All mouse protocols were reviewed and approved by the Institutional Animal Care and Use Committee of the University of North Carolina at Chapel Hill (Chapel Hill, NC, USA). Animals were maintained in a specific pathogen-free facility. Six- to 8-week-old wild-type (WT) C57Bl/6J mice were purchased from The Charles River Laboratories (strain #027). 6- to 8-week-old MD4 (strain #002595), µMT (strain #002288), Prkd2S707A/S711A (strain #017285), CD19^Cre/+ (strain #006785), and Myd88^Fl/Fl (strain #008888) mice on C57Bl/6J background were purchased from Jackson Laboratories. Ebi3^Tom/Tom mice were obtained from D. A. Vignali (University of Pittsburgh, Pittsburgh, PA) (19). Il12a^GFP/GFP mice were generated at the University of North Carolina at Chapel Hill (Chapel Hill, NC, USA) (18). Ebi3^Tom/Tom; Il12a^GFP/GFP mice were generated by breeding Ebi3^Tom/Tom mice with Il12a^GFP/GFP mice until progeny were homozygous for both reporter alleles. CD19^Cre/+; Myd88^Fl/Fl mice were generated by breeding CD19^Cre/+ mice with Myd88^Fl/Fl mice until progeny were homozygous for the floxed Myd88 allele. CD19^+/+; Myd88^Fl/Fl littermates lacking Cre expression were used as controls.

The murine PDAC cell line KPC4662 (p48 ^Cre/+; LSL-Kras ^G12D/+; LSL-Tp53 ^R173/+)was derived from a primary pancreatic tumor of C57Bl/6J KPC mice by Dr. Robert Vonderheide’s laboratory (University of Pennsylvania Perelman School of Medicine, Philadelphia, PA.) (28). Cells were maintained at 37°C and 5% CO2 in DMEM (#11995-065, Gibco) containing 10% Fetal calf serum (Corning) and 1x penicillin– streptomycin (#15140-122, Gibco). Cells were confirmed to be Mycoplasma and endotoxin free. KPC4662 Cells that were injected orthotopically were at <16 passages from original derivation.

For intrapancreatic injection of KPC4662 cancer cells, mice were anesthetized using a ketamine (100 mg/kg; Med-Vet International) and xylazine (10 mg/kg; Med-Vet International) cocktail. The depth of anesthesia was confirmed by verifying an absence of response to toe pinch. An incision in the left flank was made, and 1 x 10⁵ KPC4662 cells in ice-cold PBS mixed at 1:1 dilution with Matrigel (#354234, Corning) in a volume of 50 µL were injected using a 28-gauge needle into the tail of the pancreas. The wound was closed in two layers, with running 5-0 Vicryl RAPIDE sutures (Ethicon) for the body wall, and 5-0 PROLENE sutures (Ethicon) for the skin. All mice were given buprenorphine (0.1 mg/kg; Med-Vet International) subcutaneously after orthotopic surgery for pain relief. Tumors were grown in mice for 21 days unless otherwise noted. Tumor volume was measured using electronic calipers after animal sacrifice and calculated using the formula: Volume = Length * (Width²)/0.52.

For CD4⁺ and CD8⁺ T-cell depletion studies, 200 µg of αCD4 (#BP0003-1, clone GK1.4, Bio X Cell) and 200 µg of αCD8 (#BP0004-1, clone 53-6.7, Bio X Cell) or an IgG isotype control rat IgG2b (#BE0086, clone MPC-11, Bio X Cell) and rat IgG2a (#BE0089, clone 2A3, Bio X Cell), respectively, were administered intraperitoneally daily starting 3 days prior to tumor cell injection and twice a week after tumor cell injection. Mice were sacrificed 21 days after tumor implantation. Depletion of T cells was confirmed by flow cytometry at time of animal sacrifice.

Monitoring of orthotopic pancreatic tumors was performed weekly on the Vevo 2100 Imaging System (VisualSonics Inc.) using the MS550D transducer. Mice were anaesthetized using isoflurane (2%) throughout the procedure. Hair was removed from left flank of each mouse by electric razor and hair removal cream (Nair, Church & Dwight). Images were taken at 11 mm image depth and measurements were calculated using the Vevo LAB software (VisualSonics Inc.).

Single-cell suspensions were prepared from tumors, spleens, and draining lymph nodes isolated from orthotopic models. Spleens and draining lymph nodes were mechanically disrupted using the plunger end of a 5 mL syringe and resuspended in 2% FCS/PBS. Pelleted samples were then depleted of red blood cells using RBC lysis buffer (eBioscience; 00-4333-57) and resuspended in 2% FCS/PBS. For isolation of tumor-infiltrating lymphocytes, tumor tissue was thoroughly minced using sterile razor blades and digested with collagenase IV (1.25 mg/mL; #LS004188, Worthington), 0.1% soybean trypsin inhibitor (#T9128, Sigma), hyaluronidase (1 mg/mL; #LS002592, Worthington), and DNase I (100 mg/mL; #LS002007, Worthington) in complete DMEM for 30 minutes at 37°C. Cell suspensions were passed through a 70µm cell strainer (#352350, Falcon) and resuspended in RPMI 1640 media (Gibco). Leukocytes were isolated from processed tumor tissues by Ficoll-Paque PLUS (#17-1440-03, GE Healthcare) density gradient centrifugation and resuspended in 2% FCS/PBS. For FACS sorted cell populations, cells were stained with fluorophore-labeled antibodies for 30 minutes on ice in FACS buffer (2% FCS/PBS). After staining, cells were washed twice with FACS buffer and resuspended in cell sorting buffer (1% FCS/PBS). Lymphocyte populations were sorted using a BD FACS ARIA III sorter to isolate CD19⁺CD21^HiCD5⁺CD1d^Hi Breg cells, CD19⁺CD21^LoCD5^-CD1d^Lo Bcon cells, CD4⁺ T cells, and CD8⁺ T cells. Cells were collected in complete RPMI media containing 10% FCS with 1x penicillin–streptomycin (#15140-122, Gibco) antibiotics and 1x 2-β-Mercaptoethanol. For assays using total B cells, B cells were isolated from tissues using the EasySep Mouse B cell Isolation Kit (#19854, StemCell Technologies) according to manufacturer’s instructions. For adoptive transfer experiments, 6- to 8-week-old µMT mice were intravenously reconstituted with 1 x 10⁷ purified DMSO or CRT0066101 (PKDi) treated WT splenic B cells 16 hours prior to orthotopic injection. B cells were purified by negative magnetic selection (#19854, StemCell Technologies) and purity was >95% by flow cytometry. Both male and female mice were used in all studies.

Unless otherwise noted, B cells were activated with αCD40 (1 µg/mL, clone HM40-3, BioLegend), αIgM, µ chain (10µg/mL, #115-006-075, Jackson Immuno Research) and/or LPS (2 µg/mL, #L4391-1MG, Sigma) in complete RPMI for 48 hours as previously described (5, 6). For CD8⁺ and CD4⁺ T-cell culture, 1 x 10⁵ T cells were cultured in a 96-well plate with plate bound αCD3 (1 µg/mL, #BE0001-1, Clone 145-2C11, Bio X Cell) and soluble αCD28 (2 µg/mL, #BE0015-1, Clone 37.51, Bio X Cell) in complete RPMI for 48 hours. For detection of cytokines in B cells, 1 x 10⁵ cells were cultured using conditions indicated in figure legends for 48 hours in complete RPMI medium followed by the addition of PMA (50 ng/mL; #P8139, Sigma), Ionomycin (200 ng/mL; #9995S, Cell Signaling Technologies), and Brefeldin A (1x, #420601, BioLegend) for the final 5 hours at 37°C. For detection of cytokines in T cells, 1x 10⁵ cells were cultured with αCD3 and αCD28 for 48 hours with 1x Brefeldin A being added for the final 5 hours of culture. For transwell co-culture assays, a total of 1 x 10⁵ CD19⁺CD21^HiCD5⁺CD1d^Hi B cells and 1 x 10⁵ CD8⁺ T cells or 1 x 10⁵ CD4⁺ T cells were co-cultured in 96-well Transwell plates (#3381; Corning) with B cells occupying the top chamber and CD4⁺ or CD8⁺ T cells in the bottom chamber for 48 hours. For experiments utilizing kinase inhibitors, splenic B cells were cultured for 45 minutes with the indicated inhibitor prior to 48-hour stimulation in vitro. The following inhibitors were used: Selleckchem - ERKi (SCH772984, #S7101), PI3Ki (Idelalisib, #S2226), BTKi (Ibrutinib, #S2680); Cell Signaling Technologies – NFATi (FK-506, #9974S); Sigma – LYNi (Dasatinib, CDS023389-25MG); NF-kBi (Compound A) was a gift from Dr. Albert Baldwin (29). Inhibitor concentration was determined by the lowest amount of inhibitor required to decrease target activation of at least 75% from DMSO treated condition. Inhibition was quantified using ImageStudio software (LI-COR).

Isolated immune cells were washed and blocked with αCD16/CD32 (100ug/100,000 cells, #101320, BioLegend) for 5 minutes on ice and then stained with labeled antibodies against surface markers or cytokines on ice for 30 minutes in FACS buffer (2% FCS/PBS). Intracellular staining for Foxp3 was performed using a Foxp3/Transcription Factor Staining Kit (#00-5523, eBioscience) following surface marker staining and was used according to manufacturer’s instructions. The following monoclonal antibodies directed against mouse antigens were used for flow cytometry: BioLegend - CD45 (clone 30-F11), CD3 (17A2), CD4 (GK1.5), Foxp3 (FJK-16S), CD8 (53-6.7), NK1.1 (PK136), CD11c (N418), CD19 (6D5), CD5 (53-7.3), CD21/35 (7E9), IgD (11–26), IgM (RMM-1), CD23 (B3B4), CD138 (281-2); BD Biosciences - CD1d (clone 1B1), CD93 (AA4.1), CD11b (M1/70). Viability of cells was determined by staining with either Zombie NIR (#423105, BioLegend) or Zombie Red (#423113, BioLegend) viability dye. All samples were acquired on LSRII Fortessa (BD Bioscience) at UNC Flow Cytometry Core Facility and analyzed by FlowJo version 10.2 (TreeStar, Inc.).

Cells were washed and blocked with αCD16/CD32 (100ug/100,000 cells, #101320, BioLegend) for 5 minutes on ice. Cells were then washed and stained with labeled antibodies against surface markers on ice for 30 minutes in FACS buffer (2% FCS/PBS). After surface staining, cells were fixed and permeabilized using the Cytofix/Cytoperm Kit (#554714, BD Biosciences) for 20 minutes at 4°C. Intracellular staining was performed using fluorophore-conjugated cytokine antibodies for 1 hour at 4°C. Cells were then washed twice and resuspended in cell staining buffer (#420201, BioLegend). The following antibodies directed against mouse antigens were used for flow cytometry: BioLegend - IL-10 (clone JES5-16E3), IFNy (clone XMG1.2), TNFα (clone MPX-XT22); eBioscience - p35 (clone 4D10p35); R&D Systems - EBi3 (clone 355022). All samples were acquired on LSRII Fortessa (BD Bioscience) at the UNC Flow Cytometry Core Facility and analyzed by FlowJo version 10.2 (TreeStar, Inc.).

Cells were lysed using RIPA Buffer (#89901, Thermo Fisher) containing HALT phosphatase/protease inhibitors (#78440, Thermo Fisher) on ice and then centrifuged for 10 minutes at 13000rpm for 10 minutes to extract supernatant. Lysates were quantified using Pierce BCA Protein Assay (#23225, Thermo Fisher) according to manufacturer’s instructions. 40µg of total protein was boiled for 5 minutes at 95°C, mixed with 6x laemmli loading buffer, and then loaded and into polyacrylamide gels. Protein was transferred onto LF-PVDF membranes (#162-0264, Bio-Rad) followed by overnight primary antibody incubation. Primary antibody was probed with fluorescent secondary antibodies and detected using the LI-COR Odyssey imager (LI-COR). Primary and secondary antibodies were diluted in 5% BSA/TBS for blocking. The following primary antibodies were used for immunoblotting: Cell Signaling Technologies – pErk1/2 Thr202/Tyr204 (D13.14.4E), Erk1/2 (L34F12), Nfat1 (D43B1), pP65 Ser536 (93H1), P65 (L846), pAkt Ser473 (D9E), Akt (40D4), pSyk Tyr525/526 (C87C1), pSrc Tyr416 (D49G4), Src (L4A1), pBtk Tyr223 (D9T6H), Btk (D6T2C), pPKD/PKCμ Ser744/748 (#2054S), β-Actin (13E5). The following secondary antibodies were used for primary antibody detection: IRDye 800CW Goat anti-Rabbit IgG (#925-32211, LI-COR), IRDye 680RD Goat anti-Mouse IgG (#925-68070, LI-COR).

RNA was prepared from splenic, tumoral, and draining lymph node B cells of CD19^Cre/+; Myd88^Fl/Fl and CD19^+/+; Myd88^Fl/Fl tumor-bearing mice using the RNeasy Micro Kit (#74004, Qiagen). cDNA was generated from 1µg of total RNA using Maxima First-Strand cDNA synthesis RT Kit (#K1672, Thermo Fisher). qPCR analysis was performed using Perfecta SYBR Green FastMix (95072-250, Quantabio) on the QuantStudio 6 instrument (Applied Biosystems). Results were normalized to the expression of Tbp as an internal control, and each sample was run in triplicate. Gene expression was determined by the ΔΔCt method (2–ΔΔCt). The following primers were used for PCR amplification: Tbp FWD 5’-AGAACAATCCAGACTAGCAGCA–3’, Tbp REV 5’-GGGAACTTCACATCACAGCTC-3’, Il10 FWD 5’-GCTCTTACTGATGGCATGAG-3’, Il10 REV 5’-CGCAGCTCTAGGAGCATGTG-3’, Ebi3 FWD 5’- CTTACAGGCTCGGTGTGGC-3’, Ebi3 REV 5’-GTGACATTTAGCATGTAGGGCA-3’, Il12a FWD 5’-CATCGATGAGCTGATGCAGT-3’ Il12a REV 5’-CAGATAGCCCATCACCCTGT-3’, Prkd1 FWD 5’-GGGGGCATCTCGTTCCATC-3’, Prkd1 REV5’- GTGCCGAAAAAGCAGGATCTT, Prkd2 FWD 5’- GGGGTCTCCTTCCATATCCAG-3’, Prkd2 REV 5’-ACGATAGAACAGGCTAGTTGC-3’, Prkd3 FWD 5’- GTCTGTCAAATGTATCTCTGCCA-3’, Prkd3 REV 5’- GGTGAGTATGTGACTCTTCACTG-3’, Il10 FWD 5’-CCCATTCCTCGTCACGATCTC-3’, Il10 REV 5’-TCAGACTGGTTTGGGATAGGTTT-3’, Tgfb1 FWD 5’-CTCCCGTGGCTTCTAGTGC-3’, Tgfb1 REV 5’-GCCTTAGTTTGGACAGGATCTG-3’, Pdcd1 FWD 5’-ACCCTGGTCATTCACTTGGG-3’, Pdcd1 REV 5’-CATTTGCTCCCTCTGACACTG-3’, Pdcd1l1 FWD 5’-GCTCCAAAGGACTTGTACGTG-3’, Pdcd1l1 REV 5’-TGATCTGAAGGGCAGCATTTC-3’, Il12b FWD 5’-TGGTTTGCCATCGTTTTGCTG-3’, Il12b REV 5’-ACAGGTGAGGTTCACTGTTTCT-3’, Il27a FWD 5’-CTGTTGCTGCTACCCTTGCTT-3’, Il27a REV 5’-CACTCCTGGCAATCGAGATTC-3’.

Statistical analysis was performed using GraphPad Prism software and the statistical test used is indicated in the figure legends. P value of < 0.05 was considered statistically significant. The following denotations were used to further clarify statistical significance: ns=p>0.05, *=p<0.05, **=P<0.01, ***=p<0.005, ****=p<0.001. Unpaired student’s t-test was used to compare experiments with exactly two groups. One-way ANOVA was used to compare experiments with more than two groups. Two-way ANOVA was used to compare experiments with more than two groups tested under more than one condition. Sizes of animal or treatment groups are indicated in figure legends.

To understand how the expression of IL-35 subunits is regulated at single cell resolution, we generated a bona fide IL-35 reporter mouse by crossing our Il12a-GFP reporter mouse (18) to a previously generated and validated Ebi3-TdTomato reporter mouse (19). We tested the responsiveness of primary IL-35-reporter B cells to signals previously linked to either IL-10 and/or IL-35 expression in vitro (12, 17, 18). We previously identified the CD19⁺CD1d^HiCD21^HiCD5⁺ subset of B cells (regulatory B cells, Breg) as being the dominant B cell source of IL-35 expression in PDAC (5, 6). We utilized primary splenic CD19⁺CD1d^HiCD21^HiCD5⁺ B cells from IL-35 reporter mice for ex vivo stimulation assays featuring combinations of BCR, TLR4, and CD40 stimulation ( Figure 1A ). Using this model, we determined that BCR stimulation (αIgM) alone is not sufficient to induce IL-35 expression as the reporter signal intensity was not significantly increased after stimulation, due to fluctuation in the frequencies of individual TdTomato⁺ and GFP⁺ populations ( Figure 1B, C ). However, LPS stimulation or CD40 stimulation alone in the absence of a BCR signal was sufficient to expand a population of GFP⁺TdTomato⁺ Breg cells, with most significant changes in GFP MFI signal ( Figures 1B–E ). Combining LPS and CD40 stimulation led to the most significant increase in GFP⁺TdTomato⁺ cell frequency, number, and reporter signal intensity ( Figures 1B–E ). The use of αIgM instead of LPS as the main antigenic signal accompanying CD40 stimulation did generate a significant GFP⁺TdTomato⁺ cell population, but not to the same magnitude of LPS + αCD40 stimulation. Changes in the frequency of GFP⁺TdTomato⁺ reporter cells in the presence of LPS and CD40 activation were primarily driven by increase in TdTomato signal. We also examined the effect of combining BCR and TLR stimulation in the presence or absence of CD40 co-stimulation. Interestingly, combined BCR and TLR stimulation resulted in similar reporter signal intensity to LPS stimulation, but the combination resulted in significantly more double reporter-positive cells (p=0.02) and higher frequency (p=0.05) ( Figures 1B–E ). This suggests that BCR stimulation helps further drive the expansion of Bregs but requires TLR co-stimulation to do so.

Intracellular cytokine staining mirrored reporter findings to show that BCR stimulation alone was not sufficient to induce IL-35 (p35⁺EBi3⁺) expression in the regulatory B cell subset ( Supplementary Figures 1A, B and Figures 1F, G ). We observed a significantly lower percentage of stained p35⁺Ebi3⁺ cells compared to GFP⁺TdTomato⁺ B cells in each condition, but the trends in IL-35⁺ cell frequency was conserved between the two approaches. The addition of αCD40 stimulation to αIgM was able to induce a population of IL-35⁺ Breg cells ( Figures 1F, G ). Furthermore, BCR + LPS stimulation with or without CD40 co-stimulation was able to induce IL-35 expression in regulatory B cells, but not CD19⁺CD1d^LoCD21^LoCD5^- Bcon cells ( Figures 1F, G ). This suggests that ex vivo, BCR stimulation promotes the expansion of double positive regulatory cells only in the presence of either TLR and/or CD40 stimulation, and that inhibition of either the BCR or TLR signaling arm could potentially impact regulatory B cell function.

We also asked whether B cells derived from healthy mice would have the same responses to different stimuli than B cells from a tumor-bearing animal. When exposed to various stimuli ex vivo, splenic B cells derived from tumor-bearing mice generated a somewhat greater frequency of GFP⁺TdTomato⁺ B cells in response to LPS or αIgM + αCD40 compared to tumor-naïve B cells ( Figure 1H and Supplementary Figure 1C ). Examination of GFP and TdTomato signal intensity revealed that the increased frequency is due to GFP signal and not TdTomato ( Figure 1I ). The only instance we observed a significant increase in TdTomato signal intensity due to the influence of PDAC was with LPS + αCD40 stimulation, which did not have an increase in GFP intensity ( Figure 1I ). This is very interesting because Ebi3 expression is more highly regulated of the two subunits, but PDAC is primarily modulating Il12a expression in the spleen. Overall, this does suggest that pancreatic tumors can pre-condition B cells to have augmented IL-35 responses to stimuli.

To understand the role of BCR activation in pancreatic tumor growth in vivo, we utilized the MD4 mouse model (30). B cells derived from MD4 mice have a fixed transgenic BCR that is reactive to hen egg lysozyme, therefore incapable of engaging host or pancreatic tumor-generated antigens. Autoimmune studies in MD4 mice linked disease remittance and inflammatory T cell activity to decreased suppressive abilities of B cells, thus we hypothesized that the inability to sense antigen may affect the suppressive abilities of B cells in PDAC (8). Tumor-naïve MD4 mice did not show any significant alterations in the development of immune cell populations or B cell subsets compared to WT mice ( Supplementary Figures 2A, B ). Syngeneic PDAC cells orthotopically injected into the pancreas of MD4 mice formed significantly smaller tumors than WT controls, indicating that BCR signaling can promote tumor growth ( Figure 2A ). Examination of the B cell infiltrate revealed a significant decrease in the frequency of IL-35⁺ and IL-10⁺ B cells in MD4 mice as compared to WT, indicating that BCR signaling in response to PDAC induces an immunosuppressive B cell phenotype in vivo ( Figure 2B and Supplementary Figure 3A ). The frequency of the Breg cell subset was not altered in MD4 tumors suggesting that the BCR signaling plays a role in regulatory B cell function and not expansion of the regulatory B cell subset in response to PDAC ( Figure 2C ). We recently showed that decrease in IL-35 production by B cells in PDAC results in significantly increased effector T cell responses coupled with decreased tumor growth (6). Consistent with this, we detected an increased frequency of IFNγ⁺CD4⁺ and IFNγ⁺CD8⁺ T cells in MD4 tumors coupled with a decreased frequency of Foxp3⁺ T cells compared to WT tumoral immune infiltrates ( Figure 2D and Supplementary Figure 3B ). Furthermore, we observed increased effector CD4⁺ and CD8⁺ T cell activity in both the spleen and draining lymph nodes of MD4 mice indicating the presence of a systemic response in secondary lymphoid organs as well ( Figures 2E, F and Supplementary Figures 3C, D ). To understand if decreased tumor growth in MD4 mice was dependent on T cell responses, we depleted CD4⁺ or CD8⁺ T cells ( Figure 2G ). Depletion of CD4⁺ T cells from MD4 mice resulted in significantly increased pancreatic tumor growth compared to control MD4 tumors, whereas depletion of CD8⁺ T cells showed a non-significant trend towards increase in tumor growth compared to isotype controls ( Figure 2H ). Collectively these data suggest that BCR activation promotes pancreatic cancer growth largely through suppression of CD4⁺ T cell responses in vivo.

TLR-mediated B cell suppression of T cell responses in models of bacterial infection and autoimmune disease has been linked mainly to expression of IL-10 (10, 31), and is highly dependent on signaling through the TLR adapter protein MyD88. Recently, TLRs have also been discovered to induce IL-35 expression in B cells in the same disease models (12), leading us to question the roles of TLRs on B cells in PDAC progression and T cell suppression. To understand the role of B cell specific TLR activation in pancreatic tumor growth in vivo, we generated CD19^Cre/+; Myd88^Fl/Fl mice. Healthy mice with Myd88 recombination specifically in B cells did not have altered immune cell development or B cell development, congruent with a similar Cd79a^Cre; Myd88^Fl/Fl model ( Supplementary Figures 1A, B ) (32). Orthotopic pancreatic tumors in CD19^Cre/+; Myd88^Fl/Fl mice displayed trends towards decreased growth as compared to CD19^+/+; Myd88^Fl/Fl controls, but the change in growth was less robust than the tumor growth change displayed in MD4 mice ( Figures 3A , 2B ). We observed that IL-10 expression in B cells was significantly downregulated in CD19^Cre/+; Myd88^Fl/Fl mice ( Figure 3B ). However, expression of neither IL-35 subunits was affected, indicating that B cell expression of IL-10 is at least partially dependent on MyD88 in PDAC, but IL-35 could primarily be regulated by another antigenic pathway, most likely the BCR ( Figure 3B ). The loss of MyD88 signaling in B cells resulted in increased CD8⁺ T cell effector responses in the spleen but not in the tumor nor the tumor-draining lymph node upon tumor challenge, suggesting insufficient T cell priming ( Figures 3C–E ). Inversely, CD4⁺ T cell effector responses were significantly increased in the tumor and tumor-draining lymph node of CD19^Cre/+; Myd88^Fl/Fl mice and unaltered in the spleen ( Figures 3C–E ). Collectively these data demonstrate that MyD88 signaling in B cells does not play a prominent role in promoting PDAC growth.

Since BCR signaling was dominant in controlling pancreatic tumor growth in vivo, we examined how key signaling molecules in the BCR signaling pathway could contribute to IL-35 expression. To understand how the BCR could affect IL-35 expression we performed a screen with small molecule inhibitors of the BCR signaling pathway using primary IL-35-reporter B cells isolated from Il12a ^GFP/GFP; Ebi3 ^Tom/Tom spleens ( Figure 4A ). Inhibitor efficacy was assessed by Western blotting ( Supplementary Figure 4A ) and analyzed for reporter fluorescence via flow cytometry. We reasoned that the effector pathways that may most significantly affect IL-35 production may include pathways that promote cross talk between BCR and CD40 signaling, mimicking B cell and T cell interactions in vivo. Protein Kinase D (PKD) is a serine/threonine kinase that in response to BCR stimulation promotes synergy between the BCR and CD40 in a TRAF2-dependent manner (33, 34). PKD is required in B cells for complete proliferation, IL-6, TNFα, and IgM secretion responses from antigen and CD40 co-stimulation (33, 34). We tested the role of PKD in IL-35 expression as well as upstream BCR kinases Bruton’s tyrosine kinase (Btk), Phosphoinositide 3-kinase delta (PI3Kδ), and Lyn. Additionally, we tested key downstream signaling effectors: extracellular signal-regulated kinase 1/2 (ERK1/2), Nuclear factor of activated T-cells (NFAT), and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) in immunosuppressive cytokine production. We chose to investigate the roles of these effectors in the context of not only BCR stimulation, but also TLR (LPS) stimulation as well, since recent work demonstrated that effective TLR signaling activates and requires components of the BCR signaling pathway to fully signal (35).

We observed significant decreases in the frequency of GFP⁺TdTomato⁺ B cells when several kinases were individually inhibited ( Figure 4B ). Additionally, we observed significant changes in the MFI of GFP and TdTomato reporter signal with different inhibitors ( Figures 4C, D ). Inhibition of the BCR signaling phosphatase Lyn did not result in significant alteration of GFP or TdTomato intensity in response to all stimuli and significantly increased activity in most instances where BCR (αIgM) stimulation was used ( Figures 4B–D and Supplementary Figures 6D, 7D ). This is likely due to Lyn’s role in the negative regulation of the BCR signaling pathway and is consistent with previous data showing expansion of immunosuppressive IL-10 producing B cells in Lyn^-/- mice (36). Inhibition of BCR signaling kinases PKD, PI3Kδ and Btk, on the other hand, resulted in the most significant downregulation of GFP⁺TdTomato⁺ cell frequency and TdTomato signal intensity in almost all stimulation conditions ( Figures 4B–D and Supplementary Figures 5A-C, 6A-C ). In conditions containing αIgM stimulation with these kinase inhibitors, PKDi was the most efficient at reducing GFP⁺TdTomato⁺ cell frequency and TdTomato signal intensity ( Figures 4B–D and Supplementary Figures 5A-C, 6A-C ). At the same time, Btk or PI3Kδ inhibition was more toxic than PKD inhibition in αIgM stimulation conditions ( Figure 4E and Supplementary Figures 4B, C ). This is likely related to the wider range of signaling pathways supporting survival that are activated downstream of Btk or PI3Kδ than those downstream of PKD. Overall, these observations suggest that PKD inhibition may be the most promising therapeutic target for IL-35 inhibition.

With regards to the signaling effector pathways downstream of BCR and CD40 signaling, the most prominent effect on reporter signal intensity was seen with inhibition of NF-κB signaling. Inhibition of IKKα/β (NF-κBi) resulted in near complete abolition of both GFP and Tomato signals regardless of stimulation conditions ( Supplementary Figures 5E, 6E, 7E ). This agrees with prior reports stating the requirement of NF-kB signaling in the expression of Ebi3 in B cells (37) and Il12a in antigen presenting cells (38). However, NF-kB inhibition in vitro over 48 hours resulted in high cellular toxicity regardless of the stimulation condition, thus making it an unlikely candidate for therapeutic use. Inhibition of ERK1/2 did not result in significant alterations in GFP or TdTomato reporter signal intensity ( Supplementary Figures 6F, 7F ). ERK1/2 inhibition did however decrease the frequency of GFP⁺TdTomato⁺ B cells leading us to conclude that ERK signaling does not play a role in regulating IL-35 expression, but rather aids in the proliferation of IL-35 expressing cells ( Supplementary Figure 5F ). This is consistent with the effect of MEK inhibition on IL-10⁺ regulatory B cell expansion (39). Inhibition of NFAT signaling, which regulates BCR-dependent IL-10 expression (40), resulted in significant decreases in GFP⁺TdTomato⁺ cell frequency primarily through inhibition of TdTomato reporter intensity ( Supplementary Figure 5G, 6G ). The regulation of Ebi3 transcription by NFAT is entirely plausible as there is an NFAT binding region in the Ebi3 promoter region but was determined to be non-essential for activation in BMDCs (37). We continued to evaluate the role of PKD in B cell signaling by examining the effects on ERK, NFAT, and NF-kB activation in comparison to Btk inhibition, which achieves similar IL-35 reporter inhibition ( Figure 4F ). Inhibition of Btk with Ibrutinib and PKD with CRT0066101 (41) resulted in significant inhibition of ERK, NFAT, and NF-kB signaling to relatively similar extents ( Figure 4G ). Collectively, this data points to IL-35 expression, with inhibition of PKD pathway being the most efficient in shutting down expression of IL35 and maintaining cellular viability.

We focused our efforts on further evaluating the role of PKD in IL-35 expression by B cells. CRT0066101 (PKDi) inhibits all 3 PKD isoforms, but the relative contributions of each isoform on IL-35 expression is unknown. In B lymphocytes, the PKD2 isoform is predominantly expressed (42) whereas PKD3 is expressed at lower levels and PKD1 expression was not detected by RT-PCR analysis ( Figure 5A ). Next, we aimed to understand the relative contributions of PKD2 and PKD3 in B cell signaling. To do this, we used B cells isolated from a mouse model harboring a knock-in of S707A and S711A mutations into the WT Prkd2 locus, which result in a lack of PKD2 catalytic activity (42). Therefore, B cells derived from this model only have active PKD3 signaling. PMA stimulation of homozygous Prkd2-S707A/S711A splenic B cells resulted in an almost complete absence of pan-PKD serine phosphorylation in the catalytic domain leading us to conclude that PKD2 and not PKD3 is the dominant active PKD isoform in primary B cells ( Figure 5B ). PKD signaling is not only active in B cells but is also active in Kras mutant pancreatic cancer cells (43). Prior studies revealed PKD function significantly impacts pancreatic cancer cell growth and metastatic potential (41, 44). RT-PCR analysis revealed that KPC cells express all PKD isoforms, predominantly PKD1, which is consistent with other PDAC cell lines (41, 45, 46) ( Figure 5C ). To understand the role of PKD signaling in KPC4662, we used CRT0066101 and performed MTT proliferation assays in vitro ( Figure 5D ). Consistent with other PDAC lines previously tested (41), KPC cell proliferation is significantly inhibited with loss of PKD function ( Figure 5D ).

In vivo, we observed a significant delay in orthotopic tumor growth of immunocompetent mice treated with CRT0066101 inhibitor ( Figure 5E ). However, this experiment could not discern the potential contribution of PKD inhibition in cancer cells versus immune cells on tumor growth. To parse apart the effects of immune PKD function from cancer-cell autonomous PKD function in PDAC, we orthotopically injected KPC4662 cells into Prkd2-S707A/S711A mice. We observed no significant changes in tumor growth in the absence of host PKD2 function ( Supplementary Figure 8A ) suggesting that cancer cell PKD function is the key driver of PDAC tumor growth. However, we hypothesized that inhibition of PKD function in T cells, may have hindered anti-tumor immunity, resulting in increased tumor growth. PKD contributes to the expression of the effector cytokine IFNγ in response to TCR stimulation (42, 47). We also observed that activated Prkd2-S707A/S711A CD4⁺ and CD8⁺ T cells have significantly reduced IFNγ production ex vivo suggesting that T cell PKD function also plays a significant role in tumor growth ( Supplementary Figures 9A, B ). Regarding B cells, examination of the tumor immune infiltrate in both PKDi and Prkd2-S707A/S711A tumors by flow cytometry revealed that the frequency of infiltrating B cells was significantly decreased ( Figure 5F ) ( Supplementary Figure 8B ). Inversely, we observed a substantial increase in the frequency of CD11b⁺ myeloid cells upon PKD pharmalogical or genetic inhibition ( Figure 5F ) ( Supplementary Figure 8B ). Suppression of PKD in myeloid cells prevents pro-inflammatory cytokine expression in response to TLR9 agonists (48) and bacterial infection (49), but the effects in response to cancer are not known. CD4⁺ and CD8⁺ T cell frequencies were not decreased with PKDi, but CD19⁺ B cell frequencies and overall cell number were significantly reduced ( Figure 5G ). We further examined if the decreased tumoral B cell infiltrate was attributable to a specific B cell subset and found that PKD inhibition primarily decreased the frequency of follicular (CD19⁺CD93^-CD21^LoIgD^Hi) B cells within the tumor ( Supplementary Figure 10A ). Collectively, this data suggests that the decrease in tumor growth is due to inhibition of cancer-cell autonomous PKD, and the role of B cell and/or T cell PKD signaling in PDAC tumor immunity needs to be further understood in vivo.

To elucidate the effects of PKD2 function on IL-35 specifically in the regulatory B cell subset, we stimulated WT and Prkd2^S707A/S711A CD19⁺CD1d^HiCD21^HiCD5⁺ Bregs in vitro and analyzed p35 and EBi3 expression by flow cytometry ( Figure 6A ). Deficiency in PKD2 function led to a significant decrease in the frequency of p35⁺EBi3⁺ (IL-35⁺) Breg cells, confirming that PKD2 is a critical regulator of IL-35 expression in regulatory B cells ( Figure 6B ). Furthermore, we tested how deficiency in PKD2 function affected the expression of other IL-12 family cytokines, IL-12 and IL-27. Il27a and Il12b are not widely expressed by regulatory B cells (5). PKD2 functional deficiency does not enhance expression of Il12b under varying stimulation conditions, but it does significantly enhance Il27a under BCR and CD40 stimulation suggesting that PKD2 represses IL-27 production by regulatory B cells in certain conditions ( Supplementary Figure 11A ). We also tested the expression of other immunosuppressive cytokines, Il10 and Tgfb1 in the context of PKD2 functional deficiency. While Tgfb1 expression is not altered with PKD2 function, Il10 expression was significantly enhanced in certain stimulation conditions indicating a potential role for PKD2 in regulating IL-10 as well ( Supplementary Figure 11B ). To test the functional ability of Prkd2^S707A/S711A Bregs to suppress T cell function we established in vitro transwell co-culture assays, containing activated WT and Prkd2-S707A/S711A CD19⁺CD1d^HiCD21^HiCD5⁺ Bregs and activated WT CD4⁺ and CD8⁺ T cells ( Figure 6C ). Prkd2^S707A/S711A regulatory B cells were significantly less effective at suppressing effector T cell cytokine expression in both CD4⁺ and CD8⁺ T cells ( Figure 6C ).

To understand the role of B cell specific PKD2 function on tumor growth we reconstituted mice lacking mature B cells (µMT) with splenic B cells treated ex vivo with either CRT0066101 (PKDi) or DMSO followed by orthotopic injection of KPC cells. Mice reconstituted with PKDi-treated B cells had significantly smaller tumors than mice reconstituted with DMSO-treated B cells ( Figure 6D ). Examination of the tumor infiltrating lymphocytes revealed that the number of B cells in tumors was not altered by PKD inhibition ( Figure 6E ). However, PKD inhibition in B cells led to a significant increase in intratumoral CD4⁺ and CD8⁺ T cells as compared to DMSO controls ( Figure 6F ). Collectively this data suggests that PKD2 plays an essential role in not only IL-35 expression by B cells, but also effector T cell function and ultimately regulates balance of tolerogenic and anti-tumor immunity in pancreatic tumor growth.