Selective translation initiation by ribosome jumping in adenovirus-infected and heat-shocked cells.

Abstract

Translation initiation on eukaryotic mRNAs usually occurs by 5'-processive scanning of 40S ribosome subunits from the m7GTP-cap to the initiating AUG. In contrast, picornavirus and some specialized mRNAS initiate translation by internally binding ribosomes. A poorly described third mechanism of initiation, referred to as ribosome shunting or jumping, involves discontinuous scanning by 40S ribosome subunits, in which large segments of the 5' noncoding region are bypassed. Ribosome shunting has only been observed to date on a cauliflower mosaic virus mRNA. In this report we show that the family of adenovirus late mRNAs, which are preferentially translated during infection, use a ribosome jumping mechanism to initiate protein synthesis. Late adenovirus mRNAs contain a common 5'-noncoding region known as the tripartite leader, which confers preferential translation by reducing the requirement for the rate-limiting initiation factor eIF-4F (cap-binding protein complex). Adenovirus inhibits cell protein synthesis largely by inactivating eIF-4F. We show that the tripartite leader directs both 5' linear ribosome scanning and ribosome jumping when eIF-4F is abundant but exclusively uses a ribosome jumping mechanism during late adenovirus infection or heat shock (stress) of mammalian cells, when eIF-4F is altered or inactivated. Shunting is directed by a complex group of secondary structures in the tripartite leader and is facilitated by one or more unidentified viral late gene products. We propose that shunting may represent a widespread mechanism to facilitate selective translation of specialized classes of capped mRNAs, including some stress and developmentally regulated mRNAs, which possess little requirement for eIF-4F but do not initiate by internal ribosome binding.

Construction of regulatable picornavirus IRESes as a test of current models of the mechanism of internal translation initiation

“…There were two motivations behind our attempts to establish a system of regulated IRES-dependent expression: (1) such a system could be useful for coordinated and balanced expression of two or more proteins, and (2) it provides a test for the most generally assumed model of how ribosomes access the major translation initiation site in poliovirus and FMDV+ We have been successful in as far as we have shown that a downregulation of IRES-dependent protein synthesis can be achieved that is strictly dependent on IRP and on the presence of a functional IRE+ However, the success has been only on a limited scale in so far as the dynamic range of the regulation is somewhat limited, which suggests that although our models are correct in broad outline, they may be not quite correct in all details+ Our positive control CAT mRNA translated by the scanning ribosome mechanism behaved as expected+ Addition of sufficient IRP achieved complete inhibition, and the dose response suggested that the mRNA behaved as a uniform population, with no indication of a special subpopulation that was less sensitive to inhibition+ In contrast, when the IRE was inserted into the poliovirus 59 UTR, the results implied two distinct modes of initiation, or two populations of mRNA, one completely resistant to inhibition by IRP, and the other exhibiting a sensitivity comparable to that shown by the positive control CAT reporter+ Attempts to eliminate the resistant subpopulation by ensuring that any ribosomes that initiated at AUG 586 would terminate translation a short distance downstream and thus would not pass through the IRE motif in elongation mode were unsuccessful+ Interestingly, the level of IRES-dependent translation completely resistant to saturating IRP concentrations was very similar to the activity of an IRES with a stable stem-loop (two tandem IREs inserted in opposite orientations) at the same position as the IRE motif+ This suggests that the same level of residual IRES activity persists whether it is an IRP-IRE interaction or a stable stem-loop at nt 629 that blocks scanning past that site+ Experiments in which AUG 586 was placed in-frame with the authentic poliovirus initiation site (AUG 743 ) showed that an IRP-IRE interaction at nt 629 inhibited only the use of AUG 743 , and had no effect on initiation at the upstream AUG 586 + Because the low frequency of initiation at AUG 586 was unchanged, it seems reasonable to infer that the frequency of silent ribosome entry at that site was also unaffected by the IRP-IRE interaction+ This in turn implies that the inhibitory effect of the IRP-IRE interaction on initiation at AUG 743 is due to its blocking the transfer of ribosomes from the entry site to AUG 743 , a transfer that has usually been considered to be by a scanning mechanism+ The fact that some initiation at AUG 743 is completely resistant to an IRP-IRE interaction at nt 629 suggests that either some ribosomes actually enter downstream of that site or that even though ribosome entry may be upstream of nt 629, not all the ribosomes are transferred to AUG 743 by strictly linear scanning, but some may be transferred by a mechanism more akin to ribosome shunting (Yueh & Schneider, 1996La Torre et al+, 1998;Ryabova & Hohn, 2000)+ Doubts as to whether AUG 743 is invariably accessed by a process of strictly linear scanning from an entry site at or near nt 586 have previously been raised on the grounds that insertion of AUG codons or stem-loop structures in the region 586-743 had a smaller negative effect on initiation at AUG 743 than was thought to be expected on the basis of a linear scanning model (Hellen et al+, 1994)+ Previously, a mechanism of scanning had been inferred on largely indirect evidence: (1) no AUG t...…”

“…In the case of the IRE insertion at the most distal site of the FMDV construct (insertion at nt 42 in the numbering system of Fig+ 2, equivalent to a distance of 49 nt from the A of the L ab AUG to the landmark C residue mentioned above), inhibition by increasing doses of IRP was rather ineffective+ At the other extreme, insertion of an IRE at the closest position to the L ab AUG (at nt 15 in the numbering of Fig+ 2, or a distance of 22 nt from the A of the AUG to the landmark C residue) allowed for the greatest range of inhibition by IRP, but at the expense of a considerable reduction in IRES efficiency, possibly because the proximity of the IRE stemloop to the L ab site inhibited ribosome entry at that site+ At the two intermediate positions (nt 24 and 32 in the numbering of Fig+ 2), the effect of the IRP-IRE interaction was quite similar to that observed with the poliovirus system: addition of IRP caused a threefold reduction in IRES-dependent expression, but there appeared to be a residual activity that was completely resistant to even very high IRP concentrations+ Thus, as with the poliovirus system, it appears that although the majority of ribosomes access the L b initiation site by strictly linear scanning from an entry site in the region of the L ab AUG, some ribosomes seem to take an alternative route+ The main evidence in favor of ribosome entry around the L ab site followed either by initiation at that site or scanning to the L b site is the work of Belsham (1992), who showed that if AUGs were inserted between the two sites, the frequency of initiation at the L ab site was unaffected but that at the L b site was reduced+ However, the reduction in yield of product initiated at the L b site was less than the yield of products initiated at the inserted AUGs; the number of ribosomes productively captured by the inserted AUGs was greater than the decrease in initiation frequency at the L b AUG (Belsham, 1992)+ Thus, consideration of quantitative aspects of the data raises some doubts as to whether all the ribosomes that initiated at the L b site reached that site by strictly linear scanning from an entry site in the vicinity of the L ab site+ Similar doubts have been expressed by Lopez de Quinto and Martinez-Salas (1999) on the grounds of results of experiments manipulating the sequence between the two sites, or using antisense oligonucleotides targeted to the region around the L ab site+ It is also worth noting that although the L b initiation site is essential for infectivity, mutation of the L ab site results in a slightly small plaque phenotype rather than lethality (Cao et al+, 1995)+ Thus for both the poliovirus and FMDV systems, the fact that IRP-IRE interactions can regulate IRESdependent translation lends support to the idea that the majority of initiating ribosomes access the main initiation site in each case by entry at a site upstream of the IRE insertion point followed by linear scanning to the functional initiation codon+ On the other hand, the fact that regulation by the IRP-IRE interaction was limited in its dynamic range suggests that some ribosomes take an alternative route to the FMDV L b AUG or poliovirus AUG 743 + This alternative route taken by the minority of ribosomes could either be entry at some other site, which would have to be downstream of the site of IRE insertion, or entry at the originally proposed site followed by ribosome shunting, a form of nonlinear scanning that has been best characterized in the translation of Sendai virus P/C mRNA (La Torre et al+, 1998), cauliflower mosaic virus 35S mRNA (Ryabova & Hohn, 2000), and late adenovirus mRNAs with the tripartite leader 59 UTR (Yueh & S...…”

A stable hairpin preceded by a short open reading frame promotes nonlinear ribosome migration on a synthetic mRNA leader

“…To date, translation initiation by nonlinear ribosome migration has been reported only for a few viral mRNAs+ Viral multifunctional leaders, usually overloaded with RNA processing or packaging signals, could potentially interfere with the classical ribosome scanning+ Shunting has been described in viruses infecting animal cells, such as Sendai virus (Curran & Kolakofsky, 1988, 1989, adenovirus (Yueh & Schneider, 1996) and budgerigar fledgling disease virus, BFDV (Li, 1996), as well as in plant-infecting pararetroviruses, CaMV (Fütterer et al+, 1989(Fütterer et al+, , 1993, and rice tungro bacilliform virus, RTBV FIGURE 4. Expression of the CAT-reporter gene, uORF F, and uORF F::HA under control of structural mutations introduced in the CaMV 35S RNA leader+ A: Resolution of in vitro-translated products in 16+5% SDS-PAGE using the Tricine discontinuous buffer system+ Constructs as depicted in Figure 2 and Table 1; translation efficiency and migration of CAT, as well as migration of uORF F (wt, pMH161, pMH162, pMH163, pMH172, pMH215) or uORF F::HA (wt9, pMH1619, pMH1629, pMH1639, pMH1729, pMH2159) are indicated+ B: Immunoprecipitation of the uORF::HA from pMH163, pMH1639, and pMH2159 translation in vitro+ (Fütterer et al+, 1996)+ Expression of reporter genes under the influence of the shunt-competent CaMV and RTBV pregenomic RNA leaders has been studied in a broad range of experimental conditions: in transiently transfected plant protoplasts (Fütterer et al+, 1993), in vitro (Schmidt-Puchta et al+, 1997), and in planta (Schärer-Hernández & Hohn, 1998)+ Ribosome shunt occurs in plant tissues and cell cultures derived from both CaMV host and nonhost plants, indicating that shunting does not depend on host-specific factors+ In this work, we demonstrate that ribosome shunt occurs on a synthetic mRNA leader just as well as on the viral leader+ Although it has been documented only for a few mRNAs, nonlinear ribosome migration seems to operate in diverse systems and therefore might be considered as a general alternative mechanism for the initiation of translation in eukaryotes+ Nonlinear ribosome migration remains a rather poorly described mechanism for translation initiation in eukaryotes+ Extensive studies on the CaMV pregenomic 35S RNA have shown that ribosome shunt depends exclusively on cis-acting elements in the leader: a lowenergy, elongated hairpin structure preceded by a short ORF terminating upstream (Dominguez et al+, 1998;Hemmings-Mieszczak et al+, 1998;Pooggin et al+, 1998)+ In some of the other cases where shunting has been reported, stable stems also seem to be involved (Füt-terer et al+, 1996;Li, 1996;Yueh & Schneider, 1996)+ Based on the common structural features of the known viral shunt-competent leaders, a synthetic mRNA leader was designed, where the viral structure is precisely replaced...…”

“…Expression of the CAT-reporter gene, uORF F, and uORF F::HA under control of structural mutations introduced in the CaMV 35S RNA leader+ A: Resolution of in vitro-translated products in 16+5% SDS-PAGE using the Tricine discontinuous buffer system+ Constructs as depicted in Figure 2 and Table 1; translation efficiency and migration of CAT, as well as migration of uORF F (wt, pMH161, pMH162, pMH163, pMH172, pMH215) or uORF F::HA (wt9, pMH1619, pMH1629, pMH1639, pMH1729, pMH2159) are indicated+ B: Immunoprecipitation of the uORF::HA from pMH163, pMH1639, and pMH2159 translation in vitro+ (Fütterer et al+, 1996)+ Expression of reporter genes under the influence of the shunt-competent CaMV and RTBV pregenomic RNA leaders has been studied in a broad range of experimental conditions: in transiently transfected plant protoplasts (Fütterer et al+, 1993), in vitro (Schmidt-Puchta et al+, 1997), and in planta (Schärer-Hernández & Hohn, 1998)+ Ribosome shunt occurs in plant tissues and cell cultures derived from both CaMV host and nonhost plants, indicating that shunting does not depend on host-specific factors+ In this work, we demonstrate that ribosome shunt occurs on a synthetic mRNA leader just as well as on the viral leader+ Although it has been documented only for a few mRNAs, nonlinear ribosome migration seems to operate in diverse systems and therefore might be considered as a general alternative mechanism for the initiation of translation in eukaryotes+ Nonlinear ribosome migration remains a rather poorly described mechanism for translation initiation in eukaryotes+ Extensive studies on the CaMV pregenomic 35S RNA have shown that ribosome shunt depends exclusively on cis-acting elements in the leader: a lowenergy, elongated hairpin structure preceded by a short ORF terminating upstream (Dominguez et al+, 1998;Hemmings-Mieszczak et al+, 1998;Pooggin et al+, 1998)+ In some of the other cases where shunting has been reported, stable stems also seem to be involved (Füt-terer et al+, 1996;Li, 1996;Yueh & Schneider, 1996)+ Based on the common structural features of the known viral shunt-competent leaders, a synthetic mRNA leader was designed, where the viral structure is precisely replaced by a short, low-energy hairpin+ This particular stem (Fig+ 5A) has been previously used as a device to efficiently interfere with scanning: too stable to be unwound by the scanning complex, it does provoke the 40S subunit stalling on the 59 side of the hairpin (hp7; Kozak, 1989a)+ Incorporation of the artificial hairpin within the viral flanking sequences indeed fully blocks ribosome scanning, but promotes efficient shunting (Table 1; pMH192, pMH193, and pMH195)+…”

“…In eukaryotic cells, translation initiation usually occurs according to the ribosome scanning mechanism (Kozak, 1989b)+ This classical model assumes that, at first, the 40S preinitiation complex associates with an mRNA at the 59-terminal cap structure and later unidirectionally scans the mRNA leader until the first AUG codon positioned in an appropriate context is encountered, whereupon the 60S subunit joins and protein synthesis starts+ Translation can be strongly affected by cis-acting elements and trans-acting factors that interfere with ribosome scanning in the leader: (1) highly stable stemloop structures mechanically block scanning by stalling 40S subunits on the 59 site of the hairpin (Pelletier & Sonenberg, 1985;Kozak, 1986Kozak, , 1989a; (2) upstream open reading frames (uORFs) reduce initiation downstream due to the apparent low efficiency of reinitiation at subsequentAUG codons (Kozak, 1989b(Kozak, , 1992Hinnenbusch, 1994); (3) RNA-binding proteins associated with cap-proximal binding sites inhibit recognition of the cap structure due to the steric hindrance (Gray & Hentze, 1994;Stripecke et al+, 1994)+ Modified ribosome scanning mechanisms exist, however, that allow relatively efficient translation despite the presence of inhibitory elements in the leader+ Alternative modes of translation initiation have been broadly studied and fall into several categories: leaky scanning (Kozak, 1992), reinitiation (Geballe & Morris, 1994;Hinnenbusch, 1994), internal initiation (Pelletier & Sonenberg, 1988;Kaminski et al+, 1990;Belsham, 1992), and nonlinear ribosome scanning or ribosome shunt (Curran & Kolakofsky, 1988, 1989Fütterer et al+, 1989Fütterer et al+, , 1993Fütterer et al+, , 1996Yueh & Schneider, 1996)+ The cauliflower mosaic virus (CaMV) pregenomic 35S RNA 600-nt leader (Figs+ 1 and 2A) contains all three elements with the potential to strongly inhibit translation: (1) a low-energy elongated hairpin (Ϫ110 kcal/ mol) that has been characterized by enzymatic and chemical probing (Hemmings-Mieszczak et al+, 1997), (2) several uORFs, and (3) a purine-rich cloverleaf structure that interacts with the CaMV coat protein (O+ Guerra-Peraza, M+ de Tapia, T+ Hohn, and M+ HemmingsMieszczak, submitted)+ However, translation initiation downstream of the leader is cap dep...…”