Human METTL16 is a N ⁶‐methyladenosine (m⁶A) methyltransferase that targets pre‐mRNAs and various non‐coding RNAs

Abstract

N ⁶‐methyladenosine (m⁶A) is a highly dynamic RNA modification that has recently emerged as a key regulator of gene expression. While many m⁶A modifications are installed by the METTL3–METTL14 complex, others appear to be introduced independently, implying that additional human m⁶A methyltransferases remain to be identified. Using crosslinking and analysis of cDNA (CRAC), we reveal that the putative human m⁶A “writer” protein METTL16 binds to the U6 snRNA and other ncRNAs as well as numerous lncRNAs and pre‐mRNAs. We demonstrate that METTL16 is responsible for N ⁶‐methylation of A43 of the U6 snRNA and identify the early U6 biogenesis factors La, LARP7 and the methylphosphate capping enzyme MEPCE as METTL16 interaction partners. Interestingly, A43 lies within an essential ACAGAGA box of U6 that base pairs with 5′ splice sites of pre‐mRNAs during splicing, suggesting that METTL16‐mediated modification of this site plays an important role in splicing regulation. The identification of METTL16 as an active m⁶A methyltransferase in human cells expands our understanding of the mechanisms by which the m⁶A landscape is installed on cellular RNAs.

Introduction

Modifications in cellular RNAs (collectively termed the “epitranscriptome”) have emerged as important regulators of many aspects of gene expression. They expand the chemical and topological properties of the four basic nucleotides, thereby regulating the functions and fates of RNAs. So far, ~150 different types of RNA modifications have been identified in nature, ranging from methylations of different positions of the bases to complex modifications that are installed in multistep reactions by the coordinated action of several modification enzymes 1. While transfer RNAs (tRNAs) and ribosomal RNA (rRNAs) are the most highly modified RNA species 2, 3, the development of transcriptome‐wide mapping approaches for several RNA modifications 4, such as N ⁶‐methyladenosine (m⁶A; 5, 6, 7, 8), pseudouridine (Ψ; 9, 10, 11), N ¹‐methyladenosine (m¹A; 12, 13) and 5‐methylcytidine (m⁵C; 14, 15), has uncovered a complex landscape of modified sites in messenger RNAs (mRNAs) and many classes of non‐coding (nc) RNAs, including small nucleolar RNAs (snoRNAs), small nuclear RNAs (snRNAs), long non‐coding RNAs (lncRNAs) and microRNAs (miRNAs; reviewed in 16).

m⁶A is the most abundant internal mRNA modification and the majority of modifications lie within a RRACH sequence motif (R = A or G; H = A, C or U) in close proximity to stop codons 5, 6. Functionally, m⁶A modifications can have diverse effects on RNA secondary structure by either destabilising RNA duplexes or, when present in single‐stranded RNAs, promoting base stacking, thereby enhancing RNA stability 17. Alternatively, m⁶A modifications can be specifically recognised by proteins (“readers”), which then regulate the splicing, localisation, translation and/or stability of mRNAs 18, 19, 20, 21, 22. Furthermore, m⁶A modifications are installed dynamically and specific modifications have been suggested to be reversed by the action of the dioxygenases FTO and ALKBH5 23, 24. Such dynamic changes in the m⁶A profiles of mRNAs have been shown to influence mammalian circadian rhythms, stem cell differentiation and development, and promote tumorigenesis (e.g. 25, 26, 27, 28).

A complex composed of METTL3, METTL14, WTAP, RBM15/15B and KIAA1429 has been shown to possess m⁶A methyltransferase activity and be responsible for installing a multitude of m⁶A modifications in diverse cellular RNAs 29, 30, 31. Structural and biochemical analyses revealed that METTL3 is an active S‐adenosylmethionine (SAM)‐dependent RNA methyltransferase whereas METTL14 contributes to RNA substrate binding 32, 33. Transcriptome‐wide identification of the binding sites of the METTL3–METTL14 complex revealed an overlap with previously mapped m⁶A sites and a consensus binding motif that corresponds to the RRACH sequence 29. While the METTL3–METTL14 complex is a prominent mRNA m⁶A methyltransferase (“writer”), numerous known m⁶A modifications are present in different sequence contexts and were not identified in the METTL3–METTL14 crosslinking immunoprecipitation (CLIP) data sets, implying that other m⁶A methyltransferases exist in human cells.

The human genome encodes many putative methyltransferases, often designated as methyltransferase‐like (METTL) proteins, and here, we characterised METTL16 as an m⁶A “writer” protein. Using in vivo crosslinking and analysis of cDNA (CRAC), we provide insight into the RNA interactome and target spectrum of this protein, and uncover interactions with pre‐mRNAs, lncRNAs and several ncRNAs including the U6 snRNA. We show that METTL16 is responsible for N ⁶‐methylation of an adenine within an evolutionarily conserved U6 sequence that base pairs with 5′ splice sites of pre‐mRNAs. Furthermore, our data reveal that this modification is introduced during early stages in U6 snRNP biogenesis.

Results and Discussion

METTL16 binds directly to the U6 snRNA in vivo

The discovery that many m⁶A modifications in cellular RNAs do not map within the binding sites of the METTL3–METTL14 m⁶A “writer” complex indicates that other m⁶A methyltransferases exist in human cells. METTL16 represents a strong candidate for such activity due to its homology to the Escherichia coli YbiN protein that is responsible for N ⁶‐methylation of A1618 in the 23S ribosomal RNA 34. To identify RNA substrates of METTL16 in vivo, we performed crosslinking and analysis of cDNA experiments (CRAC; 35, 36, 37). HEK293 cells expressing C‐terminally His₆‐PreScission protease site‐2x Flag (Flag)‐tagged METTL16 or the Flag tag were crosslinked using UV light at 254 nm (UV‐CRAC) or alternatively, were grown in the presence of the photoactivatable nucleoside analogue 4‐thiouridine prior to crosslinking at 365 nm (PAR‐CRAC). Crosslinked protein–RNA complexes were isolated, and RNAs were trimmed and radiolabelled before separation of complexes by polyacrylamide gel electrophoresis. After transfer to a nitrocellulose membrane, visualisation of radiolabelled RNAs by autoradiography revealed strong signals in both the UV‐CRAC and PAR‐CRAC METTL16‐Flag samples but not in the controls (Fig 1A), demonstrating that METTL16 associates with cellular RNAs. These RNAs were isolated, ligated to adaptors, and a cDNA library was generated and subjected to Illumina deep sequencing. The obtained sequence reads were mapped on the human genome and the distribution of reads mapping to different classes of RNA was analysed (Fig 1B). Interestingly, this revealed a significant increase in the proportion of sequences mapping to snRNA genes in the METTL16‐Flag samples compared to the controls, indicating association of METTL16 with snRNA(s) (Fig 1B). We also observed a modest increase in the proportion of reads mapping to rRNA sequences; however, analysis of the distribution of reads mapping to the rDNA unit encoding the 47S pre‐rRNA transcript did not reveal any specific crosslinking site of METTL16 and this enrichment therefore likely reflects non‐specific crosslinking arising from the presence of METTL16 in the nucleolus (Fig EV1; 38).

Figure 1. METTL16 crosslinks to the U6 snRNA in vivo .

1. HEK293 cells expressing METTL16‐Flag (METTL16) or the Flag tag (Flag) were crosslinked using UV light at 254 nm (UV) or grown in the presence of 4‐thiouridine prior to crosslinking at 365 nm (PAR). Protein–RNA complexes were affinity‐purified and bound RNAs trimmed, labelled at the 5′ end with ³²P and ligated to sequencing adapters. Protein–RNA complexes were separated by polyacrylamide gel electrophoresis, transferred to a nitrocellulose membrane and visualised by autoradiography.
2. RNAs isolated from the membranes described in (A) were subjected to reverse transcription and PCR amplification to generate cDNA libraries that were analysed by Illumina deep sequencing. The obtained sequence reads were mapped to the human genome and the relative proportion of UV‐CRAC reads mapping to different classes of RNAs is shown.
3. The relative distribution of reads mapping to different snRNAs in the UV‐ and PAR‐CRAC data sets derived from cells expressing METTL16‐Flag or the Flag tag is shown. Only snRNAs to which > 5% of the total snRNA hits mapped are labelled within the columns.
4. HEK293 cells expressing METTL16‐Flag or the Flag tag were UV crosslinked in vivo and protein–RNA complexes were isolated under denaturing conditions. Crosslinked RNAs were isolated, and inputs (0.2%) and eluates were analysed by Northern blotting using probes hybridising to the snRNAs indicated.

Figure EV1. METTL16 localises to the nucleoplasm and nucleolus.

Immunofluorescence was performed using an antibody against METTL16 in HEK293 cells (αMETTL16; green; upper panel), or alternatively, HEK293 cells expressing METTL16‐GFP were fixed and directly analysed by fluorescence microscopy (METTL16‐GFP; green; lower panel). Nuclear material was visualised by DAPI staining (blue) and overlays of the different channels are presented on the right (Merge). The scale bar represents 10 μm.

To determine which snRNA was crosslinked to METTL16, we analysed the normalised number of sequence reads mapping to each of the snRNAs (Fig EV2) and the relative distribution of the sequence reads between the five snRNAs of the major spliceosome (Fig 1C). Notably, 75 and 70% of the reads mapped to the U6 snRNA in the METTL16 UV‐CRAC and PAR‐CRAC experiments, respectively, compared to only 17 and 27% in the corresponding control samples (Fig 1C), strongly suggesting that METTL16 interacts with the U6 snRNA. To confirm the specificity of this interaction, we performed additional crosslinking experiments in which cells expressing METTL16‐Flag or the Flag tag were UV crosslinked in vivo and after isolation of protein–RNA complexes, co‐precipitated RNAs were detected by Northern blotting (Fig 1D). While the U1, U2, U4 and U5 snRNAs were barely detected in any of the eluates, the U6 snRNA was strongly enriched in the eluate from the METTL16‐Flag, but not in the control. Notably, the U6atac snRNA of the minor spliceosome was also not enriched, implying a specific interaction of METTL16 with the U6 snRNA.

Figure EV2. METTL16 crosslinks to the U6 snRNA .

• A, B
  The normalised number of reads mapping to different snRNAs in the UV‐CRAC (A) and PAR‐CRAC (B) data sets derived from cells expressing METTL16‐Flag or the Flag tag is shown.

METTL16 is an m⁶A methyltransferase that modifies the U6 snRNA

The identification of the U6 snRNA as a direct interaction partner of the putative m⁶A methyltransferase METTL16 in vivo suggests that it may be a modification substrate. The U6 snRNA (but not U6atac) has been shown to carry an m⁶A at position 43 39, but the enzyme responsible for installing this modification long remained elusive. To determine whether METTL16 is responsible for N ⁶‐methylation of A43 in U6, we first examined the distribution of CRAC sequence reads mapping to the U6 snRNA (Fig 2A). This revealed that the majority of reads mapped to the central region of the U6 snRNA close to A43, supporting the hypothesis that METTL16 methylates this residue.

Figure 2. METTL16 is a human m⁶A methyltransferase.

1. The distribution of sequence reads, obtained from the METTL16‐Flag and Flag CRAC experiments, mapping to the U6 snRNA sequence is given as hits per million mapped reads. The sequence context of the m⁶A43 in the U6 snRNA is shown below.
2. HeLa cells treated with non‐target siRNAs (siNT) or siRNAs targeting METTL16 (siMETTL16) were used to prepare whole‐cell extracts. Protein levels were analysed by Western blotting using antibodies against METTL16 or actin.
3. The levels of METTL16 in siRNA‐treated cells analysed in (B) were quantified and normalised to actin. Data from three independent experiments are presented as mean ± SD.
4. Total RNA extracted from siRNA‐treated HeLa cells (as in B) was incubated with an anti‐m⁶A antibody (αm⁶A) or with non‐specific IgGs (IgG). Complexes were retrieved on Protein G Sepharose and after thorough washing, co‐precipitated RNAs were isolated and analysed by Northern blotting using probes hybridising to the U6 or U2 snRNAs. Input = 0.5%.
5. The amounts of U6 in the αm⁶A eluates were quantified, normalised to U2 levels and are shown as a bar graph. Data from three independent experiments are presented as mean ± SD.
6. Schematic view of the base pairing interactions between the U6 snRNA (red), the U2 snRNA (grey) and a pre‐mRNA (black). The position of m⁶A43 within the evolutionarily conserved ACAGAG motif of the U6 snRNA is indicated. 5′SS—5′ splice site, BP—branch point.

In order to demonstrate the methyltransferase activity of METTL16 on U6, we first established RNAi against METTL16. After transfection of siRNAs specifically targeting METTL16, Western blotting using an antibody against endogenous METTL16 confirmed reduction in the protein level by up to 97% (Fig 2B and C). We then performed a methylated RNA immunoprecipitation (Me‐RIP) using an antibody that specifically recognises m⁶A nucleotides and RNA derived from cells that had been treated with control siRNAs or siRNAs targeting METTL16. Analysis of the Me‐RIP eluates by Northern blotting using probes hybridising to the U6 and U2 snRNAs revealed specific precipitation of both snRNAs in samples containing the anti‐m⁶A antibody but not non‐specific IgGs, confirming the presence of m⁶A modifications in these RNAs (Fig 2D; 39, 40). Interestingly, upon depletion of METTL16, the amount of U6 co‐precipitated with the m⁶A antibody was reduced by ~50%, while the amount of U2 present in the eluates was unaffected (Fig 2D and E), demonstrating that METTL16 is an m⁶A methyltransferase that specifically modifies U6. This finding is supported by a parallel study 41 where the action of METTL16 in installing the m⁶A43 modification of U6 was shown by an in vitro methylation assay. The twofold reduction in the amount of U6 precipitated by the m⁶A antibody may be due to the detection of modifications installed in the long‐lived U6 snRNA 42 prior to efficient METTL16 depletion or may reflect the presence of an additional METTL3‐mediated m⁶A modification at A76 of the U6 snRNA 29, 43. In contrast to the putative m⁶A76 modification, the m⁶A43 modification installed by METTL16 does not lie within a RRACH motif, suggesting that different sequence or structural elements influence modification target recognition by the two m⁶A methyltransferases.

Together with the U2 snRNA, U6 forms the catalytic core of the spliceosome 44, 45 and the m⁶A43 modification lies within a highly conserved ACAGAGA sequence of U6 that base pairs with the 5′ splice site of pre‐mRNAs (Fig 2F; 46, 47, 48). Mutations in this sequence impede such interactions and are lethal in yeast 49, 50, implying that the presence of the m⁶A within this part of the U6 snRNA plays an important role in the regulation of pre‐mRNA splicing. The precise function of this modification remains unknown; however, based on the recent structures of the human U4/U6.U5 tri‐snRNP and various (pre‐)catalytic spliceosomes (e.g. 45, 51, 52), it seems unlikely that this modification is reversible or that it exerts its effect via a reader protein, but rather it is anticipated to influence local secondary structure or base pairing interactions of this region of the U6 snRNA. Assembly of the spliceosome onto its substrate pre‐mRNAs is achieved by a combination of relatively weak interactions of both spliceosomal proteins and snRNAs with the pre‐mRNA. This suggests that U6‐m⁶A43 may fine‐tune snRNA–pre‐mRNA interactions, thereby regulating either 5′ splice site recognition or spliceosome assembly.

METTL16 associates with oligouridylated pre‐U6 snRNA during U6 snRNP assembly

Prior to its integration into the spliceosome, the U6 snRNA assembles as a U6 mono‐snRNP, then a U4/U6 di‐snRNP and a U4/U6.U5 tri‐snRNP (reviewed in 53). To determine in which context the m⁶A43 modification is introduced, we performed native RNA immunoprecipitation experiments using extracts from HEK293 cells expressing either METTL16‐Flag or the Flag tag. Co‐precipitated snRNAs were detected by Northern blotting, and while the U6 snRNA was efficiently recovered from the METTL16‐Flag extract but not from the control, the U4 and U5 snRNAs were not significantly enriched in either of the eluates (Fig 3A). This implies that METTL16 associates with the nascent U6 RNA or a U6 mono‐snRNP during the early stages of U6 maturation.

Figure 3. METTL16 interacts with oligouridylated pre‐U6 snRNA and co‐precipitates La, LARP7 and MEPCE .

1. Extracts from HEK293 cell lines expressing METTL16‐Flag or the Flag tag (Flag) were used in native immunoprecipitation experiments. Co‐precipitated RNAs were analysed alongside total RNA inputs (20%) by Northern blotting using probes hybridising to the indicated snRNAs.
2. Extracts from HEK293 cell lines expressing METTL16‐Flag or the Flag tag (Flag) were used for immunoprecipitation experiments in the presence (+) or absence (−) of RNase A. Inputs (1%) and eluates containing co‐precipitated proteins were analysed by Western blotting using antibodies against La, LARP7, MEPCE and DHX16.
3. The number of sequence reads containing non‐genomically encoded T's (U‐tails) mapping to the 3′ end of the U6 snRNA was determined in each of the CRAC data sets and is shown as the relative number of additional T's detected.

After transcription by RNA polymerase III in the nucleus, the pre‐U6 snRNA acquires a 5′ guanosine triphosphate cap 54 and undergoes 3′ oligouridylation 55, 56, leading to recruitment of the La protein to stabilise the transcript 57, 58. Formation of the mature 3′ end of U6 causes release of the La protein and recruitment of the LSm2‐8 complex. The U6 snRNP transits through the nucleolus where eight 2′‐O‐methyl groups and three pseudouridines are added by box C/D and box H/ACA snoRNPs, respectively 59, 60, before returning to the nucleoplasm. To further define the timing of m⁶A43 installation on the U6 snRNA, we next identified protein interaction partners of METTL16 by immunoprecipitation experiments followed by mass spectrometry. This revealed significant enrichment of the chaperone‐like La protein 61, the La‐associated protein LARP7 62 and the guanosine triphosphate capping enzyme MEPCE 63 in METTL16‐Flag eluates compared to the control samples. To confirm these interactions and ascertain whether they are formed via RNA, immunoprecipitation experiments were performed in the presence or absence of RNase A, using extracts from cells expressing METTL16‐Flag or the Flag tag, and the eluates were analysed by Western blotting using antibodies against endogenous La, LARP7, MEPCE or, as a control, the spliceosomal RNA helicase DHX16 (Fig 3B). This confirmed specific binding of METTL16 with La, LARP7 and MEPCE, but revealed that these interactions are RNA‐dependent (Fig 3B), implying that rather than reflecting direct protein–protein contacts, these interactions may be mediated by the U6 snRNA or other substrate RNAs of METTL16 that are also bound by these factors. The presence of MEPCE in METTL16 complexes could suggest that 5′ end capping and N ⁶‐methylation of A43 occur simultaneously. However, in the context of the 7SK RNP, binding of LARP7 to MEPCE has been shown to inhibit its catalytic activity and MEPCE then has a non‐catalytic function in promoting the interaction between LARP7 and the 7SK RNA 64. Although it is not yet known whether LARP7 and MEPCE form similar interactions on the U6 snRNA, by analogy, this could suggest that METTL16‐mediated modification of A43 takes place downstream of U6 5′ capping.

Since both La and LARP7 preferentially bind oligo(U) stretches at the 3′ ends of RNA polymerase III transcripts 57, 58, 65, we quantified the number of reads mapping to the 3′ end of U6 that contained additional, non‐genomically encoded T's (corresponding to U's in the METTL16‐associated RNAs; Fig 3C). This revealed that while only a very small number of sequences in the control CRAC data sets contained additional T's, sequence reads containing non‐genomically encoded T's were significantly more common in the CRAC data derived from METTL16‐Flag cells, with numerous reads containing up to eight additional T's (Fig 3C). These results strongly suggest that METTL16 associates with a nucleoplasmic, 5′‐capped, oligouridylated form of pre‐U6 that is stabilised by the presence of La, LARP7 and MEPCE.

METTL16 associates with various ncRNAs, lncRNAs and pre‐mRNAs

Interestingly, METTL16 has recently been reported to interact with the cancer‐associated MALAT1 lncRNA 38, 66, and the mRNA encoding the SAM synthetase MAT2A 41, indicating that METTL16 associates with additional cellular RNAs. To determine whether these RNAs are directly bound by METTL16, we analysed the number and distribution of sequence reads corresponding to these transcripts in the CRAC data sets derived from cells expressing METTL16‐Flag. We identified specific crosslinking sites of METTL16 in the MALAT1 lncRNA, with the highest density of mapped sequence reads occurring on the 3′ region that forms the RNA triple helix element required for nuclear expression (Fig 4A) 38. Direct binding of METTL16 to this sequence is supported by the identification of mutations and microdeletions that are introduced during cDNA library preparation at nucleotides crosslinked to amino acids (Fig 4B; 67). Consistent with the findings of a parallel study 41, we observed crosslinking of METTL16 to the 3′ exon of the MAT2A pre‐mRNA with the highest peaks observed over hairpin secondary structures required for METTL16‐dependent splicing of an otherwise retained intron (Fig 4C; 41).

Figure 4. Various pre‐mRNAs, lncRNAs and ncRNAs are bound by METTL16.

• A
  The relative distribution of CRAC sequence reads (UV‐CRAC—black; PAR‐CRAC—grey) mapping to the MALAT1 lncRNA is shown as hits per million mapped reads. Numbers below the schematic representation of the transcript indicate nucleotide positions.
• B
  Secondary structure of the MALAT1 triple helix with nucleotides UV crosslinked to METTL16 highlighted with grey circles.
• C–L
  The relative distribution of CRAC sequence reads (UV‐CRAC—black; PAR‐CRAC—grey) mapping to the MAT2A mRNA (C), the XIST lncRNA (D), the 7SK ncRNA (E), and the RBM3 (F), STUB1 (G), ISYNA1 (H), RYR1 (I), PALM2 (J), THSD4 (K) and LGR6 (L) mRNAs is shown above schematic views of the introns (black lines) and exons (black boxes) present in the transcripts. For large mRNAs (I–L), a magnified view of the region of the pre‐mRNA that is crosslinked to METTL16 is shown (top, middle) above the schematic view of the whole mRNA.
• M
  Extracts prepared from HEK293 cells expressing METTL16‐Flag or the Flag tag were used in immunoprecipitation experiments and co‐precipitated RNAs were isolated. RT–qPCR was performed to detect the indicated transcripts. Log2 fold changes between the levels of each transcript detected in the eluates of the METTL16‐Flag and the Flag control immunoprecipitations are shown. Data from three independent experiments are presented as mean ± SD.

To identify other METTL16‐bound RNAs, we performed peak calling on our CRAC data sets, and identified 355 mRNAs, 68 lncRNAs and nine ncRNAs that were specifically crosslinked to METTL16 in the UV‐ and PAR‐CRAC experiments (Tables EV1, EV2 and EV3). Interestingly, METTL16 crosslinking peaks were identified in the well‐characterised XIST lncRNA that participates in X chromosome inactivation during development (Fig 4D; 68), as well as the three vault RNAs (vtRNAs) and the four Y‐RNAs (Table EV3). Additional METTL16 substrates identified include the 7SL RNA of the signal recognition particle and the 7SK ncRNA, which is bound by MEPCE, LARP7 and La during its biogenesis (Fig 4E, Table EV3). Furthermore, METTL16 crosslinking peaks were identified in numerous mRNAs and notably, 93% of these peaks lie within introns, implying that METTL16 binds to a subset of pre‐mRNAs (see, for example, Fig 4F–L).

The RNAs identified by CRAC represent potential METTL16 modification substrates; however, it was shown that the presence of METTL16 on a pre‐mRNA can affect its splicing 41, raising the possibility that some METTL16‐associated RNAs may be bound, but not methylated. To highlight RNAs modified by METTL16, we compared the METTL16 binding sites identified by CRAC to previously identified m⁶A sites in the transcriptome 7, 8, 41, 43, 69, 70. This revealed that approximately 10, 25 and 83% of the CRAC peaks in mRNAs, lncRNAs and ncRNA, respectively, overlapped with m⁶A sites, implying that these transcripts are modified by METTL16. It is likely that particularly in the case of mRNAs, this represents an underestimate as the majority of the METTL16 interaction sites lie in introns, which are, compared to mature mRNA sequences, under‐represented in total RNA and therefore also in m⁶A‐seq data sets. In addition, the relatively low overlap between the available m⁶A data sets suggests that only a portion of the m⁶A sites in the transcriptome have been mapped so far. This may be especially relevant for the identification of METTL16‐mediated modifications, as in some m⁶A mapping approaches the positions reported as modified were determined based on the presence of a GAC or AAC motif that is recognised by METTL3–METTL14, but the mechanism of target recognition by METTL16 may be different.

To confirm the association of METTL16 with the proposed modification substrates as well as additional RNAs identified in the CRAC analyses, we performed RNA immunoprecipitation experiments using extracts from HEK293 cells expressing METTL16‐Flag or the Flag tag and monitored RNA enrichment by quantitative PCR. This confirmed highly significant increases in the amounts of U6 co‐precipitated with METTL16 compared to the control, while the amounts of U4 (which served as a control for non‐specifically precipitated RNAs) in the two eluates were not significantly different (Fig 4M). Furthermore, this revealed significant increases in the amounts of the XIST and MALAT1 lncRNAs, the 7SK ncRNA and eight tested pre‐mRNAs (MAT2A, RBM3, STUB1, ISYNA1, RYR1, PALM2, THSD4 and LGR6) co‐precipitated with METTL16 (Fig 4M), confirming that the targets identified by the CRAC experiments represent bona fide METTL16‐bound RNAs. Notably, for three of the selected pre‐mRNA targets, the crosslinking site of METTL16 was found to contain m⁶A modifications that were recently shown to be reduced in the absence of METTL16 41. While a reduction in m⁶A modification at many other sites will likely be due to the role of METTL16 in regulation of splicing of the SAM synthetase MAT2A that affects cellular SAM levels and thereby RNA methylation 41, the identification of METTL16 binding sites corresponding to the methylated positions on the RBM3, STUB1 and ISYNA1 pre‐mRNAs strongly supports these as METTL16‐mediated modifications.

Taken together, we demonstrate METTL16 as an active RNA methyltransferase that is responsible for N ⁶‐methylation of A43 of the U6 snRNA (Fig 5). While the precise influence of N ⁶‐methylation of A43 of the U6 snRNA remains unknown, the conservation of the ACm⁶AGAGA sequence throughout eukaryotes, its interactions with pre‐mRNA 5′ splice sites during the first catalytic step of splicing (Fig 5) and the finding that mutation of the modified position in yeast U6 is lethal 71 strongly suggest that METTL16‐mediated modification of this site is an essential step in human spliceosome assembly or function. The finding that METTL16 targets U6 and other highly structured lncRNAs and ncRNAs suggests that METTL16 may recognise specific secondary structure features in its substrates.

Figure 5. Model of the cellular functions of METTL16.

The m⁶A methyltransferase METTL16 binds diverse cellular RNAs, such as pre‐mRNAs, lncRNAs and several ncRNAs including the U6 snRNA. METTL16 associates with La, LARP7 and the methylphosphate capping enzyme MEPCE and installs U6‐m⁶A43 during the early stages of U6 snRNP biogenesis on an oligouridylated transcript. During splicing, A43 is involved in base pairing interactions with the 5′ splice site (5′SS) of pre‐mRNAs.

Transcriptome‐wide mapping of the binding sites of METTL16 on cellular RNAs also identified numerous pre‐mRNAs as putative targets (Fig 5). Interestingly, and in contrast to the enrichment of METTL3–METTL14‐mediated m⁶A modifications in close proximity to stop codons, the majority of METTL16 crosslinking sites were found in introns, implying that the two enzymes are responsible for distinct subsets of m⁶A modifications that likely have different cellular functions. While the functions of m⁶A in introns remain relatively unexplored, in Drosophila, it has been shown that an intronic m⁶A in the Slx (sex‐lethal) pre‐mRNA leads to alternative splicing and a sex bias towards maleness 72. Interestingly, the vast majority (87%) of the pre‐mRNA introns crosslinked by METTL16 are constitutively rather than alternatively spliced, suggesting that METTL16 might have diverse functions in regulating the fate of target RNAs. The identification of an additional m⁶A methyltransferase alongside the METTL3–METTL14 complex provides important insight into how the landscape of m⁶A modifications is installed in human cells and a basis for the dissection of the cellular functions of atypical m⁶A modifications.

Materials and Methods

Human cell culture, generation of stable cell lines and RNAi

For the generation of stable cell lines, the coding sequence of METTL16 (NM_024086.3) was cloned into a pcDNA5 vector for expression of the protein with a C‐terminal His₆‐PreScission protease site‐2x Flag (Flag) tag or for expression of the protein with a C‐terminal GFP tag. HEK293 stable cell lines expressing METTL16‐Flag or the Flag tag alone were generated using the Flp‐In T‐Rex system (Thermo Fisher Scientific) according to the manufacturer's instructions. HEK293 Flp‐In T‐Rex and HeLa CCL2 cells were cultured in 1× Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% foetal calf serum (FCS) and 2 mM glutamine at 37°C with 5% CO₂. For RNAi, cells were transfected with non‐target siRNAs or siRNAs targeting METTL16 (40 nM; siNT 5′‐CGUACGCGGAAUACUUCGA‐3′, siMETTL16 5′‐ATGGCTGGTATTTCCTCGCAA‐3′) using Lipofectamine RNAiMAX (Thermo Fisher Scientific) reagent according to the manufacturer's instructions and cells were harvested after 5 days.

In vivo crosslinking experiments

UV‐ and PAR‐CRAC experiments were essentially performed as previously described 35, 36, 37. In brief, HEK293 cells expressing the Flag tag or METTL16‐Flag were grown in DMEM and crosslinked at 254 nm (UV‐CRAC), or alternatively were grown in DMEM supplemented with 100 μM 4‐thiouridine for 6 h prior to crosslinking at 365 nm (PAR‐CRAC). Samples were then processed and the obtained sequence reads were mapped to the human genome (GRCh37.p13) as previously described. Peak calling was performed using Piranha 73 with a bin size of 20 bp and a significance threshold of P < 0.05.

Detection of co‐precipitated, crosslinked snRNAs by Northern blotting was performed as previously described 74, 75 using specific probes (U1 5′‐GGTCAGCACATCCGGAGTGC‐3′, U2 5′‐CATTTAATATATTGTCCTCGG‐3′, U4 5′‐CCAGTGCCGACTATATTGC‐3′, U5 5′‐GACTCAGAGTTGTTCCTCTCC‐3′, U6 5′‐GAACGCTTCACGAATTTGCGTGTC‐3′, U6atac 5′‐GGTTAGATGCCACGAAGTAGGTG‐3′).

Immunoprecipitation of complexes

HEK293 stable cell lines were treated with 1 μg/ml doxycycline for 24 h to induce expression of METTL16‐Flag or the Flag tag. Cells were harvested and lysed by sonication in a buffer containing 50 mM Tris–HCl pH 7.4, 200 mM NaCl, 0.5 mM EDTA, cOmplete mini‐EDTA‐free protease inhibitor cocktail (Roche). Glycerol and Triton X‐100 were added to final concentrations of 10 and 0.35%, respectively, and the extract was supplemented with 1.5 mM MgCl₂. The resulting lysate was cleared by centrifugation and then incubated with anti‐Flag M2 magnetic beads (Sigma‐Aldrich) for 2 h at 4°C in the presence or absence of 2 μM RNase A. Bound proteins or protein–RNA complexes were eluted from the beads by addition of 250 μg/ml Flag peptide (Sigma‐Aldrich) for 30 min at 4°C. Proteins were precipitated by addition of 20% trichloroacetic acid (TCA), separated by SDS–PAGE and analysed either by mass spectrometry 76 or by Western blotting using antibodies against the La protein (Proteintech, 11720‐1‐AP), LARP7 (Proteintech, 17067‐1‐AP), MEPCE (Proteintech, 14917‐1‐AP) and DHX16 (Bethyl, A301‐537A). Alternatively, RNA was extracted from the eluates using phenol:chloroform:isoamylalcohol (25:24:1) and ethanol‐precipitated. RNA was then either separated on a denaturing 10% polyacrylamide gel and analysed by Northern blotting using the specific probes listed above or alternatively, converted into cDNA using random hexamer primers and Superscript III reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Co‐precipitated mRNAs/lncRNAs were detected by quantitative RT–PCR in a LightCycler 480 using SYBR Green I Master (Roche) and the following primers: U6_F 5′‐GCTCGCTTCGGCAGCACATAT‐3′, U6_R 5′‐ATATGGAACGCTTCACGAATTTGC‐3′, U4_F 5′‐GCAGTGGCAGTATCGTAGCC‐3′, U4_R 5′‐AAAAATTGCCAGTGCCGACT‐3′, MALAT1_F 5′‐CAGGTGAACATAACAGACTTGGC‐3′, MALAT1_R 5′‐TGACCCTACTGAAGAGCATTGG‐3′, MAT2A_F 5′‐GGAGAAAGCTGACTTGGCTG‐3′, MAT2A_R 5′‐TAGGGCAAGCAGTCATGGAAT‐3′, XIST_F 5′‐CCCTACTGCGCTTTTTGCTG‐3′, XIST_R 5′‐AAATGCAAGGGCGACTGGTA‐3′, 7SK_F 5′‐CCAGGGTTGATTCGGCTGAT‐3′, 7SK_R 5′‐ATACCCTTGACCGAAGACCG‐3′, RMB3_F 5′‐TCGGCATTGAGTGAACCTGT‐3′, RBM3_R 5′‐TAGGTTGGCTGGGTGAGGTA‐3′, STUB1_F 5′‐GTGGATGAGAAGAGGAAGGTGAG‐3′, STUB1_R 5′‐CAGAGGCTGAAAAGGGGCACT‐3′, ISYNA1_F 5′‐CCAGCATCAGCTCCGAGGTA‐3′, ISYNA1_R 5′‐GCTGCACTCCAGGTGGTCAT‐3′, RYR1_F 5′‐GACCTCACTCAGGAAGCAGCATGG‐3′, RYR1_R 5′‐ACCAACGAGTGGCAGAGATGGG‐3′, PALM2_F 5′‐GTCCTTTTCCAAAGATAGCATAGTGGTTAAG‐3′, PALM2_R 5′‐AAAGCATACAGTTAGTAAGTGGCAGAGG‐3′, THSD4_F 5′‐ACAGTCACGAAGGGCACCA‐3′, THSD4_R 5′‐GCTGCATAAGGTCATAGGGCTCA‐3′, LGR6_F 5′‐TGCTGTGGTGTAATGGTTAAGTGTCTAG‐3′, LGR6_R 5′‐CCCTTGTTACAACTCCAACAGATAGGG‐3′.

Microscopy

Expression of METTL16‐GFP was induced in a HEK293 stable cell line by addition of 1 ng/ml doxycycline for 24 h before cells were fixed using 4% paraformaldehyde in PBS for 10 min at room temperature. Immunofluorescence was performed using a METTL16 antibody (Origene, TA504710) as previously described 77. Cells were mounted on coverslips using Vectashield^® (Vector labs) and analysed by confocal microscopy using a ConfoCor2 (Carl Zeiss) microscope.

Anti‐m⁶A Me‐RIP

Isolation of m⁶A‐containing RNAs was essentially performed as described in Ref. 78; 20 μg of total RNA prepared from HeLa cells that had been treated with non‐target siRNAs or siRNAs against METTL16 as described above was resuspended in 50 mM Tris–HCl pH 7.4 and 750 mM NaCl and denatured by treatment at 70°C for 5 min before incubation on ice. The RNA samples were then incubated with 5 μg of anti‐m⁶A antibody (Synaptic systems, 202‐211) overnight at 4°C in IP Buffer [50 mM Tris–HCl pH 7.4, 750 mM NaCl, 0.5% Igepal CA‐630, 400 U/ml RiboLock (Thermo Fisher Scientific), cOmplete mini‐EDTA protease inhibitor cocktail (Roche)]. The samples were then incubated for 2.5 h at 4°C with rotation with 15 μl of Protein G Sepharose (GE Healthcare) that had been pre‐equilibrated in IP buffer. After thorough washing steps, co‐precipitated RNAs were extracted using TRI Reagent (Sigma‐Aldrich) and analysed by Northern blotting using probes hybridising to the U6 and U2 snRNAs.

Data availability

The primary high‐throughput sequencing data of the UV‐CRAC and PAR‐CRAC experiments have been submitted to the GEO SRA database and assigned the identifier GSE103948.

Author contributions

ASW and PH performed crosslinking experiments, and JK and MTB carried out bioinformatics analysis. ASW, JK and PH carried out immunoprecipitation experiments followed by Northern blotting, Western blotting and qPCRs. ASW generated stable cell lines and analysed protein localisation. CL and HU performed and analysed the mass spectrometry. KES, CH and MTB designed the study and analysed the data. MTB and CH acquired funding, and KES and MTB wrote the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Table EV2

Table EV3

Review Process File

EMBO Reports (2017) 18: 2004–2014