Evolution of Human Immunodeficiency Virus Type 1 in a Patient with Cross-Reactive Neutralizing Activity in Serum

The high mutation rate of HIV-1, which is the result of rapid replication dynamics (9, 36) in combination with an error-prone reverse transcriptase and lack of proofreading, contributes to its high genetic variability and the continuous emergence of new viral variants (22, 26). The genetically diverse viral quasispecies allows HIV-1 to adapt to its host environment by facilitating the escape from the host immune responses (1, 3, 4, 6, 14, 20, 32, 33, 35) and the selection of viral biological properties such as coreceptor use and replication capacity (10, 11, 12, 13, 23, 24, 31). The envelope glycoprotein of HIV-1 (Env) is highly variable, with a sequence variability of up to 10% within a single individual (7, 16, 30). The random generation of single point mutations in the viral envelope gene, together with insertions and/or deletions, facilitates escape from neutralizing antibodies by altering or shielding the antibody epitope. Viral escape variants are rapidly selected due to the humoral immune pressure eliminating the neutralization-sensitive virus variants and thereby changing the genetic composition of the viral population (3, 4, 15, 19, 25, 28, 32, 33, 35).

Recently, we reported on the intrapatient comparison of longitudinally obtained HIV-1 envelope sequences from viral RNA in serum (serum RNA), replication-competent HIV-1 clonal variants (CV) isolated from peripheral blood mononuclear cells (PBMC), and proviral DNA from PBMC (PBMC DNA) (5). In one of these patients, who had a typical clinical course of infection (Fig. 1 A), we studied in more depth the virus population evolved in two separate lineages: viral population 1 (VP-1) and viral population 2 (VP-2). Separate lineages of HIV-1 variants within one patient have been observed previously for coexisting CCR5 (R5)- and CXCR4 (X4)-using HIV-1 variants (2, 34). In the patient we studied here, R5 variants were present in both lineages while X4 variants were found only in VP-2. VP-1 was constituted by the majority of the viral serum RNA sequences from the first two time points studied (47 and 68 months postseroconversion [post-SC]) and two PBMC DNA sequences from the third time point (83 months post-SC) and lacked progeny at later stages of infection. VP-2, initially made up mainly of viral sequences obtained from PBMC, did have viral progeny at later time points in both serum and PBMC (Fig. 1B).

To understand the mechanisms contributing to the negative selection of VP-1, which formed the majority of the viral population in serum between years 4 and 7 post-SC, we compared the molecular and phenotypic properties of the initially coexisting HIV-1 populations that did (VP-2) or did not (VP-1) successfully generate progeny virus that persisted in peripheral blood.

From longitudinally obtained blood samples (9 years of seropositive follow-up, 4 different time points; Fig. 1A), a total of 29 gp160 env sequences were generated from serum RNA, 37 env sequences were generated from PBMC DNA, and 19 env sequences were generated from CV as described previously (5) (GenBank accession numbers GU455456 to GU455475 and HQ231027 to HQ231090).

Differences between the amino acid sequences of viral variants from VP-1 and VP-2 were found mainly, although not exclusively, in the first and second variable loops (V1V2) (Fig. 2A) and the third constant region of env. Some of those mutations altered the number of potential N-linked glycosylation sites (PNGS), resulting in a significantly higher number of PNGS in viruses from VP-2 than in viruses from VP-1, in particular in the V1V2 region (Fig. 2B). Additionally, the gp160 env sequence of VP-2 viruses was significantly longer than the gp160 env sequence from VP-1 viruses, even when the analysis was restricted to R5 variants in both virus populations (Fig. 2B). Phylogenetic analysis of HIV-1 sequences from another patient also revealed a small second population in serum RNA at different time points that lacked progeny at later time points (5). The env sequences of the viruses in this small second cluster were shorter and had fewer PNGS than the env sequences of the viral population that persisted over time (data not shown). Increases in envelope length and the number of PNGS have previously been reported to decrease the neutralization sensitivity of HIV-1 (3, 4, 8, 19, 21, 28, 29, 32, 35). Therefore, we tested the sensitivity of several virus variants from both VP-1 and VP-2 to neutralization by four broadly neutralizing antibodies (BrNAbs; b12, 2G12, 2F5, and 4E10), HIVIG (a pool of purified IgG obtained in 1995 from chronically HIV-1-infected individuals), and autologous serum.

Viral envelope gp160 sequences obtained from serum RNA, from CV, and from PBMC DNA (see the legend to Fig. 3 for the exact numbers per time point) were cloned into the viral NL4.3Δenv backbone to create replication-competent chimeric viruses through recombination as described previously (33). These env-NL4.3 chimeras were tested for sensitivity to neutralization by 3-fold serial dilutions of HIVIG (starting concentration, 1,500 μg/ml) and the four BrNAbs (starting concentration, 25 μg/ml) in triplicate and 2-fold serial dilutions of autologous serum obtained at 68 and 83 months post-SC (starting dilution, 1:50) in triplicate (33).

The neutralization sensitivities of BrNAbs 2F5 and 4E10 varied widely, both within and between the env-NL4.3 chimeras with either the gp160 env from VP-1 (VP-1 chimeras) or VP-2 viruses (VP-2 chimeras) (Fig. 3). Both VP-1 and VP-2 chimeras were relatively resistant to 2G12 (data not shown); however, the VP-1 chimeras showed a higher resistance to BrNAb b12 than the VP-2 chimeras did (Fig. 3). Conversely, VP-1 chimeras were more sensitive to neutralization by autologous serum and HIVIG than VP-2 chimeras (Fig. 3). Remarkably, VP-1 chimeras from the second time point were sensitive to contemporaneous serum, which suggests that these viruses were unable to fully escape autologous neutralization. This implies that the neutralizing antibody response of this patient may have played a role in the negative selection of those viral variants. There was no difference in neutralization sensitivity to the BrNAbs and autologous serum between the R5 and X4 variants or between the VP-2 chimeras obtained from different viral sources.

Serum from this individual was previously demonstrated to have cross-reactive neutralizing activity, defined as the ability to neutralize HIV-1 variants from different clades, which was first detected 23 months post-SC (33). Cross-reactive neutralizing activity is considered to be directed against epitopes in conserved and thereby potentially essential regions of the virus. Given that the higher sensitivity to autologous serum neutralization of the viral variants that did not persist in peripheral blood (VP-1) coincided with shorter env V1V2 regions with fewer PNGS, cross-reactive neutralizing activity seems to select for virus variants with longer variable loops that carry more glycans, which supports the occlusion of targeted epitopes in the conserved regions as a mechanism of viral escape (3, 4, 19, 25, 27, 28, 32, 35). The epitope specificities of the cross-reactive neutralizing activity in this patient are currently being studied and may help to elucidate whether viral escape also occurs by mutations of specific residues in the conserved epitopes themselves. Interestingly, and as observed before (4, 33), a decrease in env length and the number of PNGS was observed for VP-2 at later stages of infection and most likely coincided with fading autologous neutralizing activity.

Next we analyzed the replication kinetics of the VP-1 and VP-2 chimeras in a PBMC-based replication assay as described previously (33). Interestingly, VP-1 chimeras showed slower replication kinetics than VP-2 chimeras (Fig. 4). This suggests that the combination of slower replication kinetics and higher sensitivity to autologous neutralizing activity was detrimental to the persistence of VP-1. The slower replication kinetics of VP-1 may have prevented those viruses from acquiring the mutations that would have allowed their escape from the broadly neutralizing humoral immune response of this patient. Alternatively, or in addition, mutations required for the escape from neutralizing humoral immunity in the background of VP-1 virus variants may have come at a larger fitness cost to the virus, resulting in a viral population with lower replicating capacity which consequently was outcompeted by VP-2.

The half-life of HIV-1 in plasma is about 1.3 h (18), indicating that the virions that were present in serum must have been produced shortly before we detected them. We failed to detect, however, the cells that produced some of those viruses in vivo, as VP-1 viruses were not represented in CV and were only in low abundance in PBMC DNA, suggesting that the cellular source of VP-1 was outside the peripheral blood compartment. The sensitivity of VP-1 viruses to autologous neutralizing antibodies may have contributed to their vulnerability in the cell-free state and may have interfered with their infection of PBMC, halting their survival in peripheral blood. The persistence of these viruses, even throughout the third time point analyzed, suggests that they were indeed continuously produced, probably spreading through cell-to-cell transmission outside the peripheral blood compartment to avoid the cell-free state (17). However, we did not detect any VP-1 virus variants after 83 months post-SC. This patient developed AIDS with CD4⁺ T-cell counts below 200/ml at 84 months post-SC (Fig. 1A). At this stage of disease, target cell availability may become limiting and viral properties such as replication capacity may have an even greater impact on viral survival. The inability of VP-1 viral variants to compete with the more fit VP-2 viruses for limited target cells may explain their inability to persist in peripheral blood after progression to AIDS.

Our present findings suggest that virus variants with different sensitivities to neutralizing antibody pressure and with different replication fitness may coexist for a certain amount of time but that a changing environment in the host with progression of disease may favor the persistence of HIV-1 variants with the most fit and neutralization-resistant phenotype. The initial presence of coexisting, independently evolving HIV-1 populations with the ultimate persistence of only one virus population in two patients studied may imply that our findings potentially reflect the common adaptation of HIV-1 in response to host selective pressures. Our detailed collection of patient materials may have provided us with the unique opportunity to reveal in detail the phenotypic characteristics of an HIV-1 quasispecies and its interplay with its host's humoral immune response.

This work was financially supported by the Netherlands Organization for Scientific Research (NOW; grant 918.66.628), the European Community's Sixth Framework Programme Europrise (FP6/2007-2012; grant 037611), and the European Community's Seventh Framework Programme NGIN (FP7/2007-2013; grant 201433). The Amsterdam Cohort Studies on HIV infection and AIDS, a collaboration of the Amsterdam Health Service, the Academic Medical Center of the University of Amsterdam, the Sanquin Blood Supply Foundation, the University Medical Center Utrecht, and the Jan van Goyen Clinic, are part of the Netherlands HIV Monitoring Foundation and financially supported by the Center for Infectious Disease Control of the Netherlands National Institute for Public Health and the Environment.

The funding organizations had no role in study design, data collection and analysis, the decision to publish, or preparation of the manuscript.

We thank José Alcami for his kind gift of the pNL4-3 vector.