7-Phenoxy-Substituted 3,4-Dihydro-2H-1,2,4-benzothiadiazine 1,1-Dioxides as Positive Allosteric Modulators of α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic Acid (AMPA) Receptors with Nanomolar Potency

Article abstract of DOI:10.1021/acs.jmedchem.7b01323

We report here the synthesis of 7-phenoxy-substituted 3,4-dihydro-2H-1,2,4-benzothiadiazine 1,1-dioxides and their evaluation as AMPA receptor positive allosteric modulators (AMPApams). The impact of substitution on the phenoxy ring and on the nitrogen at

Full text of DOI:10.1021/acs.jmedchem.7b01323

Article
pubs.acs.org/jmc
Cite This: J. Med. Chem. XXXX, XXX, XXX−XXX
7‑Phenoxy-Substituted 3,4-Dihydro‑2H‑1,2,4-benzothiadiazine 1,1-
Dioxides as Positive Allosteric Modulators of α‑Amino-3-hydroxy-5-
methyl-4-isoxazolepropionic Acid (AMPA) Receptors with Nanomolar
Potency
Eric Goffin,†,⊥ Thomas Drapier,†,⊥ Anja Probst Larsen,‡ Pierre Geubelle,†,§ Christopher P. Ptak,∥
Saara Laulumaa,‡ Karoline Rovinskaja,‡ Julie Gilissen,†,§ Pascal de Tullio,† Lars Olsen,‡
Karla Frydenvang,‡ Bernard Pirotte,† Julien Hanson,*,†,§ Robert E. Oswald,*,∥
Jette Sandholm Kastrup,*,‡ and Pierre Francotte*,†
†Laboratory of Medicinal Chemistry, Center for Interdisciplinary Research on Medicines (CIRM), University of Lieg
̀
e, Quartier
̂ ̀
Hopital B36 Av. Hippocrate 15 B-4000 Liege, Belgium
‡Biostructural Research, Department of Drug Design and Pharmacology, Faculty of Health and Medical Sciences, University of
Copenhagen, DK-2100 Copenhagen, Denmark
§Laboratory of Molecular Pharmacology, GIGA-Molecular Biology of Diseases, University of Lieg
̀ ̀
e, Liege, Belgium
∥Department of Molecular Medicine, Cornell University, Ithaca, New York 14850, United States
S
* Supporting Information
ABSTRACT: We report here the synthesis of 7-phenoxy-
substituted 3,4-dihydro-2H-1,2,4-benzothiadiazine 1,1-dioxides
and their evaluation as AMPA receptor positive allosteric
modulators (AMPApams). The impact of substitution on the
phenoxy ring and on the nitrogen atom at the 4-position was
examined. At GluA2(Q) expressed in HEK293 cells (calcium
flux experiment), the most potent compound was 11m (4-
cyclopropyl-7-(3-methoxyphenoxy)-3,4-dihydro-2H-1,2,4-ben-
zothiadiazine 1,1-dioxide, EC50 = 2.0 nM). The Hill coefficient
in the screening and the shape of the dimerization curve in
small-angle X-ray scattering (SAXS) experiments using isolated
GluA2 ligand-binding domain (GluA2-LBD) are consistent
with binding of one molecule of 11m per dimer interface, contrary to most benzothiadiazine dioxides developed to date. This
observation was confirmed by the X-ray structure of 11m bound to GluA2-LBD and by NMR. This is the first benzothiadiazine
dioxide AMPApam to reach the nanomolar range.
discovery of cognitive enhancers and a potential therapeutic
tool in the management of Alzheimer’s disease.5 This strategy
was even found more promising considering their ability to
induce the release of neurotrophic factors in vivo, like BDNF.6,7
While AMPAR agonists are expected to exert excitotoxic
effects, AMPAR positive allosteric modulators (so-called
“AMPApams” or AMPAR potentiators) seem more promising,
as they are able to fine-tune AMPAR signals through their
binding at allosteric sites of the receptors and need the
presence of endogenous glutamate to be active.
INTRODUCTION
■
L-Glutamate is well-known as the key excitatory neuro-
transmitter in the central nervous system, acting through
activation of metabotropic receptors (mGluRs, coupled to G-
protein) and ionotropic receptors (iGluRs, cation channels).1,2
iGluRs are recognized to play a crucial role in the fast excitatory
synaptic transmission. They have been classified into three
subtypes named after their affinity for nonendogenous ligands:
N-methyl-D-aspartic acid (NMDA), α-amino-3-hydroxy-5-
methyl-4-isoxazolepropionic acid (AMPA), and kainic acid
(KA). Among these three subtypes, AMPA receptors
(AMPARs) have attracted attention, assuming that upregula-
tion of these receptors could lead to cognition enhancement.
This idea has been reinforced, taking into account their critical
involvement in synaptic plasticity3 and their involvement in the
mechanism of action of nootropic drugs.4 AMPARs are thus
considered as an attractive and appropriate target for the
AMPARs are homo/heterotetrameric complexes assembled
with GluA1−4 subunits.8 Each subunit contains an amino-
terminal domain involved in subunit assembly (ATD), a
clamshell-shaped domain providing the binding pocket for
Received: September 7, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
glutamate (LBD), a transmembrane domain forming the
permeation channel for cations (TMD) and a cytoplasmic
terminal domain (CTD) associated with processes such as
receptor trafficking or regulation.9 By their ability to form
bridging interactions between LBD dimers, positive allosteric
modulators stabilize the receptor in its open state and thus slow
deactivation and/or desensitization processes.5,10 This may be
achieved either with one or two potentiator molecules per
dimer, thus with two or four potentiator molecules per
AMPAR,11 depending on the occupancy of the A/B/B′/C
and C′ subsites, which constitute the binding cavity, as
previously defined.12
through characterization with SAXS, NMR, and crystallog-
raphy.
RESULTS AND DISCUSSION
■
Synthesis. The synthetic pathway used to prepare the 7-
phenoxy-substituted 3,4-dihydro-2H-1,2,4-benzothiadiazine
1,1-dioxides reported here is illustrated in Scheme 1.
The 7-phenoxy-substituted 3,4-dihydro-2H-1,2,4-benzothia-
diazine 1,1-dioxides 11 series were prepared from the
corresponding intermediates 10 series through a Chan−Lam
coupling reaction with the appropriate phenylboronic acid in
the presence of copper(II) acetate in dichloromethane. While
the intermediate 10b was obtained as previously described,16 its
4-methylated analogue 10a was prepared via an alternate
pathway, starting from 7-methoxy-4H-1,2,4-benzothiadiazine
1,1-dioxide 7.19 After alkylation on the 4-nitrogen atom using
methyl iodide in the presence of potassium carbonate in
acetonitrile and saturation of the double bond in the 2,3-
positions by means of sodium borohydride, conversion of the
methoxy at the 7-position into a hydroxyl group was achieved
with boron tribromide in dichloromethane giving access to 10a.
Substituents inserted on the nitrogen atom at the 4-position
were selected according to the previously reported structure−
activity relationships.19,20 While the introduction of the
cyclopropyl group in this position was clearly the optimal
choice and could thus not be neglected, we decided to also
investigate the impact of the insertion of the smallest alkyl
radical, namely the methyl group.
Among the different AMPAR potentiators,13 3,4-dihydro-2H-
1,2,4-benzothiadiazine 1,1-dioxides constitute probably the
most investigated chemical class, as exemplified by recent
publications (Figure 1).14−17
Measurement of GluA2-Induced Calcium Flux in
HEK293 Cells. The new set of compounds was evaluated in
vitro as AMPAR potentiators using a fluorescence-based
calcium assay performed on HEK293 cells stably expressing
the AMPAR GluA2(Q) in order to evaluate the effect of the
compounds on the glutamate-evoked channel opening. For
each compound, the pEC50 (negative logarithm of the
modulator concentration responsible for 50% of the glutamate
maximal effect at 1 mM) value was determined. Results
obtained with this newly developed pharmacological assay were
in good agreement with results obtained with previously used
methods: for example, the mean EC50 values obtained with a
previous fluorescence-based calcium measurement assay
induced by 300 μM AMPA on primary cultures of neurons
from a rat embryonic cortex21 for 5, 11h, and 11o were 0.9, 1.0,
Figure 1. Structures of selected AMPAR potentiators belonging to 3,4-
dihydro-2H-1,2,4-benzothiadiazine 1,1-dioxides.
Taking advantage of the observation by Cordi et al. that an
aryloxy or a heteroaryloxy group at the 7-position of
benzothiadiazine dioxides could favorably impact the potentia-
tor activity on AMPARs,18 the present work focused on the 7-
phenoxy-3,4-dihydro-2H-1,2,4-benzothiadiazine 1,1-dioxides,
exploring the influence of monosubstitution of the phenoxy
group introduced at the 7-position. Particular attention was
paid to the study of the binding mode of the new compounds
a
Scheme 1. Synthetic Pathway of 7-Phenoxy-Substituted 3,4-Dihydro-2H-1,2,4-benzothiadiazine 1,1-Dioxides
a
Conditions: (i) CH3I, K2CO3, CH3CN, 60°C, 3 h (81%); (ii) NaBH4, 2-propanol, rt, 45 min (78%); (iii) BBr3, CH2Cl2, rt, 6 h (76%); (iv)
adequately substituted phenylboronic acid, Cu(OAc)2, molecular sieves, CH2Cl2, 40°C, 5 h (13−74%).
B
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Table 1. Potentiating Effects of 7-Phenoxy-Substituted 3,4-Dihydro-2H-1,2,4-benzothiadiazine 1,1-Dioxides on the Calcium
Flux Induced by 1 mM Glutamate on HEK293 Cells Stably Expressing the GluA2(Q) Subunit and Hill Coefficient
a
b
c
compound
R4
R
pEC50
corresponding EC50
Hill coefficient
3
4.690 0.013 (3)
4.728 0.012 (3)
6.090 0.004 (3)
4.296 0.008 (3)
5.118 0.019 (3)
<5 (3)
20.4
18.7
0.81
50.6
7.62
10.0
0.39
0.95
>10.0
0.37
5.90
0.71
15.6
0.05
3.19
15.60
0.002
1.82
0.45
3.0 0.2 (3)
3.0 0.1 (3)
3.5 0.1 (3)
3.0 0.1 (3)
1.4 0.1 (3)
nd
4
5
12
11a
11b
11c
11d
11e
11f
11g
11h
11i
11j
11k
11l
11m
11n
11o
CH3
H
CH3
o-CH3
CH3
m-CH3
p-CH3
o-OCH3
m-OCH3
p-OCH3
H
6.408 0.015 (3)
6.020 0.013 (3)
<5 (3)
2.3 0.2 (3)
2.1 0.1 (3)
nd
CH3
CH3
CH3
6.431 0.030 (3)
5.229 0.035 (3)
6.147 0.033 (3)
4.807 0.022 (3)
7.336 0.027 (3)
5.496 0.021 (3)
4.807 0.018 (3)
8.700 0.049 (6)
5.740 0.023 (3)
6.343 0.018 (3)
2.3 0.4 (3)
1.7 0.2 (3)
1.3 0.1 (3)
1.7 0.1 (3)
1.1 0.1 (3)
1.9 0.2 (3)
2.3 0.2 (3)
1.4 0.2 (6)
1.5 0.1 (3)
CH3
CH(CH2)2
CH(CH2)2
CH(CH2)2
CH(CH2)2
CH(CH2)2
CH(CH2)2
CH(CH2)2
CH(CH2)2
o-CH3
m-CH3
p-CH3
o-OCH3
m-OCH3
p-OCH3
m-CN
1.7 0.1 (3)
a
b
pEC50 = negative logarithm of AMPAR potentiator concentration responsible for 50% of the maximal effect (mean
SEM (n)). Mean
c
concentration responsible for 50% of the maximal effect, in μM. Mean value SEM (n).
Figure 2. Measurement of GluA2-induced calcium flux in HEK293 cells. (A) Comparison of two activity curves obtained with 11d and 5. (B)
Comparison of the mean Hill coefficient obtained with the two sets of compounds. The boxes represent the distribution of the values corresponding
to the Hill coefficient collected for the two sets; they include a medium line depicting the corresponding mean value for each set and the error bars
corresponding the standard error to the mean.
and 0.4 μM, respectively (n = 3). These values are close to
those obtained here in heterologously transfected HEK293 cells
(EC50 values of 0.81, 0.71, and 0.45 μM, respectively; Table 1).
As previously noticed in other series,20 the introduction of a
cyclopropyl group on the nitrogen atom at the 4-position is in
most cases more favorable for the activity than introduction of a
methyl group (compare 11a vs 11h, 11c vs 11j or 11f vs 11m).
Exploration of the nature of the substitution of the phenoxy
moiety was limited but showed that a methyl group seemed to
have the same impact on activity as a methoxy group (see 11b/
11c/11d vs 11e/11f/11g and 11i/11k vs 11l/11n). However,
in the 4-cyclopropyl series, the methoxy group surprisingly
exerts a drastic change of the activity profile when positioned in
the meta position of the phenoxy moiety (see 11m, EC50 = 2.0
nM vs 11j, EC50 = 46.1 nM). This leads to the most powerful
AMPAR potentiator ever designed in our series, namely
compound 11m.
The position of the substitution was found critical for the
activity: the rank order of potency was observed following the
sequence meta > para > ortho. While the meta position for the
phenoxy substitution was found to be the best option, the
absence of a substitution was found to have a neutral to rather
C
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 3. GluA2-LBD-N775S dimerization using SAXS. (A) Dimerization models assuming one site per dimer interface and two sites per dimer
interface. K4 is the dimerization constant and K3 is the intrinsic binding constant. R is the GluA2-LBD-N775S monomer and M is the free modulator
concentration. The model is written to exclude cooperativity of binding. (B) The fraction of dimer observed in the SAXS experiment as a function of
the total modulator concentration using various allosteric modulators and different concentrations of GluA2-LBD-N775S. For compound 11m, the
curves are shown in different shades of blue, with the lightest corresponding to the highest concentration of protein. The change in shape of the
curve is due to the fact that the protein concentration is in the same range as the modulator concentration, so that it is particularly limiting in the
lower ranges of modulator concentration. The lowest protein concentration in this series has a shallow slope similar to that of HFMZ, suggesting one
binding site per dimer interface. The steeper slope of the 11m curve at increasing protein concentrations is caused by the modulator/protein ratio
depleting the free modulator. At higher protein concentrations, the dimer fraction becomes particularly limited by the lower range of modulator
concentrations.
positive effect comparing to ortho or para substitution (see 11a
vs 11b/11e/11g or 11h vs 11i/11k/11l/11n). This trend was
only contradicted with the methyl group in the para
substitution (11a vs 11d) in the series bearing a methyl
group on the nitrogen atom at the 4-position.
Based on the fact that the meta position of the phenoxy
group was critical, the preparation of a 7-phenoxy compound
substituted with a cyano moiety at the meta position was
envisaged in order to evaluate the influence of a strong
electron-withdrawing group. This compound (11o) was found
less active than the methyl or methoxy analogues (11j and
11m) and nearly as potent as the unsubstituted phenoxy
compound (11h). This observation could be explained by the
steric interactions of the corresponding group in the meta
position.
Looking at the slope of the concentration−response
sigmoidal curves generated in our new functional assay (see
Figure 2A for two representative profiles obtained with 11d and
reference compound 5, respectively) and keeping in mind the
observations previously made with SAXS data,11 the Hill
coefficient22 was examined for each curve. This feature of the
curve reflects the existence of multiple binding sites in the
receptor and enables determination of the binding stoichiom-
etry. This permitted to compare the new compounds 11 and
the references 3, 4, 5, and 12 (see Table 1 and Figure 2B). The
collected data suggest a distinct stoichiometry for 7-phenoxy
compounds 11 compared to reference compounds 3, 4, 5, and
12; whereas reference compounds bind in two copies on the
interface (as previously demonstrated for 523), the compounds
11 bind in one copy. This trend observed here might be
explained by a difference in the binding mode within the dimer
interface. The figures suggest that the new set of compounds
constitutes a unique subclass in our series of benzothiadiazine
dioxides designed to date (Figure 2B). Further investigations
(SAXS, crystallography, and NMR) presented hereafter
corroborate this observation.
Small-Angle X-ray Scattering (SAXS). The isolated
GluA2-LBD-N775S shows only a weak propensity for
dimerization, but the ability of allosteric modulators to stabilize
dimer formation can be used to assess both affinity and
stoichiometry.11 SAXS provides a measure of GluA2-LBD-
N775S dimerization because the scattering profile of the
GluA2-LBD-N775S changes markedly upon dimerization
allowing the contributions of monomer and dimer curves to
be determined by standard fitting routines (OLIGOMER).24
The fraction of dimer at each modulator condition is then
modeled using the schemes shown in Figure 3A. All curves
were fit simultaneously, with a common K4 value (dimerization
constant) and K3 values (intrinsic equilibrium dissociation
constants) specific to the individual modulator. The choice of
model (one or two sites per dimer interface) was based on the
slope of the curve and the fit of the model to the data. As
shown in Figure 3B, the dependence of dimerization on the
concentration of allosteric modulator can have quite different
slopes depending upon which modulator is used. As shown
previously,11 those modulators that bind to one site per dimer
interface exhibit a relatively shallow slope (HFMZ in Figure
3B),12 while those that bind to two sites show a much steeper
slope (cyclothiazide 1).25 This is not due to cooperativity of
modulator binding but rather a consequence of the fact that the
two-site model assumes that dimerization is stabilized further
with each successive occupancy. Two of the compounds in
Figure 1 (5 and 3) show a steeper slope consistent with binding
to two sites per dimer interface (consistent with the crystal
D
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
structures of 523 and 3, see below in this paper). Reference
compound 5 (K3 of 9 × 10−7 M) is somewhat more potent
than reference compound 3 (K3 of 2 × 10−6 M). This is in
contrast to the calcium flux experiments that show a
considerably larger difference between the two compounds.
The SAXS experiments are not directly comparable to
measurements of the intact receptor because the mechanism
involves the dimerization of two identical soluble proteins with
an extremely weak dimerization constant (approximately 32
mM), leading to an EC50 considerably higher than that
measured in the intact receptor. K3 is the intrinsic binding
constant for the interaction of the modulator with the GluA2-
LBD-N775S dimer. This, in principle, should be the same in
the intact receptor. The difference is that in the intact receptor,
the two monomers within the dimer are held in close proximity
by the structure, and the formation of the dimer interface can
be seen as a favorable conformation transition rather than a
weak protein−protein interaction involving the diffusion of two
independent proteins. For this reason, the EC50 measured in
the intact receptor is generally lower than that observed in the
SAXS measurements. (See the supplement to Ptak et al.11 for a
further discussion of this point.) The calcium flux measure-
ments suggest that compound 11m is several orders more
potent than the reference compounds 3 and 5. This raised a
difficulty in the SAXS measurements because typical protein
concentrations are higher than the affinity for this compound,
suggesting that modulator rather than protein would be
limiting. This was approached by using several low concen-
trations of GluA2-LBD-N775S and signal averaging three
independent data sets. The shapes of the dimerization curves
indicated that at some of the concentrations tested, the
modulator was indeed limiting. This is reflected in the shape of
the dimerization curves, which become increasingly steep as the
protein concentration increases (Figure 3B). By using a
numerical fitting routine described in the Experimental Section
that takes into account the limiting modulator concentration,
the affinity of compound 11m could be determined for a model
assuming one binding site per dimer interface. At the lowest
protein concentration, the dimerization slope was similar to
that of HFMZ, further supporting a model with one site per
dimer interface, consistent with the crystal structure described
below. The four curves for compound 11m shown in Figure 3B
were generated with a K4 of 32 mM and K3 of 5 × 10−11 M.
The change in the shape of the curve with increasing
concentration does not reflect an increase in stoichiometry
but is solely a predictable consequence of a limiting modulator
concentration. In fact, this change in shape provided a
significant constraint on the value of K3. Overall, the SAXS
measurements show that compound 3 and 5 both bind in two
copies per dimer interface and are of considerably lower affinity
than compound 11m, which occupies only one site per dimer
interface.
forming two biological dimers (molA/molB (Figure 4A) and
molC/symmetry-related molC). Glutamate is present in the
orthosteric-binding site of all three protein molecules, leading
to D1-D2 domain closures of 19−21°.
Unambiguous electron density was identified at the interface
of both dimers corresponding to 11m (Figure 4B). As
suggested by the Hill coefficient in the calcium flux assays
and the shape of the dimerization curve in the SAXS data, only
one molecule of 11m can occupy the modulator-binding sites at
a time, leading to two alternative binding modes with the BTD-
scaffold positioned in one of the two binding sites seen in the
structure of GluA2-LBD with compound 5.23 Due to the
symmetrical dimer, the binding mode of the two molecules of
11m is rotated 180° relative to each other. The occupancy of
the modulator and corresponding water molecules was refined
to 0.55/0.45 for molA/molB and 0.46 for molC. The 7-
substituent occupies the second binding site together with three
water molecules (W1−W3), occupying the same space in the
binding site as the sulfonamide of 11m in the alternate binding
mode. Compound 11m belongs to the shifted thiazide class of
modulators, with the BTD scaffold primarily located in subsite
C but the 7-(3′-methoxyphenoxy) substituent spanning subsite
A with the 3′-methoxy moiety oriented toward subsite B′
(Figure 4C). Previous studies of thiazide modulators (class i
and ii) have shown that most of these compounds do not reach
into subsite A and thus can bind two molecules in the dimer
interface.12,26
One hydrogen bond is formed from the sulfonamide N atom
of 11m to the backbone O atom of Pro515. The three water
molecules (W1−W3) found to occupy the second BTD
binding site together with the 7-(3′-methoxyphenoxy)
substituent do not form polar contacts with the 7-(3′-
methoxyphenoxy) substituent, but instead establish a network
of polar contacts from the backbone O atom of Pro515 on one
side of the dimer to the backbone N atom of Gly752 and O
atom Ile502 on the other side (Figure 4C). Multiple nonpolar
interactions are possible from 11m to surrounding residues
within 4 Å. The BTD scaffold forms contact to Lys514, Pro515,
Phe516, Met517, Ser518, Asn775, and Leu780 from one
protein chain and Ile502, Pro515, Ser518, Ser750, Lys751, and
Gly752 from the opposite protein chain in the dimer
comprising molA and molB. Almost identical interactions are
seen in the symmetry-generated dimer (molC/sym-molC);
however, in molC, the contact to Ser518 is replaced by a
contact to Leu772 in the same chain. Although no polar
contacts are established from the 7-(3′-methoxyphenoxy)
substituent to surrounding residues, the substituent engages
in nonpolar interactions with nine of the 12 same residues as
the BTD scaffold: Ser750 and Lys751 from one protein chain,
and Pro515, Phe516, Met517, Ser518, Gly752, Asn775, and
Leu780 from the opposite chain. Thus, binding of one molecule
of 11m to a large extent recreates the interactions seen when
two molecules of 5 bind in the dimer interface. In addition, the
benzene ring of the 7-phenoxy substituent is positioned to form
pi-stacking interactions with the backbone amide bonds
between Lys751-Gly752 in one subunit and Met517-Ser518
in the other subunit. These observations, combined with
smaller entropy loss upon binding of one instead of two
molecules, may explain the 40-fold better potency of 11m
(EC50 = 2.0 nM) compared to 5 (EC50 = 813 nM).
Structure Determination. To investigate the binding
mode of the highly potent compound 11m, and especially
how the larger substituent in the 7-position is located in the
binding site, the GluA2-LBD was crystallized in complex with
11m. In addition, the GluA2-LBD-L504Y−N775S was crystal-
lized with the reference compound 3. This compound differs
from previously crystallized BTDs as it contains a fluoro-ethyl
group at N-4 and chlorine at the C-7 position. The structures
were determined at 2.00 and 1.78 Å, respectively (Table 2).
GluA2-LBD with 11m. The GluA2-LBD in complex with
11m crystallized with three molecules in the asymmetric unit,
The activity for the newly synthesized series of compounds
was found to correlate with the position of the substituent on
the 7-phenoxy moiety in the following order: meta > para >
E
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Table 2. Crystal Data, Data Collection, and Refinement Statistics for the GluA2-LBD-L504Y−N775S in Complex with 3 and
GluA2-LBD in Complex with 11m
complex
3
11m
Crystal Data
PDB entry
5O9A
5OEW
space group
P21
P21212
unit cell: a, b, c (Å)
units cell: β (°)
molecules in a.u.
47.31, 98.41, 121.79
114.70, 164.37, 47.33
90.53
4
90
3
a
Data Collection
wavelength (Å)
0.979
1.000
b
resolution (Å)
49.20−1.78 (1.88−1.78)
106 755 (15 582)
3.7 (3.6)
29.27−2.00 (2.11−2.00)
61 361 (8 859)
4.1 (4.1)
no. of unique reflections
average redundancy
completeness (%)
100 (100)
99.8 (99.8)
0.087 (0.313)
6.5 (2.0)
c
Rmerge
0.063 (0.377)
10.4 (2.0)
I/σI
Wilson B (Å2)
13.6
17.5
Refinement
amino acid residues
263/264/263/260
261/259/261
3/2
glutamate/modulator
4/4
acetate/sulfate/Zn/Cl
−/16/−/7
3/1/1/1/1186
4/−/18/−
−/−/−/−/522
0.17/0.19
27/28/23
18/15
glycerol/PEG/ethylene glycol/citrate/water
d
e
Rwork /Rfree
0.15/0.18
average B values (Å2) for amino acid residues
average B values (Å2) for glutamate/modulator
average B values (Å2) for acetate/sulfate/Zn/Cl
18/25/17/24
12/15
−/60/−/53
49/47/46/30/29
46/-/64/-
−/−/−/−/34
0.03
average B values (Å2) for glycerol/PEG/ethylene glycol/citrate/water
RMSDs: bond lengths (Å)
0.01
0.79
0
RMSDs: angles (°)
Ramachandran plot (%): outliers
0.61
0
Ramachandran plot (%): favored
99.2
1.0
0
99.4
f
rotamer outliers
1.0
f
Cβ outliers (%)
0
f
clash score
0.74
0.56
a
b
c
a.u.: asymmetric unit of the crystal. Values in parentheses correspond to the outermost resolution shell. Rmerge is calculated as follows: Ii(hkl) is the
intensity of an individual measurement of the reflection with Miller indices hkl, and I(hkl) is the intensity from multiple observations. Rmerge = ∑hkl
d
∑i|Ii(hkl) − I(hkl)| /∑hkl ∑i |Ii(hkl)|. Rwork = ∑hkl |Fobs − Fcalc|/∑hkl |Fobs| where Fobs and Fcalc are the observed and calculated structure factor
e
amplitudes, respectively, for reflection hkl. Rfree is equivalent to Rwork but is calculated with 5% of reflections omitted from the refinement process.
f
MolProbity statistics.45
ortho. From the structure of 11m bound to GluA2-LBD, this
ranking might be explained. Introduction of substituents in the
ortho position, even a methyl group, would require changes in
the conformation or binding mode of the modulator or protein
rearrangements to avoid steric clash with Met517-Ser518.
Likewise, large substituents in the para position will lead to
clash with protein residues and interference with the
intermolecular hydrogen bond between the side-chain ND2
atom of Asn775 and the backbone O atom of Ser750, whereas
the meta position of the 7-phenoxy moiety holds the most
space for introduction of substituents. The m-OCH3 in 11m
has several van der Waals contacts to both protein molecules,
which would be less with an m-CH3 group (11j). This might
explain the better potency of 11m compared to 11j. However,
an m-CN as in 11o would lead to steric clash with the protein
and would require changes in the conformation or binding
mode of the modulator or protein rearrangements.
several studies addressing modulator binding (see Krintel et
al.16): the complex crystallized with four molecules in the
asymmetric unit, forming two biological dimers (molA/molB
(Figure 5A) and molC/molD). Glutamate binds in all four
protein molecules, leading to D1-D2 domain closures of 20−
22°.
Unambiguous electron density was identified at the interface
of both dimers corresponding to 3 (Figure 5B). Two molecules
of 3 bind at the dimer interface in subsite C, as seen with other
BTD containing compounds like 523 and 12.21 One hydrogen
bond is formed from the sulfonamide N atom of 3 to the
backbone O atom of Pro515 (Figure 5B) as also seen in the
GluA2-LBD structure with 11m. Ser518 adopts two different
side-chain conformations in all four protein molecules, one
pointing away from the modulator and one pointing toward it.
When the side chain of Ser518 points toward the modulator, it
allows for the formation of a hydrogen bond at the 7-chlorine
atom (distances of 3.1−3.2 Å). However, the occupancy of the
side chain in this conformation is only 30%, consistent with the
fact that the chlorine atom is expected to be a poor proton
GluA2-LBD-L504Y−N775S with 3. Compound 3 was
crystallized in complex with GluA2-LBD-L504Y−N775S in the
presence of glutamate. This mutant has previously been used in
F
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
Figure 4. Structure of GluA2-LBD in complex with glutamate and the AMPAR potentiator 11m. (A) Cartoon representation of the GluA2-LBD
dimer (molA in white and molB in dark salmon) with glutamate shown in blue sticks representation and the two binding orientations of 11m in
green and black, respectively. Nitrogen atoms are blue, oxygen atoms red, and sulfur atoms yellow. (B) Compound 11m binds to one site in the
dimer interface. However, because of the symmetry of the dimer interface, two overlapping copies of the compound are observed in the crystal
structure, only one of which is present in any given dimer. A 2Fo-Fc OMIT map (gray mesh, contoured at 0.5 σ and carved around the modulator at
2.0 Å) is shown with the two orientations of 11m. Six water molecules are shown as spheres: green water molecules are present together with 11m in
green orientation and black water molecules with 11m in black orientation. (C) Compound 11m primarily occupies subsites A and C in the
allosteric binding site. GluA2 residues are shown in line representation, green orientation of 11m in sticks representation. Hydrogen bonds are
shown as dashed black lines and water molecules within 4 Å of 11m as red spheres. The binding mode of two molecules of 5 (light green sticks) is
shown for comparison.
acceptor.27 Introduction of larger substituents at the 7-position
is likely to cause steric clash with Met517 and Ser518, in
agreement with previous observations that the potency
decreases with iodine and methyl substituents.19 However,
the introduction of a smaller substituent in the 7-position, such
as the fluorine analogue of 3 (EC2x = 11.3 μM), leads to an
increase in potency compared to 3 with chlorine in the 7-
position (EC2x = 18.8 μM).28
neighboring modulator molecule with a CCl···F bond angle
of 165° (Figure 5B). The ClF distances between the two
neighboring molecules of 3 are 4.1−4.4 Å. Chlorine, due to the
charge distribution of a soft halogen, has an electropositive tip,
called sigma-hole (Figure 5C).29,30 This leads to a favorable
interaction between the chlorine and the electronegative
fluorine. The aromatic ring system in 3 shows van der Waals
contacts to the 7-chlorine atom of the neighboring molecule of
3 with a shortest distance of 3.4 Å.
The alkyl chain of 3 is bent to a U-shaped conformation,
where the fluorine atom points toward the chlorine atom of the
G
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
NMR Spectroscopy. The GluA2-LBD-N775S was uni-
formly labeled with 15N and, on an 800 MHz spectrometer,
1H,15N-HSQC-TROSY spectra were collected in the presence
and absence of 90 μM compound 11m. A significant fraction of
the protein was dimerized as indicated by line broadening and
the presence of new broadened peaks. Peaks in the dimer
interface formed by the binding of 11m were significantly
shifted and not readily assigned in this spectrum. However, on
the periphery of the dimer interface, the change in chemical
environment was less pronounced such that the peak shifts
were smaller. An example is Gly760 (Figure 6), which clearly
Figure 6. GluA2-LBD-N775S dimerization using NMR spectroscopy.
(A) Portion of the H,15N-HSQC-TROSY spectra at 800 MHz showing
monomeric GluA2-LBD-N775S in black and the partially dimerized
spectrum bound to compound 11m in red. (B) Structure of the dimer
of GluA2-LBD-N775S (PDB entry 3DP6) with G760 shown as a blue
sphere.
shows two new peaks formed by the binding of compound 11m
and dimerization. Simple dimerization would not produce two
new peaks because the dimer interface is symmetrical.
Consistent with the crystal structure and SAXS results, the
binding of one molecule of compound 11m to the dimer
interface breaks that symmetry, producing two peaks.
Figure 5. Structure of GluA2-LBD-L504Y−N775S in complex with
glutamate and the AMPAR potentiator 3. (A) Cartoon representation
of the GluA2-LBD-L504Y−N775S dimer (molA in white and molB in
dark salmon) with glutamate shown in a blue stick representation and
two molecules of 3 in orange. Nitrogen atoms are blue, oxygen atoms
red, and sulfur atoms yellow. (B) The two molecules of compound 3
primarily occupies subsites C in the allosteric binding site. GluA2
residues are shown in line representation, 3 in stick representation, and
water molecules within 3.5 Å of modulator as red spheres. Hydrogen
bonds up to 3.2 Å are shown as dashed black lines. F−Cl distances
(4.1−4.4 Å) are indicated as red dashed lines. A 2Fo-Fc OMIT map is
shown as a gray mesh, contoured at 1 σ and carved around the
modulator at 1.8 Å. (C) The electrostatic potential map on the
electron density map of 3 is represented in rainbow colors. Potentials
below −15 kcal/mol are shown with red and above +20 in magenta.
The electropositive tip (3 kcal/mol) of chlorine can make a halogen
bond to the electronegative (−14 kcal/mol) fluorine. The electrostatic
potential of 3 was calculated using Jaguar in MAESTRO (v.10.1;
CONCLUSION
■
With the aim of exploring the influence of monosubstitution of
the phenoxy ring introduced at the 7-position of the
benzothiadiazine dioxide scaffold, we developed a synthetic
pathway giving access to the preparation of 15 compounds. We
also developed a new medium-throughput fluorimetric assay in
order to screen this series.
The first part of the work led to highlight the high potency of
11m, with an EC50 = 2.0 nM (calcium flux experiment); this in
vitro activity was found to be the highest in comparison with
our previous lead structures. Interestingly, the activity curves
obtained for the new compounds shared a much lower Hill
coefficient than the curves obtained with reference compounds
from our library. Considering this interesting activity profile,
Schrodinger, LLC, New York, NY, 2015) with the cc-PVDZ basis set
̈
and B3LYP hybrid potential.
H
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
(3.0 g, 80 mmol). After 45 min of stirring the mixture at room
temperature, the solvent was removed by distillation under reduced
pressure and the residue was suspended in water (250 mL). The
alkaline suspension was adjusted to pH 7 with 0.1 N HCl and
extracted 3-fold with chloroform (3 × 250 mL). The combined
organic layers were dried over MgSO4 and filtered. The filtrate was
concentrated to dryness under reduced pressure, and the residue of the
title compound was recrystallized in methanol/water 1:2 (4.7 g, 78%
attention was thus focused on the characterization of the
binding mode of the most active compound 11m in the series
via complementary SAXS, NMR, and crystallography.
SAXS data first corroborated the high affinity of 11m and
confirmed that this compound occupies only one site per dimer
interface, contrary to most benzothiadiazine dioxides developed
as AMPAR potentiators. Structure determination of GluA2-
LBD in complex with glutamate and 11m confirmed that only
one molecule of 11m binds at the dimer interface and revealed
the detailed binding mode. Compound 11m primarily occupies
subsites A and C. Results obtained with NMR spectroscopy
were found to be consistent with the X-ray and SAXS data.
From these observations, compound 11m behaves as two
molecules of the reference compound 5, since 11m is able to
recreate the interaction observed after two molecules of
compound 5 binds into the allosteric pocket. Altogether, the
data collected in this second part confirmed that compound
11m belongs to a singular subclass of the benzothiadiazine
dioxides developed so far.
The binding mode of compound 3 was characterized using
SAXS and crystallography. Compound 3 binds in two copies
per dimer interface and occupies subsites C within the allosteric
pocket. A chlorine−fluorine interaction between the two
neighboring molecules was found.
This study suggests that 3,4-dihydro-2H-1,2,4-benzothiadia-
zine 1,1-dioxide is (still) a very interesting scaffold to develop
new powerful AMPAR potentiators. Keeping this idea in mind,
efforts should now be focused toward the design of new BTDs
in order to further map the GluA2-LBD and develop powerful
pharmacological tools useful to refine knowledge of the
AMPAR biology.
yield): mp 127−129 °C (lit.18 126−128 °C); H NMR (DMSO-d6,
1
500 MHz) δ 2.89 (s, 3H, NCH3), 3.73 (s, 3H, 7-OCH3), 4.57 (s, 2H,
3-CH2), 6.87 (d, J = 9.2 Hz, 1H, 5-H), 7.04 (d, J = 3.0 Hz, 1H, 8-H),
7.09 (dd, J = 9.2, 3.0 Hz, 1H, 6-H), 8.05 (s, 1H, 3-H); 13C NMR
(DMSO-d6) δ 36.7 (NCH3), 55.6 (O−CH3), 62.4 (C-3), 107.4 (C-8),
116.2 (C-5), 121.0 (C-6), 123.3 (C-4a), 138.8 (C-8a), 150.5 (C-7).
7-Hydroxy-4-methyl-3,4-dihydro-2H-1,2,4-benzothiadiazine
1,1-Dioxide (10a). The solution of 7-methoxy-4-methyl-3,4-dihydro-
2H-1,2,4-benzothiadiazine 1,1-dioxide (9) (5.0 g, 21.9 mmol) in
dichloromethane (200 mL) was cooled at −30 °C and then
supplemented with boron tribromide (16 mL). After 6 h of stirring
the mixture at room temperature, water (200 mL) was carefully
poured in the reaction mixture and the organic solvent was removed
by distillation under reduced pressure. The aqueous layer was
extracted with ethyl acetate (3 × 200 mL). The collected organic
layers were dried over MgSO4 and filtered. The filtrate was evaporated
to dryness, and the residue was crystallized in dichloromethane and
dried (76% yield): mp 165−166 °C (lit.18 168−172 °C); H NMR
1
(DMSO-d6) δ 2.85 (s, 3H, NCH3), 4.52 (s, 2H, 3-CH2), 6.78 (d, J =
9.8 Hz, 1H, 5-H), 6.91 (m, 2H, 6-H/7-H), 7.99 (bs, 1H, 2-H), 9.31
(bs, 1H, 7-OH); 13C NMR (DMSO-d6) δ 36.9 (NCH3), 62.5 (C-3),
109.2 (C-8), 116.5 (C-5), 121.4 (C-6), 123.8 (C-4a), 137.6 (C-8a),
148.6 (C-7).
General Synthetic Pathway to Obtain 7-Phenoxy-Substi-
tuted 4-Alkyl-3,4-dihydro-2H-1,2,4-benzothiadiazine 1,1-Diox-
ides 11. A solution of the appropriate 4-alkyl-3,4-dihydro-2H-1,2,4-
benzothiadiazine 1,1-dioxide 10 (1.7 mmol) in dichloromethane (60
mL) and pyridine (10 drops), molecular sieves (5 g), the adequately
substituted phenyl boronic acid (2.5 mmol), and copper(II) acetate
(2.5 mmol) was stirred at 40 °C. After 5 h, dichloromethane was
added to the medium and the solid was removed by filtration. The
filtrate was evaporated under reduced pressure before the addition of
acidic water (50 mL). The resulting suspension was extracted with
ethyl acetate (3 × 50 mL). The organic phases were collected and
dried over anhydrous MgSO4. The removal of the solvent followed by
flash column chromatography on silica gel (dichloromethane) and
subsequent recrystallization (dichloromethane/petroleum ether 5:15)
afforded the pure 11 compounds (yields 13−74%).
4-Methyl-7-phenoxy-3,4-dihydro-2H-1,2,4-benzothiadiazine 1,1-
Dioxide (11a): 74% yield, mp 143.5−145.5 °C; 1H NMR (DMSO-d6)
δ 2.94 (s, 3H, NCH3), 4.65 (s, 2H, 3-CH2), 6.94 (m, 3H, 5-H/2′-H/
6′-H), 7.12 (m, 2H, 4′-H/8-H), 7.22 (dd, J = 9.1, 2.8 Hz, 1H, 6-H),
7.38 (t, J = 7.9 Hz, 1H, 3′-H/5′-H), 8.11 (bs, 1H, NH); 13C NMR
(DMSO-d6) δ 36.4 (NCH3), 62.3 (C-3), 114.3 (C-8), 116.0 (C-5),
117.5 (C-2′/C-6′), 123.0 (C-8a), 123.1 (C-4′), 125.2 (C-6), 130.0 (C-
3′/C-5′), 140.9 (C-4a), 146.2 (C-7), 157.5 (C-1′). Anal.
(C14H14N2O3S) Calcd: C, 57.91; H, 4.86; N, 9.65; S, 11.04. Found:
C, 57.96; H, 4.93; N, 10.04; S, 10.80.
EXPERIMENTAL SECTION
■
General Procedures. Melting points were determined on a Buchi
̈
Tottoli capillary apparatus and were uncorrected. The H and 13C
1
NMR spectra were recorded on a Bruker Avance (1H: 500 MHz, 13C:
125 MHz) instrument using deuterated dimethyl sulfoxide (DMSO-
d6) as the solvent with tetramethylsilane (TMS) as an internal
standard; chemical shifts were reported in δ values (ppm) relative to
that of internal TMS. The abbreviations s = singlet, d = doublet, t =
triplet, q = quadruplet, m = multiplet, and bs = broad singlet were used
throughout. Elemental analyses (C, H, N, S) were realized on a
Thermo Scientific Flash EA 1112 elemental analyzer and were within
0.4% of the theoretical values for carbon, hydrogen, and nitrogen; a
higher tolerance ( 0.75%) was admitted for sulfur, considering its
corresponding peak shape. This analytical method certified a purity of
≥95% for each tested compound. All reactions were routinely checked
by TLC on silica gel (Merck 60 F254).
7-Methoxy-4-methyl-4H-1,2,4-benzothiadiazine 1,1-Dioxide
(8). The mixture of 7-methoxy-4H-1,2,4-benzothiadiazine 1,1-dioxide
(7)19 (10.0 g, 47.2 mmol), potassium carbonate (20 g), and methyl
iodide (15 mL) in acetonitrile (300 mL) was heated at 60 °C for 3 h.
The solvent was removed by distillation under reduced pressure, and
the residue was suspended in water (400 mL). The resulting insoluble
material was collected by filtration, washed with water, dried, and
recrystallized in ethyl acetate (8.6 g, 81% yield): mp 204−206 °C
4-Methyl-7-(2-methylphenoxy)-3,4-dihydro-2H-1,2,4-benzothia-
1
diazine 1,1-Dioxide (11b): 25% yield, mp 166−168 °C; H NMR
(DMSO-d6) δ 2.20 (s, 3H, CH3), 2.93 (s, 3H, NCH3), 4.63 (s, 2H, 3-
CH2), 6.79 (d, J = 8.1 Hz, 1H, 6′-H), 6.93 (d, J = 9.2 Hz, 1H, 5-H),
6.96 (d, J = 2.9 Hz, 1H, 8-H), 7.08 (td, J = 7.4, 1.0 Hz, 1H, 4′-H), 7.15
(dd, J = 9.2, 3.0 Hz, 1H, 6-H), 7.20 (td, J = 7.6, 1.2 Hz, 1H, 5′-H), 7.32
(d, J = 7.5 Hz, 1H, 3′-H), 8.11 (s, 1H, NH); 13C NMR (DMSO-d6) δ
15.8 (CH3), 36.5 (NCH3), 62.3 (C-3), 112.3 (C-8), 116.2 (C-5),
118.2 (C-6′), 123.1 (C-8a), 123.8 (C-4′), 124.1 (C-6), 127.4 (C-5′),
128.5 (C-2′), 131.5 (C-3′), 140.3 (C-4a), 147.3 (C-7), 154.6 (C-1′).
Anal. (C15H16N2O3S) Calcd: C, 59.19; H, 5.30; N, 9.20; S, 10.53.
Found: C, 58.90; H, 5.24; N, 8.87; S, 10.21.
(lit.18 205−209 °C); H NMR (DMSO-d6, 500 MHz) δ 3.61 (s, 3H,
1
NCH3), 3.87 (s, 3H, 7-OCH3), 7.31 (d, J = 2.9 Hz, 1H, 8-H), 7.38 (dd,
J = 9.3, 2.9 Hz, 1H, 6-H), 7.49 (d, J = 9.3 Hz, 1H, 5-H), 8.01 (s, 1H, 3-
H); 13C NMR (DMSO-d6) δ 38.0 (NCH3), 56.0 (O−CH3), 106.1 (C-
8), 118.6 (C-5), 120.9 (C6), 123.6 (C-4a), 129.6 (C-8a), 150.3 (C-3),
157.4 (C-7).
7-Methoxy-4-methyl-3,4-dihydro-2H-1,2,4-benzothiadiazine
1,1-Dioxide (9). The solution of 4-methyl-7-methoxy-2H-1,2,4-
benzothiadiazine 1,1-dioxide (8) (6.0 g, 27 mmol) in 2-propanol
(300 mL) was supplemented under stirring with sodium borohydride
4-Methyl-7-(3-methylphenoxy)-3,4-dihydro-2H-1,2,4-benzothia-
1
diazine 1,1-Dioxide (11c): 35% yield, mp 173.5−175.5 °C; H NMR
I
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
(DMSO-d6) δ 2.28 (s, 3H, CH3), 2.94 (s, 3H, NCH3), 4.65 (d, J = 8.1
Hz, 2H, 3-CH2), 6.73 (dd, J = 8.1, 2.0 Hz, 1H, 4′-H), 6.76 (s, 1H, 2′-
H), 6.93 (d, J = 9.1 Hz, 2H, 5-H/6′-H), 7.11 (d, J = 2.9 Hz, 1H, 8-H),
7.20 (dd, J = 9.1, 2.9 Hz, 1H, 6-H), 7.25 (t, J = 7.8 Hz, 1H, 5′-H), 8.13
(t, J = 8.1 Hz, 1H, NH); 13C NMR (DMSO-d6) δ 20.9 (CH3), 36.4
(NCH3), 62.3 (C-3), 114.3 (C-8), 114.6 (C-4′), 116.0 (C-5), 118.0
(C-2′), 123.0 (C-8a), 123.8 (C-6′), 125.5 (C-6), 129.7 (C-5′), 139.7
(C-4a), 140.8 (C-3′), 146.3 (C-7), 157.5 (C-1′). Anal. (C15H16N2O3S)
Calcd: C, 59.19; H, 5.30; N, 9.20; S, 10.53. Found: C, 59.18; H, 5.37;
N, 9.18; S, 9.88.
(t, J = 7.5 Hz, 1H, 4′-H), 7.32 (d, J = 7.3 Hz, 1H, 6′-H), 7.33 (d, J =
9.2 Hz, 1H, 5-H), 7.94 (t, J = 8.0 Hz, 1H, NH); 13C NMR (DMSO-d6)
δ 8.4 (CH(CH2)2), 15.8 (CH3), 29.8 (CH(CH2)2), 61.1 (C-3), 112.1
(C-8), 116.7 (C-5), 118.4 (C-3′), 123.4 (C-8a), 123.7 (C-6), 123.9
(C-5′), 127.5 (C-4′), 128.6 (C-2′), 131.6 (C-6′), 140.2 (C-4a), 148.1
(C-7), 154.4 (C-1′). Anal. (C17H18N2O3S) Calcd: C, 61.80; H, 5.49;
N, 8.48; S, 9.70. Found: C, 62.09; H, 5.62; N, 8.43; S, 9.50.
4-Cyclopropyl-7-(3-methylphenoxy)-3,4-dihydro-2H-1,2,4-benzo-
thiadiazine 1,1-Dioxide (11j): 52% yield, mp 196−197.5 °C; 1H
NMR (DMSO-d6) δ 0.66 (m, 2H, CH(CH2)2), 0.90 (m, 2H,
CH(CH2)2), 2.28 (s, 3H, CH3), 2.49 (m, 1H, CH(CH2)2), 4.66 (d, J =
3.7 Hz, 2H, 3-CH2), 6.74 (dd, J = 8.1, 2.3 Hz, 1H, 4′-H), 6.78 (s, 1H,
2′-H), 6.94 (d, J = 7.5 Hz, 1H, 6′-H), 7.11 (d, J = 2.9 Hz, 1H, 8-H),
7.24 (m, 2H, 6-H /5′-H), 7.34 (d, J = 9.2 Hz, 1H, 5-H), 7.96 (bs, 1H,
NH); 13C NMR (DMSO-d6) δ 8.4 (CH(CH2)2), 20.9 (CH3), 29.8
(CH(CH2)2), 61.0 (C-3), 114.2 (C-8), 114.7 (C-4′), 116.6 (C-5),
118.2 (C-2′), 123.4 (C-8a), 123.9 (C-6′), 125.1 (C-6), 129.7 (C-5′),
139.7 (C-4a), 140.7 (C-3′), 147.1 (C-7), 157.4 (C-1′). Anal.
(C17H18N2O3S) Calcd: C, 61.80; H, 5.49; N, 8.48; S, 9.70. Found:
C, 62.13; H, 5.56; N, 8.53; S, 9.70.
4-Methyl-7-(4-methylphenoxy)-3,4-dihydro-2H-1,2,4-benzothia-
1
diazine 1,1-Dioxide (11d): 25% yield, mp 137−139 °C; H NMR
(DMSO-d6) δ 2.28 (s, 3H, CH3), 2.93 (s, 3H, NCH3), 4.63 (s, 2H, 3-
CH2), 6.86 (d, J = 8.5 Hz, 2H, 2′-H/6′-H), 6.92 (d, J = 9.2 Hz, 1H, 5-
H), 7.06 (d, J = 2.9 Hz, 1H, 8-H), 7.19 (m, 3H, 6-H/3′-H/5′-H), 8.11
(s, 1H, NH); 13C NMR (DMSO-d6) δ 20.2 (CH3), 36.5 (NCH3), 62.3
(C-3), 113.5 (C-8), 116.1 (C-5), 117.9 (C-2′/C-6′), 123.0 (C-8a),
125.0 (C-6), 130.4 (C-3′/C-5′), 132.3 (C-4′), 140.6 (C-4a), 147.0 (C-
7), 155.0 (C-1′). Anal. (C15H16N2O3S) Calcd: C, 59.19; H, 5.30; N,
9.20; S, 10.53. Found: C, 59.02; H, 5.37; N, 9.11; S, 9.92.
4-Methyl-7-(2-methoxyphenoxy)-3,4-dihydro-2H-1,2,4-benzo-
thiadiazine 1,1-Dioxide (11e): 19% yield, mp 150−151.5 °C; 1H
NMR (DMSO-d6) δ 2.96 (s, 3H, NCH3), 3.75 (s, 3H, OCH3), 4.60 (s,
2H, 3-CH2), 6.87 (d, J = 2.9 Hz, 1H, 8-H), 6.89 (d, J = 9.8 Hz, 1H, 5-
H), 6.98 (m, 2H, 5′-H/6′-H), 7.09 (dd, J = 9.2, 2.9 Hz, 1H, 6-H), 7.19
(m, 2H, 3′-H/4′-H), 8.08 (bs, 1H, NH); 13C NMR (DMSO-d6) δ 36.5
(NCH3), 55.6 (OCH3), 62.3 (C-3), 110.8 (C-8), 113.5 (C-3′), 116.0
(C-5), 121.0−121.1 (C-5′/C-6′), 122.8 (C-6), 122.9 (C-8a), 125.5
(C-4′), 139.9 (C-4a), 144.0 (C-1′), 148.3 (C-7), 151.1 (C-2′). Anal.
(C15H16N2O4S) Calcd: C, 56.24; H, 5.03; N, 8.74; S, 10.01. Found: C,
56.12; H, 5.06; N, 8.78; S, 9.95.
4-Cyclopropyl-7-(4-methylphenoxy)-3,4-dihydro-2H-1,2,4-benzo-
thiadiazine 1,1-Dioxide (11k): 35% yield, mp 188−190 °C; 1H NMR
(DMSO-d6) δ 0.65 (m, 2H, CH(CH2)2), 0.89 (m, 2H, CH(CH2)2),
2.28 (s, 3H, CH3), 2.49 (m, 1H, CH(CH2)2), 4.64 (s, 2H, 3-CH2),
6.87 (d, J = 8.5 Hz, 2H, 2′-H/6′-H), 7.05 (d, J = 2.9 Hz, 1H, 8-H),
7.19 (d, J = 8.3 Hz, 2H, 3′-H/5′-H), 7.22 (dd, J = 9.2, 2.9 Hz, 1H, 6-
H), 7.33 (d, J = 9.2 Hz, 1H, 5-H), 7.95 (s, 1H, NH); 13C NMR
(DMSO-d6) δ 8.4 (CH(CH2)2), 20.2 (CH3), 29.8 (CH(CH2)2), 61.1
(C-3), 113.4 (C-8), 116.6 (C-5), 118.0 (C-2′/C-6′), 123.4 (C-8a),
124.6 (C-6), 130.4 (C-3′/C-5′), 132.4 (C-4′), 140.5 (C-4a), 147.8 (C-
7), 154.9 (C-1′). Anal. (C17H18N2O3S) Calcd: C, 61.80; H, 5.49; N,
8.48; S, 9.70. Found: C, 61.86; H, 5.63; N, 8.79; S, 9.29.
4-Cyclopropyl-7-(2-methoxyphenoxy)-3,4-dihydro-2H-1,2,4-ben-
zothiadiazine 1,1-Dioxide (11l): 17% yield, mp 209−212 °C; 1H
NMR (DMSO-d6) δ 0.63 (m, 2H, CH(CH2)2), 0.88 (m, 2H,
CH(CH2)2), 2.46 (m, 1H, CH(CH2)2), 3.75 (s, 3H, OCH3), 4.61 (s,
2H, 3-CH2), 6.86 (d, J = 2.9 Hz, 1H, 8-H), 6.99 (m, 2H, 5′-H/6′-H),
7.13 (dd, J = 9.2, 3.0 Hz, 2H, 6-H), 7.19 (m, 2H, 3′-H/4′-H), 7.29 (d, J
= 9.2 Hz, 1H, 5-H), 7.92 (s, 1H, NH); 13C NMR (DMSO-d6) δ 8.3
(CH(CH2)2), 29.8 (CH(CH2)2), 55.6 (OCH3), 61.1 (C-3), 110.6 (C-
8), 113.5 (C-3′), 116.5 (C-5), 121.1 (C-5′/C-6′), 122.5 (C-6), 123.2
(C-8a), 125.6 (C-4′), 139.8 (C-4a), 143.8 (C-1′), 149.1 (C-7), 151.1
(C-2′). Anal. (C17H18N2O4S) Calcd: C, 58.94; H, 5.24; N, 8.09; S,
9.25. Found: C, 59.03; H, 5.43; N, 8.46; S, 8.67.
4-Methyl-7-(3-methoxyphenoxy)-3,4-dihydro-2H-1,2,4-benzo-
1
thiadiazine 1,1-Dioxide (11f): 43% yield, mp 143.5−145.5 °C; H
NMR (DMSO-d6) δ 2.95 (s, 3H, NCH3), 3.73 (s, 3H, OCH3), 4.65 (s,
2H, 3-CH2), 6.46 (dd, J = 8.1, 2.0 Hz, 1H, 6′-H), 6.53 (t, J = 2.2 Hz,
1H, 2′-H), 6.69 (dd, J = 8.2, 2.1 Hz, 1H, 4′-H), 6.93 (d, J = 9.2 Hz,
1H, 5-H), 7.15 (d, J = 2.8 Hz, 1H, 8-H), 7.22 (dd, J = 9.1, 2.9 Hz, 1H,
6-H), 7.26 (t, J = 8.2 Hz, 1H, 5′-H), 8.11 (bs, 1H, NH); 13C NMR
(DMSO-d6) δ 36.4 (NCH3), 55.2 (OCH3), 62.3 (C-3), 103.7 (C-2′),
108.6 (C-4′), 109.4 (C-6′), 114.4 (C-8), 116.0 (C-5), 123.0 (C-8a),
125.6 (C-6), 130.5 (C-5′), 140.9 (C-4a), 146.0 (C-7), 158.7 (C-1′),
160.7 (C-3′). Anal. (C15H16N2O4S) Calcd: C, 56.24; H, 5.03; N, 8.74;
S, 10.01. Found: C, 56.10; H, 4.99; N, 8.51; S, 9.65.
4-Methyl-7-(4-methoxyphenoxy)-3,4-dihydro-2H-1,2,4-benzo-
thiadiazine 1,1-Dioxide (11g): 27% yield, mp 104−106 °C; 1H NMR
(DMSO-d6) δ 2.92 (s, 3H, NCH3), 3.74 (s, 3H, OCH3), 4.62 (s, 2H,
3-CH2), 6.91 (d, J = 9.2 Hz, 1H, 5-H), 6.95 (s, 4H, 2′-H/3′-H/5′-H/
6′-H), 7.00 (d, J = 2.9 Hz, 1H, 8-H), 7.16 (dd, J = 9.2, 2.9 Hz, 1H, 6-
H), 8.10 (bs, 1H, NH); 13C NMR (DMSO-d6) δ 36.5 (NCH3), 55.4
(OCH3), 62.3 (C-3), 112.6 (C-8), 115.1 (C-3′/C-5′), 116.1 (C-5),
119.8 (C-2′/C-6′), 123.0 (C-8a), 124.3 (C-6), 140.3 (C-4a), 148.1 (C-
7), 150.2 (C-1′), 155.4 (C-4′). Anal. (C15H16N2O4S) Calcd: C, 56.24;
H, 5.03; N, 8.74; S, 10.01. Found: C, 56.29; H, 4.97; N, 8.50; S, 9.44.
4-Cyclopropyl-7-phenoxy-3,4-dihydro-2H-1,2,4-benzothiadiazine
1,1-Dioxide (11h): 57% yield, mp 198−200 °C; 1H NMR (DMSO-d6)
δ 0.66 (m, 2H, CH(CH2)2), 0.90 (m, 2H, CH(CH2)2), 2.51 (m, 1H,
CH(CH2)2), 4.66 (s, 2H, 3-CH2), 6.96 (d, J = 7.8 Hz, 2H, 2′-H/6′-H),
7.13 (m, 2H, 4′-H/8-H), 7.25 (dd, J = 9.2, 2.9 Hz, 1H, 6-H), 7.34 (d, J
= 9.2 Hz, 1H, 5-H), 7.38 (m, 2H, 3′-H/5′-H), 7.95 (bs, 1H, NH); 13C
NMR (DMSO-d6) δ 8.4 (CH(CH2)2), 29.8 (CH(CH2)2), 61.1 (C-3),
114.2 (C-8), 116.7 (C-5), 117.7 (C-2′/C-6′), 123.2 (C-4′), 123.4 (C-
8a), 125.1 (C-6), 130.0 (C-3′/C-5′), 140.8 (C-4a), 147.1 (C-7), 157.4
(C-1′). Anal. (C16H16N2O3S) Calcd: C, 60.74; H, 5.10; N, 8.85; S,
10.13. Found: C, 60.49; H, 5.24; N, 9.07; S, 9.58.
4-Cyclopropyl-7-(3-methoxyphenoxy)-3,4-dihydro-2H-1,2,4-ben-
1
zothiadiazine 1,1-Dioxide (11m): 31% yield, mp 150−152 °C; H
NMR (DMSO-d6) δ 0.66 (m, 2H, CH(CH2)2), 0.90 (m, 2H,
CH(CH2)2), 2.51 (m, 1H, CH(CH2)2), 3.74 (s, 3H, OCH3), 4.66 (s,
2H, 3-CH2), 6.47 (dd, J = 8.2, 1.7 Hz, 1H, 6′-H), 6.54 (t, J = 2 Hz, 1H,
2′-H), 6.70 (dd, J = 8.2, 1.9 Hz, 1H, 4′-H), 7.14 (d, J = 2.7 Hz, 1H, 8-
H), 7.25 (d, J = 9 Hz, 1H, 6-H), 7.27 (t, J = 8.3 Hz, 1H, 5′-H), 7.34 (d,
J = 9.2 Hz, 1H, 5-H), 7.97 (bs, 1H, NH); 13C NMR (DMSO-d6) δ 8.4
(CH(CH2)2), 29.8 (CH(CH2)2), 55.3 (OCH3), 61.1 (C-3), 103.9 (C-
2′), 108.7 (C-4′), 109.5 (C-6′), 114.4 (C-8), 116.6 (C-5), 123.4 (C-
8a), 125.2 (C-6), 130.5 (C-5′), 140.9 (C-4a), 146.8 (C-7), 158.6 (C-
1′), 160.7 (C-3′). Anal. (C17H18N2O4S) Calcd: C, 58.94; H, 5.24; N,
8.09; S, 9.25. Found: C, 58.60; H, 5.24; N, 8.44; S, 9.20.
4-Cyclopropyl-7-(4-methoxyphenoxy)-3,4-dihydro-2H-1,2,4-ben-
1
zothiadiazine 1,1-Dioxide (11n): 47% yield, mp 161−162.5 °C; H
NMR (DMSO-d6) δ 0.64 (m, 2H, CH(CH2)2), 0.89 (m, 2H,
CH(CH2)2), 2.47 (m, 1H, CH(CH2)2), 3.75 (s, 3H, OCH3), 4.63 (s,
2H, 3-CH2), 6.96 (s, 4H, 2′-H/3′-H/5′-H/6′-H), 6.99 (d, J = 2.9 Hz,
1H, 8-H), 7.20 (dd, J = 9.2, 2.9 Hz, 1H, 6-H), 7.32 (d, J = 9.2 Hz, 1H,
5-H), 7.94 (s, 1H, NH); 13C NMR (DMSO-d6) δ 8.3 (CH(CH2)2),
29.8 (CH(CH2)2), 55.4 (OCH3), 61.1 (C-3), 112.4 (C-8), 115.1 (C-
3′/C-5′), 116.6 (C-5), 119.9 (C-2′/C-6′), 123.4 (C-8a), 123.9 (C-6),
140.2 (C-4a), 148.9 (C-7), 150.0 (C-4′), 154.5 (C-1′). Anal.
(C17H18N2O4S) Calcd: C, 58.94; H, 5.24; N, 8.09; S, 9.25. Found:
C, 59.27; H, 5.29; N, 8.20; S, 9.25.
4-Cyclopropyl-7-(2-methylphenoxy)-3,4-dihydro-2H-1,2,4-benzo-
1
thiadiazine 1,1-Dioxide (11i): 44% yield, mp 244−247 °C; H NMR
(DMSO-d6) δ 0.65 (m, 2H, CH(CH2)2), 0.89 (m, 2H, CH(CH2)2),
2.20 (s, 3H, CH3), 2.47 (m, 1H, CH(CH2)2), 4.64 (d, J = 8.0 Hz, 2H,
3-CH2), 6.81 (d, J = 8.0 Hz, 1H, 3′-H), 6.95 (d, J = 2.9 Hz, 1H, 8-H),
7.09 (t, J = 7.4 Hz, 1H, 5′-H), 7.19 (dd, J = 9.2, 2.9 Hz, 1H, 6-H), 7.22
J
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
4-Cyclopropyl-7-(3-cyanophenoxy)-3,4-dihydro-2H-1,2,4-benzo-
thiadiazine 1,1-Dioxide (11o): 13% yield, mp 198−200 °C; 1H NMR
(DMSO-d6) δ 0.68 (m, 2H, CH(CH2)2), 0.91 (m, 2H, CH(CH2)2),
2.52 (m, 1H, CH(CH2)2), 4.68 (s, 2H, 3-CH2), 7.27 (m, 2H, 8-H/6′-
H), 7.31 (dd, J = 9.1, 2.8 Hz, 1H, 6-H), 7.36 (d, J = 9.2 Hz, 1H, 5-H),
7.41 (d, J = 2.4 Hz, 1H, 2′-H), 7.57 (m, 2H, 4′-H/5′-H), 7.98 (bs, 1H,
NH); 13C NMR (DMSO-d6) δ 8.4 (CH(CH2)2), 29.8 (CH(CH2)2),
61.0 (C-3), 112.6 (C-3′), 115.3 (C-8), 116.8 (C-5), 118.2 (CN), 120.5
(C-2′), 122.2 (C-6′), 123.5 (C-8a), 125.6 (C-6), 126.8 (C-4′), 131.5
(C-5′), 141.5 (C-4a), 145.6 (C-7), 158.0 (C-1′). Anal. (C17H15N3O3S)
Calcd: C, 59.81; H, 4.43; N, 12.31; S, 9.39. Found: C, 59.82; H, 4.71;
N, 11.96; S, 8.66.
Stable Transfection of GluA2o(Q) in HEK293 Cells. Human
embryonic kidney 293 (HEK293) cells were transfected using the x-
tremegene 9 3−1 kit (Sigma-Aldrich, St Louis, Missouri, USA). Briefly,
an expression vector (a mixture of 2.5 μg of pcDNA3.1.Myc-
GluA2o(Q) and 7.5 μL of x-tremegene 9 in a buffer of optiMEM)
was added on nearly confluent HEK293 cells in a 50 mm dish. A day
later, the antibiotic G418 (700 μg/mL) was added to make the
selection.
Fluorescence-Based Calcium Assay on GluA2(Q) Cells.
HEK293 cells stably expressing AMPA tetrameric channel form A2
(Q) flop isoform (GluA2o(Q)) were prepared. HEK293 cells or their
stable cell line equivalents were routinely grown in Dulbecco’s
Modified Eagle’s Medium (DMEM) containing 10% fetal bovine
serum (FBS). The measurement of the fluorescence intensity was
carried out using fluorescent microplate reader Fluoroskan Ascent FT
equipped with two dispensers (Thermo Electron Corporation,
Finland). After trypsinization, the cells were washed twice using
Hanks balanced salt solution (HBS, 120 mM NaCl, 2 mM KCl, 2 mM
CaCl2, 2 mM MgCl2, 10 mM HEPES, pH 7.4) and incubated for 1 h
with fluorescent dye Fluo-4/AM (5 μg/mL; Molecular Probes,
Invitrogen, Merelbeke Belgium). The cells were rinsed twice with
HBS, and 100 μL of the resulting cell suspension in HBS buffer was
introduced into each well of a 96-well plate (density of 150 000 cells/
well). The evaluated compounds were applied at various concen-
trations (from 3 × 10−8 M to 10−4 M, with maximal DMSO
concentration = 1%). After shaking the plate 30 s at 1200 rpm, the
emission was read at 538 nm after an excitation at 485 nm during 500
ms per measure. Measurement of a vehicle control was followed by
injection of a 10 mM glutamate solution (10 μL), a fluorescence
measurement, 15 s of shaking at 1200 rpm and a final fluorescence
measurement. The results (EC50) are the concentration required to
reach half of the maximal intensity, calculated by nonlinear regression
analysis (GraphPad Prism software) from at least three independent
concentration−response curves made in triplicate.
experimental SAXS data using the OLIGOMER program from the
ATSAS suite.24 Theoretical CRYSOL-derived24 form factors were
based on monomeric and dimeric glutamate-bound GluA2-LBD-
N775S structures.
Data were analyzed based on the equilibrium models in Figure 3A.11
For the model with one site per dimer interface the degree of
dimerization is given by
⎛
⎜
⎞
⎟
8Rt
M
K3
−1 + 1 +
1 +
(
)
K
⎝
⎠
4
R =
̅
⎛
⎞
⎟
8Rt
K
M
K3
⎜
1 + 1 +
1 +
(
)
⎝
⎠
4
where Rt is the total GluA2-LBD-N775S concentration and M is the
free modulator concentration. K4 and K3 are defined in Figure 3A. The
degree of dimerization for two sites per dimer interface is given by
2M
K3
M2
K32
⎛
⎜
⎝
⎞
⎟
⎠
8Rt
K4
−1 + 1 +
1 +
+
(
)
R =
̅
⎛
⎞
⎟
⎠
8Rt
2M
M2
⎜
1 + 1 +
1 +
+
K32
K3
(
)
K4
⎝
Since only the degree of dimerization is measured, there is no direct
measure of the free modulator. In order to appropriately fit the data,
the fitting routine generates a curve for values of K3 and K4 as a
function of M and then calculates the total modulator at each point,
generating a new curve for dimerization vs total modulator. In this
way, the predicted results can be compared to the data. This is
particularly important in cases in which the concentration of GluA2-
LBD-N775S approaches that of the modulator (see 11m in Figure
3B).
The value of K4 was fit simultaneously to all of the SAXS data, as
this parameter was common to all modulators and all protein
concentrations. As part of that global fit, the K3 was fit simultaneously
to all data (i.e., data from all protein concentrations) for a given
modulator. In that way, the K4 value was constrained by all of the data
across all modulators and protein concentrations, and K3 was
constrained by all of the data for a given modulator. This approach
was used previously,11 but in this case, the minimum for the K4 value
was much better defined because of the additional data used in the fit.
Crystallization of GluA2-LBD-L504Y−N775S with 3. To
GluA2-LBD-L504Y−N775S (4.5 mg/mL in 10 mM HEPES pH 7.0,
20 mM NaCl, 1 mM EDTA) were added glutamate to a final
concentration of 1 mM and 3 as solid compound. The solution was
equilibrated at 6 °C overnight prior to use. The protein−ligand
complex was crystallized by the hanging-drop vapor-diffusion method
at 6 °C. The reservoir volume was 500 μL and consisted of 18%
PEG4000, 0.23 M lithium sulfate, and 0.1 M phosphate citrate, pH 4.5.
Drops consisted of 1 μL of a protein solution mixed with 1 μL of the
reservoir solution. Crystals were briefly submerged into the reservoir
solution containing 20% glycerol as a cryo-protectant and flash cooled
in liquid nitrogen before storage.
Expression and Purification. Rat GluA2-LBD, comprising
segment S1 residues 413−527, a GT-linker, and segment S2 residues
653−807 (UNP P19491; numbering with signal peptide) as well as
GluA2-LBD-L504Y−N775S were expressed and purified following the
previously described procedure.31 For the SAXS experiments, the
construct was identical except that the L504Y mutation was not
present (GluA2-LBD-N775S).
Small-Angle X-ray Scattering. SAXS experiments were con-
ducted as previously described.11 Briefly, GluA2-LBD-N775S was
concentrated to 3 mg/mL (0.1 mM) in SAXS buffer (10 mM
glutamate, 25 mM NaCl, 25 mM Tris-Cl, pH 7) and subsequently
diluted to the appropriate concentration (0.005−0.1 mM) with SAXS
buffer. Loaded samples consisted of 1 μL of appropriately diluted
modulator stock solubilized (in DMSO) and 29 μL of diluted protein.
All SAXS experiments were collected at the Cornell High-Energy
Synchrotron Source (CHESS)’s G1 beamline using a dual Pilatus 100
K−S SAXS/WAXS detector.32,33 Capillary cells were robotically
loaded with the 30 μL precentrifuged samples, maintained at 4 °C,
and thoroughly washed between samples.34 For 0.005, 0.01, and 0.03
mM concentrations of 11m, each sample condition and SAXS buffer
was loaded and measured as three 30 μL replicates. The three
scattering curves were averaged together for analysis. Background
subtraction with matching buffer and data analysis were performed
with RAW software.35 The fraction of dimer was determined from
Crystallization of GluA2-LBD with 11m. To GluA2-LBD (5
mg/mL in 10 mM HEPES pH 7.0, 20 mM NaCl, 1 mM EDTA) were
added glutamate to a final concentration of 5 mM and 11m as solid
compound due to the low solubility of the compound. The solution
was equilibrated at 6 °C for 24 h prior to use. The protein−ligand
complex was crystallized by the hanging-drop vapor-diffusion method
at 6 °C. The reservoir volume was 500 μL and consisted of 15.2%
PEG4000, 0.3 M zinc acetate, and 0.1 M sodium acetate, pH 5.5.
Drops consisted of 1 μL of a protein solution mixed with 1 μL of the
reservoir solution. Crystals were briefly submerged into reservoir
solution containing 20% glycerol as a cryo-protectant and flash cooled
in liquid nitrogen before storage.
Structure Determination. X-ray diffraction data were collected at
beamline I911−3 at MAX-lab, Lund, Sweden36 at a cryogenic
temperature (100 K). The data were processed using XDS37 and
SCALA38 in the CCP4 suite of programs.39 The two structures were
solved by molecular replacement in PHASER40 using only the protein
K
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
atoms of the search model (PDB entry 3TKD (molA) for structure
with 3 and PDB entry 3TDJ (molA) for structure with 11m). Initial
model building was performed using AUTOBUILD41 in PHENIX42
and water molecules were introduced into the structures at this step.
During iterative rounds of model building and editing in COOT43 and
refinement in PHENIX, ligand molecules, other additives, and
additional water molecules were gradually built into the structures.
Coordinate files for 3 and 11m were generated in MAESTRO (v.9.4;
ent Research (Laulumaa, Kastrup), Novo Nordisk Foundation
(Kastrup), GluTarget (Larsen, Frydenvang, Kastrup), Danscatt
(Larsen, Frydenvang, Rovinskaja, Kastrup), CoNeXT (Larsen,
Kastrup), BioStruct-X (Frydenvang, Rovinskaja, Kastrup), and
NIH (NS085239 and GM068935, Oswald). CHESS is
supported by the NSF and NIH/NIGMS via the NSF award
DMR-1332208, and the MacChess resource is supported by the
NIGMS award GM-103485.
Schrodinger, LLC, New York, NY, 2013) and parameter files in
̈
eLBOW keeping the geometry.44 The structures were refined using
individual isotropic B values, TLS, NCS, and riding H atoms. The
structures were validated by build-in tools in COOT, MolProbity45 in
PHENIX, and the wwPDB Validation Service. Domain closure of
GluA2-LBD relative to the apo structure of GluA2-LBD (PDB entry
1FTO, molA) was calculated using the DynDom server.46 Figures
were prepared in PyMOL (v.1.7; The PyMOL Molecular Graphics
System, V.S., LLC).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
Heidi Peterson is thanked for the help with expression and
purification of the ligand-binding domain of GluA2. MAX-Lab,
Lund, Sweden is thanked for providing beamtime.
NMR Spectroscopy. GluA2-LBD-N775S was uniformly labeled
with 15N and prepared for NMR spectroscopy as described
previously.47 Spectra were collected on a Bruker AVANCE III HD
800 MHz NMR spectrometer at the SUNY ESF NMR facility. A total
of 1024 complex points were collected in the proton dimension
(spectral width of 14423 Hz), and 128 complex increments were
collected in the nitrogen dimension (spectral width of 3080 Hz). Data
were analyzed with NMRPipe48 and NMRFAM-SPARKY.49 Figure 6B
was prepared with Chimera.50
ABBREVIATIONS USED
■
AMPAR, AMPA receptor; ATD, amino-terminal domain;
BDNF, brain-derived neurotrophic factor; BTD, benzothiadia-
zine 1,1-dioxide; CTD, cytoplasmic terminal domain; GluA2-
LBD, ligand-binding domain of GluA2; HFMZ, hydro-
flumethiazide; HSQC-TROSY, heteronuclear single quantum
coherence transverse relaxation optimized spectroscopy; KA,
kainic acid; SAXS, small-angle X-ray scattering; TMD, trans-
membrane domain
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jmed-
chem.7b01323.
■
S
REFERENCES
■
(1) Dingledine, R.; Borges, K.; Bowie, D.; Traynelis, S. F. The
glutamate receptor ion channels. Pharmacol. Rev. 1999, 51, 7−61.
(2) Traynelis, S. F.; Wollmuth, L. P.; McBain, C. J.; Menniti, F. S.;
Vance, K. M.; Ogden, K. K.; Hansen, K. B.; Yuan, H.; Myers, S. J.;
Dingledine, R. Glutamate receptor ion channels: structure, regulation,
and function. Pharmacol. Rev. 2010, 62, 405−496.
Experimental SAXS data (PDF)
Molecular formula strings of the target compounds
(CSV)
Accession Codes
The structure coordinates and corresponding structure factor
file of GluA2-LBD-L483Y−N754S with glutamate and 3 have
been deposited in the Protein Data Bank under the accession
code 5O9A and the structure of GluA2-LBD with glutamate
and 11m with accession code 5OEW.
(3) Huganir, R. L.; Nicoll, R. A. AMPARs and synaptic plasticity: the
last 25 years. Neuron 2013, 80, 704−717.
(4) Lynch, G.; Palmer, L. C.; Gall, C. M. The likelihood of cognitive
enhancement. Pharmacol., Biochem. Behav. 2011, 99, 116−129.
(5) Partin, K. M. AMPA receptor potentiators: from drug design to
cognitive enhancement. Curr. Opin. Pharmacol. 2015, 20, 46−53.
(6) Jourdi, J.; Hsu, Y.-T.; Zhou, M.; Qin, Q.; Bi, X.; Baudry, M.
Positive AMPA receptor modulation rapidly stimulates BDNF release
and increases dendritic mRNA Transl. J. Neurosci. 2009, 29, 8688−
8697.
(7) Clarkson, A. N.; Overman, J. J.; Zhong, S.; Mueller, R.; Lynch, G.;
Carmichael, S. T. AMPA receptor-induced local brain-derived
neurotrophic factor signaling mediates motor recovery after stroke. J.
Neurosci. 2011, 31, 3766−3775.
(8) Gan, Q.; Salussolia, C. L.; Wollmuth, L. P. Assembly of AMPA
receptors: mechanisms and regulation. J. Physiol. 2015, 593, 39−48.
(9) Sager, C.; Terhag, J.; Kott, S.; Hollmann, M. C-terminal domains
of transmembrane α-amino-3-hydroxy-5-methyl-4-isoxazole propio-
nate (AMPA) receptor regulatory proteins not only facilitate
trafficking but are major modulators of AMPA receptor Function. J.
Biol. Chem. 2009, 284, 32413−32424.
(10) Arai, A. C.; Kessler, M.; Rodgers, G.; Lynch, G. Effects of the
Potent Ampakine CX614 on hippocampal and recombinant AMPA
receptors: interactions with cyclothiazide and GYKI 52466. Mol.
Pharm. 2000, 58, 802−813.
(11) Ptak, C. P.; Hsieh, C. L.; Weiland, G. A.; Oswald, R. E. Role of
stoichiometry in the dimer-stabilizing effect of ampa receptor allosteric
modulators. ACS Chem. Biol. 2014, 9, 128−133.
(12) Ptak, C. P.; Ahmed, A. H.; Oswald, R. E. Probing the allosteric
modulator binding site of GluR2 with thiazide derivatives. Biochemistry
2009, 36, 8594−8602.
AUTHOR INFORMATION
Corresponding Authors
■
*E-mail: j.hanson@uliege.be (for pharmacological assays).
*E-mail: reo1@cornell.edu (for SAXS and protein NMR).
*E-mail: jsk@sund.ku.dk (for structure determination).
*E-mail: pierre.francotte@uliege.be (for synthetic and medic-
inal chemistry).
ORCID
Christopher P. Ptak: 0000-0003-2752-0367
Lars Olsen: 0000-0002-7607-7130
Jette Sandholm Kastrup: 0000-0003-2654-1510
Pierre Francotte: 0000-0001-9252-7798
Author Contributions
⊥E.G. and T.D. contributed equally.
Funding
J.H. and T.D. are a F.R.S.-FNRS research associate and Ph.D.
́ ́
fellow, respectively. P.G. is a Televie Ph.D. fellow. This work
was supported by the Fonds pour la Recherche Scientifique
(F.R.S.-FNRS) Incentive Grant for Scientific Research
[F.4510.14], Fonds Leo
University of Liege [Fonds Spec
(Larsen, Frydenvang, Kastrup), Danish Council for Independ-
́
n Fred
́ ́
ericq (Drapier, Gilissen),
̀
́
iaux], Lundbeck Foundation
L
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
(13) Pirotte, B.; Francotte, P.; Goffin, E.; de Tullio, P. AMPA
receptor positive allosteric modulators: a patent review. Expert Opin.
Ther. Pat. 2013, 23, 615−628.
(14) Citti, C.; Battisti, U. M.; Cannazza, G.; Jozwiak, K.; Stasiak, N.;
Puja, G.; Ravazzini, F.; Ciccarella, G.; Braghiroli, D.; Parenti, C.; Troisi,
L.; Zoli, M. 7-Chloro-5-(furan-3-yl)-3-methyl-4H-benzo[e][1,2,4]-
thiadiazine 1,1-dioxide as positive allosteric modulator of α-amino-3-
hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor. the
end of the unsaturated-inactive paradigm? ACS Chem. Neurosci. 2016,
7, 149−160.
allosteric modulator: structures of GluA2 and GluA3 in complexes
with PEPA. Biochemistry 2010, 49, 2843−2850.
(27) Dunitz, J. D.; Taylor, R. Organic Fluorine hardly ever accepts
hydrogen bonds. Chem. - Eur. J. 1997, 3, 89−98.
(28) Francotte, P.; Goffin, E.; Fraikin, P.; Lestage, P.; Van Heugen, J.-
C.; Gillotin, F.; Danober, L.; Thomas, J.-Y.; Chiap, P.; Caignard, D.-H.;
Pirotte, B.; De Tullio, P. New fluorinated 1,2,4-benzothiadiazine 1,1-
dioxides: discovery of an orally active cognitive enhancer acting
through potentiation of the 2-amino-3-(3-hydroxy-5-methylisoxazol-4-
yl)propionic acid receptors. J. Med. Chem. 2010, 53, 1700−1711.
(29) Clark, T.; Hennemann, M.; Murray, J. S.; Politzer, P. Halogen
bonding: the σ-hole. J. Mol. Model. 2007, 13, 291−296.
(30) Politzer, P.; Murray, J. S.; Clark, T. Halogen bonding: an
electrostatically-driven highly directional noncovalent interaction. Phys.
Chem. Chem. Phys. 2010, 12, 7748−7757.
(31) Krintel, C.; Frydenvang, K.; Olsen, L.; Kristensen, M. T.; de
Barrios, O.; Naur, P.; Francotte, P.; Pirotte, B.; Gajhede, M.; Kastrup,
J. S. Thermodynamics and structural analysis of positive allosteric
modulation of the ionotropic glutamate receptor GluA2. Biochem. J.
2012, 441, 173−178.
(32) Skou, S.; Gillilan, R. E.; Ando, N. Synchrotron-based small-angle
X-ray scattering of proteins in solution. Nat. Protoc. 2014, 9, 1727−
1739.
(33) Acerbo, A. S.; Cook, M. J.; Gillilan, R. E. Upgrade of
MacCHESS facility for X-ray scattering of biological macromolecules
in solution. J. Synchrotron Radiat. 2015, 22, 180−186.
(34) Nielsen, S. S.; Møller, M.; Gillilan, R. E. High-throughput
biological small-angle X-ray scattering with a robotically loaded
capillary cell. J. Appl. Crystallogr. 2012, 45, 213−223.
(35) Hopkins, J. B.; Gillilan, R. E.; Skou, S. BioXTAS RAW:
improvements to a free open-source program for small-angle X-ray
scattering data reduction and analysis. J. Appl. Crystallogr. 2017, 50,
1545−1553.
(36) Ursby, T.; Unge, J.; Appio, R.; Logan, D. T.; Fredslund, F.;
Svensson, C.; Larsson, K.; Labrador, A.; Thunnissen, M. M. G. M. The
macromolecular crystallography beamline I911−3 at the MAX IV
laboratory. J. Synchrotron Radiat. 2013, 20, 648−653.
(37) Kabsch, W. XDS. Acta Crystallogr., Sect. D: Biol. Crystallogr.
2010, 66, 125−132.
(38) Evans, P. Scaling and assessment of data quality. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 2006, 62, 72−82.
(39) Winn, M. D.; Ballard, C. C.; Cowtan, K. D.; Dodson, E. J.;
Emsley, P.; Evans, P. R.; Keegan, R. M.; Krissinel, E. B.; Leslie, A. G.
W.; McCoy, A.; McNicholas, S. J.; Murshudov, G. N.; Pannu, N. S.;
Potterton, E. A.; Powell, H. R.; Read, R. J.; Vagin, A.; Wilson, K. S.
Overview of the CCP4 suite and current developments. Acta
Crystallogr., Sect. D: Biol. Crystallogr. 2011, 67, 235−242.
(40) McCoy, A. J.; Grosse-Kunstleve, R. W.; Adams, P. D.; Winn, M.
D.; Storoni, L. C.; Read, R. J. Phaser crystallographic software. J. Appl.
Crystallogr. 2007, 40, 658−674.
(15) Battisti, U. M.; Citti, C.; Rastelli, G.; Pinzi, L.; Puja, G.;
Ravazzini, F.; Ciccarella, G.; Braghiroli, D.; Cannazza, G. An
unexpected reversal in the pharmacological stereoselectivity of
benzothiadiazine AMPA positive allosteric modulators. MedChem-
Comm 2016, 7, 2410−2417.
(16) Krintel, C.; Francotte, P.; Pickering, D. S.; Juknaite, L.;
̇
Pøhlsgaard, J.; Olsen, L.; Frydenvang, K.; Goffin, E.; Pirotte, B.;
Kastrup, J. S. Enthalpy-entropy compensation in the binding of
modulators at ionotropic glutamate receptor GluA2. Biophys. J. 2016,
110, 2397−2406.
(17) Larsen, A. P.; Francotte, P.; Frydenvang, K.; Tapken, D.; Goffin,
E.; Fraikin, P.; Caignard, D. H.; Lestage, P.; Danober, L.; Pirotte, B.;
Kastrup, J. S. Synthesis and pharmacology of mono-, di-, and trialkyl-
substituted 7-chloro-3,4-dihydro-2h-1,2,4-benzothiadiazine 1,1-diox-
ides combined with x-ray structure analysis to understand the
unexpected structure-activity relationship at AMPA receptors. ACS
Chem. Neurosci. 2016, 7, 378−390.
(18) Cordi, A.; Desos, P.; Lestage, P. N. Benzothia(dia)zine derivatives
and their use as ampa modulators. U.S. Patent WO03053947 (A1),
2003.
(19) Francotte, P.; de Tullio, P.; Goffin, E.; Dintilhac, G.;
Graindorge, E.; Fraikin, P.; Lestage, P.; Danober, L.; Thomas, J.-Y.;
Caignard, D.-H.; Pirotte, B. Design, synthesis, and pharmacology of
novel 7-substituted 3,4-dihydro-2h-1,2,4-benzothiadiazine 1,1-dioxides
as positive allosteric modulators of AMPA receptors. J. Med. Chem.
2007, 50, 3153−3157.
(20) Nørholm, A. B.; Francotte, P.; Goffin, E.; Botez, I.; Danober, L.;
Lestage, P.; Pirotte, B.; Kastrup, J. S.; Olsen, L.; Oostenbrink, C.
Thermodynamic characterization of new positive allosteric modulators
binding to the glutamate receptor A2 ligand-binding domain:
combining experimental and computational methods unravels differ-
ences in driving forces. J. Chem. Inf. Model. 2014, 54, 3404−3416.
(21) Francotte, P.; Nørholm, A. B.; Deva, T.; Olsen, L.; Frydenvang,
K.; Goffin, E.; Fraikin, P.; de Tullio, P.; Challal, S.; Thomas, J. Y.; Iop,
F.; Louis, C.; Botez-Pop, I.; Lestage, P.; Danober, L.; Kastrup, J. S.;
Pirotte, B. positive allosteric modulators of 2-amino-3-(3-hydroxy-5-
methylisoxazol-4-yl)propionic acid receptors belonging to 4-cyclo-
propyl-3,4-dihydro-2h-1,2,4-pyridothiadiazine dioxides and diversely
chloro-substituted 4-cyclopropyl-3,4-dihydro-2H-1,2,4-benzothiadia-
zine 1,1-dioxides. J. Med. Chem. 2014, 57, 9539−9553.
(41) Terwilliger, T. C.; Grosse-Kunstleve, R. W.; Afonine, P. V.;
Moriarty, N. W.; Zwart, P. H.; Hung, L. W.; Read, R. J.; Adams, P. D.
Iterative model building, structure refinement and density modification
with the PHENIX AutoBuild wizard. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 2008, 64, 61−69.
(22) Weiss, J. N. The Hill equation revisited: uses and misuses.
FASEB J. 1997, 11, 835−841.
(23) Nørholm, A. B.; Francotte, P.; Olsen, L.; Krintel, C.;
Frydenvang, K.; Goffin, E.; Challal, S.; Danober, L.; Botez-Pop, I.;
Lestage, P.; Pirotte, B.; Kastrup, J. S. Synthesis, pharmacological and
structural characterization, and thermodynamic aspects of GluA2-
positive allosteric modulators with a 3,4-dihydro-2H-1,2,4-benzothia-
diazine 1,1-dioxide scaffold. J. Med. Chem. 2013, 56, 8736−8745.
(24) Petoukhov, M. V.; Franke, D.; Shkumatov, A. V.; Tria, G.;
Kikhney, A. G.; Gajda, M.; Gorba, C.; Mertens, H. D.; Konarev, P. V.;
Svergun, D. I. New developments in the ATSAS program package for
small-angle scattering data analysis. J. Appl. Crystallogr. 2012, 45, 342−
350.
́
(42) Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis,
I. W.; Echols, N.; Headd, J. J.; Hung, L. W.; Kapral, G. J.; Grosse-
Kunstleve, R. W.; McCoy, A. J.; Moriarty, N. W.; Oeffner, R.; Read, R.
J.; Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H.
PHENIX: a comprehensive Python-based system for macromolecular
structure solution. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010, 66,
213−221.
(43) Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K. Features and
development of Coot. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010,
66, 486−501.
(44) Moriarty, N. W.; Grosse-Kunstleve, R. W.; Adams, P. D.
electronic Ligand Builder and Optimization Workbench (eLBOW): a
tool for ligand coordinate and restraint generation. Acta Crystallogr.,
Sect. D: Biol. Crystallogr. 2009, 65, 1074−1080.
(25) Sun, Y.; Olson, R.; Horning, M.; Armstrong, N.; Mayer, M.;
Gouaux, E. Mechanism of glutamate receptor desensitization. Nature
2002, 417, 245−253.
(26) Ahmed, A. H.; Ptak, C. P.; Oswald, R. E. Molecular mechanism
of flop selectivity and subsite recognition for an AMPA receptor
M
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX

Journal of Medicinal Chemistry
Article
(45) Chen, V. B.; Arendall, W. B.; Headd, J. J.; Keedy, D. A.;
Immormino, R. M.; Kapral, G. J.; Murray, L. W.; Richardson, J. S.;
Richardson, D. C. MolProbity: all-atom structure validation for
macromolecular crystallography. Acta Crystallogr., Sect. D: Biol.
Crystallogr. 2010, 66, 12−21.
(46) Hayward, S.; Berendsen, H. J. C. Systematic analysis of domain
motions in proteins from conformational change; new results on
citrate synthase and T4 lysozyme. Proteins: Struct., Funct., Genet. 1998,
30, 144−154.
(47) Ptak, C. P.; Ahmed, A. H.; Oswald, R. E. NMR approaches to
functional dynamics of genetically separated iGluR domains. Neuro-
methods 2016, 106, 101−118.
(48) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.;
Bax, A. NMRPipe: a multidimensional spectral processing system
based on UNIX pipes. J. Biomol. NMR 1995, 6, 277−293.
(49) Lee, W.; Tonelli, M.; Markley, J. L. NMRFAM-SPARKY:
enhanced software for biomolecular NMR spectroscopy. Bioinformatics
2015, 31, 1325−1327.
(50) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.;
Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. UCSF Chimeraa
visualization system for exploratory research and analysis. J. Comput.
Chem. 2004, 25, 1605−1612.
N
DOI: 10.1021/acs.jmedchem.7b01323
J. Med. Chem. XXXX, XXX, XXX−XXX