Portal de Periódicos da CAPES

Genetic Code Ambiguity

2002; Elsevier BV; Volume: 277; Issue: 48 Linguagem: Inglês

10.1074/jbc.m208093200

1083-351X

Leslie A. Nangle, Valérie de Crécy‐Lagard, Volker Döring, Paul Schimmel,

Bacterial Genetics and Biotechnology

The rules of the genetic code are established in reactions that aminoacylate tRNAs with specific amino acids. Ambiguity in the code is prevented by editing activities whereby incorrect aminoacylations are cleared by specialized hydrolytic reactions of aminoacyl tRNA synthetases. Whereas editing reactions have long been known, their significance for cell viability is still poorly understood. Here we investigated in vitro and in vivo four different mutations in the center for editing that diminish the proofreading activity of valyl-tRNA synthetase (ValRS). The four mutant enzymes were shown to differ quantitatively in the severity of the defect in their ability to clear mischarged tRNAin vitro. Strikingly, in the presence of excess concentrations of α-aminobutyrate, one of the amino acids that is misactivated by ValRS, growth of bacterial strains bearing these mutant alleles is arrested. The concentration of misactivated amino acid required for growth arrest correlates inversely in a rank order with the degree of the editing defect seen in vitro. Thus, cell viability depends directly on the suppression of genetic code ambiguity by these specific editing reactions and is finely tuned to any perturbation of these reactions. The rules of the genetic code are established in reactions that aminoacylate tRNAs with specific amino acids. Ambiguity in the code is prevented by editing activities whereby incorrect aminoacylations are cleared by specialized hydrolytic reactions of aminoacyl tRNA synthetases. Whereas editing reactions have long been known, their significance for cell viability is still poorly understood. Here we investigated in vitro and in vivo four different mutations in the center for editing that diminish the proofreading activity of valyl-tRNA synthetase (ValRS). The four mutant enzymes were shown to differ quantitatively in the severity of the defect in their ability to clear mischarged tRNAin vitro. Strikingly, in the presence of excess concentrations of α-aminobutyrate, one of the amino acids that is misactivated by ValRS, growth of bacterial strains bearing these mutant alleles is arrested. The concentration of misactivated amino acid required for growth arrest correlates inversely in a rank order with the degree of the editing defect seen in vitro. Thus, cell viability depends directly on the suppression of genetic code ambiguity by these specific editing reactions and is finely tuned to any perturbation of these reactions. Aminoacyl-tRNA synthetases (AARSs) 1The abbreviations used for: AARSs, aminoacyl-tRNA synthetases; Abu, α-aminobutyrate; CP1, connective polypeptide 1; MS, mineral standard medium; ValRS, valyl-tRNA synthetase. 1The abbreviations used for: AARSs, aminoacyl-tRNA synthetases; Abu, α-aminobutyrate; CP1, connective polypeptide 1; MS, mineral standard medium; ValRS, valyl-tRNA synthetase. catalyze the attachment of amino acids to their cognate tRNAs to establish the genetic code. To obtain the high degree of accuracy that is observed in translation, these enzymes must exhibit strict substrate specificity for their cognate amino acids and tRNAs. Recognition of tRNA by AARSs is facilitated by both positive and negative identity elements contained within the RNA structure that ensure binding to the proper enzyme (1Schimmel P. Biochemistry. 1989; 28: 2747-2759Crossref PubMed Scopus (95) Google Scholar, 2Felden B. Florentz C. Westhof E. Giegé R. Biochem. Biophys. Res. Commun. 1998; 243: 426-434Crossref PubMed Scopus (11) Google Scholar, 3Sissler M. Giegé R. Florentz C. RNA. 1998; 4: 647-657Crossref PubMed Scopus (16) Google Scholar). To select the correct amino acid AARSs must discriminate between all standard and nonstandard amino acids that are available in the cell, many of which are structurally similar. (Special amino acids such as formylmethionine (4RajBhandary U.L. J. Bacteriol. 1994; 176: 547-552Crossref PubMed Google Scholar) and selenocysteine (5Böck A. Forchhammer K. Heider J. Leinfelder W. Sawers G. Veprek B. Zinoni F. Mol. Microbiol. 1991; 5: 515-520Crossref PubMed Scopus (549) Google Scholar) are formed after the amino acid recognition step and are created by special postaminoacylation reactions.) Discrimination is primarily achieved by preferential binding of cognate amino acids to the correct active site and by exclusion of larger substrates (6Fersht A. Enzyme Structure and Mechanism. 2nd Ed. W. H. Freeman and Company, New York1985: 354-358Google Scholar, 7Jakubowski H. Goldman E. Microbiol. Rev. 1992; 56: 412-429Crossref PubMed Google Scholar). For several synthetases, including valyl-tRNA synthetase (ValRS) and isoleucyl-tRNA synthetase (IleRS), the affinity difference between the cognate amino acid and sterically similar amino acids is not sufficient to prevent errors in protein synthesis on the order of 0.5–1% (8Fersht A.R. Dingwall C. Biochemistry. 1979; 18: 2627-2631Crossref PubMed Scopus (120) Google Scholar, 9Schmidt E. Schimmel P. Biochemistry. 1995; 34: 11204-11210Crossref PubMed Scopus (78) Google Scholar, 10Lin L. Hale S.P. Schimmel P. Nature. 1996; 384: 33-34Crossref PubMed Scopus (92) Google Scholar), far greater than observed in vivo (∼0.0001%) (11Loftfield R.B. Vanderjagt D. Biochem. J. 1972; 128: 1353-1356Crossref PubMed Scopus (266) Google Scholar). To avoid errors, the enzymes evolved proofreading mechanisms that are tRNA-dependent (12Baldwin A.N. Berg P. J. Biol. Chem. 1966; 241: 839-845Abstract Full Text PDF PubMed Google Scholar, 13Eldred E.W. Schimmel P.R. J. Biol. Chem. 1972; 247: 2961-2964Abstract Full Text PDF PubMed Google Scholar, 14Schreier A.A. Schimmel P.R. Biochemistry. 1972; 11: 1582-1589Crossref PubMed Scopus (141) Google Scholar, 15Yarus M. Proc. Natl. Acad. Sci. U. S. A. 1972; 69: 1915-1919Crossref PubMed Scopus (92) Google Scholar). The combination of substrate selection and tRNA-dependent editing provides the low error rate observed in protein synthesis.ValRS catalyzes specific aminoacylation of tRNAVal in a reaction initiated by the binding of valine at the active site where it is condensed with ATP to form valyl adenylate (Val-AMP). The activated amino acid is subsequently transferred to the 3′-end of tRNAVal to generate Val-tRNAVal. Current beliefs define two distinct tRNA-dependent pathways for editing of mischarged amino acids. Pretransfer editing refers to hydrolysis of the misactivated aminoacyl-AMP (16Fersht A.R. Biochemistry. 1977; 16: 1025-1030Crossref PubMed Scopus (201) Google Scholar, 17Hale S.P. Schimmel P. Tetrahedron. 1997; 53: 11985-11994Crossref Scopus (13) Google Scholar, 18Hale S.P. Auld D.S. Schmidt E. Schimmel P. Science. 1997; 276: 1250-1252Crossref PubMed Scopus (79) Google Scholar, 19Nomanbhoy T.K. Hendrickson T.L. Schimmel P. Mol. Cell. 1999; 4: 519-528Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar, 20Hendrickson T.L. Nomanbhoy T.K. de Crecy-Lagard V. Fukai S. Nureki O. Yokoyama S. Schimmel P. Mol. Cell. 2002; 9: 353-362Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar). The second pathway, termed posttransfer editing, involves transfer of the misactivated amino acid to the 3′-end of the tRNA followed by deacylation of the mischarged species (13Eldred E.W. Schimmel P.R. J. Biol. Chem. 1972; 247: 2961-2964Abstract Full Text PDF PubMed Google Scholar, 14Schreier A.A. Schimmel P.R. Biochemistry. 1972; 11: 1582-1589Crossref PubMed Scopus (141) Google Scholar). In either case, the misactivated aminoacyl species must be transferred from the enzyme's active site to the center for editing (19Nomanbhoy T.K. Hendrickson T.L. Schimmel P. Mol. Cell. 1999; 4: 519-528Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar, 20Hendrickson T.L. Nomanbhoy T.K. de Crecy-Lagard V. Fukai S. Nureki O. Yokoyama S. Schimmel P. Mol. Cell. 2002; 9: 353-362Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar, 21Bishop A.C. Nomanbhoy T.K. Schimmel P. Proc. Natl. Acad. Sci. U. S. A. 2002; 99: 585-590Crossref PubMed Scopus (51) Google Scholar).The small sizes of certain amino acids that fit into the binding pocket for valine lead to enzyme-catalyzed misactivation (8Fersht A.R. Dingwall C. Biochemistry. 1979; 18: 2627-2631Crossref PubMed Scopus (120) Google Scholar). Specifically, threonine, cysteine, and the noncanonical α-aminobutyrate present a problem (7Jakubowski H. Goldman E. Microbiol. Rev. 1992; 56: 412-429Crossref PubMed Google Scholar, 22Jakubowski H. Fersht A.R. Nucleic Acids Res. 1981; 9: 3105-3117Crossref PubMed Scopus (135) Google Scholar). (Although it is considered a non-canonical amino acid due to its absence in normal cellular proteins, α-aminobutyrate is a latent metabolite and is a component of the cellular environment to which ValRS is exposed.) Each of these amino acids can be misactivated (by ValRS) and then cleared through enzyme proofreading activity. The editing pathways for ValRS for threonine (Thr) can be depicted as shown in Reaction Scheme 1. Pretransfer editing:ValRS(Thr−AMP)·tRNAVal→ValRS+Thr+AMP+tRNAValPost­transfer editing:ValRS+Thr−tRNAVal→Thr+tRNAVal+ValRSReaction Scheme 1 These editing reactions occur at a second site located in an insertion known as CP1 that splits the active site for aminoacylation (9Schmidt E. Schimmel P. Biochemistry. 1995; 34: 11204-11210Crossref PubMed Scopus (78) Google Scholar, 10Lin L. Hale S.P. Schimmel P. Nature. 1996; 384: 33-34Crossref PubMed Scopus (92) Google Scholar, 17Hale S.P. Schimmel P. Tetrahedron. 1997; 53: 11985-11994Crossref Scopus (13) Google Scholar, 23Nureki O. Vassylyev D.G. Tateno M. Shimada A. Nakama T. Fukai S. Konno M. Hendrickson T.L. Schimmel P. Yokoyama S. Science. 1998; 280: 578-582Crossref PubMed Scopus (314) Google Scholar, 24Fukai S. Nureki O. Sekine S. Shimada A. Tao J. Vassylyev D.G. Yokoyama S. Cell. 2000; 103: 793-803Abstract Full Text Full Text PDF PubMed Scopus (237) Google Scholar). The two catalytic centers are about 30 Å apart. Thus, the misactivated substrates have to be translocated from the active site to the center for editing (19Nomanbhoy T.K. Hendrickson T.L. Schimmel P. Mol. Cell. 1999; 4: 519-528Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar).In recent work, after mutagenesis of the whole chromosome, selections were done to obtain mutant strains of Escherichia coli that could substitute cysteine for valine in cellular proteins at a frequency sufficient to suppress in vivo an inactivating Cys → Val mutation in thymidylate synthase (25Döring V. Mootz H.D. Nangle L.A. Hendrickson T.L. de Crécy-Lagard V. Schimmel P. Marlière P. Science. 2001; 292: 501-504Crossref PubMed Scopus (162) Google Scholar). Five independent mutant strains were obtained. Each of these bore a single point mutation in the coding sequence for ValRS. All of the mutations were located at one of the highly conserved residues in the CP1 editing domain. (One of these mutant enzymes ([T222P]ValRS) was subsequently investigated in more detail, but no further characterizations were done with the other four.) In preliminary studies we found that the mutations were not equivalent, that is, each affected editing to a different degree. Given that the editing domain is conserved through evolution, including the most ancient organisms in the bacterial and archaebacterial kingdoms (26Schimmel P. Ribas de Pouplana L. Cold Spring Harbor Symposium on Quantitative Biology. 66 Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2001: 161-166Google Scholar), we imagined that cell growth would be sensitive to defects in the editing activity. Thus, these mutant enzymes afforded an opportunity to assess the relationship between the degree of the editing defect observed in vitro and the growth phenotype or toxicity seenin vivo. The mutant enzymes also provided further opportunity to evaluate the chemical role that specific conserved residues might have in the editing activity.DISCUSSIONTo further validate the conclusions presented here, some preliminary experiments were done with a different assay in vivo. For this purpose the five chromosome-borne valSmutant alleles were evaluated in the "cysteine suppression test" described previously (25Döring V. Mootz H.D. Nangle L.A. Hendrickson T.L. de Crécy-Lagard V. Schimmel P. Marlière P. Science. 2001; 292: 501-504Crossref PubMed Scopus (162) Google Scholar). The tester strain is deleted of its essential chromosome-borne thyA allele and harbors athyA mutant allele on a plasmid. The mutant thyAallele introduces a valine codon at position 146 of the coding sequence for thymidylate synthase, instead of the natural codon for the catalytic cysteine. These thyA cells are dependent on dT thymidine (dT) for growth. The presence of any of the five point mutations allowed growth of the tester strain in the presence of the cysteine precursor S-carbamoyl cysteine (SCC), presumably due to misincorporation of cysteine for valine at the 146 position. Cells with the wild-type valS allele do not grow without dT even in the presence of SCC. To achieve growth, strains harboring the T222P and K277Q valS alleles needed 0.2 mm SCC. In contrast, cells bearing the three other mutant alleles needed higher concentrations of SCC (0.5 mm) for growth. These results are consistent with data in Figs. 3 A and 4 A, showing that the T222P and K277Q mutant enzymes have the highest levels of mischarging in vitro and in vivo.The results presented here provide strong evidence that the editing domain is critical for enabling the modern genetic code. The correlation between the degree of the defect in editing in vitro (as seen in enzymes bearing specific point mutations in the CP1 domain) and the sensitivity in vivo to Abu is striking. This correlation is consistent with mischarging leading directly to the accumulation of proteins that are toxic due to errors of amino acid incorporation, that is, to ambiguity in the code. Thus, the errors of aminoacylation are not corrected once they have passed the surveillance of the CP1 editing domain.Although the editing reactions of tRNA synthetases have long been known, most prior studies have focused on studying the phenomenonin vitro. The conservation of the CP1 editing domain through all three domains of life implies strong selective pressure in vivo to retain this domain and the present work is consistent with this implication (26Schimmel P. Ribas de Pouplana L. Cold Spring Harbor Symposium on Quantitative Biology. 66 Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2001: 161-166Google Scholar). Under normal conditions, cells may not be exposed to concentrations of noncognate amino acids that are as severe as those used here to unmask the toxicity of a defect in editing. However, the efficiency of mischarging by the various mutants extended over a range of more than 40-fold (compare, for example, T222P and D230N ValRS (Fig. 3 A)). The correlation between efficiency of mischarging and the halo of toxicity, while not linear, was over this entire range (Fig. 4, A and B). This observation shows that even a mutation conferring a relatively small perturbation of editing activity can be toxic. By analogy, a small number of errors occurring under normal growth conditions probably confer a slight growth disadvantage to cells bearing a mutation at a critical residue in CP1. Exerted over many generations the disadvantage would select against the mutation and thus explain the strong conservation of CP1 through evolution.All of the mutations studied here had little effect on the active site for synthesis of Val-tRNAVal (Fig. 2). Possibly the CP1 editing domain originally acted alone and later was inserted into the active site. The feasibility of this possibility was demonstrated by experiments showing that the cloned CP1 editing domain can act as a separate protein in clearing mischarged tRNAs (10Lin L. Hale S.P. Schimmel P. Nature. 1996; 384: 33-34Crossref PubMed Scopus (92) Google Scholar). In sequence databases proteins have been identified that represent pieces of tRNA synthetases. Examples include "fragments" of synthetases for alanine, aspartate, glutamate, glycine, histidine, lysine, methionine, phenylalanine, serine, and tyrosine (35Ribas de Pouplana L. Schimmel P. Trends Biochem. Sci. 2001; 26: 591-596Abstract Full Text Full Text PDF PubMed Scopus (127) Google Scholar). In some instances specific functions for these proteins have been established (36Sissler M. Delorme C. Bond J. Ehrlich S.D. Renault P. Francklyn C. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 8985-8990Crossref PubMed Scopus (116) Google Scholar, 37Dong J. Qiu H. Garcia-Barrio M. Anderson J. Hinnebusch A.G. Mol. Cell. 2000; 6: 269-279Abstract Full Text Full Text PDF PubMed Scopus (315) Google Scholar, 38Qiu H. Dong J. Hu C. Francklyn C.S. Hinnebusch A.G. EMBO J. 2001; 20: 1425-1438Crossref PubMed Scopus (67) Google Scholar). However, none of these examples includes a segment of the CP1 domain. Further expansion of the data base over time may reveal CP1 as a separate protein in certain organisms. (If this occurs, then an important question is whether one or more class I enzymes in those organisms is missing that domain.) However, because the domain is found as a part of class I enzymes in organisms like Aquifex aeolicus that are thought to be near the base of the tree of life (26Schimmel P. Ribas de Pouplana L. Cold Spring Harbor Symposium on Quantitative Biology. 66 Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2001: 161-166Google Scholar), CP1 may have been incorporated into class I enzymes at the time of appearance of the last common ancestor. This early incorporation may have been essential for the subsequent appearance of the full tree of life, which required a highly developed and accurate genetic code that had eliminated ambiguity. Aminoacyl-tRNA synthetases (AARSs) 1The abbreviations used for: AARSs, aminoacyl-tRNA synthetases; Abu, α-aminobutyrate; CP1, connective polypeptide 1; MS, mineral standard medium; ValRS, valyl-tRNA synthetase. 1The abbreviations used for: AARSs, aminoacyl-tRNA synthetases; Abu, α-aminobutyrate; CP1, connective polypeptide 1; MS, mineral standard medium; ValRS, valyl-tRNA synthetase. catalyze the attachment of amino acids to their cognate tRNAs to establish the genetic code. To obtain the high degree of accuracy that is observed in translation, these enzymes must exhibit strict substrate specificity for their cognate amino acids and tRNAs. Recognition of tRNA by AARSs is facilitated by both positive and negative identity elements contained within the RNA structure that ensure binding to the proper enzyme (1Schimmel P. Biochemistry. 1989; 28: 2747-2759Crossref PubMed Scopus (95) Google Scholar, 2Felden B. Florentz C. Westhof E. Giegé R. Biochem. Biophys. Res. Commun. 1998; 243: 426-434Crossref PubMed Scopus (11) Google Scholar, 3Sissler M. Giegé R. Florentz C. RNA. 1998; 4: 647-657Crossref PubMed Scopus (16) Google Scholar). To select the correct amino acid AARSs must discriminate between all standard and nonstandard amino acids that are available in the cell, many of which are structurally similar. (Special amino acids such as formylmethionine (4RajBhandary U.L. J. Bacteriol. 1994; 176: 547-552Crossref PubMed Google Scholar) and selenocysteine (5Böck A. Forchhammer K. Heider J. Leinfelder W. Sawers G. Veprek B. Zinoni F. Mol. Microbiol. 1991; 5: 515-520Crossref PubMed Scopus (549) Google Scholar) are formed after the amino acid recognition step and are created by special postaminoacylation reactions.) Discrimination is primarily achieved by preferential binding of cognate amino acids to the correct active site and by exclusion of larger substrates (6Fersht A. Enzyme Structure and Mechanism. 2nd Ed. W. H. Freeman and Company, New York1985: 354-358Google Scholar, 7Jakubowski H. Goldman E. Microbiol. Rev. 1992; 56: 412-429Crossref PubMed Google Scholar). For several synthetases, including valyl-tRNA synthetase (ValRS) and isoleucyl-tRNA synthetase (IleRS), the affinity difference between the cognate amino acid and sterically similar amino acids is not sufficient to prevent errors in protein synthesis on the order of 0.5–1% (8Fersht A.R. Dingwall C. Biochemistry. 1979; 18: 2627-2631Crossref PubMed Scopus (120) Google Scholar, 9Schmidt E. Schimmel P. Biochemistry. 1995; 34: 11204-11210Crossref PubMed Scopus (78) Google Scholar, 10Lin L. Hale S.P. Schimmel P. Nature. 1996; 384: 33-34Crossref PubMed Scopus (92) Google Scholar), far greater than observed in vivo (∼0.0001%) (11Loftfield R.B. Vanderjagt D. Biochem. J. 1972; 128: 1353-1356Crossref PubMed Scopus (266) Google Scholar). To avoid errors, the enzymes evolved proofreading mechanisms that are tRNA-dependent (12Baldwin A.N. Berg P. J. Biol. Chem. 1966; 241: 839-845Abstract Full Text PDF PubMed Google Scholar, 13Eldred E.W. Schimmel P.R. J. Biol. Chem. 1972; 247: 2961-2964Abstract Full Text PDF PubMed Google Scholar, 14Schreier A.A. Schimmel P.R. Biochemistry. 1972; 11: 1582-1589Crossref PubMed Scopus (141) Google Scholar, 15Yarus M. Proc. Natl. Acad. Sci. U. S. A. 1972; 69: 1915-1919Crossref PubMed Scopus (92) Google Scholar). The combination of substrate selection and tRNA-dependent editing provides the low error rate observed in protein synthesis. ValRS catalyzes specific aminoacylation of tRNAVal in a reaction initiated by the binding of valine at the active site where it is condensed with ATP to form valyl adenylate (Val-AMP). The activated amino acid is subsequently transferred to the 3′-end of tRNAVal to generate Val-tRNAVal. Current beliefs define two distinct tRNA-dependent pathways for editing of mischarged amino acids. Pretransfer editing refers to hydrolysis of the misactivated aminoacyl-AMP (16Fersht A.R. Biochemistry. 1977; 16: 1025-1030Crossref PubMed Scopus (201) Google Scholar, 17Hale S.P. Schimmel P. Tetrahedron. 1997; 53: 11985-11994Crossref Scopus (13) Google Scholar, 18Hale S.P. Auld D.S. Schmidt E. Schimmel P. Science. 1997; 276: 1250-1252Crossref PubMed Scopus (79) Google Scholar, 19Nomanbhoy T.K. Hendrickson T.L. Schimmel P. Mol. Cell. 1999; 4: 519-528Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar, 20Hendrickson T.L. Nomanbhoy T.K. de Crecy-Lagard V. Fukai S. Nureki O. Yokoyama S. Schimmel P. Mol. Cell. 2002; 9: 353-362Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar). The second pathway, termed posttransfer editing, involves transfer of the misactivated amino acid to the 3′-end of the tRNA followed by deacylation of the mischarged species (13Eldred E.W. Schimmel P.R. J. Biol. Chem. 1972; 247: 2961-2964Abstract Full Text PDF PubMed Google Scholar, 14Schreier A.A. Schimmel P.R. Biochemistry. 1972; 11: 1582-1589Crossref PubMed Scopus (141) Google Scholar). In either case, the misactivated aminoacyl species must be transferred from the enzyme's active site to the center for editing (19Nomanbhoy T.K. Hendrickson T.L. Schimmel P. Mol. Cell. 1999; 4: 519-528Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar, 20Hendrickson T.L. Nomanbhoy T.K. de Crecy-Lagard V. Fukai S. Nureki O. Yokoyama S. Schimmel P. Mol. Cell. 2002; 9: 353-362Abstract Full Text Full Text PDF PubMed Scopus (60) Google Scholar, 21Bishop A.C. Nomanbhoy T.K. Schimmel P. Proc. Natl. Acad. Sci. U. S. A. 2002; 99: 585-590Crossref PubMed Scopus (51) Google Scholar). The small sizes of certain amino acids that fit into the binding pocket for valine lead to enzyme-catalyzed misactivation (8Fersht A.R. Dingwall C. Biochemistry. 1979; 18: 2627-2631Crossref PubMed Scopus (120) Google Scholar). Specifically, threonine, cysteine, and the noncanonical α-aminobutyrate present a problem (7Jakubowski H. Goldman E. Microbiol. Rev. 1992; 56: 412-429Crossref PubMed Google Scholar, 22Jakubowski H. Fersht A.R. Nucleic Acids Res. 1981; 9: 3105-3117Crossref PubMed Scopus (135) Google Scholar). (Although it is considered a non-canonical amino acid due to its absence in normal cellular proteins, α-aminobutyrate is a latent metabolite and is a component of the cellular environment to which ValRS is exposed.) Each of these amino acids can be misactivated (by ValRS) and then cleared through enzyme proofreading activity. The editing pathways for ValRS for threonine (Thr) can be depicted as shown in Reaction Scheme 1. Pretransfer editing:ValRS(Thr−AMP)·tRNAVal→ValRS+Thr+AMP+tRNAValPost­transfer editing:ValRS+Thr−tRNAVal→Thr+tRNAVal+ValRSReaction Scheme 1 These editing reactions occur at a second site located in an insertion known as CP1 that splits the active site for aminoacylation (9Schmidt E. Schimmel P. Biochemistry. 1995; 34: 11204-11210Crossref PubMed Scopus (78) Google Scholar, 10Lin L. Hale S.P. Schimmel P. Nature. 1996; 384: 33-34Crossref PubMed Scopus (92) Google Scholar, 17Hale S.P. Schimmel P. Tetrahedron. 1997; 53: 11985-11994Crossref Scopus (13) Google Scholar, 23Nureki O. Vassylyev D.G. Tateno M. Shimada A. Nakama T. Fukai S. Konno M. Hendrickson T.L. Schimmel P. Yokoyama S. Science. 1998; 280: 578-582Crossref PubMed Scopus (314) Google Scholar, 24Fukai S. Nureki O. Sekine S. Shimada A. Tao J. Vassylyev D.G. Yokoyama S. Cell. 2000; 103: 793-803Abstract Full Text Full Text PDF PubMed Scopus (237) Google Scholar). The two catalytic centers are about 30 Å apart. Thus, the misactivated substrates have to be translocated from the active site to the center for editing (19Nomanbhoy T.K. Hendrickson T.L. Schimmel P. Mol. Cell. 1999; 4: 519-528Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar). In recent work, after mutagenesis of the whole chromosome, selections were done to obtain mutant strains of Escherichia coli that could substitute cysteine for valine in cellular proteins at a frequency sufficient to suppress in vivo an inactivating Cys → Val mutation in thymidylate synthase (25Döring V. Mootz H.D. Nangle L.A. Hendrickson T.L. de Crécy-Lagard V. Schimmel P. Marlière P. Science. 2001; 292: 501-504Crossref PubMed Scopus (162) Google Scholar). Five independent mutant strains were obtained. Each of these bore a single point mutation in the coding sequence for ValRS. All of the mutations were located at one of the highly conserved residues in the CP1 editing domain. (One of these mutant enzymes ([T222P]ValRS) was subsequently investigated in more detail, but no further characterizations were done with the other four.) In preliminary studies we found that the mutations were not equivalent, that is, each affected editing to a different degree. Given that the editing domain is conserved through evolution, including the most ancient organisms in the bacterial and archaebacterial kingdoms (26Schimmel P. Ribas de Pouplana L. Cold Spring Harbor Symposium on Quantitative Biology. 66 Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2001: 161-166Google Scholar), we imagined that cell growth would be sensitive to defects in the editing activity. Thus, these mutant enzymes afforded an opportunity to assess the relationship between the degree of the editing defect observed in vitro and the growth phenotype or toxicity seenin vivo. The mutant enzymes also provided further opportunity to evaluate the chemical role that specific conserved residues might have in the editing activity. DISCUSSIONTo further validate the conclusions presented here, some preliminary experiments were done with a different assay in vivo. For this purpose the five chromosome-borne valSmutant alleles were evaluated in the "cysteine suppression test" described previously (25Döring V. Mootz H.D. Nangle L.A. Hendrickson T.L. de Crécy-Lagard V. Schimmel P. Marlière P. Science. 2001; 292: 501-504Crossref PubMed Scopus (162) Google Scholar). The tester strain is deleted of its essential chromosome-borne thyA allele and harbors athyA mutant allele on a plasmid. The mutant thyAallele introduces a valine codon at position 146 of the coding sequence for thymidylate synthase, instead of the natural codon for the catalytic cysteine. These thyA cells are dependent on dT thymidine (dT) for growth. The presence of any of the five point mutations allowed growth of the tester strain in the presence of the cysteine precursor S-carbamoyl cysteine (SCC), presumably due to misincorporation of cysteine for valine at the 146 position. Cells with the wild-type valS allele do not grow without dT even in the presence of SCC. To achieve growth, strains harboring the T222P and K277Q valS alleles needed 0.2 mm SCC. In contrast, cells bearing the three other mutant alleles needed higher concentrations of SCC (0.5 mm) for growth. These results are consistent with data in Figs. 3 A and 4 A, showing that the T222P and K277Q mutant enzymes have the highest levels of mischarging in vitro and in vivo.The results presented here provide strong evidence that the editing domain is critical for enabling the modern genetic code. The correlation between the degree of the defect in editing in vitro (as seen in enzymes bearing specific point mutations in the CP1 domain) and the sensitivity in vivo to Abu is striking. This correlation is consistent with mischarging leading directly to the accumulation of proteins that are toxic due to errors of amino acid incorporation, that is, to ambiguity in the code. Thus, the errors of aminoacylation are not corrected once they have passed the surveillance of the CP1 editing domain.Although the editing reactions of tRNA synthetases have long been known, most prior studies have focused on studying the phenomenonin vitro. The conservation of the CP1 editing domain through all three domains of life implies strong selective pressure in vivo to retain this domain and the present work is consistent with this implication (26Schimmel P. Ribas de Pouplana L. Cold Spring Harbor Symposium on Quantitative Biology. 66 Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2001: 161-166Google Scholar). Under normal conditions, cells may not be exposed to concentrations of noncognate amino acids that are as severe as those used here to unmask the toxicity of a defect in editing. However, the efficiency of mischarging by the various mutants extended over a range of more than 40-fold (compare, for example, T222P and D230N ValRS (Fig. 3 A)). The correlation between efficiency of mischarging and the halo of toxicity, while not linear, was over this entire range (Fig. 4, A and B). This observation shows that even a mutation conferring a relatively small perturbation of editing activity can be toxic. By analogy, a small number of errors occurring under normal growth conditions probably confer a slight growth disadvantage to cells bearing a mutation at a critical residue in CP1. Exerted over many generations the disadvantage would select against the mutation and thus explain the strong conservation of CP1 through evolution.All of the mutations studied here had little effect on the active site for synthesis of Val-tRNAVal (Fig. 2). Possibly the CP1 editing domain originally acted alone and later was inserted into the active site. The feasibility of this possibility was demonstrated by experiments showing that the cloned CP1 editing domain can act as a separate protein in clearing mischarged tRNAs (10Lin L. Hale S.P. Schimmel P. Nature. 1996; 384: 33-34Crossref PubMed Scopus (92) Google Scholar). In sequence databases proteins have been identified that represent pieces of tRNA synthetases. Examples include "fragments" of synthetases for alanine, aspartate, glutamate, glycine, histidine, lysine, methionine, phenylalanine, serine, and tyrosine (35Ribas de Pouplana L. Schimmel P. Trends Biochem. Sci. 2001; 26: 591-596Abstract Full Text Full Text PDF PubMed Scopus (127) Google Scholar). In some instances specific functions for these proteins have been established (36Sissler M. Delorme C. Bond J. Ehrlich S.D. Renault P. Francklyn C. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 8985-8990Crossref PubMed Scopus (116) Google Scholar, 37Dong J. Qiu H. Garcia-Barrio M. Anderson J. Hinnebusch A.G. Mol. Cell. 2000; 6: 269-279Abstract Full Text Full Text PDF PubMed Scopus (315) Google Scholar, 38Qiu H. Dong J. Hu C. Francklyn C.S. Hinnebusch A.G. EMBO J. 2001; 20: 1425-1438Crossref PubMed Scopus (67) Google Scholar). However, none of these examples includes a segment of the CP1 domain. Further expansion of the data base over time may reveal CP1 as a separate protein in certain organisms. (If this occurs, then an important question is whether one or more class I enzymes in those organisms is missing that domain.) However, because the domain is found as a part of class I enzymes in organisms like Aquifex aeolicus that are thought to be near the base of the tree of life (26Schimmel P. Ribas de Pouplana L. Cold Spring Harbor Symposium on Quantitative Biology. 66 Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2001: 161-166Google Scholar), CP1 may have been incorporated into class I enzymes at the time of appearance of the last common ancestor. This early incorporation may have been essential for the subsequent appearance of the full tree of life, which required a highly developed and accurate genetic code that had eliminated ambiguity. To further validate the conclusions presented here, some preliminary experiments were done with a different assay in vivo. For this purpose the five chromosome-borne valSmutant alleles were evaluated in the "cysteine suppression test" described previously (25Döring V. Mootz H.D. Nangle L.A. Hendrickson T.L. de Crécy-Lagard V. Schimmel P. Marlière P. Science. 2001; 292: 501-504Crossref PubMed Scopus (162) Google Scholar). The tester strain is deleted of its essential chromosome-borne thyA allele and harbors athyA mutant allele on a plasmid. The mutant thyAallele introduces a valine codon at position 146 of the coding sequence for thymidylate synthase, instead of the natural codon for the catalytic cysteine. These thyA cells are dependent on dT thymidine (dT) for growth. The presence of any of the five point mutations allowed growth of the tester strain in the presence of the cysteine precursor S-carbamoyl cysteine (SCC), presumably due to misincorporation of cysteine for valine at the 146 position. Cells with the wild-type valS allele do not grow without dT even in the presence of SCC. To achieve growth, strains harboring the T222P and K277Q valS alleles needed 0.2 mm SCC. In contrast, cells bearing the three other mutant alleles needed higher concentrations of SCC (0.5 mm) for growth. These results are consistent with data in Figs. 3 A and 4 A, showing that the T222P and K277Q mutant enzymes have the highest levels of mischarging in vitro and in vivo. The results presented here provide strong evidence that the editing domain is critical for enabling the modern genetic code. The correlation between the degree of the defect in editing in vitro (as seen in enzymes bearing specific point mutations in the CP1 domain) and the sensitivity in vivo to Abu is striking. This correlation is consistent with mischarging leading directly to the accumulation of proteins that are toxic due to errors of amino acid incorporation, that is, to ambiguity in the code. Thus, the errors of aminoacylation are not corrected once they have passed the surveillance of the CP1 editing domain. Although the editing reactions of tRNA synthetases have long been known, most prior studies have focused on studying the phenomenonin vitro. The conservation of the CP1 editing domain through all three domains of life implies strong selective pressure in vivo to retain this domain and the present work is consistent with this implication (26Schimmel P. Ribas de Pouplana L. Cold Spring Harbor Symposium on Quantitative Biology. 66 Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2001: 161-166Google Scholar). Under normal conditions, cells may not be exposed to concentrations of noncognate amino acids that are as severe as those used here to unmask the toxicity of a defect in editing. However, the efficiency of mischarging by the various mutants extended over a range of more than 40-fold (compare, for example, T222P and D230N ValRS (Fig. 3 A)). The correlation between efficiency of mischarging and the halo of toxicity, while not linear, was over this entire range (Fig. 4, A and B). This observation shows that even a mutation conferring a relatively small perturbation of editing activity can be toxic. By analogy, a small number of errors occurring under normal growth conditions probably confer a slight growth disadvantage to cells bearing a mutation at a critical residue in CP1. Exerted over many generations the disadvantage would select against the mutation and thus explain the strong conservation of CP1 through evolution. All of the mutations studied here had little effect on the active site for synthesis of Val-tRNAVal (Fig. 2). Possibly the CP1 editing domain originally acted alone and later was inserted into the active site. The feasibility of this possibility was demonstrated by experiments showing that the cloned CP1 editing domain can act as a separate protein in clearing mischarged tRNAs (10Lin L. Hale S.P. Schimmel P. Nature. 1996; 384: 33-34Crossref PubMed Scopus (92) Google Scholar). In sequence databases proteins have been identified that represent pieces of tRNA synthetases. Examples include "fragments" of synthetases for alanine, aspartate, glutamate, glycine, histidine, lysine, methionine, phenylalanine, serine, and tyrosine (35Ribas de Pouplana L. Schimmel P. Trends Biochem. Sci. 2001; 26: 591-596Abstract Full Text Full Text PDF PubMed Scopus (127) Google Scholar). In some instances specific functions for these proteins have been established (36Sissler M. Delorme C. Bond J. Ehrlich S.D. Renault P. Francklyn C. Proc. Natl. Acad. Sci. U. S. A. 1999; 96: 8985-8990Crossref PubMed Scopus (116) Google Scholar, 37Dong J. Qiu H. Garcia-Barrio M. Anderson J. Hinnebusch A.G. Mol. Cell. 2000; 6: 269-279Abstract Full Text Full Text PDF PubMed Scopus (315) Google Scholar, 38Qiu H. Dong J. Hu C. Francklyn C.S. Hinnebusch A.G. EMBO J. 2001; 20: 1425-1438Crossref PubMed Scopus (67) Google Scholar). However, none of these examples includes a segment of the CP1 domain. Further expansion of the data base over time may reveal CP1 as a separate protein in certain organisms. (If this occurs, then an important question is whether one or more class I enzymes in those organisms is missing that domain.) However, because the domain is found as a part of class I enzymes in organisms like Aquifex aeolicus that are thought to be near the base of the tree of life (26Schimmel P. Ribas de Pouplana L. Cold Spring Harbor Symposium on Quantitative Biology. 66 Ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY2001: 161-166Google Scholar), CP1 may have been incorporated into class I enzymes at the time of appearance of the last common ancestor. This early incorporation may have been essential for the subsequent appearance of the full tree of life, which required a highly developed and accurate genetic code that had eliminated ambiguity. We thank J. Horowitz (Iowa State University, Ames, IA) for the clone of pET16valS and M. Berlyn from theE. coli Genetic Stock Center for the JC8769 strain. We thank P. Marlière for support and T. L. Hendrickson for helpful discussions and advice.