Figure 5. Identification of m⁶A methylation targets in HSCs.

(a) Heat map of transcript abundance between input, m⁶A-tagged, and unbound fractions. A total of 5,676 genes with log2(RPKM) ≥5 in ≥4 samples and stdev[log2(RPKM)]≥0.6 are shown.

(b) Volcano plot of meRIP-seq transcript expression differences between m⁶A-tagged and unbound fractions in HSCs. The x-axis specifies the log2 fold-changes (FC) and the y-axis specifies the log2 false discovery rate (FDR). Dashed vertical and horizontal lines indicate the filtering criteria (log2(FC)>1 and FDR<0.05). Orange dots represent 2,621 transcripts showing statistically significant differences between bound and non-bound factions, of which 2,073 are significantly enriched in the m⁶A-tagged fraction.

(c) meRIP-seq data showing that transcripts of some known HSC regulators are m⁶A-tagged in wild-type HSCs, and this methylation is largely eliminated in Mx1-cre; Mettl3^fl/fl (Mettl3^Δ/Δ) HSCs 10 days after Cre induction. Data were from n=3 biologically independent samples for control and Mettl3^Δ/Δ. Values are shown as individual points with mean ± s.d.

(d) Gene set enrichment analysis plots showing that Mettl3-deficient HSCs lose HSC gene signature, as determined by RNA-seq profiling. Analysis completed on gene list ranked by log₁₀ FDR and fold change sign, with enrichment determined after 1,000 permutations. Data were from n=3 biologically independent samples for control and Mettl3^Δ/Δ. The statistics was computed using GSEA, and controlled for multiple comparisons by false discovery rate.

(e) Cumulative distribution of log₂ (gene expression ratios, Mettl3-deficient/control). Genes were separated as m⁶A and non-m⁶A as assessed by meRIP-seq. Insert is the box plot of the log₂ fold change in expression of non-m⁶A and m⁶A targets in HSCs. Plot displays the mean, standard deviation and interquartile range. P value was determined by two-sided Kolmogorov-Smirnov test.

All sequencing data were from n=3 biologically independent animals for control and Mettl3^Δ/Δ.