Characterization of T-Cell Responses to Conserved Regions of the HIV-1 Proteome in BALB/c Mice

A likely requirement for a protective vaccine against human immunodeficiency virus type 1 (HIV-1)/AIDS is, in addition to eliciting antibody responses, induction of effective T cells. To tackle HIV-1 diversity by T-cell vaccines, we designed an immunogen, HIVconsv, derived from the most functionally conserved regions of the HIV-1 proteome and demonstrated its high immunogenicity in humans and rhesus macaques when delivered by regimens combining plasmid DNA, nonreplicating simian (chimpanzee) adenovirus ChAdV-63, and nonreplicating modified vaccinia virus Ankara (MVA) as vectors. Here, we aimed to increase the decision power for iterative improvements of this vaccine strategy in the BALB/c mouse model. First, we found that prolonging the period after the ChAdV63.HIVconsv prime up to 6 weeks increased the frequencies of HIV-1-specific, gamma interferon (IFN-γ)-producing T cells induced by the MVA.HIVconsv boost. Induction of strong responses allowed us to map comprehensively the H-2d-restricted T-cell responses to these regions and identified 8 HIVconsv peptides, of which three did not contain a previously described epitope and were therefore considered novel. Induced effector T cells were oligofunctional and lysed sensitized targets in vitro. Our study therefore provides additional tools for studying and optimizing vaccine regimens in this commonly used small animal model, which will in turn guide vaccine improvements in more expensive nonhuman primate and human clinical trials.

T he quest for a safe and effective vaccine against human immunodeficiency virus type 1 (HIV-1)/AIDS continues (1). Both prophylactic and particularly therapeutic vaccines will likely require induction of effective cytotoxic CD8 ϩ T cells in addition to protective antibodies. There is strong evidence showing that HIV-1-specific CD8 ϩ T cells contribute to the control of HIV-1 replication during acute and chronic stages of infection by killing virus-infected cells and by producing a number of soluble factors with antiviral activities (2). However, the initial CD8 ϩ T-cell response, though strong, is typically directed toward a few immunodominant variable epitopes (3) often driving selection of virus escape mutations (4)(5)(6) and substantially contributing to the evolution of a large number of HIV-1 quasispecies detected in most infected individuals (7).
To tackle HIV-1 diversity and escape, we designed a novel immunogen, HIVconsv, assembled from the 14 most conserved regions of the HIV-1 proteome and encompassing consensus amino acid sequences derived from the four major alternating HIV-1 clades A, B, C, and D (8,9). This immunogen was presented to the immune system using a variety of vaccine vectors such as plasmid DNA with and without electroporation, human and simian adenoviruses, poxvirus modified vaccinia virus Ankara (MVA), alphavirus Semliki Forest virus replicons, and modalities such as adjuvanted synthetic long peptides (9)(10)(11)(12)(13)(14)(15). These HIVconsv vaccines were used as a standalone delivery and more often in heterologous prime-boost regimens to enhance transgene productspecific responses while avoiding boost of responses against the delivery vectors. In human studies, conserved regions delivered by a combination of plasmid DNA pSG2.HIVconsv, simian (chimpanzee) adenovirus ChAdV63.HIVconsv, and MVA.HIVconsv elicited high frequencies of oligofunctional T-cell responses with broad specificities, which correlated with inhibition of 2 out of 8 tested HIV-1 isolates in an in vitro HIV-1 inhibition assay in the majority of vaccine recipients (16). While these initial preclinical and phase I clinical trial results are highly encouraging for the conserved region strategy, there is room for improvement, for example, in terms of the breadth of HIV-1 variant inhibition. Thus, vaccine modalities, conserved immunogen designs, regimens, routes of delivery, and adjuvantation will need to be modified and tested first in iterative preclinical studies to improve the vaccine performance.
To date, we have used predominantly a single immunodominant CD8 ϩ T-cell epitope, H-2D d -and L d -restricted RGPGRAF VTI, designated P18-I10 (17,18) or historically by us as the H epitope (19), which was added to the C terminus of candidate HIV-1-derived immunogens HIVA and HIVconsv to inform vaccine development in the BALB/c mouse model (9,20). The study presented here describes powering of this model for further vaccine and regimen improvements by detailed mapping of vaccineinduced T-cell specificities supported by functional characterization of the HIVconsv vaccine-induced cellular responses, which provides comprehensive and sensitive tools for further vaccine advances. Preparation of splenocytes. Spleens were collected, and cells were isolated by pressing organs individually through a 70-m nylon cell strainer (BD Falcon) using a 5-ml syringe rubber plunger. Following the removal of red blood cells with RBC Lysing Buffer Hybri-Max (Sigma), splenocytes were washed and suspended in R10 (RPMI 1640 supplemented with 10% fetal calf serum [FCS], penicillin-streptomycin, and ␤-mercaptoethanol).

MATERIALS AND METHODS
Peptides and peptide pools. One hundred ninety-nine HIVconsvderived peptides (15/11) were divided into 6 pools of 32 to 35 individual peptides and were used at a final concentration of 1.5 g/ml in all assays as described previously (16). Groups of truncated peptides were treated identically. All peptides were synthesized by GenScript HK Limited (Hong Kong), purified to Ն90% purity, and confirmed by high-performance liquid chromatography (HPLC)-mass spectrometry.
IFN-␥ ELISPOT assay. The enzyme-linked immunospot (ELISPOT) assay was performed using the mouse gamma interferon (IFN-␥) ELISPOT kit (Mabtech) according to the manufacturer's instructions. Spots were visualized using sequential applications of a biotin-conjugated secondary anti-IFN-␥ monoclonal antibody (MAb) (R4-6A2, rat IgG1), alkaline phosphatase, and a chromogenic substrate (Bio-Rad) and counted using the AID ELISpot reader system (Autoimmun Diagnostika). While all 15-mer peptides were tested on cells from individual mice, optimal epitope mapping employed pooled samples.
Ex vivo killing assay. Equal numbers of P815 target cells were differentially labeled with either 800 nM or 32 nM carboxyfluorescein succinimidyl ester (CFSE) according to the manufacturer's specifications. The P815 cells labeled with 800 nM CFSE were pulsed with peptides for 2 h and washed several times. Splenocytes from immunized mice were prepared as described above, mixed with the differentially CFSE-labeled target cells at an effector-to-target (ET) ratio of 10:1 or 5:1, and incubated overnight at 37°C. The cells were washed, stained with a LIVE/DEAD marker, and analyzed using flow cytometry. Cytotoxicity was calculated as follows: % specific lysis ϭ 100 ϫ (number of unpulsed control cells Ϫ number of peptide-pulsed cells)/number of unpulsed control cells.
Statistical analysis. Statistical analyses were performed using Prism 6 for Mac X version 6. Multiple comparisons utilized one-way analysis of variance (ANOVA), while group pairs were compared using the twotailed unpaired t test with Welch's correction. A P value of Ͻ0.05 was considered significant.

RESULTS
A longer interval after ChAdV63.HIVconsv administration benefits MVA.HIVconsv boost. BALB/c mice were immunized with single decreasing doses of MVA.HIVconsv (Fig. 1A) or ChAdV63.HIVconsv (Fig. 1B), and the splenocytes were tested in immunogenicity assays after 1 or 3 weeks, respectively. Immunogenicity was measured by the frequency of HIVconsv-induced T cells recognizing the immunodominant H-2K d /H-2L d -restricted HIV-1 epitope H originating from the hypervariable loop 3 of Env (residues 311 to 320) (18,19) in an IFN-␥ ELISPOT assay. We found that the frequency of peptide H-specific T cells increased with increasing dose of MVA.HIVconsv, and at the highest dose of 10 8 PFU, an average of 1,016 IFN-␥ spot-forming units (SFU)/10 6 splenocytes were detected (Fig. 1A). Likewise, following ChAdV63. HIVconsv immunization, the frequency of IFN-␥-producing cells increased steadily with increasing vaccine dose and reached an average of 378 SFU/10 6 splenocytes at a dose of 10 9 virus particles (vp) (Fig. 1B). For subsequent experiments, ChAdV63.HIVconsv and MVA.HIVconsv vaccines were used at 10 8 vp and 10 6 PFU, respectively; we did not choose the maximum doses to avoid saturating the system, which might obscure detection of possible enhancements.
The overall magnitude of responses at 1 week after ChAdV63. HIVconsv immunization was much lower than that after MVA. HIVconsv immunization. We therefore sought to define the optimal time for achieving a peak multifunctional response following ChAdV63.HIVconsv administration. Groups of mice were immunized with a single dose of 10 8 vp of ChAdV63.HIVconsv, and peptide H-specific T-cell responses in pooled PBMCs were determined using an intracellular cytokine staining (ICS) assay for IFN-␥, CD107a, and TNF-␣, at 1 to 5 weeks postimmunization. Indeed, vaccine-specific T-cell frequencies peaked at 20% of IFN-␥ ϩ CD8 ϩ cells of total CD8 ϩ cells at week 5 after immunization (Fig. 1C), suggesting prolonged antigenic stimulation, possibly due to persistence of low levels of transcriptionally active ChAdV63.HIVconsv genomes, as previously described (21). One week after the bleed, the mice were boosted with 10 6 PFU MVA. HIVconsv to assess the impact of the interval between the ChAdV63.HIVconsv prime and the MVA.HIVconsv boost on the magnitude of T-cell responses. A strong synergistic effect between the two vaccines, which peaked at 1,217 SFU/10 6 splenocytes at a 6-week gap, was observed (Fig. 1D). These experiments indicated that a 5-to 6-week interval between the ChAdV63.HIVconsv prime and MVA.HIVconsv boost should be used for eliciting high frequencies of T-cell responses, which will in turn allow detailed mapping of subdominant epitopes. Subsequent experiments were thus performed using a 6-week interval.
ChAdV63.HIVconsv-MVA.HIVconsv regimen induces oligofunctional T cells. The functional diversity of HIVconsv-specific T cells elicited by the ChAdV63.HIVconsv-MVA.HIVconsv regimen was assessed using the ICS assay. Examples of dot plots for the 3 immunodominant epitopes are shown in Fig. 3A. As expected, the total frequencies of CD8 ϩ T cells expressing CD107a, IFN-␥, or TNF-␣ functions were dominated by peptide pool P6 (Fig. 3B), which contains the immunodominant peptide H. This peptide pool showed a remarkably high frequency of specific degranulating CD107a ϩ CD8 ϩ T cells, which exceeded 20% of the total CD8 ϩ T cells. Strong responses to peptide pools P2 and P4 were also observed (Fig. 3B). Characterization of CD8 ϩ T-cell polyfunctionality for peptide pools P2, P4, and P6 revealed that a large proportion of the responding cells expressed all three functions comprising CD107a, TNF-␣, and IFN-␥ (Fig. 3C). We further assessed the cytotoxic potential of HIVconsv-induced T cells in an ex vivo killing assay using peptide-pulsed cells as targets and demonstrated that freshly harvested, unstimulated splenocytes could efficiently kill peptide-pulsed P815 target cells, achieving medians of 44%, 47%, and 79% lysis of target cells sensitized with peptides 42, 112, and H, respectively, at an effector-to-target cell ratio of 10:1 (Fig. 4). These lysis data concur with the high frequency of degranulating CD107a ϩ CD8 ϩ T cells observed in Fig.  3B and C and further define the functionality of HIVconsv-specific effector T cells.

DISCUSSION
Nonreplicating vaccine vectors for delivery of pathogen-derived subunit immunogens are in the forefront of vaccine development for many infections. These vectors are most efficiently used in heterologous prime-boost regimens to avoid buildup of antivector antibodies, which dampen induction of immune responses against the transgene product (8). Although some general rules for combining heterologous vectors into a prime-boost regimen are emerging, optimization of vaccination regimens is mostly empirical. Incremental vaccination improvements are best assessed first in small-animal models such as the BALB/c mice.
This study indicated an ongoing expansion of ChAdV63.HIVconsvelicited T-cell frequencies for at least 6 weeks postadministration. This was also reflected in the requirement for a long interval of 5 to 6 weeks after ChAdV63.HIVconsv priming to achieve high frequencies of HIVconsv-specific T cells by an MVA.HIVconsv boost and confirmed the impressive immunogenicity of a simple heter- ologous ChAdV63-prime and MVA-boost regimen observed in humans (16,25,26).
A comprehensive screening of peptides covering the entire HIVconsv immunogen identified several novel H-2 d -restricted immunogenic regions recognized by the BALB/c mice. While the specificities of the stronger epitopes were described before (18,(27)(28)(29), responses to peptides 42, 151, and 164 would likely be missed using a vaccine delivery weaker than the CM regimen. This information increases the analysis "granularity" of T cells elicited by the conserved region as well as other HIV-1 vaccine candidates. More detailed and comprehensive study of cell-mediated responses aids further iterative improvements of the magnitude, breadth, functionality, and longevity of this important arm of immune defenses against microorganisms and may help shed light on more fundamental questions such as immunodominance and antigen processing.
Given the differences between the H-2 d and HLA major histocompatibility complex molecules, only some peptides immuno- genic in the BALB/c mouse are also presented by the human HLAs. Thus, peptide H is unusual in that it has four anchor residues for H-2D d (30,31) and displays "promiscuous" binding to four different H-2D d , H-2D p , H-2 u , and H-2 q murine determinants (32) as well as human HLA-A2 (33), although in our hands, in human volunteers (16,(34)(35)(36)(37)(38)(39)(40)(41)(42)(43) or HLA-A2-transgenic HHD mice (unpublished data), such responses were never detected for either the HIVA (20) or the HIVconsv (9) immunogens. No responses were detected in 23 human recipients of the HIVconsv vaccine to peptide 42 or 112, while T cells were induced to peptides 151 and 164 (16). Although not directly comparable or translatable between mouse and humans, these results increase the confidence about the usefulness of optimizing new vaccine strategies first in a small and well-defined model.