Targeting Indoleamine 2,3-Dioxygenase in Cancer Models Using the Novel Small Molecule Inhibitor NTRC 3883-0

Indoleamine 2,3-dioxygenase (IDO1) is a key regulator of immune suppression by catalyzing the oxidation of L-tryptophan. IDO1 expression has been related to poor prognosis in several cancers and to resistance to checkpoint immunotherapies. We describe the characterization of a novel small molecule IDO1 inhibitor, NTRC 3883-0, in a panel of biochemical and cell-based assays, and various cancer models. NTRC 3883-0 released the inhibitory effect of IDO1 on CD8-positive T cell proliferation in co-cultures of IDO1-overexpressing cells with healthy donor lymphocytes, demonstrating its immune modulatory activity. In a syngeneic mouse model using IDO1-overexpressing B16F10 melanoma cells, NTRC 3883-0 effectively counteracted the IDO1-induced modulation of L-tryptophan and L-kynurenine levels, demonstrating its in vivo target modulation. Finally, we studied the expression and activity of IDO1 in primary cell cultures established from the malignant ascites of ovarian cancer patients. In these cultures, IDO1 expression was induced upon stimulation with IFNγ, and its activity could be inhibited by NTRC 3883-0. Based on these results, we propose the use of ascites cell-based functional assays for future patient stratification. Our results are discussed in light of the recent discontinuation of clinical trials of more advanced IDO1 inhibitors and the reconsideration of IDO1 as a valid drug target.


INTRODUCTION
The essential amino acid L-tryptophan (Trp) is an important regulator of cancer progression due to its regulatory role in immune cell activity (1,2). Depletion of Trp in the tumor microenvironment results in T cell anergy and inhibition of natural killer cell activity (1,2). The catabolism of Trp is regulated by two distinct, evolutionary unrelated enzymes, indoleamine 2,3-dioxygenase 1 (IDO1) (EC 1. 13.11.42), and tryptophan 2,3-dioxygenase (TDO) (EC 1. 13.11.11). Both enzymes catalyze the oxidation of Trp, resulting in the formation of N-formyl kynurenine (NFK), which is rapidly converted into kynurenine (Kyn) by the enzyme kynurenine formamidase.
IDO1 and TDO have a distinct tissue distribution and carry out different physiological roles. IDO1 is broadly expressed at low levels in normal tissues (3), but is strongly induced by proinflammatory stimuli, such as IFNg (4). In this manner, the expression of IDO1 by various cell types, including immune, endothelial and epithelial cells, provides a control mechanism to dampen the immune response (5,6). In contrast, TDO is constitutively expressed at high levels in the liver, where it functions to maintain Trp homeostasis (7). Consistent with this role, TDO has a relatively low affinity for Trp (K M,Trp , 190 µM) in comparison to IDO1 (K M,Trp , 6 µM) (8).
In cancer, expression of IDO1 has been observed in both tumor and immune cells. IDO1 expression has been correlated with poor prognosis, increased progression and reduced survival in several cancers (9,10). Moreover, it has been related to resistance to anti-PD-1 and anti-PD-L1 immunotherapies (10,11). In the past decade, several small molecule IDO1 inhibitors have been developed to enhance the efficacy of immunotherapy (12). The first-in-class and most advanced IDO1 inhibitor is epacadostat (INCB024360), which is a reversible, substrate-competitive enzyme inhibitor (13). More recently, increasingly potent IDO1 inhibitors have been described which act by displacing the heme cofactor of the enzyme, as exemplified by linrodostat (BMS-986205) (14). Phase I clinical studies with epacadostat have demonstrated that IDO1 inhibition is well tolerated in human patients and results in dose-dependent reductions of Kyn levels and Kyn/Trp ratios in plasma (15). In a phase I combination trial with the anti-PD-1 immunotherapeutic pembrolizumab, objective response was seen in 12 out of 22 melanoma patients treated with epacadostat (16). However, a large phase III combination trial with pembrolizumab (ECHO-301/KEYNOTE-252) was terminated prematurely, since there was no clinical benefit of the combination over pembrolizumab plus placebo (17).
Here we describe NTRC 3883-0, a potent, selective and orally bioavailable IDO1 inhibitor, which is structurally distinct from existing IDO1 inhibitors (12). NTRC 3883-0 was profiled side-byside with epacadostat in biochemical and cell-based assays for human (h) and mouse (m) IDO1, and the selectivity target TDO. Its immunomodulatory activity was studied in vitro in a co-culture assay of an hIDO1-overexpressing cell line with lymphocytes. Modulation of IDO1 activity in vivo was studied in a syngeneic mouse model of melanoma induced with mIDO1-overexpressing B16F10 cells. Finally, we studied the expression of IDO1 and its modulation by NTRC 3883-0 in ex vivo primary cell cultures established from the malignant ascites of ovarian cancer patients.

IDO1, TDO, and CYP Biochemical Assays
Full-length recombinant human and mouse IDO1 were expressed in Escherichia coli with a hexahistidine tag at the Nterminus, and purified by affinity chromatography to > 95% purity, as described (8). Human and mouse TDO containing a Cterminal hexahistidine tag were analogously expressed and purified to > 95% purity (8). The inhibitory activity of compounds was determined with the NFK GreenScreen assay technology under the conditions as described (8), except for the reaction buffer for IDO1, which was replaced by 50 mM NaH 2 PO 4 , pH 7.0, supplemented with 0.01% Tween-20 and 1% glycerol. Inhibition of the CYP3A4 and CYP2D6 enzymes was determined with the P450-Glo ™ Screening System with Luciferin-IPA (Promega, #V9920 and V9890). For determination of IC 50 values, the effect of inhibitors was determined in 10-point duplicate dose-response curves. Graphs were fitted to a four-parameter logistics equation in XLfit (ID Business Solutions, Ltd., Guildford, UK) from which IC 50 values were calculated.
the GL-261 cell line from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ, #ACC 802, RRID: CVCL_Y003). Each cell line was cultured in the medium recommended by the supplier, with FBS replaced by BCS for the B16F10 and GL-261 cell lines. Sublines of GripTite ™ 293 MSR stably overexpressing human IDO1 (HEK-hIDO1), GripTite ™ 293 MSR stably overexpressing human TDO2 (HEK-hTDO), B16F10 stably overexpressing mouse IDO1 (B16F10-mIDO1), and GL-261 stably overexpressing mouse TDO2 (GL-261-mTDO) were generated by transfection of the respective cells with respectively full-length hIDO1, hTDO2, mIDO1, or mTDO2 cDNA cloned in the expression vector pEF6v5 (Thermo Fisher, #V96120). Cells were transfected with Lipofectamine 3000 (Thermo Fisher) and single cell clones were selected by blasticidin selection and limited dilution in 384-well culture plates. Trp-catabolizing activity was evaluated using NFK Green or 4-(dimethylamino)benzaldehyde (pDMAB) (20). To assess the stability of IDO1 or TDO2 expression, clones were cultured for 4 weeks in the absence of blasticidin, during which they were tested weekly for Trp-catabolizing activity. Afterwards, the mRNA and protein levels, and the inhibitory potency of reference compounds were determined in the resulting sublines.
Cell-based assays with the transfected cell lines were performed in the absence of blasticidin.

Cell-Based Assays
Inhibition of the Trp-catabolizing activity in A375 cells, HEK-hTDO cells, GL-261-mTDO cells and primary patient samples (all seeded at 8000 cells/well) was determined after incubation of the cells with compound for 1 h, followed by supplementation of the culture medium with 200 µM Trp and incubation for 42 h. The A375 cells and primary patient samples were stimulated with 200 ng/ml IFNg prior to the incubation period. The remaining cell-based assays were performed analogously to the HEK-hTDO and GL-261-mTDO assays, with the following exceptions. For the SW48 cells, incubation with compound and Trp was performed for 18 h. For the HEK-hIDO1 cells, the cells were seeded at 16,000 cells/well, and no additional Trp was added to the culture medium. For the B16F10-mIDO1 cells, no additional Trp was added to the culture medium, and the cells were incubated for 66 h with compound. Inhibition of the Trpcatabolizing activity in heparinized human whole blood was determined after three-fold dilution of the blood in RPMI 1640 supplemented with 1% penicillin/streptomycin (P/S), incubation  NTRC 3883-0 and its analogues were developed by medicinal chemistry optimization from a 3-hydroxyimidazolin-4-one hit compound (1) identified by ultra-high throughput screening. Inhibitory potencies (IC 50 ) were determined in a human IDO1 biochemical assay and a cell-based assay with IFNg-stimulated A375 human melanoma cells. All analogues were inactive in a biochemical assay for TDO (IC 50 > 31.6 µM).
of the diluted blood with compound for 1 h, addition of 800 µM Trp and 200 ng/ml IFNg, and incubation for 66 h. For the A375, SW48, HEK-hIDO1, and HEK-hTDO cellbased assays, the Trp-to-NFK conversion was measured with NFK Green, as described previously (8). For the B16F10-mIDO1 cells, GL-261-mTDO cells, and heparinized human whole blood, the Trp-to-NFK conversion was measured with pDMAB (20). Briefly, 5% trichloroacetic acid in Milli-Q water was added to each well, followed by incubation at 55°C for 1 h, centrifugation for 20 min at 2,900 x g, transfer of the supernatant to a new plate, and addition of 2% pDMAB (Fisher Chemicals) dissolved in acetic acid to each supernatant. After incubation for 10 min at room temperature, the formation of L-kynurenine was determined by measurement of the absorbance at 480 nm. In parallel with the cell-based NFK Green assays, the cytotoxicity of the compounds was determined in cell viability assays using ATPlite ™ 1Step (PerkinElmer) (8). NTRC 3883-0, epacadostat and NTRC 3531-0 were not cytotoxic to any cell line used in this study.

Gene Expression Analysis
RNA was isolated from cell cultures and tumors with the RNeasy Mini kit (Qiagen). cDNA was prepared using the QuantiTect Reverse Transcription kit (Qiagen) and quantitative real-time PCR (qPCR) was performed in a Bio-Rad CFX96 cycler using SYBR ™ Select Master Mix (Thermo Fisher) following standard protocols (21). The sequences of the qPCR primers are provided in Table S1.

In Vivo Pharmacokinetics and Pharmacodynamics Models
In vivo pharmacokinetics studies were performed at Charles River Laboratories ('s-Hertogenbosch, The Netherlands). In vivo pharmacodynamic and intervention studies were performed at ProQinase (Freiburg, Germany) and Charles River Laboratories (Morrisville, NC). All experimental protocols were approved by, and performed in accordance with the guidelines of the Ethics Committee for Animal Experimentation of the respective organizations.
Pharmacokinetic properties of NTRC 3883-0 were determined in plasma following single oral (p.o.) and i.v. administration to male CD-1 mice (n = 6 per treatment group with three mice sampled per time point), Wistar rats (n = 3), cynomolgus monkey (n = 1) and Beagle dogs (n = 3). Formulations were prepared in DMSO, Kolliphor ® EL, 5% D-mannitol in Elix water at a volume ratio of 1:1:8 for i.v. administration, and in 0.5% gelatin, 5% D-mannitol in Elix water for p.o. administration. Plasma samples were collected at eight to eleven time points up to 24 h after administration. Pharmacokinetic properties and required dosing in humans were predicted by allometric scaling by Karin Jorga Life Science Consulting (Basel, Switzerland) ( Figure S2 and Table S2).
For the first CT26 mouse model study, performed at ProQinase, 5 × 10 5 CT26 cells were implanted subcutaneously into female BALB/c mice on day 0. From day 8 onwards, 50 mg/ kg of NTRC 3748-0 ( Table 1) or 100 mg/kg of epacadostat was administered p.o., respectively twice daily (b.i.d.) and once daily (q.d.), as a suspension in 0.5% gelatin, 5% mannitol in water, with n = 8 mice per treatment group. Tumor growth was monitored using caliper measurement 2 times weekly. The study was terminated at day 22 based on the tumor burden. Tumors and plasma were sampled 2 h post last dose.
For the second study, performed at Charles River Laboratories, 3 × 10 5 CT26 cells were implanted subcutaneously into female BALB/c mice on day 0. From day 8 onwards, 100 mg/kg of NTRC 3883-0 or epacadostat was administered p.o. b.i.d. as a suspension in 0.5% gelatin, 5% mannitol in water, with n = 10 mice per treatment group. Tumor growth was monitored using caliper measurement 3 times weekly. A tumor volume of 2000 mm 3 was used as the endpoint of the experiment for the individual animals. Tumors were sampled 2 h post last dose.
For development of the mIDO1-overexpressing B16F10 mouse model, performed at Charles River Laboratories, female B6D2F1 mice were implanted subcutaneously with 1 × 10 6 parental B16F10 cells or B16F10-mIDO1 cells (subline i6 or j19) on day 0 with n = 10 mice per group. A tumor volume of 2000 mm 3 was used as the endpoint of the experiment for the individual animals. To determine the optimal inoculation cell concentration, mice were implanted with 3 × 10 5 , 1 × 10 5 or 3 × 10 4 parental or mIDO1-overexpressing B16F10 cells (subline j19) with n = 5 mice per group. The same endpoint of the experiment was used as described above. This yielded on average no differences in growth rate between the parental and mIDO1-overexpressing groups, but instead resulted in a strong reduction of the take rate at the lower inoculation concentrations. For the intervention study, 3 x 10 5 cells of the B16F10-mIDO1 subline j19 were implanted on day 0, and treatment was started on day 3 with vehicle (DMSO, Kolliphor, 5% mannitol; 10/10/80, v/v/v), 100 mg/kg NTRC 3883-0 or epacadostat administered p.o. b.i.d. with n = 12 mice per treatment group. Four animals were excluded from analysis due to non-treatment-related deaths. Tumors and plasma were sampled 2 h post last dose.

Quantification of Inhibitor, Trp, and Kyn Levels
Tissues were homogenized in PBS at a ratio of 1 mg per 10 ml using micro pellet pestles mounted on a Micro-Tube Homogenizer System (Thomas Scientific). After centrifugation, proteins were removed from the homogenized tissue supernatants and plasma samples by precipitation with methanol. The concentrations of inhibitor, Trp and Kyn were determined by LC-MS/MS.

Patient Samples
Ascites was collected from patients with a primary diagnosis of advanced epithelial ovarian cancer during paracentesis for diagnostics or symptom relieve, or during primary debulking surgery. Ascites was filtered and cells were collected by centrifugation as described (23). To enrich for tumor cells, ascites cell samples were allowed to adhere overnight to tissue culture plastic. The cell samples were then cultured for 2-3 weeks in advanced RPMI medium (ThermoFisher), supplemented with 10% FBS, 1% glutamax and 1% P/S, with 2 or 3 times passaging of the cells. qPCR analyses of the MUC16, HE4, IDO1, and TDO2 genes were performed for all samples as described above. Five samples which could be cultured over multiple passages were analyzed for cell surface expression of mucin-16 and EpCAM by flow cytometry using FITC-conjugated anti-mucin-16 (AssayPro, #32189-05141) and APC-conjugated anti-EpCAM (Miltenyi, #130-111-117, RRID: AB_2657496) to assure that the cell cultures consisted of malignant cells. IDO1 mRNA and protein levels in these samples were analyzed as described above after incubation with or without 200 ng/ml IFNg for 24 h. The collection of ascites and the research described was conducted with approval of the medical ethical committee of the Radboud university medical center and informed written consent from each subject.

Statistical Analyses
Differences in plasma Trp and Kyn levels between the B16F10 mouse model groups with and without tumor development during the intervention study were analyzed using the Mann-Whitney U test. Differences in intratumoral Trp and Kyn levels between the intervention groups were analyzed by Welch's ANOVA followed by Games-Howell post hoc analysis. Statistical analyses were performed in RStudio.

Discovery of a Novel Class of IDO1 Inhibitors
A 3-hydroxyimidazolin-4-one class of IDO1 inhibitors was identified in a biochemical screen of IDO1 performed by the European Lead Factory. The initial hit, compound 1 in Table 1, inhibited IDO1 with a half-maximal inhibitory concentration (IC 50 ) of 1.2 µM and was inactive on human TDO at the highest tested concentration of 31.6 µM. Cellular activity was determined by measuring inhibition of Trp-catabolizing activity in the human A375 melanoma cell line, which expresses IDO1 after stimulation with IFNg. Analogues of compound 1 were synthesized with modifications at five positions ( Table 1) and isolated as racemic mixtures except for 3k and NTRC 3883-0. Optimization was started with the synthesis of analogs 2a-g, in which the 3-bromophenyl moiety was varied. These analogs showed that this position of the molecule possesses a very tight structure-activity relationship, since deletion (2a), substitution (2b, f, and g) or an alternative position (2c and d) of the 3-bromo group resulted in a significant loss of potency. Having established the importance of the 3-bromophenyl pharmacophore, efforts were focused on analogs of the benzoyl moiety (3a-j in Table 1). Contrary to the introduction of a fluoro group at the para (3a) or meta (3b) position, introduction of an o-fluoro (3c) group resulted in a potency improvement in both the enzymatic and cellular IDO1 assay ( Table 1). Other substituents at the ortho position (3d-f) were not tolerated, except for the o-chloro substitution, which showed a similar potency, albeit at the expense of a higher molecular weight. Addition of a second fluoro group to 3c resulted in compounds 3h-j. The strongest increase in potency was observed for compound 3j. The additional o-fluoro group in 3j presumably helps to orientate the phenyl group in a favorable position perpendicular to the 3hydroxyimidazolin-4-one core, thereby reducing the entropy loss upon binding to IDO1. After chiral separation of the two enantiomers of 3j, NTRC 3883-0 was identified as the more active enantiomer (eutomer).
In Vitro Characterization of NTRC 3883-0 NTRC 3883-0 inhibited human IDO1 with an IC 50 of 123 nM and was inactive on TDO ( Table 2). Trp-catabolizing activity in IFNg-stimulated A375 cells was inhibited with an IC 50 of 182 nM, whereas the compound had no effect on Trp-catabolizing activity in the human colorectal carcinoma SW48 cell line, which constitutively expresses TDO (8). Cellular activity was additionally determined in HEK-293 cells stably overexpressing the full-length human IDO1 (HEK-hIDO1) or TDO2 cDNA (the gene encoding the TDO enzyme; HEK-hTDO) ( Figure 1A), since these cells have no detectable endogenous expression of either IDO1 or TDO as determined by qPCR and immunoblot analysis ( Figure 1A). NTRC 3883-0 inhibited Trp-catabolizing activity in the HEK-hIDO1 cell line with an IC 50 of 119 nM and had no effect on Trp-catabolizing activity in the HEK-hTDO cells ( Figure 1B; Table 2). The effect on IDO1 activity in normal cells was determined in assays with human whole blood (hWB) from healthy donors after stimulation with IFNg. In the hWB assay, NTRC 3883-0 inhibited Trp-catabolizing activity with an IC 50 of 378 nM ( Table 2).  Potency (IC 50 ) in nM of IDO1 inhibitors NTRC 3883-0 and epacadostat in biochemical and cell-based assays for human (h) and mouse (m) IDO1 or TDO. 95% confidence intervals and number of experimental replicates (n) are given within brackets. hWB, human whole blood.
The clinical IDO1 inhibitor epacadostat (13) was profiled in the same assays for reference purposes ( Table 2). Our data confirm that epacadostat is a potent IDO1 inhibitor ( Figure  1B; Table 2). However, a striking discrepancy in its selectivity over TDO in biochemical and cell-based assays was noted. Epacadostat potently inhibited TDO in the biochemical assay with an IC 50 of 54 nM, which is in the same range as its potency in the biochemical assay for IDO1 ( Figure 1B; Table 2). In contrast, epacadostat showed more than 400 times lower activity on TDO in the HEK-hTDO assay, demonstrating that it is selective cellularly.
In vitro profiling of NTRC 3883-0 against a panel of 44 pharmacologically relevant receptors, ion channels, transporters and enzymes at a concentration of 10 µM at Eurofins-CEREP (Le Bois L'Eveque, France) revealed no cross-reactivities, further demonstrating the selectivity of NTRC 3883-0. NTRC 3883-0 also did not inhibit the activity of the cytochrome P450 (CYP) enzymes CYP2D6 and CYP3A4 at 10 µM. Since CYPs, like IDO1, contain a heme center, these data demonstrate that NTRC 3883-0 interacts selectively with the active site of IDO1.

Modulation of Cytotoxic T Cell Proliferation
To determine whether NTRC 3883-0 can modulate immune cell activity in vitro, co-culture experiments of HEK-hIDO1 cells with peripheral blood mononuclear cells (PBMCs) from healthy donors were performed (Figure 2). Since T cell anergy is reported to occur only at Trp concentrations of 1 µM and lower (6,24), the cells were co-cultured in Trp-free medium supplemented with a low concentration of Trp (i.e., 7.5 µM Trp). Cytotoxic T cell proliferation was determined by labeling PBMCs with carboxyfluorescein succinimidyl ester (CFSE), a fluorescent dye that dilutes upon proliferation of cells, followed by analysis of CD8positive T cells by flow cytometry. CD8-positive T cells stopped to proliferate when co-cultured with HEK-hIDO1 cells in medium containing 7.5 µM Trp (Figure 2A). When a high concentration of Trp (200 µM) was added to the co-culture with HEK-hIDO1 cells, the CD8-positive T cells continued to proliferate (Figure 2A). This indicates that the inhibitory effect of IDO1 is caused by depletion of Trp and is not due to the production of an immunosuppressive metabolite. Addition of 10 µM NTRC 3883-0 to co-cultures of HEK-hIDO1 and PBMCs at low Trp concentration relieved the inhibitory effect of IDO1 expression on T cell proliferation ( Figure  2B), analogous to the effect of treatment with 1 µM epacadostat ( Figure 2C). This demonstrates that NTRC 3883-0 can modulate immune cell function.

In Vivo Target Modulation in Syngeneic Mouse Models
Next, we aimed to assess the in vivo activity of NTRC 3883-0 in a syngeneic mouse model. This first required the selection of a suitable IDO1-expressing, implantable cell line. There exist only a small number of mouse cancer cell lines as compared to human cancer cell lines, and only a few have been reported to express IDO1 in vitro (25,26). In the colon carcinoma CT26 and melanoma B16F10 cell lines, the expression of IDO1 is low (25) or not detectable in cell culture (27). However, the expression of IDO1 in CT26 cells is strongly increased after implantation in syngeneic mice (25), while implanted B16F10 cells have been found to promote IDO1 expression in dendritic cells (28). This indicates that IDO1 gene expression can be induced by the tumor microenvironment, which is substantiated by studies with mice deficient for IFNg or depleted of CD8-positive T cells indicating that upregulation of IDO1 in the tumor microenvironment is regulated by IFNg released by CD8-positive T cells (29).
We initially selected the CT26 model to assess the in vivo efficacy of NTRC 3883-0, since this model has previously been used by others to evaluate several IDO1 inhibitors, including epacadostat (25). The effect of treatment with NTRC 3748-0 ( Table 1), NTRC 3883-0 and epacadostat was evaluated in two CT26 mouse model experiments performed at different contract research organizations ( Figure S3). While the first study reported inhibition of tumor growth with both NTRC 3748-0 and epacadostat ( Figure S3A), no inhibition of tumor growth was found in the second study with either NTRC 3883-0 or epacadostat ( Figure S3B). The tumors from both studies were harvested and analyzed for mIDO1 gene expression by qPCR and Kyn levels by liquid chromatography-tandem mass spectrometry (LC-MS/MS). In both studies, mIDO1 gene expression was detected in most, but not all CT26-derived tumors, while the expression levels were low and varied considerably among the mice ( Figures S3C, D). Nonetheless, in both studies a similar, considerable reduction of Kyn levels was found in the tumor ( Figures S3E, F), indicating target modulation regardless of tumor growth inhibition.
With the aim to establish a mouse model with stable mIDO1 gene expression, we next evaluated the use of an mIDO1overexpressing cell line for the generation of a mouse model. We therefore generated stable sublines of the B16F10 melanoma cell line overexpressing mouse IDO1 (B16F10-mIDO1) ( Figure  3). Transfection of full-length mIDO1 cDNA followed by subcloning resulted in two stable sublines with different mIDO1 mRNA and protein levels ( Figures 3A, B). The Trpcatabolizing activity of the sublines could be inhibited with NTRC 3883-0 and epacadostat ( Table 2), while the selective TDO inhibitor NTRC 3531-0 was inactive (i.e., IC 50 > 31.6 µM). Whereas NTRC 3883-0 and epacadostat inhibited mIDO1 activity in a biochemical assay with similar potency ( Table 2), NTRC 3883-0 was 5 times more potent compared to epacadostat in the cell-based B16F10-mIDO1 assay ( Table 2).
To determine the effect of mIDO1 expression on tumor growth, the parental B16F10 cell line and the two stable mIDO1-overexpressing sublines were implanted into female B6D2F1 syngeneic mice. The tumors grew rapidly, but there was no difference in growth rate between tumors derived from the parental cell line and the mIDO1-overexpressing sublines ( Figure 3C). After sacrifice of the mice, the tumors were collected and analyzed for mIDO1 gene expression by qPCR, while the Trp and Kyn levels were measured by LC-MS/MS. mIDO1 expression was maintained in all tumors collected from the B16F10-mIDO1 sublines i6 and j19, with the mRNA levels in the tumors deviating on average only 2.0-and 2.5-fold from their respective cell lines in culture (Figures 3A, D). mIDO1 expression resulted in strongly reduced Trp and increased Kyn levels ( Figures 3E, F), indicating that the model could be suitable for pharmacodynamic studies of IDO1 inhibitors. The subline j19 was chosen for further experiments, as this subline showed the least variation in mRNA levels ( Figure 3D) and induced on average the highest increase in kynurenine levels in the tumors ( Figure 3F). plasma Trp and higher plasma Kyn levels compared to nontumor bearing animals ( Figures 4B, C). This indicates that there is a direct effect of mIDO1 expression in the tumors on systemic Trp and Kyn concentrations. The effect was diminished upon IDO1 inhibitor treatment ( Figures 4B, C), indicating effective IDO1 inhibition. Treatment with NTRC 3883-0 or epacadostat had no effect on tumor growth in comparison to vehicle-treated animals ( Figure 4D), although a delay in tumor growth could be observed when correcting the tumor growth data for the variation in time until tumor establishment ( Figure 4E). After sacrifice of the mice, qPCR analysis revealed that the mIDO1 expression in the tumors was remarkably stable ( Figure 4F). Comparison of inhibitor concentrations revealed that NTRC 3883-0 reached on average 1.8 and 2.2 times lower levels in plasma and the tumors, respectively, compared to epacadostat ( Figure 4G). IDO1 inhibitor treatment resulted in significantly increased intratumoral Trp levels and reduced Kyn levels ( Figures 4H, I). Treatment with NTRC 3883-0 resulted in 3.6 times lower levels of Kyn compared to epacadostat, while it was also slightly more efficient in increasing Trp (Figures 4H, I). We conclude that NTRC 3883-0 can modulate IDO1 activity in vivo, thus showing target engagement.

Modulation of IDO1 Activity in Primary Ovarian Cancer Cells
We next extended our studies to primary, ascites-derived cells of ovarian cancer patients. In ovarian cancer, IDO1 has been related to disease progression (30), chemotherapy resistance (9) and impaired survival (9,31), based on expression analysis of IDO1 by in situ hybridization and immunohistochemistry of tumor specimens (9,30,31). Patients with advanced ovarian cancer often present with high volumes of malignant ascites, which is routinely collected for diagnostic purposes or relieve of symptoms. Ascites is a potential source for biomarkers to monitor disease progression and to determine chemotherapy response ex vivo.
Adherent cell samples were isolated from the ascites collected from nineteen ovarian cancer patients with different histologies ( Table S3). The presence of tumor cells in these samples was confirmed by qPCR analysis of the ovarian cancer marker genes MUC16 and HE4 ( Figure S4), while five of the nineteen cell samples were additionally analyzed for cell surface expression of mucin-16 and epithelial cell adhesion molecule (EpCAM) by flow cytometry ( Figure S5; Table S4). Basal IDO1 gene expression could be detected by qPCR in fifteen of the nineteen samples, while basal TDO2 expression was found in all nineteen samples ( Figure 5A).
Five of the nineteen samples were further analyzed for IDO1 and TDO2 gene expression, IDO1 protein levels, and Trpcatabolizing activity both in the presence and absence of IFNg, a known inducer of IDO1 gene expression (Figures 5B-D). qPCR analysis revealed that four out of the five samples showed comparable, relatively low basal IDO1 mRNA levels, which were highly increased (i.e., on average 120,000-fold) upon stimulation with IFNg ( Figure 5B). In contrast, 40-fold higher basal IDO1 expression was found in ASC 009, which was increased by 360fold upon IFNg stimulation ( Figure 5B). Immunoblot analysis of IDO1 revealed a consistent pattern ( Figure 5C). Analysis of the TDO2 mRNA levels showed clear basal TDO2 expression in four samples, albeit with slightly more variation among the samples compared to IDO1, while no TDO2 expression was detected in ASC 009 either with or without stimulation ( Figure 5B). Notably, TDO2 expression was found to be consistently downregulated 7-to 30-fold upon stimulation with IFNg ( Figure 5B).  Table S5.
Assessment of the Trp-catabolizing activity in the samples showed basal activity in four out of the five samples ( Figure 5D; + Trp/− IFNg). Treatment with the TDO-selective inhibitor NTRC 3531-0 resulted in a concentration-dependent inhibition of this activity with an average IC 50 of 300 nM ( Figure 5E; Table S5). In contrast, NTRC 3883-0 and epacadostat showed no or only minimal inhibition (Table S5). This indicates that the Trp-catabolizing activity in the absence of stimulation can be solely attributed to the activity of TDO, and not IDO1, despite the constitutive expression of IDO1 observed in most samples ( Figures 5A, B). Stimulation of the ascites cell samples with IFNg induced a further increase in Trp-catabolizing activity ( Figure 5D; + Trp/+ IFNg), with a consistent pattern as that observed for IDO1 mRNA and protein levels. In the stimulated samples, treatment with either NTRC 3883-0 or epacadostat resulted in potent inhibition of the Trpcatabolizing activity with IC 50 values of respectively 261 nM and 8.3 nM ( Figure 5E; Table S5), which are in close agreement with the potencies found in the A375-and HEK-hIDO1-based assays ( Table 2). Moreover, the Trp-catabolizing activity could not be inhibited by the TDO inhibitor NTRC 3531-0 (i.e., IC 50 > 31.6 mM). This demonstrates that the IFNg-dependent Trp-catabolizing activity of the primary patient-derived cancer cells can be attributed to IDO1 activity, and can be selectively inhibited by NTRC 3883-0.

DISCUSSION
IDO1 is a key regulator of the immune modulatory activity of Trp and is therefore a candidate drug target to increase the efficacy of checkpoint inhibitor therapy. We have described the pharmacological characterization of a novel, selective IDO1 inhibitor, NTRC 3883-0, which was profiled in various biochemical and cell-based assays alongside epacadostat. In a co-culture assay with healthy donor lymphocytes, our data confirm that IDO1 expression suppresses cytotoxic T cell proliferation in vitro by decreasing the Trp concentration. Treatment with an IDO1 inhibitor (i.e., NTRC 3883-0 or epacadostat) abolished this inhibitory effect.
In search of a mouse model to additionally profile the in vivo efficacy of NTRC 3883-0, we initially chose the frequently used CT26 mouse model, which is described to endogenously express IDO1 when grafted in mice (25). However, this model was found to be unreliable based on unreproducible tumor growth inhibition results among our two similar studies. Moreover, strong variation in the intratumoral IDO1 mRNA levels was found, with some mice having undetectably low expression. To abolish the potential effect of the low and strongly variable IDO1 expression on the reliability of the efficacy model, we developed a syngeneic mouse model using B16F10 melanoma cells stably overexpressing mIDO1. While we have demonstrated strong target modulation in this model upon treatment with NTRC 3883-0, we did not observe an effect of IDO1 expression or IDO1 inhibitor treatment on the tumor growth rate. This is in contrast with the results obtained by Holmgaard and coworkers, who describe the first use of this model, and who describe an increased tumor growth rate after IDO1 overexpression, which is reduced by IDO1 inhibitor treatment (27). The reason for this discrepancy is unclear, but may be related to subtle differences in the mIDO1-overexpressing or parental cell lines, the mouse strains used or the breeding conditions. Nonetheless, based on the demonstrated target modulation observed for both NTRC 3883-0 and epacadostat, the B16F10-mIDO1 model has proven suitable for application in pharmacodynamic studies of IDO1 inhibitors.
When our studies were still in progress, negative clinical results were reported on several IDO1 inhibitors, including epacadostat (17), EOS200271/PF-0684003 (32), and navoximod (NLG-919) (33). Several experts have commented on these results, in particular on those obtained in the phase III combination trial of epacadostat with pembrolizumab (34)(35)(36). Among the potential causes mentioned for the negative results, the dosing strategy was prominently discussed, since the dose of epacadostat may have been too low to obtain sufficient target coverage in the ECHO-301/KEYNOTE-252 clinical trial. The lack of a good patient stratification strategy was additionally criticized, since patients included in this study had not been selected on the basis of IDO1 expression, although tumor IDO1 gene expression was assessed as part of the study (17). A third possibility raised is that TDO may compensate for inhibition of IDO1, and it has been suggested that dual IDO1/ TDO inhibitors may result in more effective anti-tumor immunity (37).
Learning from these lessons, NTRC 3883-0 was not pursued for further clinical development. This was decided based on allometric scaling indicating that NTRC 3883-0 is unlikely to reach sufficient target coverage of human IDO1 at acceptable dose levels in cancer patients ( Figure S2; Table S2). However, because of its favorable inhibitory potency on mouse IDO1, its better selectivity in biochemical assays, and its higher degree of in vivo target modulation compared to epacadostat, NTRC 3883-0 may become an important tool compound to study the role of IDO1 in mouse disease models. In addition to cancer models, this may support research on neurodegenerative and infectious diseases, for which a role of IDO1 has also been implicated (38,39).
In the phase III clinical trial of epacadostat, IDO1 gene expression in tumor samples was quantified by in situ hybridization (17). Our studies with primary ovarian cancer cell cultures isolated from ascites indicate that determination of basal IDO1 gene expression may not be sufficient. We found that only after stimulation with IFNg, modulation of Trp-catabolizing activity by IDO1 inhibitors could be observed. We propose that functional cell-based assays for IDO1 are included as diagnostic tools to select patients most likely to respond to IDO1 inhibitor therapy. Our research demonstrates for the first time the use of malignant ascites as a source of biomarkers for clinical research on IDO1. The use of cells from ascites has a major advantage over the use of surgical biopsies or organoid cultures (40). The collection of ascites is minimally invasive and assays to determine the activity of compounds can be performed ex vivo prior to the start of therapy.
Notably, we could detect TDO gene expression in all nineteen primary ovarian cancer cell cultures examined. By making use of a selective inhibitor, we demonstrated basal TDO activity in four cultures. Thus far, only a few cancer cell lines that express TDO constitutively are known, while many cell lines express IDO1 upon stimulation with IFNg (8,41). TDO has been implicated in cancer, but its role is not well understood. Based on its high K M,Trp of 190 µM, it is unlikely that TDO contributes to the regulation of the low Trp concentrations that induce T cell anergy, which are around 1 µM and lower (6,24). However, TDO may regulate immune function through the metabolite Kyn, which is a ligand of the aryl hydrocarbon receptor (42). Furthermore, roles of TDO in tumor metastasis (43) and angiogenesis (44) have been described. Because of the different roles of IDO1 and TDO in cancer, it is unlikely that TDO expression may have compensated for IDO1 inhibition in the ECHO-301/KEYNOTE-252 trial. Therefore, there is no strong rationale for the development of dual IDO1/TDO inhibitors, since the possible beneficial effects of dual inhibition have to be balanced with the risk of interfering with the role of TDO in Trp homeostasis.
While many companies have de-prioritized their IDO1 programs, the development of the selective and potent IDO1 inhibitor linrodostat has continued. Linrodostat has demonstrated safety and good target coverage upon once-daily oral dosing in a phase I/IIa study (45), and is currently investigated in a Phase III combination trial with the anti-PD1 immunotherapeutic nivolumab and chemotherapy for advanced bladder cancer (46). Preliminary clinical response data hold promise for IDO1 as a valid drug target for immunotherapy (46). A positive outcome will undoubtedly fuel interest in the development of additional IDO1 inhibitors with differentiating properties. The novel IDO1 inhibitor scaffold described here as well as our in vitro and in vivo pharmacological models may support these R&D activities.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The animal study was reviewed and approved by Ethics Committee for Animal Experimentation at Charles River Laboratories Morrisville, NC, and Ethics Committee for Animal Experimentation at ProQinase Freiburg, Germany.