Lentiviral genomic RNA are more efficiently translated in absence of their 5′UTR. (A) Translation of HIV-2 rod, HIV-1 pNL4.3 and SIV_Mac Mm 251 gag gene in rabbit reticulocytes lysate in presence of ³⁵S-methionine. For each lentivirus, the RNA template translated comprises the gag entire open reading frame. Constructs with 5′UTR (HIV-1-5′UTR, HIV-2-5′UTR, SIV_MAC-5′UTR) start at the first nucleotide of the genomic RNA, constructs labelled devoid of 5′UTR (HIV-1–AUG₁, HIV-2–AUG₁, SIV_MAC–AUG₁) start with an extra non encoded guanosine before the first AUG codon. None of the transcripts were capped. Translation reactions contain 0.05 µM RNA and were carried out for 60 min as described in ‘Materials and Methods’ section. Proteins translated are referred to as P_x, X being its apparent molecular weight. (B) Schematic representation of HIV-2 rod, HIV-1 pNL4.3 and SIV_MAC Mm 251 gag gene, the initiation codons and the corresponding protein produced.

Leaderless gag mRNA is efficiently transcribed by internal entry of the ribosome. In vitro translation of HIV-2 gag compared to translation of a control gene in presence of ³⁵S methionine. The control gene (5′UTR control) is a chimeric construct comprising the 5′UTR of the human β-globin upstream of the β-galactosidase open reading frame. AUG-control corresponds to the β-galactosidase open reading frame preceded by a single G, SL-5′UTRcontrol and SL-AUG-control, respectively, correspond to 5′UTR control and AUG-control preceded by a stable 13 base pair stem-loop (see ‘Material and Methods’ section). Those controls have been designed so that the protein translated contains the same number of methionines (13) as the full length HIV-2 gag. Therefore the yield in the different protein translated directly reflects their relative efficiencies of translation. HIV-2- 5′UTR is the whole gag gene preceded by its cognate 5′ UTR, HIV-2- AUG1 is the gag open reading frame preceded by a single G, HIV-2-SL-5′UTR and HIV-2- SL–AUG1 are the corresponding constructs preceded by a stable stem-loop. Translation reactions were as described in the ‘Material and Methods’ in presence of 0.05 µM of RNA for 30 min, experiments were repeated three times independently. (A) SDS–PAGE analysis of the different proteins translated in RRL; (B) Quantitative analysis of the different proteins translated in RRL: the results of three independent experiments were quantified, normalised to the total amount of proteins (three isoforms) produced by translation of HIV-2–AUG₁, and averaged. The Y-error bars correspond to ± the standard deviation with the 5′UTR-control used as internal standard. (C) Schematic representation of the different constructs used.

A fourth isoform is translated from alternative initiation codons. (A) Schematic representation of the truncated HIV-2 transcript (HIV-2–AUG₁^*), and the resulting proteins. (B) In vitro translation of the truncated form (+1 to 1641) of HIV-2–AUG₁^* and HIV-2–AUG₁^*alt (HIV-2 mutated on the two alternative codons: CUG_692–694 and UUG_695–697 to UUC_692–694 and UCG_695–697). Translation of HIV-2–AUG₁^* yields four isoforms of 30, 25, 23 and 17 kDa. The mutated HIV-2-AUG₁^* alt no longer supports the translation of the p25 isoform. Both constructs were translated in standard conditions with 0.05 µM of RNA for 30 min, as described in ‘Material and Methods’ section.

Three initiation complexes are present on a single RNA molecule. Sucrose gradient analysis of the initiation complexes were performed as described in ‘Materials and Methods’ section. Initiation complexes formed were then analysed on a 10–50% sucrose gradient. Fractions were collected and counted. (A) HIV-2–AUG₁ was incubated in RRL pre-treated with GMP–PNP (red line) or cycloheximide (blue line) pre-treated RRL. Body labelled mRNA were incubated in RRL pretreated with GMP–PNP (to analyse 48S complexes) or cycloheximide (to analyse 80S complexes). Arrows indicate respectively, the fractions in which sediments one 80S complex (fraction 20), two 80S complexes (fraction 24) and three 80S complexes (fraction 26), as determined by the analysis of the polysome profile (Figure 4B). (B) Capped polyadenylated globin mRNA (red line) was incubated for 10 min in untreated RRL, then 4.8 mM cycloheximide was added, and the mix was analysed on a 10–50% sucrose gradient. As control, a 172-nucleotide-long mRNA containing no AUG was incubated in untreated RRL (control RNA, blue line), and separated on a 10–50% sucrose gradient. Arrows indicate respectively, the fractions in which sediments one 80S complex (fraction 20), two 80S complexes (fraction 24) and three 80S complexes (fraction 26). (C) HIV-2–AUG₁ (blue line) and HIV-2–AUG₁–CUC₂–CUC₃ (red line) were incubated in cycloheximide pre-treated RRL. (D) HIV-2–AUG₁–CUC₂–CUC₃ (blue line) and SL HIV-2–AUG₁–CUC₂–CUC₃ (red line) were incubated in cycloheximide pre-treated RRL.

Toeprinting analysis of the initiation complexes paused on HIV-2 AUG₁–Gag. (A) Toeprinting analysis of the initiation complex present on AUG₁ initiation codon of HIV-2–AUG₁ RNA. Initiation complexes were mapped on three different RNA: the 5′ UTR containing construct, HIV-2–5′UTR, the leaderless RNA, HIV-2–AUG₁ and a version of the leaderless mutated for the 5′ proximal AUG: HIV-2–CUC₁. RNA were reverse transcribed with the labelled primer HIV-2-16 (‘Materials and Methods’ section) in three different conditions: incubated in RRL in standard translation conditions (RRL+, Cycloheximide −, lanes 1,4,7), in RRL pretreated with 4.8 µM cycloheximide (Chx) (RRL+, cycloheximide +, lanes 2,5,8), incubated in water (RRL −, cycloheximide −, lanes 3,6,9). Elongation products were run along with a sequence realised on HIV-2-5′UTR with the HIV-2-16 primer. The bases of the initiation codon are indicated by red stars. Reverse transcriptase stops specific for the AUG containing constructs in presence of cycloheximide are identified on the right side of the autoradiogram. (B) Toeprinting analysis of the initiation complex present on the AUG₂ initiation codon of HIV-2–AUG₁ RNA. Initiation complexes were mapped on HIV-2–AUG₁ RNA. RNA was reverse transcribed with the labelled primer HIV-2-6 (‘Materials and Methods’ section) in three different conditions: incubated in RRL in standard translation conditions (RRL+, cycloheximide −, lane 1), in RRL pretreated with 4.8 µM cycloheximide (RRL+, cycloheximide +, lanes 2), incubated in water (RRL −, cycloheximide −, lanes 3). Elongation products were run along with a sequence realised on HIV-2–AUG₁ RNA using the labelled HIV-2-6 primer (‘Materials and Methods’ section). The bases of the initiation codon are indicated by red stars. Reverse transcriptase stops specific for the AUG containing constructs in presence of cycloheximide are identified on the right side of the autoradiogram. (C) Toeprinting analysis of the initiation complex present on the AUG₃ initiation codon of HIV-2–AUG₁ RNA. Legend is identical to panel 5B.

Toeprinting analysis of the initiation complex present on HIV-1 AUG₁ initiation codon. Same as 4B, but initiation complexes were mapped on HIV-1–AUG₁ RNA. RNA was reverse transcribed with the labelled primer HIV-1-3. Elongation products were run along with a sequence realised on HIV-1–AUG₁ RNA using the labelled HIV-1-3 primer.

Secondary structure model of HIV-2, SIV_Mac and HIV-1 coding region. Colours reflect the accessibility of the nucleotides to the different probes as indicated in the boxed area. ‘Weakly reactive’ refers to reverse transcriptase stops enhanced 2–3-fold compared to the untreated RNA, ‘strongly reactive’ position are enhanced 4 fold and over. The asterisk denotes position protected by at least 2-fold in presence of magnesium. Initiation codons are circled. Pairings are designated by P_n in order of appearance from 5′ to 3′. Segments of a same pairings are numbered P_n−x, and pairings that appeared to be non-conserved between the three virus are designated by P_x1−x2 with x₁ referring to the preceding conserved stem-loop. (A) Secondary model for HIV-1 gag coding sequences, from AUG_336–338 to G₇₈₈. (B) Secondary model for HIV-2 gag coding sequences, from AUG_545–547 to G₉₆₅. (C) Secondary model for SIV_Mac gag coding sequences, from AUG_537–539 to U₉₅₂.

P2 is a major determinant of HIV-2 Gag translation efficiency. Mutant HIV-2–AUG₁ leaderless RNA were translated in RRL for 30 min in presence of 0.05 µM RNA. The control WT RNA and the mutants were translated in parallel in presence of ³⁵S methionine. (A) SDS–PAGE analysis of the translation of HIV-2–AUG₁ mutated in P2: MutP2-5′ harbours mutations in the 5′ strand of P2 stem, MutP2-3′ harbours mutations in the 3′of P2 stem, both destabilise P2 pairing, while mutP2-5′/3′ combines the two sets of mutations which restores the predicted P2 pairing. (B) Schematic representation of the mutants and their effect on P2 pairing. (C) Quantitative analysis of the different proteins translated in RRL: the results of three independent experiments were quantified, normalised to the total amount of proteins (three isoforms) produced by translation of HIV-2–AUG₁, and averaged. The Y-error bars correspond to ± the standard deviation with HIV-2–AUG₁ used as internal standard.

P4 helical domain contains an important determinant for HIV-2 Gag translation. Autoradiogram of the SDS–PAGE of HIV-2–AUG₁^* (WT) and HIV-2–AUG₁^* Δ P4 (MutΔP4) translation in RRL in presence of ³⁵S methionine. The secondary structure of the two constructs is represented on each side of the autoradiogram. Please note that the full length isoform produced from HIV-2–AUG1* ΔP4 misses one methionine as compared to the WT.