Rigid-rod anion-π slides for multiion hopping across lipid bilayers

Article abstract of DOI:10.1039/b708337h

Shape-persistent oligo-p-phenylene-N,N-naphthalenediimide (O-NDI) rods are introduced as anion-π slides for chloride-selective multiion hopping across lipid bilayers. Results from end-group engineering and covalent capture as O-NDI hairpins suggested that

Full text of DOI:10.1039/b708337h

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Rigid-rod anion–p slides for multiion hopping across lipid bilayers†‡
Virginie Gorteau, Guillaume Bollot, Jiri Mareda and Stefan Matile*
Received 1st June 2007, Accepted 20th July 2007
First published as an Advance Article on the web 13th August 2007
DOI: 10.1039/b708337h
Shape-persistent oligo-p-phenylene-N,N-naphthalenediimide (O-NDI) rods are introduced as anion–p
slides for chloride-selective multiion hopping across lipid bilayers. Results from end-group engineering
and covalent capture as O-NDI hairpins suggested that self-assembly into transmembrane O-NDI
bundles is essential for activity. A halide topology VI (Cl > F > Br ∼ I, Cl/Br ∼ Cl/I > 7) implied
strong anion binding along the anion–p slides with relatively weak contributions from size exclusion
(
F ≥ OAc). Anomalous mole fraction effects (AMFE) supported the occurrence of multiion hopping
along the p-acidic O-NDI rods. The existence of anion–p interactions was corroborated by high-level
ab initio and DFT calculations. The latter revealed positive NDI quadrupole moments far beyond the
hexafluorobenzene standard. Computational studies further suggested that anion binding occurs at the
confined, p-acidic edges of the sticky NDI surface and is influenced by the nature of the phenyl spacer
between two NDIs. With regard to methods development, a detailed analysis of the detection of ion
selectivity with the HPTS assay including AMFE in vesicles is provided.
Introduction
1
,2
Multiion hopping
enables biological ion channels to act
3,4
1–14
selectively and fast.
This is remarkable because the tight
ion binding needed for high selectivity alone will slow down the
transport of this recognized ion, whereas the “slippery” surfaces
needed for fast diffusion are naturally not selective. The elegant
solution of this dilemma is to align multiple binding sites next
to each other for multiion hopping (Fig. 1). In the resting state,
permanent occupation of some binding sites will hinder the
passage of other ions, whereas complete occupation is unfavorable
because of charge repulsion. In a channel with three active sites,
the two peripheral sites will be filled and the central one will be
empty (Fig. 1A). If now an ion approaches this channel to move
across the membrane, the ion in the first active site will be repelled
and move to the free central active site (Fig. 1B, a and b). The
loading of the central binding site will repel the ion bound at the
other periphery and cause its release (Fig. 1C, c and d). Overall,
multiion hopping enables instantaneous release of strongly bound
ions on one side of the membrane in response to anion binding on
the other side.
Fig. 1 Multiion hopping solves the dilemma of ion channels to act in
a fast and selective manner. (A) The resting state of a chloride channel
with three aligned chloride binding sites may contain two permanently
bound chlorides. (B) Repelled by the approaching external chloride (a),
the chloride in the external binding site hops to the central site (b) to repel
the chloride in the internal active site (c) to cause internal chloride release
The classical test for the existence of multiion hopping is
1,2
the anomalous mole fraction effect (AMFE). AMFE refers
to lower than expected activity found with mixtures of ions
compared to pure ions. Its occurrence indicates that permanent
immobilization of the better binding ion to one single site of the
channel is insufficient for significant transport of this ion (and
may even hinder the transport of the less efficient ion). This, in
turn, indicates that the occupation of multiple sites along the
channel is required for efficient transport of the selected ion,
(
d). (C) Repelled by the external chloride (e), the central chloride then
hops to the internal site (f) to return to the resting state (A). Overall, the
chloride bound externally and the chloride released internally are not the
same.
i.e., the existence of multiion hopping (Fig. 1). AMFE has been
identified for many biological anion channels, including the cystic
fibrosis transmembrane conductance regulator (CFTR) chloride
Department of Organic Chemistry, University of Geneva, Geneva, Switzer-
land. E-mail: stefan.matile@chiorg.unige.ch; Fax: +41 22 379 5123; Tel: +41
5–7
8
2+
channel, the voltage-dependent chloride channel ClC-0, Ca -
2
†
‡
2 379 6523
9
10
activated chloride channels (ClCaCs), and so on.
The HTML version of this article has been enhanced with colour images.
Electronic supplementary information (ESI) available: Experimental
Multiion hopping has been confirmed on the structural level for
details. See DOI: 10.1039/b708337h
cationic arginine clusters within the 16-stranded b-barrel of the
3
000 | Org. Biomol. Chem., 2007, 5, 3000–3012
This journal is © The Royal Society of Chemistry 2007

View Article Online
11
porin Omp32 and three neutral, macrodipole-supported anion
12
binding sites lined up within the complex a-helix bundles of ClCs.
The latter is reminiscent of the four neutral cation binding sites of
13,14
potassium channels.
Remarkably, the creation of synthetic channels, carriers and
transporters with cation and anion selectivity from scratch is pos-
sible and has been demonstrated in many variations. For selected
recent reviews, please see references 15–20. Cation selectivity be-
came accessible with oligoether or oligocarbonyl arrays (including
1
7
crown ethers) on scaffolds reaching from flexible alkyl chains
21
22
23
to more rigid dimeric steroids, cucurbiturils, p-octiphenyls,
24
25
a-helices and the b-helical gramicidin. Interactions beyond
multivalent oxygen ligands used to create cation selectivity include
26
ion pairing along multiple carboxylate clusters and counterion-
mediated, pH-dependent inversion of the anion/cation selectivity
26
of oligoarginine arrays.
The more demanding anion selectivity has attracted recent
attention because of the possible medicinal use as antibiotics
and antitumor drugs, or to treat anion channelopathies, dis-
eases that are caused by malfunction of anion channels such
as CFTR, ClCaCs, ClCs, glycine and GABA receptors, and
so on (cystic fibrosis, Best disease, Startle disease, Angelman
syndrome, Bartter syndrome, myotonia, nephrolithiasis, osteopet-
Fig. 2 The concept of (A) cation–p and (B) anion–p slides for multiion
hopping as outlined in Fig. 1. Cation–p slides have been described
previously, the more demanding anion–p slides are introduced in this
48
report.
15,27–30
rosis, and more).
Biomimetic approaches to anion selectivity
have focused on fragments and derivatives of anion channel
27,28,31
29,30
3,48
proteins
cationic steroid antibiotics,
antibiotics,
or natural products such as squalamine
and other
cations.
Application of the Eisenman theory to anions gives
32,33
4
magainin and cationic peptide
seven meaningful halide sequences. One extreme is the very
3
4–37
38,39
−
−
−
−
or prodigiosin.
Anion recognition has been
common “Wright–Diamond” sequence I (I > Br > Cl > F ).
In halide I, also referred to as the Hofmeister series, selectivity
is determined exclusively by the cost of anion dehydration. At
accomplished with amide or urea arrays on scaffolds such as
4
0,41
42
macrocyclic (calixarene),
cholates.
linear or branched oligobenzyls or
Alterative to H-bonding approaches, ion pairing
along arrays of ammonium or guanidinium cations has been suc-
15,43,44
−
−
>
the other end is the exceptionally rare halide VII (F > Cl
−
−
Br > I ). Here, selectivity is determined exclusively by anion
binding to the channel. Intermediate selectivity sequences reflect
a specific balance between dehydration penalty and binding in
29,30,32,33
34
35
36
cessful on steroid,
b-peptide, peptoid, polynorbornene,
,3-polyphenyleneethynylene, cyclodextrin, or rigid-rod p-
37
45
1
26,46
oligophenyl scaffolds.
the channel. Anion selectivity sequences can be described as
topologies. However, many mismatches of size (e.g., OAc
F ) and hydration energy (e.g., OAc ≈ F ) limit the usefulness
of selectivity topologies for anions beyond halides. The direct
correlation of anion selectivity with hydration energy is often
preferable for this reason.
Complementary to the cation–p interactions in the established
rigid-rod cation–p slides, the use of anion–p interactions to create
−
The use of synthetic organic chemistry to create new ion
channels from scratch is attractive because of the possibility
to expand the chemistry at work beyond the limitations of
ꢀ
−
−
−
1
5–20
biology.
This has been accomplished for cation selectivity
47
3,4,9
with fixed and flexible cation–p interactions within calixarenes
and ligand-assembled p-septiphenyl rods (Fig. 2A), respectively.
As predicted theoretically, flexible cation–p interactions within
48
49
supramolecular arene arrays were required to obtain the biolog-
anion–p slides is more challenging because little experimental
+
+
50,51
ically relevant Eisenman IV selectivity sequence (K > Rb
>
support for this interaction is available in solution
despite
+
+
+
Cs > Na ∼ Li ). This sequence is difficult to obtain because it lies
quite extensive data from X-ray crystallography as well as from
5
2–54
55
midway between the easily accessible Eisenman I, dominated by
modeling.
Here, we introduce rigid-rod molecules with mul-
+
+
+
+
+
56–62
cation dehydration (Cs > Rb > K > Na > Li ), and Eisenman
tiple p-acidic naphthalenediimide (NDI)
binding sites along
+
+
+
XI, dominated by cation binding in the channel (Li > Na > K >
the transmembrane scaffold for this purpose (Fig. 2B). These
shape-persistent oligo-p-phenylene-N,N-naphthalenediimide (O-
NDI) rods are shown to preferably transport anions with relatively
high hydration energy. Covalent capture as O-NDI hairpins and
+
+
3
Rb > Cs ).
Anion selectivity beyond anion/cation selectivity is usually de-
scribed in anion selectivity sequences. Anion selectivity sequences
report the preference of anion selective channels/transporters
15–26
presumably unprecedented
AMFE are used to confirm strong
3,4,9
3
between different anions.
In the Eisenman theory, ion se-
binding and multiion hopping along transmembrane O-NDI p-
slides as the origins of these attractive characteristics. We also
report ab initio and DFT calculations of NDI and several model
systems based on N-substituted NDIs. In particular, we have
computed several complexes of chloride and bromide anions with
these model systems. Parts of this study have been reported in the
lectivity sequences are thought to originate from a balance
between dehydration penalty and energy gain from binding by
the transporter. With direct correlation of cation radius and
hydration energy, selectivity sequences can be plotted as Eisenman
topologies, where ion selectivity is given as a function of the
reciprocal ion radius. This naturally works best for group-one
55
preliminary communication on this topic.
This journal is © The Royal Society of Chemistry 2007
Org. Biomol. Chem., 2007, 5, 3000–3012 | 3001

View Article Online
Results and discussion
Synthesis
Rigid O-NDI rods 1–5 were prepared from naphthalenedianhy-
dride 6, aryl diamine 7 and alkyl amine 8 (Scheme 1). Treatment of
dianhydride 6 with excess diamine 7 gave the central NDI module
Design
The rigid O-NDI rods 1–5 were designed to explore the possibility
of multiion hopping along anion–p slides (Scheme 1 and Fig. 1
and 2). For this purpose, all rigid rods contain three p-acidic NDI
binding sites along their scaffold. The phenyl spacers, with their
planes perpendicular to the NDI planes, were fully methylated
to assure high solubility. Moreover, these methyls should also
increase the overall quadrupole moment of the O-NDI rod to
better attract anions and confine the p-acidic edges of the NDI
surface to possibly match the diameter of chloride and possibly
weaken anion–p interactions with the larger bromide or iodide
9
1
. Controlled introduction of amine 8 provided the imide terminus
0. Coupling of imide termini 10 and central module 9 already
afforded the desired rigid-rod scaffold.
Modifications of the O-NDI termini were initiated with the
removal of the Z-protection of rod 11. Attempts to isolate
the asymmetric monoamine intermediate under mild conditions
failed. The obtained diamine 12 was thus elongated with Boc-
protected glycine. “Ultramild” partial Boc-removal from the
obtained rod 1 gave a roughly statistical mixture of rods 1, 2
and 3, which were readily isolated and purified. Availability of
the asymmetric rod 2 was essential to prepare O-NDI hairpins.
Coupling with diacid 13 gave the neutral hairpin 4, deprotection
with TFA the dicationic hairpin 5.
(
see section on molecular modeling). To ensure a transmembrane
rod orientation, we made sure that the overall length of the rigid
O-NDI rods (3.4 nm) matched the thickness of the hydrophobic
core of EYPC bilayer membranes.
63,64
The O-NDI rods 1–3 have differently charged termini. Rod 1 is
uncharged, rod 2 has one cationic and one uncharged terminus,
and rod 3 has two cationic termini. The neutral O-NDI hairpin 4
corresponds to a dimeric bundle of neutral rods 1. The dicationic
O-NDI hairpin 5 corresponds to a parallel dimer of rods 2 with
one cationic and one neutral terminus. This series was prepared to
Activity
The activity of rigid-rod anion–p slides 1–5 was determined in
egg yolk phosphatidylcholine large unilamellar vesicles (EYPC
LUVs) that were loaded with the pH-sensitive fluorescent probe 8-
23,24
23,46,48,55,65
probe the importance of terminal charges for activity
and to
hydroxy-1,3,6-pyrenetrisulfonate (HPTS) at neutral pH.
explore the dependence of anion selectivity on the lateral proximity
of self-assembled transmembrane rods.
Then, a pH gradient DpH = 0.9 was applied with a base pulse, and
48
the changes in HPTS emission were recorded as a function of time
◦
Scheme 1 Synthesis of rigid-rod anion–p slides 1–5. (a) N,N-Dimethylacetamide, 135 C, 12 h, 90%. (b) 1. H
2
O, pH 6.4, reflux; 2. AcOH, 88% (overall).
◦
◦
(
c) N,N-Dimethylacetamide, 135 C, 12 h, 57%. (d) TFA, 50 C, 2 h, 61%. (e) Boc-Gly-OH, HBTU, TEA, rt, 2 h, 54%. (f) 2% TFA, CH
4% 2, 30% 3 (conversion yield). (g) 1. 13, oxalylchloride, 2. N-hydroxysuccinimide, 3. 2, TEA, 0 → 20 C, 20 min, 36%. (h) TFA, rt, 2 h, 79%.
2
Cl
2
, rt, 50 min,
◦
6
3
002 | Org. Biomol. Chem., 2007, 5, 3000–3012
This journal is © The Royal Society of Chemistry 2007

View Article Online
(
Fig. 3). Rods 1–5 were added next, and the velocity of the decay of
These trends supported the existence of an active structure with
membrane-spanning rods. The highest activity for rod 2 with one
charged terminus implied that sufficient water solubility to reach
the membrane on the one hand and favored partitioning and
translocation of the hydrophobic terminus to adapt a transmem-
brane active structure on the other hand are both important for
function. Insufficient water solubility with partial precipitation
before reaching the membrane could account for the decreasing
the pH gradient was monitored continuously following the change
in HPTS emission. At the end of each experiment, an excess of
gramicidin A, melittin, or triton X-100 was added for calibration.
Each experiment was recorded in two channels at different HPTS
excitation to ratiometrically eliminate eventual effects from origins
other than change in pH. Each experiment was repeated without
rods, and the normalized background traces were subtracted to
unambiguously extract the changes caused by rigid-rod anion–p
slides. Fig. 3 shows representative curves for the activity of rod 1
at different concentrations after the elimination of all unrelated
effects by background subtraction.
66
activity of rod 1 without charged termini. However, further
reduced activity of rod 3 with two charged termini and thus
unproblematic solubility in water identified the translocation of
one rod terminus, i.e., transmembrane rod orientation, as most
23
critical for activity.
The overall disappointing activities of O-NDI hairpins 4 (Fig. 4,
) and 5 (Fig. 4, ∇) were in agreement with these interpretations
᭹
in support of transmembrane active structures. The activity of
the neutral O-NDI hairpin 4 was a bit weaker than that of
neutral O-NDI rod 1, suggesting that the solubility in water may
have further decreased. The activity of doubly charged hairpin
5
was exceptionally weak. This finding was somewhat surprising
because hairpin 5 could span the membrane with both positive
charges resting at the outer membrane–water interface as with the
most active monocationic rod 2 (Scheme 1). The poor activity
of dicationic hairpin 5 may thus indicate that the translocation
of the oligoethyleneglycol loop across the membrane to adapt
a transmembrane hairpin with external charges is unfavorable.
Alternative explanations include self-assembly of dicationic O-
NDI (bola)amphiphiles 5 into micelles or vesicles rather than
binding to the lipid bilayer, and so on.
Fig. 3 Dependence of activity on the concentration of rigid O-NDI rod
1
. Fractional HPTS emission intensity I
shown during the addition of first NaOH (20 ll, 0.5 M), then rod 1
0.15 (a), 0.375 (b), 1.5 (c), 2.5 (d) and 15 lM (e) final concentration)
F
(kex 450 nm, kem 510 nm) is
(
to EYPC-LUVs⊃HPTS (1 mM HPTS, 10 mM HEPES, 100 mM NaCl,
Pseudo-linear behavior at low concentrations and saturation
behavior at high concentrations made Hill analysis of the Hill plots
◦
pH 7.0, 25 C, calibrated by final addition of excess gramicidin A or triton
X-100). Negative controls without rod were subtracted after calibration.
65–68
of rods 1–4 rather unrevealing.
from increasing rod precipitation at increasing concentrations
rather than partitioning into the membrane). The former indi-
The latter indicated interference
(
Hill plots
cated either a monomeric active structure or, more likely (see
below), the exergonic self-assembly into thermodynamically stable
active supramolecules, i.e., supramolecular anion–p slides (Fig. 2).
The HPTS assay was used to determine the dependence of the
activity on the concentration of rods 1–5. The highest activity was
found for the singly charged O-NDI rod 2 (Fig. 4, ᭺), followed by
the uncharged rod 1 (Fig. 4, ꢀ), whereas dication 3 was the least
active of all monomeric rods (Fig. 4, X).
Anion selectivity
The decay of a pH gradient measured in the HPTS assay can occur
+
n+
−
n− 65
by facilitated cation (H /M ) or anion exchange (OH /X ).
The activity of anion–p slide 1 did not change in response to the
+
+
+
+
replacement of external Na by K , Rb or Cs . This insensitivity
to external cation exchange suggested that rigid O-NDI rod 1 does
55
not transport cations. Anion–p slide 1, however, did respond to
external anion exchange (Fig. 5), implying the selective transport
of anions rather than cations. The complementary sensitivity of
cation–p slides (Fig. 2A) to cation and insensitivity to anion
48
exchange has been demonstrated previously.
The different activity of O-NDI rod 1 found with different
external anions provided access to anion selectivity sequences
(Fig. 5). To really understand ion selectivity sequences obtained
from the HPTS assay, a careful analysis of all situations was
Fig. 4 Hill plot of rods 1 (ꢀ), 2 (᭺), 3 (X), 4 (᭹) and 5 (ꢁ). The dependence
of the fractional activity Y on monomer concentration was determined in
EYPC-LUVs⊃HPTS as specified in Fig. 3. The term fractional activity Y
65,69
unavoidable.
The situation in symmetric NaCl was unprob-
lematic (Scheme 2a). Application of a base pulse before the
p-slide will create an additional pH gradient (Scheme 2b).
Addition of an anion exchanger such as rod 1 will then initiate
F
is used to compare fractional HPTS emissions I at a given time (here 200
s) after the start of the transport experiment, compare Fig. 3 for changes
of I with time.
−
−
−
−
F
transmembrane Cl /Cl and Cl /OH antiport (Scheme 2c).
This journal is © The Royal Society of Chemistry 2007
Org. Biomol. Chem., 2007, 5, 3000–3012 | 3003

View Article Online
The former is unproductive and invisible, the latter dissipates
the applied pH gradient and is reported by intravesicular pH
probes as increase in pH (Fig 5Ab). Rod addition before base
pulse will immediately turn on the unproductive and invisible
−
−
−
−
Cl /Cl antiport, together with similarly unproductive OH /Cl
−
−
−
−
antiport, Cl /OH antiport and OH /OH antiport (Scheme 2d).
Productive Cl /OH antiport will occur in response to a base
pulse, and OH influx is detectable as a biphasic increase in pH
−
−
−
(Scheme 2c, Fig. 5Bb). With identical final systems, the apparent
activities are thus expected to be independent of the sequence of
addition (Scheme 2c). However, slide addition before base pulse
decouples the slow formation of the active structure (partitioning,
intervesicular transfer, translocation, and so on) from the much
faster actual anion exchange. For this reason, the observed initial
rate is much faster for slide addition before base pulse than base
pulse before slide addition (Fig. 5C). After this initial fast anion
exchange along a high number of already preformed slides, the
kinetics for slide before base and base before slide addition were
perfectly superimposable.
The reason why the two curves in Fig. 5C are so nicely
superimposable after the second base pulse may not be easily
appreciated. To do so, it may be best to assume intervesicular
transfer as the rate limiting process of slide formation. The
observed fractional activity would then reflect, at any given time,
the number of vesicles containing p-slides. This number will
steadily increase after slide addition, independent of the time the
base pulse is applied (Fig. 5C). This increasing number of p-slide-
rich vesicles is naturally invisible without base pulse (Fig. 5C,
solid). Application of the base pulse after slide addition will then
cause an instantaneous pH jump corresponding to the number
of p-slide-rich vesicles and afterwards continue to rise slowly
following the kinetics of intervesicular slide transfer. We reiterate
all changes in HPTS emission found in the HPTS assay including
the initial jump after base pulse in the presence of slide 1 could
−
−
Fig. 5 Dependence of the anion selectivity of rod 1 in vesicles with internal
originate not only from rapid OH /X antiport but also from
−
−
−
Cl on the sequence of addition. Fractional emission I
F
(with respect to
HPTS efflux or cation antiport. The occurrence of OH /X
antiport was revealed by the dependence of all changes, including
this initial jump (Fig. 5B), on the nature of the extravesicular
−
Cl ) during the addition of either (A) first NaOH (20 ll 0.5 M) and
then rod 1 (1.5 lM final concentration) or (B) first 1 and then NaOH
to EYPC-LUVs⊃HPTS exposed to varied external anions (sodium salts,
conditions as in Fig. 3), with final lysis for calibration. (C) Overlay of
curves b (with NaCl) from A (dotted) and B (solid).
n−
anion X .
To determine anion selectivity sequences, isoosmolar external
−
−
65,69
Cl → X exchange was used.
This exchange produces
−
−
transmembrane Cl /X gradients that significantly complicate the
situation. Different processes are expected to occur with external
non-basic anions that are transported by the slide (X , Scheme 3),
, Scheme 4), and with unrecognized
anions that are not transported at all by the slide (X
HPTS assays for anion selectivity with non-basic anions X
that are transported by the slide of interest provide the following
information (Scheme 3). Base pulse before slide addition will add a
−
−
with weakly basic anions (X
B
−
U
, Scheme 5).
−
−
−
pH gradient to the Cl /X gradients (Scheme 3b). Slide addition
−
−
−
−
will then turn on Cl /X and Cl /OH antiport (Scheme 3c).
−
With the concentrations of X (100 mM) more than three orders
−
−
−
of magnitude higher than that of OH (pH ∼ 8), Cl /X
antiport will be much faster than Cl /OH antiport, even at high
−
−
Scheme 2 The HPTS assay for anion antiporters in symmetric NaCl
is independent of the sequence of addition: base pulse before antiporter
addition (top) and antiporter addition before base pulse (bottom) provide
−
−
−
−
OH > X selectivity. The Cl /X gradients will thus disappear
well before the occurrence of significant OH transport. With
overall much higher concentration of X (extravesicular exceeds
intravesicular volume by far), Cl concentrations are negligible in
−
−
−
−
access to the detection of Cl /OH antiport as increase in intravesicular
pH (vesicles = solid circles; antiporter = filled circles; added base = dashed
circles, see text for discussion).
−
transmembrane equilibrium. The process reported in the HPTS
3
004 | Org. Biomol. Chem., 2007, 5, 3000–3012
This journal is © The Royal Society of Chemistry 2007

View Article Online
−
Scheme 3 HPTS assay for anion selectivity with non-basic anions X .
−
−
With internal NaCl and external NaX, X /OH antiport is reported as
increase in intravesicular pH (d), independent of sequence of addition, i.e.,
base pulse before antiporter addition (top) and antiporter addition before
base pulse (bottom; vesicles = empty circles; antiporter = filled circles;
added base = dashed circles, other anion gradients are in italics, see text
for discussion).
Scheme 5 HPTS assay for anion selectivity with unrecognized anions
XU , i.e., anions that are not transported by the antiporter of interest.
−
−
−
In this case, Cl /OH antiport is reported independent of sequence of
−
−
addition (c, f). Unrecognized anions XU can be identified by OH influx
without applied pH gradient (d). The influence of the additional Cl (c) or
OH gradients (e, f) appears less important.
−
−
Addition of rod 1 to HF-acidified vesicles naturally removed this
HF-mediated pH gradient and made the HPTS emission return to
normal before application of the base pulse (Fig. 5Ba and 5Be).
The consequences of the internal acidification with weakly basic
−
anions X
B
on HPTS assays for anion selectivity were dependent
of the sequence of addition (Scheme 4). With the base pulse
applied first, the final effective pH gradient with weakly basic
−
−
anions X
B
(Scheme 4g) is larger than with non-basic anions X
(
Scheme 3d). With the antiporter added first, however, the final
−
effective pH gradients with weakly basic anions X
and non-basic anions X (Scheme 3d) are identical. Results on
ion selectivity with weakly and non-basic anions X
B
(Scheme 4j)
−
−
−
B
and X
obtained by slide addition before base pulse should thus be
directly comparable (Fig. 5B), whereas results obtained by base
pulse before slide addition may give activities for weakly basic
−
anions X
B
that appear higher than they are in reality (Fig. 5A;
−
please consult the ESI‡ for a more detailed discussion of the X
anomaly).
B
Scheme 4 HPTS assay for anion selectivity with weakly basic anions
Another class of anions that required special attention were
−
X
B
. HX
B
influx before base pulse and antiporter addition (top) gives
−
unrecognized anions X
U
, i.e., anions that are not transported
−
−
overestimates of X
dition (middle) but comparable data on X
addition before base pulse (bottom). X
B
/OH antiport for base pulse before antiporter ad-
by the anion antiporter (Scheme 5). In this case, addition of first
−
−
B
/OH antiport for antiporter
−
−
base and then slide will create a system where the internal Cl
B
are identified by low initial
−
is not exchanged by the unrecognized anion X
U
(Scheme 5a–c).
emission (c, d) that disappears in response to antiporter addition without
base pulse (h, i).
−
−
Activities found in the HPTS assay will thus report on Cl /OH
antiport. Often, this situation can be clarified by reversal of the
−
−
−
−
assay for anion selectivity is thus X /OH antiport (Scheme 3d).
Addition of the transporter before the base pulse leads to the
same situation (Scheme 3e and f). Independence of activities
on sequence of addition found with rigid O-NDI rod 1 for
sequence of addition (Scheme 5d–f). Cl /X
U
antiport excluded,
−
−
Cl /OH antiport could occur (Scheme 5d). The resulting increase
of intravesicular pH is detectable in the HPTS assay (Scheme 5e).
2
−
This formal proton pumping was observed with slide 1 for SO
4
,
−
−
−
−
−
−
Cl /OH , Br /OH and I /OH antiport was in agreement with
this interpretation, anion basicity is thought to account for the
although the effect was relatively weak (Fig. 5Bf; much stronger
proton pumping was confirmed to occur with other transporters
−
−
2−
−
−
70
dependence of F /OH antiport on sequence of addition (Fig. 5).
The initial HPTS emission in vesicles with weakly basic external
anions X
non-basic anions X (Fig. 5Ba). We interpreted this phenomenon
as passive influx of HX
exchange (Scheme 4). This internal acidification was observed for
weak bases such as F and OAc but not for non-basic anions such
as Br , I or ClO
in the presence of SO
4
and also NO
3
and ClO
4
). The activities
2
−
−
found for rod 1 with external SO
They thus report both on Cl /OH antiport and should not be
4
and Cl were indeed identical.
−
−
−
−
−
B
such as F and OAc was much weaker than that with
−
2−
confused with SO
4
selectivity (Fig. 5B, b and f).
−
−
B
in response to external Cl → X
B
Taken together, the lessons learned concerning HPTS assays
for anion selectivity are as follows. (a) Activities obtained by
−
−
−
external exchange with non-basic anions X report directly on
−
−
−
2−
−
−
4
(Fig. 5a and 5e; SO
4
is special, see Scheme 5).
X /OH antiport and are independent on sequence of addition
This journal is © The Royal Society of Chemistry 2007
Org. Biomol. Chem., 2007, 5, 3000–3012 | 3005

View Article Online
−
(
Scheme 2 and 3). (b) X
B
anomaly: activities obtained by external
−
−
−
exchange with weakly basic anions X
give results comparable with non-basic anions X only if the
antiporter is added before the base (Fig. 5Ba and 5Be, base
B
such as F or OAc
−
pulse before antiporter addition gives overestimates, Scheme 4,
−
Fig. 5Aa and 5Ae). (c) X
U
−
anomaly: results with unrecognized
−
−
2−
anions X
U
report on Cl /OH antiport (Scheme 5, here SO
4
,
Fig. 5Bf). We caution that this analysis is necessarily incomplete,
blockage by competitive and permeant anions, for example, is not
9,71,72
considered.
With these considerations in mind, the anion selectivity se-
quences of O-NDI rods were determined from the dependence
of the fractional activity Y relative to Cl− on the reciprocal
anion radius (Fig. 6) or the anion hydration energy (Fig. 7).
Experimental evidence suggested that permeability ratios from
Goldman–Hodgkin–Katz analysis of planar bilayer conductance
experiments and results on ion selectivity from the HPTS assay
Fig. 7 Anion selectivity of rods 1 (A, ꢀ) and 4 (B, ᭹). Fractional activity Y
−
(
relative to Cl ) as a function of the anion hydration energy was determined
−
−
with the HPTS assay with initial intravesicular Cl . Data for F and
−
OAc are in grey to highlight that they represent systematic overestimates
(
Scheme 4). For original data, see Fig. 5A and 8B (rod addition after base
46
are directly comparable. The activities of O-NDI rod 1 found
in HPTS assays decreased with increasing halide radius (Fig. 5A
and 6A, ꢀ). The uncertainty in the measure of the transport of
pulse).
−
the F made it generally difficult to discriminate between VII
−
−
−
−
−
−
−
−
(
F > Cl > Br > I ), halide VI (Cl > F > Br > I )
−
−
−
−
55
and halide V selectivity (Cl > Br > F > I ). However, all
these exceptionally rare
ubiquitous, dehydration-dominated Hofmeiser series or halide
I. According to the Eisenman theory, this finding implied
exceptionally strong anion binding along the rigid-rod O-NDI
p-slide 1. This interpretation was supported by the results for
1–10
halide topologies are opposite to the
3
,4
9
acetate. This weakly basic anion is a classical probe because it
is large but still difficult to dehydrate. High activity found for
−
−
OAc /OH exchange excluded size exclusion effects (Fig. 6A,
) and confirmed strong anion binding to slide 1 as origin of
ꢀ
selectivity (Fig. 7A, ꢀ).
Fig. 8 Anion selectivity of (A) rod 1 in vesicles with internal Br− and
−
(
B) rod 4 in vesicles with internal Cl . Fractional emission I
F
(relative to
−
Cl ) during the addition of NaOH (20 ll 0.5 M) and then 1 (1.5 lM)
or 4 (2.5 lM) to EYPC-LUVs⊃HPTS exposed to varied external anions
(
sodium salts, as in Fig. 3).
Fig. 6 Anion selectivity of rods 1 (A, ꢀ, ꢂ), 2 (B, ᭺) and 4 (B, ᭹).
−
−
found with internal Cl (Fig. 6A, ꢀ vs. ꢁ). Particularly noteworthy
Fractional activity Y (relative to Cl ) as a function of the reciprocal anion
−
−
−
−
was the independence of Cl > Br selectivity on internal Cl →
radius was determined with the HPTS assay with initial intravesicular Cl
(
−
−
−
−
Br exchange. It supported the interpretation that the presence or
ꢀ, ᭺, ᭹) and Br (ꢂ). Data for F and OAc are in grey to highlight that
−
−
they represent systematic overestimates (Scheme 4). For original data, see
Fig. 5A, 8A and 8B (rod addition after base pulse).
the absence of initial transmembrane Cl and Br gradients has
little influence on the velocity of OH /X exchange reported in
the HPTS assay (Scheme 3). Specifically, this finding supported
−
−
−
The dependence of the effective selectivity sequence determined
in the HTPS assay on the initial intravesicular anion was de-
that, after addition of rod 1, the replacement of internal Cl and
−
−
−
−
Br by external X is much faster than OH /X exchange because
−
−
−
−
termined next. For this purpose, internal Cl was replaced by
the concentration of X exceeds that of OH and Cl by far.
The anion selectivity of the most active cationic p-slide 2 with
one cationic terminus was determined as described in Fig. 5A for
−
internal Br , and anion selectivity was determined by external
anion exchange (Fig. 8A). The effective selectivity sequence found
−
55
with internal Br was nearly identical with the selectivity sequence
rod 1 (Fig. 6B, ᭺). The selectivity topology revealed decreasing
3
006 | Org. Biomol. Chem., 2007, 5, 3000–3012
This journal is © The Royal Society of Chemistry 2007

View Article Online
−
−
anion selectivity of cationic rod 2 (Cl /I = 1.9) compared to
−
−
neutral rod 1 (Cl /I = 7.6, Fig. 6A, ꢀ vs. Fig. 6B, ᭺).
Contrary to its poor activity (Fig. 4), the neutral O-NDI
hairpin 4 exhibited excellent anion selectivity (Fig. 8B). Compared
to the selectivity sequence of the neutral O-NDI monomer 1
determined under identical conditions (Fig. 5A), the reduced
−
−
−
activity found with F , OAc as well as Br was most intriguing
−
−
−
−
−
−
(
1: Cl /Br = 1.5, 4: Cl /Br ∼ Cl /I > 7). Considering the
−
intrinsic overestimate with F (Scheme 4), O-NDI hairpin 4
emerged as a very selective chloride–p slide (Fig. 6B and 7B, ᭹).
−
−
−
The found sequence was similar to a halide VI (Cl ꢀF ≥ Br
I ). Reduced activity with both F and OAc suggested that
the origin of the found chloride selectivity of dimer 4 is mainly
−
−
−
∼
−
−
energetic, although F > OAc may hint at contributions from
size exclusion. Dominant energetics would suggest that anion
binding along hairpin 4 is slightly weaker than anion binding
along rod 1, emerging sterics would agree with tighter packing of
the transmembrane rods after covalent capture.
Fig. 9 Mole fraction behavior of rod 1 (expected, dotted; found, solid) for
Increasing selectivity (4 > 2 > 1) roughly coincided with
decreasing apparent activity (1 > 2 > 4). One possible explanation
of this inverse correlation was that increasingly tight packing
of transmembrane O-NDI bundles increases the selectivity of
anion binding but hinders the hopping of anions from NDI to
NDI. Compared to a transmembrane bundle of neutral rods 2,
the addition of positive charges to the external termini would be
expected to loosen the packing of self-assembled bundles of O-
NDIs 1, whereas covalent capture in the neutral hairpin 4 would
tighten the packing of O-NDIs bundles.
−
−
−
55
−
−
(
A) Cl /I mixtures with intravesicular Cl , (B) Cl /ClO
4
mixtures with
−
−
−
−
intravesicular Cl , (C) Br /I mixtures with intravesicular Br and (D)
−
−
−
Cl /I mixtures with intravesicular Br . Fractional activity Y as a function
of the respective mole fractions x was determined in EYPC-LUVs⊃HPTS
as specified in Fig. 3.
−
−
always report on Cl /OH antiport and thus be independent of
−
ClO4 concentration, which is found to be the case. (Preliminary
results with similar O-NDI systems provided additional evidence
2
−
−
for unfavorable but, different to SO , still possible ClO4
4
70
transport. )
−
−
The AMFE with rod 1 was also detectable for Br /I mixtures
in vesicles with internal Br (Fig. 9C) despite the weaker discrim-
Mole fraction behavior
−
−
−
−
−
Over the years, the HPTS assay has been adapted to the charac-
ination between the two ions (Br /I = 3.0, Cl /I = 5.0, Fig. 6,
− −
4
6,48
46,48
terization of anion/cation selectivity,
selectivity sequences,
ꢁ). Finally, AMFE of rod 1 for Cl /I mixtures was not only
23,46
−
voltage-gated ion channels and pores,
age (Woodhull analysis), synthetic catalytic pores
voltage-sensitive block-
detectable in vesicles with internal Cl , but also, although less
73
74,75
−
and
pronounced, in vesicles with internal Br (Fig. 9D).
76–78,65
endovesiculation.
Otherwise explored exclusively in planar
1–10
bilayer conductance experiments, there was no reason why the
HPTS assay should not provide rapid and reliable access to mole
fraction behavior of ion channels and pores in vesicles.
Modeling
The cation–p stabilizing interaction between a cation and the
negative electrostatic potential on the face of an aromatic ring
is a widespread phenomenon for recognition in biological and
For this purpose, the activity of rod 1 was determined in EYPC-
LUVs⊃HPTS as described above in the presence of extravesicular
binary mixtures of fast and slow anions. The activity of mixtures of
7
9,80
chemical systems.
The nature of cation–p interactions has been
−
−
49,79–81
82,83
the fast Cl and the slow I was not as high as expected from simple
studied by Dougherty et al.
and others.
They came to the
additivity (Fig. 9A, solid vs. dotted lines). This substantial AMFE
conclusion that it can be described via the electrostatic terms.
−
−
found for Cl /I mixtures demonstrated that the transport of the
More recently it has been recognized from experimental studies
−
84
85
two ions is competitive, and that binding of more than one Cl is
in the solid state that s-triazines and s-tetrazine can interact
favorably with anions. Other experimental studies confirmed the
required to observe fast transport. Moreover, close proximity of
anion binding sites is required for operational charge repulsion in
multiion hopping. This proximity requirement excluded decisive
contributions from the otherwise quite likely additional anion
5
2,53
involvement of anion–p binding in aromatic receptors.
been also shown by theoretical studies
It has
52–54,86
that the interaction
between electron deficient aromatic p-systems and anions is
energetically favorable.
−
−
binding to the Boc-Gly termini. In other words, AMFE for Cl /I
mixtures provided experimental evidence for the occurrence of
multiion hopping from anion–p binding site to anion–p binding
site along rigid O-NDI rods, that is, the existence of anion–p
The anion–p interaction is in general dominated by electrostatic
54,86
and ion-induced polarization forces.
The electrostatic compo-
nent of the interaction is related to the magnitude of the permanent
and positive quadrupole moment (Qzz) of the aromatic system,
and the anion induced polarization correlates with the molecular
polarizability of the aromatic compounds.
55
slides.
AMFE was not observed in all cases. Transport of Cl /ClO
mixtures by rod 1, for instance, occurred with perfect additivity
−
−
4
(
Fig. 9B). This finding suggested that rod 1 almost does not
For molecular modeling studies of rigid O-NDI rods 1–5, the
NDIs 14–18 were selected as computationally tractable model
−
−
−
transport ClO
4
. The activities of Cl /ClO
4
mixtures should
This journal is © The Royal Society of Chemistry 2007
Org. Biomol. Chem., 2007, 5, 3000–3012 | 3007

View Article Online
systems (Fig. 10). The parent NDI 14 does not contain any
substituents. NDIs 15 and 16 are singly N-substituted with phenyl
and tetramethylphenyl spacers, respectively. NDIs 17 and 18
are doubly N-substituted with the same aryls. NDI 16 formally
corresponds to the termini of O-NDIs 1–5, and NDI 18 reproduces
their central part.
for 18 is expected to reflect the situation in the central part of rigid
O-NDI rods 1–5.
Electrostatic potential surfaces provide useful indications to
89
evaluate anion–p interactions. These surfaces revealed in the
present case how the phenyl and tetramethylphenyl spacers modify
the electrostatic environment for anion–p interactions. Due to
the perpendicular orientation of NDI and phenyl planes, the
positive surface of the NDI planes of 15 and 17 extended
in a continuous way to the equatorial region of the phenyl
spacer. Replacement of the phenyl spacer in 15 and 17 with
the tetramethylphenyl in 16 and 18 produced a more complex
surface. The electrostatic potential of the methyls ranges from
weakly (light blue) to distinctly positive as indicated by deep
blue caps located on the hydrogens. This unique combination of
electrostatic properties and conformational arrangement of NDI
and tetramethylphenyl spacer produced a continuous positive
surface with variable intensity of electrostatic potential. This
continuous positive surface is in excellent agreement with anion
hopping along rigid O-NDI rods 1–5.
Anion binding on the NDI-surface occurred in the region of
highest p-acidity near the imides (Fig. 11). NDI 16 was particularly
well suited to study anion–p interactions with O-NDI rigid-rods 1–
5
because the two imide nitrogens differ significantly. Whereas the
access to the substituent-free N1 is unrestricted, the space near
N2 is confined by the tetramethylphenyl spacer. The impact of
this tetramethylphenyl spacer on anion binding was investigated
by moving the anion above the p-system while restricting it to
remain in the C
geometry parameters of anion–p complexes 19–22 were optimized
with the PBE1PBE/6-311++G** method within the C symmetry
s
plane, which bisects the NDI moiety. The
s
−
−
restriction to finally obtain two minima for anions Cl and Br
alike.
Fig. 10 NDI model systems studied by the PBE1PBE/6-311++G**
method, where appropriate (C
computed electrostatic potential surfaces (right) are colored according
to an energy scale of ± 30.0 kcal mol , where blue is positive and
red is negative. Global quadrupole moments QZZ in Buckinghams were
computed with the MP2/6-311G** method.
s
and D2h) symmetry was imposed. DFT
In the chloride complex 19 with the chloride in the free space
near N1, the anion was found 2.95 A above the NDI plane,
˚
−
1
the horizontal anion displacement from N1 was d = 1.58 A˚
(
Fig. 11, Table 1). In the chloride complex 20 with the chloride
in the confined space near N2, the horizontal anion displacement
˚
increased to d = 2.16 A with respect to N2. The difference
The global quadrupole moments Qzz of NDIs 14–18 were
computed using the MP2/6-311G**//PBE1PBE/6-311++G**
method. For NDI 14, Qzz = +14.7 B was found. Clearly beyond
in horizontal anion displacement in N1 complex 19 and N2
complexes 20 indicated that steric interference from the proximal
˚
methyl in N2 complex 20 moved the chloride by 0.58 A from the
8
7
the hexafluorobenzene standard (Qzz = +9.5 B), the magnitude
of this quadrupole moment suggested anion binding by NDI 14
would be very strong. The introduction of one or two phenyls
as N-substituents further increased the Qzz of NDI 15 and 17,
respectively (Fig. 10). The negative phenyl quadrupole moment
produces the known p-basic surfaces that attract cations rather
location where the optimal anion–p interaction would occur. The
resulting weakening of anion–p interactions was further apparent
in the longer distance between chloride anion and NDI plane
˚
˚
˚
(3.04 A vs. 2.95 A). The close chloride–methyl contact of 2.50 A
a
Table 1 Anion position and interaction energies with NDI 16
49,79,80
than anions.
However, the perpendicular local quadrupole
b
˚
c
˚
d
/A
˚
e
−1f
moment oriented along the phenyl plane produces a p-acidic
Complex
Anion d/A
Dd/A
R
e
Eint/kcal mol
periphery. In DFT studies of the rotational energy profile around
−
1
2
2
9
0
1
Cl
Cl
1.58
2.16
1.53
2.32
—
0.58
—
2.95
3.04
3.00
3.28
−16.0
−18.2
−14.0
−15.7
88
the NDI-phenyl single bond, a global minimum was confirmed
for the phenyl plane oriented perpendicular to the NDI plane.
In this preferred conformation, the phenyl spacer reinforced the
global Qzz of NDI 15, where the z-axis is perpendicular to the
NDI plane. Addition of the second phenyl substituent to the
other imide slightly reduced the Qzz of the resulting NDI 17
−
−
Br
Br
−
22
0.78
a
c
b
From PBE1PBE/6-311++G** optimized geometries. See Fig. 11.
Horizontal anion displacement d between anion projection on the NDI
plane and the nitrogen N1 (19, 21) or nitrogen N2 (20, 22). Difference in
horizontal anion displacement Dd for the same anion near N2 compared
to N1. Equilibrium distance between anion and NDI plane. Interaction
energy with BSSE correction.
d
(
Fig. 10). A clear increase of the positive Qzz was achieved by the
e
f
tetramethyl substitution of the phenyl ring. Almost twice that of
hexafluorobenzene, the quadrupole moment Qzz = +16.8 B found
3
008 | Org. Biomol. Chem., 2007, 5, 3000–3012
This journal is © The Royal Society of Chemistry 2007

View Article Online
Fig. 11 PBE1PBE/6-311++G** optimized structures (C
s
symmetry) of complexes 19–22 formed between NDI 16 and chloride and bromide shown in
top (left) and side views (center, left) of ball and stick (atom color coding, anions green, center, left) and space filling representations (right).
−
−
supported the possibility of C–H · · · Cl stabilizing interactions in
Cl . Bromide binding in complex 22 was clearly affected by the
N2 complex 20. Similar anion–H interactions have been described,
confined space near N2. Namely, the methyl group from the spacer
pushed the bromide towards the center of the NDI. The difference
−
for example, in the case of linear C–H · · · X hydrogen bonds and
−
90,91
˚
trifurcated CH
3
· · · X improper hydrogen bonded complexes,
in horizontal anion displacement Dd = 0.78 A with bromide
9
2–94
˚
complexes was 0.20 A beyond that found in chloride complexes
as well as observed for anion receptors.
The comparison of BSSE corrected binding energies for the two
complexes 19 and 20 is of interest since they provide insight on
the strength of the overall binding between the anion and model
system 16. For complex 20, suffering from the direct interference
between chloride near N2 and tetramethylphenyl spacer, the
binding is stronger by 2.2 kcal mol than for interference-free
association in N1 complex 19 (Table 1). This suggested that
steric interference from the tetramethylphenyl spacer is largely
compensated by factors such as the anion–H interaction (Fig. 11).
Comparison of chloride complexes 19 and 20 with bromide
complexes 21 and 22 was of interest with regard to anion
selectivity. The horizontal anion displacement for unhindered
bromide binding near N1 in complex 21 was comparable to
(Table 1). Moreover, the distance R
e
= 3.28 of the bromide from
˚
the NDI plane in the N2 complex 22 was 0.24 A longer than
that of chloride in the N2 complex 20. Both effects suggested that
the spacial confinement imposed by tetramethylphenyl spacers
may provide access to anion–p slides with selectivity for small
anions. Similar to the N2 chloride complex 20, the N2 bromide
complex 22 revealed a close contact of 2.72 A between Br and
−
1
−
˚
CH . The energy found for bromide binding in the confined space
3
−
1
near N2 in complex 22 was 1.7 kcal mol higher than that for
unhindered bromide binding near N1 in complex 21. This was
−
1
only 0.5 kcal mol less than the energy gains for chloride binding
in the confined N2 region in complex 20.
Insights from molecular modeling thus not only confirmed
the possibility of anion binding on the p-acidic NDI surfaces
but also revealed preferential anion interaction near the imide
chloride binding in N1 complex 19. The equilibrium distance
−
˚
from the NDI plane was only 0.05 A longer with Br than with
This journal is © The Royal Society of Chemistry 2007
Org. Biomol. Chem., 2007, 5, 3000–3012 | 3009

View Article Online
nitrogen. Advanced binding sites in confined spaces produced
by the methyl substituents on the phenyl spacer were found to
the visualization of the rich complexity encountered in active O-
NDI suprastructures and must be considered with appropriate
caution.
−
1
contribute more to anion binding (1.7–2.2 kcal mol ) than to
−
−
−1
Cl > Br selectivity (0.5 kcal mol ). In general, these findings
are in excellent qualitative agreement with experimental results.
Possibly more important effects with larger anions as well as
Summary and conclusion
49
correlations with desolvation penalties remain to be explored.
Moreover, spacer engineering can be formulated as an attractive
rational approach to the design of advanced anion–p slides.
With the only intention to generate a feeling for possible active
suprastructures, various parallel tetrameric bundles of asymmetric
O-NDI rod 2 were loaded with three chloride anions and subjected
to short molecular dynamics simulations in vacuo (Fig. 12). O-
NDI bundles 23 display longitudinal displacement of anion–
p slides to produce a string of aligned aromatic binding sites.
Each site is composed of two face-to-face oriented NDIs and
two tetramethylphenyls. Adjacent aromatic rings are naturally
A small collection of rigid oligo-p-phenylene-N,N-naphth-
alenediimide (O-NDI) rods was synthesized to elaborate on
the formation and nature of transmembrane anion–p slides for
multiion hopping across lipid bilayer membranes. According to
the HPTS assay in EYPC vesicles, rods with one charged terminus
have the highest activity, whereas the activity of rods with charges
at both termini was weaker than that of uncharged rods. These
trends suggested that transmembrane rod orientation is needed for
activity. Hill plots were in support of either exergonic self-assembly
into active suprastructures or monomeric active structures.
The most active cationic O-NDI rods (2) exhibited a rather
◦
rotated by 90 . The terminal binding sites in bundles 23, composed
−
−
of two NDI and two glycine termini, each revealed possible
contributions from the mildly acidic amide, carbamate, amine and
ammonium hydrogens in the latter. Preliminary computational
studies suggested that these terminal functional groups may be
able to act as isolate anion receptors on their own and thus provide
weak (Cl /I = 1.9) selectivity similar to a halide IV topology
(Cl > Br > I > F ). This selectivity increased to a more
−
−
−
−
−
−
pronounced (Cl /I = 7.6) and quite unusual halide VI topology
−
−
−
>
with the less active neutral O-NDI rods (1, Cl > F > Br
−
I ). Covalent capture as rigid-rod O-NDI hairpins gave neutral
23,24
an attractive additional end-group engineering approach
to
anion–p slides (4) with poor activity but excellent chloride
−
−
−
−
refine the anion selectivity of O-NDI rods. However, significant
contributions from end-group engineering to multiion hopping
and anion selectivity were not supported by AMFE results, pre-
liminary results indicated preserved anion selectivity of rigid-rod
p-slides with tetraanionic end groups. More generally, we caution
that these preliminary models of possible active suprastructures
were obtained under highly simplified conditions far from reality
selectivity (Cl /I ∼ Cl /Br > 7). These trends suggested that
proximity of transmembrane rods is required for both activity and
selectivity, the former decreasing and the latter increasing with
increasing lateral rod proximity. Selective transport of anions with
high hydration energy suggested strong anion binding along the
−
−
anion–p slides. Similar selectivity for F and OAc supported
dominant energetics and excluded significant contributions from
size exclusion. The pronounced chloride selectivity of O-NDI
hairpin 4 implied slightly decreasing importance of energetics and
(
excluding, e.g., surrounding membrane and explicit water). They
were prepared with the only intention to more qualitatively assist
−
−
increasing of contributions from size exclusion (F > OAc ).
−
−
−
−
Lower than expected activity of mixtures of Cl /I and Br /I
−
−
but not Cl /ClO
4
demonstrated that more than one binding site
in close proximity is required for fast transport. These significant
AMFEs thus confirmed the existence of anion–p slides, i.e.,
multiion hopping along the rigid O-NDI rods across the bilayer
membrane.
With a significant inverted quadrupole moment of at least +14.7
B, almost twice that of hexafluorobenzene (+8.9 B) and beyond
that of significant negative counterparts such as pyrene (−13.8 B),
ab initio and DFT calculations confirmed the potential of NDIs
for anion–p interactions. Molecular models further highlighted the
role of the four methyl groups in the phenyl spacers to reduce the
available size of the p-acidic centers of the NDI surface for perfect
accommodation of small anions such as chloride or bromide, and
to magnify the NDI quadrupole moment.
The remarkable consistence of the found characteristics with
operational anion–p slides should not distract from the fact that
many aspects of introduced concept remain highly speculative.
Several interpretations and conclusions made may thus be best
considered as tentative, made with the only intention to stimulate
progress in the field. Current efforts to synthesize advanced
anion–p slides focus on hydrophilic anchoring to increase overall
Fig. 12 Optimized geometry of a notional active suprastructure after a
molecular dynamics simulation of 1500 ps in vacuo, that is, tetramer 23,
formed by four parallel monomers 2 (amine terminus up, Boc terminus
down) and loaded with three chlorides (green). The indicated NDI–NDI
distances (arrows) are 9.55 A˚ (left) and 8.35 A˚ (right).
70
activity, on varied substitution patterns on the phenyl spacers and
62
the NDI core, and on varied positioning of the NDI acceptors
along the rigid-rod scaffold.
3
010 | Org. Biomol. Chem., 2007, 5, 3000–3012
This journal is © The Royal Society of Chemistry 2007

View Article Online
Acknowledgements
36 M. F. Ilker, K. N u¨ sslein, G. N. Tew and E. B. Coughlin, J. Am. Chem.
Soc., 2004, 126, 15870–15875.
3
7 J. Rennie, L. Arnt, H. Tang, K. N u¨ sslein and G. N. Tew, J. Ind.
Microbiol. Biotechnol., 2005, 32, 296–300.
8 J. L. Sessler, L. R. Eller, W. S. Cho, S. Nicolaou, A. Aguilar, J. T.
Lee, V. M. Lynch and D. J. Magda, Angew. Chem., Int. Ed., 2005, 44,
5989–5992.
We thank D. Jeannerat, A. Pinto and S. Grass for NMR
measurements, P. Perrottet and the group of F. G u¨ la c¸ ar for ESI
MS measurements, N. Oudry and G. Hopfgartner for MALDI MS
measurements, CSCS in Manno for CPU time on their CRAY-
XT3 and IBM Power5–575 computers, P. Tecilla (Trieste) and
both referees for very helpful suggestions, and the Swiss NSF for
financial support.
3
3
4
9 P. A. Gale, Acc. Chem. Res., 2006, 39, 465–475.
0 V. Sidorov, F. W. Kotch, G. Abdrakhmanova, R. Mizani, J. C. Fettinger
and J. T. Davis, J. Am. Chem. Soc., 2002, 124, 2267–2278.
41 J. L. Seganish, J. C. Fettinger and J. T. Davis, Supramol. Chem., 2006,
8, 257–264.
1
4
2 P. V. Santacroce, O. A. Okunola, P. Y. Zavalij and J. T. Davis, Chem.
Commun., 2006, 41, 3246–3248.
3 B. A. McNally, A. V. Koulov, B. D. Smith, J. B. Joos and A. P. Davis,
Chem. Commun., 2005, 41, 1087–1089.
References
4
1
2
3
4
5
B. Hille and W. Schwarz, J. Gen. Physiol., 1978, 72, 409–442.
C. Miller, J. Gen. Physiol., 1999, 113, 783–787.
44 J. P. Clare, A. J. Ayling, J. B. Joos, A. L. Sisson, G. Magro, M. N. P e´ rez-
Pay a´ n, T. N. Lambert, R. Shukla, B. D. Smith and A. P. Davis, J. Am.
Chem. Soc., 2005, 127, 10739–10746.
45 N. Madhavan, E. C. Robert and M. S. Gin, Angew. Chem., Int. Ed.,
2005, 44, 7584–7587.
46 N. Sakai, D. Houdebert and S. Matile, Chem.–Eur. J., 2003, 9, 223–232.
47 Y. Tanaka, Y. Kobuke and M. Sokabe, Angew. Chem., Int. Ed. Engl.,
1995, 34, 693–694.
48 M. M. Tedesco, B. Ghebremariam, N. Sakai and S. Matile, Angew.
Chem., Int. Ed., 1999, 38, 540–543.
G. Eisenman and R. Horn, J. Membr. Biol., 1983, 76, 197–225.
E. M. Wright and J. M. Diamond, Physiol. Rev., 1977, 57, 109–156.
J. A. Tabcharani, J. M. Rommens, Y.-X. Hou, X.-B. Chang, L.-C. Tsui,
J. R. Riordan and J. W. Hanrahan, Nature, 1993, 366, 79–82.
P. Lindsell, J. Physiol., 2001, 531, 51–66.
6
7
8
P. Lindsell, Exp. Physiol., 2006, 91, 123–129.
M. Pusch, U. Ludewig, A. Rehfeldt and T. J. Jentsch, Nature, 1995,
3
73, 527–531.
9
Z. Qu and H. C. Hartzell, J. Gen. Physiol., 2000, 116, 825–884.
1
1
1
1
0 M. Duszyk, D. Liu, A. S. French and S. F. Man, Eur. Biophys. J., 1993,
2, 5–11.
1 U. Zachariae, V. Helms and H. Engelhardt, Biophys. J., 2003, 85, 954–
62.
2 R. Dutzler, E. B. Campbell and R. MacKinnon, Science, 2003, 300,
08–112.
49 R. A. Kumpf and D. A. Dougherty, Science, 1993, 261, 1708–1710.
50 R. M. Fairchild and K. T. Holmann, J. Am. Chem. Soc., 2005, 127,
16364–16365.
2
9
51 Y. S. Rosokha, S. V. Lindeman, S. V. Rosokha and J. K. Kochi, Angew.
Chem., Int. Ed., 2004, 43, 4650–4652.
1
52 M. Mascal, A. Armstrong and M. D. Bartberger, J. Am. Chem. Soc.,
2002, 124, 6274–6276.
3 D. A. Doyle, J. M. Cabral, R. A. Pfuetzner, A. Kuo, J. M. Gulbis, S. L.
Cohen, B. T. Chait and R. MacKinnon, Science, 1998, 280, 69–77.
4 R. Dutzler, FEBS Lett., 2004, 564, 229–233.
53 I. Alkorta, I. Rozas and J. Elguero, J. Am. Chem. Soc., 2002, 124,
1
1
8593–8598.
5 A. P. Davis, D. N. Sheppard and B. D. Smith, Chem. Soc. Rev., 2007,
54 C. Garau, D. Qui n˜ onero, A. Frontera, P. Ballester, A. Costa and P. M.
Dey a` , J. Phys. Chem. A, 2005, 109, 9341–9345.
55 V. Gorteau, G. Bollot, J. Mareda, A. Perez-Velasco and S. Matile, J. Am.
Chem. Soc., 2006, 128, 14788–14789.
56 L. L. Miller, B. Zinger and J. S. Schlechte, Chem. Mater., 1999, 11,
2313–2315.
3
6, 348–357.
1
1
6 P. Scrimin and P. Tecilla, Curr. Opin. Chem. Biol., 1999, 3, 730–735.
7 G. W. Gokel and A. Mukhopadhyay, Chem. Soc. Rev., 2001, 30, 274–
2
86.
1
1
2
8 T. M. Fyles, Chem. Soc. Rev., 2007, 36, 335–347.
9 S. Matile, A. Som and N. Sord e´ , Tetrahedron, 2004, 60, 6405–6435.
0 A. L. Sisson, M. R. Shah, S. Bhosale and S. Matile, Chem. Soc. Rev.,
57 D.-S. Choi, Y. S. Chong, D. Whitehead and K. D. Shimizu, Org. Lett.,
2001, 3, 2757–2760.
2
006, 35, 1269–1286.
58 J. Gawronski and K. Kacprzak, Chirality, 2002, 14, 689–702.
59 J. Lee, V. Guelev, S. Sorey, D. W. Hoffman and B. L. Iverson, J. Am.
Chem. Soc., 2004, 126, 14036–14042.
2
2
1 M. Yoshii, M. Yamamura, A. Satake and Y. Kobuke, Org. Biomol.
Chem., 2004, 2, 2619–2623.
2 Y. J. Jeon, H. Kim, S. Jon, N. Selvapalam, D. Y. Oh, I. Seo, C.-S. Park,
S. R. Jung, D.-S. Koh and K. Kim, J. Am. Chem. Soc., 2004, 126,
60 S. A. Vignon, T. Jarrosson, T. Iijima, H. R. Tseng, J. K. M. Sanders
and J. F. Stoddart, J. Am. Chem. Soc., 2004, 126, 9884–9885.
61 P. Mukhopadhyay, Y. Iwashita, M. Shirakawa, S. Kawano, N. Fujita
and S. Shinkai, Angew. Chem., Int. Ed., 2006, 45, 1592–2595.
62 S. Bhosale, A. L. Sisson, P. Talukdar, A. F u¨ rstenberg, N. Banerji, E.
Vauthey, G. Bollot, J. Mareda, C. R o¨ ger, F. W u¨ rthner, N. Sakai and S.
Matile, Science, 2006, 313, 84–86.
1
5944–15945.
3 N. Sakai, D. Gerard and S. Matile, J. Am. Chem. Soc., 2001, 123,
517–2524.
2
2
2
2
4 P. L. Boudreault and N. Voyer, Org. Biomol. Chem., 2007, 5, 1459–1465.
5 J. R. Pfeifer, P. Reiss and U. Koert, Angew. Chem., Int. Ed., 2006, 45,
5
01–504.
63 B. Ghebremariam, V. Sidorov and S. Matile, Tetrahedron Lett., 1999,
40, 1445–1448.
2
2
2
2
3
3
3
6 N. Sakai, N. Sord e´ , G. Das, P. Perrottet, D. Gerard and S. Matile, Org.
Biomol. Chem., 2003, 1, 1226–1231.
64 C. Ni and S. Matile, Chem. Commun., 1998, 33, 755–756.
65 S. Matile and N. Sakai, The Characterization of Synthetic Ion Channels
and Pores, in Analytical Methods in Supramolecular Chemistry, ed.
C. A. Schalley, Wiley, Weinheim, 2007, pp. 391–418.
66 S. Litvinchuk, G. Bollot, J. Mareda, A. Som, D. Ronan, M. R. Shah,
P. Perrottet, N. Sakai and S. Matile, J. Am. Chem. Soc., 2004, 126,
10067–10075.
7 M. Suzuki, T. Morita and T. Iwamoto, Cell. Mol. Life Sci., 2006, 63,
1
2–24.
8 J. R. Broughman, L. P. Shank, W. Takeguchi, B. D. Schultz, T. Iwamoto,
K. E. Mitchell and J. M. Tomich, Biochemistry, 2002, 41, 7350–7358.
9 G. Deng, T. Dewa and S. L. Regen, J. Am. Chem. Soc., 1996, 118,
8
975–8976.
0 C. W. Jiang, E. R. Lee, M. B. Lane, Y. F. Xiao, D. J. Harris and S. H.
Cheng, Am. J. Physiol.: Lung Cell. Mol. Physiol., 2001, 281, L1164.
1 R. Pajewski, R. Ferdani, J. Pajewska, N. Djedovic, P. H. Schlesinger
and G. W. Gokel, Org. Biomol. Chem., 2005, 3, 619–625.
67 A. V. Hill, Biochem. J., 1913, 7, 471–480.
68 S. Bhosale and S. Matile, Chirality, 2006, 18, 849–856.
69 N. Sakai and S. Matile, J. Phys. Org. Chem., 2006, 19, 452–460.
70 V. Gorteau, A. Perez-Velasco and S. Matile, in preparation.
71 N. Sakai, Y. Kamikawa, M. Nishii, T. Matsuoka, T. Kato and S. Matile,
J. Am. Chem. Soc., 2006, 128, 2218–2219.
2 B. Ding, N. Yin, Y. Liu, J. Cardenas-Garcia, R. Evanson, T. Orsak,
M. Fan, G. Turin and P. B. Savage, J. Am. Chem. Soc., 2004, 126,
1
3642–13648.
72 D. Ronan, N. Sord e´ and S. Matile, J. Phys. Org. Chem., 2004, 17, 978–
982.
3
3
3
3 B. Ding, U. Taotofa, T. Orsak, M. Chadwell and P. B. Savage, Org.
Lett., 2004, 6, 3433–3436.
73 Y. Baudry, D. Pasini, M. Nishihara, N. Sakai and S. Matile, Chem.
4 R. F. Epand, T. L. Raguse, S. H. Gellman and R. M. Epand,
Biochemistry, 2004, 43, 9527–9535.
Commun., 2005, 40, 4798–4800.
74 B. Baumeister, N. Sakai and S. Matile, Org. Lett., 2001, 3, 4229–4232.
75 N. Sakai, N. Sord e´ and S. Matile, J. Am. Chem. Soc., 2003, 125, 7776–
7777.
5 J. A. Patch and A. E. Barron, J. Am. Chem. Soc., 2003, 125, 12092–
1
2093.
This journal is © The Royal Society of Chemistry 2007
Org. Biomol. Chem., 2007, 5, 3000–3012 | 3011

View Article Online
7
7
6 M. Tedesco and S. Matile, Bioorg. Med. Chem., 1999, 7, 1373–
379.
7 H. Matsuo, J. Chevallier, F. Vilbois, R. Sadoul, J. Faur e´ , S. Matile,
85 B. L. Schottel, H. T. Chifotides, M. Shatruk, A. Chouai, L. M. P e´ rez,
J. Bacsa and K. R. Dunbar, J. Am. Chem. Soc., 2006, 128, 5895–5912.
86 C. Garau, A. Frontera, D. Qui C¸ onero, P. Ballester, A. Costa and P. M.
Dey a` , ChemPhysChem, 2003, 4, 1344–1348.
1
N. Sartori Blanc, J. Dubochet and J. Gruenberg, Science, 2004, 303,
5
31–534.
87 J. Hernandez-Trujillo and A. Vela, J. Phys. Chem., 1996, 100, 6524–
7
8 C. A. Winschel, A. Kalidindi, I. Zgani, J. L. Magruder and V. Sidorov,
J. Am. Chem. Soc., 2005, 127, 14704–14713.
9 D. A. Dougherty, Science, 1996, 271, 163–168.
6530.
88 G. Bollot, V. Gorteau, S. Matile and J. Mareda, in preparation.
89 D. L. Beene, G. S. Brandt, W. Zhong, N. M. Zacharias, H. A. Lester
and D. A. Dougherty, Biochemistry, 2002, 41, 10262–10269.
90 P. Hobza and Z. Havlas, Chem. Rev., 2000, 100, 4253–4264.
91 E. S. Kryachko and T. Zeegers-Huyskens, J. Phys. Chem. A, 2002, 106,
6832–6838.
92 K. I. Nattinen and K. Rissanen, CrystEngComm, 2003, 5, 326–330.
93 S. J. Coles, J. G. Frey, P. A. Gale, M. B. Hursthouse, M. E. Light, K.
Navakhun and G. L. Thomas, Chem. Commun., 2003, 568–569.
94 E. Light, P. A. Gale and K. Navakhun, Acta Crystallogr., Sect. E:
Struct. Rep. Online, 2006, 62, 1097–1098.
7
8
8
0 J. C. Ma and D. A. Dougherty, Chem. Rev., 1997, 97, 1303–1324.
1 J. P. Gallivan and D. A. Dougherty, J. Am. Chem. Soc., 2000, 122,
8
70–874.
2 E. Cubero, F. J. Luque and M. Orozco, Proc. Natl. Acad. Sci. U. S. A.,
998, 95, 5976–5980.
8
8
8
1
3 S. Tsuzuki, M. Yoshida, T. Uchimaru, M. Mikami and K. Tanabe,
J. Phys. Chem. A, 2001, 105, 769–773.
4 S. Demeshko, S. Dechert and F. Meyer, J. Am. Chem. Soc., 2004, 126,
4
508–4509.
3
012 | Org. Biomol. Chem., 2007, 5, 3000–3012
This journal is © The Royal Society of Chemistry 2007