Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Theory

A Gating Mechanism of the Serotonin 5-HT3


Receptor
Graphical Abstract Authors
Shuguang Yuan, Slawomir Filipek,
Horst Vogel

Correspondence
shuguang.yuan@gmail.com (S.Y.),
horst.vogel@epfl.ch (H.V.)

In Brief
Yuan et al. employ all-atom long-
timescale molecular dynamics
simulations to reveal activation
mechanism of the serotonin 5-HT3
receptor. Serotonin binding first induces
distinct changes in the highly conserved
ligand-binding cage, followed by tilting-
twisting movements of the extracellular
domain, which leads to opening of the
transmembrane helices and the
hydrophobic gate. Finally, the
intracellular helix bundle opens lateral
ports for ion passage.

Highlights
d Rotamer change of W156 in the highly conserved
aromatic cage

d Tilting-twisting movements captured at the extracellular


domain

d Rotamer change of L260 lead to formation of a continuous


water pathway

d Chlorine ions penetrate into the intracellular vestibule

Yuan et al., 2016, Structure 24, 816–825


May 3, 2016 ª2016 Elsevier Ltd All rights reserved
http://dx.doi.org/10.1016/j.str.2016.03.019
Structure

Theory

A Gating Mechanism of the Serotonin 5-HT3 Receptor


Shuguang Yuan,1,* Slawomir Filipek,2 and Horst Vogel1,*
1Instituteof Chemical Sciences and Engineering, Ecole Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne, Switzerland
2Laboratory of Biomodeling, Faculty of Chemistry & Biological and Chemical Research Centre, University of Warsaw, Pasteura 1,
02-093 Warsaw, Poland
*Correspondence: shuguang.yuan@gmail.com (S.Y.), horst.vogel@epfl.ch (H.V.)
http://dx.doi.org/10.1016/j.str.2016.03.019

SUMMARY expression, purification, and crystallization of these proteins in


a functional state (Bill et al., 2011).
Our recently solved high-resolution structure of the Present models of Cys-loop receptor structures are based on
serotonin 5-HT3 receptor (5-HT3R) delivered the first electron microscopy images of Torpedo acetylcholine receptor
detailed structural insights for a mammalian pentame- (AChR) (Unwin, 2013), crystal structures of the bacterial channel
ric ligand-gated ion channel. Based on this structure, homologs GLIC and ELIC (Bocquet et al., 2009; Hilf and Dutzler,
we here performed a total of 2.8-ms all-atom molecular 2008, 2009), a Caenorhabditis elegans glutamate-gated chloride
channel, GluCl (Althoff et al., 2014), and most recently, the first
dynamics simulations to unravel at atomic detail how
high-resolution structure of a mammalian pLGIC, the serotonin
neurotransmitter binding on the extracellular domain
5-HT3 receptor (5-HT3R) (Hassaine et al., 2014). Hence, a pro-
induces sequential conformational transitions in the totypical pLGIC is composed of three distinct structural and
receptor, opening an ion channel and translating a functional entities: a large b-sheet structured extracellular (EC)
chemical signal into electrical impulses across the domain comprising the ligand-binding site(s), a helical trans-
membrane. We found that serotonin binding first membrane (TM) channel-forming part, and an intracellular (IC)
induces distinct conformational fluctuations at the domain that influences ion conductance and interacts with
side chain of W156 in the highly conserved ligand- cellular scaffolding proteins (Thompson et al., 2010). The intra-
binding cage, followed by tilting-twisting movements cellular domain is missing in the non-mammalian receptor struc-
of the extracellular domain which couple to the trans- tures such as GLIC, ELIC, and GluCl and is only poorly resolved
membrane TM2 helices, opening the hydrophobic in the Torpedo AChR structure (Unwin, 2013), but is well resolved
in the 5-HT3R structure (Hassaine et al., 2014).
gate at L260 and forming a continuous transmem-
Although the recent high-resolution 3D structures of bacterial
brane water pathway. The structural transitions in
and eukaryotic pLGICs resulted in a common structure fold of
the receptor’s transmembrane part finally couple to the Cys-loop receptor family, the central question remains un-
the intracellular MA helix bundle, opening lateral ports resolved, namely, how the binding of an activating ligand at the
for ion passage. EC receptor domain induces the opening of a gate about 60 Å
away in the TM region, allowing the translocation of ions across
the receptor. As experimental details about the structural transi-
INTRODUCTION tions during channel activation are lacking, allosteric models
have been derived from crystallographic data and molecular dy-
The complex mental capacity and motor behavior of mammals namics (MD) simulations of the different states of the channel
rely on the evolution of a distinguished central and peripheral structures of GLIC, ELIC, and GluCl (Aryal et al., 2015; Calimet
nervous system in which ligand-gated ion channels mediate et al., 2013; Nury et al., 2010; Sauguet et al., 2014; Zhu and
fast intercellular signaling at synapses. Pentameric ligand- (i.e., Hummer, 2012).
neurotransmitter)-gated ion channels (pLGIC) of the Cys-loop Here we have used a total of 2.8-ms all-atom MD simulations
receptor family are the most prominent representatives, and to elucidate at atomic resolution the structural transitions of the
are also important targets for treating many neuronal disorders 5-HT3R in a lipid bilayer during activation by binding its natural
(Corringer et al., 2012; Hurst et al., 2013; Lemoine et al., 2012; neurotransmitter serotonin. Our computer experiments started
Smart and Paoletti, 2012; Thompson et al., 2010; Walstab with the crystal structure of the 5-HT3R in the closed-channel
et al., 2010). These receptors function as allosteric signal trans- form (Hassaine et al., 2014), removing the five stabilizing nano-
ducers across the plasma membrane: at a synapse, a released bodies, embedding the receptor in a lipid bilayer, then inserting
neurotransmitter binds specifically at the extracellular site of five serotonin molecules into the binding sites of the homopenta-
its LGIC receptor, inducing multiple conformational changes meric 5-HT3R before finally observing the structural transitions
to open an intrinsic ion-selective channel (Auerbach, 2014; of the receptor during the activation process toward the open-
Changeux, 2014; Corringer et al., 2012; daCosta and Baenziger, channel state. Cryoelectron tomography has shown that the
2013; Schmauder et al., 2011). However, structural and mech- crystal structure of the 5-HT3R is preserved in lipid bilayers
anistic details of this central process remain speculative, as (Kudryashev et al., 2016). To distinguish ligand-induced confor-
directly determined high-resolution structures of mammalian mational changes from stochastic conformational fluctuations,
Cys-loop receptors are rare due to difficulties in heterologous we performed in addition MD simulations of the apo-5-HT3R,

816 Structure 24, 816–825, May 3, 2016 ª2016 Elsevier Ltd All rights reserved
(5HTBP) in the crystal structure comprised serotonin in each of
the five binding sites (PDB: 2YMD, 2YME) (Kesters et al., 2013),
we also included five serotonin molecules in the 5-HT3R struc-
ture for simulations. The insertion of serotonin into the ligand-
free 5-HT3R structure (Hassaine et al., 2014) created conforma-
tional stress in the receptor’s binding cage, which initiated
during the first 30-ns equilibration phase structural rearrange-
ments in each of the five ligand-binding cages (Figure 1A): The
serotonin molecule was stabilized via p-p stacking with the
aromatic side chains of Y126 and W156; a cation-p stacking
was established between R65 and serotonin; the aromatic
side chain of W63 was oriented perpendicularly to the aromatic
plane of serotonin, forming a s-p interaction; and a salt bridge
was formed between serotonin’s amine group and E209.
During the subsequent MD simulation period, serotonin and
Figure 1. W156 Conformational Switches and Interaction Network
between Residues in the 5-HT3R
residues in the aromatic cage underwent further conformational
(A) Serotonin in the ‘‘aromatic cage’’ (dotted circle) at the interface between transitions. While the aromatic moiety of serotonin was partially
two neighboring extracellular domains. p-p stacking between F199 and Y207, the side chain of W156
(B) Configuration of serotonin in the aromatic cage of the 5-HT3R after 30 ns of switched toward a stable position (Figures 1A–1C). Such confor-
MD equilibration. mational switches, which are represented in the time course of
(C) Configuration of serotonin in the aromatic cage of the 5-HT3R after 700 ns c1/c2 angles of W156, took place consecutively in all subunits
of MD simulation.
of the serotonin-bound receptor (Figure S2). Notably, for the
apo-5-HT3R both c1 and c2 remained unchanged during the
i.e., the receptor in the absence of an activating ligand. On this entire MD simulation period (Figure S2).
basis our MD simulations delivered a realistic description of The 5-HT3R is a prototypical representative of an allosterically
how the binding of the activating neurotransmitter on the extra- regulated pLGIC where binding of the neurotransmitter serotonin
cellular site of the 5-HT3R induces a sequence of conformational activates an ion channel via a network of internal conformational
changes leading to the opening of the transmembrane ion chan- changes (Schmauder et al., 2011; Lummis, 2012). To characterize
nel, and finally to structural changes at the intracellular site of the the dynamic coupling of the serotonin-binding pocket to other re-
receptor, and are therefore of direct relevance for the under- gions of the receptor, we calculated the atomic displacement cor-
standing of neuronal signal transmission on an atomic scale. relations for each residue pair in MD simulations for the 5-HT3R
without and with bound serotonin. In the correlation network rep-
RESULTS resentation (Figure 3), nodes correspond to protein residues con-
nected by edges and weighted by the strength of their respective
Binding of the Activating Neurotransmitter Serotonin to correlation values. The relation of the correlation network to the 3D
the 5-HT3R: The Role of W156 and Conserved Motifs protein structure is also depicted in Figure 3. This method has been
All MD simulations have been performed on the 5-HT3R inte- applied elsewhere to dissect allosteric couplings in various sys-
grated into a bilayer of 1-palmitoyl-2-oleoyl-phosphatidylcholine tems (Scarabelli and Grant, 2014; Sethi et al., 2009). Correlation
(POPC), as it was shown elsewhere that the channel proper- matrices (Figure S3) show that all loops of the binding pocket (Fig-
ties of the receptor are properly reconstituted in such mem- ure 2) strongly couple to the extracellular b strands. As indicated in
branes (Hassaine et al., 2014). By performing 2 3 700-ns Figure 3, important correlated motions occur between distinct
all-atom MD simulations for both the agonist-free (apo) and structural regions of the extracellular part of the receptor. Binding
the agonist-bound 5-HT3R in a lipid bilayer, we elucidated at of the neurotransmitter serotonin affects collective motions, espe-
atomic resolution the structural transitions of the receptor during cially in loop B (where W156 is located), loop C, the bottom por-
ligand-induced activation. A wealth of functional studies on the tions of b7-b9-b10, b1-b2-b6, and b10 -b70 -b90 -b100 . While in the
5-HT3R (Lummis, 2012), the recent high-resolution crystal struc- apo-5-HT3R (Figure 3A), the bottom portions of b1-b2-b6 (yellow)
tures of the 5-HT3R (Hassaine et al., 2014), and the serotonin- and b7-b9-b10 (orange) form in both depicted subunits one highly
binding protein in complex with serotonin (Kesters et al., 2013) compiled community, they are spread into several groups in the
allowed us to establish through dedicated docking a highly agonist-bound receptor (cyan, orange, yellow, pink, and white in
reliable model on the binding of serotonin in its 5-HT3R. We Figure 3B). Ligand binding furthermore increased the size of the
found the bound neurotransmitter to be located in the highly black group consisting of loop B, loop C, and the top portion of
conserved ‘‘aromatic ligand-binding cage’’ of pLGICs, formed b7-b9-b10. These observations strongly indicate that the binding
by Y126, Y207, F199, W156, and W63, at the interface between of serotonin plays a fundamental role in the space of these regions,
adjacent receptor subunits (Hassaine et al., 2014) (Figures 1 inducing an overall looser coupling in the extracellular vestibule.
and 2), which was very similar to the structure found in the sero-
tonin-binding protein (Figure S1). There is no consensus on how Importance of Domain Tilting and Twisting for Receptor
many serotonin molecules are required to activate the 5-HT3R; Activation
between two and five serotonin molecules have been reported Our MD simulations revealed that activation of the 5-HT3R to-
(Rayes et al., 2009). As the pentameric serotonin-binding protein ward an open channel proceeds in multiple domain movements

Structure 24, 816–825, May 3, 2016 817


Figure 2. The Serotonin Binding Pocket of
the 5-HT3 Receptor
Residues lining the serotonin-binding pocket of the
5-HT3R are depicted, together with a conserved
residues logo based on multiple sequence align-
ment on all cationic Cys-loop receptors, including
four 5-HT3 serotonin receptors, 17 nicotinic acetyl-
choline receptors, and one zinc-activated ion
channel. The presence of conserved aromatic res-
idues including W63, Y126, W156, and Y207 imply
that p-p interactions play a central role for binding of
the neurotransmitter. Furthermore, the conserved
negatively charged residues at position 209 indicate
the importance of electrostatic interactions for the
binding of serotonin in the 5-HT3R and beyond for
ligand binding in the case of the other anionic Cys-
loop receptors. The size of the single letters in
the depicted sequence segments scales with its
conservation at this position. Conserved residues
participating in binding serotonin are highlighted by
dashed rectangles.

in each subunit (Figure 4). To visualize the overall motion within D4 = 5 –7 after 350 ns, subunit C after 500 ns, and subunits
the receptor we performed a principal components analysis D and E after 700 ns (Figure S5). By contrast, twist-angle fluc-
(PCA), extracting data for each subunit from 1,000 snapshots tuations in the apo form are distinctly smaller (1 –3 ).
evenly distributed over time in the MD simulations (Figure S3). Residues interaction correlation network analysis (Figure S3)
The channel-gating process visualized by the PCA is best shows that the motions of extracellular part of the receptor are
described as a superposition of radial tilting and tangential highly correlated with the M2-M3 transmembrane movements.
twisting of domains which results in a blooming-like opening of In addition, the motions within a subunit couple to the motions
the receptor channel (Figure 4; Movies S1 and S2). Several mo- of adjacent subunits (Figure S3). This indicates that the tilting
tifs including loop A, loop B, loop C, b1-b2 loop, Cys loop, M2-M3 and twisting in the extracellular receptor part leads to the gating
loop, and the M2 and MX helices follow these motions (Fig- of the TM channel. Our results are consistent with previous
ure 2A), which are also reflected by the root-mean-square fluctu- studies of pLGICs (Calimet et al., 2013) and GABAA receptor
ations (RMSF) over the backbones of the receptor’s subunits (Miller and Aricescu, 2014).
(Figure S4). Interestingly, the M2-M3 loops show by far the
largest RMSF amplitudes in subunits A and E; the M2 helices Pore Expansion during Receptor Activation
in their center show comparable RMSF amplitudes especially In the crystal structure (Hassaine et al., 2014), a constricted ring
in subunits B, C, and D. This demonstrates the direct coupling of hydrophobic amino acid residues in the central M2 region of
of the motions of M2-M3 loops with the fluctuations of the M2 the 5-HT3R blocks the pathway for translocation of water mole-
helices. The importance of ligand-induced loop motions in the cules and hydrated ions across the receptor (Figure 5). To allow
function of pLGICs has been noted by numerous structural and the passage of hydrated cations such as Na+, the hydrophobic
functional data, as summarized elsewhere (Changeux, 2014; constriction site has to function as a gate, which would be able
Corringer et al., 2012; Lummis, 2012; Thompson et al., 2010). to open and thus provide more internal space. We compared
The MD simulations have resolved in detail the temporal the flexibilities of residues in several conserved motifs with the
sequence of distinct local structural transitions in the 5-HT3R changes of the inner pore size of the receptor and found that
during agonist-induced activation. About 30 ns after agonist they are correlated during our MD simulations (Figure 5A). In
binding (during the equilibration period), the b-structured EC do- the extracellular vestibule of the receptor, structural changes of
mains have finished a rigid-body reorientation resulting in sub- loop A and loop B lead to considerable expansion (Figure 5C).
unit tilting of Dq = 2 –7 out toward helix M4 (Figures 4 and In the closed state of the 5-HT3R the extracellular vestibule
S5). The tilt angles are slightly different in each subunit but harbors a 4-Å wide constriction, which, after binding of seroto-
remain stable during the entire simulation time period, with fluc- nin, increased by 8 Å in diameter during the MD simulations
tuations of about ±1 around the mean. The outward tilting in- (Figures 5B and 5C). These changes expand to the interface con-
duces a substantial repositioning of the b1-b2 loops (5 Å) at necting the receptor’s EC domain via b1-b2 and the Cys loops
the interface between the receptor’s EC and transmembrane re- with M2-M3 loops of the TM domain (Figures 5A and 5B).
gions, and of loops C (3 Å), which move upward together with In consequence, the central part of the TM pore increases by
the other extracellular loops (Figure 4). However, in the MD sim- 2–3 Å in diameter after binding of serotonin (Figures 5C and S7).
ulations of the inactive apo form, the tilting angles only fluctuated The intracellular region of the 5-HT3R is composed of five MA
within 1 –2 (Figure S5). The structural transitions in the activated helices extending from the M4 transmembrane helices. They
EC domains are complemented by additional twisting motions form a closed vestibule in the non-activated receptor (Hassaine
(Figure 4B) with characteristic individual time traces. Subunits et al., 2014) (Figures 5A, 5B, and S8). The binding of serotonin
A and B reached a final stable state of a twist-angle change of induces an outward tilting and twisting of the M4 helices, which

818 Structure 24, 816–825, May 3, 2016


the volume of the EC vestibule expanded from 39 3 103 Å3 (initial
10-ns MD simulations) by 18%, to 46 3 103 Å3 (690–700-ns MD
simulations), the volume of the TM pore from 2.7 3 103 Å3 by
26%, to 3.4 3 103 Å3, and the volume of the IC vestibule from
7.8 3 103 Å3 by 32%, to 10.3 3 103 Å3.

Molecular Switching of the Hydrophobic Gates L260 and


V264 Open the Ion Channel
Ion-channel gating typically takes place on a submillisecond
timescale (Thompson et al., 2010). The extremely large simula-
tion system of the 5-HT3R, containing 186,500 atoms, cannot
be accessed by classical MD over milliseconds in a conventional
supercomputer (Dror et al., 2012). Instead, we used an advanced
free-energy sampling method, namely well-tempered metady-
namics, to explore the activation process of the 5-HT3R. Meta-
dynamics relies on a history-dependent potential acting on a
selected number of collective variables (CVs) and can substan-
tially speed up sampling and reconstruct the free-energy surface
associated with the CVs (Barducci et al., 2010). To choose the
correct CVs for efficient sampling of the gating process, we first
compared the crystal structure of 5-HT3R in the closed-channel
state (Hassaine et al., 2014) (PDB: 4PIR) with the bacterial pLGIC
homolog GLIC in the open-channel state (Nury et al., 2011) (PDB:
3P50). Two obvious differences were found in the structure
of these two channel proteins. (1) in the open-channel struc-
ture of GLIC in the M2 helix of each subunit, two hydrophobic
residues I233 and A237 adopt different rotamer structures
compared with the hydrophobic residues L260 and V264 at cor-
Figure 3. Simultaneous View of Community Residue Interaction responding positions in the 5-HT3R (Figure 6). (2) Such differ-
Network and 3D Structure of the Neurotransmitter Binding Pocket
ences influence water and ion penetration in the channel: in
of the 5-HT3R
(A) Apo form without serotonin.
GLIC the I233 residues form a pentagon with a cross-sectional
(B) With bound serotonin. Colors of the dots in the community analysis area of 100 Å2 (Figure 6A), while in the 5-HT3R the L260 resi-
correspond to colors in the indicated regions of the 3D protein structure. dues span a cross-sectional area of only 40 Å2. Similarly, the
A237 residues of GLIC span a regular pentagon of 118 Å2, but
couple to the MA helices to open 10-Å wide lateral portals in the the corresponding plane of the V264 residues in the 5-HT3R is
intracellular vestibule (Figure 5B, right panel; Figures 5C and S7), only 57 Å2. Therefore, we used side-chain dihedral angles c1 of
allowing the entrance of Cl ions (Movie S3). The Cl ions are first L260 and V264 in subunit D as two CVs for well-tempered meta-
captured by the positively charged residues K415, R424, and dynamics simulations.
R428 on the intracellular surface of the receptor, and afterward Our simulations revealed two characteristic conformations
move toward R416 inside the intracellular vestibule (Figures S6 for both L260 and V264, a closed state I and an open state II
and S7). Interestingly, even after binding Cl ions, the pKa values (Figure 6B). The free energies are 17 kJ/mol for state I and
of these residues remain in the range of 11–13, implying that they 33 kJ/mol for state II. In the closed state I, a hydrophobic
are still protonated. In addition, movements of the ends of the constriction is formed by adjacent pentameric rings of residues
MA helices form a hydrophobic tunnel, which becomes partly L260, V264, and I268, respectively (Figures 7A and S8A). In the
accessible by water molecules (Figure 5B). open-channel state II (Figures 7B and S8A), a continuous water
In detail, the following conformational changes in the M4-MA channel is formed, crossing the hydrophobic constriction site
helix lead to these overall structural changes in the intracellular re- (Movie S4). Residues interaction network analysis on M2 channel
gion of the receptor (Figure S8). The initial outward bending of the gating (residues 250–271) show that in the closed state I, all
M4 helix is induced by the local helix bending around residue D434 five submits interacted with each other almost identically (Fig-
which in turn breaks the salt bridge D434-R251 within one subunit ure S8B, left panel). However, in the active receptor state II,
and creates a new one, but now between D434 of one subunit and chains A, B, and C are much closer to each other than to subunits
R306 of the neighboring subunit. The other characteristic changes D and E which, on the other hand, are located closely together
occur in the bending of the MA helix around residues D417 and (Figure S8, right panel). Consequently, much larger space is
E418, leading to a break of the inter-subunit salt bridge D417- created for movement of water molecules (Movie S4).
K415 followed by a 20 torsion of the MA helix in this region and
the formation of a new inter-subunit salt bridge D417-R416. These DISCUSSION
conformational changes occur within all subunits.
All these structural transitions induce substantial volume Here, we compare our findings on the activation process of the
changes within different internal pore regions of the receptor: 5-HT3R with published data on pLGICs.

Structure 24, 816–825, May 3, 2016 819


Figure 4. Domain Movements and Hydro-
phobic Pore of 5-HT3R
(A) The normal modes of collective motions were
calculated by principal component analysis. The
sizes of the red arrows scale with the amplitudes of
the particular motions.
(B) Global twisting of the receptor’s extracellular
domain by D4 = 5 –7 , indicated by overlaying the
structure of the 5-HT3R before (bold) and after
(faint) binding of serotonin.
(C) Tilting of the b sandwich by Dq = 5 –7 .

diameter, the hydrophobic amino


acids have to perform a confor-
mational switch to drastically
reduce steric hindrance and thus
open the hydrophobic gate for
(1) Ligand binding. Although the general architecture of the water passage. This might represent a general mecha-
ligand-binding region of pLGICs has been established nism for activating pLGICs.
as a result of numerous functional, structural, and (4) Conformational changes in the intracellular receptor
modeling studies (Auerbach, 2014; Changeux, 2014; Cor- domain. Vertebrate pLGICs possess an intracellular
ringer et al., 2012; Hassaine et al., 2014; Lummis, 2012; domain of 70–150 residues important for receptor traf-
Smart and Paoletti, 2012; Thompson et al., 2010), our ficking, clustering at the synapse, and gating (Bouzat
MD simulations on the 5-HT3R delivered the first atomic et al., 1994; Thompson et al., 2010; Zuber and Unwin,
detailed description of how an activating neurotransmitter 2013). In the 5-HT3R this domain is a major determinant
binds to a synaptic pLGIC based on relevant structural of channel conductance (Kelley et al., 2003; Peters
data (Hassaine et al., 2014; Kesters et al., 2013). et al., 2010). Here the MA helices extend from the M4 he-
(2) Conformational changes in the EC region coupling to the lices to the intracellular side, creating a closed vestibule
TM channel. MD simulations have proposed various (Hassaine et al., 2014) Potential lateral portals between
models on allosteric transitions in non-mammalian LGICs MA helices are blocked by the post-M3 loop leaving
(Changeux, 2014). For the GluCl receptor from C. elegans, only a 3.3-Å narrow tunnel. Further, the MA helical bundle
MD simulations indicated that channel activation is tightens into a 17-Å-long, 4.2-Å narrow hydrophobic pore.
induced by opposite movements of the extracellular Therefore, the crystal structure of the non-activated
versus the TM domains coupled via b1-b2 and M2-M3 5-HT3R does not offer an exit pathway for the ions. Our
loops (Calimet et al., 2013). Despite the similarities with MD simulations show that the structural changes in the
the 5-HT3R, the two cases are not directly comparable. TM region during channel opening couple to the intracel-
The GluCl channel could only be stabilized in an open lular MA helices, and induce an opening of lateral portals
state when ivermectin, a synthetic small-molecule allo- at the intracellular vestibule and the formation of a hydro-
steric modulator, was bound at the interfaces between phobic pore at the end of the helical bundle. This is the
TM helices of adjacent subunits, potentiating neurotrans- first mechanistic description of how ions are released
mitter binding (Althoff et al., 2014). We, however, have from neuronal pLGICs into the intracellular space.
elucidated the direct activation of a neuronal pLGIC by
its neurotransmitter, which is the actual relevant process In summary, we combined classical all-atom, long-timescale
in fast neuronal signal transmission. MD simulations with advanced free-energy sampling methods
(3) Opening the TM channel. A hallmark of pLGICs is the to explore the activation process of the 5-HT3R integrated in a
hydrophobic constriction zone in the center of the TM lipid bilayer (Figure 8). Our MD simulation study is comparable
channel which, according to the hydrophobic gating with a stopped-flow experiment: We found that the fast binding
concept, is dehydrated in the closed-channel state of serotonin to the ‘‘aromatic ligand-binding cage’’ exerts a
and becomes hydrated when the pore diameter en- structural stress to the receptor, which relaxes by multiple struc-
larges beyond a threshold size, allowing water mole- tural transitions toward its activated state. First, the ligand-bind-
cules and finally hydrated ions to pass (Aryal et al., ing cage undergoes structural relaxation processes to form
2015; Beckstein and Sansom, 2006; Zhu and Hummer, optimal interactions with and integration of the ligand, which
2012). Until now an atomistic description of the gating finally induces distinct conformational transitions of W156
process was missing, which we here delivered for the sequentially in all subunits. The structural reorganization of the
5-HT3R. We found that an increase in the pore diameter ligand-binding cage then triggers overall movements in each of
in the hydrophobic gate is necessary but not sufficient the extracellular subunits, first a 6 tilting followed by 6 twisting,
for hydration and opening of the gate. Our MD simula- which appears as a blooming/opening of the entire EC domain
tions revealed that in addition to increasing the pore expanding substantially the EC inner pore volume. The EC

820 Structure 24, 816–825, May 3, 2016


Figure 5. Flexibility Pattern of a 5-HT3R Sub-
unit and Cross-Sectional View of the 5-HT3R
Channel before and after Activation by Sero-
tonin
(A) Flexibility of particular regions in a single submit
are indicated by wider radius of tube and a red
color. The most flexible regions are colored red,
medium flexible regions white, and least flexible
regions green.
(B) Cross section of 5-HT3R in the non-activated
state before binding of serotonin (left panel) and after
binding of serotonin reaching an open, water-filled
channel state (right panel). Hydrophobic regions are
colored in orange and hydrophilic regions in cyan.
(C) Pore diameter along z axis of the 5-HT3R for the
non-activated receptor state (two independent
MD simulations, black and cyan) and for the open,
water-filled channel state (two independent MD
simulations, red and blue).

rigid-body motions couple via the b1-b2 and the Cys loops to the using the OPLS_2005 force field. The ionization state of serotonin was calcu-
M2-M3 loops of the TM region. In consequence, the receptor’s lated with the Epik (Greenwood et al., 2010) tool employing Hammett and Taft
methods together with ionization and tautomerization tools.
TM core first expands by an overall outward movement of the
TM helices. The expansion enables the side chains of V264
Protein-Ligand Docking
and L260 of the hydrophobic gate to change conformation to The docking was performed using Glide (Friesner et al., 2004). Based on the
an open-channel structure, allowing water molecules to translo- crystal structure of the 5-HT3R (Hassaine et al., 2014), serotonin was docked
cate from the extracellular to the intracellular vestibule. Later, into the aromatic cage between two subunits close to W156 in a pose as found
beyond our present simulation time, the receptor might enlarge in the crystal structure of a serotonin-binding protein in complex with serotonin
the water channel further for the passage of Na+ ions across (5HTBP; PDB: 2YMD) (Kesters et al., 2013). Cubic boxes centered on the
ligand mass center with a radius of 8 Å for all ligands defined the docking bind-
the constriction site. The structural changes in the TM region
ing regions. Flexible ligand docking was executed for all structures. Twenty
finally couple to the MA helix, inducing a drastic expansion of poses per ligand out of 20,000 were included in the post-docking energy mini-
the intracellular vestibule together with opening of side portals mization. We also placed the initial docking pose at random pose and obtained
for the exit of the cations leaving the TM ion channel and the a similar binding mode as reported previously (Kesters et al., 2013) (Figure S1).
entrance of anions. Since the top three were found to be identical with each other, the best scored
As our simulations are based on the first high-resolution struc- pose for the ligand, which was similar to that of previous work (Kesters et al.,
2013), was chosen as the initial structure for MD simulations.
ture of a mammalian pLGIC and describe the activation process
of the whole receptor from the extracellular to the intracellular re-
3D Multiple Sequence Alignment
gion, they are of direct relevance for understanding synaptic The 3D multiple sequence alignment was done in Strap (Gille et al., 2014),
signal transmission. a JAVA-based tool. It first predicts the secondary structure of each sequence
and aligns them with the specified PDB file afterward. The conservation of
EXPERIMENTAL PROCEDURES each motif was then submitted to WebLogo for visualization (Crooks et al.,
2004).
Missing Loop Filling and Refinements
Since the small M2-M3 loop in each subunit was not resolved in the X-ray Molecular Dynamics Simulations
structure of the 5-HT3R (Hassaine et al., 2014), the loop refinement protocol We modeled the protein, POPC lipids, water molecules, and ions using the
in Modeller (Eswar et al., 2007) V9.10 was used to complete and refine this newest CHARMM 36 force field (Klauda et al., 2010), and the ligand using
structural region. A total of 5,000 loops were generated, and the conformation the CHARMM CGenFF small-molecule force field (Vanommeslaeghe et al.,
with the lowest Discrete Optimized Protein Energy score was chosen for con- 2012). The membrane system was built by the g_membed (Wolf et al., 2010)
structing the starting receptor structure. tool in Gromacs (Pronk et al., 2013) V4.6.5 with the receptor crystal structure
pre-aligned in the OPM (Orientations of Proteins in Membranes) database (Lo-
Protein Structure Preparation mize et al., 2011). Pre-equilibrated 234 POPC lipids coupled with 40,760 TIP3P
All protein models were prepared in the Schrödinger software suite using the water molecules and 0.15 M NaCl in a box 100 Å 3 100 Å 3 180 Å were used
OPLS_2005 force field. Five nanobodies were removed from the crystal struc- for building the protein/membrane system. This resulted in 186,500 atoms in
ture (Hassaine et al., 2014) (PDB: 4PIR). Hydrogen atoms were added to the the simulating system in all. The ligand geometry was submitted to the
repaired crystal structure of the 5-HT3R to reflect the physiological pH (7.0) us- Gaussian 09 program (Frisch et al., 2009) for optimization at Hartree-Fock
ing the PROPKA (Sondergaard et al., 2011) tool in Protein Preparation tool in 6-31G* level when generating force-field parameters. The system was gradu-
Maestro to obtain the optimized hydrogen-bond network. For this, the con- ally heated from 0 K to 310 K followed by 1 ns initial equilibration at a constant
strained energy minimizations were performed on the full-atomic models, volume and temperature set to 310 K. Next, an additional 30-ns constrained
with an atom allowed motion of 0.4 Å excluding hydrogens, which were free equilibration was performed at a constant pressure and temperature (310 K,
to move. 1 bar) while the force constant on each atom was trapped off gradually from
10 kcal/mol to 0 kcal/mol. All covalent bond lengths to hydrogen atoms
Ligand Structure Preparation were constrained with M-SHAKE. van der Waals and short-range electrostatic
The structure of the neurotransmitter serotonin was obtained from the interactions were cut off at 10 Å. Long-range electrostatic interactions
PubChem (Wang et al., 2012) online database. The LigPrep module in the were computed by the Particle Mesh Ewald summation scheme. 2 3 700-ns
Schrödinger 2014 software suite was introduced for geometric optimization MD simulations were produced for both the apo form of 5-HT3R and

Structure 24, 816–825, May 3, 2016 821


Figure 6. The Hydrophobic Channel Gate
(A) Comparison of dimensions of the hydrophobic constriction sites (channel
gates) of the 5-HT3R in the closed-channel state (PDB: 4PIR) with that of GLIC
in the open-channel state (PDB: 3P50). Shown are cross sections of the M2 TM
helix of 5-HT3R (green) and GLIC (gray). Five L260 and five V264 residues
(green) in the 5-HT3R were superimposed with corresponding residues I233
Figure 7. The hydrophobic Gate of the 5-HT3 Receptor
and A237 (purple) in GLIC.
(A) The cross section perpendicular to the membrane plane through the A and
(B) Left: free-energy changes in the conformational space of the dihedral an-
D subunits of the pentameric 5-HT3R (left panel) in the crystal structure in the
gles c1 of L260 and V264 show two distinct minima of the closed (I) and open
absence of serotonin reveals a closed hydrophobic gate formed by the L260
(II) channel state in the 5-HT3R. Right: distinct side-chain conformations of
and V264 residues (cross sections at L260 and V264 planes, right panel)
V264 (top) and L260 (bottom) of the 5-HT3R in the closed-channel state before
preventing the entrance of water molecules. Amino acid residues of the M2
binding of serotonin (I, green) and in the activated channel state after binding of
helices along the channel are indicated.
serotonin (II, cyan).
(B) A continuous water channel was formed by opening the gate of 5-HT3R by
movements of side chains of L260 and V264.
agonist-bound receptor. The MD simulation results were analyzed in Gromacs
(Pronk et al., 2013) and VMD (Humphrey et al., 1996). Bio3D (Grant et al., 2006)
was introduced for principal component analysis, and the pore diameters were central channel, is also needed. In addition, the user specifies a vector v in
calculated by HOLE (Smart et al., 1996). Potential Cl ions entrance pathways the direction of the channel axis (referred to as the channel direction vector).
were predicated by Caver3.0 (Chovancova et al., 2012) command line version. The program reads atoms from the pdb file and sets up a van der Waals radius
All MD simulations were conducted in Gromacs V4.6.5. for each (various sets are available). The maximum radius R(p) of a sphere
centered at a point p without overlap with the van der Waals surface of any
Helix Bending atom is calculated by
The bending of a helix with respect to the channel axis was calculated in
RðpÞ = minNi =atom
11 ½jxi  pj  vdWi ; (Equation 1)
Bendix (Dahl et al., 2012) with default settings. Helices were categorized as
linear, curved, or kinked according to their shapes (Bansal et al., 2000). In where xi is the position of atom number i, vdWi its van der Waals radius, and
the first step we created helical axes for each helix, based on a previously Natom the total number of atoms. The radius R(p) can be regarded as an objec-
developed algorithm. Axes are calculated only for helices of a minimal length tive function of the point p. By using Monte Carlo simulated annealing, adjust-
of four residues, and a local helix axis was calculated for a residue, based ing p, the radius of the sphere can be maximized within the pore of the channel.
on its backbone atomic coordinates and those of the three residues succeed- In all cases p is kept on a plane normal to the channel direction vector v (Smart
ing it in the sequence. Accordingly, axes are not calculated for the three et al., 1996).
carboxy-terminal residues of the helix. Using these axes the helix is assigned The grid filling the pore space generated by HOLE (Smart et al., 1996) was
to a linear, curved, or kinked geometry, according to bending angle criteria be- imported into UCSF Chimera for volume calculation. The ‘‘Measure Volume
tween successive axes (Dalton et al., 2003; Lee et al., 2007). and Area’’ tool reports the total surface area and enclosed volume of a surface
model. These tools compute surface area and enclosed volume from surface
Hydrophobicity Surface Calculation triangles rather than analytically (Goddard et al., 2007; Pettersen et al., 2004).
The hydrophobicity surface of 5-HT3R was calculated by UCSF Chimera: to
each amino acid residue a hydrophobicity value is assigned according to the Metadynamics Simulations
hydrophobicity scale of Kyte and Doolittle (1982). Free-energy profiles of the systems were calculated using well-tempered
metadynamics in Gromacs (Pronk et al., 2013) V4.6.5 with Plumed (Bonomi
Pore Diameter and Volume Calculation et al., 2009) V1.3 algorithm. Metadynamics adds a history-dependent potential
The HOLE (Smart et al., 1996) version 2.0 method was used for pore diameter V(s,t) to accelerate sampling of the specific CVs s(s1,s2,.,sm) (Barducci et al.,
calculation. The program requires the coordinates of the ion channel of interest 2008). V(s,t) is usually constructed as the sum of multiple Gaussians centered
in Brookhaven PDB format. An initial point p, which lies anywhere within the along the trajectory of the CVs (Equation 2).

822 Structure 24, 816–825, May 3, 2016


Correlation Network Analysis
Correlated atomic fluctuations of a particular receptor state were character-
ized as reported elsewhere (Grant et al., 2006; Scarabelli and Grant, 2014;
Skjaerven et al., 2014) using Bio3DT. The network nodes represent residues,
which are connected through edges weighted by their constituent atomic cor-
relation values. Community analysis and node centrality with Bio3D and sub-
optimal path calculation with the WISP software (Van Wart et al., 2014) were
performed on each network to characterize network properties and identify
residues involved in the dynamic coupling of distal sites. The parameters for
the suboptimal path analysis included input source and sink nodes, as well
as the total number of paths to be calculated. The latter parameter was set
to 500 paths, which was found to yield converged results in all cases (Scara-
belli and Grant, 2014).

SUPPLEMENTAL INFORMATION

Supplemental Information includes eight figures and four movies and can be
found with this article online at http://dx.doi.org/10.1016/j.str.2016.03.019.

AUTHOR CONTRIBUTIONS

S.Y. and H.V. initialized the project. S.Y., S.F., and H.V. designed the experi-
ments. S.Y. performed the MD simulations and analyzed the results. S.Y.,
S.F., and H.V. wrote the manuscript.

ACKNOWLEDGMENTS

The major part of computing was done at the Shanghai Supercomputer Cen-
ter. S.F. was funded by the National Center of Science, Poland (grant 2011/03/
B/NZ1/03204). H.V. was supported by the Swiss National Science Foundation
(grant 31003A-133141), the European Community (Project SynSignal, grant
FP7-KBBE-2013-613879), and internal funds of the EPFL.

Figure 8. Gating Mechanism of 5-HT3R Received: October 20, 2015


Side-view cross section through the 5-HT3R shows two protein subunits with Revised: January 27, 2016
two bound serotonin molecules near the W156 residues in the aromatic ligand- Accepted: March 6, 2016
binding cages. The global quaternary motions (black arrows) induce (1) the Published: April 21, 2016
entrance of water molecules (red) into the central transmembrane channel
across the hydrophobic gate represented by V264 and L260, (2) a deeper REFERENCES
penetration of sodium ions (blue), and (3) the entrance of chloride ions (green)
into the intracellular vestibule via lateral portals. Althoff, T., Hibbs, R.E., Banerjee, S., and Gouaux, E. (2014). X-ray structures of
GluCl in apo states reveal a gating mechanism of Cys-loop receptors. Nature
! 512, 333–337.
X
n X
n Y
m
½si ðtÞ  si ðtj Þ2
Vðs; tÞ = Gðs; tj Þ = wj exp  (Equation 2) Aryal, P., Sansom, M.S., and Tucker, S.J. (2015). Hydrophobic gating in ion
j=1 j=1 i=1
2s2 channels. J. Mol. Biol. 427, 121–130.
Auerbach, A. (2014). Agonist activation of a nicotinic acetylcholine receptor.
Periodically, during the simulation another Gaussian potential, whose loca-
Neuropharmacology 96, 150–156.
tion is dictated by the current values of the CVs, is added to V(s, t) (Li et al.,
2013). In our simulations, the dihedral angles of W156, c1 and c2, were as- Bansal, M., Kumar, S., and Velavan, R. (2000). HELANAL: a program to char-
signed as the CVs s1 and s2, while the width of Gaussians, s, was set to 5 . acterize helix geometry in proteins. J. Biomol. Struct. Dyn. 17, 811–819.
Similarly, in the second metadynamisc simulation, the c1 angle of L260 and Barducci, A., Bussi, G., and Parrinello, M. (2008). Well-tempered metadynam-
V264 were used as CVs. The time interval, t, was 0.09 ps. Well-tempered ics: a smoothly converging and tunable free-energy method. Phys. Rev. Lett.
metadynamics involves adjusting the height, wj, in a manner that depended 100, 020603.
on V(s, t) where the initial height of Gaussians w was 0.03 kcal/mol, the simu- Barducci, A., Bonomi, M., and Parrinello, M. (2010). Linking well-tempered
lation temperature was 310 K, and the sampling temperature DT was 298 K. metadynamics simulations with experiments. Biophys. J. 98, L44–L46.
The convergence of our simulations was judged by the free-energy difference Beckstein, O., and Sansom, M.S. (2006). A hydrophobic gate in an ion channel:
between states X and A at 10-ns intervals. Once the resulting data remained the closed state of the nicotinic acetylcholine receptor. Phys. Biol. 3, 147–159.
stable over time, the simulation was considered as converged. Each metady-
Bill, R.M., Henderson, P.J., Iwata, S., Kunji, E.R., Michel, H., Neutze, R.,
namics simulation lasted 120 ns, and the results were analyzed upon
Newstead, S., Poolman, B., Tate, C.G., and Vogel, H. (2011). Overcoming barriers
convergence.
to membrane protein structure determination. Nat. Biotechnol. 29, 335–340.
Residues Interaction Network Analysis Bocquet, N., Nury, H., Baaden, M., Le Poupon, C., Changeux, J.P., Delarue,
StructureViz (Morris et al., 2007) and RINalyzer (Doncheva et al., 2012), plugin M., and Corringer, P.J. (2009). X-ray structure of a pentameric ligand-gated
modules in Cytoscape (Shannon et al., 2003) were used for residues interac- ion channel in an apparently open conformation. Nature 457, 111–114.
tion network (RIN) analysis (Scarabelli and Grant, 2014). These modules allow Bonomi, M., Branduardi, D., Bussi, G., Camilloni, C., Provasi, D., Raiteri, P.,
the interactive analysis of a RIN in a 2D representation of amino acid residues Donadio, D., Marinelli, F., Pietrucci, F., Broglia, R.A., et al. (2009). PLUMED:
as nodes and their non-covalent interactions as edges together with the cor- a portable plugin for free-energy calculations with molecular dynamics.
responding 3D protein structures in UCSF Chimera (Pettersen et al., 2004). Comput. Phys. Commun. 180, 1961–1972.

Structure 24, 816–825, May 3, 2016 823


Bouzat, C., Bren, N., and Sine, S.M. (1994). Structural basis of the different Kelley, S.P., Dunlop, J.I., Kirkness, E.F., Lambert, J.J., and Peters, J.A. (2003).
gating kinetics of fetal and adult acetylcholine receptors. Neuron 13, 1395– A cytoplasmic region determines single-channel conductance in 5-HT3 recep-
1402. tors. Nature 424, 321–324.
Calimet, N., Simoes, M., Changeux, J.P., Karplus, M., Taly, A., and Cecchini, Kesters, D., Thompson, A.J., Brams, M., van Elk, R., Spurny, R., Geitmann, M.,
M. (2013). A gating mechanism of pentameric ligand-gated ion channels. Proc. Villalgordo, J.M., Guskov, A., Danielson, U.H., Lummis, S.C., et al. (2013).
Natl. Acad. Sci. USA 110, E3987–E3996. Structural basis of ligand recognition in 5-HT3 receptors. EMBO Rep. 14,
Changeux, J.-P. (2014). Protein dynamics and the allosteric transitions of pen- 49–56.
tameric receptor channels. Biophys. Rev. 6, 311–321. Klauda, J.B., Venable, R.M., Freites, J.A., O’Connor, J.W., Tobias, D.J.,
Chovancova, E., Pavelka, A., Benes, P., Strnad, O., Brezovsky, J., Kozlikova, Mondragon-Ramirez, C., Vorobyov, I., MacKerell, A.D., and Pastor, R.W.
B., Gora, A., Sustr, V., Klvana, M., Medek, P., et al. (2012). CAVER 3.0: a tool for (2010). Update of the CHARMM all-atom additive force field for lipids: valida-
the analysis of transport pathways in dynamic protein structures. PLoS tion on six lipid types. J. Phys. Chem. B 114, 7830–7843.
Comput. Biol. 8, e1002708. Kudryashev, M., Castano-Diez, D., Deluz, C., Hassaine, G., Grasso, L., Graf-
Corringer, P.J., Poitevin, F., Prevost, M.S., Sauguet, L., Delarue, M., and Meyer, A., Vogel, H., and Stahlberg, H. (2016). The Structure of the mouse se-
Changeux, J.P. (2012). Structure and pharmacology of pentameric receptor rotonin 5-HT3 receptor in lipid vesicles. Structure 24, 165–170.
channels: from bacteria to brain. Structure 20, 941–956. Kyte, J., and Doolittle, R.F. (1982). A simple method for displaying the hydro-
pathic character of a protein. J. Mol. Biol. 157, 105–132.
Crooks, G.E., Hon, G., Chandonia, J.M., and Brenner, S.E. (2004). WebLogo:
a sequence logo generator. Genome Res. 14, 1188–1190. Lee, H.S., Choi, J., and Yoon, S. (2007). QHELIX: a computational tool for the
improved measurement of inter-helical angles in proteins. Protein J. 26,
daCosta, C.J., and Baenziger, J.E. (2013). Gating of pentameric ligand-gated
556–561.
ion channels: structural insights and ambiguities. Structure 21, 1271–1283.
Lemoine, D., Jiang, R., Taly, A., Chataigneau, T., Specht, A., and Grutter, T.
Dahl, A.C.E., Chavent, M., and Sansom, M.S.P. (2012). Bendix: intuitive helix
(2012). Ligand-gated ion channels: new insights into neurological disorders
geometry analysis and abstraction. Bioinformatics 28, 2193–2194.
and ligand recognition. Chem. Rev. 112, 6285–6318.
Dalton, J.A., Michalopoulos, I., and Westhead, D.R. (2003). Calculation of helix
Li, J., Jonsson, A.L., Beuming, T., Shelley, J.C., and Voth, G.A. (2013). Ligand-
packing angles in protein structures. Bioinformatics 19, 1298–1299.
dependent activation and deactivation of the human adenosine A(2A) recep-
Doncheva, N.T., Assenov, Y., Domingues, F.S., and Albrecht, M. (2012). tor. J. Am. Chem. Soc. 135, 8749–8759.
Topological analysis and interactive visualization of biological networks and
Lomize, A.L., Pogozheva, I.D., and Mosberg, H.I. (2011). Anisotropic solvent
protein structures. Nat. Protoc. 7, 670–685.
model of the lipid bilayer. 2. Energetics of insertion of small molecules, pep-
Dror, R.O., Dirks, R.M., Grossman, J.P., Xu, H., and Shaw, D.E. (2012). tides, and proteins in membranes. J. Chem. Inf. Model. 51, 930–946.
Biomolecular simulation: a computational microscope for molecular biology.
Lummis, S.C. (2012). 5-HT(3) receptors. J. Biol. Chem. 287, 40239–40245.
Annu. Rev. Biophys. 41, 429–452.
Miller, P.S., and Aricescu, A.R. (2014). Crystal structure of a human GABAA re-
Eswar, N., Webb, B., Marti-Renom, M.A., Madhusudhan, M.S., Eramian, D.,
ceptor. Nature 512, 270–275.
Shen, M.Y., Pieper, U., and Sali, A. (2007). Comparative protein structure
modeling using MODELLER. Curr. Protoc. Protein Sci. Chapter 2. Unit 2 9. Morris, J.H., Huang, C.C., Babbitt, P.C., and Ferrin, T.E. (2007). structureViz:
linking cytoscape and UCSF chimera. Bioinformatics 23, 2345–2347.
Friesner, R.A., Banks, J.L., Murphy, R.B., Halgren, T.A., Klicic, J.J., Mainz,
D.T., Repasky, M.P., Knoll, E.H., Shelley, M., Perry, J.K., et al. (2004). Glide: Nury, H., Poitevin, F., Van Renterghem, C., Changeux, J.P., Corringer, P.J.,
a new approach for rapid, accurate docking and scoring. 1. Method and Delarue, M., and Baaden, M. (2010). One-microsecond molecular dynamics
assessment of docking accuracy. J. Med. Chem. 47, 1739–1749. simulation of channel gating in a nicotinic receptor homologue. Proc. Natl.
Acad. Sci. USA 107, 6275–6280.
Frisch, M.J., Trucks, G.W., Schlegel, H.B., Scuseria, G.E., Robb, M.A.,
Cheeseman, J.R., Scalmani, G., Barone, V.M.B., Petersson, G.A., Nakatsuji, Nury, H., Van Renterghem, C., Weng, Y., Tran, A., Baaden, M., Dufresne, V.,
H., Caricato, M., et al. (2009). Gaussian 09, Revision A.1 (Gaussian, Inc). Changeux, J.P., Sonner, J.M., Delarue, M., and Corringer, P.J. (2011). X-ray
structures of general anaesthetics bound to a pentameric ligand-gated ion
Gille, C., Fahling, M., Weyand, B., Wieland, T., and Gille, A. (2014). Alignment-
channel. Nature 469, 428–431.
Annotator web server: rendering and annotating sequence alignments.
Nucleic Acids Res. 42, W3–W6. Peters, J.A., Cooper, M.A., Carland, J.E., Livesey, M.R., Hales, T.G., and
Lambert, J.J. (2010). Novel structural determinants of single channel conduc-
Goddard, T.D., Huang, C.C., and Ferrin, T.E. (2007). Visualizing density maps
tance and ion selectivity in 5-hydroxytryptamine type 3 and nicotinic acetyl-
with UCSF Chimera. J. Struct. Biol. 157, 281–287.
choline receptors. J. Physiol. 588, 587–596.
Grant, B.J., Rodrigues, A.P., ElSawy, K.M., McCammon, J.A., and Caves, L.S.
Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M.,
(2006). Bio3d: an R package for the comparative analysis of protein structures.
Meng, E.C., and Ferrin, T.E. (2004). UCSF Chimera—a visualization system
Bioinformatics 22, 2695–2696.
for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612.
Greenwood, J.R., Calkins, D., Sullivan, A.P., and Shelley, J.C. (2010). Towards
Pronk, S., Pall, S., Schulz, R., Larsson, P., Bjelkmar, P., Apostolov, R., Shirts,
the comprehensive, rapid, and accurate prediction of the favorable tautomeric
M.R., Smith, J.C., Kasson, P.M., van der Spoel, D., et al. (2013). GROMACS
states of drug-like molecules in aqueous solution. J. Comput. Aided Mol. Des.
4.5: a high-throughput and highly parallel open source molecular simulation
24, 591–604.
toolkit. Bioinformatics 29, 845–854.
Hassaine, G., Deluz, C., Grasso, L., Wyss, R., Tol, M.B., Hovius, R., Graff, A., Rayes, D., De Rosa, M.J., Sine, S.M., and Bouzat, C. (2009). Number and
Stahlberg, H., Tomizaki, T., Desmyter, A., et al. (2014). X-ray structure of the locations of agonist binding sites required to activate homomeric Cys-loop re-
mouse serotonin 5-HT3 receptor. Nature 512, 276–281. ceptors. J. Neurosci. 29, 6022–6032.
Hilf, R.J., and Dutzler, R. (2008). X-ray structure of a prokaryotic pentameric Sauguet, L., Shahsavar, A., and Delarue, M. (2014). Crystallographic studies of
ligand-gated ion channel. Nature 452, 375–379. pharmacological sites in pentameric ligand-gated ion channels. Biochim.
Hilf, R.J., and Dutzler, R. (2009). Structure of a potentially open state of a pro- Biophys. Acta 1850, 511–523.
ton-activated pentameric ligand-gated ion channel. Nature 457, 115–118. Scarabelli, G., and Grant, B.J. (2014). Kinesin-5 allosteric inhibitors uncouple
Humphrey, W., Dalke, A., and Schulten, K. (1996). VMD: visual molecular dy- the dynamics of nucleotide, microtubule, and neck-linker binding sites.
namics. J. Mol. Graph. Model. 14, 33–38. Biophys. J. 107, 2204–2213.
Hurst, R., Rollema, H., and Bertrand, D. (2013). Nicotinic acetylcholine recep- Schmauder, R., Kosanic, D., Hovius, R., and Vogel, H. (2011). Correlated
tors: from basic science to therapeutics. Pharmacol. Ther. 137, 22–54. optical and electrical single-molecule measurements reveal conformational

824 Structure 24, 816–825, May 3, 2016


diffusion from ligand binding to channel gating in the nicotinic acetylcholine re- Unwin, N. (2013). Nicotinic acetylcholine receptor and the structural basis of
ceptor. Chembiochem 12, 2431–2434. neuromuscular transmission: insights from Torpedo postsynaptic mem-
branes. Q. Rev. Biophys. 46, 283–322.
Sethi, A., Eargle, J., Black, A.A., and Luthey-Schulten, Z. (2009). Dynamical
networks in tRNA:protein complexes. Proc. Natl. Acad. Sci. USA 106, 6620– Van Wart, A.T., Durrant, J., Votapka, L., and Amaro, R.E. (2014). Weighted im-
6625. plementation of suboptimal paths (WISP): an optimized algorithm and tool for
dynamical network analysis. J. Chem. Theory Comput. 10, 511–517.
Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T., Ramage, D., Amin,
N., Schwikowski, B., and Ideker, T. (2003). Cytoscape: a software environment Vanommeslaeghe, K., Raman, E.P., and MacKerell, A.D., Jr. (2012).
for integrated models of biomolecular interaction networks. Genome Res. 13, Automation of the CHARMM General Force Field (CGenFF) II: assignment of
2498–2504. bonded parameters and partial atomic charges. J. Chem. Inf. Model. 52,
3155–3168.
Skjaerven, L., Yao, X.Q., Scarabelli, G., and Grant, B.J. (2014). Integrating
Walstab, J., Rappold, G., and Niesler, B. (2010). 5-HT(3) receptors: role in dis-
protein structural dynamics and evolutionary analysis with Bio3D. BMC
ease and target of drugs. Pharmacol. Ther. 128, 146–169.
Bioinformatics 15, 399.
Wang, Y.L., Xiao, J.W., Suzek, T.O., Zhang, J., Wang, J.Y., Zhou, Z.G., Han,
Smart, T.G., and Paoletti, P. (2012). Synaptic neurotransmitter-gated recep-
L.Y., Karapetyan, K., Dracheva, S., Shoemaker, B.A., et al. (2012).
tors. Cold Spring Harb. Perspect. Biol. 4, http://dx.doi.org/10.1101/cshper-
PubChem’s bioassay database. Nucleic Acids Res. 40, D400–D412.
spect.a009662.
Wolf, M.G., Hoefling, M., Aponte-Santamaria, C., Grubmuller, H., and
Smart, O.S., Neduvelil, J.G., Wang, X., Wallace, B.A., and Sansom, M.S. Groenhof, G. (2010). g_membed: efficient insertion of a membrane protein
(1996). HOLE: a program for the analysis of the pore dimensions of ion channel into an equilibrated lipid bilayer with minimal perturbation. J. Comput.
structural models. J. Mol. Graph. 14, 354–360, 376. Chem. 31, 2169–2174.
Sondergaard, C.R., Olsson, M.H.M., Rostkowski, M., and Jensen, J.H. (2011). Zhu, F., and Hummer, G. (2012). Drying transition in the hydrophobic gate of
Improved treatment of ligands and coupling effects in empirical calculation the GLIC channel blocks ion conduction. Biophys. J. 103, 219–227.
and rationalization of pK(a) values. J. Chem. Theory Comput. 7, 2284–2295.
Zuber, B., and Unwin, N. (2013). Structure and superorganization of acetylcho-
Thompson, A.J., Lester, H.A., and Lummis, S.C. (2010). The structural basis of line receptor-rapsyn complexes. Proc. Natl. Acad. Sci. USA 110, 10622–
function in Cys-loop receptors. Q. Rev. Biophys. 43, 449–499. 10627.

Structure 24, 816–825, May 3, 2016 825

You might also like