Diverse Roles of STING-Dependent Signaling on the Development of Cancer

SHORT COMMUNICATION Diverse roles of STING-dependent signaling on the development of cancer

J Ahn, H Konno and GN Barber

Stimulator of interferon genes (STING) is a cellular sensor that controls cytosolic DNA-activated innate immune signaling. We have previously demonstrated that STING-deficient mice are resistant to carcinogen-induced skin cancer, similar to myeloid differentiation primary response gene 88 (MyD88) deficient mice, since the production of STING-dependent DNA-damage-induced proinflammatory cytokines, that likely require MyD88 signaling to exert their growth-promoting activity, are prevented. In contrast, MyD88-deficient mice are sensitive to colitis-associated cancer (CAC), since selected cytokines generated following DNA-damage also activate repair pathways, which can help prevent tumor development. Here, we demonstrate that STING signaling facilitates wound repair processes and that analogous to MyD88-deficient mice, STING-deficient mice (SKO) are prone to CAC induced by DNA-damaging agents. SKO mice harboring tumors exhibited low levels of tumor-suppressive interleukin-22 binding protein (IL-22BP) compared to normal mice, a cytokine considered critical for preventing colon-related cancer. Our data indicate that STING constitutes a critical component of the host early response to intestinal damage and is essential for invigorating tissue repair pathways that may help prevent tumorigenesis.

Oncogene (2015) 34, 5302–5308; doi:10.1038/onc.2014.457; published online 2 February 2015

INTRODUCTION However, MyD88-deficient mice are known to be susceptible to The innate immune system provides the first line of defense colitis-associated carcinogenesis (CAC) induced by drugs such as 10 against pathogen infection though can also influence pathways azoxymethane (AOM) and dextran sulfate sodium (DSS). AOM is than can control tumorigenesis.1,2 For example, it is known that the metabolite of 1,2-dimethylhydrazine (DMH) and is converted the cellular adapter MyD88 (myeloid differentiation primary to methylazoxymethanol, which mediates O-methylguanine for- 11 response gene 88) that facilitates Toll-like receptor and mation to trigger DNA damage responses. A single injection of interleukin-1 receptor (IL-1R) signaling pathway in the innate AOM into mice, followed by administration of the inflammatory immune response can regulate tumorigenesis through control of agent DSS via drinking water induces almost 100% colon cancer. nuclear factor-κB (NF-κB) activation, cytokine secretion and In this situation, MyD88 exerts a protective effect in part by inflammation.2 Mice lacking MyD88 are resistant to carcinogen facilitating the production of interleukin-18 (IL-18), in epithelial (7,12-dimethylbenz(a)anthracene (DMBA)) induced skin cancer cells, which downregulates dendritic cell production of the since MyD88-controlled cytokine and growth factor production, interleukin-22 (IL-22) binding protein (IL-22BP).12 IL-22BP sup- which augments skin tumorigenesis is averted.3 Of note is that the presses the function of IL-22 which is produced from innate mechanisms underlining carcinogen-driven cytokine production lymphoid cells in response to cellular/tissue damage and which largely remain unknown. potently stimulates the proliferation of intestinal epithelial cells. Stimulator of interferon genes (STING) is a cellular protein that is We have previously demonstrated that the cellular protein critical for activating the production of various cytokines such as STING facilitates cytosolic DNA-triggered innate immune signaling type I interferon in response to the detection of microbial double- pathways, independent of Toll-like receptor 9 or the DNA sensor stranded DNA (dsDNA) or cyclic dinucleotides (CDNs).4–6 In absent in melanoma 2 (AIM II).13 In humans, STING is a 348 amino addition, self-DNA is also capable of activating STING-dependent acid endoplasmic reticulum-associated molecule predominantly signaling and this pathway has now been shown to play a key role expressed in epithelial cells as well as cells of the hematopoietic in the cause of a variety of autoinflammatory diseases.7,8 Recently, lineage, that has been shown to play a key role in triggering it has been further reported that STING-dependent cytokine innate immune signaling pathways in response to infection by production can be induced by DNA-damaging agents, including viruses such as herpes simplex virus 1, and even bacteria.4 STING DMBA. It was observed that DMBA caused nucleosome release has also been shown to be responsible for triggering vascular and into the cytosol to trigger STING activity.9 STING was found to pulmonary syndrome, self-DNA-induced inflammatory diseases control the expression of cytokines requiring MyD88 signaling for such as Aicardi–Goutieres syndrome and perhaps forms of severe – the extended production of proinflammatory cytokines such as systemic lupus erythematosus.8,14 16 STING may associate tumor necrosis factor α.9 Thus, STING-deficient mice were with dsDNA species directly and is highly activated by CDNs observed to be resistant to DMBA-aggravated skin carcinogenesis, (cyclic di-GMP-AMP[cGAMP-c(G(2′,5′)pA(3′,5′)p)]) generated by similar to MyD88-deficient mice.9 certain bacteria or by cytosolic dsDNA triggering the activation

Department of Cell Biology and Sylvester Comprehensive Cancer Center, University of Miami School of Medicine, Miami, FL, USA. Correspondence: Dr GN Barber, Department of Cell Biology and Sylvester Comprehensive Cancer Center, University of Miami School of Medicine, 511 Papanicolaou Building, 1550 NW 10th Ave, Miami, FL 33136, USA. E-mail: [email protected] Received 30 October 2014; revised 18 November 2014; accepted 26 November 2014; published online 2 February 2015 Role of STING-dependent signaling on cancer J Ahn et al 5303 of a synthase, referred to as cGAS (cyclic GMP-AMP synthase, dependent carcinogen-induced skin cancer, we wished to C6orf150, Mab-21 domain-containing protein).6 address the role of STING in inflammation-aggravated intestinal Given that STING appears to play a pivotal role in controlling a tumorigenesis. Using the AOM/DSS model, we observed that variety of inflammation-driven events and may control MyD88- similar to MyD88, STING-deficient mice (SKO) are sensitive to CAC MEF 5 * Mock WT SKO WT SKO 4 AOM 3 Mock AOM Mock AOM Mock DMH Mock DMH EG546166 Rgs1 2 Ifit3 Ccl3 Usp18 Cxcl2 Rnf2 EG546166 1 LOC667370 Rgs1

Ifit3 Atf3 Cxcl10 fold changes 0 Usp18 Rgs1 WT SKO LOC100048346 Gdf15 9930013L23Rik Ccl4 5 Gm166 EG434760 Mock 9530096D07Rik Aifm2 AOM C530043K16Rik lgtp 4 LOC674960 Sprr2a LOC100038882 Rab10 3 V1rh12 Scyl1 5730455P16Rik Zfp566 2 Chst5 Mcm7 Fn1 4930476L21Rik 9630025C19Rik Sec11c 1

A530082L16Rik Usp18 Ifit3 fold changes S100a16 Ube2t 0 Ndufb6 Cxcl2 WT SKO Cd83 Klhdc4 4732474K08Rik Slfn2 10 * Mock Calm2 Pdhb AU019823 Serhl DMH 9130230L23Rik LOC280205 8 2200002D01Rik Rdh1 Ogfod2 Cyp11b1 6 Rnf213 Frag1 Rsad2 Josd3 4 Igf2bp1 Nkain2 Gm50 1110059G10Rik 2 LOC100039693 1810034K20Rik Peg13 Afg312 Cxcl10 fold changes Olfr67 Coro1b 0 Gm1381 Rnf121 WT SKO Igtp Snx16 Myo18b Cyr61 2700078K21Rik * Gbp3 2 Mock 2900097C17Rik Ngfb Zc3h10 LOC100038882 DMH Eif2c3 Myocd 1.5 Ensa Cdkn1a Ell2 Bet1 Efna5 Asb16 1 Chrna2 Ppm1f Kctd12b 2510040D07Rik Catnb Tapbpl 0.5 2410002F23Rik Rps6kb1 Ifit3 fold changes 0 0.35 1.04 0.35 1.04 0.69 0.69 0.00 0.35 0.00 0.35 0.69 0.69 1.04 1.04 WT SKO

FHC FHC 10 3 siNS * * siSting * 1.2 8 1.0 2 6 0.8 0.6 4 1 0.4 2 0.2 Sting fold changes Cxcl10 fold changes 0 Cxcl10 fold changes 0 0.0 Mock AOM DMH Mock AOM DMH siNS siSting

WT colon SKO colon Human normal colon Colon

Figure 1. For caption see next page.

© 2015 Macmillan Publishers Limited Oncogene (2015) 5302 – 5308 Role of STING-dependent signaling on cancer J Ahn et al 5304 indicating a protective role for STING in tumorigenesis. Our data exhibited tumor formation compared to 7/7 SKO within the same indicate that STING may be a key sensor that promotes the time period (Figures 2b and c and Supplementary Figure S3a). elimination of damaged intestinal epithelial cells. Hematoxylin and eosin analysis confirmed that AOM/DSS-treated SKO mice exhibited significant inflammatory cell infiltration and development of adenocarcinoma in the colon, compared to RESULTS AND DISCUSSION similarly treated WT mice (Figures 2d and e and Supplementary Activation of STING-dependent genes by AOM Figure S3b). However, Microarray analysis indicated that tumors Given that chronic inflammation is known to aggravate colon from WT mice exhibited higher levels of selected gene expression, cancer and that STING has been shown to influence inflammatory such as Cxcl13 and Ccr6, compared to tumors retrieved from SKO responses, especially those invoked by cytosolic self or pathogen- mice, perhaps since loss of STING suppressed immunomodulatory related DNA, we examined the role of STING in the control of transcriptional events (Figures 2f–h). Collectively, we postulate inflammatory CAC.1,2,15 Toward these objectives, we utilized the that STING may recognize damaged DNA and activate the AOM/DSS model and analyzed the effects of AOM and precursor production of cytokines that conceivably could promote tissue DMH on STING signaling. Principally, wild-type (WT) or STING- repair or stimulate the immune system to eradicate such cells. deficient (SKO) murine embryonic fibroblasts (MEFs) were treated Thus, loss of STING may enable damaged cells to escape immune in vitro with DMH or metabolite AOM for 8 h, and microarray surveillance processes and progress more readily into tumors. analysis performed to analyze the consequences to gene expression. This study indicated that AOM activated mRNA Suppression of IL-22BP expression in STING-deficient mice production of a wide array of innate immune-related genes in Although, we demonstrated that loss of STING facilitated colon cancer WT cells, including interferon-induced protein with tetratricopep- development, the tumor-suppressive mechanisms associated with tide repeats 3 (IFIT3) and chemokine (C-X-C motif) ligand 2 STING activity remain to be clarified. It remained plausible that STING (CXCL2) (Figure 1a and Supplementary Figure 1). However, there could exert direct tumor-suppressive, growth inhibitory or pro- was a marked decrease in the production of the same genes in apoptotic properties similar to tumor suppressor p53. Further, AOM SKO MEFs indicating that AOM was indeed capable of activating treatment has been known to induce frequent Ras mutations, which in the STING pathway (Figure 1a, left panel). A similar effect was the context of loss of STING, could facilitate cellular transformation.17 observed following the treatment of cells with DMH (Figure 1a, Expression of oncogenic Ras in an environment where p53 function is right panel). STING-dependent gene expression was confirmed – lost renders normal murine cells, the ability to form foci in soft agar following RT PCR analysis of mRNA representing Cxcl10 and IFIT3 and to become tumorigenic in vivo.18 To evaluate the anti-oncogenic (Figure 1b). We similarly treated normal human colon epithelial role of STING further, we transfected WT, SKO or p53-deficient MEFs, cells (FHCs) with AOM and observed a comparable induction of as positive controls, with Myc or activated Ras and monitored cellular innate immune genes, controlled by STING, including Cxcl10 transformation. MEFs lacking p53 were found to be readily (Figure 1c). The production of Cxcl10 by AOM was similarly transformed by the introduction of Myc or activated Ras reduced in FHCs treated with RNAi to STING (Figure 1d). To ’ (Supplementary Figure S4). However, MEFs lacking STING did not determine the consequences of DMH/AOM treatment on STING s appear appreciably sensitive to transformation following overexpres- ability to activate these key transcription factors, we carried out fl sion of Myc or activated Ras. Thus, the absence of STING does not immuno uorescence analysis on MEFs and FHCs treated with appear to exert an oncogenic stimulus, at least in vitro or to cooperate these drugs. This study indicated that DMH/AOM could instigate κ with Myc or Ras in the cellular transformation process. the translocation of interferon regulatory factor 3/NF- B in treated However, it has been demonstrated that mice lacking certain cells (Supplementary Figure 2). Thus, the DNA-damaging agent cytokines such as IL-18, IL-22 or the innate immune adapter DMH/AOM can invoke STING-dependent signaling. MyD88 are similarly susceptible to AOM/DSS-induced CAC.1,2,10,12 In this situation, MyD88 exerts a protective effect by facilitating Loss of STING renders mice susceptible to CAC the production of IL-18 through the IL-18R, which is required to – Our data indicate that DMH/AOM can activate STING in vitro.To inhibit IL-22BP.3,12,19 22 IL-22BP is necessary to suppress IL-22 examine the consequences of this in vivo, we treated mice once function, which can promote the proliferation of intestinal with AOM and subsequently orally with four treatments of DSS. epithelial cells following damage by carcinogens or inflammatory Prior to this, we analyzed STING expression in the intestine by agents.12 Mice lacking IL-18 or IL-22BP are highly susceptible to immunohistochemistry analysis. This study showed that STING CAC, similar to STING-deficient mice.12 To examine whether AOM/ was expressed in lamina propria cells as well as in endothelial and DSS treatment could affect IL-18 and IL-22BP expression in SKO epithelial cells of the gastrointestinal tract (Figure 1e). After mice, we analyzed IL-18 and IL-22BP levels in untreated normal 13 weeks the mice were analyzed for tumor development in the mice (control) or tumor tissue taken from the AOM/DSS-treated colon. Surprisingly, we observed that SKO mice developed colonic WT or SKO mice. We observed that IL-18 expression was slightly tumors at a much higher frequency compared to WT mice upregulated in tumor tissue retrieved from WT mice treated with (Figures 2a–c and Supplementary Figure S3). Indeed, 4/7 WT mice AOM/DSS (Figure 2i). We observed that IL-22BP expression levels

Figure 1. Activation of STING-dependent genes by AOM. (a) Gene array analysis of wild-type (WT) and STING-deficient (SKO) mouse embryonic fibroblasts (MEFs) treated with AOM at 0.14 mM for 8 h (left) and DMH at 1 mM for 8 h (right). MEFs were obtained from embryos from WT and SKO mice at embryonic 15 days by a standard procedure as described (Ishikawa 2008). Gene array analysis was examined by Illumina Sentrix BeadChip array (Mouse WG6 version 2). Highest variable genes are shown. Rows represent individual genes and columns represent individual samples. Pseudo-colors indicate transcript levels below (green), equal to (black) or above (red) the mean. Scale represents the intensity of gene expression (log2 scale ranges between − 2.4 and 2.4). (b) Quantitative real-time PCR analysis (qPCR) of Cxcl10 and Ifit3 in MEFs treated with AOM and DMH same as a. Total RNA was reverse transcribed using moloney murine leukemia virus (M-MLV) reverse transcriptase. qPCR was performed using TaqMan gene expression assay (Cxcl10: Mm00445235, Ifit3: Mm0170846). (c) qPCR analysis of Cxcl10 in human epithelial cell (FHC) treated with AOM and DMH at 1 mM for 24 h. (d) FHCs were transfected with STING or control siRNA for 72 h followed by AOM and DMH treatment same as c, and were then subjected to Cxcl10 mRNA expression (left). STING expression level after siRNA treatment was determined by qPCR (right). Data are representative of at least two independent experiments. Error bars indicate s.d. *Po0.05, Student’s t-test. (e) Immunohistochemistry staining of the colon tissue from WT and SKO mice (left) and human (right). All images were shown at original magnification, x200.

Oncogene (2015) 5302 – 5308 © 2015 Macmillan Publishers Limited Role of STING-dependent signaling on cancer J Ahn et al 5305 WT Cont WT AOM/DSS DSS DSS DSS DSS 160 SKO Cont SKO AOM/DSS 140 W1 W4 W7 W10 W13 120

AOM Tumor 100 injection evaluation 80 Body weight (%) 60 0 4 8 23 28 36 45 49 62 72 77 106 121 WT SKO 6 Cont DW DSS/AOM DW DSS/AOM DSS/AOM 4

Micro- polyps 2 Number of endoscopy 0

4 Control 3 DSS/AOM HE 2 1

Inflammation score 0 WT SKO WT SKO SKO- Symbol WT-DSS/AOM SKO-DW DSS/AOM 1810030J14Rik 148.69 31.43 5.58 1810030J14Rik 58.01 12.10 3.17 Control DSS/AOM Control DSS/AOM Atp12a 28.63 9.28 18.66 LOC673501 28.13 5.32 5.11 1810030J14Rik Faim3 24.57 1.19 2.42 1810030J14Rik LOC100047788 17.66 -1.19 1.09 Atp12a Cd79b 16.98 1.09 1.56 LOC673501 Fcrla 16.82 1.04 1.64 Faim3 LOC100047788 Ighg 16.59 -1.30 1.31 Cd79b Insl5 15.42 3.84 1.79 Fcrla Cxcl13 14.57 1.14 1.40 lghg AI324046 14.06 1.06 1.00 lnsl5 H2-Ab1 13.72 1.01 5.73 Cxcl13 Dlk1 13.20 1.92 1.08 Al324046 LOC100046120 13.03 1.77 10.48 H2-Ab1 LOC100047815 12.86 1.36 1.59 Dlk1 Slpi 12.44 1.64 18.37 LOC100046120 Cd52 12.37 -1.06 3.56 LOC100047815 Cd52 11.53 -1.41 2.68 Slpi Cd74 11.17 -1.19 6.07 Cd52 AI854703 11.16 5.50 4.12 Cd52 Cd79b 11.13 -1.00 1.42 Cd74 Al854703 Ela1 10.91 4.65 4.55 Cd79b D930028F11Rik 10.88 -1.44 1.16 Ela1 Klhl6 10.86 -1.05 2.16 D930028F11Rik Klhl6 10.75 -1.04 1.71 Klhl6 Cd74 10.44 -1.16 5.37 Klhl6 H2-Oa 10.41 1.11 1.85 Cd74 Lyz2 10.16 -1.25 6.48 H2-Oa Ttr 10.15 3.68 3.02 Lyz2 Fcrla 10.11 -1.06 1.35 Ttr B3gnt7 9.54 3.77 5.13 Fcrla Adh1 9.51 2.01 1.41 B3gnt7 Tnfrsf13c 9.50 1.05 1.39 Adh1 B3gnt5 9.48 2.88 10.19 Tnfrsf13c Ctse 9.42 3.35 10.23 B3gnt5 Ctse Coro1a 9.21 -1.13 2.84 Coro1a AI324046 9.19 -1.09 -1.08 Al324046 Igh-6 9.19 -1.22 1.51 lgh-6 Fcrla 9.17 1.05 1.42 Fcrla Ccr6 9.10 1.14 1.79 Ccr6 Dok3 9.06 1.19 2.40 Dok3 H2-Ob 8.95 1.11 1.23 H2-Ob Mfge8 8.82 1.16 4.50 Mfge8 Igh-6 8.42 -1.29 1.38 lgh-6 Arhgdib 8.39 -1.22 2.88 Arhgdib Coro1a 8.08 -1.35 2.70 Coro1a H2-DMb2 7.93 -1.12 2.99 H2-DMb2 Hvcn1 7.91 1.12 1.95 Hvcn1 Nt5e 7.89 3.48 12.67 2.00 1.00 3.00 3.00 0.00 1.00 Nt5e 2.00 Mfge8 7.86 1.11 4.28 Mfae8

18 * Control 100 5 8 * Control * Control Control * 16 AOM/DSS AOM/DSS AOM/DSS 7 AOM/DSS 80 4 14 6 12 * 5 10 60 3 4 8 40 2 6 3 4 20 1 2

2 IL18 Fold changes 1 CCR6 Fold changes Cxcl13 Fold changes 0 0 0 IL22bp Fold changes 0 WT SKO WT SKO WT SKO WT SKO Figure 2. For caption see next page.

© 2015 Macmillan Publishers Limited Oncogene (2015) 5302 – 5308 Role of STING-dependent signaling on cancer J Ahn et al 5306 were also elevated in the tumors of AOM/DSS-treated WT mice. of the IL-18/IL-22BP axis. First, we confirmed the influence of In contrast, we noted the significantly reduced expression of IL-18 STING on IL-18 expression, since we additionally noted that the and IL-22BP in the tumors of SKO mice, when compared to WT promoter of this cytokine is known to harbor numerous sites mice (Figure 2i). The expression of IL-22 was seen to remain recognized by innate immune gene-activating transcription relatively unaffected (data not shown). It was surprising to note factors such as signal transducer and activator of transcription 1, that IL-22BP levels were elevated in the presence of IL-18. NF-κB, IRF1 and IRF7 (Supplementary Figure S5). Our analysis However, it has been reported that the downregulation of IL-22BP indicated that IL-18, which is expressed in a wide variety of cell can occur even in the absence of IL-18, indicating that alternate types, is a STING inducible gene, as determined following STING-induced cytokines may clearly contribute to the regulation treatment of MEF cells and bone marrow-derived dendritic cells of IL-22BP.12 (BMDCs) with dsDNA or STING-activating CDNs (cGAMP; Figure 3a, In addition, we noted from our microarray analyses that IL-18 middle and right panel). A similar study indicated that DMH/AOM levels were reduced in SKO MEFs treated with STING-activating could also trigger the production of IL-18 in dendritic cells, in a dsDNA (dsDNA90 base pairs; Figure 3a, left panel). We therefore STING-dependent manner (Figure 3b). To further this study, we investigated the involvement of STING on the possible regulation examined IL-22BP expression in BMDCs stimulated with IL-18

MEF BMDCs Gene array RT PCR RT PCR 4 8 * 7 WT * WT WT * SKO 7 SKO 6 SKO 3 6 * * 5 5 4 2 4 3 3 1 2 2 lL18 fold changes lL18 fold lL18 Fold changes lL18 Fold changes lL18 Fold 1 1 0 0 0

lFN Cont Mock Mock cGAMP cGAMP dsDNA90 dsDNA90 dsDNA90

BMDCs BMDCs

6 Mock 16 WT lL18 * 14 SKO 5 12 4 10 3 8 * 6 2 4

lL18 Fold changes lL18 Fold 1 2 changes lL22bp Fold 0 0 No AOM DMH WT SKO Figure 3. IL-18 controls IL-22BP in STING-dependent manner. (a) Fold changes from gene array analysis of IL-18 in WT and SKO MEFs administered with 4 ug/ml of dsDNA90 and IFNβ for 8 h (left). qPCR analysis of IL18 in WT and SKO MEFs, and bone marrow-derived dendritic cells (BMDCs) transfected with 4 μg/ml of dsDNA90 and cyclic-di-GMP-AMP (cGAMP) for 8 h (middle). (b) qPCR analysis of IL-18 in WT and SKO BMDCs with 1 mM of AOM and 1 mM of DMH for 8 h. (c) qPCR analysis of IL-22BP in WT and SKO BMDCs treated with 10 ng/ml of recombinant mouse IL-18 protein (MBL B002-5) for 24 h. Total RNA was reverse transcribed using M-MLV reverse transcriptase. qPCR was performed using TaqMan gene expression assay (IL-18: Mm00434225, IL-22BP: Mm01192969). Data are the mean of at least ﬁve mice. Error bars indicate s.d. *Po0.05, Student’s t-test.

Figure 2. Loss of STING renders mice susceptible to CAC. (a) Schematic representation and body weight of AOM/DSS-induced colitis model. WT (n = 7) and SKO (n = 7) mice (B6;129 mix) were intravenously injected with AOM at a dose 10 mg/kg. After 1 day, mice were fed 5% DSS in drinking water for 7 days. This cycle was repeated four times. Normal drinking water was used for control group. At 91 days, micro-endoscopic procedure was performed in a blinded manner for counting number of polyps. Mice were killed at day 121 and the colon was resected, flushed with phosphate-buffered saline, fixed in formalin for histology or froze for RNA expression analysis. Representative photographs of macro-endoscopic colon tumors (b) and number of polyps (c). Hematoxylin and eosin staining (d)ofWT(n = 7) and SKO (n = 7) mice either AOM/DSS treated or normal water treated and inflammation score (0: normal to 3: most severe; e). (f) Gene array analysis of normal tissue (control) and tumor tissue (DSS/AOM) from WT and SKO mice treated same as a. Rows represent individual genes and columns represent individual samples. Pseudo-colors indicate transcript levels below (green), equal to (black) or above (red) the mean. Scale represents the intensity of gene expression (log2 scale ranges between − 2.4 and 2.4). (g) Highest variable gene lists are shown. (h) qPCR of Cxcl13 and Ccr6 in normal tissue (control) and tumor tissue (DSS/AOM) from WT and SKO mice treated same as a.(i) qPCR analysis of IL-18 and IL-22BP (IL-18: Mm00434225, IL-22BP: Mm01192969) from normal tissue (control) and tumor tissue (DSS/AOM) from WT and SKO mice. Data are the mean of at least five mice. Error bars indicate s.d. *Po0.05, Student’s t-test.

Oncogene (2015) 5302 – 5308 © 2015 Macmillan Publishers Limited Role of STING-dependent signaling on cancer J Ahn et al 5307 recombinant protein and observed the upregulation of IL-22BP Our data here indicate that IL-18 and a variety of other in WT BMDCs compared to SKO BMDCs (Figure 3c and cytokines and chemokines are inducible by dsDNA or CDN’s, or by Supplementary Figure S6). These data are similar to Figure 2i AOM/DMH in a STING-dependent manner. Similar to the situation showing that IL-18 and IL-22BP expression levels were also lower with IL-22, we propose that intestinal damage triggers STING in the tumors of AOM/DSS-treated SKO mice compared to the activity (as a consequence of DNA damage or even from microbial tumors of WT mice (Figure 2i). These data suggest that STING may ligands such as secreted CDNs or genomic DNA). This results in the regulate IL-22BP expression by IL-18. downregulation of IL-22BP, which would enable IL-22 to promote Taken together, it is conceivable that DNA damage or the tissue repair. However, similar to the situation with IL-22, long- sensing of microbial ligands that may invade damaged colon term loss of STING may delay wound repair, facilitate microbial tissue can trigger STING activity leading to the activation of invasion and chronic inflammation, which would actually aggra- inflammatory wound repair initiating cytokines as well as the vate tumorigenesis.1,2,12 In addition, loss of STING may enable indirect suppression of growth inhibitory IL-22BP. Our data thus damaged cells to escape antitumor immunosurveillance.2 It was indicate that similar to mice lacking MyD88, IL-18 or IL-22BP, noted that IL-18 expression was not totally ablated in tumors from STING-deficient mice are also prone to CAC induced by AOM/DSS. SKO mice, presumably since the expression of this cytokine could Although, we demonstrate here a protective role for STING in be induced by other pathways.12 Despite this, IL-22BP levels the prevention of CAC induced by AOM/DSS carcinogenic remained low in SKO mice evidently demonstrating the impor- treatment, these findings are in contrast to our previous data tance of STING in IL-22BP regulation. Collectively, our data indicate indicating that STING-deficient mice are resistant to DMBA- that STING plays a key role in controlling intestinal tissue damage induced skin cancer.9 In this scenario, DNA-leaked from the and CAC in part through regulating IL-22BP’s suppression of nucleus of carcinogen damaged dermal-associated cells activates IL-22. Our data also provide a mechanistic explanation for STING-dependent cytokine production, intrinsically, to attract carcinogenic-induced immunomodulatory effects that can infiltrating phagocytes. Such phagocytes engulf dying cells to influence tumorigenesis. activate extrinsic STING-dependent cytokine production, which further drives inflammatory processes that promote tumor development.9 This situation is reminiscent of similarly treated CONFLICT OF INTEREST MyD88-deficient mice, which are also resistant to DMBA- The authors declare no conflict of interest. aggravated skin cancer. Our data indicate that STING triggers the production of cytokines that bind to receptors requiring ACKNOWLEDGEMENTS MyD88-dependent signaling for additional cytokine production, such as tumor necrosis factor α.3 Our data here conversely indicate We thank Dr Masayuki Fukata for micro-endoscopy analysis; Ms Delia Gutman and Ms that STING-driven cytokine production can protect against the Auristela Rivera for mice breeding; Dr Tianli Xia for helping with confocal analysis; Dr development of other types of tumors, such as carcinogen- Biju Issac of the Sylvester Comprehensive Cancer Center Bioinformatics Core Facility influenced colon cancer.23 This event may occur in part through for Gene expression array analysis; Dr Phillp Ruiz and Ms Dayami Hernandez for helping with immunohistochemistry. STING's ability to control the production of IL-22BP, although STING clearly controls the production of numerous other regulatory proteins and cytokines. Following tissue damage, for REFERENCES example by DSS, it has been reported that IL-22 is induced and 1 Goldszmid RS, Trinchieri G. The price of immunity. Nat Immunol 2012; 13: manifests protective, wound-healing effects, including the promo- 932–938. 1,2,10,12 tion of tissue regeneration. However, if left uncontrolled, 2 Nowarski R, Gagliani N, Huber S, Flavell RA. Innate immune cells in inflammation 12 IL-22 can also endorse tumor development. IL-22 is therefore and cancer. Cancer Immunol Res 2013; 1:77–84. tightly regulated by secreted IL-22BP, which is expressed by 3 Cataisson C, Salcedo R, Hakim S, Moffitt BA, Wright L, Yi M et al. IL-1R-MyD88 CD11c+ dendritic cells.12 The importance of IL-22BP in controlling signaling in keratinocyte transformation and carcinogenesis. J Exp Med 2012; 209: IL-22 has been emphasized through observing that IL-22BP- 1689–1702. fi 4 Ishikawa H, Ma Z, Barber GN. STING regulates intracellular DNA-mediated, type I de cient mice are also susceptible to AOM/DSS-induced CAC, 461 – similar to STING-deficient mice.12 Nevertheless, IL-22 may have interferon-dependent innate immunity. Nature 2009; : 788 792. 5 Burdette DL, Monroe KM, Sotelo-Troha K, Iwig JS, Eckert B, Hyodo M et al. STING is dual functions since mice lacking IL-22 have also been reported to a direct innate immune sensor of cyclic di-GMP. Nature 2011; 478:515–518. exhibit enhanced inflammatory responses when treated repeat- 6 Cai X, Chiu YH, Chen ZJ. The cGAS-cGAMP-STING pathway of cytosolic DNA edly with DSS, plausibly because complete loss of IL-22 may cause sensing and signaling. Mol Cell 2014; 54: 289–296. a delay in intestinal repair, which in turn may actually aggravate 7 Ahn J, Ruiz P, Barber GN. Intrinsic self-DNA triggers inflammatory disease inflammatory processes.1,2,10,12 The production of IL-22BP can be dependent on STING. J Immunol 2014; 193: 4634–4642. suppressed by IL-18, which is known to be induced early after 8 Gall A, Treuting P, Elkon KB, Loo YM, Gale M Jr, Barber GN et al. Autoimmunity DSS-induced intestinal damage. Accordingly, IL-18-deficient mice initiates in nonhematopoietic cells and progresses via lymphocytes in an 36 – are also susceptible to colon cancer, presumably through chronic interferon-dependent autoimmune disease. Immunity 2012; :120 131. 9 Ahn J, Xia T, Konno H, Konno K, Ruiz P, Barber GN. Inflammation-driven carci- suppression of IL-22 activity, by unregulated IL-22BP, which may 5 fi 12 nogenesis is mediated through STING. Nat Commun 2014; : 5166. mimic the situation observed with IL-22-de cient mice. Never- 10 Salcedo R, Cataisson C, Hasan U, Yuspa SH, Trinchieri G. MyD88 and its divergent theless, the control of IL-22BP remains to be fully clarified since Toll in carcinogenesis. Trends Immunol 2013; 34: 379–389. downregulation of IL-22BP has also been reported to occur in the 11 De Robertis M, Massi E, Poeta ML, Carotti S, Morini S, Cecchetelli L et al. The AOM/ 12 absence of IL-18. In addition, it is known that loss of the Toll-like DSS murine model for the study of colon carcinogenesis: from pathways to receptor and IL-1R/IL-18R adapter MyD88 also renders mice diagnosis and therapy studies. J Carcinog 2011; 10:9. sensitive to CAC, in part due to loss of IL-18R signaling.20,21 Finally, 12 Huber S, Gagliani N, Zenewicz LA, Huber FJ, Bosurgi L, Hu B et al. IL-22BP is susceptibility to AOM/DSS-induced CAC has been shown to be regulated by the inflammasome and modulates tumorigenesis in the intestine. 491 – enhanced in mice lacking caspase-1, the adapter PYCARD Nature 2012; :259 263. (apoptosis-associated speck-like protein containing a caspase 13 Ishikawa H, Barber GN. STING is an endoplasmic reticulum adaptor that facilitates innate immune signalling. Nature 2008; 455: 674–678. activation and recruitment domain; ASC) or nucleotide-binding 14 Liu Y, Jesus AA, Marrero B, Yang D, Ramsey SE, Montealegre Sanchez GA et al. domain, leucine rich repeat and pyrin domain-containing proteins Activated STING in a vascular and pulmonary syndrome. N Engl J Med 2014; 371: 3 and 6 (NLRP3/6), presumably since Pro-IL-18 produced by 507–518. epithelial cells or dendritic cells requires cleavage prior to 15 Ahn J, Gutman D, Saijo S, Barber GN. STING manifests self DNA-dependent – secretion into an active form.24 26 inflammatory disease. Proc Natl Acad Sci USA 2012; 109: 19386–19391.

© 2015 Macmillan Publishers Limited Oncogene (2015) 5302 – 5308 Role of STING-dependent signaling on cancer J Ahn et al 5308 16 Namjou B, Kothari PH, Kelly JA, Glenn SB, Ojwang JO, Adler A et al. Evaluation of 21 Zenewicz LA, Yancopoulos GD, Valenzuela DM, Murphy AJ, Stevens S, Flavell RA. the TREX1 gene in a large multi-ancestral lupus cohort. Genes Immun 2011; 12: Innate and adaptive interleukin-22 protects mice from inflammatory bowel 270–279. disease. Immunity 2008; 29: 947–957. 17 Jacoby RF, Llor X, Teng BB, Davidson NO, Brasitus TA. Mutations in the K-ras 22 Rakoff-Nahoum S, Medzhitov R. Regulation of spontaneous intestinal tumor- oncogene induced by 1,2-dimethylhydrazine in preneoplastic and neoplastic rat igenesis through the adaptor protein MyD88. Science 2007; 317:124–127. colonic mucosa. J Clin Invest 1991; 87:624–630. 23 Zhu Q, Man SM, Gurung P, Liu Z, Vogel P, Lamkanfi M et al. Cutting edge: 18 Serrano M, Lin AW, McCurrach ME, Beach D, Lowe SW. Oncogenic ras provokes STING mediates protection against colorectal tumorigenesis by governing the premature cell senescence associated with accumulation of p53 and p16INK4a. magnitude of intestinal inflammation. J Immunol 2014; 193: 4779–4782. Cell 1997; 88:593–602. 24 Fukata M, Vamadevan AS, Abreu MT. Toll-like receptors (TLRs) and Nod-like 19 Swann JB, Vesely MD, Silva A, Sharkey J, Akira S, Schreiber RD et al. receptors (NLRs) in inflammatory disorders. Semin Immunol 2009; 21:242–253. Demonstration of inflammation-induced cancer and cancer immuno- 25 Allen IC, TeKippe EM, Woodford RM, Uronis JM, Holl EK, Rogers AB et al. editing during primary tumorigenesis. Proc Natl Acad Sci USA 2008; 105: The NLRP3 inflammasome functions as a negative regulator of tumorigenesis 652–656. during colitis-associated cancer. J Exp Med 2010; 207: 1045–1056. 20 Salcedo R, Worschech A, Cardone M, Jones Y, Gyulai Z, Dai RM et al. 26 Elinav E, Strowig T, Kau AL, Henao-Mejia J, Thaiss CA, Booth CJ et al. NLRP6 MyD88-mediated signaling prevents development of adenocarcinomas of the inflammasome regulates colonic microbial ecology and risk for colitis. Cell 2011; colon: role of interleukin 18. J Exp Med 2010; 207: 1625–1636. 145:745–757.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)