Nucleic Acids Research, 2001, Vol. 29, No. 23 4767-4782
© 2001 Oxford University Press
Misreading of termination codons in eukaryotes by natural nonsense suppressor tRNAs
Institut für Biochemie, Bayerische Julius-Maximilians-Universität, Biozentrum, Am Hubland, D-97074 Würzburg, Germany
Received July 20, 2001; Revised and Accepted October 3, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONCONCLUSIONREFERENCES |
|---|
Translational stop codon readthrough provides a regulatory mechanism of gene expression that is extensively utilised by positive-sense ssRNA viruses. The misreading of termination codons is achieved by a variety of naturally occurring suppressor tRNAs whose structure and function is the subject of this survey. All of the nonsense suppressors characterised to date (with the exception of selenocysteine tRNA) are normal cellular tRNAs that are primarily needed for reading their cognate sense codons. As a consequence, recognition of stop codons by natural suppressor tRNAs necessitates unconventional base pairings in anticodon–codon interactions. A number of intrinsic features of the suppressor tRNA contributes to the ability to read non-cognate codons. Apart from anticodon–codon affinity, the extent of base modifications within or 3' of the anticodon may up- or down-regulate the efficiency of suppression. In order to out-compete the polypeptide chain release factor an absolute prerequisite for the action of natural suppressor tRNAs is a suitable nucleotide context, preferentially at the 3' side of the suppressed stop codon. Three major types of viral readthrough sites, based on similar sequences neighbouring the leaky stop codon, can be defined. It is discussed that not only RNA viruses, but also the eukaryotic host organism might gain some profit from cellular suppressor tRNAs.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONCONCLUSIONREFERENCES |
|---|
Regulation of gene expression is found at different stages, one of which operates at the level of termination in protein synthesis. Normally, termination of mRNA translation is signalled by the occurrence of any of the three nonsense (stop) codons UAG, UAA or UGA at the ribosomal A site. Protein release factors specifically bind to these codons (for which cognate tRNAs are not available) and subsequently mediate release of the nascent polypeptide chain from the ribosome. However, there are several processes known that can circumvent nonsense codons, for example ribosomal frameshifting and suppression by either mutated or natural cellular tRNAs. In the case of frameshifting, the reading frame is shifted in the 5' or 3' direction and as a consequence the suppressed stop codon is read as a sense codon (1–3). Mutated suppressor tRNAs carry an altered anticodon allowing the tRNA to misread a stop codon by normal base pair interactions. A bulk of chemically induced nonsense suppressors have been isolated in the past from Escherichia coli and Saccharomyces cerevisiae and employed in genetic and biochemical studies, but will not be the subject of this article.
Natural nonsense suppression, however, means the reading of stop codons as sense codons by normal cellular tRNAs which are called natural suppressors. This review describes primarily the structure and function of such tRNAs—with the exception of selenocysteine tRNA—in higher eukaryotes. It does not cover missense (i.e. the reading of sense codons by non-cognate tRNAs) and nonsense suppression in bacteria for which excellent reviews exist (4–6). Furthermore, this article deals with codon context effects, i.e. primary sequences and secondary structures in the vicinity of ‘leaky’ stop codons that influence the efficiency of suppression. A number of recent reports describing related topics are recommended to the reader (7–13).
IDENTIFICATION OF STOP CODON READTHROUGH
Translational readthrough provides a regulatory mechanism of gene expression by permitting the differential production of more than one polypeptide from a single gene. Notably, RNA viruses make use of this potential to expand the genetic information of their relatively small genomes (Fig. 1).
|
The first type of stop codon readthrough was detected in E.coli infected with RNA phage Qß. It was shown that normal tRNATrp with CCA anticodon stimulates readthrough over the UGA stop codon at the end of the coat protein cistron, resulting in an elongated coat protein that is essential for the formation of infective Qß particles (14–16). It took more than a decade to detect natural suppressor tRNAs also in eukaryotic cells. Pelham (17) postulated the presence and ‘leakiness’ of a UAG codon at the end of the 126 kDa cistron in tobacco mosaic virus (TMV) RNA (Fig. 2), since he found stimulation of the in vitro synthesis of the 183 kDa readthrough protein in the presence of yeast amber suppressor tRNA. The elucidation of the complete TMV RNA sequence confirmed the existence of an internal UAG stop codon (18). Moreover, the observation that the 183 kDa polypeptide was synthesised in TMV-infected tobacco protoplasts and tobacco leaves (19,20) supported the notion that the ‘leaky’ UAG codon is very likely suppressed by a cellular tobacco tRNA. The 126 kDa protein contains a putative methyltransferase and a helicase domain, whereas the 183 kDa readthrough product harbours the highly conserved GDD motif, responsible for replicase activity (10). Both polypeptides are essential for TMV multiplication (21).
|
The second well-known example for translational readthrough is found upon expression of the murine leukemia virus (MuLV) RNA in animal tissue. In MuLV-infected mouse cells the in-frame UAG stop codon at the end of the gag cistron is suppressed to yield a gag–pol fusion polypeptide of 180 kDa, which is subsequently cleaved into the gag protein, a protease of 13 kDa and the reverse transcriptase (Fig. 3). Thus, the readthrough product is the only source of reverse transcriptase, and consequently UAG suppression is necessary for normal multiplication of MuLV. Yoshinaka et al. (22) showed that the protease that overlaps the readthrough region contains glutamine inserted at the site of the UAG codon, indicating that a cellular tRNAGln acts as a UAG suppressor. Many (but not all) members of the alphavirus genus contain a ‘leaky’ UGA stop codon within their genomic RNA separating the non-structural proteins nsP3 and nsP4 (Fig. 4). The nsP4 protein shares homologous amino acid sequences with the RNA-dependent RNA replicase of poliovirus and plant RNA viruses, and is supposed to contain the elongating activity of the sindbis virus replicase (23).
|
|
Besides TMV, a bulk of plant RNA viruses from at least 14 groups utilise stop codon readthrough to generate either a functional polymerase or an extended coat protein. The latter is important for the assembly of the virion and/or vector transmission. Of all known ‘leaky’ stop codons, UAA is by far the least frequently used. Some of the multicomponent viruses, such as beet soil-borne virus and potato mop-top virus contain diverse ‘leaky’ stop codons on different RNA species (Fig. 1).
To date, very few DNA viruses are known that employ nonsense suppression as a means of regulating gene expression. For instance, the 58 kDa virion protein of the human cytomegalovirus may be synthesised via translational readthrough of a UGA termination codon separating two open reading frames on the 1.6 kb late mRNA (24).
As mentioned above, the readthrough polypeptides have been shown in many cases to be essential for virus production. Furthermore, the ratio of the two synthesised products appears to be critical. Thus, the substitution of the ‘leaky’ UAG by a CAG glutamine codon in MuLV-RNA resulted in the inability to produce virus (25,26). An unbalanced synthesis of the 126 and 183 kDa polypeptides in TMV-infected tobacco plants also caused a low yield of progeny (21).
UAG/UAA SUPPRESSORS
Cytoplasmic tRNATyr
The first eukaryotic natural UAG suppressor tRNA was isolated from tobacco leaves and adult Drosophila on the basis of its ability to promote readthrough over the ‘leaky’ UAG codon of TMV RNA in a messenger-dependent reticulocyte lysate and in Xenopus oocytes, respectively (19,27). The highly purified suppressor tRNA was sequenced by the fragment analysis of partially digested tRNA according to Stanley and Vassilenko (28). Its amino acid acceptance and anticodon sequence identified the UAG suppressor as cytoplasmic tRNATyr with a G
A anticodon (Fig. 5). Tobacco and wheat leaves contain only tRNATyr isoacceptors with a GA anticodon, whereas 
85% of the tRNATyr species from wheat germ and lupin seeds have a QA anticodon (29–31). Interestingly, tRNATyr with a QA anticodon is totally unable to stimulate UAG readthrough (27,29,32). The highly modified Q nucleoside is synthesised as the free queuosine base (Q), which is then inserted into tRNA by a transglycosylase to replace guanine (33). Bacterial tRNAs are normally completely modified with respect to Q, whereas yeast is totally lacking Q-containing tRNAs (34). Yeast tRNATyr with a GA anticodon is likewise able to suppress the TMV-specific UAG codon in vitro (H.Beier, unpublished result) although its nucleotide sequence differs considerably from plant and animal tRNAsTyr (Fig. 5).
|
Animal tRNAs exhibit a variable Q content depending on the developmental state. In differentiated liver only tRNATyr with Q
A anticodon is present (35), but fetal liver, reticulocytes and tumour tissue contain significant amounts of tRNAs with G in place of Q in the first position of the anticodon (36,37). Adult Drosophila possess tRNATyr isoacceptors with GA and a QA anticodon (27), and relevant changes in the Q content in tRNAs isolated from different ontogenetic stages of Drosophila have been observed (38).
Another interesting attribute of tRNATyr is the pseudouridine (35) at the second position of the anticodon, a unique feature of all eukaryotic cytoplasmic tRNAsTyr. An in vitro synthesised tRNATyr with GUA instead of GA anticodon was unable to suppress the ‘leaky’ UAG codon of TMV RNA (39), indicating that the 35 modification in the tRNATyr anticodon is necessary for the unconventional codon reading. Pseudouridine can form a classical base pair with adenosine, but is more versatile in its hydrogen bonding interactions, possibly resulting in stabilisation of codon–anticodon interactions (40). Hence, there is a remarkable influence of base modifications on UAG suppression. The presence of the hypermodified Q at the first anticodon position of tRNATyr prevents, whereas at the second anticodon position enhances, the unconventional base pairing (Fig. 5).
Plant cytoplasmic tRNATyr with a GA anticodon is quite an efficient UAG suppressor. Under optimal conditions 10–30% of readthrough over the TMV-specific ‘leaky’ UAG are observed in vitro (19,39,41). The same tRNATyr is able to read UAA, but not the UGA stop codon, albeit with lower efficiency (39). Although tRNATyr(GA) is a very potent UAG suppressor in vitro, and very likely also in vivo, it cannot be regarded as a universal suppressor because of its limited presence in various tissues as outlined above.
Cytoplasmic tRNAsGln
The second class of eukaryotic UAG/UAA suppressors are cytoplasmic tRNAsGln. Two isoacceptors with a CUG or U*UG anticodon exist virtually in all pro- and eukaryotes (34). A number of observations had implied their putative ability to read UAG and/or UAA codons. For instance, in the yeast S.cerevisiae, transformation with high copy numbers of a tRNAGln isoacceptor with a CUG anticodon resulted in the suppression of a number of UAG mutations in the yeast genome (42–44). Likewise, Pure et al. (45) reported that overexpression of yeast tRNAGln with a UUG anticodon weakly suppressed internal UAA codons. Furthermore, glutamine is inserted at the site of the ‘leaky’ UAG codon in MuLV RNA (22).
Two cytoplasmic tRNAGln isoacceptors were isolated from mouse liver and tobacco leaves, and their sequences were determined (Fig. 6). The minor tRNAGln species from mouse liver with a UmUG anticodon was shown to stimulate readthrough over the ‘leaky’ UAG codon in TMV RNA in a reticulocyte lysate (46). Noticeably, the amount of this suppressor tRNAGln was greatly increased in NIH3T3 and Ehrlich ascites cells infected with Mo-MuLV, as reported by Kuchino et al. (46). In contrast, Feng et al. (47) observed equivalent amounts of the two tRNAGln isoacceptors in MuLV-infected and uninfected NIH3T3 cells. The reasons for this discrepancy are unknown.
|
Both cytoplasmic tRNAGln isoacceptors from tobacco stimulated readthrough over the TMV-specific UAG codon in a wheat germ extract partially depleted of endogenous tRNAs. In this system, Nicotiana tRNAGln with a UmUG anticodon was a less efficient UAG suppressor than tRNAGln with a CUG anticodon (48). One of the major tRNAGln isoacceptors from the unicellular ciliate Tetrahymena thermophila has a UmUG anticodon and comprises a high sequence similarity to its animal and plant counterparts (Fig. 6). This tRNAGln isoacceptor is a very potent UAG and UAA suppressor in wheat germ extract, provided appropriate amounts of a Tetrahymena synthetase preparation are added to the extract (41). Most of the sequenced cellular tRNAGln species carry an unmodified A residue immediately 3' to the anticodon at position 37 (Fig. 6). Interaction of the two tRNAGln isoacceptors with UAG and/or UAA requires an unconventional G:U base pairing at the third anticodon position (see below). Presumably, an unmodified A adjacent to the anticodon facilitates non-canonical base pairing at the third anticodon position, since a number of reports have conversely shown that a hypermodified A, like i6A (N6-isopentenyladenosine) or ms2i6A (2-methylthio-N6-isopentenyladenosine), impedes non-Watson–Crick interactions at this position (49–51).
Cytoplasmic tRNAsLeu
Two amber suppressor tRNAs have been isolated from calf liver that read the ‘leaky’ UAG codon of TMV RNA and RNA-2 of beet necrotic yellow vein virus (BNYVV) in a messenger-dependent reticulocyte lysate (52). They are neither tRNATyr nor tRNAGln species, but tRNALeu isoacceptors of cytoplasmic origin with CAA and CAG anticodon (Fig. 7). The recognition of the UAG codon by these suppressors requires an unusual A:A pairing in the second, and also a G:U pairing in the third, position of the anticodon. It is not known whether the corresponding plant tRNALeu isoacceptors also comprise UAG suppressor activity in vitro.
|
UGA SUPPRESSORS
Chloroplast and cytoplasmic tRNAsTrp
Readthrough over the ‘leaky’ UAG in TMV RNA was employed for the characterisation of virtually all eukaryotic UAG suppressors described above. In order to identify UGA suppressors, a second plant RNA virus was utilised in most cases. The tobravirus, tobacco rattle virus (TRV), has a bipartite genome. RNA-1 molecules contain the genetic information needed for RNA replication and intercellular movement, whereas RNA-2 specifies the coat protein. In vitro and in vivo translation studies have demonstrated that RNA-1 directs the synthesis of two polypeptides of 134 and 194 kDa (53). Expression of the 194 kDa replicase protein requires readthrough over a ‘leaky’ UGA stop codon at the end of the 134 kDa cistron (Fig. 8)
|
Two natural UGA suppressors were isolated from uninfected tobacco plants on the basis of their ability to promote readthrough over the ‘leaky’ UGA codon in TRV RNA-1 in a wheat germ extract (54). Their amino acid acceptance and nucleotide sequences identified the two UGA suppressors as chloroplast and cytoplasmic tryptophan-specific tRNAs with the anticodon CmCA (Fig. 9). Unexpectedly, it was found that chloroplast tRNATrp suppresses the UGA codon more efficiently than cytoplasmic tRNATrp. It has been speculated that structural features and/or base modifications of the former contribute to higher suppression. Thus, a mutant strain from E.coli with increased UGA suppressor activity turned out to contain a tRNATrp species with an unaltered anticodon, but a single nucleotide exchange at position 24 leading to an A24:U11 instead of a G24:U11 base pair in the D-stem (14; Fig. 9). Presumably, the mutation at position 24 reduces the rate at which the ribosome rejects non-cognate tRNAs, more easily permitting unconventional base pairing (55,56). The chloroplast tRNATrp from tobacco contains the same A24:U11 base pair as the mutated E.coli tRNATrp, whereas cytoplasmic tRNATrp has a G:C pair at this postion (Fig. 9). It should be noted, however, that it is not known whether eukaryotic ribosomes express similar constraints. Another difference between the two tRNATrp isoacceptors lies in the nature of the modified nucleoside at position 37. Chloroplast tRNATrp carries i6A or ms2i6A, whereas cytoplasmic tRNATrp has a m1G (1-methylguanosine) at this position. It has been proposed that i6A and its derivatives stabilise anticodon–codon interactions by increasing the stacking effect of the anticodon on neighbouring nucleotides, thus supporting non-canonical base interactions at the first anticodon position (57,58).
|
A) at position 24, and exhibits increased UGA suppressor activity as compared to the wild type (Given the high UGA suppressor activity of chloroplast tRNATrp, the question arises whether fidelity of chloroplast synthesis is impaired. However, close inspection of the 24 protein-coding genes in the Nicotiana chloroplast genome that terminate with a UGA codon, reveals that in
50% of the cases the stop codon is flanked at the 3' side by a second one within 0–3 codons (59) and in other cases it is embedded in an unfavourable codon context (see below), suggesting that deleterious effects are negligible. Several reports indicate that tRNATrp isoacceptors with UGA suppressor activity are also present in higher vertebrates. A tRNATrp was purified from rabbit reticulocytes and was shown to efficiently suppress the UGA codon at the end of the ß-globin gene in vitro (60). However, the sequence of this tRNA was not determined and, thus, the possibility cannot be ruled out that the suppressor was in fact mitochondrial tRNATrp with a U*CA anticodon, the only tRNATrp species encoded by the rabbit mitochondrial genome (34), and not cytoplasmic tRNATrp with a CCA anticodon. Furthermore, a tRNATrp with a CmCA anticodon was isolated from avian sarcoma virus virions whose sequence and nucleoside modifications were identical to cytoplasmic tRNATrp of uninfected cells, yet the former was a more active UGA suppressor in vitro as compared to its cellular counterpart (61). The reason for this unusual observation is yet unknown. The authors speculate that two conformations of avian tRNATrp may exist that influence the biological activity of the tRNA.
Chloroplast and cytoplasmic tRNAsCys
A number of observations made the existence of cysteine-accepting UGA suppressors very likely. Feng et al. (62) demonstrated that replacement of the ‘leaky’ UAG by a UGA stop codon in MuLV RNA (Fig. 3) directs the incorporation of cysteine in addition to tryptophan and arginine at the corresponding position of the readthrough product upon translation in a reticulocyte lysate.
Chloroplast and cytoplasmic tRNACys isoacceptors with a GCA anticodon (Fig. 10) were isolated from uninfected tobacco leaves on the basis of aminoacylation assays with [35S]cysteine in the presence of an unfractionated wheat germ aminoacyl-tRNA synthetase preparation. Subsequently, the readthrough activity of the two isoacceptors was studied in wheat germ extract programmed with in vitro synthesised transcripts containing the ‘leaky’ UGA region of TRV RNA-1 (Fig. 8). Chloroplast tRNACys turned out to suppress the UGA in the TRV as well as in the TMV codon context more efficiently than the cytoplasmic counterpart (63), an observation already made for chloroplast tRNATrp (see above). No data are available about putative UGA suppressor activity of cytoplasmic tRNACys from animal cells. However, as deduced from the overall sequence homology of tRNACys gene sequences in the mouse and human genome with the corresponding plant genes (64) it is reasonable to assume that animal tRNAsCys also have the potential to act as UGA suppressors.
|
An interesting situation applies for the unicellular ciliate Euplotes octocarinatus. In this organism deviations from the universal genetic code have been described. The UGA stop codon is translated as cysteine within coding regions and does not function as a termination codon in protein synthesis (65). Presumably, only a single tRNACys isoacceptor exists in Euplotes cells which contains a GCA anticodon like all known tRNAsCys species (64; Fig. 10). Plant tRNAsCys have turned out to be relatively inefficient UGA suppressors in vitro (63). One major argument in favour of a more active UGA-decoding Euplotes tRNACys(GCA) is the existence of a release factor in this organism that has lost the capability to bind to UGA codons (66,67), so that competition between the latter and tRNACys is greatly reduced.
Cytoplasmic tRNAsArg
The third class of UGA suppressors has been exclusively characterised in plants on the tRNA level. The six arginine codons CGN and AGR are read by four isoacceptors in E.coli and S.cerevisiae, and by five isoacceptors in the cytosol of higher eukaryotes. Of these five isoacceptors, tRNAArg with a U*CG anticodon is not found in E.coli and yeast (68,69). It is noteworthy that it is just this isoacceptor that is a potent UGA suppressor. The major tRNAArg species have been isolated from wheat germ. In vitro translation of transcripts containing the UGA codon in the context of TRV RNA-1 (Fig. 8) in wheat germ extract in the presence of either of the wheat tRNAArg isoacceptors revealed that mainly tRNAArg with U*CG, and to a lesser extent tRNAArg with ICG anticodon (Fig. 11), stimulated UGA readthrough. Moreover, tRNAArg with a U*CG anticodon was found to also suppress the ‘leaky’ UGA codon in the pea enation mosaic and sindbis virus context (70). Studies of Feng et al. (62) and Chittum et al. (71) indicate that tRNAsArg are also potential UGA suppressors in animals. They identified the amino acid arginine (among others) inserted at the site of UGA in the MuLV-specific context and at the UGA terminating the ß-globin cistron upon translation in rabbit reticulocytes in vitro and in vivo
|
Sindbis virus RNA contains an in-frame UGA termination codon in the region separating the non-structural proteins nsP3 and nsP4 (Fig. 4). Readthrough over this UGA codon has been observed in cultured cells of chicken, human and insect origin (23). Remarkably, in the closely related semliki forest virus (SFV) there is no UGA, but instead a CGA arginine codon at this position (72), suggesting that possibly the ‘leaky’ UGA in sindbis virus RNA is recognised preferentially by tRNAArg(U*CG), which routinely reads the CGA arginine codon. Li and Rice (73) have presented indirect evidence that tryptophan (and not arginine) is incorporated at the site of the ‘leaky’ UGA in RNA transcripts containing the sindbis virus-specific readthrough region upon in vitro translation in a rabbit reticulocyte lysate. In this connection it should be emphasised that reticulocytes represent a highly specialised type of cells in which the nuclei, and to some extent also the mitochondria, are disintegrated, resulting in the accumulation of mitochondrial tRNATrp with U*CA anticodon in the cytosol. This tRNA isoacceptor utilises normal base interactions to read UGA and consequently may easily out-compete any other natural suppressor.
UNCONVENTIONAL BASE INTERACTONS
Crick (74) had postulated that G:U, U:G and I:C/U/A base pairs at the first anticodon position (also called the ‘wobble’ position) would not affect the fidelity of protein synthesis, due to the degeneracy of the genetic code. Later, some restrictions from this rule were observed in different organisms. For instance, in the yeast Schizosaccharomyces pombe, nonsense suppression has been shown to be strictly codon specific. The ability of ochre suppressors with a U*UA anticodon (where U* is mcm5U or mcm5s2U) to read only UAA and not the UAG stop codon was attributed to the modification present at the first anticodon position, which obviously reduces the interaction with a G residue (75).
In all types of stop codon recognition by natural suppressor tRNAs non-canonical base pairing occurs mainly at the first or the third anticodon position. Thus, the misreading of UAG or UAA by tRNATyr(GA) and of the UGA stop codon by tRNACys(GCA) involves a G:G and G:A base interaction, respectively, whereas the reading of UGA by tRNATrp(CmCA) depends upon a Cm:A mismatch at the same position. On the other hand, G:U base pairs are required at the third anticodon position in the unconventional codon reading by tRNAsGln and tRNAsArg (Fig. 12A). The tRNALeu isoacceptor with a CAA anticodon (Fig. 7) that is capable of suppressing the ‘leaky’ UAG codon of TMV RNA (52) necessitates an unusual A:A interaction at the second anticodon position. A few examples exist in prokaryotes in which unconventional base pairing has been reported in the middle position of the anticodon (76). Strigini and Brickman (77) have shown that E.coli tRNATrp(CCA) misreads in vivo not only UGA (14) but also the UAA codon, requiring two C:A base pairs in the latter case. Likewise, yeast cytoplasmic tRNATrp appears to be able to suppress internal UAG codons in vivo, as demonstrated indirectly by amino acid sequence analysis of the translation product produced by UAG readthrough (78). The comprehensive study by Fearon et al. (78) further revealed that the UAG codon—placed within a favourable nucleotide context within the yeast Ste6 gene—directed not only the incorporation of tyrosine and tryptophan, but also of lysine. Misreading of UAG by tRNALys with a CUU anticodon requires an unorthodox U:U interaction at the third anticodon position (Fig. 12A). Readthrough over UAA and UGA codons—placed at identical sites as the UAG codon—has also been observed in vivo, but the nature of the incorporated amino acids has not been elucidated.
|
The mechanism of this unorthodox codon reading is not clear. Based on investigations by Topal and Fresco (79), an increasing number of non-Watson–Crick base pairs have been identified in DNA and RNA helices, some of which were proven by thermodynamic and/or X-ray crystallographic analyses (80–83). Such unconventional base pairs can be formed provided the nucleosides assume minor conformations (syn) and/or minor tautomeric or protonated forms, accompanied often by a slight displacement of the glycosidic bond as in the classical G:U pair (Fig. 12B). Whether these non-Watson–Crick pairs also occur in anticodon–codon interactions remains to be resolved. However, as mentioned above, tRNATyr with the hypermodified nucleoside Q at the first anticodon position does not promote UAG readthrough (27,29,32), emphasising that the guanosine at this position must play a selective role in stop codon suppression (Fig. 12B).
CODON CONTEXT EFFECTS
It has been reported by many groups that translational readthrough of a termination codon is affected by the nucleotide sequences surrounding the ‘leaky’ stop codon in the mRNA of prokaryotic and eukaryotic species (84–86). The codon context can simply consist of only 1–6 nt at the 3' side of the suppressed stop codon or may involve more complex signals like stem–loop or pseudoknot structures.
A comparison of sequences from a number of plant and animal viral RNAs that harbour a ‘leaky’ stop codon reveals that they have similar sequences around the stop codon, preferentially at the 3' side. These readthrough regions can be classified into three groups according to the sequence homology. The type I readthrough region is found exclusively in plant viral RNAs of quite unrelated origin (Fig. 13A). These viruses share the six nucleotides CAA UYA at the 3' side of the suppressed UAG or UAA codon. A number of nucleotide exchanges within this conserved sequence—embedded in a zein pseudogene—were introduced by site-directed mutagenesis. It was found that changes at each position completely abolished or greatly reduced UAG suppression by tRNATyr(GA) in a mRNA- and tRNA-depleted wheat germ extract (39), indicating that this tRNA species needs a very specific codon context for its suppressor activity. The hexanucleotide motif is apparently effectual in quite different systems. Thus, in vitro transcripts containing the TMV-specific readthrough region were translated in a messenger-dependent rabbit reticulocyte lysate and the importance of all 6 nt was shown by measuring UAG suppression via a chloramphenicol acetyltransferase reporter gene (87). Skuzeski et al. (88) had previously come to the same conclusion by studying the transient expression of a ß-glucuronidase reporter gene containing part of the TMV-specific readthrough region in tobacco protoplasts. The effects on suppression of all possible single-base mutations in the hexanucleotide motif indicated that the consensus sequence of the form CAR YYA confers leakiness to all three stop codons.
|
Readthrough regions of type II are found in plant and animal RNA viruses (Figs 13B and 14A). They have in common either a CGG in almost all plant virus and some alphavirus RNAs, or a CUA codon at the 3' side of the suppressed UGA (rarely UAA) codon. Extensive mutational analyses of these triplets provided evidence that 1–3 nt are sufficient to stimulate UGA readthrough. Thus, it has been found that single nucleotide exchanges of either of the three positions in the CGG codon adjacent to the ‘leaky’ UGA in TRV RNA-1 (Fig. 13B) had only marginal effects on UGA suppression by tRNATrp(CmCA) upon translation in a wheat germ extract, whereas a pronounced influence on UGA readthrough was only seen if 2 or 3 nt were replaced simultaneously (89). As a consequence of the more flexible codon context accepted by tRNATrp(CmCA), as compared to tRNATyr(G
A), this tRNA species is able to misread the UGA in nucleotide environments other than the TRV-specific codon context, like those of tobacco mosaic and sindbis virus and to a minor extent that of pea enation mosaic virus RNA (89; Figs 13 and 14). Similarly, it was established that cytoplasmic tRNAArg(U*CG) is capable of reading the UGA in quite different contexts, best of all in the pea enation mosaic and sindbis virus contexts (70). Although natural suppressor tRNAs differ in their efficiency to recognise ‘leaky’ stop codons in various codon contexts, none of them are capable of misreading ‘genuine’ stop codons at the end of open reading frames. For instance, the three natural UGA suppressors, tRNATrp(CmCA), tRNAArg(U*CG) and tRNACys(GCA), are absolutely incompetent for stimulating readthrough over the UGA at the end of the ß-globin cistron in plant and animal in vitro systems (54,63,70,89).
|
The UGA codon and the nucleotides flanking this stop codon in sindbis virus RNA were subcloned into a heterologous sequence and readthrough efficiency was measured by cell-free translation of RNA transcripts in a rabbit reticulocyte lysate. Mutagenesis of residues in the CUA triplet downstream of the suppressed UGA (Fig. 14A) revealed that only the replacement of the cytidine by any of the other three nucleotides drastically reduced the readthrough efficiency (73).
A more complex signal is required for efficient readthrough of the internal UAG of MuLV RNA located between the gag and pol coding regions (Fig. 3). The role of the nucleotide context was investigated in a rabbit reticulocyte lysate and the results indicated that readthrough is mediated by at least 50 nt on the 3' side of the UAG codon (90–92). Within this region, a purine-rich spacer sequence consisting of 8 nt immediately following the UAG and a pseudoknot structure spanning the next 49 nt have been found to be essential for MuLV-specific UAG suppression (Fig. 15). The linear octanucleotide sequence of this bipartite signal is conserved in diverse gammaretroviral RNAs, representing type III of the listed readthrough regions (Fig. 14B). Alteration of 6 conserved nt within this motif by mutational analyses either eliminated or significantly reduced UAG suppression (91). Likewise, all of these viruses can form similar pseudoknot structures on the 3' side of their UAG codons. Substitutions of several conserved nucleotides in loop 2 of the MuLV-specific pseudoknot resulted in detrimental effects on UAG readthrough in vitro, showing the importance of structural elements for efficient suppression (92).
|
Several plant RNA viruses, among them members of the carmovirus genus, exhibit a conserved octanucleotide sequence at the 3' side of the suppressed UAG (or UGA) codon that is highly homologous to the corresponding motif in MuLV RNA (Figs 13C and 14B). However, a pseudoknot structure downstream of this sequence that is similar to the one present in retroviruses has not been detected (93), so it remains open whether these plant viruses utilise a similar strategy, i.e. a bipartite signal for stop codon suppression. Barley yellow dwarf virus (BYDV), from the luteovirus group, utilises stop codon suppression to produce an extended coat protein which becomes part of the virion. Analysis of the cis-acting sequences required for UAG readtrough in oat protoplasts, wheat germ extract and in reticulocyte lysate revealed two essential regions at the 3' side of the stop codon: a C-rich region in the vicinity of the UAG, and another more distal sequence located
700 nt downstream. It is speculated that the distal element interacts with the C-rich region by long-distance base pairing (94). An influence of sequences immediately upstream of the termination codon on the efficiency of translational readthrough has been observed in the yeast S.cerevisiae (95,96), but appears to be of minor importance in plant and animal systems (73,87,89).
The precise molecular mechanism explaining the influence of downstream nucleotides and secondary structures on stop codon readthrough is still obscure. It has been suggested that the major forces involve (i) tRNA selection through stabilisation of the A-site tRNA:mRNA interaction by stacking effects; (ii) interaction between the stop codon and the rRNA; and (iii) interaction between the stop codon and the polypeptide chain release factor (85). Clearly, the competition between the suppressor tRNA and the release factor is of great significance. It has been proposed that the latter binds, in fact, to a tetranucleotide sequence (97). While the preferred termination signal differs in prokaryotic and eukaryotic species, a common feature appears to be a strong bias against a cytidine residue following a termination codon in all organisms (98,99), implying that any termination codon followed by a C is a weak stop codon. Consistent with this assumption is the presence of a C residue at the 3' side of all readthrough regions of type I and type II (Figs 13A and B and 14A). The nucleotide following the stop codon at the 3' side of type III readthrough regions is a G residue, which is the most favoured nucleotide at this position for efficient termination in the eukaryotic tetranucleotide stop signal (99), suggesting that readthrough by natural suppressor tRNAs is impaired. Possibly as a consequence of this unfavourable tetranucleotide signal, putative secondary structures beyond this position at the 3' side are required for efficient readthrough as is the case for the UAG suppression in MuLV and BYDV RNA (92,94).
All three types of codon context sequences stimulating efficient stop codon readthrough in vivo and in vitro are not restricted to the suppression of a single stop codon. For example, UAG, UAA and UGA codons in the TMV-specific context (type I) can be suppressed in tobacco protoplasts (88), rabbit reticulocyte lysate (87) and wheat germ extract (39,41); UGA as well as UAA and UAG codons in the sindbis-specific context (type II) are suppressed in cultured cells of animal origin (23); and all three termination codons in the MuLV-specific context are effective in reticulocyte lysate and infected cells (100). Furthermore, a particular stop codon is not read by a single suppressor tRNA isoacceptor. Thus, it has been shown that the ‘leaky’ UGA codon in the TRV-specific context (Fig. 13B) is recognised in vitro by tobacco tRNATrp(CmCA), wheat tRNAArg(U*CG) and tobacco tRNACys(GCA), albeit with quite different efficiencies (54,63,70). Similarly, the ‘leaky’ UAG within TMV RNA is read by tobacco and wheat tRNATyr(GA), tobacco tRNAGln(CUG), tobacco and mouse tRNAGln(UmUG) and calf tRNALeu(CAA/CAG) in different in vitro systems (19,29,46,48,52).
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONCONCLUSIONREFERENCES |
|---|
The dogma of the non-ambiguity of the genetic code is shattered by the observation that a stop codon can have two or more meanings. Thus, the UGA codon may mediate termination of polypeptide synthesis or trigger the incorporation of tryptophan, arginine or cysteine. Likewise, the UAG termination codon may provoke the incorporation of tyrosine, glutamine or leucine. The misreading of termination codons is achieved by a variety of naturally occurring suppressor tRNAs. All of the nonsense suppressors characterised to date (with the exception of selenocysteine tRNA) are normal cellular tRNAs that are primarily needed for reading their cognate sense codons. As a consequence, recognition of stop codons by suppressor tRNAs necessitates unconventional base pairings in anticodon–codon interactions (Fig. 12).
Direct methods for the characterisation of potential suppressor tRNAs involve the purification of a specific suppressor tRNA from any organism or tissue, followed by examination of its suppressor activity in an in vitro system programmed with RNA transcripts that contain defined viral stop codon readthrough regions (19,48,52,54,63,70). Indirect procedures for the detection of presumable suppressor activities utilise the isolation of readthrough proteins from suitable systems and identification of the amino acid(s) incorporated at the stop codon site by protein sequencing. This allows the prediction of the nature of the putative tRNA species involved in the suppression event (62,71,78). In most cases the same natural suppressor tRNAs have been identified by direct and indirect means, with two exceptions. The incorporation of lysine in response to an internal UAG codon within the yeast Ste6 gene (78) and of serine at the UGA terminating the ß-globin mRNA (71) suggests the existence of further suppressors not yet characterised in detail.
The observation that more than one natural UAG/UAA and UGA suppressor exist in eukaryotes raises the question whether misreading of a stop codon by suppressor tRNAs is regulated or not. One factor that influences readthrough activity is simply the abundance of a specific tRNA suppressor in a given tissue. This is most obvious for the distribution of tRNATyr(GA) versus tRNATyr(QA) [the latter not being a UAG suppressor (Fig. 5)], in various plant and animal tissues. In addition to the available amount, a number of intrinsic features of the suppressor tRNA contribute to the ability to read non-cognate codons. Apart from anticodon–codon affinity, the degree of base modifications within the anticodon or in the vicinity of the anticodon may up- or down-regulate the efficiency of misreading.
In order to out-compete the polypeptide chain release factor for their common target, an absolute prerequisite for the action of natural suppressor tRNA is a suitable nucleotide context, preferentially at the 3' side of the suppressed stop codon. This context may consist of only 3–6 downstream nt, or may involve more complex signals like stem–loop and pseudoknot structures.
Without doubt, a bulk of RNA viruses rely on stop codon readthrough to express part of their genetic information (Fig. 1). Hence, the question arises whether the eukaryotic host organisms also profit from their tRNAs that read termination codons and, thus, facilitate the multiplication of viral pathogens or, in other words, whether these natural suppressor tRNAs have any essential or at least beneficial effect on cellular biosynthesis. One of the very few examples of a natural cellular readthrough protein is the rabbit ß-globin readthrough polypeptide that has been identified in reticulocyte lysate by immunoblotting (71,101). This ß'-globin is generated by readthrough over the ‘genuine’ UGA stop codon at the end of the ß-globin cistron (flanked by a G at the 3' side), but it is unknown whether it has an essential function. However, an increasing number of reports have been published recently indicating that natural suppressor tRNAs have apparently corrected deleterious nonsense mutations within open reading frames of genes from the human and yeast genome in vivo (e.g. 44,78,102,103). Although this is a mostly unpredictable and inefficient event, the low synthesis of a full-length product might be sufficient in special cases to ensure viability of the organism.
|
ACKNOWLEDGEMENTS |
|---|
This review article is dedicated to Professor Hans J. Gross on the occasion of his 65th birthday. About 20 years ago he initiated the characterisation of natural suppressor tRNAs, at a time when their general importance was still unpredictable. H.B. is grateful for his ever lasting advice and encouragement and takes this opportunity to thank all of her former coworkers who have committed a great deal of effort in the field of nonsense suppressors. H.B. also thanks the Deutsche Forschungsgemeinschaft for long-standing support.
|
FOOTNOTES |
|---|
* To whom correspondence should be addressed. Tel: +49 931 8884031; Fax: +49 931 8884028; Email: grimm{at}biozentrum.uni-wuerzburg.de
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONCONCLUSIONREFERENCES |
|---|
-
1 Hatfield,D.L., Levin,J.G., Rein,A. and Oroszlan,S. (1992) Translational suppression in retroviral gene expression. Adv. Virus Res., 41, 193–239.[ISI][Medline]
2 Rohde,W., Gramstat,A., Schmitz,J., Tacke,E. and Prüfer,D. (1994) Plant viruses as model systems for the study of non-canonical translation. J. Gen. Virol., 75, 2141–2149.
3 Farabaugh,P.J. and Björk,G.R. (1999) How translational accuracy influences reading frame maintenance. EMBO J., 18, 1427–1434.[ISI][Medline]
4 Murgola,E.J. (1985) tRNA, suppression, and the code. Annu. Rev. Genet., 19, 57–80.[ISI][Medline]
5 Eggertsson,G. and Söll,D. (1988) Transfer ribonucleic acid-mediated suppression of termination codons in Escherichia coli. Microbiol. Rev., 52, 354–374.
6 Engelberg-Kulka,R. and Schoulaker-Schwarz,R. (1988) Stop is not the end: physiological implications of translational readthrough. J. Theor. Biol., 131, 477–485.[ISI][Medline]
7 Valle,R.P.C. and Morch,M.D. (1988) Stop making sense or regulation at the level of termination in eucaryotic protein synthesis. FEBS Lett., 235, 1–15.[ISI][Medline]
8 Hatfield,D.L., Smith,D.W.E., Lee,B.J., Worland,P.J. and Oroszlan,S. (1990) Structure and function of suppressor tRNAs in higher eukaryotes. Crit. Rev. Biochem. Mol. Biol., 25, 71–96.[ISI][Medline]
9 Santos,M.A.S. and Tuite,M.F. (1993) New insights into mRNA decoding – implications for heterologous protein synthesis. Trends Biotechnol., 11, 500–505.[ISI][Medline]
10 Maia,I.G., Seron,K., Haenni,A.-L. and Bernardi,F. (1996) Gene expression from viral RNA genomes. Plant Mol. Biol., 32, 367–391.[ISI][Medline]
11 Gesteland,R.F. and Atkins,J.F. (1996) Recoding: dynamic reprogramming of translation. Annu. Rev. Biochem., 65, 741–768.[ISI][Medline]
12 Bertram,G., Innes,S., Minella,O., Richardson,J.P. and Stansfield,I. (2001) Endless possibilities: translation termination and stop codon recognition. Microbiology, 147, 255–269.
13 Baranov,P.V., Gurvich,O.L., Fayet,O., Prere,M.F., Miller,W.A., Gesteland,R.F., Atkins,J.F. and Giddings,M.C. (2001) Recode: a database of frameshifting, bypassing and codon redefinition utilized for gene expression. Nucleic Acids Res., 29, 264–267.
14 Hirsh,D. (1971) Tryptophan transfer tRNA as the UGA suppressor. J. Mol. Biol., 58, 439–458.[ISI][Medline]
15 Weiner,A.M. and Weber,K. (1973) A single UGA codon functions as a natural termination signal in the coliphage Qß coat protein cistron. J. Mol. Biol., 80, 837–855.[ISI][Medline]
16 Hofstetter,H., Monstein,H.-J. and Weissmann,C. (1974) The readthrough protein A1 is essential for the formation of viable Qß particles. Biochim. Biophys. Acta, 374, 238–251.[Medline]
17 Pelham,H.R.B. (1978) Leaky UAG termination codon in tobacco mosaic virus RNA. Nature, 272, 469–471.[Medline]
18 Goelet,P., Lomonossoff,G.P., Butler,P.J.G., Akam,M.E., Gait,M.J. and Karn,J. (1982) Nucleotide sequence of tobacco mosaic virus RNA. Proc. Natl Acad. Sci. USA, 79, 5818–5822.
19 Beier,H., Barciszewska,M., Krupp,G., Mitnacht,R. and Gross,H.J. (1984) UAG readthrough during TMV RNA translation: isolation and sequence of two tRNAsTyr with suppressor activity from tobacco plants. EMBO J., 3, 351–356.[ISI][Medline]
20 Watanabe,Y., Emori,Y., Ooshika,I., Meshi,T., Ohno,T. and Okada,Y. (1984) Synthesis of TMV-specific RNAs and proteins at the early stage of infection in tobacco protoplasts: transient expression of the 30K protein and its mRNA. Virology, 133, 18–24.
21 Ishikawa,M., Meshi,T., Motoyoshi,F., Takamatsu,N. and Okada,Y. (1986) In vitro mutagenesis of the putative replicase genes of tobacco mosaic virus. Nucleic Acids Res., 14, 8291–8305.
22 Yoshinaka,Y., Katoh,I., Copeland,T.D. and Oroszlan,S. (1985) Murine leukemia virus protease is encoded by the gag-pol gene and is synthesized through suppression of an amber termination codon. Proc. Natl Acad. Sci. USA, 82, 1618–1622.
23 Li,G. and Rice,C.M. (1989) Mutagenesis of the in-frame opal termination codon preceding nsP4 of sindbis virus: studies of translation readthrough and its effect on virus replication. J. Virol., 63, 1326–1337.
24 Lahijani,R.S., Otteson,E.W. and St.Jeor,S.C. (1992) A possible role for nonsense suppression in the synthesis of a human cytomegalovirus 58-kDa virion protein. Virology, 186, 309–312.[ISI][Medline]
25 Felsenstein,K.M. and Goff,S.P. (1992) Mutational analysis of the gag-pol junction of Moloney murine leukemia virus: requirements for expression of the gag-pol fusion protein. J. Virol., 66, 6601–6608.
26 Jones,D.S., Nemoto,F., Kuchino,Y., Masuda,M., Yoshikura,H. and Nishimura,S. (1989) The effect of specific mutations at and around the gag-pol gene junction of Moloney murine leukemia virus. Nucleic Acids Res., 17, 5933–5945.
27 Bienz,M. and Kubli,E. (1981) Wild-type tRNATyr(G) reads the TMV RNA stop codon, but Q base-modified tRNATyr(Q) does not. Nature, 294, 188–190.
28 Stanley,J. and Vassilenko,S. (1978) A different approach to RNA sequencing. Nature, 274, 87–89.[Medline]
29 Beier,H., Barciszewska,M. and Sickinger,H.-D. (1984) The molecular basis for the differential translation of TMV RNA in tobacco protoplasts and wheat germ extracts. EMBO J., 3, 1091–1096.[ISI][Medline]
30 Barciszewski,J., Barciszewska,B., Suter,B. and Kubli,E. (1985) Plant tRNA suppressors: in vivo readthrough properties and nucleotide sequence of yellow lupin seeds tRNATyr. Plant Sci., 40, 193–196.
31 Beier,H., Zech,U., Zubrod,E. and Kersten,H. (1987) Queuine in plants and plant tRNAs: differences between embryonic tissue and mature leaves. Plant Mol. Biol., 8, 345–353.
32 Junker,V., Teichmann,T., Hekele,A., Fingerhut,C. and Beier,H. (1997) The tRNATyr-isoacceptor and their genes in the ciliate Tetrahymena thermophila: cytoplasmic tRNATyr has a QA anticodon and is coded by multiple intron-containing genes. Nucleic Acids Res., 25, 4194–4200.
33 Okada,N., Noguchi,S., Kasai,H., Shindo-Okada,N., Ohgi,T., Goto,T. and Nishimura,S. (1979) Novel mechanism of post-transcriptional modification of tRNA. J. Biol. Chem., 254, 3067–3073.
34 Sprinzl,M., Horn,C., Brown,M., Ioudovitch,A. and Steinberg,S. (1998) Compilation of tRNA sequences and sequences of tRNA genes. Nucleic Acids Res., 26, 148–153.
35 Johnson,G.D., Pirtle,I.L. and Pirtle,R.M. (1985) The nucleotide sequence of tyrosine tRNAQ*A from bovine liver. Arch. Biochem. Biophys., 236, 448–453.[ISI][Medline]
36 Okada,Y., Shindo-Okada,N., Sato,S., Itoh,Y.H., Oda,D.-I. and Nishimura,S. (1978) Detection of unique tRNA species in tumor tissues by Escherichia coli guanine insertion enzyme. Proc. Natl Acad. Sci. USA, 75, 4247–4251.
37 Landin,R.-M., Boisnard,M. and Petrissant,G. (1979) Correlation between the presence of tRNAHis(CUG) and the erythropoietic function in foetal sheep liver. Nucleic Acids Res., 7, 1635–1648.
38 White,B.N. and Tener,G.M. (1973) Activity of a transfer RNA modifying enzyme during the development of Drosophila and its relationship to the su(s) locus. J. Mol. Biol., 74, 635–651.[ISI][Medline]
39 Zerfass,K. and Beier,H. (1992) Pseudouridine in the anticodon GA of plant cytoplasmic tRNATyr is required for UAG and UAA suppression in the TMV-specific context. Nucleic Acids Res., 20, 5911–5918.
40 Griffey,R.H., Davis,D., Yamaizumi,Z., Nishimura,S., Bax,A., Hawkins,B. and Poulter,C.D. (1985) 15N-labeled Escherichia coli tRNAfMet, tRNAGlu, tRNATyr, and tRNAPhe. J. Biol. Chem., 260, 9734–9741.