 48~72h transfection. PCR amplification of the CAG repeats in ATXN3, and subsequent T7EN1 assay resulted in multiple bands in both treated and untreated control cells, 630 amplified bands were produced in untreated cells. 116bp and 514bp bands appeared after sgRNA1 cleavage, while 253bp and 377bp bands were presented after sgRNA2 cleavage. The cleavage efficiency of sgRNA1 and sgRNA2 was 19.3% and 22.6% respectively (Fig. 1e). HEK293T cells contained 14 and 24 CAG duplicates in both alleles of ATXN3 (Additional file 1: Figure S1). After treated with sgRNA1 and sgRNA2, the PCR amplification of targeting site and sequencing analysis showed bimodal and mixed signals (Fig. 1f). These results indicated that the designed sgRNAs have significant cleavage efficiency in vitro experiments.
Gene correction of SCA3/MJD patient-derived iPSCs
To correct the disease mutation in SCA3/MJD-iPSCs and generate isogenic control lines, we adopt the CRISPR/Cas9 and Cre-loxP-mediated HR based genome-editing methods. We employed targeting donor construct (loxP-pGK-Puro-loxP), the donor cassette contains a 1919bp left arm and 3264bp right arm containing 17 CAG repeats based on the sgRNA1 cleavage site (intron 9), and successfully cloned into a pFlexible-DT vector (donor repair template), and puromycin-resistance gene (Puro) for positive clone screening (Fig. 2a). Firstly, we transfected 3ug paired sgRNAs/Cas9n (sgRNA1+sgRNA2) and 5ug donor repair template into the SCA3/MJD-iPSCs carrying 31/74 CAG expansions. The paired sgRNAs/Cas9n transfecting efficiency was 2.8% by flow cytometry (Fig. 2b). After post-electroporation for 72h, 300ng/ml Puro was added for selecting corrected clones. Targeted clones were selected for further culture in 13~18 days (Fig. 2c). Positive clones were identified by PCR using P1~P2 primers. The successfully corrected cell lines contained 510bp and 3090bp, and identified 4 corrected lines (C3, C11, C12, C13), all did not detect any mutant bands (Fig. 2d). In addition, the positive clones were further verified by PCR (P3~P4 primers), and only two cell lines (C3, C12) contained 3787bp target bands (Additional file 2: Figure S2). We confirmed that paired sgRNAs effectively targeted CAG expansions in exon 10 of ATXN3.
Successful correction of the mutant ATXN3 allele was verified by western blot using antibodies for ataxin3 protein (H9, MAB5360) (Fig. 2e). Of the 116 clones screened, 14 were targeted CAG expansions, of which 2 clones (C3 and C12) were confirmed by PCR screening (P1~P2 and P3~P4 primers) and western blot, accounting for about 1.7% of the screened positive clones. The HR rate is consistent with previously reported in HD-iPSCs  (Fig. 2f). Capillary electrophoresis and fragment length analysis showed that the corrected SCA3/MJD-iPSCs did not contain visible disease-causing ATXN3 mutations (ie, CAG74). Meanwhile, the corrected SCA3/MJD-C3 and SCA3/MJD-C12 maintained 17/31 CAG repeats in ATXN3 (Additional file 3: Figure S3).
Remarkably, ten potential OTs (Additional file 9: Table S2) for ATXN3-sgRNA/Cas9n were predicted by silico analysis using GT-Scan (http://gt-scan.braembl.org.au) . The potential OTs of each sgRNAs were PCR-amplified and analyzed with T7EN1 assays. Our results showed no detectable OTs examined in the ten sites (Additional file 4: Figure S4).
Corrected SCA3/MJD-iPSC remaining pluripotent characteristics
Previous study showed that parental SCA3/MJD-iPSCs retained disease-associated mutations and normal karyotype, expressing pluripotency markers, as well as have the potential to differentiate into three germ layers . The pluripotent characteristics also kept in the genetically control SCA3/MJD-iPSCs (C3 and C12) (Fig. 3 and Additional file 5: Figure S5). Specifically, normal karyotype (Fig. 3d) and pluripotency markers of NANOG, SOX2, SSEA4 were measured by immunoﬂuorescence staining (Fig. 3a) and flow cytometry (Fig. 3b). Endogenous expressing pluripotency markers of NANOG, SOX2 and OCT4 were evaluated by RT-qPCR (Fig. 3c). In vivo teratoma assay, the corrected clones showed the potential to differentiate into three germ layers, as shown by positive hematoxylin dyeing for glandular structure (endoderm), cartilage (mesoderm), and neural rosettes (ectoderm) (Fig. 3e).
Differentiation of SCA3/MJD and isogenic control iPSCs into forebrain cortical neurons and hindbrain Purkinje progenitor cells
To generate mature neurons from SCA3/MJD-iPSCs, Koch et al.  differentiated iPSCs into long-term self-renewing neuroepithelium stem cells. The differentiation system has cortical neural and glial mixed populations after long-term proliferation. Given that studying iPSCs-derived forebrain neurons may shed light on the pathogenesis of SCA3/MJD, we used the monolayer culture method of cortical neurons, which experienced the stage of neural rosettes and mature neural differentiation . Typical and mature neurons can be observed after about 60 days of neural induction (Fig. 4a). SCA3/MJD-iPSCs (74 CAG repeats), corrected SCA3/MJD-iPSCs (C3 and C12, containing 17 and 31 CAG repeats, respectively) and control-iPSCs (Ctr1, 29 CAG repeats) were efficiently differentiated into forebrain NSCs after 16 days of neural induction. The NSCs markers of PAX6, NESTIN, FOXG1, SOX1 and OTX1D were highly expressed on day 16 detected by RT-qPCR (Fig. 4b and Additional file 6: Figure S6). Moreover, PAX6 and NESTIN were positive staining by immunofluorescence (Figure 4c-f). At this stage, there was no significant difference in the NSCs markers expression among each group.
Using our protocols, 30~60 days after neuronal differentiation, all cells expressed mature NCs markers, including the majority of β-III-tubulin, MAP2 and GABA, as well as a small portion of glial fibrillary acidic protein (GFAP in astrocytes). The ratio of neurons to astrocytes (TUJ1/GFAP) was 2:1 (Fig. 5a and 5c). At this stage, there were no significant differences in TUJ1/GFAP and MAP2/GABA positive cells among the Ctrl-NCs, SCA3/MJD-NCs and corrected SCA3/MJD-NCs (C3, C12) groups (Fig. 5a-f).Therefore, our protocols of iPSCs differentiated into neurons and astrocytes was consistent with previous studies, in which the transformation of iPSCs led to a mixture of cultured neurons and astrocytes [41, 42]. Further analysis revealed that SYP1/PSD95, which is the pre- and post-synaptic marker of synaptic development, expressed similarly in four groups at days 40~60 differentiation as previously reported (Fig. 5g-h). Moreover, IC2 NIIs aggregates were only detected in SCA3/MJD-NCs compared with other groups (Fig. 5i).
Former studies have shown the susceptibility of hindbrain neurons in SCA3/MJD patients . To conduct the cerebellar neurons, we adopted a specific developmental model of cerebellar tissue, by differentiating iPSCs into cerebellar Purkinje progenitor cells based on previous protocols (Additional file 7: Figure S7a) [43-46]. We detected the up-regulation of midbrain/hindbrain patterning markers, such as the KIRRLE2, FGF8, WNT1, GBX2 and OTX2, by RT-qPCR on days 24 of differentiation (Additional file 7: Figure S7b). Immunofluorescence showed cerebellar precursor cells expressing KIRREL2/TUJ1 on days 24~32 (Additional file 7: Figure S7c). After 24~30 days of differentiation, the cerebellar progenitor cells in heterogeneous culture were sorted by KIRREL2+, and the selected KIRREL2+ Purkinje progenitor cells accounted for 19.2% by flow cytometry (Additional file 7: Figure S7d). However, the purified KIRREL2+ cerebellar precursor cells need to be co-cultured with purified cerebellar granule cells, which came from newborn mice, for 2~3 months until the mature Purkinje cells (PCs) generation. We could not obtain enough vigorous cell populations after fluorescence-activated cell sorting. In the future, we are planning to obtain more vigorous cell populations by optimizing the Purkinje progenitor cells purification scheme, or selecting THY1+ cell subpopulations for further exploring mature PCs differentiation strategies .
Electrophysiological functions of SCA3/MJD-iPSCs differentiated cortical neurons
Next, we explored electrophysiological properties of the differentiated neurons. Whole-cell patch-clamp recording techniques were used to measure the intrinsic electrophysiological excitability of these differentiated cells for 6~7 weeks. Evoked action potentials (APs) displaying the excitable properties under current clamp were recorded in SCA3/MJD-NCs (Fig. 6a). Robustly Na+, K+ and Ca2+ inward and outward currents were detected in Ctr1-iPSCs, SCA3/MJD-iPSCs, and SCA3/MJD-C3-iPSCs derived NCs (Fig. 6b). Besides, the inward Na+ and Ca2+ currents can be blocked by TTX (1μM) and Cdcl2 (0.1mM) ion antagonist, respectively (Fig. 6c). Under the action of the depolarized current pulse, there was no significant difference in the inward or outward current peaks among the three groups (Fig. 6b). In order to assess the effect of ATXN3 mutation on glutamatergic synaptic formation and excitatory synaptic transmission, both spontaneous glutamatergic sEPSCs and spontaneous GABAergic sIPSCs were recorded in whole-cell voltage-dependent recordings. Results showed that sEPSCs could be blocked by CNQX (AMPA receptor antagonist, 10μM), or MK801 (NMDA receptor antagonist, 10μM), respectively (Fig. 6d). Altogether, these data indicated that Glu mediated excitatory synaptic and GABA mediated inhibitory synaptic, input onto iPSC-derived cortical neurons, a study supported by previous studies . The frequency and amplitude of sEPSCs were very similar in each group (Fig. 6e-f). In addition, the membrane capacitances, input resistances and RMPs also showed no significant difference among three groups (Fig. 6g-i). These findings suggested that TUJ1/MAP2-positive neurons exhibit electrophysiological properties, and there was no obvious difference across the three different iPSCs derived neurons.
Reversal of mitochondrial dysfunction and oxidative stress activation in corrected SCA3/MJD- iPSCs derived NCs
Mitochondrial membrane potential is an effective method to evaluate mitochondrial function. In this study, mitochondrial membrane potential decreased significantly in SCA3/MJD-NCs compared with Ctr1-NCs, and corrected SCA3/MJD-NCs rescued the level of membrane potential decline (Fig. 7a-b). Studies showed that abnormal changes in intracellular ROS and Ca2+ levels were correlated with various neurodegenerative diseases [47-50]. In the present study, the ROS and intracellular Ca2+ levels were significantly higher in SCA3/MJD-NCs compared with Ctr1-NCs, while obviously decreased in the corrected SCA3/MJD-NCs (Fig. 7c-d). In addition, the expression of MDA increased, and GSH decreased in SCA3/MJD-NCs, but could be rescued in the corrected NCs (Fig. 7e-f). This findings indicated that the increased generation of oxygen free radicals, activated oxidative stress, and decreased antioxidant capacity in SCA3/MJD-NCs, were related to neuronal dysfunction [50, 51]. Moreover, ROS were activated in SCA3/MJD-NCs over time following H2O2 stimulation compared with Ctrl-NCs and isogenic control SCA3/MJD-NCs (Fig. 7g). Altogether, this study highlighted the likely contribution of mitochondrial dysfunction and oxidative stress activation to the pathogenesis of SCA3/MJD.