An unexpected reversal in the pharmacological stereoselectivity of benzothiadiazine AMPA positive allosteric modulators

Article abstract of DOI:10.1039/c6md00440g

Benzothiadiazine type compounds (BTDs) have gained great attention for their potential therapeutic activity as nootropic and neuroprotective agents. BTDs, acting as AMPA positive allosteric modulators, potentiate the glutamatergic neurotransmission without the side effects typically associated with direct agonists. Studies regarding the binding mode of racemic BTDs into the receptor binding pocket demonstrated that one enantiomer establishes a more favourable interaction and possesses a higher biological activity with respect to the other one. The S enantiomer was proved to be the eutomer for both IDRA21 and S18986, two of the most studied BTD AMPA positive allosteric modulators. However, recent data highlighted an opposite stereoselectivity for some substituted BTDs (7-chloro-9-(furan-3-yl)-2,3,3a,4-tetrahydro-1H-benzo[e]pyrrolo[2,1-c][1,2,4]thiadiazine 5,5-dioxide and 7-chloro-2,3,4-trimethyl-3,4-dihydro-2H-benzo[e][1,2,4]thiadiazine 1,1-dioxide) showing unexpected structure-activity relationships. In this work, the synthesis and configuration assignment of the stereoisomers of 7-chloro-5-(3-furanyl)-3-methyl-3,4-dihydro-2H-1,2,4-benzothiadiazine 1,1-dioxide, one of the most active BTDs, are reported. Electrophysiological tests demonstrated that the R form is the eutomer. Docking and molecular dynamics simulations on the AMPA GluA2 binding site revealed new insights into the stereodiscrimination process. Lastly, metabolic studies disclosed a stereoselective hepatic metabolization of this chiral BTD.

Full text of DOI:10.1039/c6md00440g

MedChemComm
View Article Online
View Journal
RESEARCH ARTICLE
An unexpected reversal in the pharmacological
stereoselectivity of benzothiadiazine AMPA
positive allosteric modulators†‡
Cite this: DOI: 10.1039/
c6md00440g
Umberto M. Battisti,§ab Cinzia Citti,§cd Giulio Rastelli,a Luca Pinzi,a Giulia Puja,a
Federica Ravazzini,a Giuseppe Ciccarella,cd
Daniela Braghiroli and Giuseppe Cannazza
a
ad
*
Benzothiadiazine type compounds (BTDs) have gained great attention for their potential therapeutic activity
as nootropic and neuroprotective agents. BTDs, acting as AMPA positive allosteric modulators, potentiate
the glutamatergic neurotransmission without the side effects typically associated with direct agonists. Stud-
ies regarding the binding mode of racemic BTDs into the receptor binding pocket demonstrated that one
enantiomer establishes a more favourable interaction and possesses a higher biological activity with respect
to the other one. The S enantiomer was proved to be the eutomer for both IDRA21 and S18986, two of
the most studied BTD AMPA positive allosteric modulators. However, recent data highlighted an opposite
stereoselectivity
for
some
substituted
BTDs
(7-chloro-9-(furan-3-yl)-2,3,3a,4-tetrahydro-1H-
benzoije]pyrroloij2,1-c]ij1,2,4]thiadiazine 5,5-dioxide and 7-chloro-2,3,4-trimethyl-3,4-dihydro-2H-benzoije]-
ij1,2,4]thiadiazine 1,1-dioxide) showing unexpected structure–activity relationships. In this work, the synthesis
and configuration assignment of the stereoisomers of 7-chloro-5-(3-furanyl)-3-methyl-3,4-dihydro-2H-
1,2,4-benzothiadiazine 1,1-dioxide, one of the most active BTDs, are reported. Electrophysiological tests
demonstrated that the R form is the eutomer. Docking and molecular dynamics simulations on the AMPA
GluA2 binding site revealed new insights into the stereodiscrimination process. Lastly, metabolic studies
disclosed a stereoselective hepatic metabolization of this chiral BTD.
Received 1st August 2016,
Accepted 26th September 2016
DOI: 10.1039/c6md00440g
www.rsc.org/medchemcomm
these three subclasses, AMPA receptors (AMPARs) modulate
fast excitatory postsynaptic signaling and are involved in
1. Introduction
L-Glutamate is the major excitatory neurotransmitter in the
mammalian central nervous system (CNS) activating meta-
botropic glutamate receptors (coupled to G proteins) and
ionotropic glutamate receptors (iGluRs).1–3 On the basis of
their sensitivity to selective agonists, iGluRs have been fur-
ther classified into three subclasses: kainic acid (KA),
N-methyl-D-aspartic acid (NMDA) and α-amino-3-hydroxy-5-
methyl-4-isoxazolepropionic acid (AMPA) receptors.1–3 Among
long-term potentiation, a process related to learning and
memory establishment.4,5 For these reasons, AMPARs repre-
sent an interesting target for the development of cognitive en-
hancers. In fact, there is rapidly growing interest in AMPA
positive allosteric modulators (AMPA PAMs) as potentiators
of the glutamatergic function.6 Compared to direct AMPA re-
ceptor agonists, allosteric modulators are anticipated to pos-
sess finer tuning to increase glutamatergic function since
they have no effects in the absence of the natural ligand in
the synapse. Different studies highlighted a possible role for
AMPA PAMs in the therapeutic strategy to treat central ner-
vous system (CNS) disorders,7 such as schizophrenia,8
Alzheimer's disease,9 attention-deficit/hyperactivity disorder
(ADHD),10 and depression.11
Several chemical classes of AMPA PAMs have been de-
scribed in the past years; among them, benzothiadiazines
(BTD) represent one of the most investigated classes.6,12 Clas-
sical examples of BTD modulators are cyclothiazide (1,
Fig. 1), IDRA21 (2, Fig. 1) and S18986 (S-3, Fig. 1). Cyclo-
thiazide was one of the first reported BTDs active as AMPA
PAM, and it is still widely used as a pharmacological tool.
a Department of Life Sciences, University of Modena & Reggio Emilia, Via Campi
103, 41125 Modena, Italy. E-mail: giuseppe.cannazza@unimore.it;
Fax: +39 059 2055131; Tel: +39 059 2058575
b Department of Medicinal Chemistry, School of Pharmacy, Virginia
Commonwealth University, Richmond, Virginia 23298, USA
c Department of Biological and Environmental Sciences and Technologies,
University of Salento, Via per Monteroni, 73100 Lecce, Italy
d CNR-NANOTEC, Campus Ecotekne of the University of Salento, Via per
Monteroni, 73100 Lecce, Italy
† The authors declare no competing interests.
‡ We dedicate this work to Professor Carlo Parenti on the occasion of his retire-
ment, who, with his great experience and dedication to organic synthesis, has
given important contributions to scientific research.
§ These authors contributed equally to this work.
This journal is © The Royal Society of Chemistry 2016
Med. Chem. Commun.

View Article Online
Research Article
MedChemComm
IDRA21 has attracted particular clinical interest despite its
poor in vitro activity since it is effective in increasing learning
and memory performances in behavioural tests, representing
an important lead compound able to cross the blood–brain
barrier (BBB). Compound S-3, from Servier, is the only BTD
clinically evaluated as an AMPA PAM.
Recent crystallographic studies regarding the binding
mode of BTD derivatives on S1S2 GluA2 subunits of AMPAR
suggested that these compounds bind to the dimer inter-
face, leading to inhibition of the receptor desensitization by
stabilization of the ligand-binding domain within a dimer
interface.13 In particular, a detailed analysis of the crystal
structures of IDRA21 and cyclothiazide in complex with
S1S2 GluA2 subunits revealed that, notwithstanding they
share the same 1,2,4-benzothiadiazine 1,1-dioxide scaffold,
different substituents in the 3-position of the thiadiazine
ring lead to a shifted binding mode of 2 compared to that
of 1.14
the problem of having reliable structure–activity relationships
as well as a consistent binding mode prediction for BTD com-
pounds in the S1S2 GluA2 ligand binding site. In this work,
we reported the synthesis of the stereoisomers of 4 and their
configuration assignment and biological activity in order to
unambiguously identify the bioactive form. Docking and mo-
lecular dynamics simulations were then employed to rational-
ize the biological results. Finally, the hepatic stereoselective
metabolization of compound 4 was evaluated by using a chi-
ral LC-MS/MS method.
2. Results and discussion
2.1 Chemistry
The synthetic pathway employed is shown in Scheme 1.15
Halogenation of 2-amino-5-chlorobenzenesulfonamide (7) un-
der acidic conditions gave 8, which subsequently afforded
intermediate 9 by Pd-catalyzed cross-coupling reaction with
3-furanylboronic acid. Racemic 4 was finally obtained by ring
closure with acetaldehyde under catalytic acidic conditions.
The single enantiomers of compound 4 were then obtained
by chiral semipreparative HPLC.
Since previous studies on benzothiadiazines structurally
similar to 4 suggested that protic solvents could trigger a race-
mization process, the enantioseparation of 4 was performed
on ChiraSpher® NT column employing n-hexane/THF 75 : 25
(v/v) as an eluent (Fig. 2).19 The enantiomeric excess was eval-
uated by a chiral LC-MS/MS protocol, as described in the
methods section, using an analytical Chiralcel® OD-RH. The
first and second eluted isomers showed an enantiomeric ex-
cess of 96% and 97%, respectively.
The first eluted enantiomer gave an [α]D of +134.0 (2 mg
mL−1, acetone, T = 24 °C), whereas the second eluted enantio-
mer gave an [α]D of −140.0 (2.2 mg mL−1, acetone, T = 24 °C).
The difference between the two optical rotations was proba-
bly due to the slightly different enantiomeric excess. Both cir-
cular dichroism (CD) and UV spectra of the two enantiomers
were acquired (Fig. 3). Whilst the two UV spectra perfectly
overlapped, the CD spectra of the two enantiomers showed
an opposite cotton effect. The comparison between the CD
spectra of the two enantiomers and the CD spectra of
enantiomers of other BTD type compounds similar to 4
allowed for the assignment of the absolute configuration of
the enantiomers under investigation: the S configuration was
assigned to the dextrorotatory isomer and the R configuration
to the levorotatory one.
By exploiting these crystallographic data, we designed
and synthesized 7-chloro-5-(3-furanyl)-3-methyl-3,4-dihydro-
2H-1,2,4-benzothiadiazine 1,1-dioxide (4, Fig. 1); this mole-
cule exhibited an interesting pharmacological activity as
AMPA PAM as well as good BBB permeability.15 Recently,
we synthesized the single stereoisomers of 7-chloro-9-
(furan-3-yl)-2,3,3a,4-tetrahydro-1H-benzoije]pyrroloij2,1-c]ij1,2,4]-
thiadiazine 5,5-dioxide.16 The latter was designed as
chimeric compound between 3 and 4. Among the two
stereoisomers, (R)-7-chloro-9-(furan-3-yl)-2,3,3a,4-tetrahydro-
a
1H-benzoije]pyrroloij2,1-c]ij1,2,4]thiadiazine 5,5-dioxide (R-5,
Fig. 1) proved to be the eutomer. A similar finding was
reported by Larsen et al. for (R)-7-chloro-2,3,4-trimethyl-3,4-
dihydro-2H-benzoije]ij1,2,4]thiadiazine 1,1-dioxide (BPAM-321,
R-6, Fig. 1).17
The crystallographic data proved unambiguously that the
R enantiomer is the bioactive species of the latter compound.
Both findings are in complete disagreement with all the evi-
dence and data gathered for IDRA21 and S18986, their re-
spective lead compounds, for which the active enantiomer is
reported to be the S form.14,18 These new findings give rise to
2.2 Biological activity
The activity of compound 4 and its enantiomers as allosteric
modulators of AMPA/kainate receptors was evaluated by the
patch-clamp technique in primary cultures of cerebellar neu-
rons. Kainate (KA)-evoked currents were mainly mediated by
AMPAR activation because application of GYKI 53655 (100
μM), an AMPAR antagonist, almost completely abolished the
currents (data not shown). Since IDRA21 is one of the most
Fig. 1 Chemical structures of BTD AMPA PAMs: cyclothiazide (1), IDRA
21 (2), S18986 (S-3) 7-chloro-5-(3-furanyl)-3-methyl-3,4-dihydro-2H-
1,2,4-benzothiadiazine 1,1-dioxide (4), (R)-7-chloro-9-(furan-3-yl)-
2,3,3a,4-tetrahydro-1H-benzoije]pyrroloij2,1-c]ij1,2,4]thiadiazine 5,5-di-
oxide (R-5), (R)-7-chloro-2,3,4-trimethyl-3,4-dihydro-2H-benzoije]-
ij1,2,4]thiadiazine 1,1-dioxide (R-6).
Med. Chem. Commun.
This journal is © The Royal Society of Chemistry 2016

View Article Online
MedChemComm
Research Article
Scheme 1 Reagents and conditions: (i) Br2, AcOH, ACN; (ii) Na2CO3, 3-furanylboronic acid, tetrakis(triphenylphosphine)palladium(0), H2O/dioxane,
110 °C; (iii) CH3CHO, HCl, iPrOH; (iv) ChiraSpher® NT column (250 × 10 mm I.D., 5 μm), 200 μL loop, flow rate: 5 mL min−1, mobile phase:
n-hexane/THF (75 : 25, v/v). Concentration of 4: 10 μM.
obtained indicated that compound 4 and IDRA21 potentiated
by 1603 512% (n = 4) and 8 11% (n = 4) the KA-evoked
currents at 10 μM, respectively (Fig. 4). In order to highlight
possible stereospecific interactions of 4 with AMPA/kainate
receptors, R-4 and S-4 were also tested. The data obtained in-
dicated a marked difference in activity: R-4 enhanced the cur-
rents by 2586 682% while S-4 by 488 370%. These surpris-
ing results indicated that R-4 is almost five times more active
than S-4 confirming a stereospecific interaction with the
receptor.
Recently, our research group has demonstrated that com-
pound 4 is configurationally unstable in protic aqueous sol-
vents, such as physiological solution. Anyway, the racemiza-
tion rate of compound 4 is much lower than those of other
benzothiadiazine type compounds (t1/2 = 60 min at pH 7.4
Fig. 2 Semipreparative enantioseparation of 4. Column: ChiraSpher®
and 37 °C).19
NT (250 × 10 mm I.D., 5 μm), 200 μL loop, mobile phase: n-hexane :
THF (75 : 25, v/v), flow rate:
temperature.
5 , λmax = 254 nm, room
mL min−1
2.3 Modeling
A
possible explanation of the enantiomeric preference
showed by compound 4 was sought by modeling the R and S
enantiomers of this compound into the AMPA allosteric bind-
ing site. To this end, the R and S enantiomers of 4 were
docked into two crystal structures of the AMPAR, one in
Fig. 3 Experimental CD (top) and UV (bottom) spectra of the single
enantiomers of 4 obtained by semipreparative chiral chromatography.
Concentration: 0.1 mg mL−1 in acetone : methanol (1 : 20, v/v); cell
length: 1 mm (green line, S-4; blue line, R-4).
Fig. 4 Biological activity of the selected compounds. Potentiation of
R-4, S-4, 4 and 2 on KA-evoked currents (10 μM). Each data point is
the mean SEM of at least five cells.
relevant BTD derivatives reported in the literature as AMPA
PAM, it was selected as a reference compound. The data
This journal is © The Royal Society of Chemistry 2016
Med. Chem. Commun.

View Article Online
Research Article
MedChemComm
complex with an analogue of 4 with S stereochemistry (com-
pound IDRA21, PDB code 3IL1) and the other with R stereo-
chemistry (compound BPAM-321, PDB code 5BUU).14,17
Docking into the two crystal structures provided almost iden-
tical results (RMSD between ligand poses obtained in the two
crystal structures being less than 0.8 Å). Moreover, both
enantiomers of compound 4 displayed a binding mode very
similar to those of IDRA21 and BPAM-321 in the correspond-
ing 3IL1 and 5BUU crystal structures, RMSD of the common
skeleton atoms being less than 0.6 Å.
also run as a control. In the resulting MD trajectories, the ori-
entation of the two ligands in the binding pocket was
maintained throughout the simulation, the average RMSD
values with respect to the initial docking pose being always
less than 1.3 Å. The average hydrogen bond distances be-
tween the 2-NH groups of compounds S-4 and S-2 and the
Pro494 carbonyl were 2.1 0.4 Å, while no hydrogen bond
could be observed for the R enantiomers. Rather, in the latter
stereochemistry, the 2-NH group was engaged in a hydrogen
bond with a water molecule.
Fig. 5A and B show the binding modes of R-4 and S-4 pre-
dicted by docking in the 3IL1 crystal structure. Both enantio-
mers established a network of interactions analogous to
those of IDRA21 in the 3IL1 crystal structure. Moreover, se-
quential docking into the two adjacent allosteric pockets pro-
vided almost identical ligand poses, in agreement with avail-
able crystal structures. In both enantiomers, one
sulfonamide oxygen hydrogen bonded with the Gly731 back-
bone amino group. A further hydrogen bond with the Pro494
carbonyl was observed only for S-4. Interestingly, superimpo-
sition of the R-4 and S-4 binding modes (Fig. 5C) showed that
the two enantiomers adopt a very similar orientation. How-
ever, while the 2-NH group of the S enantiomer hydrogen
bonds with the backbone carbonyl of Pro494, it is solvent ex-
posed in the R enantiomer. The docking scores could not dis-
criminate the different binding ability of the two
enantiomers.
Binding free energy predictions are reported in Table 1.
Free energies of binding (ΔGbind) were calculated as the
sum of the molecular mechanics contribution of the inter-
nal, electrostatic, and van der Waals contribution to bind-
ing in vacuo (ΔGvacuo) and the solvation free energy contri-
bution to binding expressed as the sum of polar and
nonpolar solvation free energies (ΔGsolv). Comparison of cal-
culated free energies of binding suggested that the
R
stereoisomer of compounds 4 and 2 binds 6.6 kcal mol−1
more favorably and 2.5 kcal mol−1 less favorably with re-
spect to the corresponding S stereoisomer (Table 1). These
data are in agreement with the inversion of the eutomer
preferences for compounds 2 and 4 observed experimen-
tally. Interestingly, interaction energies in vacuo were in fa-
vor of the R stereoisomer for compound 4 (ΔGvacuo of −2.2
kcal mol−1) but not for compound 2, which favored the S
stereoisomer by 5.6 kcal mol−1. Solvation free energies
(ΔGsolv) were always in favor of the R stereochemistry, likely
because in this stereochemistry the 2-NH is solvent exposed
and does not hydrogen bond to the proline, thus it re-
quires less energy to desolvate. In the case of compound 4
both energy terms cooperated to favor the R enantiomer,
thus explaining the overall effect. This interaction with
Pro494 was proved to be non-essential for the binding and
the activity of BTD compounds.17,32 However, the data here
presented and the crystallographic structure reported by
Larsen et al. show that this residue plays a relevant role in
stereodiscrimination. In fact, compounds able to form addi-
tional van der Waals interactions (e.g. 4 vs. 2, or R-6 vs. 2,
R-5 vs. S-3) seem to prefer the R stereoisomer that can
probably fit better into the receptor cavity. The loss of the
polar interaction with Pro494, related to the R configura-
tion, appears to be overcome by van der Waals interactions
and/or polar interactions with water molecules.
Therefore, molecular dynamics (MD) simulations in ex-
plicit solvent followed by binding free energy predictions
made with MM-PBSA31 were performed starting from the
docking complexes of R-4 and S-4 described above. MD simu-
lations on both stereoisomers of IDRA21 (2), which showed
preference for the S enantiomer in the crystal structure, were
Table 1 Binding free energy predictions (kcal mol−1) made with MM-
PBSA for compounds 4 and 2 in the S and R stereochemistry. For each
free energy component, energy differences are always relative to those
of the S enantiomer. ΔGbind is calculated as ΔGvacuo + ΔGsolv
Fig. 5 Binding mode of compounds S-4 (A) and R-4 (B) predicted by
docking in the 3IL1 crystal structure. Panel C shows a superimposition
of the S-4 and R-4 binding modes, highlighting that the hydrogen
bond with the Pro494 backbone carbonyl is possible only for S-4. The
S-4 and R-4 enantiomers are colored in black and wheat, respectively.
The two monomers (chain B and chain E) are colored in red and green,
respectively. The image is created using PyMol (The PyMOL Molecular
Graphics System, Version 1.8, Schrödinger, LLC).
S-4
R-4
S-2
R-2
5.6
ΔGvacuo
ΔGsolv
ΔGbind
0
0
0
−2.2
−4.4
−6.6
0
0
0
−3.1
2.5
Med. Chem. Commun.
This journal is © The Royal Society of Chemistry 2016

View Article Online
MedChemComm
Research Article
2.4 Stereoselective hepatic metabolism of 7-chloro-5-(3-
furanyl)-3-methyl-3,4-dihydro-2H-1,2,4-benzothiadiazine
1,1-dioxide (4)
area of the peaks obtained by the LC-MS/MS injection of
the sample at t0 and that of the sample at t60. The loss
was about 33% for the R enantiomer and 62% for the S
enantiomer. This result suggested that enantiomer R-4 is
scarcely metabolized to the unsaturated derivative by he-
patic CYP450 compared to S-4. In the light of the results
obtained from the biological tests, these findings assume
great relevance considering that a large amount of the bio-
active enantiomer could reach the CNS to exert its function.
To the best of our knowledge, this is the first time that a
stereoselective metabolism for BTD compounds has been
reported. Moreover, recent cerebral microdialysis studies
carried out by our research group have highlighted that
when racemic 4 was administered in vivo the concentration
of the R enantiomer in the CNS was slightly higher than
that of the S enantiomer.32 This is most likely due to the
stereoselective hepatic metabolism suggested by the results
reported herein.
A simple procedure to evaluate BTD metabolism using rat
liver microsomes, a subcellular fraction containing major
drug-metabolizing enzymes, including the cytochrome P450
(CYP450) family and flavin monooxygenase has been recently
reported.29,30 This protocol was successfully applied to race-
mic 4 indicating a fast metabolization to its unsaturated
derivative
7-chloro-5-(furan-3-yl)-3-methyl-4H-benzoije]ij1,2,4]-
thiadiazine 1,1-dioxide, which demonstrated to retain most of
the activity of the parent compound 4.30 However, since it
was possible to isolate and assign the configuration of the
stereoisomers and given the difference observed in the bio-
logical activity, it is of great importance to develop a chiral
version of the metabolization protocol in order to assess a
possible stereoselective metabolization. Briefly, 5 μL of a 10
μM solution of racemic compound 4 was incubated with 432
μL of 0.1 M phosphate buffer, pH 7.4, 50 μL of 10 mM
NADPH, and 13 μL of rat liver microsomes at a protein con-
centration of 20 mg mL−1 in the same phosphate buffer. The
incubation at 37 °C was carried out at t0 (stopped right after
the addition of microsomes with ice-cold methanol) and t60
(stopped 60 min after the addition of microsomes). The two
media (at t0 and t60) were analyzed by the same LC-MS/MS
method used for the determination of the enantiomeric ex-
cess of S-4 and R-4 after semipreparative enantioseparation.
The area of the peak corresponding to the R enantiomer was
higher than that of the S enantiomer by about 50% after 60
minutes of incubation with the microsomes (Fig. 6). The loss
(%) of the parent compound was obtained by the ratio of the
3. Experimental
3.1 Chemicals and reagents
All chemicals and reagents, except those specifically noted,
were purchased from Sigma-Aldrich (Milan, Italy). LC/MS
grade water and acetonitrile (ACN) were also purchased from
Sigma-Aldrich (Milan, Italy).
3.2 Chemistry
2-Amino-5-chloro-3-(3-furanyl)benzenesulfonamide
(8).
Compound 9 was obtained as described by Battisti et al.15
1
Yield: 70%, (28 mg, two steps), m.p.: 81–83 °C. H NMR (400
MHz, CDCl3); δ = 5.10 (s, broad, 2H), 5.12 (s, broad, 2H), 6.56
(s, 1H), 7.32 (d, J = 2.5 Hz, 1H), 7.57 (t, J = 1.5 Hz, 1.6 Hz,
1H), 7.63 (s, 1H), 7.75 (d, J = 2.5 Hz, 1H). GC-MS (70 eV): m/z
272 (84) [M+], 191 (67), 163 (60), 128 (100), 101 (30).
( )-7-Chloro-5-(3-furanyl)-3-methyl-3,4-dihydro-2H-1,2,4-
benzothiadiazine 1,1-dioxide (4). Compound 4 was obtained
from compound 9 as described by Battisti et al.15 Yield: 99%
(30 mg, one step), m.p.: 183–185 °C. 1H NMR (400 MHz,
CDCl3): δ = 1.54 (d, J = 6.1 Hz, 3H), 4.65 (s, 1H, broad), 4.69
(s, 1H), 5.02–5.04 (m, 1H), 6.55 (s, 1H), 7.25 (d, J = 2.5 Hz,
1H), 7.53 (d, J = 2.5 Hz, 1H), 7.58 (t, J = 1.5 Hz, 1H), 7.63 (s,
1H). GC–MS (70 eV): m/z 298 (100) [M+], 283 (57), 192 (98),
163 (55), 128 (77), 101 (35).
R-(−)-7-Chloro-5-(3-furanyl)-3-methyl-3,4-dihydro-2H-1,2,4-
benzothiadiazine 1,1-dioxide (R-4). Compound R-4 was
obtained by semipreparative chiral chromatography from 4
(conditions: ChiraSpher® NT column (250 × 10 mm I.D., 5
μm; Merck, Darmstadt, Germany), 200 μL loop, flow rate: 5
mL min−1, mobile phase: n-hexane/THF (75 : 25, v/v), tempera-
ture: 25 °C).
Fig. 6 MS chromatograms (positive ionization mode, SRM: 299 → 218)
of the microsomal extracts after incubation of racemic 4 with the
microsomal enzymes at t0 (top) and t60 (bottom). Chromatographic
conditions: Chiralcel® OD-RH column (150 × 4.6 mm I.D., 5 μm); mo-
bile phase: water (0.1% formic acid) : ACN (50 : 50, v/v); flow rate: 0.5
mL min−1. The decrease in the peak area was about 62% for S-4 and
about 33% for R-4, indicating a stereoselective metabolization of 4 by
hepatic CYP450.
Yield: 81% (4 mg, 97% ee), m.p.: 183–185 °C. 1H NMR
(400 MHz, CDCl3): δ = 1.54 (d, J = 6.1 Hz, 3H), 4.65 (s, 1H,
broad), 4.69 (s, 1H), 5.02–5.04 (m, 1H), 6.55 (s, 1H), 7.25 (d, J
= 2.5 Hz, 1H), 7.53 (d, J = 2.5 Hz, 1H), 7.58 (t, J = 1.5 Hz, 1H),
This journal is © The Royal Society of Chemistry 2016
Med. Chem. Commun.

View Article Online
Research Article
MedChemComm
7.63 (s, 1H). GC–MS (70 eV): m/z 298 (100) [M+], 283 (57), 192
side chains and hydrogens were added by adjusting the ioni-
zation and tautomerization states of the protein at physiolog-
ical pH. Finally, the structures were refined with a restrained
minimization with the OPLS2005 force field up to a final root
mean square distance (RMSD) of 0.30 Å with respect to the
input protein coordinates. Ligand structures were drawn with
Maestro and then minimized by using the OPLS2005 force
field.22 Docking was performed by using the standard preci-
sion (SP) docking protocol available in Glide with default pa-
rameters.23 Considering that the AMPAR acts as a dimer with
two allosteric ligands bound in two adjacent and symmetri-
cally equivalent binding sites, sequential docking in 3IL1 and
5BUU structures was performed by centering the grid once
on the centroid of the first crystallographic ligand and then
on the centroid of the second one. The docking protocol was
validated by redocking IDRA21 and BPAM-321 in their respec-
tive crystal structures.
(98), 163 (55), 128 (77), 101 (35). [α]D = −140.0 (2.2 mg mL−1
acetone; 24 °C).
;
S-(+)-7-Chloro-5-(3-furanyl)-3-methyl-3,4-dihydro-2H-1,2,4-
benzothiadiazine 1,1-dioxide (S-4). Compound S-4 was
obtained by semipreparative chiral chromatography from 4
(conditions: ChiraSpher® NT column (250 × 10 mm I.D., 5
μm; Merck, Darmstadt, Germany), 200 μL loop, flow rate: 5
mL min−1, mobile phase: n-hexane/THF (75 : 25, v/v), tempera-
ture: 25 °C).
Yield: 72% (3.6 mg, 96% ee), m.p.: 183–185 °C. 1H NMR
(400 MHz, CDCl3): δ = 1.54 (d, J = 6.1 Hz, 3H), 4.65 (s, 1H,
broad), 4.69 (s, 1H), 5.02–5.04 (m, 1H), 6.55 (s, 1H), 7.25 (d, J
= 2.5 Hz, 1H), 7.53 (d, J = 2.5 Hz, 1H), 7.58 (t, J = 1.5 Hz, 1H),
7.63 (s, 1H). GC–MS (70 eV): m/z 298 (100) [M+], 283 (57), 192
(98), 163 (55), 128 (77), 101 (35). [α]D = +134.0 (2 mg mL−1
acetone; 24 °C).
;
Molecular dynamics (MD) simulations were performed
starting from the docking complexes obtained as described
above. Considering that docking into 3IL1 and 5BUU pro-
vided almost identical results and the two protein structures
are very similar, MD simulations were run only with the com-
plexes obtained with 3IL1. Simulations were run on the di-
meric structure of the receptor with two allosteric ligands,
one bound to each subunit. Simulations were performed also
for the analog without the furan ring (IDRA21), for
comparison.
MD simulations were carried out with the Amber 14 soft-
ware package, using the ff14SB force field for the protein and
the GAFF force field for the ligands. Ligands were prepared
with the Antechamber module by assigning AM1-BCC atomic
charges and GAFF atom types.24 The system was neutralized
by adding chloride ions (Cl−) according to the Coulomb po-
tential grid as calculated with Leap. Afterwards, the system
was solvated with a rectangular box of TIP3P water molecules
placed within 10 Å distance from the solute.25 Energy mini-
mization and MD were performed with the Particle Mesh
Ewald Molecular Dynamics (PMEMD) CUDA version software,
employing a 12 Å cut-off for non-bonded interactions and the
SHAKE algorithm and a 2 fs time step.26 3000 steps of
steepest descent and conjugate gradient energy minimization
were performed on water molecules and ions first, keeping
the protein and the two ligands in place using harmonic con-
straints. Then, 1000 steps of energy minimization were
conducted without constraints.
3.3 Electrophysiological tests
Primary cultures of cerebellar granule neurons were prepared
from 7 day old Sprague–Dawley rats as reported in the litera-
ture.20 Briefly, cells from the cerebellum were dispersed with
trypsin (0.24 mg mL−1; Sigma Aldrich, Milan, Italy) and
plated at a density of 0.8 × 106 cells per mL on 35 mm Falcon
dishes coated with poly-L-lysine (10 μg mL−1, Sigma Aldrich).
Cells were plated in basal Eagle's Medium (BME; Celbio, Mi-
lan, Italy), supplemented with 10% fetal bovine serum
(Celbio), 2 mM glutamine, 25 mM KCl and 100 μg mL−1
gentamycin (Sigma Aldrich) and maintained at 37 °C in 5%
CO2. After 24 h in vitro, the medium was replaced with a 1 : 1
mixture of BME and Neurobasal medium (Celbio, Milan)
containing 2% B27 supplement, 1% antibiotic, and 0.25%
glutamine (Invitrogen). At 5 days in vitro (DIV5), cytosine
arabinofuranoside (Ara-C) was added at a final concentration
of 1 μM. Recordings were performed at room temperature,
under voltage-clamp in the whole-cell configuration of the
patch-clamp technique on cells. Electrodes were pulled from
borosilicate glass (Heidelberg, FRG) using a vertical puller
(PB-7, Narishige) and had a resistance of 5–6 MOhm. Cur-
rents were amplified using an Axopatch 1D amplifier (Axon
Instruments, Foster). The recording chamber was continu-
ously perfused at 5 mL min−1 with an artificial extracellular
solution composed of (mM): 145 NaCl, 5 KCl, 1 CaCl2, 5
Hepes, 5 glucose, 20 sucrose, pH 7.4 with NaOH. The
electrode intracellular solution contained (mM): 140 KCl, 3
MgCl2, 5 EGTA, 5 Hepes, 2 ATP-Na, pH 7.3 with KOH. Drugs
were applied directly by gravity using a Y-tube perfusion
system.
The system was heated from 0 to 300 K with 100 ps con-
stant volume Langevin MD and a collision frequency of 2.0
ps−1, keeping the protein and the ligands in place using a 5
kcal mol−1 Å2 harmonic constraint. Then, constraints were
gradually reduced in 2 ns turning to constant pressure MD,
and equilibration was performed for 0.5 ns without con-
straints. After equilibration, a 20 ns MD production run was
performed for each complex and coordinates were collected
every 100 ps. For each system, MD simulations were run in
three replicates. Trajectories were analyzed with the CPPTRAJ
module available in Amber.27
3.4 Modeling
Protein structures were downloaded from the Protein Data
Bank (PDB codes 3IL1 and 5BUU). The structures were then
processed with the Schrödinger Suite 2014-3.21 The com-
plexes were prepared for docking by using the Protein Prepa-
ration Wizard module available in the Maestro suite. Missing
Med. Chem. Commun.
This journal is © The Royal Society of Chemistry 2016

View Article Online
MedChemComm
Research Article
Free energies of binding were calculated by using the par- 4. Conclusions
allel version of the MMPBSA.py module available in Amber.28
In conclusion, the isolation of the single enantiomers of 4 by
semipreparative chiral chromatography has been reported.
Electrophysiological results, which showed unexpectedly a
higher in vitro AMPA PAM activity for R-4 compared to S-4,
are in contrast to those reported for the lead compound
IDRA21, for which the S form is the eutomer. Docking stud-
ies and molecular dynamics simulations were employed to
understand the different stereoselectivities. The data
obtained suggested a relevant role of an H-bond with Pro494
in the stereodiscrimination process. Bulkier molecules might
interact more favourably with an R configuration in order to
fit into the binding pocket. The binding of the R isomer leads
to the loss of the interaction with Pro494. However, this defi-
cit seems to be overcome by the contribution of van der
Waals interactions and/or by H-bonding with water mole-
cules. Binding free energy predictions suggested that in the
Free energies were calculated with MM-PBSA on 200 equally
spaced frames collected during the 20 ns production run. All
water molecules and counterions were removed except for six
(three for each subunit) water molecules that make important
hydrogen bond bridges between the protein and the ligand.
These water molecules (W317, W459, W485, W595, W610,
W766, the numbering refers to the 3IL1 PDB crystal struc-
ture) are present in most crystal structures of the AMPA re-
ceptor in complex with ligands.
Free energies reported in Table 1 are the averages of the
energies obtained in the three replicates.
3.5 Metabolic studies
The experiments were performed at t0 (the reaction was
stopped right after the addition of microsomes) and t60 (the
reaction was stopped after 60 minutes from the addition of
microsomes) following a literature known procedure.29 The
experiments were carried out using the test compound and a
negative control (without NADPH). Each experiment at t0 and
t60 except for the negative control was performed in dupli-
cate. Each Eppendorf tube contained: 432 μL of 0.1 M phos-
phate buffer (pH 7.4) pre-warmed at 37 °C, 50 μL of 10 mM
NADPH (in the negative control, it was replaced with phos-
phate buffer) and 5 μL of the test compound (10 μM solution
in phosphate buffer). After 5 min incubation at 37 °C, 13 μL
of microsomes at a concentration of 20 mg protein mL−1 were
added. The reactions at t0 were stopped with 250 μL of ice-
cold methanol, and then the mixture was vortexed well. The
reactions at t60 were incubated at 37 °C for 60 minutes, then
stopped with 250 μL of ice-cold methanol and vortexed well.
The samples were centrifuged at 10 000g for 5 min at 4 °C.
The supernatants were placed in HPLC vials and analysed
with the suitable LC-MS/MS method. The pellets were stored
in the freezer until all the samples were analysed.30
case of compound
4 gas-phase interaction energy and
desolvation energy cooperate to favour the R enantiomer. The
metabolism of the single enantiomers of compound 4 was
also investigated. The protocol outcome underlined a stereo-
selective hepatic metabolism for the S enantiomer. The re-
sults obtained in this work highlight that not only the stereo-
chemistry of the chiral centre determines the stereoselective
binding with the target receptor, but also structural chemical
elements far from the stereogenic centre are responsible for
the stereodiscrimination process and consequently for the
biological activity.
References
1 (a) R. Dingledine, K. Borges, D. Bowie and S. Traynelis,
Pharmacol. Rev., 1999, 51, 7–61; (b) M. L. Mayer and N.
Armstrong, Structure and function of glutamate receptor ion
channels, Annu. Rev. Physiol., 2004, 66, 161–181.
2 G. L. Collingridge and J. C. Watkins, The NMDA Receptor,
Oxford University Press, Oxford, 1994.
3.6 LC-MS/MS method for the determination of R-4 and S-4
3 J. N. C. Kew and J. A. Kemp, Psychopharmacology, 2005, 179,
4–29.
The LC-MS/MS method developed was employed for both the
determination
of
the
enantiomeric
excess
after
4 U. Staubli, G. Rogers and G. Lynch, Proc. Natl. Acad. Sci.
U. S. A., 1994, 91, 777–781.
5 U. Staubli, Y. Perez, F. B. Xu, G. Rogers, M. Ingvar, S. Stone-
Elander and G. Lynch, Proc. Natl. Acad. Sci. U. S. A.,
1994, 91, 11158–11162.
semipreparative enantioseparation of R-4 and S-4 and their
quantification in the microsomal media. The samples were
injected on the Chiralcel® OD-RH column (150 × 4.6 mm I.
D.,
5 μm; Chiral Technologies Europe, Bd Gonthier
d'Andernach, Illkirch-Cedex, France); the mobile phase was
composed of water with 0.1% formic acid and ACN (50 : 50, v/
v), pumped in isocratic mode at a flow rate of 0.5 mL min−1.
An Agilent 6410 triple quadrupole-mass spectrometer with an
electrospray ionization source operating in positive mode was
used to evaluate the enantiomeric excess and to analyse the
microsomal extracts. The analyses were followed in selected
reaction monitoring (SRM) mode to improve the sensitivity:
299 → 218 (154) was the transition selected for both R-4 and
S-4. The chromatograms were integrated using Agilent Mass
Hunter software (B.06.00).
6 (a) P. Francotte, P. de Tullio and P. Fraikin, Recent Pat. CNS
Drug Discovery, 2006, 1, 239–246; (b) S. E. Ward and M.
Harries, Curr. Med. Chem., 2010, 17, 3503–3513; (c) B.
Pirotte, P. Francotte, E. Goffin and P. De Tullio, Expert Opin.
Ther. Pat., 2013, 23, 615–628; (d) S. E. Ward, L. E. Pennicott
and P. Beswick, Future Med. Chem., 2015, 7, 473–491; (e) B.
Cox and M. Gosling, Ion Channel Drug Discovery, RSC Drug
Discovery Series, 2015; ( f ) K. M. Partin, Curr. Opin.
Pharmacol., 2015, 20, 46–53.
7 (a) M. J. O'Neill, D. Bleakman, D. M. Zimmerman and E. S.
Nisenbaum, Curr. Drug Targets: CNS Neurol. Disord., 2004, 3,
This journal is © The Royal Society of Chemistry 2016
Med. Chem. Commun.

View Article Online
Research Article
181–194; (b) C. A. Zarate and H. K. Manji, Exp. Neurol.,
MedChemComm
New York, NY, 2014; Prime version 3.7, Schrödinger, LLC,
New York, NY, 2014.
2008, 211, 7–10.
8 D. C. Goff, L. Leahy, I. Berman, T. Posever, L. Herz, A. C.
Leon, S. A. Johnson and G. Lynch, J. Clin. Psychopharmacol.,
2001, 21, 484–487.
9 (a) G. Lynch, Curr. Opin. Pharmacol., 2004, 4, 4–11; (b) M. J.
O'Neill and S. Dix, IDrugs, 2007, 10, 185–192.
10 L. A. Adler, R. A. Kroon, M. Stein, M. Shahid, F. I. Tarazi, A.
Szegedi, J. Schipper and P. Cazorla, Biol. Psychiatry,
2012, 72, 971–977.
11 K. Hashimoto, Brain Res. Rev., 2009, 61, 105–123.
12 B. Pirotte, P. Francotte and E. Goffin, Curr. Med. Chem.,
2010, 17, 3575–3582.
13 (a) A. I. Sobolevsky, M. P. Rosconi and E. Gouaux, Nature,
2009, 462, 745–756; (b) Y. Sun, R. Olson, M. Horning, N.
Armstrong, M. Mayer and E. Gouaux, Nature, 2002, 417,
245–253; (c) R. Jin, S. Clark, A. M. Weeks, J. T. Dudman, E.
Gouaux and K. M. Partin, J. Neurosci., 2005, 25, 9027–9036.
14 C. P. Ptak, A. H. Ahmed and R. E. Oswald, Biochemistry,
2009, 48, 8594–8602.
15 U. M. Battisti, K. Jozwiak, G. Cannazza, G. Puia, G. Stocca,
D. Braghiroli, C. Parenti, L. Brasili, M. M. Carrozzo, C. Citti
and L. Troisi, ACS Med. Chem. Lett., 2012, 3, 25–29.
16 M. M. Carrozzo, U. M. Battisti, G. Cannazza, G. Puia, F.
Ravazzini, A. Falchicchio, S. Perrone, C. Citti, K. Jozwiak, D.
Braghiroli, C. Parenti and L. Troisi, Bioorg. Med. Chem.,
2014, 22, 4667–4676.
17 A. P. Larsen, P. Francotte, K. Frydenvang, D. Tapken, E. Goffin,
P. Fraikin, D.-H. Caignard, P. Lestage, L. Danober, B. Pirotte
and J. S. Kastrup, ACS Chem. Neurosci., 2016, 7, 378–390.
18 K. Bernard, L. Danober, J.-Y. Thomas, C. Lebrum, C. Munoz,
A. Cordi, P. Desos, P. Lestage and P. Morain, CNS Neurosci.
Ther., 2010, 16, e193–e212.
19 (a) G. Cannazza, U. M. Battisti, M. M. Carrozzo, A. S.
Cazzato, D. Braghiroli, C. Parenti and L. Troisi,
J. Chromatogr. A, 2014, 1363, 216–225; (b) G. Cannazza,
M. M. Carrozzo, U. Battisti, D. Braghiroli and C. Parenti,
J. Chromatogr. A, 2009, 1216, 5655–5659; (c) G. Cannazza, U.
Battisti, M. M. Carrozzo, L. Brasili, D. Braghiroli and C.
Parenti, Chirality, 2011, 23, 851–859; (d) G. Cannazza, M. M.
Carrozzo, U. Battisti, D. Braghiroli, C. Parenti, A. Troisi and
L. Troisi, Chirality, 2010, 22, 789–797; (e) M. M. Carrozzo, G.
Cannazza, U. Battisti, D. Braghiroli and C. Parenti, Chirality,
2010, 22, 389–397; ( f ) G. Cannazza, M. M. Carrozzo, D.
Braghiroli and C. Parenti, J. Chromatogr., B: Anal. Technol.
Biomed. Life Sci., 2008, 875, 192–199; (g) G. Cannazza, D.
Braghiroli, P. Iuliani and C. Parenti, Tetrahedron: Asymmetry,
2006, 22, 3158–3162.
22 W. L. Jorgensen, D. S. Maxwell and J. Tirado-Rives, J. Am.
Chem. Soc., 1996, 118, 11225–11236.
23 (a) R. A. Friesner, J. L. Banks, R. B. Murphy, T. A. Halgren,
J. J. Klicic, D. T. Mainz, M. P. Repasky, E. H. Knoll, D. E.
Shaw, M. Shelley, J. K. Perry, P. Francis and P. S. Shenkin,
J. Med. Chem., 2004, 47, 1739–1749; (b) T. A. Halgren, R. B.
Murphy, R. A. Friesner, H. S. Beard, L. L. Frye, W. T. Pollard
and J. L. Banks, J. Med. Chem., 2004, 47, 1750–1759; (c)
Small-Molecule Drug Discovery Suite 2014-4: Glide, version
6.5, Schrödinger, LLC, New York, NY, 2014.
24 (a) D. A. Case, V. Babin, J. T. Berryman, R. M. Betz, Q. Cai,
D. S. Cerutti, T. E. Cheatham III, T. A. Darden, R. E. Duke,
H. Gohlke, A. W. Goetz, S. Gusarov, N. Homeyer, P.
Janowski, J. Kaus, I. Kolossváry, A. Kovalenko, T. S. Lee, S.
LeGrand, T. Luchko, R. Luo, B. Madej, K. M. Merz, F.
Paesani, D. R. Roe, A. Roitberg, C. Sagui, R. Salomon-Ferrer,
G. Seabra, C. L. Simmerling, W. Smith, J. Swails, R. C.
Walker, J. Wang, R. M. Wolf, X. Wu and P. A. Kollman,
AMBER 14, University of California, San Francisco, 2014; (b)
V. Hornak, R. Abel, A. Okur, B. Strockbine, A. Roitberg and
C. Simmerling, Proteins, 2006, 65, 712–725; (c) J. Wang,
R. M. Wolf, J. W. Caldwell, P. A. Kollamn and D. A. Case,
J. Comput. Chem., 2004, 25, 1157–1174; (d) B. Wang and
K. M. J. Merz, J. Chem. Theory Comput., 2006, 2, 209–215; (e)
A. Jakalian, B. L. Bush, D. B. Jack and C. I. Bayly, J. Comput.
Chem., 2000, 21, 132–146; ( f ) A. Jakalian, D. B. Jack and C. I.
Bayly, J. Comput. Chem., 2002, 23, 1623–1641.
25 W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W.
Impey and M. L. Klein, J. Chem. Phys., 1983, 79,
926–935.
26 (a) A. W. Goetz, M. J. Williamson, D. Xu, D. Poole, S. L.
Grand and R. C. Walker, J. Chem. Theory Comput., 2012, 8,
1542–1555; (b) R. Salomon-Ferrer, A. W. Goetz, D. Poole,
S. L. Grand and R. C. Walker, J. Chem. Theory Comput.,
2013, 9, 3878–3888; (c) R. C. Walker, Comput. Phys.
Commun., 2013, 184, 374–380.
27 D. R. Roe and T. E. Cheatham, J. Chem. Theory Comput.,
2013, 9, 3084–3095.
28 B. R. Miller III, T. D. McGee Jr., J. M. Swails, N. Homeyer, H.
Gohlke and A. E. Roitberg, J. Chem. Theory Comput., 2012, 8,
3314–3321.
29 J. R. Hill, in Current Protocols in Pharmacology, John Wiley &
Sons, Inc., 2003, ch. 7.8, vol. 23, pp. 1–11.
30 C. Citti, U. M. Battisti, G. Cannazza, K. Jozwiak, N. Stasiak,
G. Puja, F. Ravazzini, G. Ciccarella, D. Braghiroli, C. Parenti,
L. Troisi and M. Zoli, ACS Chem. Neurosci., 2016, 7, 149–160.
31 M. D. Parenti and G. Rastelli, Biotechnol. Adv., 2012, 30,
244–250.
20 K. Murase, P. D. Ryu and M. Randic, Neurosci. Lett.,
1989, 103, 56.
21 Schrödinger Release 2014-3; Schrödinger Suite 2014-3 Pro-
tein Preparation Wizard; Epik version 2.9, Schrödinger, LLC,
New York, NY, 2014; Impact version 6.4, Schrödinger, LLC,
32 U. M. Battisti, C. Citti, M. Larini, G. Ciccarella, N. Stasiak, L.
Troisi, D. Braghiroli, C. Parenti, M. Zoli and G. Cannazza,
J. Chromatogr. A, 2016, 1443, 152–161.
Med. Chem. Commun.
This journal is © The Royal Society of Chemistry 2016