Manuscript 2004-02-15590C, revised
File name: Jacobson-supplementary data-FINAL.doc
Format: Microsoft Word 2000
Toeprint analyses of normal and premature termination. Whereas toeprints were readily detectable from premature termination codons in the absence of CHX in wild-type extracts (e.g., Fig. 1b), no toeprints were obtained under the same conditions when the normal termination codons of the Fusion and ADE2 RNAs were analyzed (Supplementary Fig. 1a). These results suggest that normal termination is a more efficient process than premature termination. The efficiency of normal termination could, however, be reduced sufficiently in sup45-2 extracts to allow detection of toeprints from normal terminators (Fig. 3c).
The distance limits of retroreinitiation were assessed by constructing nonsense-containing RNAs that harbor AUG codons 21 or 32 nt upstream of the premature terminator (Supplementary Fig. 1b). Control RNAs (Fusion-21 and Fusion-32), containing the -21 and -32 AUGs, respectively, but not the premature stop codon, were also constructed. Toeprint analyses revealed strong bands, sensitive to cap analog, in the UAA-M-21 and UGA-M-21 RNAs at positions 16-18 nt downstream of the –21 AUG (lanes 3-6, see asterisks) and other bands at comparable positions in the UAA-M-32 and UGA-M-32 RNAs (lanes 9-12). Additional bands ~19 nt downstream of the –32 AUGs (lanes 9 and 11), bands upstream of the -21 toeprints (lanes 3 and 5, arrow a), and bands downstream (lanes 3, 5, 9, and 11, arrow b) may arise from translocating ribosomes, ribosomes scanning backwards that collide with those already engaged at the initiation codon, or the binding of non-ribosomal factors to the RNA. The nonspecific band indicated by arrow c in the UGA-M-21 and UGA-M-32 samples may reflect the binding of factors specific to AUGs at least 21 nt upstream of the UGA since this band is absent in the UGA, UGA-M, UGA-DS and UGA-M-DS RNAs (data not shown). It is important to note that none of the aforementioned toeprint bands are detected with the Fusion-21 and Fusion-32 control RNA (lanes 1, 2, 7, and 8). These experiments thus demonstrate that yeast ribosomes can reinitiate in vitro at least 32 nt 5' to a premature terminator.
The experiments of Supplementary Fig. 1c compare the relative efficiencies of upstream and downstream reinitiation. Analysis of the UAA-DS RNA (containing the –11 AUG and another AUG 5 nt downstream of the stop codon) showed only the cap analog-sensitive +6 toeprint (lanes 3 and 4). A derivative of this RNA lacking the –11 AUG (UAA-M-DS RNA) exhibited a new translation-dependent toeprint signal 24-25 nt downstream of the stop codon (lanes 1 and 2) that corresponds to ribosomes stalled with their P site on the +5 AUG. Similar results were obtained with derivatives of the UGA RNA. Mutation of only the –11 AUG in the No-11-UGA RNA led to elimination of the +6 toeprint and maintenance of the +17 toeprint (lanes 5 and 6). Analysis of an RNA containing the +5 AUG as well as the upstream -11 and -1 AUGs (UGA-DS RNA) showed only the +6 and +17 toeprints (lanes 7 and 8). However, mutation of both upstream AUGs but inclusion of a +5 AUG (UGA-M-DS RNA) eliminated the retroreinitiation toeprints but generated a translation-dependent signal 24-25 nt downstream of the UGA stop codon (lanes 9 and 10). Since toeprints are only detected at downstream AUGs when proximal upstream AUGs are eliminated, post-termination ribosomes exiting premature termination codons must have a propensity for backwards scanning.
Ribosomes can reinitiate translation at AUG codons upstream or downstream of the stop codon. The propensity for backwards scanning has been independently confirmed by analyses of luciferase activity obtained in vitro from RNAs harboring in-frame LUC fusions to upstream or downstream AUGs (Supplementary Fig. 2). Six constructs were made in which the LUC ORF was in frame with AUGs at either -11 in the Fusion, UAA, UAA-W, and UAA-DS RNAs or at +5 in the UAA-M-DS and UAA-DS constructs (designated Inf-Fusion, Inf-UAA, Inf-UAA-W, Inf-UAA-DS, Inf-UAA-M+5D and Inf-UAA+5DS, respectively; Supplementary Fig. 2). A seventh construct, used as a control, contained the UAA stop codon, but lacked the –11 and +5 AUGs (Inf-UAA-M). Constructs containing the -11 AUG in frame with the LUC ORF were made by insertion of an A at position 4 downstream of the stop codon, or its equivalent position in the case of the control Inf-Fusion RNA. Constructs containing the +5 AUG in frame with the LUC ORF were made by insertion of an A at position 10 downstream of the stop codon, plus a deletion of the G at position 4 in the case of the Inf-UAA+5DS. Translation of these RNAs in vitro showed that ribosomes effectively reinitiate at the upstream -11 AUG, or at the downstream +5 AUG, and are able to produce luciferase activity that is dependent on the presence of both the termination codon and flanking AUG codons (Supplementary Fig. 2). Consistent with the toeprint data of Supplementary Fig. 1c, ribosomes exiting the premature stop codon prefer retroreinitiation to downstream reinitiation (compare Inf-UAA-DS to Inf-UAA+5DS). Moreover, as expected from the toeprinting results of Fig. 1f, the in vitro translational yield of all RNAs is markedly reduced in upf1Δ extracts (Supplementary Fig. 2).
Toeprint analyses in sup45-2 extracts. Results from Supplementary Fig. 3a show that, regardless of the site of post-termination reinitiation, all nonsense-containing RNAs yielded almost exclusively toeprints corresponding to ribosomes stalled with the stop codon in their A sites when translated in sup45-2 extracts. RNAs having weak terminators yield strong translation-dependent toeprints 12-14 nt downstream of the premature stop codon (lanes 5, 6, and 13-22 and 25-28) and RNAs with strong terminators yield faint bands at the same position (lanes 3, 4, 7-12). No comparable toeprints are obtained with the Fusion RNAs (lanes 1, 2 and 23, 24). Unlike wild-type extracts, those from sup45-2 cells show no differences with RNAs that do or do not contain upstream or downstream AUGs (Fig. 2 and Supplementary Fig. 1b and c vs. Supplementary Fig. 3a) and are thus incapable of scanning either backward or forward after premature termination. The epistasis of the toeprints obtained in sup45-2 extracts to those obtained in wild-type extracts provides additional evidence that the aberrant toeprints do not arise from leaky scanning and suggests that, prior to any reinitiation event, a premature stop codon in the ribosomal A site must be recognized by eRF1 and trigger peptide hydrolysis in the adjoining P site15,18. Interestingly, translation of the mini RNAs in sup45-2 extracts in the presence of CHX did not yield detectable cap analog-sensitive toeprints, reinforcing the notion that normal termination is different from premature termination (Supplementary Fig. 3b).
Tethered Pab1p stabilizes nonsense-containing mRNAs. The in vivo stability of mRNAs bearing an MS2 coat protein binding-site 3' to premature terminators in the CAN1 and PGK1 mRNAs was assessed in cells expressing an MS2-Pab1p fusion. Pab1p tethered 37-73 nt 3' to premature UAA, UGA, or UAG codons promoted 5- to 11-fold increases in mRNA abundance and stability (Supplementary Fig. 4a-b). MS2 dimer tethered at the same position of the PGK1 mRNA had no effect on mRNA abundance (Supplementary Fig. 4a).
MS2-Pab1p-mediated stabilization was: a) specific for mRNA containing the MS2-binding site since the CYH2 pre-mRNA, an NMD substrate, was not detectable on these blots (Supplementary Fig. 4b), b) not attributable to non-specific effects on translation since the average number of ribosomes associated with the PGK1-MS2 mRNA does not change in the presence or absence of tethered Pab1p (Supplementary Fig. 4c), and c) partially manifested by tethered fragments of Pab1p (Supplementary Fig. 4d). The ability of the latter fusion proteins, including those containing only RRMs1-4 or just the C-terminus of Pab1p, to partially stabilize the PGK1-MS2 mRNA suggests that distinct domains of Pab1p may contribute to the stabilizing effects.
The experiments of Fig. 4b demonstrate that tethered MS2-Pab1p is able to recruit Sup35p to the PGK1-MS2 mRNA. To evaluate the specificity of this IP, we performed an independent experiment using MS2-Pab1p lacking the Sup35p-interacting domain30. Underscoring the specificity of the original IP, this experiment failed to detect co-immunoprecipitated Sup35p (Supplementary Fig. 4e). As an additional test for the specificity of Sup35p recruitment, inactivation of Upf1p was utilized to provide an independent means of stabilizing the nonsense-containing mRNA. Western blotting of extracts from upf1Δ cells demonstrated that this alternative mode of stabilizing the PGK1-MS2 mRNA did not enhance recovery of endogenous Sup35p in an MS2-Sxl IP (Supplementary Fig. 4f), thereby implicating a specific MS2-Pab1p:Sup35p interaction.
Supplementary figure legends
Supplementary Fig. 1. Toeprint analyses of normal and premature termination. (a) No toeprints are detectable at normal terminators of the Fusion or ADE2 RNAs in wild-type extracts in the absence of CHX. Sequence ladders on the left side of each panel correspond to the normal terminator region of each mRNA. (b) Ribosomes can migrate to AUG codons 21 or 32 nt upstream of premature stop codons. The Fusion-21 and Fusion-32 constructs, used as controls, lack PTCs, but contain the -21 or -32 AUGs, respectively. Sequences indicate mutational changes to the Fusion, UAA-M, or UGA-M RNAs (see Figs. 1a and 2 for original sequences). (c) Reinitiation at upstream AUGs is favored over downstream reinitiation. Toeprint analyses of UAA RNA derivatives containing an additional downstream AUG 5 nt from the PTC with (UAA-DS) or without (UAA-M-DS) the -11 AUG, and UGA RNA derivatives lacking the –11 AUG (No-11-UGA), containing an additional AUG 5 nt 3’ to the stop codon (UGA-DS), or containing the +5 AUG, but lacking the –11 and –1 AUGs (UGA-M-DS). Numbers under the AUGs denote position relative to the PTCs and numbers on the right of each panel indicate the positions of toeprints relative to PTCs.
Supplementary Fig. 2. Ribosomes can reinitiate translation at AUG codons upstream or downstream of the stop codon. Translation reactions in wild-type (n=4) or upf1∆ (n=3) extracts were programmed with the respective in-frame LUC fusion (Inf) mRNAs and analyzed for luciferase activity. Error bars in graphs depict standard deviations. Sequences of pertinent segments of the Inf constructs are shown.
Supplementary Fig. 3. Toeprint analyses in sup45-2 extracts. (a) RNAs translated in sup45-2 extracts in the presence of CHX yield only +12 toeprints. UAA and UGA RNAs and their derivatives with or without AUG codons 5’ or 3’ to the respective terminators (see Fig. 1a and 2, and Supplementary Fig. 1b and c) were subjected to toeprint analysis. UAA-M-W RNA is a construct containing the UAA codon in a weak termination context (CAA UAA CAA)12. The sequence ladders depicted are derived from the UAA template. (b) mini RNAs translated in sup45-2 extracts in the presence of CHX yield no cap analog-sensitive toeprints.
Supplementary Fig. 4. Tethered Pab1p stabilizes nonsense-containing mRNAs. (a) Cellular steady-state levels of PTC-containing CAN1 (lanes 1-4) and PGK1 (lanes 5 and 6) mRNAs with or without tethered Pab1p. Cells expressed UAA or UGA alleles of CAN1/LUC (see Fig.1a) or the mini-PGK1 UAG allele19, all of which contained insertions of MS2 coat protein binding sites. Simultaneous expression of MS2-Pab1p fusion protein or MS2-dimer is indicated. Levels of all transcripts were determined by northern blotting. The * denotes decay intermediates generated by 5’ to 3' degradation up to the site of bound MS2-Pab1p. (b) Stabilization of the PGK1-MS2 mRNA is dependent on the presence and proximity of tethered Pab1p. Half-lives of the PGK1-MS2, PGK1-3’-MS2, and CYH2 mRNAs were assessed by northern blotting at different times (min) after transcription inhibition19. Cells expressed the mini-PGK1 allele19 containing MS2 coat protein binding sites inserted 37 nt (PGK1-MS2) or 164 nt (PGK1-3’-MS2) 3’ of the premature UAG. Simultaneous expression of MS2 fusion proteins is indicated. (c) Tethered Pab1p does not affect translation of the PGK1-MS2 mRNA. Extracts from cells expressing the PGK1-MS2 mRNA with (right) or without (left) co-expressed MS2-Pab1p were fractionated on sucrose gradients and subsequently analyzed by northern blotting. Top: OD254 profiles, with sedimentation proceeding from right to left. Bottom: northern analysis of the gradient fractions for PGK1-MS2 mRNA and control CYH2 mRNA. (d) Tethered fragments of Pab1p partially stabilize the PGK1-MS2 mRNA. MS2 fusion proteins encompassing Pab1p full-length, RRM1-4 (residues 1-405); C-terminus only (residues 406-577); and RRM2 through the C-terminus (residues 119-577) were expressed and assessed as in part (a) for their effects on the steady-state level of PGK1-MS2 mRNA. (e) Sup35p does not co-IP with MS2-Pab1p lacking the Sup35p-interacting domain. MS2-Pab1p harboring only PAB1 RRM1-4 (see part [c]) was co-expressed with PGK1-MS2 mRNA and analyzed for co-IP of Sup35p as in Fig. 4c. (f) Sup35p does not co-IP with mRNPs stabilized by deletion of UPF1. Extracts of upf1Δ cells expressing PGK1-MS2 mRNA and MS2-Sxl fusion protein were immunoprecipitated with anti-MS2 antibodies and characterized as in Fig. 4b.