2H-1,2,3-Triazole-Based Dipeptidyl Nitriles: Potent, Selective, and Trypanocidal Rhodesain Inhibitors by Structure-Based Design

Article abstract of DOI:10.1021/acs.jmedchem.7b01870

Macrocyclic inhibitors of rhodesain (RD), a parasitic cysteine protease and drug target for the treatment of human African trypanosomiasis, have shown low metabolic stability at the macrocyclic ether bridge. A series of acyclic dipeptidyl nitriles was developed using structure-based design (PDB ID: 6EX8). The selectivity against the closely related cysteine protease human cathepsin L (hCatL) was substantially improved, up to 507-fold. In the S2 pocket, 3,4-dichlorophenylalanine residues provided high trypanocidal activities. In the S3 pocket, aromatic residues provided enhanced selectivity against hCatL. RD inhibition (K_i values) and in vitro cell-growth of Trypanosoma brucei rhodesiense (IC₅₀ values) were measured in the nanomolar range. Triazole-based ligands, obtained by a safe, gram-scale flow production of ethyl 1H-1,2,3-triazole-4-carboxylate, showed excellent metabolic stability in human liver microsomes and in vivo half-lives of up to 1.53 h in mice. When orally administered to infected mice, parasitaemia was reduced but without complete removal of the parasites.

Full text of DOI:10.1021/acs.jmedchem.7b01870

Subscriber access provided by UNIV OF DURHAM
2H-1,2,3-Triazole-Based Dipeptidyl Nitriles: Potent, Selective, and
Trypanocidal Rhodesain Inhibitors by Structure-Based Design
Maude Giroud, Bernd Kuhn, Sarah Saint-Auret, Christoph Kuratli, Rainer E Martin, Franz
Schuler, François Diederich, Marcel Kaiser, Reto Brun, Tanja Schirmeister, and Wolfgang Haap
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01870 • Publication Date (Web): 29 Mar 2018
Downloaded from http://pubs.acs.org on March 29, 2018
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 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 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.
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 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
Hꢀ1,2,3ꢀTriazoleꢀBased Dipeptidyl Nitriles: Potent,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Selective, and Trypanocidal Rhodesain Inhibitors by
StructureꢀBased Design
†
‡
‡
‡
‡
Maude Giroud, Bernd Kuhn, Sarah SaintꢀAuret, Christoph Kuratli, Rainer E. Martin,
‡
†
§,¥
§,¥
‖
Franz Schuler, François Diederich, Marcel Kaiser, Reto Brun, Tanja Schirmeister, *
‡
Wolfgang Haap *
†
Laboratorium für Organische Chemie, ETH Zurich, VladimirꢀPrelogꢀWeg 3, 8093 Zurich,
Switzerland
‡
Roche Pharma Research and Early Development (pRED), Roche Innovation Center Basel, F.
HoffmannꢀLa Roche Ltd., Grenzacherstrasse 124, 4070 Basel, Switzerland
§
Swiss Tropical and Public Health Institute, Socinstrasse 57, 4051 Basel, Switzerland
¥
University of Basel, Petersplatz 1, 4003 Basel, Switzerland
‖//
Institut für Pharmazie und Biochemie, Johannes GutenbergꢀUniversität Mainz, Staudinger
Weg 5, 55128 Mainz, Germany
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
KEYWORDS
rhodesain – African sleeping sickness – dipeptidyl nitrile – trypanosome – triazole – cysteine
protease inhibitor – SBDD.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
ACS Paragon Plus Environment

Page 3 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ABSTRACT: Macrocyclic inhibitors of rhodesain (RD), a parasitic cysteine protease and drug
target for the treatment of human African trypanosomiasis, had shown low metabolic stability at
the macrocyclic ether bridge. A series of acyclic dipeptidyl nitriles was developed using
structureꢀbased design (PDB ID: 6EX8). The selectivity against the closely related cysteine
protease human cathepsin L (hCatL) was substantially improved, up to 507ꢀfold. In the S2
pocket, 3,4ꢀdichlorophenylalanine residues provided high trypanocidal activities. In the S3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
pocket, aromatic residues provided enhanced selectivity against hCatL. RD inhibition (K values)
i
and in vitro cellꢀgrowth of Trypanosoma brucei rhodesiense (IC values) were measured in the
5
0
nanomolar range. Triazoleꢀbased ligands, obtained by a safe, gramꢀscale flow production of ethyl
1Hꢀ1,2,3ꢀtriazoleꢀ4ꢀcarboxylate, showed excellent metabolic stability in human liver microsomes
and in vivo halfꢀlives of up to 1.53 h in mice. When orally administered to infected mice,
parasitaemia was reduced but without complete removal of the parasites.
INTRODUCTION
Human African trypanosomiasis (HAT, or African sleeping sickness) is a protozoan infection
caused by two subspecies of Trypanosoma brucei (T. b.): T. b. gambiense, responsible for the
1
chronic form of HAT, and T. b. rhodesiense, responsible for the more acute form of the disease.
HAT consists of two clinical stages, the haemolymphatic stage, which, if left untreated,
2
,3
progresses into stage 2, the neurological stage, when the parasites enter the brain. To date, only
four therapies exist against HAT which are unsatisfactory because of their costs, toxicity, poor
1
bioavailability, long treatment, and lack of efficacy. Pentamidine and suramin are used against
stage 1 HAT, whereas eflornithine, melarsoprol, and, since 2009, the combination therapy
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
nifurtimoxꢀeflornithine (NECT) are applied to target stage 2. Melarsoprol, an organoarsenic
1
compound, remains the only effective treatment against stage 2 T. b. rhodesiense HAT. Because
the current treatments present nonꢀneglectable toxicity, there is a need to develop new therapies
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
against this disease, and the Drugs for Neglected Diseases initiative (DNDi) has set up a target
5
product profile with requirements that need to be met for the development of new compounds.
6
7
Two compounds are currently in clinical trials: fexinidazole, and oxaborole. A third compound,
8
,9
pafuramidine, failed in clinical trial. Because of the potential resistance of the parasite towards
these new compounds and uncertain outcome of the clinical trials, more research in this field is
1
0–14
needed.
In this context, we were interested in targeting rhodesain (RD), a trypanosomal cysteine
15
protease implicated in vital functions of the parasite. RD is a human cathepsin L (hCatL)ꢀlike
cysteine protease produced by T. b. rhodesiense and involved in the parasite’s iron
1
6–18
homeostasis.
It was also shown that RD is involved in parasite invasivity, allowing it to cross
1
9
20
the bloodꢀbrain barrier (BBB) of the human host, validating RD as a drug target against HAT.
20
Several classes of inhibitors have been developed against this enzyme, which bind to the
protease in either a covalent or a nonꢀcovalent manner.
Dipeptidyl nitrile ligands have been described to inhibit a variety of cysteine proteases in a
21,22
23
24
25
26
reversibleꢀcovalent manner, such as the human cathepsins B,
C, K, L (e.g. 1), S, and
2
7
28
the parasitic cysteine proteases cruzain (e.g. 1) or falcipainꢀ2 (Figure 1). Recently, dipeptidyl
nitrile 2 was reported to potently inhibit the target enzyme RD (K (RD) = 9.2 nM), with
i
significant selectivity against hCatL (SI = 102), however with modest in vitro antitrypanosomal
2
9
activity (8 µM). The electrophilic nitrile functions as head group for nucleophilic attack by the
4
ACS Paragon Plus Environment

Page 5 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
catalytic cysteine, forming a reversibleꢀcovalent thioimidate bond with the protein (schematically
3
0
shown for 2 in Figure 1).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Known dipeptidyl nitriles inhibiting the cysteine proteases hCatL and cruzain (1), and
RD (2). The thioimidate bond formed upon nucleophilic attack by Cys25 of RD is exemplified
for 2.
We recently discovered macrocyclic dipeptidyl nitriles, such as 3, as potent, trypanocidal RD
3
1
inhibitors (Figure 2). This ligand class is highly potent on both RD (inhibitory constants K
i
down to subꢀnanomolar activities) and T. b. rhodesiense (halfꢀmaximal inhibitory concentration
IC50 in the subnanomolar range). It did however not prevent complete recovery of the
trypanosomes in vivo, notably because of the metabolic instability of the ligands, which was
3
1
attributed to the extended carbon chain of the macrocycle. We therefore aimed at simplifying
this scaffold, removing the readily metabolized ether bridge of the macrocycle to increase the in
vivo exposure and at the same time greatly simplifying the synthetic route. Xꢀray crystal
3
1
structures of RD in complex with several macrocycles (PDB IDs: 6EX8, 6EXO, and 6EXQ)
enabled the use of structureꢀbased drug design in this quest for new RD inhibitors.
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Simplification of the macrocyclic scaffold, as in 3, towards dipeptidyl nitriles 4.
Herein, we report the structureꢀguided design, synthesis, and biological evaluation of 2Hꢀ
,2,3ꢀtriazoleꢀbased dipeptidyl nitriles 4 as potent RD inhibitors (Figure 2). Extensive
1
modifications of the substituents addressing the S3 pocket (5a–r), the S2 pocket (6a–j, 7a–c), or
both (8a–d) were performed. This required the development of an efficient flow synthesis of a
1
,2,3ꢀtriazole building block, allowing for gramꢀscale preparation within short reaction times.
The selectivity of the RD inhibitors against the structurally related enzyme hCatL was assessed
and the trypanocidal activities determined in vitro using a panel of parasites. Selected inhibitors
were further evaluated in doseꢀresponse recovery assays, their absorption, distribution,
metabolism, excretion, and toxicity (ADMET) properties measured, and mouse
3
2
pharmacokinetics (PK) studies performed. The Pꢀglycoprotein (Pꢀgp) efflux ratio (ER),
a
crucial parameter for targeting stage 2 HAT, was determined together with the in vivo evaluation
of the most promising compounds in infected female mice.
RESULTS AND DISCUSSION
Ligand Design. Based on our published crystal structure of macrocycle 3 in complex with
3
1
RD (PDB ID: 6EX8, 1.6 Å resolution), we removed the –O(CH ) O– linker of 3, which was
2 4
6
ACS Paragon Plus Environment

Page 7 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
31
previously identified as a site of metabolism. This modification resulted in a typical dipeptidyl
nitrile scaffold 4, exemplified with 5a (Figure 3). The design of new substituents on scaffold 4
occupying the S2 and S3 pockets of RD was performed with MOLOC using the MAB force
3
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
field. During the modeling, we kept the protein fixed but allowed for limited flexibility of
selected solventꢀexposed side chains such as Phe61 in the S3 pocket. The crystal structure of
macrocycle 3 in complex with RD revealed a nonꢀcovalent binding mode, due to reversible
oxidation of the catalytic cysteine to a sulfenic acid (CysꢀSOH) under the crystallization
31
conditions. Previously reported crystal structures of dipeptidyl nitriles in complex with the
2
4
27
cysteine proteases hCatL (PDB ID: 3HHA, 1.27 Å resolution) and cruzain (PDB ID: 4QH6,
3.13 Å resolution) suggest however, that this type of scaffold binds in a reversibleꢀcovalent
manner to Cys25, forming a thioimidate bond (Figure S1). Therefore, a covalent bond between
3
4
Cys25 of RD and the nitrile electrophile with a geometry taken from PDB ID 2XU1 was
installed for the modeling studies (Figure 3). The modeled dipeptidyl nitrile scaffold 5a overlays
well with macrocycle 3 in the crystal structure. In the S2 pocket of RD, which preferentially
35
accommodates lipophilic residues, a trifluoromethoxy group was introduced on the phenyl ring
in para position, aiming at mimicking a part of the macrocyclic linker and stabilizing the phenyl
ring against metabolic degradation. The 3ꢀchloro substituent was retained, as it proved to be the
31
best substituent in our macrocyclic series. In our binding mode model, this substituent
undergoes favorable dispersive contacts (≤ 4.0 Å) with Leu67 (d(CLeu67···C–Cl) = 3.4 Å,
d(CLeu67···Cl) = 3.6 Å), Met68 (d(CMet68···Cl) = 3.5 Å), and Ala208 (d(CAla208···Cl) = 4.0 Å). The
phenyl ring engages in C–H···π interactions with Leu67 on one side and with Ala138 and Leu160
on the other side, at C···C distances between 3.4 to 3.6 Å (Figure S2).
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
In the macrocyclic series, e.g. in 3, we had identified an intramolecular hydrogen bond
(
hereafter referred to as IMHB) between the amide NH of the S3 vector and the adjacent phenol
3
6,37
ether Oꢀatom.
We rationalized this IMHB to be in part responsible for the superior
3
1
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
trypanocidal activity of the macrocyclic series in comparison to other ligand classes tested and
therefore looked at various heterocycles which could offer a similar hydrogen bonding pattern. In
particular, the 1,2,3ꢀtriazole motif appeared to be suitable. Our calculations suggest that in the
2
unbound state the IMHB decreases the polar surface area (PSA) of ligand 5a by ~10 Å in
comparison to the predicted binding mode (Section S3 in the SI). In addition, the preferred
2
torsion angle
τ
of N ꢀphenylꢀ1,2,3ꢀtriazoles is nearly planar (
τ
(N–N–C –C ) = 0°–10°)
phe
phe
(
Figure S3). The agreement with CSD statistics for the torsional preferences of this single bond
is support for the binding mode model, but not for the IMHB. The fluoro substituent on the
phenyl ring was introduced to provide selectivity against hCatL, as it will closely and
unfavorably approach the backbone C=O of Gly61 while the analogous backbone C=O of Ile59
34
in RD is further away (Figure S4). The fluorophenyl moiety is involved in parallelꢀshifted π···π
stacking with Phe61 at a distance of 3.3–3.5 Å. Furthermore, the triazole N(3) undergoes an
orthogonal dipolar interaction with the backbone C=O of Gly66 at the entrance of the pocket at
d(N···C=OGly66) = 3.6 Å, α(N···C=OGly66) = 62°).
8
ACS Paragon Plus Environment

Page 9 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. (A) Schematic representation of designed dipeptidyl nitrile 5a; (B) predicted binding
31
mode of 5a in complex with RD (protein coordinates taken from PDB ID: 6EX8, 1.6 Å
3
1
resolution); (C) overlay of the crystal structure of macrocycle 3 in complex with RD (PDB ID:
6EX8, 1.6 Å resolution) with the model of 5a. Color code: C magenta, C green, O red, N blue,
3
5a
F cyan. The protein surface around the active site of RD in the crystal structure is represented in
gray.
Synthesis: General Approach. The approach to synthesize the dipeptides 9 consisted in
coupling warhead fragment 10 with a carboxylic acid 11, allowing for rapid derivatization
(Scheme 1).
a
Scheme 1. General Approach for the Preparation of Dipeptides 9.
O
O
O
O
H
2
N
HN
N
H
N
H
+
N
N
OH
10
11
9
a
General coupling conditions: HATU, iPr NEt, DMF. Blue ball: S2 vector; red ball: S3
2
vector. HATU = 1ꢀ[bis(dimethylamino)methylene]ꢀ1Hꢀ1,2,3ꢀtriazolo[4,5ꢀb]pyridinium 3ꢀoxide
hexafluorophosphate; DMF = N,Nꢀdimethylformamide.
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Synthesis of 3ꢀChloroꢀ4ꢀOꢀ(trifluoromethyl)ꢀLꢀtyrosineꢀBased Dipeptides. The 3ꢀchloroꢀ
ꢀOꢀ(trifluoromethyl)ꢀLꢀtyrosineꢀbased warhead fragment 12 was prepared in a threeꢀsteps
4
synthesis (Scheme 2). Commercially available 3ꢀchloroꢀ4ꢀOꢀ(trifluoromethyl)ꢀLꢀtyrosine
hydrochloride (13) was Bocꢀprotected to give 14 in 65% yield. The nitrile warhead was
subsequently introduced via amide coupling of aminocyclopropylcarbonitrile hydrochloride and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
4 in the presence of HATU and Hünig base to give 15 (65%), which was deprotected using neat
formic acid at 23 °C, yielding amine building block 12 quantitatively.
a
Scheme 2. Preparation of Building Block 12.
a
Reagents and conditions: (i) Boc O, 1 M NaOH, H O/1,4ꢀdioxane 1:1, 16 h at 22 °C, 65%; (ii)
2
2
1
ꢀaminocyclopropanecarbonitrile hydrochloride, HATU, iPr NEt, DMF, 2 h at 22 °C, 65%; (iii)
2
HCOOH, 2.5 h at 22 °C, quant. Boc = tertꢀbutoxycarbonyl.
In order to address the S3 pocket of RD with various triazole derivatives, we looked at
efficient ways of substituting a 1,2,3ꢀtriazole building block at the 2ꢀposition. Preparing ethyl
1
Hꢀ1,2,3ꢀtriazoleꢀ4ꢀcarboxylate (16) and subsequent derivatization appeared as a promising
3
8–41
42
approach. Based on literature precedence
and due to safetyꢀrelated considerations, we opted
4
3
for a scalable flow preparation of triazole 16, starting from commercially available
trimethylsilyl azide (TMSN , 17) and ethyl propiolate (18) (Figure 4). Although we initially
3
4
3,44
aimed for running the cycloaddition reaction under classical copper catalysis,
it became
10
ACS Paragon Plus Environment

Page 11 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
apparent that the reaction proceeds best in the absence of any copper species, thus avoiding the
precipitation of insoluble copper salts that eventually lead to blockage of the system. After rapid
optimization, we came up with a plug flow process in which both starting materials trimethylsilyl
azide (17) and ethyl propiolate (18) were injected neat via loops (2 mL) every 40 min using
acetonitrile as a carrier solvent. Three interlinked perfluoroalkoxy polymer (PTA) reactor coils
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(each 10 mL) equipped at the exit with a 250ꢀpsi backꢀpressure regulator to maintain system
pressure were operated at a temperature of 140 °C. The flow rate of each individual channel was
–1
set to 0.5 mL min , resulting in a total residence time of 30 min. The product stream was
collected, and trituration with diethyl ether provided the triazole 16 as a white solid in 81%
isolated yield.
18
16
17
Figure 4. Plug flow process for the production of ethyl 1Hꢀ1,2,3ꢀtriazoleꢀ4ꢀcarboxylate (16).
2
Procedures for the preparation of N ꢀsubstituted triazoles 19 remain scarce in the literature,
4
5
and the methods described often lack reliability. We opted for four different approaches:
46
47–49
50
BuchwaldꢀHartwigꢀtype couplings, ChanꢀLam couplings,
Mitsunobu reaction, and
3
8
alkylation. NꢀAlkylation or Nꢀarylation of 16 with the corresponding aryl or alkyl halides,
1
2
2
3
boronic acids, or alcohols 20c–q yielded mixtures of N /N or N /N regioisomers, respectively,
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
which could easily be separated by flash chromatography (Scheme 3). Buchwaldꢀtype
2
46
1
2
3
aminations led, as reported, exclusively to N ꢀsubstituted triazoles. The N ꢀ, N ꢀ, or N ꢀ
substitution pattern of Nꢀsubstituted 1,2,3ꢀtriazoleꢀ4ꢀ or 5ꢀcarboxylates was determined based on
5
1–55
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(i) existing reports,
(ii) the lower polarity of the N ꢀsubstituted triazoles in comparison to
1
13
15
their corresponding regioisomers, and (iii) H and C NMR, as well NꢀHMBC and NOESY
experiments (see Section S4 in the SI for a more detailed analysis). The 2Hꢀsubstituted triazolyl
ethyl esters 19c–q were subsequently saponified with LiOH in 2:2:1 H O/THF/MeOH, giving
2
carboxylic acids 21c–q. Amide coupling of commercially available 21a,b or 21c–q with amine
1
2 in the presence of HATU and Hünig base gave dipeptidyl nitriles 5a–q. At last, 5q was Nꢀ
Bocꢀdeprotected to give compound 5r.
a
Scheme 3. Preparation of Dipeptidyl Nitriles 5a–r
a
Reagents and conditions: (i) Me tBuXPhos, [Pd (dba) ], K PO , toluene, 3 h at 90 °C, 4–
4
2
3
3
4
2
3%; (ii) Cu(OAc) , pyridine or KOtBu, DMF or DMA, 48–72 h at 23 °C, 4–12%; (iii) K CO ,
2
2
3
MeCN, 18 h at 50 °C, 24%; (iv) 2ꢀ(tributylphosphoranylidene)acetonitrile, toluene 1–3 h at 90
12
ACS Paragon Plus Environment

Page 13 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
°
2
C, 19–38%; (v) tBuOH, H SO , TFA, 5 h at 23 °C, 32%; (vi) LiOH, THF/MeOH/H O 2:2:1, 1–
2
4
2
h at 23 °C, 70–100%; (vii) 12, HATU, iPr NEt, DMF, 1–16 h at 23 °C, 26–74%; (viii)
2
HCOOH, 1 h at 23 °C, 40%. XPhos = 2ꢀdicyclohexylphosphinoꢀ2′,4′,6′ꢀtriisopropylbiphenyl;
dba = (1E,4E)ꢀ1,5ꢀdiphenylpentaꢀ1,4ꢀdienꢀ3ꢀone; Ac = acetate; DMA = N,Nꢀdimethylacetamide;
TFA = trifluoroacetic acid; THF = tetrahydrofuran.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Preparation of 2ꢀ(4ꢀFluorophenyl)ꢀ2Hꢀ1,2,3ꢀTriazoleꢀBased Dipeptides. For this series,
the 2ꢀ(4ꢀfluorophenyl)ꢀ2Hꢀ1,2,3ꢀtriazole motif was kept constant and Lꢀtyrosyl, Lꢀphenylalanyl,
and Lꢀleucyl derivatives were introduced. The NꢀBoc protected Lꢀamino acids 22a–j were
coupled with 1ꢀaminocyclopropanecarbonitrile hydrochloride in the presence of HATU and
Hünig base to give 23a–j (Scheme 4). Subsequent deprotection of 23a–j with neat HCOOH gave
the corresponding Nꢀ(1ꢀcyanocyclopropyl)ꢀLꢀamino acid derivatives 24a–j, which were coupled
with commercially available 2ꢀ(4ꢀfluorophenyl)ꢀ2Hꢀ1,2,3ꢀtriazoleꢀ4ꢀcarboxylic acid (21a),
yielding final compounds 6a–j.
a
Scheme 4. Synthesis of Dipeptidyl Nitriles 6a–j
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Reagents and conditions: (i) 1ꢀaminocyclopropanecarbonitrile hydrochloride, HATU, iPr NEt,
2
DMF, 1–18 h from 0 to 23 °C, 39–98%; (ii) HCOOH, 2.5 h at 23 °C, 83–100%; (iii) 2ꢀ(4ꢀ
fluorophenyl)ꢀ2Hꢀ1,2,3ꢀtriazoleꢀ4ꢀcarboxylic acid, HATU, iPr NEt, DMF, 2 h at 0 °C, 25–94%.
2
OꢀAlkylated Lꢀtyrosine derivatives 7a–c were prepared in five steps, following an approach
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
1
similar to the one previously reported (Scheme 5). 3ꢀChloroꢀLꢀtyrosine derivative 25 was
treated with Cs CO and the corresponding alkyl bromides, to give bisꢀalkylation products 26a–c
2
3
in 54–70% yields. Saponification using LiOH afforded carboxylic acids 27a–c in 80% to
ꢁ
quantitative yield. Amide coupling with the nitrile head group led to ꢀ ꢀprotectedꢀOꢀalkylated
Lꢀtyrosine derivatives 28a–c in 44–86% yields. Deprotection with formic acid gave amines 29a–
c in 24–52% yields, which were subsequently coupled with 2ꢀ(4ꢀfluorophenyl)ꢀ2Hꢀ1,2,3ꢀ
triazoleꢀ4ꢀcarboxylic acid (21a) to yield dipeptidyl nitriles 7a–c in 24–37% yield.
a
Scheme 5. Synthesis of Dipeptidyl Nitriles 7a–c
a
Reagents and conditions: (i) a. Cs CO , 5 min at 22 °C, b. RCH Br, 3–16 h at 22 °C, 54–70%;
2
3
2
(ii) LiOH, MeOH/THF/H O 2:2:1, 1 h at 22 °C, 80–100%; (iii) 1ꢀaminocyclopropanecarbonitrile
hydrochloride, HATU, iPr NEt, DMF, 2 h at 0 °C, 44–86%; (iii) HCOOH, 2–4 h at 22 °C, 24–
2%; (iv) 2ꢀ(4ꢀfluorophenyl)ꢀ2Hꢀ1,2,3ꢀtriazoleꢀ4ꢀcarboxylic acid, HATU, iPr NEt, DMF, 2 h at
°C, 24–37%.
2
2
5
0
2
14
ACS Paragon Plus Environment

Page 15 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Preparation of 3,4ꢀDichloroꢀLꢀphenylalanineꢀbased Dipeptides. Amide coupling between
carboxylic acids 21l–n,p and Lꢀphenylalanine derivative 24h provided 8a–d in 32–56% yield
(Scheme 6).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Scheme 6. Synthesis of Dipeptidyl Nitriles 8a–d
a
Reagents and conditions: (i) HATU, iPr NEt, DMF, 2 h at 22 °C, 32–56%.
2
Biological and Physicochemical Studies. The binding affinities of all ligands reported herein
were tested against the cysteine proteases RD and hCatL in a fluorimetric assay (assay buffer for
RD: 50 mM sodium acetate pH 5.5, 5 mM EDTA, 200 m
M
NaCl, and 0.005% Brij35 at 25 °C;
assay buffer for hCatL: 50 m
M
Tris pH 6.5, 5 m
M
EDTA, 200 m NaCl, and 0.005% Brij35 at
M
35,56–58
2
5 °C), as previously reported.
A selectivity index (SI), defined as SI = K (hCatL)/K
i i
(
RD), was determined. In addition, cellꢀgrowth inhibition of T. b. rhodesiense (IC (T. b. rhod.))
50
®
and the cytotoxicity in Lꢀ6 rat myoblast cells (IC (Lꢀ6)) were determined in the Alamar Blue
5
0
5
9
©
60
growth inhibition assays. The lipophilicity (logD ) was determined in a CAMDIS assay
7
.4
61
and the solubility in a lyophilization assay (LYSA). The predicted lipophilicity (clogD ) was
7
.4
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
calculated with a Roche inꢀhouse developed machineꢀlearning tool based on a treeꢀbased
ensemble method.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
SAR Exploration of 2ꢀSubstituted Triazoleꢀbased Dipeptidyl Nitriles. Modifying the 2ꢀ
substituents on the 1,2,3ꢀtriazole ring led to potent ligands 5a–p,r of RD (K = 2.9–93 nM),
i
which efficiently inhibited the cellꢀgrowth of T. b. rhodesiense (IC (T. b. rhodesiense) = 0.2–67
50
nM), without being significantly cytotoxic (IC (Lꢀ6) = 7 to > 100 µM) (Table 1, Table S7, and
5
0
Figure S5 in the SI). Selectivity against hCatL could be obtained by introducing phenyl residues
at the 2ꢀposition of the triazole ring as in 5a–h (SI = 10–59). Introducing smaller, aliphatic
substituents as in 5i–p,r led to a decrease or complete loss of selectivity against hCatL. In the
paraꢀsubstituted phenyl series 5a,c–e, the binding affinities in RD decreased consistently upon
increase of the size of the paraꢀsubstituent (H<F<Cl<Br) of the phenyl ring. The same trend was
reflected in the binding affinities of hCatL and the cell growth inhibition of T. b. rhodesiense. In
our binding site models, the paraꢀsubstituent is in close contact with the backbone C=O of
Ile59(RD)/Gly61(hCatL) (e.g. for 5a in RD: d(C=OIle59···F–C) = 3.6 Å, α(C=OIle59···F) = 131°).
SAR and models suggest that there could be electrostatic repulsion for the fluorine substituent
with the backbone carbonyl oxygen and potential steric clashes with the larger para substituents
5
6,57
(Figure S4).
Clearly, the halide SAR does not support any involvement of halogenꢀbonding
interactions. In the pyridyl series 5f–h, the selectivity against hCatL was reduced (SI = 5–10) in
comparison to the phenyl series 5a–e (SI = 28–59) (Table 1). These compounds 5f–h were
however more potent in the cellꢀbased assay, with 4ꢀpyridine derivative 5h inhibiting in vitro cell
growth of T. b. rhodesiense in the subnanomolar range (IC = 0.6 nM). Introduction of aliphatic
50
and heteroalicyclic substituents in 5i–5p,r on the triazole ring resulted in improved solubility
16
ACS Paragon Plus Environment

Page 17 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
compared to phenyl derivatives 5a–e. However, any selectivity against hCatL was lost (Figure
S5). Pyrrolidine ligand 5r was very potent in the in vitro cellꢀgrowth inhibition assay (IC (T. b.
5
0
–
1
rhodesiense) = 0.2 nM), well soluble (253 µg mL ), and possessed the highest lipophilic
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
efficiency (LipE = 5.7) of the whole series.
Table 1. Predicted and Experimental Lipophilicity (c)logD , LYSA Solubility, Inhibition
7
.4
of RD and hCatL, Selectivity Indices (SI), Lipophilic Efficiency (LipE) for RD, Growth
Inhibition of T. brucei rhodesiense, and Cytotoxicity on Lꢀ6 cells of Inhibitors 5a–p,r.
R
N
N
N
O
O
HN
N
H
N
F
F
F
O
Cl
[
g]
[d]
[d]
IC50
(T.
Ki
Ki
[f]
[i]
[c]
LipE
b. rhod.) IC50
[nM]
[a]
[b]
LYSA
µg/mL]
(RD) (hCatL)
[e]
Cpd
R =
clogD7.4
logD7.4
SI
(Lꢀ6)
[
(
RD)
[µM]
18.8
17.9
8.3
[nM] [nM]
[h]
(
SI )
9
5
a
3.7
3.6
3.7
3.8
3.8
n.d.
n.d.
n.d.
n.d.
n.d.
<0.3
<0.3
<0.3
<0.3
<0.3
7.4
2.9
10
266
36
28
47
43
59
4.4
4.9
4.3
3.8
3.9
(
2089)
5
5b
81.7
467
(
3580)
2
2
5
c
(
377)
2
8
5d
25
1076
1116
7.6
(271)
6
7
5
e
19
>10
[
i]
(
n.a. )
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
6
5f
3.8
3.7
3.7
3.1
4.5
4.6
3.9
2.6
3.7
3.8
3.4
1.3
n.d.
0.5
4.2
5.2
53
4
14
36
16
23
18
31
11
31
41
12
18
93
143
353
82
15
4
10
10
5
4.0
3.7
4.1
4.5
3.2
2.9
4.1
4.9
3.7
4.1
4.3
5.7
7
(1167)
1
5
5
5
5
g
3.70
3.69
3.09
4.61
4.87
3.88
2.58
3.89
n.d.
>10
>10
>10
>10
7.6
48
(
n.a.)
.6
n.a.)
40
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
h
i
(
0.7
0.2
0.3
1
(
n.a.)
4
j
(
n.a.)
4
5k
0.9
11
4
8
(1900)
1
5
5
5
5
l
11
25
76
82
112
14
(
48000)
3
m
n
o
0.8
1.5
6.8
6.2
0.2
96
(32000)
9
2.2
0.3
90
253
50
(
5556)
6
31
(
5167)
13
5p
3.33
1.33
>100
56
(
n.a.)
0
.2
5
r
(
280000)
[
a] clogD calculated with an inꢀhouse developed machineꢀlearning tool based on a treeꢀbased
7.4
ensemble method; [b] logD7.4 = intrinsic distribution coefficient between octanol and aqueous
buffer (pH 7.4), measured in a CAMDIS assay; [c] Solubility determined by lyophilization
solubility assays (LYSA) at pH 6.5; [d] Average of two measurements, each performed in
duplicate. Standard deviations < 10%; [e] SI = selectivity index = K (hCatL)/K (RD); [f] LipE =
pK – clogD ; [g] Average of at least 2 measurements, standard deviations < 10%. T. b.
rhodesiense strain STIB 900, bloodstream form (trypomastigotes); [h] SI = selectivity index
relative to the Lꢀ6 cells; [i] Average of at least 2 measurements, standard deviations < 10%. Rat
skeletal myoblast cell Lꢀ6 strain; [j] n.a. = not applicable.
©
i
i
i
7.4
18
ACS Paragon Plus Environment

Page 19 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
SAR Exploration of the Vector Addressing the S2 Pocket of RD. This series of compounds
a–j and 7a–c led to larger variations in the enzymatic and parasitic activities than those seen in
6
the S3 pocket SAR (Table 2, for all ligands, see Table S8 in the SI). On the enzymatic level,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
binding affinities ranged from K (RD) = 7.4 to 597 nM, with selectivity indices ranging from 18
i
up to 507. Removing the 3,4ꢀsubstituents on the phenyl ring of dipeptidyl nitrile 5a as in 6a led
to a 2.3ꢀfold loss of affinity in RD, 2.4ꢀfold loss in hCatL affinity, and a 57ꢀfold loss in cell
growth inhibition, suggesting that the 3,4ꢀsubstitution pattern of the phenyl ring is critical to
achieve reasonable trypanocidal activity. This series 6a–j, 7a–c however possesses relatively
high logD values, and relatively low solubility, except for Lꢀleucyl ligand 6j.
7
.4
2ꢀSubstituted Lꢀphenylalanine derivatives 6b–d led to a decrease in binding affinity in RD
and even more pronounced in hCatL in comparison to the unsubstituted derivative 6a (Table S8).
This loss in affinity was also reflected in the in vitro cellꢀgrowth inhibition, suggesting that 2ꢀ
substitution of the ring is not ideal for enzymatic or trypanocidal activity. According to
molecular modeling, the pocket is narrow in this part of the binding site, potentially leading to
repulsive contacts for nonꢀhydrogen substituents (Figure S6 in the SI). 4ꢀSubstituted Lꢀ
phenylalanine derivatives 6f and 6g exhibited excellent selectivity over hCatL (SI = 507 for 6f,
SI = 87 for 6g) and good cellꢀgrowth inhibition (6f: IC (T. b. rhodesiense) = 6 nM, 6g: IC (T.
5
0
50
b. rhodesiense) = 19 nM) (see Section S5.2 in the SI for a more detailed analysis of the high
selectivity of 6f). 3ꢀChloroꢀ4ꢀsubstituted Lꢀphenylalanine or Lꢀtyrosine derivatives 6h–j, 7a–b
were potent rhodesain inhibitors (K (RD) = 2–16 nM). In terms of potency (K (RD) = 2 nM,
i
i
LipE = 5.1), selectivity against hCatL (SI = 66), and cellꢀgrowth inhibition (IC (T. b.
5
0
rhodesiense) = 7 nM), the 3,4ꢀdichlorophenyl derivative 6h was the best inhibitor of the whole
series. On the other hand, Lꢀtyrosine derivatives 7a–c, although efficient inhibitors of RD (Ki
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
RD) = 9–16 nM) and moderate inhibitors of hCatL (K (hCatL) = 181–314 nM), were less potent
i
in the in vitro cellꢀbased assay (IC (T. b. rhodesiense) = 27–190 nM) compared to the rest of
5
0
this series. The Lꢀleucyl derivative 6j did not inhibit RD (K (RD) = 23 nM, LipE = 4.3) or the
i
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
parasitic cell growth (IC (T. b. rhodesiense) = 2.2 µM) as efficiently as the Lꢀtyrosine or Lꢀ
5
0
phenylalanine derivatives, although it was superior in terms of lipophilicity (logD7.4 = 3.3) and
–
1
solubility (LYSA = 42 µg mL ).
In summary, 6f, 6h, and 6i were the most promising compounds of this SAR. In comparison to
macrocycle 3, they were not only more active on RD, but also more selective against hCatL. In
addition, they exhibited high trypanocidal activity (IC (T. b. rhodesiense) < 10 nM).
5
0
Table 2. Predicted and Experimental Lipophilicity (c)logD , Inhibition of RD and hCatL,
7
.4
Selectivity Indices (SI), Lipophilic Efficiency (LipE) for RD, Growth Inhibition of T. brucei
rhodesiense, and Cytotoxicity on Lꢀ6 cells of Selected Inhibitors. LYSA Solubility of
–1
Inhibitors Reported in this Table Is < 0.3 µg mL .
F
N
N
N
O
O
HN
N
H
N
R
[
f]
IC50
[
h]
IC50
[
c]
[c]
[a
Ki
Ki
[e]
clogD7.4
(T. b. rhod.)
LipE
[
b] (RD) (hCatL)
[d]
Cpd
5a
R =
logD7.4
SI
(Lꢀ6)
[µM]
]
(RD)
[nM]
[
nM] [nM]
[g]
(
SI )
9
F
F
F
3.7
n.d.
7.4
266
36
4.4
18.8
O
(2089)
Cl
20
ACS Paragon Plus Environment

Page 21 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
6
6
f
3.7
3.6
3.97
n.d.
3
1522
1473
507
87
4.8
4.2
>10
>10
[i]
Cl
(n.a. )
1
9
6
g
17
O
(
n.a.)
7
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
h
3.6
3.5
3.6
n.d.
3.63
n.d.
2
7
131
253
314
66
36
20
5.1
4.6
4.2
9.7
>10
7.6
Cl
(
1386)
6
Cl
6
i
N
(n.a.)
Cl
2
7
7
b
16
O
(
281)
Cl
[
a] clogD calculated with an inꢀhouse developed machineꢀlearning tool based on a treeꢀbased
7.4
ensemble method; [b] logD7.4 = intrinsic distribution coefficient between octanol and aqueous
buffer (pH 7.4), measured in a CAMDIS assay; [c] Average of two measurements, each
performed in duplicate. Standard deviations < 10%; [d] SI = selectivity index = K (hCatL)/K
RD); [e] LipE = pK – clogD ; [f] Average of at least 2 measurements, standard deviations <
0%. T. b. rhodesiense strain STIB 900, bloodstream form (trypomastigotes); [g] SI = selectivity
index relative to the Lꢀ6 cells; [h] Average of at least 2 measurements, standard deviations <
0%. Rat skeletal myoblast cell Lꢀ6 strain; [i] n.a. = not applicable.
©
i
i
(
i 7.4
1
1
SAR Exploration of 3,4ꢀDichloroꢀLꢀphenylalanine Derivatives. The 3,4ꢀdichloroꢀLꢀ
phenylalanine derivative 6h (K (RD) = 2 nM, IC (T. b. rhodesiense) = 7 nM) was identified
i
50
from the S2 pocket SAR as a promising candidate for further optimization. In addition, the 3,4ꢀ
dichloroꢀLꢀphenylalanine possesses the advantage of being a much cheaper building block than
the corresponding 3ꢀchloroꢀ4ꢀtrifluoromethoxy moiety. We therefore varied the S3 substituents
on the 1,2,3ꢀtriazole ring. In this subseries, we confirmed that introducing saturated substituents
in the S3 pocket decreases the selectivity over hCatL (Table 3). With the exception of 8d (Ki
(
RD) = 4 nM, LipE = 5.7), the dipeptidyl nitriles featuring a cyclobutyl (8a, K (RD) = 71 nM,
i
LipE = 3.5), oxetanyl (8b, K (RD) = 33 nM, LipE = 5.4), or difluorocyclobutyl (8c, K (RD) = 18
i
i
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
nM, LipE = 4.3) moiety on the triazole were slightly less active on RD than the corresponding
inhibitors possessing a 4ꢀfluorophenyl ring (6h, K (RD) = 2 nM, LipE = 5.1) in the S3 pocket.
i
However 8a–c were highly trypanocidal, with IC values < 10 nM, suggesting that the 3,4ꢀ
50
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
dichloroꢀLꢀphenylalanine motif is particularly good for in vitro cellꢀgrowth inhibition. With
clogD7.4 = 2.1–3.6, they are also less lipophilic than the corresponding 3ꢀchloroꢀ4ꢀOꢀ
trifluoromethoxyꢀLꢀphenylalanine inhibitors 6l–m,p (clogD = 2.6–3.9).
7.4
Table 3. Predicted clogD , Inhibition of RD and hCatL, Selectivity Index (SI), Lipophilic
7
.4
Efficiency (LipE) for RD, Growth Inhibition of T. b. rhodesiense, and Cytotoxicity on Lꢀ6
Cells of Inhibitors 8a–d.
R
N
N
N
O
O
HN
N
H
N
Cl
Cl
[b]
[e]
Ki
(RD)
[b]
IC50 (T. b.
Ki
(hCatL)
nM]
[d]
[g]
[a]
[c]
LipE
RD)
rhod.) [nM]
IC50 (Lꢀ6)
Cmpd
8a
R =
clogD7.4
SI
(
[µM]
[
[f]
[
nM]
(SI )
4
3.6
2.1
3.4
2.7
71
7
0.1
2.4
3.6
23
3.5
5.4
4.3
5.7
58.2
(
14550)
8
8
b
33
18
4
78
64
93
>100
54.75
>100
[h]
(
n.a. )
8
8c
(6844)
2
3
8
d
(
n.a.)
[
a] clogD calculated with an inꢀhouse developed machineꢀlearning tool based on a treeꢀbased
7.4
ensemble method; [b] Average of two measurements, each performed in duplicate. Standard
22
ACS Paragon Plus Environment

Page 23 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
deviations < 10%; [c] SI = selectivity index = K (hCatL)/K (RD); [d] LipE = pK – clogD ; [e]
i
i
i
7.4
Average of at least 2 measurements, standard deviations < 10%. T. b. rhodesiense strain STIB
00, bloodstream form (trypomastigotes); [f] SI = selectivity index relative to the Lꢀ6 cells; [g]
9
Average of at least 2 measurements, standard deviations < 10%. Rat skeletal myoblast cell Lꢀ6
strain; [h] n.a. = not applicable.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Reversibility of Drug Effect. In addition to the IC50 determination in the standard 72 h
6
2
assay, the reversibility of drug effect was examined. Pulse incubation of T. b. rhodesiense with
a and 5b shows that at least a 24 h drug pressure is required to produce an irreversible effect on
parasite survival (Table 4). Melarsoprol, one of the few commercially available drugs against
5
6
3
HAT, was used as a standard. Inhibitor 5a (IC (48 h) = 0.064 µM) was still inhibiting the cellꢀ
5
0
growth after 48 h incubation, comparable to melarsoprol (IC (48 h) = 0.026 µM), and superior
5
0
to inhibitor 5b (IC (48 h) = 0.462 µM).
5
0
Table 4. Doseꢀresponse Recovery of 5a,b Against T. brucei rhodesiense.
[a]
IC (T. b. rhodesiense) [µM]
50
Cpd
2 h
>1
>1
6 h
>1
>1
16 h
>1
>1
24 h
0.45
0.556
0.044
48 h
0.064
0.462
0.026
5
5
a
b
Melarsoprol
>0.18
0.099
0.086
[
a] Average of at least 2 measurements, standard deviations < 10%. T. b. rhodesiense strain
STIB 900, bloodstream form (trypomastigotes)
Drug Metabolism and Pharmacokinetics. Key physicochemical parameters (Table 5), and
mouse pharmacokinetics (Table 6) were determined for a set of dipeptidyl nitriles with
promising in vitro antitrypanosomal activities.
These compounds possess low in vitro human and mice microsomal clearances, consistent
with good metabolic stability (Table 5). Their octanolꢀwater distribution coefficient (logD ) was
7
.4
5
lower than the threshold of 5 set by the DNDi’s target product profile, however they were
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
poorly soluble (LYSA < 1 ꢀg/mL), except for 5i (LYSA = 53 ꢀg/mL). The PAMPA values
–
1
–1
showed a relatively low permeability, except for 5k (28.6 nm s ) and 6h (10.7 nm s ). The
compounds did not significantly inhibit CYP3A4 or CYP2C9, but were inhibiting CYP2D6 in
the low ꢀM range.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
2
According to the target product profile, the PSA should be ≤ 85 Å to promote good brain
3
3
penetration. Polar surface areas were generally calculated using the MOLOC companion
33
program Msrfvl based on a lowꢀenergy conformation generated with the MOLOC program
Mol3d (Table 5, see Table S1 in the SI for all data). To explore how useful the Mol3d based
conformations are for PSA calculations, we investigated the dipeptidyl nitrile 5a in more detail.
Figure 5 shows model conformations representative of the bound and unbound states,
respectively. In particular, Figure 5A shows a model of the bound state in RD derived from the
2
predicted binding mode in Figure 3. The PSA is 108.9 Å , and no intramolecular hydrogen bond
is apparent. Figure 5B shows the low energy conformation in the unbound state generated with
Mol3d, which is using a molecular mechanics (MM) based representation. Two intramolecular
2
hydrogen bonds are visible and the predicted PSA is 107.3 Å . We compared this with a more
sophisticated approach using a MM conformational search (Maestro version 2016ꢀ4, OPLS3
force field, implicit solvent CHCl ) followed by a quantumꢀmechanical (QM) optimization
3
(
Gaussian 09 Rev. D.01, B3LYP/ccꢀpVDZ, implicit solvent CHCl ) of all conformers below an
3
energy cutoff ꢁEMM: 3.0 kcal/mol (RMSD cutoff: 0.5 Å). The resulting conformer with lowest
QM energy is depicted in Figure 5C. While the overall conformation is different compared to the
pure MM approach (Figure 5B) the same two intramolecular hydrogen bonds are formed. The
2
2
PSA for the QM calculated conformation is 99.9 Å , which is roughly 10 Å smaller compared to
64
the bound state (Figure 5A). In comparison, a 2D TPSA calculation for this molecule yields
24
ACS Paragon Plus Environment

Page 25 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
2
1
21.9 Å , which is strongly overestimating the 3D QMꢀbased value. From this analysis, we
conclude that 2D TPSA values are not suitable for this system involving several intramolecular
2
hydrogen bonds. The PSA was reduced by about 10 Å when replacing the 4ꢀtrifluoromethoxy
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
group as in 5a by a 4ꢀchloroꢀsubstituent as in 6h, while other properties were very similar (Table
5
), suggesting that the 3,4ꢀdichloroꢀLꢀphenylalanine motif could be the most suitable choice for
further optimizations.
A
B
C
2
2
2
PSA = 108.9 Å
PSA = 107.3 Å
PSA = 99.9 Å
Figure 5. Conformation of dipeptidyl nitrile 5a in the (A) bound conformation (extracted from
the predicted binding mode in RD, protein coordinates taken from PDB ID: 6EX8, 1.6 Å
3
resolution); (B) low energy conformation in the unbound state generated with Mol3d; (C)
unbound conformation using a MM conformational search with subsequent QM
optimization. Polar surface areas were calculated with the MOLOC companion program
Msrfvl. Color code: green C5a,bound, cyan C5a,unbound, red O, blue N, lime Cl, light blue F.
Distances represented by dotted black lines and given in Å.
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 5. In Vitro Data for Selected Compounds.
5
a
5f
5g
5i
5j
5k
6h
[a]
IC50
nM]
(T. b. rhod.)
9
6
1
40
4
4
7
[
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
[
b]
IC50 (Lꢀ6 cells) [µM] >10
6.99
<10
>10
<10
>10
<10
>10
11
7.63
<10
9.67
<10
[
c]
hCl
<10
[
c]
mCl
<10
n.d.
<0.3
n.d.
<10
n.d.
0.5
11
<10
3.09
53
26
17
<10
[
d]
logD7.4
3.70
4.2
4.61
4
4.87
0.9
n.d.
[e]
LYSA
<0.3
10.72
[f]
PAMPA
2.91
4.47
1.2
1.38
28.62
CYP
3A4/2D6/2C9
IC50
[µM]
>
50/3/9
47/1/11
8/2/7
>50/6/>50 n.d.
>50/0.5/5
>50/5/4
[
g]
2
PSA [Å ]
107.3
119.1
2
116.1
2
109.2
105.2
107.2
98.9
2
[h]
HBD
HBA
2
9
2
9
2
9
2
9
[i]
10
10
8
[
a] Average of at least 2 measurements, standard deviations < 10%. T. b. rhodesiense strain
STIB 900, bloodstream form (trypomastigotes); [b] Average of at least 2 measurements, standard
deviations < 10%. Rat skeletal myoblast cell Lꢀ6 strain; [c] Microsomal intrinsic clearances [µ
L
–1
–1
min mg ] in human (h) and mouse (m); [d] logD = intrinsic distribution coefficient between
7
.4
©
octanol and aqueous buffer (pH 7.4), measured in a CAMDIS assay; [e] Solubility determined
by lyophilisation solubility assays (LYSA) at pH 6.5; [f] Membrane permeability [nm s ]
derived from the PAMPA assay; [g] Polar Surface Area [Å ] calculated using Mol3d/Msrfvl; [h]
Number of hydrogenꢀbond donors; [i] Number of hydrogenꢀbond acceptors.
–
1
2
The PK profile of these compounds was further characterized in mice, following intravenous
administration (Table 6, Section S6 in the SI). Compounds 5a and 6h possessed relatively flat
curves, and therefore no complete area under the curve could be determined (Section S6 in the
SI). For the remaining dipeptidyl nitriles, the halfꢀlifes (t ) spanned from 0.94–1.53 h, and the
1/2
–
1
–1
clearances (CL) from 10.2–32.9 mL min kg . Finally, the unbound fraction of plasma protein
31
binding was between 1.48 % for 5f to 22.64 % for 5i. In comparison to our macrocyclic series,
26
ACS Paragon Plus Environment

Page 27 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the halfꢀlife was slightly reduced. All compounds possess a fraction unbound (%f ) in the plasma
u
protein binding assay which is in the typical range for good drugs.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 6. Mouse Pharmacokinetic Profile of Selected Compounds.
5
a
5f
5g
5i
5j
5k
6h
Route
i.v.
i.v.
i.v.
i.v.
i.v.
i.v.
1.3
1.30
17
i.v.
–
1
Dose [mg kg ]
1.2
0.8
1.5
1.5
1.5
1.3
t1/2 [h]
n.d.
n.d.
n.d.
1.42
10.2
1.2
0.94
22.6
0.8
1.52
22.8
2.7
1.53
32.9
3.0
>6.4
<4.6
<2.5
>3611
n.d.
–1
–1
CL [mL min kg ]
–
1
V [L kg ]
ss
1.7
983
n.d.
–
1
AUCpo/D [ng h mL ] n.d.
1627
1.48
738
3.95
730
22.64
507
8.26
[
a]
PPB human %fu
n.d.
[
a] Plasma protein binding, fraction unbound.
Brain Penetration of Dipeptidyl Nitriles. The ability of a drug to cross the BBB is crucial if
6
5
intended to treat stage 2 HAT. The Pꢀglycoprotein (Pꢀgp) efflux ratio (ER) (one of the
parameters used to evaluate brain penetration) of dipeptidyl nitrile 5a was determined in
LLCPK1 cells stably expressing human MDR1 (Table 7). This ER was improved by a factor 4.3
in human for 5a (ER = 6 in human) in comparison to macrocycle 3 (ER = 26), and by a factor 2.7
in mice. This improvement is remarkable, especially taking into account the predicted PSA of the
2
2
respective compounds (PSA = 107.3 Å for 5a, and PSA = 80.2 Å for 3).
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 78
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 7. PꢀGlycoprotein (Pꢀgp) Efflux Ratio (ER) and Estimated PSA of Selected
Inhibitors.
[a]
[b]
2
Pꢀgp Human ER
Pꢀgp Mouse ER
PSA [Å ]
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
[
c]
3
26
6
53
20
80.2
[c]
5
a
107.3
[
a] Human efflux ratio (ER) in LLCPK1 cells stably expressing human MDR1; [b] Mouse
2
efflux ratio (ER) in LLCPK1 cells stably expressing mouse Mdr1a; [c] PSA [Å ] = polar surface
area.
Antiparasitic Activity Against Other Parasites and Cytotoxicity. The in vitro cellꢀgrowth
inhibition of dipeptidyl nitriles 5c–p,r, 6a–j, 7a–c, 8a–d was determined for a panel of protozoan
parasites consisting, in addition to T. b. rhodesiense (Section S7 in the SI), of Leishmania
6
6
donovani, T. cruzi, and Plasmodium falciparum, the causative agents of Leishmaniasis,
6
7
68
Chagas’ disease, and malaria tropica, respectively. The IC values obtained by using the
5
0
corresponding typical antiprotozoal drugs are reported as mean of comparison. All compounds
presented good to excellent selectivity against the other parasites, suggesting that our inhibitors
2
7
selectively target RD, despite the structural similarities with the parasitic enzymes cruzain
2
8
66
(
from T. cruzi), falcipainꢀ2 (from P. falciparum), and Leishmania cysteine proteases. In
addition, they all possess moderate to low cytotoxicity, as demonstrated by the IC50 values
against Lꢀ6 rat myoblast cells.
In Vivo Evaluation of Selected Dipeptidyl Nitriles. Based on the reported in vitro data, we
selected inhibitors 5f, 5h, 5i, 5m, 5r, 6h, 8b, and 8c for in vivo evaluation. The compounds were
dissolved in DMSO/H O and orally administered in female mice on three consecutive days at a
2
dose of 50 mg/kg/day, beginning 24 h post infection (Table 8, for full data set, see Section S8 in
the SI). Despite their in vitro potencies, none of the compounds cured any infected mice. 6 out of
28
ACS Paragon Plus Environment

Page 29 of 78
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
8
in vivo tested compounds exhibited some activity (5f, 5h, 5i, 5m, 6h, and 8b) and did cause a
reduction in parasitaemia of 45–54 % at 24 h after the final dose. This limited activity may be
due to insufficient systemic drug exposure despite the rather high daily doses. Based on their in
vitro solubility, permeability and metabolic stability data it can be assumed that the compounds
are incompletely absorbed and/or too rapidly eliminated which would result in subꢀefficacious
exposure over the dosing interval. In order to improve these results, increasing the in vivo
exposure of the compounds would be required. In particular, replacing one amide bond by an
amide isoster, as performed for the development of cruzain inhibitors, would be a next step to be
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
7
undertaken.
Table 8. In Vivo Evaluation of a Set of Dipeptides Orally Administered to Female Mice
Infected with T. b. rhodesiense (STIB900). Values are means of 2 mice, and 4 mice were
included in the untreated control group.
–1
Parasites / µL of
blood
[a]
Cpd
Dose [mg kg ]
% of control
% of activity
Ctrl
5f
untreated
3 x 50
3 x 50
3 x 50
3 x 50
3 x 50
3 x 50
3 x 50
3 x 50
224500
132850
127250
104000
127710
223500
118330
122890
189000
59.2
56.7
46.4
56.9
99.6
52.8
54.7
84.2
40.8
43.3
53.6
43.1
0.4
5
h
5
i
5
m
5
r
h
b
6
47.2
45.3
15.8
8
8c
[a] Activity represents the reduction of parasitaemia compared with untreated controls.
29
ACS Paragon Plus Environment