Conserved motifs in prokaryotic and eukaryotic polypeptide release factors: tRNA-protein mimicry hypothesis.

Abstract

Translation termination requires two codonspecific polypeptide release factors in prokaryotes and one omnipotent factor in eukaryotes. Sequences of 17 different polypeptide release factors from prokaryotes and eukaryotes were compared. The prokaryotic release factors share residues split into seven motifs. Conservation of many discrete, perhaps critical, amino acids is observed in eukaryotic release factors, as well as in the C-terminal portion of elongation factor (EF) G. Given that the C-terminal domains of EF-G interacts with ribosomes by mimicry of a tRNA structure, the pattern of conservation of residues in release factors may reflect requirements for a tRNA-mimicry for binding to the A site of the ribosome. This mimicry would explain why release factors recognize stop codons and suggests that all prokaryotic and eukaryotic release factors evolved from the progenitor of EF-G. domain motifs and is involved in omnipotent suppression of nonsense codons (for a review, see ref. 11).Can the current computer programs used for sequence comparison, as designed, predict conserved amino acids at discrete positions in comparisons of multiple random sequences? It seems unlikely to us that the currently used computer programs would recognize single conserved amino acids when the number and diversity of protein sequences is increased, because the algorithms used are essentially based on one-to-one comparison of letters or words of finite length. Here, we approach this problem by identifying first "by computer" the conserved amino acids in prokaryotic RFs, and then asking "by eye" whether these residues are also present in eukaryotic RFs. This approach provided us with clues that lead to universally conserved motifs in RFs, part of which may reflect requirements for molecular mimicry of a tRNA structure.

C-terminal interaction of translational release factors eRF1 and eRF3 of fission yeast: G-domain uncoupled binding and the role of conserved amino acids

“…Truncated eRF1 proteins and their binding to the C-terminal domain of S. pombe eRF3+ The activity of the eRF3C segment (amino acid positions 482-662) for binding to different truncated eRF1 polypeptides was examined by an in vivo two-hybrid system as well as by an in vitro pull-down analysis, as shown in Figure 4, and the data are summarized+ ϩ: binding, 6: weak binding; Ϫ: no binding+ The number refers to the amino acid position from the translation start site+ Sequence motifs homologous to domains III (acceptor stem mimicry), IV (anticodon helix mimicry), and V (T stem mimicry) of elongation factor EF-G are assigned (Ito et al+, 1996(Ito et al+, , 1998a)+ interaction of the C-terminal one-third domain of eRF3 per se with eRF1+ The present study has also confirmed the importance of the C-terminal acidic amino acid stretch of S. pombe eRF1 for binding to eRF3 (Ito et al+, 1998a)+ Following near completion of the revised manuscript after submission, we became aware of a recent study that warrants mention+ Merkulova et al+ (1999) have reported the similar C-terminal domain activity of human eRF3 for binding to eRF1+…”

“…The "RF-tRNA mimicry" hypothesis predicts that, of the three tRNA-mimicry domains III-V, eRF1 regions equivalent to domains III and V of EF-G should mimic the acceptor stem and the T stem of tRNA (Ito et al+, 1996)+ We have previously assigned two eRF3-contact sites on S. pombe eRF1, one to the internal region between amino acid positions 187-247, and the other to the C-terminal region as the primary and strongest binding site for eRF3 (Ito et al+, 1998a)+ The importance of the latter for binding to eRF3 was confirmed in this work+ Given that these two sites correspond to domains III and V, respectively, the C-terminus of eRF1 may mimic the T stem of tRNA+ The C-terminal region of eRF1 conserves a number of amino acids, particularly acidic residues, in eukaryotic RFs, but not in prokaryotic RFs (see Fig+ 3)+ The three-dimensional structure of the ternary complex of Phe-tRNA, EF-Tu, and GDPNP has revealed that the contacts are located in three regions: (1) binding of the CCA-Phe end to domain 2 of EF-Tu and its interface to domain 1; (2) binding of the 59 end and a part of the acceptor stem at the intersection of the three domain interfaces and to the GTPase switch regions; and (3) binding of one side of the T-stem to the surface of domain 3 of EF-Tu (Nissen et al+, 1996)+ These features of aminoacyl-tRNA have been interpreted to define a general aminoacyl-tRNA motif that EF-Tu:GTP recognizes on all ordinary elongator aminoacyl-tRNA molecules+ The importance of the former two regions of EF-Tu have been demonstrated by mutational analyses by several investigators (as described above)+ Of these contacts, the T stem-EF-Tu interaction is thought to be important for recognition of all ordinary elongator aminoacyl-tRNA molecules, which was shown first by a three-dimensional study (Nissen et al+, 1996)+ The detailed resolution of atoms involved in this T stem-EF-Tu interaction has revealed that the negatively charged phosphate backbone of tRNA is essential for binding to EF-Tu (Nissen et al+, 1996;Nakamura & Ito, 1998), suggesting that the stretch of negatively charged amino acids at the C-terminus of eRF1 may mimic the negatively charged phosphate backbone of the T stem of tRNA to bind eRF3 (Ito et al+, 1998a)+ FIGURE 8. In vitro binding of eRF1 and eRF3* proteins carrying contact site mutations+ Experimental procedures and conditions are described in Materials and Methods+ A: Immobilized wild-type or mutant GST-eRF3* were mixed with wildtype or mutant His 6 -eRF1 proteins, and the coprecipitated proteins were analyzed by Western blotting after SDS-PAGE+ The immunoblots of bound His 6 -eRF1 (lower) and GST-eRF3* (upper) proteins were detected by chemiluminescence+ eRF3* proteins: lanes 1-3: wild-type; lanes 4-6: Y577A mutant; lanes 7-9: D647A mutant; lanes 10-12: Y577A-D647A double mutant+ eRF1 proteins: lanes 1, 4, 7, 10, and 13: wild-type; lanes 2,5,8,11,lanes 3,6,9,12,and 15: eRF1-⌬Cb+ Lanes 13-15 are in-put eRF1 controls+ B: Binding efficiency of eRF1 proteins to immobilized eRF3* proteins in the presence of amino acid substitutions+ The intensity of His 6 -eRF1 and GSTeRF3* bands was quantified, and the bindin...…”

“…In eukaryotes, one translational release factor, eRF1, recognizes three stop codons (class-I), and another factor, eRF3, stimulates eRF1 activity and binds guanine nucleotides (class-II)+ The mechanism by which the eRF1 protein reads the stop codon and the G protein, eRF3, controls the mode of termination have been coding and translational problems for the three decades since the discovery of the genetic code (for a review, see Nakamura et al+, 1996)+ Prokaryotes have two class-I release factors, RF1 and RF2, that recognize UAG/UAA and UGA/UAA, respectively+ From a sequence comparison of release factors of different organisms, we have proposed a model in which the class-I release factors mimic the shape of tRNA for binding to the decoding site (A site) of the ribosome and mimic a tRNA anticodon for reading the stop codon ("RF-tRNA mimicry" hypothesis; Ito et al+, 1996)+ The mimicry of tRNA by protein has been identified by means of structural studies of bacterial elongation factors EF-G and EF-Tu complexed with guanine nucleotide(s) and aminoacyl-tRNA+ The three-dimensional structure of Thermus thermophilus EF-G comprises five subdomains; the C-terminal part, domains III-V (AEvarsson et al+, 1994;Czworkowski et al+, 1994), appears to mimic the shapes of the acceptor stem, the anticodon helix, and the T stem of tRNA, respectively (Nissen et al+, 1995)+ Class-I release factors share homology with domain IV of EF-G (Ito et al+, 1996)+ Mutational studies have provided evidence that bacterial RF1 and RF2 encode a putative protein anticodon moiety (Ito et al+, 1998b; for a review, see Nakamura & Ito, 1998)+ The model of RF-tRNA mimicry predicts that a class-II factor, eRF3, may be an EF-Tu-like vehicle protein to bring class-I proteins to the A site of the ribosome+ Several lines of evidence support this view; eRF3 shows considerable C-terminal homology to EF-1a (for a review, see Stansfield & Tuite, 1994), and eRF3 and eRF1 bind in vivo and in vitro and exist as a heterodimer complex in yeast cell lysates (Stansfield et al+, 1995;Zhouravleva et al+, 1995;Ito et al+, 1998a)+ To extend the analysis of eukaryotic release factor function and interaction, we have cloned the eRF1 (identical to Sup45) and eRF3 (identical to Sup35) genes of the fission yeast Schizosaccharomyces pombe (Ito et al+, 1996(Ito et al+, , 1998a)+ These genes are homologous to Saccharomyces cerevisiae counterparts and complement temperature-sensitive mutations in sup45 and sup35 of S. cerevisiae, respectively+ S. pombe eRF3 is a protein with a molecular mass of 72+5 kDa (composed of 662 amino acids), and the deduced protein sequence of the C-terminal 430 amino acids is highly similar to that of S. cerevisiae eRF3 as well as to EF-1a+ However, the 230 N-terminal amino acids do not share any sequence homology with S. cerevisiae eRF3+ The C-terminal two-thirds are essential for viability and translation termination, while the N-terminal one-third is not conserved and is not essential ...…”

How protein reads the stop codon and terminates translation

“…Third, eRF3 and eRF1 bind in vivo and in vitro and exist as a heterodimer in S. cerevisiae cell lysates (Stansfield et al 1995;Zhouravleva et al 1995;Ito et al 1998a), while bacterial RF3 does not bind stably to RF1 or RF2 (Y. Kawazu, K.I. & Y.N., unpublished data; see Ito et al 1996). Fourth, eRF3 shows considerable C-terminal homology to EF-1␣ (for a review, see Stansfield & Tuite 1994), while RF3 has a long C-terminal polypeptide compared with EFTu, which shows significant homology to domain III and part of domain IV of EF-G .…”

Terminating eukaryote translation: Domain 1 of release factor eRF1 functions in stop codon recognition

“…Although a direct role for the ribosome and rRNA in stop codon recognition has been proposed in the past, evidence now points to a direct recognition model for RF stop codon discrimination+ First, prokaryote RF2 can be directly UV crosslinked to the stop codon and downstream nucleotides in vitro, inferring the release factor is in intimate contact with the termination signal (Brown & Tate, 1994;Poole et al+, 1998)+ Second, overexpression of either prokaryote RF1 or eukaryote eRF1 acts to out-compete suppressor tRNA species for stop codon binding+ This so-called antisuppressor phenotype indicates both tRNAs and RFs are cognate species in direct competition for stop codon binding (Weiss et al+, 1984;Stansfield et al+, 1995a;Legoff et al+, 1997)+ Direct recognition models like these imply that RFs might act in a tRNA-like manner to discriminate between codons+ This idea was supported by the discovery that the structure of domain IV/V of elongation factor G complexed with GDP is very similar to that of a tRNA molecule when part of a ternary complex with EF-Tu and GTP, prompting the proposal that protein elongation factors might mimic tRNA molecules (Nissen et al+, 1995)+ On the basis of limited RF sequence similarity to EF-G, the concept of structural tRNA mimicry by the central domain of class I release factors was developed (Ito et al+, 1996)+ The recent solution of the crystal structure of eukaryote eRF1 has allowed a reappraisal of this model for eukaryote RFs, and it is now apparent that, although the central domain of eRF1 at least does not represent a tRNA-like structure (Song et al+, 2000), the Y-shaped eRF1 molecule does have both similar shape and overall dimensions to a tRNA+ The N-terminal domain 1 of eRF1 may represent a potential anticodonlike region, on the basis of its position relative to the peptidyl-release triggering GGQ motif (analogous to the tRNA CCA acceptor stem; Song et al+, 2000)+ Thus the eRF1-tRNA mimicry model is now supported by direct structural evidence, although it seems likely that the bacterial RFs may be structurally dissimilar to eRF1 because their predicted secondary structures are unalike+ It cannot, however, be ruled out that the bacterial RF overall shape may still mimic that of a tRNA+ Recently, the crystal structure of another ribosomal A siteinteracting protein, the bacterial ribosome recycling factor (RRF), has been solved, revealing it, too, has a tRNA-like shape, and strengthening the tRNA mimicry proposal (Selmer et al+, 1999)+ How then is stop codon recognition achieved by a tRNA-analog protein RF? In a recent study, mixed RF1/ RF2 domain hybrid proteins were constructed and screened for RF1 molecules with RF2-like stop codon specificity+ Tripeptide motifs were identified from the central D domain of both release factors that conferred codon specificity, with the first and third amino acids of this peptide discriminating the second and third purine bases of the stop codons (Ito et al+, 2000)+ These findings reinforce the proposal that bacterial RFs directly recognize the stop codon+ Eukaryote eRF1 from Tetrahymena, recently cloned (Karamyshev et al+, 1999), may exhibit natural altered stop codon recogn...…”