The comprehensive interactomes of human adenosine RNA methyltransferases and demethylases reveal distinct functional and regulatory features

Abstract

N6-methyladenosine (m6A) and N6,2'-O-dimethyladenosine (m6Am) are two abundant modifications found in mRNAs and ncRNAs that can regulate multiple aspects of RNA biology. They function mainly by regulating interactions with specific RNA-binding proteins. Both modifications are linked to development, disease and stress response. To date, three methyltransferases and two demethylases have been identified that modify adenosines in mammalian mRNAs. Here, we present a comprehensive analysis of the interactomes of these enzymes. PCIF1 protein network comprises mostly factors involved in nascent RNA synthesis by RNA polymerase II, whereas ALKBH5 is closely linked with most aspects of pre-mRNA processing and mRNA export to the cytoplasm. METTL16 resides in subcellular compartments co-inhabited by several other RNA modifiers and processing factors. FTO interactome positions this demethylase at a crossroad between RNA transcription, RNA processing and DNA replication and repair. Altogether, these enzymes share limited spatial interactomes, pointing to specific molecular mechanisms of their regulation.

Figure 1.

BioID experimental set-up and validation. (A) Schematic representation of the BirA*-fusion protein constructs. (B) Overview of the BioID experimental design: stable inducible 293T cell lines expressing the BirA*-fusion proteins and the control BirA* and NLS-BirA* were prepared. Fusion protein expression and in vivo biotinylation was allowed for 24 h, cell lysates were prepared and biotinylated proteins were isolated by streptavidin-affinity purification. Protein pull-downs were analyzed by LC–MS/MS. Raw MS/MS data were analyzed independently and in parallel by Limma and SAINT analyses. (C) Enrichment (log2 fold change of METTL3-BirA* relative to BirA* control) of the known METTL3 complex auxiliary components on METTL3 C- and N- terminal BirA* fusion proteins affinity purifications. All shown protein hits have adjusted P-values < 0.01. (D) Protein neighbours network of METTL3 identified by BioID. Shared hits from filtered Limma and SAINT data are depicted.

Figure 2.

METLL16 BioID reveals proteins involved in the biogenesis of RNAs transcribed by the three nuclear RNA polymerases. (A) Representative biological processes (BP) gene ontology (GO) terms significantly enriched among the protein hits identified by METTL16 BioID. (B) Protein hits components of RNP particles or involved in the processing and/or modification of RNAs transcribed by RNAPI, II and III. (C) Protein neighbors network of METTL16 identified by BioID. Shared hits from filtered Limma and SAINT data are depicted.

Figure 3.

PCIF1 interacts with factors involved in RNAPII transcription initiation and co-transcriptional snRNA biogenesis. (A) Representative biological processes (BP) gene ontology (GO) terms significantly enriched among the protein hits identified by PCIF1 BioID. (B) Protein neighbours network of PCIF1 identified by BioID. Shared hits from filtered Limma and SAINT data are depicted. Proteins binding RNAPII are marked with an asterisk. Interactions detected also by Strep II-tag pull-downs are connected by grey lines. (C) Strep II-tag pull-down of PCIF1 or eGFP fusion proteins from whole-cell extracts from 293T cells. Western blot analysis show efficient pull-down and interaction between PCIF1 and Ser5-phosphorylated RNAPII. FT: unbound fraction. (D) Volcano plot representing enrichment (log₂ fold change) versus significance (–log₁₀ adjusted P-value) of RNAse A resistant interactions identify in LC-MS/MS analysis of Strep II-tag PCIF1 relative to Strep II-tag eGFP control. Complete list of significant interactions is in Supplementary Table S7. (E) RNAPII CTD serine (Ser) phosphorylation pattern during transcription initiation, elongation and termination. Protein hits bound to a specific type of Ser phosphorylation repeat are depicted. The position of RPRD2 on the gene body is only speculative.

Figure 4.

FTO on the crossroad between RNAPII transcription, pre-mRNA processing and DNA repair pathways. (A) Representative biological processes (BP) gene ontology (GO) terms significantly enriched among the protein hits identified by FTO BioID. (B) Protein neighbours network of FTO identified by BioID. Shared hits from filtered Limma and SAINT data are depicted. (C) Functional states of the fully assembled spliceosome and protein hits identified in FTO BioID. (D) 293T wildtype (WT) and FTO KO cells were treated with 2 mM hydroxyurea (HU) for the indicated hours followed by immunoblotting with antibodies against the indicated proteins (left). Quantification of DNA damage marker γ-H2AX signal normalized to tubulin signal from two independent experiments (mean ± S.D.) (right). (E) 293T WT and FTO KO cells (left) and U2OS cells transfected with FTO or non-target (#) siRNAs (right) were treated with 1 μM camptothecin (CPT) for the indicated hours followed by immunoblotting with antibodies against the indicated proteins.

Figure 5.

ALKBH5 interacts with mRNA export and pre-mRNA processing factors. (A) Representative biological processes (BP) gene ontology (GO) terms significantly enriched among the protein hits identified by ALKBH5 BioID. (B) Protein neighbours network of ALKBH5 identified by BioID. Shared hits from filtered Limma and SAINT data are depicted. Proteins reported to localize in the nuclear speckles are marked with an asterisk (white, identified by (112); orange, Uniprot database). Proteins containing an RNA recognition motif domain are underlined in red. Interactions detected also in Strep II-tag pull-downs are connected by grey lines. (C) Schematic overview of mRNA co- and post-transcriptional processing (left side). Proteins involved in pre-mRNA processing steps enriched in ALKBH5 BioID are listed (right side). (D) Strep II-tag pull-down of ALKBH5 or eGFP fusion proteins from whole-cell extracts from 293T cells. Western blot analysis show efficient pull-down. FT: unbound fraction. (E) Volcano plot representing enrichment (log2 fold change) versus significance (–log₁₀ adjusted p-value) of RNAse A resistant interactions identify in LC–MS/MS analysis of Strep II-tag ALKBH5 relative to Strep II-tag GFP control. Proteins detected also in BioID analysis are labelled. Complete list of significant interactions in Supplementary Table S7.

Figure 6.

m⁶A and m⁶Am MTs and DMTs have distinctive protein contacts in vivo. (A) Venn diagram representing the overlap of enriched BP GO terms for the protein hits of the four studied protein baits. The intersection of the four baits is represented with white borders; three baits with white and dashed borders and two baits with all dashed borders. (B) Enriched BP GO terms for the protein hits shared by two protein baits. Red, shared by PCIF1 and FTO. Yellow, shared by METTL16 and ALKBH5.