The Promise and Perils of Compound Discovery Screening with Inducible Pluripotent Cell-Derived Neurons

Abstract

Neurological diseases comprise more than a thousand ailments that adversely affect the brain and nervous system. When grouped together, these neurological conditions impact an estimated 100 million individuals in the United States and up to a billion people worldwide, making drug discovery efforts imperative. However, recent research and development efforts for these neurological diseases, including Alzheimer's disease and amyotrophic lateral sclerosis, have been exceedingly disappointing and typify the challenges associated with translating in vitro and cell-based discoveries to successful preclinical models and subsequent human clinical trials. Our viewpoint is that neuronal progenitor cells and neurons derived from inducible pluripotent stem cells afford an innovative translational bridge, with higher pathological relevancy than previous cellular models. We outline some of the opportunities and challenges associated with their evolving usage in drug discovery and development.

Introduction

Neurological diseases globally cause significant disability and death. Although some neurological conditions are common and well known (e.g., migraine headaches, traumatic brain injury, chronic pain, and stroke), most are rare, with few individuals afflicted (e.g., Alexander Disease, Bobble Head Doll Syndrome) (Table 1). Of the more than 7,000 known rare diseases,^* about 360 are primarily neurological.¹

Table 1.

Common and Rare Neurological Diseases

Common neurological disorders	Estimated cases (global)	Rare neurological disorders	Estimated cases (global)
Migraine headache	1.04 billion42	Amyotrophic lateral sclerosis	222,80143
Traumatic brain injury	69,000,00044	Guillain-Barré syndrome	100,00049
Epilepsy	65,000,00045	Batten disease	14,000a
Alzheimer's disease and other dementias	47,000,000b	Familial dysautonomia (Riley-Day syndrome)	595c
Stroke	15,000,000d	Alexander disease	500e
Parkinson's disease	6,100,00046	Bobble head doll syndrome	5847
Multiple sclerosis	2,100,00048
Traumatic spinal cord injury	250,000–500,000b

^aFrom BatCure.

^bFrom World Health Organization.

^cFrom Familial Dysautonomia Foundation, Inc.

^dwww.world-stroke.org

^ehttps://ghr.nlm.nih.gov/condition/alexander-disease#statistics

According to a recent study, the incidence of neurological diseases and disorders is expected to increase in the United States from 43 million in 2011 to 84 million in 2050.¹ This may be unsurprising, in some respects, given that the average U.S. life expectancy is approaching 80 years according to the CDC and neurological disorders are more prevalent in older populations.¹ Nonetheless, this represents an unprecedented increase in the number of people affected by neurological maladies. Thus, it is more important than ever to find new treatments that can modify, treat, and/or prevent neurological diseases.

Historically, parental neuroactive chemotypes or substances have been discovered through behavioral phenotyping, often serendipitously, using whole organisms,² as has been the case with many neuropsychiatric drugs or substances.² However, contemporary drug discovery is predominantly performed on the “microscale/nanoscale” and is propelled by cell-based phenotypic and/or target-based (e.g., in vitro biochemical) screening, which precedes in vivo evaluation (Table 2).

Table 2.

Benefits and Challenges of Compound Identification Strategies Through Phenotypic and Target-Based Screening

Screening approach	Screening assay components	Disease molecular basis	Benefits	Challenges
Phenotypic (forward pharmacology)	• Whole organism	Unknown	• Generally, results in more first-in-class drugs	Target deconvolution required
	• Cell based
			• Identification of cell-permeable compounds
Target-based (reverse pharmacology)	• Cell-based	Known	• Often leads to follow-on and/or best-in-class drugs	May have issues with cell permeability
	• In vitro biochemical
			• Target is already identified

There are strengths and weaknesses to these screening-based approaches for drug discovery. Specifically, while data suggest that more first-in-class drugs are discovered using phenotypic assays, more “follow-on” drugs and/or “best-in-class” drugs are identified using target-based approaches.³ Accordingly, compounds identified using phenotypic screens will most likely be subject to more target deconvolution steps. Nonetheless, drug discovery and development across all disease modalities are challenging and, unfortunately, the pathophysiology of many neurological disorders is often not well-characterized, making efforts even more complicated.

Recent well-publicized neurological disease drug failures^4,5 highlight the frustrating drug discovery landscape and underscore that current in vitro and in vivo model systems do not accurately recapitulate the complexities and progression of a neurological disorder/disease in humans. However, efforts to promote brain research, such as the National Institutes of Health's Brain Research through Advancing Innovative Neurotechnologies (BRAIN) Initiative, the Human Brain Project, the Cure Alzheimer's Fund, and the Allen Institute for Brain Science, are accelerating our understanding of the human brain and neurological disorders.

Multiple resources are accessible to researchers through these programs and include grants, opportunities for collaboration, searchable databases for data mining, and training. Key to progress will be the integration of research across scientific disciplines (e.g., imaging, nanoscience, stem cell technologies, bioinformatics, pharmacology, neuroscience, immunology, and technology development), underscoring the need for team-based science and the development of more physiologically representative neuronal cell-based assay systems.

Use of Inducible Pluripotent Stem Cells in Discovery Efforts

Historically, the heavy reliance on cancer cell lines for the discovery of neurologically targeted chemical probe and drugs has created, perhaps, an unavoidable “disconnect” between the model system and physiological relevancy. Conversely, primary rodent neuronal cultures, while physiologically relevant, have limited translatability to human systems and are not necessarily amenable to the conversion to large-scale compound or genetic screening assay formats. These challenges can now largely be circumvented by the use of neural progenitor cells (NPCs) derived from inducible pluripotent stem cells (iPSCs) for the creation of novel, neurologically relevant cell-based screening assays.

Recent studies have supported the critical inroads that inducible pluripotent stem (iPS) NPCs and neurons have had on screening efforts. For example, human iPS-derived NPCs were foundational in a high-content screening (HCS) assay used to identify drug candidates that not only inhibit Zika virus infection but also rescue Zika virus-induced growth and differentiation defects in NPCs.⁶ Moreover, modulators of neurite outgrowth, critical for neural network generation and nerve regeneration, were identified using an iPS neuron-based HCS assay.⁷

For Alzheimer's disease (AD), a nonautomated, three-dimensional (3D) human neuron cell culture model has recently been described that mimics in vivo amyloid β (Aβ) and tau pathology.⁸ In this study, human neural iPSCs were engineered to overexpress APP and PSEN1 with human familial AD mutations. In vitro, these cells expressed extracellular Aβ aggregates and Aβ-induced pathological tau aggregation; however, this model system did not recapitulate the neuroinflammatory environment observed in human AD patients and AD mouse models. To address this deficiency, this same research group then developed a 3D human in vitro cell culture model using iPS neurons, astrocytes, and microglia in a 3D microfluidic platform.⁹ This model maintained key human AD pathologies (i.e., Aβ aggregation and phospho-tau accumulation) and introduced the AD neuroinflammatory environment (i.e., microglial recruitment, inflammatory cytokine release, axonal cleavage, and nitric oxide release). To date, data from HCS screening with these two AD-iPS neuron relevant assay formats have not been published. Thus, the impact of iPS neurons on high-throughput screening, lead identification and optimization, and drug development has yet to be fully realized. Nonetheless, iPS NPCs and neurons have expanded our ability to interrogate neuronal functional endpoints beyond neuronal toxicity, when damage is too great for therapeutic repair.

It seems likely, however, that next-generation HCS assays will capitalize on these complex co-culture systems as well as other innovative human iPSC-based technologies such as 3D neural or cerebral organoids (i.e., “mini-brains”), which mimic brain development, cell-cell and cell-extracellular matrix interactions, and cellular organization in neurotypical or neuroatypical brain states.¹⁰ These mini-brains may be useful for screening applications, although perhaps not with the throughput previously available with two-dimensional (2D) cell culture platforms.¹¹ Nonetheless, cerebral organoids can reach sizes of up to a few millimeters and represent a variety of pathologically important brain regions, including dorsal cortex, ventral forebrain, midbrain-hindbrain boundary, choroid plexus, hippocampus, and retina.¹²

These innovative 3D models have been supplemented with microphysiological systems comprising “brains-on-chips” or engineered 3D constructs using multiple cell types, including human iPS neurons, which have the potential to expand pharmacological and toxicological screening strategies as well as studies of the blood–brain barrier.^13,14 It is noted these more physiologically and highly complex model systems have not yet yielded tangible successes in the field of drug discovery and development.

Benefits of iPSC-derived NPCs and Neurons for Assay Development and Screening

There are multiple benefits to working with iPSC-derived NPCs and neurons, and foremost, these cell populations circumvent any of the ethical issues, as well as the experimental challenges, which exist with the use of human embryonic stem cells.¹⁵ iPSC-derived NPCs can be maintained in culture and scaled up, in a reasonable amount of time, enabling the creation of low passage number cell banks and a reliable source of cells for experimentation, a critical component of HCS. iPS-derived NPCs and neurons can also be engineered, by eukaryotic expression vector transfection or lentiviral transduction, to ectopically express a variety of proteins.^8,9,16 Moreover, there are a variety of reagents available that can be used to track the maturation of neurons (from NPCs), including antibodies to neuronal nuclear antigen (NeuN), microtubule-associated protein 2 (MAP2), and Milli-Mark™ Fluor-pan, a monoclonal antibody blend that binds to multiple neuronal proteins^17–20 (Fig. 1).

Fig. 1.

Inducible pluripotent NPCs differentiated for 6 weeks. NPCs were cultured in neural differentiation medium for 6 weeks in 96-well microtiter plates. Neurons were fixed, permeabilized, and processed for immunofluorescence using antibodies for mature neurons (i.e., NeuN and MAP2) and DAPI to visualize the nucleus. Images were captured by a Perkin Elmer Operetta high-content screening confocal microscope (63 × objective) using Harmony software. Scale bar = 50 μm. MAP2, microtubule-associated protein 2; NeuN, neuronal nuclear antigen; NPC, neural progenitor cell. Color images are available online.

Critically, differentiation protocols are also available to derive specific neuronal populations, including motor, sensory, interneurons, and neurons in the brain (i.e., dopaminergic, GABAnergic, cholinergic, glutamatergic, and serotonergic neurons), allowing the derivation of complex co-culture systems, which may better mimic the human brain.^21–27 Since human iPS-derived NPCs and neurons can also be used without artificially enforcing protein expression, the proteins and receptors they express are likely to be found at more physiological levels, which could assist the translational fidelity of screening assays using these cells.

Examples also exist of disease-specific iPSC-derived NPCs and neurons, which should enable clarification of the molecular basis of human brain diseases and facilitate the identification of new drug targets,^22,23,28 as well as to provide a screening platforms for small molecules that can reverse or ameliorate the disease phenotype in vitro and, ultimately, in vivo.^29,30 Moreover, it is conceivable that iPSCs, NPCs, and neurons can be generated from patients at different disease stages, including those before overt symptom onset, providing experimental strategies to identify new disease and disease-stage biomarkers, as well as molecular drug targets.³¹

Challenges Associated with iPSC-derived NPCs and Neurons in Assay Development and Screening

Despite the significant theoretical and experimental benefits of using iPSC-derived NPCs and neurons for compound screening, there are considerable challenges associated with their wide acceptance. For example, iPS NPCs and neurons are often used to study diseases of old age (i.e., AD and Parkinson's Disease), but there is increasing evidence suggesting that iPSC reprogramming resets the epigenetic state, making the cells more epigenomically similar to embryonic neurons.^32–34 Moreover, while NPCs are more facile to culturing than the differentiated neurons, they are growth factor dependent, requiring frequent medium changes, and may be subject to selection pressures when passaged for extended time periods or at low confluency. However, there is evidence that suggests extended passaging actually enables the complete reprogramming of iPSCs to neural populations.³⁵ Nonetheless, NPCs, although critical, have limited application as they have not yet been differentiated into neurons.

iPS Neuron Differentiation Medium Composition

Conversely, iPS neurons offer more challenges with respect to culturing and, subsequently, assay validation and optimization. Overall, the number of protocols for differentiating iPSC-derived NPCs into specific neuronal lineages is limited and they are not standardized, constraining systematic transferability and potential reproducibility.

Neuronal differentiation medium composition is also problematic. It varies among research groups and commercial sources, and there is evidence that the base medium commonly used for iPS neuronal differentiation may not be optimal. The critical nature of the neuronal differentiation medium composition is highlighted in Bardy et al.³⁶ In this study, it was demonstrated that culturing iPS neuronal cells in traditional basal medium (i.e., Dulbecco's modified Eagle medium and Neurobasal) and serum enabled neuron survival and differentiation; however, it also resulted in impaired action potentials, synaptic communication, and basic neuronal function.

Moreover, Neurobasal medium was also independently demonstrated to induce neurotoxicity in primary hippocampal neuronal cultures.³⁷ Thus, while commonly used base media may provide nutrient environments that support neuronal viability, other facets of neuronal differentiation and function appear to be adversely affected, biasing results. Further complicating, the neuronal differentiation medium composition is the proprietary nature of some medium supplement (e.g., B27) formulations, making medium deconvolution and optimization difficult.

Timing of Neuronal Differentiation

Additional complications focus on the lack of consistency with respect to the timing of differentiation to yield mature neurons. As a result, the “differentiated neurons” may actually represent an immature stage of neuronal development^38,39 and because they typically do not express a full complement of adult neuronal proteins, they may lack functional activities associated with fully mature neurons. For example, synaptogenesis (i.e., formation of synapses) is generally observable at 6 weeks postdifferentiation in vitro,⁴⁰ yet there are many instances when iPS neurons are differentiated for shorter or undisclosed time periods, so the maturation state of the neurons is obscured.^31,32

These knowledge gaps reflect a significant need for further development of standardized neuron subtype-specific differentiation protocols, phenotypic characterization, and functional correlation with bona fide mature iPS neurons. Thus, once model systems have been optimized and validated, they will be an invaluable source for drug discovery (e.g., biomarkers, drug targets, pharmacodynamic endpoints, and screening assays).

Logistics of iPS Neuron Differentiation, Assay Development and Batch Mode Screening

Since iPS neuronal differentiation may take weeks in culture, a major challenge associated with iPS neurons will be the potential variation that might and most likely will occur during each neuronal differentiation induction. Thus, methodical assay development and optimization will be complex and more assay variation should be expected. As a result, this variation will almost assuredly require any high-throughput or HCS processes to occur in batch mode (i.e., with multiple “batches” of iPS neurons). Batch-mode screening is not uncommon for HCS assays as screening campaigns often take weeks to complete, depending on throughput and compound library size.

Complicating the issue, however, is that iPS neurons cannot be easily scaled up like cancer cell lines as the neurons (and their accompanying neurites) fail to thrive if passaged after neuron morphological differentiation. Thus, NPCs usually are seeded directly into the microtiter plates for screening and then induced (through medium changes) along the differentiation pathway. However, there is some evidence that NPCs can be differentiated en masse for up to 7 days before seeding in microtiter plates.³⁸ Nonetheless, the remaining time during differentiation will include multiple medium changes and edge effects will be a constant concern.

Lessons Learned from Neurological Drug Failures

Similar to drug discovery efforts in other disease areas, our failure to identify and develop efficacious neurological disease-focused therapeutics reflects an incomplete understanding of disease etiology and pathogenesis. However, it also reflects a disconnect between when drugs are patient tested versus when the drug would be temporally efficacious (i.e., “right drug-right time³¹”). It also suggests that the assay systems we use to identify and validate molecular targets as well as to screen for drugs or compounds may not be physiologically relevant since they do not represent a physiologically relevant disease state for therapeutic discovery. Moreover, we now understand that our assay systems need to not only reflect relevant disease stages or progression but also measure target engagement, especially if efficacious therapeutic intervention is needed before disease presentation.³¹

We therefore posit to build upon this “right time-right drug”³¹ reasoning to “right cell type-right assay format,” which will lay the foundational rigor and reproducibility needed to move potential drug leads and candidates through the developmental pipeline (Fig. 2). We understand that neurological diseases are highly complex and are influenced by genetics, brain injury, and inflammation and that therapeutic options may need to be combined with life style changes.³¹ However, if we build assay systems that better recapitulate the neurological disease pathogenesis, which can be used to demonstrate target engagement, we should be able to accelerate the identification of useful chemical probes and drugs. iPSC-derived NPCs and neurons, which can be derived from patients, may be able to help bridge the translation gap between preclinical and clinical drug development by offering more predictive cell-based model systems for screening.³¹

Fig. 2.

Basic neuronal screening processes. iPS NPCs and neurons are foundational to the basic neuronal screening process and lead to the first decision point–assay platform or readout. To enhance the chance of identifying meaningful chemical probes or drugs, assays should be disease and disease stage specific. 2D, two dimensional; three dimensional; iPS, inducible pluripotent stem. Color images are available online.

iPSC-derived NPCs and neurons enable the development of more physiologically relevant and complex 2D and 3D assay systems. These high physiocontextual⁴¹ assay systems, which yield rich data sets, may necessitate an evolution in screening strategies from larger to smaller compound library-based screens to accommodate the fundamental complexity of the brain (as well as data analysis and interpretation). Nonetheless, iPS NPCs and neurons are a valuable resource, but they are essentially “undiscovered country” with refinements in methodologies occurring almost simultaneously with implementation limiting the realization of their full potential. Thus, it is critical that investigators using iPSC-derived NPCs and neurons carefully codify their model development, optimization, and standardization processes to ensure robust, reproducible, and transferable results.

Abbreviations Used

2D

two dimensional

3D

three dimensional

AD

Alzheimer's disease

Aβ

amyloid β

HCS

high-content screening

iPS

inducible pluripotent stem

iPSCs

inducible pluripotent stem cells

MAP2

microtubule-associated protein 2

NeuN

neuronal nuclear antigen

NPCs

neural progenitor cells

Disclosure Statement

No competing financial interests exist.

Funding Information

The authors greatly appreciate the generous financial support of the Cure Alzheimer's Fund (E.R.S., G.S.B., and J.S.L.), the Owens Family Foundation (G.S.B. and J.S.L.), National Institutes of Health (AG051085 to G.S.B., AG063400 to E.R.S., and CA044579 and OD021723 to J.S.L.), Alzheimer's Association (Grants 4079 and ZEN-16-363266 to G.S.B.), The Virginia Chapter of the Ladies Auxiliary of the Fraternal Order of Eagles (G.S.B.), The University of Virginia President's Fund for Excellence (G.S.B.), and the Fiske Drug Discovery Fund (J.S.L. and E.R.S.).