Molecular tuning of fast gating in pentameric ligand-gated ion channels.

Abstract

Neurotransmitters such as acetylcholine (ACh) and glycine mediate fast synaptic neurotransmission by activating pentameric ligand-gated ion channels (LGICs). These receptors are allosteric transmembrane proteins that rapidly convert chemical messages into electrical signals. Neurotransmitters activate LGICs by interacting with an extracellular agonist-binding domain (ECD), triggering a tertiary/quaternary conformational change in the protein that results in the fast opening of an ion pore domain (IPD). However, the molecular mechanism that determines the fast opening of LGICs remains elusive. Here, we show by combining whole-cell and single-channel recordings of recombinant chimeras between the ECD of alpha7 nicotinic receptor (nAChR) and the IPD of the glycine receptor (GlyR) that only two GlyR amino acid residues of loop 7 (Cys-loop) from the ECD and at most five alpha7 nAChR amino acid residues of the M2-M3 loop (2-3L) from the IPD control the fast activation rates of the alpha7/Gly chimera and WT GlyR. Mutual interactions of these residues at a critical pivot point between the agonist-binding site and the ion channel fine-tune the intrinsic opening and closing rates of the receptor through stabilization of the transition state of activation. These data provide a structural basis for the fast opening of pentameric LGICs.

Full text

Page 1

Molecular tuning of fast gating in pentameric
ligand-gated ion channels
Thomas Grutter*†§, Lia Prado de Carvalho*†, Virginie Dufresne*, Antoine Taly*, Stuart J. Edelstein¶,
and Jean-Pierre Changeux*§
*Laboratoire ‘‘Re ´cepteurs et Cognition,’’ Institut Pasteur, 25 Rue du Dr. Roux, 75724 Paris Cedex 15, France; and¶Department of Biochemistry,
University of Geneva, 30 Quai Ernest-Ansermet, CH-1211 Geneva 4, Switzerland
Contributed by Jean-Pierre Changeux, October 17, 2005
Neurotransmitters such as acetylcholine (ACh) and glycine mediate
fast synaptic neurotransmission by activating pentameric ligand-
gated ion channels (LGICs). These receptors are allosteric trans-
membrane proteins that rapidly convert chemical messages into
electrical signals. Neurotransmitters activate LGICs by interacting
with an extracellular agonist-binding domain (ECD), triggering a
tertiary?quaternary conformational change in the protein that
results in the fast opening of an ion pore domain (IPD). However,
the molecular mechanism that determines the fast opening of
LGICs remains elusive. Here, we show by combining whole-cell and
single-channel recordings of recombinant chimeras between the
ECD of ?7 nicotinic receptor (nAChR) and the IPD of the glycine
receptor (GlyR) that only two GlyR amino acid residues of loop 7
(Cys-loop) from the ECD and at most five ?7 nAChR amino acid
residues of the M2-M3 loop (2–3L) from the IPD control the fast
activation rates of the ?7?Gly chimera and WT GlyR. Mutual
interactions of these residues at a critical pivot point between the
agonist-binding site and the ion channel fine-tune the intrinsic
opening and closing rates of the receptor through stabilization of
the transition state of activation. These data provide a structural
basis for the fast opening of pentameric LGICs.
allosteric proteins ? chimeric receptor ? Cys-loop receptor ? transition state
P
anionic glycine receptor (GlyR), mediate fast excitatory or
inhibitory chemical neurotransmission between neurons (1–6).
A unique feature of these receptors is that they activate the ion
channel, a process known as gating, in less than a ms. For
nicotinic receptors, a detailed single-channel analysis has re-
cently established a speed limit for the opening of the ion
channel in the ?s time range (7). Perturbations of this rapid
transmission pathway by natural mutants lead to severe diseases
such as congenital myasthenic syndromes (8), hereditary hy-
perekplexia (9), or epileptic disorders (10).
Pentameric LGICs, or Cys-loop receptors, are composed of
five homologous subunits, sharing a common structural organi-
zation, arranged (pseudo)symmetrically around the central ion
pore (1, 2). All subunits are made of two distinct topological
domains: the extracellular (ECD) and the ion pore domains
(IPD). First, the ECD is folded into a twisted ?-sandwich core,
as revealed by x-ray crystallographic studies of the mollusk
acetylcholine-binding protein (AChBP), a soluble pentameric
protein homologous to the extracellular domain of LGICs
(11–14).Second,electronmicroscopyimagesofTorpedonAChR
at 4-Å resolution revealed that the four transmembrane seg-
ments (M1 to M4) of the IPD are folded into ?-helices joined by
linking loops of variable lengths (15). By combining these
structural data, we built a 3D model of the full ?7 nAChR (16).
In this model, the coupling zone located at the interface between
the two domains is framed by flexible loops. As noted earlier for
GABAAreceptors (17), loops 2 and 7 (the so-called Cys-loop)
from the ECD are close (?5 Å) to the C-terminal end of ?-helix
M2 from the IPD. Even though structural information is now
entameric ligand-gated ion channels (LGICs), such as the
cationic nicotinic acetylcholine receptor (nAChR) and the
available for the linking of the ECD and IPD, the molecular
mechanism coupling the agonist-binding site and ion channel
resulting in the fast opening of the ion channel remains elusive.
In this article, we show, by combining whole-cell and single-
channel recordings of recombinant chimeras between the ECD
of the homomeric ?7 nicotinic receptor (nAChR) and the IPD
of the homomeric ?1 glycine receptor (GlyR), that only two
GlyR amino acid residues of the Cys-loop from the ECD and at
most five ?7 nAChR amino acid residues of the M2-M3 loop
(2–3L) from the IPD fine-tune the fast activation rates of the
?7?Gly chimera and WT GlyR. We propose that mutual inter-
actions of these residues with their WT counterparts in the
LGICs result in the acceleration of both the intrinsic opening
and closing rate constants through stabilization of the transition
state of activation. These data provide a structural basis for the
fast opening of pentameric LGICs.
Methods
Mutagenesis. Chick ?7 cDNA encoding the ECD of ?7 nAChR
was fused with the human ?1 cDNA encoding the IPD of ?1
glycine receptor. Chimeric cDNAs were subcloned into pMT3
containing the peptide signal sequence of the ?7 subunit.
Mutants of the Cys-loop region were performed on the synthetic
gene of the ?7 subunit as described (18), whereas, for the 2–3L
region and WT GlyR, mutations were created by using the
QuikChange kit (Stratagene) according to the manufacturer’s
instructions. All mutant and chimeric cDNAs were verified by
restriction enzyme analysis and sequencing.
Whole-Cell Recordings. Human embryonic kidney cells (HEK 293)
were grown and transiently transfected as described (18). Re-
cordings were made 48–72 h after transfection at room temper-
ature (25°C ? 3°C), by using the whole-cell patch-clamp tech-
nique (19). Cells were maintained at a holding potential of ?60
mV. External solution contained 140 mM NaCl, 2.8 mM KCl, 2
mM CaCl2, 2 mM MgCl2, 10 mM Glucose, and 10 mM Hepes–
NaOH (pH 7.3); Ca2?and Mg2?were removed just after the
whole-cell configuration was obtained. To determine chloride
permeability of the channel, extracellular chloride was reduced
through total replacement of NaCl of the external solution by
140 mM sodium isethionate. Patch pipette (1–2 M?) solutions
contained 140 mM CsCl2, 2 mM NaATP, 10 mM Hepes–CsOH,
and 10 mM EGTA (pH 7.3). Current-voltage relationships were
determined by the application of two inverted voltage ramps
(?50 to ?100 mV in 200 ms, from an initial holding potential of
?60 mV) during the steady-state phase of an ACh-evoked
Conflict of interest statement: No conflicts declared.
Abbreviations: LGIC, ligand-gated ion channel; nAChR, nicotinic acetylcholine receptor;
GlyR, glycine receptor; ECD, extracellular agonist-binding domain; IPD, ion pore domain;
2–3L, M2-M3 loop.
†T.G. and L.P.d.C. contributed equally to this work.
§To whom correspondence may be addressed. E-mail: grutter@pasteur.fr or changeux@
pasteur.fr.
© 2005 by The National Academy of Sciences of the USA
www.pnas.org?cgi?doi?10.1073?pnas.0509024102PNAS ?
December 13, 2005 ?
vol. 102 ?
no. 50 ?
18207–18212
NEUROSCIENCE

Page 2

response. Leak currents were subtracted. External solutions
containing the agonist (? antagonist) were delivered through
three parallel tubes placed immediately above the cell. These
tubes are displaced horizontally with the aid of a computer-
driven system (SF 77A Perfusion fast step, Warner) that ensured
a 10–90% solution exchange in 5–10 ms, as measured by changes
of tip potential of an open electrode in external solution and in
1:10 dilution of external solution with water. The perfusion
system thus determines accurately apparent on-rates kapp ?
1,000 s?1. Macroscopic currents were analyzed by using PULSEFIT
software (HEKA Electronics, Lambrecht?Pfalz, Germany). The
following single exponential function was fitted to the initial
rising (10–90%) agonist-evoked currents: I(t) ? I0 ? I?[1 ?
exp(?kapp? t)] where I0and I?are the initial and steady-state
currents, respectively, and kapp is the apparent constant of
activation. From the simple conformational change that follows
binding reactions between the agonist and its receptor, one can
derive kappas a function of agonist concentration from:
kapp? kopen?
[agonist]
rEC50? [agonist]?
N
? kclosed,
[1]
where kopenis an approximation of the opening rate constant,
kclosedis an approximation of the closing rate constant, rEC50is
the concentration of agonist that gives half of the theoretical
opening rate constant, and N the actual number of binding sites.
For all ?7?Gly chimeras, N was fixed to five as previously shown
for WT ?7 nAChR (20). Dose-response curves were obtained by
measuring the steady-state amplitude of the currents evoked by
different concentrations of agonist, and data were fitted with the
Hill equation to yield EC50and Hill coefficient, nH. All data
reported are means ? SD. Statistical differences were deter-
mined by using the Mann–Whitney U test with STATVIEW 5.0.
Single-Channel Recordings. Recordings were obtained in the cell-
attached configuration at 25°C ? 3°C at a holding potential of
?60 mV. Extracellular and patch pipette (around 5 M?) solu-
tions contained 140 mM NaCl, 2.8 mM KCl, 10 mM glucose, and
10 mM Hepes–NaOH (pH 7.3). Single-channel currents were
recorded by using an EPC-10 amplifier (HEKA), sampled at
100-?s intervals, recorded to the computer hard disk, and
detected by the half-amplitude threshold criterion by using the
program PULSETOOLS (HEKA) at a final bandwidth of 1 kHz.
Open-time histograms were plotted by using a logarithmic
abscissa and a square-root ordinate (21) and fitted to the sum of
exponentials with an imposed time resolution of 200 ?s.
Binding. ACh competition measurements against the initial rate
of ?-125I-bungarotoxin binding were as determined previously
(18, 22), and data were fitted with the Hill equation.
Results
Design of ?7?Gly Chimeras. To unravel the molecular basis of the
activation mechanism of LGICs, we constructed recombinant
chimeric subunits containing the two topological distinct do-
mains: the ECD of the nicotinic homomeric ?7 nAChR and the
IPD of the homomeric ?1 GlyR. These homomeric ?7?Gly
chimeras, homologous to the previously described ?7?5HT3
chimera (23), are novel because the ECD of a cationic receptor
is coupled to an anionic channel. Three different cDNA con-
structs were made and transfected in HEK 293 cells that resulted
in cell-surface expression of functional receptors. Only one
chimera, named ?7?Gly, is detailed in this article (for details, see
Supporting Materials and Methods and Fig. 4, which are published
as supporting information on the PNAS web site).
?7?Gly Chimera Is Gated by ACh and Selective to Chloride. The
?7?Gly chimera was gated by ACh, producing large whole-cell
inward currents (1–5 nA) with very little, if any, desensitization
(Fig. 1a). This chimeric receptor displayed the pharmacological
properties, as well as the allosteric site for calcium potentiation
(24), expected from its ?7 nAChR ECD and was selective for
chloride as expected from its GlyR ion channel.
First, ACh-evoked current amplitudes increased in a dose-
dependent manner (Fig. 1c); the apparent affinity (effective
concentration for half maximal response, EC50? 300 ?M; see
Table 1, which is published as supporting information on the
PNAS web site) was in the same range as described for ?7?5HT3
chimeric receptor (22). Second, the competitive ?7 receptor
antagonist dihydro-?-erythroidine (DH?E) reversibly inhibited
responses evoked by ACh (Fig. 1a). Third, in cells expressing
?7?Gly, external application of 4 mM Ca2?potentiated the
response evoked by a saturating ACh concentration (Fig. 1b).
This calcium potentiation was observed at all concentrations of
ACh, as described for ?7?5HT3chimera (24) (Fig. 1c). Finally,
substitution of external sodium chloride by sodium isethionate
resulted in the expected right-shift of reversal potential indica-
tive of a chloride permeability (Fig. 1d). These results demon-
strate the functional coupling of a cationic ECD to an anionic
IPD and strongly suggest that a common activation or gating
mechanism is shared by all members of the pentameric LGIC
superfamily.
Fig.1.
(1 mM ACh) through ?7?Gly and its reversible inhibition by 300 ?M dihydro-
?-erythroidine(DH?E).(b)Currentpotentiationby4mMCa2?duringa10-mM
ACh application. Recorded from a different cell. (c) Dose-response curves in
theabsence(opencircles)orpresence(filledcircles)of4mMCa2?.Currentsare
normalizedto10mMACh-evokedcurrents.(d)Current-voltagerelationships.
External sodium chloride (control condition) substituted by sodium isethion-
ate(Isecondition).(e)Typicalcurrentsevokedby100?M(bottomtrace),1mM
(middle trace), or 3 mM (top trace) ACh concentrations. kapp, derived from
singleexponentialfunctions,areindicated.(f)kappplottedasfunctionofACh
concentrations.DatawerefittedwithEq.1.Forclarity,onlypositiveerrorbars
are represented.
Functionalcharacterizationof?7?Glychimera.(a)Whole-cellcurrent
18208 ?
www.pnas.org?cgi?doi?10.1073?pnas.0509024102Grutter et al.

Page 3

?7?Gly Displays Slow Activation Rates. Interestingly, applications
longer than 1 s were necessary to reach a steady-state response
even at saturating ACh concentrations (Fig. 1 a, b, and e). This
slow activation contrasts with that of LGICs that activate in less
than a ms. We fitted the time course of the currents with single
exponential functions (Fig. 1e and Methods). The apparent
activationrates(kapp)increasedinadose-dependentmannerand
reached limiting values at around 3 mM (Fig. 1f). No linear
dependency was observed, ruling out a simple bimolecular
reactionbetweenAChandthechimera.Theexistenceoflimiting
kapp values for high concentrations indicates a more complex
reaction. A model in which a slow conformational transition
follows the rapid binding of ACh described the data (Fig. 1f and
Methods), yielding initial estimates of the channel opening
(kopen? 2.3 ? 0.6 s?1) and closing rates (kclosed? 0.20 ? 0.07
s?1; see Table 1). This model was successfully used to describe
kinetic properties of the Torpedo nAChR (25) and ?1 GlyR (26).
The kopenvalues derived for ?7?Gly were ?400-fold slower than
the time resolution of our perfusion system (?1,000 s?1; see
Methods) excluding possible limitations due to solution ex-
change. Therefore, these results suggest that the apparent ki-
netics of ?7?Gly activation are specifically altered.
The Cys-Loop Plays a Major Role in Accelerating the Kinetics of
Activation. Given that both WT ?7 nAChR and GlyR are fast
gating receptors, we hypothesized that a poor coupling between
the two domains resulted in the slow activation kinetics in
?7?Gly chimera. We built a molecular model of the ?7?Gly
chimera by homology modeling using the previously described
?7model(16).ThemodelshowsthattheCys-loopfromtheECD
plunges into the extracellular face of the IPD close to the four
M1-M4 ?-helices (Fig. 2a). Comparison of sequences shows that
some residues are not conserved in the Cys-loop between
nAChR and GlyR (see Fig. 5, which is published as supporting
information on the PNAS web site). Thus, an attractive hypoth-
esis is that in the ?7?Gly chimera interactions between Cys-loop
Fig. 2.
(blue marine). ?7 Cys-loop and numbers are highlighted in red. (b) Normalized whole-cell currents (10 mM ACh) through ?7?Gly (gray trace) and ?7(Cys-L)?Gly
(dark trace). (c) Dose-response curves for ?7?Gly (open circles) and ?7(Cys-L)?Gly (filled triangles). Whole-cell currents are normalized to 10 mM ACh-evoked
currents. (d) ACh competition against ?-125I-bungarotoxin binding to ?7?Gly (open circles, Kp ? 540 ?M) and ?7(Cys-L)?Gly (filled triangles, Kp ? 600 ?M).
Maximalbinding(Bmax)isdefinedwithoutACh.(e)Concentrationdependencyofkappfor?7?Gly(opencircles)and?7(Cys-L)?Gly(filledtriangles).Dataarefitted
withEq.1.For?7?Gly,dataarethesameasforFig.1f.Forclarity,onlypositiveerrorbarsarerepresented.(f)Single-channelcurrentsthrough?7?Gly(graytrace)
and ?7(Cys-L)?Gly (dark trace) from cell-attached patches (10 ?M ACh). Currents are displayed at 1-kHz bandwidth, with channel openings as downward
deflections. (g) Open time histograms. Simulated probability density functions are indicated for each unliganded (A0) and liganded open state (A1, . . . , A5). The
sumofthesefunctionsisindicatedingreentraces(fordetails,seeSupportingMaterialsandMethods).(h)Comparisonofcomputed(sim)andexperimental(exp)
whole-cell currents (10 mM ACh) through ?7?Gly (gray traces) and ?7(Cys-L)?Gly (dark traces). (i) Gating free energy diagram for ?7?Gly (gray trace) and
?7(Cys-L)?Gly (dark trace). ??Gclosedis arbitrarily set to zero, and vertical bar represents 10 kcal?mol.
Kinetic characterization of ?7(Cys-L)?Gly chimera. (a) 3D model of the ?7?Gly interface (one subunit shown) with ?7 ECD (pale yellow) and GlyR IPD
Grutter et al.PNAS ?
December 13, 2005 ?
vol. 102 ?
no. 50 ?
18209
NEUROSCIENCE

Page 4

residues from ?7 and IPD residues from GlyR might not be
optimal, conferring a poor coupling between the two domains.
To test this hypothesis, we replaced in ?7?Gly its chick ?7
Cys-loop by the homologous human ?1 GlyR Cys-loop. This
substitution resulted in a new chimera ?7(Cys-L)?Gly that
displayed enhanced activation rates at saturating concentrations
of ACh concomitant with an increase of the apparent desensi-
tization rate (Fig. 2b). These data demonstrate that the Cys-loop
plays a major role in the fast activation rate in the ?7(Cys-L)?Gly
chimera.
The Cys-Loop Stabilizes the Transition State of Activation. These
results might be explained by changes in agonist association
rates, channel-gating rates, and?or conductance levels. The
following observations support a change in the channel-gating
rates only.
First, whole-cell dose-response curves revealed no significant
differences of EC50values between ?7?Gly and ?7(Cys-L)?Gly
chimeras (Fig. 2c). Furthermore, no differences were noticed in
the apparent dissociation constants of ACh as measured by
competition against ?-125I-bungarotoxin binding to intact cells
(Fig. 2d). Second, compared with ?7?Gly, a parallel increase of
kapp was observed for ?7(Cys-L)?Gly (Fig. 2e). As a result,
significant increases in both kopen (16.4 ? 3.9 s?1) and kclosed
(1.77 ? 0.68 s?1) were observed (7-fold for kopenand 9-fold for
kclosed,P?0.01;forotherkineticdetails,seeSupportingMaterials
and Methods). Third, single-channel recordings revealed that 10
?M ACh elicited shorter openings in ?7(Cys-L)?Gly as com-
pared with ?7?Gly chimera (Fig. 2f). Indeed, open-time histo-
grams were fitted with three exponential components for
?7(Cys-L)?Gly (?1? 0.89 ms, a1? 0.35; ?2? 8.5 ms, a2? 0.35;
?3? 65 ms, a3? 0.30) and with four components for ?7?Gly
(?1? 0.75 ms, a1? 0.20; ?2? 5.9 ms, a2? 0.21; ?3? 42 ms, a3?
0.28; ?4? 193 ms, a4? 0.31; Fig. 2g). A left-shift to shorter
open-time values of all components was observed for ?7(Cys-
L)?Gly compared with ?7?Gly that resulted in a 3-fold decrease
of the mean open time (?mean? 23 ms for ?7(Cys-L)?Gly versus
?mean ? 73 ms for ?7?Gly). This decrease implies that the
shortest component for ?7(Cys-L)?Gly equivalent to that ob-
served for ?7?Gly is likely to be missed, explaining that only
three exponential components were sufficient to fit the data.
Because the number of channels in patches is unknown, com-
parison of the closed time histograms between chimeras was
judged to be uninformative and was not included. Fourth, no
significant modifications on the conductance levels were ob-
served for both chimeras (data not shown). Finally, to help
illustrate the experimental data, a set of theoretical parameters
derived from a kinetic-based allosteric model (see Fig. 6, which
is published as supporting information on the PNAS web site)
(27) simulated adequately the macroscopic currents (Fig. 2h)
and open-time distribution of ?7?Gly (Fig. 2g; for modeling
details,seeSupportingMaterialsandMethods).Itshouldbenoted
thatthesesimulationsillustrateonlythefunctionalresultsanddo
not result from a fitting procedure through the data. Simulations
show that changing only the intrinsic opening and closing rates
in ?7(Cys-L)?Gly resulted in the expected increase of the
activation rate observed in whole-cell recordings (Fig. 2h) and
decrease of the mean open time observed in single-channel
recordings (Fig. 2g). In term of free energy, simulations pre-
dicted a stabilization of the transition state between the closed
andopenconformations(??G‡)by?1.36kcal?mol(Fig.2i).We
concluded from these data that the Cys-loop substitution in-
creases the intrinsic opening and closing rates by stabilization of
the transition state, without modifying substantially the ACh
binding or the ion conduction properties.
Only Two Residues from the Cys-Loop Fine-Tune the Kinetics of
Activation in ?7?Gly and WT GlyR. We next investigated which
amino acids within the Cys-loop are responsible for the fast
activation rate. A systematic analysis of the Cys-loop was un-
dertaken by introducing microcassettes or single residues of the
?1 GlyR Cys-loop into ?7?Gly. For each construction, kopenand
kclosedrates were as previously derived (for a detailed description
of each construction, see Fig. 7, which is published as supporting
information on the PNAS web site, and Supporting Materials and
Methods). This study showed that the double mutation V131L?
W133N was sufficient to confer a fast activation rate, similar to
that of ?7(Cys-L)?Gly (Fig. 3a). Then, to test the reciprocity of
the mutation, we produced the converse double mutant L142V?
N144W in the WT GlyR. This mutant displayed responses to
glycine that were significantly slower (7-fold, P ? 0.05) than
those of the WT GlyR (Fig. 3 a and b). This difference was in
the same range (?10-fold) as observed after the Cys-loop
exchange in ?7?Gly chimera (Fig. 3a). Consistent with the fact
that ?7(Cys-L)?Gly was not as fast as the WT GlyR, the double
mutant L142V?N144W in GlyR remained faster than ?7?Gly
chimera (Fig. 3a). Overall, our results highlight the functional
Fig. 3.
indicated constructions. Sequences of the Cys-loop and 2–3L region are indi-
cated. Residues from ?7 nAChR are indicated in gray and GlyR in bold-type
characters. (b) Normalized whole-cell currents (10 mM glycine) through GlyR
WT(graytrace)andL142V?N144Wmutant(darktrace).(c)Normalizedwhole-
cell currents (10 mM ACh) through ?7?Gly (gray trace) and ?7?(2–3L)Gly (dark
trace). (d) Mapping of the identified residues on the ?7?Gly 3D model (red
spheres for Cys-L and cyan for 2–3L). The inner and outer blocks are colored in
orangeandgreen,respectively.Forclarity,onlyfoursubunitsarerepresented.
(Inset) Enlarged view of the coupling zone.
Tuning the activation kinetics in LGICs. (a) kopen bar plot of the
18210 ?
www.pnas.org?cgi?doi?10.1073?pnas.0509024102Grutter et al.

Page 5

importance of the two residues from the Cys-loop in tuning the
kinetics of activation in both the chimeric and WT receptors and
suggest that other regions from the ECD might also contribute
to the fast activation rate.
The M2-M3 Loop also Plays a Major Role in Accelerating the Kinetics
of Activation. Finally, the 3D model of ?7?Gly showed that the
side chains of the two identified residues of the Cys-loop are in
close contact with those of M2-M3 loop (2–3L) (Fig. 3d). This
loop links the two ?-helices M2?M3 and is not conserved
between nAChRs and GlyRs?GABAAreceptors (Fig. 5). We
reasoned that replacing the human ?1 GlyR residues of 2–3L in
?7?Gly (Y279-A282) by the homologous residues of chick ?7
nAChR (D265-L269) should restore a homogeneous coupling
betweentheCys-loopand2–3L.Therefore,weproducedthenew
chimera ?7?(2–3L)Gly, which indeed displayed a significantly
accelerated activation of the current evoked by saturating con-
centration of ACh, yielding kopen values close to those of
?7(Cys-L)?Gly (Fig. 3 a and c). These data show that the
introduction of at most five ?7 nAChR residues in the 2–3L is
sufficient to rescue a homogeneous coupling in ?7?Gly chimera,
as did the exchange of two ?1 GlyR amino acids in the Cys-loop.
Thus, our results demonstrate that both the Cys-loop and the
2–3L region mediate bidirectional allosteric coupling between
the agonist-binding and the ion channel domains in LGICs.
Discussion
In the present study, we demonstrate that the Cys-loop and 2–3L
region fine-tune the speed of the signal transduction in LGICs.
We reached this conclusion by designing a chimeric receptor
made from the ECD of ?7 nAChR and the IPD of GlyR. This
chimera is functional but displays abnormally slow activation
rates. By swapping either the Cys-loop or the 2–3L region with
their WT counterparts, we succeeded in accelerating the slow
activation rate of the ?7?Gly chimera. A mutational analysis
indicated that only a double mutation in the Cys-loop (V131L?
W133N)sufficestorestorethefastactivationrateinthechimera.
We then succeeded in slowing the WT GlyR by producing the
converse double mutant (L142V?N144W). Therefore, the Cys-
loop and 2–3L play critical roles in the fine-tuning of fast gating
in LGICs. Our results also show that the mutated ?7?Gly
chimeras were still not as fast as the WT GlyR and that the
mutated GlyR was not as slow as ?7?Gly, suggesting that other
regions, yet unidentified (possibly loops 2, 9, and pre-M1, and
the C-terminal end of M4), might contribute to the control of the
fast gating.
The Cys-loop (28–32) and 2–3L region (17) have been pre-
viously identified as important loops for the allosteric coupling;
our present work demonstrates their role in the control of the
fast opening of the ion channel. Furthermore, sequence analysis
shows that the 2–3L region identified here differs from that
found earlier in the GABAAreceptor (17). This discrepancy may
be explained by local changes in the structure of the 2–3L region
in the GABAAreceptor or alternatively by differences in the
construction of the 3D model. Indeed, on the basis of the recent
EM images of the ion channel, the M2 segment was interpreted
as a 40-Å-long ?-helix (15), longer than it was commonly
assumed. This extension inevitably brings the 2–3L loop to a
more C-terminal region that corresponds to the region we
identified in the present study.
Our results, supported by the 3D model of the receptor,
suggest that the Cys-loop and 2–3L region interact coopera-
tively to produce fast gating (Fig. 3d). Three general conclu-
sions can be drawn about their mutual interactions in the
gating process.
First, although a common activation mechanism is shared by
cationic and anionic LGICs, fine structural complementarity at
the interface region is likely to be a general rule for optimal
allosteric coupling, as already suggested by chimeras made from
the acetylcholine-binding protein (AChBP) and the 5HT3 re-
ceptor (18, 32). Indeed, for GlyR coupling, fast gating is medi-
ated by only two nonconserved amino acid residues from the
Cys-loop whereas for ?7 nAChR fast coupling occurs through at
most five nonconserved residues from the 2–3L region. There-
fore, optimal coupling in LGICs is mediated by specific inter-
actions, which depend on residues found specifically in anionic
versus cationic receptors. The specific coupling interactions
between GlyR residues depicted in Fig. 3d should therefore be
different from those of ?7 nAChR. Further experiments are
needed to identify the amino acids from the GlyR 2–3L region
that interact with the two Cys-loop residues and the amino acids
from the ?7 nAChR Cys-loop that interact with the five 2–3L
residues, to assign more precisely the minimal cluster of residues
that mediate fast gating in the LGICs.
Second, the complementary interaction between the Cys-
loop and 2–3L results probably in the stabilization of the
transition state of gating. In turn, this stabilization allows the
fast opening of the ion channel necessary to mediate fast
synaptic neurotransmission. For such large multimeric pro-
teins, subtle changes in the transition state energy barrier
(?1.36 kcal?mol) might explain their swiftness of activation.
We propose that fast gating of LGICs is thus achieved by the
interactions of some critical residues that lower the energy
barrier by a few kcal?mol.
Third, the interaction of the Cys-loop with 2–3L region
facilitates the physical opening of the ion pore. This last con-
clusion is supported by several structural gating mechanisms (15,
16, 33). In particular, normal mode analysis recently proposed
that the conformational transition that causes the opening of the
ion pore is essentially a quaternary twist motion of the protein
accompanied by discrete tertiary changes of each subunit (16).
These tertiary changes correspond to the relative motions of two
rigid blocks (Fig. 3d). One of these, the inner block, is essentially
composed of the inner ?-sheet of the ECD that carries the
agonist-binding site and ?-helix M2 that frames the ion channel
wall (16). Therefore, signal transmission between the agonist-
binding site and the ion channel is mediated by the intrinsic
concerted movement of the rigid inner block. The two flexible
loops, the Cys-loop and 2–3L region, undergo discrete confor-
mational changes, enabling the concerted motions of the two
rigid blocks associated with a minimal energy cost, facilitating
the rapid twist of all five ?-helix M2 necessary to open the ion
channel (see Movies 1 and 2, which are published as supporting
information on the PNAS web site). According to this structural
gating mechanism, the Cys-loop and 2–3L region are critically
positioned at a pivot point between the ACh-binding site and the
ion channel (Fig. 3d and Fig. 8, which is published as supporting
information on the PNAS web site).
Finally, a natural mutation that causes severe myasthenia was
recently found within the Cys-loop in the nAChR ?-subunit (34).
The mutation, homologous to the V131L mutation described in
the current work, displayed abnormal kinetics of activation.
Therefore, beyond the basic understanding of the molecular
details of fast activation processes of LGICs, the identified
region may be viewed as a potential target for new therapeutic
agents that could act as allosteric modulators able to rescue
abnormal activation occurring in natural mutants.
We thank B. Molles for critical reading of the manuscript, A. Marty for
fruitful discussion on single-channel analysis, H. Betz (Max-Planck-
Institute for Brain Research, Frankfurt, Germany) for providing the
human ?1 GlyR clone, and M. Soudan and Y. Archambeau for technical
assistance. This work was supported by the Association Franc ¸aise Contre
les Myopathies, the Commission of the European Communities, the
Association pour la Recherche Contre le Cancer, the Colle `ge de France,
and the Centre National de la Recherche Scientifique.
Grutter et al.PNAS ?
December 13, 2005 ?
vol. 102 ?
no. 50 ?
18211
NEUROSCIENCE

Page 6

1. Corringer, P. J., Le Nove `re, N. & Changeux, J. P. (2000) Annu. Rev. Pharmacol.
40, 431–458.
2. Karlin, A. (2002) Nat. Rev. Neurosci. 3, 102–114.
3. Changeux, J. P. & Edelstein, S. J. (2005) Science 308, 1424–1428.
4. Lester, H. A., Dibas, M. I., Dahan, D. S., Leite, J. F. & Dougherty, D. A. (2004)
Trends Neurosci. 27, 329–336.
5. Connolly, C. N. & Wafford, K. A. (2004) Biochem. Soc. Trans. 32, 529–534.
6. Absalom, N. L., Lewis, T. M. & Schofield, P. R. (2004) Exp. Physiol. 89,
145–153.
7. Chakrapani, S. & Auerbach, A. (2005) Proc. Natl. Acad. Sci. USA 102, 87–92.
8. Engel, A. G., Ohno, K. & Sine, S. M. (2003) Nat. Rev. Neurosci. 4, 339–352.
9. Laube, B., Maksay, G., Schemm, R. & Betz, H. (2002) Trends Pharmacol. Sci.
23, 519–527.
10. Bertrand, D., Picard, F., Le Hellard, S., Weiland, S., Favre, I., Phillips, H.,
Bertrand, S., Berkovic, S. F., Malafosse, A. & Mulley, J. (2002) Epilepsia 43,
112–122.
11. Smit, A. B., Syed, N. I., Schaap, D., van Minnen, J., Klumperman, J., Kits, K. S.,
Lodder, H., van der Schors, R. C., van Elk, R., Sorgedrager, B., et al. (2001)
Nature 411, 261–268.
12. Brejc, K., van Dijk, W. J., Klaassen, R. V., Schuurmans, M., van der Oost, J.,
Smit, A. B. & Sixma, T. K. (2001) Nature 411, 269–276.
13. Celie, P. H. N., van Rossum-Fikkert, S. E., van Dijk, W. J., Brejc, K., Smit, A. B.
& Sixma, T. K. (2004) Neuron 41, 907–914.
14. Unwin, N. (2005) J. Mol. Biol. 346, 967–989.
15. Miyazawa, A., Fujiyoshi, Y. & Unwin, N. (2003) Nature 423, 949–955.
16. Taly, A., Delarue, M., Grutter, T., Nilges, M., Le Novere, N., Corringer, P. J.
& Changeux, J. P. (2005) Biophys. J. 88, 3954–3965.
17. Kash, T. L., Jenkins, A., Kelley, J. C., Trudell, J. R. & Harrison, N. L. (2003)
Nature 421, 272–275.
18. Grutter, T., Prado de Carvalho, L., Dufresne, V., Taly, A., Fischer, M. &
Changeux, J. P. (2005) C. R. Biol. 328, 223–234.
19. Hamill, O. P., Marty, A., Neher, E., Sakmann, B. & Sigworth, F. J. (1981)
Pflu ¨gers Arch. 391, 85–100.
20. Palma, E., Bertrand, S., Binzoni, T. & Bertrand, D. (1996) J. Physiol. (London)
491, 151–161.
21. Sigworth, F. J. & Sine, S. M. (1987) Biophys. J. 52, 1047–1054.
22. Grutter, T., Prado de Carvalho, L., Le Nove `re, N., Corringer, P. J., Edelstein,
S. & Changeux, J. P. (2003) EMBO J. 22, 1990–2003.
23. Eisele ´, J. L., Bertrand, S., Galzi, J. L., Devillers-Thie ´ry, A., Changeux, J. P. &
Bertrand, D. (1993) Nature 366, 479–483.
24. Galzi, J. L., Bertrand, S., Corringer, P. J., Changeux, J. P. & Bertrand, D. (1996)
EMBO J. 15, 5824–5832.
25. Heidmann, T. & Changeux, J. P. (1980) Biochem. Biophys. Res. Commun. 97,
889–896.
26. Grewer, C. (1999) Biophys. J. 77, 727–738.
27. Edelstein, S. J., Schaad, O., Henry, E., Bertrand, D. & Changeux, J. P. (1996)
Biol. Cybern. 75, 361–379.
28. Sala, F., Mulet, J., Sala, S., Gerber, S. & Criado, M. (2005) J. Biol. Chem. 280,
6642–6647.
29. Absalom, N. L., Lewis, T. M., Kaplan, W., Pierce, K. D. & Schofield, P. R.
(2003) J. Biol. Chem. 278, 50151–50157.
30. Schofield, C. M., Jenkins, A. & Harrison, N. L. (2003) J. Biol. Chem. 278,
34079–34083.
31. Schofield, C. M., Trudell, J. R. & Harrison, N. L. (2004) Biochemistry 43,
10058–10063.
32. Bouzat, C., Gumilar, F., Spitzmaul, G., Wang, H. L., Rayes, D., Hansen, S. B.,
Taylor, P. & Sine, S. M. (2004) Nature 430, 896–900.
33. Law, R. J., Henchman, R. H. & McCammon, J. A. (2005) Proc. Natl. Acad. Sci.
USA 102, 6813–6818.
34. Shen, X. M., Ohno, K. J., Tsujino, A., Brengman, J. M., Gingold, M., Sine, S. M.
& Engel, A. G. (2003) J. Clin. Invest. 111, 497–505.
18212 ?
www.pnas.org?cgi?doi?10.1073?pnas.0509024102Grutter et al.