Research Article
Disease-modifying treatment options for individuals living with neurodevelopmental disorders are limited, and there are no cures. To address this, efforts to define disease mechanisms and novel therapeutic targets are necessary. In this review, we detail several techniques that are leading the way in these efforts.
Neurodevelopmental disorders (NDDs) affect 317 million children and young people globally. While NDDs are a diverse group of disorders that impact multiple organ systems, epilepsy, autism spectrum disorder (ASD), and/or intellectual disability (ID) are common outcomes. Unfortunately, there are few disease-modifying therapies and no cures for NDDs. Recently, however, gene editing, cellular/molecular assays, and stem cell models have emerged as promising techniques to deeply understand the root causes of NDD, categorize genetic variants, and more precisely model human disease processes. Additionally, several techniques, particularly gene editing, have already had clinical success. In this review, we detail several novel methods for functionally validating genetic variants, stem cell models of NDDs, and novel gene editing approaches. Understanding and applying these methods is crucial for uncovering novel disease mechanisms and therapeutic targets for NDDs, as well as for advancing technical approaches from the bench to the clinic.
Neurodevelopmental disorders (NDDs) affect 317 million children and young people globally1. These disorders can include cerebral palsy, fragile X syndrome, Angelman syndrome, and tuberous sclerosis complex, among many others. While NDDs are a diverse group of disorders that impact multiple organ systems, epilepsy, autism spectrum disorder (ASD), and/or intellectual disability (ID) are frequent outcomes2. Indeed, one study reported that children in Kenya with epilepsy and cognitive, vision, hearing, or motor impairments died at a rate of 3 to 4 times that of those in the non-NDD group1.
A paradigm shift for the study of NDDs occurred with the completion of the Human Genome Project in 20033. Further progress was made by defining the polygenic risk scores through GWAS, and the characterization of "neurodevelopmental disorders" by the DSM-5 in 20134. While there is meaningful progress in understanding the genetic causes of NDDs, few targeted therapies exist for specific NDDs, and there are no disease-modifying therapies for the ASD and/or ID associated with them.
The gene editing, electrophysiological, and cellular/molecular assays and methods described in this review are at the forefront of NDD research. While we have not exhaustively covered all assays, these methods represent cutting-edge approaches to identifying disease-modifying therapies. One example, Deubiquitylating enzymes (DUBs), is a group of enzymes that remove ubiquitin from proteins. Because mutations within DUBs are linked to NDDs, targeting their modification could be a potential therapy for NDDs5,6. Further, circuit-level pathways, such as those involving the GABAA receptor subunit, and other channelopathies are being interrogated to define disease mechanisms associated with epilepsy, ASD, and ID7. Another salient method, CRISPR/Cas genome editing (Figure 1), is a powerful tool used to directly alter the genome to create new experimental models and, recently, treat rare genetic disorders8,9,10. The CRISPR toolkit continues to expand, and four approaches, including CRISPRi (CRISPR Interference/Inhibition), CRISPRa (CRISPR Activation), prime editing, and base editing, have emerged as impactful research and clinical innovations. Lastly, this article details advances in the use of induced pluripotent Stem cells (iPSCs). iPSCs, and cellular derivations thereof, have allowed investigators to more accurately model many NDDs using patient-derived cells9,11,12. Importantly, the development of organoids from iPSCs has enabled the study of the developmental process that underly many NDDs in a way much more relevant to human disease than rodent models13,14.
This brief review details and expands upon the techniques introduced above. Additionally, several recent studies leveraging these techniques to gain greater insights into NDDs, gaps in the current literature, and future directions are discussed.
Induced pluripotent stem cells and associated experimental paradigms
Induced pluripotent stem cells have emerged as a powerful system in which to model NDDs. Indeed, many developmental processes occur in iPSC models that are similar to those observed during human brain development and that do not occur during rodent brain development15. Therefore, investigators have coupled stem cell models, including 2D cultures, organoids, and assembloids, to other experimental methods to better understand disease mechanisms in NDDs11,13,14,16.
Recently, assembloids were used to model a key feature of Down syndrome (DS)- impaired neurogenesis during fetal development16. Using both DS hiPSCs and isogenic euploid hiPSCs, neurogenesis was induced, followed by differentiation into neurons. After 85 days, immunocytochemistry and imaging were used to observe the cell cycle. They observed that DS neural progenitor cells did not exit the cell cycle and were unable to differentiate into mature neurons, and also found that cell cycle defects change the neurogenic phase, causing impaired neurogenesis in DS16.
While many stem cell models have focused on neurons, other cell types, including glial cells, play an important role in NDD pathology17. A recent report focused on the role of gliogenesis and astrocytes in fragile X syndrome (FXS). Using hiPSCs, astrocyte progenitor cells were cultured and differentiated into forebrain-specific astrocytes that lacked fragile X messenger ribonuclear protein11. They observed that these astrocytes had dysregulated glycolytic and mitochondrial metabolism. Prior work by the same team also demonstrated that proper mitochondria function is necessary to control action potential firing in neurons derived from patients with FXS11,18. While more work is required to understand the mechanism by which astrocyte metabolism and neuron firing are altered in FXS, these studies demonstrate that astrocytes can both drive disease pathology and be a therapeutic target.
Genome editing
Francisco Mojica laid the groundwork for genome editing with the discovery of CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats), an adaptive immune system for prokaryotes, in the early 1990's19. Building upon this, Jennifer Doudna and Emmanuelle Charpentier pioneered a method called CRISPR/Cas9 that leverages CRISPR coupled to Cas9 endonuclease to selectively isolate and change genes20. Not long after the publication of the initial CRISPR/Cas9 gene editing studies, investigators began manipulating the CRISPR/Cas system to reveal novel genomic information or edit genes in a unique way. One novel method is CRISPRi- a way to silence a gene of interest21,22. By using a catalytically inactive Cas9 nuclease in conjunction with cr- and tracr-RNAs, RNA polymerase's access to the promoter is obstructed, turning off gene expression for a specific gene or set of genes22,23. CRISPRi screens have been used to identify several risk genes for NDDs as well as identify the functional changes associated with loss-of-function in those genes. For example, a recent CRISPRi screen in iPSC-derived assembloids assayed 425 NDD genes for their impact on interneuron development and revealed that loss of several candidate genes impacted cytoskeletal and endoplasmic reticulum structure24. Further, a CRISPRi screen revealed that loss of ADNP, an ASD risk gene, resulted in changes in microglial pruning of synapses25. Thus, CRISPRi screens have revealed several convergent and divergent disease mechanisms in NDD.
Complementary to CRISPRi, CRISPRa is a tool used for increasing gene expression. This technique couples a catalytically inactive Cas9 to transcriptional activators and is targeted by a guide RNA. CRISPRa has also been leveraged to understand how activation of ASD and NDD risk genes can alter disease phenotypes. While there are few studies using CRISPRa to rescue NDD phenotypes, CRISPRa may have great therapeutic potential. For example, a recent study demonstrated that activation of ITGB3 (haploinsufficiency of which is associated with ASD) rescued network function and ASD phenotypes in rodents23. Indeed, many NDD gene variants cause disease due to haploinsufficiency, raising the possibility that CRISPRa could be used to increase expression of the unaffected allele in a disease-modifying manner.
Base editing and prime editing are precise methods within the CRISPR toolbox to edit specific nucleotides within the genome. Both cleave a single strand of DNA rather than a double-strand break (DSB) like CRISPR/Cas910,26. While base editing is more specific than standard CRISPR/Cas9 editing, it is less adaptable and therefore less translationally relevant than prime editing. The efficiency of base editing lies in its ability to correct point mutations by substituting cytosine for thymine or adenine for guanine and vice versa10. Because of this, base editing is restricted in the types of mutations it can make. Thus, although useful, only transition mutations can be executed, not insertions or deletions10. However, prime editing is a highly adaptable form of genome editing with many possible applications. Prime editing leverages a prime editing guide RNA (pegRNA) and a Cas9 nickase coupled to reverse transcriptase that can facilitate insertions, deletions, or changes in single nucleotides10. Unfortunately, because it is more technically complex and can carry out larger gene alterations, prime editing can be less efficient than base editing. Therefore, both can be used as complementary tools and are useful in genomic editing for neurodevelopmental disorders. Prime editing has already been leveraged to treat a patient with a rare genetic disorder27 and has shown promise in rescuing disease phenotypes in a rodent model of alternating hemiplegia of childhood28.
Cellular and molecular assays
NDDs are caused by genetic variants that impact cell function in different ways29,30,31. A major challenge facing those who study NDDs is how to determine whether or not a specific patient variant causes a deleterious functional change to the cell. Thus, there is a great need for assays that can rapidly and accurately determine whether or not a specific genetic variant is disease-causing or benign. A promising new assay is the ubiquitin cleavage assay32. Deubiquitylating enzymes (DUB) are a group of enzymes that remove ubiquitin from proteins, preventing degradation. DUBs are both potential treatments and causes of NDDs. So far, mutations in 12 DUBs have been identified as causally linked to NDDs, including Hao-Fountain syndrome, pseudo-TORCH syndrome, microcephaly-capillary malformation syndrome, and others5. To determine whether or not specific patient variants cause disease, in vitro ubiquitin cleavage assays have become the gold standard to measure DUB enzymatic activity to functionally validate mutations. This method can be used to assess gene variants' effect on DUBs. A recent study focused on a DUB that is mutated in X-linked intellectual disability 105, DUB Ubiquitin Specific Protease 27 (USP27X). The investigators successfully demonstrated a method for the expression and purification of recombinant USP27X and an in vitro ubiquitin chain cleavage assay. This approach allows for the ascertainment of enzymatic activity that regulates protein turnover in neurons and, therefore, may be important in understanding mechanisms underlying altered synaptic pruning, neurodevelopmental critical periods, and protein function32. Intriguingly, there is a paucity of published papers on DUBs and their role in disease, and thus, the full extent of applicability of this assay remains untested and open to further inquiry.
Because many NDDs are also classified as rare disorders, clinical genetic ascertainment of variants frequently identifies gene changes as 'variants of unknown significance' (VUS) due to a lack of prior published clinical or functional studies on those variants7,33,34. Thus, assays to rapidly determine whether a VUS is pathogenic are crucial to ascertaining the genetic causes of NDDs. One example is variants in GABAA receptors. Genetic variants in the GABAA receptor are a leading cause of genetic epilepsies and are associated with many seizure types and developmental epileptic encephalopathies7,35. A recent study investigated GABAA receptor subunits and attempted to identify whether or not the variants caused functional changes within the cell using a multi-modal approach. In silico tests were used to identify VUS and predict pathogenic γ2 GABAA subunit variations. The molecular consequences of the variants were then defined and included structural, physicochemical, and biophysical markers7. Each of these domains was defined using literature analysis, in silico modeling, and patch-clamp electrophysiology. Once a reference region was selected, pyramidal neuronal modeling was used to test the effects of GABAA receptor-mediated inhibition. They observed that two ß3 subunit mutations (ß3N110D and ß3T288N) and two γ2 subunit mutations impaired GABAergic inhibition7. This protocol demonstrates that a multi-modal approach is often useful in determining the pathogenicity of VUS. While not high-throughput, this approach may also be useful for other NDD-associated VUS that lack a specific test to functionally validate variants (e.g., DUBs)- particularly if they are associated with electrophysiological changes.
While not a novel approach, patch-clamp electrophysiology remains the gold standard for understanding the electrical properties of neurons, and thus how genetic changes impact the basic properties of neurons as well. Of relevance to NDDs are channelopathies36. While there are many types of channelopathies, variants in SCN1A have been well studied and are the primary cause of Dravet syndrome- a particularly devastating developmental epileptic encephalopathy36. Functional validation of SCN1A variants has been achieved using a combinatorial approach of automated patch-clamp and in silico simulations. Leveraging this approach, investigators were able to resolve several VUS and make genotype-phenotype correlations based on functional changes to neurons37. In a separate study using a similar approach, iPSC models were used with automated patch-clamp electrophysiology and in silico models to classify gene variants in T-Type calcium channels38. Focusing on CACNA1G, a gene encoding Cav3.1 in various types of neurons, CACNA1G was mutated to make 18 variants of Cav3.1 that were then transiently transfected into neurons. Following this, automated patch-clamp recordings were taken, followed by manual patch-clamp and computational approaches. They observed that automated patch-clamp and manual patch-clamp yielded similar results and that computational approaches were able to ascertain the functional changes induced by specific genetic variants38. While computational models provide a relatively rapid way to predict the functional consequences of gene variants, electrophysiological approaches remain time-consuming and require highly specialized skills that limit the utility of those approaches when attempting to validate large numbers of variants.
Multi-electrode arrays (MEAs) are used to measure electrical activity around small groups or across large networks of cells. This method differs from patch-clamp electrophysiology in that it focuses on multicellular models, whereas patch-clamp is centered on a sole channel or cell39,40. Because of this, MEAs can give insight into how different brain regions interact with each other and can be used to study disease on a network scale. Many studies have used MEAs to investigate experimental models of epilepsy and ASD14,39. A key feature of MEAs is that they are able to determine the ways in which different types of neuronal cells interact (e.g., glutamatergic neurons, GABAergic interneurons, progenitors, and glial cells)14. This makes MEAs an excellent approach for studying brain assembloids. One study, performed by Pan et al., used this method to study neuronal network development. The authors used human pluripotent stem cells (hPSC) from foreskin fibroblasts14 that were first virally labeled and then cultured to generate assembloids. Then, the assembloids were adhered to MEA plates and recorded14. Importantly, because the assembloids could be stabilized, recorded, and subjected to pharmacological experiments for approximately 50 days, this paradigm could be used for analyzing therapeutic drugs to treat epilepsy associated with NDDs. Indeed, this approach could be used to assay any NDDs in which changes in the network properties of neurons are known or suspected14.
NDDs can be devastating disorders and have few targeted treatments and no cures. Indeed, the majority of therapies are aimed at managing disease symptoms (e.g., anti-seizure medications for epilepsy). However, recent advances in technology have led to novel gene editing techniques, assays, and model systems that allow scientists to interrogate the root causes of NDDs (Figure 1). This review has detailed some of the approaches at the leading edge of efforts to find novel treatments for NDDs and related pathologies. Gene editing techniques will allow investigators to target mutations directly to better understand disease mechanisms and potentially treat patients with precision gene therapy. Novel assays allow for the rapid functional classification of genetic variants, and stem cell models will enable scientists to understand the early developmental mechanisms that drive NDDs as well as test novel therapeutics.The techniques detailed in this review form the foundation on which future studies can be performed with the aim of gaining a deeper understanding of NDD mechanisms, as well as leading to novel clinical approaches to treat them. As the field continues to evolve, future studies should integrate techniques such as prenatal testing, fetal therapies, high-content imaging, and pharmacogenomics with those detailed in this review.
Figure 1: Summary of key strengths and weaknesses of technical approaches. Methods detailed in the article and featured in the companion methods collections are summarized. A major strength and weakness are detailed for each technique. A key reference for each technique is provided under the method column. Please click here to view a larger version of this figure.
The authors have no conflicts of interest to declare.
Funding was provided by NIH NINDS RO1NS131223 to PHI. Figure 1 was prepared using BioRender.
