Splice site m6A methylation prevents binding of U2AF35 to inhibit RNA splicing

Abstract

The N⁶-methyladenosine (m⁶A) RNA modification is used widely to alter the fate of mRNAs. Here we demonstrate that the C. elegans writer METT-10 (the ortholog of mouse METTL16) deposits an m⁶A mark on the 3' splice site (AG) of the S-adenosylmethionine (SAM) synthetase pre-mRNA, which inhibits its proper splicing and protein production. The mechanism is triggered by a rich diet and acts as an m⁶A-mediated switch to stop SAM production and regulate its homeostasis. Although the mammalian SAM synthetase pre-mRNA is not regulated via this mechanism, we show that splicing inhibition by 3' splice site m⁶A is conserved in mammals. The modification functions by physically preventing the essential splicing factor U2AF35 from recognizing the 3' splice site. We propose that use of splice-site m⁶A is an ancient mechanism for splicing regulation.

Keywords: 3' splice site; METT-10; METTL16; SAM homeostasis; SAM synthetase; U2AF35/65; U6 snRNA; m(6)A; spermatogenesis; splicing.

Graphical abstract

Figure S1

Distribution of m⁶A in the worm transcriptome, related to Figure 1 (A) Protein sequence alignment of the methyltransferase domain of METTL16. h, Homo sapiens (NP_076991.3); m, Mus musculus (NP_080473.1); g, Gallus gallus (NP_001026773.1); x, Xenopus laevis (NP_001085334.1); z, Danio rerio (NP_001003611.1); c, Caenorhabditis elegans (NP_499247.2). Secondary structure features from the human METTL16 core methyltransferase domain (PDB: 6GT5) are indicated: α helices, β strands and η-3₁₀ helix. (B) Equimolar amounts of total or poly(A)+ RNA from the adult mouse testes and adult worms (C. elegans) were pre-mixed together before performing m⁶A-IP-seq. This allowed us to compare the m⁶A distribution between the species. The worm and the mouse RNAs reveal a similar amount of m⁶A-enriched sequences but only very low number of worm reads pile up as m⁶A peaks. Mean values ± s.d. are plotted (n = 3). (C) Analysis of mouse m⁶A peaks (peak counts are indicated within brackets). (D) Analysis of worm m⁶A peaks (peak counts are indicated within brackets). (E) A consensus sequence identified in the small number (176) of m⁶A peaks identified in worm poly(A)+ RNA. Its significance is not known. (F) RNA-seq analysis of wild-type (WT) and mett-10 (ok2204) knockout (KO) mutant worms showing loss of RNA coverage from the 5′ end of the mett-10 gene in the KO, consistent with the genomic deletion in the mutant. Biological replicas (n = 3) are plotted separately. (G) Multiple worm U6 snRNA transcripts were identified based on sequence homology to mouse Rnu6. The METTL16/METT-10 methylation consensus sequence and position of m⁶A (red arrowhead) are indicated. (H) Detection of m⁶A methylation in U6 snRNA from total RNA using the SCARLET method (STAR Methods). The method allows interrogation of site-specific methylation status (red arrowhead indicates the nucleotide position we examined). The thin-layer-chromatography (TLC) assay used in the protocol is shown. The total RNA is from wild-type (WT) or mett-10 KO worms, grown on nutrient-high or nutrient-low plates. m⁶A, refers to synthetic RNA oligos without (0%) or with (100%) m⁶A (Table S3), used here as positive controls for the experiment (see STAR Methods). A part (dotted box) of this image is reproduced as Figure 1J. (I) The loss of U6 snRNA methylation in the mett-10 KO results in slight increase of cellular U6 snRNA levels. Three input replicas are plotted separately for each tested genotype. (J) The loss of U6 snRNA methylation in the mett-10 KO does not result in overall change in counts of reads covering splice junctions, therefore has no drastic effect on general splicing. Three input replicas are plotted separately for each genotype.

Figure 1

Worm METT-10 is an m⁶A writer for U6 snRNA and SAM synthetase mRNA (A) Domain organization of the m⁶A writers: mammalian METTL16 and Caenorhabditis elegans METT-10. MTase, methyltransferase domain; VCR, vertebrate-conserved region. See also Figure S1A. (B) Quantification of RNA modifications in total and poly(A)⁺ RNA from mouse (Mus musculus), insect (silkworm, Bombyx mori), and worm (C. elegans) using liquid chromatography-tandem mass spectrometry (LC-MS/MS). The barplot shows the level of m⁶A in poly(A)⁺ RNA. (C) Scheme for mapping m⁶A sites catalyzed by worm METT-10 with m⁶A-IP-seq. Mouse testes RNA is used as an internal control. See also Figure S1B. (D) The METTL3/METTL14 methylation consensus motif (RRACH) is found on the majority of the mouse m⁶A peaks (total number of peaks in brackets). (E) Meta-analysis of the distribution of m⁶A reads over mouse and worm transcripts. (F) Scheme for identification of m⁶A targets of C. elegans METT-10 by m⁶A-IP-seq. See also Figure S1F. (G) Based on decreased m⁶A enrichment in mett-10 KO worms compared with the control wild type (WT), we identified the indicated transcripts to be targets of METT-10. See also Figure S1G. (H) Worm U6 snRNA is enriched in m⁶A-IP with total and poly(A)⁺ RNA, and this enrichment is lost in the mett-10 KO . The normalized counts (reads per million [rpm]) are plotted separately for biological replicates (n = 3). (I) Coverage of m⁶A-enriched reads along the worm U6 snRNA sequence identifies the adenosine (red arrowhead), which is part of the conserved UACm⁶AGAGAA motif, that is methylated. Methylation is lost in mett-10 KO worms. The normalized coverages (rpm) from three biological replicates are plotted separately. (J) Detection of U6 snRNA m⁶A (red arrowhead) in total RNA from WT control or mett-10 KO worms (in biological duplicates). The thin-layer chromatography (TLC) analysis used in the SCARLET method (STAR Methods) is shown. See also Figure S1H.

Figure 2

A 3′ splice site m⁶A inhibits splicing of SAM synthetase pre-mRNA (A) Mapping of m⁶A reads identifies the 3′ splice site adenosine (red arrowhead) of intron 2 in the sams-3/4 pre-mRNA as being methylated, and this methylation is lost in mett-10 KO worms. The METT-10 methylation consensus motif is highlighted. The normalized coverages (rpm) from three biological replicates are plotted separately. See also Figures S2A and S2B. The barplot shows quantification (rpm) of the reads mapping to the sams3/4 genomic window . (B) Normalized read coverage (rpm) along the sams-3 genomic locus shows uniformly increased exonic coverage and lower intron 2 coverage in the mett-10 KO, suggesting more efficient splicing. Three biological replicates are plotted separately. (C) Three sams-3 isoforms that differ in utilization of the methylated 3′ splice site are annotated in ENSEMBL. Quantification of the different sams-3 splice isoforms (rpm; STAR Methods) in WT and mett-10 KO worms shows an increase in the mature, fully spliced PC isoform in the KO. PC, protein-coding; AS, alternative splice; IR, intron-retained. Mean values ± SD are plotted (n = 3). The p values were calculated using t tests. ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001. (D) Read counts (DESeq2 normalized) for the three different sams genes in the poly(A)⁺ transcriptome from WT and mett-10 KO worms show an overall increase in the KO. The three biological replicates are plotted separately. See also Figure S2C.

Figure S2

The worm m⁶A writer METT-10 methylates the 3′ ss in an intron of the three sams homologs, related to Figure 2 (A) Three highly similar sams duplicated genes are present in the C. elegans genome. Splicing isoforms that differ in utilization of the methylated 3′ splice site (indicated with red arrowhead) are annotated in ENSEMBL. The cartoon shows the sams-4 genomic locus and the sams-3/-5 transcripts mapped to the sams-4 locus using BLAT. All sams-3/-4/-5 loci encode for the protein-coding isoform which uses the methylated 3′ splice site for splicing, but also non-coding variants where this splice site is not used. Coverage of m⁶A along the intron-exon boundary identifies the methylated adenosine at the 3′ splice site of the three SAM synthetase homologs sams-3, sams-4 and sams-5. The m⁶A-IP-seq coverage has insufficient resolution to identify the methylated adenosine (red arrowhead) in sams-5. However, methylation is completely absent for all three homologs in the mett-10 KO. Biological replicas (n = 3) are plotted separately. The METT-10 methylation consensus motif is highlighted. (B) Detection of m⁶A methylation at the 3ʹ splice site in the sams-3 pre-mRNA using poly(A)+ RNA from adult worms grown on a nutrient-high diet. The SCARLET method was used to specifically probe the 3ʹ splice site adenosine in intron2, but was undetectable. The thin-layer-chromatography (TLC) assay used in the protocol is shown. The same procedure was carried out with a dilution series of control synthetic RNA oligos (an equal mixture of oligos were the target adenosine is either methylated or unmethylated) mimicking the sams-3 target sequence. Methylation of the oligos can be detected, but efficiency drops with a decrease in the amount of the oligos used. (C) Loss of 3′ splice site methylation results in increased expression of SAM synthetase (sams) genes. Compare input reads in WT versus mett-10 KO from the m⁶A-IP-seq experiment. Biological replicas (n = 3) are plotted separately. The input data from this plot is reproduced in Figure 2D.

Figure 3

A conserved stem-loop structure containing the 3′ splice site identifies it for methylation by METT-10 (A) Position of m⁶A marks introduced by human METTL16 on the 3ʹ UTR of human MAT2A SAM synthetase as well as C. elegans METT-10 on the 3ʹ splice site of worm sams-3 pre-mRNA. A 30-nt RNA fragment (RNA-1; Table S3) spanning the intron 2-exon 3 boundary of the worm sams-3 gene is predicted to fold into a stem-loop structure, with the METT-10 methylation motif UACm⁶AGAAAC (red) present in the loop region. This is very similar to the substrate requirement of mammalian METTL16. (B) Purification of recombinant worm METT-10 and human METTL16 proteins for in vitro methylation assays. Shown are in vitro methylation assays with METT-10 and the indicated RNA substrates, based on the sams-3 intron 2/exon 3 junction sequence, using radioactive ¹⁴C-SAM as a methyl donor. The UACAGAAAC motif (red) and residues that were mutated (blue) are highlighted. The reaction products were resolved by PAGE and exposed to detect the radioactivity (¹⁴C) signal. (C) In vitro methylation with recombinant METT-10 and the RNA substrates, based on the sams-3 intron 2/exon 3 junction sequence, carrying mutations in the stem region. (D) Splicing of WT and mutant (MUT) transgene reporter constructs injected into worm gonads. A MUT construct with triple mutations (AAC→CUU) within the methylation consensus motif (in the exon 3 part) increases 3′ splice site use, producing higher amounts of the PC isoform. Barplots depict the mean relative proportion of individual isoforms ± SD (n = 4). The p values were calculated using t tests. ^∗p ≤ 0.05, ^∗∗∗p ≤ 0.001. See also Figure S3A for transgene analysis in the mett-10 KO background. (E) METT-10 consensus motif (red) and regions allowing secondary structure formation (yellow) are conserved in various worm species. Changes (green) in C. japonica are compensatory. (F) Sequence alignment of the genomic region at the intron-exon boundary of the SAM synthetase gene from different organisms. The METT-10/METTL16 methylation consensus motif is highlighted (blue). Shown are in vitro methylations with ~30-nt RNAs corresponding to the intron-exon boundary sequence, carried out with recombinant human METTL16. The reaction products were resolved by PAGE and exposed to detect the radioactivity (¹⁴C) signal. See also Figure S3B for the same reactions carried out with worm METT-10.

Figure S3

m⁶A methylation of a specific 3′ ss in SAM synthetase pre-mRNA requires a stem-loop structure, related to Figure 3 (A) Wild-type (WT) transgene reporter constructs based on the sams-3 gene were injected into mett-10 KO worm gonads and multiple independent progeny lines stably expressing them were derived. Splicing patterns were analyzed by RT-PCR analysis using primers specific to the reporter. A mutated (MUT) construct with triple mutations (AAC →CUU) within the methylation consensus motif (in the exon3) was also tested. Lack of 3′ splice site m⁶A methylation in the KO worms results in similar isoform levels from both WT and MUT constructs. Barplots depict mean relative proportion of individual isoforms ± s.d. (n = 3). See also Figure 3D. PC, protein-coding; AS, alternatively spliced; IR, intron-retained. (B) In vitro methylation assay using recombinant worm METT-10 protein and synthetic RNAs. The RNAs correspond to the intron-exon boundary of the SAM synthetase pre-mRNA from the indicated organisms, where the 3′ splice has the METT16/METT-10 methylation consensus motif. Note that the corresponding intron-exon boundary sequence in mouse Mat2a pre-mRNA has no consensus motif, unlike the confirmed METTL16 target site in its 3′ UTR. See also Figure 3F for the in vitro methylations with human METTL16. It appears that the worm METT-10 is inefficient on targets other than its own sams target site, while human METTL16 is active on all targets carrying the methylation consensus motif.

Figure 4

Worms methylate the 3′ splice site of the SAM synthetase transcripts to downregulate their expression in response to a nutrient-high diet (A) WT or mett-10 KO worms were grown on plates that were high or low in nutrients. Splicing of intron 2 in the sams-3 gene was monitored by RT-PCR analysis (biological duplicates are shown). Splicing of intron 2 in sams-3 is different between WT and KO worms only under nutrient-high diet conditions. Barplots depict the mean relative proportion of individual isoforms ± SD (done in biological duplicates). The p values were calculated using t tests. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01. See also Figure S4B for RNA-seq data. (B) Mapping of m⁶A-IP-seq reads (n = 3) from WT and mett-10 KO worms fed on nutrient-high or nutrient-low plates. The m⁶A coverage on the intron 2-exon 3 boundary of the sams-3 gene is shown. The normalized coverages (rpm) from three biological replicates are plotted separately. See also Figures S4A–S4C. The barplot shows quantification (rpm) of the reads mapping to the sams-3/4 genomic window shown. The read counts from three biological replicates are plotted separately. (C) A nutrient-high diet inhibits splicing of sams-3 intron 2 (RT-PCR analysis) in WT worms, as shown by an increased level of the AS isoform. Supplementing a nutrient-low diet with free methionine or vitamin B12 increases splicing inhibition. The barplots (mean ± SD) show quantification of the AS isoform band from three independent biological replicates. The nutrient-low and peptone-rich, nutrient-high media contained OP50 or the NA22 strain of E. coli. The p values were obtained by Tukey’s HSD after ANOVA. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001. (D) A simplified scheme showing the methionine and folate cycles. (E) Metabolomics analysis detecting the indicated metabolites. The p values were calculated using t tests and adjusted using Benjamini-Hochberg correction. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001. (F) Western blot analysis of knockin worms expressing SAMS-3-HA or METT-10-FLAG proteins under different diet conditions. One of the worm lines has intron 2 deleted (Δintron) in the sams-3-HA gene locus. See also Figures S4D–S4F. (G) Analysis of brood size in worms of the indicated genotypes (Table S5) when grown on nutrient-low or nutrient-high plates. Δintron, deletion of intron having the METT-10 methylated 3ʹ splice site. n = 3 independent experiments, each done in 2–5 technical replicates. The p values were calculated by two-way ANOVA followed by Tukey’s HSD. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001.

Figure S4

Diet-dependent change in m⁶A RNA methylation of the 3′ ss of SAM synthetase pre-mRNA in C. elegans regulates SAMS protein levels, related to Figure 4 (A) The m⁶A-IP-seq read coverages over the identical sams-3 and sams-4 intron 2/exon 3 boundary are shown. It shows a difference in methylation between the wild-type (WT) worms grown on nutrient-high and nutrient-low diets. Only the WT worms grown on nutrient-high plates show strong methylation of the 3′ splice site. When WT worms are grown on nutrient-low plates, the methylation is strikingly reduced. In mett-10 KO the methylation is absent. (B) Quantification of the RNA-seq reads mapping to the various sams-3 splice isoforms. PC, protein-coding isoform produced by correct use of the 3′ splice site in intron2; AS, alternatively spliced isoform due to use of an upstream 3′ cryptic splice site in intron2; IR, intron-retained isoform due to failure to use the 3′ splice site leading to intron2 being retained. All counts were normalized to library sizes (reads per million, rpm). Mean values are plotted ± s.d. (n = 3). Since there is almost no 3′ splice site methylation in WT worms grown on the nutrient-low plates, the removal of mett-10 therefore has little effect on sams-3 splicing (PC isoform). Consequently, both WT and mett-10 KO worms (under nutrient-low conditions) use the site for splicing and produce predominantly the correctly spliced protein-coding (PC) version of sams-3, at levels comparable to KO worms grown on nutrient-high plates. (C) Transgene reporter constructs based on worm sams-3 sequence were injected into wild-type worms to establish transgenic lines with stable expression. The constructs used had the wild-type (WT) sams-3 sequence or had mutations (MUT: AAC→CUU) in the methylation consensus motif (on the part that sits on exon 3). This mutation is shown to abolish 3′ splice site m⁶A methylation by recombinant worm METT-10 in vitro (Figure 3B). Three independent transgenic isolates expressing the constructs were used in the experiment. The worms were grown on either nutrient-high or nutrient-low plates. Splicing patterns were analyzed by RT-PCR with transgene-specific primers, and quantifications are shown below where mean relative proportions of individual isoforms are plotted ± s.d. A representative ethidium bromide-stained agarose gel showing the resolved cDNA products is shown. On nutrient-high plates, levels of the protein-coding (PC) isoform from WT construct is lower than that seen from the MUT construct, presumably due to 3′ splice site m⁶A methylation in the former. The levels of the PC isoform from both constructs are similar in the nutrient-low plates, presumably, as the former is not methylated under these conditions. (D) Western analysis of SAMS-3-HA expressed from knock-in worm lines with or without intron2 in the sams-3-HA genomic locus. The worms were grown in nutrient-high or nutrient-low plates (Table S4). Three biological replicates are shown. Quantified HA signal normalized to that from endogenous histone H3 levels is shown below, with the value in nutrient-low diet being set to 1. Levels of SAMS-3-HA is reduced in nutrient-high diet condition, and this reduction is attenuated in the absence of intron2 in the sams-3-HA locus. The lysate from replicate #2 was re-run in the gel shown in Figure 4F. (E) Western analysis of SAMS-3-HA expressed from knock-in worm lines, in the mett-10 background. Worms were grown on a nutrient-high or nutrient-low diet. Three biological replicates are shown. The lysate from replicate #2 was re-run in the gel shown in Figure 4F. (F) Western analysis of METT-10-FLAG expressed from knock-in worms grown in nutrient-high and nutrient-low plates. Three biological replicates are shown. Part of this image (replicate #1, dotted box) was reproduced in Figure 4F.

Figure 5

A 3′ splice site m⁶A inhibits splicing in human cells and blocks its recognition by U2AF35 (A) Worm transgene reporter constructs based on sams-3 were transfected into human HeLa cells, and splicing patterns were analyzed by RT-PCR. A MUT construct with triple mutations (AAC→CUU) within the methylation consensus motif (in the exon 3 part) increases 3′ splice site use, producing lower amounts of the AS isoform. The barplot depicts the mean relative proportion of the AS isoform to the sum of all isoforms ± SD (n = 3). The p value was calculated using a t test. ^∗∗p ≤ 0.01. See also Figure S5B for all replicates. (B) In vitro splicing assay with HeLa S3 nuclear extracts. The human β-globin pre-mRNA substrate is spliced correctly, whereas the same substrate with an m⁶A methylated 3′ splice site (ss) remains unspliced. The presence of the methyl mark on the exonic part does not inhibit splicing. A band corresponding to the lariat intermediate is visible in lanes where the substrate is spliced correctly. Substrates were incubated for different durations (time in minutes) with the extracts. See also Figure S5C. (C) ITC experiments reveal that the full-length (FL) yeast U2AF35 (stabilized with a fragment of yeast U2AF65; STAR Methods) strongly binds an unmethylated RNA substrate mimicking the 3′ ss (AG), whereas the presence of an m⁶A mark decreases affinity. The quality of the recombinant protein used is shown. See also Figure S6. (D) Splicing assays with the MINX pre-mRNA substrate. 3′ ss m⁶A does not inhibit splicing of this substrate, which has a strong polypyrimidine tract. (E) Mutations that weaken the polypyrimidine tract in MINX pre-mRNA make it sensitive to inhibition by 3′ ss m⁶A. The presence of the methyl mark on the exonic part does not inhibit splicing. (F) Sequence of the 3ʹ end of the intron in the splicing substrates, showing the polypyrimidine tract (bold) and the 3ʹ ss. A similar region from worm sams-3 pre-mRNA is also shown, with the consensus ss motif shown (bold). (G) Model showing how 3′ ss m⁶A methylation under nutrient-high conditions prevents binding of U2AF35, leading to inhibition of splicing of sams pre-mRNA in worms.

Figure S5

The 3′ ss m⁶A methylation-mediated splicing inhibition is conserved in human cells, related to Figure 5 (A) In vitro methylation assay using recombinant human METTL16 or worm METT-10 proteins and radioactive ¹⁴C-SAM as the methyl donor, using RNAs (two different lengths) corresponding to the intron 2-exon 3 boundary of the worm sams-3 gene. The methylation consensus motif (red) and target adenosine (in bold) are shown. The reaction products were resolved by PAGE, the gel was stained with Methylene Blue to reveal the RNAs (to assure similar levels), and exposed to detect the radioactivity signal (¹⁴C). The human METTL16 is able to recognize and methylate the worm sams-3 target site, allowing us to test worm transgene reporter constructs in human HeLa cells. See also Figure 5A, and below. (B) RT-PCR analysis of the transcripts produced from worm sams-3 transgene construct transfected into HeLa cells. Wild-type (WT) construct with the 3′ splice site which can be methylated by human METTL16 shows different splicing pattern when compared to the construct with mutations (MUT: AAC→CUU) in the methylation consensus motif (on the part that sits on exon 3). Compare ratios of alternatively spliced (AS) and correctly spliced protein-coding (PC) isoforms. Three biological replicates, each with three technical replicates, were used to quantify the individual isoforms and produce the barplot in Figure 5A. Part of this panel (replicate #1, dotted box) is reproduced in Figure 5A. (C) In vitro splicing assay shows that an artificially introduced 3′ splice site (3′ ss) m⁶A within the human beta-globin pre-mRNA abolishes its splicing in human HeLa nuclear extracts, with neither the fully spliced product nor the lariat intermediate being detected. Presence of a single exonic m⁶A has no effect on splicing. See also Figure 5B. A major RNA band (indicated with an asterisk) below the unspliced RNA substrate is an irrelevant non-ligated species leftover from production of the splint-ligated RNA substrate (see STAR Methods)

Figure S6

3′ ss m⁶A methylation blocks splicing by hindering its recognition by the U2AF35 splicing factor, Related to Figure 5 (A) Comparison of U2AF35 protein sequence among different species. The protein complex used for ITC experiments consists of the full-length U2AF23 (S. pombe U2AF35) and 93-161 aa of U2AF59 (S. pombe U2AF65). The Zinc Finger 1 (ZF1) and ZF2 domains in the yeast protein are highly similar to that in other organisms. The secondary structure features of S. pombe U2AF23 (PDB: 4YH8) is shown above the alignment: α helices, β strands and η-3₁₀ helix. The asterisks at the bottom of the alignment indicate residues coordinating the zinc ion, while the residues we mutated are indicated on the top. (B-D) Isothermal calorimetry (ITC) experiments reveal that the yeast U2AF35 (in complex with a fragment of U2AF65) strongly binds an unmethylated RNA substrate (5′CUAGG) mimicking the 3′ splice site AG, while presence of an m⁶A mark decreases the affinity. See Figure 5C. Two mutations of Arginine 35 that is involved in recognition of the splice site adenosine were made. A conservative mutation to a positively charged lysine (R35K) or to a non-conservative mutation to uncharged serine (R35S). The R35K mutation was made to see if the shorter side-chain of lysine could allow recognition of m⁶A. We also made a mutation in the serine 34, which is frequently mutated to phenylalanine or tyrosine in human cancers, so we tested the S34Y mutant. Importantly, the two mutations replacing arginine 35 reduced binding to the unmethylated RNA, all three mutations did not bind to the methylated RNA.

Figure S7

Creation of a mouse KI and cKO mutants for Mettl16, related to Figure 6 (A) Strategy for generation of the Mettl16 knockin (KI) point mutant mice. A part of the genomic sequence of the Mettl16 exon 5 and the predicted protein sequence encoded are shown. A single guide RNA (gRNA) targeting this region was used to guide Cas9 endonuclease activity and homology-mediated repair to introduce nucleic acid mutations that eventually result in the following amino acid changes: F187G (RNA binding mutant) and PP185-186AA (catalytic dead mutant). Sequence of part of the repair templates bringing the mutations (in red) are shown. (B) Examples of Sanger sequencing of genomic PCR to detect the WT, F187G and PP185-186AA Mettl16 alleles (from mouse tail DNA). Representative ethidium bromide-stained agarose gels showing resolved PCR products is shown. Primer sequences are provided in Table S3. (C) Strategy for creation of the floxed Mettl16 allele. The mouse line with floxed Mettl16 allele and an inserted FRT-flanked selection markers cassette (LacZ and neomycin) was obtained from the KOMP repository at UC, Davis. Animals were crossed to remove the selection markers (STAR Methods). Using further crosses, we then brought together the floxed (Mettl16^loxP) allele and the Mettl16 null allele (Mettl16 -). Crosses between Mettl16^loxP/- and Mettl16^loxP/+; vasa-Cre partners gave us the Mettl16^loxP/-; vasa-Cre mice = conditional knockout (cKO) mutant. In the cKO, the gene is deleted in the male and female germline (starting from embryonic day E14.5 in the male germline). Representative ethidium bromide-stained agarose gels showing resolved PCR products detecting the different alleles and Cre driver is shown. Primer sequences are provided in Table S3.

Figure 6

RNA m⁶A methylation activity of mouse METTL16 is essential for development and has the potential to methylate the 3′ ss of target RNAs (A) Analysis of knockin (KI) mouse mutants for Mettl16, with mutations abolishing catalytic activity or RNA binding. A structural model of human METTL16 (PDB: 6GFK) shows the two prolines (PP185–PP186) of the NPPF catalytic motif close to the bound SAH molecule, and a model of human METTL16 in complex with bound MAT2A hairpin RNA (PDB: 6DU4) shows the F187 that flips in to interact with the target adenosine upon substrate RNA binding. Introduced mutations are indicated. See also Figures S7A and S7B. Shown are genotypes of animals recovered in born litters from crosses between heterozygous Mettl16 knockin (KI) parents (Mettl16^KI/+). Homozygous KI mutants were not obtained for either mutation, indicating lethality. HET, heterozygous; HOM, homozygous KI. (B) Multiple-tissue western blot showing tissue-specific expression of mouse METTL16. A loading control is provided by detection of PARK7. (C) Representative picture of atrophied testes from a mouse with conditional (Vasa-Cre) deletion of Mettl16 in the germline. Such animals are infertile. See also Figure S7C. (D) Histology of adult mouse testes showing complete absence of germ cells in seminiferous tubules from mice with conditional (Vasa-Cre) deletion of Mettl16 in the germline. cKO, conditional KO. The control HET testis shows all different stages of germ cells, including post-meiotic round spermatids and elongated spermatids. (E) Scheme showing identification of putative targets of mammalian METTL16 on 3′ ss. The total numbers of 3′ ss checked and those recovered with the METTL16/METT-10 motifs are given. The predicted secondary structure of one such RNA (intron-exon boundary with 3′ ss) is shown. (F) In vitro methylation assays with recombinant human METTL16 and the indicated RNAs. The RNA sequence for mouse Mat2a is from the 3′ UTR, whereas for other mouse genes it spans the intron-exon boundary (sequences are shown below). Reactions were resolved by PAGE, and the radioactivity (¹⁴C) signal was detected. (G) Two transcripts that show increased splice junction reads specifically in Mettl16 KO embryos (morulae at E2.5 or blastocysts at E3.5), indicating increased use of that ss in the absence of METTL16. Genomic coordinates of the 3′ ss and the underlying sequence on the Crick strand are shown. In vitro methylation assays with RNAs spanning the intron-exon boundary show methylation of the 3′ ss by mammalian METTL16.

Figure 7

Conserved targets of METTL16-mediated m⁶A methylation activity and specialization of the C-terminal VCR in vertebrates SAM levels are highly regulated in vivo, and this is achieved by splicing regulation of the SAM synthetase RNA (sams-3 or MAT2A). Under high-SAM conditions, METT-10 m⁶A methylates a 3ʹ ss in sams-3 pre-mRNA to directly inhibit splicing, whereas methylation in the 3ʹ UTR of MAT2A by mammalian METTL16 leads to intron retention/decay of the RNA. Under low-SAM conditions, mammalian METTL16 binds hairpins in the 3ʹ UTR of MAT2A and uses its C-terminal VCR to stimulate splicing of the terminal intron, whereas in nematodes, absence of ss methylation allows normal splicing to proceed. The different mechanisms also highlight the different approaches to regulation of SAM levels: nematode METT-10 turns off SAM production, whereas mammalian METTL16 actively turns on SAM production.