The Indoleamine-2,3-Dioxygenase (IDO) Inhibitor 1-Methyl-D-tryptophan Upregulates IDO1 in Human Cancer Cells

1-methyl-D-tryptophan (1-D-MT) is currently being used in clinical trials in patients with relapsed or refractory solid tumors with the aim of inhibiting indoleamine-2,3-dioxygenase (IDO)-mediated tumor immune escape. IDO is expressed in tumors and tumor-draining lymph nodes and degrades tryptophan (trp) to create an immunsuppressive micromilieu both by depleting trp and by accumulating immunosuppressive metabolites of the kynurenine (kyn) pathway. Here we show that proliferation of alloreactive T-cells cocultured with IDO1-positive human cancer cells paradoxically was inhibited by 1-D-MT. Surprisingly incubation with 1-D-MT increased kyn production of human cancer cells. Cell-free assays revealed that 1-D-MT did not alter IDO1 enzymatic activity. Instead, 1-D-MT induced IDO1 mRNA and protein expression through pathways involving p38 MAPK and JNK signalling. Treatment of cancer patients with 1-D-MT has transcriptional effects that may promote rather than suppress anti-tumor immune escape by increasing IDO1 in the cancer cells. These off-target effects should be carefully analyzed in the ongoing clinical trials with 1-D-MT.


Introduction
In recent years tryptophan (trp) degradation has received increasing attention as a potent immunosuppressive mechanism involved in the maintenance of immunological tolerance. The trpdegrading enzyme indoleamine-2,3-dioxygenase (IDO) has been implicated in maternal tolerance towards allogeneic concepti [1], controlling autoimmune diseases [2,3] and chronic infection [4], as well as promoting tumor immune escape [5,6,7]. IDO-mediated trp degradation is not restricted to tumor cells [7] but is also detected in tumor-draining lymph nodes [8]. In both tumordraining lymph nodes and tumors, IDO1 creates local tolerance by directly suppressing T-cell responses and enhancing immunosuppression mediated by regulatory T cells (Treg) [6]. IDO is chronically activated in many cancer patients [9] and its expression or enzyme activity correlates with a poor prognosis in patients with various cancers such as ovarian carcinoma [10,11], endometrial carcinoma [12,13], hepatocellular carcinoma [14] and colorectal carcinoma [15].
Despite the bulk of evidence supporting a role for IDO in promoting tumor formation and tumor immune escape, there have been studies showing an anti-tumor activity of IDO1. The induction of IDO1 has been described as a mechanism by which Interferon (IFN)-c inhibits proliferation of malignant cells [16,17]. Some animal experiments demonstrated that IDO1 expression was positively associated with the elimination of malignant cells [18,19]. These findings were corroborated in clinical studies showing that despite being a strong inducer of IDO1, IFN-c was effective in the therapy of ovarian carcinoma and bladder cancer [20,21,22]. In addition, IDO1 expression in hepatocellular carcinoma specimens and in endothelial cells of renal cell carcinoma positively correlated with progression-free survival and long-term survival, respectively [23,24]. There thus remains uncertainty about the clinical relevance of IDO1 expression in tumors.
In preclinical studies the IDO-inhibitor 1-methyl-tryptophan (1-MT) reduced the tumor volume of mice preimmunized with a tumor antigen [7] and -in combination with chemotherapeutic agents -caused regression of established murine breast cancers [5]. Inhibition of IDO in combination with chemotherapy or as a vaccine adjuvant therefore represents an attractive approach for cancer immunotherapy [5,6,7,25,26]. Recently a novel IDO isoform, termed IDO2 was discovered, which -like IDO1 -is expressed in tumors and tumor-draining lymph nodes [27]. The third trp-degrading enzyme in humans, tryptophan-2,3-dioxygenase (TDO) is mainly expressed in the liver and regulates trp concentrations after nutritional trp uptake. The IDO inhibitor 1-MT exists as two stereoisomers, 1-D-MT and 1-L-MT. Most preclinical studies have employed the racemic mixture 1-D/L-MT to inhibit IDO. Recent studies have shown that IDO1 is the preferential target of 1-L-MT, while 1-D-MT preferentially inhibits IDO2 [26,27,28,29]. 1-D-MT is currently used in phase I clinical trials as an adjunct to conventional chemotherapy based on preclinical studies in mouse models of cancer. We were interested in the immunomodulatory effects of 1-D-MT in IDO1positive human cancer cells.

1-D-MT induces immunosuppression of human cancer cells
SKOV-3 cells constitutively degrade trp to kyn and express high levels of IDO1 mRNA while IDO2 and TDO mRNA are expressed at low levels ( Fig. 1A). Knockdown of IDO1 by siRNA in SKOV-3 cells decreased IDO1 mRNA expression by 87.5% (Fig. 1B) leading to a strong reduction in IDO1 protein expression as evidenced by Western Blot (Fig. 1C) and immunocytochemistry (Fig. 1D). Finally, kyn production was inhibited by 91.4% in the IDO1 knockdown cells in comparison to the mean kyn production of SKOV-3 cells transfected without siRNA or with a nontargeting siRNA control (Fig. 1E), suggesting that IDO1 is mainly responsible for the constitutive kyn production in SKOV-3 cells.
To determine the effect of 1-MT treatment on the immunomodulatory phenotype of cancer cells, SKOV-3/MLR coculture experiments were performed. Addition of kyn ( Fig. 2A) or the presence of SKOV-3 cells (Fig. 2B) inhibited alloreactive T cell proliferation in MLR. Knockdown of IDO1 by siRNA not only reversed the SKOV-3 cell mediated suppression of T cell proliferation, but even increased T cell proliferation (Fig. 2C), probably due to additional allogeneic stimulation of the T cells by the IDO-deficient SKOV-3 cells. Next we tested the effects of the two stereoisomers of 1-MT. Addition of 1-methyl-L-tryptophan (1-L-MT) also reversed the suppression of T cell proliferation in the SKOV-3/MLR coculture experiments (Fig. 2D). Surprisingly, T cell proliferation was not enhanced but inhibited in cocultures treated with 1-methyl-D-tryptophan (1-D-MT, Fig. 2E). Addition of trp did not alter this inhibition, indicating that trp depletion is not involved in this paradoxical effect of 1-D-MT (Fig. 2F). Next we analyzed the effect of 1-D-MT on the cell cycle progression and proliferation of SKOV-3 cells, as an inhibitory effect of 1-D-MT on SKOV-3 cells could explain the reduced 3 H thymidine uptake in the coculture experiments (Fig. 3). However, 1-D-MT altered neither 3 H thymidine uptake (Fig. 3A) nor cell cycle progression of SKOV-3 cells (Fig. 3B). To rule out, that 1-D-MT might have inhibited 3 H thymidine uptake of SKOV-3 cells only when these were cultured in the MLR, T cell proliferation in cocultures of SKOV-3 cells with MLR in the presence of different concentrations of 1-D-MT was measured by CFSE staining and flow cytometry (Fig. 3C). 1-D-MT concentration-dependently inhibited T cell proliferation also in these assays (Fig. 3C), thus confirming that 1-D-MT inhibits T cell proliferation and not the proliferation of SKOV-3 cells in the cocultures.

1-D-MT increases kyn production in human cancer cells
We then tested the effect of 1-MT on the kyn production of SKOV-3 cells. Surprisingly, 1-D-MT concentration-dependently increased kyn formation (Fig. 4A), while its stereoisomer 1-L-MT inhibited kyn formation as expected (Fig. 4A). The racemic mixture of 1-MT, which has been used in many studies, including those that have established IDO1 as an immunosuppressive enzyme, inhibited kyn formation, albeit less than 1-L-MT alone (Fig. 4A). As trp concentrations in the media may have limited the increase in kyn production, we also measured the kyn concentrations produced by SKOV-3 cells in response to 1-D-MT in the presence of increasing trp concentrations. Under these conditions much higher kyn concentrations were reached (Fig. 4B), suggesting that the plateau observed above concentrations of about 250 mM 1-D-MT (Fig. 4A) was due to limited trp availability. Trp concentrations in cell culture media usually vary between 12 and 20 mM while concentrations in human serum range between 50 and 70 mM. Kyn formation in cells treated with 1-D-MT was more pronounced when trp concentrations present in human serum (62.5 mM) rather than trp concentrations present in the cell culture media (15 mM) were used (Fig. 4C). However, the fold increase in kyn by addition of trp was equal in cells treated with or without 1-D-MT (Fig. 4C). To further test whether 1-D-MT directly influences IDO1 enzymatic activity we measured IDO1-mediated kyn production in SKOV-3 cell extracts. 1-D-MT did not alter kyn formation regardless whether trp was present at a fixed concentration of 100 mM (Fig. 4D) or at concentrations equimolar to 1-D-MT (Fig. 4E), suggesting that the increase in kyn formation by 1-D-MT in SKOV-3 is not mediated by a direct effect of 1-D-MT on IDO1 enzymatic activity.

IDO1 expression is enhanced by 1-D-MT in human cancer cells
Next, we investigated whether 1-D-MT influenced the expression of trp-metabolizing enzymes. To our surprise, we found that 1-D-MT increased IDO1 mRNA and protein in SKOV-3 cells, while IDO2 and TDO remained unaltered (Fig. 5A). Upregulation of IDO1 mRNA was concentration-dependent ( Fig. 5B) and was first detected after 16 h of incubation with 1-D-MT (Fig. 5C). Importantly, the IDO1-promoting effects were not restricted to SKOV-3 cells. While 1-D-MT did not induce de novo IDO1 mRNA expression and kyn production in IDO1-negative HeLa cervical carcinoma cells, it increased IDO1 mRNA and kyn production after induction of IDO1 expression and kyn production by IFN-c (Fig. 6A,B). Interestingly, the 1-D-MT-mediated upregulation of IDO1 mRNA in many IDO1-negative cancer cells was differentially dependent of the concentration of IFN-c that was used to induce de novo expression of IDO1 (

1-D-MT-mediated IDO1 expression involves JNK and p38 MAPK
We then explored signalling pathways involved in the upregulation of IDO1 in response to 1-D-MT treatment. IFNmediated STAT1 phosphorylation is involved in the induction of IDO1 in many different cells and tissues [30], but knockdown of STAT1 by siRNA did not decrease the kyn production of 1-D-MT-treated cells (Fig. 7A). In line with this result, 1-D-MT did not induce the mRNA expression of IFN-b or IFN-c (Fig. 7B). Mitogen activated protein kinase (MAPK) pathways have been reported to be modulated by the racemic mixture of 1-MT and thereby influence the polarization of dendritic cells (DC) [31]. We therefore tested whether inhibition of of MAPK signalling affected the 1-D-MT-mediated increase in IDO1 expression. Inhibition of ERK phosphorylation by PD98059 affected IDO1 mRNA expression and kyn release neither in untreated nor in 1-D-MT treated cells (Fig. 8A). While inhibition of p38-MAPK phosphorylation by SB203580 [32] slightly reduced IDO1 mRNA in untreated cells, it almost completely mitigated the increase in IDO1 mRNA expression in response to 1-D-MT (Fig. 8B). The slight inhibition of IDO1 transcipt in untreated cells did not translate into significantly reduced kyn release, while the reduction in kyn release became significant in 1-D-MT treated cells (Fig. 8B). Similar results were obtained when inhibiting JNK by SP600125 (Fig. 8C) [33]. Collectively, these data suggest, that p38 MAPK and JNK signalling are involved in mediating the induction of IDO1 in response to 1-D-MT.

Discussion
In the past IDO inhibition was mostly achieved using the racemic mixture of 1-MT [34]. As it became apparent that IDO inhibition may be a promising target for cancer therapy, the individual stereoisomers of 1-MT were investigated in more detail [5,35]. Although 1-L-MT was shown to more effectively inhibit IDO1 in enzyme assays and in cancer cell lines, 1-D-MT showed superior anti-tumor activity in mouse models and was therefore chosen for clinical trials [35]. A subsequent study suggested that the superior anti-tumor activity of 1-D-MT may result from inhibition of the IDO2 isoform [27]. Recent studies indicated that 1-D-MT inhibits IDO activity neither in dendritic cells nor in tumor cells [26,28] and does not effectively restore IDO-induced arrest of T-cell proliferation [36]. In our study, 1-D-MT suppressed T cell proliferation when constitutively IDO1-expressing SKOV-3 cells were cocultured with mixed lymphocyte reactions (Fig. 2E,F). Investigation of the underlying mechanisms surprisingly revealed that 1-D-MT increased the kyn production of cancer cells with IDO1 activity (Fig. 4, 6) due to upregulation of IDO1 mRNA and protein expression (Fig. 5, 6). The upregulation of IDO1 expression and activity was observed only in cancer cells with either constitutive or IFN-c-induced IDO1 expression (Fig. 5, 6). Upregulation of IDO1 by 1-D-MT in many cancer cells was most prominent at moderate concentrations of IFN-c that are likely to resemble physiological concentrations (Fig. 6C). Low concentrations of IFN-c may not have induced sufficient IDO1 expression (Fig. 6C), possibly explaining why 1-D-MT did not increase IDO1 at these concentrations, while stimulation with very high concentrations of IFN-c may have resulted in maximal IDO1 induction (Fig. 6C), thus explaining why further increase by 1-D-MT was not observed. In all of the examined IDO1-expressing cells an increase in IDO1 expression and kyn release was observed in response to treatment with 1-D-MT, indicating a general mechanism involved in 1-D-MT-mediated IDO1 induction (Fig. 4, 5, 6).
In a previous study, the racemic mixture of 1-MT has been reported to modify the polarization of dendritic cells (DC) by modulating MAPK [31]. Inhibition of p38 MAPK phosphorylation prevented the increase in IDO1 mRNA and kyn production by 1-D-MT (Fig. 8B), suggesting that p38-MAPK participates in 1-D-MT mediated signalling. In line with this result, p38 MAPK has previously been shown to contribute to the induction of IDO1 in a leukaemia cell line and in DC [37,38]. Also inhibition of JNK signalling mitigated the induction of IDO1 mRNA and kyn release in the presence of 1-D-MT (Fig. 8C). Inhibition of JNK phosphorylation has recently been described to decrease IDO1 expression induced by LPS in mouse microglia [39]. 1-D-MT mediated modulation of p38-MAPK and JNK suggests that 1-D-MT may have more effects than just the upregulation of IDO1.
To our knowledge this is the first report of 1-D-MT-mediated effects on gene expression in human cells. The stereoisomer of 1-D-MT, 1-L-MT was recently reported to suppress the IFN-c-induced expression of IDO1 in mouse rectal carcinoma cells [40]. In addition, it has previously been described that 1-MT influences the maturation of DC independently of IDO [31]. However the racemic mixture of 1-MT was used in this study and it is therefore not known which stereoisomer was responsible for the observed effects [31]. It remains to be elucidated whether 1-D-MT exerts more effects on gene expression than the regulation of IDO1. In contrast to IDO1, the trp-catabolising enzyme IDO2 was not induced by 1-D-MT. This finding underlines the notion that IDO1 and IDO2 are differentially regulated on a transcriptional level [27]. Possible other effects of 1-D-MT on gene expression may contribute to the high anti-tumor efficacy of 1-D-MT observed in mouse tumor models [35]. Significant differences in IDO expression and regulation exist between humans and mice [29,41], which could account for the observed discrepancies regarding 1-D-MT action in mouse models and human cells. In a study using 1-D-MT in SIV-infected rhesus macaques, a model more closely resembling humans than mouse models, Boasso and colleagues observed that kyn plasma levels were not reduced but rather induced during treatment with 1-D-MT [42]. Increased IDO1 mRNA expression in lymph nodes of macaques after 1-D-MT treatment was interpreted as a compensatory counterregulatory mechanism activated by 1-D-MT, which may have accounted for the lack of effect on plasma kyn [42]. Our data show that upregulation of IDO1 and kyn take place also in isolated human cells, in which a counterregulatory mechanism that mediates immunosuppression is unlikely.
As there is evidence that IDO1 may restrict tumor growth as a mediator of tumoricidal IFN-c in experimental models and patients [16,17,18,19,20,21,22] and as IDO1 expression in tumors positively correlated with progression-free survival and long-term survival in some studies [23,24] it is tempting to speculate that the induction of IDO1 by 1-D-MT may actually account for some of the anti-tumor effects of 1-D-MT.
In conclusion, we have identified the upregulation of IDO1 expression in human cancer cells as a profound effect of 1-D-MT, a compound currently used in clinical studies in patients with relapsed or refractory solid tumors with the aim of inhibiting (IDO)-mediated tumor immune escape. IDO1 expression is known to be immunosuppressive and may enhance tumor immune escape, but it has also been implicated in direct anti-tumor effects. More studies are needed to better understand the role of IDO in cancer biology and the potential use of 1-D-MT as an anti-cancer agent.

High performance liquid chromatography (HPLC)
High performance liquid chromatography (HPLC) analysis was performed according to [44] using a Beckman HPLC with photodiode array (PDA) detection and Lichrosorb RP-18 column (250 mm64 mm ID, 5 mm, Merck, Darmstadt, Germany). Kyn release and trp degradation were measured in the medium of 3 * 10 5 cells in 2 mL McCOY's 5A Medium (BioConcept) containing 10% FBS (Perbio), 100 U/mL penicillin and 100 mg/mL streptomycin (PAA Laboratories) supplemented with 0-125 mM L-trp (Sigma-Aldrich). The medium was harvested from 6 well plates at the indicated time points, centrifuged and frozen until further analysis. After thawing, the samples were supplemented with trichloroacetic acid for protein precipitation, centrifuged and 100 ml of the supernatant was analyzed by HPLC. Standard curves were generated with L-kyn and L-trp (Sigma-Aldrich) in the same medium. Since FBS contains kyn, low kyn concentrations (,1 mM) were detected in all samples and medium without cells was always measured for comparison.

Quantitative (q)RT-PCR
Total RNA was isolated with the Qiagen RNAeasy RNA isolation kit (Hilden, Germany) and DNA was synthesized with the Applied Biosystems reverse-transcription-Kit (Foster City, CA, USA) according to manufacturer's instructions. QRT-PCR was preformed in an ABI 7000 thermal cycler with SYBR Green PCR Mastermix (Applied Biosystems) according to standard protocols.  ON-TARGETplus siCONTROL Non-targeting Pool (D-001810-10-05, Dharmacon) and a transfection without siRNA were used as negative controls.
For transfection of siRNA, the Amaxa Nucleofector Kit V (Amaxa biosystems, Koeln, Germany) was used. Briefly, 3 * 10 5 cells were resuspended in 100 ml of the nucleofector solution V and mixed with 1.5 mg of siRNA, then electroporated using program V005. Medium was changed after 24 h, analysis of the kyn content of the medium by HPLC, harvesting of the cells for RNA extraction or the generation of lysates and immunocytochemical analysis were performed after 48 h.

Cell cycle analysis
Cells were incubated with 10 mM BrdU for 1 h, harvested and then fixed with 70% (v/v) methanol (VWR). To uncover the DNA, cells were treated with a PBS -based buffer (PAA) containing 0.1 M hydrochloric acid (VWR) and 0.3% Triton X-100 (AppliChem) and boiled in water. Finally, cells were incubated with Alexa-Fluor 647 mouse anti BrdU antibody (clone 3D4, BD Biosciences, Heidelberg, Germany). 20 mg/mL DAPI (Sigma-Aldrich) was added 1 minute before analysis by flow cytometry (FACSCantoII, BD Biosciences).  and were used for normalization. Proliferation was analysed using FlowJo Flow Cytometry Analysis Software (Treestar Inc. Ashland, OR, USA).

Statistical analysis
Data are expressed as mean 6 SEM. Experiments were repeated at least three times with similar results. Analysis of significance was performed using the Student's t-test (SigmaPlot, Systat Software Inc., San Jose, CA, USA). P values,0.05 were considered significant *.