Systematic investigation of halogen bonding in protein-ligand interactions

Article abstract of DOI:10.1002/anie.201006781

Halogen bonding triggers activity: Increasing binding affinity was observed for a series of covalent human Cathepsin L inhibitors by exchanging an aryl ring H atom with Cl, Br, and I, which undergo halogen bonding with the C=O group of Gly61 in the S3 pocket of the enzyme. Fluorine, in contrast, strongly avoids halogen bonding (see scheme). The strong distance and angle dependence of halogen bonding was confirmed for biological systems. Copyright

Full text of DOI:10.1002/anie.201006781

Communications
DOI: 10.1002/anie.201006781
Halogen Bonding
Systematic Investigation of Halogen Bonding in Protein–Ligand
Interactions**
Leo A. Hardegger, Bernd Kuhn, Beat Spinnler, Lilli Anselm, Robert Ecabert, Martine Stihle,
Bernard Gsell, Ralf Thoma, Joachim Diez, Jꢀrg Benz, Jean-Marc Plancher, Guido Hartmann,
David W. Banner,* Wolfgang Haap,* and Franꢁois Diederich*
Halogen bonding (XB) refers to the noncovalent interaction
of general structure DX···A between halogen-bearing com-
pounds (DX: XB donor, where X = Cl, Br, I) and nucleo-
philes (A: XB acceptor).[1,2] Since the first observation in
cocrystal structures of 1,4-dioxane and Br2 by Hassel and
Hvoslef in 1954,[3] XB has been widely used in crystal
engineering and solid-state supramolecular chemistry.[4–6]
The nature of the interaction and the underlying electronic
prerequisite, the s hole in the XB donor, have been the
subject of extensive theoretical studies.[1,2,7–9] Most recently,
the attractive nature of XB between 1-iodoperfluoroalkanes
and various donors has also been demonstrated and quanti-
fied in solution studies.[10,11]
Novel inhibitors of human Cathepsin L (hCatL) were
discovered[12] which bind covalently to the side chain of the
catalytic Cys25 residue in the S1 pocket under formation of
thioimidates, which are stabilized by the oxyanion hole of the
protease. These ligands form hydrogen bonds to the backbone
=
NH and C O groups of Gly68 and Asp162, respectively, and
fill the S2 and S3 pockets, thereby interacting with the enzyme
through multiple lipophilic contacts. During the course of this
research, we obtained an indication of an XB contact between
a 4-chlorophenyl moiety of a ligand, whose binding affinity
was enhanced by a factor of 13 compared to the unsubstituted
Figure 1. Binding mode of covalent inhibitors at the active site of
hCatL with its three pockets. The substituent at position 4 of the
=
phenyl ring in the S3 pocket, which approaches the C O group of
Gly61, is highlighted in green. If X=Cl, Br, or I, XB (red dashed line)
with the backbone carbonyl oxygen atom of Gly61 increases the
binding affinity.
=
phenyl derivative, and the backbone C O group of Gly61 in
the S3 pocket (Figure 1). This finding stimulated the prepa-
ration of compounds (ꢀ)-1 to (ꢀ)-40 (Table 1), which were
subjected to a comprehensive investigation of XB in a
biological environment.
[*] L. A. Hardegger, Prof. Dr. F. Diederich
Laboratorium fꢀr Organische Chemie, ETH Zꢀrich
Wolfgang-Pauli-Strasse 10, HCI, 8093 Zꢀrich (Switzerland)
Fax: (+41)44-632-1109
The synthesis of the inhibitors is depicted in Scheme 1 (for
details, see the Supporting Information). Enantiopure
4-hydroxyproline derivative (2S,4S)-41 was transformed into
thioether (2S,4R)-42, which was oxidized to the correspond-
ing sulfone and subsequently saponified. The resulting acid
was coupled with 1-aminocyclopropanecarbonitrile hydro-
chloride to afford amide (2S,4R)-43. Deprotection of the
N atom (!(2S,4R)-44) and amide coupling with a-aryl acids
45 (see the Supporting Information) afforded the target
molecules (2S,4R)-46. A second ligand class with a 2-chloro-4-
(2,2,2-trifluoroethoxy)phenylsulfonyl moiety instead of
2-chlorophenylsulfonyl was also prepared and investigated
(see the Supporting Information).
E-mail: diederich@org.chem.ethz.ch
Dr. B. Kuhn, B. Spinnler, L. Anselm, R. Ecabert, M. Stihle, B. Gsell,
Dr. R. Thoma, Dr. J. Benz, Dr. J.-M. Plancher, Dr. G. Hartmann,
Dr. D. W. Banner, Dr. W. Haap
F. Hoffmann-La Roche AG
Grenzacherstrasse 124, Bau 92, 4070 Basel (Switzerland)
Fax: (+41)61-688-8714
E-mail: david.banner@roche.com
wolfgang.haap@roche.com
Dr. J. Diez
Expose GmbH, Grabenstrasse 11, 5313 Klingnau (Switzerland)
[**] This work was supported by a Novartis scholarship to L.A.H., a
grant from the ETH Research Council, and F. Hoffmann-La Roche
AG, Basel. We thank Dr. B. B. Bernet for proofreading and Dr. M.
Stahl and Dr. H. Mauser for helpful discussions.
In both ligand classes, the aryl ring in the S3 pocket was
substituted with H, Me, F, Cl, Br, I, and CF3 groups to probe
=
the importance of XB interactions with the C O group of
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201006781.
Gly61 in the S3 pocket. The aryl moiety was either a phenyl,
314
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 314 –318

Table 1: Covalent inhibitors of hCatL.[a]
pyridine, or thiophenyl ring with
one or two additional substituents,
such as F, CF3, or Cl (Table 1, as
well as Table 1SI in the Supporting
Information).
X
H
Me
F
Cl
Br
I
X
(ꢀ)-1
(ꢀ)-15 (ꢀ)-18 (ꢀ)-22
(+)-34
0.012
2.96
(+)-38
0.0065
3.23
(ꢀ)-40
0.095
3.12
IC50
0.29
0.13
2.57
0.34
2.36
0.022
2.73
logD 2.11
The IC50 values for binding to
hCatL were determined in a fluo-
rescence assay by detecting the
change in emission intensity
caused by hCatL-mediated cleav-
age of the substrate Z-Val-Val-Arg-
AMC (for definitions and details,
see the Supporting Information).
The binding affinity of both
ligand classes changed, as expected,
for XB interactions as a function of
the substituent at position 4 of the
aryl ring. For example, the
IC50 values in the series of 4-sub-
(ꢀ)-2
(ꢀ)-19 (+)-23
(+)-35
0.0065
2.75
(+)-39[b]
0.0043
3.00
IC50
0.32
0.35
2.02
0.030
2.63
logD 1.98
(ꢀ)-3
(ꢀ)-20 (ꢀ)-24
(+)-36[b]
0.030
3.08
IC50
0.52
0.93
2.46
0.022
2.98
logD 2.37
(ꢀ)-4
(ꢀ)-25
0.25
1.69
(+)-37
0.14
1.85
IC50
1.48
logD 0.85
stituted
phenyl
derivatives
(Table 1) remained essentially
unchanged when moving from
(ꢀ)-1 (X = H; 0.29 mm) to (ꢀ)-18
(X = F; 0.34 mm), as the fluorine
substituent is not able to engage in
s-hole bonding. In contrast, the
IC50 values decreased for the heavi-
er halogens, in correlation with
increasing XB donor strength, to
0.022 mm ((ꢀ)-22, X = Cl), 0.012 mm
((+)-34, X = Br), and 0.0065 mm
((+)-38, X = I). The binding affin-
ity in ligand class 1 (Table 1) on
changing from H to Cl increases by
a factor of 12 ꢁ 9 for all the sub-
stitution patterns in Table 1.
Assuming competitive inhibition,
this increase corresponds to a gain
in the binding free enthalpy of
ꢀDDG = 1.5 ꢁ 1.3 kcalmolꢀ1.[13] In
(ꢀ)-5
(ꢀ)-16
0.41
2.49
(ꢀ)-26
0.16
2.89
IC50
0.16
logD 2.03
(+)-6
0.69
logD 2.22
(+)-21 (+)-27
IC50
0.36
2.45
0.022
3.0
(ꢀ)-7
(2S,4R)-28
0.055
2.98
IC50
0.88
logD 2.51
(+)-8
0.30
logD 2.42
(+)-29
0.023
3.19
IC50
ligand
class 2
(Table S1),
a
(ꢀ)-9
(+)-30
0.032
3.35
medium gain in binding affinity of
14 ꢁ 20 is found on changing from
H to Cl; this gain corresponds to a
gain in binding free enthalpy of
IC50
0.34
logD 2.7
ꢀDDG = 1.5 ꢁ 1.8 kcalmolꢀ1
for
(ꢀ)-10
0.46
logD 2.94
(ꢀ)-31
0.18
3.38
competitive inhibition.
IC50
The
IC50 values
further
decreased by a factor of approxi-
mately 2 and 4, respectively, upon
changing the 4-position to Br or I.
Thus, the I-substituted compound
(+)-39 (IC50 value: 0.0043 mm) is
the most active inhibitor of the
entire ligand class. A methyl sub-
stituent (compounds (ꢀ)-15 to
(+)-17), which is most similar in
size to Cl, does not enhance the
binding affinity significantly. A sur-
(ꢀ)-11 (+)-17
IC50
0.52
0.22
2.82
logD 2.14
Angew. Chem. Int. Ed. 2011, 50, 314 –318
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
315

Communications
Table 1: (Continued)
substituent of (ꢀ)-22 (Figure 2a)
clearly forms the shortest contact
to the backbone carbonyl oxygen
atom of Gly61 (3.1 ꢀ), three addi-
tional interactions to CH groups
(Glu63 Cg, Gly68 Ca, Tyr72 Ce2) at
distances between 3.7 and 4.1 ꢀ are
made with the Cl atom. Moreover,
the energetics of the stacking inter-
action of the aryl ring with the
planar peptide fragment Gly67–
Gly68, at the bottom of the S3
pocket, might be altered by chang-
ing X. There is no correlation
between the logD value (logarith-
mic distribution coefficient octanol/
water at pH 7.4) and the binding
affinity (see Section 4 in the Sup-
porting Information). While it can
X
H
Me
F
Cl
Br
I
X
(ꢀ)-12
IC50
logD 2.86
0.52
(ꢀ)-13
0.97
logD 1.62
(ꢀ)-32
0.56
2.48
IC50
(ꢀ)-14
0.39
logD 2.48
(ꢀ)-33
0.024
>3.0
IC50
[a] Top row: compound number; middle row: IC50 values (mm); bottom row: logD values. Compounds of
the second ligand class show similar behavior. For details of the determination of IC50 and logD values,
see the Supporting Information. The IC50 values were obtained from two or three measurements and
have an uncertainty on average of 2–30%. [b] The IC50 values were obtained from eight measurements.
=
be expected that the C O group of
Gly61 is solvated in the apo struc-
ture, the replacement of water
cannot explain the large gain in
binding upon introducing Cl or heavier halides compared to F
or Me substituents. Substitution of the 4-X-phenyl ring by one
or even two additional electron-withdrawing substituents (as
in (+)-29) resulted in only a small effect on binding affinity.
Higher substitution patterns would have been desirable, but
were not compatible with the ligand synthesis employed.
Much insight into the nature of XB interactions in the
S3 pocket of hCatL was gained when a series of four X-ray
cocrystals was solved (Figure 2). The X-aryl moieties stack, as
expected, on the peptide backbone of Gly67–Gly68 and
=
orient the X substituent towards the C O group of Gly61.
The Cl substituent in bound chlorophenyl derivative (ꢀ)-22
(1.45 ꢀ resolution, PDB code: 2xu1; Figure 2a and Fig-
ure 3SI) shows a nearly ideal XB interaction, with the O···Cl
distance (3.1 ꢀ) below the sum of the van der Waals radii
(3.27 ꢀ)[14] and the angle O···Cl C (1748) close to 1808. For
ꢀ
ꢀ
electrostatic reasons, XB is especially sensitive to the O···X C
angle, which should be close to 1808.[2,7,8,15–17] There are four
independent protein–ligand complexes in the unit cell, for
which we take the observed distances and angles as inde-
pendent measurements and use the average (d(O···Cl) =
Scheme 1. Synthesis of target molecules (2S,4R)-46: a) 3-nitrobenzene-
1-sulfonyl chloride (Nos-Cl), Et3N, CH2Cl2, 0!228C, 10 h; b) 2-chloro-
benzenethiol, Et3N, propionitrile, 1008C, 5.5 h, 90% (2 steps);
c) mCPBA, CH2Cl2, 0!228C, 68 h; d) LiOH, THF/H2O (1:1.5), 228C,
1.5 h; e) HATU, iPr2EtN, 1-aminocyclopropanecarbonitrile hydrochlo-
ride, DMF, 228C, 14.5 h, 79% (3 steps); f) HCO2H, 228C, 2.5 h, 80%;
g) HATU, iPr2EtN, amine (2S,4R)-44, DMF, 228C. Alternatively: SOCl2,
CH2Cl2, then iPr2EtN, amine (2S,4R)-44, CH2Cl2, 228C. For substituents
Ar and R, see Table 1 and Table 1SI. mCPBA: meta-chloroperbenzoic
acid; HATU: O-(7-azabenzotriazol-1-yl)-N,N,N’,N’-tetraethyluronium
hexafluorophosphate; Boc: tert-butyloxycarbonyl.
ꢀ
3.08 ꢁ 0.11 ꢀ; angle O···Cl C = 173.6 ꢁ 1.18; see Section 5.1
in the Supporting Information).
The 5-chlorothiophen-2-yl derivative (ꢀ)-26 (IC50 value:
0.16 mm) did not show stronger binding than the unsubstituted
control compound (ꢀ)-5 (IC50 value: 0.16 mm). The reason
became apparent when the cocrystal structure of (ꢀ)-26 with
hCatL was solved (0.9 ꢀ resolution, PDB code: 2xu3; Fig-
ure 2b and Figure 4SI). Two different conformations of the
ligand were observed. In the conformer populated by 75%,
the geometry is rather favorable for an XB interaction
ꢀ
prisingly strong affinity was found in one case for a CF3-
substituted ligand ((ꢀ)-40; IC50 value: 0.095 mm).
(d(O···Cl) = 3.1 ꢀ; angle O···Cl C = 1668). However, this gain
in interaction energy seems to be compensated by intra-
molecular ligand strain, as indicated by a short, repulsive
contact (3.0 ꢀ) between the thiophenyl C atom attached to
the cyclopropyl ring and the unsubstituted C atom adjacent to
It is important to note that the gain in binding affinity
upon replacement of X = H by X = Cl or higher halides
presumably does not arise from XB only. While the Cl
316
www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 314 –318

ꢀ
vorable O···Cl C alignment is most probably also at the origin
of the low affinity of chlorophenyl derivative (ꢀ)-9
(IC50 value: 0.34 mm) compared to (ꢀ)-22 (IC50 value:
0.022 mm).
The cocrystal structure of 4-fluorophenyl derivative
(ꢀ)-18 in a complex with hCatL (1.12 ꢀ resolution, PDB
code: 2xu4; Figure 2c and Figure 5SI) strongly confirms
earlier reports that organofluorine substituents avoid regions
of high electron density and avoid pointing directly at the
O atoms of peptidic C O bonds.[18] The F atom is moved away
=
from the carbonyl group to avoid electrostatic repulsion,
resulting in a O···F distance of 4.5 ꢀ (sum of the van der
Waals radii: 2.99 ꢀ),[14] with a water molecule bridging this
contact.
The methyl group in the cocrystal structure of (ꢀ)-15 with
hCatL (1.6 ꢀ resolution, PDB code: 2xu5; Figure 2d and
=
Figure 6SI) points toward the C O group, but has a consid-
erably longer O···C distance (3.6 ꢀ) compared to the Cl
derivative. Interestingly, the S3 pocket is widened through
side chain shifts of Glu63, Leu69, and Tyr72 compared to the
Cl structure, thus resulting in reduced interaction of the
methyl group with the protein.
An overlay of all four crystal structures (Figure 2e and
Figure 7SI) shows the unique mechanism that allows these
adaptations of the X-aryl moiety in the S3 pocket. While all
four ligands maintain identical binding geometries in the S1
and S2 pocket, the puckering of the central pyrrolidine ring in
the ligands changes slightly. This does not involve much
change in the conformational energy or in the intermolecular
interactions in this region of the protein. This small change in
the puckering of the five-membered ring, however, translates
into larger differences in the penetration of the X-aryl moiety
into the S3 pocket.
Figure 2. a) Cocrystal structure of (ꢀ)-22 with hCatL at 1.45 ꢁ resolu-
tion (PDB code: 2xu1). The amino acids of the S3 pocket are
highlighted, as well as the XB interaction between the backbone
carbonyl group of Gly61 and the chlorine atom. b) The cocrystal
structure of (ꢀ)-26 in a complex with hCatL at 0.9 ꢁ resolution (PDB
code: 2xu3) shows two different binding modes: 75% (green) undergo
XB but suffer from intramolecular repulsion (green dashed line), 25%
(pink) bind without apparent intramolecular strain, but undergo
poorer XB. c) Cocrystal structure of (ꢀ)-18 with hCatL at 1.12 ꢁ
resolution (PDB code: 2xu4). The repulsion between the carbonyl
oxygen atom and the fluorine atom as well as the distances to the
bridging water are highlighted. d) Cocrystal structure of (ꢀ)-15 with
hCatL at 1.6 ꢁ resolution (PDB code: 2xu5). Distances are given in ꢁ.
Color code: C: gray (enzyme), green or pink (inhibitor); O: red; N:
blue; S: yellow; Cl: lemon; F: light blue. e) Overlay of (ꢀ)-15
(magenta), (ꢀ)-18 (green), (ꢀ)-22 (turquoise), and the two binding
modes of (ꢀ)-26 (yellow with strong XB, violet with weak XB). The
adjustment of the phenyl moiety in the S3 pocket is accommodated by
a slight change of the puckering of the five-membered ring.
We compared our experimental results with the energetics
=
of an isolated C O···X-Phe interaction. Thus, we performed
quantum mechanical calculations at an adequate theoretical
level (MP2/aug-cc-pVDZ//B3LYP/aug-cc-pVDZ, for details
see the Supporting Information) to determine interaction
energy curves for different monosubstituted phenyl deriva-
tives with N-methylacetamide. Model geometries were con-
strained to the relative orientation found in the X-ray
complex structure of compound (ꢀ)-22. In line with previous
calculations of related, unconstrained model systems,[8] we
find attractive energy profiles for Cl, Br, and I, with a
common minimum distance of 3.1 ꢀ and well depths of ꢀ1.3,
ꢀ2.2, and ꢀ3.5 kcalmolꢀ1, respectively (Figure 3, Figure 8SI).
The observed O···Cl distance of 3.1 ꢀ in the complex
structure of (ꢀ)-22 is close to its optimal value, further
supporting a stabilizing effect of the XB interaction in the
complex of the Cl derivative. XB interactions with the Br and
especially I analogues should be stronger, which is reflected in
the approximately two- and fourfold lower IC50 values for
hCatL binding.
the pyrrolidine N atom (green dashed line in Figure 2b). The
second conformer, populated by 25%, features a much less
ꢀ
favorable geometry for XB interactions (angle O···Cl C =
In contrast to the heavier halides, the interaction of the
fluorine derivative is repulsive in nature, and the energy curve
shows no minimum. The loss of binding affinity and the
change in binding mode to a much larger intermolecular
distance are in good agreement with this. The computed
interaction energies for a close to linear arrangement are even
1398 at d(O···Cl) = 3.0 ꢀ), but shows no apparent intra-
molecular strain. According to quantum-mechanical energy
profile calculations (see below), the difference in the XB
interaction energy for the two geometrical arrangements
should be approximately 1 kcalmolꢀ1 (Figure 9SI). An unfa-
Angew. Chem. Int. Ed. 2011, 50, 314 –318
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
317

Communications
sum of the van der Waals radii and a strong dependence on
ꢀ
the O···X C angle. Establishing a halogen bond might
enhance protein–ligand interactions by a factor of as much
as 74 ((ꢀ)-2 versus (+)-39), which translates into a gain in free
enthalpy of ꢀDDG = 2.6 kcalmolꢀ1. In view of this favorable
energetic balance, it is predictable that XB will increasingly
be used to enhance protein–ligand binding.
Received: October 28, 2010
Keywords: halogen bonding · medicinal chemistry ·
.
molecular recognition · protein–ligand interactions ·
structure–activity relationships
[1] P. Politzer, P. Lane, M. C. Concha, Y. Ma, J. S. Murray, J. Mol.
Model. 2007, 13, 305 – 311.
[2] T. Clark, M. Hennemann, J. S. Murray, P. Politzer, J. Mol. Model.
2007, 13, 291 – 296.
[3] a) O. Hassel, J. Hvoslef, Acta Chem. Scand. 1954, 8, 873; b) O.
Hassel, Science 1970, 170, 497 – 502.
[4] G. R. Desiraju, Angew. Chem. 1995, 107, 2541 – 2558; Angew.
Chem. Int. Ed. Engl. 1995, 34, 2311 – 2327.
Figure 3. Calculated interaction energies for monosubstituted phenyl
derivatives with N-methylacetamide as a peptide backbone mimic.
Torsional angles from the crystal structure of (ꢀ)-22 were applied as
constraints to enforce the relative orientation at the active site of
hCatL (see the Supporting Information).
[5] a) P. Metrangolo, G. Resnati, Chem. Eur. J. 2001, 7, 2511 – 2519;
b) P. Metrangolo, H. Neukirch, T. Pilati, G. Resnati, Acc. Chem.
Res. 2005, 38, 386 – 395; c) P. Metrangolo, G. Resnati, Halogen
Bonding: Fundamentals and Applications, Springer, Berlin,
2008; d) P. Metrangolo, F. Meyer, T. Pilati, G. Resnati, G.
Terraneo, Angew. Chem. 2008, 120, 6206 – 6220; Angew. Chem.
Int. Ed. 2008, 47, 6114 – 6127; e) P. Metrangolo, G. Resnati,
Science 2008, 321, 918 – 919.
[6] For some examples, see a) A. Sun, J. W. Lauher, N. S. Goroff,
Science 2006, 312, 1030 – 1034; b) A. C. B. Lucassen, A. Karton,
G. Leitus, L. J. W. Shimon, J. M. L. Martin, M. E. van der Boom,
Cryst. Growth Des. 2007, 7, 386 – 392.
[7] J. P. M. Lommerse, A. J. Stone, R. Taylor, F. H. Allen, J. Am.
Chem. Soc. 1996, 118, 3108 – 3116.
[8] a) K. E. Riley, K. M. Merz, Jr., J. Phys. Chem. A 2007, 111,
1688 – 1694; b) K. E. Riley, P. Hobza, J. Chem. Theory Comput.
2008, 4, 232 – 242.
[9] a) P. Politzer, J. S. Murray, P. Lane, Int. J. Quantum Chem. 2007,
107, 3046 – 3052; b) P. Politzer, J. S. Murray, T. Clark, Phys.
Chem. Chem. Phys. 2010, 12, 7748 – 7757.
[10] R. Cabot, C. A. Hunter, Chem. Commun. 2009, 2005 – 2007.
[11] M. G. Sarwar, B. Dragisic, L. J. Salsberg, C. Gouliaras, M. S.
Taylor, J. Am. Chem. Soc. 2010, 132, 1646 – 1653.
[12] R. Alvarez Sanchez, D. Banner, S. M. Ceccarelli, U. Grether, W.
Haap, P. Hartman, G. Hartmann, H. Hilpert, H. Kuehne, H.
Mauser, J.-M. Plancher (F. Hoffmann-La Roche AG),
US20100267722(A1), 2010.
[13] Y.-C. Cheng, W. H. Prusoff, Biochem. Pharmacol. 1973, 22,
3099 – 3108.
[14] A. Bondi, J. Phys. Chem. 1964, 68, 441 – 451.
[15] K. E. Riley, J. S. Murray, P. Politzer, M. C. Concha, P. Hobza, J.
Chem. Theory Comput. 2009, 5, 155 – 163.
more unfavorable with the CF3 derivative. We expect that the
binding mode will be different from the chlorine compound
and that the good binding affinity of (ꢀ)-40 (IC50 value:
0.095 mm) is due to intermolecular interactions of the CF3
=
group that are unrelated to the C O group of Gly61. Weak
ꢀ
O···H C interactions play a role for the methyl derivative,
and the calculations predict a slightly weaker interaction
energy than the chlorine derivative with the carbonyl oxygen
atom at the observed distances. However, this difference in
“solvation” is too small to fully explain the sixfold higher
IC50 value of the methyl derivative. Widening of the S3 pocket
and fewer intermolecular interactions are seen in the crystal
structure of the methyl derivative, thus suggesting that the
methyl substituent is less well accommodated in this “CH-
rich” region, compared to the more polarizable Cl atom.
Apparently, both a favorable XB interaction and an excellent
general fit to the S3 pocket of hCatL contribute to the
increases in the affinity of the non-fluorine-containing halide
compounds.
In summary, we have presented the first systematic study
on XB in protein–ligand complexes and show that XB can
indeed serve as a powerful tool, comparable to hydrogen
bonding, to enhance the binding affinity and certainly also
affect the binding selectivity, as proven recently[19] in biolog-
ical molecular recognition (for earlier examples of potential
XB in biological complexes, see Section 7 of the Supporting
Information). Our study confirms several theoretical predic-
tions and also the recent results on model systems. XB
increases in strength with the mass of the halide substituent
(Cl < Br< I) but is non-existent with organofluorine com-
pounds. The interaction has high geometrical requirements,
such as a distance between the interacting atoms below the
[16] P. Auffinger, F. A. Hays, E. Westhof, P. S. Ho, Proc. Natl. Acad.
Sci. USA 2004, 101, 16789 – 16794.
[17] C. Ouvrard, J.-Y. Le Questel, M. Berthelot, C. Laurence, Acta
Crystallogr. Sect. B 2003, 59, 512 – 526.
[18] K. Mꢁller, C. Faeh, F. Diederich, Science 2007, 317, 1881 – 1886.
[19] S. Baumli, J. A. Endicott, L. N. Johnson, Chem. Biol. 2010, 17,
931 – 936.
318
www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 314 –318