Biology Reference
In-Depth Information
contrast to the surface-based interactions described earlier. Therefore, adequately
representing the membrane environment in MD simulations of GPCRs is a key factor
to successfully characterize the communication between GPCRs and membrane lipids.
In the following section, we give an overview of GPCR modeling, the process of
embedding GPCRs into membranes, facts about simulating receptor-membrane
complexes, and a detailed description of assessing relevant lipid-GPCR interactions.
4.2.1
Modeling GPCRs
All GPCRs share a general architecture in which seven TM helices are connected by
intra- and extracellular loops. However, upon the lack of structural information on a
particular target receptor, a first milestone is the construction of a reliable three-
dimensional (3D) structure by homology modeling. Homology modeling involves
the generation of an all-atom model of the target structure based on experimentally
derived high-resolution structures of a closely related (homologous) protein (tem-
plate). Continuous advances in solving high-resolution GPCR structures keep on
providing excellent templates for the prediction of 3D structures. At the time of writ-
ing this chapter, 44 GPCR structures of 16 different receptor subtypes were deposited
at the PDB (
Berman et al., 2000
). According to a recent assessment of GPCR model-
ing (GPCR DOCK2010), the selection of a correct template is extremely important
for an accurate prediction of the target structure and should have at least 35-40%
sequence identity with the target protein (
Kufareva, Rueda, Katritch, Stevens, &
Abagyan, 2011
). It should be stressed, however, that there is a lack of reliable tem-
plates for certain GPCRs, even between some class A GPCRs such as the cannabi-
noid receptor.
The modeling procedure generally starts with a sequence alignment, where the
sequence of the target GPCR, typically retrieved from the UniProt database
(
http://www.uniprot.org
)
, is aligned to the sequence of the selected template struc-
ture using a sequence alignment tool (e.g., ClustalW—
http://www.ebi.ac.uk/Tools/
msa/clustalw2/
)
. Usually, the resulting alignment needs to be manually refined to
guarantee a perfect alignment of the highly conserved residues of the GPCR super-
family. In a next step, the obtained alignment along with the 3D structure of a tem-
plate serves as an input for specific modeling software, for example, MODELLER
software (
Eswar et al., 2007
), which is able to yield a pool of initial 3D structural
models of the target GPCR. Disulfide bonds, such as the highly conserved one estab-
lished between cysteines C3.25 (Ballesteros-Weinstein nomenclature (
Ballesteros &
Weinstein, 1995
)) at the beginning of the TM helix 3 and the one located in the mid-
dle of the extracellular loop 2, have to be assigned and maintained as constraints for
geometric optimization. Best models can be selected by both using the MODELLER
objective function and visual inspection. An issue of high relevance that could affect
the dynamic properties or the stability GPCR target structures in later simulation ex-
periments is the protonation state of titratable groups. A useful method that can help
in assessing this issue is based on the PROPKA algorithm (
Li, Robertson, & Jensen,
2005
), which is also available as a web server application (
http://propka.ki.ku.dk/
)
.
