|Home | About | Journals | Submit | Contact Us | Français|
Tumor-cell vaccination with irradiated autologous tumor cells is a promising approach to activate tumor-specific T cell responses without the need for tumor antigen identification. However, uptake of dying cells by DC is generally a non-inflammatory or tolerizing event in order to prevent the development of autoreactive immune responses. Here we describe the mechanisms that confer the potent T cell priming capacity of a recently identified a population of DC (merocyticDC, mcDC) that potently primes both CD8+ and CD4+ T cells to cell-associated antigens upon uptake of apoptotic cells.
mcDCs acquired cell-associated materials though a process of merocytosis that is defined by the uptake of small particles that are stored in non-acidic compartments for prolonged periods, sustained antigen presentation, and the induction of type I IFN. T cells primed by mcDC to cell-associated antigens exhibit increased primary expansion, enhanced effector function and increased memory formation. Using transgenic T cell transfer models and endogenous models, we show that treatment of tumor-bearing mice with mcDC that have been exposed to dying tumor cells results in tumor suppression and increased host survival through the activation of naïve tumor-specific CD8+ T cells as well as the revigoration of tumor-specific T cells that had been rendered non-responsive by the tumor in vivo.
The potent capacity of mcDCs to prime both CD4+ and CD8+ T cells to cell-associated antigens under immunosuppressive conditions makes this DC subset an attractive target for tumor therapies as well as interventional strategies for autoimmunity and transplantation.
The use of DCs loaded with tumor antigens is one of the most promising approaches to induce tumor-specific immune responses (reviewed in (1–4)). The underlying premise of this approach is that the efficiency of, and control over, the vaccination process provided by ex vivo manipulation of the DCs can generate and optimally potent APC and a superior method for stimulation anti-tumor immunity, compared with more conventional direct vaccination methods. Although there is a large body of literature involving both patient trials and animal models of tumor immunity in which DCs loaded with tumor-associated epitopes/proteins are able to induce immune responses to solid and blood-borne tumors, the efficacy of DC therapy in tumor-bearing hosts is still low.
DCs are phenotypically and functionally heterogeneous. Based on surface marker expression at least six DC populations have been described in mice and humans: plasmacytoid DCs (pDCs), three blood-derived populations (CD4+DCs, CD8α+DCs, and CD4−CD8−CD11b+ DCs(5, 6)), and two tissue-derived subsets (Langerhans’ cells and dermal/interstitial DCs)— all of which appear to be distinct sublineages and not precursor-product-related(7–9). However, recent research classifying DC populations based on their functional capacities, implies the existence of several additional DC populations that are found in common lymphoid tissues or associated with specific organs(10–13).
The potency of the DC used in therapeutic settings is of critical importance as the DC need to activate naïve T cells as well as reactivate or reinvigorate tumor-specific T cells that have been rendered anergic by the immune suppressive environment created by the tumor. Most DC-based tumor therapies require the identification and characterization of tumor antigens for each specific tumor, or –when tumor antigen-derived peptides are used– the appropriate amino acid sequence for each MHC haplotype of the patient(14). The use of apoptotic, autologous tumor cells for DC loading would eliminate the need for antigen identification and present a natural way to target antigens into the MHC pathways of the DC.
The majority of human and mouse DC populations can activate CD4+ and CD8+ T cells when pulsed with cognate peptides. However, although nearly all human and mouse DC populations can capture protein and cell-associated antigens, few populations are able to process and present the Ag-derived epitopes to either CD4+ T or CD8+ T cells(15, 16). In particular, cross-presentation of cell-associated antigens to CD8+ T cells has been thought to be limited to one or two DC populations (16, 17).
Besides the fact that only few DC subsets can present both CD4+ and CD8+ T cell epitopes from cell-associated antigens, both human and mouse studies have shown that in vitro and in vivo uptake of apoptotic cells generally induces a tolerogenic state in DCs(18). Phagocytosis of apoptotic material prevents DC maturation and strongly inhibits the production of pro-inflammatory cytokines(19, 20). In addition, the uptake of apoptotic cells has been shown to induce immunoregulatory factors that actively dampen adaptive immune responses, including IL-10, TGF- β, PGE2, and IDO(20–23). Although this is crucial for preventing the development of autoreactive immune responses after removal of dying cells during normal tissue homeostasis, tissue repair after injury, and removal of aging and senescent cells, it clearly is an impediment to DC tumor therapy.
Here we compare the therapeutic capacity of a naturally occurring DC population that lacks the conventional DC subset markers, CD8α, CD11b and PDCA-1 and is identified based on its functional capacities. This CD11c+CD11b-CD8α-PDCA-1- DC population represents ~5% of the splenic DC population and is not susceptible to tolerance induction by apoptotic tumor cells. In contrast, this DC subset produces pro-inflammatory type I IFNs upon encounter with irradiated or Fas-treated (tumor) cells in vitro(13), and shows efficient activation of both CD4+ and CD8+ T cells to cell-associated-Ag. As this CD11c+CD11b-CD8α-PDCA-1-DC subset only acquired small particles of the dying cells via a process that resembled “nibbling” (13)- in contrast to other DCs that engulfed dead and dying cells- we named this subset merocytic DC, or mcDC (μεροσ = particle).
C57Bl/6J, B6.PL-Thy1a/CyJ (B6/CD90.1), B6.C-H2bm1/ByJ(Kbm1), and B6.SJL.Ptpcra (B6/CD45.1) mice were purchased from The Jackson Laboratory. Ifnar−/−, B6.C-H2bm1/ByJ (B6.Kbm1), MHC class I Kb−/−, OT-1/CD45.1, OT-1/CD45.1/Kbm1, and OT-2/CD90.1, ActmOVAxKbm1 transgenic mice (24) were bred in our facility. Mice were maintained under specific pathogen-free conditions in accordance with guidelines by the Association for Assessment and Accreditation of Laboratory Animal Care International.
B16F10 and B16F10-OVA (resp. B16 and B16-OVA(25)) and B3Z (26) were cultured in IMDM; (Invitrogen, CA) supplemented with 10% FCS, 50 μM 2-mercaptoethanol, 2 mM L-glutamine, 20 U/mL penicillin, and 20 μg/mL streptomycin. OVA257–264 (SIINFEKL), tyrosine-related protein-2 TRP-2180–188 (SVDYDFFDWL), LCMV GP33–41 (KAVYNFATC), OVA323–339 (ISQAVHAAHAEINEAGR) and LCMV GP61–80 (GLKGPDIYKGVYQFKSVEFD) were obtained from A&A laboratories (San Diego, CA).
DC were isolated from spleens of naive mice or mice treated for 9 days with 10ug human recombinant (hr)Flt3L as described before (27)(Peprotech, Rocky Hill, NJ). Data in the manuscript are generated with DC from non-treated naïve mice, unless stated otherwise. DCs were subsorted based on their expression of CD11c, CD11b, CD8α, or PDCA-1 by flowcytometry to purity of >95% and viability >95% (7-AAD staining).
OT-1 and OT-2 T cells were isolated using CD8α or CD4 microbeads (Miltenyi Biotec, Auburn, CA) and labeled with CFSE (Molecular Probes, Eugene, OR) as described before(13). Purity of sorted cells was >98% and viability was >97% as determined by CD4/CD8α /Vα2/Vβ5 expression and 7-AAD staining.
DC were analyzed for the expression of CD4, CD8α, CD11b, CD11c, CD40, CD54, CD80, CD86, Kb, Db, and I-A/E by flowcytometric analysis (antibodies/isotype controls; eBioscience/Biolegend, San Diego, CA.
For phagocytosis studies, purified DCs were incubated with CFSE-labeled irradiated splenocytes. After 20 hr, apoptotic cells were removed and DCs were either analyzed immediately, or after an additional 24 hours of culture. DC were stained with antibodies to CD11c, CD11b, CD8α, and nuclear dye Draq5 and analyzed by Imagestream (Amnis, Seattle, WA) (28). At least 10,000 events were acquired for each condition. Images of fixed cells were analyzed using ImageStream Data Exploration and Analysis Software. The number and size of phagocytosed particles was determined using the spot counting and spot size features after tight masking on the brightfield image in order to exclude membrane-associated extra-cellular particles.
Measure of phagosomal pH over time was performed as described before (29) in splenic DC stained with CD11c, CD11b, CD8 and B220. For pH acceleration studies, 10 •M of DPI was added to the DC after 1 hour of chase and the pH measurement was performed 15 min later. For Imagestream analysis of lysosomal FITC degradation, purified DC were stained with CD11c, CD11b, CD8 and B220, incubated with polybeads, and “chased” as described in (29). To calculate the degree of FITC degradation, specific DC populations were selected based on surface markers after which a spot mask was applied to identify single cells containing only a single bead. For each beads the ratio of FITC:A647 was determined using the “intensity features” that calculates the sum of the pixel intensities in the spot mask with subtraction of the background for each fluorochrome.
For antigen retention studies, DC were incubated with irradiated actmOVA-Kbm1 T cells (1500 rad) in a 1:3 ratio, sorted by flowcytometry after 20 hr of co-culture and plated at 1×105 cells/well. At indicated time-points 1×105 OVA257–264-specific B3Z hybrodoma cells were added and their activation was determined 24 hr later by CPRG conversion assay. OVA257–264-pulsed DC were used as positive control for each population. In parallel studies DC were treated with 10 •M of DPI for 2 hr before B3Z hybrodoma cells were added.
To test the capacity for naive T cell priming, 1×105 purified DCs were cultured with irradiated actmOVA-Kbm1 T cells in a 1:3 ratio in 96 well U bottom plates. After 24 hr, 1×105 CFSE-labeled OT-1 or OT-2 T cells were added to the wells. In these set up only the DC are able to present the OVA257–264 peptide to the B3Z/OT-1 T cells, as the mutations in the Kbm1 binding groove do not allow adequate binding of the peptide on the actmOVA-Kbm1 cells(30). In addition, the lack of I-Ab on the T cells prevents direct activation of the OT-2 T cells. As positive control DC were pulsed with OVA peptide for 10 minutes and thoroughly washed. DC viability over time was determined by Annexin-V and 7-AAD staining.
OT-1 and OT-2 T cell proliferation and survival was determined after 70 hr by analysis of CFSE dilution together with staining for Vα2, CD4/CD8α, Annexin-V, and 7-AAD. Expansion of OT-1/OT-2 T cells was determined by dividing the number of live T cells at the end of the culture by the number of cells added at the start of culture. Cytokine production in the supernatant was determined by standard sandwich ELISA for IL-2, IL-4, TNF-α, and IFN-γ (Biolegend, San Diego, CA) and reporter assay for type I IFN (13). Intracellular cytokine production was determined in T cells after a 5 hr stimulation with cognate peptide in the presence of Brefeldin A. Surface staining and intracellular cytokine staining for IFN-γ, IL-2, and TNF-α were performed using a Cytofix/Cytoperm Kit (BDPharmingen, San Diego, CA) according to the manufacturer’s directions. In vitro cytolytic activity was evaluated by a JAM test as previously described(32)
B6 mice i.v. received 106 purified DC that were incubated with irradiated actmOVA-Kbm1 T cells as described above. Apoptotic cells were removed from the DC populations using the apoptotic cell removal kit (Miltenyi Biotec). At indicated time-points CD4+ T cell and CD8+ T cell responses were determined by Elispot and intracellular cytokine. CD8+ T cell memory was determined upon challenge with 103 OVA-expressing Listeria moncytogenes (13).
For the transgenic transfer model, Kbm1CD45.2 recipients received 1×106 CFSE-labeled purified OT-1-Kbm1/CD45.1 cells together with 1×106 CFSE-labeled non-transgenic purified CD8+-Kbm1/CD45.2 cells that served as an internal control. The next day mice received i.v. 106 DC (exposed to irradiated actmOVA-Kbm1 cells). Three days later, OT-1-Kbm1/CD45.1 proliferation and expansion was determined based on CFSE dilution and the ratio of OT-1-Kbm1/CD45.1/CFSE control cells.
In vivo cytolytic activity was determined as described before(32), using CFSE labeled B6.CD45.1. and TRP-2180–188 and OVA257–264 peptide.
B6/CD45.2 recipients received 1×106 tumor cells s.c. in the left flank. As soon as palpable tumor had formed, mice received 106 purified congenically identifiable OT-1 T cells with 106 non-transgenic purified CD8α/ 1 T cells (internal control). At indicated time-points mice were received i.v. 106 DCs (exposed to apoptotic actmOVA-Kbm1 cells). At different time-points tumor draining LN were isolated and the ratio of OT-1-CD45.1/CD90.1 control T cells was determined. Activation status of the OT-1 T cells was determined by CD44 staining. Cytokine production by OT-1 T cells was determined by intracellular cytokine staining as described above. Tumor growth was measured every 2nd day with vernier calipers. Tumor size was calculated as the product of bisecting tumor diameters.
Mice received 106 B16 cells s.c. in the left flank. As soon as palpable tumors had formed, mice received i.v. DCs (106 cells/mouse) that had been exposed to apoptotic B16 cells. Tumor growth was monitored as described above.
Tumor cells and tumor-infiltrating lymphocytes (TILs) were isolated and analyzed as described before(33). Quantitative PCR was performed on resected tumors using SybrGreen and the following primers: mL32 fwd 5′-GAAACTGGCGGAAACCCA-3′, mL32 rev 5′-GGATCTGGCCCTTGAACCTT-3, 5′-CTGAATGGCCCAGGTCTGA-3′, β-actin rev 5′-CCCTGGCTGCCTCAACAC-3′, CD8α fwd 5′-CCGTTGACCCGCTTTCTGT-′3, CD8α rev 5′-CGGCGTCCATTTTCTTTGGAA-′3, TRP-1 fwd 5′-GCTGCAGGAGCCTTCTTTCTC-′3, TRP-1 rev 5′-AAGACGCTGCACTGCTGGTCT-′3. The ratio of CD8 α mRNA:TRP-1 mRNA was determined after normalization to β-actin and L32 for each sample(33).
Unless stated otherwise, the data are expressed as means (standard error of the mean (SEM). Survival responses were analyzed by Kaplan-Meyer using a log-rank test. All other data was evaluated using an analysis of variance followed by a Dunnett test. A probability value of P<0.05 was considered statistically significant.
Although mcDC resemble CD8α+DC and CD11b+DC with regard to MHC class I/II and costimulatory molecule expression (fig. 1a), they exhibit distinct features after interacting with dying cells. To further characterize the mcDCs, different DC populations were purified from spleens of naïve mice, exposed to irradiated CFSE-labeled cells and analyzed for CFSE uptake by ImageStream, which allows for analysis of high numbers of individual cells with the capacity to discriminate between true and false phagocytosis (intracellular presence versus binding to the surface of cells)(28).
The mcDC population contained a significant higher frequency of cells with intracellular CFSE-labeled particles compared to the CD11b+DC population, and a slightly increased frequency compared to CD8α+DCs (fig. 1b,c). In addition, mcDCs contained more CFSE-particles per phagocytosing cell than CD11b+ or CD8α+ DCs. Importantly, the CFSE-particles in mcDCs were considerably smaller than the particles in CD8α+ or CD11b+ DCs (fig. 1d). Additional studies indicated that the difference in detected CFSE-particles was not necessarily reflective of differences in phagocytic capacity, but could in part be contributed to by different intracellular handling and processing of antigen (35, 36). When pulse-chase experiments were performed, CD11b+ and CD8α+ DCs exhibited quenching of the pH sensitive dye CFSE over time, indicating transportation of the phagocytosed particles to low pH organelles (fig. 2a). In contrast, mcDCs showed only a small decrease in both the frequency of cells with phagocytosed material and the number of phagocytosed particles per mcDCs over time, suggesting that the phagocytosed material was not subjected to lysosomal degradation. Pulse-chase experiments using beads dually-labeled with a pH sensitive (FITC) and pH-stable (A647) dye demonstrated that 20 hr after bead uptake mcDC retained beads at higher pH than other DC populations (fig. 1b,c). Comparable results were seen when DC were analyzed by Imagestream; mcDC had a significantly increased ratio of FITC:A647 intensity feature (i.e. ratio of the sum of the pixel values within the spotmask, background substracted), indicating the persistence of the FITC signal over time (fig. 2d,e).
Importantly, the antigen persistence in mcDCs resulted in a prolonged capacity to prime Ag-specific T cells. mcDCs exposed to irradiated OVA-expressing cells on a Kbm1 background (actmOVA-Kbm1) were able to activate the OVA257–264-specific T cell hybridoma B3Z for at least 3 days after removal of the irradiated cells (fig. 2f). In contrast CD8α+ DCs, currently considered the dominant player in the priming of CD8+ T cell responses to cell-associated antigens, showed already a significant reduction in T cell priming capacity 48 hr after removal of irradiated cells. This was not a result of decreased DC viability, as peptide pulsed CD8α+ DCs and mcDCs induced comparable B3Z activation that decreased in time (fig. 2g and data not shown). Treatment of mcDC with diphenylene iodonium (PDI), a specific inhibitor of flavin-containing enzymes that has been shown to accelerate lysosomal acidification(29, 37), resulted in increased lysosomal acidification in mcDC that correlated with the loss of their prolonged antigen presenting capacity of cell-associated OVA to B3Z cells (fig. 2h). The DPI effect was not mediated by decreased mcDC viability or altered MHC turnover as DPI treatment did not affect B3Z activation by OVA257–264 pulsed mcDC (fig. 2h).
To further assess the mcDC capacity for T cell activation, purified DCs were cultured with irradiated T cells from actmOVA-expressing Kbm1 mice. Twenty-four hr later, CFSE- labeled OVA-specific OT-1 or OT-2 T cells were added to the culture. When irradiated T cells from control Kbm1 mice were used, none of the DCs induced activation or proliferation of the OT-1 and OT-2 T cells. After incubation with apoptotic actmOVA-Kbm1T cells, CD11b+ DCs induced activation of most of the OT-1 and OT-2 T cells, but only a small population of T cells proliferated (fig 3a,c and data not shown). CD8α+ DCs induced more activation and proliferation in OT-1 and OT-2 T cells than CD11b+ DCs did. However, only mcDCs were able to induce proliferation in all OT-1 and OT-2 T cells (fig 3a,c). In addition, both OT-1 and OT-2 cells activated by mcDCs underwent more rounds of division and showed increased accumulation over T cells activated by CD11b+DC or CD8α+DC. pDC failed to induce activation and proliferation in OT-1 T cells. When DCs were pre-pulsed with OVA257–264 or OVA323–339 peptide, all DCs were capable to activate and induce proliferation in naïve OT-1 and OT-2 T cells (fig. 3a, and data not shown). This observation indicated that all DCs were able to prime naïve T cells, but that their capacity to do so against cell-associated antigens from apoptotic cells was differently regulated in the distinct DCs.
Importantly, the majority of the T cells activated by mcDCs were able to produce cytokines upon restimulation with their cognate peptide, while only a small fraction of the cells activated by CD11b+ and CD8α+ DCs produced cytokines upon restimulation with peptide (fig. 3d, and supplemetal figure 1). Comparable results were found when cytolytic activity was assessed. OT-1 T cells primed by mcDCs showed greater capacity to kill OVA257–264 pulsed target cells at lower effector/target ratios than OT-1 T cells activated by CD11b+ and CD8α+ DCs (fig. 3b).
Comparable results were seen when priming of endogenous CD4+ and CD8+ T cells responses was assessed. Transfer of mcDCs that had been exposed to apoptotic actmOVA-Kbm1 T cells in vitro induced a higher frequency of OVA323–339-specific IFNγ-producing CD4+ T cells than transfer of CD8α+ or CD11b+ DCs (fig. 4a). Similarly, transfer of actmOVA-Kbm1-exposed mcDCs resulted in an increased frequency of OVA257–264-specific IFNγ-producing CD8+ T cells than did transfer of CD8α+ or CD11b+ DCs. OVA257–264-specific CD8+ T cells induced by mcDCs showed comparable kinetics in expansion and contraction as CD8+ T cells induced by CD8α+ DCs, but had significantly improved memory expansion when challenged with OVA-expressing Listeria monocytogenes in vivo (fig. 4b).
To determine if T cell priming after mcDC transfer was restricted to the mcDCs or resulted from antigen transfer to other DCs(15), we used an in vivo model in which antigen presentation was restricted to a specific DC population (38). In this model, CFSE-labeled OT-1 CD45.1 cells from Kbm1 mice were transferred together with CFSE-labeled naive CD45.2 Kbm1 control CD8+ T cells into Kbm1 CD45.2 recipients. One day later, purified DCs from WT (Kb+/+) mice (exposed to irradiated actmOVA-Kbm1 cells) were transferred into the Kbm1 mice (fig. 5a). In this model the transferred DCs are the only cells capable of presenting OVA257–264 to the OT-1 cells, as the Kbm1 recipient mice cannot bind the OVA257–264 peptide sufficiently to allow OT-1 activation via cross-presentation. Comparable to the in vitro data, transfer of Kb+/+mcDCs resulted in increased OT-1 T cell proliferation compared to CD8α+ DC transfer (fig 5b,c), suggesting that T cell priming resulted from a direct interaction of the T cells with the mcDCs without a need for antigen presentation by other DCs. Similarly, transfer of mcDCs that were either irradiated or lacked MHC class I Kb expression failed to induce endogenous OVA257–264-specific CD8+ T cell responses in WT recipients (fig. 5d), indicating that mcDCs directly interacted with endogenous CD8+ T cells.
All DC population can produce type I IFNs upon activation with TLR ligands, however, only mcDCs produce type I IFNs after interacting with irradiated cells (fig 6a and data not shown). We recently showed that priming of endogenous T cell responses to cell-associated antigens was dependent on type I IFNs (13). We therefore set out to determine the relative contribution of type I IFN production by mcDC in T cell priming. Transfer of type I IFN receptor deficient (ifnar−/−) mcDCs that had been exposed to irradiated actmOVA-Kbm1 cells into WT recipients significantly reduced the induction of endogenous OVA257–264-specific CD8+ T cells compared to transfer of WT mcDCs (fig. 6b). In addition, transfer of WT-mcDCs into ifnar−/− recipients resulted in a comparable decrease in OVA257–264-specific CD8+ T cell frequency. Transfer of ifnar−/−-mcDCs into ifnar−/− recipients resulted in an even more profound decrease in CD8+ T cell responses, illustrating both an autocrine and paracrine role for mcDC derived type I IFNs in the priming of T cell responses to cell-associated antigens (fig. 6b).
The increased capacity to prime T cells to cell associated antigens prompted us to study mcDC in the context of DC tumor therapy. As DC therapy is generally performed after the immune system has been exposed to tumor antigens and the immunosuppressive orchestrations of the tumor, DC therapy not only needs to be able to activate naïve tumor-specific T cells, but also to re-activate tumor-specific T cells rendered anergic or non-responsive by the tumor and/or its environment.
To study the efficacy of DC therapy in the re-activation of anergic/non-responsive tumor-specific T cells we used a model in which B6.PL mice (CD45.2+/CD90.1+) were inoculated with B16-OVA cells and received CD45.1+/CD90.2+ OT-1 T cells (together with CD45.2+/CD90.2+ CD8+ control T cells) as soon as palpable tumors had formed (fig. 7a). In this model, up to 75% of the OT-1 T cells, but not the control T cells, are actively deleted within 8 days in both the spleen and tumor-draining LN (TDLN)(fig. 7b). The remaining OT-1 T cells expressed high levels of the activation marker CD44, but failed to produce cytokines upon stimulation with peptides in vitro, indicating the induction of non-responsiveness (fig. 7c).
Treatment of tumor-bearing mice 9 d after OT-1 T cell transfer with CD11b+ DCs (exposed to apoptotic actmOVA-Kbm1 cells) did not significantly inhibit tumor growth or increase survival compared to untreated mice (fig. 7d); nor did it result in an increase in the OT-1 T cell populations in the spleen or TD-LN compared to untreated mice (fig. 7e). Treatment with CD8α+ DCs resulted in inhibition of tumor growth, delayed mortality, and tumor clearance in 17–20% of the animals. This protective effect correlated with an increased frequency of OT-1 T cells compared to untreated mice (3.4±1.1 fold increase). Importantly, mcDC treatment resulted in a 12.4±3.8 fold increase in OT-1 T cell numbers concomitant with a significant survival advantage— with 75% mice being tumor free for at least 80 days after inoculation (fig. 7d,e). In addition, the majority of the OT-1 T cells primed by mcDCs produced IFN-γ or IFN-γ/TNF-α after ex vivo stimulation, while only few OT-1 T cells primed by CD8α+ DCs produced these cytokines (fig. 7f)
As DC therapy was performed 9 days after OT-1 T cell transfer –when most of the OT-1 T cells had been depleted or rendered anergic/non-responsive- these data indicate that mcDCs can potently activate protective anti-tumor responses in an otherwise tolerogenic environment.
Recent research suggest that models using transfer of transgenic T cells might not be completely representative of endogenous T cell responses, as endogenous CD8+ T cells are generally present at lower frequencies in naïve mice and have a wide range of TCR affinity for the epitopes of tumor-specific and tumor-associated antigens(39, 40). In addition, the dominant tumor-associated rejection antigens consist of self-antigens (in contrast to the tumor-specific OVA on the B16-mOVA tumor) and the available endogenous T cell repertoire in the periphery therefore predominantly consists of low-affinity T cells that have escaped selection in the thymus. We therefore tested whether mcDC were able to induce a protective endogenous CD8+ T cell response to tumor-associated self-antigens.
When naïve mice were treated with CD11b+ DCs, CD8α+ DCs or mcDCs that had been exposed to irradiated B16 cells, cytolytic CD8+ T cell responses to the dominant TRP-2 epitope could be seen in an in vivo cytotoxicity assay 7 days later (fig. 8). mcDC treatment resulted in the most efficient cytolytic response, which correlated with the number of TRP-2 specific CD8+ T cells (fig. 5 and data not shown). Treatment with CD8α+ or CD11b+ DCs also resulted in significant TRP-2-specific cytolytic responses, but these were not as robust as seen as after treatment with mcDCs. Importantly, when this experiment was performed in B16 tumor-bearing mice, the TRP-2-specific cytolytic CD8+ T cell responses were significantly reduced. CD11b+ DC treatment did not result in any detectable killing of the TRP-2 targets, while CD8α+DC treatment resulted in a minimal amount of killing (14 ± 9%), Importantly, mcDC treatment still resulted in considerable killing of the target cells (51 ± 8%) (fig. 8). As all mice started with a comparable endogenous repertoire and tumor burden, these data indicate that mcDC were able to activate TRP-2 specific CD8+ T cells under tolerizing conditions, while CD11b+ DC and –to a lesser degree- CD8α+ DC failed to do so.
When B16 tumor bearing mice were treated with TRP-2 peptide pulsed DCs no tumor regression/inhibition was observed in any of the treatment groups (data not shown). However, a single treatment with mcDCs that had been exposed to irradiated B16 cells resulted in a significant survival advantage over treatment with CD11b+ or CD8α+ DCs (fig. 9a and data not shown). Although most mice exhibited only a temporary inhibition of tumor growth, some mice showed complete tumor clearance. Mice that showed complete tumor clearance generally showed depigmentation of the hairs at the tumor site, an indication of a melanin-specific response (fig. 9b).
Analysis of dissected tumors 16 d post-inoculation showed a strong correlation of survival advantage with tumor-infiltrating lymphocyte (TIL) numbers. Tumors from mcDC-treated mice contained a greater frequency and absolute number of CD8+ T cells among the TILs compared to CD8α+ DCs and control-treated mice (fig. 9c). To obtain a more quantitative assessment of the T cell populations infiltrating the tumor, quantitative RT-PCR was performed and the ratio of TCRβ mRNA to tumor antigen (TRP-1) mRNA was determined (fig. 9d). No significant difference in the TCRβ:TRP-1 mRNA ratio was observed between control and CD8α+ DC-treated mice. Corresponding with the flowcytometry data, mcDC treatment significantly increased the TCRβ:TRP-1 mRNA ratio, demonstrating increased infiltration of TCRβ expressing cells in the tumor.
Here we show that a specific DC population with potent capacity to prime both CD4+ and CD8+ T cells to cell-associated antigens can be exploited to induce protective tumor-specific T cell responses in tumor bearing hosts. mcDCs have a distinct phenotype in the internalization and processing of cell-associated antigens, characterized by antigen storage in compartments with reduced lysosomal degradation resulting in prolonged antigen presentation. mDC-driven activation of T cells to cell-associated antigens endows T cells with a greater capacity for primary expansion, enhanced effector function and increased memory development, even in an otherwise immunosuppressive environment. As a result, treatment of tumor-bearing mice with mcDCs was more effective in the induction of protective tumor-specific CD8+ T cell responses, inhibition of tumor growth and increased survival than was treatment with other types of DC.
DCs are a phenotypically and functionally heterogenous population of leukocytes and, although the intrinsic properties of each DC subset may dictate their functional specificity, their final maturation and functional capacities are also influenced by the tissue environment and the cell-types they interact with (reviewed in (41)). The mcDCs described in this study account for ~5% of the total splenic DC population and lack the conventional DC classification markers CD8 and CD11b. Recently it has been suggested that the CD11b-CD8- DCs that express cystatin C and are expanded by FLT3L are immediate pre-cursors of CD8α+ DCs, as transfer of CD11b-CD8- DCs results in rapid conversion to CD8α+ DC (10). The mcDCs described here share many characteristics with CD11b-CD8- DCs in that they express cysteine protease inhibitor cystatin C and are expanded by FLT3L (data not shown). However, our data indicates that all characteristic features of mcDCs, including type I IFN production after phagocytosis of cell-associated materials, and CD4+ and CD8+ T cell priming to cell-associated antigens, neither required nor affected CD8α expression on mcDCs. Moreover, transfer of mcDCs on a CD45.1 background into a CD45.2 host, showed that only a small population of mcDCs expressed CD8α around 3 d after transfer (supplemental figure 2). Given that CD8α+ DCs do not produce type I IFNs after uptake of cell-associated material and show relatively poor capacity to prime CD4+ T cells to cell-associated antigens, the mcDC population could be regarded as a functionally distinct population.
Several mechanisms may contribute to the effective CD8+ T cell priming capacity of the mcDC, including the distinct internalization, processing and presentation of antigen, the production of type I IFNs by the mcDC upon uptake of cell-associated material, and increased induction of CD4+ T help.
Pulse-chase studies showed that mcDCs were capable of activating T cells to cell-associated antigens over a longer time-span than CD8α+ DCs. Importantly, when DCs were pulsed with the high affinity CD8+ T cell OVA epitope SIINFEKL, both mcDCs and CD8α+ DCs showed comparable priming, which was significantly less sustained over time than the priming by mcDCs presenting cell-associated antigens. Similarly, transfer of peptide-pulsed mcDCs and CD8α+ DCs into naïve mice resulted in comparable CD8+ T cell responses that did not protect mice from subsequent tumor challenge (data not shown). Together with the observation that mcDCs “store” cell-associated antigen in a compartment with relatively little lysosomal degradation, these data suggest that mcDCs may use these compartments as a source to continuously supply MHC ligands for presentation. This is in line with the observations of Van Montfoort, et al., who showed the formation of long-term storage compartments in DCs after receptor-mediated endocytosis of IgG-OVA(35). These compartments were lysosome-like organelles, distinct from MHC class II compartment and the recently described early endosomal loading compartments, and functioned as antigen depots which enabled the DCs to maintain high MHC-peptide levels on their surface over time.
Recent studies indicated that cross-presenting CD8α+ DCs limit endophagocytic proteolysis through low expression and low recruitment of proteolytic enzymes to phagosomes, limited acidification by the V-ATPase and active alkalinization of the endosomal and phagosomal lumen by the NAPDH oxidase NOX2(29). In CD8α+ DCs, RAC2-mediated assembly of the NOX2 complex to phagocomes resulted in the production of reactive oxygen species (ROS), which causes alkalinization of the phagosome and prevents antigen degradation(37). mcDCs and CD8α+ DCs have comparable ROS production in the resting and activated states (data not shown). In addition, DNA arrays showed that mcDCs express mRNA for all genes of the NAPDH complex at similar or even higher levels than and CD8α+ DCs (data not shown), suggesting that mcDC scould possible exploit a comparable mechanism as CD8α+ DCs to delay antigen degradation and enhance their antigen presenting capacity.
MHC-antigen density and length of T cell stimulation significantly affect clonal burst size, acquisition of effector function, and development of memory(42–46). The sustained antigen presentation by mcDCs therefore could result in a longer, and therefore more robust activation on a per cell level. In addition, maintenance of sufficient MHC-peptide complexes on the mcDCs would provide the mcDC with a greater timeframe to encounter and activate T cells.
A second adjuvant mechanism in the enhanced priming capacity of mcDCs could be assigned to their production of type I IFNs after uptake of cell-associated materials. Our transfer data shows that type I IFNs act in both an autocrine and paracrine fashion, as transfer of ifnar−/− mcDCs into WT recipients, and WT mcDCs into ifnar−/− recipients, significantly reduced antigen-specific CD8+ T cell responses. Type I IFNs have been shown to mature DCs and enhance the processing and presentation of antigenic peptides by increasing proteasome components involved in the generation of peptides and components that target the peptides for interaction with MHC class I and class II molecules(47). In addition, type I IFNs have been shown to enhance MHC expression and soluble and membrane-associated costimulatory molecules that affect T cell priming (48, 49). Besides direct effects on DCs, type I IFNs can act directly on T cells and enhance their proliferation and accumulation by inhibition of apoptosis(50, 51). Additional paracrine effects on bystander cells may result in the induction of cytokines and chemokines that would provide a favorable milieu for T cell priming(49, 52).
The ligand that induces type I IFNs in mcDC after uptake of cell-associated materials is still unknown, but recent research indicates that apoptotic cell-derived nucleotide structures that escape lysosomal degradation can induce type I IFNs(53). Yoshida, et al., showed that mice deficient in DNase II, the lysosomal enzyme that digests the chromosomal DNA of apoptotic cells and expelled nuclei, die in feto due to anemia caused by type I IFN production(54). Mice deficient in both the DNase II and ifnar genes, however, were born relatively healthy. In light of the fact that mcDCs show prolonged antigen retention and delayed lysosomal degradation of cell-associated materials, it is not unlikely that nucleotide structures can escape from such retained material. Our previous work showed that this type I IFN induction was not dependent on TLR signaling (13) which suggests a possible role for cytosoli nucleotide sensors.
The capacity of the mcDC to prime strong CD4+ T cell responses to cell-associated antigens might also be instrumental in the induction of the potent anti-tumor effect. We and others have shown that CD4+ T cell help during priming of CD8+ T cells is required for optimal CD8+ T cell activation, primary expansion, acquisition of effector function, and the development of memory(32, 55, 56). Increasing CD4+ T cell help through transfer of (transgenic) CD4+ T cells or pre-immunization of mice has been shown to enhance the induction of CD8+ T cell responses(57, 58). In addition, ample studies indicate that CD4+ T cell help plays a supporting role in the maintenance, reactivation, and expansion of existing memory cells(59–61).
DCs provide an attractive mechanism for therapeutic manipulation of the immune system in the priming of protective anti-tumor responses. However, the complexity of the DC system underscores the necessity for its rational manipulation in order to achieve protective and therapeutic immunity. Although current therapies with ex-vivo generated DCs have been proven feasible, there are still many obstacles to overcome to improve clinical trial success and offset the cost and complexity of customized cell therapy. Here we show that DC therapy using DCs that, after uptake of irradiated tumor cells, exhibit prolonged antigen retention, sustained antigen presentation, production of pro-inflammatory type I IFNs, and the capacity to prime both CD4+ and CD8+ T cells, is efficient in the induction of protective anti-tumor responses in vivo in tumor bearing hosts. Targeting DCs with these characteristics in humans, or manipulation of other DCs that would induce similar behavior, could be instrumental in the design of effective therapeutic and preventive DC-based cancer vaccines.
This work is supported by NIH/NIAID grant AI079545 and NIH/NCI grant CA138617 to EMJ.
The authors wish to thank the core facilities at CCHMC for their support in animal and cell services. We wish to think Drs Christopher Karp, David Hildeman, Kasper Hoebe, and Jonathan Katz for critical review of the manuscript.