VIRMA mediates preferential m6A mRNA methylation in 3'UTR and near stop codon and associates with alternative polyadenylation

Abstract

N⁶-methyladenosine (m⁶A) is enriched in 3'untranslated region (3'UTR) and near stop codon of mature polyadenylated mRNAs in mammalian systems and has regulatory roles in eukaryotic mRNA transcriptome switch. Significantly, the mechanism for this modification preference remains unknown, however. Herein we report a characterization of the full m⁶A methyltransferase complex in HeLa cells identifying METTL3/METTL14/WTAP/VIRMA/HAKAI/ZC3H13 as the key components, and we show that VIRMA mediates preferential mRNA methylation in 3'UTR and near stop codon. Biochemical studies reveal that VIRMA recruits the catalytic core components METTL3/METTL14/WTAP to guide region-selective methylations. Around 60% of VIRMA mRNA immunoprecipitation targets manifest strong m⁶A enrichment in 3'UTR. Depletions of VIRMA and METTL3 induce 3'UTR lengthening of several hundred mRNAs with over 50% targets in common. VIRMA associates with polyadenylation cleavage factors CPSF5 and CPSF6 in an RNA-dependent manner. Depletion of CPSF5 leads to significant shortening of 3'UTR of over 2800 mRNAs, 84% of which are modified with m⁶A and have increased m⁶A peak density in 3'UTR and near stop codon after CPSF5 knockdown. Together, our studies provide insights into m⁶A deposition specificity in 3'UTR and its correlation with alternative polyadenylation.

Fig. 1

Proteomic identification of new m⁶A methyltransferase complex components and evaluation of their effects on mRNA m⁶A modification distribution. a Overlap of protein interactomes of METTL3, METTL14, and WTAP. Stable expression HeLa cells with dual-tagged (N-term tandem Flag and HA) METTL3, METTL14, and WTAP were subjected to tandem affinity purification with Flag and HA antibodies and further mass spectrometry identification. b Validation of selected common targets of METTL3, METTL14, and WTAP by western blotting. After Flag IP in the dual Flag-HA-tagged METTL3, METTL14, and WTAP stable cell lines, western blotting was performed to validate the interactions using endogenous antibodies. Because endogenous ZC3H13 antibody is not good for western blotting, Flag-tagged full-length ZC3H13 and its truncated forms N-terminal (N-ZC3H13) and C-terminal (C-ZC3H13) were overexpressed in the METTL3, METTL14, and WTAP stable cell lines for validation using Flag antibody after HA IP. c Effects of siRNA knockdown of VIRMA, HAKAI, ZC3H13, TRIM28, and HNRNPH on m⁶A level in HeLa cell line. Two different siRNA constructs were tested for each gene. Data are represented as means ± SEM of n = 3. P values calculated using one-tailed Student’s t-test between control and siRNA knockdown samples are <0.05, except for TRIM28 knockdown samples (P > 0.05). d Effects of VIRMA siRNA knockdowns on WTAP expression and polyadenylated RNA m⁶A levels. Overexpression (oe) of WTAP under VIRMA siRNA knockdowns cannot recover m⁶A level. Data are represented as means ± SEM of n = 3. P values calculated using one-tailed Student’s t-test between control and siRNA knockdown samples are <0.001, while P values between siVIRMA and siVIRMA/WTAP oe are larger than 0.05. e Comparison of m⁶A levels in polyadenylated RNAs in between control and VIRMA^mut/− HeLa cell lines. The VIRMA^mut/− HeLa cell line was generated by cas9 gene-editing system. Data are represented as means ± SEM of n = 3. P values calculated using one-tailed Student’s t-test between control and VIRMA^mut/− samples are <0.001. f, g Comparison of mRNA and protein expression levels of METTL3, METTL14, and WTAP between control and VIRMA^mut/− HeLa cell lines using RT-qPCR (f) and western blotting (g), respectively. GAPDH serves as control. P values calculated using one-tailed Student’s t-test between control and VIRMA^mut/− RT-qPCR samples are >0.05

Fig. 2

VIRMA affects cellular functions. a, b Effects of depletion of VIRMA (a) and METTL3, METTL14, and WTAP (b) by siRNA knockdowns on the profiles of m⁶A peak density along mRNA transcript. c Effect of partial VIRMA knockout (VIRMA^mut/−) on the profiles of m⁶A peak density along mRNA transcript. d Differentially expressed genes (fold change > 2) were identified by the RNA-seq in between VIRMA^mut/− and control cell lines. The number of overlapped upregulated and downregulated genes are shown. e Gene ontology (GO) and enrichment analysis of the differentially expressed and m⁶A-containing genes. f Estimate of VIRMA depletion on cell proliferation by MTS assay. Data are represented as means ± SEM of n = 3

Fig. 3

VIRMA modulates abundance and lifetime of mRNAs through mediating m⁶A installation. a,b Validation of selected mRNA targets with upregulated expression (a) and decreased m⁶A level (b) upon VIRMA depletion in HeLa cell line. P values calculated using one-tailed Student’s t-test between control and VIRMA depletion samples; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Data are represented as means ± SEM of n = 3. c, d Measurement of lifetimes of selected mRNAs IGFBP5 (c) and NOTCH1 (d). Data are represented as means ± SEM of n = 3

Fig. 4

VIRMA recruits catalytic components METTL3/METTL14/WTAP in order to guide mRNA m⁶A methylation at specific site. a Schematic of domain architecture (aa, amino acids) of VIRMA, N terminus of VIRMA (N-VIRMA, aa 1–1130), and C terminus of VIRMA (C-VIRMA, aa 1131–1812). b Co-immunoprecipitation experiments in HeLa cells in order to dissect interactions of different domains of VIRMA with other methyltransferase components. RNase was added in order to determine if the interaction was RNA-dependent. HSV-tagged C-ZC3H13 was co-expressed for immunoprecipitation experiment and corresponding HSV-tag antibody was used in the western blotting. c Construct of the tethering reporter assay. The mRNA reporter consists of a firefly luciferase sequence as the coding region and five Box B sequence at 3′UTR (F-Luc-5BoxB). There exists a GGACU motif near the stop codon. Both N-VIRMA and C-VIRMA were fused with λ peptide (N-VIRMA-λ and C-VIRMA-λ), which recognizes Box B RNA with a high affinity. d Measurement of m⁶A level of F-Luc-5BoxB mRNAs under co-expression with different truncated forms of VIRMA. The synonymously mutated construct with GGAUU was tested for comparison. Renilla luciferase was used as an internal control to normalize the F-Luc signal. e Overlap of VIRMA RIP-seq mRNA targets with m⁶A-seq targets in HeLa cells. Endogenous VIRMA antibody was used for direct RNA immunoprecipitation. RIP-seq-enriched genes were defined as FDR ≤ 0.05 and log₂(IP/Input) ≥ 1. f Comparison of m⁶A peak density profiles between RIP-seq-enriched and control genes. RIP-seq control genes (1909) were defined as FDR ≤ 0.05 and log₂(IP/Input) < 1

Fig. 5

VIRMA associates with polyadenylation cleavage factors. a Western blots to validate potential proteomic targets of VIRMA, including polyadenylation cleavage factors CPSF5 and CPSF6, cleavage stimulation factor CSTF2T, poly(A)-binding protein PABP1, RNA polymerase II subunit POLR2B, and transcriptional intermediary factor TRIM28. b Effect of siRNA knockdown of CPSF5 and CPSF6 on m⁶A level of polyadenylated RNAs. P values calculated using one-tailed Student’s t-test between control and knockdown samples are >0.05. Data are represented as means ± SEM of n = 3. c, d Scattering plots of PDUIs in between control and VIRMA (c) and METTL3 (d) mutant cells, where 3′UTRs of mRNAs are significantly shortened or lengthened with FDR ≤ 0.05, absolute ΔPDUI ≥ 0.2, and at least twofold change of PUDIs. e Overlap of 3′UTR-lengthened genes regulated by VIRMA and METTL3 depletion effects. f Scattering plots of PDUIs in between siControl and siCPSF5 cells, where 3′UTRs of mRNAs are significantly shortened or lengthened with FDR ≤ 0.05, absolute ΔPDUI ≥ 0.2, and at least twofold change of PUDIs. g Pie chart shows percentages of m⁶A and non-m⁶A genes with 3′UTR shortening under the CPSF5 knockdown. Eighty-four percent of shortened genes are enriched with m⁶A modification. Chi-square test gives a P value < 1e-16. h The m⁶A peak density profiles of m⁶A genes in HeLa siControl and siCPSF5 cells grouped by their 3′UTR shortening induced by CPSF5 knockdown. i Differential gene expressions for 3′UTR-shortened m⁶A and non-m⁶A genes, respectively, after CPSF5 knockdown. RPKM stands for reads per kilobase per million mapped reads

Fig. 6

Correlation of 3′UTR m⁶A methylation with alternative polyadenylation. a The m⁶A peak density profiles of m⁶A genes in HeLa siControl and siCPSF5 cells. b m⁶A GGACU motif coverage versus motif distance from proximal APA site along mRNA transcript. c Proposed model for VIRMA-regulated m⁶A methylation specificity in 3′UTR and near stop codon and correlation of m⁶A methylation with APA