STING Ligand-Mediated Priming of Functional CD8+ T Cells Specific for HIV-1-Protective Epitopes from Naive T Cells

ABSTRACT Functional HIV-1-specific CD8+ T cells primed from naive T cells are expected to act as effector T cells in a “shock-and-kill” therapeutic strategy for an HIV-1 cure since less functional HIV-1-specific CD8+ T cells are elicited from memory T cells in HIV-1-infected individuals on combined antiretroviral therapy (cART). CD8+ T cells specific for HIV-1 conserved and protective epitopes are candidates for such T cells. We investigated the priming with STING ligand of CD8+ T cells specific for HLA-B*52:01 or HLA-C*12:02-restricted protective epitopes from naive T cells. STING ligand 3′3′-cGAMP effectively primed CD8+ T cells specific for 3 of 4 HLA-B*52:01-restricted epitopes but failed to prime those specific for all 3 HLA-C*12:02-restricted epitopes from the naive T cells of HIV-1-uninfected individuals having an HLA-B*52:01-C*12:02 protective haplotype. These HLA-B*52:01-restricted CD8+ T cells had a strong ability to suppress HIV-1 replication and expressed a high level of cytolytic effector molecules. The viral suppression ability of these T cells was significantly correlated with the expression level of perforin and showed a trend for a positive correlation with the expression level of CD107a. The present study highlighted the priming with STING ligand of functional CD8+ T cells specific for protective epitopes, which T cells would contribute as effector T cells to a shock-and-kill therapy. IMPORTANCE The current “shock-and-kill” therapeutic strategy for HIV cure has been directed toward eliminating latent viral reservoirs by reactivation of latent reservoirs with latency-reversing agents followed by eradication of these cells by immune-mediated responses. Although HIV-1-specific T cells are expected to eradicate viral reservoirs, the function of these T cells is reduced in HIV-1-infected individuals with long-term cART. Therefore, priming of HIV-1-specific T cells with high function from naive T cells is to be expected in these individuals. In this study, we demonstrated the priming with STING ligand 3′3′-cGAMP of CD8+ T cells specific for HIV-1-protective epitopes from naive T cells. cGAMP primed CD8+ T cells specific for 3 HLA-B*52:01-restricted protective epitopes, which cells expressed a high level of cytolytic effector molecules and effectively suppressed HIV-1 replication. The present study suggested that the priming with STING ligand of functional CD8+ T cells specific for protective epitopes would be useful in a therapy for an HIV-1 cure.

HLA-A*24:02 1 individuals were stimulated with Flt3 ligand for 24 h, followed by stimulation with 3939-cGAMP and NefRF10 peptide or with NefRF10 peptide only. The induction of NefRF10-specific T cells was assessed by a flow cytometric analysis using HLA-A*24:02-NefRF10 tetramers after a 10-day culture period (Fig. 1A). Specific T cells were found in culture cells stimulated with 3939-cGAMP and NefRF10 peptide from all of the individuals (mean frequency of NefRF10-specific T cells6 2 standard deviations [SD], 0.206% 6 0.186%), whereas they were not detected in those stimulated with only RF10 peptide (mean frequency of NefRF10-specific T cells 6 2 SD, 0.011% 6 0.006% [ Fig. 1B]).
We next investigated the ability of these epitope peptides to bind to HLA-B*52:01 by performing an HLA stabilization assay using RMA-S-HLA-B*52:01 cells ( Fig. 2A). GagRI8 and PolSI8 peptides bound to HLA-B*52:01 with similar binding affinities, whereas the affinity of the GagWV8 peptide for HLA-B*52:01 was significantly weaker than that of GagRI8 and PolSI8 (Fig. 2B). In contrast, the affinity of the GagMI8 peptide for HLA-B*52:01 was much weaker than that of the other 3 epitopes (Fig. 2B). Thus, the binding affinity of these epitope peptides was correlated with the efficacy of specific CD8 1 T-cell induction from naive T cells in this priming system.
Failure of STING ligand-mediated priming of CD8 + T cells specific for HLA-C*12:02-restricted HIV-1 epitopes. We further investigated the ability of the STING ligand 3939-cGAMP to prime CD8 1 T cells specific for HLA-C*12:02-restricted ones from naive T cells derived from the same individuals (indicated above) who had been analyzed for the priming of HLA-B*52:01-restricted CD8 1 T cells. In addition to 2 protective and immunodominant epitopes (PolIY11, NefMY9), HLA-C*12:02-restricted CTL immunodominant epitope EnvRL9 (36) was also used in this study. Specific CD8 1 T cells for these epitopes were analyzed by using HLA-C*12:02-tetramers for each epitope. CD8 1 T cells specific for these 3 epitopes were not primed from any of the 8 HIV-seronegative individuals having the HLA-B*52:01-C*12:02 haplotype (Fig. 1D).
Since it is well known that the expression level of HLA-C molecules on the cell surface is lower than that of HLA-A and -B ones (37)(38)(39), we speculated that the low expression level of HLA-C*12:02 molecules on the antigen-presenting cells resulted in failure of priming of CD8 1 T cells specific for HLA-C*12:02-restricted epitopes from naive T cells. So, we investigated the expression levels of HLA-C*12:02 and HLA-B*52:01 on dendritic cells (DCs). First, to quantify the relative strength of binding of anti-B5 (HLA-B*51 and -B*52) monoclonal antibody (MAb) (4D12) and anti-HLA-C MAb (DT9), we determined the concentration of these MAbs that provided the same affinity by using TP25.99 anti-HLA class I a3 domain MAb (Table 1). We then analyzed the expression levels of HLA-B*52:01 and HLA-C*12:02 on plasmacytoid DCs (pDCs), myeloid DCs (mDCs), and T cells from 3 HIV-1-seronegative individuals homozygous for the HLA-B*52:01-HLA-C*12:02 haplotype by using these 4D12 and DT9 MAbs (Fig. 3A). The expression of HLA-C*12:02 on both DCs and T cells was approximately 3 to 5 times lower than that of HLA-B*52:01 on those cells (Fig. 3B). We further analyzed the expression levels of HLA-B*52:01 and HLA-C*12:02 on DCs and T cells 24 h and 48 h after stimulation with 3939-cGAMP. The expression of HLA-C*12:02 on both DCs and T cells was approximately 3 to 10 times lower than that of HLA-B*52:01 on those cells (Fig. 3C). These results taken together suggest that a low expression level of HLA-C*12:02 molecules may be one of the important reasons why CD8 1 T cells specific for HLA-C*12:02restricted epitopes were not primed from naive T cells.
Ability of 3939-cGAMP-primed CD8 + T cells specific for protective epitopes to suppress HIV-1 replication. Since the frequency of GagRI8-, GagWV8-, and PolSI8-specific CD8 1 T cells among total primed T cells was too low to analyze the function of the primed T cells (Fig. 1), we established CD8 1 T-cell lines specific for these epitopes for further functional analysis. Epitope-tetramer 1 CD8 1 T cells were sorted and then cultured for 2 additional weeks. GagRI8-, GagWV8-, and PolSI8-specific CD8 1 T-cell lines were successfully established from 4, 2, and 5 HIV-1-seronegative individuals, respectively, having the HLA-B*52:01-C*12:02 haplotype (Fig. 4A). We investigated the ability of these primed CD8 1 T-cell lines to suppress the replication of HIV-1, which suppression is an indicator of the efficacy of CD8 1 T cells to control HIV-1. The results of the viral suppression assay showed that primed T-cell lines specific for all of these epitopes effectively suppressed HIV-1 replication in primary CD4 1 T cells infected with NL4-3 ( Fig. 4B and Table 2). The viral suppression ability of these primed T-cell lines was evaluated by comparing it with that of a positive-control CTL clone, 12B (GagRI8-specific), D3 (PolSI8-specific), or C5 (GagWV8-specific), which showed a very strong viral suppression ability in vitro (27,34). Most of the primed T-cell lines specific for GagRI8, GagWV8, or PolSI8 epitopes exhibited high viral suppression ability, which was approximately 80% of that of the positivecontrol clone (Fig. 4C). Taken together, these results showed that 3939-cGAMP could effectively prime highly functional CD8 1 T cells specific for HLA-B*52:01-restricted protective immunodominant epitopes from naive T cells.
We next investigated the ability of the primed T-cell lines specific for epitope to produce cytokines such as IFN-g and tumor necrosis factor alpha (TNF-a), and the chemokine MIP-1b, which exert antiviral, inflammatory, and regulatory functions against HIV-1 (40-42). We used HIV-1-infected CD4 1 T cells as stimulator cells to evaluate the ability of these T cells to recognize epitopes near the in vivo condition. These T cells had the ability to effectively produce these cytokines and chemokine in response to HIV-1-infected CD4 1 T cells ( Fig. 5A and B). We next analyzed the correlation between the viral suppression ability of and cytokine/chemokine production by these primed T cells. No significant correlation between these 2 abilities    was found (Fig. 5C), suggesting that the ability of these primed T cells to produce these cytokines and chemokine may not have been mainly involved in their viral suppression ability. Production of cytolytic effector molecules in 3939-cGAMP-primed CD8 + T cells specific for protective epitopes. To clarify other potent effectors for their viral suppression ability, we further analyzed the expression levels of intracellular granzyme B and perforin in primed T-cell lines and compared them with those in a positive-control T-cell clone with strong viral suppression ability ( Fig. 6A and B). T-cell lines specific for these 3 epitopes expressed granzyme B and perforin, though the levels of these molecules in the T-cell lines were relatively lower than those in the control clone. We next investigated whether the amount of intracellular perforin or granzyme B in the T-cell lines was correlated with their viral suppression ability. The viral suppression ability of the T-cell lines was significantly correlated with the expression level of perforin and showed a trend for a positive correlation with that of granzyme B (Fig. 6C). Thus, the expression of these effector molecules was correlated with the viral suppression ability of the primed T-cell lines.
To verify that intracellular granzyme B and perforin could be released from primed T cells in an antigen-specific manner, we investigated the expression of the degranulation marker CD107a, which is associated with the release of lytic granule proteins (43,44). After 6 h of stimulation with HIV-1-infected CD4 1 T cells, CD107a expression was measured. All of epitope-specific T-cell lines expressed CD107a in response to HIV-1-infected CD4 1 T cells ( Fig. 7A and B), indicating that these T cells had released cytotoxic effector molecules such as perforin and granzyme B immediately upon antigen-specific stimulation. The expression level of CD107a showed a trend for a positive correlation with the viral suppression ability of the cells (Fig. 7C). These results together suggest that these T cells primed with 3939-cGAMP could suppress HIV-1 replication mainly via lytic granule-mediated killing.

DISCUSSION
Our previous study demonstrated that the STING ligand 3939-cGAMP enables priming of HLA-A*24:02-restricted HIV-1 NefRF10-specific CD8 1 T cells with a strong effector function from naive T cells via STING ligand-mediated production of type I IFN (24). This study suggested that 3939-cGAMP acts as an adjuvant to induce more potent T-cell immunity against HIV-1. However, it remained unclear whether this ligand could induce highly functional CD8 1 T cells specific for other HIV-1 epitopes from naive T cells, especially those protective ones. As T cells specific for HIV-1-protective epitopes have a strong ability to suppress HIV-1 replication, priming of these T cells from naive T cells would be expected to serve as effector cells in a cure treatment such as a shock-and-kill therapy. In the present study, we therefore investigated the 3939-cGAMP-mediated priming of CD8 1 T cells specific for 4 HLA-B*52:01 protective epitopes as well as those for 2 protective epitopes restricted by HLA-C*12:02 and showed that CD8 1 T cells specific for 3 HLA-B*52:01-restricted epitopes were primed from  Since all of HLA-B*52:01-restricted CD8 1 T cells specific for these 3 protective epitopes exhibited a strong ability to suppress HIV-1 replication, they would be expected to be elicited in individuals with this haplotype if they were immunized with this epitope antigen together with cGAMP. A previous study on CD8 1 T cells specific for 3 hepatitis C virus (HCV) epitopes showed that the frequency of naive precursor T cells specific for the HCV epitopes is related to that of the specific T cells in chronically HCV-infected patients (45). We therefore speculated that the efficacy of HIV-1-specific T-cell induction from naive T cells would be associated with the immunodominance (breadth and magnitude) of HIV-1-specific T cells in HIV-1-infected individuals. Our previous study showed that T-cell responses to GagRI8, GagMI8, GagWV8, and PolSI8 were detected at 60% (60/100), 42% (42/100), 43% (43/100), and 43% (43/100), respectively, in HLA-B*52:01 1 individuals chronically infected with HIV-1 subtype B (26). However, CD8 1 T cells specific for the MI8 epitope were not induced from naive T cells in any of the HIV-1-seronegative individuals in the present study. The binding affinity of the MI8 peptide for the HLA-B*52:01 molecule was much lower than that of the other 3 peptides. This result suggested that the failure of induction of MI8-specific T cells from naive T cells resulted from lower presentation of this epitope by DCs. Since MI8-specific CD8 1 T cells were efficiently elicited in chronically HIV-1-infected individuals, the binding ability of MI8 peptide to HLA-B*52:01 may not critically affect the efficacy of T-cell induction in a chronic infection. A previous study showed that the magnitude of the primary T-cell response is influenced by the amount and duration of presentation of the relevant peptide-HLA ligands detected by antigen-presenting cells (46)(47)(48)(49). Since a continuous high viral load is seen during chronic HIV-1 infection, persistent exposure to HIV-1 antigens may compensate for the small amount of antigen presentation due to weak binding of MI8 epitope peptide to HLA-B*52:01. Several studies showed that the immunodominance of HIV-1-specific CD8 1 T-cell responses in a chronic infection is different from that in an acute one (50,51), suggesting that duration of antigen exposure may be involved in this difference.
A previous study showed that T-cell responses to PolIY11 and NefMY9 are detectable in 25% of HLA-C*12:02 1 individuals chronically infected with HIV-1 subtype B but that those to EnvRL9 are found in 65% of these individuals (36). Since EnvRL9-specific T cells were elicited more frequently than those specific for PolIY11 and NefMY9, we also analyzed the priming of EnvRL9-specific T cells. However, CD8 1 T cells specific for all 3 HLA-C*12:02-restricted epitopes were not primed from any of the HIV-1-seronegative individuals tested, even though those specific for HLA-B*52:01-restricted epitopes were successfully primed from the same individuals in 6 cases. Since cell surface expression levels of HLA-C are lower than those of HLA-B and HLA-A on uninfected cells and even on cells infected with HIV-1 (37)(38)(39), antigen presentation by HLA-C molecules could be weaker than that by HLA-A or HLA-B ones. We therefore speculated that it would be difficult to prime CD8 1 T cells specific for HLA-Crestricted epitopes from naive T cells due to lower presentation by the HLA-C allele. Indeed, we confirmed that the expression of HLA-C*12:02 on DCs and T cells from an HIV-1-seronegative individual was lower than that of HLA-B*52:01 on these cells.
On the other hand, persistent antigen exposure may enable the induction of HLA-C-restricted CD8 1 T cells in HIV-1-infected individuals as discussed above. The expression level of HLA-C alleles is positively correlated with the frequency of HLA-C-restricted T-cell responses in Europeans and African-Americans (52), suggesting that higher expression of HLA-C molecules results in a more efficient primary response of the HLA-C-restricted T cells in HIV-1-infected individuals. The expression level of HLA-C*12:02 remains unknown, but it might be the same as that of HLA-C*12:03, which is expressed at a medium level among HLA-C alleles (52). It can be expected that the STING ligand may prime HIV-1-specific CD8 1 T cells restricted by other HLA-C alleles such as HLA-C C*01:02 or HLA-C C*14:02, which are highly expressed on the cell surface (52).
In the majority of HIV-1-infected individuals treated with long-term cART, HIV-1-specific effector CD8 1 T cells are lost due to weak antigen stimulation under very low viral loads (10,11). In addition, the remaining HIV-1-specific memory CD8 1 T cells have ineffective or dysfunctional antiviral functions (13,14). Therefore, it would be expected that the induction of HIV-1-specific CD8 1 T cells with high function from naive T cells would be required for eliminating HIV-1-infected cells in individuals on cART. cGAMPprimed HLA-B*52:01-restricted CD8 1 T cells specific for 3 epitopes showed high expression levels of perforin and had the ability to suppress HIV-1 replication. These findings allow us to expect that these HLA-B*52:01-restricted primed T cells might contribute to the elimination of the latency-reversing agent-reactivated HIV-1 reservoir in HIV-1infected individuals on cART. We previously identified other HLA-restricted protective T-cell epitopes specific for conserved Gag and Pol regions (26)(27)(28)(29)(30). cGAMP-primed CD8 1 T cells specific for these epitopes together with those for the 3 HLA-B*52:01 epitopes described here may be useful as effector T cells for eradication of latent HIV-1infected cells in a broad number of HIV-1-infected individuals.
A previous study showed that HIV-1-specific memory CD8 1 T cells are detectable during cART but exhibit lower cytolytic function, even HLA-B*57 or HLA-B*27 protective allele-restricted T cells (13,14). Therefore, it is important to induce functional HIV-1-specific T cells from naive T cells in individuals under cART. In the present study, we demonstrated that HLA-B*52:01-restricted protective epitope-specific CD8 1 T cells with strong effector functions were primed with STING ligand 3939-cGAMP from naive T cells, suggesting that these primed T cells may contribute to HIV-1 prevention and cure. The present study highlighted the priming with STING ligand of functional CD8 1 T cells specific for protective epitopes and suggested that the priming of HIV-1-protective epitope-specific T cells would contribute to a shock-and-kill therapy. Priming of HIV-1-specific T cells from naive T cells in individuals with cART is important for shock-and-kill therapy, since the naive T-cell repertoire is smaller in HIV-1-infected individuals than that in uninfected ones (53). A study to recover the function of HIV-1-specific memory and effector T cells would also be important for this therapy. Thus, we expect to study the effect of STING ligand on HIV-1-specific memory and effector T cells in the near future.

MATERIALS AND METHODS
Subjects. Eight HIV-1-seronegative individuals having the HLA-B*52:01-C*12:02 haplotype and 7 ones having HLA-A*24:02 were recruited for this study. This study was approved by the Ethical Committee of Kumamoto University, Japan. Written informed consent was obtained from all subjects for the collection of blood and their subsequent analysis according to the Declaration of Helsinki. Peripheral blood mononuclear cells (PBMCs) were separated from whole blood by the use of Ficoll-Paque Plus. HLA genotypes of all individuals were determined by the Luminex microbead method at the HLA laboratory (Japan).
Intracellular staining of perforin and granzyme B. Epitope-specific CD8 1 T-cell lines or CTL clones stained with PE-conjugated epitope-specific tetramer, PE-Cy7-conjugated CD8 MAb, and 7-AAD were fixed with 4% paraformaldehyde solution and then rendered permeable with 0.1% saponin buffer. Thereafter, the cells were stained with Alexa Fluor 647-labeled anti-Granzyme B (BD Pharmingen) and Alexa Fluor 488-labeled anti-perforin (BD Pharmingen). All stained cells were analyzed by using the FACS Canto II.
Statistics. Statistical analyses were performed by using Prism software (GraphPad). Groups were compared by using the paired t test or unpaired t test. Correlations were determined with the Spearman rank test. P values of ,0.05 were considered significant.