rotamer trials of the residues, and moved back together). Next, it
samples and scores different interfaces, using a function taking into
account membrane score terms.
Memdock is an algorithm developed specifically forα-helical
transmembrane proteins [86]. In this approach, the docking poses
are generated using rigid docking, with subsequent structure
refinement carried out through side-chain rotations and a normal
mode analysis-based backbone refinement. In both the steps the
proteins complexes are kept in a membrane consistent orientation.
The scoring of obtained complexes is carried out using a function
that takes into account membrane properties, distinguishing its
specific sublayers. The weights for scoring were optimized using a
training set of available membrane proteins.
Due to their impressive speed in generating and scoring poses,
protein–protein docking methods are an attractive tool, when
studying GPCR-GPCR dimerization. Using these methods, it is
possible to quickly obtain reliable structures of membrane protein
multimers, without having to rely on computationally demanding
lengthy molecular dynamics simulations. Currently, the main chal-
lenge appears to be optimizing protocols for membrane proteins.
First of all, it is necessary to limit the pose generation step, so it
takes into account only complexes where both of the proteins are in
the membrane. Secondly, the energetic scoring functions need to
be adapted to properly score binding events occurring in the
hydrophobic environment of the membrane (seeNote 3).
The latter point raises the question as to the type of implicit
membrane that should be utilized in scoring algorithms as the
membrane environment has been shown to impact GPCR-GPCR
dimerization [87]. In particular, receptor aggregation is sensitive to
the membrane thickness which depends on the composition of the
membrane. For instance, membranes with a high content of unsat-
urated lipids are condensed and thicker, whereas fully saturated
lipids yield thinner membranes [88]. While it is easy to address
this issue using full-atom simulations, in terms of docking we are
still away from a unified algorithm that can take into account
membrane diversity.
It is important to mention that docking provides a static snap-
shot of the interaction between the two proteins. A reliable oligo-
mer model can help rationalize the impact of certain mutations or
signaling diversity on the frequency of protein–protein contacts, it
can be also used as a tool for the rational design of ligands targeting
GPCR multimers, but to observe how formation of such a unit
impacts signaling, it is crucial to utilize techniques that have a
bigger temporal resolution, and better account for protein
fluctuations.298 Agnieszka A. Kaczor et al.