Study participants and cell preparation
We obtained up to 300 ml of peripheral blood from 19 HIV-1 positive participants (one on two occasions); three were ultimately excluded from the data presented, because their cells did not show evidence of latency reversal after stimulation with antibodies to CD3 and CD28 (the positive T cell activation control). All participants had a CD4 T cell count of at least 500 cells/µl of peripheral blood and fully suppressed viremia (less than 20 copies/ml) for at least one year. For the participants for whom data are available, HIV-1 infection was diagnosed five to over twenty-seven years before their blood was obtained for this study (Table 1).
We isolated peripheral blood mononuclear cells (PBMC) by density centrifugation. Constitutive expression of MHC class-II molecules is confined to professional antigen presenting cells (APCs). To ensure optimal antigen presentation to CD4 T cells in the context of MHC class-II, we kept professional APCs - B cells and monocytes - in the cell population by depleting CD8 T cells from the PBMC rather than isolating the CD4 T cells. Our rationale for removing CD8 T cells was to avoid potential killing of reactivated reservoir cells as well as the potential secretion of suppressive mediators. CD8 T cell depletion from participant PBMC was confirmed by flow cytometry, and residual CD8 T cells were usually < 0.1 % (data not shown).
Identification of antigen specific CD4 T cell responses by IFN-γ ELISpot assay
Production of interferon (IFN)-γ by CD4 T cells is a hallmark of the Th1-type CD4 T cell phenotype and is typically associated with an effective host defense against intracellular pathogens (reviewed in [26]). Therefore, we first used an IFN‑γ enzyme-linked immune absorbent spot (ELISpot) assay to evaluate the ability of peptide pools to activate and induce an immune response in the CD4 T cells within the CD8 depleted cell populations.
We screened IFN-γ production in response to peptide pools containing 15mer overlapping peptides from: 1) Cytomegalovirus (CMV) matrix protein 65 (pp65); 2) Candida albicans mannoprotein MP65; 3) a mixture of 14 previously identified optimal MHC class-II restricted epitopes from human cytomegalovirus (HHV-5), Epstein-Barr virus (HHV-4), influenza A, and Clostridium tetani toxoid (pool named CEFT); 4) HIV-1 Gag; 5) HIV-1 Pol; 6) HIV-1 Env; and 7) HIV-1 Nef. For each condition, we stimulated 100,000 CD8 depleted PBMC with 1 µg/ml per individual peptide. We used anti-CD3/CD28-coated beads as well as phorbol myristate acetate in combination with ionomycin (PMA/Iono) as a positive control for maximal T cell stimulation. Both positive controls typically saturated the signals on the ELISpot plate and did not allow quantification (Figure 1A). We used DMSO as a negative control, because the peptide pools and individual peptides were reconstituted in DMSO. Wells treated with DMSO typically showed <10 spot forming cells (SFC)/million cells. IFN‑γ responses of varying magnitudes were measured for these peptide pools in CD8 depleted PBMCs from a subset of participants (Figure 1A). For three participants, we tested the 123 Gag 15mer peptides singly at a concentration of 12.5 µg/ml per peptide with 100,000 CD8 depleted PBMC per well to identify the specific peptides within the pool that induced IFN-γ production in the CD4 T cells (Figure 1B). Individual peptides that yielded >50 SFC/million CD8 depleted PBMC were subsequently pooled to generate a participant-customized ELISpot selected Gag peptide pool used in some of the latency reactivation experiments (Figure 1B; indicated by asterisks). We verified for one participant that the observed IFN-γ production was not a result of contamination with CD8 T cells by comparing two ELISpot assays, one with CD8 depleted PBMC and one with isolated CD8 T cells. The ELISpot assay with the isolated CD8 T cells showed IFN-γ production in response to different peptides compared to the assay with CD8 depleted PBMC, suggesting that the different epitopes were recognized in the context of MHC class-I and MHC class‑II (Figure 1C). Overall, these data indicated that with the possible exception of the peptides derived from Candida MP65, the peptide pools activated IFN-γ production in CD4 T cells, presumably in an antigen-specific manner.
Virion-release from latently infected cells after antigenic stimulation measured by real-time RT-qPCR
For nine participants, we stimulated five million CD8 depleted PBMC with each peptide pool in quadruplicate for seven days in the presence of the integrase inhibitor raltegravir [1 µM]. Raltegravir prevented viral spread and ensured that the measured cell-free, virion-associated (cf)-RNA was released directly from reactivated reservoir cells and not a consequence of propagation of virus in the culture. Depending on the yield of isolated cells from each participant, we stimulated the cells with overlapping peptide pools from Gag, CMV, Candida, CEFT, Pol, Env, and Nef at a concentration of 1 µg/ml per peptide. DMSO was used as negative control and plate‑bound antibodies against CD3 and CD28 were used as a maximum-stimulation control.
We first assessed the activity of the peptide pools with respect to immune activation by measuring the early activation marker CD69. After two days of culture, an aliquot of ~50,000 cells from each condition was stained for cell-surface CD69 and CD4 and analyzed by flow cytometry. The DMSO controls showed baseline CD69 positive CD4 T cells ranging from 1% to 12%, whereas the plate‑bound anti-CD3 and anti-CD28-stimulated cells showed the maximum T cell activation for each participant, which ranged from 45% to 87% CD69 positive CD4 T cells (Figure 2A). In general, none of the peptide pools activated more than 20% of the CD4 T cells.
After seven days of antigen-stimulation in the presence of raltegravir, we measured released virion-associated RNA to assess reversal of latency. The culture supernatants were cleared of debris and cellular contaminants by low-speed centrifugation, and the virions were isolated by ultracentrifugation through 20% sucrose. Cell-free (cf)‑RNA was isolated and cf-HIV-1 Gag RNA was measured using real-time RT-qPCR (Figure 2B). The results were normalized to the cf-RNA yields after maximum stimulation with antibodies to CD3 and CD28 for each participant. One participant was evaluated twice, at times approximately one year apart (Figure 2B, filled circles and open circles). At the first evaluation, this participant's cells showed relatively high fractional levels of latency-reversal in response to several peptide pools including the Gag and CEFT pools, despite a limited up-regulation of CD69 (Figure 2A, filled circles). At this time, this participant's cells yielded cf-RNA in response to the C. albicans peptide pool, although this pool did not show activity in the IFN-γ ELISpot assay using cells from other participants (Figure 1A). Also, at this time, this participant's cells had the highest baseline expression of cf‑RNA in the DMSO control. At the later time (open circles in Figure 2B), this participant's cells showed a different pattern of latency reversal: under 5% of the positive control values in all cases except for the Nef peptide pool, which was over 10% of the positive control value. Cells from the other participants shown in Figure 2 yielded modest latency reversal after exposure to the peptide pools (less than 15% of the positive control values). The HIV-1 Nef peptide pool appeared the most consistent among multiple participants, although the fraction of latency reversal relative to the positive control was less than 15%.
Viral cell-associated mRNA induction after antigenic stimulation measured by droplet digital (dd)PCR
To confirm and extend the results obtained measuring cf-RNA, we evaluated latency reversal in a second group of participants by measuring cell-associated (ca) HIV-1 mRNA using a ddPCR assay that detects multiply spliced Tat/Rev transcripts [27]. We stimulated nine million CD8 depleted PBMC from six participants with each peptide pool, again in the presence of 1 µM raltegravir to prevent viral spread. Depending on the yields of participant cells, CD8 depleted PBMC were stimulated with peptide pools from CMV, CEFT, Gag, or a participant-customized ELISpot-selected Gag peptide pool at a concentration of 1 µg/ml and plated in three-fold limiting dilutions and six replicates. As before, we used DMSO as the negative control and plate‑bound antibodies to CD3 and CD28 as a maximum-stimulation control. Cellular activation was measured by the up-regulation of CD69 after 48 hours of incubation. Consistent with the results above (Figure 2A), we observed that the peptide pools did not activate more than 20% of the CD8 depleted PBMC (Figure 3A).
After five days of culture in the presence of 1 µM raltegravir, we collected the cells and isolated ca‑RNA to measure the amounts of multiply spliced Tat/Rev mRNA using the ddPCR assay. Overall, the ddPCR mRNA data showed no consistent induction of the expression of multiply spliced Tat/Rev mRNA by the different peptide pools. With some exceptions the peptide pools reversed latency only modestly when compared to the positive control (Figure 3B). In cells from one participant (indicated by open diamond), the Gag peptide pool induced Tat/Rev mRNA to 75% of the positive control value, but for this participant the DMSO control value was also unusually high (24%). Notably, in cells from another participant (indicated by "x"), the ELISpot-selected Gag-peptide pool induced Tat/Rev mRNA to 50% of the positive control value, while the DMSO control value in this case was low (under 3%). This participant also had partial latency reversal in response to peptide pools of CMV (18% of the positive control) and the CEFT mixture (29% of the positive control).
Antigen presentation by autologous monocyte-derived dendritic cells (DC)
We considered that suboptimal antigen presentation might render the above experiments less sensitive to the potential activity of antigens as LRAs. Therefore, we isolated, differentiated, and matured autologous monocyte-derived DC from two additional participants and used these mature DC as antigen-presenting cells. We also used complete proteins in addition to peptide pools as antigens to better simulate natural antigen processing and presentation (Figure 4 and Table 2). In these co-culture experiments, like those above, we included the integrase inhibitor raltegravir to block the spread of infection during the seven day incubation (Figure 4B), but for one participant we also omitted the raltegravir and incubated the cells for 18 days to determine how allowing viral propagation would affect the results (Table 2).
Consistent with our experiments using CD8 depleted PBMC, the use of autologous DCs to present antigen did not cause substantial T cell activation measured by CD69 up-regulation (Figure 4A), nor did it yield consistent latency reversal by any peptide or protein antigen (Figure 4B). For both participants, we used cf-RNA for the initial readout (Figure 4B). The data of one participant indicated that in the presence of raltegravir, the Gag peptide pool reversed latency to almost 40% of the positive control value, but the other peptide pools and complete proteins were inactive.
For the participant in whom the Gag peptide pool appeared active (square symbol in Figure 4B), we also omitted raltegravir from parallel cultures: in this condition the intact p55 Gag protein yielded substantial cf-RNA (1 x 1010 copies/ml) reflecting marked viral outgrowth, but neither SIV Gag nor any other antigen did (Table 2). Notably, although the Gag peptide pool did not yield amounts of cf-RNA comparable to maximum-stimulation or the p55 Gag protein, it differed from the other conditions, yielding 3 x 104 copies/ml of cf-RNA compared to the low or undetectable copies for DMSO and the other antigens (Table 2). For this participant (square symbol in Figure 4B), we also used ddPCR of ca Tat/Rev mRNA as the readout (Table 2). The data indicated that in the presence of raltegravir, the CMV, CEFT, and Gag peptide pools each reversed latency substantially. The activity of the Gag peptide pool (64% of maximum-stimulation) measured by the ca-RNA (Tat/Rev mRNA) readout in the presence of raltegravir was consistent with that measured using the cf-RNA (virion-RNA) readout (35% of maximum-stimulation). In contrast, when raltegravir was omitted, only HIV‑1 p55 protein yielded detectable induction of Tat/Rev mRNA: stimulation with p55 yielded 7.5 x 106 copies/ml compared to 1.2 x 109 copies/ml after maximum stimulation. Overall, these results suggested that antigen presentation by DCs did not markedly increase the degree or consistency of latency reversal by these peptides and proteins. The results also suggested that prolonged culture can change the conclusions of these latency reversal experiments relative to short-term, "single-cycle" readout, either amplifying or reducing the signals.