Dear Editor, Dynamic and reversible N6-methyladenosine (m6A) RNA methylation has been found to greatly impact gene expression, leading to the field of epitranscriptomics. 1 Unlike m6A that is an internal modification, a terminal modification at mRNA cap in higher eukaryotes exists, termed as N6, 2′-O-dimethyladenosine (m6Am)(Fig. 1 a). The first and sometimes the second nucleotide after the N7-methylguanosine (m7G) cap can be methylated at the 2′-hydroxyl group; and when the first nucleotide is 2′-O-methyladenosine (Am), it can be further methylated at the N6 position to become m6Am. m6Am was first identified in animal cells and virus mRNA in 1975 2; several years later the methyltransferase was partially purified and was proposed to be a species whose molecular weight is~ 65 KD. 3 Only very recently, m6Am was found to be reversible as well: the first m6A demethylase FTO also catalyzed the demethylation of m6Am, depending on its subcellular localizations. 4, 5 By changing FTO levels, m6Am at mRNA cap was also suggested to impair DCP2-mediated mRNA decapping. 4 However, the methyltransferase of m6Am is not unambiguously identified, significantly hindering the functional and mechanistic study of m6Am. To clearly identify the methyltransferase, we fractioned the cell lysates of HEK293 cells, which contain robust N6-methylation activity (Supplementary information, Fig. S1a). This activity was assayed by incubating the column fractions with a 25 nt, synthetic vaccinia virus RNA probe (Probe-1, see Supplementary information) that begins with m7GpppAm. We modified the purification route of cell lysates, based on the procedure originally reported 3 (Supplementary information, Fig. S1b), and subjected the fractions of high N6-methylation activity to protein identification by sensitive mass spectrometry. We then searched for proteins with putative methyltransferase domain or sequence motif in the list of more than 100 proteins detected by MS, and found a protein named “phosphorylated CTD-interacting factor 1”(or PCIF1)(Fig. 1 b; Supplementary information, Fig. S1c), which was bioinformatically proposed to be a DNA/RNA N6-adenosine methyltransferase. 6 PCIF1 was originally identified and named due to its ability to directly bind to the phosphorylated C-terminal domain of RNA polymerase II via its WW domain 7; hence it was speculated to play a role in mRNA biogenesis. However, no enzymatic activity has been reported for PCIF1.

To test whether PCIF1 possesses methyltransferase activity in vivo, we first knocked down PCIF1 in HEK293 cells by two independent siRNAs and confirmed the knockdown efficiency by qRT-PCR (Supplementary information, Fig. S2a). We then measured the level of m6Am in polyA+ RNA fraction after decapping using LC-MS/MS. We were able to observe a reduction of m6Am level upon PCIF1 knockdown (Fig. 1 c; Supplementary information, Fig. S2b); importantly, the level of the internal m6A modification remained unchanged (Fig. 1 d),