Using Antigen-Specific B Cells to Combine Antibody and T Cell–Based Cancer Immunotherapy

Cancer immunotherapy has demonstrated clinical success in the treatment of hematologic and solid malignancies (1). The majority of developmental efforts focus on the induction of T cell–mediated antitumor immunity. Antibodies against immune checkpoint molecules such as PD-1 or CTLA-4 have demonstrated therapeutic efficacy (2). Furthermore, direct genetic engineering of T cells such as introduction of chimeric antigen receptors, which allows for MHC-unrestricted tumor-directed T-cell activation, has also shown clinical activity in defined indications (3). Here, we report an alternative approach, devoid of challenges resulting from genetic engineering and able to induce specific T-cell activity by vaccination.

CD40 activation of normal B cells improves antigen presentation (4, 5), leading to induction of a specific naïve or memory CD4⁺ (6–8) and CD8⁺ (4, 5) T-cell response. In contrast to DC vaccines, CD40-activated B cells (CD40B cells) can be greatly expanded from small amounts of peripheral blood (4, 5, 9). These CD40B cells are home to tumor-draining lymph nodes (TDLN; ref. 10) and induce antitumor immunity in mice (11, 12). However, so far only B cells of unknown and polyclonal specificity have been used for CD40 activation and tumor targeting, disregarding the two main advantages of B cells: the high affinity of the specific B-cell receptor (BCR) for its antigen and the ability to produce large amounts of specific antibodies. Development of CD40 activators that can be produced in GMP grade (13, 14) has overcome an obstacle for the use of CD40B cells in a clinical trial. We have thus hypothesized that a CD40B cellular vaccine based on tumor antigens could represent a promising approach for cancer therapy. Here, we report that murine and human B cells with a defined specificity can be enriched and used for bifunctional cancer immunotherapy that combines B cell–mediated T-cell activation and antibody-mediated antitumor effects.

For buffy coat preparations donors gave their consent. All experiments were approved by our institutional ethical board. C57BL/6 mice were purchased from JANVIER Labs. Luc⁺C57BL/6 mice were kindly provided by Robert Zeiser (University Hospital Freiburg, Germany). OT-I and OT-II mice were purchased from The Jackson Laboratory. Mice were housed under specific pathogen-free conditions. All animal experiments were approved by our regional animal care committee.

Human peripheral blood mononuclear cells (PBMC) or murine splenocytes were isolated using Pancoll density-gradient centrifugation (Pan-Biotech). B cells were enriched using immunomagnetic selection with human or murine CD19 microbeads (Miltenyi Biotec) according to the manufacturer's protocol. Murine T cells were purified by using EasySep Mouse T-cell Enrichment Kit (Stem Cell Technologies) according to the manufacturer's protocol.

Antigens were biotinylated by using the EZ-Link NHS-Biotin Reagent (Thermo Scientific) according to the manufacturer's protocol. Ovalbumin (OVA) and KLH were purchased from Sigma Aldrich, Hepatitis B surface antigen (HBV)-protein, and TRP-2 protein were purchased from Abcam, recombinant human protein CEA was purchased from Sino Biological and recombinant human protein NY-ESO-1 was purchased from OriGene. Antigen-specific B cells were enriched by labeling with 0.2 μg/μL biotinylated antigen and subsequent selection using the EasySep Biotin Selection Kit (Stem Cell Technologies) according to the manufacturer's protocol.

Murine CD34⁺ progenitor cells were purified from bone marrow of mice by positive selection with EasySep Biotin Selection Kit (Stem Cell Technologies) by using an anti-murine CD34-biotinylated antibody (BioLegend). Cells were cultivated at a concentration of 0.25 × 10⁶ cells/mL in VLE-RPMI medium (Biochrom) supplemented with 5% FBS, 50 μmol/L ß-mercaptoethanol and 500 U/mL of murine GM-CSF (Immunotools) and 1 U/mL murine IL4 for 7 days. For maturation, medium was supplemented with 1 μg/mL anti-mouse CD40L antibody (HM40-3, Acris Antibodies).

Cell culture with feeder cells for activation and expansion of B cells was performed as described previously (15). CD40L-expressing tmuCD40L HeLa feeder cells were kindly provided by Clemens Wendtner (Klinikum Schwabing, Munich, Germany). Briefly, B cells were seeded at a density of 1 × 10⁶ cells/mL on lethally irradiated feeder cells. Cell passaging of feeder cells was done twice per week. B cells were cultures in DMEM medium (Life Technologies) supplemented with 580 μg/mL glutamine, 10% fetal bovine serum, 1% HEPES, 1% MEM and 15 μg/mL gentamicin. Recombinant murine IL4 (50 U/mL; Immunotools) was freshly added.

Cell culture with soluble CD40 ligand was performed as described previously (14). Human B cells were seeded at a density of 1 × 10⁶ cells/mL in IMDM medium (Life Technologies) supplemented with 10% heat-inactivated human AB serum (Thermo Fisher Scientific), 5 μg/mL human insulin Actrapid (Novo Nordisk Pharma GmbH), and 15 μg/mL gentamicin. Recombinant human IL4 (50 U/mL; Immunotools) was added freshly. On days 0 and 5, the medium was supplemented with cross-linked human CD40L (Miltenyi Biotec): Recombinant CD40L (0.5 μg/mL) was preincubated with cross-linking antibody (10 μg/mL) for 30 minutes at room temperature. On days 7 and 11, cells were reseeded as described above.

Cell phenotypes were evaluated using the following fluorophore-conjugated antibodies: human B cells with CD19, CD80, CD86, CD138, IgD, IgG1, IgM (BioLegend), CD20 (Life Technologies), HLA-DR and HLA-A2 (eBioscience); murine B cells with CD62L, CD80, CD86, CD138, CCR7, CXCR4, CXCR5, IgD, IgG1, IgM, H2K^b, streptavidin (BioLegend), B220 and CD19 (Life Technologies); murine T cells with CD3e, CD4, CD8, CD25 (BioLegend) and SIINFEKL-H2K^b Tetramer (Glycotope Biotechnology). Murine DCs with CD11b, CD11c, CD80, CD83, and CD86 (BioLegend). Foxp3 was stained with Foxp3 Fix/Perm Buffer set and anti-mouse Foxp3 antibody (BioLegend). Data were collected on a Gallios flow cytometer (Beckman Coulter) and analyzed using FlowJo (TreeStar) or Kaluza (Beckman Coulter) software.

For in vitro generation of antibody-secreting plasma cells, B cells from mice were resuspended at a concentration of 1 × 10⁶ cells/mL DMEM medium (Life Technologies) supplemented with different stimuli, including IL4 (1 U/mL, Immunotools), IL21 (50 ng/mL, Immunotools), antibody to mouse CD40 (1 μg/mL, Acris Antibodies), and OVA-biotin tetramers (0.2 μg/mL OVA-Biotin + 0.5 μg/μL streptavidin-PE, BioLegend). Cells were incubated for 72 hours.

Murine APCs were incubated with OVA-protein at a concentration of 75 nmol/L for 24 hours prior to incubation with T cells. Human APCs were incubated with CEA-protein at a concentration of 100 nmol/L for 1 hours at 37°C prior to incubation with T cells. APCs were irradiated once with 26 Gy. T cells were stained with 10 μmol/L CFSE. Murine APCs were mixed at a ratio of 1:1 with negatively isolated T cells from spleens of OT-I or OT-II mice. Human APCs were mixed at a ratio of 1:1 with autologous negatively isolated T cells from PBMCs. Cocultures were incubated for 5 days. Subsequently, T-cell proliferation was assessed by flow cytometry.

Negatively isolated T cells (0.2 × 10⁶) from PBMCs were cocultured together with 0.1 × 10⁶ protein-pulsed (0.1 μmol/L) CD40B cells on precoated anti-human IFNγ FITC fluorospot plates (Mabtech). AIM-V medium was supplemented with 0.1 μg/mL anti-CD28. An antibody to CD3 (Mabtech) was used as positive control in a dilution of 1:1000. Fluorospot analysis was performed on an AID reader 20 hours later.

For quantification of OVA-specific antibodies, assays were performed according to the sandwich-ELISA protocol from BioLegend. In brief, high-binding plates were incubated with OVA-protein at a concentration of 20 μg/mL. Anti-OVA IgG1, IgG2a (BioLegend), and IgE (Novus Biologicals) served as standards. Mouse detection antibodies were obtained from BioLegend and applied at a concentration of 100 ng/mL. Avidin-horseradish peroxidase (BioLegend) was used at a dilution of 1:1,000. The affinity of IgG1 OVA-specific antibodies was determined by preincubating supernatants of differentiation cultures with decreasing concentrations of soluble OVA for 1 hours. Competition ELISAs were performed as described above.

EG.7 lymphoma cells (0.3 × 10⁵) were labeled with CFSE and subsequently cocultured with 3 × 10⁵ splenocytes isolated from spleens of C57BL/6 mice by Pancoll. The percentage of dead tumor cells was quantified by staining for 7AAD⁺ CFSE⁺ cells in flow cytometry analyses. NK cell activation in these cultures was measured by determining the percentage of NK1.1⁺ CD3⁻ CD69⁺ cells.

C57BL/6 or Luc⁺ mice were immunized i.p. with 100 μL of 20 μmol/L OVA-Protein in PBS + Incomplete Freund's Adjuvant (IFA; Sigma Aldrich) for generation of OVA-specific B cells. For immunization with cell subsets, APCs were exogenously loaded with 10 μmol/L OVA protein for 1 hour at 37°C.

As target cells, lymphocytes were isolated from spleens of C57BL/6 mice by Pancoll and labeled with either 2 μmol/L CFSE or 20 μmol/L CFSE. The 20 μmol/L CFSE population was pulsed with 10 μmol/L OVA-peptide (SIINFEKL) for 1 hour at 37°C. A 1:1 mixture of low and high CFSE lymphocytes were then injected i.p. into immunized mice on day 21. Twenty-four hours later, spleens were removed and analyzed for killing of peptide-pulsed target cells. The ratio of unpulsed versus pulsed (RatioUP) target cells was determined by dividing the percentage of CFSE low cells by the percentage of CFSE high cells. Specific lysis was calculated by the following formula:

% Specific lysis = (1 − (RatioUP Negative Control/RatioUP Immunized) × 100.

Murine CD19⁺ B cells or CD40B cells from Luc⁺ mice were injected i.v. into syngeneic mice. Detection of Luc⁺ B cells was performed by injecting 7.5 mg D-Luciferin in 250 mL 1× PBS i.p. Imaging was performed in the Xenogen IVIS 200 (Perkin Elmer). Mice were constantly kept under narcosis with 1.5–4% Isofluran at 37°C. Bioluminescence pictures were analyzed with the Living Image Software (Perkin Elmer).

The tumor cell line E.G7 was kindly provided by Tomo Saric (University Hospital Cologne, Germany). The tumor cell line Panc02-OVA was cultivated as previously described (16). For tumor formation, cells were harvested and resuspended at a concentration of 4 × 10⁶ cells/mL or 20 × 10⁶/mL 1× PBS, respectively. Cells (100 μL) were injected s.c. into the right flank of mice. Tumor size was determined daily from day 7 after inoculation by measuring tumor diameter in two dimensions using a vernier caliper. The tumor volume was calculated using the following formula: Tumor volume = 0.5 × (length × width²) mm.

Five and 25 days after tumor challenge, mice received either a mixture of each 250 μg of monoclonal anti-CD4 clone YTS191.1 and clone YTA3.1 or 500 μg of monoclonal anti-CD8 clone YTA 169.4. Efficient depletion was controlled by taking 100 μL blood from the tail vein 5 days after antibody injection and determining the percentage of CD4⁺ or CD8⁺ T cells by flow cytometry.

Significant differences were calculated by a two-tailed t-test, one-way ANOVA, or two-way ANOVA where appropriate using GraphPad Prism Software. P values of statistical significance were marked with asterisks as indicated in figure legends. Mean values and standard deviations (SD) were calculated from at least three independent experiments.

To test our hypothesis that tumor antigen-specific B cells can be used for cancer immunotherapy, we chose a mouse model that uses OVA as tumor antigen. We first established a method for detecting and enriching antigen-specific B cells with antigen-biotin tetramers (Fig. 1A; ref. 17). After immunizing mice with OVA in IFA, the OVA-specific B-cell population made up ∼2% of splenic CD19⁺ B220⁺ B cells compared with ∼0.60% in nonimmunized mice (Fig. 1B). Enrichment by antistreptavidin microbeads resulted in 85% pure CD19⁺ B220⁺ B cells of which ∼65% were OVA specific.

These OVA-specific B cells (OVA-B cells) showed higher levels of activation markers CD86, MHCI and MHCII compared with non-OVA-specific B cells of the same mice (Fig. 1C). OVA-B cells upregulated the expression of surface IgG1 and downregulated expression of IgD and IgM (Fig. 1D), indicative of an activated and class-switched phenotype. These results were not restricted to OVA-B cells because we obtained similar purities and activation profiles for other antigens such as keyhole limpet hemocyanin (KLH)-specific and tyrosinase-related protein 2 (TRP-2)-specific B cells (Supplementary Fig. S1).

Next, we investigated whether antigen-specific B cells provide superior T-cell stimulation compared with nonspecific B cells. Theoretically, antigen-specificity would have several advantages because it enables B cells to take up antigen via the specific BCR whereas antigen uptake by polyclonal nonspecific B cells occurs primarily by pinocytosis (18). BCR affinity is directly proportional to the capacity of B cells to present antigen to CD4⁺ T cells (19). Furthermore, antigen uptake by the BCR leads to changes in the antigen processing machinery that facilitate the intracellular trafficking of antigen and MHC-class II molecules and the generation of peptide-MHCII complexes (20).

First, OVA-B cells were incubated on CD40L-expressing feeder cells to induce activation and to enhance their antigen-presenting capacity. After 24 hours, B cells were analyzed for the expression of activation markers. OVA-CD40B cells showed higher expression of activation markers such as CD80, CD86, MHCI and MHCII (Fig. 2A) compared with unstimulated controls.

To confirm the enhanced antigen-presenting function of OVA-CD40B cells, their ability to stimulate an antigen-specific T-cell response was investigated in vitro. Cocultures of OVA protein-pulsed purified OVA-CD40Bs with both antigen-specific CD4⁺ and CD8⁺ T cells from OT-II or OT-I mice, respectively, increased proliferation of T cells compared with polyclonal CD40B cells (Fig. 2B). Induction of T-cell proliferation by OVA-CD40B cells was similar to that of conventional professional APCs, as exemplified by anti-CD40 matured dendritic cells (DCs). Maturation of DCs was confirmed by flow cytometry (Supplementary Fig. S2). T-cell stimulation by OVA-CD40B cells or DCs resulted in a similar degree of T-cell activation that was higher than that achieved by polyclonal B cells (Supplementary Fig. S3).

To further characterize the capacity of OVA-CD40B cells to induce specific T-cell responses, we next assessed their antigen-presenting function in vivo using an in vivo cytotoxicity assay. Mice were immunized i.v. with 5 × 10⁶ protein-pulsed APCs, i.e., polyclonal CD40B cells, OVA-CD40B cells or anti-CD40 matured DCs. On day 21, CFSE⁺ OVA-peptide pulsed syngeneic lymphocytes were injected as target cells into immunized mice. The in vivo-specific lysis by antigen-specific CD8⁺ T cells was calculated by determining the ratio between unpulsed and pulsed target cells in spleens. The specific lysis was higher in mice immunized with OVA-CD40B cells compared with mice immunized with polyclonal CD40B cells (Fig. 2C). Immunization with OVA-CD40B cells and DCs resulted in a similar degree in specific lysis.

Spleens of vaccinated mice were stained for the presence of OVA-peptide specific CD3⁺ CD8⁺ T cells (Fig. 2D). Mice vaccinated with OVA-CD40B cells had higher percentages of OVA-specific CD8⁺ T cells than mice that were treated with polyclonal CD40B cells or negative controls. There was no difference in CD8⁺ T-cell induction compared with DC-based vaccinations. The effect was specific to CD8⁺ T cells and did not lead to compensatory induction of CD4⁺CD25⁺Foxp3⁺Tregs (Supplementary Fig. S4). The induction of antigen-specific T-cell immunity was dose-dependent, as increasing the injected number of APCs for vaccination also resulted in an enhanced specific lysis in all three different APCs (Fig. 2E). These results demonstrate that antigen-specific CD40B cells are potent inducers of antigen-specific T-cell responses.

To further understand the nature and dynamics of B- and T-cell encounters, the in vivo migration kinetics of luciferase (Luc⁺) CD40B cells were assessed. OVA-specific Luc⁺ CD40B cells appeared in the spleen within 12 hours after injection and then migrated to the abdominal lymph nodes within the next 5 days (Fig. 3A).

In mice with OVA-expressing E.G7-lymphoma, polyclonal CD40B cells homed to the spleen, but preferentially accumulated in TDLN relative to abdominal lymph nodes (Fig. 3B). OVA-CD40B cells migrated to the spleen and TDLN over a period of 36 hours (Fig. 3C). In contrast to polyclonal CD40B cells, 1/10 of injected OVA-CD40B cells appeared at the tumor site where they still could be detected on day 5. The majority of these Luc⁺ cells was detectable in the spleen and TDLN.

Encounter of APCs and T cells is regulated by chemokine gradients in T-cell areas of secondary lymphoid organs (21). Therefore, we studied the chemokine receptor pattern of OVA-CD40B cells and could indeed confirm that they display a pattern of chemokine receptors that is consistent with migration to secondary lymphoid organs (10, 22). CXCR4 and CCR7 were upregulated in OVA-B cells compared with non–OVA-B cells of the same mice (Supplementary Fig. S5). On the other hand, CXCR5 and CD62L were not significantly altered in OVA-B cells. Compared with an unstimulated control, activation by CD40L resulted in further upregulation of CCR7 and CD62L on OVA-B cells. CXCR4 expression did not vary over the observed period, whereas CXCR5 was reduced by 64.61% ± 13.59% (Fig. 3D). This expression pattern is consistent with homing to T-cell-rich areas in the secondary lymphoid organs (23, 24).

The classic effector function of B cells is the secretion of antibodies upon differentiation into plasma cells (PC). We reasoned that the combination of this effector function with antigen-presentation offers additional advantages for cancer immunotherapy. To induce differentiation into antibody-secreting PCs, we developed a protocol in which antigen-specific B cells were stimulated with a combination of cytokines and a BCR agonist. PC differentiation was confirmed by analysis of CD138 expression (Fig. 4A). We obtained the most efficient differentiation when stimulating OVA-B cell cultures with OVA, IL4, IL21, and anti-CD40 (44.24% ± 0.76). IL21 and anti-CD40 were crucial, because stimulation without either of the two stimuli resulted in decreased percentages of CD138⁺ PCs. Based on these results, OVA-B cells were stimulated with the combination of OVA-protein, IL4, IL21, and anti-CD40 to produce PCs for subsequent in vivo experiments. Treatment of naïve polyclonal B cells also resulted in differentiation into CD138⁺ PCs; however, maturation of polyclonal B cells to PCs was less efficient than for OVA-B cells. To further characterize these OVA-specific PCs, we next assessed their ability to produce OVA-specific antibodies. In accordance with the observed changes in CD138⁺ B cells, OVA-B cells secreted OVA-specific IgG1 antibodies when treated with the cytokine cocktail (6.68 ng/mL ± 0.34; Fig. 4B). In addition, antibodies of the subclasses IgG2b, IgG3, or IgM (Fig. 4C) could be detected after stimulation with the cytokine combination. Polyclonal B cells did not secrete OVA-specific antibodies (Fig. 4B and C).

The affinity of these OVA-specific IgG1 antibodies in supernatants of cultures was determined by competition ELISAs. Secreted OVA-specific IgG1 antibodies had a higher affinity for the soluble antigen than IgG1 clone TOSG1C6 (Biolegend), a commercially available standard antibody (LogIC₅₀: −1.397 vs. −1.523; Fig. 4D). In cocultures with splenocytes and OVA-expressing E.G7 cells, the secreted antibodies induced NK cell activation (Fig. 4E) and killing of tumor cells (Fig. 4F), thereby demonstrating their functional activity and suggesting a mechanism for the observed antibody-mediated cytotoxicity.

Next, the influence of a preventive immune intervention with tumor antigen-specific CD40B cells on tumor growth was investigated. For this purpose, protein-pulsed APCs (0.1–2 × 10⁶) were intravenously injected into mice thrice every 7 days. On day 28, immunized mice were inoculated s.c. with E.G7-lymphoma cells. Tumor growth in mice treated with OVA-CD40B cells was reduced and survival was prolonged compared with mice that were immunized with DCs or polyclonal CD40B cells. The combined treatment with OVA-CD40B cells and PCs was superior in protecting mice from tumor challenge and delayed tumor growth compared with mice that were immunized with DCs or OVA-CD40B cells alone (Supplementary Fig. S6A and S6B). Survival (Supplementary Fig. S6C) was also prolonged. The presence of OVA-specific IgG1 antibodies in serum of treated mice was confirmed by ELISAs before and after vaccination (Supplementary Fig. S6D). Mice treated with the combination of OVA-CD40B cells and PCs had more specific antibodies in their serum than mice treated with DCs. Mice that stayed tumor-free until day 50 after the first tumor inoculation were rechallenged with E.G7 cells and tumor growth was observed until day 84 (Supplementary Fig. S6E). Mice that were treated with OVA-CD40B cells alone or in combination with PCs were protected from renewed tumor challenge.

To extend our findings to other disease models and provide evidence for therapeutic efficacy of our concept, mice were challenged s.c. with Panc02OVA tumor cells and treated twice with 0.1–2 × 10⁶ APCs as soon as the tumor became palpable. Tumors grew less in mice that were treated with the combination of OVA-CD40B cells and PCs compared with controls (Fig. 5A). OVA-CD40B cells alone reduced tumor growth as effectively as treatment with DCs. Three of four mice that were treated with the combination of OVA-CD40B cells and PCs, and two of four mice that were treated with either cell type alone remained tumor-free until day 65 (Supplementary Fig. S7A). In contrast, tumors of mice treated with DCs grew more slowly, but they grew out in four of five mice. Accordingly, mice treated with OVA-CD40B cells and PCs survived longer than controls (Fig. 5B). In mice treated with PCs alone or the combination with OVA-CD40B cells, the therapeutic activity was positively correlated with the amount of OVA-specific IgG1 antibodies (Fig. 5C). To explore the mechanism of the induced antitumor immunity, mice were treated with CD4⁺ or CD8⁺ T cell–depleting antibodies on day 5 and 25 after tumor inoculation. Efficient depletion of T cells was determined in blood samples (Supplementary Fig. S7B) before mice were either treated with OVA-CD40B cells or with PBS as negative control. Depletion of CD8⁺ T cells resulted in increased tumor growth (Fig. 5D) and reduced survival (Fig. 5E) compared with nondepleted mice. Depletion of CD4⁺ T cells had contrary effects. CD4⁺ T-cell depletion resulted in reduced tumor growth and prolonged survival. This effect might be due to depletion of Foxp3⁺ Tregs, because Panc02-OVA tumors possess a high percentage of Foxp3⁺ cells (12.54 % ± 1.4%) in the tumor-infiltrating CD4⁺ T-cell compartment (Supplementary Fig. S7C and ref. 25), which was also targeted by the CD4⁺ T cell–depleting antibody (Supplementary Fig. S7C). Tumors in CD4⁺ T cell–depleted negative controls showed recurrent tumor growth from day 29 onward, whereas mice additionally treated with CD40B cells stayed tumor free until the end of the experiment. Taken together, these results support the value of coordinated exploitation of synergistic B-cell effector functions for cancer immunotherapy.

As a proof-of-principle for translation into clinical application, we developed a protocol for isolation of human HBV-specific B cells (HBV-B cells) from vaccinated donors based on the method established for murine antigen-specific B cells (Fig. 6A). The CD20⁺ CD19⁺ B-cell population consisted of ∼3% HBV-B cells. Selection of HBV-B cells via HBV-biotin tetramers resulted in a 10-fold enrichment of the antigen-specific B-cell population. In addition, B cells specific for the two shared self/tumor antigens CEA and NY-ESO-1 were also detected by this method in single-cell suspensions of TDLN obtained from patients with colorectal or gastric esophageal cancer, respectively (Fig. 6B).

Purified HBV-B cells were characterized by flow cytometry. Analogous to murine antigen-specific B cells, they showed an activated phenotype by upregulating the activation marker CD86 and MHCII (Fig. 6C). Next, human antigen-specific B cells were incubated with soluble CD40L to stimulate their antigen-presenting function. On day 7 or 11 of culture, the phenotype of the B cells was assessed. HBV-CD40B cells upregulated the activation markers CD80, CD86, and HLA-DR (Fig. 6D) compared with unstimulated cells. Like HBV-CD40Bs, CEA or NY-ESO-1-specific B cells also proliferated when activated with the CD40L (Fig. 6B). IFNγ fluorospot analysis of cocultures of NY-ESO-1–specific CD40B cells from TDLN cells together with T cells of the same donor confirmed the immunostimulatory capacity of NY-ESO-1–specific CD40B cells (Fig. 6E and Fig. 6F). Although all patients showed NY-ESO-1–specific B cells in flow cytometry analyses, they showed variable T-cell responses in fluorospot analyses and thereby were classified as low and high responders. Within low responders, the T-cell response induced by protein-pulsed CD40B cells from TDLN was higher than that induced by CD40B cells from PBMCs. In one donor, B cells were also isolated and cultivated from single cell suspensions of the tumor. These CD40B cells induced similar T-cell responses when pulsed with the specific protein as B cells from TDLN or PBMCs in high responders (Fig. 6F). Similar results were obtained in IFNγ fluorospot analyses with CEA-protein pulsed CD40B cells from patients with colorectal cancer (Supplementary Fig. S8A and S8B). Antigen-presentation confirmed our results obtained from fluorospot analyses by showing induction of T-cell proliferation (Supplementary Fig. S8C).

After decades of rather disappointing results, cancer immunotherapy has experienced a clinical breakthrough. Immune checkpoint blockade with anti-CTLA4 or anti–PD-1 and CAR T-cell therapies have yielded favorable results in clinical trials (1). Unfortunately, only about 20% of patients show a response to checkpoint blockade. However, approximately 75% of patients who respond show a durable response (26). The remaining 25% relapse because the tumor acquires resistance to immunotherapy through several distinct mechanisms, including defects in immune signaling or antigen presentation (27). Therefore, despite advances in immunotherapy, there still is the need for improvement. Complementary immunotherapeutic approaches may be useful.

The present study reports a cellular immunotherapy based on tumor antigen-specific B cells. The method combines the secretion of specific antibodies with the induction of T-cell immunity and offers the possibility of translation to different kinds of cancer or infectious diseases with a known antigen.

B cells play an important role in antitumoral immunity (28). Whereas regulatory B-cell subsets can impair tumor immunity, activated B cells can also contribute to immunosurveillance of cancers (29). In animal models, the antitumoral activities of B cells are mediated by several immunologic mechanisms including antibody-dependent effects and stimulation of tumor-specific T-cell responses (30). We aimed to exploit the immunotherapeutic potential of B cells by combining both their antibody-mediated and the T-cell stimulatory antitumoral effector mechanisms.

Published approaches on B cell–based immunotherapy (31, 32) have used polyclonal, and not antigen-specific, B lymphocytes. The role of antigen-specificity of B cells has been poorly studied, likely due to difficulties in isolation and characterization of antigen-specific B cells (17, 33). As a consequence, isolation of antigen-specific B cells and use of their APC function have not been combined or tested in vivo. Here, we report the isolation and enrichment of both murine and human antigen-specific B cells by using antigen tetramers, and subsequent differentiation of these B cells into APCs and antibody-producing PCs for use in cancer immunotherapy.

We demonstrated that OVA-CD40B cells were superior to polyclonal CD40B cells in the induction of a T-cell response resulting from affinity of the BCR for its antigen (19). The antigen concentration for pulsing of OVA-CD40B cells was 0.03% of that of the minimal concentration necessary for presentation after fluid-phase pinocytosis of antigen, typical for antigen uptake by polyclonal CD40B cells (19). Although DCs have been considered to be more potent APCs than B cells (34), we found that OVA-CD40B cells and DCs had a similar APC capacity in vitro and in vivo.

APCs must physically encounter T cells in order to induce immune responses. In vivo migration experiments confirmed homing of polyclonal and OVA-CD40B cells into the secondary lymphoid organs of mice. After entry into lymphoid tissue, CD40B cells accumulate both in B-cell follicles and at the edge of the T-cell zone (35), enabling them to interact with T cells. This would constitute an advantage of CD40B cells over DCs, which have been shown to poorly migrate to secondary lymphoid organs after injection and to instead stay at the site of injection (36). In tumor-bearing mice, tumor antigen-specific CD40B cells appeared in the tumor three days after injection. These results are supported by studies that detected tumor antigen-specific antibodies in human breast cancer (37), thereby suggesting that at least some of the tumor-infiltrating B cells are antigen-specific. At the same time, about 90% of injected CD40B cells stayed in TDLN, allowing interaction with T cells.

Tumor-infiltrating B cells and antibody-secreting PCs play a crucial role in the context of tumor immunity. Tumor-infiltrating B cells are associated with improved survival (38, 39). With the expectation to enhance the antitumor immune response, we combined cellular and humoral immune functions of B cells by treating mice with a combinatorial immunotherapeutic approach using antigen-specific CD40B cells as APCs in combination with antibody-secreting PCs. This dual approach seems to represent an improvement over conventional immunotherapeutic strategies because it combines two distinct immunologic mechanisms.

Tumor-specific antibodies can mediate effective antitumor immunity (40). In patients with cancer, the presence of certain antibodies against tumor antigens is associated with improved survival (41). We demonstrated that the induced PCs secrete OVA-specific antibodies of the major subclass IgG1 but also IgG2b, IgG3, and IgM, which were functional and capable of inducing NK cell–mediated cytotoxicity. All four subclasses were shown to be induced by IL21 and anti-CD40 stimulation and represent a T_H2-polarized immune response (42). In addition, antigen-presenting B cells can efficiently induce T-cell responses against tumor antigen. Activated B cells are relatively resistant to inhibition by tumor-associated immunosuppressive molecules such as IL10, TGF-β, and VEGF (43). Although some studies failed to demonstrate an effect of therapeutic treatment with polyclonal CD40B cells on tumor growth (44, 45), we showed that antigen-specific B cells are superior to nonspecific B cells and that combined immunotherapy with tumor antigen-specific CD40B cells and PCs resulted in the remission of established tumors. The combination of two distinct antitumoral immune mechanisms may also prove to be more resistant to immune escape. Our strategy might be usefully combined with checkpoint blockade.

Although the murine studies provided a proof-of-principle for our bifunctional B cell–based vaccine platform, we also performed additional experiments to evaluate the translational potential of this immunotherapeutic approach. We demonstrated that antigen-specific B cells can be isolated from TDLN of cancer patients and that these tumor antigen-specific B cells can be differentiated into potent APCs. Although lymphocytes are usually more tolerant against shared self/tumor antigens, we show that CD40B cells from cancer patients induce strong T-cell responses against these antigens. These results are in line with earlier studies showing antigen presentation of cancer testis antigens and induction of T-cell responses with polyclonal CD40B cells isolated from PBMCs (4, 6). Nevertheless, in contrast to TDLN, we could not detect tumor antigen-specific B cells in PBMCs of cancer patients. Moreover, T-cell responses were lower when induced by protein-pulsed CD40B cells derived from PBMCs than from TDLN. We therefore conclude that TDLN are a more promising source of tumor antigen-specific B cells with regard to antigen-presentation and T-cell induction. Although tumor-infiltrating B cells did not stain for CEA or NY-ESO-1, they showed induction of T-cell responses in fluorospot analysis. We conclude that tumor antigen-specific B cells were present, but could not be stained with antigen-tetramers because the BCR of tumor-infiltrating B cells was saturated with antigen.

Taken together, our findings demonstrate that the combination of cognate presentation of tumor antigens and production of tumor-reactive antibodies generates antitumor immunity that is capable of inducing tumor remission. Our method allows simultaneous generation of tumor antigen-specific antigen-presenting B cells and antibody-producing plasma cells from small amounts of peripheral blood B cells. This study presents a proof-of-principle in murine tumor models and provides a rationale for the clinical evaluation of this B cell–based immunotherapeutic platform.

H. Abken is a consultant/advisory board member for Miltenyi Biotec. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K. Wennhold, H. Abken, A. Shimabukuro-Vornhagen, M. von Bergwelt-Baildon

Development of methodology: K. Wennhold, O. Utermöhlen, M. von Bergwelt-Baildon

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Wennhold, M. Thelen, H.A. Schlößer, N. Haustein, S. Reuter, M. Garcia-Marquez, A. Lechner, S. Kobold, F. Rataj

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Wennhold, M. Thelen, H.A. Schlößer, N. Haustein, S. Reuter, M. Garcia-Marquez, A. Lechner, S. Kobold, G. Chakupurakal, S. Theurich, H. Abken, A. Shimabukuro-Vornhagen, M. von Bergwelt-Baildon

Writing, review, and/or revision of the manuscript: K. Wennhold, H.A. Schlößer, S. Kobold, F. Rataj, G. Chakupurakal, S. Theurich, M. Hallek, H. Abken, A. Shimabukuro-Vornhagen, M. von Bergwelt-Baildon

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Wennhold, M. Garcia-Marquez, O. Utermöhlen, M. von Bergwelt-Baildon

Study supervision: G. Chakupurakal, A. Shimabukuro-Vornhagen, M. von Bergwelt-Baildon

We thank Holger Spiegel from the Fraunhofer IME in Aachen, Germany, for his expertise in antibody characterization.

K. Wennhold was supported by the Graduate Program for Pharmacology and Experimental Therapeutics. This work was supported by grants from the Else Kröner-Fresenius Stiftung, Köln Fortune and the German Ministry for Innovation, Science, Research and Technology of North Rhine-Westphalia. S. Kobold was supported by the international doctoral program “i-Target: Immunotargeting of cancer” funded by the Elite Network of Bavaria, the Marie-Sklodowska-Curie “Training Network for the Immunotherapy of Cancer (IMMUTRAIN),” the Deutsche Krebshilfe, the Wilhelm Sander Stiftung, the Ernst Jung Stiftung, and the Else Kröner Fresenius Stiftung.

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.