Molecular biology

A fix for RNA

It has long been known that cells repair chemically or physically damaged DNA. But the discovery that damaged RNA can also be repaired may come as a surprise. What's more, some of the same enzymes are involved.

The 'central dogma' of biology states that DNA makes RNA makes protein, and that this relationship governs the flow of genetic information in almost all living organisms. Damage to any one of these molecules can subvert the normal information flow. Because DNA provides the ultimate repository for genetic information, however, and because DNA damage can lead to permanent sequence changes (mutations), the study of how cells repair damaged DNA has prevailed, to the extent that the repair of damaged RNA (and proteins) has been considered unlikely to even exist. This is in part because, unlike the genome, of which there are just two to four copies per cell, RNAs and proteins are present at tens to thousands of copies and are readily replaced. Damaged RNAs and proteins have thus been considered dispensable.

That view must now be revised, as we learn — from Aas et al.1 on page 859 of this issue — of the repair of damaged RNA in vivo. This exciting finding comes hard on the heels of the discovery of a brand new DNA-repair mechanism2,3,4, catalysed in the bacterium Escherichia coli by the enzyme AlkB, and in humans by related proteins. The RNA repair proceeds by the same mechanism.

DNAs, RNAs and proteins have all been reported to be chemically damaged by aberrant methylation — the addition of CH3 groups. These methyl groups might come from cellular methyl donors such as S-adenosylmethionine, or from environmental pollutants such as tobacco smoke. The consequences of RNA and protein methylation have not been well studied, but aberrant DNA methylation can have profound effects, ranging from the induction of mutations to programmed cell death and concomitant genome degradation (Fig. 1). Fortunately, cells are equipped with several means of repairing DNA that contains methylation damage5.

Figure 1: Major biological molecules and the consequences of damage to them.

DNA is needed to make RNA, which in turn is needed to make protein, and this relationship governs the flow of genetic information in almost all living organisms. Information can also flow back from RNA to DNA in the case of RNA viruses; and DNA molecules can be replicated in order to propagate life. DNA is the ultimate repository of information, so it is vital that it is repaired when damaged. If it is not repaired it may cause mutations, or it may be degraded as part of a programme of cell death. It has been thought that RNAs and proteins are dispensable, in part because new copies can be easily generated. But Aas et al.1 have found that RNA repair can take place, and may be physiologically important.

Recently, researchers discovered a new mechanism by which E. coli repairs its methylated DNA, namely oxidative demethylation of damaged cytosine (3-methylcytosine) and adenine (1-methyladenine) bases in DNA2,3 by the AlkB protein (Fig. 2). In working out how AlkB functions, the researchers were helped by a prior computational analysis by Aravind and Koonin6 that predicted how AlkB might fold in its active conformation. This analysis identified AlkB as a possible member of the iron-dependent dioxygenase family of enzymes, some of which catalyse oxidative demethylation reactions. Database searches revealed putative AlkB relatives not only in the human genome, but also in the genomes of several RNA viruses, leading Aravind and Koonin to predict that AlkB also acts on methylated RNA. Almost 20 years ago, Karran7 reported that methylated RNA bases can act as a substrate for DNA-repair methyltransferases in vitro, but the in vivo relevance of this was not pursued.

Figure 2: Mechanism for repairing damaged RNA and DNA.

Recently2,3, a new mechanism was discovered by which Escherichia coli repairs aberrantly methylated adenine and cytosine bases in its DNA. This process, shown here, involves the AlkB protein, which removes the methyl (CH3) groups. Aas et al.1 have now shown that the same protein can also fix similar errors in RNA. AlkB has three human relatives, one of which (hABH2) can repair DNA, and another (hABH3) RNA. Of the highlighted atoms, H is found in DNA and OH in RNA. Oxygen, iron and 2-oxoglutarate are also required for these reactions.

Aas et al.1 now show that Aravind and Koonin were correct in their prediction, and also that RNA repair is indeed relevant in vivo. The authors find, remarkably, that bacterial AlkB catalyses oxidative demethylation reactions on chemically methylated RNA as well as DNA. They also provide in vivo evidence that AlkB ensures the survival of chemically methylated bacterial viruses that have either RNA-based or DNA-based genomes.

As regards humans, three AlkB relatives have now been identified1,4,8 — hABH1, hABH2 and hABH3. Aas et al. also show that hABH2 and hABH3 have AlkB-like activity both in vivo and in vitro (Fig. 2). Biochemically, hABH3 works primarily on single-stranded RNA (and to a certain extent on single-stranded DNA), and hABH2 on single- and double-stranded DNA. The fact that these two enzymes have different subcellular localizations in human cells supports the idea that they have distinct cellular roles. hABH2 is located exclusively in the nucleus; in non-dividing cells it concentrates in nuclear regions called nucleoli, but during the DNA-replication phase of the cell-division cycle it redistributes to sites of DNA synthesis, where DNA transiently becomes single-stranded. hABH3 is located mainly in the nucleus, but also outside it. Its localization in the nucleus is diffuse, and Aas et al. presume that it is present at sites at which genes are active, and are being copied into messenger RNAs, in case these mRNAs become damaged.

In essence, then, the activities of these two human proteins add up to that of one bacterial protein, as E. coli AlkB repairs both DNA and RNA. It seems that, for each human protein, parameters have evolved to distinguish between RNA and DNA; structural studies will surely shed light on these parameters. Another area for future work will be to generate cells with defects in hABH2-mediated DNA repair or hABH3-mediated RNA repair, in an attempt to identify the relative contributions of each activity in protecting against methylation-induced toxicity.

Why, though, should it be necessary to repair damaged RNA? The answer could be that although DNA is the final arbiter of genetic information, RNA is essential for the most basic biological processes. RNA-based primer sequences are required for DNA replication; and mRNAs, transfer RNAs (tRNAs) and ribosomal RNAs (rRNAs) are all needed during the elaborate process of protein synthesis. Even the formation of peptide bonds by ribosomes (the cell's protein-making machines) turns out to require catalysis mediated by rRNAs9. Moreover, a battery of small, non-protein-coding RNAs regulates a variety of other cellular processes10.

So maintaining RNA integrity is important for proper cellular function. And repairing damaged RNA may be more efficient than destroying it and starting again. Ribosome assembly is a complex, energy-intensive process, and it is not hard to imagine that the thrifty repair of damaged rRNA would be preferable to disassembling or discarding an entire ribosomal particle. Likewise, it takes several hours to produce full-length mRNA copies of large genes, so the repair of damaged copies might make energetic sense. The repair of truncated tRNAs has been documented11, so even the repair of single chemically damaged bases in tRNAs does not seem unreasonable. Most of these RNAs exist as partial duplexes; it will be interesting to see whether AlkB and its relatives repair double-stranded RNA as efficiently as they do the single-stranded variety.

One further question is raised by the fact that many tRNAs and rRNAs have both 1-methyladenine and 3-methylcytosine bases as natural, enzyme-mediated modifications12. Can AlkB and hABH3 distinguish between these biological modifications and chemical damage? If so, how? And do these enzymes have any role in controlling the normal modification of RNAs? It might be that the RNA-demethylation activity of AlkB-like proteins evolved to regulate biological RNA methylation, and that the repair of aberrant, chemical methylation is fortuitous. Finally, the fact that similar mechanisms are used for both RNA and DNA repair suggests that these nucleic acids might overlap in other ways, with the activation of signal-transduction pathways being a distinct possibility. Damaged DNA is known to signal cell-cycle arrest and programmed cell death. In light of the work by Aas et al.1, we suggest that RNA damage might do the same.


  1. 1

    Aas, P. A. et al. Nature 421, 859–863 (2003).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Falnes, P. O., Johansen, R. F. & Seeberg, E. Nature 419, 178–182 (2002).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Trewick, S. C., Henshaw, T. F., Hausinger, R. P., Lindahl, T. & Sedgwick, B. Nature 419, 174–178 (2002).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Duncan, T. et al. Proc. Natl Acad. Sci. USA 99, 16660–16665 (2002).

    ADS  CAS  Article  Google Scholar 

  5. 5

    Friedberg, E. C., Walker, G. C. & Siede, W. DNA Repair and Mutagenesis (Am. Soc. Microbiol., Washington DC, 1995).

    Google Scholar 

  6. 6

    Aravind, L. & Koonin, E. V. Genome Biol. 2 (3), 1–8 (2001).

    Article  Google Scholar 

  7. 7

    Karran, P. Proc. Natl Acad. Sci. USA 82, 5285–5289 (1985).

    ADS  CAS  Article  Google Scholar 

  8. 8

    Wei, Y. F., Carter, K. C., Wang, R. P. & Shell, B. K. Nucleic Acids Res. 24, 931–937 (1996).

    CAS  Article  Google Scholar 

  9. 9

    Nissen, P., Hansen, J., Ban, N., Moore, P. B. & Steitz, T. A. Science 289, 920–930 (2000).

    ADS  CAS  Article  Google Scholar 

  10. 10

    Gottesman, S. Genes Dev. 16, 2829–2842 (2002).

    CAS  Article  Google Scholar 

  11. 11

    Reichert, A. S. & Morl, M. Nucleic Acids Res. 28, 2043–2048 (2000).

    CAS  Article  Google Scholar 

  12. 12

Download references

Author information



Corresponding author

Correspondence to Leona D. Samson.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Begley, T., Samson, L. A fix for RNA. Nature 421, 795–796 (2003).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.