Non-AUG HIV-1 uORF translation elicits specific T cell immune response and regulates viral transcript expression

Abstract

Human immunodeficiency virus type-1 (HIV-1) is a complex retrovirus that relies on alternative splicing, translational, and post-translational mechanisms to produce over 15 functional proteins from its single ~10 kb transcriptional unit. Using ribosome profiling, nascent protein labeling, RNA sequencing, and whole-proteomics of infected CD4 + T lymphocytes, we characterized the transcriptional, translational, and post-translational landscape during infection. While viral infection exerts a significant impact on host transcript abundance, global translation rates are only modestly affected. Proteomics data reveal extensive transcriptional and post-translational regulation, with many genes showing opposing trends between transcript/ribosome profiling and protein abundance. These findings highlight a complex regulatory network orchestrating gene expression at multiple levels. Viral ribosome profiling further uncovered extensive non-AUG translation of small peptides from upstream open reading frames (uORFs) within the 5' long terminal repeat, which elicit specific T cell responses in people living with HIV. Conservation of uORF translation among retroviruses, along with TAR sequences, shapes DDX3 dependency for efficient translation of the main viral open reading frames.

Fig. 1. Transcriptional and translational changes in HIV-1 infected cells.

A Schematic representation of the procedure to monitor transcript abundance and translation in HIV-1 infected cells (Created in BioRender. Ricci, E. (2022) BioRender.com/k47z556). Briefly, SupT1 cells were infected or not (Mock) with HIV-1 (NL4.3 strain) at MOI 5. At 0, 1, 12, 24 and 36 hours post infection (hpi), cells were lysed to recover the cytoplasmic fraction and prepare ribosome profiling and RNA-seq libraries subjected to high-throughput sequencing on the Illumina Hiseq platform (n = 4 independent experiments). Total cell lysates were recovered for mass spectrometry analysis (n = 4 independent experiments). B Scatter-plot of the fold-change (log2) in cytoplasmic RNA-seq and Ribo-Seq of the Mock-infected and HIV-1 infected cells at each time point of infection. Orange dots (“Transcript-level only”) corresponds to genes exclusively regulated at the transcript abundance level. Blue dots (“Translation only”) correspond to genes which display differences in ribosome occupancy while transcript abundance remains unchanged. Green dots (“Translationally regulated”) correspond to genes with significant changes in transcript abundance and significantly further changes, in the same direction, in ribosome occupancy upon infection. Red dots (“Translationally buffered”) correspond to transcripts displaying significant changes in transcript abundance but for which there is compensation at the translational level to maintain unchanged ribosome occupancy levels upon infection. C Gene ontology analysis of differentially expressed genes at each time point.

Fig. 2. Transcriptional and translational changes in HIV-1 infected cells.

A (Left panel) Scheme describing nascent protein labeling using O-propargyl-puromycin (OPP). Briefly, cells are incubated with OPP and fixed in paraformaldehyde before a fluorophore is covalently linked through a click-reaction. Cells are then analyzed by flow cytometry to monitor signal intensity at a single-cell level. B Flow cytometry analysis (n = 3 independent experiments) of OPP signal in SupT1 cells infected with HIV-1, 12, 24 and 36 hpi. C OPP signal in SupT1 cells infected with single-round recombinant HIV-1 virions coding for GFP at 24, 48, 72 and 96 hpi (n = 3 independent experiments). D OPP signal in primary Human CD4 + T cells infected with HIV-1, 48 and 72 hpi and either positive or negative for p24 expression (p24+ or p24-). Results from four independent donors are displayed separately for each time point tested (n = 5 independent experiments). B–D Multiple paired t tests were performed to compare OPP incorporation signals between control and infected samples, * corresponds to a p-value < 0.05. Barplots represent the average value of all biological replicates and error bars correspond to the standard-deviation. Source data are provided as a Source Data file.

Fig. 3. Multi-omics analysis of gene expression changes in HIV-1 infected cells.

A Scatter plots of log2 fold changes in ribo-seq (x axis), RNA-seq (y axis) and protein (color coded) abundance at 12, 24 and 36 hours upon infection (n = 4 independent experiments). B Gene clustering analysis taking into account changes in RNA-seq, ribo-seq and protein abundance at all time points of infection. Mean trajectories (bold lines) and standard-deviation (light colored surfaces) are depicted (top panels) as well as gene ontology analysis (bottom panels) for each cluster.

Fig. 4. Relative and absolute expression of viral transcripts during infection.

A (Left panel) Relative cytoplasmic amount of viral transcripts (the sum of all transcripts at any given time point corresponding to 100%) in SupT1 cells, 1, 12, 24 and 36 hpi. (Right panel) Overall abundance of viral transcripts within RNA-seq libraries displayed in transcripts per million (TPM) (n = 4 independent experiments). B Relative (left panel) and overall (right panel) abundance of viral transcripts in U937 cells bearing a latent HIV-1 provirus integrated in their genome, 3, 6, 9, 12, 15, 18 and 24 hours after induction of proviral DNA expression using PMA and ionomycin (n = 3 independent experiments). Error bars in figures correspond to the Standard Error of the Mean (SEM).

Fig. 5. Translational landscape of viral transcripts.

A HIV-1 genomic structure. B Distribution of RNA-seq (red) and ribosome profiling (blue) reads across the HIV-1 genome in SupT1 cells at 1, 12, 24 and 36 hpi (n = 4 independent experiments). C Translation efficiency of the genomic RNA (across both Gag and Pol coding sequences) at each time point of infection (n = 4 independent experiments). Boxplots are defined as minima, 1st quartile, median, 3th quartile and maxima. D Translation of incoming viral genomic RNAs as tested by infecting cells pre-incubated or not with cycloheximide or puromycin with a recombinant replication-competent HIV-1 virus bearing a GFP sequence within Gag (n = 4 independent experiments). Barplots correspond to the mean value of all biological replicates. A one tailed, paired t test was performed to compare samples. Source data are provided as a Source Data file (Partially created in BioRender. Ricci, E. (2022) BioRender.com/k47z556 and using an Illustration from NIAID NIH BIOART Source https://bioart.niaid.nih.gov/bioart/160). E Translation efficiency of canonical viral mRNAs, 12, 24 and 36 hpi (n = 4 independent experiments). Points correspond to the mean value of all biological replicates and error bars correspond to the Standard Error of the Mean (SEM). F Percentage of Gag-Pol ribosome frameshifting at each time point of infection (n = 4 independent experiments). Boxplots are defined as minima, 1st quartile, median, 3th quartile and maxima.

Fig. 6. Translation initiation sites in viral transcripts.

A Distribution of ribosome P-sites around annotated start and stop codons in all cellular transcripts in harringtonine (Green and Red for each biological replicate) and cycloheximide (Blue) libraries. B Distribution of ribosome profiling reads across the HIV-1 genome obtained from harringtonine and cycloheximide treated cells. C Distribution of ribosome profiling reads in the first 500 nucleotides of the Gag CDS obtained from harringtonine (Green and Red for each biological replicated) and cycloheximide (Blue) treated cells (y axis correspond to reads per million values - RPM). The canonical AUG start codon of Gag (p55 isoform) at position +336, as well as the position of other out-of-frame start codons predicted by Ribocode and lastly the position of the Gag (p40 isoform) start codon at position +759 are annotated in the figure.

Fig. 7. Non-AUG translation initiation sites in the 5’UTR of viral transcripts.

A (Top panel) Distribution of ribosome profiling reads across the HIV-1 genome from infected U937 (Green, n = 3 independent experiments), SupT1 (Red, n = 4 independent experiments) and primary CD4 + T lymphocytes (Blue, n = 2 independent experiments). All y axis correspond to reads per million values - RPM. (Bottom panel) Close-up view of the unspliced HIV-1 5’UTR showing the distribution of ribosome profiling reads and the position of putative non-AUG start codons as well as the open-reading frames of putative peptides produced from non-AUG start codons. The y-axis corresponds to reads per million of sequenced reads (RPM). B Distribution of FLOSS (Fragment Length Distribution Score) values for cellular and viral transcripts computed from ribosome profiling reads from SupT1 (left panel), U937 (middle panel) or primary CD4 T cells (right panel).

Fig. 8. Productive translation from uORFs in viral transcripts and role of DDX3 in alleviating the negative impact of uORFs on translation from downstream main ORFs.

A Ribosome profiling reads across the 5′UTR of unspliced viral mRNAs in U937 lysates incubated with or without 1 M KCl, Ribolace experiments in SupT1 cells, and RPL7a IPs from HEK293T cells (n = 2 independent experiments). Reads are expressed as RPM. B Distribution of ribosome profiling reads on luciferase coding reporter transcripts bearing the wild-type HIV-1 5’UTR (WT 5’UTR), a mutant version in which all non-AUG uORF start codons were mutated to UAA stop codons (no uORFs 5’UTR; uORFs with mutated initiation codons are labeled as discontinued lines), a mutant in which all non-AUG uORF start codons were mutated into AUG codons (AUG uORFs 5’UTR; uORFs with mutated initiation codons are labeled as bold lines) (n = 2 independent experiments). Reads are expressed as RPM. C Relative renilla luciferase activity (normalized against the Globin 5’UTR reporter mRNA) of the different 5’UTR reporter mRNAs upon in vitro translation in the rabbit reticulocyte lysate (n = 3 independent experiments) (D), Western-blot analysis of β-actin (left panel) and DDX3 (right panel) expression in HEK293T cells transfected with plasmids coding for the Cas9 and a sgRNA targeting the GFP sequence (sgGFP) or the DDX3 coding sequence (sgDDX3). (n = 3 independent experiments with similar results). E Relative luciferase activity of 5′UTR reporters in sgDDX3 or sgGFP-transfected HEK293T cells (left), and fold-change in luciferase activity upon DDX3 knockdown (right) (n = 3 independent experiments). F Relative renilla luciferase activity (normalized against the Globin 5’UTR reporter mRNA) from viral reporter mRNAs in which the TAR loop is present or absent, transfected into HEK293T cells in which DDX3 expression was knocked-down (sgDDX3) or not (sgYFP) using CRISPR-Cas9 (n = 3 independent experiments). For panels (C, E and F), a two-tailed Student t-test was performed to assess the differences between the mean values of compared conditions. Barplot values correspond to the mean value of all biological replicates and error bars correspond to the Standard Error of the Mean (SEM). Source data are provided as a Source Data file.

Fig. 9. Identification of polypeptides that initiate specific T cell responses in treated and untreated HIV infected individuals.

(A to D), Peptides potentially encoded by uORFs (see peptides “A02; A03; A06; A07; A08” in panel (A) of Fig. 8) were synthesized and used to screen for T cell responses in PBMCs of HIV-infected individuals. PBMCs from treated (ART) and untreated but aviremic (so called elite controllers, EC) individuals were stimulated with a pool of uORF-derived peptides (POOL) and cultured in the presence of IL-2 and IL-7 cytokines in order to expand peptide-specific memory T cells. On day 7 (not shown) and day 12, T cell responses against the POOL were assessed using IFNγ-Elispot. In addition, on day 12, except for donor EC-3, individual peptides of the pool (A02; A03; A06; A07; A08) were assessed using IFNγ-Elispot. As positive control for T cell expansion and activation, a pool of immunogenic peptides from HIV Env and Gag proteins was also used. A Number of uORF-derived peptides recognized by each HIV-infected individuals. The color code indicates from which uORF the peptides are derived (A02= Blue; A03= Orange; A07&A08= Red). B, C and D left panels, detailed IFNγ-Elispot data from the 3 individuals presenting T cell responses, expressed as spot forming units (SFU) per million PBMCs; right panels pictures of the raw data from the Elispot plates in technical triplicate (n = 3 technical replicates) (B and D), or duplicate (n = 2 technical replicates) (C), POOL (−) and POOL (+): PBMCs expanded with the pool of uORF-derived peptides but restimulated on the day of the Elispot with medium or the POOL, respectively. A02; A03; A06; A07 and A08 name of individual uORF-derived peptides used for re-stimulation. HIV (-) and HIV ( + ): PBMC expanded with the pool of Env- and Gag-derived peptides but re-stimulated for the Elispot assay with medium or the same pool of HIV peptides, respectively. Responses were considered positive when IFNγ production was superior to 20 spots/106 PBMCs and at least twofold higher than background production from cells re-stimulated with medium (dotted lines). SAT: saturated signal, where counts cannot be estimated due to overwhelm IFNγ secretion by activated T cells. Source data are provided as a Source Data file. Error bars correspond to the standard deviation of the mean.