Research Article

Protumor Steering of Cancer Inflammation by p50NF-kB Enhances Colorectal Cancer ProgressionChiara Porta1, Alessandro Ippolito1, Francesca Maria Consonni1,Lorenzo Carraro1, Giuseppe Celesti2, Carmen Correale2, Fabio Grizzi2,Fabio Pasqualini2, Silvia Tartari2, Maurizio Rinaldi1, Paolo Bianchi2, Fiorella Balzac3,Stefania Vetrano2, Emilia Turco3, Emilio Hirsch3, Luigi Laghi2, and Antonio Sica1,2

Abstract

Although tumor-associated macrophages (TAM) display a M2-skewed tumor-promoting phenotype in most cancers, in colorec-tal cancer, both TAM polarization and its impact remain contro-versial. We investigated the role of the M2-polarizing p50 NF-kBsubunit in orchestrating TAM phenotype, tumor microenviron-ment composition, and colorectal cancer progression. We firstdemonstrated, by parallel studies in colitis-associated cancer(CAC) and in genetically driven ApcMin mouse models, that thep50-dependent inhibition of M1-polarized gut inflammationsupported colorectal cancer development. In accordance withthese studies, p50–/– mice displayed exacerbated CAC with fewerand smaller tumors, along with enhanced levels of M1/Th1cytokines/chemokines, including IL12 and CXCL10, whose

administration restrained CAC development in vivo. The inflam-matory profile supporting tumor resistance in colons from p50–/–

tumor bearers correlated inversely with TAM load and positivelywith both recruitment of NK, NKT, CD8þ T cells and number ofapoptotic tumor cells. In agreement, myeloid-specific ablation ofp50 promoted tumor resistance in mice, whereas in colorectalcancer patients, a high number of p50þ TAMs at the invasivemargin was associated with decreased IL12A and TBX21 expres-sion and worse postsurgical outcome. Our findings point to p50involvement in colorectal cancer development, through itsengagement in the protumor activation of macrophages, andidentify a candidate for prognostic and target therapeutic inter-vention. Cancer Immunol Res; 6(5); 578–93. �2018 AACR.

IntroductionColorectal cancer development in inflammatory bowel disease

patients is almost 3-fold higher than the general population,making inflammation the third most common colorectal cancerrisk factor after the hereditary colorectal cancer syndromes, famil-ial adenomatous polyposis coli, and hereditary nonpolyposiscolon cancer (1). Even in sporadic and familial colorectal cancertumors, a "smoldering inflammation" is present despite the lackof an obvious preexisting inflammation (2, 3). As a predominantimmune population infiltrating the tumor, macrophages exert apivotal role in the orchestration of cancer inflammation (4), bytriggering either M1 or M2 activation states in response to micro-environmental signals (5). Selected pathways and transcriptionfactors have been described to control the polarization state of

macrophages (6). Among these, the transcription factorNF-kB is akey regulator of gut inflammation and functions, and aberrantactivation of NF-kB has been frequently observed in humancolorectal cancer, correlating with worse outcome. An IKKb-dependent activation of NF-kB supports colorectal cancer cells'proliferation, survival, and epithelial–mesenchymal transition(7, 8), as well as the expression of the tumor-promoting cytokineIL6 bymyeloid cells. In turn, NF-kB activates additional pathwaysof cancer cell proliferation and survival (e.g., STAT3; refs. 9, 10).Interestingly, IKKa ablation promoted IKKb-driven NF-kB acti-vation in intestinal epithelial cells and the subsequent recruitmentof M1-polarized antitumor myeloid cells (11). However, despitethis promising premise, targeting NF-kB remains a major chal-lenge in colorectal cancer therapy, as its systemic inhibition leadsto severe adverse effects (12).

Tumor-infiltrating leukocytesmay exert beneficial or detrimen-tal activities, depending on their functional state (13). Experi-mental and clinical studies indicate that the immunoscore,defined by type, density, and location of T cells, is an importantprognostic factor in colorectal cancer, whichhas been suggested toovercome AJCC/UICC TNM classification, although this pointremains debated (14, 15). Tumor-associated macrophages (TAM;refs. 13, 16) largely express an M2-skewed phenotype, associatedwith suppression of adaptive immune functions and promotionof angiogenesis and invasion (13). Accordingly, a high number ofTAMs has been associated with poor prognosis in several humancancers (17–19). However, TAMs can also express antitumoractivities (20), and their role in colorectal cancer remains con-troversial (21–23). In colorectal cancer intratumoral macro-phages appear to play tumor-promoting roles (24), whereas those

1Department of Pharmaceutical Sciences, Universit�a degli Studi del PiemonteOrientale "Amedeo Avogadro," Novara, Italy. 2Humanitas Clinical and ResearchCenter, Rozzano, Italy. 3Department of Molecular Biotechnology and HealthSciences, Molecular Biotechnology Center, Torino, Italy.

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

Corresponding Authors: Antonio Sica, Universit�a degli Studi del PiemonteOrientale "Amedeo Avogadro," Largo Donegani 2, Novara 28100, Italy. Phone:39-0321-375881; Fax: 39-0321-375621; E-mail: antonio.sica@uniupo.it; andChiara Porta, Department of Pharmaceutical Sciences, Universit�a degli Studidel Piemonte Orientale "Amedeo Avogadro," via Bovio 6, 28100 Novara, Italy.Phone: 39-0321-375883; Fax: 39-0321-375821; E-mail: chiara.porta@uniupo.it

doi: 10.1158/2326-6066.CIR-17-0036

�2018 American Association for Cancer Research.

CancerImmunologyResearch

Cancer Immunol Res; 6(5) May 2018578

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

located at the invasive front exert beneficial activities (25).Although the tumor microenvironment sculpts the phenotypeof TAMs (13), the pathways driving their responses in theinner versus peripheral tumor remain unknown. M1 and M2macrophage populations are dynamically recruited and function-ally modulated during colorectal cancer development in mice(26), and simultaneous accumulation of M1 (e.g., NOS2þ) andM2 (e.g., CD163þ) and mixed macrophage populations areobserved in human colorectal cancer tumors (27). The molecularbasis and the clinical relevance of this dynamic reprogrammingand heterogeneity of TAMs are still unclear.We have reported thatprogressive nuclear accumulation of p50 NF-kB in macrophagespromotes tolerance and M2-polarized activation (28, 29). Basedon this evidence, we investigated the impact of p50NF-kB–driveninflammation in colorectal cancer development and progression.

Materials and MethodsPatients

The study was approved by the Institute Ethical Committee,and written informed consent was obtained from patients. Thestudy was conducted on 47 specimens of primary tumor from awell-studied, consecutive series of patients resected for stage II/IIIcolorectal cancer at the Humanitas Research Hospital between1997 and 2006, for which both frozen tissues and paraffin-embedded specimens were available (30). Patients who hadreceived neoadjuvant radiochemotherapy were excluded fromthe study. Patient demographics and clinical and pathologicalfeatures (Supplementary Table S1) were available at the hospitalintranet system, which was also used for retrieving clinical dataconcerning patients' postsurgical outcome.

MiceC57BL/6J wild-type (WT) mice (Charles River Laboratories),

p50 NF-kB–deficient mice (28), originally donated by Drs.Michael Karin and Giuseppina Bonizzi (University of California,San Diego School of Medicine, San Diego, CA), ApcMin mice (TheJackson Laboratory) were on the sameC57BL/6J background. TheApcMin mice were crossed with p50–/– mice in order to generateApcMinp50–/–mice.Mice carrying theNFKB1floxed allele (p50Fl/Fl

mice) were generated as described below.The targeting construct to generate a NFKB1flox/flox (p50Fl/Fl)

mice was designed as follows: the eighth exon of NFKB1 gene wasflankedwith loxP sites, and the neomycin resistance gene, flankedwith Flippase Recognition Target (FRT) sites to permit its excision,was inserted in the seventh intron. This construct was introducedby electroporation into mouse ES cells. Homologous recombi-nation was confirmed by Southern blot, and ES cells carryingNFKB1fl-neo allele were injected into C57 blastocysts to generategermline chimeras. The neomycin resistance cassette in the target-ing construct was removed by crossing heterozygous NFKB1Fl/þ-neo mice with C57BL6 mice carrying the FLP recombinaseunder the control of the actin promoter (kindly provided byDr. Rolf Sprengel Heidelberg University, Heidelberg, Germany)to produce mice carrying the NFKB1 floxed allele (p50Fl/þ mice).p50Fl/Fl mice were crossed with B6.129P2-Lyz2tm1(cre)Ifo/Jand B6.Cg-Tg(Vil1-cre) mice (Jackson Laboratories) to generatep50Fl/Fl;Lyz2Cre and p50Fl/Fl;Villin-Cre mice, respectively.Genotyping of mice was performed by PCR using the followingprimers: p50FloxF1: 50- CTGGGCTCCTAGCGGGGGAG-30 andp50FloxR1, 50- GGCTGACATGTGGGTCCTCAGC-30, which gaverise to anampliconof 282bp for thefloxed allele andof 182bp for

the WT allele. To detect Lyz2Cre and Villin-Cre transgenes, weadopted the PCR protocols provided by Jackson Laboratories.

To confirm p50 NF-kB ablation in myeloid cells (p50Fl/Fl;Lyz2Cre mice) and enterocytes (p50Fl/Fl;Villin-Cre mice), thiogly-colate-elicited peritoneal macrophages from p50Fl/Fl;Lyz2Cremice and littermate controls (p50Fl/Fl mice) were analyzed forp50 expression by Western blot as previously described (ref. 28;Supplementary Fig. S1A). Sections of formalin-fixed, paraffin-embedded colons from p50Fl/Fl;Villin-Cre mice and littermatecontrols (p50Fl/Fl mice) were evaluated for p50 expression byimmunohistochemistry with rabbit monoclonal anti-p50 (cloneE381; Abcam; Supplementary Fig. S1B). Detailed information isprovided in the "Immunohistochemistry or immunofluores-cence" section.

The study was designed in compliance with principles set outin the following laws, regulations, and policies governing thecare and use of laboratory animals: Italian Governing Law(Legislative Decree 116 of January 27, 1992); EU directivesand guidelines (EEC Council Directive 86/609, OJ L 358,December 12, 1986); Legislative Decree September 19, 1994,n. 626 (89/391/CEE, 89/654/CEE, 89/655/CEE, 89/656/CEE,90/269/CEE, 90/270/CEE, 90/394/CEE, 90/679/CEE); the NIHGuide for the Care and Use of Laboratory Animals (1996edition), and it was approved by the scientific board of Huma-nitas Clinical and Research Center. In all experiments, sex- andage-matched mice bred in the same specific-pathogen-free(SPF) animal facility were used. Mice were monitored dailyand euthanized when displaying excessive discomfort. To assessoverall survival, ApcMin and ApcMinp50–/– mice were continu-ously monitored for a period of up to 40 weeks.

Azoxymethane (AOM)/DSS-induced colorectal cancerSix- to 8-week-old mice (WT and p50–/–, p50Fl/Fl;Lyz2Cre and

p50Fl/Fl littermates, and p50Fl/Fl;Villin-Cre and p50Fl/Fl litter-mates) and 16-week-old mice (WT and p50–/– mice that weresubjected to bonemarrow (BM) transplantation at 8weeks of age)were injected intraperitoneally (i.p.) with a single dose (10mg/kgof body weight) of the mutagenic agent azoxymethane (Sigma).Five days later, mice received water with 2% dextran sodiumsulfate (DSS; MP Biomedicals, molecular mass: 40 kDa) over 5days, followed by 16 days of regular water. This cycle of DSStreatment was repeated two more times, and mice were sacrificed3 weeks after the last cycle (31). The clinical course of colitis wasevaluated by monitoring mice body weight during the course ofthe experiment and by measuring colon length at necroscopy. Atthe time of harvest, mice were euthanized, colons were resected,flushed with saline solution, opened longitudinally, and macro-scopically evaluated for tumor numbers. This protocol was usedin all the experiments unless otherwise specified.

When specified, starting fromday 15 (e.g., recovery phase of thefirst DSS cycle), WTmice underwent i.p. administration of 100 ngIL12 (BioLegend) diluted in 100 mL of 1% BSA phosphate-buffered saline (PBS) solution or intrarectal injection of 500 ngCXCL10 (BioLegend), once a week. For CXCL10 treatment, micewere anesthetized by i.p. injection of Ketamina (Imalgene; 100mg/kg) and Xilazina (Rompun; 10 mg/kg). Next, 500 ng ofCXCL10 (BioLegend) diluted in 50 mL of 1% BSA PBS solutionwas rectally administered via a flexible plastic feeding tubeinserted 3 cm proximally to the anus. After rectal administration,micewere kept in an inverted position for 30 seconds. As controls,mice received vehicle only.

p50 NF-kB Shapes Colorectal Cancer–Promoting Inflammation

www.aacrjournals.org Cancer Immunol Res; 6(5) May 2018 579

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

When specified, starting from the day before the first DSStreatment, WT and p50–/– mice received an i.p. injection of 0.3mg anti-mouse CD4 (clone GK1.5; Bio X Cell) and 0.3 mg anti-mouse CD8 (clone 2.43; Bio X Cell), once a week for the entireexperimental period. FACS analysis of peripheral blood samplesconfirmed CD4þ and CD8þ cells depletion for 7 days.

Histologic analysisAt the end of AOM/DSS experiments, mice were euthanized,

colons were resected, flushed with PBS, opened longitudinally,and rolled up. At the indicated age, ApcMin andApcMinp50–/–micewere euthanized. Both small gut and colons were harvested,flushed with PBS, and prepared according to swiss and rolltechnique. Gut samples were fixed in 10% neutral bufferedformalin for 24 hours and paraffin-embedded. Four micrometersof serial tissue sections was stained with hematoxylin and eosinstaining (H&E) and analyzed in a blinded fashion by a pathol-ogist. Histologic evaluation of grade of colitis was performedaccording to the score of Cooper and colleagues (32), and Suzukiand colleagues (33) only slightly modified to adapt it to thefindings of present study. 0 ¼ Normal colonic mucosa, 1 ¼shortening and loss of the basal one third of the crypts with mildinflammation in themucosa, 2¼ loss of thebasal two thirds of thecrypts with moderate inflammation in the mucosa, occasionallyextending into submucosa, 3¼ loss of the entire crypts with severeinflammation in the mucosa and submucosa but with the retain-ment of the surface epithelium, 4 ¼ presence of mucosal ulcerwith severe inflammation in themucosa, submucosa, submucosa,tunica muscularis, and/or subserosa. The scoring of colitis wasmade at 40�magnification on the entire colon swiss roll with orwithout proliferative lesions and expressed asmean score/mouse.Additionally, the number of ulcers, and total number and size ofthe neoplastic lesions was recorded.

BM transferBM cells were flushed from femurs and tibia of CD45.1 WT or

p50–/– mice. Recipient CD45.2þp50–/– or WT mice were lethallyirradiated (900 cGy). Five hours later, 5 � 106 red blood cell–depleted BM cells were injected intravenously in recipient mice.BM reconstitution was evaluated 6 weeks after transplantation byflow cytometry on peripheral blood.

Analysis of p50 nuclear accumulation in murine and humanTAMs

Mouse colonic sections (untreated and AOM/DSS-treated WTmice) of 10 mmthickness and colorectal cancer specimens of 3 mmthickness were deparaffinized and rehydrated. Human slides wereexposed to UV radiation overnight. Antigen unmasking wascarried out by incubation in a decloaker chamber at 125�C for3 minutes and 90�C for 10 minutes in Diva Decloaker retrievalsolution (Biocare). Unspecific binding sites were blockedwith 2%BSA, 0.1% Triton X-100 in 0.05% Tween 20 phosphate buffersolution (PBST) for 1 hour at room temperature. Next, murinesamples were stained at room temperature in a humidifiedchamber with mouse monoclonal anti-p65 NF-kB (clone L8F6;Cell Signaling Technology; 1/100dilution in PBST) for 1hour andthen incubated with rabbit monoclonal anti-p50 NF-kB (cloneE381; Abcam; 1/250 dilution in PBST) for 4 hours. Samples werethen incubated with rat monoclonal anti-mouse F4/80 (clone CI:A3-1; AbD Serotec; 1/100 diluted in PBST) for 1 hour. After eachstaining, slides were washed with PBST. Next, goat anti-mouse

Alexa Fluor 555–conjugated (Cat. # A28180; Thermo FisherScientific), goat anti-rabbit Alexa Fluor 647–conjugated (Cat. #A27040; Thermo Fisher Scientific), and goat anti-rat AlexaFluor 488–conjugated (Cat. # A-11006; Thermo Fisher Scien-tific; 1/1,000 dilution in PBST) were incubated for 1 hour atroom temperature in a humidified chamber. Human sampleswere stained at room temperature in a humidified chamberwith mouse monoclonal anti-human CD68 (clone M0814;Dako; 1/200 dilution in PBST) for 1 hour, followed by rabbitmonoclonal anti-p50 NF-kB (clone E381; Abcam; 1/250 dilu-tion in PBST) for 4 hours. After each staining, slides werewashed with PBST, followed by addition of goat anti-mouseAlexa Fluor 488–conjugated and goat anti-rabbit Alexa Fluor647–conjugated (both diluted 1/1000 in PBST) and incubatedfor 1 hour at room temperature in a humidified chamber.Nuclei were stained with DAPI (Life Technologies) and thenmounted with ProLong Antifade Gold Reagent (Life Technol-ogies). Slides were analyzed with Olympus Fluorview FV1000confocal microscope with 60� (N.A. 0.4), and single-cell countwas performed by lab personnel under the supervision ofimaging core facility personnel (n � 8 field for every sample;n � 3 for every condition).

Real-time PCR analysisAt the end of AOM/DSS treatment, colons fromWT and p50–/–

mice were washed with ice-cold PBS, and then macroscopictumors and the adjacent healthy tissue were harvested and main-tained in RNA stabilization solution (RNAlater, Ambion). Sim-ilarly, colons were collected from untreated WT and p50–/–

mice and from AOM/DSS-treated WT mice 9 days after thebeginning of first cycle of DSS treatment. Similarly, macroscopiccolonic tumors and the adjacent healthy tissue were harvestedfrom 20- to 24-week-old ApcMin and ApcMinp50–/– mice. TotalRNA was extracted from mouse tissues using the TissueLyser II(Qiagen) and RNeasy Lipid Tissue Mini Kit (Qiagen) and fromhuman fresh stage II and III colorectal cancer surgical specimensby using TRIzol Reagent (Thermo Fisher Scientific), accordingto the manufacturer's instructions. RNA quality was assessed as260/280 and 260/230 OD ratio >1.5. 1 mg of RNA was reversetranscribed by the cDNA Archive kit (Applied Biosystems), andthen 20 ng of template was amplified using GOTAQ qPCRMasterMix (Promega) and detected by the CFX96 Real-Time System(Bio-Rad). All sampleswere run in triplicate. Expression datawerenormalized to Actin or 18S mRNA expression and were analyzedby the 2�DDCt method. Results are expressed as fold upregulationwith respect to the control cell population specified in the figurelegend. The sequences of gene-specific primers were obtainedfrom previous published papers (10, 29, 34) or were designedby using https://www.ncbi.nlm.nih.gov/tools/primer-blast/. Theprimers are synthetized by Invitrogen and are reported in Sup-plementary Table S2.

Immunohistochemistry or immunofluorescenceFormalin-fixed, paraffin-embedded colons from untreated and

AOM/DSS-treated mice (WT and p50–/–), as well as from p50Fl/Fl

and p50Fl/Fl;Villin-Cre mice, were cut in sections of 10 mmthickness. Colonic slides were deparaffinized and rehydrated.Antigen unmasking was carried out by incubation in a decloakerchamber at 125�C for 3minutes and 90�C for 10minutes in DivaDecloaker retrieval solution (Biocare Medical). Endogenous per-oxidases were blocked with peroxidized 1 (Biocare Medical) for

Porta et al.

Cancer Immunol Res; 6(5) May 2018 Cancer Immunology Research580

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

Figure 1.

Association between distinct inflammatory gene clusters and clinical outcome. A, Expression of selected inflammatory genes was analyzed in total RNA extractedfrom colons of mice with colitis (after the first cycle of DSS administration), tumors, and adjacent healthy tissue at the end of the AOM/DSS experiment.Total RNA from colon of untreated mice was used as control. Normalized qPCR results shown as fold increase over control. Cluster 1: genes similarly upregulated incolitis and tumor stages. Cluster 2: genes reaching the maximum peak of expression at the tumor stage. Box: the 25th–75th percentiles; line: the median;whiskers: range. � , P�0.05 by two-tailed Kruskal–Wallis with Dunn correction formultiple comparisons. Control (�): white boxes,N¼ 4 differentmice; colitis: greenboxes, N ¼ 9 different mice; tumor: blue boxes, N ¼ 13 different mice; healthy (AOM/DSS): gray boxes, N ¼ 5 different mice. B, Immunofluorescence of nuclearp50 (white) and p65 (red) in colonic and TAMs (F4/80þ cells; green). N ¼ 326 single F4/80þ cells analyzed from 7 total images of colons from 2 untreatedWT mice; N ¼ 1314 single F4/80þ cells analyzed from 52 total images of 16 tumors from 5 AOM/DSS-treated WT mice. Error bars, SEM. � , P < 0.05 by two-tailedunpaired t test. Representative images shown. Scale bars: 10 mm.

p50 NF-kB Shapes Colorectal Cancer–Promoting Inflammation

www.aacrjournals.org Cancer Immunol Res; 6(5) May 2018 581

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

Porta et al.

Cancer Immunol Res; 6(5) May 2018 Cancer Immunology Research582

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

20 minutes, then unspecific binding sites were blocked with 2%BSA PBS at room temperature for 1 hour.

Immunohistochemistrywas performed at room temperature: ratanti-mouse F4/80 (clone CI:A3-1; AbD Serotec; 1/100 dilution inPBST) for 1 hour; rat anti-mouse Ly6G (clone 1A8; BD Bioscence;1/200 dilution in PBST) for 1 hour; rat anti-mouse Ly6C (clone ER-MP20; Thermo Fisher Scientific; 1/100dilution in PBST) for1 hour;rabbit polyclonal anti-human CD3 (cat. #A0452; DAKO; 1/100dilution in PBST) for 1 hour; rabbit polyclonal antibody againstcleaved-caspase 3 antigen (cat. #Asp175; Cell Signaling Technolo-gy; 1/200dilution inPBST) for 1 hour; and rabbitmonoclonal anti-p50NF-kB(cloneE381;Abcam;1/250dilution inPBST) for1hour.MACH1 Universal HRP Polymer (Biocare Medical) was used assecondary antibody for 30 minutes. The reaction was developedwith 3,30-diaminobenzidine (DAB), and nuclei were counter-stained with hematoxylin and thenmounted with Eukitt. The totalantigenþ area (mm2) and fraction area (total antigenþ area/totalarea of field at 200�) was evaluated by lab personnel under thesupervision of imaging core facility personnel using the ImageJanalysis program (http://rsb.info.nih.gov/ij/) in 200�microscopicfields selected within the neoplastic lesions ("tumor") and theadjacent nonneoplastic mucosa ("nontumor").

Colons from untreated and AOM/DSS-treated mice (WT andp50–/–) were frozen in OCT, and then sections were cut at 8 mmthickness. Colonic slides were fixed for 3minutes in cold acetone:chloroform (3:1), followed by incubated with 5% donkey serum,2% BSA, 0.1% Triton X-100, 0.2% NP-40, PBST solution for 1hour at room temperature to block unspecific binding sites.

Immunofluorescence staining was carried out at room temper-ature in humidified chamber with polyclonal rabbit anti-humanCD3 (cat. #A0452; DAKO; 1/100 dilution) for 1 hour andpolyclonal goat anti-mouse NKp46 (cat. #AF2225; R&D; 1/100dilution) for 1 hour. After being washed with PBST, slides wereincubated at room temperature in humidified chamber for 1 hourwith donkey anti-goat Alexa Fluor 488–conjugated and goat anti-rabbit Alexa Fluor 647–conjugated (both 1/1000 dilution) inPBST.Nucleiwere stainedwithDAPI (Life Technologies) and thenmounted with ProLong Antifade Gold Reagent (Life Technolo-gies). Slideswere analyzed by lab personnel under the supervisionof imaging core facility personnel using an Olympus FluoviewFV1000 laser scanning confocal microscope with 60� (N.A. 0.4),and single-cell count was performed (n� 5 field for every sample;n � 3 for every condition).

Fluorescence-activated cell sorting (FACS) analysisColons harvested from AOM/DSS-treated and untreated mice

(WT and p50–/–) were longitudinally cut and washed with PBSto remove feces and debris. Tumors and tumor-free colontissues were cut in 0.5 cm pieces and incubated under rotationin HBSS containing 5% FBS, 10 mmol/L HEPES, 2.5 mmol/LEDTA for 20 minutes at 37�C (twice). Tissues were then finelyminced and incubated under rotation in HBSS containing 5%FBS, 10 mmol/L HEPES, dispase/collagenase (1 mg/mL; Roche),collagenase IV (250mg/mL; Serva), andDNAse (40mg/mL; Roche)for 20 minutes at 37�C. Single-cell suspensions were sequentiallyfiltered through a 100-mm and 70-mm cell strainers and centri-fuged10minutes at 1300 rpm.Cellswere incubatedwith TruStainfcX (anti-CD16/32) antibody (BioLegend), according to theman-ufacturer's instructions. Afterward, 0.5 � 106 to 1 � 106 cellswere stained with antigen-specific antibodies in the presence ofLIVE/DEAD Fixable Violet (Invitrogen) to evaluate cell viability.Cells were stained in 0.5% FBS, HBSS solution with: anti-CD45–FITC or -PerCP (clone 30-F11), CD11b–PerCP or -APC (cloneM1/70), F4/80–PE-Cy7 or -PE (clone BM8), Ly6C-PE (cloneHK1.4), NK1.1-APC (clone PK136), CD4–PE-Cy7 (clone GK1.5),CD8-APC (clone 53-6.7), CD3-PE (clone 145-2C11), TNFa-APC(clone MP6-XT22), IL12–PE-Cy7 (clone C15.6), IFNg-FITC(clone XMG1.2; all from BioLegend), unconjugated polyclonalrabbit anti-mouse NOS2 (Abcam) followed by incubation withsecondary goat anti-rabbit Alexa Fluor 488–conjugated (Invitro-gen, Molecular Probes).

Cytofix/Cytoperm and Permwash staining kits (BD Pharmin-gen) were used for intracellular staining (TNFa, IL12, IFNg ,iNOS), according to the manufacturer's instructions. Expres-sion levels of cytokines (IL12, TNFa, IFNg) were evaluated after3 hours of stimulation with pherbol-12 13-acetate (PMA; 40ng/mL) and ionomycin (1 mg/mL) in the presence of brefeldinA (5 mg/mL; Sigma). Cells were acquired using FACSCanto II orLSR Fortessa (BD Bioscience), and data were analyzed using9.3.2 FlowJo software (Tree Star, Inc.). Mean fluorescenceintensity (MFI) was normalized to isotype control or fluores-cence minus one controls.

Statistical analysisData are expressed as mean � SEM. Statistical significance

between two groups was assessed by unpaired, one- or two-tailed Student t test or Mann–Whitney test, and multiple groups

Figure 2.p50 NF-kB promotes CAC by hampering M1/Th1 inflammation. To induce CAC, mice were treated with AOM and DSS. A, Top: Body weight monitored every2 to 3 days during the experimental period (multi t test, starting from day 10; � , FDR < 0.05;N¼ 16). Bottom: Colon lengthmeasured at the time ofmice sacrifice (day100). Data shown are mean � SEM of different mice from two independent experiments. ��� , P < 0.001 by two-tailed Mann–Whitney test. Untreated WT: N ¼ 6;untreated p50–/–: N ¼ 6; AOM/DSS-treated WT: N ¼ 17; AOM/DSS-treated p50–/–: N ¼ 12. Center: Ulceration and degree of inflammation (colitis score)analyzed on colon swiss roll sections by H&E. Data shown are mean � SEM of different mice from two independent experiments. � , P < 0.05 by two-tailedMann–Whitney test.WT:N¼ 11; p50–/–:N¼9.B,Colonswere longitudinally opened andpolypswere counted. Data shownaremean�SEMofdifferentmice from twoindependent experiments. ��� ,P<0.001 by two-tailedMann–Whitney test.WT:N¼ 11; p50–/–:N¼9. Representative images are shown. The total number and the sizeof tumors were counted and tumor burden was calculated for each animal. Data shown are mean � SEM of different mice or tumors from two independentexperiments. � , P < 0.05; ��� , P < 0.001 by two-tailed Mann–Whitney test. N ¼ 11 different WT mice; N ¼ 9 different p50–/– mice; N ¼ 52 WT tumors; N ¼ 9 p50–/–

tumors). Representative images are shown. Scale bars: 1,000 mm. C, Top: Total RNA from tumors of AOM/DSS-treated WT and p50–/– mice analyzedfor the expressionof the genes belonging to clusters 1 and2; bottom: additional Th1/M1 genes. Results are expressedas fold inductionoverWT tumor expression.Datashown aremean� SEMof differentmice from two independent experiments. � ,P<0.05 by one-tailedMann–Whitney testN¼ 13WTmice;N¼ 15 p50–/–mice.D, Top:FACS analysis of colorectal cancer lesions for selective M1 gene products in TAMs (CD11bþF4/80þ cells) and bottom: the frequency of IFNg-expressingT cells. Intracellular expression of cytokines (IL12, TNFa, and IFNg) wasmeasured upon 3 hours of stimulationwith 50 ng/mLPMAand 1mg/mL ionomycin in presenceof 5 mg/mL Brefeldin A. Data shown are mean � SEM. � , P < 0.05; �� , P < 0.01 by two-tailed t test. N ¼ 6 WT mice; N ¼ 5 p50–/– mice. Center: RepresentativeFACS histograms of M1 markers are shown.

p50 NF-kB Shapes Colorectal Cancer–Promoting Inflammation

www.aacrjournals.org Cancer Immunol Res; 6(5) May 2018 583

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

were analyzed by Kruskal–Wallis test or one-way ANOVA(GraphPad Prism software). P � 0.05 was considered statisti-cally significant.

The mean percentage (GraphPad Prism software) of p50þ

TAMs associated with the occurrence of postsurgical metastasiswas used to divide patients with high (above mean) and lownumbers of p50þ TAMs used in Kaplan-Meier curves for disease-free survival (DFS) and colorectal cancer-specific survival (CRC-SS). P values were calculated by log-rank test. Receiver operatingcharacteristic (ROC) curves were used to determine cutoff valuesfor the investigated transcripts associated to the occurrence ofpostsurgical progression. These cutoffs were used to dividepatients with high and low mRNA expression in Kaplan–Meircurves.

ResultsColitis-associated cancer development is paralleled by M2polarization

Whereas type 1 immune responses restrain colorectal cancerprogression (35), the molecular events driving the balancebetween antitumor and tumor-promoting inflammation in colo-rectal cancer remain unclear. We first addressed this issue in achemical model of colitis-associated colorectal cancer (CAC; 31).We analyzed the mRNA expression of relevant inflammatory

genes in colons from WT mice, obtained 9 days (colitis) and80 to 90 days (established tumor development) after the first DSSadministration.

As compared with control colons, we identified a gene cluster(gene cluster 1) that was similarly upregulated in both phases,colitis and tumor,whereas a second cluster of genes (cluster 2)wasmore expressed in established tumors (Fig. 1A). Other genestested included Ptgs2, which was more expressed in tumorscompared with control colons, and ArgI, which was moreexpressed in tumors than colitis. An increased expression of Ccl5was observed in colitis rather than in tumors, and Il12a, Prf1,Gzmb, Fasl, and Il21 were not significantly upregulated in bothcolitis and tumors (Supplementary Fig. S2). In the healthy muco-sa, adjacent to the excised tumor lesions, expression of both geneclusters was comparable to those observed in the colon ofuntreated control mice (Fig. 1A). The higher expression of cluster2 in tumor tissues, containing the tumor promoting genes Tnf andIl23a (3), along with markers of Th2/M2 polarized inflammation(Il10, Tgfb1, Ccl17, and Ccl22; ref. 36; Fig. 1A), represented atumor-promoting activity. In contrast, cluster 1 was comprised ofgenes associated with an antitumor Th1/M1-skewed immuneprofile (Il1b, Il6, Il12b, Il27, Ebi3, Cxcl9, Cxcl10, Nos2, andIfng; Fig. 1A; ref. 35). These results indicate that a shift in thepolarized inflammatory response from type 1 to type 2 occurredduring the transition from colitis to cancer.

Figure 3.

Administration of M1 cytokines inhibits CAC development. A, Treatment regimen with IL12 and CXCL10 during CAC induction. B, Body weight loss during thetreatment. � , FDR < 0.05 by multi t test of treated versus vehicle. N ¼ 10 vehicle-treated mice; N ¼ 7 IL12-treated mice; N ¼ 8 CXCL10 treated mice. C, Colonlength of mice treated with vehicle, IL12, or CXCL10, and D, analysis of tumor development (number of polyps) was performed at day 80 on longitudinally openedcolons. Representative images are shown. Data shown are mean � SEM. � , P < 0.05 by one-tailed Mann–Whitney test. N ¼ 5 vehicle-treated mice; N ¼ 6IL12-treated mice; N ¼ 6 CXCL10-treated mice.

Porta et al.

Cancer Immunol Res; 6(5) May 2018 Cancer Immunology Research584

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

To investigate the molecular basis of this transcriptionalreprogramming, we focused on macrophages. Because nuclearaccumulation of p50 in macrophages promotes an M2-liketranscriptional program (28, 29), we analyzed the nuclear

levels of the p50 and p65 NF-kB subunits, in both laminapropria macrophages and TAMs. Confocal microscopy showeda selective accumulation of nuclear p50 over p65 in TAMscompared with lamina propria macrophages from control mice

Figure 4.

Lack of p50 in ApcMin mice inhibits spontaneous intestinal tumor development. A, Histological analysis of small gut and colons harvested from ApcMin andApcMinp50–/–mice at 12 or 18 weeks of age. Number and size of tumorswasmonitored, and tumor burden calculated for eachmouse. Data shown aremean� SEM ofdifferent mice or tumors. � , P < 0.05; �� , P < 0.01; ��� , P < 0.001 by two-tailed Mann–Whitney test. N ¼ 7 ApcMin mice 12 weeks old; N ¼ 8 ApcMinp50–/– mice12 weeks old; N ¼ 13 ApcMin mice 18 weeks old; N ¼ 15 ApcMinp50–/– mice 18 weeks old; N ¼ 76 tumors from ApcMin mice 12 weeks old; N ¼ 39 tumors fromApcMinp50–/–mice 12 weeks old;N¼ 155 tumors fromApcMin mice 18 weeks old;N¼ 133 tumors fromApcMinp50–/–mice 18 weeks old). B, Survival time of ApcMin andApcMinp50–/– mice. ��� , P < 0.0001 by log-rank Mantel–Cox test. N ¼ 54 ApcMin mice; N ¼ 26 ApcMinp50–/– mice. C and D, expression of gene clusters 1 and 2, andadditional Th1/M1 genes in colorectal cancer lesions from ApcMin mice versus C, the adjacent healthy colonic mucosa and D, colorectal cancer lesions fromApcMinp50–/– mice. Data shown are mean � SEM. � , P < 0.05 by one-tailed, Mann–Whitney test. N ¼ 6 ApcMin mice; N ¼ 6 ApcMinp50–/– mice.

p50 NF-kB Shapes Colorectal Cancer–Promoting Inflammation

www.aacrjournals.org Cancer Immunol Res; 6(5) May 2018 585

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

Figure 5.

Modulation of gut-associated leukocyte populations by p50 NF-kB. A, Formalin-fixed and paraffin-embedded colons from AOM/DSS-treated and untreated mice(WT and p50–/–) were evaluated for the number of macrophages (F4/80þ), monocytes (Ly6Cþ), neutrophils (Ly6Gþ), and T lymphocytes (CD3þ) byimmunohistochemistry. � , P < 0.05; �� , P < 0.01; ��� , P < 0.001 one-tailed Mann–Whitney test of WT (N ¼ 5) versus p50–/– (N ¼ 4) mice. B, Immunofluorescenceanalysis of frozen colonic samples were performed to evaluate NK (CD3–NKp46þ) and NKT (CD3þNKp46þ) cells. Data shown are mean � SEM of differenttumors (WT N¼ 15; p50–/–N¼ 5) or fields (WT N¼ 23; p50–/–N¼ 23) of different mice. � , P < 0.05; �� , P < 0.01 one-tailed Mann–Whitney test of WT (N¼ 6) versusp50–/– (N¼ 3)mice.C, FACS analysis of colorectal cancer lesions for the frequency of the depicted immune cell populations. Data shown aremean� SEMof differentmice. �� , P < 0.01; ��� , P < 0.001 two-tailed t test. N ¼ 4 WT mice; N ¼ 4 p50–/– mice. D, Transcripts of genes encoding for markers of different leukocytepopulations from healthy colons and tumors of untreated and AOM/DSS-treated mice. Results are shown as fold induction over healthy untreated WT mice. Datashown are mean � SEM. � , P < 0.05 by one-tailed Mann–Whitney test. N ¼ 4 untreated WT mice; N ¼ 4 untreated p50–/– mice; N ¼ 5 AOM/DSS-treated WTmice; N ¼ 6 AOM/DSS-treated p50–/– mice). E, T cells were depleted from WT and p50–/– mice during the entire experimental period by i.p. injections of anti-CD4(a-CD4) and anti-CD8 (a-CD8). Control mice received vehicle (�) only. Tumor development was evaluated. Representative images of longitudinally openedcolons at day 80 are shown. Data shown are mean � SEM. �, P < 0.05 two-tailed Kruskal–Wallis test N ¼ 11 vehicle-treated WT mice; N ¼ 10 vehicle-treatedp50–/– mice; N ¼ 10 anti-CD4/CD8–treated WT mice; N ¼ 7 anti-CD4/CD8–treated p50–/– mice).

Porta et al.

Cancer Immunol Res; 6(5) May 2018 Cancer Immunology Research586

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

(Fig. 1B). These results indicate that the tumor-promotingreprogramming of colorectal cancer–associated inflammationoccurred during tumor development, in conjunction withnuclear accumulation of p50 NF-kB in TAMs.

p50 NF-kB deficiency restrains CAC developmentTo investigate the impact of p50 NF-kB ablation in the inflam-

matory reprogramming supporting CAC development, WT andp50–/– mice were treated with AOM/DSS. We initially examined

Figure 6.

Lack of p50 increases apoptosis of both colonic epithelial and tumor cells after AOM/DSSadministration.A,Activated cleaved caspase-3 in colon sections fromAOM/DSS-treated and untreated mice by immunohistochemistry and digital image analysis. Representative images are shown. Data shown are mean� SEM of differentfields. Scale bars, 20 mm. � , P < 0.05; �� , P < 0.01 by one-tailed Mann–Whitney test. N¼ 4 fields from 1 untreatedWTmice; N¼ 4 fields from 1 untreated p50–/–mice;N ¼ 16 total tumor fields and N ¼ 14 total nontumor fields from 5 different AOM/DSS-treated WT mice; N ¼ 8 total tumor fields and N ¼ 14 total nontumorfields from 3 different AOM/DSS-treated p50–/– mice. B, Transcripts of survival genes were evaluated in total RNA isolated from colon and tumor lesions ofuntreated and AOM/DSS-treated mice. Normalized qPCR results are shown as fold induction over healthy untreated WT mice. Data shown are mean � SEM.� , P < 0.05 by one-tailed Mann–Whitney test. N ¼ 2 untreated WT mice; N ¼ 2 untreated p50–/– mice; N ¼ 5 AOM/DSS-treated WT mice; N ¼ 6 AOM/DSS-treated p50–/– mice.

p50 NF-kB Shapes Colorectal Cancer–Promoting Inflammation

www.aacrjournals.org Cancer Immunol Res; 6(5) May 2018 587

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

Cancer Immunol Res; 6(5) May 2018 Cancer Immunology Research588

Porta et al.

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

the inflammatory response developed by WT and p50–/– miceafter a single round of DSS. No differences in colon length wereobserved betweenWT andp50–/– controlmice drinking untreatedwater (Supplementary Fig. S3B, right), whereas with DSS treat-ment, lack of p50 resulted in a higher mortality rate (Supplemen-tary Fig. S3A), severe body weight loss (Supplementary Fig. S3B,left), and colon shortening (Supplementary Fig. S3B, right),indicating that the p50 NF-kB subunit is an essential regulatorof chemically induced intestinal inflammation.

To examine the role of p50 NF-kB in CAC, WT and p50–/–

mice were treated with AOM in combination with three roundsof DSS treatment. In keeping with the acute model of colitis,lack of p50 resulted in increased weight loss (Fig. 2A, top) and ahigher degree of intestinal inflammation, in both number ofulcers in colon tissues and colitis score, over the entire exper-imental period (Fig. 2A, middle). Compared with WT mice, atnecropsy, p50–/– mice showed a significant decrease in colonlength (Fig. 2A, bottom). Conversely, compared with WT mice,both macroscopic and histologic analysis of p50–/– colonsshowed a significant decrease in the number of neoplasticlesions, together with reduced size (Fig. 2B). Collectively, theseresults indicate a tumor-promoting role for p50-driven inflam-mation in CAC development.

Hence, we examined the expression of genes belonging toclusters 1 and 2 (Fig. 1A) in the tumor-resistant, p50-deficientmice. Compared with their WT counterpart, we observed a stronginhibition of Il23a expression in tumors, paralleled by a signif-icant upregulation of Th1/M1 inflammatory genes (Il12b, Il27,Ebi3, Cxcl9, Cxcl10, Nos2, and Ifng;Fig. 2C, top). Consistently, thelack of p50 also resulted in higher expression of Prf1, GzbB, Fasl,Il21, and Ccl5(Fig. 2C, bottom). FACS analysis of tumor immuneinfiltrates confirmed that genetic ablation of p50 in TAMsenhanced expression of M1-related gene products (e.g., TNFa,IL12, iNOS; Fig. 2D, top), along with increased frequency of IFNgexpressingCD8þ andCD4þT cells (Fig. 2Dbottom). These resultsindicate that boosting of CACdevelopment by p50was paralleledby its ability to restrain Th1/M1 immune responses.

To strengthen the association between antitumor resistanceand gene cluster 1, AOM/DSS-treated mice underwent systemic(intraperitoneal) or local (intrarectal) administration of selectedTh1/M1 cytokines/chemokines, IL12 (100 ng) and CXCL10 (500ng), once aweek (Fig. 3A).Despite a similar bodyweight loss (Fig.3B) and colon length (Fig. 3C) in the untreated and treated

groups, both IL12 and CXCL10 induced a significant reductionin tumormultiplicity (Fig. 3D), confirming the antitumor activityof the Th1/M1-polarized immune response.

Because the molecular determinants of intestinal carcino-genesis may differ in models of CAC compared with geneticallydriven colorectal cancers (37, 38), we evaluated the role of p50NF-kB in the ApcMin mouse model of spontaneous intestinalcarcinogenesis (39). To this aim, ApcMin and p50–/– mice werecrossed to generate double ApcMinp50–/– mice. Subsequently,small intestines and colons were harvested from both ApcMin

and ApcMinp50–/– mice at different time points (12 and 18weeks) and were analyzed for the number of tumor lesions. Asexpected, tumor multiplicity and size increased over time.However, compared with ApcMin mice, a significant inhibitionof both tumor incidence and growth was observed in theApcMinp50–/– group (Fig. 4A), which was associated withincreased survival from 6 (mean 22 weeks) to 10 months(mean 41 weeks; Fig. 4B).

We then investigated whether the inflammatory responsedetected in the CAC tumormodel was also activated in the ApcMin

mice. Because ApcMin mice largely develop tumors in the smallintestine, colorectal cancer lesions were obtained from the colonof 23-week-oldmice, when tumors arise also in the large intestine.Similar to the CACmodel, most of the genes belonging to clusters1 and 2 were upregulated in tumor tissues compared with theadjacent healthymucosa (Fig. 4C), whereas lack of p50 confirmedthe selective reduction in the tumor-promoting Il23a gene tran-script, along with the enhanced expression of Th1/M1 genes (Fig.4D). The results obtained in both the CAC and the ApcMinmodelsindicate that, irrespective of the etiological events, p50 promotesintestinal cancer development by hampering Th1/M1-dependentantitumor responses.

To gain additional insights in the p50-driven M2 reprogram-ming of TAMs, we adopted the immunogenic MC38 transplant-able colorectal cancer model, characterized by an abundant TAMinfiltrate (40). As expected, tumor growth was significantly inhib-ited in p50–/–mice (Supplementary Fig. S4A). Comparedwith theWT counterparts, TAMs purified from p50–/– mice by magneticcell sorting (Supplementary Fig. S4B) displayed enhanced mRNAexpression of selected M1 genes (Cxcl9, Il12b, Ifng, Il1b, Tnf, andIl23a; Supplementary Fig. S4C), along with increased proteinlevels of TNFa and CXCL9 (Supplementary Fig. S4D). Consistentwith the tumor-promoting activity of p50-driven, M2-polarized

Figure 7.Correlation between the number of p50NF-kBþ TAMs and the clinical response of colorectal cancer patients.A,Histologic analysis of colons fromAOM/DSS-treatedchimericmice (p50–/–BM inWT recipients,N¼6andWTBM in p50–/– recipients,N¼ 7)with respect to controlmice (N¼ 14 lethally irradiatedWTmice reconstitutedwith WT BM). � , P < 0.05 by one-tailed Mann–Whitney test. B, Analysis of colorectal cancer lesions in AOM/DSS-treated p50Fl/Fl (N ¼ 6) and p50Fl/Fl;Lyz2Cre(N ¼ 7; top) and p50Fl/Fl (N ¼ 6); p50Fl/Fl;Villin-Cre (N ¼ 6; bottom) mice. Data shown are mean � SEM. � , P � 0.05 by one-tailed Mann–Whitney test. C,Immunofluorescence of nuclear p50þTAMs (CD68þ cells) in colorectal cancer lesions from26 colorectal cancer patients, stage II and III. Left, analysis of the density ofTAM with p50þ nuclei at the invasive margin in relation with disease recurrence within 4 years from surgery. Two-tailed t test; �� , P < 0.01. Center and right, theKaplan–Meier analysis shows both DFS and colorectal cancer-specific survival (CRC-SS) of colorectal cancer patients, in relation to high (N ¼ 10; >mean;black line) or low (N ¼ 16; <mean; gray line) number of p50þTAMs at the invasive margin. P values calculated by log-rank Mantel–Cox test; P < 0.05 consideredsignificant. D, Expression of selected type I inflammatory genes in total RNA obtained from 26 colorectal cancer patients (stage II and III colorectal cancer)in relation to high (>mean; black dots) or low (<mean; gray squares) number of p50þTAMs located at the invasivemargin. Data cleaned by outliers passedD'Agostinoand Pearson omnibus normality test and analyzed by two-tailed t test; � , P < 0.05; �� , P < 0.01. E and F, Expression of selected type 1 inflammatory genes intotal RNA from 47 stage II and III human colorectal cancer specimens. Results are normalized over the housekeeping gene b-actin. For each gene transcript, cutoffvalue was extrapolated by ROC curve analysis (IL12A 4.28� 10�5; IL12B 7.88� 10�5; TBX21 3.35� 10�4; CXCL9 3.72� 10�3; CXCL10 8.32� 10�3; IL21 1.13� 10�4). TheKaplan–Meier curves show (E) DFS and (F) CRC-SS of colorectal cancer patients in relation with the expression of the selected type 1 inflammatory genesin tumor samples. Black line: low gene transcripts (<cutoff value; IL12A N¼ 18; IL12B N¼ 37; TBX21 N¼ 32; CXCL9 N¼ 26; CXCL10 N¼ 29; IL21 N¼ 25); gray line: highgene transcripts (>cutoff value; IL12A N ¼ 29; IL12B N ¼ 10; TBX21 N ¼ 15; CXCL9 N ¼ 21; CXCL10 N ¼ 18; IL21 N ¼ 22). P values were calculated by log-rankMantel–Cox test; P < 0.05 considered significant.

www.aacrjournals.org Cancer Immunol Res; 6(5) May 2018 589

p50 NF-kB Shapes Colorectal Cancer–Promoting Inflammation

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

activation, coinjection of magnetically sorted WT TAMs signifi-cantly enhanced the growth of the MC38 tumors in p50-deficientmice compared with coinjection of p50–/– TAMs (SupplementaryFig. S4E).

Lack of p50 restrains TAM accumulation and favors T-cellinfiltration

Taken together, data from the preclinical models indicate thatp50 NF-kB impairs antitumor immunity. Hence, we investigatedthe composition of the immune infiltrate in untreated and AOM/DSS-treated conditions, in both WT and p50-deficient mice. Nochanges were observed in the number of tumor and laminapropria neutrophils (Ly6Gþ cells), whereas lack of p50 resultedin reduced tumor and lamina propria monocytes (Ly6Cþ cells)and macrophages (F4/80þ cells) and an increased number of Tlymphocytes (CD3þ cells), NK (NKp46þCD3–), and NKT(NKp46þCD3þ) cells in both tumors and adjacent healthy tissues(Fig. 5A and B; Supplementary Fig. S5A and S5B). SupplementaryFig. S6 shows that in the absence of p50, the number of mucosalmonocytes (Ly6Cþ) andmacrophages (F4/80þ) was significantlyreduced, even in untreated mice, whereas colonic NK, NKT, and Tcells were unaffected. FACS analysis of colorectal cancer lesionsisolated from AOM/DSS-treated mice further confirmed that p50deficiency impairs TAM accumulation, while increasing the fre-quency of lymphoid effector cells (NK, NKT, CD8þ T cells; Fig.5C). This observation was corroborated through the expressionanalysis of genes encoding for lineage-specific markers, demon-strating that lack of p50 promotes an increased mRNA expressionof both the Cd8 and Tbx21 (Tbet) genes, whereas the transcriptsfor Cd4 and the transcription factors Gata3, Rorc2 (RORgt), andFoxp3, respectively, guiding differentiation of Th2/ILC2, Th17/ILC3, and Tregs did not change (Fig. 5D). These results furtherindicate that p50 NF-kB was involved in controlling the homingof lymphoid and myeloid cells to the gut and, hence, playing acrucial role in the control of intestinal innate and adaptiveimmune response. To assess the role of T cells in antitumorimmunity consequent to p50 ablation, animalswere next injectedonce a week with depleting anti-CD4 and anti-CD8 during theentire AOM/DSS treatment. Depletion of CD4þ and CD8þ T cellsresulted in a consistent augmentation of tumor burden in p50knockouts (Fig. 5E), confirming that p50 reduced both recruit-ment and antitumor activity of T cells.

Lack of p50 NF-kB impairs survival of colorectal cancer cellsOur data support the concept that the absence of p50 NF-kB

increases cytotoxic immune functions, thus preventing tumormultiplicity and growth. We then investigated colonic cancer cellproliferation and survival in WT and p50–/– mice. Colons wereharvested from untreated and AOM/DSS treatedmice and immu-nostained with anti-Ki67 and anti-active caspase-3. Although inboth untreated and AOM/DSS-treated conditions, lack of p50generated longer crypts with a higher proliferation rate of colonicepithelial cells, it did not affect cancer cell proliferation (Supple-mentary Fig. S7A). In accordance with this finding, p50 deficiencyincreased the expression of genes associated with cell-cycle pro-gression in the normal colonic mucosa from either untreated orAOM/DSS-treated mice, whereas no differences were foundbetween WT and p50–/– cancer cells (Supplementary Fig. S7B).AlongwithNF-kB, STAT3 is a key orchestrator of cell proliferationand survival, which is upregulated in AOM/DSS-induced colo-rectal cancer (9, 10). Immunohistochemical analysis of WT and

p50–/– colon tissues showed similar STAT3 phosphorylationbetween WT and p50–/– neoplastic cells (Supplementary Fig.S7C), suggesting that, in this context, p50 acts independentlyfrom STAT3. In contrast, colons from AOM/DSS-treated p50–/–

mice displayed an increased number of cells expressing theactivated form of caspase-3, indicating that absence of p50impairs epithelial and cancer cell survival (Fig. 6A). Accordingly,tumors harvested from p50–/– mice expressed higher proapopto-tic Bak1 and lower prosurvival Bcl2l1 (Fig. 6B), whereas nodifferences were found in the expression of Bcl2 (antiapoptotic),Survivin (antiapoptotic), and Bax (proapoptotic) genes (Fig. 6B).No differences were detected in the number of apoptotic cells(Fig. 6A) and gene transcripts (Fig. 6B) between the colons of WTand p50–/–-na€�ve mice. Overall, these results indicate that inflam-matory conditions arising in the absence of p50 impaired survivalof both colonic epithelial and cancer cells.

Density of p50þ TAMs is directly associated with a worsecolorectal cancer patient outcome

To assess the role of p50 NF-kB in both macrophages andintestinal epithelial cells in CACdevelopment, p50–/– andWTBMcells were respectively transplanted in sublethally irradiated WTand p50–/– recipients. Compared with control mice (WT micereconstituted with WT BM), only chimeric mice carrying hemato-poietic-specific ablation of p50 displayed a tumor resistant phe-notype in response to AOM/DSS treatment (Fig. 7A). We thengeneratedmice carrying a specific ablation of p50 inmyeloid cells(p50Fl/Fl;Lyz2Cre) and in enterocytes (p50Fl/Fl;Villin-Cre). Fol-lowing AOM/DSS treatment, p50Fl/Fl;Lyz2Cre mice developedless colorectal cancer lesions than littermate controls (p50Fl/Fl),whereas p50Fl/Fl;Villin-Cre and p50Fl/Fl mice showed a similarnumber of tumors (Fig. 7B). These results demonstrate the role ofmyeloid-specific p50 in tumor promotion, fostering the analysisof its impact on colorectal cancer patients' outcome. To this aim,we assessed the nuclear expression of p50 in CD68þ TAMs fromstage II and III colorectal cancer patients (Supplementary TableS1). A direct correlation was observed between the percentage ofp50þCD68þ TAMs within the tumor and those at the invasivefront (Correlation coefficient (r) ¼ 0,733161; r2 ¼ 0,537524).Such correlation was similar in stage II and stage III colorectalcancers.

The percentage of nuclear p50þCD68þ TAMs at the invasivefront (but not of those within the tumor) was significantlyhigher (two-tailed unpaired t-test, P ¼ 0.007) in patientswith postsurgical progression (Fig. 7C, left and Supplemen-tary Fig. S8B, left). Consistently, the higher density of p50þ

TAMs at the invasive front (but not of thosewithin the tumor)wasassociated with a significantly worse patient outcome, assessedboth as relapse (DFS, log-rank test, P ¼ 0.0032) and as colorectalcancer–specific survival (CRC-SS, log-rank test, P ¼ 0.0010;Fig. 7C, center and right; Supplementary Fig. S8B, center andright).

Next, we investigated whether nuclear accumulation of p50in peritumoral TAMs correlated with inhibition of M1-drivenantitumor immunity. To this aim, we analyzed the expressionof mRNAs coding for type 1 inflammatory genes (IL12A, IL12B,CXCL9, CXCL10, IL21, and TBX21) in tumor tissues of stage IIand III colorectal cancers. We observed that the mRNA expres-sion of most inflammatory genes was significantly correlatedamong themselves, with a clear inverse correlation betweenIL12A mRNA and the density of p50þCD68þ TAMs at the

Cancer Immunol Res; 6(5) May 2018 Cancer Immunology Research590

Porta et al.

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

invasive front (Kendall tau, �0.37; P ¼ 0.0072). In agreement,comparing M1/Th1 gene expression in groups of patientscharacterized by high and low numbers of p50þ TAMs, wefound an inverse correlation between the density of p50þ TAMsand the expression of both the IL12A and TBX21 genes (datacleaned by outliers passed D'Agostino and Pearson omnibusnormality test and were analyzed by two-tailed t-test; Fig. 7D).We observed that the mRNA of CXCL9, CXCL10, and IL21consistently tended to be lower in the group with a highnumber of p50þ TAMs (Fig. 7D). Higher expression of theinflammatory genes IL12A, TBX21, CXCL9, CXCL10, and IL21were associated with increased DFS and lower disease progres-sion (Fig. 7E), as well as with improved CRC-SS (Fig. 7F). Ourresults indicate that the accumulation of p50 NF-kB in TAMslocated at the invasive edge of colorectal cancer and converselylow levels of type I inflammation are associated with a worseclinical outcome.

DiscussionFollowing the recognition of infiltrating immune cells as

determinants of tumor progression (41), the concept that robustadaptive immune responses in the tumor microenvironment cancounteract colorectal cancer progression toward systemic dissem-ination, and, thus positively affect patient outcome, has beenexemplified by a prognostic immunoscoremeasuring the amountof T lymphocytes (14). Differently, the accumulation of TAMs,acting as orchestrator of cancer-related inflammation (4, 16), isusually associated with an M2-polarized, tumor-promoting phe-notype (36) and poor prognosis inmost human tumors (17–19).However, TAMs' impact on colorectal cancer remains controver-sial. Here, we report that accumulation of p50þ TAMs at theinvasive margin negatively affected the outcome of patients withstage II/III colorectal cancer. A high number of p50þ TAMs wasassociated with low expression of both IL12A and TBX21 mRNAand, in general, with decreased expression of Th1/M1genes. Thesefindings support the concept that nuclear accumulation of p50 inTAMs restrains Th1/M1-dependent antitumor responses andmark a subgroup of TAMs with protumor features in colorectalcancer thatmay negatively affect outcome. Several evidences fromour preclinical models also support the M2 relevance of the p50þ

TAMs.Although the M1 to M2 switch of TAM functions has been

largely debated (11, 13, 36, 42), the molecular basis and clinicalrelevance of this dynamic reprogramming has been only partiallyelucidated. The results of our study, obtained using either theAOM/DSS or the ApcMin mouse models, both of which developmultiple polyps (19, 39), indicate that independent of the etio-logical trigger, the dynamic and p50-dependent inhibition of M1polarization ofmacrophages is a crucial event for intestinal tumorgrowth. Our data indicate that p50 accumulation in tumorsparalleled increased expression of a gene cluster (cluster 2) con-taining both M2-related (Il10, Tgfb1, Ccl17, and Ccl22) andselected tumor-promoting (Tnf and Il23a) genes. This dynamic,transcriptional transition appears to create tumor-promotingconditions by hampering the cytotoxic actions of the Th1/M1-polarized inflammatory and cytotoxic gene cluster 1.

Analysis of the leukocyte infiltrate in WT and p50-deficienttumor-bearingmice revealed a strong reduction of both TAMs andlamina propria macrophages. As gut macrophages are constantlyreplenished by BM-derived monocytes (43), our results suggest

that p50 is required for intestinal monocyte/macrophage turn-over, suggesting additional roles of p50 in macrophage biology,which underlines the need of future studies to further dissect therole of p50 in myeloid cell biology and tumor development.

At steady state, NK, NKT, and T cells were similarly presentin the lamina propria of WT and p50–/– mice, whereas theyincreased in both the adjacent healthy mucosa and tumor lesionsof AOM/DSS-treated mice. In accordance with these findings,p50–/– tumors expressed higher type 1-specific chemokines (e.g.,Cxcl9 and Cxcl10), cytokines (Il12a, Il12b, Il27, and Il21), andeffector molecules (Ifng, Prf1, Gzmb, and Fasl). Although IL21 cansupport colorectal cancer by promoting Th17-driven inflamma-tion (44–46), this cytokine can also elicit antitumor effects(47, 48) by favoring the development, expansion, and cytotoxicactivities of CD8þ T cells, NKT, and NK cells (49). Of relevance,increased Il21 expression in p50–/– tumor-bearing mice did notcorrelate with enhanced expression of neither Il17a nor Rorc2(RORgt), but rather with increased genes expressed by cytotoxiclymphoid cells [e.g., Ifng, Prf1, Gzmb, Fasl, Tbx21 (Tbet)] andfrequency of IFNg expressing CD8þ and CD4þ T cells. Hence, inline with previous reports (50), lack of p50 appears to selectivelyenhance the antitumor effects of IL21, without engaging protu-moral Th17-driven inflammation.

In response to gutmicrobial products,myeloid cells orchestratetumor-promoting IL23 and IL17 responses (51). In particular,IL23 drives Th17 cells expansion and functions (52) and inhibitsNK cell effector functions (53). Accordingly, the increased expres-sion of IL23p19, occurring during the AOM-DSS–driven transi-tion from colitis to tumor, was prevented in p50–/– colorectalcancer and ApcMinp50–/– mice. Overall, our results identify amechanism by which p50 NF-kB promotes colorectal cancerprogression by limiting Th1/M1 antitumor functions. This eventis paralleled by both decreased accumulation of effector NK, NKT,andT cells and specific antitumor immunity. Although ablationofp50 may also impair colorectal cancer cell survival in a cell-autonomous manner (54), our conclusion is supported by theobservation that depletion of CD4þ and CD8þ T lymphocytesabolished the resistance of p50-deficientmice to colitis-associatedcolorectal cancer and that adoptive transfer of WT TAMs rescuedMC38 tumor growth in p50–/–mice. The expression of p50NF-kBitself, like that of the related cytokines and chemokines, has clearprognostic implications and deserves proper assessment in largeclinical studies. Although blocking the activity of negative reg-ulators of the inflammatory response, including p50, may poten-tially elicit adverse reactions (55), administration of type 1immune stimulators, such as IL12 and CXCL10, could offsetadverse effects and restore antitumor immunity.

Disclosure of Potential Conflicts of InterestE. Hirsch is founder of Kither Biotech. No potential conflicts of interest were

disclosed by the other authors.

Authors' ContributionsConception and design: C. Porta, A. Ippolito, A. SicaDevelopment of methodology: C. Porta, A. Ippolito, G. Celesti, C. Correale,S. Tartari, F. Balzac, E. TurcoAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): C. Porta, A. Ippolito, F.M. Consonni, F. Grizzi,S. Tartari, F. Balzac, E. Turco, E. Hirsch, L. LaghiAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): C. Porta, A. Ippolito, F.M. Consonni, L. Carraro,G. Celesti, F. Grizzi, M. Rinaldi, P. Bianchi, L. Laghi

www.aacrjournals.org Cancer Immunol Res; 6(5) May 2018 591

p50 NF-kB Shapes Colorectal Cancer–Promoting Inflammation

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

Writing, review, and/or revision of the manuscript: C. Porta, L. Carraro, P.Bianchi, S. Vetrano, L. Laghi, A. SicaStudy supervision: A. SicaOther (carried out experiment): F. Pasqualini

AcknowledgmentsThis work was supported by Ministero Universita Ricerca, Italy

(#20103FMJEN_003 and #RBAP11H2R9_005), Associazione Italiana Ricercasul Cancro, Italy (Nos. 1558 and 19885), Ministero della Salute, Italy (129/GR-2011-02349580), Universit�a degli Studi del Piemonte Orientale (researchfunding 2016).

The authors acknowledge Dr. Camilla Recordati within Fondazione Filarete,Milano, andDr. VincenzoArenaUniversita Cattolica del SacroCuore, Roma, foranimal histopathology.

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 January 19, 2017; revised November 27, 2017; accepted March 9,2018; published first March 27, 2018.

References1. Rogler G. Chronic ulcerative colitis and colorectal cancer. Cancer Lett

2014;345:235–41.2. Quante M, Varga J, Wang TC, Greten FR. The gastrointestinal tumor

microenvironment. Gastroenterology 2013;145:63–78.3. Wang K, Karin M. Tumor-elicited inflammation and colorectal cancer.

Adv Cancer Res 2015;128:173–96.4. Mantovani A, Allavena P, Sica A, Balkwill F. Cancer-related inflammation.

Nature 2008;454:436–44.5. Ruffell B, Affara NI, Coussens LM. Differential macrophage program-

ming in the tumor microenvironment. Trends Immunol 2012;33:119–26.

6. Porta C, Riboldi E, Ippolito A, Sica A. Molecular and epigenetic basis ofmacrophage polarized activation. Semin Immunol 2015;27:237–48.

7. Schwitalla S, Ziegler PK, Horst D, Becker V, Kerle I, Begus-Nahrmann Y,et al. Loss of p53 in enterocytes generates an inflammatory microenviron-ment enabling invasion and lymph nodemetastasis of carcinogen-inducedcolorectal tumors. Cancer Cell 2013;23:93–106.

8. Greten FR, Eckmann L, Greten TF, Park JM, Li ZW, Egan LJ, et al. IKKbetalinks inflammation and tumorigenesis in a mouse model of colitis-associated cancer. Cell 2004;118:285–96.

9. Grivennikov S, Karin E, Terzic J, Mucida D, Yu GY, Vallabhapurapu S,et al. IL-6 and Stat3 are required for survival of intestinal epithelial cellsand development of colitis-associated cancer. Cancer Cell 2009;15:103–13.

10. Bollrath J, Phesse TJ, von Burstin VA, Putoczki T, Bennecke M, Bateman T,et al. gp130-mediated Stat3 activation in enterocytes regulates cell survivaland cell-cycle progression during colitis-associated tumorigenesis. CancerCell 2009;15:91–102.

11. Goktuna SI, Canli O, Bollrath J, Fingerle AA, Horst D, Diamanti MA, et al.IKKalpha promotes intestinal tumorigenesis by limiting recruitment ofM1-like polarized myeloid cells. Cell Rep 2014;7:1914–25.

12. Pasparakis M. Regulation of tissue homeostasis by NF-kappaB signalling:implications for inflammatory diseases.Nat Rev Immunol 2009;9:778–88.

13. Mantovani A, Sica A. Macrophages, innate immunity and cancer: balance,tolerance, and diversity. Curr Opin Immunol 2010;22:231–7.

14. Galon J,Mlecnik B, BindeaG, AngellHK, Berger A, Lagorce C, et al. Towardsthe introduction of the `Immunoscore' in the classification of malignanttumours. J Pathol 2014;232:199–209.

15. Laghi L, Bianchi P, Miranda E, Balladore E, Pacetti V, Grizzi F, et al. CD3þcells at the invasive margin of deeply invading (pT3-T4) colorectal cancerand risk of post-surgical metastasis: a longitudinal study. Lancet Oncol2009;10:877–84.

16. Qian BZ, Pollard JW. Macrophage diversity enhances tumor progressionand metastasis. Cell 2010;141:39–51.

17. Bingle L, Brown NJ, Lewis CE. The role of tumour-associated macrophagesin tumour progression: implications for new anticancer therapies. J Pathol2002;196:254–65.

18. Steidl C, Lee T, Shah SP, Farinha P, Han G, Nayar T, et al. Tumor-associatedmacrophages and survival in classic Hodgkin's lymphoma. N Engl J Med2010;362:875–85.

19. Dannenmann SR, Thielicke J, Stockli M,Matter C, von Boehmer L, CecconiV, et al. Tumor-associated macrophages subvert T-cell function and cor-relate with reduced survival in clear cell renal cell carcinoma.Oncoimmunology 2013;2:e23562.

20. Zhang QW, Liu L, Gong CY, Shi HS, Zeng YH, Wang XZ, et al. Prognosticsignificance of tumor-associated macrophages in solid tumor: a meta-analysis of the literature. PLoS One 2012;7:e50946.

21. Erreni M, Mantovani A, Allavena P. Tumor-associated macrophages(TAM) and inflammation in colorectal cancer. Cancer Microenviron2011;4:141–54.

22. Braster R, Bogels M, Beelen RH, van Egmond M. The delicate balance ofmacrophages in colorectal cancer; their role in tumour development andtherapeutic potential. Immunobiology 2017;222:21–30.

23. Norton SE, Ward-Hartstonge KA, Taylor ES, Kemp RA. Immune cellinterplay in colorectal cancer prognosis. World J Gastrointest Oncol2015;7:221–32.

24. Kang JC, Chen JS, Lee CH, Chang JJ, Shieh YS. Intratumoral macrophagecounts correlate with tumor progression in colorectal cancer. J Surg Oncol2010;102:242–8.

25. Forssell J, Oberg A, Henriksson ML, Stenling R, Jung A, Palmqvist R. Highmacrophage infiltration along the tumor front correlates with improvedsurvival in colon cancer. Clin Cancer Res 2007;13:1472–9.

26. WangW, Li X, Zheng D, Zhang D, Peng X, Zhang X, et al. Dynamic changesand functions of macrophages and M1/M2 subpopulations during ulcer-ative colitis-associated carcinogenesis in an AOM/DSS mouse model. MolMed Rep 2015;11:2397–406.

27. Edin S, Wikberg ML, Dahlin AM, Rutegard J, Oberg A, Oldenborg PA, et al.The distribution ofmacrophages with aM1 orM2 phenotype in relation toprognosis and the molecular characteristics of colorectal cancer. PLoS One2012;7:e47045.

28. Saccani A, Schioppa T, Porta C, Biswas SK, Nebuloni M, Vago L, et al. p50nuclear factor-kappaB overexpression in tumor-associated macrophagesinhibits M1 inflammatory responses and antitumor resistance. Cancer Res2006;66:11432–40.

29. Porta C, Rimoldi M, Raes G, Brys L, Ghezzi P, Di Liberto D, et al. Toleranceand M2 (alternative) macrophage polarization are related processesorchestrated by p50 nuclear factor kappaB. Proc Natl Acad Sci U S A2009;106:14978–83.

30. Miranda E, Bianchi P, Destro A, Morenghi E, Malesci A, Santoro A, et al.Genetic and epigenetic alterations in primary colorectal cancers and relatedlymph node and liver metastases. Cancer 2013;119:266–76.

31. Okayasu I, Ohkusa T, Kajiura K, Kanno J, Sakamoto S. Promotion ofcolorectal neoplasia in experimental murine ulcerative colitis. Gut 1996;39:87–92.

32. Cooper HS,Murthy SN, Shah RS, Sedergran DJ. Clinicopathologic study ofdextran sulfate sodium experimental murine colitis. Lab Invest 1993;69:238–49.

33. Suzuki R, KohnoH, Sugie S, Tanaka T.Dose-dependent promoting effect ofdextran sodium sulfate on mouse colon carcinogenesis initiated withazoxymethane. Histol Histopathol 2005;20:483–92.

34. Johnston RJ, Poholek AC, DiToro D, Yusuf I, Eto D, Barnett B, et al. Bcl6and Blimp-1 are reciprocal and antagonistic regulators of T follicularhelper cell differentiation. Science 2009;325:1006–10.

35. Galon J, Costes A, Sanchez-Cabo F, Kirilovsky A, Mlecnik B, Lagorce-PagesC, et al. Type, density, and location of immune cells within humancolorectal tumors predict clinical outcome. Science 2006;313:1960–4.

36. Sica A, Mantovani A. Macrophage plasticity and polarization: in vivoveritas. J Clin Invest 2012;122:787–95.

37. Salcedo R, Worschech A, Cardone M, Jones Y, Gyulai Z, Dai RM, et al.MyD88-mediated signaling prevents development of adenocarcinomas ofthe colon: role of interleukin 18. J Exp Med 2010;207:1625–36.

38. Rakoff-Nahoum S, Medzhitov R. Regulation of spontaneous intestinaltumorigenesis through the adaptor protein MyD88. Science 2007;317:124–7.

Cancer Immunol Res; 6(5) May 2018 Cancer Immunology Research592

Porta et al.

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

39. Moser AR, Pitot HC, Dove WF. A dominant mutation that predisposes tomultiple intestinal neoplasia in the mouse. Science 1990;247:322–4.

40. Ries CH, CannarileMA,Hoves S, Benz J,Wartha K, Runza V, et al. Targetingtumor-associated macrophages with anti-CSF-1R antibody reveals a strat-egy for cancer therapy. Cancer Cell 2014;25:846–59.

41. Balkwill F, Mantovani A. Inflammation and cancer: back to Virchow?Lancet 2001;357:539–45.

42. Hefetz-Sela S, Stein I, Klieger Y, Porat R, Sade-Feldman M, Zreik F, et al.Acquisition of an immunosuppressive protumorigenic macrophage phe-notype depending on c-Jun phosphorylation. Proc Natl Acad Sci U S A2014;111:17582–7.

43. Bain CC, Scott CL, Uronen-Hansson H, Gudjonsson S, Jansson O, Grip O,et al. Resident and pro-inflammatory macrophages in the colon representalternative context-dependent fates of the same Ly6Chi monocyte pre-cursors. Mucosal Immunol 2013;6:498–510.

44. Jauch D, Martin M, Schiechl G, Kesselring R, Schlitt HJ, Geissler EK, et al.Interleukin 21 controls tumour growth and tumour immunosurveillancein colitis-associated tumorigenesis in mice. Gut 2011;60:1678–86.

45. Stolfi C, Rizzo A, Franze E, Rotondi A, Fantini MC, Sarra M, et al. Involve-ment of interleukin-21 in the regulation of colitis-associated colon cancer.J Exp Med 2011;208:2279–90.

46. De Simone V, Ronchetti G, Franze E, Colantoni A, Ortenzi A, Fantini MC,et al. Interleukin-21 sustains inflammatory signals that contribute tosporadic colon tumorigenesis. Oncotarget 2015;6:9908–23.

47. Ugai S, ShimozatoO,Kawamura K,Wang YQ, Yamaguchi T, SaishoH, et al.Expression of the interleukin-21 gene in murine colon carcinoma cells

generates systemic immunity in the inoculated hosts. Cancer Gene Ther2003;10:187–92.

48. Bindea G, Mlecnik B, Tosolini M, Kirilovsky A, Waldner M, Obenauf AC,et al. Spatiotemporal dynamics of intratumoral immune cells reveal theimmune landscape in human cancer. Immunity 2013;39:782–95.

49. Spolski R, Leonard WJ. Interleukin-21: a double-edged sword with ther-apeutic potential. Nat Rev Drug Discov 2014;13:379–95.

50. Larghi P, Porta C, Riboldi E, Totaro MG, Carraro L, Orabona C, et al. Thep50 subunit of NF-kappaB orchestrates dendritic cell lifespan and activa-tion of adaptive immunity. PLoS One 2012;7:e45279.

51. Grivennikov SI, Wang K, Mucida D, Stewart CA, Schnabl B, Jauch D, et al.Adenoma-linked barrier defects and microbial products drive IL-23/IL-17-mediated tumour growth. Nature 2012;491:254–8.

52. Langrish CL, Chen Y, Blumenschein WM, Mattson J, Basham B, SedgwickJD, et al. IL-23 drives a pathogenic T cell population that induces auto-immune inflammation. J Exp Med 2005;201:233–40.

53. TengMW, AndrewsDM,McLaughlinN, von Scheidt B, Ngiow SF,Moller A,et al. IL-23 suppresses innate immune response independently of IL-17Aduring carcinogenesis and metastasis. Proc Natl Acad Sci U S A 2010;107:8328–33.

54. Southern SL, Collard TJ, Urban BC, Skeen VR, Smartt HJ, Hague A,et al. BAG-1 interacts with the p50-p50 homodimeric NF-kappaBcomplex: implications for colorectal carcinogenesis. Oncogene2012;31:2761–72.

55. Erdman S, Fox JG, Dangler CA, Feldman D, Horwitz BH. Typhlocolitis inNF-kappa B-deficient mice. J Immunol 2001;166:1443–7.

www.aacrjournals.org Cancer Immunol Res; 6(5) May 2018 593

p50 NF-kB Shapes Colorectal Cancer–Promoting Inflammation

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036

2018;6:578-593. Published OnlineFirst March 27, 2018.Cancer Immunol Res   Chiara Porta, Alessandro Ippolito, Francesca Maria Consonni, et al.   Colorectal Cancer Progression

B EnhancesκProtumor Steering of Cancer Inflammation by p50 NF-

  Updated version

  10.1158/2326-6066.CIR-17-0036doi:

Access the most recent version of this article at:

  Material

Supplementary

  http://cancerimmunolres.aacrjournals.org/content/suppl/2018/03/27/2326-6066.CIR-17-0036.DC1

Access the most recent supplemental material at:

   

   

  Cited articles

  http://cancerimmunolres.aacrjournals.org/content/6/5/578.full#ref-list-1

This article cites 55 articles, 15 of which you can access for free at:

  Citing articles

  http://cancerimmunolres.aacrjournals.org/content/6/5/578.full#related-urls

This article has been cited by 3 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

  .pubs@aacr.orgat

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

  Permissions

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

.http://cancerimmunolres.aacrjournals.org/content/6/5/578To request permission to re-use all or part of this article, use this link

on July 4, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst March 27, 2018; DOI: 10.1158/2326-6066.CIR-17-0036