br Results Disease modeling using iPSCs can be

Results Disease modeling using iPSCs can be confounded by phenotypic variation resulting from genetic differences between individual iPSC lines (Merkle and Eggan, 2013). To overcome this obstacle, we sought to generate isogenic cell lines by targeted correction of the point mutation in the SOD1 iPSC using the CRISPR-Cas9 system (Hendriks et al., 2016). We designed a guide RNA to specifically target the mutant allele, taking advantage of the observation that the A > G mutation in the SOD1 locus creates a PAM recognition sequence not present in the reference allele (Figure 1A). In addition, the guide RNA was chosen such that Cas9 would create a double-stranded break within 5–6 bp of the targeted mutation, thereby increasing the efficiency of incorporating the reference allele provided via a donor DNA oligonucleotide (Figure 1A and Experimental Procedures). Correction of the heterozygous point mutation in the SOD1 gene was confirmed by PCR amplification of the targeted genomic regions followed by capillary sequencing (Figure 1B). In addition, the corrected iPSCs displayed a normal karyotype (Figure S1A). We differentiated the diseased and corrected iPSCs as well as an iPSC line derived from a healthy individual (designated as 80a) into spinal motor neurons by adapting a recently published protocol capable of generating MNs with high efficiency (Figure 1C; Maury et al., 2015). For all the iPSC lines used in our study, we were able to generate OLIG2+ motor neuron progenitors at day 10 and post-mitotic ISL1+/TUJ1+ MNs at day 14 (Figures S1B and S1C). At day 30 of differentiation, all iPSC lines generated ISL1+/TUJ1+ spinal MNs with efficiencies >70%, similar to that observed previously (Figure 1D). In addition, greater than 80% of the ISL1+ MNs at day 30 also expressed CHAT, a marker of mature MNs, and both ISL1+ MNs as well as ISL1− non-MNs expressed MAP2, a pan-neuronal marker expressed in mature neurons (Figures 1E and S1D; Sances et al., 2016). Further, our iPSC-derived MNs were electrophysiologically active and responded to optogenetic stimulation (Figure 1F). Lastly, when co-cultured with rat cortical neurons (Figure S1E), MNs displayed synchronized neural activity as measured by the genetically encoded Ca+ indicator GCaMP6 (Figure S1F). Importantly, the observed MN neural activity was synchronous with the cortical neuronal activity (Figure S1G), indicating that our MNs were capable of accepting synaptic input from the cortical neurons. Taken together, the marker SGX-523 and neural activity indicated that our iPSC-derived MNs were mature and functionally active. Since the over-arching phenotype observed in ALS patients is the specific loss of MNs, we asked whether ALS MNs would reveal disease-associated decline of survival in vitro. ALS is an adult-onset disease, therefore we monitored survival of MNs after they had attained maturity. Accordingly, we followed MN survival in low-density cultures from day 30 to day 44 (Figure 2A). ALS mutant MNs showed a significant decline in survival compared with the healthy control MNs (Figure 2B). On the other hand, genetic correction of the disease MNs significantly improved survival to a level comparable with wild-type MNs, indicating that correction of the mutation had ameliorated the disease phenotype (Figure 2B). Importantly, this decline in survival was not observed in ISL1−/TUJ1+ non-MN (Figure 2C), thereby recapitulating the MN-specific loss observed in ALS. In addition, we observed increased cleaved caspase activity in disease MN cultures (day 37) compared with healthy or isogenic control cultures, indicating that the loss of diseased MNs was, in part, due to apoptosis (Figure 2D). We observed that increased apoptosis in diseased MNs was also accompanied by morphological changes that were consistent with observations of postmortem spinal tissue from ALS patients (Kiernan and Hudson, 1991). Morphometric analysis of our in vitro ALS model revealed a reduction in the soma size, maximum neurite length, as well as average neurite tree length in mutant MNs at day 44 compared with the control MNs, while genetic correction of the mutation improved these morphological characteristics in the isogenic MNs (Figures 2E–2G). We also observed significantly higher levels of the tumor suppressor p53 (TP53) in the nuclei of mutant SOD1 MNs compared with both the healthy control and isogenic MNs (Figure 2H), which is concordant with activated p53 observed in ALS postmortem spinal tissue as well as rodent models of ALS (Qiu et al., 2014; Ranganathan and Bowser, 2010). Studies on the SOD1 G93A mouse model of ALS, as well as a recent iPSC model of SOD1 ALS, have revealed heightened endoplasmic reticulum (ER) stress in ALS MNs (Kiskinis et al., 2014; Nishitoh et al., 2008). In cells undergoing ER stress, IRE1 splices XBP1 to generate the active spliced form of the transcription factor (sXBP1), while PERK activates the transcription factor ATF4, which leads to upregulation of ATF3 and CHOP (Jiang et al., 2004; Xiang et al., 2016). To assay whether our in vitro model revealed an increase in ALS-related ER stress, we measured the ratio of spliced XBP1 to total XBP1 transcript levels (sXBP1:XBP1) via qRT-PCR. We found an increased sXBP1:XBP1 ratio in SOD1 MNs compared with the control MNs that normalized upon genetic correction (Figure 2I). In addition, we observed significantly higher levels of ATF3 and CHOP in SOD1 MN cultures compared with both the control and isogenic MNs (Figures S2A and S2B), indicating that the ALS iPSC-derived MNs displayed an increased ER stress response.