Synaptic dysfunction in human neurons with autism-associated deletions in PTCHD1-AS

Abstract

BACKGROUND:

The Xp22.11 locus that encompasses PTCHD1, DDX53, and the long noncoding RNA (lncRNA) PTCHD1-AS is frequently disrupted in males with autism spectrum disorder (ASD), but the functional consequences of these genetic risk factors for ASD are unknown.

METHODS:

To evaluate the functional consequences of PTCHD1 locus deletions, we generated induced pluripotent stem cells (iPSCs) from unaffected controls and three ASD subjects with microdeletions affecting PTCHD1-AS/PTCHD1, PTCHD1-AS/DDX53, or PTCHD1-AS alone. Function of iPSC-derived cortical neurons was assessed using molecular approaches and electrophysiology. We also compiled novel and known genetic variants of the PTCHD1 locus to explore the roles of PTCHD1 and PTCHD1-AS in genetic risk for ASD and other neurodevelopmental disorders. Finally, genome editing was used to explore the functional consequences of deleting a single conserved exon of PTCHD1-AS.

RESULTS:

iPSC-derived neurons from the ASD subjects exhibited reduced miniature excitatory post-synaptic current (mEPSC) frequency and NMDA receptor hypofunction. We found that 36 ASD-associated deletions mapping to the PTCHD1 locus disrupt exons of PTCHD1-AS. We also report a novel ASD-associated deletion of PTCHD1-AS exon 3, and we show exon 3 loss alters PTCHD1-AS splicing without affecting expression of the neighboring PTCHD1 coding gene. Finally, targeted disruption of PTCHD1-AS exon 3 recapitulated diminished mEPSC frequency, supporting a role for the lncRNA in the etiology of ASD.

CONCLUSIONS:

Our genetic findings provide strong evidence that PTCHD1-AS deletions are risk factors for ASD, and human iPSC-derived neurons implicate these deletions in the neurophysiology of excitatory synapses and in ASD-associated synaptic impairment.

INTRODUCTION

Autism spectrum disorder (ASD) is a common neurodevelopmental disorder that is characterized by impaired social interactions and repetitive, inflexible behaviors (1). Presentation and severity of ASD features varies widely between individuals, suggesting etiological heterogeneity. Genetic factors play an important role in the development of ASD, and rare genetic variants in protein coding genes have implicated altered synaptic function in ASD development (2–7). However, much remains unknown regarding the functional consequences of ASD-associated genetic risk factors and their effects on neuronal circuitry.

Induced pluripotent stem cell (iPSC) technology enables production of neurons that are genetically matched to people with ASD, and can be used to identify ASD-associated neuronal phenotypes (8). Neurotransmitter release was reduced in human neurons with heterozygous mutation of NRXN1 (9). SHANK3 haploinsufficient neurons had impairments in dendrite complexity (10) and synaptic function (11). iPSC-derived neurons from ASD subjects exhibited changes in dendritic morphology and formed fewer synapses (12, 13). iPSCs from people with idiopathic ASD and macrocephaly overproduced inhibitory GABAergic neurons (14). iPSC technology therefore enables functional sub-classification of ASD risk genes with respect to their effects on synapse function and neuronal circuitry, which could facilitate design and interpretation of clinical trials for ASD therapeutics.

Genetic variants of the “PTCHD1 locus” on chromosome Xp22.11 are among the most common and penetrant genetic risk factors for ASD and other neurodevelopmental disorders but the functional consequences of these variants remain unknown (3, 15). Although females can carry PTCHD1 locus microdeletions with no obvious damaging effects, these deletions are highly penetrant ASD risk factors in males (15, 16) and account for <1% of ASD cases (3). PTCHD1 (Patched domain containing 1) encodes a transmembrane protein with a patched domain, and its involvement in neurodevelopmental disorders is supported by microdeletions and frameshift mutations in individuals with neurodevelopmental delay (NDD), intellectual disability (ID), and ASD (15–21). However, recently described Ptchd1 mutant mice had impairments in attention and cognition (22–24), but did not overtly exhibit ASD-associated behaviors. Also, many ASD-associated PTCHD1 locus microdeletions are upstream of the PTCHD1 protein-coding gene and disrupt exons of the neighboring brain-enriched lncRNA PTCHD1-AS (PTCHD1 antisense RNA) (15). Some upstream deletions also encompass the protein-coding gene DDX53 (DEAD-box helicase 53), but this gene reportedly has limited expression in the brain (15, 17). The goal of this study was to evaluate the effects of PTCHD1 locus deletions on neuronal circuitry, and to explore the roles of PTCHD1 locus genes in mediating the cellular phenotypes that we observed.

Here we link upstream genomic rearrangement of the PTCHD1 locus with ASD and employ iPSCs and genome editing to determine the functional consequences of PTCHD1 locus deletions in human neurons. We generated iPSCs from controls and three males with ASD and deletions of the PTCHD1 locus. iPSC-derived neurons from the ASD subjects exhibited similar impairments in excitatory synaptic function, and synaptic impairment was also observed in neurons with engineered disruption of PTCHD1-AS. Our findings strengthen the connection between synaptic dysfunction and ASD and argue that disruption of PTCHD1-AS is a compelling ASD risk factor.

MATERIALS AND METHODS

Induced pluripotent stem cells

iPSC work was approved by the CIHR Stem Cell Oversight Committee. iPSCs were generated from dermal fibroblasts or from CD34⁺ blood cells, which were obtained with informed consent at The Hospital for Sick Children, with SickKids Research Ethics Board approval. Fibroblasts were reprogrammed with retrovirus vectors (25) and characterized/cultured (26, 27) as described. Blood cells were reprogrammed with Sendai virus and characterized at the Centre for the Commercialization of Regenerative Medicine. Teratoma experiments were approved by SickKids Animal Care Committee and complied with guidelines of the Canadian Council on Animal Care. For details, see the Supplemental Methods.

Neuronal differentiation

Neuronal differentiation procedures have been described (27, 28). For analyses of neuronal synaptic connections we co-cultured iPSC-derived neurons with exogenous astrocytes (9, 10, 11). For AMPAR-mEPSC recordings NPCs were seeded on human astrocytes (ScienCell) or mouse astrocytes (29) and allowed to mature for 12-16 weeks. For morphometric analyses, neurons were pre-differentiated for 3 weeks, dissociated, reseeded on coverslips with mouse astrocytes, and analyzed 5 weeks later. For some gene expression analyses, NPCs and astrocytes were depleted by magnetic cell sorting (MACS) as described (30) and neurons were reseeded on Matrigel (Corning). For details, see the Supplemental Methods and Table S1.

RNA analyses

RNA was harvested using TRIzol and reverse transcribed using SuperScript II or III (Thermo-Fisher Scientific). Primers are listed in Table S1. RNA fractionation (31, 32) and half-life assays (33) were performed using published protocols. For details, see the Supplemental Methods.

Immunocytochemistry and imaging

Antibodies for immunocytochemistry are listed in Table S3. Excitatory synapses were quantified as overlapping SYN1/HOMER1-punctae/10μm of MAP2⁺ dendrite (13, 34) in confocal Z stacks. Dendrites in individual neurons were labeled by low-efficiency transfection of the plasmid pL-SIN-EF1a-using Lipofectamine 2000 (Thermo-Fisher Scientific). Measurements of total dendrite length and complexity were performed using the Simple Neurite Tracer plugin for ImageJ. For details, see the Supplemental Methods.

Microarrays and RNA-seq

Copy number variations (CNVs) (35) and RNA expression (36) were analyzed using microarrays as described. Data are deposited in GEO (www.ncbi.nlm.nih.gov/geo/): accession GSE83089 (CNVs), GSE81624 (expression microarray), GSE123753 (high coverage RNA-seq), and GSE1129808 (ASD-70 RNA-seq). For details, see the Supplemental Methods.

Patch-clamp recordings in human iPSC-derived neurons

Whole-cell patch-clamp recordings were performed at room temperature in human iPSC-derived neurons – cultured with human or mouse astrocytes (12-16 weeks old) or without astrocytes (8 weeks old) – as described (28). For details, see the Supplemental Methods and Table S1.

iPSC genome editing

Genome editing was performed as described (37) in iPSCs from the unaffected male to replace PTCHD1-AS exon 3 (ex3) with two tandem polyadenylation sequences (38). Oligonucleotide sequences are in Table S2. For details, see the Supplemental Methods.

Statistical analyses

Statistical analyses were primarily performed using GraphPad Prism software, with N being the number of biological replicates from 2-4 independent experiments. Biological replicates were defined as individual neurons (electrophysiology/dendrites), coverslips (synapses), or cultures (qRT-PCR). N is typically displayed with the format A/B (A = biological replicates / B = iPSC lines). Normally distributed data were displayed as mean ± standard error of the mean (SEM) and analyzed using parametric statistical tests. Non-normal data were displayed as median ± 95% confidence interval (CI) and analyzed using nonparametric statistical tests. For details, see the Supplemental Methods.

RESULTS

Generation of iPSCs for functional analyses of PTCHD1 locus deletions

To test the functional consequences of PTCHD1 locus deletions, we generated iPSCs from two male probands with ASD (Figure 1A and S1A). Proband 1 (Prb1) has a deletion that eliminates the promoters and first exons of PTCHD1 and PTCHD1-AS (15). Proband 2 (Prb2) has a deletion that is upstream of PTCHD1, and eliminates the conserved (15) third exon of PTCHD1-AS and DDX53. As controls, we used iPSCs from the unaffected mother of Prb1 (28) and from an unaffected, unrelated male (39–41) (Table S4 and Figure S1A).

Figure 1. Expression of PTCHD1 locus transcripts in iPSC-derived neurons.

(A) Schematic representation of the PTCHD1 locus, indicating the locations of transcripts and the ASD-associated microdeletions of probands (Prb) 1 and 2.

(B) iPSCs were differentiated to NPCs (left) and 6 week-old neurons (right), which were analyzed by immunocytochemistry with the indicated antibodies and counterstained with DAPI. Scale bars: 50 μm (left) and 20 μm (right).

(C) Gene expression was analyzed in NPCs or 3 week-old neurons by qRT-PCR with the indicated primer sets (normalized to GAPDH). Data display the mean and SEM. N=biological-replicates/iPSC lines. *P < 0.05, **P < 0.01, ****P < 0.0001, t-test.

(D) Gene expression was analyzed by qRT-PCR in 2-8 week-old female control iPSC-derived neurons that were left unstimulated or were stimulated for 6h with 55mM KCl (expression was normalized to ACTB). *P < 0.05, t-test.

iPSCs were thoroughly characterized (Table S4). PTCHD1 was expressed in most iPSC lines derived from the unaffected mother of Prb1 (Figure S1B), who is a carrier of the PTCHD1 microdeletion. Endonuclease accessibility analyses revealed clonal X chromosome inactivation in PTCHD1-expressing female lines (Figure S1C), suggesting that the X chromosome carrying the PTCHD1 deletion was inactivated. iPSCs synthesized pluripotency-associated proteins (Figure S1D) and were functionally pluripotent in embryoid body and teratoma assays (Figures S1E and S1F). Although PTCHD1-null iPSC lines tended to have abnormal karyotypes (Table S5), analyses of karyotype (Figure S2A) and CNVs (Figure S2B) revealed that lines chosen for modeling ASD were normal and exhibited no marked genetic instability.

Expression of PTCHD1 and PTCHD1-AS in functional human neurons

PTCHD1 and PTCHD1-AS are expressed in the cortex and cerebellum (15), and cortical PTCHD1 expression coincides with synaptogenesis and is enriched in deep layer excitatory neurons (42). iPSCs were differentiated into neural precursor cells (NPCs) and cortical neurons (Figure 1B) using our neural rosette-based protocol (27). ASD subjects and controls exhibited similar expression of FOXG1 and PAX6 in NPCs (Figure 1C), and immunocytochemistry revealed no significant change in the number of PAX6-positive NPCs (Figure S3A and S3B). PTCHD1-AS was nearly undetectable in NPCs, but was expressed in neurons (Figure 1C). Neurons from Prb1 did not express PTCHD1-AS, and neurons from Prb2 expressed the preserved PTCHD1-AS2 exon 1, but not the deleted ex3 (Figure 1C). In addition to steady-state gene expression, we also tested whether these iPSC-derived neurons could support activity-dependent transcription (43). Neuronal depolarization with KCl resulted in a ~2-fold induction of PTCHD1 and a ~5-fold induction of the positive control BDNF (Figure 1D), suggesting that – like other genes implicated in neurodevelopmental disorders (44) – PTCHD1 expression is stimulated by neuronal activity.

Having determined that neurons express PTCHD1 and PTCHD1-AS, we next verified that these neurons were functional. Neurons from controls and both ASD subjects generated spontaneous (Figure 2A) or evoked (Figure 2B) action potentials, and electrophysiological recordings revealed no substantial differences in intrinsic membrane properties in proband neurons (Table S6). These neurons also formed structural excitatory synapses (Figure 2C and S4A). Synapse quantification revealed no difference for Prb1 and a ~40% increase in the number of excitatory synapses in Prb2 compared to the unaffected controls (Figure 2D and S4B). Generation of action potentials and formation of synapses indicate that these neurons are suitable for modeling synaptic function (and dysfunction) in ASD.

Figure 2. Synaptic function is impaired in neurons from ASD subjects with PTCHD1-AS microdeletions.

(A) Representative traces showing that iPSC-derived neurons (14-16 weeks old) generate spontaneous action potentials.

(B) Typical traces showing evoked action potentials in iPSC-derived neurons (8 weeks old).

(C) Excitatory synapses were visualized in 8 week-old iPSC-derived neurons as overlapping punctae of SYN1 (magenta; pre-synaptic marker) and HOMER1 (green; post-synaptic marker of excitatory neurons) on MAP2 (blue)-positive dendrites (scale bars: 5 μm). Overlapping magenta and green punctae are white in appearance. Images from the blue, green, and magenta channels are displayed individually in Figure S3C.

(D) Quantification of excitatory synapse density in iPSC-derived neurons. Excitatory synapses were detected by immunocytochemistry with anti-SYN1 (pre-synaptic marker) and anti-HOMER1 (post-synaptic marker of excitatory synapses). Graphs display median and 95% CI. N=coverslips/iPSC lines (9 dendrite segments per coverslip, 2 coverslips per line, 2 replicate experiments). **P < 0.01, Kruskal-Wallis with Dunn’s posthoc test. Ctrl displays pooled data from the male and female controls, which are displayed separately in Figure S4B.

(E) Representative traces showing AMPAR-mEPSCs in 14-16 week old neurons. Inset displays average mEPSCs in neurons from the left panel.

(F) Scatter plots displaying all data points for the frequency (top) and amplitude (bottom) of AMPAR-mEPSCs. Graphs also display median and 95% CI. N = neurons/iPSC lines. ***P < 0.001, Kruskal-Wallis with Dunn’s posthoc test. Ctrl displays pooled data from the male and female controls, which are displayed separately in Figure S4C.

(G) Quantification of total dendrite length in neurons from controls and ASD probands. Graphs display mean, SEM, and all data points. N = neurons/iPSC lines. Ctrl displays pooled data from the male and female controls, which are displayed separately in Figure S4E.

(H) Dendrite complexity of iPSC-derived neurons was determined by Sholl analysis. Graphs display the mean number of dendrite crossings (and SEM) at the indicated distances from the soma of neurons that were analyzed in Figure 3G. Ctrl displays pooled data from the male and female controls, which are displayed separately in Figure S4F.

Excitatory synaptic function is decreased in ASD subject neurons

To examine functional excitatory synapses, we measured miniature excitatory post-synaptic currents (mEPSCs), which are primarily mediated by AMPA receptors (AMPARs) and NMDA receptors (NMDARs) (45). We first examined AMPAR-mEPSCs (Figure 2E) by blocking NMDARs with AP-5. We observed a ~50% decrease in the frequency of AMPAR-mEPSCs in neurons from both Prb1 and Prb2 (Figures 2E and S4C), with no change in amplitude compared with controls. To explore a potential mechanism for decreased synaptic activity we examined dendrite morphology (Figure S4D), which revealed no difference in total dendrite length (Figures 2G and S4E) and only subtle changes dendrite complexity (Figure 2H and S4F). Therefore, neurons with deletions of the PTCHD1 locus exhibit decreased excitatory synaptic activity without any change in dendrite morphology.

NMDAR function is impaired in ASD subject neurons

NMDARs play a key role in synaptic transmission and plasticity (46), and they have also been implicated in ASD (47). To examine NMDAR function, we performed whole-cell voltage-clamp recordings, which revealed a ~30-50% decrease in NMDA-evoked current amplitude in proband neurons compared with controls (Figure 3A and 3B), with no change in the reversal potential of the currents. NMDARs have a property of voltage-dependent blockade by extracellular Mg²⁺ (48), and loss of this blockade may contribute to neurodevelopmental disorders (49). Extracellular Mg²⁺ led to a voltage-dependent blockade in NMDA-evoked currents in control and proband neurons (Figure 3C), and NMDA-evoked currents were reduced by ~45-55% in proband neurons held at +60 mV in the presence of extracellular Mg²⁺ (Figure 3D). Taken together, decreases in both the frequency of AMPAR-mEPSCs and the amplitude of NMDA-evoked currents suggest pronounced impairments in excitatory neurotransmission in neurons from ASD subjects with PTCHD1 locus deletions.

Figure 3. NMDAR function is impaired in neurons from ASD subjects with PTCHD1-AS microdeletions.

(A) Representative traces displaying NMDA-evoked currents recorded at membrane potentials of −60 mV in iPSC-derived neurons (8 week-old) in the absence of extracellular Mg²⁺ (left). A plot showing current-voltage relationship of NMDA-evoked currents in the three neurons displayed in the left panel, recorded in the absence of extracellular Mg²⁺ (right).

(B) Scatter plots displaying all data points for NMDA-evoked currents recorded at −60 mV in the absence of extracellular Mg²⁺. Graphs also display mean and SEM. N=neurons/iPSC lines. *P < 0.05; ***P < 0.001, one-way ANOVA with Dunnett’s posthoc test.

(C) Representative traces of NMDA-evoked currents from 8 week-old iPSC-derived neurons, recorded at membrane potentials from −60 mV to +60 mV in the presence of extracellular Mg²⁺ (1 mM).

(D) Current-voltage relationship of NMDA-evoked currents recorded in the presence of extracellular Mg²⁺ at 1 mM. Graphs display mean and SEM. N=neurons/iPSC lines. *P < 0.05; **P < 0.01, one-way ANOVA with Dunnett’s posthoc test, compared at +60 mV.

PTCHD1-AS is an ASD candidate gene

Our findings indicate that neurons from both ASD subjects displayed similar synaptic phenotypes, and PTCHD1-AS is the only gene disrupted in both individuals (Figure 1A). Therefore, we explored the structure and regulation of the PTCHD1-AS gene. The PTCHD1-AS lncRNA is spliced into at least 3 known variants (PTCHD1-AS1, PTCHD1-AS2, and PTCHD1-AS3) (15) and is divergently transcribed away from the PTCHD1 protein-coding gene (Figure 4). The NCBI reference transcript for PTCHD1-AS (NR_073010.2, March 2016) includes all PTCHD1-AS2 exons and additional downstream exons (Figure 4), although only exons 1-5 are deleted in ASD cases (15, 17).

Figure 4. Genetic evidence for PTCHD1-AS as an ASD candidate gene.

Displayed are PTCHD1 locus transcripts and published, deposited, or novel genetic variants that disrupt genes within the locus. NR_073010.2 is the NCBI reference transcript for PTCHDI-AS. PTCHDI-ASI, -AS2, and -AS3 were previously described by Noor et al (2010). PTCHDI-AS2 and -AS3 TSSs are indicated as promoters. PTCHDI-ASI is likely a truncated partial transcript, and the 5’ end of the deposited sequence is too short to be unambiguously mapped (indicated by ‘?’ symbol). Hashed grey vertical lines correspond to exons of interest from PTCHD1 locus transcripts; hashed magenta lines correspond to novel exons of PTCHD1-AS reported in this study. The hashed red line indicates the location of the PTCHD1-AS2 transcription TSS, which is 20 kb away from the nearest PTCHD1 deletion. Genetic variants that spare this TSS and all downstream exons are predicted to leave PTCHD1-AS transcripts preserved. All displayed genetic variants were identified in males. Aside from frameshift mutations (indicated by *) and four duplications, all displayed variants are deletions. NDD/ID indicates individuals with neurodevelopmental delays and/or intellectual disability who were not diagnosed with ASD.

To gain insight into the consequences of PTCHD1-AS deletions, we mined publicly available genomics data. CAGE-seq data (50) revealed two transcriptional start sites (TSS) for PTCHD1-AS, which are separated by ~40 kb and map to the first exons of PTCHD1-AS2 and PTCHD1-AS3 (Figure 4 and Figure S5A). ChIP-seq data (51) from human neurons suggest that the PTCHD1-AS3 TSS shares a bi-directional promoter with PTCHD1 (Figure S5B) that bears histone markers of both promoters (H3K4me3) and enhancers (H3K27Ac, H3K4me1) (Figure S5C). The second TSS corresponds to PTCHD1-AS2 and is associated with enhancer markers. Together these data indicate that PTCHD1-AS transcription is initiated at two spatially distinct sites (Figure 4).

To survey the evidence for PTCHD1-AS as an ASD candidate gene we compiled 75 published and unpublished genomic variants of the PTCHD1 locus from males (Figure 4 and Table S7), 71 of which are associated with ASD and/or NDD/ID (6, 15, 17, 20, 21). We found 14 variants that truncate or delete PTCHD1 while preserving PTCHD1-AS (i.e., at least one promoter and all downstream exons), and 2 of these individuals had ASD (Figure 4). However, 51 deletions that eliminated both promoters or disrupted exons of PTCHD1-AS2 were found, and 35 of these individuals had ASD or ASD features (Figure 4). These findings show that 69% of variants that disrupt PTCHD1-AS are associated with ASD or ASD features, in contrast to 14% of variants that exclusively interrupt PTCHD1. These findings suggest disruption of PTCHD1-AS is more likely to confer ASD risk than disruption of PTCHD1 (p = 0.0005, Fisher’s exact test).

We also identified several novel or unreported microdeletions. One noteworthy novel deletion disrupted PTCHD1-AS in three brothers with ASD (Figure 4: 8257_001, 8257_004, and 8257_005) and partially overlaps the deletion found in Prb2 (Figure 4: 5298_3). Both of these deletions encompass the third exon of the annotated PTCHD1-AS transcript (Figure 4), which is evolutionarily conserved and is expressed in syntenic transcripts from rodents (15). Finally, we found PTCHD1-AS deletions in four male general-population controls from databases (Figure 4), although none of these individuals would have been assessed for ASD. Together these refined genetic data build upon our previous findings (15) and strongly suggest that PTCHD1-AS deletions are ASD risk factors.

PTCHD1-AS is alternatively-spliced and localized to the nucleus

To gain insights into the molecular basis of synaptic phenotypes in proband neurons we characterized splicing and localization of PTCHD1-AS RNA. Transcripts arising from exons 1-6 of the >1 Mb PTCHD1-AS gene were alternatively spliced, ranged in size from 320-827 nucleotides (Figure S6A and Table S8), and included several undescribed exons (Figure 4 and Table S9). In control neurons ex3 was detected in most cloned splice variants from both PTCHD1-AS2 and PTCHD1-AS3, but exon 4 was absent (Figures 5A, S6A, and S6B). In contrast, the only consistent difference in exon usage in Prb2 neurons (which lack ex3) was inclusion of the fourth exon (Figure 5A). When analyzed by subcellular fractionation (31), control transcripts OIP5-AS1 and MALAT1 (52) were primarily localized in the cytoplasm and nucleus, respectively (Figure 5B). PTCHD1-AS was primarily found in the nucleus – and mostly chromatin-associated – in neurons from both the control and from Prb2 (Figure 5B). Enrichment in the chromatin fraction was not due to transcript instability (53), as PTCHD1-AS had a half-life of >5h (Figure S7). Therefore, PTCHD1-AS is alternatively spliced and localizes to the nucleus, and loss of ex3 leads to a shift in splicing patterns without a change in subcellular localization.

Figure 5. PTCHD1-AS ex3 is essential for normal excitatory synaptic function in gene-edited iPSC-derived neurons.

(A) Heat map of exon usage in cloned PTCHD1-AS2 splice variants from the unaffected female control and from Prb2.

(B) Relative abundance of transcripts in the cytoplasm, nucleoplasm, and chromatin fractions of 7-8 week-old neurons from the unaffected female control and from Prb2 (N=biological replicates from 1 line each).

(C) qRT-PCR analysis of gene expression (normalized to GAPDH/18s) in MACS-enriched neurons (4 weeks old). The sample labelled Astro/NPC consisted of that were cells enriched by MACS with antibodies targeting CD44 and CD184. N = biological-replicates/iPSC lines. Graphs display mean and SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ANOVA with Tukey posthoc test.

(D) Hierarchical clustering of differentially expressed genes in 4 week-old MACS-enriched neurons from unaffected controls and ASD subjects. Displayed data are from multiple iPSC lines from two independent experiments (a, b).

(E) Representative traces showing AMPAR-mEPSCs from 14-16 week old genome-edited iPSC-derived neurons and isogenic controls (Iso ctrl). Inset displays average mEPSCs from the left panel.

(F)Quantification of frequency (left) and amplitude (right) of AMPAR-mEPSCs from CRISPR-edited neurons and isogenic controls. All data points are displayed along with median and 95% CI. N is the total number of neurons analyzed from two independent experiments. *P < 0.05, ***P < 0.001, Mann-Whitney U test.

(G) Representative traces showing AMPAR-mEPSCs from 12-13 week old iPSC-derived neurons from ASD-70 and controls. Inset displays average mEPSCs from the left panel.

(H)Quantification of frequency (left) and amplitude (right) of AMPAR-mEPSCs from iPSC-derived neurons from ASD-70 and unaffected controls 68 and 50. All data points are displayed along with median and 95% CI. N is the number of neurons/iPSC lines from three independent experiments. *P < 0.05, ns: P > 0. 5, Mann-Whitney U test.

PTCHD1 expression is not affected in cis by PTCHD1-AS ex3 deletion

Some nuclear lncRNAs act in cis to regulate expression of neighboring genes (54), so we tested whether deletion of PTCHD1-AS ex3 in Prb2 affected expression of the neighboring protein coding gene PTCHD1 (Figure 4). qRT-PCR analyses of MACS-enriched neurons revealed similar robust expression of the neuronal marker TUBB3 and low levels of the astrocyte marker GFAP (Figure 5C). PTCHD1-AS was not expressed in Prb1 and lacked ex3 in Prb2 (Figure 5C). DDX53 was detectable by qRT-PCR in control neurons (Figure 5C). When we mined high-coverage RNA-seq from MACS-enriched control iPSC-derived neurons we detected PTCHD1-AS ex3 in all replicates. However, reads mapping to DDX53 did not span its full exon and had different distributions in each replicate (Figure S8A), suggesting that DDX53 is expressed at low levels. Genetic deletion of DDX53 in Prb2 eliminated its expression (Figure 5C), but the intact DDX53 gene was not expressed in Prb1 neurons, suggesting that transcription through the PTCHD1-AS gene may promote DDX53 expression. Neurons from Prb2 expressed PTCHD1 levels that were nearly identical to unaffected controls (Figure 5C). Similarly, we found no difference in KCl-induced expression of PTCHD1 in Prb2 neurons (Figure S8B). Therefore, synaptic phenotypes in Prb2 neurons are not mediated by misregulation of PTCHD1 mRNA expression.

Global gene expression is subtly affected by PTCHD1 locus deletions

To test for potential trans effects (54) of PTCHD1-AS we examined global gene expression in MACS-enriched neurons from multiple control and ASD subject iPSC lines (Tables S1 and S4). Hierarchical clustering of microarray data showed that gene expression patterns segregated by clinical presentation, experimental subject, and iPSC line (Figure 5D). Few genes were misregulated in common in the two ASD subjects: only 18 transcripts from 14 genes exhibited a fold-change of >1.25 and achieved statistical significance in both ASD subjects (Table S10), although none of these genes have known neuronal functions. We also detected no consistent expression changes in select genes with known roles in hedgehog signaling, regional specification of neurons, or neurotransmitter subtypes (Table S10). Therefore, gene expression is only subtly affected in iPSC-derived neurons with PTCHD1 locus deletions.

Deletion of PTCHD1-AS ex3 impairs synaptic function

Our genetic and electrophysiological data implicate PTCHD1-AS in neuronal function. However, Prb1 has a deletion of PTCHD1, and concurrent whole genome sequencing of blood DNA from Prb2 revealed a de novo missense variant in the ASD risk gene SHANK2 (p.A1352T) (6), which may also potentially contribute to the phenotype.

To further test the consequence of PTCHD1-AS disruption in neurons we used CRISPR/Cas9 genome editing in iPSCs from the unaffected male. Guided by functional data from Prb2 and supporting genetic data from a novel ASD-associated microdeletion (Figure 4: 8257_001, _004, and _005), we focused on PTCHD1-AS ex3 (Figures S9A). We designed CRISPR guide RNAs to cut within ex3 (Figure S9B), facilitating homologous recombination to replace ex3 with two tandem polyadenylation sequences (Figure S9C) with the intent of prematurely terminating transcription of PTCHD1-AS. To complement this isogenic pair of lines, we also obtained blood for reprogramming from the 8257 family with the ex3 microdeletion that leaves DDX53 intact (Figure 4). iPSCs were generated and characterized from this ASD subject (8257_005, ASD-70) and his unaffected father (8257_001, Ctrl 68) (Table S11).

To examine expression of genes in the PTCHD1 locus we differentiated the CRISPR-edited line into neurons. These neurons did not express PTCHD1-AS ex3 as expected, but also exhibited decreased expression of other PTCHD1 locus transcripts (Figure S9D). iPSCs from ASD-70 and controls (Ctrl 68 and an unrelated unaffected male Ctrl 50B) were differentiated using an adherent neuronal differentiation protocol and neuron identity was confirmed by RNA-seq detection of cortical layer markers (Figure S9E). ASD-70 neurons did not express PTCHD1-AS ex3, and expressed other PTCHD1 locus transcripts at similar levels to control neurons (Figure S8F).

Synaptic function was examined in neurons with PTCHD1-AS ex3 deletions by recording AMPAR-mEPSCs. We found a decrease in the frequency of AMPAR-mEPSCs in the CRISPR-edited neurons as compared to the isogenic controls (Figure 5E and 5F) and in neurons from ASD-70 as compared to control neurons (Figure 5G and 5H). mEPSC amplitude was decreased in CRISPR-edited neurons (Figure 5F), but not in ASD-70 neurons (Figure 5H). These data reveal that targeted mutation of PTCHD1-AS ex3 impairs synaptic function, consistent with our results from Prb1 and Prb2, thereby implicating this lncRNA in the etiology of ASD.

DISCUSSION

We describe genetic and functional data that strengthen the link between the PTCHD1-AS lncRNA and ASD. Our genetic findings, including the identification of novel ASD-associated microdeletions, strongly suggest a role for PTCHD1-AS in conferring ASD risk in males. We also used a combination of cellular reprogramming and genome editing to show that disruption of PTCHD1-AS impairs excitatory neurotransmission.

Disruption of the PTCHD1 locus impairs excitatory synaptic function

PTCHD1-AS deletions led to decreased AMPAR-mEPSC frequency in Prb1 and Prb2 with no consistent morphometric changes. Prb2 had more structural excitatory synapses than did the control neurons, which may be due to the SHANK2 p.A1352T missense mutation, consistent with our recent report of increased synapse numbers in SHANK2-haploinsufficient neurons (41). The apparent discordance between synapse numbers and mEPSC frequency may be caused structural synapses that are postsynaptically silent (55, 56). It is also possible that co-culturing neurons with exogenous normal astrocytes may have obscured potential morphometric phenotypes, which could be revealed by using only endogenous mutant astrocytes produced at defined ratios during differentiation (34). Despite the increased synapse number in Prb2, we observed a reduction in AMPAR-mEPSC frequency. Overall, reduced mEPSC frequency was consistently observed in all 4 mutations studied, being corroborated in the PTCHD1-AS ex3 deleted CRISPR isogenic pair and in the ASD-70 neurons compared to control neurons. Our findings show that hypoactivity is cell autonomous in PTCHD1-AS neurons because it is detected even when control human astrocytes are present in the cultures. It would be interesting to explore potential effects of PTCHD1-AS in inhibitory neurons or non-cell autonomous effects by astrocytes in future work. Such impaired excitatory neurotransmission has also been observed in other human pluripotent stem cell-derived neurons with SHANK3 haploinsufficiency (10, 11) or with targeted mutations of NRXN1 (9). Together these findings suggest that reduced function of excitatory synapses plays an important role in the development of ASD.

In addition to decreased AMPAR-mEPSC frequency we also found that neurons from Prb1 and Prb2 had diminished responses to NMDA, implicating hypofunction of NMDARs which, should be verified in future studies of neurons with PTCHD1-AS deletions. Future studies will also address the mechanism of NMDAR dysfunction, which could result from changes in NMDAR subunit expression, localization, or phosphorylation (28, 47). NMDAR function was impaired in mice with heterozygous deletion of the ASD candidate gene Tbr1, likely via altered expression of its target Grin2B (57), which is also an ASD candidate gene (47). Interestingly, ASD-associated behavioral deficits in Tbr1 +/− mice were rescued by treatment with D-cycloserine (57), which is a partial agonist of NMDARs. NMDAR hypofunction may therefore represent a therapeutic target in some ASD cases, including individuals with deletions of PTCHD1-AS.

Role of PTCHD1-AS in ASD

The accumulating data indicate that the PTCHD1-AS lncRNA is an important target for early and/or confirmatory diagnosis of ASD (58). In fact, two of the most frequently observed CNVs in ASD cases are microdeletions of the PTCHD1-AS locus in males and chromosome 16p11.2 in both sexes (3, 6). Combined data from published reports and new subjects described here indicate that 69% of male carriers of PTCHD1-AS microdeletions present with ASD, while PTCHD1 likely has an independent role in susceptibility to NDD/ID. Further resolving the relative contributions of PTCHD1-AS and PTCHD1 in ASD- and NDD-associated synaptic phenotypes will be a priority for future experiments.

Another future priority is elucidating the molecular functions of PTCHD1-AS. Due to its low abundance, PTCHD1-AS is not efficiently detected in conventional low coverage RNA-seq studies, and this has hindered efforts to define exon use in naturally-occurring splice variants. It is therefore not clear which of the many PTCHD1-AS transcript variants is most appropriate for transgene rescue experiments. Moreover, rescue of a cis-acting nuclear lncRNA may require synthesis at the specific genomic locus (59). Finally, the low abundance and inherent diversity of PTCHD1-AS transcripts makes them challenging for biochemical enrichment to identify bound proteins or target genes (60). Identification of the molecular functions of PTCHD1-AS will likely benefit from studies of knockout mouse models, where brain regions that express the transcripts can be identified and neurons collected in abundance for biochemistry experiments.

Heterogeneity in genetic and functional underpinnings of ASD represent a major challenge for identifying novel therapeutics, as this likely leads to inconsistent treatment responses in clinical trials (61). By identifying neurophysiological deficits associated with specific ASD risk loci, human iPSC models will facilitate stratification of treatment groups for improved trial design. Our findings, therefore, help to highlight new therapeutic targets to the physiological pathways affected by PTCHD1-AS deletions, and suggest that individuals with these deletions may benefit from treatments that enhance excitatory synaptic function.

KEY RESOURCES TABLE

Resource Type	Specific Reagent or Resource	Source or Reference	Identifiers	Additional Information
Add additional rows as needed for each resource type	Include species and sex when applicable.	Include name of manufacturer, company, repository, individual, or research lab. Include PMID or DOI for references; use “this paper” if new.	Include catalog numbers, stock numbers, database IDs or accession numbers, and/or RRIDs. RRIDs are highly encouraged; search for RRIDs at https://scicrunch.org/resources.	Include any additional information or notes if necessary.
Antibody	For antibodies, see Table S3	This paper	For RRIDs/suppliers, see Table S3
Cell Line	iPSC lines, see Tables S4 and S11	This paper and references in Tables S4 and S11	N/A
Cell Line	293T cells	ATCC	ATCC Cat# CRL-3216, RRID:CVCL_0063
Cell Line	Plat-E cells	PMID 10871756	RRID:CVCL_B488
Commercial Assay Or Kit	CytoTune-iPS 2.0 Sendai Reprogramming Kit	ThermoFisher	A165517
Deposited Data; Public Database	GSE83089, GSE81624, GSE1129808, GSE123753	NCBI GEO datasets	https://www.ncbi.nlm.nih.gov/gds/
Deposited Data; Public Database	MSSNG database	https://www.mss.ng/	N/A
Deposited Data; Public Database	DECIPHER database	http://decipher.sanger.ac.uk/	RRID:SCR_006552
Deposited Data; Public Database	Simons Simplex Collection	https://sfari.org/resources/simons-simplex-collection	RRID:SCR_004644
Recombinant DNA	Plasmids for lentivirus packaging	PMID 20010937	N/A
Recombinant DNA	Plasmids for retrovirus/EOS reprogramming	PMID 20010937	N/A
Recombinant DNA	pL-SIN-EF1a-eGFP	Addgene 21320	RRID:Addgene_21320
Sequence-Based Reagent	For oligonucleotides, see Table S2	This paper and references in Table S2
Software; Algorithm	Fiji (ImageJ 2.0.0-rc-59/1.51k)	https://fiji.sc/	RRID:SCR_002285
Software; Algorithm	Simple neurite tracer	https://imagej.net/Simple_Neurite_Tracer	RRID:SCR_016566
Software; Algorithm	Mini Analysis Software	Synaptosoft Inc	RRID:SCR_002184
Software; Algorithm	Prism 8	Graphpad	RRID:SCR_002798