Fluorine Scan of Inhibitors of the Cysteine Protease Human Cathepsin L: Dipolar and Quadrupolar Effects in the π-Stacking of Fluorinated Phenyl Rings on Peptide Amide Bonds

Article abstract of DOI:10.1002/cmdc.201600132

The π-stacking of fluorinated benzene rings on protein backbone amide groups was investigated, using a dual approach comprising enzyme-ligand binding studies complemented by high-level quantum chemical calculations. In the experimental study, the phenyl substituent of triazine nitrile inhibitors of human cathepsin L (hCatL), which stacks onto the peptide amide bond Gly67-Gly68 at the entrance of the S3 pocket, was systematically fluorinated, and differences in inhibitory potency were measured in a fluorimetric assay. Binding affinity is influenced by lipophilicity (clog P), the dipole and quadrupole moments of the fluorinated rings, but also by additional interactions of the introduced fluorine atoms with the local environment of the pocket. Generally, the higher the degree of fluorination, the better the binding affinities. Gas phase calculations strongly support the contributions of the molecular quadrupole moments of the fluorinated phenyl rings to the π-stacking interaction with the peptide bond. These findings provide useful guidelines for enhancing π-stacking on protein amide fragments. The introduction of fluorine substituents on benzene rings of triazine nitrile inhibitors for human cathepsin L (hCatL) enhances the π-stacking of these rings with peptide amide bonds at the entrance of the S3 pocket. The degree and pattern of fluorine substitution matter, which is explained in an experimental/computational study by changes in lipophilicity, dipolar and quadrupolar interactions, and additional protein-ligand interactions beyond the arene-amide stacking.

Full text of DOI:10.1002/cmdc.201600132

DOI: 10.1002/cmdc.201600132
Communications
Fluorine Scan of Inhibitors of the Cysteine Protease
Human Cathepsin L: Dipolar and Quadrupolar Effects in
the p-Stacking of Fluorinated Phenyl Rings on Peptide
Amide Bonds
Maude Giroud,[a] Michael Harder,[a] Bernd Kuhn,[b] Wolfgang Haap,[b] Nils Trapp,[a] W.
Bernd Schweizer,[a] Tanja Schirmeister,*[c] and FranÅois Diederich*[a]
The p-stacking of fluorinated benzene rings on protein back-
bone amide groups was investigated, using a dual approach
comprising enzyme–ligand binding studies complemented by
high-level quantum chemical calculations. In the experimental
study, the phenyl substituent of triazine nitrile inhibitors of
human cathepsin L (hCatL), which stacks onto the peptide
amide bond Gly67ÀGly68 at the entrance of the S3 pocket,
was systematically fluorinated, and differences in inhibitory po-
tency were measured in a fluorimetric assay. Binding affinity is
influenced by lipophilicity (clogP), the dipole and quadrupole
moments of the fluorinated rings, but also by additional inter-
actions of the introduced fluorine atoms with the local envi-
ronment of the pocket. Generally, the higher the degree of flu-
orination, the better the binding affinities. Gas phase calcula-
tions strongly support the contributions of the molecular
quadrupole moments of the fluorinated phenyl rings to the p-
stacking interaction with the peptide bond. These findings pro-
vide useful guidelines for enhancing p-stacking on protein
amide fragments.
Gly61 in the S3 pocket of hCatL was systematically investigat-
ed and quantified.[5] In another study, which focused on the in-
hibition of rhodesain (RD), a parasitic cysteine protease pro-
duced by Trypanosoma brucei rhodesiense and a potential drug
target against human African trypanosomiasis (sleeping sick-
ness), extensive structure–activity relationships (SARs) of tria-
zine nitrile ligands inhibiting both RD and hCatL were per-
formed.[6] A co-crystal structure of triazine nitrile 1 binding to
hCatL was solved (Figure 1).[6c] The catalytic Cys25 residue un-
dergoes reversible covalent binding, reacting with the activat-
ed nitrile to form a thioimidate stabilized by the oxyanion
hole. The active site of hCatL is subdivided into three sub-
pockets S1–S3. The S1 pocket is solvent exposed and tolerates
a wide range of substituents without much affecting the over-
all ligand binding affinities. The S2 pocket is rather hydropho-
bic, and a diversity of lipophilic residues, such as cyclohexyl de-
rivatives, were successfully introduced.[6c] The S3 pocket is sol-
vent exposed, and two glycines (Gly67 and Gly68) form a flat
dipeptide fragment at the entrance of the pocket. The 4-chlo-
robenzyl ring of ligand 1 stacks on the Gly67ÀGly68 peptide
bond at a distance of ~3.6 Š (Figure 1).[6c] In this ligand class,
the chlorine substituent is too far apart from the C=O of Gly61
to form a strong halogen bond.
Human cathepsin L (hCatL, EC 3.4.22.15) is a cysteine protease
explored as a target against atherosclerosis and abdominal
aortic aneurysms,[1] various cancers,[2] and as a predictive
marker for pancreatic cancer.[3] It also mediates the degrada-
tion of amyloid precursor proteins in Alzheimer’s disease.[4]
This enzyme has been an ongoing subject of biological molec-
ular recognition studies in our collaborative research. Halogen
bonding (XB) of halophenyl-substituted ligands to the C=O of
Organofluorine is increasingly used to tune the physico-
chemical[7] and ADMET (administration, distribution, metabo-
lism, excretion, toxicity) properties of drugs.[8] Previous work
also demonstrated that the introduction of organofluorine may
enhance binding affinity and selectivity.[9] The underlying ex-
perimental studies involved fluorine scans[10] of inhibitors for
a diversity of enzymes, mapping the fluorophilicity and fluoro-
phobicity of their active sites.[11–13]
[a] M. Giroud, Dr. M. Harder, Dr. N. Trapp, Dr. W. B. Schweizer,
Prof. Dr. F. Diederich
In 2013, we reported a computational study revealing that
p-stacking of heteroarenes on peptide amide bonds becomes
increasingly favorable with: 1) decreasing p-electron density of
the heterocycle and 2) stronger local dipole moments of the
heterocycle when aligned in an antiparallel orientation to the
local dipole moment of the peptide bond.[14] Prediction 1 was
recently validated in experimental model studies.[15] The p-elec-
tron density of benzene rings is also strongly decreased with
increasing fluorine substitution, and we became interested in
evaluating how fluorine substitution on a phenyl ring would
affect p-stacking on peptide amide bonds. With the large
dipole moments of Csp2ÀF bonds and of appropriately fluorine-
substituted phenyl rings,[16] we also expected effects of the ori-
Laboratorium für Organische Chemie, ETH Zürich,
Wolfgang-Pauli-Strasse 10, HCI, 8093 Zürich (Switzerland)
E-mail: diederich@org.chem.ethz.ch
[b] Dr. B. Kuhn, Dr. W. Haap
Small Molecule Research, Roche Innovation Center Basel,
F. Hoffmann–La Roche AG, Grenzacherstrasse 124, Building 92, 4070 Basel
(Switzerland)
[c] Prof. Dr. T. Schirmeister
Institut für Pharmazie und Biochemie,
Johannes Gutenberg-Universität Mainz, Staudinger Weg 5, 55128 Mainz
(Germany)
E-mail: schirmei@uni-mainz.de
Supporting information for this article can be found under http://
dx.doi.org/10.1002/cmdc.201600132.
ChemMedChem 2016, 11, 1042 – 1047
1042
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Figure 1. Co-crystal structure of triazine nitrile 1 with hCatL (PDB ID: 4AXM,[6c] resolution 2.80 Š). Color code: gray—Cenzyme, green—Cligand, red—O, blue—N,
yellow—S, lime—Cl.
entations of the local substituted ring dipoles relative to the
peptide bond dipole.
ing mode of 2 was considered, in which the fluorinated ring
would bind to the S2 pocket (Figure S2). However, the lipophil-
ic cyclopentyl ring would then be fully exposed to bulk sol-
vent, and hence this binding mode was excluded.
Herein we report a systematic fluorine scan—from mono- to
pentafluorination—of the benzyl substituent of the triazine ni-
trile inhibitors reaching into the S3 pocket of hCatL and p-
stacking on the Gly67ÀGly68 peptide bond at the entrance of
the pocket. Adding a computational study, we explain the ex-
perimentally measured large differences in inhibitory activity
by a combination of lipophilicity, dipolar and quadrupolar ef-
fects, as well as interactions of the fluorinated rings with the
local environment of the S3 pocket. A similar analysis for com-
plexation by the structurally related cysteine protease rhode-
sain is included in the Supporting Information (SI).
A series of 20 differently fluorinated triazine nitriles 2–21
was prepared (Table 1), employing a previously developed
three-step synthesis,[6c] which is exemplified for inhibitor 2 in
Scheme 1 (see the SI for all synthetic details). X-Ray crystal
structures of six triazine nitriles confirmed their proposed
structures (Section S5 in the SI). Noteworthy is that the S2 and
S3 pocket vectors in the single-molecule crystal structures do
not always adopt the same orientation as modeled in the
active site. For the triazine nitrile ligands containing 0 (3), 1 (4–
6), and 2 (8) fluorine atoms on the ring, the benzyl and cyclo-
pentyl rings switch positions relative to the predicted enzyme-
bound geometry by rotation of 1808 around the exocyclic CÀN
bond. The differences in geometry are possibly due to crystal
We selected a series of triazine nitrile analogues of 1
(Ki (hCatL)=13 nm, Figure 1), in which the benzylic ring (with-
out the 4-chloro substituent) was sequentially fluorinated. The
morpholine substituent in 1 for the S1 pocket was kept,
whereas the cyclohexyl derivative for the S2 pocket was ex-
changed for a cyclopentyl ring. This modification decreases the
binding affinity, thereby enlarging the activity range in which
assay data during the fluorine scan could be obtained at
a high confidence level. Modeling with MOLOC,[17] using the
MAB force field, suggested that the various fluorinated ligands
in the active site of hCatL adopt a similar binding geometry to
that observed in the X-ray co-crystal structure of 1 (Figure 1,
Figure S1 in the SI). In the modeling, the protein coordinates,
with the exception of the side chain of Glu63, were fixed as
well as the position of the cyano group, covalently bonded as
thioimidate to Cys25. In the previous XB study with hCatL,
a large number of co-crystal structures had revealed that the
side chain of Glu63, located at the surface of the protein, is
highly flexible and can adopt various conformations. Substitu-
tion of the benzyl ring with up to five fluorine atoms did not
cause steric repulsion in our model. In the complex of inhibitor
2 (see Table 1 for the structures of all compounds), the penta-
fluorophenyl ring is predicted to stack on the peptide bond
Gly67ÀGly68 at a distance of ~3.5 Š (Figure S1). Another bind-
Scheme 1. Synthesis of inhibitor 2. a) 1) CH2Cl2, molecular sieves (4 Š), 258C,
1 h, 2) NaBH(OAc)3, 258C, 16 h, 59%; b) 1) cyanuric chloride, iPr2NEt, CH2Cl2,
08C, 2 h, 2) morpholine, iPr2NEt, 258C, 16 h, 78%; c) KCN, DABCO, Me2SO/
H2O (10:1), 808C, 18 h, 50%. DABCO=1,4-diazabicyclo[2.2.2]octane.
ChemMedChem 2016, 11, 1042 – 1047
www.chemmedchem.org
1043
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
200 mm NaCl, and 0.005% Brij35) at 298 K.[18] The results are
summarized in Table 1. The difference in binding affinity be-
tween the best (15) and the weakest inhibitor (19) is a factor
of 13, which corresponds to a change of approximatively
1.8 kcalmolÀ1 in Gibbs energy. The reported pKi values result
from the mean Ki values of at least two independent assays,
each performed in duplicate. The data for the binding of 2 re-
sulted from seven independent measurements. Uncertainties
in the inhibition constant Ki are listed in Table S4 in the SI.
With the exception of the outlier 19, all fluorinated inhibitors
show higher binding affinity than benzyl derivative 3. As a gen-
eral trend (except 19), the higher the degree of fluorination,
the better the binding affinity. The degree and pattern of fluo-
rination, however, affect the binding affinities in a more com-
plex way (Figure 2).
Table 1. Fluorination pattern of triazine nitriles 2–21, predicted clogP
values, and pKi values for hCatL inhibition in aqueous buffer[a] at 298 K.
R
Compd
Fluorination
pattern
Predicted
clogP[b]
pKi
(hCatL)
3
–
0.87
6.3
4
5
6
2
3
4
0.93
0.93
0.93
6.5
6.7
6.3
7
8
9
10
11
12
2,3
2,4
2,5
2,6
3,4
3,5
0.89
1.02
1.05
1.02
0.89
1.02
6.7
6.9
6.6
7.0
6.5
6.5
13
14
15
16
17
18
2,3,4
2,3,5
2,3,6
2,4,5
3,4,5
2,4,6
1.04
1.04
1.04
1.04
1.04
1.14
6.5
6.7
7.2
6.5
6.4
7.1
19
20
21
2,3,4,5
2,3,4,6
2,3,5,6
0.95
1.08
1.11
6.1
7.0
7.0
Figure 2. Plot of the pKi values for hCatL against the number of fluorine
atoms on the phenyl ring.
2
2,3,4,5,6
1.06
6.9
Adding fluoro substituents to a molecule generally enhances
its lipophilicity, which in turn should favor protein bind-
ing.[7,9,12c,19] For our set of compounds, the predicted octanol/
water partition coefficients (clogP) were calculated using ACD/
Labs software[20] (Table 1). The clogP values only span from
0.87 (3) to 1.14 (18), whereas pKi values vary from 6.1 (19) to
7.2 (15). Figure S3 (SI) shows the weak correlation (R2 value=
0.36 for a linear correlation) between inhibition potency (pKi)
and clogP values. Besides lipophilicity, other physical parame-
ters clearly contribute to the measured changes in hCatL bind-
ing potency upon fluorination.
[a] 50 mm Tris (pH 6.5), 5 mm EDTA, 200 mm NaCl, 0.005% Brij35.
[b] Values calculated with ACD/Labs Version 12.5 (ꢀ1994–2012 ACD/Labs).
packing effects. To better quantify the energy difference be-
tween these two conformations we performed NMR measure-
ments and quantum chemical calculations (Section S5 in the
SI). In CDCl3, mixtures of the two rotamers in 6:4 or 7:3 ratios
are observed. A torsional scan of 4 and 10 performed at the
B3LYP/6-31G(d) level further suggests a 75:25 rotamer distribu-
tion for 4 at 298 K and a 40:60 distribution for 10, with the
most stable conformer for 4 adopting the geometry as in the
modeled enzyme–ligand complex, while 10 possesses the op-
posite orientation of the S2 and S3 vectors. In summary, the
occurrence of two main conformations with respect to the
aminotriazine bond shows that the ligands are not fully preor-
ganized for protein binding, which presumably results in an
entropic penalty for complexation.
We first considered the effects of local dipole moment vec-
tors of the fluorinated rings. Based on the theoretical study by
Harder et al.,[14] we expected that rings with higher dipole mo-
ments in antiparallel alignment to the Gly67ÀGly68 peptide
bond will lead to enhanced ligand binding. The dipole mo-
ments for fluorinated toluene derivatives (as models for the
benzylic S3 substituent) and, for comparison, the correspond-
ing benzene derivatives were calculated at the MP2/aug-cc-
pVDZ level of theory (Table S1 in Section S1 in the SI). In the p-
stacking complex geometry, only the model for the 2,6-disub-
stituted benzyl ring in 10 shows a significant local dipole
Ligands 2–21 were tested against hCatL in a fluorimetric
assay in aqueous buffer (50 mm Tris pH 6.5, 5 mm EDTA,
ChemMedChem 2016, 11, 1042 – 1047
www.chemmedchem.org
1044
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
moment (1.3 D) antiparallel to the local dipole moment of the
Gly67ÀGly68 bond (model: N-methylacetamide (NMAC): 3.7 D).
Ligand 10 is among the best binders (pKi =7.0, Table 1). How-
ever, the situation is complicated by the fact that a second flat
peptide bond fragment, Gly68ÀLeu69, with an opposite direc-
tion of its local dipole moment is in close proximity and may
also affect dipolar interactions (Figure S4 in the SI). While in
the modeling, all benzyl substituents mainly stack on the
Gly67ÀGly68 bond, the influence of the second amide bond in
the dipeptide at the entrance of the S3 pocket cannot be ig-
nored. The plot of pKi values as a function of dipole moment
does not show a significant correlation (Figure S5 in the SI).
Furthermore, fluorinated benzyl rings with a larger dipole
moment are expected to be better solvated in the aqueous
buffer, resulting in a higher desolvation penalty upon binding.
Next, we investigated the influence of the quadrupole mo-
ments of the fluorinated benzyl rings on binding affinity. The
attractive pentafluorophenyl/phenyl and hexafluorobenzene/
benzene eclipsed stacking interactions, based on large oppo-
site-sign quadrupole moments, are well documented in supra-
molecular chemistry and crystal engineering[21] and have also
been quantified in a protein environment.[22] Hernµndez-Trujillo
and Vela have calculated the molecular quadrupole moments
for a large series of fluorobenzenes,[16a] and we plotted Qzz (the
quadrupole moment tensor component perpendicular to the
aromatic ring) against the pKi values of our ligands. A poor cor-
relation (R2 =0.17) was observed (Figure S6). However, when
only inhibitors with dipole moments close to zero (m<1 D, Fig-
ure S7) were considered, a clear linear dependence of the
binding affinity on the quadrupole moment appears (R2 =
0.75). The same linear correlation is even stronger in the case
of RD binding (R2 =0.97; Figure S18).
placed arrangement approximates best our modeled binding
mode. This study for the first time reveals the importance of
quadrupolar aromatic stacking interactions with peptide amide
bonds.
The 2,3,4,5,6-pentafluorobenzyl derivative 2, which approxi-
mates the corresponding model compound, is very potent
against hCatL but not the best binder in the series. This can be
explained by the exposure of individual fluoro substituents to
the residual protein environment of the S3 pocket.
The protein surrounding the S3 pocket, beyond the flat
Gly67ÀGly68ÀLeu69 dipeptide fragment, features regions of
markedly different polarity, as revealed by surfaces generated
with MOE[23] for the complex with pentafluorophenyl derivative
2 (Figure 4). The electrostatic (A) and lipophilic surfaces (B)
show that the hydrophobic Leu69 side chain at the entrance
of the pocket creates a hydrophobic wall, which is a favorable
environment for organofluorine. On the other hand, the oppo-
site pocket side has high negative electrostatic potential, re-
sulting largely from the flexible side chain of Glu63, which is
a fluorophobic environment.
2,3,6-Trifluorophenyl derivative 15 (pKi =7.2) is the best
binder of the whole series. We propose that the 2,3-F atoms fa-
vorably interact with the Leu69 side chain, whereas the 6-F
atom does not point into the region of negative electrostatic
potential created by the Glu63 side chain. In contrast, 2,3,5-tri-
fluoro derivative 16 is a much weaker binder (pKi =6.5), be-
cause the 5-F atom points into the electronegative potential
region. Indeed, most of the ligands with a 5-F atom, postulated
by modeling to point in this region, are on the weaker affinity
side (e.g. 9, 12, 16, 17, and 19). Compound 19 differs from the
general trend. Its activity should be similar to that of the other
fourfold fluorinated compounds based on the dipole and
quadrupole moments of its phenyl ring. We attribute this dis-
crepancy to the fact that 19 does not possess a 2,6-fluorination
pattern, which generally leads to higher binding affinities. Also,
its clogP value (Figure S3 in the SI) is substantially lower than
those of the other tetrafluoro derivatives.
We calculated the gas-phase interaction energies between
NMAC as model for the peptide bond and a set of fluorinated
benzenes using quantum chemical calculations at the SCS-
MP2/aug-cc-pVDZ//LMP2/cc-pVDZ level of theory.[14] Both par-
allel-displaced (Figure 3) and parallel (“eclipsed”) alignments
(Figure S10) were calculated.[16] The trend in binding energies
is similar for the two alignments. The best interaction energies
are calculated for hexafluorobenzene with the largest molecu-
lar quadrupole moment of opposite sign to benzene (Qzz =
6.09 B) in both “eclipsed” (DE=À3.6 kcalmolÀ1) and “parallel-
displaced” stacking (DE=À4.1 kcalmolÀ1). The parallel-dis-
An enzyme–ligand network interaction analysis was per-
formed for the binding of the pentafluorophenyl ligand 2 to
hCatL (Figures S11, S12).[24] This analysis recognizes the p–p in-
teraction of the phenyl ring with the amide backbone and sug-
gests that the fluorine atoms in positions 2, 3 (interactions
with the Leu69 side chain), and 5 (interaction with backbone
carbonyl carbon of Asn66) specifically contribute to binding.
However, according to the experimental data, the latter inter-
action is unfavorably compensated by the location of this fluo-
rine in a region of high negative electrostatic potential.
Finally, we performed a Matched Molecular Pair (MMP) anal-
ysis of our data set focusing on transformations with changes
by one heavy atom (Table S2).[25] The transformations with the
biggest affinity gains almost exclusively involve addition of
a fluorine at positions 2 or 6 on the benzyl ring (DpKi =0.8 for
2 vs. 19, DpKi =0.6 for 15 vs. 9 and for 8 vs. 6, and DpKi =0.5
for 15 vs. 7, for 10 vs. 4, and for 20 vs. 13). According to this
analysis, a fluorine atom in the ortho position to the scaffold
increases the hCatL binding affinity most efficiently. This can
be rationalized by the fact that 1) the fluorine atom can inter-
Figure 3. Binding energies (SCS-MP2 calculations) for the parallel-displaced
arrangement, reported with dipole moments[16b] and quadrupolar moments
Qzz.[16a] B=Buckingham; a [8] is defined as the angle between the dipole
moment vectors of 1,2,3-trifluorobenzene and NMAC.
ChemMedChem 2016, 11, 1042 – 1047
www.chemmedchem.org
1045
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Figure 4. A) Electrostatic surface of the S3 pocket of hCatL bound to triazine nitrile 2. B) Lipophilic surface of hCatL with inhibitor 2. Surfaces generated with
MOE.[23] Color code: blue—Cenzyme, gray—Cligand, red—O, blue—N, green—F. Surfaces: red=negative, blue=positive, pink=lipophobic, green=lipophilic.
act with the side chain of Leu69, 2) the additional bond dipole
Keywords: cysteine protease inhibitors · dipolar interactions ·
component is roughly antiparallel to the Gly67ÀGly68 amide
fluorine scans · peptide bonds · quadrupolar interactions ·
dipole, and 3) the quadrupolar moment Qzz becomes slightly
quantum chemical calculations · p-stacking
more positive.
In summary, the binding of triazine nitrile inhibitors bearing
[1] B.-J. Lv, J. S. Lindholt, J. Wang, X. Cheng, G.-P. Shi, Atherosclerosis 2013,
benzyl rings with different degrees and patterns of fluorination
to fill the S3 pocket of hCatL was investigated. A similar analy-
sis is reported in the Supporting Information (SI) for binding to
rhodesain.[26] In general, fluorination enhances binding relative
to the unsubstituted benzyl derivative and provides a valuable
means to enhance aryl stacking interactions with protein back-
bone amide bonds. But the degree and pattern of fluorination
matter significantly, leading to a difference in Gibbs energy of
1.8 kcalmolÀ1 (factor of 13) between the best and the weakest
binder. A complex analysis, including a high-level computation-
al study, suggests that binding affinity is affected by 1) the lip-
ophilicity of the S3 pocket substituent (clogP values), 2) favora-
ble and unfavorable dipolar interactions, 3) favorable quadru-
polar interactions, and 4) favorable and unfavorable organo-
fluorine interactions with the S3 pocket environment. We be-
lieve that these findings provide valuable guidelines for future
work on ligands that undergo p-stacking with aromatic rings
on peptide amide bonds, and further highlight the importance
of fluorine in ligand design.
230, 100–105.
[2] a) L. Zhang, L. Wei, G. Shen, B. He, W. Gong, N. Min, L. Zhang, Y. Duan, J.
Xie, H. Luo, X. Gao, Mol. Med. Rep. 2015, 11, 468–474; b) D. R. Sudhan,
D. W. Siemann, Pharmacol. Ther. 2015, 155, 105–116; c) O. C. Olson, J. A.
Joyce, Nat. Rev. Cancer 2015, 15, 712–729.
[3] N. Singh, P. Das, S. Gupta, V. Sachdev, S. Srivasatava, S. Datta Gupta,
R. M. Pandey, P. Sahni, S. S. Chauhan, A. Saraya, World J. Gastroenterol.
2014, 20, 17532–17540.
[4] H. Wang, N. Sang, C. Zhang, R. Raghupathi, R. E. Tanzi, A. Saunders, Bio-
chemistry 2015, 54, 2806–2816.
[5] a) L. A. Hardegger, B. Kuhn, B. Spinnler, L. Anselm, R. Ecabert, M. Stihle,
B. Gsell, R. Thoma, J. Diez, J. Benz, J.-M. Plancher, G. Hartmann, D. W.
Banner, W. Haap, F. Diederich, Angew. Chem. Int. Ed. 2011, 50, 314–318;
Angew. Chem. 2011, 123, 329–334; b) L. A. Hardegger, B. Kuhn, B.
Spinnler, L. Anselm, R. Ecabert, M. Stihle, B. Gsell, R. Thoma, J. Diez, J.
Benz, J.-M. Plancher, G. Hartmann, Y. Isshiki, K. Morikami, N. Shimma, W.
Haap, D. W. Banner, F. Diederich, ChemMedChem 2011, 6, 2048–2054.
[6] a) V. Ehmke, C. Heindl, M. Rottmann, C. Freymond, W. B. Schweizer, R.
Brun, A. Stich, T. Schirmeister, F. Diederich, ChemMedChem 2011, 6,
273–278; b) V. Ehmke, J. E. Q. Quinsaat, P. Rivera-Fuentes, C. Heindl, C.
Freymond, M. Rottmann, R. Brun, T. Schirmeister, F. Diederich, Org.
Biomol. Chem. 2012, 10, 5764–5768; c) V. Ehmke, E. Winkler, D. W.
Banner, W. Haap, W. B. Schweizer, M. Rottmann, M. Kaiser, C. Freymond,
T. Schirmeister, F. Diederich, ChemMedChem 2013, 8, 967–975.
[7] a) D. O’Hagan, Chem. Soc. Rev. 2008, 37, 308–319; b) Q. A. Huchet, B.
Kuhn, B. Wagner, N. A. Kratochwil, H. Fischer, M. Kansy, D. Zimmerli,
E. M. Carreira, K. Müller, J. Med. Chem. 2015, 58, 9041–9060.
[8] a) S. Purser, P. R. Moore, S. Swallow, V. Gouverneur, Chem. Soc. Rev.
2008, 37, 320–330; b) Fluorine in Pharmaceutical and Medicinal Chemis-
try (Eds.: V. Gouverneur, K. Müller), Imperial College Press, London,
2012; c) E. P. Gillis, K. J. Eastman, M. D. Hill, D. J. Donnelly, N. A. Mean-
well, J. Med. Chem. 2015, 58, 8315–8359; d) Y. Zhou, J. Wang, Z. Gu, S.
Wang, W. Zhu, J. L. AceÇa, V. A. Soloshonok, K. Izawa, H. Liu, Chem. Rev.
2016, 116, 422–518.
Acknowledgements
This research was supported by F. Hoffmann–La Roche, Basel,
and the ETH Research Council. We thank Ulrike Nowe, Sabine
Maehrlein, and Nicole Heindl from the Johannes-Gutenberg Uni-
versität Mainz for support with the biological assays, and Mattia
Valentini for the preparation of one final compound. We are
grateful to Dr. Bruno Bernet for correcting the Experimental Sec-
tion in the Supporting Information and to Dr. Oliver Dumele for
theoretical calculations. Dr. Christian Kramer is acknowledged for
the MMP analysis program.
[9] K. Müller, C. Faeh, F. Diederich, Science 2007, 317, 1881–1886.
[10] For an early computational fluorine scan, see: B. Kuhn, P. A. Kollman, J.
Am. Chem. Soc. 2000, 122, 3909–3916.
[11] a) J. B. Doyon, A. Jain, Org. Lett. 1999, 1, 183–185; b) C.-Y. Kim, J. S.
Chang, J. B. Doyon, T. T. Baird Jr., C. A. Fierke, A. Jain, D. W. Christianson,
J. Am. Chem. Soc. 2000, 122, 12125–12134.
[12] a) J. A. Olsen, D. W. Banner, P. Seiler, U. Obst-Sander, A. D’Arcy, M. Stihle,
K. Müller, F. Diederich, Angew. Chem. Int. Ed. 2003, 42, 2507–2511;
ChemMedChem 2016, 11, 1042 – 1047
www.chemmedchem.org
1046
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Communications
Angew. Chem. 2003, 115, 2611–2615; b) J. Olsen, P. Seiler, B. Wagner, H.
Fischer, T. Tschopp, U. Obst-Sander, D. W. Banner, M. Kansy, K. Müller, F.
Diederich, Org. Biomol. Chem. 2004, 2, 1339–1352; c) E. Schweizer, A.
Hoffmann-Rçder, K. Schärer, J. A. Olsen, C. Fäh, P. Seiler, U. Obst-Sander,
B. Wagner, M. Kansy, F. Diederich, ChemMedChem 2006, 1, 611 –621;
d) M. Morgenthaler, J. D. Aebi, F. Grüninger, D. Mona, B. Wagner, M.
Kansy, F. Diederich, J. Fluorine Chem. 2008, 129, 852–865.
ham, P. M. O’Neill, Annu. Rev. Pharmacol. Toxicol. 2001, 41, 443–470;
d) H.-J. Bçhm, D. Banner, S. Bendels, M. Kansy, B. Kuhn, K. Müller, U.
Obst-Sander, M. Stahl, ChemBioChem 2004, 5, 637–643.
[20] ACD/Labs, Version 12.01, I. Advanced Chemistry Development Inc., Tor-
onto, ON (Canada), 2009.
[21] a) J. H. Williams, Acc. Chem. Res. 1993, 26, 593–598; b) F. Cozzi, F. Ponzi-
ni, R. Annunziata, M. Cinquini, J. S. Siegel, Angew. Chem. Int. Ed. 1995,
34, 1019–1020; Angew. Chem. 1995, 107, 1092–1094; c) G. W. Coates,
A. R. Dunn, L. M. Henling, D. A. Dougherty, R. H. Grubbs, Angew. Chem.
Int. Ed. Engl. 1997, 36, 248–251; Angew. Chem. 1997, 109, 290–293;
d) K. Kishikawa, Isr. J. Chem. 2012, 52, 800–808.
[22] H. Zheng, J. Gao, Angew. Chem. Int. Ed. 2010, 49, 8635–8639; Angew.
Chem. 2010, 122, 8817–8821.
[23] Molecular Operating Environment (MOE), 2013.08, Chemical Computing
Group Inc., 1010 Sherbooke St. West, Suite #910, Montreal, QC, H3A2R7
(Canada), 2016.
[24] B. Kuhn, J. E. Fuchs, M. Reutlinger, M. Stahl, N. R. Taylor, J. Chem. Inf.
Model. 2011, 51, 3180–3198.
[25] C. Kramer, J. E. Fuchs, S. Whitebread, P. Gedeck, K. R. Liedl, J. Med. Chem.
2014, 57, 3786–3802.
[26] The Supporting Information (SI) includes further analyses of the contri-
butions of fluorination of the benzylic S3 substituent to hCatL and RD
binding affinity, inhibition assays, small-molecule X-ray analysis, experi-
mental procedures, and NMR spectra of the new ligands.
[13] a) H. Hilpert, W. Guba, T. J. Woltering, W. Wostl, E. Pinard, H. Mauser,
A. V. Mayweg, M. Rogers-Evans, R. Humm, D. Krummenacher, T. Muser,
C. Schnider, H. Jacobsen, L. Ozmen, A. Bergadano, D. W. Banner, R.
Hochstrasser, A. Kuglstatter, P. David-Pierson, H. Fischer, A. Polara, R.
Narquizian, J. Med. Chem. 2013, 56, 3980–3995; b) I. Hyohdoh, N. Furui-
chi, T. Aoki, Y. Itezono, H. Shirai, S. Ozawa, F. Watanabe, M. Matsushita,
M. Sakaitani, P.-S. Ho, K. Takanashi, N. Harada, Y. Tomii, K. Yoshinari, K.
Ori, M. Tabo, Y. Aoki, N. Shimma, H. Iikura, ACS Med. Chem. Lett. 2013, 4,
1059–1063; c) Y. Lou, Z. K. Sweeney, A. Kuglstatter, D. Davis, D. M. Gold-
stein, X. Han, J. Hong, B. Kocer, R. K. Kondru, R. Litman, J. McIntosh, K.
Sarma, J. Suh, J. Taygerly, T. D. Owens, Bioorg. Med. Chem. Lett. 2015, 25,
367–371.
[14] M. Harder, B. Kuhn, F. Diederich, ChemMedChem 2013, 8, 397–404.
[15] M. Harder, M. A. Carnero Corrales, N. Trapp, B. Kuhn, F. Diederich, Chem.
Eur. J. 2015, 21, 8455–8463.
[16] a) J. Hernµndez-Trujillo, A. Vela, J. Phys. Chem. 1996, 100, 6524–6530;
b) J. R. Trudell, Biophys. Chem. 1998, 73, 7–11; c) K. Shimizu, M. F. Costa
Gomes, A. A. H. Pµdua, L. P. N. Rebelo, J. N. Canongia Lopes, J. Phys.
Chem. B 2009, 113, 9894–9900.
[17] P. R. Gerber, K. Müller, J. Comput.-Aided Mol. Des. 1995, 9, 251–268.
[18] S. Ludewig, M. Kossner, M. Schiller, K. Baumann, T. Schirmeister, Curr.
Top. Med. Chem. 2010, 10, 368–382.
Received: March 3, 2016
Revised: March 23, 2016
[19] a) D. O’Hagan, H. S. Rzepa, Chem. Commun. 1997, 645–652; b) B. E.
Smart, J. Fluorine Chem. 2001, 109, 3–11; c) B. K. Park, N. R. Kittering-
Published online on April 20, 2016
ChemMedChem 2016, 11, 1042 – 1047
www.chemmedchem.org
1047
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim