The simplest form of evolutionary continuity is provided by conservation of the anticodon hairpin (stem and loop) of tRNAs as the most ancient adaptor, which was charged at position 37 of the modern tRNA molecule, adjacent to the 3'-end of the anticodon (Woese, 1972). The CCH hypothesis has adopted this view (Szathmary, 1996, 1999). As discussed in those papers, such a hairpin offers an ideal transient binding by ribozymes through complementary structures (e.g. in the form of 'kissing' hairpins). Here we deal with two questions: the nature and synthesis of the chemical bond between the adaptor and the amino acid, and the growth of the tRNA molecule to its present form.
As suggested by Woese (1972) and Wong (1991), the link must have been established by a stable N-bond. Inspection of relevant contemporary modified bases (Fig. 8) suggests that the nature of this primordial form was through a carbamoyl group.
NHC ON HCHC O2H
NHC ON HCHC O2H
NHC ON HC H2CO2H
NHC ON HC H2CO2H
HH HO OH
Fig. 8 Nucleosides modified by amino acids in tRNA molecules (from Grosjean et al., 2004). (a) N6-threonylcarbamoyladenosine (hn6 A: N6-hydroxynorvalylcarbamoyladenosine); (b) N6-glycylcarbamoyladenosine; (c) glutamylqueuosine
Regarding the early plausibility of this link, we call attention to the experimental investigations of Taillades et al. (1998) who suggested that ^-protected N-carbamoyl-a-amino acids rather than free a-amino acids formed in the primitive hydrosphere, which serve as eminent starting points also for peptide synthesis. This type of metabolism may have been present even in the RNA world. Remarkably, a formal peptide bond arises by the coupling between the amino acid and adenine through the carbamoyl link!
If the original charging site was in the anticodon loop, then there must have been a stage in evolution when amino acids were still charged to the old position in this loop and, at the same time, already to the new 3'-end of the full tRNA molecule (Szathmary, 1999). Some of the initial charging persisted as modern tRNA modifications, provided translation evolved using tRNA molecules still charged in the loop (Woese, 1972). Put differently, translation has adapted to these molecules being charged, which partly explains why removal of these modifications disturbs translation today. It is remarkable that the prediction of dual charging of tRNA by synthetases turned out to be correct for tRNAAsp in Escherichia coli (Dubois et al., 2004). A paralog of a glutamyl-tRNA synthetase charges Glu to tRNAAsp at position 34 (wobble) to queuosine (Fig. 8c). This example is quite suggestive, even if the chemistry and exact location is different from what we consider important for the CCH scenario. We predict that other synthetase paralogs with tRNA modifying activity will be found in the future. In agreement with the view advocated here, Grosjean et al. (2004, p. 519) comment that 'this modified nucleoside might be a relic of an ancient code'.
We wish to comment further on position 37. It is adjacent to the third base of the anticodon, which is complementary to the first base of the codon. As mentioned above, there is good correlation between first codon base identity and amino acid biosynthetic family membership (Taylor and Coates, 1989). If it is true that position 37 was an important ancient charging site, then the third anticodon base was closest to it. If some of the amino acid transformations (analogous to the modern tRNA: synthetase system) took place in the context of the ancient adaptor-ribozyme synthetase relation, then in agreement with the co-evolution theory (Wong, 1975) we suggest that some amino acids could have been biosynthetically transformed while bound to the ancient adaptor (the anticodon hairpin). The immediate and amino acid-specific neighbourhood on the adaptor would have been position 36, to which the transforming ribozymes would have been sensitive. The correlation of the third anticodon base with biosynthesis could be a relic of the most ancient genetic code and adaptor charging.
We should also explain how the ancient adaptor could have grown to its current size (tRNA). Again, position 37 seems to convey a message. It is between position 37 and 38 that tRNA genes for Glu, His, Gln, and initiator Met has been found in the archeon Nanoarchaeum equitans (Randau et al., 2005a,b). The splicing endo-nuclease splices other intron-coding tRNA genes as well as the transcripts of the two halves of the split tRNA genes (Randau et al., 2005c). The latter is made possible by overhangs at the 3' and at the 5'-ends of the two transcripts, respectively. This archeon remains the only known organism that functions with split tRNA genes, so this is likely to be a derived (rather than ancestral) phenomenon, in contrast to the interpretation of Di Giulio (2006).
Nevertheless, an adjacent position (between 36 and 37) seems to be important. In Eubacteria there is a Group I self-splicing intron in the same position in several tRNA genes (Reinhold-Hurek and Shub, 1992; Haugen et al., 2005). It is this type of intron that can bind arginine with codonic binding sites (Yarus, 1989). Szathmary (1993) accepted the idea of Ho (1988) that Group I introns had once been primordial synthetases. This idea still seems promising to us, and experimental attempts at demonstrating such a function in some (mutant) version would be welcome. Preservation of these introns could be due to fortuitous self-insertion (reverse self-splicing) of these molecules into some anticodon loops (Szathmary, 1993).
Regarding the evolutionary growth of the anticodon hairpin to a full-blown tRNA, we think that ideas resting simply on single, major hairpin duplication (e.g. Widmann et al., 2005) are remote from what we want because half a tRNA is much bigger than a single anticodon loop. This question warrants careful investigation; here we merely call attention to the fact that Bloch et al. (1985) and Nazarea et al. (1985) found statistical evidence for tandem repeats of units of length 8-10 (centred around 9) in both tRNAs and rRNAs. Note that the Group I intron splits the anticodon hairpin into one piece of 10 nucleotides and another one of seven nucle-otides. We do not yet know what this could mean. Yet, the most direct evidence for the primitive ancestry of the anticodon arm is that the anticodon arms of tRNAs with complementary anticodons are also complementary, which is not true for the acceptor stem (Rodin et al., 1993).
Last but not least, we point out that the proposed ancestral mechanism for adaptor charging can explain a strange feature found in the acceptor stem. There seems to be a vestigial anticodon-codon pair in the 1-2-3 position, and opposite to in the tRNA acceptor stem (Rodin et al., 1996). The same investigation revealed that there are several tandem repeats in tRNAs, e.g. those of the -DCCA motif (D is the overhanging 'discriminator base' at the 3'end) and its complementary sequence. Szathmary (1999) presented a somewhat artificial scenario for the evolutionary growth of the anticodon arm to a tRNA with Rodin's anticodon-codon pair in the acceptor stem. Finally, we mention a resolution of the apparent conflict between the primitive ancestry of the anticodon arm and the idea that the anticodon-binding part of present day synthetases is regarded younger than the part binding the acceptor stem (Woese et al., 2000). Whereas ribozymes charged tRNAs both at the old (anticodon) and at the new (acceptor stem) positions by the cognate amino acids, most of the emerging protein synthetases charged them only at the new site.
Was this article helpful?