Optimization of Triazine Nitriles as Rhodesain Inhibitors: Structure-Activity Relationships, Bioisosteric Imidazopyridine Nitriles, and X-ray Crystal Structure Analysis with Human CathepsinL

Article abstract of DOI:10.1002/cmdc.201300112

The cysteine protease rhodesain of Trypanosoma brucei parasites causing African sleeping sickness has emerged as a target for the development of new drug candidates. Based on a triazine nitrile moiety as electrophilic headgroup, optimization studies on the substituents for the S1, S2, and S3 pockets of the enzyme were performed using structure-based design and resulted in inhibitors with inhibition constants in the single-digit nanomolar range. Comprehensive structure-activity relationships clarified the binding preferences of the individual pockets of the active site. The S1 pocket tolerates various substituents with a preference for flexible and basic side chains. Variation of the S2 substituent led to high-affinity ligands with inhibition constants down to 2nM for compounds bearing cyclohexyl substituents. Systematic investigations on the S3 pocket revealed its potential to achieve high activities with aromatic vectors that undergo stacking interactions with the planar peptide backbone forming part of the pocket. X-ray crystal structure analysis with the structurally related enzyme human cathepsinL confirmed the binding mode of the triazine ligand series as proposed by molecular modeling. Sub-micromolar inhibition of the proliferation of cultured parasites was achieved for ligands decorated with the best substituents identified through the optimization cycles. In cell-based assays, the introduction of a basic side chain on the inhibitors resulted in a 35-fold increase in antitrypanosomal activity. Finally, bioisosteric imidazopyridine nitriles were studied in order to prevent off-target effects with unselective nucleophiles by decreasing the inherent electrophilicity of the triazine nitrile headgroup. Using this ligand, the stabilization by intramolecular hydrogen bonding of the thioimidate intermediate, formed upon attack of the catalytic cysteine residue, compensates for the lower reactivity of the headgroup. The imidazopyridine nitrile ligand showed excellent stability toward the thiol nucleophile glutathione in a quantitative invitro assay and fourfold lower cytotoxicity than the parent triazine nitrile.

Full text of DOI:10.1002/cmdc.201300112

CHEMMEDCHEM
FULL PAPERS
DOI: 10.1002/cmdc.201300112
Optimization of Triazine Nitriles as Rhodesain Inhibitors:
Structure–Activity Relationships, Bioisosteric
Imidazopyridine Nitriles, and X-ray Crystal Structure
Analysis with Human Cathepsin L
Veronika Ehmke,[a] Edwin Winkler,[a] David W. Banner,*[b] Wolfgang Haap,[b] W.
Bernd Schweizer,[a] Matthias Rottmann,[c, d] Marcel Kaiser,[c, d] Cꢀline Freymond,[c, d]
Tanja Schirmeister,*[e, f] and FranÅois Diederich*[a]
The cysteine protease rhodesain of Trypanosoma brucei para-
sites causing African sleeping sickness has emerged as a target
for the development of new drug candidates. Based on a tria-
zine nitrile moiety as electrophilic headgroup, optimization
studies on the substituents for the S1, S2, and S3 pockets of
the enzyme were performed using structure-based design and
resulted in inhibitors with inhibition constants in the single-
digit nanomolar range. Comprehensive structure–activity rela-
tionships clarified the binding preferences of the individual
pockets of the active site. The S1 pocket tolerates various sub-
stituents with a preference for flexible and basic side chains.
Variation of the S2 substituent led to high-affinity ligands with
inhibition constants down to 2 nm for compounds bearing cy-
clohexyl substituents. Systematic investigations on the S3
pocket revealed its potential to achieve high activities with ar-
omatic vectors that undergo stacking interactions with the
planar peptide backbone forming part of the pocket. X-ray
crystal structure analysis with the structurally related enzyme
human cathepsin L confirmed the binding mode of the triazine
ligand series as proposed by molecular modeling. Sub-micro-
molar inhibition of the proliferation of cultured parasites was
achieved for ligands decorated with the best substituents iden-
tified through the optimization cycles. In cell-based assays, the
introduction of a basic side chain on the inhibitors resulted in
a 35-fold increase in antitrypanosomal activity. Finally, bioiso-
steric imidazopyridine nitriles were studied in order to prevent
off-target effects with unselective nucleophiles by decreasing
the inherent electrophilicity of the triazine nitrile headgroup.
Using this ligand, the stabilization by intramolecular hydrogen
bonding of the thioimidate intermediate, formed upon attack
of the catalytic cysteine residue, compensates for the lower re-
activity of the headgroup. The imidazopyridine nitrile ligand
showed excellent stability toward the thiol nucleophile gluta-
thione in a quantitative in vitro assay and fourfold lower cyto-
toxicity than the parent triazine nitrile.
Introduction
Parasitic cysteine proteases have emerged as attractive targets
for drug discovery due to their indispensable roles in disease
propagation.[1] The tropical parasitic diseases malaria and Afri-
can sleeping sickness (human African trypanosomiasis, HAT)
offer several cysteine proteases suitable for the development
of innovative chemotherapies to overcome the limitations of
drugs in current use. Proteolytic pathways, however, often rely
on closely related enzymes, leading to general concerns about
cytotoxicity and target-related side effects in the mammalian
host. Protease-based drug discovery has proven challenging,
but new therapeutic approaches are continuously emerging to
exploit the potential of new protease targets.[2]
[a] Dr. V. Ehmke, E. Winkler, Dr. W. B. Schweizer, Prof. Dr. F. Diederich
Laboratorium fꢀr Organische Chemie
ETH Zꢀrich, Wolfgang-Pauli-Strasse 10
HCI, 8093 Zꢀrich (Switzerland)
E-mail: diederich@org.chem.ethz.ch
[b] Dr. D. W. Banner, Dr. W. Haap
F. Hoffmann-La Roche AG
Grenzacherstrasse 124, Bau 92, 4070 Basel (Switzerland)
[c] Dr. M. Rottmann, Dr. M. Kaiser, C. Freymond
Swiss Tropical and Public Health Institute
Socinstrasse 57, 4002 Basel (Switzerland)
[d] Dr. M. Rottmann, Dr. M. Kaiser, C. Freymond
Universitꢁt Basel, Petersplatz 1, 4003 Basel (Switzerland)
[e] Prof. Dr. T. Schirmeister
Bloodstream Trypanosoma brucei parasites express two
papain family (EC 3.4.22.2) cysteine proteases: the human cath-
epsin L (hCatL)-like rhodesain and an hCatB-type enzyme
(TbCatB). These seem to be part of an essential pathway for
host protein degradation with overlapping activity profiles.[3]
Cysteine protease inhibitors have been shown to kill T. brucei
parasites both in cell culture and in animal models.[4] The inter-
play between rhodesain and TbCatB, however, remains poorly
Institut fꢀr Pharmazie und Lebensmittelchemie
Universitꢁt Wꢀrzburg, Am Hubland, 97074 Wꢀrzburg (Germany)
[f] Prof. Dr. T. Schirmeister
Institut fꢀr Pharmazie und Biochemie
Johannes Gutenberg-Universitꢁt Mainz
Staudinger Weg 5, 55128 Mainz (Germany)
E-mail: schirmei@uni-mainz.de
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201300112.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
1
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
understood, and elucidation of their individual roles would be
beneficial for the development of new HAT drug candidates.
The release of X-ray crystal structures of rhodesain in 2009
and 2010 (Protein Data Bank (PDB) codes: 2P7U and 2P86)[3a,5]
enabled the application of structure-based inhibitor design. In
our previous studies, computer-aided modeling guided the de-
velopment of potent nitrile inhibitors for rhodesain and the re-
lated falcipain-2 protein of the malarial parasite Plasmodium
falciparum.[6] A triazine nitrile scaffold was identified as both
a privileged central core to position the S1, S2, and S3 sub-
stituents and as an electrophilic anchor to form a covalent thi-
oimidate adduct with the catalytic cysteine. As a drawback,
however, this ligand class displayed moderate cytotoxicity and
only low target selectivity. We showed that these off-target ef-
into the binding mode of this inhibitor class and the hydration
state of the active site.
Results and Discussion
Structure–activity relationships on rhodesain
The overall structure of rhodesain shows the common two-
domain fold of the papain family and is highly similar to that
of hCatL. To analyze the binding preferences of rhodesain, we
used the software MOLOC[7] and the recent X-ray co-crystal
structure of rhodesain (PDB code: 2P86, resolution 1.16 ꢂ)[3a]
for molecular modeling. The optimization studies on the S1,
S2, and S3 pockets were based on triazine nitrile 1 with inhibi-
tion constants (Ki values) of 8 nm for rhodesain and 33 nm for
hCatL (selectivity ratio (SR)=4.1, calculated as the ratio Ki-
(hCatL)/Ki(rhodesain)).[6a] Lead compound 1 is decorated with
a morpholine S1 substituent, a cyclohexyl residue addressing
the hydrophobic S2 pocket, and a 1,3-benzodioxol-5-ylmethyl
group to stack on the flat peptide backbone formed by Gly65–
Gly66 in the solvent-exposed S3 pocket (see figure 1SI in the
Supporting Information (SI)).
The S1 pocket of rhodesain is located entirely at the surface
of the enzyme and is very similar to that of hCatL (figure 2SI).
This part of the active site should preferentially accommodate
polar or no vectors due to its high exposure to bulk solvent.
Triazine nitriles 2–(Æ)-6 were designed with various S1 sub-
stituents such as hydrogen or methoxy, cyclopropyl, 2-methox-
yethylamino, or the chloroquine-derived 5-diethylaminopent-2-
ylamino side chains (see Table 1 for structures). The corre-
sponding S2 and S3 substituents of lead compound 1 were
maintained to ensure a comparable binding mode (figure 3SI).
The synthetic strategy toward triazine nitriles 2–(Æ)-6 relied on
a sequence of reductive amination, nucleophilic aromatic sub-
stitution on cyanuric chloride, and cyanation as previously re-
ported (scheme 1SI).[6] Ligands 2–(Æ)-6 were subjected to bio-
Figure 1. Structure and binding affinities of triazine nitrile 1 functionalized
with substituents that address the S1, S2, and S3 pockets of rhodesain.
fects originate mainly from the high electrophilicity of the ni-
trile, which is promoted by the electron-deficient triazine
system, and that the electronic properties of the central core
directly influence the biological activity of the ligand.[6b]
Herein we report the structure-based design, synthesis, bio-
logical evaluation, and enzyme
X-ray crystal structure analysis of
new nitrile-derived rhodesain in-
Table 1. Binding affinities of ligands 1–(Æ)-6 with various S1 vectors for rhodesain (RD) and hCatL.
hibitors. To systematically ex-
plore the binding preferences of
the individual pockets, detailed
optimization studies on the S1,
S2, and S3 pockets were per-
formed based on the previously
developed triazine nitrile ligand
1[6a] (Figure 1). A 1-methyl-1H-
imidazo[4,5-c]pyridine-4-carboni-
trile core was introduced as
Compd
R
clogD[a]
Ki [nm][b]
hCatL
RD
8
1[6a]
4.16
33
2[6b]
3[6b]
4.01
4.93
19
9
63
2
4
5
4.98
4.55
68
28
76
61
a
suitable bioisosteric central
scaffold aiming at decreasing
the nitrile electrophilicity with-
out loss of binding affinity. Final-
ly, X-ray crystal structures of
hCatL, both in complex with a tri-
azine nitrile ligand and unligand-
ed, provided valuable insight
(Æ)-6
3.82
7
54
[a] Calculated with ACD/Labs version 12.5 (1994–2012 ACD/Labs) at pH 7.4. [b] Values are the average of at
least two independent measurements, each performed in duplicate.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
2
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
logical tests with rhodesain from T. brucei rhodesiense and
hCatL (see the SI). The variation of the S1 vector did not sub-
stantially affect the binding affinity (Table 1). Omitting the S1
vector led to only a slightly decreased binding affinity (Ki =
19 nm) for ligand 2, whereas the methoxy-substituted triazine
3 showed single-digit nanomolar inhibition of both rhodesain
(Ki =9 nm) and hCatL (Ki =2 nm). In contrast, a more than 3.5-
fold activity loss was observed for triazine nitrile ligands 4 and
5 with a cyclopropylamino or 2-methoxyethylamino group.
The chloroquine-derived side chain was identified as the best
S1 substituent, with ligand (Æ)-6 (Ki =7 nm) being equipotent
to lead compound 1, but with slightly improved selectivity
(SR=7.7). These findings suggest that the S1 pocket tolerates
a variety of substituents and can thus serve as a useful handle
to gain selectivity and tune the pharmacokinetic properties of
the ligands.
vector revealed the 4-(n-propyl) substituent as the most suita-
ble vector with Ki =2 nm for compound 9. Further increasing
the size of the substituent resulted in a drastic loss of affinity
for compounds 10 and 11 (Ki >470 nm). Replacement of the
cyclohexyl moiety with a phenyl group as in ligand 12 led to
a promising inhibition constant of 17 nm, whereas compound
13 with an indanyl moiety is a much weaker inhibitor (Ki =
410 nm). Excellent selectivity toward hCatL was found for the
4-trans-n-propylated ligand 9 (SR=345). In the absence of X-
ray crystal structure information, however, the large affinity dif-
ference for rhodesain and hCatL exhibited by compound 9
cannot be satisfyingly rationalized (figure 6SI). The established
structure–activity relationship (SAR) underlines the potential of
the S2 pocket to gain binding affinity with a preference for
suitably sized aliphatic substituents.
The last step in the optimization process concerned the S3
pocket, characteristically shaped by the exposed peptide back-
bone of Gly65–Gly66 (Gly67–Gly68 in hCatL, figure 7SI).[9] Ac-
cording to modeling, the aromatic 1,3-benzodioxol-5-yl group
of lead compound 1 undergoes stacking interactions with the
planar p surface of the peptide backbone with distances (d)
between the two planes <4.0 ꢂ. (figure 8SI). Our group recent-
ly showed that the introduction of heteroatoms into aromatic
rings, which stack on flat peptide segments in the S1 pocket
of factor Xa, significantly influences the strength of the stack-
ing interactions, and thus, the binding affinities of the inhibi-
tors.[10] We furthermore investi-
The hydrophobic and well-conserved S2 pocket of hCatL-like
enzymes is known as the main contributor to binding affinity
and substrate specificity (figure 4SI).[8] Ligands 7–13, bearing
trans-disubstituted cyclohexyl groups or aromatic substituents,
were designed to address this pocket (figure 5SI, see Table 2
for structures). The syntheses of triazine nitriles 7–13 relied on
the strategy described above (schemes 2SI and 3SI). Determi-
nation of their biological activity against rhodesain confirmed
the preference of the S2 pocket for extended hydrophobic res-
idues (Table 2). The stepwise elongation of the cyclohexyl S2
gated heteroaromatic stacking
on protein amide fragments
Table 2. Binding affinities of ligands 7–13 with various S2 vectors for rhodesain (RD) and hCatL.
using quantum chemical calcu-
lations on model systems and
the PDB and formulated guide-
lines to improve the stacking
energy, particularly through fa-
vorable dipolar interactions.[11]
Triazine nitriles 14–26, featuring
various aliphatic and aromatic
S3 vectors and sharing a binding
mode similar to that of lead
compound 1, were designed
and synthesized (see Table 3 for
structures; see schemes 4SI and
5SI for synthesis). Biological
evaluation of the ligand series
highlighted the crucial role of
the S3 substituent (Table 3). Tri-
azine nitrile 14 without an S3
vector was sevenfold less active
against rhodesain (Ki =53 nm)
than lead compound 1. The in-
troduction of an ethyl or isopro-
pyl group, as in ligands 15 and
16,[6a] resulted in a further activi-
ty decrease (Ki >74 nm). In con-
trast, good to excellent inhibi-
tion was observed for almost all
triazine nitriles 17–26 having ar-
Compd
R
clogD[a]
Ki [nm][b]
RD
hCatL
1[6a]
4.16
4.69
5.20
5.71
8
33
7
22
8
120
42
8
9[6a]
2
690
10
5.97
>80000
>80000
11
12
13
2.95
3.63
4.90
470
17
690
35
410
>80000
[a] Calculated with ACD/Labs version 12.5 (1994–2012 ACD/Labs) at pH 7.4. [b] Values are the average of at
least two independent measurements, each performed in duplicate.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
3
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
chain (Asn62 in hCatL). Finally,
thienyl ligand 26 showed an ex-
cellent inhibitory constant of
3 nm. This high activity can be
attributed to a favorable posi-
tioning of the thienyl ring on
top of the peptide backbone
(d<3.6 ꢂ, figure 9SI). Triazines
23, 24, and 26 showed im-
proved selectivities toward
hCatL (SR=5.0–10.7). The SAR
on the S3 pocket highlights the
potential of this pocket to gain
binding affinity with aromatic
substituents undergoing effi-
cient stacking interactions with
the exposed peptide backbone
of the enzyme.
Table 3. Binding affinities of ligands 14–26 with various S3 vectors for rhodesain (RD) and hCatL.
Compd
R
clogD[a]
Ki [nm][b]
RD
1[6a]
4.16
8
33
14
2.40
3.07
53
74
180
200
15[6a]
16[6a]
17
3.42
4.19
100
25
160
88
18
19
4.28
4.77
29
3
130
13
In accordance with the SARs
on the S1, S2, and S3 pockets,
the preferred vectors were com-
bined to give ligands (Æ)-27
and (Æ)-28 (Figure 2, scheme 6
SI). The interplay of these vec-
tors resulted in high affinities
for rhodesain paired with good
selectivities for both (Æ)-27 (Ki =
3 nm, SR=150) and (Æ)-28 (Ki =
5 nm, SR=15). In addition, the
ligands also displayed inhibition
of the related malarial cysteine
protease falcipain-2 in the nano-
molar range (Ki =84 and
200 nm). Evaluation of both
compounds in cell-based assays
showed sub-micromolar median
inhibitory concentration (IC50)
values against the malarial para-
sites (IC50 =0.73–0.95 mm). Li-
gands (Æ)-27 and (Æ)-28 were
more than 20-fold more effec-
tive inhibitors of the cell prolif-
20
21
22
5.04
3.84
5.00
55
13
27
12
31
14
23
24
4.69
4.87
6
6
30
62
25
26
4.76
4.01
>80000
>80000
3
32
[a] Calculated with ACD/Labs version 12.5 (1994–2012 ACD/Labs) at pH 7.4. [b] Values are the average of at
least two independent measurements, each performed in duplicate.
omatic S3 substituents (Ki =3–55 nm), with the para-chloro-
phenyl-substituted ligand 19 being one of the most potent
rhodesain inhibitors of this study (Ki =3 nm). Modeling sug-
gests that this affinity gain originates from favorable interac-
tions of the chlorine atom with the side chain NH of Asn70
and the backbone NH of Phe61 and an optimal parallel stack-
ing on the Gly65–Gly66 amide backbone (d=3.4 ꢂ, figure 8SI).
Among the series of ligands with biaryl-type S3 substituents,
compounds 23 and 24 with pyrazole groups showed similarly
high affinities for rhodesain (Ki =6 nm for both ligands) and
hCatL (Ki =30 and 62 nm, respectively) as those of lead com-
pound 1. Replacement of the C(3) methyl group with a trifluor-
omethyl moiety in ligand 25 led to complete loss of inhibition
(Ki >80000 nm), which might be due to strongly repulsive in-
teractions of the fluorine atoms with the polar Asp60 side
eration of T. brucei rhodesiense (IC50 =0.88 and 1.36 mm) as
compared to triazine 9 with a morpholine side chain. The en-
hancement of cell-based activity by introduction of a basic
amine side chain is in accordance with results of Coterꢃn
et al.[12] and with the concept of lysosomotropism.[13]
Imidazopyridine nitriles
We previously reported on the influence of nitrile headgroup
electrophilicity of cysteine protease inhibitors on biological ac-
tivity.[6b] Inhibitors with a highly activated aromatic nitrile, as
present in electron-deficient triazine systems, were very potent
inhibitors but also exhibited significant cytotoxicity, whereas li-
gands with less reactive headgroups displayed both lower ac-
tivity and cytotoxicity. To prevent side reactions and off-target
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
4
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
bond between the imidazopyri-
dine N(3) and the imino group
(d(N(3)···(H)N=C)=3.0 ꢂ)
does
not affect the positioning of the
thioimidate moiety at the oxy-
anion hole.
The
synthetic
sequence
toward imidazopyridine 29
started with the chlorination of
citrazinic acid (30) using phos-
phoryl trichloride and methanol
to afford methyl ester 31 in
73% yield (Scheme 1).[16] Sapo-
nification using lithium hydrox-
ide hydrate provided carboxylic
acid 32 in 96% yield. Conver-
sion into 4-aminopyridine 33[16]
was achieved by activation of
Figure 2. Antiparasitic activities and cytotoxicities of ligands (Æ)-27 and (Æ)-28. Values are the average of at least
two independent measurements, each performed in duplicate. P. falciparum (P. falc.) strain NF54, T. brucei rhode-
siense (T. b. rhod.) strain STIB 900, trypomastigote stage, and rat skeletal myoblasts (L-6 cells) were used in the
cell-based assays. For comparison, the IC50 values for lead compound 1 are 2.9 mm (P. falc.), 30.1 mm (T. b. rhod.),
and 11.1 mm (L-6 cells).
effects induced by an over-activated headgroup, modification
of the central motif was necessary. Cai and co-workers recently
the carboxylic acid functionality using oxalyl chloride, followed
by formation of the corresponding acyl azide, and subsequent
Curtius rearrangement[17] in 57% yield over three steps. N-Ni-
tration using sulfuric and nitric acid at 08C quantitatively gave
nitramide 34. Acid-promoted rearrangement of the nitro group
furnished 2,6-dichloro-3-nitro-4-aminopyridine (35) in 75%
yield.[18] The nitrile functionality at C(2) was introduced by
treatment with copper(I) cyanide in N-methyl-2-pyrrolidone
(NMP) at 1808C to quantitatively yield nitrile 36.[15a] Reaction
with methyl iodide in acetonitrile afforded the corresponding
N-methylated derivative 37 in 71% yield. X-ray analysis of crys-
tals obtained for compound 37 confirmed its assigned struc-
ture (see the SI). Reduction of the nitro functionality with stan-
nous chloride dihydrate quantitatively gave 3,4-diaminopyri-
dine 38, which was subsequently cyclized using triethyl ortho-
formate in the presence of catalytic amounts of para-toluene-
sulfonic acid to afford imidazopyridine 39 in 44% yield.
Introduction of the amino substituent at position C(6) of the
imidazopyridine core by palladium-catalyzed aromatic amina-
tion proved to be challenging. Acyclic aliphatic secondary
amines are sterically demanding and prone to b-hydride elimi-
nation, making them difficult substrates for Buchwald–Hartwig
couplings.[19] In a screening of reaction conditions, the best re-
sults were obtained by using potassium carbonate as base,
palladium(II) acetate as catalyst, and 2,2’-bis(diphenylphosphi-
no)-1,1’-binaphthyl ((Æ)-BINAP) as ligand in dry toluene at
908C for 48–72 h (table 1SI). Full conversion, however, could
not be achieved, and the amino-substituted imidazopyridine
29 was only obtained in low yield.
developed
1H-imidazo[4,5-c]pyridine-4-carbonitrile-based
hCatS ligands with improved stability toward unselective nu-
cleophiles.[14] This ligand class combines the features of purine
nitriles with sufficiently high electrophilicity and a pyridine
system, which is stable toward unselective nucleophiles due to
its low electron deficiency.[15] To achieve high potency for aro-
matic nitriles, it has been postulated that an intramolecular hy-
drogen bond between the imidazopyridine N(3) and the thioi-
midate NH leads to additional transition-state stabilization.[14]
This energy gain should compensate for the low nitrile activa-
tion by the pyridine subunit and simultaneously avoid a loss in
binding affinity (Figure 3).
In biological tests, imidazopyridine 29 showed a binding af-
finity for rhodesain of Ki =190 nm and an enhanced selectivity
toward hCatL (SR=18.7) as compared with triazine 1. Although
the calculated electrophilicity of the nitrile headgroup of an
Figure 3. Transition-state stabilization of imidazopyridine nitriles by intramo-
lecular hydrogen bonding upon nucleophilic attack of Cys25 (papain num-
bering).
Based on lead compound 1, we designed imidazopyridine
29 with a cyclohexyl and 1,3-benzodioxol-5-ylmethyl substitu-
ent (Scheme 1). A methyl substituent was selected as S1 vector
to improve the metabolic stability of the ligand.[14] The binding
mode of ligand 29 as obtained by modeling is comparable to
that of triazine 1 (figure 10SI). The intramolecular hydrogen
imidazopyridine is considerably lower (DH=À6.3 kcalmolÀ1
;
calculated as the difference in formation enthalpies of the thio-
imidate adduct and methanethiol and the nitrile; for DFT calcu-
lations on nitrile electrophilicities see the SI) than that of the
highly reactive nitrile of triazine ligand 1 (DH=À8.8 kcal
molÀ1),[6b] nanomolar inhibition could be maintained. Stabiliza-
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
5
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
X-ray crystal structure analysis
with hCatL
To verify the proposed binding
mode, X-ray crystal structure in-
formation with rhodesain in
complex with our ligands was
desirable. Unfortunately, all at-
tempts to obtain suitable crys-
tals failed. Gratifyingly, crystalli-
zation experiments with hCatL
were successful, and a co-crystal
structure of the enzyme cova-
lently bound to triazine nitrile
19 (PDB code: 4AXM, resolution
2.80 ꢂ) and an uncomplexed
hCatL structure (PDB code:
4AXL, resolution 1.92 ꢂ) could
be solved (see the Experimental
Section and the SI for details).
Due to the high structural simi-
larity between hCatL and rhode-
sain, these X-ray crystal struc-
tures are suitable to discuss the
Scheme 1. Synthesis of compound 29. Reagents and conditions: a) POCl3, Me4NCl, MeOH, 1308C, 24 h, 73%;
b) LiOH·H2O, THF, 258C, 20 min, 96%; c) 1. (COCl)2, CH2Cl2/THF, 258C, 4 h; 2. NaN3, acetone/H2O, 258C, 2 min, then
trifluoroacetic acid, benzene, 908C, 21 h; 3. K2CO3, MeOH, H2O, 258C, 7.5 h, 57% over three steps; d) H2SO4, HNO3,
08C, 2 h, quant.; e) H2SO4, 1008C, 1 h, 75%; f) CuCN, N-methyl-2-pyrrolidone, 1808C, 40 min, quant.; g) MeI, MeCN,
808C, 3 h, 71%; h) SnCl2·2H2O, EtOH, 258C, 3 h, quant.; i) (EtO)3CH, p-toluenesulfonic acid, 1208C, 3 h, 44%; j) N-
(1,3-benzodioxol-5-ylmethyl)cyclohexanamine,[6a] K3PO4, 2 mol% [Pd(OAc)2], 4 mol% (Æ)-BINAP, toluene, 908C,
72 h, 2%.
tion of the thioimidate adduct by the intramolecular hydrogen
bond compensates for the low nitrile reactivity. For compari-
son, the analogous pyridine ligand with a similarly low electro-
binding mode of the triazine ligand series at the active site of
rhodesain.
The complex of triazine ligand 19 and hCatL crystallized in
the P212121 space group, with six protein chains of mature
enzyme in the asymmetric unit. The best electron density for
the ligand was found for chain B with 219 amino acids. No
water molecules are resolved in the proximity of the active
site. The thioimidate intermediate formed upon attack of
Cys25 is stabilized through the oxyanion hole by weak hydro-
gen bonding to the side chain of Gln19 (d(N(H)Gln19···N=C)=
3.4 ꢂ, Figure 4). The binding mode of ligand 19 in the X-ray
philicity (DH=À5.4 kcalmolÀ1 [6b]
)
as that of imidazopyridine 29
was inactive against rhodesain. Imidazopyridine 29 was also
tested in cell-based assays and found to be a low micromolar
inhibitor of P. falciparum proliferation (IC50 =3.6 mm), but
showed no antitrypanosomal activity. The inconsistency be-
tween the cell-based data and the enzymatic activity suggests
that inhibition of rhodesain might not be the only determinant
of parasitic activity, as discussed above. Assays with rat skeletal
myoblasts revealed a fourfold decrease in cytotoxicity for imi-
dazopyridine 29 (IC50 =48.9 mm) relative to triazine 1 (IC50 =
11.1 mm).
To monitor the reactivity of nitrile-containing ligands toward
unspecific nucleophiles and to investigate a potential toxicity
risk, we used a quantitative l-glutathione (GSH)-based assay
(see the SI).[20] Liquid chromatography–mass spectrometry (LC–
MS) was used to analyze the stability of triazine nitrile 1 and
imidazopyridine nitrile 29 toward GSH under physiological
conditions (pH 7.4, 378C, 5 mm GSH). GSH adduct formation
and the simultaneous decrease in the amount of free nitrile li-
gands over time were monitored and the half-lives (t1/2) deter-
mined (see the SI). Triazine nitrile 1 readily reacted with GSH in
accordance with the high nitrile electrophilicity (t1/2 =13 h),
whereas imidazopyridine nitrile 29 was stable under the assay
conditions even after 120 h. The ligand stabilities correlate very
well with the calculated electrophilicities of the nitriles and the
selectivities determined toward hCatL (SR(ligand 1)=4.1, SR(li-
gand 29)=18.7). The GSH adduct assay underscores the po-
tential of the imidazopyridine nitriles as a promising ligand
class for future optimization studies on rhodesain, which are
currently ongoing in our research program.
Figure 4. X-ray co-crystal structure of triazine nitrile 19 covalently bound at
the active site of hCatL (PDB code: 4AXM, resolution 2.80 ꢂ). Distances given
in ꢂ; color code: gray Cenzyme, green Cligand, red O, blue N, yellow S, lime Cl.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
6
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
co-crystal structure is in excellent agreement with the one ob-
tained by modeling, validating our design concept (figure 11
SI). The morpholine group addresses the S1 pocket, where it
weakly interacts with the carbonyl groups of Cys65 (d(C=
O
Cys65···C)=4.2 ꢂ) and Asn66 (d(C=OAsn66···C)=4.2 ꢂ, figure 12SI).
The electron-deficient triazine ring is positioned in the proximi-
ty of Gly67 undergoing CÀH···p interactions (d(CCys67···N)=
4.0 ꢂ). The cyclohexyl vector is deeply buried in the S2 pocket
and engaged in several hydrophobic interactions with the side
chains of Gly68, Ile69, and Ala135 (d<4.1 ꢂ, figure 12SI). The
S3 pocket accommodates the para-chlorobenzyl substituent
(figure 13SI). The chlorine atom weakly interacts with the car-
bonyl group of Gly61 (d(C=OCys61···Cl)=3.8 ꢂ, angle(O···ClÀC)=
1608).[21] The aromatic ring undergoes stacking interactions
with the planar peptide fragment of Gly67–Gly68 (d<3.8 ꢂ).
During the crystallization experiments, we also obtained an
uncomplexed hCatL structure, which represents the first exam-
ple in the PDB of mature and active hCatL without any ligand
bound. The only other released hCatL “apo” structure bears
a mutation of the active site Cys25 residue (PDB code: 3IV2,
resolution 2.20 ꢂ).[22] The obtained uncomplexed hCatL struc-
ture crystallized in space group P6122 with one protein chain
of 219 amino acids in the asymmetric unit. A zinc ion unex-
pectedly present at the active site interacts with His163
(d(NHis163···Zn)=2.1 ꢂ) and bridges to the side chain of Glu141
(d(C=OHis163···Zn)=2.1 ꢂ) of the next protein in the crystal (fig-
ure 14SI). In addition, two water molecules occupy the coordi-
nation sphere of zinc (d<2.3 ꢂ). The side chains of Cys25 and
Glu141 are more distal to the metal center, forming only weak
interactions (d>2.5 ꢂ).
Figure 5. Superposition of the X-ray co-crystal structure of hCatL (gray, PDB
code: 4AXM, resolution 2.80 ꢂ, ligand 19 omitted for clarity) and the uncom-
plexed hCatL structure (cyan, PDB code: 4AXL, resolution 1.92 ꢂ). The active
site region shows the positioning of the water molecules and the zinc ion of
the uncomplexed structure. Color code: gray Cenzyme–4AXM, cyan Cenzyme–4AXL
red O, blue N, yellow S, magenta Zn. Water molecules are shown as red
spheres.
,
Conclusions
We used structure-based design to optimize the triazine nitrile
ligand class as potent rhodesain inhibitors and to establish
SARs on the individual pockets. Variation of the S1 pocket sub-
stituent had only minimal impact on binding affinity and selec-
tivity, with slight preference for a basic chloroquine-derived
side chain. Investigations on the hydrophobic S2 pocket led to
the identification of cyclohexyl-based substituents as the most
promising vectors. The SAR on the S3 pocket revealed a clear
preference for aromatic vectors to undergo efficient stacking
interactions with the peptide backbone shaping this pocket,
leading to several rhodesain inhibitors with single-digit nano-
molar inhibitory activity. Ligands bearing a basic chloroquine-
derived side chain were up to 35-fold more active against
T. brucei rhodesiense in cell-based assays. Replacement of the
triazine core with a bioisosteric imidazopyridine scaffold result-
ed in a nanomolar rhodesain inhibitor with a less reactive ni-
trile headgroup. Due to stabilization of the thioimidate adduct
by intramolecular hydrogen bonding, the electrophilic proper-
ties of the nitrile could be adjusted, and the imidazopyridine
nitrile ligand showed excellent stability toward GSH in a kinetic
LC–MS assay and a fourfold lower cytotoxicity than that of the
parent triazine nitrile ligand. X-ray crystal structure analysis
with hCatL confirmed the binding mode of the triazine nitrile
ligand series as obtained by modeling, and an uncomplexed
hCatL structure allowed a detailed analysis of the hydration
state of the enzyme in its unbound state. These results provide
the basis for further optimization studies on rhodesain, focus-
ing on imidazopyridine nitrile-derived ligands with improved
enzymatic and cell-based activity.
The careful analysis of the hydration state of a binding
pocket in the unbound state is important for structure-based
design, because water molecules might be replaceable or con-
sidered as part of the protein structure depending on the hy-
drogen bond network formed and their direct enzymatic envi-
ronment.[23] In the uncomplexed hCatL structure, a network of
eleven water molecules fills the binding site. Several water
molecules are hydrogen bonded to each other and the sur-
rounding amino acids (d<3.5 ꢂ), providing a certain structural
rigidity to the enzyme (Figure 5). In general, most of the water
molecules are only loosely bound at the active site, with each
water molecule engaged in less than two hydrogen bonds to
the enzyme. The overall structure of the uncomplexed hCatL
enzyme shows remarkably high analogy with the co-crystal
structure in complex with ligand 19, in particular at the active
site region. Upon ligand complexation, the water molecules lo-
cated in proximity of the catalytic center must be displaced.
Due to the small number of contacts between the water mole-
cules and the enzyme and the hydrophobicity of the binding
site, desolvation upon ligand binding should be energetically
favored.[23] The uncomplexed hCatL structure shows high struc-
tural complementarity with the other hCatL “apo” structure
(PDB code: 3IV2,[22] figure 15SI). The rigidity of hCatL in its unli-
ganded state is a desirable prerequisite for reliable modeling
and structure-based drug design.
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
7
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
N-methyl-2-pyrrolidone (80 mL) was stirred at 1808C for 40 min.
The suspension was cooled to 258C, diluted with EtOAc (500 mL),
and filtered. H2O (500 mL) was added. The organic layer was sepa-
rated, washed with brine (3ꢄ300 mL), dried over Na2SO4, filtered,
and evaporated to give crude 36 (8.96 g, quant.) as a brown solid,
which was directly used in the next step. Rf =0.36 (cyclohexane/
EtOAc 1:1); HRMS-EI: m/z [M]+ calcd for C6H337ClN4O2+: 199.9910,
found: 199.9909 (31%); [M]+ calcd for C6H335ClN4O2+: 197.9939,
found: 197.9935 (100%).
Experimental Section
X-ray crystal structure determinations: Crystals of hCatL were
produced and the data measured as previously described.[21,24] Pro-
tein at a concentration of 2 mm in 20 mm Bis-Tris methane pH 7,
100 mm NaCl, 10% glycerol, 5 mm DTT, and 0.01% NaN3 was incu-
bated with the ligands at 10-fold molar excess overnight at 48C
under argon. Prior to the crystallization experiments, the protein
was concentrated to 26–35 mgmLÀ1 and centrifuged (20000 g).
The crystallization droplets were set up at 228C by mixing 0.15 mL
of 30 mgmLÀ1 protein solution with 0.3 mL reservoir solution (Index
Screen, Hampton Research) in microbatch experiments under Al’s
oil (Hampton Research). Crystals were harvested and flash frozen
with paraffin oil as cryoprotectant. Data were processed with
XDS[25] and scaled with SADABS (from Bruker AXS). The structure
was solved by molecular replacement using PHASER[26] and a start-
ing model generated from the structure with PDB code 2YJ2.[24]
Model building was done with Coot[27] and refinement with
Refmac5[28] from the CCP4 suite[29] and BUSTER (from Global Phas-
ing Ltd. 2011, version 2.11.2). The final refinement was performed
in Refmac5 with hydrogen atoms added at riding positions and
TLS refinement,[30] but without NCS restraints. Coordinates have
been deposited at the Protein Data Bank with PDB codes 4AXM
(resolution 2.80 ꢂ, co-crystal structure with triazine nitrile 19) and
4AXL (resolution 1.92 ꢂ, uncomplexed hCatL structure). See the SI
for more details.
6-Chloro-4-(methylamino)-3-nitro-2-pyridinecarbonitrile (37):
A
suspension of 36 (9.00 g, 45.32 mmol) and MeI (7.05 mL,
113.30 mmol) in MeCN (100 mL) was stirred at 808C for 3 h. Addi-
tional MeI (2.82 mL, 45.32 mmol) was added, and the solution was
stirred at 808C for an additional 2 h. The solution was cooled to
258C and diluted with EtOAc (200 mL) and H2O (150 mL). The or-
ganic layer was separated, washed with H2O (2ꢄ200 mL), dried
over Na2SO4, filtered, and evaporated to give 37 (6.86 g, 71%) as
a beige solid. Rf =0.34 (cyclohexane/EtOAc 2:1); mp: 155–1578C;
1H NMR (300 MHz, CDCl3): d=3.04–3.11 (m, 3H; Me), 6.93 (s, 1H; H-
C(5)), 8.13 ppm (brs, 1H; NH); 13C NMR (100 MHz, CDCl3): d=30.20
(Me), 109.79 (C(5)), 113.91 (CN), 131.32 (C(3)), 150.56 (C(6)),
155.10 ppm (C(4)); CCN signal not visible; IR (ATR): n˜ =3376, 3085,
2944, 1594, 1551, 1508, 1496, 1455, 1432, 1389, 1350, 1284, 1232,
1167, 1101, 1059, 953, 900, 861, 836, 775, 663, 636, 622 cmÀ1
;
HRMS-EI: m/z [M]+ calcd for C7H537ClN4O2+: 214.0067, found:
214.0064 (35%); [M]+ calcd for C7H535ClN4O2+: 212.0096, found:
212.0094 (100%).
General details and synthetic procedures: In the following, the
experimental details for the syntheses of compounds 29, 36, 37,
38, and 39 are described (Scheme 1). See the SI for the atom num-
3-Amino-6-chloro-4-(methylamino)-2-pyridinecarbonitrile (38): A
suspension of 37 (6.86 g, 32.27 mmol) in EtOH (150 mL) was treat-
ed with SnCl2·2H2O (25.48 g, 112.94 mmol) and stirred at 258C for
3 h. The solution was diluted with EtOAc (200 mL) and an aqueous
NH3 solution (10%, 200 mL). The aqueous layer was extracted with
EtOAc (5ꢄ200 mL). The combined organic layers were washed
with brine (2ꢄ300 mL), dried over Na2SO4, filtered, and evaporated
to give 38 (5.90 g, quant.) as a brown solid. Rf =0.24 (cyclohexane/
1
bering used to assign the H and 13C NMR signals. All other experi-
mental and analytical data are included in the SI, as well as details
about the X-ray crystal structure analysis of compounds 37
(Scheme 1) and 58 (scheme 4SI). CCDC 916055 and CCDC 916056
contain the supplementary crystallographic data for this paper.
These data can be obtained free of charge from The Cambridge
Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_re-
quest/cif.
1
EtOAc 1:1); mp: 167–1698C; H NMR (400 MHz, CD3OD): d=2.89 (s,
3H; Me), 6.45 ppm (s, 1H; H-C(5)); 13C NMR (100 MHz, CD3OD): d=
28.46 (Me), 103.08 (C(3)), 103.23 C(5)), 115.77 (CN), 138.67 (C(2)),
141.22 (C(6)), 146.08 ppm (C(4)); IR (ATR): n˜ =3441, 3363, 3326,
3265, 2949, 2926, 2223, 1651, 1581, 1542, 1516, 1478, 1425, 1353,
6-[(1,3-Benzodioxol-5-ylmethyl)(cyclohexyl)amino]-1-methyl-1H-
imidazo[4,5-c]-pyridine-4-carbonitrile (29): A suspension of 39
(500 mg, 2.60 mmol), [Pd(OAc)2] (12 mg, 0.05 mmol), (Æ)-BINAP
(65 mg, 0.10 mmol), and K3PO4 (1650 mg, 7.79 mmol) in toluene
(4 mL) was treated with N-(1,3-benzodioxol-5-ylmethyl)cyclohexa-
namine[6a] (1820 mg, 7.79 mmol). The suspension was degassed by
bubbling argon for 15 min, the flask sealed, and stirring continued
at 908C for 72 h. The suspension was cooled to 258C, filtered
through Celite, and the filtrate evaporated. Purification by FC (SiO2;
cyclohexane/THF 1:2 and CH2Cl2/EtOAc 1:2) gave 29 (20 mg, 2%)
1261, 1149, 1121, 1069, 977, 923, 873, 829, 806, 740, 656 cmÀ1
;
HRMS-ESI: m/z [M+H]+ calcd for C7H837ClN4+: 185.0403, found:
185.0403 (41%); [M+H]+ calcd for C7H835ClN4+: 183.0432, found:
183.0432 (100%).
6-Chloro-1-methyl-1H-imidazo[4,5-c]pyridine-4-carbonitrile (39):
Compound 38 (3.91 g, 21.39 mmol) was treated with triethyl ortho-
formate (21.35 mL, 128.34 mmol) and para-toluenesulfonic acid
(814 mg, 4.28 mmol), and the mixture was stirred at 1208C for 3 h
in a distillation apparatus to remove the forming EtOH. The solu-
tion was cooled to 258C and diluted with EtOAc (150 mL) and an
aqueous Na2CO3 solution (10%, 150 mL). The organic layer was
separated, dried over Na2SO4, filtered, and concentrated. Purifica-
tion by FC (SiO2; cyclohexane/EtOAc 1:5) gave 39 (1.81 g, 44%) as
a beige solid. Rf =0.11 (cyclohexane/EtOAc 1:2); mp: 189–1918C;
1H NMR (400 MHz, CDCl3): d=3.93 (s, 3H; Me), 7.60 (s, 1H; H-C(2)),
8.10 ppm (s, 1H; H-C(7)); 13C NMR (100 MHz, CDCl3): d=31.73 (Me),
108.99 (C(7)), 114.00 (CN), 124.02 (C(3a)), 142.46 and 142.70 (C(7a),
C(4)), 143.97 (C(6)), 148.46 ppm (C(2)); IR (ATR): n˜ =3086, 3026,
2957, 2241, 1788, 1608, 1570, 1492, 1436, 1425, 1344, 1305, 1251,
1207, 1097, 1053, 984, 896, 885, 863, 797, 709, 666, 653, 625,
612 cmÀ1; HRMS-ESI: m/z [M+H]+ calcd for C8H637ClN4+: 195.0246,
1
as a pale-yellow foam. Rf =0.38 (EtOAc); H NMR (300 MHz, CDCl3):
d=0.86–1.85 (m, 10H; H-C(2’-6’)), 3.62 (s, 3H; Me), 4.51 (s, 2H;
CH2N), 4.67–4.75 (m, 1H; H-C(1’)), 5.92 (s, 2H; CH2O), 6.26 (s, 1H;
H-C(7)), 6.72–6.74 (m, 3H; 3 arom. H), 7.71 ppm (s, 1H; H-C(2));
13C NMR (100 MHz, CDCl3): d=25.78 (C(4’)), 26.03 (2C, C(3’,5’)),
30.71 (2C, C(2’,6’)), 30.91 (Me), 47.85 (CH2N), 55.47 (C(1’)), 89.09
(C(7)), 100.97 (CH2O), 106.79 (arom. CH), 108.31 (arom. CH), 115.84
(CN), 119.17 (arom. CH), 121.72 (C(3a)), 133.20 (arom. C), 137.15
(C(4)), 143.26 (C(7a)), 145.50 (C(2)), 146.38 (arom. C), 147.95 (arom.
C), 154.97 ppm (C(6)); IR (ATR): n˜ =2930, 2855, 2230, 1733, 1627,
1568, 1503, 1489, 1446, 1367, 1314, 1237, 1133, 1039, 1006, 924,
807, 729, 617 cmÀ1; HRMS-ESI: m/z [M+H]+ calcd for C22H24N5O2
390.1925, found: 390.1914 (100%).
+
:
4-Amino-6-chloro-3-nitro-2-pyridinecarbonitrile (36): A suspen-
sion of 35 (9.41 g, 45.24 mmol) and CuCN (8.10 g, 90.48 mmol) in
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
8
&
ÞÞ
These are not the final page numbers!

CHEMMEDCHEM
FULL PAPERS
www.chemmedchem.org
found: 195.0238 (35%); [M+H]+ calcd for C8H635ClN4+: 193.0276,
found: 193.0266 (100%).
[11] M. Harder, B. Kuhn, F. Diederich, ChemMedChem 2013, 8, 397–404.
[12] J. M. Coterꢃn, D. Catterick, J. Castro, M. J. Chaparro, B. Dꢆaz, E. Fernꢇn-
dez, S. Ferrer, F. J. Gamo, M. Gordo, J. Gut, L. de Las Heras, J. Legac, M.
Marco, J. Miguel, V. MuÇoz, E. Porras, J. C. de La Rosa, J. R. Ruiz, E. San-
doval, P. Ventosa, P. J. Rosenthal, J. M. Fiandor, J. Med. Chem. 2010, 53,
6129–6152.
Acknowledgements
[13] W. C. Black, M. D. Percival, ChemBioChem 2006, 7, 1525–1535.
[14] J. Cai, D. J. Bennett, Z. Rankovic, M. Dempster, X. Fradera, J. Gillespie, I.
Cumming, W. Finlay, M. Baugh, S. Boucharens, J. Bruin, K. S. Cameron,
W. Hamilton, J. Kerr, E. Kinghorn, G. McGarry, J. Robinson, P. Scullion,
J. C. M. Uitdehaag, M. van Zeeland, D. Potin, L. Saniere, A. Fouquet, F.
Chevallier, H. Deronzier, C. Dorleans, E. Nicolai, Bioorg. Med. Chem. Lett.
2010, 20, 4447–4450.
[15] a) J. Cai, M. Baugh, D. Black, C. Long, D. J. Bennett, M. Dempster, X. Fra-
dera, J. Gillespie, F. Andrews, S. Boucharens, J. Bruin, K. S. Cameron, I.
Cumming, W. Hamilton, P. S. Jones, A. Kaptein, E. Kinghorn, M. Maid-
ment, I. Martin, A. Mitchell, Z. Rankovic, J. Robinson, P. Scullion, J. C. M.
Uitdehaag, P. Vink, P. Westwood, M. van Zeeland, L. van Berkom, M. Bas-
tiani, T. Meulemans, Bioorg. Med. Chem. Lett. 2010, 20, 4350–4354; b) J.
Cai, X. Fradera, M. van Zeeland, M. Dempster, K. S. Cameron, D. J. Ben-
nett, J. Robinson, L. Popplestone, M. Baugh, P. Westwood, J. Bruin, W.
Hamilton, E. Kinghorn, C. Long, J. C. M. Uitdehaag, Bioorg. Med. Chem.
Lett. 2010, 20, 4507–4510.
This research was supported by the ETH Research Council and
F. Hoffmann-La Roche Ltd., Basel (Switzerland). Financial support
by the Deutsche Forschungsgemeinschaft (DFG) is acknowledged
(T.S.). V.E. is grateful to the German Fonds der Chemischen Indus-
trie for a Kekulꢂ PhD Fellowship. We thank Anna Kucharski (Uni-
versitꢁt Wꢀrzburg), Sabine Mꢁhrlein, Nicole Heindl, and Ulrike
Nowe (all Johannes Gutenberg-Universitꢁt Mainz) for performing
the enzyme assays, Dr. Pablo Rivera-Fuentes (ETH Zꢀrich) for the
computational studies, and Bernhard Gsell, Martine Stihle, and
Dr. Jçrg Benz for the hCatL protein and crystals.
Keywords: antiprotozoal agents · cysteine proteases · ligand
design · nitriles · structure–activity relationships
[16] J. V. Mello, N. S. Finney, Org. Lett. 2001, 3, 4263–4265.
[17] J. R. Pfister, W. E. Wymann, Synthesis 1983, 38–40.
[1] R. Vicik, M. Busemann, K. Baumann, T. Schirmeister, Curr. Top. Med.
Chem. 2006, 6, 331–353.
[18] A. M. Palmer, G. Mꢅnch, C. Brehm, P. J. Zimmermann, W. Buhr, M. P.
Feth, W. A. Simon, Bioorg. Med. Chem. 2008, 16, 1511–1530.
[19] “Metal-catalyzed Cross-Coupling Reactions”: L. Jiang, S. L. Buchwald in
Palladium-Catalyzed Aromatic Carbon–Nitrogen Bond Formation, Vol. 2
(Eds.: A. de Meijere, F. Diederich), Wiley-VCH, Weinheim, 2004, pp. 699–
760.
[20] a) E. D. Clarke, D. T. Greenhow, D. Adams, Pestic. Sci. 1998, 54, 385–393;
b) P. A. MacFaul, A. D. Morley, J. J. Crawford, Bioorg. Med. Chem. Lett.
2009, 19, 1136–1138.
[21] L. A. Hardegger, B. Kuhn, B. Spinnler, L. Anselm, R. Ecabert, M. Stihle, B.
Gsell, R. Thoma, J. Dietz, J. Benz, J.-M. Plancher, G. Hartmann, D. W.
Banner, W. Haap, F. Diederich, Angew. Chem. 2011, 123, 329–334;
Angew. Chem. Int. Ed. 2011, 50, 314–318.
[22] M. A. Adams-Cioaba, J. C. Krupa, C. Xu, J. S. Mort, J. Min, Nat. Commun.
2011, 2, 197.
[2] M. Drag, G. S. Salvesen, Nat. Rev. Drug Discovery 2010, 9, 690–701.
[3] a) I. D. Kerr, P. Wu, R. Marion-Tsukamaki, Z. B. Mackey, L. S. Brinen, PLoS
Neglected Trop. Dis. 2010, 4, e701; b) Z. B. Mackey, T. C. O’Brien, D. C.
Greenbaum, R. B. Blank, J. H. McKerrow, J. Biol. Chem. 2004, 279,
48426–48433; c) L. Redecke, K. Nass, D. P. DePonte, T. A. White, D. Re-
hders, A. Barty, F. Stellato, M. Liang, T. R. M. Barends, S. Boutet, G. J. Wil-
liams, M. Messerschmidt, M. M. Seibert, A. Aquila, D. Arnlund, S. Bajt, T.
Barth, M. J. Bogan, C. Caleman, T.-C. Chao, R. B. Doak, H. Fleckenstein,
M. Frank, R. Fromme, L. Galli, I. Grotjohann, M. S. Hunter, L. C. Johans-
son, S. Kassemeyer, G. Katona, R. A. Kirian, R. Koopmann, C. Kuptiz, L.
Lomb, A. V. Martin, S. Mogk, R. Neutze, R. L. Shoeman, J. Steinbrener, N.
Timneanu, D. Wang, U. Weierstall, N. A. Zatsepin, J. C. H. Spence, P.
Fromme, I. Schlichting, M. Duszenko, C. Betzel, H. N. Chapman, Science
2013, 339, 227–230.
[23] C. Bissantz, B. Kuhn, M. Stahl, J. Med. Chem. 2010, 53, 5061–5084.
[24] L. A. Hardegger, B. Kuhn, B. Spinnler, L. Anselm, R. Ecabert, M. Stihle, B.
Gsell, R. Thoma, J. Diez, J. Benz, J.-M. Plancher, G. Hartmann, Y. Isshiki, K.
Morikami, N. Shimma, W. Haap, D. W. Banner, F. Diederich, ChemMed-
Chem 2011, 6, 2048–2054.
[25] W. Kabsch, Acta Crystallogr. Sect. D 2010, 66, 125–132.
[26] A. J. McCoy, R. W. Grosse-Kunstleve, P. D. Adams, M. D. Winn, L. C. Storo-
ni, R. J. Read, J. Appl. Crystallogr. 2007, 40, 658–674.
[27] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
[28] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr. Sect. D
1997, 53, 240–255.
[29] Collaborative Computational Project, Number 4, Acta Crystallogr. Sect. D
1994, 50, 760–763.
[4] S. Scory, C. R. Caffrey, Y. D. Stierhof, A. Ruppel, D. Steverding, Exp. Parasi-
tol. 1999, 91, 327–333.
[5] I. D. Kerr, J. H. Lee, C. J. Farady, R. Marion, M. Rickert, M. Sajid, K. C.
Pandey, C. R. Caffrey, J. Legac, E. Hansell, J. H. McKerrow, C. S. Craik, P. J.
Rosenthal, L. S. Brinen, J. Biol. Chem. 2009, 284, 25697–25703.
[6] a) V. Ehmke, C. Heindl, M. Rottmann, C. Freymond, W. B. Schweizer, R.
Brun, A. Stich, T. Schirmeister, F. Diederich, ChemMedChem 2011, 6,
273–278; b) V. Ehmke, J. E. Q. Quinsaat, P. Rivera-Fuentes, C. Heindl, C.
Freymond, M. Rottmann, R. Brun, T. Schirmeister, F. Diederich, Org.
Biomol. Chem. 2012, 10, 5764–5768.
[7] P. R. Gerber, K. Mꢅller, J. Comput.-Aided Mol. Des. 1995, 9, 251–268.
[8] M. K. Ramjee, N. S. Flinn, T. P. Pemberton, M. Quibell, Y. Wang, J. P.
Watts, Biochem. J. 2006, 399, 47–57.
[30] M. D. Winn, M. N. Isupov, G. N. Murshudov, Acta Crystallogr. Sect. D
2001, 57, 122–133.
[9] S. Cotesta, M. Stahl, J. Mol. Model. 2006, 12, 436–444.
[10] a) E. A. Meyer, R. K. Castellano, F. Diederich, Angew. Chem. 2003, 115,
1244–1287; Angew. Chem. Int. Ed. 2003, 42, 1210–1250; b) L. M. Salo-
nen, M. C. Holland, P. S. J. Kaib, W. Haap, J. Benz, J.-L. Mary, O. Kuster,
W. B. Schweizer, D. W. Banner, F. Diederich, Chem. Eur. J. 2012, 18, 213–
222.
Received: March 13, 2013
Published online on && &&, 0000
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&
9
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
V. Ehmke, E. Winkler, D. W. Banner,*
W. Haap, W. B. Schweizer, M. Rottmann,
M. Kaiser, C. Freymond, T. Schirmeister,*
F. Diederich*
Out of the HAT: SAR studies on rhode-
sain inhibitors deciphered the binding
preferences, giving Ki values in the
single-digit nanomolar range. A bioiso-
steric imidazopyridine scaffold led to
lower electrophilicity of the nitrile head-
group, resulting in reduced off-target ef-
fects. X-ray crystal structures of hCatL in
complex with a triazine nitrile inhibitor
and in the apo form shed light on the
binding mode and hydration state of
the active site.
&& – &&
Optimization of Triazine Nitriles as
Rhodesain Inhibitors: Structure–
Activity Relationships, Bioisosteric
Imidazopyridine Nitriles, and X-ray
Crystal Structure Analysis with Human
Cathepsin L
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 0000, 00, 1 – 10
&10
&
ÞÞ
These are not the final page numbers!