Modeling non-syndromic autism and the impact of TRPC6 disruption in human neurons

Patient ascertainment

Analysis of genomic copy number variations

Genomic DNA was hybridized to the HumanHap300 Genotyping BeadChip from Illumina (La Jolla, CA, USA) according to manufacturer’s protocol to detect possible CNVs in the ASD individual. The data were analyzed using PennCNV ¹⁰ and QuantiSNP ¹¹ software, and the results were compared to the database of genomic variants (http://projects.tcag.ca/variation/) to classify the identified CNVs as rare or common variants.

Fluorescent in situ hybridization

Chromosomes for fluorescent in situ hybridization (FISH) analysis were prepared from colchicine-treated lymphocytes of the proband. Bacterial artificial chromosomes (BACs) encompassing the genomic regions of interest were selected from the RPCI-11 library (Roswell Park Cancer Institute) using the UCSC genome browser (http://genome.ucsc.edu/, assembly Mar. 2006, NCBI36/hg18). The BACs were fluorescently labeled by nick translation and hybridized to the metaphase spreads using standard protocols ¹².

Exome sequencing

Exome sequencing and analysis were performed by BGI Tech (Shenzhen, China). Briefly, genomic DNA samples were randomly fragmented into segments with a base-pair peak of 150 to 200 bp, and library enrichment for exonic sequences was performed using Agilent SureSelect Human All Exon 51M (for individual and mother) or Agilent SureSelect Human All Exon 71M (for the father). The captured libraries were loaded on Hiseq2000, and the sequences of each individual were generated as 90-bp paired-end reads. The coverage for the three individuals was 80-fold. Burrows-Wheeler Aligner (BWA) was used for the alignment. Single nucleotide polymorphisms (SNPs) were identified by SOAPsnp, small insertion/deletion (InDels) were detected by Samtools/GATK, and single nucleotide variants (SNVs) were detected by 1/35 Varscan.

Isolation and culture of human DPCs

DPC lineages were obtained as described elsewhere ¹³. Briefly, dental pulp tissues were digested in a solution of 0.25% trypsin for 30 minutes at 37°C. The cells were cultivated in DMEM/F12 media (Gibco) supplemented with 15% fetal bovine serum (Hyclone, TX), 1% penicillin/streptomycin, and 1% non-essential amino acids and maintained under standard conditions (37°C, 5% CO₂). The DPC control lineages used for the whole-genome expression analysis were donated by Dr. Daniela Franco Bueno and Gerson Shigueru Kobayashi of the University of São Paulo. One of the DPC control lineages used for iPSC generation was a kind gift from Dr. Songtao Shi (University of Southern California).

RNA extraction

RNA samples were extracted from lymphocytes, DPCs, and iPSCs using Trizol reagent (Invitrogen, CA) and treated with Turbo DNA-free (Ambion). Sample concentrations and quality were evaluated using a Nanodrop 1000 and gel electrophoresis.

Microarray studies

For microarray experiments, 100 ng of RNA was converted to cDNA, amplified, labeled, and hybridized to the Human Gene 1.0ST chip from Affymetrix following the manufacturer’s protocol. The chips were scanned using the GeneChip® Scanner 3000 7G System, and a quality control was processed using Affymetrix® Expression Console™ Software. The data were normalized using the robust multi-array average (RMA) method ¹⁴, and the differentially expressed genes were selected with the significance analysis of microarrays method (SAM) ¹⁵ and RankProd ¹⁶. To select DEGs, we used a p-value < 0.05 adjusted for the false discovery rate (FDR) and 3,000 permutations. Functional annotation, canonical pathways, and networks analyses were performed using Ingenuity Pathways (http://www.ingenuity.com/). The CREB target genes database (http://natural.salk.edu/CREB/search.htm ¹⁷) was used to determine whether the DEGs found are regulated by the transcription factor CREB.

Gene expression analyses by qPCR

RNA samples were reversed transcribed into cDNA using the Super Script III First Strand Synthesis System (Invitrogen, CA) according to the manufacturer’s instructions. The reactions were run on an Applied Biosystem 7500 sequence detection system using SYBR Green master mix (Applied Biosystems, CA). The primers were designed using PrimerExpress v. 2.0 software (Applied Biosystems, CA), and specificity was verified by melting curve analysis using 7500 System SDS v. 1.2 Software (Applied Biosystems, CA). Quantitative analysis was performed using the comparative threshold cycle method ¹⁸. GeNorm (www.medgen.ugent.be/genorm/) was used to determine the stability of the reference genes GAPDH, HPRT1, SDHA, and HMBS and to generate a normalization factor for the expression values of the target genes. The principles of analysis of geNorm have been described ¹⁹. Microarray validation was performed using the one-tailed unpaired t-test with Welch’s correction to compare the qPCR expression values obtained for the ASD individual and controls. A concentration of 10 µM hyperforin was used to treat the DPCs of a control sample for 15 and 30 minutes and 1, 3, 6, 24, and 48 hours. The samples were prepared in triplicate, and the results were normalized by the values obtained for an untreated sample. Primers used on this work are described in Table S1.

Western blotting

Rabbit anti-TRPC6 (ProScience, 1:250; Sigma, 1:1000); mouse anti-TRPC6 (Abcam, 1:1000); rabbit anti-CREB (Cell Signaling, 1:500); rabbit anti-P-CREB (Cell Signaling, 1:500); and mouse anti-β-actin (Ambion, 1:5000) antibodies were used as primary antibodies. Horseradish peroxidase-conjugated goat anti-rabbit and goat anti-mouse (Promega, 1:2000) antibodies were used as secondary antibodies. ECL Plus (Amersham) was used for signal detection. For Semiquantitative analysis of p-CREB signal, intensity was corrected with respect to CREB/β-actin relative quantification. A paired t-test analysis with a p-value < 0.05 was used to compare the control and ASD individual p-CREB signal intensity normalized data.

Cellular reprogramming

The iPSCs were obtained from the DPCs of the ASD individual and a control. Briefly, DPCs were transduced with retroviruses containing OCT4, SOX2, KFL4, and MYC to induce overexpression of these genes ⁵. Two days after transduction, the cells were transferred to a co-culture system with murine embryonic fibroblasts (mEFs) maintained with DMEM/F12 (Invitrogen, CA), 20% Knockout Serum Replacement (Invitrogen, CA), 1% non-essential amino acids, and 100 µM beta-mercaptoethanol and treated with 1 mM valproic acid (Sigma) for 5 days. The iPSC colonies were identified after approximately 2 weeks in this culture system, transferred to Matrigel (BD Biosciences)-coated plates, and maintained in mTeSR media (Stem Cell Technologies).

Immunocytochemistry

The cells were fixed with PBS containing 4% paraformaldehyde for 10 minutes and then incubated at room temperature for 1 hour in a blocking solution containing 5% donkey serum and 0.1% Triton X-100. The primary antibodies were incubated overnight at 4°C, followed by incubation with secondary antibodies (Jackson ImmunoResearch) for 1 hour at room temperature. Images were captured with a Zeiss microscope. The primary antibodies used included the following: Tra-1-81 (1:100, Chemicon); Nanog and Lin28 (1:500, R&D Systems); Sox2 (1:250; Chemicon); human Nestin (1:100, Chemicon); Tuj1 (1:500, Covance); MAP2 (1:100; Sigma); VGLUT1 (1:200, Synaptic Systems); GABA (1:100, Sigma); Musashi (1:200, Abcam); Ctip2 (1:200, Abcam); and Tbr1 (1:200, Abcam).

Teratoma formation

iPSC colonies from five semi-confluent 100 mm dishes (1–3 ×10⁶ cells) were harvested after treatment with 0.5 ng/mL dispase, pelleted, and suspended in 300 µL Matrigel. The cells were injected subcutaneously into nude mice; 5 to 6 weeks after injection, teratomas were dissected, fixed overnight in 10% buffered formalin phosphate, and embedded in paraffin. The sections were stained with hematoxylin and eosin for further analysis. The protocols were approved by the University of California San Diego Institutional Animal Care and Use Committee.

Fingerprinting and karyotype

Standard G-banding karyotype and DNA fingerprinting analysis were performed by Cell Line Genetics (Madison, WI).

Neuronal differentiation

The iPSC colonies were plated on Matrigel (BD Biosciences)-coated plates and maintained for 5 days in mTSeR media (Stem Cell Technologies). On the 5^th day, the media was changed to N2 media [DMEM/F12 media supplemented with 1X N2 supplement (Invitrogen) and 1 µM dorsomorphin (Tocris). After 2 days, the colonies were removed from the plate and cultured in suspension as embryoid bodies (EBs) for 2–3 weeks using N2 media with dorsomorphin during the entire procedure. The EBs were then gently dissociated with accutase (Gibco), plated on Matrigel-coated dishes, and maintained in NBF media (DMEM/F12 media supplemented with 0.5X N2, 0.5X B7 supplements, 20 ng/mL FGF and 1% penicillin/streptomycin). The rosettes that emerged after 3 or 4 days were manually selected, gently dissociated with accutase, and plated in dishes coated with 10 µg/mL poly-ornithine and 5 µg/mL laminin. This NPC population was expanded using NBF media. To differentiate the NPCs into neurons, the cells were re-plated with 10 µM ROCK inhibitor (Y-27632, Calbiochem) in the absence of FGF, with regular media changes every 3 or 4 days.

Ca²⁺ influx studies

Intracellular Ca²⁺ levels were monitored using Fluo-4 AM. The cells were incubated for 45 minutes at 37°C with 2.5 µM Fluo-4 AM and superfused for 5 minutes with HBSS buffer before the beginning of the recording. A concentration of 10 µM hyperforin (a kind gift from Dr. Willmar Schwabe GmbH & Co, Karlsruhe, Germany) was used in combination with 100 µM FFA (Sigma-Aldrich) for TRPC6 activation. Images were captured at 6-second intervals for 30 minutes using a Biorad MRC 1024 confocal system attached to an Olympus BX70 microscope. The drugs were applied at the 3^rd minute using a perfusion system. A triplicate of each individual was analyzed. The average fluorescence of the individual cells was quantified and normalized to the resting fluorescence level for each cell. The plugins MultiMeasure and MeasureStacks from ImageJ software were used to measure fluorescence intensity. The analyses were performed blinded to avoid bias.

Cell cycle analysis

A total of 1×10⁶ NPCs were harvested from a single-cell suspension with PBS washing buffer (PBS and 1% serum) and fixed in 75% EtOH for at least 2 hours at 4°C. After washing twice with washing buffer, the cells were stained with 200 µL propidium iodine (PI) solution (20 µg/mL propidium iodide, 200 µg/mL RNase A, and 0.1% Triton X-100). Multiple NPC samples from the TRPC6-mutant individual and controls were analyzed by fluorescence-activated cell sorting (FACS) on a Becton Dickinson LSRI, and cell cycle gating was examined using FLOWJO-Flow Cytometry Analysis Software.

Quantification of neuronal morphology and synaptic puncta

Neuronal tracing was performed on neurons for which the shortest dendrite was at least three times longer than the cell soma diameter using a semi-automatic ImageJ plug-in (NeuroJ). Spines and VGLUT1 puncta were quantified after three-dimensional reconstruction of z-stack confocal images. The same density of neurons was plated in each condition. Final cell density was confirmed by DAPI and Synapsin-EGFP-positive cells. Only Synapsin-EGFP positive neurons with spines were scored. Images were taken randomly for each individual and from two different experiments, using at least two different clones. Quantification was performed blind to the cell genotype. The total dendritic length includes the summed length of all dendrites per neuron and dendritic segment count represents the total number of dendritic segments per neuron. No distinction was made between different types of spines due to the unviability of this assessment using the presented method. All experiments were performed with independent clones and different controls. All analyses were performed blinded to avoid bias. For the rescue experiments, 10 ng/mL IGF1 (Peprotech) or 0.5µM hyperforin was added to neuronal cultures for 2 weeks.

Construction and characterization of retroviruses

Self-inactivating murine oncoretroviruses were engineered to express short-hairpin RNAs (shRNAs) under the control of the U6 promoter and green fluorescent protein (GFP) or the Discosoma sp. red fluorescent protein DsRed under the control of the Ef1 alpha promoter. shRNAs against TRPC6 and a non-silencing scrambled control shRNA were cloned into retroviral vectors as previously described ²². The following shRNA sequences were selected and cloned into retroviral vectors: shRNA-control, 5’-TTCTCCGAACGTGTCACGT-3’; shRNA-TRPC6-1, 5’-TCGAGGACCAGCATACATG-3’; and shRNA-TRPC6-3, 5’-CTCAGAAGATTATCATTTA-3’.

For rescue experiments, a resistant form of murine TRPC6 (TRPC6-WT^R) was engineered to harbor six silent mutations in the region targeted by shRNA-TRPC6-1. The TRPC6 targeting sequence was mutated from AAT CGA GGA CCA GCA TAC ATG to AAC CGC GGC CCT GCT TAT ATG by site-directed mutagenesis. The resistant form of TRPC6 was cloned into a retroviral vector driven by the ubiquitin promoter followed by a bicistronic expression of GFP and a WPRE stabilization sequence. The specificity and efficiency of shRNA-control, shRNA-TRPC6-1, shRNA-TRPC6-3, and the TrpC6-WT constructs were verified by co-transfection into HEK-293 cells. Cell lysates were collected and analyzed by western blot analysis with anti-TRPC6 antibodies (Sigma).

Primary hippocampal cultures

Hippocampal neuronal cultures were prepared from C57BL/6 E18 embryonic mice. Briefly, hippocampi were dissected, dissociated with trypsin, and plated at a density of 300 cells/mm² on glass coverslips coated with poly-L-lysine and laminin. The hippocampal neurons were maintained in Neurobasal medium (Gibco) supplemented with B27 (Invitrogen). Neurons were treated with either shRNA scramble control or shRNA targeting TRPC6 at DIV12-14 and were fixed for further analysis at DIV21.

In vivo stereotaxic injection of engineered retroviruses into the dentate gyrus of adult mouse hippocampus

High titers of engineered retroviruses were produced by cotransfection of retroviral vectors and vesicular stomatitis viral envelope into the 293 GP cell line as described previously ²³. Supernatants were collected 24 hours post transfection, filtered through 45-µm filters, and ultracentrifuged. The viral pellet was dissolved in 14 µL of PBS and stereotaxically injected into the hilus of anesthetized mice at four sites (0.5 µL per site at 0.25 µL/minute). The following coordinates were used: posterior = 2 mm from the bregma, lateral = ±1.6 mm, ventral = ±2.5 mm; posterior = 3 mm from the bregma, lateral = ±2.6 mm, ventral = ±3.2 mm. Adult C57BL/6 mice (6–8 weeks old, female) were used for the study. All procedures followed institutional guidelines.

Immunostaining and confocal analysis

Coronal brain sections (40 µm thick) were prepared from retrovirus-injected mice. Images of GFP⁺ cells were acquired on a META multiphoton confocal system. Neuronal positioning was analyzed by acquiring a single-section confocal image of a GFP⁺ cell body stained with DAPI and assigning it to one of the four domains as illustrated. A minimum of 10 GFP⁺ cells were randomly chosen from the each animal, and at least three animals were used for each experimental condition, as previously described ²⁴. Statistical significance was determined by ANOVA. Dendritic development was analyzed by via a three-dimensional reconstruction of the entire dendritic tree from Z-series stacks of confocal images. The images were converted to two-dimensional projections for analysis of dendritic length and branch number using NIH ImageJ software and the NeuronJ plugin, as described previously ²⁴. As a measure of arborization, Sholl analysis was performed by counting the number of dendritic crossings at a series of concentric circles at 10-µm intervals from the cell body using the Sholl analysis plugin.

Slice electrophysiology

Mice housed under standard conditions were anesthetized at 3 weeks post-retroviral injection, and acute coronal slices were prepared as previously described ²⁵. The brains were removed and placed in an ice-cold cutting solution containing the following: 110 mM choline chloride; 2.5 mM KCl; 1.3 mM KH₂PO₄; 25 mM NaHCO₃; 0.5 mM CaCl₂; 7 mM MgCl₂; 10 mM dextrose; 1.3 mM sodium ascorbate; 0.6 mM sodium pyruvate; and 5 mM kynurenic acid. Slices were cut into 300-µm-thick sections with a vibratome (Leica VT1000S) and transferred to a chamber containing ACSF: 125 mM NaCl; 2.5 mM KCl; 1.3 mM KH₂PO₄; 25 mM NaHCO₃; 2 mM CaCl₂; 1.3 mM MgCl₂; 1.3 mM sodium ascorbate; 0.6 mM sodium pyruvate; and 10 mM dextrose (pH 7.4, 320 mOsm), saturated with 95% O₂, 5% CO₂ at 35°C for 20 minutes. The slices were then maintained at room temperature for at least 45 minutes prior to placement in the recording chamber. The slices were maintained at room temperature and used for the following 4 hours. Electrophysiological recordings were performed at 34°C. Microelectrodes (4–6 MΩ) were filled with a solution containing the following: 120 mM potassium gluconate; 15 mM KCl; 4 mM MgCl₂; 0.1 mM EGTA; 10.0 mM HEPES; 4 mM MgATP; 0.3 mM Na₃GTP; and 7 mM phosphocreatine (pH 7.4, 300 mOsm). The whole-cell patch-clamp configuration was used in the current-clamp mode. Approximately 10–20 giga-ohm seals were obtained with borosilicate glass microelectrodes. The electrophysiological recordings were obtained at 32–34°C. Neurons and dendrites were visualized through differential interference contrast microscopy. The data were collected using an Axon Instruments 200B amplifier and acquired via a Digidata 1322A at 10 kHz.

Electrophysiology recordings using cultured human iPSC-derived neurons

Whole-cell patch clamp recordings were performed using cells cultured in the absence of astrocytes after approximately 6 weeks of differentiation. Before the recordings, the growth media were removed and replaced with a bath solution comprising the following: 130 mM NaCl; 3 mM KCl; 1 mM CaCl₂; 1 mM MgCl₂; 10 mM HEPES; and 10 mM glucose (pH 7.4) at room temperature (22–24°C). The electrodes for whole-cell recordings were pulled on a Flaming/Brown micropipette puller (Model P-87, Sutter Instrument, Novato, CA) from filamented borosilicate capillary glass (1.2 mm OD, 0.69 mm ID, World Precision Instruments, Sarasota, FL). The electrodes were fire polished, and the resistance values were typically 2–5 MΩ for the voltage-clamp experiments and 7–9 MΩ for the current-clamp experiments. The pipette solution contained the following: 138 mM KCl; 0.2 mM CaCl₂; 1 mM MgCl₂; 10 mM HEPES (Na⁺ salt); and 10 mM EGTA, (pH 7.4). The osmolarity of all solutions was adjusted to 290 mOsM. All chemicals were purchased from Sigma-Aldrich (St. Louis, MO), with the exception of MgCl₂ (J.T. Baker, Phillipsburg, NJ). Current traces in voltage clamp were leak-subtracted. Liquid junction potentials were nulled for each individual cell with the Axopatch 1C amplifier (Molecular Devises, Sunnyvale, CA). The analyses were performed in a double-blinded manner to avoid bias.

Behavioral tests in mice

The three-chamber test was used to evaluate the social behavior of TRPC6 wild type (WT), heterozygous (HET) and knockout (KO) mice. To evaluate repetitive behavior, the mice were initially observed for 10 minutes in the dark, and the time spent in grooming and freezing behavior was measured. After 5 minutes of habituation under a light condition, a small cage with a never-met animal was introduced to one side of the box, and an empty cage was introduced to the other side. The time spent in each chamber and the time spent during nose-to-nose interaction between the animals was measured. Adult mice (6–8 weeks old, male) with a C57BL/6 background were used for the study. At least 12 animals per group were utilized as biological replicates. The experimenter was blind to the genotypes. The data were analyzed using the non-parametric Kruskal-Wallis ANOVA. The analyses were performed in a double-blinded manner. All procedures followed institutional guidelines.

Mutation screening of TRPC6

Characterization of the t(3;11)(p21;q22) translocation breakpoint and exome sequencing

We identified an 8-year-old male autistic individual carrying a de novo 46, XY, t(3;11)(p21;q22) translocation by G-banding karyotyping of lymphoblastoid cells. No gain or loss of genetic material was observed near the breakpoint areas via a genome-wide array analysis (Figure 1a). Only a duplication (104.225.150 bp - 104.339.273 bp) on chromosome 14 was identified, which was previously shown to be a common copy number variant (CNV; http://projects.tcag.ca/variation/). Fluorescent in situ hybridization (FISH) analysis revealed that BAC probes RP11-780O20 and RP11-109N8 span the breakpoint on chromosome 3p21, while probes RP11-3F4 and RP11-1006P7 map distal and proximal to the breakpoint, respectively (Figure 1b, c). This narrowed the breakpoint to an interval of approximately 15 kb spanning the gene encoding the Vpr-binding protein (VPRBP), indicating that this gene was disrupted. Similarly, the breakpoint on chromosome 11q22 was mapped to a region spanned by probes RP11-141E21 and RP11-153K15, distal to RP11-315B9 and proximal to RP11-942D19 (Figure 1d, e), suggesting disruption of the TRPC6 gene, which was confirmed by the use of additional strategies.

We first measured the expression levels of TRPC6 exons 4, 6, 12, and 13 in the lymphocytes of the ASD individual, his parents, and six non-affected control individuals by quantitative real-time PCR (qPCR) (Figure S1a). In the ASD individual’s parents and in six other individual controls, exons 6, 12, and 13 had similar expression levels as exon 4. In the ASD individual, however, the expression levels of exons 12 and 13 were reduced by 60% compared to exon 4. After sequencing all TRPC6 exons, we found that the individual was heterozygous for two common polymorphisms: one mapping to exon 6 (rs12366144) and the other to exon 13 (rs12805398). However, sequencing of cDNA from the individual’s lymphocytes revealed heterozygosity only for the polymorphism in exon 6 (Figure S1b). Parentage was confirmed through genotyping of microsatellite markers (Figure S1c). These results demonstrate that TRPC6 has biallelic expression and that the heterozygosity loss in exon 13 in the individual’s cDNA can be explained by TRPC6 disruption. Accordingly, TRPC6 is transcribed up to the breakpoint, which is located between exons 6 and 12. We did not identify any pathogenic change in TRPC6 exons upon sequencing the individual’s DNA (data not shown). We also did not identify any extra band in the protein extracts from individual’s cells using a N-terminal antibody, indicating that a truncated TRPC6 form is unlikely to be a byproduct of the translocation (Figure S1d).

Disruption of TRPC6, VPRBP, and several other unknown genes might contribute to the ASD phenotype. To identify other genetic alterations in this ASD individual, we performed exome sequencing on the individual and compared the result to those for his parents. Exome sequencing analysis revealed 50 de novo, rare, nonsynonymous variants and three frameshift insertions/deletions in the individual. Consultation with AutismKB ²⁷ indicated that none of the other genes harboring genetic variants are associated with ASD, with the exception of the cyclic adenosine monophosphate-specific (PDE4A) gene, for which lower levels of expression have been observed in the brains of autistic individuals ²⁸. We also observed an alteration in the ATXN3 gene, linked to the spinocerebellar ataxia-3 disease in humans. All genetic variants are presented in Table S3.

TRPC6 disruption leads to transcriptional alterations and dysregulation of CREB phosphorylation

To determine gene transcription due to genetic perturbations in the ASD individual carrying the novel chromosomal translocation, we conducted a global expression analysis comparing the individual’s dental pulp cells (DPCs) to six control samples. DPCs can be easily isolated from the deciduous teeth of ASD individuals via a non-invasive procedure ²⁹. DPCs have an ectodermic neural crest origin, express several neuronal markers, and have proven to be a useful model to study ASD ^30–32. We identified 67 differentially expressed genes (DEGs) between the ASD individual and non-affected controls (P<0.05; Table S4). Functional annotation analysis revealed that 16 (24%) of these genes have a role in nervous system development and function (Table 1). We confirmed the reduction of TRPC6 expression (P<0.01) but not VPRBP (Figure 2a). The reduced level of TRPC6 expression is likely due to nonsense-mediated decay or rearrangement of regulatory elements caused by the translocation. Moreover, PDE4A, ATXN3 and DOCK3 (another neuronal gene ³³ present near the break point on chromosome 3, Figure 1b) were also not differentially regulated in the individual’s DPCs (Figure 2a). TRPC6 is a Ca²⁺-permeable, nonselective, cation channel involved in neuronal survival, growth cone guidance, and spine and synapse formation, biological processes that have previously been implicated in ASD etiology ^{8, 9, 34–36}. The function of VPRBP (Vpr-binding protein) is less clear and may include DNA replication, S-phase progression, and cellular proliferation ³⁷. Given the time-consuming nature of additional functional analyses, we elected to focus on additional genetic and functional studies of TRPC6, which has not been previously associated with ASDs.

Using the CREB-target genes database (http://natural.salk.edu/CREB/), we determined that 8 of the 16 functionally relevant DEGs are regulated by CREB, a transcription factor that is activated upon Ca²⁺ influx through TRPC6 ⁸. Of the functionally relevant DEGs, we evaluated 6 CREB-target genes (INA, MAP2, NPTX1, CLDN11, PCDH10, and EPHA4) and two other genes (SEMA3A and CDH6) by quantitative PCR (qPCR) to validate the microarray experiments. We measured dysregulated expression of CDH6 (-2.68-fold, P<0.05), INA (−2.64-fold, P<0.05), MAP2 (−2.79-fold, P<0.05), and CLDN11 (4.07-fold, P<0.001) in the individual compared with controls in the same direction observed in the microarray analysis (Table 1). To validate that TRPC6 haploinsufficiency is leading to transcriptional dysregulation of these genes, we treated a control DPC culture with hyperforin plus flufenamic acid (FFA) and measured the expression levels of the same candidate genes over 48 hours (Figure 2b, c and Figure S1e). Hyperforin specifically activates TRPC6 and FFA increases the amplitude of the currents through this channel ^38–40. If the candidate genes are regulated through TRPC6 signaling, we expect a change in their expression levels opposite to the observed change in the TRPC6-mut individual. After a 48-hour treatment, we observed the expected correlation for five of the eight genes. While the expression levels of SEMA3A, EPHA4, and CLDN11 were significantly reduced (−28-fold, −3.2-fold and −4.76-fold, respectively), MAP2 and INA displayed 20- and 10-fold increases in expression, respectively. These results validate the microarray data and support the hypothesis that the selected genes are regulated by the TRPC6 pathway.

We measured CREB phosphorylation in the DPCs of the individual and a control to assess the functional effect of TRPC6 disruption. Stimulation of DPCs with hyperforin plus FFA induced a significantly reduced level of increased phosphorylated CREB (p-CREB) in the individual’s DPCs (30.3±0.7%) compared to control (48.6±2.3%; P<0.005) after 15 minutes. After 30 minutes, p-CREB levels in the individual’s DPCs (6.3±2.1%) were also significantly lower compared to the control (20.9±2.1%; P<0.05) (Figure 2d and e). Taken together, these results demonstrate that several of the functionally relevant DEGs identified in the microarray studies are controlled via TRPC6 signaling, likely through CREB phosphorylation, suggesting that TRPC6 disruption influences neuronal cell function.

Generation of neural cells from ASD individuals

To further evaluate the effect of TRPC6 haploinsufficiency on neural cell function, we generated iPSCs from DPCs from the ASD individual and two control individuals (Figure S2 and Table S5 and S6). We chose to reprogram DPCs because these cells develop from the same set of early progenitors that generate neurons. Furthermore, the neurons derived from iPSCs generated from DPCs express higher levels of forebrain genes, many of which are implicated in ASD ⁴¹. We fully characterized three clones from each individual and used at least two different clones for follow up experiments. A summary of the clones used for each experiment can be found in Table S6. Neural progenitor cells (NPCs) and cortical neurons from iPSCs were obtained using a modified protocol from our previous publication ⁶. Briefly, iPSC colonies on Matrigel were treated with dorsomorphin under FGF-free conditions until confluence. Pieces of iPSC colonies were grown in suspension for 2–3 weeks as embryoid bodies (EBs) in the presence of dorsomorphin (Figure 3a). The EBs were then dissociated and plated to form rosettes. The rosettes were manually selected and expanded as NPCs (Figure 3a). These NPCs were negative for the pluripotent marker OCT4 and positive for early neural-specific markers such as Musashi-1 and Nestin (Figure 3b and Figure S3a). To obtain mature neurons, NPCs were plated with ROCK inhibitor and maintained for 3–4 additional weeks under differentiation conditions. At this stage, the cells were positive for the pan neuronal marker Tuj1 (β-III-Tubulin) and expressed the more mature neuronal markers synapsin I (SYN1) and microtubule-associated protein 2 (MAP2; Figure 3c). These cells expressed genes typically found in the cortex, including CTIP2, important for the differentiation of subcortical projecting neurons; TBR1, critical for cortical development and ABAT, a marker for GABAergic neurons, encoding for the 4-aminobutyrate aminotransferase protein and responsible for the catabolism of GABA neurotransmitter. (Figure S3a). Expression of NESTIN indicates the presence of NPCs and the expression of S100B and GFAP are indicative of glia cells, suggesting a mixed cell population at this stage (Figure S3a). In our cultures, the presynaptic SYN1 puncta were frequently adjacent to the postsynaptic marker HOMER1, suggesting the presence of developed synapses (Figure S3b). Using immunostainning, we also detected expression of the inhibitory neurotransmitter γ-aminobutyric acid (GABA) in 13% of the neurons, and 22% were positive for vesicular glutamate transporter-1 (VGLUT1), a marker for excitatory neurons, in both controls and ASD subjects (Figure 3c–e). Our protocol generated a consistent population of forebrain neurons, confirmed by the co-localization of pan-neuronal and subtype-specific cortical markers, such as 16% of Ctip2 (Layers V and VI) and 6% of Tbr1 (Layers I and VI; Figure 3d–e). Expression of peripherin and En1, markers for peripheral and midbrain neurons, respectively, was not detected. We did not observe a significant variability in these subtypes of neurons between the control and ASD backgrounds (Figure S3a). Next, we determined the functional maturation of the iPSC-derived neurons using electrophysiological methods. Whole-cell recordings were performed using cells that had differentiated for at least 6 weeks in culture. Both controls and ASD- neurons showed action potentials evoked by somatic current injections (Figure 3f–h and Figure S3c and d). Therefore, our data indicate that somatic cell reprogramming did not affect the ability of iPSC-derived neurons to mature and become electrophysiologically active.

TRPC6 disruption does not affect NPC proliferation

TRPC1, another member of the transient receptor potential channel family, is involved in NPC proliferation mediated by FGF ⁴². Therefore, we investigated whether reduction of TRPC6 expression levels affects the cell cycle profile. No difference was observed when comparing the percentage of cells in G1 (56.2±5.0% and 47.8±10.5%, P>0.2), S (30.6±3.0% and 36.0±6.4%, P>0.2), and G2/M (10.1±1.4% and 14.3±4.8%; P>0.2) phases between control and TRPC6-mut iPSC-derived NPCs, indicating that TRPC6 likely does not play a role in NPC proliferation, in contrast to TRPC1 (Figure S4a).

Ca²⁺ influx is reduced in TRPC6-mutant NPCs

The role of TRPC6 in dendritic spine formation depends on a pathway that involves Ca²⁺ influx through the channel ⁸. To test if changes in intracellular Ca²⁺ levels might be altered in TRPC6-mut neural cells upon TRPC6 activation, we stimulated iPSC-derived NPCs from the TRPC6-mut individual and a control with hyperforin plus FFA. This combination of drugs induced transient and repetitive increases in intracellular Ca²⁺ concentrations in both TRPC6-mut- and control-derived NPCs. The TRPC6 activation-induced Ca²⁺ oscillation peak was significantly higher in control NPCs compared with TRPC6-mut NPCs (Figure 4a). The average amplitude of the Ca²⁺ increase over baseline in the 100 cells analyzed was reduced by 30% in the TRPC6-mut NPCs (1.9±0.08-fold) compared with the control sample (2.7±0.2-fold; P<0.001) when stimulated with hyperforin and FFA (Figure 4b).

TRPC6 signaling regulates gene expression in neuronal cells

To validate our DPC findings, we examined the expression of some neuronal genes in NPCs in response to TRPC6 activation (Figure S4b). After a 48h hyperforin treatment, SEMA3A expression was reduced (0.6±0.05-fold, P<0.05), whereas INA and MAP2 again showed increased expression (2.6±0.09-fold and 1.8±0.1-fold; P<0.001). These results parallel our DPC expression analysis and support the hypothesis that TRPC6 signaling is important for the regulation of genes involved in neuronal function.

TRPC6 disruption alters the neuronal phenotype

To determine if TRPC6 disruption influences spine formation and synaptogenesis, we investigated neurons derived from TRPC6-mut and control iPSCs. To avoid variability from reprogramming, all experiments were performed with different iPSC clones and independent experiments. All biological replicates and iPSC clones used in each experiment are summarized in Table S6. The neurons derived from this ASD individual exhibited a 60% reduction (P<0.01) in TRPC6 protein levels as measured by western blot (Figure 4c and d). We first examined neuron morphology by infecting cells with a previously described lentiviral vector containing the EGFP sequence under the control of the synapsin gene promoter (syn::EGFP) ⁶. By measuring the size of neurites and their ramifications, we verified that the TRPC6-mutant neurons are shorter in total length (1782±101.2 and 2666±153.7 pixels; P<0.001) and less arborized (3.7±0.2 and 8.7±0.5 vertices; P<0.001) than the controls (Figure 4e–g). Moreover, the density of dendritic spines in TRPC6-mutant neurons was reduced (7.4±0.5 spines per 20µm of dendrite length) compared with control neurons (12.9±0.8 spines; P<0.001) derived from several individuals (Figure 4h–j, Figure S3e). TRPC6 expression was previously shown to regulate spine density ⁸. Thus, to confirm that the alterations observed in this ASD individual were caused by TRPC6 haploinsufficiency, we downregulated TRPC6 expression in control neurons using a specific, pre-validated shRNA in a lentiviral vector. Neurons derived from control iPSCs expressing shTRPC6 exhibited a significant reduction in spine density (6.0±0.5 spines) compared with control neurons expressing a scrambled shRNA (12.5±0.7 spines; P<0.0001; Figure 4j). Even further, restoring TRPC6 expression in the TRPC6-mut neurons using a lentiviral vector expressing wild-type TRPC6 (Figure S4c–d) rescued these morphological alterations, increasing total neuronal length (3051±133.4 pixels; P<0.001), arborization (8.9±0.6 vertices; P<0.001), and dendritic spine density (11.9±0.9 spines; P<0.001) to control levels (Figure 4e, f, and h). Interestingly, the specific activation of the wild type TRPC6 in mutant neurons by hyperforin was also sufficient to rescue these morphological phenotypes in our culture conditions (Figure 4e, f and h).

TRPC6 is mainly expressed in glutamatergic synapses and its loss interferes with synapsin I cluster density in pre-synaptic sites of hippocampal neurons, suggesting that this gene has an important role in the regulation of excitatory synapse strength ⁹. Quantifying VGLUT1 puncta in MAP2-labeled neurons confirmed that the TRPC6-mutant neurons had a significantly lower density of VGLUT1 puncta (4.6±0.3 puncta per 20µm of dendrite length) compared with independent clones isolated from several independent controls (10.3±0.4 puncta; P<0.001) (Figure 4k–l, Figure S3g). To determine if TRPC6 haploinsufficiency contributed to the lower density of VGLUT1 puncta, we treated TRPC6-mut neurons with hyperforin to specifically stimulate TRPC6. After 2 weeks of treatment, the neurons exhibited a significant increase in the number of VGLUT1 puncta compared with untreated cells (7.4±0.5 puncta; P<0.05) (Figure 4k). Control neurons expressing shTRPC6 also exhibited a lower density of VGLUT1 puncta, indicating that loss of TRPC6 function affects the formation of glutamatergic synapses (P<0.01) (Figure 4m–n). In addition, overexpression of TRPC6 in the TRPC6-mut neurons was able to rescue synapse numbers (8.0±0.6 puncta per 20µm of dendrite length; P<0.001) to control levels, as measured by synapsin I puncta (Figure 4o–p). Finally, electrophysiological recordings revealed that the Na⁺ currents of TRPC6-mutant neurons (28.38±7.5 pA) were impaired compared to controls (154.4±45.9 pA; P<0.0001) (Figure 4q–r, Figure S3e).

TRPC6 and MeCP2 share a similar molecular pathway

Certain neuronal phenotypes (reduction of spine density and glutamatergic synapses) associated with TRPC6 function loss are similar to those previously described for loss of MeCP2 function in human neurons ⁶. MeCP2 genetic alterations have been recognized in several non-syndromic ASD individuals ^43–50, and reduced MeCP2 expression in brains of autistic individuals has been reported ^{51, 52}. In addition, two independent studies have reported that MeCP2 regulates TRPC6 expression in the mouse brain, likely through an indirect mechanism ^{53, 54}. Thus, we investigated whether MeCP2 acts upstream of TRPC6 in human neurons. We used two iPSC clones from a female RTT patient carrying the T158M MeCP2 mutation, which results in persistent X chromosome inactivation ⁶. Each clone expresses a different MeCP2 allele, a wild type or mutant version of the MeCP2 gene. We then differentiated both clones into neurons and evaluated TRPC6 protein expression levels. The TRPC6 expression level was reduced by 40% in the clone carrying the non-functional version of MeCP2 compared to the wild type control clone (61.67±6.0% and 99.3±1.2%; P<0.01), indicating that MeCP2 levels affect TRPC6 expression in human neurons (Figure 4s). This observation supports MeCP2 acting upstream of TRPC6 in the same molecular pathway to affect neuronal morphology and synapse formation. We next investigated whether MeCP2 could occupy regions of the human TRPC6 promoter. Chromatin immunoprecipitation (ChIP) followed by quantitative PCR (qPCR) revealed high levels of MeCP2 in association with the TRPC6 promoter region in human neurons, suggesting a potential mechanism of transcriptional regulation (Figure 4t, Figure S4f).

Our data suggest that the molecular pathway involving MeCP2 and TRPC6 is a rate-limiting factor in regulating glutamatergic synapse number in human neurons. Administration of insulin-like growth factor-1 (IGF-1) promotes the reversal of RTT-like symptoms in a mouse model ⁵⁵ and of molecular alterations in RTT human neurons ⁶, and is currently in clinical trials for RTT patients ⁵⁶. To investigate whether the potential convergence of molecular mechanisms underlying RTT and non-syndromic autism suggests shared therapeutic benefits, we treated TRPC6-mutant neurons with full-length IGF-1 (10 ηg/mL). Interestingly, we observed a significant increase in TRPC6 protein levels after treatment. Moreover, Psd-95 and synapsin I protein levels were also upregulated by IGF-1 (P<0.01; Figure 4c and d). IGF-1 treatment also rescued the glutamatergic synapse number in TRPC6-mutant neurons as measured by VGLUT1 puncta (9.2±0.6 puncta per 20µm of dendrite length; P<0.01), suggesting that the drug treatment could correct this neuronal phenotype (Figure 4k).

TRPC6 downregulation compromises neuronal development in vivo

In vitro experiments in rodent primary neurons have shown that Trpc6 levels affect spine density and excitatory synapses ^{9, 57}. To corroborate our findings from human derived neurons, we looked to examine the effect of Trpc6 loss in a rodent model. We validated two shRNAs (#1 and #3) against mouse Trpc6 by western blot analysis and used both for further experiments (Figure S4g–h). Using this shRNA targeting Trpc6, we transduced mouse primary hippocampal neurons. The neurons expressing shRNA targeting Trpc6 demonstrated reduced spine density (7.5±0.5 and 11.4±0.8 spines per 20µm of dendrite length; P<0.001) and fewer synapses (5.6±0.5 and 7.7±0.5 puncta per 20µm of dendrite length; P<0.01) versus neurons transduced with a shRNA scramble control (Figure 5a and b). Thus, as described above, we determined that TRPC6 downregulation causes similar neuronal alterations in human and rodent neurons. We next looked to validate the cell autonomous effect of TRPC6 loss of function in vivo by taking advantage of adult neurogenesis in the hippocampus ⁵⁸. Using retroviruses to target newborn neurons, we delivered the shRNAs against mouse Trpc6. Trpc6 downregulation led to migration defects and reduced neuronal dendritic arborization (Figure 5c–f). Moreover, whole-cell patch clamping to record action potentials revealed a significant reduction in the firing rate of neurons expressing shRNAs against Trpc6 compared with controls (Figure 5g–i). To demonstrate the contribution of Trpc6 to these phenotypes in vivo, we rescued the migration defects by co-transfecting the shRNA with an expression construct for an shRNA-resistant form of TRPC6-WT (TRPC6-WT^R; Figure 5d, Figure S4h).

TRPC6 knockout (KO) mice ⁵⁹ display reduced exploratory activity in a square open field and elevated star maze compared with control siblings ⁶⁰. Limited environmental exploration is commonly associated with ASD individuals ⁶¹. Thus, we decided to investigate whether the TRPC6 KO mouse displays other ASD-like behaviors. We assessed the social interaction and repetitive behaviors of these animals, but observed no significant differences between wild-type controls (WT) and heterozygotes (HET) or WT and KO mice (Figure 5j). Together, these data demonstrate loss or reduction of TRPC6 in a rodent model induces neuronal abnormalities paralleling our findings in the TRPC6-mut iPSC-derived neurons, such as reduced neuronal arborization, spine density and synapse numbers.

Mutation screening of TRPC6

Based on the initial observation of TRPC6 disruption by a chromosomal breakpoint, we established a narrow hypothesis focusing on TRPC6 to conduct a single gene case/control association study. We screened targeted high-throughput sequencing data from all coding exons and splice sites of TRPC6 in 1041 ASD cases from the Simons Simplex Collection (SSC) ²⁶ and 942 ancestrally matched controls from the NINDS Neurologically Normal Caucasian Control Panel (http://ccr.coriell.org/Sections/Collections/NINDS/). A summary of the quality control metrics of the high-throughput sequencing is presented in Table S7. We focused on novel splice sites, missense, and nonsense mutations that were observed only once across all of our cohorts and not present in the dbSNP137 and 6503 exomes available from the Exome Variant Server (EVS, v.0.0.15). We reasoned that these variants were most likely to be deleterious and subject to purifying selection. Moreover, the study of variants observed only once, in combination with case-control matching for ancestry, represents a more rigorous approach to protecting against population stratification ⁶². Table S8 lists all such variants in TRPC6. We observed significantly more novel nonsynonymous singleton mutations in cases compared with controls (10/1041 cases versus 1/942 controls; p = 0.013, OR = 9.127, 95% CI = 1.211-191.027, Fisher exact test, two-tailed). To confirm the low mutation rate observed in this control sample, we examined the whole exome-sequencing data from an in-house database and identified an additional 1930 northern European (NE) controls who clustered tightly with the HapMap CEU cohort. We evaluated the coding exons and splice sites of TRPC6 and, to maximize sensitivity, did not set a minimum read threshold to identify all novel nonsynonymous singleton variants, which are listed in Table S5. An omnibus analysis revealed an even more significant over-representation of such variants in cases (10/1041 cases versus 4/2872 controls; p = 0.001, OR = 6.954, 95% CI = 2.008-26.321, Fisher exact test, two-tailed). Because our results indicate that TRPC6 disruption leads to haploinsufficiency of the corresponding protein, two of the case variants are particularly noteworthy: M1K, which disrupts the start codon; and Q3X, which is a very early premature stop codon. Unfortunately, live cells from these individuals were not available for follow-up functional studies. No TRPC6 mutations affecting the start codon or nonsense mutations were identified in a total of 7445 controls: 942 NINDS neurologically normal Europeans and 6503 exomes from the EVS (4300 European-American, 2203 African-American). Segregation analysis of the case variants revealed that each was inherited from an apparently unaffected parent, suggesting that these variants are incompletely penetrant, as has been previously observed for a wide range of ASD mutations such as Shank3 ⁶³ and CNTNAP2 ⁶⁴. Thus, although these genetic variations cannot be considered as causal mutations, they might represent risk factors for ASD. No TRPC6 CNVs have been described in ASD (http://projects.tcag.ca/autism_500k).