Neutrophil Extracellular Traps Modulate Recruitment and Immunosuppression of Macrophages in Pancreatic Adenocarcinoma

Abstract

Pancreatic adenocarcinoma (PDAC) has a dismal survival rate due to limited effective therapies. Although studies have focused on the influence of innate immune cells on adaptive immune cell functions, few have explored interactions between innate immune cells, which modulate the PDAC tumor microenvironment (TME). Macrophages are responsible for the clearance of neutrophil-mediated inflammation in physiologic, resolving immune responses; however, both of these cell types coexist in the TME, suggesting a failure of macrophages to clear neutrophils in PDAC. We sought to determine how neutrophil extracellular traps (NET), neutrophil release of decondensed chromatin, and intracellular contents affect monocyte/macrophage populations in the PDAC TME. Utilizing samples from patients with PDAC, we demonstrated elevated levels of the monocyte chemokine CCL2 in plasma, as well as elevated NET citrullinated histone H3 and the pan-macrophage marker CD68 in the PDAC TME via fluorescent IHC. To determine how NETs affected macrophage populations in the PDAC TME, we targeted NETs with DNase I treatment to digest extracellular DNA released from NETs or with genetic knockout of PAD4, an enzyme required for NET formation. NET depletion resulted in an elevation in the pan-macrophage marker F4/80. The depletion led to an increased T-cell stimulatory signal, CD80, whereas the protumor macrophage marker CD206 was decreased. We further demonstrated that macrophages in the NET-deficient PDAC TME may be recruited through the CCL2/CCR2 axis, and CCL2 was released from tumor cells and macrophages in the presence of IFNγ. Taken together, our findings reveal that inhibition of NETs can prime the innate immune response toward an antitumor phenotype.

Introduction

The 5-year survival rate for patients with pancreatic adenocarcinoma (PDAC) is only 12% (1). Although there have been significant improvements in recent years, there are still limited therapeutic options for patients with PDAC. Despite recent advances in chemotherapy, the current standard-of-care chemotherapy regimens only increase survival for metastatic disease to just over 11 months (2, 3). Given that PDAC is projected to become the second leading cause of cancer deaths by the year 2030 (4), therapeutic options to improve patient outcomes are needed.

In recent years, immune checkpoint blockade has drastically improved outcomes for patients with several cancer types (5). Despite success in other cancers, immune checkpoint inhibitors have not been successful in PDAC (5). To harness these therapies, we need a better understanding of the unique, suppressive immune cell functions regulated by the PDAC tumor microenvironment (TME). Adaptive immune cells have been a primary focus in PDAC research; however, innate immune cells have also been implicated in PDAC progression. Specifically, blood monocytes are elevated in patients with PDAC, and a high ratio of circulating CCR2⁺ inflammatory monocytes to their bone marrow precursors is a negative prognostic indicator for patients (6). Similarly, a high systemic inflammatory index [platelets × (neutrophil/lymphocytes)] is also a negative prognostic indicator (7–9). Neutrophils, which can release decondensed chromatin into their surroundings, known as neutrophil extracellular traps (NET), are also elevated in PDAC (10–12) and have negative prognostic implications for patients with PDAC (11, 13).

Although many studies have focused on how innate immune cells affect the adaptive immune system in the PDAC TME (6, 14, 15, 16), few studies have focused on interactions between different innate immune cells (15). The relationship between neutrophils and monocytes/macrophages in tumor models, in particular, is understudied. In a murine model of PDAC, inhibition of CCR2⁺ monocytes caused a compensatory increase in CXCR2⁺ neutrophils (15). This may be explained by monocytes specifically inhibiting neutrophil recruitment, which has been demonstrated in a cervical carcinogenesis model (17). This correlation is reciprocal, as inhibition of CXCR2⁺ neutrophils increases infiltrating monocytes (15). Despite this interplay when one population is inhibited, the prevalence of both cell types is overall increased in the PDAC TME compared with the normal pancreas. Using The Cancer Genome Atlas (TCGA) database, we have previously shown that genes for neutrophils and macrophages of different ontogeny are concomitantly elevated in PDAC relative to the normal pancreas, which was unique compared with other cancer types (18). These data suggest a critical need to better understand the complex interactions between these cell types in pancreatic cancer.

Although neutrophils and monocytes have been targeted in a limited number of studies in PDAC (15), the interactions between NETs and monocytes in PDAC have not been studied. In this study, we investigated how NETs influence the recruitment of monocytes into the PDAC tumor. When NETs were removed from the PDAC TME, there was an increase in F4/80⁺ macrophages in the PDAC TME. We showed higher CD80 (antitumor phenotype) and lower CD206 (protumor phenotype) immunostaining in the absence of NETs. We also suggested that these macrophages may be recruited via the CCL2/CCR2 axis. These findings demonstrate a relationship between NETs and monocytes/macrophages in PDAC, with implications in the PDAC TME and in response to treatment.

Materials and Methods

Patient plasma and tissue collection

Prior to sample collection, approval was obtained from the West Virginia University Institutional Review Board #1903496995. Patient studies were conducted in accordance with the Declaration of Helsinki. All patients signed informed consent prior to participation. Seventy-two patients undergoing pancreaticoduodenectomy between May 2019 and May 2022 for PDAC or a noncancerous pancreatic pathology were included in the analysis. Diagnosis was confirmed by pathology. Blood was obtained immediately prior to surgery in EDTA tubes and centrifuged at 1,000 × g for 10 minutes at 20°C to collect plasma. Plasma was aliquoted and stored at −80°C until use. PDAC tumor tissue or noncancerous pancreatic tissue was preserved in optimal cutting temperature and stored at −80°C until use.

Murine orthotopic PDAC model

All animal studies were approved by the Institutional Animal Care and Use Committee. A Panc02 model of murine pancreatic cancer was utilized as previously described (10). C57BL/6J mice (000664) were purchased from The Jackson Laboratory. PAD4^−/− mice were bred according to Animal Care and Use Committee guidelines. Briefly, 1 × 10⁶ Panc02 (NCI repository, 2008) cells were injected into the tail of the pancreas through a limited laparotomy in male and female mice, ages 8 to 10 weeks. Animals were euthanized for tissue collection 24 days after tumor inoculation (or 21 days after the start of the treatment). Blood was collected by cardiac puncture into EDTA. For plasma, blood was centrifuged at 1,000 × g for 10 minutes at 20°C. Plasma was stored at −80°C until analyzed. Pancreata were collected and stored in O.C.T. Compound (Fisher Healthcare, 23-730-571; for immunofluorescent staining) at −80°C. Pancreatic tumors used for flow cytometry were collected for immediate dissociation and staining. Animal treatment and euthanasia were carried out in strict accordance with protocols approved by the Institutional Animal Care and Use Committee at West Virginia University and in compliance with the Association for Assessment and Accreditation of Laboratory Animal Care International.

Fluorescent IHC staining

Ten-micrometer sections of optimal cutting temperature–embedded frozen tissues were fixed in 10% formalin (Fisher Scientific, SF984) and washed with Dulbecco’s PBS (DPBS; Gibco, 14190144), then permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, T-9284) for 10 minutes, followed by a DPBS wash. Tissues were blocked for 1 hour at room temperature with 5% BSA (MACS BSA Stock Solution, Miltenyi Biotec, 130-091-376) and 10% goat serum (Fisher Scientific, NB036772) in DPBS. Primary antibodies were added overnight at 4°C [rat polyclonal anti-CK19 (DSHB, TROMA-III), 1:200; mouse anti-human CD68 (BD Biosciences, clone Y1/82A, BDB556059), 1:100; rabbit polyclonal anti-Histone H3 (Abcam, ab5103), 1:200; rat anti-mouse F4/80 (Novus Biologicals, NBP1-60140), 1:100; rabbit polyclonal anti-CD80 (Invitrogen, PA5-85913), 1:100; and rat anti-mouse CD206 (Invitrogen, MA5-16871), 1:100]. Tissues were washed with DPBS, and secondary antibodies were added for 1 hour at room temperature (Alexa Fluor 488 and Alexa Fluor 647 goat anti-mouse, Invitrogen, A-21235; Alexa Fluor 647 goat anti-rabbit, Invitrogen, A-27040; Alexa Fluor 647 goat anti-rat, Invitrogen, A-21247), then washed with DPBS. The Vector TrueVIEW Autofluorescence Quenching Kit (Vector Laboratories, SP-8400-15) was prepared per the manufacturer’s instructions and added for 5 minutes. After the DPBS wash, tissues were stained with 4’,6-diamidino-2-phenylindole (DAPI; Thermo Scientific, 62248), 1:1,000, and washed. VECTASHIELD Vibrance Antifade Mounting Media (patient samples and DNase treatment experiments; Vector Laboratories, H-1700) or SlowFade Gold Antifade Mountant (PAD4^−/− experiments; Thermo Fisher Scientific, S36936) was used to mount coverslips. All antibodies were diluted in 1% BSA and 10% goat serum in DPBS. Imaging was performed on a Nikon A1R confocal on a Nikon ECLIPSE Ti microscope using Nikon Confocal NIS-Elements AR version 5.21.03 software with the Plan Apo 20x DIC N2/0.75 objective, WD 1.0, and photomultiplier and gallium arsenide phosphide (GaAsP) detectors. For quantification, 3 to 6 images were taken at 20^ magnification randomly throughout the tissue section. Images were quantified using ImageJ software by splitting the color channels on raw images into red, green, and blue, using the Automated Threshold function (MaxEntropy, Li) and the Ignore Black and White selection. After thresholding the red channel (all stains for quantification were in the red channel), the Analyze Particles function was used to generate the percentage of area presented in figures. Assembled representative images captured from the Nikon Confocal NIS-Elements AR software were opened in Adobe Photoshop 2024 and brightened using the Image menu > Adjust > Levels feature. The same value was applied to all figures to maintain consistency across images.

NET depletion in mice

DNase I treatment to remove NETs was employed as previously described (11). Briefly, DNase I (MilliporeSigma, 11284932001) treatment (2,000 U/mg) was started 72 hours after Panc02 tumor inoculation at a dose of 5 mg per kg mouse in PBS, pH 7.4 (Gibco, 10-010-049; e.g., a 25 g mouse would receive 0.125 mg DNase, which is 250 U) and was administered daily by intraperitoneal injection until euthanasia. The vehicle control was PBS (pH 7.4). Final DNase I and vehicle control were administered 1 hour prior to euthanasia and tissue collection. C57BL/6J PAD4^−/− (026708) male and female mice obtained from The Jackson Laboratory were orthotopically implanted with 10⁶ Panc02 tumor cells as described above (10). PAD4^−/− mice were also perfused with Lectin DyLight 488 (Vector Laboratories, NC0358674) prior to exsanguination.

Cytokine quantification

Patient plasma levels of CCL2 were quantified with the Human CCL2/MCP-1 Quantikine ELISA Kit (R&D Systems, DCP00) following the manufacturer’s protocol. Murine plasma CCL2 was measured via the ELISA MAX Deluxe Set Mouse MCP-1 (BioLegend, 432704) per the manufacturer’s protocol. In vitro CCL2 was measured with a Quantikine ELISA kit (Bio-Techne, MJE00B). Murine plasma IFNγ was measured via the ELISA MAX Deluxe Set Mouse IFNγ (BioLegend, 430804) per the manufacturer’s protocol. Absorbance was measured using a SpectraMax iD3 plate reader and SoftMax Pro software version 7.1.0.

Flow cytometry

To isolate circulating leukocytes, 500 μL of EDTA-treated whole blood was added to 10 mL of ACK Lysing Buffer (Gibco, A1049201) at room temperature for 5 minutes and washed in cold PBS. Bone marrow cells were isolated from femurs as previously described (19). Briefly, bone marrow was washed from femurs with cold RPMI and centrifuged at 1,200 RPM for 5 minutes. To isolate tumor leukocytes, tumors were excised, dissociated on a gentleMACS Dissociator (Miltenyi Biotec), pushed through a 70 μm filter, and centrifuged at 300 × g for 7 minutes according to the Miltenyi Mouse Tumor Dissociation Kit protocol (Miltenyi Biotec, 130-096-730). Anti-CD16/CD32 (BD Pharmingen, 553142) was added to the isolated cell suspension to block Fc receptors for 10 minutes. The EasySep Mouse TIL (CD45⁺) Positive Selection Kit (STEMCELL Technologies, 100-0350) and STEMCELL Technologies EasySep Magnets were used to positively select for tumor leukocytes. Isolated blood leukocytes, tumor leukocytes, or bone marrow cells were then surface stained in 96-well plates in 100 μL FACS buffer (PBS + 1% FBS) for 30 minutes at 4°C using the following reagents: LIVE/DEAD Violet (Invitrogen, 1:5,000 dilution), CD45-APC/Cy7 (clone 30-F11, BioLegend, 1:200), CD11b-FITC (clone M1/70, BioLegend, 1:200), Ly6C-BV510 (clone HK1.4, BioLegend, 1:200), Ly6G-PerCP/Cy5.5 (clone IA8, BioLegend, 1:200), CCR2-PE/Cy7 (clone SA203G11, BioLegend, 2.5 mL), F4/80-APC (clone BM8, BioLegend, 1:100), CD80-PE (clone 16-10A1, BioLegend, 1:100), or CCR2-BV711 (clone SA203G11, BioLegend, 1:20). Cells were fixed in 2% paraformaldehyde (Thermo Fisher Scientific, AAJ19943K2) for 20 minutes at 4°C. Tumor cells were further stained intracellularly by permeabilization using the eBioscience FoxP3/Transcription Factor Staining Buffer Set (Invitrogen, 00-5523-00), followed by staining with CD206-PE/Cy7 (clone C068C2, BioLegend, 1:100) for 30 minutes. Flow cytometric analysis of blood leukocytes and bone marrow cells was done with the BD LSRFortessa, and analysis of tumor leukocytes was done with the Cytek Aurora. Data were analyzed using FlowJo version 9 (BD Life Sciences; for blood/bone marrow flow cytometry) and FCS Express 7 (De Novo Software; for intratumor flow cytometry).

NET media generation and measurement

Protocol for neutrophil isolation was adapted from previously described methods (20). The bone marrow of femurs was washed with RPMI 1640 through a 100 μm cell strainer using a 10-cm³ syringe and a 27-gauge needle. Bone marrow was washed with 10 mL of RPMI, then centrifuged at 1,400 RPM for 7 minutes at 4°C. The cells were washed, and the resulting pellet was resuspended in 1 mL of cold DPBS. Neutrophils were collected by density centrifugation with Histopaque-1119 (Sigma-Aldrich, 11191) and Histopaque-1077 (Sigma-Aldrich, 10771) overlay and washed with RPMI 1640. Neutrophils were resuspended in Hank’s Balanced Saline Solution (Thermo Fisher Scientific, J67763.AP) and plated in a T75 flask for 30 minutes. Then, NETs were stimulated by plateletactivating factor (Sigma-Aldrich, 511075) to a final concentration of 50 μmol/L. After 3 hours, Hank’s Balanced Saline Solution was removed and replaced with RPMI 1640 with 1% penicillin/streptomycin (Cytiva HyClone, SV30010). NETs were removed from the plate using a cell scraper. The resulting media was centrifuged at 1,200 RPM for 10 minutes at 20°C to collect cell-free supernatants. Supernatants were stored at −80°C until use. The resulting NET DNA concentration was quantified using the QuantiT PicoGreen dsDNA Assay Kit (Invitrogen, P7589) per the manufacturer’s protocol. Fluorescence was measured using a SpectraMax iD3 plate reader and SoftMax Pro software version 7.1.0.

Cell release of CCL2

Panc02 murine cells were cultured in RPMI 1640 (Corning, 10-041-CV), 10% FBS (Corning, 35-010-CV), and 1% penicillin/streptomycin (Cytiva HyClone, SV30010). Cells were plated in a 24-well plate at 30,000 cells per well overnight. The cells were then treated with IFNγ (Bio-Techne, 485-MI-100) diluted in the corresponding serum-free medium. Media were collected after 24 hours and centrifuged at 1,200 RPM for 10 minutes at 20°C to collect cell-free supernatants. Supernatants were stored at −80°C until analysis.

Bone marrow–derived macrophages (BMDM) were differentiated from C57BL/6J mice as previously described (21). Briefly, the bone marrow of femurs was washed with RPMI 1640 through a 70 μm cell strainer using a 10-cm³ syringe and a 27-gauge needle. Cells were washed and plated at 1,000,000 cells/well in a 24-well plate and cultured in RPMI 1640, 10% low endotoxin FBS, 1% penicillin/streptomycin, 10 ng/mL polymyxin B (Fisher Scientific, 501779478), and 20 ng/mL murine recombinant CSF1 (Bio-Techne, 416-ML). On the second day, cells were spiked with additional CSF1 (5 ng/mL). After 4 days, the media were changed to RPMI 1640, 10% low endotoxin FBS (Bio-Techne, S11550), and 1% penicillin/streptomycin (Cytiva HyClone, SV30010). Macrophages were treated with IFNγ media diluted in RPMI 1640 and 1% penicillin/streptomycin for 24 hours. Media were then collected and centrifuged at 1,200 RPM for 10 minutes at 20°C to collect cell-free supernatants. Supernatants were stored at −80°C until analysis.

GEPIA2021 analysis

Gene Expression Profiling Interactive Analysis (GEPIA2021) (22) is an open-access online tool and was used to determine the cell type differential gene expression of CCL2 and IFNG from deconvoluted immune cell populations in PDAC tumors in the TCGA-PAAD and Genotype-Tissue Expression (GTEx) pancreas bulk RNA-seq datasets. The GEPIA2021 tool (http://gepia2021.cancer-pku.cn/) was utilized without modification with the bulk data deconvolution data accessible at https://github.com/zwj-tina/GEPIA2021 developed by Li and colleagues (22). According to Li and colleagues, GEPIA2021 data were downloaded from the UCSC Xena project (23). All sample analyses were performed via the GEPIA2021 tool. In the Sub-Expression Analysis tool, the parameters for analysis were as follows: TCGA tumor was pancreatic adenocarcinoma (PAAD), TCGA normal was PAAD, GTEx was pancreas, and the deconvolution tool was Estimating the Proportions of Immune and Cancer cells (EPIC) or quanTIseq. In the EPIC deconvolution matrix, B cells, CD4 T cells, CD8 T cells, NK cells, and macrophages were selected. In the quanTIseq deconvolution matrix, M1 macrophages, M2 macrophages, and monocytes were selected. The subexpression of CCL2 and IFNG was measured as reported. Figures, log(transcripts per million + 1) values, and F and P values were generated by the GEPIA2021 tool.

Statistics

Statistics were performed using GraphPad Prism. Outliers were removed using the Grubbs or ROUT test for outliers in murine and in vitro models. A Student t test, Mann–Whitney U test, or ordinary one-way ANOVA was performed as indicated in figure legends.

Results

NET marker citrullinated histone H3, macrophage marker CD68, and monocyte chemokine CCL2 are elevated in patients with PDAC

Previous studies have shown that NETs (13) and circulating proinflammatory monocytes (6) portend a worse prognosis in patients with PDAC and are concomitantly elevated in PDAC (18). Circulating NET markers are elevated in PDAC and correlate with disease stage (11, 24). We hypothesized that the classical monocyte chemokine CCL2 would also be elevated in the circulation of patients with PDAC. We collected whole blood from patients with PDAC (n = 55) or noninvasive pancreatic pathology (n = 17) prior to surgical resection via pancreaticoduodenectomy. We observed no differences in age, body mass index, and smoking status between groups, but there was a significant difference in sex between the PDAC and noninvasive pathology groups (Supplementary Table S1). The noninvasive pathologies included both benign and pre-malignant conditions, including intraductal papillary mucinous neoplasia (Supplementary Fig. S1A). A total of 38.2% of patients with PDAC underwent neoadjuvant therapy prior to sample collection (Supplementary Fig. S1B), with all patients receiving folate, 5-fluorouracil, oxaliplatin, and irinotecan or gemcitabine-based therapies (Supplementary Fig. S1C). We measured the monocyte chemokine CCL2 in plasma and found that it was significantly elevated in patients with PDAC compared with those with noninvasive pancreatic pathologies (Supplementary Fig. S2A). We observed no difference in CCL2 levels between tumor stage or node status or in patients receiving neoadjuvant chemotherapy (Supplementary Fig. S3). Taken together, these findings support established work demonstrating that CCL2 is elevated in the circulation of patients with PDAC compared with other pancreatic conditions.

To determine if our findings in the plasma translated to increased NETs and macrophages in PDAC tissue, we measured the NET marker citrullinated histone H3 (CitH3; refs. 25, 26) and the pan-macrophage marker CD68 in resected frozen noninvasive pancreatic tissue and PDAC tumors via fluorescent IHC (FL-IHC). Consistent with existing literature (10, 13, 27), we found that the percentage of the positive area of CitH3 immunostaining was significantly greater in PDAC compared with noncancerous tissues (Supplementary Fig. S2B and S2C). We further found that the percentage of the positive area of CD68 immunostaining was also significantly greater in PDAC compared with noninvasive tissues (Supplementary Fig. S2D and S2E). Overall, these findings support that NETs and macrophages coexist in the PDAC TME to a greater extent than in noninvasive pancreatic tissues and support that circulating levels of the macrophage chemokine CCL2 are elevated in circulation.

DNase I treatment removes NETs and increases F4/80 immunostaining

To further determine if NETs can affect macrophage presence in the setting of PDAC, we first sought to remove NETs with DNase I in an orthotopic Panc02 murine model of PDAC. Three days after orthotopic implantation, we treated the mice with daily intraperitoneal (IP) DNase I (at 2,000 U/mg) or PBS, pH 7.4 (vehicle), injecting 5 mg per kg of mouse for 21 days (Fig. 1A). As shown in previous reports demonstrating that NETs support tumor growth (11, 28), DNase I treatment decreased pancreas weight (Fig. 1B). Tumors were then resected, and the percentage of the positive area of CitH3 immunostaining via FL-IHC was compared. We observed decreased CitH3 in the tumors of DNase I–treated mice compared with the vehicle-treated group (Fig. 1C and D), validating the ablation of NETs with DNase I treatment. To determine how removing NETs influences overall macrophage numbers within the PDAC TME, we immunostained for F4/80, a murine pan-macrophage marker, in the PDAC tumors (Fig. 1E). F4/80 immunostaining was greater in the DNase I–treated group than in the control group (Fig. 1F), suggesting increased monocyte infiltration or resident macrophage expansion.

Figure 1.

DNase I treatment decreases intratumoral CitH3 and increases intratumoral F4/80. A, Experimental design for the murine orthotopic model and treatment. B, Pancreas weight was compared between Panc02 orthotopic murine cohorts with PBS and DNase I treatment (n = 10 per group). C, Representative CitH3 staining in Panc02 orthotopic tumors in PBS- and DNase I–treated cohorts. D, Quantification of CitH3 staining between PBS- and DNase I–treated groups (n = 5 per group). E, Representative F4/80 staining in Panc02 orthotopic tumors in PBS- and DNase I–treated cohorts. F, Quantification of F4/80 staining compared between PBS- and DNase I–treated cohorts (n = 5 per group). Values are reported as mean ± SD. Student’s unpaired t test was used to compare groups. *, P < 0.05.

Removal of NETs promotes an antitumor phenotype in the PDAC TME

Next, given the detrimental role of tumor-associated macrophages (TAM) in PDAC, including the promotion of fibrosis (29), immunosuppression (30), and metastasis through epithelial-to-mesenchymal transition (31), or increased microvessel density (32), we sought to determine if the depletion of NETs affects the pro- or antitumor phenotype of macrophages. We immunostained PDAC tissue for the antitumor, T cell–activating costimulatory molecule CD80, which lowers the activating threshold of T cells (33) and is predominantly expressed by M1-like (antitumor) macrophages (34), and the M2-like (protumor) macrophage mannose receptor CD206, commonly expressed on TAMs and associated with worse prognosis in many tumor types, including PDAC (35). Our data indicate that the normal pancreas did not express CD80 or CD206 (Fig. 2A). DNase I treatment resulted in a significantly higher percentage of the positive area of CD80 by FL-IHC (Fig. 2B and C) and increased CD80 geometric mean fluorescence intensity as measured by flow cytometry gated on F4/80-high macrophages (Fig. 2D). Flow cytometry gating of tumor leukocytes is detailed in Supplementary Fig. S4. Additionally, there was a significantly lower percentage of the positive area of CD206 (Fig. 2E and F). Although costaining for these markers was not performed, the markers are predominantly expressed on macrophages in the PDAC TME (36–39). These data suggest that NETs may also have a role in the regulation of macrophage pro- and antitumor phenotypes in PDAC. Further supporting these findings, we measured the proinflammatory cytokine IFNγ, known to promote an antitumor, M1-like macrophage phenotype (40). In mice treated with DNase I, IFNγ levels were elevated (Fig. 2G). These results support a potential shift toward an antitumor macrophage phenotype in the absence of NETs.

Figure 2.

Anti- and protumor markers are altered by DNase I treatment. A, Representative CD80 and CD206 immunofluorescent staining in normal murine pancreas. B, Representative CD80 immunofluorescent staining in Panc02 orthotopic tumors in PBS- and DNase I–treated cohorts. C, Quantification of CD80 immunofluorescent staining between PBS- and DNase I–treated cohorts (n = 5 per group). D, Flow cytometric quantification of CD80 geometric mean fluorescence intensity (gMFI) on CD45⁺CD11b⁺F4/80^high cells isolated from tumors compared between PBS- and DNase I–treated groups (PBS, n = 13; DNase, n = 14). E, Representative CD206 immunofluorescent staining in Panc02 orthotopic tumors in PBS- and DNase I–treated groups. F, Quantification of CD206 staining between PBS- and DNase I–treated cohorts (n = 5 per group). G, IFNγ quantification in plasma of Panc02 orthotopic mice compared between PBS- and DNase I–treated cohorts (n = 4 per group). Values are reported as mean ± SD. A Student unpaired t test was used to compare groups. *, P < 0.05; **, P < 0.01.

Ablation of NETs from the PDAC TME alters circulating CCL2 levels

As the number of macrophages increased in DNase I–treated tumors and those macrophages expressed markers suggestive of an antitumor phenotype, we sought to determine the source of the increased macrophages, which reside in the pancreas from both infiltration of circulating monocytes (6) and proliferation of tissue-resident macrophages (29) in PDAC. One prevalent source of macrophages in the PDAC TME has been reported to be peripheral blood monocytes recruited via the CCL2/CCR2 axis (6, 41); however, clinical trials inhibiting the recruitment of monocytes via CCR2 antagonism only modestly improved tumor responses (14, 42). Thus, we hypothesized that NETs, which are known to promote immunosuppressive T cells (28, 43), may also play a role in restricting the CCL2/CCR2 recruitment axis and may represent an upstream target for modulating the activity of monocytes and macrophages. To determine how NETs influence the CCL2/CCR2 recruitment of monocytes, we first compared plasma levels of CCL2 between DNase I– and PBS-treated groups. Our data suggest an increase in plasma CCL2 in the DNase I–treated mice compared with the PBS-treated group (Fig. 3A). This suggests that the inhibition of NETs leads to an increase in the activity of the CCL2/CCR2 axis, similar to the increased recruitment of CCR2⁺ monocytes following neutrophil inhibition in PDAC murine models (15).

Figure 3.

CCL2/CCR2 axis is altered in the absence of NETs. A, CCL2 quantification in plasma of Panc02 orthotopic mice compared between PBS- and DNase I–treated cohorts (PBS, n = 7; DNase, n = 10). B, Representative plots of Ly6C and Ly6G staining of blood leukocytes isolated from Panc02 orthotopic mice for PBS- and DNase–treated cohorts. C, Flow cytometric quantification of Ly6C^hiLy6G⁻ monocytes isolated from blood leukocytes of Panc02 orthotopic mice compared between PBS- and DNase I–treated cohorts (PBS, n = 7; DNase, n = 8). D, Quantification of Ly6C^loLy6G⁺ neutrophils isolated from blood leukocytes of Panc02 orthotopic mice compared between PBS- and DNase I–treated cohorts (PBS, n = 7; DNase, n = 8). E, Representative plots for CCR2 staining on Ly6C^hiLy6G⁻ circulating monocytes isolated from blood leukocytes in PBS- and DNase I–treated Panc02 orthotopic tumor–bearing mice. F, Flow cytometric quantification of CCR2 staining on Ly6C^hiLy6G⁻ monocytes isolated from blood leukocytes of Panc02 tumor–bearing mice compared between PBS- and DNase I–treated cohorts (PBS, n = 7; DNase, n = 8). G, Flow cytometric quantification of CCR2 staining on CD45⁺CD11b⁺F4/80^high macrophages isolated from Panc02 orthotopic tumors (PBS, n = 7; DNase, n = 6). H, Flow cytometric quantification of CCR2 staining on antitumor (CD80⁺) CD45⁺CD11b⁺F4/80^high macrophages isolated from Panc02 orthotopic tumors (PBS, n = 5; DNase, n = 6). I, Flow cytometric quantification of CCR2 staining on protumor (CD206⁺) CD45⁺CD11b⁺F4/80^high macrophages isolated from Panc02 orthotopic tumors (PBS, n = 7; DNase, n = 6). Values are reported as mean ± SD. A Student unpaired t test was used to compare groups. *, P < 0.05; **, P < 0.01; ns, not significant.

As CCL2 is the primary chemokine responsible for the mobilization of monocytes into PDAC (6), we next sought to determine if circulating monocytes are affected by the increase in plasma CCL2 when NETs are degraded. Peripheral blood and bone marrow cells were isolated from DNase I– and PBS-treated mice after IP injection of DNase I to compare monocyte and neutrophil populations. The gating strategy for flow cytometry of blood and bone marrow populations is included in Supplementary Fig. S5. We found that depleting NETs with DNase I significantly increased Ly6C^hiLy6G⁻ monocytes in circulation (Fig. 3B and C), whereas we observed no significant changes in Ly6C^loLy6G⁺ neutrophil populations (Fig. 3B and C) and no difference in CD11b⁺ cells (Supplementary Fig. S6). We found that CCR2⁺ monocytes in the Ly6C^hiLy6G⁻ population were significantly decreased in the DNase I–treated group (Fig. 3E and F), suggesting that monocytes expressing CCR2 are being recruited out of the blood and into tissue—a classical function of monocytes in circulation. We also found a trend toward decreased CCR2⁺ monocytes in the bone marrow Ly6C^hiLy6G⁻ population; however, there were no significant changes in CD11b⁺ cells, Ly6C^hiLy6G⁻ monocytes, or Ly6C⁻Ly6G⁺ neutrophils (Supplementary Fig. S7). To validate that CCR2⁺ monocytes in DNase I–treated mice were being recruited to the tumor, flow cytometry was performed to evaluate CCR2 expression on F4/80⁺ macrophages as CCR2⁺ monocytes are expected to rapidly differentiate into them once they reach the tumor. Although there was no significant difference in the overall percentage of CCR2⁺ tumor macrophages (Fig. 3G), there was a significant increase in the percentage of antitumor macrophages (CD80⁺) that express CCR2 after DNase I treatment (Fig. 3H), suggesting that the CCR2⁺ monocytes recruited to the tumor after DNase I treatment are driven toward an antitumor phenotype. There was no significant difference in the percentage of CCR2⁺ protumor macrophages (CD206⁺) after DNase I treatment (Fig. 3I). Together, these findings point to the activation of the CCL2/CCR2 axis to enhance CCR2⁺ monocyte mobilization and infiltration into the TME in the absence of NETs.

Genetic ablation of NETs also confers increased CCL2 and macrophage infiltration

To corroborate the findings seen after DNase I treatment, we also evaluated tumors from mice with genetic depletion of NETs based on PAD4, which citrullinates histones to allow for unwinding and expulsion from the neutrophil. PAD4^−/− mice, which are deficient in NET formation, expressed little CD80, CD206, and F4/80 in their normal pancreas by immunostaining (Fig. 4A), similar to DNase I–treated mice. There was also a similar trend in increased F4/80 immunostaining in PAD4^−/− mice compared with wild-type (WT) C57BL/6J mice (Fig. 4B and C). Taken together, these findings suggest that removing NETs increases F4/80⁺ macrophages in the PDAC TME. We also noted increased CD80 and decreased CD206 in PAD4^−/− mouse tumors with genetic ablation of NETs compared with WT mice (Fig. 4D–H). Plasma CCL2 levels are similarly increased in PAD4^−/− mice compared with WT cohorts (Fig. 4I). These findings validate the effects seen after NET inhibition with DNase I treatment in a genetic model.

Figure 4.

Macrophage staining and CCL2 measurements in the PAD4^−/− murine model of genetic ablation of NETs. A, Representative images of CD80, CD206, and F4/80 immunofluorescent staining in ealthy (nontumor-bearing) PAD4^−/− pancreas. B, Representative F4/80 immunofluorescent staining of Panc02 orthotopic tumors in PAD4^−/− and WT mice. C, Quantification of F4/80 immunofluorescent staining of Panc02 orthotopic tumors between WT (n = 5) and PAD4^−/− (n = 6) cohorts. D, Representative CD80 immunofluorescent staining of Panc02 orthotopic tumors in PAD4^−/− and WT mice. E, Quantification of CD80 immunofluorescent staining in Panc02 orthotopic tumors between WT (n = 4) and PAD4^−/− (n = 6) mice. F, Flow cytometric quantification of CD80 geometric mean fluorescence intensity (gMFI) gated on CD45⁺CD11b⁺F4/80^high macrophages (PAD4^−/−, n = 6; WT, n = 8) isolated from Panc02 orthotopic tumors. G, Representative CD206 immunofluorescent staining of Panc02 orthotopic tumors in PAD4^−/− and WT mice. H, Quantification of CD206 immunofluorescent staining in Panc02 orthotopic tumors between WT (n = 5) and PAD4^−/− (n = 5) cohorts. I, CCL2 measured via ELISA in the plasma of PAD4^−/− (n = 3) and WT (n = 3) Panc02 orthotopic mice. Values are reported as mean ± SD. A Student unpaired t test was used to compare groups. *, P < 0.05; **, P < 0.01. KO, konckout.

Tumor cells and macrophages release CCL2 in response to IFNγ stimulus

Given our findings of increased activity of the CCL2/CCR2 axis in the NET-deficient PDAC TME, we sought to uncover a predominant source of CCL2 from the tumor. Previous studies have indicated that pancreatic tumor cells are the main source of CCL2 (6, 41, 44, 45). Thus, we measured the release of CCL2 from the Panc02 murine cell line in culture. In the absence of a stimulus, little CCL2 was released from the Panc02 murine tumor cell line used to generate the PDAC tumors in vivo. However, the addition of the proinflammatory cytokine IFNγ (45), which is elevated in NET-deficient mice (Fig. 2G), increased the release of CCL2 from the tumor cells (Fig. 5A). Another potential source of CCL2 in the TME is macrophages (21, 46). Macrophages are one of the most abundant immune cells in solid tumors, including PDAC (47). To test macrophages as a major source of CCL2 in our model, we measured CCL2 released from BMDMs. Although untreated quiescent BMDMs released some CCL2, there was a significantly higher amount of CCL2 released in a dose-dependent manner in response to IFNγ (Fig. 5B). To determine if NETs specifically modulate the release of CCL2 from tumor cells and macrophages, we treated the Panc02 cell line and BMDMs with IFNγ and NETs. Panc02 cells showed no synergistic effect for CCL2 release when IFNγ and NETs were added compared with IFNγ alone (Supplementary Fig. S8). We noted a modest increase in CCL2 release from BMDMs treated with NETs and IFNγ compared with IFNγ alone; however, NETs alone did not stimulate CCL2 release from the BMDMs or Panc02 cell line (Supplementary Fig. S8). These findings suggest that NETs modulate CCL2 release by increasing IFNγ and not through direct induction of release from tumor cells or BMDMs. Taken together, these findings point to a possible feed-forward mechanism in which removing NETs increases CCL2 release from tumor cells and macrophages through IFNγ, which can recruit additional macrophages into the tumor (48).

Figure 5.

CCL2 is released from tumor cells and macrophages. A, Quantification of CCL2 via ELISA measured in the supernatant of the Panc02 murine tumor cell line(n = 4 per group) with IFNγ treatment. Graphs represent one of three to four independent replicates. Mean ± SD is reported. Ordinary one-way ANOVA was usedto compare groups with control (0 ng/mL). B, Quantification of CCL2 via ELISA measured in the supernatant of BMDMs (n = 4 per group) with IFNγ treatment. Graphs represent one of three to four independent replicates. Mean ± SD is reported. Ordinary one-way ANOVA was used to compare groups with control (0 ng/mL). C, CCL2 gene expression in immune cells (B cells, CD4 T cells, CD8 T cells, NK cells, and macrophages) identified by the EPIC deconvolution matrix in GEPIA2021. F and P values are calculated using GEPIA2021. D, CCL2 gene expression in M1 macrophages, M2 macrophages, and monocytes identified by the quantiSeq deconvolution matrix in GEPIA2021. F and P values are calculated using GEPIA2021. TPM, transcripts per million.

CCL2 Is Produced by M1-Like Macrophages in Human PDAC

As we determined that BMDMs can secrete higher levels of CCL2 in the presence of the proinflammatory cytokine IFNγ, we explored CCL2 mRNA expression by immune cells in human PDAC samples using GEPIA2021 (22) analysis of the TCGA-PAAD dataset to substantiate our preclinical findings. Utilizing the EPIC deconvolution matrix in GEPIA2021, we compared CCL2 expression in B cells, CD4 T cells, CD8 T cells, NK cells, and macrophages (Fig. 5C). We found that CCL2 was expressed by macrophages in the PAAD tumor and GTEx normal pancreas tissue. Next, we used the quantiSeq deconvolution matrix in GEPIA2021 to stratify M1 macrophages, M2 macrophages, and monocytes in the TCGA-PAAD dataset. We found that CCL2 expression was highest in the M1 macrophage population in the PAAD tumor and GTEx pancreas tissue (Fig. 5D), corroborating our in vitro findings that CCL2 is released from macrophages in the presence of the proinflammatory cytokine IFNγ (Fig. 5B). Furthermore, we used the GEPIA2021 tool to determine the possible source of IFNγ in the PDAC TME. Using the EPIC deconvolution matrix, we determined that IFNG was most highly expressed by CD4 and CD8 T cells (Supplementary Fig. S9). These findings suggest interactions among NETs, macrophages, and T cells in the PDAC TME, which warrants further exploration (Fig. 6).

Figure 6.

Removing NETs from the PDAC TME promotes an antitumor immune response. The summary figure represents the suggested interactions between NETs and the recruitment of macrophages in the PDAC TME. The presence of NETs in the PDAC TME decreases CCL2 in circulation, thereby promoting a protumor immune response in macrophages. In contrast, removing NETs from the PDAC TME promotes an antitumor immune response in recruited monocytes driven by CCL2 and IFNγ.

Discussion

In the present study, we explore the relationship between NETs and macrophages in the setting of PDAC. The PDAC TME is known to be uniquely fibrotic, hypoxic, and immunosuppressive. Although many studies focus on how the innate immune system can influence the adaptive immune system in PDAC (6, 49), the interaction between innate immune cells is understudied (15). As neutrophils and monocytes have a reciprocal relationship in PDAC and other cancer types (15, 17), we sought to determine how NETs can influence macrophages in PDAC. First, we examined NET markers and macrophage markers in patients with PDAC compared with those with noncancerous pathologies. We show that the monocyte chemokine CCL2 is elevated in the plasma of patients with PDAC. Using intratumoral staining, we also show higher pan-macrophage marker CD68 and NET marker CitH3 in PDAC. Overall, these findings validate a role for NETs and macrophages in PDAC progression that has been previously shown (6, 13, 15).

Despite several studies trying to elucidate the interactions between NETs and macrophages (50–54), the relationship between these two cell types is still not fully understood. For example, monocyte-derived macrophages have been shown to clear NETs without an increase in cytokine release in an in vitro setting. DNase I degradation of the NETs promoted macrophage uptake and lysosomal degradation of the NETs (50). However, NETs in sepsis can cause caspase-1–driven pyroptosis (cell death) in macrophages through the receptor for advanced glycation end products signaling pathway (53). Furthermore, NETs can cleave integrin receptors from the cell surface of macrophages, preventing efferocytosis (uptake of dead and dying cells) in an extracellular cold-inducible RNA-binding protein model of sepsis (52). Of particular interest is how NETs influence macrophages in wound healing, as cancer is commonly described as “a wound that never heals.” In a cardiac healing model of acute myocardial infarction, NETs promoted an M2-like wound healing response with increased IL10 and decreased IL6 and TNFα expression in macrophages (51), whereas in a diabetic atherosclerosis model, NET⁺ staining areas of the plaque demonstrated more immunostimulatory phenotypic characteristics (54). Thus, the relationship between NETs and macrophages deserves additional attention, especially in the setting of PDAC, in which both NETs and macrophages are elevated (18).

To determine how NETs can influence macrophages in the PDAC TME, we used IP DNase I to remove NETs from the PDAC TME and validated these results with data from genetic ablation of NETs (PAD4^−/−). When NETs are removed, cells expressing the pan-macrophage marker F4/80 are increased, suggesting elevated macrophage numbers. Macrophages infiltrating the PDAC TME have been shown to promote fibrosis (29), immunosuppression (30), and metastasis (31, 32), so we sought to determine the effects of removing NETs on macrophage phenotypes. In the absence of NETs, expression of the antitumor macrophage marker CD80 is increased, whereas expression of the macrophage marker CD206, generally protumor, is decreased. Although our current dataset does not include costaining for CD80/CD206 and macrophage markers, these markers are most frequently expressed on macrophages in the pancreas. Additionally, to combat this limitation, we performed flow cytometry so that the expression of CD80/CD206/CCR2 could be analyzed specifically in F4/80 high macrophage populations. Previous studies have shown that CD206-expressing TAMs release TGFβ and promote epithelial-to-mesenchymal transition and migration of pancreatic tumor cells (31). In contrast, CD80 is a costimulatory molecule that binds to CD28 on T cells and promotes T-cell activation (33), indicating that in the absence of NETs, the macrophage tumor immune phenotype may be more antitumor (55, 56). The proinflammatory cytokine IFNγ, which is known to promote polarization of macrophages to an M1-like, antitumor phenotype (57), also increased without NETs in the TME. Our analysis of the TCGA-PAAD database suggests that IFNγ may be produced by CD4 and CD8 T cells in the PDAC tumor, which may indicate an interplay among NETs, macrophages, and T cells in PDAC. Although the immune microenvironment in PDAC may be primed to be immunostimulatory when NETs are removed, additional checkpoint blockade molecules may be needed to improve T-cell function (58). Thus, these findings suggest that removing NETs may represent an option for promoting an antitumor innate immune response and would be interesting to test in combination with checkpoint immunotherapy.

Given the previous lackluster performance of CCR2 antagonism in clinical trials (14, 42), we next sought to measure the activity of the CCL2/CCR2 axis in the absence of NETs. When NETs are removed, there is a corresponding increase in CCL2 and inflammatory monocyte numbers in the circulation of PDAC tumor–bearing mice. Previously, a similar result was seen when inhibition of neutrophil CXCR2 via a CXCR2 neutralizing antibody increased CCR2⁺ monocyte mobilization (15). Earlier clinical trials utilizing CCR2 antagonism with standard-of-care folate, 5-fluorouracil, oxaliplatin, and irinotecan (14) and gemcitabine (42) resulted in modest tumor response or pulmonary toxicity, respectively. In the context of the present study, inhibition of the CCL2/CCR2 axis alone may not provide the optimal antitumor response as NETs enhance both the CCL2/CCR2 axis and macrophage phenotype. Similar to previous reports, we demonstrate that both tumor cells (6, 41, 44, 45) and macrophages (21, 46) can be a source of CCL2, particularly in the presence of the cytokine IFNγ (45, 59). CCL2 and IFNγ together promote an antitumor macrophage phenotype in PDAC when a CD40 agonist is administered in PDAC murine models (46). Our findings point to a feedforward mechanism with the potential to stimulate an antitumor immune response, in which removing NETs allows for the recruitment of monocytes via the CCL2/CCR2 axis, which can in turn release additional CCL2 to recruit additional monocytes.

The current study has potential limitations. In this study, the effect of depleting NETs is likely multifactorial, and all downstream impacts of NET inhibition are likely not included here, making mechanistic studies challenging. We demonstrate an increase in IFNγ in the absence of NETs. Although we also show a correlation between IFNγ stimulation and CCL2 release from BMDMs and Panc02 cells, the in vitro setting does not fully replicate the TME, which can include additional factors such as hypoxia and a broad cytokine milieu. We have not definitively isolated the source of increased IFNγ, although our analysis of available data with GEPIA21 suggests that T cells are the main source. However, CD80⁺ macrophages are known to release more IFNγ in response to stimuli, so these, as well as tumor cells, are possible additional sources (60). Our data strongly suggest that removing NETs increases IFNγ and therefore CCL2; however, additional studies of the pancreatic tumor cells and macrophages in the TME are needed to further validate these findings. Although an antitumor phenotype is demonstrated by canonical markers using FL-IHC staining, the functional characteristics of these cells have not been otherwise elucidated. The literature suggests that these M1- and M2-like macrophage markers are strongly associated with functional phenotype (58, 60, 61, 62, 63, 64); however, we have not validated these findings in the current article. Future studies will include a more comprehensive characterization of macrophage function, including cytokine production and T-cell stimulation/suppression (6, 15, 65), angiogenesis (44), and collagen deposition (29) within the PDAC TME with and without NETs. Furthermore, CD80, an M1-like marker, has also been shown to limit CD4 T-cell infiltration via binding to CTLA-4 in PDAC, a mechanism that was not thoroughly explored in the present study (58). In addition to characterizing macrophages within the TME, this study explores the effect of NETs on the recruitment of circulating monocytes via the CCL2/CCR2 axis but does not demonstrate CCR2⁺ monocyte staining within the PDAC tumor, as CCR2 expression can be lost rapidly during monocyte maturation (66), making quantification of CCR2⁺ monocyte infiltration challenging to capture. Instead, we used flow cytometry to detect CCR2⁺ macrophages within the PDAC TME. Although these CCR2⁺ macrophages in the TME are likely a product of CCR2⁺ monocyte recruitment and differentiation into macrophages, they remain a surrogate marker of CCR2⁺ monocyte recruitment. Complicating this further, we have previously shown that CCL2 knockout mice upregulate other pathways of macrophage recruitment, including CCL5 and CCL7, when CCL2 is targeted, suggesting that compensatory pathways may be activated to regulate macrophage trafficking in these mice that would confound our results (21). Additionally, the presented data suggest that the elevated number of circulating monocytes may represent a source of macrophages, but we did not explore yolk sac–derived tissue-resident macrophages, which are a known source of macrophages in PDAC (29, 46, 67). This study also does not explore CCL2/CCR2 recruitment of dendritic cells (68), myeloid-derived suppressor cells (42), and regulatory T cells (69), which have also been shown to be recruited via this axis although to a lesser extent. Of these immune cell populations, both dendritic cells and macrophages may express CD80 (58). Such considerations may be important for teasing apart how NETs can influence myeloid subpopulations in the PDAC TME. Additionally, multiple direct and indirect mechanisms after NET inhibition that alter the PDAC immune microenvironment are likely at play. It is challenging to separate the effects of NETs directly on T cells (49) from NET-induced changes to the immune milieu, as outlined in this article, that would promote antitumor T-cell function. Our group has shown that NETs can activate pancreatic stellate cells (11), which are critical drivers in TME immunosuppression (65, 70). Reversal of NET-induced pancreatic stellate cell activation is another potential indirect mechanism that could explain the findings in the current article. Nonetheless, the current study provides a mechanism through which NET degradation limits tumor growth and promotes antitumor immunity in preclinical PDAC.

These results support previous findings that NETs are detrimental in PDAC and further reveal that depletion of NETs increases macrophages within the PDAC TME. NET depletion also promotes an antitumor phenotype and activates the CCL2/CCR2 axis. These data support NET removal as a method for modulating macrophages in the PDAC TME, which may represent a therapeutic strategy for patients with PDAC.

Supplementary Material

Supplementary data for this article are available at Cancer Immunology Research Online (http://cancerimmunolres.aacrjournals.org/).

Data availability

The data generated in this study are available from the corresponding author (B.A. Boone) upon request.