Active Site Mapping of Human CathepsinF with Dipeptide Nitrile Inhibitors

Article abstract of DOI:10.1002/cmdc.201500151

Cleavage of the invariant chain is the key event in the trafficking pathway of major histocompatibility complex classII. CathepsinS is the major processing enzyme of the invariant chain, but cathepsinF acts in macrophages as its functional synergist which is as potent as cathepsinS in invariant chain cleavage. Dedicated low-molecular-weight inhibitors for cathepsinF have not yet been developed. An active site mapping with 52 dipeptide nitriles, reacting as covalent-reversible inhibitors, was performed to draw structure-activity relationships for the non-primed binding region of human cathepsinF. In a stepwise process, new compounds with optimized fragment combinations were designed and synthesized. These dipeptide nitriles were evaluated on human cysteine cathepsinsF, B, L, K and S. Compounds 10 (N-(4-phenylbenzoyl)-leucylglycine nitrile) and 12 (N-(4-phenylbenzoyl)leucylmethionine nitrile) were found to be potent inhibitors of human cathepsinF, with K_i values <10nM. With all dipeptide nitriles from our study, a 3D activity landscape was generated to visualize structure-activity relationships for this series of cathepsinF inhibitors. Mapping with nitriles: For human cathepsinF, low-molecular-weight inhibitors have not been developed so far. Therefore, a library of 52 dipeptide nitriles, known to interact in a covalent but reversible manner with the active site cysteine, was evaluated for cathepsinF inhibition. With the kinetic data in hand, optimized candidates were designed, synthesized, and tested to improve the activity profile as cathepsinF inhibitors.

Full text of DOI:10.1002/cmdc.201500151

DOI: 10.1002/cmdc.201500151
Full Papers
Active Site Mapping of Human Cathepsin F with Dipeptide
Nitrile Inhibitors
Janina Schmitz,[a, b] Norbert Furtmann,[a, c] Moritz Ponert,[a] Maxim Frizler,[a] Reik Lçser,[a, d]
Ulrike Bartz,[b] Jürgen Bajorath,[c] and Michael Gütschow*[a]
Cleavage of the invariant chain is the key event in the traffick-
ing pathway of major histocompatibility complex class II. Cath-
epsin S is the major processing enzyme of the invariant chain,
but cathepsin F acts in macrophages as its functional synergist
which is as potent as cathepsin S in invariant chain cleavage.
Dedicated low-molecular-weight inhibitors for cathepsin F
have not yet been developed. An active site mapping with 52
dipeptide nitriles, reacting as covalent–reversible inhibitors,
was performed to draw structure–activity relationships for the
non-primed binding region of human cathepsin F. In a stepwise
process, new compounds with optimized fragment combina-
tions were designed and synthesized. These dipeptide nitriles
were evaluated on human cysteine cathepsins F, B, L, K and S.
Compounds 10 (N-(4-phenylbenzoyl)-leucylglycine nitrile) and
12 (N-(4-phenylbenzoyl)leucylmethionine nitrile) were found to
be potent inhibitors of human cathepsin F, with Ki values
<10 nm. With all dipeptide nitriles from our study, a 3D activi-
ty landscape was generated to visualize structure–activity rela-
tionships for this series of cathepsin F inhibitors.
Introduction
Cysteine cathepsins are lysosomal proteases that play crucial
roles in various physiological processes, such as bone remodel-
ing, osteoporosis, neurological disorders, autoimmune re-
sponse and cancer, and represent the largest and best-charac-
terized group of cathepsins.[1,2] One of its representatives, cath-
epsin F, is predominantly expressed in macrophages, but not
in dendritic cells or B cells.[3,4] As other cysteine cathepsins,
cathepsin F is translated as an inactive precursor, targeted to
the lysosomes and activated by cleavage and dissociation from
its pro-region. It is an active endopeptidase and contains a cys-
tatin-like domain in the elongated N-terminal pro-region,
a unique feature among the cysteine cathepsins.[5]
rough endoplasmic reticulum (RER) and is assembled with
a type II transmembrane protein referred to as the chaperone
invariant chain. MHC class II mediated antigen presentation re-
quires the participation of endosomal and lysosomal proteases
for two proteolytic events: 1) cleavage of the antigens to small
antigenic peptides and 2) degradation of the invariant chain.
The MHC class II binding site is blocked by the invariant chain,
prohibiting preterm peptide loading. The complex traffics from
the RER through the trans-Golgi network to the early endo-
somes. During transport, the invariant chain is degraded by en-
dosomal and lysosomal proteases. The remaining fragment of
the invariant chain in the MHC II binding groove, Iip10, is
cleaved to a class II-associated invariant chain peptide (CLIP)
by certain cysteine proteases. These CLIP fragments can be re-
placed by antigenic peptides. Loaded MHC class II molecules
are transported to the cell surface, allowing the display of anti-
genic peptides to Tcells. When the processing of Iip10 to CLIP
is discontinued, the transport of MHC II to the cell surface is
delayed due to its retention in the lysosomes. Thus, the cleav-
age of Iip10 is thought to be the key event in the antigen-pre-
sentation pathway of MHC class II complexes. The major proc-
essing enzyme in dendritic cells, macrophages, and B cells in
this pathway is cathepsin S, but cathepsin L can carry out this
function in thymic epithelial cells.[4,6] Notably, it was observed
that macrophages of mice, deficient in cathepsins S and L,
were still capable of processing the invariant chain and loading
peptides onto MHC class II complexes. It was shown for puri-
fied cathepsin F that the enzyme is as potent as cathepsin S in
invariant chain processing and CLIP formation.[6] Moreover,
with an inhibitor of the cysteine cathepsin family (I,
Figure 1),[7,8] invariant chain processing could be fully blocked
The major histocompatibility (MHC) class II pathway allows
the presentation of exogenous antigens via the endosomal
route. MHC class II is an a/b heterodimer synthesized in the
[a] J. Schmitz, N. Furtmann, M. Ponert, Dr. M. Frizler, Dr. R. Lçser,
Prof. Dr. M. Gütschow
Pharmaceutical Institute, Pharmaceutical Chemistry I
University of Bonn, An der Immenburg 4, 53121 Bonn (Germany)
E-mail: guetschow@uni-bonn.de
[b] J. Schmitz, Prof. Dr. U. Bartz
Department of Natural Sciences
University of Applied Sciences Bonn-Rhein-Sieg
von-Liebig-Straße 20, 53359 Rheinbach (Germany)
[c] N. Furtmann, Prof. Dr. J. Bajorath
Department of Life Science Informatics, B-IT
LIMES Program Unit Chemical Biology and Medicinal Chemistry
University of Bonn, Dahlmannstraße 2, 53113 Bonn (Germany)
[d] Dr. R. Lçser
Helmholtz-Zentrum Dresden-Rossendorf
Institute of Radiopharmaceutical Cancer Research
Bautzner Landstraße 400, 01328 Dresden (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cmdc.201500151.
in catSÀ/À/catLÀ/À mice. Vinyl sulfone
I is selective for
ChemMedChem 2015, 10, 1365 – 1377
1365
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
former amino acid attacking the carbonyl atom of the scissile
peptide bond, leading to the hydrolytic cleavage of the pepti-
dic substrate in the course of an acyl transfer reaction. Nitrile-
based inhibitors can be designed by starting from the specific
substrate structure of the target cysteine protease by introduc-
ing a nitrile group as reactive warhead in place of the carbonyl
group and additional peptidomimetic modifications. Such pep-
tide nitriles and azapeptide nitriles have attracted much atten-
tion as highly potent inhibitors.[2,8,13–20] The covalent reaction
of peptide nitriles with cysteine cathepsins involves attack of
the active-site thiolate at the inhibitor’s nitrile carbon atom
and reversible formation of a thioimidate adduct (Scheme 1).[2]
Scheme 1. Reaction of dipeptide nitriles with cysteine cathepsins and rever-
sible formation of thioimidates (CG=capping group).
Figure 1. Structures of peptidomimetic cathepsin inhibitors.
It has been shown for several cysteine cathepsins that di-
peptide nitriles represent promising therapeutics. Cathepsin K
is the major processing enzyme in osteoclastic bone resorption
and is responsible for the degradation of bone matrix. Balicatib
(II), a basic and hence lysosomotropic cathepsin K inhibitor,
shows the characteristic features of dipeptide nitrile inhibitors
(Figure 1). The replacement of the P3–P2 amide bond by a tri-
fluoroethylamine fragment and the introduction of a 1,1-cyclo-
propane ring at the P1 position led to the front-runner odana-
catib (III) with decreased metabolic liabilities. Odanacatib is
currently being developed as a once-weekly treatment for os-
teoporosis.[15,21]
cathepsin S at low concentrations (10 nm), whereas at the
higher concentration used, additional cysteine proteases were
inhibited, which demonstrated a further cysteine protease,
namely cathepsin F, to be involved in the degradation pro-
cess.[3]
In catFÀ/À mice, severe neurological disorders were observed
starting after 12–16 months, involving incoordination, weak-
ness, and premature death, as well as lysosomal storage de-
fects. Large amounts of lipofuscin, characteristic of neuronal
ceroid lipofuscinosis (NCL), were found to accumulate in the
central nervous system of cathepsin F-deficient mice.[9] Such
storage defects in humans may be induced by cathepsin F
gene mutations leading to storage disorders such as Type B
Kufs disease, an adult-onset NCL.[10]
The inhibitor interactions with the S2 and S3 binding sites
are essential to achieve selectivity toward other cathepsins.
This was also taken into account in the design of cathepsin S
inhibitors, among which IV (Figure 1) exhibits high potency, se-
lectivity, and bioavailability.[13] This compound was developed
by the incorporation of an isobutylsulfonylcysteine building
block in P2, and a para-fluoro-substituted phenyl residue at
the P3 position, advantageous for occupying the deep S2
pocket and the rather small S3 binding site, respectively.
The crystal structure of human cathepsin F has been deter-
mined in complex with the vinyl sulfone inhibitor V.[22] The
structure of cathepsin F adopts a typical fold of papain-like en-
zymes with some unique features in the S2 and S3 binding
sites. The S2 pocket of cathepsin F is defined by Leu67 and
Met205; in cathepsin S these amino acids are replaced by two
phenylalanines. The methionine residue in cathepsin F points
toward the leucine residue, decreasing the size of the binding
pocket, while the S3 pocket is wider and shallower than the S2
pocket.[5,22]
Furthermore, cathepsin F has been implicated in atherogen-
esis. The enzyme appeared to be responsible for the degrada-
tion of low-density lipoprotein (LDL) and an enhancement of
the binding of LDL to proteoglycans. The resulting generation
and accumulation of extracellular lipid droplets in the arterial
intima may yield to atherogenesis.[11]
The implication of cathepsin F in various physiological condi-
tions should provide the impetus for the development of low-
molecular-weight inhibitors or activity-based probes as possi-
ble therapeutic agents or tool compounds, respectively, to fur-
ther elucidate the physiological role(s) of this protease. Given
that cathepsins F and S are the most important processing en-
zymes in antigen-presenting macrophages, their proteolytic
synergism might be addressed in the design of inhibitors as
potential therapeutics for macrophage-related disorders. Much
effort has been made to develop inhibitors for human cathe-
psin S, which hold potential for the treatment of autoimmune
diseases,[8,12] but little is known about inhibition of cathepsin F
by low-molecular weight compounds.
In this study, a library of dipeptide nitriles was applied for
active-site mapping of the non-primed binding region of
human cathepsin F. Stepwise structural modifications were
aimed at developing optimized inhibitors and at elucidating
structure–activity relationships.
The core of the active site of cathepsin F is characterized by
Cys25 and His159, with the nucleophilic thiolate group of the
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1366
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
more advantageous than phenylalanine (2 and 3 versus 1,
Table 1). Bulky capping groups at position P3 as present in
compounds 5 and 6 clearly increased the potency (5 versus 1;
6 versus 3, Table 1). Moreover, introduction of the methionine
side chain caused a small improvement in activity (4 versus 1,
Table 1).
Results and Discussion
As a starting point of this project, we investigated a library of
52 dipeptide nitriles with various aminonitrile moieties at posi-
tion P1, different amino acids at position P2, and various cap-
ping groups at the N terminus. The general inhibitor structure
VI is depicted in Scheme 1, and the individual formulae are
given in the Supporting Information (Table S1). These dipep-
tide nitriles have already been evaluated as inhibitors of cathe-
psins L, S, K, and in some cases cathepsin B,[14,18–20] but have
not been investigated as inhibitors of cathepsin F.
The following syntheses were carried out to provide further
cathepsin F inhibiting dipeptide nitriles in a stepwise optimiza-
tion process. First, to corroborate that leucine is more favora-
ble at the P2 position than cyclohexylalanine (3 versus 2,
Table 1), a new pair of Boc-protected dipeptides, both with
methionine nitrile at the P1 position, were designed. The prep-
aration of these compounds, 7 and 8, is depicted in Scheme 2.
Boc-methionine (13) was treated with ammonia in a mixed--
anhydride procedure, and the resulting carboxamide 14 was
deprotected and subsequently coupled with Boc-protected cy-
clohexylalanine or leucine to obtain dipeptides 16 or 17, re-
spectively. The nitrile group of the desired products 7 and 8
was formed upon treatment with cyanuric chloride.[23] The ki-
netic data of 7 and 8 are listed in Table 2. The corresponding Ki
values for cathepsin F inhibition confirmed cyclohexylalanine
to be a more suitable P2 building block than phenylalanine (7,
Table 2 versus 4, Table 1), but to be less advantageous than
leucine (7 versus 8, Table 2). The two new methionine deriva-
tives were more active than their glycine counterparts (7 and
8, Table 2 versus 2 and 3, Table 1).
Thus, in this study, the library was used to provide initial
structure–activity relationship information for dipeptide nitrile
inhibitors of cathepsin F. Human cathepsin F was assayed with
a fluorogenic peptide substrate, and Ki values were obtained
with the Cheng–Prusoff equation. The observed linearity of the
progress curves in the presence of the library compounds indi-
cated a ‘fast binding’ behavior, reflecting rapid formation of
the covalent and reversible enzyme–inhibitor thioimidate
adduct VII (Scheme 1). However, inhibition constants of the 52
dipeptide nitriles span five orders of magnitude (Table S1 in
the Supporting Information). For example, Boc-glycylglycine ni-
trile (35, Table S1) was a particularly poor inhibitor, with a Ki
value of 1.2 mm, whereas the strongest dipeptide nitrile (6,
Table 1) exhibited a Ki value of 12 nm. The most significant re-
sults of this screening campaign are depicted in Table 1. We
considered the Ki values of the six compounds to deduce the
influence of structural fragments on the inhibitory activity
against human cathepsin F. The data illustrate that cyclohexyla-
lanine or leucine at position P2 resulted in Ki values in the low-
micromolar range. These amino acids appeared somewhat
To obtain new dipeptide inhibitors with the bi- or triaryl cap-
ping groups present in 5 and 6, the required carboxylic acids
21 and 24 were prepared (Scheme 3). Synthesis of building
block 21 involved acylation of an intermediate amidoxime to
20 and a cyclocondensation to produce the internal oxadiazole
Table 1. Inhibition of human cathepsin F by dipeptide nitriles 1–6.
Compd
Ki [mm][a]
Compd
Ki [mm][a]
1
8.48Æ0.35
4
4.36Æ0.19
2
3
4.40Æ0.15
1.50Æ0.04
5
6
0.687Æ0.046
0.0124Æ0.0006
[a] IC50 values were obtained from duplicate measurements in the presence of five different inhibitor concentrations. Progress curves were followed over
20 min and analyzed by linear regression. IC50 values were determined by nonlinear regression using the equation: vs =v0/(1+[I]/IC50), in which vs is the
steady-state rate, v0 is the rate in the absence of inhibitor, and [I] is the inhibitor concentration. Standard error of the mean (SEM) values refer to this non-
linear regression. Ki values ÆSEM were calculated from IC50 values by applying the equation Ki =IC50/(1+[S]/KM), in which [S] is the substrate concentra-
tion.
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1367
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Table 2. Inhibition of human cathepsins by dipeptide nitriles 7–12.
Compd
Ki [mm][a]
cat F
cat B
cat L
cat K
cat S
7
8
9
1.33Æ0.03
16.5Æ0.8
1.88Æ0.13
2.62Æ0.35
0.203Æ0.008
0.0806Æ0.0052
0.106Æ0.009
0.510Æ0.027
0.139Æ0.016
42.6Æ1.6
0.854Æ0.085
0.781Æ0.116
0.0288Æ0.0014
0.154Æ0.011
4.94Æ1.31
10
11
0.00779Æ0.00074
0.0118Æ0.0010
1.79Æ0.14
1.36Æ0.07
1.03Æ0.05
0.563Æ0.067
0.000561Æ0.000048
0.000479Æ0.000035
0.000261Æ0.000019
0.0387Æ0.0032
0.0347Æ0.0016
0.0171Æ0.0021
0.801Æ0.043
0.333Æ0.025
12
0.00728Æ0.00063
[a] See Table 1 footnotes for details.
Scheme 2. Synthesis of 7 and 8. Reagents and conditions: a) 1. N-methylmor-
pholine (NMM), ClCO2iBu, THF, 2. NH3, À258C!RT, overnight, 79%; b) tri-
fluoroacetic acid (TFA), CH2Cl2, 08C, 2 h, quant.; c) Boc-Cha-OH or Boc-Leu-
OH, N,N-diethylisopropylamine (DIPEA), N-[(dimethylamino)-1H-1,2,3-triazolo-
[4,5-b]pyridin-1-ylmethylene]-N-methylmethanaminiumhexa-fluoro-phos-
phate N-oxide (HATU), CH2Cl2, RT, overnight, 99% (16), 56% (17); d) cyanuric
chloride, DMF, RT, 4 h, 62% (7), 28% (8).
Scheme 3. Synthesis of 21[18] and 24. Reagents and conditions: a) 1. oxalyl
chloride, CH2Cl2, DMF, RT, 1 h, 2. tert-butanol, pyridine, RT, 6 h, 77%;
b) 1. NH2OH·HCl, Et3N, EtOH, reflux, 3 h, 2. 2-thenoyl chloride, Et3N, MeCN,
RT, 5 h, 76%; c) 1. AcOH, 808C, overnight, 2. TFA, CH2Cl2, RT, 6 h, 83%;
d) 1. Pd(PPh3)4, 1,2-dimethoxyethane (DME), RT, 10 min, 2. phenylboronic
acid, Na2CO3, 1208C, 4 h, 52%; e) LiOH, THF/H2O (2:1), 08C!RT, 3 h, 99%.
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1368
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Scheme 4. Synthesis of 9 and 10. Reagents and conditions: a) 1. NMM,
ClCO2iBu, THF, 2. H2NCH2CN·0.5H2SO4, 2n NaOH, À258C!RT, overnight,
32% (27), 42% (28); b) 1. methanesulfonic acid, THF, 08C!RT, overnight,
2. N-4-[5-(2-thienyl)-1,2,4-oxadiazol-3-yl]benzoic acid (21) or biphenyl-4-car-
boxylic acid (24), EDC, DMAP, DIPEA, THF, RT, overnight, 22% (9), 30% (10).
Scheme 5. Synthesis of 11. Reagents and conditions: a) biphenyl-4-carboxylic
acid (24), DIPEA, HATU, CH2Cl2, RT, overnight, 73%; b) 1. TFA, CH2Cl2, 08C,
2 h, 2. compound 15, DIPEA, HATU, CH2Cl2, RT, overnight, 81%; c) cyanuric
chloride, DMF, RT, 4 h, 33%.
ring,[18] and 24 was produced through Suzuki coupling.[24] Boc-
cyclohexylalanine (25) and Boc-leucine (26) were reacted with
aminoacetonitrile to form dipeptide nitriles 27 and 28
(Scheme 4), which were then deprotected with TFA and finally
coupled with the aforementioned carboxylic acids 21 or 24. By
combination of the triaryl capping group and the cyclohexyla-
lanine P2 building block, we obtained the dipeptide nitrile 9
and, by combination of the biphenyl group and leucine, the di-
peptide nitrile 10. With 9 and 10 in hand and after evaluating
their cathepsin F inhibitory activities (Table 2), further struc-
ture–activity relationships could be drawn. Importantly, the
corresponding Ki values revealed that the biphenyl group is
more favorable than the triaryl capping group (10, Table 2
versus 6, Table 1). A comparison of the two triaryl-containing
analogues confirmed the leucine–cyclohexylalanine exchange
to be disadvantageous, this time leading to a 10-fold loss in
potency (6, Table 1 versus 9, Table 2). Obviously, when accom-
modating an inhibitor with a more voluminous P3 moiety,
cathepsin F became more susceptible to a change in the P2
substructure. Such a trend was also observed for the biphenyl
pair, in which the expected improvement due to replacement
of phenylalanine at the P2 position with leucine led a 100-fold
lower Ki value (10, Table 2 versus 5, Table 1). Moreover, 9 and
10 (Table 2) are clearly more active than the Boc-capped ana-
logues 2 and 3 (Table 1).
Scheme 6. Synthesis of 12. Reagents and conditions: a) compound 15, DIPEA,
HATU, CH2Cl2, RT, overnight, 50%; b) 1. H2, 10% Pd/C, MeOH, RT, 7 h, 2. bi-
phenyl-4-carboxylic acid (24), DIPEA, HATU, CH2Cl2, RT, overnight, 95%; c) cy-
anuric chloride, DMF, RT, 4 h, 31%.
Next, we intended to examine whether glycine nitrile might
be favorably replaced by methionine nitrile. Leucine tert-butyl
ester was acylated with biphenyl-4-carboxylic acid to obtain 30
(Scheme 5). This amino acid derivative was deprotected and
activated with HATU. The activated intermediate could easily
undergo oxazolone formation because of the nucleophilic po-
tential of the acyl oxygen, leading to racemization. Thus, the
coupling reaction with methionine amide trifluoroacetate (15)
produced the diastereomeric mixture of carboxamide 31,
which was finally dehydrated to the dipeptide nitrile 11. After
verifying the biological activity of 11 (Table 2), the diastereo-
merically pure compound 12 was synthesized (Scheme 6). In
comparison with the aforementioned sequence (introduction
of the capping group, peptide coupling, dehydration), the
order of the two first steps was reversed. Reaction of the car-
bamate-protected leucine 32 with 15 yielded the dipeptide 33
without racemization. Hydrogenolytic removal of the Cbz
group followed by introduction of the biphenyl capping group
and treatment of the resulting 34 with cyanuric chloride af-
forded the desired dipeptide nitrile 12. Whereas the diastereo-
meric mixtures of compounds 31 and 11 (Scheme 5) exhibited
double sets of NMR signals, this was not the case for the coun-
terparts 34 and 12 (Scheme 6), indicating their diastereomeric
purity. Expectedly, dipeptide 11 had lower activity against cath-
epsin F and the other cathepsins listed in Table 2 than the
(S,S)-configured analogue 12.
Dipeptide nitrile 12 showed a Ki value for inhibition of
human cathepsin F of 7.3 nm. Although the introduction of the
methionine side chain did not further lead to a notable im-
provement of cathepsin F inhibitory activity (12 versus 10,
Table 2), it was the most potent cathepsin F inhibitor obtained
in the course of this study.
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1369
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. Shown are putative binding modes of covalent inhibitors of human cathepsin F. A) The predicted binding mode of compound 12 (cyan) within the
active site of cathepsin F. Active site residues are depicted in magenta and the surface is rendered transparent. Binding pockets are annotated according to
the Schechter and Berger nomenclature. B) Overlay of the predicted binding mode of 12 (cyan) and the crystallographic vinyl sulfone inhibitor V (green). The
active site surface is rendered non-transparent. C) Overlay of the predicted binding modes of compounds 1–10 and 12 (all in blue) within the active site of
cathepsin F using a non-transparent surface representation.
Figure 2A shows the putative binding mode of inhibitor 12
in the active site of human cathepsin F. Through formation of
a thioimidate (VII, Scheme 1) the nitrile function of the inhibi-
tor is covalently bound to the side chain of residue Cys25.
Such a covalent docking approach is not appropriate to pro-
vide information about the initial enzyme–inhibitor complex
and the orientation of the nitrile warhead relative to the active
site cysteine nucleophile, but to model the likely binding pose
of the inhibitor within the active site focusing on the comple-
mentary occupancy of subsites with substructures of the inhib-
itor after formation of the covalent enzyme–inhibitor complex.
In the case of inhibitor 12, the thioimidate nitrogen atom is in
potential hydrogen bonding distance to the backbone NH
group of residue Cys25, while the methionine side chain occu-
pies the S1 pocket. The leucine residue at the P2 position of
12 is directed into the S2 pocket and might form van der
Waals contacts with the side chains of residues Leu67, Ala133,
and Ala160. The shallow S3 pocket of cathepsin F hosts the
large biphenyl moiety of the inhibitor. The side chains of resi-
dues Leu67 and Lys61, which determine the width of the S3
pocket,[22] might form lipophilic interactions with the biphenyl
moiety. In addition, the backbone of the dipeptide inhibitor is
in a favorable position to form several hydrogen bonding in-
teractions to peptide backbone atoms of residues Asp158,
Gly65, and Gly66. An overlay of the predicted binding mode of
compound 12 and the crystallographic vinyl sulfone inhibitor
V (Figure 1) is shown in Figure 2B. The backbone peptide
structures of both inhibitor compounds display a similar orien-
tation. The propyl substituent at position P1 and the phenyl
moiety at position P2 of the crystallographic inhibitor V
occupy the S1 and S2 pockets of cathepsin F, respectively, in
a manner similar to the corresponding side chains in com-
pound 12. In contrast, the substituents at the P3 position
show different orientations. While the biphenyl moiety of in-
hibitor 12 fits very closely into the S3 pocket, the P3 residue
of the crystallographic inhibitor is directed out of the pocket
and shows a high degree of solvent exposure. In Figure 2C, an
overlay of the putative binding modes of inhibitors 1–10 and
12 within the active site of cathepsin F is presented. All com-
pounds adopt a similar orientation such that the S1, S2, and S3
pockets of the enzyme are occupied by the corresponding P1,
P2, and P3 substituents of the inhibitor compounds. The me-
thionine side chain at the P1 position is well accommodated in
the S1 pocket. This is reflected by the kinetic data; compounds
with glycine at this position, which leave the S1 pocket empty,
are less active than the methionine analogues. However, these
activity differences were not striking. According to the predict-
ed docking poses, the smaller leucine is preferred over phenyl-
alanine and cyclohexylalanine in the narrow S2 pocket, the size
of which is determined by the position of the side chains of
residues Leu67, Ile157, and Met205.[22] In contrast to the well-
defined S2 pocket of cathepsin F, its S3 binding site is spatially
less restricted. Accordingly, bulky P3 residues such as the bi-
phenyl substituent of 12 fit nicely into the wide open S3
pocket, which accounted for a further increased binding affini-
ty. Thus, for the compounds presented herein, plausible bind-
ing modes were generated that help to rationalize structure–
activity relationships. The orientations of the peptide back-
bones of the inhibitors are very similar to the crystallographic
vinyl sulfone inhibitor V, and side chains are directed into the
S1, S2, and S3 pockets in an equivalent manner.
Inhibitors 7–12 were also evaluated at the human cathe-
psins B, L, K and S (Table 2). None of these dipeptide nitriles
was selective for cathepsin F over cathepsin K. In fact, Ki values
for cathepsin K inhibition by compounds 8 and 10–12 were
more than 10-fold lower. These results show that the intended
optimization of cathepsin F inhibitors was accompanied by
a parallel improvement of their affinity for cathepsin K and re-
flect a high similarity of the addressed binding region in both
cysteine cathepsins. However, cathepsin K is mainly expressed
in osteoclasts and hence has a different tissue distribution
than cathepsin F.[4] Compounds 7–12 exhibited cathepsin L in-
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1370
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
hibitory activity with Ki values that differ by only a factor of 3.
By contrast, a clear successive improvement of the inhibitory
activity toward cathepsins B and S was observed in ascending
order from 7 to 12. Selectivity for cathepsin F over S is of par-
ticular interest, because these two proteases share the catalytic
activity to cleave the invariant chain of the MHC II complex,
a process which enables antigen presenting cells to display an
antigen fragment on their membrane, leading to recognition
by Tcells and the subsequent events in the immune response.
Whereas compounds 7 and 8 with the Boc capping group
were more active at cathepsin S, nitrile inhibitors 10 and 12 al-
ready exhibited a preference for cathepsin F over S. Further in-
vestigations are needed for the development of cathepsin F-se-
lective nitrile-based inhibitors. Such tool compounds are ex-
pected to be valuable for investigating the overlapping roles
of cathepsins F and S in macrophages.
these compounds only lead to moderate changes in the bio-
logical activity. For example, the phenylalanine residue at the
P2 position of compound 1 is replaced by cyclohexylalanine in
compound 2, which leads to a minor change in potency
against cathepsin F of about factor 2. Furthermore, the com-
pounds are arranged in clusters that are defined by substitu-
ents at the P2 and P3 positions. Compounds in the area of in-
hibitors 1 and 4 all contain aromatic side chains at the P2 posi-
tion and a tert-butyl residue of the Boc capping group at posi-
tion P3. In contrast, the region defined by inhibitors 2, 3, 7,
and 8 represents compounds with aliphatic and non-aromatic
residues at the P2 position and again a P3 tert-butyl group.
The set of compounds on the lower left side of the landscape
with low potency values displays a trace of structure–activity
relationship progression to a peak region with the highly
active Cbz-leucylglycine nitrile (40, Table S1 in the Supporting
Information). All compounds involved in this progression have
the benzyl residue of the N-terminal Cbz capping group. An
area of discontinuous structure–activity relationships is present
in the upper part of the landscape. This can be seen, for exam-
ple, if the N-terminal tert-butyloxy group of 3 is replaced by
a biphenyl moiety in 10, improving the potency by a factor of
~200. Compounds in the upper part of the landscape contain
large aromatic P3 substituents such as a biphenyl capping
group and belong to the most potent inhibitors within this
series including 6, 9, 10, and 12. In this area of the activity
landscape, most of the compounds contain leucine at the P2
position and have been synthesized in the course of this study,
attempting to structurally optimize cathepsin F inhibitors.
Three dimensional (3D) activity landscape representations
are suitable tools to illustrate structure–activity relationships of
large compound data sets. The activity cliff concept has at-
tracted much interest in medicinal chemistry because it can
reveal whether small chemical changes lead to significant dif-
ferences in potency.[25] However, activity landscape modeling
and cliff analysis have thus far been rarely applied to peptidic
drugs. In the course of this study, a small library of enzyme in-
hibitors of the same chemotype and hence with a comparably
high similarity was analyzed. Figure 3 shows the 3D activity
landscape representation for the 57 cathepsin F inhibitors
(compound 11 was not included).
The front part of the landscape represents an area of contin-
uous structure–activity relationships. Compounds in this area
have low to medium potency against cathepsin F, which is in-
dicated by the green color, and structural changes among
Conclusions
Despite the involvement of human cathepsin F in important
physiological and pathophysiological processes, this enzyme
has thus far not been a focal point of drug development ef-
forts. This prompted us to perform an active site mapping by
employing a library of peptide nitriles known to interact with
cysteine cathepsins in a covalent and reversible manner. A
stepwise structural optimization afforded two highly potent
cathepsin F inhibitors, 10 and 12, with Ki values <10 nm. The
structure–activity relationships obtained for 57 dipeptide ni-
triles have been illustrated by 3D activity landscape representa-
tions. The inhibitors’ affinities clearly reflect the characteristics
of the non-primed binding region of human cathepsin F. Our
data and, in particular, the structures of inhibitors 10 and 12
might represent a starting point for the future design of thera-
peutics against autoimmune diseases. Moreover, these pepti-
domimetic inhibitors are expected to be valuable tools for in-
vestigations on the role of cathepsin F in macrophage-related
disorders, atherogenesis, or lysosomal storage defects.
Experimental Section
Figure 3. 3D activity landscape representation for the cathepsin F inhibitor
series. Euclidean fingerprint distances were calculated for the 57 inhibitors
using the ECFP2 fingerprint. The surface is colored according to interpolated
surface elevation using a continuous spectrum from green (lowest potency)
to red (highest potency). Black spheres represent compound data points
and are labeled with compound numbers for inhibitors 1–10, 12, and 40.
Amino acid derivatives 13, 25, 26, 29, 32 as well as compounds 18
and 22 were obtained from Bachem (Bubendorf, Switzerland),
Acros (Geel, Belgium) and Aldrich (Steinheim, Germany). Solvents
were used without additional purification. Preparative column
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1371
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
chromatography was performed on silica gel 60, 0.060–0.200 mm
(Acros Organics). Mass spectra were recorded on an API 2000 mass
spectrometer (electrospray ion source, Applied Biosystems, Darm-
stadt, Germany) coupled with an Agilent 1100 HPLC system using
a Phenomenex Luna HPLC C18 column (50”2.00 mm, particle size:
3 mm). The purity of the tested compounds was determined by
HPLC-DAD obtained on an LC–MS instrument (HPLC Agilent 1100)
using the procedure as follows: dissolving the compounds at a con-
centration of 1.0 mgmLÀ1 in MeOH and, if necessary, sonication to
complete dissolution. Then, 10 mL of the substance solution was in-
jected into the Phenomenex Luna C18 column, and elution was car-
ried out with a gradient of H2O/MeOH containing 2 mm ammoni-
um acetate from 90:10 up to 0:100 for 30 min at a flow rate of
250 mLminÀ1, starting the gradient after 10 min. UV absorption was
detected from l 220 to 400 nm using a diode array detector. All
1H), 4.82–4.87 (m, 1H), 6.96 (d, J=8.2 Hz, 1H), 8.62 ppm (d, J=
7.6 Hz, 1H); 13C NMR (125 MHz, [D6]DMSO): d=14.6, 21.7, 23.0,
24.4, 28.3, 28.8, 31.5, 38.9, 52.8, 78.2, 119.3, 155.5, 173.0 ppm, one
carbon signal is obscured by the DMSO signal; LC/ESI-MS (m/z):
negative mode 342 [MÀH]À, positive mode 344 [M+H]+; purity,
96%.
N-{4-[5-(2-Thienyl)-1,2,4-oxadiazol-3-yl]benzoyl}cyclohexylalanyl-
glycine nitrile (9): Compound 27 (0.19 g, 0.60 mmol) was dissolved
in dry THF (5 mL). Under ice cooling, methanesulfonic acid (0.35 g,
3.60 mmol) was added. The resulting mixture was stirred at RT
overnight. The solvent was evaporated under reduced pressure
and the crude product, cyclohexylalanylglycine nitrile methanesul-
fonate, was used without further purification. 4-[5-(2-Thienyl)-1,2,4-
oxadiazol-3-yl]benzoic acid (21, 0.16 g, 0.60 mmol) was dissolved in
dry THF (30 mL). EDC (93 mg, 0.60 mmol) and DMAP (4 mg,
0.03 mmol) were added and stirred for 10 min at RT. DIPEA (0.39 g,
0.53 mL, 3.00 mmol) and the cyclohexylalanylglycine nitrile metha-
nesulfonate were added and the solution was stirred overnight at
RT. After evaporation of the solvent, the residue was extracted with
EtOAc (3”100 mL). The combined organic layers were washed
with 10% KHSO4 (60 mL), H2O (60 mL), sat. NaHCO3 (2”60 mL),
H2O (60 mL) and brine (60 mL). The solvent was dried (Na2SO4) and
evaporated. The crude product was purified by column chroma-
tography on silica gel using PE/EtOAc (1:2) as eluent to obtain 9 as
a white solid (61 mg, 22% over two steps); mp: 168–1708C;
1H NMR (500 MHz, CDCl3): d=0.90–1.04 (m, 2H), 1.22–1.25 (m, 3H),
1.36–1.44 (m, 1H), 1.62–1.89 (m, 7H), 4.08–4.12 (m, 1H), 4.21 (dd,
J=17.4 Hz, J=5.7 Hz, 1H), 4.75–4.79 (m, 1H), 6.86 (d, J=7.9 Hz,
1H), 7.21 (dd, J=5.0 Hz, J=3.8 Hz, 1H), 7.56 (t, J=5.5 Hz, 1H), 7.65
(dd, J=5.1 Hz, J=1.3 Hz, 1H), 7.87 (d, J=8.5 Hz, 2H), 7.94 (dd, J=
3.6 Hz, J=1.1 Hz, 1H), 8.19 ppm (d, J=8.6 Hz, 2H); 13C NMR
(125 MHz, CDCl3): d=26.0, 26.1, 26.3, 27.6, 32.8, 33.5, 34.2, 39.0,
51.4, 115.8, 125.6, 127.7, 127.9, 128.6, 130.3, 132.1, 132.3, 135.4,
167.4, 167.9, 171.7, 172.4 ppm; LC/ESI-MS (m/z): negative mode
462 [MÀH]À, positive mode 464 [M+H]+; purity, 99%.
(S,S)-N-(4-Phenylbenzoyl)leucylglycine nitrile (10): Compound 28
(0.16 g, 0.60 mmol) was dissolved in dry THF (5 mL). Under ice
cooling, methanesulfonic acid (0.35 g, 3.60 mmol) was added. The
resulting mixture was stirred at RT overnight. The volume of the
solvent was reduced, Et2O was added, the crude product, leucinyl-
glycine nitrile methanesulfonate, was filtered off and used without
further purification. Biphenyl-4-carboxylic acid (24, 56 mg,
0.30 mmol) was dissolved in dry THF (10 mL). EDC (47 mg,
0.30 mmol) and DMAP (2 mg, 0.02 mmol) were added and stirred
for 10 min at RT. DIPEA (0.47 g, 0.63 mL, 3.60 mmol) and the leuci-
nylglycine nitrile methanesulfonate were added and the solution
was stirred overnight at RT. After evaporation of the solvent, the
residue was extracted with EtOAc (3”50 mL). The combined or-
ganic layers were washed with 10% KHSO4 (30 mL), H2O (30 mL),
sat. NaHCO3 (2”30 mL), H2O (30 mL) and brine (30 mL). The sol-
vent was dried (Na2SO4) and evaporated. The crude product was
purified by column chromatography on silica gel using PE/EtOAc
(1:2) as eluent to obtain 10 as a white solid (31 mg, 30%); mp:
173–1758C; 1H NMR (500 MHz, CDCl3): d=0.96 (d, J=6.3 Hz, 3H),
0.99 (d, J=6.3 Hz, 3H), 1.71–1.77 (m, 2H), 1.82–1.88 (m, 1H), 4.08
(dd, J=17.5 Hz, J=5.8 Hz, 1H), 4.19 (dd, J=17.3 Hz, J=5.7 Hz, 1H),
4.75 (q, J=7.5 Hz, 1H), 6.72 (d, J=7.9 Hz, 1H), 7.36–7.39 (m, 1H),
7.42–7.45 (m, 2H), 7.56–7.58 (m, 2H), 7.59–7.60 (m, 1H), 7.64 (d,
J=8.6 Hz, 2H), 7.83 ppm (d, J=8.5 Hz, 2H); 13C NMR (125 MHz,
CDCl3): d=22.1, 22.9, 24.9, 27.5, 40.3, 51.8, 115.8, 127.2, 127.4,
127.6, 128.2, 129.0, 131.7, 139.7, 145.1, 167.9, 172.4 ppm; LC/ESI-MS
(m/z): negative mode 348 [MÀH]À, positive mode 350 [M+H]+;
purity, 98%.
1
tested compounds possessed a purity of no less than 95%. H and
13C NMR spectra were recorded on a Bruker Avance 500 MHz NMR
spectrometer with [D6]DMSO as solvent. NMR spectra were record-
ed at 308C. Chemical shifts are given in parts per million (ppm) rel-
ative to the remaining protons of the deuterated solvent used as
internal standard. Coupling constants J are given in Hertz, and spin
multiplicities are given as s (singlet), d (doublet), t (triplet), q (quar-
tet), m (multiplet), and br (broad). Melting points were determined
on a Büchi 510 oil bath apparatus and were uncorrected. Com-
pounds 19–21 were prepared as described.[18]
Chemistry
(S,S)-N-(tert-Butoxycarbonyl)cyclohexylalanylmethionine nitrile
(7): Compound 16 (0.40 g, 1.00 mmol) was dissolved in dry DMF
(20 mL). Cyanuric chloride (0.18 g, 1.00 mmol) was added, and the
mixture was stirred for 4 h at RT. The solvent was evaporated
under reduced pressure. The resulting solid was treated with 10%
NaHCO3 (20 mL) and stirred for 10 min at RT. The aqueous suspen-
sion was extracted with EtOAc (3”50 mL). The combined organic
layers were washed with 10% KHSO4 (30 mL), H2O (30 mL), sat.
NaHCO3 (30 mL), H2O (30 mL), and brine (30 mL). The solvent was
dried (Na2SO4) and evaporated. The crude product was purified by
column chromatography on silica gel using EtOAc/petroleum ether
(PE) (3:7) as eluent to obtain 7 as a white solid (0.24 g, 62%); mp:
74–758C; 1H NMR (500 MHz, [D6]DMSO): d=0.80–0.91 (m, 2H),
1.09–1.18 (m, 3H), 1.23–1.26 (m, 1H), 1.36 (s, 9H), 1.38–1.42 (m,
2H), 1.59–1.69 (m, 5H), 1.99–2.06 (m, 5H), 2.51–2.56 (m, 2H), 3.92–
3.95 (m, 1H), 4.83–4.86 (m, 1H), 6.95 (d, J=6.6 Hz, 1H), 8.62 ppm
(d, J=6.6 Hz, 1H); 13C NMR (125 MHz, [D6]DMSO): d=14.7, 25.9,
26.0, 26.2, 28.3, 28.9, 31.5, 33.1, 33.2, 33.8, 38.9, 40.3, 52.2, 78.3,
119.4, 155.6, 173.1 ppm; LC/ESI-MS (m/z): negative mode 382
[MÀH]À, positive mode 384 [M+H]+; purity, 100%.
(S,S)-N-(tert-Butoxycarbonyl)leucylmethionine nitrile (8): Com-
pound 17 (0.90 g, 2.50 mmol) was dissolved in dry DMF (20 mL).
Cyanuric chloride (0.46 g, 2.50 mmol) was added, and the mixture
was stirred for 4 h at RT. The solvent was evaporated under re-
duced pressure. The resulting solid was treated with 10% NaHCO3
(20 mL) and stirred for 10 min at RT. The aqueous suspension was
extracted with EtOAc (3”50 mL). The combined organic layers
were washed with 10% KHSO4 (30 mL), H2O (30 mL), sat. NaHCO3
(30 mL), H2O (30 mL) and brine (30 mL). The solvent was dried
(Na2SO4) and evaporated. The crude product was purified by
column chromatography on silica gel using EtOAc/PE (1:2) as
eluent to obtain 8 as a white solid (0.24 g, 28%); mp: 68–698C;
1H NMR (500 MHz, [D6]DMSO): d=0.85 (d, 3H, J=6.6 Hz), 0.87 (d,
3H, J=6.7 Hz), 1.36 (s, 9H), 1.40–1.49 (m, 2H), 1.55–1.60 (m, 1H),
1.99–2.03 (m, 2H), 2.04 (s, 3H), 2.51–2.57 (m, 2H), 3.89–3.93 (m,
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1372
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
N-(4-Phenylbenzoyl)-(R,S)-leucyl-(S)-methionine nitrile (11): Com-
pound 31 (0.22 g, 0.50 mmol) was dissolved in dry DMF (10 mL).
Cyanuric chloride (92 mg, 0.50 mmol) was added and the mixture
was stirred for 4 h at RT. The solvent was evaporated under re-
duced pressure. The resulting solid was treated with 10% NaHCO3
(20 mL) and stirred for 10 min at RT. The aqueous suspension was
extracted with EtOAc (3”50 mL). The combined organic layers
were washed with 10% KHSO4 (30 mL), H2O (30 mL), sat. NaHCO3
(30 mL), H2O (30 mL) and brine (30 mL). The solvent was dried
(Na2SO4) and evaporated. The crude product was purified by
column chromatography on silica gel using CH2Cl2/MeOH (9:1) as
eluent to obtain 11 as a white solid (70 mg, 33%); mp: 134–1358C;
1H NMR (500 MHz, [D6]DMSO, mixture of diastereomers, ratio ap-
proximately 3:4 according to 1H NMR, w=weak, refers to minor
diastereomer, i=intense, refers to major diastereomer): d=0.88–
0.90 (w+i overlapping; m, 3H (w), 3H (i)), 0.92–0.94 (w+i overlap-
ping; m, 3H (w), 3H (i)), 1.51–1.58 (w+i overlapping; m, 1H (w),
1H (i)), 1.65–1.77 (w+i overlapping, m, 2H (w), 2H (i)), 2.00–2.10
(w+i overlapping, m, 5H (w), 5H (i)), 2.51–2.59 (w+i overlapping,
m, 2H (w), 2H (i)), 4.48–4.53 (w+i overlapping, m, 1H (w), 1H (i)),
4.83–4.90 (w+i overlapping, m, 1H (w), 1H (i)), 7.39–7.42 (w+
i overlapping, m, 1H (w), 1H (i)), 7.47–7.50 (w+i overlapping, m,
2H (w), 2H (i)), 7.72–7.73 (w+i overlapping, m, 2H (w), 2H (i)),
7.76–7.78 (w+i overlapping, m, 2H (w), 2H (i)), 7.99–8.02 (w+
i overlapping, m, 2H (w), 2H (i)), 8.55–8.57 (w+i overlapping, m,
1H (w), 1H (i)), 8.77 ppm (d, J=7.9 Hz, 1H (i)), 8.82 (d, J=7.9 Hz,
1H (w)); 13C NMR (125 MHz, [D6]DMSO, w=weak, refers to minor
diastereomer, i=intense, refers to major diastereomer): d=14.6
(w), 14.7 (i), 21.5 (i), 21.5 (w), 23.0 (w), 23.1 (i), 24.6 (i), 24.6 (w), 28.9
(w+i), 31.4 (w), 31.4 (i), 39.0 (w), 39.1 (i), 40.3 (w+i), 51.9 (i), 52.0
(w), 119.3 (w+i), 126.5 (w+i), 127.0 (w+i), 128.2 (w+i), 128.4 (w+
i), 129.2 (w+i), 132.9 (w+i), 139.3 (w+i), 143.0 (w+i), 166.3 (w+i),
172.7 (i), 172.7 ppm (w); negative mode 422 [MÀH]À, positive
mode 424 [M+H]+; purity, 100%.
tion of N-methylmorpholine hydrochloride occurred. The mixture
was allowed to warm to RT within 30 min and was stirred over-
night. After evaporation of the solvent the resulting aqueous resi-
due was extracted with EtOAc (3”50 mL). The combined organic
layers were washed with 10% KHSO4 (30 mL), H2O (30 mL), sat.
NaHCO3 (2”30 mL), H2O (30 mL) and brine (50 mL). The solvent
was dried (Na2SO4) and evaporated. The crude product was recrys-
tallized from EtOAc to obtain 14 as a white solid (1.56 g, 79%);
mp: 116–1188C, lit.[19] mp: 118–1198C; 1H NMR (500 MHz,
[D6]DMSO): d=1.37 (s, 9H), 1.70–1.77 (m, 1H), 1.81–1.87 (m, 1H),
2.02 (s, 3H), 2.40–2.45 (m, 2H), 3.90–3.95 (m, 1H), 6.80 (d, J=
8.2 Hz, 1H), 6.94 (s, 1H), 7.20 ppm (s, 1H); 13C NMR (125 MHz,
[D6]DMSO): d=14.8, 28.3, 30.0, 31.9, 53.5, 78.1, 155.5, 173.9 ppm.
LC/ESI-MS (m/z): negative mode 247 [MÀH]À, positive mode 249
[M+H]+; purity, 100%.
(S)-Methionine amide trifluoroacetate (15): Compound 14
(0.25 g, 1.00 mmol) was dissolved in a mixture of dry CH2Cl2 and
TFA (10 mL, 1:1) and stirred for 2 h at 08C. The solvent was evapo-
rated and the oily crude product was used without further purifica-
tion.
(S,S)-N-(tert-Butoxycarbonyl)cyclohexylalanylmethionine amide
(16): Boc-Cha-OH (0.27 g, 1.00 mmol) was dissolved in dry CH2Cl2
(10 mL). Compound 15 (0.26 g, 1.00 mmol), DIPEA (0.65 g, 0.88 mL,
5.00 mmol) and HATU (0.38 g, 1.00 mmol) were added. The solu-
tion was stirred overnight at RT. After evaporation of the solvent,
the resulting aqueous residue was extracted with EtOAc (3”
50 mL). The combined organic layers were washed with 10%
KHSO4 (30 mL), H2O (30 mL), sat. NaHCO3 (30 mL), H2O (30 mL) and
brine (50 mL). The solvent was dried (Na2SO4) and evaporated. The
crude product was recrystallized from EtOAc to obtain 16 as
a white solid (0.40 g, 99%); mp: 166–1678C; 1H NMR (500 MHz,
[D6]DMSO): d=0.79–0.90 (m, 2H), 1.08–1.18 (m, 3H), 1.24–1.29 (m,
1H), 1.37 (s, 9H), 1.40–1.44 (m, 2H), 1.59–1.70 (m, 5H), 1.73–1.80
(m, 1H), 1.87–1.95 (m, 1H), 2.01 (s, 3H), 2.35–2.43 (m, 2H), 3.91–
3.96 (m, 1H), 4.25–4.29 (m, 1H), 6.94 (d, J=7.9 Hz, 1H), 7.06 (s,
1H), 7.25 (s, 1H), 7.72 ppm (d, J=7.9 Hz, 1H); 13C NMR (125 MHz,
[D6]DMSO): d=14.8, 25.8, 26.0, 26.2, 28.3, 29.6, 32.1, 32.2, 33.3,
33.7, 39.2, 51.6, 52.6, 78.3, 155.6, 172.5, 173.0 ppm; LC/ESI-MS
(m/z): negative mode 400 [MÀH]À, positive mode 402 [M+H]+;
purity, 100%.
(S,S)-N-(4-Phenylbenzoyl)leucylmethionine nitrile (12): Com-
pound 34 (0.13 g, 0.30 mmol) was dissolved in dry DMF (10 mL).
Cyanuric chloride (55 mg, 0.30 mmol) was added and the mixture
was stirred for 4 h at RT. The solvent was evaporated under re-
duced pressure. The resulting solid was treated with 10% NaHCO3
(20 mL) and stirred for 10 min at RT. The aqueous suspension was
extracted with EtOAc (3”50 mL). The combined organic layers
were washed with 10% KHSO4 (30 mL), H2O (30 mL), sat. NaHCO3
(30 mL), H2O (30 mL) and brine (30 mL). The solvent was dried
(Na2SO4) and evaporated. The crude product was purified by
column chromatography on silica gel using CH2Cl2/MeOH (50:1) as
eluent to obtain 12 as a white solid (39 mg, 31%); mp: 152–
1538C; 1H NMR (500 MHz, [D6]DMSO): d=0.89 (d, J=6.3 Hz, 3H),
0.93 (d, J=6.7 Hz, 3H), 1.53–1.58 (m, 1H), 1.65–1.77 (m, 2H), 2.03–
2.10 (m, 5H), 2.51–2.59 (m, 2H), 4.48–4.52 (m, 1H), 4.86–4.90 (m,
1H), 7.38–7.42 (m, 1H), 7.47–7.50 (m, 2H), 7.71–7.73 (m, 2H), 7.75–
7.78 (m, 2H), 7.98–8.01 (m, 2H), 8.57 (d, J=7.9 Hz, 1H), 8.77 ppm
(d, J=7.6 Hz, 1H); 13C NMR (125 MHz, [D6]DMSO): d=14.7, 21.5,
23.1, 24.6, 29.0, 31.4, 39.1, 51.9, 119.3, 126.5, 127.0, 128.2, 128.4,
129.2, 132.9, 139.3, 143.1, 166.3, 172.7 ppm, one carbon signal is
obscured by the DMSO signal; negative mode 422 [MÀH]À, posi-
tive mode 424 [M+H]+; purity, 97%.
(S,S)-N-(tert-Butoxycarbonyl)leucylmethionine amide (17): Boc-
Leu-OH (1.12 g, 4.50 mmol) was dissolved in dry THF (20 mL) and
cooled at À258C. To the stirred solution, NMM (046 g, 4.50 mmol)
and isobutyl chloroformate (0.61 g, 4.50 mmol) were added con-
secutively. Compound 15 (1.18 g, 4.50 mmol) was dissolved in 2n
NaOH (2.25 mL, 4.50 mmol) and given to the reaction mixture
when the precipitation of N-methylmorpholine hydrochloride oc-
curred. The mixture was allowed to warm to RT within 30 min and
stirred overnight at RT. After evaporation of the solvent, the result-
ing aqueous residue was extracted with EtOAc (3”50 mL). The
combined organic layers were washed with 10% KHSO4 (30 mL),
H2O (30 mL), sat. NaHCO3 (30 mL), H2O (30 mL) and brine (50 mL).
The solvent was dried (Na2SO4) and evaporated. The crude product
was recrystallized from EtOAc to obtain 17 as a white solid (0.90 g,
56%); mp: 154–1558C; 1H NMR (500 MHz, [D6]DMSO): d=0.84 (d,
J=6.6 Hz, 3H), 0.86 (d, J=6.7 Hz, 3H), 1.37 (s, 9H), 1.41 (t, J=
7.4 Hz, 2H), 1.56–1.61 (m, 1H), 1.73–1.80 (m, 1H), 1.88–1.95 (m,
1H), 2.02 (s, 3H), 2.34–2.43 (m, 2H), 3.88–3.93 (q, J=6.6 Hz, 1H),
4.25–4.29 (m, 1H), 6.96 (d, J=8.2 Hz, 1H), 7.05 (s, 1H), 7.26 (s, 1H),
7.72 ppm (d, J=8.2 Hz, 1H); 13C NMR (125 MHz, [D6]DMSO): d=
14.8, 21.7, 23.1, 24.4, 28.3, 29.6, 32.2, 40.5, 51.6, 53.2, 78.3, 155.6,
(S)-N-(tert-Butoxycarbonyl)methionine amide (14): Boc-Met-OH
(13, 1.99 g, 8.00 mmol) was dissolved in dry THF (20 mL) and
cooled at À258C. To the stirred solution, NMM (0.81 g, 8.00 mmol)
and isobutyl chloroformate (1.09 g, 8.00 mmol) were added con-
secutively. Concentrated aqueous ammonia solution (2.99 mL,
40.0 mmol) was given to the reaction mixture when the precipita-
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1373
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
172.5, 173.0 ppm; LC/ESI-MS (m/z): negative mode 360 [MÀH]À,
positive mode 362 [M+H]+; purity, 100%.
ture. It was allowed to warm to RT within 30 min and stirred over-
night at RT. After evaporation of the solvent, the residue was ex-
tracted with EtOAc (3”50 mL). The combined organic layers were
washed with 10% KHSO4 (30 mL), H2O (30 mL), sat. NaHCO3 (2”
30 mL), H2O (30 mL) and brine (30 mL). The solvent was dried
(Na2SO4) and evaporated. The crude product was recrystallized
from EtOAc to obtain 28 as a white solid (0.45 g, 42%); mp: 113–
114 8C, lit.[20] mp: 114–1168C; 1H NMR (500 MHz, [D6]DMSO): d=
0.84 (d, J=6.7 Hz, 3H), 0.86 (d, J=6.7 Hz, 3H), 1.37 (s, 9H), 1.40–
1.47 (m, 2H), 1.57 (sept, J=6.6 Hz, 1H), 3.93–3.97 (m, 1H), 4.09 (dd,
J=5.7 Hz, J=1.0 Hz, 2H), 6.95 (d, J=7.9 Hz, 1H), 8.51 ppm (t, J=
5.5 Hz, 1H); 13C NMR (125 MHz, [D6]DMSO): d=21.5, 23.0, 24.4,
27.2, 28.3, 40.6, 52.6, 78.2, 117.7, 155.5, 173.4 ppm; LC/ESI-MS
(m/z): negative mode 268 [MÀH]À, positive mode 270 [M+H]+;
purity, 100%.
Methyl biphenyl-4-carboxylate (23): To a suspension of Pd(PPh3)4
(0.11 g, 0.09 mmol) in dry DME (20 mL) in an autoclave, methyl-4-
bromobenzoate (22, 0.65 g, 3.00 mmol) was added. The mixture
was stirred for 10 min at RT. Phenylboronic acid (0.55 g, 4.5 mmol)
and aqueous Na2CO3 (0.67 g, 6.3 mmol, in 5 mL H2O) were added.
The mixture was stirred at 1208C for 4 h. The solvent was evapo-
rated under reduced pressure and the residue was suspended in
H2O (50 mL). The aqueous suspension was extracted with EtOAc
(3”30 mL). The combined organic layers were washed with 10%
KHSO4 (2”30 mL), H2O (30 mL), sat. NaHCO3 (2”30 mL), H2O
(30 mL) and brine (30 mL). The solvent was dried (Na2SO4) and
evaporated. The crude product was recrystallized from MeOH to
obtain 23 as a white solid (0.33 g, 52%); mp: 115–1168C, lit.[24] mp:
114–1158C; 1H NMR (500 MHz, [D6]DMSO): d=3.87 (s, 3H), 7.40–
7.44 (m, 1H), 7.48–7.52 (m, 2H), 7.72–7.74 (m, 2H), 7.80–7.83 (m,
2H), 8.02–8.04 ppm (m, 2H); 13C NMR (125 MHz, [D6]DMSO): d=
52.3, 127.1, 127.1, 128.5, 128.6, 129.2, 129.9, 139.0, 144.8, 166.2 ppm;
LC/ESI-MS (m/z): positive mode 213 [M+H]+; purity, 99%.
tert-Butyl (S)-N-(4-phenylbenzoyl)leucine (30): Biphenyl-4-carbox-
ylic acid (24, 0.20 g, 1.00 mmol) was dissolved in dry CH2Cl2
(10 mL). H-Leu-OtBu hydrochloride (29, 0.22 g, 1.00 mmol), DIPEA
(0.35 mL, 2.00 mmol) and HATU (0.38 g, 1.00 mmol) were added.
The solution was stirred overnight at RT. After evaporation of the
solvent the resulting aqueous residue was extracted with EtOAc
(3”50 mL). The combined organic layers were washed with 10%
KHSO4 (30 mL), H2O (30 mL), sat. NaHCO3 (30 mL), H2O (30 mL) and
brine (50 mL). The solvent was dried (Na2SO4) and evaporated. The
crude product was recrystallized from EtOAc to obtain 30 as
a white solid (0.27 g, 73%); mp: 105–1068C; 1H NMR (500 MHz,
[D6]DMSO): d=0.88 (d, J=6.6 Hz, 3H), 0.93 (d, J=6.3 Hz, 3H), 1.41
(s, 9H), 1.53–1.59 (m, 1H), 1.67–1.78 (m, 2H), 4.37–4.42 (m, 1H),
7.39–7.42 (m, 1H), 7.47–7.51 (m, 2H), 7.71–7.73 (m, 2H), 7.76–7.78
(m, 2H), 7.96–7.98 (m, 2H), 8.59 ppm (d, J=7.9 Hz, 1H); 13C NMR
(125 MHz, [D6]DMSO): d=21.4, 23.0, 24.7, 27.8, 38.4, 51.8, 80.5,
126.6, 127.0, 128.2, 128.3, 129.1, 133.0, 139.3, 143.0, 166.3,
172.0 ppm; negative mode 366 [MÀH]À, positive mode 368 [M+
H]+; purity, 88%.
Biphenyl-4-carboxylic acid (24): Compound 23 (0.32 g, 1.50 mmol)
was dissolved in a mixture of THF/H2O 2:1 (30 mL). LiOH (0.16 g,
3.75 mmol) was added under ice cooling and stirred for 3 h. The
solvent was evaporated under reduced pressure. The mixture was
adjusted with concentrated HCl to pH 1–2. The aqueous solution
was extracted with EtOAc (3”30 mL) and washed with brine
(30 mL). The crude product was recrystallized from MeOH to
obtain 24 as a white solid (0.30 g, 99%); mp: 227–2288C, lit.[26] mp:
224–2258C; 1H NMR (500 MHz, [D6]DMSO): d=7.40–7.43 (m, 1H),
7.47–7.51 (m, 2H), 7.71–7.73 (m, 2H), 7.77–7.80 (m, 2H), 8.00–8.03
(m, 2H), 12.92 ppm (bs, 1H); 13C NMR (125 MHz, [D6]DMSO): d=
126.9, 127.1, 128.4, 129.2, 129.8, 130.1, 139.2, 144.4, 167.3 ppm; LC/
ESI-MS (m/z): positive mode 199 [M+H]+; purity, 98%.
N-(tert-Butyloxycarbonyl)-b-cyclohexylalanylglycine nitrile (27):
Boc-Cha-OH (1.09 g, 4.00 mmol) was dissolved in dry THF (10 mL)
and cooled at À258C. To the stirred solution, NMM (0.40 g,
4.00 mmol) and isobutyl chloroformate (0.55 g, 4.00 mmol) were
added consecutively. Aminoacetonitrile monosulfate (0.62 g,
4.00 mmol) was dissolved in 2n NaOH (2.00 mL, 4.00 mmol) and
given to the reaction mixture. It was allowed to warm to RT within
30 min and stirred overnight at RT. After evaporation of the sol-
vent, the residue was extracted with EtOAc (3”50 mL). The com-
bined organic layers were washed with 10% KHSO4 (30 mL), H2O
(30 mL), sat. NaHCO3 (2”30 mL), H2O (30 mL) and brine (30 mL).
The solvent was dried (Na2SO4) and evaporated. The crude product
was recrystallized from EtOAc to obtain 27 as a colorless oil
N-(4-Phenylbenzoyl)-(R,S)-leucyl-(S)-methionine amide (31): Com-
pound 30 (0.26 g, 0.70 mmol) was dissolved in a mixture of dry
CH2Cl2 and TFA (10 mL, 1:1) and stirred for 2 h at 08C. The solvent
was evaporated and the oily crude product, (S)-N-(4-phenylben-
zoyl)leucine, was used without further purification and dissolved in
dry CH2Cl2 (10 mL). (S)-Methionine amide trifluoroacetate (15,
0.18 g, 0.70 mmol), DIPEA (1.21 mL, 7.00 mmol) and HATU (0.27 g,
0.70 mmol) were added. The solution was stirred overnight at RT.
After evaporation of the solvent the resulting aqueous residue was
extracted with EtOAc (3”50 mL). The combined organic layers
were washed with 10% KHSO4 (30 mL), H2O (30 mL), sat. NaHCO3
(30 mL), H2O (30 mL) and brine (50 mL). The solvent was dried
(Na2SO4) and evaporated. The crude product was recrystallized
from MeOH to obtain 31 as a white solid (0.25 g, 81% over two
steps); mp: 231–2328C; 1H NMR (500 MHz, [D6]DMSO, mixture of
diastereomers, ratio approximately 1:3 according to 1H NMR, w=
weak, refers to minor diastereomer, i=intense, refers to major dia-
stereomer): d=0.88–0.90 (w+i overlapping; m, 3H (w), 3H (i)),
0.93 (w+i overlapping; m, 3H (w), 3H (i)), 1.55–1.60 (w+i overlap-
ping; m, 1H (w), 1H (i)), 1.66–1.75 (w+i overlapping; m, 2H (w),
2H (i)), 1.76–1.84 (w+i overlapping; m, 1H (w), 1H (i)), 1.91–1.98
(w+i overlapping; m, 1H (w), 1H (i)), 2.01 (w+i overlapping; s, 3H
(w), 3H (i)), 2.37–2.46 (w+i overlapping; m, 2H (w), 2H (i)), 4.27–
4.31 (w+i overlapping; m, 1H (w), 1H (i)), 4.47–4.51 (w+i overlap-
ping; m, 1H (w), 1H (i)), 7.05 (s, 1H (i)), 7.11 (s, 1H (w)), 7.26 (s, 1H
(i)), 7.31 (s, 1H (w)), 7.39–7.42 (w+i overlapping; (m, 1H (w), 1H
(i)), 7.47–7.50 (w+i overlapping) (m, 2H (w), 2H (i)), 7.71–7.73 (w+
i overlapping; m, 2H (w), 2H (i)), 7.76–7.78 (w+i overlapping; m,
1
(0.40 g, 32%); mp: 82–838C, lit.[20] mp: 81–838C; H NMR (500 MHz,
[D6]DMSO): d=0.78–0.92 (m, 2H), 1.08–1.18 (m, 3H), 1.23–1.29 (m,
1H), 1.37 (s, 9H), 1.41 (t, J=7.1 Hz, 2H), 1.58–1.71 (m, 5H), 3.98 (q,
J=7.7 Hz, 1H), 4.09 (dd, J=5.6 Hz, J=2.7 Hz, 2H), 6.93 (d, J=
8.2 Hz, 1H), 8.50 ppm (t, J=5.5 Hz, 1H); 13C NMR (125 MHz,
[D6]DMSO): d=25.8, 26.0, 26.2, 27.2, 28.3, 31.8, 33.2, 33.7, 51.9,
78.2, 117.7, 155.5, 173.5 ppm, one carbon signal is obscured by the
DMSO signal; LC/ESI-MS (m/z): negative mode 308 [MÀH]À; purity,
100%.
(S,S)-N-(tert-Butoxycarbonyl)leucylglycine nitrile (28): Boc-Leu-OH
(0.93 g, 4.00 mmol) was dissolved in dry THF (10 mL) and cooled at
À258C. To the stirred solution, NMM (0.40 g, 4.00 mmol) and isobu-
tyl chloroformate (0.55 g, 4.00 mmol) were added consecutively.
Aminoacetonitrile monosulfate (0.62 g, 4.00 mmol) was dissolved
in 2n NaOH (2.00 mL, 4.00 mmol) and given to the reaction mix-
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1374
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
2H (w), 2H (i)), 7.92 (d, J=8.2 Hz, 1H (i)), 7.97–7.99 (w+i overlap-
ping; m, 2H (w), 2H (i)), 8.29 (d, J=8.5 Hz, 1H (w)), 8.55 (d, J=
7.9 Hz, 1H (i)), 8.60 ppm (d, J=7.0 Hz, 1H (w)); 13C NMR (125 MHz,
[D6]DMSO, w=weak, refers to minor diastereomer, i=intense,
refers to major diastereomer): d=14.7 (w), 14.8 (i), 21.6 (i), 22.0 (w),
22.9 (w), 23.2 (i), 24.6 (w+i), 29.7 (i), 29.9 (w), 31.3 (w), 32.0 (i), 51.7
(w), 51.8 (i), 52.4 (i), 52.8 (w), 126.6 (w+i), 127.0 (w+w+i), 128.2
(w+i), 128.4 (i), 129.2 (w+w+i), 132.9 (w), 133.1 (i), 139.3 (i), 143.0
(w+i), 166.4 (i), 166.7 (w), 172.2 (i), 172.5 (w), 173.1 (i), 173.4 ppm
(w), one carbon signal is obscured by the DMSO signal; negative
mode 440 [MÀH]À, positive mode 442 [M+H]+; purity, 87%.
(S,S)-N-Benzyloxycarbonylleucylmethionine amide (33): Cbz-Leu-
OH (32, 0.53 g, 2.00 mmol) was dissolved in dry CH2Cl2 (20 mL).
Compound 15 (0.52 g, 2.00 mmol), DIPEA (1.29 g, 1.74 mL,
10.00 mmol) and HATU (0.76 g, 2.00 mmol) were added. The solu-
tion was stirred overnight at RT. After evaporation of the solvent
the resulting aqueous residue was extracted with EtOAc (3”
50 mL). The combined organic layers were washed with 10%
KHSO4 (30 mL), H2O (30 mL), sat. NaHCO3 (30 mL), H2O (30 mL) and
brine (50 mL). The solvent was dried (Na2SO4) and evaporated. The
crude product was recrystallized from MeOH to obtain 33 as
a white solid (0.39 g, 50%); mp: 199–2018C, lit.[27] mp: 210–2118C;
1H NMR (500 MHz, [D6]DMSO): d=0.84 (d, J=6.7 Hz, 3H), 0.86 (d,
J=6.6 Hz, 3H), 1.43–1.45 (m, 2H), 1.57–1.65 (m, 1H), 1.74–1.81 (m,
1H), 1.87–1.95 (m, 1H), 2.01 (s, 3H), 2.35–2.46 (m, 2H), 4.00–4.04
(m, 1H), 4.25–4.29 (m, 1H), 5.02 (s, 2H), 7.04 (s, 1H), 7.26 (s, 1H),
7.28–7.37 (m, 5H), 7.44 (d, J=7.9 Hz, 1H), 7.84 ppm (d, J=8.2 Hz,
1H); 13C NMR (125 MHz, [D6]DMSO): d=14.8, 21.6, 23.1, 24.3, 29.7,
32.1, 40.6, 51.7, 53.5, 65.5, 127.8, 128.5, 127.9, 137.2, 156.1, 172.3,
173.0 ppm; negative mode 394 [MÀH]À, positive mode 396 [M+
H]+; purity, 95%.
tion was 360 nm and for emission 440 nm. The reactions were fol-
lowed at 258C over 20 min. Human cathepsins B and L were as-
sayed spectrophotometrically on a Cary 50 Bio, Varian, at 405 nm.
The reactions were followed at 378C over 20 min. The 10 mm in-
hibitor solutions were prepared in DMSO. The final concentration
of DMSO in all assays was 2%. Progress curves were analyzed by
linear regression. Results given in Tables 1 and 2 were obtained
from experiments performed in duplicate with five different inhibi-
tor concentrations. IC50 values were determined by nonlinear re-
gression using the equation v=v0/(1+[I]/IC50), in which v and v0
are the rates in the presence and absence of the inhibitor, respec-
tively, and [I] is the inhibitor concentration. Ki values were calculat-
ed from IC50 values by applying the equation Ki =IC50/(1+[S]/KM).
Cathepsin F inhibition assay: A human recombinant cathepsin F
(Enzo Life Sciences, Lçrrach, Germany) stock solution
(0.014 mgmLÀ1) in 20 mm NaOAc buffer pH 5.0, 2.5 mm EDTA,
250 mm NaCl, and 20 mm l-cysteine was diluted (1:50) with assay
buffer (100 mm sodium phosphate pH 6.0, 100 mm NaCl, 5 mm
EDTA, 0.01% Brij 35) containing 5 mm dithiothreitol (DTT) and incu-
bated for 2 h at 378C. A 10 mm stock solution of the fluorogenic
substrate Cbz-Phe-Arg-AMC was prepared in DMSO. The assay was
performed with a final cathepsin F concentration of 22.4 ngmLÀ1
.
The final concentration of substrate was 40 mm (2.1KM). Into a cuv-
ette containing 900 mL assay buffer, the inhibitor solution and
DMSO in a total volume of 16 mL, and 4 mL of the substrate solu-
tion (10 mm) were added and thoroughly mixed. The reaction was
initiated by adding 80 mL of the cathepsin F solution (280 ngmLÀ1).
Cathepsin B inhibition assay:[18] Assay buffer was 100 mm sodium
phosphate buffer pH 6.0, 100 mm NaCl, 5 mm EDTA, 0.01% Brij 35.
A stock solution of human isolated cathepsin B (Calbiochem, Darm-
stadt, Germany) of 1.81 mgmLÀ1 in 20 mm sodium acetate buffer
pH 5.0, 1 mm EDTA was diluted 1:500 with assay buffer containing
5 mm DTT and incubated for 30 min at 378C. A 100 mm stock solu-
tion of the chromogenic substrate Cbz-Arg-Arg-pNA was prepared
with DMSO. The final concentration of the substrate was 500 mm
(0.45KM). Assays were performed with a final cathepsin B concen-
tration of 72 ngmLÀ1. Into a cuvette containing 960 mL assay
buffer, inhibitor solution and DMSO in a total volume of 15 mL, and
5 mL of the substrate solution were added and thoroughly mixed.
The reaction was initiated by adding 20 mL of the cathepsin B solu-
tion.
(S,S)-N-(4-Phenylbenzoyl)leucylmethionine amide (34): Com-
pound 33 (0.20 g, 0.50 mmol) was dissolved in MeOH (20 mL) and
10% Pd/C (20 mg) was added. The reaction mixture was hydrogen-
ated at 44 psi for 7 h. Pd/C was filtered off and the solvent was
evaporated. The crude product, (S,S)-leucinylmethionine amide,
was used without further purification. Biphenyl-4-carboxylic acid
(24, 0.10 g, 0.50 mmol) was dissolved in dry CH2Cl2 (10 mL). The
(S,S)-leucinylmethionine amide, DIPEA (65 mg, 0.08 mL, 0.50 mmol)
and HATU (0.19 g, 0.50 mmol) were added. The solution was stirred
overnight at RT. After evaporation of the solvent the resulting
aqueous residue was extracted with EtOAc (3”50 mL). The com-
bined organic layers were washed with 10% KHSO4 (30 mL), H2O
(30 mL), sat. NaHCO3 (30 mL), H2O (30 mL) and brine (50 mL). The
solvent was dried (Na2SO4) and evaporated. The crude product was
recrystallized from MeOH to obtain 34 as a white solid (0.21 g,
95%, over two steps); mp: 245–2478C; 1H NMR (500 MHz,
[D6]DMSO): d=0.88 (d, J=6.3 Hz, 3H), 0.93 (d, J=6.3 Hz, 3H),
1.54–1.60 (m, 1H), 1.67–1.75 (m, 2H), 1.76–1.83 (m, 1H), 1.92–1.99
(m, 1H), 2.01 (s, 3H), 2.38–2.46 (m, 2H), 4.27–4.31 (m, 1H), 4.47–
4.51 (m, 1H), 7.05 (s, 1H), 7.25 (s, 1H), 7.39–7.42 (m, 1H), 7.47–7.50
(m, 2H), 7.72–7.73 (m, 2H), 7.76–7.78 (m, 2H), 7.91 (d, J=8.2 Hz,
1H), 7.97–7.99 (m, 2H), 8.54 ppm (d, J=7.6 Hz, 1H); 13C NMR
(125 MHz, [D6]DMSO): d=14.8, 21.6, 23.2, 24.6, 29.7, 32.0, 39.2,
51.8, 52.3, 126.6, 127.0, 128.2, 128.3, 129.2, 133.1, 139.3, 143.0,
166.4, 172.2, 173.1 ppm; negative mode 440 [MÀH]À, positive
mode 442 [M+H]+; purity, 96%.
Cathepsin L inhibition assay:[18] Assay buffer was 100 mm sodium
phosphate buffer pH 6.0, 100 mm NaCl, 5 mm EDTA, and 0.01%
Brij 35. A stock solution of human isolated cathepsin L (Enzo Life
Sciences, Lçrrach, Germany) of 135 mgmLÀ1 in 20 mm malonate
buffer pH 5.5, 400 mm NaCl, and 1 mm EDTA was diluted 1:100
with assay buffer containing 5 mm DTT and incubated for 30 min
at 378C. A 10 mm stock solution of the chromogenic substrate
Cbz-Phe-Arg-pNA was prepared with DMSO. The final concentra-
tion of the substrate was 100 mm (5.88KM). Assays were performed
with a final cathepsin L concentration of 54 ngmLÀ1. Into a cuvette
containing 940 mL assay buffer, inhibitor solution and DMSO in
a total volume of 10 mL, and 10 mL of the substrate solution were
added and thoroughly mixed. The reaction was initiated by adding
40 mL of the cathepsin L solution.
Cathepsin K inhibition assay:[18] A human recombinant cathepsin K
(Enzo Life Sciences, Lçrrach, Germany) stock solution of 23 mgmLÀ1
in 50 mm sodium acetate buffer pH 5.5, 50 mm NaCl, 0.5 mm EDTA,
5 mm DTT was diluted 1:100 with assay buffer (100 mm sodium cit-
rate pH 5.0, 100 mm NaCl, 1 mm EDTA, 0.01% CHAPS) containing
5 mm DTT and incubated for 30 min at 378C. A 10 mm stock solu-
Enzyme inhibition assays
Human cathepsins F, K, and S were assayed fluorimetrically on
a Monaco Safas spectrofluorimeter flx. The wavelength for excita-
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1375
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tion of the fluorogenic substrate Cbz-Leu-Arg-AMC was prepared
with DMSO. The final concentration of the substrate was 40 mm
(13.3KM). Assays were performed with a final cathepsin K concen-
tration of 5 ngmLÀ1. Into a cuvette containing 960 mL assay buffer,
inhibitor solution and DMSO in a total volume of 16 mL, and 4 mL
of the substrate solution were added and thoroughly mixed. The
reaction was initiated by adding 20 mL of the cathepsin K solution.
points. Accordingly, white areas in the landscape are interpolated
and not populated with data points.
Acknowledgements
The authors thank Tianwei Li, Adela Dudic, Karina Scheiner, Erik
Gilberg, and Robert Sellier (all University of Bonn, Germany) for
assistance. J.S. is supported by the Gender Equality Center of the
Bonn-Rhein-Sieg University of Applied Sciences, and N.F. by a fel-
lowship from the Jürgen Manchot Foundation (Düsseldorf, Ger-
many).
Cathepsin S inhibition assay:[28] Assay buffer was 100 mm sodium
phosphate buffer pH 6.0, 100 mm NaCl, 5 mm EDTA, and 0.01%
Brij 35. A stock solution of human recombinant cathepsin S (Enzo
Life Sciences, Lçrrach, Germany) of 70 mgmLÀ1 in 100 mm MES
buffer, pH 6.5, 1 mm EDTA, 50 mm l-cysteine, 10 mm DTT, 0.5%
Triton X-100 and 30% glycerol was diluted 1:100 with a 50 mm
sodium phosphate buffer pH 6.5, 50 mm NaCl, 2 mm EDTA, 0.01%
Triton X-100 and 5 mm DTT and incubated for 60 min at 378C. A
10 mm stock solution of the fluorogenic substrate Cbz-Phe-Arg-
AMC was prepared in DMSO. The assay was performed with a final
substrate concentration of 40 mm (0.74KM) and a final cathepsin S
concentration of 42 ngmLÀ1. Into a cuvette containing 920 mL
assay buffer, inhibitor solution and DMSO in a total volume of
16 mL, and 4 mL of the substrate solution were added and thor-
oughly mixed. The reaction was initiated by adding 60 mL of the
cathepsin S solution.
Keywords: 3D activity landscapes
human cathepsins · nitrile inhibitors
· cysteine proteases ·
[1] a) J. Reiser, B. Adair, T. Reinheckel, J. Clin. Invest. 2010, 120, 3421–3431;
b) V. Turk, V. Stoka, O. Vasiljeva, M. Renko, T. Sun, B. Turk, D. Turk, Bio-
ˇ
chim. Biophys. Acta 2012, 1824, 68–88; c) A. Pislar, J. Kos, Mol. Neurobiol.
2014, 49, 1017–1030.
[2] M. Frizler, M. Stirnberg, M. T. Sisay, M. Gütschow, Curr. Top. Med. Chem.
2010, 10, 294–322.
[3] B. G. Shi, R. A. Bryant, R. Riese, S. Verhelst, C. Driessen, Z. Li, D. Brçmme,
H. L. Ploegh, H. A. Chapman, J. Exp. Med. 2000, 191, 1177–1185.
[4] a) N. Obermajer, B. Doljak, J. Kos, Expert Opin. Biol. Ther. 2006, 6, 1295–
Docking studies
ˇ ´
ˇ
1309; b) M. Perisic Nanut, J. Sabotic, A. Jewett, J. Kos, Front. Immunol.
The X-ray structure of human cathepsin F (PDB ID: 1M6D)[22] bound
to an irreversible vinyl sulfone inhibitor (V, Figure 1) was obtained
from the RCSB Protein Data Bank.[29] The structure was used as
template for covalent ligand docking with AutoDock 4.2.[30] Prior to
the docking calculations, the bound inhibitor and crystallographic
water molecules were removed from the template structure, and
hydrogen atoms were added using Molecular Operating Environ-
ment (MOE 2012.10).[31] AutoDock Tools[30] was used to calculate
atomic partial charges. By forming a covalent thioimidate bond to
their nitrile function, candidate compounds were linked to the
active site residue Cys25. The side chain of the residue Cys25 in-
cluding the attached inhibitor structures was treated flexibly to
predict possible binding modes of selected compounds with Auto-
Dock 4.2. All other residues were fixed in their crystallographic con-
formation for the docking process. Putative compound binding
modes were selected after visual inspection of high-scoring dock-
ing poses.
2014, 5, 616.
ˇ
[5] M. Fonovic, D. Brçmme, V. Turk, B. Turk, Biol. Chem. 2004, 385, 505–
509.
[6] a) J. A. Villadangos, R. A. Bryant, J. Deussing, C. Driessen, A. M. Lennon-
DumØnil, R. J. Riese, W. Roth, P. Saftig, G. P. Shi, H. A. Chapman, C.
Peters, H. L. Ploegh, Immunol. Rev. 1999, 172, 109–120; b) K. Honey,
A. Y. Rudensky, Nat. Rev. Immunol. 2003, 3, 472–482; c) L. C. Hsing, A. Y.
Rudensky, Immunol. Rev. 2005, 207, 229–241.
[7] J. T. Palmer, D. Rasnick, J. L. Klaus, D. Brçmme, J. Med. Chem. 1995, 38,
3193–3196.
[8] A. Lee-Dutra, D. K. Wiener, S. Sun, Expert Opin. Ther. Pat. 2011, 21, 311 –
337.
[9] C. H. Tang, J. W. Lee, M. G. Galvez, L. Robillard, S. E. Mole, H. A. Chap-
man, Mol. Cell. Biol. 2006, 26, 2309–2316.
[10] K. R. Smith, H. H. Dahl, L. Canafoglia, E. Andermann, J. Damiano, M.
Morbin, A. C. Bruni, G. Giaccone, P. Cossette, P. Saftig, J. Grçtzinger, M.
Schwake, F. Andermann, J. F. Staropoli, K. B. Sims, S. E. Mole, S. France-
schetti, N. A. Alexander, J. D. Cooper, H. A. Chapman, S. Carpenter, S. F.
Berkovic, M. Bahlo, Hum. Mol. Genet. 2013, 22, 1417–1423.
[11] a) K. Öçrni, M. Sneck, D. Brçmme, M. O. Pentikäinen, K. A. Lindstedt, M.
Mäyränpää, H. Aitio, P. T. Kovanen, J. Biol. Chem. 2004, 279, 34776–
34784; b) H. D. Urbina, F. Debaene, B. Jost, C. Bole-Feysot, D. E. Mason,
P. Kuzmic, J. L. Harris, N. Winssinger, ChemBioChem 2006, 7, 1790–1797.
[12] a) J. J. M. Wiener, S. Sun, R. L. Thurmond, Curr. Top. Med. Chem. 2010, 10,
717–732; b) R. Lçser, Expert Opin. Ther. Pat. 2011, 21, 585–591.
[13] J. Y. Gauthier, W. C. Black, I. Courchesne, W. Cromlish, S. Desmarais, R.
Houle, S. Lamontagne, C. S. Li, F. MassØ, D. J. McKay, M. Ouellet, J. Robi-
chaud, J. F. Truchon, V. L. Truong, Q. Wang, M. D. Percival, Bioorg. Med.
Chem. Lett. 2007, 17, 4929–4933.
Generation of a 3D activity landscape
For the inhibitor set, a 3D activity landscape was generated as de-
scribed previously.[32] The ECFP2[33] fingerprint was used to calculate
Euclidean distance relationships between compounds (constituting
a coordinate-free chemical reference space), which were projected
onto a 2D plane by dimension reduction using multidimensional
scaling (MDS).[34] As a third dimension, the compound potency of
the 57 cathepsin F inhibitors was added. By interpolating with the
krige function,[35] coherent potency surface was created. The color
of the surface was adjusted to the interpolated potency values
(surface elevation) using a color gradient from green (indicating
lowest potency values) over yellow to red (highest potency). The
density of experimental potency measurements is reflected by the
transparency of the activity landscape. Opaque colored grid points
represent close proximity to original data points, whereas fully
transparent grid points are furthest away from experimental data
[14] M. Frizler, F. Lohr, M. Lülsdorff, M. Gütschow, Chem. Eur. J. 2011, 17,
11419–11423.
[15] W. C. Black, Curr. Top. Med. Chem. 2010, 10, 745–751.
[16] a) R. Lçser, M. Frizler, K. Schilling, M. Gütschow, Angew. Chem. Int. Ed.
2008, 47, 4331–4334; Angew. Chem. 2008, 120, 4403–4406; b) J. Y.
Gauthier, N. Chauret, W. Cromlish, S. Desmarais, T. Duongle, J. P. Fal-
gueyret, D. B. Kimmel, S. Lamontagne, S. LØger, T. LeRiche, C. S. Li, F.
MassØ, D. J. McKay, D. A. Nicoll-Griffith, R. M. Oballa, J. T. Palmer, M. D.
Percival, D. Riendeau, J. Robichaud, G. A. Rodan, S. B. Rodan, C. Seto, M.
ThØrien, V. L. Truong, M. C. Venuti, G. Wesolowski, R. N. Young, R. Zam-
boni, W. C. Black, Bioorg. Med. Chem. Lett. 2008, 18, 923–928; c) J.
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1376
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Schmitz, A. M. Beckmann, A. Dudic, T. Li, R. Sellier, U. Bartz, M. Güt-
schow, ACS Med. Chem. Lett. 2014, 5, 1076–1081.
[17] M. Frizler, J. Schmitz, A. C. Schulz-Fincke, M. Gütschow, J. Med. Chem.
2012, 55, 5982–5986.
[27] S. Bajusz, Acta Chim. Acad. Sci. Hung. 1964, 42, 383–391.
[28] M. D. Mertens, J. Schmitz, M. Horn, N. Furtmann, J. Bajorath, M. Mares,
M. Gütschow, ChemBioChem 2014, 15, 955–959.
ˇ
[29] RCSB Protein Data Bank: www.rcsb.org.
[18] M. Frizler, F. Lohr, N. Furtmann, J. Kläs, M. Gütschow, J. Med. Chem.
2011, 54, 396–400.
[19] R. Lçser, M. Gütschow, J. Enzyme Inhib. Med. Chem. 2009, 24, 1245–
1252.
[30] G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner, R. K. Belew, D. S. Good-
sell, A. J. Olson, J. Comput. Chem. 2009, 30, 2785–2791.
[31] Molecular Operating Environment (MOE), ver. 2012.10, Chemical Com-
puting Group Inc., Montreal, QC (Canada).
[20] R. Lçser, K. Schilling, E. Dimmig, M. Gütschow, J. Med. Chem. 2005, 48,
7688–7707.
[32] L. Peltason, P. Iyer, J. Bajorath, J. Chem. Inf. Model. 2010, 50, 1021–1033.
[33] D. Rogers, M. Hahn, J. Chem. Inf. Model. 2010, 50, 742–754.
[34] I. Borg, P. J. F. Groenen, Modern Multidimensional Scaling. Theory and Ap-
plications, 2nd ed., Springer, New York, NY, 2005.
[35] N. Cressie, Statistics for Spatial Data, revised ed., Wiley-VCH, New York,
NY, 1993.
[21] J. Y. Reginster, A. Neuprez, C. Beaudart, M. P. Lecart, N. Sarlet, D. Bernard,
S. Disteche, O. Bruyere, Drugs Aging 2014, 31, 413–424.
[22] J. R. Somoza, J. T. Palmer, J. D. Ho, J. Mol. Biol. 2002, 322, 559–568.
[23] G. A. Olah, S. C. Narang, A. P. Fung, B. G. Gupta, Synthesis 1980, 657–
658.
[24] S. Schneider, W. Bannwarth, Helv. Chim. Acta 2001, 84, 735–742.
[25] D. Stumpfe, J. Bajorath, J. Med. Chem. 2012, 55, 2932–2942.
[26] H. Berger, A. E. Siegrist, Helv. Chim. Acta 1979, 62, 1411–1428.
Received: April 7, 2015
Published online on June 26, 2015
ChemMedChem 2015, 10, 1365 – 1377
www.chemmedchem.org
1377
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim