B cells in pancreatic cancer stroma

Francesca Romana Delvecchio, Michelle R Goulart, Rachel Elizabeth Ann Fincham, Michele Bombadieri,Hemant M Kocher

Abstract Pancreatic cancer is a disease with high unmet clinical need. Pancreatic cancer is also characterised by an intense fibrotic stroma, which harbours many immune cells. Studies in both human and animal models have demonstrated that the immune system plays a crucial role in modulating tumour onset and progression.In human pancreatic ductal adenocarcinoma, high B-cell infiltration correlates with better patient survival. Hence, B cells have received recent interest in pancreatic cancer as potential therapeutic targets. However, the data on the role of B cells in murine models is unclear as it is dependent on the pancreatic cancer model used to study. Nevertheless, it appears that B cells do organise along with other immune cells such as a network of follicular dendritic cells (DCs),surrounded by T cells and DCs to form tertiary lymphoid structures (TLS). TLS are increasingly recognised as sites for antigen presentation, T-cell activation, Bcell maturation and differentiation in plasma cells. In this review we dissect the role of B cells and provide directions for future studies to harness the role of B cells in treatment of human pancreatic cancer.

Key Words: B cells; Pancreatic cancer; Cancer immunology; Tertiary lymphoid structures;Anti-tumour immunoglobulins; Plasma cells

lNTRODUCTlON

Pancreatic cancer and its immune infiltration

The majority (about 95 %) of pancreatic cancers (adenocarcinomas) arise from the exocrine pancreas,most likely from the epithelial cells lining the pancreatic duct, to form gland-like structures, and hence,are commonly referred to as pancreatic ductal adenocarcinoma (PDAC), though mucinous tumours are the second most common histological type of pancreatic cancer[1 ]. PDAC is the gastrointestinal tumour with the poorest prognosis, with 80 % of the patients presenting with advanced disease. A mere 15 %-20 % of the patients are suitable for surgical resection, which currently represents the only curative option for pancreatic cancer. For advanced PDAC, the most common systemic treatment is single-agent gemcitabine which is increasingly being replaced with a combination of chemotherapeutics (e.g.,FOLFIRINOX or gemcitabine-nab-paclitaxel), at least in patients with good performance status as firstline treatment[2 ]. Although immunotherapies have gained success in other cancers, there are no approved immunotherapies for PDAC[2 ].

Many immuno-therapeutic approaches are under investigation for PDAC. Immune-checkpoint inhibition has shown clinical benefit in 2 % of PDAC patients harbouring a DNA mismatch repair(MMR) deficiency[3 ,4 ]. Vaccination strategies are also being tested including “personalised” dendritic cell (DC)-vaccines loaded with the antigen[5 ,6 ]. GVAX [granulocyte-macrophage colony-stimulating factor (GM-CSF)-secreting, allogeneic PDAC vaccine, NCT01417000 , NCT00727441 , NCT00084383 ] are being investigated further[7 ,8 ]. Furthermore specific use of immune cells is being explored by adoptive transfer of T cells carrying chimeric antigen receptors[9 -12 ], or recover the immuno-suppression and chemo-sensitivity using Ibrutinib, the inhibitor of Bruton’s tyrosine kinase (BTK), a member of the B-cell receptors (BCR) signalling pathway (NCT02436668 ), targeting regulatory B cells and macrophages.

PDAC is conventionally known as a “cold tumour”, due to low inflammatory cytokine profile and hypoxia, low mutational load and exclusion of infiltrating lymphocytes[13 ,14 ]. Recent research has identified an “immunogenic subtype” enriched in genes associated to B-cell signalling, CD4 + and CD8+T cells, and antigen presentation[13 ]. Furthermore, the combination of genetic, stromal, and immunological features of PDAC can lead to further definition of novel immune-subtypes which may have prognostic value and the possibility of identifying tumours with immuno-therapeutic potential[15 ].Whilst spatial distribution and infiltration of T cells and the formation of clusters with B cells is associated with better outcome in human and murine models of PDAC[16 ,17 ],in vivostudies of B cell depletion in murine models of PDAC describe a pro-tumorigenic role of B cells[18 -21 ]. These discordant findings can be ascribed to the different tumour sub-types analysed and to the use of dissimilar murine models. For example, mice that are genetically lacking in B cells might behave differently to those where depletion of B cells is conducted by a depleting antibody[22 ]. In this review, we critically discuss the evidence for the perceived dichotomous role of B cells in pancreatic cancer.

TUMOUR-SUPPRESSlNG ROLE OF B CELLS lN SOLlD CANCERS

Immuno-histochemical analysis using CD20 , and metagene analysis for B-cell signature, showed a positive correlation between B-cell infiltration and patient prognosis in many different cancer types. For example, work in primary cutaneous melanoma (n= 106 , immunostaining, multivariate analysis)demonstrated that intra- and peri-tumoral B cells are important, in particular CD20 +/OX40+cell density[23 ]. In high-grade serous ovarian cancer (n = 70 , immunostaining of tissues and FACS of peripheral blood), suggested a role for CD27 -memory B cells[24 ]. In basal-like breast cancer (n = 728 , breast cancer,TCGA dataset, B-cell mRNA signature) and non-small cell lung cancer (NSCLC) (n= 74 , untreated patients with early-stage NSCLC and 122 patients with treated advanced-stage NSCLC; immunostaining and FACS analysis) demonstrated a prognostic value for follicular B cells[25 ,26 ]. In sarcoma (n=608 , soft-tissue sarcomas; gene expression profiles) led to the identification of different immunophenotypes, and the B-cell enriched demonstrated improved survival and response to immunotherapy[27 ]. Some studies included the organisation of tumour-infiltrating B cells into tertiary lymphoid structures (TLS) in addition to the B-cell density[28 ,29 ]. B cells are known to act as antigen-presenting cells (APCs) or antibody-producing cells[30 ]. Thus, presence of B cells or at least their subsets or organisation within cancer tissues seem to confer prognostic benefit suggesting a role for humoral immunity in the anti-tumour response mounted by the host[31 ].

Tumours can express antigens recognised as non-self by the immune system to induce a specific antitumour immune response, collectively referred to as the “cancer immunome”[32 ]. In this context, B cells with high affinity for a specific tumour-associated antigen (TAA), engulf and process the antigen to display it on their cell surface; thus, acting as APCs. This complex is recognised by activated T helper cells, which induce B-cell proliferation and clonal expansion. Some B cells may serve as memory cells whilst others act as effector cells that differentiate into antibody-producing plasma cells[33 ]. The antibody-TAA binding also initiates the destruction of the tumour cells expressing the TAA by several mechanisms, such as opsonisation and macrophage recognition and phagocytosis, or blocking of the receptors associated with tumour cell proliferation and survival, or uptakeviaFcγ receptors, leading to antigen cross-presentation and vigorous CD4 + and CD8+T cell responses, complement-dependent cytotoxicity (CDC), or antibody-dependent cellular cytotoxicity (ADCC).

Antibody-production

B cell affinity maturation and differentiation to plasma cells have been described within TLS in several cancers, in addition to the usual places of maturation such as lymph nodes[34 ]. Tumour-specific B cells may acquire somatic hyper-mutations (SHMs) in TLS and extra-follicular B cells maturation has been described[35 -37 ]. Furthermore expansion of tumour-specific B cells without SHMs may reflect a mechanism of T cell-independent or T cell-dependent but germinal centre-independent B cell activation[38 ]. For example, in gastric cancer, tumour-infiltrating B cells showed broad variations in the degrees of SHMs, with some producing functional antibodies directed against sulfated glycosaminoglycan with, at least, tumour growth-suppressive propertiesin vitro[39 ].

Since cancer is driven by mutations in “self-proteins”, cancer-associated auto-antibodies are detectable[40 ]. These antibodies may be in response to “self-antigens” which are either over-expressed [e.g., human epidermal growth factor receptor 2 (HER2 /neu)] or aberrantly expressed (e.g., cancer-testis antigen) during tumorigenesis. Mechanisms for secretion of cancer-related auto-antibodies include changes in the expression levels, altered protein structures, presentation of dying tumour cells (due to chemo/radiotherapy for example) to the immune system leading an abnormal exposure of autologous intracellular antigens[40 ]. Antigen load and duration of exposure may induce humoral immune responses since antibodies against several TAA (such as p53 , New York esophageal squamous cell carcinoma-1 (NY-ESO-1 ), surviving, tyrosinase) were more frequently found in advanced tumour stages[41 ]. Antibodies produced by tumour infiltrating B cells may induce lysis of cancer cells by ADCC or CDC, leading to the direct killing of the cancer cells[42 ]. Murine models demonstrate binding of tumour B-cell antibodies to mouse tumours in an antigen-specific manner and complement-dependent lysis[43 -45 ]. Binding of C3 components to CD21 (the complement receptor 2 ) induces B-cell activation to promote anti-cancer responses[46 ].

Promoting T cell response

B cells may represent the most abundant APC since DCs are scarce in the tissue[47 ]. Tumour infiltrating B cells can also provide antigen-independent help to cytolytic T cells (CTLs) within the tumour, by interaction between CD27 expressed on helper B cells and CD70 on CTLs, promoting their antigenindependent survival and proliferation of T cells[48 ].

Activation of bystander B cells

B cells can also be stimulated by transactivation of bystander B cells not in direct contact with the antigen,viatransfer of human leukocyte antigen-peptide complexes or BCRs contained in exosomes or cytonemes[49 ]. These activated bystander B cells canper seproduce antibodies and/or serve as APC, but also release T-cell activating cytokines, thus amplifying the cellular and humoral immune response,even with a limited antigen load[49 ].

Interaction with T follicular helper cells, in intra-tumour TLS

Presence of TLS within the tumour parenchyma correlates with better patient survival[50 ,51 ]. Within TLS, B cells in close proximity to T cells and interact with T follicular helper cells and follicular DCs and promote germinal centre (GC) reaction, which results in B-cell differentiation into memory B cells and long-term surviving plasma cells. Within TLS, B cells can act as APCs and produce anti-tumour antibodies, exhibiting tumour-specific humoral responsesin situ[26 ,36 ,37 ]. NSCLC-infiltrating B cells were shown to producein vitroimmunoglobulin (Ig) G and IgA directed against tumour antigens(MAGE, LAGE-1 , NY-ESO-1 , P53 )[26 ]. Micro-dissected TLS-derived B cells from breast cancer showed poly-clonality and high mutation rate, suggestive of an affinity maturation occurring within TLS[36 ].Moreover colorectal cancer-infiltrating B cells were shown to produce IgG which bound epitopes on the cell membrane of different tumour cell lines[37 ,52 ]. TLS may also be artificially induced by neo-adjuvant treatment such as with anti-programmed cell death protein (PD) 1 in NSCLC, or vaccination against human papilloma virus (HPV) in cervical cancer patients[53 ,54 ]. Furthermore, presence of TLS is associated with response to immuno-therapy in NSCLC, melanoma and sarcoma patients[27 ,50 ,55 ,56 ].

TUMOUR-PROMOTlNG ROLE OF B CELLS lN SOLlD CANCERS

Immuno-histochemical characterisation of the tumoral immune infiltrate has shown a negative correlation between B cell/plasma cell infiltration with patient survival in melanoma, prostate cancer,lung cancer and ovarian cancer[57 -60 ]. Furthermore, the detection of tumour specific (auto)-antibodies in the sera of cancer patients was associated with poor prognosis[61 ]. Depending on the tumour type studied and murine model investigated, a number of mechanisms for the pro-tumorigenic nature of B cells have been suggested.

Antibody production

Whilst several human studies show a positive correlation between antibodies directed against Her2 /neu or mucin 1 (MUC-1 ) with favourable patient prognosis, high serum anti-p53 antibody levels are associated with poor prognosis[61 -64 ]. It has been speculated that this may be due to high antigen load and exposure rather than a reflection of poor immune activity. Antibodies activate the complement system once they have bound the antigen in the immune-complexes[65 ]. However, murine studies showed that, counter-intuitively, some antibodies might contribute to the progression of tumours by formation of circulating immune-complexes (CICs). These CICs can bind to myeloid cells within tumours, and activate their Fcγ receptors to induce myeloid suppressor cell activity which promotes tumorigenesis[65 ,66 ]. Immune-complexes formation can lead also to the activation of complement cascades resulting in formation of C3 and C5 a anaphylatoxins, which can induce the recruitment of inflammatory cells which, in turn, may provide a rich pro-angiogenic and pro-tumoral environment[66 ]. Deposition of complement components per se does not induce chronic inflammation during tumorigenesis in HPV16 /recombination activating gene 2-/-murine model of skin cancers. However,transfer of competent B cells as well as serum from immuno-competent animals could enhance premalignant to malignant transformation for skin cancer, raising the speculation that B-cell derived antibodies home into the neoplastic tissue and activate the complement cascade, mediating recruitment of innate immune cells; thus, modulating a tumour-promoting chronic inflammation[66 ].

Furthermore, different IgG subclasses have distinct biological function[67 ]. IgG4 is associated with chronic antigen exposure, typical of cancer disease, andin vitroandin vivostudies have demonstrated that this subclass counteracts anti-tumour immunity by antagonising IgG1 -mediated immunity[68 ]. The presence of IgG4 in tumour microenvironment (TME) not only prevents IgG1 -FcR-mediated effector functions, contributing to tumour evasion to humoral immunity, but could also impair therapeutic antibody effector function[69 ].

Cytokine production by B cells

B cells have been shown to directly inhibit cytotoxic T-cell responses in several tumour modelsviathe production of B-cell-derived factor[70 ,71 ]. The negative correlation between high tumour-infiltrating B cells and prognosis in prostate cancer was ascribed to the production of lymphotoxin by tumoral B cells recruited by chemokine (C-X-C motif) ligand 13 (CXCL13 ) signalling, after androgen ablation by castration in a mouse prostate cancer model[58 ,72 ]. Lymphotoxin activates non-canonical and canonical nuclear factor kappa-B signalling and signal transducer and activator of transcription 3 in the remaining cancer cells, resulting in androgen-refractory growth and tumour progression[72 ].

B-regulatory functions

Akin to T-cell subtypes, phenotypically and functionally distinct B-cell subpopulations have been identified. In presence of chronic exposure to the antigen and chronic inflammation, such auto-immune encephalomyelitis or colitis, and cancer, B cells may acquire a regulatory phenotype[73 -75 ]. This subset of B cells have been shown to have immunosuppressive properties, alongside with myeloid-derived suppressor cells or T-regulatory cells (Tregs), thus expanding the team of the suppressive immune players within the TME[76 ,77 ]. These B-regulatory cells act as tumour promoters by affecting the function of other immune cells, through immunosuppressive factors, such as transforming-growth factor (TGF)-β, interleukin (IL)-4 , and IL-10 , which are associated with Th2 skewing of T cells, IL-13 and IL-35 , that support tumour-cell growth as well as M2 polarisation of tumour-associated macrophages(TAMs). Moreover, immunosuppression is further induced through PD1 expression, which, by binding to PD-L1 on the surface of tumour cells, can abrogate tumour recognition and killing. In addition to these indirect mechanisms, B regulatory cells can be directly pro-tumourigenic, for example, B-cell derived TGF-β promotes epithelial-mesenchymal transition in colorectal cancer, or through CD40 /CD154 signalling pathway drives primary liver cancer[78 ,79 ]. These distinct B-cell phenotypes and mechanisms may account for the paradoxical tumour-promoting role of B cells observed in human studies and murine models.

However, the depletion of B cells using a B-cell depleting antibody, for treatment of renal cell carcinoma, melanoma or colorectal cancer, did not show any clinical benefit[80 ,81 ]. In particular, in an old early phase clinical trial involving patients with advanced colorectal cancer (n= 14 ), a reduction of the tumour size was observed after treatment with Rituximab, a humanised monoclonal antibody directed against human CD20 , and was associated with a reduction of hyper-positive CD21 B cells in peripheral blood[81 ]. Surprisingly, this observation has not been further explored in later phase clinical trials. Nevertheless this observation is substantiated byin vivostudies using syngeneic tumour implantation models[82 ]. The vast majority of these studies, using genetically deficient murine models for B cells, show that B-cell infiltration within the TME produces worse outcomes in mouse models[71 ].In contrast, acute B-cell depletion using anti-CD20 antibody did not recapitulate these findings[83 ]. It is important to note that B-cell deficient mice manifest several secondary immune abnormalities that may contribute to their tumour-suppressive phenotype[83 ].

B cell exhaustion

Akin to T cell exhaustion, recent reports describe a reversible state of B cell dysfunction, different from anergy and senescence, named B-cell exhaustion. Exhausted B cells, identified in viremic HIV patient blood[84 ], and described in older and auto-immune patients[85 ], are phenotypically characterised by low CD21 and CD27 expression, high expression of inhibitory receptors, and deficient effector functions[86 ]. In NSCLC and breast cancer, exhausted B cells, also named tissue-like memory B cells, were found to correlate with T regulatory cells and exhausted PD1 + CD4 +/CD8 + T cells[87 ,88 ].

B CELLS lN PDAC

Similar to other cancers, the role of B cells in pancreatic cancer is controversial, perhaps due to model selection in various studies. B cells are generally associated with an improved outcome in PDAC patients[14],and yet often, their presence correlates with tumour growth and shorter survival in murine models of pancreatic cancer, with various mechanisms offered for this pro-tumorigenic role[18 -21 ]. For example, immunoglobulins produced by splenic B cells may form immune complexes, that can bind TAMs and induce M2 polarisation, consequently suppressing the CD8+T-cell cytotoxic activity; thus,driving tumour progression (Figure 1 ). Both B cells and macrophages were shown to express activated Bruton’s tyrosine kinase (BTK).In vitrothe use of the BTK inhibitor Ibrutinib blocked the M2 polarisation of macrophages that occurred following co-culture with B cells, suggesting that B cells promote the pro-tumorigenic macrophage phenotype, and that BTK signalling is tumour-promoting in both these immune cell types. The use of the BTK inhibitor Ibrutinib in orthotopic pancreatic tumoursin vivoreduced tumour growth[18 ]. A subset of regulatory B cells, called B1 (identified as CD1 dhi , CD5+,10 % of all B cells within the murine tumours), express relatively high levels IL-12 a and Ebi3 transcripts,which encode for IL-35 : An immunosuppressive cytokine (Figure 1 )[19 ]. The injection of IL12 a-/-B cells was unable to restore tumour growth, implying that B-cell-derived IL-35 drives tumour cell proliferation[19 ]. Moreover in the presence of hypoxia, induced by HIF1 α deletion, fibroblasts secrete CXCL13 , which recruits B cells to the tumour site; in particular the B1 regulatory B cells, which promote tumour growth (Figure 1 )[20 ].

However, these immuno-suppressive B cells represent a mere 10 % of the entire B-cell population in PDAC. Therefore, their pro-tumorigenic role might be overcome by the presence of a much larger proportion of pro-inflammatory B cells. The genetic analysis of bulk intra-tumoral B-cell population showed a pro-inflammatory and immuno-stimulatory phenotype in both orthotopic and the KPC(KrasG12 D-Pdx1 -Cre) genetic models of PDAC[22 ]. Indeed, the phenotype of splenic B cells differs from the intra-tumour B-cell phenotype[22 ]. Since Gunderson and colleagues used splenic and not intratumour B cells, in co-culture experiments with bone-marrow-derived macrophages, the immunesuppressive role of B-cells described by them might be irrelevant within the tumour microenvironment[18 ]. Furthermore, in independent experiments, it appears that the regulatory phenotype is not acquired in the tumour microenvironment. CD1 dhi CD5+B cells isolated from a healthy spleen and injected into a μMT mice (genetic depletion of B cells from birth) before orthotopic cancer cell injection rapidly restored tumour cell growth[19 ].

Figure 1 B-cell role in pancreatic cancer. B cells mature in plasma cells, which can produce immunoglobulin G, and are able to reprogram the M1 macrophage phenotype to M2 via Bruton’s tyrosine kinase activation. B regulatory cells are able to produce immune-suppressive cytokines, which inhibit the antitumour immune response, leading to tumour growth. Furthermore, in presence of hypoxia, stromal fibroblasts can secrete chemokine (C-X-C motif) ligand 13 , which recruit B regulatory cells (CD1 dhiCD5 +) and B1 B cells, resulting in faster tumour growth. Clusters of B cells, with follicular dendritic cells and T cells, are sites for T cell priming and B cell maturation and differentiation into antibody-producing cells, with anti-tumoral effect. TLS: Tertiary lymphoid structures; IL-35 : Interleukin-35 ; CXCL-13 : Chemokine (C-X-C motif) ligand 13 .

Interestingly, most studies investigating the role of B cells in cancer immunity were conducted in Bcell-deficient mice, where the absence of B cells restricted tumour growth in a variety of tumour models,generally suggesting that B cells inhibit rather than enhance spontaneous anti-tumour immunity[18 -20 ,82 ,83 ,89 ]. On the other hand, the majority of models using an acute B-cell depletion in an established tumour (for example, achieved by treatment with a B-cell depleting antibody, anti-CD20 ) enhanced tumour growth, suggesting that B cells may have an anti-tumoral role[89 -91 ]. Since, this anti-tumoral aspect is not confirmed if B-cell depletion occurs before initiation of tumour growth, we can speculate that B cells play an initial immunosuppressive/pro-tumoral role; perhaps a role played by circulating or peripheral B cells. However, over the course of tumour development, as B cells infiltrate tumours, they form TLS and acquire a pro-inflammatory phenotype that sustains DC recruitment and activation and antigen presentation, resulting in an anti-tumoral role[14 ,89 ,92 ]. Of note, depletion of B cells earlier in PDAC development in a more relevant pre-clinical model of PDAC, KPC (KrasG12 D-Pdx1 -Cre) mice,did not impact disease progression[22 ]. In contrast, B-cell compartment is competent before and during human PDAC tumourigenesis. Lastly, it is now well understood that B-cell-deficient murine models harbour several immune abnormalities, such as defects in myeloid subsets, which may render those mice tumour-resistant[93 ]. Therefore, acute B-cell depletion in tumour-bearing mice may represent a more reliant model to study the effect of B cells in cancer (Figure 1 ) [14 ,22 ,92 ].

Based on these considerations, B-cell depletion may prevent TLS formation, suggesting that removing B cells in PDAC patients may be detrimental, as the tumours are deprived of sites of DC localisation and anti-tumour immune response[34 ,94 ]. Presence of TLS has been shown to be associated with improved patient survival in PDAC[16 ,17 ,95 ]. The location of TLS (peri-tumoral and intra-tumoral) may be important since those with intra-tumoral TLS had better outcome[16 ]. PDAC tissues with intra-tumoral TLS showed significantly higher infiltration of T and B cells and lower infiltration of immunosuppressive cells, as well as significantly higher expression of Th1 - and Th17 -related genes.

It is possible that the dual behaviour of B cells in non-metastatic PDAC patients is dependent on their spatial organisation[17 ]. Favourable clinical outcome was observed when B cells were organised in TLS,whilst worse patient survival was observed when B cells were scattered at the tumour-stroma edge. The two studies show a different TLS distribution, probably due at the different approaches used for the identification. Hiraokaet al[16 ] demonstrate a near-universal presence of TLS within human PDAC tissue based on H&E staining, whilst Castinoet al[17 ] identify the aggregate pattern only in a subset of patients. This apparent discrepancy, described also in other cancers, can be resolved thorough TLS functional characterisation, such as activation status and composition, through use of key phenotypic markers; thus, rendering them more useful in predicting patients’ outcome[95 -97 ].

In the KPC transgenic murine model, more closely mimicking human cancer, sporadic presence of TLS was observed, but in the orthotopic model of PDAC, lacking the characteristic desmoplastic stroma,TLS were not observed[17 ]. Not only TLS developed spontaneously within the tumour parenchyma of the KPC mice, but also their formation could be enhanced by injection with the immunotherapeutic DNA-vaccine encoding the glycolytic enzyme ENO1 . The vaccination induced a higher number of TLS,PD1 + GC formation and increased antigen-specific T-cells infiltration[17 ]. Furthermore injection of chemokine (C-C motif) ligand 21 (CCL21 ) in a subcutaneous PDAC murine model showed a beneficial effect, by inhibiting tumour growth, decreasing distant metastasis, and recruiting T and DCs within the TME[98 ]. In keeping with these observations, NSCLC patients are receiving intra-tumoral injections of CCL21 -transduced autologous DCs in a phase I clinical trial (NCT00601094 , NCT01574222 )[99 ].

Several studies report the development of TLS after anti-tumour vaccination protocols, including pancreatic cancer[6 ,95 ,99 ]. Lutz et al[95 ] used an irradiated, GVAX given as a single agent or in combination with low-dose cyclophosphamide to deplete regulatory T cells, showing a way to convert a“non-immunogenic” neoplasm such as PDAC, into an “immunogenic” neoplasm, by inducing infiltration of T cells and development of TLS in the TME. The study describes the presence of TLS as defined by a core of B cells and follicular DCs, Ki67 positivity, suggesting the presence of a germinal centre, and CD3 + T cells. Among these, there were CD4 + cells in close vicinity to mature DCs (CD83+and DCLAMP+), and monocyte/macrophages, suggesting that these aggregates exhibited adaptive immunity[95 ]. A better characterisation of the T-cell subsets suggested the presence of negative regulatory signals in the aggregates: most of the aggregates presented FoxP3 + cells and upregulated PD-L1 expression.Thus, the activities of GVAX included both the recruitment of effector T cells into the TME and the upregulation of immunosuppressive regulatory mechanisms, specifically the expression of PD-L1 and T-regs infiltration. But the net impact of the infiltration of both T-effector (T-eff) and Tregs, expressed as ratios of interferon γ-producing Teff/Tregs, were higher in vaccinated patients, suggesting that GVAX can alter the balance of T-eff and T-regs, in favour of an anti-tumour response. The number of TLS resulted increased after combination of GVAX with cyclophosphamide[95 ].

TLS are also known to be site for the formation of antigen-specific B cells and development of memory response and represent an “antibody factory” within non-lymphoid tissues. Intra-tumoral B cells have been shown to produce high-affinity anti-tumour antibodies, mostly IgG, in several human and murine model studies, providing evidence that tumour-specific humoral responses can be generatedin situ, within TLS[26 ,36 ,37 ,52 ]. Such evidence of humoral response is provided by the presence of germinal centres and follicular DC network in PDAC-associated TLS[16 ]. Furthermore there is evidence of antibody production by intra-tumour derived B cells in PDAC[100 ]. IgG production against wild type and mutant KRAS targets (a common occurrence in human PDAC) was assessed to study the antigen specificity of PDAC infiltrating B cells[100 ]. Incubation of tumour infiltrating B-cell supernatant suggests that B-cell responses targeting mutant and not wild-type KRAS are present in the parenchyma of PDAC, yet not detectable in the serum[100 ].

The identification of TAA-directed immunoglobulins in PDAC would be of great use in cancer therapy. For example, IgG1 antibody PAM4 , identified by vaccination of mice with mucin purified from human pancreatic cancer cells, has been applied in radio-immunotherapy and diagnosis[101 ]. In PDAC,serum titre of MUC-1 specific immunoglobulins correlates with improved patients survival[62 ].Examples of anti-MUC1 antibody-based therapeutics developed against pancreatic cancer and that are in clinical trials are huPAM4 , PankoMab-GEX (Gatipotuzumab), AR20 .5 [102 -104 ]. Many other pancreatic cancer specific antigens could serve as valid clinical targets[11 ,105 -108 ].

CONCLUSlON

B cells play a different role in human and murine cancers. In PDAC, high B-cell infiltrate is associated with better prognosis, especially when those B-cells cluster in TLS (Figure 2 ). Yet this is discordant with data obtained using orthotopic models of PDAC, where B-cell depletion suggests an early, pro-tumour function of B cells (Figure 2 ). This apparent paradox can be explained with B cells playing different roles as the tumour progresses and evolves. Firstly, there are differences between intra-tumour and peripheral immune-responses, as demonstrated byin vivostudies in both PDAC[22 ] and other cancers[109 ,110 ]. The more complex cell-cell interactions within the TME may influence B cell phenotype. There are inherent difficulties to recapitulate these features in murine models where the desmoplasia, a characteristic feature of human PDAC may not always be present[111 -113 ]. Furthermore, as with T cells,multiple B-cell subsets have been extensively described in murine models of cancer, but not in human PDAC; and this would be the new frontier of investigation. Despite the current failure of immunotherapy in PDAC, exploring new successful immuno-therapeutic avenues may still be possible. For example, immuno-therapy with immune-checkpoint inhibitors appears effective in the small percentage of PDAC patients harbouring MMR deficiency. Target immunotherapy should be considered for the different PDAC (immune)-subtypes, and should aim to enhance the potential in situ anti-tumour response, which arises within some tumours (TLS+ve patients), with a possibility to revert the immunesuppressive TME. Current immuno-therapeutics under investigation in PDAC in relation to B-cell modulation, include promoting the anti-tumour response [the GVAX vaccine induces thein situformation of active clusters of T and B cells (TLS)] or inducing the immuno-tolerance (Ibrutinib, BTK inhibitor). Combination with other stromal modulating approaches may yield substantial benefits[111 ,114 ,115 ]. An extensive immuno-genetic and immuno-phenotypic profiling of tumour infiltrating B cells may pave the way towards the understanding of integrated tumoral immune system in PDAC and generate crucial new therapeutic insights.

Figure 2 Mechanisms by which B cells regulate tumour growth. Different conclusions drawn from human and mouse studies can be accommodated within this model which takes into account the different intra-tumour and peripheral immune-responses. In vitro and in vivo studies usually measure the functional immune response in secondary lymphoid organs or blood, rather than studying the infiltration and the spatial organisation of different immune cells within the tumour microenvironment. In the pancreas, B cells can form clusters with T cells, named tertiary lymphoid structures, which are sites of antigen presentation, CD78 activation and antibody production. However, in secondary lymphoid organ the presence of B cells during T cell priming can skew the immune response towards Th2 ,attenuating Type 1 response. Furthermore, B-regulatory cells can produce immune-suppressive cytokines, which inhibit the anti-tumour immune-response. Finally, a positive correlation is found between serum immunoglobulin G (IgG) 1 and increased survival. However, repeated isotype switching within IgG subclasses generates in human IgG4 , an isotype that has been linked to regulatory functions, in mouse models IgG2 a, with pro-inflammatory functions. TLS: Tertiary lymphoid structures;IL: Interleukin; TGF-beta: Transforming-growth factor-β.

FOOTNOTES

Author contributions:Delvecchio FR review the literature, Delvecchio FR and Kocher HM drafted the original manuscript; Delvecchio FR composed the original figures; Delvecchio FR, Goulart MR, Fincham REA, Bombardieri M and Kocher HM contributed to the review and editing of the manuscript.

Supported byFrancesca Romana Delvecchio is supported by Cancer Research UK Post-doctoral fellowship; Michelle Goulart is supported by PCRF post-doctoral fellowship; Rachel Elizabeth Ann Fincham is supported by PhD

studentship awarded by Barts Charity (London, UK) and A*STAR (Singapore).

Conflict-of-interest statement:There are no conflicts of interest to declare.

Open-Access:This article is an open-access article that was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution NonCommercial (CC BYNC 4 .0 ) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is noncommercial. See: https://creativecommons.org/Licenses/by-nc/4 .0 /

Country/Territory of origin:United Kingdom

ORClD number:Francesca Romana Delvecchio 0000 -0002 -3661 -1212 ; Michelle R Goulart 0000 -0001 -8333 -3908 ; Rachel Elizabeth Ann Fincham 0000 -0002 -4699 -5002 ; Michele Bombadieri 0000 -0002 -3878 -5216 ; Hemant M Kocher 0000 -0001 -6771 -1905 .

S-Editor:Chang KL

L-Editor:A

P-Editor:Chang KL