The Evolution of Stem Cells, Disease Modeling, and Drug Discovery for Neurological Disorders

Abstract

Human neurological disorders are among the most challenging areas of translational research. The difficulty of acquiring human neural samples or specific representative animal models has necessitated a multifaceted approach to understanding disease pathology and drug discovery. The dedifferentiation of somatic cells to human induced pluripotent stem cells (hiPSCs) for the generation of neural derivatives has broadened the capability of biomedical research to study human cell types in neurological disorders. The initial zeal for the potential of hiPSCs for immediate biomedical breakthroughs has evolved to more reasonable expectations. Over the past decade, hiPSC technology has demonstrated the capacity to successfully establish “disease in a dish” models of complex neurological disorders and to identify possible novel therapeutics. However, as hiPSCs are used more broadly, an increased understanding of the limitations of hiPSC studies is becoming more evident. In this study, we review the challenges of studying neurological disorders, the current limitations of stem cell-based disease modeling, and the degrees to which hiPSC studies to date have demonstrated the capacity to fill essential gaps in neurological research.

Introduction

The study and treatment of neurological disorders have lagged behind the advancements of other biomedical disciplines. In essence, the need for models of neurological disorders that can generate genuine treatments has been thwarted by obstacles inherent to studying the human brain. However, the generation of human induced pluripotent stem cells (hiPSCs) by Yamanaka and colleagues in 2007 has allowed for new cell-based models of heritable diseases to be developed [1]. The unique value of the technique is the ability to derive hiPSCs from individuals affected by a heritable disease and construct an in vitro platform to study their respective disease. The guiding principles behind utilizing hiPSCs for “disease in a dish” modeling are that these cells have the same genetic background as the affected individual and can be experimented on in a controllable disease-relevant paradigm (Fig. 1). Before the development of reprogramming technologies, the study of many diseases was heavily restricted due to the inability to investigate the human cells the disease manifested in. This is especially true for neurological disorders, where access to human material is limited to postmortem brain samples, the availability of which is often quite low, and likely includes artifacts that skew results. Alternatively, although vital for controlled experiments, in vivo animal models may only partially or vaguely recapitulate human behavior, genetics, and anatomical pathology [2,3]. Translating drug discoveries from animal models to the clinic has been further limited by issues such as species-related differences in drug-metabolism and blood–brain barriers [4]. The addition of hiPSC applications to neuropsychiatric research fills previous gaps by providing the potential for controlled experiments using human neural cells that reflect the complexities of underlying disease pathology. Diseases are often the manifestation of the interplay of cumulative atypical genetic polymorphisms, chromatin structures, DNA methylation, and mitochondrial genomes. Therefore, simple gene replacement strategies using less specific immortalized human cell lines are unlikely to be sufficient at unraveling the multifaceted components of these diseases.

FIG. 1.

hiPSC approaches for neurological “disease in a dish” modeling. Human somatic cells (fibroblasts) are harvested from affected and unaffected individuals and are grown in vitro. Human fibroblasts are reprogrammed into hiPSCs via increased transcription of pluripotency genes. Multiple hiPSC clones are made from each individual, and are then differentiated into the disease relevant neural cell types. Various comparative molecular and cellular analyses can now be performed on relevant human neural cultures for a variety of distinct categories of neurological diseases. Hypothetical experimental approaches and limitations are presented for studying the different disease types. Neurological disorders that have been modeled with stem cells are listed by disease category. In order from left to right, neurological disorders that have been studied via hiPSCs are as follows: amyotrophic lateral sclerosis [93], dystonia [47], Lesch-Nyhan [53], Machado-Joseph disease [54], multiple sclerosis [56], Niemann-Pick disease [58], Spinal muscular atrophy [14], Alzheimer's disease [94], frontotemporal dementia [49], Huntington's disease [95], oliviopontocerebral atrophy [59], Parkinson's disease [96], tauopathy [65], transthyretin amyloidosis [68], aneuploidy syndrome [43], autism spectrum disorders [41], Down syndrome [97], Dravet's syndrome [98], familial dysautonomia [19], fragile X syndrome [99], Prader-Willi syndrome [62], Rett syndrome [100], Timothy syndrome [67], Spinal cord injury [63], bipolar disorder [44], and schizophrenia [101]. hiPSC, human induced pluripotent stem cell. Color images are available online.

Pluripotent stem cells are a unique class of cells characterized by their capacity for self-renewal, and their ability to differentiate into all three germ layers [5]. In human development, pluripotent stem cells appear during embryogenesis in the inner cell mass of a burgeoning blastocyst [6]. hiPSCs mimic embryonic stem cells (ESCs) in their ability to differentiate and self-renew, but carry less ethical quandaries for the biomedical field than embryonically derived stem cells. The starting materials of hiPSCs are nucleated somatic cells, predominantly from the skin or blood, which are dedifferentiated to a stem cell-like state (Fig. 2). Currently, the most common genes utilized for reprogramming are c-MYC, KLF-4, SOX-2, and OCT-4 [1]. Since 2007, reprogramming methodologies have evolved from the original technique of integrating viral vectors to include more advanced techniques such as micro-RNA and nonintegrating episomal approaches [7,8]. Reprogrammed cells may be maintained indefinitely in a pluripotent state, without entering senescence, and can be differentiated into most somatic cell types for study. To date, hiPSCs have demonstrated value in disease modeling alongside in vivo models, clinical studies, and in conjunction with animal transplantation models. The first such demonstration of modeling diseases with stem cells via transplantation occurred before the discovery of iPSCs. Coined as a “poor man's transgenic mouse,” murine brains were engrafted with primary neural stem cells (NSCs) which acted as vehicles for infectious virus production, establishing a novel model of acute spongiform encephalopathy [9,10]. Alternatively, animal models may be used to demonstrate functionality or therapeutic replacement potential of hiPSC derivatives, as has been demonstrated in rat models of Parkinson's disease (PD) [11]. Thus, transplantation, both for disease modeling and validating cellular function, may be a powerful strategy for understanding disease-associated phenotypes in hiPSC studies.

FIG. 2.

Human cell reprogramming and differentiation. Diagram of the workflow from patient-isolated somatic cells to mature neural cultures. Far left image is phase contrast of skin fibroblast cells. The second image from the left is of somatic cells that are reprogrammed into induced pluripotent stem cells and stained for the pluripotency marker OCT-4. The second from the right image is of neural precursor cells differentiated from hiPSCs and are stained for the nuclear ectodermal marker PAX-6. The far right image is of mature neural cells differentiated from hiPSCs and are stained with the mature neuron cytoskeletal marker MAP2 and the astrocyte marker GFAP. Scale bars are 10 μm. Color images are available online.

Before the discovery of induced pluripotent stem cells, researchers tried for decades to obtain human NSCs (hNSCs) to study brain development and neurological disorders. Early efforts derived cells from postmortem specimens; however, these cells were unreliable and were largely replaced by NSCs isolated from olfactory epithelial biopsies. The caveats of using biopsies as a pool for NSCs were that they required an invasive medical procedure and required additional steps to isolate the NSCs from heterogeneous mixtures of cell contained in the biopsies [12]. There have also been roles for genetically modified human ESC lines, but that approach is overall less relevant to disease, although much of the work contributed to understanding and manipulating hiPSCs [13]. Nevertheless, hiPSC technology facilitates access to an unlimited amount of disease-state NSCs and neural cells at a reasonable cost and provides an advantageous platform for studying neurological disorders (Fig. 2).

hiPSC Disease Modeling Discoveries in Neurological Disorders

Owing to the fact that hiPSCs carry the identical genetic background as the source, the first neurological hiPSC studies focused on monogenic diseases. In the seminal study to establish that hiPSC models could faithfully emulate a heritable phenotype in vitro, Ebert et al. in 2009 [14] examined hiPSC-derived motor neurons from individuals affected with spinal muscular atrophy (SMA) caused by an aberration in a single gene called survival motor neuron 1 (SMN1). These motor neurons not only reflected the expected abrogation of SMN1 transcription but also were diminished in size compared with controls. In addition, the phenotype could be rescued with valproic acid and tobramycin treatment [14]. Further study of SMA hiPSC-derived neurons identified new disease characteristics and possible drug targets. In 2012, Sareen et al. found aberrant Fas ligand-mediated apoptosis in hiPSC-derived SMA motor neurons, and discovered that the dysfunction is rescued by a Fas blocking antibody or a caspase-3 inhibitor [15]. Since then, further hiPSC studies have found additional compounds such as thyrotropin-releasing hormone analog, 5-oxo-l-prolyl-l-histidyl-l-prolinamide, which was able to increase SMN1 protein levels reportedly through glycogen synthase kinase-3 beta (GSK3β) inhibition [16]. An innovative hiPSC study for SMA was centered around SMN2, paralogous gene in humans, which normally produces low levels of SMN protein. In the study led by Naryshkin et al., they identified that the alternative splicing compound RG7800 can increase SMA protein levels via the SMN2 gene, which is now undergoing clinical trials [17]. SMA, as well as other monogenic diseases, is conducive to hiPSC studies due to its established connection between pathology and disease-specific genotypes. Monogenic diseases, however, constitute only a small portion of neurological disorders, and the robustness of hiPSC technology will be evaluated by its capability to model complex polygenic disorders and produce translational breakthroughs.

Familial dysautonomia

Another monogenic neurological disorder, familial dysautonomia (FD), has been studied extensively by Lee et al. [19] since 2009. A subtype of FD is caused by a mutation in the IKBKAP gene and leads to the loss of neurons in the peripheral nervous system [18]. Utilizing hiPSC-derived peripheral neurons and neural crest precursor cells, Lee et al. found tissue-specific transcriptional aberrations of IKBKAP mRNA in vitro, signifying a possible breakthrough in understanding previously unknown pathophysiological mechanism of FD [19]. Transcriptional and cellular analysis of FD in vitro revealed deficiencies in migration and neural differentiation, which positioned researchers to screen for drugs. In a 2012 follow-up study, Lee et al. [20] used FD hiPSC-derived neural cells as a drug-screening platform for validating the efficacy of compounds capable of reversing the mis-splicing-mediated reduction of IKBKAP transcription, neural differentiation deficits, and migratory defects. The screen identified a compound capable of inducing IKBKAP transcription via cAMP and CREB phosphorylation, as well as rescuing diminished IKAP levels in FD hiPSC-derived autonomic neurons, which is related to FD's pathological loss of autonomic neurons [20]. Other approaches for treating FD include inhibiting 26S proteasomes as they contribute to lowering levels of IKBKAP via protein degradation. A French study found that inhibiting proteasomes with Food and Drug Administration (FDA) and European Medicines Agency approved drugs could increase IKBKAP mRNA levels in ecto-mesenchymal stem cells from an FD patient. Specifically, the proteasome inhibitor carfilzomib demonstrated the best results, including regulation of kinetin in improving IKBKAP isoform ratios that counterbalanced the drug's toxicity [21].

Amyotrophic lateral sclerosis

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease where there is a progressive loss of motor neurons throughout the body, resulting in paralysis and death [22]. Understanding the mechanism underlying ALS has been difficult due to the inaccessibility of diseased human motor neurons. In 2012, Egawa et al. [23] made inroads into elucidating familial ALS related to mutation of the Tar DNA-binding protein-43. The study was able to produce hiPSC-derived motor neurons that exhibited the disease phenotypes of cytosolic aggregates and diminished neurite length. Molecular analysis of the hiPSC-derived motor neurons found increases in expression of RNA metabolism genes and decreased expression of cytoskeletal genes. After screening compounds on the ALS motor neurons, the study discovered a histone acetyltransferase inhibitor, anacardic acid, capable of rescuing the cellular diseased phenotype observed in ALS patient hiPSC-derived motor neurons [23]. In addition, research performed on mutant SOD1 ALS hiPSC-derived motor neurons by Yang et al. found that a potent dual GSK3 and HGK inhibitor, Kenpaullone, could extend the survival of ALS motor neurons longer than two compounds used in ALS clinical trials [24]. Naujock et al. found that ALS motor neurons have imbalanced sodium and potassium ratios which may cause endoplasmic reticulum stress, and these deficits can be rescued with the FDA-approved compound 4-aminopyridine [25]. As it stands, with the wide array of discoveries being made regarding ALS with hiPSCs, researchers are surely advancing toward a clinical breakthrough. The closest hiPSC discovered drug to the clinic for ALS may be the repurposed antiepileptic drug, ezogabine, which has advanced through phase 2 trials for ALS [26]. Stem cell transplantations are also showing promise for future ALS treatments. A recent phase 1 trial showed microtransplanting hNSC into the spinal cords of ALS patient decreased disease progression [27]. On a broader note, this is an exciting study because it is the first of its kind to transplant highly standardized, not immortalized, and reproducible hNSC in a safe manner, which will positively influence hNSC transplantation studies for other conditions.

Alzheimer's disease

Alzheimer's disease (AD) is the most prevalent neurodegenerative disorder with no disease modifying treatment and represents one of the most daunting health care issues facing modern medicine. The need for scientific breakthroughs that can lead to the development of therapeutic treatments for AD is paramount, and stem cell-based studies have begun to provide discoveries due to their capacity to model AD in vitro. In 2011, Yagi et al. [28] were able to establish a hiPSC model of the monogenic autosomal-dominant version of familial AD, where there is a mutation in either of the presenilin genes. While differentiating the familial AD hiPSC lines into neurons, the study observed increased levels of amyloid-beta 42 secretion, a known molecular phenotype of the disease. More importantly, Yagi et al. [28] were able to demonstrate that the aberrant increased secretion of amyloid-beta could be ameliorated by using gamma-secretase inhibitors. These findings highlight the potential of hiPSC-based studies for screening, identifying, and validating candidate drugs for the treatment of AD [28]. One of the major hurdles for discovering therapeutics for AD, and neurological diseases in general, is that the plurality of active compounds developed so far have toxic off-target effects. Work by Ben Halima et al. has attempted to address the toxic effects of beta-secretase inhibitors, by designing endosomal specific beta-secretase inhibitors that will preferentially impact the AD pathology of amyloid-precursor protein (APP), and not non-AD beta-secretase substrates such as neuregulin [29]. Furthermore, AD research and modeling studies are now starting to successfully investigate therapeutic possibilities outside traditionally perceived AD drug targets. One such hiPSC AD study found that the plant-derived polyphenol, apigenin, was able to protect human neurons from AD-associated excitotoxicity and apoptosis via global downregulation of nitric oxide and cytokine release [30]. To optimize lead identification, hiPSC are also being applied to generating drug screening platforms. Kondo et al. established an in vitro pathogenic amyloid-beta production assay with AD patient iPSC-derived neurons, and was able to identify multiple novel lead compounds capable of lowering amyloid beta levels in both familial and sporadic AD neurons [31]. Paired with a recent report that a novel retromer stabilization compound reduces amyloid-beta levels and independently inhibits tau-hyperphosphorylation, these AD studies are providing momentum toward a possible medical breakthrough [32].

An interesting aspect about AD pathology is its association with other significant disorders, such as Down syndrome, which is caused by trisomy of chromosome 21. Individuals with Down syndrome are at increased risk of developing AD because chromosome 21 carries major AD-associated genes, specifically APP. In 2012, Shi et al. were able to validate that Down syndrome hiPSC and ESC-derived neurons recapitulated key AD molecular pathology and could be a useful platform for screening possible AD candidate drugs and designing novel AD intervention therapies [33]. In addition, Shi et al.’ [33] use of Down syndrome stem cell lines in modeling AD demonstrated that stem cell technologies could model polygenic late-stage neurological disorders, as disease symptoms do not manifest clinically until decades into human life.

Schizophrenia

The use of hiPSC technology for disease modeling and drug discovery is also being tested in relationship to complex disorders, particularly psychiatric disorders. Initial focus of psychiatric hiPSC research has been schizophrenia (SCZD), a psychotic disorder which can include symptoms of hallucinations, delusions, cognitive dysfunction, and affective blunting [34]. Analysis of SCZD hiPSC-derived neurons showed reduced neuronal connectivity due to diminished neurite number when compared to healthy controls, a typical phenotype observed in postmortem SCZD brains. Further investigation revealed reduced PSD95 and glutamate receptor levels, as well as altered expression profiles of genes in the cAMP and WNT signaling pathways. It was demonstrated that the antipsychotic medication, Loxapine, could partially rescue the diminished neuronal connectivity associated with SCZD in vitro [35]. Several SCZD hiPSC studies have been attempted to further elucidate the underlying molecular pathology and its relationship to the mechanisms of drug actions. One such hiPSC study performed by Paulsen et al. [36] in 2012 identified cellular respiration and oxidative stress as possible core factors in SCZD pathogenesis. These pathways have been previously implicated in SCZD postmortem studies. Specifically, increased levels of reactive oxygen species were observed in the SCZD hiPSC-derived neurons, which were rescued with the known antipsychotic, valproic acid [36]. Stem cells studies are now showing their clinical utility beyond drug discovery. Nakazawa et al. [37] studied the transcriptomic differences between hiPSCs-derived neurons from schizophrenic twins discordant for responding to the antipsychotic drug clozapine. While genes associated with cell adhesion molecules were observed to differentially regulate according to clozapine responsiveness, the study also noted that its findings could also be the basis for identifying a biomarker for diagnosing SCZD and predicting drug responsiveness [37]. Interestingly, iPSC-derived cell products may be a viable treatment for SCZ. As hyperexcitable phenotypes are being discovered in SCZ studies [38], transplantation of inhibitory GABAergic interneurons is now being proposed as a viable strategy [39,40].

In recent years, there has been an increase in hiPSC studies of neurological disorders, demonstrating the power of hiPSC technology for modeling complex diseases that so far have been difficult to study via other methods. Other neurological disorders that have been studied with stem cells include autism spectrum disorder [41,42], aneuploidy syndrome [43], bipolar disorder [44], Down syndrome [45], Dravet's syndrome [46], dystonia [47], fragile X syndrome [48], dementia [49,50], Huntington's disease [51,52], Lesch-Nyhan syndrome [53], Machado-Joseph disease [54], mucopolysaccharidosis [55], multiple sclerosis [56,57], Niemann-Pick disease [58], oliviopontocerebral atrophy [59], Rett syndrome [60,61], Prader-Willi syndrome [62], spinal cord injury [63,64], tauopathy [65,66], Timothy syndrome [67], transthyretin amyloidosis [68], and Wilson's disease [69].

The Limitations of Reprogramming and hiPSC Studies

The first transgenes utilized for transcriptional reprogramming were identified by Yamanaka and colleagues as KLF-4, SOX-2, c-MYC, and OCT-4, but other gene combinations are now possible to create hiPSCs [70]. The initial zeal of hiPSCs revolved around the notion that researchers could finally study any disease in vitro because hiPSC technology provided unfettered access to previously unattainable live human cells harboring disease pathology. As more hiPSC studies have been conducted, the aforementioned specious optimism moderated to more reasonable expectations. An ongoing limitation in the field is reproducible differentiation of hiPSCs to relevant cell types. Even with well-established protocols for the most accessible derivatives, idiosyncratic differences from individual clones and hiPSC lines are observable. Early reprogramming techniques used viruses that integrated into the host genome, causing unintended perturbations in the genetic sequence [71]. To overcome the myriad of confounds that came from studies, which utilized integrative viral reprogramming, the field moved to non-DNA altering methods. Some of the different nonintegrating methods for generating hiPSCs include Sendai viruses [72], episomal vectors [73], nonintegrative adenoviruses [74], micro-RNAs [8], and piggyBac transposons [75]. Although nonintegrative reprogramming methods address the previous genetic heterogeneity found in hiPSC-derived cells, it is unclear to what extent epigenetic changes are inherent to a dedifferentiated state specific to the genotype versus stochasticism related to reprogramming, with the most heavily studied epigenetic dissimilarity being methylation profiles [76]. The extent these epigenetic discrepancies have on preventing hiPSC-derived cells from accurately modeling diseases has only been marginally studied, with the full impact being far from known.

Every step in reprogramming and differentiating hiPSCs generates variability that can confound future results. Several studies have demonstrated that reprogramming somatic cells into hiPSCs generates random changes to the original genetic sequence in the form of copy number variants (CNVs), indels, or chromosomal anomalies [77,78]. The abnormalities induced by the reprogramming and differentiation process can undermine the disease-specific genetics that hiPSC “disease in a dish” modeling is intended to conserve. Undoubtedly, many of the de novo genetic mutations that arise in hiPSC-derived cells could affect gene expression directly or indirectly. This can in turn create numerous false positive and false negative results via the application of a competitive growth bias over unvaried cells, a selection bias over unvaried cells, or unwitting creation of a heterogeneous clonal population. The CNV [71,79] and methylomic [80] aberrations inherent to hiPSC disease modeling have already been intensely studied, but a remaining underinvestigated confound of hiPSC neurological disease modeling studies is transposable elements (TEs).

Transposable elements

TEs—also known as transposons—are DNA sequences capable of moving from one location in the genome to another. TEs comprise over half of the human genome [81]. Active TEs can also create CNVs, and this phenomenon has been associated with neurodevelopmental disorders such as autism spectrum disorder [82,83] as well as neuropsychiatric disorders such as SCZD [84]. TEs contribute to neuronal diversity and plasticity during normal development, which is a key dysregulated mechanism in many neurological disorders [85]. Active TEs during development have the capability to produce genomically heterogeneous neurons within the same individual. This leaves open the possibility that for some neurological disorders, only individual populations of cells may be diseased while the rest of the subject's cells are unaffected. A study highlighted this concern, demonstrating how ample genetic mosaicism exists across individual neurons in a single human brain [86]. Although this study was not performed in brains affected by disease pathology, it also revealed that hiPSC-derived neurons contained more CNVs than the initial starting fibroblasts of origin, as well as more genomic aberrations than the initial neural progenitor cells. A present apprehension of hiPSC technology is the possibility that some neurological diseases cannot be hiPSC-modeled with appropriate fidelity and that findings from hiPSC disease modeling studies may be masked or lost by the reprogramming and differentiation process. The association CNVs [87] and indels [88] have with neurological disorders has been demonstrated by myriad studies. Nonetheless, as hiPSC disease modeling technology advances, more intensive measures of genomic hiPSC verification will become necessary. hiPSC-based models of neurological disorders should attempt to maintain the genome of the disease, the individual, and the impacted cells of interest. However, somatic mutations appear inherent to reprogramming mechanisms, including nonintegrating approaches [87]. Therefore, the effect reprogramming has on creating bona fide hiPSC lines with regard to CNVs, TEs, and indels requires continued study. A technique that could theoretically overcome this epigenetic reprogramming confound is somatic cell nuclear-transfer (SCNT) generated stem cell lines, but experiments need to be performed to verify such an assumption. SCNTs are similar to induced pluripotent stem cells, but instead of reprogramming somatic cells via transcription of pluripotency genes, a somatic cell's nucleus is injected into an enucleated oocyte arrested at metaphase-II, which then enters an ESC-like state with the genome of interest [89]. In a study comparing the genetics and epigenetics of ESCs, hiPSCs, and SCNT ESCs, the researchers found the SCNT ESCs more faithfully mimicked ESCs than hiPSCs, however, transposon profiles were not specifically addressed [77].

Isogenic hiPSC lines

A more realistic approach that hiPSC studies are undertaking to control for confounds related to the discrepancies between the genetic fidelity of a disease and disease-generated hiPSC lines is the use of isogenic paired controls. Isogenic paired hiPSC lines include a monogenic mutant line, and a disease-free line from the same individual. The assumption is that the isogenic line's genome should only differ at the gene of interest, which therefore should control for background genomic aberrations caused by reprogramming. The isogenic paired lines are created with DNA editing techniques such as zinc-finger nucleases or CRISPRs, where either a disease-specific mutation is introduced into a healthy hiPSC line to create a paired diseased line, or a diseased hiPSC line's mutation is edited to create a healthy wild-type line. Isogenic paired hiPSC approaches have already been successfully applied to studying neurological disorders. Rett syndrome is a monogenic autism spectrum disorder caused by a mutation in the MECP2 gene and is a debilitating developmental X-linked chromosomal disorder. Ananiev et al. created several isogenic paired Rett syndrome lines and found the diseased hiPSC-derived neurons had smaller nuclear sizes compared to healthy wild-type controls, a known disease phenotype [90]. Further demonstration of the utility of isogenic hiPSC studies was achieved by Ryan et al.’ [91] research into PD. Isogenic pairs were established for a well-known pathogenic PD mutation in the alpha-synuclein gene, and the group studied the relationship the mutant alpha-synuclein neurons had with toxin-induced nitrosative-oxidative stress in impacting MEF2C-controlled transcription. The study was able to glean mechanistic understanding into environmentally mediated PD pathogenesis, by demonstrating that redox stress inhibits the MEF2C pathway, and contributes to mitochondrial damage [91]. Aided by the decrease in experimental variance isogenic-paired lines provide, when a high-throughput screen was performed to find small molecules, it confirmed the MEF2C transcriptional network as a drugable target for treating PD. The screen identified the molecule isoxazole to be an activator of MEF2 and PGC1-alpha transcription in hiPSC-derived PD neurons and found it protective against mitochondrial toxin driven apoptosis. Isogenic drug screening has been effective for other diseases as well. A group at the Gladstone Institute has demonstrated isogenic AD lines can be used in high-throughput screening. Utilizing a library of pharmacologically active compounds, they identified a class of adrenergic receptor agonist compounds capable of reducing endogenous human tau [92].

Discussion

The ultimate goal of hiPSC disease modeling is to faithfully and feasibly model human diseases to make previously unattainable medical and therapeutic discoveries. The utility of hiPSC technology for neurological disorders is in its earliest stages. Significant considerations still need to be paid though, and additional confounds must be addressed for future hiPSC studies. These issues include, but are not limited to, the correspondence of reprogrammed and differentiated genomes compared to the disease's genome, the reproducibility of techniques and results, establishment of relevant disease pathology assays, and designing financially realistic large scale studies. The time and efficiency of neural differentiation protocols for generating mature, clinically relevant cell types are currently not optimal enough to perform large-scale studies to model most neurological disorders. Understanding the advantages and limitations of patient-specific neural derivatives in the context of the disease pathophysiology is critical to enhancing the benefit of such studies which is essential given the cost and time requirements involved.

At the current forefront of hiPSC technology is in vitro organoids and their use for modeling tissue-wide phenotypes of a disease. The appeal of modeling disease effects during brain development, in an organ like structure, is already changing the landscape of the hiPSC field. With the development of organoid approaches come additional hurdles and pitfalls. It will be incumbent on researchers to demonstrate that their organoids are not just a meaningless collection of cells that can be assumed to recapitulate every aspect of the brain, but a faithful representation of a specific process the study is attempting to investigate.

Indeed, neurological diseases, in which a known specific neural derivative or neuronal subtype largely underlies the pathophysiology, are well suited for human hiPSC studies and potentially drug discovery. Alternatively, for monogenic neurologic disorders, a specific genetic lesion may be utilized as a starting point for understanding its effects on neurogenesis, morphology, neuronal function, migration, and cell–cell connectivity. This approach may be required in lieu of focused anatomical or cell-type pathology. For complex diseases such as psychiatric disorders, this is generally not a tenable strategy as these entities are more likely to be polygenic, episodic, or triggered by environmental stimuli. Nevertheless, hiPSC derivatives may provide a new tool to investigate or isolate the effects of candidate single-nucleotide polymorphisms and CNVs. Conversely, defining specific neural and neuronal phenotypes has been challenging as many morphological parameters are nonspecifically altered in a wide range of pathophysiological processes, and cellular and anatomical phenotypes can be subtle, heterogeneous, or altogether elusive to advanced imaging techniques. With increasingly sensitive and focused imaging studies, new leads in functional neuroanatomy may ultimately lend to development of in vitro neural microenvironments. Although this requires the ongoing development of more sophisticated in vitro methods for recapitulating the neural milieu, including interaction with immune cells and inflammatory signaling components, this is particularly important for approaching unchartered hiPSC disease modeling territory such as traumatic, infectious, and immunologic neurological processes that have to date been more amenable to transplantation studies in relevant animal models. Bioengineering approaches that should improve the tractability of hiPSC modeling are also being rapidly developed. These include vessels for coculture systems for examining interactions of heterogeneous cell populations, three-dimensional scaffolds for recapitulating cellular structures and compound diffusion and uptake, fluidic chambers providing for highly controlled culture environments, and electrode plates allowing for real-time measurement of neural connectivity in culture. Ultimately, these devices may facilitate more advanced drug screening assays. A rapidly expanding area of focus for finding novel neurological therapeutics is the repurposing of pharmacologic compounds for unrelated diseases. An alternative approach to repurposing is expanding the range of therapeutic compounds that modulate established drug targets or synergistically enhance the action of proven neuropsycho-pharmaceuticals. Neurons derived from hiPSCs provide a physiologically relevant screening tool with amenable assays for cell proliferation and apoptosis, protein level or posttranslational modification, cellular phenotypes such as neurite or organelle morphology, and calcium kinetics. There are a number of experimental pitfalls in utilizing hiPSC derivatives, including genomic integrity as well as growth and differentiation proclivities idiosyncratic to particular cell lines. In addition, practical limitations to sample sizes in hiPSC studies related to clinical availability, labor, and expense of maintenance of independent cell lines are slowly improving. However, these caveats are less likely to be impactful or misleading if cross validation, confirmation, and investigation with other models is applied to such findings, as it is necessary to demonstrate the significance of the findings for translational endpoints, particularly in neuropsychiatric diseases.

The capability to model neurological disorders and make translational breakthroughs with patient hiPSC-derived cell lines is an exciting progression in the field of biomedical sciences. However, only a portion of neurological disorders that have been modeled with hiPSCs have produced breakthrough results. As the limitations of current hiPSC studies become addressed by new reprogramming and differentiation technology, more diseases will become amenable to “disease in a dish modeling.” The research reviewed thus far has established the value of hiPSC technology and its cemented role in the field of biomedical research.

Ethics Statement

The authors have also confirmed that this article is unique and not under consideration or published in any other publication and that they have permission from rights holders to reproduce any copyrighted material.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

Funding was provided by National Institute of Mental Health (RC2MH090011-02 and Library of Integrated Network-based Cellular Signatures Program; California Institute of Regenerative Medicine training grants; International Bipolar Foundation; and the Viterbi Foundation Neuroscience Initiative.