An allosteric ligand-binding site in the extracellular cap of K2P channels /631/45/269/1151 /631/45/612/1237 /631/92/555 article

Article abstract of DOI:10.1038/s41467-017-00499-3

Two-pore domain potassium (K2P) channels generate leak currents that are responsible for the maintenance of the resting membrane potential, and they are thus potential drug targets for treating diseases. Here, we identify N-(4-cholorphenyl)-N-(2-(3,4-dihydrosioquinolin-2(1H)-yl)-2-oxoethyl)methanesulfonamide (TKDC) as an inhibitor of the TREK subfamily, including TREK-1, TREK-2 and TRAAK channels. Using TKDC as a chemical probe, a study combining computations, mutagenesis and electrophysiology reveals a K2P allosteric ligand-binding site located in the extracellular cap of the channels. Molecular dynamics simulations suggest that ligand-induced allosteric conformational transitions lead to blockage of the ion conductive pathway. Using virtual screening approach, we identify other inhibitors targeting the extracellular allosteric ligand-binding site of these channels. Overall, our results suggest that the allosteric site at the extracellular cap of the K2P channels might be a promising drug target for these membrane proteins.

Full text of DOI:10.1038/s41467-017-00499-3

A RT ICLE
DOI: 10.1038/s41467-017-00499-3
OPEN
A n a ll os te r ic lig a n d -b in di ng s it e in t he ext r ace llul a r
ca p o f K2 P ch a n n e ls
1, 2
1, 2
1, 2
a1,2
i 1,2
g1,2
i1,2
g 3
Q
ic
h
a
o
Lu
o
,
Li
p
i
n
g
C
h
e
n
,
X
i
C
he
ng
,
Y
uq
i
n
M
,
Xi
a
o
na
L
,
B
i
n
g
Zh
an
,
L
i
L
,
Sh
i
le
i
Zh
an
,
Fe
i
Gu
o 1, 2
,
Ya
n
g
Li
1, 2
& H uai yu Y an
g1, 2, 4
Two -p o re d o ma i n p o ta s si um (K 2P ) c han ne l s g e ne r at e le ak cu r re n ts t h at ar e r es po n sib le f or
th e ma in te na n ce o f th e re s t in g m e m bra ne po te n ti al, an d t h ey ar e t h us p ot e nt ia l d ru g t ar g et s
f or tr e ati n g d is e a s e s. H e re , we id e nt ify N -(4 -c ho lor ph en yl) -N -( 2- (3 , 4- di hy d ro s io q ui no l in -2
(
1H) -y l ) -2 - o xo e th y l) m e th a ne s ul fo na m id e (T KD C) as an in h ibi to r o f t h e T RE K su b fa mily ,
in clu d in g TR E K- 1, TR E K-2 an d T RA A K c han ne l s. Usi ng T KD C as a ch e mica l p ro b e, a
st u dy co m b in i n g co m pu ta ti ons , m ut ag e n e si s an d e le ct r o ph ys iol og y r ev eals a K2P allo s te r ic
lig a nd- b i n d in g s it e l o ca te d in t he e xt ra ce ll ula r c ap of t h e ch an n els . Mo le cu lar d yn amic s
si mu la ti o ns s u g g e s t th at li g and -in du c ed al los t e ric c onf o rm ati on al t ra ns it io ns le ad t o b lo ckag e
o f th e i on co n d u ctiv e pa th way . Us in g vi rtu al s cr e e nin g ap p ro ach , w e id en t if y o t he r in h ibi to r s
tar g e t in g th e e x tra ce llu lar al los t e ric li g and -b ind in g si te o f t h ese ch an n els . O ve ral l, o ur
re s u lts s u g g e s t th a t th e al los t e ric s it e at t he e xt ra ce ll ular cap o f t h e K2P ch an n els mi g ht b e a
p rom i si n g d ru g ta rg e t fo r t he s e m e m bra ne pr ot e in s.
1
State Key Laboratory of Drug Research and Key Laboratory of Receptor Research, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, 555
Zuchongzhi Road, Shanghai 201203, China. 2 University of Chinese Academy of Sciences, No. 19A Yuquan Road, Beijing 100049, China. 3 Jiangsu Key
Laboratory of Translational Research and Therapy for Neuro-Psycho-Diseases & Department of Medicinal Chemistry, College of Pharmaceutical Sciences,
4
Soochow University, 199 Ren’ai Road, Suzhou 215123, China. Shanghai Key Laboratory of Regulatory Biology, Institute of Biomedical Sciences and School of
Life Sciences, East China Normal University, 500 Dongchuan Road, Shanghai 200241, China. Qichao Luo, Liping Chen, Xi Cheng and Yuqin Ma contributed
equally to this work. Correspondence and requests for materials should be addressed to Y.L. (email: liyang@simm.ac.cn) or to
H.Y. (email: hyyang@bio.ecnu.edu.cn)
N A TU RE C OMM U N IC AT I ON S |8: 3 7 8
|D OI : 1 0 . 1 0 3 8/ s 4 1 46 7 - 0 1 7 - 0 0 4 99- 3 |w ww . n atu r e . c om /n a tu r e co m mu n i ca ti o n s
1

A R T I C L E
N AT URE CO M MU N ICATI ON S | D O I: 10 . 10 38/s4 14 67 -0 17 - 00499- 3
ifteen two-pore domain potassium (K2P) channels have
a Compound 1 (TKDC)
c
TREK-1
been identified in the human genome1, 2. They contribute to
O
O
O
S
N
F
the background leak currents responsible for the main-
+80 mV
tenance of the resting membrane potential. Linked to several
pathologies, K2P channels represent important clinical targets
in the treatment of cardiovascular disease and neurological
N
CI
–
80 mV –60 mV
3
disorders, including pain and depression . For example, the
b
2
00 pA
TREK-1 channel contributes to the perception of pain, regulation
400 ms
of mood, anesthetic responses, cardiac mechanoelectric feedback
and vasodilation4–9 and is involved in the glutamate conductance
10 µM TKDC
DMSO
1
0–12
and the regulation of blood–brain-barrier permeability
.
d 1
.0
.8
.6
.4
.2
.0
e 100
80
Therefore, modulators targeting K2P channels would be
therapeutically useful for the design of drugs treating relevant
diseases. To progress toward a successful rational drug design
targeting K2P channels, a basic understanding of how ligands
interact with these proteins is necessary.
The currently available crystal structures of K2P channels
have revealed information about how these channels respond
to ligands. In these structures, K2P channels are homogenous
dimers. Each monomer includes two extracellular helices (E1 and
TREK-1
TREK-2
TRAAK
0
0
0
0
0
**
6
0
0
4
20
0
–
7
–6 –5 –4 –3 –2
Log (M)
E2), two-pore domains (P1 and P2), and four transmembrane
Fig. 1 I nhibi t ion of T RE K subf ami ly c hanne l s by T KD C in CH O ce ll s.
a Che mic al st r uct ur e of T KD C. b, c T ypi cal whol e -c e ll c urr e nt t ra ce s
re c or de d f rom CH O c e lls ove r e xpr e ssing t he T RE K-1 c hanne l wi t h 1 0 μM
T KD C b or D M SO appl ic at io n c. Cur re nt s we r e e li ci t e d by de p ol ar iz i n g
vol t ag e st e ps f rom a hol ding pot e nt ia l of −80 mV t o + 80 mV i n 2 0 mV
inc r e me nt s, f oll owe d by st e ppi ng down t o −60 mV. d D ose -de p en den t
inh ibit i on of T KD C on T RE K-1 , T RE K-2 and T RAAK c hanne l s. e T he
st at i st ic s of t he hal f- inhi bit or y c onc e nt rat ions of T KD C f or T RE K- 1 ( n = 7 ) ,
T RE K-2 ( n = 7 ), and T RAAK (n = 8) c hanne l s. I C5 0 va lue s we r e o bt a in ed vi a
dose–re s ponse fit t ing . T he Kr usk al–W all is t e st wa s use d f or st a t ist i ca l
ana lysi s; ** ind ic at e s P < 0 .0 1 . T he dat a ar e show n as t he me a n ± s.e .m
helices (M1-M4)1
3–18
. In the transmembrane domain formed by
the M2-M4 helices, there are prominent fenestrations connecting
the inner pore with the milieu of the membrane. These fenes-
trations could be occupied by lipid acyl chains or small molecular
ligands that project into the intracellular ion conducting pore,
thus contributing to a non-conductive channel 5, 16, 18.
1
A rather unique structural feature of K2P channels is the
extracellular cap formed by the E1 and E2 helices, which is not
observed in other ion channels. In some K2P channels, an apical
1
9–21
disulfide bridge stabilizes the E1 and E2 helices
. This
extracellular domain defines two tunnel-like side portals as the
extracellular ion pathway and partially obstructs the direct
movement of ions into the extracellular milieu22–25. Compared We then investigated whether other TREK subfamily members
with classical potassium channels, K2P channels offer bilateral were also inhibited by TKDC. During TKDC treatments,
extracellular access to the selectivity filter. This distinguishing significant inhibition (IC : 5.2 ± 1.9 μM) was observed in
50
extracellular ion pathway explains the insensitivity of K2P CHO cells expressing TREK-2 (Fig. 1d, e). However, TKDC
channels to the classical potassium channel pore blockers, such as showed a weak inhibitory effect on TRAAK with an IC50 value
2
6, 27
tetraethylammonium, 4-aminopyridine, and cesium ion
.
of 65.9 ± 7.0 μM, which is ~11-fold higher than that for the
In this study, we find that through interactions with the extra- inhibition of the TREK-1 channel (Fig. 1d, e). These results
cellular cap, N-(4-cholorphenyl)-N-(2-(3,4-dihydrosioquinolin-2 suggest that TKDC is a common inhibitor for all three TREK
(1H)-yl)-2-oxoethyl)methanesulfonamide (TKDC, Fig. 1a) is able to channels with different potencies.
inhibit all three members of the TREK subfamily (TREK-1, TREK-2 In the electrophysiological experiments investigating TREK-1,
and TRAAK). Using computational modeling, mutagenesis, and the incubation of cells with 10 μM TKDC significantly decreased
electrophysiology with chemical probes, we characterize the binding the current, which suggested the possibility of the internalization
mode of TKDC to TREK-1 and provide a molecular explanation of TREK-1. Therefore, we used confocal microscopy to investigate
for the TKDC-induced allosteric conformational transitions. the co-localization of TREK-1 with Rab5a (an early endosome
We discover more inhibitors by applying virtual screening to this marker protein) under the same conditions. Compared with the
binding site, which further supports the idea that the extracellular DMSO control, TKDC did not significantly increase TREK-1
cap of K2P channels is a functionally important drug target. internalization (Supplementary Fig. 1a, b). Then, we used
Our results suggest that the allosteric conformational transitions flow cytometry to examine TREK-1 cell membrane expression.
induced by the interaction of inhibitors with the extracellular cap TREK-1 positive cells did not obviously lose the cell surface signal
of K2P channels may provide a molecular basis for the development after TKDC incubation (Supplementary Fig. 1c). Both results
of drugs targeting K2P channels.
exclude the possibility that the TKDC-induced endocytosis of
TREK-1 yielded the significant decrease in channel current
observed in the electrophysiological experiments.
Results
To further validate the direct binding of TKDC to the TREK-1
Inhibition of TREK channels by TKDC. In the experimental channel, we tested the TKDC-induced inhibition of TREK-1 in
screening, we identified TKDC as an inhibitor of TREK-1 from inside-out and outside-out patches. The application of TKDC to
an in-house library of approximately 1000 small molecules the extracellular face of the membrane (outside-out recording)
developed over time. In the whole-cell voltage clamp experiments consistently inhibited TREK-1 (Supplementary Fig. 2a). In
on CHO cells that were transiently transfected with TREK-1, contrast, the application of TKDC to the intracellular side of
pronounced decreases in current were generated by perfusion the channel did not significantly inhibit TREK-1 activity
with 10 μM TKDC (Fig. 1b, c). The inhibitory effect of TKDC on (Supplementary Fig. 2b). These results provide evidence that
the TREK-1 channel was dose-dependent with a half-maximal TKDC directly inhibits TREK-1 and the inhibition requires access
inhibition concentration (IC ) of 4.9 ± 0.6 μM (Fig. 1d, e). to the extracellular cap domain.
5
0
2
N A TU RE CO MMU N I CA TI ON S |8: 3 7 8
|D OI : 1 0 . 1 0 3 8/s 4 1 4 67 - 0 1 7 - 0 0 4 99- 3 |w ww . na tu r e . c o m /n atu r e c o m m u n ic a tio n s

NA TUR E COMM UNICAT IO NS | D OI: 1 0.1 0 38/ s4 1 467 - 01 7 - 004 99-3
A R T I C L E
Binding mode of TKDC to TREK-1. To investigate the possible
binding modes of TKDC to the TREK channels, we adopted a
two-stage computational approach. In the first stage, molecular
docking methods (Schrodinger Glide and Autodock ) were
applied to the TREK family crystal structures to identify possible
binding site(s) of TKDC across the entire family. We located
a
b
2
8
29
E1
E2
Q83
L102
3
6 potential TKDC-binding sites (termed sites 1–36) with
low docking scores (Glide G-scores < −3.0 or Autodock binding
T79 I80
score < −6.0) in the twelve crystal structures of TREK channels
Q105
(
Supplementary Fig. 3). Some of these predicted binding sites
M4
M3
were equivalent in the different crystal structures. For example,
sites 1, 2, 4–8, 11 and 21 were all binding pockets consisting of the
M1, M2, M4 and P1 helices and shared conserved residues
M2 M1
Q76
(Supplementary Fig. 4a–f). Thus, according to the locations and
compositions of residues, we assigned all 36 predicted binding
sites to 11 groups (groups A–K). All sites were ranked based on
the Glide G-score and Autodock binding score, respectively
(
Supplementary Fig. 3b). Regardless of docking methods (Glide
c
d 4
1
0
.0
.8
0
0
and Autodock), the predicted binding sites in groups A and B
always had the lowest docking scores (Glide G-scores < −7.5 or
Autodock binding score < −9.0). In group A sites (sites 1, 2, 4–8,
**
3
0.6
0.4
WT
1
1 and 21), TKDC was surrounded by hydrophobic residues,
Q76A
I 80A
20
*
which provided an unfavorable environment for the charged
sulfonyl group of TKDC (Supplementary Fig. 4a–f). Therefore,
these sites may not bind to TKDC. In group B, there was only one
member, i.e., site 3, which had the third lowest G-score and the
lowest Autodock binding score. Site 3 was located in the extra-
cellular cap domain consisting of two helices from different
subunits in TREK-1 (Fig. 2a). This site was only observed in a
crystal structure of TREK-1 (Protein Data Bank (PDB) code
0
0
.2
.0
L102A
*
1
0
0
–
7
–6
Log (M)
–5
–4
Fig. 2 Bi nd i ng mo d e of T K DC t o T R EK- 1. a B in di ng si t e of TKD C i n th e
e xtr ac el l ul a r c ap of the T RE K- 1 c ha nnel . b D e t ail e d v iew of t he int er ac t io ns
b e tw e e n T KD C a n d t h e T REK- 1 ch a n ne l. T K DC a n d t he re s id u es in v olv e d i n
b in d in g are s hown as s t ic ks. T he p ro t e in is sh o wn in a c art o o n and sur fa ce
d e pic tio n . c D os e- de pe nde nt in hibi ti on of T K DC on t he WT and mut an t
T REK-1 channels. d H is to g ram s s u m m ar izin g t h e ha lf- in h ib it o ry
c on ce nt rat io n s f or WT (n = 7), L 1 02A ( n = 10), Q 76A (n = 6 ) a n d I 80A
n = 3) mu ta nt TR EK- 1 c ha nne ls. IC 50 value s we re obt ained via dose–r es po n se
fit t in g. T he K ru s ka l –Wa ll is te st wa s u se d f or st a ti sti c al ana ly s is; * indi c a tes
P < 0. 05 , ** i n di c a te s P < 0. 01 . T he da ta a re s ho w n a s th e me an ± s. e. m
4
TWK). No inhibitors of K2P channels have been reported to
bind to site 3. In site 3, TKDC showed a more reasonable binding
pose with its hydrophobic groups inserted into a gap between the
extracellular helices and the polar sulfonyl group exposed to the
extracellular solvent (Supplementary Fig. 4g, h). Thus, site 3 was
considered as the possible binding site of TKDC in TREK-1 and
was selected for further analysis. Notably, site 3 is an extracellular
site, which is consistent with the results of the inside-out and
outside-out patch experiments. In the second stage, considering
(
the receptor flexibility and environmental effects, RosettaLi-
the WT TREK-1 channel (Supplementary Table 1), indicating
that the polar side chains of these residues were not critical for
TKDC binding.
gand3
0–33
was applied to accurately characterize the docking
mode of TKDC into the identified binding site, i.e., site 3. The top
ten docking models with the lowest binding energy exhibited
good structural convergence and were consistent with the first-
stage docking results. In the docking model with the lowest Mutagenesis of TRAAK extracellular cap. Compared with
binding energy (Fig. 2), the bicyclic group of TKDC was TREK-1 and TREK-2, TRAAK was less sensitive to TKDC.
embedded in a deep cavity formed primarily by residues I80, Q83, According to the multiple sequence alignment of the extracellular
L102 and Q105. Near the bottom of the extracellular cap domain, caps of TREK channels, TRAAK has three unique residues, A35,
the chlorophenyl group of TKDC was positioned in a shallow E38 and V42, which are situated in the binding site of
cavity formed by Q76, T79 and I80. The binding pose of TKDC in the extracellular cap (Fig. 3a). We substituted these residues
TREK-2 was similar to that in TREK-1 (Supplementary Fig. 5).
(A35, E38 and V42) of TRAAK with the corresponding residues
The molecular docking of the TKDC/TREK-1 complex (Q76, T79 and Q83, respectively) in TREK-1, and tested the
suggests that TKDC might bind to the extracellular cap of TKDC-induced inhibition of these mutant TRAAK channels. As
TREK-1. To verify this possibility, we mutated the residues in the shown in Fig. 3b, c, the substitution of these residues resulted
putative TKDC-binding site and examined the inhibition of in enhanced inhibition of TRAAK by TKDC, indicating that
mutants by TKDC. Mutagenesis manipulation revealed that TKDC bound to the proposed binding site in the extracellular
the kinetics of TREK-1 was intact (Supplementary Fig. 6a). cap of TRAAK.
Compared with wild-type (WT) TREK-1, the Q76A, I80A, and
The available crystal structures of TRAAK do not display the
L102A mutants were less sensitive to TKDC (Fig. 2c and proposed extracellular binding site. To further investigate the
Supplementary Fig. 6c, d). The mutation of these residues into structural basis for the enhanced TKDC-induced inhibition in
alanine significantly increased the IC50 for TKDC-induced mutants, we built homology models of three TRAAK mutants
inhibition by at least -twofold (Fig. 2d). These results are (A35Q, E38T and V42Q) using the TREK-1 crystal structure as
consistent with our docking results and suggest that Q76, I80 and the template and performed docking of TKDC to these mutant
L102 are important residues for TKDC binding to the TREK-1 models. In the docking models of the A35Q mutant, the outward-
channel. The T79A, Q83A, and Q105A mutants showed no facing Q35 with a larger side chain could provide a larger contact
significant changes in TKDC-induced inhibition compared with surface with TKDC than the alanine in WT TRAAK (Fig. 3d).
N A TU RE C OMM U N IC AT I ON S |8: 3 7 8
|D OI : 1 0 . 1 0 3 8/ s 4 1 46 7 - 0 1 7 - 0 0 4 99- 3 |w ww . n atu r e . c om /n a tu r e co m mu n i ca ti o n s
3

A R T I C L E
N AT URE CO M MU N ICATI ON S | D O I: 10 . 10 38/s4 14 67 -0 17 - 00499- 3
a
E1
E2
29
38
48
58
68
TRAAK
TREK-1
TREK-2
b
c 120
1.0
0.8
0.6
0.4
0.2
0.0
100
80
WT
4
0
0
0
A35Q
E38T
V42Q
**
**
2
**
–
7
–6
–5
–4
–3
WT TRAAK
L61
A35Q
E38T
V42Q
Log (M)
d
e
f
V42
L61
Q42
L39
L61
V42
E38
L39
T38
E38
A35
L39
Q35
E64
E64
A35
E64
TKDC in A35Q mutant
TKDC in V42Q mutant
TKDC in E38T mutant
Fig. 3 S ub st i t ut io ns of k ey r e si due s i n t he e x t ra ce l l ul ar ca p of T RA AK c hanne l . a M ult i ple se que nc e al ig nme nt f or t he e xt r ac e llul ar c aps of T RE K- 1, T RE K- 2 ,
a n d T RAA K c ha n ne ls . b D ose - de pende nt i nhi bi t io n of T KD C on t he W T and mut ant T RAAK c hanne l s. c H ist og r ams summa riz ing t he hal f- in hi bi to ry
c o nc en tra t io ns of TK DC fo r t he W T (n = 8) , A35 Q ( n = 8) , E 38T ( n = 6) and V4 2 Q (n = 7 ) mut ant T RAAK c hanne l s. M ut at ion s of t he se re si d ue s i n t h e
ex tr a cel l ular c a p s ho wed e nha nce d i nhi bi t or y e ff e ct s of T KD C. IC 5 0 va lue s we r e obt ai ne d vi a dose–re s ponse fit t ing . O ne -wa y AN O VA wi t h po st h oc LS D
t es t wa s us e d f or s t a ti s ti c al an al ys i s [ F (3 , 2 0 ) = 1 8.5 5 1]; ** i ndi c at e s P < 0 .0 1 . T he dat a ar e show n as t he me an ± s. e .m. d, e D oc ki ng mode s o f T KD C i n
A 3 5Q TR AA K d, in V 42 Q T RAAK e, an d i n E 38T T RAAK f. T he pr ot e in is show n as a c art oo n. T KD C and k ey re s idue s ar e show n as st i ck s
These interactions might help stabilize the insertion of TKDC in comparable to those observed for TKDC-induced inhibitions
the extracellular cap. In the docking models of the V42Q mutant, (Fig. 4c). These observations indicate that the sulfonyl group
in contrast to a valine interacting with nearby hydrophobic makes a negative contribution to the binding of TKDC to
residues (e.g., L61), Q42 flipped outward to yield an expanded TRAAK, but is acceptable for TREK-1 and TREK-2, both of
cavity between the E1 and E2 helices. As a result, the bicyclic which lack negatively charged residues in the center of the
hydrophobic group of TKDC could insert deeply into the channel binding site. In the docking models of TRAAK/28NH, 28NH
cap, as shown in Fig. 3e. In the WT TRAAK, E38 was located in displayed a reasonable binding pose with its hydrophobic groups
the center of the gap between the E1 and E2 helices. This deeply inserted into the proposed extracellular site of TRAAK
negatively charged residue might repel TKDC, which contains a (Fig. 4d).
negatively charged sulfonyl group. As shown in Fig. 3d–f, TKDC
and the side chains of the mutants were arranged to maximize the
distance between E38 and the sulfonyl oxygens in both docking Ligand-induced blockage of ion conductive pathway. In
models. In the docking models of the E38T mutant, the negatively the experiments mentioned above, we identified the binding
charged E38 was substituted with a polar threonine residue. site and characterized the binding mode of TKDC to TREK-1.
Because of the reduced electronic repulsion in E38T, TKDC With this knowledge, we aimed to simulate the process of
might have a higher affinity with the extracellular binding site. the TKDC-induced inhibition of TREK-1 using molecular
These electronic interactions could be critical in the binding of dynamics (MD) simulation approach. A TREK-1/TKDC complex
ligand to the K2P channel at the extracellular cap; this topic is simulation system (termed Scomplex) was built by inserting the
discussed later.
docking model of TREK-1/TKDC into an explicit membrane
environment. To investigate the influence of TKDC,
system without TKDC (termed Sapo) was also established as a
a
Ligand-binding mode of TRAAK examined by a TKDC derivative. control. To build the Sapo system, we removed TKDC from
The docking models of A35Q and V42Q TRAAK suggest that the docking model of TREK-1/TKDC and placed the protein
ionic interactions between E38 and the sulfonyl group of TKDC in the same environment as that of the Scomplex system. Each
might affect the binding of TKDC to TRAAK. We designed a system was replicated to perform two to three independent
ligand termed 28NH by excluding the sulfonyl group in TKDC simulations. In the Scomplex simulations, the interactions between
(Fig. 4a). 28NH significantly inhibited the TRAAK channel with TKDC and TREK-1 were stable. The dominant pose of TKDC
an IC50 of 11.8 ± 2.1 μM, which was almost sevenfold lower than (Supplementary Fig. 7) was very similar to its starting pose
the value obtained with TKDC (Fig. 4b, c). However, the IC50 (Fig. 2b). To further characterize the binding mode of TKDC to
values for 28NH-induced inhibition of TREK-1 and TREK-2 were TREK-1, we calculated the frequency of interaction between a
4
N A TU RE CO MMU N I CA TI ON S |8: 3 7 8
|D OI : 1 0 . 1 0 3 8/s 4 1 4 67 - 0 1 7 - 0 0 4 99- 3 |w ww . na tu r e . c o m /n atu r e c o m m u n ic a tio n s

NA TUR E COMM UNICAT IO NS | D OI: 1 0.1 0 38/ s4 1 467 - 01 7 - 004 99-3
A R T I C L E
a
b
a
b
1
0
.0
.8
Compound 2 (28NH)
O
H
0.6
N
N
E1
E2
0
0
0
.4
.2
.0
TKDC
CI
28NH
–
7
–6 –5 –4 –3
Log (M)
Conductive
pathway
c
d
1
00
Apo TREK-1
TREK-1/TKDC complex
TKDC
28NH
**
V42 L61
8
6
4
2
0
0
0
0
0
L39
Fig. 5 D if f e re nt c onf ormat ion s of T RE K-1 obt ai ne d f rom M D simu l ati on s.
a Conf or mat i on of T RE K-1 in t he apo sys t e m Sa po. b Conf or mat io n o f T RE K- 1
in t he c omple x sys t e m S com ple x wi t h T KD C. T he pr ot e in is sho wn a s a
c art oo n. T he e xt r ac ell ular E 2 he li x (cyan) and t he se le c t iv it y fil t er b el ow
E38
A34
E64
(
green and yellow) ar e show n as sur f ace s . T he filt e r in g re e n an d t h e E 2
he li x in c yan ar e in t he same subu nit . T he ye l low filt e r is in t he o t he r
subu nit . T KD C is show n as st i ck s. T he re d das he d ar row re pr e se nt s t h e
c onduc t iv e pat hw ay. T he re d sol id ar row s ind ic at e t he mov eme nt o f t h e E 2
he li x unde r t he influe nc e of T KD C. All il lust r at ion s ar e f rom t h e en d o f
si mulat i on ru ns
Fig. 4 In hib it i on of TR EK su bfa mi ly ch ann el s by 2 8N H i n CH O c e lls .
a Ch em ic al s t ru c tu r e of 2 8N H. b D ose - de pe nde nt i nhi bi t io n of T RAAK by
T KDC a nd 2 8NH. c Th e s ta t is t ic s of IC50 v al ue s fo r t he i nhi bi t io n of T RE K-1 ,
T REK-2 a nd TR AA K c ha nn el s by T KD C an d 2 8N H. IC 50 v al ue s we r e
o b ta i ne d v ia do s e–r e sp ons e fit t in g. T he unp ai r e d t- te s t w as us e d f or
s ta t i st i c a l a na l y si s [ t ( 7) = 1 .02 7 an d n = 9 fo r T REK- 1, t (4 ) = 0 .9 10 and
n = 6 f or TR EK-2 , a nd t ( 6 ) = 5 .7 2 4 an d n = 8 fo r T RAAK ]; ** i nd ic at e s
P < 0 .0 1. Th e nu mbe rs in t he ba r s i ndi ca te t he numb e r of ce l l s s t udie d pe r
c o nd i ti on . Th e da t a a r e s how n as t he me an ± s.e .m. d Bi ndi ng m ode l of
chemicals from the SPECS database were screened, and
5 hits were selected for bioassay (Fig. 6a). Two compounds
with different chemotypes from TKDC (TKN1 and TKN2)
displayed significant inhibitory effects on the current of TREK-1
Fig. 6). Their IC50 values were revealed to be 6.5 ± 1.7 μM and
.8 ± 0.6 μM, respectively (Fig. 6c). These two compounds also
2
(
3
showed inhibitory effects on TREK-2 and TRAAK (Supplemen-
tary Fig. 9 and Supplementary Table 3). These two chemotypes
have not been reported as inhibitors of TREK channels. This
result also demonstrates that the extracellular cap can be a
binding target of inhibitors belonging to different chemotypes.
2
8NH t o TR AA K. Th e pr o t e in i s sh own as a ca r t oon. 2 8N H an d k ey
r es i d ues a r e s ho wn a s s t ic ks
particular residue and TKDC in the Scomplex system simulations.
The ‘interaction’ is defined based on distance: when a residue has
one or more atoms within 0.4 nm of TKDC, it interacts directly
with TKDC. In the last 200 ns trajectory of each Scomplex system
simulation, I80 and L102 interacted directly with the bicyclic
group of TKDC at a high frequency (more than 0.85). Each of
Q76, T79 and I80 interacted with the chlorophenyl group of
TKDC at a frequency greater than 0.6. This result is highly
consistent with our docking and mutagenesis studies.
In the last 200 ns simulations, the conformations of the
TREK-1 channel in the system Scomplex were different from the
conformations observed in the system Sapo (Fig. 5). Compared
with the apo system Sapo without an inserted ligand, Scomplex had
a wider gap between the E1 and E2 helices, especially at
their bottoms near the selectivity filter (Supplementary Table 2).
In Scomplex, the average minimal distance between the bottom
of the E2 helix (cyan in Fig. 5b) and the filter loop upstream
pore helix P1 in the other subunit (yellow in Fig. 5b) was only
Antidepressant-like effects of TKDC in mice. To examine the
antidepressant-like effect of TKDC, we chronically administered
it to naive mice for 10 days. In the forced swimming test, the
chronic administration of TKDC resulted in a decrease in
immobility compared with the results in the vehicle-treatment
group (Fig. 7a). In contrast, fluoxetine did not show anti-
−1
depressant effects even at a higher dose (10 mg kg ). In the tail
suspension test, we observed that TKDC induced a dose-
dependent decrease in immobility (Fig. 7b). In the open field
test, mice receiving TKDC were more likely to move into the
central area (Fig. 7c). There was no obvious difference in the total
distance traveled among groups, which excluded the possibility
that the psychomotor substances may elicit false positive results
(
Fig. 7d). All tests showed that 10-day chronic treatment
with TKDC is able to induce antidepressant-like effects at low
doses (1 and 5 mg kg ), whereas fluoxetine treatment requires a
0
.30 ± 0.08 nm, which suggested that the bottom of E2 directly
−1
contacted the outer mouth of the selectivity filter and obstructed
the extracellular ion conductive pathway of TREK-1 (Supple-
mentary Fig. 8). In contrast, in the apo system Sapo in
which TKDC did not occupy the binding site, this distance was
longer time and higher doses. Similarly, a fast antidepressant-like
effect of TKDC was also observed in the acute treatment
(Supplementary Fig. 10).
0
.56 ± 0.17 nm, indicating that helix E2 left a clear extracellular
Discussion
conductive pathway in the simulation (Fig. 5a and Supplementary
In this research, we show that TKDC is an inhibitor of TREK
channels. Using this compound as a probe, we find that the
extracellular caps of the TREK channels contain small-molecule
Fig. 8).
Other inhibitors targeting the identified pocket. If the binding binding sites. We apply a hybrid approach to validate the
site of TKDC is reasonable, there may be more inhibitors identified binding site from different perspectives. First, outside-
recognizing this site. To further verify the binding site of out patch experiments indicate that TKDC acts directly on the
TKDC in the extracellular cap, we performed a structure-based extracellular side of TREK-1. Second, molecular docking is used
virtual screening targeting the identified site. More than 200,000 to predict potential binding sites, and mutagenesis is applied to
N A TU RE C OMM U N IC AT I ON S |8: 3 7 8
|D OI : 1 0 . 1 0 3 8/ s 4 1 46 7 - 0 1 7 - 0 0 4 99- 3 |w ww . n atu r e . c om /n a tu r e co m mu n i ca ti o n s
5

A
R
T
I
C
L
E
N AT URE CO M MU N ICATI ON S | D O I: 10 . 10 38/s4 14 67 -0 17 - 00499- 3
a
b
c
O
O
1.0
N
0
.8
0.6
.4
H
OH
Compound 3 (TKN1)
0
HO
O
TNK1
TNK2
0.2
0.0
NH
N
S
O
–
7
–6
–5
–4
Compound 4 (TKN2)
Log (M)
Fig. 6 In hib it or s of TR EK-1 i de nt i fie d i n st r uc t ur e -base d v i rt u al s cr e e ning . a Car t oon show ing supe r impos ed bin ding mode l s of 2 5 hit s . b Che mi ca l
s tr u ct u r es of t he di s c ov er ed i nhi bi t or s of T REK- 1. c D ose - de pe nde nt inh ibit i on of T RE K-1 by T KN 1 and T KN 2 . I C 5 0 va lue s we r e obt ai ne d vi a dose –re spo ns e
fit ti n g
characterize the residues within the site. Third, a chemical 5-HT2B receptor and can also inhibit TREK channels7, 16. The
probe derived from TKDC is designed to further investigate the crystal structure of the TREK-2/fluoxetine complex suggests that
extracellular binding site of the TRAAK channel and validate the fluoxetine binds to a fenestration adjacent to the pore filter
2
proposed binding model. Finally, virtual screening is performed entrance of TREK-2 (ref. ). Different from the binding site in
to discover more small-molecule active compounds targeting this the extracellular cap, this fenestration is highly conserved. The
site. The discovery of various inhibitors demonstrates that the residues interacting with fluoxetine are mostly identical among
extracellular cap may accommodate small-molecule modulators. TREK channels (Supplementary Fig. 11). In contrast, the residues
The extracellular ion pathway is functionally important in composing the binding site in the cap are not conserved
the ligand-induced inhibition of K2P channels. Earlier studies of among the K2P family (Supplementary Fig. 11). Within the
K2P channels show that ruthenium red binds to the residues TREK subfamily, different members have varied residue
(
D135 in TREK-2 and E70 in TASK-3) lining the wall of the compositions in the cap. Considering the unique residue E38 in
2
2–25
extracellular ion pathway to inhibit TREK-2 and TASK-3
.
TRAAK, we were able to design a chemical probe, 28NH, which
Braun et al. reports that TREK-1 is not sensitive to ruthenium lacks a negatively charged sulfonyl group and can significantly
red. However, ruthenium red inhibits the I110D TREK-1 mutant, inhibit TRAAK (IC50 = 11.8 ± 2.1 μM). Each K2P channel has
which has one isoleucine mutated to aspartate within its distinctive features in the binding site of the extracellular cap. For
2
2
extracellular ion pathway . In the current study, the proposed example, TASK-5, THIK-1 and THIK-2 have multiple positively
inhibitors, e.g., TKDC, do not interact directly with the residues charged residues but few negatively charged ones in their binding
within the extracellular ion pathway, but obstruct the pathway by sites. They might have high binding affinities to a ligand with a
inducing conformational changes of the extracellular helices. negatively charged group (Supplementary Fig. 11). TRAAK,
Our MD simulations suggest when TKDC binds to the groove TWIK-3 and TALK-2, which have more negatively charged
consisting of the E1 and E2 helices, the E2 helix moves towards residues in the binding sites, might favor a ligand with a positively
the selectivity filter and directly interacts with the pore region on charged group (Supplementary Fig. 11). The distinctive structural
the different subunit to block the extracellular ion pathway. An features of the proposed binding site can be directly applied to
earlier crosslinking study of the other K2P channel TASK-3 by the design of specific inhibitors.
Clarke et al. shows that the E70 residue at the bottom of E2 helix
2
3
and the H98 residue in the pore are close to each other . By
Methods
mutating one E70 residue and one H98 residue to cysteine in Molecular biology and cell transfection. cDNA of the human TREK-1, TREK-2
TASK-3, they found that crosslinking could occur between the and TRAAK channels were subcloned into the pEGFPN1 expression vector
Cys-70 and Cys-98 residues on different subunits of the dimer.
This finding supports our assumption that the E2 helix can access
the pore region and physically block the extracellular ion pathway
in a ligand-induced allosteric transition.
(
Invitrogen), respectively. Mutations were introduced into TREK-1 or TRAAK by
PCR using the QuickChange XL sitedirected mutagenesis kit (Agilent Technolo-
gies) and were subsequently confirmed by DNA sequencing.
Chinese hamster ovary (CHO) cells were provided by the Cell Bank of Chinese
Academy of Sciences (Shanghai, China). CHO cells were maintained in DMEM/
F12 (Gibco) supplemented with 10% fetal bovine serum (Gibco) and 100 μg ml
penicillin-streptomycin (Cellgro) in a humidified incubator at 37 °C (5% CO ). For
electrophysiology experiments, cells plated in 35 mm tissue culture dishes (Corning
Incorporated) were transiently transfected with the plasmids using PolyJet™ reagent
(SignaGen) according to the manufacturer’s instructions. Primers used in this
study are listed in Supplementary Table 4.
−
1
In this work, we identify a TKDC-binding site in the extra-
cellular cap. However, the proposed binding site is only shown in
the crystal structure of TREK-1 (PDB code 4TWK), but not in the
2
1
3–15, 17
structures of the other K2P channels, e.g., TRAAK
.
Mutations of the residues A35, E38 and V42 within the binding
site in the extracellular cap of TRAAK significantly increased
the TKDC-induced inhibition, indicating that the proposed site Electrophysiology. All electrophysiological recordings were obtained using
patch-clamp recordings in two different configurations: (1) whole cell for
macroscopic current dose–response curves; and (2) outside-out and inside-out
for macroscopic currents recordings.
All patch-clamp recordings were conducted 36–96 h after transfection with an
was involved in modulating channels. Proteins are dynamic;
crystal structures represent some of their numerous dynamic
conformations. TKDC might insert into the extracellular cap of
TRAAK in different conformations from the known crystal Axopatch-200B amplifier (Molecular Devices) at room temperature (23–25 °C).
The microelectrodes fashioned from 1.5 mm thin-walled borosilicate glass with
structures.
filament were pulled from Flaming/Brown type micropipette puller (P-97;
As an inhibitor targeting TREK-1, TKDC exhibited obvious
antidepressant-like effects, which is consistent with previous
SUTTER INSTRUMENT) with the resistances of 3–7 MΩ (whole cell) or 7–15 MΩ
(
2
outside-out) when filled with a solution containing (in mM): 140 KCl, 2 MgCl , 10
studies showing antidepressant behaviors in TREK-1 knockout EGTA, 1 CaCl
2
, 10 HEPES (pH 7.3, adjusted with KOH). External solution
34, 35
contained (in mM): 150 NaCl, 5 KCl, 0.5 CaCl , 1.2 MgCl , 10 HEPES (pH 7.3,
mice
. In the animal behavior tests performed in this work,
2
2
adjusted with NaOH). The current signals were filtered at 1 kHz and digitized at a
the antidepressant-like effects of TKDC were observed and
compared with the effects of a known antidepressant drug,
10 kHz sampling frequency by using DigiData 1440 A.
In the outside-out patch recordings and inside-out patch recordings, the
fluoxetine. Fluoxetine is well-known as an antagonist of the inhibitor (10–30 μM TKDC) was applied via dish perfusion. The bath level was
6
N A TU RE CO MMU N I CA TI ON S |8: 3 7 8
|D OI : 1 0 . 1 0 3 8/s 4 1 4 67 - 0 1 7 - 0 0 4 99- 3 |w ww . na tu r e . c o m /n atu r e c o m m u n ic a tio n s

NA TUR E COMM UNICAT IO NS | D OI: 1 0.1 0 38/ s4 1 467 - 01 7 - 004 99-3
A R T I C L E
a
b
Forced swimming test
Tail suspension test
2
1
1
00
50
00
1
20
**
*
*
8
0
0
0
*
1
2
12
12
12
12
**
4
5
0
0
11
12
12
12
11
Veh
10
0.5
1.0
5.0
Veh
10
0.5
1.0
5.0
FLX
TKDC (mg/kg)
FLX
TKDC (mg/kg)
c
d
Open field test
Open field test
25
20
15
10
5
0
5
00
00
*
*
**
4
300
200
100
0
1
2
10
10
12
12
12
10
10
12
12
Veh
10
0.5
1.0
5.0
Veh
10
0.5
1.0
5.0
FLX
TKDC (mg/kg)
FLX
TKDC (mg/kg)
Fig. 7 Ch ro ni c a dmi ni s tr at i on of T KD C i n mi ce . a T ime sp en t i mmobil e in t he f orc e d swi mming t e st af t e r admi nistra t ion of T KD C and fluox et i ne ( FLX )
[
5
o n e- wa y ANO VA wi th pos t hoc LS D t e st , F (4 ,5 5 ) = 4.2 0 ]. V eh ind ic at e s t he ve hi c le -t re at me nt g roup . T KD C wa s admi nist e r ed at dose s o f 0. 5 , 1 a nd
m g k g . F luo xe t ine wa s ad min is t e red at a dos e of 1 0 mg kg . b T ime spe nt immob ile in t he t ai l sus pe nsion t e st af t e r admi nist r at ion of T KD C a nd
−1
−1
flu o xeti ne [on e -wa y ANOV A w it h pos t hoc LS D t e st , F (4 ,5 3) = 2 .5 5 ]. c P e rc e nt ag e of dis t anc e t ra ve le d in t he c e nt e r of t he fie ld ove r t he t o ta l d ist a nc e
t ra ve l ed a f t er a dmi ni s tr a t i on of T KD C an d fluox e t i ne i n t he ope n fie ld t e st (one - way AN O VA wi t h post hoc L SD t e st , F (4 ,5 1 ) = 3. 81). d T ota l d ist a nc e
t ra ve l ed a f t er a dmi ni s tr at i on of T KD C an d fluox e t i ne i n t he ope n fie ld t e st (one - way AN O VA wi t h post hoc L SD t e st , F (4 ,5 1 ) = 2 .0 2 ). T he num be rs i n t h e
b a rs i nd ic a t e t he nu mbe r of ce l l s st u di ed pe r co ndi t io n. T he r e sul t s ar e show n as t he me an ± s. e .m; * ind ic at e s P < 0 .0 5 , ** ind ic at e s P < 0 .0 1
kept as low as possible to reduce noise. Currents were recorded at a pipette
potential of + 80 mV.
4.23 (s, 2H), 3.72 (t, J = 5.8 Hz, 1H), 3.67 (t, J = 6.0 Hz, 1H), 2.91 (t, J = 5.7 Hz, 1H),
2.78 (t, J = 5.8 Hz, 1H).
Then, to a solution of 2-bromo-1-(3,4-dihydroisoquinolin-2(1H)-yl)ethanone
(
254 mg) and 4-chloroaniline (140 mg, 1.1 equiv) in DMF (5 ml) was added K
2 3
CO
Flow cytometry analysis. CHO-K1 cells were transfected with pFLAG-TREK1
for 24 h. After washed with phosphate-buffered saline (PBS), the cells were
disassociated with a non-enzyme cell dissociation solution (Sigma). After 10 μM
TKDC incubation for 20, 40 min and 1 h at room temperature, cell surface TREK-1
expression in CHO-K1 cells was analyzed by flow cytometry. Transfected cells
expressing FLAG-tagged TREK-1 were incubated with monoclonal ANTI-FLAG
M2-FITC antibody (Sigma). Data were collected with a flow cytometer (FACS
Calibur, BD Bioscience) and analyzed with Flow Jo.
(
276 mg, 2 equiv). The solution was heated for 1 h at 60 °C. The cooled solution
was diluted with ethyl acetate (20 ml), water was added thereto, followed by
extraction with ethyl acetate. The ethyl acetate layer was washed with brine and
then dried over anhydrous sodium sulfate, and the solvent was removed by
evaporation. The separation by silica gel chromatography column with petroleum
ether/ethyl acetate (15/1) gave the desired product, compound 2 (147 mg, 49%).
LC-MS: m/z (M + H) 300.9. H NMR (400 MHz, CDCl
6
3
®
1
3
) δ 7.33–7.12 (m, 6H),
.58–6.55 (m, 2H), 4.98 (br, 1H), 4.78 (s, 1H), 4.60 (s, 1H), 3.93–3.87 (m, 3H),
.66 (t, J = 5.9 Hz, 1H), 2.95 (t, J = 6.0 Hz, 1H), 2.89 (t, J = 6.0 Hz, 1H).13C NMR
) δ 167.79, 167.74, 146.02, 134.98, 133.89, 132.99, 131.74, 129.18,
29.08, 128.38, 127.33, 122.94, 126.92, 126.74, 126.68, 126.21, 122.19, 114.08, 45.99,
5.65, 45.55, 44.61, 42.09, 40.29, 29.28, 28.48; HPLC (Agilent HC-C18, CH CN:
(
101 MHz, CDCl
3
Confocal microscopic analysis of channel internalization. In eight-well
chamber (Thermo), CHO-K1 cells were co-transfected with PEGFP-N1-TREK1
and OFP-Rab5a. After 24 h, cells were treated with 10 μM TKDC for 20 min at
room temperature and DMSO as control. At the end of incubation, cells were
washed with PBS, fixed with 4% paraformaldehyde and washed with PBS. Images
were taken at × 100 with scan zoom 1.74 using Leica SP8 and 3D reconstruction
was done using Imaris software (Bitplane) and co-localization analysis was quan-
tified using Imaris software. An automatic threshold of red and green channels was
selected. The data represent images taken from 18 different areas (n = 18).
1
4
3
−1
water = 70:30, flow rate = 1.0 mL min , λ = 254 nm), purity = 97.2%.
Molecular docking. The crystal structures of TREK-1 (Protein Data Bank (PDB)
code 4TWK), TREK-2 (PDB codes 4BW5, 4XDK, 4XDJ and 4XDL) and TRAAK
(
PDB code 4I9W, 4WFH, 4WFG, 4WFF, 4WFE, 4RUE and 4RUF) were used to
1, 2, 4, 7
detect ligand-binding pockets
. TKDC was docked to each pocket in all 2
structures using Schrodinger Glide software in SP mode with default parameters 8.
Pockets binding to TKDC with Glide G-scores below −3.0 were considered as
possible binding sites. To validate the docking models generated using Glide, we
also docked TKDC into all pockets using Autodock4 (ref. 2 ). Searches were
performed using Lamarckian Genetic Algorithm with default settings. Pockets
binding to TKDC with Autodock binding scores below −6.0 were also considered
Chemicals. All the tested compounds are known structures belong to our in-house
library developed over time. The in-house library includes approximately
9
1000 small molecules purchased or synthesized for our previous and current
research projects of drug discovery.
TKDC. (N-(4-cholorphenyl)-N-(2-(3,4-dihydrosioquinolin-2(1H)-yl)-2-
oxoethyl)methanesulfonamide) and other compounds applied in virtual screening
were purchased from SPECS. We synthesized 28NH, (2-((4-chlorophenyl)amino)-
as possible binding sites. Considering the energetic effects of the solvent
environment and the receptor flexibility, RossettaLigand application3
0–33
was
1
-(3,4-dihydroisoquinolin-2(1H)-yl)ethanone).
To synthesize 28NH, we first synthesized 2-bromo-1-(3,4-dihydroisoquinolin-2
applied to generate accurate molecular docking models of TKDC to the crystal
structure of TREK-1 (PDB code 4TWK). TKDC was initially placed in the center of
the proposed binding pocket identified in the Glide docking. Its center of mass was
constrained to move within 1 nm diameter sphere, where it was allow moving
freely during the docking process. Residue side chains in the binding site were
repacked using a rotamer library. The docking model with the lowest binding
energy was selected for analysis. Modeller was used to create homology models of
human wild-type TREK-2, wild-type TRAAK and three mutant TRAAK channels
based on the crystal structure of TREK-1. In each case, model with the lowest root
mean square deviation from the structure with PDB code 4TWK was applied to
ligand docking using RossetaLigand application.
(
1H)-yl)ethanone. To a solution of 1,2,3,4-tetrahydroisoquinoline (980 mg) and N,
N-dimethylaniline (932 mg, 1.1 equiv) in DCM (35 ml) was added 2-bromoacetyl
bromide (1.49 g, 1.1 equiv), followed by stirring under ice-cooling for 30 min. After
completion of the reaction, the solution was diluted with DCM (50 ml), water was
added thereto, followed by extraction with DCM. The DCM layer was washed with
brine and then dried over anhydrous sodium sulfate, and the solvent was removed
by evaporation. The separation by silica gel chromatography column with
3
6
petroleum ether/ethyl acetate (10/1) gave the desired product, compound 5
1
(
1.42 g, 80%). H NMR (400 MHz, CDCl
3
) δ 7.21–7.19 (m, 4H), 4.70–4.51 (m, 2H),
N
A
T
U
R
E
C
O
M
M
U
N
I
C
A
T
I
O
N
S
|
8
:
3
7
8
|D OI : 1 0 . 1 0 3 8/ s 4 1 46 7 - 0 1 7 - 0 0 4 99- 3 |w ww . n atu r e . c om /n a tu r e co m mu n i ca ti o n s
7

A R T I C L E
N AT URE CO M MU N ICATI ON S | D O I: 10 . 10 38/s4 14 67 -0 17 - 00499- 3
Molecular dynamics simulations. The crystal structure of TREK-1 (PDB code
of the testing period; the first 2 min were excluded. Because most mice are very
active at the beginning of the test, and the potential effects of the treatment can be
minimized during the first 2 min.
4TWK) and the best docking model of TREK-1/TKDC were used to build the
models of TREK-1 with and without TKDC. Missing atoms and loops were
added back in most favored position without clashes using Discovery Studio 3.037.
The models of the TREK-1 with and without TKDC were inserted in a POPC
Forced swimming test. The mice were performed under normal light. The
apparatus was a transparent Plexiglas cylinder with 30 cm height × 10 cm diameter
and the water temperature at 23–25 °C were set to prevent mice from touching the
bottom with their limbs. Mice were allowed to swim for 8 min, and their activity
was videotaped. There was an extra 15-min swimming in the previous day. The
duration of immobility defined as floating or remaining motionless was assessed.
To avoid extremely active at the beginning of swimming, the first 2 min of
swimming was also excluded as the tail suspension test.
(
1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) lipid bilayer to establish a
ligand-binding complex system and an apo system without TKDC, respectively. In
each MD simulation system, the protein model and lipid bilayer were solvated in a
periodic boundary condition box (10 nm × 10 nm × 13 nm) filled with TIP3P
8
(
transferable intermolecular potential 3 P) water molecules3 and 0.15 M KCl.
The apo system was replicated to generate two independent simulations. The
ligand-binding system was replicated to generate three independent simulations.
All MD simulations were performed using GROMACS version 4.52 9, 40,
3
4
1, 42
which employs the CHARMM all-atom force field
. After 40 ns equilibration,
a 1-μs-long production run was carried out for each apo system. For the ligand-
binding system, a 40 ns equilibration and a 500 ns production were performed for
each simulation. All productions were performed in the NPT ensemble at a
temperature of 300 K and a pressure of 1 atm. Temperature and pressure were
Drug administration. For the in vitro study, the compounds (TKDC and 28NH)
were dissolved in dimethyl sulfoxide. For the in vivo study, TKDC was dissolved in
5‰ tween-80 and intraperitoneally administered to animals. Used as a positive
control, fluoxetine (Santa Cruz Biotechnology, USA) was dissolved in water
intraperitoneally administered at a dose of 10 mg kg . The vehicle, 5‰ tween-80,
was administered to control mice as described above. All animals were treated with
the drugs 30 min before the behavioral experiments.
4
3
−1
controlled using the velocity-rescale thermostat and the Parrinello–Rahman
4
4, 45
barostat
with isotropic coupling, respectively. Equations of motion were
integrated with a 2 fs time step, and the LINCS algorithm was used to constrain
4
6
bonds lengths . Nonbonded pairlists were generated every 10 steps using a
distance cutoff of 1.4 nm. A cutoff of 1.2 nm was used for Lennard-Jones (excluding
scaled 1–4) interactions, which were smoothly switched off between 1 and 1.2 nm.
Electrostatic interactions were computed using the Particle-Mesh-Ewald
Data analysis and statistics. Data of whole-cell voltage clamp experiments were
fitted using doseResp function. The half inhibition concentrations were derived
from fits of the dose–response curves to the function:
4
7
algorithm with a real-space cutoff of 1.2 nm. Analysis of the simulations was
performed using GROMACS and HOLE48
.
I
A À A
2
1
¼
A1 þ
;
ð1Þ
ðlog x0ÀxÞP
I0
1 þ 10
Virtual screening for TREK-1. A structure-based virtual screening strategy
2
8
where I
and A
0
and I are current amplitudes before and after application of inhibitors; A
are constants between 0 and 1; x is the concentration of inhibitor; x is the
1
was performed using Schrodinger Glide software . Approximately 200,000
commercially available compounds from the SPECS database were screened in the
first run using the high-throughput virtual screening (HVS) module in Glide. The
human TREK-1 crystal structure (PDB code 4TWK) was used as the docking
receptor and residues in the extracellular cap are defined as the binding site. Glide
G-scores was used to rank the result list and the top-ranked 20,000 candidates were
selected. Since HVS applying simplified scoring function, these selected candidates
were screened again using a standard docking module SP in Glide. The top-ranked
2
o
concentration when 50% inhibition was reached (IC50), P is the Hill constant.
In statistical analyses, the experimental data were expressed as the
mean ± s.e.m. The averaged IC50 was collected from the fitting results of each patch
lasting for five inhibitor concentrations), and then was used to compare between
(
different groups. Unless specified, an unpaired t-test was used when two means are
compared; a one-way analysis of variance (ANOVA) along with post hoc LSD test
was used when more than two means are compared. When the data distribution
was skewed, the independent-sample Kruskal–Wallis test (non parametric test)
followed by Dunn–Bonferroni post hoc test was performed. The results of animal
behavioral tests (forced swimming test, tail suspension test and open field test)
were analyzed using one-way ANOVA with post hoc LSD tests. P < 0.05 was
considered statistically significant.
10,000 candidates were rescored by the CSCORE (consensus score) module of
SYBYL 6.8 (Tripos Inc.). Compounds that had a consensus score of four or five or
that were ranked in top 10% by at least four out of five scoring functions were
selected. Using the Cluster Molecules module in Pipeline Pilot 7.5 (Scitegic, Inc.),
these selected compounds were clustered based on their two-dimensional
structures. To ensure the structural diversity of the compounds, two to three
compounds with good drug-like properties (molecular weight < 500, log P < 5
and polar surface area < 140 Å) were selected from each cluster for electro-
physiological assay.
Data availability. The authors declare that the main data supporting the findings
of this study are available within the article and its Supplementary Information files
and available from the corresponding authors upon reasonable request.
Animal behavior tests. Male C57BL/6 mice, aged 8 weeks (weighing 18–20 g)
were used for acute and chronic drug administration and behavioral tests. The mice
were housed four per cage under a 12-h light–dark cycle. Mice were raised under
stable conditions with food and water ad libitum. All animal studies and experi-
mental procedures were approved by the Animal Care Committees of the Shanghai
Institute of Materia Medica, Chinese Academy of Sciences, and experiments were
carried out in accordance with EU Directive 2010/63/EU on the protection of
animals used for scientific purposes. For acute drug administration, mice were −
Received: 30 September 2016 Accepted: 4 July 2017
1
randomly divided into five groups and singly injected (i.p) with vehicle, 10 mg kg
References
−
1
−1
−1
fluoxetine, 0.5 mg kg TKDC, 1 mg kg TKDC and 5 mg kg TKDC. For chronic
drug administration, mice were also randomly divided into five groups and i.p.
1. Piechotta, P. L. et al. The pore structure and gating mechanism of K2P
channels. EMBO J. 30, 3607–3619 (2011).
−
1
−1
−1
administrated with vehicle, 10 mg kg fluoxetine, 0.5 mg kg TKDC, 1 mg kg
2. Goldstein, S. A., Bockenhauer, D., O’Kelly, I. & Zilberberg, N. Potassium leak
channels and the KCNK family of two-P-domain subunits. Nat. Rev. Neurosci.
2, 175–184 (2001).
−
1
TKDC and 5 mg kg TKDC for 10 days (once per day). The mice were removed to
a sound-attenuating testing room 3–4 h before testing. The behavioral tests were
carried out 30 min and 24 h after acute and chronic drugs administration,
respectively.
3. Enyedi, P. & Czirjak, G. Molecular background of Leak K+ currents: two-pore
domain potassium channels. Physiol. Rev. 90, 559–605 (2010).
4.
Alloui, A. et al. TREK-1, a K+ channel involved in polymodal pain perception.
Open field test. The apparatus was made of opaque Plexiglas (50 cm × 50 cm) and
placed in a room with sound proof. A camera was positioned above the apparatus
to monitor the animal behaviors. The mice were randomly placed into the center of
the sheet following 30 min drugs administration and their behaviors in the open
field arena were recorded for 10 min using the camera. The surface of the sheet was
cleaned with 75% ethyl alcohol after each trial to remove the permeated odors from
previous animals. Locomotor activity and exploratory behavior were measured in
the open field test and each animal was placed individually at the center for a 10
min session. The total distance moved and the number of rearing was measured in
this experiment.
EMBO J. 25, 2368–2376 (2006).
5. Dominguez-Lopez, S., Howell, R. & Gobbi, G. Characterization of serotonin
neurotransmission in knockout mice: implications for major depression.
Rev. Neurosci. 23, 429–443 (2012).
6. Garry, A. et al. Altered acetylcholine, bradykinin and cutaneous pressure-
induced vasodilation in mice lacking the TREK1 potassium channel: the
endothelial link. EMBO Rep. 8, 354–359 (2007).
7. Kennard, L. E. et al. Inhibition of the human two-pore domain potassium
channel, TREK-1, by fluoxetine and its metabolite norfluoxetine. Br. J.
Pharmacol. 144, 821–829 (2005).
8.
Liu, X. X. et al. Stretch-activated potassium channels in hypotonically induced
Tail suspension test. Mice were secured on the shelf by using adhesive tape placed
approximately 1 cm from the tip of the tail. The mice were then suspended by
placing the other end of the tape on the edge of shelf at the height of 80 cm above
the floor. Six-minute session was performed for each animal. A blinded experiment
was conducted to record the amount of time spent immobile during the last 4 min
blebs of atrial myocytes. J. Membr. Biol. 226, 17–25 (2008).
9. Noel, J. et al. The mechano-activated K+ channels TRAAK and TREK-1 control
both warm and cold perception. EMBO J. 28, 1308–1318 (2009).
10. Bittner, S. et al. Endothelial TWIK-related potassium channel-1 (TREK1)
regulates immune-cell trafficking into the CNS. Nat. Med. 19, 1161–1165 (2013).
8
N A TU RE CO MMU N I CA TI ON S |8: 3 7 8
|D OI : 1 0 . 1 0 3 8/s 4 1 4 67 - 0 1 7 - 0 0 4 99- 3 |w ww . na tu r e . c o m /n atu r e c o m m u n ic a tio n s

NA TUR E COMM UNICAT IO NS | D OI: 1 0.1 0 38/ s4 1 467 - 01 7 - 004 99-3
A R T I C L E
1
1
1
1. Hwang, E. M. et al. A disulphide-linked heterodimer of TWIK-1 and TREK-1
mediates passive conductance in astrocytes. Nat. Commun. 5, 3227 (2014).
2. Woo, D. H. et al. TREK-1 and Best1 channels mediate fast and slow glutamate
release in astrocytes upon GPCR activation. Cell 151, 25–40 (2012).
40. Van der Spoel, D. et al. GROMACS: fast, flexible, and free. J. Comput. Chem.
26, 1701–1718 (2005).
41. Guvench, O. et al. CHARMM additive all-atom force field for carbohydrate
derivatives and its utility in polysaccharide and carbohydrate-protein modeling.
J. Chem. Theory Comput. 7, 3162–3180 (2011).
42. MacKerell, A. D. et al. All-atom empirical potential for molecular
modeling and dynamics studies of proteins. J. Phys. Chem. B 102, 3586–3616
(1998).
43. Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through velocity
rescaling. J. Chem. Phys. 126, 014101 (2007).
3. Brohawn, S. G., Campbell, E. B. & MacKinnon, R. Domain-swapped chain
connectivity and gated membrane access in a Fab-mediated crystal of the
+
human TRAAK K channel. Proc. Natl Acad. Sci. USA 110, 2129–2134 (2013).
1
4. Brohawn, S. G., Campbell, E. B. & MacKinnon, R. Physical mechanism for
gating and mechanosensitivity of the human TRAAK K+ channel. Nature 516,
126–130 (2014).
1
5. Brohawn, S. G., del Marmol, J. & MacKinnon, R. Crystal structure of the
44. Nose, S. & Klein, M. L. Constant pressure molecular-dynamics for molecular-
systems. Mol. Phys. 50, 1055–1076 (1983).
human K2P TRAAK, a lipid- and mechano-sensitive K+ ion channel. Science
45. Parrinello, M. & Rahman, A. Polymorphic transitions in single-crystals - a new
molecular-dynamics method. J. Appl. Phys. 52, 7182–7190 (1981).
3
35, 436–441 (2012).
1
6. Dong, Y. Y. et al. K2P channel gating mechanisms revealed by structures of
TREK-2 and a complex with Prozac. Science 347, 1256–1259 (2015).
7. Lolicato, M., Riegelhaupt, P. M., Arrigoni, C., Clark, K. A. & Minor, D. L. Jr.
Transmembrane helix straightening and buckling underlies activation of
mechanosensitive and thermosensitive K2P channels. Neuron 84, 1198–1212
46. Hess, B., Bekker, H., Berendsen, H. J. C. & Fraaije, J. G. E. M. LINCS: a linear
constraint solver for molecular simulations. J. Comput. Chem. 18, 1463–1472
1
(
1997).
4
7. Darden, T., York, D. & Pedersen, L. Particle mesh Ewald - an N.Log(N) method
for Ewald sums in large systems. J. Comput. Chem. 98, 10089–10092 (1993).
8. Smart, O. S., Goodfellow, J. M. & Wallace, B. A. The pore dimensions of
gramicidin-A. Biophys. J. 65, 2455–2460 (1993).
(
2014).
4
1
1
8. Miller, A. N. & Long, S. B. Crystal structure of the human two-pore domain
potassium channel K2P1. Science 335, 432–436 (2012).
9. Gonzalez, W. et al. An extracellular ion pathway plays a central role in the
cooperative gating of a K2P K+ channel by extracellular pH. J. Biol. Chem. 288,
Acknowledgements
The research was supported in part by the National Natural Science Foundation of China
5
984–5991 (2013).
+
2
0. Lesage, F. et al. Dimerization of TWIK-1 K channel subunits via a disulfide
bridge. EMBO J. 15, 6400–6407 (1996).
(
(
21422208), the Ministry of Science and Technology of the People’s Republic of China
2013CB910601 and 2013CB910604), the “Personalized Medicines—Molecular Sig-
2
1. Niemeyer, M. I., Cid, L. P., Valenzuela, X., Paeile, V. & Sepulveda, F. V.
Extracellular conserved cysteine forms an intersubunit disulphide bridge in the
nature-based Drug Discovery and Development”, Strategic Priority Research Program of
the Chinese Academy of Sciences (XDA12040302), the National Key Scientific Instru-
ment & Equipment Development Program of China (2012YQ0306010), the E-Institutes
of Shanghai Municipal Education Commission (E09013) and the Special Program for
Applied Research on Super Computation of the NSFC-Guangdong Joint Fund
(the second phase) under Grant No.U1501501. We also thank the computer center of
East China Normal University for computational resources.
+
KCNK5 (TASK-2) K channel without having an essential effect upon activity.
Mol. Membr. Biol. 20, 185–191 (2003).
2
2
2
2
2. Braun, G., Lengyel, M., Enyedi, P. & Czirjak, G. Differential sensitivity of
TREK-1, TREK-2 and TRAAK background potassium channels to the
polycationic dye ruthenium red. Br. J. Pharmacol. 172, 1728–1738 (2015).
3. Clarke, C. E., Veale, E. L., Wyse, K., Vandenberg, J. I. & Mathie, A. The M1P1
loop of TASK3 K2P channels apposes the selectivity filter and influences
channel function. J. Biol. Chem. 283, 16985–16992 (2008).
Author contributions
Q.L., L.C., X.C. and Y.M. conducted the majority of the experiments, including
electrophysiological assays, molecular docking, molecular dynamics simulation and
virtual screening. X.L. conducted the flow cytometry analysis and confocal microscopy
analysis. B.Z., F.G. and L.L. carried out the animal behavioral experiments. S.Z.
conducted the chemical synthesis experiments. L.C., X.C., Q.L., Y.M., Y.L. and H.Y.
prepared the manuscript. Y.L. and H.Y. conceived and supervised the project. All authors
approved the manuscript.
4. Czirjak, G. & Enyedi, P. Ruthenium red inhibits TASK-3 potassium channel by
interconnecting glutamate 70 of the two subunits. Mol. Pharmacol. 63, 646–652
(
2003).
+
5. Musset, B. et al. Effects of divalent cations and spermine on the K channel
TASK-3 and on the outward current in thalamic neurons. J. Physiol. 572,
639–657 (2006).
+
26. Fink, M. et al. A neuronal two P domain K channel stimulated by arachidonic
acid and polyunsaturated fatty acids. EMBO J. 17, 3297–3308 (1998).
2
7. Lesage, F. et al. TWIK-1, a ubiquitous human weakly inward rectifying K+
channel with a novel structure. EMBO J. 15, 1004–1011 (1996).
2
Additional information
Supplementary Information accompanies this paper at doi:10.1038/s41467-017-00499-3.
8. Friesner, R. A. et al. Glide: a new approach for rapid, accurate docking and
scoring. 1. Method and assessment of docking accuracy. J. Med. Chem. 47,
Competing interests: The authors declare no competing financial interests.
1739–1749 (2004).
2
3
3
3
3
3
3
3
9. Morris, G. M. et al. AutoDock4 and AutoDockTools4: Automated docking with
selective receptor flexibility. J. Comput. Chem. 30, 2785–2791 (2009).
0. Fleishman, S. J. et al. RosettaScripts: a scripting language interface to the
Rosetta macromolecular modeling suite. PLoS ONE 6, e20161 (2011).
1. Davis, I. W. & Baker, D. RosettaLigand docking with full ligand and receptor
flexibility. J. Mol. Biol. 385, 381–392 (2009).
Reprints and permission information is available online at http://npg.nature.com/
reprintsandpermissions/
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
2. Lemmon, G. & Meiler, J. Rosetta ligand docking with flexible XML protocols.
Methods Mol. Biol. 819, 143–155 (2012).
3. Yarov-Yarovoy, V., Schonbrun, J. & Baker, D. Multipass membrane protein
structure prediction using Rosetta. Proteins 62, 1010–1025 (2006).
4. Heurteaux, C. et al. Deletion of the background potassium channel TREK-1
results in a depression-resistant phenotype. Nat. Neurosci. 9, 1134–1141 (2006).
5. Gordon, J. A. & Hen, R. TREKing toward new antidepressants. Nat. Neurosci.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
licenses/by/4.0/.
9
, 1081–1083 (2006).
6. Sali, A. & Blundell, T. L. Comparative protein modelling by satisfaction of
spatial restraints. J. Mol. Biol. 234, 779–815 (1993).
7. DSV. Discovery Studio Visualizer v3.0. Articles (2010).
8. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L.
Comparison of simple potential functions for simulating liquid water. J. Chem.
Phys. 79, 926–935 (1983).
3
3
3
9. Hess, B., Kutzner, C., van der Spoel, D. & Lindahl, E. GROMACS 4: algorithms
for highly efficient, load-balanced, and scalable molecular simulation. J. Chem.
Theory Comput. 4, 435–447 (2008).
©
The Author(s) 2017
N A TU RE C OMM U N IC AT I ON S |8: 3 7 8
|D OI : 1 0 . 1 0 3 8/ s 4 1 46 7 - 0 1 7 - 0 0 4 99- 3 |w ww . n atu r e . c om /n a tu r e co m mu n i ca ti o n s
9