The process of translation termination is of critical importance in protein synthesis. Misinterpretation of a stop codon leads to read through and translation beyond the natural end of the coding sequence and into at best non-functional or even harmful additions to the final to the peptide chain. Obviously such illegitimate C-terminal extensions of an amino acid can lead to misfolding and consequently to lack of protein function. Potentially the protein could acquire new and perhaps deleterious activity resembling the way fusion proteins created through chromosomal translocations can acquire novel carcinogenic functions. The mechanism of translation termination presents a long-standing problem of considerable interest. As a result of the identification of the RF-encoding genes, functional studies of mutants involving RF and ribosomal subunits, and the high-level expression of the corresponding gene products for biochemical assays important clues are now emerging that may provide the keys into unlocking the puzzle of the termination of protein synthesis.
The final product of transcription translocates into the cytoplasm from the nucleus into the cytoplasm containing an abundance of ribosomes. In the cytoplasm the process of translation or protein synthesis takes place. It is initiated by methionyl-tRNA in eukaryotes or fmet-tRNA in other organisms placing the first amino acid residue, Met or fMet, opposite to the first coding position of mRNA. This is a first of many amino acid additions which ultimately leads to an elongated, fully functional protein. Triplet signals, similar to triplet signals which code for amino acids, are built into a messenger RNA’s coding sequence which tell the ribosome when to stop translating the mRNA. These are the so-called stop codons UAA, UAG and UGA, and they signal the termination of mRNA translation by facilitating the binding of polypeptide release factor to the ribosome to the ribosome, thereby stimulating hydrolysing of the final peptidyl-transfer RNA linkage and releasing the completed polypeptide from the ribosome.
Termination of protein synthesis or translation occurs on ribosomes as a response to a stop, rather than sense, codon in the decoding site also known as the A site. Translation termination requires two classes of polypeptide release factors (RFs) which interact to form a heterodimer that mediates termination. Class-I factors including codon-specific RFs, RF1 and RF2 in prokaryotes and eRF1 in eukaryotes. Class-II factors are codon non-specific RFs such as RF3 in prokaryotes and eRF3 in eukaryotes (). RF1 is codon specific for the stop codons UAG and UAA. RF2 is specific for the stop codons UGA and UAA. eRF1 is specific for all the stop codons, UAA, UGA, and UAG.
This codon specificity suggested that they must interact directly with the codon on the mRNA. Translation termination involves protein-RNA recognition instead of the well understood codon-anticodon pairing during the mRNA-tRNA interaction (Tate and Brown, 1992). This model was supported by protein-RNA crosslinking data that provide evidence for close contact between the stop codon and RF (Tate and Brown, 1992).
An intimate relationship is involved in translational stopping which requires interactions between the ribosome (rRNA and proteins), mRNA and RF in response to a stop, rather than sense, codon in the decoding site of the ribosome.
For some time now, it has been thought that RFs might interact directly with the codon, until recently evidence has been lacking for such a direct contact. The formation of a double strand structure between the stop codon and the rRNA might cause a confirmational change of ribosome, allowing binding of the ribosome and RF. A zero length crosslinker composed of a thiouracil residue in a stop codon within a designed RNA was efficiently able to form covalent complex with RF, and less efficiently with rRNA (Tate et al. 1990) providing support for a close contact between the stop codon and RF.
The genes encoding prokaryotic RF1 and RF2 have been identified as prfA and prfB respectively (Weiss et al. 1984). Mutations in these genes often cause misreading of stop signals, increased frameshifting, inefficient termination and read through, or a temperature sensitive growth of the cells (reviewed by Ryden & Isaksson 1984 and others).
RF3 stimulates the activities of the class-I factors RF1 and RF2 in prokaryotes whereas eRF3 stimulates eRF1. Class-II factors are known to bind guanosine nucleotides in the form of GTP, and furthermore is not codon specific (Caskey 1969). Furthermore, gene for RF3 was discovered allowing detailed analysis of its action (Mikuni and Grentzmann 1994). The gene encodes the RF3 protein which contains guanine nucleotide binding motifs and has significant sequence homology to the elongation factors EF-G and EF-Tu. Mutations in RF3 cause misreading of all three stop codons, and an excess of RF3 stimulates the formation of ribosomal termination complexes and increases RF1 and RF2 activity (Matsumara 1996). Recent publications indicate that eRF3 also known as mammalian GTP-binding protein GSPT, which has a carboxy-terminal sequence homologous to the eukaryotic elongation factor EF1a . GSPT binds to eRF1.
tRNA mimicry hypothesis
Sequences of RF from different organisms have revealed the conservation of various protein motifs in both eukaryotic and prokaryotic release factors. Some of these conserved domains have sequence homology to domains of EF-G. Elongation factor G (EF-G) is a translocase protein that forwards peptidyl tRNA from the A site to the P site on the ribosome during elongation in translation. The 3-D structure of EF-G in thermus thermophilus comprises 5 subdomains: the C-terminal region and domains III to V appear to mimic the shape of the acceptor stems in the anticodon helix of the T stem of tRNA (AEvarsson 1994).
It has been shown that a RF region shares homology with domain IV of EF-G, therefore constituting a ‘tRNA-mimicry domain’ necessary for RF protein binding to a ribosomal A site . By deducing that RF1 and RF2 are protein mimicing tRNA analogues, RF3 can be imagined as a EF-G protein possessing translocase-like activity or possibly an EF-Tu-like vehicle protein. Furthermore it has been shown that the N-terminal domains of EF-G are similar to RF3 (Ito 1996).
By considering the structural mimicry shown in the figure one may see that the RF1/RF2:RF3 protein complex closely resembles the EF-G:GTP:aminoacyl-tRNA ternary complex (Ito 1996). Therefore peptide chain termination closely resembles the elongation process, except that a stop codon, instead of a sense codon is decoded in the A or decoding site of the ribosome.
There is considerable anticipation for the completion of the crystollographic study of RFs which would further add support for the hypothesis.
Evidence for Protein Anticodon
The RF-tRNA mimicry could explain how release factors have the ability to recognize the stop codon by asuuming an anticodon-mimicry element in the protein. Evidence has accumulated that this in fact may be the case. Mutagenesis of amino acids in the RF2 region equivalent to domain IV of EF-G illustrates several amino acids whose alterations are lethal to cells due to abnormal terminations at the sense codon(s) or the disruption of normal termination. These amino acids seem to be topologically proximal by mimicry to the tip of the anticodon stem.
Data from hydroxy radical probing has provided evidence of EF-Gs location and orientation on the ribosome which gives convincing evidence that EF-G is bound to the A-site on the ribosome as well as the proximity of domain IV of EF-G to the 30S decoding site. This provides topological support for the proposed molecular mimicry model in that RF has a similar domain IV (Wilson 1998).
There is a difference in the shared domain IV however which seems to stem from the fact that RF and tRNAs recognize specific codons whereas EF-G does not (Nissen 1995). A model has emerged in which the dominant toxic position of RF mimic the anticodon loop of tRNA to recognize the three stop codons. Furthermore, stop codon specificity can be altered by swapping the predicted anticodon regions(s) between RF1 and RF2, providing convincing evidence for the anticodon mimicry (K.I. & Y.N. unpublished data). However, the stop codon may be recognized as an alternate genetic code for frameshifting, readthrough or selenocysteine incorporation - referred to as ‘recoding’ for reprogrammed genetic decoding (review by Gesssteland and Atkins 1996). Translational termination can now be thought of either as the general and fundamental event resulting in the release of the protein product, or as a pause or yield for more specialized events beyond the constraints of the genetic code (Nakamura et al. 1990; Tate and Brown 1992).
Ribosomal RNA (rRNAs) play a major role in every aspect of translation, and also have been implicated with the processes of translation termination. A significant finding involved the observation that both bacterial and mammalian ribosomes can carry out the hydrolysis of peptidyl-tRNA without RF in the presence of 30% acetone and tRNA (Caskey 1971). Various antibiotics are inhibitors of peptidyl transferase and inhibit hydrolysis. Other antibiotics inhibit peptidyl-transferase activity but stimulate hydrolysis. These results suggested that the ribosome could alone catalyse the hydrolysis of peptidyl-tRNA and the catalytic site of the ribosomal hydrolytic activity is either the same as the ribosome site that carries out peptidyl transfer or is situated very close to peptidyl transferase (Arkov 1998). Both of the reactions require the same peptidyl-tRNA ester bond needs to be cleaved, either leading to a new peptide bond during peptide transfer or resulting in the release of a free polypeptide from the ribosome during peptide chain termination. A common catalytic site for both elongation and termination can be thus be imagined ().
The difference one can deduce involves the binding of differennt factors at the A or decoding site corresponding to different codons. A sense codon would normally facilitate the binding of elongation factors such as EF-G and EF-Tu promoting the cleavage of the peptide bond and the formation of a new peptide bond during transfer or the binding or RF in response to a stop codon which trigger hydrolysis of the peptidyl-tRNA bond and releasing the peptide.
Both subunits have been implicated in catalysis of peptidyl-tRNA hydrolysis during translation termination (Arkov 1998). Mutations in both subunits, 16S and 23S rRNA caused substantial defects in catalysis of peptidyl-tRNA hydrolysis.
Studies suggest the involvement of 16S rRNA in ribosomes in release factor (RF) binding, which triggers the hydrolysis of peptidyl-tRNA (Caskey 1977). There is indirect evidence that peptidyl transferase may be involved in termination (Vogel 1969) and 23S rRNA has been shown to be the central element of peptidyl transferase (12,42).
The ribosomal mutations analyzed by Arkov et al. caused major defects in the RF2-dependent peptidyl-tRNA hydrolysis. RF1 hydrolysis however remained normal suggesting that the rRNA sites at which the mutations occurred were not involved in hydrolysis of the peptide-tRNA but was involved in the binding of RF2 factors. These mutations possibly could prevent the correct positioning of RF2 causing it to be unable to activate the catalytic site. A further possibility is that the mutations prohibit a conformational change of the RF2 so that is necessary for triggering hydrolysis (Arkov et al., 1998).
From studies on mutations in the rRNA small subunit, sites which are critical for translation termination were identified. C1054A mutations in 16S, located in helix 34 of the ribosome were isolated as suppressors of UGA nonsense mutations, however these mutations were not seen to affect other misreading events (7,8). In eukaryotic 18S mutations corresponding to C1054A had similar effects (13). Another region of 16S rRNA, helix 44, has been implicated in termination. It was shown in E.Coli that cloacin DF13 cleavage of the part of helix that contributes to the A site inhibits formation of a complex between ribosomes, UAA, RF1 and RF2 (21). Furthermore termination was substantially inhibited in vitro (20). C1054 of helix 34, the A-site portion of helix 4, and the three nucleotides of the A-site codon are very closely spatially arranged next to each other in the 30S rRNA subunit. This lead to the speculation that this site may form a structurally-unique surface which is recognized by a small region on release factors (RFs) (Murgola et al.1999). Nucleotides adjacent to the stop codon may modulate the strength of the recognition (15,26,29) and furthermore P-site bound tRNA (28,30).
Mutational studies in the large rRNA subunit have identified sites critical for peptide chain termination. A GTPase center has been discovered and implicated in termination and the process of GTP hydrolysis of EF-G and EF-Tu in elongation(31,32).
Under physiological conditions, either RF1 and RF2 in association with the ribosome activate the hydrolytic site. The mechanism of this signal transduction is unknown. Genetic evidence is present for an interaction between the GTPase center in domain II and the entrance to domain Vof 23S during termination that may be part of the activation process. This interaction may be important for termination since domain V is implicated in peptidyl transferase activity (12). One may speculate that an interaction between the GTPase center domain II and the entrance to domain V is important for the transformation of peptidyl transferase into peptidyl-tRNA hydrolase this may involve cross-linking using release factors as signal transducers.
STOP CODONS
Translation stop signals in the genetic code are defined by 3 of the 64 ‘standard’ codons, UAA, UAG, and UGA, all consisting of three nucleotides each. However, there are strong biases in the upstream and downstream nucleotides surrounding stop codons. Studies have shown that termination is greatly affected by the identity of the +4 nucleotide immediately downstream of the stop codon. These biases may reflect sites of contact between the tRNA-mimicking RFs and the mRNA. This fact alone that protein:mRNA interaction was involved allowed the possibility of the stop signal being greater than just the three nucleotides specified by the code. This suggests that the stop codon encompasses an area greater than the three nucleotide stop codon previously thought and may include the addition of surrounding nucleotides, the terminal amino acids present on the aminoacyl-tRNA, and possibly the tRNA itself. Due to the fact that stop codons are surrounded by other nucleotides in natural mRNAs they are thus subjected to influences by context in both directions, and therefore it cannot be determined if they functinon in vivo as triplets. However, with the availability of the crystal structure of the release factor with its recognition substrate will clearly define the exact length of the recognized stop signal.
TransTerm is a database which was recently created of sequence contexts of start and stop codons has been established which includes over 39 500 coding sequences for over 150 organisms and is available on the World Wide Web (Dalphin 1996). This TransTerm database has proved to be a invalube tool for studying the stop codon context, and larger encompassing studies of prokaryotic and eukaryotic genes have shown that there is a very marked bias in the bases surrounding the stop codon tri-nucleotide. The general conclusions which have been drawn include a bias in the (+4) nucleotide following the stop codon and lesser biases in other positions upstream and downstream of the stop codon, when a comparison is made on the occurance of the nucleotides in the non-coding region adjacent to the termination sites (Tate 1996).
This data and the conclusion drawn illustrate that the length of the termination signal may be more than the codon alone. Poorly performing stop codons have been analyzed and they have been implicated in alternative genetic events such as readthrough, selenocyteine incorporation, and translational hoping which use these weak signals for specific genetic regulation in the cell. Stonger stop signals are involved in higly expressed genes. Highly expressed genes selected from the TransTerm database have extreme biases where translational pressures for the rapid decoding of signals in all the three phases of protein synthesis. Highly expressed E.Coli genes ~10% of all genes UAA is the predominant codon and 70-80% use UAAU or UAAG as the stop signal. In the case of mammals, selection of a subgroup of highly expressed genes such as immunoglobulins, albumins etc. revealed that over 90% of such genes have A or Gin the +4 position of the stop signal (McCaughan 1995). Analyses such as these illustrated that there was a scale of translation termination signals composed of different strengths, and that the +4 nucleotide possibly was a key determinant in the rapidity of signal decoding. The 3’ or +4 nucleotide termination strength (competitiveness against a frameshift event) is in the order U>G>A>C, as seen in the high incidence of U and G in highly expressed genes where rapid and efficient termination is required.
UAAU is the most common stop signal in highly expressed E.Coli genes and it is the strongest of the 12 four-base signals. Purines (G and A), however are favoured in the +4 position in highly expressed mammalian genes and these give the strongest signals. These differences suggest different interactions in binding and signal transduction between prokaryotic RF1 and RF2 which are specific for UAA and eukaryotic eRF which is also specific for UAA.
The statistical bias seem in the upstream nucleotides may be present for reasons other than those of the downstream regions but may be still relevant in peptide chain termination (Arkov 1995). Upstream biases may reflect the preference for amino acids of a particular character at the C-terminal end of the nascent peptide - which specifies the last two sense codons and amino acids -to ensure that efficient termination occurs (Bjornsson 1996). This has suggested that the release factor has the ability to spatially position itself close to the aminoacyl-tRNA interface in the P site of the ribosome. Although evidence suggests that ribosomal RNA has the peptidyl-tRNA hydrolase function, there might be an interaction between the release factor and the last two peptidyl-tRNA (Varshavsky 1992). Basic amino acids were found to ensure efficient termination at UGA versus acidic residues which are inefficient (Bjornsson 1996). Furthermore, biases in the last amino acid were determined from various studies in which a non-randomness occurence in the last (-1) sense codon position was found (Buckingham 1990). Many amino acids were found to be favourable, although in this position only polar (charged) amino acids are over-represented (Bjornsson 1996).
