Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
Cancer Tumor Biology and Immunology Research
IL22RA1/STAT3 Signaling Promotes Stemness and Tumorigenicity in Pancreatic Cancer Weizhi He1, Jinghua Wu1, Juanjuan Shi1, Yan-Miao Huo2, Wentao Dai3, Jing Geng1, Ping Lu1, Min-Wei Yang2, Yuan Fang4, Wei Wang4, Zhi-Gang Zhang5, Aida Habtezion6, Yong-Wei Sun2, and Jing Xue1
Abstract
Chronic inflammation is a feature of pancreatic cancer, but bored higher stemness potential and tumorigenicity. Notably, little is known about how immune cells or immune cell–related IL22 promoted pancreatic cancer stemness via IL22RA1/STAT3 signals affect pancreatic cancer stemness and development. Our signaling, establishing the mechanism of regulation of cancer previous work showed that IL22/IL22RA1 plays a vital role in stemness by microenvironmental factors. Moreover, STAT3 was acute and chronic pancreatitis progression by mediating cross- indispensable for the maintenance of IL22RA1hi cells. Overall, talk between immune cells and acinar cells or stellate cells, these findings provide a therapeutic strategy for patients with respectively. Here, we find IL22RA1 is highly but heteroge- PDAC with high expression of IL22RA1. neously expressed in pancreatic cancer cells, with high expres- Significance: IL22RA1/STAT3 signaling enhances stemness sion associated with poor prognosis of patients with pancreatic and tumorigenicity in pancreatic cancer. Cancer Res; 78(12); cancer. The IL22RA1hi population from pancreatic cancer har- 3293–305. 2018 AACR.
Introduction lating CSCs and developing novel CSC-targeted therapies are an urgent need. Pancreatic cancer is a highly aggressive disease with few effec- IL22, a well-defined ligand for IL22RA1, belongs to the tive therapies. Pancreatic ductal adenocarcinoma (PDAC) is the IL10 cytokine family and plays a critical role in host defense, most common form of pancreatic cancer (1–3). Despite recent tissue repair, and carcinogenesis in a context-dependent manner advances in surgical and early diagnosis techniques, survival has (8–10). IL22 signals via the IL22 receptor that consists of IL22RA1 not changed much in the past two decades. Pancreatic cancer stem and IL10RB (11). IL10RB is expressed ubiquitously in various cells (CSC), first identified in 2007, are characterized by their organs (12), whereas IL22RA1 has more restricted expression in abilities of self-renewal and differentiation, in vivo tumorigenicity, different tissues and pathologic conditions (13). IL22 and and the ability to drive metastasis (4, 5). Most importantly, CSCs IL22RA1 are expressed on leukocytes and epithelial cells, respec- are believed to contribute to tumor initiation, relapse, and ther- tively, allowing for cross-talk between the two cell types. Our apeutic resistance (6, 7). Therefore, revealing mechanisms regu- previous studies have elucidated the role of IL22/IL22RA1 sig- naling in acute and chronic pancreatitis (14, 15); in this study, we have explored the role of this signaling axis in pancreatic cancer 1State Key Laboratory of Oncogenes and Related Genes, Stem Cell Research development and progression. Center, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, The interactions between tumor cells and immune cells pro- Shanghai, China. 2Department of Biliary-Pancreatic Surgery, Renji Hospital, mote tumor development and progression (16–18). It is well School of Medicine, Shanghai Jiao Tong University, Shanghai, China. 3Shanghai accepted that both intrinsic and external signals from microen- Center for Bioinformation Technology & Shanghai Engineering Research Center vironment define the phenotype of CSCs (19–21). Chronic 4 of Pharmaceutical Translation, Shanghai, China. Department of General Sur- inflammation is a feature of PDAC and is believed to be an gery & Research Institute of Pancreatic Disease, Ruijin Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China. 5State Key Labora- independent risk factor for pancreatic cancer initiation and pro- – tory of Oncogenes and Related Genes, Shanghai Cancer Institute, Shanghai Jiao gression (22, 23). However, how immune cells or immune cell Tong University, Shanghai, China. 6Division of Gastroenterology and Hepatol- related signaling modulate CSCs is poorly defined in pancreatic ogy, Stanford University School of Medicine, Stanford, California. cancer. Note: Supplementary data for this article are available at Cancer Research Here, we show that high expression of IL22RA1 is associated Online (http://cancerres.aacrjournals.org/). with poor prognosis in pancreatic cancer. PDAC cells with high Corresponding Authors: Jing Xue, Renji Hospital, School of Medicine, Shanghai expression of IL22RA1 possess stemness characteristics, including Jiao Tong University, Shanghai 200127, China. Phone: 8621-6838-3922; Fax: self-renewal, tumorigenicity, and metastasis. Notably, IL22 pro- 8621-6838-3916; E-mail: [email protected]; Yong-Wei Sun, [email protected] motes pancreatic cancer stemness via IL22RA1/STAT3 signaling, ; and Aida Habtezion, Department of Medicine, Division of Gastroenterology & identifying the mechanism of regulation of cancer stem cells by Hepatology, Stanford University School of Medicine, 300 Pasteur Drive, Stanford, microenvironmental factors. Moreover, STAT3 activation is indis- CA 94305. E-mail: [email protected] pensable for the maintenance of IL22RA1hi cells, both in the doi: 10.1158/0008-5472.CAN-17-3131 absence or presence of IL22, providing a therapeutic target for 2018 American Association for Cancer Research. patients with PDAC with high expression of IL22RA1.
www.aacrjournals.org 3293
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
He et al.
Materials and Methods with 10% FBS and 1% penicillin–streptomycin in a humidified Pancreatic cell lines and primary tumor cells chamber at 37 C with 5% CO2. Pancreatic patient-derived xeno- Human pancreatic cancer cell lines (BXPC3, SW1990, PANC1, graft (PDX)-derived cancer cells (PDC), including PDC0034, Miapaca-2, AsPC-1, and HPAC) and normal pancreatic ductal cell PDC0049, and PDC0001, were isolated from pancreatic PDX lines (HPDE and hTERT-HPNE) were obtained from ATCC or Cell tumors and cultured in complete RPMI medium plus 10 ng/mL Bank of Chinese Academy of Sciences (Shanghai, China). PDAC EGF and 1% ITS as described previously (24). Primary tumor cells cells were cultured in DMEM or RPMI medium supplemented from a patient with PDAC samples were directly isolated by
Figure 1. IL22RA1 is upregulated in pancreatic cancer, and its level is negatively associated with patient prognosis. A, Immunoblotting analysis of IL22RA1 protein level among human pancreatic cancer cell lines (HPAC, AsPC-1, Miapaca-2, PANC1, SW1990, and BxPC3) and normal pancreatic ductal cells (HPDE and hTERT-HPNE). B, Representative immunofluorescence images of PDAC tissues from three different patients with pancreatic cancer, costained with IL22RA1, CK19, and DAPI (nuclei). Scale bar, 50 mm. C, Flow cytometry analysis for IL22RA1 expression among PDAC cell lines, PDCs, and primary tumor cells (pregated on liveþEpCAMþ cells). D, The association between IL22RA1 transcript level and patient overall survival. The analyses were conducted in 50 patients with pancreatic cancer (GSE102238; log-rank test, P ¼ 0.031). E, The association between IL22RA1 transcript level and patient overall survival. The analyses were conducted in 177 patients with pancreatic cancer from TCGA database (log-rank test, P ¼ 0.022).
3294 Cancer Res; 78(12) June 15, 2018 Cancer Research
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
IL22RA1/STAT3 Promotes Stemness in PDAC
collagenase digestion (collagenase IV, 2 mg/mL, 30–60 minutes). Flow cytometry Human pancreatic tissues from patients with PDAC were Cells were prepared as a single-cell suspension for FACS obtained from the Renji Hospital (Shanghai, China) with Local staining. For surface staining, the following antibodies were Ethics Committee approval and patient consent. used: PE/Cy7-CD24, FITC-CD44, PE-ESA/EpCAM, APC-CD3,
Figure 2. Pancreatic cancer cells with high IL22RA1 expression harbor stemness properties. A, Morphology of SW1990 cells was grown as monolayer or spheres. Western blot analysis of IL22RA1, NANOG, SOX2, OCT3/4, and b-actin in spheres as compared with adherent cells. B, qPCR analysis of stemness-associated genes and IL22RA1 in spheres versus adherent cells among different PDAC cell lines. Data are normalized to GAPDH expression and are presented as fold change in gene expression relative to adherent cells. Data are mean SEM from three independent experiments. , P < 0.05; , P < 0.01; , P < 0.001; (unpaired two-tailed Student t test). C and D, ALDEFLUOR analyses of IL22RA1hi and IL22RA1 cells from SW1990 and primary PDAC cells. DEAB, a specific inhibitor of ALDH1, was used as a control. ALDH1þ cells were gated as shown. E, IL22RA1hi and IL22RA cells were sorted with flow cytometer. The relative expression of IL22RA1 and stemness-associated genes was determined by qPCR assay. Data are mean SEM, n ¼ 3; , P < 0.05 (multiple unpaired Student t test). F, IL22RA1hi and IL22RA1 cell population from SW1990 and PDC0034 was sorted for sphere assay. Representative image of sphere and mean number of sphere are shown. Scale bar, 50 mm. Data, mean SEM (unpaired two-tailed Student t test). G, Cell viability of IL22RA1hi and IL22RA1 cells was examined upon titrated dose of gemcitabine treatment for 72 hours. Data represent mean SD of three replicates from one representative experiment (two-way ANOVA test and Bonferroni post hoc test were performed to compare the doses effects between two groups; , P < 0.05).
www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3295
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
He et al.
A B Cell injected 1,000 Group – 1K 10K 100K IL22RA1 800 IL22RA1hi – 0/10 4/10 5/5 600 IL22RA1 P (0%) (40%) (100%) 400 < 0.001 2/10 8/10 5/5 IL22RA1hi 200 (20%) (80%) (100%) Proliferation (%) 0 *P < 0.05 D0 D1 D2 D3
IL22RA1– C 8 D IL22RA1hi 2,000 6 IL22RA1– 1,500 IL22RA1hi 4 1,000
KPC1199 2 P < 0.001 500
Iso IL22RA1 0 Poliferation (%) 0 Relative gene expressior D0 D1 D2 D3 Sox2 II22ra1NanogPou5f1 Epcam Orthotopic model E F
H&E CK19 IL22RA1
IL22RA1hi IL22RA1– hi P 0.4 = 0.004
0.3 IL22RA1
0.2 –
0.1 Tumor weight (g) Tumor 0.0 IL22RA1 hi –
IL22RA1 IL22RA1
G Liver metastasis model H&E CK19
4× 20× 4× 20× P < 0.001 25 hi 20
IL22RA1 15
10 – Metastatic lesion (per 4× field) + 5 IL22RA1
CK19 0 hi –
IL22RA1 IL22RA1
3296 Cancer Res; 78(12) June 15, 2018 Cancer Research
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
IL22RA1/STAT3 Promotes Stemness in PDAC
PE/cy7-CD45, and BV421-CCR6 (BioLegend). For IL22RA1 IHC and immunofluorescence staining, cells were stained first with the IL22RA1 antibody Pancreas tissues were fixed in 10% buffered formalin or frozen (ab5984, Abcam) for 30 minutes at 4 C and second with AF647 in Tissue-Tek OCT compound. Fixed tissues were sectioned and goat anti-rabbit IgG for 30 minutes at 4 C. For p-STAT3 used for hematoxylin and eosin (H&E) staining and CK19 IHC intracellular staining, sorted cells were fixed and permeabilized staining (Troma-III, DSHB). Frozen tissues were also sectioned for with methanol and stained with FITC-conjugated pSTAT3 anti- immunofluorescence staining with the following antibodies: body (clone 4/P-STAT3, BD) for 30 minutes at 4 C. ALDH rabbit anti-human CK19 (Santa Cruz Biotechnology) and activity was determined using the ALDEFLUOR Kit and follow- IL22RA1 (Abcam). Cells were mounted with DAPI and ProLong ing the manufacturer's instructions (STEMCELL). For IL22 Gold Antifade Reagent (Life Technology) before analysis with a intracellular cytokine staining, immediately after isolation, confocal microscope. cells were cultured in RPMI complete medium and stimulated with phorbol myristate acetate (50 ng/mL), ionomycin (1 mg/ Chromatin immunoprecipitation mL), and brefeldin A (10 mg/mL, eBioscience) for 4 hours at Chromatin immunoprecipitation (ChIP) was performed 37 C. The cells were washed and stained with surface markers according to the EZ-ChIP Kit protocol (Millipore). In brief, cross- (CD45, CD3, and CCR6) and subsequently fixed and permea- linking was performed with 1% formaldehyde at room temper- bilized using the eBioscience Kit following the manufacturer's ature for 10 minutes; then, glycine was added to quench unreacted guidelines (14). PE-IL22 (BioLegend) was used for intracellular formaldehyde. Cells were lysed with SDS lysis buffer and soni- staining. Dead cells were excluded from the analysis using a cated with the Sonics VibraCell Sonicator for shearing crosslinked viability stain (Invitrogen). The stained cells were acquired for DNA to about 200 to 1,000 bp. Protein and DNA complexes were analysis or sorting using the LSRFortessa or AriaII (BD Bios- precipitated with specific antibodies against STAT3 and IgG con- ciences). Flow cytometry data was analyzed with FlowJo soft- trol (Cell Signaling Technology). To reverse DNA–protein cross- ware (Tree Star Inc.). links, NaCl was added, and lysates were incubated at 65 C for 4 to 5 hours. ChIP-enriched chromatin was used for qPCR with SYBR Sphere formation assay Green MasterMix, normalizing to input. Specific primers are listed PDAC spheres were generated as described previously (4, 25). in the Supplementary Information (Supplementary Table S1). Briefly, cells were suspended in DMEM: F12 medium containing 2%B27(Gibco),20ng/mLbFGF(PeproTech),andplatedin Lentiviral transduction 24-well ultralow attachment plates (Corning). Growth factors Several lentiviral vectors were used to transduce PDAC cells and were replenished every 3 days. After 7 days, spheres with a establish stable cell lines. diameter >50 mm were quantified. The vectors included pGIPZ lentiviral vector encoding gene- specific shRNAs for STAT3, IL22RA1, or scrambled shRNA. The þ Quantitative RT-PCR stable cell lines were obtained through GFP (coexpressed with Cells were lysed with TRIzol (Invitrogen) and RNA extracted the lentiviral vector) cell sorting or puromycin selection. Knock- following standard protocol. cDNA was synthesized with the down efficiency was assessed by qPCR and immunoblotting. High-Capacity cDNA Reverse Transcription Kits (Applied Biosys- tems). Specific primers are included in the Supplementary Infor- Cell viability analysis mation (Supplementary Table S1). Quantitative RT-PCR reaction Cell proliferation and sensitivity assay to gemcitabine and was performed using SYBR Green Master Mix (Roche) and the STAT3 inhibitors were all performed with CellTiter-Glo 2.0 StepOnePlus system (Applied Biosystems). Relative expression of reagent (Promega) according to the manufacturer's instructions. target genes was calculated according to the Ct value with nor- malization to GADPH. In vivo tumor formation Serial numbers (103,104, and 105) of sorted IL22RA1 and Immunoblotting IL22RA1hi SW1990 cells were subcutaneously injected into each Immunoblotting was performed using the following anti- flank of 4-week-old male nude mice (BALB/cA-nu/nu). Tumor bodies: human STAT3 and phosphorylated STAT3 (Cell Signaling incidence was monitored within 8 weeks. KPC1199 mouse pan- Technology); OCT3/4, NANOG, SOX2, and IL22RA1 (Abcam); creatic cancer cells were sorted into IL22RA1 and IL22RA1hi and b-actin (Sigma). Signals were detected by ECL reagents subpopulations and injected either orthotopically into the pan- (Thermo Fisher Scientific). creas tail (104/20 mL) or intraspenically (5 104/50 mL) into
Figure 3. IL22RA1hi subpopulation of pancreatic cancer cell possesses higher capability for tumorigenicity and metastasis. A, Tumor-initiation assay was performed and compared between IL22RA1hi and IL22RA subpopulations with SW1990 cells. Cells from the indicated populations were injected into nude mice after FACS sorting. The number of successful tumor initiations after 8 weeks is shown. Representative images of H&E staining of indicated xenografts are shown. Scale bar, 50 mm. B, Sorted IL22RA1hi and IL22RA cells from SW1990 were examined by proliferation assay (mean SD; , P < 0.05, one-way ANOVA and Bonferroni post hoc test). C and D, KPC1199 representative FACS plots of IL22RA1 staining are shown. IL22RA1hi and IL22RA cells from KPC1199 were sorted and examined for indicated genes expression and proliferation capability. Data, mean SEM; , P < 0.05 (unpaired two-tailed Student t test). E, IL22RA1hi and IL22RA KPC1199 cells (104 each) were sorted and injected into pancreatic tail of C57BL/6J mice. Representative images and weight of orthotropic tumors originated from indicated group (n ¼ 6 of each group; mean SEM, unpaired two-tailed Student t test). F, Representative images of H&E, IHC (CK19), and immunofluorescence (IL22RA1) staining from indicated group are shown. G, IL22RA1hi and IL22RA cells (5 104) from KPC1199 were sorted and intrasplenic cells injected into C57BL/6J mice. Representative images of H&E and CK19 staining on the liver of mice injected with indicated cell. Scale bar, 100 mm. Quantification of CK19þ liver metastatic lesion (per 4 filed) from indicated population. Data are mean SEM, n ¼ 12 pooled from two independent experiments (unpaired two-tailed Student t test).
www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3297
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
He et al.
C57BL/6J mice. For the orthotopic model, tumors were harvested expression (Fig. 1E). Altogether, these data strongly suggest that 3 to 4 weeks after injection. Tumor weights were measured and the increased tumor IL22RA1 abundance is an important predictor of tumors processed for H&E, IHC (CK19), and immunofluores- poor survival in pancreatic cancer. cence (IL22RA1) analysis. For the liver metastasis model, liver tissues were harvested 4 to 6 weeks after injection and processed Pancreatic cancer cells with high IL22RA1 expression harbor þ for H&E and CK19 staining. CK19 staining was quantified and stemness properties represents the metastatic index. To elucidate the function of IL22RA1 in pancreatic cancer progression, we first assessed gene coexpression with IL22RA1 Statistical analysis using the TCGA pancreatic cancer database. We found that Pearson correlation together with P value was computed to IL22RA1 expression was positively correlated with developmental assess the degree of association between indicated genes. Survival and stemness genes, like PDX1, SOX9, HES1, and CD24 (Supple- curves were estimated by Kaplan–Meier methods. Two-way mentary Fig. S2A), which indicated that increased IL22RA1 ANOVA or one-way ANOVA together with Bonferroni post hoc expression might be associated with cancer stemness. Pancreatic test was used for multiple groups analysis. Unpaired Student t test CSCs can be enriched in vitro using the anchorage-independent was used to determine statistical significance for the remaining sphere formation assay (4, 25). Compared with adherent mono- experiments, and a P value of less than 0.05 was considered layer cells, IL22RA1 together with multiple core stem cell genes fi (NANOG, POU5F1, and SOX2) were dramatically elevated in signi cant. Values are expressed as mean SEM or mean SD þ þ þ (Prism 6; GraphPad Software). Unless indicated, results are from sphere cultures (Fig. 2A and B). CD44 CD24 ESA triple positive at least two or three independent experiments. cells are characterized as pancreatic CSCs (5); we observed that IL22RA1 was upregulated in these cells compared with non-CSCs (CD44 CD24 ESA ; Supplementary Fig. S2B). These results Results indicate that IL22RA1 is associated with cancer stemness in PDAC. IL22RA1 is upregulated in pancreatic cancer and is negatively We determined whether IL22RA1hi cells had cancer stemness associated with prognosis of patients with PDAC properties. ALDH1 is a functional marker of stemness. IL-22RA1hi In previous studies, we have shown that the IL22/IL22RA1 cells possessed higher ALDH activity compared with IL22RA1 signaling axis plays a critical but distinct role in the pathogenesis cells in multiple PDAC cell lines and primary tumor cells (Fig. 2C of acute and chronic pancreatitis (14, 15). It is widely accepted and D; Supplementary Fig. S3A and S3B). This was comparable þ þ þ that chronic inflammatory conditions of the pancreas are risk with the analysis of CD44 CD24 ESA cells as shown in Sup- factors for the development of PDAC (22, 26). To extend our plementary Fig. S3C. Furthermore, IL22RA1 and IL22RA1hi cells understanding of IL22RA1 in pancreatic disease, here we inves- were sorted from SW1990 and PDC0034 pancreatic cancer cell tigated the function of IL22RA1 in pancreatic cancer. We first lines and qRT-PCR analysis revealed that the expression of core compared IL22RA1 protein levels between PDAC cell lines and stemness genes (SOX2, NANOG, POU5F1, STAT3) was upregu- normal pancreatic ductal cell lines (HPDE and hTERT-HPNE). lated in IL22RA1hi cells compared with IL22RA1 cells (Fig. 2E). IL22RA1 was elevated in most PDAC cells compared with normal To support this, the sphere formation assay showed that the ductal cells (Fig. 1A). IL22RA1 is mainly expressed by pancreatic clonogenic potential of IL22RA1hi cells was superior to that of acinar cells and stellate cells under homeostatic and inflammatory IL22RA1 cells (Fig. 2F). Moreover, IL22RA1hi cells possessed conditions, respectively (14, 15). To determine the distribution of stronger resistance to the chemotherapy drug gemcitabine (Fig. IL22RA1 expression in tumors, human pancreatic cancer tissues 2G; Supplementary Fig. S3D). These results demonstrated that were analyzed by immunofluorescence staining of IL22RA1. pancreatic cancer IL22RA1hi cells display stemness properties. IL22RA1 was dominantly expressed by PDAC cells, demonstrated by its coexpression with CK19 (Fig. 1B). Moreover, the expression IL22RA1hi subpopulation of pancreatic cancer cells possesses of IL22RA1 was heterogeneous among the PDAC tumor cells. To higher capability for tumorigenicity and metastasis confirm the intratumoral heterogeneity of IL22RA1, PDAC cells The in vivo tumorigenicity of IL22RA1hi cells compared with from primary tumors were isolated and analyzed by flow cyto- IL22RA1 cells was explored using limiting-dilution tumor assays metry staining for IL22RA1. Consistent with immunofluorescence in nude mice. IL-22RA1hi cells possessed enhanced tumor-initi- staining, intratumoral heterogeneity of IL22RA1 was shown ating capability in vivo (Fig. 3A) despite showing significantly (about 4%–32% depending on the antibody isotype). In addi- lower proliferative capacity in vitro compared with IL22RA1 cells tion, a similar IL22RA1 expression pattern was observed in PDAC (Fig. 3B). Xenografts derived from these two groups had no cell lines and our established pancreatic PDCs (Fig. 1C; Supple- obvious histology differences (Fig. 3A). mentary Fig. S1A and S1B; Supplementary Table S2). IL22, the major ligand of IL22RA1 is only expressed by immune Next, we examined the clinical significance of elevated IL22RA1 cells (27). Given the immunodeficiency of nude mice, we further in pancreatic cancer. When patients were divided into "high" and examined the tumorigenicity of IL22RAhi cells using a PDAC "low/median" IL22RA1 expression based on the median mRNA orthotopic model in immunocompetent C57BL/6J mice. To this value of IL22RA1, we observed that elevated IL22RA1 expression end, IL22RA1hi and IL22RA1 cells sorted from KPC1199, a was associated with poor patient survival in the Renji cohort (Fig. mouse pancreatic cancer cell line generated from KPC mice 1D). This cohort includes 50 patients with pancreatic cancer with (C57BL/6J background), were examined by qRT-PCR analysis. clinical and pathologic information. Tumor differentiation state is Consistent with human PDAC cells, IL22RA1hi cells from also positively associated with IL22RA1 expression (Supplemen- KPC1199 cells expressed higher core stemness genes, including tary Table S3). Comparable results were observed from The Nanog, Pou5f1, Sox2, and Epcam (Fig. 3C). In line with this, these Cancer Genome Atlas (TCGA) pancreatic cancer database, in cells exhibited lower proliferation in vitro, but enhanced tumor- which overall survival was shorter in patients with high IL22RA1 igenicity in vivo (Fig. 3D and E). Interestingly, the orthotopic
3298 Cancer Res; 78(12) June 15, 2018 Cancer Research
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
IL22RA1/STAT3 Promotes Stemness in PDAC
Figure 4. IL22/IL22RA1 axis promotes stemness of pancreatic cancer cells. A, SW1990 cells were treated with vehicle (VE) or different IL22RA1 ligands (IL22, IL20, and IL24) for 24 hours before ALDEFLUOR analyses. B, Sphere assay was performed in SW1990 under vehicle or IL22 treatment. Representative images, number, and diameter of spheres are shown (n 3; mean SEM). C, SW1990 and PDC0034 were cultured with vehicle or IL22 for 6 hours. The relative mRNA of indicated genes was detected by qPCR assay. Data, mean SEM ( , P < 0.01; , P < 0.001, compared with vehicle). D, SW1990 cells were treated with vehicle or IL22 for indicated time. The expression of SOX2, NANOG, IL22RA1, and b-actin was examined by immunoblotting. E, Sphere assay was performed with shIL22RA1 or shCON- transfected SW1990 cells in the presence of IL22. Results are shown as sphere images and quantification of sphere number in triplicate (mean SD). IL22RA1 knockdown efficiency was confirmed with immunoblotting. F, IL22RA1hi and IL22RA cells from SW1990 were sorted for sphere assay under vehicle or IL22 treatment. Representative images and quantification of sphere number are shown (mean SEM). G, IL22RA1hi and IL22RA cells from SW1990 were sorted and treated with vehicle or IL22. The relative expression of indicated genes was detected by qPCR after 6 hours. Mean SD; , P < 0.05; , P < 0.01; , P < 0.001. H, IL22 level in adjacent normal tissue and pancreatic tumor was examined by flow cytometry. Cells were pregated with live/dead and CD45 staining. n ¼ 8–10; data are mean SEM; , P < 0.01; ns, nonsignificant.
www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3299
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
He et al.
3300 Cancer Res; 78(12) June 15, 2018 Cancer Research
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
IL22RA1/STAT3 Promotes Stemness in PDAC
tumor derived from IL22RA1hi cells possessed similar histologic enhanced stemness of cells with high expression of IL22RA1 (Fig. features with primary pancreatic cancer (Fig. 3F), indicating 4F and G; Supplementary Fig. S5E). More importantly, the IL22RA1hi cells have self-renewal and repopulation abilities in expression of IL22RA1 could be upregulated by IL22 in IL22RA1hi the pancreatic tumor microenvironment. cells (Fig. 4G), providing positive feedback for stemness mainte- Cancer cells with stemness features play a vital role in metas- nance in IL22RA1hi cells. tasis in a broad spectrum of malignancies (4, 7). We next deter- Having established that IL22 promotes pancreatic cancer stem- mined whether PDAC IL22RA1hi cells had higher metastatic ness, we determined whether IL22 was elevated in pancreatic capability compared with IL22RA1 cells. The epithelial–mesen- cancer. Elevated IL22 has been reported in primary pancreatic chymal transition (EMT) is a process by which epithelial cells lose cancer previously (29, 30). Consistent with these previous studies, þ their cell polarity and cell–cell adhesion to initiate cancer cell we confirmed that IL22 leukocytes were increased and IL22 was þ metastasis (25). IL22RA1hi cells showed higher expression of dominantly expressed by CD4 T cells in the pancreatic cancer EMT-related transcript factors than IL22RA1 cells (Supple- microenvironment (Fig. 4H; Supplementary Fig. S5F). mentaryFig.S4).IL22RA1hi and IL22RA1 KPC1199 cells (5 104) were injected intrasplenically into C57BL/6J mice. After IL22 promotes pancreatic cancer stemness and IL22RA1 þ 35 days, liver tissues were examined for histology and CK19 expression via STAT3 activation PDAC lesions. Mice injected with IL22RA1hi cells had more and Next, we explored the molecular mechanisms by which IL22 þ larger CK19 metastatic lesions in the liver when compared promotes pancreatic cancer stemness. STAT3 plays a critical role with mice injected with IL22RA1 cells (Fig. 3G). Taken togeth- in bridging the cross-talk between cancer cells and immune er, these results revealed the enhanced malignancy of IL22RA1hi cells in the tumor microenvironment (31–34). Moreover, cells in vivo and prompted us to further characterize their the majority of PDAC show constitutive activation of STAT3 regulation in vitro. (34, 35). In line with previous reports, using immunoblotting and flow cytometry assays, we observed that IL22 activated IL22/IL22RA1 signaling axis promotes stemness of pancreatic STAT3 in pancreatic cancer cells (Fig. 5A and B). Moreover, cancer cells STAT3 expression was elevated upon long-term IL22 exposure IL22RA1 executes its function through activation by various (Fig. 5C). We sought to examine whether the effect of IL22 in ligands, including IL22, IL20, and IL24. As IL22 has been reported PDAC stemness was STAT3 dependent. To this end, we manip- to promote cancer stemness in colon cancer (28), here we ulated STAT3 activation (using specific inhibitors) or expres- hypothesized that IL22 might enhance stemness in pancreatic sion (through shRNA knockdown) in pancreatic cancer cells. As cancer. In support of this, IL22 significantly increased ALDH shown, both inhibition of STAT3 activity and decreased STAT3 activity in PDAC cells compared with other ligands (Fig. 4A). expression resulted in reduced sphere number upon IL22 Next, we examined whether IL22 can enhance sphere formation of treatment (Fig. 5D; Supplementary Fig. S6A). Interestingly, IL6 PDAC cells. IL22 increased the number and size of spheres for all activated STAT3 but failed to promote stemness of PDAC cells tested cells (Fig. 4B; Supplementary Fig. S5A and S5B) without (Supplementary Fig. S6B–S6D). IL6/STAT3–triggered synthesis altering cell proliferation (Supplementary Fig. S5C). IL22 of SOCS3 is expected to shut down IL6-mediated (36) but not enhanced mRNA (Fig. 4C) and protein (Fig. 4D) expression of IL22-mediated STAT3 activation in the long term, which might multiple core stem cell genes, including NONOG, SOX2, result in the observed differences between IL6 and IL22 (Sup- and POU5F (OCT3/4), as well as IL22RA1 in SW1990 and plementary Fig. S6E). These observations suggest that IL22- PDC0034 cells. Moreover, IL22RA1 knockdown abrogated mediated STAT3 activation is necessary and relatively specificin IL22-induced sphere formation and core stem cell gene expression promoting PDAC stemness. (Fig. 4E; Supplementary Fig. S5D). These data showed that the We further investigated the effects of IL22 activation of STAT3 IL22/IL22RA1 signaling axis contributes to stemness maintenance on the expression of stemness-associated genes. STAT3 knock- in PDAC cells. down reduced the expression of core stemness genes in the We have shown that IL22 promoted pancreatic cancer stemness presence of IL22 (Fig. 5E). Herein, we speculated that STAT3 via the IL22/IL22RA1 signaling pathway. We next determined might directly bind to promoters of stemness-associated genes whether only IL22RA1hi cells, which have stemness features, and subsequently induce their expression. We found several could respond to activation by IL22. IL22RA1hi and IL22RA1 predictions for STAT3 binding to SOX2 and NANOG promoter cells from SW1990 cells were treated with IL22. IL22 preferentially regions (http://www.sabiosciences.com/chipqpcrsearch.php). ChIP
Figure 5. IL22 promotes pancreatic cancer stemness and IL22RA1 expression via STAT3 activation. A, SW1990 cells were treated with IL22 (50 ng/mL) for indicated time points. The amount of phosphorylated STAT3, STAT3, and b-actin protein was detected by immunoblotting. B, SW1990 cells were treated with IL22 (50 ng/mL) for 15 minutes, and phosphorylated STAT3 (Y705) was determined by flow cytometry. Data are representative FACS plot and quantification of p-STAT3 level (mean SEM). C, SW1990 cells were treated with IL22 (50 ng/mL) for 24 and 48 hours. STAT3 and b-actin protein were detected by immunoblotting. D, Sphere assay was performed with shSTAT3 or scrambled shRNA (shCON)-transfected SW1990 cells in the presence of IL22. STAT3 knockdown efficiency was examined by immunoblotting. Representative sphere images and quantification of sphere number from two independent experiments are shown (mean SD; , P < 0.01). E, SW1990 cells expressing shSTAT3 or shCON were cultured with vehicle (VE) or IL22 for 6 hours. Cells were prefasted before IL22 stimulation. The relative mRNA level of NANOG and SOX2 was normalized with vehicle from shCON group. Mean SD; , P < 0.05; , P < 0.01. F, STAT3-ChIP assay was performed to examine STAT3 binding at SOX2 and NANOG promoters in SW1990 cells upon vehicle or IL22 treatment. One of two independent experiments is shown. G, IL22RA1 mRNA expression was detected in SW1990 cells after vehicle and STAT3 inhibitors (stattic or S3I-201) treatment in the presence of IL22. The relative mRNA level of IL22RA1 was normalized with vehicle from shCON group. Data, mean SD; , P < 0.05; , P < 0.01. H, STAT3-ChIP assay was performed to examine STAT3 binding at IL22RA1 promoters in SW1990 cells upon vehicle or IL22 treatment. One of two independent experiments is shown. Unpaired two-tailed Student t test was used for all data analyses. ns, nonsignificant.
www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3301
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
He et al.
3302 Cancer Res; 78(12) June 15, 2018 Cancer Research
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
IL22RA1/STAT3 Promotes Stemness in PDAC
demonstrated that IL22 increased STAT3 binding to several sites Discussion of the SOX2 and NANOG promoters (Fig. 5F), suggesting that Chronic inflammation is a key feature of pancreatic cancer STAT3 might directly activate stemness genes. Moreover, IL22 (22, 23). Many immune cells exist in tumor microenvironment failedtoinduceIL22RA1 expression in the presence of STAT3 (41), indicating their indispensable role in PDAC initiation and inhibitors stattic or S3I-201 (Fig. 5G). In addition, several progression. Elucidation of the cross-talk between immune and STAT3 potential binding sites were predicted on the IL22RA1 PDAC cells, especially pancreatic CSCs, would enable better promoter, and direct binding of STAT3 to the IL22RA1 pro- understanding of pancreatic cancer progression and provide a moterwasfurtherverified using ChIP (Fig. 5H). Thus, IL22 potential therapeutic strategy for targeting PDAC. In our current promotes pancreatic cancer stemness and IL22RA1 expression study, we aimed to investigate how IL22/IL22RA1 contributed to via STAT3 activation. PDAC development and progression. IL22 can be produced by Th17 cells, Th22, gd T cells, NKT cells, STAT3 activity is required for IL22RA1hi population and innate lymphoid cells (42–45). Knowledge of IL22 biology maintenance has evolved rapidly since its discovery in 2000 (46), and a role for STAT3 is essential for stemness maintenance in pancreatic IL22 has been identified in numerous tissues, including the cancer (37–40). As showed in Fig. 2B, STAT3 mRNA expression intestines, lung, liver, kidney, thymus, pancreas, and skin. The was elevated in IL22RA1hi populations. Moreover, STAT3 activa- expression IL22 and its receptor IL22RA1 is limited to hemato- tion was significantly higher in IL22RA1hi PDAC cells (Fig. 6A). poietic cells and epithelial/stromal cells, respectively (27). The Therefore, we next determined whether STAT3 was required for IL22/IL22RA1 signaling axis enables communication between expression of IL22RA1 in the absence of IL22. The percentage of immune and epithelial/stromal cells. In our previous studies, we þ IL22RA1 cells was significantly reduced by STAT3 inhibitors (Fig. found IL22RA1 is dominantly expressed on pancreatic acinar cells þ 6B). In addition, STAT3 inhibitors downregulated IL22RA1 during acute pancreatitis, which enabled IL22 immune cells to mRNA expression together with core stem cell genes in IL22RA1hi promote tissue repair by targeting acinar cells (14). However, we þ cells (Fig. 6C). As showed in Fig. 5E, STAT3 inhibitors did not also found IL22 immune cells, elevated in patients with chronic downregulate basal levels of IL22RA1, which is partially due to pancreatitis with a smoking history, promoted fibrosis via trig- low levels of STAT3 activity caused by serum-free pretreatment. gering pancreatic stellate cell activation (15). In all, we observed The above results demonstrated that STAT3 activation was indis- that the function of IL22 under different pathologic conditions pensable for the maintenance of IL22RA1hi cells, both in the could be different even in the same tissue, which was partly absence or presence of IL22 (Figs. 6B and C and 5G). Thus, STAT3 dependent on where IL22RA1 is expressed. could be a potential target for those cells with high expression of Two recent reports identified IL22 and IL22RA1 were elevated IL22RA1. We therefore determined whether STAT3 inhibition in pancreatic cancer (29, 30). However, little is known about how could sensitize IL22RA1hi cells to chemotherapy. The STAT3 the IL22/IL22RA1 signaling axis is involved in pancreatic cancer inhibitor stattic specifically inhibited phosphorylated STAT3 development. Here, we confirmed that IL22RA1 expression was (Y705) in a dose-dependent manner (0.1–1 mmol/L; Supplemen- elevated in pancreatic cancer using PDAC cell lines, and elevated tary Fig. S6F). IL22RA1hi cells were exposed to gemcitabine alone, expression of IL22RA1 was associated with poor survival in both titrated concentrations of stattic alone, or both. IL22RA1hi PDAC our Renji and TCGA pancreatic tumor cohorts. To further under- cells could be targeted by combined treatment with gemcitabine stand how IL22RA1 is involved in PDAC development, we ana- and stattic in vitro (Fig. 6D). The synergistic effect of gemcitabine lyzed genes coexpressed with IL22RA1 in the TCGA pancreatic and stattic on IL22RA1hi cells was examined in vivo using an tumor database. We observed that multiple stemness-related orthotopic model. Combined treatment with gemcitabine and genes were tightly associated with IL22RA1, which suggested that stattic significantly reduced IL22RA1hi cell-derived tumor pro- IL22RA1 may contribute to cancer stemness. To confirm our gression in vivo (Fig. 6E) compared with each drug alone. Besides hypothesis, pancreatic CSCs were enriched via sphere formation þ þ þ maintaining stemness, IL22RA1hi cells would be expected to and CD44 CD24 ESA cell sorting (5, 25). IL22RA1 together differentiate in vivo. Thus, it was not surprising to see the inhib- with core stemness genes were upregulated in the enriched CSCs. itory effect of gemcitabine or stattic alone on IL22RA1hi cell- By sorting and analyzing IL22RA1hi cells, we found IL22RA1hi derived tumors. Taken together, these data show the significance PDAC cells harbored stemness features including higher expres- of STAT3 in maintaining the IL22RA1hi population and in medi- sion of core stemness genes, sphere formation ability, ALDH ating IL22/IL22RA1–promoted stemness and provides a thera- activity, and gemcitabine resistance when compared with peutic strategy for eliminating IL22RA1hi PDAC cells as depicted IL22RA1 cells. In addition, in vivo studies further confirmed the by the schematic model (Fig. 6F). higher capability for tumor formation and metastasis of
Figure 6. STAT3 activity is required for IL22RA1hi population maintenance. A, Phosphorylated STAT3 (p-STAT3) level was detected in sorted IL22RA1 and IL22RA1hi cells from SW1990 with intracellular staining. Representative FACS plot is shown. B, Flow cytometry analysis of IL22RA1 expression in SW1990 cells after treatment þ of vehicle (VE) and STAT3 inhibitors (stattic 1 mmol/L; S3I-201 50 mmol/L). Representative FACS plots and quantification of IL22RA1 cells (%) from three independent experiments are shown (mean SEM; , P < 0.05, unpaired two-tailed Student t test). C, IL22RA1hi cells from SW1990 were sorted for vehicle or stattic treatment (1 mmol/L, 8 hours). The relative expression of indicated genes was detected by qPCR assay. Data, mean SD (one-way ANOVA test). D, IL22RA1hi cells from SW1990 were treated with titrated doses of stattic (0–1 mmol/L) together with or without gemcitabine (100 nmol/L) for 72 hours. Percentage of cell viability was determined (mean SD; , P < 0.05, one-way ANOVA test). ns, nonsignificant. E, IL22RA1hi sorted from KPC1199 cells (104 each) were injected into pancreatic tail of C57BL/6J mice. After 8 days, mice were grouped for indicated treatments. Representative images of orthotopic tumors and H&E staining from indicated group are shown. F, The scheme of IL22/IL22RA1/STAT3 signaling on IL22RAhi population maintenance and the targeting strategy for IL22RAhi cells with CSC-like potentials.
www.aacrjournals.org Cancer Res; 78(12) June 15, 2018 3303
Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2018 American Association for Cancer Research. Published OnlineFirst March 23, 2018; DOI: 10.1158/0008-5472.CAN-17-3131
He et al.