Figure 4: Mettl3 promotes RNA destabilization during pausing.

a. RNAs with increased m⁶A in pausing (as defined in Fig. 3e) are significantly more downregulated than RNAs with decreased m⁶A. This pattern is Mettl3-dependent, as analyzing the same RNAs in Mettl3^−/− ESCs shows no effect. RNA-seq data as shown in Fig. 2f (n=3 biological replicates per group).

b. Differences in expression (log₂FC paused/Ctrl) between exonic and intronic RNA-seq data indicate a global decrease in RNA stability in Mettl3^+/+ ESCs upon pausing. This effect is absent in Mettl3^−/− ESCs. c. Schematic of the measurement of RNA degradation kinetics by SLAM-seq (left). In the paused state, Mettl3^−/− ESCs display an overall longer half-life of the transcriptome compared to Mettl3^+/+ ESCs (right). Half-lives were measured using 4 time points, with samples collected over 2 independent experiments (see Extended Data Fig. 5d). S4U: 4-thiouridine. d. Changes in RNA expression during pausing in wild-type ESCs (Paused/Ctrl, fold-change > 1.5) are anti-correlated with changes in RNA half-life in paused Mettl3^−/− ESCs (as measured in Fig. 4c). e. RNAs with increased m⁶A in pausing (as defined in Fig. 3e) are enriched among RNAs stabilized in Mettl3^−/− ESCs (half-life fold-change > 1.5). f. Increased RNA stability in control E3.5 blastocysts compared to diapaused blastocysts, as measured by treatment with actinomycin D for 2 hours followed by RT-qPCR (n = 4 biological replicates). Ribosomal 28S as negative control for RNA decay. All data are mean ± SD.

P-values (as indicated on figure) by two-tailed Student’s t-tests (a-b, f), one-way ANOVA (d) and two-proportion z-tests (e). Boxes in the box plots define the interquartile range (IQR) split by the median, with whiskers extending to the most extreme values within 1.5 × IQR beyond the box.