Synthetic transmembrane channels: Functional characterization using solubility calculations, transport studies, and substituent effects

Article abstract of DOI:10.1021/ja962694a

Dibenzyldiaza- 18-crown-6 (PhCH₂CH₂Ph, 1), di(dodecyldiaza-18- crown-6(C₁₂H₂₅C₁₂H₂₅, 2), HOOC(CH₂)₁₁(CH₂)₁₁COOH (3), <18N>(CH₂)₁₂(CH₂)₁₂ (4), (CH₂)₁₂(CH₂)₁₂ (5), C₁₂H₂₅(CH₂)₁₂(CH₂)₁₂C₁₂H₂₅ (6), PhCH₂(CH₂)₁₂(CH₂)₁₂CH₂Ph(7), 4-(p- MeOC₆H₄CH₂C₁₂)₂ (8), (p-NO₂C₆H₄CH₂C₁₂)₂ (9), and [chol-O-(CH₂)₂C₁₂]₂(N18N) (10) were studied. Octanol- water partition coefficients were determined for 1, 6, 7, 8, 10, and 3- cholestanyl-OCOCH₂(CH₂)₁₂(CH₂)₁₂COCH₂O-3- cholestanyl (11). All were found to favor octanol, and by implication the phospholipid bilayer membrane, by at least 10⁴-fold. Transport of Na⁺ was assessed in both a phospholipid bilayer and in a bulk CHCl₃ membrane phase. Addition of ionophores to the latter was found in some cases to strongly enhance CHCl₃ phase hydration. An attempt to correlate transport rates determined in the two systems failed, suggesting that the carrier mechanism, required in the CHCl₃ phase, does not apply to the tris(macrocyclic) compounds in the bilayer. Sodium transport rates were also assessed for these compounds by using the bilayer clamp technique. Although Na⁺ flux rates thus determined for 7-9 in the phospholipid bilayer did not correlate with results obtained by the ²³Na-NMR technique, the traces are similar to those obtained with protein channels, further supporting the function of tris(macrocycle)s as channel formers.

Full text of DOI:10.1021/ja962694a

5540
J. Am. Chem. Soc. 1997, 119, 5540-5549
Synthetic Transmembrane Channels: Functional
Characterization Using Solubility Calculations, Transport
Studies, and Substituent Effects
Oscar Murillo, Iwao Suzuki, Ernesto Abel, Clare L. Murray, Eric S. Meadows,
Takashi Jin, and George W. Gokel*
Contribution from the Bioorganic Chemistry Program and Department of Molecular Biology &
Pharmacology, Washington UniVersity School of Medicine, 660 South Euclid AVenue,
Campus Box 8103, St. Louis, Missouri 63110
ReceiVed August 2, 1996. ReVised Manuscript ReceiVed March 27, 1997X
Abstract: Dibenzyldiaza-18-crown-6 (PhCH2 N18N CH2Ph, 1), di(dodecyldiaza-18-crown-6 (C12H25 N18N C12H25,
2), HOOC(CH2)11 N18N (CH2)11COOH (3), 18N (CH2)12 N18N (CH2)12 N18 (4), N18N (CH2)12 N18N (CH2)12 N18N
(5), C12H25 N18N (CH2)12 N18N (CH2)12 N18N C12H25 (6), PhCH2 N18N (CH2)12 N18N (CH2)12 N18N CH2Ph (7),
4-(p-MeOC6H4CH2 N18N C12)2 N18N (8), (p-NO2C6H4CH2 N18N C12)2 N18N (9), and [chol-O-(CH2)2 N18N -
C12]2 N18N (10) were studied. Octanol-water partition coefficients were determined for 1, 6, 7, 8, 10, and
3-cholestanyl-OCOCH2 N18N (CH2)12 N18N (CH2)12 N18N COCH2O-3-cholestanyl (11). All were found to favor
octanol, and by implication the phospholipid bilayer membrane, by at least 104-fold. Transport of Na+ was assessed
in both a phospholipid bilayer and in a bulk CHCl3 membrane phase. Addition of ionophores to the latter was found
in some cases to strongly enhance CHCl3 phase hydration. An attempt to correlate transport rates determined in the
two systems failed, suggesting that the carrier mechanism, required in the CHCl3 phase, does not apply to the
tris(macrocyclic) compounds in the bilayer. Sodium transport rates were also assessed for these compounds by
using the bilayer clamp technique. Although Na+ flux rates thus determined for 7-9 in the phospholipid bilayer did
not correlate with results obtained by the 23Na-NMR technique, the traces are similar to those obtained with protein
channels, further supporting the function of tris(macrocycle)s as channel formers.
Introduction
have been prepared. In these cases, the amphiphilic monomers
are presumably in flux within their respective bilayer leaflets
(lateral relaxation) and conduction occurs when two elements
occupy positions opposite to each other. The monomer systems
of Tabushi,8 Menger,9 Kobuke,10 and Regen11 are representative
of this approach.
One of the most intensely studied areas of modern biology
concerns the function and mechanism of transmembrane ion
channels.1 Numerous proteins span bilayer membranes and
effect the transport of cations, anions, and molecules across the
boundaries of cells and organelles. During the past decade, an
enormous amount of information has accumulated on which
proteins are responsible for the transport of ionic or molecular
species2 but physical chemical details of transport and selectivity
remain largely obscure.3
This ardent biological interest has fostered attempts to develop
chemical models designed to help to understand the structural
and kinetic requirements of cation transport. The various model
systems reflect unique approaches4 to the design and synthetic
problems. Synthetic peptides have been prepared that span
membranes and associate into “bundles” capable of ionophoretic
activity.5 A number of “half-channel elements,” perhaps
inspired by the natural ionophores gramicidin6 or amphotericin,7
A number of these compounds have been designed as “tunnel-
like” structures intended to emulate the structure of bacterio-
rhodopsin,12 which forms a pore from seven transmembrane
helical protein segments. Notable examples of this include the
pioneering designs of Fyles13 and Lehn.14 Recently Ghadiri and
co-workers15 have constructed cyclic, peptidic “nanotubes” that
conduct cations. Voyer and Robitaille16 have used an R-helical
amino acid backbone to which are appended appropriately
(5) (a) Mutter, M.; Tuchscherer, G. G.; Miller, C.; Altman, K.-H.; Carey,
R. I.; Wyss, D. F.; Labhardt, A. M.; Rivier, J. E. J. Am. Chem. Soc. 1992,
114, 1463. (b) Åkerfeldt, K. S.; Lear, J. D.; Wasserman, Z. R.; Chung, L.
A.; DeGrado, W. F. Acct. Chem. Res. 1993, 26, 191-197.
(6) (a) Urry, E. W.; Goodall, M. C.; Glickson, J. D.; Meyers, D. F. Proc.
Natl. Acad. Sci. U.S.A. 1971, 68, 1907-1911. (b) Urry, D. W. Proc. Natl.
Acad. Sci. U.S.A. 1971, 68, 672-676.
(7) Bolard, J. Biochim. Biophys. Acta 1986, 864, 257-304.
(8) Tabushi, I.; Kuroda, Y.; Yokota, K. Tetrahedron Lett. 1982, 4601-
4604.
(9) Menger, F. M.; Davis, D. S.; Persichetti, R. A.; Lee, J.-J. J. Am.
Chem. Soc. 1990, 112, 2451-2452.
(10) (a) Kobuke, Y.; Ueda, K.; Sokabe, M. J. Am. Chem. Soc. 1992,
114, 7618-1722. (b) Tanaka, Y.; Kobuke, Y.; Sokabe, M. Angew. Chem.,
Int. Ed. Engl. 1995, 34 (6), 693-694.
X Abstract published in AdVance ACS Abstracts, June 1, 1997.
(1) Nicholls, D. G. Proteins, Transmitters, and Synapses; Blackwell
Science: Oxford, U.K., 1994.
(2) (a) Henderson, R.; Baldwin, J. M.; Ceska, T. A.; Zemlin, F.;
Beckmann, E.; Downing, K. H. J. Mol. Biol. 1990, 213, 899-929. (b)
Deisenhofer, J.; Epp, O.; Miki, K.; Huber, R.; Michel, H. Nature 1985,
318, 618-624. (c) Baldwin, J. M. EMBO J. 1993, 12, 1693-1703.
(3) (a) Stein, W. D. Channels, Carriers, and Pumps; Academic Press:
New York, 1990. (b) Hille, B. Ionic Channels of Excitable Membranes;
Sinauer Press: Sunderland, MA, 1992.
(4) (a) Neevel, J. G.; Nolte, R. Tetrahedron Lett. 1984, 25 (21), 2263-
2266. (b) Nolte, R. J. M.; Beijnen A. J. M.; Neevel, J. G.; Zwikker, J. W.;
Verkley, A. J.; Drenth, W. Israel. J. Chem. 1984, 24, 297-301. (c) Kragten,
U. F.; Roks, M. F. M.; Nolte, R. J. M. J. Chem. Soc., Chem. Commun.
1985, 1275-1276. (d) Fuhrhop, J.-H.; Liman, U.; David, H. H. Angew.
Chem., Int. Ed. Engl. 1985, 24, 339-340. (e) Jullien, L.; Lehn, J.
Tetrahedron Lett. 1988, 29, 3803-3806.
(11) (a) Stadler, E.; Dedek, P.; Yamashita, K.; Regen, S. J. Am. Chem.
Soc. 1994, 116, 6677-6682. (b) Deng, G.; Merritt, M.; Yamashita, K.;
Janout, V.; Sadownik, A.; Regen, S. L. J. Am. Chem. Soc. 1996, 118, 3308.
(12) (a) Henderson, R.; Baldwin, J. M.; Ceska, T. A.; Zemlin, F.;
Beckmann, E.; Downing, K. H. J. Mol. Biol. 1990, 213, 899-929. (b)
Deisenhofer, J.; Epp, O.; Miki, K.; Huber, R.; Michel, H. Nature 1985,
318, 618-624. (c) Baldwin, J. M. EMBO J. 1993, 12, 1693-1703.
S0002-7863(96)02694-7 CCC: $14.00 © 1997 American Chemical Society

Synthetic Transmembrane Channels
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5541
positioned crown ethers so that a “tunnel” of macrocycles is
available to the transmigrating cation.
Scheme 1
Our own approach involved the use of crown ether com-
pounds to function both as “head groups” and as a cation relay
positioned within the phospholipid bilayer.17 The most impor-
tant feature of our design is the use of a “flexible framework”
which possesses what we believe are the essential elements for
function. The system still has enough flexibility to adapt to
unanticipated requirements. This approach is based upon a
recognition that the most rigid and stable structures often lack
the dynamics required to achieve the desired function. We have
recently reported a number of structural variations on our
original design, and we have correlated these changes with
differences in sodium cation flux.18 The emergence of “single-
strand” channels such as the Isk protein19 suggests that nontu-
bular, membrane-spanning monomers are an important natural
alternative to bacteriorhodopsin or the acetylcholine receptor.
The “7-TM” proteins clearly form a tunnel-like physical
arrangement in the bilayer that is impossible with shorter
peptides. Even so, certain single-strand compounds exhibit ion
channel function. Although many channel compounds have
been characterized in remarkable detail, the mechanism of
transport for either multipass proteins or single-strand peptides
remains unknown. It is therefore unclear whether the mecha-
nism is the same or different.
In the present work, we have undertaken a three-part study
in an effort to further characterize our synthetic channel model
system. First, we have measured and/or calculated octanol-
water partition coefficients for several simple crown ethers and
channel model compounds. It is possible that the tris(macro-
cyclic) amphiphiles transport cations to the same extent and
the differences observed in activity reflect only differential
solubility. Second, we have examined transport rates for
channel formers and known carriers in both lipid bilayers and
in a bulk organic membrane system. Transport in the latter
system cannot occur by a channel mechanism because the
membrane thickness is, by molecular standards, enormous.
Assessment of transport rates using the two membrane systems
and comparisons among compounds shed light on the function
of these synthetic ionophores. Third, we have used the bilayer
clamp technique to assess the Na+ flux rates for three benzyl-
substituted tris(macrocycle) derivatives having different sub-
stituents in the benzyl group’s 4-position. The single-channel
conductivity data were acquired by using an asolectin phos-
pholipid bilayer membrane and a patch clamp amplifier
equipped with a bilayer head stage. The conductivity traces
are identical in most respects to those observed for protein
channels.
Results and Discussion
Compounds Studied. Several of the compounds used in the
present study possess somewhat complicated structures. We
use here a shorthand system that we have employed previously20
to describe these and related systems. A crown is represented
using angle brackets and a number to indicate ring size. Thus,
18 is shorthand for 18-crown-6. Aza-15-crown-5 can be
represented as N15 or 15N and hydrogen (normally under-
stood) or an additional abbreviation may be added to indicate
the presence of a substituent, e.g. 15N H or 15N R. 4,13-
Diaza-18-crown-6, the most common macrocycle used in this
study, is represented as N18N .
Gramicidin, a naturally occurring channel former, and vali-
nomycin, a mitochondrial potassium carrier,21 were obtained
commercially. Dibenzyldiaza-18-crown-6 (PhCH2 N18N CH2Ph,
1), and didodecyldiaza-18-crown-6 (C12H25 N18N C12H25, 2),
18N (CH2)12 N18N (CH2)12 N18 (4), N18N (CH2)12 N18N -
(CH2)12 N18N (5), C12H25 N18N (CH2)12 N18N (CH2)12 N18N -
C12H25 (6), [PhCH2 N18N (CH2)12]2 N18N (7), [chol-O-
(CH2)2 N18N C12]2 N18N (10), and [chol-O-CH2CO N18N -
C12]2 N18N (11) have been previously reported.17 The car-
boxy-terminated derivative of 2, HOOC(CH2)11 N18N -
(CH2)11COOH (3), and the p-methoxy and p-nitro derivatives
of 7, i.e., 8 and 9, respectively, are described in the Experimental
Section.
Synthetic Access. The tris(macrocycle)s described herein
were prepared by a sequence similar to that previously
reported.17 Thus, 4,13-diaza-18-crown-6 was treated with 1,2-
dibromododecane to afford Br(CH2)12 N18N (CH2)12Br. 4,13-
Diaza-18-crown-6 was monoalkylated by treatment with 4-ni-
trobenzyl bromide, sodium carbonate, and potassium iodide in
butyronitrile. Along with the expected dialkylated derivative,
the desired product was obtained in 44% yield. Reaction of 2
equiv of ArCH2 N18N H with Br(CH2)12 N18N (CH2)12Br,
Na2CO3 and KI in CH3CH2CH2CN afforded tris(macrocycle)
9 in 39% yield as a yellow solid, mp 61.5-62 °C. The approach
is shown in Scheme 1.
Compound 3 (HOOC(CH2)11 N18N (CH2)11COOH), was
prepared from 12-bromododecanoic acid, which was protected
as its benzyl ester. Bis(alkylation) gave the diester PhCH2OOC-
(CH2)11 N18N (CH2)11COOCH2Ph. Hydrogenolysis afforded
the diacid (3), which could be further functionalized but was
used as obtained in the present studies.
(13) (a) Carmichel, V. E.; Dutton, P.; Fyles, T.; James, T.; Swan, J.;
Zojaji, M. J. Am. Chem. Soc. 1989, 111, 767-769. (b) Fyles, T.; James,
T.; Kaye, K. Can. J. Chem. 1990, 68, 976-978. (c) Fyles, T.; Kaye, K.;
James, T.; Smiley, D. Tetrahedron Lett. 1990, 1233. (d) Kaye, K.; Fyles,
T. J. Am. Chem. Soc. 1993, 115, 12315-12321. (e) Fyles, T.; James, T.;
Pryhitka, A.; Zojaji, M. J. Org. Chem. 1993, 58, 7456-7468.
(14) (a) Jullien, L.; Lehn, J.-M. Tetrahedron Lett. 1988, 3803-3806.
(b) Jullien, L., Lehn, J.-M. J. Inclusion Phenom. 1992, 12, 55-74. (c) Pregel,
M. J.; Jullien, L.; Lehn, J.-M. Angew. Chem., Int. Ed. Engl. 1992, 31, 1637-
1640 (d) Canceill, J.; Jullien, L.; Lacombe, L.; Lehn, J.-M. HelV. Chim.
Acta 1992, 75, 791.
(15) (a) Ghadiri, M. R.; Granja, J. R.; Buehler, L. K. Nature 1994, 369,
301-304. (b) Khazanovich, N.; Granja, J. R.; McRee, D. E.; Milligan, R.
A.; Ghadiri, M. R. J. Am. Chem. Soc. 1994, 116, 6011-6012.
(16) (a) Voyer, N.; Robitaille, M. J. Am. Chem. Soc. 1995, 117, 6599-
6600. (b) Voyer, N.; Lamothe, J. Tetrahedron 1995, 51, 9241.
(17) Nakano, A.; Xie, Q.; Mallen, J.; Echegoyen, L.; Gokel, G. W. J.
Am. Chem. Soc. 1990, 112, 1287.
Compound Solubility Properties. It might be argued that
the compounds of the present study are all amphiphiles that
behave as detergents. The mode of action would presumably
be to create a defect in the bilayer. Differences in transport
might then be due to differential membrane solubility rather
(18) Murillo, O.; Watanabe, S.; Nakano, A.; Gokel, G. W. J. Am. Chem.
Soc. 1995, 117, 7665-7679.
(19) (a) Takuma, T. NIPS 1993, 8, 175. (b) Takuma, T.; Ohkubo, H.;
Nakanishi, S. Science 1988, 242, 1042.
(20) Hernandez, J. C.; Trafton, J. E.; Gokel, G. W. Tetrahedron Lett.
1991, 6269.
(21) Brockmann, H.; Schmidt-Kastner, G. Chem. Ber. 1955, 88, 57.

5542 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Murillo et al.
Table 1. Octanol-Water Partitioning Coefficients for Crown
Ether Derivatives
no.
compd
Poct(exptl) Poct(calcd)d
12-crown-4
15-crown-5
18-crown-6
benzo-15-crown-5
benzo-33-crown-11
p-tert-butylbenzo-27-crown-9
p-tert-butylbenzo-30-crown-10
dibenzo-18-crown-6
dibenzo-24-crown-8
dibenzo-27-crown-9
5,5-dibenzo-30-crown-10
2,8-dibenzo-30-crown-10
4,7-dibenzo-33-crown-11
8,8-dibenzo-48-crown-16
N,N-dibenzyl-4,13-diaza-18-crown-6
C12 N18N C12 N18N C12 N18N C12
[PhCH2 N18N C12]2 N18N
(p-MeOC6H4CH2 N18N C12)2 N18N
(p-O2NC6H4CH2 N18N C12)2 N18N
0.92a
0.33a
0.21a
1.60c
0.24c
1.89b
1.69b
2.20c
3.54c
3.30c
1.94c
3.32c
3.39c
1.43c
4.21
-
0.57
0.34
0.17
0.91
0.23
2.15
1.65
3.70
2.11
1.63
1.80
1.82
1.45
0.52
4.10
18.48
11.23
11.26
11.76
30.08
30.83
1
6
7
8
9
-
-
-
10 [chol-O-(CH2)2 N18N C12]2 N18N
11 (chol-O-CO-CH2 N18N C12)2 N18N
-
-
a Hendrixson, R. R.; Mack, M. P.; Palmer, R. A.; Ottolenghi, A.;
b
Ghiradelli, R. G. Toxic. Appl. Pharm. 1978, 44, 263. Stolwijk, T.
B.; Vos, L. C.; Sudholter, E. J. R.; Reinhoult, D. N. J. Am. Chem. Soc.
1989, 111, 6321. c Stolwijk, T. B.; Vos, L. C.; Sudholter, E. J. R.;
Reinhoult, D. N. Recl. TraV. Chim. Pays-Bas 1989, 108, 103.
d Calculations were made using the HINT module of the Sybyl
molecular modeling package. Molecules were minimized using TRI-
POS force field and Gasteiker-Huckel charges. All atoms were taken
into account (including hydrogens), the solvent condition was neutral,
the polar proximity was calculated using a through-space function
me(-nr), where e ) exponential, r ) radius, m ) 5 and n ) 1.
30-crown-10.23 We experimentally determined the partition
coefficient for dibenzyldiaza-18-crown-6. These solubilities
were then compared with the experimentally determined values
for calibration. Keeping in mind that P values are reported on
a logarithmic scale, the agreement apparent in the data shown
in Table 1 is excellent.
The agreement between calculated and experimentally de-
termined values for simple crown ethers is generally good. In
particular, the agreement for dibenzyl-4,13-diaza-18-crown-6
(1, 4.2 Vs 4.1) is excellent. A value of 4 means that the
compound favors octanol over water by a factor of 104. Thus,
a compound having a P value of 2 would partition 99% into
n-octanol and one with a value of 4 would partition 99.99%
into octanol. In terms of ionophore function, any value greater
than 2 would leave almost no ionophore in the aqueous phase.
We have been unable to experimentally determine log P
values for channel compounds (6-11) because there is no
detectable concentration of them in water. The calculated values
show clearly that this is expected. Compounds 6-11 all favor
the hydrophobic solvent by >1010. The suggestion that differ-
ences in transport efficacy are due to differential solubility of
the ionophores in the bilayer is therefore not supported.
Picrate Extraction Constants. The ability of the compounds
studied here to complex Na+ can be assessed by the picrate
extraction method.24 In this experiment, sodium picrate is
contacted by a mixture of H2O and CHCl3. The yellow picrate
salt remains in water. Addition of a host that can complex
cations and partition into CHCl3 causes a yellow color to develop
in the lipophilic phase. The amount of picrate anion in CHCl3
can be assessed quantitatively by UV-vis spectroscopy. The
than any intrinsic transport efficacy. Whatever variations in
solubility may be observed for these compounds, it is clear that
the ionophore must partition into the membrane to effect
transport. Relatively few standard solubility data are available
in the literature for crown systems, and none is recorded for
the compounds of interest here. The determination of solubility
data for all-aliphatic channel 6 is problematic because it lacks
UV- or fluorescence-active residues that would permit the
detection of very low concentrations of material in water.
Solubility was determined by a comparison of experimental
determined octanol-water partition coefficients (“P values”)
with values calculated by using the HINT (solubility) module
of the SYBYL molecular modeling package. The calculations
take into account all atoms (including H). The “solvent
condition” parameter in this program was set to neutral, and
the “polar proximity” was calculated using the through-space
function me(-nr), where e ) exponential, m ) 5, n ) 1, and r
) radius.
Five octanol-water partition coefficients for crown ethers
are reported in the literature. These are for 3n-crown-n (n )
4-6),22 p-tert-butyl-benzo-27-crown-9, and p-tert-butyl-benzo-
(23) (a) Stolwijk, T. B.; Vos, L. C.; Sudholter, E. J. R.; Reinhoult, D.
N. J. Am. Chem. Soc. 1989, 111, 6321. (b) Stolwijk, T. B.; Vos, L. C.;
Sudholter, E. J. R.; Reinhoult, D. N. Recl. TraV. Chim. Pays-Bas 1989,
108, 103.
(22) Hendrixson, R. R.; Mack, M. P.; Palmer, R. A.; Ottolenghi, A.;
Ghiradelli, R. G. Toxicol. Appl. Pharmacol. 1978, 44, 263.

Synthetic Transmembrane Channels
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5543
Table 2. Sodium Cation Binding and Transport by Ionophoresa
rate × 108
mol‚h-1
rel rate
(CHCl3)
picrate
extn %
rel rate
(bilayer)
no.
ionophore
valinomycin
gramicidin
PhCH2 N18N CH2Ph
C12 N18N C12
HOOC(CH2)11 N18N (CH2)11COOH
18N C12 N18N C12 N18
H N18N C12 N18N C12 N18N H
C12 N18N C12 N18N C12 N18N C12
(PhCH2 N18N C12)2 N18N
(p-MeOC6H4CH2 N18N C12)2 N18N
(p-O2NC6H4CH2 N18N C12)2 N18N
4.17
0.08
2.22
2.01
NDb
2.4
1.13
1.1
1.9
1.0
16.6
na
13.7
17
0.14
1.00
0.01
0.01
0.01
0.02
0.28
0.28
0.38
0.43
0.30
0.02
0.02
0.53
0.48
0
0.58
0.27
0.26
0.46
0.43
0.44
0.62
1
2
3
4
5
6
7
8
9
1.5
37.3
27.9
37
26.2
26
16.4
29.1
1.8
1.85
2.6
c
10
[chol-O-(CH2)2 N18N C12]2 N18N
b
a See text for details. ND is “not determined.” c “Chol” is 3-dihydrocholestanyl.
amount of Na+ is assumed to be equimolar, so its concentration
is known as well. Extraction constants are normally expressed
as a percentage of available cation. Note that the extraction
constant tells only how much of the available Na+ is extracted
(complexed) and not how many of the macrorings are involved
in the process.
Extraction constants were determined for valinomycin and
compounds 1-10. The method used 3.0 mL each of H2O and
CHCl3; the compound under study was 1.0 mM in CHCl3. The
aqueous sodium picrate solution was 1.0 mM in Na+(picrate)-
and 0.0975 M in NaOH. The UV spectrum (λ ) 354 nm) was
evaluated quantitatively to obtain the extraction constants which
are recorded in Table 2. Further details of the procedure may
be found in the Experimental Section.
We may use diaza-18-crown-6 derivatives 1 and 2 as
“baseline” compounds for our consideration of cation binding.
These two structures possess the diaza-18-crown-6 framework
that is the basic structural unit of the other systems. Whether
the sidearms are benzyl (1) or n-dodecyl (2), the extraction
constant is 15 ( 2%. Adding two aza-18-crown-6 units at the
end of each dodecyl chain increases the extraction constant to
∼37%, or about 2.5-fold. A similar increase is observed when
N18N C12 is added to the end of each dodecyl chain (f6).
Surprisingly, the corresponding substitution by N18N CH2Ph
(f7) leads to an extraction constant of only 26%. A similar
value is found if benzyl is altered to 4-methoxybenzyl (8, 26%)
or if the chain is eliminated entirely and the crown left as
N18N H (5, 28%). The cholestanyl sidearm of 10 gave slightly
increased (29%) extraction, but the only major difference
resulted from attachment of a 4-NO2 group to 7 to give 9 (16%).
Transport Experiments. We have previously reported that
sodium transport in a phospholipid bilayer is effected by
tris(macrocycle)s such as 5 and 6.20 The inability of 4 to
function as a channel may be due to the lack of an additional
residue or functional group that would stabilize it within the
bilayer. In 6, this residue is the N-dodecyl chain that is held in
the membrane. In 7, it is a benzyl group that may interact with
phospholipid head groups. Compound 5 may be stabilized at
the membrane head group by protonation of a nitrogen atom,
only one of which is available in 4.
observed for even the best carriers.25 The difference in efficacy
is apparent from a comparison of bilayer transport rates for two
naturally-occurring peptides: gramicidin (channel) and valino-
mycin (carrier). We have measured the valinomycin-mediated
transport rate for Na+ in a phospholipid bilayer. To our
knowledge, this has never before been reported. The observed
rate in the present experimental system is ∼25 s-1
. This
corresponds to about 14% of the rate exhibited by gramicidin
under identical conditions.
Second, compounds such as 6 are amphiphilic and are known
to form stable membranes26 using the diazacrown residues as
head groups. The mobility of an amphiphile in a membrane is
normally high within a bilayer leaflet (lateral relaxation). If
the present compounds are functioning as carriers, they would
be required to pass through to the opposite leaflet, an obviously
unfavorable process akin to transverse relaxation. Such chemi-
cal processes are known to occur in membranes, but the
difference in rates for the two is usually many powers of 10
(lateral > transverse).
Third, if 6 [C12 N18N C12 N18N C12 N18N C12] functions
as a carrier, then analogues of it such as C12 N18N C12 N18N C12
should be almost as active. We have previously shown that
transport of Na+ by this bolaamphiphile (NMR method) does
not occur at a detectable rate in the phospholipid bilayer
system.18 The same is true for C12 N18N C12 (2).
Notwithstanding these arguments, we felt it would be
worthwhile to assess the transport ability of this family of
compounds in a bulk organic membrane since many potential
ionophores have been studied using this technique (see Table
2).
Techniques Used in Transport Studies. The ability of
valinomycin, gramicidin, and compounds 1-10 to transport Na+
was assessed by using two different techniques. Sodium cation
transport through a phosphatidylcholine/phosphatidyl glycerol
bilayer can be assessed directly by using the 23Na-NMR
technique developed originally by Riddell and co-workers.27 The
vesicle system is prepared in the presence of NaCl and then
Dy3+ is added to the external medium. Two signals are
observed for 23Na, their exchange rate can be measured, and
(25) Stryer, L. Biochemistry, 4th ed.; W. H. Freeman and Co.: New
York, 1995; p 274.
(26) Mun˜oz, S.; Malle´n, J. V.; Nakano, A.; Chen, Z.; Echegoyen, L.;
Gay, I.; Gokel, G. W. J. Chem. Soc., Chem. Commun. 1992, 520-522. (b)
Mun˜oz, S.; Malle´n, J.; Nakano, A.; Chen, Z.; Gay, I.; Echgoyen, L.; Gokel,
G. W. J. Am. Chem. Soc. 1993, 115, 1705-1711.
(27) (a) Riddell, F. G.; Hayer, M. K. Biochim. Biophys. Acta 1985, 817,
313-317. (b) Buster, D. C.; Hinton, J. F.; Millett, F. S.; Shungu, D. C.
Biophys. J. 1988, 53, 145-152. (c) Riddell, F. G.; Arumugam, S.; Brophy,
P. J.; Cox, B. G.; Payne, M. C. H.; Southon, T. E. J. Am. Chem. Soc. 1988,
110, 734-738. (d) Riddell, F. G.; Arumugam, S. Biochim. Biophys. Acta
1989, 984, 6-10. (e) Riddell, F. G.; Tompsett, S. J. Biochim. Biophys. Acta
1990, 1024, 193-197.
We expected 6 to transport cations by a channel, rather than
by a carrier, mechanism for three reasons. First, cation transport
mediated by protein channels is usually 103-104 greater than
(24) (a) Araki, T.; Tsukube, H. Liquid Membranes: Chemical Applica-
tions; CRC Press: Boca Raton, FL, 1990. (b) Izatt, R. M.; Clark, G. A.;
Bradshaw, J. S.; Lamb, J. D.; Christensen, J. J. Separation Purif. Methods
1986, 15, 21. (c) Christensen, J. J.; Lamb, J. D.; Brown, P. R.; Oscarson,
J. L.; Izatt, R. M. Separation Sci. Technol. 1981, 16, 1193. (d) Wong, K.
H.; Yagi, K.; Smid, J. J. Membrane Biol. 1974, 18, 379. (e) Strzelbicki, J.;
Charewicz, W. A.; Liu, Y.; Bartsch, R. A. J. Inclusion Phenom. 1989, 7,
349.

5544 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Murillo et al.
an exchange rate constant can be obtained in the presence of a
suitably active ionophore. In all cases, gramicidin was run
concurrently with any new sample and rate constants obtained
were normalized to a value of 1.00 corresponding to its observed
exchange rate (∼175 s-1).
Bulk liquid membrane transport rates were measured using
the concentric tube apparatus developed by Izatt and co-
workers.28 The device consists of a hollow glass tube (10 mm
o.d.) inserted into a tall-form beaker (25 mL). The bulk organic
membrane used was either CHCl3 or CDCl3 (6 mL). The source
phase was aqueous sodium picrate (1 mM) in the presence of
excess NaOH (100 mM). The receiving phase was initially
distilled, deionized water. The membrane was stirred magneti-
cally. Aliquots were withdrawn at the specified intervals and
analyzed by UV-visible spectroscopy.
The ability of such compounds as dibenzyldiaza-18-crown-6
(2) and didodecyldiaza-18-crown-6 (3) to transport cations
through a bulk liquid membrane is well-established by numerous
experimental studies.29 As expected, both 2 and 3 transported
sodium picrate (as judged by measurement of changes in picrate
concentration) with rates of 2.22 × 10-8 and 2.01 × 10-8
mol‚h-1, respectively. These values compare with an experi-
mentally determined rate of 4.17 × 10-8 mol‚h-1 for valino-
mycin. Under identical conditions, gramicidin exhibited a
negligible transport rate of 0.08 × 10-8 mol‚h-1, which was
barely above simple diffusion (blank value: 0.013 × %10-8
mol‚h-1). Channel former 6 was also found to transport Na+,
but its rate (1.1 × 10-8 mol‚h-1) was less than one-half that
for either 2 or 3. Compounds 4 and 5, which constitute major
fragments of the channel structure, were found to transport
sodium picrate with rates of 2.4 × 10-8 and 1.13 × 10-8
mol‚h-1, respectively. It is noteworthy that although 5 and 6
function similarly as carriers and are both inferior to 4, exactly
the opposite relationship is observed in a phospholipid bilayer
(see column 4 of Table 2).
Figure 1. Plot of H2O content in CHCl3 as a function of time in the
absence and in the presence of compounds 5-7.
(∼22 °C), “dry” CHCl3 was found to quickly hydrate to an
equilibrium water concentration of ∼140 mM. Our system is
in contact with aqueous sodium picrate, and this could have a
2-fold effect on hydration, although contact with a salt might
be expected to reduce the chloroform aquation. A slight
solubility of the picrate in the absence of carrier might account
for a small, additional amount of water brought into the
chloroform phase along with the sodium cation.
In the presence of 5, a similar extent of hydration was
observed. In the presence of compound 7 (PhCH2 N18N C12-
N18N C12 N18N CH2Ph), the membrane hydrated to the extent
of about 250 mM but had reached the equilibrium level of 140
mM by 24 h. The most dramatic result was observed for 6
(C12 N18N C12 N18N C12 N18N C12), which caused the mem-
brane to reach a water concentration of ∼750 mM within 30
min. The water was present as ∼3000 Å droplets as judged by
laser light scattering. After several hours, the concentration of
water diminished and, by 24 h, the H2O and CHCl3 were
essentially in equilibrium.
Bulk Membrane Hydration. During the course of transport
studies with 6, a distinct cloudiness developed in the bulk CHCl3
membrane. It seemed reasonable to assume that water was
partitioning into the CHCl3 phase. This membrane hydration
was not apparent in most other cases, so the extent and rate of
hydration were determined experimentally by using NMR
methods.
There are two obvious questions related to these observations.
First, why does the membrane hydrate, and second, why does
the amount of water diminish in an apparent return to equilib-
rium? The answer to the second question is probably that the
water is gradually “salted out” of the membrane by an increasing
concentration of sodium picrate in the receiving side phase. Why
the membrane hydrates so extensively in the presence of 6 is a
more difficult and intriguing question, the answer to which is
currently unclear but may be due to the formation of a hydrated
pseudo-micelle.
The extent of hydration was quantified as follows. During
the transport experiment involving 1 mM ionophore, 100-µL
aliquots were withdrawn from the CHCl3 (CDCl3 in this case)
phase. The samples were diluted with 500 µL of additional
CDCl3, and a stock solution (25 µL) of 1,4-dinitrobenzene (75
mM) in CDCl3 was added. Comparison of the integrals of water
and the single, downfield (8.4 ppm) resonance of dinitrobenzene
permitted quantitative measurement of the former. The data
are plotted in Figure 1 for compounds 5-7 along with data for
the blank.
Comparison of Transport Rates in a Phospholipid Bilayer
and in Chloroform. An important purpose of the studies
described here was to determine if the use of a bulk liquid
(CHCl3) membrane transport system could provide any informa-
tion about the chemical behavior of single-stranded, synthetic
channel compounds. In previous work in our own group, we
studied a series of 12-, 15-, and 18-membered ring azamacro-
cycles having alkyl-substituted ester or amide sidearms appended
at nitrogen [e.g., 15N CH2COO(CH2)9CH3].20 The sidearms
ranged from 5 to 18 carbon atoms in length. In that study,
picrate extraction constants and homogeneous equilibrium
binding constants (log KS) were both found to correlate
reasonably well with transport rates in a bulk chloroform
membrane. The correlation (r factor) between transport and
Chloroform is reported30 to absorb temperature-dependent
percentages of water. In the range 20-25 °C, the equilibrium
amount of water reported to be present in CHCl3 (presumably
ethanol-free) is 60-80 mM. In our own experimental system
(28) Lamb, J. D.; Christensen, J. J.; Izatt, S. R.; Bedke, K.; Astin, J. S.;
Izatt, R. M. J. Am. Chem. Soc. 1980, 102 (10), 3399. (b) Lamb, J. D.; Izatt,
R. M.; Garrick, D. G.; Bradshaw, J. S.; Christensen, J. J. J. Membr. Sci.
1981, 9, 83-107.
(29) Moyer, B. A. Complexation and transport. In ComprehensiVe
Supramolecular Chemistry; Gokel, G. W., Ed.; Pergamon Press: Oxford,
U.K., 1996; Vol. 1, Chapter 10, p 377.
(30) (a) Stephen, H., Stephen, T. Eds. Solubilities of Inorganic and
Organic Compounds; Pergamon Press: New York, 1963; Vol. 1, Part 1.
(b) Linke, W. F. Solubilities, Inorganic and Metal-Organic Compounds,
4th ed.; D. Van Nostrand Co., Inc.: New Jersey, 1958; Vol 1.

Synthetic Transmembrane Channels
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5545
either extraction constants or binding constants was, respectively,
0.89 (10 points) and 0.93 (17 log KS values).
Among compounds 1-10, valinomycin is expected to be the
best carrier in CHCl3 and gramicidin is expected to be the most
effective channel former in a bilayer. All rate data are therefore
normalized to values of 1.0 for these respective standards. We
anticipated that compounds such as N,N′-dibenzyl-4,13-diaza-
18-crown-6 (1) and N,N′-didodecyl-4,13-diaza-18-crown-6 (2)
would be excellent carriers when studied in this system.
Gramicidin, in order to transport cations, must dimerize to form
a membrane-spanning, channel-like structure within a bilayer.
Since this is impossible in the bulk CHCl3 membrane, grami-
cidin was expected to function poorly in this experimental
system.
Figure 2. Plot of transport rate Vs extraction constant for compounds
1-10.
Dialkyl crowns such as 1 and 2 were not expected to be as
effective channels in a bilayer, but compounds such as 6
(C12 N18N C12 N18N C12 N18N C12) might exhibit efficacy in
both types of membranes. From the data in Table 1, it is clear
that certain of the above expectations are met. It is especially
interesting to note that valinomycin exhibits transport activity
in a lipid bilayer that is about 14% that of gramicidin. To our
knowledge, valinomycin transport has not been previously
studied in a bilayer using NMR methods. Also as anticipated,
the efficacy of gramicidin in CHCl3 is poor (8% that of
valinomycin).
A comparison of compounds 4 and 5 is particularly interest-
ing. The Na+ transport rate by 5 ( N18N C12 N18N C12 N18N )
in a bilayer is identical to that of 6. On the other hand, 4
( 18N C12 N18N C12 N18 ) does not transport cations in the
bilayer system. The order of efficacy in the bilayer is 5;6 > 4.
In contrast, in the CHCl3 membrane, the order is 4 > 5;6.
Compound 7 (PhCH2 N18N C12 N18N C12 N18N CH2Ph) shows
transport activity in the bilayer about 40% of gramicidin and
about 35% better than for 6. In the bulk membrane system, its
rate is about 46% that of valinomycin, nearly 6-fold that of
gramicidin, and ∼75% better than for 6. These differences in
relative efficacy in the various membrane systems strongly imply
that different mechanisms operate.
One of the most dramatic examples of this efficacy difference
is observed for 11. In the lipid bilayer, its activity is below
our ability to detect even though structurally it is quite similar
to 6. In the bulk CHCl3 membrane, its activity is second only
to valinomycin itself (66%). A carrier mechanism operating
in both membrane systems is expected to give similar, if not
identical, efficacy orders.
A plot of relative transport rate Vs picrate extraction constant
is shown in Figure 2. The correlation coefficient for a calculated
linear fit of the 10 points shown is 17%. Unlike data obtained
from a variety of previous comparisons, this is essentially a
scatter plot. When data for nonchannel compounds 1-3,
gramicidin, and valinomycin are removed, the best linear fit of
the data is a line through them having a slope of 0.
It might be argued that the difference in media (CHCl3 Vs
phospholipid bilayers) is too great for there to be any meaningful
relationship between the function of a set of compounds in
either. Using the family of compounds noted above [e.g.,
15N CH2COO(CH2)9CH3], we studied transport in the bilayer31
and demonstrated that this did, indeed, correlate to carrier
transport determined by the bulk membrane approach.32
Figure 3. Comparison of Na+ transport effected by 1-10 in a CHCl3
membrane or a phospholipid bilayer.
The data obtained for transport of Na+ by this group of
compounds either in a bulk chloroform membrane or in a
phospholipid bilayer show that there is no correlation between
the two. If the tris(macrocycle) conducts cations by functioning
as a carrier in CHCl3 and a “supercarrier” in the bilayer, there
ought to be a clear rate correlation. It is simply unreasonable
to think that carrier transport is fundamentally different in a
bilayer than it is in a bulk membrane. The lack of correlation
is shown clearly in Figure 3 in which rates determined for a
range of compounds in both the bilayer and chloroform
membranes are compared. In this figure, values are normalized
to 1 Vs gramicidin in the bilayer and Vs valinomycin in CHCl3.
The construction of the concentric tube transport device is
such that the shortest distance between the two aqueous phases
through CHCl3 is ∼108C. It is clearly impossible for any of
these compounds to bridge this distance as gramicidin spans a
∼30 Å biological bilayer membrane by dimerizing. The only
known transport mechanism available to the compounds under
study in this system is the carrier mechanism. The complete
absence of any correlation between Na+ transport efficiencies
assessed in the CHCl3 and phospholipid membranes suggests
that the transport mechanisms differ.
The channel mechanism is clearly impossible in the CHCl3
membrane. The carrier mechanism may be possible for the
tris(macrocycle)s in the bilayer, but the behavior of valinomycin
(31) Xie, Q.; Gokel, G. W.; Hernandez, J. C.; Echegoyen, L. J. Am.
Chem. Soc. 1994, 116, 690-696.
(32) Arnold, K. A.; Hernandez, J. C.; Li, C.; Mallen, J. V.; Nakano, A.;
Schall, O. F.; Trafton, J. E.; Tsesarskaja, M.; White, B. D.; Gokel, G. W.
Supramol. Chem. 1995, 5, 45-60.

5546 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Murillo et al.
results obtained by this method correlate with those previously
obtained. Still, this correlation is neither quantitative nor the
expected one and the meaning of open time as a parameter is
currently unclear.
Another difficulty we encountered in comparing data obtained
by the bilayer clamp and NMR methods involves the use of
gramicidin as an internal standard. In our previous efforts, we
have determined Na+ flux mediated by gramicidin in parallel
with the corresponding determination for a tris(macrocycle). This
was done to minimize experimental variations in these relatively
complex analyses. It was found, however, that the flux rates
determined for gramicidin by using the bilayer clamp method
varied depending upon the concentration of NaCl present in
the medium. Thus, when [NaCl] varied from 100-500 mM,
transport of Na+ by gramicidin varied sufficiently that it was
either faster or slower than the nearly constant rate for 7.
The use of gramicidin also presents a problem in the 23Na-
NMR method of which we were not previously aware. In the
dynamic 23Na-NMR experiment developed by Riddell, line
Figure 4. Bilayer clamp data trace for 7 in a phospholipid membrane.
argues against it. Valinomycin is the most efficient of all the
carrier compounds studied. Compounds 1, 4, and 10 exhibit
about one-half of its activity in CHCl3. Valinomycin’s natural
environment is the mitochondrial membrane, a phospholipid
bilayer. In the model phospholipid bilayer, valinomycin shows
a transport efficacy only 14% that of gramicidin. Compounds
1, 4, and 10 show no detectable activity. Our experimental
system is such that we could easily detect one-half the activity
of valinomycin, i.e. 7% of the activity of gramicidin. That the
best synthetic carriers are the poorest channels does not comport
with the postulate that these compounds may function by a
carrier mechanism. Obviously, this does not prove that the
remaining compounds are channel formers but the implication
is clear.
Attempted Correlation of Transport Rates to Substituent
Constants. We have previously shown that tris(macrocycle)s
7-9, which differ only by the para-substituent present on the
distal benzyl groups, exhibit Na+ transport rates that correlate
with the appropriate Hammett parameters.33 The previously
reported rates were obtained by using the dynamic 23Na-NMR
method described above. Our plan was to determine whether
data obtained by using the bilayer clamp method would give
corresponding results.
The bilayer clamp method uses a patch clamp amplifier and
a bilayer head stage34 to assess single ion currents in a
phospholipid bilayer. Results obtained in this way therefore
differ from those obtained by using the NMR method because
the latter affords an “average” or “macroscopic” result. A
typical data trace obtained for [PhCH2 N18N C12]2 N18N (7)
is shown in Figure 4. The data were obtained at different
applied potentials in the presence of 500 mM NaCl. The
discrete, integral opening and closing behavior observed for this
system is also typical of naturally occurring protein channels.
The conductance traces are strongly suggestive of the channel
mechanism. This is important evidence that bears directly on
the channel mechanism, notwithstanding the original intent only
to demonstrate a substituent effect.
The cation flux for the channel molecules in the bilayer can
be determined by dividing the observed peak height (amperage)
by the applied voltage. In this case, a value of 12.8 picosiemens
(pS) was obtained for 7. It was anticipated, based upon the
results obtained by the 23Na-NMR method, that different,
substituent-dependent fluxes would be observed for 7, 8, and
9. It was also expected that these flux rates would correlate to
the Taft σ0 constant. In fact, very little difference was observed
in the cation fluxes. Instead, preliminary studies suggest that
the open times observed for the channels are different. Thus,
open times appear to be in the order p-CH3O > p-H > p-NO2
(8 > 7 > 9). In principle, the longer a channel is open, the
greater the flux of sodium is expected to be. In this sense, the
broadening of the Na+ is observed relative to [ionophore].
(in)
In our system, the Na+ line width is typically determined at
4-8 ionophore concentrations (0∼20 µM). Changes in the half-
height line width (multiplied by π) are plotted against ionophore
concentration. The rate is determined from the equation for
the line and expressed as its “standard” value at 10 µM.
It is of interest to consider the slope of the log log plot of
the data obtained as described above. According to Hinton,35
this should give the kinetic order of the reaction. For most of
the examples we have studied, the slope of the plot of log10
k
[or log10 (1/τ)] Vs log10 [ionophore] lies in the range 1-1.2.
The value of approximately 1 was not surprising to us since
our tris(macrocycle)s were designed to function as unimolecular
ion transporters. The fact that the slopes we observed were in
a range rather than exactly unity was troubling, but we were
comforted by the fact that the slope reported for the gramicidin
dimer is 1.8 rather than 2.0.35 Unfortunately, in our work,
gramicidin gave a slope of approximately 1 rather than nearly
2 even though the experiment was repeated more than 50 times.
In an effort to understand this discrepancy, we examined the
published data for the gramicidin experiment. The graph
presented in the report shows seven points plus zero while the
data table shows only five points plus zero. We plotted the
five data points and obtained, as published, an excellent line.
The corresponding log log plot gave a slope of ∼1.8, as
previously published. Surprisingly, R2 for the original data set
plotted as 1/τ Vs [ionophore] was 1.00. Unfortunately, none of
our data were of this remarkably high quality.
The similarity of the rates observed by using the bilayer clamp
method is demonstrated for compounds 7-9 in the current-
voltage plot shown as Figure 5. The applied voltages ranges
from (60 to (140 mV, and the calculated conductance was
∼13 pS.
The above results show that 23Na-NMR method has proved
to be experimentally cumbersome and that the data that are
obtained do not correlate exactly with those obtained by using
the bilayer clamp method. Our inability to corroborate the NMR
method Hammett study with bilayer clamp data makes suspect
the previous results. The fact that the bilayer clamp data appear
typical in most ways of data obtained from Na+-conducting
protein channels strongly suggests that the tris(macrocycle)s
function as channels rather than as carriers or by some other,
as yet undefined, mechanism.
Conformation of 6 in the Phospholipid Bilayer. Previous
studies have confirmed that the central macrocycle does not
(35) Buster, D. C.; Hinton, J. F.; Millett, F. S.; Shungu, D. C. Biophys.
J. 1988, 53, 145-152.
(33) Murillo, O.; Suzuki, I.; Abel, E.; Gokel, G. W. J. Am. Chem. Soc.
1996, 118, 7628-7629.
(34) Warner Instrument PC-505 amplifier equipped with a HB-202 50
gΩ head stage.

Synthetic Transmembrane Channels
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5547
tris(macrocyclic) systems span the bilayer and function as model
cation channels.
Experimental Section
1H-NMR spectra were recorded at 300, 500, or 600 MHz in CDCl3
solvent and are reported in ppm (δ) downfield from internal (CH3)4Si
unless otherwise specified. 13C-NMR were recorded at proportional
frequencies as noted above. Infrared spectra were recorded on a Perkin-
Elmer 1710 Fourier transform infrared spectrophotometer and were
calibrated against the 1601 cm-1 band of polystyrene. Melting points
were determined on a Thomas Hoover apparatus in open capillaries
and are uncorrected. Thin layer chromatographic (TLC) analyses were
performed on aluminum oxide 60 F-254 neutral (Type E) with a 0.2-
mm layer thickness or on silica gel 60 F-254 with a 0.2-mm layer
thickness. Preparative chromatography columns were packed with
activated aluminum oxide (MCB 80-325 mesh, chromatographic grade,
AX 611) or with Kieselgel 60 (70-230 mesh). Rotating disk
chromatography was performed on a Harrison Research Model 7924
Chromatotron with 2-mm thick circular plates prepared from Kieselgel
60 PF-254.
Figure 5. Plot of voltage Vs amperage for compounds 7-9.
All reactions were conducted under dry N2 unless otherwise stated.
All reagents were the best (non-LC) grade commercially available and
were distilled, recrystallized, or used without further purification, as
appropriate. Molecular distillation temperatures refer to the oven
temperature of a Kugelrohr apparatus. Combustion analyses were
performed by Atlantic Microlab, Inc., Atlanta, GA, and are reported
as percents.
Reagents. Phophatidylcholine (CHCl3 solution), phosphatidylglyc-
erol (CHCl3:CH3OH solution), Na2HPO4, KH2PO4, NaCl, gramicidin
D (85% gramicidin A, 15% gramicidin B and C), and sodium
tripolyphosphate (98%) were purchased from Sigma, St. Louis, MO,
and used without further purification. Diethyl ether (anhydrous) was
purchased from Mallinckrodt and distilled over sodium/benzophenone
prior to use. The water used in the buffer solution was distilled and
deionized. 2,2,2-Trifluoroethanol (TFE, NMR grade), DyCl3‚H2O
(99.99%), D2O, KCl, CHCl3 (HPLC grade), picric acid, NaOH
volumetric solution (0.0975 M), and 1,4-dinitrobenzene (98%) were
purchased from Aldrich and used without further purification. Chlo-
roform-d (99.8%) was purchased from Isotec, Ohio. Polycarbonate
membranes (1.0 µm, 2.5 cm diameter) and membrane holders were
obtained from Poretics Corp., Livermore, CA. The sonication was done
in a water sonication bath, Branson, model 1200, or in a Branson,
Bransonic 12. The mean diameters of the vesicles were determined in
a Coulter, submicron particle analyzer, model N4 ND.
Figure 6. Postulated channel conformation for 7 in a phospholipid
bilayer.
require to be parallel to the distal rings. The distal macrocycles
are expected to be head groups in these structures. This is
anticipated from studies of crown ethers as head groups for
stable vesicles.36 Upon the basis of the evidence reported here
and accumulated thus far, we suggest the conformation shown
in Figure 6 as the most likely arrangement of 7 in a lipid bilayer.
Presumably, water infiltrates the channel and assists in transport
of the cation through the transmembrane “groove.”
Conclusions
The present study addressed three basic issues concerning
the tris(macrocyclic) Na+-transporting system. The possibility
that differences in Na+ transport rate may correlate with
differential membrane solubility of the ionophores was ad-
dressed. Calculations, verified where possible, of the octanol-
water partition coefficients show that all of the channel formers
favor a hydrophobic environment by >1010. Second, Na+
transport rates were determined in a bulk CHCl3 membrane.
The channel mechanism is not possible in this case due to the
large distance separating the source and receiving aqueous
phases. A plot of transport rates determined in CHCl3 and in
a bilayer showed no correlation. It was interesting to note that
several of the tris(macrocyclic) ionophores caused rapid hydra-
tion of the CHCl3 phase although water was lost during
equilibration over several hours. Finally, a postulated channel-
like conformation for the tris(macrocycle)s should show “chan-
nel-like” conductivity when studied by using a bilayer clamp
method. Such data traces were, in fact, obtained and show
remarkable similarity to those obtained for natural, Na+-
conducting protein channels. We thus conclude that the
Compounds Studied. Gramicidin and valinomycin were purchased
from Sigma and Aldrich and used as received.
N,N′-Dibenzyl-4,13-diaza-18-crown-6 (1) was prepared as previ-
ously reported.37
N,N′-Didodecyl-4,13-diaza-18-crown-6 (2) was prepared as previ-
ously reported.38
N,N′-Bis(11-carboxyundecyl)-4,13-diaza-18-crown-6 (3). Benzyl
12-Bromodedecanoate (3A). To a mixture of 12-bromododecanoic
acid (4.75 g, 17.0 mmol), DCC (3.86 g, 18.7 mmol), and DMAP (0.23
g, 1.87 mmol) in ether (50 mL) was added benzyl alcohol (2.02 g,
18.7 mmol) at room temperature. After 2 h, the precipitate that formed
was removed by filtration and the filtrate was washed successively with
5% NaHCO3 (2 × 15 mL), water (2 × 15 mL), 10% citric acid (2 ×
15 mL), water (15 mL), and brine (15 mL). The ether layer was dried
(MgSO4), concentrated, and chromatographed (SiO2 column) to give
3A (4.03 g, 64%) as a colorless oil, bp 197-200 °C (0.5 Torr). 1H-
NMR: 1.20-1.35 (12H, m, alkyl), 1.35-1.50 (2H, m, BrCH2-
CH2CH2CH2), 1.57-1.70 (2H, m, COCH2CH2CH2CH2), 1.851 (2H,
quintet, J ) 7.3 Hz, BrCH2CH2CH2CH2), 2.354 (2H, t, J ) 7.5 Hz,
COCH2CH2CH2), 3.408 (2H, t, J ) 6.9 Hz, BrCH2CH2), 5.116 (2H, s,
ArCH2O), 7.32-7.38 (5H, m, aromatics). IR (neat film): 3066, 3034,
2927, 2854, 1737, 1498, 1456, 1383, 1353, 1256, 1214, 1164, 1120,
1003, 908, 736, 698 cm-1. Anal. Calcd for C19H29BrO2: C, 61.79;
H, 7.91; Br, 21.63%. Found: C, 61.93; H, 7.91; Br, 21.48%.
(36) Mun˜oz, S.; Malle´n, J.; Nakano, A.; Chen, Z.; Gay, I.; Echegoyen,
L.; Gokel, G. W. J. Am. Chem. Soc. 1993, 115, 1705-1711.
(37) Gatto, V. J.; Miller, S. R.; Gokel, G. W. Org. Synth. 1989, 68, 227.
(38) Gatto, V. J.; Arnold, K. A.; Viscariello, A. M.; Miller, S. R.; Morgan,
C. R.; Gokel, G. W. J. Org. Chem. 1986, 51, 5373.
N,N′-Bis(11-benzyloxycarbonylundecyl)-4,13-diaza-18-crown-6 (3B).
A mixture of 4,13-diaza-18-crown-6 (1.20 g, 4.57 mmol), 3A (3.38 g,

5548 J. Am. Chem. Soc., Vol. 119, No. 24, 1997
Murillo et al.
9.15 mmol), Na2CO3 (9.69 g, 91.4 mmol), and KI (30 mg, 0.18 mmol)
in n-PrCN (50 mL) was heated at reflux for 24 h. After the mixture
was cooled, the insoluble materials were removed by filtration and the
resulting solution was concentrated to an oil which was dissolved in
CH2Cl2 (40 mL), washed with water (4 × 15 mL), dried (MgSO4),
and concentrated. Toluene was added and evaporated to assure
complete removal of n-PrCN. The crude, oily product was purified
by chromatography (alumina, 3% i-PrOH-CH2Cl2) and afforded 3B
(3.04 g, 79%) as colorless oil, mp <25 °C, bp0.5 > 300 °C. 1H-NMR:
1.20-1.36 (28H, m, alkyl), 1.36-1.48 (4H, br, NCH2CH2CH2), 1.638
(4H, quintet, J ) 7.4 Hz, COCH2CH2CH2), 2.352 (4H, t, J ) 7.6 Hz,
COCH2CH2), 2.475 (4H, t, J ) 7.8 Hz, NCH2CH2CH2), 2.771 (8H, t,
J ) 6.0 Hz, NCH2CH2O), 3.56-3.64 (16H, m, CH2OCH2), 5.114 (4H,
s, ArCH2O), 7.31-7.37 (10H, m, aromatics). IR (neat film): 3065,
3034, 2927, 2854, 1737, 1456, 1352, 1255, 1163, 1127, 994, 751, 698
cm-1. Anal. Calcd for C50H82N2O8: C, 71.56; H, 9.85; N, 3.34%.
Found: C, 71.60; H, 9.88; N, 3.25%.
1297, 1265, 1240, 1172, 1125, 1072, 1039, 966, 942, 881, 844, 831,
817, 803, 718, 605, 573 cm-1
. Anal. Calcd for C76H138N6O14: C,
67.12; H, 10.23; N, 6.18. Found: C, 67.02; H, 10.21; N, 6.12%.
4,13-Bis[12-{13-(4-Nitrobenzyl)-4,13-diaza-18-crown-6-4-yl}do-
decyl]-4,13-diaza-18-crown-6 (9). N-(4-Nitrobenzyl)-4,13-diaza-18-
crown-6 (9A). A mixture of 4,13-diaza-18-crown-6 (3.54 g, 13.5
mmol), 4-nitrobenzyl bromide (2.62 g 12.2 mmol), Na2CO3 (14.3 g,
135 mmol), and KI (48 mg, 0.3 mmol) in n-PrCN (300 mL) was heated
at reflux for 4.5 h. After being cooled, the mixture was filtered and
the filtrate was concentrated to a yellow oil. Toluene was added and
then evaporated (2×) to assure removal of n-PrCN. The resulting oil
was chromatographed over alumina (3% MeOH-CH2Cl2). Compound
9A eluted second; solvent was removed, and the yellow oil thus obtained
(2.36 g, 44% based on 4,13-diaza-18-crown-6) solidified on standing.
1H-NMR: 2.802 (4H, t, J ) 4.7 Hz, NCH2CH2O), 3.168 (4H, t, J )
3.9 Hz, NHCH2CH2O), 3.52-3.66 (13H, m, NH, CH2OCH2CH2O),
3.756 (2H, s, PhCH2N), 3.875 (4H, t, J ) 4.2 Hz, HNCH2CH2O), 7.584
(2H, d, J ) 8.1 Hz), 8.185 (2H, d, J ) 7.8 Hz). IR (KBr disk) 3467,
2884, 2499, 1625, 1605, 1515, 1459, 1346, 1278, 1244, 1106, 1084,
974, 937, 855, 829, 802, 744, 737, 708, 648, 523, 442 cm-1. Compound
9A was used in the next step without further purification.
4,13-Bis[12-{13-(4-nitrobenzyl)-4,13-diaza-18-crown-6-4-yl}do-
decyl]-4,13-diaza-18-crown-6 (9). The mixture of 9A (1.077 g, 2.71
mmol), 4,13-bis(12-bromododecyl)-4,13-diaza-18-crown-6 (1.00 g, 1.32
mmol), Na2CO3 (2.875 g, 27.1 mmol), and KI (20 mg, 0.12 mmol) in
n-PrCN (20 mL) was heated under reflux for 7.5 h. After cooling, the
mixture was filtered and the filtrate was concentrated to a yellow oil.
Toluene was added and evaporated (2×) to assure removal of n-PrCN.
The resulting yellow oil was chromatographed (alumina, 10% i-PrOH:
hexanes, then i-PrOH-hexanes-CH2Cl2 (2:40:10)). Evaporation of
the solvent gave a yellow oil which solidified on standing. Crystal-
lization (95% EtOH) gave 9 (0.72 g, 39% based on 4,13-bis(12-
bromododecyl)-4,13-diaza-18-crown-6) as a yellow solid, mp 61.5-
62 °C. 1H-NMR: 1.22-1.28 (32H, pseudo-s, alkyl), 1.42-1.56 (8H,
br, NCH2CH2CH2), 2.50-2.60 (8H, br, NCH2CH2CH2), 3.58-3.70
(48H, m, CH2OCH2CH2OCH2), 3.794 (4H, s, PhCH2O), 7.553 (4H, d,
J ) 8.7 Hz), 8.162 (4H, d, J ) 8.7 Hz). IR (KBr disk) 2918, 2870,
1607, 1518, 1473, 1356, 1296, 1239, 1125, 1072, 1045, 965, 843, 740,
718, 694, 605, 422 cm-1. Anal. Calcd for C74H132N8O16: C, 63.95;
H, 9.57; N, 8.06%. Found: C, 63.85; H, 9.56; N, 8.07%.
N,N′-Bis(11-carboxyundecyl)-4,13-diaza-18-crown-6 (3). A pres-
sure bottle was charged with 3B (2.00 g, 2.38 mmol) in i-PrOH (70
mL). To this solution was carefully added water (20 mL) to keep the
solution clear. The vessel was purged with N2, 10% Pd on carbon
was added, and the resulting suspension was shaken on a Parr
hydrogenator under H2 (2 atm) at room temperature. The catalyst was
removed (pad of Celite), and the mixture was evaporated to dryness,
leaving 3 (1.54 g, 97%) as a colorless solid, mp 77-79 °C. 1H-NMR:
1.20-1.38 (28H, m, alkyl), 1.48-1.66 (8H, br, NCH2CH2CH2,
COCH2CH2CH2), 2.269 (4H, t, J ) 6.8 Hz, COCH2CH2), 2.746 (4H,
t, J ) 8.0 Hz, NCH2CH2CH2), 3.066 (8H, t, J ) 5.3 Hz, NCH2CH2O),
3.592 (8H, s, NCH2CH2O), 3.726 (8H, t, J ) 5.4 Hz, OCH2CH2O). IR
(KBr disk): 3837, 2914, 2850, 1698, 1472, 1358, 1343, 1266, 1127,
1062, 1016, 974, 866, 832, 805, 720, 600, 562, 513 cm-1. Anal. Calcd
for C36H70N2O8: C, 65.62; H, 10.71; N, 4.25%. Found: C, 65.36; H,
10.66; N, 4.25%.
N,N′-Bis[(N-4,13-diaza-18-crown-6)dodecyl]-4,13-diaza-18-crown-
6 (4) was prepared as previously reported.
Compounds 5-7 were prepared as previously described.17
4,13-Bis[12-{13-(4-methoxybenzyl)-4,13-diaza-18-crown-6-4-yl}-
dodecyl]-4,13-diaza-18-crown-6 (8). N-(4-Methoxybenzyl)-4,13-
diaza-18-crown-6 (8A). A mixture of 4,13-diaza-18-crown-6 (3.54
g, 13.5 mmol), 4-methoxybenzyl chloride (1.90 g 12.2 mmol), Na2CO3
(14.3 g, 135 mmol), and KI (48 mg, 0.3 mmol) in n-PrCN (300 mL)
was heated at reflux for 2 h. After cooling, the mixture was filtered
and the filtrate was concentrated to leave a slightly yellow oil. Toluene
was added and then evaporated (3×) to remove any residual n-PrCN.
The resulting oil was chromatographed over alumina and eluted with
2% MeOH-CH2Cl2. After removal of the solvent, 8A was obtained
(2.38 g, 46% based on 4,13-diaza-18-crown-6) as a slightly yellow oil
which gradually gelled. 1H-NMR: 2.699 (4H, t, J ) 4.8 Hz,
NCH2CH2O), 3.123 (4H, t, J ) 4.5 Hz, NHCH2CH2O), 3.50-3.96
(21H, m, NH, CH2OCH2CH2OCH2, PhCH2N, CH3OPh), 6.910 (2H, d,
J ) 8.4 Hz), 7.236 (2H, d, J ) 8.4 Hz). IR (KBr): 3467, 3241, 3056,
3025, 2876, 1612, 1582, 1513, 1474, 1458, 1367, 1352, 1301, 1268,
1242, 1173, 1113, 1033, 954, 836, 793, 756, 574, 524 cm-1. Compound
8A was used in the next step without additional purification.
NMR Studies of Transport in Bilayers. The procedure is as
reported in ref 17.
log Poct Determination for N,N′-Dibenzyldiaza-18-crown-6. 1-Oc-
tanol (200 mL, ACS spectroscopic grade) was saturated with deionized
water. To saturate the octanol, 200 mL of alcohol were placed in a
1.0-L bottle with 400 mL of deionized water. The two-phase system
was gently shaken for 3 min and then placed in a thermostatic water
bath at 21.0 °C overnight. Doubly deionized water was similarly
saturated with 1-octanol.
A solution of N,N′-dibenzyldiaza-18-crown-6 (10-2 M, 3.00 mL) in
water-saturated 1-octanol was placed in a separatory funnel, and 100
mL of the water-saturated octanol were added. The funnel was gently
shaken for 3 min. The aqueous layer was drained, and the octanol
layer was removed with the aid of a Pasteur pipette. A 0.2-mL sample
of this octanol layer was placed in a spectroscopic cuvette and diluted
with 1-octanol (1.8 mL). The partition coefficient was determined by
comparison of the absorbance of this solution with that of a freshly
prepared 10-3 M solution of N,N′-dibenzyldiaza-18-crown-6 in 1-oc-
tanol.
Bulk Membrane Transport Experiments. UV-vis spectra were
recorded on a Beckman DU-8 spectrophotometer. To obtain the
transport rate of sodium picrate, the previously described procedure
has been followed with the following modifications: 20-mL Beakers
were used instead of the 18-mm vials. The internal diameter of the
concentric tube was 12 mm. Sodium picrate transport was followed
by measurement of the %T in the receiving phase at a wavelength of
354 nm. A layer of 6.0 mL of CHCl3 in a 20-mL beaker was stirred
(synchronous stirrer, power 1) using a 7-mm Teflon-coated magnetic
bar. The source and the receiving phases were separated by a glass
tube (10 mm i.d.) suspended 5 mm above the bottom of the beaker,
but below the surface of the CHCl3 layer. The source phase was placed
inside the concentric tube. The receiving phase was in the external
4,13-Bis[12-{13-(4-methoxybenzyl)-4,13-diaza-18-crown-6-4-yl}-
dodecyl]-4,13-diaza-18-crown-6 (8). A mixture of 8A (1.04 g, 2.71
mmol), 4,13-bis(12-bromododecyl)-4,13-diaza-18-crown-6 (1.00 g, 1.32
mmol), Na2CO3 (2.875 g, 27.1 mmol), and KI (20 mg, 0.12 mmol) in
n-PrCN (20 mL) was heated under reflux for 7.5 h. After being cooled,
the mixture was filtered and the filtrate was concentrated to a yellow
oil. Toluene was added and then evaporated (2×) to assure removal
of n-PrCN. The resulting oil was chromatographed over alumina (i-
PrOH-hexanes-CH2Cl2 (1:20:5)). Evaporation of the solvent gave the
crude product which solidified on standing. Repeated crystallization
(95% EtOH) gave 8 (0.158 g, 9% based on 4,13-bis(12-bromododecyl)-
4,13-diaza-18-crown-6) as a slightly yellow powder, mp 68.5-69 °C.
1H-NMR: 1.22-1.28 (32H, pseudo-s, alkyl), 1.42-1.54 (8H, br,
NCH2CH2CH2), 2.482 (8H, t, J ) 6.0 Hz, NCH2CH2CH2), 2.74-2.82
(24H, m, NCH2CH2O) 3.58-3.64 (52H, m, CH2OCH2CH2OCH2,
PhCH2O), 3.795 (6H, s, CH3OPh), 6.834 (4H, d, J ) 8.7 Hz), 7.234
(4H, d, J ) 8.7 Hz). IR (KBr): 2918, 2871, 1619, 1518, 1536, 1333,

Synthetic Transmembrane Channels
J. Am. Chem. Soc., Vol. 119, No. 24, 1997 5549
compartment. 1.0 mL of the sodium picrate solution (1.0 mM in 0.100
M NaOH) (source phase) and 6.0 mL of distilled and deionized water
(receiving phase) were placed on top of the chloroform layer. The
concentration of all the compounds studied was 1.0 mM. The transport
rate is reported as moles transported/24 h. Ion transport was determined
by following the change in absorbance (%T) of the receiving phase at
a wavelength of 354 nm. The concentration of sodium picrate was
determined from the calibration curve.29
Procedure for Bilayer Clamp Studies. Phospholipid membranes
were prepared by the painted bilayer method.39 A lipid solution was
prepared by dissolving 30 mg of asolectin40 in 1.0 mL of decane. This
solution was used to pretreat the pinhole in the Teflon cup of the bilayer
setup. Normally, 1-5 µL was placed on top of the pinhole and allowed
to dry. The cup and cup-holder setup was then assembled. A NaCl
solution (0.5 M in H2O) was added to both the cup and cupholder,
taking care that the water level was the same on both sides. A glass
rod was dipped into the lipid/decane solution, and the tip of the rod
was rubbed across the pinhole in the Teflon cup. Bilayer formation
was detected by a sudden drop in conductance. A 10-mV square wave
train pulse was applied to the membrane to monitor the membrane
capacitance. All of the membrane systems studied exhibited a
capacitance of more than 100 pF and a conductance of less than 3.0
pA when a +100-mV holding voltage was applied.
The aqueous solution in the cup holder was carefully stirred, and
10-50 µL of a 10-6 M channel solution (in TFE) was added. The
solution was continuously stirred for 1-10 min. A holding potential
was applied and channel responses recorded. Single-channel signals
were detected and recorded using a Digidata 1200A A/D converter and
the data acquisition software Axoscope. The holding voltage was
changed manually to determine the channel’s response at different
voltages.
Single-channel current amplitudes and standard deviation values were
also determined using Axoscope software.
Acknowledgment. We thank the NIH for a grant (GM
36262) that supported this work.
(39) Mueller, P.; Rudin, D. O.; Tien, H. T.; Wescot, W. D. Nature
(London) 1962, 194, 979.
(40) Asolectin is a mixture of phospholipids obtained from soya beans.
Choline content is 20-30%.
JA962694A