Breaking Breaking
Nature Biotechnology

Trimodal single-cell profiling of transcriptome, epigenome and 3D genome in complex tissues with scHiCAR

myndfocal
The three-dimensional (3D) organization of cis-regulatory elements (CREs) is critical in transcription control. However, capturing transcriptome, epigenome and 3D genome from the same single cells remains challenging. Here we present scHiCAR (single-cell Hi-C with assay for transposase-accessible chromatin and RNA sequencing), a plate-based combinatorial barcoding method that simultaneously profiles mRNA, open chromatin and chromosome conformation capture from the same cells. Compared to existing single-cell 3D genome methods, scHiCAR more efficiently enriches long-range cis-interactions anchored at candidate CREs (cCREs). Applied to 1.62 million mouse brain cells and complemented with a deep-learning-based loop caller, scHiCAR accurately defines cell-type-specific transcriptomes, accessible cCREs and 5-kb-resolution enhancer–promoter pairs across 22 brain cell types. scHiCAR also performs robustly in challenging tissues such as skeletal muscle, enabling trimodal single-cell-level analysis of gene regulation dynamics during muscle stem cell regeneration. By providing a scalable and cost-effective system for single-cell trimodal analysis of gene-regulatory landscapes in complex tissues, scHiCAR reveals gene-locus-specific regulatory roles of 3D genome reorganization in transcriptional control. Trimodal single-cell analysis reveals gene-regulatory architecture in complex tissues.

a. Proportion of CRE–gene pairs that cross topologically associating domain (TAD) boundaries. Comparisons were made between pairs predicted by BICCN snATAC-seq based on co-accessibility (light blue) and pairs identified by scHiCAR chromatin interactions (dark blue) across different brain cell types. For fair comparison, the number of co-accessible pairs (n = 418,528) was comparable to the number of scHiCAR pairs (n = 440,710) by selecting top-ranked pairs by co-accessibility score. b. Enrichment of VISTA-validated enhancers among different sets of distal cCREs. Compared were: (i) snATAC-seq distal cCREs linked to gene promoters by the ABC model (ABC score > 0.05) and (ii) snATAC-seq distal cCREs linked by scHiCAR chromatin interactions. Odds ratios were calculated using Fisher’s exact test. For fair comparison, both sets contained similar numbers of cCREs (47,332 for ABC and 43,290 for scHiCAR). c. Genome browser view showing a VISTA-validated enhancer (mm1451.0) that interacts with the Grk3 promoter in astrocytes. Chromatin interactions called by scDeepLUCIA are shown as black squares on the contact matrix. d. Log-normalized expression of genes with promoter accessibility linked to cCREs (red) versus unlinked genes (pink) across brain cell types. Box plots indicate the median, 25th–75th percentiles, and 1.5× IQR whiskers. P-values from two-sided unpaired Wilcoxon tests. The number of samples from left to right are 7,385; 5,066; 7,275; 4,990; 7,269; 5,227; 5,477; 4,212; 7,327; 4,501; 7,616; 5,143; 6,731; 5,012; 5,555; 3,756; 7,269; 4,710; 7,020; 5,135; 6,551; 5,435; 2,009; 2,448; 2,243; 1,708; 7,361; 4,609; 7,857; 4,859; 6,211; 4,109; 1,901; 1,611; 6,548; 5,473; 7,136; 5,421; 5,748; 5,960; 5,741; 5,707; 4,861; 5,419. The p-values are 1.37e-38, 2.45e-35, 6.65e-41, 3.05e-22, 1.03e-30, 7.47e-40, 1.32e-37, 2.78e-14, 1.89e-27, 1.88e-47, 2.10e-50, 1.01e-19, 0.00024, 3.78e-24, 6.45e-24, 6.14e-19, 3.23e-10, 2.21e-36, 4.91e-28, 2.85e-33, 7.86e-54, 1.04e-28. e. An example of positive association between gene expression and the number of scHiCAR chromatin interactions in L23IT-1 neurons. Shown are RNA expression (red) and chromatin accessibility (blue) tracks. f. Scatterplot showing correlation between changes in the number of scHiCAR chromatin interactions and changes in gene expression between two brain cell types. The dashed line indicates the linear regression fit. The Pearson correlation coefficient (PCC) and associated p-value were calculated using a two-sided Pearson correlation test. g. Density plot showing the distribution of Pearson correlation coefficients (PCCs) between gene expression and chromatin accessibility at distal CREs for all CRE–gene pairs. A subset of 20,270 pairs with the strongest positive correlations (FDR < 0.05) was defined as high-confidence enhancer–gene pairs (red). Randomized pairings (gray) serve as a control. h. Comparison of enhancer–gene correlation scores between scHiCAR and Paired-seq. Top 34,473 enhancer–gene pairs predicted by scHiCAR were compared with the same number of pairs from Paired-seq based on mRNA and H3K27ac signal correlation. Distribution of Spearman correlation coefficients is shown. P-value calculated by two-sided unpaired Wilcoxon test.

Source data

— Source: Nature Biotechnology (https://www.nature.com/articles/s41587-026-03013-7)

Biomedical Research Health & Biotech
Read original on Nature Biotechnology →