Activation of a dendritic cell–T cell axis by Ad5 immune complexes creates an improved environment for replication of HIV in T cells

Ad5-specific memory T cells and Ad5 NAbs are an important component of Ad5 preexisting immunity (11), and interaction between Ad5 vector and NAbs certainly occurs rapidly after the administration of vectors in vivo. Ad5 IC are also well characterized in vitro (12, 13). Thus, we investigated the effects of Ad5 IC on DCs. For these purposes, we generated Ad5 IC by mixing Ad5 vectors with sera containing Ad5 NAbs. Formation of ICs was determined by assessing C1q (complement protein 1q) binding (14). To reproduce the scenario in the STEP trial, the Ad5 vectors used in the present study are also E1/E3 deleted. To preclude the possibility that the effects were caused by complement activity, heat-inactivated sera (30 min at 56°C) were used throughout this study. The effects of Ad5 IC on DCs were assessed initially by analyzing the induction of the expression of costimulatory molecules (CD40 and CD86). Consistent with previous studies (15), Ad5 vectors alone did not induce a substantial increase in the expression of costimulatory molecules by immature DCs (Fig. 1, a–c). The dose of Ad5 vector used, 2.5 × 10⁴ physical particles (pp)/cell was consistent with that used in previous studies (11, 16). To determine the effects of Ad5 IC, ICs were generated by progressively increasing the volume of serum that contained high Ad5 NAbs titers (∼512) in the presence of a fixed dose of Ad5 vector. The neutralizing activity of the sera was confirmed by determining the levels of Ad5 vector transduction as previously described (14) and using DCs. The percent of AdGFP-transduced DCs was inversely proportional to the volume of serum, thus demonstrating the presence of Ad5 NAbs (Fig. S1).

Up-regulation of CD86 expression was observed 48 h after treatment of immature DCs with Ad5 IC, and the up-regulation paralleled the increase in serum volume and thus in Ad5 IC formation (Fig. 1 a). Similarly, up-regulation of CD86 was also observed using a fixed volume of serum and increasing the amount of Ad5 vector (Fig. 1 b). No significant increase in the expression of CD86 was observed in the presence of serum or with Ad5 vector alone (Fig. 1, a–d). Significantly (P < 0.05), increased expression of CD40 and CD86 were observed in the presence of Ad5 IC and not in the presence of Ad5 vector alone or after treatment of DCs with Ad5 plus a nonneutralizing serum (Fig. 1, c and d).

To test the functional maturation of DCs induced by Ad5 IC, we assayed their ability to take up antigens (Ags). Immature DCs efficiently capture Ags primarily through macropinocytosis and mannose receptor-mediated endocytosis, but they lose this function during maturation (17). Immature DCs and DCs treated with Ad5 vector alone, with Ad5 IC, or with LPS were incubated with FITC-labeled dextran, and the level of Ag uptake was evaluated as a function of time (Fig. 1 e). Consistent with functional immaturity, mock-treated immature DCs maintained their ability to internalize the dextran. DCs incubated with Ad5 vector alone induced modest DC maturation and, indeed, their ability to Ag uptake was partially preserved. Consistent with the induction of maturation, Ag uptake by DCs treated with Ad5 IC or LPS was lost.

To better define the effects of Ad5 IC on activation and maturation of DCs, we determined the cytokine profile of DCs treated with Ad5 vector, sera ± NAbs, Ad5 incubated with a nonneutralizing serum (NNS) or Ad5 IC. The panel of cytokines investigated included TNF-α, IL-12p70, IL-10, IFN-I, IFN-γ, and IL-1β. DCs treated with Ad5 vector, Ad5/NNS, or neutralizing serum, did not induce the release of any significant levels of the cytokines tested (Fig. 2, a and b; and not depicted). In contrast, when assaying Ad5 IC our results were consistent with those found when testing the effects of Ad5 IC on the induction of the expression of costimulatory molecules (Fig. 1). Progressive increments in secretion of TNF-α and IFN-I from DCs were observed either at a fixed dose of Ad5 vector together with increasing volume of serum or vice versa (Fig. 2, a and b). No secretion of the other investigated cytokines was observed (unpublished data). We then investigated the relationship between the NAb titers and the levels of cytokines secretion. Interestingly, we found a significant correlation (P < 0.05) between the levels of TNF-α and IFN-I secreted and the Ad5 NAb titers (Fig. 2, c and d). Collectively, these results indicate that Ad5 IC induced greater activation and maturation of DCs and, more importantly, the anti-Ad5 NAb titer was a key determinant in the induction of DCs activation.

According to previous studies, cross-linking of the FcγRs by immobilized Igs induces expression of costimulatory molecules (18). To better define the role of FcγRs, DCs were treated with Ad5 IC and the induction of TNF-α and IFN-I secretion was determined in the presence or absence of blocking Abs against FcγRI, IIa, and III. The three FcγRs were differently expressed on DCs with the FcγRIIa the most expressed (Fig. S2). We found that blocking FcγRI, III, and particularly IIa reduced the secretion of TNF-α and IFN-I (Fig. 3 a). These data demonstrate that FcγRs are involved in delivering an Ad5 IC–derived activation signal to DCs. To further exclude the effects of unknown serum components, we used purified Igs (14) to generate Ad5 IC and confirmed the efficient induction of DC activation (unpublished data). The purified Igs were predominantly IgG (IgM and IgA were a minor fraction), and IgG₁ and IgG₂ were the predominant isotypes (unpublished data).

Signaling through FcγRs may be caused by either direct cross-linking or by downstream events of receptor-mediated internalization. To discriminate between these two possibilities, DCs were cultured in wells coated with purified Ad5-specific Abs. No significant secretion of TNF-α, IL-12p70, IL-10, IFN-I, IFN-γ, or IL-1β was observed (unpublished data), thus suggesting that cross-linking of FcγRs is not an activation signal sufficient to induce the release of cytokines in DCs despite the fact that it induces expression of costimulatory molecules. Therefore, these results suggested that downstream events of receptor-mediated Ad5 IC internalization were involved in the delivery of activation signals.

On the basis of the observations detailed in the previous section, it is possible that FcγR-mediated Ad5 IC internalization preferentially delivers some Ad components (proteins, virus-associated RNA, or the double-stranded DNA genome) to distinct subcellular compartments. To test this hypothesis, we incubated DCs with a helper-dependent Ad5 vector alone or complexed with sera containing NAbs (helper-dependent [HD] IC). Although the genomes of helper-dependent vectors are deleted in all viral genes, the virions contain the same core and capsid proteins and virus-associated RNAs (19). We found that neither the helper-dependent vector alone nor the HD IC promoted the release of TNF-α or IFN-I (Fig. 3 b), suggesting that the genome of the E1/E3-deleted Ad5 vector harbors critical signals for complete DC activation and maturation.

In this regard, nucleic acids of viral or mammalian origin can be recognized by intracellular pathogen recognition receptors. For example, human lupus Ab–DNA complexes activate DCs through the interaction between FcγRII and TLR9 (20). Furthermore, several potential immunoregulatory sequences (IRSs) can inhibit the release of IFN-I after stimulation with TLR9 and TLR7 agonists (21). Among these, IRS 869 mediated the most potent inhibition of IFN-I secretion in TLR9-stimulated plasmacytoid DC (21). We therefore asked whether IRS 869 could suppress TNF-α and/or IFN-I production induced by Ad5 IC. We found that IRS 869 suppressed both TNF-α and IFN-I production after stimulation with Ad5 IC (>90% inhibition; Fig. 3 c). LPS, which stimulates DCs through the cooperation of CD14 with TLR4 (22), and CpG (TLR9 agonist) (21) were used as controls. IRS 869 did not inhibit TNF-α or IFN-I production by DCs stimulated with LPS, whereas it inhibited cytokine production of DCs stimulated with CpG (Fig. 3 c).

Consistent with previous studies (23, 24), our data suggest that Ad5 IC are internalized via FcγRIIa and are likely delivered to intracellular compartments where the vector genome interacts with TLR9, which in turn leads to activation of DCs. Importantly, these data argue against a major role for Ad capsid proteins and RNA.

To investigate the effects of Ad5 IC on the stimulation/activation of Ad5-specific CD4 and CD8 T cells in Ad5 seropositive subjects, we developed an ex vivo stimulation assay of Ad5-specifc T cells that may mimic those operating in vivo. DCs were generated from five subjects with high (>1,000) Ad5 NAb titers. Immature DCs were incubated with Ad5 vector alone or with Ad5 IC for 24 h. Two sera were used to generate Ad5 ICs for each DC population, autologous serum (from the donor-providing DCs) and heterologous serum with high (4,096) Ad5 NAb titer. At the end of the incubation period, Ad5 vector or Ad5 IC–treated DCs were used to stimulate Ad5-specific CD4 and CD8 T cells from autologous blood mononuclear cells. We then assessed secretion of IL-2, IFN-γ, and TNF-α (6 h after stimulation) using polychromatic flow cytometry. No significant difference (P > 0.05) was observed in the percent of total, IL-2, IFN-γ, and TNF-α–secreting Ad5-specific CD4 T cells in blood cells stimulated with Ad5 vector– or Ad5 IC–treated DCs (Fig. 4 a). In contrast, a highly significant (P < 0.001) approximately threefold difference was found in the percentage of total cytokine-secreting Ad5-specific CD8 T cells stimulated with Ad5 IC–treated DCs compared with Ad5 vector–treated DCs (Fig. 4 a). Furthermore, major changes in the cytokine profile of Ad5-specific CD8 (Fig. 4 b), but not CD4, T cells (Fig. S3) were observed between the two stimulation conditions. In particular, we detected a significant (P < 0.05) increase in the dual IFN-γ⁺/TNF-α⁺ (Fig. 4 b) and a significant (P < 0.05) decrease in the IL-2–secreting (IL-2⁺/IFN-γ⁺/TNF-α⁺ IL-2⁺/TNF-α⁺ and single IL-2⁺; Fig. 4 b) CD8 T cell populations. With regard to Ad5-specific CD4 T cells, the cytokine profile was modestly different between Ad5 vector alone and Ad5 IC–treated DCs These results indicate that Ad5 IC induced stimulation of higher frequencies of Ad5-specific memory CD8 T cells with an effector cytokine profile.

We next determined the effects of Ad5 IC on the proliferation capacity of Ad5-specific memory T cells and consistently observed a trend toward increase proliferation (about twofold) of CD4 and CD8 T cells stimulated with Ad5 IC–treated DCs, although these differences did not reach statistical significance (P > 0.05; Fig. S4). We also determined the cytotoxic profile of the proliferating (CFSE low) Ad5-specific CD8 T cells based on the expression of perforin and granzyme B. The proportion of Ad5-specific CD8 T cells expressing perforin and granzyme B was ∼2.5-fold greater (P < 0.05) when Ad5 IC–treated DCs were used as stimuli (Fig. 4 c).

Collectively these results indicate that Ad5 IC induced substantial anti-Ad5 vector effector CD8 T cells and a trend toward a larger expansion of Ad5-specifc T cells. The more effective stimulation of Ad5-specific CD8 T cells likely results from the capacity of DCs to efficiently process and present exogenous Ags, such as IC, to CD8 T cells via cross-presentation (24).

It was critical to investigate whether the increased T cell activation induced by Ad5 IC enhanced HIV infection. To address this issue we used an ex vivo model that likely mimics the cellular interactions that occurred during the STEP trial. We stimulated DCs with Ad5 vector or Ad5 IC and then incubated these cells with autologous blood mononuclear cells from four subjects to reactivate memory Ad5-specific T cells. Autologous or heterologous (4,096 titer) NS sera were used to generate Ad5 IC. The cocultures were then incubated with HIV-1_LAV. Ex vivo HIV infection and propagation was assessed by measuring p24 in the culture supernatants. It is important to mention that DCs, through the formation of conjugates with CD4 T cells, facilitate productive HIV-1 infection in cell cultures even in the absence of polyclonal or Ag-specific activation of T cells (25). Indeed, we observed productive HIV-1 infection in cell cultures containing mock-treated DCs mixed with autologous blood mononuclear cells (Fig. 5). Nonetheless, we consistently observed approximately threefold higher levels of HIV replication in the cell cultures containing blood mononuclear cells stimulated with DCs treated with Ad5 IC as compared with Ad5 vector or cultured with DCs alone (Fig. 5). The mean p24 level at day 4 was 3,230 pg/ml in the culture of blood mononuclear cells stimulated with DCs treated with Ad5 vector alone versus 10,800 pg/ml in the culture of blood mononuclear cells stimulated with DCs treated with Ad5 IC (P < 0.05). At day 7, the mean p24 level was 64,400 pg/ml in the culture of blood mononuclear cells stimulated with DCs treated with Ad5 vector alone versus 207,600 pg/ml in the culture of blood mononuclear cells stimulated with DCs treated with Ad5 IC (P < 0.001). Therefore, Ad5 IC likely induced greater activation of T cells that translated into enhanced levels of HIV replication in tissue culture.

Our study advances our understanding of the likely immunological events operating after Ad5 vector-based vaccination and suggests a series of events that could explain the increase in acquisition of HIV infection in Ad5 seropositive subjects in the STEP trial. During Ad5 vector–mediated gene transfer, the vectors likely infect several target cells, including professional Ag-presenting cells like DCs. Infection of these latter cells will eventually activate memory Ad5-specific CD4 and CD8 T cells and hopefully generate an HIV-specific T cell response. However, the presence of high Ad5 NAbs titers is associated with the formation of Ad5 IC. Compared with Ad5 vector alone, Ad5 IC induced “hyperactivation” and maturation of DCs, as indicated by the increased expression of costimulatory molecules, the loss of the capacity of Ag uptake, and the secretion of TNF-α and IFN-I. The activation signal to DCs mediated by Ad5 IC results in part from the cooperation between FcγRs and TLR9.

The survival signals and clonal expansion of memory T cells mediated by IFN-I (26, 27) may explain the effect of Ad5 IC on the increase of Ad5-specifc memory effector CD8 T cells, likely through cross-priming and the trend toward larger expansions of both Ad5-specific memory CD4 and CD8 T cells. The reactivation of Ad5-specific memory T cells likely prevents effective generation of the primary immune response against the vector-encoded HIV Ags through two nonexclusive mechanisms: one, an unfavorable cytokine environment resulting from the inflammatory response caused by the Ad5 IC formation; and two, the killing of DCs. Because DCs will express not only HIV but also Ad5 Ags (28), they become a target of the reactivated Ad5-specific CD8 T cells, thus resulting also in the reduction of the DC pool presenting HIV Ags. In support of this hypothesis, preliminary data from the STEP trial indicate that the percent of responders to HIV proteins (measured by IFN-γ ELISpot) was less (∼25%) in subjects with Ad5 NAb titers >200, and the magnitude of the HIV-specific T cell response (frequency of IFN-γ–secreting cells in blood) was also reduced by ∼50% (www.HVTN.org). Finally, increased survival of activated memory Ad5-specific CD4 T cells mediated by IFN-I (26) may transiently enlarge this pool of memory CD4 T cells that effectively support HIV replication and spreading, thus facilitating susceptibility to HIV infection. Our results also provide new insights into the type of experimental strategies and assays that can be instrumental for the preclinical assessment of the safety of other viral vector-based vaccines in addition to Ad5.

In conclusion, we have shown that ICs containing E1/E3-deleted Ad5 vectors and NAbs cause a pattern of innate/adaptive inflammatory activation that, together with a possible persistence of the Ad5 vector, may provide the basis for a chronic permissive environment for HIV-1 infection, thus helping to explain the increased acquisition of HIV-1 infection among the Ad5 seropositive vaccine recipients in the STEP trial. Moreover, our results also suggest that the delivery/formulation of Ags through ICs may result in a powerful immunization strategy to stimulate T cells.

Blood mononuclear cell isolation (from 14 healthy subjects) and monocyte-derived DC generation were performed as previously described (15).

Two E1/E3-deleted Ad5 vectors (Adβgal and AdGFP) have been previously described (15). Adβgal harbors a lacZ expression cassette and was used in T cell stimulation, Ag uptake, Fc receptor blocking, in vitro HIV infection, and costimulatory molecule experiments. AdGFP harbors a GFP expression cassette and was used for the assessment of NAb titers, in DC transduction assays, and in costimulatory molecule experiments. The helper-dependent Ad5 vector (19) is deleted in all viral genes and contains a lacZ expression cassette (provided by M. Castro, Gene Therapeutics Research Institute, Cedars Sina Medical Center, University of California, Los Angeles, Los Angeles, CA). Ad5 vector titers were measured in virus pp/cell as described by Mittereder et al. (29).

The neutralizing activity of a panel of sera (n = 76) was determined by transduction inhibition assays (14) using the 911 cell line. With regard to the DC transduction assay, immature DCs (2.5 × 10⁵ cells) were resuspended in medium containing 2.5 × 10⁴ pp/cell and 0.016–20 μl of serum, incubated for 48 h, and then analyzed by flow cytometry.

DCs from 14 donors were incubated with AdGFP, Ad5 IC, or 50 ng/ml LPS (Escherichia coli 0127 B5; Sigma-Aldrich) for 48 h. Supernatants were collected and analyzed for the presence of TNF-α, IL-12p70, IL-1β, IFN-γ, and IL-10 by ELISA (BD). IFN-I quantification was performed as previously described (15) using cells that contain a luciferase expression cassette under the control of IFN-I–inducible promoter. The NAb titers and number of sera analyzed for each titer were the following: 0, 1; 4, 3; 8, 2; 16, 5; 32, 1; 64, 6; 128, 4; 512, 2; and titer 1,024, 4 sera.

Ag uptake was determined using FITC-labeled dextran (MW 10,000; Sigma-Aldrich) as previously described (15). In brief, immature DCs were incubated with Adβgal, Ad5 IC, or 50 ng/ml LPS for 48 h. DCs were then washed and incubated with 1 mg/ml FITC-labeled dextran for 10, 30, or 90 min. Control DCs were incubated at 4°C. Assays were performed twice.

The following monoclonal Abs were used for functional assays: CD16 (FcγRIII; 3G8; BD), CD32a (FcγRIIa IV.3), CD32 (FcγRII AT10), and CD64 (FcγRI; 10.1) (20).

DCs were cultured either alone or with Adβgal vector or Adβgal mixed with 20 μl of neutralizing serum and 50 ng/ml LPS for 48 h. The expression of costimulatory molecules was assessed on HLA-DR–positive cells using an HLA-DR Alexa Fluor 700–conjugated Ab (LN3; eBioscience). Five DC donors and eight sera were used.

DCs were first treated for 24 h with Ad5 vector alone or with Ad5 IC. Cell cultures were then extensively washed, mixed with autologous blood mononuclear cells (10⁶ cells), and cultured for 6 h in 1 ml of complete media containing 1 μl/ml GolgiPlug (BD). Intracellular cytokine staining was performed as previously described (30). The Abs used were the following: CD4-APC-H7, CD8-PerCP-Cy5.5, CD3-ECD, IFN-γ–AF700, IL-2–PE, and TNF-α–PE-CY7. Data were acquired on an LSRII three laser (488, 633, and 405 nm; BD) and analyzed using FlowJo and SPICE 4.1.5 software (M. Roederer, Vaccine Research Center, NIAID, Bethesda, MA).

DCs were first treated for 24 h with Ad5 vector alone or with Ad5 IC. Cell cultures were then extensively washed, mixed with autologous blood mononuclear cells, and cultured for 6 d. Blood mononuclear cells proliferation was assessed using CFSE (Invitrogen) as previously described (9, 30). Cells were stimulated with Ad5 IC formed with autologous NS and heterologous serum. Stimulation with 100 ng/ml Staphylococcus aureus enterotoxin B was used as a positive control. All sera tested had Ad5 NAb titers >1,000. The percent of proliferating CD4⁺ and CD8⁺ T cells, i.e., CFSE-low cells, was determined in the CD3⁺ cell population. The criteria for scoring as positive the proliferating cell cultures included both CFSE-low cells >1% after subtracting the percent of CFSE-low cells in unstimulated cell cultures and stimulation index >3. The stimulation index was calculated by the fold increase between stimulated versus unstimulated cell cultures. Expression of perforin and granzyme B expression on CFSE-low proliferating CD8 T cells was performed using APC-conjugated anti-perforin (BioLegend) and AF700-conjugated anti-granzyme B (BD).

DCs were first treated for 24 h with Ad5 vector alone or with Ad5 IC. Cell cultures were then extensively washed, mixed with autologous blood mononuclear cells, and cultured for 48 h. Cell cultures of 10⁶ cells were then inoculated with 30 pg HIV-1_LAV for 3 h. At the end of the incubation, cells were washed, replated at 10⁶ cells per well, and HIV replication was assessed by measuring p24 in the culture supernatant. Mock-infected cultures were used as a negative control. HIV infection was also determined in cultures in which blood mononuclear cells were mixed with untreated DCs. p24 measures were performed in triplicates and measurements were performed at days 0, 4 and 7 using an ELISA (Immunogenetics).

Phosphorothioate-protected IRS 869 and control oligonucleotide (ODN; Sigma-Aldrich) were resuspended to make a final concentration of 500 μM. TLR9 immunostimulatory sequence 2216 was purchased from InvivoGen. In brief, 2.5 × 10⁵ immature DCs were seeded in 500 μl in 48-well plates and treated with IRS 869 or control ODN (final concentration of 5 μM) for 2 h. Immunostimulatory sequence 2216 (also at 5 μM) or Ad5 IC prepared using 20-μl serum (NAbs titer >1,000) for 15 min at room temperature were then incubated with the DCs for ∼18 h at 37°C in 5% CO₂. The supernatants were collected at the indicated times and tested for TNF-α or IFN-I.

Statistical significance (p-values) was calculated by using a Student's t test for the comparisons, a Spearman rank test for the correlations, and a Wilcoxon rank test for the SPICE analyses. A two-tail p-value of <0.05 was considered significant. The analyses of multiple comparisons have been taken into account for the calculation of statistical significance.

The four figures show data on Ad5 vector transduction efficiency, expression of different FcγRs on DCs, and effects of Ad5 IC stimulation on CD4 T cells and T cell proliferation.

We are grateful to the Etablissement Français du Sang (Montpellier, France), to L. Poupon for technical help, M. Castro for the helper-dependent vector, and M. Pelegrin, H.A. Michaud, F. Mennechet, V. Pinet, and P. Lebon for discussion.

Part of this work was funded by the Association Française contre les Myopathies, Vaincre les Maladies Lysosomales, La Region Languedoc Roussillion and the Centre National de la Recherche Scientifique.

The authors have no conflicting financial interests.