Modeling ALS and FTD with iPSC-derived Neurons

Abstract

Recent advances in genetics and neuropathology support the idea that amyotrophic lateral sclerosis (ALS) and frontotemporal lobar dementia (FTD) are two ends of a disease spectrum. Although several animal models have been developed to investigate the pathogenesis and disease progression in ALS and FTD, there are significant limitations that hamper our ability to connect these models with the neurodegenerative processes in human diseases. With the technical breakthrough in reprogramming biology, it is now possible to generate patient-specific induced pluripotent stem cells (iPSCs) and disease-relevant neuron subtypes. This review provides a comprehensive summary of studies that use iPSC-derived neurons to model ALS and FTD. We discuss the unique capabilities of iPSC-derived neurons that capture some key features of ALS and FTD, and underscore their potential roles in drug discovery. There are, however, several critical caveats that require improvements before iPSC-derived neurons can become highly effective disease models.

1. The Expanding Landscape of Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Dementia (FTD)

1.1 ALS and FTD: Two Ends of A Disease Spectrum

Amyotrophic lateral sclerosis (ALS), also known as Lou Gehrig’s disease, is an adult-onset neurodegenerative disease that affects upper and lower motor neurons. As initially described by Jean-Martin Charcot more than 140 years ago, the key clinical features in ALS patients include muscle wasting, and progressive loss of spinal motor neurons and upper motor neurons and their axons in the lateral columns of the spinal cord. This century-old classic definition, however, has been under scrutiny due to the emerging appreciations that some ALS patients who exhibit deficits in higher cognitive functions at the early stage of their clinical course eventually develop behavioral variant frontotemporal dementia (bvFTD). Indeed, it has been estimated that ~15% of FTD patients develop features of ALS (Lomen-Hoerth et al., 2002; Ringholz et al., 2005), and up to 50% of ALS patients show abnormal neuropsychological testing indicative of frontal lobe dysfunctions (Lillo et al., 2011; Lomen-Hoerth et al., 2003). Furthermore, patients with FTD-ALS usually have shortened life span compared to patients with pure FTD or ALS (Olney et al., 2005). Together, these studies provide a new framework to re-evaluate the diagnostic criteria for ALS (Strong et al., 2009), and a new clinical paradigm in which ALS and FTD are linked within a disease spectrum (Figure 1).

Figure 1. A Schematic Diagram Illustrating the Disease Spectrum, Genetics and Proteinopathy of Amyotrophic Lateral Sclerosis (ALS) and Frontotemporal Dementia (FTD).

Clinical manifestations indicate that ALS and FTD are two ends of a disease spectrum. It is estimated that ~15% of FTD patients develop ALS and up to 50% of ALS patients show impairments in frontal executive functions. This new clinical paradigm is further supported by the findings that patients with FTD-ALS and ALS share similar genetic mutations, in particular mutations that involve the GGGGCC hexanucleotide expansion in the non-coding region of the C9ORF72 gene and two RNA/DNA binding protein TDP-43 and FUS. FTD-ALS and ALS patients with mutations in the same gene (C9ORF72, TARDBP or FUS) often exhibit similar neuropathological features with neurons containing distinct protein aggregates of TDP-43 or FUS proteins. Therefore, the neuropathology of these patients can be categorized by the nature of the proteinopathy into FTLD (frontotemporal lobar degeneration)-tau (FTLD-tau), FTLD-TDP, FTLD-FUS, ALS-TDP, ALS-FUS or ALS-SOD1.

1.2. Unraveling the Genetics and Neuropathology of ALS and FTD

The idea that FTD and ALS are a disease spectrum is further supported by genetic evidence that patients with familial FTD-ALS and ALS often carry mutations in the same genes (Ling et al., 2013). In fact, of all the different subtypes of FTD, FTD-ALS cases have such a high propensity (~30%) of familial inheritance that genetic counseling is now considered as the standard of care for these patients (Goldman et al., 2005). Among a growing number of genes implicated in familial ALS and FTD-ALS (Figure 1), mutations in three genes account for the majority of cases. These mutations include missense mutations in genes encoding two RNA/DNA binding proteins, TDP-43 (TARDBP or TAR-DNA-binding protein-43) and FUS/TLS (fused in sarcoma/translocation in liposarcoma or FUS) and the GGGGCC hexanucleotide expansions in C9ORF72 gene (Lee et al., 2012; Ling et al., 2013). The discovery of TDP-43 as a component in the ubiquitin-positive, tau-negative insoluble protein aggregates in neurons and glia represents a major breakthrough in FTD research (Arai et al., 2006; Neumann et al., 2006). Moreover, the impact of this discovery goes beyond the identification of a single disease gene and essentially ushers in a new era of research that focuses on the potential contributions of transcription, RNA splicing and RNA metabolism on neurodegenerative diseases.

TDP-43 is originally identified to bind to the TAR DNA sequence in HIV-1 genome to regulate viral gene expression (Ou et al., 1995). Under physiological conditions, TDP-43 is a ubiquitous nuclear protein. In FTD patients, however, TDP-43 proteins form aggregates that are found predominantly in neuronal cytoplasm and dystrophic neuronal processes (Arai et al., 2006; Neumann et al., 2006). This distinctive feature, also known as TDP-43 proteinopathy, defines a major neuropathological diagnosis entity in frontotemporal lobar degeneration (FTLD-TDP) and in sporadic ALS (ALS-TDP)(Figure 1)(Mackenzie et al., 2011; Mackenzie et al., 2010). Several subsequent studies show that dominant mutations in the TARDBP gene can also be identified in familial ALS and FTD patients (Lattante et al., 2013). The identification of autosomal dominant mutations in the FUS gene in large kindred of familial ALS (FALS) further expanded the genetic and neuropathological landscape of ALS (Kwiatkowski et al., 2009; Vance et al., 2009). Similar to TDP-43, FUS proteins reside primarily in the neuronal nuclei, but in ALS-FUS patients FUS proteins form large aggregates in the cytoplasm. The morphology of FUS proteinopathy in FALS ranges from diffuse and dense cytoplasmic aggregate present in late onset cases, to basophilic inclusions commonly found in juvenile FALS with FUS-P525L mutation. Finally, in 2011 two groups independently reported the GGGGCC hexanucleotide repeat expansions in the noncoding region of the C9ORF72 gene as causal links to ALS and FTD (DeJesus-Hernandez et al., 2011; Renton et al., 2011). Although TDP-43 proteinopathy can be detected in FTLD and ALS patients with C9ORF72 mutations, the neuropathological features in these cases are quite heterogeneous and also include prominent ubiquitin and p62 positive, but TDP-43 negative intracytoplasmic and intranuclear inclusions (Bigio, 2012; Mackenzie et al., 2014). Interestingly, TDP-43 proteinopathy has also been identified in protein aggregate myopathies, such as inclusion body myositis (Olive et al., 2009). Together, these results raise the possibility that TDP-43 proteinopathy may not be specific for FTLD or ALS.

It is estimated that mutations in TARDBP and FUS each account for ~5% of FALS, whereas the GGGGCC expansion mutations in C9ORF72 account for 20-40% of familial ALS and FTD-ALS cases, depending on the population studied. Given the large number of cases with C9ORF72 mutations, there have been tremendous interests in understanding the underlying mechanisms. While several mechanisms, including RNA toxicity and dipeptide accumulation (Ash et al., 2013; Donnelly et al., 2013; Mori et al., 2013), have been proposed for C9ORF72 mutations, the rapidly progressing research on this subject will definitively bring many more surprises in the future. One important feature noted in a recent study indicates that the age of disease onset for FALS caused by FUS, TARDBP and C9ORF72 mutations differ quite drastically in that mutations in FUS account for ~35% of FALS in patients younger than 40 years old, whereas mutations in C9ORF72 are much more common in patients older than 50 years of age (Millecamps et al., 2012). Indeed, meta-analyses of 154 ALS cases with FUS mutations (including FALS and SALS with de novo FUS mutations) show an average disease onset of 43.8 ± 17.4 years (Deng et al., 2014; Lattante et al., 2013). More than 60% of cases with FUS mutations show disease onset before 45 years of age, with many juvenile ALS cases presenting with disease onset in late teens and early 20’s (Baumer et al., 2010; Huang et al., 2010). These findings are similar to those from another study using smaller sample size, and show that the average disease onset for FUS, SOD1 or TARDBP mutations is 43.6 ± 15.8, 47.7 ± 13.0 and 54.7 ± 15.3, respectively (Yan et al., 2010). Kaplan-Meier survival analysis shows statistically significant differences in the trend of age of onset among these three genes. This distinctive feature of FUS mutations raises the intriguing hypothesis that, despite the similarities between TDP-43 and FUS proteins, mutations in FUS may target divergent mechanisms that perturb the development and maintenance of synaptic homeostasis in the nervous system in early postnatal life and during the aging process (Qiu et al., 2014; Sephton et al., 2014).

2. Induced Pluripotent Stem Cells (iPSCs) as Models for ALS

Recent advances in stem cell biology have provided exciting and unprecedented opportunities to develop disease-specific cell types that allow us to understand and explore mechanisms that contribute to pathogenesis of disease. In particular, the ability to reprogram somatic cells into induced pluripotent stem cells (iPSCs) provide obvious advantages to generate patient-specific iPSCs that carry the exact same genetic makeup, including mutations that may contribute to the disease process (Figure 2)(Takahashi et al., 2007; Takahashi and Yamanaka, 2006). These iPSCs can be differentiated into cell types that provide strong correlations to clinical benchmarks and pathology (Han et al., 2011). In addition, patient-specific iPSCs may also serve as powerful resources for personalized medicine, including drug discovery, genetic testing, and ultimately cell replacement therapy (Figure 2). Despite these unique advantages, one important caveat with the iPSC models is that a myriad of epigenetic changes due to reprogramming and the lengthy process of cell selection and in vitro cultures might contribute to the variations in this system. Here we focus on the recent progress on using induced pluripotent stem cell (iPSC)-derived neurons as the new models to investigate disease mechanisms for ALS and FTD. The salient features of these studies are summarized in Table 1.

Figure 2.

A schematic diagram illustrating the generation of disease-specific induced pluripotent stem cells (iPSCs) and its applications in modeling disease phenotypes and in identifying therapeutic targets.

Table 1.

Summary of the results from iPSC-derived neurons as disease models for ALS and FTD.

Genetic mutation	Neuron types	Analyses	Phenotypes & mechanisms	Gene targeted corrections, drug screening or other intervention s	Reference s
SOD1D90A	Motor neurons	Cell biology & immune-gold EM	• Neurofilament  aggregates &  inclusions  • Mutant SOD1  binds to 3’UTR  of NF-L mRNA  & destabilizes  NF-L mRNA	TALEN- mediated homologous recombinatio n	Chen et al., 2014
SOD1A4V C9ORF72	Motor neurons	RNA-seq gene expression profiling	• Increased  apoptosis,  reduced soma  size and  shorter &  fewer  processes • Defects in  mitochondrial  morphology &  motility • Unfolded  protein  response  (UPR) & ER  stress	Zinc finger nuclease (ZFN) correction of SOD1A4V mutation improves survival and soma size	Kiskinis et al., 2014
SOD1G93A, SOD1L144F	Motor neurons	MN survival assay, Electrophysiolog y	• Improves  motor neuron  survival via  cell-  autonomous  mechanisms  by Inhibition of  GSK3α/β&Tak  1-MKK4-JNK-  c-Jun	Small molecule screening	Yang et al., 2013
SOD1A4V SOD1D90A SOD1G85S C9ORF72 FUSM511FS FUSH517Q	Motor neurons	Extracellular multielectrode array recording (MEAs) & whole-cell patch clamp	• Hyperexcitabili  ty that can be  blocked by  Retigabe	n.d.	Wainger et al., 2014
TDP- 43M337V	Motor neurons & astrocytes	Selective cellular vulnerability by longitudinal fluorescent microscopy	•TDP-43M337V  neurons show  selective  cellular vulnerability to  PI3K inhibitors • TDP-43M337V  neurons show  increase in  detergent-  insoluble TDP-  43 proteins • TDP-43M337V  astrocytes  show increase  in TDP-43, but  no non-cell  autonomous  cytotoxicity to  iPSC neurons	TDP-43M337V allele- specific siRNA knock-down reduces cytosolic TDP-43	Bilican et al., 2012 Serio et al., 2013 Nishimura et al., 2014
TDP- 43Q343R TDP- 43M337V TDP-G298S 43	Motor neurons	Cell biology, gene expression profiling & drug screening	• Reduced NFM  & NFL  expression •  Increase in detergent-  insoluble TDP-  43 in ALS  neurons • Increased  TNFα/NFκB  signaling  pathway &  sensitivity to  arsenite-  induced cell  death • Gene  expression  profiling  reveals  increases in  RNA  metabolism-  related genes  in ALS  neurons	Candidate drug screenings identify HAT inhibitor Anacardic acid to decrease TDP-43 mRNA, reduce insoluble TDP-43 and increase NFM/NFL expression	Egawa et al., 2012
TDP- 43A90V TDP- 43M337V	Tuj1+ neurons	Electrophysiolog y, cellular response to stress and microRNA expression	• Cytoplasmic  localization of  TDP-43 in  TDP-43A90V  neurons • Reduced total  TDP-43  protein level in  TDP-43A90V  neurons • Decrease in  miR-9 and  precursor	n.d.	Zhang et al., 2013
TDP- 43M337V TDP- 43G298S TDP-A315T 43	Motor neurons	Axonal transport of RNA beacons, live imaging of RNA transport	• Axonal  transport of  target mRNAs	n.d.	Alami et al., 2014
C9ORF72	Mixed neuron- glia cultures	FISH for RNA foci, RAN translation pathology, identification of RNA binding proteins for GGGGCC RNA and ASO treatment	• RNA foci and RAN  translation  products detected • Screening for  RNA binding  proteins that  bind GGGGCC  and  characterize  ADARB2 • Gene  expression  profiling • Increase in  glutamate  cytotoxicity	ASOs targeting C9ORF72 rescues glutamate cytotoxicity & reverses disease- specific transcription al changes	Donnelly et al., 2013
C9ORF72	Motor neurons	Southern blot analyses for GGGGCC expansion, FISH for RNA foci, gene expression profiling and ASO treatment	• RNA foci  detected, but  not RAN  translation  products • Support gain-  of-function  properties in  RNA foci to  sequester  RNA binding  proteins and  affect splicing  and  transcription • Gene  expression  profiling  reveals  enrichment in  genes related  to cell  adhesion,  synaptic  transmission  and neural  differentiation • C9-ALS  neurons show  reduced  excitability  upon  depolarization	ASOs targeting C9ORF72 reverses disease- specific transcription al changes	Sareen et al., 2013
C9ORF72	Telencephalon neurons (FOXG1- expressing)(80 % MAP2+, 30% VGlut1+, 10% GABA+)	Southern blot analyses for GGGGCC expansion, RNA foci and stress	• RNA foci and  RAN translation  products  detected • Reduced cell  viability in the  presence of  autophagy  inhibitors	n.d.	Almeida et al., 2013
C9ORF72 TDP-43M337V	Motor neurons	Whole-cell patch clamping recordings	• ALS motor  neurons  display initial  hyperexcitabilit  y, followed by  a progressive  loss in action  potential and  synaptic  activity	n.d.	Devlin et al., 2015
Sporadic ALS	Motor neurons	Cell biology, drug screening & comparison with postmortem neuropathology	• Validation of  TDP-43  protein  aggregates in  iPSC-derived  motor neurons  from SALS  cases.	High content screening for TDP-43 aggregation inhibitors	Burkhardt et al., 2013
MAPT A152T	MAP2+ neurons	Isogenic iPSC using zinc-finger nuclease- mediated gene editing	• Tau protein  fragmentation,  Tau protein  phosphorylatio  n, Degeneration  of neuronal  process	Zinc finger nuclease mediated gene editing to correct the mutation	Fong et al., 2013
GrnS116X & Sporadic FTD	Neurons (80% MAP2+) & microglia	Electrophysiolog y, PGRN level, sensitivity to cellular stress, and gene expression profiling	• Grn S116X  neurons show  PGRN  haploinsufficie  ncy • GrnS116X  neurons show  increased  sensitivity to  ER stress,  staurosporine  & kinase  inhibitors, and  down-  regulation of  S6K2 • Pharmacologic  al suppression  of sortilin 1  (SORT1)  increases  extracellular  PGRN in  GrnS116X  neurons	Lentivirus- mediated expression of PGRN rescues cellular sensitivity to stress	Almeida et al., 2012 Lee et al., 2014
GrnIVS1+5G> C	Cortical neurons (CTIP2, FOXP2, TBR1+)	Genome-wide transcriptomes analyses	• Aberrantly  activated Wnt  signaling  pathway	Genetic correction using homologous recombinatio n with zinc finger nucleases (ZFNs)	Raitano et al., 2014

Notes: n.d., not done.

2.1. Modeling SOD1 Mutations

Autosomal dominant mutations in the superoxide dismutase 1 (SOD1) are the first to be causally linked to familial ALS. Following this discovery, several transgenic rodent models overexpressing SOD1 mutant proteins become widely used to reveal a plethora of disease mechanisms, including axonal transport defects, oxidative stress, protein misfolding, mitochondrial dysfunction and excitotoxicity. In addition, non-cell autonomous pathogenic mechanisms involving astrocytes, microglia and NG2 cells, have also been identified to contribute to neurodegeneration caused by mutant SOD1 overexpression. The availability of iPSC-derived motor neurons from patients with SOD1 mutations provide unique opportunities to capture salient neurodegenerative features of ALS-SOD1 at more realistic expression levels and to re-evaluate disease mechanism(s) that may lead to novel therapeutic insights. By reprogramming skin fibroblasts from two patients with SOD1^A4V mutation (SOD1^A4V/+), Kiskinis and colleagues showed that iPSC-derived ISL+/HB9+ SOD1^A4V/+ motor neurons exhibit a higher propensity to undergo apoptosis, similar to those observed in the SOD1^G93A mouse model (Kiskinis et al., 2014). In addition, iPSC-derived SOD1^A4V/+ motor neurons show significant reductions in soma size, and fewer and shorter neuronal processes. These pathological features appear to be specific to motor neurons because ISL− SOD1^A4V/+ neurons in the same culture conditions do not show similar defects. To further demonstrate the specificity of these phenotypes, the authors used zinc finger nuclease (ZFN)-mediated gene targeting to correct the SOD1^A4V/+ mutation and showed that motor neurons with corrected mutation (SOD1^+/+) show significant rescues in both survival and soma size.

Genome-wide transcriptome analyses in FACS-purified SOD1^A4V/+ motor neurons using RNA-seq further reveal altered expression in genes related to cytoskeleton organization. Consistent with these findings, another study shows that iPSC-derived motor neurons carrying SOD1^D90A mutation progressively develop neurofilament (NF) inclusions such that by Day 31 in cultures >60% of SOD1^D90A motor neuron cell body and 25% of neurites contain NF inclusions (Chen et al., 2014). The development of NF inclusions in SOD1^D90A motor neurons correlates with degeneration of neurites. The underlying mechanism for the formation of NF inclusion in SOD1^D90A motor neurons appears to be caused by mutant SOD1 proteins that bind to the 3’UTR of NF-L mRNA, destabilize NF-L mRNA and thereby change the proportion of NF subunit composition. In addition to the defects in cytoskeleton organization, gene expression profiling in SOD1^A4V/+ motor neurons also reveals defects in mitochondrial function and structure. In support of these findings, ultrastructural analyses show distortion of mitochondrial morphology in SOD1^A4V/+ motor neurons. In addition, live cell imaging shows that mitochondrial movement is reduced in the axons of SOD1^A4V/+ neurons (Chen et al., 2014).

Another major group of genes misregulated in SOD1^A4V/+ motor neurons are those related to ER stress and unfolded protein response (UPR). While the exact cause(s) leading to the activation of ER stress and the UPR pathways remain unclear, at least two mechanisms may contribute to these processes. First, the progressive accumulation of misfolded mutant SOD1 proteins may trigger ER stress and UPR in SOD1^A4V/+ motor neurons. In support of this idea, SOD1^A4V/+ motor neurons show a significant increase in the insoluble mutant SOD1 proteins when the neurons are treated with proteasome inhibitor MG-132. Second, additional evidence indicates that spinal cord motor neurons appear to have inherently higher level of ER stress, which may be interconnected with the unique electrophysiological properties endowed in motor neurons. Interestingly, compared to control motor neurons, SOD1^A4V/+ motor neurons have reduced delayed-rectifier potassium current amplitudes, which may contribute to their hyperexcitability properties (Wainger et al., 2014). Remarkably, the Kv7 channel activator retigabine blocks the hyperexcitability in SOD1^A4V/+ motor neurons and improves their survival in vitro. These results support the idea that the hyperexcitable properties of SOD1^A4V/+ motor neurons may directly contribute to neurodegeneration in ALS. Intriguingly, similar up-regulation of ER stress-related genes and hyperexcitability can also be detected in iPSC-derived motor neurons that carry C9ORF72 or FUS mutations, suggesting that protein misfolding and the ensuring consequences of ER stress and UPR may have broader pathogenic roles in different subtypes of ALS (Wainger et al., 2014). Finally, one recent study tests the utility of iPSC-derived motor neurons with SOD1 mutations in drug screening (Yang et al., 2013). The results from this study show that blocking HGK (hepatocyte progenitor kinase like or MAP4K4) with inhibitor kenpaullone can promote the survival of SOD1^L144F human iPSC-derived motor neurons as well as mouse ESC-derived SOD1^G93A motor neurons. These encouraging results support the idea that iPSC-derived motor neurons can be used as effective tools to identify novel therapeutic targets for ALS.

2.2. Modeling C9ORF72 Mutations

At least three mechanisms, including the formation of RNA foci, repeat-associated non-ATG-initiated (RAN) translation products, and loss-of-function in C9ORF72 gene product, have been postulated to contribute to the neurodegeneration in C9ORF72 hexanucleotide repeat mutations (Gendron et al., 2014). Several groups have generated iPSC-derived neurons and developed RNA fluorescent in situ hybridization (FISH) to show that ~20-35% of iPSC-derived neurons contain intranuclear GGGGCC RNA foci (Almeida et al., 2013; Donnelly et al., 2013; Sareen et al., 2013). The presence of RNA foci in C9ORF72 iPSC-derived neurons recapitulates a key neuropathological feature of C9ORF72 mutations, and further allows characterization of how the intranuclear RNA foci might affect RNA metabolism. For instance, several RNA binding proteins, including ADARB2, hnRNPA1, hnRNPA1B2, Pur-a, FUS, and TDP-43, have been shown to potentially interact with GGGGCC RNA repeats (Donnelly et al., 2013; Sareen et al., 2013). However, so far only ADARB2 co-localizes with intranuclear RNA foci, suggesting that such association might have functional consequences (Donnelly et al., 2013). It is important to note that the various C9ORF72 iPSC models have implicated different sets of RNA binding proteins in the development of RNA foci. This variability may reflect the inherent clonal variations in the iPSC models, differences in the GGGGCC repeat lengths or other unknown factors (Table 1). Finally, despite the attention to RNA foci, it remains unclear whether these foci are toxic to neurons or simply reflect the disease state.

In addition to the intranuclear RNA foci, a higher percentage of C9ORF72 iPSC-derived neurons contain cytoplasmic RNA foci, suggesting that RAN translation may lead to accumulation of high molecular weight cytoplasmic dipeptide protein products, including poly-(Gly-Ala), poly-(Gly-Pro) and poly-(Gly-Arg)(Ash et al., 2013; Mori et al., 2013). In support of this idea, immunohistochemical staining confirms the presence of poly-(Gly-Pro) dipeptide protein products in C9ORF72 iPSC-derived neurons (Almeida et al., 2013; Donnelly et al., 2013). However, despite the presence of RNA foci and RAN translation protein products in iPSC-derived neurons, how these phenotypes directly contribute to human disease remain unclear. For instance, the variations in the quantity of RNA foci per cell in different lines of iPSC-derived neurons suggest that the formation of RNA foci may not correlate directly with the number of GGGGCC repeat, but perhaps may be influenced by other genetic or epigenetic factors (Almeida et al., 2013). Similarly, the levels of RAN translation protein products appear to be independent of the GGGGCC repeat length. In at least one study, the RAN translation protein products and aberrant increase of proteasome marker p62 are not consistently detected (Sareen et al., 2013). These results have also been further extended by another recent study (Gomez-Deza et al., 2015), which shows a distinct lack of RAN translation protein products in spinal motor neurons in FALS patients with C9ORF72 mutations. In the future, it will be important to use the iPSC models to determine how RAN proteins induce p62 pathology and whether p62 pathology is causally linked to neuronal degeneration and functional deficits.

The C9ORF72 iPSC-derived neurons provide a convenient and effective tool to further interrogate the pathogenesis of ALS. There are several directions to achieve this goal. First, transcriptome analyses in iPSC-derived motor neurons reveal misregulation of genes that are involved in cell adhesion, synaptic transmission and neural differentiation (Donnelly et al., 2013; Sareen et al., 2013). Interestingly, however, direct comparison of the transcriptomes of C9ORF72 iPSC-derived neurons and motor cortex of C9ORF72 ALS patients shows very limited overlapping. This difference may be due to the contribution of the glial cells in the motor cortex. Alternatively, it is possible that the difference may reflect that the iPSC-derived neurons represent a much earlier stage in the disease manifestations. Second, C9ORF72 iPSC-derived neurons exhibit elevated sensitivity to stress-mediated cytotoxicity. For instance, C9ORF72 iPSC-derived neurons (mixed population with 30-40% HB9+ neurons) exhibit more vulnerability to glutamate-mediated cytotoxicity in a dose-dependent manner (Donnelly et al., 2013). Similarly, another study indicated that C9ORF72 iPSC-derived neurons show reduced viability when exposed to chloroquine, an inhibitor that blocks the autophagy pathway (Almeida et al., 2013). Finally, C9ORF72 iPSC-derived neurons should also provide a convenient tool to investigate the potential loss-of-function effect of C9ORF72 gene product in the regulation of endosomal trafficking and in the pathogenesis of TDP-43 proteinopathy, which are common neuropathological features in these cases (Farg et al., 2014; Gomez-Deza et al., 2015).

In addition to revealing disease mechanisms, C9ORF72 iPSC-derived neurons provide an efficient tool to identify potential therapeutics. Based on the success of using antisense oligonucleotides (ASOs) approach to reverse RNA toxicity due to myotonic dystrophy (MD1)(Mulders et al., 2009), two studies showed that ASOs that targets the GGGGCC repeat containing transcripts can reduce RNA foci, normalize disease-specific transcriptional changes and rescue the increased vulnerability to glutamate toxicity, without affecting the viability of iPSC-derived neurons (Donnelly et al., 2013; Sareen et al., 2013). These results confirm that RNA toxicity, rather than loss-of-function in C9ORF72, is likely to be the major contributing factor to the pathogenesis of C9 ALS and provide a promising blueprint for future development of therapeutics.

2.3. Modeling TDP-43 Mutations

The identification of TDP-43 proteinopathy as a key neuropathological feature in sporadic ALS and FTLD represents a major advance in neurodegenerative disease research (Arai et al., 2006; Neumann et al., 2006). Subsequent studies show that autosomal dominant mutations in the TDP-43 gene TARDBP can indeed be identified in both sporadic and familial ALS patients (Lagier-Tourenne and Cleveland, 2010), though the incidence of these mutations is quite rare. To further understand how mutations in TARDBP might alter the subcellular localization of TDP-43 proteins, several groups have generated iPSC-derived motor neurons that carry TDP-43 mutations, including TDP-43^M337V, TDP-43^Q343R, TDP-43^G298S, TDP-43^A90V or TDP-43^A315T. These iPSC lines provide unique resources to determine whether these mutations directly contribute to the pathogenesis of disease, or merely represent risk factors. For instance, in control motor neurons wild type TDP-43 proteins are present predominantly in the soluble fraction. In contrast, TDP-43^M337V and TDP-43^Q343R motor neurons show increases of mutant TDP-43 proteins in the detergent-resistant insoluble fractions (Bilican et al., 2012; Egawa et al., 2012). These results suggest that the increases in mutant TDP-43 proteins are probably due to post-translational modifications rather than transcriptional mechanisms. Although the exact cause(s) that promote the accumulation of TDP-43 in the insoluble fractions remain unclear, there is evidence that mutant TDP-43 can form more stable complex with splicesome factor SNRPB2 (Egawa et al., 2012), suggesting that mutant TDP-43 may disrupt the splicing machinery through gain-of-function properties. Interestingly, automated live microscopy shows that TDP-43^M337V motor neurons have increased relative risks to undergo cell death. Together, these results are in agreement with previously reported cellular and transgenic TDP-43 models of elevated levels of cytoplasmic, but not nuclear, TDP-43 correlates with cellular toxicity (Barmada et al., 2010). The accumulation of cytoplasmic TDP-43 proteins in TDP-43^M337V motor neurons shows a significant reduction when these neurons are treated with allele-specific siRNA to knock-down the mutant TDP-43^M337V allele (Nishimura et al., 2014). These results not only provide definitive evidence that the accumulation of mutant TDP-43 proteins in cytoplasm is specific, they also support the idea that RNA interference has the potential to become an effective therapeutic tool for ALS.

Several studies further take advantage of the iPSC-derived motor neurons to interrogate the roles of mutant TDP-43 in cell autonomous and non-cell autonomous mechanisms that cause motor neuron dysfunctions. For instance, TDP-43^A90V and TDP-43^M337V iPSC-derived neurons exhibit increased vulnerability to stress-induced cytotoxicity and reduced microRNA-9 expression (Zhang et al., 2013). The findings in TDP-43^A90V iPSCs are quite intriguing because TDP-43^A90V has been considered as a risk factor for ALS. Hence, results from this study suggest that other additional factors, such as genetic background or environmental factors, may have contributed to the initiation and progression of disease. With respect to TDP-43^M337V mutation, gene expression profiling in the purified TDP-43^M337V iPSC-derived motor neurons further reveals misregulation of genes that are implicated in RNA metabolism and cytoskeleton functions (Egawa et al., 2012). In particular, the expression of neurofilament-medium (NEFM) and neurofilament-light (NEFL) chain is significantly reduced in TDP-43^M337V iPSC-derived motor neurons. These results are similar to those reported in SOD1^D90V motor neurons (Chen et al., 2014), suggesting that different genetic mutations may share the same target genes leading to ALS phenotype. In another study that combines three model systems, namely Drosophila motor neurons, mouse cortical neurons and human iPSC-derived motor neurons, Alami and colleagues provide compelling evidence that wild type TDP-43 proteins form cytoplasmic mRNP granules and facilitate the bidirectional, microtubule-dependent transport of these mRNP granules in axons. In contrast, mutant TDP-43 proteins, either TDP-43^M337V or TDP-43^A315T, consistently cause impairments in the transport of NEFL (neurofilament-L chain) mRNP granules in axons (Alami et al., 2014). Given the pluripotent nature of iPSCs, these cells provide rather convenient tools to determine whether glial cells expressing mutant TDP-43 proteins might cause cytotoxicity to motor neurons. Indeed, neural precursors from TDP-43^M337V iPSCs are treated with epidermal growth factor (EGF) and leukemia inhibitory factor (LIF) for 4-6 weeks, followed by 14 days treatment of ciliary neurotrophic factor (CNTF) to promote differentiation to astrocytes. Using motor neuron-astrocyte co-cultures, Serio and colleagues show that TDP-43^M337V iPSC-derived astrocytes do not exhibit non-cell autonomous cytotoxic effects to motor neurons (Serio et al., 2013).

Finally, TDP-43^M337V iPSC-derived motor neurons also provide a unique tool to identify potential therapeutic targets that may mitigate the ALS phenotypes. To this end, Egawa and colleagues show that when exposed to arsenite to induce oxidative stress, TDP-43^M337V iPSC-derived motor neurons show increase in insoluble TDP-43 and reduced survival. This arsenite-induced motor neuron death assay provides a convenient tool to identify anacardic acid that can promote the survival of these neurons. Anacardic acid reportedly reduces TDP-43 mRNA expression in TDP-43^M337V iPSC-derived motor neurons by 147-fold compared to untreated motor neurons, and reduces the amount of insoluble TDP-43, but not those in the soluble fractions. In addition, anacardic acid also increases the neurite length, increases the expression of NEFM in treated motor neurons, down-regulates the RNA metabolism-related genes, and reverses the changes in TNFa/NFKB signaling pathway (Egawa et al., 2012).

2.4. Modeling Sporadic ALS

Although TDP-43 proteinopathy is a common feature in sporadic ALS, the cause-effect relationship between TDP-43 protein aggregates and the pathogenesis of ALS remains unclear. This is complicated by the observations that mutations in TDP-43 gene TARDBP are quite rare in sporadic ALS cases, and that TDP-43 proteinopathy can also be identified in other neurodegenerative diseases, such as Alzheimer’s disease (Amador-Ortiz et al., 2007). Moreover, none of the rodent models that express wild type or mutant TDP-43 show definitive TDP-43 proteinopathy that recapitulates human pathology. Given these limitations, the iPSC-derived motor neurons provide an ideal and highly relevant model to address this important question. Currently, only one study reports the isolation and characterization of iPSC-derived motor neurons from sporadic ALS cases. While the results require further verifications, this study shows that under basal culture conditions, motor neurons in sporadic ALS patients inherently have a higher propensity to develop intranuclear TDP-43 aggregates even without the presence of stress (Burkhardt et al., 2013). Although these findings are encouraging, these inclusions only capture one feature of TDP-43 proteinopathy in sporadic ALS. There is no evidence that TDP-43 aggregates are identified in the cytoplasm or neuronal processes in sporadic ALS iPSC-derived motor neurons. It is also unclear if these intranuclear TDP-43 aggregates are ubiquitinated or hyperphosphorylated just like those identified in human tissues. Using intranuclear TDP-43 aggregates as readouts, this study further identifies several compounds that potentially may inhibit the formation of TDP-43 protein aggregates.

2.5. Modeling Progranulin Mutations

Autosomal dominant mutations in human progranulin (GRN) gene have been causally linked to FTD. Although it is well-established that non-sense-mediated decay of mutant progranulin mRNA contributes to haploinsufficiency, the exact mechanisms of neurodegeneration caused by PGRN deficiency are not entirely clear. Patient-specific iPSC-derived neurons that carry GRN^S116X mutation show about 50% loss of PGRN mRNA, as well as secreted and intracellular PGRN proteins. GRN^S116X iPSC-derived neurons exhibit increased sensitivity to a number of cellular stress conditions, including treatment with ER stress inducer tunicamycin and inhibitors for PI3 kinase (PI3K), AKT, and ERK/MAPK signaling pathways (Almeida et al., 2012). Interestingly, lentivirus-mediated expression of PGRN in GRN^S116X iPSC-derived neurons rescues the elevated sensitivity to cellular stress conditions. In addition, gene expression profiling show that Grn^S116X iPSC-derived neurons have reduced expression of ribosomal protein S6 kinase beta-2 (RPS6KB2), a member of the S6 kinase family that has been implicated in both PI3K/AKT and MEK/MAPK signaling pathways (Fenton and Gout, 2011).

Although GRN^S116X iPSC-derived neurons have no problem with neuronal differentiation, iPSCs derived from another mutation GRN^IVS1+5G>C show reduced differentiation efficiency into CTIP2−, FOXP2− or TBR1-positive cortical neurons (Raitano et al., 2015). This defect can be corrected by zinc finger nuclease-mediated rescue of GRN cDNA. Interestingly, transcriptome analyses in GRN^IVS1+5G>C iPSC-derived cortical neurons reveal that genes in the Wnt signaling pathway are significantly misregulated (Rosen et al., 2011), and treatment of GRN^IVS1+5G>C iPSCs with Wnt inhibitors can restore the neuronal differentiation phenotype.

2.6. Modeling MAPT Mutations

Mutations in human MAPT gene are a major contributing factor to the pathogenesis of familial FTD (Wolfe, 2009). The neuropathology features in these patients are characterized by the profound tau protein aggregates in the cytoplasm and processes of neurons and glia (frontotemporal lobar degeneration-tau or FTLD-tau). The large number of tau protein isoforms, alternatively spliced products, and various post-translational modifications make it quite challenging to decipher the mechanisms of tau protein aggregate formation and its contribution to disease process. The majority of MAPT mutations are missense mutations that cluster around the microtubule binding domain, suggesting that these mutations most likely perturb the ability of tau to bind to microtubules and cause neuronal dysfunction or even death. Several groups have reported the isolation of iPSC-derived neurons, including one with MAPT^A152T mutation from a patient with progressive supranuclear palsy (PSP), one with MAPT^N279K and MAPT^V337M from FTDP-17 patients, and one with 10+16 splice site mutation in MAPT gene (Ehrlich et al., 2015; Fong et al., 2013; Sposito et al., 2015). The study by Fong and colleagues showed that MAPT^A152T neurons exhibit distinct degenerative features, characterized by breaks, bends and bulges along neuronal processes and reduced neuronal survival (Fong et al., 2013). Interestingly, MAPT^A152T iPSC-derived neurons show phosphorylated tau protein aggregates in cytoplasm and neuronal processes, though no abnormal tau protein aggregates are identified in astrocytes derived from MAPT^A152T iPSCs. Using zinc finger nuclease-mediated gene editing, the authors showed that the phenotypes in MAPT^A152T iPSC-derived neurons can be further aggravated when the mutation is present in both allele (MAPT^152T/T). Conversely, genetic correction of the MAPT^A152T allele to wild type (MAPT^152A/A) eliminates the neuronal phenotypes in isogenic iPSC-derived neurons. Furthermore, MAPT^152A/T and MAPT^152T/T iPSC-derived neurons show very low percentage of dopaminergic neurons, which can be reversed by genetic correction. Together, these results suggest that the dopaminergic neurons may be more vulnerable to neurotoxicity caused by MAPT^A152T mutation. These interesting findings support the idea that MAPT^A152T is a disease risk factor that affects the development and maintenance of neuronal processes.

The effects of MAPT mutations in the development of neuronal processes are further confirmed in neurite outgrowth assays using iPSC-derived neurons with MAPT^N279K or MAPT^V337M mutation (Ehrlich et al., 2015). In addition, iPSC-derived neurons with MAPT^N279K or MAPT^V337M mutation are much more vulnerable to oxidative stress-induced toxicity, which is supported by an increase in unfolded protein response (UPR) and distinct “disease-specific” gene expression profiles. Finally, it is well-recognized that different isoforms of tau can be generated via alternative splicing. Using iPSCs from controls and two patients with 10+16 splice-site mutation in MAPT, Sposito and colleagues show that control iPSC-derived neurons express 3R tau only during the first 100 days in culture (Sposito et al., 2015). Interestingly, prolonged culturing of control neurons switches the tau expression from 3R tau to a diverse complement of tau isoforms. In contrast, iPSC-derived neurons with 10+16 splice site mutation in MAPT express both 3R and 4R tau during the first 100 days in culture. These results suggest that FTD-related splice site mutation in MAPT can override the developmental program that determines splicing-mediated generation of tau isoforms. Given the robust post-translational modifications of tau, it will be useful to take advantage of patient-specific iPSCs to investigate the signaling mechanisms that regulate phosphorylation and acetylation of tau, and how perturbations to these mechanisms contribute to disease.

3. Concluding Remarks and Future Directions

3.1. Conclusions and Potential Caveats

The large numbers of studies using iPSCs derived from ALS and FTD patients have led to several important conclusions. First, it is highly feasible to generate patient-specific iPSCs and harness the genetic editing technology to correct mutations. Second, most iPSC-derived neurons that carry disease-causing mutations can recapitulate certain key features of human disease, including proteinopathy and RNA toxicity. Third, the phenotypes observed in iPSC-derived neurons provide confidence that they can be quite effective models for ALS and FTD. Finally, these studies also support the feasibility of using iPSC-derived neurons as platforms for the discovery of disease-specific therapeutic targets.

Despite these promising results, however, there are several caveats that deserve our attention before entering the next phase of research using iPSC-derived neurons. For example, there is sufficient evidence that phenotypic variations do exist in iPSC-derived neurons. In the case of C9ORF72 mutations, at least two studies have reported distinct differences in the electrophysiological properties in iPSC-derived neurons (Devlin et al., 2015; Sareen et al., 2013). Despite using the same reagents to characterize C9ORF72 iPSC-derived neurons, not all the studies are able to identify RAN translation protein products (Sareen et al., 2013). Furthermore, although all disease-specific iPSC-derived neurons exhibit increased vulnerability to stress conditions, it is unclear if these phenotypes are disease-specific, nor do we know the underlying mechanisms that contribute to these phenotypes. Looking ahead, as we improve and refine iPSC-derived neurons as bona fide disease models, it will be important to compare and contrast the similarities and differences with human neuropathological features. Furthermore, by expanding the repertoire of neuron subtypes from iPSCs, we can begin to determine the molecular mechanisms that define “selective vulnerability” that is a consistent feature in neurodegenerative disease.

3.2. Future Directions

The diverse phenotypes reported in iPSC-derived neurons that carry pathogenic mutations causally linked to ALS or FTD pose challenges for future efforts to identify therapeutic targets. While the ASO-mediated knock down of RNA foci caused by C9ORF72 GGGGCC expansions provide promising therapeutic strategies, it could be a daunting task to restore the hundreds, or thousands, of disrupted RNA processing events that may occur following sequestration of RNA processing factors by repeat expansion RNAs. For example, many proteins involved in RNA metabolism, RNA transport and splicing have been isolated in the interactomes of GGGGCC RNA repeats, but it remains unclear whether ASO-based therapeutic approaches will be sufficient to mitigate the gain-of-function properties of RNA repeats (Lagier-Tourenne et al., 2013; Wheeler et al., 2012). Another challenge for ASO-based approach is to develop highly efficient delivery of ASO into brain and spinal cord. Considering these obstacles, small molecule inhibitors that aim at disrupting specific RNA-protein interactions might circumvent the technical challenges confronting ASO-based therapy and be more effective in releasing endogenous proteins from the gain-of-function properties of RNA repeats (Arambula et al., 2009; Jahromi et al., 2013; Warf et al., 2009). In this context, C9ORF72 iPSC-derived neurons should serve as effective tools to screen for small molecule inhibitors. Given the variations and diversity in human disease, it might be more prudent to develop patient-specific iPSC-derived neurons for the purpose of screening for the best and most effective ASO and small molecule candidates.

Another therapeutic potential is to use iPSC-derived neurons in cell-based therapy to replace disease-damaged neurons and to restore the normal function of specific neural circuits. Several studies in rodents and in chicks have provided proof-of-principle evidence that iPSC-derived neurons can indeed be transplanted into cerebral cortex or spinal cord, and demonstrate functional integration into the existing neural circuits (Espuny-Camacho et al., 2013; Sareen et al., 2014; Toma et al., 2015). Given the available technology to correct the genetic mutations, it is possible that patient-specific iPSC-derived neurons might be able to reduce immune response and improve the chance of graft survival.

Highlights.

• Amyotrophic lateral sclerosis (ALS) and frontotemporal lobar dementia (FTD) are two ends of a disease spectrum

• Comprehensive summary of studies that use iPSC-derived neurons to model ALS and FTD

• Balanced discussions on the unique capabilities of iPSC-derived neurons that capture some key features of ALS and FTD

• Highlight the potentials of iPSC-derived neurons in drug discovery and therapeutics

• Identify critical caveats that require improvements before iPSC-derived neurons can become highly effective disease models