Blocking M2-Like Macrophage Polarization Using Decoy Oligodeoxynucleotide-Based Gene Therapy Prevents Immune Evasion for Pancreatic Cancer Treatment

The tumor microenvironment (TME) is a dynamic and complex milieu surrounding the tumor, which plays a critical role in regulating cancer progression and affecting therapeutic outcomes (1). The TME contains not only tumor cells but also diverse cell types, such as Th1, Th2, B lymphocytes, tumor-associated macrophages (TAMs), dendritic cells, natural killer (NK) cells, neutrophils, myeloid-derived suppressor cells, and fibroblasts (2). Among them, TAMs represent the prominent immune cell type, capable of suppressing the innate and adaptive immunity, assisting tumor cells in evading immune attack, and supporting tumors inclined toward cold tumors (3, 4). In pancreatic cancer, TAMs are the most abundant immune cells infiltrating the TME, which orchestrate tumor-promoting inflammation, and have been associated with a poor prognosis (5, 6).

Past studies have shown that macrophages are a diverse group of immune cells with different properties and functions that may originate from distinct cell differentiation pathways and environments (7). The macrophage plasticity in different states of macrophage activation has been proposed, like M1-like, M2a-alternative, M2b-type 2, M2c-deactivated, and M2d (8). Furthermore, it is well known that M1-like macrophages with antitumor activity produce inflammatory cytokines such as IFNγ, CCL2, IL1β, IL12p40, and reactive nitrogen and oxygen species (RNS and ROS) to promote Th1 responses and increase antimicrobial ability against various types of bacteria and viruses (9, 10). Macrophages with M2-like phenotype exhibit protumor activity and release TGFβ, VEGFA, CCL17, and CCL21 to exert anti-inflammatory effects (9, 10). Considerable evidence has indicated that M2-like TAMs exert immunosuppressive effects by secreting various chemokines, cytokines, and enzymes to downregulate immune cells activation (11). For example, M2-like TAMs can secret IL10 to activate regulatory T cells that suppress the killing effect of T and NK cells (12). In addition to soluble inhibitory molecules, M2-like TAMs reportedly express well-known immune checkpoint proteins to negatively regulate immune responses, such as classical and nonclassical MHC class I molecules, cytotoxic T-lymphocyte antigen 4 ligands, and programmed cell death protein ligand 1 (PD-L1; refs. 13, 14). Therefore, blocking of immune checkpoints is considered a promising approach for activating therapeutic antitumor immunity (15).

Reactive oxygen species (ROS) are the most common oxygen radicals in the body, such as superoxide, hydroxide, and non-radicals (16). Much research has shown that ROS, specifically produced by the NADPH oxidase complex (NOX, also called Cybb or gp91phox), are a critical factor impacting host defense and cell differentiation in multicellular organisms (17). NOX2 is the major isoform of NOX in phagocytes and dendritic cells. NOX2, p22phox, p67phox, p47phox, p40phox, and Rac can form active enzyme complexes to produce ROS. Among the six subunits, gp91phox and p22phox are membrane proteins mainly constituting the flavin cytochrome b558 (cyt b558) and the catalytic core (17, 18). A recent report showed that an antioxidant drug, butylated hydroxyanisole, efficiently suppressed monocyte–macrophage differentiation, the occurrence, and tumorigenesis in a mouse cancer model, suggesting that ROS are essential for the formation (19).

To date, targeting the signaling pathways such as CCL2-CCR2, CSF1-CSF1R, CD47-SIRPα, and CXCL12-CXCR4 has therapeutic potential to limit monocyte recruitment, reduce the activation of M2-like TAMs, and reverse the M2-like phenotype to the antitumor M1-like phenotype (10, 20–22); however, the systemic adverse effects of these treatments have been a challenging issue. For example, blocking CCL2-CCR2 signaling that drives the recruitment of circulating inflammatory monocytes to inflammatory centers can reduce TAM population at primary and metastatic sites (21, 23, 24). However, a previous clinical trial for pancreatic cancer has shown that treatment with the small molecule CCR2 inhibitor, PF-04136309 decreased CD14⁺CCR2⁺ inflammatory monocytes in the peripheral blood and TAMs formation in the TME, but this was accompanied by severe side effects such as pulmonary toxicity, dysesthesia, diarrhea, and hypokalemia (23, 25). To solve this problem, using double-stranded “decoy” oligonucleotides to target the action of transcription factors (TFs) is a promising alternative approach because of their good tolerance and low toxicity (26–28). Therefore, the objective of this study was to investigate the mechanism underlying M2 polarization and design decoy oligodeoxynucleotide (ODN)-based gene therapy to reduce M2-like formation and reactivate antitumor immunity for pancreatic cancer treatment.

Diphenyleneiodonium, 8λ³-iodatricyclo[7.4.0.0²,⁷]trideca-1(13),2,4,6,9,11-hexaen-8-ylium chloride (DPI; Sigma-Aldrich; D2926; ref. 29) was used to treat cells at a concentration of 5 μmol/L for 24 hours. Lipopolysaccharide (LPS; Sigma-Aldrich; L2387; ref. 30) was used to treat cells at a concentration of 100 nmol/L for 24 hours. IL4-neutralizing antibodies (GeneTex; GTX15760; clone name: 25D2; RRID: AB_3099661) were used at 1 mg/mL for 72 hours (31). IL4 recombinant protein (proteintech; Cat No: HZ-1004) was used at a concentration of 20 ng/mL at various time points (32). The following concentrations of drugs were used to pretreat cells for 24 hours: Ruxolitinib, (R)-3-(4-(7H-pyrrolo[2,3-d]pyrimidin-4-yl)-1H-pyrazol-1-yl)-3-cyclopentylpropanenitrile (Rux; Cat No: S1378; ref. 33) at 10 μmol/L; Tofacitinib, 3-[(3R,4R)-4-methyl-3-[methyl(7H-pyrrolo[2,3-d]pyrimidin-4-yl)amino]piperidin-1-yl]-3-oxopropanenitrile (Tof; Cat No: S2789; ref. 34) at 50 μmol/L; AS1517499, 4-(benzylamino)-2-[2-(3-chloro-4-hydroxyphenyl)ethylamino]pyrimidine-5-carboxamide (AS; Cat No: S8685; ref. 35) at 250 nmol/L; Trichostatin A, (2E,4E,6R)-7-[4-(dimethylamino)phenyl]-N-hydroxy-4,6-dimethyl-7-oxohepta-2,4-dienamide (TSA; Cat No: S1045; ref. 36) at 10 μmol/L; and LBH589, (E)-N-hydroxy-3-[4-[[2-(2-methyl-1H-indol-3-yl)ethylamino]methyl]phenyl]prop-2-enamide (LBH; Cat No: S11030; ref. 37) at 50 μmol/L. All drugs were purchased from Selleckchem.

The human pancreatic cancer cell lines AsPC-1 (Catalog No. CRL-1682; RRID: CVCL_0152), BxPC3 (Catalog No. CRL-1687; RRID: CVCL_0186), PANC-1 (Catalog No. CRL-1469; RRID: CVCL_0480), and MiaPaCa2 (Catalog No. CRL-1420; RRID: CVCL_0428), and the monocytic U937 (Catalog No. CRL-1593.2; RRID: CVCL_0007), and THP-1 (Catalog No. TIB-202; RRID: CVCL_0006) cell lines were obtained from the American Type Culture Collection (ATCC) and have been validated by the Center for Genomic Medicine, National Cheng Kung University (NCKU), using the ATCC short tandem repeat database. The mouse pancreatic cancer cell line KPC was established from Pdx1-Cre; Kras^G12D/+; Trp53^-/- mice kindly gifted by Dr. Kuang-Hung Cheng (Institute of Biomedical Sciences, National Sun Yat-sen University, Kaohsiung, Taiwan). KPC cells were transfected with the pGL4 Luciferase Reporter plasmid to establish a stable luciferase-expressing clone, KPC^LUC. All cells were cultured in RPMI 1640 medium (Hyclone; SH30027.02) supplemented with 10% fetal bovine serum (Hyclone; SH30071.03), 2 mmol/L L-glutamine (Simply; Cat. No CC515-0100), 100 units/mL Antibiotic-Antimycotic (Simply; Cat. No CC501-0100), and 1 mol/L HEPES (Simply; Cat. No CC519-0100) at 37°C in a humidified atmosphere of 5% CO₂. All cell lines were routinely detected for the presence of mycoplasma using the Mycoplasma Detection Kit (BioSmart, BSMP-101) performed every 6 months.

We collected U937 cells after 72 hours of LPS treatment or 72 hours of coculture with PANC-1 cells for microarray analysis and submitted them to the Gene Expression Omnibus database (GSE269815). The gene microarray data were analyzed using the Agilent Human whole genome oligo 4*44k chip (array kit serial number: US00082833) and Agilent’s gene expression microarray software, including Feature Extraction and GeneSpring GX. Feature Extraction software extracts data from microarray images and provides workflows for analysis. GeneSpring GX offers statistical tools for data analysis and visualization (Welgene Biotech), and the genes of interest were drawn as a heatmap.

U937 cells were seeded into cell culture transwell inserts (3 × 10⁵ cells per insert) with pore size of 0.4 μm (Falcon) packed in a six-well plate (GeneDireX), where PANC-1 cells were grown (3 × 10⁵ cells per well). The cocultures were incubated at 37°C for 72 hours.

Bone marrow cell suspensions were separated by flushing femurs and tibias of 8 to 12 weeks old wild-type mice with complete RPMI1640 media. Pipetting dislodged aggregates, and debris was removed using a 70 μm nylon web. Cells were washed using HBSS, adjusted to a 10⁶ cells/mL suspension, and seeded on 10 cm low attachment surface dishes. Cells were treated with 20 ng/μL rmM-CSF (PeproTech, 315-03) and maintained in a humidified incubator at 37°C and 5% CO₂.

RNA sequencing (RNA-seq) of 100 pancreatic tumors was conducted and the data were analyzed by Allbio Life Co., Ltd. We evaluated the correlation between CYBB expression and the M2 (CD204)/M1 (CD86) ratio using their FPKM values.

The experimental protocols and procedures for the animal study were approved by the Institutional Animal Care and Use Committee of NCKU (IACUC No.: 103150 and 110331). Luciferase-expressing KPC cells (KPC^LUC, 1 × 10⁷ cells in 50 μL serum-free medium) were orthotopically injected into the pancreas of 8 to 12 weeks old male C57BL/6 mice. Two weeks after the cancer cell injection, the tumor-bearing mice were treated with decoy ODNs (1 μg/mL in normal saline, intravenously) once weekly for 3 weeks.

Paraffin-embedded sections were deparaffinized and antigens were unmasked by autoclaving at 121°C for 10 minutes in sodium citrate buffer (10 mmol/L; pH 6.0). Sections were incubated with primary antibodies at 4°C overnight, followed by incubation with secondary antibodies at room temperature for 30 minutes. Immunoreactivity was visualized using the DAB chromogen system (DAKO). Information regarding the antibodies and their dilution ratios used in this study is shown in Supplementary Table S1.

After approval from the Institutional Review Board of NCKU Hospital, a pancreatic cancer tissue microarray (TMA) was constructed from paraffin-embedded blocks of 187 patients with pancreatic cancer resected at NCKU Hospital. TMA paraffin blocks were cut into 5 μm sections. TMA sections were stained with primary antibodies at 4°C overnight followed by incubation with fluorescent-conjugated secondary antibodies at room temperature for 1 hour. The nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI). Panoramic images of immunofluorescence (IF)-stained TMAs were acquired using the FACS-like Tissue Cytometry System (Tissue Gnostics, Vienna, Austria). The percentage of stained cells was quantified using the TissueQuest system software TissueFAX (Tissue Gnostics, Vienna, Austria). Information regarding the antibodies and their dilution ratios used in this study is shown in Supplementary Table S1.

Total RNA was isolated from the cultured cells using TRIzol (Invitrogen). The purity and concentration of the RNA were determined using a Nanodrop spectrophotometer (Thermo Fisher Scientific; ND-1000). Total RNA from each sample was reverse transcribed into 2 μg cDNA using Deoxyt HiSpec Reverse Transcriptase (Yeastern Biotech). Quantitative assessment of mRNA levels was performed by qPCR using KAPA qPCR SYBR Green Master Mix (KAPA) on a StepOne Real-Time PCR System (Applied Biosystems). The reaction conditions were as follows: initial denaturation at 95°C for 5 minutes; 40 cycles of denaturation at 95°C for 10 seconds; and annealing at 60°C for 10 seconds. mRNA expression was determined using the 2^−ΔΔCt method (fold difference) and normalized to that of GAPDH. The sequences of the human and mouse primers used in this study are listed in Supplementary Table S2.

STAT6, HDAC2, and nontarget short hairpin RNA (shRNA) vectors were purchased from the National RNAi Core Facility, Academia Sinica, Taipei, Taiwan. Lentiviral transfection of U937 cells with shRNAs was performed in the presence of 8 μg/mL polybrene (Sigma-Aldrich, AL-118). Puromycin (P9620; Sigma-Aldrich) was used to select permanent cell lines.

The harvested cells were washed twice with PBS and lysed in ice-cold radioimmunoprecipitation assay (RIPA) lysis buffer (Millipore) for 30 minutes. Lysates were cleared by centrifugation at 14,000 rpm for 10 minutes at 4°C, and the protein concentration was measured using the Bradford assay (Bio-Rad Laboratories, 500-0006). For Western blot analysis, cell lysates were boiled for 5 minutes in sample buffer before being resolved on SDS-polyacrylamide gels. After separation, the proteins were transferred to the PVDF membranes (Millipore, IPVH00010). The membrane was blocked with 5% nonfat dried milk in TBST buffer (20 mmol/L Tris-HCL, pH 7.4, 150 mmol/L NaCl, 0.1% Tween 20) for 1 hour and then incubated overnight at 4°C with specific primary antibodies. Subsequently, the membrane was washed with TBST and incubated with horseradish peroxidase–conjugated secondary antibodies (LEADGENE) for 1 hour at room temperature. The blot signals were developed using an enhanced chemiluminescence kit (GeneTex, GTX14698) and captured using the iBright 1000 Series Image System (Thermo Fisher). Information regarding the antibodies and their dilution ratios used in this study is shown in Supplementary Table S1.

According to the manufacturer’s protocol, chromatin immunoprecipitation (ChIP) was performed using the EZ-ChIP Chromatin Immunoprecipitation Kit (Millipore). Briefly, the cells were fixed with 1% formaldehyde for 15 minutes, and unreacted formaldehyde was quenched by adding 125 mmol/L glycine for 15 minutes. Chromatin was sonicated to an average length of 200 to 1,000 bp. Samples were reacted with anti-STAT6 (1:50, Cell Signaling Technology, #5397S) and anti-HDAC2 (1:50, Cell Signaling Technology, #57156S). The DNA sample precipitated with the target antibody was detected by PCR using the primers covering the −896/−719 region of the CYBB promoter: forward 5′- TGA​CAC​AAT​CTC​GGC​TCA​CTG​CAA-3′ and reverse 5′-TCA​CGC​CTG​TAA​TCC​CAG​CAC​TTT-3′.

The CYBB-WT– and CYBB-mutated reporter constructs were generated by standard gene synthesis and cloning of the promoter sequences in the luciferase reporter system (pGL4.17 vector, Promega). U937 cells were transfected with the above reporter plasmids using electroporation (NEPA21 electroporator). Cells (1 × 10⁶) were suspended in OPTI-MEM (100 μL) and mixed with 10 μg plasmid DNA in an electrode chamber. Twenty-four hours after transfection, cells were stimulated with or without IL4 for 8 hours, and then were harvested and lysed for luciferase and Renilla measurements using the Dual-Luciferase Assay System (Promega). To assess transfection efficiency, cotransfection with the pRL-TK vector as an internal control allowed normalization of transfection by Renilla luciferase activity.

The decoy ODNs were synthesized by Genomics (New Taipei City, Taiwan). STAT6-CYBB decoy ODNs, mutated decoy ODNs, and scrambled ODNs were synthesized using selected sequence targets (38, 39). The STAT6-CYBB decoy ODN is a double-stranded phosphonothioate 28mer DNA that acts as an antagonistic inhibitor of the TF STAT6 and binding site of consensus sequence is underlined. We designed the sequences to act as a signal-strand DNA as follows: STAT6-CYBB decoy ODN, 5′-CTG​ACT​CCC​AGGTTC​AAG​TGA​TTCT​CCT-3′ and 3′ -GAC​TGA​GGG​TCCAAG​TTC​ACT​AAGA​GGA-5′; mutated decoy ODN, 5′-CTG​ACT​CCC​GAG​TTC​AAG​TGA​TTC​CCC​T-3′ and 3′-GAC​TGA​GGG​CTC​AAG​TTC​ACT​AAG​GGG​A-5′; scrambled decoy ODN, 5′-CGA​AAA​TTC​GTT​AAA​TCA​CTA​GCT​TAC​C-3′ and 3′-GCT​TTT​AAG​CAA​TTT​AGT​GAT​CGA​ATG​G-5′. Synthetic ODNs were dissolved in sterile annealing buffer (100 mmol/L Tris-HCl pH 7.5, 10 mmol/L EDTA pH 8.0, 1 mol/L NaCl, and ddH₂O) and were annealed for 3 hours during which the temperature was reduced from 90°C to 25°C. After annealing, the reaction mixture was maintained at 4°C.

Each experiment was performed at least three times independently under identical conditions, and the data are expressed as the mean ± standard error of the mean (SEM). Student t test and one-way ANOVA were used to analyze differences. Statistical analyses were performed using GraphPad Prism 9.0 (GraphPad Software Inc., La Jolla, CA, USA) and SigmaPlot 12.0 (Systat Software Inc.). Statistical probability (P) is expressed as ^∗∗∗, P < 0.001; ^∗∗, P < 0.01; ^∗, P < 0.05.

The authors confirm that the data supporting the findings of this study are available in the article and the supplementary information. Raw data supporting the findings of this study are available for the corresponding author upon reasonable request.

Previous studies have shown that monocytic U937 cells can polarize to M2-like macrophages in response to tumor-derived factors (40, 41). Using gene expression microarray analysis, we have identified the potential genes that contribute to macrophage polarization. Compared to LPS treatment that induced M1-like polarization as a control, 72 hours of coculture with PANC-1 cells led to significantly increased expression of M2-like macrophage markers but decreased expression of M1-like macrophage markers and ROS-associated CYBB in U937 cells (Fig. 1A). The microarray data were confirmed by qPCR (Fig. 1B). As CYBB expression was downregulated in M2-like macrophages, we sought to determine the clinical significance of CYBB. We assessed the correlation between CYBB expression and the M2-like/M1-like ratio from our RNA-seq data (FPKM values) of 100 pancreatic tumors. A significantly negative correlation between CYBB expression and the M2-like/M1-like ratio was observed (r = −0.269, P = 0.0068; Fig. 1C), and the clinicopathological characteristics of these patients are shown in Supplementary Table S3. Next, we performed IF staining for CYBB on 180 human pancreatic cancer TMAs. The clinicopathological characteristics of these patients are shown in Supplementary Table S4. After quantification of the area percentages of staining, we found that pancreatic tumor tissues had lower CYBB expression than unpaired or paired normal tissues (Fig. 1D). We further divided the patients into the CYBB^Low group (staining percentage <50%) and the CYBB^High group (staining percentage ≥50%, Supplementary Fig. S1A). The CYBB^Low group showed significantly worse overall survival (OS; P = 0.002, Supplementary Fig. S1B) and disease-free survival (DFS; P < 0.001, Supplementary Fig. S1C) than the CYBB^High group, consistent with the data showing that patients with high CD204 M2-like expression also had shorter OS (P = 0.057, Supplementary Fig. S1D) and DFS (P = 0.217, Supplementary Fig. S1E). Furthermore, the combination of CYBB^Low and CD204^High M2-like conferred the worst OS (P < 0.001, Fig. 1E) and DFS (P = 0.018, Fig. 1F). Taken together, these results indicate that CYBB is a key regulator of M2-like macrophages polarization and is associated with worst patient survival.

IL4 can promote the proliferation and survival of cancer cells and is a key inducer of M2-like polarization (42). During the coculture with U937 cells, expression of IL4 mRNA was increased in various pancreatic cancer cell lines (Fig. 1G), and the expression of IL4 protein was increased in U937 cells cocultured with PANC-1 cells at a time-dependent manner (Supplementary Fig. S2). However, at 72 hours after coculture, blockade of IL4 signaling with neutralizing antibodies reversed the downregulation of CD86 and CYBB mRNA levels but suppressed CD204, MRC1 (CD206), and ARG1 mRNA expression in U937 cells (Fig. 1H; Supplementary Fig. S3). Similarly, CD86 protein expression was increased but CD204 protein expression was reduced by addition of IL4 neutralizing antibodies (Fig. 1I). These data indicated the importance of IL4 for M2-like polarization. Furthermore, we also found that CYBB mRNA expression was decreased by IL4 treatment in both U937 cells and mouse BMDMs (Fig. 1J), suggesting that CYBB expression can be regulated by IL4. As STAT6 is a key mediator of macrophage polarization and a major downstream effector of the IL4 pathway, we next investigated the involvement of STAT6 in IL4-mediated CYBB downregulation. After treatment with IL4, STAT6 and its upstream kinases JAK1 and JAK3 were phosphorylated, followed by a decrease in CYBB expression at 8 to 24 hours (Fig. 2A). As ROS have been reported to play an important role in TAMs differentiation, we also examined whether ROS contribute to the effect of IL4. Interestingly, after pretreatment with DPI (43), a flavoenzyme inhibitor that inhibited CYBB activity, to attenuate ROS production, IL4-induced phosphorylation of JAK1, JAK3, and STAT6 in U937 cells was significantly suppressed (Fig. 2B), suggesting that ROS produced by CYBB are required for IL4-mediated M2 polarization. In addition, pharmacological inhibition of JAK1, JAK3, and STAT6 restored IL4-mediated downregulation of CYBB, suggesting that IL4 reduces CYBB expression to promote M2-like polarization through STAT6 activation (Fig. 2C). To study whether ROS alone are sufficient to induce M2-like polarization, we added hydrogen peroxide (H₂O₂) to generate ROS in U937 cells. After treatment with H₂O_2, the phosphorylation of JAK1, JAK3, and STAT6 and expression of CYBB did not significantly change (Supplementary Fig. S4). However, cotreatment with IL4 and H₂O₂ could significantly induce CD204 expression and the phosphorylation of JAK1, JAK3, and STAT6 (Fig. 2D). Taken together, these results indicate that ROS plays an essential role in regulating the activation of the IL4/JAK/STAT6 pathway.

The epigenetic regulator HDAC modulates inflammatory responses in the TME and affects the polarization of M1-like/M2-like macrophages (44–46). In our microarray data, we found that HDAC2 expression was increased in M2-like TAMs (Fig. 2E), suggesting that HDAC2 plays a role in M2-like polarization. Coculture with pancreatic cancer cells increased HDAC2 mRNA expression in U937 cells and mouse BMDMs (Fig. 2F). Treatment with IL4 to activate STAT6 increased HDAC2 expression and decreased CYBB expression at the mRNA (Supplementary Fig. S5) and protein levels (Fig. 2G) in a time-dependent manner. Direct interactions between HDACs and STATs have been observed in previous research (47). By co-immunoprecipitation assay, there was a direct interaction between STAT6 and HDAC2 in U937 cells after IL4 treatment (Fig. 2H), suggesting that the two proteins cooperate to regulate CYBB transcription. Inhibiting STAT6 with AS1517499 and inhibiting HDAC2 with TSA or LBH589 reduced IL4-induced HDAC2 expression and restored IL4-mediated downregulation of CYBB at the mRNA (Supplementary Fig. S6) and protein levels (Fig. 2I). Similar results were also observed in the coculture experiments in which treatment with STAT6 and HDAC2 inhibitors restored the coculture-induced CYBB downregulation in U937 cells (Supplementary Fig. S7). Furthermore, RNAi targeting STAT6 and HDAC2 in U937 cells reduced M2-like polarization (IL10^Hgh) and restored mRNA and protein expression of CYBB but did not affect M1-like polarization (IL12p40^High; Supplementary Fig. S8A–C; Fig. 2J). Similar results were also observed in the coculture experiments in which RNAi targeting STAT6 and HDAC2 reduced the coculture-induced M2-like polarization but did not affect M1-like polarization and restored the coculture-induced CYBB expression in U937 cells (Supplementary Fig. S9). The IL4/STAT6 signaling pathway was similar like in acute monocytic leukemia cell (THP-1; Supplementary Fig. S10). Collectively, these data indicate that HDAC2 functions as a key downstream regulator of IL4/STAT6 signaling to modulate M2-like polarization.

To study how STAT6 and the corepressor HDAC2 regulate expression of CYBB, we performed ChIP assay to determine whether STAT6 and HDAC2 binds to the CYBB promoter. Several ChIP PCR primers were designed for the CYBB promoter, as shown in Fig. 3A. PCR products covering the D primer (−896/−719) region of the CYBB promoter were detected using ChIP-PCR and ChIP-QPCR after IL4 stimulation (Fig. 3B and C), suggesting a potential binding site for STAT6 and HDAC2 within the −896/−719 region. IL4 treatment triggered the binding of STAT6 to the promoter region, but this effect was inhibited by STAT6 knockdown (Fig. 3B and C). Taken together, these results indicate that IL4 stimulated the binding of STAT6 and HDAC2 to the CYBB promoter.

Next, we constructed the luciferase reporter plasmids with the WT CYBB promoter and mutated CYBB promoters in which the STAT6-binding sequence was changed (Fig. 3D). The reporter assay showed that IL4 decreased the luciferase activity of the WT CYBB promoter reporter but not that of the mutated CYBB promoter reporter (Fig. 3E). Silencing STAT6 effectively blocked the effect of IL4 (Fig. 3E). Taken together, these data suggest that IL4 from pancreatic cancer cells and ROS in macrophages can cooperate to activate STAT6 and HDAC2, and thus M2-like polarization, which contributes to the development of cold TME (noninflamed, immunologically ignorant) of pancreatic cancer (Fig. 3F).

TF decoy ODNs have been applied to interfere with the interaction between TFs and their corresponding genomic binding sites through competitive suppression, selectively regulating gene expression in cells as a therapeutic agent with great potential (28, 48). Therefore, we investigated whether decoy ODNs can inhibit the binding of STAT6 to the CYBB promoter and thus reduce M2-like polarization. A 28mer double-strand ODN was designed to inhibit STAT6 binding to the CYBB promoter. The reporter assay showed that IL4 treatment could repress CYBB expression by reducing its promoter activity; however, treatment with the STAT6-CYBB decoy ODN reversed the reduction of WT CYBB promoter-driven luciferase activity by IL4. (Fig. 4A). Furthermore, in U937 cells, the STAT6-CYBB decoy ODN also increased the expression of CYBB mRNA (Fig. 4B) and the expression of M1-like cytokines including IL12p40, IFNγ, and CCL2, but decreased the expression of M2-like cytokines including IL10, VEGFA, and TGFβ1 after IL4 treatment (Fig. 4C and D) or coculture with PANC-1 cells (Fig 4C and D). Treatment with the STAT6-CYBB decoy ODN also decreased the mRNA expression of PD-L1 and PD-L2 in U937 cells in the presence of IL4 or when cocultured with PANC-1 cells (Fig. 4E). U937 cells treatment with the STAT6-CYBB decoy ODN after coculturing with PANC-1 cells also increased the mRNA expression of IL2, Granzyme B, and Profilin 1 in the Jurkat T cells (Fig. 4F), thereby increasing the proliferation of Jurkat T cells and αCD3⁺ PBMC cells in the CFSE assay (Supplementary Fig. S11). These results suggest that STAT6-CYBB decoy ODNs can significantly prevent the formation of M2-like TAMs, increase T-cell proliferation, and have the potential as an effective therapeutic approach for pancreatic cancer.

Because STAT6-CYBB decoy ODNs effectively impeded the formation of M2-like macrophages in vitro, we evaluated the therapeutic effects of STAT6-CYBB decoy ODNs on pancreatic cancer in vivo. Twenty days after the inoculation of KPC^LUC cells into the pancreas, mice were intravenously injected with scrambled ODNs, STAT6-CYBB decoy ODNs, and mutated STAT6-CYBB decoy ODNs. Three weeks after treatment with decoy ODNs, the mice were sacrificed to collect tumors. Mice treated with STAT6-CYBB decoy ODNs had the lowest tumor volume and tumor weight (P = 0.0003; Fig. 5A). In addition, compared with scrambled ODNs and mutated STAT6-CYBB decoy ODNs, STAT6-CYBB decoy ODNs significantly extended the survival of pancreatic tumor-bearing mice (P < 0.0001; Fig. 5B). Immunohistochemistry (IHC) staining showed that STAT6-CYBB decoy ODNs significantly increased the expression of iNOS and CYBB but reduced the expression of CD206 and YM-1 (Fig. 5C; Supplementary Fig. 12). Compared to the tumors receiving scrambled or mutated decoy ODNs, the STAT6-CYBB decoy ODNs group exhibited increased expression of caspase 3, cleaved caspase 3, and PARP, and decreased expression of the antiapoptotic marker BCL-XL and the cell proliferation marker ki67 (Fig. 5D), suggesting that STAT6-CYBB decoy ODNs reduced pancreatic tumor growth by promoting cell apoptosis and suppressing cell proliferation. Therefore, treatment with STAT6-CYBB ODN also suppresses stromal or epithelial-type cell activation (Supplementary Fig. S13). Serum levels of glutamic oxaloacetic transaminase (GOT), glutamic pyruvic transaminase (GPT), albumin (ALB), and total protein (TP) for liver function monitoring (Fig. 5E) and the serum levels of blood urea nitrogen (BUN), uric acid (UA), creatinine (CRE), and lactate dehydrogenase (LDH) for renal function monitoring (Fig. 5F) in mice were not affected by decoy ODNs compared with sham treatment.

Compared with scrambled and mutated decoy ODNs, STAT6-CYBB decoy ODNs also decreased neutrophil levels but increased lymphocyte and monocyte percentages in the blood circulation (Fig. 6A). In tumors, STAT6-CYBB decoy ODNs downregulated the neutrophil marker Ly6G and the regulatory T-cell marker Foxp3 but upregulated the total lymphocyte marker CD3, the cytotoxic T marker CD8, the NK-cell marker CD56, and the B-cell marker CD19 (Fig. 6B). Moreover, we also confirmed that the cytotoxic marker Granzyme B was upregulated in tumors treated with STAT6-CYBB ODN (Supplementary Fig. S14). These results indicate that STAT6-CYBB decoy ODNs without toxicity have great potential to inhibit M2-like polarization and thus reduce immunosuppression and pancreatic cancer growth (Fig. 6C).

A high density of M2-like macrophages in the pancreatic TME confers poor prognosis. Interrupting the polarization of M2-like TAMs is a potential strategy for pancreatic cancer therapy. Here, we showed that cancer cell-derived IL4 in combination with ROS to activate STAT6/HDAC2 in macrophages could repress CYBB transcription and skew M2-like in the TME of pancreatic cancer. STAT6-CYBB decoy ODNs blocked STAT6/HDAC2 binding to the CYBB promoted and prevented M2-like polarization. This action can maintain the transcription of CYBB to produce ROS, thus maintaining the M1-like phenotype in macrophages and reactivating antitumor immunity to attack cancer cells. Our preclinical animal study provided strong evidence that gene therapy of STAT6-CYBB decoy ODNs could effectively turn “cold” tumors into “hot” ones by inhibiting the formation of M2-like TAMs and upregulating antitumor immune responses in pancreatic TME.

As a multifunctional cytokine in the TME, IL4 not only participates in alternative macrophage priming but also accelerates pancreatic cancer progression (49). By binding to IL4R or IL13R, IL4 activates STAT6 signaling to polarize macrophages to the M2-like phenotype; however, the detailed mechanism has not been fully elucidated (50). It has been reported that IL4 stimulates STAT6 activation and the upregulation of immunoglobulin heavy chain via histone modification in B-cell differentiation, suggesting the involvement of epigenetic regulation in IL4 signaling–associated cell differentiation (51). Consistently, our results showed that IL4 promoted M2-like polarization by upregulating HDAC2 to diminish CYBB gene expression.

Previously, HDAC inhibitors have been reported to inhibit inflammation and promote M2-like macrophage formation (45); however, the underlying mechanism is unclear. Our microarray analysis showed that the expression of HDAC2 was increased in U937 cells after coculture with PANC-1 cells. Furthermore, we demonstrated that the cooperation of IL4 with CYBB-derived ROS strongly activated the STAT6/HDAC2 pathway to inhibit CYBB gene transcription, thus promoting the formation of M2-like. During the conversion of the M1-like phenotype to the M2-like phenotype, CYBB is critical at the beginning because of its function in ROS production; however, after M2-like formation, CYBB function is suppressed, suggesting that CYBB-derived ROS are not necessary for M2-like function and their maintenance.

Previous studies have shown that decoy ODNs focusing on myc-associated zinc finger and high mobility group AT-hook 1 (HMGA-1) inhibit the activation of KRAS transcription and HMGA-1-associated gemcitabine resistance in pancreatic cancer, respectively (52, 53). Additionally, a recent clinical trial has investigated the treatment effectiveness of STAT3 decoy ODNs in head and neck cancer (27). The decoy of TFs provides a novel strategy for interrupting gene expression and has several substantive advantages over knockout animal studies. In other words, decoy ODN have been recognized as an important tool in gene therapy for accurately targeting human diseases. Our study showed that the use of STAT6-CYBB decoy ODNs could markedly suppress M2-like polarization and reduce pancreatic tumor growth. Furthermore, no obvious hepatotoxicity or nephrotoxicity was observed in treated mice.

In conclusion, we clearly demonstrated that, in the TME of pancreatic cancer, cancer cells secrete IL4 to skew M2-like polarization via activation of CYBB-ROS/STAT6/HDAC2 signaling in macrophages. Treatment with STAT6-CYBB decoy ODNs can block M2-like formation to enhance antitumor immune responses in the TME and reduce tumor growth, suggesting a novel gene therapy tool for the treatment of pancreatic cancer.

C.-J. Chen reports grants from National Science and Technology Council, R.O.C during the conduct of the study. H.-C. Wang reports grants from National Science and Technology Council, R.O.C during the conduct of the study. No disclosures were reported by the other authors.

C.-J. Chen: Data curation, formal analysis, investigation, methodology, writing–original draft. H.-C. Wang: Data curation, formal analysis, supervision, investigation, writing–original draft. Y.-C. Hou: Conceptualization, data curation, supervision. Y.-Y. Wu: Data curation, investigation, methodology. C.-C. Shieh: Conceptualization, resources, supervision, writing–review and editing. Y.-S. Shan: Conceptualization, resources, supervision, funding acquisition, project administration, writing–review and editing.

This study was approved by the Ethical Committee and Institutional Review Board of National Cheng Kung University Hospital (IRB number: A-ER-105-459) and the Institutional Animal Care and Use Committee of National Cheng Kung University (IACUC NO.:103150 and of 110331).

All the authors in the study have consented to publish this paper.

We are grateful for technical support from 1. Human Biobank, 2. Immunobiology Core, 3. Flow Cytometry Core Facilities, Clinical Medicine Research Center, National Cheng Kung University Hospital, and technical support from the Laboratory Animal Center, College of Medicine, National Cheng Kung University, and Taiwan Animal Consortium. This study was supported by the National Cheng Kung University Hospital (grant number NCKUH-10301002) and the Ministry of Science and Technology, Taiwan (MOST 106-2314-B-006 -059; MOST 107-2314-B-006 -004; MOST 108-2314-B-006 -001).