Antigen loading of DCs with irradiated apoptotic tumor cells induces improved anti-tumor immunity compared to other approaches
- First Online:
- Cite this article as:
- Fry, T.J., Shand, J.L., Milliron, M. et al. Cancer Immunol Immunother (2009) 58: 1257. doi:10.1007/s00262-008-0638-7
- 107 Views
Dendritic cells (DCs) serve as central regulators of adaptive immunity by presenting antigens and providing necessary co-signals. Environmental information received by the DCs determines the co-signals delivered to the responding adaptive cells and, ultimately, the outcome of the interaction. DCs loaded with relevant antigens have been used as therapeutic cellular vaccines, but the optimal antigen loading method has not been determined. We compared different methods to load class I and class II epitopes from the male antigenic complex, HY, onto DCs for the potency of the immune response induced in vivo. Co-incubation of female DCs with HY peptides, RNA or cell lysate from HY expressing tumor induced immune responses equivalent to male DCs. In contrast, female DCs incubated with irradiated, apoptotic HY expressing tumor cells (or male B cells) generated a stronger immune response than male DCs or female DCs loaded using any of the other methods. DC loading with apoptotic tumor resulted in complete protection against high dose HY-expressing tumor challenge whereas 100% lethality was observed in groups receiving DCs that were loaded with peptides, RNA, or lysate. We conclude that signals provided to the DCs by apoptotic cells substantially augment the potency of DC vaccines.
KeywordsDendritic cell vaccinesCancer vaccinesAntitumor immunity
Dendritic cells (DCs) orchestrate adaptive immune responses by interpreting signals received by the surrounding environment at the time of antigen presentation and modulating costimulatory signals to shape the outcome of the T cell:DC encounter [1, 2]. Host-derived inflammatory signals, pathogen-derived products and immunosuppressive mediators together provide information to the DC that are capable of initiating or suppressing an immune response to a cognate antigen [3, 4]. DC vaccination can induce potent immune responses in vivo and clinical trials using DC-based immunotherapeutic targeting of tumors are underway . While recent studies of DC vaccines have shown some preliminary activity in the setting of minimal residual disease , little clinical activity has been observed in the setting of established bulky tumors . Thus, although DC vaccination remains a promising approach for the immunotherapy of malignancy, strategies to improve efficacy are needed. Integration of the growing knowledge of DC biology into future DC vaccine trials provides a strategy to achieve this goal.
A number of methods have been used to load tumor associated antigens onto DCs including transfection, dendritic cell:tumor cell fusion, and loading with peptides, tumor RNA, DNA, whole cell lysate and apoptotic cells. While each of these techniques has inherent advantages and has shown activity in diverse model systems, very few studies have compared techniques for antigen loading in the same model . Given the importance of environmental signals delivered to the DC for adaptive immune response induction, we hypothesized that antigen-loading methods might alter the potency of DC vaccination. To examine this, we utilized immune responses to the model antigen complex HY following DC vaccination using different strategies of antigen loading. Endpoints were quantitative immune responses as measured by tetramer and ELISPOT and protection against the growth of an HY-expressing tumor. Remarkably, antigen loading with apoptotic cells was more potent than all of the other approaches tested, resulting in the highest measured immune responses and the most effective protection from tumor challenge. Therefore, the method used to load antigens onto DCs for vaccination in clinical settings significantly impacts vaccine effectiveness.
Generation of bone marrow dendritic cells
Bone marrow was flushed from femurs and tibiae of C57BL/6 mice, filtered through a 70 µM nylon mesh (BD Falcon), and erythrocytes were lysed by ammonium chloride (BioWhittaker). Cells expressing the lineage specific antigens CD5, CD45R, CD11b, Gr-1 and Ter-119 were depleted by magnetic activated cell sorting (MACS, Lineage Depletion Kit, Miltenyi Biotec), followed by positive selection for sca-1 (Sca-1 Microbeads, Miltenyi Biotec). Sca-1+/lin− cells were cultured for 7 days in complete media (RPMI, 10% fetal calf serum (FCS, Harlan Technologies), penicillin/streptomycin, Hepes buffer, sodium pyruvate, non-essential amino acids, 2 mercaptoethanol (Life Technologies) supplemented with GM-CSF 1,000 U/mL and IL-4 1,000 U/mL (PeproTech, Inc.) and the hematopoietic progenitor cell expansion cytokines Flt-3 ligand (25 ng/mL), SCF (100 ng/mL) and IL-3 (20 ng/mL) based on the method of Jackson et al. . The media was changed and additional cytokines were added on days 3, 5 and 7 of the culture (GM-CSF and IL-4 only on days 3 and 7, all cytokines on day 7). Cultures were maintained at 37°C, 5% CO2 in tilted flasks.
DCs were matured with 20 µg/mL anti-CD40 mAb (R&D Systems, Clone I-C10) and 5 µg/mL CpG-ODN 1555 (GCTAGACGTTAGCGT, gift of Cindy Leifer, National Cancer Institute). For antigen loading using peptides, DCs were activated 20–22 h prior to peptide pulsing. For antigen loading using RNA, lysate and apoptotic bodies which required antigen processing, DCs were activated 20–22 h following antigen loading.
Dendritic cell characterization
DC surface phenotype on day 1, 5, 7 and 8 of culture was measured by flow cytometry using fluorescein isothiocyanate (FITC) conjugated MHC class II (I-Ab) monoclonal antibody, phycoerythrin (PE) conjugated CD11b, CD40, CD54, CD80, and CD86 monoclonal antibody and allophycocyanin (APC) conjugated CD11c, in addition to appropriate isotype (IgG2a) controls (BD Pharmingen). Cells were analyzed using a dual-laser FACS Caliber (BD) with Cellquest® software.
On day 7 of culture using the conditions described above, CD11c + DCs were positively selected with magnetic beads (Miltenyi Biotec), then incubated for 1 h with 10 µg of FITC conjugated albumin (Sigma) at either 37 or 0°C. The uptake reaction was stopped by washing three times with ice-cold FACS staining buffer (PBS plus 2 mM EDTA, 10% FCS and 0.1% sodium azide (Life Technologies). The percent of live, CD11c + DCs incorporating FITC albumin was determined by flow cytometry.
Intracellular IL-12 expression
Expression of intracellular IL12 was evaluated in activated versus nonactivated DCs on day 8 using flow cytometry. Following positive selection for CD11c, protein export was stopped by adding 1 µg/mL brefeldin A (GolgiPlug, BD Pharmingen) for 10 h. Cells were fixed and permeabilized (Cytofix/Cytoperm, PermWash buffer, BD Pharmingen) prior to intracellular staining. DCs stained with the PE-conjugated antibody preincubated with purified IL12 and appropriate PE isotype controls served as negative controls.
Activated female DCs were washed twice in complete media and placed in serum-free media [HL-1 (Cambrex) plus 1% penicillin–streptomycin with glutamine] at 2 × 107 cells/mL, then incubated with 1 µM HPLC-purified (90–95% pure, prepared from 1,000 µM stock solutions diluted in serum-free media) peptides (Bachem) for 2 h at 37°C: Uty (H2-Db restricted class I immunodominant, WMHHNMDLI), Smcy (H2-Db restricted class I subdominant, KCSRNRQYL), Dby (H2-Ab class II immunodominant, NAGFNRNRANSSRSS) or irrelevant peptides (for class I, E7, RAHYNIVTF; for class II, Trypanosoma cruzii (TC) surface protein, SHNFTLVASVIEEA). Following peptide pulsing, DCs were washed twice in complete media then resuspended at 1 × 106 cells/mL.
Whole cell preparations
MB49 a bladder epithelial carcinoma that naturally expresses the male antigenic complex, HY (generously provided by Dr. Edward Lattime) , was grown to confluence, and cells were harvested by trypsinization. Tumor RNA was isolated from 2 × 106 MB49 by phenol extraction followed by ethanol precipitation. The final concentration of RNA was determined by spectroscopy and 10 µg of RNA was added to 1 × 106 female dendritic cells in serum-free media for 1 h at 37°C. MB49 tumor lysates were prepared according to the methods previously described by Mule et al. . Briefly, MB49 cell suspensions of 2 × 106 cells/mL were frozen at −80°C for 20 min, thawed at 37°C for 10 min. After three freeze–thaw cycles, the lysate was aliquoted and incubated with 1 × 106 DCs (lysate:DC ratio = 3 tumor cells:DCs) overnight at 37°C in complete media. To generate apoptotic bodies, MB49 was harvested by trypsinization, then placed in single cell suspensions at 1 × 107 cells/mL, then exposed to 10,000 cGy gamma irradiation. Following irradiation, cells were washed once, counted and incubated in a 1:1 ratio with 1 × 106 DCs for 4 h at 37°C in complete media. Apoptosis was verified 30 min post-irradiation by Annexin V staining (BD Pharmingen).
Measurement of immune responses
Tetramer and ELISPOT assays were performed as previously described . Briefly, RBC-lysed splenocyte single cell suspensions were incubated for 1 h at room temperature with H2-Db tetramer loaded with peptides derived from Uty, Smcy or the irrelevant H2Db binding peptide (E7 or TC peptides). Cells were washed in FACS buffer and incubated with anti-CD8, anti-CD44 and anti-CD4 antibodies (BD-Pharmingen) then analyzed. After gating on CD8+/CD4− cells, CD44+/tetramer positive cells were quantified. The total number of specific Uty- and Smcy-reactive cells were calculated following subtraction of E7 binding cells. For the interferon γ ELISPOT assay, 1 × 106 splenocytes were co-incubated with peptide-pulsed stimulators in a 1:1 ratio in 96-well membrane plates (Millipore) that had been pre-coated with purified anti-IFNγ capture antibody (BD-Pharmingen) for 24 h in a 1:1 ratio with peptide-pulsed stimulators. All samples were run in triplicate. The plates were washed and then incubated for an additional 24 h and biotinylated anti-IFNγ. Plates were washed again and streptavidin-alkaline phosphatase was added for 2 h followed by washing and development with substrate solution. Following drying, plates underwent automated counting (CTL). Specific spots were calculated by subtracting spot numbers in wells stimulated with E7 (for Uty and Smcy) or TC (for Dby).
Tumor protection studies
Female C57BL/6 mice were immunized on day 0 with 1 × 105 DCs injected intradermally. On day 14, MB49 cells were injected subcutaneously at the right flank. Rate of tumor growth and tumor-free survival was monitored every 2–3 days for at least 40 days.
Statistical analysis was performed using Prism (GraphPad Software). Differences between two groups were evaluated using a two-tailed, unpaired Mann–Whitney test. Survival curves were analyzed using the Wilcoxon Rank Sum statistic. P values <0.05 were considered significant.
Prior to expansion, sca+/lin− cells do not express the DC markers CD11c, CD40, CD80 or CD86. By day 5 of culture, the expanding cells express CD11c, but retain an immature DC phenotype characterized by low expression of CD40 and CD86 and no expression of CD80 (Fig. 1b). These immature DCs were capable of antigen uptake as evidenced by endocytosis of FITC-conjugated albumin (Fig. 1c). Given that immature DCs can induce tolerance , we activated the DCs using anti-CD40 and a CpG oligonucleotide on day 7, resulting in CD80 expression and upregulation of CD40, CD86 and CD11c (Fig. 1b). Furthermore, the activated DCs showed dendrite formation (Fig. 1d), an inability to endocytose FITC conjugated albumin and higher levels of intracellular IL-12 (Fig. 1e), all features consistent with a mature or activated phenotype. The combination of anti-CD40 and CpG resulted in superior activation when compared to either stimulus alone (data not shown). Thus, this system generated large numbers of immature DCs capable of antigen uptake and processing as well as the reliable DC activation into a mature phenotype associated with characteristic functional changes.
HY-expressing tumor cell RNA and tumor lysate loading of DCs expands HY reactive T cells with a potency equivalent to peptide-pulsed DCs, but DCs loaded with irradiated apoptotic tumor cells elicit stronger immune responses. One of the disadvantages associated with the clinical translation of peptide-pulsed DC vaccines is that specific epitopes for the targeted antigens must be identified for each MHC allele, requiring extensive preclinical characterization of each epitope. For this reason, peptide-based immunization has primarily targeted common class I HLA alleles and often does not incorporate class II epitopes, which are difficult to identify. To overcome this, alternative strategies have been employed to load antigens derived from whole tumor cells which are then processed and presented according to the MHC molecules present on the DC. Such techniques offer the advantage of not requiring knowledge of specific antigens, but it remains unclear whether these strategies efficiently load tumor antigens and whether these antigens can successfully compete with the self antigen repertoire present in both the DC and on the tumor itself. To explore this in our model, we used cells from MB49, a bladder epithelial carcinoma that expresses HY antigens, to prepare RNA, lysate and irradiated apoptotic cells using standard techniques. These sources for HY antigen were then added to cultures containing immature bone marrow-derived DCs. Based upon the ability for these DCs to take up FITC albumin, we inferred that they would also be capable of lysate and/or RNA uptake. To optimize presentation, the DCs were activated with anti-CD40/CpG 20–22 h following coincubation with lysate or RNA. Antigen-loaded DCs were then injected intradermally using the same day 0 and day 14 schedule employed for the peptide-loaded and male DCs. As seen in Fig. 2, HY-expressing tumor lysate or tumor RNA-loaded DCs elicited quantitatively similar CD8+ and CD4+ T cell expansion to HY peptide-pulsed DCs and male DCs. Thus the magnitude of the vaccine response induced by peptide-pulsed DCs, DCs endogenously expressing HY, tumor lysate-pulsed DCs, and RNA-pulsed DCs are equivalent in this model system. Remarkably however, immunization with DCs loaded with apoptotic HY-expressing tumor cells resulted in a statistically significant increase in CD8+ dominant, CD8+ subdominant and CD4+ responses when compared to all other methods of antigen loading. Furthermore, this method was also superior to the responses induced by male DCs endogenously expressing the HY antigen complex. This was not unique to tumor cells as a similar magnitude of expansion was seen when irradiated male spleen cells were used to load HY antigens into DCs (data not shown).
A number of reports have compared DC loading strategies, but these have generally used only two loading methods and the majority did not quantitate immune responses to specific epitopes (reviewed in ). We report a comparison of quantitative (including class I dominant and subdominant and class II responses) and functional immunity to a model tumor antigen system following a diverse array of antigen loading methods that encompass many of the common DC loading methods in current clinical use. Although all methods induced CD4 and CD8 T cell expansion, DC loading with irradiated tumor cells induced higher quantitative responses to the three epitopes studied which also translated into improved antitumor activity. While a technical explanation based upon improved protein stability/recovery following irradiation as compared to the other methods could be invoked to explain the superiority of the irradiated cells compared to peptide, RNA or lysate loading, the fact that apoptotic cell loading also afforded enhanced protection over male DCs endogenously expressing HY provides evidence that protein integrity does not entirely explain the data presented here. Rather, the results suggest that apoptotic tumor cells provide additional signals to DCs that optimize antigen presentation beyond that conferred by the maturational stimuli already provided by anti-CD40 in this model system.
These results are remarkable in light of the substantial data in the literature demonstrating that apoptotic cell death is associated with tolerogenic immune signals [15, 16]. Numerous studies have demonstrated that naturally occurring apoptosis in normal development, is interpreted as immunologically “bland” resulting in tolerogenic DCs which appear to contribute to the maintenance self-tolerance [17–20]. In contrast, necrotic cell death, associated with tissue destruction in vivo activates immunity via DCs, presumably as a result of endogenous and/or exogenous cofactors associated with the tissue damage. This paradigm does not predict that apoptotic tumor cells delivered to the dendritic cell would enhance anti-tumor reactivity. However, conclusions drawn from non-inflammatory models may not be entirely relevant to a model system in which exogenous stimuli are simultaneously provided to the DCs and induce IL-12 production and DC maturation. In such settings, substantial data has demonstrated that the tolerogenic impact of apoptotic cells can be overridden by the immune activating stimuli provided by DCs. Consistent with these findings, several studies have demonstrated that apoptotic cells are immune activating when provided in the context of tumors [21, 22] or infection  in vivo or ex vivo if they are presented via activated mature DCs [24–26]. It is important to note that the method used to induce apoptosis is critical in determining whether the cell death is immunogenic or tolerogenic [27, 28]. Thus, the results presented here are consistent with previous data demonstrating that activated mature DCs induce immune activation when apoptotic cells or apoptotic cell fragments are released. However, this is the first data to demonstrate that apoptotic cells actually enhance immune activation beyond that provided by anti-CD40 alone. It remains unclear whether this results from more effective access of epitopes acquired via phagocytosis of apoptotic cells as compared to those acquired via other means or whether this reflects improved antigen presentation by cells that have ingested apoptotic bodies. The fact that gamma irradiation-induced apoptosis has been shown to be immunogenic further supports the notion that the irradiated cells used in this model provided additional activation signals beyond the anti-CD40 and CpG rather than simply improving antigen loading .
These findings have a number of clinical implications for DC vaccination-based strategies. First, while acquisition of apoptotic tumor cells from solid tumors may be technically challenging in the clinical setting, leukemic blasts can be readily harvested and exposed to gamma irradiation ex vivo to generate apoptotic bodies. Our results suggest that apoptotic tumor cells, generated via irradiation, should be tested in the context of DC vaccines targeting leukemia and solid tumors where possible, since apoptotic bodies may be more effective than antigen loading via tumor lysates, peptides or tumor RNA. Second, even if loading of DCs with apoptotic cells is not possible, or if a more restricted immune response is desired (such as might be the case following allogeneic stem cell transplantation for neoplastic disease where a broad repertoire of recipient antigens on tumor vaccines could induce graft vs. host disease), future work should seek to identification the mechanism by which apoptotic bodies enhance the efficiency of immune priming in this system. The ability to mimic such signals in the context of alternative loading strategies could potentially improve the efficacy of DC vaccines. Finally, these results emphasize that strategies seeking to generate potent antigen-specific adaptive immunity must optimize the signals delivered by the innate immune system in order to reap the greatest possible clinical benefit since the anti-tumor effects resulting from a DC vaccine differed substantially based solely upon the method used to antigen load the DC. Future work should focus upon optimizing signals received by DCs in the context of tumor vaccines in an effort to improve the effectiveness of DC-based cellular vaccination for cancer or infection.
This work was funded by the Intramural Research Program of the National Cancer Institute. TF performed experiments, analyzed data and partially wrote the manuscript; JS performed experiments, analyzed data and partially wrote the manuscript; MM and ST performed experiments and analyzed data; CM provided intellectual and financial support for the experiments shown, analyzed data and partially wrote the manuscript. The authors thank Dr. Martin Guimond for his careful review of this manuscript. This research was supported by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research. ST and JK were supported by the Howard Hughes Medical Institute Research Scholars Program.