Acute-Phase CD4⁺ T Cell Responses Targeting Invariant Viral Regions Are Associated with Control of Live Attenuated Simian Immunodeficiency Virus

Studies defining effective cellular immune responses to human immunodeficiency virus (HIV) and SIV have largely focused on a rare population that express specific MHC class I alleles and control virus replication in the absence of antiretroviral treatment. This leaves in question whether similar effective immune responses can be achieved in the larger population. The majority of HIV-infected individuals mount CD8⁺ T cell responses that target variable viral regions that accumulate high-frequency escape mutations. Limiting T cell responses to these variable regions and targeting invariant viral regions, similar to observations in rare “elite controllers,” may provide an ideal strategy for the development of effective T cell responses in individuals with diverse MHC genetics. Therefore, it is of paramount importance to determine whether T cell responses can be redirected toward invariant viral regions in individuals without protective MHC alleles and if these responses improve control of virus replication.

ABSTRACT

We manipulated SIVmac239Δnef, a model of major histocompatibility complex (MHC)-independent viral control, to evaluate characteristics of effective cellular responses mounted by Mauritian cynomolgus macaques (MCMs) that express the M3 MHC haplotype, which has been associated with poor control of pathogenic simian immunodeficiency virus (SIV). We created SIVΔnef-8x to test the hypothesis that effective SIV-specific T cell responses targeting invariant viral regions can emerge in the absence of immunodominant CD8⁺ T cell responses targeting variable epitopes and that control is achievable in individuals lacking known “protective” MHC alleles. Full-proteome gamma interferon (IFN-γ) enzyme-linked immunospot (ELISPOT) assays identified six newly targeted immunogenic regions following SIVΔnef-8x infection of M3/M3 MCMs. We deep sequenced circulating virus and found that four of the six newly targeted regions rarely accumulated mutations. Six animals infected with SIVΔnef-8x had T cell responses that targeted at least one of the four invariant regions and had a lower set point viral load than two animals that did not have T cell responses that targeted any invariant regions. We found that MHC class II molecules restricted all four of the invariant peptide regions, while the two variable regions were restricted by MHC class I molecules. Therefore, in the absence of immunodominant CD8⁺ T cell responses that target variable regions during SIVmac239Δnef infection, individuals without protective MHC alleles developed predominantly CD4⁺ T cell responses specific for invariant regions that may improve control of virus replication. Our results provide some evidence that antiviral CD4⁺ T cells during acute SIV infection can contribute to effective viral control and should be considered in strategies to combat HIV infection.

IMPORTANCE Studies defining effective cellular immune responses to human immunodeficiency virus (HIV) and SIV have largely focused on a rare population that express specific MHC class I alleles and control virus replication in the absence of antiretroviral treatment. This leaves in question whether similar effective immune responses can be achieved in the larger population. The majority of HIV-infected individuals mount CD8⁺ T cell responses that target variable viral regions that accumulate high-frequency escape mutations. Limiting T cell responses to these variable regions and targeting invariant viral regions, similar to observations in rare “elite controllers,” may provide an ideal strategy for the development of effective T cell responses in individuals with diverse MHC genetics. Therefore, it is of paramount importance to determine whether T cell responses can be redirected toward invariant viral regions in individuals without protective MHC alleles and if these responses improve control of virus replication.

INTRODUCTION

During human immunodeficiency virus (HIV) infection, virus-specific CD8⁺ T cell responses are associated with resolution of peak viremia. These responses exert substantial immune pressure that often results in rapid selection for viral escape variants, suggesting that limiting viral escape is beneficial and may prove critical in the design of immunotherapies for HIV (1–4). There is evidence that the control of viremia associated with individuals expressing specific “protective” major histocompatibility complex (MHC) class I alleles may be attributed to CD8⁺ T cells that target specific peptide epitopes within highly invariant regions where mutations are likely to impose a significant fitness cost (5–8). However, this rare population of “elite controllers” is estimated at less than 1% of the infected population, while the majority of HIV-infected individuals do not express protective MHC alleles and more frequently mount CD8⁺ T cell responses that target viral regions that tolerate escape mutations easily (9, 10). Recently, multiple lines of evidence also indicate a nontraditional cytolytic role of HIV-specific CD4⁺ T cells that cooperate with HIV-specific CD8⁺ T cells to mediate suppression of virus replication and may be predictive of disease outcome (11–13). It is essential for the design of vaccines and therapeutics to determine if virus-specific CD8⁺ and/or CD4⁺ T cell responses can be mounted by individuals not expressing protective MHC alleles and if these responses are effective at controlling viremia. For an intervention to be truly effective, a universal approach that can contend with the extraordinary sequence diversity of HIV in people with and without protective HLA alleles is needed.

Using nonhuman primates (NHPs), we can determine if acute-phase T cell responses targeting invariant viral regions can control primary viremia. Similar to humans, some macaques express protective MHC class I alleles associated with control of simian immunodeficiency virus (SIV) replication (14–16). However, these studies have yet to define why control of virus replication in individuals expressing protective MHC alleles is incompletely penetrant, and they do not address how to induce viral control in animals without protective MHC alleles (17). Far less is known about the specificity of SIV-specific CD4⁺ T cell responses and whether they may also directly suppress virus replication. Only recently has there been interest in developing immunogens to elicit antiviral T cells targeting conserved viral regions across individuals with diverse MHC alleles in vivo (18–20). Mauritian cynomolgus macaques (MCMs) are ideal for studying pathogen-specific T cells because they have extremely restricted MHC class I and II genetics, so that nearly all of their MHC alleles can be accounted for by 7 common haplotypes, termed M1 to M7 (21). As a result, animals with identical MHC alleles with the potential to present identical T cell peptide epitopes can be selected for studies (21, 22).

Our group and others have reported that M3/M3 MCMs poorly control infection with pathogenic SIVmac239, making them a good example of individuals with “nonprotective” MHC alleles in which to characterize favorable immune responses that could be elicited in a greater proportion of the population (23, 24). Unlike pathogenic SIVmac239, replication of live-attenuated SIVmac239Δnef is controlled in nearly every infected animal, regardless of host MHC genetics (25, 26). Control of SIVmac239Δnef replication in a host with nonprotective MHC alleles may provide a more favorable environment in which to find the characteristics of effective immune responses that control pathogenic virus replication in the broader population. Therefore, this unique model of MHC-independent control in M3/M3 MCMs may allow the characterization of effective T cell responses in animals without protective MHC alleles.

Previously, our group reported data suggesting that control of SIVmac239Δnef relied on immunodominant CD8⁺ T cell responses that select for escape mutations (25). However, at the time of our previous study, the CD8⁺ T cell responses restricted by MCMs expressing the M3 haplotype were incompletely known, and no SIV-specific M3-restricted CD4⁺ T cell responses had been identified. Additionally, the m3KOΔnef virus used in that study included additional mutations outside known M3-restricted epitopes with unknown impacts on virus replication (25). We wanted to improve upon the m3KOΔnef virus by creating a virus in which only known epitopes were disturbed and mutations in other regions of the virus were avoided. Since that time, we have improved our understanding of M3-restricted CD8⁺ T cell epitopes and now know of 10 epitopes in SIVmac239 that select for high-frequency mutations (22, 25, 27, 28).

In the current study, we used this new information to create a variant of SIVmac239Δnef, termed SIVΔnef-8x, that ablated the eight M3 MHC class I-restricted epitopes that accumulate mutations during infection with SIVmac239Δnef. We hypothesized that limiting the development of CD8⁺ T cell responses targeting highly variable epitopes might promote the development of alternate T cell responses that target invariant regions to suppress SIVmac239Δnef replication in animals with nonprotective MHC class I alleles. We identified six immunogenic regions in SIVΔnef-8x whose immunogenicity had not previously been defined in SIV-infected M3/M3 MCMs. Four of these regions did not accumulate mutations, despite eliciting detectable responses. Interestingly, all four invariant regions were restricted by M3 MHC class II molecules and were made exclusively by animals that controlled replication of SIVΔnef-8x. These data suggest that viral control is achievable in animals with nonprotective MHC alleles even when immunodominant CD8⁺ T cell responses that are normally elicited during SIVmac239Δnef infection are absent. Our findings provide support for the inclusion of immunogens able to elicit CD4⁺ T cell responses that target invariant viral antigens in the design of an effective HIV vaccine, as this approach may be applied for widespread use across individuals with a diverse array of MHC genetics.

(This article was submitted to an online preprint archive [29].)

RESULTS

Construction of SIVΔnef-8x.

We engineered a variant of SIVmac239Δnef with point mutations in eight CD8⁺ T cell epitopes restricted by MHC class I molecules expressed by the M3 haplotype (Fig. 1a). All eight epitopes are highly immunogenic and accumulate high-frequency escape mutations in response to immunodominant CD8⁺ T cell pressure elicited during infection with either SIVmac239 or SIVmac239Δnef (16, 27, 28). For each epitope, we used sequence data from nine M3/M3 MCMs chronically infected with SIVmac239 to identify common variants in the replicating virus population (27). We performed gamma interferon (IFN-γ) enzyme-linked immunospot (ELISPOT) assays with the variant peptides of each of the 8 epitopes. Using peripheral blood mononuclear cells (PBMC) collected from several SIV-infected M3/M3 MCMs, we identified variant peptides for each epitope that elicited an IFN-γ ELISPOT response in vitro significantly lower than responses against the corresponding wild-type peptide (data not shown). We then used either these identified variant peptide sequences or the variant peptide sequences we included in the m3KOΔnef virus in SIVmac239Δnef virus (25). We referred to the resulting virus as SIVΔnef-8x to indicate its origin as an SIVmac239Δnef derivative with point mutations aimed at disrupting the immunogenicity of the eight variable M3-restricted CD8⁺ T cell epitopes present in SIVmac239Δnef (Fig. 1a).

FIG 1.

Construction of SIVΔnef-8x. (a) Diagram of live-attenuated SIV with epitope locations (top) and amino acid sequence differences between SIVΔnef-8x and SIVmac239Δnef (bottom). Point mutations were engineered into 8 M3-restricted epitopes known to accumulate high-frequency mutations during SIV infection. (b) SIVΔnef-8x and SIVmac239Δnef in vitro coculture fitness assay. For each assay, we included BCVΔnef as a reference at a 1:9, 1:1, or 9:1 ratio of p27 content relative to the query virus, either wild-type SIVmac239Δnef or SIVΔnef-8x, in the inoculum. The numbers of copies of the query virus and the barcoded virus were determined by qRT-PCR at each time point. A ratio of the number of copies of query virus to that of barcoded virus was determined and compared to the ratio present in the inoculum. All the data represent means with standard deviation of triplicate values. Unpaired t tests were performed at each time point. *, P < 0.05.

To test if the epitope variants incorporated into SIVΔnef-8x affected viral fitness, we performed in vitro coculture competition assays with a barcoded SIVmac239Δnef (BCVΔnef) containing 10 synonymous changes in gag that can be detected by a separate quantitative-PCR (qPCR) assay (25, 30). Using different ratios of BCVΔnef relative to the query virus (SIVmac239Δnef or SIVΔnef-8x), our data indicated that the eight variant epitopes we incorporated into SIVΔnef-8x did not substantially alter viral fitness in vitro (Fig. 1b).

Engineered variant epitopes in SIVΔnef-8x prevent the development of responses to wild-type epitopes and are minimally immunogenic in vivo.

We infected eight M3/M3 MCMs with SIVΔnef-8x, six M3/M3 MCMs with SIVmac239Δnef, and three M4/M6 MCMs with SIVΔnef-8x (Table 1). The T cell responses present in M4/M6 MCMs infected with SIVΔnef-8x should mirror those that would develop in M4/M6 MCMs infected with SIVmac239Δnef, as these animals do not express any of the alleles of the M3 MHC haplotype (21). To determine if the mutations we incorporated into SIVΔnef-8x were sufficient to prevent the development of the expected responses targeting the eight wild-type epitopes, we performed IFN-γ ELISPOT assays using wild-type and variant peptides with PBMC collected throughout infection. At 3 weeks postinfection, PBMC from all six M3/M3 animals infected with SIVmac239Δnef recognized up to five of the eight M3-restricted wild-type epitopes (Fig. 2a, blue). In contrast, only one of eight M3/M3 animals infected with SIVΔnef-8x had positive IFN-γ ELISPOT responses for any of the wild-type epitopes (Fig. 2a, black). None of the M4/M6 MCMs infected with SIVΔnef-8x made responses specific for the eight wild-type M3-restricted epitopes (data not shown). Similar results were observed in all the animals at 8 weeks post-SIVΔnef-8x infection (Fig. 2b, black) and 12 weeks post-SIVmac239Δnef infection (Fig. 2b, blue). Therefore, M3/M3 MCMs infected with SIVΔnef-8x did not develop the previously reported CD8⁺ T cell responses that target variable epitopes during SIVmac239Δnef infection.

TABLE 1.

Animals used in this study

Animal ID	Gender	MHC haplotype	Infecting virus
cy0684	Female	M3/M3	SIVmac239Δnef
cy0686	Female	M3/M3	SIVmac239Δnef
cy0687	Male	M3/M3	SIVmac239Δnef
cy0689	Male	M3/M3	SIVmac239Δnef
cy0749	Male	M3/recM1M3a	SIVmac239Δnef
cy0752	Female	M3/recM2M3a	SIVmac239Δnef
cy0748	Male	M4/M6	SIVΔnef-8x
cy0751	Female	M4/M6	SIVΔnef-8x
cy0754	Male	M4/M6	SIVΔnef-8x
cy0685	Female	M3/M3	SIVΔnef-8x
cy0688	Male	M3/M3	SIVΔnef-8x
cy0690	Male	M3/M3	SIVΔnef-8x
cy0750	Male	M3/M3	SIVΔnef-8x
cy0753	Female	M3/M3	SIVΔnef-8x
cy0755	Female	M3/M3	SIVΔnef-8x
cy0756	Female	M3/M3	SIVΔnef-8x
cy0757	Female	M3/M3	SIVΔnef-8x

^aExpressed the major MHC class I A and B alleles present in the M3 MHC haplotype, but also expressed minor MHC class I A alleles of the M1 (cy0749) and M2 (cy0752) MHC haplotypes. Transcriptionally abundant (major) MHC-I transcripts are responsible for restricting SIV-specific CD8⁺ T cell responses (27), and both animals expressed the major MHC class I A allele present in the M3 haplotype (A1*063). Accordingly, we include these functional M3/M3 animals in our group of M3/M3 MCMs infected with SIVmac239Δnef.

FIG 2.

M3-restricted CD8⁺ T cell responses elicited during SIVmac239Δnef infection are silenced in M3/M3 MCMs infected with SIVΔnef-8x. (a and b) IFN-γ ELISPOT assays using peptides that matched the wild-type epitope sequences were performed at week 3 (a) and week 8 or 12 (b) postinfection with SIVmac239Δnef (blue) or SIVΔnef-8x (black). (c and d) IFN-γ ELISPOT assays using peptides that matched the variant epitope sequences present in SIVΔnef-8x were performed at week 3 (c) and week 8 (d) post-SIVΔnef-8x infection.

To determine if the variants incorporated in SIVΔnef-8x were immunogenic, we performed parallel IFN-γ ELISPOT assays using peptides that matched the variant epitope sequences engineered into SIVΔnef-8x. At 3 weeks post-SIVΔnef-8x infection, we observed positive IFN-γ ELISPOT responses to only two of eight variant epitopes (Fig. 2c). One variant epitope, Env338-346RF9 K3R/W8R, was immunogenic in all eight animals at 3 weeks postinfection. Another variant epitope, Tat42-49QA8 R4H, elicited a response in two animals. These responses were diminished 8 weeks postinfection, with only one animal detecting the variant epitope Env338-346RF9 K3R/W8R (Fig. 2d).

Similar in vivo pathogenicities of SIVΔnef-8x and SIVmac239Δnef in M3/M3 MCMs.

Following infection, we assessed the pathogenicity of SIVΔnef-8x compared to SIVmac239Δnef by measuring plasma viremia and memory CD4⁺ T cell subsets. The limit of detection for the viral load assay was 100 copies/ml. We did not observe the peak viral load to be significantly different (P = 0.158) in M3/M3 MCMs infected with SIVΔnef-8x and those infected with SIVmac239Δnef (Fig. 3a). While peak viremia of SIVΔnef-8x was slightly lower (P = 0.046) in M4/M6 MCMs (n = 3) than in M3/M3 MCMs (n = 8), it was not significantly different (P = 0.275) from that of SIVmac239Δnef in M3/M3 MCMs (n = 6). We also evaluated the set point viral load, calculated as the geometric mean of viral loads between 14 and 30 weeks postinfection with SIVmac239Δnef or SIVΔnef-8x (Fig. 3b). Five of six M3/M3 MCMs infected with SIVmac239Δnef established a set point viral load at or near an undetectable level (median = 104 vRNA copies/ml), while one animal (cy0687) had a set point viral load of 18,300 vRNA copies/ml. In contrast, virus levels by week 30 were more diverse in SIVΔnef-8x-infected M3/M3 MCMs and ranged from nearly undetectable to over 7,000 vRNA copies/ml (median = 468 vRNA copies/ml). Virus replication was controlled below 1,000 vRNA copies/ml in six animals, four of which maintained a set point viral load below 250 vRNA copies/ml. In contrast, two animals (cy0690 and cy0755) failed to control replication during the chronic phase of infection and had circulating virus levels that exceeded 3,500 vRNA copies/ml. A set point viral load was undetectable in two M4/M6 MCMs and fluctuated around 1,000 vRNA copies/ml in one M4/M6 MCM. Set point viral load comparisons also revealed no significant differences between groups.

FIG 3.

Similar in vivo pathogenicities of SIVΔnef-8x and SIVmac239Δnef in M3/M3 MCMs. Plasma viral load and memory CD4⁺ T cell populations were measured for 30 weeks after infection: M3/M3 MCMs infected with SIVΔnef-8x (black), M3/M3 MCMs infected with SIVmac239Δnef (blue), and M4/M6 MCMs infected with SIVΔnef-8x (red). (a) Individual peak viral loads in animals infected with SIVΔnef-8x or SIVmac239Δnef displayed as geometric means with 95% confidence intervals. (b) Comparisons of set point viral loads (geometric mean of viral loads between 14 and 30 weeks postinfection). (c) Longitudinal assessment of changes in the absolute count of memory CD4⁺ T cell subsets over time in animals infected with SIVΔnef-8x and SIVmac239Δnef. Effector memory CD4⁺ T cells were defined as CD3⁺ CD4⁺ CD8⁻ CD95⁺ CD28⁻ CCR7⁻, and central memory CD4⁺ T cells were defined as CD3⁺ CD4⁺ CD8⁻ CD95⁺ CD28⁺ CCR7⁺. An unpaired Student t test was used to calculate significant differences for the peak viral load and set point viral load.

To determine if CD4⁺ T cells were depleted following infection with SIVΔnef-8x compared to infection with SIVmac239Δnef, we evaluated changes in the absolute counts of central memory CD4⁺ T cells and effector memory CD4⁺ T cells in the blood at select time points (Fig. 3c). Though transient changes to both memory CD4⁺ T cell populations existed, we did not observe substantial differences between groups during peak viremia or the establishment of set point viral loads. Taken together, these results suggest that animals infected with SIVΔnef-8x did not exhibit enhanced peripheral depletion of CD4⁺ T cells compared to animals infected with SIVmac239Δnef.

A majority of M3/M3 MCMs control SIVΔnef-8x.

Next, we compared viral load trajectories and the time to control virus replication in MCMs infected with SIVΔnef-8x or SIVmac239Δnef. Plasma viral loads were measured in M3/M3 MCMs infected with SIVΔnef-8x (Fig. 4a, black) or SIVmac239Δnef (blue) and in M4/M6 MCMs infected with SIVΔnef-8x (red) for 30 weeks following infection.

FIG 4.

Similar in vivo pathogenicities of SIVΔnef-8x and SIVmac239Δnef in M3/M3 MCMs. Plasma viral loads and memory CD4⁺ T cell populations were measured for 30 weeks after infection: M3/M3 MCMs infected with SIVΔnef-8x (black), M3/M3 MCMs infected with SIVmac239Δnef (blue), and M4/M6 MCMs infected with SIVΔnef-8x (red). (a) Individual peak viral loads in animals infected with SIVΔnef-8x or SIVmac239Δnef displayed as geometric means with 95% confidence intervals. (b) Comparisons of set point viral loads (geometric mean of viral loads between 14 and 30 weeks postinfection). (c) Longitudinal assessment of changes in the absolute count of memory CD4⁺ T cell subsets over time in animals infected with SIVΔnef-8x and SIVmac239Δnef. Effector memory CD4⁺ T cells were defined as CD3⁺ CD4⁺ CD8⁻ CD95⁺ CD28⁻ CCR7⁻, and central memory CD4⁺ T cells were defined as CD3⁺ CD4⁺ CD8⁻ CD95⁺ CD28⁺ CCR7⁺. An unpaired Student t test was used to calculate significant differences for the peak viral load and set point viral load.

We compared the initial times required to control virus replication below a specific threshold and modeled this with Kaplan-Meier survival curves. We included all three groups in this analysis, even though the M4/M6 group had only three animals. The M4/M6 animals were used only to ensure that the in vivo fitness of SIVΔnef-8x was similar to that of SIVmac239Δnef, but our interest lay in comparing viral control in the two M3/M3 groups. Using a threshold for control similar to that of elite controller macaques infected with SIVmac239 (<1,000 vRNA copies/ml), we observed no statistically significant differences in the times required to control virus replication among the three cohorts (Fig. 4b) (14, 15, 24, 30, 31). We then evaluated time to control using a threshold matching the limit of detection for the viral load assay (<100 vRNA copies/ml) because, in most animals, SIVmac239Δnef was controlled at or near the limit of detection (25, 26). Using this threshold, we observed a statistically significant delay in the time required to control SIVΔnef-8x compared to SIVmac239Δnef in M3/M3 MCMs (Fig. 4c). Thus, when either threshold was set for viral control, at least half of the M3/M3 MCMs controlled virus replication. This suggests that M3/M3 MCMs can control virus replication even when they do not develop acute CD8⁺ T cells that select for escape mutations.

Acute-phase T cell responses elicited in M3/M3 MCMs infected with SIVΔnef-8x target several viral regions that do not accumulate mutations.

To determine if there were T cell responses present in M3/M3 MCMs infected with SIVΔnef-8x that targeted previously undefined T cell epitopes, we performed full-proteome IFN-γ ELISPOT assays with overlapping peptide pools to scan the entire SIVmac239 proteome. Positive peptide pools were then deconvoluted to assess responses to individual peptides. During acute infection (week 3) and post-peak infection (weeks 7 to 9), we identified six immunogenic regions in the SIVΔnef-8x proteome that elicited T cell responses to peptide sequences in Gag (n = 5) and Env (n = 1) that were not previously characterized in SIV-infected M3/M3 MCMs. During acute infection, all six animals that controlled SIVΔnef-8x replication responded to at least one of these new regions. Four out of these six animals made responses against 3 or more of the new regions (Fig. 5a, left). We observed similar results at 8 weeks postinfection, when five out of the six animals controlling virus replication below 1,000 vRNA copies/ml made one or more new responses (median = 2) (Fig. 5a, right). No responses were detected to any of these new regions during acute or post-peak infection in the two animals (cy0690 and cy0755) that did not control SIVΔnef-8x replication below either threshold.

FIG 5.

A majority of newly targeted regions do not accumulate high-frequency mutations during SIV infection. (a) Full-proteome IFN-γ ELISPOT assays were performed at an acute time point (week 3) and a postpeak time point (weeks 7 to 9). Peptide pools were deconvoluted to assess responses to individual peptides, resulting in the identification of responses targeting six viral regions previously undefined in M3/M3 MCMs. Only positive responses are shown, with solid bars representing assays using freshly isolated PBMC and hatched bars representing assays using frozen PBMC. (b) Virus populations were deep sequenced from plasma from M3/M3 MCMs during acute infection and post-peak infection with SIVΔnef-8x. (c) The six regions were also analyzed in M3/M3 MCMs chronically infected with SIVmac239 (>52 weeks) from a previous study by our group. The heat maps represent the percent sequence identity in relation to the inoculum. Darker colors correspond to higher sequence identity. n.c, no sequence coverage.

We deep sequenced viral populations in parallel to determine if the six newly targeted regions accumulated point mutations. During acute infection, only one of the six new regions, Gag_57–71CG15, accumulated high-frequency mutations in a majority of animals (Fig. 5b, left). When we examined the sequences from virus populations isolated during post-peak infection (weeks 7 to 9), four of the six new regions remained nearly identical to the inoculum, while Gag_57–71CG15 and Env_329–347VG19 had accumulated high-frequency mutations (Fig. 5b, right). To determine whether these 4 apparently invariant regions could accumulate mutations during pathogenic SIV infection, we examined variant accumulation in these 4 regions in virus populations isolated and sequenced from 9 M3/M3 MCMs that had been infected with SIVmac239 for ∼52 weeks for a previous study by our group (Fig. 5c) (27). In 3 of these regions, we found that 70 to 99% of the sequences (median = 97%) matched wild-type SIVmac239 in all nine animals. In eight animals, more than 80% of Gag_25–39GN15 sequences matched wild-type SIVmac239, while one animal had only 30% of sequences that matched wild-type SIVmac239. Taken together, our data suggest that it is possible to exert viral control when animals develop acute-phase T cell responses targeting regions that do not readily accumulate mutations.

CD4⁺ T cells targeting regions that do not accumulate mutations are common during SIVΔnef-8x infection.

We wanted to determine whether CD4⁺ or CD8⁺ T cells were responsible for targeting the six immunogenic regions identified in the IFN-γ ELISPOT assays. We grew polyclonal T cell lines specific for each of the six peptides and mapped MHC restriction. Of the four invariant viral regions that elicited responses, Mafa-DRA*01:02:01/DRB1*10:02 restricted both Gag_249–263WY15 and Gag_297–315Y19, while Mafa-DPA1*13:01/DPB1*09:02 restricted both Gag_25–39GN15 and Gag_413–427GC15 (Fig. 6a). MHC restriction was also mapped for the two responses targeting viral regions that accumulated high-frequency mutations during SIVΔnef-8x infection. Using the optimal peptides contained within the regions, we found that Gag_57–71CG15 was restricted by Mafa-B*011:01 while Env_329–347VG19 was restricted by Mafa-A1*063:02 (Fig. 6b). A summary of the newly targeted regions and their restricting alleles is shown in Table 2. Thus, acute-phase CD4⁺ T cells targeting viral regions that do not accumulate mutations are common during acute infection in animals without favorable MHC genetics that ultimately control virus replication.

FIG 6.

Characterization of M3-restricted SIVΔnef-8x T cell responses. ICS assays were performed to determine MHC class II restriction for CD4⁺ T cell lines specific for the four identified invariant regions (a) and MHC class I restriction for CD8⁺ T cell lines specific for the two identified variable regions (b). The data represent the percentages of cells positive for IFN-γ and/or TNF-α for each T cell line. MHC-matched 721.221 cells or K562 cells expressing the indicated M3 MHC class I alleles or RM3 cells expressing the indicated M3 MHC class II alleles were used as antigen-presenting cells. Peptide-pulsed untransfected 721.221 cells, K562 cells, or RM3 cells were used as negative controls.

TABLE 2.

Characterization of M3-restricted SIVΔnef-8x T cell responses

Immunogenic region	Invariant	Epitope sequence	Allele specificity
Gag25-39GN15	Yes	GKKKYMLKHVVWAAN	DPA1*13:01/DPB1*09:02
Gag57-71CG15	No	CQKILSVLAPLVPTG	B*011:01
Gag249-263WY15	Yes	WMYRQQNPIPVGNIY	DRA*01:02:01/DRB1*10:02
Gag297-315YK19	Yes	YVDRFYKSLRAEQTDAAVK	DRA*01:02:01/DRB1*10:02
Gag413-427GC15	Yes	GCWKCGKMDHVMAKC	DPA1*13:01/DPB1*09:02
Env329-347VG19	No	VFHSQPINDRPKQAWCWFG	A1*063:02

DISCUSSION

Rare individuals who express protective MHC alleles control HIV replication without antiretroviral treatment and often make CD8⁺ T cell responses that target highly invariant, possibly evolutionarily conserved, viral regions (7, 8, 32). In contrast, responses to invariant viral regions are frequently subdominant in HIV-infected individuals who do not express protective MHC alleles (10). It is hypothesized that an effective vaccine will need to limit T cell responses to highly immunogenic variable regions in order to maximize the likelihood of developing T cell responses against invariant viral regions (33–35). Using SIVΔnef-8x, we directly tested this hypothesis in M3/M3 MCMs that did not express protective MHC alleles and observed that control of SIVmac239Δnef could still be achieved.

Cellular responses that target Gag during acute HIV infection have been associated with a low viral load and improved disease outcome (2, 8, 9). Out of the six regions that elicited T cell responses in M3/M3 MCMs infected with SIVΔnef-8x, five were located within Gag. All six animals that controlled SIVΔnef-8x replication made T cell responses during acute infection that recognized one or more of the following five regions that span the majority of Gag: Gag_249–263WY15 and Gag_297–315YK19 are located within the p27 capsid (CA), Gag_57–71CG15 and Gag_25–39GN15 are located within the p15 matrix (MA), and Gag_413–427GC15 is located within p6. Of note, the first 9 amino acids of Gag_297–315YK19 are present in an evolutionarily conserved motif known as the major homology region (MHR) and share three of the four residues identified as highly conserved within the MHR (36). The MHR is also contained within the C-terminal subdomain of CA (CA-CTD), one of only two regions of HIV-1 Gag that are absolutely required for assembly, suggesting the region may represent an ideal target for T cell-based vaccines (37). In all six animals that controlled SIVΔnef-8x, four of the five Gag regions that were targeted did not accumulate mutations. Interestingly, the same four regions did not accumulate high-frequency mutations in nearly all nine M3/M3 MCMs chronically infected with SIVmac239 (>52 weeks) that were sequenced by our group for a previous study (Fig. 5c) (27).

Besides the four invariant regions, we found two regions that accumulated mutations during acute SIVΔnef-8x infection. Gag_57–71CG15 accumulated both Q58R and V63A mutations by 3 weeks postinfection that were present at a frequency of 7% to 37% and 4% to 88%, respectively. By 9 weeks postinfection, less than 1% of the circulating virus matched the inoculum in Gag_57–71CG15 for all eight M3/M3 MCMs infected with SIVΔnef-8x (Fig. 5b, right). Even though the V63A mutation in Gag_57–71CG15 was observed coincident with breakthrough viremia in an elite controller rhesus macaque infected with SIVmac239 (31), we found the V63A mutation was present in virus populations from both controllers and noncontrollers of SIVΔnef-8x (data not shown). Our results suggest that this mutation alone does not confer breakthrough replication, so perhaps targeting this region of Gag may still offer some immunological benefit.

We also observed high-frequency mutations in Env_329–349VG19, a region that is located at the C terminus of the V3 loop (38). This region also contains the variant epitope Env_338–346RF9 K3R/W8R that we engineered into SIVΔnef-8x and was immunogenic during acute infection in all eight animals (Fig. 2c). The most common mutation we observed within the newly targeted Env_329–349VG19 was an R-to-W change at position 345 of Env that restored one of the two mutations in Env_338–346RF9 to the original SIVmac239 sequence. It is possible that the R-to-W reversion was a consequence of the development of T cells targeting Env_338–346RF9 K3R/W8R during acute infection. Alternatively, it is possible that tryptophan at position 345 may be more favorable than the arginine we engineered into the virus. Given the proximity of this arginine to the C311-C344 link that serves as the base of the V3 loop, it seems possible that the polarity of the amino acid at position 345 may impact the formation of the V3 loop. The relatively conserved nature of the V3 loop, as well as its role in determining coreceptor tropism, further supports this region as a site for effective T cell responses to exert their antiviral function during acute infection (38–40).

There is growing evidence that HIV-specific CD4⁺ T cells may play a bigger cytolytic role in control of virus replication than previously thought (11, 32, 41, 42). SIV-specific CD4⁺ responses have been previously detected in macaques infected with pathogenic SIV, as well as SIVmac239Δnef (43–45). While observed to be typically subdominant in comparison to many CD8⁺ T cell responses, immune pressure imparted by CD4⁺ T cells has been sufficient to select for escape variants in certain cases (30, 31, 46). We found four immunogenic regions that elicited acute-phase CD4⁺ T cells in M3/M3 animals infected with SIVΔnef-8x. All four of these were located in Gag and did not accumulate mutations by 9 weeks postinfection despite eliciting positive IFN-γ ELISPOT responses, suggesting they may represent invariant viral regions that are immunogenic. Notably, we deep sequenced virus populations in M3/M3 MCMs infected with SIVmac239 for >52 weeks as part of a previous study by our laboratory and found that the same four regions of Gag did not routinely accumulate mutations, though it is unknown whether these regions elicited virus-specific CD4⁺ T cells (Fig. 5c) (27). Although we were unable to dissect a mechanism of CD4⁺ T cell-mediated control of virus replication during acute SIVΔnef-8x infection, cytolytic CD4⁺ T cells have been previously implicated in control of viral infections (11, 41, 47). This argument is in contrast to virus-specific CD4⁺ T cells as preferential targets of infection and may be explained by distinct transcriptional and functional signatures of cytolytic CD4⁺ T cells that mirror CD8⁺ T cells more than Th1 CD4⁺ cells (13, 48, 49). Together, we provide evidence that acute-phase CD4⁺ T cells may improve control of SIV replication, and we provide a model through which to further explore this mechanism in future studies.

It is certainly possible that infection with SIVΔnef-8x did not deplete peripheral CD4⁺ T cells, in contrast to reports of macaques infected with SIVmac239 (50–52). As a result, this could increase the likelihood that SIV-specific CD4⁺ T cell responses were preserved and elicited following infection with SIVΔnef-8x. While changes to the numbers of effector memory and central memory CD4⁺ T cells could not account for the differential disease progression observed in animals infected with SIVΔnef-8x, it is possible that the activation states of these populations may have been an influencing factor. Previously, macaques infected with SIVmac239 have been shown to have increased activation of memory CD4⁺ T cells (52). In contrast, macaques infected with SIVmac239Δnef typically do not generate activated memory CD4⁺ T cells, and disease progression rarely occurs (45, 52). Still, there is evidence that animals infected with SIVmac239Δnef can exhibit some limited T cell proliferation during early infection (51). Thus, future studies of animals infected with SIVΔnef-8x may warrant an examination of T cell activation in the periphery and mucosal tissues to assess if there is an association with viral control.

We found that two M3/M3 MCMs infected with SIVΔnef-8x (cy0690 and cy0755) and one M3/M3 MCM infected with SIVmac239Δnef (cy0687) were unable to control virus replication, similar to what was seen in the animals of the Harris et al. study (25). Both cy0690 and cy0755 did not have T cell responses detectable by IFN-γ ELISPOT assays. When we sequenced virus populations isolated from cy0690 and cy0687, we found an additional deletion in nef that had previously been shown restore the reading frame and to result in increased pathogenicity (data not shown) (53). This observation prompted us to reexamine the sequences of virus populations replicating during chronic infection from the M3/M3 MCMs of the Harris et al. study. We found similar sequence changes within the same region of nef in virus populations isolated from these animals (data not shown). The precise mechanism by which nef is restored in certain animals and whether functional advantages are conferred remains to be determined and requires further investigation.

Together, our study suggests that perhaps expanding subdominant virus-specific CD4⁺ T cells toward invariant viral regions during early infection may improve viral control. Even though not every animal in our study was able to mount these responses and elicit viral control, our data provide compelling evidence that CD4⁺ T cell responses targeting MHC class II-restricted epitopes have the potential to be effective and can be mounted by individuals without protective MHC alleles. To our knowledge, this is the first identification of MHC class II-restricted SIV-specific T cell responses in MCMs. Future nonhuman primate studies may consider focusing on effective SIV-specific CD4⁺ T cell responses and evaluating their direct contribution to controlling virus replication in vivo.

MATERIALS AND METHODS

Animal care and use.

Seventeen MCMs were purchased from Bioculture Ltd. and were housed and cared for by the Wisconsin National Primate Research Center (WNPRC) according to protocols approved by the University of Wisconsin Graduate School Animal Care and Use Committee. The animals were chosen based on expression of particular MHC alleles as previously described (54–58). Eight M3/M3 MCMs and three M4/M6 MCMs were infected intravenously with 10 ng of p27 SIVΔnef-8x. Four M3/M3 MCMs and two functional M3/M3 MCMs were infected intravenously with 10 ng p27 from SIVmac239Δnef. Four of these MCMs were infected as part of a previous study (26), and two animals (cy0749 and cy0752) were functionally M3/M3 (Table 1).

Creation of virus stocks.

We created plasmids (pUC57) containing the 5′ and 3′ viral genomes of SIVmac239Δnef by custom gene synthesis (GenScript, Piscataway, NJ), as was done previously (25). For SIVΔnef-8x, 10 substitutions were then incorporated into SIVmac239Δnef by site-directed mutagenesis (GenScript, Piscataway, NJ). The plasmids containing the 5′ and 3′ halves of the corresponding genomes were digested with SphI, treated with Antarctic phosphatase, precipitated, and ligated together. Vero cells were transfected with the ligated products and cocultured with CEMx174 cells for 48 h. The infected CEMx174 cells were grown for ∼2 weeks to produce high-titer viruses and harvested daily during the last week. Plasma SIV loads were determined by quantitative reverse transcription (qRT)-PCR, and the p27 content of the virus stocks was determined by enzyme-linked immunosorbent assay (ELISA) (ZeptoMetrix Corp., Buffalo, NY) according to the manufacturer's protocol. The p27 content of SIVmac239Δnef was 327 ng/ml, and the viral load was 2.19e9 copies/ml. The p27 content of SIVΔnef-8x was 245 ng/ml, and the viral load was 1.39e9 copies/ml.

In vitro coculture competition fitness assays.

In triplicate, SIVmac239Δnef or SIVΔnef-8x was mixed with BCVΔnef at p27 content ratios of 1:1, 1:9, and 9:1. Each virus mixture was incubated with 1e6 CEMx174 cells at 37°C for 4 h. After washing, 5e5 cells were plated and grown for 1 week, with supernatant sampled at days 3, 5, and 7. A discriminating qPCR assay was used to quantify the copies of SIVmac239Δnef or SIVΔnef-8x and BCVΔnef, as was done previously (25, 30). Briefly, viral RNA (vRNA) was isolated from the inoculum and each supernatant and then quantified with a SuperScript III Platinum one-step quantitative-PCR kit (Invitrogen, Carlsbad, CA). In one reaction, primers and probes targeting an 84-bp region of gag were used to quantify SIVmac239Δnef and SIVΔnef-8x. A separate reaction with a distinct set of primers and probes was used to quantify BCVΔnef. The p27 content ratio of SIVΔnef-8x or SIVmac239Δnef to BCVΔnef in each supernatant was normalized to the ratio that was present in the inoculum, and replicative differences between viruses were assessed at each time point with unpaired Student t tests (GraphPad Prism, La Jolla, CA). All the data represent means with standard deviations of triplicate values and are plotted on a log₂ scale.

Plasma viral load analysis.

Plasma was isolated from undiluted whole blood by Ficoll-based density centrifugation and cryopreserved at −80°C. SIV gag loads were determined as previously described (26). Briefly, vRNA was isolated from plasma, reverse transcribed, and amplified with the Superscript III Platinum one-step qRT-PCR system (Invitrogen). The detection limit of the assay was 100 vRNA copy equivalents per ml of plasma. The limit of detection (100 vRNA copies/ml) was reported when the viral load was at or below the limit of detection.

Peptides.

The NIH AIDS Research and Reference Reagent Program (Germantown, MD) provided 15-mer peptides overlapping by 11 amino acid positions spanning the full SIVmac239 proteome. Additional peptides used for mapping were created by custom synthesis (GenScript, Piscataway, NJ). All the peptide sequences were derived from the SIVmac239 proteome.

IFN-γ ELISPOT assays.

IFN-γ ELISPOT assays were performed using fresh and frozen PBMC, as previously described (26, 27). Each of the eight variant epitopes was selected using frozen PBMC collected from at least five M3/M3 MCMs during acute or chronic SIV infection. Fresh PBMC were isolated from EDTA-anticoagulated blood by Ficoll-based density centrifugation. A precoated monkey IFN-γ ELISPOTplus plate (Mabtech, Mariemont, OH) was blocked, and individual peptides were added to each well at a final concentration of 1 μM. Multiple peptide pools containing 15-mer peptides that spanned the full SIVmac239 proteome, each overlapping by 11 amino acids, were used to assess new responses that emerged during infection. The peptide pools totaled 10 μM (1 μM each peptide) and were added to cells at a final pool concentration of 1 μM. Each peptide or peptide pool was tested in duplicate, as was concanavalin A (10 μM), which was used as a positive control. Four to 10 wells did not receive any peptides and served as a negative control for calculating background reactivity. Assays were performed according to the manufacturer's protocol, and wells were imaged with an ELISPOT reader (AID Autoimmun Diagnostika GmbH). Positive responses were determined using a one-tailed t test and an α level of 0.05, where the null hypothesis was that the background level would be greater than or equal to the treatment level (59). Positive responses were considered valid only if each duplicate well had a value of at least 50 spot-forming cells (SFCs) per 10⁶ PBMC. If statistically positive, the reported values represent the average of the test wells minus the average of all negative-control wells.

Generation of M3/M3 BLCLs.

MHC-matched B lymphoblastoid cell lines (BLCLs) were generated as previously described (28). Briefly, PBMC were isolated by density-based centrifugation from whole blood containing EDTA. B cells were then immortalized with medium from an S549 cell line containing herpesvirus papio. The cells were maintained in R-10 medium and sequenced to verify the presence of appropriate MHC alleles.

Generation of peptide-specific T cell lines.

Polyclonal CD8⁺ and CD4⁺ T cell lines were generated from whole-blood PBMC that were isolated by Ficoll-based density centrifugation. Based on the availability of blood, approximately 3 to 5 peptide-specific CD8⁺ T cell lines were generated from animals infected with SIVΔnef-8x, as previously described (28). Briefly, 5e6 freshly isolated PBMC were incubated with 2 μM peptide in complete medium (RPMI 1640 medium supplemented with 15% fetal calf serum, 1% antibiotic/antimycotic, and 1% l-glutamine) with 100 IU of interleukin-2 (NIH AIDS Research and Reference Reagent Program). After 1 week, the cell lines were restimulated weekly with peptide-pulsed irradiated (9,000 rads) BLCLs, as previously described (28, 60). Approximately 2 or 3 CD4⁺ T cell lines per peptide were generated and cultured as previously described (61). Briefly, freshly isolated PBMC were depleted of CD8⁺ cells using NHP-specific anti-CD8 microbeads (Miltenyi Biotech, San Diego, CA) via magnetic separation according to the manufacturer's protocol. After separation, 5e6 CD8-depleted cells were used to generate CD4⁺ T cell lines and cultured in complete medium similarly to CD8⁺ T cell lines, with the addition of interleukin-7 (BioLegend, San Diego, CA) at a final concentration of 50 ng/ml. In select cases, we attempted to generate multiple CD4⁺ T cell lines using whole PBMC but were unable to establish peptide-specific polyclonal T cell lines similar to those established following magnetic separation of CD8⁺ cells.

Intracellular-cytokine-staining (ICS) assay.

To determine T cell line peptide specificity and the restricting MHC class I or class II molecule, we measured intracellular expression of IFN-γ and tumor necrosis factor alpha (TNF-α) as previously described (28, 61). Briefly, 2e5 peptide-pulsed BLCLs or MHC class I or class II transferents were incubated for 1.5 h with peptide at 37°C, washed twice with RPMI medium containing 10% fetal bovine serum (FBS) (R10), and then combined with 2e5 cells from corresponding CD4⁺ or CD8⁺ T cell lines for an additional 4 h at 37°C in the presence of brefeldin A (Sigma-Aldrich, St. Louis, MO). The cells were incubated with LIVE/DEAD fixable near-infrared (IR) dead cell stain for 15 min before being surface stained with CD3-AF700 (clone SP34-2; BD Biosciences), CD4-allophycocyanin (APC) (clone M-T466; Miltenyi Biotec), and CD8-Pacific Blue (clone RPA-T8; BD Biosciences) for 30 min in the dark at room temperature. The cells were fixed with 2% paraformaldehyde (PFA) for at least 20 min and then permeabilized with 0.1% saponin and stained with IFN-γ (clone 4S.B3; BD Biosciences) and TNF-α–peridinin chlorophyll protein (PerCP)/Cy5.5 (clone MAb11; BD Biosciences) for 30 min in the dark at room temperature. Flow cytometry was then performed on an LSR II instrument (BD Biosciences) using 2% PFA-fixed cells, and data were analyzed using FlowJo version 9.9.6 (Treestar, Ashland, OR).

Deep sequencing of SIV.

Replicating virus populations were subjected to genome-wide deep sequencing, as previously described (27). Briefly, viral RNA was isolated from plasma with a MinElute virus spin kit (Qiagen), and amplification of cDNA was performed using a Superscript III one-step reverse transcription (RT)-PCR system with high-fidelity Platinum Taq (Invitrogen). This resulted in four overlapping amplicons spanning the entire SIV coding sequence, which were then purified with the MinElute gel extraction kit (Qiagen) and quantified using a Quant-IT double-stranded DNA (dsDNA) HS assay kit (Invitrogen). Pooled amplicons (1 ng) were used to generate uniquely tagged libraries using a Nextera XT kit (Illumina). In select cases, cDNA was generated using a SuperScript IV First Strand synthesis system (Invitrogen) and PCR amplified with Q5 high-fidelity DNA polymerase (NEB) in duplicate to generate 37 overlapping amplicons spanning the entire SIV coding sequence. The amplified products were quantified using the Quant-IT dsDNA HS assay kit (Invitrogen) and diluted to 3 ng/μl for library preparation with TruSeq Nano HT (Illumina). Tagged libraries were then quantified using the Quant-IT dsDNA HS assay kit, and the fragment size distribution was assessed using a high-sensitivity Agilent bioanalyzer chip. The libraries were then pooled and sequenced on an Illumina MiSeq.

All sequences were analyzed with Geneious (Biomatters, Ltd.). Paired reads were initially quality trimmed using the BBDuk (decontamination using kmers) plugin, a part of the BBTools package by Brian Bushnell (Joint Genome Institute, Walnut Creek, CA) and then mapped with high sensitivity to the SIVmac239 sequence; gaps up to 500 bp were allowed when mapping to the SIVmac239 reference sequence (GenBank accession number M33262) in order to identify any additional deletions or insertions. Variant nucleotides were called at a threshold of 1%. The variants detected in the analyzed virus populations were then compared to the mutant sites originally incorporated into SIVΔnef-8x.

Statistics.

Student's t test was used evaluate the significance of differences between the peak viral load and set point viral load (the geometric mean of viral loads from 15 to 30 weeks postinfection). All viral load measurements were log₁₀ transformed. Kaplan-Meier survival analyses were used to model the time to control virus replication in M3/M3 MCMs infected with SIVΔnef-8x (n = 8) or SIVmac239Δnef (n = 6) and M4/M6 MCMs infected with SIVΔnef-8x (n = 3); a log rank (Mantel-Cox) test was then used to determine the significance of differences. All statistical analyses were performed using GraphPad (La Jolla, CA) Prism.

Accession number(s).

The genome-sequencing and assembly data determined in this study are available in GenBank under BioProject number PRJNA479845.