Electrostatics of cell membrane recognition: Structure and activity of neutral and cationic rigid push-pull rods in isoelectric, anionic, and polarized lipid bilayer membranes

Article abstract of DOI:10.1021/ja003141+

Design, synthesis, and structural and functional studies of rigid-rod ionophores of different axial electrostatic asymmetry are reported. The employed design strategy emphasized presence of (a) a rigid scaffold to minimize the conformational complexity, (b) a unimolecular ion-conducting pathway to minimize the suprastructural complexity and monitor the function, (c) an extended fluorophore to monitor structure, (d) variable axial rod dipole, and (e) variable terminal charges to create axial asymmetry. Studies in isoelectric, anionic, and polarized bilayer membranes confirmed a general increase in activity of uncharged rigid push-pull rods in polarized bilayers. The similarly increased activity of cationic rigid push-pull rods with an electrostatic asymmetry comparable to that of α-helical bee toxin melittin (positive charge near negative axial dipole terminus) is shown by fluorescence-depth quenching experiments to originate from the stabilization of transmembrane rod orientation by the membrane potential. The reduced activity of rigid push-pull rods having an electrostatic asymmetry comparable to that in α-helical natural antibiotics (a positive charge near the positive axial dipole terminus) is shown by structural studies to originate from rod "ejection" by membrane potentials comparable to that found in mammalian plasma membranes. This structural evidence for cell membrane recognition by asymmetric rods is unprecedented and of possible practical importance with regard to antibiotic resistance.

Full text of DOI:10.1021/ja003141+

J. Am. Chem. Soc. 2001, 123, 2517-2524
2517
Electrostatics of Cell Membrane Recognition: Structure and Activity
of Neutral and Cationic Rigid Push-Pull Rods in Isoelectric, Anionic,
and Polarized Lipid Bilayer Membranes
Naomi Sakai,* David Gerard, and Stefan Matile*
Contribution from the Department of Organic Chemistry, UniVersity of GeneVa,
CH-1211 GeneVa, Switzerland
ReceiVed August 22, 2000
Abstract: Design, synthesis, and structural and functional studies of rigid-rod ionophores of different axial
electrostatic asymmetry are reported. The employed design strategy emphasized presence of (a) a rigid scaffold
to minimize the conformational complexity, (b) a unimolecular ion-conducting pathway to minimize the
suprastructural complexity and monitor the function, (c) an extended fluorophore to monitor structure, (d)
variable axial rod dipole, and (e) variable terminal charges to create axial asymmetry. Studies in isoelectric,
anionic, and polarized bilayer membranes confirmed a general increase in activity of uncharged rigid push-
pull rods in polarized bilayers. The similarly increased activity of cationic rigid push-pull rods with an
electrostatic asymmetry comparable to that of R-helical bee toxin melittin (positive charge near negative axial
dipole terminus) is shown by fluorescence-depth quenching experiments to originate from the stabilization of
transmembrane rod orientation by the membrane potential. The reduced activity of rigid push-pull rods having
an electrostatic asymmetry comparable to that in R-helical natural antibiotics (a positive charge near the positive
axial dipole terminus) is shown by structural studies to originate from rod “ejection” by membrane potentials
comparable to that found in mammalian plasma membranes. This structural evidence for cell membrane
recognition by asymmetric rods is unprecedented and of possible practical importance with regard to antibiotic
resistance.
Introduction
anism from pertinent discussions in recent reports on this topic
(Figure 2).1-3,6-8 The key characteristic of this mechanism is
that the formation of host-guest complex 1 with magainin-
like rods in bacteria-type membranes with a high inside negative
potential (ca. -200 mV)7a requires TM cation translocation for
favorable axial dipole alignment (Figure 2, arrow a). Reduced
Why is melittin a toxin and magainin 2 a natural antibiotic?1
Both are R-helical cationic polypeptides that act by mediating
ion transport across bilayer membranes (Figure 1). They share
central structural characteristics: A semirigid “rod” of a length
sufficient to span the hydrophobic core of bilayer membranes
and facial amphiphilicity for self-assembly into transmembrane
(TM) bundles with central hydrophilic channels. They differ,
however, in axial electrostatic asymmetry. The positively
charged amino acid residues are accumulated near the negative
end of the axial rod dipole in melittin and near the positive
dipole terminus in magainin. Following Merrifields original
suggestion,2 these two simplified situations of electrostatic rod
asymmetry are referred to as “melittin-like” and “magainin-
like”.3
(4) Recent references on peptide-based artificial ion channels and channel
models: (a) Baumeister, B.; Sakai, N.; Matile, S. Angew. Chem., Int. Ed.
2000, 39, 1955-1958. (b) Schrey, A.; Vescovi, A.; Knoll, A.; Rickert, C.;
Koert, U. Angew. Chem., Int. Ed. 2000, 39, 900-902. (c) Dieckmann, G.
R.; Lear, J. D.; Zhong, Q.; Klein, M. L.; DeGrado, W. F.; Sharp, K. A.
Biophys. J. 1999, 76, 618-630. (d) Clark, T. D.; Buehler, L. K.; Ghadiri,
M. R. J. Am. Chem. Soc. 1998, 120, 651-656. (e) Lear, J. D.; Schneider,
J. P.; Kienker, P. K.; DeGrado, W. F. J. Am. Chem. Soc. 1997, 119, 3212-
3217. (f) Voyer, N.; Potvin, L.; Rousseau, E. J. Chem. Soc., Perkin Trans.
2 1997, 1469-1472. (g) Meillon, J.-C.; Voyer, N. Angew. Chem., Int. Ed.
Engl. 1997, 36, 967-969.
(5) Selected reviews on nonpeptide ion channels and channel models:
(a) Kobuke, Y. In AdVances in Supramolecular Chemistry; Gokel, G. W.,
Ed.; JAI: Greenwich, London, 1997, Vol. 4, 163-210. (b) Koert, U.; Chem.
Unserer Z. 1997, 31, 20-26. (c) Gokel, G. W.; Murillo, O. Acc. Chem.
Res. 1996, 29, 425-432. Recent references: (d) Renkes, T.; Scha¨fer, H. J.
Siemens, P. M.; Neumann, E. Angew. Chem., Int. Ed. 2000, 39, 2512-
2516. (e) Forman, S. L.; Fettinger, J. C.; Pieraccini, S.; Gottarelli, G.; Davis,
J. T. J. Am. Chem. Soc. 2000, 122, 4060-4067. (f) Otto, S.; Janout, V.;
DiGiorgio, A. F.; Young, M. C.; Regen, S. L. J. Am. Chem. Soc. 2000,
122, 1200-1204. (g) Pe´rez, C.; Esp´ınola, C. G.; Foces-Foces, C.; Nu´n˜ez-
Coello, P.; Carrasco, H.; Mart´ın, J. D. Org. Lett. 2000, 2, 1185-1188. (h)
Kobuke, Y.; Nagatani, T. Chem. Lett. 2000, 298-299. (i) Das, S.; Kurcok,
P.; Jedlinski, Z.; Reusch, R. N. Macromolecules 1999, 32, 8781-8785. (j)
Abel, E.; Maguire, G. E. M.; Murillo, O.; Suzuki, I.; De Wall, S. L.; Gokel,
G. W. J. Am. Chem. Soc. 1999, 121, 9043-9052. (k) Qi, Z.; Sokabe, M.;
Donowaki, K.; Ishida, H. Biophys. J. 1999, 76, 631-641. (l) Hall, C. D.;
Kirkovits, G. J.; Hall, A. C. Chem. Commun. 1999, 1897-1898. (m) Fritz,
M. G.; Walde, P.; Seebach, D. Macromolecules 1999, 32, 574-580.
Compare also: (n) Fyles, T. M.; Look, D.; Zhou, X. J. Am. Chem. Soc.
1998, 120, 2997-3003. (o) Nolte, R. J. M. Chem. Soc. ReV. 1994, 11-19.
The importance of appropriate length and facial amphiphi-
licity for ion transport activity has been exemplified extensively
with many natural and artificial molecular rods.4,5 Elucidation
of the importance of electrostatic rod asymmetry for cell
membrane recognition proved more difficult. Nevertheless, it
is possible to extract a simplified, tentative recognition mech-
(1) Sakai, N.; Matile, S. Chem. Eur. J. 2000, 6, 1731-1737 and
references therein.
(2) Juvvadi, P.; Vunnam, S.; Merrifield, R. B. J. Am. Chem. Soc. 1996,
118, 8989-8997 and references therein.
(3) Note that the terms magainin-like and melittin-like axial rod
asymmetry are defined here as “excess positive charge near the positive
dipole terminus” and “excess positive charge near the negative dipole
terminus”, respectively. Particularly, the example of magainin 2 illustrates
clearly that, for example, (a) poor charge localization and (b) contributions
of negative charges near negative the axial dipole terminus may not be
negligible in future, refined definitions.2,8,21
10.1021/ja003141+ CCC: $20.00 © 2001 American Chemical Society
Published on Web 02/22/2001

2518 J. Am. Chem. Soc., Vol. 123, No. 11, 2001
Sakai et al.
Figure 2. Tentative cell membrane recognition mechanism of molec-
ular rods with axial electrostatic asymmetry.1-3,6-8 Emphasis is on the
electrostatic interactions between rods and membrane potential:6 rods
with magainin-8 (top) and melittin-like asymmetry (bottom) on one side
and unpolarized bilayer membranes common in model systems (left),
weakly polarized bilayer membranes similar to mammalian plasma
membranes (center), and highly polarized bilayer membranes similar
to bacterial plasma membranes (right) on the other. (a) TM rod
orientation, including TM cation translocation. (b) TM rod orientation,
excluding TM cation translocation. Schematic diagrams for electrostatic
rod asymmetry are as defined in Figure 1 and Table 1 (forward block
arrow, axial rod dipole;3 forward open arrow, direction of mammalian
membrane potential (up to -150 mV); forward open arrow, direction
of bacterial membrane potential (ca. -200 mV). Arrows, except a and
b, are pointing from plus to minus. Hydrophobic cores of lipid bilayers
are in gray.
Figure 1. Above: uniform alignment of the backbone carbonyl dipoles
accounts for the inherent, invariable axial dipole of R-helical membrane-
spanning peptides. Below: schematic diagrams for axial electrostatic
asymmetry, structure, and charge distribution for antibiotic magainin
2, toxic melittin, and synthetic retro melittin in R-helical conformation.
One-letter abbreviations are used for amino acids.
toxicity to mammals may, thus, originate from insufficient
polarization of mammalian plasma membranes to promote such
charge translocation. Because reversed TM rod orientation is
unlikely due to destructive dipole-potential interactions (Figure
2, arrow b), an inactive host-guest complex 2 may result for
magainin-like rods in mammal-type membranes. The formation
of active host-guest complexes 3 and 4 in the unpolarized
membranes that are frequently used in biophysical model
systems, however, is conceivable for rods with magainin- and
melittin-like axial asymmetry.8 Preferential TM orientation of
melittin-like rods in unpolarized (i.e., 4), mammal-type (i.e., 5)
and bacteria-type bilayer membranes (i.e., 6) is consistent with
poor cell membrane recognition (i.e., toxicity).1
by magainin-like electrostatic asymmetry (Figure 1), selective
formation of inactive complex 2 in weakly, and active complex
1 in highly, polarized membranes is conceivable to account for
cell membrane recognition by retro melittin (Figure 2). Recent
reports dealing with antibiotic activity of cationic helical
â-peptides9 underscored that a better understanding of cell mem-
brane recognition by electrostatic fine-tuning may, indeed,
provide rational design strategies for novel antibacterials having
promising properties concerning antibiotic resistance.1,2,7
Despite high scientific and commercial interest in the anti-
biotic activity of magainin-like rods, insights into the electro-
static principles underlying their recognition mechanism remain
elusive on the molecular structural level.1 In other words, it turns
out to be very difficult to actually “see” the membrane potential-
induced rod reorientations assumed in host-guest complexes
1-3 (Figure 2). We have suggested recently that fluorescent,
ionophoric rigid p-octiphenyl rods may be of use to elucidate
the subtle balance of constructive and destructive electrostatic
interactions underlying cell membrane recognition.1 This concept
emerged from a preliminary model study with rigid push-pull
rod 7 (Figure 3).10 p-Octiphenyl 7 comprises four lateral
azacrowns to mediate TM ion transport, a terminal π-donor and
a terminal π-acceptor, to provide a permanent axial rod dipole.
Asymmetric rod 7 was shown to depolarize polarized bilayer
membrane more efficiently than the symmetric rod 8, which
has a nearly identical global structure. On the basis of these
results, five design principles have been elaborated for the
construction of “universal” probes for comprehensive clarifica-
tion of the electrostatic principles underlying cell membrane
recognition:1 (1) a rigid scaffold to minimize the conformational
complexity, (2) a unimolecular TM ion-conducting pathway to
simplify the suprastructural complexity and to monitor functions,
(3) an extended fluorophore to monitor structures, (4) variable
Insightful studies from the Merrifield lab provided persuasive
support for the simplified recognition mechanism depicted in
Figure 2.2 Retro melittin, for example, was shown to exhibit
reduced hemolytic, but maintained antibiotic, activity as com-
pared to isomeric melittin. Because retro melittin is characterized
(6) With respect to results described in this work, the tentative recognition
mechanism in Figure 2 emphasizes stabilization of TM rods by favorable
alignment of the rod dipole to inside negative membrane potentials and
neglects contributions from multivalent (compare, e.g., (a) Mammen, M.;
Choi, S.-K.; Whitesides, G. M. Angew. Chem., Int. Ed. 1998, 37, 2754-
2794) interfacial ion pair formation between anionic lipid headgroups and
cationic rod charges. Further simplifications include the exclusion of tilted
rods, partial rod insertions, mixed populations, membrane composition and
phase behavior, and so on (compare, e.g., (b) Schanck, P. J.; Lins, L.;
Brasseur, R. J. Theor. Biol. 1999, 198, 173-181).
(7) (a) Gabay, J. E. Science 1994, 264, 373-374, other articles in this
special issue of Science, and references therein. Recent review: (b) Kourie,
J. I.; Shorthouse, A. A. Am. J. Physiol. Cell Physiol. 2000, 278, C1063-
C1087. For alternative approaches to cell membrane recognition focusing
on their molecular composition, phase behavior, thickness, and so on, see,
for example: (c) Breukink, E.; Wiedemann, I.; van Kraaij, C.; Kuipers, O.
P.; Stahl, H.-G.; de Kruijff, B. Science 1999, 286, 2361-2364. (d) Bolard,
J. Biochim. Biophys. Acta 1986, 864, 257-304, and references therein, in
refs 1 and 5f.
(8) Note that the orientation of magainin 2 itself in unpolarized
membranes is presumably not TM, as in 3, but is interfacial, as in 2
(compare, e.g., (a) Matsuzaki, K.; Murase, O.; Tokuda, H.; Funakoshi, S.;
Fujii, N.; Miyajima, K. Biochemistry 1994, 33, 3342-3349. (b) Matsuzaki,
K. Biochim. Biophys. Acta 1999, 1462, 1-10). This alternative structure
of 3 with interfacial magainin 2 is, however, in excellent agreement with
the mechanism in Figure 2, because magainin 2 contains two anions near
the negative dipole terminus in addition to multiple cations near the positive
dipole terminus. Disfavored TM transfer of these terminal anions may, thus,
account for interfacial magainin 2 with reduced activity in unpolarized
membranes. Studies with the corresponding push-pull rods with one anionic
and one cationic terminus are ongoing.
(9) (a) Porter, E. A.; Wang, X.; Lee, H. S.; Weisblum, B.; Gellman, S.
H.; Nature 2000, 404, 565-565. (b) Hamuro, Y.; Schneider, J. P.; DeGrado,
W. F. J. Am. Chem. Soc. 1999, 121, 12200-12201.
(10) (a) Winum, J.-Y.; Matile, S. J. Am. Chem. Soc. 1999, 121, 7961-
7962. (b) Supporting Information of 10a.

Electrostatics of Cell Membrane Recognition
J. Am. Chem. Soc., Vol. 123, No. 11, 2001 2519
Deiodinated p-septiphenyls were observed as minor side prod-
ucts (overall, 11%).
Among multiple methods for sulfide oxidation,12 the follow-
ing conditions were selected on the basis of qualitative model
studies with sulfide 24 (Scheme 2). Initial partial oxidation with
mCPBA yielded almost quantitative formation of sulfoxide 25
in addition to sulfide 24. Subsequent oxidation of this mixture
with KMnO412b selectively converted sulfoxide 25 into sulfone
26 with nearly negligible oxidation of sulfide 24. The remaining
sulfoxide 25 that was obtained from the incomplete oxidation
could be selectively reduced with TFAA/NaI12c to ultimately
give a 1:1 mixture of sulfide 24 and sulfone 26.
Figure 3. Structure of neutral rigid push-pull rod 7 and symmetric
pull-pull rod 8.
Application of the three-step protocol in Scheme 2 to the
symmetric disulfide 9 yielded a clean mixture of push-push rod
9, push-pull rod 10, and pull-pull rod 11 that was readily
separated by semipreparative RP-HPLC. The same procedure
was also compatible with the more complex situation of
asymmetric push-push rod 17. The four desired rods 17-20
could be separated and isolated in excellent yield by semi-
preparative RP-HPLC; deprotection with TFA gave the cationic
rigid-rod ionophores 12-15.
During the interpretation of the usual data collected to confirm
the correct structure of all new compounds (see Supporting
Information), we observed peculiar multiple satellites in the ESI-
MS (electrospray ionization mass spectrometry) of cationic rigid-
rod ionophores (Figure 5). As illustrated for melittin-like push-
pull rod 13, dominant molecular ion peaks with two or three
sodium cations were followed by several smaller signals with
constant additional m/z corresponding to the sodium salt of TFA.
This finding, thus, unexpectedly illustrated the capacity of rigid-
rod ionophores to bind multiple cations and, thus, (possibly)
mediate TM ion transport by the same cation “hopping” or
“wire” mechanism that emerges as crucial for functional ion-
conducting pathways in biological ion channels.13,14
The previously communicated synthesis of p-sexiphenyl 21
from biphenyl crowns 27 and 28 turned out to be less practical
than the route through “insoluble” p-sexiphenyl 29, prepared
directly from bianisoles 30 and 31, and solubilized14 tert-butyl
ester 32 (Scheme 3, see Supporting Information for discus-
sion).15 Subunit 22 leading to the cationic rod termini in 12-
15 was prepared from amine 33 (Scheme 4). BOC protection
(to give bromide 34) and condensation with arylthiol 35 gave
bromide 36 that was converted into boronate 22.
Figure 4. Design principles for rigid-rod ionophores with variable axial
electrostatic asymmetry.
axial rod dipole, and (5) variable terminal charges to precisely
tune axial electrostatic asymmetry (Figure 4).
The call for rigid-rod scaffolds, extended fluorophores and
variable axial rod dipoles was of particular interest, because
these structural characteristics are incompatible with R-helical
peptides and were apparently unique for rods 7 and 8 until now.
Here we report the synthesis and study of a neutral push-
pull rod (10), cationic push-pull rods with magainin- (14) and
melittin-like (13) electrostatic asymmetry, and the corresponding
neutral (9, 11) and cationic (12, 15) p-octiphenyl ionophores
without axial dipole (Scheme 1 and Table 1). Functional and
structural studies in isoelectric, anionic, and polarized bilayer
membranes under comparable conditions are described to afford
unprecedented insights into the molecular structural level of
membrane recognition by electrostatically asymmetric rods.
Results and Discussion
Design. As described for model rods 7 and 8 (Figure 3),10
the design of rigid-rod ionophores 9-20 focused on a rigid
scaffold, a unimolecular ion-conducting pathway, and an
extended fluorophore, as well as variable axial rod dipole and
terminal charges, to create axial asymmetry (vide supra, Figure
4).1 In contrast to 7 and 8, ionophores 9-20 contain six instead
of four lateral azacrowns. This change was made to improve
the unimolecular TM ion-conducting pathway (Scheme 1 and
Table 1).4f,g,5a-c,j,l,o Moreover, the terminal π-donors and π-ac-
ceptors in 7 and 8 were replaced by sulfides and sulfones. This
second change was necessary for the attachment of cationic side
chains to the terminal π-donors and acceptors. Facile synthetic
access to variable cationic termini was crucial for, for example,
functional and structural studies with push-pull rods with
melittin- (13) and magainin-like electrostatic asymmetry (14).
Ion Transport Activity of Neutral and Cationic Rods in
Unpolarized Bilayer Membranes. The capacity of rods 9-15
to mediate ion transport across unpolarized bilayer membranes
with isoelectric, zwitterionic surfaces was assessed using
uniformly sized EYPC (egg yolk phosphatidylcholine)-SUVs
(small unilamellar vesicles) with internal pH-sensitive fluores-
cent probe HPTS (8-hydroxypyrene-1,3,6-trisulfonic acid, tri-
sodium salt) (Figure 6).17 A TM pH gradient of ∆pH ≈ 1.0
(11) (a) Suzuki, A. J. Organomet. Chem. 1999, 576, 147-168. (b)
Murata, M.; Watanabe, S.; Masuda, Y. J. Org. Chem. 1997, 62, 6458-
6459. (c) Ishiyama, T.; Murata, M.; Miyama, N. J. Org. Chem. 1995, 60,
7508-7510.
(12) (a) Uemura, S. ComprehensiVe Organic Synthesis; Trost, B. M.,
Ed.; Pergamon Press: Oxford, 1991; Vol. 6, pp 757-787. (b) Henbest, H.
B.; Khan, S. A. J. Chem. Soc., Chem. Commun. 1968, 1036-1036. (c)
Drabowicz, J.; Oae, S. Synthesis 1977, 404-405.
(13) (a) Aqvist, J.; Luzhkov, V. Nature 2000, 404, 881-884. (b) Doyle,
D. A.; Morais Cabral, J.; Pfuetzner, R. A.; Kuo, A.; Gulbis, J. M.; Cohen,
S. L.; Chait, B. T.; MacKinnon, R. Science 1998, 280, 69-77. (c) Gulbis,
J. M.; Zhou, M.; Mann, S.; MacKinnon, R. Science 2000, 289, 123-127.
(14) Sakai, N.; Majumdar, N.; Matile, S. J. Am. Chem. Soc. 1999, 121,
4294-4295.
Synthesis. Rigid-rod molecules 9-20 were prepared from
p-sexiphenyl 21 (Scheme 1). Suzuki-coupling11 with equimolar
amounts of sulfides 22 and 23 and separation of the product
mixture by semipreparative RP-HPLC (reverse-phase high-
pressure liquid chromatography) afforded push-push rods 9, 17,
and 16 in excellent 15%, 31%, and 15% yield, respectively.
(15) Robert, F.; Winum, J.-Y.; Sakai, N.; Gerard, D.; Matile, S. Org.
Lett. 2000, 2, 37-39.

Scheme 1a
a (a) PdCl2(dppf), Na2CO3; isolated 15% (9), 31% (17), 15% (16). (b) 1, mCPBA; 2, KMnO4, MgSO4; 3, TFAA/NaI; isolated 26% (11), 40% (10), 31% (9). (c) 1, mCPBA; 2, KMnO4, MgSO4; 3,
TFAA/NaI; isolated 25% (20), 25% (19), 22% (18), 23% (17). (d) TFA, 90% (12), 71% (13), 89% (14), 91% (15). Lateral rod substitution as in Scheme 3.

Electrostatics of Cell Membrane Recognition
J. Am. Chem. Soc., Vol. 123, No. 11, 2001 2521
Scheme 3a
Figure 5. ESI-MS (MeOH) of cationic rigid push-pull rod 13.
a (a) 4,4,5,5-tetramethyl-[1,3,2]-dioxaborolane,11c PdCl2(dppf), 69%.
(b) Pd(PPh3)4, Na2CO3, 10%. 1, BBr3; 2, tert-butylbromoacetate,
Cs2CO3, 84%. (d) 1, TFA/CH2Cl2; 2, 18-azacrown-6, HBTU, Et3N,
69% (revised from ref 15).
Table 1. Schematic Diagrams for the Electrostatic Asymmetry of
Rigid-Rod Ionophores Described in This Study (Scheme 1)
Scheme 4a
a (a) BOC2O, 84%. (b) K2CO3, 87%. (c) bis(pinacolato)diboron,
PdCl2(dppf), 47% (conversion yield, 61%).11c
Scheme 2
Figure 6. Assay for ion transport activity of neutral and cationic rods
in unpolarized bilayer membranes:17,18 top, schematic representation
of the assay; bottom, changes in fluorescence intensity of intravesicular
HPTS after ionophore (10 µM) addition as a function of time (relative
was established by the increase of the external pH using base.
The ionophores were added at time 0 and the rate of pH gradient
collapse was monitored by following the excitation and emission
intensity of internal HPTS as a function of time (Figure 6).18
emission intensity ) (It - I0)/(I∞ - I0) × 100%; λem ) 510 nm, λex
)
460 nm). Complete lysis (I∞ ) was achieved by excess Triton X-100 or
melittin17 (0.25 mM EYPC-SUVs in 10 mM potassium phosphate
buffer, 100 mM KCl). Curves for rods given in parentheses were
identical to those of the parent compound and are not shown for clarity.
(16) Dourtoglou, V.; Gross, B. Synthesis, 1984, 572-574.
(17) Weiss, L. A.; Sakai, N.; Ghebremariam, B.; Ni, C.; Matile, S. J.
Am. Chem. Soc. 1997, 119, 12142-12149.
Compared to the previous model rods 7 and 8 (Figure 3),10
the activity of the corresponding neutral rods 9-11 (Figure 4)
was not substantially higher (Figure 6). This was surprising,
because the presence of six instead of four lateral azacrowns
was expected to result in increased activity. Our suspicion that
relatively poor water-solubility and incorporation into bilayer
membranes, rather than the number of lateral azacrowns, chiefly
govern the activity of neutral rods 7-11 under these conditions
was confirmed by the substantially increased activity of all
cationic rods (12-15, Figure 6).
(18) Note that this interpretation is only valid because HPTS efflux can
be excluded (compare section on lytic activity). When compared to other
pH-sensitive probes such as CF, the advantage of HPTS is compatibility
with double-channel fluorescence kinetics. In these experiments, fluores-
cence intensity change was simultaneously monitored at another wavelength
(λem ) 510 nm, λex ) 405 nm) to identify and eliminate possible changes
in emission that do not originate from changes in pH (compare: Haugland,
R. P. Handbook of Fluorescent Probes and Research Chemicals; 6th ed,
Molecular Probes: Eugene, OR, 1996). However, influence from the
fluorescence of the rods was much greater in this channel; thus, the
ratiometric normalization was not applied.

2522 J. Am. Chem. Soc., Vol. 123, No. 11, 2001
Sakai et al.
The highest activity was observed for magainin-like rod 14.
Structural studies suggested that this effect does not originate
from improved incorporation or TM orientation (compare to
section on structural studies). A likely explanation, thus,
considers favorable influence of maximal global electrostatic
asymmetry, as compared to all of the other rods (i.e., the sum
of axial dipole plus terminal charges in rod 14) on the properties
of the adjacent ion-conducting pathway.4e
Nearly identical activities were observed for ion transport
across unpolarized bilayer membranes having anionic surfaces
prepared by doping EYPC-SUVs with 50% EYPG (egg yolk
phosphatidylglycerol) (not shown). This suggested that, surpris-
ingly,1,2,7,8 constructive electrostatic interactions between the
membrane surfaces and the rod termini are irrelevant for ion
transport activity under these conditions. Multivalent interactions6a
are presumably required for recognition of negatively charged
membrane surfaces. Further studies with anionic membranes
were, thus, redundant.
The concentration dependence of all of the observed flux rates
in unpolarized SUVs was roughly linear (not shown). This trend
is in support of the expected presence of monomeric active
structures.5f,4a,10b No indications for rod self-assembly were seen
during all of the reported structural and functional studies
(compare, for example, ESI-MS in Figure 5) with one excep-
tion: all fluorescence kinetic measurement with pull-pull rods
11 and 15 were hampered by the occasional appearance of large
noise. These peculiar observations, made with disulfone rods
only, may stem from light scattering and, thus, originate from
self-assembly of disulfone rods in water. Because of minor
interest for studies of membrane recognition, this phenomenon
was not further investigated, and studies of the interaction of
disulfone rods with lipid bilayers were discontinued.
Ion Transport Activity of Neutral and Cationic Rods in
Polarized Bilayer Membranes. Doubly labeled, polarized
EYPC-SUVs with internal HPTS and external safranin O were
prepared as described previously (Figure 7).10 In brief, uniformly
sized EYPC-SUVs were prepared by dialytic detergent removal
using a buffer containing 100 mM K+. External K+ was
subsequently exchanged with isoosmolar Na+ by dialysis and
gel filtration.10b An inside negative membrane potential was
established by the addition of K+ carrier valinomycin at
concentrations sufficient for slow K+ transport without rapid
K+ gradient collapse (Figure 7, background curve).10b Standard
calibration of the emission intensity of the potential-sensitive
dye safranin O indicated that the maximal potassium diffusion
potentials obtained by this procedure were about -150 mV
(corresponding to a relative emission intensity of 0 in Figure
7).19b This value indicated the presence of ∼300 µM residual
external K+. A maximal Nernst potential around -150 mV is
comparable to highly polarized mammalian plasma membranes
but below that of Gram-positive and -negative bacteria.7a
Figure 7. Assay for ion transport activity of neutral and cationic rods
in polarized bilayer membranes:10,19,20 top: schematic representation of
the assay; bottom, changes in fluorescence intensity of extravesicular
safranin O (10 µM) during the addition of valinomycin (0.6 µM, for
membrane polarization), base (for ∆pH ≈ 1.0), rods (10 µM), and final
complete depolarization (excess melittin). Relative emission intensity
) (It - I0)/(I∞ - I0) × 100%; λem ) 581 nm, λex ) 522 nm. Relative
emission intensity at 0% relates to a potential of about -150 mV, a
relative emission intensity at 100% to 0 mV. (Conditions: 0.5 mM
EYPC-SUVs with 10 mM potassium phosphate 100 mM KCl inside,
in 10 mM sodium phosphate buffer, 100 mM NaCl.) Curves for rods
given in parentheses were identical to those of the parent compound
and are not shown for clarity.
potentials appears, thus, to be, indeed, a general mechanism
for cell membrane recognition, independent of the chemical
nature of the terminal π-donor and π-acceptors and tolerant to
modifications of the lateral ion-conducting pathway.
In unpolarized EYPC-SUVs, cationic rods were generally
more active than neutral rods (Figure 6). In polarized EYPC-
SUVs, this trend did not hold (Figure 7). Only the cationic push-
pull rod 13 with melittin-like electrostatic axial asymmetry
exhibited the activity in the range of the neutral push-pull rod
10. The most distinct decrease in activity was observed for
magainin-like push-pull rod 14, dropping from the highest
activity in unpolarized membranes to the second-lowest activity
in polarized membranes. Only the symmetric, neutral rod 9
remained less active than rod 14 with the highest global axial
asymmetry.20
Structural Studies of Cationic Rigid Push-Pull Rods in
Unpolarized and Polarized Bilayer Membranes. The activities
of melittin-like rod 13 and magainin-like rod 14 in polarized
and unpolarized membranes were consistent with the biological
activities of the corresponding R-helical rods.21 In contrast to
the situation with their biological counterparts, elucidation of
The capacity of ionophoric rods to depolarize these mammal-
type bilayer membranes was reported by changes of safranin O
emission as a function of time after addition (Figure 7).10,19 In
contrast to their roughly identical activity in unpolarized EYPC-
SUVs, and despite their almost identical global structure, the
capacity of neutral push-pull rod 10 to depolarize polarized
membranes exceeded that of neutral push-push rod 9 clearly.
This result was identical to that obtained previously for the
neutral push-pull rod 7 and push-push rod 8 (Figure 3).10
Favorable alignment of axial rod dipoles with membrane
(20) Control experiments demonstrated that valinomycin does not affect
the ion transport activity of rods in EYPC-SUVs with H+ but without K+
and Na+ gradients (for a discussion of this effect, see ref 17). The high
sensitivity of this assay toward minimal changes in temperature, that is,
highest importance for strict temperature control (25 ( 0.5°, using a
thermostated sample holder) for successful studies, must be emphasized.
For details on calibration of the assay with regard to valinomycin and
safranin O concentration,19 melittin experiments, and simultaneous pH-
gradient detection using intravesicular HPTS by triple channel fluorescence
kinetics, compare to ref 10b.
(19) (a) Woolley, G. A.; Deber, C. M. Biopolymers 1989, 28, 267-272.
(b) Singh, A. P.; Nicholls, P. J. Biochem. Biophys. Methods 1985, 11, 95-
108.

Electrostatics of Cell Membrane Recognition
J. Am. Chem. Soc., Vol. 123, No. 11, 2001 2523
guest complexes in unpolarized, isoelectric bilayer membranes
(i.e., 4 and 3, Figure 2), fluorescence-depth quenching results
remained partially ambiguous and were suggestive of a more
complex situation (i.e., 38 and 39, Figure 8).
The application of an inside negative membrane potential by
the identical method used for functional studies (Figure 7) had
a profound effect on this comparably poor rod organization in
unpolarized EYPC-SUVs (Figure 8c). A straightforward (Q1 )
Q2) situation was induced for melittin-like 13 (Figure 8B). This
fully corroborated favorable alignment of the TM axial rod
dipole with mammal-type membrane potentials (i.e., host-guest
complex 5, Figures 2, 8). In sharp contrast, negligible quenching
was observed for isomer 14 (Figure 8D). This proved, on the
structural level, inhibition of TM orientation of magainin-like
rods by mammal-type membrane potentials (i.e., host-guest
complex 2, Figures 2, 8) which confirms both unfavorable
charge translocation (Figure 8a) and alignment of dipole with
potential (Figure 8b).
The, thus far, unique variability of axial electrostatic asym-
metry in rigid push-pull rods combined with their facile and
reliable detectability in bilayer membranes, thus, revealed indeed
valuable and novel insights into the structural basis of cell
membrane recognition. Namely, increased activity of melittin-
like rods 13 in polarized bilayers (Figures 6 and 7) was
demonstrated to originate from the stabilization of TM rod
orientation by inside negative potentials (Figure 8A,B). Reduced
activity of magainin-like rods 14 in polarized membranes, on
the other hand, was in excellent agreement with experimental
structural evidence for voltage-induced inhibition of rod binding
in TM orientation.
Lytic Activity of Neutral and Cationic Rods in Unpolar-
ized Bilayer Membranes. The above studies provided unprec-
edented direct structural support that electrostatic rod asymmetry
may account for, for example, the biological activity of the toxic
melittin and the antibiotic magainins. Although consequently
of minor importance for cell membrane recognition, the mech-
anism of ion transport by push-pull rigid-rod ionophores was,
nevertheless, elucidated by dye leakage and planar bilayer
conductance experiments (see Supporting Information). In view
of the capacity of melittin and magainins to mediate transport
of relatively large molecules across bilayer membranes, an
assessment of lytic activity of the corresponding TM push-pull
rods was of particular interest. For this reason, dye leakage from
EYPC-SUVs loaded with CF (5(6)-carboxyfluorescein) at high,
self-quenching concentrations was measured by monitoring
changes in CF fluorescence as a function of time after the
addition of ionophores 9-14 (Figure 9).4a,17 Corroborating
absence of lysis and TM pores >10 Å, no significant CF-efflux
was detectable with all rods. These findings confirmed that cell
membrane recognition and (hemo)lytic activity are, at least in
this case,2,9 unrelated.
Figure 8. Assay for structural studies of cationic rods in unpolarized
and polarized bilayer membranes.17 Quenching of the emission of
fluorescent rods 13 (A, B) and 14 (C, D) in unpolarized (A, C) and
polarized EYPC-SUVs (B, D). Relative emission intensities (I0 - I)
/I0 × 100% were calculated from the emission of rods (0.1 µM) in
unlabeled EYPC-SUVs (I0, 0.25 mM) and EYPC-SUVs doped with
8.9 mol % of either 12-DOXYL-PC (Q1) or 5-DOXYL-PC (Q2). Inside
negative potentials were applied with a K+ gradient and valinomycin
as described for function (Figure 7). Potential-induced structural changes
(c) were not observed in the absence of valinomycin.
the structural basis of their bilayer membrane recognition
mechanism under comparable experimental conditions was
expected to be unproblematic using simplified fluorescence-
depth quenching techniques in polarized and unpolarized EYPC-
SUVs.1
The usefulness of extended rigid fluorophores to simplify
structural studies in bilayer membranes by fluorescence-depth
quenching at relevant concentrations has been demonstrated
extensively.1,4a,17 In brief, EYPC-SUVs were labeled with 8.9
mol % lipids containing a quencher either near the middle (i.e.,
a central quencher Q1, dQ ) 5.8 Å) or near the surface of the
bilayer (i.e., an interfacial quencher Q2, dQ ) 12.2 Å) (Figure
8E). Measurement of the quenching efficiency by Q1 and Q2
on the emission of fluorescent rods in comparison to that in
unlabeled EYPC-SUVs reveals central rod location for (Q1 .
Q2)-efficiency, interfacial rods for (Q1 , Q2), and TM rods for
(Q1 ≈ Q2).1
In unpolarized EYPC-SUVs, quenching of melittin-like rod
13 and its isomer 14 varied between 23% and 34% and did not
depend much on the location of the quencher (Figure 8A,C).
Although this (Q1 ≈ Q2)-situation was in general support of
TM rod orientation, the small differences seen between central
and interfacial quenchers were above experimental error (5%)
and reproducible. According to parallax analysis,22 these dif-
ferences were, however, not sufficiently distinct to indicate
interfacial or central rod location (6.9 Å for 13 and 11.2 Å for
14 from the center of the bilayer). The (Q1 g Q2) situation with
melittin-like 13 may, thus, be best explained by the presence
of a minor population of central rods in addition to a major
population of TM rods (i.e., 38, Figure 8A). The complementary
(Q1 e Q2) situation with isomer 14 would then originate from
the presence of minor interfacial populations in addition to
mainly TM rods (i.e., 39, Figure 8C). The presence of partially
inserted or tilted rods can, however, not be fully excluded.6b
Thus, although roughly consistent with the expectable host-
Conclusions
The objective of this study was to assess the potential of
rationally designed universal probes for comprehensive elucida-
tion of the electrostatic principles underlying cell membrane
recognition on the molecular structural level. Probe design
emphasized the presence of an invariable rigid scaffold, a
unimolecular ion-conducting pathway, and extended fluorophore
in combination with systematically variable axial dipole and
terminal charges. The above results demonstrate that this unique
approach permits direct correlation of functional studies in
unpolarized, anionic, and polarized bilayer membranes with
straightforward structural studies under nearly identical condi-
tions.
(21) For a comparison of these results with magainin 2 and other natural
antibiotics, see: Kobayashi, S.; Takeshima, K.; Park, C. B.; Matsuzaki, K.
Biochemistry 2000, 39, 8648-8654.
(22) Abrams, F. S.; London, E. Biochemistry 1992, 31, 5312-5322.

2524 J. Am. Chem. Soc., Vol. 123, No. 11, 2001
Sakai et al.
an electrostatic axial asymmetry similar to that of the R-helical
bee toxin melittin showed an increased activity in polarized
bilayer membranes that coincided with structural evidence for
a favorable alignment of the axial rod dipole with potentials.
Reduced activity of the isomeric rods with electrostatic axial
asymmetry similar to that of natural antibiotics in polarized
bilayers, on the other hand, was compellingly evidenced to
originate from voltage-induced inhibition of rod incorporation
into membranes with mammal-type polarization. These two
findings provide, to the best of our knowledge, the first direct
experimental evidence of the molecular structural level for
operational cell membrane recognition by electrostatically
asymmetric nanorods.1,23
Experimental Section
See Supporting Information.
Acknowledgment. We thank B. Ghebremariam, Dr. J.-Y.
Winum, Dr. F. Robert, and S. Biltresse for synthetic assistance,
A. Pinto and J.-P. Saulnier and the group of Prof. Gu¨lac¸ar for
NMR and MS, respectively, and Swiss NSF (21-57059.99) for
financial support.
Figure 9. Assay for lytic activity of neutral and cationic rods in
unpolarized bilayer membranes. Top, schematic representation of the
assay. Bottom, changes in fluorescence intensity of intravesicular CF
(50 mM) during the addition of rods (10 µM) and final lysis (excess
Triton X-100). Relative emission intensity ) (It - I0)/(I∞ - I0) × 100%;
λem ) 514 nm, λex ) 492 nm. (Conditions: 0.25 mM EYPC-SUVs in
10 mM HEPES buffer, 107 mM NaCl).
Supporting Information Available: Experimental proce-
dures, RP-HPLC profiles for product mixtures 9/16/17, 9/10/
11 and 17/18/19/20 (Figure 10), vesicle preparation, flux assays,
and single ion channel characteristics in planar bilayer conduc-
tance experiments (Figure 11). This material is available free
of charge via the Internet at http://pubs.acs.org.
In conclusion, general importance for cell membrane recogni-
tion was identified for (a) the interaction between the axial rod
dipole and membrane potential and (b) the TM charge trans-
location but not for (c) the membrane surface charge and (d)
the ion transport mechanism. Two findings were particularly
noteworthy within a consistent set of data collected for several
neutral and cationic rigid-rod ionophores of different electrostatic
axial asymmetry. On the one hand, rigid push-pull rods with
JA003141+
(23) Potential-induced translocation but not reorientation has been
reported previously for asymmetric lantibiotic epilancin K7 rods: Driessen,
A. J. M.; van den Hooven, H. W.; Kuiper, W.; van de Kamp, M.; Sahl,
H.-G.; Konings, R. N. H.; Konings, W. N. Biochemistry 1995, 34, 1606-
1614.