Unraveling Mechanisms of Patient-Specific NRXN1 Mutations in Neuropsychiatric Diseases Using Human Induced Pluripotent Stem Cells

Heterozygous deletions in neurexin 1 (NRXN1) show statistically robust associations with diverse neuropsychiatric and developmental conditions, including schizophrenia, mood disorders, autism, and intellectual disability [1 –3]. However, the molecular and cellular mechanisms linking patient NRXN1 mutations and disease phenotypes remain obscure. Human induced pluripotent stem (hiPS) cells provide a means to study and experimentally manipulate patient neural cells [4]. However, despite their potential, few studies to date have used hiPS cell approaches to clarify the mechanisms by which patient mutations influence neural function.

In Nature Genetics, Flaherty and colleagues utilized hiPS cells derived from patients to explore the molecular and cellular mechanisms by which rare heterozygous mutations in NRXN1 (NRXN1^+/−) might increase risk for psychiatric disorders. Crucially, this strategy enables the investigation of patient mutations in a human- and disease-relevant context. Given that NRXN1 undergoes extensive alternative splicing, the authors surveyed the repertoire of a subset of NRXN1 isoforms (the NRXN1α family) in hiPS cell-derived neurons and human postmortem brain tissue by using short- and long-read sequencing.

To validate their hiPS cell neuron system, the authors first demonstrated that the NRXN1 isoform profile of hiPS cell-derived neurons largely recapitulates that seen in human postmortem brain tissue. They generated hiPS cell lines from eight different individuals: four patients with psychosis (harboring large-scale [>100 kb] heterozygous deletions in the NRXN1 locus; NRXN1^+/−), as well as one related and three unrelated healthy controls. Two of the patient-derived hiPS cell lines harbored 5′-NRXN1^+/− mutations and two carried 3′-NRXN1^+/− mutations. Strikingly, hiPS cell-derived neurons from NRXN1^+/− patients with psychosis had a different NRXN1α isoform repertoire compared with controls: patient-derived hiPS cell-derived neurons demonstrated a reduction in half of wild-type NRXN1α isoforms, and also expressed many previously unreported isoforms from the mutant allele.

Flaherty and colleagues next sought to characterize the molecular features of patient hiPS cell-derived neurons in more detail. Examination of single-cell RNA-seq data and electrophysiological measures indicated that neurons derived from NRXN1^+/− patients were immature compared with those from controls. Thus, NRXN1^+/− patient neurons had reduced expression of genes associated with nervous system development, axon guidance, and regional patterning. At a functional level, NRXN1^+/− neurons showed reductions in the number of spontaneous action potentials, as well as decreased neurite number and neuronal length. These findings are consistent with observations that NRXN1 haploinsufficiency disrupts synaptic activity [5]. However, future studies are needed to disentangle the mechanistic relationships between neuronal immaturity and synaptic dysregulation. Moreover, it remains unclear whether the observed phenotypes are dependent on the differentiation strategy employed. Impressively, the authors demonstrated the consistency of the neuronal immaturity phenotype across three different approaches for producing hiPS cell-derived neurons [6,7]. These protocols included conventional growth factor-mediated forebrain neuronal differentiation, as well as transcription factor-mediated induction of glutamatergic fate, by NGN2 overexpression, or GABAergic identity, by ASCL1/DLX2 induction. However, it would be of significant interest to know whether the genotype effect remains as pronounced if NRXN1^+/− hiPS cell-derived neurons were cocultured with astrocytes, which promote neuronal maturation [8,9].

What is the functional relevance of the observed isoform dysregulation? A standard technique for analyzing the biological relevance of mutant gene products is to attempt to “correct” the disease phenotype in patient cells, by normalizing the molecular profile, while conversely, investigating whether the introduction of the disease-related molecular change induces the observed disease phenotype in wild-type cells. Impressively, the authors successfully employed both approaches: neuronal activity was rescued in patient hiPS cell-derived neurons by overexpressing wild-type NRXN1α isoforms and, reciprocally, the activity of control neurons was reduced following expression of mutant NRXN1α isoforms. Therefore, these data support a model in which NRXN1^+/− mutations cause neuronal dysfunction by dominant-negative activity.

The remarkable finding that lentiviral overexpression of single wild-type NRXN1α isoform (four different wild-type isoforms were successfully used) is sufficient to rescue the phenotype in patient cells raises the tantalizing possibility of targeting alternative splicing for the treatment of patients harboring the NRXN1^+/− mutations. However, given that the key cellular phenotype influence by NRXN1^+/− mutations is neuronal maturity, any therapeutic manipulation would probably need to target the developing brain, meaning that any clinical implications are likely distant at present.

A key limitation of this study pertains to the “maturity” of the hiPS cell-derived neurons, particularly since the key phenotype observed in NRXN1^+/− neurons is immaturity observed at both gene expression and neuronal activity levels. Gene expression analyses indicate that hiPS cell-derived neurons are immature compared with those in the adult human brain and instead exhibit a transcriptome that most closely resembles fetal brain tissue [10]. Indeed, in this study, the NRXN1α isoform repertoire in control hiPS cell-derived neurons largely recapitulated that of fetal prefrontal cortex (PFC) tissue. Although caution should be exercised, given the small sample sizes involved, the authors compared NRXN1α isoforms across fetal and adult PFC tissue, they observed that as many as 55% of fetal isoforms could not be identified in adult PFC samples, suggestive of developmental regulation. Therefore, while useful for studying developmental phenotypes, the fetal properties of hiPS cell-derived neurons likely compromise their ability to model postnatal components of neuropsychiatric diseases. The development of new experimental strategies to program hiPS cell-derived neurons toward a more adult identity will improve disease models. For example, it might be of potential interest to examine the NRXN1α isoform repertoire and gene expression profiles of hiPS cell-derived neurons that have been “aged” by progerin overexpression [11].

Furthermore, although hiPS cell-derived neurons have been used to model early neural development, a critical issue with 2D systems is their inability to reconstruct higher order features of the human brain, including the more complex organization of neuronal structures. The advent of cerebral “organoid” methodologies aims to resolve this issue [12,13]. Indeed, cerebral organoids, when compared with neural cells obtained through conventional 2D culture systems, show more sophisticated patterning of neural cell types more closely reminiscent of the human brain [14,15]. However, like 2D culture systems, cerebral organoids also possess some limitations regarding their accuracy as models of human brain development [16]. Nonetheless, it will, therefore, be of interest to determine whether the altered NRXN1α isoform profiles and immature neuronal gene expression patterns observed in 2D cultures can also be recapitulated in 3D human brain organoid culture systems and, if so, whether these phenotypes “normalize” as cultures mature.

Although the findings here strongly suggest that mutant NRXN1α isoforms induce the changes in neuronal activity, and can be rescued by expression of “wild-type” isoforms, the underlying mechanisms remain to be determined. Furthermore, the technical approach used means that only mutations in the 3′ end of the NRXN1 gene could be examined. Therefore, it is not clear whether the findings observed here will generalize to all NRXN1 pathogenic mutations, particularly those occurring in 5′ regions. Studies employing both novel technical approaches and larger number of samples will likely be required to resolve these outstanding questions [17].

As well as producing insights into NRXN1-related conditions such as schizophrenia and autism, Flaherty and colleagues' work represents a landmark study illustrating the power of using hiPS cell-derived neurons to elucidate hitherto elusive molecular and cellular consequences of patient-specific mutations. Future insights into the isoform repertoire of genes of interest to psychiatry, many of which are likely to be woefully undercharacterized at present [18], could eventually lead to the identification of new drug targets, including targeting alternative splicing itself.