Adenovirus-Transduced Dendritic Cells Injected into Skin or Lymph Node Prime Potent Simian Immunodeficiency Virus-Specific T Cell Immunity in Monkeys ¹

Dendritic cell (DC) ³-based immunotherapy trials have been initiated for a wide range of human malignancies, and interest is growing in using DC-based approaches to control infectious causes of chronic disease, including HIV (1, 2, 3). Immunotherapeutic approaches to HIV infection in particular face unique hurdles, as the virus has a strong propensity to mutate and escape from cytotoxic T cell-mediated control (4). Thus, a central aim of DC-based therapy for HIV is to increase the strength and breadth of virus-specific T cell immunity (5). Ideally, this would include induction of T cell responses to novel epitopes that may not be detected during natural infection, but may be important in controlling virus. A key component in achieving this goal is the selection of Ag for DC loading.

Numerous clinical trials in the cancer field have used DC pulsed with tumor-associated peptides as vaccines (6, 7, 8, 9, 10, 11). However, this method is limited by existing knowledge of epitopes and patient MHC molecules, and may not satisfactorily induce broad T cell responses, as a finite number of peptide epitopes can be used. More complex antigenic sources such as inactivated virus particles have the potential to provide a wide range of viral Ag for processing by DC. Indeed, treatment of SIV-infected monkeys with virus-loaded DC had a significant therapeutic benefit, although the breadth of immune response was not evaluated (12). However, inactivated virus particles carry significant concerns relating to safety and production that may limit the enthusiasm for clinical trials. An attractive alternative to these and other sources of Ag is the use of recombinant viral vectors to deliver complex viral or tumor Ag to DC (13). Replication-defective viral vectors can be engineered to express full-length proteins containing known and unknown epitopes, and transduced DC can be used as vaccines without concern for patient MHC haplotype expression. Moreover, given the appropriate facilities, recombinant viral vectors can readily be produced to clinical grade for therapeutic application. Canarypox vectors may be unsuitable for DC-based vaccination, as they induce apoptosis in transfected immature DC (14). In contrast, DC remain healthy after transduction with recombinant adenoviral vectors and stimulate robust CD8⁺ T cell responses to viral transgenes in vitro (15, 16, 17). Murine studies have demonstrated that adenovirus-transduced DC induce protective antitumor responses, even in the presence of virus-specific immune responses (18, 19). These findings support the development of DC-based immunotherapy protocols using recombinant adenovirus as an Ag source.

To test the potential for adenovirus-transduced DC to stimulate Ag-specific T cell responses to HIV, we immunized normal rhesus macaque monkeys with DC-expressing SIV Gag Ag via infection with recombinant adenovirus. Monkeys that were SIV naive were chosen to evaluate the strength and breadth of T cell response to vaccination in the absence of any pre-existing T cell immunity to virus, and to determine whether vaccination could prime responses to novel epitopes not previously identified through SIV infection. To determine the optimal means of immunization, DC were delivered via intradermal or intranodal injection. Finally, the effect of repeated vaccination on vector-specific immune responses and DC trafficking was evaluated.

Nine SIV-naive adult rhesus macaques (Macaca mulatta) were used in this study and housed at the University of Pittsburgh Primate Facility for Infectious Disease Research in compliance with the appropriate institutional regulatory committees. Molecular MHC class I typing was done through a contract with the Wisconsin National Primate Research Center (Madison, WI). All animals had undetectable titers of neutralizing Ab to adenovirus serotype 5 before vaccination (data not shown).

E1/E3-deleted adenovirus serotype 5 vector was used to generate two viruses, Adp17 and Adp45, expressing p17 MA protein and the remaining p45 region of SIVmac239 Gag, respectively (47). Briefly, overlapping primers designed using an algorithm to introduce Kozak consensus sequences, optimize codon usage, and eliminate AU-rich elements were used in repeated cycles of PCR to generate gag p17 and p45 cDNA based on the published SIVmac239 sequence. A SalI-NotI fragment containing the p17 or p45 sequence was inserted into the pAdlox shuttle vector, with subsequent cotransfection with ψ5 helper virus into the CRE8 packaging cell line to generate recombinant replication-defective adenovirus, as described previously (20).

DC were cultured from purified blood monocytes of normal monkeys, as described (21). For adenoviral infection of DC, viruses were added directly to cells in culture at a multiplicity of infection of 200 and incubated for 45 min at room temperature in a reduced volume of medium. Cells were cultured for 24 h, washed to remove free virus, and then cultured with GM-CSF and IL-4 and recombinant human trimeric CD40 ligand (1 μg/ml; Immunex, Seattle, WA) for an additional 16 h (21). Alternatively, for infection before analysis by Western blot, HEK 293 human embryonic kidney cells or DC were infected in a pellet with adenoviruses at a multiplicity of infection of 10 or 100, respectively, and then cultured for 48 h.

Cell lysates were separated by gel electrophoresis and transferred to polyvinylidene difluoride membrane (Bio-Rad, Hercules, CA). Gag p17 and p45 were detected by probing with mAb KK59 and 2F12, respectively (National Institutes of Health AIDS Research and Reference Reagent Program, National Institutes of Health, Bethesda, MD), followed by HRP-conjugated anti-mouse IgG and development using an ECL kit (Amersham Pharmacia Biotech, Piscataway, NJ). For immunofluorescence analysis, infected DC were settled onto glass slides and processed, as described (22). Cells were incubated with primary Ab overnight, followed by goat anti-mouse IgG conjugated to Alexa 488 (Molecular Probes, Eugene, OR) for 2 h. Nuclei were labeled with Hoescht 33342, and slides were coverslipped and imaged using a Zeiss (Oberkochen, Germany) Axiovert 135 microscope.

Blood volumes of 50–100 ml were taken from each animal depending on body size for DC culture. Mature DC expressing Gag p17 and p45 proteins were combined in one syringe in a volume of 200 μl saline for injection. For intranodal injections, cells were administered under the capsule of the inguinal or axillary lymph nodes using a 27 G needle following surgical exposure. For intradermal injections, cells were injected into skin overlying the inguinal or axillary lymph node using a 27 G needle.

Ninety-six-well membrane-coated plates (Multiscreen-IP; Millipore, Bedford, MA) were incubated with 10 μg/ml mAb to rhesus IFN-γ (MD-1; U-Cytech, Utrecht, The Netherlands) in 0.1 M carbonate buffer overnight. Previously frozen PBMC were thawed and plated at 1 × 10⁵ to 2 × 10⁵ cells/well in medium containing 10% human AB serum. In some experiments, CD8⁺ T cells or CD4⁺ T cells were depleted from PBMC using microbeads coated with Ab to CD8 (SK1; Miltenyi Biotec, Auburn, CA) or CD4 (M-T321; Miltenyi Biotec), respectively. Depletions were confirmed to be greater than 90% by flow cytometry. Individual 15-mer peptides overlapping by 11 aa representing the Gag sequence of SIVmac239 (National Institutes of Health AIDS Research and Reference Reagent Program) were dissolved in DMSO at 10 mg/ml and used in pools of 30–32 peptides (final concentration 3.1–3.3 μg/ml) or 8 peptides (3.9 μg/ml), or as individual peptides (5 μg/ml). Peptides, DMSO, or staphylococcal enterotoxin B (1 μg/ml; Sigma-Aldrich, St. Louis, MO) were added to cells and incubated at 37°C for 24 h. Alternatively, autologous DC were infected with Adp45 or Adψ5 and matured with CD40 ligand for use as stimulators with PBMC at a ratio of 1:10. Wells were washed free of cells and then incubated with successive overnight incubations of biotinylated Ab to monkey IFN-γ (10 μg/ml; U-Cytech) and streptavidin-alkaline phosphatase conjugate (Bio-Rad). Spots were developed using chromogenic alkaline phosphatase substrate (Bio-Rad) and enumerated using an AID ELISPOT reader (Cell Technology, Columbia, MD).

Cells were infected with adenovirus-encoding enhanced green fluorescence protein (EGFP) and matured with CD40 ligand before injection under the capsule of the axillary lymph node, as described above. Twenty-four hours after injection, the lymph node was excised and prepared, as described (23). Sections were stained with mouse mAb to p55 (55K-2; DAKO, Carpenteria, CA) or CD83 (HB15a; Immunotech, Miami, FL), followed by Cy3-conjugated goat anti-mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA) and rabbit anti-human CD3 (DAKO), followed by Cy5-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories). Slides were imaged using an Olympus Fluoview BX61 laser-scanning microscope (Melville, NY).

We first analyzed protein expression by DC following transduction with recombinant adenoviruses expressing SIV Gag proteins. Immature DC infected with Adp17 expressed the Gag p17 protein, as determined by Western blotting of cell lysates with the SIV p17-specific mAb KK59. A protein of similar size and specificity was detected in Adp17-transduced HEK 293 cells (Fig. 1,a). Lysates of DC infected with the control vector Adψ5 did not label with KK59 (Fig. 1,a). Similarly, DC and HEK 293 cells infected with Adp45, but not DC infected with adenovirus encoding the tumor Ag gp100 (20), expressed a 45-kDa protein that labeled with the SIV p27-specific mAb 2F12 (Fig. 1,a). To determine the relative efficiency of transduction of DC, we did immunofluorescence analysis of adenovirus-infected cells. The majority of DC infected with Adp17 stained brightly with KK59, but not with 2F12 (Fig. 1,b). Conversely, DC infected with Adp45 stained with 2F12, but not with KK59 (Fig. 1 c). These data indicate that adenovirus-transduced DC express the appropriate Gag proteins at a high efficiency, consistent with previous reports (17).

We next immunized two groups of animals with a series of injections of Gag-expressing DC either by intradermal or intranodal injection. In addition, one animal received two intranodal injections of autologous DC infected with control Adψ5 vector (Table I). We used in vitro matured DC as opposed to immature DC, as the latter have been shown to inhibit Ag-specific T cell function in vivo (9, 24). T cell responses were measured using pools of overlapping peptides pulsed onto uncultured PBMC without expansion to measure effector as opposed to memory T cell function. All animals responded to vaccination, as determined by IFN-γ release, with peak responses occurring from 2 to 8 wk after the first immunization (Fig. 2). Responses were generally directed against more than one peptide pool, indicating a broad T cell response. However, few animals had p17-specific responses following vaccination, as measured by IFN-γ release in response to peptides p1-p31 (Fig. 2), despite the fact that the p17 protein was expressed by approximately one-half of all cells injected (Table I). As expected, animal M2598, which received intranodal injections of DC infected with the control Adψ5, had background responses to Gag following immunization (Fig. 2). Surprisingly, the mean T cell response of animals in the intranodal and intradermal vaccination groups to peptide pools spanning the Gag protein was of a similar specificity and intensity, suggesting that immunization by either route is effective at inducing T cell immunity using transduced DC (Fig. 3).

To determine the breadth of T cell response at the individual peptide level, we examined the responses of animals M13601 and M1901 in detail. The peak overall frequency of Gag-specific T cells in these animals was very high at ∼1/500 uncultured PBMC, despite the fact that they were immunized by different routes (Fig. 4,a). As for other animals, the peak response to vaccination in M13601 and M1901 followed the first or second vaccination, and boosting responses were generally limited (Fig. 4,a). To identify specific T cell epitopes, we used peptide pools in a matrix design (25, 26), focusing on peptides p32-p95 based on the dominant response to this region in the bulk assays (Fig. 4,a). Sixteen pools of eight peptides each were generated, in which overlapping peptides were not contained in the same pool and each peptide was represented in two pools. By testing peptides in this matrix design, responder peptides can be identified out of a larger pool and subsequently tested to confirm reactivity (26). Using this method on PBMC from animal M13601 at wk 11, we identified 4 responder peptides from the total of 64, being p35 (Gag_137–151), p36 (Gag_141–155), p94 (Gag_373–387), and p70 (Gag_277–291) (Fig. 4,b). A similar, but less intense response was noted at wk 28 in the same animal, demonstrating a sustained response to vaccination (Fig. 4,b). Peptides p35 and p36 share a common 11-aa stretch (GGNYVHLPLSP) that most likely contains a single immunodominant sequence. By a similar matrix analysis, we identified specific responses in PBMC from M1901 at wk 2 to peptides p35 and p66 (Gag_261–275) (Fig. 4,b). Examination of PBMC from this animal at wk 20, 4 wk after the boost at wk 16, showed positive responses to the same two peptides, demonstrating that boosting resulted in expansion of the same peptide-specific T cells as stimulated by initial vaccination (Fig. 4,b). At least one other peptide epitope was also recognized by M1901 at wk 2, based on the response to pool p96-p125 in the bulk assay, although response to this pool was reduced at wk 20 (Fig. 4,b). To determine whether these T cell responses were mediated by CD4⁺ or CD8⁺ T cells, we did Ab depletion experiments. IFN-γ responses of PBMC from M13601 specific for peptides p36 and p94 were mediated by CD8⁺ T cells, as depletion of CD8⁺ T cells reduced the mean response to 20 and 13%, respectively, compared with that of unseparated PBMC, whereas depletion of CD4⁺ T cells did not change significantly the response relative to PBMC for each peptide (Fig. 4,c). Similarly, depletion of CD8⁺ T cells, but not CD4⁺ T cells, abolished the IFN-γ response of M1901 to peptide p35 (Fig. 4 c). These findings indicate that the Gag-specific response to DC-based immunization in these animals was mediated by CD8⁺ T cells.

The effectiveness of direct adenoviral delivery of tumor Ag to cancer patients is substantially limited by existing neutralizing Ab responses to the vector (27), and studies in mice suggest that immune responses generated by repeated injections of adenoviral vectors may limit gene therapy in a lung model (28). To determine whether immunity to adenovirus limited DC-based vaccine responses in our nonhuman primate model, we analyzed vector-specific responses in M13601 before and after vaccination. To evaluate the T cell response to adenovirus, we transduced DC with Adp45 or the control Adψ5 vector and used these cells as stimulators for PBMC in the ELISPOT assay. DC infected with either Adp45 or Adψ5 stimulated a frequency of IFN-γ-secreting cells substantially greater than did untransduced CD40 ligand-matured DC in PBMC collected before vaccination (Fig. 5,a). This is most likely due to immunostimulatory effects of adenoviral infection on DC (29) rather than pre-existing cross-reactive T cell responses to adenovirus, given the lack of contiguous stretches of homologous sequences in simian and human adenoviruses (30). The response to DC infected with the control vector was unchanged at wk 26, while the response to DC infected with Adp45 was significantly increased, demonstrating induction of Gag-specific, but not vector-specific, T cell responses to vaccination (Fig. 5,a). Similarly, PBMC at wk 26 were responsive to the peptide pool p64-p95 (Fig. 5,a), and immunization at this time induced a modest increase in Gag-specific T cells (Fig. 4 a). Surprisingly, while T cell responses to the vector appear to be limited, low titers of neutralizing Ab specific for adenovirus serotype 5 were induced in three of four animals given intranodal DC (data not shown). In animals M1699 and M11300, these responses were sustained, with neutralizing Ab to adenovirus serotype 5 still detectable at 14 wk following the final boost at wk 16 (data not shown).

To gain further insight into any potential influence previous vaccination may have on the function of adenovirus-transduced DC in vivo, we analyzed the capacity for adenovirus-transduced DC to localize to the T cell-rich paracortex following injection in an animal that had received multiple prior vaccinations. Mature DC transduced with AdEGFP were injected into the axillary lymph node of M1401 at wk 45, following a complete intradermal vaccination course. EGFP-expressing DC were identified within the injected lymph node paracortex 24 h after injection in a similar location to endogenous interdigitating DC, identified by labeling with mAb to p55 (31) (Fig. 5,b). Importantly, injected EGFP-expressing DC associated with and interposed between CD3⁺ T cells in the paracortex and maintained expression of the activation marker CD83 (Fig. 5, c and d). Injected cells residing in the lymph node appeared healthy, with no evidence of cell death 24 h after injection (Fig. 5 d). These results indicate that repeated injections of adenovirus-transduced DC do not impair the capacity of these DC to localize to the T cell-rich area of lymph nodes.

This study provides the first validation in a primate model for the use of adenoviral vectors to induce robust T cell immunity to a virus via DC, and supports the findings of previous in vitro and murine studies showing the efficacy of this approach (17, 18, 19, 32). Vaccination generated potent SIV Gag-specific CD8⁺ T cell responses in naive animals with effector T cells reaching an unprecedented frequency of 1/500 freshly isolated PBMC after a single injection. When examined at the level of individual peptide epitopes, the frequency of Ag-specific T cells was similar to that reported in human volunteers following boosting injections of DC loaded with the immunodominant influenza matrix peptide (33). This indicates that adenoviral delivery of Ag to DC provides a potent means to stimulate broad virus-specific T cell responses. Responses to specific peptides in our study were mediated by CD8⁺ T cells, as has been shown when adenoviral vectors are injected directly to primates (34). Given the importance of CD4⁺ T cell help in viral immunity (35), the generation of a mixed CD4⁺ and CD8⁺ T cell response may be ideal, and this could most likely be achieved by including sequences in the viral vectors to target Ag to the endosomal/lysosomal pathway, as has been done in the RNA transfection system (36).

The response to Gag induced by vaccination was at least as broad as that seen in monkeys experimentally infected with SIV. In the majority of animals, the response to vaccination was directed toward more than one peptide pool, and in the animals examined in detail, responses were directed toward two to three peptide epitopes within the Gag protein. A similar breadth of response to Gag is found in Mamu-A*01-expressing monkeys infected with SIV, with a dominant response directed toward the CM9 Gag_181–189 epitope (37) and minor responses directed against three other epitopes of Gag, identified based on class I-binding motifs (38). In Mamu-A*02-expressing monkeys vaccinated with a DNA/modified vaccinia virus Ankara regimen and subsequently infected with SIVmac239, cytotoxic T cell responses to Gag appear to be restricted to a single epitope in p17 (Gag_71–79 GY9) (39). Interestingly, in the two monkeys that expressed this class I allele in our study (M2001 and M1699), there was no evidence of a p17-specific response (peptides 1–31), and instead robust responses to at least two regions within p27 were induced based on pooled peptide assays. This may suggest that vaccination with DC expressing adenovirally encoded genes induces T cell responses to epitopes not generally recognized during the course of infection, which may translate into a broadening of T cell immunity when DC are used therapeutically in infected animals. Indeed, the peptides identified in our study to be targets of Gag-specific CD8⁺ T cells appear to contain novel epitopes not previously identified. Of particular interest is the common region shared by peptides 35 and 36 (Gag_141–151), which includes an immunodominant epitope recognized by the two animals studied in detail. This sequence (GGNYVHLPLSP) is near the region containing the Mamu-A*01-restricted peptide LW9 (38) and has not previously been recognized as being immunogenic. The MHC restriction of this epitope cannot currently be determined, as monkey M1901, which had a prominent response to the sequence, does not express any of the known Mamu class I alleles. This highlights the value of adenoviral vectors in providing Ag to DC for stimulation of potent T cell responses without knowledge of animal or patient MHC molecules.

In previous cancer trials, administration of adenoviral vectors expressing tumor Ag directly to patients was significantly limited by the presence of high titer neutralizing Ab to the vector (27). In contrast, murine studies indicate that immunization with adenovirus-transduced DC is unaffected by pre-existing vector-specific neutralizing Ab (18), suggesting that delivery of tumor or viral Ag via DC transduction may circumvent the problem of vector-specific immunity. Our findings in the nonhuman primate suggest that weak vector-specific immune responses, in particular neutralizing Ab responses, are generated in response to repeated vaccination with adenovirus-transduced DC, and that these responses may limit the effectiveness of repeated boosting injections when given multiple times. The high multiplicity of infection used to transduce DC in culture most likely resulted in the passive transfer of a small percentage of adenoviral particles to the animal via adherence to DC, sufficient to induce neutralizing Ab responses. Lower multiplicities of infection may minimize this transfer and reduce the induction of neutralizing Ab to the vector. Alternatively, transduction of cells with a second unrelated recombinant adenovirus such as adenovirus serotype 35 would completely avoid adenovirus 5-specific immune responses, and such vectors are now becoming available (40). It should be noted that adenovirus-transduced DC localized to and persisted in the lymph node paracortex in an animal that had received multiple prior injections, suggesting that vector-specific responses are not so strong as to result in the rapid elimination of transduced DC.

The optimal route of administration of Ag-loaded DC to induce T cell responses in patients remains controversial. Intravenous administration of DC leads to accumulation of cells in lung, liver, and spleen, whereas delivery into skin leads to homing of DC preferentially to draining lymph nodes (21, 41, 42, 43). However, the vast majority of DC injected into skin remains at the injection site in monkeys and humans (21, 41, 43), leading to the hypothesis that intralymphatic or intranodal delivery of Ag-loaded DC may be optimal. In our study, Ag-expressing DC injected into the lymph node localized to the paracortex, similar to findings in human cancer patients using radiolabeled DC (43). However, we found no advantage of intranodal delivery over intradermal delivery of adenovirus-transduced DC, as animals in both groups responded to vaccination with essentially the same frequency of IFN-γ-secreting T cells directed toward the same regions of Gag. Similarly, in a study comparing i.v., intradermal, and intralymphatic routes of injection of DC in human prostate cancer patients, intradermal and intralymphatic delivery generated superior and almost indistinguishable responses, as measured by Ag-specific IFN-γ release and proliferation (44). Hence, there is evidence that intradermal and intranodal delivery of DC may be equally effective at inducing T cell responses to vaccination. However, the situation in AIDS and potentially other chronic infectious diseases may be different from cancer patients and normal monkeys. Studies by us and others in the monkey model indicate that migration of Langerhans cells from skin to lymph node is normal during acute SIV infection, but is markedly suppressed during AIDS, possibly related to an alteration in the lymph node chemokine environment (31, 45). Given that migration of exogenously delivered DC also depends on normal chemokine expression, it is possible that effective DC-based immunotherapy in patients with AIDS will require direct injection of Ag-loaded DC into lymph nodes. In contrast, initiation of DC-based therapy early after infection may potentially be effective when DC are injected into skin, as has recently been demonstrated in the rhesus macaque SIV model (12).

Our study provides a proof-of-principle for using adenoviral vectors to deliver complex Ag to DC to prime broad T cell responses in vivo. Vaccination with DC expressing Gag Ag exclusively is unlikely to induce long-term protection against virus challenge (46). Hence, future studies will be aimed at immunizing monkeys with DC transfected with adenoviral vectors expressing a range of SIV genes in the setting of vaccination or immunotherapy.

We thank D. McClemens-McBride, D. Meleason, H. Warnock, and S. Casino for assistance with animal procedures; D. Rehrauer for molecular typing of rhesus macaque MHC class I alleles; C. Rinaldo for access to the ELISPOT reader; N. Pedersen, K. Kent, and C. Arnold for contributing mAb acquired through the AIDS Research and Reference Reagent Program of the National Institutes of Health; and Schering-Plough Research Institute (Kenilworth, NJ) and Immunex for gifts of reagents.