molecules

Article
A Complete Assessment of Dopamine Receptor-
Ligand Interactions through Computational Methods
Beatriz Bueschbell 1 , Carlos A. V. Barreto 2 , António J. Preto 2 , Anke C. Schiedel 1 and
Irina S. Moreira 2,3, *
1 PharmaCenter Bonn, Pharmaceutical Institute, Pharmaceutical Chemistry I, University of Bonn,
D-53121 Bonn, Germany; bueschbell@uni-bonn.de (B.B.); schiedel@uni-bonn.de (A.C.S.)
2 Center for Neuroscience and Cell Biology, UC- Biotech Parque Tecnológico de Cantanhede, Núcleo 04,
Lote B, 3060-197 Cantanhede, Portugal; cbarreto@cnc.uc.pt (C.A.V.B.); martinsgomes.jose@gmail.com (A.J.P.)
3 Institute for Interdisciplinary Research, University of Coimbra, 3004-531 Coimbra, Portugal
* Correspondence: irina.moreira@cnc.uc.pt; Tel.: +351-231-249-730




Received: 5 February 2019; Accepted: 23 March 2019; Published: 27 March 2019


Abstract: Background: Selectively targeting dopamine receptors (DRs) has been a persistent challenge
in the last years for the development of new treatments to combat the large variety of diseases
involving these receptors. Although, several drugs have been successfully brought to market,
the subtype-specific binding mode on a molecular basis has not been fully elucidated. Methods:
Homology modeling and molecular dynamics were applied to construct robust conformational
models of all dopamine receptor subtypes (D1 -like and D2 -like). Fifteen structurally diverse ligands
were docked. Contacts at the binding pocket were fully described in order to reveal new structural
findings responsible for selective binding to DR subtypes. Results: Residues of the aromatic
microdomain were shown to be responsible for the majority of ligand interactions established to
all DRs. Hydrophobic contacts involved a huge network of conserved and non-conserved residues
between three transmembrane domains (TMs), TM2-TM3-TM7. Hydrogen bonds were mostly
mediated by the serine microdomain. TM1 and TM2 residues were main contributors for the coupling
of large ligands. Some amino acid groups form electrostatic interactions of particular importance for
D1 R-like selective ligands binding. Conclusions: This in silico approach was successful in showing
known receptor-ligand interactions as well as in determining unique combinations of interactions,
which will support mutagenesis studies to improve the design of subtype-specific ligands.

Keywords: dopamine receptors; molecular docking; molecular dynamics; receptor-ligand interactions

1. Introduction

1.1. Dopamine Receptors

The dopaminergic system has been intensively studied over the past 75 years due to the
(patho)physiological role in modulating cognitive and motor behaviour [1,2]. The importance of
dopamine has dramatically emerged from being just an intermediate in the formation of noradrenaline
to having a celebrity status as the most important mammalian neurotransmitter [3]. Dopamine binds to
five distinct dopamine receptors (DRs; D1–5 Receptor), grouped into two classes —D1 -like and D2 -like
receptors— that differ in their physiological effects and signal transduction. The D1 -like receptors,
D1 R and D5 R, are principally coupled to Gs proteins and enhance the activity of adenylyl-cyclase,
whereas D2 -like receptors, D2–4 R, are primarily coupled to inhibitory Gi proteins and suppress
the activity of adenylyl cyclases [1,4]. The DRs belong to the G Protein-Coupled Receptor family
(GPCRs), the largest and most diverse protein family in humans with approximately 800 members [5,6].

Molecules 2019, 24, 1196; doi:10.3390/molecules24071196 www.mdpi.com/journal/molecules

Molecules 2019, 24, x FOR PEER REVIEW 2 of 26

adenylyl-cyclase, whereas D2-like receptors, D2–4R, are primarily coupled to inhibitory Gi proteins
Molecules
and 2019, 24, the
suppress 1196 activity of adenylyl cyclases [1,4]. The DRs belong to the G Protein-Coupled 2 of 26
Receptor family (GPCRs), the largest and most diverse protein family in humans with approximately
800 members [5,6]. GPCRs share a conserved overall fold of seven transmembrane helices (TMs)
GPCRs share a conserved overall fold of seven transmembrane helices (TMs) linked by three
linked by three intracellular loops (ICLs) and three extracellular loops (ECLs). Around 30–40% of all
intracellular loops (ICLs) and three extracellular loops (ECLs). Around 30–40% of all available
available pharmacotherapeutics target this protein family [7].
pharmacotherapeutics target this protein family [7].
Many severe neuropsychiatric and neurodegenerative disorders such as Tourette’s syndrome,
Many severe neuropsychiatric and neurodegenerative disorders such as Tourette’s syndrome,
schizophrenia, Parkinson’s disease and Huntington’s disease are believed to occur as a result of
schizophrenia, Parkinson’s disease and Huntington’s disease are believed to occur as a result
imbalances and alterations in dopamine signaling [8–10]. Moreover, also a wide array of psychiatric
of imbalances and alterations in dopamine signaling [8–10]. Moreover, also a wide array of
disorders such as hallucinations, paranoia, bipolar disorder, gambling, alcoholism, mania,
psychiatric disorders such as hallucinations, paranoia, bipolar disorder, gambling, alcoholism, mania,
depression, eating disorders, movement and hyperactivity disorders are linked to malfunctioning
depression, eating disorders, movement and hyperactivity disorders are linked to malfunctioning
dopaminergic transmission [3,11–13]. The discovery of chlorpromazine, the first antipsychotic drug,
dopaminergic transmission [3,11–13]. The discovery of chlorpromazine, the first antipsychotic
in the 1950s, was the hallmark of the development of a collection of antipsychotics [14] which were
drug, in the 1950s, was the hallmark of the development of a collection of antipsychotics [14]
later reported to commonly bind to the D2R subtype (with different affinity) [4,11]. The “first-
which were later reported to commonly bind to the D2 R subtype (with different affinity) [4,11].
generation/classical” antipsychotics came along with significant motor side effects such as tardive
The “first-generation/classical” antipsychotics came along with significant motor side effects such
dyskinesia, extrapyramidal symptoms and related conditions. These problems were not eliminated
as tardive dyskinesia, extrapyramidal symptoms and related conditions. These problems were not
in “second-generation/atypical” antipsychotics, and others such as weight gain and the “metabolic
eliminated in “second-generation/atypical” antipsychotics, and others such as weight gain and the
syndrome” also appeared [11,15–17]. It was then later discovered that these multiple clinical and
“metabolic syndrome” also appeared [11,15–17]. It was then later discovered that these multiple clinical
adverse effects of several antipsychotics depended on the combination of occupied receptors from
and adverse effects of several antipsychotics depended on the combination of occupied receptors from
other systems such as cholinergic, histaminergic and serotoninergic receptors (but always including
other systems such as cholinergic, histaminergic and serotoninergic receptors (but always including the
the D2R) resulting in non-selective profiles and therefore in an insufficient explanation of the
D2 R) resulting in non-selective profiles and therefore in an insufficient explanation of the mechanism
mechanism of action [11,15]. Until today, no drug has yet been identified with antipsychotic action
of action [11,15]. Until today, no drug has yet been identified with antipsychotic action without
without a significant affinity for the D2R [15]. However, antipsychotics remain a necessary first-line
a significant affinity for the D2 R [15]. However, antipsychotics remain a necessary first-line treatment
treatment for the management of a variety of the already mentioned psychiatric disorders (Figure 1).
for the management of a variety of the already mentioned psychiatric disorders (Figure 1). In fact,
In fact, it is difficult to directly target one of these diseases with one specific antipsychotic, since there
it is difficult to directly target one of these diseases with one specific antipsychotic, since there are
are also numerous cases of non-responding patients to first-line or any antipsychotic treatment or
also numerous cases of non-responding patients to first-line or any antipsychotic treatment or which
which become resistant to treatment over time [11,18].
become resistant to treatment over time [11,18].

Figure 1. Examples for first-line treatments of neurological diseases and selective dopamine receptor
Figure 1. Examples for first-line treatments of neurological diseases and selective dopamine receptor
ligands. Drugs are classified in typical and atypical antipsychotics [19]. The targets of the selective
ligands. Drugs are classified
ligands haloperidole, in typical
7-OH-DPAT, and atypical
SCH23390 antipsychotics
and SKF38393 [19]. The
were colored targets of the selective
in blue.
ligands haloperidole, 7-OH-DPAT, SCH23390 and SKF38393 were colored in blue.
The search for DR subtype selective (foremost D2 R-selective) therapeutics is an ongoing field of
research [15] as current drugs have D2 R/D3 R-affinity or affinity for all DRs [19,20]. It has been proposed
Molecules 2019, 24, 1196 3 of 26

that substituted 4-phenylpiperazine compounds distinguish between D3 R and D2 R selectivity [21,22].

In addition, the aminotetraline derivative 7-OH-DPAT was identified as a selective D3 R agonist [23,24],
whereas it was shown that most D4 R available therapeutics are not selective [22], with only one
exception, haloperidole [25] (Figure 1). Also, finding D1 -like DR targeting ligands, a more challenging
aim [26,27], may improve antipsychotic treatment, as D1 R was also shown to be relevant for modulating
behaviour in health and disease (reviewed in O’Sullivan et al. [28]). So far, SKF38393 was the only
selective agonist attained for the D1 R, while D5 R completely lacks a selective ligand [29,30]. SCH23390
was proposed as the only D1 -like DR selective antagonist [31] (Figure 1).

1.2. Computer-Aided Drug Design

The strive for finding new and effective therapeutics led to a growing interest in the use of
Computer-Aided Drug Design (CADD). Originally developed for high-throughput screening (HTS)
of compound libraries, the use of CADD nowadays plays an important role in drug discovery [32].
The CADD pipeline can be classified into two general categories: structure-based and ligand-based,
dependent on the available information about the topic of investigation [33]. A structure-based CADD
is used when the target, e.g., a receptor, is known and compound libraries can be screened to find
suitable structures for the target. In contrast, ligand-based CADD relies on known active and inactive
compounds with their affinities in order to construct quantitative pharmacophore models and to
perform virtual screening that is carried out target-independently [32]. Both CADD approaches are
only fruitful if sufficient information is present. Structure-based approaches rely on the availability
of the target crystal structure or homologs proteins whereas pharmacophore and other ligand-based
methodologies rely on the existence of a sufficient number of ligands. For example, for GPCRs
potentially involved in Parkinson’s disease, a variety of molecular docking studies were carried out
using resolved crystal structures to which self-synthesized ligands were docked (reviewed in Lemos et
al., [34]). Vice versa, inspection of known ligands was used to build pharmacophores or Quantitative
Structure-Activity-Relationships (QSAR) to screen for new bioactive molecules (reviewed in Lemos et
al., [34]). All in all, CADD is capable of addressing many challenges in hit-to-lead-development and is
currently widely used in the pharmaceutical industry [34,35].

1.3. Aim
Modeling GPCRs remains problematic due to the complex structure of these membrane proteins
and the lack of structural information about the desired receptor to target. However, the recent
boom of resolved X-ray crystallography structures leads to a more promising application of CADD
approaches to this receptor. Herein, we used tools of structure-based CADD to investigate the
receptor-ligand properties of all DR-subtypes with marketed DR therapeutics. In particular, we: (i)
applied homology modeling by using the resolved X-ray crystallography structures of the dopamine
receptors D2 R, D3 R and D4 R (D2 R bound to the atypical antipsychotic risperidone, PDBid: 6CM4 [36];
D3 R bound to D2 R/D3 R-selective antagonist eticlopride, PDBid: 3PBL [37]; and D4 R in complex with
D2 R/D3 R-selective antagonist nemonapride, PDBid: 5WIU [38]) in order to provide models with
structural ligand-free properties; (ii) performed Molecular Dynamics (MD) of the five model structures,
and (iii) performed molecular docking studies of 15 ligands targeting different conformational
rearrangements’ of DR subtypes. The binding energies, number of conformations as well as the
distances between ligands and receptor interacting residues of the binding pocket were calculated for
all complexes. The interaction between ligands and receptors were further analyzed using an in-house
pipeline that takes advantage of BINding ANAlyzer (BINANA), a Python- implemented algorithm
for analyzing ligand binding [39]. BINANA was shown to successfully atomically characterize key
interactions between protein amino acids and ligand atoms, and as such it is a promising approach
to map such interactions in GPCRs [39]. The main goal was to reveal new structural findings to help
explain the mode of binding of the selected ligands and their selectivity for certain DR-subtypes.
Molecules 2019, 24, 1196 4 of 26

2. Results

2.1. Homology Models of Dopamine Receptors D1 R-D5 R Are Stable

The ligand-free D2 -like homology models were generated using the resolved ligand-bound crystal
structures of the D2 R (PDBid: 6CM4) [36], D3 R (PDBid: 3PBL) [37] and D4 R (PDBid: 5WIU) [38]
(over 90% identity). The 3D crystal structures of DRs are typically incomplete, lacking key regions
for intracellular partner coupling such as intracellular loop 3 (ICL3). In contrast, D1 -like DRs lack
their own templates and therefore, the most suitable template to each DR was selected according
to the percentage of similarity obtained upon sequence alignment by BLAST [40] in combination
with ClustalOmega [41]. In fact, the D3 R crystal structure was chosen as template for D1 R (35.0%
identity with BLAST and 39.5% with ClustalOmega), and the D4 R crystal structure was chosen for
D5 R models (total similarity of 35.0% BLAST/39.1% ClustalOmega, check Materials and Methods
section). We also calculated the similarity of the TMs in relation to the respective template and the
results are summarized in Table S1. All TMs of the D2 -like subtypes showed almost 100% identity
with their crystal structure templates, which is also in line with the total similarity. For D1 R an average
TM similarity with its template was 41.0%, compared to a total similarity of 39.5%, while for D5 R
36.0% TM identity was calculated compared to the total similarity of 39.1%. For the D1 -like subtypes
no differences between the TM similarity and the total similarity with their template were obtained.
Furthermore, for D1 R the highest similarity between the model and its template was observed for
TM1-3, whereas for D5 R it was achieved for TM2, TM3 and TM7. Consequently, TM2 and TM3 seem
to be very conserved among all DR subtypes.
The combination of different metrics and scores were used to choose the most accurate models in
order to perform MD and molecular docking: (i) the Discrete Optimized Protein Energy (DOPE) [42]
score, MODELLER’s standard metrics, (ii) Protein Structure Analysis (ProSA-web) [43,44] and (iii)
Protein Quality (ProQ)-LGscore and MaxSub score [45,46]. All final DR models (check Materials and
Methods section) achieved LGscores > 4 and MaxSub scores > 0.5. The highest z-score was obtained
for D4 R model, whereas the lowest were counted for D1 -like DR models.
MD simulations of 100 ns were briefly run for each ligand-free modelled receptor and analyzed
to confirm their stability. Root-Mean-Square-Deviations (RMSD) of Cα atom mean values ranged
from 0.3 nm and 0.5 nm (Figure S1). Each DR model showed good overall stability. However, D1 -like
models showed slightly higher RMSD values than D2 -like models: D1 R (0.48 ± 0.07 nm) and D5 R
(0.49 ± 0.06 nm) vs D2 R (0.35 ± 0.09 nm), D3 R (0.34 ± 0.04 nm) and D4 R (0.36 ± 0.09 nm). This behavior
is justified by the higher homology scores attained for the D2 -like subfamily. Additionally, RMSD of
Cα carbons for individual TM was computed and their average values were listed in Table S2. The low
values obtained for TM’s RMSD further supports the good overall stability of the model structures.

2.2. Dopamine Receptor Binding Pocket Definition

In this work, we used the comprehensive review of Floresca and Schnetz [47], highly cited [48–50],
as a base for the definition of the binding pocket of all DRs (Figure 2). Furthermore, by applying
Ballesteros and Weinstein numbering (B&W) [51], the position of considered critical residues was
more easily comparable between all receptors. This nomenclature is based on the presence of a highly
conserved residue in each TMs [51], the so called X.50, in which X varies between 1 and 7 depending
on the TM helix. The remaining residues were numbered according to their position relative to the
most conserved one.
Molecules 2019, 24, 1196 5 of 26
Molecules 2019, 24, x FOR PEER REVIEW 5 of 26

Figure2.2. Key
Figure Key residues
residues for
formolecular
moleculardocking
dockingofofDopamine
Dopamineto toDR
DRmodels.
models. Residues
Residues were
weredesignated
designated
in
inbold
boldand
andcolored
coloredinincyan,
cyan,TMs
TMsin
inbold
boldand
anditalic.
italic. This
This set-up
set-up was
was kept
kept for
for Figures
Figures S3–S7
S3–S7 and
and highly
highly
corresponds
corresponds to the definition of the binding pocket of Floresca and Schnetz [47]. The area of the
to the definition of the binding pocket of Floresca and Schnetz [47]. The area of the
orthosteric
orthostericand
andsecondary
secondarybinding
bindingpockets
pocketswere
werecolored
coloredand
andviolet,
violet,respectively.
respectively.

Mutagenesis
Mutagenesisstudies studiesinduced
inducedthe thebelieve
believe that forfor
that dopamine
dopamine binding,
binding,thethe
endogenous
endogenous agonist of the
agonist of
DR, a negatively charged aspartate (3.32Asp) forms an ionic bond interaction
the DR, a negatively charged aspartate (3.32Asp) forms an ionic bond interaction with the protonable with the protonable amine
of dopamine
amine [2,50,52].
of dopamine Moreover,
[2,50,52]. it was shown
Moreover, it wasthatshown this that
effectthis
was crucial
effect wasfor ligandfor
crucial binding
ligandand that
binding
this
andamino-acid was not only
that this amino-acid was not conserved among the
only conserved amongDR, butthe also in all
DR, but biogenic
also amine GPCRs
in all biogenic amine [53,54].
GPCRs
Also, a serine microdomain in TM5 (5.42Ser, 5.43Ser, 5.46Ser) was considered
[53,54]. Also, a serine microdomain in TM5 (5.42Ser, 5.43Ser, 5.46Ser) was considered as an important as an important feature
for dopaminergic
feature for dopaminergicbinding binding
in all DRs in as
allitDRs
is believed that the serines
as it is believed that theform hydrogen
serines bonds with
form hydrogen the
bonds
catechol
with thehydroxyls of dopamine,
catechol hydroxyls increasingincreasing
of dopamine, the bindingthe affinity
bindingandaffinity
orientingandligands
orienting in the orthosteric
ligands in the
binding
orthosteric binding pocket [47,52,55–57]. While 5.42Ser seems to be critical, 5.43Ser plays a [47].
pocket [47,52,55–57]. While 5.42Ser seems to be critical, 5.43Ser plays a less important role less
A further microdomain,
important role [47]. A furtherthe aromatic microdomain,
microdomain, consisting
the aromatic of 6.48Trp, consisting
microdomain, 6.51Phe, 6.52Phe and
of 6.48Trp,
6.55His/Asn
6.51Phe, 6.52Phe was andreported to triggerwas
6.55His/Asn thereported
activationtooftriggerthe dopamine receptor.
the activation All dopamine
of the amino-acids in this
receptor.
microdomain share the same hydrophobic face in the water-assessable
All amino-acids in this microdomain share the same hydrophobic face in the water-assessable binding-site crevice, indicating
that any reorientation
binding-site of these residues
crevice, indicating that anyby binding to aofligand
reorientation these would
residues cause steric clashes
by binding and would
to a ligand would
therefore force the residues to reorient themselves in a domino-like
cause steric clashes and would therefore force the residues to reorient themselves in a domino-like fashion, which lastly leads to
the so-called “rotamer toggle switch” [47,50,53,58]. In addition, 6.48Trp
fashion, which lastly leads to the so-called “rotamer toggle switch” [47,50,53,58]. In addition, 6.48Trp was reported together with
6.55His to stabilize
was reported togetherthe position
with 6.55His of thetoligand
stabilize in the
thebinding
positionpocket
of the via π-π-stacking
ligand in the binding[47,58]. Therefore,
pocket via π-
6.48Trp and 6.55His as well as one phenylalanine (6.51Phe) were chosen
π-stacking [47,58]. Therefore, 6.48Trp and 6.55His as well as one phenylalanine (6.51Phe) were chosen for the docking protocol to
mimic
for thethe ligand-binding
docking protocol to onmimic
TM6. the Dependent on the ligand
ligand-binding on TM6. properties
Dependent other onresidues
the ligand of TM3 were
properties
also considered, such as 3.33Val and 3.36Cys. 3.36Cys is believed
other residues of TM3 were also considered, such as 3.33Val and 3.36Cys. 3.36Cys is believed to be to be part of a deeper subpocket
below
part ofthe orthosteric
a deeper subpocketbindingbelow pocket (OBP) [36].
the orthosteric Additionally,
binding Ericksen
pocket (OBP) [36]. et al. reportedEricksen
Additionally, that this et
cysteine wasthat
al. reported a relevant residue
this cysteine wasfora benzamide
relevant residue binding [49]. Regarding
for benzamide binding 3.33Val, it was reported
[49]. Regarding 3.33Val,to
show
it wasinteraction
reported to with
showN-methylspiperdone by Moreira et al. [53]by
interaction with N-methylspiperdone asMoreira
well as with
et al. the
[53]methoxy
as well as ring
withof
nemonapride, determined in the crystal
the methoxy ring of nemonapride, determined in the crystalstructure of the D 4 R [38]. Different authors hypothesized
structure of the D4R [38]. Different that
DRs havehypothesized
authors a secondary that binding DRspocket
have a(SBP)secondarynext to the OBP,
binding which
pocket wasnext
(SBP) confirmed by thewhich
to the OBP, resolvedwas
crystal structures together with computational analyses [37,38,59].
confirmed by the resolved crystal structures together with computational analyses [37,38,59]. Crystal structures of D 2 R (PDBid:
Crystal
6CM4) [36] of
structures and D2D R (PDBid:
R3(PDBid: 3PBL)[36]
6CM4) [37]andandDcomputational
3R (PDBid: 3PBL) data[37]
suggest that 7.43Tyr is also
and computational dataa suggest
crucial
amino-acid
that 7.43Tyr is also a crucial amino-acid for interaction in the SBP [17,36,37]. 2.57Val was shownfor
for interaction in the SBP [17,36,37]. 2.57Val was shown to form a hydrophobic pocket to
antagonists like clozapine
form a hydrophobic pocketand forhaloperidole
antagonists [57]. However, and
like clozapine sincehaloperidole
the OBP is widely explored since
[57]. However, through the
OBP is widely explored through experimental, computational and crystal structure data, there could
be other residues important in the SBP. Detailed information about the literature, mostly regarding
Molecules 2019, 24, 1196 6 of 26

experimental, computational and crystal structure data, there could be other residues important in the
SBP. Detailed information about the literature, mostly regarding D2 -like DR can be found on Table S3.
In Molecules
order to2019,
compare
24, x FORall DRs
PEER ligand-binding properties and specificity, we focused on the mentioned
REVIEW 6 of 26
residues in the OBP. The residues considered flexible in the different dockings were listed in Methods
D2-like DR can be found on Table S3. In order to compare all DRs ligand-binding properties and
and Materials section.
specificity, we focused on the mentioned residues in the OBP. The residues considered flexible in the
2.3.different dockings of
Proof-of-Concept were listed inDocking
Molecular Methods and Materials section.
Success

2.3.Ten conformational
Proof-of-Concept rearrangements
of Molecular were chosen every 5 ns upon a 50 ns stabilization MD run.
Docking Success.
These 10 plus the initial model (time 0 ns) were then subjected to molecular docking of 15 different
Ten conformational rearrangements were chosen every 5 ns upon a 50 ns stabilization MD run.
ligands.
These 10 Theplusresults of the
the initial molecular
model (time 0 docking
ns) were thenweresubjected
evaluated to by AutoDock4.2,
molecular docking which ranks the
of 15 different
possible binding positions by energy level and clusters these positions
ligands. The results of the molecular docking were evaluated by AutoDock4.2, which ranks by RMSD of 2 Å. In addition,
the
thepossible
total number
bindingofpositions
conformations
by energy (NoC)
level in
and these clusters
clusters thesewere counted.
positions by RMSDBinding
of 2 poses with more
Å. In addition,
than
thefive
totalconformations per cluster (NoC)
number of conformations were considered as a valid
in these clusters were ligand
counted. position,
Bindingdespite
poses withthe binding
more
energy (BE) of this pose. All results of the docking can be checked in the
than five conformations per cluster were considered as a valid ligand position, despite the binding Supplementary Information
(Tables
energy S4–S8).
(BE) of this pose. All results of the docking can be checked in the Supplementary Information
As proof
(Tables of concept, redocking of the co-crystalized ligands to the crystal structure templates of
S4–S8).
the D2 R, AsDproof
3 R and of D 4 R [36–38]
concept, was conducted
redocking (Figure S2, ligands
of the co-crystalized Table S9). Receptors
to the and ligands
crystal structure coordinates
templates of
the retrieved
were D2R, D3R and fromD4R [36–38]
PDB files.was conducted
Top clusters(Figure
achievedS2, Table S9). pose
a ligand Receptors and ligands
equivalent to thecoordinates
pose in the
were retrieved
correspondent from PDB
crystal, files. Top
presenting very clusters
small achieved
RMSD values.a ligand pose these
Lastly, equivalent
resultstowere
the pose in the to
compared
correspondent crystal, presenting very small RMSD values. Lastly,
the docking poses of the corresponding DR-models and ligands at time point 0 ns. The binding these results were compared to
the docking poses of the corresponding DR-models and ligands at time
energies of the two sets were found to fall within a similar range. This is a further evidence of docking point 0 ns. The binding
energies
protocol of the two sets were found to fall within a similar range. This is a further evidence of docking
reliability.
protocol
For a reliability.
general overview, dopamine docking was analyzed in detail (Figure 3A) as it is the
For a general overview, dopamine docking was analyzed in detail (Figure 3A) as it is the
endogenous ligand of the DRs and its binding mode is well-known compared to the other ligands [47].
endogenous ligand of the DRs and its binding mode is well-known compared to the other ligands
However, we have to stress out the lack of a crystal structure with the dopamine-bound DR as the
[47]. However, we have to stress out the lack of a crystal structure with the dopamine-bound DR as
ligand’s structural properties are not suitable for crystallization (too small, not suitable for stabilizing
the ligand’s structural properties are not suitable for crystallization (too small, not suitable for
a GPCR). We observed that the binding energy of D2 R was the most stable at different analyzed MD
stabilizing a GPCR). We observed that the binding energy of D2R was the most stable at different
conformations, while for the other subtypes it oscillated more frequently. Over time the average binding
analyzed MD conformations, while for the other subtypes it oscillated more frequently. Over time
energy for all binding
the average DRs wasenergy foundfor to all at −9was
be DRs kcal/mol.
found toThe be athighest NoC during
−9 kcal/mol. all MD
The highest NoC conformations
during all
were obtained
MD conformations for D R and D
4 were obtained
1 R (up to > 80 for D R at 95 ns), whereas for D R around
for D4R and4 D1R (up to > 80 for D4R at2 95 ns), whereas 30 conformations
for D2R
were counted for all conformational arrangements (Figure 3B). Lastly,
around 30 conformations were counted for all conformational arrangements (Figure 3B). Lastly, for for all DRs complexed with
dopamine, the first or the second cluster with the lowest binding energy also
all DRs complexed with dopamine, the first or the second cluster with the lowest binding energy also contained the highest NoC,
indicating
containedthat thethe docking
highest NoC,ofindicating
dopamine was
that theindeed
docking stable and reliable
of dopamine was(Tables
indeed S4–S8).
stable and In reliable
summary,
the(Table
binding energy
S4–S8). In and 3D positions
summary, of dopamine
the binding energy docking may demonstrate
and 3D positions of dopaminethe binding
dockingmode may of
dopamine
demonstrate to DRs. theAccording
binding mode to Floresca and Schnetz,
of dopamine these
to DRs. features are
According crucial forand
to Floresca dopamine’s binding
Schnetz, these
features
affinity and are
DR crucial for dopamine’s
activation but may not binding affinity be
necessarily andtrue
DRforactivation but may notligands
all dopaminergic necessarily be true
(selective and
for all dopaminergic
non-selective) [47]. ligands (selective and non-selective) [47].

A B
-12 40
Number of conformations
Binding energy [kcal/mol]

D1R D1R
D2R 30
D2R
-10 D3R D3R
D4R 20 D4R

-8 D5R D5R
10

-6 0
0 55 60 65 70 75 80 85 90 95 100 0 55 60 65 70 75 80 85 90 95 100
Time points [ns] Time points [ns]

Figure 3. Results of the molecular docking of dopamine to all DR subtypes at all MD time steps.
(A)Figure 3. Results
The average of theenergy
binding molecular docking
of the of dopamine
three lowest to of
energies alldopamine
DR subtypes
wasatcalculated.
all MD time(B)steps. A.
The mean
of The average binding
the number energy of the
of conformations three
of the lowest
three energies
clusters of the
with dopamine
lowestwas calculated.
binding B. The
energies are mean
shown offor
thetime
each number of and
point conformations
receptor. of the three clusters with the lowest binding energies are shown for each
time point and receptor.
Molecules 2019, 24, 1196 7 of 26

The binding position of dopamine to all DR complexes was stable over time namely, the protonable
amine was always directed towards the aspartic acid on TM3 (3.32Asp) and the hydroxy groups were
facing the serine microdomain (5.42Ser, 4.32Ser and 4.46Ser), in agreement with Floresca and Schetz [47]
Molecules 2019, 24, x FOR PEER REVIEW 7 of 26
andMolecules
Durdagi2019, 24,
et xxal.
FOR [60]PEER REVIEW S3–S8). As known from the literature dopamine’s interaction
(Figures 7 of 26 with
Molecules 2019, 24, FOR PEER REVIEW 7 of 26
Molecules
The2019, 24, x FOR
binding PEER REVIEW
position of typically
dopaminerequires to all DR complexes was stable overtotime namely, 7 ofthe
26
the serine microdomain only two of the
The binding position of dopamine to all DR complexes was stable over time namely, the serines binding the hydroxy groups [47].
The
protonable binding
amine position
was always of dopamine
directed to
towards all DR
the complexes
aspartic acid was
on TM3stable over
(3.32Asp) time
and namely,
the hydroxy the
0 nsThe
At protonable binding
dopamine
amine was position
was located
always of directed
dopamine intotheallOBP
planartowards DRaspartic
the complexes
in the acid was
positionon TM3 stable
described over above.
(3.32Asp) time
and thenamely,
Notably,
hydroxy the D2 R and
protonable
groups
protonablewereamine
facingwas the always directed
serine microdomain towards the
the aspartic
(5.42Ser, 4.32Ser andacid on
on TM3
4.46Ser), (3.32Asp)
in agreement and the
the hydroxy
with Floresca
D4groups
R wereamine
hydroxyl facing
groups was always
thewere
serinemore directed
microdomain towards
directed (5.42Ser,
towards aspartic
4.32Ser
serine acid
and TM3 in
4.46Ser),
microdomain (3.32Asp)
agreement
(Figure andwith
S3). hydroxy
Floresca
At 55 ns torsions
groups
and
groups were
Schetz
were[47]facing
[47] andthe
facing the serine
serine microdomain
Durdagi et al. [60] (Figures
microdomain (5.42Ser,
(5.42Ser, 4.32Ser
S3–S8). As and
4.32Ser known
and 4.46Ser),
fromin
4.46Ser), agreement
the
inthe literaturewith
agreement Floresca
dopamine’s
with Floresca
and
were Schetz
observed and
forthe Durdagi
dopamine et al.
bounded [60] (Figures S3–S8).
to alltypically
DR, As
which known
included from literature dopamine’s
and
and Schetz
interaction [47]
[47] and
with
Schetz with and Durdagi
serine
Durdagi et
et al.
al. [60]
microdomain [60] (Figures
only
(Figures S3–S8). As
As known
requires
S3–S8). requires known two from
fromof atheswitch
the ofbinding
literature
serines
theserines
literature
interactions
dopamine’sto the with the
dopamine’s
interaction
serines at TM5 for the
D serine
R, since microdomain
it is known only
that typically
dopamine is onlytwo of the
capable of binding
interacting to thetwo of the
with
interaction
hydroxy
interaction with
groups
with the the serine
[47]. At 0 ns
3 microdomain
dopamine was only typically
located requires
planar two
in the OBPof the serines
in the binding
position to
to the
described
hydroxy groups [47].serineAt 0 ns microdomain
dopamine was onlylocated
typically requires
planar twoOBP
in the of the
in theserines binding
position described the
three serines
hydroxy
above.
hydroxy [47].
groups
Notably,
groups DAt
[47]. 60At
Atns00 dopamine
ns
ns dopamine
dopamine is shifted
was
was more
located
located to
planar
planar thein
2R and D4R hydroxyl groups were more directed towards serine microdomain
[47]. in serine
the
the OBP
OBP andin
in aromatic
the
the position
position microdomain
described
described (TM6)
above. Notably, D2R and D4R hydroxyl groups were more directed towards serine microdomain
above.
(Figure
for(Figure
all DRs
above. Notably,
S3).
in At D
55 ns
a different
Notably, 2R and D
torsions R
manner.
4 hydroxyl
were groups
observed
However, were
for dopamine
only at Dmore directed
bounded
4 R abounded towards
to all DR,
strong direction serine
which
of microdomain
included protonable
dopamine’s a
S3). At 55D2ns R and D4R were
torsions hydroxyl groups
observed forwere more
dopamine directed towards
to all DR, serine
which microdomain
included a
(Figure
switch
amine ofS3).
towards
(Figure S3). At
At 55
interactions
3.32Asp
55 ns
ns torsions
with
was the
torsions were
serines
observed.
were observed
at At
TM5
observed 65 forfor
ns
for 3dopamine
Ddopamine
dopamine bounded
R, since it isbounded
known that
bounded to
to
to all
all
all DR,
dopamine
DRs
DR, which
was
which included
is only
locatedcapable
included aa planar
again
switch of interactions with the serines at TM5 for D3R, since it is known that dopamine is only capable
switch
of
switch of
of interactions
interacting with two
interactions with the
of the
with serines
three
theindividual
serines at
at TM5
serinesTM5 for
[47].
forAt D
D33R,
At 60
R, since it
it is
is known
ns dopamine
since known that
that dopamine
is shifted more
dopamine to is only
the capable
serine and
in the
of OBP (Figure
interacting withS4). two Small
of the three serines torsions
[47]. were
60 observed
ns dopamine isduring
shifted the
more to is
period only
the of capable
70–90
serine andns (Figures
of
of interacting
aromatic
interacting with
microdomain
with two
two of
of the
(TM6)
the three
for
three serines
all DRs
serines [47].
in
[47].a At 60
different
At 60 ns
ns dopamine
manner.
dopamine is shifted
However,
is shifted more
only
more toat
to the
D
the 4Rserine
a
serine and
strong
and
S4 aromatic
and S5). microdomain
Interestingly,(TM6) at 95 ns fordopamine
all DRs in was strongly
a different involved
manner. in the aromatic
However, only at D4microdomain
R a strong (TM6)
aromatic
direction
aromatic microdomain
of dopamine’s(TM6)
microdomain (TM6) for
for all
protonable DRs
DRs in
amine
all amine aa different
towards
intowards
different manner.
3.32Asp
manner. wasHowever,
observed.only
However, At 65
only at D
D44R
ns
at ns aa strong
dopamine
Rdopamine
strong
direction
at all DR, of dopamine’s
which was then protonable
vanished especially for 3.32Asp
D R at
3(Figurewas
was
100 ns.observed.
The large At 65
decrease in D 4 R binding
direction
bounded
direction to of
to
ofalldopamine’s
all DRs was located
dopamine’s protonable
protonableagainamineplanar towards
amine in the OBP
towards 3.32Asp
3.32Asp S4). observed.
was Small individual
observed. At
At 65 ns
ns dopamine
torsions
65 torsionsdopaminewere
bounded
energy at 90 ns DRs
can wasexplained,
be located again bynsplanar
the in the OBP
approximation (Figure of S4). Small individual
dopamine to953.32Asp were from the
distance
bounded
observed
bounded to to all
during DRs
all DRs the was located
period of again
70–90 planar
(Figurein the
S4 OBP
and (Figure
S5). S4). Small
Interestingly, individual
at ns torsions
dopamine were
was
observed during thewas located
period again ns
of 70–90 planar
(Figurein theS4 OBP
and (Figure S4). Small individual
S5). Interestingly, torsions were
at 95 ns dopamine was
serine microdomain
observed
strongly during
during the
observedinvolved in(Figure
the period
the aromatic
period
S5).
of
of 70–90 ns
ns (Figure
microdomain
70–90 (Figure(TM6)S4
S4 and
and S5).
at all DR,
S5). Interestingly,
which was then
Interestingly, at
at 95 ns
ns dopamine
95vanished
dopamine was
especially
was
strongly involved in the aromatic microdomain (TM6) at all DR, which was then vanished especially
strongly
for D3R at
strongly involved
100 ns.inThe
involved the
the aromatic microdomain
large decrease in D4R (TM6)
binding at all
all DR,
DR, which
energy at 90 nswas then
can bevanished
explained, especially
by the
100 ns.inThe aromatic microdomain
in D4R (TM6) at energy which wascan
then
bevanished especially
2.4.for
for
D3R at
Docking
D 3R at of
approximation
for D3R at 100 of
Various
100 ns.
of The
dopamine
ns.dopamine
large
Ligands
largeto
decrease
to DR
decrease
3.32Asp
The largeto decrease
Models
in D
distance
binding
4R binding
from
in D4R binding the energy
serine
at 90 ns
at 90 ns
microdomain can be
explained,
explained,
(Figure S5).
by the
by
by the
approximation 3.32Asp distance from theenergy at 90 ns can (Figure
serine microdomain be explained,
S5). the
approximation
approximation of
of dopamine
dopamine to
to 3.32Asp
3.32Asp distance
distance from
from the
the serine
serine microdomain
microdomain (Figure
(Figure S5).
S5).
Since non-selective agonistic activity was already covered by dopamine docking, chlorpromazine
2.4. Docking of Various Ligands to DR Models.
was2.4.chosen
Dockingas of Various Ligands to DR
a non-selective Models. [61,62]. Herein, we also selected the following ligands:
antagonist
2.4.
2.4. Docking
Docking of
of Various
Various Ligands
Ligands to
to DR
DR Models.
Models.
Since non-selective agonistic activity was already covered by dopamine docking,
SKF38393 as selective
Since non-selective D1 R agonist
agonistic [27,30] and SCH23390
activity was already as Dcovered
1 -like DRby antagonist
dopamine [31,63], apomorphine
docking,
Since
chlorpromazine
Since non-selective
was
non-selective chosen agonistic
as a
agonistic activity
non-selective
activity was
was already
antagonist
already covered
[61,62].
covered by
Herein,
by dopamine
we also selected
dopamine docking,
the
docking,
chlorpromazine
as selective was
D2 R agonist chosen as a
[60], 7-OH-DPAT non-selective antagonist
as selective [61,62].
D3 R[61,62]. Herein,
agonistHerein, we also
[23], nemonapride selected
as Dthe2 R and D3 R
chlorpromazine
following ligands:
chlorpromazine was chosen
SKF38393 asas a non-selective
selective D 1R agonistantagonist
[27,30] and SCH23390 as we
D also
1-like DRselected
antagonist the
following
selective ligands:was
antagonist [64]
chosenasasselective
SKF38393and lastly
a non-selective
D1R agonist
haloperidole,
antagonist
due[27,30]
to its
[61,62].
and
affinity
Herein,
SCH23390
for D
we
asRD alsoDRselected
1-like
[25]. This antagonist
set of
the
ligands was
following
[31,63],
following ligands:
ligands: SKF38393
apomorphine as
as selective
as selective
SKF38393 D2R D
selective 1R agonist
agonist [60],[27,30] and
and SCH23390
7-OH-DPAT as
as D
4 1-like
as selective D3RDR antagonist
agonist [23],
[31,63], apomorphine as selective D2R D 1R agonist
agonist [60],[27,30]
7-OH-DPAT SCH23390 D1-like
as selective D3RDR antagonist
agonist [23],
chosen
[31,63],as example as D2of ligands withDdifferent DR[64]selectivitylastly(Table 1). Thedueobtained bindingfor energies
[31,63], apomorphine
nemonapride
nemonaprideapomorphine R and
as D2R and
as Dselective
as D 3R selective
selective
3R selectiveD22R agonist
agonist [60],
antagonist
Rantagonist [60], 7-OH-DPAT
and
7-OH-DPAT
[64] and
as
as selective
haloperidole,
selective
lastly haloperidole,
Dto
dueDto
3R its
agonist
3R its
affinity
agonist [23],
affinity[23],
for
andnemonapride
D 4NoC
R [25].in
nemonapride these
This as
setD
as D 2Rligands
of and
clusters D R
are
3wasselective
summarized
chosen antagonist
as in
example [64]
Figure
of and 4
ligands lastly haloperidole,
(graphical
with output
different DR due
of to
the its
selectivity affinity
other for
ligands
(Table 1). can be
D 4R [25]. This set of2Rligands
and D3wasR selective
chosen antagonist
as example[64] and lastly
of ligands withhaloperidole,
different DRdue to its affinity
selectivity (Table for1).
D
The
found 4R [25].
in
D4R [25].theThis
obtained set
appendix:
This bindingof
binding ligands
set of ligandsFigurewas
energies chosen
and
S7).
was and
chosenNoCas example
in
as in these
example of ligands
clusters
of ligands with
are different
summarized DR
in selectivity
Figure 4 (Table
(graphical 1).
The obtained energies NoC these clusters arewith different DR
summarized in selectivity (Table 1).
Figure 4 (graphical
The
The obtained
output binding
of the other
obtained binding energies
ligands can be
energies and
and NoC
found
NoCininin
the
in these clusters
appendix: are
are summarized
Figure S7). in
in Figure
Figure 44 (graphical
output of the other ligands can be found thethese clusters
appendix: Figure summarized
S7). (graphical
output
output ofof the other
other ligands
the Table 1. Ligands
ligands can be
be found
canused in
in the
the appendix:
for molecular
found dockingFigure
appendix: Figure S7).
and information
S7). on their function.
Table 1. Ligands used for molecular docking and information on their function.
Table 1. Ligands used for molecular docking and information on their function.
Table 1. LIGAND
Ligands used for molecular docking and information on their function.
FUNCTION BP REFERENCES
Table 1. Ligands
LIGANDused for molecular docking and information on their
FUNCTION BP function.
REFERENCES
LIGAND FUNCTION BP REFERENCES
LIGAND FUNCTION
LIGAND Endogenous
Endogenous
FUNCTION agonist of BP
agonist BP
REFERENCES
REFERENCES
DOPAMINE
DOPAMINE Endogenous agonist OBP OBP [47,52,65][47,52,65]
DOPAMINE of all DR
all
Endogenous DR
agonist OBP [47,52,65]
DOPAMINE of all DR
Endogenous agonist OBP [47,52,65]
DOPAMINE of all DR OBP [47,52,65]
of all DR
Synthetic D3R
7-OH-DPAT Synthetic
Synthetic D 3D
R3Rselective OBP [47,65,66]
7-OH-DPAT
7-OH-DPAT selective
Synthetic agonist
D3R OBP OBP [47,65,66][47,65,66]
7-OH-DPAT selective
Synthetic agonist
D3R
agonist OBP [47,65,66]
7-OH-DPAT selective agonist OBP [47,65,66]
selective agonist

APOMORPHINE D2R selective agonist OBP [47,52,65]

APOMORPHINE D2R selective agonist OBP [47,52,65]
APOMORPHINE
APOMORPHINE DD2R
2Rselective agonist
selective agonist OBP OBP [47,52,65][47,52,65]
APOMORPHINE D2 R selective agonist OBP [47,52,65]

BROMOCRIPTINE D2R selective agonist OBP [47,65]

BROMOCRIPTINE D2R selective agonist OBP [47,65]
BROMOCRIPTINE D2R selective agonist OBP [47,65]
BROMOCRIPTINE
BROMOCRIPTINE DD2R
2 Rselective agonist
selective agonist OBP OBP [47,65] [47,65]
Molecules 2019, 24, 1196 8 of 26

Molecules 2019, 24, x FOR PEER REVIEW 8 of 26

Molecules 2019, 24, x FOR PEER REVIEW Table 1. Cont. 8 of 26
Molecules 2019, 24, x FOR PEER REVIEW 8 of 26
Molecules 2019, 24, x FOR PEER REVIEW 8 of 26
Molecules 2019, 24, x FOR PEER REVIEW
LIGAND Table 1. Cont. FUNCTION BP 8 of 26
REFERENCES
Molecules 2019, 24, x FOR PEER REVIEW Table 1. Cont. 8 of 26
Molecules 2019, 24, x FOR PEER REVIEW Table 1. Cont. 8 of 26
Molecules 2019, 24, x FOR PEER REVIEW Table 1. Cont. 8 of 26
Molecules 2019, 24, x FOR PEER REVIEW Table 1. Cont. 8 of 26
Molecules 2019, 24, x FOR PEER REVIEW Table 1. Cont.“DIRTY DRUG”, 8 of 26
Molecules 2019, 24, x FOR PEER REVIEW Table 1. Cont.“DIRTY DRUG”, 8 of 26
Table 1. Cont.
“Dirty MULTIPLE
“DIRTY drug”, DRUG”, multiple
CLOZAPINE
CLOZAPINE Table 1. Cont. MULTIPLE
“DIRTY DRUG”, OBP OBP [47,65,67,68] [47,65,67,68]
CLOZAPINE Table 1. Cont. RECEPTOR
receptor
MULTIPLE binding OBP [47,65,67,68]
CLOZAPINE “DIRTY
RECEPTOR DRUG”, OBP [47,65,67,68]
Table 1. Cont. MULTIPLE
“DIRTYBINDING
RECEPTOR DRUG”,
CLOZAPINE MULTIPLE
“DIRTYBINDING DRUG”, OBP [47,65,67,68]
CLOZAPINE RECEPTOR
MULTIPLE OBP [47,65,67,68]
CLOZAPINE “DIRTY BINDING
RECEPTOR DRUG”, OBP [47,65,67,68]
MULTIPLE
BINDING
“DIRTY DRUG”,
CLOZAPINE RECEPTOR
MULTIPLE OBP [47,65,67,68]
CLOZAPINE “DIRTY BINDING
RECEPTOR DRUG”, OBP [47,65,67,68]
“DIRTYMULTIPLE
BINDING
RECEPTOR DRUG”,
CLOZAPINE D2R/D MULTIPLE3R selective
BINDING OBP [47,65,67,68]
CLOZAPINE
NEMONAPRIDE D RECEPTOR
MULTIPLE OBP
OBP + SBP [47,65,67,68]
[38,47,55,65]
CLOZAPINE D22R/DR/D
BINDING
RECEPTOR3R selective
antagonist 3 R selective OBP OBP [47,65,67,68]
NEMONAPRIDE
NEMONAPRIDE D2R/D 3R selective
BINDING
RECEPTOR + SBP
OBP + SBP[38,47,55,65]
[38,47,55,65]
NEMONAPRIDE D2R/D antagonist
antagonist
3R selective
BINDING OBP + SBP [38,47,55,65]
NEMONAPRIDE D2R/D antagonist
BINDING
3 R selective OBP + SBP [38,47,55,65]
NEMONAPRIDE D2R/D antagonist
3R selective OBP + SBP [38,47,55,65]
NEMONAPRIDE D2R/D antagonist
3R selective OBP + SBP [38,47,55,65]
NEMONAPRIDE antagonist
D2R/D3R selective OBP + SBP [38,47,55,65]
NEMONAPRIDE antagonist
D“Dirty
2R/D 3R drug”,
selective OBP + SBP [38,47,55,65]
NEMONAPRIDE D antagonist
2“Dirty
R/D 3R drug”,
selective OBP + SBP [38,47,55,65]
SULPIRIDE
NEMONAPRIDE Dmultiple
antagonist
2“Dirty
R/D receptor
3R drug”,
selective OBP
OBP +++ SBP
SBP [47,65,66]
[38,47,55,65]
SULPIRIDE
NEMONAPRIDE multiple
“Dirty drug”,
antagonist
“Dirty receptor
drug”, multipleOBP
OBP + SBP
SBP [47,65,66]
[38,47,55,65]
SULPIRIDE
SULPIRIDE multiple binding
antagonist receptor OBP +OBPSBP + SBP[47,65,66][47,65,66]
“Dirtybinding drug”,
SULPIRIDE receptor
multiple
“Dirtybinding
binding
receptor
drug”, OBP + SBP [47,65,66]
SULPIRIDE multiple
“Dirty receptor
drug”, OBP + SBP [47,65,66]
SULPIRIDE multiple binding receptor OBP + SBP [47,65,66]
“Dirtybinding drug”,
SULPIRIDE multiple
“Dirty receptor
drug”, OBP + SBP [47,65,66]
SULPIRIDE multiple binding receptor OBP + SBP [47,65,66]
“Dirtybinding drug”,
SULPIRIDE multiple
“Dirtybinding receptor
drug”, OBP + SBP [47,65,66]
SULPIRIDE
SCH23390 multiple
D1Rbinding receptor
antagonist OBP
OBP + SBP [47,65,66]
[31,47,65,69]
SULPIRIDE
SCH23390 multiple
D1Rbinding receptor
antagonist OBP
OBP + SBP [47,65,66]
[31,47,65,69]
SCH23390 D1Rbinding
antagonist OBP [31,47,65,69]
SCH23390
SCH23390 DD 1R R
1 antagonist
antagonist OBP OBP [31,47,65,69] [31,47,65,69]
SCH23390 D1R antagonist OBP [31,47,65,69]
SCH23390 D1R antagonist OBP [31,47,65,69]
SCH23390 D1R antagonist OBP [31,47,65,69]
SCH23390 D1R antagonist OBP [31,47,65,69]
SCH23390
SKF38393 D1RDselective
1R antagonist agonist OBP
OBP [31,47,65,69]
[31,47,65,70]
SCH23390
SKF38393 D1RDselective
1R antagonist agonist OBP [31,47,65,69]
[31,47,65,70]
SCH23390
SKF38393 D1RDselective
1R antagonist agonist OBP [31,47,65,69]
[31,47,65,70]
SKF38393 D1R selective agonist OBP [31,47,65,70]
SKF38393
SKF38393 DD1R1 Rselective
selective agonist
agonist OBP OBP [31,47,65,70] [31,47,65,70]
SKF38393 D1 R selective agonist OBP [31,47,65,70]
SKF38393 D1R selective agonist OBP [31,47,65,70]
SKF38393 D R/D R
D1R selective agonist
2 3 selective OBP [31,47,65,70]
ETICLOPRIDE
SKF38393 DD1R2R/D 3R selective
selective agonist OBP + SBP
OBP [66,37]
[31,47,65,70]
ETICLOPRIDE
SKF38393 DD 1R2R/D
antagonist
3R selective
selective agonist OBP
OBP+ SBP [66,37]
[31,47,65,70]
ETICLOPRIDE
SKF38393 D antagonist
R/D
D1R selective
2 3 R selective
agonist OBP
OBP + SBP [66,37]
[31,47,65,70]
ETICLOPRIDE D2R/D antagonist
3R selective OBP + SBP [66,37]
ETICLOPRIDE D antagonist
R/D
D2 R/D R selective OBP + SBP [66,37]
ETICLOPRIDE
2 3
antagonist 3 R selective OBP + SBP
ETICLOPRIDE D2R/D 3R selective OBP + SBP [66,37] [37,66]
ETICLOPRIDE D2R/D antagonist
antagonist
3R selective OBP + SBP [66,37]
ETICLOPRIDE antagonist
D“Dirty
2R/D3R drug”, selective OBP + SBP [66,37]
ETICLOPRIDE antagonist
D2“Dirty
R/D3R drug”, selective OBP + SBP [66,37]
ETICLOPRIDE
RISPERIDONE D 2R/D antagonist
multiple 3R receptor
selective OBP + SBP
OBP+SBP [66,37]
[36,47]
ETICLOPRIDE “Dirty
antagonist drug”, OBP + SBP [66,37]
RISPERIDONE multiple
“Dirty receptor
drug”, OBP+SBP [36,47]
RISPERIDONE antagonist
multiple binding receptor OBP+SBP [36,47]
“Dirtybinding drug”,
RISPERIDONE multiple
“Dirty receptor
drug”, OBP+SBP [36,47]
RISPERIDONE multiple binding receptor OBP+SBP [36,47]
“Dirtybinding drug”,
RISPERIDONE multiple
“Dirty“Dirty drug”, receptor
drug”, multipleOBP+SBP [36,47]
RISPERIDONE
RISPERIDONE multiple binding receptor OBP+SBP OBP+SBP [36,47] [36,47]
RISPERIDONE “Dirty
Partial binding
receptor
multiple D 2R
drug”,
agonist,
binding OBP+SBP
receptor [36,47]
“Dirty
Partial binding
D drug”,
R agonist,
RISPERIDONE multiple
“Dirty
D
2
R/D
binding
2D receptor
drug”,
3R OBP+SBP [36,47]
RISPERIDONE
ARIPIPRAZOLE Partial
multiple 2R agonist,
receptor OBP+SBP
OBP + SBP [36,47]
[66,71]
RISPERIDONE Partial
multiple D2R/D
Dbinding 2R 3R
agonist,
receptor OBP+SBP [36,47]
ARIPIPRAZOLE heterodimer
D R/D
binding R OBP + SBP [66,71]
ARIPIPRAZOLE Partial DR/D
2
heterodimer
D 2R agonist,
3
3R OBP + SBP [66,71]
ARIPIPRAZOLE Partial binding
2D
antagonist
heterodimer 2R agonist,
OBP + SBP [66,71]
Partial D2DR/D
antagonist 3R
2R agonist,
ARIPIPRAZOLE heterodimer
D R/D R OBP + SBP [66,71]
ARIPIPRAZOLE Partial D2R agonist,
2
antagonist
heterodimer
3
Partial
Partial D2R/D
antagonist D RR agonist, OBP
2R23agonist,
+ SBP [66,71]
ARIPIPRAZOLE Partial
D D2D
heterodimer
2RD
R/D
antagonist 3R
2R agonist,
selective OBP + SBP [66,71]
ARIPIPRAZOLE
ARIPIPRAZOLE DPartial
2 R/Dheterodimer
D R
R/D heterodimer
3R OBP +OBPSBP + SBP [66,71] [66,71]
ARIPIPRAZOLE D D3
antagonist
R
heterodimer
2 2 R agonist,
2selective
OBP + SBP [66,71]
HALOPERIDOLE D
antagonist, 2 R/D
antagonist
Dheterodimer
2R
3
selectiveRD 4 R OBP+SBP [47,65,67,72,73]
ARIPIPRAZOLE antagonist OBP + SBP [66,71]
HALOPERIDOLE D 2D
antagonist, R/D3RD4R
antagonist
R2selective
heterodimer OBP+SBP [47,65,67,72,73]
ARIPIPRAZOLE
HALOPERIDOLE antagonist
antagonist, D R OBP +
OBP+SBP SBP [66,71]
[47,65,67,72,73]
D antagonist
2R selective
heterodimer
antagonist
4
HALOPERIDOLE antagonist,
Dantagonist
2R selective D4R OBP+SBP [47,65,67,72,73]
HALOPERIDOLE antagonist
antagonist,
antagonist D4R OBP+SBP [47,65,67,72,73]
Dantagonist
2R selective
HALOPERIDOLE antagonist,
Dantagonist
2R selective D 4 R OBP+SBP [47,65,67,72,73]
HALOPERIDOLE antagonist,
D D R R D4R
selective
selective OBP+SBP [47,65,67,72,73]
HALOPERIDOLE antagonist
2 2
antagonist, D4R OBP+SBP [47,65,67,72,73]
Dantagonist
2R selective
HALOPERIDOLE
HALOPERIDOLE antagonist,
antagonist,
Dantagonist
2R selective D4D R4R OBP+SBPOBP+SBP [47,65,67,72,73]
[47,65,67,72,73]
HALOPERIDOLE
SPIPERONE antagonist,
Affinity for all D4DR R OBP+SBP
OBP + SBP [47,65,67,72,73]
[47,65,66]
HALOPERIDOLE
SPIPERONE Affinityantagonist
antagonist
antagonist, for all D4DR R OBP+SBP
OBP + SBP [47,65,67,72,73]
[47,65,66]
SPIPERONE antagonist
Affinity for all DR OBP + SBP [47,65,66]
SPIPERONE antagonist
Affinity for all DR OBP + SBP [47,65,66]
SPIPERONE Affinity for all DR OBP + SBP [47,65,66]
SPIPERONE Affinity for all DR OBP + SBP [47,65,66]
SPIPERONE Affinity for all DR OBP + SBP [47,65,66]
SPIPERONE Affinity for all DR OBP + SBP [47,65,66]
SPIPERONE Affinity for all DR OBP + SBP [47,65,66]
SPIPERONE
SPIPERONE Antagonist
Affinity
Affinity forfor onall
all all DR OBP +OBP
DR SBP + SBP[47,65,66][47,65,66]
CHLORPROMAZINE
SPIPERONE Antagonist
Affinity for on DR
all all OBP
OBP + SBP [47,65,74]
[47,65,66]
CHLORPROMAZINE Antagonist DR on all OBP [47,65,74]
CHLORPROMAZINE Antagonist DR on all OBP [47,65,74]
CHLORPROMAZINE Antagonist DR on all OBP [47,65,74]
CHLORPROMAZINE Antagonist DR on all OBP [47,65,74]
CHLORPROMAZINE Antagonist DR on all OBP [47,65,74]
CHLORPROMAZINE
Abbreviations: DR-dopamine receptors, BP-binding pocket,
Antagonist DR
OBP-orthosteric
on all OBP
binding pocket, [47,65,74]
SBP-
CHLORPROMAZINE
Abbreviations: DR-dopamine receptors, BP-binding pocket,
Antagonist DR on all
OBP-orthosteric OBP
binding [47,65,74]
pocket, SBP-
CHLORPROMAZINE
secondary
Abbreviations:binding pocket.
DR-dopamine receptors,
BP-binding pocket,
Antagonist DR
OBP-orthosteric
on all OBP
binding [47,65,74]
pocket, SBP-
CHLORPROMAZINE
Abbreviations:
secondary binding DR-dopamine
pocket. receptors,
BP-binding pocket,
Antagonist DR
OBP-orthosteric
on all OBP
binding [47,65,74]
pocket, SBP-
CHLORPROMAZINE
CHLORPROMAZINE
Abbreviations:
secondary binding DR-dopamine
pocket. receptors, Antagonist
BP-binding pocket, DR on all DR OBP
OBP-orthosteric bindingOBP [47,65,74]
pocket, SBP- [47,65,74]
secondary binding
Abbreviations: pocket.
DR-dopamine receptors, BP-binding pocket, DR OBP-orthosteric binding pocket, SBP-
Abbreviations:
secondary DR-dopamine
binding pocket. receptors, BP-binding pocket, OBP-orthosteric binding pocket, SBP-
Abbreviations: DR-dopamine
secondary binding pocket. receptors, BP-binding pocket, OBP-orthosteric binding pocket, SBP-
Abbreviations:
secondary DR-dopamine
binding pocket. receptors, BP-binding pocket, OBP-orthosteric binding pocket, SBP-
secondary binding
Abbreviations: pocket.
DR-dopamine receptors, BP-binding pocket, OBP-orthosteric binding pocket, SBP-
Abbreviations: DR-dopamine
secondary binding
Abbreviations: pocket.
DR-dopamine receptors, BP-binding
receptors, BP-binding pocket,
pocket, OBP-orthosteric
OBP-orthosteric binding
bindingpocket,
pocket,SBP-
SBP-secondary
secondary binding pocket.
secondary
binding binding pocket.
pocket.
Molecules 2019, 24, 1196 9 of 26

Molecules 2019, 24, x FOR PEER REVIEW 9 of 26

For 7-OH-DPAT we observed a low and stable binding energy upon binding to all DRs.
For 7-OH-DPAT we observed a low and stable binding energy upon binding to all DRs. For
For apomorphine, a decrease in the binding energy was determined for D2 R at 65 ns (−11 kcal/mol),
apomorphine, a decrease in the binding energy was determined for D2R at 65 ns (−11 kcal/mol),
whereas an increase at 85 ns was shown for D R (−9 kcal/mol). Stable binding energies around
whereas an increase at 85 ns was shown for D4R4 (−9 kcal/mol). Stable binding energies around −10
−10 kcal/mol were observed for DR:nemonapride complexes, however a massive increase was
kcal/mol were observed for DR:nemonapride complexes, however a massive increase was observed
observed for the D5 R at 100 ns. For SCH23390, but not for SKF38393 the binding energy was stable
for the D5R at 100 ns. For SCH23390, but not for SKF38393 the binding energy was stable over time at
over time at −9 kcal/mol for all DRs. The binding energy of SKF38393 at D R and D4 R increased
−9 kcal/mol for all DRs. The binding energy of SKF38393 at D2R and D4R2 increased at 85 ns.
at 85 ns. Haloperidole displayed the most interesting docking-profile: while the binding energies of
Haloperidole displayed the most interesting docking-profile: while the binding energies of DRs were
DRs were stable at −10 kcal/mol, only for D4 R a massive increase was observed at 55 ns and 80–90 ns
stable at −10 kcal/mol, only for D4R a massive increase was observed at 55 ns and 80–90 ns into the
into the positive range, meaning these binding positions were extremely unfavorable for haloperidole.
positive range, meaning these binding positions were extremely unfavorable for haloperidole. Lastly,
Lastly, the chlorpromazine binding energy was increased only for D R at 70 ns up to −3 kcal/mol.
the chlorpromazine binding energy was increased only for D1R at 701 ns up to −3 kcal/mol.
Similar to dopamine binding, the NoC of 7-OH-DPAT decreased at all DRs from 0 to 65 ns.
Similar to dopamine binding, the NoC of 7-OH-DPAT decreased at all DRs from 0 to 65 ns. For
For apomorphine, the lowest binding energies were obtained for D1 R and D2 R. Lesser NoC were
apomorphine, the lowest binding energies were obtained for D1R and D2R. Lesser NoC were counted
counted for nemonapride in total at all DRs (max. 30 at 85 ns for D2 R). The NoC for SKF38393 were
for nemonapride in total at all DRs (max. 30 at 85 ns for D2R). The NoC for SKF38393 were the lowest
the lowest over 70–85 ns period for D R, D R and D3 R. In contrast to the BE of haloperidole, the NoC
over 70–85 ns period for D1R, D2R and1 D3R.2In contrast to the BE of haloperidole, the NoC was found
was found to be stable over time except for D1 R with up to 40 conformations at 60 ns. In addition,
to be stable over time except for D1R with up to 40 conformations at 60 ns. In addition, most
most conformations were counted for the D4 R especially at 0–70 ns for haloperidole.
conformations were counted for the D4R especially at 0–70 ns for haloperidole.
7-OH-DPAT 7-OH-DPAT
-12 50
Number of conformations
Binding energy [kcal/mol]

D1R D1R
D2R 40 D2R
-10 D3R D3R
30
D4R D4R
D5R 20 D5R
-8
10

-6 0
0 55 60 65 70 75 80 85 90 95 100 0 55 60 65 70 75 80 85 90 95 100
Time points [ns] Time points [ns]

Apomorphine Apomorphine
-12 50
Number of conformations
Binding energy [kcal/mol]

D1R D1R
D2R 40 D2R
-10 D3R D3R
30
D4R D4R
D5R 20 D5R
-8
10

-6 0
0 55 60 65 70 75 80 85 90 95 100 0 55 60 65 70 75 80 85 90 95 100
Time points [ns] Time points [ns]

Nemonapride Nemonapride
-12 50
Number of conformations
Binding energy [kcal/mol]

D1R D1R
D2R 40 D2R
-10
D3R D3R
30
-8 D4R D4R
D5R 20 D5R
-6
10

-4 0
0 55 60 65 70 75 80 85 90 95 100 0 55 60 65 70 75 80 85 90 95 100
Time points [ns] Time points [ns]

SCH23390 SCH23390
-12 50
Number of conformations
Binding energy [kcal/mol]

D1R D1R
-10 D2R 40 D2R
D3R D3R
-8 30
D4R D4R
-6 D5R 20 D5R

-4 10

-2 0
0 55 60 65 70 75 80 85 90 95 100 0 55 60 65 70 75 80 85 90 95 100
Time points [ns] Time points [ns]

Figure 4. Cont.
Molecules 2019, 24, x FOR PEER REVIEW 10 of 26

Molecules 2019, 24, 1196 10 of 26

Figure 4. Cont.

SKF38393 SKF38393
-12 50

Binding energy [kcal/mol]

Number of conformations
D1R D1R
D2R 40 D2R
-10 D3R D3R
30
D4R D4R
D5R 20 D5R
-8
10

-6 0
0 55 60 65 70 75 80 85 90 95 100 0 55 60 65 70 75 80 85 90 95 100
Time points [ns] Time points [ns]

Haloperidol Haloperidol
50

Number of conformations
-12 D1R
Binding energy [kcal/mol]

D1R
-10
40 D2R
-8 D2R
-6 D3R
-4
D3R 30
-2 D4R D4R
0 20 D5R
2 0 55 60 65 70 75 80 85 90 95 100 D5R
4 10
6
8
10 Time points [ns] 0
0 55 60 65 70 75 80 85 90 95 100
Time points [ns]
Chlorpromazine Chlorpromazine
-12 50
Number of conformations
Binding energy [kcal/mol]

D1R D1R
-10 D2R 40 D2R
D3R D3R
-8 30
D4R D4R
-6 D5R 20 D5R

-4 10

-2 0
0 55 60 65 70 75 80 85 90 95 100 0 55 60 65 70 75 80 85 90 95 100
Time points [ns] Time points [ns]

Figure 4. Results of the molecular docking of 7-OH-DPAT, apomorphine, nemonapride, SCH23390,

Figure 4. Results
SKF38393, of the and
haloperidole molecular docking of
chlorpromazine for7-OH-DPAT, apomorphine,
all DR subtypes nemonapride,
at time points SCH23390,
[ns]. The average of the
SKF38393,
three lowesthaloperidole and chlorpromazine
binding energies of dopamine were forcalculated
all DR subtypes at plots.
in the left time points [ns].ofThe
The mean theaverage
number of
of
the three lowest binding energies of dopamine were calculated in the left plots. The mean
conformations of the three clusters with the lowest binding energies were plotted for each time point of the
number of conformations
and receptor (right plot). of the three clusters with the lowest binding energies were plotted for each
time point and receptor (right plot).
We also calculated the distance between the center of mass of the ligand and the alpha carbon of
We alsopocket
the binding calculated the distance
residues. Overall between
results the center
of all of mass of the
ligand-residue ligand and the
measurements alpha5)carbon
(Figure showed of
the
thatbinding
3.32Asppocket was the residues.
closest Overall
residue forresults of all ligand-residue
the majority of ligands. The measurements
ligand center (Figure 5) showed
of mass-residue
that
alpha 3.32Asp
carbonwas the closest
distance residue
was lower thanfor7–8theÅ,majority of ligands.
particularly for D1The
R (<6 ligand
Å). We center
noted of an
mass-residue
increase in
alpha carbonbetween
the distance distance 3.32Asp
was lower and than 7–8 Å,
several particularly
ligands for D4 R. forThe
D1Rdistances
(< 6 Å). We noted 3.32Asp
between an increase andinboth
the
distance between 3.32Asp and several ligands for D R. The distances
SKF38393 and SCH23390 ligands were larger at for D3 R, D4 R and D5 R, but also D2 R. This effect might
4 between 3.32Asp and both
SKF38393
occur dueand to theSCH23390
fact, thatligands
SCH23390 were andlarger at for Dare
SKF38393 3R,reported
D4R and to D5be
R, but also D2R. This
D1 R-selective effect
[30,63]. might
Subtype
occur
specificdue to the fact,
tendencies thatobserved
were SCH23390 forand
the SKF38393 are reported
serine microdomain. to be D
5.42Ser 1R-selective
was shown to[30,63].
be mostSubtypedistant
specific
at D1 -liketendencies
receptorswere observed
and 5.43Ser forfor
D2the
R and serine
D3 Rmicrodomain.
(D2 -like). These 5.42Ser was shown
differences to be
are less most distant
accentuated for
at D1-like receptors
dopamine, 7-OH-DPAT, and 5.43Ser for D2Rand
apomorphine andbromocriptine.
D3R (D2-like). These differences are less accentuated for
dopamine, 7-OH-DPAT,
For 7-OH-DPAT, apomorphine
a known and bromocriptine.
D3 R selective agonist, distances between ligand and the defined pocket
For 7-OH-DPAT,
are higher a knownand
for D1 -like receptors D3Rdistinctive
selective agonist, distancesDbetween
residue between 2 -like seems ligand
to beand the defined
6.52Phe, that is
pocket
closer to are
thehigher
ligandfor onDD1-like
3 R. receptors
The same and
pattern distinctive
was visibleresidue
with between
apomorphine, D 2-likea seems
selective toD be2 R6.52Phe,
agonist,
that
where is closer
distances to the
in Dligand
1 -like on
areDhigher,
3R. Thealthough
same pattern was visible
distinction withinwith apomorphine,
D2 -like family is less a selective
pronounced. D2R
agonist,
Clozapine, where distances
sulpiride and in D1-like areare
risperidone higher,
known although
as “dirty distinction within of
drugs” because D2-like
their family is less
non-selective
pronounced.
profile, and for Clozapine,
that reason sulpiride
none ofand theserisperidone
ligands showedare known as “dirty
distinctive drugs” between
differences because of DRtheir non-
subtypes.
selective profile, and for that reason none of these ligands showed distinctive
Likewise, residues 3.32Asp and 3.33Val/Ile were the closest to clozapine in all five subtypes, suggesting differences between
DR
thatsubtypes.
these residuesLikewise, residues
are crucial for 3.32Asp
this ligand’sand binding.
3.33Val/Ile were the closest
Haloperidole, to clozapine
categorized as D2 Rinselective
all five
subtypes,
antagonist suggesting
with some activitythat theseon D4residues are crucial
R, has distinctive for thisbetween
differences ligand’sD1binding.
-like and D Haloperidole,
2 -like family,
categorized
being closeras toDthe
2R selective antagonistwithin
second (although with some activity
D2 -like family onthere
D4R, ishasnodistinctive differences
great differences between
on distances
D 1-like and
pattern). D2-like family,
Spiperone being closer tohave
and chlorpromazine the second
affinity(although
for all DRwithin
subtypes,D2-like
which family there
agrees withis nothegreat
lack
differences on distances pattern). Spiperone and chlorpromazine have
of significant differences in the measured distances. Finally, nemonapride and eticlopride, described affinity for all DR subtypes,
 Molecules 2019, 24, x FOR PEER REVIEW 11 of 26
Molecules 2019, 24, 1196 11 of 26
which agrees with the lack of significant differences in the measured distances. Finally, nemonapride
and eticlopride, described as D2R/D3R selective antagonists, were located closest to D2-like DR
as D2 R/D3 R selective antagonists, were located closest to D2 -like DR residues compared to the D1 -like
residues compared to the D1-like DR, however it seemed as these two ligands demonstrated
DR, however it seemed as these two ligands demonstrated preference for D4 R.
preference for D4R.

D1R Å D2R Å
2.57Val 4 2.57Val 4
3.32Asp 5 3.32Asp 5
3.33Val/Ile 3.33Val/Ile
6 6
3.36Cys 3.36Cys
7 7
5.42Ser 5.42Ser
5.43Ser 8 5.43Ser 8
5.46Ser 9 5.46Ser 9
6.48Trp 10 6.48Trp 10
6.51Phe 6.51Phe 11
11
6.52Phe 6.52Phe
12 12
6.55His/Asn 6.55His/Asn
7.43Tyr/Trp blank 7.43Tyr/Trp blank

o m ro l
ro o P e
oc h T
N Cl rip ine
on ap ne
Suapr ne
SC lp ide
SK 23 de
R ticl 83 0
A pe pri 3
H ip done
al ra e
or pi ri e

az ne
e
ro o P e

N Cl rip ine
on ap ne
Suapr ne
SC lp ide
SK 23 de
R ticl 383 0
A pe pri 3
H ip done
al ra e
or pi ri e

az ne
e
om ro l
oc h T

pr pe do
pr pe do

m rp A
B m -D n

E F3 39
is o 9
rip ri d

in
hl S pe zol
m rp A
B m -D n

E F 39
is o 9
rip ri d

in
hl S pe zol

po H i

em o ti
i

iri
po H i

em oz ti
i

iri

A 7-O pam
A 7-O am

H
H
op

o
o

o
D
D

C
C

D3R Å D4R Å
2.57Val 4 2.57Val 4
3.32Asp 5 3.32Asp 5
3.33Val/Ile 3.33Val/Ile
6 6
3.36Cys 3.36Cys
7 7
5.42Ser 5.42Ser
5.43Ser 8 5.43Ser 8
5.46Ser 9 5.46Ser 9
6.48Trp 10 6.48Trp 10
6.51Phe 11 6.51Phe
11
6.52Phe 6.52Phe
12 6.55His/Asn 12
6.55His/Asn
7.43Tyr/Trp blank 7.43Tyr/Trp blank
N Cl rip ine

om ro l
ro o P e
oc h T

on ap ne
Suapr ne
SC lp ide
SK 23 de
R ticl 83 0
A pe pri 3
H ip done
a l ra e
or pi ri e

az ne
e
ro o P e

N Cl rip ine
on ap ne
Suapr ne
SC lp ide
SK 23 de
R ticl 383 0
A pe pri 3
H ip done
al ra e
or pi ri e

az ne
e
om ro l
oc h T

pr pe do
pr pe do

m rp A

E F3 39
is o 9

hl S pe zol
m rp A

B m -D n

rip ri d

in
B m -D n

rip ri d

in
hl S pe zol
E F 39
is o 9

po H i

em oz ti
i

i ri
po H i

em oz ti
i

iri

A 7-O pam
A 7-O pam

H
H

o
o

o
o

D
D

C
C

D5R Å
2.57Val 4
3.32Asp 5
3.33Val/Ile
6
3.36Cys
7
5.42Ser
5.43Ser 8
5.46Ser 9
6.48Trp 10
6.51Phe 11
6.52Phe
12
6.55His/Asn
7.43Tyr/Trp blank
ro o P e

N Cl rip ine
on ap ne

al ra e
Suapr ne
SC lp ide
SK 23 de
R ticl 383 0
A pe pri 3
H ip done

or pi ri e

az ne
e
om ro l
oc h T

pr pe do
m rp A
B m -D n

r i p ri d

in
hl S pe zol
E F 39
is o 9
po H i

em o ti
i

iri
A 7-O am

H
op

o
D

Figure 5. Summary of the distances between ligands and residues used in molecular docking for
allFigure 5. Summary of the distances between ligands and residues used in molecular docking for all
DR subtypes. For each ligand-residue-distance [Å], we calculated the mean of all time points
DR subtypes. For each ligand-residue-distance [Å], we calculated the mean of all time points of the
of the conformational models (11) of the three best docked clusters ranked by binding energy
conformational models (11) of the three best docked clusters ranked by binding energy [kcal/mol]
[kcal/mol] Noteworthy is that not all ligands were set to interact with all residues shown in the
Noteworthy is that not all ligands were set to interact with all residues shown in the x-axis in the
x-axis in the molecular docking. (e.g., only clozapine and aripiprazole were set to interact with
molecular docking. (e.g. only clozapine and aripiprazole were set to interact with 3.33Val). The
3.33Val). The distances are color coded: while dark colors indicate short distances, light colors indicate
distances are color coded: while dark colors indicate short distances, light colors indicate wider
wider distances.
distances.
2.5. The Type of Pairwise Interactions Between Receptor Amino-Acids and Ligand is Relevant for Binding
2.5. The Type of Pairwise Interactions Between Receptor Amino-Acids and Ligand is Relevant for Binding.
In-house scripts using the BINANA algorithm (software used in other non-GPCR
In-house scripts using the BINANA algorithm (software used in other non-GPCR studies [39,75–
studies [39,75–78]) were constructed to identify the type of interactions established between the ligands
78]) were constructed to identify the type of interactions established between the ligands and binding
and binding pocket amino-acids [39]. We measured close contacts between receptor and ligands below
or equal 2.5 Å and below or equal 4.0 Å, hydrogen bonds (HB), hydrophobic contacts (hydrocontacts)
Molecules 2019, 24, 1196 12 of 26

and salt-bridges (SB) as well as π-interactions, further subdivided into cation-π-interactions (cat-π),
aromatic superpositions (π-π-stack) and perpendicular interactions of aromatic rings also referred
to as edge-face-interactions (T-stack) [39]. For a first overview, all interactions despite their type
and ligand were summarized and compared between the DR-subtypes (Results section at SI and
Figure S8). Moreover, detailed mapping of pairwise interactions for each receptor is displayed in
Figure 6. Figures S9–S23 show the change of interaction pattern over time for each ligand. Furthermore,
the pairwise analysis highlighted the role of key receptor residues. By assorting those for each ligand
at all DRs (time points summarized), patterns but also unique receptor-ligand interactions were
highlighted (Tables S10–S14).

2.6. 2.5 Å-Interactions

2.5 Å-interactions, very short (closer) contacts are especially relevant for ligand binding and are
described in more detail herein. For dopamine the number of these interactions increased for D1 -like
DRs, while for 7-OH-DPAT the highest number of interactions observed in total only occurred for D3 R.
For bromocriptine, 2.5 Å-interactions were significantly higher for D4 R. Also, haloperidole seemed to
have a higher number of established interactions with D4 R as well as eticlopride. Only risperidone had
a higher number of interactions with D2 R. Chlorpromazine had the lowest number of compared to all
ligands with no preference for any DR-subtype. All in all, 2.5 Å-interactions seemed to be particularly
relevant for the ligand binding to D4 R.

2.7. Hydrogen Bonds and Hydrophobic Contacts

Charge-reinforced hydrogen bonds are reported to be much stronger than the neutral hydrophobic
contacts [79]. Moreover, it was reported that HBs determine the specificity of receptor-ligand
binding [79]. Nevertheless, hydrophobic contacts also contribute to ligand binding, and a balance
between HB and hydro contacts is required for drug-like molecules [79]. It was not surprising
that a large number of hydro contacts was observed for all ligands, while HB were less common.
Hydro contacts were preferably formed for D1 R and achieved their lowest value for D3 R. These contacts
were particularly relevant for one ligand, bromocriptine (Figure 6). Moreover, a large hydrophobic
network involving conserved and non-conserved residues of all TMs were found for all DRs (less
pronounced for D5 R). The “dirty drugs” were the second in line with the highest number of
hydro contacts.
Most interesting were the HB interactions. For dopamine a different set-up was presented at each
DR. While the D2 -like DRs and D5 R HB were formed by the serine microdomain (5.42Ser, 5.43Ser and
5.46Ser), for D1 R the serine microdomain was not involved at all. 3.32Asp appeared as interaction
partner for all DRs. For D5 R, an HB between 5.38Tyr and dopamine was stressed out as unique for all
ligands. However, 5.38Tyr was found at the D4 R to form HB with 7-OH-DPAT. Not more than 2 HB
were found at any DR bounded to 7-OH-DPAT. Lastly, chlorpromazine does not seem to form any HB
in any DR complex.

2.8. Salt-Bridges
Most stable SB interactions were unsurprisingly achieved by dopamine for all DR sub-types.
7-OH-DPAT, SB were found for D1 R (three in total), while for the other subtypes, contacts ranged
between one and three over time. The same trend was observed for nemonapride and SKF38393.
SCH23390 formed the highest number of SB with D5 R and with D2 R between 70 and 85 ns.
Haloperidole seemed to establish a higher number of SB with D1 -like DR and D2 R, while none were
formed with D3 R and D4 R. Spiperone seemed to preferably form SB with D1 -like DRs. The following
ligands did not form any salt-bridges at any time point: apomorphine, bromocriptine, clozapine,
risperidone, aripiprazole and chlorpromazine.
Undoubtedly, 3.32Asp was always involved in the establishment of SB in all DRs. However,
at D1 R, 74Pro located on ECL1 appeared also to establish relevant SB interactions. In addition,
Molecules 2019, 24, 1196 13 of 26

D3 R SB-bonding for spiperone was found to occur involving 1.44Leu and 75Ser (ECL1) rather than
3.32Asp. All in all, salt-bridges were found to be highly conserved regarding the residues involved.

2.9. Cat-π- and π-π-Stacking Interactions

Cationic-π and π-π-stacking are considered as natural key non-covalent interactions [80]. They are
important as solitary effects, but also their interplay omnipresent in many biological systems [81].
In the DR-ligand system frequent oscillations between different receptor conformations were noted for
some ligands, depicted in Figures S9–S13.
Dopamine, for example, showed the highest cat-π-interactions for D2 R, oscillating from 2–4
interactions/time point. Cation-π-interactions seemed to be more relevant for D4 R and were less
common and mainly formed by conserved residues on TM6 (6.42Gly, 6.31Thr, 6.30Glu, 6.39Val) for
D1 R. Bromocriptine (3.28Trp, 6.51Phe), nemonapride (6.48Trp, 6.51Phe, 6.52Phe), sulpiride (2.61Lys,
6.48Trp, 6.51Phe) and SCH23390 (6.48Trp) showed one cat-π-contacts to D5 R each. For risperidone,
cat-π-interactions were mainly formed with D4 R, while π-π-stacking was mostly related to D3 R
complexes. Aripiprazole seemed to preferably form cat-π-interactions with D4 R, while increasing
π-π-stacking-interactions were observed with D1 R between 65 and 80 ns. Haloperidole seemed to
prefer π-π-stacking-interactions with D2 R, maybe important for its selectivity towards this receptor.
For chlorpromazine, no cat-π-interactions were observed at D1 -like DRs (D1 R and D5 R), while many
interactions were counted with D2 R between 65 and 75 ns, with D3 R at 95 ns and with D4 R at 60 ns.
The π-π-interactions were rather rare compared to the other interaction types. Some ligands did
not form π-π-stacking interactions with DR subtypes (e.g., D1 R binding to dopamine, 7-OH-PAT and
sulpiride; D2 R binding to sulpiride either, D4 R binding to eticlopride and haloperidole; D5 R binding
to nemonapride). It was also obvious that the residues of the aromatic microdomain (6.48Trp, 6.51Phe,
6.52Phe, 6.55His) were responsible for the majority of ligands interactions to all DRs. However,
different residue partners were determined for π-π- compared to T-stacking such as residues from
TM5 (5.38Tyr, 5.47Phe). For aripiprazole, residues 7.43Tyr (D2 R-D4 R) and 7.34Thr (D1 R) seemed also
to be important for this type of interaction. Most interesting was the interaction pattern for sulpiride:
while for D1 R and D2 R no π-π-stacking was detected, for D3 R and D5 R only a few residues seemed to
be relevant (2.43Val, 2.44Val, 2.48Val, 38Thr, 5.38Phe, 6.51Phe, 6.52Phe for D3 R; 3.28Trp and 6.48Trp for
D5 R) while for D4 R, 27 residues from all TMs were involved in contact network formation. This may
be explained by the different possible binding poses of sulpiride on the different D4 R conformations.

2.10. T-Stacking Interactions

T-stacking-interactions were similar to cat-π- and π-π-interactions, yet more frequent fluctuations
in the number of interacts between ligands and receptor were observed in total. Especially
for risperidone, which showed the highest number of T-stacking-contacts, preferably with D2 R.
Haloperidole and spiperone also seemed to have a D2 R-preference, while chlorpromazine formed
a large number of interactions with D5 R. Despite the ligand, T-stack-contacts involved mainly
conserved residues (6.39Val, 6.42Gly, 6.43Val) or residues from the aromatic microdomain (6.48Trp,
6.51Phe, 6.52Phe, 6.55His). An exception was bromocriptine and sulpiride for D2 R, haloperidole for
D4 R and spiperone for D5 R. Unique interactions were found for risperidone binding to D4 R with
6.44Phe and for chlorpromazine binding to D1 R with 6.30Glu. However, other residues from other
TMs were also involved in forming T-stack-contacts: for example, 7-OH-DPAT unique interaction with
2.47Ala and SKF38393 with 35Ala (ICL1) were found at D3 R. For risperidone, a unique interaction
with 231Phe (ICL3) was determined for D1 R. Whereas for spiperone 1.35Tyr and 159Ile (ECL2) seemed
to be relevant for D4 R, 2.14Tyr was relevant for chlorpromazine coupling.
However, TM7 residues were also involved in T-stack-formation: 7.34Thr (D1 R) and 7.35Tyr
(D2 -like)/7.35Phe(D5 R), 7.43Tyr(D2 -like). Residues on TM2 were also relevant for T-stack-formation
(2.41Tyr, 2.43Val, 2.45Ser, 2.46Leu, 2.47Ala, 2.50Asp) but only for D3 R. For D4 R and D5 R, only residues
from TM6 and TM7 were involved in T-stack-contacts, except for SKF38393 where 5.47Phe was
Molecules 2019, 24, 1196 14 of 26
Molecules 2019, 24, x FOR PEER REVIEW 14 of 26

established meaningful interactions with nemonapride, sulpiride, SCH23390, aripiprazole and

relevant for binding to D R. Lastly, for D1 R and D2 R TM3 (3.28Trp(D1 R)/3.28Phe(D2 R)) residues
spiperone. Although these4 residues (especially on TM2 and TM7) are more related to the SBP than to
also established meaningful interactions with nemonapride, sulpiride, SCH23390, aripiprazole and
the OBP (herein TM6 is the most relevant TM), contact formation was also observed for smaller
spiperone. Although these residues (especially on TM2 and TM7) are more related to the SBP than
ligands (7OH-DPAT, SCH23390, SKF38393). It was not expected that these ligands would access the
to the OBP (herein TM6 is the most relevant TM), contact formation was also observed for smaller
SBP. Noteworthy is also the fact, that dopamine exclusively formed T-stack-contacts with the
ligands (7OH-DPAT, SCH23390, SKF38393). It was not expected that these ligands would access the
conserved aromatic microdomain for all DR. Finally, it was also obvious that the variety of T-stack-
SBP. Noteworthy is also the fact, that dopamine exclusively formed T-stack-contacts with the conserved
contacts was also limited by the number of aromatic rings of the ligand (e.g. dopamine only contacted
aromatic microdomain for all DR. Finally, it was also obvious that the variety of T-stack-contacts was
3 different sequential residues). In brief, our results also pinpoint for the fact that T-stacking-
also limited by the number of aromatic rings of the ligand (e.g., dopamine only contacted 3 different
interactions seem to be relevant for large ligands, primary in antagonists binding than in agonists
sequential residues). In brief, our results also pinpoint for the fact that T-stacking-interactions seem to
case.
be relevant for large ligands, primary in antagonists binding than in agonists case.

D1R D2R
2.5 Å high 2.5 Å high
> 100 > 100
4Å 75 - 100 4Å 75 - 100
Cat-π 50 - 75 Cat-π 50 - 75
25 - 50 25 - 50
HB 10 - 25 HB 10 - 25
Hydrocontacts 8 - 10 Hydrocontacts 8 - 10
6-8 6-8
π-π-stack 4-6 π-π-stack 4-6
SB 2-4 SB 2-4
1-2 1-2
T-stack 0-1 T-stack 0-1
ro o P e

N Cl rip ine
on ap ne
Su pr e
SC lp ide
SKH2 rde
R ticl 383 0
A pe pri 3
H ipi ridode
o a e
or pi id e
om ro e
az ne
e
oc h T

oc h T
ro o P e

N Cl rip ine
on ap ne
Su pr e
SC lp ide
SK 2 rde
R cl 3 0
A pe pri 3
H ipi ridode
o a e
or pi id e
om ro e
az ne
e
m rp A

hl S per zol
pr pe ol
E F 39
is o 9

in
B m -D n

a in

al pr n

m rp A

E t F3 3 9
is o 9
B m -D n

a in

al pr n
hl S per zol
pr pe ol

in
po H i

em o ti

po H i

em o t i
A 7-O pam

A 7 -O p a m
i

i
3

3
8
z

H
o

i
D

D
r

r
C

C
D3R D4R
2.5 Å high 2.5 Å high
> 100 > 100
4Å 75 - 100 4Å 75 - 100
Cat-π 50 - 75 Cat-π 50 - 75
25 - 50 25 - 50
HB 10 - 25 HB 10 - 25
Hydrocontacts 8 - 10 Hydrocontacts 8 - 10
6-8 6-8
π-π-stack 4-6 π-π-stack 4-6
SB 2-4 SB 2-4
1-2 1-2
T-stack 0-1 T-stack 0-1
ro o P e

N Cl rip ine

SC lp ide

R ticl 383 0
A pe pri 3

o a e

o m ro e
on ap ne
Su pr e

SK 2 rde

H ipi ridode

or pi id e

az ne
e
oc h T
ro o P e

N Cl rip ine
on ap ne
Su pr e
SC lp ide
SKH2 rde
R ticl 383 0
A pe pri 3
H ipi ridode
o a e
or pi id e
om ro e
az ne
e
oc h T

B m -D n

a in

al p r n

in
m rp A

E F 39
is o 9

hl S per zol
pr pe ol
m rp A

hl S per zol
p r pe o l
E F 39
is o 9
B m -D n

a in

al pr n

in

po H i

em o ti
po H i

e m o ti

A 7-O pam
A 7-O pam

i
i

3
3

z
z

H
o
o

D
D

r
r

C
C

D5R
2.5 Å high
> 100
4Å 75 - 100
Cat-π 50 - 75
25 - 50
HB 10 - 25
Hydrocontacts 8 - 10
6-8
π-π-stack 4-6
SB 2-4
1-2
T-stack 0-1
ro o P e

N Cl rip ine
on ap n e
Su pr e
SC lp ide
SKH2 rde
R ticl 383 0
A p e p ri 3
H ipi ridode
o a e
or pi id e
o m ro e
az ne
e
oc h T
m rp A

pr pe ol
E F 39
B m -D n

a in

is o 9

al pr n

in
hl S per zol
po H i

em o ti
A 7-O pam

i
3
z
o
D

Figure 6. Interaction types counted for each ligand at DR-subtypes. Data are summarized for each
Figure 6. all
ligand at Interaction types
time points. counted
Total for of
numbers each
theligand at for
contacts DR-subtypes. Data type
each interaction are summarized for each
are color-coded: few
ligand at all were
interactions time points.
coloredTotal numbers
white, of theinteractions
while many contacts forwere
each colored
interaction type
dark. arecells
Grey color-coded: few
indicate that
these values were
interactions are outside
coloredthe scale,while
white, whichmany
was only the case were
interactions for bromocriptine
colored dark.atGrey
the D 4 R with
cells 360 four
indicate that
Å-interactions.
these values are outside the scale, which was only the case for bromocriptine at the D4R with 360 four
Å-interactions.

3. Discussion

One of the major research efforts in the research of dopamine receptors is the design of DR-
subtype selective ligands [82]. However, most predictive studies have been performed on D2R ligand
Molecules 2019, 24, 1196 15 of 26

3. Discussion
One of the major research efforts in the research of dopamine receptors is the design of DR-subtype
selective ligands [82]. However, most predictive studies have been performed on D2 R ligand specificity,
as this receptor is the most crucial in neurotransmission [17,57,83]. Herein, we present a comprehensive
in silico approach, which reveals important interactions between DRs key residues and ligands in
a more detailed way when compared with available literature [55,57,59,60,84,85].

3.1. Validation of the In Silico Pipeline

Homology modeling and TM definition of all DR subtypes showed that there were smaller
structural differences among the “classical” TMs (TM3, TM5, TM6), which are known to be key for
ligand binding. Yet, as expected, structural differences between the subtypes were observed in the
intracellular and extracellular loops, where some are important for ligand binding (ECL2) or for
intracellular signaling (ICL2 and ICL3) [86]. This was particularly true for D1 -like receptors, due to
their much larger ICL3. Although no crystal structure was available for the D1 -like DRs, the high
sequence similarity among all DR helped to find suitable models for molecular docking. Validating
the docking performance by low binding energies and high NoC by cluster also showed that the
homology modeling-docking approach was suitable and reproducible. In fact, the combination of the
different software and in-house scripting resulted in a straightforward in silico approach which can
certainly be applied for studying other GPCRs. Data is also in line with experimental information,
which corroborates the conceptual framework of this analysis protocol [47].

3.2. Pairwise Interactions Analysis Was Able to Determine Key Amino-Acids and Types of Interaction
A clear D2 -like selectivity or binding preference was only found for apomorphine, while for
others either D2 R and D5 R seemed to form a lower number of 4 Å-interactions such as nemonapride
(D2 R/D3 R-antagonist [87]), SCH23390 (D1 -like antagonist [88]), SKF38393 (D1 R-antagonist [30]) or
D1 R and D4 R were highly preferred (higher number of meaningful interactions). In other cases, such as
for eticlopride (D2 R/D3 R antagonist [37]) and spiperone (D2 R-antagonist [64]), the D3 R was the least
attractive DR for interaction. It was shown that the “classical” conserved residues e.g., 3.32Asp,
the serine microdomain 5.42Ser, 5.45Ser, 5.46Ser and the aromatic microdomain 6.48Trp, 6.51Phe,
6.52Phe, 6.55His were relevant for all ligands and formed specific interactions, electrostatic (cat-π, π-π,
T-stack), salt-bridges and hydrogen bonds. These residues were omnipresent in all our analyses. Yet,
the distances for the most conserved OBP residues (3.32Asp, serine residues and 6.48Trp), distinct
differences were observed between agonists and antagonists. For example, dopamine was constantly
close to OBP, indicating its receptor activating properties as described by Floresca and Schnetz [47],
while risperidone was found distant from these residues according to its antagonistic properties.
This was also the case for the other antagonists such as haloperidole, nemonapride and the biased
ligand aripiprazole. In addition, we described other TM residues involved in binding of these ligands,
as previously described by Kalani et al. for D2 R [57].
It was not surprising that the “classical” TMs, e.g., TM3, TM5, TM6 and TM7 were involved
in many different interaction types. TM3 residues such as 3.35Cys, 3.36Ser, 3.33Val or 3.33Ile and
3.39Ser were often found forming different interactions with different ligands. This was also in
concordance with previous studies regarding the involvement of other conserved residues on TM2 and
TM7 (and TM3) [57,82,83,85], which was also described as part of a SBP only assessable for ligands
with piperazine-moieties [59]. Residues on TM4 were not contributing to receptor-ligand interaction,
except for D4 R complexes. By comparing large ligands such as spiperone or haloperidole with rather
compact ligands such as dopamine, SCH23390 or clozapine, it was possible to point out a larger
number of TM1 and TM2 residues involved in establishing meaningful interactions. Author’s had
already hypothesized that these residues could belong to a SBP, only accessed by large ligands [57,85].
Furthermore, there was a clear higher network contact formation with D4 R. Except for that fact that
Molecules 2019, 24, 1196 16 of 26

the D4 R is physiologically distant compared to D2 R and D3 R [82], no further explanation could be

found for this trend.
A systematic study by De Freitas and Schapira [89] showed that the most frequent type of
non-covalent interactions for protein-ligand complexes were hydrophobic contacts, followed by
hydrogen bonding, π-stacking, salt-bridges, amide-stacking (corresponds to T-stack) and lastly
cation-π-stacking. The same ranking of frequency of interaction type was found in our study.
As also described by Davis and Teague [79] hydrophobic contacts are the most common type of
receptor-ligand-interactions as they not only enhance binding affinity but also are sometimes favored
over tight, charged hydrogen bonds [79]. In addition, they can be formed with different ligand-atoms
such as carbons, halogens or sulphurs [89]. As reviewed in Davis and Teague [79] most docking studies
fail to count in the hydrophobicity for their ligands. However, the balance between polarity (causing
hydrogen bonds) and lipophicity (causing hydrophobic contacts) is the main drive to make a ligand
“drug-like” [79]. Our study was successful to determine not only the hydrogen bonds but also the large
hydrophobic network of each “drug-like” ligand (as well as of the marketed drugs). Hydrophobic
contacts appeared to form a huge network of conserved and non-conserved residues that stabilized
ligand positions during binding. This network was spread across TM2-TM3-TM7. Residues from TM1
and TM2 were shown to be relevant for binding large ligands such as nemonapride. Lastly, T-stacking
interactions revealed as especially relevant for some large ligands such as apomorphine, risperidone
or aripiprazole.
Conserved residues in the OBP were found to be clustered in microdomains, stabilizing
ligand-binding through the formation of a HB network. Indeed, HBs where mostly mediated by
the serine microdomain (5.42Ser, 5.34Ser and 5.46Ser especially at D2 R and D5 R). Interestingly,
these residues were not relevant for D1 R, although a study by Hugo et al. mentioned 5.46Ser as
key residue for activating D1 R [90]. In this study, 3.37Trp was also proposed to be mediator of the
D1 R-activation [90]. We were not able to confirm these findings in our study, only bromocriptine and
spiperone were interacting 3.37Trp at D1 R, while at D5 R we did not observe any interaction with this
residue. 3.37Thr D2 R was found to interact with 7-OH-DPAT, indicating that these residues may not be
D1 R-specific. Salt-bridges were exclusively formed by 3.32Asp but appeared to involve also residues
from ECL1 for spiperdone for D1 R and D3 R. For “bulky” ligands such as clozapine or bromocriptine
no salt-bridges were formed.
Frontera et al. reported that the strength of cation-π-interactions is also influenced by the presence
of weaker interactions such as hydrogen or hydrophobic bonds [81]. For instance, it is well known
that H-bonding is highly contributing to the bond strength of π-stacking [81]. But not only weaker
interactions benefit π-interactions, cat-π and π-π-stacking were also found to be cooperative for
each other [81]. Such combinations where cat-π and π-π-stacking were simultaneously present,
were indisputably found for D2 -like rather than for D1 -like DRs. In addition, these residues and those
of the TM6 aromatic microdomain (6.48Trp, 6.51Phe, 6.52Phe, 6.55His/Asn) were mostly involved
in forming π-interactions (cat-π, π-π or T-stack). Phenylalanine, tyrosine and tryptophan interacting
ligands could indeed be further extended in order to design a new selective SAR for D5 R, as they
were found to be exclusively involved in π-interactions and π-stacking formation at this DR subtype.
Since for the D1 R-like DR SCH23390 and SKF38393 are the only known selective ligands, a closer look at
the interacting residues of these ligands revealed that cat-π-interactions (6.30Glu, 6.39Val, 6.42Gly) were
only present at the D1 R for SCH23390, the antagonist at the D1 -like DR [88]. Moreover, these residues
were not the “classical” TM6 residues usually involved in binding, while this was true for the other
ligands. This encourages the search for D1 R- or D5 R-selective ligands, which should ideally form
cat-π-interactions with certain amino-acids, as they were found in this ligand set. From a structural
basis SCH23390 and SKF38393 are more related to the benzodiazepines, compared to the other
ligands that are either small molecules or longer ligands with piperidine moieties [91]. Lastly, another
difference found between SCH23390 and SKF38393 binding to D5 R were that SKF38393 established
more interactions with residues from different TMs and a variety of neighboring residues of the
Molecules 2019, 24, 1196 17 of 26

“classical” interacting residues; whereas SCH23390-receptor-interactions were more limited to a smaller

number of residues. These observations were not found for both ligands at the D1 R. Reported by
Bourne, who discovered SCH23390, this compound is the 3-methyl, 7-chloro analogue of the D1 R
agonist SKF38393, which is furthermore enantioselective [88]. In addition, it was stated that the
phenyl ring in the benzodiazepine-derivatives and the receptors was involved in electrostatic forces,
important for binding [88,92]. Mapping the full electrostatic potential of the D5 R using ligands with
benzodiazepine properties may be useful to find D5 R-selective SAR.
In order to find future SARs for DRs and improve subtype selectivity, we should not only
considerer the known “classical” residues and binding motifs such as the “DRY” motif, but also
conserved neighboring amino acids as shown herein. For sure, this would improve the treatment with
antipsychotics of many patients.

4. Materials and Methods

4.1. Homology Modeling

4.1.1. General Approach

The apo-DR models were generated with MODELLER 9.19 (version MODELLER 9.19, released Jul
25th, 2017) (University of California San Francisco, San Francisco, CA, USA) [93]. For D2 -like receptors
we used their corresponding crystal structures as templates: the D2 R complexed with risperidone
(PDBid: 6CM4) [36], the D3 R complexed with D2 R-antagonist eticlopride, (PDBid: 3PBL) [37] or D4 R
complexed with D2R/D3R-antagonist nemonapride (PDBid: 5WIU) [38]. Depending on the sequence
similarity obtained with Basic Local Alignment Search Tool (BLAST, NCBI, Rockville, MD, USA) [40]
and ClustalOmega (EMBL-EBI, Cambridgeshire, UK) [41] and listed in Table 2, either D3 R (for the D1 R)
or D4 R (for the D5 R) were chosen as template to model the D1-like DRs. Due to the length of the ICL3,
this was cut and substituted with two or four alanine residues, for D2- and D1-like receptors. Water
and co-crystalized compounds were removed from the template structures. In the modeling protocol
the lengths of the TMs and the perimembrane intracellular helix (HX8) were specified. In addition,
disulphide bonds were constricted in the known pairs of cysteines, in particular between 3.25Cys and
a non-conserved cysteine in ECL2 and between two non-conserved cysteines in the ECL3. Furthermore,
loop refinement was performed for extracellular and intracellular loops for all DR using the module
“loop refinement” of MODELLER 9.19. The number of models calculated with MODELLER [93] was
set to 100.

Table 2. Identity between DRs in study and their corresponding templates calculated with BLAST [40]
and ClustalOmega [41].

DOPAMINE RECEPTOR TEMPLATE BLAST [%] CLUSTALOMEGA [%]

D1 R 3PBL 35.0 39.5
D2 R 6CM4 97.0 100.0
D3 R 3PBL 93.0 99.3
D4 R 5WIU 93.0 100.0
D5 R 5WIU 35.0 39.1

4.1.2. Model Evaluation/Methods of Quality

There are several approaches to validate homology models such as built-in metrics of
open-source [52] and licensed softwares [94]. In a preliminary study we experienced [50]
that the combination of different independent metrics provided adequate models suitable for
molecular docking. For instance, the combination of MODELLER’s metrics [95], ProSA-web [43,44]
and ProQ [45,46] revealed to be a promising and reliable protocol to create valid models for
molecular docking.
Molecules 2019, 24, 1196 18 of 26

Discrete Optimized Protein Energy (DOPE) [42] scores are MODELLER’s standard metrics and
were utilized in combination with visual inspection to initially remove incorrect models. DOPE is
specific for a given target sequence, e.g., it accounts for the finite and spherical shape of native protein
states with the lowest free energy [42]. It should be noted, that although DOPE is not an absolute
measure, it helps to rank the proposed models. Then, out of a small set of potential candidates (selection
of 5–10), ProSA web service [44] and the online ProQ prediction server [46] were used to determine
the final models with the best combination of scores. For the z-score provided by ProSA-web analysis
values around −4 are suggested as acceptable. It was only used for error recognition, as it indicates
overall model quality with respect to an energy distribution derived from random conformations for
globular proteins [44] The ProQ analysis (LGscore [95] and MaxSub [96]) provides absolute measures
based on a neural network, which were set as base for the more detailed evaluation of the models.
Regarding the LGscore, values > 3, for MaxSub values > 0.5 are typically considered as “good”.
Additionally, ProQ allows to include secondary structure information calculated with PSIPRED [97],
improving the prediction accuracy and increasing the model quality up to 15%. The ProQ analysis was
only carried out, if z-scores around 2–4 were achieved using the ProQ protocol.
We could not compare our models with other authors as metrics scores are mostly not
shown [43,98]. D1 -like models, without a known crystal structure and D2 -like models for which
there are 3D crystal structure, showed similar quality (Table 3).

Table 3. Metrics and scores of the final DR homology models used herein.

DR DOPE LGscore LGscore + PSIPRED MaxSub MaxSub + PSIPRED z-Score

D1 R −39070.82 2.53 4.26 0.18 0.53 −2.14
D2 R −39284.66 2.52 4.22 0.21 0.52 −2.22
D3 R −39458.37 3.14 4.19 0.27 0.55 −3.12
D4 R −36738.05 3.33 4.25 0.25 0.59 −3.90
D5 R −38356.05 2.60 4.14 0.15 0.57 −1.49

4.2. Molecular Dynamics

4.2.1. System Setup

It is well known that GPCRs take an infinite number of conformations over time. As such we
performed MD simulations of modelled apo-forms to verify the effect of punctual fluctuations into
the overall binding arrangements of ligands. Before setting up the system, the selected DR models
were subjected to the Orientations of Proteins in Membranes (OPM) web-server [99–102] to calculate
spatial orientations respecting to the Membrane Normal defined by the z-axis. In addition, the state
of titratable residues was calculated by Propka 3.1 [103,104] within the PDB2PQR web-server [105]
at a pH of 7.0. The prepared receptor structures were inserted into a previously constructed lipid
bilayer of POPC: Cholesterol (9:1). Insertion of the receptors in the membrane was performed with
g_membed package of GROMACS [106]. System was then solvated with explicitly represented water.
Sodium and chloride ions were added to neutralize the system until it reached a total concentration of
0.15 M. The final systems dimensions were 114 × 114 × 107 Å and included approximately 370 POPC,
40 cholesterols, 125 sodium ions, 139 chloride ions and 28500 water molecules, with small variations
from receptor to receptor.

4.2.2. Molecular Dynamics Simulation Protocol

CHARMM36 force field, with an implemented CMAP correction, was used for ions, water
(TIP3P model), lipids and protein parameters [107]. Prior to MD simulation, the systems were
relaxed to remove any possible steric clashes by a set of 50,000 steps of Steepest Descent energy
minimization. Equilibration was performed afterwards as following: the system was heated using
Nosé-Hoover thermostat from 0 to 310.15 K in the NVT ensemble over 100 ps with harmonic restraints
Molecules 2019, 24, 1196 19 of 26

of 10.0 kcal/mol. Then systems were subjected through a first step of NPT ensemble of 1 ns with
semi isotropic pressure coupling and a pressure of one bar. Further equilibration was performed with
sequential release of membrane lipids and protein’s atoms with a final step of NPT ensemble with
harmonic restraints on the protein of 1.0 kcal/mol, for a total of 5 ns of restrained equilibration.
MD simulations of all DR models were performed with the periodic boundary condition to
produce isothermical-isobaric ensembles using GROMACS 5.1.1 [106]. The Particle Mesh Ewald (PME)
method [108] was used to calculate the full electrostatic energy of a unit cell in a macroscopic lattice of
repeating images. Temperature was regulated using the Nosé-Hoover thermostat at 310.15 K. Pressure
was regulated using the Parrinello-Rahman algorithm. The equations of motion were integrated using
leapfrog algorithm with a time step of 2 fs. All bonds, involving hydrogen atoms within protein
and lipid molecules were constrained using the LINear Constraint Solver (LINCS) algorithm [109].
Additionally, a cut-off distance of 12 Å was attributed for Coulombic and van der Waals interactions.
Then a single independent simulation of 100 ns was initialized from the final snapshot of the restrained
equilibration from each DR, for a total of 5 simulations. Trajectory analysis was performed by in-house
scripting using GROMACS [106] and Visual Molecular Dynamics (VMD) [110]. Trajectory snapshots
were saved every 5 ns. The snapshots after the first 50 ns MD stabilization were used for molecular
docking studies.

4.3. Molecular Docking

4.3.1. Ligand Dataset

The following ligands were docked into the receptor decoys: dopamine,
7-hydroxy-N,N-dipropyl-2-aminotetralin (7-OH-DPAT), apomorphine, bromocriptine, clozapine,
nemonapride, sulpiride, SCH23390, SKF38393, eticlopride, risperidone, aripiprazole, haloperidole,
spiperone and chlorpromazine (Table 1). All structures were obtained from the DrugBank database
(https://www.drugbank.ca) or from ChemSpider (http://www.chemspider.com) [111].

4.3.2. Docking Procedure

DR binding pocket was defined in several experimental and computational
studies [2,47,52,55,57,59,85]. Herein, we used the comprehensive review by Floresca and Schetz [47]
as a base for exploration of the DR binding pocket, since it contains detailed experimental data.
A summary of the procedure can be better reviewed in Bueschbell et al. [50]. AutoDock4.2 (version
AutoDock 4.2.6, released in 2009) was used to perform ligand docking [112]. DR hydrogens were
added and Kollman united atom charges were assigned. Hydrogens were also added to the ligand
and Gasteiger-Marsili was used to calculate charges. Before docking an energy, grid was created using
AutoGrid (version AutoGrid 4.2.6, released 2009) with a box-size varying with the times step and
ligand. For each docking simulation 100 independent Lamarckian genetic algorithm (LGA) runs
were performed with the number of energy evaluations set to 10.000.000, the population size set to
200 and the maximum number of generations set to 27.000. Default settings were maintained for
the rest of the parameters. Docked conformations within a RMSD of 2 Å were clustered. The most
populated and lowest energy cluster (Gibbs free energy of binding) was used for conformational
analysis. To find the local energy minimum of the binding site with a limited search space to that
region, a low-frequency local search method was used. The 100 conformations obtained from docking
were clustered by low-energy and RMSD. The top-ranked conformations within the best 3 clusters
were visually inspected. The docking parameters were not changed for any ligand, only the residues
treated as flexible in the docking protocol differed between the ligands. The flexible residues for each
DR model are summarized in Table 4.
Molecules 2019, 24, 1196 20 of 26

Table 4. Flexible residues used in the molecular docking different ligands.

LIGAND FLEXIBLE RESIDUES IN B&W NUMBERING

DOPAMINE
3.32Asp, 5.42Ser, 5.43Ser, 5.46Ser, 6.48Trp, 6.51Phe, 6.52Phe, 6.55His/Asn
7-OH-DPAT
APOMORPHINE
3.32Asp, 3.36/3.35Cys, 5.42Ser, 5.43Ser, 5.46Ser, 6.48Trp, 6.51Phe, 6.52Phe, 6.55His/Asn
BROMOCRIPTINE
CLOZAPINE 3.32Asp, 3.33Val, 3.36Cys, 5.42Ser, 5.43Ser, 5.46Ser, 6.48Trp, 6.55His/Asn
NEMONAPRIDE 2.57Val, 3.32Asp, 5.42Ser, 5.43Ser, 5.46Ser, 6.48Trp, 6.51Phe, 6.52Phe, 7.43Tyr
SULPIRIDE 3.32Asp, 6.48Trp, 5.42Ser, 5.43Ser, 5.46Ser, 6.55His/Asn, 7.43Tyr, 6.51Phe
SCH23390
3.32Asp, 6.48Trp, 5.42Ser, 5.43Ser, 5.46Ser, 6.55His/Asn, 6.51Phe, 6.52Phe
SKF38393
ETICLOPRIDE 3.32Asp, 6.48Trp, 5.42Ser, 5.43Ser, 5.46Ser, 6.55His/Asn, 7.43Tyr, 6.51Phe, 6.52Phe
RISPERIDONE 3.32Asp, 6.48Trp, 3.36Cys, 6.55His/Asn, 2.57Val, 5.42Ser, 5.43Ser, 5.46Ser
ARIPIPRAZOLE 3.32Asp, 6.48Trp, 3.33Val, 5.42Ser, 5.43Ser, 5.46Ser, 7.43Tyr, 6.55His/Asn
HALOPERIDOLE 3.32Asp, 6.48Trp, 6.51Phe, 6.52Phe, 3.36Cys, 2.57Val, 5.42Ser, 5.43Ser, 5.46Ser
SPIPERONE 3.32Asp, 6.48Trp, 5.42Ser, 5.43Ser, 5.46Ser, 3.36Cys, 6.55His/Asn, 2.57Val
CHLORPROMAZINE 3.32Asp, 6.48Trp, 5.42Ser, 5.43Ser, 5.46Ser, 6.55His/Asn, 3.36Cys, 6.51Phe

4.3.3. Analysis of Molecular Docking

In this study, 15 DR ligands were docked to the homology model and to different conformational
arrangements retrieved at every 5 ns for the 55–100 ns range for each DR simulation (825 dockings
in total). All distances between the center of mass of the ligand and the alpha-C-atom (Cα) of
the residues, treated as flexible in the docking protocol, were calculated using in-house PyMOL
scripts [2,17,47,52,57,59] as well as previously published work [50]. We also develop in-house
BINANA scripts to predict the main receptor-ligand interactions [39]. BINANA is an open-source
python-implemented algorithm which uses output files from AutoDock [112] for the analysis of
interactions and visualizes them in the free molecular-visualization program VMD [110]. Key binding
characteristics such as hydrogen bonds, hydrophobic contacts, salt-bridges and π-interactions were
calculated with BINANA.

Supplementary Materials: The following are available online. Figure S1—RMSD throughout the 100 ns of
simulation for all DR models; Figure S2—Redocking of ligands with their respective DR and bound ligand;
Figure S3—Molecular docking of Dopamine at the D1R–D5R during 0–60 ns; Figure S4—Molecular docking of
Dopamine at the D1R–D5R during 65–75 ns; Figure S5—Molecular docking of Dopamine at the D1R–D5R during
80–90 ns; Figure S6—Molecular docking of Dopamine at the D1R–D5R during 95 and 100 ns; Figure S7—Results
of the molecular docking of bromocriptine, clozapine, sulpiride, eticlopride, risperidone, aripiprazole and
risperidone for all DR subtypes at time points [ns]; Figure S8—Total interactions counted for each DR over time
points [ns]; Figure S9—Pairwise interaction results for dopamine; Figure S10—Pairwise interaction results for
7-OH-DPAT; Figure S11—Pairwise interaction results for apomorphine; Figure S12—Pairwise interaction results
for bromocriptine; Figure S13—Pairwise interaction results for clozapine; Figure S14—Pairwise interaction results
for nemonapride; Figure S15—Pairwise interaction results for sulpiride; Figure S16—Pairwise interaction results
for SCH23390; Figure S17—Pairwise interaction results for SKF38393; Figure S18—Pairwise interaction results
for eticlopride; Figure S19—Pairwise interaction results for risperidone; Figure S20—Pairwise interaction results
for aripiprazole; Figure S21—Pairwise interaction results for haloperidole; Figure S22—Pairwise interaction
results for spiperone; Figure S23—Pairwise interaction results for chlorpromazine. Table S1—Comparison
between the total and transmembrane specific identity [%] of the DR model with their crystal structure
templates calculated with Clustal Omega; Table S2—Averages RMSD values for TM throughout the simulations;
Table S3—Summary of the structures used in literature for defining the binding pocket for the D2 R and source
(experimental and computational); Table S4—Docking results for the D1 R; Table S5—Docking results for the D2 R;
Table S6—Docking results for the D3 R; Table S7—Docking results for the D4 R; Table S8—Docking results for the
D5 R; Table S9—Docking results for the crystal structure templates of D2 R (PDBid: 6CM4), D3 R (PDBid: 3PBL)
and D4 R (PDBid: 5WIU) docked with their co-crystalized ligands; Table S10—D1 R residues with Ballesteros &
Weinstein-numbering participating in different interaction types sorted by ligands; Table S11—D2 R residues with
Ballesteros & Weinstein-numbering participating in different interaction types sorted by ligands; Table S12—D3 R
Molecules 2019, 24, 1196 21 of 26

residues with Ballesteros & Weinstein-numbering participating in different interaction types sorted by ligands;
Table S13—D4 R residues with Ballesteros & Weinstein-numbering participating in different interaction types
sorted by ligands; Table S14—D5 R residues with Ballesteros & Weinstein-numbering participating in different
interaction types sorted by ligands.
Author Contributions: B.B., A.J.P. and C.A.V.B. performed the experiments; B.B, A.J.P. and C.A.V.B. analyzed the
data; B.B., A.C.S and I.S.M conceived and designed the experiments; all authors wrote the paper.
Funding: B.B. and A.C.S. were supported by the German Federal Ministry of Education and Research (BMBF
project) of the Bonn International Graduate School in Drug Sciences (BIGS DrugS). Irina S. Moreira acknowledges
support by the Fundação para a Ciência e a Tecnologia (FCT) Investigator programme - IF/00578/2014
(co-financed by European Social Fund and Programa Operacional Potencial Humano). This work was also
financed by the European Regional Development Fund (ERDF), through the Centro 2020 Regional Operational
Programme under project CENTRO-01-0145-FEDER-000008: BrainHealth 2020. We also acknowledge the grant
PTDC/QUI-OUT/32243/2017 financed by national funds through the FCT / MCTES and/or State Budget.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Beaulieu, J.-M.; Gainetdinov, R.R. The Physiology, Signaling, and Pharmacology of Dopamine Receptors.
Pharmacol. Rev. 2011, 63, 182–217. [CrossRef]
2. Platania, C.B.M.; Salomone, S.; Leggio, G.M.; Drago, F.; Bucolo, C. Homology Modeling of Dopamine D2
and D3 Receptors: Molecular Dynamics Refinement and Docking Evaluation. PLoS ONE 2012, 7, e44316.
[CrossRef]
3. Marsden, C.A. Dopamine: The rewarding years. Br. J. Pharmacol. 2006, 147 (Suppl. 1), 136–144. [CrossRef]
4. Leggio, G.M.; Bucolo, C.; Platania, C.B.M.; Salomone, S.; Drago, F. Current drug treatments targeting
dopamine D3 receptor. Pharmacol. Ther. 2016, 165, 164–177. [CrossRef]
5. Rosenbaum, D.M.; Rasmussen, S.G.F.; Kobilka, B.K. The structure and function of G protein-coupled
receptors. Nature 2009, 459, 356–363. [CrossRef]
6. Maurice, P.; Guillaume, J.L.; Benleulmi-Chaachoua, A.; Daulat, A.M.; Kamal, M.; Jockers, R. GPCR-Interacting
Proteins, Major Players of GPCR Function. In GPCR—Interacting Proteins, Major Players of GPCR Function,
1st ed.; Elsevier Inc.: New York, NY, USA, 2011; Volume 62.
7. Thimm, D.; Funke, M.; Meyer, A.; Müller, C.E. 6-Bromo-8-(4-methoxybenzamido)-4-oxo-4
H-chromene-2-carboxylic Acid: A powerful tool for studying orphan G protein-coupled receptor
GPR35. J. Med. Chem. 2013, 56, 7084–7099. [CrossRef]
8. Jaber, M.; Robinson, S.W.; Missale, C.; Caron, M.G. Dopamine receptors and brain function.
Neuropharmacology 1997, 35, 1503–1519. [CrossRef]
9. Seeman, P. Atypical Antipsychotics: Mechanism of Action. Focus 2002, 47, 27–38. [CrossRef]
10. Rangel-Barajas, C.; Coronel, I.; Florán, B. Dopamine Receptors and Neurodegeneration. Aging Dis. 2015, 6,
349. [CrossRef]
11. Amato, D.; Vernon, A.C.; Papaleo, F. Dopamine, the antipsychotic molecule: A perspective on mechanisms
underlying antipsychotic response variability. Neurosci. Biobehav. Rev. 2017, 85, 146–159. [CrossRef]
12. Noble, E.P. D2 dopamine receptor gene in psychiatric and neurologic disorders and its phenotypes. Am. J.
Med. Genet. 2003, 116B, 103–125. [CrossRef]
13. Zhang, A.; Neumeyer, J.L.; Baldessarini, R.J. Recent progress in development of dopamine receptor
subtype-selective agents: Potential therapeutics for neurological and psychiatric disorders. Chem. Rev.
2007, 107, 274–302. [CrossRef]
14. Miller, R. Mechanisms of action of antipsychotic drugs of different classes, refractoriness to therapeutic effects
of classical neuroleptics, and individual variation in sensitivity to their actions: Part II. Curr. Neuropharmacol.
2009, 7, 315–330. [CrossRef]
15. Mauri, M.C.; Paletta, S.; Maffini, M.; Colasanti, A.; Dragogna, F.; Di Pace, C.; Altamura, A.C. Clinical
pharmacology of atypical antipsychotics: An update. EXCLI J. 2014, 13, 1163–1191.
16. Sykes, D.A.; Moore, H.; Stott, L.; Holliday, N.; Javitch, J.A.; Lane, J.R.; Charlton, S.J. Extrapyramidal side
effects of antipsychotics are linked to their association kinetics at dopamine D2 receptors. Nat. Commun.
2017, 8, 763. [CrossRef]
Molecules 2019, 24, 1196 22 of 26

17. Salmas, R.; Serhat Is, Y.; Durdagi, S.; Stein, M.; Yurtsever, M. A QM protein–ligand investigation of
antipsychotic drugs with the dopamine D2 Receptor (D2 R). J. Biomol. Struct. Dyn. 2018, 36, 2668–2677.
[CrossRef]
18. Loebel, A.; Citrome, L.; Correll, C.U.; Xu, J.; Cucchiaro, J.; Kane, J.M. Treatment of early non-response in
patients with schizophrenia: Assessing the efficacy of antipsychotic dose escalation. BMC Psychiatry 2015,
15, 1–7. [CrossRef]
19. Behere, B.P.; Das, A.; Behere, A.P. Antipsychotics. In Clinical Psychopharmacology; Springer: Singapore, 2019;
pp. 39–87.
20. Moritz, A.E.; Benjamin Free, R.; Sibley, D.R. Advances and challenges in the search for D2 and D3 dopamine
receptor-selective compounds. Cell. Signal. 2018, 41, 75–81. [CrossRef]
21. Banala, A.K.; Levy, B.A.; Khatri, S.S.; Furman, C.A.; Roof, R.A.; Mishra, Y.; Gri, S.A.;
Sibley, D.R.; Luedtke, R.R.; Newman, A.H. N-(3-Fluoro-4-(4-(2-methoxy or 2,3-dichlorophenyl)piperazine-
1-yl)arylcarboxamides as selective dopamine D3 receptor ligands: Critical role of the carboxamide linker for
D3 recetpor selectivity. J. Med. Chem. 2011, 54, 3581–3594. [CrossRef]
22. Newman, A.H.; Beuming, T.; Banala, A.K.; Donthamsetti, P.; Pongetti, K.; LaBounty, A.; Levy, B.A.; Cao, J.;
Michino, M.; Luedtke, R.R.; et al. Molecular determinants of selectivity and efficacy at the dopamine D3
receptor. J. Med. Chem. 2012, 55, 6689–6699. [CrossRef]
23. Damsma, G.; Bottema, T.; Westerink, B.H.C.; Tepper, P.G.; Dijkstra, D.; Pugsley, T.A.; MacKenzie, R.G.;
Heffner, T.G.; Wikström, H. Pharmacological aspects of R-(+)-7-OH-DPAT, a putative dopamine D3 receptor
ligand. Eur. J. Pharmacol. 1993, 249, 9–10. [CrossRef]
24. Lévesque, D.; Diaz, J.; Pilon, C.; Martres, M.P.; Giros, B.; Souil, E.; Schott, D.; Morgat, J.L.; Schwartz, J.C.;
Sokoloff, P. Identification, characterization, and localization of the dopamine D3 receptor in rat brain using
7-[3H]hydroxy-N,N-di-n-propyl-2-aminotetralin. Proc. Natl. Acad. Sci. USA 1992, 89, 8155–8159. [CrossRef]
25. Sampson, D.; Zhu, X.Y.; Eyunni, S.V.K.; Etukala, J.R.; Ofori, E.; Bricker, B.; Lamango, N.S.; Setola, V.; Roth, B.L.;
Ablordeppey, S.Y. Identification of a new selective dopamine D4receptor ligand. Bioorg. Med. Chem. 2014, 22,
3105–3114. [CrossRef]
26. Zhang, J.; Xiong, B.; Zhen, X.; Zhang, A. Dopamine D1 receptor ligands: Where are we now and where are
we going. Med. Res. Rev. 2009, 29, 272–294. [CrossRef]
27. Conroy, J.L.; Free, R.B.; Sibley, D.R. Identification of G Protein-Biased Agonists That Fail to Recruit β-Arrestin
or Promote Internalization of the D1 Dopamine Receptor. ACS Chem. Neurosci. 2015, 6, 681–692. [CrossRef]
28. O’Sullivan, G.J.; Roth, B.L.; Kinsella, A.; Waddington, J.L. SK&F 83822 distinguishes adenylyl cyclase from
phospholipase C-coupled dopamine D1 -like receptors: Behavioural topography. Eur. J. Pharmacol. 2004, 486,
273–280.
29. Butini, S.; Nikolic, K.; Kassel, S.; Brückmann, H.; Filipic, S.; Agbaba, D.; Gemma, S.; Brogi, S.; Brindisi, M.;
Campiani, G.; et al. Polypharmacology of dopamine receptor ligands. Prog. Neurobiol. 2016, 142, 68–103.
[CrossRef]
30. Lee, S.M.; Kant, A.; Blake, D.; Murthy, V.; Boyd, K.; Wyrick, S.J.; Mailman, R.B. SKF-83959 is not
a highly-biased functionally selective D1dopamine receptor ligand with activity at phospholipase C.
Neuropharmacology 2014, 86, 145–154. [CrossRef]
31. Arimitsu, E.; Ogasawara, T.; Takeda, H.; Sawasaki, T.; Ikeda, Y.; Hiasa, Y.; Maeyama, K. The ligand binding
ability of dopamine D1 receptors synthesized using a wheat germ cell-free protein synthesis system with
liposomes. Eur. J. Pharmacol. 2014, 745, 117–122. [CrossRef]
32. Sliwoski, G.; Kothiwale, S.; Meiler, J.; Lowe, E.W. Computational methods in drug discovery. Comput.
Methods Drug Discov. 2014, 66, 334–395. [CrossRef]
33. Jain, A. Computer Aided Drug Design & QSAR. J. Phys. Conf. Ser. 2017, 884, 012072.
34. Lemos, A.; Melo, R.; Preto, A.J.; Almeida, J.G.; Moreira, I.S.; Natália, M.D.S. In silico studies targeting
G-protein coupled receptors for drug research against Parkinson’s disease. Curr. Neuropharmacol. 2018, 16,
786–848. [CrossRef]
35. Shin, W.H.; Christoffer, C.W.; Kihara, D. In silico structure-based approaches to discover protein-protein
interaction-targeting drugs. Methods 2017, 131, 22–32. [CrossRef]
36. Wang, S.; Che, T.; Levit, A.; Shoichet, B.K.; Wacker, D.; Roth, B.L. Structure of the D2 dopamine receptor
bound to the atypical antipsychotic drug risperidone. Nature 2018, 555, 269. [CrossRef]
Molecules 2019, 24, 1196 23 of 26

37. Chien, E.Y.T.; Liu, W.; Zhao, Q.; Katritch, V.; Won Han, G.; Hanson, M.A.; Shi, L.; Newman, A.H.; Javitch, J.A.;
Cherezov, V.; et al. Structure of the Human Dopamine D3 Receptor in Complex with a D2 /D3 Selective
Antagonist. Science 2010, 330, 1091–1095. [CrossRef]
38. Wang, S.; Wacker, D.; Levit, A.; Che, T.; Betz, R.M.; Mccorvy, J.D.; Venkatakrishnan, A.J.; Huang, X.-P.;
Dror, R.O.; Shoichet, B.K.; et al. D4 dopamine receptor high-resolution structures enable the discovery of
selective agonists. Science 2017, 358, 381–386. [CrossRef]
39. Durrant, J.D.; McCammon, J.A. BINANA: A novel algorithm for ligand-binding characterization. J. Mol.
Graph. Model. 2011, 29, 888–893. [CrossRef]
40. Altschul, S.F.; Gish, W.; Miller, W.; Myers, E.W.; Lipman, D.J. Basic local alignment search tool. J. Mol. Biol.
1990, 215, 403–410. [CrossRef]
41. Sievers, F.; Wilm, A.; Dineen, D.; Gibson, T.J.; Karplus, K.; Li, W.; Lopez, R.; McWilliam, H.; Remmert, M.;
Söding, J.; et al. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal
Omega. Mol. Syst. Biol. 2011, 7, 539. [CrossRef]
42. Shen, M.; Sali, A. Statistical potential for assessment and prediction of protein structures. Protein Sci. 2006,
15, 2507–2524. [CrossRef]
43. Hou, J.; Charron, C.L.; Fowkes, M.M.; Luyt, L.G. Bridging computational modeling with amino acid
replacements to investigate GHS-R1a-peptidomimetic recognition. Eur. J. Med. Chem. 2016, 123, 822–833.
[CrossRef]
44. Wiederstein, M.; Sippl, M.J. ProSA-web: Interactive web service for the recognition of errors in
three-dimensional structures of proteins. Nucleic Acids Res. 2007, 35, 407–410. [CrossRef]
45. Wallner, B.; Elofsson, A. Can correct protein models be identified? Protein Sci. 2003, 12, 1073–1086. [CrossRef]
46. Wallner, B.; Elofsson, A. Identification of correct regions in protein models using structural, alignment,
and consensus information. Protein Sci. 2006, 15, 900–913. [CrossRef]
47. Floresca, C.Z.; Schetz, J.A. Dopamine Receptor Microdomains Involved in Molecular Recognition and the
Regulation of Drug Affinity and Function. J. Recept. Signal Transduct. 2004, 24, 207–239. [CrossRef]
48. Cummings, D.F.; Ericksen, S.S.; Goetz, A.; Schetz, J.A. Transmembrane Segment Five Serines of the D4
Dopamine Receptor Uniquely Influence the Interactions of Dopamine, Norepinephrine, and Ro10-4548. J.
Pharmacol. Exp. Ther. 2010, 333, 682–695. [CrossRef]
49. Ericksen, S.S.; Cummings, D.F.; Teer, M.E.; Amdani, S.; Schetz, J.A. Ring Substituents on Substituted
Benzamide Ligands Indirectly Mediate Interactions with Position 7.39 of Transmembrane Helix 7 of the D4
Dopamine Receptor. J. Pharmacol. Exp. Ther. 2012, 342, 472–485. [CrossRef]
50. Bueschbell, B.; Preto, A.J.; Barreto, C.A.V.; Schiedel, A.C.; Moreira, I.S. Creating a valid in silico Dopamine
D2 -receptor model for small molecular docking studies. In MOL2NET, International Conference Series on
Multidisciplinary Sciences; SCIFORUM: Basel, Switzerland, 2017; Volume 3, pp. 1–6.
51. Ballesteros, J.A.; Weinstein, H. Integrated methods for the construction of three dimensional models and
computational probing of structure-function relations in G-protein coupled receptors. Methods Neurosci.
1995, 25, 366–428.
52. Salmas, R.E.; Yurtsever, M.; Stein, M.; Durdagi, S. Modeling and protein engineering studies of active and
inactive states of human dopamine D2 receptor (D2R) and investigation of drug/receptor interactions. Mol.
Divers. 2015, 19, 321–332. [CrossRef]
53. Moreira, I.S.; Shi, L.; Freyberg, Z.; Ericksen, S.S.; Weinstein, H.; Javitch, J.A. Structural Basis of Dopamine
Receptor Activation. In The Dopamine Receptors; Neve, K.A., Ed.; Humana/Springer: Berlin, Germany, 2010;
pp. 47–73.
54. Huang, E.S. Construction of a sequence motif characteristic of aminergic G protein-coupled receptors. Protein
Sci. 2003, 12, 1360–1367. [CrossRef]
55. Tschammer, N.; Dörfler, M.; Hübner, H.; Gmeiner, P. Engineering a GPCR-ligand pair that simulates the
activation of D 2L by dopamine. ACS Chem. Neurosci. 2010, 1, 25–35. [CrossRef]
56. Kling, R.C.; Tschammer, N.; Lanig, H.; Clark, T.; Gmeiner, P. Active-state model of a dopamine D2
receptor—Galpha-i complex stabilized by aripiprazole-type partial agonists. PLoS ONE 2014, 9, e100069.
[CrossRef]
Molecules 2019, 24, 1196 24 of 26

57. Kalani, M.Y.S.; Vaidehi, N.; Hall, S.E.; Trabanino, R.J.; Freddolino, P.L.; Kalani, M.A.; Floriano, W.B.;
Kam, V.W.T.; Goddard, W.A. The predicted 3D structure of the human D2 dopamine receptor and the
binding site and binding affinities for agonists and antagonists. Proc. Natl. Acad. Sci. USA 2004, 101,
3815–3820. [CrossRef]
58. Holst, B.; Nygaard, R.; Valentin-Hansen, L.; Bach, A.; Engelstoft, M.S.; Petersen, P.S.; Frimurer, T.M.;
Schwartz, T.W. A conserved aromatic lock for the tryptophan rotameric switch in TM-VI of
seven-transmembrane receptors. J. Biol. Chem. 2010, 285, 3973–3985. [CrossRef]
59. Männel, B.; Jaiteh, M.; Zeifman, A.; Randakova, A.; Möller, D.; Hübner, H.; Gmeiner, P.; Carlsson, J.
Structure-guided screening for functionally selective D2 dopamine receptor ligands from a virtual chemical
library. ACS Chem. Biol. 2017, 12, 2652–2661. [CrossRef] [PubMed]
60. Durdagi, S.; Salmas, R.E.; Stein, M.; Yurtsever, M.; Seeman, P. Binding Interactions of Dopamine and
Apomorphine in D2High and D2Low States of Human Dopamine D2 Receptor Using Computational and
Experimental Techniques. ACS Chem. Neurosci. 2016, 7, 185–195. [CrossRef] [PubMed]
61. Boyd, K.N.; Mailman, R.B. Dopamine receptor signaling and cirrent and future antipyschotic drugs. Handb.
Exp. Pharmacol. 2012, 212, 53–86.
62. Bergman, J.; Madras, B.K.; Spealman, R.D. Behavioral effects of D1 and D2 dopamine receptor antagonists in
squirrel monkeys. J. Pharmacol. Exp. Ther. 1991, 258, 910–917. [PubMed]
63. Chen, T.; Hu, Y.; Lin, X.; Huang, X.; Liu, B.; Leung, P.; Chan, S.O.; Guo, D.; Jin, G. Dopamine signaling
regulates the projection patterns in the mouse chiasm. Brain Res. 2015, 1625, 324–336. [CrossRef]
64. Hidaka, K.; Matsumoto, M.; Tada, S.; Tasaki, Y.; Yamaguchi, T. Differential effects of [3H]nemonapride and
[3H]spiperone binding on human dopamine D4 receptors. Neurosci. Lett. 1995, 186, 145–148. [CrossRef]
65. Seeman, P.; Van Tol, H.H. Dopamine receptor pharmacology. Trends Pharmacol. Sci. 1994, 15, 264–270.
[CrossRef]
66. Lawler, C.P.; Prioleau, C.; Lewis, M.M.; Mak, C.; Jiang, D.; Schetz, J.A.; Gonzalez, A.M.; Sibley, D.R.;
Mailman, R.B. Interactions of the novel antipsychotic aripiprazole (OPC-14597) with dopamine and serotonin
receptor subtypes. Neuropsychopharmacology 1999, 20, 612–627. [CrossRef]
67. Lindsley, C.W.; Hopkins, C.R. Return of D4 Dopamine Receptor Antagonists in Drug Discovery. J. Med.
Chem. 2017, 60, 7233–7243. [CrossRef]
68. Newton, C.L.; Wood, M.D.; Strange, P.G. Examining the effects of sodium ions on the binding of antagonists
to dopamine D2 and D3 receptors. PLoS ONE 2016, 11, e0158808. [CrossRef]
69. Zhang, B.; Yang, X.; Tiberi, M. Functional importance of two conserved residues in intracellular loop 1
and transmembrane region 2 of Family A GPCRs: Insights from ligand binding and signal transduction
responses of D1 and D5 dopaminergic receptor mutants. Cell. Signal. 2015, 27, 2014–2025. [CrossRef]
70. Andringa, G.; Drukarch, B.; Leysen, J.E.; Cools, A.R.; Stoof, J.C. The alleged dopamine D1 receptor agonist
SKF 83959 is a dopamine D1 receptor antagonist in primate cells and interacts with other receptors. Eur. J.
Pharmacol. 1999, 364, 33–41. [CrossRef]
71. Burris, K.D.; Molski, T.F.; Xu, C.; Ryan, E.; Tottori, K.; Kikuchi, T.; Yocca, F.D.; Molinoff, P.B. Aripiprazole,
a novel antipsychotic, is a high-affinity partial agonist at human dopamine D2 receptors. J. Pharmacol. Exp.
Ther. 2002, 302, 381–389. [CrossRef]
72. López-Muñoz, F.; Alamo, C. The consolidation of neuroleptic therapy: Janssen, the discovery of haloperidol
and its introduction into clinical practice. Brain Res. Bull. 2009, 79, 130–141. [CrossRef] [PubMed]
73. Madras, B.K. History of the discovery of the antipsychotic dopamine D2 receptor: A basis for the dopamine
hypothesis of schizophrenia. J. Hist. Neurosci. 2013, 22, 62–78. [CrossRef]
74. Abhijnhan, A.; Adams, C.E.; David, A.; Ozbilen, M. Depot fluspirilene for schizophrenia. Cochrane Database
Syst. Rev. 2007, 1, 1–45. [CrossRef] [PubMed]
75. Valizade Hasanloei, M.A.; Sheikhpour, R.; Sarram, M.A.; Sheikhpour, E.; Sharifi, H. A combined Fisher
and Laplacian score for feature selection in QSAR based drug design using compounds with known and
unknown activities. J. Comput. Aided Mol. Des. 2018, 32, 375–384. [CrossRef]
76. Kumar, S.P. PLHINT: A knowledge-driven computational approach based on the intermolecular H bond
interactions at the protein-ligand interface from docking solutions. J. Mol. Graph. Model. 2018, 79, 194–212.
[CrossRef]
Molecules 2019, 24, 1196 25 of 26

77. Trisciuzzi, D.; Nicolotti, O.; Miteva, M.A.; Villoutreix, B.O. Analysis of solvent-exposed and buried
co-crystallized ligands: A case study to support the design of novel protein–protein interaction inhibitors.
Drug Discov. Today 2018, 24, 551–559. [CrossRef]
78. Tanina, A.; Wohlkönig, A.; Soror, S.H.; Flipo, M.; Villemagne, B.; Prevet, H.; Déprez, B.; Moune, M.;
Perée, H.; Meyer, F.; et al. A comprehensive analysis of the protein-ligand interactions in crystal structures of
Mycobacterium tuberculosis EthR. Biochim. Biophys. Acta Proteins Proteom. 2019, 1867, 248–258. [CrossRef]
79. Davis, A.M.; Teague, S.J. Hydrogen bonding, hydrophobic interactions, and failure of the rigid receptor
hypothesis. Angew. Chem. Int. Ed. 1999, 38, 736–749. [CrossRef]
80. Bosch, E.; Barnes, C.L.; Brennan, N.L.; Eakins, G.L.; Breyfogle, B.E. Cation-Induced π-stacking. J. Org. Chem.
2008, 73, 3931–3934. [CrossRef]
81. Frontera, A.; Quiñonero, D.; Deyà, P.M. Cation-π and anion-π interactions. Wiley Interdiscip. Rev. Comput.
Mol. Sci. 2011, 1, 440–459. [CrossRef]
82. Wang, Q.; MacH, R.H.; Luedtke, R.R.; Reichert, D.E. Subtype selectivity of dopamine receptor ligands:
Insights from structure and ligand-based methods. J. Chem. Inf. Model. 2010, 50, 1970–1985. [CrossRef]
83. Simpson, M.M.; Ballesteros, J.A.; Chiappa, V.; Chen, J.; Suehiro, M.; Hartman, D.S.; Godel, T.; Snyder, L.A.;
Sakmar, T.P.; Javitch, J.A. Dopamine D4 /D2 receptor selectivity is determined by A divergent aromatic
microdomain contained within the second, third, and seventh membrane-spanning segments. Mol. Pharmacol.
1999, 56, 1116–1126. [CrossRef]
84. Salmas, R.E.; Yurtsever, M.; Durdagi, S. Atomistic molecular dynamics simulations of typical and atypical
antipsychotic drugs at the dopamine D2 receptor (D2R) elucidates their inhibition mechanism. J. Biomol.
Struct. Dyn. 2016, 35, 1–17. [CrossRef]
85. Sukalovic, V.; Soskic, V.; Sencanski, M.; Andric, D.; Kostic-Rajacic, S. Determination of key receptor-ligand
interactions of dopaminergic arylpiperazines and the dopamine D2 receptor homology model. J. Mol. Model.
2013, 19, 1751–1762. [CrossRef]
86. Moreira, I.S. Structural features of the G-protein/GPCR interactions. Biochim. Biophys. Acta Gen. Subj. 2014,
1840, 16–33. [CrossRef]
87. Scarselli, M.; Novi, F.; Schallmach, E.; Lin, R.; Baragli, A.; Colzi, A.; Griffon, N.; Corsini, G.U.; Sokoloff, P.;
Levenson, R.; et al. D2 /D3 Dopamine Receptor Heterodimers Exhibit Unique Functional Properties. J. Biol.
Chem. 2001, 276, 30308–30314. [CrossRef]
88. Bourne, J.A. SCH23390: The First Selective Dopamine D1-Like Receptor Antagonist. CNS Drug Rev. 2006, 7,
399–414. [CrossRef]
89. Ferreira De Freitas, R.; Schapira, M. A systematic analysis of atomic protein-ligand interactions in the PDB.
Medchemcomm 2017, 8, 1970–1981. [CrossRef]
90. Hugo, E.A.; Cassels, B.K.; Fierro, A. Functional roles of T3.37 and S5.46 in the activation mechanism of the
dopamine D1 receptor. J. Mol. Model. 2017, 23, 142. [CrossRef]
91. Zarrindast, M.R.; Honardar, Z.; Sanea, F.; Owji, A.A. SKF 38393 and SCH 23390 inhibit reuptake of serotonin
by rat hypothalamic synaptosomes. Pharmacology 2011, 87, 85–89. [CrossRef]
92. Pettersson, I.; Gundertofte, K.; Palm, J.; Liljefors, T. A Study on the Contribution of the 1-Phenyl Substituent
to the Molecular Electrostatic Potentials of Some Benzazepines in Relation to Selective Dopamine D-1
Receptor Activity. J. Med. Chem. 1992, 35, 502–507. [CrossRef]
93. Webb, B.; Sali, A.; Francisco, S. Comparative Protein Structure Modeling Using MODELLER. Curr. Protoc.
Bioinforma 2017, 54, 1–55.
94. Halgren, T.A.; Murphy, R.B.; Friesner, R.A.; Hege, S.B.; Klicic, J.J.; Mainz, D.T.; Repasky, M.P.; Knoll, E.H.;
Shelley, M.; Perry, J.K.; et al. Glide: A New Approach for Rapid, Accurate Docking and Scoring. 1. Method
and Assessment of Docking Accuracy. J. Med. Chem. 2004, 47, 1739–1749. [CrossRef]
95. Cristobal, S.; Zemla, A.; Fischer, D.; Rychlewski, L.; Elofsson, A. A study of quality measures for protein
threading models. BMC Bioinform. 2001, 2, 5. [CrossRef]
96. Siew, N.; Elofsson, A.; Rychlewski, L.; Fischer, D. MaxSub: An automated measure for the assessment of
protein structure prediction quality. Bioinformatics 2000, 16, 776–785. [CrossRef]
97. McGuffin, L.J.; Bryson, K.; Jones, D.T. The PSIPRED protein structure prediction server. Bioinformatics 2000,
16, 404–405. [CrossRef]
Molecules 2019, 24, 1196 26 of 26

98. Gaete-Eastman, C.; Morales-Quintana, L.; Herrera, R.; Moya-León, M.A. In-silico analysis of the structure and
binding site features of an α-expansin protein from mountain papaya fruit (VpEXPA2), through molecular
modeling, docking, and dynamics simulation studies. J. Mol. Model. 2015, 21, 1–12. [CrossRef]
99. Lomize, M.A.; Pogozheva, I.D.; Joo, H.; Mosberg, H.I.; Lomize, A.L. OPM database and PPM web server:
Resources for positioning of proteins in membranes. Nucleic Acids Res. 2012, 40, 370–376. [CrossRef]
100. Lomize, A.L.; Pogozheva, I.D.; Lomize, M.A.; Mosberg, H.I. Positioning of proteins in membranes:
A computational approach. Protein Sci. 2006, 15, 1318–1333. [CrossRef]
101. Lomize, A.L.; Pogozheva, I.D.; Lomize, M.A.; Mosberg, H.I. The role of hydrophobic interactions in
positioning of peripheral proteins in membranes. BMC Struct. Biol. 2007, 7, 1–30. [CrossRef]
102. Lomize, A.L.; Pogozheva, I.D.; Mosberg, H.I. Anisotropic solvent model of the lipid bilayer. 2. Energetics
of insertion of small molecules, peptides and proteins in membranes Andrei. J. Chem. Inf. Model. 2011, 51,
930–946. [CrossRef]
103. Søndergaard, C.R.; Olsson, M.H.M.; Rostkowski, M.; Jensen, J.H. Improved Treatment of Ligands and
Coupling Effects in Empirical Calculation and Rationalization of pKa Values. J. Chem. Theory Comput. 2011,
7, 2284–2295. [CrossRef]
104. Olsson, M.H.M.; Søndergaard, C.R.; Rostkowski, M.; Jensen, J.H. PROPKA3: Consistent Treatment of Internal
and Surface Residues in Empirical pKa Predictions. J. Chem. Theory Comput. 2011, 7, 525–537. [CrossRef]
105. Dolinsky, T.J.; Nielsen, J.E.; McCammon, J.A.; Baker, N.A. PDB2PQR: An automated pipeline for the setup of
Poisson-Boltzmann electrostatics calculations. Nucleic Acids Res. 2004, 32, 665–667. [CrossRef]
106. Berendsen, H.J.C.; van der Spoel, D.; van Drunen, R. GROMACS: A message-passing parallel molecular
dynamics implementation. Comput. Phys. Commun. 1995, 91, 43–56. [CrossRef]
107. Vanommeslaeghe, K.; Hatcher, E.; Acharya, C.; Kundu, S.; Zhong, S.; Shim, J.; Darian, E.; Guvench, O.;
Lopes, P.; Vorobyov, I.; et al. CHARMM general force field: A force field for drug-like molecules compatible
with the CHARMM all-atom additive biological force fields. J. Comput. Chem. 2010, 31, 671–690. [CrossRef]
108. Darden, T.; York, D.; Pedersen, L. Particle mesh Ewald: An N*log(N) method for Ewald sums in large
systems Tom. J. Chem. Phys. 1993, 98, 10089. [CrossRef]
109. Hess, B.; Bekker, H.; Berendsen, H.J.C.; Fraaije, J.G.E.M. LINCS: A linear constraint solver for molecular
simulations. J. Comput. Chem. 1997, 18, 1463–1472. [CrossRef]
110. Humphrey, W.; Dalke, A.; Schultern, K. VMD—Visual Molecular Dynamics. J. Mol. Graph. Model. 1996, 14,
33–38. [CrossRef]
111. O’Boyle, N.M.; Banck, M.; James, C.A.; Morley, C.; Vandermeersch, T.; Hutchison, G.R. Open Babel: An Open
chemical toolbox. J. Cheminform. 2011, 3, 1–14. [CrossRef]
112. Morris, G.M.; Ruth, H.; Lindstrom, W.; Sanner, M.F.; Belew, R.K.; Goodsell, D.S.; Olson, A.J. AutoDock4
and AutoDockTools4: Automated docking with selective receptor flexibility. J. Comput. Chem. 2009, 30,
2785–2791. [CrossRef]

Sample Availability: Samples of the compounds are not available from the authors.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).