Structure-Activity Relationship, Pharmacological Characterization, and Molecular Modeling of Noncompetitive Inhibitors of the Betaine/γ-Aminobutyric Acid Transporter 1 (BGT1)

Article abstract of DOI:10.1021/acs.jmedchem.7b00924

N-(1-Benzyl-4-piperidinyl)-2,4-dichlorobenzamide 5 (BPDBA) is a noncompetitive inhibitor of the betaine/GABA transporter 1 (BGT1). We here report the synthesis and structure-activity relationship of 71 analogues. We identify 26m as a more soluble 2,4-Cl substituted 3-pyridine analogue with retained BGT1 activity and an improved off-target profile compared to 5. We performed radioligand-based uptake studies at chimeric constructs between BGT1 and GAT3, experiments with site-directed mutated transporters, and computational docking in a BGT1 homology model based on the newly determined X-ray crystal structure of the human serotonin transporter (hSERT). On the basis of these experiments, we propose a binding mode involving residues within TM10 in an allosteric site in BGT1 that corresponds to the allosteric binding pocket revealed by the hSERT crystal structure. Our study provides first insights into a proposed allosteric binding pocket in BGT1, which accommodates the binding site for a series of novel noncompetitive inhibitors.

Full text of DOI:10.1021/acs.jmedchem.7b00924

Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES
Article
Structure-activity relationship, pharmacological characterization
and molecular modeling of non-competitive inhibitors
of the betaine/#-aminobutyric acid transporter 1 (BGT1).
Lars Jørgensen, Anas Al-Khawaja, Stefanie Kickinger, Stine Byskov Vogensen, Jonas Skovgaard-
Petersen, Emil Rosenthal Jensen, Nrupa Borkar, Rebekka Löffler, Karsten K Madsen, Hans Bräuner-
Osborne, Arne Schousboe, Gerhard F. Ecker, Petrine Wellendorph, and Rasmus Prætorius Clausen
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b00924 • Publication Date (Web): 09 Oct 2017
Downloaded from http://pubs.acs.org on October 9, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 46
Journal of Medicinal Chemistry
1
2
3
4
Structure-activity
relationship,
pharmacological
5
6
7
8
characterization and molecular modeling of non-competitive
inhibitors of the betaine/γ-aminobutyric acid transporter 1
(BGT1).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Lars Jørgensena*, Anas Al-Khawajaa*, Stefanie Kickingerb*, Stine B. Vogensena*, Jonas
Skovgaard-Petersena, Emil Rosenthala, Nrupa Borkara, Rebekka Löfflera, Karsten K. Madsena,
Hans Bräuner-Osbornea, Arne Schousboea, Gerhard F. Eckerb, Petrine Wellendorpha, and
Rasmus P. Clausena
a
University of Copenhagen, Department of Drug Design and Pharmacology, Faculty of Health
b
and Medical Sciences, 2100 Copenhagen, Denmark. University of Vienna, Department of
Pharmaceutical Chemistry, Althanstrasse 14, 1090 Vienna, Austria.
* Contributed equally to this paper
Keywords
Betaine/γ-aminobutyric acid transporter 1, BGT1, GABA transporters, non-competitive
inhibition, BPDBA, computational docking, mutagenesis
ABSTRACT
N-(1-Benzyl-4-piperidinyl)-2,4-dichlorobenzamide 5 (BPDBA) is a non-competitive inhibitor of
the betaine/GABA transporter 1 (BGT1). We here report the synthesis and structure-activity
relationship of 71 analogues. We identify 26m as a more soluble 2,4-Cl substituted 3-pyridine
analogue with retained BGT1 activity and an improved off-target profile compared to 5. We
performed radioligand-based uptake studies at chimeric constructs between BGT1 and GAT3,
experiments with site-directed mutated transporters and computational docking in a BGT1
homology model based on the newly determined X-ray crystal structure of the human serotonin
transporter (hSERT). Based on these experiments we propose a binding mode involving residues
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 46
1
2
3
4
5
6
7
8
9
within TM10 in an allosteric site in BGT1 that corresponds to the allosteric binding pocket
revealed by the hSERT crystal structure. Our study provides first insights into a proposed
allosteric binding pocket in BGT1, which accommodates the binding site for a series of novel
non-competitive inhibitors.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1 INTRODUCTION
The principal inhibitory neurotransmitter in the mammalian brain is γ-aminobutyric acid
(GABA).1 GABAergic neurotransmission is crucial for most CNS functions, and a number of
CNS disorders are proposed to arise from malfunctions in this system.2–4 Thus, compounds that
increase the GABAergic inhibitory signalling have been developed to treat various neurological
diseases. One example, related to this work, is the anticonvulsant GABA-uptake inhibitor, [(R)-
N-[4,4-bis(3-methyl-2-thienyl)-3-butenyl]nipecotic acid 1 (Tiagabine), Fig. 1],5 which has
provided proof-of-concept for pharmacological inhibition of GABA uptake in epilepsy. In
addition, pharmacological intervention with GABA uptake has also been proposed to have
clinical potential in other neurological conditions, including stroke and anxiety.6–8
Four subtypes of GABA transporters (GATs) have been cloned from the mammalian brain that
all belong to the solute carrier 6 (SLC6) family of sodium-dependent neurotransmitter
transporters.9 According to the International Union of Basic and Clinical Pharmacology
(IUPHAR), these are named GAT1, GAT2, GAT3 and BGT1 (betaine/GABA transporter 1) and
will be referred to as such, regardless of the species.10 Highly selective inhibitors of GAT1 have
been developed (e.g. 1)11 and successfully used for understanding the physiological function and
therapeutic relevance of GAT1.12 The development of inhibitors with comparable potency and
2
ACS Paragon Plus Environment

Page 3 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
selectivity for the other GAT subtypes has, however, been difficult, which is related to the high
sequence identity, especially between GAT2, GAT3 and BGT1.13 We have previously reported
(RS)-4-[N-[1,1-bis(3-methyl-2-thienyl)but-1-en-4-yl]-N-methylamino]-4,5,6,7-
9
tetrahydrobenzo[d]isoxazol-3-ol 2 (EF1502) as a mixed GAT1/BGT1 inhibitor,14 and extensive
pharmacological investigation of the compound has indicated that selective inhibition of BGT1 is
implicated in seizure management.15 We also reported the further development of 2 into [4,4-bis-
(3-methyl-thien-2-yl)-but-3-enyl]-(2-carboxy-cyclohex-2-enyl)-N-methylamine 3 (RPC425) with
an improved subtype-selectivity among the GATs for BGT1.16 Some preference for BGT1 was
also reported for 1-[3-(9H-carbazol-9-yl)propyl]-4-(2-methoxyphenyl)-4-piperidinol 4 (NNC 05-
2090),17 but this compound was, however, also demonstrated to act at other targets not related to
GABA.18 This has limited its use as a tool compound, and no further optimization of the
compound has been described. Recently, the first BGT1 inhibitor with high selectivity among
GATs and with nanomolar potency was identified. This compound, (1S,2S,5R)-5-
aminobicyclo[3.1.0]hexane-2-carboxylic acid, is a GABA analogue, which displays 129- and
169-fold higher potency at BGT1 (IC50 590 nM) than at GAT3 and GAT1/2, respectively.19 To
allow for the identification of other lead structures, we performed a pharmacological screening of
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a
commercial
compound
library and
identified
N-(1-benzyl-4-piperidinyl)-2,4-
dichlorobenzamide (BPDBA, 5) as a non-competitive inhibitor (IC50 32 µM), with selectivity for
BGT1 among the GABA transporters and representing a new structural class possibly acting at
an allosteric binding site.20 Furthermore, two analogues of 5, compounds 6 and 7, showed
activity at both BGT1 and GAT2, while having no activity at GAT1 and GAT3.20 This represents
an unusual pharmacological profile, since GAT2 and GAT3 typically display a similar
pharmacology due to their high sequence identity,13 at least with regards to the orthosteric
binding pocket. Thus, we found 5 to be an interesting lead structure for further development,
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 46
1
2
3
4
5
6
7
8
9
focusing on increasing the potency and rather poor solubility of the compound. Moreover, we
aimed at exploring the so far elusive binding site for 5 in BGT1, by investigating the structure-
activity relationship (SAR) and relating this to structural modeling and mutational studies. We
here report our efforts through the synthesis of 71 analogues through systematic structural
variations, their biological evaluation, and propose a binding site based on the SAR, uptake
experiments at chimeric transporter constructs and structure-based modeling studies.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2 RESULTS AND DISCUSSION
2.1 Synthesis
Most of the compounds were made from two principal routes (Scheme 1): Route I starting with
acylation of the 4-amino group of 1-Boc-4-aminopiperidine, followed by removal of the Boc
group and alkylation of the secondary amine in the 4-amidopiperidines; and Route II starting
with alkylation of the secondary amine in 4-(N-Boc-amino)piperidine, followed by removal of
the Boc group and acylation of the primary 4-amino group. Acid chlorides were employed for
acylations, and alkylbromides for alkylations. Compound 14 was made by hydrolysis of tert-
butyl ester 15. Esters 33a and 33b were made by acylation of 1-benzyl-4-hydroxypiperidine. 34
was made by reduction of 5 with LiAlH4.
2.2 Structure-activity relationship
The structure of 5 can be divided into a dichlorobenzoic amide part (A) and a benzyl amino part
(B), connected via a piperidine ring (Fig. 1). Systematic as well as simultaneous variations were
introduced in each part. The structures are shown in Fig. 2 and Fig. 3, and the pharmacological
activities are listed in Table 1 and 2. The compounds were tested in a [3H]GABA uptake assay
using recombinant cells transiently expressing either the human (h) or mouse (m) GAT subtypes.
4
ACS Paragon Plus Environment

Page 5 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
The first series of compounds had variations in the benzyl amino part (B). The benzyl group was
exchanged for a range of alkyl groups (8–12) and alkyl groups capped with various functional
groups such as a cyano, carboxylic acid and ester as well as phosphonate ester (13–16), but all
compounds were inactive in concentrations up to 1,000 µM. In a second series, substituents were
introduced at the aromatic ring of the benzyl group. Introducing a nitro (17a and 17b) or cyano
(17c and 17d) group in the ortho/para position was detrimental for activity at the highest tested
concentrations of 100 µM and 50 µM, respectively. Other substituents in the para-position, such
as iodo (17e), carboxylic methyl ester (17f) and trifluoromethoxy (17g), also led to inactive
compounds. Moving the trifluoromethoxy to the meta-position (17h), however, led to some
activity at mBGT1 (IC50 160 µM). Introducing a trifluoromethyl (17i) or a fluoro moiety (17j) in
this position led to compounds with potency similar to that of 5. A chloro substituent in the meta-
position (17k) was not tolerated, whereas the m-methoxy substituted 17l was equipotent with 5.
Methyl (17m) substituent in this position led to inactive compounds. Di-substitution in the 3- and
5-positions with methyl groups (17n) or chloro atoms (17o) gave inactive compounds, whereas
methoxy groups (18) in these positions were tolerated, yielding a compound equipotent with 5 at
mBGT1. A range of aromatic analogues exchanging the phenyl ring of the benzyl group with 2-,
3- and 4-pyridyl (19–21), styryl (22), naphthyl (23), quinolonyl (24) and benzoyl substituents
(25) led to seemingly inactive compounds (at least for 23 and 24 when tested in concentrations
up to 50 µM).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Another series of compounds were based on variation in the dichlorobenzoic amide part (A).
Exchanging the dichlorophenyl ring for an isopropyl group gave the inactive compound 26a,
whereas the unsubstituted benzamide 26b displayed inhibitory activity at hBGT1 with an order
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 46
1
2
3
4
5
6
7
8
9
of magnitude lower potency than 5. Reintroducing the chloro substituent in the 2- and 4- position
(26c and 26d, respectively) gave compounds with reduced potencies at mBGT1 compared to 5
(IC50 151 and 77 µM, respectively). Addition of a third chlorine atom in 5 gave 26e, which
showed an apparent 3-fold increase in potency, giving the most potent compound in the series
(IC50 11 µM). Introduction of other substituents, such as p-F (26f), o-OEt (26g) m-CF3 (26h) and
the 2,3-dimethoxy substituted (26i) led to slightly increased potency (IC50 100 µM, 180 µM, 170
µM, respectively) compared to the unsubstituted 26b, but still lower potency than 5. Exchange of
the phenyl ring for a 3-pyridine ring (26j) was detrimental for activity, and introduction of 2-Cl
substituent at the pyridine ring (26k) was not beneficial for activity. Nevertheless, introduction
of the chloro substituent in the 4-position (26l) of the 3-pyridine ring increased the potency quite
dramatically (IC50 39 µM), however, at the expense of mBGT1-selectivity over mGAT2 and
mGAT3. Introduction of two chlorine atoms gave the 2,4-Cl substituted 3-pyridine analogue
26m with a potency similar to 26l, but with a better subtype-selectivity for mBGT1 compared to
mGAT2 and mGAT3. The 4-pyridyl analogue 26n had much weaker activity, but addition of a
3-Cl (26o) increased the activity slightly (IC50 300 µM). The 2-thienyl analogue (26p) showed
activity similar to 26b.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We observed a highly improved solubility of 26m compared to 5, and this was experimentally
determined to 610 µg/ml and 40 µg/ml, respectively. Compound 26m was therefore used as a
new starting point for variation. Exchanging the phenyl group of the benzyl amino moiety for a
2-thienyl group (27) gave an inactive compound, whereas exchange for a 3-indolyl (28) or a
benzo[d][1,3]dioxol-5-yl group (29) was tolerated, yielding equipotent compounds. The biphenyl
substituted 30 turned out to be inactive. We then introduced a range of substituents in the meta
position of the phenyl ring of the benzyl amino moiety. Methoxy- (31a) and ethoxy- (31b)
6
ACS Paragon Plus Environment

Page 7 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
substitution in this position was tolerated (IC50 115 and 111 µM, respectively), but 2-
hydroxyethoxy- (31c), nitro- (31d), methoxycarbonyl- (31e), carboxyl- (31f), carbamoyl- (31g),
hydroxyl- (31h), and 4’fluorophenoxy- (31i) were all inactive. The activity of m-methoxy
substituted 17l and 31a prompted further investigation. Removal of both chlorine atoms in 31a
gave the inactive compound 32a, and reintroducing the 2-substituted Cl (32b) was not beneficial
for potency. The 4-substituted 32c had low activity (IC50 115 µM), and as with 26l, we also
started to see some activity at mGAT2 and mGAT3 (IC50 540 and 460 µM, respectively). The 4-
pyridyl substituted 32d and the 2-thienyl substituted 32e were inactive. Two compounds probed
the exchange of the amide bond. The ester 33a had a low potency (IC50 106 µM) and displayed
some activity at mGAT2 and mGAT3 (IC50 349 and 415 µM, respectively), while the dichloro
analogue 33b was inactive at all GAT subtypes. Removal of the amide carbonyl oxygen in 5
gave diamino 34 that displayed reduced potency (IC50 112 µM). The profile of compounds 17i,
26e, 31a and 31b prompted us to determine the solubility of these compounds too. Compound
26e had a solubility of 40 µg/mL that is the same as 5. Compound 17i had a solubility of only 3
µg/ml, whereas 31a and 31b solubility was 240 and 130 µg/ml, respectively. Thus, 26m (610
µg/ml) was clearly the most soluble among these compounds.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2.3 Selectivity profiling of 5 and 26m at neuroreceptor and transporter panel
5 and 26m were screened for activity at 43 other neuroreceptors and transporters by the National
Institute of Mental Health’s Psychoactive Drug Screening Program (NIMH-PDSP)
(Supplemental Information, Table S1). 5 was found to bind to at least eight CNS targets in the
nanomolar and low micromolar ranges, including the serotonin type 1D and 2B receptors, the
dopamine D1, D2 and D4 receptors, the α2C adrenergic receptor and the histamine H1 receptor.
Interestingly, the analogue 26m had effects at fewer off-targets. Most notably, 26m was not
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 46
1
2
3
4
5
6
7
8
active at the dopamine D1, D2 and D4 receptors in contrast to 5. However, 26m showed
pronounced affinity in the nano- and micromolar range for sigma and serotonin type 2B
receptors, respectively.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2.4 Binding site characterization
We have previously demonstrated that 5 is a non-competitive inhibitor of GABA uptake at
BGT1,20 which suggests that its binding site is different from the substrate binding site. In the
present study, two strategies were pursued to guide mutations in hBGT1 with the aim of
identifying and validating the putative binding site of 5: an experimental approach based on
chimeric hBGT1/hGAT3 constructs and a computational docking study.
2.4.1 Chimeric study
Chimeras between related transporters that show distinct pharmacological properties have
previously been used to successfully delineate the importance of specific protein domains for
ligand binding.21–23 In this study, we generated four chimeric transporter constructs consisting of
varying parts of hBGT1 and hGAT3 (Fig. 4) to elucidate the binding region of 5. The hBGT1-
specific sequence in the N-terminus was proven necessary for functionality (data not shown), and
thus, all chimeras contained hBGT1 sequence at the N-terminus and hGAT3 sequence at the C-
terminus. The chimeras were created by standard overlap extension PCR24 with parts from
hBGT1 and hGAT3 spliced together at positions that allowed an overlap of six identical amino
acids in regions of sequence conservation. The chimeras were designated A, B, C and D with
junction sites placed in the transmembrane (TM) segments 3, 6a and 8, and intracellular loop
(IL) 5, respectively (Supplemental Fig. S1). Chimera A contained the least amount of hBGT1
sequence, whereas Chimera B, C and D contained progressively more and more hBGT1
8
ACS Paragon Plus Environment

Page 9 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
sequence. The chimeras were transiently expressed in tsA201 cells and tested in the [3H]GABA
uptake assay.
6
7
8
9
Chimera A turned out to be non-functional, which precluded further studies using this construct.
Whether the lack of function was related to dysfunctional transport properties or faulty protein
translation/surface trafficking was not further investigated. On the other hand, chimeras B, C and
D were functional as they were able to transport [3H]GABA. The activity for GABA at chimera
B and D was in the same range as at wildtype (wt) hBGT1, whereas it was significantly lower at
chimera C (Table 3). 5 had no activity at chimera B similar to wt hGAT3, while it displayed
inhibitory activity at chimera C that was not significantly different from wt hBGT1. This
suggested that the amino acid sequence differences between chimera B and C, i.e. TM6b–TM8a
of BGT1 (Fig. 4), are necessary for the BGT1-selectivity of 5. Furthermore, the inhibitory
activity of 5 at chimera D was markedly higher than wt hBGT1, implying that residues within
TM8b–10 could also be involved in the region of important molecular interactions.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Examining further TM6b–TM8, we identified 19 amino acid residues in this region that differed
between hBGT1 and hGAT3, according to the alignment presented in Fig. S1. These residues
were mutated to the corresponding hGAT3 amino acids by site-directed mutagenesis. I386 in
hBGT1 was the only residue that differed not only from hGAT3, but also from hGAT2, and was
therefore mutated to the corresponding hGAT2 residue, generating hBGT1 I386F. All 20 hBGT1
single-point mutants were functional in the [3H]GABA uptake assay, implying surface
expression. However, a diminished total [3H]GABA uptake was seen at hBGT1 Q299L and
K310N compared to wt hBGT1. To examine whether the variable functionalities were related to
differences in surface expression, we introduced a human influenza hemagglutinin (HA) epitope
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 46
1
2
3
4
5
6
7
8
9
tag in hBGT1 and hGAT3 that enabled us to quantify surface expression of the mutated
transporters. The tag was inserted in the loop between TM3 and TM4 similar to an HA-tagged
version of the dopamine transporter.25 Pharmacological characterization of HA-tagged hBGT1
and hGAT3 in the [3H]GABA uptake assay revealed that they were functional and displayed
wildtype-like properties (data not shown). Using the enzyme-linked immunosorbent assay
(ELISA), we could see a correlation between the level of surface expression and maximal
transport capacity (data not shown). Therefore, the diminished [3H]GABA uptake by hBGT1
Q299L and K310N was inferred to be a consequence of a reduced surface expression.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The potency of GABA and 5 at all 20 hBGT1 mutants was either comparable to or higher than
wt hBGT1 (Table 4). Similar results were seen for 26m at hBGT1 Q299L and I386F (Table 4).
Although we did not see an attenuation of 5 or 26m activity at the hBGT1 mutants, we saw an
increase in the potency of 26m at hGAT3 L314Q, corresponding to Q299L in hBGT1, compared
to wt hGAT3 (Table 4). Overall, the mutational study suggested that none of the 19 residues in
hBGT1 TM6b–TM8a, which differed from hGAT3, are individually responsible for the BGT1
GABA transporter subtype-selectivity of 5. However, this does not rule out that this region of
BGT1 contains important determinants for the binding of 5, especially seen in the light that
function, not binding, is the measurable output of the available [3H]GABA uptake assay.
Therefore, perturbing the interaction with a specific residue by a mutation may not translate into
a disrupted uptake inhibitory activity by 5. Additionally, single-point mutations may not be
sufficient to obstruct 5 principal interactions, which could be shaped by several amino acids in
combination. Furthermore, the loss of an interacting residue due to a mutation could also be
masked as a result of compensatory interactions in that area. Analogously, it has been suggested
that compensatory mechanisms, including backbone interactions, govern the resilience of a non-
10
ACS Paragon Plus Environment

Page 11 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
nuclease inhibitor of reverse transcriptase against drug resistance mutations.26 Moreover, the fact
that increased 5 activities are observed for several of the mutants (Table 4) further supports this
notion.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In the lack of an available radioligand-based binding assay for BGT1, we therefore, expanded on
the findings presented by the chimeric studies using in silico techniques to model the binding
mode of 5 with the aim of identifying the molecular determinants for the selectivity of 5 for
BGT1 over the other GATs.
2.4.2 Computational docking study
As discussed above, the increased activity of 5 seen at chimera D compared to wt hBGT1
suggested that TM8b–10 contain residues important for the BGT1-subtype selectivity. Within
this region of hBGT1, however, almost 30 residues differ from hGAT3 (Fig. S1). Therefore, we
wanted to utilize a more rational approach to identify promising residues for new mutational
experiments. As no crystal structure of hBGT1 has been published yet, the studies were based on
a homology model.
The crystal structure of the bacterial leucine transporter (LeuT)27 has so far been the most
frequently used template for modeling the outward-open state of GATs.28–31 Although the LeuT
homolog shares only about 20% sequence identity with hBGT1 (Fig. S2), the highly conserved
overall fold in the protein has made it a decent template for modeling GATs.9,32 The recently
published crystal structures of the Drosophila dopamine transporter (dDAT)33 and the human
serotonin transporter (hSERT),34 however, not only displays substantially higher sequence
identity with BGT1 (47% and 43%, respectively, cf. Figure S2), but also notable structural
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 46
1
2
3
4
5
6
7
8
9
differences compared to LeuT within the regions of EL2 and TM12. It is noteworthy to add that
the structure of hSERT was co-crystallized with a ligand, revealing an allosteric binding site
accommodated by TMs 1, 6, 10 and 11.34 We have previously postulated that the corresponding
site in hGAT3 contains the binding pocket of isatin analogues, shown to act as non-competitive
inhibitors selective for GAT3 among GATs.35 Considering the non-competitive nature of 5,20 the
high homology between hSERT and hBGT1, and the fact that the hSERT crystal structure has
recently been used as a template to model rat GAT1,36 we regarded this crystal structure as an
appropriate template for generating a hBGT1 homology model (Fig. 5A).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Cavity analysis of the TM8b–10 region of the hBGT1 model with MOE Site Finder37 suggested
that only the TM10 domain is involved in a possible binding cavity positioned in the
extracellular vestibule, corresponding to the allosteric pocket in the hSERT crystal structure (Fig.
S3). Therefore, we hypothesized that the allosteric binding site observed in hSERT could also
exist in hBGT1 and accommodate 5.
Hence, analogues of 5 were docked into the postulated allosteric binding site of hBGT1 between
TM regions 1, 6, 10 and 11. Following our experimental data-guided docking approach,38–40 ten
active compounds (5, 17i, 17j, 17l, 18, 26d, 26e, 26i, 26l, 26m) with IC50 values of less than 100
µM at BGT1 were selected and docked with flexible side chains using the software package
GOLD.41 A total of 100 poses per compound were generated and assembled into 88 clusters
according to an in-house protocol for common scaffold clustering with a root-mean-square
deviation (RMSD) of less than 3 Å as a similarity threshold.40,41 Highly populated clusters that
contained poses of all docked compounds were considered the most promising. Five out of the
six most populated clusters (more than 50 poses) contained poses of all docked compounds and
12
ACS Paragon Plus Environment

Page 13 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
were selected for further analysis (Fig. S4 and S5). Of the essence for selecting the most
promising cluster was the SAR observed for compounds 17a–o and 18. Different substituents in
the ortho, meta and para position of 5’s benzyl amino part (B) showed that substituents are only
tolerated in the meta, but not in ortho and para position. Therefore, we concluded that the sub-
pocket accommodating part B of the ligand provides space for meta substituents. Out of the six
most populated clusters only two clusters (4 and 10) contained poses that were accommodated in
such a tight sub-pocket. Cluster 4 was finally selected because its poses could explain the SAR
of 17a–17l and 18.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accordingly, the 2,4-dichlorobenzoic amide ring of 5 (part A) fits into a mixed hydrophilic and
lipophilic sub-pocket that is formed by T289, F293, Y453, S457, G458, I459 and S516. The
dichlorobenzoic amide is stabilized by π-π interactions with Y453, hydrogen bonding of the
carbonyl group with the sidechain of T289, and possible interaction of ortho and para Cl with the
side chains of S457 and S516, respectively (Fig. 5B). The 1-benzyl-4-piperidinyl ring (part B)
fits into the opposite part of the pocket where the benzyl ring fills a tight lipophilic sub-pocket
that is formed by W60, L56, Y132, Y133, I136 and F448. This part of the pocket is connected
with the orthosteric site.
The meta trifluoromethyl, methoxy and dimethoxy substituents (part B) of 17i, 17l and 18 are all
accommodated in the pocket by allowing W60 to move in order to form a small sub-pocket with
P360 (Fig. 5B and Fig. S6). 17l’s methoxy group forms a hydrogen bond with the sidechain of
R61, which replaces the interaction of the carbonyl oxygen with T289, resulting in similar
activity compared to 5. The meta fluorine substituent of 17j is accommodated in a similar
fashion without the need for W60 to adapt a different rotamer. Compounds 26d, 26e, 26i and 26l
show an orientation in the pocket analogous to 5, but the carbonyl group undergoes hydrogen
bonding with R61 instead of T289 (Fig. S6). The sub-pocket that accommodates part A of the
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 46
1
2
3
4
5
6
7
8
ligands offers enough space to accommodate orto and meta dimethoxy (26i) as well as a third
chlorine in the ortho position (26e). Compound 26m shows the same binding mode as 5 and also
the same activity (Fig. S6).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Y453 was selected as an interesting residue for mutational experiments, because interactions
with this residue were not only observed for almost all poses in cluster 4, but also for all six
highly populated clusters. Furthermore, Y453 is part of TM10, which was suggested by the
chimeric studies to be involved in the BGT1 GAT-selectivity of 5. Y453 is conserved among the
GATs, except for GAT3, where it corresponds to S468 (Fig. S1). Interestingly, we already
proposed, in an independent study of non-competitive GAT3 selective inhibitors, GAT3 S468 to
be a key residue involved in binding.35 Since 5 shows no inhibitory activity at GAT3, we
expected the hBGT1 Y453S and hBGT1 Y453A mutants to show a decreased 5 inhibitory
activity. On the other hand, we expected the corresponding hGAT3 S468Y mutant to introduce
activity for 5 in GAT3. The three single-point mutated transporters were functional in the
[3H]GABA uptake assay, but did not alter the inhibitory activities of 5 and 26m (Table 5). Also
the double-mutant hBGT1 Y453S+Y454A, which addressed possible compensatory interactions
by the adjacent Y454, did not lead to a disruption of the BGT1 activity of 5 or 26m (Table 5).
Coleman et. al showed that the allosteric site is highly flexible.34 Therefore, the postulated loss
of interactions within the double-mutant may be compensated by other residues.
Although the mutational studies were unsuccessful in delineating the specific residues involved
in the BGT1 subtype selectivity of 5, the postulated binding hypothesis within the suggested
allosteric site was able to explain the SAR of selected BGT1-active 5 analogues. Thus, the
attachment of a methylenedioxy group to the benzamide ring (part A) as in 6 is tolerated (IC50 of
14
ACS Paragon Plus Environment

Page 15 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
42 µM) since the sub-pocket that contains part A offers enough space to accommodate the
additional residue. Attaching a naphthamide (7) slightly increases activity (IC50 15 µM), because
the additional conjugated double bonds not only fit nicely into the pocket, but can also undergo
π-π-stacking with Y453. Exchanging the dichlorobenzene ring with propyl deprives 26a of its
activity, because the π-π interaction with Y453 is lost. Once the phenyl ring is reintroduced
(26b), weak activity (IC50 290 µM) is regained. Sequential attachment of chlorine in the ortho
and para position gradually increases activity, as shown in the series of the analogues 26c, 26d
and 26e (IC50 151, 77 and 11 µM, respectively), because the chlorine substituents in the ortho
and para position not only increase the van der Waals interaction by nicely filling the pocket, but
can also interact with the side chains of S516 and T512. Exchanging the chlorine substituents
with ortho and meta methoxy (26i), ortho ethoxy (26g) or meta trifluoromethyl (26h) is tolerated
by a slight decrease of activity (IC50 91, 180 and 170 µM, respectively), because the sub-pocket
offers enough space to accommodate these residues, but no additional interactions are formed.
Displacing the phenyl ring of 5 with 3-pyridine gives 26m, which shows the same activity (IC50
37 µM) as 5 because of its identical binding mode. Detaching 2-Cl from 26m gives 26l, which
shows similar activity (IC50 39 µM), but selectivity is lost, which is also observed for 32c and
26d. Surprisingly, the attachment of only one chlorine atom in the ortho position yields the
inactive compound 26k (cf. 26c). We hypothesize that the possible interaction of ortho chlorine
with S457 might trigger selectivity. Although S457 is conserved among the GATs, it is part of
the three amino acid motif ASS457, which corresponds to AAS452, AAS472 and SAS456 in
GAT1, GAT2 and GAT3, respectively. S457 in BGT1 resembles a one-amino acid insertion in
the middle of an unwound region in TM10, which is unique for the GATs and the highly related
taurine and creatine transporter.42 The selectivity drop of 26l, 26d and 32c might be linked to this
three-amino acid motif, as for 26l and 32c activity is introduced for GAT2 and GAT3 that share
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 46
1
2
3
the same motif.
4
5
6
7
8
9
8–16 lack the benzyl ring of part B and therefore, do not fully occupy the lipophilic pocket and
thus, show no activity. 17a–17o contain different substituents in the ortho, para, and meta
position. Residues D452, Y132, Y133 and W60 of the tight lipophilic sub-pocket that
accommodates part B sterically hinder the attachment of functional groups in the ortho and para
position, which could explain the lack of activity of 17a–17b, 17e–17g. Lipophilic meta
substituents are tolerated with decreasing volume (17h–17j, 17l, 31a–b, IC50 160, 51, 34, 26, 115
and 11 µM, respectively), as they can be accommodated in the lipophilic sub-pocket due to the
conformational flexibility of W60. Methoxy and ethoxy substituents are also tolerated in the 26m
analogues series, as shown for 31a and 31b (IC50 115 and 111 µM, respectively). Polar
substituents in the meta position are not tolerated in the lipophilic sub-pocket, which explains the
inactivity of 31c–h. However, one has to note, that the inactivity of 17m–o, which contain small
lipophilic substituents in the meta position, cannot be explained by the binding hypothesis. 19–
21 show no activity because of the introduction of a nitrogen atom into the benzylamine ring,
which might not be tolerated in the highly lipophilic sub-pocket. Actually, the same should
account for indolyl 28, which, however, shows activity (IC50 58 µM). Inactive compounds 20–24
contain sterically demanding substituents instead of the benzylamine part, which may lead to
steric clashes in the lipophilic sub-pocket.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3 CONCLUSION
Within this study, we have designed and tested 71 analogues of 5 for their inhibitory profile at
the four GATs. Because of their non-competitive nature, and in analogy to the recently resolved
structure of hSERT, we propose a new allosteric pocket in BGT1 as a possible binding cavity.
16
ACS Paragon Plus Environment

Page 17 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Computational docking studies of a series of 10 analogues of 5, supported by chimeric
experiments, postulated residues within TM10 as important for the BGT1-selectivity among the
GATs. Unfortunately, efforts to validate the binding hypothesis by single- and double-point
mutations were unsuccessful, possibly due to the inherent flexibility of transporter proteins and
the lack of an established BGT1 binding assay that can directly assess how inhibitor binding is
affected by mutations. Nevertheless, the postulated binding hypothesis allows to explain key
elements of the SAR observed in the data set, and thus represents a valid starting point for further
studies. Ideally, such studies would be strongly aided by the availability of a BGT1 inhibitor-
radioligand, which would permit testing of the binding hypotheses directly. Furthermore, the 2,4-
Cl substituted 3-pyridine derivative, 26m, exhibits improved solubility and specificity when
compared to the original lead structure 5, proposing this analogue as a better tool compound than
BDBPA for studying BGT1. In conclusion, our study provides the first insights into a proposed
allosteric binding pocket in BGT1, which accommodates the binding site for a series of novel
non-competitive inhibitors, and could prove to be instrumental in terms of designing new and
better tool compounds for further studying the physiological function and therapeutic relevance
of BGT1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4 EXPERIMENTAL SECTION
Materials
Dulbecco's Modified Eagle Medium (DMEM) with GlutaMAX-I, Dulbecco's Phosphate
Buffered Saline (DPBS), fetal bovine serum (FBS), penicillin-streptomycin (P/S), trypsin and
Hank's Balanced Salt Solution (HBSS) were purchased from Life Technologies (Paisley, UK).
PolyFect Transfection Reagent was purchased from Qiagen (West Sussex, UK). Poly-D-lysine
(PDL), HEPES (4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid) and GABA were
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 46
1
2
3
4
5
6
7
8
purchased from Sigma-Aldrich (St. Louis, MO, USA). [2,3-3H(N)]GABA (35.0 Ci/mmol) and
MicroScint-20 were purchased from PerkinElmer (Boston, MA, USA). Blasticidin-S was
purchased from InvivoGen (San Diego, CA, USA).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Molecular biology
Generation of hBGT1/hGAT3 chimeric transporters. The four chimeric constructs were
generated using standard molecular biology techniques and a standard overlap extension PCR
strategy 24; first amplification of the first part of the chimera (PCR-1), then amplification of the
second part of the chimera (PCR-2) and finally an overlap PCR connecting the two pieces. The
junction sites for chimera A, B, C and D were positioned in TM3 (129-134), TM6a (289-294),
TM8 (392-397) and IL5a (480-485), respectively (hBGT1 amino acid numbering, Fig. S1). The
sequences of the external and internal primers for chimera A, B, C and D with the BGT1 N-
terminus and the GAT3 C-terminus (Fig. S1) are given below.
External primers:
BGT1 forward primer, 5’-gccatggacgggaaggtggc-3’; GAT3 reverse primer, 5’-
gccatggacgggaaggtggc-3’.
Internal primers:
Chimera A, forward primer, 5’-atttgaatgtctactacatcatcatcctggcatgg-3’; reverse primer: 5’-
gccaggatgatgatgtagtagacattcaaatatgactcg-3’.
Chimera
B,
forward
primer:
5’-
acccagatcttcttctcctatgccatttgcctgg-3’; reverse primer, 5’-tggcataggagaagaagatctgggtgcccgcatcc-
3’. Chimera C, forward primer, 5’-ttcctagggctggacagccagtttgtgtgtgtgg-3’; reverse primer, 5’-
acacacacaaactggctgtccagccctagg-3’.
Chimera
D,
forward
primer,
5’-
18
ACS Paragon Plus Environment

Page 19 of 46
Journal of Medicinal Chemistry
1
2
3
gaccgtttctatgacaacattgaagacatgattggctacc-3’; reverse primer, 5’-gtcttcaatgttgtcatagaaacggtcc-3’.
4
5
6
7
8
9
The PCR strategy is exemplified for chimera A as follows: PCR-1: BGT1 forward primer and
chimera A reverse primer with hBGT1 subcloned into the pcDNA5/FRT vector43 as template.
PCR-2: Chimera A forward primer and GAT3 reverse primer with hGAT3 (pcDNA5/FRT)43 as
template. PCR-overlap: BGT1 forward primer and GAT3 reverse primer with PCR-1 and PCR-2
as template. Subsequently, the final PCR product was inserted into the mammalian
pcDNA5/FRT vector according to the protocol of the manufacturer (pcDNA5/FRT/V5-His
TOPO® TA Expression Kit, Invitrogen, Paisley, UK).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Epitope tagging of hBGT1 and hGAT3. To enable evaluation of surface expression of mutated
BGT1 and GAT3 constructs using ELISA, we inserted an epitope tag accessible from the
extracellular side of the plasma membrane. Inspired by the work of Sorkina and co-authors,25
where a human influenza hemagglutinin (HA) peptide tag (YPYDVPDYA) was inserted in the
extracellular loop between TM3 and TM4 in the human dopamine transporter (hDAT), we
inserted an HA tag in hBGT1 and hGAT3 subcloned into the pcDNA5/FRT vector43 in positions
corresponding to the location in hDAT. Using overlap extension PCR, the HA tags were
introduced after predicted N-glycosylation sites between TM3 and EL2 in hBGT1 and hGAT3
(as predicted using the CBS NetNGlyc 1.0 server).44 In hBGT1, the HA tag was inserted
immediately after S173 with the deletion of the subsequent 6 amino acids (GAGTVT), while it
was placed immediately after N190 in hGAT3 with the deletion of the subsequent 7 amino acids
(YSHVSLQ) (Fig. S1).
The final PCR product was inserted into the FRT/pcDNA5 vector per the manufacturer’s
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 46
1
2
3
instructions (pcDNA5/FRT/V5-His TOPO® TA Expression Kit, Invitrogen).
4
5
6
7
8
9
Site-directed mutagenesis of hBGT1 and hGAT3. The HA-tagged hBGT1 Y453S, Y453A,
Y453S+Y454A and hGAT3 S468Y constructs were obtained from GenScript (Piscataway, NJ,
USA). The other mutations were introduced in-house into either HA-tagged hBGT1 or untagged
hGAT3 using the QuikChange II Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA,
USA).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The sequences of the chimeric, HA-tagged and mutated constructs were confirmed by DNA
sequencing (GATC Biotech AG, Konstanz, Baden-Württemberg, Germany), and all primers
were obtained from TAG Copenhagen (Copenhagen, Denmark).
Pharmacology
Cell culture and transfection. DMEM with GlutaMAX-I supplemented with 10% FBS and 1%
P/S was used as growth medium for the tsA201 cells and for the four human embryonic kidney
293 (HEK293) cell lines each stably expressing one of the four mGAT subtypes.45 mGAT-
HEK293 cells were selected with blasticidin-S (5 µg/mL). tsA201 cells were transfected with
43
DNA constructs encoding the four hGATs subcloned into the pcDNA5/FRT vector
using
PolyFect according to the manufacturer’s instructions (Qiagen, West Sussex, UK) with one
minor modification (40 µL PolyFect instead of 80 µL in a 10 cm culture dish). All cell cultures
were propagated at 37 °C in a humidified atmosphere of 95% O2 and 5% CO2.
[3H]GABA uptake assay. The assay was performed as previously described.46 Cells expressing
recombinant wildtype or mutated GAT constructs were plated into PDL-coated white 96-well
20
ACS Paragon Plus Environment

Page 21 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
polystyrene cell culture microplates (PerkinElmer). The growth medium was removed
approximately 24 hours after, and the cells were washed once with assay buffer (HBSS
supplemented with 20 mM HEPES, 1 mM CaCl2 and 1 mM MgCl2, pH 7.4). Then, assay buffer
supplemented with 30 nM [3H]GABA and various concentrations of test compound was added to
each well. After three minutes of incubation at 37 °C, the cells were washed three times with ice-
cold assay buffer, and MicroScint-20 scintillation liquid was added to each well. The plate was
shaken for at least one hour, and the radioactivity was subsequently measured for three minutes
in a TopCount NXT Microplate Scintillation & Luminescence Counter (PerkinElmer).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ELISA. Approximately 16 hours after transfection, tsA201 cells were plated into PDL-coated
clear 24-well tissue culture plates (PerkinElmer). One day after, the cells were washed with ice-
cold assay buffer (PBS supplemented with 1 mM CaCl2) and subsequently fixed with 4%
paraformaldehyde (12 minutes on ice). The subsequent steps were performed at room
temperature. Cells were washed twice with assay buffer and incubated with blocking solution
(3% dry milk in 50 mM Tris-HCl supplemented with 1 mM CaCl2, pH 7.5) for 20 minutes. Cells
were then incubated with mouse Anti-HA antibody (HA1.1 monoclonal antibody; Nordic
BioSite, Copenhagen, Denmark; or Covance, San Diego, Ca, USA) diluted 1:1,000 in blocking
solution. Next, the cells were washed three times with assay buffer and incubated with blocking
solution for 20 minutes followed by 45 minutes of incubation with goat anti-mouse IgG
horseradish peroxidase-conjugated secondary antibody (Invitrogen) diluted 1:400 in blocking
solution. Finally, cells were washed three times with assay buffer, and transporter expression was
quantified using the 3,3’,5,5’-tetramethylbenzidine (TMB) liquid substrate system (Sigma-
Aldrich) according to the protocol of the manufacturer. Absorbance was measured at 450 nm in
an EnSpire 2300 Multilabel Reader (PerkinElmer). HA-tagged hBGT1 was used as a positive
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 46
1
2
3
4
5
6
7
8
control, while untagged wt hBGT1 or hGAT3 were used as negative controls. All experiments
were performed in triplicate or quadruplicate measurements in at least three independent
experiments.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Data analysis. Data were analyzed and statistically evaluated in GraphPad Prism 5.0a or 7.02
(GraphPad Software, San Diego, CA, USA). Concentration-response curves generated from the
uptake assay were fitted by non-linear regression using the equation for sigmoidal concentration-
response with variable slope: Y=Bottom+(Top-Bottom)/(1+10^((logIC50-X)×HillSlope)), where
Y is the response, X is the logarithm of the concentration, Top and Bottom are the plateaus in
same units as Y, logIC50 is the concentration giving a response halfway between Bottom and
Top, and the HillSlope is the steepness of the curve. Unpaired Student’s t-test and One-way
ANOVA followed by Dunnett’s multiple comparison test were performed when appropriate and
indicated in the figure captions. Data are presented as means
independent pooled experiments performed in triplicates.
S.E.M. of at least three
Chemistry
All starting materials were obtained from commercial sources and used without further
purification. Dry column vacuum chromatography (DCVC) was performed using Fischer Silica
60A (20-45 µm). Reverse phase chromatography was performed using Supelco Discovery®
DSC-C8. NMR spectra were recorded on a 300/75 MHz Varian spectrometer at room
temperature using the solvent indicated below. Chemical shifts (δ) are quoted in ppm relative to
residual solvent peaks. MS data was recorded using electrospray ionization liquid
chromatography-mass spectrometry (ESI-LC/MS) on an Agilent 6410 Triple Quadrupole Mass
Spectrometer instrument coupled to an Agilent 1200 HPLC system using a C18 reverse phase
22
ACS Paragon Plus Environment

Page 23 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
column (Agilent XDB-C18, 4.6 ´ 50 mm, 1.8 mm) with a linear gradient of the binary solvent
system of H2O/MeCN/formic acid (A: 95:5:0.1 and B: 5:95:0.043) with a flow rate of 1 mL/min.
During analysis, evaporative light scattering (ELS) traces were obtained with a Sedere Sedex 85
Light Scattering Detector used for purity evaluation. Elemental analysis was performed at the
University of Vienna and are within 0.4 of the theoretical values if not stated otherwise.
Melting points were determined in an open capillary tube and are uncorrected. Final compounds
are more than 95% according to elemental analysis and NMR. The experimental procedures for
synthesizing all compounds can be found in the supporting information.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
N-(1-Benzylpiperidin-4-yl)-2,6-dichloronicotinamide (26m)
4-Amino-1-benzylpiperidine (0.50 mL, 2.45 mmol) was dissolved in dichloromethane (10 mL)
and triethylamine (0.44 mL, 3.18 mmol) was added. 2,6-Dichlorpyridine-3-carbonyl chloride
(619 mg, 2.94 mmol) was added and the mixture was stirred at room temperature for 1 hour. The
reaction mixture was filtered through a plug of silica, eluting with ethyl acetate. The solvent was
evaporated. The residue was re-crystallized from ethyl acetate and heptane yielding white
crystals of 26m (766 mg, 2.10 mmol, 86 %). 1H NMR (CDCl3): δ 8.11 (d, J = 8.0 Hz, 1 H), 7.40
(d, J = 8.0 Hz, 1 H), 7.37-7.27 (m, 5 H), 6.43 (d, J = 7.7 Hz, 1 H), 4.14-4.00 (m, 1 H), 3.57 (s, 1
H), 2.90-2.84 (m, 2 H), 2.31-2.21 (m, 2 H), 2.12-2.05 (m, 2 H), 1.74-1.58 (m, 2 H). 13C NMR
(CDCl3): δ 162.7, 151.3, 146.0, 141.9, 138.1, 129.8, 128.9, 128.1, 126.9, 123.3, 63.0, 51.9, 47.6,
31.9. Mp 134-135 °C. Anal. Calcd for C18H19Cl2N3O: C, H, N. LC/MS: Calcd for C18H19Cl2N3O:
396.1; found: 397.2 [M+H]+, 395,3 [M-H]-.
Computational Methods
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 46
1
2
3
4
5
6
7
8
Homology modeling. The PDB crystal structure (PDB code 5I73) of hSERT with a co-
crystallized escitalopram in the allosteric site was selected as template for homology modeling.34
The sequence alignment was conducted with the online tool PROMALS3D
9
(http://prodata.swmed.edu/promals3d/promals3d.php).47 Models were generated with Modeller
9.16,48 ranked according the the DOPE score49, and checked with Ramachandran plots and the
online service ProQM (http://www.bioinfo.ifm.liu.se/ProQM/index.php?about=proqm, Fig. S7
and S8).50 Special attention was paid to modeling the insertion in TM10 where no structural
template is available. The top ranked models were selected to undergo a loop refinement with
Modeller 9.16 allowing the two residues before and after the insertion to be independently
modeled from the template. The models with the refined TM10 were ranked again according to
DOPE score, and the top 10 were visually inspected. The model with the most frequently
occurring TM10 fold was selected as a final model. The authors will release the atomic
coordinates upon article publication.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Docking. Flexible sidechain docking (W60, R61, Q290, F293, D452, Y453) was performed with
GOLD 5.2.2.41 The binding site was determined by superposing the hBGT1 model onto the
hSERT crystal structure with MOE 2016.0837 and selection of the corresponding residues in the
vicinity of the allosteric escitalopram ligand (4.5 Å). Protein and ligand preparations was
performed with PrepWiz and LigPrep, respectively, in the Schrödinger Suit 2015-2.51,52 100
poses per ligand were generated and hierarchically clustered according to an in-house protocol
for common scaffold clustering with a maximal distance of 3 Å within one cluster.40 Poses were
analyzed and visualized with MOE 2016.0837. The top most scored poses of the most populated
clusters were considered as most promising.
24
ACS Paragon Plus Environment

Page 25 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 46
1
2
3
4
5
6
7
8
Corresponding authors:
Chemistry: Associate Professor Rasmus P. Clausen
Tel.: +45 3533 6566
9
E-mail address: rac@sund.ku.dk
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pharmacology: Associate Professor Petrine Wellendorph
Tel.: +45 3533 6397
E-mail address: pw@sund.ku.dk
Modeling: Professor Gerhard F. Ecker
Tel. +431 4277 55110
E-mail address: gerhard.f.ecker@univie.ac.at
FUNDING SOURCES
This work was supported by the Lundbeck Foundation (grant R133-A12270), the Austrian
Science Fund (grants # F03502 and W1232), the Drug Research Academy of the University of
Copenhagen, and the A.P. Møller Foundation for the Advancement of Medical Sciences.
ACKNOWLEDGMENTS
Mark Svan and Kristian K. Thomsen are acknowledged for preparing some of the compounds
and Christa Ancher Marvig for determining solubility. Furthermore, we acknowledge Margot
Ernst for providing her guidance and expertise for generating and selecting the homology model.
The authors would like to thank Bolette Kragholm for generating BGT1/GAT3 chimeric
constructs and Maja Michelle Hansen for generating BGT1 mutants.
ABBREVIATIONS
26
ACS Paragon Plus Environment

Page 27 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
Betaine/γ-aminobutyric acid transporter 1, BGT1;
N-(1-benzyl-4-piperidinyl)-2,4-
7
8
dichlorobenzamide, BPDBA; human serotonin transporter, hSERT; transmembrane, TM;
9
domains γ-aminobutyric acid, GABA; GABA transporters, GATs.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ASSOCIATED CONTENT
Supporting Information. Experimental data for compounds (8 -34) and additional tables and
figures, molecular formula strings and three dimensional coordinates of the homology model of
BGT-1. This material is available free of charge via the Internet at http://pubs.acs.org
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 46
1
2
3
4
5
(1) Mody, I.; Pearce, R. A. Diversity of Inhibitory Neurotransmission through GABAA Receptors.
Trends Neurosci. 2004, 27 (9), 569–575.
6
7
8
9
(2) Ting Wong, C. G.; Guin Ting Wong, C.; Bottiglieri, T.; Carter Snead, O. GABA, γ-Hydroxybutyric
Acid, and Neurological Disease. Ann. Neurol. 2003, 54 (S6), S3–S12.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(3) Fatemi, S. H.; Hossein Fatemi, S.; Stary, J. M.; Earle, J. A.; Araghi-Niknam, M.; Eagan, E.
GABAergic Dysfunction in Schizophrenia and Mood Disorders as Reflected by Decreased Levels of
Glutamic Acid Decarboxylase 65 and 67 kDa and Reelin Proteins in Cerebellum. Schizophr. Res.
2005, 72 (2-3), 109–122.
(4) Inan, M.; Petros, T. J.; Anderson, S. A. Losing Your Inhibition: Linking Cortical GABAergic
Interneurons to Schizophrenia. Neurobiol. Dis. 2013, 53, 36–48.
(5) Rowley, N. M.; Madsen, K. K.; Schousboe, A.; Steve White, H. Glutamate and GABA Synthesis,
Release, Transport and Metabolism as Targets for Seizure Control. Neurochem. Int. 2012, 61 (4),
546–558.
(6) Clarkson, A. N.; Huang, B. S.; Macisaac, S. E.; Mody, I.; Carmichael, S. T. Reducing Excessive
GABA-Mediated Tonic Inhibition Promotes Functional Recovery after Stroke. Nature 2010, 468
(7321), 305–309.
(7) Schwartz, T. L.; Nihalani, N. Tiagabine in Anxiety Disorders. Expert Opin. Pharmacother. 2006, 7
(14), 1977–1987.
(8) Lie, M.; Al-Khawaja, A.; Damgaard, M.; Haugaard, A. S.; Schousboe, A.; Clarkson, A. N.;
Wellendorph, P. Glial GABA Transporters as Modulators of Inhibitory Signalling in Epilepsy and
Stroke. In Glial Amino Acid Transporters; Ortega, A., Schousboe, A., Eds.; Springer; p in press.
(9) Kristensen, A. S.; Andersen, J.; Jørgensen, T. N.; Sørensen, L.; Eriksen, J.; Loland, C. J.;
Strømgaard, K.; Gether, U. SLC6 Neurotransmitter Transporters: Structure, Function, and
Regulation. Pharmacol. Rev. 2011, 63 (3), 585–640.
(10) Alexander, S. P.; Kelly, E.; Marrion, N.; Peters, J. A.; Benson, H. E.; Faccenda, E.; Pawson, A. J.;
Sharman, J. L.; Southan, C.; Davies, J. A.; CGTP Collaborators. The Concise Guide to
28
ACS Paragon Plus Environment

Page 29 of 46
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
PHARMACOLOGY 2015/16: Transporters. Br. J. Pharmacol. 2015, 172 (24), 6110–6202.
(11) Nielsen, E. B.; Suzdak, P. D.; Andersen, K. E.; Knutsen, L. J.; Sonnewald, U.; Braestrup, C.
Characterization of Tiagabine (NO-328), a New Potent and Selective GABA Uptake Inhibitor. Eur.
J. Pharmacol. 1991, 196 (3), 257–266.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(12) Sałat, K.; Kulig, K. GABA Transporters as Targets for New Drugs. Future Med. Chem. 2011, 3 (2),
211–222.
(13) Damgaard, M.; Haugaard, A. S.; Kickinger, S.; Al-Khawaja, A.; Lie, M.; Ecker, G. F.; Clausen, R.
P.; Frølund, B. Development of Non-GAT1-Selective Inhibitors: Challenges and Achievements. In
Glial Amino Acid Transporters; Ortega, A., Schousboe, A., Eds.; Springer: Cham, 2017, pp 315-332.
(14) Clausen, R. P.; Moltzen, E. K.; Perregaard, J.; Lenz, S. M.; Sanchez, C.; Falch, E.; Frølund, B.;
Bolvig, T.; Sarup, A.; Larsson, O. M.; Schousboe, A.; Krogsgaard-Larsen, P. Selective Inhibitors of
GABA Uptake: Synthesis and Molecular Pharmacology of 4-N-Methylamino-4,5,6,7-
Tetrahydrobenzo[d]isoxazol-3-Ol Analogues. Bioorg. Med. Chem. 2005, 13 (3), 895–908.
(15) White, H. S.; Watson, W. P.; Hansen, S. L.; Slough, S.; Perregaard, J.; Sarup, A.; Bolvig, T.;
Petersen, G.; Larsson, O. M.; Clausen, R. P.; Frølund, B.; Falch, E.; Krogsgaard-Larsen, P.;
Schousboe, A. First Demonstration of a Functional Role for Central Nervous System
Betaine/{gamma}-Aminobutyric Acid Transporter (mGAT2) Based on Synergistic Anticonvulsant
Action among Inhibitors of mGAT1 and mGAT2. J. Pharmacol. Exp. Ther. 2005, 312 (2), 866–874.
(16) Vogensen, S. B.; Jørgensen, L.; Madsen, K. K.; Borkar, N.; Wellendorph, P.; Skovgaard-Petersen, J.;
Schousboe, A.; White, H. S.; Krogsgaard-Larsen, P.; Clausen, R. P. Selective mGAT2 (BGT-1)
GABA Uptake Inhibitors: Design, Synthesis, and Pharmacological Characterization. J. Med. Chem.
2013, 56 (5), 2160–2164.
(17) Thomsen, C.; Sørensen, P. O.; Egebjerg, J. 1-(3-(9H-Carbazol-9-Yl)-1-Propyl)-4-(2-
Methoxyphenyl)-4-Piperidinol, a Novel Subtype Selective Inhibitor of the Mouse Type II GABA-
Transporter. Br. J. Pharmacol. 1997, 120 (6), 983–985.
(18) Dalby, N. O.; Thomsen, C.; Fink-Jensen, A.; Lundbeck, J.; Søkilde, B.; Man, C. M.; Sørensen, P. O.;
29
ACS Paragon Plus Environment