We report the relative stabilities, in the form of complex lifetimes, of complexes between two tRNAs complementary, or nearly so, in their anticodons. The results show striking parallels with the genetic coding rules, including the wobble interaction and the role of modified nucleotides sAU and V (a 5-oxyacetic acid derivative of U). One important difference between the genetic code and the pairing rules in the tRNAtRNA interaction is the stability in the latter of the short wobble pairs, which the wobble hypothesis excludes. We stress the potential of U for translational errors, and suggest a simple stereochemical basis for ribosome-mediated discrimination against short wobble pairs. Surprisingly, the stability of anticodon-anticodon complexes does not vary systematically with GC composition, unlike all other known double helices. How (10)(11)(12)(13)(14). Therefore, we selected this model system for more intensive study of base pairing rules; the observed strong parallels with the genetic coding rules justify our choice a posteriori."Recognition" between two tRNA anticodon loops can be expressed quantitatively by the association equilibrium constant, or by the kinetic constants for the association-dissociation reaction. Because the association rate constants are found to vary only moderately (by about a factor of five; unpublished results), the association equilibrium constant is approximately inversely proportional to the dissociation rate constant. Hence, we have selected a single parameter, the lifetime of the anticodonanticodon complex, determined by the temperature-jump technique, as a measure of the strength of association and therefore of the extent of recognition.In general we find that a "correct" base triplet interaction (as determined by the genetic coding rules) has a long lifetime. However, the lifetime of correct complexes frequently is not more than two orders of magnitude longer than the lifetime of some of the incorrect complexes that have mismatching bases. Therefore, to avoid errors, the ribosome must amplify the small difference in lifetime to produce a greater distinction between correct and incorrect pairings in protein synthesis. Specific mechanisms for such an amplification have been discussed under the heading of kinetic proofreading (15)(16)(17). MATERIALS AND METHODSThe procedures used in this work were described in a preceding paper (14). The concentration of each tRNA was 1.3-1.8 ,uM, based on amino acid acceptance activity. Temperature jumps of up to 100 were used, and relaxations were monitored at 266 nm. Control experiments on single samples of tRNA (not mixed with their complement) showed no slow relaxation signals between 00 and 150, except as reported explicitly below.The lifetimes (T*) of anticodon-anticodon complexes at 90 were determined from the relaxation time r, using the equation

Numerous misacylations occur on heterologous systems containing unfractionated tRNAs from yeast or from Bacillus stearothermophilus and pure valyl-tRNA synthetase from B. stearothermophilus or from yeast, when special aminoacylation conditions are used. I n the homologous system and in heterologous systems where the unfractionated tRNAs and the enzyme originate from prokaryotic organisms (B. stearothermophilus and Escherichia coli), the errors are seldom. This phenomenon is explained by competition effects between the cognate tRNAVa1 and noncognate tRNAs for the valyl-tRNA synthetase. But using pure tRNA species, errors can be observed in such systems, even under classical assay conditions; in particular it was shown that the valyl-tRNA synthetase from B. stearothermophilus catalyses the misacylation of E. coli tRNAIle and tRNA,Met and of yeast tRNA,Met and tRNAPhe. These reactions are characterized by K , values slightly increased as compared to the value obtained in the cognate system and by V values decreased by a factor of about 40 to 3000 compared to the cognate tRNA species. I n the presence of dimethylsulfoxide, the rate and the extent of those misacylation reactions are enhanced. I n the case of E . coli tRNAPhe, the misacylation occurs only in the presence of the organic solvent. I n no case however, new aminoacylation errors are induced a t high temperature (50-75 "C) in the presence of the thermostable valyl-tRNA synthetase from B. stearothermophilus ; only an increase of the rates of the aminoacylation reactions which already occur a t 30 "C has been observed a t higher temperature. Thus organic solvent and heat must have distinct effects on the essential parameters determining the specificity of the tRNA aminoacylation reactions.Also it has been observed that the most easily misacylated tRNA species by valyl-tRNA synthetase from B. stearothermophilus are the same as those which are misacylated by the valyltRNA synthetases from E . coli and from yeast. This observation suggests the existence of a family of tRNAs containing besides of tRNAVB', other tRNAs such as tRNATle, tRNAyet and tRNAPhe, and which are likely to be related from a phylogenic point of view. Moreover these tRNA species have also been found to be easily misacylated by other aminoacyl-tRNA synthetases namely the enzymes specific for isoleucine and phenylalanine, thus suggesting more generally the existence of interacting tRNA-aminoacyl-tRNA synthetase families.The possibility of the tRNAs to be enzymatically mischarged with a wrong amino acid seems to be a rather general feature as numerous incorrect aminoacylation systems have now been described [i -31 (and references therein). These mischarging reactions have especially been detected in vitro, under various experimental conditions, either in heterologous systems where the enzyme and the tRNA came from different sources, but also in homologous systems, where one pure tRNA species interacts Enzyme. Valyl-tRNA synthetase (EC 6.1.1.9). Eur. J. Biochem. 45 (1974)with one non-cognate pure ...

We have investigated the specificity of the tRNA modifying enzyme that transforms the adenosine at position 34 (wobble position) into inosine in the anticodon of several tRNAs. For this purpose, we have constructed sixteen recombinants of yeast tRNAAsp harboring an AXY anticodon (where X or Y was one of the four nucleotides A, G, C or U). This was done by enzymatic manipulations in vitro of the yeast tRNAAsp, involving specific hydrolysis with S1-nuclease and RNAase A, phosphorylation with T4-polynucleotide kinase and ligation with T4-RNA ligase: it allowed us to replace the normal anticodon GUC by trinucleotides AXY and to introduce simultaneously a 32P-labelled phosphate group between the uridine at position 33 and the newly inserted adenosine at position 34. Each of these 32P-labelled AXY "anticodon-substituted" yeast tRNAAsp were microinjected into the cytoplasm of Xenopus laevis oocytes and assayed for their capacity to act as substrates for the A34 to I34 transforming enzyme. Our results indicate that: 1/ A34 in yeast tRNAAsp harboring the arginine anticodon ACG or an AXY anticodon with a purine at position 35 but with A, G or C but not U at position 36 were efficiently modified into I34; 2/ all yeast tRNAAsp harboring an AXY anticodon with a pyrimidine at position 35 (except ACG) or uridine at position 36 were not modified at all. This demonstrates a strong dependence on the anticodon sequence for the A34 to I34 transformation in yeast tRNAAsp by the putative cytoplasmic adenosine deaminase of Xenopus laevis oocytes.

We have investigated the specificity of the enzyme tRNA (wobble guanosine 2′‐O‐)methyltransferase which catalyses the maturation of guanosine‐34 of eukaryotic tRNAPhe to the 2′‐O‐methyl derivative Gm‐34. This study was done by micro‐injection into Xenopus laevis oocytes of restructured yeast tRNAPhe in which the anticodon GmAA and the 3′ adjacent nucleotide ‘Y’ were substituted by various tetranucleotides. The results indicate that the enzyme is cytoplasmic; the chemical nature of the bases of the anticodon and its 3′ adjacent nucleotide is not critical for the methylation of G‐34; the size of the anticodon loop is however important; structural features beyond the anticodon loop are involved in the specific recognition of the tRNA by the enzyme since Escherichia coli tRNAPhe and four chimeric yeast tRNAs carrying the GAA anticodon are not substrates; unexpectedly, the 2′‐O‐methylation is not restricted to G‐34 since C‐34, U‐34 and A‐34 in restructured yeast tRNAPhe also became methylated. It seems probable that the tRNA (wobble guanosine 2′‐O‐)methyltransferase is not specific for the type of nucleotide‐34 in eukaryotic tRNAPhe; however the existence in the oocyte of several methylation enzymes specific for each nucleotide‐34 has not yet been ruled out.

A combination of several enzymes, RNase-T1, nuclease S1, T4-polynucleotide kinase and T4-RNA ligase were used to prepare and modify different fragments of yeast tRNAAsp (normal anticodon G U C). This allowed us to reconstitute, in vitro, a chimeric tRNA that has any of the four bases G, A, U or C, as the first anticodon nucleotide, labelled with (32p) in its 3' position. Such reconstituted (32p) labelled yeast tRNAAsp were microinjected into the cytoplasm or the nucleus of the frog oocyte and checked for their stability as well as for their potential to work as a substrate for the maturation (modifying) enzymes under in vivo conditions. Our results indicate that the chimeric yeast tRNAsAsp were quite stable inside the frog oocyte. Also, the G34 was effectively transformed inside the cytoplasm of frog oocyte into Q34 and mannosyl-Q34; U34 into mcm5s2U and mcm5U. In contrast, C34 and A34 were not transformed at all neither in the cytoplasm nor in the nucleus of the frog oocyte. The above procedure constitutes a new approach in order to detect the presence of a given modifying enzyme inside the frog oocyte; also it provides informations about its cellular location and possibility about its specificity of interaction with foreign tRNA.