Similar but different: Thermodynamic and structural characterization of a pair of enantiomers binding to acetylcholinesterase

Article abstract of DOI:10.1002/anie.201205113

Take a closer look: Unexpectedly, a pair of enantiomeric ligands proved to have similar binding affinities for acetylcholinesterase. Further studies indicated that the enantiomers exhibit different thermodynamic profiles. Analyses of the noncovalent interactions in the protein-ligand complexes revealed that these differences are partly due to nonclassical hydrogen bonds between the ligands and aromatic side chains of the protein. Copyright

Full text of DOI:10.1002/anie.201205113

.
Angewandte
Communications
DOI: 10.1002/anie.201205113
Molecular Recognition
Similar but Different: Thermodynamic and Structural Characterization
of a Pair of Enantiomers Binding to Acetylcholinesterase**
Lotta Berg, Moritz S. Niemiec, Weixing Qian, C. David Andersson, Pernilla Wittung-Stafshede,
Fredrik Ekstrçm,* and Anna Linusson*
In a previous high-throughput screening campaign, we
identified the racemate C5685 (1) as an inhibitor of Homo
sapiens acetylcholinesteraste (hAChE, Figure 1).[1] Structural
analysis of the complex formed between 1 and Mus musculus
AChE (mAChE) revealed that the substituted phenyl ring
interacts with the peripheral anionic
site (PAS) of mAChE, and that the
chiral N-ethylpyrrolidine moiety of
1 is directed towards the catalytic
site (CAS). However, owing to a dis-
ordered electron density map we
were unable to model the part of
1 that binds to CAS. Furthermore,
the electron density map of the side
Figure 1. The chemical
structure of 1 identified
chain of Tyr337 indicated two con-
formers. These results prompted us
to prepare enantiomerically pure
1 and characterize both enantiomers
at the kinetic, thermodynamic, and
structural levels, as reported herein.
as an inhibitor of hAChE
with a half-maximal
inhibitory concentration
(IC50) of 1.3 mm.
Scheme 1. Synthesis of (R)-6 and (S)-6. a) MeOH, H2SO4, reflux, 96%;
b) MeI, NaH, DMSO, RT, quantitative; c) NaOH, dioxane, 558C, 93%;
d) (COCl)2, DMF, MeCN, 08C!RT; e) (R)-2-aminomethyl-1-ethylpyrro-
lidine, Et3N, CH2Cl2, RT, 72% over two steps; f) (S)-2-aminomethyl-1-
ethylpyrrolidine, Et3N, CH2Cl2, RT, 74% over two steps.
The synthesis of the enantiomers started with deacetyla-
tion of 2,[2] followed by N-methylation to transform 3 into 4
(Scheme 1).[3] An ester hydrolysis gave the benzoic acid
derivative 5, which was further transformed into the acid
chloride and subsequently reacted with (R)- or (S)-2-(amino-
methyl)-1-ethylpyrrolidine to give enantiomerically pure (R)-
C5685 ((R)-6) and (S)-C5685 ((S)-6), respectively.
The two enantiomers have nearly identical dose–response
relationships with hAChE and mAChE (Figure 2 and
Table 1); this result was unexpected, since previous studies
[*] L. Berg, M. S. Niemiec, Dr. C. D. Andersson,
Prof. Dr. P. Wittung-Stafshede, Dr. A. Linusson
Department of Chemistry, Umeꢀ University
90187 Umeꢀ (Sweden)
L. Berg, Dr. C. D. Andersson, Dr. A. Linusson
Computational Life Science Cluster (CLiC), Umeꢀ University
90187 Umeꢀ (Sweden)
E-mail: anna.linusson@chem.umu.se
Dr. W. Qian
Laboratories for Chemical Biology Umeꢀ
Chemical Biology Consortium Sweden
Department of Chemistry, Umeꢀ University
90187 Umeꢀ (Sweden)
Dr. F. Ekstrçm
Swedish Defence Research Agency, CBRN Defence and Security
90182 Umeꢀ (Sweden)
E-mail: fredrik.ekstrom@foi.se
Figure 2. Dose–response curves used for determining IC50 values of
(R)-6 and (S)-6 for wild-type hAChE and mAChE (with the mAChE
concentration adjusted to a lower maximum activity than the hAChE
concentration, for clarity). The enzymatic activity is measured as the
increase in absorbance (DA) at 412 nm over time as a result of the
enzymatic hydrolysis of acetylthiocholine.[3] Means ꢀS.D. from tripli-
cate dose response curves (goodness of fit, R2 >0.99). Dose–response
[**] Acknowledgement is made to the Swedish Research Council and
Umeꢀ University for financial support. We thank Dr. D. Jansson
(Swedish Defence Research Agency) for assistance with the chiral
HPLC, Dr. M. Thunnissen (MAX-lab) for assistance during X-ray
data collection, and HPC2N (High Performance Computing Center
North) for HPC resources and technical support.
&
*
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201205113.
curves from the top: (R)-6 hAChE ( ), (S)-6 hAChE ( ), (S)-6 mAChE
~
( ), (R)-6 mAChE (&).
12716
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 12716 –12720

Angewandte
Chemie
Table 1: IC50 values and thermodynamic parameters for the binding of (R)-6 and (S)-6 to Mus musculus
and Homo sapiens AChE.[3]
together, there are four conformers
of the (R)-6/mAChE complex to be
considered: A and B conformers
with respect to Tyr337 and also A
and B conformers with respect to
the configuration of the pyrrolidine
nitrogen of the ligand. In contrast,
there is only one conformer of the
(S)-6/mAChE complex, as we do
[a]
Species
Ligand IC50
KD
[mm]
DG
DH
DS
ꢁTDS
[mm]
[kcalmolꢁ1
]
[kcalmolꢁ1
]
[calmolꢁ1 Kꢁ1
]
[kcalmolꢁ1
]
Mus musculus (R)-6
Mus musculus (S)-6
Homo sapiens (R)-6
Homo sapiens (S)-6
0.7
0.7
1.3
1.4
0.9
1.3
ꢁ8.1
ꢁ7.9
ꢁ5.6
ꢁ7.5
8.5
1.4
ꢁ2.5
ꢁ0.4
ꢁ0.4
3.6
1.7[b] ꢁ7.8
ꢁ7.4[b]
1.5[b]
2.2[b] ꢁ7.7
ꢁ11.3[b]
ꢁ12.1[b]
[a] Mean values from triplicate dose response curves. [b] Mean values from two ITC experiments.
of enantiomers binding to AChE have indicated that the
binding site exhibits a stereochemical preference.[4] To further
investigate the binding properties, we determined the ther-
modynamic binding profiles by using isothermal titration
calorimetry (ITC). The results of the ITC experiments
confirm that the enantiomers have similar binding strength
to both mAChE and hAChE (Table 1). However, the
enthalpic and entropic contributions to the two enantiomersꢀ
free energy of binding (DG) differ (Table 1). (S)-6 has a more
favorable binding enthalpy for both mAChE and hAChE, but
the entropic contribution is more pronounced for (R)-6, thus
giving them similar DG values. These observations are in
agreement with the entropy–enthalpy compensation phe-
nomenon reported for protein–ligand complexation.[5] Nota-
bly, here such compensation is observed for enantiomers with
similar ligand flexibility and desolvation.
To elucidate the structural basis of the difference in the
enantiomersꢀ binding properties, we determined the X-ray
crystal structures of the complexes (R)-6/mAChE and (S)-6/
mAChE at resolutions of 2.5 and 2.25 ꢁ, respectively
(Figure 3).[3,6] The electron density maps indicate that the
two enantiomers coordinate to the PAS in similar conforma-
tions, in accordance with the previously published structure.[1]
The data show that the complexes (R)-6/mAChE and (S)-6/
mAChE differ primarily in terms of their conformations in the
CAS (Figure 3).
The N-ethyl pyrrolidine moieties of the inhibitors extend
towards Trp86 at the bottom of the active-site gorge, in
conformations that are specific for each enantiomer. It is not
possible to distinguish between the two tetrahedral config-
urations of the amine nitrogen from the electron density map
of the complex with (R)-6, and so one major and one minor
conformer were included in the final structure (A-(R)-6 and
B-(R)-6, Figure 3).[7] In both cases the N-ethyl substituent of
(R)-6 is directed toward residues Gly121, Gly122, Tyr124, and
Phe297. In contrast, the N-ethyl substituent of (S)-6 occupies
a pocket formed by the aromatic rings of Tyr337, Phe338, and
Tyr341 (Figure 3b). The complexes formed between the two
enantiomers and mAChE also differ with respect to the
conformation of Tyr337. In the (R)-6/mAChE complex, the
side chain of Tyr337 was refined in two conformers of similar
occupancy (conformers A and B),[8] whereas only one (which
was equivalent to the A conformer) was observed for (S)-6/
mAChE (Figure 3b). In the B conformer, Tyr337 is directed
toward Phe338, closing off the pocket occupied by the N-ethyl
substituent of (S)-6. The A conformer of Tyr337 is common in
crystal structures of AChE, while the B conformer has been
previously observed in only a few structures.[1,9] Taken
Figure 3. Overlay of the (R)-6/mAChE and (S)-6/mAChE crystal struc-
tures, showing the binding conformations of the two enantiomers a) in
the binding cavity of mAChE and b) using a stick model and a subset
of the adjacent amino acids. The carbon atoms of (S)-6 are shown in
yellow, while those of the two conformers A/B-(R)-6 are shown in
green. A and B conformers of residue Tyr337 in the (R)-6/mAChE
complex are shown green, while Tyr337 in the (S)-6/mAChE complex is
shown in yellow.
not observe any conformational mobility of Tyr337 and/or the
ligand.
Noncovalent interactions in the crystal structures were
then explored by using quantum mechanics.[3] More specifi-
cally, the geometries of reduced all-atom systems were
optimized using density functional theory (DFT) with the
M06-2X hybrid functional.[10] The electrostatic potentials
were mapped on the calculated electron density surfaces
and close interaction points were identified and analyzed.[11]
The conformations of (S)-6 and (R)-6 appear to be stabilized
in the CAS by a number of noncovalent interactions
Angew. Chem. Int. Ed. 2012, 51, 12716 –12720
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 12717

.
Angewandte
Communications
Figure 4. Identification of noncovalent interactions in the (S)-6/mAChE complex. a) Geometries of reduced all-atom systems (see text for
modeling details), with ligand carbon atoms colored in yellow. b) Electrostatic potential mapped on the calculated electron density surface
visualized by a spectrum from red (electron-rich) through orange, yellow, green, and blue to purple (electron-poor). c) Geometry-optimized
structures shown using stick models with close interaction points highlighted by dashed lines. Residues Trp86, Tyr124 (right), Tyr337 (left), and
water molecule W2211 are shown. Atoms of the ligand and Tyr124 that were not included in the DFT calculations have been added for clarity. Red
and blue dashed lines indicate hydrogen bonds and van der Waals or electrostatic interactions, respectively.
(Figures 4 and 5),[3] including van der Waals and electrostatic
interactions as well as both classical and nonclassical (here of
CH···Y type) hydrogen bonds. Hydrogen bonds[12] of the
CH···Y type (where Y represents for example arene or
oxygen) are common in biological systems, and their role and
importance have been widely discussed.[13] The geometry
optimizations of the all-atom systems using M06-2X clearly
resulted in a fine-tuning of the hydrogen-bond orientations in
terms of both distances and angles compared to the crystal
structures with added hydrogen atoms using molecular
mechanics.
In the case of the (R)-6/mAChE complex, specific
interaction patterns were established for each of the four
systems based on the different conformers identified from the
crystal structures. Note that the DFT geometry optimizations
converged to similar conformations (within experimental
error) of both conformers of the ligand and Tyr337.[3] This
result supports the anticipated stability of the four geometries
(Figure 5). The N-ethyl pyrrolidine group of (R)-6 interacts
Figure 5. Noncovalent interactions in CAS between (R)-6 and mAChE.
with Tyr337, Trp86, Tyr124, and a water molecule through
hydrogen bonds and/or electrostatic and van der Waals
interactions in all four conformations. In the A-(R)-6-
ATyr337 model, an internal hydrogen bond between the
amide oxygen atom and the amine group of the pyrrolidine
ring was observed (Figure 5a). In addition, three hydrogen
bonds of the CH···O type were observed between the N-ethyl
pyrrolidine group and both hydroxy groups of Tyr124 and
Tyr337 and a water molecule. Van der Waals interactions with
Trp86 were also observed (see Figure 5b–d for the interaction
patterns of other conformers). We propose that the entropy
change for the binding of (R)-6 is more favorable than that for
the binding of (S)-6, because in the former case, the ligand
and protein can form attractive interaction patterns that allow
an ensemble of multiple conformations, thus suggesting
a greater degree of flexibility. However, it should be noted
that the experimental X-ray data do not allow for detailed
water analysis to elucidate solvation/desolvation effects. The
importance of conformational entropy has been highlighted
Geometry-optimized structures are shown using stick models (ligand
carbon atoms colored green) with close interaction points highlighted
by dashed lines. (a) and (b) show the complexes featuring the A
conformer of Tyr337 and the two alternative configurations of the
amine (i.e. A-(R)-6-ATyr337 and B-(R)-6-ATyr337), while (c) and (d)
show the B conformer of Tyr337 and the corresponding two config-
urations of the amine (A-(R)-6-BTyr337 and B-(R)-6-BTyr337). Residues
Trp86, Tyr124 (right), Tyr337 (left), and water molecule W2126 are
shown. Atoms of the ligand and Tyr124 that were not included in the
DFT calculations have been added for clarity. Red and blue dashed
lines indicate hydrogen bonds and van der Waals or electrostatic
interactions, respectively.
in several recent studies,[5,14] which have shown cases where
ligand preorganization by conformational restriction lead to
entropic penalties in protein–ligand complex formation
(these entropic penalties are contrary to previous expect-
ations).[14a,b] A possible explanation is that the reduced
flexibility of the ligand limits the conformational freedom of
the formed complex. Both (R)-6 and the protein examined in
12718 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 12716 –12720

Angewandte
Chemie
the present study can adopt different conformers that are
stabilized by several relatively weak interactions. The net
result is that the binding of (R)-6 to mAChE is more
entropically favorable than that of (S)-6.
observed entropy–enthalpy compensation. Analyses of the
noncovalent interactions revealed that nonclassical hydrogen
bonds of the CH···O and CH···arene types have a major
influence on the enantiomersꢀ binding properties. We believe
that the identification and characterization of nonclassical
hydrogen bonds will play increasingly important roles in
elucidating the formation of protein–ligand complexes gen-
erally, and interactions involving AChE in particular.
The stabilizing interactions between the single conforma-
tion of (S)-6 and the CAS are similar to those observed for
(R)-6 (Figure 4c). The amide oxygen atom forms both an
internal hydrogen bond to the protonated amine of the
pyrrolidine ring and a hydrogen bond with the hydroxy group
of Tyr124. (S)-6 also forms interactions with Trp86 and CH···O
type hydrogen bonds with a water molecule. The positioning
of the N-ethyl group between residues Tyr337 and Phe338 is
stabilized by the internal hydrogen bond (Figure 3b). Geom-
etry optimization of the system derived from the (S)-6/
mAChE complex revealed the presence of an NCH···arene
hydrogen bond between the aromatic ring of Tyr337 and one
of the hydrogen atoms on the methylene group a to the
nitrogen atom of the pyrrolidine ring (Figure 6). The distance
between the methylene hydrogen atom and the plane of the
Received: June 29, 2012
Revised: October 12, 2012
Published online: November 19, 2012
Keywords: aromatic interactions ·
.
density functional calculations · molecular recognition ·
nonclassical hydrogen bonds · stereoselectivity
[1] L. Berg, C. D. Andersson, E. Artursson, A. Hçrnberg, A.-K.
Tunemalm, A. Linusson, F. Ekstrçm, PLoS ONE 2011, 6,
e26039.
[2] D. Renneberg, P. B. Dervan, J. Am. Chem. Soc. 2003, 125, 5707.
[3] For further details, see the Supporting Information.
[4] a) M. McKinney, J. H. Miller, F. Yamada, W. Tuckmantel, A. P.
Kozikowski, Eur. J. Pharmacol. 1991, 203, 303; b) A. Saxena, N.
Qian, I. M. Kovach, A. P. Kozikowski, Y. P. Pang, D. C. Vellom,
Z. Radic, D. Quinn, P. Taylor, B. P. Doctor, Protein Sci. 1994, 3,
1770; c) P. Camps, R. ElAchab, J. Morral, D. MuÇoz-Torrero, A.
Badia, J. E. BaÇos, N. M. Vivas, X. Barril, M. Orozco, F. J.
Luque, J. Med. Chem. 2000, 43, 4657.
[5] C. H. Reynolds, M. K. Holloway, ACS Med. Chem. Lett. 2011, 2,
433.
[6] The crystal structures have been deposited in the Protein Data
Bank (PDB), PDB codes 4ARA and 4ARB.
[7] The maximal atom deviation of the two conformers is 0.82 ꢁ and
the apparent occupancies are 0.89 and 0.11.
Figure 6. Detailed illustration of the interaction between Tyr337 of
mAChE and the N-ethyl pyrrolidine group of (S)-6. a) The electrostatic
potential mapped on the calculated electron density surface visualized
by a spectrum from red (electron-rich) through orange, yellow, green,
and blue to purple (electron-poor). b) Reduced stick model with close
interactions indicated with dashed lines. The CH···arene hydrogen
bond is highlighted in red.
[8] The maximal atom deviation of the two Tyr337 conformers is
3.99 ꢁ and the apparent occupancies are 0.44 and 0.56.
[9] a) Y. Bourne, H. C. Kolb, Z. Radic, K. B. Sharpless, P. Taylor, P.
Marchot, Proc. Natl. Acad. Sci. USA 2004, 101, 1449; b) F.
Ekstrçm, A. Hçrnberg, E. Artursson, L. G. Hammarstrçm, G.
Schneider, Y. P. Pang, PLoS ONE 2009, 4, e5957.
[10] Y. Zhao, D. Truhlar, Theor. Chem. Acc. 2008, 120, 215.
[11] The noncovalent interactions were determined by visual inspec-
tion of the electrostatic potential mapped on the calculated
electron density at an isovalue of 0.007 electrons/Bohr3, as well
as distances and angles. Further details and all close contacts in
the complexes can be found in the Supporting Information.
[12] a) E. Arunan, G. R. Desiraju, R. A. Klein, J. Sadlej, S. Scheiner,
I. Alkorta, D. C. Clary, R. H. Crabtree, J. J. Dannenberg, P.
Hobza, H. G. Kjaergaard, A. C. Legon, B. Mennucci, D. J.
Nesbitt, Pure Appl. Chem. 2011, 83, 1619; b) E. Arunan, G. R.
Desiraju, R. A. Klein, J. Sadlej, S. Scheiner, I. Alkorta, D. C.
Clary, R. H. Crabtree, J. J. Dannenberg, P. Hobza, H. G.
Kjaergaard, A. C. Legon, B. Mennucci, D. J. Nesbitt, Pure
Appl. Chem. 2011, 83, 1637.
[13] a) G. R. Desiraju, T. Steiner, The weak hydrogen bond in
structural chemistry and biology, Oxford University Press, New
York, 1999; b) S. Sarkhel, G. R. Desiraju, Proteins Struct. Funct.
Bioinf. 2004, 54, 247; c) G. R. Desiraju, Chem. Commun. 2005,
2995; d) C. Bissantz, B. Kuhn, M. Stahl, J. Med. Chem. 2010, 53,
5061; e) P. Hobza, K. Muller-Dethlefs, Non-covalent interactions.
Theory and experiment, The Royal Society of Chemistry, Cam-
bridge, 2010; f) W. A. Herrebout, M. A. Suhm, Phys. Chem.
Chem. Phys. 2011, 13, 13858; g) S. Melandri, Phys. Chem. Chem.
areneꢀs p system was 2.5 ꢁ and the C–H–p plane angle was
1658. We suggest that this hydrogen bond between the
activated C–H group and the aromatic ring of Tyr337 is the
primary reason for (S)-6 having a greater enthalpy of binding
than its enantiomer. CH···arene interactions typically have
energies of approximately 1.5–2.5 kcalmolꢁ1.[15] The mean
CH/p plane distance for CH···arene hydrogen bonds in small
organic molecules is 2.5–2.8 ꢁ, and the mean C–H–p plane
angle is 148–1578, depending on the strength of the proton
donor involved.[15a,16] The closer the angle is to 1808, the
stronger the interaction and the shorter the distance.[12]
CH···arene interactions have also been detected in protein–
ligand complexes,[13b,17] and their potential importance for
ligand binding in AChE has been previously considered.[18]
Here, we have identified a hydrogen bond between a C–H
group in (S)-6 and a tyrosine in mAChE; this hydrogen bond
may explain the greater enthalpy upon complexation of (S)-6,
relative to its enantiomer.
In this study we have shown that two enantiomers have
similar binding affinities for AChE while differing in their
thermodynamic profiles. By combining crystallography and
computational chemistry, we were able to rationalize the
Angew. Chem. Int. Ed. 2012, 51, 12716 –12720
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 12719

.
Angewandte
Communications
Phys. 2011, 13, 13901; h) S. Scheiner, Phys. Chem. Chem. Phys.
2011, 13, 13860.
[15] a) M. Nishio, Phys. Chem. Chem. Phys. 2011, 13, 13873; b) L. M.
Salonen, M. Ellermann, F. Diederich, Angew. Chem. 2011, 123,
4908; Angew. Chem. Int. Ed. 2011, 50, 4808.
[16] O. Takahashi, Y. Kohno, S. Iwasaki, K. Saito, M. Iwaoka, S.
Tomoda, Y. Umezawa, S. Tsuboyama, M. Nishio, Bull. Chem.
Soc. Jpn. 2001, 74, 2421.
[17] a) T. Steiner, G. Koellner, J. Mol. Biol. 2001, 305, 535; b) M.
Harigai, M. Kataoka, Y. Imamoto, J. Am. Chem. Soc. 2006, 128,
10646; c) S. K. Panigrahi, G. R. Desiraju, Proteins Struct. Funct.
Bioinf. 2007, 67, 128.
[18] a) H. Dvir, H. L. Jiang, D. M. Wong, M. Harel, M. Chetrit, X. C.
He, G. Y. Jin, G. L. Yu, X. C. Tang, I. Silman, D. L. Bai, J. L.
Sussman, Biochemistry 2002, 41, 10810; b) G. Koellner, T.
Steiner, C. B. Millard, I. Silman, J. L. Sussman, J. Mol. Biol.
2002, 320, 721.
[14] a) A. P. Benfield, M. G. Teresk, H. R. Plake, J. E. DeLorbe, L. E.
Millspaugh, S. F. Martin, Angew. Chem. 2006, 118, 6984; Angew.
Chem. Int. Ed. 2006, 45, 6830; b) J. E. DeLorbe, J. H. Clements,
M. G. Teresk, A. P. Benfield, H. R. Plake, L. E. Millspaugh, S. F.
Martin, J. Am. Chem. Soc. 2009, 131, 16758; c) B. Baum, L.
Muley, M. Smolinski, A. Heine, D. Hangauer, G. Klebe, J. Mol.
Biol. 2010, 397, 1042; d) C. Diehl, O. Engstrçm, T. Delaine, M.
Hꢂkansson, S. Genheden, K. Modig, H. Leffler, U. Ryde, U. J.
Nilsson, M. Akke, J. Am. Chem. Soc. 2010, 132, 14577; e) J. M.
Ward, N. M. Gorenstein, J. Tian, S. F. Martin, C. B. Post, J. Am.
Chem. Soc. 2010, 132, 11058.
12720 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 12716 –12720