N6-methyladenosine-dependent regulation of messenger RNA stability

Abstract

N(6)-methyladenosine (m(6)A) is the most prevalent internal (non-cap) modification present in the messenger RNA of all higher eukaryotes. Although essential to cell viability and development, the exact role of m(6)A modification remains to be determined. The recent discovery of two m(6)A demethylases in mammalian cells highlighted the importance of m(6)A in basic biological functions and disease. Here we show that m(6)A is selectively recognized by the human YTH domain family 2 (YTHDF2) 'reader' protein to regulate mRNA degradation. We identified over 3,000 cellular RNA targets of YTHDF2, most of which are mRNAs, but which also include non-coding RNAs, with a conserved core motif of G(m(6)A)C. We further establish the role of YTHDF2 in RNA metabolism, showing that binding of YTHDF2 results in the localization of bound mRNA from the translatable pool to mRNA decay sites, such as processing bodies. The carboxy-terminal domain of YTHDF2 selectively binds to m(6)A-containing mRNA, whereas the amino-terminal domain is responsible for the localization of the YTHDF2-mRNA complex to cellular RNA decay sites. Our results indicate that the dynamic m(6)A modification is recognized by selectively binding proteins to affect the translation status and lifetime of mRNA.

Figure 1. YTHDF2 selectively binds m⁶A-containing RNA

a, Illustration of m⁶A methylatransferase, demethylase, and binding proteins. RRACH is the extended m⁶A consensus motif, where R is G or A and H is not G. b, LC-MS/MS showing m⁶A enrichment in GST-YTHDF2-bound mRNA while depleted in the flow-through portion. Error bars, mean ±s.t.d., n = 2, technical replicates. c, Overlap of peaks identified through YTHDF2-based PAR-CLIP and the m⁶A-seq peaks in the same cell line. d, Binding motif identified by MEME with PAR-CLIP peaks (p = 3.0 e−46, 381 sites were found under this motif out of top 1000 scored peaks). e, Pie chart depicting the region distribution of YTHDF2-binding sites identified by PAR-CLIP, TTS (200 bp window from the transcription starting site), stop codon (400 bp window centered on stop codon).

Figure 2. YTHDF2 destabilizes its cognate mRNAs

a–d, Cumulative distribution of mRNA input (a), ribosome-protected fragments (b), and mRNA lifetime log₂ fold changes (Δ, c) between siYTHDF2 (YTHDF2 knockdown) and siControl (knockdown control) for non-targets (grey), PAR-CLIP targets (blue), and PAR CLIP-RIP common targets (red). The mRNA lifetime log₂ fold changes were further grouped and analyzed based on the number of CLIP sites on each transcript (d). The increased binding of YTHDF2 on its target transcript correlates with reduced mRNA lifetime. P values were calculated using two-sided Mann-Whitney or Kruskal-Wallis test (rank-sum test for the comparison of two or multiple samples, respectively). Detailed statistics were presented in Extended Data Fig. 3c. e, Western-blotting of flag-tagged YTHDF2 on each fraction of 10–50% sucrose gradient showing that YTHDF2 does not associate with ribosome. The fractions were grouped to non-ribosome mRNPs, 40–80S, and polysome. f, Quantification of the m⁶A/A ratio of the total mRNA, non-ribosome portion, 40–80S, and polysome by LC-MS/MS. Noticeable increases of the m⁶A/A ratio of the total mRNA, mRNA from 40–80S, and mRNA from polysome were observed in the siYTHDF2 sample compared to control after 48 h. A reduced m⁶A/A ratio of mRNA isolated from the non-ribosome portion was observed in the same experiment. P values were determined using two-sided Student’s t-test for paired samples. Error bars, mean ± s.t.d, for poly(A)-tailed total mRNA input, n = 10 (five biological replicates × two technical replicates), and for the rest, n = 4 (two biological replicates × two technical replicates).

Figure 3. YTHDF2 affects SON mRNA localization in processing body (P-body)

a, Schematic of the domain architecture (aa stands for amino acids) of YTHDF2, N-terminal of YTHDF2 (N-YTHDF2, aa 1–389, blue) and C-terminal of YTHDF2 (C-YTHDF2, aa 390-end, red). b, Over-expression of full-length YTHDF2 led to reduced levels of m⁶A after 24 h, while over-expression of N-YTHDF2 or C-YTHDF2 increased the m⁶A/A ratio of the total mRNA. P values were determined using two-sided Student’s t-test for paired samples. Error bars, mean ± s.t.d., n = 4 (two biological replicates × two technical replicates). c–e, Fluorescence in situ hybridization of SON mRNA and fluorescence immunostaining of DCP1a (P-body marker), flag-tagged YTHDF2 (c), flag-tagged C-YTHDF2, (d) and flag-tagged N-YTHDF2 (e). Full-length YTHDF2 and C-YTHDF2 co-localize with SON mRNA (bearing m⁶A) while the full-length YTHDF2 significantly increases the P-body localization of SON mRNA compared to N-YTHDF2 and C-YTDF2. The numbers shown above figures are Pearson correlation coefficients of each channel pair with the scale of the magnified region (white frame) set as 2 µm × 2 µm. f, Tethering N-YTHDF2-λ to a mRNA reporter F-luc-5BoxB led to a ~40% reduction of the reporter mRNA level compared to tethering N-YTHDF2 or λ alone (green) and controls without BoxB (F-luc, yellow). P values were determined using two-sided Student’s t-test for paired samples. Error bars, mean ± s.t.d., n = 6 (F-luc-5BoxB) or 3 (F-luc). g, A proposed model of m⁶A-dependent mRNA degradation mediated through YTHDF2. The three states of mRNAs in cytoplasm are defined by their engagement with ribosome using the sedimentation coefficient range in sucrose gradient: >80S for actively translating polysome; 40–80S for translatable pool; 20–35S for non-ribosome mRNPs.