Natural variation in Vif: differential impact on APOBEC3G/3F and a potential role in HIV-1 diversification

Abstract

The HIV-1 Vif protein counteracts the antiviral activity exhibited by the host cytidine deaminases APOBEC3G and APOBEC3F. Here, we show that defective vif alleles can readily be found in HIV-1 isolates and infected patients. Single residue changes in the Vif protein sequence are sufficient to cause the loss of Vif-induced APOBEC3 neutralization. Interestingly, not all the detected defects lead to a complete inactivation of Vif function since some mutants retained selective neutralizing activity against APOBEC3F but not APOBEC3G or vice versa. Concordantly, independently hypermutated proviruses with distinguishable patterns of G-to-A substitution attributable to cytidine deamination induced by APOBEC3G, APOBEC3F, or both enzymes were present in individuals carrying proviruses with completely or partly defective Vif variants. Natural variation in Vif function may result in selective and partial neutralization of cytidine deaminases and thereby promote viral sequence diversification within HIV-1 infected individuals.

Figure 1. Naturally Occurring vif Sequence Variation

The phylogenetic relationships among 79 independent vif sequences derived from patients (P1, P2, P3) and viral isolates (V1, V2, V3, V4) were analyzed using the Neighbor-joining method. Seven subtype B reference sequences and a consensus subtype B sequence were also included. A cluster of hypermutated vif sequences found in P3 is indicated. The 40 distinct protein variants selected for functional testing are identified by •. For each patient and isolate, an individual vif consensus sequence was generated and the percent divergence of each vif consensus sequence from the NL4–3 vif referenced is shown.

Figure 2. Activity of Vif Variants from Patients and Viral Isolates

(A) Infectivity of HIV-1 vector particles generated by transient transfection of 293T cells in the presence or absence of fixed amounts of APOBEC3G or APOBEC3F and the indicated amounts of HIV-1 NL4–3 Vif expression plasmids was determined, as described in the Material and Methods. Representative results from one out of three independent experiments are depicted. Infectivity measurements were performed in duplicate assays, and the error bars represent the standard deviation of the RLU values. RLU, relative light units. (B) APOBEC3G neutralization by Vif proteins from LTNPs (P1, P2, P3) and viral isolates (V1, V2, V3, V4). The infectivity of particles generated in the presence of APOBEC3G and each Vif protein is expressed relative to the infectivity of particles generated in the absence of APOBEC3G and Vif. Each Vif protein is identified by its source (e.g., P1, V1) and a variant number (e.g., P1–2). The data represents the average infectivity values of at least three independent experiments, with the error bars showing the standard deviations. (C) Summary of the properties of Vif variants. Independent sequences were defined as alleles that were derived from different PCR reactions or had different nucleotide sequences. Because some changes are synonymous, not all independent sequences encode variant proteins. Vif variants with gross defects (e.g., premature stop codons) as well as those that were found to be inactive in the functional assay are designated “defective Vif.” The overall frequency of inactive Vif proteins is expressed as a percentage relative to the number of independent sequences.

Figure 3. Closely Related Vif Proteins Display Distinct Functional Properties

(A) Summary of the residues implicated as causing Vif defects by comparison of functional and non-functional Vif variants. Amino acid substitutions that occurred exclusively in non-functional Vif proteins are depicted relative to the NL4–3 Vif sequence. Changes identified by * are caused by G-to-A mutations. (B, C) Function of closely related vif alleles assessed by quantitation of the infectivity of particles produced in the presence of APOBEC3G (B) or APOBEC3F (C). Amino acid substitutions in the non-functional partner of the “matched” functional Vif variant are given in parentheses in (B). The dotted line in (C) indicates the level of infectivity observed for the vector generated in the presence of APOBEC3F and in the absence of Vif. The data represent the average infectivity values of at least three independent experiments, with the error bars showing the standard deviations.

Figure 4. Effect of Naturally Occurring Single Amino Acid Substitutions on NL4–3 Vif APOBEC3G and APOBEC3F Neutralization Activity

The HIV-1 vector infectivity generated in the presence of APOBEC3G (A) and APOBEC3F (B) and a fixed dose of NL4–3 Vif mutant was measured. The dotted line in (B) indicates the level of infectivity observed for the vector generated in the presence of APOBEC3F and in the absence of Vif. The data represent the average infectivity as determined by at least three independent experiments, with the error bars showing the standard deviations. (C) Protein expression levels of the NL4–3 Vif mutants were determined by Western blotting of transfected 293T cell lysates. (D) Infectivity of HIV-1 vector generated in the presence of APOBEC3G (filled symbols) and APOBEC3F (open symbols) and varying levels of selected Vif mutants.

Figure 5. Analysis of Gag-pol Sequences Derived from LTNP and Viral Isolates

(A) Graphic representation of the G-to-A changes (compared to HIV-1 NL4–3) present in Gag-pol. Analysis was performed using the HYPERMUT program [48]. (B) Quantitative summary of the observed changes. “# diff” is the number of positions at which the patient sequence differs from NL4–3. “# G-A (%)” represents the absolute number and percentage of all substitutions that are G-to-A changes. The dinucleotide context (GG, GA, GC, GT) reflects the two contiguous bases, with G-to-A mutations occurring in the first position. The numbers of differences (“# diff”) for the APOBEC3G- and APOBEC3F-induced changes in the in vitro assay are given as an average (± standard deviation) of those occurring in 10 to 15 clones of hrGFP (450 nucleotides).

Figure 6. Phylogenetic Relationships and Hypermutation in p17MA Sequences

(A) Neighbor joining tree representing the phylogenetic relationships among 70 independent p17MA sequences derived from patients (P1, P2, P3) and viral isolates (V1, V2, V3, V4). Additional subtype B reference sequences were also included. (B) Graphic representation of the G-to-A changes (compared to HIV-1 NL4–3) present in p17MA sequences of P2 and P3. Analysis was performed using the HYPERMUT program [48]. All possible G-to-A substitutions in the context of the reference sequence (NL4–3) are shown with the dinucleotide context color coded (bottom four sequences).