Programmable m6A modification of cellular RNAs with a Cas13-directed methyltransferase

Abstract

N⁶-Methyladenosine (m⁶A) is the most widespread internal messenger RNA modification in humans. Despite recent progress in understanding the biological roles of m⁶A, the inability to install m⁶A site specifically in individual transcripts has hampered efforts to elucidate causal relationships between the presence of a specific m⁶A and phenotypic outcomes. In the present study, we demonstrate that nucleus-localized dCas13 fusions with a truncated METTL3 methyltransferase domain and cytoplasm-localized fusions with a modified METTL3:METTL14 methyltransferase complex can direct site-specific m⁶A incorporation in distinct cellular compartments, with the former fusion protein having particularly low off-target activity. Independent cellular assays across multiple sites confirm that this targeted RNA methylation (TRM) system mediates efficient m⁶A installation in endogenous RNA transcripts with high specificity. Finally, we show that TRM can induce m⁶A-mediated changes to transcript abundance and alternative splicing. These findings establish TRM as a tool for targeted epitranscriptome engineering that can reveal the effect of individual m⁶A modifications and dissect their functional roles.

Figure 1.. Overview of cellular modification of adenine to m⁶A in mRNA, and design of a targeted RNA methylation system.

(a) METTL3 and METTL14 form a “writer” complex that catalyzes S-adenosyl methionine (SAM)-dependent methylation of the N⁶ of adenine in cellular mRNA. Additional components influence the formation and activity of this core complex. FTO and ALKBH5 (“erasers”) remove the methyl group of m⁶A. “Readers” recognize the mark on RNA and direct it to various outcomes. (b) Proposed strategy for targeted RNA methylation (TRM). A programmable RNA-binding protein such as dCas13 when fused to an appropriate methyltransferase complex mediates the guide RNA-specified methylation of A to m⁶A site-specifically in a target transcript.

Figure 2.. Validation of targeted RNA methylation (TRM) in E. coli.

(a) METTL3^273-580 (M3) or METTL3^359-580–(GGS)₁₀–METTL14^111-456 (M3M14) methyltransferases were fused to catalytically-impaired PspCas13b Δ984-1090 (dCas13) to generate dCas13–M3 or dCas13–M3M14 TRM editors. (b) Plasmids transformed into E. coli encode IPTG-inducible TRM editors with a guide RNA targeting a synthetic transcript. After targeted methylation, cellular total RNA was purified, fragmented, and immunoprecipitated with anti-m⁶A antibodies. Enrichment of m⁶A was measured by target-specific RT-qPCR (MeRIP-RT-qPCR) or transcriptome-wide high-throughput sequencing (MeRIP-seq). (c) Target transcript m⁶A abundance measured by MeRIP-RT-qPCR under four conditions: induced TRM editor with transcript-targeting guide, non-induced editor with transcript-targeting guide, induced editor with a non-target guide, and cells lacking TRM editor and guide plasmid. The methyltransferase-inactive control was dCas13–M3^D395A (dCas13–dM3). Values and error bars reflect the mean±s.e.m. of n=3 independent biological replicates. (d) Top: differential m⁶A enrichment of all methylated adenines in E. coli expressing synthetic transcript-targeting guide RNAs and the indicated TRM editors. For the comparisons above, only differentially methylated sites with statistical significance (P<0.001) are shown. Bottom: Venn diagrams depicting overlap of all methylated m⁶A sites for the above comparisons. 42,418 total m⁶A motifs (RRACH) are susceptible to modification within the E. coli transcriptome. MeRIP-seq analysis was performed with n=3 independent biological replicates. Statistical significance was calculated using a logs likelihood-ratio test with false discovery rate correction.

Figure 3.. Methylation of reporter transcripts in human cells.

(a) HEK293T cells were transfected with plasmids encoding a TRM editor, Cas13 guide RNA, and the targeted Cluc reporter transcript containing a 3’ UTR methylation target. (b) Methylation of a synthetic 3’ UTR fused to Cypridina luciferase (Cluc–syn) by TRM editors. (c) Methylation of the Socs2 3’ UTR fused to Cypridina luciferase (Cluc–Socs2) by TRM editors. Inactive TRM editors contain a methyltransferase-inactivating D395A mutation within M3. Values and error bars reflect the mean±s.e.m. of n=3 independent biological replicates.

Figure 4.. Cellular localization of TRM editors and targeted methylation of endogenous transcripts in human cells.

(a) Left: nucleus- and cytoplasm-localized TRM editor variants. NES = nuclear export signal; NLS = nuclear localization signal. Right: representative immunofluorescence images of HEK293T cells transfected with HA-tagged TRM editors and non-targeting or Actb-targeting guide RNAs. Green = HA tag; blue = DAPI. Scale bars represent 10 μm. The experiment was independently performed twice with similar results. (b-e) Methylation by nucleus- and cytoplasm-localized TRM editors targeting endogenous (b) Actb A1216, (c) Gapdh A690, (d) Foxm1 A3488 and A3504, and (e) Sox2 A1398 and A1405. Guide RNAs used for targeting each transcript are shown in purple. Inactive TRM editors contain a methyltransferase-inactivating D395A mutation within M3. Values and error bars reflect the mean±s.e.m. of n=6 (b), n=4 (c), or n=5 (d,e) independent biological replicates. P values were calculated using a two-tailed Student’s t-test.

Figure 5.. Specificity and off-target methylation of TRM editors.

(a,b) Gapdh read coverage of m⁶A-immunoprecipitated (m⁶A IP) RNA from MeRIP-seq of transfected HEK293T cells. Gapdh A690 was targeted with (a) dCas13–M3nls or (b) dCas13–M3M14nes with the following conditions: active editor with a target guide RNA, inactive editor with a target guide RNA, and active editor with a non-target guide RNA. All DRACH motifs susceptible to TRM modification are shown as black tick marks underneath the IGV track. The target guide RNA (purple) and targeted A690 site (grey) are shown. (c,d) Differential m⁶A enrichment of >21,000 methylated sites in HEK293T cells transfected with (c) dCas13–M3nls or (d) dCas13–M3M14nes and Gapdh A690-targeting or non-target guide RNAs. Top: differential methylation of m⁶A sites between conditions indicated above. Only differentially methylated sites with statistical significance (P<0.001) are shown. Bottom: Venn diagrams depicting overlap of all methylated m⁶A sites for the above comparisons. (e) Overlay of dCas13–M3nls (blue) and dCas13–M3M14nes (grey) differential methylation with the following comparisons: left, active editors with Gapdh guide RNAs vs. inactive editors with Gapdh guide RNAs; right, active editors with non-target guide RNAs vs. inactive editors with Gapdh guide RNAs. The targeted Gapdh A690 site is outlined in black. Inactive TRM editors contain a methyltransferase-inactivating D395A mutation within M3. MeRIP-seq analysis was performed with n=5 independent biological replicates. Statistical significance was calculated using a logs likelihood-ratio test with false discovery rate correction.

Figure 6.. TRM effect on RNA abundance and alternative splicing.

Differential RNA expression of HEK293T cells transfected with (a) dCas13–M3nls or (b) dCas13–M3M14nes and either Actb A1216-targeting or non-targeting guide RNAs. The targeted Actb gene is marked in red. Inactive TRM control indicates editors containing a methyltransferase-inactivating M3 D395A mutation with an Actb-targeting guide RNA. Non-target gRNA indicates methyltransferase-active TRM editors with a non-targeting guide RNA as a control. Differential RNA-seq analysis was performed with n=5 independent biological replicates. Statistical significance was calculated using a two-tailed Student’s t-test with false discovery rate correction. Alternative splicing of (c) Brd8 and (d) Znf638 in HEK293T cells transfected with dCas13–M3nls and the indicated guide RNAs (purple) for targeting m⁶A sites shown (red). Exon exclusion or inclusion indicates the percentage of RNA isoform lacking or containing the alternatively-spliced exon, respectively. Values and error bars reflect the mean±s.e.m. of n=3 (c) or n=6 (d) independent biological replicates. P values were calculated using a two-tailed Student’s t-test.