Ligand binding cooperativity: Bioisosteric replacement of [Formula presented] with SO2among thrombin inhibitors

Article abstract of DOI:10.1016/j.bmcl.2016.07.024

Ligand–protein binding is a complex process that involves the formation of number of non-covalent interactions, e.g. H-bonds and hydrophobic interactions, between the ligand and the protein host. Upon binding, ligand functional groups can act synergistica

Full text of DOI:10.1016/j.bmcl.2016.07.024

Accepted Manuscript
Ligand Binding Cooperativity: Bioisosteric Replacement of C=O With SO2
Among Thrombin Inhibitors
Ahmed M. Said, David G. Hangauer
PII:
S0960-894X(16)30737-5
DOI:
http://dx.doi.org/10.1016/j.bmcl.2016.07.024
BMCL 24068
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
29 May 2016
7 July 2016
8 July 2016
Please cite this article as: Said, A.M., Hangauer, D.G., Ligand Binding Cooperativity: Bioisosteric Replacement of
C=O With SO2 Among Thrombin Inhibitors, Bioorganic & Medicinal Chemistry Letters (2016), doi: http://
dx.doi.org/10.1016/j.bmcl.2016.07.024
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Ligand Binding Cooperativity: Bioisosteric Replacement of C=O With
SO2 Among Thrombin Inhibitors
Ahmed M. Said a,b and David G. Hangauer a
a Department of Chemistry, University at Buffalo, The State University of New York, Buffalo,
NY 14260
b
Department of Pharmaceutical Organic Chemistry, Faculty of Pharmacy, Helwan University,
Helwan, Egypt
Corresponding Authors: Correspondences should be addressed to A M S at
ahmedmoh@buffalo.edu or D H at hangauer@buffalo.edu
Abstract: Ligand-protein binding is a complex process that involves the formation of number of
non-covalent interactions, e.g. H-bonds and hydrophobic interactions, between the ligand and the
protein host. Upon binding, ligand functional groups can act synergistically (positive
cooperativity) to improve the overall ligand binding affinity beyond what would be expected
from their individual contributions. In this study, using thrombin as a protein model system, we
evaluated the effect of the bioisosteric replacement of a carbonyl functionality with a sulphonyl
functionality on positive cooperativity between their H-bonds with thrombin and hydrophobic
binding in the adjacent S3 pocket. The positive cooperativity observed was greatly reduced when
replacing the carbonyl group with a sulphonyl group. Evaluating how bioisosteric replacements
affect cooperativity is important for making better informed ligand optimization SAR decisions.
Keywords: Ligand-protein binding; Cooperativity; Nonadditivity; Bioisosteric replacement;
Hydrogen bonding; Hydrophobic binding; Thrombin.
One of the challenges facing medicinal chemists in drug discovery is to optimize the binding
affinity of the ligands for their protein target.1-4 Unfortunately, accurately predicting a ligand’s
1

binding affinity remains a significant problem. This is mainly because some of the fundamental
non-covalent factors underlying the binding process are still not fully understood and quantified.
These factors include enthalpy-entropy compensation, the detailed role of water, and the
intricacies of the hydrophobic effect and cooperativity. 5-12
Scoring functions used to predict ligand binding affinity typically make the inaccurate
assumption that the contribution of ligand functional groups can be treated in additive manner.13-16
They usually don’t take into consideration that ligand functional groups can have a mutual effect
on each other i.e. synergistic or antagonistic. Cooperativity among ligand functional groups is
observed when the differential binding energy obtained from non-covalent interactions when
acting together is more favorable (positive cooperativity) or less favorable (negative
cooperativity) than the sum of the differential binding energies obtained when these groups act
individually. The potential error in ligand binding affinity predictions for not including
cooperativity between ligand functional groups can be 1-3 orders of magnitude.17-20 This suggests
that scoring function predictions of ligand binding affinities could be improved if cooperativity
were properly included. Kuhn and coworkers have started to address this significant deficiency
by incorporating some elements of cooperativity into their scoring function.21
Bioisosteric replacement is a useful tactic for medicinal chemists when seeking to improve
potency, enhancing selectivity, alter physical properties, reduce or redirect metabolism, eliminate
or modify toxicophores, or to acquire new intellectual property without making major changes in
the chemical structure of a molecule.1-2 Bioisosterism also finds utility in structure based drug
design. There are now a large number of available crystal structures of ligand-protein complexes,
as well as for the protein hosts themselves. The availability of this structural information allows
the medicinal chemist to design new molecules while attempting to determine what changes
may, or may not, be tolerated in a particular location on the ligand. As a result, the decision to
2

use a functional group bioisostere, and if so which bioisostere, can be guided by the ligand’s
structural information available.
Hydrogen bonds are considered one of the mostly important non-covalent interactions between
the ligand and protein when forming the complex.22-23 In this study, the bioisosteric replacement
of a H-bond accepting carbonyl group (C=O) with a H-bond accepting sulfonyl group (SO2) was
performed and the effect on functional group cooperativity with an adjacent hydrophobic
interaction was determined.
Thrombin is considered a good model protein for fundamental structure-based ligand design
studies. This is mainly because thrombin has a relatively rigid and well-defined binding site with
numerous deposited x-ray crystal structures containing a variety of bound inhibitors. The active
24
site of the thrombin consists of four pockets: (1) The S1 pocket which is a deep hydrophobic
pocket with the carboxylate of Asp189 and two backbone carbonyls at the bottom of the pocket.
This Asp189 carboxylate anion is responsible for strong ionic and bifurcated H-bonding
interactions with positively charged ligand amidine containing P1 residues; (2) The S2
hydrophobic pocket which mainly consists of Tyr60A and Trp60D side chains of the 60-
insertion loop as well as the isobutyl group of Leu99; (3) The S3-pocket which is a well-defined
hydrophobic pocket, consisting mainly of the side chains of Trp215, Ile174, and Leu99.
In the current study the effect of a bioisosteric replacement of the ligand carbonyl moiety with
a sulfonyl moiety, both of which can accept an H-bond from thrombin Gly216, on cooperativity
with the adjacent hydrophobic side chains binding in the S3 pocket was evaluated. In our
previous study, this carbonyl group was found to be engaged in positive cooperativity with the
hydrophobic side chain which accounted for -3.9 kJ/mol enhanced binding affinity.18 The current
study address’s two important questions for predicting ligand binding affinities: 1) Is
cooperativity the same when substituting one H-bond acceptor with another? 2) How does the
3

H-bond acceptor change affect the magnitude of the hydrophobic binding in an adjacent pocket?
To answer these questions, three series of thrombin inhibitors were designed, synthesized, and
biologically tested. The structural features and the expected non-covalent interactions with
thrombin for these inhibitors are shown in Figure 1.
Series I: X = -C=O
Series II: X = -SO2-
Series III: X = -CH2-
R = -CH3, -CH2CH3, -CH(CH3)2, -
CH2CH(CH3)2, -benzyl
Figure 1. The structural features, expected non-covalent interactions, and the binding mode of
the p-benzamidine containing inhibitors inside the thrombin active site. The X group is carbonyl
or sulphonyl or methylene.
As shown in Figure 1, thrombin inhibitors of series I have X = carbonyl group, which
acts as a H-bond acceptor with the Gly216 –NH- residue. In series II the carbonyl group of series
I is replaced with a sulphonyl group to act as a H-bond acceptor bioisostere, and which is also
expected to form a H-bond with the Gly216 –NH- residue. In series III, the H-bond accepting
moiety is removed altogether by replacing it with a methylene group. The common structural
features between all of these inhibitors are: 1) A proline moiety selected to bind under Tyr60A
and Trp60D in the S2 pocket, as does the natural substrate.25 2) The R hydrophobic side chain
that binds in the thrombin S3 pocket and ranges from a methyl group to a benzyl group. 3) A p-
benzamidine moiety to bind in the thrombin S1 pocket. This moiety forms strong ionic
interactions with the protein residues through a bifurcated salt bridge with the Asp 189
4

carboxylate at the bottom of the S1 pocket. The amidine also forms H-bonds with the Gly219
carbonyl residue as well as with a nearby crystallographic water molecule.
The synthetic route for series II is outlined in Scheme 1. The synthesis starts with an
amide coupling in DMF between commercially available N-Boc-L-proline and 4-cyano benzyl
amine using EDC / HOAt in presence of DIEA to provide 1a. The Boc protecting group was
then removed by stirring in 3 N methanolic HCl to provide intermediate 1b as hydrochloride salt.
The general synthesis continues with a coupling between the commercially available sulphonyl
chlorides and 1b using triethylamine (TEA) as a base in DCM. The resulting sulfonamides 1c-5c
were stirred with hydroxylamine hydrochloride and DIEA in anhydrous methanol for 16 h to
form the corresponding hydroxyamidines. The hydroxyamidines were carried on directly by
stirred with acetic anhydride in acetic acid for 30-45 min. 10% Pd/C was then added and the
mixture was hydrogenated. The resulting products were purified by reverse phase HPLC to give
the final compounds 1-5 with a purity of > 95%. We published the synthetic details for series I
and III previously.18
Scheme 1. Synthesis of thrombin inhibitors series II
5

Reagents and conditions: (a) EDCI, HOAT, DIEA, DCM, rt, 16 h; (b) 3 N methanolic HCl, 3 h,
0 °C ; (c) TEA, DCM, 24 h; (d) Hydroxylamine HCl, DIEA, anhydrous MeOH, 16 h; (e) Ac2O,
AcOH, 30 min then H2, 10% Pd/C, 12 h
All the ligands reported herein were tested for thrombin inhibition using a standard
kinetic photometric assay at pH 7.4 using Pefachrome-tPA as a chromogenic substrate.26 The
IC50 values for the tested ligands were determined from the dose-response curves. These values
were then utilized to obtain the inhibition constants “Ki’s” using the Cheng-Prusoff equation,27
and the inhibition constants were used to calculate the binding free energies “ΔG’s” of the tested
ligands using the equation ΔG = RT ln(Ki). In this study, the hydrophobic contact surface areas
between the designed inhibitors and thrombin active site were calculated using SYBYL-X
following our previously published protocol.18-19 The molecular modeling of the series II and III
ligands was carried out starting with the thrombin bound crystal structures of 7 (pdb: 2ZI2) and
10 (pdb: 2ZHQ).19 The resulting modeled complexes were then energy minimized to
convergence using the Tripos force field. More details regarding the molecular modeling
procedure are given in the Supplementary Information. Once the ligands were docked into the
6

thrombin active site, the hydrophobic contact surface areas were calculated. The experimentally
determined Ki values, the corresponding Gibbs free energy ΔG˚ values, and the calculated
hydrophobic contact surface areas for the series I, II and III inhibitors are provided in Table 1.
Table 1. The Ki values, the corresponding standard Gibbs free energy ∆G˚ and the calculated
hydrophobic contact surface area for thrombin inhibitors 1-15
Inhibitor
No.
Series No.
(X group)
R
Ki
(µM)
± sd
∆G˚
(kJ/mol)
± sd
Hydrophobic
contact
surface area
(Å2)
II (SO2)
II (SO2)
II (SO2)
II (SO2)
II(SO2)
Methyl
Ethyl
117
12
12
-22.4
-28.1
-31.2
-31.8
-31.9
0.3
0.1
0.2
0.4
0.1
171 ± 4.0
1
2
3
4
5
0.40
0.30
0.46
0.11
243 ± 0.9
270 ± 1.5
265 ± 2.2
339 ± 5.0
Isopropyl
Isobutyl
Benzyl
3.4
2.7
2.5
I (C=O)
I (C=O)
I (C=O)
I (C=O)
I (C=O)
III (CH2)
III (CH2)
III (CH2)
III (CH2)
III (CH2)
Methyl
Ethyl
14
5.0
4.0
1.0
0.2
63
0.1
0.2
0.8
0.04
0.02
2.0
6.0
7.0
5.0
1.0
-27.6
-30.3
-31.1
-34.2
-37.9
-23.9
-24.3
-24.6
-26.6
-28.0
0.2
0.1
0.5
0.1
0.2
0.1
0.3
0.4
0.5
0.2
298 ± 0.5
324 ± 6.0
360 ± 2.0
369 ± 0.6
389 ± 2.0
286 ± 1.2
322 ± 3.2
344 ± 0.5
364 ± 1.6
388 ± 0.9
6
7
Isopropyl
Isobutyl
Benzyl
Methyl
Ethyl
8
9
10
11
12
13
14
15
54
Isopropyl
Isobutyl
Benzyl
48
22
12
7

Bioisosteric replacement of the carbonyl group with a sulphonyl group: What is the effect on
binding affinity, adjacent hydrophobic binding, and cooperativity? The binding affinity (Ki) of
the key inhibitors 2, 5, 7, 10, 12 and 15 are compared first. These inhibitors form matched sets
wherein the side chain R is either ethyl or benzyl, respectively, and the H-bonding moiety (X)
with both side chains is either C=O (7 & 10), SO2 (2 & 5), or CH2 (12 & 15). The absolute
binding affinities of the three inhibitor series when the larger benzyl side chain is present
decreased in the range of an order-of-magnitude per change in the sequence X = C=O (10, Ki =
0.2 M), to X = SO2 (5, Ki = 2.5 M) and then to X = CH2 (15, Ki = 12 M). The 10-fold
reduction in activity going from X = C=O, to X = SO2 can be only partly traced back to the 2-
fold reduction in binding affinity with the small ethyl side chain between series I and II as
demonstrated by comparing 7 (X = C=O, Ki = 5.0 M) to 2 (X = -SO2, Ki = 12 M). In contrast,
the 5-fold reduction in binding from X = SO2 to X = CH2 with a benzyl side chain is mirrored by
the 5-fold reduction in binding with an ethyl side chain, i.e. compare 2 (X = -SO2, Ki = 12 M)
to 12 (X = CH2, Ki = 54 M). Increasing the size of the R side chain from an ethyl group in 12 to
a benzyl group in 15 (X = CH2) results in a 5-fold improvement in the ligand binding affinity.
Inhibitor 7 (X= C=O, has a H-bonding capability with Gly216, R= ethyl group) has a binding
affinity of 5 µM. Increasing the size of its R side chain to a benzyl group in 10 now results in a
~25-fold improvement in the ligand binding affinity compared to only 5-fold improvement when
X=CH2. Replacing the X=carbonyl with a X=sulphonyl while R=ethyl results in a slight (~2-
fold) reduction in the ligand binding affinity to 12 µM (inhibitor 2 compared to 7). When the R
side chain is again increased to a benzyl group from an ethyl group the improvement in binding
affinity is now only ~5-fold, the same as when X= CH2. This shows that the nature of the X
group’s interaction with Gly216 is affecting the hydrophobic binding contribution of the R side
8

chain in the adjacent S3 pocket. If the H-bond is a carbonyl group, the contribution of the P3
benzyl side chain to the binding affinity is high, whereas if the H-bond acceptor is absent, i.e. X
= CH2, the contribution of the P3 benzyl side chain on the binding affinity is weak. This is in
spite of the fact that the hydrophobic contact surface across the side chain series is very similar
for X = C=O vs. X = CH2 as shown in Table 1. As described in more detail later, when X = SO2
the positioning of the SO2 H-bond acceptor and the P3 side chain can be mutually compromised
to non-optimal orientations.
In order to gain more insight into the effect of the H-bonding properties of group X on
cooperativity with the adjacent hydrophobic side chains, the hydrophobic contact surface area
between the ligand’s P3 hydrophobic side chain and the protein was calculated and plotted
against the corresponding Gibbs free energy for each series of inhibitors from Table 1 as shown
in Figure 2.
Figure 2. Plot of Gibbs binding energy vs hydrophobic contact surface area for series I (red
squares), II (blue diamonds) and III (green triangles). Standard deviation of both Gibbs free
energy and the hydrophobic contact surface area are incorporated in the plot.
This plot demonstrates a linear relationship between the binding free energy and the size
of the R side chain within each series of inhibitors, analogous to earlier studies we have
published.17 The slope of the line for series I (X = C=O) is -0.1007 (kJ/mol) / Å2 which is 2.5-
9

fold more than that of -0.0409 (kJ/mol) / Å2 in series III (X = CH2). This is the result of the
existence of a positive cooperativity between the carbonyl group and the adjacent hydrophobic
side chain wherein the incremental increase in hydrophobic binding per Å2 is magnified by the
presence of an adjacent H-bonding interaction as previously demonstrated.17 In addition, the
linear relationship indicates that the amount of cooperativity between the carbonyl H-bond and
the hydrophobic side chain does not change as the size of R side chain is systematically
increased. On the other hand, the slope of the line for series II (X = SO2) is -0.0501 (kJ/mol) / Å2
which is only 1.2-fold more than that of -0.0409 (kJ/mol) / Å2 for series III (X = CH2), even
though the absolute binding affinity is increased by 5-fold when comparing series II to series III.
This means that the strong positive cooperativity between the X = C=O group and the P3 side
chain is lost when the carbonyl is replaced by the sulphonyl bioisostere.
Finally the data can be analyzed thermodynamically using double mutant cycles.
Generally in a double mutant cycle, cooperativity is determined by obtaining the difference
between two differential binding free energies on either the right/left or the bottom/top arrows in
the analysis diagram (Figure 3). A difference with a negative sign indicates positive
cooperativity (synergism), a difference with a positive sign indicates negative cooperativity
(antagonism), while a difference close to zero indicates additivity. To reveal the effect of
functional group bioisosteric replacement on cooperativity, two double mutant cycles were
constructed as shown in Figure 3. The double mutant cycle shown in Figure 3a reveals the
previously discovered positive cooperativity between the carbonyl group H-bond and the
adjacent P3 phenyl side chain.18 This positive cooperativity was found to account for 3.9 kJ/mol
improvement in the binding free energy. Replacing the carbonyl with a sulphonyl, as shown in
10

the double mutant cycle shown in Figure 3b, results in a reduction in this cooperativity to only -
0.1 kJ/mol thereby becoming essentially additivity.
Figure 3. a) Double mutant cycle showing the positive cooperativity between the carbonyl group
and the P3 side chain accounts for -3.9 kJ/mol; b) Double mutant cycle showing the effect of the
bioisosteric replacement of the carbonyl with a sulphonyl instead greatly reducing the positive
cooperativity to be only -0.1 kJ/mol suggesting that the sulphonyl will act in an additive manner
with the adjacent P3 side chain
11

What is the cause of the connection between the X group and the engagement in positive
cooperativity with the adjacent P3 side chain? In order to address this question molecular
modeling studies were performed and relevant thrombin-ligand complex crystal structures were
analyzed. Molecular modeling of inhibitors 2 and 5 was carried out, as describe in the
Supplementary Information, starting with the thrombin bound crystal structure of the similar
inhibitors 7 (pdb: 2ZI2) and 10 (pdb: 2ZHQ), respectively. As shown in Figure 4, the binding
modes of the inhibitors with a carbonyl group and with the modeled sulphonyl group are overall
very similar except for some key differences. The carbonyl or the sulphonyl group form a similar
H-bond with the thrombin Gly216-NH- residue having a distance of 3.2 Å or 3.3 Å, respectively,
when the P3 side chain is an ethyl group. However, the positioning of this ethyl side chain in the
hydrophobic S3 pocket is different between the two inhibitors. In the presence of the carbonyl
group (inhibitor 7), the ethyl side chain was found to be directed towards the hydrophobic
residues of the S3 pocket while in presence of the sulphonyl group (inhibitor 2) the ethyl side
chain is directed away from the hydrophobic residues of the S3 pocket.
The binding mode from the crystal structure of inhibitor 10 (X = C=O, R = benzyl side
chain), and the analogous modeled structure of inhibitor 5 (X = SO2, R = benzyl side chain),
again showed that the two inhibitors also have an overall similar binding mode (Figure 4).
However, the H-bonds to Gly216-NH- residue are now are significantly different. In case of
inhibitor 10, the carbonyl forms a shorter and stronger H-bond with the Gly216-NH- residue at a
distance of 3.2 Å while the sulphonyl group forms a longer and weaker H-bond with the Gly216-
NH- residue at a distance of 3.6 Å. In addition the P3 hydrophobic side chains are positioned
somewhat differently in the thrombin S3 pocket. An analysis of the thrombin bound x-ray
crystal structure of inhibitor 10 (Figure 4) showed that the presence of the carbonyl group results
12

in the P3 benzyl side chain being positioned close to the hydrophobic side chain residue Leu99
(with an average distance of 4.08 Å) in the S3 pocket, thereby forming strong hydrophobic
interactions with the protein. On the other hand, the modeled structure of inhibitor 5, where the
carbonyl was replaced with a sulphonyl, indicated that the benzyl side chain will be located
farther (average distance of 4.92 Å) from the S3 hydrophobic surface than is the case for
inhibitor 10. The non-optimal placement of the benzyl side chain in the S3 pocket is also
reflected by the decrease in hydrophobic contact surface area from 389 Å2 with inhibitor 10 to
339 Å2 for inhibitor 5. Overall this analysis indicates that for the carbonyl group containing
inhibitors 7 and 10 an H-bond with the thrombin Gly216 residue can be established at the same
time as the P3 hydrophobic side chain is positioned to bind properly in the S3 pocket. On the
other hand, when the carbonyl is replaced with a sulfonyl group as in inhibitors 2 and 5 the
positioning of the P3 side chain in the S3 pocket, and the H-bond with the thrombin Gly216
residue are put in conflict with each other. In order to establish the optimal interaction of one a
less than optimal interaction of the other is needed and hence in the case of the benzyl side chain
inhibitor 5 both interactions are compromised. Hence the positive cooperativity that would have
otherwise been obtained is negated by this conflict.
b)
a)
Trp60D
Trp60D
3.6 Å 3.2 Å
3.2 Å
3.3 Å
Gly216
Gly216
Figure 4: a) Overlay of the crystal structure of inhibitor 7 (blue) and the modeled structure of
inhibitor 2 (red) in the thrombin active site. The carbonyl of inhibitor 7 forms a H-bond with the
13

protein Gly216 at a distance of 3.2 Å, while in inhibitor 2, one of the oxygens of the sulphonyl
forms the same H-bond with a distance of 3.3 Å but with the ethyl side chain redirected to
outside of the S3 pocket; b) Overlay of the crystal structure of inhibitor 10 (orange) and the
modeled structure inhibitor 5 (white) in the thrombin binding site. The carbonyl of inhibitor 10
forms an H-bond with the protein with a distance of 3.2 Å, while in inhibitor 5, one of the
oxygens of the sulphonyl forms an H-bond with a distance of 3.6 Å
In conclusion, this study demonstrates that the accuracy of SAR predictions would be
enhanced by the inclusion of the more detailed aspects of cooperativity that can be gleaned from
a structural analysis, an example of which is provided herein. The current study shows that the
enhancement in binding affinity between an H-bonding group and an adjacent hydrophobic
interaction through positive cooperativity can be highly dependent on the functional group’s
effect on each other’s ability to become optimally positioned within the active site. This study
also revealed a linear contribution of each Å2 of hydrophobic contact surface area of the
hydrophobic side chain. The carbonyl group allows a good positioning of the P3 side chain and
as a result each Å2 hydrophobic contact surface area of the P3 side chain contributes -0.1007
(kJ/mol) / Å2 to the binding affinity. The sulphonyl is a bioisosteric H-bond acceptor, but does
not allow the P3 side chain to be properly positioned in the S3 pocket, and as a result each Å2
hydrophobic contact surface area of the P3 side chain contributes only -0.0501 (kJ/mol) / Å2,
which is only about half of that contributed when the H-bond acceptor is a carbonyl group. More
studies of this type, including more crystal structures, are needed to reveal the factors affecting
the presence and magnitude of cooperativity. This information is important to further refine the
accuracy of SAR predictions when bioisostere tactics are employed for ligand optimization
efforts.
Conflict of interest: The authors declare that they have no conflict of interest.
14

Funding: This research did not receive any specific grant from funding agencies in the public,
commercial, or not-for-profit sectors.
References:
1. Bioisosteres in Medicinal Chemistry, First Edition. Edited by Nathan Brown, Published 2012
by Wiley-VCH Verlag GmbH & Co. K
2. Ciapetti, P. and Giethlen, B. in The Practice of Medicinal Chemistry, 2008, 3rd edition,
Elsevier.
3. Langdon, S. R.; Ertl, P.; Brown, N., ‎Mol. Inform 2010, 29, 366.
4. Wermuth, C. G., Drug Discov Today 2006, 11, 348.
5. Nasief, N. N.; Hangauer, D. J. Med Chem. 2014, 57, 2315.
6. Whitesides, G. M.; Krishnamurthy, V. M. Q. Rev. Biophys. 2005, 38, 385.
7. Lemieux, R. U. Acc. Chem. Res., 1996, 29, 373.
8. Wang, L.; Berne, B. J.; Friesner, R. A. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 1326.
9. Englert, L.; Biela, A.; Zayed, M.; Heine, A.; Hangauer, D.; Klebe, G. Biochem. et Biophys.
Acta. 2010, 1800, 1192.
10. Snyder, P. W.; Mecinovic, J.; Moustakas, D. T.; Tomas, S. W., III; Harder, M.; Mack, E. T.;
Lockett, M. R.; Heroux, A.; Sherman, W.; Whitesides, G. M. Proc. Natl. Acad. Sci. U.S.A.
2011, 108, 17889.
11. Stahl, M.; Rarey, M., J. Med. Chem. 2001, 44, 1035.
12. Williams, D. H.; Stephens, E.; O'Brien, D. P.; Zhou, M., Angew. Chem. Int. Ed. 2004, 43,
6596.
15

13. Warren, G. L.; Andrews, C. W.; Capelli, A.-M.; Clarke, B.; LaLonde, J.; Lambert, M. H.;
Lindvall, M.; Nevins, N.; Semus, S. F.; Senger, S.; Tedesco, G.; Wall, I. D.; Woolven, J. M.;
Peishoff, C. E.; Head, M. S., J. Med. Chem. 2005, 49, 5912.
14. Cheng, T.; Li, X.; Li, Y.; Liu, Z.; Wang, R., J. Chem. Inf. Model. 2009, 49, 1079.
15. Wang, R.; Lu, Y.; Wang, S., J. Med. Chem. 2003, 46, 2287.
16. Li, Y.; Han, L.; Liu, Z.; Wang, R., J. Chem. Inf. Model.2014, 54, 1700.
17. Said, A. M.; Hangauer, D. G. Eur. J. Med. Chem. 2015, 96, 405.
18. Muley, L.; Baum, B.; Smolinski, M.; Freindorf, M.; Heine, A.; Klebe, G.; Hangauer, D. J.
Med. Chem. 2010, 53, 2126.
19. Nasief, N.; Hangauer, D. Eur. J. Med. Chem. 2015, 90, 897.
20. M. T. Khayat, ProQuest, UMI Dissertations Publishing, 2013.
21. Kuhn, B.; Fuchs, J. E.; Reutlinger, M.; Stahl, M.; Taylor, N. R., J Chem Inf Model. 2011, 51,
3180.
22. Williams, M. A.; Ladbury, J. E. Methods Princ. Med. Chem. 2003, 19, 137.
23. Bissantz, C.; Kuhn, b.; Stahl, M. J. Med. Chem. 2010, 53, 5061.
24. ilsson, . m l inen, M.; Ivarsson, M.; Gottfries, J.; Xue, Y.; Hansson, S.; Isaksson, R.;
Fex, T., J. Med. Chem. 2009, 52, 2708.
25. Stubbs, M. T.; Oschkinat, H.; Mayr, I.; Huber, R.; Angliker, H.; Stone, S. R.; Bode,W. Eur.
J. Biochem. 1992, 206, 187.
26. Sturzebecher, J.; Sturzebecher, U.; Vieweg, H.; Wagner, G.; Hauptmann, J.; Markwardt, F.
Thromb. Res. 1989, 54, 245.
27. Cheng, Y.; Prusoff, W. H.; Biochem. Pharmacol. 1973, 22, 3099.
16

Graphical Abstract
17