TI - DISCUSSION . AB - The results presented here further extend the SUBstrate specificity of the AlkB proteins , by demonstrating that also 1-meG lesions in RNA and 3-meT lesions in DNA are efficiently demethylated by AlkB . Thus , it now appears likely that all N1-methylated purine bases and N3-methylated pyrimidine bases in nucleic/ acids ( shown in Figure 1 ) are AlkB SUBstrates ( it has not yet been tested whether 3-methyluracil is demethylated by AlkB ) . 3-meT has previously been shown to block DNA polymerases when present in the DNA template or in the incoming deoxynucleotide ( 25 ) , and may therefore be a cause of both genomic instability and mutation . The introduction of 3-meT in DNA upon exposure to methylating agents is quite inefficient , compared with the level of 1-meA and 3-meC ( 26 ) . However , numerous studies have reported the introduction of 3-meT in DNA in vitro ( 16,18,19,27 ) , and also in vivo ( 21 ) , so it appears likely that this lesion poses a threat to genomic integrity in living cells . Compared with the 1-meA and 3-meC containing subSTRate ( obtained by DMS treatment ) , the 3-meT-containing subSTRate was a poorer subSTRate for AlkB and hABH3 ( hABH2 was approximately equally efficient on both SUBstrates ) . Thus , although the frequency of introduction of 3-meT lesions is quite low , this lesion may be as deleterious as the more frequent 1-meA and 3-meC lesions , due to its inefficient repair . Here , we have shown that [14C] MeI treatment of tRNA introduces 1-meG lesions , and that these lesions are repaired by AlkB . Previous studies have shown that 1-meG is the major methylation product obtained by MeI treatment of guanosine under basic conditions ( 28 ) , and that other methylating agents can induce 1-meG in RNA and DNA in vitro ( 16-18 ) and in vivo ( 22 ) . We also attempted to introduce 1-meG residues in DNA by treatment of a G-rich oligonucleotide with [14C] MeI , but were unsuccessful ( P O Falnes and R F Johansen , unpublished data ) . In our previous studies , we have titrated the amounts of AlkB and hABH3 required to demethylate the [14C] MeI-treated tRNA SUBstrate used in Figure 4B ( 15 ) , and we have also performed similar titrations for the [3H] methylated RNA oligonucleotide used in Figure 4A ( 23 ) . Several conclusions can be drawn by combining our previous data with those presented here . When treating the [14C] methylated tRNA subSTRate with hABH3 , the maximal extent of demethylation corresponded to the plateau level obtained with E.coli AlkB ( Figure 5 in ref 15 ) , indicating that 1-meG is also a SUBstrate for hABH3 . Furthermore , the titration curve of the activity of AlkB on the [14C] methylated tRNA subSTRate containing the AlkB SUBstrates 1-meG , 1-meA and 3-meC was not biphasic ( Figure 5 in ref 15 ) , and very similar to that obtained when using the 1-meA and 3-meC containing [3H] methylated RNA oligonucleotide ( Figure 4A in ref 23 ) . This may suggest that 1-meG , 1-meA and 3-meC in RNA are demethylated at similar efficiencies by AlkB . Kroger and Singer ( 29 ) studied the transcription of homopolymeric DNA templates and found that the introduction of 1-meG lesions completely abrogated transcription , while 1-meA or 3-meC only decreased its rate . These data demonstrate that methylation of guanosine in the N1-position severely compromises the ability of DNA to act as template for RNA polymerases , and suggest to us that 1-meG may also block DNA replication and mRNA translation , as is the case for 1-meA and 3-meC ( 6,15 ) . Similarly to 1-meA and 3-meC , 1-meG exists as a naturally occurring modified base in tRNA . The presence of 1-meG residues in some tRNAs is important for maintaining the proper reading frame in translation ( 30 ) , and bacterial mutants deficient in incorporating 1-meG in tRNA display decreased growth rates ( 31 ) . Thus , the ability of AlkB to demethylate 1-meG appears counterproductive in this context , and it will be interesting to study whether naturally occurring bases in tRNA are protected from AlkB -mediated demethylation , or whether a low level of such demethylation is actually tolerated . A recent article reported that the bases and nucleosides corresponding to the lesions 1-meA and 3-meC strongly stimulated the AlkB -mediated uncoupled decarboxylation of 2-oxoglutarate to succinate ( 12 ) . Here , we have measured the effect of 1-meG and 3-meT on this reaction but were not able to detect any stimulation . These data indicate that the interaction of the neutral bases 1-meG and 3-meT with the AlkB active site may be weaker than in the case of 1-meA and 3-meC , which are positively charged at physiological pH ( Figure 1 ) . This is also supported by the observation that AlkB was more efficient at relieving a replication block at the DMS-treated SUBstrate ( containing 1-mA and 3-meC ) , than at the 3-meT-containing oligo ( cf Figure 2C and F ) . Mechanistic insight on several enzymes that bind methylated bases in DNA and RNA has been obtained by studying the 3D structure of the protein in complex with a free methylated base or nucleoside ( 32,33 ) . This approach may also be useful in the case of AlkB , particularly , since bases and nucleosides corresponding to 1-meA and 3-meC apparently can interact with the AlkB active site , but are not AlkB SUBstrates . However , 1-meG and 3-meT may not be equally useful for this purpose ; their inability to stimulate the uncoupled , AlkB -mediated decarboxylation of 2-oxoglutarate may suggest that their interaction with the AlkB active site is weaker than in the case of 1-meA and 3-meC . Since the initial discovery that 1-meA and 3-meC in DNA are demethylated by AlkB ( 6,7 ) , the number of known AlkB SUBstrates has increased considerably ( 10,11 ) . Similarly , the family of putative human AlkB homologues has grown . For many years , only one such homologue was known ( 34 ) , but recent searches in protein databases have identified eight putative human AlkB proteins , denoted hABH1-hABH8 ( 8,10,13,14 ) . Of these , an enzymatic activity has only been demonstrated in the case of hABH2 and hABH3 , which , like E.coli AlkB , are able to demethylate 1-meA and 3-meC lesions in nucleic/ acids ( 10,14 ) . A wide range of roles can be imagined for the remaining proteins , such as repair of other lesions in nucleic/ acids or reversal of nucleic/ acid or protein methylations that regulate gene activity . These unsolved questions represent important challenges for future research .