Biology Reference
In-Depth Information
to find a combination of such hidden parameters apparently allowing a
fine estimation of
G
bind
for a given system. Of course, the transferability
of such results to other systems is questionable. Nevertheless, MM-
PB(GB)SA proved to be useful for several applications less sensitive to
the choice of hidden parameters, such as the determination of relative
affinities for different small ligands in DD applications, comparison of
relative stabilities of macromolecular conformations, and estimation of
the effect of mutations on association processes and fold stability.
Although some studies aimed at determining absolute
∆
G
bind
for
ligand-protein association, MM-PB(GB)SA is generally used to estimate
relative affinities for different ligands targeting the same protein. This
allows additional approximations, like the neglect of the entropy terms
for ligands of similar masses binding to the same site. Also, despite the
fact that this approach is expected to tackle chemically diverse ligands, it
is often applied to a series of chemically related ligands. This clearly sim-
plifies the problem thanks to the additional cancellation of errors, but it
also reflects the usual DD processes that generally focus on families of
similar ligands. A recent and detailed review of the numerous studies
using MM-PB(GB)SA in the context of DD can be found elsewhere.
55
MM-PB(GB)SA has given variable results, ranging from poor correla-
tions between experimental and calculated
∆
G
bind
to very good ones, with
correlation coefficients up to 0.96.
55
The performance seems to be a
function of the nature of the targeted protein and of the range of activi-
ties encompassed by the ligands. Not surprisingly, the ranking is better
for a broader range of affinities.
56
MM-PB(GB)SA has been found to perform well at determining the
effect of mutations on association processes, and at identifying the “hot
spots” of protein-protein complexes.
42,50,57-60
Two main approaches exist.
First, it is possible to perform a so-called computational alanine scanning
(CAS).
57,58
The alanine mutation is introduced by modifying the frames
extracted from the MD simulation of the wild-type system. The differ-
ence in
∆
G
bind
between the wild-type system and the mutants may be
compared directly to the results of an experimental alanine scanning
(AS).
57,58
The second possibility is to perform a binding free energy
decomposition (BFED)
42
for the wild-type system. This process aims at
calculating the contributions to
∆
∆
G
bind
arising from each atom or groups
