A RIPK3-PGE Circuit Mediates Myeloid-Derived Suppressor ... · Chem Express), or ONO-AE3-208 (5 mg/kg, catalog no. 402473-54-5, Med Chem Express) was injected intraperitone- ally

Tumor Biology and Immunology

A RIPK3-PGE2 Circuit Mediates Myeloid-DerivedSuppressor Cell–Potentiated ColorectalCarcinogenesisGuifang Yan1,2, Huakan Zhao1,2, Qi Zhang1,2, Yu Zhou1,2, Lei Wu1,2, Juan Lei1,2,Xiang Wang1,2, Jiangang Zhang1,2, Xiao Zhang1,2, Lu Zheng3, Guangsheng Du4,Weidong Xiao4, Bo Tang5, Hongming Miao6, and Yongsheng Li1,21

Abstract

Receptor-interacting protein kinase 3 (RIPK3) isessential for mucosal repair in inflammatory boweldiseases (IBD) and colorectal cancer. However, itsrole in tumor immunity is unknown. Here, wereport that decreased RIPK3 in colorectal cancercorrelates with the accumulation of myeloid-derived suppressor cells (MDSC). Deficiency ofRIPK3 boosted tumorigenesis via accumulation andimmunosuppressive activity of MDSCs. Reductionof RIPK3 in MDSC and colorectal cancer cells eli-cited NFkB-transcribed COX-2, which catalyzed thesynthesis of prostaglandin E2 (PGE2). PGE2 exacer-bated the immunosuppressive activity of MDSCsand accelerated tumor growth. Moreover, PGE2suppressed RIPK3 expression while enhancingexpression of NFkB and COX-2 in MDSCs andcolorectal cancer cells. Inhibition of COX-2 or PGE2receptors reversed the immunosuppressive activityof MDSCs and dampened tumorigenesis. Patientdatabases also delineated the correlation of RIPK3and COX-2 expression with colorectal cancer sur-vival. Our findings demonstrate a novel signalingcircuit by which RIPK3 and PGE2 regulate tumorimmunity, providing potential ideas for immuno-therapy against colorectal cancer.

Significance: A novel signaling circuit involvingRIPK3 and PGE2 enhances accumulation andimmunosuppressive activity of MDSCs, implicatingits potential as a therapeutic target in anticancerimmunotherapy.

GraphicalAbstract:http://cancerres.aacrjournals.org/content/canres/78/19/5586/F1.large.jpg.CancerRes; 78(19);5586–99.�2018AACR.

© 2018 American Association for Cancer Research

GzmBIFN-γ

CD8+ T CellProliferation Activation

A RIPK3-PGE2 signaling circuit in the tumor microenvironment mediates MDSC-potentiatedtumorigenesis of colorectal cancer.

COOH

COOH

O

HO

PGE2

OH

RIPK3Arg-1

AAAccumulation

Immunosuppressiveactivity

G

Tumor Cell

COX-2NFκBMDSC

COX-2

1Institute of Cancer, Xinqiao Hospital, Third Military Medical University, Chongq-ing, China. 2Clinical Medicine Research Center, Xinqiao Hospital, Third MilitaryMedical University, Chongqing, China. 3Department of Hepatobiliary Surgery,Xinqiao Hospital, Third Military Medical University, Chongqing, China. 4Depart-ment of General Surgery, Xinqiao Hospital, Third Military Medical University,Chongqing, China. 5Department of Gastroenterology, Xinqiao Hospital, ThirdMilitary Medical University, Chongqing, China. 6Department of Biochemistry andMolecular Biology, Third Military Medical University, Chongqing, China.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

G. Yan, H. Zhao, and Q. Zhang contributed equally to this article.

Corresponding Authors: Hongming Miao, Department of Biochemistry andMolecular Biology, Third Military Medical University, Chongqing 400037, China.Phone: 8623-6877-4705; Fax: 8623-6877-4705; E-mail:[email protected]; and Yongsheng Li, Third Military Medical University,2 Xinqiao Street, Shapingba District, Chongqing 400037, China. Phone: 8623-6877-4705; Fax: 8623-6877-4705; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-17-3962

�2018 American Association for Cancer Research.

CancerResearch

Cancer Res; 78(19) October 1, 20185586

on June 6, 2020. © 2018 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 16, 2018; DOI: 10.1158/0008-5472.CAN-17-3962

http://crossmark.crossref.org/dialog/?doi=10.1158/0008-5472.CAN-17-3962&domain=pdf&date_stamp=2018-9-20

http://cancerres.aacrjournals.org/content/canres/78/19/5586/F1.large.jpg

http://cancerres.aacrjournals.org/

IntroductionColorectal cancer is one of the most common malignant

tumors and the third leading cause of cancer-associated mor-tality in the world (1). The 5-year survival rate is only approx-imately 10% for the patients with advanced colorectal cancer(2). Exploring the pathogenesis and effective therapeutic targetsof colorectal cancer is of great clinical significance. Inflamma-tory bowel diseases (IBD) are recognized as precancerous dis-eases of colorectal cancer. The colorectal cancer–infiltratingimmune cells, including myeloid-derived suppressor cells(MDSC) promote the carcinogenesis, which is one of the mostimportant causes for tumor progression and therapeutic failure(3–5). MDSCs can induce immune tolerance by overexpressingarginase 1 (Arg-1), inducible nitric oxide synthase (iNOS orNOS2), and reactive oxygen species (ROS) to suppress theactivation of CTLs. MDSCs can also differentiate towardtumor-associated macrophages and promote the immunosup-pressive function of regulatory T cells. In addition, MDSCssecrete prostaglandin E2 (PGE2), calcium-binding proteinS100A8/A9, FGFs, matrix metalloproteinases, TGFb, VEGF, andother cytokines to promote tumor proliferation, angiogenesis,and metastasis (6). Therefore, addressing the mechanismsregulating MDSCs will provide new ideas for the immunother-apy of colorectal cancer.

Inflammation initiates necroptosis, which parallels withcaspase-mediated apoptosis and NFkB-mediated proliferation,and plays an essential role in carcinogenesis (7). Receptor-interacting protein kinase 3 (RIPK3) is a central regulatorymolecule for necroptosis (8), whereas its role in tumor immu-nity remains unknown. It has been reported that RIPK3 pro-motes the mucosal repair in IBD (9). More importantly, RIPK3also inhibits the tumorigenesis of colorectal cancer and theexpression of proinflammatory factors including S100A8, che-mokine (C-X-C motif) ligand 1 (CXCL1), IL1b, IL6, and TNFa(10). Because these proinflammatory factors correlate with theaccumulation and maintenance of MDSCs (11), the abovestudies suggest that RIPK3 may regulate the tumor-infiltratingMDSCs.

Here, we demonstrate that the downregulation of RIPK3 intumor-infiltrating MDSCs potentiates NFkB activation and COX-2–derived PGE2 production. PGE2, in turn, further reduces RIPK3and promotes the immunosuppressive activity of MDSCs andcarcinogenesis. Therapy targeting this signaling circuit involvingRIPK3 and PGE2 potently blunts the accumulation and activity ofMDSCs andprotects the colorectumagainstmalignancy.Our dataprovide molecular basis for RIPK3 regulating MDSCs and tumorimmunity, and suggest potential immunotherapeutic idea forcolorectal cancer.

Materials and MethodsHuman databases

The correlations between RIPK3 gene expression and colorectalcancer were determined through analysis of Kaiser colon andSkrzypczak colorectal cancer datasets, respectively, which areavailable at Oncomine (http://www.oncomine.org/).

The National Center for Biotechnology Information GeneExpression Omnibus databases GSE21510 (12) and GSE17536(13) containing 148 and 177 patients with colorectal cancer wereevaluated for the correlation of RIPK3 and indicated genes andsurvival, respectively.

Animal experimentsC57BL/6 mice were purchased from the Chinese Academy of

Medical Sciences (Beijing, China). RIPK3 knockout (KO) micewith C57BL/6 background were kindly provided by XiaodongWang andZhirong Shen (National Institute of Biological Sciences,Beijing, China). All wild-type (WT) and KOmice were age and sexmatched, and cages were randomly assigned to the treatmentgroups. All animal procedures were conducted in accordancewiththe National and International Guidelines for the Care andUse ofLaboratory, approved by the Animal Care and Use Committee ofThird Military Medical University (Chongqing, China) and com-plied with the Declaration of Helsinski.

The model of acute IBD was established by feeding C57BL/6mice with 2% dextran sodium sulfate (DSS) dissolved in sterilepure water for 6 days. For colorectal cancer induction, 6-week-old C57BL/6 mice were injected intraperitoneally with 10 mg/kg azoxymethane (AOM, catalog no. 25843-45-2, Sigma-Aldrich) and after 7 days, they were fed with sterile pure watercontaining 2% DSS (catalog no. 0216011080, MW 40,000-50,000, MPbio) for three cycles. Colons and spleens wereremoved upon sacrifice at indicated interval. Macroscopictumors were measured with calipers. Portions of the distalcolons were either frozen at �80�C or fixed with formaldehydeand embedded in paraffin for histologic analysis. For someindicated experiments, GSK872 (0.75 mg/kg, catalog no. 2673,BioVision), AH-6809 (5 mg/kg, catalog no. HY-10418, MedChem Express), or ONO-AE3-208 (5 mg/kg, catalog no.402473-54-5, Med Chem Express) was injected intraperitone-ally every 2 days until the mice were sacrificed. Anti-Gr-1 (12.5mg/kg, catalog no. BE0075, BioXcell) or CXCR2 antagonist(CXCR2-a) SB225002 (4 mg/kg, catalog no. 182498-32-4, MedChem Express) was injected intraperitoneally every 2 days fromthe third cycle until the mice were sacrificed. In the in vivoaspirin (ASA) treatment experiments, mice were subjected0.02% ASA (catalog no. 50-78-2, Med Chem Express) contain-ing water during the colorectal cancer induction.

Body weight, stool consistency, and rectal bleeding were mon-itored daily. The values before DSS exposure were recorded asbaseline. The diarrhea scores were calculated as follows: 0, stoolformed pellets; 1, diarrhea; 2, hematochezia; 3, serious hemato-chezia or archoptosis; 4, die. The average total scores were calcu-lated in each cycle for 21 days.

For the chimerism experiments, C57BL/6 WT or RIPK3-KOmice were irradiated (850 cGy) and injected intravenously with1� 107 bone marrow cells from congenic WT or RIPK3-KOmice,respectively. For two weeks after engraftment, mice were givenantibiotic water (containing trimethoprim and sulfamethoxa-zole). After 7 weeks, the peripheral MDSCs were analyzed toconfirm chimerism.

Flow cytometryThe single-cell suspension was prepared by mechanic disper-

sion and enzymatic digestionof indicated tissues. For extracellularstaining of target proteins, cells (1 � 106/mL) were preincubatedin a mixture of PBS, 1% FBS, and 0.1% (w/v) sodium azidewith FcgIII/IIR-specific antibody to block nonspecific bindingand stained with different combinations of fluorochrome-cou-pled antibodies including CD45 (catalog no. 103108), CD11b(catalog no. 101208/101224), Gr-1 (catalog no. 108426), Ly6G(catalog no. 127618), Ly6C (catalog no. 128007), F4/80 (catalogno. 123110), CD11c (catalog no. 117308), CD206 (catalog no.

RIPK3-PGE2 Circuit Regulates MDSCs

www.aacrjournals.org Cancer Res; 78(19) October 1, 2018 5587



http://www.oncomine.org/


141704), CD3 (catalog no. 100206),CD8a (catalog no. 301030),CD4 (catalog no. 100422), CD33 (catalog no. 303304), HLA-DR(catalog no. 307606), and APC Annexin V Apoptosis DetectionKit with 7-AAD (catalog no. 640930) from BioLegend, FixableViability Dye Efluor 780 (catalog no. 65-0865-14) fromeBioscience, ROS (catalog no. s0033-1) from Beyotime. Forintracellular staining of RIPK3 (catalog no. 95702, Cell SignalingTechnology; ab152130, Abcam), COX-2 (catalog no. 12282, CellSignaling Technology), Arg-1 (catalog no. 42284, GeneTex),NOS2 (MA5-17139, Thermo Fisher Scientific), IFNg (catalog no.505810, BioLegend), and granzyme B (GzmB, catalog no.515403, BioLegend), we followed the manufacturers' protocolsafter cells were treated with PGE2 (catalog no. 363-24-6, Cay-man Chemical), GSK872 (catalog no. 2673, BioVision), AH-6809 (catalog no. HY-10418, Med Chem Express), Caffeic AcidPhenethyl Ester (catalog no. S7414, Selleck Chemicals), ASA(catalog no. 50-78-2, Med Chem Express), N-Hydroxy-nor-L-arginine (NHNL, catalog no. 399275, Calbiochem), bevacizu-mab (Roche), cetuximab (Merck), nimotuzumab (BiotechPharma), irinotecan (CPT-11, Pfizer), oxaliplatin (catalog no.HY-17371, Med Chem Express), 5-Fluorouracil (5-FU, catalogno. HY-9006, Med Chem Express), or gemcitabine (Eli Lilly andCompany), respectively. The fluorescence were determined on aFACSCanto II system (BD Biosciences) and analyzed withFlowJo software (Tree Star).

MDSC induction in vitroMDSCs were isolated and induced as indicated previously (14,

15). Briefly, bone marrow cells from WT or RIPK3�/� mice werestained by anti-mouse Gr-1 particles (catalog no. 558111, BDBiosciences), which were optimized for positive selection andcollected together using the BD IMag Cell Separation Magnet.Freshly harvested MDSCs were incubated in RPMI1640 mediumcontaining 5% FBSwith GM-CSF (20 ng/mL, 315-03, PeproTech)for 48 hours.

Cell cultureThemouse colorectal cancer CT26 cell line was purchased from

ATCC and authenticated via short tandem repeat profiling. Cellswere cultured in RPMI1640 (Gibco) supplemented with 10% FBS(Gibco) and 1% penicillin–streptomycin (Gibco). The CT26 cellswere routinely verifiedMycoplasma-free using MycAwayTM-ColorOne-Step Mycoplasma Detection Kit (Yeasen Bio-technol) andthe most recent date of testing was April 5, 2018. Cells were usedwithin 12 passages following thawing in all experiments.

CD8þ T-cell isolation, purification, and proliferation assayCD8þ T cells were isolated from the spleen of C57BL/6mice by

CD8þ T-cell Isolation Kit (catalog no. 480007, BioLegend). Bonemarrow–derived MDSCs were cocultured with CFDA-SE (56-carboxyfluorescein diacetatesuccinimidyl ester, CFSE; catalog no.2011-11-2, Dojindo) labeled CD8þ T cells (1� 106) at 10:1 in themedium containing anti-CD3 (1 mg/mL) and anti-CD28 (1 mg/mL). At day 3 post cocultivation, cells were harvested andCFSEþCD8þ T cells were detected by flow cytometry (FCM).

For the analysis of CD8þ T-cell function, CD8þ T cells werecocultured withMDSCs (5:1, 10:1, or 16:1) andwere harvested tostimulate using Cell Stimulated Cocktail (catalog no. 4303372,eBioscience) for 4 hours and then were collected for the deter-mination of GzmB and IFNg by FCM.

Western blot analysisSamples were lysed in RIPA buffer containing phenylmethyl-

sulfonylfluoride. The protein quantification was determined byBCA protein assay (catalog no. P0068, Beyotime), and equalamounts of proteins (40 mg) were subjected to SDS-PAGE(12% gels). After electrophoresis, proteins were transferred ontopolyvinylidene difluoride membranes (0.45 mm) in runningbuffer with 20% methanol. Nonspecific sites were blocked with5% (w/v) nonfat dried skimmed milk powder in TBST (2 mol/LTris-HCl buffer, pH 7.6; 0.05 mol/L NaCl; and 0.05% Tween-20)for 60 minutes at 37�C. The membranes were then incubatedovernight at 4�C with the following antibodies, which werediluted in TBST: anti-RIPK3 (1:1,000; catalog no. 2283, ProSci),anti-COX-2 (1:500; catalog no. 12282, Cell Signaling Technolo-gy), anti-p65 (1:500; catalog no. 41556, GeneTex), anti-PKA(1:1,000; catalog no. ab76238, Abcam), anti-Actin (1:1,000;catalog no. A1978, Sigma-Aldrich), Cell Signaling Technologyantibodies including anti-CREB (1:500; catalog no. 9197), anti-p-CREB (1:500; catalog no. 9198), anti-p-stat3 (1:1,000; catalog no.9145), anti-stat3 (1:1,000; catalog no. 4904), anti-p-stat6(1:1,000; catalog no. 56554s), and anti-stat6 (1:1,000; catalogno. 5397). After four washes in TBST, the membranes wereincubated with horseradish peroxidase–conjugated secondaryantibodies (catalog no. A0562, Beyotime) for 1 hour in TBST(dilution of 1:5,000). Protein bands were visualized by usingEnhanced Chemiluminescence (ECL) plus Western blottingdetection kit (catalog no. P0018-2, Beyotime).

Confocal microscopyMice or human tissues were fixed and permeabilized with

Fixation & Permeabilization Buffers (BD Biosciences) for 15minutes and then incubated with FC-block (BD Biosciences) for30minutes at room temperature. Subsequently, cells were stainedwith Gr-1 (1:50, catalog no. MAB1037, R&D Systems), RIPK3(1:100, catalog no. 95702, Cell Signaling Technology), or CD33(1:100, catalog no. ab213050, Abcam), RIPK3 (1:50, catalog no.ab152130, Abcam) for overnight at 4 �C, washed thrice with PBSbefore incubation with fluorochrome-associated secondary anti-bodies for 30 minutes with Alexa-488, and 647 (Bioss). After-wards, sections were washed thrice with wash buffer (BD Bios-ciences), and then were incubated with DAPI and mountedon slides using Prolong Gold antifade reagent (Beyotime). Thesectionswere imagedwith a Leica TCS SP5 laser scanning confocalmicroscope (LeicaMicrosystems). The colocalization and averageintensity were assessed by using Leica LASX (Microsystemssoftware).

IHC analysisColon and tumor tissues were fixed with formaldehyde. Par-

affin sections were stained with hematoxylin and eosin or sub-jected to IHC for Gr-1 (catalog no. MAB1037, R&D Systems),RIPK3 (catalog no. 2283, ProSci), and COX-2 (catalog no. 12282,Cell Signaling Technology).

Real-time PCRReal-time PCR was performed as previously (16). Total RNA

was extracted from cells with RNA queous Mico Kit (catalog no.00490515, Invitrogen). Real-time quantitative PCR was per-formed on a CFX384 system (Bio-Rad). The following primerswere used in this study: RIPK3-F: CAGTGGGACTTCGTGTCCG,RIPK3-R: CAA GCT GTG TAG GTA GCA CAT C; EP1-F: CTT AAC

Yan et al.

Cancer Res; 78(19) October 1, 2018 Cancer Research5588




CTG AGC CTA GCG GAT, EP1-R: ATG TGC CAT TAT CGC CTGTTG; EP2-F: GGA GGACTG CAAGAG TCG TC, EP2-R: GCG ATGAGA TTC CCC AGA ACC; EP3-F: GCT CAT GGG GAT CAT GTGTGT, EP3-R: CAC CAC CCCGAAGAT GAA CAT; EP4-F: ACC ATTCCT AGA TCG AAC CGT, EP4-R: CAC CAC CCC GAA GAT GAACAT; ACTIN-F: TGA CAGGATGCAGAAGGAGA; ACTIN-R: GTACTT GCG CTC AGG AGG AG.

PGE2 determinationCell supernatants were collected for evaluating PGE2 concen-

tration with UPLC/MS-MS as previously (16). Prior to sampleextraction, d4-PGE2 (500 pg) was added to permit quantification.Extracted samples were separated by an Acquity UPLC I-Classsystem (Waters) and mass spectrometry was performed on an ABSciex 6500 QTRAP. PGE2 was analyzed using scheduled multiplereaction monitoring (MRM). Data acquisitions were performedusing Analyst 1.6.2 software (Applied Biosystems).

Cell proliferation assayCell Counting Kit-8 (CCK8) assay was used to assess the

proliferation of MDSCs and CT-26 cells. For indicated experi-ments, 5 � 105 bone marrow–derived MDSCs or 5 � 103 CT-26cells were seeded in 96-well plates. After 48 hours, a batch of cellsin 100-mL medium was stained with 10 mL of CCK8 reagent(Dojindo) at 37�C for 2 hours. The data were quantified withan automatic plate reader (Thermo Fisher Scientific) at 450 nm.

Statistical analysisThe number of animals used in the experiments was estimated

to give sufficient power (>90%) on the basis of the effect sizesobserved in our preliminary data. The statistical analysis wasperformed using Excel (Microsoft), Origin 9.1 (OriginLab), orGraphPad Prism 7 (GraphPad Software). Statistical significancefor binary comparisons was assessed by 2-tailed Student t test. Forcomparison of more than 2 groups, ANOVA with Sidak multiplecomparisons test was used. For correlation analysis, Pearson'scorrelation coefficient was applied. Overall survival was calculat-edusing theKaplan–Meiermethod, and thedifferences in survivalcurves were analyzed using the log-rank test. All data are reportedas mean � SEM. The P value of 0.05 or less was consideredsignificant.

ResultsRIPK3 is downregulated in colorectal cancer–infiltratingMDSCs

The RIPK3 expression was first evaluated in colorectal cancerpatient databases with Oncomine, which showed a consistentdecrease of RIPK3 in colorectal cancer tissue (SupplementaryFig. S1A). We next employed AOM plus DSS-induced mousecolorectal cancer model (Fig. 1A, left; 17). The body weightreduced during the DSS treatment and rebound subsequentlyafter DSS withdrawn at each cycle (Supplementary Fig. S1B).Upon sacrifice after Day 90, the colorectum was collected andtumors were separated (Fig. 1A, right). We compared thepercentage of immune cells in tumor and colorectal tissue andfound that both leukocytes (CD45þ) and MDSCs (CD11bþGr-1þ) were significantly higher in tumor than in colorectal tissues(Fig. 1B and C). The tumor-infiltrating MDSCs showed lowerRIPK3, compared with MDSCs in the colorectum of tumor-bearing mice (Fig. 1D and E).

We collected clinical colorectal cancer and adjacent normaltissues and found that the accumulation of MDSCs was muchhigher. Consistently, RIPK3 expression inMDSCswas significantlysuppressed in the tumor microenvironment (TME) than in adja-cent tissue (Fig. 1F). Of interest, we found that RIPK3 expressionwas highest in the colorectum of IBD mice, whereas, was lower inthe colorectum of colorectal cancer mice and lowest in tumortissues (Supplementary Fig. S1C-S1E), suggesting a differentialRIPK3 expression pattern during the development of colorectalcancer. In addition, we evaluated other immune cells in colorectaland tumor tissue of colorectal cancer mice. CD8þ T cells anddendritic cells (DC, CD11cþ) were less while macrophages (Mf,F4/80þ) were more abundant in tumor than in colorectal tissues(Supplementary Fig. S1F). We also determined RIPK3 expressionbut did not found significant changes in DCs, Mfs, or T cellswhen compared with that in colorectal tissues (SupplementaryFig. S1G). We hence wondered whether the down-regulation ofRIPK3 in tumor-infiltrating MDSCs was caused by factors fromTME. We stimulated mouse bone marrow cells with supernatantsfrom CT26 colorectal cancer cells in vitro and found that the per-centage of MDSCs increased significantly, whereas, the expressionof RIPK3 was downregulated significantly (Fig. 1G). Together,these results indicated that RIPK3 was downregulated in tumortissues and colorectal cancer–infiltrating MDSCs.

Enhanced MDSC accumulation and tumorigenesis in RIPK3-deficient mice

To investigate the role of RIPK3 in the tumorigenesis ofcolorectal cancer, we employed RIPK3 KO mice. These miceshowed decreased body weight, higher diarrhea score, shortenedcolorectum length, increased tumor number in colorectum,heavier spleen, and significantly reduced survival, comparedwith WT mice (Fig. 2A–E; Supplementary Fig. S2A). The accu-mulation of MDSCs in tumor, colorectum, and spleen alsoincreased in KO mice (Fig. 2F–I). Of note, only the granulocyticMDSCs (g-MDSC, CD11bþLy6Gþ) increased in the tumorcompared with that in the colorectal tissues, which was notshared by the monocytic MDSCs (m-MDSC, CD11bþLy6Cþ;18), Mfs, DCs, or T cells (Fig. 2J and K).

We also used GSK872, a specific RIPK3 inhibitor (19) to treatthe WT mice. GSK872 significantly aggregated AOM plus DSSinduced weight loss, colorectum shortening, tumormass, spleno-megaly, and MDSC accumulation, while it did not alter theinfiltration ofMfs andDCs (Supplementary Fig. S2B-S2H). Theseresults demonstrated that RIPK3 deficiency promoted colorectalcarcinogenesis and MDSC infiltration.

Deficiency of RIPK3 promotes the proliferation andimmunosuppressive activity of MDSCs in vitro

The role of RIPK3 onMDSCswas next sought in vitro. We foundthat the percentage of MDSCs was much higher in RIPK3-KOgroup than inWT after mouse bonemarrow cells were stimulatedby GM-CSF (Fig. 3A), although there was no difference betweenWT and RIPK3-KO groups without stimulation (SupplementaryFig. S3A). RIPK3 absence in MDSCs also resulted in a modesthigher proliferation (Fig. 3B) but did not show significant changein cell death (Fig. 3C), whichwere consistent with the results fromGSK872-treated MDSCs (Supplementary Fig. S3B). The differen-tiation of MDSCs was also assessed in vitro. After induction withGM-CSF (20 ng/mL) for 48 hours, the percentage of Mfs, espe-cially M2 type Mfs (F4/80þCD206þ) were significantly higher in






Figure 1.

RIPK3 is downregulated in MDSCs of colorectal cancer tissues. A, Schematic of mice treated with AOM and DSS (Model 1, 2% DSS drinking for 5 days each cycle).The entity image of the colorectum after mice sacrifice is shown on the right. B, The percentage of leukocytes (CD45þ) in the tumor and colorectal tissue.C, The percentage of MDSCs (CD11bþGr-1þ) in CD45þ cells in the tumor and colorectal tissue. D and E, RIPK3 expression in MDSCs in the tumor and colorectal tissueby FCM (D) and confocal microscopy (E; RIPK3, red; Gr-1, green). F, RIPK3 expression in MDSCs in human colorectal cancer and adjacent tissue. G, After mousebone marrow cells were treated with PBS or tumor supernatants (from CT26 cells) for 48 hours, the percentage of CD11bþGr-1þ cells in bone marrow cells andRIPK3þ cells in CD11bþGr-1þ cells was determined. Data of B–G are expressed as mean � SEM (� , P < 0.05; �� , P < 0.01; �� , P < 0.0001, by Student t test).

Yan et al.





KO group, while DCs were lower as compared withWT (Fig. 3D),suggesting that RIPK3 absence in MDSCs promoted the M2-likedifferentiation upon GM-CSF induction.

In addition, Arg-1 but notNOS2orROS increased in theRIPK3-KO and GSK872-treated MDSCs (Fig. 3E and F; SupplementaryFig. S3C). The cocultivation of RIPK3-KO and GSK872-treatedMDSCs significantly dampened anti-CD3 andanti-CD28 inducedproliferation of CD8þ T cells (Fig. 3G), as well as the expression ofGzmB and IFNg (Fig. 3H; Supplementary Fig. S3D), comparedwith that cocultured with WT MDSCs. Administration of NHNL,an Arg-1 inhibitor, significantly rescued the activity of CD8þ

T cells (Fig. 3I). Of note, GSK872-treated T cells also showed

moderately impaired expression of GzmB and IFNg (Supplemen-tary Fig. S3E), suggesting that RIPK3 deficiency in CTL may alsocontribute to colorectal carcinogenesis. Furthermore, the super-natant from RIPK3-KO MDSCs enhanced the proliferation ofCT26 cells (Fig. 3J). These results indicated that RIPK3 deficiencyenhanced proliferation and immunosuppressive function ofMDSCs.

Carcinogenesis is accelerated after RIPK3-deficient bonemarrow chimerism

To further test the role of RIPK3 on MDSC function in vivo, wegenerated chimeras by infusing WT or KO bone marrow cells into

Figure 2.

RIPK3 deficiency promotes MDSC infiltration and colorectal cancer tumorigenesis. A–K, WT and RIPK3-KO (KO) mice model were established with theprotocol as Model 1 in Fig. 1A. The body weight (A), the tumor number (B), colorectum length (C), spleen weight (D), survival (E), MDSC infiltration intumor by immunofluorescence staining (F), MDSCs percentage in tumor (G), colorectum (H), and spleen (I) were monitored; the percentages of immunecells (J), CD11bþLy6Gþ and CD11bþLy6Cþ cells (K) in the tumor tissue upon sacrifice were assessed. Data are expressed as mean � SEM (� , P < 0.05;�� , P < 0.01; �� , P < 0.001, WT vs. KO, by Student t test). ns, nonsignificant.






WT or KO recipient mice after irradiation. The presence of chime-rism after 7 weeks was confirmed using FCM. These animals weresubsequently induced colorectal cancer using model 2 (Supple-mentary Fig. S3F).We found that theWT recipients exhibitedmoresevere weight loss, higher mortality and tumor formation ratio,shorter colorectum length, splenomegaly, andmoreMDSCs in thecolorectum and spleen when engrafted with cells from RIPK3-KO

donors, compared with that from WT donors (Fig. 4A–G). Thepercentage of g-MDSCs were higher, while Mfs and DCs showedno significant difference in the colorectum and spleen of micereceived KO bone marrow, compared with that of mice receivedWT bone marrow. The leukocytes in colorectal tissues were higherinmice received KO bonemarrow, while their percentages did notshow significant change in spleen (Fig. 4H and I; Supplementary

Figure 3.

RIPK3 absence in MDSCs enhances the immunosuppressive activity in vitro. A–C, After bone marrow cells from WT and RIPK3 KO mice were treated withGM-CSF (20 ng/mL) for 48 hours, the proportion (A), proliferation (B), and death (C) of MDSCs were examined. After treatment with GM-CSF (20 ng/mL)for 48 hours, the differentiation of WT and KO MDSCs into Mfs and DCs (D), as well as the expression of Arg-1, NOS2, and ROS (E and F) were examined.G, CD8þ T cells were cocultured with WT/RIPK3-KO MDSCs (10:1) for 3 days; the proliferation of CD8þ T cells was determined by CFSE. H, After CD8þ T cellswere cocultured with WT/RIPK3-KO MDSCs (10:1) for 48 hours, the expression of GzmB and IFNg was assessed with FCM. I, After MDSCs were treated withvehicle (PBS) or NHNL (30 mmol/L), CD8þ T cells were cocultured with MDSCs (5:1) for 48 hours, and the expression of GzmB and IFNg was assessed.J, After CT26 colorectal cancer cells were cultured with conditioned medium (supernatant of WT/RIPK3–KO-MDSC: culture medium ¼ 1:1) for 48 hours, theproliferation was assessed. Data are expressed as mean � SEM (� , P < 0.05; �� , P < 0.01; �� , P < 0.001, by Student t test). ns, nonsignificant.

Yan et al.





Fig. S3G and S3H). Moreover, although KO recipients showed amoderate weight loss, increased mortality and modest highertumorigenicity compared with WT recipients after engrafted withWT bone marrow cells, the colorectum length, spleen weight,

MDSC infiltration, andMfs andDCs in the colorectumand spleendid not show significant change (Fig. 4A–I).

To further validate the essential role of MDSCs in the tumor-igenesis of colorectal cancer, we administrated anti-Gr-1 (to

Figure 4.

RIPK3 in myeloid-derived cells is essential for inhibiting colorectal tumorigenesis. A–I, The chimeric mice and colorectal cancer model were established as indicatedin Materials and Methods. The body weight (A), mortality (B), tumor formation ratio (C), colorectum length (D), spleen weight (E), MDSC percentage incolorectum (F), MDSC percentage in spleen (G), g-MDSC percentage in colorectum (H), and g-MDSC percentage in spleen (I) upon sacrifice were assessed.J–O, Colorectal cancer model of MDSC depletion and CXCL1 receptor CXCR2 antagonist treatment was established as indicated in Materials and Methods.The body weight (J), mortality (K), tumor number (L), colorectum length (M), MDSC percentage in colorectum (N), and in spleen (O) were assessed.Data are expressed as mean � SEM (� , P < 0.05; �� , P < 0.001, by Student t test). ns, nonsignificant.






deplete MDSCs) and CXCR2-a SB225002 (to inhibit MDSCchemotaxis) every 2 days from the third cycle of DSS untilthe mice were sacrificed. Both anti-Gr-1 and CXCR2-a reversedthe weight loss, mortality, tumor formation, colorectum length,and the MDSC infiltration in the colorectum and spleen com-pared with KO control (Fig. 4J–O). Together, these findingssupported the conclusion that RIPK3 deficiency in MDSCspromoted tumorigenesis.

NFkB/COX-2/PGE2 axis is upregulated in RIPK3-deficientMDSCs

We next explored the underlying mechanism by which RIPK3regulated MDSCs. Aforementioned, PGE2 is a proinflammatoryand immunosuppressive lipid mediator that potentiates MDSCsactivity and tumor growth (20). COX-2 is an essential enzyme forthe production of PGE2. We found that COX-2 expression wassignificantly upregulated in the tumor-infiltrating MDSCs than incolorectal MDSCs (Fig. 5A), but no significant difference of COX-2 expressionwas observed in CD45� cells of tumor and colorectaltissues (Supplementary Fig. S4A). ComparedwithWTmice, COX-2 expression were upregulated in the tumor-infiltrating MDSCs(Fig. 5B) and in CD45� cells of both tumor and colorectal tissuesof RIPK3-KO mice (Supplementary Fig. S4B and S4C). UsingUPLC/MS-MS, we found that RIPK3-KO MDSCs produced morePGE2 than that inWTMDSCs (Fig. 5C),whichwas consistentwiththe results of COX-2 expression.

Given that NFkB is a well-known transcription factor of COX-2and an essential controller for the immunosuppressive activity ofMDSCs (21, 22), we examined the expression of NFkB inMDSCs.Significant upregulated NFkB p65 and COX-2 were observed inRIPK3-KO MDSCs, compared with WT (Fig. 5D). We adminis-trated aspirin (ASA, COX inhibitor) and caffeic acid phenethylester (CAPE, NFkB inhibitor; ref. 23) and found that they bothinhibited PGE2 production from RIPK3-KO MDSCs but only

showed a trend in decreasing PGE2 production from GSK-872–treated CT26 cells (Supplementary Fig. S4D and S4E). Wealso assessed other key signaling molecules that drive the accu-mulation and function of MDSCs including stat3 and stat6.However, they showed no difference between WT and KOMDSCs (Supplementary Fig. S4F). These findings demonstratedthat RIPK3 reduction in MDSCs promoted the activation ofNFkB/COX-2/PGE2 axis.

Inhibitors targeting COX-2 and EP2 blunt theimmunosuppressive activity of MDSCs and carcinogenesis

PGE2 exerts its function by binding to its receptors includingEP1-4. We found that the RIPK3-KO MDSCs showed higher EP2and EP4, compared with WT MDSCs (Supplementary Fig. S5A).Therefore, the colorectal cancer mice model was treated with ASAor EP inhibitors (EP1 and EP2 inhibitor AH6809 and EP4 inhib-itor ONO-AE3-208; ref. 24). We found that ASA significantlyprotected the mice against tumorigenesis and reduced the accu-mulation and COX-2/Arg-1 expression of MDSCs (Fig. 6A–E;Supplementary Fig. S5B). AH6809 but not ONO-AE3-208 atten-uated AOM/DSS–induced tumorigenesis and MDSC accumula-tion (Fig. 6F–L; Supplementary Fig. S5C).

In vitro, PGE2 significantly enhanced Arg-1 expression inMDSCs and the differentiation toward M2 macrophages, whichwere reversed by ASA and AH6809 (Fig. 6M and N). Antagonistsof NFkB/COX-2/PGE2/EPs signaling pathway consistently res-cued the CD8þ T-cell activation dampened by RIPK3-KO MDSCcocultivation (Fig. 6O and P).

In addition, PGE2 promoted the proliferation of CT26 cells,which was blunted by AH6809 (Supplementary Fig. S5D, left).The cocultivation with supernatant from RIPK3-KO MDSC alsoenhanced the proliferation of CT26 cells, which was reversed byASA or AH6809 pretreatment in these MDSCs (SupplementaryFig. S5D, right). Together, these results indicated that antagonists

Figure 5.

NFkB/COX-2/PGE2 signaling is enhanced in RIPK3-deficient MDSCs. A, The expression of Gr-1 and COX-2 in tumor and colorectal tissue. B, COX-2 expressionin tumor-bearing WT/RIPK3-KO MDSCs was evaluated by FCM. ns, nonsignificant. C, MRM chromatograms, MS/MS spectrum, and production of PGE2 from bonemarrow–derivedWT/RIPK3-KOMDSCswere identifiedwithUPLC/MS-MS.D,The expression of COX-2 andp65 in bonemarrow–derivedWT/RIPK3-KOMDSCs. Dataare expressed as mean � SEM. � , P < 0.05; �� , P < 0.01; �� , P < 0.001, by one-way ANOVA with Sidak multiple comparisons test (A) or Student t test (B and C).

Yan et al.





Figure 6.

Blockade of COX-2 or EP2 attenuates tumorigenesis.A–E,WTandRIPK3-KOmice colorectal cancermodel 2 (Supplementary Fig. S3F) treatedwith orwithout 0.02%ASA containing drinking water as indicated in Materials and Methods; the body weight, tumor number, colorectum length, and spleen weight upon sacrificewere determined (A–D). The accumulation of MDSCs and COX-2 expression in MDSCs of tumor, colorectum, and spleen were analyzed (E). F–L,WT and RIPK3-KOmice colorectal cancer model 2 treated with or without AH-6809 (5 mg/kg) as indicated in Materials and Methods; the body weight (F), tumor number, colorectumlength, spleen weight (G–I), and the percentage of MDSCs in tumor, colorectum, and spleen upon sacrifice (J–L) were determined. M and N, After bone marrow–

derived MDSCs from WT or RIPK3-KO mice were treated with vehicle (PBS), PGE2 (10 mmol/L), ASA (10 mmol/L), or AH6809 (10 mmol/L) for 48 hours, Arg-1expression (M), and percentage of differentiation toward M2 macrophages (N) were assessed with FCM. O and P, After bone marrow–derived MDSCs from WT orRIPK3-KOmicewere treatedwith/without ASA (10 mmol/L), CAPE (2 mmol/L), or AH-6809 (10 mmol/L) for 48 hours, CD8þ T cells were coculturedwith MDSCs (16:1)for 24 hours. The expression of IFNg (O) and GzmB (P) in CD8þ T cells was assessed with FCM. Data are expressed as mean � SEM. � , P < 0.05; �� , P < 0.01;�� , P < 0.001; �� , P < 0.0001, by Student t test (A–L) or one-way ANOVA with Sidak multiple comparisons test (M–P).






targetingNFkB/COX-2/PGE2 signaling improved theprognosis ofcolorectal cancer.

RIPK3-PGE2 circuit in tumor microenvironment potentiatesmalignancy

We next questioned whether PGE2 in turn regulated RIPK3 andthe downstream NFkB/COX-2 signaling. Bone marrow–derivedMDSCswere treatedwith or without PGE2, andwe, indeed, foundthat PGE2 significantly suppressed RIPK3 while enhancing theexpression of p65 and COX-2 in MDSCs (Fig. 7A).

It is well-known that EP receptors are G-protein–coupledreceptors that activate cAMP-dependent protein kinase A (PKA)and promote the subsequent translocation of the transcriptionfactor cAMP-responsive element binding protein (CREB; 25).Of note, a recent report indicated that CREB reduced the pro-moter activity of RIPK3 (26). Here we found that the PGE2-downregulated RIPK3 could be rescued by both H89 (PKA inhib-itor) and AH6809 in MDSCs (Fig. 7B and C). ASA also couldupregulate RIPK3 expression in MDSCs (Fig. 7D), indicating thatPGE2 suppressed RIPK3 via PKA-CREB signaling. Consistently,PGE2 decreased RIPK3 via PKA–CREB pathway while bothPGE2 and GSK872 promoted the expression of p65 and COX-2in CT26 cells (Supplementary Fig. S6A–S6D). These results iden-tified a novel circuit involving RIPK3, NFkB, COX-2, and PGE2 inthe tumor microenvironment.

To explore the correlation of clinical chemotherapy and tar-geted therapy with RIPK3 expression, we treated MDSCs withsome common drugs used in cancer. The expression of RIPK3 inMDSCs were upregulated by bevacizumab or cetuximab (Fig. 7E)but was downregulated by CPT-11, oxaliplatin, 5-FU, and gemci-tabine (Fig. 7F).

We also evaluated the clinical relevance of the RIPK3-PGE2circuit in patients with colorectal cancer. We found that withthe development of colorectal cancer clinical stage, the expressionof RIPK3 in tumor-infiltrating MDSCs decreased (Fig. 7G;Supplementary Table S1). The relationship between RIPK3 andindicated gene transcripts were examined in 148 patientswith colorectal cancer from the database of National Centerfor Biotechnology Information Gene Expression Omnibus(GSE21510; ref. 13). Our results demonstrated that RIPK3 expres-sion negatively correlated with CD33 and S100A8, which areMDSC markers (Fig. 7H; ref. 27). Importantly, RIPK3 also neg-atively correlated with PTGS2 (COX-2; Fig. 7I). Because thedatabase GSE21510 lacked the survival results, we analyzed thecorrelation of RIPK3 and survival of patients with colorectalcancer with another one (GSE17536), which involved 177 pati-ents. We divided patients into "low" and "high" groups basedon the median values of RIPK3 and PTGS2. We found thatpatients with RIPK3highPTGS2low showed longest survival whilelow RIPK3 and high PTGS2 was associated with poor survival(Fig. 7J; Supplementary Table S2). Therefore, RIPK3 down-regulation and COX-2/PGE2 upregulation in the tumor micro-environment formed a circuit that promoted the accumulationof immunosuppressive MDSCs and colorectal carcinogenesis.

DiscussionThe infiltration of MDSCs in tumor microenvironment is

closely related to poor prognosis (3, 11). Here, we found thatthat the downregulation of RIPK3 promoted the infiltration andimmunosuppressive activity of MDSCs in tumor microenviron-

ment. The chimeric mice experiment also indicated the pivotalrole of RIPK3 on colorectal cancer–infiltrating MDSCs. Therefore,we identified that RIPK3 regulated tumor immunity by modu-lating MDSCs.

MDSCs play an immunosuppressive function mainly viamultiple signal pathways. First, the lipid metabolite PGE2derived from arachidonic acid via COX-2 catalysis in tumormicroenvironment stimulates the expression of Arg-1, IL6,VEGF, and other cancer-promoting molecules in MDSCs (11,28). Of note, MDSCs also express COX-2, which promotes theirown immunosuppressive activity (29). Second, NFkB activa-tion in MDSCs can promote the proliferation and inhibit thedifferentiation of MDSCs (22). Third, the secretion of VEGFfromMDSCs is also promoted by stat signaling pathway, whichenhances angiogenesis (11). This study showed that the expres-sion of COX-2 in RIPK3-KO MDSCs of colorectal cancer tissueswas significantly enhanced, compared with that in WT MDSCs.In vitro experiments also demonstrated that RIPK3-deficientMDSCs exhibited increased COX-2 expression and PGE2 secre-tion. However, we did not observe significant changes in ROS,NOS2, stat3, and stat6 between WT and RIPK3-KO MDSCs.These data suggested that the loss of RIPK3 in tumor-infiltratingMDSCs promoted the immunosuppressive function by activat-ing COX-2/PGE2.

A previous study reported a necrosis-independent pathwayof IBD by regulating DCs (9). They showed that in DSS-inducedIBD, RIPK3 deficiency impaired the NFkB activation and cas-pase 1–mediated processing of IL1b in DCs, thereby dampen-ing the tissue repair. Actually, we also observed that RIPK3 wasupregulated in the colorectal tissue during the IBD acuteinduction of DSS, while it significantly reduced at the stage ofcolorectal cancer. Furthermore, we showed that the RIPK3-KOMDSCs did not tend to differentiate into DCs. These cellspossessed higher immunosuppressive function and were proneto differentiate toward M2 macrophages. Another recent reportshowed that RIPK3 deficiency enhanced the lipopolysaccharide(LPS)-induced IL1b and TNFa expression in macrophages (30).Because LPS is an endotoxin that was found in the outermembrane of Gram-negative bacteria such as E.coli (31) andNFkB also transcripts IL1b and TNFa (32), this study wasconsistent with our results that RIPK3 deficiency in MDSCsenhanced the NFkB activation that in turn upregulated COX-2expression and PGE2 production during the carcinogenesis ofcolorectal cancer.

The mechanisms of RIPK3 upregulating NFkB are complex,which we did not investigate in detail in this study, but isdiscussed below. Aforementioned NFkB is a parallel proliferationpathway of RIPK3-mediated necroptosis and caspase-associatedapoptosis (7, 33). Knockout of RIPK3 is supposed to lead to acompensatory promotion of NFkB pathway. A recent study indi-cated that absent RIPK3 activated NFkB via a MLKL-independentpathway (34).Moreover, ubiquitination degradation is an impor-tant mechanism for the downregulation of multiple transcriptionfactors in cells. It has been shown that Cullin-RING E3 ligases(CRL) can mediate NFkB ubiquitination degradation and reduceits entry into nuclei (35). Hence RIPK3 may also phosphorylateCRLs by mimicking MLKL activation, thereby promoting ubiqui-tination degradation of NFkB.

Of note, the downregulation of RIPK3 and the subsequentNFkB/COX-2/PGE2 signaling was also identical in colorectalcancer cells in vivo and in vitro, which demonstrated the negative

Yan et al.





Figure 7.

PGE2 negatively regulates RIPK3 and upregulates NFkB and COX-2 in MDSCs and colorectal cancer cells. A, Bone marrow–derived MDSCs were treated with vehicle(PBS) or PGE2 (10 mmol/L) for 48 hours; the expression of RIPK3, p65, and COX-2 was assessed with Western blot analysis. B and C, Bone marrow–derived MDSCswere treated with vehicle (PBS), PGE2 (10 mmol/L), H89 (20 mmol/L), or AH6809 (10 mmol/L) for 48 hours; the expression of PKA, CREB, p-CREB, and RIPK3was determined with Western blot analysis (B), and the RIPK3 mRNA level was determined with qPCR (C). D, Bone marrow–derived MDSCs were treated withvehicle (PBS) or ASA (10 mmol/L) for 48 hours, and the protein expression of RIPK3 was assessed. E and F, RIPK3 expression of MDSCs from bonemarrow cells of WTmicewas treatedwith vehicle (PBS), bevacizumab (2.5mg/mL; Bev), cetuximab (Cet; 0.5mg/mL), nimotuzumab (Nim;0.5mg/mL;E), CPT-11 (300nmol/L), oxaliplatin(OXA; 30 nmol/L), 5-FU (250 nmol/L), or gemcitabine (GEM; 100 nmol/L; F). ns, nonsignificant. G, Confocal microscopy determination of RIPK3 expression inMDSCs in human polypus and cancer tissues at different stages. Data are expressed asmean� SEM (� , P < 0.05, by one-way ANOVAwith Sidakmultiple comparisonstest). H and I, Pearson's correlation coefficient was used to determine the correlation between RIPK3 and indicated genes, including CD33, S100A8 (H), and PTGS2(COX-2; I), in 148 patientswith colorectal cancer was examined (GSE21510). J,Patient survival data were obtained fromGEO database (GSE17536), and overall survivalprobability was then calculated using the Kaplan–Meier method, and the differences in survival curves were analyzed using the log-rank test. ns, nonsignificant.






correlationbetweenRIPK3expression and tumorigenesis. Admin-istrationof PGE2 inhibited RIPK3 expressionbut enhancedNFkB/COX-2 signaling in both MDSCs and CT26 colorectal cancer cellsby activating PKA–CREB signaling, indicating an unappreciatednegative signaling circuit that aggregate the malignancy. More-over, PGE2 was reported to directly and indirectly blunt theactivation of CD8þ T cells (20, 36). Our results showed thatinhibition of COX-2 or the PGE2 receptors significantly reversedthe downregulated RIPK3 and attenuated the immunosuppres-sive activity of MDSCs, thereby dampening the tumorigenesis ofcolorectal cancer.

In summation, this study identified a novel RIPK3-PGE2 circuitthat regulated the infiltration and function of MDSCs and thetumorigenesis of colorectal cancer. RIPK3 reduction led to NFkBactivation and upregulation of downstream COX-2, which cata-lyzed the synthesis of PGE2. PGE2, in turn, further inhibitedRIPK3, and promoted NFkB/COX-2 and Arg-1 expression inMDSCs. This signaling circuit also existed in colorectal cancercells and accelerated tumor growth. Importantly, using com-pounds or drugs targeting this signaling circuit clarified theimmunoregulatory role of RIPK3 and attenuated the carcinogen-esis of colorectal cancer. Daily consumption of low-dose ASA hasnow been applied to efficiently prevent and cure colorectal cancer(37, 38), delineating the significance of PGE2 blockade in thetumor microenvironment. PGE2 was reported to directly stimu-late colorectal cancer cells to secrete CXCL1, which bound toCXCR2 to recruit MDSCs into TME (39, 40). Of note, colorectalcancer cells also express CXCR2, which predicts poor prognosis.TheCXCR2 antagonist inhibits the proliferation andmetastasis ofcolorectal cancer cells (41, 42). Because the RIPK3-PGE2 circuitexists in both MDSCs and colorectal cancer cells, our resultsdemonstrated a mutual role of PGE2 blockade and CXCR2 antag-onist in inhibiting colorectal cancer tumorigenesis in RIPK3-KOmice. Moreover, our data demonstrated that cetuximab andbevacizumab upregulated RIPK3, while CPT-11, oxaliplatin, 5-FU, and gemcitabine suppressed RIPK3 in MDSCs. Of interest,cetuximab or bevacizumab are both the first-line targeted drugsfor patients with colorectal cancer and they were reported toinhibit COX-2 (43, 44). Therefore, targeting RIPK3 in MDSCsmight be considered for rational use of chemotherapeutic andtargeted drugs, which is essential for reeducating the immuno-

suppressive TME and enhance the antitumor immunity. Thesefindings provided the molecular basis and potential ideas for theimmunotherapy of colorectal cancer.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: Y. LiDevelopment of methodology: G. Yan, Q. Zhang, Y. Zhou, J. Lei, X. Zhang,G. Du, Y. LiAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): G. Yan, H. Zhao, Q. Zhang, Y. Zhou, L. Wu, J. Lei,X. Wang, J. Zhang, X. Zhang, G. Du, W. Xiao, Y. LiAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): G. Yan, H. Zhao, Q. Zhang, Y. Zhou, L. Wu, J. Lei,X. Wang, J. Zhang, X. Zhang, G. Du, B. Tang, H. Miao, Y. LiWriting, review, and/or revision of themanuscript:G. Yan,Q. Zhang,H.Miao,Y. LiAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): G. Yan, Q. Zhang, Y. Zhou, J. Lei, X. Zhang,G. Du, H. Miao, Y. LiStudy supervision: Y. LiOther (contributed to discussion, performed experiments and analyzeddata): L. Zheng

AcknowledgmentsThe work was supported by Youth 1000 Talent Plan (to Y. Li) and the

National Natural Science Foundation of China (81472435 and 81671573to Y. Li) and cstc2017jcyjBX0071 (to H. Miao) from the Foundation andFrontier Research Project of Chongqing. We sincerely thank XiaodongWang and Zhirong Shen (National Institute of Biological Sciences,Beijing, China) for providing RIPK3 knockout (KO) mice. We extremelyappreciate Chunyan Hu for her support on FCM. We are also very gratefulto Rong Xin and Lu Jiang for their valuable assistance in IHC and IFprocedure, and to Jin Peng and Qian Chen for confocal microscopyexperiments.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must, therefore, be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received December 22, 2017; revised June 9, 2018; accepted July 10, 2018;published first July 16, 2018.

References1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin

2017;67:7–30.2. Brenner H, Kloor M, Pox CP. Colorectal cancer. Lancet 2014;383:1490–

502.3. Medina-Echeverz J, Aranda F, Berraondo P. Myeloid-derived cells are

key targets of tumor immunotherapy. Oncoimmunology 2014;3:e28398.

4. Katoh H, Wang D, Daikoku T, Sun H, Dey SK, Dubois RN. CXCR2-expressing myeloid-derived suppressor cells are essential to promotecolitis-associated tumorigenesis. Cancer Cell 2013;24:631–44.

5. Tauriello DVF, Batlle E. Targeting the microenvironment in advancedcolorectal cancer. Trends Cancer 2016;2:495–504.

6. Gabrilovich DI, Nagaraj S. Myeloid-derived suppressor cells as regulatorsof the immune system. Nat Rev Immunol 2009;9:162–74.

7. Han J, Zhong CQ, Zhang DW. Programmed necrosis: backup to andcompetitor with apoptosis in the immune system. Nat Immunol2011;12:1143–9.

8. Weinlich R, Oberst A, Beere HM, Green DR. Necroptosis in development,inflammation and disease. Nat Rev Mol Cell Biol 2017;18:127–36.

9. Moriwaki K, Balaji S, McQuade T, Malhotra N, Kang J, ChanFK. The necroptosis adaptor RIPK3 promotes injury-inducedcytokine expression and tissue repair. Immunity 2014;41:567–78.

10. Bozec D, Iuga AC, Roda G, Dahan S, Yeretssian G. Critical function of thenecroptosis adaptor RIPK3 in protecting from intestinal tumorigenesis.Oncotarget 2016;7:46384–400.

11. Ostrand-Rosenberg S, Sinha P. Myeloid-derived suppressor cells: linkinginflammation and cancer. J Immunol 2009;182:4499–506.

12. Tsukamoto S, Ishikawa T, Iida S, Ishiguro M, Mogushi K, Mizushima H,et al. Clinical significance of osteoprotegerin expression in human colo-rectal cancer. Clin Cancer Res 2011;17:2444–50.

13. Smith JJ, Deane NG, Wu F, Merchant NB, Zhang B, Jiang A, et al.Experimentally derived metastasis gene expression profile predictsrecurrence and death in patients with colon cancer. Gastroenterology2010;138:958–68.

14. LechnerMG, Liebertz DJ, Epstein AL. Characterization of cytokine-inducedmyeloid-derived suppressor cells from normal human peripheral bloodmononuclear cells. J Immunol 2010;185:2273–84.

Yan et al.





15. Dolcetti L, Peranzoni E, Ugel S, Marigo I, Fernandez Gomez A, Mesa C,et al. Hierarchy of immunosuppressive strength among myeloid-derived suppressor cell subsets is determined by GM-CSF. Eur J Immu-nol 2010;40:22–35.

16. Li Y, Dalli J, Chiang N, Baron RM, Quintana C, Serhan CN. Plasticity ofleukocytic exudates in resolving acute inflammation is regulated byMicro-RNA and proresolving mediators. Immunity 2013;39:885–98.

17. Neufert C, Becker C, Neurath MF. An inducible mouse model of coloncarcinogenesis for the analysis of sporadic and inflammation-driven tumorprogression. Nat Protoc 2007;2:1998–2004.

18. OuzounovaM, Lee E, Piranlioglu R, El Andaloussi A, Kolhe R, Demirci MF,et al. Monocytic and granulocytic myeloid derived suppressor cells differ-entially regulate spatiotemporal tumour plasticity during metastatic cas-cade. Nat Commun 2017;8:14979.

19. Qiu X, Klausen C, Cheng JC, Leung PC. CD40 ligand inducesRIP1-dependent, necroptosis-like cell death in low-grade serousbut not serous borderline ovarian tumor cells. Cell Death Dis 2015;6:e1864.

20. Zelenay S, van der Veen AG, Bottcher JP, Snelgrove KJ, Rogers N, Acton SE,et al. Cyclooxygenase-dependent tumor growth through evasion of immu-nity. Cell 2015;162:1257–70.

21. Nguyen LK, Cavadas MA, Kholodenko BN, Frank TD, Cheong A. Speciesdifferential regulation of COX2 can be described by an NFkappaB-dependent logic AND gate. Cell Mol Life Sci 2015;72:2431–43.

22. Yu J, Wang Y, Yan F, Zhang P, Li H, Zhao H, et al. Noncanonical NF-kappaB activation mediates STAT3-stimulated IDO upregulation inmyeloid-derived suppressor cells in breast cancer. J Immunol 2014;193:2574–86.

23. Natarajan K, Singh S, Burke TR Jr., Grunberger D, Aggarwal BB. Caffeicacid phenethyl ester is a potent and specific inhibitor of activation ofnuclear transcription factor NF-kappa B. Proc Natl Acad Sci USA 1996;93:9090–5.

24. Rosch S, Ramer R, Brune K, Hinz B. Prostaglandin E2 induces cyclooxy-genase-2 expression in human non-pigmented ciliary epithelial cellsthrough activation of p38 and p42/44 mitogen-activated protein kinases.Biochem Biophys Res Commun 2005;338:1171–8.

25. Alvarez Y, Municio C, Alonso S, Sanchez Crespo M, Fernandez N. Theinduction of IL-10 by zymosan in dendritic cells depends on CREBactivation by the coactivators CREB-binding protein and TORC2 andautocrine PGE2. J Immunol 2009;183:1471–9.

26. GuidaN, Laudati G, Serani A,Mascolo L,Molinaro P,Montuori P, et al. Theneurotoxicant PCB-95 by increasing the neuronal transcriptional repressorREST down-regulates caspase-8 and increases Ripk1, Ripk3 and MLKLexpression determining necroptotic neuronal death. Biochem Pharmacol2017;142:229–41.

27. HeinemannAS, Pirr S, Fehlhaber B,Mellinger L, Burgmann J, BusseM, et al.In neonates S100A8/S100A9 alarmins prevent the expansion of a specificinflammatory monocyte population promoting septic shock. FASEB J2017;31:1153–64.

28. Condamine T, GabrilovichDI.Molecularmechanisms regulatingmyeloid-derived suppressor cell differentiation and function. Trends Immunol2011;32:19–25.

29. Rodriguez PC, Hernandez CP, QuicenoD, Dubinett SM, Zabaleta J, OchoaJB, et al. Arginase I inmyeloid suppressor cells is induced by COX-2 in lungcarcinoma. J Exp Med 2005;202:931–9.

30. Moriwaki K, Bertin J, Gough PJ, Chan FK. A RIPK3-caspase 8 complexmediates atypical pro-IL-1beta processing. J Immunol 2015;194:1938–44.

31. Wang JH, Manning BJ, Wu QD, Blankson S, Bouchier-Hayes D, RedmondHP. Endotoxin/lipopolysaccharide activates NF-kappa B and enhancestumor cell adhesion and invasion through a beta 1 integrin-dependentmechanism. J Immunol 2003;170:795–804.

32. Banerjee D, Liou HC, Sen R. c-Rel-dependent priming of naive T cells byinflammatory cytokines. Immunity 2005;23:445–58.

33. Vlantis K, Wullaert A, Polykratis A, Kondylis V, Dannappel M, SchwarzerR, et al. NEMO prevents RIP kinase 1-mediated epithelial cell deathand chronic intestinal inflammation by NF-kappaB-dependent and-independent functions. Immunity 2016;44:553–67.

34. Najjar M, Saleh D, Zelic M, Nogusa S, Shah S, Tai A, et al. RIPK1 andRIPK3 kinases promote cell-death-independent inflammation by toll-likereceptor 4. Immunity 2016;45:46–59.

35. Schweitzer K, Bozko PM,DubielW,NaumannM.CSN controlsNF-kappaBby deubiquitinylation of IkappaBalpha. EMBO J 2007;26:1532–41.

36. Weinlich R, Bortoluci KR, Chehab CF, Serezani CH, Ulbrich AG, Peters-GoldenM, et al. TLR4/MYD88-dependent, LPS-induced synthesis of PGE2by macrophages or dendritic cells prevents anti-CD3-mediated CD95Lupregulation in T cells. Cell Death Differ 2008;15:1901–9.

37. Liao X, Lochhead P, Nishihara R, Morikawa T, Kuchiba A, Yamauchi M,et al. Aspirin use, tumor PIK3CA mutation, and colorectal-cancer survival.N Engl J Med 2012;367:1596–606.

38. Algra AM, Rothwell PM. Effects of regular aspirin on long-term cancerincidence and metastasis: a systematic comparison of evidence fromobservational studies versus randomised trials. Lancet Oncol 2012;13:518–27.

39. Kumar V, Donthireddy L, Marvel D, Condamine T, Wang F, Lavilla-AlonsoS, et al. Cancer-associated fibroblasts neutralize the anti-tumor effect ofCSF1 receptor blockade by inducing PMN-MDSC infiltration of tumors.Cancer Cell 2017;32:654–68 e5.

40. WangD,WangH, Brown J,Daikoku T,NingW, ShiQ, et al. CXCL1 inducedby prostaglandin E2 promotes angiogenesis in colorectal cancer. J ExpMed2006;203:941–51.

41. Jamieson T, Clarke M, Steele CW, Samuel MS, Neumann J, Jung A, et al.Inhibition of CXCR2 profoundly suppresses inflammation-driven andspontaneous tumorigenesis. J Clin Invest 2012;122:3127–44.

42. WuW, SunC, XuD, Zhang X, ShenW, Lv Y, et al. Expression of CXCR2 andits clinical significance in human colorectal cancer. Int J Clin Exp Med2015;8:5883–9.

43. Kaliberova LN, Kusmartsev SA, Krendelchtchikova V, Stockard CR, GrizzleWE, BuchsbaumDJ, et al. Experimental cancer therapy using restoration ofNADþ -linked 15-hydroxyprostaglandin dehydrogenase expression. MolCancer Ther 2009;8:3130–9.

44. Half E, Sun Y, Sinicrope FA. Anti-EGFR and ErbB-2 antibodies attenuatecyclooxygenase-2 expression and cooperatively inhibit survival of humancolon cancer cells. Cancer Lett 2007;251:237–46.






2018;78:5586-5599. Published OnlineFirst July 16, 2018.Cancer Res Guifang Yan, Huakan Zhao, Qi Zhang, et al. Potentiated Colorectal Carcinogenesis

− Circuit Mediates Myeloid-Derived Suppressor Cell2A RIPK3-PGE

Updated version

10.1158/0008-5472.CAN-17-3962doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2018/07/13/0008-5472.CAN-17-3962.DC1

Access the most recent supplemental material at:

Overview

Visual

http://cancerres.aacrjournals.org/content/78/19/5586/F1.large.jpgA diagrammatic summary of the major findings and biological implications:

Cited articles

http://cancerres.aacrjournals.org/content/78/19/5586.full#ref-list-1

This article cites 44 articles, 12 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/78/19/5586.full#related-urls

This article has been cited by 2 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/78/19/5586To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-17-3962

http://cancerres.aacrjournals.org/content/suppl/2018/07/13/0008-5472.CAN-17-3962.DC1

http://cancerres.aacrjournals.org/content/78/19/5586/F1.large.jpg

http://cancerres.aacrjournals.org/content/78/19/5586.full#ref-list-1

http://cancerres.aacrjournals.org/content/78/19/5586.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/78/19/5586


Documents

A RIPK3-PGE Circuit Mediates Myeloid-Derived Suppressor ... · Chem Express), or ONO-AE3-208 (5 mg/kg, catalog no. 402473-54-5, Med Chem Express) was injected intraperitone- ally