On the permeation of large organic cations through the pore of ATP-gated P2X receptors

Article abstract of DOI:10.1073/pnas.1701379114

Pore dilation is thought to be a hallmark of purinergic P2X receptors. The most commonly held view of this unusual process posits that under prolonged ATP exposure the ion pore expands in a striking manner from an initial small-cation conductive state to a dilated state, which allows the passage of larger synthetic cations, such as N-methyl-D-glucamine (NMDG⁺). However, this mechanism is controversial, and the identity of the natural large permeating cations remains elusive. Here, we provide evidence that, contrary to the time-dependent pore dilation model, ATP binding opens an NMDG⁺-permeable channel within milliseconds, with a conductance that remains stable over time. We show that the time course of NMDG⁺ permeability superimposes that of Na⁺ and demonstrate that the molecular motions leading to the permeation of NMDG⁺ are very similar to those that drive Na⁺ flow. We found, however, that NMDG⁺ percolates 10 times slower than Na⁺ in the open state, likely due to a conformational and orientational selection of permeating molecules. We further uncover that several P2X receptors, including those able to desensitize, are permeable not only to NMDG⁺ but also to spermidine, a large natural cation involved in ion channel modulation, revealing a previously unrecognized P2X-mediated signaling. Altogether, our data do not support a time-dependent dilation of the pore on its own but rather reveal that the open pore of P2X receptors is wide enough to allow the permeation of large organic cations, including natural ones. This permeation mechanismhas considerable physiological significance.

Full text of DOI:10.1073/pnas.1701379114

On the permeation of large organic cations through the
pore of ATP-gated P2X receptors
Mahboubi Harkata,b,1, Laurie Peverinia,b,1, Adrien H. Cerdanb,c, Kate Dunninga,b, Juline Beudeza,b, Adeline Martza,b
Nicolas Calimetb,c, Alexandre Spechta,b, Marco Cecchinib,c, Thierry Chataigneaua,b, and Thomas Gruttera,b,2
,
aLaboratoire de Conception et Application de Molécules Bioactives, Équipe de Chimie et Neurobiologie Moléculaire, Faculté de Pharmacie, Centre National
de la Recherche Scientifique (CNRS), UMR 7199, F-67400 Illkirch, France; bUniversité de Strasbourg, F-67000 Strasbourg, France; and cLaboratoire
d’Ingénierie des Fonctions Moléculaires, Institut de Science et d’Ingénierie Supramoléculaires (ISIS), UMR 7006, CNRS, F-67000 Strasbourg, France
Edited by Christopher Miller, Howard Hughes Medical Institute, Brandeis University, Waltham, MA, and approved March 31, 2017 (received for review January
25, 2017)
Pore dilation is thought to be a hallmark of purinergic P2X receptors.
The most commonly held view of this unusual process posits that
under prolonged ATP exposure the ion pore expands in a striking
manner from an initial small-cation conductive state to a dilated
state, which allows the passage of larger synthetic cations, such as
N-methyl-D-glucamine (NMDG+). However, this mechanism is contro-
versial, and the identity of the natural large permeating cations
remains elusive. Here, we provide evidence that, contrary to the
time-dependent pore dilation model, ATP binding opens an
NMDG+-permeable channel within milliseconds, with a conductance
that remains stable over time. We show that the time course of
NMDG+ permeability superimposes that of Na+ and demonstrate
that the molecular motions leading to the permeation of NMDG+
are very similar to those that drive Na+ flow. We found, however,
that NMDG+ “percolates” 10 times slower than Na+ in the open
state, likely due to a conformational and orientational selection of
permeating molecules. We further uncover that several P2X recep-
tors, including those able to desensitize, are permeable not only to
NMDG+ but also to spermidine, a large natural cation involved in ion
channel modulation, revealing a previously unrecognized P2X-mediated
signaling. Altogether, our data do not support a time-dependent dila-
tion of the pore on its own but rather reveal that the open pore of P2X
receptors is wide enough to allow the permeation of large organic
cations, including natural ones. This permeation mechanism has con-
siderable physiological significance.
A remarkable feature of P2X receptors is that a longer ATP
application causes a striking time-dependent pore dilation of the
channel. This process, named pore dilation, is observed only for
selected subtypes, notably homomeric P2X2, P2X4, and P2X7 and
heteromeric receptors P2X2/3 and P2X2/5 (17–21). Different
possible mechanisms have been proposed to explain pore dilation
(22, 23), but the most prevailing one posits that the open pore of
the initial I1 state progressively dilates for several seconds to form
an enlarged pore, denoted I2 state, that is permeable to large
organic cations such as N-methyl-D-glucamine (NMDG+) or
fluorescent dyes such as ethidium bromide and YO-PRO-1, a
carbocyanine DNA binding dye (2). As these cations are larger
than small inorganic ions, they are believed to be impermeable to
the I1 state. This belief has been supported by many biophysical
studies, including atomic force microscopy (AFM), patch-clamp
electrophysiology, fluorescent dye uptakes, and real-time confor-
mational change measurements (17, 18, 24–31). However, the
molecular mechanism of pore expansion is still unclear.
Very recently, a study has challenged the pore dilation para-
digm (32). The study suggests that the slow transition toward the
I2 state is not caused by a progressive change in the permeability
ratio of NMDG+ relative to Na+, which is usually determined by
measuring the hallmark shift in equilibrium potentials by patch-
clamp electrophysiology under bi-ionic conditions but rather by a
dramatic, unappreciated change of the concentrations of these
ions inside the cell. Moreover, this study showed that NMDG+
permeability is rapidly activated and can be inhibited when a
pore dilation purinergic receptor photoswitches YO-PRO uptake
|
|
|
|
spermidine
Significance
2X receptors are nonselective cation channels activated by
Padenosine 5′-triphosphate (ATP). They are integral mem-
brane proteins and form a family of trimeric receptors com-
posed of seven subunits (P2X1–P2X7) that are involved in
a wide range of physiological and pathological processes, in-
cluding pain sensation, hearing protection, taste, modulation
of neurotransmitter release, hypertension, inflammation, and
neuropathic pain (1–5). P2X receptors have, therefore, attrac-
ted attention as promising therapeutic targets (5, 6). Supported
by recent X-ray structures (7–11), a functional receptor is
composed of three subunits that assemble in the cell membrane
as a homo- or heterotrimer to form a central transmembrane
pore (12, 13). Each subunit comprises two transmembrane he-
lices, named TM1 and TM2, which are linked by an extracel-
lular domain, where the ATP-binding sites are nestled. In
response to a short application of ATP, the pore rapidly opens
on the millisecond timescale—a process known as gating—to a
state that is selective to small inorganic cations, such as Na+, K+,
and Ca2+ ions. The flow of these cations is estimated to occur at
relatively high conduction rates, from 6 to 20 × 106 ions per
second at a given driving force amplitude (14–16). This rapidly
affects the ion balance of the cell and consequently initiates
signal transduction. This open state is sometimes referred to as
I1 state (2).
Unlike many ion channels whose pore conductances remain
relatively stable over time, it is thought that prolonged ATP
applications to P2X receptors cause a striking increase over
time in the permeability of large molecules, a process dubbed
pore dilation. However, this mechanism remains poorly un-
derstood and highly controversial. Here, we use different
methods spanning single-channel recordings, photochemistry,
molecular biology, and computations to show that contrary to
longstanding view, rapid activation by ATP allows the stable
passage of large cations through the P2X pore. We further
discover that spermidine, a large natural cation known to
modulate other ion channels, is able to transit through many
P2X receptors, including those thought to be nondilating. Our
data thus reveal an unacknowledged P2X-mediated signaling.
Author contributions: A.S., M.C., and T.G. designed research; M.H., L.P., A.H.C., K.D., J.B.,
A.M., N.C., A.S., M.C., and T.C. performed research; M.H., L.P., A.H.C., K.D., J.B., A.M., N.C.,
A.S., M.C., T.C., and T.G. analyzed data; and T.G. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1M.H. and L.P. contributed equally to this work.
2To whom correspondence should be addressed. Email: grutter@unistra.fr.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1701379114/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1701379114
PNAS Early Edition
|
1 of 10

pore-lining cysteine residue is reacted with methanethiosulfonate
(MTS) reagents, suggesting that the open state of P2X receptors
exhibits measurable and stable permeability to large cations. In
line with this work, pore dilation has never been observed at the
single-channel level (16, 33), and no crystal structure of a dilated
state has been reported to date, making the study of this peculiar
state particularly difficult. Finally, the physiological relevance of
P2X pore dilation remains unclear, especially due to the un-
known identity of the natural molecules that permeate through
the dilated state. A recent work has, however, shown that ge-
netically determined P2X7 dilated pore formation regulates
variability in chronic pain sensitivity (34). The study revealed
that in women having undergone a mastectomy there is a genetic
association between lower pain intensity and the hypofunctional
His270 (rs7958311) allele of P2RX7 gene, suggesting that se-
lectively targeting P2X pore formation may be a new strategy for
individualizing the treatment of chronic pain. The concept of
pore dilation of P2X receptors has remained a matter of con-
troversy for almost 20 years, and as such, understanding of this
molecular mechanism with new approaches represents a chal-
lenging and pertinent issue.
further reveal by molecular dynamics (MD) simulations a complex
mechanism for NMDG+ permeation, which involves both a con-
formational and orientational selection of the permeating mole-
cules. Finally, we uncover that desensitizing P2X receptors have the
capacity to transiently conduct NMDG+ ions in response to ATP
binding and identify spermidine as a natural large molecule able to
permeate selected human P2X pores. Our results underscore a
previously unappreciated P2X signaling.
Results
ATP Gates a Pore That Is Immediately Permeable to NMDG+. In a
recent study carried out in P2X2 receptors, it has been shown that
ATP rapidly gates macroscopic currents in symmetric NMDG+
solutions (32). To determine whether this occurs at the level of the
single channel, we used the outside-out configuration of the patch-
clamp technique that enables fast solution exchange at the cell
membrane on the millisecond time scale. We used symmetric
NMDG+ solutions, in which NMDG+ is present in both the ex-
ternal (outside) and internal (inside the pipette) solutions
(NMDG+out/NMDG+in) and thus represents the only carrier of
cationic currents. To increase seal resistance and patch stability,
fluoride ions (F−) were used as internal counterions to NMDG+
cations (Materials and Methods). We transiently transfected hu-
man embryonic kidney (HEK-293) cells with plasmids encoding
the rat P2X2–3T, which is a cysteine-less mutated receptor that
In this paper, we discover the molecular mechanism of per-
meation of large organic cations in P2X receptors. Contrary to the
prevailing assumptions, we provide evidence by single-channel
recordings that the pore does not undergo a time-dependent di-
lation upon opening but rather enters rapidly, in response to ATP retains wild-type P2X2 functionality (35) but displays increased
binding, a state that allows the stable passage of large organic single-channel conductance (36). In a first series of experiments,
cations. In addition, we designed and used photo-switchable cross-
we used excised outside-out patches that contained multiple
linkers of different lengths to probe conformational changes of channels and observed robust inward NMDG+ currents that
engineered cysteine-substituted P2X2 receptors that are associated developed rapidly following fast perfusion of 3 μM ATP, while
with NMDG+ and YO-PRO-1 permeability. We demonstrate that holding the voltage to –120 mV (activation time constant
the molecular motions leading to the permeation of these large
τ
NMDG+ = 203 47 ms, n = 9 patches; Fig. 1A, Left and Fig. S1 A
organic cations are very similar to those that drive Na+ flow. We
and B). NMDG+ currents remained constant during the 6-s time
NMDG+out/NMDG+
Na+out/NMDG+
in
A
B
in
ATP
ATP
ATP
5 pA
2 s
50 pA
2 s
100
1000
Time (ms)
5000
NMDG+out/NMDG+
Na+out/NMDG+
6
in
in
C
D
f
c (1 kHz)
c
c
Na+
fc (1 kHz)
ATP
ATP
43 pS
o1
0
5 pA
-10
0
5 pA
5 pA
500 ms
0.3
fc (100 Hz)
NMDG+
fc (100 Hz)
4.0 pS
o2
o1
c
o21
c
o21
o
o
0
-1.5
0
0.5 pA
Current (pA)
Fig. 1. Rapid ATP activation of P2X2–3T receptors induces instantaneous and stable permeation of NMDG+. (A) Fast application of ATP (3 μM, blue traces) to
a multiple channel-containing outside-out patch from HEK-293 cells expressing the P2X2–3T receptor evokes rapid NMDG+ (Left) and Na+ (Right) currents
recorded at –120 mV. The patch was first bathed in extracellular NMDG+ solution (NMDG+out/NMDG+in, Left) and then rapidly exchanged to extracellular Na+
solution (Na+out/NMDG+in, Right). Note the difference in current scale. (B) Superposition of normalized NMDG+ (black trace) and Na+ (gray trace) currents
shown in A in logarithmic timescale. (C) Single-channel recordings from an outside-out patch recorded at –120 mV of currents elicited by 1 μM ATP, first in
NMDG+out/NMDG+in solution (Left), and then in Na+out/NMDG+in solution (Right). Data were sampled at 10 kHz and filtered at a final corner frequency (fc) of
1 kHz (Top traces) or 100 Hz (Bottom traces). Channel openings (o) are downward deflections indicated by the red dashed lines. Baseline currents, which
correspond to closed channels (c), are indicated by the black dashed lines. (D) All-points amplitude histograms of single-channel Na+ (Top, fc = 1 kHz) and
NMDG+ currents (Bottom, fc = 100 Hz) obtained from the patch shown in C. Distributions were fit by a sum of four Gaussians for Na+ currents and three
Gaussians for NMDG+ currents (blue lines). Gaussians centered at 0 pA represent closed channels. For this patch, the mean unitary conductance of Na+
currents in Na+out/NMDG+ is 43 pS, whereas that of NMDG+ currents in NMDG+out/NMDG+ is 4.0 pS.
in
in
2 of 10
|
www.pnas.org/cgi/doi/10.1073/pnas.1701379114
Harkat et al.

application. Increasing ATP concentration consistently decreased probe the gating mechanism of P2X2 receptors by using light to
activation time constants (τNMDG+ = 36 4 ms, n = 10 patches at
open and close the pore (36). This strategy entails the use of a
10 μM; Fig. S1B). Consistent with a recent study (32), these data synthetic semirigid azobenzene cross-linker, 4,4′-bis(maleimido-
show that ATP rapidly gates an NMDG+ conductance, which glycine) azobenzene, called MAM (named hereafter MAM-3),
remains stable over time.
which is covalently tethered between a pair of engineered
cysteine-substituted residues located at an appropriate distance
apart. Light stimulation at specific wavelengths can then be used
to force parts of the protein to come closer together or move
farther apart due to isomerization of the azobenzene between
defined trans and cis configurations (Fig. 2 A and B). These
light-driven motions provide a faithful readout of the molecular
movements induced by ATP (36, 37).
We next compared these NMDG+ currents to those carried by
Na+ at the same potential, by rapidly exchanging (less than 1 s)
external NMDG+ for Na+ (i.e., Na+out/NMDG+in), and then
challenging again the same patch with 3 μM ATP. Robust inward
Na+ currents were recorded that were larger than NMDG+
currents (INMDG+/INa+ = 9.7
1.7%, n = 9 patches; Fig. 1A,
Right). This enhancement is expected, as the electrochemical
driving force is in favor of Na+ influx in Na+out/NMDG+ so-
We focused on the region of the pore that is particularly ef-
fective for manipulating molecular motions by light—that is,
around residues I328 and S345, which can be cross-linked hori-
zontally or vertically relative to the membrane plane by MAM-3
(36) (Fig. 2C). To increase the chance of trapping any in-
cremental change of the pore diameter, we synthesized two
shorter, more rigid photo-switchable cross-linkers, named MAM-1
(also known as BMA) (37) and MAM-2, in which the cysteine-
reactive maleimides are either directly attached to both ex-
tremities of the azobenzene core (MAM-1) or indirectly at one
of the two extremities by a glycine unit (MAM-2) (Fig. 2A and
Figs. S2–S4). As a result, the two end-to-end maleimides are
incrementally separated by 2 Å between any two consecutive
MAMs in their trans configuration (Fig. S5). The end-to-end
distance distributions in the cis configuration were significantly
broader and more difficult to interpret. We produced 13 single
and double cysteine mutants in the P2X2–3T background that
were expressed in HEK-293 cells and individually treated with
each photo-switchable cross-linker, giving rise to 42 different
combinations, including controls on the P2X2–3T (Fig. 2E).
Effects of light on receptor activity were then assayed using
whole-cell patch-clamp recordings on 365-nm (80 ms or 2 s,
17.7 mW/mm2), 530-nm (2 or 4 s, 13.3 mW/mm2), or 455-nm
(2 s, 39.5 mW/mm2) illumination cycles to elicit photocurrents
in symmetric NMDG+ solutions. These currents were normal-
ized to those carried by Na+ (INMDG+/INa+), following the rapid
in
lution. Comparison of normalized ATP-gated Na+ currents to
ATP-gated NMDG+ currents revealed similar onsets of inward
currents (Fig. 1B), quantified by similar activation rates (τNa+
=
235 74 ms at 3 μM ATP, n = 9; Fig. S1 A and B). Increasing
ATP concentration also consistently decreased activation time
constants (τNa+ = 27 4 ms at 10 μM, n = 10 patches; Fig. S1B),
with no substantial change of the ratio INMDG+/INa+ (11.4
0.6%, n = 9). These results indicate that the ATP-gated
P2X2 open pore is simultaneously permeable to both Na+ and
NMDG+. The similarity of the activation rates determined in
both solutions at a given ATP concentration suggests similar
gating kinetics and that the difference in the current ratio must
be due to different rates of permeation of these ions.
To determine NMDG+ permeation rates, we measured uni-
tary conductance of NMDG+ current from outside-out patches
that contained single channels, using the same protocols as
described for patches that contained multiple channels. In
NMDG+out/NMDG+in, barely visible single openings and closings
were detected following 1 μM ATP application, whereas robust
single-channel currents were recorded in Na+out/NMDG+ (Fig.
in
1C, Top) with a mean conductance of 44 8 pS (n = 7 patches,
data filtered at 1 kHz). However, when the same recordings were
further filtered at a much lower bandwidth (100 Hz; Materials and
Methods), discernable unitary currents were then resolved in
symmetric NMDG+ solution (Fig. 1C, Bottom). These small uni-
tary currents were always recorded in patches that responded to
exchange of cells to an Na+out/NMDG+ solution.
in
ATP in Na+ (16 out of 24 examined patches; the remaining
The screening revealed four important findings. First, all
mutants responded to ATP and were permeable to NMDG+
when ATP was used as an agonist, with INMDG+/INa+ ratios that
were similar to those recorded in outside-out patches (between
4% and 15%; Fig. 2E, Fig. S6A, and Table S1). In control ex-
periments, no light-gated currents were observed with P2X2–3T
incubated with any MAM (Fig. 2E and Fig. S6B). Second, only a
few of the cysteine mutants (7 combinations out of 39) showed
reliable light-induced NMDG+ currents, with a INMDG+/INa+
ratio > 5 (mean 9 3%) that was close to that of ATP controls
(Fig. 2E). Indeed, most of the cysteine mutants responded to
light following incubation with MAMs (27/39), but control ex-
periments revealed that many of the light-gated currents origi-
nating from double mutants (12/19) had activation profiles that
were similar to those of one of their single-mutant counterparts
(asterisk-labeled boxes in Fig. 2E). In addition, two other com-
binations—I332C/F346C incubated with MAM-2 or MAM-3—
which displayed the highest INMDG+/INa+ ratios, were discarded
because MAM treatment appeared to dramatically reduce light-
gated Na+ currents, which, in turn, may introduce uncertainty
regarding ratio measurements (Fig. 2 D and E).
out
8 patches were unresponsive to ATP and no unitary NMDG+
currents were detected), and they were not observed in the ab-
sence of ATP (Fig. S1D). All-points histogram analysis revealed
that the mean conductance of these unitary currents was 3.3
0.6 pS (mean amplitude of 0.40 0.04 pA, n = 6 patches), which
represented 7.5% of the unitary Na+ currents, a value that was
close to that of the ratio INMDG+/INa+ determined from multi-
channel currents (Fig. 1D and Fig. S1C). Increasing ATP con-
centration to 10 or 30 μM did not change the ratio of the mean
conductance of NMDG+ relative to Na+, suggesting that a near
saturating concentration of ATP does not increase NMDG+
conductance (Fig. S1C). Compared with recordings carried out in
symmetric Na+ solution (43 6 pS, n = 4 patches at 1 μM ATP;
Fig. S1E), the unitary conductance of NMDG+ currents was about
13 times lower than that of Na+ currents. From these values, we
conclude that NMDG+ ions flow through the ATP-gated open
pore at extremely low rates (∼2.5 × 106 NMDG+ ions per second
per channel at –120 mV with 132.6 mM NMDG+) compared with
Na+ (∼32 × 106 Na+ ions per second per channel at –120 mV with
132.6 mM Na+). Given the fact that NMDG+ can rapidly transit
through the ATP-gated receptor channel, our data demonstrate
that the open pore is wide enough to allow its passage on the
millisecond time scale.
Third, two phenotypes were observed in light-induced NMDG+
currents: Horizontally cross-linked single mutants (I328C treated
with MAM-2 or MAM-3, and I332C treated with MAM-1) were
activated in the trans configuration of the azobenzene, leading to
an NMDG+ permeability, whereas vertically cross-linked double
mutants (I328C/S345C treated with any MAMs, and I332C/S345C
treated with MAM-2) were activated in the cis configuration
(Fig. 2 D and E). Conversely, backward isomerization of the
Molecular Motions Underlying NMDG+ Permeation Are Similar to
Those Underlying Na+ Permeation. We next sought to determine
the molecular motions that drive permeation of NMDG+. We
used our recent “opto-tweezers” strategy, which enabled us to
Harkat et al.
PNAS Early Edition
|
3 of 10

NMDG+out/NMDG+
Na+out/NMDG+
in
O
A
D
in
cis
N
trans
O
N
N
O
I328C + MAM-1
N
trans MAM-1
10 pA
2 s
100 pA
200 pA
O
O
trans
N
I328C + MAM-2
O
O
N
N
O
20 pA
100 pA
50 pA
N
N
H
O
trans MAM-2
O
H
N
I328C + MAM-3
N
O
1000 pA
N
N
O
O
O
N
cis trans
N
H
O
trans MAM-3
I328C/S345C + MAM-2
500 pA
40 pA
R'
B
N
N
365 nm
N
N
I332C/F346C + MAM-2
P2X2-3T
R
R
R'
trans
530 nm
cis
40 pA
or 455 nm
C
E
31
200
I328C
P329C
I332C
S345C
F346C
L347C
P329
I328
I332
P329
Out
5
10
0
I328
I332
0
TM2 (A)
TM2 (B)
I328C/S345C
I328C/F346C
I328C/L347C
P329C/S345C
I332C/S345C
I332C/F346C
I332C/L347C
*
*
*
*
*
*
*
*
*
*
F346
S345
TM2 (C)
L347
In
*
*
Fig. 2. NMDG+ permeation operates with similar molecular motions to Na+ permeation. (A) Chemical structures of MAMs in the trans state. (B) Cis–trans
isomerization of the azobenzene induced by the indicated wavelengths of light. The backward cis–trans isomerization is induced by irradiation at 455 nm for
MAM-1 or 530 nm for MAM-2 and MAM-3. (C) Side view of the three transmembrane TM2 helices of the hP2X3 ATP-bound X-ray structure [Protein Data Bank
(PDB) ID code 5SVK] (10) shown in ribbons and color-coded by subunit. Residues selected for cysteine substitution are indicated by spheres on monomers A
and B, and for clarity, they were converted to equivalent rP2X2 numbering. Double red arrows indicate possible horizontal and vertical MAMs cross-linking,
relative to the membrane plane, between engineered cysteines. (D) Macroscopic light-gated currents recorded at –60 mV from HEK-293 cells expressing the
indicated mutants treated with the indicated MAMs. Cells were first bathed in extracellular NMDG+ solution (NMDG+out/NMDG+in, Left), which was then
rapidly exchanged for extracellular Na+ solution (Na+out/NMDG+in, Right). Illumination was carried out, as indicated, at 365 nm (violet bars) and 455 nm (blue
bars) or 530 nm (green bars) to elicit the cis and trans states of the azobenzene, respectively. Short illuminations (80 ms) are indicated by violet arrows. For
MAM-treated single mutants, cells were briefly preirradiated at 365 nm (80 ms) before recordings. (E) Heat map of NMDG+ currents normalized to Na+
currents (Left) and of Na+ currents density (Right) recorded in indicated conditions. All light-gated currents recorded for single mutants were induced by the
trans configuration, whereas all light-gated currents recorded for double mutants were elicited by the cis configuration, except those indicated by asterisks,
which responded to the trans configuration, similarly to their single mutant counterparts I328C, P329C, or I332C. Gray boxes indicate that INMDG+/INa+ cannot
be determined. In controls, 30 μM ATP, applied on untreated cells, elicited responses that range between 32% and 92% of the maximal response (see Table
S2) (for each box, n = 4–7 cells).
azobenzene from trans to cis, in the case of horizontally cross-
The reason for this discrepancy is unknown, but given that
linked mutants, or cis to trans configuration, for vertically cross- NMDG+ permeability had been obtained by measuring Erev from
linked mutants, closed the pore, and an additional illumination
cycle revealed full reversibility of the light-gated NMDG+ currents.
Supported by intersubunit cross-linking (Fig. S6C), these data
demonstrate that the opening of the NMDG+-permeable pore
involves two molecular motions: (i) a specific vertical shortening of
the distance between extracellular and intracellular ends of adja-
cent TM2 helices, and (ii) an outward separation of the extracel-
lular ends from two adjacent TM2 helices.
bi-ionic experiments (36), it remains possible that its permeability
was underestimated, further stressing the need to use symmetric
solutions to measure direct permeation events.
Occurrence of Partially Open States. The correlation between light-
gated NMDG+ and Na+ currents was, however, not perfect. The
results revealed that in 3 out of 11 combinations that displayed size-
able light-gated Na+ currents (current density > 3 pA/pF), no or very
small light-gated NMDG+ currents were recorded (INMDG+/INa+ were
equal or close to zero; Fig. 2E). These data were observed in both
vertical (I332C/S345C treated with MAM-3) and horizontal
(I328C and P329C treated with MAM-1) cross-linking. Of note,
increasing the length of the “tweezers” in the horizontal I328C
cross-linking from MAM-1 to MAM-2 to MAM-3 increased the
ratio INMDG+/INa+ from 0% to ∼7% (Fig. 2D). Likewise, de-
creasing the length of the tweezers in the vertical I332C/S345C
Fourth, a clear correlation between light-gated NMDG+ and Na+
currents was observed, whereby Na+ currents were always observed
in the case of NMDG+ currents, and inversely, no light-gated
NMDG+ currents were recorded when no light-gated Na+ cur-
rents were recorded, indicating that the molecular mechanism un-
derlying NMDG+ permeation is very similar to that underlying Na+
permeation.
Unlike our previous work (36), we found substantial NMDG+
permeability in mutant I328C cross-linked horizontally with MAM-3. cross-linking from MAM-3 to MAM-2 also increased the ratio
4 of 10
|
www.pnas.org/cgi/doi/10.1073/pnas.1701379114
Harkat et al.

INMDG+/INa+ from ∼0.2% to ∼6%. Although we cannot rule out
the hypothesis that the presence of MAM is specifically interfering
with the bulkier NMDG+, these data suggest the existence of
partially, Na+-selective open states that are not permeable
to NMDG+.
ATP
+100 s
A
-200 s
-100 s
+200 s
+300 s
365 nm
Light-Induced Motions Gate Dye Uptake in Physiological Conditions.
To further confirm data obtained by patch-clamp electrophysi-
ology, we measured YO-PRO-1 dye uptake in physiological
conditions, where Na+ replaced NMDG+. We focused on the
pair I328C and S345C treated with MAM-2, which gave both the
highest INMDG+/INa+ ratio and the most robust light-gated Na+
current density. Robust dye uptake following a pulse of 5 s of
light at 365 nm was observed for the double mutant I328C/S345C
(Fig. 3 A and C). These uptakes were light-dependent and were
similar to those induced by 30 μM ATP on the P2X2–3T back-
ground (Fig. 3 A and B). In contrast, no dye uptake was observed
in cells expressing the single mutant I328C or S345C incubated
with MAM-2 that were illuminated at 530 nm, following a brief
preirradiation at 365 nm before YO-PRO-1 application to reset
the azobenzene in the cis state (Fig. 3 D and E). Although these
fluorescence data were in agreement with patch-clamp data, the
lack of YO-PRO-1 intake for I328C seemed to be in contrast to
the recorded NMDG+ permeability. However, a careful analysis
showed a rapid inactivation of light-gated currents during and
after irradiation of cells expressing the I328C mutant treated
with MAM-2, whereas stable light-gated currents were recorded
for the double mutant I328C/S345C incubated with the same
photo–cross-linker (Fig. S7). The transient light-gated activation
of the single mutant likely prevents substantial accumulation of
YO-PRO-1 into cells, a hypothesis that readily explains the ap-
parent lack of dye uptake. All together, these results provide
evidence that YO-PRO-1 uptake induced by the cis configura-
tion of MAM-2 tethered to the I328C/S345C double mutant
occurs in physiological conditions.
0
1
Fluorescence (A.U.)
I328C/S345C
B
D
C0.4
P2X2-3T
0.6
0.0
+hν
+ATP
365 nm
0
30 μM
-hν
-ATP
300
0.0
-200
300
-200
0
Time (s)
Time (s)
E 0.4
S345C
0.4
I328C
530 nm
530 nm
0
0.0
0.0
-200
300
-200
0
300
Time (s)
Time (s)
Fig. 3. Structural changes of TM2 helices induce cellular dye uptake in
physiological conditions. (A) Selected time series of YO-PRO-1 dye uptake in
HEK-293 cells expressing the P2X2–3T receptor (Upper) and double mutant
I328C/S345C treated with MAM-2 (Lower). Fluorescence (in arbitrary units)
was acquired before and during ATP (30 μM) or light (365 nm for 5 s) acti-
vation that started at 0 s as indicated. On the left are shown the corre-
sponding microphotographs under transmitted light. (Scale bar, 20 μm.) (B)
Corresponding rate of YO-PRO-1 dye uptake in cells (n = 15 cells) expressing
the P2X2–3T receptor in the absence (orange) or presence (red) of ATP. The
arrow indicates the time at which ATP was applied. (C) Corresponding rate
of dye uptake in cells expressing the double mutant I328C/S345C treated
with MAM-2 in the absence (light blue) or presence (dark blue) of irradiation
at 365 nm (n = 18 cells). The arrow indicates the time at which cells were
briefly irradiated. (D) Same protocol as in C for cells expressing the single
mutant I328C treated with MAM-2, except that irradiation occurred at
530 nm (n = 12 cells). (E) Same protocol as in D for cells expressing the single
mutant S345C treated with MAM-2 (n = 18–21 cells). For D and E, cells were
briefly preirradiated at 365 nm for 2 s just before YO-PRO-1 application.
Shaded areas denote mean SEM.
MD Simulations of NMDG+ Permeation. To obtain insights into the
permeation mechanism of large organic cations in P2X recep-
tors, NMDG+ conductance was explored by all-atom MD. For
this purpose, an atomistic model of the open state of zfP2X4
equilibrated with three MAM-2 vertically cross-linked between
I336C and N353C (i.e., equivalent to I328C and S345C in
rP2X2) was simulated in the presence of a membrane potential
generated by a constant electric field (Materials and Methods).
To capture NMDG+ permeation events on the simulation time
scale (i.e., <50 ns), the ion channel devoid of the extracellular
domain was simulated in the presence of a large membrane
potential (up to –2 V) in the absence of MAM and with har-
monic restraints on the backbone atoms to preserve its open-
pore conformation. The MD results show that our open-state
model of zfP2X4 is permeable to NMDG+ (Fig. 4 and Movie
S1) with a permeation rate of 7 × 107 s−1, which is approximately
one order of magnitude lower than that simulated in the pres-
ence of Na+ (Table S1). Of note, these results closely match the
ratio measured from single-channel recordings, although the
absolute values obtained from modeling were largely higher than
those determined experimentally. This is likely due to the non-
physiological value of the electrochemical driving force that was
used in MD. These simulations reveal that the difference in
permeability observed for Na+ versus NMDG+ is due to a more
complex permeation mechanism for the latter, which involves a
selection for permeation based on both the molecular confor-
mation and the orientation of the organic cation relative to the
pore axis. In fact, by monitoring the end-to-end distance and the
orientation of NMDG+ along a series of successful permeation
events sampled by MD, we found that to be able to cross the
constriction point, the organic cation must adopt a fully extended
conformation (d > 7.5 Å) and have the charged nitrogen atom
facing downward along the electrochemical gradient (θ < −50°)
(Fig. 4). In addition, the simulations indicated that before pop-
ulating a permeable conformation, the flexible NMDG+ needs to
sample several conformations and orientations, which signifi-
cantly hinders its permeability, consistent with low unitary con-
ductance of single-channel NMDG+ currents. Hence, the simulation
results confirm that the open-channel state elicited by MAM is
permeable to NMDG+ and provide an atomistic picture of the
permeation mechanism.
Permeation of YO-PRO-1 was also investigated by MD;
however, in sharp contrast to results obtained for NMDG+,
no permeation event was sampled under similar simulation
conditions (see Table S1).
Harkat et al.
PNAS Early Edition
|
5 of 10

ATP in cells expressing the human P2X2 and P2X3 receptors while
the membrane was held at –60 mV (Fig. 5B). Of note, MD sim-
ulations show that the ATP-bound, zfP2X4 open state is also
permeable to spermidine (Movie S2). In addition, the simulations
indicate that its permeation rate (5 × 108 s−1) is sevenfold faster
than that of NMDG+ under the same membrane potential. The
symmetrical structure of spermidine in addition to the presence of
positive charges at the extremities make both the conformational
and orientational barriers for permeation almost vanish, thus
drastically enhancing its permeability relative to NMDG+ despite
their similar size. These data therefore uncover a previously un-
appreciated P2X signaling in which large molecules can rapidly
permeate through the pore in response to ATP binding.
1
2
3
4
A
B
Out
In
20
10
Out
Discussion
2
2
1
1
In this article we uncover the molecular mechanism underlying the
permeability of P2X receptors to large organic cations and identify
an important natural ion channel modulator able to flow through
the ATP-gated pore. Our data tackle the pore dilation paradigm,
which has remained for nearly 20 years one of the most enigmatic
features of ionotropic purinergic receptors (19, 20). Until recently,
the dogma concerning pore dilation was that the channel un-
dergoes a progressive expansion, over time reaching a diameter
wide enough to allow permeation of large cations, such as NMDG+
and YO-PRO-1. However, a recent study challenged this model by
elegantly demonstrating that P2X channels rapidly develop an
0
3
3
-10
-20
-30
4
4
In
7
6
7
8
9
-90 -45
0
45 90 3 5
End-to-end distance (Å)
Horizontality (°)
Pore radius (Å)
Fig. 4. Mechanism of NMDG+ permeation revealed by MD simulations.
(A) Snapshots at different simulation times. (B) End-to-end distance of NMDG+
(Left) and angle formed by the longitudinal axis of NMDG+ cation with the axis
parallel to the membrane plane (Middle) are displayed per frame of the sim-
ulation for all permeant molecules. Data were collected from four different
NMDG+ concentration and membrane potential simulation setups: 1 M at –1 V
(red dots), 0.15 M at –1 V (green), 0.15 M at –1.5 V (blue), and 0.15 M at –2 V
(cyan). Molecular snapshots of NMDG+ conformations and orientations cor-
responding to extreme values for these two observables are shown on top.
Indicated numbered dots in yellow are snapshots from A. HOLE profile of the
zfP2X4 open-state model after 50 ns of equilibration (Right). The simulation
results indicate that the ion pore of the ATP-bound state is sufficiently wide to
allow for the passage of large organic cations.
rP2X3
A
B
ATP
hP2X2
NMDG+out/NMDG+
ATP
Na+out/NMDG+
in
in
600 pA
2 s
0.1 nA
1 s
1 nA
hP2X3
rP2X4
Selected Homomeric P2X Receptors Conduct Large Natural Molecules
in Response to ATP. Our data reveal that the permeability to large
cations develops within milliseconds upon ATP application, with
the same time course as permeability to small cations. We thus
asked whether this process is also true for other P2X members,
especially for the fast-desensitizing P2X1 and P2X3 receptors,
which were presumed to be “nondilating.” To this end, we
recorded ATP-gated currents in HEK cells transfected with rat
P2X1, P2X3, P2X4, P2X5, P2X7, and human P2X2 and
P2X3 receptors in symmetric NMDG+ solutions (Fig. 5A and
Fig. S8). We observed fast and robust NMDG+ currents for all
these P2X receptors, expect for P2X1, which for unknown reasons
was not functional in symmetric NMDG+ solution (no ATP-gated
current was observed in the control solution Na+out/NMDG+in).
Importantly, we provide evidence that NMDG+ can permeate the
rat and human P2X3 receptors, demonstrating that these desen-
sitizing P2X receptors do indeed carry the ability to rapidly enter a
state that is wide enough to allow the passage of large molecules.
Finally, having shown that rapid activation of P2X receptors
allows permeation of NMDG+, we sought to determine which
natural compounds can transit through the pore. We focused on
positively charged compounds that share a similar size to NMDG+.
We selected spermidine, which is a natural polyamine known to
modulate many ion channels (38), and produced symmetric sper-
midine solutions (Materials and Methods). We observed robust in-
ward spermidine whole-cell currents following rapid application of
10 pA
30 pA
2 s
100 pA
1 s
rP2X7
100 pA
2 s
500 pA
Fig. 5. P2X receptors are permeable to natural organic cations. (A) Mac-
roscopic ATP-gated currents recorded at –60 mV from HEK-293 cells expressing
the indicated rat P2X receptors. Recordings were first made in symmetric
NMDG+out/NMDG+in solution (black traces) and then in Na+out/NMDG+in (gray
traces) solution. ATP concentration was 30 μM for rP2X3 and rP2X4 and
300 μM for rP2X7 receptors. For rP2X3 receptors, ATP applications were spaced
at least by 3 min. (B) Macroscopic ATP-gated currents recorded in symmetric
spermidine solution at –60 mV from HEK-293 cells expressing the indicated
human P2X receptors. ATP concentration was 30 μM. Spermidine current
density was as follows: hP2X2, 43 4 pA/pF (n = 5 cells); hP2X3, 6 2 pA/pF
(n = 7 cells). For all traces, ATP application lasted 5 s.
6 of 10
|
www.pnas.org/cgi/doi/10.1073/pnas.1701379114
Harkat et al.

NMDG+ current upon ATP binding and lack the slow phase of
pore dilation (32). Our results are consistent with this work, and
taken together, these two studies firmly establish that, in striking
contrast to earlier beliefs, the open state that is reached in milli-
seconds following ATP binding is also permeable to large mole-
cules. This important finding raises considerable questions
concerning the previously predominant view that suggested that the
initial I1 open state is only selective to small cations, such as Na+,
K+, and Ca2+. The immediate implication of our results is that the
structure of the open pore of the ATP-bound state must be suffi-
ciently wide to accommodate large molecules. From the X-ray
structures of the ATP-bound zfP2X4 receptor (8) and very re-
cently the ATP-bound human P2X3 (hP2X3) receptor (10), this
seems to be the case, as the minimal cross-section of an extended
NMDG+ (6 Å × 6 Å × 12.5 Å) is less than the diameter of the open
pore (∼7 Å for zfP2X4 and 6.4 Å for hP2X3). However, reliable
NMDG+ permeation by MD simulations cannot be made on the
zfP2X4 X-ray structure because the pore systematically collapses
within a few nanoseconds (<5) of equilibration (36), likely due to
the lack of the intracellular domain that was removed for crystal-
lization purposes (8). By using an improved model of the open
likely through a change of the helical structure. Importantly, such
similar molecular motions were also identified during channel
gating (36), suggesting that ATP binding drives the rapid open-
ing of a pore that is simultaneously selective to both small and
large cations. As we do not have evidence for a time-dependent
increase of the permeability of large cations, our data thus
support the conclusion that pore dilation is not an intrinsic
property of the channel itself. However, we do not rule out the
possibility that pore dilation does exist, but if this were the case,
it must require a regulatory element that is external to the P2X
pore and that would be lost in our experimental conditions.
One intriguing feature of the mechanism revealed by the opto-
tweezers approach is that it raises the possibility to open the
channel in a state that is not fully open. We found that an in-
sufficient horizontal outward separation or inappropriate vertical
motions of two adjacent TM2 helices failed to open NMDG+
conductance but not Na+ conductance, suggesting that the pore
is partially open. A further variation, by only a few Å, of adjacent
helices induced an NMDG+ permeation in addition to the Na+
flow, thus allowing the pore to become fully opened. The phys-
iological relevance of these partially open states remains unclear,
state of zfP2X4 (36), which was equilibrated by 50-ns MD in a but they might be related to dynamic changes of other bio-
physiological environment and stabilized by three vertically cross- physical properties of P2X receptors, such as ATP potency and
linked MAM-2 molecules, we provide clear evidence that both
NMDG+ and the natural cation spermidine may flow through the
ion pore of the ATP-bound state. We found, however, that the flow
rectification, as reported previously (40, 41). Another possibility
would be that these partially open states might correspond to the
actual I1 state. However, we do not favor this hypothesis because
of NMDG+ was significantly lower than that of Na+, as the former we provide no evidence that ATP binding naturally drives the
needs to “snake” through the permeating pathway in a fully linear
conformation, with the positively charged nitrogen head group
pointing downward along the electrochemical gradient. The con-
formational confinement to this extended form, along with specific
orientational constraints in the narrowest region of the pore, in-
troduces significant barriers that hinder cation permeability, thus
revealing a complex mechanism for NMDG+ permeation. These
molecular requirements therefore cause a decrease in the overall
rate of NMDG+ flow. This conclusion is fully consistent with the
low unitary conductance of single-channel NMDG+ currents that
we (present study) and others (16, 33, 39) have measured for
P2X2 and P2X7 receptors.
opening of the channel in partially open states that are only
selective to small cations.
Finally, we have identified spermidine as a natural cation able
to permeate through the ATP-gated open state. Polyamines are
well-known to modulate the activity of many ion channels, in-
cluding synaptic ligand-gated ion channels (38). Importantly, we
show that even desensitizing receptors, such as hP2X3, which
have been considered thus far as nondilating pores, are able to
briefly activate their pores in an open state, allowing for a tran-
sient flow of spermidine. This finding has considerable physio-
logical significance because it discloses an unsuspected role of
polyamines in P2X signaling and more generally because it raises
the possibility that activation of P2X receptors may allow for the
exchange of other physiological molecules between cells, such
as amino acids. Notably, a recent study exploited the large-pore
property of P2X receptors to deliver small membrane-impermeable
drugs to diseased retina cells (42). We thus propose that besides
the critical role of the permeation of inorganic cations, the passage
of small-sized metabolites, like spermidine, through the ATP-
gated open P2X pore may contribute to alternative physiological
responses. These findings open up new horizons in P2X signaling.
In contrast to NMDG+, we were unable to sample permeation
events for YO-PRO-1 in MD simulations. Because the minimal
cross-section of the dye (7 Å × 8 Å × 19 Å) reaches the outer
limits of the open pore (∼7 Å), it is possible that YO-PRO-1
needs to sample more conformations and orientations before a
successful permeation can occur, a process that would take
considerably more time. As a result, the flux of this fluorescent
dye would be extremely low, likely below that of NMDG+. Such a
molecular sampling-limited step may help to explain the appar-
ent delayed entry of fluorescent dyes that were typically used for
monitoring dye uptake through P2X receptors.
Materials and Methods
Chemical Synthesis. All chemicals were purchased from Sigma-Aldrich, Fluka,
Across, or Alfa Aesar in analytical grade. An Agilent LC–MS RRLC 1200SL/ESI
QTof 6520 was used for ESI analysis. 1H NMR and 13C NMR were run at
400 and 100 MHz, respectively, on an AvanceIII 400 NMR spectrometer from
Brucker. Coupling constants (J) are quoted in Hz and chemical shifts (δ) are
given in parts per million (ppm) using the residue solvent peaks as reference
relative to tetramethylsilane (TMS).
For MAM-1 synthesis, 498 mg (2.348 mmol) of (E)-4,4’-(diazene-1,2-diyl)
dianiline and 462.3 mg (4.715 mmol, 2 eq) of maleic anhydride were mixed
in 20 mL of anhydrous THF and left 1 h at 4 °C. The resulting product was
centrifuged (4 min, 8,000 g), and the precipitate was resolubilized in 20 mL
of anhydrous THF. Then, 0.5 mL (5.324 mmol, 2.3 eq) of acetic anhydride and
50.2 mg of sodium acetate were added. The mixture was heated under micro-
wave conditions (20 min, 110 °C). Distilled water was added (100 mL), and after
centrifugation (10 min, 8,000 g), the precipitate was resolubilized in 100 mL of
methanol and an orange solid was obtained by slowly adding cold water. After
filtration, purification was carried out by Flash-column chromatography (silica)
with dichloromethane and ethyl acetate (gradient 100:0 ≥ 95:5). After evapora-
tion of the solvent, the pure MAM-1 was obtained as a bright orange solid (36%
yield): NMR 1H (CDCl3):δ (ppm), 8.01 (4H, d, J = 8.8 Hz), 7.55 (4H, d, J = 8.8 Hz), and
An important finding of this study is that the molecular mo-
tions driving NMDG+ conductance are very similar to those that
lead to Na+ flow. By using our recently reported opto-tweezers
approach (36), we tested the ability of three photo-switchable
cross-linkers of different lengths to optically control NMDG+
permeability of 13 cysteine mutants. We observed reliable
NMDG+ permeation with only 7 out of the 39 tested combina-
tions and found a clear correlation between light-gated NMDG+
currents and light-gated Na+ currents. The specificity of cross-
linked residues and the size dependence of the MAMs strengthen
the conclusion that the used tweezers do not induce disorder in
the protein but rather can be used as mechanical actuators to
justly monitor naturally relevant motions. We identified two
molecular motions that lead to permeation of large organic
cations: a horizontal outward separation of the extracellular ends
of TM2 helices and a vertical motion, in which the N and C
termini of TM2 helices from adjacent subunits come closer to-
gether or change their orientations relative to one another, most
Harkat et al.
PNAS Early Edition
|
7 of 10

6.88 (4H, s); NMR 13C (CDCl3):δ (ppm), 169.16, 151.23, 134.42, 133.72, 126.22, and
123.70; and (ESI-HMRS):(m/z, [M+H]+), calculated for C20H13N4O4+, 373.0859;
found, 373.0932.
pH 7.3 with NaOH. Normal internal solution (NIS) contained 140 mM KCl,
5 mM MgCl2, 5 mM EGTA, 10 mM Hepes, adjusted to pH 7.3 with NaOH. For
MAMs incubation, cells were incubated before recordings at room temper-
ature with, respectively, MAM-1 (10 μM for 5 min, 1% final concentration of
DMSO in NES), MAM-2 (10 μM for 10 min, 1% DMSO in NES), and MAM-3
(3 μM for 20 min, 1% DMSO in NES) and 3 μM ATP. After treatment, cells
were washed out with NES. Patch pipettes contained 140 mM NMDG, 10 mM
Hepes, and 10 mM EDTA, pH adjusted with HCl to 7.3. To measure NMDG+
permeation, extracellular solution contained 140 mM NMDG and 10 mM
Hepes, pH 7.3. The solution was then exchanged to NaCl solution containing
140 mM NaCl and 10 mM Hepes, pH 7.3. For control experiments, these
solutions were supplemented with 30 μM ATP. For spermidine permeation
experiments, whole-cell recordings were performed 24–48 h after trans-
fection. Patch pipettes contained 33 mM spermidine, 10 mM Hepes, and
10 mM EDTA, pH adjusted to 7.3 with HCl. The extracellular solution con-
tained 33 mM spermidine and 10 mM Hepes, pH 7.3, supplemented with
30 μM ATP. Osmolarity of all these solutions was adjusted as described
above. Current density was obtained by dividing the current by the cell
membrane capacitance.
For synthesis of intermediate 1, 398 mg (1.875 mmol) of (E)-4,4’-(diazene-
1,2-diyl)dianiline and 726.8 mg (4.7 mmol, 2.5 eq) of 2-(2,5-dioxo-2,
5-dihydro-1H-pyrrol-1-yl)acetic acid were solubilized in a mixture of an-
hydrous DMF/acetonitrile. We then added 1.108 g (4.7 mmol, 2.5 eq) of
HATU and 0.65 mL of anhydrous triethylamine (4.7 mmol, 2.5 eq). The
mixture was agitated at room temperature for 20 h. After extraction
(NaHCO3, 3× ethyl acetate), the crude product was washed with acetone.
The supernatant was purified by flash-column chromatography (silica)
with ethyl acetate and heptane (60:40). An orange product was obtained
(compound 1, 51% yield): NMR 1H (acetone-d6):δ (ppm), 9.66 (1H, s), 7.78
(2H, d, J = 9.3 Hz), 7.76 (2H, d, J = 9.3 Hz), 7.71 (2H, d, J = 8.9 Hz), 7.01 (2H,
s), 6.78 (2H, d, J = 8.9 Hz), and 4.41 (2H, s); NMR 13C (DMSO-d6):δ (ppm),
170.64, 165.11, 152.47, 139.70, 138.55, 136.20, 134.95, 128.15, 125.90,
123.50, 122.23, 118.86, 112.85, 112.52, 68.49, 55.81, 32.08, and 29.58.
For synthesis of MAM-2, 200 mg (0.5725 mmol) of intermediate 1 was
mixed with 112.3 mg (1.145 mmol, 2 eq) of maleic anhydride and heated
under microwave conditions (110 °C, 90 min) in acetone. The obtained
precipitate was filtered and resuspended in acetone and then heated 5 min
at 60 °C with 0.12 mL of triethylamine (0.8588 mmol, 1.5 eq). We then added
0.54 mL of acetic anhydride (5.725 mmol, 10 eq) with a catalytic amount of
manganese acetate (III), and the mixture was heated under microwave
conditions (90 min, 110 °C). After addition of water and filtration, 46.1 mg of
MAM-2 was obtained (19% yield): NMR 1H (DMSO-d6):δ (ppm), 10.66 (1H, s),
7.97 (2H, d, J = 8.8 Hz), 7.92 (2H, d, J = 8.8 Hz), 7.79 (2H, d, J = 8.8 Hz), 7.58
(2H, d, J = 8.8 Hz), 7.23 (2H, s), 7.16 (2H, s), and 4.34 (2H, s); NMR 13C (ace-
tone-d6):δ (ppm), 171.30, 170.38, 166.17, 152.11, 149.51, 142.76, 135.71,
135.60, 135.25, 127.78, 124.83, 123.80, 120.53, and 41.46; (ESI-HMRS):(m/z,
[M+H]+), 429.1073 calculated for C22H15N5O5+; found, 429.1069.
Cell-Surface Cross-Linking. Cross-linking of cell-surface receptors was per-
formed as follows. TSA-201 cells in dishes were transfected with pcDNA3.1(+)
vectors containing the mutant constructs. After 24 h or 48 h, cells in dishes
were washed with ice-cold PBS that contained 154 mM NaCl, 2.68 mM KCl,
4.2 mM Na2HPO4, 1.47 mM KH2PO4, pH 7.0, supplemented with 1 mM MgCl2
and 0.4 mM CaCl2. Then, cells were incubated under gentle agitation with
50 μM photo-switchable cross-linker in the presence of 3 μM ATP in ice-cold
PBS, for 20 min (MAM-3) or 15 min (MAM-1, MAM-2). Quenching of un-
reacted MAM solution was carried out by a 10-min incubation with 10 mM
N-acetyl-L-cysteine methyl ester (Sigma-Aldrich), in ice-cold PBS, pH 8.0.
Dishes were rapidly washed with PBS and incubated with a thiol-cleavable,
membrane-impermeant reagent sulfosuccinimidyl-2-(biotinamido)ethyl-1,
3-dithiopropionate (Sulfo-NHS-SS-Biotin; ThermoFisher Scientific) in PBS at
pH 8.0 for 30 min under gentle agitation. Unreacted Sulfo-NHS-SS-Biotin was
quenched by incubation with 20 mM Tris(hydroxymethyl)aminomethane
(Biosolve Chemicals) in ice-cold PBS, pH 8.0, for 10 min. Cells were solubilized
in lysis buffer, and the supernatant was incubated overnight with neutravidin-
agarose beads (ThermoFisher Scientific) as previously described (43). Protein
samples were separated on 4–15% SDS/PAGE gels in Tris/Glycine/SDS running
buffer (Bio-Rad). Samples were transferred to a nitrocellulose membrane as
described (43), which was then incubated in TPBS (PBS supplemented with
1% nonfat dry milk, 0.5% BSA, and 0.05% Tween 20) to block the mem-
brane. The membrane was incubated in TPBS buffer overnight at 4 °C with
mouse anti–c-Myc antibody (ThermoFisher Scientific) diluted at 1:2,500.
After three washes with TPBS, the blot was incubated with peroxidase-
The synthesis of MAM-3 was carried out as previously described (36).
Molecular Biology. Cysteine mutations were introduced into a rP2X2 receptor
background in which Cys9, Cys348, and Cys430 were mutated to threonine
(P2X2–3T) (35) using KAPA HiFi HotStart PCR kit (Cliniscience). All mutations
were confirmed by DNA sequencing (GATC-Biotech). All P2X encoding genes
were subcloned in pcDNA3.1 vector, except that encoding hP2X2, which was
subcloned in the vector pCMV6-AC-mGFP (OriGene). hP2X2 contains mGFP
at its C terminus.
Expression in Cultured Cells. HEK-293 and TSA-201 cells were cultured and
transiently transfected using phosphate calcium procedure with the
pcDNA3.1(+) vectors (0.05–0.1 μg for single channel recordings, 0.3 μg for
whole-cell recordings, and 10 μg for cell surface cross-linking) and a vector
encoding a green fluorescent protein (0.3 μg), as previously described (43).
conjugated sheep anti-mouse antibody for
2 h (dilution 1:10,000; GE
Healthcare life Sciences) at room temperature and washed a further three
times with TPBS and developed using Amersham ECL Prime Western blotting
detection reagent (Dominique Dutscher).
Patch-Clamp Electrophysiology. Single-channel recordings using outside-out
configuration were carried out using HEK-293 cells at room temperature
24 h after transfection. Recording pipettes pulled from borosilicate glass
(Harvard Apparatus) were coated with Sylgard 184 (Dow Corning Co.) and fire
polished to yield resistances of 10–20 MΩ (Sutter model p-97). The holding
potential was –120 mV. The extracellular solution contained 132.6 mM NaCl
or NMDG (Sigma), 0.3 mM CaCl2, 0.25 mM MgCl2, 10 mM Hepes, pH 7.3,
adjusted with NaOH (for NaCl solution) or HCl solution (for NMDG solution).
The intracellular solution contained either 132.6 mM NMDG, 9.46 mM
Hepes, and 10 mM EDTA, adjusted to pH 7.3, first approximately with a 5%
HF solution, then more precisely with 0.5% HF with Polypropylene (PP) pi-
pettes (Dominique Dutscher) or 132.6 mM NaF, 9.46 mM Hepes, and 10 mM
EDTA, adjusted to pH 7.3 with NaOH. Osmolarity was adjusted to 290–
310 mOsmol·kg−1 with glucose. Data were acquired with a patch-clamp
amplifier (HEKA EPC 10) using PATCHMASTER software (HEKA Co.), sam-
pled at 4–10 kHz, and low-pass filtered at 2.9 kHz. For offline analysis, data
were refiltered to give a cascaded filter corner frequency (fc) of either 1 kHz
or 100 Hz. For data analyses, FitMaster (HEKA Electroniks, v2 × 73.2) and
IGOR PRO (WaveMetrics, v6.37A) softwares were used. Channel events were
detected by using TAC software (Bruxton Co.), and conductance levels were
measured by all-points amplitude histograms fitted to Gaussian distributions.
Fitting procedures to access the time constant were based on the single-
exponential decay equation function, It = I0 + A exp(–t/τ), where It is the in-
stantaneous current; I0 and A are the residual current and maximal amplitude,
respectively; t is the time in seconds; and τ is the time constant in seconds.
Whole-cell recordings were performed 24–48 h after transfection in HEK-
293 cells. Normal external solution (NES) contained 140 mM NaCl, 2.8 mM
KCl, 2 mM CaCl2, 2 mM MgCl2, 10 mM glucose, 10 mM Hepes, adjusted to
Fluorescence Measurements. Fluorescence was measured using an Olympus
IX73 (Olympus LUCPlanFLN 20×/0.45 PH1) with ProgRes MF-cool camera. Im-
ages were captured at 0.5 Hz. For each experiment, YO-PRO-1 (ThermoFisher)
fluorescence was measured from three single cells per field with excitation at
455 nm (ET-EGFP filter, Chroma). For the double-mutant I328C/S345C, cells
were first incubated with 10 μM MAM-2 in the presence of 3 μM ATP for
10 min and washed with NES buffer. Then, cells were incubated with
10 μM YO-PRO-1 (4-[(3-methyl-1,3-benzoxazol-2(3H)-ylidene)methyl]-1-[3-
(trimethylammonio)propyl]quinolinium diiodide) in NES, and following
10 min of incubation, cells were irradiated at 365 nm for 5 s. For single-
mutant I328C and S345C, the same protocol was carried out, except that
activation was achieved by 530 nm irradiation and that cells were briefly
preirradiated at 365 nm before YO-PRO-1 incubation to reset the azobenzene in
the cis state. In control experiments with the P2X2–3T, the incorporation of
10 μM YO-PRO-1 was measured in response to 30 μM ATP.
Molecular Modeling. The end-to-end distances for the free MAM-1 and
MAM-2 molecules in solution (∼10,000 atoms with water molecules) were
obtained from all-atom MD simulations performed with ACEMD (44). Eight 100 ns-
long unrestrained MD simulations were computed in the NVT ensemble at 310 K
for the cis and trans configurations and for the R/R, S/R, R/S, and S/S stereoisomers.
The mean distances between the S–S atoms were computed by averaging over the
four stereoisomers for a total of n = 200,000 per cis or trans configuration. Nor-
malized probability distributions of the S–S distance were obtained by clustering all
distance values using a bin width of 0.5 Å. The permeation mechanism of Na+,
8 of 10
|
www.pnas.org/cgi/doi/10.1073/pnas.1701379114
Harkat et al.

NMDG+, YO-PRO-12+, and spermidine3+ in P2X was explored by all-atom MD
simulations starting from a relaxed open-state model of zfP2X4 stabilized by three
vertical MAM-2 photo-switchable cross-linkers fused at the positions I336C/N353C
in cis configuration. The latter model (∼138,000 atoms) was produced following
the procedure described previously (36), with a few differences: (i) Three MAM-2
(instead of MAM-3) cross-linkers were fused to zfP2X4 with the glycine unit placed
downward near N353C; (ii) the 2-ns equilibration MD was followed by a pro-
duction of 50 ns with no positional restraints; (iii) in addition to the distance re-
straints mimicking the internal TM1/TM2 Cd2+ binding site (45), the symmetry of
the P2X trimer was loosely controlled by using the “Symmetry Restraints” com-
mand in NAMD with force constants of 0.25 and 1.0 kcal·mol−1·Å−2 for the ex-
tracellular and the transmembrane domains, respectively; and (iv) the side chain of
L351 involved in the new TM2–TM2 interface was simulated using four non-
interacting copies (46) together with the R/R, R/S, S/R, and S/S stereoisomers
of MAM-2.
brated for 2 ns in the NPT ensemble at 1 atm pressure. Production runs were
performed in NVT ensemble imposing a membrane potential to increase the
ion permeation probability on the simulation timescale. The membrane po-
tential Vm was generated by introducing a constant electric field Ez on all
atoms along the z axis perpendicular to the membrane plane (52), Ez = Vm/Lz,
where Lz is the size of the simulated system in the z direction. The membrane
potential was set to the following voltages: –2, –1.5, and –1 V. Analysis of ion
permeation was done using Tcl scripting in VMD, whereas the pore profiles
were computed by the program HOLE (53). The End-to-End distance of per-
meant NMDG+ was computed at each simulation step as the distance between
the two terminal carbons. To compute the horizontality, we selected a vector
from the center of mass of the molecule to the terminal carbon linked to the
nitrogen, and then the complement of the polar angle was computed. The z
coordinate of the center of mass of permeant NMDG+ was plotted as a
function of the two previous observables.
For the MD simulations of cation permeation, two modifications were
introduced in the resulting model of the P2X open state. First, the MAM-2
linkers were removed (while keeping the mutated residues into cysteines) to
mimic the physiological conditions. Second, the extracellular domain of the
receptor was deleted to reduce the size of the system and to enhance the
sampling of permeation events on the simulation timescale. This modification
was done by introducing a peptide bond between D59 and F333 at the top of
the transmembrane domain. The resulting structure was energy-minimized
for 5,000 steps with NAMD 2.11 (47) using the CHARMM force-field ver-
Data Analysis. All experiments were performed at least four times from at least
two transfections, and values are presented as mean SEM. For modeling, values
of distribution of cis and trans configurations are presented as mean SE. The
number of cells or patches used for the experiments is provided in the text or
corresponding figure legends.
ACKNOWLEDGMENTS. We thank Professor Hongbo Zhao and Dr. François
Rassendren for providing human P2X2 and human P2X3 receptors, respec-
tively; and Dr. Frederic Bolze and Romain Vauchelles for imaging advising.
This work was supported by Agence Nationale de la Recherche Grant ANR-
14-CE11-0004-01 (to T.G.), the Ministère de la Recherche, and the Fondation
Pierre et Jeanne Spiegel. Financial support from the International Center for
Frontier Research in Chemistry (icFRC) and the Agence Nationale de la
Recherche through the LabEx project Chemistry of Complex Systems (ANR-
10-LABX-0026 CSC; to M.C.) is gratefully acknowledged. This work was
granted access to the High Performance Computing resources of the
Centre de Calcul Recherche et Technologie/Centre Informatique National
de l’Enseignement Supérieur/Institut du Développement et des Ressources
en Informatique Scientifique under the allocation 2016-[076644] made by
the Grand Equipement National de Calcul Intensif.
sion 36 (48). NMDG+, YO-PRO-12+ and spermidine3+ parameters were
,
obtained from the CHARMM general force field (49). During all simulations,
harmonic restraints (5 kcal·mol−1·Å−2) on the backbone atoms of the protein
were applied to preserve the configuration of the transmembrane domain
as in the MAM2-equilibrated model. The receptor was then embedded into
a pre-equilibrated 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC)
lipid bilayer, solvated with TIP3P water molecules and NaCl at 0.15 M or 1 M
concentration using VMD (50). To study permeation of large organic cations,
Na+ ions were replaced by NMDG+, YO-PRO-12+, or spermidine3+ (51). All
simulations were performed with periodic boundary conditions and Particle
Mesh Ewald long-range electrostatics. The system (∼90,000 atoms) was min-
imized during 5,000 steps, briefly thermalized (600ps) to 300K, and equili-
1. Idzko M, Ferrari D, Eltzschig HK (2014) Nucleotide signalling during inflammation.
Nature 509:310–317.
21. Compan V, et al. (2012) P2X2 and P2X5 subunits define a new heteromeric receptor
with P2X7-like properties. J Neurosci 32:4284–4296.
2. Khakh BS, North RA (2012) Neuromodulation by extracellular ATP and P2X receptors
in the CNS. Neuron 76:51–69.
22. Jiang LH, et al. (2005) N-methyl-D-glucamine and propidium dyes utilize different per-
meation pathways at rat P2X(7) receptors. Am J Physiol Cell Physiol 289:C1295–C1302.
3. Wirkner K, Sperlagh B, Illes P (2007) P2X3 receptor involvement in pain states. Mol
Neurobiol 36:165–183.
4. Yan D, et al. (2013) Mutation of the ATP-gated P2X(2) receptor leads to progressive
hearing loss and increased susceptibility to noise. Proc Natl Acad Sci USA 110:
2228–2233.
23. Pelegrin P, Surprenant
A (2006) Pannexin-1 mediates large pore formation and
interleukin-1beta release by the ATP-gated P2X7 receptor. EMBO J 25:5071–5082.
24. Eickhorst AN, Berson A, Cockayne D, Lester HA, Khakh BS (2002) Control of P2X(2)
channel permeability by the cytosolic domain. J Gen Physiol 120:119–131.
25. Fisher JA, Girdler G, Khakh BS (2004) Time-resolved measurement of state-specific
P2X2 ion channel cytosolic gating motions. J Neurosci 24:10475–10487.
26. Khakh BS, Egan TM (2005) Contribution of transmembrane regions to ATP-gated
P2X2 channel permeability dynamics. J Biol Chem 280:6118–6129.
27. Chaumont S, Khakh BS (2008) Patch-clamp coordinated spectroscopy shows
P2X2 receptor permeability dynamics require cytosolic domain rearrangements but
not Panx-1 channels. Proc Natl Acad Sci USA 105:12063–12068.
5. Pijacka W, et al. (2016) Purinergic receptors in the carotid body as a new drug target
for controlling hypertension. Nat Med 22:1151–1159.
6. North RA, Jarvis MF (2013) P2X receptors as drug targets. Mol Pharmacol 83:759–769.
7. Kawate T, Michel JC, Birdsong WT, Gouaux E (2009) Crystal structure of the ATP-gated
P2X(4) ion channel in the closed state. Nature 460:592–598.
8. Hattori M, Gouaux E (2012) Molecular mechanism of ATP binding and ion channel
activation in P2X receptors. Nature 485:207–212.
9. Kasuya G, et al. (2016) Structural insights into divalent cation modulations of ATP-
gated P2X receptor channels. Cell Reports 14:932–944.
10. Mansoor SE, et al. (2016) X-ray structures define human P2X(3) receptor gating cycle
and antagonist action. Nature 538:66–71.
11. Karasawa A, Kawate T (2016) Structural basis for subtype-specific inhibition of the
P2X7 receptor. eLife 5:5.
12. Browne LE, Jiang LH, North RA (2010) New structure enlivens interest in P2X recep-
tors. Trends Pharmacol Sci 31:229–237.
13. Habermacher C, Dunning K, Chataigneau T, Grutter T (2016) Molecular structure and
function of P2X receptors. Neuropharmacology 104:18–30.
14. Ding S, Sachs F (1999) Single channel properties of P2X2 purinoceptors. J Gen Physiol
113:695–720.
ꢀ
28. Yan Z, Li S, Liang Z, Tomic M, Stojilkovic SS (2008) The P2X7 receptor channel pore
dilates under physiological ion conditions. J Gen Physiol 132:563–573.
29. Shinozaki Y, et al. (2009) Direct observation of ATP-induced conformational changes
in single P2X(4) receptors. PLoS Biol 7:e1000103.
30. Khadra A, et al. (2012) Gating properties of the P2X2a and P2X2b receptor channels:
Experiments and mathematical modeling. J Gen Physiol 139:333–348.
31. Browne LE, Compan V, Bragg L, North RA (2013) P2X7 receptor channels allow direct
permeation of nanometer-sized dyes. J Neurosci 33:3557–3566.
32. Li M, Toombes GE, Silberberg SD, Swartz KJ (2015) Physical basis of apparent pore
dilation of ATP-activated P2X receptor channels. Nat Neurosci 18:1577–1583.
33. Ding S, Sachs F (1999) Ion permeation and block of P2X(2) purinoceptors: Single
channel recordings. J Membr Biol 172:215–223.
34. Sorge RE, et al. (2012) Genetically determined P2X7 receptor pore formation regu-
lates variability in chronic pain sensitivity. Nat Med 18:595–599.
15. Evans RJ (1996) Single channel properties of ATP-gated cation channels (P2X recep-
tors) heterologously expressed in Chinese hamster ovary cells. Neurosci Lett 212:
212–214.
16. Riedel T, Schmalzing G, Markwardt F (2007) Influence of extracellular monovalent
cations on pore and gating properties of P2X7 receptor-operated single-channel
currents. Biophys J 93:846–858.
35. Li M, Chang TH, Silberberg SD, Swartz KJ (2008) Gating the pore of P2X receptor
channels. Nat Neurosci 11:883–887.
36. Habermacher C, et al. (2016) Photo-switchable tweezers illuminate pore-opening
motions of an ATP-gated P2X ion channel. eLife 5:e11050.
37. Browne LE, et al. (2014) Optical control of trimeric P2X receptors and acid-sensing ion
channels. Proc Natl Acad Sci USA 111:521–526.
38. Guerra GP, Rubin MA, Mello CF (2016) Modulation of learning and memory by nat-
ural polyamines. Pharmacol Res 112:99–118.
39. Pippel A, et al. (2017) Localization of the gate and selectivity filter of the full-length
P2X7 receptor. Proc Natl Acad Sci USA 114:E2156–E2165.
17. Virginio C, MacKenzie A, Rassendren FA, North RA, Surprenant A (1999) Pore dilation
of neuronal P2X receptor channels. Nat Neurosci 2:315–321.
18. Khakh BS, Bao XR, Labarca C, Lester HA (1999) Neuronal P2X transmitter-gated cation
channels change their ion selectivity in seconds. Nat Neurosci 2:322–330.
19. Rokic MB, Stojilkovic SS (2013) Two open states of P2X receptor channels. Front Cell
Neurosci 7:215.
40. Clyne JD, Brown TC, Hume RI (2003) Expression level dependent changes in the
properties of P2X2 receptors. Neuropharmacology 44:403–412.
20. Wei L, Caseley E, Li D, Jiang LH (2016) ATP-induced P2X receptor-dependent large
pore formation: How much do we know? Front Pharmacol 7:5.
Harkat et al.
PNAS Early Edition
|
9 of 10

41. Fujiwara Y, Kubo Y (2004) Density-dependent changes of the pore properties of the
P2X2 receptor channel. J Physiol 558:31–43.
47. Phillips JC, et al. (2005) Scalable molecular dynamics with NAMD. J Comput Chem 26:
1781–1802.
42. Tochitsky I, et al. (2016) How azobenzene photoswitches restore visual responses to
the blind retina. Neuron 92:100–113.
43. Jiang R, et al. (2010) A putative extracellular salt bridge at the subunit interface
contributes to the ion channel function of the ATP-gated P2X2 receptor. J Biol Chem
285:15805–15815.
44. Harvey MJ, Giupponi G, Fabritiis GD (2009) ACEMD: Accelerating bio-
molecular dynamics in the microsecond time scale. J Chem Theory Comput 5:
1632–1639.
48. Best RB, et al. (2012) Optimization of the additive CHARMM all-atom protein force
field targeting improved sampling of the backbone φ, ψ and side-chain χ(1) and χ(2)
dihedral angles. J Chem Theory Comput 8:3257–3273.
49. Vanommeslaeghe K, et al. (2010) CHARMM general force field: A force field for drug-
like molecules compatible with the CHARMM all-atom additive biological force fields.
J Comput Chem 31:671–690.
50. Humphrey W, Dalke A, Schulten K (1996) VMD: Visual molecular dynamics. J Mol
Graph 14:27–38.
45. Heymann G, et al. (2013) Inter- and intrasubunit interactions between trans-
membrane helices in the open state of P2X receptor channels. Proc Natl Acad Sci USA
110:E4045–E4054.
46. Roitberg A, Elber R (1991) Modeling side chains in peptides and proteins: Application
of locally enhanced sampling and simulated annealing methods to find minimum
energy conformations. J Chem Phys 95:9277–9287.
51. Onasch F, et al. (1984) The interactions between nucleic acids and polyamines. III.
Microscopic protonation constants of spermidine. Biophys Chem 19:245–253.
52. Roux B (2008) The membrane potential and its representation by a constant electric
field in computer simulations. Biophys J 95:4205–4216.
53. Smart OS, Neduvelil JG, Wang X, Wallace BA, Sansom MS (1996) HOLE: A program for the
analysis of the pore dimensions of ion channel structural models. J Mol Graph 14:354–360, 376.
10 of 10
|
www.pnas.org/cgi/doi/10.1073/pnas.1701379114
Harkat et al.