Optimal protein synthesis requires bases in transfer RNAs to be modified. A key modification has been shown to involve an unusual two-step mechanism that entails the sequential activities of two enzymes. See Letter p.494
A central dogma of molecular biology is that DNA and the corresponding RNA are complementary. But this complementarity is rewired by a process called RNA editing, which recodes genomic information by modifying, deleting or inserting nucleotides in RNA transcripts. RNA editing has wide-ranging roles in various cellular processes and has clear implications in human disease1. The molecular and biochemical mechanisms underlying RNA editing have proved a challenge to scientists since its discovery 30 years ago in trypanosomes, a type of protozoan2. On page 494, Rubio et al.3 now provide intriguing evidence that editing of transfer RNAs is an unconventional two-step process, thus reasserting a valuable mechanistic concept for the field4.
An evolutionarily well-conserved editing process called base deamination converts RNA components known as nucleosides into other nucleosides: more specifically, it converts cytidine (C) into uridine (U), and adenosine (A) into inosine (I). In tRNAs, such conversions generate 'wobble' base pairs that are crucial for proper translation of the genetic code, and that maintain tertiary structure5,6,7. The enzyme that edits tRNA in archaea7 (which, along with bacteria, constitute the microorganisms known as prokaryotes) has been identified, but its counterpart in eukaryotes (organisms that include plants, animals, fungi and protozoa) has remained elusive.
Previous studies8 demonstrated that downregulation in vivo of the gene that encodes TbADAT2/3 — an enzyme that catalyses the deamination of A, converting it to I, in tRNAs in Trypanosoma brucei — leads to decreased C-to-U conversion in the tRNA (tRNAThr) that carries the amino acid threonine to nascent proteins during protein synthesis. This suggests that the deaminase TbADAT2/3 acts on C as well as A. However, purified TbADAT2/3 does not convert C to U in tRNAThr in vitro8.
On the basis of the observation that tRNAThr nucleoside residues can be methylated in various eukaryotes9, Rubio et al. posited that TbADAT2/3 might deaminate a methylated C (m3C) in tRNAThr. This would generate methylated uridine (m3U), and explain why TbADAT2/3 is unable to deaminate unmethylated tRNAThr (ref. 8). Indeed, the authors observed that purified tRNAThr from T. brucei contains both the m3C and m3U modifications.
Rubio and colleagues went on to identify the TbRM140a enzyme10 as the protein that potentially methylates tRNAThr. Intriguingly, they observed that TbRM140a alone was not sufficient to methylate tRNAThrin vitro, but that the m3C and m3U modifications both formed if TbADAT2/3 was added. Moreover, TbADAT2/3 could deaminate methylated tRNAThr only in the presence of TbRM140a, suggesting that these enzymes are mutually dependent on one another for activity. The authors hypothesized that the enzymes must interact directly for the observed methylation and deamination to occur, and performed in vivo and in vitro binding experiments that provided evidence of this interaction.
An earlier study8 had demonstrated that TbADAT2/3 could deaminate DNA in the bacterium Escherichia coli. Rubio et al. hypothesized that the interaction with TbRM140a might help to regulate TbADAT2/3 activity and prevent aberrant DNA deamination. Sure enough, they observed that coexpression of TbRM140a and TbADAT2/3 in E. coli leads to the bacteria generating fewer mutations than when TbADAT2/3 is expressed alone. This supports the idea that the TbRM140a–TbADAT2/3 interaction might prevent 'wholesale deamination' of the genome by TbADAT2/3 when it is localized within T. brucei nuclei.
Rubio and co-workers' findings suggest a provocative model for the mutual dependence of two enzymes that target a single tRNA residue for editing (Fig. 1), and raise several questions regarding the relationship between TbRM140a and TbADAT2/3. Strikingly, m3U is not generated during the initial reaction with TbRM140a and TbADAT2/3 that catalyses m3C formation, but only when m3C is incubated again with the two enzymes. Does the methylation activity of TbRM140a require the catalytic activity of TbADAT2/3, and vice versa? Studies featuring catalytically inactive mutants of the enzymes and mutations that disrupt direct binding of the enzymes to each other would address whether these proteins act as mutual scaffolds. Likewise, determining the affinity of the TbRM140a–TbADAT2/3 interaction, the stoichiometry of the TbRM140a–TbADAT2/3 complex and the specific activities of TbRM140a and of TbADAT2/3 alone will add invaluable insight to this stepwise mechanism for deamination.
The idea that formation of the TbRM140a–TbADAT2/3 complex prevents wholesale genome deamination in T. brucei nuclei suggests exciting future studies to work out the specificity of the reaction. The well-characterized RNA-editing enzyme APOBEC1 requires a cofactor to direct deamination to a specific residue on its target messenger RNA11. Do TbRM140a and TbADAT2/3 provide specificity for each other? APOBEC1 and another deaminase, AID, can deaminate single-stranded DNA (ref. 11). Is this also a feature of the TbRM140a–TbADAT2/3 complex? In which case, what regulatory mechanisms protect the cell from rampant methylation and deamination of single-stranded DNAs by TbRM140a–TbADAT2/3, both of which could have major detrimental consequences for cellular fitness?
In addition to its role of directly recoding RNA transcripts, RNA editing has been implicated in biological processes such as RNA silencing, splicing and innate immunity12. Hundreds of RNA-editing sites have been detected in humans13, and aberrant RNA editing is implicated in numerous neurological disorders1, emphasizing the need to understand the fundamental molecular and biochemical mechanisms involved. Methylation has previously been shown4 to occur before deamination in the A-to-I editing of prokaryotic tRNAs, but Rubio et al. show that this also occurs in C-to-U editing in eukaryotes. In doing so, they provide a valuable lesson for researchers hypothesizing mechanisms for RNA editing: sometimes nature doesn't take the most direct route to the desired destination.Footnote 1
Farajollahi, S. & Maas, S. Trends Genet. 26, 221–230 (2010).
Benne, R. et al. Cell 46, 819–826 (1986).
Rubio, M. A. T. et al. Nature 542, 494–497 (2017).
Grosjean, H. et al. Biochimie 78, 488–501 (1996).
Lonergan, K. M. & Gray, M. W. Science 259, 812–816 (1993).
Gerber, A. P. & Keller, W. Science 286, 1146–1149 (1999).
Randau, L. et al. Science 324, 657–659 (2009).
Rubio, M. A. et al. Proc. Natl Acad. Sci. USA 104, 7821–7826 (2007).
Jühling, F. et al. Nucleic Acids Res. 37, D159–D162 (2009).
Fleming, I. M. et al. Sci. Rep. 6, 21438 (2016).
Conticello, S. G., Langlois, M. A., Yang, Z. & Neuberger, M. S. Adv. Immunol. 94, 37–73 (2007).
Savva, Y. A., Rezaei A., St Laurent, G. & Reenan, R. A. Front. Genet. 7, 100 (2016).
Li, J. B. et al. Science 324, 1210–1213 (2009).