Content uploaded by Francesca Spyrakis
Author content
All content in this area was uploaded by Francesca Spyrakis on May 27, 2014
Content may be subject to copyright.
Current Topics in Medicinal Chemistry, 2011, 11, 000-000 1
1568 0266/11 $55 00+ 00 © 2011 Bentham Science Publishers Ltd
Protein Flexibility and Ligand Recognition: Challenges for Molecular
Modeling
Francesca Spyrakis1,2,*, Axel BidonChanal3, Xavier Barril3,4 and F. Javier Luque3,*
1Department of General and Inorganic Chemistry, University of Parma, Parma, Italy, 2INBB, Biostructures and Biosys-
tems National Institute, 3Departament de Fisicoquímica and Institut de Biomedicina (IBUB), Facultat de Farmàcia,
Universitat de Barcelona, 08028, Barcelona, Spain, 4ICREA
Abstract: The intrinsic dynamics of macromolecules is an essential property to relate the structure of biomolecular sys-
tems with their function in the cell. In the field of ligand-receptor recognition, numerous evidences have revealed the limi-
tations of the lock-and-key theory, and the need to elaborate models that take into account the inherent plasticity of bio-
molecules, such as the induced-fit model or the existence of an ensemble of pre-equilibrated conformations. Depending on
the nature of the target system, ligand binding can be associated with small local adjustments in side chains or even the
backbone to large-scale motions of structural fragments, domains or even subunits. Reproducing the inherent flexibility of
biomolecules has thus become one of the most ch allenging issues in molecular modeling and simulation studies, as it has
direct implications in our understanding of the structure-function relationships, but even in areas such as virtual screening
and structure-based drug discovery. Given the intrinsic limitation of conventional simulation tools, only events occurring
in short time scales can be reproduced at a high accuracy level through all-atom techniques such as Molecular Dynamics
simulations. However, larger structural rearrangements demand the use of enhanced sampling methods relying on modi-
fied descriptions of the biomolecular system or the potential surface. This review illustrates the crucial role that structural
plasticity plays in mediating ligand recognition through representative examples. In addition, it discusses some of the
most powerful computational tools developed to characterize the conformational flexibility in ligand-receptor complexes.
Keywords: Molecular recognition, conformational flexibility, ligand-receptor complex, induced-fit, molecular dynamics, drug
design.
INTRODUCTION
Biological processes are essentially based on the ability
of biomolecules to recognize each other within a noisy envi-
ronment, where many look-alike molecules are present. Ex-
amples are given, for instance, by antibodies recognizing
antigenic peptides [1, 2], regulatory proteins binding specific
DNA sequences [3, 4], or enzymes catalyzing and transform-
ing their substrate [5]. A binding event, like the formation of
a protein-ligand complex, normally occurs in aqueous solu-
tion and, under thermodynamic equilibrium conditions, the
standard free energy of binding determines the reaction ten-
dency. The Gibbs’s free energy is decomposable into enthal-
pic and entropic contributions. The former is phenome-
nologically attributed to the formation of non-bonded inter-
actions such as salt bridges, hydrogen bonds, hydrophobic
contacts, cation-pi interactions and metal complexation. The
latter is given by the loss of conformational mobility of both
the protein and the ligand, and by the release of water mole-
cules from hydrophobic surfaces to the solvent [6].
The ability of the scoring functions implemented in dock-
ing programs to retain the delicate balance between enthalpy
and entropy is critical for the prediction of the binding of
*Address correspondence to these authors at the Department of General and
Inorganic Chemistry, University of Parma, Parma, Italy, INBB, Biostruc-
tures and Biosystems National Institute; Tel: +390521905669;
Fax: +390521905556; E-mail; francesca.spyrakis@unipr.it
Departament de Fisicoquímica and Institut de Biomedicina (IBUB), Facultat
de Farmàcia, Universitat de Barcelona, 08028, Barcelona, Spain;
Tel: +34934024557; Fax: +34934035987; E-mail; fjluque@ub.edu
ligands to a macromolecular target [7]. Though there is con-
sensus that the problem of generating reasonable ligand ori-
entations is quite satisfactory, recognizing near-native geo-
metries and predicting their affinities is still achieved with
limited success [8-11]. Besides the specific nature of the
interactions that mediate ligand binding, several factors ac-
count for the challenging question of developing a physi-
cally-based accurate scoring function. First, one cannot ne-
glect the fundamental role played by water molecules, which
rearrange around the ligand and the protein active site upon
the complex formation. The inclusion of conserved water
molecules in the active site of proteins and the correct esti-
mation of their energetic contribution is crucial for docking
analysis. Thus, one should be able to simulate the insertion
of a ligand into a solvated pocket, displace only irrelevant
waters and estimate the total binding free energy including
the contribution of the retained ones. Numerous efforts have
been made to predict structural waters [12-16], improving
both binding poses and energetic predictions [16-18]. Never-
theless, a general strategy for the solvation treatment in
docking and virtual screening is still elusive. Another aspect
to be considered is the proper assignment of the ionization
and tautomerization states of the ligands and of residues in
the active sites, which can be affected by the microenviron-
ment at the binding site [19]. At this point, it is worth noting
that waters can buffer the active site changing the number of
hydrogen-bond donors and acceptors [20, 21]. Finally, one
cannot forget that proteins are inherently flexible systems
able to undergo functionally relevant conformational transi-
2 Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 Spyrakis et al.
tions even in native state conditions [22-26], and that flexi-
bility is often essential for the protein function [24].
It is generally assumed that the recognition process is
regulated by the interplay between the interaction energy
gained from the proper alignment of the ligand in the binding
site, and the elastic energy required to deform the interacting
molecules. Savir and Tlusty suggested that the conforma-
tional changes experienced by both ligand and target could
arise from an evolutionary optimization of the binding proc-
ess, since the target-ligand mismatch improves the recogni-
tion specificity in a noisy environment as a conformational
proofreading mechanism [27]. Moreover, as reported by Mit-
tag et al. [28], the conformational disorder may provide an
electrostatic and steric advantage due to the generation of
fluctuating electrostatic fields, allowing disordered proteins
to bind several surfaces of the partners by altering the buried
surface area. In this context, the conformational variability
increases plasticity and malleability, since the same molecule
can interact with different partners, and provides an evolu-
tionary advantage, facilitating alternative splicing, domain
shuffling, protein modularity and mutations.
Knowledge of the intrinsic dynamics of macromolecules
has direct implications in our understanding of the relation-
ship between structure and function, and particularly in areas
such as virtual screening and structure-based drug discovery.
Numerous evidences have revealed the limitations of the
lock-and-key theory and the need to elaborate more sophisti-
cated models that account for the plasticity of biomolecules.
In this context, this work pursues to illustrate the implica-
tions between ligand binding and target flexibility. A series
of representative examples, involving either small local ad-
justments in specific residues or even large-scale motions of
structural fragments, will be discussed. Moreover, the broad
range of structural rearrangements that might mediate ligand
binding raises challenging issues for the capability of model-
ing techniques to capture the conformational flexibility of
biomolecules. Given the intrinsic limitation of conventional
modeling tools, only events occurring in short time scales
can be reproduced by all-atom simulation methods, and
larger structural rearrangements demand the use of enhanced
sampling techniques. This review also discusses some of the
most powerful computational tools developed to characterize
the role of structural flexibility in the formation of ligand-
receptor complexes.
PROTEIN FLEXIBILITY AND LIGAND BINDING
The energy landscape of a protein is usually displayed as
a folding funnel, lined by unfavorable states collapsing
through various routes into few favorable folded states, made
by sub-ensembles of structurally and energetically equivalent
conformations of the protein [14, 23, 29-32]. The use of a
single or even few structures (typically provided by X-ray
crystallography or NMR) neglects or poorly represents the
existence of these conformational substates, which are dy-
namically interchanging depending on “external” conditions
such as ionic strength, pH, temperature and the presence of
ligands. Generally the highest flexibility is related to con-
formational rearrangements in loops, but there is increasing
evidence about the occurrence of larger structural modifica-
tions in the structural elements that delineate the binding site
or even global rearrangements. Depending on the degree of
flexibility, proteins could be basically classified into three
main categories, i) rigid proteins, showing only small side-
chain adjustments; ii) flexible proteins, able to undergo
larger motions around hinge points or structural elements of
the binding site, and iii) intrinsically unstable proteins, which
are only stabilized by the entrance of a ligand into the bind-
ing pocket.
The strong belief that the protein conformational diver-
sity represents an evolutionary and functional advantage
pushed the lock-and-key model further away, and now an
appropriate handling of the binding pocket flexibility, or
more generally of the protein dynamics, is fundamental to
shed light into the protein-ligand binding. Unfortunately,
predicting which conformation a flexible target will adopt
upon the binding of a specific molecule is still a challenging
question and, whether the occurring conformational changes
have to be attributed to the ligand entry (i.e, induced-fit
model), or depend on the intrinsic structural and dynamical
properties of the protein is a subject of current interest.
This dilemma was first addressed in the late fifties by
Koshland, who proposed the induced-fit theory, where the
ligand plays a key role in inducing the target conformational
change [33], and later by Monod, who suggested the exis-
tence of a preexisting ensemble of conformations, where the
ligand only selects the most suitable one and shifts the con-
formational equilibrium [34]. According to the induced-fit
model, the ligand binding appreciably reduces the conforma-
tional entropy of the target bringing an energetic penalty to
the overall binding process (Fig. (1)). On the contrary, in the
model of equilibrium conformational states, the ligand se-
lects and stabilizes one of the available conformations, with-
out significant reduction in the conformational entropy but
favorably contributing to the binding free energy (Fig. (1))
[35]. The induced-fit model could easily account for local
motions as side-chain reorientations or transitions between
isomeric states. Nevertheless, cooperative changes or rear-
rangements of entire domains can be hardly attributed to the
entrance of small ligands into a slightly flexible binding
pocket.
Experimental and computational studies have strongly
supported the view of an existing equilibrium of conforma-
tional substates in native conditions [36, 37], as originally
reported by Monod, Wyman and Changeux [22, 30, 38-42],
relating the different conformational states to an intrinsic
predisposition and dynamics of proteins. This approach
would also agree with the view of a dynamic energy land-
scape in which the protein’s “open” form corresponds to the
lowest energy minimum, while the “closed” form is trapped
in a less stable minimum [43-45]. As an example (see be-
low), Xu et al. have suggested that binding of ligands to ace-
tylcholinesterase (AChE) likely affects the pre-existing equi-
librium dynamics selecting the most suitable conformation
rather than promoting the formation of non-native structures
[46].
More recently, Okazaki and Takada [47], and Sullivan
and Holyoak [48] sustained that protein flexibility could be
better explained by joining both intrinsic dynamics and in-
duced-fit theories, since the selection of a pre-existing
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 3
Fig. (1). Modification of the protein energy landscape upon ligand
binding in the models of (a) induced fit and (b) conformational
states. In the induced-fit model the ligand promotes a structural
rearrangement in the binding site of the receptor, which reduces the
conformational entropy. In the model of conformational states the
ligand selects and stabilizes one of the pre-existing conformations,
shifting the conformational equilibrium towards the bound state.
conformation could be reasonably followed by additional
rearrangements induced by a specific molecule [49]. In par-
ticular, to shed more light on the dualism between induced-
fit and population-shift mechanisms, Okazaki and Takada
developed and applied an implicit ligand binding model
combined with a double basin Hamiltonian, finding that both
models have their own range of applicability. Thus, strong
and long-range interactions favor essentially an induced-fit
mechanism, as in the case of protein-protein or protein-DNA
interactions [50-52], whereas weak and short-range contacts
better support the population-shift model, as observed for
antigen-antibody binding and substrate binding to enzymes
[53-55]. The real situation likely resembles both mecha-
nisms, since the ligands might first select a partially closed
conformation, and then induce conformational changes up to
the closed state [47].
While these issues deserve further clarifications, drug
design approaches have to somehow predict and simulate the
intrinsic flexibility of proteins, trying to find the best com-
promise between accuracy, reliability and computational
resources. This explains the intense effort paid to the devel-
opment of docking algorithms dealing with flexible binding
sites [24]. Simple and less time-consuming methods include
only binding pocket side-chain flexibility [56-59], while
more complex approaches use constrained geometric simula-
tions [60], elasticity network theory [61, 62], docking into
ensembles of conformations derived from experimental or
computational techniques [17, 63-67], or even use of mo-
lecular dynamics (MD) simulations to post-process rigid
docking solutions [68]. The availability of a large number of
structures a priori represents the best strategy to dock
ligands into the most suitable and reliable conformation,
while predicting protein-ligand interactions by modeling a
single receptor structure would only be useful for ligands
targeting a single particular narrow state of the conforma-
tional ensemble. Nevertheless one should be aware that -due
to current methodological limitations- ensemble docking
does not necessarily translate into better predictions and can,
in fact, degrade the results [69, 70]. Therefore, a judicious
choice of the target conformations seems necessary for a
successful identification of new hits in virtual screening stu-
dies.
LOCAL STRUCTURAL CHANGES INDUCED UPON
BINDING
Most protein side chains undergo small structural ad-
justments during ligand binding. Thus, the inspection of the
side chains motions promoted by the ligand in 63 complexes
involving 20 proteins supports the assumption that protein
side chains move as little as necessary in order to achieve a
collision-free complex [59]. In fact, Zavodszky and Kuhn
reported that ligand binding did not typically involve
changes between rotamers, but mostly small (<15 degrees)
changes in the torsional angles. In turn, these findings sug-
gest that docking into the apo forms of their binding targets
should a priori lead to satisfactory predictions about the best
poses of ligands when combined with simple energy minimi-
zation or steric optimization procedures.
In spite of the preceding comments, local adjustment of
the side chains of residues have been noticed in certain cases,
such as the low-molecular-weight protein tyrosine phospha-
tase (LMW-PTP), where Trp49 is responsible for the sub-
strate specificity but also for the conformational exchange
quenching associated with the monomer-dimer transition
[71], or in the copper-binding proteins superoxide dismutase
[72] and azurin [73], where a single histidine regulates the
accessibility of the copper-binding site. In these cases, the
inclusion of local flexibility would be mandatory for the suc-
cessful prediction of the ligand pose. Thus, binding sites can
be seen as a combination of preorganized regions, whose
structure is well preserved due to a network of contacts with
neighboring residues, and other regions exhibiting a larger
structural plasticity likely due to a limited involvement in
intramolecular contacts or to a large exposure to the solvent
[21].
Acetylcholinesterase, an enzyme that catalyzes the hy-
drolysis of the neurotransmitter acetylcholine, is well suited
to illustrate the subtle balance between structurally preserved
and flexible regions in the binding site. The 3D structure of
Torpedo californica AChE (initially solved at 2.8 Å resolu-
tion [74], and later refined to 2.5 Å resolution [75]) shows a
deep and narrow gorge (around 20 Å long) that penetrates
halfway into the enzyme and widens out close to its base.
Fourteen highly conserved aromatic residues line a substan-
tial portion of the surface of the gorge. The catalytic active
site is located at the bottom of the gorge, and the pi-electron
distribution of Trp84 assists binding of the quaternary moi-
ety of choline through cation-pi interactions. In addition,
4 Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 Spyrakis et al.
another triptophane residue (Trp279) at the entrance of the
gorge defines a peripheral site, which might act as an initial
binding site for substrate entry.
The structural complexity of the active site gorge ac-
counts for the large diversity of AChE inhibitors, which ex-
hibit notable differences in the mechanism of inhibition and
selectivity [76, 77]. In spite of such chemical diversity, the
main structural features of the residues at the catalytic and
peripheral binding sites are generally preserved in the differ-
ent AChE complexes with reversible inhibitors, except for
Phe330 at the catalytic site and Trp279 at the peripheral one.
Thus, these residues adopt drastic conformational changes
for different inhibitors (Fig. (2)). In the catalytic pocket, both
tacrine [78] and huprine X [79] bind the enzyme through
stacking of the aminoacridine ring with the aromatic rings of
Trp84 and Phe330. In particular, the side chain of this latter
residue is characterized by a N-C!-C"-C# ($1) torsional angle
of ~160°. The interaction of (%)-huperzine A [75] involves a
cation-pi interaction between the amino group of the inhibi-
tor with the aromatic rings of Trp84 and Phe330. In this
case, the Phe330 side chain adopts a conformation defined
by $1 ~ -170°, which thus mimics the conformation found for
the AChE complex with edrophonium [78]. Finally, donepe-
zil extends from the catalytic site to the peripheral one along
the gorge, and its benzyl ring interacts through stacking in-
teraction with Trp84, while the piperidine ring lies onto the
benzene moiety of Phe330, whose side chain shows a tor-
sional angle of ~ -130° [80]. A similar orientation is found
for the binding of decamethonium [78].
The structural features of the binding of AChE inhibitors
at the peripheral binding site are more delicate due to the
conformational flexibility of Trp279, as noted in the X-ray
crystallographic structures available for several peripheral
site ligands [81] and dual binding site inhibitors [82-85]. In
fact, three main arrangements of the indole ring of Trp279
can be identified upon inspection of the X-ray crystallo-
graphic structures (Fig. (2)). The first orientation is charac-
terized by dihedral angles N-C!-C"-C3 ($1) and C!-C"-C3-C3a
($2) close to -60° and -80°, respectively, as found in the apo
form of the enzyme [74]. Similar values are observed in
complexes with catalytic binding site inhibitors such as
huprine X [79] or (-)-huperzine A [75], and with peripheral
binding site inhibitors such as propidium, decidium and gal-
lamine [81]. Moreover, this arrangement is also found in
complexes with dual binding site inhibitors, such as de-
camethonium [78], donepezil [80], tacrine(10)-hupyridone
[84], and anti-TZ2PA6 [82], as well as in the complex with
fasciculin [86]. Dihedral angles close to -120° ($1) and +50°
($2) are found in the complexes with tacrine(8)-4-amino-
quinoline [85] and NF595 [83]. Finally, an alternative orien-
tation defined by dihedral angles close to -160° ($1) and -
120° ($2) is found in the complex with syn-TZ2PA6 [82].
As a final remark, it is worth noting how an apparently
minor chemical change, such as the different length of the
methylenic tether in bis(5)-tacrine and bis(7)-tacrine, leads
to a different arrangement of the indole ring of Trp279 [85],
as the former binds to the peripheral site by imposing an
apo-like conformation for Trp279 ($1: -76.3°, $2: -87.0°),
whereas this residue is completely reoriented in the latter ($1:
Fig. (2). Representation of selected residues in both catalytic and
peripheral sites of acetylcholinesterase (AChE). Top: residues
Phe330 and Trp84 at the catalytic binding site for AChE complexes
with tacrine, (%)-huperzine A and donepezil. Bottom: residue
Trp279 in the peripheral site for the AChE complexes with donepe-
zil, bis(7)-tacrine and syn-TZ2PA6.
-121.4°, $2: +43.7°). As noted by Xu et al. [46], the confor-
mational plasticity observed for Trp279 should not be con-
ceived as the result of an induced-fit mechanism, but rather
as the preferential binding of the inhibitor to pre-existing
conformational states, as those conformations are accessed in
MD simulations of the apo form of the enzyme [46]. In this
context, rational drug design could largely benefit from the
identification of regions with a high intrinsic conformational
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 5
plasticity in the binding site, facilitating the right choice of
conformational states for successful docking studies [87].
SMALL-AMPLITUDE BACKBONE MOVEMENTS
Small-molecule binding sites in proteins usually tend to
be more rigid than the rest of the protein. For instance, in the
case of enzymes, changes in the backbone skeleton are usu-
ally not significant and the functional atoms move generally
less than 1 Å upon ligand binding [88]. Nevertheless, some
enzymes and receptors, as cytochromes or the liver X recep-
tors [89], have adaptable binding sites, which enable them to
recognize a wider range of ligands. Though large conforma-
tional changes in the backbone are relatively rare, the struc-
tural features of the binding site may be completely trans-
formed when they occur. Moreover, these changes mainly
affect regions in the binding site where flexibility plays a
functional role, as will be illustrated here for different pro-
teins.
Heat-Shock Protein 90
The oncology target Hsp90, and in particular its ATP
binding site, has attracted much interest in recent years.
Early structures of the N-terminus domain already identified
a loop that could adopt two different conformations (open
and closed) depending on the nature of the bound ligand
(Fig. (3)) [90]. Later, the inhibitor PU3 -initially thought to
bind the open conformation [91]- revealed the existence of a
third conformation, known as helical. While the open-closed
transition affects the periphery of the binding site, the helical
conformation results in the creation of a large lipophilic
pocket at the bottom of the cavity, which changes substan-
tially the volume and the physicochemical properties of the
binding site [92]. Interestingly, the crystallographic structure
of the full-length dimer has shown that the flexible region is
part of a larger loop that undergoes a major conformational
transition (about 40 Å) as part of the chaperoning cycle [93].
This suggests that the flexibility of this loop, which is a func-
tional requirement, may be exploited by non-natural ligands
to reshape the binding site and obtain better complementar-
ity. Unfortunately the knowledge of where conformational
changes may take place cannot be used (at present) to predict
which molecule will induce what change, and it is down to
serendipity that ligands are discovered (e.g. random screen-
ing, naïf design) and that the conformational change is eluci-
dated (as it may be incompatible with crystal formation).
This system also illustrates how experimentally observed
conformations may not be treated as equal. Thus, due to its
larger volume and lipophilicity, the helical conformation
offers many more binding opportunities but is energetically
penalized and, unless the conformational strain is taken into
account, docking results may degrade [69].
Renin
This protease, which converts angiotensinogen into an-
giotensin I, has been studied as an antihypertensive target for
many years. The structure of the protein and its complex
with the peptidic substrate was used to design peptidomimet-
ics, often of high molecular weight and suboptimal pharma-
cokinetics (e.g. the oral bioavailability of Aliskiren is
Fig. (3). Superimposition of open, closed and helical conformations
of Hsp90. In the open conformation, the binding site becomes con-
siderably larger due to the opening of a hydrophobic pocket (black
surface). Upon dimerization, this loop (shown in the center, above
the black surface) undergoes a major conformational rearrange-
ment, closing onto the binding site.
approximately 2.5% in humans [94]). More recently, a small
molecule of unrelated structure (identified by means of high-
throughput screening) was shown to bind in a newly created
cavity. The ligand inserts itself between residues Trp40 and
Phe114, causing a partial melting of two !-strands that form
a lid over the binding site [95]. Molecules exploiting this
cryptic site can achieve better ligand efficiency and pharma-
cokinetic properties than most peptidomimetics [96].
Kinase DFG-Out Binders
A common mechanism of control of kinases is the phos-
phorylation of the so-called activation loop (AL). Recogni-
tion of the AL by an upstream kinase is thought to involve a
partial unfolding, and its regulatory function is also likely to
involve some conformational diversity. The flexibility of this
protein segment can also be attested by the lack of electron
density and, therefore, missing residues in some X-ray struc-
tures. This flexibility has been exploited by several kinase
inhibitors that bind to a hydrophobic pocket created by the
displacement of a phenylalanine in the conserved DFG se-
quence in the AL. DFG-out ligands were initially reported
for p38 [97], and they are often referred to as allosteric in-
hibitors or indirect competitors because, although the inhibi-
tor binds in the same pocket as ATP, the substrate is com-
peted off by the conformational change induced by the
ligand rather than by the ligand itself. As there is more se-
quence diversity in the pocket created by DFG-out binders,
these inhibitors can confer increased selectivity, a property
of critical importance in the kinase field. As the displace-
ment of the AL happens at relatively long timescales, an-
other property of DFG-out ligands is slower association and
6 Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 Spyrakis et al.
dissociation rates, which may affect the pharmacokinetics of
the compounds [98].
Protein Tyrosine Phosphatase 1B
PTP1B is a drug target for the treatment of type II diabe-
tes and obesity that has proven to be extremely challenging.
Many inhibitors acting on the phosphotyrosine binding site
have been described, but potency is heavily dependent on the
presence of a negative charge, which greatly damages the
pharmacokinetics properties [99]. In this context, the discov-
ery of allosteric inhibitors opened a new window of opportu-
nity [100]. Previously, two distinct conformations had been
described for the so-called WPD loop (residues 179-184),
which lines the binding site. In the apo form, it adopts an
open conformation, whereas substrate binding induces a
closing of the loop, thus reducing the size of the cavity, now
tightly fitting around the phosphotyrosine [101]. Interest-
ingly, this conformational change is coupled to a larger am-
plitude transition in the !7-helix (residues 287-295), which
is located about 20 Å away (Fig. (4)). In the WPD-closed
conformation this helix is ordered and in contact with the
!3-helix, but in the WPD-closed conformation it is disor-
dered and separated from the rest of the protein. The allos-
teric inhibitors bind to the latter form, occupying a hydro-
phobic pocket that appears upon displacement of Trp291
(part of the !7-helix). It would seem then that these ligands
stabilize a conformation able to preclude the closing of the
WPD-loop. In addition to providing a completely different
chemotype, in this case the allosteric inhibitors bind in an
area that is poorly conserved amongst phosphatases, making
these compounds highly selective for PTP1B [100]. Very
recently, the WPD-open conformation (apo form) has been
exploited to identify non-competitive inhibitors. Although
they bind to the phosphotyrosine binding site rather than
competing with the substrate, they simply stabilize the inac-
tive conformation, reducing the concentration of the catalyti-
cally competent enzyme. Unlike most direct inhibitors, these
molecules do not bear a negative charge and can thus cross
membranes, achieving cellular activity [102].
Overall, even though the rigidity of small-molecule bind-
ing sites has certainly been one of the factors contributing to
the success of structure-based drug design, molecules ex-
ploiting other mechanisms of action -such as protein-protein
inhibitors or allosteric modulators- tend to bind in areas that
are much more flexible, thus adding another layer of uncer-
tainty and complexity to rational design [103, 104]. Consid-
ering our increased knowledge about biological systems and
the benefits provided by molecules acting on non-typical
binding sites (e.g. expansion in the number of druggable
proteins, increased selectivity, potential to activate as well as
inhibit) these type of effectors –as well as protein flexibility–
are expected to play an ever-increasing role in drug design
[105, 106].
LARGE-SCALE PROTEIN MOTIONS
Many proteins experience significant conformational
transitions through hinge bending motions in order to adopt
the active state and perform their function. As reported by
Kumar et al. in an outstanding review [107], three type of
Fig. (4). Superposition of active (light grey) and inactive (dark
grey) conformations of PTP-1B. In the active form, the active site
loop is closed and !7-helix (in the top of the figure) is packed
against the rest of the protein. In the inactive form, the loop is open
(shown in the center), leaving a large and shallow binding site.
Allosteric inhibitors compete with the !7-helix for binding, thereby
stabilizing the inactive conformation.
movements can be identified: i) motions involving fragments
of the protein chains, ii) hinge-bending motions involving
protein domains, and iii) hinge-bending motions between
covalently unconnected subunits. Generally these move-
ments are not attributed to an induced-fit mechanism, but as
resulting from the conformational variability experienced by
proteins. Accordingly, the transition from an “open” to a
“closed” conformation is not forced by the incoming ligand,
but by an equilibrium shift towards an already available con-
formation. Interestingly, the physical principles that regulate
the shifting of the conformational equilibrium at the bottom
of a binding funnel should be comparable to the principles
driving the identification of the native structures in folding
funnels [107].
Relevant conformational adjustments have been reported
in a variety of cases, such as the acyl carrier protein [108] or
the glutamate receptor [109], where the ligand binding trans-
forms a local adjustment into a global allosteric effect. Other
systems where large-scale conformational transitions take
place are the NS3 RNA helicase from hepatitis C virus
[110], the protein kinase Zap70 involved in T-cell activation
[111], the DNA-binding domain of PU1 [112], calsensin
[113], the ATP-binding cassette transporters [114], the het-
erodimeric transcription factor CBF [115], the insulin-like
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 7
growth factor binding protein [116], the C-terminal zinc fin-
gers of human MTF-1 [117], the HIV-1 nucleocapsid NCp7
[118], or the CTP:glycerol-3-phosphate cytidylyl-transferase
dimeric enzyme [119], as reported in detail by Valente et al.
[35]. Few selected cases will be examined in more detail in
the following.
Kinase Proteins
Proteins belonging to the kinase family represent a nota-
ble example of systems undergoing large-scale movements.
These proteins share a well preserved catalytic core for the
transfer of the !-phosphate of ATP to serine, threonine or
tyrosine in different target enzymes [120]. This activity is
regulated at different levels, such as phosphorylation by
other kinases, membrane and organelle localization via scaf-
folding proteins, protein-protein or domain-domain interac-
tions through regulator modules [121]. Protein kinases could
be viewed as molecular switches that can adopt two extreme
states: an “on” (active) conformation and an “off” (inactive or
minimally active) one. While all kinases show structurally
very similar active conformations, many differences have
been observed in the off states, where the formation of a cata-
lytically active pocket is not required [120]. From a struc-
tural viewpoint, these proteins contain two distinct lobes (the
smaller N lobe and the larger C lobe), forming a deep cleft at
the interface where ATP is bound (Fig. (5)). A highly con-
served phosphate loop, also known as the P loop, contains a
flexible Gly-rich sequence, which can thus closely contact
the phosphates and easily bind small inhibitors in the ab-
sence of ATP. The AL occupies a more central position and
needs to be phosphorylated to allow the protein adopting the
active conformation. Since a proper regulation of kinases is
crucial for cellular growth and development, misregulation
can lead to cell transformation or even cancer. Therefore,
understanding the conformational flexibility of these struc-
tural elements is fundamental for both a regulatory and a
drug therapy perspective.
Hyeon and co-workers computationally investigated the
dynamics of nucleotide binding in the catalytic domain of
protein kinase A (PKA) [121], as previously done by differ-
ent groups for other kinases as adenylate kinase (AK) [53,
122, 123], c-Src [124, 125], Abl [126], and cyclin-dependent
kinase 5 [127]. PKA is activated by the phosphorylation of a
threonine in the AL and repressed by the binding of a regula-
tory subunit. Its catalytic domain has been crystallized in
diverse conformations that reveal differences in both ATP
and substrate-binding lobes, and significant movements in
many secondary structure elements [128-130]. Activation of
PKA involves an open-close conformational transition in-
duced by ATP binding. Interestingly, the major changes are
found in regions distant from the binding pocket. The com-
putational analyses showed that PKA maintains an open con-
formation until ATP enters the catalytic domain. This event
leads to a clockwise rotation of the C-terminal tail that al-
lows Phe327 to contact ATP and enhances the compaction
between small and large lobes of the catalytic domain. These
local dynamical effects occur at faster rate than the global
dynamics, suggesting that the allosteric signal propagates
from the binding pocket to the exterior sites through a net-
work of native contact pairs. When the nucleotide occupies
Fig. (5). Superposition of the open (inactive; dark grey) and closed
(catalytically active; light grey) forms of (top) Protein Kinase A and
(bottom) Adenylate Kinase. ATP is shown in space fill cartoon. The
key regions are labeled and indicated by black arrows.
its binding site, the native contacts are temporarily removed
to accommodate the incoming substrate. Some of the secon-
dary motifs partially unfold to subsequently refold in the so-
called cracking mechanism (occurring in a 1-5 ms time
range), which implies a non-monotonic change in flexibility
along the transition route. In particular, local unfoldings are
exploited to relieve specific areas of high stress during con-
formational deformations and to overcome high energy bar-
riers [131].
An unfolding (cracking) model was also proposed for
AK [123], though an innovative approach has been sug-
gested by Hanson and co-workers [54]. This enzyme cata-
lyzes the conversion of ATP and AMP in two ADP mole-
cules. It can adopt distinct (open and closed) limit conforma-
8 Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 Spyrakis et al.
tions. The large hinge movement assisting the transition be-
tween open and closed forms is critical for function [132],
since the binding of ATP and AMP induces the closing mo-
tion of both the ATP lid domain and the nucleoside mono-
phosphate binding (NMP bind) domain (Fig. (5)), which thus
keeps water molecules far from the catalytic pocket during
the phosphoryl transfer reaction. The authors observed that
the enzyme experiences a dynamical equilibrium between
open and closed forms, with a preference towards the closed
state, even in the absence of substrates, which appears to be
unnecessary to induce the formation of the active-site cavity.
In turn, this large-scale conformational motion seems to be
an intrinsic feature of the protein. The natural flexibility and
the major stability of the closed state undoubtedly help in
minimizing the energy required to shift the equilibrium to-
wards the active configuration characterized by a higher
catalytic rate. Accordingly, ligand binding does not block the
enzyme in a single conformation, but shifts the ATP lid do-
main distribution doubling its closing rate, and restricts the
range of conformational changes. Once the reaction has
taken place, the products can be released following the lid
opening, which indeed represents the limiting step. This en-
zyme thus appears to be designed to exploit large-amplitude
conformational dynamics to achieve the fast catalytic turn-
over.
Calmodulin
CaM is a calcium binding protein able to bind and regu-
late a number of different targets as kinases, phosphatases,
NO synthase, Ca2+ pumps, proteins active in motility and
many others. Thus, it can participate in multiple biological
processes, such as inflammation, metabolism, apoptosis,
muscle contraction, etc. Calmodulin consists of two structur-
ally similar N- (nCaM) and C-terminal (cCaM) globular do-
mains bonded by a flexible interdomain linker. The two do-
mains display quite different properties, in particular in
terms of flexibility and Ca2+ affinity [133]. The Ca2+ binding
to the apo/closed form induces significant conformational
changes leading to the holo/open form (Fig. (6); [134, 135]).
NMR and MD simulations studies demonstrated that the
cCaM has greater Ca2+ affinity and is more open than nCaM
when bound to Ca2+ [136-138]. It was also suggested that
cCaM could be partially unfolded in native conditions [139,
140], and that the conformational exchange may imply a
more complicated mechanism than a simple two-state (open-
closed) process, as suggested by the analysis performed by
Tripathi and Portman using a coarse-grained variational
model [133]. While the transition from the closed to the open
form of the more flexible nCaM domain does not require a
cracking event, the same shifting in the cCaM domain only
occurs via a transient unfolding in the helix linker between
the binding loops. Similar processes also occur upon binding
of CaM with its biological peptidic substrates (Fig. (6);
[141]). The large-scale transition leading from the holo CaM
form to the complexed structure occurs via a hinge-based
motion, involving the breaking of the long helix-shaped
linker, which partly unfolds around the hinge and generates
two different helices. This high flexibility represents an in-
teresting strategy for enabling the binding of a number of
peptides with different sequences, and is the expression of a
complex energy funnel characterized by a large number of
isoenergetic minima around the funnel bottom [107].
Glutamate Dehydrogenase
A significant hinge motion has also been observed for
glutamate dehydrogenase (GDH), which catalyzes the re-
versible oxidative deamination of L-glutamate into 2-
oxoglutamate and ammonia using NAD+ as a cofactor. GDH
is formed by two trimers stacked on top of each other, whose
subunits are, at least, composed of three domains (Fig. (7)).
The NAD+ binding domain is located in the upper part of the
subunits, overhung by a long protrusion known as antenna.
This last part likely plays a relevant role in regulating the
enzyme, allosterically controlled by numerous ligands and
cofactors as GTP, ATP, ADP, or by hydrophobic compounds
as palmitoyl-CoA and steroid hormones ([142] and refer-
ences therein). GDH undergoes significant conformational
rearrangements during each catalytic cycle (Fig. (7)). When
the NADH enters and binds into the coenzyme binding
Fig. (6). Different conformations of calmodulin: (a) apo structure, (b) holo structure (Ca2+ ions are represented by spheres), and (c) structure
of a calmodulin-peptide complex.
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 9
pocket, the coenzyme domain experiences a 18° rotation
with respect to the other domain, via a hinge bending move-
ment, to close down both the coenzyme and the natural sub-
strate (i.e. glutamate). At the same time, the ascending helix
of the antenna moves towards the pivot helix of the adjacent
subunit, which performs a counterclockwise rotation along
the helical axis. These movements induce a compression of
the entire hexamer, since the upper three pairs of subunits
move as a unique rigid entity towards the lower subunits,
compressing the cavity at the interface.
COMPUTATIONAL STUDIES AND MODELING OF
CONFORMATIONAL FLEXIBILITY
Much of our knowledge on the structural features of
biomolecular targets comes from high-resolution experimen-
tal techniques. Structural models obtained from X-ray crys-
tallography are usually static, time- and space-average struc-
tures solved at extremely low temperatures. Therefore, they
provide information on one or at most few structural con-
formations of the proteins, but they cannot account for a
Fig. (7). Top: Superposition of the open (dark grey) and closed (light grey) forms of Glutamate Dehydrogenase. Bottom, left: Inset of the
open apo conformation. ATP is represented in space fill cartoons. Bottom, right: Inset of the closed catalytically active form. The GTP in-
hibitor, the NADH coenzyme and the substrate (glutamate; shown in dark grey) are represented in space fill cartoons.
10 Current Topics in Medicinal Chemis try, 2011, Vol. 11, No. 2 Spyrakis et al.
comprehensive understanding of the inherent protein flexi-
bility. Nevertheless, in the last few years technical advances
such as time-resolved measurement on single crystals have
demonstrated to be extremely valuable to shed light into dy-
namical aspects of fast processes, such as the breathing mo-
tions of cavities coupled to ligand migration [143]. NMR
techniques provide ensembles of low-energy conformations,
and are particularly useful for proteins or polypeptides diffi-
cult to crystallyze or that provide too disordered crystals
[144]. Moreover, it has the large advantage of being per-
formed under conditions and in solutions that mimic the bio-
logical environment. Unfortunately, the intrinsic experimen-
tal limitations of NMR make this technique applicable to a
limited set of targets. Even if the equilibrium between ener-
getically close conformational states may be investigated,
larger movements can be hardly predicted.
MD simulations represent an alternative strategy to ex-
plore the inherent flexibility of proteins and the impact of
structural fluctuations on the binding of ligands or in protein-
protein interactions [145]. In essence, classical MD simula-
tions rely on the iterative numerical calculation of instanta-
neous forces acting on the atoms of a system, which in turn
consists of particles that move in response to their interac-
tions according to the equations of motion defined in classi-
cal mechanics. MD can a priori be routinely applied to in-
vestigate a wide range of dynamic processes, and coupled to
statistical mechanics permits to derive thermodynamic prop-
erties of the system. Accordingly, one can predict, for in-
stance, changes in the binding free energy of a ligand or the
mechanisms and energetic consequences of conformational
alterations in proteins. Nevertheless, applications are largely
limited in practice by the time scale of the dynamical proc-
esses, which in biomolecular systems can range from femto-
seconds to hours.
Conventional atomistic MD simulations are well suited to
study local elastic vibrations of atoms/groups of atoms and
rotations of side chains, and can therefore assist in identify-
ing small structural rearrangements triggered upon ligand
binding. For instance, MD simulations of AChE complexed
with syn-TZ2PA6 [146] revealed a change in the side chain
of Trp286 (numbering in mouse AChE; this residue is
equivalent to Trp279 in Torpedo californica AChE) from the
orientation found in the apo form of the enzyme to that seen
in the X-ray crystallographic complex with the inhibitor
[82]. This change, which took place in a 2 ns window, makes
the Trp286 side chain shooting off the hydrophobic core and
adopting another conformation, which permits a stable stack-
ing with the inhibitor. Moreover, this change was accompa-
nied by other major changes in His287, Trp86, Tyr133,
Tyr337 and Phe338. A similar structural rearrangement of
the tryptophan residue at the peripheral site has also been
reported in an independent study of the AChE complex with
indole-tacrine hybrids, which were designed as dual binding-
site inhibitors [147].
MD simulations provide an efficient way to generate
unbiased structures to be used in docking or virtual screening
analyses [148-150]. While inexpensive and fast docking al-
gorithms can be used to scan large compound libraries and
reduce their size, more accurate MD simulations can be ap-
plied when few ligand candidates remain. Thus, MD simula-
tions can be used not only for the refinement of docked
complexes, but also during the preparation of the protein
receptor before docking in order to optimize its structure and
account for protein flexibility [151]. As examples, we quote
two independent studies on the docking of small ligands,
which report a receptor-based pharmacophore method that
relies on a collection of multiple protein structures to ac-
count for protein flexibility in HIV-1 protease [152], and to
the molecular mechanism of resistance of the mutant
T66I/M154I HIV-1 integrase to the inhibitor L-731,988
[153]. In particular, this strategy can be relevant for studies
dealing with membrane-bound proteins (i.e., G-protein cou-
pled receptors), which have proven very difficult to crystal-
lize for examination by X-ray crystallography, and are often
too insoluble for NMR analysis [154, 155]. Nevertheless, the
success of MD simulations coupled to docking will depend
on the structural diversity of the sampled conformations,
which in turn will be an issue of particular relevance in those
proteins with a high degree of conformational flexibility and
when docking involves compounds with diverse chemical
scaffolds.
The suitability of atomistic MD techniques to character-
ize slower motions such as bending of domains through
hinge regions or allosterism is more questionable, as the time
scales of those conformational changes are not accessible via
conventional MD techniques. To address this issue one has
to resort to enhanced sampling techniques [156-161]. One
possibility is the use of simplified representations of the
biomolecular system that enhance sampling of certain de-
grees of freedom at the expense of others (typically those
related to the fastest motions). The goal is to enhance sam-
pling of the “slow” motions via simplified descriptions
through integration of the “fast” degrees of freedom into a
few (i.e., coarse graining) ones [162]. To this end, the system
is modeled as a collection of beads. As the number of beads
decreases, the simulation is less expensive and the system
that can be simulated is larger. However, developing an ac-
curate and transferable parametrized force field capable of
describing the general dynamics of systems becomes in-
creasingly difficult as the graining is ‘coarser’. Coarse-
grained MD approaches, where some of the fine atomistic
details of the system are smoothed over or averaged out, are
promising to analyze large-scale structural changes, such as
reshaping of binding pockets. Nevertheless, while these
techniques can be informative about the general trends of
protein dynamics, the absence of atomistic details limits se-
riously the impact of the sampled conformations for docking
experiments. Nevertheless, interesting hybrid methods com-
bining coarse-grained models of proteins with an atomistic
description of the active site have been recently developed,
to enhance the potential of these approaches in examining
molecular recognition processes [163].
Enhanced sampling can also be achieved through tech-
niques that rely on the modification of the conventional MD
sampling. In this case, the system is forced to explore the
conformational space by facilitating the escape from local
energy minimum wells by using non-Boltzmann sampling.
This is achieved, for instance, in the Locally Enhanced Sam-
pling method by considering a given number of non-
interacting copies of the fragment to be explored (i.e., a resi-
due side chain or a ligand), whereas the interaction of each
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 11
fragment with the rest of the system is reduced by a suitable
factor from their original magnitudes [164]. Another exam-
ple is the use of multiple copies of the system that are simul-
taneously explored by MD, such as Replica Exchange (RE)
MD [165]. Alternatively, enhanced sampling can also be
attempted via the explicit modification of the potential sur-
face. The aim is to reduce the time spent by the simulated
system in a local energy minimum well, which would thus
favor the transition from the conformational region corre-
sponding to such well to another. Accordingly, the alteration
of the potential surface reduces the propensity of energy
wells to act as conformational traps and promotes the system
to sample the rest of the available conformational space. This
approach includes techniques such as umbrella sampling
[166] and accelerated MD [167].
In the following we examine in more detail a selection of
computational techniques that have been applied to explore
the impact of conformational flexibility on the interaction
between ligands and their biomolecular targets.
Replica Exchange Methods
The original RE method, first reported by Swendens and
Wang in 1986 [168], was developed to enhance conforma-
tional transitions by crossing energy barriers in the rugged
energy landscape of biomolecules [169, 170]. This tech-
nique, also known as parallel tempering, involves running a
number of simulations at different temperature, and exchang-
ing temperature and coordinates value every fixed number of
steps based on Metropolis criteria (Fig. (8)). Thus, improved
samplings are obtained by exchanging complete configura-
tions and allowing low temperature systems to access a rep-
resentative set of low-temperature regions of phase space.
Since the simulation of multiple replicas requires a propor-
tional higher computational effort, the set of temperatures
should be chosen to ensure that no replica becomes trapped
in local minima, and the number of replicas should be large
enough to ensure the swapping of adjacent replicas.
RE methods have been generally applied to investigate
folding mechanisms and intermediate state structures (see for
instance refs. [171-175]). The number of studies devoted to
rational drug design has increased in the last years. For in-
stance, Verkhivker et al. [176] used parallel tempering simu-
lations to rank the different complexes according to the aver-
age ligand-protein interaction energies in the binding of
SB203386 inhibitor to HIV-1 protease. This approach
yielded better sampling of the inhibitor conformational space
during the binding process compared to MC simulations,
where the protein was trapped in local minima even at high
temperature. Nagashima and co-workers developed a new
software framework enhancing the efficiency of conforma-
tional sampling by RE [177]. The toolkit employed object-
oriented design to overcome the complexity associated with
biomolecules and parallelization, and to better interact with
other software components, in the perspective of being ap-
plied to receptor-drug docking analyses.
Different RE variants have been reported in the literature.
Berne and co-workers developed a RE method with solute
tempering variant able to explicitly include the solvent con-
tribution, but also to reduce the number of replicas by avoid-
ing the evaluation of solvent-solvent interactions [178]. The
strategy allows the potential energy to scale with the tem-
perature in a way that the molecule becomes hotter while the
waters stay cold. The system is thus divided into two parts,
the protein and the waters or, eventually, the protein and its
solvation shell as the central group, and the remaining water
as the bath. Then, the temperature of the central group is
changed, while the temperature of the bath is kept at the
same target value. In this way the number of required repli-
cas dramatically reduces, since the acceptance probability is
independent from the solvent-solvent interaction energy.
Fig. (8). Schematic representation of replica exchange formalism. Several replicas of the system are simulated at different temperatures. En-
hanced sampling is achieved by allowing the systems at different temperatures to exch ange complete configurations. The inclusion of higher
temperature systems ensures that the lower temperature ones can access a representative set of low-temperature regions of phase space.
12 Current Topics in Medicinal Chemis try, 2011, Vol. 11, No. 2 Spyrakis et al.
Simmerling et a l. applied a hybrid implicit/explicit solvation
model, a REMD variant where simulations are performed
with a fully explicit solvent, but the exchange probability is
determined by combining the first solvation shell and a con-
tinuum solvation model [179]. The consequent reduction of
the system size significantly reduces the computational cost,
while maintaining a very good agreement with results ob-
tained from the standard explicit solvent REMD.
Instead of investigating a set of different temperatures, a
set of modified potentials is used in Hamiltonian REMD, as
proposed by Kwak et a l. [180]. Analogously the technique
designed by Hritz and Oostenbrik is based on a Hamiltonian
REMD scheme, but adopts soft-core interactions between
parts of the systems mostly contributing to define high en-
ergy barriers [181]. Differently from other Hamiltonian
REMD algorithms, a lower number of replicas is required in
this case, with locally larger differences between the indi-
vidual Hamiltonians.
An alternative RE method, known as distance replica
exchange method, has been used by Lou and Cukier to in-
vestigate the conformational fluctuations of the apo form of
AK [122], which undergoes a large fluctuation in the transi-
tion between open and closed conformations linked to the
adjustment of the AMP-binding and lid domains. The ap-
proach is used to sample the system along a reaction coordi-
nate monitoring the distance between the AMP-binding and
core domains. As the reaction coordinate decreases, the lid
domain reduces its mobility moving toward more con-
strained fluctuations, until reaching a stable closed-form con-
formation, while the more stable core structure is well main-
tained at all stages along the reaction coordinate. These re-
sults suggest that a closed form state can be sampled even in
the absence of ligand. When compared to other techniques,
this method demonstrated to greatly accelerate the rate and
the extent of configurational sampling.
More recently, Rodinger et al. applied the distributed
replica method as an efficient Boltzmann sampling of con-
formational space that enables large-scale computing [182].
Thus, one of the fundamental drawbacks of REMD is the
large CPU time required to simulate a complex biological
system, since the replicas must run synchronously and a
dedicated and homogeneous cluster is necessary for an effi-
cient implementation of the algorithm. In the distributed rep-
lica method multiple replica of the system covering a range
of temperature or of reaction coordinate are independently
simulated, but the replicas, instead of experiencing a pair-
wise exchange, are stochastically moved one at a time.
Finally, RE techniques have also been combined with
free energy perturbation (FEP) and finite difference thermo-
dynamic integration (FDTI) in order to design new methods,
respectively identified as REFEP and RETI, able to estimate
the relative free energies of systems involving large reor-
ganizations of the environment [183]. These techniques have
been applied to the calculation of the hydration free energy
of methane and water. Compared to other standard methods,
REFEP and RETI showed lower random sampling and stan-
dard errors and very little hysteresis, plus an excellent
agreement with experimental data, with a little or no extra
computational cost.
Accelerated Molecular Dynamics
McCammon and coworkers developed the Accelerated
Molecular Dynamics method [167, 184], based on previous
works by Voter [185], to improve and enhance the explora-
tion of the free energy landscape of large biological systems.
In this method a bias potential is added to the true one reduc-
ing the energy barriers between states by raising the potential
energy in the minima. A threshold energy is defined so that
below this energy, called the boost energy, the simulation is
performed on the modified potential, and when the system
has an energy above the threshold, the simulation is per-
formed on the true potential.
When the bias potential is applied, the probability to es-
cape from a potential basin increases depending on the boost
energy value and the form of the bias potential. In the formu-
lation proposed by Hamelberg et al., the boost bias potential
is of the form:
() ()()
()()
rVE
rVE
rV
!+
!
="
#
2
,
where E is the boost energy, and ! is a tunable parameter
that controls the deepness of the modified potential energy
basins.
High values of the parameter ! will produce small boost
bias potentials, and accordingly the modified potencial en-
ergy basins will resemble the ones of the normal potential
and no effective sampling gain will be achieved. After con-
sidering different combinations of the boost energy and !
parameters, Hamelberg et a l. concluded that the best choice
combines an ! value close to E–Vmin, where Vmin is the value
of the potential energy minimum nearest to the starting struc-
ture, and a value of E greater than Vmin but sufficiently small
to prevent the random walk of the system over the potential
energy surface.
This technique has been used by Markwick and cowork-
ers to explore the intrinsic conformational flexibility of pro-
tein GB3 [186], and by Grant et al. to investigate the con-
formational switching in Ras protein [187]. Moreover, it has
been used to examine the folding mechanism of a beta-
hairpin, trpzip2, thus enabling the generation of multiple
protein folding and unfolding trajectories in relatively short
simulations [188], and the identification of folding pathways.
Metadynamics
This technique [189] has been successfully applied to a
wide variety of systems in different fields ranging from bio-
physics [190-194] to materials science [195-199], crystal
structure prediction [200-203] or chemistry [204-208], as a
useful tool to reconstruct free energy surfaces and to simu-
late rare events. In this method, the evolution of the system
is biased by adding a history-dependent potential energy
function of a determined set of collective variables to the
Hamiltonian of the system. The potential is constructed
through successive addition of Gaussian functions that act as
repulsive potentials to prevent the system from revisiting
points of the free energy surface (Fig. (9)). Consequently, the
system can escape the wells in the rugged landscape and
efficiently explore it. The sum of Gaussian functions de-
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 13
posed along the trajectory of the system is then used to re-
construct the free energy.
A delicate aspect of the method is to identify the vari-
ables that are of particular relevance for the motion of the
biomolecular system and that are difficult to sample by con-
ventional simulation techniques [189]. Ideally the collective
variables should i) clearly distinguish between the initial
state, the final states and the intermediates, ii) should de-
scribe all the slow events that are relevant to the process of
interest, and iii) their number should not be too large in order
to avoid a too long simulation time to fill the free energy
surface. Unfortunately, there is no a priori recipe for finding
the correct set of collective variables, and in many cases a
trial-and-error process is required to detect the most adequate
choice. These variables are functions of the coordinates of
the system. The simplest definition involves geometry-
related variables, such as interatomic distances, angles or
torsions. However, more elaborate definitions are also valid,
such as the potential energy, which is particularly useful in
phase transitions [209] definitions, normal modes and essen-
tial coordinates derived from essential dynamics, which have
been used to explore the conformational space of peptides
[210], or specific variables in proteins, such as the ‘helicity
of the backbone’ or the ‘dihedral correlation’ [211]. A his-
tory-dependent dynamics is constructed in the space of these
variables in order to compensate the underlying free energy.
As stated before, the use of methods with atomic detail,
like all-atom explicit MD simulations, provide sufficient
accuracy to investigate large fluctuations and conformational
changes, but the computational time needed to get enough
sampling is often too high. With metadynamics, the time
required to explore large portions of conformational space is
drastically reduced as the biasing potential forces the system
to evolve to a fixed final state through non-previously sam-
pled conformations. Thus, it is possible for example to inves-
tigate large conformational changes, like the folding energy
landscape of proteins [212], or the transition between open
and closed states of a kinase protein (CDK5; [127]) with
reasonable computational resources.
Another scenario where metadynamics have proved to be
valuable is the inclusion of the receptor flexibility in docking
studies [213] and the evaluation of small ligand-induced con-
formational changes. As metadynamics is coupled to an MD
simulation, it is possible, in principle, to reproduce the bind-
ing pathway of a ligand with a receptor from the solvent to
the binding site, thus allowing the necessary rearrangements
in the side chains of residues to occur. For example, the
method has been applied to study the folding inhibition of
the HIV-1 protease with a small peptide [214], where the
large number of rotatable bonds and the necessity to include
the flexibility of the receptor limits the use of simple MD or
common docking methods.
Although metadynamics simulations imply a substantial
computational cost, its advantage relies in the fact that it
makes possible the identification of bottlenecks or unfavor-
able interactions that occur along the ligand’s binding path.
This is clearly illustrated by studies of the translocation
process of ampicillin through OmpF in Escherichia coli
[215], or of the trimethylamonium migration through the
gorge of acetylcholinesterase [216]. In both cases, a good
description of the mechanism and energetics of ligand migra-
tion is obtained, which in turn could be used for the design
of more potent and novel drugs.
CONCLUSIONS
Protein flexibility undoubtedly plays a critical role in
determining molecular recognition. The interaction between
Fig. (9). Schematic representation of time-dependent reconstruction of the multidimensional free energy. The potential is constructed as a
sum of gaussians centered along the trajectory of the collective variables. The dynamic evolution is labeled by the number of gaussians.
14 Current Topics in Medicinal Chemis try, 2011, Vol. 11, No. 2 Spyrakis et al.
a drug-like compound and its macromolecular target can
involve either local changes in the conformational distribu-
tion of the side chain of specific residues in the binding site,
small amplitude readjustements in the backbone of structural
elements, or even large-scale conformational alterations af-
fecting the spatial location of domains in the protein. Model-
ing efforts in drug design generally disregard these effects,
as they represent a formidable challenge to computational
and simulation studies. Nevertheless, it is also clear that a
precise knowledge of the structural plasticity of the target
and its implication in mediating the binding of a drug is ex-
tremely valuable to assist the development of more potent,
selective compounds in structure-based drug design. Impor-
tantly, such knowledge can also lead to the discovery of
novel effective strategies to modulate the activity of a bio-
molecular target, either by identifying alternative binding
pockets present in specific conformations of the protein or
by exploiting distinct pharmacophoric features of the binding
site associated with conformational transitions. Overall, this
information will be translated into new opportunities for
drug discovery, such as enhanced rates of hit detection in
pharmacophore-guided virtual screening or the identification
of lead compounds incorporating novel scaffolds.
MD techniques are extremely powerful tools that allow
us to gain insight into the link between target flexibility and
ligand binding. They provide a firm basis on which structural
information and biological data may be reconciled. All-atom
molecular dynamics simulations have become a standard
tool to examine the fine structural details of ligand-receptor
interactions, as they can reveal subtle structural rearrange-
ments due to the mutual accommodation between the ligand
and the residues that define the walls of the pocket upon
ligand binding in few nanoseconds. The challenge, neverthe-
less, consists of extending the capability of MD techniques
to describe conformational transitions that require longer
time scales. The rapid progress in algorithms and computa-
tional resources make it feasible to shed light into the mo-
lecular mechanism underlying those large-scale conforma-
tion transitions and their functional implications, opening
new avenues for drug discovery. Enhanced sampling tech-
niques contribute decisively to the progress in this field,
however, the synergy between experiments and simulations
shall undoubdtedly represent the fundamental step towards
the understanding of the role played by dynamics properties
in a variety of biochemical systems.
ACKNOWLEDGEMENTS
We acknowledge the financial support from the Ministe-
rio de Innovación y Ciencia (MICINN; SAF2008-05595 and
SAF2009-08811) and Generalitat de Catalunya (2009-
SGR00298).
ABBREVIATIONS
AK = Adenylate Kinase
AL = Activation Loop
AChE = Acetylcholinesterase
CaM = Calmodulin
FDTI = Finite Difference Thermodynamic Inte-
gration
FEP = Free Energy Perturbation
GDH = Glutamate Dehydrogenase
HSP = Heat-Shock Protein
LMW-PTP = Low-Molecular-Weight Protein Tyro-
sine Phosphatase
MD = Molecular Dynamics
NMP = Nucleoside Monophosphate
PKA = Protein Kinase A
PTP = Protein Tyrosine Phosphatase
RE = Replica Exchange
REFERENCES
[1] Jimenez, R.; Salazar, G.; Baldridge, K. K.; Romesberg, F. E. Flexi-
bility and molecular recognition in the immune system. Proc. Natl.
Acad. Sci. USA, 2003, 100, 92-97.
[2] Sundberg, E. J.; Mariuzza, R. A. Molecular recognition in anti-
body-antigen complexes. Adv. Protein Chem., 2002, 61, 119-160.
[3] Pabo, C. O.; Sauer, R. T. Protein-DNA recognition. Annu. Rev.
Biochem., 1984, 53, 293-321.
[4] Pabo, C. O.; Sauer, R. T. Transcription factors: structural families
and principles of DNA recognition. Annu. Rev. Biochem., 1992, 61,
1053-1095.
[5] Hammes, G. G. Multiple conformational changes in enzyme ca-
talysis. Biochemistry, 2002, 41, 8221-8228.
[6] Hunter, C. A. Quantifying intermolecular interactions: guidelines
for the molecular recognition toolbox. Angew. Chem. Int. Ed. Engl.,
2004, 43, 5310-5324.
[7] Gohlke, H.; Klebe, G. Approaches to the description and prediction
of the binding affinity of small-molecule ligands to macromolecu-
lar receptors. Angew. Chem. Int. Ed. Engl., 2002, 41, 2644-2676.
[8] Bissantz, C.; Folkers, G.; Rognan, D. Protein-based virtual screen-
ing of chemical databases. 1. Evaluation of different dock-
ing/scoring combinations. J. Med. Chem., 2000, 43, 4759-4767.
[9] Charifson, P. S.; Corkery, J. J.; Murcko, M. A.; Walters, W. P.
Consensus scoring: A method for obtaining improved hit rates from
docking databases of three-dimensional structures into proteins. J.
Med. Chem., 1999, 42, 5100-5109.
[10] Spyrakis, F.; Cozzini, P.; Kellogg, G. E. Docking and scoring in
drug discovery. In: Burger's Medicinal Chemistry; Eds.; Wiley-
VCH: New York, 2010.
[11] Spyrakis, F.; Kellogg, G. E.; Amadasi, A.; Cozzini, P. Scoring
functions for virtual screening. In Frontiers in Drug Design and
Discovery; Eds.; Bentham Science Publisher, 2007, pp 317-379.
[12] Moitessier, N.; Westhof, E.; Hanessian, S. Docking of aminoglyco-
sides to hydrated and flexible RNA. J. Med. Chem., 2006, 49,
1023-1033.
[13] Schnecke, V.; Kuhn, L. A. Virtual screening with solvation and
ligand-induced complementarity. Perspect. Drug Discov. Des.,
2000, 20, 171-190.
[14] Shoemaker, B. A.; Wang, J.; Wolynes, P. G. Structural correlations
in protein folding funnels. Proc. Natl. Acad. Sci. USA, 1997, 94,
777-782.
[15] van Dijk, A. D.; Bonvin, A. M. Solvated docking: introducing
water into the modelling of biomolecular complexes. Bioinformat-
ics, 2006, 22, 2340-2347.
[16] Verdonk, M. L.; Chessari, G.; Cole, J. C.; Hartshorn, M. J.;
Murray, C. W.; Niss ink, J. W.; Taylor, R. D.; Taylor, R. Modeling
water molecules in protein-ligand docking using GOLD. J. Med.
Chem., 2005, 48, 6504-6515.
[17] Osterberg, F.; Morris, G. M.; Sanner, M. F.; Olson, A. J.; Goodsell,
D. S. Automated docking to multiple target structures: incorpora-
tion of protein mobility and structural water heterogeneity in
AutoDock. Proteins, 2002, 46, 34-40.
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 15
[18] Rarey, M.; Kramer, B.; Lengauer, T. The particle concept: placing
discrete water molecules during protein-ligand docking predictions.
Proteins, 1999, 34, 17-28.
[19] Czodrowski, P.; Dramburg, I.; Sotriffer, C. A.; Klebe, G. Devel-
opment, validation, and application of adapted PEOE charges to es-
timate pKa values of functional groups in protein-ligand com-
plexes. Proteins, 2006, 65, 424-437.
[20] Brenk, R.; Vetter, S. W.; Boyce, S. E.; Goodin, D. B.; Shoichet, B.
K. Probing molecular docking in a charged model binding site. J.
Mol. Biol., 2006, 357, 1449-1470.
[21] Deng, W.; Verlinde, C. L. Evaluation of different virtual screening
programs for docking in a charged binding pocket. J. Chem. Inf.
Model., 2008, 48, 2010-2020.
[22] Bahar, I.; Chennubhotla, C.; Tobi, D. Intrinsic dynamics of en-
zymes in the unbound state and relation to allosteric regulation.
Curr. Opin. Struct. Biol., 2007, 17, 633-640.
[23] Carlson, H. A.; McCammon, J. A. Accommodating protein flexibil-
ity in computational drug design. Mol. Pharmacol., 2000, 57, 213-
218.
[24] Cozzini, P.; Kellogg, G. E.; Spyrakis, F.; Abraham, D. J .; Costan-
tino, G.; Emerson, A.; Fanelli, F.; Gohlke, H.; Kuhn, L. A.; Morris,
G. M.; Orozco, M.; Pertinhez, T. A.; Rizzi, M.; Sotriffer, C. A.
Target flexibility: an emerging consideration in drug discovery and
design. J. Med. Chem., 2008, 51, 6237-6255.
[25] Henzler-Wildman, K.; Kern, D. Dynamic personalities of proteins.
Nature 2007, 450, 964-972.
[26] Tsai, C. J.; Kumar, S.; Ma, B.; Nussinov, R. Folding funnels, bind-
ing funnels, and protein function. Protein Sci., 1999, 8, 1181-1190.
[27] Savir, Y.; Tlusty, T. Conformational proofreading: the impact of
conformational changes on the specificity of molecular recognition.
PLoS One, 2007, 2, e468.
[28] Mittag, T.; Kay, L. E.; Forman-Kay, J. D. Protein dynamics and
conformational disorder in molecular recognition. J. Mol. Recog-
nit., 2010, 23, 105-116.
[29] Freire, E. The propagation of binding interactions to remote sites in
proteins: analysis of the binding of the monoclonal antibody D1.3
to lysozyme. Proc. Natl. Acad. Sci. USA, 1999, 96, 10118-10122.
[30] Ma, B.; Kumar, S.; Tsai, C. J.; Nussinov, R. Folding funnels and
binding mechanisms. Protein Eng., 1999, 12, 713-720.
[31] Onuchic, J. N. Contacting the protein folding funnel with NMR.
Proc. Natl. Acad. Sci. USA, 1997, 94, 7129-7131.
[32] Rejto, P. A.; Freer, S. T. Protein conformational substates from X-
ray crystallography. Prog. Biophys. Mol. Biol., 1996, 66, 167-196.
[33] Koshland, D. E. Application of a Theory of Enzyme Specificity to
Protein Synthesis. Proc. Natl. Acad. Sci. USA, 1958, 44, 98-104.
[34] Monod, J.; Wyman, J.; Changeux, J. P. On the nature of allosteric
transitions: A plausible model. J. Mol. Biol., 1965, 12, 88-118.
[35] Valente, A. P.; Miyamoto, C. A.; Almeida, F. C. Implications of
protein conformational diversity for binding and development of
new biological active compounds. Curr. Med. Chem., 2006, 13,
3697-3703.
[36] Bosshard, H. R. Molecular recognition by induced fit: how fit is the
concept? News Physiol. Sci., 2001, 16, 171-173.
[37] James, L. C.; Tawfik, D. S. Conformational diversity and protein
evolution-a 60-year-old hypothesis revisited. Trends Biochem. Sci.,
2003, 28, 361-368.
[38] Kern, D.; Zuiderweg, E. R. The role of dynamics in allosteric regu-
lation. Curr. Opin. Struct. Biol., 2003, 13, 748-757.
[39] Mittermaier, A.; Kay, L. E. New tools provide new insights in
NMR studies of protein dynamics. Science, 2006, 312, 224-228.
[40] Showalter, S. A.; Bruschweiler, R. Quantitative molecular ensem-
ble interpretation of NMR dipolar couplings without restraints. J.
Am. Chem. Soc., 2007, 129, 4158-4159.
[41] Tang, C.; Schwieters, C. D.; Clore, G. M. Open-to-closed transition
in apo maltose-binding protein observed by paramagnetic NMR.
Nature, 2007, 449, 1078-1082.
[42] Tobi, D.; Bahar, I. Structural changes involved in protein binding
correlate with intrinsic motions of proteins in the unbound state.
Proc. Natl. Acad. Sci. USA, 2005, 102, 18908-18913.
[43] Gulukota, K.; Wolynes, P. G. Statistical mechanics of kinetic
proofreading in protein folding in vivo. Proc. Natl. Acad. Sci. USA,
1994, 91, 9292-9296.
[44] Kumar, S.; Ma, B.; Tsai, C. J.; Sinha, N.; Nussinov, R. Folding and
binding cascades: dynamic landscapes and population shifts. Pro-
tein Sci., 2000, 9, 10-19.
[45] Tsai, C. J.; Ma, B.; Nussinov, R. Folding and binding cascades:
shifts in energy landscapes. Proc. Natl. Acad. Sci. USA, 1999, 96,
9970-9972.
[46] Xu, Y.; Colletier, J. P.; Jiang, H.; Silman, I.; Sussman, J. L.; Weik,
M. Induced-fit or preexisting equilibrium dynamics? Lessons from
protein crystallography and MD simulations on acetylcho-
linesterase and implications for structure-based drug design. Pro-
tein Sci., 2008, 17, 601-605.
[47] Okazaki, K.; Takada, S. Dynamic energy landscape view of cou-
pled binding and protein conformational change: induced-fit versus
population-shift mechanisms. Proc. Natl. Acad. Sci. USA, 2008,
105, 11182-11187.
[48] Sullivan, S. M.; Holyoak, T. Enzymes with lid-gated active sites
must operate by an induced fit mechanism instead of conforma-
tional selection. Proc. Natl. Acad. Sci. USA, 2008, 105, 13829-
13834.
[49] Bakan, A.; Bahar, I. The intrinsic dynamics of enzymes plays a
dominant role in determining the structural changes induced upon
inhibitor binding. Proc. Natl. Acad. Sci. USA, 2009, 106, 14349-
14354.
[50] Bui, J. M.; McCammon, J. A. Protein complex formation by ace-
tylcholinesterase and the neurotoxin fasciculin-2 appears to involve
an induced-fit mechanism. Proc. Natl. Aca. Sci. USA, 2006, 103,
15451-15456.
[51] Levy, Y.; Onuchic, J. N.; Wolynes, P. G. Fly-casting in protein-
DNA binding: frustration between protein folding and electrostatics
facilitates target recognition. J. Am. Chem. Soc., 2007, 129, 738-
739.
[52] Sugase, K.; Dyson, H. J.; Wright, P. E. Mechanism of coupled
folding and binding of an intrinsically disordered protein. Nature,
2007, 447, 1021-1025.
[53] Arora, K.; Brooks, C. L., 3rd. Large-scale allosteric conformational
transitions of adenylate kinase appear to involve a population-shift
mechanism. Proc. Natl. Acad. Sci. USA, 2007, 104, 18496-18501.
[54] Hanson, J. A.; Duderstadt, K.; Watkins, L. P.; Bhattacharyya, S.;
Brokaw, J.; Chu, J. W.; Yang, H. Illuminating the mechanistic roles
of enzyme conformational dynamics. Proc. Natl. Acad. Sci. USA,
2007, 104, 18055-18060.
[55] James, L. C.; Roversi, P.; Tawfik, D. S. Antibody multispecificity
mediated by conformational diversity. Science, 2003, 299, 1362-
1367.
[56] Claussen, H.; Buning, C.; Rarey, M.; Lengauer, T. FlexE: efficient
molecular docking considering protein structure variations. J. Mol.
Biol., 2001, 308, 377-395.
[57] Huey, R.; Morris, G.; Olson, A.; Goodsell, D. A semiempirical free
energy force field with charge-based desolvation. J. Comput.
Chem., 2007, 28, 1145-1152.
[58] Verdonk, M. L.; Cole, J. C.; Hartshorn, M. J.; Murray, C. W.; Tay-
lor, R. D. Improved protein-ligand docking using GOLD. Proteins,
2003, 52, 609-623.
[59] Zavodszky, M. I.; Kuhn, L. A. Side-chain flexibility in protein-
ligand binding: the minimal rotation hypothesis. Protein Sci., 2005,
14, 1104-1114.
[60] Rosenfeld, R. J.; Goodsell, D. S.; Musah, R. A.; Morris, G. M.;
Goodin, D. B.; Olson, A. J. Automated docking of ligands to an ar-
tificial active site: augmenting crystallographic analysis with com-
puter modeling. J. Comput. Aided Mol. Des., 2003, 17, 525-536.
[61] Zacharias, M.; Sklenar, H. Harmonic modes as variables to ap-
proximately account for receptor flexibility in ligand-receptor
docking simulations: applications to DNA minor groove ligand
complex. J. Comput. Chem., 1999, 20, 287-300.
[62] Zavodszky, M. I.; Lei, M.; Thorpe, M. F.; Day, A. R.; Kuhn, L. A.
Modeling correlated main-chain motions in proteins for flexible
molecular recognition. Proteins, 2004, 57, 243-261.
[63] Cavasotto, C. N.; Abagyan, R. A. Protein flexibility in ligand dock-
ing and virtual screening to protein kinases. J. Mol. Biol., 2004,
337, 209-225.
[64] Cavasotto, C. N.; Kovacs, J. A.; Abagyan, R. A. Representing
receptor flexibility in ligand docking through relevant normal
modes. J. Am. Chem. Soc., 2005, 127, 9632-9640.
[65] Kallblad, P.; Dean, P. M. Efficient conformational sampling of
local side-chain flexibility. J. Mol. Biol., 2003, 326, 1651-1665.
[66] Nabuurs, S. B.; Wagener, M.; de Vlieg, J. A flexible approach to
induced fit docking. J. Med. Chem., 2007, 50, 6507-6518.
16 Current Topics in Medicinal Chemis try, 2011, Vol. 11, No. 2 Spyrakis et al.
[67] Totrov, M.; Abagyan, R. Flexible ligand docking to multiple recep-
tor conformations: a practical alternative. Curr. Opin. Struct. Biol.,
2008, 18, 178-184.
[68] Krol, M.; Tournier, A. L.; Bates, P. A. Flexible relaxation of rigid-
body docking solutions. Proteins, 2007, 68, 159-169.
[69] Barril, X.; Morley, S. D. Unveiling the full potential of flexible
receptor docking using multiple crystallographic structures. J. Med.
Chem., 2005, 48, 4432-4443.
[70] Craig, I.; Essex, J.; Spiegel, K. Ensemble docking into multiple
crystallographically-derived protein structures: an evaluation based
on the statistical analysis of enrichments. J. Chem. Inf. Model.,
2010, 50, 511-524.
[71] Akerud, T.; Thulin, E.; Van Etten, R. L.; Akke, M. Intramolecular
dynamics of low molecular weight protein tyrosine phosphatase in
monomer-dimer equilibrium studied by NMR: a model for changes
in dynamics upon target binding. J. Mol. Biol., 2002, 322, 137-152.
[72] Banci, L.; Bertini, I.; Cantini, F.; D'Onofrio, M.; Viezzoli, M. S.
Structure and dynamics of copper-free SOD: The protein before
binding copper. Protein Sci., 2002, 11, 2479-2492.
[73] Korzhnev, D. M.; Karlsson, B. G.; Orekhov, V. Y.; Billeter, M.
NMR detection of multiple transitions to low-populated states in
azurin. Protein Sci., 2003, 12, 56-65.
[74] Sussman, J. L.; Harel, M.; Frolow, F.; Oefner, C.; Goldman, A.;
Toker, L.; Silman, I. Atomic structure of acetylcholinesterase from
Torpedo californica: a prototypic acetylcholine-binding protein.
Science, 1991, 253, 872-879.
[75] Raves, M. L.; Harel, M.; Pang, Y. P.; Silman, I.; Kozikowski, A.
P.; Sussman, J. L. Structure of acetylcholinesterase complexed with
the nootropic alkaloid, (-)-huperzine A. Nat. Struct. Biol., 1997, 4,
57-63.
[76] Barril, X.; Kalko, S. G.; Orozco, M.; Luque, F. J. Rational design
of reversible acetylcholinesterase inhibitors. Mini Rev. Med. Chem.,
2002, 2, 27-36.
[77] Muñoz-Torrero, D. Acetylcholinesterase inhibitors as disease-
modifying therapies for Alzheimer’s disease. Curr. Med. Chem.,
2008, 15, 2433-2455.
[78] Harel, M.; Schalk, I.; Ehret-Sabatier, L.; Bouet, F.; Goeldner, M.;
Hirth, C.; Axelsen, P. H.; Silman, I.; Sussman, J. L. Quaternary
ligand binding to aromatic residues in the active-site gorge of ace-
tylcholinesterase. Proc. Natl. Acad. Sci. USA, 1993, 90, 9031-9035.
[79] Dvir, H.; Wong, D. M.; Harel, M.; Barri l, X. ; Orozco, M.; Luque,
F. J.; Munoz-Torrero, D.; Camps, P.; Rosenberry, T. L.; Silman, I.;
Sussman, J. L. 3D structure of Torpedo californica acetylcho-
linesterase complexed with huprine X at 2.1 A resolution: kinetic
and molecular dynamic correlates. Biochemistry, 2002, 41, 2970-
2981.
[80] Kryger, G.; Silman, I.; Sussman, J. L. Structure of acetylcho-
linesterase complexed with E2020 (Aricept): implications for the
design of new anti-Alzheimer drugs. Structure, 1999, 7, 297-307.
[81] Bourne, Y.; Taylor, P.; Radic, Z.; Marchot, P. Structural insights
into ligand interactions at the acetylcholinesterase peripheral ani-
onic site. EMBO J., 2003, 22, 1-12.
[82] Bourne, Y.; Kolb, H. C.; Radic, Z.; Sharpless, K. B.; Taylor, P.;
Marchot, P. Freeze-frame inhibitor captures acetylcholinesterase in
a unique conformation. Proc. Natl. Acad. Sci. USA, 2004, 101,
1449-1454.
[83] Colletier, J. P.; Sanson, B.; Nachon, F.; Gabellieri, E.; Fattorusso,
C.; Campiani, G.; Weik, M. Conformational flexibility in the pe-
ripheral site of Torpedo californica acetylcholinesterase revealed
by the complex structure with a bifunctional inhibitor. J. Am.
Chem. Soc., 2006, 128, 4526-4527.
[84] Haviv, H.; Wong, D. M.; Greenblatt, H. M.; Carlier, P. R.; Pang, Y.
P.; Silman, I.; Sussman, J. L. Crystal packing mediates enantiose-
lective ligand recognition at the peripheral site of acetylcho-
linesterase. J. Am. Chem. Soc., 2005, 127, 11029-11036.
[85] Rydberg, E. H.; Brumshtein, B.; Greenblatt, H. M.; Wong, D. M.;
Shaya, D.; Williams, L. D.; Carlier, P. R.; Pang, Y. P.; Silman, I.;
Sussman, J. L. Complexes of alkylene-linked tacrine dimers with
Torpedo californica acetylcholinesterase: Binding of Bis5-tacrine
produces a dramatic rearrangement in the active-site gorge. J. Med.
Chem., 2006, 49, 5491-5500.
[86] Kryger, G.; Harel, M.; Giles, K.; Toker, L.; Velan, B.; Lazar, A.;
Kronman, C.; Barak, D.; Ariel, N.; Shafferman, A.; Silman, I.;
Sussman, J. L. Structures of recombinant native and E202Q mutant
human acetylcholinesterase complexed with the snake-venom toxin
fasciculin-II. Acta Crystallogr. D Biol. Crystallogr., 2000, 56,
1385-1394.
[87] Camps, P.; Formosa, X.; Galdeano, C.; Munoz-Torrero, D.; Rami-
rez, L.; Gomez, E.; Isambert, N.; Lavilla, R.; Badia, A.; Clos, M.
V.; Bartolini, M.; Mancini, F.; Andrisano, V.; Arce, M. P.; Rodri-
guez-Franco, M. I.; Huertas, O.; Dafni, T.; Luque, F. J. Pyrano[3,2-
c]quinoline-6-chlorotacrine hybrids as a novel family of acetylcho-
linesterase- and beta-amyloid-directed anti-Alzheimer compounds.
J. Med. Chem., 2009, 52, 5365-5379.
[88] Gutteridge, A.; Thornton, J. Conformational changes observed in
enzyme crystal structures upon substrate binding. J. Mol. Biol.,
2005, 346, 21-28.
[89] Barril, X.; Fradera, X. Incorporating protein flexibility into docking
and structure-based drug design. Exp. Opin. Drug Discov., 2006, 1,
335-349.
[90] Stebbins, C. E.; Russo, A. A.; Schneider, C.; Rosen, N.; Hartl, F.
U.; Pavletich, N. P. Crystal structure of an Hsp90-geldanamycin
complex: targeting of a protein chaperone by an antitumor agent.
Cell, 1997, 89, 239-250.
[91] Chiosis, G.; Timaul, M. N.; Lucas, B.; Munster, P. N.; Zheng, F.
F.; Sepp-Lorenzino, L.; Rosen, N. A small molecule designed to
bind to the adenine nucleotide pocket of Hsp90 causes Her2 degra-
dation and the growth arrest and differentiation of breast cancer
cells. Chem. Biol., 2001, 8, 289-299.
[92] Wright, L.; Barril, X.; Dymock, B.; Sheridan, L.; Surgenor, A.;
Beswick, M.; Drysdale, M.; Collier, A.; Massey, A.; Davies, N.;
Fink, A.; Fromont, C.; Aherne, W.; Boxall, K.; Sharp, S.; Work-
man, P.; Hubbard, R. E. Structure-activity relationships in purine-
based inhibitor binding to HSP90 isoforms. Chem. Biol., 2004, 11,
775-785.
[93] Ali, M. M.; Roe, S. M.; Vaughan, C. K.; Meyer, P.; Panaretou, B.;
Piper, P. W.; Prodromou, C.; Pearl, L. H. Crystal structure of an
Hsp90-nucleotide-p23/Sba1 closed chaperone complex. Nature,
2006, 440, 1013-1017.
[94] Staessen, J. A.; Li, Y.; Richart, T. Oral renin inhibitors. Lancet,
2006, 368, 1449-1456.
[95] Vieira, E.; Binggeli, A.; B reu, V.; Bur, D.; Fischli, W.; Guller, R.;
Hirth, G.; Marki, H. P.; Muller, M.; Oefner, C.; Scalone, M.;
Stadler, H.; Wilhelm, M.; Wostl, W. Substituted piperidines--
highly potent renin inhibitors due to induced fit adaptation of the
active site. Bioorg. Med. Chem. Lett., 1999, 9, 1397-1402.
[96] Bezencon, O.; Bur, D.; Weller, T.; Richard-Bildstein, S.; Remen,
L.; Sifferlen, T.; Corminboeuf, O.; Grisostomi, C.; Boss, C.; Prade,
L.; Delahaye, S.; Treiber, A.; Strickner, P.; Binkert, C.; Hess, P.;
Steiner, B.; Fischli, W. Design and preparation of potent, nonpep-
tidic, bioavailable renin inhibitors. J. Med. Chem., 2009, 52, 3689-
3702.
[97] Pargellis, C.; Tong, L.; Churchill, L.; Cirillo, P. F.; Gilmore, T.;
Graham, A. G.; Grob, P. M.; Hickey, E. R.; Moss, N.; Pav, S.;
Regan, J. Inhibition of p38 MAP kinase by utilizing a novel allos-
teric binding site. Nat. Struct. Biol., 2002, 9, 268-272.
[98] Regan, J.; Pargellis, C.; Cirillo, P. F.; Gilmore, T.; Hickey, E. R.;
Peet, G.; Proto, A.; Swinamer, A.; Moss, N. The kinetics of binding
to p38MAP kinase by analogues of BIRB 796. Bioorg. Med. Chem.
Lett., 2003, 13, 3101-3104.
[99] Abad-Zapatero, C. Ligand efficiency indices for effective drug
discovery. Exp. Opin. Drug Discov., 2007, 2, 469-488.
[100] Wiesmann, C.; Barr, K. J.; Kung, J.; Zhu, J.; Erlanson, D. A.; Shen,
W.; Fahr, B. J.; Zhong, M.; Taylor, L.; Randal, M.; McDowell, R.
S.; Hansen, S. K. Allosteric inhibition of protein tyrosine phospha-
tase 1B. Nat. Struct. Mol. Biol., 2004, 11, 730-737.
[101] Jia, Z.; Barford, D.; Flint, A. J.; Tonks, N. K. Structural basis for
phosphotyrosine peptide recognition by protein tyrosine phospha-
tase 1B. Science, 1995, 268, 1754-1758.
[102] Liu, S.; Zeng, L. F.; Wu, L.; Yu, X.; Xue, T.; Gunawan, A. M.;
Long, Y. Q.; Zhang, Z. Y. Targeting inactive enzyme conforma-
tion: aryl diketoacid derivatives as a new class of PTP1B inhibitors.
J. Am. Chem. Soc., 2008, 130, 17075-17084.
[103] Betts, M. J.; Sternberg, M. J. An analysis of conformational
changes on protein-protein association: implications for predictive
docking. Protein Eng., 1999, 12, 271-283.
[104] Hardy, J. A.; Wells, J. A. Searching for new allosteric sites in en-
zymes. Curr. Opin. Struct. Biol., 2004, 14, 706-715.
[105] Lindsley, J. E.; Rutter, J. Whence cometh the allosterome? Proc.
Natl. Acad. Sci. USA, 2006, 103, 10533-10535.
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 17
[106] Whitty, A.; Kumaravel, G. Between a rock and a hard place? Nat.
Chem. Biol., 2006, 2, 112-118.
[107] Kumar, S.; Ma, B.; Tsai, C. J.; Wolfson, H.; Nussinov, R. Folding
funnels and conformational transitions via hinge-bending motions.
Cell Biochem. Biophys., 1999, 31, 141-164.
[108] Li, Q.; Khosla, C.; Puglisi, J. D.; Liu, C. W. Solution structure and
backbone dynamics of the holo form of the frenolicin acyl carrier
protein. Biochemistry, 2003, 42, 4648-4657.
[109] Valentine, E. R.; Palmer, A. G., 3rd. Microsecond-to-millisecond
conformational dynamics demarcate the GluR2 glutamate receptor
bound to agonists glutamate, quisqualate, and AMPA. Biochemis-
try, 2005, 44, 3410-3417.
[110] Seeliger, M. A.; Spichty, M.; Kelly, S. E.; Bycroft, M.; Freund, S.
M.; Karplus, M.; Itzhaki, L. S. Role of conformational heterogene-
ity in domain swapping and adapter function of the Cks proteins. J.
Biol. Chem., 2005, 280, 30448-30459.
[111] Folmer, R. H.; Geschwindner, S.; Xue, Y. Crystal structure and
NMR studies of the apo SH2 domains of ZAP-70: two bikes rather
than a tandem. Biochemistry, 2002, 41, 14176-14184.
[112] McKercher, S. R.; Lombardo, C. R.; Bobkov, A.; Jia, X.; Assa-
Munt, N. Identification of a PU.1-IRF4 protein interaction surface
predicted by chemical exchange line broadening. Proc. Natl. Acad.
Sci. USA, 2003, 100, 511-516.
[113] Venkitaramani, D. V.; Fulton, D. B.; Andreotti, A. H.; Johansen, K.
M.; Johansen, J. Solution structure and backbone dynamics of
Calsensin, an invertebrate neuronal calcium-binding protein. Pro-
tein Sci., 2005, 14, 1894-1901.
[114] Wang, C.; Karpowich, N.; Hunt, J. F.; Rance, M.; Palmer, A. G.
Dynamics of ATP-binding cassette contribute to allosteric control,
nucleotide binding and energy transduction in ABC transporters. J.
Mol. Biol., 2004, 342, 525-537.
[115] Yan, J.; Liu, Y.; Lukasik, S. M.; Speck, N. A.; Bushweller, J. H.
CBFbeta allosterically regulates the Runx1 Runt domain via a dy-
namic conformational equilibrium. Nat. Struct. Mol. Biol., 2004,
11, 901-906.
[116] Yao, S.; Headey, S. J.; Keizer, D. W.; Bach, L. A.; Norton, R. S. C-
terminal domain of insulin-like growth factor (IGF) binding protein
6: conformational exchange and its correlation with IGF-II binding.
Biochemistry, 2004, 43, 11187-11195.
[117] Giedroc, D. P.; Chen, X.; Pennella, M. A.; LiWang, A. C. Confor-
mational heterogeneity in the C-terminal zinc fingers of human
MTF-1: an NMR and zinc-binding study. J. Biol. Chem., 2001,
276, 42322-42332.
[118] Ramboarina, S.; Srividya, N.; Atkinson, R. A.; Morellet, N.;
Roques, B. P.; Lefevre, J. F.; Mely, Y.; Kieffer, B. Effects of tem-
perature on the dynamic behaviour of the HIV-1 nucleocapsid
NCp7 and its DNA complex. J. Mol. Biol., 2002, 316, 611-627.
[119] Stevens, S. Y.; Sanker, S.; Kent, C.; Zuiderweg, E. R. Delineation
of the allosteric mechanism of a cytidylyltransferase exhibiting
negative cooperativity. Nat. Struct. Biol., 2001, 8, 947-952.
[120] Huse, M.; Kuriyan, J. The conformational plasticity of protein
kinases. Cell, 2002, 109, 275-282.
[121] Hyeon, C.; Jennings, P. A.; Adams, J. A.; Onuchic, J. N. Ligand-
induced global transitions in the catalytic domain of protein kinase
A. Proc. Natl. Acad. Sci. USA, 2009, 106, 3023-3028.
[122] Lou, H.; Cukier, R. I. Molecular dynamics of apo-adenylate kinase:
a principal component analysis. J. Phys. Chem. B, 2006, 110,
12796-12808.
[123] Whitford, P. C.; Miyashita, O.; Levy, Y.; Onuchic, J. N. Conforma-
tional transitions of adenylate kinase: switching by cracking. J.
Mol. Biol., 2007, 366, 1661-1671.
[124] Faraldo-Gomez, J. D.; Roux, B. On the importance of a funneled
energy landscape for the assembly and regulation of multidomain
Src tyrosine kinases. Proc. Natl. Acad. Sci. USA, 2007 , 104, 13643-
13648.
[125] Yang, S.; Roux, B. Src kinase conformational activation: thermo-
dynamics, pathways, and mechanisms. PLoS Comput. Biol., 2008,
4, e1000047.
[126] Levinson, N. M.; Kuchment, O.; Shen, K.; Young, M. A.; Koldob-
skiy, M.; Karplus, M.; Cole, P. A.; Kuriyan, J. A Src-like inactive
conformation in the abl tyrosine kinase domain. PLoS Biol., 2006,
4, e144.
[127] Berteotti, A.; Cavalli, A.; Branduardi, D.; Gervasio, F. L.; Reca-
natini, M.; Parrinello, M. Protein conformational transitions: the
closure mechanism of a kinase explored by atomistic simulations.
J. Am. Chem. Soc., 2009, 131, 244-250.
[128] Johnson, D. A.; Akamine, P.; Radzio-Andzelm, E.; Madhusudan,
M.; Taylor, S. S. Dynamics of cAMP-dependent protein kinase.
Chem. Rev., 2001, 101, 2243-2270.
[129] Knighton, D. R.; Bell, S. M.; Zheng, J.; Ten Eyck, L. F.; Xuong, N.
H.; Taylor, S. S.; Sowadski, J. M. 2.0 Å refined crystal structure of
the catalytic subunit of cAMP-dependent protein kinase complexed
with a peptide inhibitor and detergent. Acta Cryst. D Biol. Cryst.,
1993, 49, 357-361.
[130] Zheng, J.; Knighton, D. R.; Xuong, N. H.; Taylor, S. S.; Sowadski,
J. M.; Ten Eyck, L. F. Crystal structures of the myristylated cata-
lytic subunit of cAMP-dependent protein kinase reveal open and
closed conformations. Protein Sci., 1993, 2, 1559-1573.
[131] Miyashita, O.; Onuchic, J. N.; Wolynes, P. G. Nonlinear elasticity,
proteinquakes, and the energy landscapes of functional transitions
in proteins. Proc. Natl. Acad. Sci. USA, 2003, 100, 12570-12575.
[132] Yan, H.; Tsai, M. D. Nucleoside monophosphate kinases: structure,
mechanism, and substrate specificity. Adv. Enzymol. Relat. Areas
Mol. Biol., 1999, 73, 103-134.
[133] Tripathi, S.; Portman, J. J. Inherent flexibility determines the transi-
tion mechanisms of the EF-hands of calmodulin. Proc. Natl. Acad.
Sci. USA, 2009, 106, 2104-2109.
[134] Chattopadhyaya, R.; Meador, W. E.; Means, A. R.; Quiocho, F. A.
Calmodulin structure refined at 1.7 A resolution. J. Mol. Biol.,
1992, 228, 1177-1192.
[135] Kuboniwa, H.; Tjandra, N.; Grzesiek, S.; Ren, H.; Klee, C. B.; Bax,
A. Solution structure of calcium-free calmodulin. Nat. Struct. Biol.,
1995, 2, 768-776.
[136] Chou, J. J.; Li, S.; Klee, C. B.; Bax, A. Solution structure of
Ca(2+)-calmodulin reveals flexible hand-like properties of its do-
mains. Nat. Struct. Biol., 2001, 8, 990-997.
[137] Evenas, J.; Forsen, S.; Malmendal, A.; Akke, M. Backbone dynam-
ics and energetics of a calmodulin domain mutant exchanging be-
tween closed and open conformations. J. Mol. Biol., 1999, 289,
603-617.
[138] Vigil, D.; Gallagher, S. C.; Trewhella, J.; Garcia, A. E. Functional
dynamics of the hydrophobic cleft in the N-domain of calmodulin.
Biophys. J., 2001, 80, 2082-2092.
[139] Chen, Y. G.; Hummer, G. Slow conformational dynamics and
unfolding of the calmodulin C-terminal domain. J. Am. Chem. Soc.,
2007, 129, 2414-2415.
[140] Lundstrom, P.; Mulder, F. A.; Akke, M. Correlated dynamics of
consecutive residues reveal transient and cooperative unfolding of
secondary structure in proteins. Proc. Natl. Acad. Sci. USA, 2005,
102, 16984-16989.
[141] Meador, W. E.; Means, A. R.; Quiocho, F. A. Target enzyme rec-
ognition by calmodulin: 2.4 A structure of a calmodulin-peptide
complex. Science, 1992, 257, 1251-1255.
[142] Li, M.; Smith, C. J.; Walker, M. T.; Smith, T. J. Novel inhibitors
complexed with glutamate dehydrogenase: allosteric regulation by
control of protein dynamics. J. Biol. Chem., 2009, 284, 22988-
23000.
[143] Tomita, A.; Sato, T.; Ichiyanagi, K.; Nozawa, S.; Ichikawa, H.;
Chollet, M.; Kawai, F.; Park, S. Y.; Tsuduki, T.; Yamato, T.; Ko-
shihara, S. Y.; Adachi, S. Visualizing breathing motion of internal
cavities in concert with ligand migration in myoglobin. Proc. Natl.
Acad. Sci. USA, 2009, 106, 2612-2616.
[144] Billeter, M.; Wagner, G.; Wuthrich, K. Solution NMR structure
determination of proteins revisited. J. Biomol. NMR, 2008, 42, 155-
158.
[145] Adcock, S. A.; McCammon, J. A. Molecular dynamics: survey of
methods for simulating the activity of proteins. Chem. Rev., 2006,
106, 1589-1615.
[146] Senapati, S.; Bui, J. M.; McCammon, J. A. Induced fit in mouse
acetylcholinesterase upon binding a femtomolar inhibitor: a mo-
lecular dynamics study. J. Med. Chem., 2005, 48, 8155-8162.
[147] Munoz-Ruiz, P.; Rubio, L.; Garcia-Palomero, E.; Dorronsoro, I.;
del Monte-Millan, M.; Valenzuela, R.; Usan, P.; de Austria, C.;
Bartolini, M.; Andrisano, V.; Bidon-Chanal, A.; Orozco, M.; Lu-
que, F. J.; Medina, M.; Martinez, A. Design, synthesis, and bio-
logical evaluation of dual binding site acetylcholinesterase inhibi-
tors: new disease-modifying agents for Alzheimer's disease. J.
Med. Chem., 2005, 48, 7223-7233.
[148] Kua, J.; Zhang, Y.; McCammon, J. A. Studying enzyme binding
specificity in acetylcholinesterase using a combined molecular dy-
namics and multiple docking approach. J. Am. Chem. Soc., 2002,
124, 8260-8267.
18 Current Topics in Medicinal Chemis try, 2011, Vol. 11, No. 2 Spyrakis et al.
[149] Lin, J. H.; Perryman, A. L.; Schames, J. R.; McCammon, J. A.
Computational drug design accommodating receptor flexibility: the
relaxed complex scheme. J. Am. Chem. Soc., 2002, 124, 5632-
5633.
[150] Lin, J. H.; Perryman, A. L.; Schames, J. R.; McCammon, J. A. The
relaxed complex method: Accommodating receptor flexibility for
drug design with an improved scoring scheme. Biopolymers, 2003,
68, 47-62.
[151] Alonso, H.; Bliznyuk, A. A.; Gready, J. E. Combining docking and
molecular dynamic simulations in drug design. Med. Res. Rev.,
2006, 26, 531-568.
[152] Meagher, K. L.; Carlson, H. A. Incorporating protein flexibility in
structure-based drug discovery: using HIV-1 protease as a test case.
J. Am. Chem. Soc., 2004, 126, 13276-13281.
[153] Brigo, A.; Lee, K. W.; Fogolari, F.; Mustata, G. I.; Briggs, J. M.
Comparative molecular dynamics simulations of HIV-1 integrase
and the T66I/M154I mutant: binding modes and drug resistance to
a diketo acid inhibitor. Proteins, 2005, 59, 723-741.
[154] Deupi, X.; Dolker, N.; Lopez-Rodriguez, M. L.; Campillo, M.;
Ballesteros, J. A.; Pardo, L. Structural models of class a G protein-
coupled receptors as a tool for drug design: insights on transmem-
brane bundle plasticity. Curr. Top. Med. Chem., 2007, 7, 991-998.
[155] Fanelli, F.; De Benedetti, P. G. Computational modeling ap-
proaches to structure-function analysis of G protein-coupled recep-
tors. Chem. Rev., 2005, 105, 3297-3351.
[156] Berne, B. J.; Straub, J. E. Novel methods of sampling phase space
in the simulation of biological systems. Curr. Opin. Struct. Biol.,
1997, 7, 181-189.
[157] Hansmann, U. H.; Okamoto, Y. New Monte Carlo algorithms for
protein folding. Curr. Opin. Struct. Biol., 1999, 9, 177-183.
[158] Lei, H.; Duan, Y. Improved sampling methods for molecular simu-
lation. Curr. Opin. Struct. Biol., 2007, 17, 187-191.
[159] Mitsutake, A.; Sugita, Y.; Okamoto, Y. Generalized-ensemble
algorithms for molecular simulations of biopolymers. Biopolymers,
2001, 60, 96-123.
[160] Okamoto, Y. Generalized-ensemble algorithms: enhanced sampling
techniques for Monte Carlo and molecular dynamics simulations. J.
Mol. Graph. Model., 2004, 22, 425-439.
[161] Trebst, S.; Troyer, M.; Hansmann, U. H. Optimized parallel tem-
pering simulations of proteins. J. Chem. Phys., 2006, 124, 174903.
[162] Tozzini, V. Coarse-grained models for proteins. Curr. Opin. Struct.
Biol., 2005, 15, 144-150.
[163] Neri, M.; Anselmi, C.; Cascella, M.; Maritan, A.; Carloni, P.
Coarse-grained model of proteins incorporating atomistic detail of
the active site. Phys. Rev. Lett., 2005, 95, 218102.
[164] Czerminski, R.; Elber, R. Computational studies of ligand diffusion
in globins: I. Leghemoglobin. Proteins, 1991, 10, 70-80.
[165] Sugita, Y.; Okamoto, Y. Replica-exchange molecular dynamics
method for protein folding. Chem. Phys. Lett., 1999, 314, 141-151.
[166] Chipot, C.; Pohorille, A., Eds. Free Energy Calculations: Theory
and Applications in Chemistry and Biology. Springer, Berlin, 2007.
[167] Hamelberg, D.; Mongan, J.; McCammon, J. A. Accelerated mo-
lecular dynamics: a promising and efficient simulation method for
biomolecules. J. Chem. Phys., 2004, 120, 11919-11929.
[168] Swendsen, R. H.; Wang, J. S. Replica Monte Carlo simulation of
spin glasses. Phys. Rev. Lett., 1986, 57, 2607-2609.
[169] Earl, D. J.; Deem, M. W. Parallel tempering: theory, applications,
and new perspectives. Phys. Chem. Chem. Phys., 2005, 7, 3910-
3916.
[170] Liwo, A.; Czaplewski, C.; Oldziej, S.; Scheraga, H. A. Computa-
tional techniques for efficient conformational sampling of proteins.
Curr. Opin. Struct. Biol., 2008, 18, 134-139.
[171] Garcia, A. E.; Onuchic, J. N. Folding a protein in a computer: an
atomic description of the folding/unfolding of protein A. Proc.
Natl. Acad. Sci. USA, 2003, 100, 13898-13903.
[172] Sanbonmatsu, K. Y.; Garcia, A. E. Structure of Met-enkephalin in
explicit aqueous solution using replica exchange molecular dynam-
ics. Proteins, 2002, 46, 225-234.
[173] Zhou, R. Trp-cage: folding free energy landscape in explicit water.
Proc. Natl. Acad. Sci. USA, 2003, 100, 13280-13285.
[174] Zhou, R. Exploring the protein folding free energy landscape:
coupling replica exchange method with P3ME/RESPA algorithm.
J. Mol. Graph. Model, 2004, 22, 451-463.
[175] Zhou, R.; Berne, B. J. Can a continuum solvent model reproduce
the free energy landscape of a beta-hairpin folding in water? Proc.
Natl. Acad. Sci. USA, 2002, 99, 12777-12782.
[176] Verkhivker, G.; Rejto, P.; Bouzida, D.; Arthurs, S.; Colson, A.;
Freer, S. T.; Gehlhaar, D.; Larson, V.; Luty, B.; Marrone, T.; Rose,
P. Parallel simulated tempering dynamics of ligand-protein binding
with ensembles of protein conformations. Chem. Phys. Lett., 2001,
337, 181-189.
[177] Goto, H.; Obata, S.; Kamakura, T.; Nakayama, N.; Sato, M.; Naka-
jima, Y.; Nagashima, U.; Watanabe, T.; Inadomi, Y.; Ito, M.;
Nishikawa, T.; Nakano, T.; Nilsson, L.; Tanaka, S.; Fukuzawa, K.;
Inagaki, Y.; Hamada, M.; Chuman, H. Drug discovery using grid
technology. In Modern Methods for Theoretical Physical Chemis-
try of Biopolymers; Starikov, E., Lewis, J. and Tanaka, S.; Eds.; El-
sevier: Amsterdam, 2006, pp 227-248.
[178] Liu, P.; Kim , B.; Friesner, R. A.; Berne, B. J. Replica exchange
with solute tempering: a method for sampling biological systems in
explicit water. Proc. Natl. Acad. Sci. USA , 2005, 102, 13749-
13754.
[179] Cheng, X.; Cui, G.; Hornak, V.; Simmerling, C. Modified replica
exchange simulation methods for local structure refinement. J.
Phys. Chem. B, 2005, 109, 8220-8230.
[180] Kwak, W.; Hansmann, U. H. Efficient sampling of protein struc-
tures by model hopping. Phys. Rev. Lett., 2005, 95, 138102.
[181] Hritz, J.; Oostenbrink, C. Hamiltonian replica exchange molecular
dynamics using soft-core interactions. J. Chem. Phys., 2008, 128,
144121.
[182] Rodinger, T.; Howell, P. L.; Pomes, R. Calculation of absolute
protein-ligand binding free energy using distributed replica sam-
pling. J. Chem. Phys., 2008, 129, 155102.
[183] Woods, C.; Essex, J.; King, M. The development of replica-
exchange-based free-energy methods. J. Phys. Chem, B, 2003, 107,
13703-13710.
[184] Hamelberg, D.; McCammon, J. A. Accelerating conformational
transitions in biomolecular systems. Ann. Rep. Comp. Chem., 2006,
2, 221-232.
[185] Voter, A. A method for accelerating the molecular dynamics simu-
lation of infrequent events. J. Chem. Phys., 1997, 106, 4665-4677.
[186] Markwick, P. R.; Bouvignies, G.; Blackledge, M. Exploring multi-
ple timescale motions in protein GB3 using accelerated molecular
dynamics and NMR spectroscopy. J. Am. Chem. Soc., 2007, 129,
4724-4730.
[187] Grant, B. J.; Gorfe, A. A.; McCammon, J. A. Ras conformational
switching: simulating nucleotide-dependent conformational transi-
tions with accelerated molecular dynamics. PLoS Comput. Biol.,
2009, 5, e1000325.
[188] Yang, L.; Shao, Q.; Gao, Y. Q. Thermodynamics and folding
pathways of trpzip2: an accelerated molecular dynamics simulation
study. J. Phys. Chem. B, 2009, 113, 803-808.
[189] Laio, A.; Parrinello, M. Escaping free-energy minima. Proc. Natl.
Acad. Sci. USA, 2002, 99, 12562-12566.
[190] Biarnes, X.; Ardevol, A.; Planas, A.; Rovira, C.; Laio, A.; Par-
rinello, M. The conformational free energy landscape of beta-D-
glucopyranose. Implications for substrate preactivation in beta-
glucoside hydrolases. J. Am. Chem. Soc., 2007, 129, 10686-10693.
[191] Domene, C.; Klein, M. L.; Branduardi, D.; Gervasio, F. L.; Par-
rinello, M. Conformational changes and gating at the selectivity fil-
ter of potassium channels. J. Am. Chem. Soc., 2008, 130, 9474-
9480.
[192] Fiorin, G.; Pastore, A.; Carloni, P.; Parrinello, M. Using metady-
namics to understand the mechanism of calmodulin/target recogni-
tion at atomic detail. Biophys. J., 2006, 91, 2768-2777.
[193] Petraglio, G.; Bartolini, M.; Branduardi, D.; Andrisano, V.; Reca-
natini, M.; Gervasio, F. L.; Cavalli, A.; Parrinello, M. The role of
Li+, Na+, and K+ in the ligand binding inside the human acetyl-
cholinesterase gorge. Proteins, 2008, 70, 779-785.
[194] Rohrig, U. F.; Laio, A.; Tantalo, N.; Parrinello, M.; Petronzio, R.
Stability and structure of oligomers of the Alzheimer peptide
Abeta16-22: from the dimer to the 32-mer. Biophys. J., 2006, 91,
3217-3229.
[195] Behler, J.; Martonak, R.; Donadio, D.; Parr inello, M. Metadynam-
ics simulations of the high-pressure phases of silicon employing a
high-dimensional neural network potential. Phys. Rev. Lett., 2008,
100, 185501.
[196] Di Pietro, E.; Pagliai, M.; Cardini, G.; Schettino, V. Solid-state
phase transition induced by pressure in LiOH x H2O. J. Phys.
Chem. B, 2006, 110, 13539-13546.
Protein Flexibility and Ligand Recognition Current Topics in Medicinal Chemistry, 2011, Vol. 11, No. 2 19
[197] Donadio, D.; Bernasconi, M.; Boero, M. Ab initio simulations of
photoinduced interconversions of oxygen deficient centers in
amorphous silica. Phys. Rev. Lett., 2001, 87, 195504.
[198] Prestipino, S.; Giaquinta, P. V. Liquid-solid coexistence via the
metadynamics approach. J. Chem. Phys., 2008, 128, 114707.
[199] Zipoli, F.; Bernasconi, M.; Martonak, R. Constant pressure reactive
molecular dynamics simulations of phase transitions under pres-
sure: The graphite to diamond conversion revisited. Eur. Phys. J. B,
2004, 39, 41-47.
[200] Karamertzanis, P. G.; Raiteri, P.; Parrinello, M.; Leslie, M.; Price,
S. L. The thermal stability of lattice-energy minima of 5-
fluorouracil: metadynamics as an aid to polymorph prediction. J.
Phys. Chem. B, 2008, 112, 4298-4308.
[201] Martonak, R.; Donadio, D.; Oganov, A. R.; Parrinello, M. Crystal
structure transformations in SiO2 from classical and ab initio
metadynamics. Nat. Mater., 2006, 5, 623-626.
[202] Martonak, R.; Laio, A.; Bernasconi, M.; Ceriani, C.; Raiteri, P.;
ZIpoli, F.; Parrinello, M. Simulation of structural phase transitions
by metadynamics. Zeitschrift für Kristallographie 2005, 220, 489-
498.
[203] Oganov, A. R.; Martonak, R.; Laio, A.; Raiteri, P.; Parrinello, M.
Anisotropy of Earth's D'' layer and stacking faults in the MgSiO3
post-perovskite phase. Nature, 2005, 438, 1142-1144.
[204] Blumberger, J.; Ensing, B.; Klein, M. L. Formamide hydrolysis in
alkaline aqueous solution: insight from Ab initio metadynamics
calculations. Angew. Chem. Int. Ed. Engl., 2006, 45, 2893-2897.
[205] Ensing, B.; De Vivo, M.; Liu, Z.; Moore, P.; Klein, M. L. Metady-
namics as a tool for exploring free energy landscapes of chemical
reactions. Acc. Chem. Res., 2006, 39, 73-81.
[206] Ensing, B.; Laio, A.; Gervasio, F. L.; Parrinello, M.; Klein, M. L. A
minimum free energy reaction path for the E2 reaction between
fluoro ethane and a fluoride ion. J. Am. Chem. Soc., 2004, 126,
9492-9493.
[207] Michel, C.; Laio, A.; Mohamed, F.; Krack, M.; Parrinello, M.;
Milet, A. Free energy ab initio metadynamics: A new tool for the
theoretical study of organometallic reactivity? Example of the C!C
and C!H reductive eliminations from platinum(IV) complexes. Or-
ganometallics, 2007, 26, 1241-1249.
[208] Schreiner, E.; Nair, N. N .; Marx, D. Influence of extreme thermo-
dynamic conditions and pyrite surfaces on peptide synthesis in
aqueous media. J. Am. Chem. Soc., 2008, 130, 2768-2770.
[209] Trudu, F.; Donadio, D.; Parrinello, M. Freezing of a Lennard-Jones
fluid: from nucleation to spinodal regime. Phys. Rev. Lett., 2006,
97, 105701.
[210] Spiwok, V.; Lipovova, P.; Kralova, B. Metadynamics in essential
coordinates: free energy simulation of conformational changes. J.
Phys. Chem. B, 2007, 111, 3073-3076.
[211] Piana, S.; Laio, A. A bias-exchange approach to protein folding. J.
Phys. Chem. B, 2007, 111, 4553-4559.
[212] Bussi, G.; Gervasio, F. L.; Laio, A.; Parrinello, M. Free-energy
landscape for beta hairpin folding from combined parallel temper-
ing and metadynamics. J. Am. Chem. Soc., 2006, 128, 13435-
13441.
[213] Gervasio, F. L.; Laio, A.; Parrinello, M. Flexible docking in solu-
tion using metadynamics. J. Am. Chem. Soc., 2005, 127, 2600-
2607.
[214] Bonomi, M.; Gervasio, F. L.; Tiana, G.; Provasi, D.; Broglia, R. A.;
Parrinello, M. Insight into the folding inhibition of the HIV-1 pro-
tease by a small peptide. Biophys. J., 2007, 93, 2813-2821.
[215] Ceccarelli, M.; Danelon, C.; Laio, A.; Parrinello, M. Microscopic
mechanism of antibiotics translocation through a porin. Biophys. J.,
2004, 87, 58-64.
[216] Branduardi, D.; Gervasio, F. L.; Cavalli, A.; Recanatini, M.; Par-
rinello, M. The role of the peripheral anionic site and cation-pi in-
teractions in the ligand penetration of the human AChE gorge. J.
Am. Chem. Soc., 2005, 127, 9147-9155.
Received: December 1, 2009 Accepted: April 15, 2010
