Ligand-gated synthetic ion channels

Article abstract of DOI:10.1002/chem.200500516

Supramolecular π-stack architecture is fundamental in DNA chemistry but absent in biological and synthetic ion channels and pores. Here, a novel rigid-rod π-stack architecture is introduced to create synthetic ion channels with characteristics that are at

Full text of DOI:10.1002/chem.200500516

FULL PAPER
DOI: 10.1002/chem.200500516
Ligand-Gated Synthetic Ion Channels
Pinaki Talukdar, Guillaume Bollot, Jiri Mareda, Naomi Sakai, and Stefan Matile*[a]
Abstract: Supramolecular p-stack ar-
chitecture is fundamental in DNA
chemistry but absent in biological and
synthetic ion channels and pores. Here,
a novel rigid-rod p-stack architecture is
introduced to create synthetic ion
channels with characteristics that are at
the forefront of rational design, that is,
twist inactive electron-poor helical p-
stacks without internal space into open
barrel-stave ion channels. Conductance
experiments in planar lipid bilayers
corroborate results from spherical bi-
layers and molecular modeling: Highly
cooperative and highly selective ligand
gating produces small, long-lived,
weakly anion selective, ohmic ion chan-
nels. Structural studies conducted
under conditions relevant for function
provide experimental support for
helix–barrel transition as origin of
ligand gating. Control experiments
demonstrate that minor structural
changes leading to internal decrowding
suffice to cleanly annihilate chiral self-
organization and function.
Keywords: bioorganic chemistry ·
ion channels · pi interactions · self-
assembly · supramolecular chemis-
try
ligand gating by
a conformational
change of the functional supramole-
cule. Namely, the intercalation of elec-
tron-rich aromatics is designed to un-
Introduction
pore–membrane interactions resulted in pore activation in
response to chemical stimulation that could be used for the
continuous detection of chemical reactions. This synergism
was further proven on the structural level to originate from
ligand-mediated change in partitioning, presumably without
conformational change of the pore. Indeed, we failed so far
to find a strategy that would produce closed rigid-rod b-
barrel architecture without internal space which could physi-
cally open up by a conformational change in response to
chemical stimulation. Therefore, replacement of the b-sheet
hoops in rigid-rod b-barrels by p-stacks appeared as a
choice. In the resulting “p-barrel,” the repeat distance of
4.9 Š of the p-octophenyl stave should exceed the repeat
distance of around 3.4 Š between “p-hoops” (Figure 1).
Therefore, a rigid-rod p-barrel should twist into a rigid-rod
After the creation of unifunctional ion channels as such,[1]
scientific attention is continuously shifting toward “smart”
supramolecular architecture with refined, multiple func-
tions.[2–19] Early breakthroughs on the recognition of non-
trivial ions[2] and specific characteristics of lipid bilayer
membranes such as polarization[3] were followed by many
remarkable studies on ion and membrane selectivity. The
more recent combination of molecular translocation with
molecular recognition[4] and transformation[5] lead to the
concept of synthetic multifunctional pores and practical ap-
plications[6,7] in sensing and catalysis.[8–10] Current research
on molecular recognition by not only synthetic[1–19] but also
bioengineered[20,21] and biological[22] ion channels and pores
as sensors focuses, however, almost exclusively on blockage.
The more demanding pore opening in response to molec-
ular recognition was achieved last year with a synthetic mul-
tifunctional pore from the classical rigid-rod b-barrel
series.[7] External pore design focusing on the modulation of
“p-helix” with matched hoop-stave repeats (e.g. 1nH
,
Figure 1). Such a notional rigid-rod p-helix was envisioned
as the desired closed ion channel where aromatic ligands
could intercalate, initiate the untwisting helix–barrel transi-
tion and give an open barrel-stave ion channel (e.g. 1n·2m,
Figure 1).
The envisioned p-stack architecture and intercalation
chemistry is reminiscent of chemistry and biology of oligo-
nucleotides such as DNA duplexes, but rare, if not unknown
as architecture of biological or synthetic ion channels and
pores.[1–22] The compatibility of p-stack architecture with the
cylindrical interface between aqueous interior and non-polar
surrounding of ion channels and pores was unknown and
[a] P. Talukdar, G. Bollot, Dr. J. Mareda, Dr. N. Sakai, Prof. S. Matile
Department of Organic Chemistry, University of Geneva
Geneva (Switzerland)
Fax : (+41)22-379-3215
E-mail: stefan.matile@chiorg.unige.ch
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author: Experimental
Section.
Chem. Eur. J. 2005, 11, 6525 – 6532
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6525

therefore reason for concern. However, due to remarkable
progress in supramolecular chemistry, a marvelous collection
of p-stack architectonics was on hand to address these ques-
tions. One of the best-explored motifs is the purple charge-
transfer (CT) complex formed by dialkoxynaphthalene
(DAN) donors and naphthalenediimide (NDI) acceptors.
These complexes have already been of use to create single-
and double-stranded oligo-DAN–NDI foldamers,[23–25]
switchable DAN–NDI rotaxanes[26,27] and DAN–NDI cate-
nanes.[28–30]
between the intercalating NDI hoops and in DAN–NDI CT
complexes.
1) To promote self-assembly into cylindrical supramolecular
oligomers (and to prevent linear self-assembly into
supramolecular polymers), it was important to place cor-
ners in the suprastructure. Rigid p-octiphenyl rods with
seven adaptable, chiral biphenyl torsion angles w ¼ 0
per stave are considered to be perfect for this pur-
pose.[8,9]
Encouraged by these spectacular applications to refined
architecture, we decided to use the DAN–NDI motif for the
creation of ligand-gated synthetic ion channels. The elec-
tron-poor NDI acceptors rather than electron-rich DAN
donors were used in the envisioned p-stack architecture be-
cause of their preference for face-to-face self-assembly.[31]
The p-octiphenyls 1 with eight NDIs along the rigid-rod
2) Amides were installed at both ends of the NDI to form
hydrogen-bonded chains along the NDI stacks oriented
parallel to the rigid-rod staves (Figure 1A). These hydro-
gen-bonded chains were expected to position the p-
stacks like hoops in the barrel-stave architecture.[32–34] In
geometry-minimized molecular models, these hydrogen-
bonded chains were intact after the intercalation of
DAN ligands. In hydrophobic bilayer membranes, the
hydrophilic alkylammonium tails consequently point
toward the interior of the barrel (Figure 1A).
scaffold were expected to self-assemble into p-helices 1nH
.
Helix–barrel transition in response to the intercalation of
DAN donors 2 should then give open ion channel 1n·2m.
Ligand 2 was equipped with an anionic tail on one and an
alkyl tail on the other side to support CT complex formation
by ion pairing within the cationic interior of the notional p-
barrel 1n·2m and external hydrophobic contacts with the sur-
rounding bilayer membrane, respectively (Figure 1B).
Four additional design principles were applied to assure
the formation of rigid-rod p-stack architecture of 1nH and
1n·2m despite the poor directionality of aromatic interactions
3) Internal crowding[8,9] was expected to promote the self-
assembly of higher hollow oligomers by steric hindrance
between bulk at the inner surface (and to prevent the
self-assembly of dimers without internal space). Internal
crowding was introduced with the alkylammonium tails;
variation of their length should vary the magnitude of in-
ternal crowding (see below for experimental support).
Figure 1. Ligand-gated opening of notional rigid-rod p-helix 1nH into ion channel 1n·2m by intercalation of DAN ligands 2 with magnified views of A) an
NDI–NDI stack to illustrate internal crowding, internal charge repulsion and flanking H-bonded chains, and B) an NDI–DAN–NDI CT complex. All
suprastructures are simplifications that are, however, consistent with experimental data and molecular models (Figure 3); stoichiometries of supramole-
cules are unknown.
6526
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6525 – 6532

Ion Channels
FULL PAPER
4) According to the ICR model,[35] internal charge repul-
sion is another key parameter to promote the self-assem-
bly of higher hollow oligomers like barrel-stave supra-
molecules.[8,9] Intermediate internal charge repulsion, de-
pendent on proximity effects, pH and ionic strength, was
taken care of by the partially deprotonated[35] alkylam-
monium tails as well.
from the HPTS assay in vesicles have already been de-
scribed in the preliminary communication on this topic.[36]
Single-channel conductance: In planar EYPC membranes,
single-channel currents could be observed only when rod 1
and ligand 2 were added together (Figure 2). The presence
of several conducting states was consistent with the supra-
molecular nature of ion channels 1n·2m. Considering the su-
prastructural complexity of the challenged synergism, the
found single-current levels were surprising homogeneous
and inert. The most frequently observed single channels
were long-lived (Figure 2C) and ohmic (Figure 2D). The re-
versal potential Vr =+4.6 mV found at a 2.0m ! 0.5m KCl
gradient gave the permeability ratio PClÀ/PK+ = 1.38. This
weak anion selectivity was in agreement with the selectivity
of ligand-gated ion channels 1n·2m determined with the
HPTS assay in spherical EYPC membranes.
The two frequently observed single-channel conductances
showed ohmic behavior (Figure 2D). The inner diameter of
the ligand-gated ion channel 1n·2m was calculated from the
dominant single-channel conductance g=94 pS using the
Hille equation with Sansomꢁs correction factor of 5.61.[37,38]
A corrected[38] inner channel diameter d=3.5 Š was ob-
tained. This value was in agreement with the inner diame-
ters of geometry-minimized molecular models of ligand-
gated ion channels 1n·2m with about twelve (14·212, d=5.4 Š)
to sixteen ligands per tetramer (14·216, d=3.0 Š, Figure 3B).
Moreover, the inner diameter d=3.5 Š of ligand-gated ion
channels 1n·2m calculated from the single-channel conduc-
tance in planar bilayers was in excellent agreement with
size-exclusion experiments in vesicles.
In the following, we first use conductance measurements
in planar lipid bilayers to characterize single ligand-gated
ion channels 1n·2m as small, long-lived, weakly anion selec-
tive and ohmic. The characteristics found are in good agree-
ment with those from functional studies in vesicles described
in the preliminary communication on this topic.[36] We then
continue with structural studies in support of a conforma-
tional change of the rigid-rod p-stack architecture, that is
the helix–barrel transition from 1nH to 1n·2m, as origin of
ligand gating to finish up with a study on internal decrowd-
ing, demonstrating that a remarkably minor change in mon-
omer structure suffices to cleanly annihilate all supramolec-
ular architecture and function.
A brief comment concerning the term “ligand gating”
may be appropriate to avoid eventual misunderstandings.
The use of this term with regard to function (i.e., ligand-
mediated increase in “channel activity”) appears to be un-
problematic in the context of this study. The specific struc-
tural implications often associated with “ligand gating” (i.e.,
ligand-mediated changes in channel conformation) are eval-
uated in this study and shown to apply within reason. Differ-
ences to “biological ligand gating” beyond completely abio-
tic architecture and gating mechanism include the use of
more than one ligand to open the channel. Moreover, the
enzyme-regulated reversibility of “biological ligand gating,”
although in principle unproblematic,[8] has so far not been
studied experimentally.
Results and Discussion
The synthesis of the target molecule 1 from commercial
starting materials in overall 13 steps has been communicat-
ed.[36] Initially, ligand gating was determined in spherical egg
yolk phosphatidylchline (EYPC) bilayers (i.e., large unila-
mellar vesicles, LUVs) by using the HPTS assay. According
to the HPTS assay, the opening of the poorly active p-heli-
ces 1nH by ligand 2 occurs in a highly cooperative manner
(n=6.5) at an effective concentration EC50 =13.7 mm. Negli-
gible activity found with DAN controls lacking either the
anionic or the hydrophobic tail as well as intercalators such
as AMP or GMP indicated that ligand-gated formation of
ion channels 1n·2m is highly selective. Inability to mediate
the efflux of large organic anions and cations indicated that
ion channels 1n·2m are small. Competition experiments re-
vealed that, according to the HPTS assay, ion channels 1n·2m
Figure 2. Planar EYPC bilayer conductance in the presence of A) 1 (5 mm
cis), B) 2 (2.4 mm cis) and C) 1 (5 mm cis) and 2 (2.4 mm cis) at +50 mV
(trans at ground) in 2m KCl. D) IV profile of two different frequent
single-channel current levels (open and closed circles) with linear curve
fit.
2À
are anion selective with an inhibition sequence SO4
>
À
NO3 ꢀ IÀ > ClÀ ꢀ BrÀ > AcOÀ > FÀ. These results
Chem. Eur. J. 2005, 11, 6525 – 6532
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6527

S. Matile et al.
Figure 3. Optimized geometry of molecular models of complexes 14·212 (left, barrel length l=38 Š, inner diameter d=5.4 Š), 14·216 (middle, l=42 Š, d=
3.0 Š) and 14·228 (right, l=45 Š, d=2.0 Š) in side (bottom) and axial view (top). p-Octiphenyl staves (dark green) and DAN ligands (red) are both in
ball-and-stick representation, NDI hoops (light green) are in wire representation, H omitted, 50% (left) to 100% (middle, right) amine protonation, 0%
sulfate protonation (14·212 is adapted from[36] with permission, Copyright 2005 Am. Chem. Soc.).
Structural studies: To minimize the probability to produce
irrelevant, misleading findings, structural studies were strict-
ly limited to conditions relevant for function, that is, com-
plexes 1n·2m at low micromolar concentrations in the pres-
ence of EYPC bilayer membranes.[8,9] Fluorescence and cir-
cular dichroism (CD) spectroscopy were compatible with
these restrictive conditions. The NDI chromophore[39] of
channel 1n and the DAN fluorophore[40] of ligand 2 were
available as intrinsic probes. The otherwise invaluable blue
emission of the p-octiphenyl stave[7–9] was completely
quenched by the NDI hoops.
The interaction of ligand 2 with lipid-bilayer membranes
was studied first (Figure 4). In the absence of rod 1, ligand 2
exhibited increasing excimer emission[40] with increasing
membrane concentration. Conventional data treatment[41]
gave a partition coefficient Kx =8.0 ꢁ 0.6”105 (Figure 4,
inset). This finding suggested that ligands 2 would approach
rod 1 in the membrane rather than in the water, and pre-
sumably not as monomers.
The self-organization of rod 1 in the presence and absence
of lipid bilayer membranes was studied next (Figure 5). The
CD spectrum of rod 1 in water and the “membrane-mimet-
ic” 2,2,2-trifluoroethanol (TFE) exhibited a strong band-I
Cotton effect with vibrational fine structure overlapping the
exciton band splitting (Figure 5A, black). The observed bi-
signate band-I Cotton effect originates from exciton cou-
Figure 4. Emission spectra of 2 (20 mm) with increasing concentration of
EYPC LUVs [0 (a) ! 330 mm EYPC (c), lex 325 nm]. Inset: Frac-
tional emission intensity I410/I343 as a function of lipid concentration with
curve fit to the White equation.[41]
pling of non-parallel N,N-polarized electric p–p* transitions
between two NDI chromophores that are positioned close
in space with a well-defined and significant twist.[39,42] The
magnitude of the observed CD amplitude around A =
À24mÀ1 cmÀ1 was comparable to that of pertinent covalent
CD model compounds like the classical 1R,2R-cyclohexane-
diamine derivative (A=De1 (391 nm) À De2 (~353 nm)).[39]
This similarity to ideal covalent positioning of perfectly
twisted proximal NDIs demonstrated that the self-organiza-
tion of rod 1 is significant. Indeed, monomeric control 1b
6528
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6525 – 6532

Ion Channels
FULL PAPER
(Figure 5A, grey) but also control rod 1a with nearly identi-
cal monomer structure (below) were almost CD silent. The
significant helical twist seen in CD spectrum of rod 1 was in
excellent agreement with the notional p-helix 1nH, the nega-
tive sign of the CD amplitude demonstrated M helicity
(Figure 1).[39,42]
helix 1nH (Figure 5B, bottom). This asymmetric change im-
plied that thermal denaturation caused the transformation
into a new thermophilic suprastructure rather than the
simple self-disorganization of notional p-helix 1nH. Thermal
denaturation was irreversible. This irreversibility offered
sample decomposition on the structural rather than supra-
structural level as alternative but naturally not further ex-
plored explanation of the observed changes in the CD spec-
tra at high temperature.
Chemical denaturation of notional p-helix 1nH was ex-
plored with guanidinium chloride (GuCl). A rapid decrease
with increasing denaturant concentration saturating at A =
À15mÀ1 cmÀ1 was observed (Figure 5D). Considering possi-
bly incomplete self-organization without denaturant, a self-
organization energy ÀDGH O ꢃ1.3 kcalmolÀ1 was calculated.
2
For comparison, protein denaturation under identical condi-
tions occurs within ÀDGH O = 5–15 kcalmolÀ1.[44,45] In con-
2
trast to thermal denaturation, the CD ratio R ꢀ1 did not
change with decreasing CD amplitude during chemical de-
naturation.
The CD spectrum of notional p-helix 1nH was independent
of presence or absence of lipid-bilayer membranes (Fig-
ure 5E). Partitioning and eventual suprastructural conforma-
tional changes upon binding to bilayer membranes were
therefore not detectable by this method.
The CD amplitude decreases with decreasing angle be-
tween the exciton-coupled transition moments, from a maxi-
mum around 708 to CD silence for parallel transition mo-
ments (Figure 6A).[42] The helix–barrel transition during the
ligand-gated opening of ion channel 1n·2m should, therefore,
be characterized by CD silencing (Figure 1). This CD silenc-
ing was observed during the formation of the purple charge-
transfer complex 1n·2m (Figure 6A; the band-I absorption
was practically not affected by the appearance of the weak
and very broad charge-transfer band between 500 and
650 nm). The CD amplitude decreased to A=À8mÀ1 cmÀ1,
an unprecedented value clearly below the one found for
both chemical and thermal denaturation (Figure 5F and 6B,
Figure 5. A) CD (top) and UV/Vis spectra (bottom) of 1 in buffer (c)
and TFE( g) compared with monomer 1b (50 mm, grey) with correla-
tion of amplitude sign with the absolute sense of twist of coupled
NDIs[39] and dependence of CD amplitude A (top) and CD ratio R (De
391 nm/De 353 nm, bottom) on B) the concentration of 1; C) the temper-
ature and the presence of increasing concentrations of D) guanidinium
chloride, E) vesicles and F) ligand
2 (compare Figure 6); constants:
*
). This comparison with denaturation experiments corrobo-
10 mm HEPES, 100 mm NaCl, pH 7.9, 258C, 200 mm EYPC (D, F). The
dotted line in B–F (top) indicates the CD amplitude of complex 1n·2m
rated that the increase in average distance between NDI
chromophores by DAN intercalation was likely to contrib-
ute to but unlikely to fully account for such a dramatic CD
silencing. The increase in CD ratio from R ꢀ1 to R=2.3
confirmed that a conformational change of the exciton-cou-
pled NDIs occurred (Figure 5F). The decrease in CD ampli-
tude caused by this conformational change was in excellent
support of the untwisting motion of the NDI stacks in the
notional p-helix 1nH during the ligand-gated opening into
barrel-stave ion channel 1n·2m (Figure 1).
(F).
The bisignate band-I Cotton effect in the CD spectrum of
rod 1 was concentration independent (Figure 5B). This find-
ing demonstrated that the expected self-assembly into no-
tional p-helix 1nH is exergonic and that the number of mono-
mers per supramolecule cannot be determined from the cM
profile.[43] Thermal denaturation of notional p-helix 1nH re-
vealed stability up to 508C followed by a steady decrease to
A = À16mÀ1 cmÀ1 at 908C. Considering the possibility of in-
complete denaturation near boiling, the found melting tem-
perature was best described as TM ꢂ708C. For comparison, a
DNA duplex of similar length has a TM = 458C (d(GT)7/
d(AC)7).[4] The first CD Cotton effect at 391 nm decreased
much more than the second one at 353 nm. The result was a
drop of the CD ratio from original values around R=1.0 to
remarkably low R=0.2 for thermally denatured notional p-
The excimer emission of membrane-bound ligand
2
(Figure 4) decreased during the formation of complex 1n·2m
(Figure 6B). This decrease was consistent with the separa-
tion of the (excited) dimers 22 during the intercalation into
notional p-helix 1nH. Both trends, that is, silencing of the ex-
citon-coupled CD of the NDIs and excimer emission of the
DANs, were combined to obtain a Job plot of complex
1n·2m. The effective mole fraction x50 ꢀ0.66 was compatible
Chem. Eur. J. 2005, 11, 6525 – 6532
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6529

S. Matile et al.
control rod 1a with a truncated methylamine tail was syn-
thesized in 15 steps (Scheme 1, and Supporting Information
for Experimental Section): The transformation of diprotect-
ed diaminopropionic acid 3 into amide 4 followed by Z de-
protection gave amine 5. As for the original functional rod
1,[36] anhydride 6 was then treated with amines 5 and 7 to
give the asymmetric NDI 8. Removal of the Z group gave
amine 9, which was coupled with the carboxylic acids along
the scaffold of rigid rod 10 and Boc-deprotected to afford
the target molecule 1a.
Figure 6. A) CD spectra of 1 (5 mm) in EYPC LUVs with increasing con-
*
centration of 2 [0 (c) ! 36 mm (a)]. B) CD amplitude of 1 ( ) and
Scheme 1. a) Boc2O, NH4HCO3, MeCN, pyridine, 16 h, RT, 96%; b) H2,
Pd(OH)2/C, MeOH, 6 h, RT, 89%; c) triethylamine, DMF, 48 h, 408C,
50%; d) H2, Pd(OH)2/C, MeOH, 6 h, RT, 90%; e) see ref. [5];
f) 1) HATU, DMF, 2,6-di-tert-butyl pyridine, 24 h, RT, 28%; 2) TFA/
CH2Cl2 1:1, 75%.
*
fluorescence emission I410/I343 of 2 ( ) in EYPC LUVs with increasing
relative concentrations of 2 and 1, respectively (10 mm HEPES, 100 mm
NaCl, pH 7.9, 258C, 200 mm EYPC).
with an onset of helix–barrel transition with 1–3 ligands per
rods as in notional ion channel 1n·23n. Helix–barrel transition
around 1–3 ligands per rods matched channel opening by
n=6.5 ligands per notional tetrameric p-helix 14H reasonably
well,[36] particularly since ligands 2 may exist as dimers in bi-
layer membranes (i.e., n ꢀ13).[43]
In the HPTS assay, ligand gating of ion channel 1n·2m was
demonstrated following the collapse of an applied pH gradi-
ent by the change in emission of the intravesiclar, pH-sensi-
tive 8-hydroxypyrene-1,3,6-trisulfonate probe in the pres-
ence of 1, 2 and both. Nearly no activity was observed for 1
without 2 (Figure 7A, d) and 2 without 1 (Figure 7A, e),
whereas 1 and 2 together exhibited significant synergistic ac-
tivity (Figure 7A, a).[36] Ligand 2 failed to activate control
rod 1a (Figure 7A, c) under identical conditions (Figure 7A,
b) and at higher concentrations (not shown).
Structural support for ligand gating by a helix–barrel tran-
sition from notional p-helix 1nH to notional p-barrel 1n·2m
was obtained from the disappearance of the exciton chirality
of twisted p-stacks of self-organized rod 1 upon addition of
ligand 2 (Figure 6A). In striking contrast, control rod 1a
with very similar monomer structure was nearly CD silent
(Figure 7B).
The maximal mole fraction xmax ꢀ0.9 suggested that full
loading with seven ligands per rod is possible. Geometry-
minimized molecular models of the original tetramers 14·212
and higher tetramers 14·216 and 14·228 confirmed that exces-
sive ligand intercalation is conceivable (Figure 3). However,
increasing hoop–stave mismatch upon excessive intercala-
tion accounted for an increasing disorder in the overloaded
14·228, with some ligands being expelled from the p-stack
during optimization. The total length of the p-barrels 1n·2m
naturally increased with increasing number of intercalators,
whereas the inner diameter decreased. The fully loaded tet-
ramer 14·228 was not hollow anymore. Molecular modeling,
therefore, suggested that ion channels 1n·2m that are open at
intermediate ligand intercalation may close in response to
an overdose.
Taken together, internal decrowding of the notional rigid-
rod p-stack architecture 1nH and 1n·2m resulted in the poor
activity expected for the contraction into dimeric complexes
1a2 without internal space and helicity. The differences in
structure and function were striking considering the similar
Internal decrowding: To probe the relevance of internal
crowding for the design of ligand-gated ion channel 1n·2m,
6530
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6525 – 6532

Ion Channels
FULL PAPER
excellent agreement found between functional and structur-
al studies (under meaningful conditions) is remarkable and
must be appreciated with appropriate reservation.
Acknowledgements
We thank D. Jeannerat, A. Pinto and J.-P. Saulnier for NMR measure-
ments, P. Perrottet and the group of F. GülaÅar for MS measurements, H.
Eder for elemental analyses, and the Swiss NSF for financial support (in-
cluding the National Research Program “Supramolecular Functional Ma-
terials” 4047–057496).
Figure 7. Comparison of A) function and B) suprastructure of 1 and 1a.
(A) Fractional HPTS emission I (lex 450 nm, lem 510 nm) as a function of
time after addition of base (DpH 0.9) followed by 2 (a, b, e: 20 mm, c, d:
0 mm) and then 1 (a, d: 1.6 mm, from ref. [36]) or 1a (b, c: 3.2 mm) at t=0 s
to EYPC–LUVs ꢄ HPTS (10 mm HEPES, 100 mm NaCl, pH 7.0; up to
100 mm 2 did not significantly increase the activity of 1a+2 in b). B) CD
spectrum of 1 (g) and 1a (c; 10 mm each, 10 mm HEPES, 100 mm
NaCl, pH 7.9, 258C, 200 mm EYPC).
[1] I. Tabushi, Y. Kuroda, K. Yokota, Tetrahedron Lett. 1982, 23, 4601–
4604.
[2] Y. Tanaka, Y. Kobuke, M. Sokabe, Angew. Chem. 1995, 107, 717–
719; Angew. Chem. Int. Ed. Engl. 1995, 34, 693–694.
[3] T. M. Fyles, D. Loock, X. Zhou, J. Am. Chem. Soc. 1998, 120, 2997–
3003.
[4] N. Sakai, B. Baumeister, S. Matile, ChemBioChem 2000, 1, 123–125.
[5] B. Baumeister, N. Sakai, S. Matile, Org. Lett. 2001, 3, 4229–4232.
[6] G. Das, P. Talukdar, S. Matile, Science 2002, 298, 1600–1602.
[7] V. Gorteau, F. Perret, G. Bollot, J. Mareda, A, N. Lazar, A. W. Co-
leman, D.-H. Tran, N. Sakai, S. Matile, J. Am. Chem. Soc. 2004, 126,
13592–13593.
[8] N. Sakai, J. Mareda, S. Matile, Acc. Chem. Res. 2005, 38, 79–87.
[9] N. Sakai, S. Matile, Chem. Commun. 2003, 2514–2523.
[10] S. Matile, Chem. Soc. Rev. 2001, 30, 158–167.
[11] R. S. Hector, M. S. Gin, Supramol. Chem. 2005, 17, 129–134.
[12] S. Matile, A. Som, N. Sordꢂ, Tetrahedron 2004, 60, 6405–6435.
[13] Y. J. Jeon, H. Kim, S. Jon, N. Selvapalam, D. H. Oh, I. Seo, C.-S.
Park, S. R. Jung, D.-S. Koh, K. Kim, J. Am. Chem. Soc. 2004, 126,
15944–15945.
[14] Synthetic Ion Channels (Ed.: U. Koert), Bioorg. Med. Chem. 2004,
12, 1277–1350, whole Issue.
[15] J. Zhang, B. Jing, S. L. Regen, J. Am. Chem. Soc. 2003, 125, 13984–
13987.
[16] J. M. Boon, B. D. Smith, Curr. Opin. Chem. Biol. 2002, 6, 749–756.
[17] G. W. Gokel, A. Mukhopadhyay, Chem. Soc. Rev. 2001, 30, 274–
286.
[18] G. J. Kirkovits, C. D. Hall, Adv. Supramol. Chem. 2000, 7, 1–47.
[19] P. Scrimin, P. Tecilla, Curr. Opin. Chem. Biol. 1999, 3, 730–735.
[20] H. Bayley, P. S. Cremer, Nature 2001, 413, 226–230.
[21] S. Terrettaz, W.-P. Ulrich, R. Guerrini, A. Verdini, H. Vogel, Angew.
Chem. 2001, 113, 1790–1793; Angew. Chem. Int. Ed. 2001, 40, 1740–
1743.
monomer structure of p-octiphenyls 1 and 1a. They can be
interpreted as support for the importance of internal crowd-
ing for the design of higher hollow barrel-stave architecture.
Molecular models were in agreement with this interpreta-
tion.
Conclusion
A colorful, surprisingly detailed and remarkably consistent
portrait of a ligand-gated ion channel emerges from a rich
collection of experimental facts on functional rigid-rod p-
stack architecture, made from scratch. All the characteristics
observed in spherical or planar lipid-bilayer membranes by
using fluorescent probes or conductance measurements
were significant and as planned. Ligand gating was found to
be highly cooperative as well as extremely selective, that is,
minor changes in both ligand and channel structure cleanly
canceled all function. The open ion channels were small,
long-lived, surprisingly homogenous, ohmic, and anion selec-
tive.
In research focusing on the design of function, it is most
important to obtain the experimental evidence for the de-
sired function. However, structural insights on at least the
two key concepts of the introduced design strategy appeared
desirable (as long as they could be secured under conditions
relevant for function). Because of their purple color, pres-
ence of charge-transfer complexes in open ion channels was
readily detectable with the naked eye. Direct experimental
support for the otherwise more elusive helix–barrel transi-
tion as structural basis of ligand gating was obtained based
on the principles of the exciton chirality method.[42] Corro-
borated by a series of control experiments, CD silencing by
an untwisting rotation of coupled NDI chromophores was in
excellent agreement with the conformational change from a
CD-active p-helix to a CD-silent barrel-stave ion channel.
Because the most stable suprastructure detected in structur-
al studies does not have to (and often does not) correspond
to the one responsible for function,[43] we reiterate that the
[22] D. W. Deamer, D. Branton, Acc. Chem. Res. 2002, 35, 817–825.
[23] R. S. Lokey, B. L. Iverson, Nature 1995, 375, 303–305.
[24] G. J. Gabriel, B. L. Iverson, J. Am. Chem. Soc. 2002, 124, 15174–
15175.
[25] J. Lee, V. Guelev, S. Sorey, D. W. Hoffman, B. L. Iverson, J. Am.
Chem. Soc. 2004, 126, 14036–14042.
[26] S. A. Vignon, T. Jarrosson, T. Iijima, H.-R. Tseng, J. M. K. Sanders,
J. F. Stoddart, J. Am. Chem. Soc. 2004, 126, 9884–9885.
[27] T. Iijima, S. A. Vignon, H. R. Tseng, T. Jarrosson, J. M. K. Sanders,
F. Marchioni, M. Venturi, E. Apostoli, V. Balzani, J. F. Stoddart,
Chem. Eur. J. 2004, 10, 6375–6392.
[28] D. G. Hamilton, N. Feeder, S. J. Teat, J. M. K. Sanders, New J. Chem.
1998, 22, 1019–1021.
[29] R. L. Furlan, S. Otto, J. K. M. Sanders, Proc. Natl. Acad. Sci. USA
2002, 99, 4801–4804.
[30] G. D. Fallon, M. A.-P. Lee, S. J. Langford, P. J. Nichols, Org. Lett.
2004, 6, 655–658.
[31] M. S. Cubberley, B. L. Iverson, J. Am. Chem. Soc. 2001, 123, 7560–
7563.
[32] F. Würthner, B. Hanke, M. Lysetska, G. Lambright, G. S. Harms,
Org. Lett. 2005, 7, 967–970.
Chem. Eur. J. 2005, 11, 6525 – 6532
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6531

S. Matile et al.
[33] K. Sugiyasu, N. Fujita, S. Shinkai, Angew. Chem. 2004, 116, 1249–
1252; Angew. Chem. Int. Ed. 2004, 43, 1229–1233.
[34] W. Zhang, D. Horoszewski, J. Decatur, C. Nuckolls, J. Am. Chem.
Soc. 2003, 125, 4870–4873.
[35] B. Baumeister, A. Som, G. Das, N. Sakai, F. Vilbois, D. Gerard, S. P.
Shahi, S. Matile, Helv. Chim. Acta 2002, 85, 2740–2753.
[36] P. Talukdar, G. Bollot, J. Mareda, N. Sakai, S. Matile, J. Am. Chem.
Soc. 2005, 127, 6528–6529.
[37] B. Hille, Ionic Channels of Excitable Membrane, 2nd ed., Sinauer
Associate, Sunderland, MA, 1992.
[38] O. S. Smart, J. Breed, G. R. Smith, M. S. P. Sansom, Biophys. J. 1997,
72, 1109–1126.
[40] J. Reichwagen, H. Hopf, A. Del Guerzo, J. P. Desvergne, H. Bouas-
Laurent, Org. Lett. 2004, 6, 1899–1902.
[41] S. H. White, W. C. Wimley, A. S. Ladokhin, K. Hristova, Methods
Enzymol. 1998, 295, 62–87.
[42] K. Nakanishi, N. Berova, The Exciton Chirality Method. In Circu-
lar Dichroism—Principles and Applications (Eds: K. Nakanishi, N.
Berova, R. W. Woody), VCH, Weinheim, Germany, 1994, 361–398.
[43] S. Litvinchuk, G. Bollot, J. Mareda, A. Som, D. Ronan, M. R. Shah,
P. Perrottet, N. Sakai, S. Matile, J. Am. Chem. Soc. 2004, 126,
10067–10075.
[44] F. Ahmad, C. C. Bigelow, J. Biol. Chem. 1982, 257, 12935–12938.
[45] G. Das, S. Matile, Chirality 2001, 13, 170–176.
[39] J. Gawronski, M. Brzostowska, K. Kacprzak, H. Kolbon, P. Skowro-
nek, Chirality 2000, 12, 263–268.
Received: May 9, 2005
Published online: August 24, 2005
6532
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2005, 11, 6525 – 6532