5'-untranslated regions with multiple upstream AUG codons can support low-level translation via leaky scanning and reinitiation

Abstract

Upstream AUGs (uAUGs) and upstream open reading frames (uORFs) are common features of mRNAs that encode regulatory proteins and have been shown to profoundly influence translation of the main ORF. In this study, we employed a series of artificial 5'-untranslated regions (5'-UTRs) containing one or more uAUGs/uORFs to systematically assess translation initiation at the main AUG by leaky scanning and reinitiation mechanisms. Constructs containing either one or two uAUGs in varying contexts but without an in-frame stop codon upstream of the main AUG were used to analyse the leaky scanning mechanism. This analysis largely confirmed the ranking of different AUG contextual sequences that was determined previously by Kozak. In addition, this ranking was the same for both the first and second uAUGs, although the magnitude of initiation efficiency differed. Moreover, approximately 10% of ribosomes exhibited leaky scanning at uAUGs in the most favourable context and initiated at a downstream AUG. A second group of constructs containing different numbers of uORFs, each with optimal uAUGs, were used to measure the capacity for reinitiation. We found significant levels of initiation at the main ORF even in constructs containing four uORFs, with nearly 10% of ribosomes capable of reinitiating five times. This study shows that for mRNAs containing multiple uORFs/uAUGs, ribosome reinitiation and leaky scanning are efficient mechanisms for initiation at their main AUGs.

Figure 1

Flow cytometry analysis of constructs with one or two uAUGs of various strengths and no in-frame stop codon upstream of the main AUG in transiently transfected mammalian cells. (A) Schematic of constructs containing only one uAUG with various strengths. The uORF specified by these uAUGs terminates 1 nt downstream of the GFP start codon. The 20 nt leader sequence is indicated by an arrow and includes 14 nt provided by the pEGFP-N1 vector as shown in the insert box. The uAUG is underlined and the beginning of the GFP ORF is indicated by a bent arrow. The uAUG contexts in each construct differ from each other in positions –3 and +4 as indicated in bold and highlighted. The sequence between the uAUG and main AUG is shown in the insert box. (B) Schematic of constructs containing two uAUGs within the 5′-UTR. The two uAUGs are underlined. The first uAUG has a fixed strength (G^–3) and the second uAUG has differing strengths as indicated. The sequence between the second uAUG and main AUG is shown in the insert box. (C) Graphical representation of relative GFP intensities determined by flow cytometry of the constructs shown in (A) and (B). The values were normalized using a control construct containing the same sequence as uATG_e but with the uAUG mutated to UUG. ANOVA statistical analysis confirmed significant differences (P < 0.001) between all constructs within each grouping.

Figure 2

Flow cytometry analysis of constructs with one uAUG in the most optimal context and a longer leader of 94 nt and constructs containing a stem–loop near the 5′-cap site in transiently transfected mammalian cells. (A) Schematic showing constructs containing one uAUG in the most optimal context with a short leader of 20 nt (uATG_b), a leader of 94 nt (uATG_bL) or containing a stem–loop 14 nt downstream of the 5′-cap site (uAUG_bL-SL). The control constructs lacking uAUGs, with (pEGFP-N1-SL) and without (pEGFP-N1) stem–loops, are shown. The various leader lengths between the 5′-cap and first AUG in these constructs are indicated by arrows. The uAUG is underlined and the beginning of the GFP ORF is indicated by a bent arrow. The sequence between the uAUG and main AUG is shown in the insert box. (B) GFP fluorescence intensity histograms compiled from the analysis of 20 000 cells per sample for each construct in transfected HeLa cells. It shows that construct uATG_bL expresses GFP fluorescence levels that are similar to construct uATG_b. Both stem–loop constructs (uAUG_bL-SL and pEGFP-N1-SL) display GFP fluorescence levels that are similar to the untransfected control. (C) Graphical representation of GFP intensities equating to translation efficiency for each construct in transfected HeLa cells.

Figure 3

Northern analysis of mRNA levels in transfected cells. (A) Total RNA was prepared from cell cultures transfected with constructs uATG_b, uATG_d, a positive control (a construct with the same leader length but lacking a uAUG) and a negative control (untransfected control) as indicated on top of each lane. Similar intensities were detected for the positive control and uATG_b and uATG_d constructs when adjusted for RNA loading levels (28S and 18S RNA). (B) Northern hybridization of transcripts produced by constructs pEGFP-N1, pEGFP-N1-SL, uATG_b, uATG_bL and uATG_bL-SL transfected into HeLa cells. Similar GFP mRNA levels were seen for all transfected constructs.

Figure 4

Flow cytometry analysis of constructs containing multiple uORFs with weak uAUGs and a 12 nt intercistronic spacer between ORFs in transiently transfected mammalian cells. (A) Schematic showing the 5′-UTR structure of constructs containing differing numbers of uORFs with weak uAUGs. The 20 nt leader sequence is indicated by an arrow and includes 14 nt provided by the pEGFP-N1 vector. The insert box shows the context of uAUGs, the sequences of the uORFs, the intercistronic spacer between uORFs and the spacer between the ultimate uORF and the main ORF. The beginning of the GFP ORF is indicated by a bent arrow. The control construct 1uORF^TTG has the same sequence and length as 1uORF but with the uAUG mutated to uUUG. (B) Graphical representation of GFP intensities determined by flow cytometry of constructs shown in (A). All values are relative to construct 1uORF^TTG.

Figure 5

Flow cytometry analysis of constructs containing multiple uORFs with the Kozak consensus sequence flanking uAUGs and a 12 nt spacer between ORFs in transiently transfected mammalian cells. (A) Schematic showing the 5′-UTR structure of constructs containing differing numbers of uORFs with ‘strong’ uAUGs. Construct design was as for Figure 4A except that the Kozak sequence flanked uAUGs. The insert box shows the context of uAUGs, the sequences of the uORFs, the intercistronic spacer between uORFs and the spacer between the ultimate uORF and the main ORF. (B) Graphical representation of GFP intensities determined by flow cytometry of constructs shown in (A). All values are relative to construct 1uORF^TTG.

Figure 6

Constructs containing multiple uORFs with the Kozak consensus sequence flanking uAUGs and a 52 nt spacer between ORFs and their analysis in transiently transfected mammalian cells. (A) Schematic showing the 5′-UTR structure of constructs containing differing numbers of uORFs with ‘strong’ uAUGs and a longer intercistronic spacer. Construct design was as for Figure 5A except with longer spacers between uORFs. Each construct contains a 53 nt spacer between the main AUG of the GFP reporter and the preceding uORF. The insert box shows the context of uAUGs, the sequences of the uORFs, the intercistronic spacer between uORFs and the spacer between the ultimate uORF and the main ORF. (B) Graphical representation of GFP intensities determined by flow cytometry of constructs shown in (A). All values are relative to construct 1uORF^TTG.