Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Reduced CYFIP1 in Human Neural Progenitors Results in Dysregulation of Schizophrenia and Epilepsy Gene Networks

  • Rebecca A. Nebel,

    Affiliation Department of Genetics, Albert Einstein College of Medicine, Bronx, New York, United States of America

  • Dejian Zhao,

    Affiliation Department of Neurology, Albert Einstein College of Medicine, Bronx, New York, United States of America

  • Erika Pedrosa,

    Affiliation Department of Psychiatry and Behavioral Sciences, Albert Einstein College of Medicine, Bronx, New York, United States of America

  • Jill Kirschen,

    Affiliation Department of Medicine, Albert Einstein College of Medicine, Bronx, New York, United States of America

  • Herbert M. Lachman,

    Affiliations Department of Genetics, Albert Einstein College of Medicine, Bronx, New York, United States of America, Department of Psychiatry and Behavioral Sciences, Albert Einstein College of Medicine, Bronx, New York, United States of America, Department of Medicine, Albert Einstein College of Medicine, Bronx, New York, United States of America, Department of Neuroscience, Albert Einstein College of Medicine, Bronx, New York, United States of America

  • Deyou Zheng,

    Affiliations Department of Genetics, Albert Einstein College of Medicine, Bronx, New York, United States of America, Department of Neurology, Albert Einstein College of Medicine, Bronx, New York, United States of America, Department of Neuroscience, Albert Einstein College of Medicine, Bronx, New York, United States of America

  • Brett S. Abrahams

    brett.abrahams@einstein.yu.edu

    Affiliations Department of Genetics, Albert Einstein College of Medicine, Bronx, New York, United States of America, Department of Neuroscience, Albert Einstein College of Medicine, Bronx, New York, United States of America

Abstract

Deletions encompassing the BP1-2 region at 15q11.2 increase schizophrenia and epilepsy risk, but only some carriers have either disorder. To investigate the role of CYFIP1, a gene within the region, we performed knockdown experiments in human neural progenitors derived from donors with 2 copies of each gene at the BP1-2 locus. RNA-seq and cellular assays determined that knockdown of CYFIP1 compromised cytoskeletal remodeling. FMRP targets and postsynaptic density genes, each implicated in schizophrenia, were significantly overrepresented among differentially expressed genes (DEGs). Schizophrenia and/or epilepsy genes, but not those associated with randomly selected disorders, were likewise significantly overrepresented. Mirroring the variable expressivity seen in deletion carriers, marked between-line differences were observed for dysregulation of disease genes. Finally, a subset of DEGs showed a striking similarity to known epilepsy genes and represents novel disease candidates. Results support a role for CYFIP1 in disease and demonstrate that disease-related biological signatures are apparent prior to neuronal differentiation.

Introduction

The identification of novel disease loci through genome-wide characterization of copy number variations (CNVs) has been important in many regards, including the discovery of genotype-phenotype relationships and genetic counseling. Personalized treatment, however, requires a clear understanding of underlying mechanisms, including which genes within disease-associated loci contribute to risk and how effects are mediated across development. Given that penetrance is often incomplete and expressivity is highly variable with CNVs, the ability to model this complexity is likewise critically important.

In this context, the breakpoint 1 to breakpoint 2 deletion region at 15q11.2 (BP1-2), associated with increased risk for schizophrenia and epilepsy, is exemplary [13]. The deletion spans ~ 500 kb and encompasses four genes: tubulin, gamma complex associated protein 5 (TUBGCP5), cytoplasmic FMR1 interacting protein 1 (CYFIP1), non imprinted in Prader-Willi/Angelman syndrome 2 (NIPA2), and non imprinted in Prader-Willi/Angelman syndrome 1 (NIPA1). The deletion is common in the general population, occurring at a frequency of ~ 1/500, and is often transmitted to affected children by clinically unaffected parents [4]. While a growing body of work suggests that altered CYFIP1 expression impacts neuronal morphology and function [511], relatively little has been done to evaluate the possible impact of reduced expression at earlier developmental time points, particularly in a human model system. One study using induced pluripotent stem cells (iPSCs) from BP1-2 deletion carriers determined that reduced CYFIP1 dosage resulted in cell polarity defects in neural rosettes [12]. However, genome-wide transcriptional profiling was not performed in this study, and inter-individual differences, a defining feature of BP1-2 deletions, were not evaluated.

To investigate an early developmental role for CYFIP1 in disease and explore inter-individual variation in a human system, we developed a knockdown model using lentiviral shRNAs in neural progenitor cells (NPCs) derived from human iPSCs with two copies of each gene at the BP1-2 interval (BP1-2 copy number neutral). NPCs are proliferating cells that resemble radial glia and have the capacity to differentiate into neurons, astrocytes, and oligodendrocytes. In support of their use, work comparing NPCs derived from individuals with idiopathic schizophrenia and controls found abnormalities in cytoskeletal remodeling and the oxidative stress response in patient-derived samples [13]. Similarly, cell proliferation and expression of ion channels were found to be abnormal in NPCs derived from individuals with bipolar disorder [14].

We demonstrate here that CYFIP1 plays a critical role in cytoskeletal remodeling in human NPCs, as we observe dysregulation of cytoskeleton-related genes, reduced levels of WAVE1/2, and diminished F-actin following knockdown. CYFIP1 knockdown cells tended to be bigger and had significantly larger nuclei as well. Moreover, reduced CYFIP1 also disrupted normal expression of fragile X mental retardation protein (FMRP) targets and postsynaptic density (PSD) genes, each previously implicated in schizophrenia [1517]. Providing direct support for involvement of CYFIP1 in BP1-2 mediated neurodevelopmental disease (NDD), we observed marked dysregulation of schizophrenia and epilepsy risk genes, disorders associated with the deletion. Mirroring the clinical variability associated with BP1-2 deletions, the presence and magnitude of these disease effects varied between donor lines. Effects were also highly specific in that enrichment was not seen for genes implicated in complex diseases or traits unrelated to the BP1-2 deletion. We also identified novel candidate disease genes among a subset of altered mRNAs that are unexpectedly similar to known epilepsy genes. The in vitro model described here will enable identification of compounds that normalize BP1-2 disease-related endpoints and as such represents an important next step towards personalized treatment of deletion carriers.

Results

Generation of human neural progenitor cells with reduced CYFIP1 expression

We sought to model, in a human system, the impact of reduced CYFIP1 expression with regard to BP1-2-related NDDs. To validate the model, we first confirmed in lymphoblastoid cell lines from BP1-2 deletion carriers that CYFIP1 mRNA levels were indeed significantly reduced relative to cell lines from BP1-2 copy number neutral individuals (39% reduction; p = 1.4 x 10−3; Fig 1A). We next differentiated NPCs from previously characterized human iPSC lines [1823], and transduced cells with a lentivirus carrying either a non-silencing control shRNA (NS) or an shRNA targeting CYFIP1 (CYFIP1KD). To study CYFIP1-specific contributions, we limited these and subsequent experiments to lines prepared from BP1-2 copy number neutral male adults with no history of NDDs. We determined that CYFIP1 levels were reduced significantly in response to knockdown (81% reduction; p = 2.2x 10−3; Fig 1B). The extent of CYFIP1 reduction in our model system falls within the range of what is observed in neural cells derived from BP1-2 deletion carriers [12,24]. mRNA levels of the closely related CYFIP2 gene were indistinguishable between NS and CYFIP1KD NPCs. Paralleling CYFIP1 mRNA levels, quantification of CYFIP1 protein levels showed a significant reduction in CYFIP1KD NPCs relative to NS controls (70% reduction; p = 4.4 x 10−14; Fig 1C and 1D).

thumbnail
Fig 1. Similar to lymphoblastoid cell lines from individuals with BP1-2 deletions, CYFIP1 levels are reduced following shRNA knockdown in BP1-2 copy number neutral human neural progenitor cells (NPCs).

(A) qPCR shows that relative to BP1-2 neutral controls (n = 7), CYFIP1 mRNA levels are reduced in lymphoblastoid cell lines from deletion carriers (n = 7). (B) qPCR on BP1-2 copy number neutral NPCs shows that relative to non-silencing shRNA transduced cells (NS, black), transduction with an shRNA targeted against CYFIP1 (CYFIP1KD, grey) results in a reduction of CYFIP1 but not CYFIP2 mRNA (n = 6 per group). (C, D) Relative to control NPCs, western blot shows that CYFIP1KD results in a reduction of CYFIP1 protein (n = 15 per group; 3 or more samples from each of 3 separate transductions). One-tailed Student’s t-test, **p≤ 0.01; ****p≤0.0001.

https://doi.org/10.1371/journal.pone.0148039.g001

Transcriptional profiling in CYFIP1KD NPCs identifies hundreds of dysregulated genes

To gain insight into the molecular effects of reduced CYFIP1 expression, we performed RNA-seq on NS and CYFIP1KD NPCs prepared from three different iPSC lines (C2, C4, and C5). Because each line was generated from a different donor, we reasoned that this experimental design would allow us to evaluate line-dependent and line-independent effects. For each line, RNA from six independent transductions (three CYFIP1KD and three NS) was sequenced, and an average of 23.2 million properly mapped pairs was obtained (S1 Table). For each donor line, the average of the three CYFIP1KD samples was compared to the average of the three isogenic NS samples. RNA-seq and qPCR validation showed that CYFIP1 was similarly reduced in NPCs from all three lines (Fig 2A). Confirming that the cells we transduced were indeed in a progenitor state, all lines showed high levels of neural progenitor markers and much lower expression of neuronal and glial markers (Fig 2B).

thumbnail
Fig 2. CYFIP1 knockdown results in dysregulation of hundreds of genes in human neural progenitor cells (NPCs).

(A) RNA-seq and qPCR show that compared to non-silencing control transduced NPC lines (NS, black), cells transduced with an shRNA targeting CYFIP1 (CYFIP1KD, grey and hatched) show a significant reduction in CYFIP1 mRNA in lines from three different BP1-2 copy number neutral individuals (C2, red; C4, green; C5, blue, n = 3 per group throughout). P-values for RNA-seq analysis- DESeq2; qPCR- one-tailed Student’s t-test; *p≤ 0.05; ****p≤0.0001. (B) RNA-seq shows that mRNAs for NPC markers are expressed 10 to 100-fold higher than mRNAs for neuronal and glial markers. TPM values are averaged across NS and CYFIP1KD samples. (C, D) Only a handful of genes are similarly dysregulated in response to CYFIP1 knockdown across all three NPC lines evaluated. (E) Unsupervised hierarchical clustering of genes showing nominal differential expression (p<0.05) points to marked between-line differences. (F) RNA-seq and qPCR on mRNA from control and CYFIP1KD NPCs (CYFIP1KD RNA-seq, black; CYFIP1KD qPCR, grey) show reduced levels of genes downregulated in all three cell lines. Abbreviations: CCND1- cyclin D1, GFRA1- GDNF family receptor alpha-1, GPR133- G-protein coupled receptor 133, L1CAM- L1 cell adhesion molecule, PGPEP1- pyroglutamyl-peptidase I, TIPARP- TCDD-inducible poly(ADP-ribose) polymerase.

https://doi.org/10.1371/journal.pone.0148039.g002

We next sought to identify genes sensitive to reduced levels of CYFIP1 (differentially expressed genes, DEGs). Using a stringent cutoff (transcripts per million (TPM)>1 in NS samples and FDR-adjusted p<0.05), 2212 unique genes were found to be dysregulated in NPCs from one or more lines, with 637 DEGs in C2, 183 DEGs in C4, and 1553 DEGs in C5 (Fig 2C and 2D, S2 and S3 Tables). Unsupervised hierarchical clustering of genes showing nominal differential expression (p<0.05) in all three lines (n = 153) revealed that samples clustered first by line, and then by knockdown status (Fig 2E). Using the more stringent cutoff, only seven DEGs (including CYFIP1) showed dysregulation in all three lines. Excluding CYFIP1, evaluation of the six remaining genes altered between all three lines (CCND1, GFRA1, GPR133, L1CAM, PGPEP1, TIPARP) in line C2 by qPCR confirmed DEG status for each (Fig 2F).

Genes involved in M phase of cell cycle and cytoskeletal remodeling are overrepresented among dysregulated mRNAs

We next reasoned that while the majority of genes dysregulated in NPCs prepared from each of the three lines differed, the underlying signaling pathways altered in each might overlap. To test this hypothesis, we carried out a series of gene ontology (GO) analyses on the upregulated and downregulated DEGs from each line. Our analyses identified an enrichment of genes involved in M phase of the cell cycle in lines C4 and C5 (Table 1). Also consistent with dysregulation of shared processes was the overrepresentation of cytoskeletal-related GO terms among downregulated genes from each of the three lines (Fig 3A). C2 was enriched for cell motion and other actin based processes while C4 and C5 were enriched for cell adhesion, another cytoskeletal based function [25,26]. Together these results demonstrate that while the particular genes dysregulated in response to CYFIP1KD differed between NPCs from each line, similar processes were impacted in response to reduced expression. Furthermore, these studies are the first to show that reduced levels of CYFIP1 affect transcriptional control of cytoskeletal genes.

thumbnail
Table 1. Genes involved in M phase of cell cycle are overrepresented among transcripts upregulated in response to CYFIP1 knockdown in human neural progenitors.

https://doi.org/10.1371/journal.pone.0148039.t001

thumbnail
Fig 3. Dysregulation of cytoskeletal remodeling in response to CYFIP1 knockdown in human neural progenitor cells (NPCs).

(A) Genes involved in cytoskeletal remodeling are overrepresented among transcripts downregulated in response to CYFIP1 knockdown. (B, C) Western blot shows that relative to non-silencing shRNA transduced NPCs (NS, black), transduction with an shRNA targeted against CYFIP1 (CYFIP1KD, grey) results in a significant 54% reduction of WAVE1/2 protein levels. n = 10 per group. (D, E, F) Phalloidin staining shows that F-actin levels are significantly reduced (11% reduction) in CYFIP1KD NPCs relative to control cells. n = 12 wells per group. (G, H, I) Nuclear area, quantified by DAPI, is significantly increased by 9% in CYFIP1KD NPCs relative to control cells. n = 12 wells per group. (J, K, L) Cell area, captured by phalloidin staining, is nominally greater in CYFIP1KD NPCs relative to control cells (34% increase). n = 12 wells per group. Two-tailed Student’s t-test, *p<0.05 **p≤0.01 ***p≤0.001.

https://doi.org/10.1371/journal.pone.0148039.g003

Functional studies corroborate molecular findings for aberrant cytoskeletal remodeling in CYFIP1KD NPCs

We next aimed to validate a subset of these molecular findings functionally, focusing on cytoskeletal remodeling. Although CYFIP1 is well known to play a role in cytoskeletal remodeling in neurons [6,7,9,10,27], its role in NPCs has received less attention. To ensure that our findings here were line-independent, this work was carried out in NPCs prepared from iPSCs that were not used in our RNA-seq experiments (see Materials and Methods). We first measured WAVE1/2 protein levels in CYFIP1KD and NS NPCs. CYFIP1 and WAVE proteins are members of the Wave Regulatory Complex (WRC), a key mediator of cytoskeletal dynamics. Previous studies have shown that inactivation of individual components of the WRC leads to reduced expression of the other WRC members [9,28]. Moreover, neural rosettes prepared from BP1-2 deletion carrier iPSCs show diminished WAVE2 levels as a consequence of reduced CYFIP1 expression [12]. Consistent with what is seen in neural cells derived from BP1-2 deletion carriers, examination of WAVE1/2 protein levels in our system showed a significant reduction in CYFIP1KD NPCs relative to controls (54% reduction; p = 1.4 x 10−4; Fig 3B and 3C). Because the WRC is a critical regulator of actin polymerization, we next stained CYFIP1KD and NS NPCs with phalloidin to permit quantification of F-actin levels. These experiments showed that CYFIP1KD resulted in a significant reduction of F-actin levels (11% reduction; p = 5.0 x 10−3; Fig 3D–3F). In the course of these experiments we also determined that nuclear size was significantly increased in CYFIP1KD NPCs when compared to NS cells using a DAPI stain (9% increase; p = 3.4 x 10−2 Fig 3G–3I). CYFIP1KD NPCs also tended to be larger than NS NPCs although this apparent difference was not significant (34% increase; p = 0.08; Fig 3J–3L). These findings corroborate RNA-seq results and underscore a prominent role for CYFIP1 in cytoskeletal regulation in NPCs.

CYFIP1KD in NPCs results in transcriptional dysregulation of FMRP targets and postsynaptic density genes

To further explore how reduced levels of CYFIP1 may impact NPC function, we looked for overrepresentation of FMRP targets and PSD genes (S4 Table) among DEGs. CYFIP1 is known to physically interact with FMRP to repress translation of FMRP-associated mRNAs [5,6]. Alterations in CYFIP1 levels have also been shown to cause changes to dendritic morphology in differentiated neurons, where proteins of PSD genes localize [6,7,10]. Our analyses determined that FRMP targets were significantly enriched among DEGs across all lines (C2: p = 6.1x10-3, Fold Enrichment (FE) = 1.6; C4: p = 1.5x10-3, FE = 2.2; C5: p = 1.7x10-6, FE = 1.6; Table 2 and S3 Table), showing for the first time that CYFIP1 exerts not only a translational control over FMRP targets, but a transcriptional control as well. As with FMRP targets, enrichment analyses revealed a line-independent overrepresentation of PSD genes among DEGs (C2: p = 9.9x10-7, FE = 2.2; C4: p = 1.0x10-3, FE = 2.5; C5 p = 1.6x10-7, FE = 1.8; Table 2 and S3 Table). Additional GO analyses of FMRP and PSD-related DEGs showed that PSD genes dysregulated in response to CYFIP1KD were significantly enriched for genes involved in cytoskeleton-based processes (S5 Table), providing a satisfying explanation for this finding. Given that candidate mutations in both FMRP targets and PSD genes are present at a significantly greater frequency in individuals with schizophrenia compared to typically developing controls [1517], these results suggest that reduced levels of CYFIP1 might increase risk for schizophrenia at least in part through dysregulation of one or both of these functional networks. Results also support the idea that abnormalities in NPCs might in some cases predict neuronal dysfunction.

thumbnail
Table 2. Dysregulation of FMRP targets and post-synaptic density genes in response to CYFIP1 knockdown in human neural progenitor cells.

https://doi.org/10.1371/journal.pone.0148039.t002

CYFIP1KD results in dysregulation of genes implicated in schizophrenia and epilepsy, but not diseases or traits unrelated to BP1-2 deletions

BP1-2 deletions are well established to increase risk for both schizophrenia and epilepsy [13]. Clinical presentation, however, is highly variable and many carriers do not meet diagnostic criteria for any NDD [29,30]. Given evidence for dysregulation of disease-related pathways by CYFIP1KD, we sought to determine if reduced levels of CYFIP1 in NPCs might alter the expression of BP1-2 deletion-related disease gene networks. To test this hypothesis, we identified lists of genes associated with schizophrenia and epilepsy (S4 Table) and looked for overrepresentation among CYFIP1KD DEGs (Table 3 and S3 Table). Overrepresentation of schizophrenia risk genes was seen in C2 (p = 4.6 x10-2, FE = 1.5) and C5 (p = 4.9x10-4, FE = 1.5) but not C4 (p = 1.0, FE = 1.0). With regard to epilepsy, overrepresentation of CYFIP1KD DEGs was observed in two of three lines, although the identity of these lines differed from those in which the schizophrenia effects were observed. Overrepresentation here was seen in C4 (p = 3.2x10-8, FE = 10.7) and C5 (p = 6.0x10-3, FE = 2.0) but not C2 (p = 0.58, FE = 1.2). Effects were not associated with either the number of DEGs per line or efficiency of CYFIP1 knockdown. These data demonstrate that reduction of CYFIP1 levels can alter BP1-2 deletion-related disease gene networks, but in contrast to dysregulation of cytoskeleton genes, FMRP targets, and PSD genes, the presence and magnitude of these disease effects were variable, much like clinical outcomes in deletion carriers.

thumbnail
Table 3. Genes implicated in schizophrenia and epilepsy, but not conditions unrelated to BP1-2 deletions, are overrepresented among CYFIP1 knockdown dysregulated mRNAs.

https://doi.org/10.1371/journal.pone.0148039.t003

To evaluate the specificity of these effects, we assembled gene lists for diseases and traits with no known association to BP1-2 deletion status (S4 Table). As outlined in Table 3 and S6 Table, in contrast to findings for schizophrenia and epilepsy where strong effects were seen, nothing comparable was observed for: rheumatoid arthritis (pmin = 0.35, FE = 1.5 for C2), breast cancer (pmin = 0.13, FE = 1.6 for C5), type II diabetes (pmin = 0.03, FE = 1.8 for C5), Crohn's disease (pmin = 0.01, FE = 2.6 for C2), body mass index (pmin = 0.24, FE = 0.4 for C5), or blood pressure (pmin = 0.16, FE = 1.9 for C2). An effect for height was observed in one line (pmin = 1.5x10-3, FE = 1.7 for C5). Additionally, while case-only studies have suggested relationships between deletion status and both autism and intellectual disability, these results are confounded by ascertainment bias, and no clear effects have been seen in population based studies [4,3133]. Consistent with the absence of a relationship, we did not see an enrichment for genes implicated in autism (pmin = 0.06, FE = 1.4 for C5) or intellectual disability (pmin = 0.14, FE = 1.4 for C5). Finally, if considered together, a disproportionate number of nominally significant effects are from analyses for schizophrenia and epilepsy as opposed to diseases and traits with no evidence for an association with deletion status (4/6 vs. 3/27; p = 0.01, two-sided Fisher’s exact test).

Interrogation of schizophrenia and epilepsy genes elucidates mechanisms and identifies novel disease candidates

To gain a better understanding of what disease genes that are sensitive to reduced levels of CYFIP1 do functionally, we performed additional GO analyses. With regard to schizophrenia, we reasoned that such analyses might provide additional mechanistic insights beyond possible contributions from FRMP targets and PSD genes. However, this was not the case; analysis of the schizophrenia-associated genes included within our list of CYFIP1KD DEGs did not reveal a significant enrichment of any GO terms (S7 Table). Meanwhile, GO analysis of epilepsy-related CYFIP1KD DEGs showed a strong enrichment for genes involved in synaptic transmission (Table 4). A closer examination of these genes revealed that they primarily code for ion channels such as CACNA1H, KCNA1, and SCN1A, providing additional support for the idea that signs of neuronal dysfunction may be present at the NPC stage. GO analyses also found evidence for genes involved in mitochondrial electron transport (Table 4).

thumbnail
Table 4. Epilepsy genes dysregulated in response to CYFIP1 knockdown are involved in synaptic transmission.

https://doi.org/10.1371/journal.pone.0148039.t004

Given the marked overrepresentation of well-established epilepsy genes among CYFIP1KD DEGs, we reasoned that a subset of these might represent novel disease candidates. Employing the same epilepsy gene list used in our overrepresentation analysis as a training set, we used ToppGene [34] to identify candidates. The top 10 genes from each line (Table 5; complete output- S8 Table), corresponding to 28 unique genes in total, showed unexpected similarity to known disease genes (C2: p<3.2x10-5; C4: p<1.1x10-4; C5: p = 4.0x10-7). Two genes, CLIC4 and LOC102724434, appeared in the top 10 ranked genes for both C2 and C5. Validating the concept that these 28 epilepsy-like CYFIP1KD DEGs are good disease gene candidates, a post-analysis literature review determined that CACNA1C and MT-ATP6, although not included in our training set, are in fact known to cause epilepsy [35,36]. Likewise, 12/28 genes have been implicated in epilepsy or have a close family member established to increase risk including SCN2B, CALM1, CALM2, and APP [3746]. These 12 genes, and the additional 14 not previously linked to disease, represent new epilepsy candidates that can be evaluated in the near term through sequencing studies of patients and controls.

thumbnail
Table 5. A subset of genes dysregulated as a result of CYFIP1 knockdown show striking similarity to known epilepsy genes, and as such are strong candidate disease genes.

https://doi.org/10.1371/journal.pone.0148039.t005

Discussion

We present a novel human model for studying early developmental abnormalities associated with BP1-2 deletion mediated disease risk. Results from molecular and cellular analyses support an important role for CYFIP1 in cytoskeletal regulation in NPCs and provide evidence for transcriptional regulation of these processes. Although the specific genes dysregulated in each line differed, the underlying signaling pathways perturbed were largely shared across lines. Similarly, RNA-seq data show that CYFIP1KD resulted in line-independent overrepresentation of both FMRP-targets and PSD genes among DEGs.

CYFIP1KD also resulted in a significant dysregulation of BP1-2 disease gene networks, but not in an overrepresentation of genes associated with randomly selected disorders or traits. Like clinical presentation in deletion carriers, however, clear between-line differences were evident. A variety of explanations could account for the line—to—line differences we observed, but like BP1-2 mediated risk for NDD, a multifactorial explanation is most probable. For example, it could be that the genomes of the individuals that our iPSC lines were generated from harbor additional genetic risk factors that act in an additive fashion to alter gene expression of disease-associated genes. Consistent with such a mechanism, children who harbor a BP1-2 deletion and a second large CNV are more likely to show developmental delay than those who carry only the deletion [47]. Also relevant is that behavioral deficits seen in mice heterozygous for a mutation in FMR1 are highly dependent on genetic background [4850]. Alternatively, between-line differences may reflect the effects of environmental exposures unique to individual donors. It could also be the case that differences are the result of random genetic or epigenetic events that arose during reprogramming or subtle procedural differences during cell culture. This question will be clarified through characterization of additional clones generated from the same donors as the lines studied here and through study of additional clones from unrelated individuals.

Nevertheless, the overarching approach of utilizing materials from multiple individuals and focusing on differences as well as similarities is likely to be of great value. So far as we know, this is the first study to use iPSC derived cells from multiple healthy individuals to study NDDs. Previous studies that used an approach similar to ours looked at materials from only a single donor [23,51,52]. As our data reveal, results can vary between lines, and so caution must be exercised when interpreting results from studies that only use one line for their experiments. This is particularly true with regard towards translational efforts for risk factors like BP1-2 deletions where inter-individual variability is a key feature.

In terms of mechanisms, dysregulation of FMRP targets and PSD genes is intriguing given previous studies that suggest both pathways are related to schizophrenia risk [1517]. Because FMRP and PSD effects were observed in all lines, including line C4 where we did not observe an overrepresentation of schizophrenia candidate genes, alteration of these signaling pathways cannot fully explain the observed schizophrenia effect. Additional work will be required to identify additional contributory factors. Meanwhile, genes involved in regulation of M phase might be relevant to understanding dysregulation of epilepsy gene networks; such transcripts were overrepresented among DEGs from C4 and C5, the same two lines in which an overrepresentation of epilepsy genes was observed.

PSD and synaptic transmission effects were initially surprising given that these gene networks are typically associated with differentiated neurons. Subsequent analyses, however, highlighting a large number of cytoskeleton-related genes within the set of PSD genes dysregulated by CYFIP1KD provide some clarity here. Results suggest that disease-related biological signatures are evident long before neuronal differentiation and support the notion that early development insults may confer risk for disease. The study of iPSCs derived neurons has emerged as a popular means of modeling disease [5356], but less attention has been given to NPCs. Support for the use of NPCs has come from studies of schizophrenia, bipolar disorder, and Rett Syndrome [13,14,5759]. For example, Topol et al. discovered altered WNT signaling in NPCs derived from schizophrenia patients, a pathway previously known to be dysregulated in post-mortem brain of individuals with schizophrenia and thought to be a target of many antipsychotic drugs [60,61]. Our findings provide further support for the idea that NPCs can be useful in studying NDD risk and may in some cases circumvent the time and cost associated with study of differentiated neurons. However, it will be interesting to determine the effects of CYFIP1 knockdown on differentiated neurons and the differentiation process in future experiments.

Lastly, although dysregulation of CYFIP1 alone may not account for all aspects of BP1-2 associated risk, our data point to the importance of this gene. For example, results from our CYFIP1 knockdown model are consistent with what is seen in neural cells derived from BP1-2 deletion carriers. Additionally, in each of the three lines we evaluated, CYFIP1KD resulted in the dysregulation of one or more disease gene networks. In the face of these data, high-throughput drug screening in our model system could be useful in the identification of candidate compounds that could be evaluated in deletion carriers.

Materials and Methods

Ethics Statement

Subjects who provided us with materials to generate lymphoblastoid cell lines were enrolled through Einstein and provided written informed consent. For minors enrolled in this study, informed written consent was obtained via a parent. Subjects from which iPSCs were generated were recruited from two settings; Einstein and the National Institutes of Mental Health (NIMH), Child Psychiatry Branch. For the NIMH subjects, all participants provided written assent or consent. These studies, consent forms, and procedures for obtaining informed consent were approved by the Einstein or the NIMH Institutional Review Board.

Lymphoblastoid Cell Lines

Eight lymphoblastoid cell lines, four from BP1-2 neutral subjects and four from deletion carriers, were obtained from the Simons Simplex Collection [62]. An additional six lines, three from neutral subjects and three from deletion carriers, were generated at Einstein. Peripheral blood was collected from subjects and lymphoblastoid cells were immortalized by Epstein Barr virus transformation. See S9 Table regarding clinical information for subjects.

BP1-2 Copy Number Variant Genotyping

Before experimentation, a TaqMan assay Hs01476346_cn (Life Technologies) was used to determine copy number at the BP1-2 region in all lymphoblastoid and NPC lines used. Assays were performed in triplicate with TaqMan Gene Expression Master Mix (Life Technologies) following the standard TaqMan Copy Number Assay protocol. An RNase P probe (Cat# 4403326, Life Technologies) was used as a reference. Copy number variants were called using Copy Caller Software (v2.0, Life Technologies).

iPSC Lines and NPC Differentiation

iPSC lines were generated from fibroblast cells taken from skin biopsies of healthy adults with no reported NDDs. iPSC lines used in this study have been described previously [1823]. The lines used for RNA-seq in this study were: 553.C2 (C2), iPS5.C4 (C4), 690.C5 (C5). These lines were generated from a 31 year old Caucasian male, a 33 year old Caucasian male, and a 27 year old Caucasian male, respectively. Lines iPS2.C4 and iPS6.C4, used for functional studies, were generated from a 57 year old Caucasian male and a 46 year old Caucasian male, respectively. Reprogramming was carried out by nucleofection of non-integrating plasmids containing OCT4, SOX2, KLF4, L-MYC, LIN28, and a p53 shRNA vector (Cat# 27077, Addgene). All lines have normal karyotypes, express stem cell markers, and can be differentiated into all three germ layers. Differentiation into NPCs was performed was as previously described [1820,22,23]. All lines were confirmed by Taqman assay Hs01476346_cn (see above) to not carry a BP1-2 deletion or duplication.

Cell Culture and Lentiviral Transduction

All cultures were maintained in humidified incubators at 37°C and 5% CO2 in air. Lymphoblastoid cell lines were cultured in RPMI 1640 with 15% heat inactivated fetal bovine serum, 1% Glutamate, and 1% penicillin/streptomycin (pen/strep) (Life Technologies). NPCs were maintained on polyornithine/laminin coated plates and were fed every other day with NBF media (DMEM/F12 supplemented with 0.5% N-2, 0.5% B-27, and 1% pen/strep (Life Technologies)) and fresh 20ng/ml FGF2. FGF2 from Gemini was used for all RNA-seq studies. For functional studies FGF2 from Gemini or R&D Systems was used.

NPCs were transduced with GIPZ lentivirus carrying either an shRNA targeting CYFIP1 (Oligo ID- V2LHS_258463, sense sequence- GCAAAGATGAGATTATTAA, Open Biosystems) or a non-silencing control (Cat#- RHS 4346, sense sequence- CTCGCTTGGGCGAGAGTAA, Open Biosystems). Lentiviruses were made in HEK293T cells using 2nd generation packaging plasmids. Transductions were carried out when NPCs were 50% confluent. Briefly, 1.0–1.5x106 viral particles were added to NPCs in transduction media (NBF media, 1 μg/ml polybrene (Santa Cruz), and 20ng/ml FGF2). Twenty-four hours after transduction, media was replaced with fresh NBF and fresh FGF2. To select for transduced cells, media was changed again 48 hours later with fresh NBF/FGF2 but with the addition of 0.5 μg/ml puromycin (Santa Cruz). This selection media was replaced every other day.

RNA-seq

When NPCs were 100% confluent, total RNA was isolated using a miRNAeasy Kit (Qiagen) according to the manufacturer’s protocol. An additional DNaseI digestion step was performed to ensure that samples were not contaminated with genomic DNA (Qiagen). RNA quality was assessed via an Agilent bioanalyzer. One hundred bp paired-end RNA-seq was carried out on an Illumina 2000 HiSeq at the JP Sulzberger Columbia Genome Center at Columbia University. The 100 bp paired-end reads were processed as previously described [21]. Briefly, sequence quality for each sample was evaluated with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) (S1 Table). No evidence of adapter contamination was observed and average read quality scores > 30 for all samples. Reads were then aligned with TopHat (v 2.0.4) to the human genome (hg19) [63]. HTSeq (v 0.6.1) [64] was used to obtain read counts for each gene (Ensembl release 74) and DESeq2 utilized to determine differentially expressed genes [65]. Relative expression was measured in transcripts per million (TPM) [66]. Unless otherwise stated, only genes with a TPM>1 in non-silencing control samples and a FDR-adjusted p<0.05 were considered differentially expressed. RNA-seq data has been submitted to the Gene Expression Omnibus (GEO) repository [67] (accession number- GSE70935).

Quantitative real-time PCR (qPCR)

Total RNA was isolated from cells as above. RNA was reverse-transcribed to cDNA with the SuperScript III First Strand Synthesis System for RT-PCR (Life Technologies). qPCR reactions were performed in triplicate with Power SYBR Green Master Mix (Life Technologies). Genes of interested were normalized to the housekeeping gene, B2M, and relative changes were calculated using the Pfaffl method. Primers used for qPCR can be found in S10 Table.

Gene Ontology Enrichment Analyses

The Database for Annotation, Visualization, and Integrated Discovery (DAVID) (http://david.abcc.ncifcrf.gov/home.jsp) was used for gene ontology analyses [68,69]. Ensembl IDs were used as input. For each line, genes expressed in non-silencing control NPC samples with a TPM>1 were used for background (S2 Table). DEGs used for GO analyses are noted in S3 Table. GO terms with Benjamini p-values <0.05 were considered significantly enriched.

Western Blot

Cells were homogenized in RIPA lysis buffer containing a protease and phosphatase inhibitor cocktail (Thermo Scientific). Lysates were incubated on ice for 10 min, sonicated for 10 min, and centrifuged at 15,000 x g for 15min at 4°C. Protein concentration was determined using a Pierce BCA Protein Assay Kit (Thermo Scientific). An equal amount of protein for each sample was run on a 4–20% gradient gel (BioRad) and transferred onto a PVDF membrane. Membranes were incubated with primary antibodies overnight at 4°C, incubated with secondary antibodies for 1 hr at room temperature, and visualized with SuperSignal West Dura Chemiluminescent Substrate (Thermo Scientific). Primary antibody dilutions were as follows: rabbit polyclonal anti-CYFIP1 1:1000 (Cat#07–531, Millipore), rabbit polyclonal anti-WAVE1/2 1:1000 (Cat#sc-5557, Santa Cruz), mouse monoclonal anti-GAPDH 1:20000 (Cat#10R-G109a, Fitzgerald).

Immunocytochemistry and Microscopy

NPCs were grown on polyornithine/laminin coated coverslips. Cells were fixed in 4% formaldehyde for 10 min at room temperature and permeabilized with 0.1% Triton X-100 in PBS for 5 min at room temperature. Coverslips were incubated with a rhodamine phalloidin probe (Life Technologies) at a 1:40 dilution for 1 hour at room temperature. Samples were then mounted with ProLong Gold Antifade with DAPI (Life Technologies). Images were taken on a Zeiss AxioImager at 20x magnification. Images were analyzed for F-actin intensity and nuclear size (area) with Volocity (v 6.3). Cell size (area) was analyzed with ImageJ.

Gene Lists

FMRP targets and PSD genes, provided in S4 Table, were from Supplementary S2 Table from [70] and Supplementary S2 Table from [71], respectively. Lists of disease and trait-related genes, also provided in S4 Table, were as follows: Schizophrenia- de novo hits from [15]; Epilepsy- the union of GeneEpi (http://epilepsy.hardwicklab.org/) and BrainSpan (http://www.brainspan.org) [72]; Rheumatoid Arthritis, Breast Cancer, Type II Diabetes, Crohn’s Disease, Body Mass Index, Blood Pressure, and Height- the NHGRI-EBI Catalog of published genome-wide association studies (https://www.ebi.ac.uk/gwas/) [73]; Autism- the union of categories S through 5 from SFARI Gene (https://gene.sfari.org/autdb/Welcome.do) [74]; Intellectual Disability- Supplementary Table 6 from [75].

ToppGene

ToppGene (https://toppgene.cchmc.org) was used for prioritization of candidate genes as previously described [34]. The union of genes from GeneEpi (http://epilepsy.hardwicklab.org/) and BrainSpan (http://www.brainspan.org) [72] that was used for our epilepsy enrichment analyses was employed as our training set. Differentially expressed genes from each line were each used as separate test sets. Default cutoffs and gene limits were used for analyses with the addition of Interaction and Gene Family categories.

Statistical Analysis

Differentially expressed genes were determined by DESeq2 using a cutoff of TPM>1 in non-silencing control samples and an FDR-adjusted p<0.05. Two-sided Fisher’s exact tests were used for overrepresentation analyses. For qPCR, western blot, and fluorescent-based analyses, results are based on the combination of data from three separate transductions. Two-tailed Student’s t-tests were used to analyze results from these experiments, except for quantification of CYFIP1 mRNA and protein levels, where one-tailed Student’s t-tests were used. Fold enrichment was calculated as the fraction of DEGs that are within a given category (ex. FMRP targets) divided by the fraction of total expressed genes that are within that same given category. All results presented in graphs are expressed as fold change ± SEM. The number of samples used in each experiment is included in the figure legends.

Supporting Information

S1 Table. Quality control metrics for RNA-seq data.

https://doi.org/10.1371/journal.pone.0148039.s001

(XLSX)

S2 Table. Complete list of expressed genes in at least one donor line (TPM>1 in non-silencing control samples) from RNA-seq experiments.

https://doi.org/10.1371/journal.pone.0148039.s002

(XLSX)

S3 Table. Complete list of genes identified as differentially expressed in response to CYFIP1 knockdown with corresponding status with regard to FMRP target, postsynaptic density, schizophrenia, and epilepsy.

https://doi.org/10.1371/journal.pone.0148039.s003

(XLSX)

S4 Table. Lists of FMRP target and PSD genes along with disease and trait-related genes used for enrichment analyses.

https://doi.org/10.1371/journal.pone.0148039.s004

(XLSX)

S5 Table. Postsynaptic density genes dysregulated in response to CYFIP1 knockdown are involved in cytoskeletal remodeling.

https://doi.org/10.1371/journal.pone.0148039.s005

(XLSX)

S6 Table. Results for seven diseases or traits not known to be associated with BP1-2 deletion status.

https://doi.org/10.1371/journal.pone.0148039.s006

(XLSX)

S7 Table. Results from gene ontology analyses for schizophrenia genes dysregulated in response to CYFIP1 knockdown.

https://doi.org/10.1371/journal.pone.0148039.s007

(XLSX)

S8 Table. All significant ToppGene results for genes dysregulated in response to CYFIP1 knockdown.

https://doi.org/10.1371/journal.pone.0148039.s008

(XLSX)

S9 Table. Clinical information for BP1-2 deletion and BP1-2 neutral subjects from which lymphoblastoid cell lines were generated.

https://doi.org/10.1371/journal.pone.0148039.s009

(XLSX)

Acknowledgments

Thanks to all individuals who provided the biomaterials that enabled us to carry out the experiments described here and to the Simon’s Foundation for Autism Research for granting access to a subset of lymphoblastoid cell lines used. We also thank the Molecular Cytogenetic Core at Einstein for generating lymphoblastoid cell lines, the Analytical Imaging Facility at Einstein for advice and training on image analysis software, and the shRNA Core Facility at Einstein for generating lentiviruses.

Author Contributions

Conceived and designed the experiments: RAN BSA. Performed the experiments: RAN EP D. Zhao. Analyzed the data: RAN D. Zhao BSA HML D. Zheng. Contributed reagents/materials/analysis tools: HML. Wrote the paper: RAN BSA. Oversaw patient recruitment: JK.

References

  1. 1. Stefansson H, Rujescu D, Cichon S, Pietiläinen OPH, Ingason A, Steinberg S, et al. Large recurrent microdeletions associated with schizophrenia. Nature. 2008;455: 232–236. pmid:18668039
  2. 2. Zhao Q, Li T, Zhao X, Huang K, Wang T, Li Z, et al. Rare CNVs and tag SNPs at 15q11.2 are associated with schizophrenia in the Han Chinese population. Schizophrenia Bulletin. 2013;39: 712–719. pmid:22317777
  3. 3. de Kovel CGF, Trucks H, Helbig I, Mefford HC, Baker C, Leu C, et al. Recurrent microdeletions at 15q11.2 and 16p13.11 predispose to idiopathic generalized epilepsies. Brain. 2010;133: 23–32. pmid:19843651
  4. 4. Burnside RD, Pasion R, Mikhail FM, Carroll AJ, Robin NH, Youngs EL, et al. Microdeletion/microduplication of proximal 15q11.2 between BP1 and BP2: a susceptibility region for neurological dysfunction including developmental and language delay. Hum Genet. 2011;130: 517–528. pmid:21359847
  5. 5. Napoli I, Mercaldo V, Boyl PP, Eleuteri B, Zalfa F, De Rubeis S, et al. The Fragile X Syndrome Protein Represses Activity-Dependent Translation through CYFIP1, a New 4E-BP. Cell. 2008;134: 1042–1054. pmid:18805096
  6. 6. De Rubeis S, Pasciuto E, Li KW, Fernández E, Di Marino D, Buzzi A, et al. CYFIP1 Coordinates mRNA Translationand Cytoskeleton Remodelingto Ensure Proper Dendritic Spine Formation. Neuron. 2013;79: 1169–1182. pmid:24050404
  7. 7. Oguro-Ando A, Rosensweig C, Herman E, Nishimura Y, Werling D, Bill BR, et al. Increased CYFIP1 dosage alters cellular and dendritic morphology and dysregulates mTOR. Molecular Psychiatry. 2015;20: 1069–1078. pmid:25311365
  8. 8. Schenck A, Bardoni B, Moro A, Bagni C, Mandel JL. A highly conserved protein family interacting with the fragile X mental retardation protein (FMRP) and displaying selective interactions with FMRP-related proteins FXR1P and FXR2P. Proc Natl Acad Sci USA. 2001;98: 8844–8849. pmid:11438699
  9. 9. Bogdan S, Grewe O, Strunk M, Mertens A, Klämbt C. Sra-1 interacts with Kette and Wasp and is required for neuronal and bristle development in Drosophila. Development. 2004;131: 3981–3989. pmid:15269173
  10. 10. Pathania M, Davenport EC, Muir J, Sheehan DF, López-Doménech G, Kittler JT. The autism and schizophrenia associated gene CYFIP1 is critical for the maintenance of dendritic complexity and the stabilization of mature spines. Transl Psychiatry. 2014;4: e374. pmid:24667445
  11. 11. Bozdagi O, Sakurai T, Dorr N, Pilorge M, Takahashi N, Buxbaum JD. Haploinsufficiency of Cyfip1 Produces Fragile X-Like Phenotypes in Mice. Dawson TM, editor. PLoS ONE. 2012;7: e42422. pmid:22900020
  12. 12. Yoon K- J, Nguyen HN, Ursini G, Zhang F, Kim N- S, Wen Z, et al. Modeling a Genetic Risk for Schizophreniain iPSCs and Mice Reveals Neural Stem Cell Deficits Associated with Adherens Junctions and Polarity. Stem Cell. 2014;15: 79–91.
  13. 13. Brennand K, Savas JN, Kim Y, Tran N, Simone A, Hashimoto-Torii K, et al. Phenotypic differences in hiPSC NPCs derived from patients with schizophrenia. Molecular Psychiatry. 2015;20: 361–368. pmid:24686136
  14. 14. Madison JM, Zhou F, Nigam A, Hussain A, Barker DD, Nehme R, et al. Characterization of bipolar disorder patient-specific induced pluripotent stem cells from a family reveals neurodevelopmental and mRNA expression abnormalities. Molecular Psychiatry. 2015;20: 703–717. pmid:25733313
  15. 15. Purcell SM, Moran JL, Fromer M, Ruderfer D, Solovieff N, Roussos P, et al. A polygenic burden of rare disruptive mutations in schizophrenia. Nature. 2015;506: 185–190.
  16. 16. Fromer M, Pocklington AJ, Kavanagh DH, Williams HJ, Dwyer S, Gormley P, et al. De novo mutations in schizophrenia implicate synaptic networks. Nature. 2015;506: 179–184.
  17. 17. Glessner JT, Reilly MP, Kim CE, Takahashi N, Albano A, Hou C, et al. Strong synaptic transmission impact by copy number variations in schizophrenia. Proc Natl Acad Sci USA. 2010;107: 10584–10589. pmid:20489179
  18. 18. Pedrosa E, Sandler V, Shah A, Carroll R, Chang C, Rockowitz S, et al. Development of patient-specific neurons in schizophrenia using induced pluripotent stem cells. J Neurogenet. 2011;25: 88–103. pmid:21797804
  19. 19. Lin M, Pedrosa E, Shah A, Hrabovsky A, Maqbool S, Zheng D, et al. RNA-Seq of Human Neurons Derived from iPS Cells Reveals Candidate Long Non-Coding RNAs Involved in Neurogenesis and Neuropsychiatric Disorders. Yoshikawa T, editor. PLoS ONE. 2011;6: e23356–12. pmid:21915259
  20. 20. Lin M, Hrabovsky A, Pedrosa E, Wang T, Zheng D, Lachman HM. Allele-Biased Expression in Differentiating Human Neurons: Implications for Neuropsychiatric Disorders. de Erausquin GA, editor. PLoS ONE. 2012;7: e44017–10. pmid:22952857
  21. 21. Lin M, Zhao D, Hrabovsky A, Pedrosa E, Zheng D, Lachman HM. Heat Shock Alters the Expression of Schizophrenia and Autism Candidate Genes in an Induced Pluripotent Stem Cell Model of the Human Telencephalon. Westerheide SD, editor. PLoS ONE. 2014;9: e94968–11. pmid:24736721
  22. 22. Zhao D, Lin M, Chen J, Pedrosa E, Hrabovsky A, Fourcade HM, et al. MicroRNA Profiling of Neurons Generated Using Induced Pluripotent Stem Cells Derived from Patients with Schizophrenia and Schizoaffective Disorder, and 22q11.2 Del. Yue J, editor. PLoS ONE. 2015;10: e0132387–24. pmid:26173148
  23. 23. Chen J, Lin M, Hrabovsky A, Pedrosa E, Dean J, Jain S, et al. ZNF804A Transcriptional Networks in Differentiating Neurons Derived from Induced Pluripotent Stem Cells of Human Origin. Hu VW, editor. PLoS ONE. 2015;10: e0124597–23. pmid:25905630
  24. 24. Das DK, Tapias V, D'Aiuto L, Chowdari KV, Francis L, Zhi Y, et al. Genetic and Morphological Features of Human iPSC-Derived Neurons with Chromosome 15q11.2 (BP1-BP2) Deletions. Mol Neuropsychiatry. 2015;1: 116–123. pmid:26528485
  25. 25. Mitchison TJ, Cramer LP. Actin-Based Cell Motility and Cell Locomotion. Cell. 1996;84: 371–379. pmid:8608590
  26. 26. Hall A. Rho GTPases and the actin cytoskeleton. Science. 1998;279: 509–514. pmid:9438836
  27. 27. Schenck A, Bardoni B, Langmann C, Harden N, Mandel JL, Giangrande A. CYFIP/Sra-1 Controls Neuronal Connectivity in Drosophila and Links the Rac1 GTPase Pathway to the Fragile X Protein. Neuron. 2003;38: 887–898. pmid:12818175
  28. 28. Kunda P, Craig G, Dominguez V, Baum B. Abi, Sra1, and Kette Control the Stability and Localization of SCAR/WAVE to Regulate the Formation of Actin-Based Protrusions. 2003;13: 1867–1875.
  29. 29. Stefansson H, Meyer-Lindenberg A, Steinberg S, Magnusdottir B, Morgen K, Arnarsdottir S, et al. CNVs conferring risk of autism or schizophrenia affect cognition in controls. Nature. Nature Publishing Group; 2015;505: 361–366.
  30. 30. Hashemi B, Bassett A, Chitayat D, Chong K, Feldman M, Flanagan J, et al. Deletion of 15q11.2(BP1-BP2) region: Further evidence for lack of phenotypic specificity in a pediatric population. Am J Med Genet. 2015;167: 2098–2102.
  31. 31. Doornbos M, Sikkema-Raddatz B, Ruijvenkamp CAL, Dijkhuizen T, Bijlsma EK, Gijsbers ACJ, et al. Nine patients with a microdeletion 15q11.2 between breakpoints 1 and 2 of the Prader-Willi critical region, possibly associated with behavioural disturbances. European Journal of Medical Genetics. 2009;52: 108–115. pmid:19328872
  32. 32. Chaste P, Sanders SJ, Mohan KN, Klei L, Song Y, Murtha MT, et al. Modest Impact on Risk for Autism Spectrum Disorder of Rare Copy Number Variants at 15q11.2, Specifically Breakpoints 1 to 2. Autism Res. 2014;7: 355–362. pmid:24821083
  33. 33. Malhotra D, Sebat J. CNVs: Harbingers of a Rare Variant Revolution in Psychiatric Genetics. Cell. 2012;148: 1223–1241. pmid:22424231
  34. 34. Chen J, Bardes EE, Aronow BJ, Jegga AG. ToppGene Suite for gene list enrichment analysis and candidate gene prioritization. 2009;37: W305–W311.
  35. 35. Splawski I, Timothy KW, Sharpe LM, Decher N, Kumar P, Bloise R, et al. CaV1.2 Calcium Channel Dysfunction Causes a Multisystem Disorder Including Arrhythmia and Autism. Cell. 2004;119: 19–31. pmid:15454078
  36. 36. Holt IJ, Harding AE, Petty RK. A new mitochondrial disease associated with mitochondrial DNA heteroplasmy. The American Journal of Human Genetics. 1990;46: 428–433.
  37. 37. Kirby DM, McFarland R, Ohtake A, Dunning C, Ryan MT, Wilson C, et al. Mutations of the mitochondrial ND1 gene as a cause of MELAS. Journal of Medical Genetics. 2004;41: 784–789. pmid:15466014
  38. 38. Schwindinger WF, Giger KE, Betz KS, Stauffer AM, Sunderlin EM, Sim-Selley LJ, et al. Mice with deficiency of G protein gamma3 are lean and have seizures. Molecular and Cellular Biology. 2004;24: 7758–7768. pmid:15314181
  39. 39. Buntup D, Skare Ø, Solbu TT, Chaudhry FA, Storm-Mathisen J, Thangnipon W. β-Amyloid 25–35 Peptide Reduces the Expression of Glutamine Transporter SAT1 in Cultured Cortical Neurons. Neurochem Res. 2007;33: 248–256. pmid:18058230
  40. 40. Puschmann A, Wszolek ZK, Farrer M, Gustafson L, Widner H, Nilsson C. Parkinsonism and Related Disorders. Parkinsonism and Related Disorders. 2009;15: 390–392.
  41. 41. Mirza N, Vasieva O, Marson AG, Pirmohamed M. Exploring the genomic basis of pharmacoresistance in epilepsy: an integrative analysis of large-scale gene expression profiling studies on brain tissue from epilepsy surgery. Human Molecular Genetics. 2011;20: 4381–4394. pmid:21852245
  42. 42. Baum L, Haerian BS, Ng H-K, Wong VCN, Ng PW, Lui CHT, et al. Case—control association study of polymorphisms in the voltage-gated sodium channel genes SCN1A, SCN2A, SCN3A, SCN1B, and SCN2B and epilepsy. Hum Genet. 2013;133: 651–659. pmid:24337656
  43. 43. Born HA, Kim JY, Savjani RR, Das P, Dabaghian YA, Guo Q, et al. Genetic Suppression of Transgenic APP Rescues Hypersynchronous Network Activity in a Mouse Model of Alzeimer's Disease. Journal of Neuroscience. 2014;34: 3826–3840. pmid:24623762
  44. 44. Sima X, Xu J, Li J, Zhong W, You C. Expression of β-amyloid precursor protein in refractory epilepsy. Molecular Medicine Reports. 2014;9: 1242–1248. pmid:24566726
  45. 45. Crotti L, Johnson CN, Graf E, De Ferrari GM, Cuneo BF, Ovadia M, et al. Calmodulin mutations associated with recurrent cardiac arrest in infants. Circulation. 2013;127: 1009–1017. pmid:23388215
  46. 46. Gurses C, Azakli H, Alptekin A, Cakiris A, Abaci N, Arikan M, et al. Mitochondrial DNA profiling via genomic analysis in mesial temporal lobe epilepsy patients with hippocampal sclerosis. Gene. 2014;538: 323–327. pmid:24440288
  47. 47. Girirajan S, Rosenfeld JA, Coe BP, Parikh S, Friedman N, Goldstein A, et al. Phenotypic heterogeneity of genomic disorders and rare copy-number variants. N Engl J Med. 2012;367: 1321–1331. pmid:22970919
  48. 48. Spencer CM, Alekseyenko O, Hamilton SM, Thomas AM, Serysheva E, Yuva-Paylor LA, et al. Modifying behavioral phenotypes in Fmr1KO mice: genetic background differences reveal autistic-like responses. Crawley J, DiCicco-Bloom E, Bailey AJ, editors. Autism Res. 2011;4: 40–56. pmid:21268289
  49. 49. Moy SS, Nadler JJ, Young NB, Nonneman RJ, Grossman AW, Murphy DL, et al. Social approach in genetically engineered mouse lines relevant to autism. Genes, Brain and Behavior. 2009;8: 129–142.
  50. 50. Pietropaolo S, Guilleminot A, Martin B, D'Amato FR, Crusio WE. Genetic-Background Modulation of Core and Variable Autistic-Like Symptoms in Fmr1 Knock-Out Mice. Cushing B, editor. PLoS ONE. 2011;6: e17073–11. pmid:21364941
  51. 51. Zeng L, Zhang P, Shi L, Yamamoto V, Lu W, Wang K. Functional Impacts of NRXN1 Knockdown on Neurodevelopment in Stem Cell Models. Arking DE, editor. PLoS ONE. 2013;8: e59685–11. pmid:23536886
  52. 52. Sugathan A, Biagioli M, Golzio C, Erdin S, Blumenthal I, Manavalan P, et al. CHD8 regulates neurodevelopmental pathways associated with autism spectrum disorder in neural progenitors. Proc Natl Acad Sci USA. 2014;111: E4468–77. pmid:25294932
  53. 53. Germain ND, Chen P- F, Plocik AM, Glatt-Deeley H, Brown J, Fink JJ, et al. Gene expression analysis of human induced pluripotent stem cell-derived neurons carrying copy number variants of chromosome 15q11-q13.1. Mol Autism. 2014;5: 44. pmid:25694803
  54. 54. Paşca SP, Portmann T, Voineagu I, Yazawa M, Shcheglovitov A, Paşca AM, et al. Using iPSC-derived neurons to uncover cellular phenotypes associated with Timothy syndrome. Nat Med. 2011;17: 1657–1662. pmid:22120178
  55. 55. Bavamian S, Mellios N, Lalonde J, Fass DM, Wang J, Sheridan SD, et al. Dysregulation of miR-34a links neuronal development to genetic risk factors for bipolar disorder. Molecular Psychiatry. 2015;20: 573–584. pmid:25623948
  56. 56. Robicsek O, Karry R, Petit I, Salman-Kesner N, ller F-JMU, Klein E, et al. Abnormal neuronal differentiation and mitochondrial dysfunction in hair follicle-derived induced pluripotent stem cells of schizophrenia patients. Molecular Psychiatry. 2013;18: 1067–1076. pmid:23732879
  57. 57. Topol A, Zhu S, Tran N, Simone A, Fang G, Brennand KJ. Altered WNT Signaling in Human Induced Pluripotent Stem Cell Neural Progenitor Cells Derived from Four Schizophrenia Patients. Biological Psychiatry. 2015;78: e29–34. pmid:25708228
  58. 58. Weick JP, Held DL, Bonadurer GF, Doers ME, Liu Y, Maguire C, et al. Deficits in human trisomy 21 iPSCs and neurons. Proc Natl Acad Sci USA. 2013;110: 9962–9967. pmid:23716668
  59. 59. Marchetto MCN, Carromeu C, Acab A, Yu D, Yeo GW, Mu Y, et al. A model for neural development and treatment of Rett syndrome using human induced pluripotent stem cells. Cell. 2010;143: 527–539. pmid:21074045
  60. 60. Hur E-M, Zhou F-Q. GSK3 signalling in neural development. Nat Rev Neurosci. 2010;11: 539–551. pmid:20648061
  61. 61. Freyberg Z, Ferrando SJ, Javitch JA. Roles of the Akt/GSK-3 and Wnt Signaling Pathways in Schizophrenia and Antipsychotic Drug Action. American Journal of Psychiatry. 2010;167: 388–396. pmid:19917593
  62. 62. Fischbach GD, Lord C. The Simons Simplex Collection: A Resource for Identification of Autism Genetic Risk Factors. Neuron. 2010;68: 192–195. pmid:20955926
  63. 63. Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, Salzberg SL. TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biology. 2013;14: R36. pmid:23618408
  64. 64. Anders S, Pyl PT, Huber W. HTSeq—a Python framework to work with high-throughput sequencing data. Bioinformatics. 2015;31: 166–169. pmid:25260700
  65. 65. Anders S, Huber W. Differential expression analysis for sequence count data. Genome Biology. 2010;11: R106. pmid:20979621
  66. 66. Wagner GP, Kin K, Lynch VJ. Measurement of mRNA abundance using RNA-seq data: RPKM measure is inconsistent among samples. Theory Biosci. 2012;131: 281–285. pmid:22872506
  67. 67. Edgar R, Domrachev M, Lash AE. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Research. 2002;30: 207–210. pmid:11752295
  68. 68. Huang DW, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2008;4: 44–57.
  69. 69. Huang DW, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Research. 2009;37: 1–13. pmid:19033363
  70. 70. Darnell JC, Van Driesche SJ, Zhang C, Hung KYS, Mele A, Fraser CE, et al. FMRP Stalls Ribosomal Translocation on mRNAs Linked to Synaptic Function and Autism. Cell. 2011;146: 247–261. pmid:21784246
  71. 71. Bayés À, van de Lagemaat LN, Collins MO, Croning MDR, Whittle IR, Choudhary JS, et al. Characterization of the proteome, diseases and evolution of the human postsynaptic density. Nature Neuroscience. 2010;14: 19–21. pmid:21170055
  72. 72. BrainSpan: Atlas of the Developing Human Brain [Internet]. Funded by ARRA Awards 1RC2MH089921-01, 1RC2MH090047-01, and 1RC2MH089929-01. © 2011. Available from: http://developinghumanbrain.org.
  73. 73. Hindorff LA, MacArthur J (European Bioinformatics Institute), Morales J (European Bioinformatics Institute), Junkins HA, Hall PN, Klemm AK, and Manolio TA. A Catalog of Published Genome-Wide Association Studies. Available at: www.genome.gov/gwastudies. Accessed [15 May 2015].
  74. 74. Abrahams BS, Arking DE, Campbell DB, Mefford HC, Morrow EM, Weiss LA, et al. SFARI Gene 2.0: a community-driven knowledgebase for the autism spectrum disorders (ASDs). Mol Autism. 2013;4: 36. pmid:24090431
  75. 75. Neale BM, Kou Y, Liu L, Ma’ayan A, Samocha KE, Sabo A, et al. Patterns and rates of exonic de novo mutations in autism spectrum disorders. Nature. 2012;485: 242–245. pmid:22495311