Molecular recognition at the active site of factor Xa: Cation-π Interactions, stacking on planar peptide surfaces, and replacement of structural water

Article abstract of DOI:10.1002/chem.201102571

Factor Xa, a serine protease from the blood coagulation cascade, is an ideal enzyme for molecular recognition studies, as its active site is highly shape-persistent and features distinct, concave sub-pockets. We developed a family of non-peptidic, small-molecule inhibitors with a central tricyclic core orienting a neutral heterocyclic substituent into the S1 pocket and a quaternary ammonium ion into the aromatic box in the S4 pocket. The substituents were systematically varied to investigate cation-π interactions in the S4 pocket, optimal heterocyclic stacking on the flat peptide walls lining the S1 pocket, and potential water replacements in both the S1 and the S4 pockets. Structure-activity relationships were established to reveal and quantify contributions to the binding free enthalpy, resulting from single-atom replacements or positional changes in the ligands. A series of high-affinity ligands with inhibitory constants down to K_i=2nM were obtained and their proposed binding geometries confirmed by X-ray co-crystal structures of protein-ligand complexes. Factor Xa is an ideal enzyme to undertake molecular recognition studies at atomic level resolution as its active site is completely conserved in complexes with designed ligands. Cation-π interactions, water replacements, and stacking interactions with flat peptide fragments were investigated, revealing large changes in binding affinity resulting from single-atom mutations or positional shifts of heteroatoms in the ligands. Copyright

Full text of DOI:10.1002/chem.201102571

FULL PAPER
DOI: 10.1002/chem.201102571
Molecular Recognition at the Active Site of Factor Xa: Cation–p Interactions,
Stacking on Planar Peptide Surfaces, and Replacement of Structural Water
Laura M. Salonen,[a] Mareike C. Holland,[a] Philip S. J. Kaib,[a] Wolfgang Haap,*[b]
Jçrg Benz,[b] Jean-Luc Mary,[b] Olivier Kuster,[b] W. Bernd Schweizer,[a]
David W. Banner,[b] and FranÅois Diederich*[a]
Abstract: Factor Xa, a serine protease
from the blood coagulation cascade, is
an ideal enzyme for molecular recogni-
tion studies, as its active site is highly
shape-persistent and features distinct,
concave sub-pockets. We developed a
family of non-peptidic, small-molecule
inhibitors with a central tricyclic core
orienting a neutral heterocyclic sub-
stituent into the S1 pocket and a qua-
ternary ammonium ion into the aro-
matic box in the S4 pocket. The sub-
stituents were systematically varied to
investigate cation–p interactions in the
S4 pocket, optimal heterocyclic stack-
ing on the flat peptide walls lining the
S1 pocket, and potential water replace-
ments in both the S1 and the S4 pock-
ets. Structure–activity relationships
were established to reveal and quantify
contributions to the binding free en-
thalpy, resulting from single-atom re-
placements or positional changes in the
ligands. A series of high-affinity ligands
with inhibitory constants down to Ki =
2 nm were obtained and their proposed
binding geometries confirmed by X-ray
co-crystal structures of protein–ligand
complexes.
Keywords: cation–p interactions ·
enzyme inhibitors
ligand conformations · molecular
recognition
· factor Xa ·
AHCTUNGTRENNUNG
Introduction
Its active site is highly shape-persistent and features distinct,
concave sub-pockets. As a central enzyme in the blood
coagulation cascade, factor Xa cleaves prothrombin to form
thrombin initiating the common pathway, which ultimately
leads to the formation of fibrin clots. The pharmaceutical in-
dustry has undertaken large efforts in the development of
direct factor Xa inhibitors as new antithrombotic agents,[3]
and the first drug rivaroxaban (Xarelto)[3f] has received
market approval.
Factor Xa inhibitors are typically L-shaped, with a central
core orienting the virtually orthogonal S1 and S4 vectors to
their respective pockets (Figure 1). The active site of factor
Xa is located mainly on the surface of the enzyme, except
for the S1 binding pocket, which reaches towards the core
with Asp189 and Tyr228 at the bottom. This pocket was
We decipher and quantify protein–ligand interactions at the
level of individual atoms in an approach comprising struc-
ture-based design and synthesis of non-peptidic, small-mole-
cule enzyme inhibitors, analysis of their biological affinities,
and X-ray crystallography of protein–ligand co-crystal struc-
tures. The resulting insight into molecular recognition pro-
cesses is not only useful in the development and optimiza-
tion of active compounds in the pharmaceutical and crop
protection industry, but also in many other areas of chemical
research, such as supramolecular chemistry and catalyst
development.
Previously, we identified the enzyme factor Xa, a serine
protease from the human blood coagulation cascade,[1] as an
excellent model system for molecular recognition studies.[2]
[a] L. M. Salonen, M. C. Holland, P. S. J. Kaib, Dr. W. B. Schweizer,
Prof. Dr. F. Diederich
Laboratorium fꢀr Organische Chemie
ETH Zꢀrich
Wolfgang-Pauli-Strasse 10, HCI, 8093 Zꢀrich (Switzerland)
Fax : (+41) 44-632-1109
E-mail: diederich@org.chem.ethz.ch
[b] Dr. W. Haap, Dr. J. Benz, J.-L. Mary, O. Kuster, Dr. D. W. Banner
Pharma Division
Prꢁklinische Forschung
F. Hoffmann-La Roche AG
4070 Basel (Switzerland)
E-mail: wolfgang.haap@roche.com
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201102571.
Figure 1. Schematic representation of the active site of factor Xa and
ligand design.
Chem. Eur. J. 2012, 18, 213 – 222
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
213

filled in earlier inhibitor development with cationic phenyl-
ACHTUNGTRENNUNGamidinium substituents interacting with the anionic side
We first undertook the optical resolution of (ꢁ)-1 to
verify which of the (+)- or (ꢀ)-enantiomers corresponds to
the 3aS,4R,8aS,8bR-configured one bound at the active site
of factor Xa in the X-ray co-crystal structure of (ꢁ)-1, and
to determine the degree of enantioselectivity of the binding
process. The endo,trans-configured[2] racemate (ꢁ)-3 is one
of four possible diastereoisomeric pairs of enantiomers
formed in the 1,3-dipolar cycloaddition between malein-
chain of Asp189, thereby mimicking the interaction of
Asp189 with the side chain of Arg in the complex of the nat-
ural substrate prothrombin.[3d] Currently, the more favored
approach is targeting the S1 pocket with halogenated arenes
and heteroarenes, with the halide substituent approaching
the side chain of Tyr228 at the bottom of the pocket, which
ultimately enabled the discovery of drug-like compounds.[3]
The other main binding pocket, the S4 pocket, is shaped
by the side chains of Tyr99, Phe174, and Trp215. This
“aromatic box” is an attractive site to study cation–p
interactions.[4] One of our earlier inhibitors featuring a tricy-
clic core similar to the one in the new ligands, binding to the
S4 pocket with a quaternary ammonium moiety (Me3N+-,
see Figure 1SI in the Supporting Information) and to the S1
pocket with a phenylamidinium substituent (“needle“), was
active against factor Xa with an inhibitory constant Ki =
280 nm, whereas its neutral tert-butyl (Me3C-) counterpart
binds only weakly (Ki =29 mm). This single-atom replace-
ment (N+/C) allowed the quantification of cation–p interac-
tions in the S4 pocket as DDGC!N+ =ꢀ2.8 kcalmolꢀ1.[2a]
When the phenylamidinium needle was replaced with an
isoxazolyl-chlorothienyl moiety ((ꢁ)-1, Scheme 1),[2b] the
onium ion ligand was bound much more strongly with Ki =
9 nm, whereas the tert-butyl derivative (ꢁ)-2 gave Ki =
550 nm (DDGC!N+ =ꢀ2.5 kcalmolꢀ1, about ꢀ0.8 kcalmolꢀ1
per aromatic ring lining the aromatic box).
AHCTUNGTRENNUNG
imide 4, l-proline (5), and aldehyde 6 (Scheme 1).[6] It can
be isolated by rather tedious column chromatography
(SiO2).
Here, we provide further insight into the strength of
cation–p interactions in the S4 pocket at the active site of
factor Xa. Additionally, we identify a chain of conserved
water molecules at the back of the S4 pocket, as seen in the
X-ray co-crystal structures, and present our first approaches
to replace these by ligand parts. In the S1 pocket, we probe
the stacking interactions of the ligand with the flat peptide
backbone segments lining the pocket and address the role of
the chlorine substituent approaching the side chain of
Tyr228.
Scheme 1. Synthesis and biological activities of new inhibitors of factor
Xa. Compounds (ꢁ)-1 and (ꢁ)-2 have been previously reported.[2b]
a) MeCN, 808C, 24 h, 4%; b) Me3N (4.2m in EtOH), EtOH, RT, 46 h,
93%. Inhibitory constants are averages of 1–3 duplicate measurements.
Enantiomeric resolution of (ꢁ)-3 was achieved by HPLC
on a chiral stationary phase (Chiralpak AD), and the enan-
tiomers were subsequently converted in excellent yield into
the trimethylammonium derivatives (+)-1 and (ꢀ)-1 by re-
action with Me3N. The enantiomers were tested for their
biological activity towards factor Xa in a spectrophotometric
assay,[7] and for selectivity against thrombin[8] (for descrip-
tion of the assays and thrombin binding data, see the Sup-
porting Information). Enantiomer (+)-1 (Ki =5 nm) has an
approximately 50-fold higher affinity towards factor Xa than
(ꢀ)-1 (Ki =271 nm), and consequently, we can unambiguous-
ly assign the absolute configuration of (+)-1 as
3aS,4R,8aS,8bR.
Results and Discussion
Cation–p interactions: We previously reported the X-ray co-
crystal structure of (ꢁ)-1 with factor Xa (PDB code: 2JKH,
1.25 ꢃ resolution; Figure 1SI),[2b] which shows the
3aS,4R,8aS,8bR-configured enantiomer bound at the active
site as predicted by molecular modeling (Moloc).[5] The qua-
ternary ammonium ion sits in the center of the aromatic box
in a very similar manner to that seen in the X-ray co-crystal
structure of the phenylamidinium inhibitor with factor Xa
(PDB code: 2BOK, 1.64 ꢃ resolution; Figure 1SI).[2a] How-
ever, the position of the tricyclic core of the inhibitor
changes drastically shifting from the back (phenylamidini-
um) to the front ((ꢁ)-1) of the active site (Figure 1SI). The
heterobiaryl needle in (ꢁ)-1 undergoes efficient p-stacking
interactions with the highly polar walls of the S1 pocket,
and the Cl atom is directed towards Tyr228 (see below).
Next, we prepared the trimethylphosphonium derivative
(ꢁ)-7 by treating (ꢁ)-3 with a solution of Me3P in THF. We
found that (ꢁ)-7 binds to factor Xa with Ki =7 nm
AHCTUNGTREG(NNNU Scheme 1), in a similar range of activity to trimethylammo-
nium derivative (ꢁ)-1. Thus, the change in the nature and
the size of the cation does not significantly affect the bind-
ing affinity. Such results have been reported in the literature
for resorcinarene and pyrogallolarene cavitands, where the
214
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 213 – 222

Molecular Recognition at the Active Site of Factor Xa
FULL PAPER
larger tetramethylphosphonium cation (115 ꢃ3) was bound
with Me3N. The inhibitory constant for iodide (ꢁ)-14 was
determined to be Ki =18 nm, in a similar range to that of
bromide (ꢁ)-1. Thus, in our system, exchanging the counter-
anion seems to have a negligible influence on the strength
of the cation–p interaction under the conditions of the assay
(buffer solution at pH 7.8). As the aromatic box is open to-
wards the surface of the protein, we presume that the onium
salts bind as solvent-separated ion pairs.
with
a similar association constant to that of the
tetramethyl
ACHTUNGTRENNUNG
ammonium guest (105 ꢃ3) in the aromatic
cavity.[9]
In the previous study,[2b] we prepared primary, secondary,
and tertiary amines (ꢁ)-8–(ꢁ)-10 (Figure 2), which are fully
protonated under the conditions of the biological assay at
pH 7.8, and found the degree of methylation to have a dra-
Water replacement in the S1
pocket: In the X-ray co-crystal
structure of (ꢁ)-1 bound to
factor Xa, the Cl atom on the
thienyl ring is directed towards
Tyr228 in the S1 pocket
(d(Cl···C-OH)=3.6 ꢃ,
angle
(Cl···C-O)=828) and undergoes
van der Waals interactions with
the surrounding amino acids
Ala190, Val213, Gly226, and
Figure 2. Biological activities of the ammonium ion series (ꢁ)-8–(ꢁ)-10, which are all fully protonated under
the conditions of the assay at pH 7.8, quaternary ammonium (ꢁ)-1, and control compounds (ꢁ)-11–(ꢁ)-13 and Ile227 (d
ACHTUNGRETN(NUNG Cl···C/N)=3.5–3.8 ꢃ)
(ꢁ)-2. All inhibitory constants are averages of 1–3 duplicate measurements. Due to solubility problems in the
case of the neutral (ꢁ)-11–(ꢁ)-13 and (ꢁ)-2, the measurements have large uncertainties.
(Figure 2SI). In a search of the
Protein Data Bank (PDB)
using Relibase,[11] we found the
location of the Cl atom and its
matic effect on the strength of the cation–p interaction, with
a stepwise energetic gain (ꢀDDG) of 1.2–1.8 kcalmolꢀ1 per
methylation. We now synthesized the neutral, alkyl-bearing
ligands (ꢁ)-11-(ꢁ)-13 to function, in addition to (ꢁ)-2,[2b] as
control compounds (Figure 2; for synthetic procedures, see
the Supporting Information). It should be noted that the
compounds in this neutral series have low water-solubility,
which results in substantial uncertainties in their experimen-
tally determined binding activities.
An enormous loss in binding affinity was observed when
the S4 vector was entirely omitted: N-methyl ligand (ꢁ)-11
binds with Ki =10.3 mm, showing that occupying the S4
pocket is crucial to maintain the high activity of the inhibi-
tors. Activity increases by a factor of ten with the incorpora-
tion of an n-pentyl vector as in (ꢁ)-12 (Ki =1.4 mm). Slight
enhancements in affinity are observed with additional
methyl groups, with the 4-methylpentyl-bearing ligand
(ꢁ)-13 binding with Ki =0.75 mm, and finally the 4,4-dime-
thylpentyl (“tert-butyl“) derivative (ꢁ)-2 with Ki =0.55 mm.
The binding affinity is enhanced by approximately a factor
of three when moving from the n-pentyl to the 4,4-dimethyl-
pentyl derivative. In sharp contrast, binding affinity increas-
es by a factor of 1000 when moving from the primary (Ki =
9.8 mm) to the quaternary ammonium ion inhibitor (Ki =
9 nm), corroborating the increasingly strong cation–p inter-
action in that series (Figure 2).
approach angle to the phenol ring to be highly conserved
among the X-ray co-crystal structures of factor Xa inhibi-
tors.[2b,12] We believe that the position of the Cl atom in the
co-crystal structures of factor Xa ligands with terminally
chlorinated S1 needles reflects favorable dipolar interactions
with the phenolic C–O dipole[13] while avoiding close Cl···p
contacts.[12] Remarkably, the specific position of the Cl atom
closely resembles that of a structural water molecule seen in
both the co-crystal structure of the earlier phenylamidinium
inhibitor (PDB code: 2BOK; for overlay, see Figure 1SI)[14]
and the X-ray crystal structure of factor Xa without a bound
ligand (resolution 2.2 ꢃ, PDB code: 1HCG).[1] In the phe-
nylamidinium structure, this water molecule undergoes dipo-
lar interactions with the C–O dipole of the phenolic ring of
Tyr228 (d
hydrogen bonding to Ile227 (d
to the phenylamidinium S1 needle of the inhibitor (d-
(N···(H)OH)=3.1 ꢃ). We therefore propose that the Cl sub-
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
stituent in (ꢁ)-1 displaces this water[15] and substitutes the
dipolar interaction with the phenolic C–O(H) group, while
undergoing favorable van der Waals interactions with the
surrounding lipophilic amino acid residues. The replacement
of conserved water molecules seen in X-ray co-crystal struc-
tures of protein–ligand complexes is of substantial interest
in contemporary structure-based drug design.[16]
We prepared a series of compounds with different thienyl
substituents ((ꢁ)-15-(ꢁ)-19, Figure 3, for synthesis, see the
Supporting Information) and studied their biological activi-
ty.[17] The Cl atom indeed makes a large contribution to the
binding free enthalpy of (ꢁ)-1. Substitution of Cl by H
((ꢁ)-15) increases the inhibitory constant by a factor of 70
to Ki =612 nm, corresponding to a loss of binding free
Counteranion effects on cation–p interactions have been
investigated with model systems in organic solvents.[10] To
address potential counteranion effects in the binding of the
quaternary ammonium ion inhibitors to factor Xa, we pre-
pared iodide (ꢁ)-14 (Scheme 1) by a Finkelstein reaction of
(ꢁ)-3 with NaI and subsequent treatment of the product
Chem. Eur. J. 2012, 18, 213 – 222
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
215

F. Diederich, W. Haap et al.
Figure 3. Influence of thienyl ring substituents on binding affinity. Inhibi-
tory constants are averages of 1–3 duplicate measurements.
Figure 4. Influence of heteroatom positions in the S1 needle on binding
affinity. Inhibitory constants are averages of 2–3 duplicate measurements.
enthalpy of DDG=2.6 kcalmolꢀ1, in agreement with reports
in the literature.[12] While the brominated ligand (ꢁ)-18
(Ki =6 nm) is of similar potency to (ꢁ)-1, the iodine sub-
stituent in (ꢁ)-19 seems to be too large and affinity drops
by nearly an order of magnitude (Ki =47 nm). The same re-
duction in binding is measured for fluorinated (ꢁ)-17 (Ki =
48 nm), presumably as a result of the unfavorable proximity
of the F atom to the phenolic p-surface of Tyr228,[18] and
less favorable van der Waals contacts due to the smaller size
of the F atom compared to the larger organohalogens.[19]
These unfavorable effects seem to overcompensate the
stronger dipolar interaction with the phenolic C–O, expected
complex with factor Xa (Figure 5; PDB code: 2Y5G for
(ꢁ)-20 at 1.33 ꢃ resolution and 2Y5H for (ꢁ)-21 at 1.29 ꢃ
resolution). An overlay of the three co-crystal structures of
(ꢁ)-1, (ꢁ)-20, and (ꢁ)-21 is shown in Figure 6. The active
ꢀ
for the C F bond. A large loss in affinity is observed for the
methoxy-substituted ligands ((+)-16: Ki =509 nm, (ꢀ)-16:
Ki =3540 nm). The methoxy group is presumably unable to
interact with Tyr228 in a favorable geometry, thus resulting
in a loss of binding affinity. Additionally, a higher desolva-
tion penalty as compared to the iodinated analogue, for ex-
ample, may play a role. However, a different binding mode,
with the MeO substituent pointing to Asp189 at the bottom
of the S1 pocket, as reported in the literature,[3a] cannot be
excluded in this case.
Stacking on planar peptide backbone fragments in the S1
pocket: Backbone planar peptides line the walls of the
narrow S1 pocket, Ala190-Cys191-Gln192 on one side, and
Trp215-Gly216 on the other (Figure 1). The X-ray co-crystal
structure of (ꢁ)-1 with factor Xa showed that these back-
bone surfaces undergo stacking interactions at van der
Waals distance with the isoxazolyl-chlorothienyl needle (see
Figure 3SI).[2b] To investigate these interactions in greater
detail, we alternated the positions of the heteroatoms on the
isoxazole ring of (ꢁ)-1 and prepared the oxazole-containing
ligands (ꢁ)-20 and (ꢁ)-21 (Figure 4; for synthesis, see the
Supporting Information). The compounds were tested for
their biological activity and, to our great surprise, were
found to be much weaker factor Xa inhibitors than (ꢁ)-1,
with a Ki value of 146 nm for (ꢁ)-20 and Ki =1620 nm for
(ꢁ)-21.
Figure 5. X-ray co-crystal structures of inhibitors a) (ꢁ)-20 (turquoise)
and b) (ꢁ)-21 (magenta) bound at the active site of factor Xa (PDB
codes: 2Y5G and 2Y5H, resolutions 1.33 ꢃ and 1.29 ꢃ, respectively).
Distances are shown in ꢃ. Color code: gray Cenzyme, turquoise C(ꢁ)-20, ma-
genta C(ꢁ)-21, red O, blue N, yellow S, lime Cl.
To clarify the reasons for this large loss in activity when
replacing the isoxazole with an oxazole ring, we obtained
X-ray co-crystal structures of both (ꢁ)-20 and (ꢁ)-21 in
216
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 213 – 222

Molecular Recognition at the Active Site of Factor Xa
FULL PAPER
to the isoxazole ring of (ꢁ)-1. This flip apparently induces
the tricyclic core to shift to the back of the active site, with
its location now resembling more closely that of the earlier
phenylamidinium inhibitor bound to factor Xa (PDB code:
2BOK, see Figure 1SI). Significant differences in the con-
tacts of the tricyclic core with the protein are, however, not
apparent in the different geometries, as the core resides in a
spacious part of the active site, open to the surface.
The strongly reduced affinity and the different conforma-
tion of bound (ꢁ)-20 and (ꢁ)-21, as compared to (ꢁ)-1,
must originate from the interplay of several factors, which
taken alone cannot fully explain the experimental findings.
i) Gas-phase conformational analysis (DFT B3LYP/6-
311GACTHNUTRGENU(GN d,p)) of the free ligands showed that in the global
minimum of free (ꢁ)-1, the isoxazole ring adopts the
same alignment with respect to the tricyclic core as seen
in the X-ray co-crystal structure. The energetic mini-
mum with a flipped geometry is found higher in energy
by up to 2.9 kcalmolꢀ1, presumably due to lone-pair re-
pulsion between the tricyclic amine center and the isox-
azole N atom. In the medium-strength ligand (ꢁ)-20,
minima for both orientations of the oxazole ring are ob-
served within 1 kcalmolꢀ1. The analysis of the weakly
binding compound (ꢁ)-21 shows a clear preference for
the flipped geometry, as the lone pair of the tricyclic
amine N atom interacts favorably with a C–H moiety of
the oxazole ring. The flipping of the oxazole ring as
compared to the isoxazole ring is, therefore, partially a
result of the intrinsic conformational preferences of li-
gands (ꢁ)-20 and (ꢁ)-21. When the various energetic
minima geometries of the three ligands, identified in the
conformational search, were docked into the active site,
the key binding interactions of both the onium ion and
the chlorothienyl moiety were only established properly
when the conformers resembling those seen in the X-
ray co-crystal structures were docked in and positioned
as seen experimentally. In other words, the tricyclic core
in (ꢁ)-20 and (ꢁ)-21 apparently needs to change posi-
tion to efficiently establish these key contacts.
Figure 6. Overlay of inhibitors (ꢁ)-1 (green), (ꢁ)-20 (turquoise), and
(ꢁ)-21 (magenta) bound at the active site of factor Xa (PDB codes:
2JKH, 2Y5G, and 2Y5H, respectively). Only the enzyme of 2JKH is
shown. a) View onto the entire inhibitor. b) View from the top on the tri-
cyclic core. Color code: gray Cenzyme, green C(ꢁ)-1, turquoise C(ꢁ)-20, magen-
ta C(ꢁ)-21, red O, blue N, yellow S, lime Cl.
site of the enzyme is completely conserved in all three struc-
tures (Figure 4SI). In all three complexes, the key binding
interactions are fully established: the onium ion is located in
the center of the S4 pocket, and the chlorothienyl needle
undergoes favorable interactions at the bottom of the S1
pocket. The van der Waals contacts between the Cl atom
and the surrounding amino acids are in the same range in
each case. The thiophene ring stacks at a similar distance
onto Gly216-Trp215 in all three structures (see Figure 3SI,
6SI, 7SI) and the Cl atom is positioned in a similar manner
(d(Cl···C-O=3.6 ꢃ, 3.6 ꢃ, and 3.9 ꢃ and angle Cl···C-O=
828, 868, and 868 for (ꢁ)-1, (ꢁ)-20, and (ꢁ)-21, respectively).
The small displacement of the Cl atom in (ꢁ)-21 away from
Tyr228 can account for some of the loss in activity as sug-
gested in the literature,[20] but this is clearly not the sole
reason.
ii) Another relevant matter may be the surprisingly differ-
ent angle between the two bonds departing from the
isox
N
and (ꢁ)-21. A CSD search (Figure 8SI) showed that the
angle in 3,5-disubstituted isoxazoles varies from 1538 to
1628, with a predominance around 1578. In contrast, the
angles in the two oxazoles are much smaller. The angle
in 2,4-disubstituted oxazoles, as in (ꢁ)-20, varies be-
tween 1278 and 1378, with a maximum of occurrences
around 1308, whereas in 2,5-disubstituted oxazoles as in
(ꢁ)-21 it varies between 1328 and 1528, with a prefer-
ence for 145–1498. In the co-crystal structures, the angle
for the isoxazole ring in (ꢁ)-1 is 1478, 1338 for the oxa-
zole ring in (ꢁ)-20, and 1478 for the oxazole ring in (ꢁ)-
21.
Figure 6 shows that the overall binding conformation of
the inhibitors is different for (ꢁ)-20 and (ꢁ)-21 as compared
to (ꢁ)-1: the oxazole ring in both ligands has flipped relative
iii) The oxazole rings in (ꢁ)-20 and (ꢁ)-21 are nearly co-
planar with the neighboring thiophene ring, whereas the
Chem. Eur. J. 2012, 18, 213 – 222
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
217

F. Diederich, W. Haap et al.
isoxazole ring of (ꢁ)-1 is twisted by approximately 308
out of this plane (Figure 5SI). In the structure of (ꢁ)-21,
a repulsive electrostatic contact between the O atom of
the oxazole ring and the carbonyl O atom of Gly218 is
established, (d(O···O=C–Gly218)=3.5 ꢃ, Figure 5b).
As a result of the twist out of the thienyl plane, the isox-
of one or more of these water molecules, a methyl group of
the onium ion moiety of inhibitor (ꢁ)-1 was exchanged for a
H-bond-donating substituent (Figure 8). We envisioned a
hydroxyl or amido group as first target vectors, and accord-
ing to modeling with Moloc,[5] the hydrogen-bond donor of
the inhibitors extends to the back of the aromatic box re-
placing a water molecule.
azole ring of (ꢁ)-1 (Figure 3SI) establishes more favora-
ble contacts to Gln192 than the oxazole rings in (ꢁ)-20
(Figure 6SI) or (ꢁ)-21 (Figure 7SI), with closest contacts
between heavy atoms of d((is)oxazole···Gln192)=3.3
((ꢁ)-1), 3.8 ((ꢁ)-20), and 3.7 ((ꢁ)-21) ꢃ, respectively.
iv) The positions of positively (d+) and negatively (dꢀ) po-
larized atoms in the heterocycles are shifted when re-
placing the isoxazole ring with either of the oxazoles.
This may affect the stacking interaction between the
heterocycle and the flat polarizable peptide surfaces
lining the S1 pocket, with their distinct d+ and dꢀ cen-
ters. The importance of properly aligning centers of op-
posite polarity on stacking interactions has been clearly
established in model systems.[21] Examination of the co-
crystal structures shows that complementary intermolec-
ular d+···dꢀ contacts are best established by the isoxa-
zole ring in (ꢁ)-1. This observation is certainly worth
validating in further systematic study involving hetero-
cyclic S1 needles with differently polarized frameworks.
Figure 8. Inhibitors with polar groups departing from the onium ion. All
inhibitory constants are averages of 1–4 duplicate measurements.
We prepared compounds (ꢁ)-22–(ꢁ)-25 bearing substitu-
ents for potential water replacement (Figure 8; for inhibitor
synthesis, see the Supporting Information). Ligand (ꢁ)-22
(Ki =26 nm) with an amide functionality is a weaker inhibi-
tor of factor Xa than (ꢁ)-1. Likewise, inhibitors (ꢁ)-23 and
(ꢁ)-24 (Ki =14 nm and 76 nm, respectively) have lower bind-
ing affinities, also compared to the corresponding, previous-
ly reported ethyldimethylammonium- and pyridinium-bear-
ing inhibitors, which lack the hydrogen bond-donating moi-
eties (Ki =8 and 30 nm, respectively).[2b] The loss of activity
may be partly attributed to higher desolvation costs, as the
inhibitors bearing the H-bond-donating residues are more
hydrophilic than the corresponding control compounds. The
X-ray crystal structure of the free ligand (ꢁ)-24 was solved
and revealed a homochiral dimeric assembly, in which the
two molecules interact by forming two halogen bonds[22] be-
tween the chloro substituents on the thiophene rings and
imide C=O groups (Figure 10SI).
Opportunities for water replacement at the back of the S4
pocket: When superimposing the X-ray co-crystal structures
of (ꢁ)-1, (ꢁ)-20, and (ꢁ)-21, we identified a conserved
chain of five water molecules located at the back of the S4
pocket (Figure 7). The water molecules are hydrogen-
bonded to each other and the surrounding amino acids
Ser173,
Ile175,
Glu97,
and
Thr98
(Figure 9SI,
d(O(H)···(H)O/N)=2.7–2.9 ꢃ). Aiming at the replacement
The N-(2-hydroxyethyl)pyrrolidinium derivative (ꢁ)-25 is
our most potent factor Xa inhibitor to date with Ki =2 nm.
The high binding affinity was initially thought to stem from
successful water replacement. We obtained the X-ray co-
crystal structure of (ꢁ)-25 in complex with factor Xa
(Figure 9, PDB code: 2Y5F, resolution 1.29 ꢃ), which shows
the 3aS,4R,8aS,8bR-configured enantiomer bound at the
active site in a binding mode resembling closely that of (ꢁ)-
1 (Figure 11SI). However, instead of reaching to the back of
the S4 pocket to replace the first of the five water molecules
of the chain, which remain conserved, the hydroxyethyl
moiety is directed towards the surface of the protein and un-
dergoes water-mediated hydrogen-bonding interactions to
Figure 7. Conserved water molecules identified at the back of the S4
pocket. Only the enzyme of 2JKH is shown. Color code: gray Cenzyme
,
green C(ꢁ)-1 and water, turquoise C(ꢁ)-20 and water, magenta C(ꢁ)-21 and
water, red O, blue N.
218
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 213 – 222

Molecular Recognition at the Active Site of Factor Xa
FULL PAPER
to undergo favorable stacking interactions with the flat pep-
tide walls lining the pocket and to generate optimal inter-
molecular interactions between oppositely polarized
(d+ and dꢀ) centers. The change in the overall binding ge-
ometry from the complex of (ꢁ)-1 to the complexes of (ꢁ)-
20 and (ꢁ)-21 is quite remarkable when merely substituting
an isoxazole by oxazole rings. While replacements of one or
more of the five conserved water molecules at the back of
the S4 pocket by polar ligand parts have so far not been suc-
cessful, the crystallographic analysis has shown the way to
accomplish such replacements in future work. Factor Xa is
clearly an ideal enzyme for precise molecular recognition
studies at atomic level resolution, as its active site is com-
pletely conserved in all co-crystal structures discussed here,
a prerequisite for assigning incremental energetic changes to
single-atom mutations. Thus, this study allows gaining
knowledge that can ultimately be generally applied to struc-
ture-based drug and crop protection agent design, as well as
to many other areas of supramolecular chemistry.
Figure 9. X-ray co-crystal structure of (ꢁ)-25 bound at the active site of
factor Xa (PDB code: 2Y5F, resolution 1.29 ꢃ). Distances are shown in
ꢃ. Color code: gray Cenzyme, peach C(ꢁ)-25, red O, blue N, yellow S, lime
Cl.
the carbonyl of Glu97 (d
(H2O···(H)OCH2–(ꢁ)-25)=2.8 ꢃ;
Experimental Section
d(Glu97–C=O···(H)OH)=2.9 ꢃ). As a control, we prepared
the N-methylpyrrolidinium ligand (ꢁ)-26, and determined
its biological activity to be Ki =5 nm. Evidently, the hydro-
gen-bonding interactions of (ꢁ)-25 provide some additional
stabilization to the complex, compensating for the increased
desolvation costs due to the polar hydroxyethyl chain.
These results clearly show that the hydroxyethyl chain is
too flexible for a water-replacement vector and that confor-
mationally more rigidly positioned polar substituents, such
as those of spirocyclic quaternary ammonium ions, will be
required to accomplish the displacement of one or more of
the conserved water molecules in the chain at the back of
the S4 pocket.
X-ray co-crystal structure determinations: Crystals of short form factor
Xa (Arg150Glu mutant) were produced and data measured as previously
described,[15g] except that data were collected at the Swiss Light Source.
Data were processed with XDS[23] and SADABS (obtained from Bruker
AXS), and data reduction used the CCP4 package.[24] The structures
were solved by using the isomorphous 2jkh.pdb as starting model. Model
building of the ligand was performed with Moloc,[5] and structures were
refined with cycles of Coot[25] and Refmac5.[26] Coordinates have been de-
posited at the Protein Data Bank with the PDB codes: 2Y5G ((ꢁ)-20),
2Y5H ((ꢁ)-21), and 2Y5F ((ꢁ)-25). For more details, see the Supporting
Information.
General details and synthetic procedures: In the following, the experi-
mental details for the syntheses of compounds (+)-1 and (ꢀ)-1, (ꢁ)-7,
(ꢁ)-20, (ꢁ)-21, and (ꢁ)-22–(ꢁ)-25 are described. All other synthetic de-
tails and experimental data are given in full in the Supporting Informa-
tion. CC=column chromatography; RT=room temperature.
Conclusion
3-[(3aS,4R,8aS,8bR)-4-[5-(5-Chloro-2-thienyl)-1,2-oxazol-3-yl]-1,3-dioxo-
A
ACHTUNGTRNE(NUNG 3H)-yl]-N,N,N-trimethyl-1-pro-
AHCTUNGTRENNUNG
This study confirms that potent inhibitors need to properly
fill both the aromatic box in the S4 pocket and the S1
pocket of factor Xa. Quaternary ammonium ions are far su-
perior substituents to fill the aromatic box than comparable
alkyl substituents, with the single-atom (N+ ! C) mutation
in EtOH (1 mL) at RT, 4.2m Me3N in EtOH (200 mL, 0.83 mmol) was
added. The mixture was stirred at RT for 46 h and evaporated in vacuo.
The residue was dissolved in MeOH. Precipitation from Et2O gave (+)-1
(20 mg, 89%) as off-white crystals. M.p. 2148C (decomp); ½aꢂ2D5 : +141.2
1
(c=1, CH3OH); H NMR (600 MHz, CD3OD): d=1.80–1.89 (m, 2H, H-
C(7), H-C(8)), 2.01–2.09 (m, 2H, +NCH2CH2), 2.11–2.19 (m, 2H, H-
C(7), H-C(8)), 2.87 (ddd, J=12.7, 8.3, 4.3 Hz, 1H, H-C(6)), 2.97 (ddd,
J=12.6, 8.7, 6.9 Hz, 1H, H-C(6)), 3.15 (s, 9H, (CH3)3+N), 3.34–3.36 (m,
2H, +NCH2), 3.49 (dd, J=8.1, 1.5 Hz, 1H, H-C(8b)), 3.56 (qt, J=14.9,
6.5 Hz, 2H, +NCH2CH2CH2), 3.73–3.76 (m, 1H, H-C(8a)), 3.84 (t, J=
8.5 Hz, 1H, H-C(3a)), 4.43 (d, J=8.8 Hz, 1H, H-C(4)), 6.63 (s, 1H, Ar),
7.11 (d, J=4.0 Hz, 1H, Ar), 7.42 ppm (d, J=4.0 Hz, 1H, Ar); 13C NMR
(150 MHz, CD3OD): d=22.8, 24.2, 30.2, 36.7, 50.6, 50.8, 52.4, 53.6 (t),
62.7, 65.3 (t), 69.6, 101.4, 128.1, 129.0, 129.1, 134.2, 164.8, 165.1, 177.4,
179.9 ppm; IR (ATR): n˜ =3386 (b), 2957, 1772, 1694, 1614, 1524, 1476,
1423, 1402, 1351, 1319, 1283, 1183, 1146, 1067, 1030, 1000, 965, 923, 878,
strongly reducing binding affinity by up to DDGN+!C
=
2.5 kcalmolꢀ1 as a result of the loss of cation–p interactions.
Optimal filling of the S1 pocket is challenging. The termi-
nal chlorine substituent on the S1 needle must be properly
positioned to undergo dipolar interactions with the phenol
C–O(H) of Tyr228, to establish van der Waals contacts with
neighboring lipophilic amino acid side chains, and to dis-
place a water molecule interacting in a similar position with
the phenolic ring in the absence of ligand.[1] The single-atom
mutation on the thiophene ring (Cl ! H) reduces binding
affinity by DDGCl!H =2.6 kcalmolꢀ1. In addition, the hetero-
cyclic ring linking the tricyclic ligand core with the chloro-
thienyl moiety requires proper positioning in the S1 pocket
793, 751, 666, 628 cmꢀ1
; HR-MALDI-MS: m/z (%): 465.1536 (42),
[MꢀBr]+, calcd for C22H2837ClN4O3S+: 465.1537, 463.1570 (100),
[MꢀBr]+, calcd for C22H2835ClN4O3S+: 463.1565.
3-[(3aR,4S,8aR,8bS)-4-[5-(5-Chloro-2-thienyl)-1,2-oxazol-3-yl]-1,3-dioxo-
ACHUTNGRENUNoG ctahydrodipyrrolo[1,2-a:3’,4’-c]pyrrol-2ACHTUNTGREN(NUGN 3H)-yl]-N,N,N-trimethyl-1-pro-
Chem. Eur. J. 2012, 18, 213 – 222
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
219

F. Diederich, W. Haap et al.
A
N,N,N-Trimethyl-3-[(3aRS,4RS,8aSR,8bSR)-4-[2-(5-Chloro-2-thienyl)-
1,3-oxazol-4-yl]-1,3-dioxooctahydrodipyrrolo[1,2-a:3’,4’-c]pyrrol-2ACHTUNGTRENNUNG(3H)-
in EtOH (1 mL) at RT, 4.2m Me3N in EtOH (200 mL, 0.83 mmol) was
added. The mixture was stirred at RT for 46 h and evaporated in vacuo.
The residue was dissolved in MeOH. Precipitation from Et2O gave (ꢀ)-1
(21 mg, 94%) as off-white crystals. M.p. 2148C (decomp); ½aꢂ2D5 : ꢀ137.2
yl]-1-propanaminium bromide ((ꢁ)-21): To (ꢁ)-28 (12 mg, 0.02 mmol) in
EtOH (0.25 mL) at RT, 4.2m Me3N in EtOH (120 mL, 0.5 mmol) was
added. The mixture was stirred for six days at RT and the solvent evapo-
rated in vacuo. The residue was dissolved in MeOH. Precipitation from
Et2O gave (ꢁ)-21 (20 mg, quant.) as a white solid. M.p. 1808C (decomp);
1H NMR (600 MHz, CDCl3): d=1.81–1.87 (m, 2H, H-C(7), H-C(8)),
1.99–2.08 (m, 2H, +NCH2CH2), 2.09–2.18 (m, 2H, H-C(7), H-C(8)),
2.85–2.90 (m, 1H, H-C(6)), 2.94–2.98 (m, 1H, H-C(6)), 3.09 (s, 9H,
(CH3)3+N), 3.33–3.36 (m, 2H, +NCH2), 3.45–3.49 (m, 2H, H-C(8b),
+NCH2CH2CH2), 3.56 (m, 1H, +NCH2CH2CH2), 3.79–3.82 (m, 2H, H-
C(3a), H-C(8a)), 4.28 (d, J=8.9 Hz, 1H, H-C(4)), 7.09 (d, J=4.0 Hz, 1H,
Ar), 7.51 (d, J=4.0 Hz, 1H, Ar), 7.79 ppm (d, J=0.7 Hz, 1H, Ar);
13C NMR (150 MHz, CDCl3): d=22.8, 24.3, 30.6, 36.8, 50.6, 51.1, 52.4,
53.6 (t), 62.6, 65.4 (t), 69.1, 128.9, 129.0, 129.4, 134.4, 137.9, 141.5, 158.0,
177.6, 180.1 ppm; IR (ATR): n˜ =3383 (b), 2959, 1774, 1694, 1593, 1478,
1431, 1403, 1351, 1319, 1241, 1182, 1066, 1026, 982, 929, 906, 877, 803,
725, 663 cmꢀ1; HR-MALDI-MS: m/z (%): 465.1526 (33), [MꢀBr]+, calcd
for C22H2837ClN4O3S+: 465.1541, 463.1564 (100), [MꢀBr]+, calcd for
C22H2835ClN4O3S+: 463.1565.
(c=1, CH3OH); H NMR (600 MHz, CD3OD): d=1.79–1.89 (m, 2H, H-
C(7), H-C(8)), 2.02–2.09 (m, 2H, +NCH2CH2), 2.11–2.19 (m, 2H, H-
C(7), H-C(8)), 2.87 (ddd, J=12.7, 8.4, 4.4 Hz, 1H, H-C(6)), 2.97 (ddd,
J=12.7, 8.7, 7.0 Hz, 1H, H-C(6)), 3.14 (s, 9H, (CH3)3+N), 3.33–3.36 (m,
2H, +NCH2), 3.48 (dd, J=8.1, 1.5 Hz, 1H, H-C(8b)), 3.56 (qt, J=15.0,
6.5 Hz, 2H, +NCH2CH2CH2), 3.73–3.76 (m, 1H, H-C(8a)), 3.84 (t, J=
8.5 Hz, 1H, H-C(3a)), 4.42 (d, J=8.8 Hz, 1H, H-C(4)), 6.63 (s, 1H, Ar),
7.10 (d, J=4.0 Hz, 1H, Ar), 7.42 ppm (d, J=3.9 Hz, 1H, Ar); 13C NMR
(150 MHz, CD3OD): d=22.8, 24.2, 30.2, 36.7, 50.7, 50.8, 52.4, 53.7 (t),
62.7, 65.3 (t), 69.6, 101.4, 128.1, 129.0, 129.1, 134.2, 164.9, 165.1, 177.5,
179.9 ppm; IR (ATR): n˜ =3385 (b), 2957, 1772, 1694, 1524, 1476, 1423,
1402, 1351, 1319, 1283, 1246, 1206, 1183, 1145, 1068, 1030, 1000, 965, 923,
898, 878, 794, 750, 665, 627 cmꢀ1; HR-MALDI-MS: m/z (%): 465.1528
(41), [MꢀBr]+, calcd for C22H2837ClN4O3S+: 465.1537, 463.1562 (100),
[MꢀBr]+, calcd for C22H2835ClN4O3S+: 463.1565.
{3-[(3aRS,4RS,8aSR,8bSR)-4-[5-(5-Chloro-2-thienyl)-3-isoxazolyl]-1,3-
dioxo-octahydrodipyrrolo[1,2-a:3’,4’-c]pyrrol-2ACTHNUGTRNEUNG(3H)-yl]propyl}-(trime-
N-(2-Amino-2-oxoethyl)-3-[(3aRS,4RS,8aSR,8bSR)-4-[5-(5-chloro-2-
thienyl)-3-isoxazolyl]-1,3-dioxooctahydrodipyrrolo[1,2-a:3’,4’-c]pyrrol-2-
thyl)phosphonium bromide ((ꢁ)-7): A mixture of 1m Me3P in THF
(1.24 mL, 1.24 mmol) and (ꢁ)-3 (30 mg, 0.06 mmol) was stirred at RT
under N2 for 16 h, treated with additional 1m Me3P in THF (1 mL,
1.00 mmol), and stirred for five days. The precipitate was filtered and
washed with THF. The residue was dissolved in MeOH. Precipitation
from Et2O gave (ꢁ)-7 (24 mg, 69%) as a white solid. M.p. 2178C
(decomp); 1H NMR (400 MHz, CD3OD): d=1.77–1.90 (m, 4H, H-C(7),
H-C(8), +PCH2CH2), 1.88 (d, 9H, J=14.5 Hz, (CH3)3P+) 2.10–2.25 (m,
4H, H-C(7), H-C(8), +PCH2), 2.86 (ddd, J=12.4, 8.1, 4.3 Hz, 1H, H-
C(6)), 2.97 (ddd, J=12.6, 8.6, 6.9 Hz, 1H, H-C(6)), 3.49 (dd, J=8.2,
1.5 Hz, 1H, H-C(8b)), 3.51–3.63 (m, 2H, +PCH2CH2CH2), 3.75 (br. t, J=
7.5 Hz, 1H, H-C(8a)), 3.84 (t, J=8.5 Hz, 1H, H-C(3a)), 4.42 (d, J=
8.8 Hz, 1H, H-C(4)), 6.62 (s, 1H, Ar), 7.11 (d, J=4.0 Hz, 1H, Ar),
7.42 ppm (d, J=4.0 Hz, 1H, Ar); 13C NMR (100 MHz, CD3OD): d=7.7
(d, 1JCP =55.1 Hz), 20.9 (d, 2JCP =3.7 Hz), 21.6 (d, 1JCP =54.1 Hz), 24.1,
30.2, 39.8 (d, 3JCP =18.4 Hz), 50.65, 50.71, 52.4, 62.7, 69.6, 101.3, 128.1,
128.98, 129.02, 134.1, 164.9, 165.0, 177.4, 180.0 ppm; 31P NMR (172 MHz,
CD3OD): d=28.0 ppm; IR (ATR): n˜ =2956, 1769, 1694, 1604, 1527, 1475,
1424, 1398, 1370, 1336, 1306, 1280, 1245, 1203, 1159, 1103, 1063, 1029,
990, 973, 901, 873, 798, 768, 756, 727, 682, 667, 654, 623 cmꢀ1; HR-
MALDI-MS: m/z (%): 482.1238 (39), [MꢀBr]+, calcd for
C22H2837ClN3O3PS+: 482.1248, 480.1272 (100), [MꢀBr]+, calcd for
C22H2835ClN3O3PS+: 480.1272.
(3H)-yl]-N,N-dimethyl-1-propanaminium bromide ((ꢁ)-22): To a solution
of (ꢁ)-3 (23 mg, 0.05 mmol) in EtOH (1.5 mL) at RT, 29 (63 mg,
0.62 mmol) was added. The mixture was stirred at RT for three days and
at 408C for four days. The mixture was evaporated in vacuo, dissolved in
MeOH, and precipitation from Et2O gave (ꢁ)-22 (20 mg, 71%) as an off-
white solid. M.p. 2078C (decomp); 1H NMR (600 MHz, CDCl3): d=
1.80–1.88 (m, 2H, H-C(7), H-C(8)), 2.02–2.09 (m, 2H, +NCH2CH2CH2),
2.11–2.19 (m, 2H, H-C(7), H-C(8)), 2.86 (ddd, J=12.7, 8.4, 4.3 Hz, 1H,
H-C(6)), 2.96 (ddd, J=12.6, 8.6, 7.0 Hz, 1H, H-C(6)), 3.28 (s, 3H, CH3),
3.30 (s, 3H, CH3), 3.49 (dd, J=8.1, 1.5 Hz, 1H, H-C(8b)), 3.52–3.63 (m,
4H, +NCH2CH2CH2, +NCH2CH2CH2), 3.74 (br. t, J=7.7 Hz, 1H, H-
C(8a)), 3.84 (t, J=8.4 Hz, 1H, H-C(3a)), 4.07 (s, 2H, N+CH2CO), 4.41
(d, J=8.7 Hz, 1H, H-C(4)), 6.64 (s, 1H, Ar), 7.11 (d, J=4.0 Hz, 1H, Ar),
7.43 ppm (d, J=4.0 Hz, 1H, Ar); 13C NMR (150 MHz, CDCl3): d=22.5,
24.2, 30.3, 36.6, 50.6, 50.8, 52.2, 52.5, 52.7, 62.5, 62.7, 64.3, 69.6, 101.2,
128.1, 129.0, 129.1, 134.2, 164.9, 165.1, 166.9, 177.5, 180.0 ppm; IR (ATR):
n˜ =3219, 3106, 2928, 1774, 1705, 1595, 1532, 1470, 1441, 1426, 1396, 1356,
1324, 1303, 1242, 1218, 1204, 1164, 1080, 1063, 1026, 997, 952, 907, 898,
871, 824, 789, 737, 667, 638, 609 cmꢀ1
; HR-MALDI-MS: m/z (%):
506.1631 (100), [MꢀBr]+, calcd for C23H2935ClN5O4S+: 506.1623.
3-[(3aSR,4RS,8aSR,8bRS)-4-[5-(5-Chloro-2-thienyl)-3-isoxazolyl]-1,3-di-
oxooctahydrodipyrrolo[1,2-a:3’,4’-c]pyrrol-2ACTHNUTRGNE(NUG 3H)-yl]-N-(2-hydroxyethyl)-
N,N-dimethyl-1-propanaminium bromide ((ꢁ)-23): To a solution of (ꢁ)-3
(30 mg, 0.06 mmol) in EtOH (1 mL) at RT, N,N-dimethylethanolamine
(125 mL, 1.24 mmol) was added and the mixture stirred at RT for 24 h
and at 408C for further 24 h. The mixture was evaporated in vacuo. The
residue was dissolved in MeOH. Precipitation from Et2O gave (ꢁ)-23
3-[(3aSR,4RS,8aSR,8bRS)-4-[5-(5-Chloro-2-thienyl)-1,3-oxazol-2-yl]-1,3-
dioxooctahydrodipyrrolo[1,2-a:3’,4’-c]pyrrol-2ACTHNUTRGNEUG(N 3H)-yl]-N,N,N-trimethyl-
1-propanaminium bromide ((ꢁ)-20): To a solution of (ꢁ)-27 (17 mg,
0.04 mmol) in EtOH (2 mL) at RT was added 4.2m Me3N in EtOH
(170 mL, 0.70 mmol). The mixture was stirred at RT for five days and
evaporated in vacuo. The residue was dissolved in MeOH. Precipitation
from Et2O gave (ꢁ)-20 (14 mg, 74%) as a yellow solid. M.p. 2238C
(decomp); 1H NMR (600 MHz, CDCl3): d=1.77–1.89 (m, 2H, H-C(7),
H-C(8)), 2.01–2.08 (m, 2H, +NCH2CH2), 2.10–2.19 (m, 2H, H-C(7), H-
C(8)), 2.87 (ddd, J=12.6, 8.2, 4.4 Hz, 1H, H-C(6)), 2.97–3.01 (m, 1H, H-
C(6)), 3.13 (s, 9H, (CH3)3+N), 3.33–3.37 (m, 2H, +NCH2), 3.49 (dd, J=
8.2, 1.5 Hz, 1H, H-C(8b)), 3.54–3.59 (m, 2H, +NCH2CH2CH2), 3.75 (ddd,
J=9.2, 7.3, 1.6 Hz, 1H, H-C(8a)), 3.91 (t, J=8.5 Hz, 1H, H-C(3a)), 4.47
(d, J=8.8 Hz, 1H, H-C(4)), 7.03 (d, J=4.0 Hz, 1H, Ar), 7.22 (d, J=
3.9 Hz, 1H, Ar), 7.30 ppm (s, 1H, Ar); 13C NMR (150 MHz, CDCl3): d=
22.7, 24.2, 30.4, 36.8, 50.5, 50.7, 52.4, 53.6 (t), 63.6, 65.2 (t), 69.6, 122.9,
125.4, 128.6, 129.5, 131.6, 147.9, 162.6, 177.5, 179.8 ppm; IR (ATR): n˜ =
2953, 1773, 1704, 1562, 1478, 1434, 1404, 1357, 1321, 1245, 1226, 1205,
1183, 1146, 1127, 1087, 1068, 1031, 960, 929, 884, 818, 777, 748, 668,
629 cmꢀ1; HR-MALDI-MS: m/z (%): 465.1531 (41), [MꢀBr]+, calcd for
C22H2837ClN4O3S+: 465.1537, 463.1567 (100), [MꢀBr]+, calcd for
C22H2835ClN4O3S+: 463.1565.
1
(26 mg, 73%) as a pink solid. M.p. 1808C (decomp); H NMR (600 MHz,
CD3OD): d=1.79–1.89 (m, 2H, H-C(7), H-C(8)), 2.00–2.07 (m, 1H,
+NCH2CH2CH2), 2.09–2.19 (m, 3H, H-C(7), H-C(8), +NCH2CH2CH2),
2.96 (ddd, J=12.7, 8.8, 7.0 Hz, 1H, H-C(6)), 2.87 (ddd, J=12.7, 8.3,
4.3 Hz, 1H, H-C(6)), 3.16 (s, 3H, CH3), 3.20 (s, 3H, CH3), 3.34–3.61 (m,
7H, H-C(8b), +NCH2CH2CH2, +NCH2CH2CH2, CH2CH2OH), 3.70–3.73
(m, 1H, H-C(8a)), 3.84 (t, J=8.4 Hz, 1H, H-C(3a)), 3.90–3.95 (m, 1H,
CH2OH), 4.12 (ddq, J=14.0, 7.2, 2.4 Hz, 1H, CH2OH), 4.44 (d, J=
8.8 Hz, 1H, H-C(4)), 6.66 (s, 1H, Ar), 7.11 (d, J=4.0 Hz, 1H, Ar),
7.44 ppm (d, J=3.8 Hz, 1H, Ar); 13C NMR (150 MHz, CD3OD): d=22.5,
24.1, 30.2, 36.6, 50.4, 50.7, 52.2, 52.5 (t), 52.6 (t), 56.8, 62.7, 63.9, 66.3,
69.5, 101.4, 128.1, 128.96, 129.02, 134.2, 164.7, 165.1, 177.4, 179.8 ppm; IR
(ATR): n˜ =3283, 3008, 2955, 2882, 1768, 1696, 1600, 1530, 1476, 1428,
1400, 1351, 1313, 1233, 1188, 1156, 1067, 1000, 963, 923, 895, 802, 742,
667, 632 cmꢀ1; HR-MALDI-MS: m/z (%): 495.1650 (37), [MꢀBr]+, calcd
for C23H3037ClN4O4S+: 495.1644, 493.1678 (100), [M-Br]+, calcd for
C23H3035ClN4O4S+: 493.1671; elemental analysis calcd (%) for
C23H30BrClN4O4S (572.09): C 48.13, H 5.27, N 9.76; found C 47.84, H
5.32, N 9.58.
220
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 213 – 222

Molecular Recognition at the Active Site of Factor Xa
FULL PAPER
1-{3-[(3aSR,4RS,8aSR,8bRS)-4-[5-(5-Chloro-2-thienyl)-3-isoxazolyl]-1,3-
2005, 117, 4474–4479; Angew. Chem. Int. Ed. 2005, 44, 4400–4404;
b) L. M. Salonen, C. Bucher, D. W. Banner, W. Haap, J.-L. Mary, J.
Benz, O. Kuster, P. Seiler, W. B. Schweizer, F. Diederich, Angew.
Chem. 2009, 121, 825–828; Angew. Chem. Int. Ed. 2009, 48, 811–
814.
dioxooctahydrodipyrrolo[1,2-a:3’,4’-c]pyrrol-2
ACHTUNGTREN(NUNG 3H)-yl]propyl}-3-
(hydroxy
ACHTUNGTRENNUNG
(30 mg, 0.06 mmol) in EtOH (1 mL) at RT, 3-(hydroxymethyl)pyridine
(120 mL, 1.24 mmol) was added and the mixture stirred at RT for 29 h
and further 67 h at 408C. The mixture was evaporated in vacuo and the
residue dissolved in MeOH. Precipitation from Et2O gave (ꢁ)-24 (35 mg,
95%) as yellow crystals. M.p. 1888C (decomp); 1H NMR (400 MHz,
CD3OD): d=1.80–1.94 (m, 2H, H-C(7), H-C(8)), 2.10–2.21 (m, 2H, H-
C(7), H-C(8)), 2.23–2.30 (m, 2H, +NCH2CH2), 2.87 (ddd, J=12.5, 8.2,
4.3 Hz, 1H, H-C(6)), 3.00 (ddd, J=12.5, 8.6, 6.9 Hz, 1H, H-C(6)), 3.46–
3.61 (m, 3H, H-C(8b), +NCH2CH2CH2), 3.79 (ddd, J=9.1, 7.3, 1.6 Hz,
1H, H-C(8a)), 3.88 (t, J=8.5 Hz, 1H, H-C(3a)), 4.45 (d, J=8.7 Hz, 1H,
H-C(4)), 4.63 (td, J=7.6, 1.2 Hz, 2H, +NCH2), 4.84 (s, 2H, CH2OH),
6.62 (s, 1H, Ar), 7.09 (d, J=4.0 Hz, 1H, Ar), 7.40 (d, J=4.0 Hz, 1H, Ar),
8.07 (dd, J=8.0, 6.1 Hz, 1H, pyridinium), 8.55 (d, J=8.1 Hz, 1H, pyridi-
nium), 8.92 (d, J=6.1 Hz, 1H, pyridinium), 8.99 ppm (s, 1H, pyridinium);
13C NMR (100 MHz, CD3OD): d=24.2, 30.4, 30.6, 36.4, 50.8, 52.4, 60.4,
61.1, 62.6, 69.5, 101.2, 128.1, 128.9, 129.0, 134.1, 143.9, 144.5, 144.6, 145.5,
164.9, 165.0, 177.5, 180.0 ppm (2 C not visible); IR (ATR): n˜ =3303, 3021,
2950, 2868, 1769, 1699, 1612, 1474, 1427, 1397, 1344, 1297, 1207, 1173,
1065, 998, 920, 878, 796, 687 cmꢀ1; HR-MALDI-MS: m/z (%): 515.1337
(43), [MꢀBr]+, calcd for C25H2637ClN4O4S+: 515.1332, 513.1359 (100),
[MꢀBr]+, calcd for C25H2635ClN4O4S+: 513.1358; elemental analysis calcd
(%) for C25H26BrClN4O4S (592.05): C 50.56, H 4.41, N 9.43; found: C
50.83, H 4.63, N 9.29.
[3] a) M. Nazarꢄ, D. W. Will, H. Matter, H. Schreuder, K. Ritter, M.
Urmann, M. Essrich, A. Bauer, M. Wagner, J. Czech, M. Lorenz, V.
Laux, V. Wehner, J. Med. Chem. 2005, 48, 4511–4525; b) S. Roehrig,
A. Straub, J. Pohlmann, T. Lampe, J. Pernerstorfer, K.-H. Schlemm-
er, P. Reinemer, E. Perzborn, J. Med. Chem. 2005, 48, 5900–5908;
c) M. Ieko, T. Tarumi, T. Nakabayashi, M. Yoshida, S. Naito, T.
Koike, Front. Biosci. 2006, 11, 232–248; d) L. Anselm, D. W.
Banner, J. Benz, K. Groebke Zbinden, J. Himber, H. Hilpert, W.
Huber, B. Kuhn, J.-L. Mary, M. B. Otteneder, N. Panday, F. Ricklin,
M. Stahl, S. Thomi, W. Haap, Bioorg. Med. Chem. Lett. 2010, 20,
5313–5319; e) D. J. P. Pinto, J. M. Smallheer, D. L. Cheney, R. M.
Knabb, R. R. Wexler, J. Med. Chem. 2010, 53, 6243–6274; f) Y.-K.
Lee, M. R. Player, Med. Res. Rev. 2011, 31, 202–283; g) E. Perzborn,
S. Roehrig, A. Straub, D. Kubitza, F. Misselwitz, Nat. Rev. Drug Dis-
covery 2011, 10, 61–75; h) A. Straub, S. Roehrig, A. Hillisch,
Angew. Chem. 2011, 123, 4670–4686; Angew. Chem. Int. Ed. 2011,
50, 4574–4590.
[4] For reviews on cation–p interactions, see: a) J. C. Ma, D. A. Dough-
erty, Chem. Rev. 1997, 97, 1303–1324; b) H. A. Lester, M. I. Dibas,
D. S. Dahan, J. F. Leite, D. A. Dougherty, Trends Neurosci. 2004, 27,
329–336; c) E. T. Kool, M. L. Waters, Nat. Chem. Biol. 2007, 3, 70–
73; d) E. A. Meyer, R. K. Castellano, F. Diederich, Angew. Chem.
2003, 115, 1244–1287; Angew. Chem. Int. Ed. 2003, 42, 1210–1250;
e) L. M. Salonen, M. Ellermann, F. Diederich, Angew. Chem. 2011,
123, 4908–4944; Angew. Chem. Int. Ed. 2011, 50, 4808–4842.
[5] a) P. R. Gerber, K. Mꢀller, J. Comput.-Aided Mol. Des. 1995, 9,
251–268; b) Gerber Molecular Design, http://www.moloc.ch.
[6] a) U. Obst, P. Betschmann, C. Lerner, P. Seiler, F. Diederich, V.
Gramlich, L. Weber, D. W. Banner, P. Schçnholzer, Helv. Chim.
Acta 2000, 83, 855–909; b) E. Schweizer, A. Hoffmann-Rçder, K.
Schꢁrer, J. A. Olsen, C. Fꢁh, P. Seiler, U. Obst-Sander, B. Wagner,
M. Kansy, F. Diederich, ChemMedChem 2006, 1, 611–621.
[7] L. Anselm, K. Groebke Zbinden, W. Haap, J. Himber, C. M. Stahl,
S. Thomi, US 2005/0215599, 2005.
1-{3-[(3aSR,4RS,8aSR,8bRS)-4-[5-(5-Chloro-2-thienyl)-1,2-oxazol-3-yl]-
1,3-dioxooctahydrodipyrrolo[1,2-a:3’,4’-c]pyrrol-2ACHTUNGTRENNUNG
droxyethyl)pyrrolidinium bromide ((ꢁ)-25): To
a
(20 mg, 0.041 mmol) in EtOH (1 mL) at RT, N-(2-hydroxyethyl)pyrroli-
dine (97 mL, 0.825 mmol) was added. The mixture was stirred at RT for
17 h, at 408C for four days, and evaporated to dryness. The residue was
dissolved in MeOH. Precipitation from Et2O gave (ꢁ)-25 (3 mg, 12%) as
1
colorless crystals. M.p. 1828C; H NMR (600 MHz, CDCl3): d=1.79–1.90
(m, 2H, H-C(7), H-C(8)), 1.96–2.03 (m, 1H, pyrrolidinium), 2.12–2.25
(m, 5H, H-C(7), H-C(8), +NCH2CH2CH2N, pyrrolidinium), 2.88 (ddd,
J=12.7, 8.3, 4.3 Hz, 1H, H-C(6)), 2.95 (ddd, J=12.8, 8.9, 7.1 Hz, 1H, H-
C(6)), 3.34–3.76 (m, 9H, H-C(8a), H-C(8b), +NCH2CH2CH2N,
+NCH2CH2CH2N, pyrrolidinium, CH2CH2OH), 3.82 (t, J=8.4 Hz, 1H,
H-C(3a)), 3.87–3.90 (m, 1H, CH2OH), 4.14–4.17 (m, 1H, CH2OH), 4.43
(d, J=8.7 Hz, 1H, H-C(4)), 6.68 (s, 1H, Ar), 7.12 (d, J=4.0 Hz, 1H, Ar),
7.44 ppm (d, J=4.0 Hz, 1H, Ar), protons hidden under solvent peak;
13C NMR (150 MHz, CDCl3): d=22.5, 22.6, 23.2, 24.1, 30.0, 36.6, 50.1,
50.7, 52.0, 57.3, 59.7, 61.9, 62.7, 64.6, 65.0, 69.5, 101.5, 128.2, 128.9, 129.1,
134.2, 164.5, 165.2, 177.4, 179.7 ppm; IR (ATR): n˜ =3246, 2950, 1764,
1700, 1696, 1607, 1518, 1472, 1430, 1418, 1399, 1366, 1232, 1151, 1182,
1089, 1060, 1026, 995, 953, 922, 868, 811, 789, 671 cmꢀ1; HR-MALDI-MS:
m/z (%): 521.1791 (41), [MꢀBr]+, calcd for C25H3237ClN4O4S+: 521.1801,
519.1827 (100) [MꢀBr]+, calcd for C25H3235ClN4O4S+: 519.1827, 334.2123
18, [MꢀC7H4BrClNOS]+, calcd for C18H28N3O3+: 334.2125.
[8] K. Hilpert, J. Ackermann, D. W. Banner, A. Gast, K. Gubernator, P.
Hadvꢅry, L. Labler, K. Mꢀller, G. Schmid, T. B. Tschopp, H. van de
Waterbeemd, J. Med. Chem. 1994, 37, 3889–3901.
[9] a) A. Shivanyuk, J. C. Friese, S. Dçring, J. Rebek, Jr., J. Org. Chem.
2003, 68, 6489–6496; b) N. K. Beyeh, A. Valkonen, K. Rissanen,
Supramol. Chem. 2009, 21, 142–148.
[10] a) S. Bartoli, S. Roelens, J. Am. Chem. Soc. 1999, 121, 11908–11909;
b) S. Bartoli, S. Roelens, J. Am. Chem. Soc. 2002, 124, 8307–8315;
c) C. A. Hunter, C. M. R. Low, C. Rotger, J. G. Vinter, C. Zonta,
Proc. Natl. Acad. Sci. USA 2002, 99, 4873–4876; d) C. A. Hunter,
C. M. R. Low, C. Rotger, J. G. Vinter, C. Zonta, Chem. Commun.
2003, 834–835; e) P. Sarri, F. Venturi, F. Cuda, S. Roelens, J. Org.
Chem. 2004, 69, 3654–3661.
[11] a) M. Hendlich, A. Bergner, J. Gꢀnther, G. Klebe, J. Mol. Biol.
2003, 326, 607–620; b) J. Gꢀnther, A. Bergner, M. Hendlich, G.
Klebe, J. Mol. Biol. 2003, 326, 621–636.
[12] H. Matter, M. Nazarꢄ, S. Gꢀssregen, D. W. Will, H. Schreuder, A.
Bauer, M. Urmann, K. Ritter, M. Wagner, V. Wehner, Angew.
Chem. 2009, 121, 2955–2960; Angew. Chem. Int. Ed. 2009, 48, 2911–
2916.
Acknowledgements
This work was supported by a grant from the ETH Research Council and
by F. Hoffmann-La Roche AG, Basel. We thank David Wechsler and
Daniel Zimmerli, Roche Basel, for the enantiomeric resolutions. We are
grateful to Pablo Rivera-Fuentes (ETH) for DFT calculations on the free
ligands. We thank Moritz Hunkeler and Sandra Kienast (ETH) for their
contributions to the syntheses of some of the S1 needles.
[13] R. Paulini, K. Mꢀller, F. Diederich, Angew. Chem. 2005, 117, 1820–
1839; Angew. Chem. Int. Ed. 2005, 44, 1788–1805.
[14] For a similarly positioned conserved water molecule in co-crystal
structures of phenylamidinium ligands bound to factor Xa, see: S.
Maignan, J.-P. Guilloteau, S. Pouzieux, Y. M. Choi-Sledeski, M. R.
Becker, S. I. Klein, W. R. Ewing, H. W. Pauls, A. P. Spada, V. Mikol,
J. Med. Chem. 2000, 43, 3226–3232.
[15] For a similar proposal, see: a) S. Maignan, J.-P. Guilloteau, Y. M.
Choi-Sledski, M. R. Becker, W. R. Ewing, H. W. Pauls, A. P. Spada,
V. Mikol, J. Med. Chem. 2003, 46, 685–690; b) M. Nazarꢄ, H.
[1] K. Padmanabhan, K. P. Padmanabhan, A. Tulinsky, C. H. Park, W.
Bode, R. Huber, D. T. Blankenship, A. D. Cardin, W. Kisiel, J. Mol.
Biol. 1993, 232, 947–966.
[2] a) K. Schꢁrer, M. Morgenthaler, R. Paulini, U. Obst-Sander, D. W.
Banner, D. Schlatter, J. Benz, M. Stihle, F. Diederich, Angew. Chem.
Chem. Eur. J. 2012, 18, 213 – 222
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
221

F. Diederich, W. Haap et al.
Matter, O. Klinger, F. Al-Obeidi, H. Schreuder, G. Zoller, J. Czech,
M. Lorenz, A. Dudda, A. Peyman, H. P. Nester, M. Urmann, A.
Bauer, V. Laux, V. Wehner, D. W. Will, Bioorg. Med. Chem. Lett.
2004, 14, 2801–2805; c) M. Nazarꢄ, M. Essrich, D. W. Will, H.
Matter, K. Ritter, M. Urmann, A. Bauer, H. Schreuder, J. Czech, M.
Lorenz, V. Laux, V. Wehner, Bioorg. Med. Chem. Lett. 2004, 14,
4197–4201; d) J. T. Kohrt, C. F. Bigge, J. W. Bryant, A. Casimiro-
Garcia, L. Chi, W. L. Cody, T. Dahring, D. A. Dudley, K. J. Filipski,
S. Haarer, R. Heemstra, N. Janiczek, L. Narasimhan, J. McClanahan,
J. T. Peterson, V. Sahasrabudhe, R. Schaum, C. A. Van Huis, K. M.
Welch, E. Zhang, R. J. Leadley, J. J. Edmunds, Chem. Biol. Drug
Des. 2007, 70, 100–112; e) J. X. Qiao, C.-H. Chang, D. L. Cheney,
P. E. Morin, G. Z. Wang, S. R. King, T. C. Wang, A. R. Rendina,
J. M. Luettgen, R. M. Knabb, R. R. Wexler, P. Y. S. Lam, Bioorg.
Med. Chem. Lett. 2007, 17, 4419–4427; f) Y. Shi, D. Sitkoff, J.
Zhang, H. E. Klei, K. Kish, E. C.-K. Liu, K. S. Hartl, S. M. Seiler,
M. Chang, C. Huang, S. Youssef, T. E. Steinbacher, W. A. Schu-
macher, N. Grazier, A. Pudzianowski, A. Apedo, L. Discenza, J.
Yanchunas, Jr., P. D. Stein, K. S. Atwal, J. Med. Chem. 2008, 51,
7541–7551; g) K. Groebke Zbinden, L. Anselm, D. W. Banner, J.
Benz, F. Blasco, G. Dꢄcoret, J. Himber, B. Kuhn, N. Panday, F. Rick-
lin, P. Risch, D. Schlatter, M. Stahl, S. Thomi, R. Unger, W. Haap,
Eur. J. Med. Chem. 2009, 44, 2787–2795.
len, C. Oostenbrink, Curr. Top. Med. Chem. 2010, 10, 55–66; i) R.
Abel, N. K. Salam, J. Shelley, R. Farid, R. A. Friesner, W. Sherman,
ChemMedChem 2011, 6, 1049–1066; j) S. E. Wong, F. C. Lightstone,
Expert Opin. Drug Discovery 2011, 6, 65–74.
[17] For related investigations, see references [12] and [15a].
[18] K. Mꢀller, C. Faeh, F. Diederich, Science 2007, 317, 1881–1886.
[19] B. E. Smart, J. Fluorine Chem. 2001, 109, 3–11.
[20] H. G. Wallnoefer, T. Fox, K. R. Liedl, C. S. Tautermann, Phys.
Chem. Chem. Phys. 2010, 12, 14941–14949.
[21] a) A. D. Hamilton, D. Van Engen, J. Am. Chem. Soc. 1987, 109,
5035–5036; b) A. V. Muehldorf, D. Van Engen, J. C. Warner, A. D.
Hamilton, J. Am. Chem. Soc. 1988, 110, 6561–6562; c) for a review,
see reference [4d].
[22] a) G. R. Desiraju, Angew. Chem. 1995, 107, 2541–2558; Angew.
Chem. Int. Ed. Engl.Angew. Chem. Int. Ed. 1995, 34, 2311–2327;
b) P. Metrangolo, H. Neukirch, T. Pilati, G. Resnati, Acc. Chem. Res.
2005, 38, 386–395; c) P. Metrangolo, F. Meyer, T. Pilati, G. Resnati,
G. Terraneo, Angew. Chem. 2008, 120, 6206–6220; Angew. Chem.
Int. Ed. 2008, 47, 6114–6127; d) L. A. Hardegger, B. Kuhn, B. Spinn-
ler, 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. Die-
derich, Angew. Chem. 2011, 123, 329–334; Angew. Chem. Int. Ed.
2011, 50, 314–318.
[16] For selected publications, see: a) R. Abel, T. Young, R. Farid, B. J.
Berne, R. A. Friesner, J. Am. Chem. Soc. 2008, 130, 2817–2831;
b) P. Ball, Chem. Rev. 2008, 108, 74–108; c) B. C. Roberts, R. L.
Mancera, J. Chem. Inf. Model. 2008, 48, 397–408; d) N. Huang,
B. K. Shoichet, J. Med. Chem. 2008, 51, 4862–4865; e) H. Yu, S. W.
Rick, J. Am. Chem. Soc. 2009, 131, 6608–6613; f) T. Ritschel, P. C.
Kohler, G. Neudert, A. Heine, F. Diederich, G. Klebe, ChemMed-
Chem 2009, 4, 2012–2023; g) M. Ellermann, R. Jakob-Roetne, C.
Lerner, E. Borroni, D. Schlatter, D. Roth, A. Ehler, M. G. Rudolph,
F. Diederich, Angew. Chem. 2009, 121, 9256–9260; Angew. Chem.
Int. Ed. 2009, 48, 9092–9096; h) S. B. A. de Beer, N. P. E. Vermeu-
[23] W. Kabsch, J. Appl. Crystallogr. 1993, 26, 795–800.
[24] Collaborative Computational Project, Number 4, Acta Crystallogr.
Sect. D 1994, 50, 760–763.
[25] P. Emsley, K. Cowtan, Acta Crystallogr., Sect. D: Biol. Crystallogr.
2004, 60, 2126–2132.
[26] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr., Sect.
D: Biol. Crystallogr. 1997, 53, 240–255.
Received: September 1, 2011
Published online: December 9, 2011
222
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 213 – 222