Tumor Cell–Derived IL1β Promotes Desmoplasia and Immune Suppression in Pancreatic Cancer

Pancreatic ductal adenocarcinoma (PDA) is a highly lethal malignancy with a mortality rate approaching the rate of incidence (1). In addition to lack of efficient early diagnosis methods, disease survival is compromised by resistance to conventional chemotherapy and immunotherapeutic strategies that are proving effective in the treatment of other cancers (2, 3). It is becoming increasingly recognized that this recalcitrance is largely attributable to an elaborate network of tumor–stromal interactions that are orchestrated by paracrine factors released by the tumor epithelium, activated fibroblasts, and immune cells (4, 5). Identification and functional characterization of such factors, and the processes they control, is therefore an essential prerequisite for rational development of strategies that can circumvent therapeutic barriers and improve immune responsiveness of PDA tumors.

The cytokine IL1β is an inflammatory mediator that is frequently upregulated in a variety of cancers and its production is associated with poor prognosis (6, 7). Upregulation of either IL1B expression or posttranslational processing in head and neck squamous carcinoma, breast cancer, lung cancer, and melanoma results in increased tumor infiltration of immunosuppressive macrophages and myeloid-derived suppressor cells (MDSC), thereby promoting immune evasion and tumor development (8–10). Other protumorigenic effects of IL1β have been attributed to the induction of neoangiogenesis (11) and the regulation of expression in stromal cells of soluble mediators that enhance tumor cell survival and metastasis (7). These effects are mediated by IL1β-dependent signaling cascades that under conditions of IL1β overabundance result in the sustained activation of NFκB and MAPK pathways (6).

Several lines of evidence suggest a role for IL1β in pancreatic cancer development and progression. Increased pancreatic levels of IL1β are observed in association with pancreatitis, a well-established PDA risk factor (12). High intratumoral and serum IL1β levels in patients with pancreatic cancer correlate with poor overall survival and increased chemoresistance (13–15). In mouse models of PDA, adipocyte-secreted IL1β is found to promote obesity-induced pancreatic carcinogenesis and drug resistance through recruitment of tumor-associated neutrophils (16). In addition, regulatory pathways that control IL1β production in PDA-associated myeloid cells have been reported to support tumor progression by promoting immune tolerance (17, 18). Overall, several lines of evidence suggest a heterotypic distribution of IL1β expression in PDA with implications in disease pathogenesis. Thus, in this study, we sought to elucidate the mechanisms underlying the regulation and function of IL1β in PDA, with an eye on assessing its potential as a therapeutic target.

Here, we identify the tumor cell compartment as a prominent source of IL1β production in human and mouse PDA through activation of the TLR4–NLRP3 inflammasome signaling pathway. Targeted depletion of IL1β in established mouse models demonstrates acute dependency of pancreatic cancer evolution on tumor cell–derived IL1β through protumorigenic modulation of the stroma and immune microenvironment. Overall, our study identifies IL1β as an attractive target that may improve PDA response to therapeutic strategies, including immunotherapy.

All mouse protocols were reviewed and approved by the Institutional Animal Care and Use Committee of the New York University (NYU) Grossman School of Medicine (New York, NY). The LSL-Kras^G12D/+, LSL-Trp53^R172H/+, and p48^Cre/+ mice strains have been described previously (19, 20). Eight- to 10-week-old wild-type (WT) C57BL/6 (stock 027) mice were purchased from The Charles River Laboratories. Eight- to 10-week-old IL1r1^−/− (stock # 003245) mice were purchased from The Jackson Laboratories. All mice were on a C57BL/6 genetic background. Female mice were used for orthotopic injections of KRas^G12D-PDEC and KPC cells (21, 22). Briefly, mice were anesthetized using a ketamine (100 mg/kg)/Xylazine (10 mg/kg) cocktail administered via intraperitoneal injection. After making a small incision on the left abdominal wall, either 10⁶ KRas^G12D-PDEC or 5 × 10⁴ KPC cells in ice-cold PBS mixed at 1:1 dilution with Matrigel (#354234, Corning) in a volume of 50 μL were injected into the tail of the pancreas using a 28-gauge needle. For pancreatic stellate cell (PSC) coimplantation experiments, 10⁶ KRas^G12D-PDEC cells and 10⁵ PSCs were mixed in 50 μL of ice-cold PBS: Matrigel (1:1) and injected into the tail of the pancreas using a 28-gauge needle. The incision was closed using 5-0 Vicryl RAPIDE sutures (Ethicon) for the body-wall and 4-0 PROLENE sutures (Ethicon) for the skin. All animals were given buprenorphine (0.1 mg/kg) for pain relief directly after surgery and once a day for three days postsurgery. Mice were euthanized by carbon dioxide–induced narcosis 2 weeks postimplantation of KRas^G12D-PDEC. For KPC cells, mice were euthanized 2 weeks and 4 weeks postimplantation for flow cytometry analysis and tumor volume assessment, respectively. KPC tumors were measured using digital caliper (VWR) at the endpoint and tumor volume was calculated using the formula πLW²/6.

Mice were orthotopically injected into the pancreas with 5 × 10⁴ KPC cells. On day 7 postinjection, mice were intraperitoneally administered either 10 mg/kg anti-PD-1 (Novartis), 10 mg/kg anti-IL1β (Novartis), or IgG control (Novartis) antibody diluted in 200 μL of sterile PBS. Thereafter, anti-PD-1 antibody was administered on days 9, 11, and 16 and anti-IL1β was administered every 2 days. For CD8⁺ T-cell depletion, 200 μg of anti-CD8 (BioXCell, clone 53-6.7) or an IgG isotype control antibody (BioXCell, clone 2A3) diluted in 200 μL of sterile PBS were administered intraperitoneally daily starting 3 days prior to tumor cell injection and every 5 days after tumor cell injection. Efficiency of CD8⁺ T-cell depletion was assessed by flow cytometry.

Mouse gut microbiota depletion was performed as described previously (23). Briefly, 6-week-old WT mice were administered 0.2 mL of 0.1 mg/mL amphotericin-B (Sigma) by oral gavage every 12 hours for 3 days. Subsequently, water flasks were supplemented with 1g/L ampicillin (Fisher Bioreagents) and antibiotic cocktail containing 0.2 mL of 5 mg/mL vancomycin (Cayman Chemical Company), 10 mg/mL neomycin (Sigma), and 10 mg/mL metronidazole (Sigma) was administered by oral gavage once a day for 2 weeks presurgery and then for the duration of the experiment. Fresh antibiotic cocktail was mixed every day and ampicillin and water was renewed every seventh day. To assay for microbial depletion, fecal pellets were collected from mice at day 0 and day 14 of antibiotic treatment. DNA was isolated from fecal pellets with QIAamp DNA Stool Mini Kit (Qiagen) as per manufacturer's instructions. Bacterial 16S DNA gene quantification was assessed by quantitative PCR as described previously (24).

Isolation, culture, and adenoviral infection of PDEC were carried out as described previously (21). The KPC cell line (line 4662) was a kind gift from Dr. R.H. Vonderheide. The immortalized PSC cell line was a kind gift from Dr. A.C. Kimmelman. Isolation and culture of primary PSCs was carried out as described previously (25). Cell lines were not authenticated and were tested for Mycoplasma contamination every 4 months. Scramble control and shRNAs against Il1β, Nlrp3, and Tlr4 were cloned into the lentiviral pLKO.1 hygro vector obtained from Addgene (#24150). shRNA sequences used were as follows—scramble: GCGACATCCTCATCTCGTTAGTA; IL1β-sh1: GTGGTCAGGACATAATTGACTTC, IL1β-sh2: GCAGCACATCAACAAGAGCTTCA; NLRP3-sh1: AGCCTGAGCTGACTATAGTCTTC, NLRP3-sh2: CTTGAAGATGTGGACCTCAAGAA; TLR4-sh1: GCCAATCCTAAGAATGCTATA. Lentiviral particles were generated by transfecting HEK-293T cells with the pLKO.1 vector, the packaging construct (psPAX2), and the envelope plasmid (pMD2G). Supernatants containing viral particles were collected over a period of 48 hours and stored at 4°C. Following final collection, supernatants were filtered through a 0.45-μm syringe filter and concentrated using 100 MWCO Amicon Ultra centrifugal filters (Millipore). A multiplicity of infection of 15 was used for lentiviral infection of KRas^G12D-PDEC or KPC cells in the presence of 10 μg/mL polybrene (Chemicon) and infected cells were selected using 150 μg/mL hygromycin (Sigma).

130 human PDA tumor (n = 75) and adjacent normal (Adj Norm; n = 55) mRNA expression profiles generated on the same array (Affymetrix GeneChip Human Genome U133 Plus 2.0) were downloaded from GEO (https://www.ncbi.nlm.nih.gov/geo/; GSE15471, GSE16515). Adj Norm samples clustering with PDA tumors, PDA tumor profiles clustering with subsets of Adj Norm samples, and duplicates were discarded (as described previously; ref. 26), for a remainder of n = 74 tissues (n = 50 PDA tumor and n = 24 Adj Norm). Raw data were processed and normalized in one batch using a GC-content background correction robust multi-array average (RMA) algorithm (GC-RMA), performed in R: A language and environment for statistical computing. Unpaired Student t tests were generated in GraphPad Prism (GraphPad Software; www.graphpad.com).

For the purposes of analyzing IL1B, NLRP3, and TLR4 expression patterns, we examined 8–10 patient PDA lesions and corresponding adjacent normal tissue samples. Samples consisted of 5-μm sections that were cut from formalin-fixed, paraffin embedded blocks provided by the Center for Biospecimen Research and Development at NYU Langone Health. This study was conducted in accordance with the Declaration of Helsinki; all samples were anonymized prior to being transferred to the investigator's laboratory and therefore met exempt human subject research criteria.

Mouse pancreata were fixed in 10% buffered formalin (Thermo Fisher Scientific) overnight and embedded in paraffin as described earlier (21). Trichrome staining was performed at NYU Grossman School of Medicine Histopathology Core Facility. For IHC, deparaffinized sections (6 μm) were rehydrated, quenched in 2% hydrogen peroxide/methanol for 15 minutes, and antigen retrieval was performed in 10 mmol/L sodium citrate/0.05% Tween-20 (pH 6.0) for 15 minutes in a microwave oven. Blocking was done in 10% serum/1% BSA/0.5% Tween-20 for 1 hour at room temperature, followed by incubation with the primary antibodies diluted in 2% BSA overnight at 4°C. Primary antibodies are detailed in Supplementary Data. After incubating with secondary biotinylated antibodies and ABC solution (both from Vector Laboratories), sections were developed with DAB peroxidase substrate kit (Vector Laboratories) according to the manufacturer's instructions. After counterstaining with Harris hematoxylin (Sigma), slides were subjected to an alcohol dehydration series and mounted with Permount (Thermo Fisher Scientific). Slides were examined on a Nikon Eclipse 80i microscope and images were analyzed to measure stained area using ImageJ software.

Formalin-fixed, paraffin-embedded sections were deparaffinized and rehydrated, permeabilized with TBS/0.1% Tween-20, and washed in PBS. Citrate buffer antigen retrieval (10 mmol/L sodium citrate/0.05% Tween-20, pH 6.0) was performed in a microwave for 15 minutes. Blocking was performed in 10% serum/1% BSA/0.5% Tween-20/PBS for 1 hour at room temperature. Primary antibodies were diluted in 2% BSA/0.5% Tween-20/PBS and incubated on sections overnight at 4°C. Primary antibodies are detailed in Supplementary Data. Secondary antibodies (Alexa Fluor–labeled; 1:1,000, Invitrogen) were diluted in 2% BSA/PBS, and incubated on sections for 1 hour at room temperature. Sections were washed with PBS and stained with DAPI. Slides were examined and imaged on a Nikon Eclipse Ti2 microscope.

Single-cell suspensions were prepared from pancreas as described previously (27). For isolation of tumor-infiltrating lymphocytes, tumor tissue was minced into 1 to 2 mm pieces and digested with collagenase IV (1.25 mg/mL, Worthington) and 0.1% trypsin inhibitor from soybean (Sigma), in complete RPMI for 25 minutes at 37°C. For isolation and FACS analysis of epithelial and fibroblast compartments, minced tumor tissue was digested with Pronase (0.2 mg/mL, Roche), Collagenase P (0.5 mg/mL, Roche), and DNase I (0.5 mg/mL, Roche). Cells were suspended in 1%FBS/PBS, passed through a 70-μm strainer and treated with RBC lysis buffer (eBiosciences). Single-cell suspensions were blocked with anti-CD16/CD32 antibody (Fc Block, BD Biosciences) for 5 minutes on ice and labeled with mAbs against mouse antigens as detailed in Supplementary Data. All samples were acquired on LSR II (BD Biosciences) at NYU Flow Cytometry Core Facility and analyzed by FlowJo version 10.2 (TreeStar, Inc.). Cell sorting using a BD FACS ARIA II sorter was performed to isolate Ep-CAM⁺ cells, CD140a⁺ fibroblasts and CD45⁺ cells, and >95% purity of sorted cells was achieved.

For RNA isolation from tumors, pancreata processed to single-cell suspension were stained for flow cytometry. CD45⁻CD34⁻CD140a⁺Ep-CAM⁻ fibroblasts were FACS sorted using a 100-μm nozzle into the lysing reagent RLT and total RNA was extracted as per the manufacturer's instructions (RNeasy Mini Kit, Qiagen). To check knockdown in KRas^G12D-PDEC and KPC cells, 10⁵ cells were lysed in 350 μL RLT reagent and total RNA was extracted as per the manufacturer's instructions (RNeasy Mini Kit, Qiagen). Total RNA (1 μg) was reverse-transcribed using the Quantitect Reverse Transcription Kit (Qiagen). Subsequently, specific transcripts were amplified by SYBR Green PCR Master Mix (USB) using a Stratagene Mx 3005P thermocycler. Where fold expression is specified, comparative C_t method was used to quantify gene expression. Expression was normalized to GAPDH. Primers used for qPCR are detailed in Supplementary Data.

For cytokine analysis of mouse pancreata, the tissues were harvested, minced with a sterile razor blade, and incubated in 500 μL of complete media for 24 hours before supernatant collection. Mouse IL1β protein levels were determined by Mouse IL1β Quantikine ELISA Kit (R&D Systems) as per manufacturer's instructions.

At least 7 to 15 mice were used in each group, and the experiments were repeated a minimum of two times to validate reproducibility. Group means were compared with Student t tests. Significance in variations between two groups was determined by an unpaired Student t test (two-tailed). Statistical analyses were performed using GraphPad Prism software (version 7.0d), and data are presented as mean ± SD. P < 0.05 was considered statistically significant.

To assess IL1β production in PDA, we first examined microarray data from 50 PDA patient tumors and 24 adjacent normal tissue samples. Our analysis revealed significant upregulation of IL1β expression in PDA tumors relative to normal adjacent pancreatic tissue (Fig. 1A). We next assessed the distribution pattern of IL1β by IHC staining of tumor sections from patient PDA samples. Consistent with previous reports documenting the expression of IL1β by innate immune cells and fibroblasts (15, 16, 18), robust IL1β staining was detected in the tumor stroma (Supplementary Fig. S1A). However, unexpectedly, we also observed significant IL1β staining in the ductal epithelium (Fig. 1B). Similarly, robust expression of IL1β in the CK8⁺ epithelial tumor cell compartment was revealed by immunofluorescence staining of pancreata from the slowly progressive Kras^G12D; p48^Cre (KC) mouse model of pancreatic neoplasia (19) and the Kras^G12D;p53^R172H;p48^Cre (KPC) invasive PDA mouse model (Fig. 1C and D; Supplementary Fig. S1B; ref. 20). The relative levels of IL1β production by stromal (CD140a⁺ fibroblasts and CD45⁺ immune cells) and tumor cells (EpCAM⁺ epithelium) from KC mice was further analyzed by flow cytometry (Fig. 1E; Supplementary Fig. S1C). In agreement with the IHC data, the epithelial compartment displayed the highest levels of IL1β production (Fig. 1E).

Next we sought to investigate the functional significance of tumor cell–derived IL1β. Utilizing a RNAi strategy, two independent short hairpin (sh) sequences targeting IL1B were introduced into pancreatic ductal epithelial cells derived from either Kras^G12D mice (KRas^G12D-PDEC) or Kras^G12D;Trp53^R172H;p48^Cre mice (KPC; refs. 21, 22). Knockdown efficiency was ascertained by qPCR and immunofluorescence staining (Supplementary Fig. S1D–S1F). Orthotopic injection of IL1β-sh KRas^G12D-PDEC into pancreata of syngeneic mice led to grafts that displayed a significant reduction of CK8⁺ pancreatic intraepithelial neoplasia (PanIN)-like lesions, relative to scramble control (Fig. 1F; Supplementary Fig. S1G). A role for tumor cell–derived IL1β was also evident in the context of more advanced lesions, with IL1β-depleted KPC cells forming significantly smaller tumors upon orthotopic implantation (Fig. 1G and H). Furthermore, IL1β knockdown increased the survival of mice bearing orthotopic KPC tumors (Fig. 1I). Overall, these results establish a prooncogenic role for tumor cell–derived IL1β in pancreatic cancer.

Given the well-established role of IL1β as an inflammatory mediator, we tested whether tumor cell–derived IL1β promotes pancreatic oncogenesis through its interactions with the tumor microenvironment (TME) by implanting KRas^G12D-PDEC or KPC cells into the pancreas of Il1β receptor (Il1r1) null mice (28). Absence of IL1β signaling in the pancreatic stroma phenocopied the depletion of tumor cell–derived IL1β, with reduced growth of orthotopic KRas^G12D-PDEC grafts and KPC tumors in IL1r1-null mice relative to wild-type control (Supplementary Fig. S1H–S1K). In addition, the overall survival of KPC cell–implanted animals was extended in IL1r1-null mice, relative to wild-type control (Supplementary Fig. S1L). Notably, surface expression of IL1R1 is nearly undetectable in Kras^G12D-PDEC cells (Supplementary Fig. S1M). These observations suggest a paracrine role for tumor cell–derived IL1β and therefore prompted us to investigate the fibro-inflammatory effects of tumor cell–derived IL1β on the TME.

Flow cytometric analysis of pancreatic grafts formed by IL1β-sh KRas^G12D-PDEC revealed a pronounced alteration of the TME immune landscape, relative to the scramble control. Specifically, depletion of tumor cell–derived IL1β significantly decreased stromal accumulation of CD11b⁺F4/80⁺ tumor-associated macrophages (TAM; Fig. 2A), CD11b⁺Gr1⁺ myeloid-derived suppressor cells (MDSC; Fig. 2B), CD11b⁺Ly6G⁺ tumor-associated neutrophils (TAN; Fig. 2C), CD1d^hiCD5⁺ regulatory B cells (B_reg; Fig. 2D), and CD4⁺RORγt⁺ Th17 cells (Fig. 2E). In addition, knockdown of tumor cell–derived IL1β also decreased the CD206⁺ M2-polarized state of stromal TAMs (Fig. 2F). No significant changes were observed in stromal recruitment of CD4⁺FoxP3⁺ regulatory T cells, CD4⁺ Th cells, and CD19⁺ B cells, upon IL1β knockdown (Supplementary Fig. S2A–S2C).

The immune cell profile resulting from the suppression of IL1β production is consistent with a role for tumor cell–derived IL1β in constraining antitumor immunity in PDA. For instance, TANs have been reported to downregulate T-cell infiltration in the PDA stroma (29). In addition, M2 TAMs, MDSCs, and CD1d^hiCD5⁺ B_reg are immunosuppressive cell populations that have been shown to inhibit the tumor lytic activity of CD8⁺ T_c cells (21, 27, 30). Indeed, loss of stromal immunosuppressive subpopulations in IL1β-depleted KRas^G12D-PDEC grafts was accompanied by a significant increase in tumor infiltration (Fig. 2G) and activation of CD8⁺ T_c cells, as measured by IFNγ and granzyme B expression (Fig. 2H and I). Similar immune changes were also observed in tumors formed by orthotopic transplantation of IL1β-sh KPC cells (Supplementary Fig. S2D–S2L), indicating a role for IL1β in shaping the immune microenvironment in advanced disease as well.

Together, these observations implicate tumor cell–derived IL1β in promoting the establishment of an immunosuppressive microenvironment. Notably, we observed a reduction in IL1β-expressing CD45⁺ immune cells present in IL1β-sh KRas^G12D-PDEC pancreatic grafts relative to the scramble control (Supplementary Fig. S2M), suggesting a feedforward mechanism wherein the tumor-derived IL1β could dictate the abundance of stromal-derived IL1β.

In addition to the immune changes described above, we also observed a significant decrease in stromal fibrosis, as detected by collagen deposition, in IL1β-sh KRas^G12D-PDEC pancreatic grafts (Supplementary Fig. S3A). Because the most prominent source of collagen deposition in the extracellular matrix are cancer-associated fibroblasts (CAF) that are generated by the activation of PSCs, we assessed the state of PSC activation in IL1β-sh pancreatic grafts by assaying for the activation marker αSMA (31). While prominent αSMA staining was observed in the microenvironment of scramble control KRas^G12D-PDEC grafts and KPC tumors, the IL1β-sh KRas^G12D-PDEC grafts and IL1β-sh KPC tumors were largely devoid of αSMA staining (Fig. 3A and B). In contrast, no change in the abundance of vimentin-positive fibroblasts was detected in the IL1β-sh KRas^G12D-PDEC grafts (Fig. 3A), indicating that tumor cell–derived IL1β is required for PSC activation and not viability. The potential relevance of this activation mechanism to our observations is supported by the findings that these cells display surface expression of IL1β receptor as determined by FACS analysis (Supplementary Fig. S3B).

To distinguish between direct and indirect effect of tumor-derived IL1β on PSC activation, we isolated primary PSCs from wild-type or Il1r1-null mice and coimplanted each with KRas^G12D-PDECs in Il1r1-null mice. Under these conditions, all components of the host stroma are Il1r1 null and therefore by definition unresponsive to tumor cell–derived IL1β. We found that coimplanted wild-type PSCs successfully underwent activation, as detected by αSMA staining (Supplementary Fig. S3C). These results implicate tumor cell–derived IL1β as the primary driver of PSC activation. Consistent with this interpretation, the Il1r1-null PSCs derived from pancreata of Il1r1-null mice failed to undergo activation when coimplanted with tumor cells (Supplementary Fig. S3C). The functional significance of the tumor-derived IL1β/PSCs axis for pancreatic tumor growth is indicated by the observation that the growth defect of KRas^G12D-PDEC pancreatic grafts in Il1r1-null mice could be rescued by coimplanted wild-type PSCs but not Il1R1-null PSCs (Supplementary Fig. S3D).

The protumorigenic effects of CAFs is well documented and is mediated by multiple paracrine mechanisms (31, 32). In addition to secreting extracellular matrix proteins and growth factors, the transition of PSCs from a quiescent to an activated state has been shown to be accompanied by the induction of an inflammatory program with upregulation of cytokines and chemokines such as IL6, CCL2, CCL5, and CCL8 (32). To investigate the effect of tumor cell–derived IL1β depletion on the inflammatory secretome of CAFs, we sorted CAFs from scramble control and IL1β-sh KRas^G12D-PDEC pancreatic grafts using the fibroblast marker CD140 (Supplementary Fig. S3E) and analyzed them for expression of previously characterized cytokines (33). Loss of tumor cell–derived IL1β significantly downregulated the expression of several inflammatory cytokines, relative to scramble control (Fig. 3C). Moreover, flow cytometric analysis of CD140a⁺ CAFs derived from IL1β-sh KRas^G12D-PDEC grafts showed a significant decrease in Ly6C expression, a surface glycoprotein that marks cytokine-producing CAFs (Supplementary Fig. S3F; ref. 34). Our results thus indicate that tumor cell–derived IL1β promotes the activation and shapes the secretory phenotype of CAFs.

Many of the cytokines produced by CAFs in response to IL1β are known modulators of immune cell function (35). We therefore postulated that tumor cell–derived IL1β-mediated PSC activation may, in turn, contribute to the establishment of an immune-suppressive TME. To test this hypothesis, we coimplanted immortalized CAFs isolated from KPC tumors (36) with IL1β-sh KRas^G12D-PDEC in wild-type mice pancreata. Successful coimplantation was verified by restoration of stromal αSMA staining (Fig. 3D), increase in Ly6C⁺ CD140a⁺ cell population (Supplementary Fig. S3F) and increased collagen deposition (Supplementary Fig. S3G). Significantly, CAF coimplantation with IL1β-sh KRas^G12D-PDEC specifically rescued the decrease in macrophage recruitment and M2 TAM polarization induced by loss of tumor cell–derived IL1β (Fig. 3E and F; Supplementary Fig. S3H–S3K) as well as restored the inactive state of CD8⁺ Tc cells and the decrease in CD8⁺ Tc cell infiltration (Fig. 3G and H). This phenotype is consistent with the observed IL1β-dependent production by CAFs of CCL2 and CCL5 (Fig. 3C), which promote macrophage infiltration and M2 polarization (37, 38), and CXCL12 (Fig. 3C), which is known to impede tumor infiltration of CD8⁺ T_c cells (39). To determine the functional significance of CD8⁺ T-cell exclusion in the protumorigenic role of tumor cell–derived IL1β, we depleted CD8⁺ T cells in mice prior to orthotopic implantation of scramble or IL1β-sh KRas^G12D-PDECs (Supplementary Fig. S4A). CD8⁺ T-cell depletion completely rescued tumor growth defect of IL1β-sh KRas^G12D-PDEC pancreatic grafts (Fig. 4A and B; Supplementary Fig. S4B), indicating that the oncogenic role of tumor cell–derived IL1β is mediated through immune suppression of CD8⁺ T-cell infiltration and activity.

The poor response of pancreatic tumors to immune checkpoint blockade has been primarily attributed to its immunosuppressive microenvironment and poor CD8⁺ T-cell infiltration (3). Because depletion of tumor-derived IL1β significantly increases CD8⁺ T-cell infiltration and activity, we reasoned that IL1β neutralization may sensitize PDA tumors to PD-1 checkpoint blockade. To this end, orthotopic KPC tumor–bearing mice were treated with neutralizing antibodies against IL1β and PD-1 (Fig. 4C). Indeed, addition of α-IL1β treatment significantly enhanced the antitumor activity of α-PD-1 (Fig. 4D; Supplementary Fig. S4C). As predicted, combined treatment of α-IL1β and α-PD-1 resulted in increased tumor infiltration of CD8⁺ T cells, relative to vehicle control or α-PD-1 alone (Fig. 4E).

Having established its importance in pancreatic tumorigenesis, we next wanted to dissect the molecular pathway regulating IL1β production in tumor cells. In innate immune cells, IL1β mRNA is translated to produce an inactive precursor pro-IL1β form, which is further processed to yield the mature secreted form of the cytokine by a multimeric protein complex called the inflammasome (6). The most well-characterized inflammasomes are comprised of a Nod-like receptor protein family pyrin-domain containing (NLRP) protein that serves as an activation sensor, which associates with apoptosis-associated speck-like proteins containing a CARD complex (ASC) protein (40). In complex with NLRP, ASC recruits procaspase-1 that autocatalyzes its cleavage to active caspase-1. Active caspase-1, in turn, cleaves pro-IL1β to produce the functional IL1β protein. Of the various NLRP proteins that can form inflammasomes, the NLRP3 inflammasome appears most relevant to our study because its activation was found to be necessary for induction of pancreatitis (41), a major risk factor for PDA development. In addition, NLRP3 inflammasome activity is associated with malignancies such as colon cancer and melanoma (42). We therefore analyzed the activation status of the NLRP3 inflammasome axis in the pancreatic tumor epithelium. A robust presence of NLRP3 was detected in the tumor epithelial compartment of KC mice pancreata (Fig. 5A). NLRP3 expression in these tumor cells strongly correlated with the expression of cleaved caspase-1 (Fig. 5B), a product of active inflammasomes. Moreover, NLRP3 was found to colocalize with phospho-ASC (Y223) in speck-like aggregates (43) in tumor cells of both KC and KPC mouse pancreata (Fig. 5C; Supplementary Fig. S5A) as well as in human PDA samples (Fig. 5D), further validating the presence of active NLRP3 inflammasomes in these cells.

To determine whether the NLRP3 inflammasome is the primary source of pro-IL1β–processing in tumor cells, we knocked down NLRP3 expression in KRas^G12D-PDEC and KPC cells using two independent short hairpins (Supplementary Fig. S5B and S5C). Depletion of NLRP3 in the transformed ductal epithelia significantly reduced cleaved caspase-1 expression and IL1β production in IL1β-sh KRas^G12D-PDEC pancreatic grafts, relative to scramble control (Fig. 5E–G; Supplementary Fig. S5D). This was accompanied by a decrease in growth of IL1β-sh KRas^G12D-PDEC pancreatic grafts (Fig. 5H; Supplementary Fig. S5E) as well as decreased tumor growth and increased overall survival of orthotopic IL1β-sh KPC tumor-bearing mice (Fig. 5I–K). Together, these results implicate the NLRP3 inflammasome in the production of tumor cell–derived IL1β and define a tumor-supportive role for NLRP3 in pancreatic cancer.

NLRP3 inflammasome assembly and the posttranslational processing of IL1β that ensues is predominantly regulated by Toll-like receptors (TLR) through induction of IL1B and NLRP3 expression in response to pathogens or cellular damage (44). Members of the TLR family are expressed in various cancers and have been shown to promote tumor growth (45). In pancreatic cancer, while most TLRs have been shown to be expressed largely in stromal cells, elevated TLR4 expression has been reported in the tumor cell compartment of patient PDA samples and shown to be correlated with reduced survival (46, 47). Consistent with these findings, analysis of a panel of patient PDA samples revealed significant upregulation of tumor-associated TLR4 expression, relative to adjacent normal tissue (Fig. 6A), and IHC analysis demonstrated robust TLR4 expression in tumor cells in human PDA as well as KC and KPC mouse pancreata (Fig. 6B and C; Supplementary Fig. S6A). To determine whether IL1β production in tumor cells is TLR4-driven, we employed RNAi to stably knockdown TLR4 expression in KRAS^G12D-PDEC (Supplementary Fig. S6B). Transformed ductal epithelium of TLR4-sh KRas^G12D-PDEC pancreatic grafts had significantly reduced IL1β production, relative to scramble control (Fig. 6D and E). Moreover, TLR4 knockdown in KRas^G12D-PDEC and KPC cells decreased growth of orthotopic KRas^G12D-PDEC pancreatic grafts (Fig. 6F; Supplementary Fig. S6C) and KPC tumors, respectively (Fig. 6G and H; Supplementary Fig. S6D) and increased overall survival of TLR4-sh KPC orthotopic tumor bearing mice, relative to scramble control (Fig. 6I). We conclude TLR4 thus serves as a critical regulator of tumor cell–derived IL1β production and pancreatic tumorigenesis.

Having identified TLR4 as the receptor that controls IL1β production in pancreatic tumor cells, we next searched for possible cues in the pancreatic microenvironment that can induce TLR4 signaling. Recent reports on the existence of a complex pancreatic microbiome (23) prompted us to hypothesize that microbial-derived ligands that are known to activate TLR4 signaling (17) could be responsible for inducing IL1β production in pancreatic tumor cells. To test this hypothesis, we treated wild-type mice with an antibiotic cocktail for 3 weeks to ablate their microbiome prior to implantation of KRas^G12D-PDEC (Supplementary Fig. S6E). KRas^G12D-PDEC grafts formed in antibiotic-treated mice indeed displayed a significant reduction in tumor cell IL1β expression without affecting TLR4 level, relative to vehicle-treated control mice (Fig. 6J and K; Supplementary Fig. S6F). Overall, our results indicate a role for the pancreatic microbiome in initiating a signaling cascade, likely through TLR4, to activate IL1β production in pancreatic tumor cells.

In this study, we demonstrate a role for tumor cell–derived IL1β in promoting pancreatic oncogenesis by paracrine induction of heterotypic stromal interactions. Specifically, we show that tumor-derived IL1β is critical for shaping the tolerogenic immune landscape of PDA by promoting stromal accumulation of immunosuppressive cell populations. These include M2-polarized macrophages, tumor-associated neutrophils, IL17-producing Th17 cells, MDSCs, and CD1d^hiCD5⁺ regulatory B cells. In addition, we report that tumor-derived IL1β regulates PDA-associated desmoplasia by promoting activation of quiescent PSCs.

IL1β is a member of the IL1 family of proinflammatory cytokines, which also includes the cofounding member, IL1α (7). Both IL1α and IL1β are critical immune regulators that signal through a common cell surface receptor (IL1R1-IL1RAcP) to activate two main pathways: IKK–IκB–NF-κB and/or MKK–MAPK/JNK/ERK (6). Despite considerable functional homology, the two cytokines differ appreciably in several aspects. While IL1α is predominantly a cytosolic or membrane-bound protein constitutively expressed in epithelial, endothelial, and immune cells, IL1β is a secretory protein chiefly produced by immune cells only in response to inflammatory cues (7). In PDA, previous studies have predominantly categorized IL1 signaling into tumor cell production of IL1α (48) and stromal production of IL1β (15, 16, 18). In fact, IL1β protein is reportedly undetectable in PDA cell lines and organoids in vitro (33, 49). Consistent with these reports, we too did not detect IL1β production by KRas^G12D-PDEC and KPC cells cultured ex vivo. We did, however, observe a robust in vivo production of the IL1β protein in the tumor cell compartment of human and murine PDA. This suggests the existence within the pancreatic tumor microenvironment of regulatory cues that can induce the activation of the toll-like receptor signaling pathway in tumor cells to drive IL1β expression and posttranslational processing. This conclusion is supported by our finding that the TLR4/NLRP3 inflammasome signaling axis is active in pancreatic tumor cells and is required for the production of IL1β by these cells.

The TLR4 receptor is a specific sensor of exogenous microbial ligands such as lipopolysaccharides (LPS) as well as endogenous ligands termed damage-associated molecular patterns (DAMP), derived from host tissue or cells (46). Significantly, the pancreatic microenvironment has been shown to be rich in such endogenous TLR4 ligands including HMGB-1 and S100A9 that can activate TLR4 signaling in tumor cells (17). Furthermore, the recently described PDA-associated microbiome has been shown to be rich in microbial ligands capable of activating the toll-like receptor pathway (23). Our finding that bacterial dysbiosis leads to inhibition of tumor cell–derived IL1β production indicates that the pancreatic microbiome plays a significant role in inducing IL1β production in transformed cells, likely through the TLR pathway. Moreover, it raises the possibility that the prooncogenic role of the microbiome in pancreatic cancer (23) could be, in part, mediated by the activation of TLR4-mediated IL1β production in the tumor cell compartment.

We have established that the posttranslational mechanisms that drive IL1β processing and maturation in the tumor cells require the NLRP3 inflammasome. The precise nature of the TLR4-induced signals that promote the assembly and activation of the NLRP3 inflammasome in pancreatic tumor cells remain to be determined. In monocytes, TLR4 ligation can induce ATP release, which, in turn, triggers NLRP3 inflammasome assembly via the ATP-gated ionotropic receptor P2X7 (P2RX7) (40). Interestingly, P2RX7 is highly expressed in pancreatic cancer cells (50), suggesting that a cooption by the tumor cells of this immune cell–specific signaling axis might be responsible for the NLRP3 inflammasome activation.

As indicated by our loss-of-function and rescue experiments, the secretion by pancreatic tumor cells of IL1β instigates sweeping changes in the fibroinflammatory pancreatic milieu, in part, by modulating PSC function. PSCs have been implicated in the regulation of a plethora of protumorigenic processes including tumor cell growth and metabolic adaptation and metastasis (31, 32). Recently, Öhlund and colleagues have described the presence of two distinct intratumoral CAFs subpopulations in PDA: myofibroblastic cancer-associated fibroblasts (myCAF) with elevated αSMA expression and inflammatory cancer-associated fibroblasts CAF (iCAF) expressing an array of cytokines and chemokines (51). Our data indicate a role for IL1β in regulating the secretory phenotype of inflammatory CAFs. Specifically, we demonstrate the dependence of stromal CAFs on tumor-derived IL1β for the production of cytokines and chemokines with documented roles in subverting antitumor immunity. These include the chemokines CCL2 and CCL5 that regulate chemotaxis of monocytes, and, in the context of pancreatic cancer, have been found to regulate macrophage infiltration and M2 polarization (37, 38), as well as the chemokine CXCL12, which is known to inhibit intratumoral accumulation of CD8⁺ T cells (39). In accordance with this IL1β-dependent secretory profile, we found that upregulation of M2-TAMs and restriction of CD8⁺ T_c cell tumor infiltration is dependent on stromal PSCs. Overall, our study delineates epistatic interactions between tumor cell–derived IL1β and PSCs that are critical for the establishment of immune tolerance in pancreatic cancer.

The low immunogenicity of pancreatic cancer due to poor tumor infiltration of CD8⁺ T_c cells is considered a major factor responsible for the failure of checkpoint immunotherapy in PDA (3). As demonstrated by our studies, neutralizing IL1β promotes intratumoral CD8⁺ T_c-cell infiltration and function and sensitizes PDA to checkpoint immunotherapy. Hence, therapeutic strategies that target IL1β may increase the efficacy of immune checkpoint inhibitors in pancreatic cancer. It is noteworthy that in a recent analysis of pancreatic cystic neoplasms (PCN) in patients, intracystic bacterial load as well as increased IL1β protein levels were detected in cystic precursors to pancreatic cancer called intraductal papillary mucinous neoplasms (IPMN), relative to non-IPMN PCNs (52). This study in combination with our current findings suggest that IL1β production might be an early event in pancreatic tumorigenesis, the targeting of which could be used to impede disease progression.

No potential conflicts of interest were disclosed.

Conception and design: S. Das, D. Bar-Sagi

Development of methodology: S. Das, D. Bar-Sagi

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Das, B. Shapiro, S. Vogt

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Das, B. Shapiro, E.A. Vucic

Writing, review, and/or revision of the manuscript: S. Das, E.A. Vucic, S. Vogt, D. Bar-Sagi

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Das

Study supervision: D. Bar-Sagi

The authors thank L.J. Taylor for help with article preparation, Mark Choi for technical assistance, and members of the Bar-Sagi lab for valuable discussions and comments. The KPC cell line (line 4662) was a kind gift from Dr. R.H. Vonderheide. The immortalized CAF cell line isolated from KPC tumors was a kind gift from Dr. A.C. Kimmelman. We thank Novartis for providing the mouse anti-PD1 and anti-IL1β antibodies and isotype controls for this study. The TROMA-I antibody against cytokeratin-8 developed by Brulet P and colleagues, Institut Pasteur, was obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology. The authors also thank NYU Langone's Cytometry and Cell Sorting Laboratory and the Experimental Pathology Research Laboratory, which are supported, in part, by grant P30CA016087 from the NIH/National Cancer Institute, for providing cell sorting/flow cytometry technologies and histochemistry support, respectively, as well as the Center for Biospecimen Research and Development for providing patient pancreatic tumor tissue sections. This work was supported by NIH/NCI grant CA210263 (D. Bar-Sagi) and by a Stand Up To Cancer-Lustgarten Foundation Pancreatic Cancer Convergence Dream Team Name Translational Research Grant (SU2C-AACR-DT14-14). Stand Up to Cancer is a division of the Entertainment Industry Foundation administered by the American Association for Cancer Research, the Scientific Partner of SU2C. B. Shapiro was supported by NIH grant T32GM115313. E.A. Vucic was supported by a Canadian Institutes of Health Research Fellowship (146792).

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