IL-4 and IL-17A Cooperatively Promote Hydrogen Peroxide Production, Oxidative DNA Damage, and Upregulation of Dual Oxidase 2 in Human Colon and Pancreatic Cancer Cells

Recent studies suggest that reactive oxygen species (ROS) play an important role as mediators of inflammation-related malignancies by enhancing tumor cell proliferation and angiogenesis (1–3). ROS are produced in human tumors by a variety of mechanisms, including by members of the mitochondrial electron transport chain (4) and the NADPH oxidase (NOX) gene family (5, 6).

The seven NOX family members share certain structural homologies while retaining distinct tissue specificities and mechanisms of activation (7–9). DUOX2 is a membrane-localized glycoprotein composed of six transmembrane helices bearing a cytosolic C-terminal FAD/NADPH binding domain, two cytosolic EF-hands for calcium binding, and an extracellular N-terminal peroxidase-like domain. In the presence of its cognate maturation factor DUOXA2, these structural components regulate the transfer of electrons from NADPH to molecular oxygen to generate H₂O₂ (7, 10–12), which is important for the synthesis of thyroid hormone (13) and plays an important host defense role against pathogens in the airway and gastrointestinal mucosal epithelia (10, 14, 15).

Increasing evidence suggests an etiologic role for DUOX2 in gastrointestinal malignancies; it is overexpressed in premalignant chronic inflammatory states of the colon (16, 17), adenomatous large intestinal polyps (18), and in chronic pancreatitis (19) as well as early stages of pancreatic ductal adenocarcinoma (PDAC) (20). The proinflammatory microenvironment of these diseases, including overexpression of T_H17 and T_H2 cytokines (IL-17A and IL-4/IL-13) and their corresponding receptors, has been demonstrated to significantly alter epithelial tumor cell proliferation, survival, and metastatic potential (21–24). Furthermore, prior studies from our group demonstrated that LPS and the T_H1 cytokine IFN-γ stimulate DUOX2 expression and activity in pancreatic cancer cells by activating signaling through NF-κB and STAT1 (19, 25, 26).

In the present work, we found significant upregulation of DUOX2 and DUOXA2 in surgically resected colon cancer specimens compared with adjacent normal colonic epithelium and have shown that DUOX2 expression is significantly associated with poor prognosis in pancreatic cancer clinical samples from The Cancer Genome Atlas (TCGA). To understand the potential role of DUOX2 in adenocarcinomas of the pancreas and colon, we examined the mechanisms of proinflammatory cytokine-related ROS production by DUOX2 in human pancreatic and colonic cancer cells. We report in this study that IL-4—alone or in combination with IL-17A—strongly induces DUOX2/DUOXA2 mRNA and DUOX protein expression, leading to significantly increased extracellular H₂O₂ accumulation and ROS-related DNA damage. DUOX2 upregulation by IL-4 occurred as a consequence of STAT6 activation and translocation to the nucleus, suggesting a role for STAT6 in the transcriptional regulation of DUOX2 following proinflammatory stimulation. Enhanced expression of DUOX2 by IL-17A appears to be a consequence of NF-κB–related signal transduction. These results suggest that DUOX2 and associated ROS play a role in inflammation-related gastrointestinal malignancies.

Human rIFN-γ (285-IF), IL-17A (317-ILB), IL-4 (204-IL), and IL-13 (213-ILB) were purchased from R&D Systems (Minneapolis, MN). All cytokines were dissolved in 0.1% BSA in sterilized PBS. Actinomycin D (A9415), cycloheximide (C7968), and catalase/polyethylene glycol (PEG-catalase; lyophilized powder, 40,000 U/mg, C4963-2MG) were from Sigma-Aldrich (St. Louis, MO); and ionomycin (calcium salt, 407952-1MG) and LPS (437627) were from Calbiochem (MilliporeSigma, Billerica, MA). Anti-human DUOX Ab, which reacts with both DUOX1 and DUOX2, was previously developed by Creative Biolabs (Port Jefferson Station, NY) and characterized by our laboratory (26). Anti-STAT6 (catalog no. 611290) and anti–p-STAT6_Y641 (catalog no. 611566) used for Western blot analysis were from BD Transduction Laboratories (San Diego, CA). Abs against human lamin A/C (no. 2032), p-histone H2AX (no. 2577S), p-STAT1_Y701 (no. 9167), STAT3 (79D7, no. 4904), p-STAT3_Y705 (no. 9145), and NF-κB (p65) (no. 8242) were purchased from Cell Signaling Technology (Beverly, MA). STAT1 p84/p91 (sc-346), STAT3 (C-20, sc-482), goat anti-rabbit IgG/HRP (sc-2004), and goat anti-mouse IgG/HRP (sc-2005) were from Santa Cruz Biotechnology (Dallas, TX). All human primer and probe sets, β-actin (Hs99999903_m1), DUOX2 (Hs00204187_m1), DUOXA2 (Hs01595310_g1), STAT6 primer (Hs00598625_m1), IL-17RA primer (Hs01064648_m1), and IL-4Rα (Hs00166237_m1), were from Life Technologies (Carlsbad, CA). Negative Control No. 1 siRNA (AM4635) and human DUOX2 siRNA (S27012) were obtained from Applied Biosystems (Carlsbad, CA). ON-TARGETplus Human STAT6 siRNA SMARTpool (L-006690-00-0005) was purchased from GE Healthcare Dharmacon (Lafayette, CO).

Level 3–normalized (RNASeq V2) gene expression datasets were generated by TCGA Research Network (http://cancergenome.nih.gov) and downloaded from the cBioPortal for Cancer Genomics (27, 28). Overall, DUOX2 mRNA expression levels in 9121 samples from 30 studies were retrieved and log2 transformed. For the survival analysis of patients with pancreatic cancer, only primary solid tumors with measured DUOX2 expression and known survival data (n = 177) were included in the analysis. The samples were stratified into low- and high-DUOX2 expression groups using Cutoff Finder to determine the optimal expression cutoff point for statistical significance by log-rank test (29). DUOX2, IL-4R, IL-17RA, and IL-17A gene expression levels in primary solid tumors of the pancreas (n = 178) and colon (n = 380) from TCGA were retrieved using cBioPortal.

The tissue samples analyzed for this study were from patients diagnosed with colon adenocarcinoma and treated by the Department of General Surgery, University Medicine Gottingen (Gottingen, Germany). These data were previously described by Camps et al. (30, 31), and the complete gene expression dataset was previously deposited into the National Center for Biotechnology Information Gene Expression Omnibus database, accession number GSE10402 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE10402). As described in Camps et al. (30), a subset of samples was selected based on also having oligonucleotide array data, and then by applying a 90th interpercentile range procedure to equalize the spread of Cy3 measurement per array (in log2 scale). The resulting subset contained 28 samples: 23 primary colon tumors and 5 normal colon mucosae samples. RNA extraction, quantification, quality assessment, expression microarrays, and data analysis were previously reported for the 23 primary tumors (31). RNA expression levels for 11 oxidases or related proteins were expressed for the 23 colon adenocarcinoma samples relative to the pool of five normal colon mucosae from noncancer patients (31). Unsupervised clustering analysis with BRB-ArrayTools (32) was used to generate a heat map.

For other comparative analyses between normal and malignant tissues, primary human colon and pancreatic cancers and adjacent nonmalignant tissue samples were acquired from the National Cancer Institute–sponsored Cooperative Human Tissue Network (Eastern, Mid-Western, and Mid-Atlantic Divisions) in compliance with the Office of Human Subjects Research at the National Institutes of Health, Bethesda, MD. Specimens were selected without regard to age, race/ancestry, or sex and were obtained from patients who had not received chemotherapy or irradiation treatment prior to surgical intervention. Tumors were preserved by snap freezing in liquid nitrogen within 60 min of surgery. RT-PCR methods are described in the corresponding section below.

Immunohistochemical (IHC) staining was performed on 1203 formalin-fixed, paraffin-embedded tissue microarray blocks from US Biomax. These included the following: colon disease tissue arrays no. BC05002, no. BC051110, and no. CO809a; small intestine tissue array no. SM2081; breast invasive ductal carcinoma tissue array no. BC08118; breast cancer tissue array no. BR1505b; lung cancer tissue array no. LC2085b; midadvanced-stage ovary cancer tissue array no. OV8010; prostate cancer tissue array no. PR2085b; and a tissue array of gastritis with intestinal metaplasia and gastric carcinoma no. IC00011. The tissue microarrays were deparaffinized in alcohol and rehydrated with graded alcohol prior to Ag retrieval. A validated DUOX Ab developed in mouse and described previously (26) was used. Slides were then counterstained with hematoxylin, dehydrated, and coverslipped. All samples were processed in parallel with a no-primary-Ab control to evaluate possible artifactual nonspecific staining from the secondary Ab. Isotype control staining was prepared with normal rabbit IgG (Cell Signaling Technology, Danvers, MA) at a comparable concentration to the primary Ab. The verification of staining performance was confirmed on a series of cancer tissue samples. In addition, a series of normal, nontumor tissues were evaluated to establish immunoreactivity and assay specificity. Evaluation and comparison of staining on sections exposed to the primary and secondary Abs were compared with negative control sections that were not exposed to the primary Ab. Each slide was digitally imaged using an Aperio ScanScope. Tissues were scored as positively stained only if they exhibited a staining pattern with the primary Ab that was significantly different from that found when omitting the primary Ab. Those that did not demonstrate a significant difference between primary and omitting primary staining were graded as 0+ no stain and 0% cells stained. Tissues that demonstrated a significant difference between the two conditions were graded as follows. The assay was interpreted with a scoring system of 0+, 1+, 2+, and 3+ for staining intensity, corresponding to negative, weak, moderate, and strong DUOX staining, respectively. The percentage of stained tumor/lesion cells (distribution) was estimated for each patient, and 0 to <10% was considered negative.

Human colon cancer cell lines T84 (CCL-248) and LS513 (CRL-2134) and human pancreatic cancer cell lines BxPC-3 (CRL-1687), AsPC-1 (CRL-1682), and CFPAC-1 (CRL-1918) were obtained from American Type Culture Collection (Manassas, VA). T84 cells were cultured in a 1:1 mixture of Ham’s F12 medium and DMEM with 2.5 mM l-glutamine and 5% FBS; LS513, BxPC-3, and AsPC-1 cells were cultured in RPMI 1640 medium (SH30255.01; GE Healthcare HyClone, Logan, UT) with 1.0% sodium pyruvate and 10% FBS (100-106; Gemini Bio-Products, Sacramento, CA). CFPAC-1 cells were cultured in IMDM with 10% FBS. The identity of each cell line was confirmed by the Genetic Resources Core Facility at Johns Hopkins University (Baltimore, MD). To establish starvation conditions before each experiment, cells were cultured overnight in the same medium without FBS. Starvation conditions were used because DUOX2 induction by different cytokines is stronger after serum starvation, as noted previously (25). In all cases, cells were cultured in a humidified incubator at 37°C in an atmosphere of 5% CO₂ in air.

For analysis of human tissues, samples ranging in size from 200 to 750 mg were homogenized on ice, and RNA was isolated using the RNeasy Plus Universal Mini Kit (catalog no. 73404; QIAGEN, Germantown, MD) according to the manufacturer’s protocol. Two micrograms of total RNA isolated from each specimen was used for cDNA synthesis in a 20 μl reaction system with the following cycles: 25°C for 5 min, 42°C for 50 min, and 75°C for 5 min. After the reaction was complete, the synthesized cDNA was diluted with H₂O to 100 μl prior to quantitative RT-PCR. For the analysis of human tumor and adjacent normal surgical specimens, β-actin and 18S rRNA were used as housekeeping control genes for colon cancer and pancreatic cancer patient samples, respectively.

For in vitro measurements, total RNA was extracted from 1 × 10⁶ cells using the RNeasy Mini Kit (catalog no. 74104; QIAGEN, Valencia, CA) following the manufacturer’s instructions. Two micrograms of total RNA were then used for cDNA synthesis, along with SuperScript III Reverse Transcriptase (18080-044) and random primers (48190-011), both from Life Technologies, in a 20 μl reaction. The cDNA synthesis steps consisted of cycles of 25°C for 10 min, 42°C for 50 min, and 75°C for 10 min. The synthesized cDNA was diluted to 100 μl with diethylpyrocarbonate-treated H₂O, and quantitative PCR was conducted in 384-well plates in a 20-μl volume consisting of 2 μl diluted cDNA, 1 μl primers, 7 μl H₂O, and 10 μl TaqMan Universal PCR Master Mix (4364340; Life Technologies). The PCR was performed using the default cycling conditions (50°C for 2 min and 95°C for 10 min and 40 cycles of 95°C for 15 s and 60°C for 10 min) with the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA). Triplicate samples were used for the quantitative RT-PCR, and the mean values were calculated. The data in all figures represent three independent experiments. Relative gene expression was calculated from the ratio of the target gene expression to the internal reference gene (β-actin or 18S rRNA) expression based on the cycle threshold values.

For the analysis of human tumor surgical specimens, NOX gene expression levels were measured by quantitative RT-PCR with Power SYBR Green technology (Applied Biosystems). For each RT-PCR, 300 ng cDNA was used. PCR was done with the default variables of the Applied Biosystems’ Prism 7000 sequence detector with a total reaction volume of 25 μl. Primers were obtained from Operon Technologies (Huntsville, AL). Each sample was analyzed in triplicate, and each data point was calculated as the median of the three measured cycle threshold values.

To prepare extracts, cells were lysed in 1× RIPA Lysis Buffer (catalog no. 20-188; MilliporeSigma) supplemented with a phosphatase inhibitor tablet (04-906-837001) and a protease inhibitor tablet (11-836-153001), both from Sigma-Aldrich, to generate whole-cell extracts (WCEs). Nuclear extracts were additionally prepared using the NE-PER Nuclear and Cytoplasmic Extraction Kit (78833; Thermo Fisher Scientific, Rockford, IL). The protein concentrations of both types of extracts were measured using the Pierce BCA Protein Assay Kit (23227; Thermo Fisher Scientific). The extracts were then combined with an equal volume 2× SDS Protein Gel Loading Solution (351-082-661; Quality Biological, Gaithersburg, MD). Next, unless otherwise indicated, 50 μg of WCE or 20 μg of nuclear extract was electrophoretically separated on a 4–20% Tris-Glycine Gel (EC6028; Life Technologies) and transferred to nitrocellulose membranes using an iBlot Transfer Stack (IB 3010-01; Life Technologies). The membranes were blocked in 5% nonfat milk in 1× TBST (TBS with 0.1% Tween 20) buffer for 1 h at room temperature and then incubated overnight with the indicated primary Abs in TBST. After three washes in TBST, the membranes were incubated with the appropriate HRP-conjugated secondary Abs for 1 h at room temperature on a shaker. SuperSignal West Pico Luminol/Enhancer Solution (1856136; Thermo Fisher Scientific) was then applied to visualize the proteins of interest.

Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit (catalog no. A22188; Life Technologies) was employed to detect extracellular H₂O₂ release. Cells were washed twice with 1× PBS, trypsinized, and counted. Next, for BxPC-3 and AsPC-1 cells, 20 μl cell suspension containing 2 × 10⁴ live cells in 1× Krebs–Ringer phosphate glucose buffer was mixed with 100 μl solution containing 50 μM Amplex Red and 0.1 U/ml HRP in Krebs–Ringer phosphate glucose buffer with 1 μM ionomycin and incubated at 37°C for the indicated times. The fluorescence of the oxidized 10-acetyl-3,7-dihydroxyphenoxazine was then measured at excitation and emission wavelengths of 530 and 590 nm, respectively, using a SpectraMax Multi-Mode Microplate Reader (Molecular Devices, Sunnyvale, CA), and the amount of extracellular H₂O₂ was determined using a standard curve from 0 to 2 μM H₂O₂. Each value in the figures is the mean value for triplicate or quadruplicate experiments.

Immunofluorescence staining was performed following the manufacturer’s protocol (Trevigen, Gaithersburg, MD) with some modifications described as follows. A total of 6 × 10⁴ AsPC-1 cells were seeded onto four-well glass chamber slides, then treated with either solvent or IL-4 (50 ng/ml) plus IL-17A (50 ng/ml) for 48 h. The primary Abs were incubated overnight at 37°C on a humidified chamber with dilution 1:2000 in 1× PBS containing 1% BSA and 0.01% Tween 20. The secondary Abs were incubated using a dilution of 1:400 in 1× PBS containing 1% BSA for 1 h at room temperature in the dark. Slides were mounted using VECTASHIELD Antifade Mounting Medium (catalog no. H-1200; Vector Laboratories, Burlingame, CA) with DAPI for counterstaining of nuclei. The Abs for immunofluorescence staining of 8-oxo-7,8-dihydro-2′-deoxyguanosine (8-oxo-dG) were a mouse monoclonal 8-oxo-dG Ab, catalog no. 4354-MC-050, from Trevigen and a goat anti-mouse secondary Ab conjugated with Alexa Fluor 488, catalog no. A-11001, from Thermo Fisher Scientific. Fluorescence images were captured using a confocal microscope (Zeiss LSM 780); ZEN Blue version 2.3 software was used for image quantitation.

Data represent the mean ± SD from three or more independent experiments. Statistical differences between the mean values of samples were assessed using the two-tailed Student t test or the Wilcoxon signed-rank test. Statistical significance was defined as *p < 0.05, **p < 0.01, and ***p < 0.001.

To assess the potential importance of DUOX2 expression in tumorigenesis, we examined the relative expression levels of DUOX2 across a broad range of human malignancies using human tumor datasets from TCGA (Fig. 1A). DUOX2 mRNA expression varied widely in human tumors, with thyroid, pancreas, and colorectal cancers exhibiting the highest levels. These results are consistent with our recent report demonstrating significantly increased DUOX2 expression in patients with early stage PDAC (20). The relative mRNA expression levels of members of the NOX gene family (NOX1/4/5, DUOX1/2, and their accessory proteins) were also examined in surgical samples from 23 colon cancer patients relative to normal colon mucosae from five noncancer patients. As shown in Fig. 1B, NOX1 and DUOX2 were both highly expressed in colon cancers. Some tumors had either increased expression of NOX1 or DUOX2 (versus normal colonic mucosa), whereas other colon cancers demonstrated concurrent overexpression of both genes.

To confirm these findings, we evaluated DUOX protein expression by IHC staining of tissue microarrays from normal tissues, as well as benign and malignant tumors. Generally, DUOX expression was observed in epithelial cells, often with a cytoplasmic pattern. The most significant overexpression relative to normal tissue was identified in gastrointestinal tumors such as colonic adenomas and adenocarcinomas (***p < 0.001; Table I), with DUOX located mainly in the cytoplasm with increasing intensity at the brush border of the malignant glands (Fig. 1Ca); in normal colon mucosa, DUOX protein was detected in neuroendocrine cells (Fig. 1Cb, inset) and at the tips of villi. DUOX protein expression was also increased in small intestinal adenocarcinomas (***p < 0.001; Table I); it was localized in the cytoplasm and the brush border and apical membrane of malignant cells (Fig. 1Cc); normal small intestinal mucosa was mostly negative for DUOX (Fig. 1Cd). High levels of DUOX staining were also observed but less frequently in prostate adenocarcinomas (**p < 0.01 versus normal prostate; Table I); staining was principally cytoplasmic and at the luminal surface of malignant cells (Fig. 1Ce), whereas adjacent nonmalignant tissue showed mild focal DUOX staining (Fig. 1Cf). DUOX protein was also significantly overexpressed in some invasive ductal carcinomas of the breast (**p < 0.01; Fig. 1Cg, Table I), compared with normal ducts and breast tissues (Fig. 1Ch). Finally, we found that 65% (n = 32) of cases with chronic gastritis and 35% (n = 8) of those with gastric carcinomas demonstrated moderate or intense (2+ and 3+) DUOX epithelial staining (Table I). Recently, we reported similar results in patients with chronic pancreatitis (19). These data suggest that chronic inflammation, as well as frank malignancy, may be associated with enhanced DUOX2 expression in multiple organs.

To interrogate possible associations between DUOX2 and patient survival, we performed a Kaplan–Meier analysis and Cox regression of next-generation sequencing data from the TCGA pancreatic cancer cohort. The dataset was stratified into two groups representing high- and low-DUOX2 expression, with the DUOX2 expression cutoff optimized by log-rank analysis (29). We discovered a significant association between high-DUOX2 mRNA expression and worse overall survival (Fig. 1D; hazard ratio = 1.66 and *p < 0.05). These results support the hypothesis that DUOX2 expression could contribute to the development or outcome of gastrointestinal malignancies known to occur in the setting of inflammatory stress (1, 33).

IL-4 activates IL-4Rα, which, in turn, heterodimerizes with IL-13Rα to form the type II IL-4R (34). Several reports indicate that the type II IL-4R is upregulated and activated in various epithelial tumors, including malignant glioma, ovarian, lung, breast, pancreas, and colon carcinomas (35, 36). We therefore screened colon and pancreatic cancer cell lines for the RNA expression of both chains of the type II IL-4R and found measurable RNA expression of IL-4Rα and IL-13Rα1 by quantitative RT-PCR in AsPC-1 and BxPC-3 pancreatic cancer cells and in LS513 and T84 colon cancer cell lines (Supplemental Fig. 1). To characterize whether inflammatory cytokines modulate DUOX2 expression in tumor cells, BxPC-3 pancreatic cancer cells were exposed to 25 ng/ml IFN-γ, IL-4, or IL-13 for 24 h. In BxPC-3 cells, IFN-γ increased DUOX protein expression and the phosphorylation of STAT1 at Tyr701 (p-STAT1_Y701), consistent with our previous report (25). In contrast, IL-4 and IL-13 increased DUOX protein in concert with p-STAT6_Y641, but not p-STAT1_Y701 (Fig. 2A). At the RNA level, the IL-4‒mediated upregulation was specific for DUOX2, and to a lesser extent, DUOX1; the expression levels of NOX1/2/4/5 were unchanged by IL-4 exposure for 24 h in BxPC-3 cells (Supplemental Fig. 2A).

A time course experiment with IL-4 (20 ng/ml) in BxPC-3 cells revealed that after 3 h, IL-4 significantly increased DUOX2 and DUOXA2 RNA expression (**p < 0.01 versus 0 h), with a progressive increase from 3 to 24 h (Fig. 2B, Supplemental Fig. 3A). The effect of 12 h IL-4 treatment was mitigated by 30-min pretreatment with either the transcription initiation inhibitor actinomycin D (100 ng/ml) or the protein synthesis inhibitor cycloheximide (1 μg/ml; ***p < 0.001). Western analysis demonstrated similar results for DUOX protein induction following IL-4 treatment (Fig. 2B, lower panel). Analysis of the kinetics of activation of the IL-4R–mediated signaling pathway revealed that IL-4 selectively induced maximal phosphorylation of STAT6 (but not p-STAT1_Y701 or p-STAT3_Y705) at 1 h, which was sustained up to 24 h (Fig. 2B).

Upregulation of DUOX2/DUOXA2 mRNA and protein levels were associated with enhanced histone H2AX phosphorylation at serine 139 (γH2AX), a marker of DNA double-strand breaks, 12–24 h after IL-4 treatment. A concentration-response evaluation of 24 h IL-4 exposure demonstrated that 1 ng/ml IL-4 was sufficient to significantly increase both DUOX2 and DUOXA2 mRNA and DUOX protein expression in BxPC-3 cells, and to activate p-STAT6_Y641 (Fig. 2C, Supplemental Fig. 3B); maximal induction of DUOX protein expression and the p-STAT6_Y641 signal occurred with 10 ng/ml IL-4 (Fig. 2C, lower panel). Moreover, BxPC-3 cells treated with IL-4 produced significantly more H₂O₂ over time compared with solvent-treated cells (***p < 0.001 at 15–180 min) as measured by Amplex Red assay (Fig. 2D). The addition of membrane permeable PEG-catalase to the reaction system decreased extracellular H₂O₂ accumulation from both solvent-treated and IL-4‒exposed BxPC-3 cells below the limit of detection (data not shown).

We also assessed the expression level of the receptor for the T_H17 proinflammatory cytokine IL-17A, IL-17RA, and found that it was expressed in our panel of cell lines (Supplemental Fig. 1, dark gray bars). In BxPC-3 cells, treatment with IL-17A alone (25 ng/ml) increased the expression of DUOX protein as well as γH2AX, analogous to treatment with IL-4 (Fig. 2E). However, this effect was independent of p-STAT6_Y641 activation. The enhanced γH2AX signal produced by exposure to IL-4 or IL-17A, was attenuated by pretreatment of these cells with 2000 U/ml PEG-catalase (Fig. 2E). The catalase treatment, however, did not prevent cytokine-enhanced DUOX protein expression.

To define the role of STAT6 in the regulation of IL-4–mediated DUOX2 expression further, we performed RNA interference experiments in the BxPC-3 pancreatic cancer cell line. Transfecting BxPC-3 cells with human STAT6-specific small interfering RNA (siRNA) resulted in ≈75% inhibition of STAT6 mRNA expression compared with endogenous levels (Fig. 3A). STAT6 knockdown was accompanied by significant attenuation of IL-4‒induced DUOX2 (Fig. 3B; ***p < 0.001 versus control siRNA with IL-4) and DUOXA2 mRNA expression (Fig. 3C; ***p < 0.001 versus control siRNA with IL-4). Western analysis demonstrated that STAT6-specific siRNA also inhibited STAT6 protein expression, but had no effect on the expression of total STAT1 and STAT3 (Fig. 3D). Furthermore, silencing STAT6 also inhibited IL-4–induced DUOX protein expression in BxPC-3 cells. These results suggest that IL-4–mediated DUOX2/DUOXA2 gene regulation occurs via STAT6 signaling in BxPC-3 pancreatic cancer cells.

To investigate the nuclear distribution of STAT6 following treatment with IL-4 (20 ng/ml), nuclear extracts of BxPC-3 cells treated for different durations (0, 1, 24, and 48 h) were prepared for Western analysis. Lamin A/C served as a loading control. After 1 h exposure to IL-4, nuclear translocation of STAT6 was observed (Fig. 3E). After 24 and 48 h, nuclear STAT6 signal persisted, albeit to a lesser extent compared with 1 h of treatment. Three-kilobase segments of human DUOX2 promoter were scanned to identify a potential STAT6 binding site matching the canonical STAT6 binding motif 5′-TTCN_(3–4)GAA-3′ (37), in which the underlined letters correspond to conserved palindromes and N denotes any of the 4 nt. We identified such a site in the human DUOX2 promoter region with sequence 5′-TTCACTGAA-3′, localized at around –4594 bp from the DUOX2 transcription start site.

To explore the effects of IL-4 and IL-17A on DUOX2 expression in a different model of human pancreatic cancer, AsPC-1 cells were treated for 1 or 24 h with IL-4 (50 ng/ml) or IL-17A (50 ng/ml) alone or in combination. The stimulatory effects of IL-4 and IL-17A on DUOX protein expression, p-STAT6_Y641, and γH2AX described above for BxPC-3 cells were confirmed in AsPC-1 cells (Fig. 4). Whereas 50 ng/ml IL-4 alone or with IL-17A (50 ng/ml) induced a robust p-STAT6_Y641 signal as early as 1 h, cytokine-induced upregulation of DUOX protein and production of γH2AX occurred later and could be detected at 24 h (Fig. 4A). As demonstrated for BxPC-3 cells, RNA interference experiments confirmed that the STAT6/siRNA not only significantly decreased STAT6 mRNA (***p < 0.001; Fig. 4B, left) and protein (Fig. 4C) expression by more than 50%, but also diminished IL-4 plus IL-17A‒mediated upregulation of DUOX2 mRNA (***p < 0.001; Fig. 4B, right) and DUOX protein (Fig. 4C). These results support a role for STAT6 signaling in the cytokine-mediated overexpression of DUOX2 in pancreatic cancer cells.

To corroborate the association of DUOX2-derived H₂O₂ with DNA damage in this model, AsPC-1 cells were preincubated with the flavin dehydrogenase (and NOX) enzyme inhibitor diphenylene iodonium (DPI) or the reduced thiol N-acetyl-l-cysteine (NAC). Both DPI and NAC decreased the IL-4 plus IL-17A–mediated activation of γH2AX but had no effect on DUOX or p-STAT6_Y641 protein levels (Fig. 4D). The role of DUOX2 in the induction of DNA damage was also evaluated by addition of a DUOX2-specific siRNA to AsPC-1 cells; in these experiments, DUOX2-specific siRNA abrogated DUOX protein expression and activation of γH2AX following IL-4 plus IL-17A exposure (Fig. 4E, lane 4 compared with lane 2). However, the DUOX2 siRNA had no inhibitory effect on total STAT6 protein expression or Y641 phosphorylation (Fig. 4E). To evaluate the involvement of DUOX2 in the activation of DNA damage repair, we examined whether IL-4 and IL-17A enhanced the enzymatic activity of DUOX2 in AsPC-1 cells. In this cell line, the combination of IL-4 with IL-17A and the calcium ionophore ionomycin significantly enhanced the production of H₂O₂ compared with cells that had not been exposed to the cytokine combination (***p < 0.001; Fig. 4F).

Finally, to evaluate the potential mutagenic effect of cytokine-enhanced DUOX2 expression (38, 39), we measured the production of the promutagenic oxidized DNA base 8-oxo-dG in AsPC-1 cells following a 48-h exposure to the combination of IL-4 and IL-17A (Fig. 4G). Treatment with the cytokine combination substantially increased 8-oxo-dG levels in AsPC-1 cells (p = 0.053).

IL-17A regulates immune function through its cognate receptor IL-17RA and downstream activation of MAPK and NF-κB signaling (40, 41). To understand the molecular mechanism underlying IL-17A–mediated DUOX2 regulation in human pancreatic cancer cells, the potential of IL-17A alone or in combination with IFN-γ/IL-4 to enhance DUOX2 mRNA expression was compared with that of LPS (Fig. 5). The latter has been previously shown to activate NF-κB signaling and induce DUOX2 expression either alone or in combination with IFN-γ in CFPAC-1 pancreatic cancer cells (19). As demonstrated in Fig. 5A, LPS, IFN-γ, and IL-17A independently induce DUOX2 mRNA expression (***p < 0.001 versus solvent treated). The combination of either LPS or IL-17A with IFN-γ increased DUOX2 mRNA expression >8-fold compared with LPS or IL-17A alone and >20-fold compared with IFN-γ alone in CFPAC-1 cells (Fig. 5A; ***p < 0.001 versus either cytokine alone). However, the cotreatment of LPS or IL-17A with IL-4 did not result in a further increase in DUOX2 mRNA expression compared with LPS/IL-17A exposure alone (Fig. 5B). Western analysis using whole-cell lysates from CFPAC-1 cells treated with LPS or IL-17A for different durations revealed that although neither cytokine affected the expression of the p65 subunit of NF-κB, both induced robust DUOX protein expression after 24 h exposure (Fig. 5C, lane 4 for LPS; lane 7 for IL-17A). We then compared NF-κB signal transduction in CFPAC-1 cells exposed to LPS or IL-17A. Nuclear extracts prepared from CFPAC-1 cells treated with LPS or IL-17A for 1 or 24 h were analyzed (Fig. 5D); 1 h exposure to either LPS (1 μg/ml; lane 2) or IL-17A (50 ng/ml; lane 3) resulted in an increase in the nuclear translocation of p65 compared with solvent-treated cells (lane 1). At 24 h, the enhanced nuclear localization of p65 was maintained in LPS-treated cells, but not in IL-17A‒treated cells, revealing ligand-dependent differences in the kinetics of NF-κB signaling activation.

Because DUOX2 is overexpressed in surgically resected tumors from a substantial fraction of colorectal cancer patients (Fig. 1A, 1B, Table I), we evaluated the effects of IL-4 and IL-17A on DUOX2 mRNA and protein expression in colon cancer lines. For T84 colon cancer cells (Fig. 6A), 24 h exposure to IL-4 (50 ng/ml) significantly upregulated both DUOX2 and DUOXA2 mRNA expression (***p < 0.001 versus control). Western analysis confirmed the increase in DUOX protein levels and the activation of p-STAT6_Y641 following IL-4 exposure (Fig. 6B). Furthermore, the combination of IL-4 and IL-17A significantly enhanced the induction of DUOX2 and DUOXA2 mRNA (***p < 0.001 versus control and IL-4 alone) and increased DUOX protein expression in this colon cancer model (Fig. 6A, 6B). The effect of the IL-4 and IL-17A combination on NOX expression in T84 cells was observed for DUOX2, and to a much lesser extent DUOX1 and NOX1; these cytokines did not affect the mRNA expression of other NOX family enzymes (Supplemental Fig. 2B). Likewise, in another colon cancer cell line that expresses NOX1 (Supplemental Fig. 4A), LS513, IL-4 alone (24 h, 25 ng/ml) significantly upregulated DUOX2 mRNA (***p < 0.001; Supplemental Fig. 4B), as well as DUOX protein and p-STAT6_Y641 (Supplemental Fig. 4C). The addition of IL-17A to IL-4 further enhanced the DUOX2 RNA and protein expression over IL-4 treatment alone (Supplemental Fig. 4B, 4C). In T84 cells, the upregulation of DUOX protein by IL-4 and IL-17A treatment was accompanied by a DNA damage response in the form of an increase in γH2AX expression (Fig. 6B), as described above in pancreatic cancer cell lines BxPC-3 and AsPC-1. Moreover, in this model of colon cancer as well, STAT6-specific siRNA significantly inhibited both STAT6 RNA (Fig. 6C, left) and protein (Fig. 6D), in addition to IL-4 plus IL-17A‒induced DUOX2 upregulation (Fig. 6C [right], Fig. 6D).

To assess the clinical relevance of IL-4‒mediated upregulation of DUOX2 in pancreatic and colon cancer cells, we measured the expression of DUOX2 and IL-4R in 13 surgically resected PDACs and their adjacent normal pancreatic tissues, and in 19 surgically resected colon adenocarcinomas and associated normal colon (42) (Figs. 7, 8). In the pancreatic cancer cohort, the expression of DUOX2, IL-4R, and IL-17RA was significantly higher in tumors compared with adjacent normal tissues (*p < 0.05, **p < 0.01; Fig. 7A–C). In the colon cancer resection specimens, DUOX2 (but not IL-4R) was strongly overexpressed in tumor samples compared with normal tissue (***p < 0.001; Fig. 8A, 8B). DUOXA2 expression levels in these samples followed the same pattern as DUOX2 (data not shown).

In an evaluation of tumor expression profiles from the TCGA, we found significant positive correlations between the expression of DUOX2 and the expression of IL-4R in large pancreatic (n = 178; Pearson r = 0.37; ***p < 0.001) and colon (n = 380; Pearson r = 0.20; ***p < 0.001) cancer cohorts (Figs. 7D, 8C, respectively). Additionally, in the TCGA pancreatic cancer dataset, DUOX2 and IL-17RA expression were also significantly correlated (Pearson r = 0.29; ***p < 0.001; Fig. 7E); the expression of the IL-17RA ligand IL-17A was found to be significantly correlated with the expression of DUOX2 in TCGA colorectal cancer cohort (Pearson r = 0.41; ***p < 0.001; Fig. 8D). These results support a relationship between proinflammatory signaling via IL-4 and IL-17A, and DUOX2 in human pancreatic and colon cancers.

DUOX2, which plays an important role in gastrointestinal and respiratory host defense (10, 13–15), is upregulated during chronic inflammation in humans and in multiple cancers (Fig. 1A) (18, 25, 26, 43, 44). Using our human DUOX mAb (26), we recently demonstrated that DUOX is highly expressed in most early stage PDACs as well as pancreatic intraepithelial neoplasms (20). We report in this study that evaluation of over 1200 tumors and normal samples by IHC revealed that, in addition to PDAC, DUOX is most frequently overexpressed in adenocarcinomas of the large and small bowel (Fig. 1C, Table I). These results are supported by a bioinformatic examination of expression levels of DUOX2 mRNA across the TCGA (Fig. 1A), as well as our two newly reported datasets from surgical samples comparing tumor to adjacent normal tissue levels of DUOX2 expression by expression array profiling (Fig. 1B) or by quantitative RT-PCR (Figs. 7A–C, 8A, 8B).

Because we found a significant association between high-DUOX2 mRNA expression and poor prognosis for patients with PDAC (Fig. 1D), and because genetically engineered mouse models have established an intimate link between chronic inflammation because of IL-4 and IL-17A and pancreatic and bowel carcinogenesis (45–49) as well as a role for DUOX2 in the pathogenesis of IBD (50), we initiated the current studies to examine potential mechanistic interactions between these proinflammatory cytokines, NOX isoforms, and reactive oxygen production. In a recent study, we reported that IL-4 exposure induces NOX1 expression in HT-29 and WiDr colon cancer cells (42). Our demonstration in the present work that IL-4 can also increase DUOX2 expression in colorectal cancer lines previously known to express only NOX1, including T84 (Fig. 6A) and LS513 cells (Supplemental Fig. 4), is consistent with the data in Fig. 1B demonstrating that surgically resected colorectal tumors can express either NOX1 or DUOX2 or both at high levels compared with normal colonic mucosa.

In PDAC and colon cancer cells, IL-4 significantly increased DUOX2 mRNA and DUOX protein expression through a STAT6-dependent mechanism (Figs. 2–4, 6, Supplemental Fig. 4). Kinetic analysis of IL-4–mediated signaling revealed that STAT6 activation occurred first (after 1 h in BxPC-3 cells), suggesting that this step is upstream of the transcriptional regulation of DUOX2 (Fig. 2B). IL-4 promotes the nuclear translocation of STAT6 (Fig. 3E); the IL-4–mediated DUOX2 response can be silenced by STAT6 siRNA (Fig. 3A–C) supporting this thesis.

IL-17 is often described as an potentiator of inflammatory response because it promotes the production of cytokines such as IL-6, chemokines such as CCL5, and other inflammatory mediators and, as such, has been shown to synergize with other cytokines to increase inflammation (51, 52). Addition of IL-17A to IL-4 further increased DUOX2 overexpression in pancreatic (Fig. 4A, 4B) and colon cancer (Fig. 6A, 6B) cells, which may be NF-κB related (Fig. 5). Interactions and cross-talk between STAT6 and NF-κB have been studied by other groups. For instance, Shen et al. (53) provided the proof of concept for the functional interaction of STAT6 and NF-κB using HEK293 and I.29μ B lymphoma cells and an IL-4–inducible reporter gene containing cognate binding sites for both STAT6 and NF-κB. Upon stimulation with IL-4, the transcription factors cooperatively activated the transcription of the IL-4 response gene (53). Correspondingly, the DUOX2 promoter contains canonical binding motifs for both STAT6 and NF-κB. We have previously demonstrated that the p65 component of the NF-κB complex binds the DUOX2 promoter about −2700 bp from the transcription start site (19).

Cytokine-enhanced ROS formation and resulting DNA damage play an important role in chronic inflammation-associated gastrointestinal malignant transformation (19, 54–56). Production of H₂O₂ by IL-4/IL-17A-enhanced DUOX2 expression led to DNA double-strand breaks in BxPC-3, AsPC-1, and T84 cells (Figs. 2, 4, 6), as well as the production of the oxidized DNA base 8-oxo-dG in the AsPC-1 line. Oxidant DNA damage could help to explain at least part of the genetic instability that accompanies chronic pancreatic and bowel inflammation. In all three cell lines, IL-4 and IL-17A increased γH2AX in concert with DUOX protein; this effect was mitigated by PEG-catalase in BxPC-3 cells (Fig. 2E) and by the flavoprotein and NOX inhibitor DPI and the reduced thiol NAC in AsPC-1 cells (Fig. 4D). These results suggest that DUOX2-associated ROS play a role in the induction of DNA damage following cytokine treatment and that an increase in DUOX2 activity could promote malignant transformation in response to inflammatory stimuli.

Binding of IL-4 to the IL-4R initiates a signaling cascade that activates STAT6 (57, 58). Engagement of IL-17 with IL-17RA mediates NF-κB and MAPK activation, leading to production of proinflammatory cytokines and chemokines, and subsequent recruitment of myeloid cells to the inflamed tissues (21, 41). Therefore, we posit that the combination of T_H2 and T_H17 cytokines synergistically enhance oxidative stress by activating distinct regulatory pathways to increase DUOX2 expression in pancreatic and colon cancer. Our observations of the significant correlation between the mRNA expression of DUOX2 and IL-4R in TCGA pancreatic and colorectal cancer cohorts, as well as the correlation of DUOX2 expression with IL-17RA in pancreatic tumors and IL-17A in colorectal tumors, provide support for a pathogenetic role of IL-4 and/or IL-17‒mediated upregulation of DUOX2 in those cancers (Figs. 7, 8).

In summary, we have shown that IL-4 alone, or combined with IL-17A, strongly induces DUOX2 and its cognate maturation factor DUOXA2, leading to long-lasting H₂O₂ production and DNA damage in human pancreatic and colon tumor cells. These studies, in concert with mRNA and IHC data demonstrating high-level DUOX protein expression in colon and pancreatic cancer and associated adverse impact on overall survival in the latter case suggest that DUOX2 may be an important therapeutic target for these two inflammation-related malignancies.

We thank Dr. Thomas Ried (Cancer Genetics Branch, National Cancer Institute, National Institutes of Health) for providing access to the gene expression data for the 23 patients with colon adenocarcinoma.

The authors have no financial conflicts of interest.