Biology Reference
In-Depth Information
Chen, Zhang, &Huang, 2013 ), showing dimerization interfaces previously described
in computational studies. Therefore, despite all the challenges, assessing GPCR olig-
omerization by MD simulations allows studying certain aspects of this phenomenon
that today we cannot evaluate experimentally.
Nowadays, two main concerns govern the study of GPCR oligomerization by all-
atom standard MD simulations. On the one hand, a large amount of simulation data is
needed to assess the rotational and diffusional energy profile of the two GPCRmono-
mers. Therefore, describing the energetic landscape of the GPCR dimerization pro-
cess requires very long timescales, namely, in the order of milliseconds ( Sadiq et al.,
2013 ). Such timescales are not reachable by the current technological resources and,
particularly, when using highly descriptive simulations such as all-atom standard
MD. However, certain attempts have been made to overcome the timescale problem
such as reducing the level of description of the simulation (e.g., CG simulations;
Periole et al. (2007) ) or just performing a reduced sampling by biasing such
simulation (e.g., umbrella sampling ( Provasi, Johnston, & Filizola, 2010 ) and/or
metadynamics ( Johnston et al., 2011 )). On the other hand, a very limited number
of GPCR dimers have been solved and deposited in the PDB. These crystal structures
contain relevant data on potentially physiological dimerization interfaces between
GPCRs that could help in reducing the timescale needed to characterize the energy
of the different binding modes. The lack of such information hampers the possibility
of adequately biasing the simulation towards low-energy dimers (i.e., physiological
interfaces).
4.3.2 Simulating GPCR heteromers: all-atom versus CG simulations
Both factors, namely, high timescales and lack of dimer structures, have tipped the
scale in favor of CG simulations when dealing with GPCR dimerization. In CG force
fields, a certain number of heavy atoms (including water molecules) are represented
as a single interaction center, for example, the four-to-one mapping used by the
MARTINI force field ( Periole & Marrink, 2013 ). Such underrepresentation allows
CG simulations to reach much longer timescales when compared to all-atom simu-
lations, although a prudent interpretation of the results is needed. The main drawback
of CG simulations applied to receptor-receptor interactions lies on the set of re-
straints that one needs to use to preserve the secondary structure of proteins
( Periole, Cavalli, Marrink, & Ceruso, 2009 ). Since the secondary structure needs
to be restrained, CG simulations of this nature do not account for any protein-folding
event or any change in the predefined secondary structure. In contrast, CG simula-
tions are very useful to measure protein-protein aggregation ( Javanainen et al., 2013 )
or even reveal preferred dimerization interfaces, as recently shown for rhodopsin
oligomers in an inspiring computational study made by Periole, Knepp, Sakmar,
Marrink, and Huber (2012) .
The high performance of CG simulations gives way to a better sampling of the
energetic landscape, and, therefore, this method can simulate much more heteroge-
neous systems when compared to all-atom simulations. Thus, the MARTINI force
Search WWH ::




Custom Search