Influenza Virus and Poly(I:C) Inhibit MHC Class I-Restricted Presentation of Cell-Associated Antigens Derived from Infected Dead Cells Captured by Human Dendritic Cells¹

Dendritic cells (DCs)³ have the ability to present exogenous Ags via MHC class I in a process called cross-presentation (1, 2, 3). This mechanism is thought to be particularly important for the generation of CD8⁺ cytotoxic T lymphocyte (CTL) responses against certain viruses that do not infect DCs, or against tumors (4, 5). In the context of viral infection, DCs can capture dead infected cells, cross-present viral Ag derived from infected dead cells, and prime virus-specific CTLs. However, DCs that capture dying cells might also cross-present cell-associated Ags of dead cells leading to generation of autoreactive T cells.

Conceivably, DCs regulate activation of cell-associated Ag-specific T cells both through induction of T cell tolerance as well as regulating presentation of cell-associated antigenic peptides on MHC class I and II. The first concept stems from studies on the role of DCs in the maintenance of peripheral tolerance. There, it has been demonstrated that immature DCs can present Ag in a tolerogenic fashion (6, 7, 8, 9). A hypothesis has been put forward that DCs induce peripheral tolerance to cellular Ags associated with dead noninfected cells generated during the normal tissue turnover (9, 10, 11). Consequently, the T cells have been tolerized in the steady-state, long before the DCs are called upon to initiate immunity to viral Ags. Such preemptive strategy would therefore lower the risk of activating autoreactive T cells. Inherent to this hypothesis is, however, an assumption that all and each of cell-associated Ags have been presented by DCs before the first viral infection of the host. Furthermore, new peptides might be generated from cell-associated Ags in the context of infection but not necessarily during the normal tissue turnover (12). Presentation of these peptides by DCs activated in the course of infection could lead to expansion of autoreactive T cells. Further studies demonstrated that activated DCs found a way to deal with the issue of Ags derived from captured dead cells by developing a capacity to activate and expand CD4⁺ regulatory T cells (9, 10, 11, 13). Thus, activated DCs could support regulatory T cells that have been generated specifically to prevent autoreactive response (9, 10, 11, 13). Nevertheless, this mechanism does not provide an explanation for how Ags presented by MHC class I are dealt with.

One such model for how DCs might regulate presentation of cell-associated Ag is based on the concept of phagosome-autonomy whereby Ag presentation to MHC class II-restricted CD4⁺ T cells is enhanced when DCs captured dead cells which carry a TLR agonist, but not for concomitantly captured dead cells without a TLR agonist (14). This model thus allows DCs to discriminate between cell-associated Ags and microbial Ags based on the TLR signals that accompany the specific cargo they engulf for presentation by MHC class II (14). However, it remains to be determined how human DCs regulate presentation of captured Ags for MHC class I presentation. This is important because understanding of such mechanisms could offer therapeutic targets, for example a tool to enhance immunogenicity of DC vaccines. Indeed, mouse studies showed enhanced cross-priming of transgenic T cells against OVA upon immunization with poly(I:C)- and OVA-expressing cells (15).

In this study, we have taken advantage of our in vitro system in which loading human DCs with dead allogeneic melanoma cells permits generation of melanoma-specific CTLs (16, 17, 18). Using this strategy, we found that influenza virus or poly(I:C) within dead cells inhibits MHC class I-restricted presentation of cell-associated Ags by human DCs in vitro thereby not allowing generation of CD8⁺ T cell immunity to these Ags.

HLA-A*0201⁺ Me290 melanoma cells were a gift from Drs. J.-C. Cerottini and D. Rimoldi (both from Ludwig Institute for Cancer Research, Lausanne, Switzerland). T2 cells were purchased from American Type Culture Collection (Manassas, VA). Cell lines were maintained in RPMI 1640 (Invitrogen), 1% l-glutamine, 1% penicillin/streptomycin, and 10% heat-inactivated FCS. HLA-A*0201-restricted CD8⁺ CTL clones M26 (specific for MART-1 peptide 27–35; AAGIGILTV) and G154 clone (specific for gp100 peptide 154–162; KTWGQYWQV) were provided by Dr. Cassian Yee (Fred Hutchinson Cancer Center, Seattle, WA).

GM-CSF was from Bayer HealthCare Pharmaceuticals and IFN-α was from Schering. CD40L, IL-2, IL-7, and IL-4 were all purchased from R&D Systems. Betulinic acid (BA) was purchased from Sigma-Aldrich. Poly(I:C) was purchased from Invivogen. Abs used to assess DC maturation (anti-HLA-DR, HLA-ABC, CD40, CD80, and CD86) were purchased from BD Biosciences. Peptides were synthesized by Biosynthesis. Anti-type I IFN Abs and influenza A virus M1 (FluM1) protein conjugated anti-DCIR (DC immunoreceptor) humanized IgG4 mAb were generated in house.

Poly(I:C) bodies were prepared by electroporation of Me290 melanoma cells at 4 × 10⁶ cells/ml with 100 μg/ml poly(I:C) in Opti-MEM medium (Invitrogen) at 300 V for 500 μs. Cells were then plated and after 24-h culture at 37°C, cultured at 42°C for 4 h. Cells were then treated with 10 μg/ml BA for an additional 24 h at 37°C and nonadherent dead cells were harvested the next day and washed several times with complete RPMI 1640 medium plus 10% FCS. Mock bodies were prepared using Me290 cells that were mock electroporated (no poly(I:C)).

Lipofected bodies were prepared by plating Me290 melanoma cells at 3–4 × 10⁵ cells/well in a 24-well plate with 0.5 ml Opti-MEM. After 24 h at 37°C, cells were treated with a 100 μl mixture of 10 μg of poly(I:C) in Lipofectamine 2000 reagent (Invitrogen) prepared according to manufacturer’s instructions. After 24 h of lipofection at 37°C, cells were heat shocked and treated with BA as described above. Dead cells were then harvested and washed with complete RPMI 1640 medium plus 10% FCS. Mock bodies were prepared from mock lipofected cells.

Bodies were analyzed for poly(I:C) content after preparation by immunofluorescent microscopy using K1 mAb (English & Scientific Consulting) that recognizes large (over 40 bp) dsRNA.

Flu-infected bodies were prepared by infecting Me290 melanoma cells with influenza virus A/PR8/34 (Charles River Laboratories) at 10⁴ hemagglutinin U/1 million cells. Control cells were left uninfected. Cells were then heat shocked and treated with BA as described above and then dead cells harvested. Both Flu-infected and control bodies were further irradiated for 20 min (6600 cGy) to prevent virus replication in the DCs.

Monocytes were enriched by apheresis from HLA-A*0201⁺ healthy volunteers and cultured in CellGenix medium (CellGenix) with GM-CSF (100 ng/ml), and IL-4 (10 ng/ml). Cells were fed with fresh cytokines at day 2 and 4 postculture. For some experiments (e.g., peptide priming), 3-day-old DCs prepared with GM-CSF (100 ng/ml) and IFN-α (500 U/ml) were used. DCs were harvested and loaded with killed melanoma cells at a 2:1 ratio for 24 h at 37°C. After 24 h, cells were washed, harvested, and used to prime autologous naive CD8⁺ T cells. For some experiments, DCs were loaded with CFSE-labeled tumor bodies (1 μM CFSE for 10 min at room temperature; washed with complete RPMI 1640 plus 10% FCS) followed by sorting of CD11c⁺, CFSE⁺ DCs. DCs were then pulsed with 10 μg/ ml MART-1 peptide analog (ELAGIGILTV) and gp100 peptide (IMDQVPFSV), Influenza-M1 peptide (GILGFVFTL), or FluM1 protein conjugated to anti- DCIR IgG4 mAb, or soluble FluM1 protein.

Lymphocytes were enriched by apheresis from HLA-A*0201⁺ healthy volunteers. Naive CD8⁺ T cells were sorted as CD8⁺, CD45RA⁺, CCR7⁺ or CD27⁺, CD4/CD19/CD56⁻ cells (>95% purity) and cocultured with DCs at a 20:1 ratio in a 24-well plate with 2 ml of complete RPMI 1640 plus 10% human AB serum plus 200 ng/ml soluble CD40L and 5 ng/ml IL-7. At day 7 of coculture, cells were split and fresh CD40L and IL-7 added along with 10 U/ml IL-2. CD8⁺ T cells were harvested for ⁵¹Cr release assay at day 10 of coculture.

Targets were labeled with Na⁵¹CrO₄ for 1 h at 37°C. A 4-h standard killing assay was performed as previously described. The mean of triplicate wells for each sample was calculated, and the percentage of specific ⁵¹Cr release was determined according to the following formula: % specific ⁵¹Cr release = 100 × (experimental ⁵¹Cr release – spontaneous release)/(maximum ⁵¹Cr release – spontaneous release).

The iTAg MHC HLA-A*0201 tetramers, MART-1 (ELAGIGILTV), gp100 (IMDQVPFSV), and Influenza-M1 (GILGFVFTL), were purchased from Beckman Coulter. Primed T cells were stained with PE-conjugated tetramer and with anti-CD8/anti-CD3 mAb for 45 min at room temperature. Tetramer binding to MHC HLA-A*0201 tetramer HIV gag (SLYNTVATL) was used as negative control.

T cells were labeled with 1 μM CFSE for 10 min at room temperature and washed three times with complete RPMI 1640 plus 10% human AB serum. For CD4⁺ T cell proliferation, cell were cocultured with tumor bodies-loaded DCs for 7 days at 37°C and CFSE dilution analyzed by FACS analysis. Naive CFSE-labeled CD8⁺ T cells were cocultured with tumor bodies-loaded DCs in the presence of 5 ng/ml IL-7 and 200 ng/ml CD40L. CD8⁺ T cell proliferation was read out at day 4, 5, and 7 of coculture by FACS analysis.

One hundred thousand CTL clones were cocultured with 5000 tumor body-loaded DCs in a 96-well plate in 200 μl of complete RPMI 1640 medium with 10% human AB serum and 200 ng/ml CD40L. After 8–16 h of coculture, cells were centrifuged and supernatant harvested for assessment of IFN-γ and IP-10 levels by Luminex assay.

DC supernatant was collected 24 h after loading with tumor bodies. Supernatant from CD4⁺ T cell-DC coculture was collected at day 7. DC supernatant was analyzed for IL-1α, IL-1β, IL-3, IL-6, IL-7, IL-8, IL-10, IL-12p40, IL-12p70, IL-15, TNF-α, MCP-1, MIP-1α, IFNα, and IP-10 by Luminex. T cell coculture supernatant was analyzed for IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-10, IL-12p40, IL-12p70, IL-13, IL-15, TNF-α, IFNα, and IFN-γ by Luminex.

DCs were harvested, washed with PBS, and fixed overnight with 1% paraformaldehyde at 4°C. The next day, cells were permeabilized with 0.1% saponin/1% BSA/PBS. Cells were incubated with mouse anti-human MART-1, gp100, tyrosinase mAbs for 30 min. After washing with 0.1% saponin/PBS, Alexa₅₆₈-conjugated goat anti-mouse IgG Ab was added for additional 30 min in 0.1% saponin/1% BSA/PBS. After washing with 0.1% saponin/PBS, cells were stained with FITC-conjugated mouse anti-human HLA-DR for an additional 30 min in 0.1% saponin/1% BSA/PBS. Cells were then washed with PBS and mounted onto Superfrost slides with DAPI/Vectashield. Leica TCS-NT SP confocal microscopy was applied with detection channels of FITC (510–550 nm) and Alexa₅₆₈ (580–660 nm). HLA-DR⁺ DCs were then scored for intracellular melanoma Ag and data given as percentage of DCs that were positive for melanoma Ag.

Melanoma cells (Me290) were infected with influenza virus and then killed (Flu bodies) by exposure to betulinic acid as described previously (18). Flu bodies were further gamma-irradiated for 20 min (corresponding to radiation dose of 6600 cGy) to prevent virus replication within DCs. The lack of viral replication was confirmed by the progressive decrease of Flu nuclear protein staining in Flu bodies-loaded DCs over 3 days in culture (data not shown). Monocyte-derived DCs were loaded with either control noninfected bodies or with Flu bodies and cultured at a 1:20 ratio with autologous purified naive CD8⁺ T cells. DC:T cell cocultures were supplemented with CD40L, IL-7, and IL-2 and CTL generation was assessed at day 10 in a standard ⁵¹Cr release assay. As expected, control cultures generated CTLs able to kill Me290 melanoma cells (Fig. 1,A). However, loading DCs with Flu bodies resulted in the complete lack of Me290 cell killing (Fig. 1,A). Importantly, the DCs were able to elicit FluM1 response as demonstrated by the expansion of FluM1-specific CD8⁺ T cells (Fig. 1 B). Thus, capture of influenza virus-infected dead cells by DCs inhibits generation of CD8⁺ T cells specific to cell-associated Ags derived from the dead cells. As we will demonstrate hereunder, this striking observation is not an indicator of a complete blockade of DC capacity to generate T cell immunity.

To determine whether dsRNA was involved, melanoma cells were transfected with synthetic dsRNA poly(I:C) (poly(I:C) bodies) and loaded onto DCs which were then used in cultures with autologous naive CD8⁺ T cells. As expected, CD8⁺ T cells from cultures with DCs loaded with mock-transfected bodies (mock bodies) efficiently killed Me290 cells (Fig. 1,C). However, CD8⁺ T cells recovered from cultures with poly(I:C) bodies-loaded DCs were unable to kill Me290 cells (Fig. 1,C). This result was highly reproducible and observed with at least 5 different donors (Fig. 1,D). DCs loaded with 10-fold less tumor bodies (5 × 10⁴ bodies per 1 × 10⁶ DCs rather than 5 × 10⁵ bodies per 1 × 10⁶ DCs) produced the same results (Fig. 1,E). There, even when DCs were loaded with poly(I:C) bodies at levels comparable to mock bodies (Fig. 1 F), poly(I:C) bodies-loaded DCs failed to prime CTLs that kill Me290 cells. This inhibition is not restricted to Me290 cells and can also be found when other melanoma cells are used (data not shown).

In line with the absence of CTL function, autologous naive CD8⁺ T cells exposed to poly(I:C) bodies-loaded DCs did not dilute CFSE indicating the lack of CD8⁺ T cell proliferation (Fig. 2). This was contrary to cultures with mock bodies-loaded DCs, where CFSE dilution was detected already at day 4 and 7 of coculture >50% of CD8⁺ T cells diluted CFSE indicating their proliferation (Fig. 2). Thus, capture of poly(I:C) carrying-dead cells by DCs does not lead to induction of CD8⁺ T cell responses against cell-associated Ags derived from the dead cells, including alloantigens.

This lack of CTL generation was not due to insufficient uptake of poly(I:C) bodies (Fig. 3). In fact, DCs loaded with CFSE-labeled poly(I:C) bodies showed significantly more CFSE staining by flow cytometry (p < 0.001; Fig. 3, A and B). Fluorescence microscopy analysis confirmed the presence of CFSE-labeled bodies within the DCs (Fig. 3 C). Therefore, reduced CTL generation is not a result of the lack of cell-associated Ags derived from dead cells captured by the DCs.

DCs loaded with control and poly(I:C) bodies were then cultured with CFSE-labeled autologous CD4⁺ T cells. At day 7, cocultures of DCs loaded with poly(I:C) bodies displayed twice as many CD4⁺ T cells that have diluted CFSE than cocultures with DCs loaded with mock bodies (Fig. 4^,A). Enhanced autologous CD4⁺ T cell proliferation was associated with the enhanced IFN-γ secretion by CD4⁺ T cells (Fig. 4^,B), consistent with CD4⁺ T cell differentiation. Expansion of CD4⁺ T cells was consistent with DC maturation (supplementary Fig. 1, A and B)⁴ and their cytokine secretion pattern. Loading DCs with poly(I:C) bodies preferentially induces IFN-α secretion but no IL-12p70 secretion by DCs (Fig. 4,C). This pattern was due to delivery of poly(I:C) within dead cells because soluble poly(I:C) treatment induces IL-12p70 secretion but very little IFN-α secretion (Fig. 4,C). Flu bodies also fail to induce IL-12p70 secretion from DCs, but do induce IFN-α production from DCs (Fig. 4 D). Both cytokines have been implicated in Th1 skewing of helper CD4⁺ T cells (19, 20, 21, 22). Thus, DCs that captured poly(I:C)-carrying dead cells elicit proliferation and differentiation of CD4⁺ T cells, indicating that activation of autologous CD4⁺ T cells is not inhibited with poly(I:C) bodies-loaded DCs as is activation of autologous CD8⁺ T cells.

To determine whether CD4⁺ T cells could rescue CD8⁺ responses against cell-associated Ags of captured dead cells (23, 24), DCs loaded with control bodies, poly(I:C) bodies, or Flu bodies were cultured with autologous T cells containing both CD4⁺ and CD8⁺ T cells. At day 10 of culture, CD8⁺ T cells were purified and tested for their capacity to kill Me290 melanoma cells. Even in the presence of CD4⁺ T cells, poly(I:C) bodies-loaded DCs (Fig. 5,A) or Flu bodies-loaded DCs (Fig. 5 B) did not activate CD8⁺ T cells that kill Me290 cells.

Because autocrine IFN-α has been implicated in inhibited DC-mediated cross-presentation of Ags from apoptotic bodies (25), we next set to determine its role by adding IFN-α-neutralizing and IFN-αβ receptor blocking Abs at the time of DC loading and at the time of DC:T cell coculture. CD8⁺ T cells harvested from these cultures still failed to kill target Me290 cells (Fig. 5,C). We confirmed that type I IFN activity was blocked by decreased IP-10 production, which is produced in response to IFN stimulation (supplementary Fig. 2). Furthermore, Transwell experiments showed that CD8⁺ T cells expanded by mock bodies-loaded DCs effectively kill Me290 target cells and this killing was somewhat enhanced when mock bodies-loaded DCs were exposed to soluble factors produced by poly(I:C) bodies-loaded DCs (Fig. 5 D), possibly due to the soluble poly I:C that might have been released from poly(I:C) bodies-loaded DCs and/or inflammatory cytokines produced by these DCs. These results indicated that inhibited responses against cell-associated Ags of captured dead cells by poly(I:C) bodies-loaded DCs was not due to soluble factors acting on DCs in an autocrine manner.

To determine the specificity of the observed inhibition, we next analyzed whether DCs loaded with poly(I:C) bodies could generate responses from naive CD8⁺ T cells to Ags other then those associated with captured dying cells. First, loaded DCs were used to stimulate allogeneic naive CD8⁺ T cells in a MLR. As shown in Fig. 6,A, allogeneic CD8⁺ T cells cultured with poly(I:C) bodies-loaded DCs showed CFSE dilution comparable to that of control cultures, demonstrating that DCs can actually expand naive CD8⁺ T cells possibly due to preformed MHC class I/peptide complexes presented by the DCs. To further analyze the capacity of these DCs to present preprocessed Ags, in the second set of experiments the DCs loaded with the control or poly(I:C) bodies were additionally pulsed with 9–10 aa HLA-A*0201-restricted peptides. To ensure that MHC class I-restricted presentation was being assessed with DCs that actually captured poly(I:C) bodies, the DCs were loaded with CFSE-labeled tumor bodies and then sorted as CD11c⁺CFSE⁺ cells. CD11c⁺CFSE⁺ DCs were then pulsed with MART-1 and gp100 peptides and used in cultures with autologous naive CD8⁺ T cells. As shown in Fig. 6,B in a tetramer-binding assay, poly(I:C) bodies-loaded DCs that were coloaded with MART-1 peptide were able to expand MART-1 peptide-specific CD8⁺ T cells. The frequency of MART-1 specific CD8⁺ T cells in cultures with poly(I:C)-loaded DCs was comparable to that observed in cultures with control bodies-loaded DCs (Fig. 6,B). Furthermore, when the expansion of rare gp100-specific CD8⁺ T cells was analyzed, the poly(I:C) bodies-loaded DCs were far more efficient at expanding gp100-specific CTLs (Fig. 6 B), thereby suggesting the adjuvant effect of poly(I:C). Generated T cells were functional as demonstrated by their capacity to kill T2 cells pulsed with respective peptides (supplementary Fig. 3). Similar results were obtained with poly(I:C) bodies-loaded DCs pulsed with HIV gag151 peptide (data not shown).

Experiments with allogeneic CD8⁺ T cells and with autologous CD8⁺ T cells thus demonstrated that the overall capacity of DCs to generate CD8⁺ T cell responses was not altered. However, because in both cases the peptides were preprocessed or MHC class I/peptide complexes preformed, these results did not exclude the possibility that cell-associated poly(I:C) actually inhibited Ag processing for or intracellular loading onto MHC class I. Generation of CD4⁺ T cell responses shown above indicated that the global Ag processing capacity of DCs was not altered. Therefore, to determine whether poly(I:C) bodies-loaded DCs were at all able to process Ags for MHC class I presentation, we measured response to soluble proteins which require internalization and processing for presentation. To this end, sorted CD11c⁺ CFSE⁺ DCs were coloaded with FluM1 protein targeted to DCIR on DCs by way of FluM1 protein conjugated to an anti-DCIR mAb (26, 27) (E. Klechevsky, A. L. Flamar, and Y. Ca., submitted for publication). As shown in Fig. 6,C, DCs coloaded with poly(I:C) bodies and targeted FluM1 protein expanded Ag-specific CD8⁺ T cells comparable to DCs loaded with control bodies and targeted FluM1 protein. This suggested that the overall Ag processing and presentation capacity of poly(I:C) bodies-loaded DCs is not inhibited. However, DCIR could target the Ag to a unique compartment not affected by poly(I:C). Therefore, we have further analyzed the presentation of soluble FluM1 protein delivered in the nontargeted fashion. To this end, DCs were loaded with CFSE-labeled mock bodies or poly(I:C) bodies along with soluble FluM1 protein. After 24 h loading, CFSE⁺ DCs were sorted and used in coculture-purified autologous CD8⁺ T cells. No significant difference in the frequency of FluM1-tetramer binding CD8⁺ T cells in cultures with DCs loaded with mock bodies or with poly(I:C) bodies could be observed (Fig. 6,D). Importantly, adding soluble poly(I:C) to mock bodies-loaded DCs significantly enhanced the expansion of FluM1-specifc CD8⁺ T cells (Fig. 6 D). Thus, the inhibition of CD8⁺ T cell responses is specific for cell-associated Ags of captured dead cells that contain poly(I:C).

The results thus far implied that the inhibited generation of CD8⁺ T cell responses with poly(I:C) bodies-loaded DCs may be due to a lack of presentation of cell-associated Ags of captured dead cells. To test this, we made use of a melanoma Ag-specific CD8⁺ CTL clone. The HLA-A*0201-restricted G154 clone recognizes melanoma-associated gp100 peptide 154–162 (KTWGQYWQV) (28). IFN-γ or IP-10 secretion by the G154 clone after 8–16 h coculture with DCs was used as a read-out of activation. Secretion of IFN-γ by the gp100-specific CTL clone was observed when G154 cells were cocultured with mock bodies-loaded DCs (Fig. 7,A). However, no IFN-γ secretion was elicited with poly(I:C) body-loaded DCs (Fig. 7,A). Both poly(I:C) and control bodies-loaded DCs activated the gp100-specific CTL clone when the DCs were coloaded with the respective gp100 peptide, suggesting that inhibition of cross-presentation is not due to DC dysfunction (Fig. 7,B). The lack of melanoma Ag presentation by poly(I:C) bodies-loaded DCs did not appear to be due to altered kinetics of Ag digestion/processing in the DCs because when DCs were loaded with respective tumor bodies, and allowed an additional 48 h after loading to “digest” the tumor body Ag, poly(I:C) bodies-loaded DCs still failed to activate G154 cells (Fig. 7,C). Similar results were obtained with the HLA-A*0201-restricted M26 CD8⁺ CTL clone which recognizes MART-1 peptide (data not shown) (28). These results indicated the lack of presentation of cell-associated Ag when the dead cell is loaded with poly(I:C). DCs from three different donors loaded with Flu bodies also failed to present melanoma Ag to the G154 clone, whereas control bodies-loaded DCs presented endogenous cell-associated Ag to the T cell clone (Fig. 7,D). Adding soluble poly(I:C) to DCs loaded with uninfected bodies did not inhibit Ag presentation (Fig. 7 E) further attesting to a specific inhibition by cell bodies containing virus or poly(I:C).

Finally, the lack of melanoma Ag presentation was not due to the lack of melanoma Ag. Indeed, immunofluorescence analysis demonstrated the presence of melanoma proteins in loaded DCs (Fig. 7,F). As was the case with CFSE-labeled bodies (Fig. 3), a larger fraction of DCs showed staining demonstrating the capture of melanoma Ags upon loading with poly(I:C) bodies when compared with loading with mock bodies (Fig. 7, F and G).

In this study, we found that DCs that captured dead cells infected with influenza virus, or carrying poly(I:C), did not present cell-associated Ags derived from captured dead cells to autologous CD8⁺ T cells. That resulted in the remarkable inhibition of generating CD8⁺ T cell responses against these Ags. Interestingly, neither type 1 CD4⁺ T cells nor CD40 signal could overcome the specific inhibition of CD8⁺ T cell responses. Furthermore, it could not be rescued by cytokines or by innate effector cells such as NK cells (data not shown). Whereas the mechanism of this specific inhibition is yet to be determined, it is clear that the overall DC Ag-presenting capacity has not been altered. First, the capacity to expand CD8⁺ T cells and generate specific responses is maintained in poly(I:C) or Flu body-loaded DCs when MHC class I/peptide complexes are preformed, i.e., DCs can expand naive allogeneic CD8⁺ T cells. DCs that captured poly(I:C) bodies-loaded DCs can also expand autologous CD8⁺ T cells if they are further pulsed with 9–10 mer peptides that do not require processing. Finally, when poly(I:C) bodies-loaded DCs are further given a secondary Ag in the form of soluble protein or receptor-targeting complex, CD8⁺ T cells specific to secondary Ag are expanded, thus suggesting that the secondary Ag is cross-presented. Furthermore, tumor bodies-loaded DCs are capable of activating autologous CD4⁺ T cells. Thus, DCs that have captured infected or poly(I:C)-containing dead cells are capable of processing Ag for cross-presentation, but there is a selective inhibition in the cross-presentation of Ags derived from dead cells to CD8⁺ T cells.

Several observations suggest that this inhibition is DC intrinsic. First, intracellular analysis by confocal microscopy shows the presence of specific cell-associated tumor Ags within poly(I:C) bodies-loaded DCs. Indeed, loading of DCs with poly(I:C) bodies results in more total cellular Ag and specific tumor Ags than loading DCs with control bodies. Second, because activation of allogeneic CD8⁺ T cells or autologous CD8⁺ T cells specific for secondary Ags is maintained, the inhibition is not due to any direct effect on CD8⁺ T cells, such as through inhibitory cytokines or negative costimulatory molecules.

A possible explanation is that dead cells containing viral stimulus are differentially compartmentalized within DCs. It is important to note that TLR3, which recognizes double-stranded RNA, is localized within endocytic vesicles (29, 30). In our system, TLR3 signaling within an endocytic compartment may target the accompanying Ag to a compartment that precludes cross-presentation. Similar to the work of Blander et al. (14), we found that DCs loaded with dead cells containing a TLR stimulus display enhanced activation of CD4⁺ T cells. Blander et al. attributed this enhanced MHC class II-restricted presentation to enhanced “phagosomal maturation” that occurs in a phagosome-autonomous manner, i.e., Ag accompanied by a TLR stimulus targets to LAMP-1⁺ lysosomal compartments with greater efficiency than Ag not accompanied by a TLR agonist, as well as activation of MHC class II loading components (31, 32). Recent work by Burgdorf et al. (33, 34) has indicated that loading of Ag for CD4⁺ T cell activation and cross-presentation to CD8⁺ T cells occur in distinct intracellular compartments. The former occurs in LAMP-1⁺ lysosomal compartment, while the latter takes place in early endosomal compartments distinct from lysosomes (33, 34). Therefore, it is conceivable that in our studies the presence of a TLR stimulus within dead cells stimulates targeting of the dead cells antigenic material to LAMP-1⁺ lysosomes for potentiated MHC class II-restricted presentation, but not to a compartment amenable for presentation via MHC class I to CD8⁺ T cells. Many groups report that cross-presentation requires transport of Ag from endosomes to the cytosol, therefore the presence of poly(I:C) within dead cells may inhibit this transport (3). Further work is required to understand mechanistically how poly(I:C) or influenza virus within dead cells prevent cross-presentation of Ags associated to captured dead cells.

Our in vitro finding with poly I:C bodies in the human system appears in contrast with the in vivo data in mice. There, Schulz et al. (15) analyzed responses against OVA that is overexpressed in cells used for immunization and found enhanced cross-priming of transgenic T cells against OVA upon immunization with poly I:C- and OVA-expressing cells. In our model, the Ags naturally expressed by melanoma cells are analyzed. It is possible that in vivo there is leakage of Ag so that not all of it is cell associated and accompanied by a TLR stimulus. It is worth noting in our system that concomitant loading of DCs with tumor bodies and treatment with soluble poly I:C does not inhibit cross-presentation, i.e., inhibition requires the poly I:C be associated with the dead cell. This is also indicated by our experiments with poly I:C bodies-loaded DCs further loaded with a different soluble or targeted Ag. Similar reasoning applies to observations by Cui et al. (35, 36) where repeated vaccination of mice with the freeze-thaw lysates of tumor cells that were transduced with poly I:C led to enhanced tumor rejection. Thus, when the Ag is provided in vivo in excess (Schulz et al.) or in the form of soluble protein (Cui et al. and FluM1 protein herein) it can escape inhibition. Additionally, we might be faced with consequences of inherent discrepancies between the murine and human systems. Indeed, the human counterpart of murine CD8α⁺ DCs (15, 37, 38, 39), which appear responsible for enhanced CTL cross-priming in the work described by Schulz et al., remains to be identified.

A lack of cross-presentation of cell-associated Ag by human DCs that have captured infected dead cells has a direct implication on activation of both viral-specific and self-Ag specific CD8⁺ T cells. It is conceivable such inhibited cross-presentation would prevent activation of destructive self-Ag specific CD8⁺ T cells in the context of viral infection, however it raises the inquiry of how virus-specific CD8⁺ T cells are maintained. One possibility is that abundant amounts of viral-derived peptides are generated in infected cells that are more readily loaded onto recycling surface MHC class I as dead cells are captured. Changes in self-peptide processing have been described in infected cells due to autocrine type I IFN, which may prevent availability of self-peptides for MHC class I presentation (12). Another possibility is that DCs that capture infected dead cells pass on antigenic material for cross-presentation to secondary DCs. Carbone et al. (40) have shown that in the context of skin HSV infection, migratory DCs such as Langerhans cells do not cross-present HSV Ag but rather transfer the Ag to lymph node resident DCs. Recent work by Randolph et al. has indicated a similar Ag transfer with human DCs that capture influenza-infected dead cells (41). At this stage, our results cannot be interpreted as differential cross-presentation of self and foreign Ags by DCs that captured dead infected cells.

Lastly, the inhibition of cell-associated Ag cross-presentation has significant implication for the use of DCs in cancer immunotherapy. Because generation of tumor-specific CD8⁺ T cell responses is one of the goals of such immunotherapy, our work highlights issues that arise when using whole dead tumor cells as a source of Ag.

We thank Dr. Sangkon Oh and Sandra Zurawski for discussion and construction of fusion proteins and; Flow cytometry and Luminex Core at BIIR; and Dr. Cassian Yee at Fred Hutchinson Cancer Research Center for providing melanoma-specific CD8⁺ T cell clones. We thank Cindy Samuelsen, Carson Harrod, and Nicolas Taquet for continuous help. We thank Drs. Michael Ramsay and William Duncan for continuous support.

The authors have no financial conflict of interest.