Detailed analysis and follow-up studies of a high-throughput screening for indoleamine 2,3-dioxygenase 1 (IDO1) inhibitors

Article abstract of DOI:10.1016/j.ejmech.2014.06.078

Indoleamine 2,3-dioxygenase 1 (IDO1) is a key regulator of immune responses and therefore an important therapeutic target for the treatment of diseases that involve pathological immune escape, such as cancer. Here, we describe a robust and sensitive high-throughput screen (HTS) for IDO1 inhibitors using the Prestwick Chemical Library of 1200 FDA-approved drugs and the Maybridge HitFinder Collection of 14,000 small molecules. Of the 60 hits selected for follow-up studies, 14 displayed IC₅₀ values below 20 μM under the secondary assay conditions, and 4 showed an activity in cellular tests. In view of the high attrition rate we used both experimental and computational techniques to identify and to characterize compounds inhibiting IDO1 through unspecific inhibition mechanisms such as chemical reactivity, redox cycling, or aggregation. One specific IDO1 inhibitor scaffold, the imidazole antifungal agents, was chosen for rational structure-based lead optimization, which led to more soluble and smaller compounds with micromolar activity.

Full text of DOI:10.1016/j.ejmech.2014.06.078

European Journal of Medicinal Chemistry 84 (2014) 284e301
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Detailed analysis and follow-up studies of a high-throughput
screening for indoleamine 2,3-dioxygenase 1 (IDO1) inhibitors
a
a
,
, e
Ute F. Rohrig , Somi Reddy Majjigapu b, Marc Chambon c, Sylvian Bron a, Luc Pilotte d
,
€
,
,
Didier Colau d e, Benoît J. Van den Eynde d e, Gerardo Turcatti c, Pierre Vogel b,
Vincent Zoete a *, Olivier Michielin a
,
, f, g, *
a
^
ꢀ
Swiss Institute of Bioinformatics, Molecular Modeling Group, Quartier Sorge e Batiment Genopode, CH-1015 Lausanne, Switzerland
Laboratory of Glycochemistry and Asymmetric Synthesis, Ecole Polytechnique Federale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland
Biomolecular Screening Facility, Ecole Polytechnique Federale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland
b
c
ꢀ ꢀ
ꢀ ꢀ
d
ꢀ
de Duve Institute and the Universite catholique de Louvain, B-1200 Brussels, Belgium
e Ludwig Institute for Cancer Research, B-1200 Brussels, Belgium
f Department of Oncology, University of Lausanne and Centre Hospitalier Universitaire Vaudois (CHUV), CH-1011 Lausanne, Switzerland
g Ludwig Center for Cancer Research of the University of Lausanne, CH-1015 Lausanne, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
Indoleamine 2,3-dioxygenase 1 (IDO1) is a key regulator of immune responses and therefore an
important therapeutic target for the treatment of diseases that involve pathological immune escape, such
as cancer. Here, we describe a robust and sensitive high-throughput screen (HTS) for IDO1 inhibitors
using the Prestwick Chemical Library of 1200 FDA-approved drugs and the Maybridge HitFinder
Collection of 14,000 small molecules. Of the 60 hits selected for follow-up studies, 14 displayed IC50
Received 11 April 2014
Received in revised form
26 June 2014
Accepted 27 June 2014
Available online 10 July 2014
values below 20 mM under the secondary assay conditions, and 4 showed an activity in cellular tests. In
view of the high attrition rate we used both experimental and computational techniques to identify and
to characterize compounds inhibiting IDO1 through unspecific inhibition mechanisms such as chemical
reactivity, redox cycling, or aggregation. One specific IDO1 inhibitor scaffold, the imidazole antifungal
agents, was chosen for rational structure-based lead optimization, which led to more soluble and smaller
compounds with micromolar activity.
Keywords:
Cancer immunotherapy
Enzyme inhibition
High throughput screening
Indoleamine 2,3-dioxygenase
In silico drug design
© 2014 Elsevier Masson SAS. All rights reserved.
Molecular dynamics simulations
Structureeactivity relationship
Tryptophan metabolism
1. Introduction
tryptophan catabolism [5], leading to the depletion of tryptophan
and to the production of kynurenine metabolites, both responsible
The kynurenine pathway of tryptophan metabolism has been
recognized as a central component of both the nervous [1] and the
immune system [2e4]. The enzyme indoleamine 2,3-dioxygenase 1
(IDO1, EC 1.13.11.52) catalyzes the initial and rate-limiting step in
for local immunosuppression [6]. The immunosuppressive function
of IDO1 [7,8] and the functionally related enzyme tryptophan 2,3-
dioxygenase (TDO) [9,10] is exploited by tumor cells to escape a
potentially effective immune response [6]. As high IDO1 expression
is associated with poor prognosis in a variety of cancer types [11], it
has been selected as a therapeutic target for pharmacological in-
terventions [12,13]. Results from in vitro and in vivo studies
demonstrate indeed that the efficacy of therapeutic vaccination or
chemotherapy may be improved by concomitant administration of
an IDO1 inhibitor [7,14,15].
Abbreviations: IDO1, indoleamine 2,3-dioxygenase 1; MHC, Maybridge Hit-
Finder Collection; NFK, N-formyl kynurenine; PCL, Prestwick Chemical Library; PIM,
4-phenylimidazole; TDO, tryptophan 2,3-dioxygenase.
* Corresponding author. Swiss Institute of Bioinformatics, Molecular Modeling
^
ꢀ
Group, Quartier Sorge e Batiment Genopode, CH-1015 Lausanne, Switzerland.
€
E-mail addresses: ute.roehrig@isb-sib.ch (U.F. Rohrig), somireddy.majjigapu@
Intense efforts to develop potent, selective IDO1 inhibitors are
ongoing in academia and in pharmaceutical companies and have
yielded different active scaffolds, such as quinones, indoles,
hydroxyamidines, triazoles, and indoles [16]. Two compounds have
entered clinical trials, the hydroxyamidine INCB024360 developed
epfl.ch (S.R. Majjigapu), marc.chambon@epfl.ch (M. Chambon), sylvian.bron@unil.
ch (S. Bron), luc.pilotte@bru.licr.org (L. Pilotte), didier.colau@bru.licr.org
(D. Colau), benoit.vandeneynde@bru.licr.org (B.J. Van den Eynde), gerardo.
turcatti@epfl.ch (G. Turcatti), pierre.vogel@epfl.ch (P. Vogel), vincent.zoete@isb-
sib.ch (V. Zoete), olivier.michielin@isb-sib.ch (O. Michielin).
http://dx.doi.org/10.1016/j.ejmech.2014.06.078
0223-5234/© 2014 Elsevier Masson SAS. All rights reserved.

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
285
by Incyte [15,17,18] (NCT: NCT01604889, NCT01685255,
NCT01822691, NCT01961115), and more recently the imidazole
NLG919 from New Link Genetics [19]. High-throughput screening
(HTS) has become a routine method for the discovery of new in-
hibitors, but care must be taken to avoid compound-dependent
assay interference artifacts [20]. A few screening campaigns for
IDO1 inhibitors have been described. A yeast growth restoration
assay conducted with 2500 compounds from the NCI Diversity Set
and from algal extracts identified one IDO1 inhibitor with a Ki of
compounds were detected by structural filters, or, in case of strong
electrophiles, subject to an enzymatic assay in presence of high
concentrations of a nucleophile. Interference with the optical
readout was circumvented by using HPLC separation in the sec-
ondary assay. Confirmed hit compounds were subject to a cellular
test for murine IDO1 inhibition and counter-screened for murine
TDO inhibition. One confirmed active scaffold, the imidazole anti-
fungal agents, was chosen for computer-aided structure-based lead
optimization, which we have successfully applied to the design of
new IDO1 inhibitors before [40,41]. Here, this approach led to more
soluble and smaller compounds with micromolar activity.
1.5 mM [21]. 50,000 compounds of the ChemBridge DiverSet com-
pound library were subject to a yeast-based screening assay [22],
validated by a low hit rate (0.2%) and a high confirmation rate (76%).
However, no chemical structures were disclosed. Amgen's com-
pound library (245,000 compounds) has been screened in an
enzymatic assay against IDO1 [23], yielding a hit rate of 1.4%. The
2. Results and discussion
2.1. High throughput screening
structure of one hit, Amg-1 with an IC50 value of 3 mM, was dis-
closed. Another screen of the NCI Diversity Set with a new fluo-
rescence detection method [24] yielded 35 hits without disclosing
their structures. Recently, a library of 352 fragments was screened
on IDO1 using differential scanning fluorimetry as detection
method [25]. 51 hits were disclosed, and lead optimization
concentrated on the hydrazine scaffold, identifying phenyl-
hydrazine as a nanomolar IDO1 inhibitor.
Review of the screening literature on IDO1 reveals that only
selected fractions of the results have been published, for example
omitting the identification of promiscuous hits, and often con-
cealing the chemical structures of the active scaffolds. While this is
understandable in view of the advantage of nondisclosure when
developing new therapeutics, it hampers the discovery of new se-
lective IDO1 inhibitors, as manifest by numerous recent publica-
tions describing compounds probably acting through an unspecific
inhibition mechanism on IDO1. Here, we describe a HTS for IDO1
inhibitors including a detailed analysis of the hits and extensive
follow-up studies to distinguish between specific and promiscuous
IDO1 inhibitors. We only leave out the description of two active
compounds to allow for their potential further development as
therapeutic agents.
The employed HTS protocol was similar to commonly employed
enzymatic assay protocols [41]. Briefly, the reaction mixture con-
tained IDO1, the substrate
ascorbate and methylene blue, and the library compound at a
concentration of 10 M. After incubation for 60 min at room tem-
L-tryptophan, the reducing agents
m
perature, the enzymatic reaction was stopped by addition of a
hydrogen peroxide solution. Some features of the protocol need to
be taken into account when interpreting the screening results.
Firstly, to obtain a high-throughput detection method of the
enzymatic reaction product N-formyl kynurenine (NFK), no HPLC
separation was carried out. Instead, the light absorption of the re-
action mixtures was measured at a wavelength of 320 nm before
and after incubation with IDO1. However, this wavelength is not
very specific for NFK. Screening compounds that undergo chemical
modifications during incubation or after addition of hydrogen
peroxide may also cause absorbance changes in this region. Sec-
ondly, the absence of catalase, which is normally added to the in-
cubation medium [42], could facilitate the oxidation of active,
ferrous IDO1 to inactive, ferric IDO1. False positives due to these
effects were removed during the secondary assay, when kynur-
enine was quantified after HPLC separation and catalase was added
to the incubation medium.
In an academic screening effort, we screened two chemical li-
braries against IDO1, the Prestwick Chemical Library (PCL), which is
composed of 1200 FDA-approved drugs, and the Maybridge Hit-
Finder Collection (MHC) of 14,000 compounds. The structural
overlap between the Maybridge HitFinder Collection and previ-
ously screened libraries for which data is publicly available (NCI
Diversity Set, Chembridge DiverSet) is very low (23 and 77 com-
pounds respectively). Therefore the possibility of discovering new
inhibitor scaffolds through our screening is warranted. Although
some FDA approved compounds have previously been tested for
IDO1 inhibition [26], no systematic screening has been carried out
yet. The advantage of screening a library of FDA-approved com-
pounds is the large amount of available information. This is of value
not only for the purpose of drug repositioning [27] but also for
evaluating probable modes of action. As many of our screening hits
turned out to be false positives, we used known criteria for filtering
out HTS interference compounds [28e34] as well as different sec-
ondary assays for detecting unspecific inhibition. Oxidative pro-
teins are known to be difficult screening targets, because library
compounds capable of redox cycling may indirectly inhibit target
activity depending on assay conditions [29,35e37]. False positives
can also result from phenomena such as aggregation [28,38], co-
valent reactivity [34], or interference with the optical readout [20].
We used quantum-chemical calculations to investigate the link
between the reduction and oxidation potentials of known redox-
cycling compounds and their activity on IDO1. Aggregation was
controlled in a secondary assay by addition of small amounts of
detergent to the incubation medium [39]. Chemically reactive
An example of the screening result for one 384-well plate is
given in Fig.1A. The first 32 wells were used for the negative control
(no inhibitor present, cyan squares), while the last 32 wells were
used for the positive control (reference inhibitor present, green
squares). Here, we used our previously developed inhibitor 4-(3-
bromophenyl)-1,2,3-triazole [41] (IC50 ¼ 2.0
mM) at a concentra-
tion of 50 M as positive reference. At this concentration a com-
m
plete inhibition of IDO1 activity was achieved, as we previously
confirmed by HPLC analysis. A HTS score of zero was assigned to the
average absorbance change at 320 nm of the negative control
samples, while a score of one was assigned to the average absor-
bance change of the positive control samples.
The blue shaded area in Fig. 1A denotes the average 3 standard
deviations of the negative control, while the green shaded area
denotes the average 3 standard deviations of the positive control.
The calculated Z0 value [43] for this plate is 0.80 and demonstrates
the excellent reproducibility and sensitivity of the employed
screening protocol. The Z0 values for each individual screening plate
ranged from 0.74 to 0.86. The cumulative interplate Z0 was 0.8,
illustrating the good quality and robustness of the screening assay.
Using a HTS score cutoff of 0.2 to define a hit, 36 hits were obtained
for the PCL (3.0%) and 183 hits for the MHC (1.3%).
2.2. Structural and functional analysis of the PCL hits
Based on the large amount of literature available for the FDA-
approved compounds of the PCL, we developed the hypothesis

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
286
Fig. 1. A. HTS results for one sample plate. Each dark blue spot corresponds to one library compound. B. Results of the single-point assay for the 60 purchased compounds from the
PCL and the MHC. Compounds producing less than 5% of inhibition (dashed line) were classified as inactives and are thus false positives of the HTS. C. Correlation between the HTS
score and the score of the secondary assay calculated at a concentration of 10
color in this figure legend, the reader is referred to the web version of this article.)
m
M for the active compounds (correlation coefficient ¼ 0.55). (For interpretation of the references to
butoconazole (0.34), enilconazole (0.19), ketoconazole (0.15),
and bifonazole (0.14). These are known to inhibit the enzyme
that many of the identified hits act through unspecific inhibition on
IDO1 for one or more of the following reasons.
CYP51A1 (lanosterol 14
a-demethylase), a heme-containing
cytochrome P450 [46] but they could also act unspecifically
through aggregation [28].
ꢀ Redox-cycling compounds may interfere with the active
reduced state of IDO1 under the HTS assay conditions (e.g.
ebselen, primaquine, idebenone, chrysene-1,4-quinone,
anthralin, eseroline, nisoldipine, lacidipine, nifedipine, disul-
firam) [37].
ꢀ Iron chelators may interfere with the heme cofactor (e.g. ciclo-
pirox, disulfiram, deferoxamine, mercaptopurine) [44].
ꢀ Tetracycline antibiotics (e.g. lymecycline, oxytetracycline,
doxycycline) might act through a redox mechanism or as iron
chelators [45]
ꢀ Rabeprazole (0.75), a proton pump inhibitor.
ꢀ Phenelzine (0.30), an irreversible monoamine oxidase inhibitor
of the hydrazine class, which has been suggested to modulate
drug metabolism through heme binding in cytochrome P-450
[47].
ꢀ Amiodarone (0.21), an antiarrhythmic agent.
ꢀ Phentermine (0.15),
a psychostimulant drug and appetite
suppressant.
ꢀ Liranaftate (0.15), a squalene epoxidase inhibitor used as an
ꢀ Some compounds may inhibit IDO1 through chemical reactivity,
antifungal.
for example the alkylating agent dacarbazine.
ꢀ Raloxifene (0.14), a selective estrogen receptor modulator.
ꢀ Aggregators may promiscuously inhibit enzymes by denatur-
ation (miconazole, econazole, and sulconazole) [28].
ꢀ Dye molecules could produce a signal by changing absorption at
320 nm during incubation or after addition of hydrogen
peroxide (e.g. Chicago sky blue, propidium iodide, merbromin).
Our hits show a good overlap with the recently described FDA-
approved compounds active on mIDO1 [26], which included some
anti-fungal imidazoles (miconazole, econazole, clotrimazole, sul-
conazole) as well as lansoprazole, fenretinide, nifedipine, and
troglitazone.
See SI Table S1 for a list of all 23 purchased compounds from the
PCL, ranked according to their enzymatic IC50 value measured un-
der standard conditions.
To test our hypotheses, some of these compounds were chosen
for experimental follow-up studies. These include miconazole (HTS
score 0.91), sulconazole (0.91), econazole (0.83), ciclopirox (0.61),
mercaptopurine (0.61), disulfiram (0.59), nisoldipine (0.42), laci-
dipine (0.22), and nifedipine (0.09). Additionally, 14 compounds
were chosen for further testing:
2.3. Structural and functional analysis of the MHC hits
ꢀ The imidazole antifungal drugs sertaconazole (score 0.91), tio-
The hits from the MHC were also analyzed structurally and
functionally. Again, many of these compounds were identified as
conazole (0.68), oxiconazole (0.60), isoconazole (0.47),

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
287
Table 1
Results for all confirmed hits of the Prestwick Chemical Library (PCL). Name, chemical structure, number of failed structural filters (FScore), enzymatic IC50 value on hIDO under
standard conditions and with 0.01% Triton-X added, cellular IC50 value on mIDO, cellular IC50 value on mTDO, cellular LD50 value. All values are given in
ND: not determined.
mM. NI: no inhibition.
Name
Structure
FScore
Enzymatic tests
IC50/standard
Cellular tests
IC50/mIDO
IC50/Triton
IC50/mTDO
LD50
Sulconazole
0
6.4
6.7
8.1
8.4
ND
NI
NI
NI
NI
NI
10
Miconazole
1
0
0
3.8
ND
ND
NI
NI
NI
10
10
10
Econazole
Sertaconazole
Phenelzine
Bifonazole
1
0
14
23
ND
ND
NI
NI
NI
NI
>25
10
Enilconazole
1
23
ND
NI
NI
>25
Isoconazole
1
23
ND
NI
NI
10
(continued on next page)

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
288
Table 1 (continued )
Name
Structure
FScore
Enzymatic tests
IC50/standard
Cellular tests
IC50/mIDO
IC50/Triton
IC50/mTDO
LD50
Oxiconazole
1
28
ND
NI
NI
10
Ketoconazole
2
32
25
NI
NI
25
potential unspecific inhibitors because of redox-cycling capacities,
chemical reactivity, iron chelation, or strong optical absorption.
It is noteworthy that the five top-ranking hits from the MHC
were all para-quinones, which were also strongly enriched among
the screening hits (0.2% of the MHC, 5.5% of the hits). A similar
enrichment was observed for thiosemicarbazones, known iron
chelators (0.7% of the MHC, 13.7% of the hits).
We clustered compounds according to common chemical mo-
tifs, and one or a few representatives of each group were chosen for
follow-up studies. A total of 37 compounds was acquired, ranging
from 0.97 to 0.21 in the screening score (SI Table S2).
From the MHC, 17 compounds were selected for IC50 determi-
nation based on their activity in the single-point assay. The un-
confirmed hits included all five tested thiosemicarbazones.
2.4.3. Enzymatic IC50 determination
Enzymatic IC50 values were determined for the positives from
the single-point measurements. The assay conditions were the
same as for the single point measurements, except for the con-
centration of the tested chemical compounds, that was seven times
3-fold diluted from its highest soluble concentration. For 21 com-
pounds an IC50 value could be determined, and for 14 compounds
this value was below 20 mM (Tables 1 and 2). For the active com-
2.4. Experimental follow-up studies
pounds, a positive correlation was found between the HTS score
and the score in the secondary assay calculated at a compound
2.4.1. In vitro assay
concentration of 10
m
M (Fig. 1C, r ¼ 0.55), corroborating the validity
For detailed single-point and doseeresponse studies, we
quantified kynurenine production and tryptophan degradation
precisely by HPLC. Chromatographic separation of the reaction
mixture into its components allows for the use of higher
amounts of the reductive co-factors (ascorbate and methylene
blue) as well as the presence of catalase, shown to be necessary
for the removal of hydrogen peroxide from the medium [42].
Additionally, chromatographic separation prevents the interfer-
ence of colored compounds with the detection of the enzymatic
reaction product.
of the screening protocol for confirmed positives. However,
reproduction of measured IC50 values in independent experiments
was in general more challenging than in case of our previously
established inhibitor scaffolds [41], suggesting that they were
influenced by an experimental parameter that was not systemati-
cally monitored.
From the PCL, nine imidazole antifungal agents showed a
micromolar IC50 value (Table 1). The only structurally distinct hit
from the PCL was phenelzine (phenylethylhydrazine) with an IC50
value of 12 mM. It is known that hydrazine can directly bind to the
ferrous heme of cytochrome P450 (CYP) [47], so an interaction with
the active site heme of IDO1 is probable. Indeed, our docking results
suggest that the hydrazine group can bind to the heme iron of IDO1,
locating the phenyl ring in the buried hydrophobic A pocket
(Fig. 2A). This binding mode is in agreement with the recently
2.4.2. Single-point activities at highest soluble concentration
In order to be able to detect even weak IDO1 inhibitory activity
among the 60 purchased compounds, we tested each compound at
its previously determined highest soluble concentration (in phos-
phate buffer at pH 6.5 containing 5% DMSO). The results are given in
Fig. 1B, SI Tables S1 and S2. About half of the compounds (31
compounds, 52%) displayed more than 5% of IDO1 inhibitory ac-
tivity in this test, while the other half were inactive and therefore
false positives of the HTS. The HTS score does not allow filtering out
the false positives as they show scores in a wide range between 0.84
and 0.09.
From the PCL, only phenelzine and most of the imidazole anti-
fungal agents were confirmed as positives. As expected, the false
positives included the compounds ciclopirox, disulfiram, mercap-
topurine, nisoldipine, lacidipine, and nifedipine. However, also the
imidazole antifungal agents butoconazole and tioconazole as well
as the drugs amiodarone, rabeprazole, phentermine, liranaftate,
and raloxifene were not confirmed as hits.
proposed binding mode of phenylhydrazine (IC50 ¼ 0.25
mM) [25].
For eleven compounds from the MHC, an IC50 value could be
determined. The structures and results for these compounds,
except for two undisclosed compounds, are shown in Table 2.
2.4.4. Cellular IC50 and LD50 determination
We tested compounds with a low IC50 in the enzymatic assay for
their ability to inhibit tryptophan degradation and kynurenine
production in cells expressing murine (mIDO1). Murine IDO1
shares 62% of sequence identity with human IDO1 [48], and the
active site residues are 100% conserved. In an earlier work we found
an excellent correlation between cellular assay results on mIDO1
and on hIDO1 (r ¼ 0.95) [41]. Such cellular assay is informative for
drug development, as it evaluates not only the IDO1 inhibitory

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
289
Table 2
Results for the confirmed hits from the Maybridge HitFinder Collection (MHC), excluding two undisclosed compounds. Product code, chemical structure, number of failed
structural filters (FScore), enzymatic IC50 values on hIDO under standard conditions and with 5 mM GSH or 0.01% Triton-X added, cellular IC50 value on mIDO, cellular IC50 value
on mTDO, cellular LD50 value. All values are given in mM. NI: no inhibition. ND: not determined.
Code
Structure
FScore
Enzymatic tests
IC50/standard
Cellular tests
IC50/mIDO
IC50/GSH
IC50/Triton
IC50/mTDO
LD50
NRB00511
2
0.71
7.1
ND
NI
NI
5
CD02761
3
0.89
ND
1.2
2.1
NI
10
BTB13360
S14435
3
1
1.0
1.8
1.6
0.96
ND
NI
NI
NI
NI
>25
>25
130
DP00477
KM04550
2
1
7.0
11
71
6.3
4.8
NI
NI
NI
NI
>25
200
10
HTS01525
4
13
ND
15
NI
NI
>25
KM04416
NRB04258
1
2
15
54
100
31
11
53
NI
NI
10
10
0.34
4.2

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
290
Fig. 2. A. Phenelzine docked into the IDO1 X-ray structure. BeI. Imidazole antifungal agents docked into the IDO1 active site as present after molecular dynamics simulations. As
exemplified for miconazole (B, C), both the S and the R enantiomers can be accommodated in the active site.
effect of the compounds but also their capacity to permeate the cell,
their inhibition of tryptophan and kynurenine transport, and their
potential cytotoxicity. To control specificity, we also tested the
inhibitory activity of the compounds on cells expressing murine
TDO (mTDO). The same murine cell type, stably transfected with
murine IDO1 or TDO, was used for inhibition and cell viability
assays.
compounds a cellular IC50 value could be determined. In order to
better understand the effects leading to these results, we investi-
gated some of their potential causes.
2.5. Analysis of unconfirmed or promiscuous hits
2.5.1. Structural filters
Inhibition tests on mIDO1 and mTDO at a maximal concentra-
Several structural filters for the detection of problematic
screening compounds have been developed [30e34], and algo-
rithms for automated filtering are publicly available. The purpose of
the Glaxo filter [30] is to describe reactive functional groups, un-
suitable leads, and unsuitable natural products. The Pfizer LINT
tion of 25
(Table 1). One reason could be their high toxicity displayed in these
tests, resulting in LD50 values around 10 M, except for phenelzine
mM yielded negative results for all PCL compounds
m
and enilconazole. The observed toxicities for miconazole, econa-
zole, and sulconazole are slightly higher than the ones reported
earlier [26], which could be due to the different cell types (P815B
here vs. HEK), the different cell quantities (2 ꢁ 105 cells here vs.
1 ꢁ 105 cells), and the different medium (IMDM-2% FCS here vs.
Opti-MEM). In this earlier report, cellular IC50 values for these
Table 3
Number of compounds from the screening libraries, the screening hits, and the
confirmed hits failing different structural filters for the detection of problematic
screening compounds. The total number of compounds for each column is given on
top.
compounds were determined to be 18
econazole, and 30 M for sulconazole [26].
Of the eleven confirmed hits of the MHC, only four showed a
LD50 value above 25 M in the cellular assays (Table 2), but these
mM for miconazole, 11 mM for
m
Filter
Library
PCL
Hits
PCL
Confirmed hits
MHC
MHC
PCL
MHC
m
were inactive both on mIDO1 and on mTDO. Four of the remaining
compounds showed an activity on mIDO1. By addressing affinity,
selectivity, and safety, these active compounds might be pursued
for lead optimization.
Nr. of Compounds
36 183
1200
14,000
10
11
Lilly [34]
423 (35%) 5640 (40%) 20 (56%) 131 (72%) 1 (10%) 10 (91%)
Pfizer LINT [31] 518 (43%) 5850 (42%) 21 (58%) 95 (52%) 5 (50%) 6 (55%)
In summary, the attrition rate among the compounds acquired
for follow-up studies of the HTS was very high. Of the 60 acquired
compounds, 29 failed the single-point activity assay. For 21 com-
pounds an enzymatic IC50 value was obtained, but only for 4
Abbot ALARM
[32]
Glaxo [30]
PAINS [33]
170 (14%) 2265 (16%) 14 (39%) 75 (41%) 0 (0%)
4 (36%)
134 (11%) 1200 (9%)
21 (2%) 365 (3%)
5 (14%) 22 (12%) 0 (0%)
2 (6%)
1 (9%)
10 (5%) 1 (10%) 1 (9%)

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
291
filter [31] aims to detect reactive functional groups as well as
mutagenic or carcinogenic functionalities. The ALARM filter from
Abbott Laboratories is based on a specifically developed NMR assay
and predicts thiol reactivity of small molecules [32]. This is of
special interest for IDO1, as cystein modification has been shown to
inhibit enzymatic activity [49]. The “pan assay interference com-
pounds” (PAINS) filter relies on the analysis of large numbers of
screening hits and on literature data mining [33]. Most recently,
also Lilly Research Laboratories published an extensive list of rules
for identifying reactive or promiscuous compounds based on re-
view by a panel of experienced medicinal chemists [34].
screening libraries, while the other filters are more permissive. The
failure rates of all filters increase significantly among the screening
hits as compared to the full libraries, indicating that the hits contain
a higher number of problematic compounds. This trend is shown
graphically in Fig. 3A and B, where the number of failed filters per
compound among the full libraries and among the screening hits
are displayed. For the confirmed hits the filtering results are biased
by our cherry-picking procedure and the low structural diversity in
case of the PCL compounds. However, especially in case of the
confirmed hits of the MHC, the substantial number of failed filters
per compound (2 1, Tables 2 and 3) suggests that unspecific in-
hibitors are also present among the confirmed hits. Interestingly, a
more than twofold increase in the failure rate of the Abbot ALARM
As Table 3 shows, the Lilly Medchem Rules and the Pfizer LINT
filters flag the largest numbers of compounds, about 40% of the two
Fig. 3. Filtering results and doseeresponse curves of secondary enzymatic assays. Ratio of compounds that failed the indicated number of structural filters (0e5) (A) from the PCL
library (blue) and its hits (red), and (B) from the MHC library (blue) and its hits (red). Enzymatic doseeresponse curves of (C) miconazole and (D) ketoconazole without (black) and
with 0.01% Triton X-100 (red). Enzymatic doseeresponse curves of compounds (E) KM04550 and (F) S14435 without (black) and with (red) preincubation with 5 mM glutathione.
(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
292
filter is observed when passing from the full libraries (14% and 16%)
to the screening hits (39% and 41%), suggesting that thiol reactivity
plays a role in the detection of screening hits (Table 3).
The number of failed filters for each purchased screening hit is
given in SI Tables S1 and S2.
it is shown in Fig. 3E and F, preincubation with GSH strongly shifted
the IC50 values of two electrophilic compounds, KM04550 and
S14435, to higher concentrations. These results suggest that these
compounds inhibit IDO1 through covalent modification of nucle-
ophilic amino acid side chains. A similar trend was seen for com-
pounds NRB00511, DP00477, and KM04416, which all failed at least
the Lilly Medchem Rules or the Abbot ALARM thiol reactivity filter.
2.5.2. Aggregators
Assay interference by aggregation has been extensively dis-
cussed in the literature over the last decade [28,38], but the
detection of aggregation-prone compounds in silico remains chal-
lenging. Two algorithms developed in the Shoichet Laboratory
[28,50] yield widely disparate results when filtering the MHC, the
recursive partitioning filter [28] detecting 4825 compounds as
possible aggregators, the similarity filter [50] detecting only 332
compounds, and the overlap between the two methods being only
117 compounds. However, experimentally the interference of
aggregators can easily be minimized by the addition of small
amounts of non-ionic detergent, which increases their IC50 values
substantially [51,39]. Promiscuous aggregators often yield steep
doseeresponse curves with large Hill slopes [38].
Counter-screening for aggregation-based inhibition was espe-
cially important for the imidazole antifungal agents, as three of
them (miconazole, econazole, sulconazole) have been described as
promiscuous aggregating inhibitors before, while another (keto-
conazole) was not [28]. We therefore tested two of our antifungal
agents, miconazole and ketoconazole, in the absence and in the
presence of 0.01% Triton X-100 (Fig. 3C and D, Table 1). Interest-
ingly, the doseeresponse curves of miconazole and ketoconazole
display Hill slopes close to unity and do not shift to higher values in
presence of detergent, as it would be expected for an aggregation-
based inhibitor. We therefore propose that the imidazole antifungal
agents act through a specific mechanism on IDO1.
2.5.4. Redox-cycling compounds and para-quinones
Repetitively coupled reduction and oxidation reactions
involving oxygen and reactive oxygen species are called “redox
cycling”. Redox-cycling compounds have been described as prob-
lematic compounds for HTS because of their appearance as false
positives in many screens [29,36,37]. Since we observed numerous
known redox-cycling compounds among our hits, we tried to
determine if they inhibit IDO1 by entering into the active site and
oxidizing active ferrous heme into inactive ferric heme (specific
inhibition), or if they inhibit IDO1 by redox-cycling with the
reducing cofactors in solution (ascorbate, methylene blue) to
inactivate IDO1 (unspecific inhibition) [37]. This is an important
question, not only for the interpretation of our HTS results, but also
in view of the large number of published IDO1 inhibitors which
contain a quinone or a quinone-like function [40,49,52e63]. The
matter is complicated by the fact that quinones are not only redox-
cycling compounds [64], but some are also chemically reactive to-
wards nucleophilic amino acid side chains [32,65].
In the present HTS, 2 of the 7 para-quinones present in the PCL
and 10 of the 29 para-quinones present in the MHC were classified
as hits with the very high average score of 0.9 (SI Table S3).
Structurally these compounds can be divided into benzoquinones
(9 compounds, 2 hits), naphthoquinones (9 compounds, 7 hits),
anthraquinones (15 compounds, 1 hit), and other quinones (3
compounds, 2 hits). In spite of the large structural diversity of the
12 active para-quinones, exemplified by their molecular weight
(168e380 g/mol), number of rotatable dihedrals (0e12), number of
hydrogen bond acceptors (2e6), and clogP (ꢂ2.1e3.6), their HTS
scores only span the narrow range from 0.83 to 1.07 with two ex-
ceptions (0.62 and 0.60). This flat structureeactivity relationship is
indicative of an unspecific inhibition mechanism.
We also tested most of the active compounds of the MHC for
detergent-dependent inhibition (Table 2), including the three
compounds (HTS01525, DP00477, NRB04258) that fail the stricter
recursive partitioning filter for aggregators [28]. None of these
compounds showed a significant increase of IC50 value character-
istic for this type of promiscuous enzyme inhibition.
2.5.3. Chemically reactive compounds
To address the question if there exists a correlation between
redox potentials and HTS score, we calculated the redox potentials
of all screening compounds by a semi-empirical quantum-chemical
approximation. Additionally, the redox potentials of the 29 para-
quinones present in the MHC were calculated by higher level
density functional theory calculations (see SI for details). Indeed,
the calculations showed that on average the screening hits are
more prone to redox cycling than the inactive compounds, but the
differences are statistically not significant (Fig. S1A). For the para-
quinones, we did not find a correlation between their oxidation
potential and the HTS score (Fig. S1B). A similar absence of corre-
lation between calculated oxidation potentials and redox-cycling
Chemically reactive compounds may act on IDO1 activity not by
specifically binding to its active site, but rather by reacting for
example with the cofactors necessary for enzyme activation or with
nucleophilic amino acid side chains. It has been shown, e.g., that the
selenazal drug ebselen inhibits IDO1 by covalently modifying
multiple cysteine residues [49].
The Lilly Medchem Rules [34] filter out 72% of the screening hits
of the MHC, a 1.8-fold enrichment with respect to the ratio of failed
compounds in the full library (Table 3). An even higher enrichment
among the screening hits is observed for a number of specific
exclusion rules, such as quinone_para (25.5-fold enrichment), acy-
l_aromatic_nn (7.9-fold), hydrazone (4.9-fold), sulfonly_michael (3.1-
fold), enamine (2.9-fold), and vinyl_cyano_2 (2.3-fold), all detecting
potentially reactive compounds [34].
In order to experimentally test the prediction that some of the
confirmed MHC hits could react with nucleophilic protein side
chains, we carried out a secondary enzymatic assay, pre-incubating
the compounds with high concentrations of glutathione (GSH)
before adding IDO1 to the reaction mixture. Besides being a strong
nucleophile, GSH is also a reducing agent. However, as the incu-
bation medium already contained high concentrations of ascorbate,
its reducing properties should have a minor influence on the
enzymatic reaction. Indeed, the addition of GSH neither altered
IDO1 activity, nor had it any effect on the measured IC50 value of the
chemically inert compound 4-phenyl-imidazole (PIM). However, as
activity has recently been found in
a screening specifically
designed to detect redox-cycling compounds [64]. These findings
suggest that the chemical reactivity of quinones [32,65] must also
be taken into account. In fact, the 29 para-quinones on average
failed 4.3 out of the five structural filters we applied (Table S3). It
seems common medicinal chemistry knowledge that quinones do
not present valid lead-like properties.
2.6. Lead optimization of the imidazole antifungal agents
The high and uniform activity of the imidazole antifungal agents
in the enzymatic assay (Table 1) is remarkable, especially when
confronted with the complete inactivity of the structurally very
similar class of 1,2,4-triazole antifungal drugs (anastrozole,

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
293
Table 4
Derivatives of the imidazole antifungal scaffold. The modified part according to the IDO1-binding analysis, the respective substitutions, and the enzymatic IC50 values are
indicated. NI: no inhibition.
Comp.
Modification
R1
R2
R3
Enz. IC50 [mM]
2
A
NI
4a
4b
6a
6b
B
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
Imidazol-1-yl
1,2,4-Triazol-1-yl
1,2,3-Triazol-4-yl
H
H
2,4-di-Cl
e
820
NI
140
84
NI
130
NI
200
830
NI
NI
NI
A þ B
B
OCH3
OCH2-3-pyridyl
2,4-di-Cl
2,4-di-Cl
2,4-di-Cl
2,4-di-Cl
e
B
B
B
7
]O
OH
OH
OH
OH
OH
OH
OH
OH
OH
OH
10a
10b
10c
10d
10e
10f
10g
10h
10i
13
A þ B
A þ B
A þ B
A þ B
A þ B
A þ B
A þ B
Heme
Heme
2-Cl
3,4-di-Cl
3-Cl
4-Cl
2,6-di-Cl
2-OH
NI
NI
NI
2,4-di-Cl
2,4-di-Cl
4c
15
16
17
Spacer þ B
310
NI
Spacer þ B
Spacer þ A þ B
Spacer þ A þ B
NI
NI
fluconazole, itraconazole, letrozole, terconazole, voriconazole) also
present among the screening compounds. This striking difference
between the imidazole and the 1,2,4-triazole heterocycle is remi-
niscent of the discrimination of IDO1 between the regioisomers 4-
atom in position four have been described previously as IDO1 in-
hibitors [66e68]. Recently, also 4,5-disubstituted imidazoles active
on IDO1 have been discovered [69]. Here, we explore for the first
time the structureeactivity relationship of 1-substituted
imidazoles.
phenyl-1,2,3-triazole (enzymatic IC50 ¼ 83
mM) and 3-phenyl-1,2,4-
triazole (no inhibition) [41]. Together with the known binding
mode of imidazole antifungal drugs to the iron of heme proteins via
their unsubstituted imidazole nitrogen, resolved in many X-ray
structures [46], this finding suggests that also in IDO1 the imidazole
ring binds directly to the heme iron. As we described above, testing
of representative members of the imidazole antifungal agents for
aggregation-based inhibition returned negative results, suggesting
that they inhibit IDO1 through a specific mechanism. Although the
cellular IC50 values of the antifungal agents could not be deter-
mined due to their high toxicity in the MTS/PMS cell viability assay,
we selected this scaffold for optimization, expecting that the
cytotoxicity issue may be resolved based on the low number of
structural alerts (Table 1). Imidazoles substituted on the carbon
2.6.1. In silico studies
To date only two X-ray structures of IDO1 have been published,
one with cyanide and the other with the inhibitor 4-phenyl-imid-
azole (PIM) bound to the heme iron [70]. The latter has successfully
been used by ourselves [40,41] and by others [66,71,72] for the
structure-based design of new IDO1 inhibitors. In order to guide
lead optimization, we attempted to dock the imidazole antifungal
agents into the IDO1 active site. However, since PIM is much
smaller, there is not enough space in the corresponding X-ray
structure to fit the larger antifungal agents.
We therefore modeled the induced fit resulting from imidazole
antifungal agent binding by carrying out molecular dynamics (MD)

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
294
simulations of fully solvated IDO1 with miconazole bound in a
putative binding mode, and slowly removing all restraints such as
to obtain an intrinsically stable complex. The initial binding mode
of miconazole was chosen similar to the binding modes of imid-
azole antifungal agents in CYP proteins, with the free imidazole
nitrogen bound to the heme iron, one of its chlorinated phenyl rings
extending into the hydrophobic and buried A pocket, and the other
into the B pocket at the entrance of the active site.
Scheme 2. Reagents & conditions: (a) Imidazole, K2CO3, THF, reflux, 14 h.
In order to accommodate miconazole, IDO1 undergoes slight
structural rearrangements which increase the size of pocket A, but
(trimethylsilyl)-1H-imidazole and CsF in DMF (Scheme 7). All other
compounds (7, 16, 17) were bought from SigmaeAldrich.
the
overall
backbone
displacements
remain
small
(RMSD ¼ 0.95 0.09 Å for the residues within 5 Å of the ligand).
Miconazole rearranges undergoing several dihedral transitions to
accommodate one phenyl ring deeper in pocket A and the other one
deeper in pocket B. In this conformation it remains stable without
constraints for several tens of nanoseconds (Fig. 2B).
Using the central structure of the stable conformation of mi-
conazole from the MD simulations for docking with EADock-DSS
[73], a consistent binding mode for all antifungal agents was
found, with the imidazole bound to the heme iron, the phenyl ring
deeply buried inside the A pocket, and the second aromatic or
conjugated moiety located in the B pocket (Fig. 2BeI). Miconazole,
as many other imidazole antifungal agents, possesses a chiral
center. However, both enantiomers can bind to the IDO1 active site
in a very similar manner (Fig. 2B, C).
2.6.3. Structureeactivity relationship
In order to establish a structureeactivity relationship and to
increase the solubility of the imidazole antifungal family, we
chemically modified all four elements of the scaffold separately and
in combination (Table 4). All IC50 determinations were carried out
in the presence of 0.01% of Triton X-100 to minimize aggregation-
based inhibition.
Complete removal of the A-pocket group resulted in an inactive
compound (2), while complete removal of the B-pocket group
resulted in a compound with weak but detectable activity (4a,
IC50 ¼ 820
3-pyridyl group (6b) yielded a compound with an IC50 value of
84 M. Limitation of the B-pocket group to the methoxy (6a,
IC50 ¼ 140 M) both
mM). Replacement of the aromatic B-pocket moiety by a
Based on the docked structures of the antifungal agents, we
m
divided the representative structure of miconazole (IC50 ¼ 7.6
mM)
m
M) or to the hydroxyl function (10a, IC50 ¼ 130
m
into four structural entities: a heme binding motif (imidazole), a
spacer, an aromatic moiety occupying the A pocket, and an aro-
matic moiety occupying the B pocket (Table 4). All active antifungal
agents share the same heme binding motif, and, except for bifo-
nazole, all share the same spacer and A-pocket moiety. The greatest
structural variety is seen in the B pocket moiety.
yielded active compounds, while the corresponding keto com-
pound (7) was inactive. All these derivatives showed the advantage
of substantially enhanced solubility (>1 mM) with respect to mi-
conazole (z10 mM). While improving solubility, modifications of
the B-pocket group lowered the activity ten to twenty-fold. How-
ever, this region seems to be amenable to optimizations and can be
explored further.
Other modifications of the A-pocket group were tested on
compounds also lacking the aromatic B-pocket group in order to
profit from improved solubility. Removal of both chloride sub-
stituents from the A-pocket phenyl ring lead to an inactive com-
pound (10b) while removal of the 4-Cl substituent alone (10c)
2.6.2. Synthesis
The derivatives of the imidazole antifungal agents were syn-
thesized using previously published protocols or procedures
adapted from the literature [(2, 6ae6b) [74], (4ae4c) [75],
(10ae10i) [74,76], (13) [77,78] (15) [79]]. Compound 2 was obtained
by treatment of 2,4-dichlorobenzyl bromide with 2-(1H-imidazol-
1-yl)ethanol and sodium hydride in THF (Scheme 1). Compounds
4ae4c were obtained by treatment of an aryl bromide with imid-
azole and potassium carbonate in THF (Scheme 2). Compounds
6ae6b were prepared by reaction of the 1-(2,4-dichlorophenyl)-2-
(1H-imidazol-1-yl)ethanol with alkyl and aryl halides (Scheme 3).
A series of target compounds, namely the 1-aryl-2-(1H-imidazol-1-
yl)ethanols (10ae10h) and 1-(2,4-dichlorophenyl)-2-(1H-1,2,4-
triazol-1-yl)ethanol (10i) were prepared by reaction of the
respective 1-aryl-2-bromoethanones with imidazole/1H-1,2,4-
triazole (Schemes 4 and 5). Most of the required 1-aryl-2-
bromoethanones were commercially available. Compound 13 was
obtained by allylation of 2,4-dichlorobenzaldehyde with prop-
argylbromide mediated by zinc powder. The resulting ethynyl de-
rivative was reacted with TMSN3 (Scheme 6). Compound 15 was
obtained by treatment of 2,4-dichloro-1-fluorobenzene with 1-
yielded an active compound with an IC50 value of 200
substitutions on the phenyl ring in pocket A yielded less active (10d,
IC50 ¼ 830 M) or inactive compounds (10e, 10f, 10g, 10h).
mM. Other
m
Modification of the heme-binding moiety from imidazole to
either 1,2,4-triazole (10i) or 1,2,3-triazole (13) completely abolished
activity, suggesting that the 1,2,4-triazole antifungal drugs present
in the PCL are indeed inactive because of their heme-binding
moiety. Removal of the B-pocket group and simultaneous short-
ening of the spacer from two carbon atoms to one carbon atom (4c)
yielded a compound which was about 2.5 fold more active than the
corresponding compound with the original spacer (4a). However,
complete elimination of the spacer (15) yielded an inactive com-
pound. Elimination of the di-chloride substitution in pocket A of the
shortened compounds yielded inactive compounds (16 and 17), in
agreement with the inactivity of compound 4b.
In summary, our studies demonstrate that (i) the heme-binding
moiety is essential for activity, (ii) the spacer may be shortened
from two to one carbon atom, (iii) the A-pocket is sensitive to the di-
Scheme 1. Reagents & conditions: (a) 2-(1H-imidazol-1-yl)ethanol, NaH, THF, reflux,
3 h.
Scheme 3. (a) NaH, RX, DMF, 0 ꢃC-rt, 1.5 h.

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
295
Scheme 4. Reagents & conditions: (a) Imidazole, K2CO3, DMF, rt, 2e3 h (b) NaBH4, MeOH, 0 ꢃC-rt-50 ꢃC 2 h.
Scheme 5. Reagents & conditions: (a) 1H-1,2,4-triazole, K2CO3, DMF, rt, 2e3 h (b) NaBH4, MeOH, 0 ꢃC-rt-50 ꢃC 2 h.
Scheme 6. Reagents & conditions: (a) propargylbromide, Zn powder, 1 h (b) TMSN3, Cul, DMF/MeOH (9:1), 100 ꢃC, 10e12 h.
chloride substitutions in the right place, (iv) the B-pocket group is
most amenable to optimization, and (v) removal of one of the aro-
matic moieties from the A pocket or the B pocket greatly enhances
solubility. The fact that we obtained a structureeactivity relationship
covering an activity range of more than three orders of magnitude
suggests that the imidazole antifungal scaffold is amenable to opti-
mization and inhibiting IDO1 through a specific mechanism.
for topical application, but they might still be interesting in the local
treatmentof certain cancer types, such as melanoma for instance. The
reported therapeutic concentrations for example for miconazole
(40 mM) and econazole (26 mM) [28] are larger than their determined
IC50 values on IDO1 and could therefore induce significant IDO1 in-
hibition in vivo. Here, the imidazole antifungal scaffoldwas chosenfor
rational structure-based lead optimization, which led to more soluble
and smaller compounds with micromolar activity.
In summary, despite the robust screening results based on the
well-established enzymatic assay protocol for IDO1, most screening
hits turned out to be false positives or unspecific inhibitors,
establishing IDO1 as a difficult screening target. We hope that the
present work will be useful for other researchers in the field to
critically assess the specificity of novel IDO1 inhibitors.
3. Conclusions
We have screened a library of FDA-approved compounds and a
library of drug-like compounds for IDO1 inhibition. The screening
assay was based on a commonly used enzymatic assay protocol and
displayed a good sensitivity and robustness, expressed in a cumu-
lative interplate Z0 of 0.8. However, of the 60 hit compounds selected
for follow-up studies, an enzymatic IC50 value could be determined
only for 21 of them, and only four were active in cellular assays.
Through experimental and computational studies we could show
that most of the screening hits inhibit IDO1 through unspecific in-
hibition mechanisms such as covalent reactivity or redox cycling.
The only compound inhibiting mIDO1 but not mTDO in a
cellular context was the MHC compound CD02761 (Table 2).
However, this compound was flagged as problematic by 3 out of 5
structural filters because of its isothiourea functionality. S-
substituted isothioureas have been described before as IDO1 in-
hibitors in enzymatic and cellular assays [80], but their activities
might be influenced by hydrolysis to the corresponding reactive
thiols, which were also shown to be active [80].
Some FDA-approved compounds from the PCL showed
confirmed IDO1 inhibitory activity, for example phenelzine, a non-
selective and irreversible monoamine oxidase inhibitor which is
used as an antidepressant and anxiolytic. Recently, similar hydra-
zines have been described as IDO inhibitors [25]. However, phe-
nelzine has been suggested to covalently bind to the porphyrin of
the heme prosthetic group of CYP, thereby destroying it [47].
Moreover, the maximal daily dose of phenelzine (75e90 mg) might
not be enough for significant IDO1 inhibition in vivo.
4. Methods
4.1. HTS assay
The HTS was carried out on Corning UV-transparent flat-bottom
384-well plates. The reaction mixture (80
phosphate buffer at pH 6.5 (50 mM), ascorbic acid as reductant
(10 mM), methylene blue as electron-carrier (5 M), the substrate
L-Trp (100 M), the tested compound (10 M), DMSO (1%), and
IDO1 (4 g/mL). Before starting the incubation by addition of IDO1,
ml) contained potassium
m
m
m
m
the absorbance of each reaction well was recorded at a wavelength
of 320 nm. After incubation at room temperature for 60 min, the
reaction was stopped by addition of hydrogen peroxide solution
(8
ml, 1 M), and absorbance at 320 nm was remeasured. The
Scheme 7. Reagents & conditions: (a) 1-(trimethylsilyl)-1H-imidazole, CsF, DMF, 60 ꢃC,
20 h.
The imidazole antifungal agents described here provide a specific
IDO1 inhibiting scaffold. Most of these compounds are only approved

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
296
screening was carried out in duplicates. On each plate, 64 wells
were reserved for control samples.
The change in absorbance at 320 nm before and after incubation
with IDO1 is expected to be proportional to the quantity of NFK
formed by the enzymatic oxidation of L-Trp by IDO1 and thus yields
a measure of IDO1 activity in presence of the tested compound.
macinchem.org/reviews/pains/painsFilter.php). The Pfizer LINT
[31], Abbot ALARM [32], and Glaxo [30] filter calculations were
performed through the SmartsFilter web application, kindly pro-
vided by the Division of Biocomputing, Dept. of Biochemistry &
Molecular Biology, University of New Mexico, Albuquerque, NM,
(http://pasilla.health.unm.edu/tomcat/biocomp/smartsfilter). For
the Abbot ALARM filter, we used a lower cutoff of 0.3 for the thiol
reactivity index [32]. For the Glaxo filter, the three subfilters for
reactive functional groups, unsuitable leads, and unsuitable natural
products [30] were considered as a single filter. The detection of
aggregators based on the recursive partitioning filter [28] was used
as implemented in FILTER version 2.0.2. from OpenEye Scientific
Software, Santa Fe, NM. The detection of aggregators based on
similarity to known aggregators [50] is available through the
“Aggregator Advisor” web service (http://advisor.bkslab.org).
4.2. In vitro assay
The enzymatic assay was carried out on 96-well plates. The
reaction mixture (100
pH 6.5 (100 mM), ascorbic acid as reductant (20 mM), methylene
blue as electron carrier (10 M), bovine catalase to remove
hydrogen peroxide (400 units/mL), the substrate L-Trp (100 M),
the screening compound at varying concentrations, DMSO (5%),
and IDO1 (2.5 g/mL). After incubation at room temperature for
60 min, the reaction was stopped by addition of trichloroacetic acid
solution (40
l, 30% w/v), and the samples were incubated at 50 ꢃC
for 30 min to allow hydrolysis of the enzymatic reaction product,
ml) contained potassium phosphate buffer at
m
m
m
4.5. Modeling
m
We docked the hits of the two compound libraries in silico into
the IDO1 active site as provided in the PIM-bound X-ray structure
(PDB: 2D0T, chain A) [70].
NFK, to kynurenine. After centrifugation, 80
each well were used for HPLC analysis.
ml of supernatant from
The mobile phase for HPLC analysis was composed of 50% (v/v)
of methanol and 50% of sodium citrate buffer (40 mM, pH 2.35)
A consensus docking procedure based on our in-house docking
algorithms EADock [81e83], EADock DSS [73] and AttractingCav-
ities [84] was used. All three algorithms rely on the physical scoring
function of the CHARMM22 force field [85], while solvation effects
are taken into account by the FACTS model [86], which has been
shown to allow for accurate docking results [87]. Ligand force-field
parameters were derived with the SwissParam tool [88]. A Morse-
like metal binding potential (MMBP) was used to describe the in-
teractions between the heme iron of IDO1 and ligand atoms that
display a free electron pair for iron binding [89]. Two molecules of
N-cyclohexyltaurine bound at the entrance of the binding site [70]
were removed during setup, and the protein was kept fixed during
docking. For more details see Ref. [41].
An induced-fit structure of miconazole bound to IDO1 was
created by relaxing the protein structure around a putative binding
mode of the antifungal agent, forcing one phenyl ring into pocket A,
the other one into pocket B, and the imidazole binding to the heme
iron. We carried out MD simulations of the IDO1/miconazole
complex using the GROMACS code version 4.5.1 [90] in combina-
tion with the all-atom CHARMM27 force field [91,92] and the
explicit TIP3P water model [93]. A neutral system at physiological
ion concentration (z150 mmol/L) was obtained by adding sodium
and chloride ions. The box dimension of (100 Å)3 was chosen so
that the initial minimum gap between periodic images of the
protein was 30 Å. Electrostatic interactions were calculated with
the particle mesh Ewald method [94]. Bonds involving hydrogen
atoms were constrained using the P-LINCS algorithm [95]. The time
integration step was set to 2 fs. The system was kept at ambient
temperature (300 K) through velocity rescaling with a relaxation
time of 0.2 ps [96] and at ambient pressure (1 bar) by coupling to a
Berendsen bath [97] with a relaxation time of 1 ps. A distance re-
straint between the heme iron and the available imidazole nitrogen
of miconazole served to mimic a covalent bond.
containing 400
mM sodium dodecyl sulfate. An Agilent Zorbax
Eclipse XDB C18 column (150 ꢁ 4.6 mm) was used at 23 ꢃC with a
flow rate of 1 mL/min and an injection volume of 20
ml. For
detection of kynurenine, absorption at 365 nm was measured,
while remaining L-Trp was detected at a wavelength of 280 nm. All
assays were carried out in duplicates.
4.3. Cellular assays
For evaluating inhibition of murine IDO1 and murine TDO in a
cellular context, we used cell lines P815-mIDO1 clone 6 and P815-
mTDO clone 12, respectively, as described in Ref. [40]. The inhibi-
tion assays were performed in 96-well flat bottom plates seeded
with 2 ꢁ 105 cells in a final volume of 200
m
l of IMDM (Iscove's
Modified Dulbecco's Medium, Invitrogen) (80
mM L-tryptophan)
supplemented with 2% FCS and a titration of the compound ranging
from 0.024 to 25
m
M. The cells were incubated at 37 ꢃC for 18 h
(mIDO1) or for 24 h (mTDO). The reaction was stopped before 50%
of the initial amount of tryptophan was converted into kynurenine
to assure linearity of the reaction. The plates were then centrifuged
for 10 min at 300 g, and 150
HPLC analysis of tryptophan and kynurenine concentrations. Pro-
teins were precipitated by mixing 50 l of supernatant with an
equal volume of TCA 6%. After centrifugation, 70 l of supernatant
ml of the supernatant were collected for
m
m
were collected, diluted in an equal volume of water and injected
into the HPLC system (C18 column). Tryptophan was detected at an
absorption wavelength of 280 nm, and kynurenine at 360 nm.
To estimate cytotoxicity (LD50), 2 ꢁ 105 P815 cells were plated in
96-well flat bottom plates in a final volume of 200
ml of IMDM
(Invitrogen) supplemented with 2% FCS and a titration of the
compound ranging from 0.024 to 25 mM. After 24 h of incubation at
37 ꢃC, the number of living cells was evaluated with a MTS/PMS
assay (Promega) as described by the manufacturer.
The whole system was energy minimized, heated to 300 K over a
period of 100 ps, and equilibrated for 900 ps at 300 K, restraining
protein heavy atoms to their positions in the X-ray structure. For a
further 10 ns, only the positions of the heme and the ligand heavy
atoms were restrained to their initial positions, before removing all
structural restraints and continuing the simulation foranother 10 ns.
4.4. Structural filtering of screening compounds
We used the implementation of the Lilly Medchem Rules [34]
provided by one of the original authors which is available free of
charge at http://github.com/IanAWatson/Lilly-Medchem-Rules.
The PAINS filter (Tables S6eS8 from the original reference [33]) was
applied within the Filter-it software from Silicos-it (http://silicos-it.
com/), using the SMARTS list from Chris Swain (http://www.
4.6. Calculation of redox potentials
For the two libraries of 15,200 compounds in total, we opti-
mized the molecular geometries in gas phase with the semi-

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
297
empirical AM1 method [98] using the Gaussian09 code [99]. Their
reduction and oxidation potentials were approximated by the
resulting HOMO and LUMO energies, as described before [100]. We
checked that the inclusion of implicit aqueous solvation with the
conductor-like polarizable continuum model (CPCM) [101,102] did
not change the results qualitatively.
7.07 (bs, 1H), 6.98 (s, 1H), 4.56 (s, 2H), 4.17 (t, 2H, J ¼ 5.1 Hz), 3.79 (t,
2H, J ¼ 5.1 Hz) ppm. 13C NMR (100 MHz, CDCl3):
d 137.5, 133.9,
133.88, 133.3, 129.6, 129.4, 129.1, 127.0, 119.4, 70.0, 69.6, 47.0 ppm.
ESI-TOF-HRMS: m/z calcd for (M þ H) C12H12N2Cl2O 271.0405 found
271.0416.
For more detailed calculations on the 29 para-quinones of the
MHC, we adapted a recently reported procedure [103]. Each com-
pound was geometry optimized in the gas phase both in its
oxidized and in its one-electron reduced state at the density
functional theory level using the B3LYP hybrid functional [104,105]
and the 6-31G(d) basis set in the Gaussian09 code [99]. Aqueous-
phase molecular energies were calculated by using the
conductor-like polarizable continuum model (CPCM) [101,102]
with its default settings. It has been shown that the difference
between the solution-phase molecular energies of the reduced and
the oxidized species are highly correlated with experimentally
measured reduction potentials [103]. As reported recently, the en-
ergy difference computed in this way is highly correlated with the
LUMO energy of the oxidized compound [106] thus reducing the
necessary calculations per compound from four to two.
4.7.2.2. 1-(2,4-Dichlorophenethyl)-1H-imidazole
(4a)
[107].
Under a nitrogen atmosphere, a mixture of imidazole (2.1 mmol,
2.1 eq) and K2CO3 (1 mmol, 1 eq) in dry THF (6 mL) was stirred at
room temperature for 10 min. A solution of 1-(2-bromoethyl)-2,4-
dichlorobenzene (3a, 1 mmol, 1 eq) in THF (1 mL) was added, and
the mixture was heated under reflux for 14 h. The solution was
filtered and the filtrate was concentrated in vacuo. The residue was
dissolved in CH2Cl2, and the solution was washed with water (5 mL,
twice). The CH2Cl2 layer was extracted with dilute aqueous HCl
(5 mL, three times). The aqueous extract was neutralized with solid
NaHCO3, and the free base was extracted using CH2Cl2 (10 mL, three
times). The combined organic extracts were dried (Na2SO4) and
concentrated. High-vacuum drying gave 4a as an oil in 45% yield. 1H
NMR (400 MHz, CDCl3):
d 7.44e7.42 (m, 1H), 7.33 (s, 1H), 7.14 (dd,
1H, J ¼ 8.2, J ¼ 2.2 Hz), 7.05 (bs, 1H), 6.88 (d, 1H, J ¼ 8.0 Hz), 6.85 (s,
1H), 4.20 (t, 2H, J ¼ 7.0 Hz), 3.16 (t, 2H, J ¼ 7.0 Hz) ppm.
4.7. Chemistry
4.7.2.3. 1-Phenylethyl-1H-imidazole (4b) [75]. Same procedure as
for the preparation of 4a, using imidazole and (2-bromoethyl)
benzene (3b) affords 4b as an oil in 42% yield. 1H NMR (400 MHz,
4.7.1. General remarks
Materials and reagents were obtained from commercial sup-
pliers and used without further purification. For extraction and
chromatography, all solvents were distilled prior to use. Thin layer
chromatography for reaction monitoring was performed on silica
gel plates (Merck 60 F254) with detection by UV light (254 nm) and
charring with KMnO4 or Pancaldi reagent. Flash chromatography
was conducted using silica gel 60 Å, 230e400 mesh (Merck 9385).
Melting points were measured with a Mettler FP52 apparatus and
are uncorrected. IR spectra were recorded on a PerkineElmer
Paragon 1000 FT-IR spectrometer. Mass spectra were recorded on a
Nermag R 10-10C instrument in chemical ionization mode. Elec-
trospray mass analyses were recorded on a Finnigan MAT SSQ 710C
spectrometer in positive ionization mode. 1H and 13C NMR spectra
were recorded Bruker-DPX-400 or Bruker-ARX-400 spectrometer
at 400 MHz and 100.6 MHz, respectively. Data for 1H NMR spectra
are reported as follows: chemical shift, multiplicity, coupling con-
stant, and integration. Data for 13C NMR spectra are reported in
terms of chemical shifts. Chemical shifts are given in parts per
million, relative to an internal standard such as residual solvent
signals. Coupling constants are given in Hertz. High-resolution
mass spectra were recorded via ESI-TOF-HRMS. The purity of all
novel compounds was confirmed to exceed 95% by NMR and high-
resolution mass spectra.
CDCl3):
d 7.33e7.22 (m, 4H), 7.09e7.02 (m, 3H), 6.83 (s, 1H), 4.17 (t,
2H, J ¼ 7.0 Hz), 3.05 (t, 2H, J ¼ 7.0 Hz) ppm.
4.7.2.4. 1-(2,4-Dichlorobenzyl)-1H-imidazole (4c) [108]. Same pro-
cedure as for the preparation of 4a, using imidazole and 1-(bro-
momethyl)-2,4-dichlorobenzene (3c) affords 4c as a white solid in
60% yield. 1H NMR (400 MHz, CDCl3):
d 7.59 (s, 1H), 7.47 (d, 1H,
J ¼ 2.2 Hz), 7.25 (dd, 1H, J ¼ 8.2, J ¼ 2.6 Hz), 7.15 (bs, 1H), 6.96 (bs,
1H), 6.88 (d, 1H, J ¼ 8.0 Hz), 5.22 (s, 2H) ppm.
4.7.2.5. 1-(2-(2,4-Dichlorophenyl)-2-methoxyethyl)-1H-imidazole
(6a) [109]. A suspension of sodium hydride (1.5 mmol, 1.5 eq) in
DMF (5 mL) was treated with a solution of 1-(2,4-dichlorophenyl)-
2-(1H-imidazol-1-yl)ethanol (5, 1 mmol, 1 eq) in DMF (5 mL) at
0
ꢃC. The resulting mixture was stirred under a nitrogen atmo-
sphere at room temperature for 1.5 h. The mixture was cooled to
0
ꢃC and methyliodide (1 mmol, 1 eq) was added. The resulting
mixture was stirred for 1 h at room temperature. The reaction was
quenched by addition of water (15 mL). The aqueous phase was
extracted with ethyl acetate (15 mL, three times), the organic was
washed with water (15 mL), dried over Na2SO4 and the solvent was
removed in vacuo. Purification by column chromatography (SiO2,
MeOH/ethyl acetate) afforded 6a as an oil in 60% yield. 1H NMR
4.7.2. Procedures
(400 MHz, CDCl3): d 7.48e7.42 (m, 2H), 7.30e7.21 (m, 2H), 7.03 (bs,
4.7.2.1. 1-(2-((2,4-Dichlorobenzyl)oxy)ethyl)-1H-imidazole
(2).
1H), 6.92 (s, 1H), 4.78 (dd, 1H, J ¼ 7.0, J ¼ 2.6 Hz), 4.21e4.16 (m, 1H),
A suspension of sodium hydride (2 mmol, 2 eq) in THF (5 mL) was
treated with a solution of 2-(1H-imidazol-1-yl)ethanol (1 mmol,
1 eq) in THF (3 mL) at 0 ꢃC. The resulting mixture was stirred under
a nitrogen atmosphere at room temperature for 1 h. The mixture
was cooled to 0 ꢃC before addition of 2,4-dichlorobenzyl bromide
(1, 1.5 mmol, 1.5 eq). The resulting mixture was stirred under reflux
for 3 h. The reaction was quenched by addition of a saturated
aqueous solution of NH4Cl (5 mL). The aqueous phase was extracted
with ethyl acetate (15 mL, three times), the organic phase was
washed with water (10 mL), dried over Na2SO4 and the solvent was
removed in vacuo. Purification by column chromatography (SiO2,
4.04e3.97 (m, 1H), 3.25 (s, 3H) ppm.
4.7.2.6. 3-((1-(2,4-Dichlorophenyl)-2-(1H-imidazol-1-yl)ethoxy)
methyl)pyridine (6b). Synthesized from 5 and 3-(bromomethyl)
pyridine using the same procedure as for 6a to afford 6b as an oil in
62% yield. IR (film): nmax: 3366, 2923, 2868, 1918, 1717, 1588, 1504,
1468, 1229, 1090, 786 cmꢂ1. 1H NMR (400 MHz, CDCl3):
d 8.56 (dd,
1H, J ¼ 5.0, J ¼ 1.5 Hz), 8.45 (s, 1H), 7.49e7.44 (m, 3H), 7.35e7.24 (m,
3H), 7.04 (bs, 1H), 6.90 (s, 1H), 4.99 (dd, 1H, J ¼ 7.6, J ¼ 2.6 Hz),
4.50e4.45 (m, 1H), 4.30e4.18 (m, 2H), 4.09e4.01 (m,1H) ppm. 13
C
NMR (100 MHz, CDCl3):
d 149.3, 148.8, 137.7, 135.5, 135.1, 133.6,
MeOH/ethyl acetate) afforded 2 as an oil in 65% yield. IR (film): nmax
3109, 2868, 1898, 1589, 1505, 1469, 1075, 813, 737 cmꢂ1 1H NMR
(400 MHz, CDCl3): 7.55 (bs, 1H), 7.37 (s, 1H), 7.24e7.23 (m, 2H),
:
133.3, 132.4, 129.7, 129.0, 128.4, 128.0, 123.7, 119.8, 69.1, 51.3 ppm.
ESI-TOF-HRMS: m/z calcd for (M þ H) C17H15N3OCl2 348.0670 found
348.0684.
.
d

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
298
4.7.2.7. General procedure for the synthesis of imidazole alcohols
(10ae10h). A mixture of the -bromoketone (8aeh, 1 mmol, 1 eq),
ethanone (8g) according to the general procedure to afford 10g as a
white solid in 65% yield, 1H NMR (400 MHz, CD3OD):
7.63 (bs, 1H),
a
d
imidazole (3 mmol, 3 eq), and anhydrous K2CO3 (3 mmol, 3 eq) in
DMF (3 mL) was stirred at room temperature for 2e3 h. The reac-
tion mixture was diluted with H2O (10 mL) and extracted with
EtOAc (15 mL, twice). The combined organic phase was washed
sequentially with H2O (10 mL) and brine (10 mL). The combined
organic extracts were dried over Na2SO4, and evaporated. High-
vacuum drying gave crude 9aeh. A solution of crude 9aeh
(1 mmol, 1 eq) in methanol (4 mL) was added to a suspension of
sodium borohydride (1 mmol,1 eq) in anhydrous methanol (20 mL)
maintaining the temperature below 5 ꢃC under atmospheric pres-
sure. The resulting mixture was stirred at room temperature for 1 h
and then heated to 50 ꢃC for 1 h. After cooling to room temperature
the reaction mixture was concentrated and water (8 mL) was
added. The product was extracted with ethyl acetate (10 mL, twice).
The organic phase was dried over Na2SO4 and purified by flash
chromatography on silica gel using MeOHeEtOAc (6:94) as the
eluent to afford the imidazole alcohol (10ae10h); overall yields up
to 77%.
7.41e7.36 (m, 2H), 7.29e7.23 (m, 1H), 7.08 (bs, 1H), 6.93 (s, 1H),
5.75e5.69 (m, 1H), 4.69e4.61 (m, 1H), 4.49e4.42 (m, 1H) ppm.
4.7.2.15. 2-(1-Hydroxy-2-(1H-imidazol-1-yl)ethyl)phenol
(10h).
Synthesized from 2-bromo-1-(2-hydroxyphenyl)ethanone (8h)
according to the general procedure to afford 10h as a white solid in
70% yield, m.p. 149e152 ꢃC, IR (film): nmax: 3360, 2938, 1594, 1511,
1454, 1282, 820, 752 cmꢂ1, 1H NMR (400 MHz, CDCl3):
d 7.42 (bs,
1H), 7.14 (td, 1H, J ¼ 7.6, J ¼ 1.5 Hz), 7.05e7.01 (m, 1H), 6.94 (bs, 1H),
6.90 (s, 1H), 6.84e6.77 (m, 2H), 5.03e4.97 (m, 1H), 4.23e4.11 (m,
2H) ppm. 13C NMR (100 MHz, CD3OD):
d 153.9, 137.5, 128.2, 127.2,
126.8, 126.3, 120.2, 119.2, 114.6, 68.7, 52.3 ppm. ESI-TOF-HRMS: m/z
calcd for (MþH) C11H12N2O2 205.0977 found 205.0984.
4.7.2.16. 1-(2,4-Dichlorophenyl)-2-(1H-1,2,4-triazol-1-yl)ethanol
(10i) [115,116]. Synthesized from 2-bromo-1-(2,4-dichlorophenyl)
ethanone (8a) and 1H-1,2,4-triazole according to the general pro-
cedure to afford 10i as a white solid in 74% yield, 1H NMR (400 MHz,
4.7.2.8. 1-(2,4-Dichlorophenyl)-2-(1H-imidazol-1-yl)ethanol (10a)
[110]. Synthesized from 2-bromo-1-(2,4-dichlorophenyl)ethanone
(8a) according to the general procedure to afford 10a as a white
CD3OD):
d
8.43 (s, 1H), 7.96 (s, 1H), 7.56 (d, 1H, J ¼ 8.2 Hz), 7.47 (d,
1H, J ¼ 6.0 Hz) 7.36 (dd, 1H, J ¼ 8.3, J ¼ 2.1 Hz), 5.41e5.36 (m, 1H),
4.51e4.44 (m, 1H), 4.38e4.30 (m, 1H) ppm.
solid in 75% yield, 1H NMR (400 MHz, CDCl3):
d 7.61e7.57 (m, 1H),
7.44e7.39 (m, 2H), 7.34e7.30 (m, 1H), 6.93 (s, 1H), 6.92e6.87 (m,
1H), 5.30e5.25 (m, 1H), 4.28e4.20 (m, 1H), 3.94e3.85 (m, 1H) ppm.
4.7.2.17. 1-(2,4-Dichlorophenyl)-2-(1H-1,2,3-triazol-5-yl)ethanol
(13). Activated zinc powder (5.75 mmol, 1.15 eq) was placed in a
flame-dried round-bottom flask (50 mL) fitted with a magnetic stir
bar. 2,4-Dichlorobenzaldehyde (5 mmol, 1 eq) and propargyl-
bromide (5.75 mmol, 1.15 eq) were added. The resulting mixture
was vigorously stirred for 1 h at room temperature. After reaction
completion, sat. aq. soln of NH4Cl was poured into the mixture and
stirred for several minutes. Diethyl ether was added and the
organic layer was separated and dried over anhydrous MgSO4. The
residue was purified by flash chromatography on silica gel giving 1-
phenylpropargylalcohol (12). To a stirred solution of 12 (1 mmol,
1 eq) and CuI (0.05 mmol, 0.05 eq) in DMF/MeOH solution (2 mL
9:1) under an argon atmosphere was added trimethylsilyl azide
(1.5 mmol, 1.5 eq). The resulting solution was stirred at 100 ꢃC for
10e12 h. After consumption of ethynyl substrate, the mixture was
cooled to room temperature and the precipitate was filtered off and
the solution concentrated under reduced pressure. The crude res-
idue was purified by silica gel column chromatography to afford 13
as an oil in 60% yield. IR (film): nmax: 3142, 2932, 2828, 1588, 1560,
4.7.2.9. 2-(1H-imidazol-1-yl)-1-phenylethanol
(10b)
[111].
Synthesized from 2/bromo/1/phenylethanone (8b) according to the
general procedure to afford 10b as a white solid in 70% yield, 1H
NMR (400 MHz, CDCl3):
d 7.43e7.29 (m, 6H), 7.02 (bs, 1H), 6.92 (s,
1H), 4.98e4.93 (m, 1H), 4.20e4.08 (m, 2H) ppm.
4.7.2.10. 1-(2-Chlorophenyl)-2-(1H-imidazol-1-yl)ethanol
(10c)
[112]. Synthesized from 2-bromo-1-(2-chlorophenyl)ethanone
(8c) according to the general procedure to afford 10c as a white
solid in 72% yield, 1H NMR (400 MHz, CDCl3):
d 7.60e7.54 (m, 2H),
7.40e7.37 (m, 1H), 7.35e7.27 (m, 2H), 7.02 (bs, 1H), 6.98 (s, 1H), 5.35
(dd, 1H, J ¼ 7.8, J ¼ 2.2 Hz), 4.34e4.27 (m, 1H), 4.03e3.96 (m, 1H)
ppm.
4.7.2.11. 1-(3,4-Dichlorophenyl)-2-(1H-imidazol-1-yl)ethanol (10d)
[113,114]. Synthesized from 2-bromo-1-(3,4-dichlorophenyl)etha-
none (8d) according to the general procedure to afford 10d as a
1467, 1069, 1043, 818, 789 cmꢂ1. 1H NMR (400 MHz, CD3OD):
d 8.20
white solid in 75% yield, 1H NMR (400 MHz, CDCl3):
d 7.49e7.41 (m,
(s, 1H), 7.54e7.47 (m, 2H), 7.38e7.33 (m, 1H), 5.37e5.30 (m, 1H),
3H), 7.14 (dd, 1H, J ¼ 8.3, J ¼ 2.0 Hz), 7.01 (bs, 1H), 6.91 (s, 1H), 4.95
3.39e3.33 (m, 1H), 3.20e3.12 (m, 1H) ppm. 13C NMR (100 MHz,
(m, 1H), 4.18e4.01 (m, 2H) ppm.
CD3OD):
d 140.5, 133.2, 132.1, 128.5, 128.4, 127.0, 68.5, 32.4 ppm.
ESI-TOF-HRMS: m/z calcd for (MþH) C10H9N3Cl2O 258.0201 found
4.7.2.12. 1-(3-Chlorophenyl)-2-(1H-imidazol-1-yl)ethanol
(10e)
258.0191.
[113,114]. Synthesized from 2-bromo-1-(3-chlorophenyl)ethanone
(8e) according to the general procedure to afford 10a as a white
solid in 77% yield, 1H NMR (400 MHz, CDCl3):
d 7.42 (bs, 1H), 7.38 (s,
4.7.2.18. 1-(2,4-Dichlorophenyl)-1H-imidazole
(15)
[79].
1H), 7.34e7.28 (m, 2H), 7.22e7.16 (m, 1H), 6.99 (bs, 1H), 6.91 (s, 1H),
4.96e4.90 (m, 1H), 4.18e4.03 (m, 2H) ppm.
Synthesized from 2,4-dichloro-1-fluorobenze and 1-(trime-
thylsilyl)-1H-imidazole, according to reported procedure [79] to
afford 15 as a solid in 70% yield. 1H NMR (400 MHz, CDCl3):
d 7.70 (s,
4.7.2.13. 1-(4-chlorophenyl)-2-(1H-imidazol-1-yl)ethanol (10f) [76].
Synthesized from 2-bromo-1-(4-chlorophenyl)ethanone (8f) ac-
cording to the general procedure to afford 10f as a white solid in
1H), 7.60 (d, 1H, J ¼ 2.2 Hz), 7.40 (dd, 1H, J ¼ 8.3, J ¼ 2.2 Hz), 7.32 (d,
1H, J ¼ 8.4 Hz), 7.25 (bs, 1H), 7.16 (bs, 1H) ppm.
68% yield, 1H NMR (400 MHz, CD3OD):
d 7.50 (bs, 1H), 7.35e7.27 (m,
4H), 7.07 (bs, 1H), 6.90 (s, 1H), 4.95e4.90 (m, 1H), 4.26e4.12 (m, 2H)
Disclosure of potential conflict of interest
ppm.
B.J. Van den Eynde is a cofounder of and received consultancy
fees from iTeos Therapeutics. No potential conflicts of interest were
disclosed by the other authors.
4.7.2.14. 1-(2,6-Dichlorophenyl)-2-(1H-imidazol-1-yl)ethanol (10g)
[113,114]. Synthesized
from
2-bromo-1-(2,6-dichlorophenyl)

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
299
[11] J. Godin-Ethier, L.-A. Hanafi, C.A. Piccirillo, R. Lapointe, Indoleamine 2,3-
dioxygenase expression in human cancers: clinical and immunologic per-
spectives, Clin. Canc. Res. 17 (2011) 6985e6991, http://dx.doi.org/10.1158/
1078-0432.CCR-11-1331.
[12] J.B. Katz, A.J. Muller, G.C. Prendergast, Indoleamine 2,3-dioxygenase in T-cell
tolerance and tumoral immune escape, Immunol. Rev. 222 (2008) 206e221,
http://dx.doi.org/10.1111/j.1600-065X.2008.00610.x.
[13] G.C. Prendergast, Immune escape as a fundamental trait of cancer: focus on
IDO, Oncogene 27 (2008) 3889e3900, http://dx.doi.org/10.1038/
onc.2008.35.
[14] A.J. Muller, J.B. DuHadaway, P.S. Donover, E. Sutanto-Ward, G.C. Prendergast,
Inhibition of indoleamine 2,3-dioxygenase, an immunoregulatory target of
the cancer suppression gene bin1, potentiates cancer chemotherapy, Nat.
Med. 11 (2005) 312e319, http://dx.doi.org/10.1038/nm1196.
Acknowledgments
We would like to thank Miquel Busquets from the EPFL Bio-
molecular Screening Facility for screening automation and data
analysis and the NCCR Chemical Biology-ACCESS for providing ac-
cess to the PCL collection. We are much obliged to the Center of
Integrative Genomics (University of Lausanne), especially to Gilles
Boss for assistance with the enzymatic assay. For computational
resources and support we would like to thank the Vital-IT team at
the Swiss Institute of Bioinformatics and the computing services at
the University of Lausanne. We would like to thank Michel Cuendet,
[15] X. Liu, N. Shin, H.K. Koblish, G. Yang, Q. Wang, K. Wang, L. Leffet,
M.J. Hansbury, B. Thomas, M. Rupar, P. Waeltz, K.J. Bowman, P. Polam,
R.B. Sparks, E.W. Yue, Y. Li, R. Wynn, J.S. Fridman, T.C. Burn, A.P. Combs,
R.C. Newton, P.A. Scherle, Selective inhibition of IDO1 effectively regulates
mediators of antitumor immunity, Blood 115 (2010) 3520e3530, http://
dx.doi.org/10.1182/blood-2009-09-246124.
ꢀ
Antoine Daina, Aurelien Grosdidier, and Daniela Caldelari-Guex for
assistance with the computational and experimental aspects of this
work. Molecular graphics and analyses were performed with the
UCSF Chimera package [117]. Instant JChem 6.1, 2013, ChemAxon
(http://www.chemaxon.com), was used for structure database
management, search and prediction. Marvin 6.1, 2013, ChemAxon
(http://www.chemaxon.com) was used for drawing, displaying and
characterizing chemical structures. We are grateful to OpenEye
Scientific Software, Inc. (http://www.eyesopen.com) for the aca-
demic license agreement. This work was supported by the Clinical
Discovery Program of the Ludwig Institute for Cancer Research. The
computational aspects were supported by funds 310030e130857
and 310030e146971 to O. M. and V. Z. from the Swiss National
Science Foundation.
ꢁ ꢀ
ꢀ ꢀ
[16] E. Dolusic, R. Frederick, Indoleamine 2,3-dioxygenase inhibitors: a patent
review (2008e2012), Expert Opin. Ther. Targets 23 (2013) 1367e1381,
http://dx.doi.org/10.1517/13543776.2013.827662.
[17] E.W. Yue, B. Douty, B. Wayland, M. Bower, X. Liu, L. Leffet, Q. Wang,
K.J. Bowman, M.J. Hansbury, C. Liu, M. Wei, Y. Li, R. Wynn, T.C. Burn,
H.K. Koblish, J.S. Fridman, B. Metcalf, P.A. Scherle, A.P. Combs, Discovery of
potent competitive inhibitors of indoleamine 2,3-dioxygenase with in vivo
pharmacodynamic activity and efficacy in a mouse melanoma model, J. Med.
Chem. 52 (2009) 7364e7367, http://dx.doi.org/10.1021/jm900518f.
[18] H.K. Koblish, M.J. Hansbury, K.J. Bowman, G. Yang, C.L. Neilan, P.J. Haley,
T.C. Burn, P. Waeltz, R.B. Sparks, E.W. Yue, A.P. Combs, P.A. Scherle, K. Vaddi,
J.S. Fridman, Hydroxyamidine inhibitors of indoleamine-2,3-dioxygenase
potently suppress systemic tryptophan catabolism and the growth of ido-
expressing tumors, Mol. Cancer Ther. 9 (2010) 489e498, http://dx.doi.org/
10.1158/1535-7163.MCT-09-0628.
Appendix A. Supplementary data
[19] M.R. Mautino, F.A. Jaipuri, J. Waldo, S. Kumar, J. Adams, C. Van Allen,
A. Marcinowicz-Flick, D. Munn, N.N. Vahanian, C.J. Link, NLG919, a novel
indoleamine-2,3-dioxygenase (IDO)-pathway inhibitor drug candidate for
cancer therapy [abstract]Proceedings of the 104th Annual Meeting of the
American Association for Cancer Research; 2013 Apr 6e10; Washington, DC,
Cancer Res. vol. 73 (2013). Philadelphia (PA).
Supplementary data associated with this article can be found in
the online version, at http://dx.doi.org/10.1016/j.ejmech.2014.06.
078. These data include MOL files and InChiKeys of the most
important compounds described in this article.
[20] N. Thorne, D.S. Auld, J. Inglese, Apparent activity in high-throughput
screening: origins of compound-dependent assay interference, Curr. Opin.
Chem.
Biol.
14
(2010)
315e324,
http://dx.doi.org/10.1016/
j.cbpa.2010.03.020.
References
[21] E. Vottero, A. Balgi, K. Woods, S. Tugendreich, T. Melese, R.J. Andersen,
A.G. Mauk, M. Roberge, Inhibitors of human indoleamine 2,3-dioxygenase
identified with
282e288, http://dx.doi.org/10.1002/biot.200600001.
[22] M. Cerejo, G. Andrade, C. Roca, J. Sousa, C. Rodrigues, R. Pinheiro,
S. Chatterjee, H. Vieira, P. Calado, A powerful yeast-based screening assay for
the identification of inhibitors of indoleamine 2,3-dioxygenase, J. Biomol.
a target-based screen in yeast, Biotechnol. J. 1 (2006)
ꢀ
ꢀ
€
[1] L. Vecsei, L. Szalardy, F. Fülop, J. Toldi, Kynurenines in the CNS: recent ad-
vances and new questions, Nat. Rev. Drug Dis. 12 (2013) 64e82, http://
dx.doi.org/10.1038/nrd3793.
[2] H. Barth, S. Raghuraman, Persistent infectious diseases say e IDO. Role of
indoleamine-2,3-dioxygenase in disease pathogenesis and implications for
therapy, Crit. Rev. Microbiol. 40 (2014) 360e368, http://dx.doi.org/10.3109/
1040841X.2012.742037.
Screen.
17
(2012)
1362e1371,
http://dx.doi.org/10.1177/
1087057112452595.
[23] D. Meininger, L. Zalameda, Y. Liu, L.P. Stepan, L. Borges, J.D. McCarter,
C.L. Sutherland, Purification and kinetic characterization of human indole-
amine 2,3-dioxygenases 1 and 2 (IDO1 and IDO2) and discovery of selective
IDO1 inhibitors, Biochim. Biophys. Acta 1814 (2011) 1947e1954, http://
dx.doi.org/10.1016/j.bbapap.2011.07.023.
[24] P. Tomek, B.D. Palmer, J.U. Flanagan, S.-P.S. Fung, D.J.A. Bridewell, J.F. Jamie, L.-
M. Ching, Formation of an N-formylkynurenine-derived fluorophore and its
use for measuring indoleamine 2,3-dioxygenase 1 activity, Anal. Bioanal.
Chem. 405 (2013) 2515e2524, http://dx.doi.org/10.1007/s00216-012-6650-y.
[25] S.-P.S. Fung, H. Wang, P. Tomek, C.J. Squire, J.U. Flanagan, B.D. Palmer,
D.J. Bridewell, S.M. Tijono, J.F. Jamie, L.-M. Ching, Discovery and characteri-
sation of hydrazines as inhibitors of the immune suppressive enzyme,
[3] T.W. Stone, N. Stoy, L.G. Darlington, An expanding range of targets for
kynurenine metabolites of tryptophan, Trends Pharmacol. Sci. 34 (2013)
136e143, http://dx.doi.org/10.1016/j.tips.2012.09.006.
[4] D.H. Munn, A.L. Mellor, Indoleamine 2,3 dioxygenase and metabolic control
of immune responses, Trends Immunol. 34 (2013) 137e143, http://
dx.doi.org/10.1016/j.it.2012.10.001.
[5] S. Yamamoto, O. Hayaishi, Tryptophan pyrrolase of rabbit intestine. D- and L-
tryptophan-cleaving enzyme or enzymes, J. Biol. Chem. 242 (1967)
5260e5266.
[6] M. Platten, W. Wick, B.J. Van den Eynde, Tryptophan catabolism in cancer:
beyond IDO and tryptophan depletion, Cancer Res. 72 (2012) 5435e5440,
http://dx.doi.org/10.1158/0008-5472.CAN-12-0569.
indoleamine 2,3-dioxygenase
1 (IDO1), Bioorg. Med. Chem. 21 (2013)
ꢀ
[7] C. Uyttenhove, L. Pilotte, I. Theate, V. Stroobant, D. Colau, N. Parmentier,
7595e7603, http://dx.doi.org/10.1016/j.bmc.2013.10.037.
T. Boon, B.J. Van den Eynde, Evidence for a tumoral immune resistance
mechanism based on tryptophan degradation by indoleamine 2,3-
[26] S.M. Bakmiwewa, A. Fatokun, A. Tran, R.J. Payne, N.H. Hunt, H.J. Ball, Iden-
tification of selective inhibitors of indoleamine 2,3-dioxygenase 2, Bioorg.
Med. Chem. Lett. 22 (2012) 7641e7646, http://dx.doi.org/10.1016/
j.bmcl.2012.10.010.
[27] A. Mullard, Drug repurposing programmes get lift off, Nat. Rev. Drug Dis. 11
(2012) 505e506, http://dx.doi.org/10.1038/nrd3776.
dioxygenase, Nat. Med.
nm934.
9 (2003) 1269e1274, http://dx.doi.org/10.1038/
[8] C. Du, Y. Wang, The immunoregulatory mechanisms of carcinoma for its
survival and development, J. Exp. Clin. Cancer Res. 30 (2011) 12, http://
dx.doi.org/10.1186/1756-9966-30-12.
[28] J. Seidler, S.L. McGovern, T.N. Doman, B.K. Shoichet, Identification and pre-
diction of promiscuous aggregating inhibitors among known drugs, J. Med.
Chem. 46 (2003) 4477e4486, http://dx.doi.org/10.1021/jm030191r.
[29] K.M. Soares, N. Blackmon, T.Y. Shun, S.N. Shinde, H.K. Takyi, P. Wipf, J.S. Lazo,
P.A. Johnston, Profiling the NIH Small Molecule Repository for compounds
that generate H2O2 by redox cycling in reducing environments, Assay Drug
Dev. Technol. 8 (2010) 152e174, http://dx.doi.org/10.1089/adt.2009.0247.
[30] M. Hann, B. Hudson, X. Lewell, R. Lifely, L. Miller, N. Ramsden, Strategic
pooling of compounds for high-throughput screening, J. Chem. Inf. Comput.
Sci. 39 (1999) 897e902, http://dx.doi.org/10.1021/ci990423o.
[9] C.A. Opitz, U.M. Litzenburger, F. Sahm, M. Ott, I. Tritschler, S. Trump,
T. Schumacher, L. Jestaedt, D. Schrenk, M. Weller, M. Jugold, G.J. Guillemin,
C.L. Miller, C. Lutz, B. Radlwimmer, I. Lehmann, A. von Deimling, W. Wick,
M. Platten, An endogenous tumour-promoting ligand of the human aryl
hydrocarbon receptor, Nature 478 (2011) 197e203, http://dx.doi.org/
10.1038/nature10491.
[10] L. Pilotte, P. Larrieu, V. Stroobant, D. Colau, E. Dolusic, R. Frederick, E.D. Plaen,
C. Uyttenhove, J. Wouters, B. Masereel, B.J. Van den Eynde, Reversal of tu-
moral immune resistance by inhibition of tryptophan 2,3-dioxygenase, Proc.
Natl. Acad. Sci. U. S. A. 109 (2012) 2497e2502, http://dx.doi.org/10.1073/
pnas.1113873109.
ꢁ ꢀ
ꢀ ꢀ

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
300
[31] J.F. Blake, Identification and evaluation of molecular properties related to
[56] G. Carr, M.K.W. Chung, A.G. Mauk, R.J. Andersen, Synthesis of indole-
amine 2,3-dioxygenase inhibitory analogues of the sponge alkaloid exi-
guamine a, J. Med. Chem. 51 (2008) 2634e2637, http://dx.doi.org/
10.1021/jm800143h.
[57] S. Kumar, W. Malachowski, J. Duhadaway, J. Lalonde, P. Carroll, D. Jaller,
R. Metz, G. Prendergast, A. Muller, Indoleamine 2,3-dioxygenase is the
anticancer target for a novel series of potent naphthoquinone-based in-
hibitors, J. Med. Chem. 51 (2008) 1706e1718, http://dx.doi.org/10.1021/
jm7014155.
preclinical optimization and clinical fate, Med. Chem. 1 (2005) 649e655,
http://dx.doi.org/10.2174/15734060577459808.
[32] J.R. Huth, R. Mendoza, E.T. Olejniczak, R.W. Johnson, D.A. Cothron, Y. Liu,
C.G. Lerner, J. Chen, P.J. Hajduk, ALARM NMR: a rapid and robust experi-
mental method to detect reactive false positives in biochemical screens,
J. Am. Chem. Soc. 127 (2005) 217e224, http://dx.doi.org/10.1021/ja0455547.
[33] J.B. Baell, G. Holloway, New substructure filters for removal of pan assay
interference compounds (PAINS) from screening libraries and for their
exclusion in bioassays, J. Med. Chem. 53 (2010) 2719e2740, http://
dx.doi.org/10.1021/jm901137j.
[58] G. Carr, W. Tay, H. Bottriell, S.K. Andersen, A.G. Mauk, R.J. Andersen, Plec-
tosphaeroic acids a, b, and c, indoleamine 2,3-dioxygenase inhibitors pro-
[34] R.F. Bruns, I. Watson, Rules for identifying potentially reactive or promis-
cuous compounds, J. Med. Chem. 55 (2012) 9763e9772, http://dx.doi.org/
10.1021/jm301008n.
duced in culture by a marine isolate of the fungus Plectosphaerella
cucumerina, Org. Lett. 11 (2009) 2996e2999, http://dx.doi.org/10.1021/
ol900972j.
ꢁ ꢀ
[35] J.T. Pearson, S. Siu, D.P. Meininger, L.C. Wienkers, D.A. Rock, In vitro modulation
of cytochrome p450 reductase supported indoleamine 2,3-dioxygenase ac-
tivity by allosteric effectors cytochrome b(5) and methylene blue, Biochem-
istry 49 (2010) 2647e2656, http://dx.doi.org/10.1021/bi100022c.
[36] L. Lor, J. Schneck, D.E. McNulty, E. Diaz, M. Brandt, S.H. Thrall, B. Schwartz,
A simple assay for detection of small-molecule redox activity, J. Biomol.
Screen. 12 (2007) 881e890, http://dx.doi.org/10.1177/1087057107304113.
[37] P. Johnston, Redox cycling compounds generate H2O2 in HTS buffers con-
taining strong reducing reagentsereal hits or promiscuous artifacts? Curr.
Opin. Chem. Biol. 15 (2011) 174e182, http://dx.doi.org/10.1016/
j.cbpa.2010.10.022.
[59] E. Dolusic, P. Larrieu, C. Meinguet, D. Colette, A. Rives, S. Blanc, T. Ferain,
L. Pilotte, V. Stroobant, J. Wouters, B. Van den Eynde, B. Masereel,
ꢀ
ꢀ
E. Delfourne, R. Frederick, Indoleamine 2,3-dioxygenase inhibitory activity of
derivatives of marine alkaloid tsitsikammamine A, Bioorg. Med. Chem. Lett.
23 (2013) 47e54, http://dx.doi.org/10.1016/j.bmcl.2012.11.036.
[60] D.J. Bridewell, J. Sperry, J.R. Smith, P. Kosim-Satyaputra, L.-M. Ching,
J.F. Jamie, M. Brimble, Natural product-inspired pyranonaphthoquinone in-
hibitors of indoleamine 2,3-dioxygenase-1 (IDO-1), Aust. J. Chem. 66 (2013)
40, http://dx.doi.org/10.1071/CH12393.
[61] R. Pasceri, D. Siegel, D. Ross, C.J. Moody, Aminophenoxazinones as inhibitors
of indoleamine 2,3-dioxygenase (IDO). Synthesis of exfoliazone and chan-
drananimycin A, J. Med. Chem. 56 (2013) 3310e3317, http://dx.doi.org/
10.1021/jm400049z.
[38] B.K. Shoichet, Screening in a spirit haunted world, Drug Discov. Today 11
(2006) 607e615, http://dx.doi.org/10.1016/j.drudis.2006.05.014.
[39] B.Y. Feng, B.K. Shoichet, A detergent-based assay for the detection of pro-
[62] H.E. Flick, J.M. Lalonde, W.P. Malachowski, A.J. Muller, The tumor-selective
miscuous inhibitors, Nat. Protoc.
1
(2006) 550e553, http://dx.doi.org/
cytotoxic agent b-lapachone is a potent inhibitor of IDO1, Int. J. Trypto-
10.1038/nprot.2006.77.
phan Res. 6 (2013) 35e45, http://dx.doi.org/10.4137/IJTR.S12094.
[63] G. Pantouris, C.G. Mowat, Antitumour agents as inhibitors of tryptophan 2,3-
dioxygenase, Biochem. Biophys. Res. Commun. 443 (2014) 28e31, http://
dx.doi.org/10.1016/j.bbrc.2013.11.037.
[64] P. Rana, R. Naven, A. Narayanan, Y. Will, L.H. Jones, Chemical motifs that
redox cycle and their associated toxicity, MedChemComm 4 (2013) 1175,
http://dx.doi.org/10.1039/c3md00149k.
[65] S. Bittner, When quinones meet amino acids: chemical, physical and bio-
logical consequences, Amino acids 30 (2006) 205e224, http://dx.doi.org/
10.1007/s00726-005-0298-2.
[66] S. Kumar, D. Jaller, B. Patel, J.M. LaLonde, J.B. DuHadaway, W.P. Malachowski,
G.C. Prendergast, A.J. Muller, Structure based development of
phenylimidazole-derived inhibitors of indoleamine 2,3-dioxygenase, J. Med.
Chem. 51 (2008) 4968e4977, http://dx.doi.org/10.1021/jm800512z.
[67] M. Mautino, S. Kumar, F. Jaipuri, J. Waldo, T. Kesharwani, N. N. Vahanian, C. J.
Link, J. Lalonde, G. Prendergast, A. Muller, W. Malachowski. IDO Inhibitors.
Patent WO 2009/132238.
€
[40] U.F. Rohrig, L. Awad, A. Grosdidier, P. Larrieu, V. Stroobant, D. Colau,
V. Cerundolo, A.J.G. Simpson, P. Vogel, B.J. Van den Eynde, V. Zoete,
O. Michielin, Rational design of indoleamine 2,3-dioxygenase inhibitors,
J. Med. Chem. 53 (2010) 1172e1189, http://dx.doi.org/10.1021/jm9014718.
€
[41] U.F. Rohrig, S.R. Majjigapu, A. Grosdidier, S. Bron, V. Stroobant, L. Pilotte,
D. Colau, P. Vogel, B.J. Van den Eynde, V. Zoete, O. Michielin, Rational design
of 4-aryl-1,2,3-triazoles for indoleamine 2,3-dioxygenase 1 inhibition, J. Med.
Chem. 55 (2012) 5270e5290, http://dx.doi.org/10.1021/jm300260v.
[42] T. Ohnishi, F. Hirata, O. Hayaish, Indoleamine 2,3-dioxygenase. Potassium
superoxide as substrate, J. Biol. Chem. 252 (1977) 4643e4647.
[43] J.-H. Zhang, A simple statistical parameter for use in evaluation and valida-
tion of high throughput screening assays, J. Biomol. Screen. 4 (1999) 67e73,
http://dx.doi.org/10.1177/108705719900400206.
[44] Y. Eberhard, S.P. McDermott, X. Wang, M. Gronda, A. Venugopal, T.E. Wood,
R. Hurren, A. Datti, R. a. Batey, J. Wrana, W.E. Antholine, J.E. Dick, J. Dick,
A.D. Schimmer, Chelation of intracellular iron with the antifungal agent
ciclopirox olamine induces cell death in leukemia and myeloma cells, Blood
114 (2009) 3064e3073, http://dx.doi.org/10.1182/blood-2009-03-209965.
[45] M.O. Griffin, E. Fricovsky, G. Ceballos, F. Villarreal, Tetracyclines: a pleitropic
family of compounds with promising therapeutic properties. Review of the
literature, Am. J. Physiol. Cell P. H. 299 (2010) C539eC548, http://dx.doi.org/
10.1152/ajpcell.00047.2010.
[46] N. Strushkevich, S. Usanov, H.-W. Park, Structural basis of human CYP51
inhibition by antifungal azoles, J. Mol. Biol. 397 (2010) 1067e1078, http://
dx.doi.org/10.1016/j.jmb.2010.01.075.
[47] S.F. Muakkassah, W.C. Yang, Mechanism of the inhibitory action of phe-
nelzine on microsomal drug metabolism, J. Pharmacol. Exp. Ther. 219 (1981)
147e155.
[48] C.J.D. Austin, F. Astelbauer, P. Kosim-Satyaputra, H.J. Ball, R.D. Willows,
J.F. Jamie, N.H. Hunt, Mouse and human indoleamine 2,3-dioxygenase
display some distinct biochemical and structural properties, Amino Acids
36 (2009) 99e106, http://dx.doi.org/10.1007/s00726-008-0037-6.
[49] A.C. Terentis, M. Freewan, T.S.S. Plaza, M.J. Raftery, R. Stocker, S.R. Thomas,
The selenazal drug ebselen potently inhibits indoleamine 2,3-dioxygenase
by targeting enzyme cysteine residues, Biochemistry 49 (2010) 591e600,
http://dx.doi.org/10.1021/bi901546e.
[68] M. Mautino, S. Kumar, F. Jaipuri, J. Waldo, T. Kesharwani, X. Zhang. Imidazole
Derivatives as IDO Inhibitors. Patent WO 2011/056652.
[69] M. Mautino, S. Kumar, J. Waldo, F. Jaipuri, T. Kesharwani, X. Zhang.
Fused Imidazole Derivatives Useful as IDO Inhibitors. Patent WO 2012/
142237.
[70] H. Sugimoto, S. Oda, T. Otsuki, T. Hino, T. Yoshida, Y. Shiro, Crystal structure
of human indoleamine 2,3-dioxygenase: catalytic mechanism of O2 incor-
poration by a heme-containing dioxygenase, Proc. Natl. Acad. Sci. U. S. A. 103
(2006) 2611e2616, http://dx.doi.org/10.1073/pnas.0508996103.
ꢁ ꢀ
[71] E. Dolusic, P. Larrieu, S. Blanc, F. Sapunaric, J. Pouyez, L. Moineaux, D. Colette,
ꢂ
V. Stroobant, L. Pilotte, D. Colau, T. Ferain, G. Fraser, M. Galleni, J.-M. Frere,
ꢀ
ꢀ
B. Masereel, B. Van den Eynde, J. Wouters, R. Frederick, Discovery and pre-
liminary SARS of keto-indoles as novel indoleamine 2,3-dioxygenase (IDO)
inhibitors, Eur. J. Med. Chem. 46 (2011) 3058e3065, http://dx.doi.org/
10.1016/j.ejmech.2011.02.049.
[72] J.R. Smith, K.J. Evans, A. Wright, R.D. Willows, J.F. Jamie, R. Griffith, Novel
indoleamine 2,3-dioxygenase-1 inhibitors from a multistep in silico screen,
Bioorg. Med. Chem. 20 (2012) 1354e1363, http://dx.doi.org/10.1016/
j.bmc.2011.10.068.
[73] A. Grosdidier, V. Zoete, O. Michielin, Fast docking using the charmm force
field with eadock dss, J. Comput. Chem. 32 (2011) 2149e2159, http://
dx.doi.org/10.1002/jcc.21797.
[50] B. Shoichet, A. Doak, K. Ziebart, Aggregator Advisor, 2013. http://advisor.
bkslab.org.
[51] A.J. Ryan, N.M. Gray, P.N. Lowe, C.-w. Chung, Effect of detergent on “pro-
miscuous” inhibitors, J. Med. Chem. 46 (2003) 3448e3451, http://dx.doi.org/
10.1021/jm0340896.
~
ꢀ
[74] E. Cuevas Yanez, A. Canul Sanchez, J.M. Serrano Becerra, J.M. Muchowski,
R. Cruz Almanza, Synthesis of miconazole and Analogs through a Carbenoid
Intermediate, Rev. Soc. Quim. Mex. 48 (2004) 49e52.
[52] H.C. Brastianos, E. Vottero, B.O. Patrick, R.V. Soest, T. Matainaho, A.G. Mauk,
R.J. Andersen, Exiguamine a, an indoleamine-2,3-dioxygenase (IDO) inhibitor
isolated from the marine sponge neopetrosia exigua, J. Am. Chem. Soc. 128
(2006) 16046e16047, http://dx.doi.org/10.1021/ja067211.
[75] J.Z. Vlahakis, C. Lazar, G. Roman, D. Vukomanovic, K. Nakatsu, W.A. Szarek,
Heme oxygenase inhibition by a-(1H-imidazol-1-yl)-u-phenylalkanes: effect
of introduction of heteroatoms in the alkyl linker, ChemMedChem 7 (2012)
897e902, http://dx.doi.org/10.1002/cmdc.201100602.
[53] R. Andersen, A. Pereira, X.-H. Huang, G. Mauk, E. Vottero, M. Roberge, A.
Balgi. Indoleamine 2,3-Dioxygenase (IDO) Inhibitors. Patent WO 2006/
005185 (01 2006).
[54] A. Pereira, E. Vottero, M. Roberge, A.G. Mauk, R.J. Andersen, Indoleamine 2,3-
dioxygenase inhibitors from the northeastern pacific marine hydroid garveia
annulata, J. Nat. Prod. 69 (2006) 1496e1499, http://dx.doi.org/10.1021/
np060111x.
[76] G. Roman, J.Z. Vlahakis, D. Vukomanovic, K. Nakatsu, W.A. Szarek, Heme
oxygenase inhibition by 1-aryl-2-(1h-imidazol-1-yl/1h-1,2,4-triazol-1-yl)
ethanones and their derivatives, ChemMedChem
http://dx.doi.org/10.1002/cmdc.201000120.
5 (2010) 1541e1555,
[77] J.-X. Wang, X. Jia, T. Meng, L. Xin, Rapid and solvent-free synthesis of
Homoallyl or Homopropargyl alcohols mediated by zinc powder, Synthesis
2005 (2005) 2838e2844, http://dx.doi.org/10.1055/s-2005-872178.
[55] R. Andersen, M. Leblanc, H. Brastianos, E. Vottero, M. Roberge, G. Mauk, G.
Carr. Substituted Quinone Indoleamine 2,3-Dioxygenase (IDO) Inhibitors and
Synthesis and Uses Thereof. Patent WO 2008/052352 (05 2008).
[78] T. Jin, S. Kamijo, Y. Yamamoto, Copper-catalyzed synthesis of N-unsub-
stituted 1,2,3-triazoles from nonactivated terminal alkynes, Eur. J. Org.
Chem. 18 (2004) 3789e3791, http://dx.doi.org/10.1002/ejoc.200400442.

€
U.F. Rohrig et al. / European Journal of Medicinal Chemistry 84 (2014) 284e301
301
[79] D. Dehe, I. Munstein, A. Reis, W.R. Thiel, Mild transition-metal-free amina-
tion of fluoroarenes catalyzed by fluoride ions, J. Org. Chem. 76 (2011)
1151e1154, http://dx.doi.org/10.1021/jo102063s.
[80] K. Matsuno, K. Takai, Y. Isaka, Y. Unno, M. Sato, O. Takikawa, A. Asai, S-
benzylisothiourea derivatives as small-molecule inhibitors of indoleamine-
2,3-dioxygenase, Bioorg. Med. Chem. Lett. 20 (2010) 5126e5129, http://
dx.doi.org/10.1016/j.bmcl.2010.07.025.
[81] A. Grosdidier, V. Zoete, O. Michielin, EADock: docking of small molecules into
protein active sites with a multiobjective evolutionary optimization, Proteins
67 (2007) 1010e1025, http://dx.doi.org/10.1002/prot.21367.
[82] A. Grosdidier, V. Zoete, O. Michielin, Blind docking of 260 protein-ligand
complexes with EADock 2.0, J. Comput. Chem. 30 (2009) 2021e2030,
http://dx.doi.org/10.1002/jcc.21202.
K.N. Kudin, V.N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari,
A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J.M. Millam,
M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts,
R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski,
R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador,
J.J. Dannenberg, S. Dapprich, A.D. Daniels, O. Farkas, J.B. Foresman, J.V. Ortiz,
J. Cioslowski, D.J. Fox, Gaussian 09 Revision A.1, Gaussian Inc, Wallingford CT,
2009.
€
[100] R.V. Bensasson, S. Sowlati-Hashjin, V. Zoete, D. Dauzonne, C.F. Matta, Phys-
icochemical properties of exogenous molecules correlated with their bio-
logical efficacy as protectors against carcinogenesis and inflammation, Int.
Rev. Phys. Chem. 32 (2013) 393e434, http://dx.doi.org/10.1080/
0144235X.2013.767669.
[83] V. Zoete, A. Grosdidier, M. Cuendet, O. Michielin, Use of the FACTS solvation
model for protein-ligand docking calculations. Application to EADock, J. Mol.
Recognit. 23 (2010) 457e461, http://dx.doi.org/10.1002/jmr.1012.
[84] V. Zoete, A. Grosdidier, T. Schuepbach, O. Michielin, Attracting Cavities
(manuscript in preparation).
[101] V. Barone, M. Cossi, Quantum calculation of molecular energies and energy
gradients in solution by a conductor solvent model, J. Phys. Chem. A 102
(1998) 1995, http://dx.doi.org/10.1021/jp9716997.
[102] M. Cossi, N. Rega, G. Scalmani, V. Barone, Energies, structures, and electronic
properties of molecules in solution with the C-PCM solvation model,
J. Comput. Chem. 24 (2003) 669, http://dx.doi.org/10.1002/jcc.10189.
[103] E.J. Lynch, A.L. Speelman, B.A. Curry, C.S. Murillo, J.G. Gillmore, Expanding
and testing a computational method for predicting the ground state reduc-
tion potentials of organic molecules on the basis of empirical correlation to
experiment, J. Org. Chem. 77 (2012) 6423e6430, http://dx.doi.org/10.1021/
jo300853k.
[85] A.D. MacKerell Jr., D. Bashford, M. Bellott, R.L. Dunbrack Jr., J.D. Evanseck,
M.J. Field, S. Fischer, J. Gao, H. Guo, S. Ha, D. Joseph-McCarthy, L. Kuchnir,
K. Juczera, F.T.K. Lau, C. Mattos, S. Michnik, T. Ngo, D.T. Nguyen, B. Prodhom,
W.E. Reiher III, B. Roux, M. Schlenkrich, J.C. Smith, R. Stote, J. Straub,
ꢀ
M. Watanabe, J. Wiorkiewicz-Kuczera, D. Yin, M. Karplus, All-atom empirical
potential for molecular modeling and dynamics studies of proteins, J. Phys.
Chem. B 102 (1998) 3586e3616, http://dx.doi.org/10.1021/jp973084f.
[86] U. Haberthür, A. Caflisch, Facts: fast analytical continuum treatment of sol-
vation, J. Comput. Chem. 29 (2008) 701e715, http://dx.doi.org/10.1002/
jcc.20832.
[104] A. Becke, Density-functional thermochemistry. iii. The role of exact ex-
change, J. Chem. Phys. 98 (1993) 5648e5652, http://dx.doi.org/10.1063/
1.464913.
[105] C. Lee, W. Yang, R. Parr, Development of the Colle-Salvetti correlation-energy
formula into a functional of the electron density, Phys. Rev. B 37 (1988)
785e789, http://dx.doi.org/10.1103/PhysRevB.37.785.
[87] V. Zoete, A. Grosdidier, M. Cuendet, O. Michielin, Use of the FACTS solvation
model for protein-ligand docking calculations. Application to EADock, J. Mol.
Recognit. 23 (2010) 457e461, http://dx.doi.org/10.1002/jmr.1012.
[88] V. Zoete, M.A. Cuendet, A. Grosdidier, O. Michielin, SwissParam: a fast force
field generation tool for small organic molecules, J. Comput. Chem. 32 (2011)
2359e2368, http://dx.doi.org/10.1002/jcc.21816.
ꢀ
ꢀ
[106] D.D. Mendez-Hernandez, P. Tarakeshwar, D. Gust, T. a. Moore, A.L. Moore,
V. Mujica, Simple and accurate correlation of experimental redox potentials
and DFT-calculated HOMO/LUMO energies of polycyclic aromatic hydro-
carbons, J. Mol. Model 19 (2013) 2845e2848, http://dx.doi.org/10.1007/
s00894-012-1694-7.
€
[89] U.F. Rohrig, A. Grosdidier, V. Zoete, O. Michielin, Docking to heme proteins,
J. Comput. Chem. 30 (2009) 2305e2315, http://dx.doi.org/10.1002/jcc.21244.
[90] B. Hess, C. Kutzner, D. van der Spoel, E. Lindahl, GROMACS 4: algorithms for
highly efficient, load-balanced, and scalable molecular simulation, J. Chem.
Theory Comput. 4 (2008) 435e447, http://dx.doi.org/10.1021/ct700301q.
[91] A.D. MacKerell, M. Feig, C.L. Brooks, Extending the treatment of backbone
energetics in protein force fields: limitations of gas-phase quantum me-
chanics in reproducing protein conformational distributions in molecular
dynamics simulations, J. Comput. Chem. 25 (2004) 1400e1415, http://
dx.doi.org/10.1002/jcc.20065.
[92] P. Bjelkmar, P. Larsson, M.A. Cuendet, B. Hess, E. Lindahl, Implementation of
the CHARMM force field in GROMACS: analysis of protein stability effects
from correction maps, virtual interaction sites, and water models, J. Chem.
Theory Comput. 6 (2010) 459e466, http://dx.doi.org/10.1021/ct900549r.
[93] W.L. Jorgensen, J. Chandrasekhar, J.D. Madura, R.W. Impey, M.L. Klein,
Comparison of simple potential functions for simulating liquid water,
J. Chem. Phys. 79 (1983) 926, http://dx.doi.org/10.1063/1.445869.
[107] W.N. Cannon, C.E. Powell, R.G. Jones, Studies of imidazoles. VII. Substituted
1-phenethylimidazoles and related compounds, J. Org. Chem. 22 (1957)
1323e1326, http://dx.doi.org/10.1021/jo01362a008.
~
[108] M.R. Cuberes, M. Moreno-Manas, A. Trius,
a-Lithiation of N-arylmethylimi-
dazoles and triazoles: a general method for the synthesis of 1,2-diaryl-1-(N-
azolyl)-ethanes, Synthesis 1985 (1985) 302e304, http://dx.doi.org/10.1055/
s-1985-31187.
[109] Y. Wahbi, C. Tournaire, R. Caujolle, M. Payard, M. Linas, J. Seguela, Aliphatic
ethers and esters of 1-(2,4-dichlorophenyl)-2-(1H-imidazolyl) ethanol:
study of antifungal activity against yeasts and hydrophobic character, Eur. J.
Med. Chem. 29 (1994) 701e706, http://dx.doi.org/10.1016/0223-5234(94)
90032-9.
[110] G.M. Dulcevscaia, V.C. Kravtsov, F.Z. Macaev, G.G. Duca, E.P. Stingachi,
S.I. Pogrebnoi, V.V. Boldescu, S.F. Clapco, J.P. Tiurina, A.A. Deseatnic-Ciloci,
J. Lipkowski, S.-X. Liu, S. Decurtins, S.G. Baca, New copper(II) complexes with
isoconazole: synthesis, structures and biological properties, Polyhedron 52
(2013) 106e114, http://dx.doi.org/10.1016/j.poly.2012.10.040.
[111] R. Torregrosa, I.M. Pastor, M. Yus, Solvent-free direct regioselective ring
opening of epoxides with imidazoles, Tetrahedron 63 (2007) 469e473,
http://dx.doi.org/10.1016/j.tet.2006.10.055.
[94] U. Essmann, L. Perera, M.L. Berkowitz, T. Darden, H. Lee, L.G. Pedersen,
A smooth particle mesh Ewald method, J. Chem. Phys. 103 (1995) 8577,
http://dx.doi.org/10.1063/1.470117.
[95] B. Hess, P-LINCS: a parallel linear constraint solver for molecular simulation,
J. Chem. Theory Comput.
ct700200b.
[96] G. Bussi, D. Donadio, M. Parrinello, Canonical sampling through velocity
rescaling, J. Chem. Phys. 126 (2007) 014101, http://dx.doi.org/10.1063/
1.2408420.
[97] H.J.C. Berendsen, J.P.M. Postma, W.F. van Gunsteren, A. DiNola, J.R. Haak,
Molecular dynamics with coupling to an external bath, J. Chem. Phys. 81
(1984) 3684, http://dx.doi.org/10.1063/1.448118.
4
(2008) 116e122, http://dx.doi.org/10.1021/
[112] Y. Iwasaki, T. Fujita, H. Yabe, H. Hirao, K. Kitagawa, S. Futaki, T. Inoue,
T. Matsuzaki, T. Akita, Synthesis and antifungal activity of a series of novel 2-
propen-1-one, Yakugaku Zasshi 108 (1988) 942e950.
[113] G. Sharma-Singh, D.A. Brown, F.S. Pullen, A.G. Wright, Open access capillary
electrophoresis. A walk up capillary electrophoresis service for the synthetic
chemist, J. Chromatogr. A 888 (2000) 219e227.
[114] E. F. Godefroi, J. Heeres. 1-(Beta-aryl)Ethyl-Imidazole Derivatives. Patent DE
1940388 (1970).
[98] M.J.S. Dewar, E.G. Zoebisch, E.F. Healy, J.J.P. Stewart, AM1: a new general
purpose quantum mechanical molecular model, J. Am. Chem. Soc. 107 (1985)
3902e3909, http://dx.doi.org/10.1021/ja00299a024.
[99] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb,
J.R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson,
H. Nakatsuji, M. Caricato, X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino,
G. Zheng, J.L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda,
J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven,
J.A. Montgomery Jr., J.E. Peralta, F. Ogliaro, M. Bearpark, J.J. Heyd, E. Brothers,
[115] Z.-F. Li, F.-Y. Luo, Jingxi Huagong 22 (2005) 619e624.
[116] K. Oh, K. Nakai, K. Yamada, Y. Yoshizawa, Synthesis of novel triazole de-
rivatives as potent inhibitor of allene oxide synthase (CYP74A), a key enzyme
in jasmonic acid biosynthesis, J. Pestic. Sci. 37 (2012) 80e84, http://
dx.doi.org/10.1584/jpestics.D11-006.
[117] E.F. Pettersen, T. Goddard, C. Huang, G. Couch, D. Greenblatt, E. Meng,
T.E. Ferrin, UCSF Chimera e a visualization system for exploratory research
and analysis, J. Comput. Chem. 25 (2004) 1605e1612, http://dx.doi.org/
10.1002/jcc.20084.