Computational Drug Discovery and Design

Newton’s second law of motion. The output is a trajectory which

describes the positions and velocities of the particles in the system

throughout the simulation as a function of time (seeNote 1). In the

following subsections, we describe the MD simulation steps for

membrane proteins, particularly focusing on the 7TM model sys-

tem (i.e., GPCRs).

2.1.1 Preparation
of the System

An all-atom MD simulation for a typical membrane protein system

consists of a membrane protein, ligand, surrounding lipid bilayer,

and water bath with ions. Prior to MD simulations, the ligand and

membrane protein structures should be prepared (for details,see

X-Ray Crystal Structures 1. MD Simulation

Agonist-bound form

Apo form

Apo form Agonist-bound form

3. Network Analysis

Residue network model

(C)

binding siteG-Protein

Ligand binding

site

(B) (A)

3D-Structure

Highest

betweennesscentrality

Highest

degree/closeness centrality

π-π stacking

adenosinewith

for agonismPotent

CWxPmotif

DRY motif &

Ionic lock

NPxxY

motif

2. Generation of Structural Ensembles

Fig. 1Overview of our methodology for the identification of allosteric hot spot residues and pathways of signal
flow in A2AAR signaling. The three major protocols implemented in our method include: (1) MD simulations of
the apo (gray) and agonist-bound (cyan) forms of A2AAR for 300 ns each; (2) Generation of the conformational
ensembles; and (3) Residue interaction network construction and centrality analysis for the receptor’s 3D
structures. The allosteric hot spots and signaling pathways of A2AAR were elucidated using the measure of
betweenness centrality (CB) for each residue in the network, glycine scanning analysis, and cross-correlation
analysis. The figures in the bottom from right to left (A–C) denote network analysis of the A2AAR structure. (A)
Network representation of A2AAR apo form. The residue interaction network was constructed using a cutoff
distance of 7 A ̊;(B) Centrality is the most common concept in network analysis. The characteristics of a node
or the whole network can be deduced via this centrality analysis. The concepts of the three most popular
centrality measures, namely, degree, closeness, and betweenness, have already been detailed (refer to
Subheading2). In the minimal energy structure, the key residues with high betweenness centralities
(CB0.05) were identified as the microswitch residues, which are important for agonistic signal flow or
ligand binding. In addition, the residues with high betweenness centralities create physically connected
networks; (C) Cross-correlation analysis is based on the network analysis for the structural ensembles derived
from MD simulation. The highly correlated, but physically distant residue pairs were extracted to quantify long-
range coupling. The represented apo- and agonist-bound structures of A2AAR show the minimum paths
between the cross-correlated residues. Long-range cross-correlations between the extracellular ligand
binding site and cytoplasmic G-protein binding site are detected for the agonist-bound structure

458 Shaherin Basith et al.