DNA methylation profiles across diverse human, chimpanzee, and hybrid cell types.

(A) Four cell types were differentiated from human, chimpanzee, and hybrid induced pluripotent stem cells. The cell types span diverse body systems including skeletal myocytes for the musculoskeletal system, dopaminergic neurons for the central nervous system, cranial neural crest cells for craniofacial development and hepatocyte progenitors for the liver. We collected DNA methylation and RNA-seq data from these cell types from humans, chimpanzees and their hybrids. Dental pulp stem cells were also collected from human and chimpanzee adult tissues. (B) Clustering analysis and (C) PCA between samples for all cell types (left) and for all iPSC-derived cell types (i.e. excluding DPSC, right). (D) Cell type specific markers of gene expression. Figure 1A was created using BioRender, and is published under a CC BY-NC-ND 4.0 license.

Contribution of cis and trans regulation to DNA methylation divergence between human and chimpanzee.

(A) Interspecies differences in methylation levels per CpG site is compared between parental and hybrid systems and assigned a regulation category based on hypothesis testing. (B) Parental and hybrid methylation ratios in dopaminergic neurons and the respective regulation category (color key in panel C). (C) Quantification of relative contribution of non-conserved regulation groups to whole-genome CpG sites across cell types. (D) Heterogeneity of the distribution of CpG sites of all regulation groups. (E) Quantification of CpG regulatory clusters by overlap with genomic features of transcripts genome-wide as well as cis-regulatory elements including promoters, enhancers and CTCF-bound cis-regulatory regions (CTCF). (F) Quantification of relative contribution of non-conserved regulation groups to regulatory clusters across cell types (color key in panel C). Figure 2A was created using BioRender, and is published under a CC BY-NC-ND 4.0 license.

Cis- and trans-factors influencing species-specific methylation patterns.

(A) HOMER motif enrichment of trans-DMRs in regions Hu>Ch and Ch>Hu methylated in humans (left panel), and differential expression of corresponding transcription factors in parents (right panel). (B) Fold-enrichments against genomic background of FOXM1, one of the top motifs enriched in Hu>Ch and Ch>Hu trans-DMRs, plotted against Hu and Ch relative gene expression levels in parents. (C) CpG-disrupting SNV and their effects on neighboring CpG methylation within ±25, 50, and 100 bp. The differential methylation patterns around CpG-disrupting SNV variants versus non-CpG-disrupting variants are assessed using two-proportion test (***: p < 0.001, **: p < 0.01, *: p < 0.05) (D) Distances of each cis-DMR vs trans-DMR to the nearest CpG-disrupting SNV site (Mann-Whitney U test; ***: p < 0.001).

Human-chimpanzee allele-specific methylation can be associated with allele-specific expression.

(A) Illustration of the transcriptional consequences of DNA methylation explored in this study. DNA methylation, as a stable cis-acting regulatory mark, often represses transcription. In a hybrid, allele-specific methylation can lead to allele-specific gene expression, although both methylation and its effects on transcription can be context-dependent. (B) We identified many promoters with allele-specific methylation in each cell type, most of which showed allele-specific methylation in only one cell type. (C) Quantification of promoters with pure-cis regulation of methylation across cell types. (D, E, F) Examples of gene expression-methylation correspondence. Methylation tracks show fractional methylation (mC/coverage × 100%) of individual CpGs. Expression values are in TMM (Trimmed Mean of M-values)-normalized CPM (counts per million). (D) CTSF as an example of species-specific, cell type-agnostic DMR, (E) LGALS8 as an example of cell type-specific allele-specific methylation, and (F) HOXA9 as an example of cell type-specific but species-agnostic methylation patterns. (G) Genome-wide profile of expression-methylation correlations (binned by distance of methylated region to the TSS) for genes with both promoter methylation and expression under cis-regulation (dark red and dark blue) or trans-regulation (light brown and light gray). Detailed statistics are provided in Supplemental File 6. Figure 4A was created using BioRender, and is published under a CC BY-NC-ND 4.0 license.

Gene sets with evidence of lineage-specific selection on methylation and gene expression.

The length of the bars indicates the number of genes in each gene set with allele-specific expression or methylation in each direction (human or chimpanzee-biased). The bars in darker colors represent the number of genes where higher methylation is associated with repressed gene expression, whereas lighter colors represent genes where higher methylation is associated with increased gene expression.

Directional bias in allele-specific methylation and gene expression identifies candidate genes for human-specific phenotypes.

Allele-specific expression (ASE) is plotted against allele-specific methylation (ASM). Orange dots represent genes consistent with repressive methylation (e.g. Hu>Ch ASE and Hu<Ch ASM) and purple dots represent genes consistent with activating methylation (e.g. Hu>Ch ASE and Hu>Ch ASM). Bar plots represent counts of genes with repressive methylation that exhibit expression changes in directions either consistent with (light green) or opposite to (light pink) what would be expected based on phenotypic differences between humans and chimpanzees. A) The “highly arched eyebrow” gene set in iPSC hybrids shows consistent Hu<Ch ASE and Hu>Ch ASM for genes with repressive methylation. B) The “growth delay” gene set in iPSC hybrids shows significant bias towards Hu<Ch ASE and Hu>Ch ASM for genes with repressive methylation. C) The “intellectual disability, moderate” gene set in DA shows consistent Hu>Ch ASE and Hu<Ch ASM for genes with repressive methylation, with genes including GRIK2, TUBB3, EGF and CC2D2A being among the genes with highest magnitude ASE and ASM. D) The “Hepatitis C” gene set in HEP shows consistent Hu<Ch ASE and Hu>Ch ASM for genes with repressive methylation. D) The “poor speech” gene set in iPSC hybrids represents a potentially compensatory pathway where most genes with repressive methylation show Hu<Ch ASE and Hu>Ch ASM, whereas the genes with highest magnitude of ASE and ASM (TUBB3 and GRIK2) show Hu>Ch ASE and Hu<Ch ASM.