- Email: [email protected]
The RNA world, the genetic code and the tRNA molecule
LETTERS
The RNA world, the genetic code and the tRNA molecule
Outlook
The RNA world, the genetic code and the tRNA molecule n the evolution of the enzyme–coenzyme complex, as presented by Szathmáry1, the fact that the (non-amino acid) coenzyme is not covalently bound to the RNA does not mean that the coenzyme is a fossil of the RNA world, a claim which is implicit in White’s model2, which covalently links the coenzyme to the RNA. In other words, two molecules that interact with weak bonds, as in Szathmáry’s model1, only weakly implies a common evolutionary history on the basis that they are not part of the same molecule. As they are not part of the same molecule we cannot conclude that nucleotide-like coenzymes are fossils of an earlier metabolic state, that is the RNA world, whereas White2 can. Consequently, Szathmáry’s model1 appears to be flawed as it does not covalently link the coenzyme to the RNA. Indeed, Fig. 2 in Szathmáry’s article1 clearly shows that the coenzyme is not a ribozyme fossil and cannot, therefore, imply the RNA world. Hence, this model differs substantially, both from White’s2 and from my own3. As Szathmáry1 suggests, the nucleotide-like coenzymes might have interacted with ribozymes by means of weak bonds. Nevertheless, the structures of several actual coenzymes, such as ATP and CoA, do not guarantee specificity in the coenzyme–ribozyme weak interaction because the basepairing appears to be confined primarily to the adenine of these coenzymes, and thus the ribozyme might have bonded to more than one coenzyme, with the risk of interfering with catalysis. For the coenzymes that have a more complex structure, such as nicotinamide adenine dinucleotide or flavin adenine dinucleotide, this interaction might have been better achieved4. However, this problem remains unresolved unless we postulate that the weak interaction also involved the non-nucleotide-like part of the coenzyme; but this would make the model meaningless, as it is the weak interaction between the nucleotide-like part of the coenzyme and ribozyme that implicates RNA enzymes and thus supports the RNA world hypothesis. However, even if this interaction was achieved, we must consider the above observations more generally; the coenzyme–ribozyme weak interaction is not a reliable indicator of a shared evolutionary history, and this is true in an absolute sense, even if this weak interaction did actually take place. More directly, the nucleotide coenzyme in Szathmáry’s model1 implicates a role for RNA only through presumed basepairing, which represents too weak a constraint to really testify to the past existence of RNA enzymes and hence the RNA world. Consequently, I believe that even if this model was effectively used by nature, we cannot use it to make inferences about the RNA world. The non-amino acid coenzyme, which in Szathmáry’s model1 represents the coenzymes that can be seen in the actual enzyme–coenzyme complexes, disagrees with the amino acid (and also nucleotide) nature of many actual coenzymes3,5,6. This casts doubt over the evolutionary bifurcation that Szathmáry1 hypothesizes for coenzymes,
I
0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01893-4
as the ones that should not have amino acids in the molecule might, in actual fact, have them. Although these coenzymes with a mixed nucleotide and amino acid nature give some support to Szathmáry’s coding coenzyme handles hypothesis, I believe that they are related to the origins of enzyme catalysis2,3,5,6 and, only in the later phases of the ribonucleopeptide world, to the origins of the genetic code3. Szathmáry goes against common belief 7 by suggesting that the anticodon loop preceded the acceptor stem of tRNAs1. If the anticodon loop did precede, in evolutionary terms, the identity determinants in the acceptor stem of tRNAs, the fact that the identity determinants are nucleotides that tell the aminoacyl-tRNA synthetases which amino acid to charge onto a specific tRNA, and the fact that two sites exist where these determinants are located (in the anticodon and the acceptor stem)8 would have made it unnecessary to transfer the identity determinants in the anticodon loop to the acceptor stem. Indeed, this would have created a second site for the identity determinants, which would have been superfluous as one already existed. (Szathmáry1 justifies this by claiming the need for ‘a relocation of the charged amino from the anticodon loop to the 39-end’; the anticodon could have continued with its function to become, as expected, the only site of identity determinants as there is no real reason to create a second site.) Whereas if, as is commonly believed7, the identity determinants present in the tRNA acceptor stem evolved before those of the anticodon, and I believe that it was the anticodons that evolved in the proximity of the 39-end of the hairpin structure’s stem9–11, then it becomes necessary to create the anticodon loop by transferring these anticodons. This was because of the need to improve the efficiency of protein synthesis, using hairpin structures10,11, by the completion of the tRNA molecule. All this took place through the direct duplication of hairpin structures12, which gave rise both to the complete tRNA molecules and to the simultaneous transfer of anticodons from the hairpin-structure stems to the anticodon loop9–11. In the light of this interpretation, the identity determinants in the acceptor stem of some tRNAs are fossils of anticodons that were once housed in the proximity of the 39-end in the hairpin structure’s stem9–11. Therefore, the evolution of tRNA from an anticodon hairpin to a longer hairpin structure with an operational code at its end, as hypothesized by Szathmáry1, does not have properties that justify certain stages of the origin of this fundamental molecule. Moreover, the longer hairpin structure1 already has an anticodon in the anticodon loop and one in the stem but it is still not a complete tRNA molecule. Its completion would entail the ‘insertion’ of two other stem-loops, which seems paradoxical as this would lead to tRNAs with different secondary structures and, thus, different from common observations. If this happened through duplication, it would indeed create a tRNA but the three loops would also introduce three TIG January 2000, volume 16, No. 1
Massimo Di Giulio [email protected] International Institute of Genetics and Biophysics, CNR, Via G. Marconi 10, 80125 Naples, Napoli, Italy. 17
Outlook
LETTERS
The RNA world, the genetic code and the tRNA molecule
anticodons. My model12 is based on the simple direct duplication of a hairpin structure that houses an anticodon in the stem but not in the loop, which immediately creates the complete tRNA molecule12, thus explaining, in mechano-evolutionary terms, the close relationship between the identity determinants in the acceptor stem of tRNAs and the anticodons9–11. In the Stop Press box following his article, Szathmáry1 also says that the Asp-tRNAAsn → Asn-tRNAAsn pathway might be analogous but not homologous between Thermus and Deinococcus. With reference to the RNA world, these pathways should be recognized as the most important molecular fossils13,14 that have ever reached us. If these pathways, such as Glu-tRNAGln → Gln-tRNAGln, which are present in at least two primary phyletic lines (Archaea and Bacteria)15 are an acquired and not an ancestral trait, then how could they have evolved? The mischarging of, for example, tRNAAsn with Asp, would have undermined the accuracy of protein synthesis, and therefore would have been highly selected against16. Moreover, if these pathways are not an ancestral trait, then we cannot understand why they should have evolved, as their function is in some cases carried out by the corresponding aminoacyl-tRNA synthetase. Thus we find ourselves facing
References 1 Szathmáry, E. (1999) The origin of the genetic code: amino acids as cofactors in an RNA world. Trends Genet. 15, 223–229 2 White, H.B. (1982) Evolution of coenzymes and the origin of pyridine nucleotides. In The Pyridine Nucleotide Coenzymes (Everse, J. et al., eds), pp. 1–17, Academic Press 3 Di Giulio, M. (1997) On the RNA world: Evidence in favor of an early ribonucleopeptide world. J. Mol. Evol. 45, 571–578 4 Burgstaller, R.R. and Famulok, M. (1994) Isolation of RNA aptamers for biological cofactors by in vitro selection. Angew. Chem. 33, 1084–1087 5 Reanney, D.C. (1977) Aminoacyl thiol esters and the origins of genetic specificity. J. Theor. Biol. 65, 555–569
an apparent paradox: pathways with an evolution that is both difficult to achieve and almost impossible to explain. Clearly, the most convincing explanation is that these pathways are the relics of the mechanism that led to the origin of genetic code organization, as envisaged by the coevolution theory17, because otherwise these pathways could not be easily interpreted in evolutionary terms. If, on the other hand, these pathways are an ancestral trait, then they imply that the ancient biosynthetic pathways between amino acids took place on tRNA-like molecules (as envisaged by the coevolution theory)17, which provided the tRNA-like molecules charged with the amino acids needed for protein synthesis. In this way, we can explain all the ‘anomalous’ observations regarding these pathways. For instance, we can eloquently explain the existence of primary phyletic lines without certain aminoacyl-tRNA synthetases, as their function becomes historically expendable because it is performed by the biosynthetic pathway taking place on the tRNA. But, above all, as these pathways are the most explicit manifestation of the transformations from precursor amino acids into product amino acids on which the coevolution theory is founded, they provide us with the interpretative key to the origin of genetic code organization17.
6 Wong, J.T. (1991) Origin of genetically encoded protein synthesis: a model based on selection for RNA peptidation. Orig. Life Evol. Biosph. 21, 165–176 7 Schimmel, P. et al. (1993) An operational RNA code for amino acids and possible relationship to genetic code. Proc. Natl. Acad. Sci. U. S. A. 90, 8763–8768 8 Giegé, R. et al. (1998) Universal rules and idiosyncratic features in tRNA identity. Nucleic Acids Res. 26, 5017–5035 9 Di Giulio, M. (1995) Was it an ancient gene codifying for a hairpin RNA that, by means of direct duplication, gave rise to the primitive tRNA molecule? J. Theor. Biol. 177, 95–101 10 Di Giulio, M. (1998) Reflections on the origin of the genetic code: a hypothesis. J. Theor. Biol. 191, 191–196
11 Di Giulio, M. (1999) The non-monophyletic origin of the tRNA molecule. J. Theor. Biol. 197, 403–414 12 Di Giulio, M. (1992) On the origin of the transfer RNA molecule. J. Theor. Biol. 159, 199–214 13 Wächtershäuser, G. (1988) Before enzymes and template: theory of surface metabolism. Microbiol. Rev. 52, 452–484 14 Danchin, A. (1989) Homeotopic transformation and the origin of translation. Prog. Biophys. Mol. Biol. 54, 81–86 15 Ibba, M. et al. (1997) Aminoacyl-tRNA synthesis: divergent routes to a common goal. Trends Biochem. Sci. 22, 39–42 16 Di Giulio, M. (1993) Origin of glutaminyl-tRNA synthetase: an example of palimpsest? J. Mol. Evol. 37, 5–10 17 Wong, J.T. (1975) A co-evolution theory of the genetic code. Proc. Natl. Acad. Sci. U. S. A. 72, 1909–1912
Reply: certain uncertainties about the origin of the genetic code i Giulio presents a concerted attack1 on the coding coenzyme handle (CCH) hypothesis2. Such debate in general is welcome, as much is at stake: our understanding of the core genetic mechanism linking genotype to phenotype. However, I disagree with almost everything he says. First, it is a mistake to think that coenzymes have a historical importance only if they have descended from molecules once covalently linked to primordial RNAs or ribonucleoprotein (RNP) particles. In fact, the methodology of biochemical retrodiction3 suggests, in agreement with previous ideas4,5, that coenzymes functioned in early metabolism by acting as catalysts without macromolecules in the first place. This implies a likely role for them serving also as reversibly bound cofactors in early evolution2. The methodology of inference about the plausible nature of the RNA world does not contain any condition on coenzymes
D
Eörs Szathmáry szathmary@ zeus.colbud.hu Dept of Plant Taxonomy and Ecology, Eötvös University, 2 Ludovika tér, H-1083 Budapest, Hungary. 18
TIG January 2000, volume 16, No. 1
being weakly or strongly bound to ribozymes6. It is worthwhile to quote Orgel7: ‘It is argued that the presence of a nucleotide in such cofactors as nicotinamide adenine dinucleotide (NAD) and flavin adenine dinucleotide (FAD) cannot be explained readily in terms of their specific functions; much simpler derivatives of nicotinamide, for example, have chemical properties very similar to the cofactor NAD. It is claimed that the only plausible explanation of the involvement of nucleotides is that the cofactors are relics of an RNA world in which catalysts were equipped with nucleotide handles to bind them by specific hydrogen bonds to polynucleotide carriers’. The postulated regular RNA handles of the CCH hypothesis stem from this idea, leading to a consistent, plausible and experimentally testable scenario. The current coenzymes of a mixed (amino acid plus nucleotide) nature can indeed ‘give some 0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01890-9
The RNA world, the genetic code and the tRNA molecule
Outlook
The RNA world, the genetic code and the tRNA molecule n the evolution of the enzyme–coenzyme complex, as presented by Szathmáry1, the fact that the (non-amino acid) coenzyme is not covalently bound to the RNA does not mean that the coenzyme is a fossil of the RNA world, a claim which is implicit in White’s model2, which covalently links the coenzyme to the RNA. In other words, two molecules that interact with weak bonds, as in Szathmáry’s model1, only weakly implies a common evolutionary history on the basis that they are not part of the same molecule. As they are not part of the same molecule we cannot conclude that nucleotide-like coenzymes are fossils of an earlier metabolic state, that is the RNA world, whereas White2 can. Consequently, Szathmáry’s model1 appears to be flawed as it does not covalently link the coenzyme to the RNA. Indeed, Fig. 2 in Szathmáry’s article1 clearly shows that the coenzyme is not a ribozyme fossil and cannot, therefore, imply the RNA world. Hence, this model differs substantially, both from White’s2 and from my own3. As Szathmáry1 suggests, the nucleotide-like coenzymes might have interacted with ribozymes by means of weak bonds. Nevertheless, the structures of several actual coenzymes, such as ATP and CoA, do not guarantee specificity in the coenzyme–ribozyme weak interaction because the basepairing appears to be confined primarily to the adenine of these coenzymes, and thus the ribozyme might have bonded to more than one coenzyme, with the risk of interfering with catalysis. For the coenzymes that have a more complex structure, such as nicotinamide adenine dinucleotide or flavin adenine dinucleotide, this interaction might have been better achieved4. However, this problem remains unresolved unless we postulate that the weak interaction also involved the non-nucleotide-like part of the coenzyme; but this would make the model meaningless, as it is the weak interaction between the nucleotide-like part of the coenzyme and ribozyme that implicates RNA enzymes and thus supports the RNA world hypothesis. However, even if this interaction was achieved, we must consider the above observations more generally; the coenzyme–ribozyme weak interaction is not a reliable indicator of a shared evolutionary history, and this is true in an absolute sense, even if this weak interaction did actually take place. More directly, the nucleotide coenzyme in Szathmáry’s model1 implicates a role for RNA only through presumed basepairing, which represents too weak a constraint to really testify to the past existence of RNA enzymes and hence the RNA world. Consequently, I believe that even if this model was effectively used by nature, we cannot use it to make inferences about the RNA world. The non-amino acid coenzyme, which in Szathmáry’s model1 represents the coenzymes that can be seen in the actual enzyme–coenzyme complexes, disagrees with the amino acid (and also nucleotide) nature of many actual coenzymes3,5,6. This casts doubt over the evolutionary bifurcation that Szathmáry1 hypothesizes for coenzymes,
I
0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01893-4
as the ones that should not have amino acids in the molecule might, in actual fact, have them. Although these coenzymes with a mixed nucleotide and amino acid nature give some support to Szathmáry’s coding coenzyme handles hypothesis, I believe that they are related to the origins of enzyme catalysis2,3,5,6 and, only in the later phases of the ribonucleopeptide world, to the origins of the genetic code3. Szathmáry goes against common belief 7 by suggesting that the anticodon loop preceded the acceptor stem of tRNAs1. If the anticodon loop did precede, in evolutionary terms, the identity determinants in the acceptor stem of tRNAs, the fact that the identity determinants are nucleotides that tell the aminoacyl-tRNA synthetases which amino acid to charge onto a specific tRNA, and the fact that two sites exist where these determinants are located (in the anticodon and the acceptor stem)8 would have made it unnecessary to transfer the identity determinants in the anticodon loop to the acceptor stem. Indeed, this would have created a second site for the identity determinants, which would have been superfluous as one already existed. (Szathmáry1 justifies this by claiming the need for ‘a relocation of the charged amino from the anticodon loop to the 39-end’; the anticodon could have continued with its function to become, as expected, the only site of identity determinants as there is no real reason to create a second site.) Whereas if, as is commonly believed7, the identity determinants present in the tRNA acceptor stem evolved before those of the anticodon, and I believe that it was the anticodons that evolved in the proximity of the 39-end of the hairpin structure’s stem9–11, then it becomes necessary to create the anticodon loop by transferring these anticodons. This was because of the need to improve the efficiency of protein synthesis, using hairpin structures10,11, by the completion of the tRNA molecule. All this took place through the direct duplication of hairpin structures12, which gave rise both to the complete tRNA molecules and to the simultaneous transfer of anticodons from the hairpin-structure stems to the anticodon loop9–11. In the light of this interpretation, the identity determinants in the acceptor stem of some tRNAs are fossils of anticodons that were once housed in the proximity of the 39-end in the hairpin structure’s stem9–11. Therefore, the evolution of tRNA from an anticodon hairpin to a longer hairpin structure with an operational code at its end, as hypothesized by Szathmáry1, does not have properties that justify certain stages of the origin of this fundamental molecule. Moreover, the longer hairpin structure1 already has an anticodon in the anticodon loop and one in the stem but it is still not a complete tRNA molecule. Its completion would entail the ‘insertion’ of two other stem-loops, which seems paradoxical as this would lead to tRNAs with different secondary structures and, thus, different from common observations. If this happened through duplication, it would indeed create a tRNA but the three loops would also introduce three TIG January 2000, volume 16, No. 1
Massimo Di Giulio [email protected] International Institute of Genetics and Biophysics, CNR, Via G. Marconi 10, 80125 Naples, Napoli, Italy. 17
Outlook
LETTERS
The RNA world, the genetic code and the tRNA molecule
anticodons. My model12 is based on the simple direct duplication of a hairpin structure that houses an anticodon in the stem but not in the loop, which immediately creates the complete tRNA molecule12, thus explaining, in mechano-evolutionary terms, the close relationship between the identity determinants in the acceptor stem of tRNAs and the anticodons9–11. In the Stop Press box following his article, Szathmáry1 also says that the Asp-tRNAAsn → Asn-tRNAAsn pathway might be analogous but not homologous between Thermus and Deinococcus. With reference to the RNA world, these pathways should be recognized as the most important molecular fossils13,14 that have ever reached us. If these pathways, such as Glu-tRNAGln → Gln-tRNAGln, which are present in at least two primary phyletic lines (Archaea and Bacteria)15 are an acquired and not an ancestral trait, then how could they have evolved? The mischarging of, for example, tRNAAsn with Asp, would have undermined the accuracy of protein synthesis, and therefore would have been highly selected against16. Moreover, if these pathways are not an ancestral trait, then we cannot understand why they should have evolved, as their function is in some cases carried out by the corresponding aminoacyl-tRNA synthetase. Thus we find ourselves facing
References 1 Szathmáry, E. (1999) The origin of the genetic code: amino acids as cofactors in an RNA world. Trends Genet. 15, 223–229 2 White, H.B. (1982) Evolution of coenzymes and the origin of pyridine nucleotides. In The Pyridine Nucleotide Coenzymes (Everse, J. et al., eds), pp. 1–17, Academic Press 3 Di Giulio, M. (1997) On the RNA world: Evidence in favor of an early ribonucleopeptide world. J. Mol. Evol. 45, 571–578 4 Burgstaller, R.R. and Famulok, M. (1994) Isolation of RNA aptamers for biological cofactors by in vitro selection. Angew. Chem. 33, 1084–1087 5 Reanney, D.C. (1977) Aminoacyl thiol esters and the origins of genetic specificity. J. Theor. Biol. 65, 555–569
an apparent paradox: pathways with an evolution that is both difficult to achieve and almost impossible to explain. Clearly, the most convincing explanation is that these pathways are the relics of the mechanism that led to the origin of genetic code organization, as envisaged by the coevolution theory17, because otherwise these pathways could not be easily interpreted in evolutionary terms. If, on the other hand, these pathways are an ancestral trait, then they imply that the ancient biosynthetic pathways between amino acids took place on tRNA-like molecules (as envisaged by the coevolution theory)17, which provided the tRNA-like molecules charged with the amino acids needed for protein synthesis. In this way, we can explain all the ‘anomalous’ observations regarding these pathways. For instance, we can eloquently explain the existence of primary phyletic lines without certain aminoacyl-tRNA synthetases, as their function becomes historically expendable because it is performed by the biosynthetic pathway taking place on the tRNA. But, above all, as these pathways are the most explicit manifestation of the transformations from precursor amino acids into product amino acids on which the coevolution theory is founded, they provide us with the interpretative key to the origin of genetic code organization17.
6 Wong, J.T. (1991) Origin of genetically encoded protein synthesis: a model based on selection for RNA peptidation. Orig. Life Evol. Biosph. 21, 165–176 7 Schimmel, P. et al. (1993) An operational RNA code for amino acids and possible relationship to genetic code. Proc. Natl. Acad. Sci. U. S. A. 90, 8763–8768 8 Giegé, R. et al. (1998) Universal rules and idiosyncratic features in tRNA identity. Nucleic Acids Res. 26, 5017–5035 9 Di Giulio, M. (1995) Was it an ancient gene codifying for a hairpin RNA that, by means of direct duplication, gave rise to the primitive tRNA molecule? J. Theor. Biol. 177, 95–101 10 Di Giulio, M. (1998) Reflections on the origin of the genetic code: a hypothesis. J. Theor. Biol. 191, 191–196
11 Di Giulio, M. (1999) The non-monophyletic origin of the tRNA molecule. J. Theor. Biol. 197, 403–414 12 Di Giulio, M. (1992) On the origin of the transfer RNA molecule. J. Theor. Biol. 159, 199–214 13 Wächtershäuser, G. (1988) Before enzymes and template: theory of surface metabolism. Microbiol. Rev. 52, 452–484 14 Danchin, A. (1989) Homeotopic transformation and the origin of translation. Prog. Biophys. Mol. Biol. 54, 81–86 15 Ibba, M. et al. (1997) Aminoacyl-tRNA synthesis: divergent routes to a common goal. Trends Biochem. Sci. 22, 39–42 16 Di Giulio, M. (1993) Origin of glutaminyl-tRNA synthetase: an example of palimpsest? J. Mol. Evol. 37, 5–10 17 Wong, J.T. (1975) A co-evolution theory of the genetic code. Proc. Natl. Acad. Sci. U. S. A. 72, 1909–1912
Reply: certain uncertainties about the origin of the genetic code i Giulio presents a concerted attack1 on the coding coenzyme handle (CCH) hypothesis2. Such debate in general is welcome, as much is at stake: our understanding of the core genetic mechanism linking genotype to phenotype. However, I disagree with almost everything he says. First, it is a mistake to think that coenzymes have a historical importance only if they have descended from molecules once covalently linked to primordial RNAs or ribonucleoprotein (RNP) particles. In fact, the methodology of biochemical retrodiction3 suggests, in agreement with previous ideas4,5, that coenzymes functioned in early metabolism by acting as catalysts without macromolecules in the first place. This implies a likely role for them serving also as reversibly bound cofactors in early evolution2. The methodology of inference about the plausible nature of the RNA world does not contain any condition on coenzymes
D
Eörs Szathmáry szathmary@ zeus.colbud.hu Dept of Plant Taxonomy and Ecology, Eötvös University, 2 Ludovika tér, H-1083 Budapest, Hungary. 18
TIG January 2000, volume 16, No. 1
being weakly or strongly bound to ribozymes6. It is worthwhile to quote Orgel7: ‘It is argued that the presence of a nucleotide in such cofactors as nicotinamide adenine dinucleotide (NAD) and flavin adenine dinucleotide (FAD) cannot be explained readily in terms of their specific functions; much simpler derivatives of nicotinamide, for example, have chemical properties very similar to the cofactor NAD. It is claimed that the only plausible explanation of the involvement of nucleotides is that the cofactors are relics of an RNA world in which catalysts were equipped with nucleotide handles to bind them by specific hydrogen bonds to polynucleotide carriers’. The postulated regular RNA handles of the CCH hypothesis stem from this idea, leading to a consistent, plausible and experimentally testable scenario. The current coenzymes of a mixed (amino acid plus nucleotide) nature can indeed ‘give some 0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01890-9