Role of a Short Open Reading Frame in Ribosome Shunt on the Cauliflower Mosaic Virus RNA Leader

The pregenomic 35 S RNA of cauliflower mosaic virus (CaMV) belongs to the growing number of mRNAs known to have a complex leader sequence. The 612-nucleotide leader contains several short open reading frames (sORFs) and forms an extended hairpin structure. Downstream translation of 35 S RNA is nevertheless possible due to the ribosome shunt mechanism, by which ribosomes are directly transferred from a take-off site near the capped 5′ end of the leader to a landing site near its 3′ end. There they resume scanning and reach the first long open reading frame. We investigated in detail how the multiple sORFs influence ribosome migration either via shunting or linear scanning along the CaMV leader. The sORFs together constituted a major barrier for the linear ribosome migration, whereas the most 5′-proximal sORF, sORF A, in combination with sORFs B and C, played a positive role in translation downstream of the leader by diverting scanning ribosomes to the shunt route. A simplified, shunt-competent leader was constructed with the most part of the hairpin including all the sORFs except sORF A replaced by a scanning-inhibiting structure. In this leader as well as in the wild type leader, proper translation and termination of sORF A was required for efficient shunt and also for the level of shunt enhancement by a CaMV-encoded translation transactivator. sORF A could be replaced by heterologous sORFs, but a one-codon (start/stop) sORF was not functional. The results implicate that in CaMV, shunt-mediated translation requires reinitiation. The efficiency of the shunt process is influenced by translational properties of the sORF. The pregenomic 35 S RNA of cauliflower mosaic virus (CaMV) belongs to the growing number of mRNAs known to have a complex leader sequence. The 612-nucleotide leader contains several short open reading frames (sORFs) and forms an extended hairpin structure. Downstream translation of 35 S RNA is nevertheless possible due to the ribosome shunt mechanism, by which ribosomes are directly transferred from a take-off site near the capped 5′ end of the leader to a landing site near its 3′ end. There they resume scanning and reach the first long open reading frame. We investigated in detail how the multiple sORFs influence ribosome migration either via shunting or linear scanning along the CaMV leader. The sORFs together constituted a major barrier for the linear ribosome migration, whereas the most 5′-proximal sORF, sORF A, in combination with sORFs B and C, played a positive role in translation downstream of the leader by diverting scanning ribosomes to the shunt route. A simplified, shunt-competent leader was constructed with the most part of the hairpin including all the sORFs except sORF A replaced by a scanning-inhibiting structure. In this leader as well as in the wild type leader, proper translation and termination of sORF A was required for efficient shunt and also for the level of shunt enhancement by a CaMV-encoded translation transactivator. sORF A could be replaced by heterologous sORFs, but a one-codon (start/stop) sORF was not functional. The results implicate that in CaMV, shunt-mediated translation requires reinitiation. The efficiency of the shunt process is influenced by translational properties of the sORF. untranslated region short open reading frame open reading frame internal ribosome entry site Kozak stem cauliflower mosaic virus the CaMV-encoded transactivator protein rice tungro bacilliform virus, SbCMV, soybean chlorotic mottle virus chloramphenicol acetyltransferase At least three distinct steps involving 40 S ribosomal subunits and the mRNA 5′-untranslated region (5′-UTR)1 precede the formation of translation competent 80 S ribosomes on most mRNAs in eukaryotic cells. First, the 40 S subunit binds to the capped 5′ end of the mRNA; second, ribosomes translocate to the start codon by linear scanning; and third, the start codon is recognized (1.Kozak M. J. Cell Biol. 1989; 108: 229-241Crossref PubMed Scopus (2789) Google Scholar, 2.Kozak M. Gene (Amst.). 1999; 234: 187-208Crossref PubMed Scopus (1116) Google Scholar, 3.Merrick W.C. Hershey J.W.B. Hershey J.W.B. Mathews M.B. Sonenberg N. Translational Control. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY1996: 31-70Google Scholar). A number of ribosome- or RNA-associated protein factors are involved in cap recognition, removal of RNA secondary structure, and initiation complex formation (4.Pestova T.V. Hellen C.U. Trends Biochem. Sci. 1999; 24: 85-87Abstract Full Text Full Text PDF PubMed Scopus (34) Google Scholar). The helix-destabilizing capacity of the scanning ribosome appears limited. Double-stranded regions with a free energy of less than −50 kcal/mol block the scanning process (5.Kozak M. Mol. Cell. Biol. 1989; 9: 5134-5142Crossref PubMed Scopus (482) Google Scholar). Besides secondary structure, alternative translation start sites could interfere with scanning. Since the composition of the scanning complex is altered upon translation initiation, most notably by “consumption” of the bound eIF2-GTP-initiator methionyl tRNA complex, it was long assumed that eukaryotic ribosomes can initiate only once on an mRNA. However, particularly after translation of a short ORF (sORF), at least a fraction of ribosomes can reinitiate translation on the same mRNA. How ribosomes regain initiation capacity, how they reach further downstream ORFs, and how all this is influenced by the nature of the first ORF is still unclear. In the yeast GCN4 system (6.Hinnebusch A.G. J. Biol. Chem. 1997; 272: 21661-21664Abstract Full Text Full Text PDF PubMed Scopus (430) Google Scholar) and in in vitro systems (7.Kozak M. Mol. Cell. Biol. 1987; 7: 3438-3445Crossref PubMed Scopus (401) Google Scholar), reinitiation is distance-dependent, suggesting that a certain time is required to reestablish initiation competence. In GCN4, this time (translated to a scanning distance) was inversely correlated to the concentration of active eIF2. An increasing number of eukaryotic mRNAs have been detected with strong secondary structures and/or alternative start sites in their 5′-UTR, particularly those of transcription factors or protooncogenes (8.Willis A.E. Int. J. Biochem. Cell Biol. 1999; 31: 73-86Crossref PubMed Scopus (116) Google Scholar). Alternative start sites may create short ORFs or N-terminally extended variants of the major ORF. Alternative start sites are not necessarily easily predictable, because “non-AUG” codons,i.e. codons deviating from AUG in one position, may act as start codons (e.g. Ref. 9.Gordon K. Fütterer J. Hohn T. Plant J. 1992; 2: 809-813PubMed Google Scholar), particularly in conjunction with downstream secondary structures that may pause scanning ribosomes (10.Kozak M. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 8301-8305Crossref PubMed Scopus (393) Google Scholar). In general, complex 5′-UTRs have negative effects on translation efficiency and it has been argued that their function could be to down-regulate expression. However, such features may also be used for more subtle post-transcriptional control of gene expression, although only a few examples have been studied in any detail (6.Hinnebusch A.G. J. Biol. Chem. 1997; 272: 21661-21664Abstract Full Text Full Text PDF PubMed Scopus (430) Google Scholar, 11.Alderete J.P. Jarrahian S. Geballe A.P. J. Virol. 1999; 73: 8330-8337Crossref PubMed Google Scholar, 12.Delbecq P. Werner M. Feller A. Filipkowski R.K. Messenguy F. Pierard A. Mol. Cell. Biol. 1994; 14: 2378-2390Crossref PubMed Scopus (60) Google Scholar, 13.Mize G.J. Ruan H. Low J.J. Morris D.R. J. Biol. Chem. 1998; 273: 32500-32505Abstract Full Text Full Text PDF PubMed Scopus (69) Google Scholar, 14.Muckenthaler M. Gray N.K. Hentze M.W. Mol. Cell. 1998; 2: 383-388Abstract Full Text Full Text PDF PubMed Scopus (206) Google Scholar, 15.Sonstegard T.S. Hackett P.B. J. Virol. 1996; 70: 6642-6652Crossref PubMed Google Scholar, 16.Wang Z. Fang P. Sachs M.S. Mol. Cell. Biol. 1998; 18: 7528-7536Crossref PubMed Scopus (53) Google Scholar). How the eukaryotic translation machinery negotiates complex 5′-UTRs is still far from being understood, even though a number of alternatives to the scanning process have been described. Direct binding to internal ribosome entry sites (IRES) was observed in many viruses and some cellular mRNAs (17.Jackson R.J. Kaminski A. RNA. 1995; 1: 985-1000PubMed Google Scholar). Entry points can be directly at the start codon or upstream of it, thus requiring a translocation step to the start codon. The other alternative to continuous scanning is the ribosome shunt, which leads to the bypass of scanning-inhibitory regions by direct transfer of ribosomal subunits between two RNA regions. Ribosome shunt has so far been described mainly for viral RNAs: cauliflower mosaic virus (CaMV) (18.Fütterer J. Gordon K. Sanfaçon H. Bonneville J-M. Hohn T. EMBO J. 1990; 9: 1697-1707Crossref PubMed Scopus (75) Google Scholar, 19.Fütterer J. Kiss-László Z. Hohn T. Cell. 1993; 73: 789-802Abstract Full Text PDF PubMed Scopus (179) Google Scholar), rice tungro bacilliform virus (RTBV) (20.Fütterer J. Potrykus I. Bao Y. Li L. Burns T.M. Hull R. Hohn T. J. Virol. 1996; 70: 2999-3010Crossref PubMed Google Scholar), Sendai virus (21.Curran J. Kolakofsky D. EMBO J. 1988; 7: 2869-2874Crossref PubMed Scopus (68) Google Scholar, 22.Latorre P. Kolakofsky D. Curran J. Mol. Cell. Biol. 1998; 18: 5021-5031Crossref PubMed Scopus (93) Google Scholar), adenovirus (23.Yueh A. Schneider R.J. Genes Dev. 1996; 10: 1557-1567Crossref PubMed Scopus (167) Google Scholar), human papillomavirus (24.Remm M. Remm A. Ustav M. J. Virol. 1999; 73: 3062-3070Crossref PubMed Google Scholar), and budgerigar fledgling disease virus (25.Li J. Molecular Analysis of Gene Expression in BFDV Ph.D. thesis. University of Giessen, Giessen, Germany1996: 1-180Google Scholar). In most of these cases, viral factors are not required, even though they may enhance the process, suggesting that ribosome shunt is a normal, cellular mechanism and might also occur on cellular RNAs. Ribosome shunt combines the efficient binding of ribosomes through the mRNA cap and the versatility of internal ribosome entry. The mechanisms that allow 40 S ribosomes to bypass RNA regions are not understood. Here we describe analysis of some of the requirements for ribosome shunt on the leader of the CaMV 35 S RNA. This polycistronic RNA begins with a long leader of 612 nucleotides preceding ORF VII. The leader is inhibitory for the scanning process, because it contains up to nine AUGs creating sORFs, 2–35 codons in length (26.Pooggin M.M. Hohn T. Fütterer J. J. Virol. 1998; 72: 4157-4169Crossref PubMed Google Scholar), and forms an extended central hairpin (27.Hemmings-Mieszczak M. Steger G. Hohn T. J. Mol. Biol. 1997; 267: 1075-1088Crossref PubMed Scopus (36) Google Scholar). Downstream translation is nevertheless possible via the ribosome shunt (18.Fütterer J. Gordon K. Sanfaçon H. Bonneville J-M. Hohn T. EMBO J. 1990; 9: 1697-1707Crossref PubMed Scopus (75) Google Scholar, 19.Fütterer J. Kiss-László Z. Hohn T. Cell. 1993; 73: 789-802Abstract Full Text PDF PubMed Scopus (179) Google Scholar). Shunting functions in various translation systems including plant protoplasts (19.Fütterer J. Kiss-László Z. Hohn T. Cell. 1993; 73: 789-802Abstract Full Text PDF PubMed Scopus (179) Google Scholar, 28.Fütterer J. Gordon K. Pfeifer P. Sanfaçon H. Pisan B. Bonneville J.-M. Hohn T. Virus Genes. 1989; 3: 45-55Crossref PubMed Scopus (46) Google Scholar), wheat germ extract (29.Schmidt-Puchta, W., Dominguez, D., Lewetag, D., and Hohn, T. (1997) 25, 2854–2860Google Scholar), rabbit reticulocyte lysate (30.Ryabova L.A. Hohn T. Genes Dev. 2000; 14: 817-829PubMed Google Scholar), and transgenic plants (31.Schärer-Hernández N. Hohn T. Virology. 1998; 242: 403-413Crossref PubMed Scopus (11) Google Scholar). It requires the formation of a stable stem section at the base of the central hairpin (stem section 1) (32.Dominguez D.I. Ryabova L.A. Pooggin M.M. Schmidt-Puchta W. Fütterer J. Hohn T. J. Biol. Chem. 1998; 273: 3669-3678Abstract Full Text Full Text PDF PubMed Scopus (39) Google Scholar, 33.Hemmings-Mieszczak M. Steger G. Hohn T. RNA. 1998; 4: 101-111PubMed Google Scholar) and the presence of an sORF (sORF A) in front of this or a similar stable structure (19.Fütterer J. Kiss-László Z. Hohn T. Cell. 1993; 73: 789-802Abstract Full Text PDF PubMed Scopus (179) Google Scholar, 32.Dominguez D.I. Ryabova L.A. Pooggin M.M. Schmidt-Puchta W. Fütterer J. Hohn T. J. Biol. Chem. 1998; 273: 3669-3678Abstract Full Text Full Text PDF PubMed Scopus (39) Google Scholar, 34.Hemmings-Mieszczak M. Hohn T. RNA. 1999; 5: 1149-1157Crossref PubMed Scopus (25) Google Scholar). The importance of sORF A has also been underscored by revertants obtained after infection of plants with CaMV mutants (26.Pooggin M.M. Hohn T. Fütterer J. J. Virol. 1998; 72: 4157-4169Crossref PubMed Google Scholar). In this report, we investigated (i) combinatorial and individual effects of the sORFs in the CaMV leader on ribosome shunt and linear scanning along the leader and (ii) the effect of the CaMV-encoded transactivator of polycistronic translation, TAV (35.Bonneville J-M. Sanfaçon H. Fütterer J. Hohn T. Cell. 1989; 59: 1135-1143Abstract Full Text PDF PubMed Scopus (166) Google Scholar, 36.Fütterer J. Hohn T. EMBO J. 1991; 10: 3887-3896Crossref PubMed Scopus (86) Google Scholar), on shunt-mediated expression. The group of sORFs constituted a major inhibitory barrier for linear scanning, whereas the most 5′-proximal sORF A, in combination with sORFs B and C, alleviated this inhibition by diverting scanning ribosomes to the shunt route. Parameters of sORF A were further investigated in a simplified shunt-competent leader. In this leader mimicking the wild type CaMV leader, proper elongation and termination of sORF A were required for efficient ribosome shunt and also for the level of shunt enhancement by TAV. sORF A could be functionally replaced by heterologous, naturally occurring sORFs; however, a start/stop codon sequence was not functional. These results support a reinitiation model of the ribosome shunt. The “wild type” plasmid used in this study as a base-line control has been previously described as pLC20 (18.Fütterer J. Gordon K. Sanfaçon H. Bonneville J-M. Hohn T. EMBO J. 1990; 9: 1697-1707Crossref PubMed Scopus (75) Google Scholar). Its expression unit consists of (i) the fragment of the CaMV strain CM4–184 genome comprising the 35 S RNA promoter and the complete leader, (ii) a CAT reporter gene fused to the ORF VII AUG, and (iii) the CaMV terminator region. All point mutations were introduced into the leader sequence by the polymerase chain reaction ligation method (37.Galvez A.F. DeLumen B.O. Plant Mol. Biol. Rep. 1995; 13: 232-242Crossref Scopus (2) Google Scholar). The polymerase chain reaction primers and conditions used as well as the detailed description of the mutations removing all the sORFs have been published elsewhere (26.Pooggin M.M. Hohn T. Fütterer J. J. Virol. 1998; 72: 4157-4169Crossref PubMed Google Scholar). The sequences of various point mutations in sORF A are depicted in the corresponding figures. The mutant versions of the leader were subcloned from plasmid pV322 (26.Pooggin M.M. Hohn T. Fütterer J. J. Virol. 1998; 72: 4157-4169Crossref PubMed Google Scholar) into pLC20 using unique sites EcoRV and ClaI naturally occurring in the promoter and near the 3′ end of the leader, respectively. Two large deletions in the leader were made taking advantage of (i) new restriction sites XhoI and HindIII created by the point mutations of sORFs B and F, respectively, yielding delXH and its derivatives delXHMAtTG, delXHMAtaG, and A::F, which carry additional point mutations in sORF A; or (ii) the sites XhoI and ClaI, yielding delXC and its derivative delXCMAtaG lacking sORF A. In both cases, plasmid MALL+A, which contains sORF A in an otherwise AUG-free leader or its derivatives with the mutated sORF A (26.Pooggin M.M. Hohn T. Fütterer J. J. Virol. 1998; 72: 4157-4169Crossref PubMed Google Scholar), was cut with the corresponding enzymes, the ends filled in with Klenow, and ligated. The simplified shunt-competent construct delXHKS was generated by replacing the XhoI-HindIII fragment of MALL+A with the desired oligonucleotides, such that theHindIII site was maintained. In the latter construct, newXhoI and AflII sites in front of sORF A and the KS sequence, respectively, were introduced by oligonucleotide-directed mutagenesis (CAT54gAg and Cg113u), yielding KSXAHA. Disruption of stem section 1 in the latter construct or modifications in the sORF A region were performed by replacing theXhoI-AflII fragment with the corresponding oligonucleotides. Restoration of stem section 1 was carried out by replacing the HindIII-ClaI fragment with an oligonucleotide carrying the desired compensatory mutations. The Kozak stem in the center of the full-length leader was created by replacing the fragment between two BglII sites (positions 218 and 238) with the self-complementary oligonucleotide 5′-GATCggggcgcgtggtggcggctgcagccgccaccacgcgcccc-3′. Protoplasts were prepared from suspension cultures ofOrychophragmus violaceus and transfected with plasmid DNA by electroporation as described previously (28.Fütterer J. Gordon K. Pfeifer P. Sanfaçon H. Pisan B. Bonneville J.-M. Hohn T. Virus Genes. 1989; 3: 45-55Crossref PubMed Scopus (46) Google Scholar) with a few modifications. Briefly, 0.7 ml of protoplasts (about 2 × 106) were mixed with plasmids on ice, transferred to 0.4-cm electrode gap cuvettes (Bio-Rad), and electroporated by discharging 960 microfarads at 450 V and 200 ohms using the Bio-Rad Gene Pulser. After 20–24 h of incubation at 27 °C, protoplasts were harvested by centrifugation and protein extract prepared by three cycles of freezing and thawing in 200 μl of GUS extraction buffer (50 mmNaH2PO4 (pH 7.0), 10 mm EDTA, 0.1% Triton X-100, 0.1% sarcosyl). 30-μl aliquots were immediately taken for reporter gene assays, or stored at −70 °C. CAT expression levels were determined using the Roche Molecular Biochemicals CAT enzyme-linked immunosorbent assay kit as recommended by the manufacturer. GUS activity was measured by a fluorimetric assay (38.Jefferson R.A. Burgess S.M. Hirsh D. Proc. Natl. Acad. Sci. U. S. A. 1986; 83: 8447-8451Crossref PubMed Scopus (816) Google Scholar). 10 μg of CAT plasmid was always cotransfected with 2.5 μg of GUS-expressing plasmid (pV594) to serve as an internal standard of transfection efficiency. For transactivation, 5 μg of plasmid pV374 expressing the TAV protein was also added. The latter plasmid has been previously described as pHELP7 (35.Bonneville J-M. Sanfaçon H. Fütterer J. Hohn T. Cell. 1989; 59: 1135-1143Abstract Full Text PDF PubMed Scopus (166) Google Scholar). For each CAT construct, transfections were repeated at least three times in duplicate (± TAV) with new DNA preparations of independent clones and in new protoplast batches. Additional transfections were performed if deviations in the normalized CAT expression were more than 10% of the average value. It should be noticed that greater deviations (maximum 20%) were observed for constructs with scanning-mediated expression. We assume that varying cell parameters in different batches of protoplasts might differentially affect scanning- versus shunt-mediated expression that was set to 100% in all experiments. The influence of CaMV 35 S RNA leader variants on downstream translation was tested in transiently transfected protoplasts. Discrimination between translation by scanning or by shunting was achieved by insertion of a stable hairpin structure, “Kozak stem” or “KS” (5.Kozak M. Mol. Cell. Biol. 1989; 9: 5134-5142Crossref PubMed Scopus (482) Google Scholar), which has been used previously as an effective inhibitor of scanning in plant cells (36.Fütterer J. Hohn T. EMBO J. 1991; 10: 3887-3896Crossref PubMed Scopus (86) Google Scholar). We also tested the influence of the CaMV-derived multifunctional protein, TAV, one of whose functions is to promote translational reinitiation on polycistronic mRNAs (36.Fütterer J. Hohn T. EMBO J. 1991; 10: 3887-3896Crossref PubMed Scopus (86) Google Scholar). The transient expression level of a chloramphenicol acetyltransferase (CAT) ORF downstream of a wild type CaMV 35 S RNA leader was set to 100% (Fig.1, line 1). The removal of the start codons of all the sORFs in the leader by point mutations increased expression about 5-fold (line 2). Additional removal of the complete leader hairpin structure further increased expression, but only by a factor of 1.7 (line 4); the particularly stable base of the leader hairpin (stem section 1) contributes about two thirds of this structural inhibition of translation (line 3). These data indicate that the sORFs in the leader are the major inhibitory elements. The inhibitory effect of some of the individual sORFs was tested by reintroduction of their start codons. sORF D′ alone was not inhibitory, sORF E“ was only slightly inhibitory, and sORF F that overlaps ORF VII::CAT was stronger than the others (Fig.2 A, lines 4, 5, and 6, respectively). Inhibition by sORF combinations was not additive and not predictable. Combination of sORFs E” and F (line 7) was more inhibitory than E“ alone but less than F alone; addition of D′ to E” and F (line 8) created a greater inhibition, although D′ alone had had no effect at all. Addition of sORFs D, E, and E′ to the latter construct only slightly reduced expression (Fig.2 B, line 9). These data could be explained by analogy to the yeast GCN4 mRNA with distance-dependent reinitiation (39.Abastado J.P. Miller P.F. Hinnebusch A.G. New Biol. 1991; 3: 511-524PubMed Google Scholar); a fraction of ribosomes that resume scanning and reinitiate after translation of E“ would bypass the start codon of F because of the short intervening distance and thus the negative effect of the latter sORF is partially alleviated. In a similar way, after translation of D′, ribosomes can efficiently reinitiate at the CAT ORF, unless they are intercepted by the intervening sORFs. In the later cases, a potential second round of reinitiation (after sORFs D, E, E′, or E”) is probably inefficient. While individual readdition of the sORFs to the 3′ half of an otherwise AUG-free leader revealed their potential for translation inhibition, their individual removal from the wild type, sORF-containing leader had no effect on expression (data not shown), indicating that, in this context, the individual sORFs have no major influence on the expression level. This contrasts with the behavior of sORFs A, B, and C; removal of any of these sORFs individually or in combination reduced expression (constructs in Fig. 2 B, compared with wild type), suggesting that they alleviate the negative effect of the rest of the leader. sORF A removal had the most striking effect, decreasing expression 5-fold (line 12). The positive function of these sORFs was only detectable in the context of the wild type leader. In combination with other sORF mutations, individual sORFs A, B, and C could also display negative effects on overall expression (e.g. Fig. 2, pairs 16/2 or 15/9 for A, 11/10 for B, and 10/9 for C). Insertion of the KS into the center of the wild type leader had only a slight effect on downstream translation (Fig.2 A), confirming our previous observations (19.Fütterer J. Kiss-László Z. Hohn T. Cell. 1993; 73: 789-802Abstract Full Text PDF PubMed Scopus (179) Google Scholar) and indicating that, in this case, translation mostly depends on shunt and only little scanning through the leader is allowed. In contrast, insertion of KS into an AUG-free leader or a leader lacking the sORFs A, B, and C greatly reduced expression (Fig. 2, lines 2 and 9 versus 1),i.e. in the absence of sORFs A, B, and C, most or all downstream translation depends on continuous, linear scanning and only a small proportion (if any) of the ribosomes perform a shunt. The presence of sORF A in the otherwise AUG-free leader allowed some ribosomes to overcome the KS barrier (Fig. 2 C,line 16 versus line 2), and the presence of sORFs A, B, and C completely restored shunting (line 17 versus line 1, with KS). In the absence of KS, translation from these mRNAs occurs by shunting and scanning. Comparison of mutants in one or several of the respective sORFs with and without KS revealed that removal of any of the first three sORFs increased the contribution of scanning to overall expression and reduced the proportion of shunting (Fig. 2 B). In the presence of the KS, the knock out mutation of sORF A (line 12) had the most drastic inhibitory effect on shunt-mediated translation (∼13-fold), while the individual mutations of sORFs B and C (lines 13 and 14) were less inhibitory (∼1.5- and 4-fold, respectively). Because of difficulties with quantification of the very low levels of reporter RNA obtained from transfected protoplasts (data not shown), we could not exclude that some mutations might affect RNA stability. However, the fact that the same combination of the sORF point mutations could have either positive or negative effects on expression, depending on the leader context and consequently on the mode of ribosome migration (Fig. 2, compare lines 1 and2, in the absence or presence of KS), suggests that mainly the rate and/or the mechanism of translation initiation are affected. Previously, it has been shown (29.Schmidt-Puchta, W., Dominguez, D., Lewetag, D., and Hohn, T. (1997) 25, 2854–2860Google Scholar) that in a shunt-competent wheat germ extract the CAT transcripts containing the wild type CaMV leader without or with the Kozak stem (see Fig. 2, line 1) were as stable as the leader-less CAT transcript. This also supports that neither the CaMV sORFs nor the Kozak stem sequence represent instability determinants. These results show that sORFs can fundamentally alter the translation properties of an mRNA not just by regulating the initiation capacity of ribosomes but also by changing the mode how ribosomes access the internal mRNA regions. sORF A was confirmed as a major determinant of efficient ribosome shunting, but its effects are influenced by the more internal sORFs B and C. The CaMV transactivator TAV acts by enhancing the reinitiation potential of ribosomes that have already translated one ORF (36.Fütterer J. Hohn T. EMBO J. 1991; 10: 3887-3896Crossref PubMed Scopus (86) Google Scholar). Here, TAV was found to enhance expression downstream of the 35 S RNA leader by a factor of 2–3 in all cases involving significant ribosome shunting (Fig. 2, seeInterpretation column), allowing expression levels approaching 30% of the maximal translation potential; the latter can be deduced from the reference construct lacking all inhibitory elements (Fig. 1, line 4). Scanning-mediated expression was influenced not at all or much less. When an RNA is translated by both mechanisms, the shunting-dependent fraction is activated by TAV whereas the scanning-dependent fraction is hardly influenced (Fig.2 C, line 17). Shunt requires sORF A and thus all constructs showing a strong effect of TAV contain this sORF. When the start codon was mutated to UAG (Fig. 2 B,line 12), downstream translation occurred only by scanning and TAV had at most a small effect; however, with the non-AUG codon UUG, a small amount of shunting is apparent in the presence of TAV (Fig. 2 B, line 11, +KS), suggesting that the UUG is active as a start codon in this context. TAV may function in restoring, at least partially, the initiation capacity of shunting ribosomes that have translated sORF A. For scanning-dependent reinitiation after sORF A translation, this function might not be required because the distance in the leader is long enough to acquire reinitiation capacity during the scanning process (e.g. Fig. 2, line 16). The data presented so far provide strong evidence that sORF A is translated. To investigate how efficiently ribosomes recognize the sORF A start codon, we fused sORF A in the short leader containing only stem section 1 to the AUG-less remnants of sORF F, which overlap the ORFVII::CAT for 35 nucleotides (Fig. 3, line 2). The mutation drastically reduced expression, suggesting that only a small proportion of scanning ribosomes skips the sORF A AUG start codon. As a control, additional mutation of the latter AUG to UAG fully derepressed translation (line 3). To study the specific role of sORF A and TAV in shunting, the complexity of translation events on the reporter RNA was reduced by replacement of the upper part of the leader hairpin (including all other sORFs) with the much shorter but energy-rich Kozak stem (Fig.4 A). In this simpler expression unit, CAT translation was almost completely dependent on the presence of sORF A. Expression levels, the effects of sORF A start codon mutations to UAG or UUG (see Fig.5 A, line 8), and the response to TAV were similar to those observed with the wild type leader (Fig. 4 A), suggesting that in both cases shunting occurs. This similarity was maintained only when the KS was present and stem section 1 intact. In the absence of KS, sORF A exhibited a 2.5-fold inhibition (Fig. 4 B), indicating that the respective RNA is translated mainly by scanning. In the presence of KS, disruption of stem section 1 by multiple point mutations (Fig.4 C) completely abolished expression as expected for a strictly scanning-dependent mRNA with a KS element. Restoration of stem section 1 by compensatory mutations on the descending arm restored expression, albeit not fully, and effective response to TAV (Fig. 4 C), which is consistent with our previous results for the full-length CaMV leader (32.Dominguez D.I. Ryabova L.A. Pooggin M.M. Schmidt-Puchta W. Fütterer J. Hohn T. J. Biol. Chem. 1998; 273: 3669-3678