The molecular mechanisms that lead to the development of the SMA pathology are unclear. For this reason, despite substantial research in the area, an effective treatment for this disease does not yet exist. As such, there is a need to identify therapeutic strategies that delay the advance of SMA pathology. Following lineage restriction of hiPSCs to generate motor neurons, previously shown to be functional 
and electrophysiologically active 
, we identified molecular markers of apoptosis in SMA-iPSC MN cultures. In the present study, we demonstrate using two independent SMA and two control iPSC lines that there were significantly fewer MNs at 10 weeks of differentiation from SMA patient-iPSCs. Importantly, we show here that this phenotype could be rescued by blocking the Fas receptor or inhibiting caspase-3. As such, our data suggest that apoptosis plays an important role in disease progression, and therapies targeting this cascade may have important clinical applications.
The lentiviral SMA type 1 line (13iSMA) 
used in this study produces similar number of SMI-32+ MNs compared to two control lines at early time points in culture, while the new virus-free SMA type I line (77iSMA) initially produces far greater MNs than all the lines combined. This would suggest that in this SMA stem cell model, the patient iPSC lines are competent in generating MN progenitors at early stages of differentiation and argue against a serious fetal developmental maturation error as suggested in a recent study examining post-mortem pathological analyses from spinal cords of SMA patients 
. Between 3 and 10 weeks of differentiation, the SMA cultures show a decline in the total number of MNs in comparison to the control cultures (). This could either represent death of the motor neurons over time, or failure to increase in number due to deficits in motor neuron progenitor pools. However, we show here that greater numbers of MNs in the SMA cultures were double labeled with caspase-3 () when compared to control cultures, suggesting that at least a proportion of MNs were actively undergoing apoptosis.
We have observed that there is inter-line variation in the propensity of hiPSCs and hESCs towards terminal MN differentiation, which is likely due to the intrinsic characteristics of the lines. This observation is confirmed by a recently published study where 16 iPSC lines were tested by independent labs for MN production using standard protocols 
. Although all cell lines were capable of generating MNs, significant and reproducible quantitative differences were revealed in the propensity of the lines for terminal differentiation. This cautions against the direct comparison of cell numbers at any stage – pluripotent stem cells, progenitors or terminally differentiated derivatives – between control and patient iPSC lines for a reproducible disease phenotype. Therefore, in cases where a disease phenotype is expected in differentiated cells, we suggest that validation of a disease phenotype should be performed by a temporal, longitudinal study collecting data at multiple time points for an intra- and inter-line comparison. Using this methodology, each cell line serves as its own control with regard to starting and ending numbers of specific types of neurons or other cell types. Here we show that MNs generated from SMA patients undergo selective degeneration in a temporal manner, and that this is associated with the activation of the Fas-mediated apoptosis. Our findings of MN degeneration in SMA-iPSC lines are also confirmed by a very recent report which shows that five clonal iPSC lines from a single SMA patient (GM09677 patient fibroblasts were used in this report analogous to the source of the 77iSMA line used in our study), made using retroviral integrating vectors, exhibit a reduced capacity to form MNs and abnormalities in neurite outgrowth 
, all of which could be rescued by ectopic expression of SMN.
The discovery that the neuronal apoptosis inhibitor protein (NAIP) on chromosome 5 was mutated in greater than half of all SMA type 1 cases 
led to early suggestions that apoptosis may be directly involved with SMA. Interestingly, deletions in the NAIP gene in the same patients may increase the severity of disease and thus act as a modifier 
. Furthermore, while the major function of SMN is the biogenesis of spliceosomal snRNPs, it has also been shown to modulate apoptosis 
. Further support for involvement of apoptosis in SMA comes from a number of in vitro
and in vivo
. Our data showing significant increases in active caspase-3 within the differentiating SMA MN cultures lends further direct support to the idea of apoptosis in the human disease. This is in contrast to a recent report from Ito and colleagues in which no significant difference in caspase-3 activation was observed between SMA and control fetal post-mortem spinal cord tissue samples 
. This could be due to a variety of reasons including timing of analysis, tissue processing, methods of detection, and intrinsic differences between in vitro
and in vivo
studies. Nevertheless, apoptotic signals are activated in our in vitro
model and warrant further investigation in SMA pathophysiology.
Clearly apoptosis is a complex process where one of the major pathways involves binding of ligands, such as TNF or FasL, to their death receptors (TNFR or FasR, respectively), leading to the recruitment of adaptor proteins and the subsequent activation of caspase-8 
. Downstream activation of caspase-3 leads to cleavage of substrates vital to cell function. Here we show that cells within MN cultures of SMA patient-iPSCs showed significant chromatin condensation as well as activation of initiator caspase-8 and executioner caspase-3. While we cannot eliminate some involvement of mitochondrial-dependent pathways as previously found in a mouse model of SMA 
, the major component appears to be death receptor-mediated. Membrane-bound FasL is a potent activator of the death-receptor mediated apoptosis, and we showed increases in membrane-bound FasL in the SMA samples corresponding to the time of significant MN loss (). Blocking this pathway during differentiation with Fas neutralizing antibody rescues the MNs, further supporting a role for this pathway being active in this disease. SMN levels do not vary temporally in this culture system over the course of MN differentiation in SMA-iPSC lines (Fig. S3
), but the current data do not identify a mechanism linking diminished SMN protein to up-regulated Fas ligand. Therefore, further studies will address FasL processing, splicing, transport, and translation specifically in relation to reduced SMN.
During development, apoptosis plays a critical role in neuron pruning such that 50% of MNs initially generated in the spinal cord die 
. Target tissues play an important role in providing necessary trophic support to maintain MNs. However, embryonic MNs have been shown to activate apoptosis through binding of FasL and activation of caspase-3 independent of trophic factor support 
. Caspase-3 activation has been observed in healthy control and SMA patient fetal post-mortem spinal cord tissue 
. Similarly, we also observe caspase-3 activation in the MN cultures from the control and SMA lines at early time points (), which is presumably due to active apoptotic signals upon the transition from a pluripotent stem cell state to a differentiated state, reflecting non-MN and MN death. If MNs are naturally primed to undergo apoptosis during development, perhaps the lack of SMN in SMA-iPSCs maintains a hyperactive apoptotic process via prolonged caspase-3 activation, and may also explain the disease phenotype observed in patients and experimental models. Alternatively, the MNs in our culture system may be lacking the appropriate target tissue or trophic support. Notably, blocking caspase-3 activation with a commercial inhibitor rescues the MN degeneration phenotype in SMA, signifying the importance of this pathway in our disease model.
The iPSC-derived cultures in this study contain a mixed population of cell types, including ChAT positive motor neurons, Tuj1 positive neurons, and GFAP positive astrocytes (Fig. S5
). At this time, technical complexities, such as purification and survival of mature MNs from iPSCs by sorting or panning, make it difficult to distinguish whether a cell autonomous or a non-cell autonomous apoptotic process is involved. The non-neuronal cells play a determinant function in the vulnerability of the neurons in several pathological conditions 
. Astrocytes, for example, have been shown to dramatically alter the health and survival of MNs in transgenic models of ALS 
and FasL is known to promote astrocyte reactivity and production of proinflammatory cytokines 
. Since SMN1 is absent in all cell types, the possibility remains that astrocytes may themselves be dysfunctional and contributing to the apoptotic process. Future studies targeting the HB9 motor neuron-specific promoter in iPSCs in conjunction with co-culture assays will address this issue. The protective effect of FasL antagonistic antibody may occur though an indirect mechanism implicating surrounding cells in our MN cultures.
During the past several years, our understanding of the mechanisms mediating cell death in neurologic diseases has improved considerably. The fact that activation of these pathways is a feature of a broad range of neurologic diseases makes them important and attractive therapeutic targets. Therefore, conferring neuroprotection by interrupting apoptosis and preserving mitochondrial integrity are being actively explored by companies and academics. For example, monoamine oxidase inhibitors, selegiline and rasagiline, in Alzheimer’s and Parkinson’s disease 
, and dexpramipexole, a lower affinity non-ergot dopaminergic autoreceptor agonist, for the potential treatment of amyotrophic lateral sclerosis 
are being investigated. Additionally, combination therapies targeting apoptosis with other pathways are being tested for Parkinson’s and Alzheimer’s disease including antioxidants, cell cycle inhibitors, JNK inhibitors, GSK3β inhibitors, and STATINS 
, which could also be suitable strategy for delaying the progression of in SMA. Despite the lack of information from clinical trials, natural product compounds such as resveratrol and melatonin are attracting considerable attention because of their antioxidant and anti-apoptotic action in addition to low toxicity in humans 
. Resveratrol has also been shown to increase full length SMN transcript and protein in SMA fibroblasts 
. The data presented here further suggest that apoptotic pathway inhibition may be a therapeutically relevant target for SMA. Furthermore, by scaling up the current system for high content screening studies 
, it may be possible to test these novel compounds for efficacy in this novel human model prior to administration to patients, and thus increase the possible chances of success.