key: cord-0005441-urmhw450
authors: Fütterer, Johannes; Hohn, Thomas
title: Translation in plants-rules and exceptions
date: 1996
journal: Plant Mol Biol
DOI: 10.1007/bf00039382
sha: 0f27bb160d422f07b8ca8687d1dac057df6a308f
doc_id: 5441
cord_uid: urmhw450

Translation processes in plants are very similar to those in other eukaryotic organisms and can in general be explained with the scanning model. Particularly among plant viruses, unconventional mRNAs are frequent, which use modulated translation processes for their expression: leaky scanning, translational stop codon readthrough or frameshifting, and transactivation by virus-encoded proteins are used to translate polycistronic mRNAs; leader and trailer sequences confer (cap-independent) efficient ribosome binding, usually in an end-dependent mechanism, but true internal ribosome entry may occur as well; in a ribosome shunt, sequences within an RNA can be bypassed by scanning ribosomes. Translation in plant cells is regulated under conditions of stress and during development, but the underlying molecular mechanisms have not yet been determined. Only a small number of plant mRNAs, whose structure suggests that they might require some unusual translation mechanisms, have been described.

Plant cells harbor three different translation machineries: the eukaryotic ribosomes in the cytoplasm and systems for translation in mitochondria and plastids. The transcriptional and post-transcriptional regulation of chloroplast gene expression has been reviewed recently [233, 282] . This review deals exclusively with translation events and mRNAs in the cytoplasm and concentrates on cis-active sequences that influence efficiency and fidelity of translation. Although the translation of only a few nuclear plant genes has been analyzed so far, it is very safe to assume that translation of plant mRNAs in general proceeds according to the rules established for other eukaryotic systems. The 'exceptions' described mostly apply to viral RNAs encoded by cytoplasmic RNA viruses or by plant pararetroviruses in the nucleus.

Most of the different translation processes have been elaborated in mammalian cell systems, in extracts derived from such cells, and more recently in yeast cells. However, an extract derived from wheat germ has been used in many experiments, in parallel with the most widely used extract of rabbit reticulocytes. In addition, in recent years a variety of experiments have been performed with transiently or stably transformed plants or plant cells. It was concluded from these experiments that most translational features are similar in plant and animal cells ( [127, 320] for review). Some plant translation factors have been characterized and the genes cloned (see review by Browning, this issue). Some of the data suggesting general similarities between translation in mammalian and in plant cells are discussed below.

The vast majority of eukaryotic mRNAs are capped at the 5' end, polyadenylated at the 3 ~ end, and monocistronic, i.e. only one open reading frame (ORF) is translated into a protein per mRNA molecule. Translation of such mRNAs is well explained by the scanning model, which states that 40S ribosomal subunitsassociated with eukaryotic initiation factors (elF) 1A, 3 and the 'ternary complex' of elF2, GTP and initiator methionine tRNA -bind near the capped 5 ~ end of the mRNA, which in turn is associated with a set of factors of the elF4 group. The function of the members of the elF4 group seems to be to prepare the RNA for ribosome binding by removing hindering secondary structures and promoting the linear migration of the bound ribosomes along the RNA in search of a translation start site (scanning). At such a start site, usually an AUG codon, the 60S ribosomal subunit joins on and the resulting 80S subunit begins the synthesis of the encoded polypeptide [ 162, 202,208-210, 238,239, 278, 279, 331] .

In the original scanning model, the 40S ribosomal subunit together with a number of initiation factors was defined as the scanning entity [202] . The precise order of the association of factors and ribosomal subunits with each other and with the mRNA is still disputed, and it is possible that more than one pathway can lead to translation initiation, depending on the prevalent conditions [76, 324] . Particularly, ribosome-independent RNA binding of elF2 has been observed [26, 188] . Alternative hypotheses with factors of the elF4 group [314] or with elF2 [332] as the scanning factor have been proposed. According to these latter models, the 40S subunit binds to the mRNA at a later step (possibly only directly at the AUG start codon). We feel that it is more difficult to explain some of the features of eukaryotic translation with these models than with the 40S ribosome scanning model, particularly some aspects of reinitiation of translation at downstream ORFs (see below). In the following, therefore, we discuss translation mainly in terms of the original scanning model.

The direct influence of the cap in plant translation systems has been demonstrated in many in vitro translation experiments with wheat germ extracts and by transfection of protoplasts with in vitro synthesized RNAs containing or lacking a cap structure [16, 48, 53, 102, 120, 123, 240, 327, 334, 337] . The competitivity of AMV (for virus acronyms, see Table 1 ) RNAs for the translation machinery was directly related to cap accessibility [134] . The cap dependence of in vitro systems in general depends on the extract used, on the precise reaction conditions [202, 224] , on the concentrations of the different initiation factors [102, 324] and on the specific RNAs ( [218, 217, 334] ; see also the review by Galtie, this issue). Stimulatory effects of the cap on expression in the range of 2-to 100-fold were observed. The wheat germ extract is generally more cap-dependent than the rabbit reticulocyte lysate [224] . In vivo, the effect of the cap structure was partially to stabilize the RNA but the main influence was clearly on translation efficiency [123, 124] .

The importance of the RNA 5' end for translation was also indicated by the strong inhibitory effect of an inserted stem structure (-125 kJ/mol) at the immediate 5' end of an RNA encoded by a transfected nuclear gene [112, 114] . Soil Turnip yellow mosaic Tymo 

S c a n n i n g

The ribosome migrates along the RNA from the first point of ribosome association to the translation initiation site. Evidence for such a scanning process is obtained from the of insertion of strong secondary structures or additional initiation sites into the scanned region. The scanning ribosome or its associated factors are apparently unable to melt stem structures of a given stability [ 14, 199] . The required minimal stability has not been determined for plant cells, but hairpins with a free energy below -180 kJ/mol [ 16, 112, 114, 204] were sufficient to significantly inhibit translation of downstream ORFs in a variety of plant protoplast and in vitro systems. Similarly, the introduction of alternative initiation sites reduces downstream translation because the ribosome 'wastes' its initiation capacity at these sites (see below). Both findings show that the ribosome encounters the inhibiting features on the way to the normal initiation site, i.e. it most likely migrates along the RNA. The region upstream of the main initiation site is termed the 'leader sequence' or '5' untranslated region', although the latter designation could be misleading since in some cases such a region might actually contain short ORFs which may be translated (see below).

Only a small number of leader sequences of nuclear plant genes have been systematically compared [186] . The analysis is complicated by the frequent lack of precise RNA data for plant genes. However, it can be concluded so far that the leader sequences of most plant genes (like those of other eukaryotic genes) are shorter than 100 nucleotides, lack strong secondary structures and short ORFs and are thus suited to the ribosome-scanning mechanism. A few exceptions will be discussed below. The leader sequences of many RNA viruses seem to have primary or secondary structures that allow a particularly efficient translation, often even in the absence of a cap structure. These features are discussed in detail by Gallie (this volume). In general, longer leaders (e.g. 80 nucleotides) result in higher translation rates than shorter leaders (e.g. 10 nucleotides). The exact difference depends on the assay system and salt conditions [201] ; in transfected plant protoplasts an approximately twofold effect was reported [126] , while the effects in the reticulocyte lysate can be greater [206] . However, mRNAs with leader sequences less than 10 nucleotides long can still be efficiently translated in plant cells (G. Chen et al., unpublished observation) and are actually found in some coat protein mRNAs of RNA viruses (e.g. BMV).

Upon initiation, the ternary complex dissociates from the ribosome as elF2-GDP. Since, according to the scanning model, the ternary complex is delivered to the initiation codon together with the ribosome, ribosomes can initiate translation only once unless a new ternary complex can be recruited. Due to the scanning process, this one-and-only initiation site is the one located closest to the 5' end of the RNA. Translation usually begins at an AUG codon, and the efficiency of AUG codon recognition is modulated by the sequence context of this codon [198] . Sequence compilation of plant translation initiation regions has suggested that the most frequent (and therefore regarded as optimal) AUG context for plant genes is similar to that of mammalian genes (AACAATGGC [55, 186, 227] ). The most crucial positions in both cases are a purine at position -3 and a guanine at position +4 (where the A of the AUG is +1). For mammalian cells, an influence of positions +5 and +6 has been recently documented [32, 144] ; other positions seem to be less important. Start codons that deviate from the optimal context at one or more of the crucial positions may be recognized less efficiently and allow the passage of ribosomes to further downstream start codons (leaky scanning). The influence of sequence context on initiation in plant cells has been directly verified by mutagenesis studies in vitro, in plant protoplasts, and in transgenic plant cells [6, 77, 87, 146, 203, 220, 253, 328] . Differences in initiation efficiency between codons in optimal or suboptimal contexts strongly depend on the conditions of the assay system [87, 203] . While some workers concluded that the context of the initiation codon in plants is of minor importance [227] , others have shown that reduction of the Mg 2+ concentration from 3 mM to below 2 mM produces significant codon context discrimination, similar to that in mammalian systems [86, 203] . However, it is unclear whether these optimal in vitro conditions also apply to translation in vivo. The (suboptimal) codon context of soybean lipoxygenase ORFs (AAAG.ATG.TTT) was found to give 10 times higher /3-glucuronidase (GUS) expression rates than the (also unfavorable) context in the standard GUS expression plasmid pBI 211 [192] . In transgenic plants, translation initiation patterns from two consecutive, in-frame AUG codons were found to be drastically different in different parts of the plant (and at different developmental stages) [ 161 ] .

The efficiency of start-codon recognition can be influenced by features that lead to a prolonged pause of the scanning ribosome at the position of the codon. A secondary structure element located 14 nt downstream of the start codon was found to increase initiation efficiency at weak AUG and at non-AUG codons [205] . A similar pause results from a downstream initiation event and can cause more efficient recognition of a properly spaced upstream AUG [87] . In contrast, initiation at a downstream codon may be negatively influenced by overlapping translation from an upstream start site, possibly because of interference by the translating 80S ribosomes [211 ] .

Decoding of the RNA occurs at the aminoacyl tRNA site (A site) of the ribosome in 3-base steps by an amino-acylated tRNA with appropriate anticodon. Peptide elongation occurs by transfer of the nascent peptide from the tRNA of the previous decoding step, which is located at the peptidyl tRNA site (P site), to the aminoacyl tRNA at the A site. The mRNA(codon)-tRNA-peptide complex is then translocated to the P site and the A site becomes free for interaction with the next tRNA. A third site (exit or E site) on the ribosome interacts with the unloaded tRNA before release. The binding of a tRNA to its codon differs in the A and P sites [272] and interactions between the tRNAs at the different ribosomal sites influence the kinetics of decoding [222, 312] . Translation elongation is promoted by eukaryotic elongation factors eEFI and eEF2 and requires the hydrolysis of GTP [281 ] . Unlike the scanning process, ribosome translocation during peptide synthesis is relatively insensitive to inhibition by RNA secondary structure. Even basepairing between long sense and anti-sense RNA stretches can be disrupted, provided it begins more than about 20 nt downstream of the AUG codon and, therefore, does not interfere with the formation of the 80S ribosome [ 185, 221, 300] . Structural elements may nevertheless cause transient pausing of translating ribosomes (see below). The speed of translation elongation can vary between less than 2 to about 10 amino acids per second per translating ribosome in eukaryotic cells, depending on the cellular conditions and the mRNA ( [289] for review). While initiation is generally the rate-limiting step in translation, elongation can become limiting with some particular ORFs [62, 94] or under certain physiologic-al conditions [289] ; for example, elongation rates are reduced at the slightly acidic pH that is induced by oxygen deprivation [354] . For plants, regulation at the level of elongation has been proposed for two oat seed proteins [37] and in response to heat-shock [10] , oxygen deprivation [99, 348, 354] , wounding [248] , and light [28, 305, 310] .

The nascent peptide is transported through a ribosome tunnel to the surface [71, 364] , where it can interact with signal recognition particles or other cytoplasmic factors [359] . In bacteria, slowing down of elongation can be caused by the nascent peptide [145, 226] , and the penultimate amino acid in particular may influence the termination step [245, 249] . These effects are probably caused by interactions between the peptide and translation factors, rRNAs or the tunnel through which it is transported to the surface of the ribosome [154] , and it is likely that similar mechanisms exist in eukaryotes. Ribosomal pause sites not ascribed to secondary structure elements have been detected for a number of genes [ 193, 194, 263, 362] . It has been proposed that codons that are recognized by rare tRNAs ('rare codons' [47] ) interfere with efficient translation [371 ] and could be a regulatory feature of gene expression [66, 291 ] . The speed of decoding could also influence the folding of the nascent peptide chain [213] , and possibly the targeting [263] . In microorganisms, rare codons tend to be avoided in highly expressed genes and adjustment of the codon bias can result in increased expression of a gene [78, 315] . In yeast [ 166] and Escherichia coli [59, 135] , a number of rare codons near the initiation site is required to substantially reduce expression of the respective ORE It is likely that all these mechanisms also apply to translation in plant cells. A compilative analysis of protein-coding sequences revealed that the codon usage in plants differs from other organisms, and differences can also be discerned between plant families [49, 252] . In plants, particularly in monocots, highly expressed genes show no special preference for frequent codons [49, 252] . Nevertheless, bacterial genes have been modified to comply with the plant codon usage and were indeed expressed to a much higher level [107, 212, 267] . During this mutagenesis, however, signals leading to potential aberrant RNA processing were also altered, and it is still not clear which of the features of the new RNA sequence are responsible for the elevated expression.

For the stop codon (UAA, UAG, UGA) usually no complementary tRNAs are available and release factors associate with the ribosome; this causes release of the polypeptide chain and termination of translation [45, 69, 338, 344, 345] . In all eukaryotes studied so far, the three stop codons are recognized by only one release factor, eRF1 [106, 372] or SUP45 in yeast [317] . Recognition or termination activity is stimulated by a second protein factor, eRF3 [372] or SUP 35 in yeast ( [317, 318] for review). The immediate sequence context of stop codons can modulate recognition efficiency. Hierarchies of termination efficiency have been established for E. coli [271] , mammals [236] , and yeast [33] , and sequence comparisons suggest the existence of similar context effects in other eukaryotes, including plants [9, 42, 43, 55, 344] . The fate of the ribosome after translation termination remains unclear. However, at least some ribosomal subunits must resume scanning after translation termination since in some cases ribosomes are able to reinitiate translation at other downstream start sites. It has been proposed that the reinitiation capacity is modulated by sequences surrounding the stop codons [142, 242] , but it is not known whether this occurs at the level of ribosome-bound factors or at the level of ribosome-RNA association.

Translation of simple, monocistronic RNAs is easily explained by the scanning model. The translation of exceptional RNAs with unconventional initiation sites or encoding more than one protein or peptide, discussed in the following sections, is not fundamentally different but requires a few modifications or additions to the original model. Most of these RNAs are derived from plant viruses [168, 243, 285] .

Translation in mammalian and insect cells can commence at codons which differ from AUG in one position. In artificial constructs, most of the possible AUG derivatives are active to some extent [203, 237, 262] . In natural mRNAs, CUG is mainly found as an alternative start codon [36, 54, 151, 273, 342, 343] , but indica-tions of the activity of AUC and ACG have also been presented [ 19, 72] . In Saccharomyces cerevisiae, non-AUG codons are recognized very inefficiently [65] . Non-AUG recognition can be enforced by mutations in elF2/3 [89] and elF2~, [90] . For plants, the activity of such codons was analyzed with artificial constructs in wheat germ extracts [203] and in transfected protoplasts [138] . In both systems, initiation at non-AUG codons was more context dependent than that at AUG codons, and in the wheat germ extract recognition was much higher at high Mg 2+ concentrations [203] . For protoplasts, a hierarchy of activity was found of CUG (30% efficiency of an AUG codon), GUG, ACG (15%), UUG, AUA, AUC, AUU (2 to 5%), and AAG, AGG (<0.1%) [138] .

It has been inferred that non-AUG initiation occurs in a mutated version of a CaMV RNA at the start of the pol gene [297] . This example is important in that it shows that mutation of the start codon of an ORF does not necessarily abolish ORF expression completely.

An AUU codon opens the first ORF of rice tungro bacilliform virus (RTBV). Translation efficiency at this codon is low in RNAs where the position is reached in a normal scanning mechanism but reaches about 10% of AUG efficiency at its natural position downstream of the long RTBV leader sequence [ 116] . This leader sequence contains a number of features that inhibit scanning and most likely a specific mechanism is required for ribosomes to reach the region downstream of the leader (see below). This mechanism may also lead to enhanced recognition of the AUU codon.

No other plant genes with translation initiation exclusively at non-AUG codons have been described. In mammals, most ORFs with a non-AUG initiation site also have a normal AUG initiation site further downstream. Additional non-AUG initiation is found mainly for ORFs encoding regulatory proteins that are located downstream of a long, structured leader sequence. The protein variants with N-terminal extensions may have regulatory effects different to the 'normal' one [2, 46] and initiation at the additional sites can be regulated by the cellular conditions ( [151, 152] for review). Similar translation events may occur in plants, but these rare and probably important events have not yet been detected because few plant genes and their protein products have been analyzed to the required extent. In soil-borne wheat mosaic virus (SBWMV), an N-terminal extension variant of a 29 kDa protein with unknown function may be produced by initiation at an as yet unidentified, non-AUG codon upstream of the first AUG codon on RNA 2 [303] .

80S ribosome formation at an AUU codon in the leader sequence of TMV RNA (f~) has been reported [101, 173, 340] . The codon is in frame with the TMV 126 kDa protein but it is not known whether it serves as an initiation site for a variant of this protein. An ORF fused to the f~ leader was found to be translated also from an upstream AUU codon with high efficiency [323] . The 5 ~ leader sequence of AMV RNA 3 even binds two 80S ribosomes, presumably also at AUU codons [268] ; one 80S ribosome also binds to an as yet unassigned position in the leaders of BMV RNA3 [5] and TYMV RNA [101] . The relevance of these binding events for translation of the respective RNAs remains unclear. Disome or trisome formation is not involved in the translation stimulating effect of these leader sequences [126, 309] .

Non-AUG initiation can occur at inconspicuous positions of an mRNA: in the leader sequence of the CaMV 35S RNA, efficient initiation at a CUG codon that leads to translation of a short ORF has been detected (disrussed by Gordon et al. [138] ). Again, it is not known whether this translation event has any functional importance, but the example shows that caution is required when the possible translation events on one RNA are only deduced from sequence data.

If the translation start site for an ORF is not the most 5 ~proximal on a given RNA, three different possibilities exist for the route by which a translation-competent ribosomal subunit may reach this codon ( Fig. 1 ).

1. Most simply, but least compatible with the scanning model, ribosomes avoid upstream initiation sites by jumping or by directly binding to a downstream site.

2. In line with the scanning model, the upstream codon may be avoided by a fraction of ribosomes due to effects of the sequence context (leaky scanning). A special case for leaky scanning are start codons that are so close to the RNA 5 r end that the decoding site of the 43S subunit is already downstream when the ribosome is bound to the 5' end [206, 298] .

3. In cases where the upstream OR1 ~ terminates before the downstream initiation site, ribosomes may continue scanning and reinitiate translation; reinitiation requires the recruitment of initiation factors by the scanning ribosomes. Examples of all these mechanisms are known for specific mammalian RNAs [207] and at least some have been described for RNAs translated in plants.

RNAs with multiple initiation sites can be grouped in three classes ( Fig. 1 ):

1. RNAs that contain one or more short ORFs (sORFs) in their leader region; the sORF may terminate before or overlap the main ORF; 2. RNAs that contain multiple, overlapping long ORFs; 3. RNAs that contain multiple, non-overlapping long ORFs. The distinction between a 'coding, long' ORF and a sORF is somewhat arbitrary. Usually a sORF is shorter than 50 codons and the putative translation product is not known to have any function besides possibly regulating the translation efficiency in cis.

Systematic comparison of the 5' leader sequences of 79 plant genes showed that 6 of these contain upstream AUG(s) [ 186] . Our own, unsystematic inspection of plant gene sequences also showed that about 10% contain one or more AUG codons in the 5' leader sequence. In many cases, it remains to be seen which of these AUG codons are present in functional mRNAs; some may be removed by splicing. Unspliced RNAs are not necessarily degraded in plant cells [136] and may be represented in cDNA preparations. In other cases, multiple transcription start sites may lead to the production of mRNAs with different 5' leader sequences. The effect of upstream sORFs on translation has been studied for a few nuclear-encoded mRNAs: The mRNA for the maize Lc protein contains a 38 codon long sORF (with a total of three AUG codons) in its 256 nucleotide long leader region, which ends 62 nucleotides upstream of the Lc initiation codon [73] . As expected from the scanning model, the presence of this sORF causes a reduction in Lc translation. In an assay that used the transcription-transactivating activity of the Lc protein for quantification of Lc translation, a 30-fold repression was observed in biolistically transfected maize aleutone cells. The repressing effect depended greatly on the intactness of the sORF sequence; all point mutations alleviated the effect at least partially. The mutations also included some which led to conservative amino acid exchanges or were even silent [73] . Examples in which the coding sequence of an sORF is important for its repressing effect are documented for yeast [242, 358] and mammalian cells ( [50, 80, 133] for review) but the stringent sequence requirements observed for the Lc leader sORF are unique. It is noteworthy, that the mutated versions of the sORF apparently had almost no inhibitory effect on downstream translation although they still contained AUG codons and should therefore be translated.

The mRNA leader of the maize opaque-2 gene contains three, partially overlapping sORFs which inhibit translation of the downstream opaque-2 ORF about fivefold in tobacco protoplasts [225] . Individual sORFs showed a similar inhibitory effect. When the leader sequence was modified such that either the first or second sORF was elongated to overlap the downstream, long ORF, inhibition was increased. This is taken as evidence that at least the first two sORF initiation sites are actually used as efficient translation start sites and that the normally observed initiation at the opaque-20RF AUG is a reinitiation event [225] . As stated above, it is possible that even a poorly translated overlapping ORF may interfere with downstream initiation [211 ] and it cannot be excluded that the opaque-2 ORF is normally translated by leaky scanning, which is inactivated in the overlapping ORF configuration.

The leader sequences of several plasma membrane proton ATPases from Nicotiana plumbaginifolia, tomato and Arabidopsis thaliana also have sORFs. The sORF in the pmal RNA is 10 codons long, contains two AUG codons in suboptimal sequence context and ends about 60 nucleotides upstream of the coding ORF. Removal of any single AUG codon had no effect on translation but removal of both increased translation about twofold in wheat germ extracts and in protoplasts [241] .

The leader sequences of several Arabidopsis thaliana homeobox genes contain sORFs. For the homeobox gene 1 (ATH1), the leader is longer than 600 nucleotides. The 5' halves of the probably two different mRNA versions have not yet been characterized but the 3 ~ half contains a number of sORFs and inhibits translation in an in vitro system about fivefold [276] .

It is unknown whether the leader is involved in the complex regulation of ATH1 activity.

An sORF in the leader of the soybean aminoalcohol-phosphotransferase 1 (AAPT1) inhibited translation of the downstream ORF in yeast and thereby interfered with complementation of a yeast mutant by this plant gene [84] . Again, it is unknown whether the sORF regulates AAPT1 expression in plants.

Negative effects of sORFs on downstream translation also have been observed in artificial constructs made for plant transformation [31, 275, 284] . The effects of several features of an upstream sORF were tested in constructs in which a reporter ORF was preceded by an artificially designed sORF. The length of the sORF was modulated by sequence multiplication and the number of potential start codons was varied. The inhibitory effect increased with the length of the sORF. Even the shortest possible sORF, consisting only of an AUG codon, reduced downstream translation by about a factor of two. Intermediate length sORFs (around 30 codons) reduced fivefold and sORFs longer than 100 codons completely abolished downstream translation [113] . In plant protoplasts, sORFs with additional internal AUG codons were not more inhibitory than those without, probably because the first start codon was in an optimal sequence context and thus all approaching ribosomes initiated there. If the sORF overlapped the downstream ORF, translation was completely precluded. A limited number of sORF sequence variants in the coding sequence or around the stop codon were also tested, but no further features influencing the inhibitory potential of the sORF could be discerned [ 113] . From all these results it is clear, that sORFs can be inhibitory because they are translated and thus reduce translation initiation at the downstream start codon. However, initiation at an sORF does not completely preclude translation reinitiation at a further downstream ORF by the same ribosome [105, 113, 275] . This process should require the recruitment of a new set of initiation factors. In the artificial constructs described by Fiitterer and Hohn [ 113] , the reinitiation efficiency was dependent on the length of the sORF and reinitiation was observed despite an intercistronic distance of only 16 nucleotides.

The effect of sORFs on translation of a further downstream ORF has also been studied in mammalian and yeast systems. A similar length dependence of the inhibitory effect of the sORF was observed in mammalian cells [228] . Besides, on the sORF's coding sequence (see above), the efficiency of downstream translation was also dependent on sequences downstream of the sORF's termination codon [142] and on the distance between sORF and downstream ORE Distances greater than 50 [200] or 200 [ 1] nucleotides were required and in the yeast system, downstream translation was strongly influenced by the physiological state of the cells ( [1, 165, 260, 358] for reviews). Tissuespecific effects of sORFs on downstream translation have also been observed in mammalian cells [375] . It is likely that similar effects also exist in plants. So far the only safe fact seems to be that an sORF-containing leader always causes lower expression than an sORFfree leader. In particular, for long leader sequences with several sORFs, the multitude of possible activities of ribosomes, such as translation, leaky scanning, stalling, translation-termination-dependent dissociation from the RNA, non-productive scanning (due to the lack of initiation factors), sequence-or distancedependent reinitiation etc., makes a precise prediction of the degree of inhibition impossible.

Up to 14 AUG codons are found in the leader sequences of the pregenomic RNAs of the plant pararetroviruses, and the leaders of the CaMV 35S RNA [17, 109] and the RTBV 35S RNA [58] are indeed inhibitory to downstream translation. For CaMV, inhibition was up to 100-fold in some protoplast systems derived from non-host plants [ 17, 109] . In other protoplast systems, including some from virus host plants, the inhibition was much less severe (around two-to six-fold; [109] ). These leader sequences can also form extensive secondary structures [108, 147, 158] . It is possible that most of the translation inhibitory effect of the leader was caused by secondary structure rather than by the sORFs, as was shown for the c-sis protooncogene RNA with its long, sORF-containing leader [170] . It is, however, likely in CaMV that some of the sORFs inhibit downstream translation since they are inhibitory when located in the supposedly unstructured leaders of truncated mRNAs. The first three CaMV sORFs are very short (2-5 codons) and their start codons are in suboptimal sequence context. Consequently, they are only slightly inhibitory for translation of a reporter gene fused to sORFs located further downstream [108, 114] . The longer sORFs in the 3' half of the leader, at least one of which has a start codon in optimal sequence context, are more inhibitory [111] . Translation of a reporter ORF positioned in the center of the leader follows the scanning model. Either enough ribosomes avoid initiation at the upstream sORFs because of the unfavorable sequence context or they regain initiation capacity after sORF translation [112, 114] . The first longer ORF downstream of the CaMV 35S RNA leader is thought to be translated by modified scanning which was termed ribosome shunt [114] . A similar process might be active on the RTBV 35S RNA [116] . In the shunt, initially scanning ribosomes are transferred directly from a donor to an acceptor site without linear scanning of the intervening region. The ribosome migration on the 35S RNA leader has been studied by insertion of strong stem-loop structures and additional ORFs at various sites. (It is noteworthy that in all cases where no ribosome shunt was involved, these additional elements produced exactly the effect that would have been predicted by the scanning model. Therefore, these experiments are also another confirmation of the general features of the scanning model in plant cells.) The shunt could also be detected in trans, albeit at low efficiency [114] and was independent of viral proteins. Since shunt efficiency showed a certain cell-type dependence, involvement of cellular factors is likely [109] . Although mutagenesis analyses have defined regions in the leader sequence of the CaMV and RTBV 35S RNAs that are important for the shunt process [58, 111, 114, 116] , the structural features that allow a shunt are only poorly defined so far. It is assumed that longrange RNA interactions are involved, probably with the assistance of associated proteins. Computer programs predicted a stable stem-loop structure for almost the complete CaMV and RTBV leader sequences [ 108, 158] , and for CaMV such a structure was also predicted by an analysis of folding parameters of the growing RNA molecule [147] . Direct analysis of the in vitro structure of the leader confirms the presence of the predicted structure or a similar one (M. Hemmings-Mieszczak, G. Steger, and T. Hohn, unpublished observations). In an alternative folding of the CaMV 35S RNA leader sequence, the presence of pseudo-knots in the acceptor as well as in the donor regions was noted (K. Gordon, pers. comm.). Which of these structures, if any, is important for the ribosome shunt has still to be determined. The location of the ribosome acceptor region was defined for CaMV and for RTBV with some precision. In both cases, it lies immediately downstream of a CT-rich region and for RTBV it involves the AUU start codon of the first ORF [ 114, 116] . The efficiency of recognition of this AUU start codon is increased above that of the same codon reached by scanning, possibly because a shunted ribosome has more time for recognition before scanning is resumed [116] . Recently, translation of CaMV 35S RNA leader constructs in wheat germ extracts produced results that suggest that a ribosome shunt can also occur in this heterologous in vitro system (W. Schmid-Puchta, D. Dominguez and T. Hohn, unpublished observations).

Ribosome shunt-like mechanisms have been described for translation of Sendai virus RNA [72] , the adenovirus tripartite leader (A. Yueh and R.J, Schneider, pers. comm.) and for a papova-virus (G. Hobom, pers. comm.).

RNAs that encode more than one protein in consecutive or overlapping ORFs have not been described for plants; however, a number of plant viruses use such RNAs (Fig. 1) and apply one or several unusual mechanism(s) for their translation (Fig. 2) .

In the most easily explainable cases, translation initiation at two start sites is regulated by the rules of leaky scanning. This requires that the first AUG is in an unfavorable sequence context. The second initiation site can be in the same or in a different reading phase, giving rise to an N-terminal deletion variant of the first protein, or a completely different protein, respectively. The degree of leakiness of an AUG codon can be high; a potential start codon in the leader of PPV was found to be recognized very inefficiently [280] while for two ORFs of the luteoviruses PLRV and BYDV-PAV ratios between 100:1 and 1:7 for translation of two overlapping ORFs have been described [86, 87, 322] . This mainly depends on the sequence context [87] and on the conditions of the translation system (cation concentration [89] ). A protoplast system was more discriminating (100-fold difference)between an optimal and a suboptimal start codon than the reticulocyte lysate [87] . In those cases where leaky scanning has been observed or postulated, the first start codon is usually in an unfavorable sequence context ( Table 2) .

As described above, the ratio of expression from two overlapping ORFs depends not only on the direct Pol is translated as ORF1-Pol fusion protein upon a frameshift linking the two ORFs from the full length RNA. ORF 4, which overlaps with CP is translated upon leaky scanning and ORF 5 is translated as CP-ORF 5 fusion protein by readthrough of an amber stop codon from the longer subgenomic RNA. A site close to the 3 t end of the two larger RNAs promotes translation in cis. CaMV: Three mRNAs were identified, the pregenomic 35S RNA, the subgenomic 19S RNA and a spliced version of the 35S RNA. The first ORF (ORF VII) of the 35S RNA is translated after a shunt mechanism. Whether translation of the first ORF (ORF III) from the spliced RNA also requires shunting or another mechanism (leaky scanning, reinitiation) is not known. Translation of the second and probably consecutive ORFs from unspliced and spliced 35S RNA occurs upon transactivation involving the translation product of ORF VI (transactivator) derived from the 19S RNA. The transactivator is the only larger CaMV protein obtained by classical translation. sequence context of the start codons but may also be influenced by structural features of the RNA. In mammalian cells, mutual influences of overlapping ORFs during the elongation phase of translation have been described, probably caused by the different speed of decoding of the two overlapping reading phases [94] . A similar effect may account for observations made with start codon mutations of the overlapping TYMV ORFs for a 69 kDa and a 206 kDa protein, where it was found that elimination of the first AUG had no positive effect on translation from the second one, while elimination of the second AUG increased translation from the first one [355] . This suggests that translation (initiation or elongation) of the downstream ORF is also rate limiting for the overlapping upstream ORE Initiation efficiency at the upstream BYDV coat protein (CP) ORF was positively influenced by efficient initiation of the 17 kDa ORF starting 40 nucleotides downstream. Probably the pause resulting from initiation at the downstream AUG results in ribosome stacking and provides more time for recognition of the upstream AUG [87] . Leaky scanning is at least partially responsible for translation of the 95 kDa protein of CPMV M RNA, which initiates about 350 nucleotides downstream of the 105 kDa protein; no AUG codon in any of the three reading phases is found in this intervening region [350] . Translation of RNAs of the rice dwarf phytoreovirus [321] , the satellite tobacco mosaic virus [244] and luteo-, tymo-and tombus viruses also probably involves leaky scanning ( [243] for review).

In at least two cases, leaky scanning is the mechanism of translation of subsequent, non-overlapping ORFs. An ORF on PCV RNA 2 is located downstream of the 620 nucleotides long coat protein ORF but is accessible to scanning ribosomes because this long upstream region is devoid of AUG codons apart from the one opening the CP ORE Leaky scanning was deduced from the negative effect on downstream translation of the insertion of stem structures or of additional AUG codons into the upstream region [ 164] . A similar, even more extreme case is found in RTBV, where the 900 nucleotides encoding ORFs I and II upstream of ORF III contain only one AUG codon that opens ORF II and has an unfavorable sequence context. A similarly peculiar bias is also observed in the genomes of the other badnaviruses. Improvement of the efficiency of RTBV ORF I translation by mutating its AUU initiation codon to AUG drastically reduced ORF II and III expression, as expected for a leaky scanning mechanism (unpublished observations).

If the upstream ORF is opened by an efficiently recognized start codon and/or if no particular bias against the presence of additional AUG codons in any of the three reading phases exists, a following ORF will normally not be translated because leaky scanning is impossible and reinitiation of translation is inefficient. In in vitro systems, translation efficiencies for downstream ORFs of 1-20% have been found [ 137] , but in plants or protoplasts downstream ORFs are expressed with considerably lower efficiency. This has generally been experienced with constructs designed to express a marker ORF downstream of another ORF in transgenic plants [8, 174, 374] . An up to 1500-fold reduction of expression efficiency of the downstream ORF was observed [8] ; however, even low expression efficiencies can be sufficient to cause phenotypes, such as antibiotic resistance [8, 174] or GUS staining [374] . In one exceptional case, the introduction of a complete bacterial arylsulfatase ORF upstream of a GUS ORF seemed not to influence GUS translation in transiently or stably transformed plant cells [63] . However, in the respective dicistronic constructs, GUS translation begins at an in-frame AUG accidentally present considerably upstream of the AUG that is used in the monocistronic reference construct, and leads to different GUS proteins with possibly different enzyme stabilities or activities. In transient expression experiments in a variety of plant protoplast systems, expression of a downstream ORF was at or below the detection limit [34, 112, 139, 294] . In Orychophragmus violaceus protoplasts, detectable expression of a CAT ORF downstream of a GUS ORF was only observed when the intercistronic distance was increased to more than 300 nucleotides [112] .

The pregenomic RNA of the caulimoviruses and probably also a number of internally spliced derivatives of these RNAs serve as polycistronic mRNAs for a number of viral proteins [34, 88, 110, 139, 196, 294] . The respective ORFs closely follow each other without long intercistronic distances, are often opened by efficiently recognized start codons, and usually also contain internal AUG codons. In plant protoplasts and in transgenic plants, most of the downstream ORFs are indeed not or only poorly expressed unless the virus encoded transactivator (TAV) is present (Fig. 2) . Transactivation activity has been demonstrated for the ORFs VI of CaMV [34, 83, 112] and FMV [139, 294, 295] and it likely resides also in the Corresponding ORFs of CERV, SoCMV, and PC1SV [83] . The TAV protein specifically enhances the translation of a downstream ORE Expression was obtained from totally artificial, polycistronic constructs as long as the ORF organization allowed a reinitiation mect~anism, i.e. when long ORF overlaps were avoided [112, 113] . In the artificial polycistronic RNAs, transactivation was particularly efficient when the first ORF was around 30 codons long; shorter and longer ORFs were less effect-ive [113] . Transactivation was observed for several ORFs following such a short ORE The polar effects of the insertion of stem-loop structures into polycistronic mRNAs and the specificity for non-overlapping ORFs suggest that transactivation causes enhanced reinitiation of translation. The dependence of transactivation efficiency on first ORF length suggests, that the transactivator directly or indirectly acts on the translating (or terminating) ribosome [ 112, 113] . The optimal length of the first ORF of 30 codons may be significant since a 30 amino acid nascent peptide is just long enough to emerge from the translating ribosome [231, 359] . It appears possible that at this stage of translation a structural change of the ribosome occurs which leads to loss of a residual reinitiation capacity and also to a loss of transactivation responsiveness. The absence of a requirement for cis-active sequences does not necessarily implicate the lack of such sequences in the CaMV genome. The presence of the CaMV ORF VII supported transactivation [34] although an artificial ORF of similar length did not [113] . The expression constructs also contained the CaMV polyadenylation signal, which contributes the terminal 200 nucleotides of the transcripts. Studies with FMV suggested that cis-active sequences are required for transactivation in this case. These sequences are located at the end of the FMV 35S RNA leader sequence [140] and in the ORF VI coding region, which was effective as a 3'-untranslated region in the respective plasmid constructs [295] . For the former, however, no distinction was made between sequences that allow ribosome access to the region downstream of the leader (a FMV ribosome shunt) and sequences that are specifically involved in transactivation, nor was the precise configuration of the sORFs with respect to the reporter ORF discussed. In CaMV, the TAV also stimulates translation directly downstream of the leader [114] . This increased expression was dependent on the shunt process and was apparently not caused by the action of the TAV on ribosomes that migrated through the leader sequence. In the case of the stimulating sequences in the 3'-'untranslated' region, it remains to be seen whether the effects are due to stimulation of translation or to RNA stabilization by TAV-induced translation of ORFs in this region. Strong reductions of mRNA levels have been reported for constructs that contained inefficiently translated 3' ORFs [8, 174] and are in general often observed for RNAs with premature stop codons and, therefore, long untranslated 3' regions in plants [347] and other eukaryotes ( [232, 259] for review).

The transactivating function could be localized to the central third of the TAV protein [83] . To achieve normal levels of transactivation with a peptide comprising only this portion (mini-TAV), 100-fold more DNA encoding the TAV polypeptide had to be transfected. It is not known whether this lower efficiency is due to a defect in transactivation itself or to problems with protein folding and stability. The mini-TAV was only active in Nicotiana plumbaginifolia protoplasts, while the full-length TAV was active in a number of dicot plant protoplasts and also in maize protoplasts [34, 139] . TAV was found to be associated with polysomes and, in an overlay binding assay, also with an 18 kDa ribosomal or ribosome-associated protein from plants and also from yeast (A. Himmelbach, Y. Chapdelaine and T. Hohn, unpublished observations). An RNAbinding activity of TAV residing outside the minimal transactivating region may enhance activity by increasing TAV concentration near the RNA [83] . Crossing of transgenic Arabodopsis lines containing either the transactivator or a dicistronic GUS reporter construct revealed that transactivation can also be obtained in transgenic plants [374] . No tissue or development specificity was observed. Arabidopsis and tobacco plants expressing the transactivator are not easy to obtain and show abnormal phenotypes [18, 325, 374] . At present it is unclear whether these phenotypes are caused by transactivation or some other activity of the protein.

Recently, TAV-dependent transactivation of polycistronic translation was also described for Saccharomyces cerevisiae [299] . This latter finding opens up many possibilities for the study of the mechanism and for the characterization of the cellular partners involved.

Downstream ORFs on a polycistronic mRNA could in principle also be translated in a scanning-independent mechanism by ribosomes that enter the RNA at an internal position (internal ribosome entry site, IRES) as was originally shown for picornaviruses ( [3, 57, 181, 189, 266, 292, 314] for reviews) and has since been described for a number of animal cellular and viral mRNAs [27, 257, 342, 343] ; it may also occur in yeast [175] . A number of plant viral mRNAs are not capped [366 for review] and, therefore, must have a cap-independent ribosome entry site. Cap-independent translation initiation might still be dependent on ribosome association with the RNA 5' end and not involve a true IRES. For CPMV, three members of the potyvir-us group, BYDV and STNV, cap-independent translation was shown to be conferred by viral sequences in vitro and partially also in vivo. In the potyviruses, the important RNA sequences are located upstream of the AUG codon that initiates translation of the polyprotein (TEV [53] , PPV [280] , PVY [218] , TuMV [16] ); in BYDV [353] and STNV [74, 334] sequences at the 5' end and near the 3 ~ end of the RNA are required, and in CPMV, the sequence resides between the two first start codons on the M RNA (positions 161 and 512 [333, 350] ). Competition experiments suggest that the function of the TEV and TuMV sequences depends on interaction with a cellular factor also involved in cap-dependent translation [16, 53] . The TuMV sequence is active independent of its orientation and, like the CPMV sequence, also downstream of a scanning-inhibiting secondary structure element [16, 333] ; however, this structural element reduced the translation by a factor of five for the TuMV leader [ 16] and expression in the CPMV case was quite low and not different from other internal initiation events on the respective reporter RNA [333] . The effect of the PVY sequence was completely abolished by an antisense oligonucleotide for the first 16 nucleotides, suggesting that the important sequence is close to the RNA 5 t end [16] . The cap independence conferred by the TEV leader depends on sequences located more than 80 nucleotides downstream of the 5 ~ end [53] . Efficient, cap-independent initiation with BYDV and STNV RNA required sequences near the 5 ~ end and near the 3 ~ end [74, 334, 353] .

The PPV, PVY and CPMV sequences were tested as internal entry sites between two ORFs. With the PPV sequence, only low levels of downstream ORF expression were observed in in vitro translation systems, which led to the conclusion that the PPV leader does not act as a true IRES [280] . Expression obtained with the PVY sequence was interpreted as proof for internal ribosome entry [218] ; however, the data are difficult to evaluate because no quantitative comparison of downstream ORF translation with translation from a monocistronic RNA was made. The CPMV sequence allowed low levels of downstream ORF translation in the reticulocyte lysate [350] , but was tested without success in an animal cell system [23] . In general, caution is required in extrapolating in vitro data to the in vivo situation. A sequence upstream of the TNV coat protein ORF was found to allow internal ribosome entry in wheat germ extract but not in tobacco protoplasts [240] . In vitro translation systems are known to accept also uncapped RNA, depending on the exact translation conditions like ionic strength and translation factor concentration. The relatively long AUG-less regions that are tested for their IRES potential would direct all initiation capacity by spurious ribosome binding to the next downstream AUG codon and may, therefore, appear more efficient than other RNA sequences. The reported cases of internal ribosome entry on plant viral RNAs seem to need more substantiation.

The leader sequences of the como-and potyviruses are not as long or as complex as those of the distantly related picomaviruses that support intemal initiation of translation. There is also no evidence so far that any of the plant viruses inhibits plant translation to create an advantage for translation of its own mRNAs. The picomaviruses inhibit cap-dependent translation by various mechanisms and in the absence of this inhibition viral mRNAs often compete only poorly with cellular RNAs [258] . Plant virus RNAs seem to compete with cellular mRNAs by reducing the requirements for or by increasing the affinity to initiation factors. The leader sequences of many plant viruses (but not all) have been found to increase translation in in vitro systems, in transfected plant protoplasts, and in transgenic plants by unknown mechanisms [53, 77, 91, 118, 119, 183, 184, 234, 255, 270, 308, 309, 311, 335, 367] (reviewed by Gallie, this issue). Translation of the uncapped STNV RNA requires elF 4F at a lower concentration than other RNAs and is not affected by the presence of cap analogues [44] , whereas translation of the capped, efficient AMV RNA 4 leader requires less eIF--4F and -4B than, for example, the/3-globin RNA, but is clearly cap-dependent [102, 153] . The enhancing effect of the TMV f~ sequence in a yeast extract is independent of the cap-binding factor eIF-4E but still requires eIF-4A (the supposed RNA-unwinding factor) [7] , although TMV RNA is naturally capped and f~ lacks strong secondary structure. The TMV f~ sequence may act differently to other translation enhancing plant viral RNA leaders since it is active in E. coli and almost all eukaryotic translation systems [119] (but not or only poorly in monocot cells [ 120] ). The activity of f~ is end-dependent in eukaryotic translation systems [197] , and two sequence motifs are crucial in plant protoplasts [126] ; for the activity in E. coli the 5 ~ end with homology to a recently defined sequence element that can functionally replace a normal Shine-Dalgarno sequence may be responsible [121, 177] . No particular sequence motifs with enhancing effect have been found in other plant viral RNA leaders. For a more detailed review of these features see the contributions of D. Gallie (this volume).

It is also likely that some cellular mRNAs have leader sequences which modulate translation due to their particular translation factor requirements. For example, the untranslated regions of the barley c~amylase specifically enhanced translation in aleurone cells [ 129] . Other candidates for such leader sequences are the mRNAs translated under stress or developmental conditions that are detrimental to the translation of most other cellular mRNAs, and where discrimination occurs at the level of translation initiation ( [127] for review).

In other eukaryotes, leader sequences and also 3'untranslated sequences can modulate translation and RNA stability in concert with mRNA-specific proteins ( [235, 316] for reviews); no example of such a mechanism has been described so far for plants.

All the examples described above demonstrate flexibility in the efficiency and location of translation initiation. Somewhat surprisingly, the 3'-untranslated region (including the poly(A) tail) alone or in conjunction with the 5 ~ leader can have a positive influence on translation initiation (reviewed by Gallie, this issue). These effects of the RNA 3' ends are probably commonplace but as they have been studied in only a few cases are treated here as exceptions. The mechanism for the enhancement is not known, but it is probably similar for the poly(A) tail and for the nonpolyadenylated 3 ~ ends of some RNA viruses [122, 125, 288] . In yeast, effects of the poly(A) tail are effective on the formation of the 80S ribosome [251, 290] and are mediated by the poly(A)-binding protein (PABP) [ 180, 250 for reviews]. The importance of the PABP was also observed in a pea in vitro translation system [304] . In plants, sequences far away from the RNA 5 ~ end in conjunction with 5 ~ leaders can cause cap-independence or at least reduce cap-dependence, as shown for STNV [334] , TMV [370] , BMV [ 191 ] and BYDV [353] ; this suggests involvement in ribosome binding to the RNA. The stimulatory effects of 3'-UTRs seem to be quite variable in different assay systems and apparently are not always reproducible (e.g. the effects of the 3'-UTR of BMV described by Gallie and coworkers [128, 217] could not be reproduced by Lahser et al. [216] ). The effects of the poly(A) tail in transfected plant protoplasts were much stronger in the presence of a 5' cap than in its absence. Exogenously added poly(A) stimulated translation of capped, non adenylated mRNAs in yeast [251] and inhibited that of uncapped RNA in yeast [251] and in plant extracts [I 30] , suggesting that poly(A) interacts with factors required for translation and that it may form a complex with the capped 5' end of the RNA, which is particularly efficient in ribosome binding (reviewed in detail by Gallie, this issue). In the presence of PABP, an oligo(A) tract in the mRNA leader is inhibitory for translation [81] . In Xenopus oocytes the poly(A) tail was found to enhance translation reinitiation [ 117] , which means that in this case it acts after the first round of translation of an ORF on ribosomes that have translated the ORF already once. It is unclear how such a recycling of ribosomes could occur, but polysomal RNAs with a circular appearance have been observed in a variety of cases [64, 92, 171, 215] , and sequence interactions between the ends of plant viral RNAs have been observed or suggested [74, 75, 103, 319, 334] . This interpretation has similarities with the ribosome shunt. It has been noted that the ribosome shunt in vivo might also serve for such a recycling since in the pararetroviral RNAs the shunt donor site is also present in the terminal redundant part at the 3' end of the RNA [ 114] . A ribosome recycling might help to increase the density of ribosomes on RNAs that can be successfully translated. In most present day translation studies, kinetics of polysome loading are not considered and only the amount of synthesized protein after the incubation period is measured. Also, in most of the studies of the effect of poly(A) tails or 3~-untranslated regions in yeast or plant systems, no discrimination between first initiation and re-initiation was made, and the kinetics of protein production with the different leader and 3'-untranslated regions still has to be investigated. Loading of mRNAs into large polysomes can be considerably slower than the transit time of a translating ribosome on such an mRNA [254] , which suggests that it is not determined by primary binding of ribosomes to an RNA, but by rebinding of ribosomes that have translated already once. Reinitiation is also differently affected by initiation inhibitors and, therefore, may be functionally different from first initiation events [254] . While the effect of a cap structure on translation efficiency in transfected plant protoplasts was evident at the earliest measured time point, the full effect of the poly(A) tail was apparent only later [ 120] . Reinitiation may also be less or not cap-dependent, which may be one explanation why in some cases the 3t-UTR together with some 5' sequences conferred cap independence (see above). If reinitiation is partially involved in the mechanisms of translation enhancement by 3' regions, it is to be expected that these effects will differ with different reporter genes and possibly also with the precise linkage of the reporter ORF to the 3~-untranslated region, since in this case the mode of translation termination and subsequent ribosome migration should be important.

In other eukaryotes, 3'-UTRs of specific mRNAs have been found also to contain a number of other signals involved in RNA transport and localization, specific translation repression, and translation-dependent or independent RNA degradation ( [79] for review). Strong differences in gene expression levels in plants have been described for reporter constructs with different 3'-untranslated regions [176] , but the cause of the effect (translation, 3~-end processing, RNA stability, transport or localization) has not been elucidated.

The translation process is in general precise and competing events such as misreading by 'wrong' tRNAs or shifts in the reading phase occur with a significant frequency only if decoding of the codon at the A site by a cognate tRNA or a termination factor is slow or disturbed [11] . In bacteria, translating ribosomes have been found to even skip longer stretches of the RNA completely [25, 172, 357] . This extreme is so far unknown in eukaryotes but smaller 'programmed errors' occur during translation elongation. The paradigm for such processes in eukaryotes is frameshifting and stop codon suppression in retroviruses and retrotransposons, which use both mechanisms to express their pol gene as a fusion protein with the upstream capsid proteins in a stoichiometrically controlled ratio ( [156, 286] for reviews). A similar mechanism is also used by other viruses of mammals [40, 182] and yeast [341 ] , but not by the plant pararetroviruses, which are relatives of the animal pararetroviruses [ 167, 286] . In these viruses, thepol ORF is either translated separately from the upstream, overlapping ORF corresponding to the retroviral gag-ORF [297] or is part of a long precursor ORF that also contains the gag functions [286] . Instead, frameshift events have been detected or proposed for pol gene expression of luteo- [85] , enamo- [82] , diantho- [195, 287] , sobemo- [230] and carlaviruses [141] . Frameshift may also occur in the BWYV ST9-associated RNA [61] and in closteroviruses [4, 190] . In all cases except the last, frameshift occurs leftwards to the -1 reading phase ( Table 3) .

The frameshift sites of BYDV-PAV [39, 85, 132] , two strains of PLRV [214, 274] , RCNMV [195] and CfMV [230] have been studied in greater detail. Like the frameshift signal of retroviruses and many retrotransposons they consist of a 'shifty' heptanucleotide sequence of the type X.XXY.YYZ (arranged in codons of the 0 frame) and a close downstream secondary structure element ( Table 3 ). The sequences of a number of retrovirai transframe-proteins have been determined and on this basis it has been suggested that the frameshift occurs when the peptidyl tRNA at the XXY codon at the P site and the aminoacyl RNA at the YYZ codon at the A site simultaneously slip backwards one nucleotide to the XXX. and YYY codons, respectively [178] . Frameshifting does not necessarily lead to a unique transframe protein but rather to a small number of variants with slightly different amino acid sequences around the frameshift site [179, 356] . Frameshifting requires that the two tRNAs at the A and the P sites can stably interact with the new codons with mismatches only at the wobble position. It is caused or enhanced by ribosome pausing due to the downstream structural element [60, 313, 329, 330, 336] and can be influenced by the translation frequency [169] . The downstream element can be a simple hairpin structure or an elaborate pseudoknot [56, 301, 329, 330] . The nature of the codons (or their cognate tRNAs) is also important: of all possible YYZ codons, so far only AAC, UUU, UUA and recently AAU have been found [156, 243, 329] .

That most of the studied frameshift events in plants follow the same mechanism was elucidated by determination of the sequence of the BYDV transframe protein [85] and by mutation analyses of the shifty heptanucleotides and the supposed downstream pause elements [132, 195, 214, 274, 363] . Frameshift sites were analyzed either in their natural contexts or by inserting the sites into a reporter ORF. Depending on the frameshift signal and the assay system, frameshift efficiencies of 1-30% were observed. High efficiencies were obtained with the Polish strain of PLRV and with CfMV in in vitro translation systems [214, 230] , while the other signals direct frameshifting at only 1-4% efficiency in different in vitro systems or in plant protoplasts. Signals consisting of A and U nucleotides tend to produce higher efficiencies than those containing G or C [141, 274] . The BWYV ST9-associated RNA may contain a particularly efficient ffameshift site, which has, however, not yet been identified [61] .

A stop codon found immediately downstream of many of the shifty sites has a positive effect in some assay systems, probably because it contributes to ribosome pausing [39, 85] . In wheat germ extract, the BYDV frameshift event is greatly stimulated by a Table 3 . Frameshift. Sequences at proven or suggested sites in plant viral mRNAs. Virus acronyms and the family are given. The frameshift site is arranged in the 0 reading frame. For the -1 frameshift sites, the given sequence starts with the first nucleotide of the shifty site (X.XXY.YYN) and ends at the beginning of the downstream pause signal. Data are compiled from the indicated references; if no reference is given, data are from the compilation by Miller et al. [243] . 190 sequence located more than 3 kb downstream of the site [353] , and sequences 40-100 nucleotides upstream of the site may also be required [246] . The frameshift efficiency of a given signal is not only determined by the cis-acting sequences but also by the available tRNAs. In infected Nicotiana benthamiana plants, shift-promoting tRNA (classes) exist that decode UUU, UUA, UUC and the asparagine codon AAU but none that decode the asparagine codon AAC [1951.

The supposed frameshift signal of PVM differs from those described above. The shifty sequence U.AGA.AAA.UGA allows only one basepair to be formed after the shift between RNA and tRNA at the P site, which is hardly sufficient to drive the shift. It was, therefore, suggested that in this case the shift occurs after peptidyl transfer and transtocation when the last codon of the ORF (AAA) together with the peptidyl tRNA occupies the P site and the ribosome awaits recognition of the stop codon at the A site by the release factor [ 141 ] . This model was supported by mutagenesis analyses of the role of the stop codon and of the preceding A stretch. The frameshift signal was less effective than those described above (about 0.3% in the reticulocyte lysate) and the postulated frameshift product was not detectable in infected plants. It is noteworthy that the downstream ORF, which should be linked by the frameshift event to the upstream coat protein ORE has its own translation start codon and is also translated separately (at least in vitro) by an unknown mechanism [141] .

A number of yeast retrotransposons use a +1 frameshift to produce the correspondent of a gag-pol fusion protein [22] . In the case of Ty3, the sequence GCG.AGU.U is decoded as Ala-Val by reading the underlined codons. No slippage of the tRNA hla is possible but the frameshift efficiency depends on the presence of the 'rare' AGU codon and certain features of the tRNA that decodes the GCG codon or (more general) other special codons located upstream of a rare codon. Artificial frameshift sites can be constructed by combining codons that are decoded by such 'shifty' tRNAs with rare codons [261, 351 ] . A similar codon configuration may be present in a suggested + 1 frameshift site between closterovirus ORFs 1 a and lb, which contains a rare CGG codon (CTV [190] ) or a stop codon (BYV [4] ) and, at least for CTV, lacks obvious secondary structure [190] . Frameshifting has still to be proven for these viruses.

The mammalian type C retroviruses use stop codon suppression instead of frameshifting to produce the gag-pol fusion protein [97, 156, 365] . Suppression occurs with an efficiency of about 5% in vitro and in vivo and is accomplished by misreading of an UAG termination codon by a glutaminyl tRNA [97] . A UGA stop codon at the same position is decoded as arginine, cysteine or tryptophan [97] . Since the stop codons . of normal cellular genes are not suppressed detectably under the same conditions, it was concluded that a specific context is responsible for the effect in retroviral RNAs. The presence of a pseudoknot structure 8 nucleotides downstream was required for efficient readthrough in vitro [98, 360, 361] , while in vivo only evidence for the requirement of a stem structure was obtained [96] .

Stop codon suppression has been suggested as a mechanism for the expression of some seed storage proteins whose ORFs are interrupted by stop codons [93, 223, 352] . It is, however, not always clear whether these ORFs are on functional mRNAs or whether they represent silent pseudogenes. It has been pointed out that storage protein genes contain a high number of CAA and CAG codons (coding for glutamine) which could mutate into a stop codon by a simple Cto-T transition [93] . A stop codon in the hordein gene A-horl-14 was suppressed in bombarded barley endosperm with an efficiency of only 0.6% [93] .

A large number of plant RNA viruses use stop codon suppression to produce components of RNA dependent RNA-polymerases or elongated coat proteins that are probably required for transmissibility by their respective vectors [243, 326, 366] . Unlike for retroviruses, no structure requirements for stopcodon suppression have been observed in plant systems. Rather readthrough is dependent on the immediate sequence context of the stop codon and on the presence of certain tRNA species that can decode such stop codons as sense codons (Table 4 ). Suppressible stop codons have been inserted into GUS ORFs and readthrough has been quantified in transfected plant protoplasts. The context of the stop codon of the tobacco mosaic virus 126 kDa ORF (CAA.UAG.CAA.UUA) allowed particularly efficient (around 5%) readthrough for all three possible stop codons [306, 307, 346] . The leakiness of this stop codon context was used to express angiotensin-Iconverting enzyme inhibitor peptide as the C-terminal extension of a subfraction of TMV coat proteins from an accordingly engineered infectious TMV clone [148] . Mutagenesis analysis of the readthrough sequence defined (C/A)(A/C)A.UAG.CAR.YYA (with R=purine, Y--pyrimidine) as optimal consensus. This consensus sequence, like the UGA readthrough signals of TRV [150] , PEBV [229] , PCV and SBW-MV [163] and of sindbis virus RNA [219] , fits to the statistical analysis of eukaryotic stop codon contexts: efficiently recognized stop codons normally avoid a C directly downstream but rather have a purine in this position [9, 42, 43, 236, 344] . In the light of this finding, the readthrough of the UAG.G signals of the carmo-and luteoviruses, and of the tombusviruses [ 159, 296] (Table 4) has to be functionally different and may require additional cis-acting sequences or specific tRNAs. The presence of a conserved CCCCA motif or repeated CCXXXX motifs downstream of the luteoand carmovirus readthrough sites has been noted but the involvement in readthrough has not been experimentally tested so far [243] . The signals of TRV, Car-MV, MCDV, BYDV and BWYV were active in in vitro systems or in infected plants [12, 52, 100, 265] but were rather ineffective in transfected tobacco protoplasts [306] .

The sequences of plant viral readthrough proteins produced in vivo have not been determined, but it is likely that the stop codons are decoded by particular tRNAs, as has been found in in vitro systems and Xenopus oocytes, where addition of tRNA Tyr with a Gff~A anticodon but not with a GUA anticodon greatly stimulated readthrough of the TMV signal [15, 20, 21, 368] . Interestingly, this tRNA is rare in young wheat leaves but is abundant in older tissue, suggesting that processes like readthrough (and possibly also frameshifting) are developmentally regulated. The Table 4 . Leaky termination codons of plant virus ORFs. The immediate sequence context of partially suppressed termination codons (underlined) and the protein resulting from translational suppression are listed. References to those viral sequences not further mentioned in this review can be found in the compilation by Miller et al. [243] .

Read-through protein References TMV readthrough site was also efficiently suppressed (20%) in transfected tobacco protoplasts by a bean tRNA leu with an artificially introduced CUA anticodon [51] and in protoplasts and transgenic plants (up to 10%) by modified trp tRNAs [104] . The fact that such suppressor tRNAs can be overexpressed without strong phenotypic changes suggests that normal stop codons are not significantly suppressed by these tRNAs. It has, however, been noted that suppression is more efficient in protoplasts than in plants, suggesting that only transgenic plants are regenerated which have a lower expression level of the suppressor tRNA [51 ] .

The leaky UGA stop codon of TRV is suppressed by a tRNA xrp with a CmCA anticodon [369]. One such tRNA originated from the chloroplast and was more efficient than that from the cytoplasm. Since the respective chloroplast and mitochondrial tRNAs are almost identical and TRV appears to be associated with mitochondria in infected cells it is possible that the virus uses the mitochondrial tRNA as suppressor in vivo [369] . At present it is unknown which tRNA suppresses the UAG.G stop codon of the luteoviruses.

Different suppressing tRNAs could be the reason that readthrough efficiencies vary with the assay system, but possible involvement of cis-active sequences distal to the readthrough sites may also account for the variation. For example, in in vitro translation experiments with the full-length TMV RNA, fourfold higher readthrough efficiencies have been obtained [264] than with the artificial constructs of Skuzeski et al. [306, 307] , and it has to be determined whether this is an effect of more distal cis-acting sequences or of the different assay systems.

In other organisms, reading of UGA stop codons as selenocysteinyl codons is a special type of stop codon suppression ( [95, 157] for reviews). Selenocysteine has been found in plant proteins [41 ] and a selenocysteine-tRNA recognizing UGA has been identified in Beta vulgaris [157] . However, it is unknown whether UGA misreading is the only way of selenocysteine incorporation into plant proteins and whether this process requires additional cis-active sequences on the mRNA, as was found in animal cells [30, 302] and E. coli [ 160] .

For retroviruses, it is still a matter of discussion whether virus infection alters the frameshift or stop codon suppression capacity of the cell. Controversial results on induction of synthesis of shifty tRNAs or on differences in the levels of tRNA modifications with consequences for the frameshift efficiency have been presented ( [ 156] for review). For plants, however, nothing comparable is known for tRNAs involved in stop codon suppression, although concentration differences in different tissues have been reported [21, 285] , and the findings of Kim and Lommel [195] may indicate that the frameshift-prone tRNAs may also be represented differently in different plants.

Many responses of plants to developmental and environmental signals or to stresses like wounding, heat shock, lack of water or oxygen have posttranscriptional components (for reviews see Gallie, this issue; [127, 143, 320] ). Generally the translation machinery concentrates on a number of specific mRNAs while others are not translated anymore and are either stored as inactive RNAs or degraded. The pattern of polysome-associated mRNAs and overall translational activity can vary widely, but little is known about the fate of specific mRNAs and about the controlling cisand trans-acting features. Translational control is exerted mainly at the levels of initiation [29, 277] and elongation [10, 28, 99, 248, 305, 310, 348, 354] . Heat-shock mRNAs are more competitive under heat stress and this feature has been attributed to their leader sequences [ 131, 269, 293] . These sequences are, however, very heterogeneous and no common motifs have been discerned so far [187] .

The association of mRNAs with architectural components of the cell or the precise location of an mRNA in the cytoplasm can influence the translational activity on that RNA [67, 149] . An alteration of such association or sequestration seems to be one level of translational control under conditions of stress or changes in developmental programs in plant cells [10, 13, 24, 70, 256, 277] , but the molecular basis of this phenomenon is unclear.

A number of features that modulate translation have been detected in mRNAs translated in animal and yeast cells. Some of these are also found in RNAs translated in the cytoplasm of plant cells, predominantly on viral mRNAs. Quantitative control of the synthesis of two (or more) proteins from one mRNA is achieved by leaky scanning, frameshifting or stop codon suppression, or transactivated reinitiation on plant viral mRNAs. With the possible exception of stop codon suppression, it is not known whether any of these mechanisms is active also in the control of nuclear gene expression or is so plant, tissue, or development specific that it could be part of the host-range control of plant viruses. The 5 t leader sequences and the 3'-untranslated region separately or in conjunction increase the cap-dependent or -independent affinity for ribosomes and probably can serve as a paradigm for cellular RNAs that are translated under special conditions. Short ORFs are found in the leaders of a number of nuclear and viral genes, but it is unknown whether these sORFs simply serve to reduce downstream translation or whether they regulate translation in response to cellular conditions. Further work is required to link these features to the translation control events observed in plant cells in response to a variety of stresses and developmental or environmental signals.

In particular, it is to be expected that further study of the influence of 5' leaders and 3t-UTRs on (capindependent) translation, of the translation factors involved in this process in plants, and further analysis of the unusual translation mechanisms of the plant pararetroviruses will contribute essential knowledge to our understanding of eukaryotic translation processes.

Suppression of ribosomal reinitiation at upstream open reading frames in amino-acid-starved cells forms the basis for GCN4 translational control

Subcellular fate of the Int-2 oncoprotein is determined by choice of initiation codon

The 5 t-untranslated region of picornaviral genomes

Beet yellows closterovirus: complete genome structure and identification of a leader papainlike thiol protease

Two-step binding of eukaryotic ribosomes to brome mosaic virus RNA 3

Development of a heat shock inducible expression cassette for plants: characterization of parameters for its use in transient expression assays

The 5 ~-leader sequence of tobacco mosaic virus RNA mediates initiation-factor-4E-independent, but still initiation-factor-4A-dependent translation.in yeast extracts

Expression of dicistronic transcription units in transgenic tobacco

Analysis of the stop codon context in plant nuclear genes

Heat shock gene expression is controlled primarily at the translational level in carrot cells and somatic embryos

Towards a genetic dissection of the basis of triplet decoding, and its natural subversion: Programmed reading frame shifts and hops

Expression of the genome of potato leafroll virus: readthrough of the coat protein termination condon in vivo

Hypoxic stress-induced changes in ribosomes of maize seedling roots

mRNA structures influencing translation in the yeast Saccharomyces cerevisiae

Plant tRNA suppressors: in vivo readthrough properties and nucleotide sequence of yellow lupin seeds tRNA Tyr

Evidence for an internal ribosome entry site within the 5 ~-untranslated region of turnip mosaic potyvirus RNA

Cauliflower mosaic virus 35S RNA leader region inhibits translation of downstream genes

Cauliflower mosaic virus gene VI produces a symptomatic phenotype in transgenic tobacco plants

Direct mapping of adeno-associated virus capsid protein B and C: a possible AUC initiation codon

UAG readthrough during TMV RNA translation: Isolation and sequence of two tRNAs Tyr with suppressor activity from tobacco plants

The molecular basis for the differential translation of TMV RNA in tobacco protoplasts and wheat germ extracts

Ribosomal frameshifling in the yeast retrotransposon Ty: tRNA slippage on a 7 nucleotide minimal site

The mechanism of translation of cowpea mosaic virus middle component RNA: no evidence for internal initiation from experiments in an animal cell transient expression assay

Synthesis of ribosomal proteins from stored mRNAs early in seed germination

Frameshifling of the E. coli trpR gene occurs by the bypassing of a segment of its coding sequence

Browning KS: mRNA binding properties of wheat germ protein synthesis initiation factor 2

An internal ribosome entry mechanism promotes translation of murine leukemia virus gag polyprotein precursors

mRNAs encoding ribulose-1,5-bisphosphate carboxylase remain bound to polysomes but are not translated in amaranth seedlings transferred to darkness

Light-mediated control of translation initiation of ribulose-l,5-bisphosphate carboxylase in amaranth cotyledons

Recognition of UGA as a selenocysteine codon in type I deiodinase requires sequences in the 3 t untranslated region

Binary Agrobacterium vectors for plant transformation

Positions +5 and +6 can be major determinants of the efficiency of non-AUG initiation codons for protein synthesis

The efficiency of translation termination is determined by a synergistic interplay between upstream and downstream sequences in Saccharomyces cerevisiae

Posttranscriptional transactivation in cauliflower mosaic virus

Nucleotide sequence of beet necrotic yellow vein virus RNA-2

Selection of CUG and AUG initiator codons for Drosophila E74A translation depends on downstream sequences

Evidence for the translational control of storage protein gene expression in oat seeds

coding density of the turnip yellow mosaic virus genome: roles of the overlapping coat protein and p206-readthrough coding regions

Translational frameshifting mediated by a viral sequence in plant cells

Mutational analysis of the RNA pseudoknot component of a coronavirus ribosomal frameshifting signal

Identification of selenocysteine in the proteins of selenate-grown Vigna radiata

Sequence analysis suggests that tetra-nucleotides signal the termination of protein synthesis in eukaryotes

The translational termination signal database

Evidence that the requirements for ATP and wheat germ initiation factors 4A and 4F are affected by a region of satellite tobacco necrosis virus RNA that is 3 ~ to the ribosomal binding site

Codon context and protein synthesisenhancements of the genetic code

Alternative initiation of translation determines cytoplasmic or nuclear localization of basic fibroblast growth factor

Coevolution of codon usage and transfer RNA abundance

Expression of mRNA electroporated in plant and animal cells

Codon usage in higher plants, green algae, and cyanobacteria

Translational inhibition by a human cytomegalovirus upstream open reading frame despite inefficient utilization of its AUG codon

Transfer RNA-mediated suppression of stop codons in protoplasts and transgenic plants

Characterization of cell-free translation products of carnation mottle virus genomic and subgenomic RNAs

Cap-independent enhancement of translation by a plant potyvirus 5 r-untranslated region

Translation of equine infectious anemia virus bicistronic tat-rev mRNA requires leaky ribosome scanning of the tat CTG initiation codon

Eukaryotic start and stop translation sites

An RNA pseudoknot and an optimal heptameric shift site are required for highly efficient ribosomal frameshifting on a retroviral messenger RNA

Initiation of protein synthesis by the eukaryotic translational apparatus on circular RNAs

Rice tungro bacilliform virus: transcription and translation in protoplasts

Role of the AGA/AGG codons, the rarest codons in global gene expression in Escherichia coli

Structural and functional studies of retroviral RNA pseudoknots involved in ribosomal frameshifting: nucleotides at the junction of the two stems are important for efficient ribosomal frameshifting

The beet western yellows virus ST9-associated RNA shares structural and nucleotide sequence homology with carmo-like viruses

A difference in the rate of ribosomal elongation balances the synthesis of eukaryotic translation initiation factor (elF)-2 and elF-2/3

Expression pattern of bacterial polycistronic genes in tobacco cells

Circular polysomes predominate on the rough endoplasmic reticulum of somatropes and mammotropes in the rat anterior pituitary

Efficiency of translation initiation by non-AUG codons in Saccharomyces cerevisiae

Codon usage in Escherichia coli may modulate translation initiation

Elongation factor 1-alpha, translation and the cytoskeleton

The complete nucleotide sequence of tobacco necrosis virus strain D

Recent advances in peptide chain termination

Stress-induced translational control in potato tubers may be mediated by polysome associated proteins

The signal sequence moves through a ribosomal tunnel into a rtoncytoplasmic aqueous environment at the ER membrane early in translocation

Scanning independent ribosomal initiation of the sendal virus X protein

An upstream open reading frame represses expression of Lc, a member of the R/B family of maize transcriptional activators

The 3 r untranslated region of satellite tobacco necrosis virus RNA stimulates translation in vitro

Sequence of the 3' untranslated region of brome mosaic virus coat protein messenger RNA

Selection of the 5 ~-proximal translation initiation site is influenced by mRNA and eIF-2 concentrations

Improved high-level constitutive foreign gene expression in plants using an AMV RNA4 untranslated leader sequence

Biased codon usage

Diversity of cytoplasmic functions for the 3' untranslated region of eukaryotic transcripts

Translational inhibition mediated by a short upstream open reading frame in the human cytomegalovirus gpULA (gp48) transcript

Autoregulation of poly(A)-binding protein synthesis in vitro

The nucleotide sequence and luteovirus-like nature ofRNA 1 of an aphid non-transmissable strain of pea enation mosaic virus

Molecular dissection of the cauliflower mosaic virus translation transactivator

The AAPT1 gene of soybean complements a cholinephosphotransferasedeficient mutant of yeast

Translational frameshifting by barley yellow dwarf virus RNA (PAV serotype) in Escherichia coli and in eukaryotic cell-free extracts

Precise mapping and in vitro translation of a trifunctional subgenomic RNA of barley yellow dwarf virus

Control of start codon choice on a plant viral RNA encoding overlapping genes

Initiation of translation of the cauliflower mosaic virus genome from a polycistronic mRNA: evidence from deletion mutagenesis

Castilho Valavicius B: Mutations at a Zn(II) finger motif in the yeast elF-2fl gene alter ribosomal start-site selection during the scanning process

Mutations in GCDI 1, the structural gene for elF2-gamma in yeast, alter translational regulation of GCN4 and the selection of the start site for protein synthesis

Plant viral leaders influence expression of a reporter gene in tobacco

Structure of globin mRNA and mRNA-protein particles: use of dark-field electron microscopy

Amber codon suppression: the in vivo and in vitro analysis of two C-hordein genes from barley

Translation ofbicistronic viral mRNA in transfected cells: Regulation at the level of elongation

Alternative readings of the genetic code

Mutational analysis of the gag-pol junction of Moloney murine leukemia virus: Requirements for expression of the gag-pol fusion protein

Identification of amino acids inserted during suppression of UAA and UGA termination codons at the gag-pol junction of Moloney murine leukemia virus

Bipartite signal for read-through suppression in murine leukemia virus mRNA: an eight-nucleotide purine-rich sequence immediately downstream of the gag termination codon followed by an RNA pseudoknot

Post-translational regulation of gene expression in oxygen-deprived roots of maize

In vivo expression and mutational analysis of the barley yellow dwarf virus readthrough gene

Binding of ribosomes to 5 rterminal leader sequences of eukaryotic messenger RNAs

The absence of a m7G cap on ~-globin mRNA and alfalfa mosaic virus RNA 4 increases the amounts of initiation factor 4F required for translation

Possible functional role of viral tRNA-like structures

Construction and expression of nonsense suppressor tRNAs which function in plant cells

Bacterial genes inserted in an engineered RNA virus. Efficient expression in monocotyledonous plant cells

A highly conserved eukaryotic protein family possessing properties of polypeptide chain release factor

Insect resistant rice generated by introduction of a modified 6-endotoxin gene of Bacillus thuringiensis

The leading sequence of caulimovirus large RNA can be folded into a large stem-lop structure

Differential inhibition of downstream gene expression by the cauliflower mosaic virus 35S RNA leader

Expression from polycistronic cauliflower mosaic virus pregenomic RNA. In: Posttranscriptional Control of Gene Expression

Positive and negative control of translation by the leader sequence of cauliflower mosaic virus pregenomic 35S RNA

Translation of a polycistronic mRNA in the presence of the cauliflower mosaic virus transactivator protein

Role of an upstream open reading frame in the translation of polycistronic mRNAs in plant cells

Non-linear ribosome migration on cauliflower mosaic virus 35S RNA

Splicing in a plant pararetrovirus

Position dependent ATT initiation during plant pararetrovirus rice tungro bacilliform virus translation

Role of the 3 ~-poly(A) sequence in translational regulation of mRNAs in Xenopus laevis oocytes

The 5 ~-leader sequence of tobacco mosaic virus RNA enhances the expression of foreign gene transcripts in vitro and in vivo

A comparison of eukaryotic viral 5 ~-leader sequences as enhancers of mRNA expression in vivo

Visualizing mRNA expression in plant protoplasts: factors influencing efficient mRNA uptake and translation

A translational enhancer derived from tobacco mosaic virus is functionally equivalent to a Shine-Dalgarno sequence

RNA pseudoknot domain of tobacco mosaic virus can functionally substitute for a poly(A) tail in plant and animal cells

The cap and poly(A) tail function synergistically to regulate mRNA translational efficiency

Posttranscriptional regulation in higher eukaryotes: The role the reporter gene in controlling expression

Functional analysis of the tobacco mosaic virus tRNA-like structure in cytoplasmic gene regulation

Identification of the motifs within the tobacco mosaic virus 5t-leader responsible for enhancing translation

Posttranscriptional regulation of gene expression in plants

The role of the 3P-untranslated region of non-polyadenylated plant viral RNAs in regulating translational efficiency

The regulation of gene expression in transformed maize aleurone and endosperm protoplastsanalysis of promoter activity, intron enhancement, and mRNA untranslated regions on expression

Poly(A) binds to initiation factors and increases cap-dependent translation in vitro

Heat shock disrupts cap and poly(A) tail function during translation and increases mRNA stability of introduced reporter mRNA

Differential response to frameshift signals in eukaryotic and prokaryotic translational systems

Initiation codons within 5/-leaders of mRNAs as regulators of translation

Cap accessibility correlates with the initiation efficiency of alfalfa mosaic virus RNAs

Consecutive low-usage leucine codons block translation only when near the 5 t end of a message in Escherichia coli

The AU-rich sequences in the introns of plant nuclear pre-mRNAs are required for splicing

ln vitro expression of cauliflower mosaic virus genes

Efficient initiation of translation at non-AUG triplets in plant cells

Gene VI offigwort mosaic virus (caulimovirus group) functions in posttranscriptional expression of genes on the full-length RNA transcript

Requirement of gene VII in cis for the expression of downstream genes on the major transcript of figwort mosaic virus

The nucleic acid-binding zinc finger protein of potato virus M is translated by internal initiation as well as by ribosomal frameshifting involving a shifty stop codon and a novel mechanism of P-site slippage

Effect of sequence context at the stop codons on efficiency of reinitiation in GCN4 translational control

Control of mRNA stability in higher plants

The immediate downstream codon strongly influences the efficiency of utilization of eukaryotic translation initiation codons

Antipeptidyl transferase leader peptides of attenuation-regulated chloramphenicol-resistance genes

Effect of two consensus sequences preceding the translation initiator codon on gene expression in plant protoplasts

The computer simulation of RNA folding pathways using a genetic algorithm

A new tobacco mosaic virus vector and its use for the systemic production of angiotensin-I-converting enzyme inhibitor in transgenic tobacco and tomato

Polyribosome targeting to microtubules -enrichment of specific mRNAs in a reconstituted microtubule preparation from sea urchin embryos

The complete nucleotide sequence of tobacco rattle virus RNA-I

Translational activation of the non-AUG-initiated c-myc 1 protein at high cell densities due to methionine deprivation

Regulation and function of non-AUG-initiated proto-oncogenes

mRNAs containing the unstructured 5 r leader sequence of alfalfa mosaic virus RNA 4 translate inefficiently in lysates from poliovirns-infected Hela cells

Peptide inhibitors of peptidyltransferase alter the conformation of domains IV and V of large subunit rRNA: a model for nascent peptide control of translation

Selenocysteyl-tRNAs recognize UGA in Beta vulgaris, a higher plant, and in Gliocladium virens, a filamentous fungus

Translational suppression in retroviral gene expression

UGA: a split personality in the universal genetic code

An analysis of the sequence of an infectious clone of rice tungro bacilliform virus, a plant pararetrovirus

The complete genome structure and synthesis of infectious RNA from clones of tomato bushy stunt virus

Coding from a distance: dissection of the mRNA determinants required for the incorpora-183

tion of selenocysteine into protein

Translation controls the expression level of a chimaeric reporter gene

Translational control in mammalian cells

Complete nucleotide sequence of peanut clump virus RNA 1 and relationships with other fungus-transmitted rod-shaped viruses

Translation of the second gene of peanut clump virus RNA 2 occurs by leaky scanning in vitro

Translational control of GCN4 -an in vivo barometer of initiation-factor activity

Codon replacement in the PGK 1 gene of Saccharornyces cerevisiae: Experimental approach to study the role of biased codon usage in gene expression

Pararetroviruses and retroviruses: a comparison of expression strategies

Transcriptional and translational control of gene expression in cauliflower mosaic virus

Translation frequency of the human T-cell leukemia virus (HTLV-2) gag gene modulates the frequency of ribosomal frameshifting

Multiple elements in the 5 ~ -untranslated region downregulate c-sis messenger RNA translation

Electron microscopic evidence for the circular form of RNA in the cytoplasm of eukaryotic cells

A persistent untranslated sequence within bacteriophage T4 DNA topoisomerase gene 60

Messenger RNA for the coat protein of tobacco mosaic virus

Expression of a downstream gene from a bicistronic transcription unit in transgenic tobacco plants

Cap-dependent and cap-independent translation by internal initiation of mRNAs in cell extracts prepared from Saccharomyces cerevisiae

Different 3 ~ end regions strongly influence the level of gene expression inplant cells

A second putative mRNA binding site on the Escherichia coli ribosome

Signals for ribosomal frameshifting in the Rous sarcoma virus gag-pol region

Varmus HE: Characterization of ribosomal frameshifting in HIV-1 gag-pol expression

Do the poly(A) tail and 3 / untranslated region control mRNA translation?

Cap-independent translation of picornavirus RNAs: structure and function of the internal ribosome entry site

RNA sequence of astrovirus: distinctive genomic organization and putative retrovirus-like ribosomal frameshifting signal that directs viral replicase synthesis

Enhanced translation of chimeric messenger RNAs containing a plant viral untranslated leader sequence

In vitro transcription and translation efficiency of chimeric SP6 messenger RNAs devoid of 5 / vector nucleotides

Targetspecific arrest of mRNA translation by antisense 2/-0 -alkyloligoribonucleotides

An inspection of the domain between putative TATA box and translation start site in 79 plant genes

5 / untranslated leader sequences of eukaryotic mRNAs encoding heat shock induced proteins

Specific binding of eukaryotic initiation factor 2 to stallite tobacco necrosis virus RNA at a 5r-terminal sequence comprising the ribosome binding site

Internal initiation of mRNA translation in eukaryotes

Complete sequence of the citrus tristeza virus RNA genome

The 3/-untranslated region of brome mosaic virus RNA does not enhance translation of capped mRNAs in vitro

A modified/~-glucuronidase gene: Sensitive detection of plant promoter activities in suspension-cultured cells of tobacco and rice

Ribosomes pause at specific sites during synthesis of membrane-bound chloroplast reaction center protein D1

Localization of in vivo ribosome pause sites

Identification and analysis of the site of -1 frameshifting in red clover necrotic mosaic virus

Splicing of cauliflower mosaic virus is essential for viral infectivity

Binding of ribosomes to linear and circular forms of the 5/-terminal leader fragment of tobacco mosaic virus

Point mutations define a sequence flanking the AUG initiator codon that modulates translation by eukaryotic ribosomes

Influences of mRNA secondary structure on initiation by eukaryotic ribosomes

Effects of intercistronic length on the efficiency of reinitiation by eukaryotic ribosomes

Leader length and secondary structure modulate mRNA function under conditions of stress

The scanning model for translation: an update

Context effects and inefficient initiation at non-AUG codons in eukaryotic cell free translation systems

Circumstances and mechanisms of inhibition of translation by secondary structure in eukaryotic mRNAs

Downstream secondary structure facilitates recognition of initiator codons by eukaryotic ribosomes

Effects of long 5 / leader sequences on initiation by eukaryotic ribosomes in vitro

Structural features in eukaryotic mRNAs that modulate the initiation of translation

A consideration of alternative models for the initiation of translation in eukaryotes

Regulation of translation in eukaryotic systems

Determinants of translational fidelity and efficiency in vertebrate mRNAs

Adherence to the first-AUG rule when a second AUG codon follows closely upon the first

Field performance of elite transgenic maize plants expressing an insecticidal protein derived from Bacillus thuringiensis

Elongation and folding of nascent ricin chains as peptidyl-tRNA on ribosomes -the effect of amino acid deletions on these processes

Structural requirements for efficient translational frameshifting in the synthesis of the putative viral RNA-dependent RNA polymerase of potato leafroll virus

Electron microscopic evidence of circular molecules of 9-S globin mRNA from rabbit reticulocytes

Contributions of the brome mosaic virus RNA-3 3/-nontranslated region to replication and translation

A phylogenetically conserved sequence within viral 3 t untranslated RNA pseudoknots regulates translation

The 5 t untranslated region of PVY RNA, even located in internal position, enables initiation of translation

The signal for translational readthrough of a UGA codon in sindbis virus RNA involves a single cytidine residue immediately downstream of the termination codon

Effects of the first intron of rice waxy gene on the expression of foreign genes in rice and tobacco protoplasts

Translation inhibition by an mRNA coding region secondary structure is determined by its proximity to the AUG initiation codon

Analysis of action of the wobble adenine on codon reading within the ribosome

Transcriptional characterization of an c~-zein gene cluster in maize

Relative importance of 7-methylguanosine in ribosome binding and translation of VSV mRNA in wheat germ and reticulocyte cell-free systems

Translation of the mRNA of the maize transcriptional activator opaque-2 is inhibited by upstream open reading frames present in the leader sequence

Nascent peptide regulation of translation

Selection of AUG codons differs in plants and animals

Efficiency of reinitiation of translation on human immunodeficiency virus type 1 mRNAs is determined by the length of the upstream open reading frame and by the intercistronic distance

Pea early browning virus RNA1 encodes four polypeptides including a putative zinc-finger protein

The putative replicase of the cocksfoot mottle sobemovirus is translated as a part of the polyprotein by -1 ribosomal frameshift

Partial resistance of nascent polypeptide chains to proteolytic digestion due to ribosomal shielding

When cells stop making sense: Effects of nonsense codons on RNA metabolism in vertebrate cells

Regulation of chloroplast gene expression

Enhancement of translational activity mediated by potyviral 5 t-untranslated sequence in vivo but not in vitro

Cytoplasmic mRNA-protein interactions in eukaryotic gene expression

Translational termination efficiency in mammals is influenced by the base following the stop codon

Initiation of translation at CUG, GUG and ACG codons in mammalian cells

Mechanism and regulation ofeukaryotic protein synthesis

Eukaryotic protein synthesis -an in vitro analysis

Subgenomic RNAs mediate expression of cistrons located internally on the genomic RNA of tobacco necrosis virus strain A

A plant plasma membrane proton-ATPase gene is regulated by development and environment and shows signs of a translational regulation

Sequences that surround the stop codons of upstream open reading frames in GCN4 mRNA determine their distinct functions in translational control

Luteovirus gene expression

Nucleotide sequence and translation of satellite tobacco mosaic virus RNA

The leader peptides of attenuation-regulated chloramphenicol resistance genes inhibit translation termination

Genes and cisacting sequences involved in replication of barley yellow dwarf virus-PAV RNA

Overlapping open reading frames revealed by complete nucleotide sequencing of turnip yellow mosaic virus genomic RNA

Biphasic stimulation of translational activity correlates with induction of translation elongation factor 1 subunit alpha upon wounding in potato tubers

The second to last amino acid in the nascent peptide as a codon context determinant

Tales of poly(A): a review

mRNA poly(A) tail, a 3 ~ enhancer of translation initiation

Codon usage in plant genes

Analysis of unstable RNA transcripts of insecticidal crystal protein genes of Bacillus thuringiensis in transgenic plants and electroporated protoplasts

Regulation of mRNA entry into polysomes: parameters affect polysome size and the fraction of mRNA in polysomes

The 5 ~ untranslated region of pea seedborne mosaic potyvirus RNA as a translational enhancer in pea and tobacco protoplasts

Cytoplasmic heat shock granules are formed from precursor particles and are associated with a specific set of mRNAs

Homeotic gene antennapedia mRNA contains 5 t-noncoding sequences that confer translation initiation by internal ribosome binding

Proteolytic cleavage of initiation factor elF-4-Gamma in the reticulocyte lysate inhibits translation of capped mRNAs but enhances that of uncapped mRNAs

The relationship between eukaryotic translation and mRNA stability -a short upstream open reading frame strongly inhibits translational initiation and greatly accelerates mRNA degradation in the yeast Saccharomyces cerevisiae

Translational control during amino acid starvation

Pulling the ribosome out of frame by +1 at a programmed frameshift site by cognate binding of aminoacy-tRNA

Translation initiation at non-AUG triplets in mammalian cells

Studies on the translocation of the amino terminus of apolipoprotein B into the endoplasmic reticulum

Leaky UAG termination codon in tobacco virus RNA

Translation of tobacco rattle virus RNAs in vitro: four proteins from three RNAs

Internal initiation of translation of eukaryotic mRNA directed by a sequence derived from poliovirus RNA

Modification of the coding sequence enhances plant expression of insect control protein genes

Binding of ribosomes to the 5 t leader sequence (N = 258) of RNA 3 from alfalfa mosaic virus

The role of the leader sequence during thermal repression of translation in maize, tobacco and carrot protoplasts

The 5t-untranslated leader sequence of potato virus X RNA enhances the expression of a heterologous gene in vivo

The identity of the base following the stop codon determines the efficiency of in vivo translational termination in Escherichia coli

Ribosomal decoding processes at codons in the A or P sites depend differently on 2t-OH groups

High molecular mass forms of basic fibroblast growth factor are initiated by alternative CUG codons

Ribosomal frameshifting in plants: A novel signal directs the-1 frameshift in the synthesis of the putative replicase of potato leafroll luteovirus

Initiation of translation of the flglucuronidase reporter gene at internal AUG codons in plant cells

The homeobox gene ATH1 of Arabidopsis is derepressed in the photomorphogenic mutants copl and detl

Methyl jasmonate represses translation initiation of a specific set of mRNAs in barley

Cap recognition and the entry of mRNA into the protein synthesis cycle

Regulation of eukaryotic protein synthesis by initiation factors

Identification of the initiation codon of plum pox potyvirus genomic RNA

Eukaryotic protein elongation factors

Post-transcriptionai steps in the expression of chloroplast genes

Infectious transcripts from cloned cucumber necrosis virus cDNA: evidence for a bifunctional subgenomic RNA

Evidence for ribosome scanning during translation initiation of mRNAs in transformed plant cells

Plant viruses as model systems for the study of non-canonical translation mechanisms in higher plants

Pararetroviruses and retroviruses: a comparative review of viral structure and gene expression strategies

Nucleotide sequence of carnation ringspot dianthovirus RNA-1

The 3t-terminal untranslated region of alfalfa mosaic virus RNA 4 facilitates the RNA entry into translation in a cell-free system~

Regulation of protein synthesis at the elongation stage

The poly(A) binding protein is required for poly(A) shortening and 60S ribosomal subunitdependent translation initiation

Differential codon usage -A safeguard against inappropriate expression of specialized genes

Basepairing with 18S ribosomal RNA in internal initiation of translation

The function of plant heat shock promoter elements in the regulated expression of chimaeric genes in transgenic tobacco

The fulllength transcript of caulimovirus is a polycistronic mRNA whose genes are transactivated by the product of gene VI

Regulation of caulimovirus gene expression and the involvement of cisacting elements on both viral transcripts

The tomato bushy stunt virus replicase proteins are coordinately expressed and membrane associated

The reverse transcriptase gene of cauliflower mosaic virus is translated separately form the capsid gene

Translation initiation at a downstream AUG occurs with increased efficiency when the upstream AUG is located very close to the 5; cap

Expression of a plant viral polycistronic mRNA in yeast Saccharomyces cerevisiae mediated by a plant virus translational transactivator

Destabilization of messenger RNA/complementary DNA duplexes by the elongating 80S ribosome

The structure of an RNA pseudoknot that causes efficient frameshifting in mouse mammary tumor virus

Structure and function of the selenium translation element in the 3; -untranslated region of human cellular glutathione peroxidase mRNA

Complete nucleotide sequence and organization of the bipartite RNA genome of soil-borne wheat mosaic virus

The influence of poly(A)-binding proteins on translation of poly(A) + RNA in a cell-free system from embryo axes of dry pea seeds

Translational control of photoinduced expression of the Cat2 catalase gene during leaf development in maize

Analysis of leaky viral translation termination codons in vivo by transient expression of improved fl-glucuronidase vectors

The signal for a leaky UAG stop codon in several plant viruses includes the two downstream codons

Characterization of the 5;-leader of tobacco mosaic virus RNA as a general enhancer of translation in vitro

Studies on the mechanism of translational enhancement by the 5;-leader sequences of tobacco mosaic virus RNA

Synthesis and turnover of the lightharvesting chlorophyll a/b-protein in Lemna gibba grown with intermittent red light: possible translational control

Translational efficiency and competitive abilit~¢ of mRNAs with 5t -untranslated o~/3-1eader of potato virus X RNA

tRNA-tRNA interactions within cellular ribosomes

Ribosomal pausing during translation of an RNA pseudoknot

Picornavirus RNA translation continues to surprise

High-level expression of sperm whale myoglobin in Escherichia coli

Regulation of translation by specific protein mRNA interactions

The products of the SUP45 (eRF1) and SUP35 genes interact to mediate translation termination in Saccharomyces cerevisiae

The end in sight: terminating translation in eukaryotes

Particularit6s structurales de l'acide ribonucl6ique extrait du virus de la mosaique janne du navet. II

Post-transcriptional regulation of nuclear-encoded genes in higher plants: the roles of mRNA stability and translation

Rice dwarf phytoreovirus segment S12 transcript is tricistronic in vitro

Characterization of a potato leafroll luteovirus subgenomic RNA: differential expression by internal translation initiation and UAG suppression

PLRV expression in potato

Effect of eukaryotic initiation factor 4F on AUG selection in a bicistronic mRNA

Cauliflower mosaic virus gene VI causes growth suppression, development of necrotic spots and expression of defence-related genes in transgenic tobacco plants

Evidence that the 75K readthrough protein of beet necrotic yellow vein virus RNA-2 is essential for transmission by the fungus Polymyxa betae

Successful expression in pollen of various plant species of in vitro synthesized mRNA introduced by particle bombardment

Optimizing the expression of chimeric genes in plant cells

RNA pseudoknots: translational frameshifting and readthrough on viral RNAs

Analysis of the role of the pseudoknot component in the SRV-1 gag-pro ribosomal frameshift signal -Loop lengths and stability of the stern regions

RE: Cap recap: the involvement of eIF-4F in regulating gene expression

Hypothesis: is eukaryotic initiation factor 2 the scanning factor

Cowpea mosaic virus middle component RNA contains a sequence that allows internal binding of ribosomes and that requires eukaryotic initiation factor 4F for optimal translation

The 5 / and 3 / untranslated regions of satellite tobacco necrosis virus RNA affect translational efficiency and dependence on a 5 ~ cap structure

Effects of sequence elements in the potato virus X RNA 5r-nontranslated &H-leader on its translation enhancing activity

Ribosomal movement impeded at a pseudoknot required for frameshifting

Continuously coupled transcription-translation system for the production of rice cytoplasmic aldolase

I: Translation -Knowing when to stop

Analysis of a translational enhancer upstream from the coat protein open reading frame of potato virus S

Multiple ribosome binding to the 5/-terminal leader sequence of tobacco mosaic virus RNA. Assembly of an 80S ribosome-mRNA complex at an AUU codon

Ribosomal frameshifting requires a pseudoknot in the Saccharomyces cerevisiae double-stranded RNA virus

Alternative translation of human fibroblast growth factor 2 mRNA occurs by internal entry of ribosomes

Alternative translation initiation of the Moloney murine leukemia virus mRNA controlled by internal ribosome entry involving the p57/PTB splicing factor

Stop making sense. Regulation at the level of termination in eukaryotic protein synthesis

Peptide chain termination

Codon context effects in virus translational readthrough. A study in vitro of the determinants of TMV and Mo-MuLV amber suppression

Translatability of plant mRNAs strongly influences its accumulation in transgenic plants

Translational arrest in hypoxic potato tubers is correlated with the aberrant association of elongation factor EF-1-alpha with polysomes

Synthesis of full-length transcripts of beet western yellows virus RNA: messenger properties and biological activity in protoplasts

The sequence between nucleotides 161 and 512 of cowpea mosaic virus M RNA is able to support internal initiation of translation in vitro

Special peptidyl-tRNA molecules can promote translational frameshift without slippage

Sequence of a maize 21 kd zein gene from maize containing an in-frame stop codon

A sequence located 4.5 to 5 kilobases from the 51 end of the barley yellow dwarf virus (PAV) genome strongly stimulates translation of uncapped RNA

Elongation and termination reactions of protein synthesis on maize root tip polysomes studied in a homologous cell-free system

Infectious TYMV RNA from cloned cDNA: Effects in vitro and in vivo of point substitutions in the initiation codons of two extensively overlapping ORFs

E. coli ribosomes re-phase on retroviral frameshift signals at rates ranging from 2-50 percent

A nascent peptide is required for ribosomal bypass of the coding gap in bacteriophage T4 gene 60

The leader peptide of yeast gene CPA1 is essential for the translational repression of its expression

A protein complex required for signal-sequence-specific sorting and translocation

Evidence that a downstream pseudoknot is required for translational readthrough of the Moloney murine leukemia virus gag stop codon

Pseudoknot-dependent read-through of retroviral gag termination codons: importance of sequences in the spacer of loop 2

Ribosome pausing and stacking during translation of a eukaryotic mRNA

Synthesis of the putative red clover necrotic mosaic virus RN A polymerase by ribosomal frameshifting in vitro

A tunnel in the large ribosomal subunit revealed by three-dimensional image reconstruction

Translational readthrough of an amber termination codon during synthesis of feline leukemia virus protease

The remarkable variety of plant virus genomes

Translation enhancing properties of the 5t-leader of potato virus X genomic RNA

Pseudouridine in the anticodon G~A of plant cytoplasmic tRNA Tyr is required for UAG and UAA suppression in the TMV specific context

The leaky UGA termination codon of tobacco rattle virus RNA is suppressed by tobacco chloroplast and cytoplasmic tRNAs Trp with CmCA anticodon

Enhancing effect of the 3 p-untranslated region of tobacco mosaic virus RNA on protein synthesis in vitro

Clustering of low usage codons and ribosomal movement

Termination of translation in eukaryotes is governed by two interacting polypeptide chain release factors, eRF1 and eRF3

Cell-free translation of beet-necrotic yellow vein virus: read-through of the coat protein cistron

Cauliflower mosaic virus gene VI controis translation from dicistronic expression units in transgenic Arabidopsis plants

Tissue specific expression of the retinoic acid receptor-beta-2: Regulation by short open reading frames in the 5~-noncoding region

We acknowledge the critical reading of the manuscript by Pat King and the steady help and interest of Ingo Potrykus. The authors are supported by biotechnology grants of the Swiss National Funds.