IDO1 and Kynurenine Pathway Metabolites Activate PI3K-Akt Signaling in the Neoplastic Colon Epithelium to Promote Cancer Cell Proliferation and Inhibit Apoptosis

Several types of cancer pathologically exploit tryptophan metabolism along the kynurenine pathway (KP) to promote growth and escape immune surveillance (1–3). Indoleamine 2, 3 dioxygenase 1 (IDO1) is the most widely studied of tryptophan-metabolizing enzymes that catalyze the initial step of the KP. Tumor overexpression of IDO1 increases local kynurenine (Kyn) concentrations and depletes tryptophan levels. These changes promote an immune-tolerant tumor microenvironment by several mechanisms including suppressing the tumor-reactive effector T-cell responses and NK-cell responses, promoting T-regulatory cell differentiation as well as the expansion and activation of myeloid-derived suppressor cells (2, 4, 5). In addition, a nonenzymatic, protolerance function is also attributed to IDO1 via its interactions with TGFβ (6).

Human and animal studies illustrate the importance of IDO1 in cancer. In preclinical models, IDO1 expression promotes greater tumor burden of several cancers including those of the colon, lung, skin, pancreas, and breast (7–14). In humans, increased IDO1 expression is associated with poor clinical prognosis across several solid tumor types (7, 15, 16). On the basis of these findings, IDO1 inhibition is currently under evaluation in clinical trials for numerous cancer types (17, 18). The IDO1 functional ortholog, tryptophan dioxygenase (TDO), and evolutionary paralog, IDO2, also metabolize tryptophan and contribute to neoplastic pathogenesis in some models; however, the potential for therapeutic targeting remain less well developed than for IDO1.

IDO1 overexpression is a common feature of human colorectal cancer, the second leading cause of cancer-related death in the United States. Pathology studies show that IDO1 expression localizes to colorectal cancer–infiltrating myeloid-derived cells as well as in the neoplastic colon epithelium (3, 19, 20). Patients with colorectal cancer also exhibit reduced serum tryptophan levels and increased KP metabolites, indicating increased IDO1 activity (21–23). Furthermore, high epithelial IDO1 expression at the tumor invasion front is an independent adverse prognostic factor for overall survival and metachronous colorectal cancer metastases, whereas high density of IDO1-expressing cells in the tumor-draining lymph nodes was associated with a reduced 5-year survival rates in patients with colon cancer (16, 20, 24). These findings, which are reviewed more thoroughly elsewhere (25, 26), highlight the relevance of IDO1 as a therapeutic target in human colorectal cancer.

We recently examined the role of IDO1 in a model of colitis-associated cancer and in the cell lines derived from patients with sporadic colorectal cancer. This study demonstrated that IDO1 is highly expressed in the neoplastic colon epithelium and that promoted tumor growth (14). Germline genetic deletion of IDO1 and administration of the first-generation IDO1 inhibitor (1-methyl tryptophan) decreased tumorigenesis. Two novel mechanistic observations arose from these studies. First, IDO1 blockade reduced tumorigenesis even in the mice lacking mature adaptive immunity. Second, IDO1 activity promoted nuclear translocation of epithelial cell β-catenin, a pivotal transcriptional regulator in colorectal cancer. Together these findings provided initial evidence that IDO1 expression is a pathogenic driver of colorectal cancer progression by a mechanism involving the neoplastic epithelium and one that is complementary to its ability to promote immune tolerance through T-cell suppression.

In this study, we sought to delineate the role of epithelial IDO1 in colon tumorigenesis and to define the regulators and mechanism of its immune-independent protumorigenic effect. To address this, we developed an epithelial-specific IDO1 knockout mouse and examined the human colorectal cancer samples, established cell lines, and human-derived colonoids and tumoroids. The data presented herein reveal that epithelial cell IDO1 is key to colon tumorigenesis and that KP metabolites rapidly activate PI3K-Akt signaling in the neoplastic epithelium to promote cellular proliferation and resistance to apoptosis. These findings are highly relevant to IDO1 inhibitors as they move through clinical trials and to the important unmet therapeutic need in advanced colorectal cancer.

All mice used in this study were on C57BL/6J background. Heterozygous Ido1 “knockout-first” tm1A allele mice; Ido1^{tm1a(EUCOMM)Wtsi} were initially acquired as heterozygotes via Material Transfer Agreement from the Wellcome Trust Sanger Institute [Ido1 (MDCT; EPD0198_1_F02)]. A schematic of the targeting vector used to create the Ido1 “knockout-first” allele (tm1a) is shown in Fig. 1A. As the figure illustrates, subsequent breeding with FLPeR mice and Cre mice with under the control of villin promoter generated mice with intestinal epithelium–specific Ido1 knockout (IDO1-iKO). A detailed protocol is given in the Supplementary Materials and Methods. Cre-littermates were used as wild-type (WT) controls.

Age (6–12 weeks)-matched littermates were used in all experiments with approximately equal representation of male and female mice. Animals were housed in specific pathogen-free barrier facility and all the protocols were performed under the regulations of and approval of Washington University's (St Louis, MO) Institutional Animal Care and Use Committee. Induction of colon carcinogenesis by azoxymethane (AOM) followed by cycles of dextran sodium sulfate (DSS), preparation of tissue section for histology and assessment of disease activity index were done as per earlier published protocols (14, 27, 28). Detailed mouse phenotyping data are shown in Fig. 1A. Additional phenotyping data and detailed procedures data are provided in Supplementary Methods and Methods and Supplementary Fig. S1A–S1D.

The collection and use of human tissue for establishing primary epithelial cell culture or organoid culture was approved by the Washington University human research protection office (Institutional Review Board) and collected by the Washington University Digestive Diseases Research Core Center Biobank Core. Informed consent was obtained from all patients. Organoid cultures were established from human and mouse tissues and maintained in Matrigel (catalog no. 354234; Corning Life Sciences) as described previously by our group and others (29–31). Human tumoroids were derived from three colorectal cancer specimens from patients with known APC mutations and familial adenomatous polyposis syndrome. Mouse normal colonoids were established from WT and iKO mice, whereas tumoroids were established from colon tumors of WT mice. Differentiation was induced in normal organoids by maintaining them in 1:10 diluted L-WRN conditioned media with Advanced DMEM/F12 supplemented with 10 μmol/L Y-27632 (ROCK inhibitor; Tocris Bioscience, R&D Systems) before using them for experiments (29). Tumor organoids were maintained in Advanced DMEM/F12 supplemented with 10 μmol/L Y-27632 (Tocris Bioscience) and 10 μmol/L SB 431542 (TGFBR1 inhibitor; Tocris Bioscience, R&D Systems). For experiments comparing human and mouse normal organoids with tumor organoids, tumor organoids were maintained in 1:10 diluted L-WRN conditioned media to match the growth conditions of normal organoids.

Human colorectal cancer cell lines HT-29 (ATCC HTB-38; MSS; KRAS WT), DLD1 (ATCC CCL-221; MSI; KRAS mutant), and HCT116 (ATCC CCL-247; MSI; KRAS and B-Catenin mutant) were purchased from ATCC at study initiation. Cells were grown in DMEM (Gibco, Thermo Fisher Scientific) with high glucose supplemented with 10% FBS (Gibco) and 1% penicillin/streptomycin (Corning Life Sciences). Data presented use cell lines from passage 3 and 30 at approximately 70%–80% confluence. Cell lines were checked for Mycoplasma contamination in September 2016 at the Tissue Culture Support Center in Washington University School of Medicine (St Louis, MO). Kyn, 3-hydroxyanthranillic acid (3HAA), quinolinic acid (QA), picolinic acid (PA), kynurenic acid (KA), xanthurenic acid (XA), and anthranilic acid (AA) were purchased from Millipore Sigma. The 3-hydroxykynurenine (3HK) was acquired from Santa Cruz Biotechnology. Each were solubilized according to manufacturer's recommendation. Recombinant Human DKK-1 (catalog no. GF170) was purchased from Millipore. Total protein was extracted from colorectal cancer cell lines using Cell Lysis Buffer (Cell Signaling Technology) as per manufacturer's recommendation. Total protein was extracted from organoids using RIPA Buffer (Sigma) as per published protocols (29). Cytoplasmic and nuclear proteins were extracted from HT-29 cells using NE-PER (Thermo Fisher Scientific). Total protein was estimated using BCA Protein Assay Kit (Thermo Fisher Scientific) before subjecting them to Western blot analyses. All phosphorylation-specific and corresponding total antibodies used in this study were purchased from Cell Signaling Technology, unless specified. Supplementary Table S1A lists all the antibodies used in this study.

Colorectal cancer cells at approximately 75% confluency and serum were starved for 24 hours then incubated with KP metabolites for defined time points followed by total protein or cytoplasmic and nuclear protein extraction. For experiments involving signaling pathway inhibition, HT-29 cells were incubated with 20 μmol/L PI3K (GDC-0941, Selleckchem) or Akt (MK2206, Selleckchem) inhibitor 4 hours prior to KP metabolite incubation. Furthermore, to determine whether the activation of Akt signaling by Kyn is intracellular, a putative Kyn transporter, L-amino acid transporter 1 (LAT1), was blocked using specific inhibitors BCH and JPH203 overnight prior to KP metabolite incubation. Organoids were first isolated from Matrigel using cell recovery solution (catalog no. 354253; Corning Life Sciences) after overnight serum depletion, then incubated with KP metabolites in serum-free DMEM/F12 followed by total protein extraction.

Colorectal cancer cell lines and organoids were plated in flat-bottom 96-well plates. Eight hours after plating, colorectal cancer cells were serum starved overnight followed by incubation with 100 μmol/L KP metabolites in DMEM containing 1% FBS. Organoids were plated in Matrigel and maintained in L-WRN (20% FBS) media for 12 hours after plating. The media were then changed to 1% FBS containing differentiation media and maintained for 12 hours followed by the addition of 100 μmol/L KP metabolites in 1% FBS differentiation media. Tumor organoids were maintained in advanced DMEM (20% FBS) for 12 hours followed by maintenance in 1% FBS-containing differentiation media for 12 hours. After 12 hours, 100 μmol/L KP metabolites were added in 1% FBS containing differentiation media. Proliferation was measured using CCK-8 assay (Dojindo Laboratories) after 72 hours of incubation with KP metabolites in both colorectal cancer cell lines and organoids.

Expression of cleaved PARP was used as a measure of apoptosis by Western blotting and immunofluorescence. HT-29 cells were grown to approximately 75% confluency before serum starving overnight. To induce apoptosis, cells with treated with 100 ng/mL TNFα (PeproTech) or 2 μmol/L Staurosporine (Cell Signaling Technology) in DMEM containing 1% FBS for 12 hours with or without combinations of Akt inhibitor (MK2206) and 100 μmol/L Kyn. Cells were pretreated with 5 ng/mL cycloheximide for 1 hour before incubating with TNFα or staurosporine (STP). Cleaved PARP was determined using rabbit anti-cleaved PARP mAb (Cell Signaling Technology) by Western blot analysis.

For immunofluorescence, cells were seeded on coverslips in 6-well plates and grown to approximately 75% confluency. Media were aspirated and coverslips were washed in cold PBS then fixed in fresh 4% paraformaldehyde in PBS for 20 minutes at room temperature. Cells were permeabilized (0.1% Triton X-100 in PBS for 10 minutes), washed in cold PBS, and then blocked with 1% BSA containing 1% normal goat serum for 60 minutes at room temperature. Cells on coverslips were then probed with anti-cleaved PARP mAb (Cell Signaling Technology) at 1:400 dilution in blocking buffer at 4°C overnight. After washing in cold PBS, goat anti-rabbit AlexaFluor 488 (Thermo Fisher Scientific) at 1:600 dilution was added in blocking buffer for 60 minutes at room temperature and mounted with Southern Biotech DAPI Fluoromount-G.

For analyzing cell viability after apoptosis induction, the CCK-8 assay was used 24 hours after TNFα or STP incubation with or without 100 μmol/L Kyn cotreatment. Terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) assay was performed to determine apoptosis in mouse tumors using In Situ Cell Death Detection Kit (Roche Applied Science).Immunofluorescence images of cleaved PARP and TUNEL assay were captured with a Zeiss Axioskop 2 MOT microscope equipped with an Apotome module. Immunofluorescence was quantified in ImageJ as follows: black and white images of the blue channel were converted to binary and nuclei were defined using nuclei watershed separation setting, followed by particle analysis with size exclusion of 0.005 in²-infinity. Green channel image thresholds were adjusted until individual nuclei were visible, and particle analysis was performed as above. Data are expressed as percent of particles in green channel (positive nuclei) divided by number of particles in blue channel (total nuclei) × 100.

Murine colon tissues were fixed in formalin and stabilized in 2% agar before embedding in paraffin as described previously (27). All antibodies are described in Supplementary Table S1. Ido1 immunostaining with rat anti-mouse IDO1 antibody (BioLegend) was used at 1:50 dilution after antigen retrieval by steaming in sodium citrate buffer (pH 6.0) for 30 minutes. Trilogy one step antigen retrieval (Millipore Sigma) was performed as per the manufacturer's instructions for pAkt SER473, cyclin D1, β-catenin, and survivin immunostaining (Cell Signaling Technology at 1:50 dilution) on mouse tissue. Tissue sections from postirradiated, surgically resected human colon adenocarcinoma were obtained from Siteman Cancer Center histopathology core. Trilogy one step heat-mediated antigen retrieval was performed for immunostaining IDO1 and pAkt SER473. The relative intensity of immunostaining for human IDO1 and pAkt SER473 was graded on the basis of a point system (no staining = 0, low expression = 1, high expression = 2). Colorectal cancer pathology samples from 5 patients were analyzed by selecting neoplastic crypts (>10/specimen, 141 total) followed by sequential evaluation of IDO1 staining and pAKT SER473. The grading was confirmed by two contributing scientists. Using this data, a correlation curve between IDO1 and pAkt SER473 expression was calculated.

Animal experiments included 5 to 10 mice in each group and were repeated twice. Cell proliferation assays involved data accumulation from five to eight wells/group and experiments were repeated at least twice. Wound healing assays included data from three wells in each group and the experiments were performed twice. All data were represented as average ± SEM. P values were calculated using two-way ANOVA or Student t test and a value less than 0.05 was considered as significantly different between groups. The Spearman correlation coefficient was used to analyze the correlation between IDO1 and Akt SER473 based on their IHC staining scores. Figures and statistical analysis were done using GraphPad Prism version 5.00 for Windows (GraphPad Software, www.graphpad.com).

We previously showed that germline deletion of Ido1 reduces tumor burden in a mouse model of colorectal cancer and that inhibition of IDO1 activity reduces colorectal cancer cell proliferation in vitro. These findings suggest that the neoplastic epithelium may be a key cellular source of IDO1 activity for driving colorectal cancer growth. To directly examine this issue, we generated a mouse with genetic knockout of Ido1 specifically in the intestinal epithelium (IDO1-iKO) using cre-lox recombination technology (Fig. 1A). Mice with a floxed Ido1 gene were crossed to mice expressing cre behind the intestinal epithelial-specific villin promoter. The resulting mouse line demonstrated effective knockout of Ido1 protein expression in the colonic epithelium, but not nonepithelial lamina propria cells (Fig. 1B–D). Colonoids established from normal colons of WT, but not iKO mice, showed change in Kyn production after IFNγ treatment (Fig. 1E), illustrating the functional knockout of Ido1 and the relative importance of IDO1 (vs. IDO2 or TDO) to the KP in the colon epithelium. Additional details on mouse modeling and Ido1 expression data are presented in Supplementary Fig. S1A–S1D.

Colon tumorigenesis was induced in IDO1-iKO and littermate WT control mice using the AOM-DSS protocol. No difference in weight loss was observed between the two groups and the disease activity index differed only briefly during the first DSS cycle (Supplementary Fig. S1B and S1C). However, IDO1-iKO mice were found to develop fewer and smaller tumors compared with WT mice (Fig. 1F and G). Nuclear β-catenin was visibly reduced in the tumors of IDO1-iKO versus WT mice, as was expression of the target gene cyclin D1 and the tumor cell proliferation index by 16% (Fig. 2A–C). These effects were not significant in the adjacent normal colon epithelium, although the colonic crypt epithelial proliferation index trended lower in the IDO1-iKO mice. Tumors from IDO1-iKO mice also demonstrated increased apoptosis (TUNEL staining) and reduced expression of Survivin (a key protein that regulates mitosis and apoptosis; Fig. 2D and E). Together, these data illustrate the importance of epithelial-based IDO1 expression as a driver of neoplastic proliferation, apoptotic resistance, and colon tumorigenesis.

We next sought to examine the signaling mechanism by which IDO1 activity promotes neoplastic epithelial proliferation, nuclear translocation of β-catenin, and resistance to apoptosis. In these experiments we examined the impact of Kyn on colorectal cancer cells using in vitro methods. We found that both Kyn rapidly promoted nuclear translocation of β-catenin from the cytoplasm in a time dependent manner (Fig. 3A and B).

Unchecked nuclear β-catenin–mediated transcriptional activity is a common feature of colorectal cancer and usually attributable to genetically driven aberrant signaling along the canonical Wnt pathway (32). Therefore, we examined whether KP metabolites act as endogenous Wnt ligands. Two approaches were taken, but neither confirmed this mechanism. First, exogenous Kyn did not promote activation of LRP5/6 (lipoprotein receptor–related protein), a coreceptor that is phosphorylated in the presence of Wnt ligands. Second, exogenous Kyn promoted enhanced colorectal cancer cell proliferation even in the presence of the Wnt antagonist Dickkopf (Fig. 3C and D).

We next evaluated whether KP metabolites activate protein kinase B (Akt), a key signaling intermediate known to inhibit apoptosis while promoting β-catenin nuclear translocation and transcriptional activity. This mechanism involves phosphorylation of Akt at Ser473 and Thr308 residues leading to increased phosphorylation of β-catenin at Ser552 and has been linked to inflammation in colorectal cancer (33, 34). Indeed, we found that exogenous Kyn application to serum-starved cells promoted rapid and dose-dependent Akt activation evidenced by increases in pAKT S472 and phosphorylated PRAS40 pT246, a direct target Akt activity (Fig. 3E).

To address the physiologic relevance of these concentrations, we assessed the intracellular Kyn levels in two colorectal cancer cell lines used in this study and found them to reach upwards of 75 μmol/L (Fig. 3F). Supporting these findings, prior studies have demonstrated intracellular concentrations of Kyn to exceed 2 mmol/L in malignant myeloid-derived cells (such as those that infiltrate the tumor) and to reach at least 100 μmol/L in WiDr colorectal cancer cells (35). On the basis of these data, we used 100 μmol/L of Kyn henceforth to examine the cell signaling pathway.

We next sought to understand the pathways upstream and downstream to Kyn-mediated Akt activation as relevant to colorectal cancer and to β-catenin activation. Upon activation, Akt signaling is known to interact with β-catenin indirectly by inactivating GSK3β (36), an enzyme that when in its active state, destabilizes β-catenin towards degradation. Indeed, we identified that initial Kyn and terminal (QA) KP metabolites rapidly inactivate GSK3β by inducing its phosphorylation at Ser9. This mechanism also applied to other colorectal cancer cell lines despite known β-catenin stabilizing genetic mutations and regardless of status for microsatellite instability or KRAS mutation (Fig. 4A). Together, these data suggest a positive synergy between IDO1 activity in the neoplastic epithelium and the β-catenin stabilizing genetic mutations that drive colon cancer development and progression.

We also aimed to define signaling upstream from Akt activation. First, we identified that intracellular Kyn was required to activate Akt as inhibition of the LAT1 prevented this effect (Fig. 4B). LAT1 is an active transporter of Kyn and known to be highly upregulated in colorectal cancer (37, 38). Akt resides at the intersection of numerous signaling pathways including PI3K, PTEN, and positive feedback from mTOR. We did find that Kyn promoted mTOR phosphorylation (pmTor Ser2448) in colorectal cancer cells, a form that binds to both raptor and rictor and is shown to be overexpressed in human colorectal cancer (39, 40). However, this occurred only after 15 minutes, suggesting this was secondary (rather than proximate) to Akt activation (Supplementary Fig. S2A). Although PTEN is a negative regulator of Akt activation, (PI3K directly activates Akt. We confirmed the involvement of PI3K with inhibitor studies and found that Kyn/QA did not induce β-catenin, GSK3β, or PRAS40 phosphorylation in the presence of Akt or PI3K inhibition (Fig. 4C). Finally, although activation of the aryl hydrocarbon receptor (AHR) activation is implicated in mediating some of the biologic effects of the IDO1-KP metabolites, we found that rapid Akt activation occurred despite AHR antagonism (Fig. 4D). Altogether, these results indicate that Kyn promote β-catenin activity via a PI3K/AKT/GSK3β signaling event, rather than AHR-mediated transcriptional activity.

Having shown that KP metabolites rapidly activate Akt signaling in colorectal cancer cell lines, we then sought to determine whether this mechanism extended in tumor epithelium in vivo. Indeed, we found expression of pAkt Ser473 to be reduced in the neoplastic colon epithelium of iKO versus WT mice (Fig. 5A). We next evaluated a series of five human colorectal cancer samples and found that crypt staining for IDO1 mirrored and significantly correlated with that of pAkt Ser473 staining (Fig. 5B and C).

IDO1 catalyzes the initial and rate-limiting step in tryptophan metabolism to Kyn initiating a cascade that generates several bioactive downstream KP metabolites (Supplementary Fig. S3A). We next evaluated the capacity of each of these metabolites to activate the Akt/GSK3β/β-catenin pathway and to increase colorectal cancer cell proliferation. The majority of KP metabolites activated the Akt/GSK3β/β-catenin pathway and promoted colorectal cancer proliferation. Only KA did not activate Akt and pPRAS40 T246 and conversely suppressed colorectal cancer proliferation (Supplementary Fig. S3B–S3D). These findings were confirmed in DLD1 colorectal cancer cells as well (Supplementary Fig. S3E and S3F). Consistent with our findings of KP metabolites potentiating malignant colorectal cancer activity, we also confirmed that Kyn and QA promote wound closure (Supplementary Fig. S4A and S4B).

PI3K/Akt pathway activation is a recognized mechanism by which colorectal cancer cells resist stress-induced apoptosis (41–43). We thus sought to determine whether KP metabolites offer protection against apoptosis in colorectal cancer cells. Indeed, Kyn cotreatment reduced TNFα-induced apoptosis as measured by cleaved PARP and enhanced colorectal cancer cell viability (Fig. 6A–D). Importantly, we found that Kyn did not prevent TNFα-induced apoptosis in the presence of an Akt inhibitor. We also found that Kyn reduced staurosporine-induced apoptosis (Supplementary Fig. S5A–S5D). These results complement the in vivo data presented in Fig. 2 and collectively indicate that KP metabolites promote tumor progression by promoting Akt-induced resistance to apoptosis and increasing neoplastic survival during inflammatory and cellular stress.

In IDO1-iKO mice, the epithelial cell proliferation index was significantly reduced in the tumor, but the difference was less pronounced in adjacent normal crypts (Fig. 2C). To further explore potential differences in how KP metabolites impact normal versus neoplastic epithelial cells and to extend the findings further to human disease, we used human-derived colon tumoroid and normal colonoid cultures. Similar to our observations in monolayer colorectal cancer cell cultures, Kyn and QA induced rapid Akt/GSK3β/β-catenin signaling in human colorectal cancer–derived tumoroids (Fig. 7A). However, in the colonoids derived from normal colon epithelial crypts, Kyn and QA provoked no rapid response in phosphorylation. Enhanced proliferation and cyclin D1 levels were also observed in colorectal cancer–derived tumoroids compared with minimal change in proliferation in normal colonoids (Fig. 7B and C). These data indicate that neoplastic transformation drives an augmented response of the colon epithelium to IDO1 pathway activity. A summary illustration of the overall findings is shown in Fig. 7D.

The immunometabolic IDO1 pathway has emerged as an important therapeutic immunooncology target for solid tumors including colon cancer (17). The recognized capacity of this enzyme to suppress antitumoral immunity provides the mechanistic basis for current clinical trials of IDO1 inhibitors. In this study, we define a novel, epithelial cell–centric mechanism and signaling cascade by which IDO1, through Kyn, interacts directly with the signaling pathways of the neoplastic cells to promote tumor growth. Using cell culture, animal models, and human colorectal cancer tissues, we illustrate that epithelial IDO1 activity and several KP metabolites directly promote colon tumorigenesis by activating the proproliferative/antiapoptotic PI3K/Akt pathway in the neoplastic epithelium. These findings provide mechanistic insight and strengthen justification for clinical trials evaluating IDO1 inhibitors as adjuvant therapeutics in colon cancer. Furthermore, these findings may have implications for the rational design of future clinical trials that would exploit a synergy of IDO1 inhibitors and conventional cancer therapies where Akt activation provides resistance such as radiation.

Linking IDO1 activity and KP metabolites to PI3K/Akt signaling provides a new mechanistic insight to how inflammation fuels neoplastic growth. Activation of Akt signaling or impaired expression of PTEN (a negative regulator of Akt) is reported in a majority of human colon cancers, whereas inhibitors of PI3K/Akt signaling are considered potential therapeutic agents (44). We demonstrated that intracellular KP activation rapidly initiates several Akt regulated protumorigenic effects including β-catenin activation and proliferative cell-cycle signaling, prevention of apoptosis from internal and extrinsic sources of cellular stress, and activation of the mTOR signaling pathway. These findings also define the upstream signaling pathway for our previous observations linking IDO1 to β-catenin activation as being mediated through PI3K/Akt signaling, which is implicated in colitis and colorectal cancer (34, 45). The precise mechanism by which KPs rapidly activate PI3K is an area of future study with our preliminary investigations not implicating the common activator Src (Supplementary Fig. S6A). However, the rapidity of these events (initiated within 5 minutes) do point to activation of cell signaling cascades rather than activating translational regulators such as AHR or induction of COX2, which Kyn has been tied to in inflammation and malignancy (46, 47). Regarding COX2, we confirmed the recent observation that Kyn induced its expression, but required a longer time course ranging from hours, rather than minutes (Supplementary Fig. S6B).

To our knowledge, this is the first study to demonstrate a specific functional role for epithelial-based IDO1 as an independent promoter of tumorigenesis using tissue-specific mouse modeling. Mice with genetic deletion of colon epithelial IDO1 demonstrate fewer tumors, significantly reduced tumor proliferation and showed higher apoptosis than WT mice. Furthermore, KP metabolites directly promoted colorectal cancer cell proliferation, resistance to apoptosis and restoration after injury. Given that the levels of intracellular Kyn within cancer cells can reach high micromolar to low millimolar concentrations (35), it should not be surprising that they drive biologic effects within the neoplastic cells. Our own analysis shows on intracellular concentrations of Kyn show that the levels can reach >70 μmol/L, which when applied exogenously activated Akt (Fig. 3). Thus, together with our published work showing IDO1 inhibition directly suppresses proliferation in colorectal cancer cells in vitro. These findings support the hypothesis that epithelial IDO1 and Kyn are both sufficient and necessary to drive colon tumor growth. The protumorigenic, epithelial cell–intrinsic effects of IDO1 activity no doubt augment the pathogenic immunosuppressive effect of IDO1 expressing bone marrow–derived monocytic cells and tumor cells.

Differences in IDO1-mediated effects between normal and neoplastic epithelial cells are also highlighted by the current data. In vivo, tumor cell proliferation is higher in WT than IDO1-iKO mice, but not significantly higher in the adjacent normal epithelial crypts. In vitro, KP metabolites robustly activate Akt and potentiate β-catenin activity to promote proliferation in colorectal cancer cell lines and colon tumoroids derived from patients with APC mutations. However, only delayed and modest effects were observed in colonoids derived from normal tissue. Prior studies had already identified that expression of IDO1 and other Kyn-producing enzymes is often high and constitutively active after neoplastic transformation, compared with the nontransformed state (3, 25, 26, 46). Intriguing studies have implicated COX2 and inflammation-associated AHR signaling as regulators of this phenomenon (47, 48). Our studies now illustrate a functional significance to this observed difference in expression by demonstrating that the progrowth, epithelial cell–centric effects of IDO1 activity are also more pronounced in neoplastic over normal epithelial cells. Taken together, these data suggest a positive synergy between acquisition of constitutive IDO1 activity and the genetic mutations that drive colon cancer progression, including those that promote stabilized β-catenin. Thus, although the in vivo AOM/DSS model of colitis-associated cancer was used in this study, we predict that the phenotype extends to APC and β-catenin mutation-driven sporadic colon cancers as well.

As an immunomodulatory pathway, there is existing rationale and clinical precedence for examining IDO1 inhibition in colorectal cancer. High tumor IDO1 expression is observed in a significant subset of with colorectal cancer and high IDO1 expression at the tumor invasion front is an independent adverse prognostic factor for overall survival and metachronous metastases (16, 20, 25, 26). As promising as this sounds, IDO1 inhibition as a monotherapy appears inadequate in colorectal and other cancers. A recently published phase I trial examined a novel hydroxyamidine small-molecule IDO1 inhibitor (Epacadostat, Incyte Corp.; refs. 11, 49) in individuals with advanced solid tumors failing prior therapies. Twenty-nine of 52 patients enrolled (56%) had colorectal cancer (50). Although the study met safety and biochemical efficacy endpoints, no patients achieved complete or partial response. Current studies with Epacadostat are examining it in combination with other immune-checkpoint inhibitors targeting CTLA-4 and the programmed cell death (PD-1) pathway.

Our findings provide new mechanistic rationale for targeting IDO1 inhibition in colorectal cancer and potentially add directionality to future clinical studies. IDO1 inhibitors may synergize with or augment the effects of cytotoxic chemo- or radiation therapy as the Akt pathway is intimately involved in preventing apoptotic or radiation-induced cell death in colorectal cancer (51). Moreover, the specificity of IDO1 expression in neoplastic versus normal colon cells may provide a target that could enhance therapeutic efficacy without enhancing the toxicity. Also relevant to cytotoxic cancer therapies, the IDO1–Kyn pathway provides an important source for de novo generation of NAD⁺ (52). NAD⁺ is an important enzymatic cofactor for enzymes involved in DNA repair and inhibitors of its generation are recognized targets in cancer therapy (53, 54). Finally, it is intriguing to think that IDO1 inhibitors might have a therapeutic role in colorectal cancer when combined with other well-tolerated small-molecule drugs, such as inhibitors of the COX2 pathway, with which our study also confirmed this association (47).

KA, unlike other KP metabolites, did not activate Akt and in HT29 cells decreased proliferation. This finding is consistent with published literature that suggested that KA may have chemopreventive effects in colorectal cancer after demonstrating that KA reduced PI3K/Akt signaling (55). Notably, these effects were observed using high (millimolar) doses of KA. Still, these findings are intriguing and potentially suggest that pharmacologic shunting of the Kyn pathway toward KA production and away from the Akt-activating KP metabolites may have a potentiating effect for IDO1 inhibitors or in combination with other therapeutics.

In summary, our results demonstrate a novel mechanistic link between IDO1 activity, Kyn pathway metabolites, and the PI3K/Akt proproliferative and antiapoptotic pathways in colorectal cancer including activated β-catenin signaling. These findings demonstrate the importance of epithelial IDO1 activity in driving colon tumorigenesis and extend from cell lines and mouse models to human colorectal cancer. Together, these findings underscore the value of investigating IDO1 inhibition in colorectal cancer, provide insights into mechanism of action, and may stimulate consideration for future clinical trials that may exploit a synergy of IDO1 inhibitors and therapies where Akt activation provides resistance.

M.A. Ciorba reports receiving a commercial research grant and is a consultant/advisory board member for Incyte. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K.S. Bishnupuri, D.M. Alvarado, B. Chen, M.A. Ciorba

Development of methodology: K.S. Bishnupuri, B. Chen, M.A. Ciorba

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.S. Bishnupuri, D.M. Alvarado, A.N. Khouri, M. Shabsovich, B. Chen, B.K. Dieckgraefe, M.A. Ciorba

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.S. Bishnupuri, D.M. Alvarado, B. Chen, B.K. Dieckgraefe, M.A. Ciorba

Writing, review, and/or revision of the manuscript: K.S. Bishnupuri, D.M. Alvarado, B. Chen, B.K. Dieckgraefe, M.A. Ciorba

Study supervision: M.A. Ciorba

M.A. Ciorba has support from a Crohn's and Colitis Foundation Daniel H Present Senior Research Award (Ref. 370763), NIH grants (DK109384, DK100737, and AI095776), philanthropic support from the Givin' it all for Guts Foundation, https://givinitallforguts.org/, and a Central Society for Clinical Research Early Career Development Award. B.K. Dieckgraefe was supported by I01 BX003072. Core support was from The Washington University Digestive Diseases Research Core Center (P30 DK052574) and Siteman Cancer Center (P30 CA91842). D.M. Alvarado was supported by DK077653 and The Lawrence C. Pakula MD IBD Innovation Fellowship. We acknowledge technical assistance from Srikanth Santhanam PhD. Finally, we are grateful to Nicholas O. Davidson MD for his unwavering and enthusiastic support of gastroenterology research at Washington University and for his insightful comments on this project.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.