Structure-based design of novel pharmacological tools for the A3
adenosine G protein-coupled receptor
Ashish Vinayak Tamhankar
Degree Project in Pharmaceutical Modelling, 45hp, Autumn/Spring Semester
2020-2021
Examiner: Dr. Christian Sköld
Host lab: Hugo Gutiérrez de Terán, Computational Biology and Bioinformatics program,
Department of Cell and Molecular Biology, BMC.
Division for Drug Design and Development Department of Medicinal Chemistry
Table of Contents
Abbreviations ... 1 Summary ... 2 Abstract ... 3 Introduction ... 4 Aim ... 8 Methods ... 9Protein preparation and receptor-ligand docking ... 9
Membrane insertion ... 10
Free energy calculations ... 10
Results and Discussion ... 15
Free energy calculations on existing mutagenesis data for A3 AR ... 15
Site-directed mutagenesis on A3 AR with pyrimidine scaffold (2g) ... 17
Conclusion ... 26
Acknowledgement ... 27
References ... 28
Abbreviations
ARs - adenosine receptors FEP - Free energy perturbations GPCR - G-protein coupled receptor
hA3 AR - Human A3 adenosine receptor
MD - Molecular Dynamics MAE - Mean absolute error RMSE - Root mean squared error SDM - Site-directed mutagenesis SEM - standard error of the mean TM - Trans-Membrane
Summary 2
Summary
Adenosine receptors (ARs) are transmembrane proteins that belong to the G protein-coupled receptor (GPCRs) family. They are involved in the signal transduction of the extracellular
nucleoside adenosine. They are of four subtypes namely A1, A2A, A2B, A3. The A2A and A1 ARs
have been extensively characterized and studied previously. The A3 AR is the most recently
characterized receptor but not yet crystallized. A3 AR is expressed poorly in all tissues except
mast cells. It is involved in various pathophysiological processes such as asthma, ischemia,
inflammation, and pain. The A3 AR antagonists can be used in the treatment of stroke,
glaucoma, inflammation, asthma, and also in cancer chemotherapy. However, due to high homology between the AR subtypes, previous studies encountered pharmacological and
chemical problems in developing A3 AR antagonists. Moreover, the signaling mechanism of
the A3 AR is still controversial in several pathophysiological processes. Therefore, there is a
need for new pharmacological tools to investigate the pathophysiological roles of this receptor.
In this project, we first established the A3 AR antagonist binding mode by performing free
energy perturbations on the experimental site-directed mutagenesis data. The comparisons between the calculated and experimental results showed a good correlation thus validating the binding mode. Thereafter, we performed in-silico site-directed mutagenesis using free energy
perturbations on the A3 AR with a potent and selective antagonist (2g) from the
4-acetamidopyrimidine series. The results displayed the impact on the binding affinity of the antagonist by residue mutations. This enabled the identification of residues involved in ligand
recognition and antagonist binding on the A3 AR. The study of novel mutations on the A3 AR
will thus further enable the development of chemically simple, potent, and selective A3 AR
antagonists as well as in the characterization and crystallization of the A3 AR. The most impact
Abstract
Adenosine receptors (ARs) are a family belonging to the GPCR superfamily, involved with multiple physiological processes and widely distributed throughout the body. In this study, we
focus on the A3 AR due to its vast scale applications in various pathophysiological conditions
such as inflammation, pain, allergic asthma, and cancer chemotherapy. We report the validation
of the binding mode of A3 AR antagonists, through a comprehensive in-silico mutagenesis
study using free energy perturbations, reproducing prior experimental site-directed
mutagenesis data for A3 AR antagonists. After validating the antagonist binding mode, we
performed an extensive in-silico site-directed mutagenesis study on the hA3 AR using potent,
selective, and structurally simple A3 AR antagonist based on the
N-(2,6-diarylpyrimid4-yl)acetamide scaffold (pyrimidine core). The results of this study will be used to design in-vitro site-directed mutagenesis performed by collaborators (University of Barcelona). Once the binding mode of this scaffold is validated, it will be the basis for the design of compounds with
two well-defined functionalities: a fluorophore moiety and bivalent ligands that target A3
dimers. The discovery of novel mutations on the hA3 AR is a step forward in the development
of both chemically simple, potent, and selective A3 AR antagonists as well as in the
Introduction 4
Introduction
G protein-coupled receptors (GPCRs) are transmembrane proteins that are involved in signal transduction across the cellular membrane and represent about three percent of the human
encoding genome1,2. They are the largest family of membrane receptors and display an
important role in cellular communication, and about 30 to 40% of marketed drugs target a
GPCR2,3. However, most potential GPCR drug targets for several indications do not have a
drug on the market1,2. GPCR activation is promoted by the binding of a ligand in the
extracellular side which stabilizes the active conformation of the receptor leading to the cellular
response through dimerization with the intracellular G protein3. GPCR ligands compete with
the endogenous agonist at the orthosteric site and either mimic (agonist) or block (antagonist)
the pharmacological effect3. Structurally, GPCRs display characteristic and conserved features
as a seven-transmembrane helical bundle, which are connected by three intracellular and three
extracellular loops2,3. Different sequence analyses showed that the largest differences of GPCR
occurred in the ligand-binding site (the extracellular side)3.
The superfamily of GPCRs is divided into classes (A-F), out of which the rhodopsin-like (class
A) are the best-characterized receptors2,3. Adenosine receptors (ARs) are of particular interest
as they play an important role in signal transduction for the extracellular nucleoside adenosine4.
The AR family is further divided into four subtypes namely A1, A2A, A2B, and A3 which mediate
the signaling of adenosine5. Initially, ligand-based drug design techniques such as quantitative
structure-activity relationship (QSAR) and computational ligand-based virtual screening, were used for assessing the chemistry of GPCRs due to the lack of structural information within the
GPCR superfamily3,6. In the last decade, due to advances in protein engineering and protein
A2A AR enabled the application of a structure-based drug design approach on ARs7. However, due to high homology between the AR subtypes which exhibit different pathophysiological functions, structure-based drug design led to pharmacological problems such as a decreased
selectivity4. To avoid these pharmacological problems highly elaborated chemical scaffolds
were generated which instead led to chemical problems (complex heterocycles with poor
pharmacokinetics)4. Therefore, despite the large-scale pharmacological interest in GPCRs only
A2A AR agonist regadenoson and A2A AR antagonist istradefylline have been marketed to
date8,9. Recent crystallization of A
1 AR with a covalently bound xanthine antagonist enabled
to improve understanding of the receptor-ligand interactions which led to high selectivity and affinity10.
A3 AR is the most recently characterized member of the AR family, albeit not yet crystallized11.
It is a high-affinity receptor but is poorly expressed in all tissues except mast cells4,12,13. A
3 AR
activation has been shown to elevate intracellular Ca2+ and IP3 levels by inhibition of adenylate
cyclase, which results in the release of inflammatory and allergic mediators from mast cells12.
A3 AR is involved in several pathophysiological processes such as inflammation, pain, and
allergic asthma, and antagonists of A3 AR are predicted to be of use in the treatment of stroke,
glaucoma, inflammation, asthma, COPD14, and in the development of antiallergic15 and
cerebroprotective drugs16. Potentially, A3 AR antagonists also have applications in cancer
chemotherapy due to the high level of expression in several cell lines17. However, the signaling
mechanism of A3 AR is still controversial, the two personalities of A3 AR being in direct
contrast, e.g., in ischemia, inflammation, and cancer (behaving in two different pathways)18,19.
Introduction 6
The host lab has been extensively involved in the discovery and elucidation of the molecular design of AR antagonists for the past decade. The work includes generation of automated workflow for small molecule and protein mutations free energy calculations in Q (QligFEP and
QresFEP)20,21, structure-based designing of AR ligands4, structural mapping of AR mutations22,
structure-based design of the four AR subtypes, and development of structurally simple, potent,
and selective A3 AR antagonists with an integrated approach combining succinct and efficient
synthetic methodologies and rational computer-based drug design12,13,23. Initial studies showed
a series of potent and selective 4-amidopyrimidines with variations at three points of diversity on the pyrimidine scaffold, with identical aryl substitutions at L2 and L3, and alkyl substitution
at L1 as shown in figure 1A12. The binding mode of A3 AR antagonists of this series was
established from this study, which showed a double hydrogen bond with N2506x55 and π-π
stacking between the central heterocycle and F16045x52 was essential for activity12 (refer SI for
residue numbering scheme). Based on the findings in this study, recently, two new series of 4-acetamidopyrimidines and 2-acetamidopyridines with aryl substitutions at L2 and L3 along
with acetamide at L1 were developed (figure 1B)13. Compound 2g which has methoxyphenyl
substitutions at L2 and L3 as shown in figure 1C from the N-(2,6-diarylpyrimidin-4-yl)acetamide series was identified as the most potent, selective, and structurally simple
antagonist of the A3 AR (Ki of 3.6 ± 0.2 nM)13. The corresponding compounds in the pyridine
series showed that replacement of the second nitrogen atom in the heterocycle with a C-H group through bioisosteric replacement reduced the binding affinity of the compounds,
suggesting the N1 and N3 in the heterocycle are essential for binding with N2506x55 together
Figure 1. (A) 4-amidopyrimidine scaffold with variations at three points of diversity L1, L2, and L3. (B) N-(2,6-diarylpyrimidin-4-yl)acetamide scaffold with identical aryl substitutions at L2 and L3 along with acetamide at L1. (C) Compound 2g based on the acetamide scaffold with methoxy aryl (4-Me-O-Ph) substitution at L2 and L3 used to explore new mutations on A3 AR in this study.
Aim 8
Aim
The project focuses on two well defined goals as described below:
1) Performing free energy perturbations (FEP) on available experimental site-directed
mutagenesis (SDM) data for A3 AR antagonists. The correlation between the
experimental and calculated results will provide validation of the hA3 AR model and
in-silico approach used in this study.
2) Performing FEP simulations on the N-(2,6-diarylpyrimidin-4-yl)acetamide scaffold (2g) generating an extensive in-silico site-directed mutagenesis study. This will be used in determining the impact on binding affinity by residue mutations thus elucidating
residue involvement in antagonist binding on A3 AR. The in-silico results obtained will
Methods
The previously published equilibrated homology model of hA3 AR with antagonist 2g was used
as a starting structure13. Experimental SDM data for antagonists of A3 AR was extracted from
GPCRdb2,24. The following computational strategy (figure 2) was followed to validate the
binding mode of A3 AR antagonists: 1) Protein preparation and receptor-ligand docking 2)
Membrane insertion 3) Free energy calculations. For proposing new mutations on A3 AR using
the selective and potent antagonist 2g, the equilibrated homology model of hA3 AR13 was
directly subjected to free energy calculations.
Figure 2. Representation of the computational strategy used in this study. Ten replicates were run for each mutation studied.
Protein preparation and receptor-ligand docking
The protein structure was prepared by removing the equilibrated waters and lipid structures
using the protein preparation wizard in Maestro25. The identification of side-chain protonation
states for Asp/Glu/His rotamers was done with Molprobity web server (http://molprobity.biochem.duke.edu)26. The reported ligand files were acquired from PubChem and the 3D coordinates of the ligands were generated with LigPrep and subsequently
docked to the prepared protein using Glide27. Refined docking poses were obtained by
Protein
preparation Ligand docking
Membrane insertion
Methods 10
superimposing to published antagonist bound crystal structures of A2A AR and A1 AR, based
on chemotype of respective ligand. The ligands were then geometrically optimized by energy minimization of the ligand using the Schrodinger suite (Macromodel Minimization, OPLS3e force field), and a docking pose with the lowest structural deviation was maintained.
Membrane insertion
These resulting protein-ligand complex structures were then subjected to membrane insertion
using PyMemDyn28. It is a python library that enables automated membrane insertion of the
proteins through molecular dynamics (MD) simulations28. The structure is embedded in a
pre-equilibrated hydrated POPC (1-Palmitoyl-2-oleoyl-phosphatidyl-choline) membrane along the vertical axis of the transmembrane (TM) bundle. A more comprehensive characterization of the model was achieved by incorporating the non-protein elements, in this case, ligand and sodium ion, in the system with appropriate force field files which were generated using
LigParGen29. This system is then solvated with bulk water and then inserted into a hexagonal
prism-shaped box and energy minimized using Gromacs 4.630 using OPLS-AA force field
parameters31 for protein and ligand while using Berger parameters for the lipids32. The starting
restraints are gradually released throughout the equilibration by decreasing the force constants
under periodic boundary conditions for 5 ns following the PyMemDyn protocol28.
Free energy calculations
These equilibrated receptor-ligand structures were then transferred to open-source MD
software Q33 for free energy perturbation (FEP) calculations under spherical boundary
conditions. The estimation of the effects of the single point mutations on the ligand binding
was done using QresFEP21,34. FEP simulations were performed for residues within 5 Å distance
of the ligand, and which had a binding affinity value reported in the experimental SDM data
on GPCRdb 2,24. A 25 Å radius sphere was centered on the center of geometry of the respective
ligand were used as the sphere center) and protein atoms within the sphere boundary had a
positional restraint of 20 kcal/mol/Å2. The solvent atoms were subjected to polarization and
radial restraints using the surface constrained all-atom solvent (SCAAS) model to mimic the
properties of bulk water at the sphere surface33,35. Solute atoms outside the sphere are excluded
from the calculation of non-bonded interactions and are tightly constrained with a force
constant of 200 kcal/mol/Å2. Electrostatic interactions of atom pairs at a distance longer than
10 Å were treated with the local reaction field method36, except for atoms undergoing FEP
transformation were the cut off was applied. Solvent bonds and angles were constrained using
the SHAKE algorithm37. Residues outside the sphere were considered in their neutral form to
avoid dielectrical screening as described elsewhere21. Residue parameters were translated from
OPLS-AA/M force field parameters38 which are already implemented in Q33, whereas the
OPLS-AA force field parameters for the ligands were obtained using LigParGen web server29.
The simulation sphere was warmed up from 0.1 to 310 K during the first equilibration period
of 0.61 ns, where initial restraint of 25 kcal/mol/Å2 was imposed on all heavy atoms and was
slowly released. After that, the system was subjected to unrestrained MD simulation, starting with a 0.25 ns unbiased equilibration period followed by FEP sampling, where the atom transformations occur between initial and end states, linearly distributed in FEP lambda windows. The sampling was replicated on 10 independent MD simulations with different initial velocities and each of them consisting of 20 ps sampling per l window using a 2 fs time step in all cases.
For compound 2g, ligand force field parameters were acquired similarly to that of other ligands and FEP simulations were performed adopting the same MD parameters as described above for all the residues within 5 Å distance of the ligand, which were identified through the
alaSCAN script in QresFEP21. Based on the initial alanine mutants the possible residues
Methods 12
alanine based on selectivity hotspots, chemical intuition, possible hydrophobic interactions, and prior experimental SDM data available for other ARs (refer table S1).
The QresFEP protocol uses a single topology for amino acid perturbations and is based on the
alanine scanning protocol21. The residue side chain transformation to alanine is divided into
different stages, starting at the most topologically distant atom from the Cb atom, and atom annihilations occur gradually for each charge group (as defined on the OPLS force field) and starts a series of subperturbations: 1) initially the partial charges are removed, 2) Lennard-Jones potentials are transformed into smoother soft-core potentials, before the annihilation of the corresponding group of atoms, and 3) restoring the partial charges of the final species (refer
figure 3)21,39. Depending on the nature of the sidechain involved the number of perturbation
stages needed for the full annihilation differ, ranging from four (Ser) to nine (Trp), where each subsequent stage is linearly distributed into 20 l windows. To complete a thermodynamic cycle, the same sidechain annihilation is simulated in the apo structure of the protein (keeping the same MD parameters), so that the energetic difference between these two processes equals the binding affinity shift due to the mutation (figure 4).
Figure 3. Overview of the alanine scanning protocol used in this study. The Trp side chain is gradually broken down in a protein system in 9 FEP subperturbations. Annihilation of partial charges and introduction of soft-core potentials are indicated on the molecular structure with gray color and orange dots, respectively21.
In all cases where the wild type (WT) residue was mutated to a non-alanine residue, the WT residue was initially mutated to desired non-alanine residue using the Pymol mutagenesis wizard, which generates desired residue rotamers based on the backbone. The rotamer with the lowest structural deviation, low clashes, and appropriate side-chain positioning was determined
as the appropriate rotamer. This mutated protein was used as the starting point for FEP simulation (e.g., MUT to Ala) with similar MD settings as described above. The two thermodynamic cycles (e.g., WT to Ala and MUT to Ala) were then combined through their common alanine leg as illustrated in figure 5, and the total binding free energy of the mutation was analyzed.
Figure 4. The thermodynamic cycle of the residue mutations for a single point mutation (alanine scanning protocol) on the protein where the WT and mutated protein are shown in blue and orange with WT and MUT denoted respectively. The ligand is represented by green color and denoted with L. FEP calculations are carried for both the holo (with the ligand present) and apo (in absence of the ligand) systems and to calculate the relative binding affinity.
Figure 5. Schematic representation of a thermodynamic cycle of a non-alanine mutation (e.g., Trp to Phe). The mutation is divided into two thermodynamic cycles, (A) represents mutation from WT (Trp) to Ala, while (B) represents the mutation from Pymol mutated (Phe) to Ala. The two thermodynamic cycles are joined by their
Methods 14
common alanine leg and the total free energy of the mutation (Trp to Phe) can be calculated using equation 3. The WT and mutated protein (Ala) is shown in blue and orange with WT and MUT1 while MUT2 denotes (Phe) and is shown in yellow. The ligand is represented by green color and denoted with L. The relative binding free energy is calculated by adding up the two thermodynamic cycles as described in equation 4.
The average binding free energies for all the residue transformations (10 replicates) were estimated by solving the thermodynamic cycle using the Bennet acceptance ratio (BAR)
method as described in equation 121.
∆𝐺! = −𝛽"#𝑙𝑛 〈#&'
!"(∆% ∆'(!)( ) 〉(+,
〈#&'+"(∆% ∆'(!)( ) 〉( + 𝐶! (1)
where the constants 𝐶i are optimized iteratively so that the two ensemble averages become
equal, yielding ∆𝐺𝑖 = 𝐶𝑖.
The effect of the mutation on the binding affinity of the ligand was estimated by calculating the difference between the average binding free energies in holo and apo structures as given in equation 2.
DDGcalc = DGholo - DGapo (2) DDGcalc = DDGcalccycleA+ (- DDGcalccycleB) (3)
In the case of non-alanine mutations, the DDGcalc of each thermodynamic cycle were calculated
using equation 2 respectively and the total binding free energy of the mutation was estimated
by calculating the difference between DDGcalc of the two thermodynamic cycles as mentioned
Results and Discussion
Free energy calculations on existing mutagenesis data for A3 AR
Based on the experimental site-directed mutagenesis (SDM) data collected from GPCRdb 2,24
for four A3 AR antagonists namely CGS15943, MRS1220, PSB11 and XAC (figure S1) 40,41.
Relative binding free energy changes between mutant and WT receptors were calculated from
Ki values for each mutant40,41 and reported in table 1 (DDG
exp). The relative binding free energy
for each mutant calculated based on 3D models of the respective receptor-ligand complexes
using FEP simulations are reported as DDGcalc (table 1). Both the experimental and calculated
binding free energy values were reported along with their standard error of mean (SEM). The
correlation coefficient (R2) between the experimental and the calculated binding free energy
values, root mean squared error (RMSE) and mean absolute error (MAE) values for each set of mutants for respective ligands are included in table 1.
The three mutants analyzed for CGS1594340 showed good agreement with the experimental
SDM data with an R2 of 0.78 and MAE = 0.44 ± 0.17 kcal/mol while showing a low RMSE of
0.53. MRS122040 was analyzed for five mutants as described in table 1, calculated binding free
energy for T86A3x36 mutant displayed discrepancy with the experimental binding free energy,
while the rest of the mutants tested for this ligand showed good correlation with the
experimental binding free energy values with MAE = 0.48 ± 0.20 kcal/mol and R2 value of
0.68. T86A3x36 mutant also showed disagreement between calculated and experimental results
for PSB1141. A R2 of 0.72 and RMSE of 0.44 were calculated excluding this mutant for
PSB1141. A low MAE of 0.64 ± 0.35 kcal/mol was observed for the mutants on PSB1141.
L236A6x49 mutant which showed a slight increase in binding affinity in experimental data,
Results and Discussion 16
XAC40 showed an excellent correlation between the experimental and calculated binding free
energy values with an R2 of 0.80 and an MAE of 0.46 ± 0.16 kcal/mol with an optimal RMSE
value of 0.53.
The discrepancies for the threonine to alanine mutations can be attributed to the fact that the annihilated side chain may or may not be replaced by a water molecule in the simulations. The
SEM of W2356x48 mutants for all the four ligands displayed large variations. This occurs due
to the large perturbations in this mutation, one way to achieve a low SEM value would be to
increase the l windows, which would then cause a slower release of the constraints which
might help in stabilizing the atoms during perturbations. A large difference in the experimental
and calculated SEM values was also observed in the H264E7x42 mutant for XAC40, which was
expected due to charge integration in the binding pocket because of the charged glutamine side chain. Overall, the calculated binding free energy values were in close correspondence with the
SDM data as illustrated in figure 5, thus validating the binding mode of the A3 antagonist on
the hA3 AR model and in-silico computational strategy used in this study.
Table 1. Experimental binding affinity values from SDM data are converted to relative binding free energy values (DDGexp kcal/mol) along with their SEM using (DDG = RTln (Kimut/KiWT) and calculated binding free energy
(DDGcalc kcal/mol) with SEM (kcal/mol) for single point mutations that were reported on GPCRdb which were
within 5Å of the ligand. R2, RMSE, and MAE (kcal/mol) values for the set of mutants for each ligand are reported.
The R2 and RMSE values calculated for PSB11 indicated with a (*) were calculated excluding the mutant
Figure 6. DDGexp and DDGcalc (kcal/mol) values for each residue point mutation for each reference ligands with
corresponding error bars indicating a close correlation between experimental and calculated binding free energies. The experimental results (DDGexp)are shown in blue while the calculated results (DDGcalc)are shown in orange. Site-directed mutagenesis on A3 AR with pyrimidine scaffold (2g)
The initial alanine scanning FEP results for all residues within 5 Å of 2g (31-residue mutations) have been reported in table 2, with their calculated relative binding free energies and SEM. Relative binding free energy value greater than ± 1 kcal/mol was considered to be significantly influencing the binding affinity, while an SEM in the range of 0.6 - 0.8 kcal/mol was considered
to be efficient21. A negative relative binding free energy value indicates an increase in binding
Ligand Mutant ΔΔGexp ΔΔGcalc R2 RMSE MAE
Results and Discussion 18
affinity, while a positive DDG value translates to be detrimental for binding affinity. Figure 7 illustrates the residue mutations with the recorded effect on the binding affinity of 2g.
Table 2. The binding free energy values calculated (DDG(kcal/mol)) along with their SEM (kcal/mol) for initial
alanine scanning mutations using QresFEP for each of the 31 mutants within 5Å of 2g. Relative binding free energy value (> ± 1 kcal/mol) was considered to be significantly influencing the binding affinity, while an SEM in the range of 0.6 - 0.8 kcal/mol was considered to be efficient. A negative relative binding free energy value indicates an increase in binding affinity, while a positive DDG value translates to be detrimental for binding affinity.
Mutant ΔΔG (kcal/mol) SEM (+/-)
Figure 7. Illustration of the equilibrated homology model of hA3 AR with residue mutations according to the
recorded effect on the binding affinity of 2g (blue). Magenta indicates a significant decrease in binding affinity (> ± 1 kcal/mol), orange displays the residues resulting in a slight decrease in binding affinity (< ± 1 kcal/mol). While Dark green indicates the residues with a significant increase in binding affinity and light green indicates residues showing a slight increase in binding affinity.
The residue T863x36 on TM3 is involved in triggering receptor activation and plays a role in
thermostabilizing the GPCR22. The mutant T86A3x36 showed a DDG value of -1.39 kcal/mol
indicating a significant increase in antagonist binding affinity. The polar side chain of threonine is involved in hydrogen bond interactions with the water network, replacement by aliphatic side chain of alanine disrupts these, but significantly increases the hydrophobicity within the
binding pocket. Experimental SDM data for A2A AR showed this mutation had a minimal
increase in binding affinity for ZM24138542. However, this response is in direct opposition
compared to the SDM data for A3 AR antagonists, where it induced a slight decrease in the
binding affinity41.The mutation at L2366x49 to alanine resulted in a DDG of -2.29 kcal/mol
indicating a large increase in binding affinity. Experimental SDM data of A3 AR antagonists
suggested mutation to alanine caused a minimal increase in binding affinity40,41. Although there
Results and Discussion 20
indicating no direct involvement in ligand recognition. Therefore, this residue was not considered interesting.
S2396x52 is a conserved residue in A3 AR on TM6 (figure 8) and is a selectivity hotspot.
S239A6x52 mutant resulted in a significant increase in binding affinity with a DDG of -1.94
kcal/mol, but with a high SEM of 0.96 kcal/mol. The aliphatic side chain of alanine increases hydrophobicity, and the shorter side chain allows more relaxed docking of the ligand (figure
S2B). Experimental SDM data for A3 AR antagonists showed an insignificant or slight decrease
in binding affinity 40,41. When mutated to histidine (Hie), it was expected to result in a negative
impact on the binding affinity due to steric clashes with the ligand, but results showed a large
increase in binding affinity (DDG = -2.88 kcal/mol) (table 3) as compared to the S239A6x52
mutant due to hydrogen bond interaction between protonated nitrogen of histidine and the 4-Me-O of 2g (figure S2C).
V642x63 is replaced by isoleucine in other AR subtypes (figure 8). Initial mutation to alanine
displayed a slight increase in binding affinity. When mutated to isoleucine, a more hydrophobic effect imparted by isoleucine created a more compact hydrophobic packing around the binding pocket which increased binding affinity significantly (DDG of -1.96 kcal/mol and SEM of 0.96
kcal/mol) in comparison to the alanine mutation. The residue V16145x53 is conserved in A3 AR
in the EL2 and is involved in the hydrophobic packing and is also associated with stabilizing
the protein. V161A45x53 mutant displayed a slight increase in binding affinity. Replacement by
glutamic acid shows water bridging hydrogen bond interactions with 2g (figure S2D) giving an increased effect in binding affinity (DDG = -2.87 kcal/mol). A high SEM was obtained due to charge integration of the protonated glutamic acid side chain (table 3).
V572x56 is involved in hydrophobic packing and V57F2x56 mutant increased the binding affinity
(DDG of -0.38 kcal/mol), as compared to V57A2x56 mutant which had a detrimental effect on
binding affinity. Although the V57F2x56 mutant increased the binding affinity, it was below the
significance limit and a high SEM = 1.15 kcal/mol indicating a large variation in the binding
affinity. When the wild type (WT) A612x60 was mutated to glycine, it resulted in decreased
binding affinity with a binding free energy of 1.35 ± 0.77 kcal/mol due to annihilation of the
aliphatic side chain, disrupting the hydrophobic packing. A612x60 is conserved throughout the
AR family and is involved in hydrophobic packing. The A61F2x60 mutant resulted in a lesser
detrimental effect on binding affinity (table 3) than the A61G2x60 mutant, due to an increase in
hydrophobicity. Substitution with serine on A612x60 increased the binding affinity as compared
to the A61G2x60 mutant, by interacting and stabilizing the hydrogen bond network through its
polar side chain, but this effect was below the significance limit (DDG = 0.88 kcal/mol). T793x29 is a conserved residue in A3 AR and is replaced by alanine in other AR subtypes.
Mutation of T793x29 to alanine decreased a binding (DDG of 0.94 kcal/mol) indicating the role
of this residue in the selectivity profile of the antagonists for A3 AR. L813x31 is a conserved
Mutant ΔΔG (kcal/mol) SEM (+/-)
Results and Discussion 22
residue in A3 AR and when mutated to alanine decreased binding affinity, however, the side
chain is pointing away from the binding site suggesting no direct involvement in ligand
recognition. The neighboring L823x32 is a conserved residue in A
3 AR and is involved in
hydrophobic packing, the L82A3x32 mutant resulted in the loss of binding affinity. When
mutated to valine, a DDG value of 0.42 kcal/mol lesser than the corresponding alanine mutant suggested an increase in binding affinity. However, this increase was not above the significance limit.
The second extracellular loop (EL2) is associated with the stability of the protein structure22
and M16445x56 fills up the hydrophobic space, mutation to alanine decreases the hydrophobicity
within the binding pocket, decreasing the binding affinity suggesting the involvement of this
residue in stabilizing the protein-ligand complex. Moreover, M16445x56 is a conserved residue
in A3 AR which may suggest the role of this residue in the selectivity profile of the ligands.
The residue M1665x37 on TM5 is present in close proximity to the orthosteric binding site and
experimental SDM data on A1 AR showed that it has decreased binding affinity43. Mutation to
alaninedecreased hydrophobicity, which caused a detrimental effect for antagonist binding
with a DDG of 1.94 kcal/mol (figure S3). The M1695x40 is a conserved residue in all ARs,
M169A5x40 decreased antagonist binding, indicating the involvement of this residue in ligand
recognition, which may be due to methionine aromatic interactions (figure S3), which provide stability and replacement with alanine reduced the binding affinity (DDG = 1.89 kcal/mol, SEM
= 0.74 kcal/mol). Experimental SDM data in A3 AR showed a complete loss of agonist and
antagonist binding suggesting the essentiality of this residue 41.
S1735x43 is conserved in A3 AR and is replaced by asparagine in other AR subtypes and
therefore, is a selectivity hotspot (figure 8). S1735x43 is involved in hydrogen interaction with
1.12 kcal/mol). The S173N5x43 increased the binding affinity with a DDG of 0.58 kcal/mol, significantly lower than the alanine mutant but had a high SEM of 1.06 kcal/mol (figure S5B). A large decrease in binding affinity (DDG = 1.96 kcal/mol, SEM = 0.81 kcal/mol) was
displayed by the F174A5x44 mutant due to drastic loss of hydrophobicity. Experimental SDM
data showed, F174A5x44 an abolished effect on the binding of antagonist XAC in A
2A AR44,
while on A3 AR, it showed a slight decrease in binding affinity41.
Figure 8. Pseudo sequence alignment of human ARs, for residues within 5Å of 2g. Residue positions are mentioned according to the GPCRdb numbering scheme. The residues for which experimental SDM was available are shown in yellow, while the residues with no experimental SDM data available are shown in grey.
I1785x47 on TM5 is deep inside the orthosteric binding pocket and is replaced by valine in other
AR subtypes (refer figure 8). I178A5x47 mutation resulted in a significant decrease in binding
affinity (DDG = 1.91 kcal/mol, SEM = 0.66 kcal/mol), due to loss in hydrophobic packing.
When mutated to glutamine increased binding affinity (table 3) as compared to the I178A5x47
mutant was observed. I178V5x47 mutant resulted in DDG = 1.56 kcal/mol which lower than the
I178A5x47 mutant suggesting a mild increase in binding affinity as compared to the alanine
mutant. F2316x44 has been associated with conformational change which causes an outward
shift of TM6 and is termed to be involved in thermostabilizing the receptor22. F231A6x44
severely decreased the binding affinity (DDG = 2.23 kcal/mol), the loss in binding affinity is due to loss of hydrophobicity in the binding pocket and also indicates the possible involvement of the aromatic sidechain in hydrophobic packing (figure S4).
2x56 2x59 2x60 2x61 2x62 2x63 3x28 3x29 3x30 3x31 3x32 3x33 3x36 45x52 45x53 45x56 5x37 5x40 5x43 5x44 5x47 6x44 6x48 6x49 6x51 6x52 6x54 6x55 6x58 7x34 7x38
A3 AR V L A I V V M T C L L L T F V M M M S F I F W L L S I N I L I
A2A AR A F A I T I I A C F V L T F E V M M N F V F W L L H I N T M I
A2B AR A F A I T I L A C F V L T F E V M M N F V F W L V H V N T M I
Results and Discussion 24
L2386x51 on TM6and mutation to alanine show detrimental effects on binding affinity (DDG =
1.62 kcal/mol). Experimental SDM data in A2A AR showed an abolished activity on antagonist
binding45. L238V6x51 resulted in slight increase in binding affinity (DDG of 1.06 kcal/mol, SEM
= 0.88 kcal/mol) in comparison to L238A6x51 mutant. The mutant I260A7x38 caused a decrease
in binding affinity (table 2) which may be due to loss of optimal hydrophobic packing in the
binding pocket. Similar results were obtained in experimental SDM data on A2A AR mutation
showed a similar drastic decrease in binding affinity for the antagonist XAC44. Both S173N5x43
and S239H6x52 mutants led to a significant increase in the binding affinity respectively (table
3), therefore a double mutant with S173N5x43 and S239H6x52 was analyzed which resulted in a
large decrease in binding affinity with a DDG = 2.85 kcal/mol and a high SEM of 1.68 kcal/mol possibly due to protein instability and steric clashes with the ligand.
The alanine mutations at residue positions L602x59, I622x61, V632x62, M783x28, C803x30, L833x33,
I2416x54, I2456x58, and L2567x24 afforded energy values below significance threshold of ± 1 kcal/mol and were not considered as interesting to elaborate on. Furthermore, binding affinity
results for mutations at W2356x48, N2426x55, and F16045x52 showed a severe decrease in binding
affinity which further supplements the validation of antagonists binding mode in A3 AR.
The in-silico site-directed mutagenesis results performed with 2g enabled to identify the role
of the residues in ligand recognition and antagonist binding on A3 AR. The binding free energy
values are indicative of the extent of involvement of the residue in antagonist binding. 2g has substitution on the extracellular side (L1 part), which allowed evaluation of residues that were not associated with ligand recognition in prior studies. Furthermore, 2g is bulkier and the L2 and L3 region bind deep within the binding pocket, which enabled evaluation of the residues deep inside the orthosteric binding site. The study also provides confirmation of the involvement of the residues that were studied prior in other AR subtypes and illustrates their
be instrumental in development of pharmacological tools (fluorophore moiety and bivalent
ligands that target A3 dimers) for the A3 AR which would enable development of simple, potent
and selective A3 AR antagonists and characterization and crystallization of the A3 AR.
The largest differences in the AR subtypes are in the orthosteric binding pocket which is to on
the extracellular side3. Although compound 2g used in this study has a substituent that extends
towards the extracellular side, the substituent is relatively small. Therefore, there is a need to perform mutagenesis studies on antagonists with a larger substituent that extends towards the extracellular side. Furthermore, the activation of the GPCRs is believed to be through dimerization with the intracellular G-protein. Therefore, there is a need to develop bivalent ligands that would connect through their substituents at L1 and allow better assessment and
evaluation of the A3 AR.
The computational strategy used in this study is robust and can be replicated to other A3 AR
antagonists. QresFEP uses single topology approach which allows accurate prediction of residue mutations but is computationally expensive. The project will act as a base for further
exploration and elucidation studies performed on A3 AR. It also focusses on the largely varied
unexplored extracellular side of the A3 AR, but the ligand used to explore this has relatively
Conclusion 26
Conclusion
In summary, through available experimental SDM data for A3 AR on GPCRdb, we established
the binding mode of the A3 AR with our homology model using FEP simulations (with
QresFEP). Thereafter, we performed an extensive site-directed mutagenesis study for A3 AR
with the N-(2,6-diarylpyrimidin-4-yl)acetamide scaffold (2g) using FEP simulations. The resulting mutagenesis data was used to detect residues involved in ligand recognition for
binding of antagonists on the A3 AR. The initial alanine mutants analyzed were then subjected
Acknowledgement
References 28
References
(1) Hauser, A. S.; Attwood, M. M.; Rask-Andersen, M.; Schiöth, H. B.; Gloriam, D. E. Trends in GPCR Drug Discovery: New Agents, Targets and Indications. Nature Reviews
Drug Discovery 2017, 16 (12), 829–842. https://doi.org/10.1038/nrd.2017.178.
(2) Munk, C.; Harpsøe, K.; Hauser, A. S.; Isberg, V.; Gloriam, D. E. Integrating Structural and Mutagenesis Data to Elucidate GPCR Ligand Binding. Current Opinion in
Pharmacology 2016, 30, 51–58. https://doi.org/10.1016/j.coph.2016.07.003.
(3) Rodriguez, D.; Gutierrez-de-Teran, H. Computational Approaches for Ligand Discovery and Design in Class-A G Protein- Coupled Receptors. Current Pharmaceutical Design
2013, 19 (12), 2216–2236. https://doi.org/10.2174/1381612811319120009.
(4) Gutiérrez-de-Terán, H.; Sallander, J.; Sotelo, E. Structure-Based Rational Design of Adenosine Receptor Ligands. Current Topics in Medicinal Chemistry 2016, 17(1), 40-58. https://doi.org/10.2174/1568026616666160719164207.
(5) Fredholm, B. B.; IJzerman, A. P.; Jacobson, K. A.; Linden, J.; Müller, C. E. International Union of Basic and Clinical Pharmacology. LXXXI. Nomenclature and Classification of Adenosine Receptors - An Update. Pharmacological Reviews 2011, 63 (1), 1–34. https://doi.org/10.1124/pr.110.003285.
(6) Kooistra, A. J.; Leurs, R.; de Esch, I. J.; de Graaf, C. From three-dimensional GPCR
structure to rational ligand discovery. Advances in experimental medicine and biology
(7) Jaakola, V. P.; IJzerman, A. P. The Crystallographic Structure of the Human Adenosine A2A Receptor in a High-Affinity Antagonist-Bound State: Implications for GPCR Drug Screening and Design. Current Opinion in Structural Biology 2010, 20 (4), 401–414. https://doi.org/10.1016/j.sbi.2010.05.002.
(8) Chen, J.; Eltzschig, H.; Fredholm, B. Adenosine receptors as drug targets — what are the challenges?. Nature Reviews Drug Discovery 2013, 12(4), 265-286. https://doi.org/10.1038/nrd3955.
(9) Dungo, R.; Deeks, E. D. Istradefylline: First Global Approval. Drugs 2013, 73 (8), 875– 882. https://doi.org/10.1007/s40265-013-0066-7.
(10) Glukhova, A.; Thal, D. M.; Nguyen, A. T.; Vecchio, E. A.; Jörg, M.; Scammells, P. J.; May, L. T.; Sexton, P. M.; Christopoulos, A. Structure of the Adenosine A1 Receptor Reveals the Basis for Subtype Selectivity. Cell 2017, 168 (5), 867-877.e13. https://doi.org/10.1016/j.cell.2017.01.042.
(11) Zhou, Q. Y.; Chuanyu, L. I.; Olah, M. E.; Johnson, R. A.; Stiles, G. L.; Civelli, O. Molecular Cloning and Characterization of an Adenosine Receptor: The A3 Adenosine Receptor. Proceedings of the National Academy of Sciences of the United States of
America 1992, 89 (16), 7432–7436. https://doi.org/10.1073/pnas.89.16.7432.
(12) Yaziji, V.; Rodríguez, D.; Gutiérrez-De-Terán, H.; Coelho, A.; Caamaño, O.; García-Mera, X.; Brea, J.; Loza, M. I.; Cadavid, M. I.; Sotelo, E. Pyrimidine Derivatives as Potent and Selective A3 Adenosine Receptor Antagonists. Journal of Medicinal
Chemistry 2011, 54 (2), 457–471. https://doi.org/10.1021/jm100843z.
N-(4,6-Diarylpyridin-2-References 30
Yl)Acetamides as Potent and Selective Antagonists. Journal of Medicinal Chemistry
2017, 60 (17), 7502–7511. https://doi.org/10.1021/acs.jmedchem.7b00860.
(14) Polosa, R.; Blackburn, M. Adenosine receptors as targets for therapeutic intervention in asthma and chronic obstructive pulmonary disease. Trends in Pharmacological
Sciences 2009, 30 (10), 528-535. https://doi.org/10.1016/j.tips.2009.07.005.
(15) Forsythe, P.; Ennis, M. Adenosine, Mast Cells and Asthma. Inflammation Research
1999, 48 (6), 301–307. https://doi.org/10.1007/s000110050464.
(16) Von Lubitz, D. K. J. E.; Carter, M. F.; Deutsch, S. I.; Lin, R. C. S.; Mastropaolo, J.; Meshulam, Y.; Jacobson, K. A. The Effects of Adenosine A3 Receptor Stimulation on Seizures in Mice. European Journal of Pharmacology 1995, 275 (1), 23–29. https://doi.org/10.1016/0014-2999(94)00734-O.
(17) Merighi, S.; Mirandola, P.; Varani, K.; Gessi, S.; Capitani, S.; Leung, E.; Baraldi, P. G.; Tabrizi, M. A.; Borea, P. A. Pyrazolotriazolopyrimidine Derivatives Sensitize Melanoma Cells to the Chemotherapic Drugs: Taxol and Vindesine. Biochemical
Pharmacology 2003, 66 (5), 739–748. https://doi.org/10.1016/S0006-2952(03)00400-3.
(18) Jacobson, K.; Klutz, A.; Tosh, D.; Ivanov, A.; Preti, D.; Baraldi, P. Medicinal Chemistry of the A3 Adenosine Receptor: Agonists, Antagonists, and Receptor Engineering.
Handbook of experimental pharmacology 2009, (193), 123-159. https://doi.org/10.1007/978-3-540-89615-9_5.
(19) Gessi, S.; Merighi, S.; Varani, K.; Leung, E.; mac Lennan, S.; Borea, P. A. The A3 Adenosine Receptor: An Enigmatic Player in Cell Biology. Pharmacology and
(20) Jespers, W.; Esguerra, M.; Åqvist, J.; Gutiérrez-De-Terán, H. Qligfep: An Automated Workflow for Small Molecule Free Energy Calculations in Q. Journal of
Cheminformatics 2019, 11 (1), 1–16. https://doi.org/10.1186/s13321-019-0348-5.
(21) Jespers, W.; Isaksen, G. v.; Andberg, T. A. H.; Vasile, S.; van Veen, A.; Åqvist, J.; Brandsdal, B. O.; Gutiérrez-De-Terán, H. QresFEP: An Automated Protocol for Free Energy Calculations of Protein Mutations in Q. Journal of Chemical Theory and
Computation 2019, 15 (10), 5461–5473. https://doi.org/10.1021/acs.jctc.9b00538.
(22) Jespers, W.; Schiedel, A. C.; Heitman, L. H.; Cooke, R. M.; Kleene, L.; van Westen, G. J. P.; Gloriam, D. E.; Müller, C. E.; Sotelo, E.; Gutiérrez-de-Terán, H. Structural Mapping of Adenosine Receptor Mutations: Ligand Binding and Signaling Mechanisms.
Trends in Pharmacological Sciences 2018, 39 (1), 75–89. https://doi.org/10.1016/j.tips.2017.11.001.
(23) Yaziji, V.; Rodríguez, D.; Coelho, A.; García-Mera, X.; el Maatougui, A.; Brea, J.; Loza, M. I.; Cadavid, M. I.; Gutiérrez-De-Terán, H.; Sotelo, E. Selective and Potent Adenosine A3 Receptor Antagonists by Methoxyaryl Substitution on the N-(2,6-Diarylpyrimidin-4-Yl)Acetamide Scaffold. European Journal of Medicinal Chemistry 2013, 59, 235–242. https://doi.org/10.1016/j.ejmech.2012.11.010.
(24) Kooistra, A. J.; Mordalski, S.; Pándy-Szekeres, G.; Esguerra, M.; Mamyrbekov, A.; Munk, C.; Keserű, G. M.; Gloriam, D. E. GPCRdb in 2021: Integrating GPCR Sequence, Structure and Function. Nucleic Acids Research 2021, 49 (D1), D335–D343. https://doi.org/10.1093/nar/gkaa1080.
References 32
Enrichments. Journal of Computer-Aided Molecular Design 2013, 27 (3), 221–234. https://doi.org/10.1007/s10822-013-9644-8.
(26) Chen, V. B.; Arendall, W. B.; Headd, J. J.; Keedy, D. A.; Immormino, R. M.; Kapral, G. J.; Murray, L. W.; Richardson, J. S.; Richardson, D. C. MolProbity: All-Atom Structure Validation for Macromolecular Crystallography. Acta Crystallographica
Section D: Biological Crystallography 2010, 66 (1), 12–21. https://doi.org/10.1107/S0907444909042073.
(27) Friesner, R. A.; Murphy, R. B.; Repasky, M. P.; Frye, L. L.; Greenwood, J. R.; Halgren, T. A.; Sanschagrin, P. C.; Mainz, D. T. Extra Precision Glide: Docking and Scoring Incorporating a Model of Hydrophobic Enclosure for Protein-Ligand Complexes.
Journal of Medicinal Chemistry 2006, 49 (21), 6177–6196. https://doi.org/10.1021/jm051256o.
(28) Gutiérrez-De-Terán, H.; Bello, X.; Rodríguez, D. Characterization of the Dynamic Events of GPCRs by Automated Computational Simulations. Biochemical Society
Transactions 2013, 41 (1), 205–212. https://doi.org/10.1042/BST20120287.
(29) Dodda, L. S.; de Vaca, I. C.; Tirado-Rives, J.; Jorgensen, W. L. LigParGen Web Server: An Automatic OPLS-AA Parameter Generator for Organic Ligands. Nucleic Acids
Research 2017, 45 (W1), W331–W336. https://doi.org/10.1093/nar/gkx312.
(30) Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. GROMACS 4: Algorithms for Highly Efficient, Load-Balanced, and Scalable Molecular Simulation. Journal of
Chemical Theory and Computation 2008, 4 (3), 435–447. https://doi.org/10.1021/ct700301q.
Accurate Quantum Chemical Calculations on Peptides. The Journal of Physical
Chemistry B 2001, 105 (28), 6474−6487. https://doi.org/10.1021/jp003919d.
(32) Berger, O.; Edholm, O.; Jähnig, F. Molecular Dynamics Simulations of a Fluid Bilayer of Dipalmitoylphosphatidylcholine at Full Hydration, Constant Pressure, and Constant
Temperature. Biophysical Journal 1997, 72 (5), 2002–2013.
https://doi.org/10.1016/S0006-3495(97)78845-3.
(33) Marelius, J.; Kolmodin, K.; Feierberg, I.; Åqvist, J. Q: A Molecular Dynamics Program for Free Energy Calculations and Empirical Valence Bond Simulations in Biomolecular Systems. Journal of Molecular Graphics and Modelling 1998, 16 (4–6), 213–225. https://doi.org/10.1016/S1093-3263(99)00012-1.
(34) Boukharta, L.; Gutiérrez-de-Terán, H.; Åqvist, J. Computational Prediction of Alanine Scanning and Ligand Binding Energetics in G-Protein Coupled Receptors. PLoS
Computational Biology 2014, 10 (4). https://doi.org/10.1371/journal.pcbi.1003585.
(35) King, G.; Warshel, A. A Surface Constrained All-Atom Solvent Model for Effective Simulations of Polar Solutions. The Journal of Chemical Physics 1989, 91 (6), 3647– 3661. https://doi.org/10.1063/1.456845.
(36) Lee, F. S.; Warshel, A. A Local Reaction Field Method for Fast Evaluation of Long-Range Electrostatic Interactions in Molecular Simulations. The Journal of Chemical
Physics 1992, 97 (5), 3100–3107. https://doi.org/10.1063/1.462997.
(37) Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. Numerical Integration of the Cartesian Equations of Motion of a System with Constraints: Molecular Dynamics of n-Alkanes.
Journal of Computational Physics 1977, 23 (3), 327–341.
References 34
(38) Robertson, M. J.; Tirado-Rives, J.; Jorgensen, W. L. Improved Peptide and Protein Torsional Energetics with the OPLS-AA Force Field. Journal of Chemical Theory and
Computation 2015, 11 (7), 3499–3509. https://doi.org/10.1021/acs.jctc.5b00356.
(39) Jespers, W., Åqvist, J. and Gutiérrez-de-Terán, H. Free Energy Calculations for Protein– Ligand Binding Prediction. In: Ballante F. (eds) Protein-Ligand Interactions and Drug
Design. Methods in Molecular Biology 2021, 203-226.
https://doi.org/10.1007/978-1-0716-1209-5_12.
(40) Gao, Z.; Chen, A.; Barak, D.; Kim, S.; Müller, C.;Jacobson, K.A. Identification by Site-directed Mutagenesis of Residues Involved in Ligand Recognition and Activation of the Human A3 Adenosine Receptor. Journal of Biological Chemistry, 2002, 277 (21), 19056-19063. https://doi.org/10.1074/jbc.M110960200.
(41) Gao, Z. G.; Kim, S. K.; Gross, A. S.; Chen, A.; Blaustein, J. B.; Jacobson, K. A. Identification of Essential Residues Involved in the Allosteric Modulation of the Human A3 Adenosine Receptor. Molecular Pharmacology 2003, 63 (5), 1021–1031. https://doi.org/10.1124/mol.63.5.1021.
(42) Lane, J. R.; Herenbrink, C. K.; van Westen, G. J. P.; Spoorendonk, J. A.; Hoffmann, C.; IJzerman, A. P. A Novel Nonribose Agonist, LUF5834, Engages Residues That Are Distinct from Those of Adenosine-like Ligands to Activate the Adenosine A2a Receptor.
Molecular Pharmacology 2012, 81 (3), 475–487.
https://doi.org/10.1124/mol.111.075937.
Efficacy. Molecular Pharmacology 2016, 90 (6), 703–714. https://doi.org/10.1124/mol.116.105007.
(44) Kim, J.; Wess, J.; van Rhee, A.; Schöneberg, T.; Jacobson, K.A. Site-directed Mutagenesis Identifies Residues Involved in Ligand Recognition in the Human A2a Adenosine Receptor. Journal of Biological Chemistry, 1995, 270 (23), 13987-13997. https://doi.org/10.1074/jbc.270.23.13987.
(45) Jaakola, V. P.; Lane, J. R.; Lin, J. Y.; Katritch, V.; IJzerman, A. P.; Stevens, R. C. Ligand Binding and Subtype Selectivity of the Human A2A Adenosine Receptor: Identification and Characterization of Essential Amino Acid Residues. Journal of Biological
Supplementary Information 36
Supplementary Information
Generic Residue Numbering
All residues and their mutations have been numbered according to the following scheme22:
A000BTx00
Where:
A = WT residue.
000 = residue number of mutated residue, this number corresponds with the residue numbering of the receptor under investigation.
B = the mutated residue.
Tx00 = the GPCRdb numbering scheme, which is a structure-based update of the Ballesteros–
Weinstein (B–W) numbering scheme. T represents the transmembrane helix number and 00 the correlative residue number. Residues in an inter- or extracellular domain of the receptor are
represented using 00x00, where the first two numbers depict the two transmembrane domains
before and after the inter- or extracellular domain.
Table S1. Represents the hypothesis for residue mutations to other residues apart from alanine. The predicted effect on binding affinity is in relation to the binding affinity observed in initial alanine scanning mutants for respective residue.
Residue Possible involvement MutaQon to Hypothesis based on Expected intensity of effect observed
Predicted effect on binding affinity
V572x56 Hydrophobic packing (water network) F A large side chain would destabilize packing/ if not, then
would increase hydrophobicity major decrease/increase A612x60 Hydrophobic packing F Increased hydrophobicity major increase A612x60 Hydrophobic packing S Interact and stabilize hydrogen bond network major increase V642x63 Hydrophobic packing I Increased hydrophobicity (ILE in A2A and A1 ARs) minor increase L823x32 Hydrophobic packing V More op\mal packing ( VAL in A2a and A1, muta\on awayfrom VAL showed major decrease of abolishment of
binding) minor increase V16145x53 Hydrophobic packing and possibly stabilizing
the protein E Charged interac\on with water (GLU in A2a and A1) major increase S1735x43 Hydrogen bond network N Increased hydrogen bonding network interac\ons (ASN in
A2A and A1 ARs) minor increase I1785x47 Hydrophobic space filling Q Hydrogen bond forma\on with the ligand (Glu N-H to 2g
O-Me) major increase
I1785x47 Hydrophobic space filling V Selec\vity hotspot for A2A/2B/1 minor decrease L2386x51 Hydrophobic space filling V Selec\vity hotspot A2B minor decrease S2396x52 Space filling and H bond interac\ons H (Hie) Selec\vity hotspot for A2A/2B/A1 major increase S1735x43_S2396x52
(Double-mutant)
Involved in hydrophobic packing and in H-bond interac\ons respec\vely
N and H respec\vely
Both S173 and S239 mutants are expected to increase binding, a combina\on of both would lead to even becer
binding
Table S2. Proposed series of ten mutants based on the binding free energy values along with SEM (kcal/mol), according to the position of the residue on the receptor, comparison with experimental SDM data on ARs to be initially tested in-vitro.
Figure S1. 2D structures of the antagonists used to validate the binding mode and the A3 homology model used.
Mutant ΔΔG (kcal/mol) SEM (+/-)
Supplementary Information 38
Figure S2. (A) WT S2396x52 interacting 2g with water bridging H-bond interactions. (B) S239A6x52 allowing
relaxed binding of the ligand. (C) H2396x52 in close proximity of 2g suggesting steric clashes, and H-bond
interactions with 2g. (D) V161E45x53 showing water bridging H-bond interactions.
Figure S3. (A) WT M1665x37 and M1695x40 in close proximity to the ligand and involved in hydrophobic packing.
Figure S4. (A) WT F2316x44 involved in hydrophobic packing and receptor stabilization. (B) F231A6x44 mutation
showing disruption of the hydrophobic packing due to annihilation of the large aromatic side chain of WT residue.
Figure S5. (A) WT S1735x43 involved in hydrophobic interactions with the water molecules imparting a
hydrophobic effect. (B) S1735x43 showing disruption of hydrophobic interactions causing decreased binding
