B Cell-Based Cancer Immunotherapy

B cells are best known for their role as producers of antibodies. Over recent decades, it has become clear that B cells serve much more diverse functions than just antibody production. B cells are an important source of cytokines and chemokines and thus contribute to the regulation of immune responses. Depending on the mode of activation, the subtype involved, or the microenvironment, B cells either contribute to upregulation of T-cell responses or they can exert immunoregulatory functions and participate in the downregulation of T-cell immunity [reviewed in 1].

In the 1980s, the ability of B cells to act as antigen-presenting cells (APCs) became increasingly appreciated. However, concurrently dendritic cells (DCs) were characterized as potent professional APCs. Due to their potent antigen-presenting capacity, DCs were regarded as the primary APCs for the induction of T-cell immunity and became the main focus for further development of cellular cancer vaccines. However, DCs possess several important drawbacks as APCs for cellular cancer vaccines. It is difficult and relatively expensive to generate sufficient amounts of DCs for repeated vaccinations. Furthermore, there are a large variety of protocols using different cytokine cocktails to generate DCs for immunotherapeutic purposes. Little is known about which protocol is optimal. Therefore, several research groups have investigated alternative cellular adjuvants.

Activated B cells become potent professional APCs only when appropriately activated. Soon after CD40 and its ligand CD40L (also named CD154) were first described, it became clear that CD40L/CD40 signaling was among the most potent stimuli for the activation of B cells [2, 3]. Classically, CD40L is expressed on activated CD4+ T cells and, thus, is essential for a thymus-dependent B-cell response and for the development of a humoral and cellular immune response. CD40L is a type II transmembrane protein, which exists as a trimer, inducing oligomerization of CD40 upon binding [4], a process that is critical for signaling via the CD40 receptor and likely accounts for the diverse biologic activities induced by different monoclonal antibodies [5]. CD40 acts a transmembrane signal transducer activating intracellular kinases and transcription factors within the cell. More specifically, recruitment of TRAF proteins to the cytoplasmic tail of CD40 activates the canonical and noncanonical NFκB pathways, MAP kinases, phosphoinositide 3-kinases, and the phospholipase Cγ pathway [reviewed in 6]. Independent of TRAF proteins, Janus family kinase 3 can directly bind to the cytoplasmic tail of CD40 inducing phosphorylation of STAT5 [7, 8]. These signaling cascades in B cells eventually promote germinal center formation, immunoglobulin isotype switch, somatic hypermutation, and formation of long-loved plasma cells or memory B cells [9-12]. Moreover, the CD40L/CD40 interaction is involved in the cellular immune response by regulating the costimulatory activity of APCs [13] and thus influences T-cell priming and effector functions. This discovery resulted in the development of cell culture systems that allow the activation and expansion of B cells from peripheral blood [14]. In the late 1990s, Schultze et al. [15] proposed in vitro-generated CD40-activated B cells (CD40B cells) as an alternative to DCs as cellular adjuvant for cancer immunotherapy. Ex vivo-generated CD40B cells possess potent immunostimulatory properties and are capable of priming CD4 and CD8 T cells in vitro and in vivo [16-18]. Over the subsequent years, the antigen-presenting function of B cells was characterized in more detail and the concept of B cell-based cancer vaccines was increasingly refined. Several experimental studies in different tumor models confirmed that vaccination with CD40B cells could induce effective antitumor CD4 and CD8 T-cell responses.

In 2005, Biagi et al. [19] reported the first small clinical trial of a cancer vaccine that used CD40B cells as cellular adjuvant. They transduced autologous leukemic B cells isolated from patients with chronic lymphocytic leukemia (CLL) with an adenoviral vector that contained the human CD40L gene and reinfused these cells together with transduced autologous CLL cells that expressed interleukin (IL)-2. Three of 9 patients demonstrated a greater than 50% reduction in lymph node size. Unfortunately, the induced T-cell responses were only transient and unable to overcome tumor-induced immunosuppression in the long term. In spite of these disappointing results, this study provided a first proof-of-concept for B cell-based cancer immunotherapy and demonstrated that antitumor T-cell responses can be induced by activated antigen-presenting B lymphocytes.

Resting B lymphocytes are poor APCs and are unable to induce strong T-cell immunity [20]. B cells can be activated by a variety of stimuli to acquire immunostimulatory capacity, including B-cell receptor (BCR) binding to antigen and toll-like receptor-mediated signals. However, signals transmitted via CD40 have consistently been found to be the most potent inducer of many features of potent APCs [2]. Several strategies have been investigated to exploit CD40-CD40L interaction for the generation of antigen-presenting B cells (summarized in Table 1 for human B cells ) [reviewed in 21]. These include the usage of recombinant soluble CD40L proteins [22-25], triggering CD40 with agonistic monoclonal CD40 antibodies [26, 27], and CD40L-expressing feeder cells [28-30]. A number of factors affect the extent of B-cell activation by CD40-mediated signals. For instance, the effect of anti-CD40 antibodies on B-cell activation is determined by the exact location of their binding to CD40 [5]. Another factor that crucially determines the extent of B cell activation is the degree of CD40 crosslinking. It has long been established that optimal bioactivity is only observed when using a multimerized form of the CD40L homotrimer, thus allowing clustering on the cell surface [31-33]. Clustering of the CD40L is not elicited by monoclonal anti-CD40 antibodies, thus only inducing activation, but not proliferation of B cells [31-33]. CD40L- expressing feeder cells naturally provide a multimerized form of the CD40L, but to avoid xenogeneic components in clinical products recombinant soluble CD40L is the preferred choice for a clinical application of B cells.

Typically, human peripheral blood mononuclear cells (PBMCs) or purified B cells are cultured for a period of at least 14 days in the presence of the soluble CD40L and IL-4 [22, 23], in which the addition of IL-4 is necessary for B-cell proliferation [34]. These culture conditions result in a profound polyclonal activation of B cells that leads to an approximately 20-fold expansion [15, 35, 36] and the acquisition of an antigen-presenting phenotype [15, 37, 38]. When PBMCs are used as the starting material, typically B-cell purities of more than 95% can be achieved. Throughout the culture period, B cells acquire a memory-like state that represents an intermediary stage between naïve B cells and plasma cells [39]. B cells that are stimulated for at least 3 days by the CD40L show a high expression of MHC class I and MHC class II molecules, the costimulatory markers CD80, CD83, and CD86, and the adhesion molecules CD54 and CD58, which remains stable throughout the subsequent culture period [15, 18, 35, 37, 38]. Combination of CD40L stimulation with CpG as proposed by some studies [18, 40] has no further impact on the expression of activation markers or proliferation of B cells, while additional stimulation with LPS further increases the activation of B cells [18, 40]. When normalized relative to cell size, expression levels of activation molecules on the cell surface of CD40B cells are equivalent to CD40L/IFN-γ or TNF-α-matured DCs [35].

Increased expression of MHC and costimulatory molecules on CD40B cells correlates with the acquisition of antigen-presenting functions. CD40 activation results in improved antigen processing and presentation [41], typically via the classical MHC class II pathway [18, 42], but also a distinct nonclassical, cytosolic MHC class II pathway [43]. Becker et al. [43] demonstrated that presentation of the model antigen CMV pp65 by CD40B cells was limited when using the proteasome inhibitor epoxomicin resulting in reduced T-cell activation and IFN-γ production. However, epoxomicin sensitivity was not observed in DCs, suggesting an antigen-processing mechanism unique to CD40B cells.

The ability of human CD40B cells to expand antigen-experienced CD4+ T cells, but also to prime naïve CD4+ T cells was demonstrated in several studies [15, 17, 22, 24, 26, 35, 37, 44, 45]. Lapointe et al. [37] showed that when pulsed with tumor lysates, CD40B cells expanded and activated tumor antigen-specific memory CD4+ T cells from the blood of cancer patients. Our group demonstrated that responses of naïve CD4+ T cells against MCH class II-restricted neoantigens could be induced when using CD40B cells as sole APCs [42]. In addition, expression of CD107a and CD40L was detected in CD4+ T cells early after activation with CD40B cells [46].

Human CD40B cells also cross-present antigen via MHC class I pathways and, thus, where shown to induce naïve and memory CD8+ T-cell responses [47-49]. Similar to the system used for CD4+ T cells, CD40B cells were used as APCs to expand antigen-specific CD8+ T cells from healthy donors and cancer patients [15, 35, 36, 45, 48-50]. Specific T-cell responses where not only detected against the memory antigens influenza A MP58, MART-1, and hTERT, but also the neoantigen RTpol from HIV [35], thus again demonstrating the ability of CD40B cells to induce naïve T-cell responses.

CD40B cells fulfill crucial requirements for their use as APCs in cancer immunotherapy: (1) they can be consistently generated from peripheral blood, (2) they are relatively insensitive towards tumor-derived immunosuppressive mechanisms, (3) they do not induce tolerance by themselves, and (4) they are well tolerated upon infusion in terms of toxic side effects.

From a practical view, CD40B cells offer several potential advantages over DCs. From a small amount of peripheral blood, one can usually obtain sufficient numbers (approximately 1 × 10⁵ to 1 × 10⁷ cells/kg body weight) of activated antigen-presenting B cells [35, 51], whereas the generation of DCs typically requires a leukapheresis [52-55]. It has been shown that this is even feasible in cancer patients [35, 51]. This aspect is particularly important considering that cancer patients typically are frequently lymphocytopenic due to the underlying disease and/or prior chemotherapy. Furthermore, the culture system for generating CD40B cells is relatively easy and inexpensive.

Tumor-derived factors mediating immunosuppression in the tumor microenvironment, such as prostaglandin E2 [56], TGF-β [57, 58], VEGF [59, 60] or IL-10 [61, 62], act in part by inhibiting DC differentiation, maturation, trafficking, and antigen presentation [62, 63]. Therefore, one might suppose that they have similar effects on antigen-presenting B cells. However, activated B cells turned out to be relatively resistant to inhibition by tumor-associated immunosuppressive molecules. In vitro, neither migration nor activation of CD40B cells was inhibited by these immunosuppressive factors, nor did they influence the ability of CD40B cells to induce proliferation of CD4+ or CD8+ T cells [64]. TGF-β and VEGF had no effect on the proliferation of CD40B cells, while IL-10 even increased their expansion. On the contrary, TGF-β actually enhances BCR-mediated antigen presentation [65]. Concerning the induction of tolerance by administration of activated B cells, the mode of activation is of considerable importance. While human B cells that were activated by bacterial stimuli induced anergy and apoptosis of CD4+ T cells in an IL-2-dependent manner [66], CD40B cells were shown to activate T cells in the presence of IL-2 besides the fact that they express CD25 [37, 44, 67]. Toxic side effects of CD40B-cell administration were not observed in in vivo in studies with mice or dogs. Wild-type mice received autologous CD40B cells in different injection routes (intravenous, subcutaneous, and intraperitoneal) and two different high concentrations (40 × 10⁶ and 40 × 10⁷ cells/kg). Body weight and survival remained unchained under all tested conditions. No abnormal lymphocytic infiltration, structural tissue injury, or indications of inflammation could be detected in histological analyses of heart, lung, liver, spleen, and kidney [68]. These results are in line with a study where administration of RNA-loaded CD40B cells was well tolerated by dogs with non-Hodgkin’s lymphoma and no long-term complications were observed in the follow-up [69].

Only few studies investigated antigen presentation by CD40B cells or the influence of their administration on tumor growth in vivo (summarized in Table 2). Sorenmo et al. [69] published results of a study using tumor RNA-loaded CD40B cells as cellular adjuvant in dogs with non-Hodgkin’s lymphoma. The authors reported positive specific immune responses as detected by IFN-γ ELISPOTs, but could not detect a statistically significant correlation between the immunological response and the clinical outcome. However, they detected a significant improvement in the rate of durable second remission and survival between vaccinated and nonvaccinated groups. Since dogs are a widely accepted animal model to evaluate safety and efficacy before proceeding to a clinical trial, this study was an important step towards a clinical application of CD40B cells.

More promising results in terms of cancer treatment were reported in mice. Vaccination of wild-type mice with LCMV-antigen pulsed CD40B cells, but not LPS-activated B cells, significantly reduced growth of LL-LCMV subcutaneous tumors [70]. In two studies, which applied RNA-transfected or OVA antigen-pulsed CD40B cells in a therapeutic setting, treatment did not result in delayed tumor growth of B16.F10 melanomas or E.G7 lymphomas, respectively [71, 72]. In two studies, B cells were isolated from tumor-draining lymph nodes (TDLN) of wild-type mice with MCA205, D5G6, or 4T1 tumors [73, 74]. After activation with anti-CD40 antibodies, they were adoptively transferred into syngeneic tumor-bearing mice. In combination with activated T cells, CD40B-cell administration resulted in the reduction of spontaneous metastases. Moreover, combining adoptive transfer of B cells with chemotherapy or total body irradiation significantly inhibited tumor growth. The generated B cells were shown to produce tumor antigen-specific IgG antibodies, indicating specificity for tumor antigens presented by B cells isolated from TDLN. However, these studies used soluble anti-CD40 antibodies for the activation of B cells, which was demonstrated to result in weaker CD40 stimulation than activation by CD40L-expressing feeder cells [21, 31]. When using CD40L-expressing feeder cells for activation, vaccination with tumor antigen-pulsed CD40B cells before B16 melanomas or E.G7 lymphomas were injected [68] resulted in significantly delayed growth in both tumor models. The rate of tumor control by CD40B cell vaccination was comparable to that induced by DCs. Using tumor antigen-specific B cells for immunotherapy seems to further improve the observed antitumor efficacy. Moutai et al. [75] isolated HEL-specific B cells and stimulated them with a combination of CD40L/BAFF-expressing feeder cells, IL-4, and IL-21. Therapeutic administration of these iGC-termed B cells resulted in the regression of pulmonary metastases of HEL-expressing B16 melanomas. These results are in line with a more recent study using tumor antigen-specific CD40B cells for therapeutic treatment of EG.7 lymphoma or Panc02OVA tumor-bearing mice [76]. This study exploited the advantage of antigen-specific B cells to take up and process antigen more efficiently via the specific BCR than polyclonal B cells do via BCR-independent mechanisms such as pinocytosis [77]. Antigen-specific B cells more efficiently induced antigen-specific T-cell responses in vitro and in vivo than polyclonal CD40B cells, subsequently resulting in complete remission in 60% of mice [76]. In addition, B cells were differentiated into antibody-secreting plasma cells supporting the antitumor immune response induced by CD40B cells.

The preclinical experiments described above provide a strong rational for the clinical application of CD40B cells as a cellular cancer vaccine. The proof-of-principle studies in several distinct murine cancer models and the more genetically diverse canine tumors demonstrate the potential of B cell-based cancer vaccines for the therapeutic treatment of established tumors. Apart from the above-mentioned clinical study of a CD40-activated B cell vaccine by Biagi et al. [19], there are only few clinical studies that assessed the use of B cells for cancer immunotherapy.

In two small clinical trials, B cells were used as part of a hybrid cell vaccination approach, in which allogeneic B cells from PBMCs of healthy donors were fused with autologous tumor cells. In the first study in patients with renal cell carcinoma, two complete and two partial responses were observed out of 11 patients. Most patients at least showed an initial response and the vaccination was well tolerated [78]. The second study was conducted in patients with metastatic melanoma. The vaccination with the hybrid vaccine induced T-cell relocation into the tumor nodules. Out of 16 patients, 1 complete and 1 partial remission and 5 cases of stable disease were observed. The vaccination proved to be safe as only minor side effects occurred [79].

Another study in humans using adoptive B-cell transfer rather focused on the ability of memory B cells to differentiate into plasma cells. Winkler and colleagues [80] developed a method to produce good manufacturing practice (GMP)-conforming purified human B cells for the treatment of patients after allogeneic stem cell transplantation to restore humoral immunity. They initiated a first-in-man phase I/IIa clinical trial to evaluate safety and tolerability of adoptively transferred donor B cells in a dose escalation study [81] (ClinicalTrials.gov identifier: NCT02007811). B cells were isolated under GMP-conditions from donor leukapheresis products in two separation steps in the CliniMACS® System including the depletion of CD3+ T cells followed by positive selection of CD19+ B cells. When the first results were reported (in 2016 at the ASH conference), the lower doses of 0.5 × 10⁶, 1 × 10⁶, and 2 × 10⁶ B cells were well tolerated without any acute adverse reactions or chronic GvHD reactions during the observation period of 4 month. As secondary endpoints, the activity of the infused donor memory B cells was evaluated. Preliminary results suggested a significant mobilization of plasma blasts in some of the patients after revaccination with a pentavalent vaccine.

The recent development of a GMP-grade CD40-activating reagent has been one of the important steps towards the clinical testing of a CD40B cell-based cancer vaccine [22]. This B cell-activating reagent has overcome some of the problems described above with other activating reagents, i.e., xenogeneic components or poor proliferation.

Apart from the effective activation and expansion of immunostimulatory B cells, the process of loading B cells with antigen is crucial for the successful application of B cell-based cancer vaccines. BCR-mediated antigen uptake is the most efficient way of antigen acquisition and leads to highly efficient antigen processing and presentation [77, 82, 83]. Other modes of antigen uptake such as pinocytosis are less effective. Therefore, several different strategies for antigen delivery to B cells have been explored. A promising approach of antigen delivery to B cells is via the targeting of antigens to CD19 [84]. Szeto et al. [85] recently reported another interesting approach using a microfluidic device for antigen delivery to B cells through a process termed mechanoporation. In this microfluidic device, B cells are passed through narrow channels. The passage through the narrow channel causes the transient formation of pores in the B-cell membrane that facilitate the intracellular uptake of proteins from the surrounding medium.

A possible alternative to the use of polyclonal B cells that have to be loaded with tumor antigens is the isolation of B cells with tumor antigen-specific BCRs from the patients’ blood or tumor tissue. Since antigen-uptake through the BCR is highly specific and results in rapid and effective antigen processing and presentation, one can circumvent the need for antigen-loading prior to reinfusion of the B-cell vaccine. At least in mice, the use of antigen-specific CD40B cells for immunotherapy was highly efficient in inducing a strong antitumor immune response resulting in complete remission [76]. However, like the use of polyclonal B cells, this method requires the prior choice of a defined tumor antigen.

Another interesting antigen-agnostic strategy for the generation of tumor antigen-specific immunostimulatory B cells is the use of B cells that were isolated from the patient’s tumor or TDLN. In murine experiments this approach proved to be successful at inducing antitumor immunity [74]. These results are in line with an in vitro study where B cells isolated from TDLN of patients with esophageal-gastric cancer or colorectal cancer were partially specific for the tumor antigens NY-ESO-1 or CEA, respectively, and induced antigen-specific T-cell responses in vitro [76].

A possible first study with CD40B cells should strive to include the aspect of antigen specificity, to ensure that their full potential is exploited. However, since patient material is limited and isolation of antigen-specific B cells by antigen-tetramers is complex and costly to be developed in GMP-grade, isolation of the whole B-cell population from TILs offers the most promising option. This B-cell population contains B cells specific for tumor antigens, is presumably loaded with tumor antigen already, but can also be further stimulated with the CD40L [76]. Thus, a tumor entity should be chosen where TILs are easily assessable, i.e., surgery is part of the standard procedure, a possible tumor antigen for pulsing is known, and where there is a great clinical need. Manufacturing of B cells under GMP-conditions comprises no obstacles anymore after today’s experience with CAR trials [86] and B cell-adoptive transfer [81], and suitable tumor antigens for pulsing have been discovered in many solid tumor entities [87]. The whole isolation and activation process of CD40B cells in general would also be suitable for an automated manufacturing process, e.g., in the CliniMACS Prodigy (Miltenyi Biotec) [88]. The most straightforward approach would thus include isolation of B cells from TILs by CD19+ microbeads, activation and expansion with the CD40L, loading with antigen after control of the activation status, and reinjection into the patient (Fig. 1). The primary objectives would of course be the feasibility, safety, and toxicity of a CD40B-cell vaccination, but surely the induction of an immune response, persistence of transfused CD40B cells, and evidence of disease control would be equally exciting secondary objectives.

Even though cancer vaccination has long been regarded as a promising approach for cancer immunotherapy, the sobering results of early clinical trials and the economic failure of the few approved cancer vaccines have led to reduced interest in the further development of cellular cancer vaccines. However, the current excitement about the success of immune checkpoint blockade in the treatment of a broad range of malignancies has sparked a renaissance of cancer vaccines [52]. Currently, several clinical trials are investigating the combination of DC vaccines with checkpoint inhibitors. Tumor-induced T-cell dysfunction seems to be the major immunologic mechanism that limits the ability of cellular vaccines to elicit an antitumor immune response [89, 90]. Thus, combining B-cell immunotherapy with drugs that reverse T-cell dysfunction appear to be a plausible future line of investigation.

In particular, combination of CD40B-cell vaccination and checkpoint inhibition represents a promising combination approach to further enhance the activity of B cell-based cancer immunotherapy. There are already several checkpoint inhibitors that are approved for clinical use and preclinical studies demonstrate that a dual strategy of active tumor vaccination and checkpoint blockade can overcome tumor-induced immune escape [91]. Furthermore, it can be expected that in the near future additional drugs that reverse T-cell dysfunction become available [92].

Taken together, the work of recent years that we summarized here strongly highlight the potential of B cells for immunotherapy and their applicability in a clinical setting. The most challenging obstacles for the use of CD40B cells in humans have been overcome in the meantime. The CD40-activation culture system is a versatile tool for the generation of activated B cells that can be used for immunotherapeutic purposes. The current success of chimeric antigen-receptor T cells will lead to a more widespread establishment of the infrastructure required for the clinical application of cellular therapies. In addition, technological advances such as small, automated, closed system cell manufacturing platforms that enable the decentralized “point-of-care” generation of cellular therapies will further ease the clinical testing of cellular immunotherapies such as CD40B-cell cancer vaccines [88]. Therefore, it can be expected that the near future will see the first clinical trials of B cell-based cancer vaccines. These trials will show if B cells deserve a place in the oncologist’s toolbox.

K.W. and A.S.-V. contributed equally to this work.