Identification of M. tuberculosis thioredoxin reductase inhibitors based on high-throughput docking using constraints

Article abstract of DOI:10.1021/jm3015734

A virtual screening campaign is presented that led to small molecule inhibitors of thioredoxin reductase of Mycobacterium tuberculosis (MtTrxR) that target the protein-protein interaction site for the substrate thioredoxin (Trx). MtTrxR is a promising dru

Full text of DOI:10.1021/jm3015734

Subscriber access provided by UQ Library
Article
Identification of M. tuberculosis thioredoxin reductase
inhibitors based on high-throughput docking using constraints
Oliver Koch, Timo Jaeger, Kristin Heller, Purushothama Chary Khandavalli,
Jette Pretzel, Katja Becker, Leopold Flohe, and Paul M. Selzer
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/jm3015734 • Publication Date (Web): 15 May 2013
Downloaded from http://pubs.acs.org on May 20, 2013
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Identification of M. tuberculosis thioredoxin
reductase inhibitors based on highꢀthroughput
docking using constraints
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Oliver Koch1,2,3*, Timo Jäger2,4, Kristin Heller2, Purushothama Chary Khandavalli2, Jette
Pretzel5, Katja Becker5, Leopold Flohé2,6, Paul M. Selzer1,7,8*
1 MSD Animal Health Innovation GmbH, Schwabenheim, Germany
2 MOLISA GmbH, Magdeburg, Germany
3Current address: Technische Universität Dortmund, Chemical Biology, Dortmund, Germany
4 Helmholtz Zentrum für Infektionsforschung, Braunschweig, Germany
5 Biochemistry and Molecular Biology, Justus Liebig University Giessen, Germany
6 OttoꢀvonꢀGuerickeꢀUniversität, Magdeburg, Germany
7
Universität Tübingen, Interfakultäres Institut für Biochemie, Tübingen, Germany
8 Wellcome Trust Centre for Molecular Parasitology and Division of Infection & Immunity,
University of Glasgow, Glasgow, UK
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 40
1
2
3
4
5
6
*Corresponding authors:
Paul M. Selzer
7
8
9
Tel: +49 6130ꢀ948ꢀ396.
Fax: +49 6130ꢀ948ꢀ517.
Eꢀmail: paul.selzer@msd.de
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
MSD Animal Health Innovation GmbH
Oliver Koch
Technische Universität Dortmund
Chemical Biology
Tel.: +49 231 133 2942
EꢀMail: oliver.koch@tuꢀdortmund.de
2
ACS Paragon Plus Environment

Page 3 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
Keywords: thioredoxin reductase inhibitors, Mycobacterium tuberculosis, proteinꢀprotein
7
8
9
interaction, highꢀthroughput docking,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Abstract
A virtual screening campaign is presented that led to small molecule inhibitors of thioredoxin
reductase of Mycobacterium tuberculosis (MtTrxR) which target the proteinꢀprotein interaction
site for the substrate thioredoxin (Trx). MtTrxR is a promising drug target, because it dominates
the Trxꢀdependent hydroperoxide metabolism and the reduction of ribonucleotides, thus
facilitating survival and proliferation of M. tuberculosis. Moreover, MtTrxR sufficiently differs
from its human homologs to suggest the possibility of selective inhibition, if the MtTrxRꢀTrx
interaction site is targeted. To this end, highꢀthroughput docking of 6.5 million virtual
compounds to the thioredoxin binding site of MtTrxR combined with constraints as filtering
steps was applied. 170 highꢀscoring compounds yielded 18 compounds that inhibited MtTrxR
with IC50 values up to the low micromolar range thus revealing that the proteinꢀprotein
interaction site of MtTrxR is indeed druggable. Most importantly, selectivity towards MtTrxR in
comparison to human TrxR (HsTrxR) is also demonstrated.
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 40
1
2
3
4
5
6
Introduction
7
8
9
Tuberculosis, thought to be largely defeated during the last century, still accounted for 1.7
million fatalities in 2009 (www.who.int/tb/), the vast majority of which occurred in the
developing world. Current firstꢀline treatment is based on only four different drugs, which by
now are more than 40 years old. A lack of novel drugs and the mandatory treatment duration of
at least 6 months inevitably led to strains with drug and multiꢀdrug resistance. With the
development and increasing spread of extensive drug resistance M. tuberculosis strains, which
often can no longer be treated by any of the available drugs1,2, tuberculosis is reꢀemerging as a
global threat3 that demands the search for new drugs with modes of action distinct from those of
known ones to minimize the risk of crossꢀresistance4.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To this end, the mycobacterial thioredoxin reductase (MtTrxR), here chosen as so far
unexploited drug target5, appeared to offer unique chances but was equally a challenge. In M.
tuberculosis, MtTrxR reduces the two typical thioredoxins B and C (MtTrxB and MtTrxC) at the
expense of NADPH. The functions of the pleiotropic thioredoxins comprise, inter alia, i) the
peroxiredoxinꢀmediated reduction of hydroperoxides and peroxynitrite considered to be pivotal
for the pathogen’s survival in macrophages6ꢀ10; and ii) likely, as in other species, the synthesis of
deoxyribonucleotides indispensable for DNA synthesis and, thus, for proliferation.11,12
Expectedly, MtTrxR was genetically validated as an essential gene/protein by transposon
analysis.13 Moreover, in other bacteria such as S. aureus14 or N. gonorrhoeae15, TrxR was also
shown to be essential for growth. Thus, TrxR inhibitors should compromise multiple pathways in
M. tuberculosis and, after elucidation of TrxR structures of several species, MtTrxR has become
a most promising target for structureꢀbased drug design16ꢀ18
.
4
ACS Paragon Plus Environment

Page 5 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The primary challenge inherent to this approach is how to inhibit MtTrxR selectively. The
human and other mammalian hosts of M. tuberculosis contain 3 homologous thioredoxin
reductases that fuel similar and equally important metabolic pathways11 and therefore must not
be simultaneously inhibited. Fortunately, all human thioredoxin reductases (HsTrxR1ꢀ3) differ
from the bacterial ones in one important aspect; their catalytic cycle involves a minimum of 3
redox centres: i) the NADPH binding site with the flavin cofactor, ii) the associated central
dithiolꢀdisulfide relay, and iii) a flexible arm with a terminal cysteineꢀselenocysteineꢀglycine
(CUG) motif19, which transfers the reduction equivalents from the central dithiolꢀdisulfide relay
to the CxxC motif of Trx20ꢀ22. In contrast, in the low molecular mass TrxRs of bacteria including
MtTrxR, the first two redox centres are conserved, whereas the selenocysteineꢀcontaining arm is
missing and the substrate Trx has to interact directly with the central dithiolꢀdisulfide relay,
which in mammalian TrxRs is inaccessible by Trx.23 Therefore, the TrxRꢀTrx interaction site in
the enzyme of M. tuberculosis differs and can be considered unique enough for selective
targeting.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
However, this encouraging perspective implies another, a technical challenge: the MtTrxs bind
to MtTrxR at a proteinꢀprotein interaction site which is considered not to be easily druggable.
Accordingly, commercial libraries generally show a low hitꢀrate regarding proteinꢀprotein
interaction inhibition.24 In order to increase the chance of detecting lowꢀmolecular weight
inhibitors, a targetꢀfocused library was created using an exhaustive inꢀsilico screening campaign
that screened a huge chemical space of 6.5 million molecules. The initial screening step here
applied, a novel highꢀthroughput docking approach that was complemented with constraints, a
priori biased the selection towards compounds with high affinity to the Trx binding site of
MtTrxR. By means of unbiased reꢀdocking, several filtering steps, scoring, and final consensus
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 40
1
2
3
4
5
6
7
8
9
scoring, the huge number of compounds was reduced to 170 candidates for real testing of
inhibitory activity. Of those, 18 were proved to inhibit MtTrxR, revealing the efficiency of the
inꢀsilico screening strategy and the druggability of a proteinꢀprotein interaction phase if a large
chemical space is explored. Moreover, the inhibitors indeed appear to interact with the Trx
binding site typical of the bacterial TrxR, since they do not affect HsTrxR.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Results
Structural comparison of EcTrxR and MtTrxR.
The bacterial TrxR is built up from two domains: an NADPHꢀbinding domain and an FADꢀ
binding domain. There is also ample evidence for substantial conformational change; the
NADPH domain rotates approximately 67° during catalysis, which complicates the identification
of substrate interaction sites.23 Most of the available Xꢀray structures (see experimental section)
disclose the FO conformation (FAD and disulphide sideꢀbyꢀside for flavin Oxidation; Figure 1 a
and b), in which typically the central redox centre is present as a buried disulphide between
Cys135 and Cys138 to be reduced by the FADH2 cofactor. Pdb 1f6m is the only Xꢀray structure
that shows the FR conformation (FAD and NADPH sideꢀbyꢀside for flavin Reduction; Figure 1 c
and d). 25 Here the active site cysteines are reduced and surfaceꢀexposed for the reduction of Trx,
while the oxidized flavin is in an optimum position for reduction by NADPH.25 Moreover, this
structure represents a deadꢀend intermediate of the catalytic cycle, revealing the binding mode of
Trx to a mutated EcTrxR (EcTrxRC135S). In this mutant, TrxR Cys 138 can still react with the
exposed cysteine of Trx (Cys32), but the downstream catalysis via thiol disulphide exchange is
no longer feasible, and Trx is arrested in its typical binding mode.
6
ACS Paragon Plus Environment

Page 7 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The most prominent interactions are seen between a Trx loop (Tyr70ꢀIle75) and a
complementary cleft on the EcTrxR surface, between Phe141 and Phe142 of the TrxR and a
hydrophobic pocket on EcTrx, a hydrogen bond between the Arg73 side chain of Trx and the
TrxR backbone carbonyl group of Ala237, and a hydrogen bond between the Asp139 side chain
of EcTrxR and the backbone amide of the Trx Ile75 (Figure 2 a and b). The importance of Arg73
in EcTrx has been validated by a loss of activity due to its mutation to Gly and Asp.26 This
presumed Trx binding site is very similar in the FO and FR conformation, but in the FR
conformation Trx access is blocked by the rotating FAD domain.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Creating a model for the FR form of MtTrxR.
The overlay of EcTrxR25 and MtTrxR27 reveals high structural similarity in the Trx binding
clefts (Figure 2 b/e). In fact, an almost identical binding mode of Trx to TrxR in M. tuberculosis
can be expected, which is characterized by hydrogen bonding to the corresponding residues
Thr242 and Asp149. Therefore, the backbone carbonyl group of Thr242 and the carboxylate of
Asp149, separated by a hydrophobic binding cleft (Figure 2e), were chosen as primary target
points for putative inhibitors. Unfortunately, this binding cleft and Asp149 are not completely
accessible in the only MtTrxR structure available, an FO conformation in which the FAD domain
partially blocks the Trx binding region. However, comparison of the FR and FO structures of
EcTrxR and MtTrxR reveals that the FAD domain simply rotates away while binding Trx,
without otherwise affecting the Trx binding site. This allows creating a model of MtTrxR that
mimicked its FR form by just removing the FAD domain (see Figure 2c and d). Additional
flexibility was introduced by allowing different conformers for Asp149 during docking, since
Asp 149 can adapt for optimal hydrogen bonding during ligand binding.
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 40
1
2
3
4
From a virtual library to active compounds
5
6
7
8
The initial highꢀthroughput docking of 6,433,706 compounds of a virtual library of
purchasable compounds yielded 151,768 structures that might bind to MtTrxR via the hydrogen
bonds suggested to be mandatory by the MtTrxR model outlined above (see Figure 3). All other
compounds were either physically unable to fulfil the constraints (e.g. only one Hꢀbond donor),
or the existing hydrogen bonds were too weak to pass the constraint filter with a minimum Hꢀ
bond geometry value of 0.8. This still unrealistically high number of possibly active compounds
was then further reduced to less than 10% via unrestrained reꢀdocking with exhaustive settings in
order to obtain reliable docking poses and then filtered again using rescoring with the same
hydrogen bonding constraints as in the first filter step. Depending on the scoring function,
different numbers of virtual hits remained: 7,334 by Goldscore, 9,931 by ASPscore, and 12,684
by Chemscore (Figure 3 b).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The use of a normalization function28, which was applied on each docking pose, led to a
broader and more balanced distribution of the molecular masses of the topꢀranking compounds
(data not shown) in comparison to the original scoring, because scoring functions alone tend to
prefer compounds with high molecular mass, while small molecules are underrepresented.
Finally, consensus scoring was applied as another approach which is especially used, in cases
where no data is available as to which scoring function performs best. As observed before
(www.ccdc.cam.ac.uk/products/life_sciences/gold/case_studies/ optimum_gold/), scoring by
Goldscore complies least with the results of Chemscore and ASPscore. Therefore, two different
combinations of the available scoring functions were used to rank the compounds based on a
consensus scoring by rank.29,30
8
ACS Paragon Plus Environment

Page 9 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The 200 topꢀranked compounds for the different consensus scoring strategies were chosen as
candidates for testing. Finally, 170 compounds were selected depending on scoring rank and ease
of availability and subjected to a primary screen for inhibition of MtTrxR. Out of these 170
compounds, 18 exhibited inhibition of MtTrxR. Table 1 shows the structures of representative
compounds that inhibited MtTrxR by at least 25% at a concentration of 45 ꢁM (Compounds 1ꢀ7
obtained from virtual screening each represent a distinct scaffold). As is evident from Table 2,
the concentration dependence of MtTrxR inhibition does not always follow expected rules, most
likely due to limited inhibitor solubility with increasing concentrations. Accordingly, IC50 values
could only be roughly extrapolated from inhibition observed at low inhibitor concentrations.
However, compound 1 (OK_TrxR_119) with an IC50 value of 12.5 ꢁM already reaches a specific
activity that appears promising enough for an optimization project of this 4ꢀaminoquinoline
scaffold.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The cluster of four compounds that share a 4ꢀ(piperidinꢀ1ꢀyl)aniline scaffold (see Table 1: 2,8ꢀ
10 and Table 3) and is represented by 2 (OK_TrxR_221) was selected for preliminary
optimization. Unrestrained reꢀdocking of all these compounds revealed binding to the preꢀ
selected Trx binding site of MtTrxR, taking advantage of hydrogen bonding to Thr242 and
Asp149 and a reasonable fit to the hydrophobic pocket in between (see Figure 4). The
4ꢀ(piperidinꢀ1ꢀyl)aniline scaffold was therefore used for synthesis of derivatives based on
structureꢀbased design and the assumed binding mode. The scaffold by itself (8) already showed
an IC50 of 33.4 ꢁM and two of the derivates (9 and 10) reached IC50 values of 15.3 and 15.8 ꢁM,
respectively (Table 4). The replacement of 4ꢀ(piperidinꢀ1ꢀyl)aniline by 4ꢀ(piperidinꢀ
1ꢀyl)benzylamine in all three compounds (8-10) leads to a complete loss of activity (data not
shown). This indicates that the aniline substructure containing a hydrogenꢀbond donor and an
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 40
1
2
3
4
5
6
aromatic ring as a hydrophobic group and capable of a πꢀπ stacking interaction is essential which
supports the assumed binding mode.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Selectivity of the TrxR inhibitors
The virtual screening was designed to exploit the differences in Trx binding between MtTrxR
and its human orthologues23 in order to identify inhibitors that selectively target the
mycobacterial enzyme. It was therefore attempted to investigate the identified inhibitors of
MtTrxR for inhibitory activity on one of its human congeners, the TrxR1 isolated from human
placenta (‘HsTrxR’) in two different test systems, using either a mutant version of the
physiological substrate human Trx1(TrxC73S; ‘HsTrx’) or the artificial acceptor substrate
DTNB. As high test compound concentrations were required, the evaluation of selectivity proved
to be an experimental challenge irrespective of the test systems. The aromatic nature of all test
compounds in some cases precluded reliable measurements at 100 ꢂM with the HsTrxꢀbased test
at a wave length of 344 nm. But also the DTNBꢀbased system with its more convenient read out
at 412 nm was sometimes disturbed by test compound concentrations near the solubility limit.
Apparent ‘enzyme activations’ (negative inhibition values in Tables 2 and 4) likely reflect little
else than timeꢀdependent changes in dispersity of system components which add an increment to
the absorbance change. Table 2 and 4 list the results for HsTrxR inhibition, as experimentally
obtained with the more reliable DTNB assay, whereby the apparent enzyme activations are rated
as ‘no inhibition’. With these precautions, it can safely be stated that only three of the MtTrxR
inhibitors (1, 3 and 9) marginally interfere with HsTrxR when tested at the maximum possible
concentration.
10
ACS Paragon Plus Environment

Page 11 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Due to the same limitations, only lower limits for selectivity indices (SI) can be estimated.
The values listed in Table 2 and 4 are based on the assumption that, presuming a typical
concentration response of a reversible inhibitor, an extrapolated IC50 for HsTrxR does not exceed
200 or 300 ꢂM, if a compound inhibits the enzyme by <30 or <10%, respectively, or less at 100
ꢂM. With these premises, all inhibitors here presented can clearly be rated as selective for
MtTrxR.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Discussion and Conclusion
The common goal of in silico screening campaigns is to reduce a huge number of available
compounds (e.g. the ZINC library (http://zinc.docking.org/) contains over 14 million purchasable
compounds) to a targetꢀfocused subset that can be tested for activity with reasonable efforts.31
This is commonly achieved by pharmacophore filtering, which preselects the hypothetical
candidates for drugꢀlike compounds that might bind to the target molecule.31 In the present
investigation, however, we faced the challenge preselecting compounds from the virtual library
that bind to the target MtTrxR in a specific way that minimizes the chance of binding to the
homologous human proteins. Moreover, the preꢀdetermined binding site was a proteinꢀprotein
interaction site which is not considered to be likely druggable. Under these premises, a
combination of highꢀthroughput docking and the use of strong constraints deemed superior to
conventional pharmacophore searches. While the latter hardly discriminate between weak and
strong hydrogen bonding interactions, the docking software GOLD36 allows to identify strong
hydrogen bonding interactions, in particular by choosing a high minimum Hꢀbond geometry
weight as a constraint (here 0.8 on a scale between 0 and 1). Thereby, only very strong hydrogen
bonding interactions were considered.
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 40
1
2
3
4
5
6
7
8
9
Despite the comparatively high computational power required for docking, the novel strategy
proved to be technically feasible. The fast settings applied in the initial docking run (see
Experimental Section) generated preliminary binding poses within ~13 sec per compound on an
AMD Opteron 250 processor with 2.4 GHz, and the computational power of the latest processors
should manage this task even faster. Depending on the scoring function applied, the huge number
of compounds of the starting database could be reduced to 0.11% and 0.20%, respectively, which
already corresponds to numbers that can be subjected to real screening. Thus, the highꢀ
throughput docking approach can, in principle, be used as an alternative to pharmacophore
filtering. As shown here, the number of candidates can be further reduced by means of consensus
scoring and normalization functions to an extent that allows highly economic testing in terms of
manpower and costs. Most importantly, a hit rate >10% compares very favourably to those
obtained for other proteinꢀprotein interfaces.32
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
It can only be speculated which of the individual screening and scoring steps had the highest
impact on this extraordinary number of hits, since activity testing was only performed with the
final compound selection. Likely, however, the initial highꢀthroughput docking already led to
substantial enrichment of active compounds, since the improvement of compound selections by
means of scoring and consensus scoring in the downstream part of the in silico campaign is a
commonly accepted and applied approach.29 The active compounds identified in this manner
were not yet rated as efficacious enough to justify an inꢀdepth characterization of their biological
activity. However, with IC50 values in the low micromolar range, the more active ones can be
considered promising enough to be used for a hitꢀtoꢀlead optimization project18 and,
interestingly, the promising hits have a comparatively low MW and could be considered as
fragments to be further modified. Our results, thus, indicate that the hits target a very important
12
ACS Paragon Plus Environment

Page 13 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
part of the interaction, a “hot spot”. Moreover, the hit validation of the compounds belonging to
the scaffold cluster 2 as well as the predicted binding modes (see Figure 4) suggests that this
compound class indeed targets the binding cleft where Arg 73 of MtTrx binds26. The selectivity
for MtTrxR, which is equally for the observed for the remaining inhibitors, justifies the
conclusion that the virtual screening strategy was successful in preꢀselecting the chemical space
for compounds that target the Trx binding cleft which is unique to the bacterial TrxRs.
In short, highꢀthroughput docking using constraints is shown to be a useful alternative for
pharmacophore filtering to handle million of compounds in a virtual screening. This approach
appears to be particularly helpful if the biomedical goal demands the exploitation of difficult
targets such as proteinꢀprotein interaction sites. The feasibility of this novel approach is
exemplified by an in silico campaign yielding the very first inhibitors of MtTrxR that specifically
target the proteinꢀprotein interaction of thioredoxins with the bacterial subfamily of thioredoxin
reductases, here specifically MtTrxR, which is distinct from the Trx interaction site of higher
organisms’ TrxRs. Simultaneously, a novel route is offered to meet the challenges of reꢀ
emerging tuberculosis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Experimental Section
Target structure of MtTrxR
Creating a target model for the FR form of MtTrxR was made feasible by comparing Xꢀray
structures of EcTrxR (FO form: pdb 1tde and 1tdf 33; FR form: Pdb 1f6m25), H. pylori (FO form :
pdb 2q0k and 2q0l34) and MtTrxR (FO form: pdb 2a8727). The FAD domain was removed from
the FO Xꢀray structure of MtTrxR yielding a preliminary ꢃ1ꢀ123, ꢃ251ꢀ335 MtTrxR structural
model in which the NADPH domain (Ala124ꢀHis250) comprising the selected Trx binding cleft
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 40
1
2
3
4
5
6
7
8
9
to be targeted exactly corresponded to pdb 2a87. The model of the truncated MtTrxR was
protonated using Protonate 3D from the program MOE35 and used as target structure
representing the FR form of MtTrxR.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Virtual screening workflow
The MtTrxR model, created as described above, was used for all docking runs. The Trx
binding site was defined to include all atoms within 15 Å of one of the aromatic hydrogen atoms
(HZ) of Phe153. The side chain of Asp149 was set flexible with six rotamers (see Table 5).
Docking was performed by means of the docking software GOLD36, and Goldscore was used as
scoring function.
For the initial highꢀthroughput docking of 6,433,706 structures, a maximum of ten different
poses were created with genetic algorithm settings set to automatic and 10% search efficiency.
The docking run for a single compound was stopped if three poses were within an RMSD (root
mean square deviation) of 1.5 Å (early termination option) or if it was physically impossible to
fulfil the constraints (part of the constraint option). A protein Hꢀbond constraint was set for
Asp149 (either one or the other oxygen of the side chain) and Thr242 (oxygen of the backbone
carbonylꢀgroup). A minimum Hꢀbond geometry value of 0.8 representing good Hꢀbond geometry
(“perfect geometry” would be a value of 1) and strong hydrogen bonding was empirically
identified. A penalty value (here 50) was added to the final score if a constraint did not fulfil this
minimum Hꢀbond geometry. Poses not passing this hydrogen bonding constraint could be easily
identified, since this penalty value was added to the docking output. Docking time per compound
averaged around 13s.
14
ACS Paragon Plus Environment

Page 15 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
For the reꢀdocking step, docking settings were changed to computationally more expensive
ones. Fifteen docking poses were created using genetic algorithm settings set to automatic and
100% search efficiency. The protein Hꢀbond constraints and the early termination option were
not used during reꢀdocking. The rest of the settings were identical to the initial highꢀthroughput
docking run. Docking time was ~270 s per compound. In order to identify the poses forming the
mandatory hydrogen bonds to Thr242 and Asp139, all poses were reꢀscored by the scoring
functions Goldscore36, Chemscore37,38 and ASPscore39 combined with the protein Hꢀbond
constraints already used in the first docking run. During rescoring, the docking poses were
minimized to the nearest local minimum of the respective scoring function. Inꢀhouse python
scripts were used to parse the output files for the constraint penalties and to extract the
compounds for the next step.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The final ranking for compound selection was based on a normalization and consensus scoring
strategy. A normalizing function described by Carta et al.28 was applied to yield a broader
distribution of the molecular weight within the topꢀranked compounds. For final consensus
scoring, two different combinations of scoring functions were applied. One consensus scoring is
based on all three scoring functions (see above), and the other is based on the combination of
29
Chemscore and ASPscore. The final ranking was based on a “rankꢀbyꢀrank” strategy , where
the averages of rank numbers were used. After removing duplicates, the 200 topꢀranked
compounds from both strategies were selected as candidates for testing.
Virtual library of purchasable compounds
The underlying virtual library consists of combined libraries from various commercial vendors
(Akos Consulting & Solutions GmbH, Basel, Switzerland; Asinex Ltd., Moscow, Russia;
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 40
1
2
3
4
5
6
7
8
9
ChemBridge Corporation, San Diego, CA, USA; Chemical Diversity Labs Inc., San Diego, CA,
USA; Enamine Ltd., Kiev, Ukraine; InterBioScreen, Moscow, Russia; LifeChemicals Inc.,
Burlington, ON, Canada; Maybridge, Cambridge, UK; Otava, Kiev, Ukraine; Specs, Delft,
Netherlands; TimTec Corp., Newark, NJ, USA; VitasꢀM Laboratory Ltd., Moscow, Russia) and
collectively contained 6,433,706 virtual compounds.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Screening of potential inhibitors against MtTrxR
MtTrxR activity was determined essentially as described in Jäger et al. (2004)9. The inhibitory
effect of each compound on MtTrxR was measured in triplicates. Obvious outliers from the
triplicates were removed. If the standard deviation of the two remaining inhibition values
exceeded 25%, the measurement was repeated.
The activity of MtTrxR was obtained by measuring initial velocities monitoring the NADPH
oxidation at 340 nm at 25 °C. 10 ꢁM TrxR was incubated with 10 ꢁM MtTrxB, 5 ꢁM MtTPx,
225 ꢁM NADPH, 0.06 ꢀ 45 ꢁM compound, 73 ꢁM tꢀbutylhydroperoxide (added after the preꢀ
incubation step) in a total volume of 100 ꢁl 50 mM HEPES pH 7.4, 1 mM EDTA for 10 min at
25 °C and started by adding hydroperoxide. No compound was present in the positive controls.
No MtTrxR was present in the negative controls. A Tecan Freedom EVO® liquid handling
platform was used for pipetting, NADPH consumption at 340 nm was measured with a
TecanGENios Pro microplate reader after adding tꢀbutylhydroperoxide. The enzymes MtTrxR,
MtTrxB, and MtTPx from M. tuberculosis were produced by heterologous expression in E.coli as
described9. NADPH, HEPES and EDTA were obtained from Carl Roth (Karlsruhe, Germany), tꢀ
butylhydroperoxide from SigmaꢀAldrich (St. Louis, MO).
16
ACS Paragon Plus Environment

Page 17 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
If not otherwise stated, IC50 values were calculated using the fourꢀparameter equation model
205 and the option "unlock" from the XLfit addꢀin (IDBS, Guildford, United Kingdom) in Excel
(Microsoft Corporation, Redmond, WA). All values are mean values from at least three
independent assays. If atypical concentration dependence at high inhibitor concentrations was
observed, estimates of IC50 values were obtained by graphically extrapolating from inhibition
data at low inhibitor concentrations assuming regular concentration responses of reversible
inhibitors (Hillꢀcoefficient = 1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Inhibition studies with HsTrxR
NADPH and 5,5'ꢀdithioꢀbisꢀ2ꢀnitrobenzoic acid (DTNB) were obtained from SigmaꢀAldrich.
Human placenta thioredoxin reductase 1 (‘HsTrxR’) and the recombinant human thioredoxin
mutant Cys73→Ser (‘HsTrx’), which does not form dimers and is stable in enzymatic assays,
were produced as described previously.40,41
Both DTNB (5,5'ꢀdithioꢀbisꢀ2ꢀnitrobenzoic acid; SigmaꢀAldrich, St Louis, MO) and Trx (as
physiological substrate) reduction assays were used to determine the effects of the selected
compounds on HsTrxR activity.40,41 All assays were conducted in 100 mM potassium phosphate,
2 mM EDTA, pH 7.4 in a total assay volume of 500 ꢁl. HsTrxR (final concentration 2.8 nM) and
200 ꢁM NADPH (SigmaꢀAldrich, St Louis, MO) were added to the buffer before the inhibitors
(up to 100 ꢁM final concentration) were added and the reaction was started with either DTNB (3
mM) or with HsTrx (20 ꢁM). The change in absorbance was monitored at 412 nm (for DTNB; ε
= 13,600 M−1cm−1) or at 340 nM (for the NADPHꢀdependent hTrx assay; ε = 6.22 M−1cm−1)
using a HITACHI Uꢀ2001 spectrophotometer. All enzyme activity assays and kinetic studies
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 40
1
2
3
4
5
6
were carried out in at least three independent experiments at 25 C. Control assays (100%
activity) were conducted in the absence of inhibitors.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Purchased compounds
Commercially available compounds were purchased from following suppliers. 1 was ordered
as 4ꢀ[(6ꢀmethoxyꢀ2ꢀmethylquinolinꢀ4ꢀyl)amino]phenol from InterBioscreen, Moscow, Russia. 2
was
ordered
as
Nꢀ[4ꢀ(4ꢀ{[2ꢀ(methylthio)ethyl]amino}ꢀ1ꢀpiperidinyl)phenyl]ꢀ5ꢀ
phenylpentanamide from ChemBridge Corporation, San Diego, CA, USA. 3 was ordered as 3ꢀ
(1Hꢀindolꢀ2ꢀyl)ꢀNꢀ(1Hꢀindolꢀ3ꢀylmethyl)alanine from VitasꢀM Laboratory Ltd., Moscow,
Russia. 4 was ordered as N'ꢀ{(E)ꢀ[2ꢀhydroxyꢀ3ꢀ(propꢀ2ꢀenꢀ1ꢀyl)phenyl]methylidene}ꢀ3ꢀ(3ꢀ
methylꢀ5ꢀoxoꢀ4,5ꢀdihydroꢀ1Hꢀpyrazolꢀ4ꢀyl)propaneꢀhydrazide from Otava, Kiev, Ukraine. 5 was
ordered as Nꢀphenylꢀ3ꢀ{1ꢀ[(3ꢀphenylꢀ1Hꢀpyrazolꢀ4ꢀyl)methyl]ꢀ4ꢀpiperidinyl}propanamide from
ChemBridge Corporation, San Diego, CA, USA. 6 was ordered as 3ꢀ(2ꢀaminoꢀ1,3ꢀthiazolꢀ4ꢀyl)ꢀ
7ꢀhydroxyꢀ2Hꢀchromenꢀ2ꢀone from InterBioscreen, Moscow, Russia. 7 was ordered as 4ꢀ[5ꢀ
aminoꢀ3ꢀoxoꢀ4ꢀ(4ꢀphenylꢀ1,3ꢀthiazolꢀ2ꢀyl)ꢀ2,3ꢀdihydroꢀ1Hꢀpyrrolꢀ1ꢀl]benzenesulfonamide from
InterBioscreen, Moscow, Russia. 8 was ordered as Nꢀ[4ꢀ(4ꢀ{[2ꢀ(3ꢀpyridinyloxy)ethyl]amino}ꢀ1ꢀ
piperidinyl)phenyl]ꢀ2ꢀpyrazinecarboxamide from ChemBridge Corporation, San Diego, CA,
USA.
9
was ordered as 2ꢀphenoxyꢀNꢀ[4ꢀ(4ꢀ{[3ꢀ(1Hꢀpyrazolꢀ1ꢀyl)benzyl]amino}ꢀ1ꢀ
piperidinyl)phenyl]acetamide from ChemBridge Corporation, San Diego, CA, USA. 10 was
ordered as Nꢀ[4ꢀ[4ꢀ(2ꢀpyridinꢀ3ꢀyloxyethylamino)piperidinꢀ1ꢀyl]phenyl]pyrazineꢀ2ꢀcarboxamide
from ChemBridge Corporation, San Diego, CA, USA. According to the suppliers’ information,
the compounds were 92ꢀ98% pure.
18
ACS Paragon Plus Environment

Page 19 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Chemical synthesis
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Solvents and chemical reagents were purchased from chemical suppliers and used without any
further purification. Dry solvents were generally purchased in sure seal bottles stored over
molecular sieves. Unless otherwise stated, all reactions were carried out under dry nitrogen
atmosphere. Progress of the reactions was monitored by thin layer chromatography (TLC)
(Merck silica gel 60F254 plates). After elution, a TLC plate was visualized under UV illumination
at 254 nm for UVꢀactive materials. Standard staining reagents were used for visualization. Flash
1
column chromatography was carried out using silica gel finer than 230ꢀ400 mesh. H NMR
spectra were recorded on a Brucker 600 or 400 MHz spectrometer in appropriate solvent using
13
tetramethylsilane (TMS) as internal standard. C NMR spectra were recorded on a 150 MHz
spectrometer. The chemical shifts are shown in δ ppm.
Qualitative and quantitative purity of the all isolated compounds was established by means of
13
1H NMR and C NMR spectroscopic analysis; also, analytical HPLC/MS (Agilent 1200 Series
HPLC equipped with Cꢀ18 column, coupled with a Sedexꢀ85, LTꢀELSD detector) was used to
demonstrate the purity of the compounds. All compounds synthesized were at least 95% pure
according to HPLC/MS chromatography.
Preparation of 4-(piperidin-1-yl)aniline (11). EtOH (64 ml) was added to 4ꢀ(piperidinꢀ1ꢀ
yl)nitrobenzene (5.21 g, 25.26 mmol) taken into a 250 ml twoꢀnecked, round bottom flask
equipped with water condenser and stirred at room temperature to generate a homogeneous
solution. Water (44 ml), Fe(s) powder (10.72 g, 191.94 mmol), and finally AcOH (8.60 ml) were
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 40
1
2
3
4
5
6
7
8
9
successively added to the reaction. The reaction mixture was heated to reflux under dry nitrogen
atmosphere, brought to room temperature after two hours, then diluted with EtOAc (50 ml) and
vigorously stirred. The resulting suspension was filtered using a short bed of CeliteꢀR545 over
sand. Combined extracts were separated from water. Organic portions were collected and dried
over anhydrous MgSO4, filtered, and concentrated. A short flash chromatographic purification on
silica gel using CH2Cl2/MeOH, (9:1) eluent system, yielded the desired product 11 (4.22 g, 95%
isolated yield). The spectral data of the compound compiled reported literature values.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
General experimental procedure for the preparation of 4-(Piperidin-1-yl)aniline scaffold
compounds. 4ꢀ(piperidinꢀ1ꢀyl)aniline (11; Figure 5) was added to a 1,2ꢀdichloroethane (DCE) (4
ml/mmol) solution of the aldehyde (1 eq) at room temperature, and finally Na(OAc)3BH (1.30
eq) was added. The reaction was performed at room temperature under nitrogen atmosphere.
After completion, saturated aqueous NaHCO3 was added and the reaction mixture was diluted
with EtOAc and vigorously stirred. Layers were separated and the aqueous phase was extracted
with EtOAc. Combined organic portions were dried over anhydrous MgSO4, filtered and
concentrated in vacuo. Flash chromatographic purification on silica gel using CH2Cl2/MeOH,
(15:1 to 4:1) eluent system yielded the desired product.
N-(1,3-benzothiazol-2-ylmethyl)-4-(piperidin-1-yl)aniline (12). Following the general
experimental protocol 10 was prepared in 50% isolated yield. Rf= 0.24 (CH2Cl2/MeOH 15:1); 1H
NMR (400 MHz, DMSOꢀd6): δ = 1.44ꢀ 1.51 (m, 2H), 1.54ꢀ1.61 (m, 4H), 2.97ꢀ3.04 (dd, J = 5.27,
5.69 Hz, 4H), 4.86 (d, J =5.69 Hz, 2H), 6.32 (t, J = 5.48, 6.54 Hz, 1H), 6.83 (d, J = 9.07 Hz, 2H),
7.35ꢀ7.42 (m, 3H), 7.48 (dd, J = 6.97, 7.42 Hz, 1H), 7.90 (d, J = 8.44 Hz, 1H), 8.07 (d, 7.80 Hz,
20
ACS Paragon Plus Environment

Page 21 of 40
Journal of Medicinal Chemistry
1
2
13
3
1H); C NMR (150 MHz, DMSOꢀd6): δ = 23.92, 24.01, 25.51, 50.37, 61.43, 116.44, 120.34,
120.36, 122.44, 124.89, 126.21, 131.28, 134.35, 148.07, 153.22, 167.86, 175.84. HPLC/MS:
C19H21N3S, [M+H]+ requires 324.15; found 324.20.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
N-(2-chloro-6-fluorobenzyl)-4-(piperidin-1-yl)aniline
(13).
Following
the
general
experimental protocol 13 was prepared in 55% isolated yield. Rf = 0.56 (CH2Cl2/MeOH 9:1); 1H
NMR (400 MHz, DMSOꢀd6): δ = 1.36 – 1.48 (m, 2H), 1.51 ꢀ 1.62 (m, 4H), 2.75 – 2.90 (m, 4H),
4.17 – 4.29 (m, 2H), 5.22 – 5.36 (m, 1H), 6.53 – 6.64 (d, J = 7.40 Hz, 2H), 6.68 – 6.78 (d, J =
13
7.90 Hz, 2H), 7.16 – 7.22 (m, 1H), 7.27 – 7.37 (m, 3H); C NMR (150 MHz, DMSOꢀd6): δ =
23.95, 25.84, 52.04, 54.01, 113.17, 114.51, 114.54, 118.73, 118.79, 125.66, 130.06, 130.13,
135.04, 142.36, 143.97, 160.62, 162.26. HPLC/MS: C18H20ClFN2, [M+H]+ requires 319.13;
found 319.15.
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 40
1
2
3
4
5
6
7
8
Ancillary Information
*Corresponding authors:
Paul M. Selzer
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Tel: +49 6130ꢀ948ꢀ396.
Fax: +49 6130ꢀ948ꢀ517.
Eꢀmail: paul.selzer@msd.de
MSD Animal Health Innovation GmbH
Oliver Koch
Technische Universität Dortmund
Chemical Biology
Tel.: +49 231 133 2942
EꢀMail: oliver.koch@tuꢀdortmund.de
Supporting Information Available: The supporting information contains representative dose
response curves for IC50 calculation. and virtual screening ranking information. This material is
available free of charge via the Internet at http://pubs.acs.org.
22
ACS Paragon Plus Environment

Page 23 of 40
Journal of Medicinal Chemistry
1
2
3
4
Funding sources
5
6
7
8
The study was supported by the Deutsche Forschungsgemeinschaft (BE1540/11ꢀ2) and the
Bundesministerium für Bildung und Forschung (0315220A).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Abbreviations used:
DTNB: 5,5'ꢀdithioꢀbisꢀ2ꢀnitrobenzoic acid
HEPES: 2ꢀ(4ꢀ(2ꢀHydroxyethyl)ꢀ 1ꢀpiperazinyl)ꢀethansulfonsäure
Hs: Homo sapiens
Mt: Mycobacterium tuberculosis
TPx: Thioredoxin Peroxidase
TrxR: Thioredoxin Reductase
Trx: Thioredoxin
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 40
1
2
3
4
References
5
6
7
8
1. Aziz, M. A.; Wright, A.; Laszlo, A.; De Muynck, A.; Portaels, F.; Van Deun, A.; Wells,
C.; Nunn, P.; Blanc, L.; Raviglione, M. Epidemiology of antituberculosis drug
resistance (the Global Project on Antiꢀtuberculosis Drug Resistance Surveillance): an
updated analysis. Lancet 2006, 368, 2142ꢀ2154.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2. WHO. Multidrug and extensively drugꢀresistant TB (M/XDRꢀTB): Multidrug and
extensively drugꢀresistant TB (M/XDRꢀTB): 2010 global report on surveillance and
response.WHO/HTM/TB: Geneva, Switzerland, 2010, 1ꢀ71.
3. WHO. XDRꢀTB ꢀ a global threat. Lancet 2006, 368, 964.
4. Koch, O.; Selzer, P. M. Biologie der Mykobakterien und neue molekulare Targets.
Pharm. Unserer Zeit. 2012, 41, 19ꢀ26.
5. Jaeger, T.; Flohé, L. The thiolꢀbased redox networks of pathogens: unexploited targets in
the search for new drugs. Biofactors 2006, 27, 109ꢀ120.
6. Bryk, R.; Griffin, P.; Nathan, C. Peroxynitrite reductase activity of bacterial
peroxiredoxins. Nature 2000, 407, 211ꢀ215.
7. Hu, Y.; Coates, A. R. Acute and persistent Mycobacterium tuberculosis infections depend
on the thiol peroxidase TpX. PLoS One 2009, 4, e5150.
8. Jaeger, T. Peroxiredoxin systems in mycobacteria. Subcell. Biochem. 2007, 44, 207ꢀ217.
9. Jaeger, T.; Budde, H.; Flohé, L.; Menge, U.; Singh, M.; Trujillo, M.; Radi, R. Multiple
thioredoxinꢀmediated routes to detoxify hydroperoxides in Mycobacterium
tuberculosis. Arch. Biochem. Biophys. 2004, 423, 182ꢀ191.
10. Master, S. S.; Springer, B.; Sander, P.; Boettger, E. C.; Deretic, V.; Timmins, G. S.
Oxidative stress response genes in Mycobacterium tuberculosis: role of ahpC in
24
ACS Paragon Plus Environment

Page 25 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
resistance to peroxynitrite and stageꢀspecific survival in macrophages. Microbiology
2002, 148, 3139ꢀ3144.
7
8
9
11. Arner, E. S.; Holmgren, A. Physiological functions of thioredoxin and thioredoxin
reductase. Eur. J. Biochem. 2000, 267, 6102ꢀ6109.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
12. Zahedi Avval, F.; Holmgren, A. Molecular mechanisms of thioredoxin and glutaredoxin
as hydrogen donors for mammalian S phase ribonucleotide reductase. J. Biol. Chem.
2009, 284, 8233ꢀ8240.
13. Sassetti, C. M.; Rubin, E. J. Genes required for mycobacterial growth defined by high
density mutagenesis. Mol. Microbiol. 2003, 48, 77ꢀ84.
14. Uziel, O.; Borovok, I.; Schreiber, R.; Cohen, G.; Aharonowitz, Y. Transcriptional
regulation of the Staphylococcus aureus thioredoxin and thioredoxin reductase genes in
response to oxygen and disulfide stress. J. Bacteriol. 2004, 186, 326ꢀ334.
15. Potter, A. J.; Kidd, S. P.; Edwards, J. L.; Falsetta, M. L.; Apicella, M. A.; Jennings, M.
P.; McEwan, A. G. Thioredoxin reductase is essential for protection of Neisseria
gonorrhoeae against killing by nitric oxide and for bacterial growth during interaction
with cervical epithelial cells. J. Infect. Dis. 2009, 199, 227ꢀ235.
16. Marhöfer, R.J.; Oellien, F.; Selzer, P.M. Drug discovery and the use of computational
approaches for infectious diseases. Future Med. Chem. 2011, 3, 1011ꢀ1125.
17. Krasky, A.; Rohwer, A.; Marhöfer, R.; Selzer, P.M. Bioinformatics and
Chemoinformatics: Key Technologies in the Drug Discovery Process. In Antiparasitic
and Antibacterial Drug Discovery; Selzer, P., Ed.; WileyꢀVCH, Weinhein, Germany,
2009; pp 45ꢀ58.
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 40
1
2
3
4
5
6
7
8
9
18. Rohwer, A.; Marhöfer, R.J.; Caffrey, C.R.; Selzer, P.M. Drug Discovery approaches
toward antiꢀparasitic agents. In Apicomplexan Parasites: Molecular Approaches
toward Targeted Drug Development; Becker, K., Ed.; WileyꢀVCH, Weinhein,
Germany, 2011; pp 3ꢀ20.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
19. Kryukov, G. V.; Castellano, S.; Novoselov, S. V.; Lobanov, A. V.; Zehtab, O.; Guigo, R.;
Gladyshev, V. N. Characterization of mammalian selenoproteomes. Science 2003, 300,
1439ꢀ1443.
20. Gromer, S.; Wissing, J.; Behne, D.; Ashman, K.; Schirmer, R. H.; Flohé, L.; Becker, K.
A hypothesis on the catalytic mechanism of the selenoenzyme thioredoxin reductase.
Biochem. J. 1998, 332, 591ꢀ592.
21. Birringer, M.; Pilawa, S.; Flohé, L. Trends in selenium biochemistry. Nat. Prod. Rep.
2002, 19, 693ꢀ718.
22. Tatyana, S.; Zhong, L.; Lindqvist, Y.; Holmgren, A., and Schneider, G.. Threeꢀ
dimensional structure of a mammalian thioredoxin reductase: Implications for
mechanism and evolution of a selenocysteineꢀdependent enzyme. Proc. Natl. Acad. Sci.
U. S. A. 2001, 98, 9533–8.
23. Williams, C. H.; Arscott, L. D.; Müller, S.; Lennon, B. W.; Ludwig, M. L.; Wang, P. F.;
Veine, D. M.; Becker, K.; Schirmer, R. H. Thioredoxin reductase two modes of
catalysis have evolved. Eur. J. Biochem. 2000, 267, 6110ꢀ6117.
24. Sperandio, O.; Reynès, C.H.; Camproux, A.C.; Villoutreix, B.O. Rationalizing the
chemical space of proteinꢀprotein interaction inhibitors. Drug Discovery Today 2010,
15, 220ꢀ229.
26
ACS Paragon Plus Environment

Page 27 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
25. Lennon, B. W.; Williams, C. H., Jr.; Ludwig, M. L. Twists in catalysis: alternating
conformations of Escherichia coli thioredoxin reductase. Science 2000, 289, 1190ꢀ
1194.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
26. Negri, A.; RodriguezꢀLarrea, D.; Marco, E.; JimenezꢀRuiz, A.; SanchezꢀRuiz, J. M.;
Gago, F. Proteinꢀprotein interactions at an enzymeꢀsubstrate interface: characterization
of transient reaction intermediates throughout a full catalytic cycle of Escherichia coli
thioredoxin reductase. Proteins: Struct., Funct., Bioinf. 2010, 78, 36ꢀ51.
27. Akif, M.; Suhre, K.; Verma, C.; Mande, S. C. Conformational flexibility of
Mycobacterium tuberculosis thioredoxin reductase: crystal structure and normalꢀmode
analysis. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2005, 61, 1603ꢀ1611.
28. Carta, G.; Knox, A. J.; Lloyd, D. G. Unbiasing scoring functions: a new normalization
and rescoring strategy. J. Chem. Inf. Model. 2007, 47, 1564ꢀ1571.
29. Oda, A.; Tsuchida, K.; Takakura, T.; Yamaotsu, N.; Hirono, S. Comparison of consensus
scoring strategies for evaluating computational models of proteinꢀligand complexes. J.
Chem. Inf. Model. 2006, 46, 380ꢀ391.
30. Verdonk, M. L.; Berdini, V.; Hartshorn, M. J.; Mooij, W. T.; Murray, C. W.; Taylor, R.
D.; Watson, P. Virtual screening using proteinꢀligand docking: avoiding artificial
enrichment. J. Chem. Inf. Comput. Sci. 2004, 44, 793ꢀ806.
31. Klebe, G. Virtual screening: strategies, perspectives and limitations. Drug Discovery
Today 2006, 11, 580ꢀ594.
32. Gautier, B.; Miteva, M. A.; Goncalves, V.; Huguenot, F.; Coric, P.; Bouaziz, S.; Seijo,
B.; Gaucher, J.ꢀF.; Broutin, I.; Garbay, C.; Lesnard, A.; Rault, S.; Inguimbert, N.;
Villoutreix, B.O.; Vidal, M. Targeting the proangiogenic VEGFꢀVEGFR proteinꢀ
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 40
1
2
3
4
5
6
protein interface with drugꢀlike compounds by in silico and in vitro screening. Chem.
Biol. 2011, 18, 1631ꢀ1639.
7
8
9
33. Waksman, G.; Krishna, T. S.; Williams, C. H., Jr.; Kuriyan, J. Crystal structure of
Escherichia coli thioredoxin reductase refined at 2 A resolution. Implications for a
large conformational change during catalysis. J. Mol. Biol. 1994, 236, 800ꢀ816.
34. Gustafsson, T. N.; Sandalova, T.; Lu, J.; Holmgren, A.; Schneider, G. Highꢀresolution
structures of oxidized and reduced thioredoxin reductase from Helicobacter pylori.
Acta Crystallogr., Sect. D: Biol. Crystallogr. 2007, 63, 833ꢀ843.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
35. Molecular Operating Environment (MOE), 2011.10; Chemical Computing Group Inc.,
1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada, H3A 2R7, 2011.
36. Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor, R. Development and validation
of a genetic algorithm for flexible docking. J. Mol. Biol. 1997, 267, 727ꢀ748.
37. Eldridge, M. D.; Murray, C. W.; Auton, T. R.; Paolini, G. V.; Mee, R. P. Empirical
scoring functions: I. The development of a fast empirical scoring function to estimate
the binding affinity of ligands in receptor complexes. J. Comput. Aided Mol. Des. 1997,
11, 425ꢀ445.
38. Baxter, C. A.; Murray, C. W.; Clark, D. E.; Westhead, D. R.; Eldridge, M. D. Flexible
docking using tabu search and an empirical estimate of binding affinity. Proteins:
Struct., Funct., Bioinf. 1998, 33, 367ꢀ382.
39. Mooij, W. T.; Verdonk, M. L. General and targeted statistical potentials for proteinꢀ
ligand interactions. Proteins: Struct., Funct., Bioinf. 2005, 61, 272ꢀ287.
28
ACS Paragon Plus Environment

Page 29 of 40
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
40. Gromer, S.; Arscott, L. D.; Williams, CH. Jr.; Schirmer, R. H.; Becker, K. Human
placenta thioredoxinreductase. Isolation of the selenoenzyme, steady state kinetics, and
inhibition by therapeutic gold compounds. J. Biol. Chem. 1998, 273, 20096ꢀ20101.
41. Urig, S.; Lieske, J.; FritzꢀWolf, K.; Irmler, A.; Becker, K. Truncated mutants of human
thioredoxin reductase 1 do not exhibit glutathione reductase activity. FEBS Lett. 2006,
580, 3595ꢀ3600.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
29
ACS Paragon Plus Environment