The external domains of the HIV-1 envelope are a mutational cold spot

Abstract

In RNA viruses, mutations occur fast and have large fitness effects. While this affords remarkable adaptability, it can also endanger viral survival due to the accumulation of deleterious mutations. How RNA viruses reconcile these two opposed facets of mutation is still unknown. Here we show that, in human immunodeficiency virus (HIV-1), spontaneous mutations are not randomly located along the viral genome. We find that the viral mutation rate experiences a threefold reduction in the region encoding the most external domains of the viral envelope, which are strongly targeted by neutralizing antibodies. This contrasts with the hypermutation mechanisms deployed by other, more slowly mutating pathogens such as DNA viruses and bacteria, in response to immune pressure. We show that downregulation of the mutation rate in HIV-1 is exerted by the template RNA through changes in sequence context and secondary structure, which control the activity of apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like 3 (A3)-mediated cytidine deamination and the fidelity of the viral reverse transcriptase.

Figure 1. HIV-1 shuttle vector system used for scoring spontaneous mutations.

A scheme of the system used for serial passaging of HIV-1 sequences in the absence of selection is shown. The shuttle vector contains the necessary elements for genomic integration (LTR) and efficient packaging (Ψ element and Rev-responsible element (RRE)), as well as the puromycin resistance gene to enable selection of cells in which integration occurs. The inserted sequence (SEQ, here env or int–vif–vpr) is carried forward by the vector. Translation of the Gag p17 protein starts at position 335 of our genomic RNA but, since a 2-nt insertion was introduced at position 355, the sequence rapidly falls out of frame and the SEQ insert is thus located many stop codons away from this translation initiation site (position 1,950). Translation could not start elsewhere because there is no internal ribosome entry site (IRES). The production of a protein from a spliced version of the genomic RNA is also excluded because, although the major HIV-1 splice donor site is present at position 289 of the vector, there are no splice acceptor sites in the inserted env sequence. Four acceptor sites are present in the int–vif–vpr sequence, but no protein synthesis can occur because of lack of initiating codons. The HIV-1 proteins Gag and Pol and the VSV envelope protein G are instead expressed from two helper plasmids. Initial transfection of the three plasmids is required to recover pseudotyped viruses, which are transduced into fresh cells where they undergo integration. The infection cycle can be restarted by transfecting the two helper plasmids only. After a given number of cycles, the DNA of the insert can be PCR-amplified from puromycin-selected cells, cloned, and sequenced. The inserted sequences contain no known functional cis-acting elements or RNA structures except for the RRE, which is required for nuclear export of viral RNA and is embedded in the env gene. However, this element was provided redundantly from the vector, thus minimizing selection. Recombination between vector (subtype B) and insert (subtype A) RRE copies was checked, and recombinant sequences were discarded from the analysis.

Figure 2. Structure of the HIV-1 envelope and location of spontaneous mutations across env.

(a) Top: map of the env gene (SP: signal peptide; V1–V5: variable regions; C1–C5: more conserved regions between the V regions; FP: fusion peptide and FP proximal region; HR: heptad repeat; DS: disulfide loop; MP: membrane-proximal ectodomain region; TM: transmembrane domain; CD; C-terminal domain). The 1 kb region encoding the extensively glycosylated outer-apical domains of gp120 is boxed. Bottom: glycosylation sites (blue dots), number of B-cell epitopes (pink histogram), and protein sequence variability calculated as the Shannon entropy averaged over a 15-residue sliding window (black skyline). Epitopes and entropy were retrieved from the HIV Immune and Sequence Databases. (b) Structure of the HIV-1 envelope trimer. Each light-blue lobe represents schematically an envelope monomer embedded in the viral membrane (yellow), and superimposed is a surface representation of the crystal structure available for a portion of the trimer including most of gp120 and segments of gp41 (PDB file: 4NCO). The five variable regions are shown in dark blue, and the more conserved segments C2–C4 also belonging to the outer-apical domains are shown in slate. The three gp41 subunits are coloured in grey tones. The various regions are labelled only in one subunit of each gp120 and gp41 for clarity. The structure shown corresponds to the closed conformation found at the surface of free virions. (c) Top: nucleotide substitutions found in env for each of the lines L1–L3 after four infection cycles. Red squares indicate individual mutations. Bottom: mutation rate (μ) averaged over 15-base sliding window (red bars). Significant mutation clusters are indicated with red circles. (d) Mutation rate in the int, vif, vpr and env genes, showing the lower mutation rate in the gp120 outer-apical domains (OA, blue). **t-test: P<0.01; NS: not significant (n=3). Error bars indicate the standard error of the mean. The exact location of each mutation is provided in Supplementary Data 3 (env) and in Supplementary Data 4 (int–vif–vpr).

Figure 3. Mutation-prone A3 targets showing significant depletion in the gp120 OA domains.

A map of the HIV-1 genome is shown on top, with each reading frame shown at a different level. The skyline plots represent the percent abundance of each motif averaged over a 0.5 kb sliding window for 100 sequences of each subtype A, B and C. The grey shaded area around the black line represents the range of variation among subtypes. The dashed box shows the 1 kb region encoding the gp120 OA domains. Genome positions correspond to the HXB2 sequence. To the right of each skyline plot is shown a boxplot of the motif abundance in the gp120 OA domains versus the rest of the genome, based on the 100 individual values from each subtype. The lines in the box indicate the first, second (median) and third quartile. Whiskers above and below the box indicate percentiles 10 and 90, and outlying points are individually plotted.

Figure 4. Effect of RNA structure on the HIV-1 mutation rate.

(a) Mapping of the shuttle vector spontaneous mutations in the env RNA structure model. Nucleotide substitutions are represented with red squares. Regions encoding gp120 and gp41 are separated with a black line. Short dashed lines delimit the following regions in the structure: SP stem (1), region encoding the gp120 OA domains (2), Rev-responsive element (RRE, 3), and a multi stem-loop structure identified in gp41 (SL, 4). For each, the pie chart indicates the fraction of sites forming base pairs, considering only sites for which the pairing status has >90% chance of being conserved across HIV-1 subtypes. (b) In vitro mutation rate of the HIV-1 RT using a stem-like template RNA obtained from PSTVd (red) versus a randomized sequence (blue). Each dot represents an individual replicate and the horizontal bars the mean rate. A list of mutations and their location is provided in Supplementary Data 5 (PSTVd) and Supplementary Data 6 (randomized).