CD40 Signaling Drives Potent Cellular Immune Responses in Heterologous Cancer Vaccinations Free

The immune system possesses the unique ability to detect and eliminate cancer cells (1, 2). To discriminate healthy from malignant cells, the immune system relies on the aberrant surface expression of peptides processed from either nonmutated or mutated antigens (3, 4). Nonmutated self-antigens represent a heterogeneous group of proteins that are expressed either in a tumor-specific or tissue-specific pattern or simply overexpressed compared to healthy tissues. In contrast, neoantigens are derived from nonsynonymous, tumor-specific mutations that create de novo epitopes for T cells that have not been subject to central tolerance (5, 6). Both tumor-associated antigens and neoantigens have been used to induce immune responses in humans (7–9) and immune responses to neoantigens have been shown to be predictors of clinical responses to checkpoint blockade (10–14), but the overall clinical success of vaccination studies that target cancer-specific antigens has so far been limited (15, 16). A major limitation of vaccines is the low immunogenicity of self-antigens when applied in the form of DNA, RNA, or peptides in combination with adjuvants (17). Bacterial and viral vectors expressing tumor antigens overcome this limitation by inducing a strongly inflammatory environment but these vaccination vectors require stable integration of the antigen of choice into the vector backbone, which limits their use for individually tailored vaccinations (18). Therefore, we sought to develop a costimulation-driven vaccination regimen that allows for the rapid induction of potent T-cell responses to any given MHC class I or II cancer epitope. For this purpose, we took advantage of studies, which demonstrated that CD8 T cells primed under conditions of low systemic inflammation adopt a memory T-cell properties that allows for potent boosting of the T-cell response less than one week after primary immunization (18–21). In these studies, weak but highly selective dendritic cell (DC) immunizations in the absence of pro-inflammatory adjuvants were induced, followed by a potent, inflammatory booster vaccination to achieve optimal CD8 T-cell expansion. To adapt this short-term vaccination protocol to the need of personalized cancer vaccines, we chose short 8-10mer peptide epitopes as well as longer polypeptide chains containing CD8 and CD4 T-cell epitopes and tested various homologous and heterologous vaccinations with peptide-pulsed DCs, peptide bound to poly-lactic co-glycolic acid (PLGA) microspheres or peptides in soluble form with co-injections of TLR agonists and costimulatory antibodies. Our results show that only a heterologous vaccination protocol with a primary DC immunization followed by a boost with soluble peptide, Poly I:C and an agonistic CD40 antibody resulted in massive CD8 T-cell expansion and therapeutic immune responses that eradicated even large transplanted tumors. The vaccination protocol was equally effective for CD8 and CD4 T-cell epitopes and for both neoantigens and nonmutated tumor antigens and could therefore serve as a universal cancer vaccination approach that is ideally suited for personalized cancer immunotherapy.

C57BL/6J mice (6–8 weeks) were purchased from Charles River Laboratories or bred at the Animal Care facility of the Medical School Hannover, Germany. After infection, mice were housed in the Animal Care facility of Medical School Hannover according to the required biosafety level. All animal experiments were performed according to German legal guidelines for animal care and experimentation (TierSchG) and were approved by institutional and governmental boards (LAVES).

Attenuated ovalbumin-expressing L. monocytogenes of an actA-deficient strain (AttLM-OVA; ref. 22) and the virulent L. monocytogenes strain 10403s (virLM) have been described before and were grown and quantified as described previously (19). For infection, mice were injected intravenously with 5 × 10⁶ cfu/mouse for attLM-OVA and 1 × 10⁴ cfu/mouse for virLM. LCMV strain Armstrong was propagated as described previously (23) and injected intravenously at a dose of 2 × 10⁵ pfu/mouse.

Splenic DCs were isolated from donor mice 10 to 14 days after subcutaneous injection of 5 × 10⁶ B16 cells expressing Flt3L as described previously (24). After DC harvest, cells were matured in vitro with LPS (0.5 μg/mL) and incubated in the presence of peptides (2 μg/mL) for approximately 6 hours. For vaccination, 1 × 10⁶ DC were injected intravenously into individual mice. Poly (lactic-co-glycolic acid; PLGA) microspheres (2-μm size) were purchased from Phosphorex and conjugated as described previously (18). Mice received 1 mg of PLGA microspheres conjugated to 100 μg of antigen. For CoAT immunizations, mice were injected intravenously with a combination of 100 μg soluble peptides (thinkpeptides), 200 μg Poly I:C (Invivogen) and 100 μg of agonistic anti-CD40 antibody (clone 1C10, hybridoma kindly provided by Frances Lund, Department of Microbiology, University of Alabama at Birmingham, AL, USA). The sequences of the peptides used are listed in Supplementary Table S1.

The following antibodies were used for FACS analysis: CD8 (53–6.7), CD11a (M17/4); CD27 (LG.7F9), CD62L (MEL-14), CD127 (A7R34), interferon gamma (XMG1.2), TNF alpha (MP6-XT22), PD-1 (J43), CD69 (H1.2F3), CD25 (PC61.5), KLRG-1 (2F1), CD90.2 (53-2.1, all eBioscience), CD90.1 (OX-7, Biolegend) and appropriate isotype controls.

Homozygous Thy1.1 and heterozygous Thy1.1/1.2 transgenic OT-I mice were bred at the animal facility of the Medical School Hannover. For adoptive transfer of naïve T cells, CD8 T cells from Thy1.1 OT-I transgenic mice were obtained from either peripheral blood or spleen and injected intravenously into naïve Thy1.2 C57BL/6J mice. To transfer memory OT-I T-cells, Thy1.1/1.2 OT-I cells were adoptively transferred into mice one day before infection with LM-OVA and sacrificed at memory time points. Percentage of OT-I T cells in total splenocytes was determined by FACS and a spleen cell mixture containing the desired number of OT-I memory T cells was injected into naïve Thy1.2 C57BL/6J mice.

Biotinylated MHC class I monomers (H-2K^b) specific for OVA_257–264 (SIINFEKL) were obtained from Glycotope Biotechnology, Heidelberg, Germany and conjugated with fluorochrome-labeled streptavidin (eBioscience) according to standard protocols.

The magnitude of the epitope-specific CD8 T-cell response was determined either by intracellular interferon gamma staining or MHC class I peptide tetramer staining as described (19). For tetramer staining, a small volume (∼50 μL) of blood was obtained via submandibular bleeding. Intracellular cytokine staining for interferon gamma and TNF alpha was performed in total splenpcytes, blood or other organs of sacrificed mice. To analyze organ-specific distribution of antigen-specific CD8 T cells, mice were sacrificed and subjected to cardiac perfusion with HBSS (PAA Laboratories) before harvesting the organs. Organs were forced through 40 μm Cell Strainers (Falcon) to obtain single-cell suspensions before intracellular interferon gamma staining.

Cytokine, chemokine, and growth factor concentrations in spleen cells were quantified by the Luminex-based multiplex technique according to the manufacturer's instructions (Bio-Rad). Standard curves and concentrations were calculated with Bio-Plex Manager 6.0, the detection sensitivity of all proteins was between 1 pg/mL and 40 μg/mL.

Serum transaminase measurements were performed according to the manufacturer's instructions using the MaxDiscovery Alanine Transaminase (ALT) Enzymatic Assay Kit (Bio Scientific).

For the establishment of subcutaneous colorectal cancers, 1 × 10⁷ MC-38 cells (kindly provided by Michael Neumaier, University of Mannheim, Germany) were injected subcutaneously into the flanks of C57BL/6J mice. Cell line origin was not authenticated. Cell line consistency was tested via STR analysis after 10-20 passages. In tumor experiments, tumor volume was calculated with the ellipsoid formula (0.5 × length × width²; ref. 25), volumes were measured twice a week and mice were sacrificed when the calculated tumor volume exceeded 2,000 mm³.

Statistical significance was assessed using the two-tailed t test with a confidence interval of >95%. Data are presented as mean (± SD or SEM as indicated). For comparison of survival curves, the Logrank test was applied. All experiments were repeated at least once to ensure reproducibility. Levels of significance are indicated by asterisks: *, P < 0.05; **, P < 0.01; ***, P < 0.005.

Potent cellular immune responses are dependent on the presence of antigen, TLR agonists and the engagement of costimulatory receptors. Using the ovalbumin-derived model antigen SIINFEKL in initial experiments and neoantigens or nonmutated tumor antigens in subsequent experiments, we sought to analyze the individual contribution of antigen, TLR agonists and costimulatory antibodies to CD8 T-cell immune responses. For this purpose, we tested various combinations of agonistic costimulatory antibodies (Co), SIINFEKL peptide antigen (A) and TLR agonists (T) in C57BL/6J mice. Optimal expansion of T-cell responses occurred after combined injection of peptide with the TLR3 agonist Poly I:C and agonistic CD40 antibodies (referred to as CoAT hereafter; Fig. 1A), with the agonistic antibody being the major determinant of a potent immune response. To achieve short-term amplification of the SIINFEKL-specific T-cell immune response, mice were primed with different combinations of soluble SIINFEKL peptide, Poly I:C and agonistic CD40 antibodies seven days before CoAT boosting. The combined prime-boost vaccination resulted in further amplification of the immune response in peripheral blood samples of mice 15 days after CoAT boosting (Fig. 1B). Administration of costimulatory antibodies for both priming and boosting led to potent T-cell expansion but subsequent death of several mice at the peak of the secondary immune response. We therefore chose a primary immunization with soluble peptide and Poly I:C (AT) followed by CoAT boosting for the establishment of a neoantigen-directed vaccination because this combination induced enhanced immune responses in the absence of detectable side effects in the immunized mice. As tumor antigen, we used the recently described Ndufs1-V491A epitope (Ndufs1^mut; ref. 26), a low affinity neoantigen derived from C57/L congenic Hepa1-6 (hepatoblastoma) and C57BL/6J CMT64 (small cell lung carcinoma) cells. Following AT/CoAT prime-boost vaccination, high frequencies of SIINFEKL-specific CD8 T cells were detected. In contrast, the cancer epitope Ndufs1^mut did not induce detectable CD8 T-cell immune responses, indicating fundamental differences between vaccines targeting the xenogen SIINFEKL and cancer vaccines targeting tumor-derived neoantigens (Fig. 1C). Since successful short-term T-cell boosting has been reported after primary immunizations under conditions of low systemic inflammation (18, 19), we next tested alternative primary immunizations in the form of either peptide-pulsed DCs or poly-lactic co-glycolic acid (PLGA) microspheres (18) conjugated to Ndufs1^mut peptide before CoAT boosting. Only the combination of DC priming with CoAT boosting induced potent Ndufs1^mut specific T-cell expansion (Fig. 1D). Similar to CD8 T-cell responses following acute infections, the immune response peaked approximately 1 week after boosting, then contracted and formed stable memory populations (Fig. 1E). As expected for secondary immune responses, the Ndufs1^mut-specific CD8 T cells exhibited a CD62L^low effector memory phenotype 40 days after boosting before converting to mixed effector (CD62L^low) and central memory (CD62L^high) T-cell populations by day 235 (Fig. 1F). In all organs studied, neoantigen-specific memory T cells represented a significant proportion of the organ-specific CD8 T-cell population, including the bronchoalveolar space (Supplementary Fig. S1), with highest frequencies in tertiary tissues but low frequency in peripheral lymph nodes (Fig. 1G).

A possible concern of costimulation-driven vaccinations in humans is overboarding immune activation because administration of costimulatory antibodies has been shown to result in antigen-independent activation of T cells (27). To address the clinical safety of DC/CoAT, we sought to analyze CD40 antibody-mediated activation of naïve and memory CD8 T cells in the absence of cognate antigen. For this purpose, we generated Thy1.1/1.2 memory OT-I T-cells in Thy1.2 C57BL/6J mice and subsequently co-transferred naïve Thy1.1 OT-I cells into the same mice one day before costimulation-assisted vaccination in the presence (CoAT) or absence (CoT) of antigen (Fig. 2A). Although CoAT vaccination induced strong upregulation of CD69 and CD25, CoT vaccination induced primarily CD69 upregulation as a consequence of Poly I:C-induced interferon alpha secretion. Compared to CoAT vaccination, changes in CD25 expression were less pronounced after CoT vaccination and preferentially present in previously activated T cells (Fig. 2B). To assess whether CD40 stimulation induces T-cell expansion, low numbers of Thy1.1/1.2 memory and Thy1.1 naïve OT-I cells were co-transferred into naïve mice before either CoT or CoAT immunization (Fig. 2C). Following these immunizations, the kinetics of the resulting primary and secondary effector CD8 T-cell populations (derived from the adoptively transferred naïve and the memory T-cell populations, respectively) were monitored in peripheral blood samples. Only CoAT immunization was followed by detectable primary and secondary CD8 T-cell expansion, while CoT-induced immune responses remained undetectable, indicating only transient activation of T cells with CD40 antibodies in the absence of antigen. To compare overall T-cell activation after DC/CoAT with immune activation after naturally occurring infections, total CD8 T-cell expansion and cytokine secretion were assessed after either DC/CoAT immunization, infection with virulent Listeria monocytogenes (virLM) or Lymphocytic choriomeningitis virus (LCMV) Armstrong infection. In vivo tracking of antigen-experienced CD11a^hiCD8^int T-cells (28) revealed similar numbers of activated CD8 T cells after DC/CoAT immunization and virLM infection, but significantly higher T-cell activation following LCMV infection (Fig. 2D). Similarly, splenic cytokine levels of interferon gamma, interleukin-6 and other chemokines (Supplementary Fig. S2) after DC/CoAT vaccination were comparable with those observed after virLM or LCMV infection (Fig. 2E). Because administration of agonistic CD40 antibody has been reported to induce liver damage (29), we measured serum transaminases 7 days after DC/CoAT vaccination and virLM or LCMV infection. Compared with the natural infections, hepatocyte damage after DC/CoAT vaccination was less pronouned and more comparable to naïve mice (Fig. 2F). These results demonstrate that DC/CoAT vaccinations do not induce antigen-independent proliferation of naïve or memory CD8 T-cell populations. Furthermore, DC/CoAT vaccinations do not induce systemic cytokine storms or liver damage.

Compared with nonmutated tumor-associated antigens, neoantigens possess several unique features that make them prime targets for vaccinations. First, mutations in anchoring positions can increase peptide affinity compared to their wildtype counterpart. Second, even mutations that do not increase peptide affinity may induce more potent T-cell responses if the neoantigen is recognized by cross-reactive naïve T cells, which have not been subject to central tolerance. In the case of Ndufs1, wildtype and mutant peptides possess a comparable, low affinity (predicted NETMHC3.4 IC50 315 nmol/L and 322 nmol/L, Supplementary Table S1). We therefore immunized mice with DC/CoAT using either Ndufs1^wt or Ndufs1^mut peptides. The magnitude of the CD8 T-cell response after vaccination with Ndufs1^mut was approximately 50-fold higher compared with Ndufs1^wt (Fig. 3A) indicating that successful DC/CoAT neoantigen vaccinations do not depend on an increase in mutant peptide affinity. These results, however, do not rule out the possibility that within the group of neoantigens high affinity peptides may be more suitable for DC/CoAT vaccinations than low affinity peptides. To compare neoantigens with different peptide affinities, we tested a panel of recently described neoantigens derived from hepatoblastoma, lung cancer (26), sarcoma (10) and colon cancer (Fig. 3B, IC₅₀ ranging from 2 nmol/L to 40,106 nmol/L, see Supplementary Table S1; ref. 30). DC/CoAT vaccination resulted in most potent T-cell expansion when high affinity peptides (IC₅₀ <50 nmol/L) were used, indicating a clear correlation between peptide affinity and the magnitude of the CD8 T-cell response (Fig. 3C). However, potent cellular responses were observed even after immunization with peptides of very low affinity (e.g., H2Q2, IC₅₀: 8,350 nmol/L). For all peptides, we noticed a strong correlation between the magnitude of the DC-induced primary immune response one day before the boost and the magnitude of the ensuing secondary effector CD8 T-cell response (Fig. 3D). These results suggest that for most potent CD8 T-cell expansion the DC-induced immunization must exceed a threshold and that DC-immunizations high-affinity peptides are better suited to reach this threshold. To further characterize the antigen requirements for DC/CoAT vaccinations we used polypeptides that require proteasomal processing for the induction of adaptive immune responses. In humans, the use of polypeptides (or whole proteins) for vaccinations represents a major advantage since this approach alleviates the need for MHC haplotype identification, thus providing a potential "off-the-shelf" approach. Mice were subjected to DC/CoAT vaccinations with either 9- or 23mer variants of the H-2D^b-restricted Adpgk^mut epitope (30). As shown in Fig. 3E, both the short and long version of the Adpgk^mut epitope induced potent systemic immune responses that were detected after intracellular cytokine staining for interferon gamma. Of interest, the difference in magnitude between the CD8 T-cell immune responses of the two groups did not differ significantly, nor did the use of the long or short peptide for the ex vivo cytokine stimulation affect the numbers of interferon gamma positive CD8 T cells. These results suggest that DC/CoAT vaccinations are not restricted to 8-9mer peptides but that antigens of variable lengths could be used for a broader, MHC haplotype-independent approach.

Given the fact that not all tumors harbor neoantigens suitable for immunotherapy, we next tested DC/CoAT in the context of the nonmutated melanoma antigen TRP-2 (31). DC/CoAT immunization with TRP-2 peptides induced a low, but detectable primary immune response and a massive secondary expansion that exceeded 40% of all peripheral blood CD8 T cells (Fig. 3F). Consistent with the role of TRP-2 in the regulation of melanin quality, DC/CoAT vaccination resulted in partial fur depigmentation while the fur color of mice receiving repeated DC vaccination remained unchanged (Fig. 3G). Thus, DC/CoAT vaccinations amplify not only neoantigen-directed CD8 T-cell responses but also adaptive immune responses targeting nonmutated tumor antigens, a target that is more frequently found in cancers than antigens derived from nonsynonymous mutations. Finally, we tested whether DC/CoAT vaccinations could also be used to induce MHC class II-restricted CD4 T-cell expansion. Because many cancer vaccines have successfully used bacteria- or virus-derived MHC class II epitopes to augment cancer-specific CD8 T-cell immune responses, we used a prototypic C57BL/6J-congenic MHC class II epitope from Mycobacterium tuberculosis (MPT59, position 239-254; ref. 32). Similar to the MHC class I epitopes, DC/CoAT induced potent expansion of MPT59-specific CD4 T-cells in the immunized mice (Fig. 3H), suggesting that MHC class II epitopes can be used either as a primary DC/CoAT target or in conjunction with MHC class I epitopes to supply T-cell help.

To evaluate the therapeutic efficacy of DC/CoAT vaccinations in vivo, we tested the prime-boost regimen in a recently described murine model of colon cancer in which both the immunogenicity of the discovered neoantigens and the presentation of these neoepitopes on the surface of the colon cancer cell line MC-38 have been verified (30). One week after subcutaneous inoculation of MC-38 cells, mice were subjected to DC/CoAT vaccinations targeting the high affinity MHC class I neoantigen Adpgk^mut (Fig. 4A). In all mice, DC/CoAT vaccination induced strong expansion of interferon-producing, Adpgk^mut-specific CD8 T-cells without apparent contraction between days 20 and 48 after tumor inoculation (Fig. 4B). Similar amplifications of Adpgk^mut-specific T-cell responses could be induced with other costimulatory agonistic antibodies, including agonistic CD137 and OX40 antibodies (Supplementary Fig. S3). Mice left untreated rapidly succumbed to the aggressively growing tumor and had to be sacrificed between days 19 and 28. In stark contrast, DC/CoAT treatment with agonistic CD40 antibodies led to rapid and complete regression of the subcutaneous tumors in all mice (Fig. 4C). Mice that received DC/CoAT vaccination survived long-term for more than 60 days (Fig. 4D) and did not show any sign of tumor recurrence up to 330 days after tumor inoculation. Despite the potent immune response, no side effects were noted in the vaccinated mice at any time point. In general, DC/CoAT vaccinations only induced mild and transient elevations of liver enyzmes, and unexpectedly liver damage after CoAT boost was less pronounced compared to CoA booster immunizations (Supplementary Fig. S4). Following the DC/CoAT vaccination in tumor-bearing mice, Adpgk^mut-specific T cells expressed both tumor necrosis factor alpha and interferon gamma after peptide stimulation ex vivo and were mostly negative for PD-1 (Fig. 4E). In some mice, the percentage of Adpgk^mut-specific CD8 T cells exceeded 60% of the total CD8 T-cell population, which is to our knowledge an unprecedented magnitude in vaccination studies targeting tumor antigens.

Current cancer vaccines primarily consist of repeated homologous injections of tumor antigens and adjuvants. These attempts have so far yielded only limited success in clinical studies, partly due to the difficulty of priming and expanding the tumor-specific T cells. The use of repeated homologous immunizations ignores an increasing body of data showing that the impact of antigen load, costimulation, inflammation, and many other environmental factors on T cells differs considerably between primary and secondary antigen stimulations (19, 33–35). These data convincingly show that high levels of systemic inflammation are able to amplify both primary and secondary immune responses but prevent primary effector T cells from acquiring an "early memory phenotype" that is required for efficient boosting. Vice versa, low levels of systemic inflammation in primary immunizations induce effector T cells with early memory T-cell characteristics but are unable to sufficiently expand the primed T cells in subsequent boosts. Heterologous vaccinations may offer a possible solution to this problem by adapting to the individual needs of both primary and subsequent immunizations with different antigen formulations. For the design of cancer vaccines, this feature could be of particular importance because cancer vaccines aim at expanding T cells of low functional avidity. In cancer vaccinations, these low-avidity T cells are easily outcompeted by bystander T cells that are primed by other auto- or alloantigens in the presence of adjuvants, costimulatory antibodies, or viral/bacterial vaccination vectors.

Our data demonstrate that the combination of a primary DC immunization followed by a booster vaccination driven by agonistic, costimulatory antibodies is ideally suited to achieve amplification of tumor-specific T cells. In our experiments, DC immunizations were best suited to achieve selective, low level expansion of tumor selective primary effector T cells, most likely because pulsing of DCs with peptides results in highly selective occupation of MHC molecules with the tumor-specific peptides. In all experiments, the quantity and quality of the primary immune responses was of utmost importance for a successful prime-boost immunization. Secondary T-cell expansion was most potent when primary immunizations became detectable in peripheral blood but failed to induce a strong systemic bystander T-cell response. These results imply that the cancer-specific peptide immunizations must be able to induce a primary T-cell response that exceeds a yet-to-define threshold after DC immunization, as shown for the high (and some low)-affinity neoantigens and the nonmutated cancer antigen TRP-2. Our results with peptides of different lengths suggest that not only 8–10 mer peptides are suitable for DC/CoAT vaccinations but also longer polypeptides or whole proteins, which require proteasomal processing as long as the induced primary immune responses meet the aforementioned criteria. These results imply that DC/CoAT vaccination can be used with antigens of variable length and without a priori knowledge of the MHC haplotype, which represents a major advantage for future clinical applications. The use of DCs for immunization, however, could represent a possible limitation for use in humans since DC immunizations require labor-intensive isolation and expansion of DCs from individual cancer patients. Nevertheless, personalized DC immunizations are nowadays offered by a number of companies and have successfully been used in a large number of immunotherapy trials.

The magnitude of the DC/CoAT-mediated tumor-specific immune responses is to our knowledge unprecedented, with more than 60% of the total CD8 T-cell population being specific for a single, tumor-derived antigen. Because of the potent T-cell amplification, DC/CoAT induced immune responses could result in increased efficacies of cancer vaccines in patients. Until now, even DC vaccines approved for therapeutic vaccinations in humans like Sipuleucel-T often remain below the threshold of detection in vivo. Conversely, numerous reports have linked the clinical effect of vaccines in humans to the magnitude of the induced immune response (15). For all the studies that use DC immunizations for cancer treatment, the results of our heterologous vaccinations suggest that CoAT boosting could represent a simple and cost-efficient means to amplify the primary, DC-induced T-cell responses, in contrast to the currently used repeated homologous DC-vaccinations. In our opinion, a potent immune response is one of the major determinants of a therapeutic vaccine but clearly not self-sufficient since even potent immune responses fail to evoke clinical effects if the target antigen is not expressed in sufficient quantities on the target cell. It is therefore important to note that our in vivo experiments took advantage of a well-characterized tumor model with an immunogenic neoantigen-derived peptide whose expression on the target cells has been verified by mass spectrometry (30).

Given the potent immunostimulatory properties of agonistic costimulatory antibodies, toxicity is a potential concern of CD40 antibody immunizations. In fact, several side effects have been reported after repetitive injections of agonistic CD40 antibody (29, 36) or administration of agonists targeting the costimulatory molecule CD28, which induced cytokine storms and multiple organ failure in humans due to CD4 T-cell activation (27). In our experiments, CD40 stimulation resulted in only weak and transient antigen-independent activation of T cells and was most pronounced in previously activated T cells. Lethal outcomes were only reported when both agonistic CD40 antibody and antigen were used for primary and secondary immunizations, a setting that should therefore be avoided for future experiments. However, no systemic side effects, cytokine storms or lethal outcomes were noted when the CoAT regimen was used in heterologous DC/CoAT immunizations, most likely because DC immunizations induce highly selective, low-magnitude primary cellular immune responses with low total numbers of activated T cells.

Our novel heterologous DC/CoAT vaccination method is specifically designed to rapidly amplify tumor-specific T cells targeting MHC class I and class II epitopes. Although most of our experiments targeted MHC class I epitopes, DC/CoAT was almost equally effective with MHC class II antigens and in vaccination approaches that targeted multiple epitopes simultaneously (data not shown). For vaccinations targeting MHC class I epitopes, the use of agonistic CD40 antibodies could represent a major advantage as CD40 stimulation can substitute for the help provided by CD4 T cells to DCs. Because of this feature of the agonistic CD40 antibody DC/CoAT might be able to induce potent expansion of MHC class I-specific CD8 T cells even in the absence of an additional MHC class II vaccination epitope.

In conclusion of our data, we propose to re-think the methodology of cancer vaccinations to enhance their clinical efficacy. Instead of searching for more potent adjuvants a more physiologic approach should be adopted that respects the physiology of cancer-specific T cells by combining a selective primary immune response with a powerful, costimulation-driven secondary immune response. Eventually, the DC/CoAT vaccination presented here may only be one of many possible heterologous vaccinations using this novel strategy to boost the clinical efficacy of future immunotherapies.

No potential conflicts of interest were disclosed.

Conception and design: S. Nimanong, T.C. Wirth

Development of methodology: S. Nimanong, J. Wingerath, N. Woller, T.C. Wirth

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Nimanong, D. Ostroumov, C.S. Falk, T.C. Wirth

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Nimanong, D. Ostroumov, J. Wingerath, S. Knocke, C.S. Falk, T.C. Wirth

Writing, review, and/or revision of the manuscript: D. Ostroumov, J. Wingerath, S. Knocke, M.P. Manns, F. Kühnel, T.C. Wirth

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Nimanong, N. Woller, E. Gürlevik, F. Kühnel, T.C. Wirth

Study supervision: S. Nimanong, T.C. Wirth

This work was supported by grants from the SFB Transregio TRR77 (T.C. Wirth., C.S. Falk, M.P. Manns, and F. Kühnel), the Deutsche Forschungsgemeinschaft (WI 3308/3-1) the Deutsche Krebshilfe (111150 to T.C. Wirth) and the SANDER Foundation (T.C. Wirth and F. Kühnel).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.