Article Figures & Data
Figures
- Figure 1. Design of cell-free chromatin immunoprecipitation sequencing (cfChIP-Seq) of plasma samples.
(A) Schematic workflow: blood samples of healthy controls and heart transplant patients were collected in Streck Cell-Free DNA BCT tubes. After an initial 1,000g centrifugation, plasma was further centrifuged at 16,000g. Cell-free DNA bound to histones was pooled by specific antibodies which were covalently bound to magnetic beads. Bound cfDNA was purified, used to generate DNA libraries, and sequenced. Sequenced reads were analyzed against controls to identify enrichment peaks, tissue-specific signatures, and biological pathways. (B) Length distribution of sequenced DNA fragments by cfChIP-seq showing nucleosomal periodicity. (C) Example sequencing saturation curves from HC6 for input DNA, H3K36me3, H3K4me1, H3K4me2, and H3K4me3 reads, highlighting the saturation achieved for H3K4me3 and H3K4me2. (D) Chromosome-based circos plots representing the distribution of cfChIP-seq peaks across the genome. Density plot of peaks found in all six healthy controls studied.
- Figure S1. Optimization of cfChIP-seq.
(A) Saturation curve of all individual samples for input DNA and chromatin (H3K36me3, H3K4me1, H3K4me2, and H3K4me3) profile. (B) Average distribution (mean and confidence interval) of cfChIP-seq signals on genes around the TSS and enhancers for with and without input normalization.
- Figure S2. Assessing replication of cfChIP-seq signals for healthy controls.
(A) Bar plots showing fraction of reads in peaks for H3K4me1 (left), H3K4me2 (middle), and H3K4me3 (right). Each bar on the x-axis indicates individual subjects. Fraction of reads in peaks was highest for H3K4me3 than for H3K4me1 and H3K4me2. (B) Bar plots representing the number of called peaks of individual samples for H3K4me1 (left), H3K4me2 (middle), and H3K4me3 (right). Number of peaks were highest for H3K4me2 than for H3K4me1 and H3K4me3.
- Figure 2. ChIP-Seq correlates with gene expression in a physiological state.
(A) Epigenome browser snapshots showing (RPGC-normalized sequence reads) H3K4me3 ChIP-seq signals at example housekeeping (TBP, GAPDH), monocyte-specific genes (FCN1, CSF3R) and silent (IL-3, CSF2) genes in HC6. (B) Comparison of genes identified by ChIP-Seq. Top: Venn diagram showing the number of promoters overlapping H3K4me3 peaks that are housekeeping genes (red). All gene lists were defined by Sadeh et al (2021) and identified in Table S3. Bottom: Venn diagram showing the overlap of non-housekeeping genes having H3K4me3 peaks associated with monocytes (orange) and/or neutrophils (green). (C) Scatterplot plot showing the correlation between H3K4me3 leukocyte ChIP-seq data and cfChIP-seq data (R2 = 0.86). Leukocyte data from Roadmap Epigenomics Consortium et al (2015).
- Figure S3. Profiles of the cell-free chromatin in heart transplants.
(A) Length distribution of sequenced DNA fragments for heart transplant patients. (B) Average reads per genomic content for H3K4me3 around the TSS site in heart transplant patients.
- Figure 3. H3K4me3 cfChIP-seq identifies relevant genes in heart transplant patients.
(A) Venn diagram showing H3K4me3 peaks in transplant patients overlapping promoters of constitutively active housekeeping genes, and non-housekeeping genes associated to monocytes (orange) and neutrophils (green). (B) Principal component analysis plot of H3K4me3 peaks showing separation of healthy controls from transplant patient samples. (C) MA scatter plot shows differential gene peaks between heart transplant and healthy controls; a subset of calcineurin genes is marked. The blue dots indicate significant differential gene signals between heart transplant and healthy controls, dots under the thick line depict genes with lower H3K4me3 signals in heart transplant subjects compared with healthy controls. (D) KEGG pathway enrichment analysis of genes whose promoters were associated with a significant negative fold-change in transplant patients relative to healthy controls. Select immune and nonimmune pathways associated with transplantation state are shown. (E) Heatmap showing H3K4me3 pattern around the TSS site of genes associated with the calcineurin pathway. Calcineurin genes were defined as being members of GO:0097720 or Reactome R-HSA-2025928. Genes above the thick line show significant difference specific to transplantation state as compared with healthy controls.
- Figure 4. cfChIP-seq identifies tissue sources of cfDNA.
(A) Dot plot showing the cell or tissue contribution of plasma cfDNA in healthy controls estimated. The radius of the circle represents Benjamin–Hochberg-adjusted P-values (q-scores) and the color represents the magnitude of normalized reads per kb. (B) Bar plot showing the proportion of cfChIP-seq signatures from different tissue types in healthy controls, calculated from the normalized counts. (C) Bar plot showing the distribution of plasma cell or tissue-specific cfDNA level for healthy controls using whole-genome bisulfite sequencing (Andargie et al, 2021). (D) The concentration of total plasma cfDNA (ng/ml) was determined by qPCR in healthy controls and transplant patients. (E) Comparison of cardiac-specific cfDNA in healthy controls and heart transplant patients, measured using cfChIP-seq. Absolute concentration was measured by multiplying cardiac-specific estimate (%) by total cfDNA concentration. (F) Comparison of cfChIP-seq hematopoietic and non-hematopoietic tissue signatures in healthy controls and heart transplant patients. (G) Comparison of cfChIP-seq tissue-specific signatures in healthy controls and heart transplant patients. For plots (E, F, G), input-normalized reads/kb were multiplied with total cfDNA concentration and comparison was done by Mann–Whitney U test. P-value <0.05 was considered statistically significant; *P < 0.05; **P < 0.01; ***P < 0.001.
Supplementary Materials
