Interaction simulation of hERG K+ channel with its specific BeKm-1 peptide: insights into the selectivity of molecular recognition.

Abstract

Potassium channels show a huge variability in the affinity when recognizing enormous bioactive peptides, and the elucidation of their recognition mechanism remains a great challenge due to an undetermined peptide-channel complex structure. Here, we employed combined computation methods to study the specific binding of BeKm-1 peptide to the hERG potassium channel, which is an essential determinant of the long-QT syndrome. By the use of a segment-assembly homology modeling method, the closed-state hERG structure containing unusual longer S5P linker was successfully constructed. It has a "petunia" shape, while four "petals" of symmetrically distributed S5P segments always decentralize. Starting from the hERG and BeKm-1 structures, a considerably reasonable BeKm-1-hERG complex structure was then screened out and identified by protein-protein docking, molecular dynamics (MD) simulations, and calculation of relative binding free energies. The validity of this predicted complex was further assessed by computational alanine-scanning, with the results correlating reasonably well with experimental data. In the novel complex structure, four considerably flexible S5P linkers are far from the BeKm-1 peptide. The BeKm-1 mainly uses its helical region to associate the channel outer vestibule, except for the S5P linker region; however, structural analysis further implies this neutral pore region with wiggling S5P linker is highly beneficial to the binding of BeKm-1 with lower positive charges. The most critical Lys18 of BeKm-1 plugs its side chain into the channel selectivity filter, while the secondarily important Arg20 forms three hydrogen bonds with spatially neighboring residues in the hERG channel. Different from the classical peptide-K+ channel interaction mainly induced by electrostatic interaction, a synergetic effect of the electrostatic and van der Waals interactions was found to mediate the molecular recognition between BeKm-1 and the hERG channel. And this specific binding process is revealed to be a dynamic change of reduction of binding free energy and conformational rearrangement mainly in the interface of both BeKm-1 and the hERG channel. All these structural and energy features yield deep insights on the high selective binding mechanism of hERG-specific peptides, present a diversity of peptide-K+ channel interactions, and also provide important clues to further study structure-function relationships of the hERG channel.

Full text

Page 1

Interaction Simulation of hERG K+Channel with Its Specific BeKm-1
Peptide: Insights into the Selectivity of Molecular Recognition
Hong Yi, Zhijian Cao, Shijin Yin, Chao Dai, Yingliang Wu,* and Wenxin Li*
State Key Laboratory of Virology, College of Life Sciences, Wuhan University, Wuhan, Hubei 430072, People’s
Republic of China
Received July 26, 2006
Potassium channels show a huge variability in the affinity when recognizing enormous bioactive
peptides, and the elucidation of their recognition mechanism remains a great challenge due to an
undetermined peptide-channel complex structure. Here, we employed combined computation methods
to study the specific binding of BeKm-1 peptide to the hERG potassium channel, which is an essential
determinant of the long-QT syndrome. By the use of a segment-assembly homology modeling method,
the closed-state hERG structure containing unusual longer S5P linker was successfully constructed. It
has a “petunia” shape, while four “petals” of symmetrically distributed S5P segments always
decentralize. Starting from the hERG and BeKm-1 structures, a considerably reasonable BeKm-1-hERG
complex structure was then screened out and identified by protein-protein docking, molecular dynamics
(MD) simulations, and calculation of relative binding free energies. The validity of this predicted complex
was further assessed by computational alanine-scanning, with the results correlating reasonably well
with experimental data. In the novel complex structure, four considerably flexible S5P linkers are far
from the BeKm-1 peptide. The BeKm-1 mainly uses its helical region to associate the channel outer
vestibule, except for the S5P linker region; however, structural analysis further implies this neutral
pore region with wiggling S5P linker is highly beneficial to the binding of BeKm-1 with lower positive
charges. The most critical Lys18 of BeKm-1 plugs its side chain into the channel selectivity filter, while
the secondarily important Arg20 forms three hydrogen bonds with spatially neighboring residues in
the hERG channel. Different from the classical peptide-K+channel interaction mainly induced by
electrostatic interaction, a synergetic effect of the electrostatic and van der Waals interactions was
found to mediate the molecular recognition between BeKm-1 and the hERG channel. And this specific
binding process is revealed to be a dynamic change of reduction of binding free energy and
conformational rearrangement mainly in the interface of both BeKm-1 and the hERG channel. All these
structural and energy features yield deep insights on the high selective binding mechanism of hERG-
specific peptides, present a diversity of peptide-K+channel interactions, and also provide important
clues to further study structure-function relationships of the hERG channel.
Keywords: A segment-assembly homology modeling • protein-protein docking • MD simulations • BeKm-1 • hERG
potassium channel • selective molecular recognition
Introduction
During cardiac repolarization, the hERG (human ether-a-go-
go-related gene1) K+channel plays an essential role in mediat-
ing the process of returning the membrane potential to its
resting status.2An inherited mutation in normal human hERG
gene could cause a disorder of cardiac repolarization, which
may directly lead to long-QT syndrome-related proarrhythmia
and sudden death.3Moreover, the similar disorder called
acquired-long QT syndrome could be triggered by many anti-
arrhythmic and noncardiovascular medication drugs via block-
age of the hERG channel.4Thus, hERG has now become a focus
target in the pharmaceutical industry for detecting this unde-
sirable side effect.5
The hERG channel shares the basic overall transmembrane
topology of the voltage-gated K+channel (Kv) family.6It is
composed of four subunits, each containing a voltage-sensing
module (S1-S4) and an ion-conducting module (S5-pore loop-
S6).2However, hERG has an unusually longer S5P linker
compared with typical Kv channels (41 amino acids versus 14-
23 amino acids),7and considerable conformational flexibility
of this S5P linker region has been found by circular dichroism
(CD) spectropolarimetry and NMR spectroscopy studies re-
cently.8,9Many studies showed that this unique S5P linker
endowed the hERG channel with novel structure-function
features. Mutagenesis studies have proposed the S5P segment
* To whom correspondence may be addressed. For W.L.: tel., ++86-(0)-
27-68752831; fax, ++86-(0)-27-68752146; e-mail, liwxlab@whu.edu.cn. For
Y.W.: e-mail, ylwu@whu.edu.cn.
10.1021/pr060368g CCC: $37.00
 2007 American Chemical Society
Journal of Proteome Research 2007, 6, 611-620
611
Published on Web 12/24/2006

Page 2

to be an important determinant of hERG, particularly rapid
inactivation kinetics,7,9,10which is critical for the role of the
channel in maintaining normal cardiac electrical activity and
suppressing arrhythmias initiated by ectopic electrical excita-
tion.11,12Besides its special electrophysiological function, the
S5P linker is also suggested to be responsible to the selective
binding of peptide ligands.13,14All these indicated that the hERG
channel contains a very unique outer pore region different from
other Kv channels,15due to the presence of a longer S5P linker.
But unfortunately, there is no information available concerning
the structural architecture of this pore-forming region yet.
Scorpion toxins are short-chain peptides extracted from
scorpion venom that could specifically block or modify potas-
sium permeability.16A considerable amount of knowledge on
the structure and function of channel extracellular surface has
been elucidated by using scorpion toxins as molecular
probes.17-19BeKm-1, a peptide isolated from the scorpion
Buthus eupeus, shares a similar molecular scaffold with other
short scorpion toxins.20It could specifically block the hERG
channel with an IC50of 3.3 nM but, interestingly, did not inhibit
other Kv channels currents even at 100 nM.21Subsequently,
mutagenesis results imply that critical amino acids Lys18 and
Arg20 of BeKm-1 associate with hERG by a novel interaction
mode, and possible contacts between Tyr11 of BeKm-1 and
residues in the S5P linker were presumed to be responsible
for this selective binding,20,22but without specification of exact
contact pairs or any other powerful evidence, thus, leading to
the question, does the seriously flexible S5P linker directly
influence or just take part in BeKm-1 binding? Because of the
absence of an hERG crystal structure, mapping the precise
composition of a channel-toxin complex could provide a
framework for a better understanding of the structure-function
relationships, especially for clarifying the role of the S5P linker
in the selective binding of hERG with peptide ligands and
channel gating.
To date, crystallization and determination of K+channel
structures, not to mention the channel-ligand complex, still
suffer from great hurdles.23Therefore, predicting protein-
protein complex structure by computational approaches, such
as docking and molecular dynamics (MD) simulations, has
become a good alternative to understand the interaction
mechanisms between K+channels and peptide inhibitors.24,25
In view of the unique pore region architecture of the hERG
channel and BeKm-1 specificity, this study was undertaken with
the intent of exploring the selective mechanism of hERG
channel recognition of the BeKm-1 peptide. With the segment-
assembling homology modeling method, a reasonable channel
pore region structure with the S5P linker was first constructed
in the closed state. Then, we screened and identified an
equilibrated reasonable structure of the Bekm-1-hERG com-
plex by using a combination of computational methods,
including protein-protein docking and MD simulations. On
the basis of the final complex structure and simulation results,
we elucidated the molecular basis underlying the selectivity of
the BeKm-1-hERG interaction from several aspects: the role
of hERG S5P in the associating process of BeKm-1, the
cooperative driving force of electrostatic and van der Waals
interactions in the blockage of BeKm-1 to the hERG channel,
and the significant conformation changes in both BeKm-1 and
hERG induced by the molecular epitope of interaction. All these
also provided some new insight into the structure-function
of the hERG channel.
Materials and Methods
Atomic Coordinates and Molecular Docking. The atomic
coordinates of BeKm-1 peptide (PDB codes: 1LGL and 1J5J)
were downloaded from the Protein Data Bank;26it has a
globular rigid structure cross-linked by three disulphide bridges
(Cys7-Cys28, Cys13-Cys33, and Cys17-Cys35).
The central pore of hERG shows high sequence identity with
the crystallized K+channel KcsA, except for the S5P segment
(Figure 1). On the basis of this, a homologous structure of the
pore region of hERG was built using a segment-assessment
method consisting of the following procedures: (a) the S5, pore
loop, and the S6 segment were separately modeled using the
structure of the KcsA channel (PDB code: 1BL8) as a template
through the SWISSMODEL server.27(b) On the basis of the
classically supposed hERG structure,8,22an appropriate S5P
segment was selected from the 5 ns unrestrained MD trajec-
tories of S5P linker structure (PDB code: 1UJL). (c) All the
segments were coassembled into a homotetramer model by
fitting into the KcsA structure; an additional 5 ns MD simulation
was performed on the model to get an equilibrated and
reasonable structure.
All 21 BeKm-1 conformations from NMR were used for
improving rigid docking performance of a molecular mocking
algorithm ZDOCK, which is a Fast Fourier Transform (FFT)-
based, initial-stage rigid-body docking algorithm.28The high
accuracy of ZDOCK on predicting protein complexes has been
proved by its good performances in the CAPRI Challenge.29
Clustering and the application of biological information along
with ZDOCK, followed by a 500-steps energy minimization for
each possible toxin-channel complex using the SANDER
module in the AMBER 8 suit of programs,30as well as calculat-
ing the ligand-receptor interaction energies with the ANAL
program of AMBER 8, were employed for identifying appropri-
ate candidate complexes for further MD simulations from
docking results.
In this work, the membrane around the channel has not
been considered during the simulation for several reasons. First,
the mutagenesis studies14,20indicated that the peptide inhibitors
bind to the extracellular part of the potassium channel, where
the interaction might not be affected by the membrane.
Furthermore, many simulation studies on the recognition
between scorpion toxins and potassium channels without
membrane included have achieved good agreements with
experimental data.24,25,31This membrane-ignored treatment also
facilitates the computations greatly.
Molecular Dynamics Simulation. All the calculations were
performed using the Amber 8 program32on a 32-CPU Dawning
TC4000L cluster. In our work, the generalized Born (GB)
solvation model in macromolecular simulations33was used
instead of explicit water during more sufficient MD simulation.
The ff99 force field (Parm99)34was applied throughout the
energy minimization and MD simulations.
As for screened docking candidate complexes, the equilibra-
tion step is relatively simple for efficiency, which is similar to
the one used to predict the ScyTx-K+channel complex
structure with the same environment settings; 50 ps GB-MD
simulations were performed. Force constant restraining for
backbone atoms was reduced from 5.0 to 0.25 (kcal/mol)/Å2.
One hundred snapshots every 0.2 ps from the last 20 ps
simulations were collected for postprocessing analysis. The
most plausible binding mode of the hERG-BeKm1 complex
for 21 conformations of BeKm-1 was screened by calculation
research articles
Yi et al.
612Journal of Proteome Research • Vol. 6, No. 2, 2007

Page 3

of binding free energy and structural analysis based on the
structure-function experiments.20
To ensure that each simulation systems is equilibrated
enough ahead of data collection and analysis,35initially, 360
ps GB-MD (IGB ) 2 in the AMBER 8) simulations were
performed on the S5P linker, BeKm-1 mutants, and the final
complex with a time step of 2.0 fs. Temperature was set at 300
K with the cutoff distance of 12 Å used for nonbonded
interaction. We started the equilibration step from a larger force
constant of 5.0 (kcal/mol)/Å2for restraining all heavy atoms
and then gradually reduced it to 0.02 (kcal/mol)/Å2for only
heavy atoms in the backbone.
To introduce enough flexibility into the side chain confor-
mations, 5 ns unrestrained GB-MD simulations were performed
on the single S5P linker and BeKm-1 mutants, as well as the
four S5P in the hERG structure model. About 8 ns unrestrained
GB-MD simulations were performed on the final complex,
except residues locating at S5 and S6 segment restrained by a
force constant of 0.02 (kcal/mol)/Å2, as these segment are
buried in the membrane and have no effect on the recognition
of BeKm-1. Such hypothesis is based on the result that no
conformational change of S5 and S6 helixes was observed by
solid-state NMR for the K+channel.36
Calculation of Binding Free Energy by MM-PBSA. In the
MM-PBSA of AMBER 8.0, the binding free energy of A + B f
AB is calculated using the following thermodynamic cycle:
Figure 1. Sequence alignments and view of modeled hERG channels. (a) Sequence alignments of the hERG channel with Kv1.2 and
KcsA channels. Red letters show identical residues, whereas gray letters show conservative substitutions. (b and c) Stereoview of the
widely recognized hERG channel model from the extracellular side and a perpendicular perspective. (d and e) Stereoview of the final
equilibrated hERG structure model from the extracellular side and a perpendicular perspective. Four subunits of tetramer are
distinguished by color.
Interaction Simulation of hERG K+Channel with BeKm-1 Peptide
research articles
Journal of Proteome Research • Vol. 6, No. 2, 2007
613

Page 4

where T is the temperature, S is the solute entropy, ∆Ggas is
the interaction energy between A and B in the gas phase, and
∆Gsolv
, ∆Gsolv
, and ∆Gsolv
and AB, which are estimated using a GB surface area (GBSA)
method,32,33,37that is, ∆Gsolv
forth. ∆GGBand ∆GSAare the electrostatic and nonpolar term,
respectively. ∆Ebond, ∆Eangle, and ∆Etorsion are contributions to
the intramolecular energy ∆Eintra of the complex. EvdW is van
der Waals (vdW) interaction energy. Because of the constant
contribution of -T∆S for each docked complex, we quote
∆G*binding, which is ∆Gbinding+ T∆S in the discussion. To verify
the quality and validity of every resulting BeKm-1-hERG
complex obtained, the relative binding free energy ∆G*binding
was calculated by using MM-PBSA for postprocessing collected
snapshots from the MD trajectories.
AB ABare the solvation free energies of A, B,
AB) ∆∆GGBSA
AB
+ ∆GGB
AB+∆GSA
AB, and so
Results and Discussion
A comparison of different open-state and closed-state K+
channel structures emphasizes two important features: (i) the
structure of the selectivity filter is highly conserved for the
mechanism of selective ion conduction and (ii) the structures
of channel domains on the extracellular side also differ very
slightly.15,38In the pore-forming region, the hERG channel is
considerably homologous with KcsA and Kv1.2, except in the
S5P linker region (Figure 1a). Therefore, it is rational to only
build a partial model of the closed-state hERG channel by using
KcsA channel as a template. To predict the BeKm-1-hERG
complex structure, the successful structure modeling of the
hERG channel with the S5P linker also becomes an obvious
prerequisite.
Structural Modeling of hERG Channel and Dynamic Char-
acteristic of S5P Linker. In this work, a segment-assembling
homology modeling method was developed for the structural
modeling of the intact hERG channel. First, we modeled the
hERG channel structure containing the S5 segment (Leu550-
Ala570) and the P-S6 segment (Val612-Thr670) by using KcsA
as a template. Second, dynamic conformation-sampling was
carried out during the 5 ns MD simulation of the S5P segment;9
several “qualified” residue fragments of S5P were used to
gradually assemble the hERG channel structure. Third, energy
minimization was performed on the intact hERG channel
structure to relieve possible side chain steric clashes and
overlaps. Finally, we obtained a widely recognized structure
of hERG (Figure 1b,c):8,22,39(i) Four symmetrically distributed
S5P segments vertically located above the channel pore region;
and (ii) the helical domains of the S5P linker all face the
symmetry axis of the channel selectivity filter so that S5P
possibly takes part in the interaction with the BeKm-1 peptide.
In aqueous solution, the S5P segment presented predomi-
nantly a random coil structure instead of well-defined second-
ary structures. Its central region (Trp585-Ile593) can take a
helical conformation only when immersed in a hydrophobic
environment.8,9Because of actual exposure of the S5P linker
to the extracellular environment, additional 5 ns unrestrained
MD simulations were performed to investigate dynamic char-
acteristics of the S5P linker in the obtained hERG structure.
Surprisingly, the spatial conformation of the final equilibrated
hERG channel resembles a “petunia”, as shown in Figure 1d,e.
More interesting, by iteratively tracking the trajectories of the
MD equilibrating process, we found that, along with the gradual
disappearance of the helical domain, the conformations of four
“petals” of symmetrically distributed S5P segments continu-
ously change without certain rule, and they always appear to
decentralize rather than group together; meanwhile, the MD
trajectories of four S5P linkers differ distinctly from each other.
Such “petunia” shape of the hERG channel is reasonable
because the decentralization of the S5P linkers will allow
BeKm-1 peptide to associate with the channel pore region and
provides possible communication between the S5P linker and
other neighboring segment linkers in hERG.8,14With the serious
flexibility of the S5P linker, an interesting question arises: does
the S5P linker influence the interaction between the BeKm-1
peptide and hERG channel?
Screening of Possible Binding Modes by ZDOCK and MD
Simulations. ZDOCK, a newly developed rigid protein-protein
docking method, had an excellent performance during the past
critical assessment of predicted interactions (CAPRI) chal-
lenge.29,40Here, we applied previous peptide-K+channel dock-
ing steps to predict BeKm-1-hERG complex structures.25Both
classical and equilibrated hERG structures are used to inves-
tigate the role of the flexible S5P linker during the BeKm-1
binding to the hERG channel. Two main candidates of BeKm-
1-hERG channel complex structures, the binding mode I
and binding mode II, were selected, where the critical Lys18
and Arg20 residues are used to associate the hERG channel
(Figure 2b,d).20
To obtain a more native-like complex structure, several
nanosecond MD simulations were performed on both candi-
date complexes with only slightly restraining channel S5 and
S6 transmembrane helixes, in order to extend the time scales
as necessary to complete the allosteric rearrangement.24,41The
initial structure of binding mode I in Figure 2b agrees well with
previous inferences from mutagenesis results,14,20,22and is
completely different from AgTx2 and ScyTx which use ? and R
domain to associate K+channel, respectively.24,25However, after
800 ps MD simulations, the BeKm-1 peptide in binding mode
I shifts unexpectedly from an initial central orientation above
the pore entrance to a position over the pore-S6 linker.
Moreover, this translocation was even stabilized by strong
electrostatic interaction between Arg20 of BeKm-1 and Glu338
of hERG channel (Figure 2b) during the rest simulations.
Meanwhile, the most essential Lys18 of BeKm-1 only contacted
a single residue, Asn207, in the channel within a contact
distance of 4.0 Å. More interestingly, one S5P linker, located
just opposite the BeKm-1 peptide, still kept upright, and other
three S5P linkers were still recumbent. In contrast to the
binding mode I, the BeKm-1-hERG complex had a more highly
stable conformation in the binding mode II during the rest 7.2
ns MD simulation time, after BeKm-1 “lays down” above the
channel pore region in the first 800 ps MD simulations (Figure
2d,e). At this time, there is obvious interaction between the
two critical residues Lys18 and Arg20 of BeKm-1 and the outer
vestibule of the hERG channel. The side chain of Lys18
physically occluded into the selectivity filter of the hERG
channel, and Arg20 of BeKm-1 was surrounded by several
channel residues within a contact distance of 4.0 Å. Here, the
comparison of two binding modes showed that the binding
mode II is more favorable and stable.
Validity of Final BeKm-1-hERG Complex. The main chal-
lenge for computational studies of protein-protein complex
structures has been to reproduce a model with sufficient
reliability. As validated by experimental results, the computa-
tional alanine scanning approach has achieved reasonable
success in the identification of near-native structures.42,43To
validate the binding mode II, we calculated the difference in
the binding free energies between mutated and wild-type
research articles
Yi et al.
614 Journal of Proteome Research • Vol. 6, No. 2, 2007

Page 5

equilibrated BeKm-1-hERG complexes (∆∆Gbinding) by the
computational alanine scanning in MM-PBSA.42The results are
listed in Table 1. Excellent agreement was found between the
calculated and experimental data for the complex. Mutating
Lys18 to Ala18 in BeKm-1 has the most dramatically affected
capacity to inhibit the HERG channel,20as it corresponds to
the biggest value of 12.32 kcal/mol in the ∆∆Gbindingamong all
computational mutations. Substitution of Arg20 in BeKm-1 to
Ala causes a significant drop of affinity to the hERG channel
(only second to Lys18Ala),20which agrees well with the bigger
value of 8.94 kcal/mol in the ∆∆Gbinding. Meanwhile, high values
of 4.18 and 3.18 kcal/mol were found for single point mutants
Figure 2. CR rmsd values of two BeKm-1-hERG binding modes from the initial structure, and the ribbon views of the initial and final
complex structures. (a) Root-mean-square deviations (Å) of the R-carbons of BeKm-1-hERG binding mode I (lower line) and binding
mode II (upper line) compared with the docked-structure. (b and c) Stereoview of the docked and equilibrated BeKm-1-hERG of binding
mode I, respectively. (d and e) Stereoview of the docked and final reasonable BeKm-1-hERG of binding mode II, respectively. The
most critical residues Lys18, Arg20, and Tyr11 are labeled in brief in all the structures.
Table 1. Computational Alanine Scanning Mutagenesis Results for the Complex of BeKm-1 with the hERG Channel (∆∆Ga)
∆Gmutant- ∆Gwild-type)
Arg1Ala Pro2Ala Asp4AlaLys6AlaGlu9AlaTyr11AlaGln12Ala Phe14AlaPhe36Ala
Final Complex
-0.180.23
-0.56
-0.34
-0.634.18 2.02 3.18
-0.01
Lys18AlaArg20Ala Phe21Ala Lys23AlaArg27Ala Val29AlaPhe32AlaAsp34Ala
Final Complex12.32 8.940.160.77 0.01 0.00
-0.02
-0.06
aAll energies are in kcal/mol.
Interaction Simulation of hERG K+Channel with BeKm-1 Peptide
research articles
Journal of Proteome Research • Vol. 6, No. 2, 2007
615

Page 6

of Tyr11 and Phe14 because such modifications disrupted the
toxin-channel interaction moderately.20As for those alanine
substitutions, Asp4Ala, Lys6Ala, and Phe36Ala, for example,
which lead to small changes in BeKm-1 affinity, the values of
∆∆Gbindingshow litter differences. Meanwhile, there are several
seeming disagreements between experimental data and alanine
scanning results, such as the Kdvalues of the Phe21Ala mutant
is 50 times higher than that of the wild-type,20while the
∆∆Gbindingof 0.16 kcal/mol cannot imply that Phe21 is impor-
tant for the binding activity. Actually, when we performed an
additional 5 ns unrestrained MD simulation on the modeled
BeKm-1-Phe21Ala and BeKm-1-Lys23Ala structures, significant
conformational changes were found not only in the disappear-
ance of secondary structure contents in R-helix and ?-sheet,
but also in the deviation of the inferred BeKm-1 binding
surface. This simulation finding is greatly supported by the CD
spectra results,20which proposed that the Phe21Ala mutant
reflected some spatial structure perturbation leading to a
decrease in the inhibitory activity. Moreover, after performing
the same MD simulation, we detected tiny structural changes
in the BeKm-1-Lys18Ala conformation, which suggested our
MD simulation for verifying the stability of protein structure
to be credible. Here, the excellent similarity between compu-
tational and experimental results strongly indicated that the
binding mode II was a considerably reasonable BeKm-1-hERG
complex structure, based on which more structural and
functional information of BeKm-1 peptide and hERG channel
could be obtained.
Implications from the Unique Interaction between BeKm-1
and hERG Channel. The outer vestibule of K+channels are
revealed to be the main determinants for specifically recogniz-
ing peptide toxin inhibitors,22which are widely used for probing
the structure of binding interface on target channels.17,44,45
Compared with a typical Kv channel, the hERG channel has
four main different features concerning peptide inhibitor
recognition: (a) the existence of an unusually longer S5P linker
(Figure 1a); (b) the S5P linker with enough flexibility (Figures
1 and 2); (c) with respect to the abundance of negatively
charged residues in the S5P linker of other Kv channels, the
hERG S5P has a much lower content (Figure 3 and Table 2),
and is far way from the channel pore (Figure 1d and Figure
2e), while in many cases, these negative charged residues locate
near the pore entrance of channels and play important roles
in recognizing peptide inhibitors by directly contacting
them;24,25,31and (d) the signature residue Asp in the K+
selectivity filter of other Kv channels was substituted by Asn629
of hERG channel (Figure 1a). All these suggest a completely
unique interaction between BeKm-1 and the hERG channel,
Figure 3. Molecular surface of hERG and Kv1.2 and possible interaction between major function residues of BeKm-1 and residues of
the hERG channel within a contact distance of 4.0 Å. (a and b) A stereoview displaying the molecular surface of the Kv1.2 and hERG
channel colored according to electrostatic potential properties, with basic residues in blue and acidic residues in red. (c) The Lys18 of
BeKm-1 plugging into the pore of the hERG channel. (d) The Arg20 of BeKm-1 surrounded by residues from the pore of the hERG
channel. The green line represents a hydrogen bond between two atoms.
research articles
Yi et al.
616 Journal of Proteome Research • Vol. 6, No. 2, 2007

Page 7

which is different from classical electrostatic interaction be-
tween ChTx/AgTx2 and Kv channels.24,46Therefore, the struc-
tures of the hERG channel and its complex will give us novel
insights into the structural and dynamic features that determine
the highly selective recognition between the BeKm-1 peptide
and hERG channel.
Role of S5P in BeKm-1 Binding. The reasonable BeKm-1-
hERG complex structure reveals that the S5P does not deter-
mine the binding mode between BeKm-1 and the hERG
channel. It is different from the previous studies that proposed
the long S5P linker to be the specific BeKm-1 binding site by
cysteine substitution experiments.14Here, our structures of
hERG channel and its complex can explain such discrepancy.
As mentioned above, the hERG S5P linker has a strong tendency
to “lie down” above the cell membrane whether in closed-state
channel or in complex with peptide inhibitors (Figure 1d and
Figure 2e). Thus, cysteine side chains (Gly584Cys, Trp585Cys,
Asn588Cys, and Leu589Cys) introduced into positions of the S5P
linker had great possibility to form intersubunit disulfide bonds,
such as forming a disulfide bond with its counterpart,8or
forming disulfide bonds with two Cys (Cys445 and Cys449) in
the long S1S2 linker of other channel subunit. Such formation
of disulfide bond will produce dimmer of channel subunits and
disrupt normal hERG channel function.8Although the S5P
linker does not finally interact with BeKm-1, it possibly affects
the BeKm-1 binding process outside the channel vestibule.
During MD simulation of biding mode I (Figure 2b,c), the
conformation of the keeping-upright S5P linker undergoes an
interesting change, it first becomes flat, together with other
three S5P linkers, and then solely return back to upright, even
move to get sometime closer to the pore entrance. Almost
simultaneously, the BeKm-1 shifts from an initial central
position above the pore entrance to a position that is just
opposite to this special S5P linker.
Different from the interaction between BeKm-1 and the
hERG channel, strong electrostatic forces between more posi-
tive charges on classical toxin peptides (e.g., ChTX and AgTx2)
and more negative charges on the shorter S5P linkers (Table 2
and Figure 3a,b) of Kv channels can efficiently orient the toxin
peptides toward the channel pore.24,46However, the unusually
longer and more flexible S5P with four Lys/Arg, and four Asp/
Glu residues (Figure 3b and Table 2) obviously cannot be
helpful in orienting ChTx toward the hERG channel so that
there is no interaction between them.47On the contrary, there
could be less influence on the binding process of specific
BeKm-1 peptide with only 2.73 positive charges. Therefore, the
S5P linker with unusual length, serious flexibility, and low
abundance of negative charged residues is an important
determinant for the specificity of BeKm-1 to the hERG channel,
and it simultaneously affects the binding process of BeKm-1
peptide.
Novel Interaction between BeKm-1 and hERG Channel. So
far, the BeKm-1 binding mode is different from known pep-
tide-K+channel interactions. Such novelty is mainly shown
as the following: (a) The outer vestibule of the hERG channel
is almost uncharged. In addition to the huge difference of
charged residue number in the S5P linkers between hERG and
Kv channels, the central pore region does not have any acidic
amino acid residue in the hERG channel (Figure 3b and Table
2). In the BeKm-1-hERG channel, there are four substituted
Asn629 residues in the channel around the pore-blocked Lys18
of BeKm-1 (Figure 3c). However, there are four acidic amino
acid residues in the Kv channels which are always around the
conserved Lys27 residues of ChTx and AgTx2 peptides, whose
residual side chain is plugged into the channel selectivity filter
(Figure 3a).24,48-50(b) The BeKm-1, with less positive charges,
mainly uses its helical domain to associate the hERG channel
(Table 2 and Figure 2e). The fact of less positive charges is a
common feature of hERG-specific peptides; for example, CsEK-
erg1 and ErgTx1 only contain 2.64 and 1.78 net charge,51,52
respectively (Table 2). Therefore, the lower positive charges of
hERG-specific peptides and the neutral outer vestibule of the
hERG channel (seen in Figure 3b) are most likely to determine
the high-selectivity of the peptide-channel interaction. In the
BeKm-1-hERG complex, the major functional residues Tyr11,
Phe14, Lys18, and Arg20 are distributed in the helical region
of BeKm-1 (Table 1 and Figure 2e). The Lys18 exerts its function
by plugging its side chain into the channel selectivity filter
(Figure 3c), which seems to be a common feature of K+channel
inhibitory peptides, such as ChTx, AgTx2, and ScyTx.24,25,48In
Figure 3d, there are three hydrogen bonds between Arg20 of
BeKm-1 and Asn629 of hERG channel, which make Arg20
residue a critical residue during BeKm-1 binding to the
channel.20Although both BeKm-1 and ScyTx mainly use the
helical region to associate target channels, the spatial distance
between the two critical Lys18 and Arg20 of BeKm-1 is much
shorter than that of the two critical R6 and R13 of ScyTx.25(c)
The electrostatic and van der Waals interactions almost equally
mediate the recognition process between BeKm-1 and the
hERG channel (Table 3). The electrostatic interaction energy
is just about 1.5-fold lower than van der Waals interaction
energy, since the main interaction between BeKm-1 and the
hERG channel takes place between polar and nonpolar residues
(Figure 3b). Such mechanism is completely different from that
Table 2. Net Charges of Representative K+ Channels and Peptide Inhibitors
name sequence total net chargea
BeKm-1
ErgTx1
CsEKerg1
(Human)hERG-S5P
(Human)hERG-Pore
ChTx
(Rat)Kv1.2-S5P
(Rat)Kv1.2-Pore
(Human)Kv1.3-S5P
(Human)Kv1.3-Pore
ScyTx
(Rat)SKCa2b-S5P
(Rat)SKCa2-Pore
MKISFVLLLTLFICSIGWSEARPTDIKCSESYQCFPVCKSRFGKTNGRCVNGFCDCF
MKVLILIMIIASLMIMGVEMDRDSCVDKSRCAKYGYYQECQDCCKNAGHNGGTCMFFKCKCA
ERDSCVEKSKCGKYGYYGQCDECCKKAGDRAGTCVYYKCKCNP
IGNMEQPHMDSRIGWLHNLGDQIGKPYNSSGLGGPSIKDKY
GFGNVSPNTN
MKILSVLLLALIICSIVGWSEAQFTNVSCTTSKECWSVCQRLHNTSRGKCMNKKCRCYS
EADERDSQFPSI
GYGDMVPTTI
EADDPTSGFSSI
GYGDMHPVTI
AFCNLRMCQLSCRSLGLLGKCIGDKCECVKH
WTVRACERYHDQQDVTSNF
GYGDMVPNTI
2.73
1.73
2.64
0.72
0.00
5.78
-12.00
-4.00
-12.00
-3.64
2.82
-3.81
-4.00
aThe total net charges of channels are calculated in four subunits of tetramer, at pH 7.0.bSKCa2, the small conductance calcium-activated potassium
channel 2.
Interaction Simulation of hERG K+Channel with BeKm-1 Peptide
research articles
Journal of Proteome Research • Vol. 6, No. 2, 2007
617

Page 8

of classical peptides recognizing target channels, whose elec-
trostatic interaction energy should be absolutely domi-
nant.24,41,46As for the interaction of ScyTx with its target
channel, the electrostatic interaction energy rises to about
3-fold lower than van der Waals interaction energy because
there is more negative charge in the outer vestibule of the K+
channel.25Such difference of interactive energies certainly
reflects the diversity of the recognition mechanism among
different peptides blocking K+channels.
In summary, the novel interaction of BeKm-1 with the hERG
channel is determined by the unique outer vestibule of hERG,
special binding mode of BeKm-1, and synergetic effect of the
electrostatic and van der Waals interactions.
Conformation Changes Induced by the Recognition. Re-
cently, Lange et al.36propose that the structural flexibility of
the K+channel and the toxin represents an important deter-
minant for the highly specific toxin-K+channel recognition.
Such conformational flexibility, induced by the interaction of
BeKm-1 with the hERG channel, was also observed in this work.
After comparing the BeKm-1-hERG complex from the ZDOCK
procedure with the final reasonable structure, we also detect
significant structural rearrangements in both BeKm-1 and the
hERG channel. The backbone root-mean-square deviation
(rmsd) is 1.58 Å for bound and unbound BeKm-1 (Figure 4a).
There is remarkable structural change in the Lys18-Asn30
segment of BeKm-1 together with the complete disappearance
of ?-sheet, which is induced by the interaction with the hERG
channel. In addition to the conformational change of BeKm-1
backbone, more significant conformational change is found in
residual side chains on the BeKm-1 interface, especially for the
Figure 4. Analysis of conformational changes induced by the interaction of BeKm-1 and hERG channel. (a) Superposition of BeKm-1
free and bound conformations, with side chains of the Lys18 and Arg20 residues, which had the most significant conformational
changes, marked in solid bonds. The free BeKm-1 is green, and the bound BeKm-1-hERG complex is pink. (b) Superposition of hERG
unbound and bound conformations, marking the residues in BeKm-1 that most possibly induce the channel conformation change. The
color scheme is the same as in panel a. (c and d) Surface plots of BeKm-1 and the hERG channel, respectively. Residues characterized
by significant MD-conformation changes upon complex formation are purple.
Table 3. Dynamic Change of Relative Binding Free Energies from Representative BeKm-1-hERG Conformations during the Binding
Processa
complex
∆Eelec
∆EvdW
∆Einter
∆∆GGB
∆∆GSA
∆Gbinding
Docking Complex
Intermediate Conformation
Final Conformation
-69.37
-91.44
-152.06
-64.76
-83.37
-100.94
-134.13
-174.81
-253.00
133.77
138.88
201.30
-7.74
-10.21
-12.57
-8.11
-46.14
-64.27
aAll energies are in kcal/mol.
research articles
Yi et al.
618 Journal of Proteome Research • Vol. 6, No. 2, 2007

Page 9

essential residues of Lys18 and Arg20 (Figure 4a). The side chain
of Lys18 rotates and extends as long as possible for matching
the limited space of selectivity filter during binding to the hERG
channel (Figure 4a). The Arg20 residue also rotates with a
certain angle and bends its side chain when recognizing its
target channel through the formation of three hydrogen bonds
(Figures 3d and 4a). Other important residues in BeKm-1 with
obvious conformational changes are also labeled in Figure 4c.
As for the hERG channel, the structure of the outer and inner
pore helixes is largely maintained after binding BeKm-1, which
was also observed in the previous studies.36In contrast,
considerable structural changes were seen in Val625-Ser636
segment, which forms the channel interface. The rmsd values
of backbone and side chain atoms are 1.88 and 2.53 Å between
bound and unbound hERG channel, respectively (Figure 4b).
Especially large side chain torsions were found for the Gly628-
Asn633 segment, which locates in the region that directly
interacts with the toxin (Figure 4d). These observations were
consistent with the experiment data that mutation of the
residue S631 and P632 in hERG channels greatly reduces
BeKm-1 affinity.14The remarkable conformational flexibility in
the GYG (or GFG) of the K+ion channels were also described
to be the key principle underlying selective K+conduction.38,53
During the interaction process of BeKm-1 and the hERG
channel, the side chains of several residues in both peptide
and channel are prealigned in their nanosecond-scale diffu-
sional encounter, and then lead to a specifically stable com-
plex.41This conformational arrangement indicated that the
contacts between BeKm-1 and hERG were strengthened gradu-
ally, and calculations of binding free energies of complexes in
the binding process greatly supported this finding. A harmoni-
ous stepwise process of energy minimizing was found to be
associated with the forming of stable complex (list in Table 3),
which demonstrated that a reduction of binding free energy
was associated with tighter interaction of BeKm-1 into the
target channel. All these strongly suggest that the binding
process is also a dynamic process of interactive energy change.
In conclusion, the change of both conformation and interac-
tive energy not only provides the basis for formation of a tight
complex with the active site of the K+channel, but also may
be a crucial prerequisite for the high-affinity ligand binding to
an ion channel.
Conclusion
On the basis of a modeled closed-state hERG channel
structure, the present work has yielded unprecedented infor-
mation on the recognition of scorpion toxin BeKm-1 with the
hERG channel. Starting from modeling the hERG channel using
the segment-assembly method, we found the pore region of
the hERG channel presents a “petunia” shape, and four “petals”
of symmetrically distributed S5P segments always decentralize.
The unusual longer S5P linker was revealed to wiggle above
the pore region of the hERG channel because of its intrinsic
flexibility. Interestingly, this considerably flexible S5P linker
likely affects the binding of BeKm-1 at certain degree, but does
not determine the final binding mode of BeKm-1 with the hERG
channel. With combined computational methods, we obtained
the reasonable BeKm-1-hERG complex structure, which is
different from that of the classically supposed model. In the
complex structure, the BeKm-1 mainly uses its helical domain
to associate target channel, while the Lys18 plugs its side chain
into the channel selectivity filter, and another critical Arg20
forms three hydrogen bonds with its spatially neighboring
residues of hERG. The binding process of BeKm-1 is equally
mediated by the electrostatic and van der Wals interactions,
which is determined by the unique and neutral outer vestibule
of hERG and lower positive charges of BeKm-1. During their
binding process, the significant conformational rearrangement
and gradual reduction of interactive energy may be a crucial
prerequisite for the highly specific BeKm-1 to target the
channel. All these structure and energy characteristics deter-
mine BeKm-1 to be a hERG-specific peptide. Simultaneously,
these findings can accelerate the research of hERG structure-
function relationship and our understanding of the role of the
hERG channel in the acquired-long QT syndrome. In addition,
our work also shows that the segment-assembly homology
modeling method and combined simulation programs are an
attractive approach to strongly study molecular mechanism and
effectively construct the network of numerous protein-protein
interactions.
Acknowledgment.
grants from the National Natural Sciences Foundation of China
to W.L., Y.W., and Z.C. (numbers: 30530140, 30500089, and
30570045), National Education Ministry Foundation of China
(number: 106108), and the Provincial Natural Sciences Foun-
dation of HuBei to Z.C. (number: 2005ABA116).
This work was supported by the
References
(1) Warmke, J. W.; Ganetzky, B. A family of potassium channel genes
related to eag in Drosophila and mammals. Proc. Natl. Acad. Sci.
U.S.A. 1994, 91, 3438-3442.
(2) Sanguinetti, M. C.; Tristani-Firouzi, M. hERG potassium channels
and cardiac arrhythmia. Nature 2006, 440, 463-469.
(3) Curran, M. E.; Splawski, I.; Timothy, K. W.; Vincent, G. M.; Green,
E. D.; Keating, M. T. A molecular basis for cardiac arrhythmia:
HERG mutations cause long QT syndrome. Cell 1995, 80, 795-
803.
(4) Recanatini, M.; Poluzzi, E.; Masetti, M.; Cavalli, A.; De Ponti, F.
QT prolongation through hERG K(+) channel blockade: current
knowledge and strategies for the early prediction during drug
development. Med. Res. Rev. 2005, 25, 133-166.
(5) Aronov, A. M. Predictive in silico modeling for hERG channel
blockers. Drug Discovery Today 2005, 10, 149-155.
(6) Yellen, G. The voltage-gated potassium channels and their
relatives. Nature 2002, 419, 35-42.
(7) Liu, J.; Zhang, M.; Jiang, M.; Tseng, G. N. Structural and functional
role of the extracellular s5-p linker in the HERG potassium
channel. J. Gen. Physiol. 2002, 120, 723-737.
(8) Jiang, M.; Zhang, M.; Maslennikov, I. V.; Liu, J.; Wu, D. M.;
Korolkova, Y. V.; Arseniev, A. S.; Grishin, E. V.; Tseng, G. N.
Dynamic conformational changes of extracellular S5-P linkers in
the hERG channel. J. Physiol. 2005, 569, 75-89.
(9) Torres, A. M.; Bansal, P. S.; Sunde, M.; Clarke, C. E.; Bursill, J. A.;
Smith, D. J.; Bauskin, A.; Breit, S. N.; Campbell, T. J.; Alewood, P.
F.; Kuchel, P. W.; Vandenberg, J. I. Structure of the HERG K+
channel S5P extracellular linker: role of an amphipathic alpha-
helix in C-type inactivation. J. Biol. Chem. 2003, 278, 42136-
42148.
(10) Clarke, C. E.; Hill, A. P.; Zhao, J.; Kondo, M.; Subbiah, R. N.;
Campbell, T. J.; Vandenberg, J. I. Effect of S5P alpha-helix charge
mutants on inactivation of hERG K+ channels. J. Physiol. 2006,
573, 291-304.
(11) Piper, D. R.; Varghese, A.; Sanguinetti, M. C.; Tristani-Firouzi, M.
Gating currents associated with intramembrane charge displace-
ment in HERG potassium channels. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 10534-10539.
(12) Lu, Y.; Mahaut-Smith, M. P.; Varghese, A.; Huang, C. L.; Kemp,
P. R.; Vandenberg, J. I. Effects of premature stimulation on HERG
K(+) channels. J. Physiol. 2001, 537, 843-851.
(13) Pardo-Lopez, L.; Zhang, M.; Liu, J.; Jiang, M.; Possani, L. D.; Tseng,
G. N. Mapping the binding site of a human ether-a-go-go-related
gene-specific peptide toxin (ErgTx) to the channel’s outer vesti-
bule. J. Biol. Chem. 2002, 277, 16403-16411.
Interaction Simulation of hERG K+Channel with BeKm-1 Peptide
research articles
Journal of Proteome Research • Vol. 6, No. 2, 2007
619

Page 10

(14) Zhang, M.; Korolkova, Y. V.; Liu, J.; Jiang, M.; Grishin, E. V.; Tseng,
G. N. BeKm-1 is a HERG-specific toxin that shares the structure
with ChTx but the mechanism of action with ErgTx1. Biophys. J.
2003, 84, 3022-3036.
(15) Long, S. B.; Campbell, E. B.; Mackinnon, R. Crystal structure of a
mammalian voltage-dependent Shaker family K+ channel. Sci-
ence 2005, 309, 897-903.
(16) Rodriguez de la Vega, R. C.; Possani, L. D. Current views on
scorpion toxins specific for K+-channels. Toxicon 2004, 43, 865-
875.
(17) Hidalgo, P.; MacKinnon, R. Revealing the architecture of a K+
channel pore through mutant cycles with a peptide inhibitor.
Science 1995, 268, 307-310.
(18) MacKinnon, R.; Cohen, S. L.; Kuo, A.; Lee, A.; Chait, B. T.
Structural conservation in prokaryotic and eukaryotic potassium
channels. Science 1998, 280, 106-109.
(19) Ruta, V.; Jiang, Y.; Lee, A.; Chen, J.; MacKinnon, R. Functional
analysis of an archaebacterial voltage-dependent K+ channel.
Nature 2003, 422, 180-185.
(20) Korolkova, Y. V.; Bocharov, E. V.; Angelo, K.; Maslennikov, I. V.;
Grinenko, O. V.; Lipkin, A. V.; Nosyreva, E. D.; Pluzhnikov, K. A.;
Olesen, S. P.; Arseniev, A. S.; Grishin, E. V. New binding site on
common molecular scaffold provides HERG channel specificity
of scorpion toxin BeKm-1. J. Biol. Chem. 2002, 277, 43104-43109.
(21) Korolkova, Y. V.; Kozlov, S. A.; Lipkin, A. V.; Pluzhnikov, K. A.;
Hadley, J. K.; Filippov, A. K.; Brown, D. A.; Angelo, K.; Strobaek,
D.; Jespersen, T.; Olesen, S. P.; Jensen, B. S.; Grishin, E. V. An
ERG channel inhibitor from the scorpion Buthus eupeus. J. Biol.
Chem. 2001, 276, 9868-9876.
(22) Rodriguez de la Vega, R. C.; Merino, E.; Becerril, B.; Possani, L.
D. Novel interactions between K+ channels and scorpion toxins.
Trends Pharmacol. Sci. 2003, 24, 222-227.
(23) Gulbis, J. M.; Doyle, D. A. Potassium channel structures: do they
conform? Curr. Opin. Struct. Biol. 2004, 14, 440-446.
(24) Eriksson, M. A.; Roux, B. Modeling the structure of agitoxin in
complex with the Shaker K+ channel: a computational approach
based on experimental distance restraints extracted from ther-
modynamic mutant cycles. Biophys. J. 2002, 83, 2595-2609.
(25) Wu, Y.; Cao, Z.; Yi, H.; Jiang, D.; Mao, X.; Liu, H.; Li, W. Simulation
of the interaction between ScyTx and small conductance calcium-
activated potassium channel by docking and MM-PBSA. Biophys.
J. 2004, 87, 105-112.
(26) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliland, G.; Bhat, T. N.;
Weissig, H.; Shindyalov, I. N.; Bourne, P. E. The Protein Data
Bank. Nucleic Acids Res. 2000, 28, 235-242.
(27) Guex, N.; Peitsch, M. C. SWISS-MODEL and the Swiss-Pdb-
Viewer: an environment for comparative protein modeling.
Electrophoresis 1997, 18, 2714-2723.
(28) Chen, R.; Li, L.; Weng, Z. ZDOCK: an initial-stage protein-docking
algorithm. Proteins 2003, 52, 80-87.
(29) Wiehe, K.; Pierce, B.; Mintseris, J.; Tong, W. W.; Anderson, R.;
Chen, R.; Weng, Z. ZDOCK and RDOCK performance in CAPRI
rounds 3, 4, and 5. Proteins 2005, 60, 207-213.
(30) Case, D. A.; Cheatham, T. E., III; Darden, T.; Gohlke, H.; Luo, R.;
Merz, K. M., Jr.; Onufriev, A.; Simmerling, C.; Wang, B.; Woods,
R. J. The Amber biomolecular simulation programs. J. Comput.
Chem. 2005, 26, 1668-1688.
(31) Cui, M.; Shen, J.; Briggs, J. M.; Fu, W.; Wu, J.; Zhang, Y.; Luo, X.;
Chi, Z.; Ji, R.; Jiang, H.; Chen, K. Brownian dynamics simulations
of the recognition of the scorpion toxin P05 with the small-
conductance calcium-activated potassium channels. J. Mol. Biol.
2002, 318, 417-428.
(32) Case, D. A. T. A. D.; Cheatham, T. E., III; Simmerling, C. L.; Wang,
J.; Duke, R. E.; Luo, R.; Merz, K. M.; Wang, B.; Pearlman, D. A.;
Crowley, M.; Brozell, S.; Tsui, V.; Gohlke, H.; Mongan, J.; Hornak,
V.; Cui, G.; Beroza, P.; Schafmeister, C.; Caldwell, J. W.; Ross, W.
S.; Kollman, P. A. Amber 8, University of California: San Fran-
cisco, CA, 2004.
(33) Tsui, V.; Case, D. A. Theory and application of the generalized
Born solvation model in macromolecular simulations. Biopoly-
mers 2001, 56, 275-291.
(34) Wang, J.; Cieplak, P.; Kollman, P. A. How well does a RESP-
(restrained electrostatic potential) model do in calculating the
conformational energies of organic and biological molecules? J.
Comput. Chem. 2000, 21, 1049-1074.
(35) Kuhn, B.; Kollman, P. A. A ligand that is predicted to bind better
to avidin than biotin: insights from computational fluorine
scanning. J. Am. Chem. Soc. 2000, 122, 3909-3916.
(36) Lange, A.; Giller, K.; Hornig, S.; Martin-Eauclaire, M. F.; Pongs,
O.; Becker, S.; Baldus, M., Toxin-induced conformational changes
in a potassium channel revealed by solid-state NMR. Nature 2006,
440, 959-962.
(37) Qiu, D.; Shenkin, P. S.; Hollinger, F. P.; Still, W. C. The GB/SA
continuum model for solvation. A fast analytical method for the
calculation approximate Born radii. J. Phys. Chem. 1997, 101,
3005-3014.
(38) Doyle, D. A.; Morais Cabral, J.; Pfuetzner, R. A.; Kuo, A.; Gulbis,
J. M.; Cohen, S. L.; Chait, B. T.; MacKinnon, R. The structure of
the potassium channel: molecular basis of K+ conduction and
selectivity. Science 1998, 280, 69-77.
(39) Xu, C. Q.; Zhu, S. Y.; Chi, C. W.; Tytgat, J. Turret and pore block
of K+ channels: what is the difference? Trends Pharmacol. Sci.
2003, 24, 446-448; author reply 448-449.
(40) Mendez, R.; Leplae, R.; Lensink, M. F.; Wodak, S. J. Assessment
of CAPRI predictions in rounds 3-5 shows progress in docking
procedures. Proteins 2005, 60, 150-169.
(41) Huang, X.; Dong, F.; Zhou, H. X. Electrostatic recognition and
induced fit in the kappa-PVIIA toxin binding to Shaker potassium
channel. J. Am. Chem. Soc. 2005, 127, 6836-6849.
(42) Massova, I.; Kollman, P. A. Computational alanine scanning to
probe protein-protein interactions: a novel approach to evaluate
binding free energies. J. Am. Chem. Soc. 1999, 121, 8133-8143.
(43) Chong, L. T.; Swope, W. C.; Pitera, J. W.; Pande, V. S. Kinetic
computational alanine scanning: application to p53 oligomer-
ization. J. Mol. Biol. 2006, 357, 1039-1049.
(44) MacKinnon, R.; Miller, C. Mutant potassium channels with altered
binding of charybdotoxin, a pore-blocking peptide inhibitor.
Science 1989, 245, 1382-1385.
(45) Gross, A.; MacKinnon, R. Agitoxin footprinting the shaker potas-
sium channel pore. Neuron 1996, 16, 399-406.
(46) Naini, A. A.; Miller, C. A symmetry-driven search for electrostatic
interaction partners in charybdotoxin and a voltage-gated K+
channel. Biochemistry (Moscow) 1996, 35, 6181-6187.
(47) Garcia, M. L.; Gao, Y.; McManus, O. B.; Kaczorowski, G. J.
Potassium channels: from scorpion venoms to high-resolution
structure. Toxicon 2001, 39, 739-748.
(48) Park, C. S.; Miller, C. Interaction of charybdotoxin with permeant
ions inside the pore of a K+ channel. Neuron 1992, 9, 307-313.
(49) Takeuchi, K.; Yokogawa, M.; Matsuda, T.; Sugai, M.; Kawano, S.;
Kohno, T.; Nakamura, H.; Takahashi, H.; Shimada, I. Structural
basis of the KcsA K(+) channel and agitoxin2 pore-blocking toxin
interaction by using the transferred cross-saturation method.
Structure 2003, 11, 1381-1392.
(50) Park, C. S.; Miller, C. Mapping function to structure in a channel-
blocking peptide: electrostatic mutants of charybdotoxin. Bio-
chemistry (Moscow) 1992, 31, 7749-7755.
(51) Nastainczyk, W.; Meves, H.; Watt, D. D. A short-chain peptide
toxin isolated from Centruroides sculpturatus scorpion venom
inhibits ether-a-go-go-related gene K(+) channels. Toxicon 2002,
40, 1053-1058.
(52) Gurrola, G. B.; Rosati, B.; Rocchetti, M.; Pimienta, G.; Zaza, A.;
Arcangeli, A.; Olivotto, M.; Possani, L. D.; Wanke, E. A toxin to
nervous, cardiac, and endocrine ERG K+ channels isolated from
Centruroides noxius scorpion venom. FASEB J. 1999, 13, 953-
962.
(53) Zhou, Y.; Morais-Cabral, J. H.; Kaufman, A.; MacKinnon, R.
Chemistry of ion coordination and hydration revealed by a K+
channel-Fab complex at 2.0 Å resolution. Nature 2001, 414,
43-48.
PR060368G
research articles
Yi et al.
620Journal of Proteome Research • Vol. 6, No. 2, 2007