Tuning the activity of known drugs via the introduction of halogen atoms, a case study of SERT ligands – Fluoxetine and fluvoxamine

Article abstract of DOI:10.1016/j.ejmech.2021.113533

The selective serotonin reuptake inhibitors (SSRIs), acting at the serotonin transporter (SERT), are one of the most widely prescribed antidepressant medications. All five approved SSRIs possess either fluorine or chlorine atoms, and there is a limited number of reports describing their analogs with heavier halogens, i.e., bromine and iodine. To elucidate the role of halogen atoms in the binding of SSRIs to SERT, we designed a series of 22 fluoxetine and fluvoxamine analogs substituted with fluorine, chlorine, bromine, and iodine atoms, differently arranged on the phenyl ring. The obtained biological activity data, supported by a thorough in silico binding mode analysis, allowed the identification of two partners for halogen bond interactions: the backbone carbonyl oxygen atoms of E493 and T497. Additionally, compounds with heavier halogen atoms were found to bind with the SERT via a distinctly different binding mode, a result not presented elsewhere. The subsequent analysis of the prepared XSAR sets showed that E493 and T497 participated in the largest number of formed halogen bonds. The XSAR library analysis led to the synthesis of two of the most active compounds (3,4-diCl-fluoxetine 42, SERT K_i = 5 nM and 3,4-diCl-fluvoxamine 46, SERT K_i = 9 nM, fluoxetine SERT K_i = 31 nM, fluvoxamine SERT K_i = 458 nM). We present an example of the successful use of a rational methodology to analyze binding and design more active compounds by halogen atom introduction. ‘XSAR library analysis’, a new tool in medicinal chemistry, was instrumental in identifying optimal halogen atom substitution.

Full text of DOI:10.1016/j.ejmech.2021.113533

European Journal of Medicinal Chemistry 220 (2021) 113533
Contents lists available at ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Tuning the activity of known drugs via the introduction of halogen
atoms, a case study of SERT ligands e Fluoxetine and fluvoxamine
ꢀ*
ꢀ
Jakub Staron , Wojciech Pietrus, Ryszard Bugno, Rafał Kurczab, Grzegorz Satała,
Dawid Warszycki, Tomasz Lenda, Anna Wantuch, Adam S. Hogendorf, Agata Hogendorf,
ꢀ
Beata Duszynska, Andrzej J. Bojarski
ꢀ
Maj Institute of Pharmacology, Polish Academy of Sciences, 12 Sme˛tna Street, 31-343, Krakow, Poland
a r t i c l e i n f o
a b s t r a c t
Article history:
The selective serotonin reuptake inhibitors (SSRIs), acting at the serotonin transporter (SERT), are one of
the most widely prescribed antidepressant medications. All five approved SSRIs possess either fluorine or
chlorine atoms, and there is a limited number of reports describing their analogs with heavier halogens,
i.e., bromine and iodine. To elucidate the role of halogen atoms in the binding of SSRIs to SERT, we
designed a series of 22 fluoxetine and fluvoxamine analogs substituted with fluorine, chlorine, bromine,
and iodine atoms, differently arranged on the phenyl ring. The obtained biological activity data, sup-
ported by a thorough in silico binding mode analysis, allowed the identification of two partners for
halogen bond interactions: the backbone carbonyl oxygen atoms of E493 and T497. Additionally, com-
pounds with heavier halogen atoms were found to bind with the SERT via a distinctly different binding
mode, a result not presented elsewhere. The subsequent analysis of the prepared XSAR sets showed that
E493 and T497 participated in the largest number of formed halogen bonds. The XSAR library analysis led
to the synthesis of two of the most active compounds (3,4-diCl-fluoxetine 42, SERT Ki ¼ 5 nM and 3,4-
diCl-fluvoxamine 46, SERT Ki ¼ 9 nM, fluoxetine SERT Ki ¼ 31 nM, fluvoxamine SERT Ki ¼ 458 nM). We
present an example of the successful use of a rational methodology to analyze binding and design more
active compounds by halogen atom introduction. ‘XSAR library analysis’, a new tool in medicinal
chemistry, was instrumental in identifying optimal halogen atom substitution.
Received 5 December 2020
Received in revised form
22 April 2021
Accepted 27 April 2021
Available online 14 May 2021
Keywords:
Serotonin
SERT
Halogen bond
XSAR
Halogen-hydrogen bond
© 2021 Elsevier Masson SAS. All rights reserved.
1. Introduction
compounds. They can form both hydrophobic and electrostatic
interactions with biological targets. The latter are mainly repre-
Serotonin transporter (SERT) is a monoamine transporter-type
protein composed of a 630 amino acid chain that forms 12 trans-
membrane domains [1,2]. It is located at the presynaptic axon ter-
minal and somatodendritic end of neurons [3] and regulates
serotonin transmission by removing serotonin from the extracel-
lular space [2]. Extensive research has shown that patients with a
functional polymorphism in the promoter region of the SERT gene
exhibit more depressive behavior in response to life events, thus
supporting the role of SERT in the regulation of mood disorders
[4,5].
sented by the halogen bond, which according to IUPAC definition is
“a net attractive interaction between an electrophilic region asso-
ciated with a halogen atom in a molecular entity and a nucleophilic
region in another, or the same, molecular entity” [6]. In particular,
the halogen bond is formed between the electron deficiency region,
called the
most often oxygen from the carbonyl group of the peptide chain
[7,8]. The size of the -hole strictly determines the strength of the
s-hole, on top of the halogen atom and a Lewis base,
s
halogen bond. It is greatest for iodine and smallest for chlorine.
Moreover, its size can be modified by the introduction of neigh-
boring electron-withdrawing groups. Several studies have shown
that additional fluorine atoms in halobenzenes greatly increase the
Halogen atoms are among the most popular substituents uti-
lized for the optimization of the properties of biologically active
size of the
s-hole [9e11]. The construction of a halogen bond
strictly defines its geometry. The optimal interaction is of near
linear nature, where the angle donor (D)-halogen (X)$$$acceptor
(A) is 160e180ꢀ and the distance between halogen and acceptor is
* Corresponding author.
E-mail address: jakubstaron@gmail.com (J. Staron).
ꢀ
https://doi.org/10.1016/j.ejmech.2021.113533
0223-5234/© 2021 Elsevier Masson SAS. All rights reserved.

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
much shorter than the sum of atomic van der Waals radii (80%) and
is equal to 3.06 Å for chlorine, 3.15 Å for bromine and 3.24 Å for
iodine [12].
derivative exhibited
a
slight increase in binding affinity
(Ki ¼ 14.4 nM [15]). Fluvoxamine substitutions with fluorine,
chlorine, and bromine can be found in the patent literature, but no
complimentary biological data allowing comparison are available
[16]. In turn, the SAR of the different halogenated modafinil analogs
can be found in a study by Okunola-Bakare et al., where substitu-
tion of chlorine with bromine slightly increased the binding affinity
to SERT (29.8 nM for chlorine derivative and 26.1 nM for bromine
derivative) [17]. Additionally, biological data for bromo-paroxetine
were also reported. Substitution of fluorine to bromine decreased
the binding affinity 13-fold (from Ki ¼ 0.08 nM, for paroxetine to
Ki ¼ 1.07 nM for Br-paroxetine); however, the new compound
maintained high activity [18]. A survey of the ChEMBL database
(release 25) revealed that out of 3529 stored SERT ligands (with
activity < 1000 nM), 145 contain the CF3 group, 768 contain fluo-
rine, 1207 contain chlorine, 101 contain bromine, and 48 contain
iodine. It is apparent that there are significantly fewer derivatives
with heavier halogens e bromine and iodine e than with fluorine,
despite their often greater affinity for SERT. The small number of
heavier halogens can be explained by a general tendency in the
pharmaceutical industry to avoid these atoms in drug chemical
structures.
All five approved selective serotonin reuptake inhibitors (SSRIs)
possess halogen atom(s) in their structure (Fig. 1); four possess
fluorine and one a chlorine atom. A limited number of biological
activity data of SSRI analogs with heavier halogens can be found,
and no consistent conclusion on the significance of halogen(s) can
be derived. Substitution of fluorine for chlorine in citalopram
caused a slight increase in binding affinity (IC50 ¼ 59 nM [13] to
IC50 ¼ 49 nM [13], respectively), whereas introduction of a bromine
decreased it (IC50 ¼ 121 nM [13]). In the case of fluoxetine (SERT
Ki ¼ 17 nM [14]), replacement of the trifluoromethyl group with
fluorine and chlorine caused
a decrease in binding affinity
(Ki ¼ 142 nM [14] and 638 nM [14], respectively), while an iodine
Even less information can be found regarding the analysis of the
role of halogen atoms in binding to SERT. The only remark about a
halogen binding pocket in SERT was published in 2009 by Zhou
et al. in a study investigating the specificity of serotonin transporter
inhibitors: sertraline and both isomers of fluoxetine [19]. Zhou et al.
found that SSRIs bind to the extracellular vestibule in the LeuT
protein (a bacterial leucine transporter) in a manner in which
halogen atoms from all compounds align in a pocket formed by L25,
G26, L29, R30, Y108, I111, and F253, which was called the halogen
binding pocket (HBP). Moreover, the HBP sequence homology
Fig. 1. Structures of FDA-approved selective serotonin reuptake inhibitors (data from
http://www.fda.gov/Drugs/DrugSafety/InformationbyDrugClass/ucm283587.htm).
Scheme 1. Synthesis of fluoxetine and fluvoxamine analogs. (a) PPh3, DIAD, THF, conc. H3PO4, rt; (b) Mg, Et2O, reflux; (c) NH2OH$HCl, NaOAc, EtOH, rt; (d) 2-chloroethylamine
hydrochloride, KOH, DMF, rt.
2

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
Table 1
Binding affinity values for synthesized fluoxetine and fluvoxamine analogs.
between LeuT, SERT, NET, and DAT is high. The only difference be-
tween NET (norepinephrine transporter), DAT (dopamine trans-
porter), and SERT is one amino acid, i.e., A77, A81, and G100,
respectively. This suggested that in human SERT, these SSRIs might
bind similarly. To test their hypothesis, Zhou et al. analyzed the
binding of SSRIs to SERT mutants with, among others, mutated
amino acids forming or neighboring the HBP. All reported muta-
tions caused a decrease in affinity.
In this study, we performed an analysis of halogen substitution
on two of the FDA-approved (Food and Drug Administration) SSRIs,
fluoxetine and fluvoxamine, to investigate the role of halogen
atoms in SERT ligand binding. In the synthesized fluoxetine and
fluvoxamine analogs, the trifluoromethyl group was substituted
with fluorine, chlorine, bromine, and iodine. Moreover, for chlorine,
bromine, and iodine derivatives, an additional fluorine atom was
introduced to alter the electron density distribution on the halogen
atom. Here, the results of a radioligand displacement binding
experiment performed for SERT and an analysis of the binding
mode based on induced fit docking and molecular dynamics are
presented.
Cmpd
R1
R2
R3
SERT Ki [nM]
2 (ST-213)
H
CF3
F
H
Cl
H
Br
I
Cl
Br
I
Cl
Br
CF3
F
Cl
Br
I
H
H
H
F
H
Cl
H
H
H
H
H
F
H
H
H
H
H
H
H
H
F
1458
31
543
349
55
66
24
14
52
41
23
31
19
458
1617
61
23
15
3
a (ST-129)
4 (ST-143)
5 (ST-215)
6 (ST-141)
7 (ST-214)
8 (ST-132)
9 (ST-137)
10 (ST-130)
11 (ST-133)
12 (ST-138)
13 (ST-131)
14 (ST-134)
33b (ST-110)
34 (ST-111)
35 (ST-112)
36 (ST-113)
37 (ST-201)
38 (ST-118)
39 (ST-121)
40 (ST-119)
41 (ST-120)
2. Chemistry
Fluoxetine, together with its analogs, was synthesized using a
two-step, one-pot procedure involving a Mitsunobu reaction of
tert-butyl N-(3-hydroxy-3-phenylpropyl)-N-methylcarbamate with
appropriate phenol and subsequent deprotection (Scheme 1). In
turn, the synthesis of fluvoxamine analogs consisted of three steps:
Grignard reaction of 1-bromo-4-methoxybutane with appropriate
nitrile, condensation with hydroxylamine, and finally, substitution
with 2-chloroethylamine (Scheme 1).
F
F
H
H
H
H
H
H
H
F
F
H
H
H
H
H
H
H
F
3. Results
Cl
Br
Cl
Br
202
137
62
F
H
H
3.1. Biological activity
F
21
Binding affinity constants for SERT were measured using a
radioligand displacement experiment (Table 1). Among the syn-
thesized analogs, the exchange of trifluoromethyl groups into
heavier halogens (chlorine, bromine, and iodine) generally causes
an increase in binding affinity (Fig. 2). Additionally, the simulta-
neous introduction of a fluorine atom results in more active com-
pounds (13, 14, 38e41). This is true for all compounds except the
2,4-substitution of fluoxetine, where the fluorine atom decreases
the binding affinity 1.6- and 1.3-fold (compounds 10 and 11,
respectively). Changing the CF3 group into a fluorine atom in both
cases (compounds 4 and 34) lowers the binding affinity 17- and 3.5-
fold, respectively. In turn, shifting the fluorine to the C-3 position in
fluoxetine analog 5 slightly increases (1.9-fold) the binding affinity.
A similar effect can be observed in the case of 3-chloro fluoxetine
analog 7, with a 1.2-fold increase.
Binding affinity, Ki, expressed as the average of at least two independent experi-
ments; the maximum S.D. did not exceed 10% (see Supplementary Information, page
20, Table 4).
a
Fluoxetine.
Fluvoxamine.
b
case of the fluoxetine complex, the phenyl ring lies in proximity to
F335 and F341. In turn, the oxygen atom of the fluvoxamines’
methoxybutyl chain forms a hydrogen bond with the hydroxyl
group of Y175. In the second mode, the compound is flipped, and
now the trifluoromethyl/halogen moiety is directed towards helix
TM10, pointing into residues E493 or T497. In this position, the
halogen atom (chlorine, bromine) was found to be engaged in a
halogen bond with the backbone oxygen of these residues. In turn,
the iodine atom of 9 positioned itself between helixes TM10 and
TM3, where it is surrounded by a net of interactions with three
residues: hydroxyl of T497, guanidine of R104, and hydroxyl of
T176. In most cases, the geometrical parameters of the formed
halogen bonds fit into the optimal boundaries (Fig. 4, Table 3). The
basic nitrogen atom still maintains a hydrogen bond with D98.
To further analyze the observed SAR, extensive molecular dy-
namics simulations were carried out for the most active derivatives
(8, 9, 35, 36, 37, 41). The starting poses were picked only from a
group representing the second mode of binding, because only there
halogen atoms may find a partner for interaction. For every
investigated analog, the halogen atom quickly forms and maintains
a directed interaction (numerical analysis of the geometrical pa-
rameters of halogen bonds is presented in Supplementary Infor-
mation). For chlorine and bromine derivatives (8, 35, 36, 41), the
3.2. Molecular modeling
An induced fit docking protocol (IFD) was used to investigate the
phenomenon behind the observed structure-activity relationship.
The protocol utilized the SERT model based on the SERT crystal
structure (PDB ID: 5I6X). Among all of the ligand poses that were
returned, each of the docked analogs was found in two different
binding modes (Fig. 3). In the first mode, the compound is located
analogously, as can be found in the published crystal structures of
SERT-SSRI complexes [20,21]. The trifluoromethyl/halogen moiety
is located in a lipophilic pocket between helixes TM3 and TM8,
surrounded by residues I172, A173, Y176, and L443, and the basic
nitrogen atom forms a hydrogen bond with aspartic acid D98. In the
3

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
values (41.8, 45.7, 45.3, and 42.8, respectively), five analogs 11, 14,
36, 38, and 41 moderate values (37.3, 36.7, 37.10, 39.5, and 37.6,
respectively) and another five analogs 8, 10, 13, 35 and 40 low
values (33.3, 33.4, 32.3, 33.5 and 33.4, respectively) of the s-hole
magnitude (a table with all of the data is shown in Supplementary
Information).
3.3. Using XSAR sets to evaluate the role of halogens in
LigandꢁProtein complexes
Most of the identified halogen bonds are formed either with the
T497 or E493 of SERT. The QPLD/GBSA (quantum-polarized ligand
docking/generalized Born and surface area solvation) docking
procedure and XSAR data were used to probe the SERT binding site
to determine the amino acids most frequently targeted by halogen
bonding. This particular methodology was previously used for a
case study of the 5-HT7 and D4R receptors [22,23]. It is based on
docking of a library of ligands to receptor proteins (either homology
models or crystal structures) and subsequent analysis of ligand-
receptor complexes with QPLD and GBSA. The obtained XSAR li-
brary for SERT was clustered into 79 sets and used to create an
XSAR matrix showing positive changes in SERT activity upon ex-
change of hydrogen to halogen (simplified, representative data
shown in Table 4, full matrix shown in Supplementary Informa-
tion). We found one primary (E493) and three secondary (T497,
S438, and S439) halogen bonding hot spots (a representative
binding mode of two selected sets of compounds is presented in
Fig. 5). The average increase in activity caused by the halogen bonds
is equal to 95.52, whereas the median activity increase is equal to
8.3. A remarkable 1885-fold change in activity was recorded for
set177, where two vicinal chlorine atoms formed two halogen
bonds with E493 and T497. Similarly, two chlorine atoms in set124
yielded a 1760-fold increase in activity.
Fig. 2. Structure-activity relationship for the synthesized fluoxetine and fluvoxamine
analogs. Binding affinity (Ki [nM]) is given above substituent symbols. Color codes
represent general binding affinity change relative to trifluoromethyl reference com-
pounds: red e large decrease, orange e moderate decrease, and green e no change or
increase.
3.4. Synthesis of 3,4-dichloro analogs of fluoxetine and fluvoxamine
The outstanding activity reported for compounds with two
vicinal chlorine atoms prompted us to synthetize 3,4-dichloro an-
alogs of fluoxetine and fluvoxamine. The synthesis of both com-
pounds was performed according to Scheme 1 (see Supplementary
Information). To our great surprise, both derivatives were found to
be more active than the best binding 4-iodo analog (Table 5). The
Xeffect of compound 42 is equal to 291 (compared to compound 2).
Compound 46 is 51-fold more active than the parent fluvoxamine
(compound 33). To visualize the halogen bond interactions
responsible for the increase in activity, induced fit docking was
performed (Fig. 6).
Fig. 3. Two ligand poses were obtained through IFD (compound 3 - fluoxetine as an
example). A cyan structure marks a pose found in the SERT crystal structures. In this
position, the trifluoromethyl group is located between helixes TM3 and TM8 and is
surrounded by residues I172, A173, Y176, and L443. A green structure marks a pose
that was a result of induced fit docking. Here, the trifluoromethyl group faces towards
helix TM10.
3.5. Quantum mechanics calculations
The intriguing binding affinity changes, observed in the syn-
thesized series of compounds, prompted us to perform a thorough
analysis of the nature of interactions formed by halogen atoms with
the transporter residues. Because of the extensiveness of the cal-
culations, only a series of mono-substituted fluoxetine analogs
(compounds 4, 6, 8, 9), supplemented by the dichloro analog
(compound 42), were selected for this task. First, the refined poses
obtained from the IFD were again refined using quantum me-
chanics methods. The obtained poses were subjected to the
extended transition state-natural orbitals for chemical valence
(ETS-NOCV) and Bader's quantum theory of atoms in molecules
(QTAIM) analyses [24e26]. As a result, decomposition of halogen
interaction energy was obtained (Table S5, Supplementary Infor-
mation). The QTAIM analysis allowed us to identify that, in addition
to halogen bonds, hydrogen bonds are also a large contributor to
interaction is formed with the backbone oxygen of G493 or T497. In
the case of iodine derivative 37, the halogen bond is also formed
with the backbone oxygen of G493. In turn, the iodine derivative of
fluoxetine 9 forms a halogen bond with the backbone oxygen of
T497 and guanidine nitrogen of R104. To further investigate the
correlation between the type of halogen atom and the strength of
ligand-receptor interactions, the magnitude of the
s-hole was
calculated (Table 2). The highest value is represented by compound
12 (45.7 kcal/mol), and the lowest value is represented by com-
pound 6 (23.5 kcal/mol). Four analogs 9, 12, 37, and 39 exhibit high
4

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
Fig. 4. Halogen bonds that were identified through the induced fit docking experiment. Next, to each figure, a 2D heat map representation of a halogen interaction is given, and the
red rectangle marks the best geometric parameters of the interaction, i.e., 140ꢀ < angle <180ꢀ and 3 Å< distance <4 Å. (A) The chlorine atom of 35 forms a halogen bond with a
backbone oxygen of T497, the distance X … O ¼ 3.12 Å, the angle CeCl/O ¼ 172.3ꢀ; (B) the bromine atom of 8 forms a halogen bond with a backbone oxygen of G493, the distance X
… O ¼ 3.09 Å, the angle CeBr/O ¼ 172.3ꢀ; (C) the bromine atom of 36 forms a halogen bond with a backbone oxygen of G493, the distance X … O ¼ 3.15 Å, the angle
CeBr/O ¼ 172.3ꢀ. The chlorine(bromine)ꢁoxygen theoretical interaction sphere illustrates the projected qualities of the formed LꢁR halogen bonds.
Table 2
Calculated values of the magnitude of
s
-hole [kcal/mol] (
s
-hole mgnd).
Fluoxetine analogs
Cmpd
6
10
13
8
11
14
9
12
Halogen substitution pattern
4Cl
29.7
4Cl2F
33.4
4Cl3F
32.3
4Br
33.5
4Br2F
37.3
4Br3F
36.7
4I
41.8
4I2F
45.7
s-hole mgnd [kcal/mol]
Fluvoxamine analogs
Cmpd
35
38
40
36
39
41
37
Halogen substitution pattern
4Cl
33.5
4Cl2F
39.5
4Cl3F
33.4
4Br
37.1
4Br2F
42.8
4Br3F
37.6
4I
45.3
s-hole mgnd [kcal/mol]
5

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
Table 3
Information pages 20e24). For the mono-chlorine derivative 6,
halogen bonds share 40% of the total interaction energy. This
contribution increases slightly with the size of the halogen atom,
reaching 43% and 54% for the bromine 8 and iodine 9 analogs,
respectively. In turn, for the most active vicinal di-Cl analog 42, the
hydrogen bonds are responsible for 61% of the total interaction
energy Apart from the above, we also observed the presence of X-
HBD (halogen e hydrogen bond donor interaction, expressed also
as HBeXB e HB enhanced XB [27]), formed between halogen atoms
and hydroxyl group of threonine T497 or tyrosine Y175. Such in-
teractions have already been recognized as being highly favorable
to the ligand-protein interactions, through the enhancement of
halogen bonds energies [28,29]. In our study, for the most active
compound 42, this interaction has the highest overall contribution
to the binding energy (ꢁ5.0 and ꢁ1.7 kcal/mol, formed with a side
chain of T497, simultaneously). Meanwhile, in the case of an iodine
analog 9, XB and X-HBD interactions have equal contribution in the
Average geometric (distance and angle) parameters of halogen bonds formed be-
tween halogens of investigated structures and amino acids of the SERT, measured
during 100 ns of performed molecular dynamics. The most optimal parameters of
halogen bonds were described as distance <4.25 Å and angle >135ꢀ for iodine;
distance <3.75 Å and angle >145ꢀ for bromine; distance <3.75 Å and angle >150ꢀ for
chlorine [11].
Compound X atom Amino acid Distance X … A [Å] Angle C-X … A [ꢀ]
8
9
Br
I
T497
T497
R104
T497
T497
E493
E493
3.74 0.41
4.14 0.38
3.76 0.37
4.43 0.64
3.97 0.75
4.95 0.97
4.17 0.87
161.78 12.88
137.88 9.62
87.34 9.19
148.08 12.03
136.34 30.24
121.70 11.33
134.63 28.08
35
36
37
41
Cl
Br
I
Br
the interaction with the transporter (a representative data for
compound 42, shown in Fig. 7, full results in Supplementary
Table 4
A representative simplified XSAR matrix (full matrix shown in Supplementary Information) generated for 79 sets showing positive changes in SERT activity upon exchange of
hydrogen with halogens: chlorine, bromine or iodine. Values depicted under aa. represent scoring functions that rate the quality of interaction; a value equal to 1 correlates
with the most favorable and 0.25 for the least favorable interaction. Orange color states for a halogen bond formed with a backbone carbonyl oxygen, green for a halogen bond
formed with a heteroatom at side chain of an amino acid and blue for a hybrid halogen-hydrogen bond formed either with a backbone or side chain. The sum of all scoring
values is equal to 104.5. At the bottom of the table, this value is used to rate the frequency of interactions with particular amino acids. Most frequently, halogen bonds are
formed with E493 (primary halogen bonds). The average activity increase for hydrophobic interactions is 4.81, for the halogen bond interaction is 95.52, whereas for halogen-
hydrogen bonds, it is 141.45.
6

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
Fig. 5. The binding modes of the two selected major increases in the biological activity of SERT ligands upon the introduction of a halogen atom. In the upper example, walking the
chlorine around the phenyl ring (XSAR set211) increased the IC50 value 131-fold (compared with the derivative with no substituents) through the formation of a halogen bond with
backbone oxygen of E493. The lower part of the figure represents a compound possessing two chlorine atoms in the meta and para positions (XSAR set937) that form two halogen
bonds, with E493 and T497. This halogen arrangement causes a 666-fold increase in activity at SERT.
Table 5
4. Discussion
Binding affinity values for synthesized 3,4-dichloro analogs of fluoxetine and
fluvoxamine.
The analysis of the data collected for the synthesized com-
Cmpd
Structure
SERT Ki [nM]
pounds shows a strong correlation between the type of introduced
halogen atom and SERT binding affinity. Chlorine analogs are
among the least active compounds, while iodine analogs are the
most active ones (Fig. 2). The influence of an additional fluorine
atom placed in the 2- or 3- position relative to the halogen atom is
more complex. From these two positions, only the ortho mutual
orientation is favorable. In the case of fluoxetine chlorine analog 6
(SERT Ki ¼ 55 nM), additional fluorine in the meta position (com-
pound 10, SERT Ki ¼ 52 nM) has a negligible influence on the
binding affinity. However, switching the fluorine to the 2-position
42
5
46
9
(compound 13) increases the binding affinity 1.77-fold.
A
Binding affinity, Ki, expressed as the average of at least two independent experi-
ments; the maximum S.D. did not exceed 10% (see Supplementary Information, page
20, Table 4).
different relationship was observed in the series of fluvoxamine
analogs. Here, all chlorine derivatives are more active than the
parent compound, with the 3-fluorine analog 38 the least active out
of three (SERT Ki ¼ 202 nM). The other two compounds, mono-
chlorine and ortho-fluorine analogs, exhibit identical binding af-
finity and are 7.4-fold more active (SERT Ki ¼ 61 nM and 62 nM,
respectively) than fluvoxamine (SERT Ki ¼ 458 nM). Interestingly,
the introduction of bromine brings both fluoxetine and fluvox-
amine analogs to an identical level of activity (average
L-P complex stabilization (ꢁ1.4, ꢁ1.3 kcal/mol, respectively). The
HBs with methyl group also seem to significantly impact the
binding because their energy ranges from ꢁ0.31 to ꢁ3.54 kcal/mol
(the highest value is observed for dichloro analog). In turn, ETS-
NOCV analysis showed that dispersion energy takes a high share
of the total stabilization energy (Table S5, Supplementary Infor-
mation), especially when HB is observed (XeHC HB takes approx-
imately 30% of the total attractive energy for Br analog 8).
Compared to that, the halogen bonds have a very low dispersion
energy component (~3%), while for X-HBD, this energy oscillates
between XB and HB (from 9% to 26%).
Ki ¼ 21.75
1.92 nM). Similar to the chlorine analogs, the only
difference can be observed among the meta-fluorine derivatives (11
and 39); in both sets of derivatives, they are the least active
(fluoxetine analog 11 SERT Ki ¼ 41 nM, fluvoxamine analog 39 SERT
Ki ¼ 137 nM). Additionally, the incorporation of iodine into the
structure of these SSRIs produces identically active compounds
(fluoxetine analog 9 SERT Ki ¼ 14 nM, fluvoxamine analog 37 SERT
7

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
Fig. 6. Ligand-receptor complexes obtained through an induced fit docking of two 3,4-dichloro analogs of fluoxetine (compounds 42, left, cyan) and fluvoxamine (compound 46,
right, magenta). Both compounds interact with the receptor via double halogen bonds formed with backbone carbonyl oxygen atoms of glutamic acid E492 and threonine T497.
Such interactions caused a substantial increase in binding affinity for the receptor. Compound 42 is 291-fold more active than the analog without any substituent (compound 3), and
it is the most active SERT ligand presented in this manuscript. Compound 46 is 51-fold more active than the parent CF3-containing fluvoxamine (compound 33), and it is the second
most active compound in this manuscript.
Fig. 7. The identified interactions of the halogen atom with the transporter residues, calculated using the QTAIM analysis, exemplified by compound 42. On the image, individual
interactions between halogen and the transporter are depicted as dotted lines. Next to each line, a value representing the share of total interaction energy is presented. The
hydrogen bonds account for 70% of total interaction energy, while the halogen bonds for 30%.
Ki ¼ 15 nM). As with the previous examples, 3-fluorine in 12 slightly
lowers (1.64-fold) the binding affinity compared to a mono-iodine
analog 9. The average geometrical parameters (Table 3) of the
halogen bonds during molecular dynamics do not fit precisely into
the boundaries of the optimum values. Instead, they oscillate on the
borders of the estimated optimum range. These results do not mean
that above this range this interaction is not present, rather its
strength is lowered by the unfavorable geometries. The increased
affinity of compounds with heavier halogens leads to the conclu-
magnitude equal to 42.8 kcal/mol, similar to the most active iodine
derivatives, is 9.8-fold less active. The additional fluorine atom,
despite increasing the magnitude of the s-hole, must be involved in
other interactions that impair the binding of a ligand to the
receptor.
The performed QTAIM analysis showed that the most important
interaction in the analyzed complexes is represented by the X-HBD
interactions. Those interactions seems to be a hybrid of XB and HB
which results in a stronger interaction than the individual ones. The
most optimum geometrical orientation for X-HBD is found for the
C-X … HBD angle equal to 90ꢀ [30]. A query of the PDB repository
allowed us to identify examples that incorporate non-halogenated
compounds (2Q6H, 2QU3, 2WHO), for which it has been experi-
mentally shown that adding a Cl or Br atom to the ligand signifi-
cantly improves the binding affinity (even 200-fold), by creating X-
sion that the size of the
affinity changes. Indeed, calculated values of the magnitude of the
-hole correlate well with the activity of mono-substituted de-
s-hole may be the factor responsible for the
s
rivatives. The introduction of a fluorine atom increases this value in
most cases, which surprisingly does not translate to a higher
binding affinity. For example, compound 39 with
a s-hole
8

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
HBD [28]. The QTAIM analysis showed that X-HBD interaction be-
tween Cl and hydroxyl group of T497 has the highest values of
energy stabilization (ꢁ1.7, ꢁ5.0 kcal/mol, for 7 and 42, respec-
tively). This feature of the chlorine atom translate to the formation
of a very strong X-HBD interactions, which were found to be
responsible for the high binding affinity of 3,4-diCl analogs. In spite
of the complex nature of X-HBD, pure HB and XB can also be
observed in the analyzed L-P complexes. In that case, importance of
XBs increase along with the halogen atom size, with simultaneous
decreased contribution of HBs (only 6% for iodine analog 9)
(Table S5, Supplementary Information). For all of the discussed
interactions the dispersion energy is a vital component, however its
share is strongly related to the type of the interaction. It is worth to
note at this point, that the obtained QM refined poses constitute of
the most optimum ligand-receptor mutual orientation and are
probably poorly populated in the real time ligand-receptor in-
teractions. Despite that, this type of analysis still allows to deeply
investigate the binding of a ligand to the receptor.
T497. This type of interaction was found to be a hybrid of XB and HB
interactions, with the latter accounting for a higher share of the
interaction energy. Subsequently, the use of an XSAR analysis
allowed to translate this to a general rule governing the binding of
SERT ligands. Especially interesting was the observation that SERT
ligands with heavier halogens bind via a markedly different,
“switched” binding mode than the parent trifluoromethyl analogs.
To the best of our knowledge, no paper to date has exclusively re-
ported two distinct binding modes for SERT ligands. Finally, the
given hypothesis of halogen bond type interactions with the re-
ceptor was confirmed by the experimental measurement of the
affinity constant for 3,4-dichloro analogs of fluoxetine and fluvox-
amine, which were the most active among all of the synthesized
compounds. The presented manuscript states as an example of the
successful use of molecular modeling for the analysis and optimi-
zation of compound activity. The utilized tool, XSAR analysis, pre-
viously used for the 5-HT7R [22] and D4R ligands [22,23], proved to
be universal and applicable for a broader type of biological target.
We anticipate that such a methodology could be successfully uti-
lized for the structure optimization of any halogen-containing
biologically active compound.
All of the above observations lead to the hypothesis that starting
from bromine, the halogen-receptor interaction becomes a domi-
nant factor influencing the activity of these two SSRIs. Chlorine
analogs 6 and 35 also possess similar binding affinity (SERT
Ki ¼ 55 nM and Ki ¼ 61 nM, respectively), but the effect is less
profound. The trifluoromethyl group interactions with the protein
are weak and indirect; thus, structural differences of fluoxetine and
Declaration of competing interest
The authors declare that they have no known competing
financial interests or personal relationships that could have
appeared to influence the work reported in this paper.
fluvoxamine produce
a very distinct binding affinity (SERT
Ki ¼ 31 nM and Ki ¼ 458 nM, respectively). The ability of heavier
halogen atoms to form a direct interaction causes different chem-
ical scaffolds of these compounds to lose their decisive impact on
the activity. The molecular modeling experiments showed two
distinct binding modes (Fig. 3). Here, the increase in activity
brought by the introduction of a halogen atom can be explained on
the ground of the formed halogen atom interactions, which are
possible only in the alternative binding mode (Fig. 3). Most
frequently, these interactions were formed with either T497 or
E493. To date, only one halogen bond in the SERT complex has been
reported. In 2019, Abramyan et al. showed a complex of bromo
paroxetine with SERT, where the bromine atom pointed in the di-
rection of the E493 and T497 backbone oxygens. However, no
further analysis was performed [31]. Therefore, we performed an
XSAR analysis, which showed explicitly that E493 was the most
frequently targeted amino acid in terms of halogen bonding with
SERT. Moreover, the proximity of T497 and E493 allowed for the
formation of two strong interactions with two vicinal halogen
atoms. Such an optimal arrangement in some cases resulted in a
spectacular thousand-fold increase in biological activity (Table 4).
This finding supported our hypothesis for the alternative binding
mode of SERT ligands bearing heavier halogen atoms. Finally, two
synthesized 3,4-dichloro analogs of fluoxetine and fluvoxamine
confirmed the theoretical assumption and showed a marked in-
crease in activity. Fluoxetine analog 42 was 3-fold more active and
fluvoxamine analog 46 was 1.7-fold more active than the best-
binding 4-iodo analogs.
Acknowledgments
The study was partially supported by grant SONATA UMO-2016/
21/D/NZ7/00620 from the Polish National Science Centre and
POWR.03.02.00e00-I013/16 from the European Union. The study
was performed with the use of PLGrid infrastructure. WP ac-
knowledges the support of InterDokMed project no.
POWR.03.02.00e00-I013.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.ejmech.2021.113533.
Methodology
Structureeactivity relationship datasets for halogenated analogs
An algorithm to find all pairs containing halogenated and cor-
responding unsubstituted structures (called the XSAR library) was
developed and used for the 5-HT7 and D4 targets in our previous
studies [22]. To describe the influence of halogenation on the bio-
logical activity of the unsubstituted (parent) molecule, the Xeffect
parameter was calculated as an activity (extracted from the
ChEMBL database during the generation of the XSAR library) ratio
of the parent compound to its halogenated derivative (an Xeffect
between 0 and 1 denotes a decrease in the activity upon haloge-
nation, and an Xeffect greater than 1 means the fold of activity
increased after halogen substitution).
5. Conclusions
In the present study, a series of fluoxetine and fluvoxamine
analogs with varying patterns of halogen substitution were syn-
thesized, and their biological activities were measured. Among the
synthesized initial series of compounds, a classical order of
increasing activity from chlorine to iodine was observed. Intro-
duced additional fluorine atoms did not have a substantial influ-
ence on the affinity for the receptor. Subsequent thorough in silico
analysis of the binding modes indicated the possibility of the for-
mation favorable X-HBD interactions with the residues E493 and
Identification of halogen bonding hot spots for SERT
Privileged amino acids (i.e., hot spots) for halogen bonding were
identified using a procedure including the following steps: clus-
tering the halogenated analogs representing each XSAR set, using
the centroids of the clusters to tune the SERT binding site by an
induced-fit docking procedure, QPLD docking of the XSAR library to
9

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
SERT conformations, and determination of the number of halogen
bonding interactions with the side chains and carbonyl oxygen
atoms of amino acids in the docking poses.
Gaussian G16 at the M06-2X/def2-tzvp [43,49] level of theory. The
energy of non-covalent bonds detected in crystal structures was
calculated with the Espinosa equation [50]:
1
2
Induced fit docking
Eint
¼
vðrÞ
The structure of the ts3 human serotonin transporter in complex
with paroxetine (PDB code 5I6X) was retrieved from the Protein
Data Bank [20]. Wild-type human SERT is unstable in detergent
solutions during crystallization, so all crystal structures in PDB re-
positories consist of thermostable mutants. The ts3 mutant of the
human serotonin transporter contains three thermostabilizing
mutations, namely, I291A, T439S, and Y110A. To obtain reliable
results from docking and molecular dynamics, the native amino
acid sequence was restored using Schrodinger software.
where Eint is the energy of interatomic interaction (a.u.) and v(r) is
kinetic energy in the bond critical point (BCP).
ETS-NOCV
To fully characterize obtained interactions we used an approach
for energy decomposition analysis. The bonding analysis was per-
formed using the ETS-NOCV approach the extended transition state
(ETS) method [51] with the natural orbitals for the chemical valence
(NOCV) scheme [52e54]. In this approach the total energy of
The three-dimensional structures of ligands were prepared us-
ing LigPrep v3.6 [32], and the appropriate ionization states at
bonding between the interacting molecules (
following contributions:
DEint) is divided into
pH ¼ 7.0
0.5 were assigned using Epik v3.4 [33]. The Protein
Preparation Wizard was used to assign the bond orders and
appropriate amino acid ionization states and to check for steric
clashes. All ligands were docked using the induced fit docking (IFD)
[34,35] protocol with XP with an OPLS3e force field. The LꢁR
complexes selected in the IFD procedure were next optimized us-
ing the QM/MM approach with QSite [34,36]. The QM area con-
taining ligand and the conserved amino acid side chain was
described by a combination of DFT hybrid functional B3LYP and
LACVP* basis set, while the rest of the system was optimized using
the OPLS2005 force field.
D
Eint
¼
D
Edist
þ
D
Eel
þ
D
EPauli
þ
DEorb
where
from their equilibrium geometry to the structure they will take up
in the complex, Eel is corresponding to the electrostatic interac-
tion between the two fragments in the supermolecule geometry,
EPauli is the repulsive interaction between occupied orbitals of the
two fragments, and the orbital interaction term and Eorb is rep-
DEdist is energy required to promote the separated fragments
D
D
D
resentation of the stabilizing component due to the final orbital
relaxation. All calculations were performed using the using the
Amsterdam Density Functional (ADF) program [55e58], imple-
menting the ETS-NOCV scheme. The BLYP-D3 [59] [16] functional
Molecular dynamics
and a standard double-z STO basis containing one set of polariza-
tion functions was adopted for all the electrons (TZP), were used in
calculations.
Molecular dynamics (MD) simulations were performed using
€
Schrodinger Desmond software [37]. Each ligandꢁreceptor com-
plex, optimized in the QM/MM procedure, was immersed into a
POPC (300 K) membrane bilayer, whose position was calculated
using the System Builder interface. The system was solvated by
water molecules described by the TIP4P potential, and the OPLS3e
[38] force field parameters were used for all atoms. NaCl (0.15 M)
was added to mimic the ionic strength inside the cell. Molecular
simulations for 100 ns (recording frames 10 ps and step of 2 fs)
using the NPAT ensemble class (constant normal pressure, tem-
perature, and lateral surface area of membranes) and OPLS3e [39]
force field were calculated for each system. Based on the obtained
trajectories, the mean geometrical distances between amino acids
and ligands were calculated using Simulation Event Analysis tools
Plotting interaction spheres for halogen bonding
To visualize (plotting interaction spheres) the possible contri-
bution of halogen bonding to LꢁR complexes, the halogen bonding
Web server was used (accessed June 20, 2020, http://www.
halogenbonding.com/) [60].
Radioligand binding replacement experiment
Cell culture and preparation of cell membranes for radioligand
binding assays
€
in Maestro Schrodinger Suit.
The HEK293 human serotonin transporter cell line (Perki-
nElmer) was maintained at 37 ꢀC in a humidified atmosphere with
5% CO2 and grown in Dulbecco's Modifier Eagle Medium containing
Magnitude of s-hole
10% dialyzed fetal bovine serum and 500 mg/mL G418 sulfate. For
MultiWFN [40,41] software was used to calculate the size of the
sigma hole for selected fluvoxamine and fluoxetine derivatives.
First, DFT [42] structure optimization was performed with the
Gaussian 16 package at the M06-2X [43]/def2-qZVP [44] level of
theory with the polarizable continuum model (PCM) [45,46]
(solvent ¼ water). The obtained wave functions were used to
calculate the values of maximum electrostatic potential over the
isodensity surface with MultiWFN as an approach to quantifying
membrane preparation, cells were subcultured in 150 cm2 flasks,
grown to 90% confluence, washed twice with phosphate buffered
saline (PBS) prewarmed to 37 ꢀC and pelleted by centrifugation
(200ꢂg) in PBS containing 0.1 mM EDTA and 1 mM dithiothreitol.
Prior to membrane preparation, pellets were stored at ꢁ80 ꢀC.
Radioligand binding assays
the
s
-holes.
Cell pellets were thawed and homogenized in 10 vol of assay
buffer using an Ultra Turrax tissue homogenizer and centrifuged
twice at 35,000ꢂg for 20 min at 4 ꢀC. The composition of the assay
buffer was as follows: 50 mM Tris-HCl pH 7.4, 120 mM NaCl, and
QTAIM analysis
For chosen complexes, the ligands and all residues in the range
of 4 Å from halogen atom were extracted and the QTAIM [47] cal-
culations (Quantum Theory of Atoms in Molecule) were performed.
The electron density topological analysis was carried out with the
AIMAll [48] program, based on electron density calculated with
5 mM KCl. The assay was incubated in a total volume of 200 ml in
96-well microtiter plates for 0.5 h at 27 ꢀC. The process of equili-
bration was terminated by rapid filtration through GF/C Unifilter
plates (PerkinElmer) with a FilterMate Unifilter 96-Harvester
10

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
(PerkinElmer). The radioactivity bound to the filters was quantified
on a Microbeta TopCount instrument (PerkinElmer). For competi-
tive inhibition studies, the assay samples contained 3 nM [3H]-
citalopram (74.5 Ci/mmol). Non-specific binding was determined
References
[1] K.P. Lesch, B.L. Wolozin, D.L. Murphy, P. Riederer, Isolation of a CDNA encoding
the human brain serotonin transporter, J. Neurochem. 60 (1993) 2319e2322.
[2] S. Montanez, L.C. Daws, G.G. Gould, A. Frazer, Serotonin (5-HT) transporter
(SERT) function after graded destruction of serotonergic neurons,
J. Neurochem. 87 (2003) 861e867.
[3] S.M. Stahl, C. Lee-Zimmetman, S. Cartwright, D.A. Morrissette, Serotonergic
drugs for depression and beyond, Curr. Drug Targets 14 (2013) 578e585.
[4] A. Caspi, K. Sugden, T.E. Moffitt, A. Taylor, I.W. Craig, H. Harrington, Influence
of life stress on depression: moderation by a polymorphism in the 5-HTT
gene, Science 301 (2003) 386e389 (80-. ).
with 10 mM imipramine. Each compound was tested in triplicate at
7 concentrations (10ꢁ10-10ꢁ4 M). The inhibition constants (Ki) were
calculated from the Cheng-Prusoff equation [61]. For all binding
assays, the results were expressed as the means of at least two
separate experiments.
[5] T.C. Eley, K. Sugden, A. Corsico, A.M. Gregory, P. Sham, P. McGuffin, Gene-
enviroment interaction analysis of serotonin system with adolescent
depression, Mol. Psychiatr. 9 (2004) 908e915.
Synthesis
[6] G.R. Desiraju, P.S. Ho, L. Kloo, A.C. Legon, R. Marquardt, P. Metrangolo,
P. Politzer, G. Resnati, K. Rissanen, Definition of the halogen bond (IUPAC
recommendations 2013), Pure Appl. Chem. 85 (8) (2013) 1711e1713.
[7] V. Andrea, H. P, The role of halogen bonding in inhibitor recognition and
binding by protein kinases, Curr. Top. Med. Chem. 7 (14) (2007) 1336e1348.
[8] L.A. Hardegger, B. Kuhn, B. Spinnler, L. Anselm, R. Ecabert, M. Stihle, B. Gsell,
R. Thoma, J. Diez, J. Benz, et al., Systematic investigation of halogen bonding in
protein-ligand interactions, Angew. Chem. Int. Ed. 50 (1) (2011) 314e318.
A general procedure for the synthesis of compounds 2e14
Tert-butyl N-(3-hydroxy-3-phenylpropyl)-N-methylcarbamate
(0.0026 mol), appropriate phenol (1 eq.), Ph3P (1.2 eq.) were added
to a round bottom flask and dissolved in anhydrous THF (50 mL,
distilled over LiAlH4) and cooled on ice. Next, DIAD (1.2 eq.) was
added dropwise. The reaction mixture was allowed to warm to
room temperature and stirred overnight. The solvent was evapo-
rated, and conc. H3PO4 (5 mL) was added; an evolution of gases was
observed. The mixture was stirred for 3 h at room temperature,
after which water was added (100 mL), and the mixture was
extracted with EtOAc (3 ꢂ 50 mL). The aqueous layer was basified
with 15% NaOH and extracted with EtOAc (3 ꢂ 50 mL). Combined
organic extracts were washed with brine and dried over anhydrous
MgSO4. The product was purified using column chromatography in
silica gel with EtOAc:MeOH (9:1).
ꢁ
ꢀ
[9] K.E. Riley, J.S. Murray, R. Jan, R.J. Sola, M.C. Concha, F.M. Ramos, P. Politzer,
ꢁ
ꢀꢁ
ꢀ
J. Fanfrlík, J. Rezac, R.J. Sola, et al., Halogen bond tunability II: the varying roles
of electrostatic and dispersion contributions to attraction in halogen bonds,
J. Mol. Model. 19 (11) (2013) 4651e4659.
[10] J. Fanfrlik, M. Kolar, M. Kamlar, D. Hurny, F.X. Ruiz, A. Cousido-Siah,
A. Mitschler, J. Rezac, E. Munusamy, M. Lepsik, et al., Modulation of aldose
reductase inhibition by halogen bond tuning, ACS Chem. Biol.
8 (2013)
2484e2492.
[11] R. Wilcken, M.O. Zimmermann, A. Lange, A.C. Joerger, F.M. Boeckler, Principles
and applications of halogen bonding in medicinal chemistry and chemical
biology, J. Med. Chem. 56 (2013) 1363e1388.
[12] P. Auffinger, F.A. Hays, E. Westhof, P.S. Ho, Halogen bonds in biological mol-
ecules, Proc. Natl. Acad. Sci. U. S. A. 101 (48) (2004) 16789e16794.
[13] J. Andersen, L. Olsen, K.B. Hansen, O. Taboureau, F.S. Jørgensen, A. Marie,
B. Bang-andersen, J. Egebjerg, K. Strømgaard, A.S. Kristensen, Mutational
mapping and modeling of the binding site for ( S ) -citalopram in the human
serotonin transporter, J. Biol. Chem. 285 (3) (2010) 2051e2063.
[14] D.T. Wong, F.P. Bymaster, E.A. Engleman, Prozac (fluoxetine, lilly 110140), the
first selective serotonin uptake inhibitor and and antidepressant drug: twenty
years since its first publication, Life Sci. 57 (5) (1995) 411e441.
[15] S. Chumpradit, M.P. Kung, C. Panyachotipun, V. Prapansiri, C. Foulon,
B.P. Brooks, S.A. Szabo, S. Tejani-Butt, A. Frazer, H.F. Kung, Iodinated tomox-
etine derivatives as selective ligands for serotonin and norepinephrine uptake
sites, J. Med. Chem. 35 (23) (1992) 4492e4497.
[16] H.B.A. Welle, V. Claassen, Aminoethyl Oximes Having Anti-depressive Activity,
Patent US4086361A, 1978.
[17] O.M. Okunola-Bakare, J. Cao, T. Kopajtic, J.L. Katz, C.J. Loland, L. Shi,
A.H. Newman, Elucidation of structural elements for selectivity across
monoamine transporters: novel 2-[(Diphenylmethyl)Sulfinyl]Acetamide
(modafinil) analogues, J. Med. Chem. 57 (3) (2014) 1000e1013.
[18] R.D. Slack, A.M. Abramyan, H. Tang, S. Meena, B.A. Davis, A. Bonifazi,
J.B. Giancola, J.R. Deschamps, S. Naing, H. Yano, et al., A novel bromine-
containing paroxetine analogue provides mechanistic clues for binding am-
biguity at the central primary binding site of the serotonin transporter, ACS
Chem. Neurosci. 10 (9) (2019) 3946e3952.
[19] Z. Zhou, J. Zhen, N.K. Karpowich, C.J. Law, M.E.A. Reith, D. Wang, Antide-
pressant specificity of serotonin transporter suggested by three LeuT e SSRI
structures, Nat. Struct. Mol. Biol. 16 (6) (2009) 652e658.
A general procedure for the synthesis of compounds 15e24
Magnesium turnings (1.1 eq., 0.47 g) were grated in a mortar and
placed in a round bottom flask under an Ar atmosphere. Anhydrous
Et2O (30 mL, distilled over LiAlH4) was added with subsequent
addition of ethylene bromide (0.1 eq., 0.155 mL). After 15 min at
room temperature,1-bromo-4-methoxybutane (3 g, 0.018 mol) was
added portionwise. After all of the bromide reacted with the
magnesium, the reaction mixture was cooled on ice, and a solution
of appropriate benzonitrile (0.6 eq., 0.0107 mol) in anhydrous Et2O
(50 mL) was added portionwise. The reaction mixture was refluxed
for 8 h. The reaction was quenched with ice-water and acidified
with 6 N HCl. The organic layer was separated, and the aqueous
phase was extracted with EtOAc (3 ꢂ 50 mL). The combined organic
layers were washed with brine (50 mL), dried over MgSO4, and
evaporated. The crude product was purified by column chroma-
tography on silica gel hexane:EtOAc (6:1) to give the title com-
pound as an oil or solid.
[20] J.A. Coleman, E.M. Green, E. Gouaux, X-ray structures and mechanism of the
human serotonin transporter, Nature 532 (7599) (2016) 334e339.
[21] J.A. Coleman, E. Gouaux, Structural basis for recognition of diverse antide-
pressants by the human serotonin transporter, Nat. Struct. Mol. Biol. 25 (2)
(2018) 170e175.
A general procedure for the synthesis of compounds 24e32
To a solution of ketone (1 eq.) in EtOH (5 mL) hydroxylamine
hydrochloride (2 eq.) and NaOAc (2 eq.) were added. After stirring
at room temperature for 3 h, the reaction mixture was quenched
with water (10 mL) and extracted with DCM (3 ꢂ 30 mL). The
combined organic phases were dried over MgSO4 and evaporated
to dryness. The residue was purified by column chromatography on
silica gel with hexane:EtOAc (5:1) to give the title compound.
[22] R. Kurczab, V. Canale, G. Satała, P. Zajdel, A.J. Bojarski, Amino acid hot spots of
halogen bonding: a combined theoretical and experimental case study of the
5-HT7 receptor, J. Med. Chem. 61 (19) (2018) 8717e8733.
ꢀ
[23] R. Kurczab, K. Kucwaj-Brysz, P. Sliwa, The significance of halogen bonding in
ligandereceptor interactions: the lesson learned from molecular dynamic
simulations of the D4 receptor, Molecules 25 (1) (2019) 91.
[24] R.F.W. Bader, Atoms in molecules, Acc. Chem. Res. 18 (1) (1985) 9e15.
[25] R.F.W. Bader, Atoms in Molecules. A Quantum Theory, Clarendon Press, Ox-
ford, UK, 1990.
A general procedure for the synthesis of compounds 33e41
An appropriate oxyme (1 mmol), 2-chloroethanamine hydro-
chloride (0.7 mmol), and KOH (1.5 mmol) were added to dry DMF
(6 mL) at 10 ꢀC. The reaction mixture was stirred overnight at room
temperature. Next, the reaction mixture was concentrated to
remove DMF, and then water was added (50 mL). The mixture was
extracted with EtOAc (3 ꢂ 30 mL). The combined organic phases
were washed with brine (20 mL), dried over MgSO4, and evaporated
to produce the target compound.
[26] K. Dyduch, M.P. Mitoraj, A. Michalak, ETS-NOCV description of
s-hole
bonding, J. Mol. Model. 19 (7) (2013) 2747e2758.
ꢀ
[27] A.M.S. Riel, R.K. Rowe, E.N. Ho, A.-C.C. Carlsson, A.K. Rappe, O.B. Berryman,
P.S. Ho, Hydrogen bond enhanced halogen bonds: a synergistic interaction in
chemistry and biochemistry, Acc. Chem. Res. 52 (10) (2019) 2870e2880.
[28] F.-Y. Lin, A.D. MacKerell, Do halogenehydrogen bond donor interactions
dominate the favorable contribution of halogens to ligandeprotein binding?
J. Phys. Chem. B 121 (28) (2017) 6813e6821.
[29] A.-C.C. Carlsson, M.R. Scholfield, R.K. Rowe, M.C. Ford, A.T. Alexander,
11

ꢀ
ꢀ
J. Staron, W. Pietrus, R. Bugno et al.
European Journal of Medicinal Chemistry 220 (2021) 113533
R.A. Mehl, P.S. Ho, Increasing enzyme stability and activity through hydrogen
bond-enhanced halogen bonds, Biochemistry 57 (28) (2018) 4135e4147.
[30] R.K. Rowe, P.S. Ho, Relationships between hydrogen bonds and halogen bonds
in biological systems, Acta Crystallogr. B 73 (2) (2017) 255e264.
[31] A.M. Abramyan, R.D. Slack, S. Meena, B.A. Davis, A.H. Newman, S.K. Singh,
L. Shi, Computation-guided analysis of paroxetine binding to HSERT reveals
functionally important structural elements and dynamics, Neuropharma-
cology 161 (2019) 107411.
with respect to nuclear coordinates: complete formulation for electrostatic
continuum solvation models, J. Chem. Phys. 110 (14) (1999) 6858e6870.
ꢁ
[46] S. Miertus, E. Scrocco, J. Tomasi, Electrostatic interaction of a solute with a
continuum. A direct utilizaion of AB initio molecular potentials for the pre-
vision of solvent effects, Chem. Phys. 55 (1) (1981) 117e129.
[47] W.A. Waters, Vistas of organic chemistry, Nature 173 (4413) (1954)
1012e1013.
[48] A. Todd, T.G.S. Keith, AIMAll. Overland Park KS: USA, 2019.
[49] M.K. Kesharwani, B. Brauer, J.M.L. Martin, Frequency and zero-point vibra-
tional energy scale factors for double-hybrid density functionals (and other
selected methods): can anharmonic force fields Be avoided? J. Phys. Chem.
119 (9) (2015) 1701e1714.
[50] E. Espinosa, E. Molins, C. Lecomte, Hydrogen bond strengths revealed by to-
pological analyses of experimentally observed electron densities, Chem. Phys.
Lett. 285 (3e4) (1998) 170e173.
[51] T. Ziegler, A. Rauk, On the calculation of bonding energies by the Hartree Fock
slater method, Theor. Chim. Acta 46 (1) (1977) 1e10.
[32] Schrodinger LigPrep, Release 2017-3, Schrodinger, LLC, NewYork, NY, 2017.
[33] Schrodinger Epik, Release 2017-3, Schrodinger, LLC, New York,NY, 2017.
€
€
[34] Schrodinger QSite, Release 2017-3, Schrodinger, LLC, New York,NY, 2017.
[35] W. Sherman, T. Day, M.P. Jacobson, R.A. Friesner, R. Farid, Novel procedure for
modeling ligand/receptor induced fit effects, J. Med. Chem. 49 (2) (2006)
534e553.
[36] R.B. Murphy, D.M. Philipp, R.A. Friesner, A mixed quantum mechanics/mo-
lecular mechanics (QM/MM) method for large-scale modeling of chemistry in
protein environments, J. Comb. Chem. 21 (16) (2000) 1442e1457.
[37] K.J. Bowers, F.D. Sacerdoti, J.K. Salmon, Y. Shan, D.E. Shaw, E. Chow, H. Xu,
R.O. Dror, M.P. Eastwood, B.A. Gregersen, et al., Molecular dynamics—scalable
algorithms for molecular dynamics simulations on commodity clusters, in: In
Proceedings of the 2006 ACM/IEEE Conference on Supercomputing e SC '06,
ACM Press, New York, New York, USA, 2006, p. 84.
[52] A. Michalak, M. Mitoraj, T. Ziegler, Bond orbitals from chemical valence the-
ory, J. Phys. Chem. 112 (9) (2008) 1933e1939.
[53] M. Mitoraj, A. Michalak, Applications of natural orbitals for chemical valence
in a description of bonding in conjugated molecules, J. Mol. Model. 14 (8)
(2008) 681e687.
[38] E. Harder, W. Damm, J. Maple, C. Wu, M. Reboul, J.Y. Xiang, L. Wang,
D. Lupyan, M.K. Dahlgren, J.L. Knight, et al., OPLS3: a force field providing
broad coverage of drug-like small molecules and proteins, J. Chem. Theor.
Comput. 12 (1) (2016) 281e296.
[39] K. Roos, C. Wu, W. Damm, M. Reboul, J.M. Stevenson, C. Lu, M.K. Dahlgren,
S. Mondal, W. Chen, L. Wang, et al., OPLS3e: extending force field coverage for
drug-like small molecules, J. Chem. Theor. Comput. 15 (3) (2019) 1863e1874.
[40] T. Lu, F. Chen, Quantitative analysis of molecular surface based on improved
marching tetrahedra algorithm, J. Mol. Graph. Model. 38 (2012) 314e323.
[41] T. Lu, F. Chen, Multiwfn: a multifunctional wavefunction analyzer, J. Comput.
Chem. 33 (5) (2012) 580e592.
[42] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb,
J.R. Cheeseman, G. Scalmani, V. Barone, G.A. Petersson, H. Nakatsuji, X. Li,
M. Caricato, A.V. Marenich, J. Bloino, B.G. Janesko, R. Gomperts, B. Mennucci,
D.J. Hratch, Gaussian 16, Revision A.03. Gaussian 16, Revis. A.03, 2016.
[43] Y. Zhao, D.G. Truhlar, The M06 suite of density functionals for main group
thermochemistry, thermochemical kinetics, noncovalent interactions, excited
states, and transition elements: two new functionals and systematic testing of
four M06-class functionals and 12 other function, Theor. Chem. Acc. 120 (1e3)
(2008) 215e241.
[54] M. Mitoraj, A. Michalak, Natural orbitals for chemical valence as descriptors of
chemical bonding in transition metal complexes, J. Mol. Model. 13 (2) (2007)
347e355.
[55] G. te Velde, F.M. Bickelhaupt, E.J. Baerends, C. Fonseca Guerra, S.J.A. van Gis-
bergen, J.G. Snijders, T. Ziegler, Chemistry with ADF, J. Comput. Chem. 22 (9)
(2001) 931e967.
[56] E.J. Baerends, P. Ros, Self-consistent molecular HartreedFockdslater calcu-
lations II. The effect of exchange scaling in some small molecules, Chem. Phys.
2 (1) (1973) 52e59.
[57] E.J. Baerends, D.E. Ellis, P. Ros, Self-consistent molecular HartreedFockds-
later calculations I. The computational procedure, Chem. Phys. 2 (1) (1973)
41e51.
[58] G. te Velde, E.J. Baerends, Numerical integration for polyatomic systems,
J. Comput. Phys. 99 (1) (1992) 84e98.
[59] J.P. Finley, Using the local density approximation and the LYP, BLYP and B3LYP
functionals within reference-state one-particle density-matrix theory, Mol.
Phys. 102 (7) (2004) 627e639.
[60] R. Wilcken, M.O. Zimmermann, A. Lange, S. Zahn, F.M. Boeckler, Using halogen
bonds to address the protein backbone: a systematic evaluation, J. Comput.
Aided Mol. Des. 26 (8) (2012) 935e945.
[44] F. Weigend, R. Ahlrichs, Balanced basis sets of split valence, triple zeta valence
and quadruple zeta valence quality for H to Rn: design and assessment of
accuracy, Phys. Chem. Chem. Phys. 7 (18) (2005) 3297.
[61] Y.-C. Cheng, W.H. Prusoff, Relationship between the inhibition constant (KI)
and the concentration of inhibitor which causes 50 per cent inhibition (I50) of
an enzymatic reaction, Biochem. Pharmacol. 22 (23) (1973) 3099e3108.
[45] B. Mennucci, R. Cammi, J. Tomasi, Analytical free energy second derivatives
12