Figure 3. DNA methylation in mSTARR-seq enhancers predicts in vivo gene expression in macrophages.

(A) Study design of the in vivo experiment, in which matched macrophage samples from 35 individuals were either left non-infected or infected with influenza A virus (IAV) for 24 hours and processed for RNA-seq and whole genome bisulfite sequencing (WGBS; Aracena et al., 2024). (B) Within individuals, DNA methylation (DNAm) levels at mSTARR-seq enhancers in non-infected cells are negatively correlated with the nearest genes’ transcriptional responses to IAV, but only in mSTARR-seq enhancers that were specific to the IFNA condition (IFNA-specific enhancers: n=1033, mean Pearson’s r=–0.170 ± 0.009 s.d., all Bonferroni-corrected p<2 x 10^–5; shared enhancers: n=1736, mean Pearson’s r=–0.049 ± 0.01 s.d., all Bonferroni-corrected p>0.1). Each colored line represents an individual, and vertical gray lines represent 95% confidence intervals (see Supplementary file 17 for full results). (C) The average within-individual correlation (r) between DNA methylation and gene expression (GE) is 2.44 times as large after infection (r=–0.261 ± 0.006) than at baseline (r=–0.106 ± 0.008) in IFNA-specific mSTARR-seq enhancers (right panel), but much less affected by infection at mSTARR-seq enhancers that are shared across conditions (left panel). Within each panel, each colored line represents an individual for the same set of enhancers (see Supplementary file 18 for full results), without and with IAV infection. (D) Across individuals, the ISG15 transcriptional response to IAV is significantly correlated with average DNAm at the mSTARR-seq enhancer chr1:1013400–1014000 in non-infected cells (R²=0.381, p=6.05 x 10^–5, q=0.084). Each dot represents an individual (see Supplementary file 19 for full results and Figure 3—figure supplement 1 for condition-specific results). (E) The mSTARR-seq enhancer predictive of ISG15 response to IAV (dark green bar) is located in the active promoter of ISG15 (as defined by ENCODE The ENCODE Project Consortium, 2012; red denotes active promoter, pink denotes weak promoter, orange denotes strong enhancer, yellow denotes weak enhancer). Three adjacent, methylation-dependent, IFNA-specific mSTARR-seq enhancers were identified (light green), but do not significantly predict ISG15 response to IAV (q>10%). The bottom 6 tracks depict non-normalized, raw read pileups for mSTARR-seq RNA replicates in either the unmethylated (open circle) or methylated (filled circle) condition, with all y-axis maximums set to 20,000.

Figure 3—figure supplement 1. Across individuals, methylation in the mSTARR-seq annotated enhancer chr1:1013400–1014000 predicts the ISG15 gene expression response to flu.

(A) Across individuals, average methylation within the mSTARR-seq annotated enhancer chr1:1013400–1014000 in non-infected baseline macrophages significantly predicts ISG15 gene expression (GE) in the non-infected condition (R²=0.324, p=2.66 x 10^–4), but (B) not in the IAV-infected condition (R²=0.001, p=0.316). Each dot represents an individual. Relative GE is log(CPM) after regressing out the effects of sequencing batch and age. (C) Individuals with relatively low methylation in the mSTARR-seq chr1:1013400–1014000 enhancer (indicated by lighter line color) in the baseline, non-infected condition tend to have higher ISG15 gene expression in the non-infected condition, ultimately resulting in a shallower ISG15 transcriptional response to IAV infection (as indicated by slope of the line). Each line represents an individual.