Binding properties of different categories of IDO1 inhibitors: a microscale thermophoresis study.

AIM
Inhibition of IDO1 is a strategy pursued in the immune-oncology pipeline for the development of novel anticancer therapies. At odds with an ever-increasing number of inhibitors being disclosed in the literature and patent applications, only very few compounds have hitherto advanced in clinical settings.


MATERIALS & METHODS
We have used MicroScale Thermophoresis analysis and docking calculations to assess on a quantitative basis the binding properties of distinct categories of inhibitors to IDO1.


RESULTS
Results shed further light on hidden molecular aspects governing the recognition by the enzyme of compounds with different mechanism of inhibition.


CONCLUSION
Results pinpoint specific binding features of distinct inhibitors to IDO1 that offer clues for the design of next-generation inhibitors of the enzyme.

In the last 40 years, L-Tryptophan (L-Trp) catabolism has gained attention because of its importance in immunity [1,2]. The 95% of dietary L-Trp (1, Figure 1) is catabolized through the kynurenine pathway (KP), with IDO1 being a heme-containing enzyme that catalyzes the first and rate-limiting step of the pathway to produce N-formyl-kynurenine (2) [3]. Many downstream products of KP have been identified as biologically active compounds (3)(4)(5), showing the capability to regulate inflammatory responses through several mechanisms [4]. An increasing body of evidence has thus thrust IDO1 under the spotlight as a key druggable target to control the balance between immune-tolerance and immune-inflammatory responses [5,6]. Noteworthy, signaling functions of IDO1 have also been discovered [7,8]. Specifically, phosphorylation on immune-receptor tyrosinebased inhibition motifs (ITIMs) of IDO1 bestows specific properties to the enzyme enabling its interaction with protein partners such as SHP-1 and -2 or SOCS-3 [9,10]. Although the former case leads to prolongation of IDO1's half-life and long-term immune-tolerance; in the latter case, IDO1 is tagged for proteasome degradation eventually leading to inflammatory response. These autoregulatory signaling functions add on the effects arising from the catalytic activity of the enzyme that are mediated by consumption of L-Trp (1) and/or production of bioactive KP metabolites (3)(4)(5) [11][12][13].
The breakthrough discovery that IDO1 plays a pivotal role in cancer immune-editing process has fostered an intense research activity aimed at developing inhibitors of the enzyme as novel anticancer agents [14][15][16]. IDO1 is indeed overexpressed in many human cancer cells as well as in dendritic cells presenting tumor antigens [17]. Cancer cells exploit IDO1 functions to propel their growth, creating an immune-privileged tolerogenic microenvironment in which cytotoxic T cells are starved by L-Trp  Research Article Greco, Coletti, Custodi et al. consumption and T reg cells are activated by KP metabolites. Inhibition of IDO1 by small molecules hampers cancer cells to evade the immune system, favoring inflammatory response in the tumor microenvironment [18]. In vitro and in vivo pharmacological studies have supported this concept, showing that combination therapy between IDO1 inhibitors and chemotherapeutic drugs synergistically improves the efficacy of blocking tumor growth [19]. Although several chemical classes of IDO1 inhibitors have been reported in the literature and patent applications, only few compounds have progressed in clinical trials [20,21]. The stagnant rate of inhibitors progressing to clinical setting has been in part ascribed to challenges associated with a correct appraisal of IDO1 inhibition [22]. The relevance of a preferential binding of inhibitors to the oxidized inactive form of IDO1 and/or to the reduced active form of the enzyme is also elusive in terms of pharmacological efficacy, cancer cell selectivity and off-target effects against other heme-containing proteins. Substrate inhibition at high concentrations of L-Trp (1) has been reported as potentially interfering with enzymatic assays aimed at investigating IDO1 inhibition kinetics of small molecules [23]. This observation was formerly explained by the presence of an accessory-binding site into IDO1 structure that may also host small-molecule enhancers of the catalytic activity [24][25][26][27]. However, more recent studies have demonstrated that substrate inhibition by L-Trp (1) is likely due to competition with oxygen in binding to heme cofactor [28,29]. Germane to this aspect, many IDO1 inhibitors reported in the literature have been described as endowed with noncompetitive or uncompetitive inhibition kinetics with respect to L-Trp (1) [3,30]. Crystallographic studies have shown that at least some of these compounds bind to the catalytic cleft of the enzyme [31][32][33]. Accordingly, their 'apparent' noncompetitive or uncompetitive profile is due to competition with binding of oxygen to heme, and not to the interaction with an accessory-binding site of the protein. In this framework, it is still questioned the mechanism of action of D-1-methyl-tryptophan (Indoximod, 6, Figure 2) [34][35][36][37], which is currently in clinical trials as bona fide IDO1 inhibitor [38]. Likewise, concerns exist on whether inhibitors bearing quinone or iminoquinone moieties block the catalytic activity of IDO1 by a specific mechanism of action or rather by an unspecific redox-cycling mechanism [39,40].
Thermophoresis is the movement of biomolecular molecules in a temperature gradient and depends on size, charge and hydration shell. Typically upon ligand/ protein interaction, at least one of these parameters will change [41]. Dissociation constants (K d ) can thus be obtained analyzing the change of movement of the unbound molecule compared with the ligandbound complex in a temperature gradient. Accordingly, herein we attempt to investigate binding properties of a selection of IDO1 inhibitors belonging to different categories of mechanism of action, by using MicroScale Thermophoresis (MST) analysis. Docking calculations were also performed to infer about the putative-binding modes of these inhibitors into the crystal structure of IDO1.

MST analysis
Ligands were purchased from different vendors (Selleckchem, Munich, Germany; Sigma Aldrich, Milan, Italy; Molport, Riga, Latvia). Tryptanthrin derivative (12) was synthesized in our laboratory following the synthetic scheme already published in the literature [42]. Recombinant human IDO1 (rhIDO1) was purchased from Proteros. Experiments to assess K d were conducted using Monolith NT.115 (NanoTemper Technologies, Munich, Germany). Fluorescence labeling of rhIDO1 was performed following the protocol for N-hydroxysuccinimide coupling of the dye NT647 (NanoTemper Technologies) to lysine residues. Briefly, 100 μl of a 20 μM solution of rhIDO1 protein in labeling buffer (130 mM NaHCO 3 , 50 mM NaCl, pH 8.2) was mixed with 100 μl of 60 μM NT647-N-hydroxysuccinimide fluorophore (NanoTemper Technologies) and incubated for 30 min at room temperature in the dark. Unbounded fluorophores were removed by sizeexclusion chromatography with MST buffer (50 mM TRIS, 150 mM NaCl, 10 mM MgCl 2 , pH 7.4, 0.05% Tween20) as running buffer. The real concentration of each element of the sample, such as protein, heme group and RED dye and the degree of labeling were determined using extinction coefficient ε280 = 51,380 M -1 cm-1 for rhIDO1, ε405 = 159,000 M -1 cm-1 for rhIDO1 heme group and ε650 = 250,000 M -1 cm-1 for NT647 fluorophore, with a correction factor (F corr ) of 0.028 at 280 nm, using C prot = [A280 -(A280 × Fcorr)/ε280 × l] and the degree of labeling resulted between 0.6 and 0.8 throughout all labeling reactions.
The stability of NT647-rhIDO1 and unmodified rhIDO1 protein was checked using circular dichroism (Supplementary Table S1). Compound predilutions were prepared for MST experiments by 16-fold 1:1 serial dilutions in assay buffer containing 4% DMSO in PCR tubes supplied by NanoTemper Technologies to yield final volumes of 10 μl. NT647-rhIDO1 solution (10 μl) was added to each compound dilution and mixed to reach a final NT647-rhIDO1 concentration of 50 nM, including 2% DMSO, 2-mM DTT and a reaction volume of 20 μl. These samples were loaded into 16 premium-coated capillaries and inserted in the chip tray of the MST instrument (Monolith NT.115) for thermophoresis analysis and the appraisal of K d values. MST signals were recorded at MST 40% (compounds 6, 7, 9, 11, 12, 13). Compounds not providing a binding curve with a good S/N ratio at 40% (8,10) were tested at MST 80%. In both cases, 20% LED power was used. K d values were calculated from compound concentration-dependent changes in normalized fluorescence (F norm ) of NT647-rhIDO1 after 21 s of thermophoresis for compounds tested at MST 40% and after 4 s for compounds tested at MST 80%. Each compound was tested in triplicate and the values reported were generated through the usage of MO Affinity Analysis software (NanoTemper Technologies); (±) K d confidence (SD) are indicated next to K d value. Specifically, confidence (SD) values define the range where the K d falls with a 68% of certainty. Binding curves of protein fraction bound versus ligand concentration (nM) of tested compounds are shown in Supplementary Materials ( Supplementary Figures 1-8).

Molecular docking
Ligands were prepared with LigPrep 3.8 generating all ionization states at pH 7 ± 2. IDO1 crystal structure with the lowest resolution factor (Protein Data Bank [PDB] code: 5EK3, res. = 2.21 Å) was downloaded from the PDB. After deleting chain B, the Protein Preparation Wizard (Maestro 10.6, Schrödinger, Inc., NY, USA) tool was used to prepare chain A adding hydrogen atoms, setting ionization states at pH 7.0 and refining the structure by energy minimization. The oxidation state of the iron atom of the heme group was set to Fe 3+ [31,[43][44]. A grid box was generated with Glide 7.1 (Schrödinger, Inc., NY, USA), locating the center on the center of mass of the cocrystallized ligand (S, R)-11 in 5EK3. The inner grid box was sized 12 × 12 × 12 Å. Docking studies were carried out using Glide 7.1 standard precision mode and storing the top five binding poses for each molecule. The best binding pose for each compound in terms of interaction energy (G-score, kcal/mol) [45][46][47] was selected in order to obtain a putative-binding mode.

Competitive inhibitors
Analogs of L-Trp were among the first competitive inhibitors reported in the literature [48]. These compounds were designed with bioisosteric replacements of the indolic nitrogen or the N-methyl substitution at the indolic ring (1-Methyl-Trp). In particular, the racemic mixture of 1-Methyl-Trp was reported to inhibit IDO with a K i of 34 μM. However, the L-(or future science group Research Article Greco, Coletti, Custodi et al.

S, 7)
isomer proved more potent than the corresponding D-(or R, 6) isomer, with the former showing 63% of inhibition at 100 μM and the latter having 12% of inhibition at 100 μM [49]. Of note, recent biochemical studies have evidenced that L-1-Methyl-Trp (7) is a slow substrate (hIDO1 k cat = 0.027 ± 0.001 s -1 ; K M = 150 ± 11 μM) rather than a competitive inhibitor of the enzyme [50]. This observation suggests that L-1-Methyl-Trp (7) may bind to the oxygen-bound activated state of ferrous IDO1, leading to the production of N-formyl-methylkynurenine.
Despite its lower IDO1 inhibition potency in biochemical assay [49], D-1-Methyl-Trp (6) was found by some authors as endowed with superior in vivo anticancer activity and in vitro T-cell activation efficacy [34]. Although these observations have contributed to advance D-1-Methyl-Trp (6) in Phase I clinical trials as combination therapy with chemotherapeutic drugs for cancer, L-1-Methyl-Trp (7) is still employed as chemical probe to unravel biological functions of IDO1 as well as its therapeutic relevance in immunological and cancer disorders [35][36][37].
To get insights into binding properties of L-1-Methyl-Trp (7) and D-1-Methyl-Trp (6) to hIDO1, we determined K d using MST. At this regard, it should be stressed that these K d , and those reported hereinafter, refer to interactions with the inactive ferric (Fe 3+ ) form of hIDO1, which is the most stable in the adopted experimental conditions, and do not rule out that different K d values for the same ligand may be associated to interactions with other forms of hIDO1, including the reduced ferrous (Fe 2+ ) state and/or oxygen-bound complex of the enzyme. As a result, a low K d was found for L-1-Methyl-Trp (7, K d = 1.38 ± 0.28 μM, Table 1, Supplementary Figure S1), whereas a higher K d was determined for D-1-Methyl-Trp (6, K d = 21.6 ± 2.8 μM, Supplementary Figure S2). The K d value of D-1-Methyl-Trp (6), in particular, suggests for the first time that it may directly bind to the ferric inactive state of the enzyme. Of note, this observation would be in agreement with the evidence that D-1-Methyl-Trp (6) is able to inhibit IDO1 catalytic activity in human and mouse dendritic cells [34]. Conversely, the poor inhibitory activity of D-1-Methyl-Trp (6) in biochemical assay would find tentative explanations in the inability of such ligand to prevent the reductive activation of IDO1 promoted by the ascorbic acid-methylene blue redox system, and/or the lack of binding affinity of D-1-Methyl-Trp (6) toward the reduced ferrous and/or oxygen-bound form of IDO1, with either events being at odds with L-1-Methyl-Trp (7).
Docking studies of these compounds into the inactive ferric form of IDO1 reveal top scored solutions with conserved binding poses. In the case of L-1-Methyl-Trp (7), the alpha amino group and carboxylic moiety make conserved hydrogen bonds with the propionate group of the heme cofactor and the carbonyl moiety of Gly262 ( Figure 3A). The N-methyl indole ring is placed almost orthogonal to the heme plane, in a hydrophobic 'sandwich' composed by Tyr126 and Phe163. A similar binding mode is also observed for D-1-Methyl-Trp ( Figure 3B). Remarkably, the similar docking score values of L-1-Methyl-Trp (7) and D-1-Methyl-Trp (6) into the catalytic cleft of hIDO1 are not able to explain the different K d (Table 1). Although a flaw of the scoring function may well explain such observation, it is also possible that K d of these ligands may be affected not only by interactions with residues of the catalytic pocket but also by interactions with other regions of hIDO1 such as the large unsolved loop located at the entrance of the catalytic site.
Hydroxyamidine-based compounds were also reported in the literature as potent competitive inhibitors of hIDO1, with preferential interactions to the active ferrous (Fe 2+ ) form of the enzyme [51]. Specifically, compounds 8 and 9 yielded IC 50 of 0. 067 μM and 0.55 μM at inhibiting hIDO1 in biochemical assay, respectively. These inhibitors proved slightly more active in HeLa cell-based assays, with cellular IC 50 of 0.019 μM for   Figure S4). K d and IC 50 values are not directly comparable, with the former parameter encoding the binding event and the latter describing the functional activity. These diverging results are likely explained with the limit of the biophysical study in which the reduced catalytically active ferrous (Fe 2+ ) form of the enzyme cannot be stabilized with the addition of reducing agents that would potentially interfere with the MST assay. Hence, MST results indirectly support the preferential interaction of 8 and 9 to the ferrous (Fe 2+ ) form of the enzyme. Although the submicromolar potency of these molecules in biochemical assay is ascribed to their interaction with the active ferrous (Fe 2+ ) form of IDO1 [51], MST results pinpoint that 8 and 9 may also bind to the inactive ferric (Fe 3+ ) form of the enzyme at higher concentrations and in a range of potency that is remarkably similar to the K d of D-1-Methyl-Trp. At this regard, it is worth noting that a high dose (75 mg/kg) of 8 was required to observe significant tumor growth inhibition in C57BL/6 mice bearing GM-CSF-secreting B16 melanoma. While the high dosage of 8 may be dictated by poor pharmacokinetic properties, the above observations raise questions about the potential contribution of the interaction to the inactive ferric form of hIDO1 to the in vivo antitumor efficacy of compound 8 and, more in general, of competitive inhibitors.
Results of docking studies of compounds 8 and 9 into the ferric form of hIDO1 are not in agreement with previous computational studies, which suggested that the hydroxyamidine group coordinates the ferrous heme iron into the catalytic pocket of the enzyme. Although different adopted docking algorithms may explain this discrepancy, it is also possible that hydroxyamidine derivatives may use two alternative binding modes to interact with the active ferrous form of the enzyme or the inactive ferric form of IDO1. In the top scored binding pose of compound 8 ( Figure 4A), the hydroxyamidine group makes a hydrogen bond with the propionate group of the heme cofactor. The phenyl moiety extends toward the entrance of the catalytic pocket, engaging residues Leu230, Phe226, Ile354 and Leu384 with hydrophobic contacts. The furazan moiety is placed into the hydrophobic pocket above the heme plane, being composed of residues Tyr126, Val130, Phe163 and Phe164. Its amino group interacts with the hydroxyl group of Ser167 making a hydrogen bond. In the top scored binding pose of compound 9 ( Figure 4B), however, the phenyl and furazan moieties invert their relative position occupying the one the hydrophobic pocket of the other; inverting the pose, 9 loses the hydrogen bond with Ser167 but it binds to the carbonyl moiety of Gly262 and Leu234. In agreement with their equipotent K d to ferric hIDO1, inhibitors 8 and 9 show similar docking score values in the top scored binding poses (Table 1).

Noncompetitive inhibitors
4-Phenylimidazole (4-PI, 10, Figure 5) is one of the earlier noncompetitive inhibitors of IDO1 disclosed in the literature with a reported IC 50 of 48 μM [52]. Compound 10 was used in crystallographic studies to solve the first ligand-bound crystal structure of the  enzyme (PDB code: 2D0T), evidencing that the distal basic nitrogen atom of the imidazole ring is involved in a coordinative interaction with the sixth coordination site of the ferric heme of IDO1 [33]. This structure was instrumental to design 4-PI analogs and carry out structure-activity relationship studies that led to identify imidazoleisoindole derivatives as a chemical class of potent noncompetitive inhibitors of IDO1 [53]. In this study, we determined K d to hIDO1 for 4-PI (10) and an imidazoleisoindole analog (11) that was recently solved in a crystallographic complex with hIDO1 (PDB code: 5EK3) [31]. MST analysis of compounds 10 and 11 reveals K d of 45.3 ± 11.7 μM and K d of 3.30 ± 0.41 μM to IDO1 (Table 1, Supplementary Figures S5, S6), respectively. At a first glance, these results are in agreement with the rank order of IC 50 potency that is reported in the literature for 4-PI (10, IC 50 = 48 μM) and 11 (IC 50 = 0.1 μM or 0.038 μM) [31,54]. However, it cannot be ruled out that the imidazoleisoindole analog (11) also binds to the reduced active ferrous (Fe 2+ ) state of hIDO1, with this latter interaction accounting for the nanomolar inhibition potency in biochemical assay. A scenario reported in the literature for another noncompetitive inhibitor of IDO1, namely Norharman, which binds to both the ferric and ferrous form of the enzyme with comparable K d (∼10 μM) [52].
Docking studies of the imidazole derivatives 10 and (S, R)-11 are in different agreement with the relative experimental crystal structures (Supplementary Tables S2, S3). Specifically, docking of 4-PI (10) is unable to provide top scored solutions consistent with its experimental crystallographic complex of IDO1 (Root-mean-square deviation [RMSD] = 2.68 Å, Figure 6A). Reason of such a failure has been discussed by us elsewhere and is out of the scope of the present work [55]. Conversely, results of docking of the imidazoleisoindole analog (S, R)-11 into the catalytic cleft of IDO1 are in good agreement with  Figure 6B). In particular, (S, R)-11 places the basic nitrogen atom of the imidazole moiety within a distance (<3 Å) and angle (about 180° considering the iron and the nitrogen of the underlying His346) that are suitable for iron coordination. The hydroxyl group takes contact through H-bond with Gly262 (or the heme propionate group in other less energetic favored solutions), the three fused rings are packed with Tyr126 and Phe163 and the cyclohexyl moiety engages Phe226 with hydrophobic contacts.

Uncompetitive inhibitors
Uncompetitive inhibitors are defined as ligands that bind to the enzyme/substrate complex with the maximal affinity. Although the substrate (L-Trp) was not added to the composition of the MST assay solution due to potential interference with the biophysical assay, we still pursued the exploration of the potential target engagement by MST of these compounds, providing an assessment of their binding properties to the substrate unbound form of IDO1.
Tryptanthrin derivatives were reported as potent uncompetitive inhibitors of IDO1 [42]. Among these, compound 12 proved low micromolar K i (= 0.161 μM) and IC 50 (= 0.534 μM) against rhIDO1, while a more potent inhibitory activity was found in HEK 293 cells (IC 50 = 0.023 μM) expressing human IDO1. Compound 12 was able to inhibit tumor growth in Lewis lung cancer cell-bearing mice, as well as reduced Foxp3 + T reg cells in the same animal model. At odds with its in vitro and in vivo potency, however, compound 12 was found with a poor dissociation constant to IDO1 in a surface plasma resonance study (K d = 46.8 μM) reported by the same authors. However, the MST analysis pinpoints a higher affinity of 12, providing a K d of 0.97 ± 0.07 μM (Table 1, Supplementary Figure S7), namely a value that is more in agreement with the reported biochemical and cellular activity of the compound. These different results of surface plasma resonance and MST are likely due to the fact that the former assay implies the immobilization of the enzyme, whereas the latter is carried out with the free enzyme in solution. As a consequence, the immobilization protocol may have affected the conformational properties of IDO1, leading to assess a poor dissociation constant of the ligand in surface plasma resonance assay.
Despite none of the trypthantrin derivatives has been cocrystallized with IDO1 and no binding mode has been proposed, we have defined its putative-binding mode through its best solution in 5EK3 ( Figure 7A). In particular, the fluorine-substituted phenyl goes into the hydrophobic cleft composed by Tyr126 and Phe163, while the opposite no-substituted phenyl is placed toward the entry channel and takes contact with Phe226 and Ile354.
Mitomycin-C (13) is a natural product that has been used as chemotherapeutic agent for several types of cancer [56][57][58]. It was reported as uncompetitive inhibitor of IDO1 with a K i of approximately 25 μM [24]. According to its mechanism of inhibition, the authors proposed that mitomycin-C (13) binds to the substrate-bound complex of IDO1, interacting with an accessory site that is shaped upon substrate binding to the enzyme. Herein, we assessed the dissociation constant of mitomycin-C (13) to IDO-1 by the MST analysis. Although it was not possible to obtain a saturated-binding curve for compound 13 (Supplementary Figure S8) due to limitation of compound solubility, a binding event was observed with an approximate K d value of 790.6 ± 117.6 μM (Table 1). Again, it should be stressed that such a high micromolar K d value is obtained against the substrate unbound ferric form of the enzyme to have an assessment of target engagement. Accordingly, MST does highlight the presence of a low-affinity accessory site for mitomycin-C (13) in the free enzyme, and it does not rule out that L-Trp binding into the catalytic site of IDO1 may adjust the shape of the accessory site to allow for hosting the large structural features of 13, thereby enabling a stronger inhibition constant with maximal affinity for the natural compound. This observation combines with the presence of a quinone moiety in the chemical structure of mitomycin-C (13). Other natural products bearing quinone groups have been reported as potent IDO1 inhibitors [59,60]. Grounding on this observation, screening of commercially available compounds containing a quinone group resulted in the identification of several potent IDO1 inhibitors [40]. However, the inhibitory activity of these compounds has been debated, questioning whether they adopt a specific mechanism of action to inhibit IDO1 or act through an unspecific redox-cycling mechanism. The inhibitory potency of mitomycin-C (13) in biochemical assay may thus arise from the coupling of specific binding event to IDO1 and unspecific redox-cycling mechanism.
Results of docking studies of mitomycin-C (13) are in accordance with the assessed K d . Indeed, compound 13 shows the worst g-score among the studied compounds. Because of its larger size, mitomycin-C (13) does not fit the catalytic cleft above the heme group, but it is placed toward the entry channel (Pocket B, Figure 7B). The bad score is in line with the uncompetitive mechanism of inhibition of mitomycin-C (13), which indeed suggests maximal interaction of 13 with the enzyme/substrate complex. Hence, a different conformation of IDO1 is required for docking studies. Specifically, the structure of the enzyme with the substrate and/or oxygen bound to the catalytic cleft for which crystallographic data are not still available.

Conclusion & future perspective
In this study, we have used the MST analysis to quantitatively assess the binding properties of distinct categories of inhibitors to IDO1, including competitive, noncompetitive and uncompetitive inhibitors. Germane to competitive inhibitors, results pinpoint specific binding properties of L-1-Methyl-Trp (8) and D-1-Methyl-Trp (7) to the ferric form of IDO1, with the former showing a stronger dissociation than the latter stereoisomer. Although the pharmacological relevance of a specific interaction of inhibitors toward the inactive ferric form of IDO1 and/or the active ferrous form of the enzyme is still elusive, these data support an IDO1-mediated anticancer activity for D-1-Methyl-Trp (7), a bona fide IDO1 inhibitor that is currently in clinical settings as singleagent therapy (Clinical Trials Id.: NCT00739609, NCT00567931, NCT01560923) or in combination therapies (Clinical Trials Id.: NCT02835729, NCT02502708, NCT02077881) for cancer disease. K d of noncompetitive inhibitors, such as 4-PI (10) and imidazoleisoindole analog (11), are in overall agreement with their biochemical and cellular IC 50 values. However, the low-micromolar K d value of 11 is somehow higher than what expected from its nanomolar inhibition potency reported in the literature. A poor dissociation constant in the high-micromolar range is also found for mitomycin-C (13), a known natural anticancer compound that has been reported as uncompetitive inhibitor with the ability to bind to the substrate-bound complex of IDO1. Taken together the data presented in this work provide further insights into the binding properties of inhibitors to the ferric form of IDO1, offering novel clues for future development of more efficient IDO1 inhibitors. Moreover, our results support the use of MST analysis as biophysical tool that, combining with available biochemical and cellular assays of IDO1, may overcome issues related to the undesired identification of lead compounds with unspecific mechanism of inhibition.

Supporting information available
Supplementary  (10)  there is a stagnant rate of compounds advancing into clinical setting due to challenges associated with a correct appraisal of IDO1 inhibition. • The mechanism of action of the anticancer drug candidate D-1-Methyl-Trp (Indoximod, 6) is still questioned as to whether it is truly mediated by IDO1.

Aim
• Binding properties of a selection of different inhibitors to IDO1 are herein investigated using the MicroScale Thermophoresis (MST) analysis. • Docking calculations are performed to infer about the putative-binding modes of selected IDO1 inhibitors.

Method
• Thermophoresis is the movement of biomolecular molecules in a temperature gradient that depends on size, charge and hydration shell typically changing upon ligand/protein interaction.
• Mitomycin-C (13) binds to IDO-1 with an approximate K d value of 790.6 ± 117.6 μM, suggesting that other factors may account for the reported inhibition potency of this natural product.

Conclusion & future perspective
• The MST analysis pinpoints specific binding properties of L-1-Methyl-Trp (8) and D-1-Methyl-Trp (7) to the ferric form of IDO1. • Results support an IDO1-mediated anticancer activity of D-1-Methyl-Trp (Indoximod, 7). • Knowledge on K d of selected inhibitors to IDO1 offer novel clues for the design and development of more efficient inhibitors. • MST is a valuable biophysical tool that may overcome issues related to the undesired identification of compounds with unspecific mechanism of IDO1 inhibition. or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
• First evidence of substrate inhibition at high concentration of L-Tryptophan is reported.