Figure 1. Evidence of selective pressures due to regulatory functionality of HARs a)Average distribution of variants by maximum AF in CG69. Average homozygosity in CG69 for all, conserved, and non-conserved nucleotides for b) common and c) rare variants. d) HARs are significantly enriched for regulatory marks in fetal and adult brain (Dunham et al., 2012; Roadmap Epigenomics et al., 2015) with the greatest activity e) in neural tissues. HARs are enriched for f) conserved TF motifs (10,000 random samples) and g) frequency of predicted motifs. h) Enrichment of TF motifs within HARs as determined against random sequences by TRANSFAC. i) Human specific alleles in HARs alter TF motifs in comparison to chimpanzee. See also Figures S1, S2, and Tables S1, S2.

Figure 2. HARs regulate dosage-sensitive genes involved in neural development a) Overlap of HAR associated and target genes with genes linked to ASD/ID and neural mouse phenotypes. b) Comparison of haploinsufficiency scores (Huang et al., 2010) across all genes, HAR associated genes, and HARs with predicted developmental brain regulatory activity (Capra et al., 2013). c) Mammalian phenotypes enriched within HAR genes. d) Enriched biological processes and mammalian phenotypes within HAR associated and target genes (Enrichr). See also Tables S3, S4.

Figure 3. Enrichment of rare de novo CNVs and biallelic point mutations in individuals with ASD a) Excess of affected individuals with de novo CNVs affecting HARs. b) Intergenic de novo duplication approximately 250kb upstream of NR2F2 with existing ChIA-PET data indicating direct interaction between HAR and gene promoter. c) Excess biallelic mutations arising from the rarest alleles, including d) conserved loci within active regulatory elements (maxAF<1%) and within e) active neural regulatory elements. Transcription factor binding enrichment for target and associated genes of rare biallelic mutations (maxAF<1%) in f) affected and unaffected individuals. g) Increased impact of rare biallelic mutations in affected individuals at conserved sites using MPRA assay in primary mouse neurospheres. See also Figure S3 and Tables S5, S6, S7.

Figure 4. Autism-linked HAR variant increases human CUX1 promoter activity in all cortical layers Two unrelated consanguineous families, a) AU-20400 and b) AU-13200 with c) homozygous mutation within long distance regulatory element of CUX1 within a methylated CpG marked by the presence of H3K4me1 in neuronal cell types. ChIA-PET (Fullwood et al., 2010; Li et al., 2010) data demonstrates interaction of HAR with promoter of CUX1. The mutation alters d) TF motifs in the reference genome by e) adding additional motifs. f) The G>A CUX1 interacting mutation results in a 2-fold increase in CUX1 promoter activity in N2A cells (neural precursor-like condition) and 2.5-fold increase in the presence of dominant-negative (DN) REST (neuronal-like condition). g) Dendritic filopodia and spines of control and Cux1 overexpressing neurons co-transfected with GFP plasmid. Bar represents 2μm. Quantifications show that Cux1 overexpression results in i) increased spine density than compared to controls and that this increase is markedly higher when treated with 4AP/BIC. Cux1 overexpression also results in h) significant increase in spine head surface area. Student’s t test p-value * ≤ 0.01; ** ≤ 0.0001 compared to control. j) Mutant A allele (Mt-A) HAR increases transcriptional activity of human CUX1 promoter linked to GFP reporter, compared to wild-type G allele (Wt-G) HAR, in E16.5 transgenic mice. k) Both Wt-G and Mt-A mutations drive expression of CUX1 across all layers of the cortical plate, with increased expression due to the mutation. See also Figure S4.

Figure 5. Homozygous INDEL within a HAR that interacts with promoter region of essential splicing regulator, PTBP2 a) Homozygous GGGTAC>A mutation between PTBP2 and DPYD with enhancer activity in fetal brain tissues (ChromHMM) and several TF motifs from TF-chip and chromatin interactions (4C-seq) with the promoter region of PTBP2. The INDEL is located within b) predicted TF motifs in the reference genome, c) causing their loss. Luciferase analysis of the mutant (Mt) and reference (Wt) HAR using a minimal promoter revealed d) no change in neural progenitor-like cells, while e) co-transfection with DN-REST resulted in a 50% decrease in activity.

Figure 6. Two independent mutations reduce regulatory activity of intronic HAR within GPC4 a) Homozygous mutations in two unrelated families with ASD b) affecting a HAR within the intron of GPC4 with enhancer activity in brain tissues (ChromHMM) where proximity and interaction data suggest interaction with GPC4 promoter. c) Loss of TF motifs due the mutations. Luciferase analysis of the mutant (Mt) and reference (Wt) HAR sequence using the GPC4 promoter revealed d) a 20–25% decrease in regulatory activity in N2A cells, while e) co-transfection with DN-REST resulted in a slightly diminished effect. T-Test p-value *≤0.05 compared to WT.