Water-templated transmembrane nanopores from shape-persistent oligocholate macrocycles

Article abstract of DOI:10.1021/ja109036z

Hydrophobic interactions normally are not considered a major driving force for self-assembling in a hydrophobic environment. When macrocyclic oligocholates were placed within lipid membranes, however, the macrocycles pulled water molecules from the aqueous phase into their hydrophilic internal cavities. These water molecules had strong tendencies to aggregate in a hydrophobic environment and templated the macrocycles to self-assemble into transmembrane nanopores. This counterintuitive hydrophobic effect resulted in some highly unusual transport behavior. Cholesterol normally increases the hydrophobicity of lipid membranes and makes them less permeable to hydrophilic molecules. The permeability of glucose across the oligocholate-containing membranes, however, increased significantly upon the inclusion of cholesterol. Large hydrophilic molecules tend to have difficulty traversing a hydrophobic barrier. The cyclic cholate tetramer, however, was more effective at permeating maltotriose than glucose.

Full text of DOI:10.1021/ja109036z

Published on Web 12/10/2010
Water-Templated Transmembrane Nanopores from
Shape-Persistent Oligocholate Macrocycles
Hongkwan Cho, Lakmini Widanapathirana, and Yan Zhao*
Department of Chemistry, Iowa State UniVersity, Ames, Iowa 50011-3111, United States
Received October 7, 2010; E-mail: zhaoy@iastate.edu
Abstract: Hydrophobic interactions normally are not considered a major driving force for self-assembling
in a hydrophobic environment. When macrocyclic oligocholates were placed within lipid membranes,
however, the macrocycles pulled water molecules from the aqueous phase into their hydrophilic internal
cavities. These water molecules had strong tendencies to aggregate in a hydrophobic environment and
templated the macrocycles to self-assemble into transmembrane nanopores. This counterintuitive
hydrophobic effect resulted in some highly unusual transport behavior. Cholesterol normally increases the
hydrophobicity of lipid membranes and makes them less permeable to hydrophilic molecules. The
permeability of glucose across the oligocholate-containing membranes, however, increased significantly
upon the inclusion of cholesterol. Large hydrophilic molecules tend to have difficulty traversing a hydrophobic
barrier. The cyclic cholate tetramer, however, was more effective at permeating maltotriose than glucose.
Introduction
More recently, Satake and Kobuke prepared nanosized pores
based on porphyrin supramolecules.6 Gong et al. described pores
Channels and pores are used in biology to permeate ions and
molecules across membranes. In addition to their important roles
in signaling, metabolism, and bacterial or viral infection,
channels and pores enable design of novel sensors for both small
and large molecules.1 Pore-forming proteins, for example, have
shown great promises in the single-molecule detection of RNAs
and DNAs.2
Although synthetic pores have the advantage of being less
expensive and less prone to denaturization than their protein
counterparts, development of nanometer-sized synthetic pores
has been a difficult challenge.3 Ghadiri et al. prepared cyclic
peptides that self-assembled into pores large enough for glucose
and glutamic acid to pass through.4 Matile and co-workers, in
a series of seminal work, reported nanometer-sized ꢀ-barrel
pores through self-assembly of oligo(phenylene) derivatives5
and demonstrated their applications in sensing5b and catalysis.5c
ca. 0.8 nm in diameter through the π-π interactions of aromatic
heterocycles.7 In addition, Fyles8 and Davis9 used amine-Pd(II)
and guanosine quartets, respectively, to construct highly con-
ducting channels consistent with nanometered pore sizes.
A big challenge in creating nanometer-sized pores within the
lipid bilayers is to keep the pore from collapsing. For this reason,
although chemists have made tremendous progress in the design
and synthesis of artificial ion channels,10 the building blocks
involved (e.g., crown ethers and open chain compounds)
typically are not amenable to nanopore formation. Despite the
advancement made in synthetic nanopores, only limited pore-
forming mechanisms exist currently. The majority of synthetic
nanopores reported so far relied on either hydrogen-bonding4,5,9
or metal-ligand coordination6,8 for stability.
Herein, we report synthetic nanopores driven by hydrophobic
interactionssa very different mechanism of pore formation from
common biological and synthetic examples. The novelty of the
approach lies in the counterintuitive design. Normally, if the
environment (i.e., lipid bilayers) is hydrophobic, hydrophobic
interactions are not expected to contribute significantly to a
supramolecular synthesis. The self-assembled pores displayed
(1) Kasianowicz, J. J.; Robertson, J. W. F.; Chan, E. R.; Reiner, J. E.;
Stanford, V. M. Annu. ReV. Anal. Chem. 2008, 1, 737–766.
(2) (a) Kasianowicz, J. J.; Brandin, E.; Branton, D.; Deamer, D. W. Proc.
Natl. Acad. Sci. U.S.A. 1996, 93, 13770–13773. (b) Akeson, M.;
Branton, D.; Kasianowicz, J. J.; Brandin, E.; Deamer, D. W. Biophys.
J. 1999, 77, 3227–3233. (c) Meller, A.; Nivon, L.; Brandin, E.;
Golovchenko, J.; Branton, D. Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
1079–1084. (d) Vercoutere, W.; Winters-Hilt, S.; Olsen, H.; Deamer,
D.; Haussler, D.; Akeson, M. Nat. Biotechnol. 2001, 19, 248–252. (e)
Howorka, S.; Cheley, S.; Bayley, H. Nat. Biotechnol. 2001, 19, 636–
639. (f) Clarke, J.; Wu, H. C.; Jayasinghe, L.; Patel, A.; Reid, S.;
Bayley, H. Nat. Biotechnol. 2009, 4, 265–270.
(6) Satake, A.; Yamamura, M.; Oda, M.; Kobuke, Y. J. Am. Chem. Soc.
2008, 130, 6314–6315.
(7) Helsel, A. J.; Brown, A. L.; Yamato, K.; Feng, W.; Yuan, L. H.;
Clements, A. J.; Harding, S. V.; Szabo, G.; Shao, Z. F.; Gong, B.
J. Am. Chem. Soc. 2008, 130, 15784–15785.
(3) (a) Matile, S.; Som, A.; Sorde´, N. Tetrahedron 2004, 60, 6405–6435.
(b) Sisson, A. L.; Shah, M. R.; Bhosale, S.; Matile, S. Chem. Soc.
ReV. 2006, 35, 1269–1286.
(8) Fyles, T. M.; Tong, C. C. New J. Chem. 2007, 31, 655–661.
(9) Ma, L.; Melegari, M.; Colombini, M.; Davis, J. T. J. Am. Chem. Soc.
2008, 130, 2938–2939.
(4) (a) Granja, J. R.; Ghadiri, M. R. J. Am. Chem. Soc. 1994, 116, 10785–
10786. (b) Sa´nchez-Quesada, J.; Kim, H. S.; Ghadiri, M. R. Angew.
Chem., Int. Ed. 2001, 40, 2503–2506.
(10) (a) Gokel, G. W.; Mukhopadhyay, A. Chem. Soc. ReV. 2001, 30, 274–
286. (b) Fyles, T. M. Chem. Soc. ReV. 2007, 36, 335–347. (c) Koert,
U.; Al-Momani, L.; Pfeifer, J. R. Synthesis-Stuttgart 2004, 1129–1146.
(d) Gokel, G. W.; Murillo, O. Acc. Chem. Res. 1996, 29, 425–432.
(e) Jung, M.; Kim, H.; Baek, K.; Kim, K. Angew. Chem., Int. Ed.
2008, 47, 5755–5757. (f) Li, X.; Shen, B.; Yao, X. Q.; Yang, D. J. Am.
Chem. Soc. 2009, 131, 13676–13680.
(5) (a) Sakai, N.; Mareda, J.; Matile, S. Acc. Chem. Res. 2005, 38, 79–
87. (b) Das, G.; Talukdar, P.; Matile, S. Science 2002, 298, 1600–
1602. (c) Sakai, N.; Sorde´, N.; Matile, S. J. Am. Chem. Soc. 2003,
125, 7776–7777.
9
10.1021/ja109036z  2011 American Chemical Society
J. AM. CHEM. SOC. 2011, 133, 141–147 141

A R T I C L E S
Cho et al.
Scheme 1. Schematic Representation of the Solvophobically Driven Folding of a Linear Oligocholate and Aggregation of 1
highly unusual behavior as a result of the counterintuitive pore-
formation. Cholesterol is known to increase the hydrophobic
thickness11 of lipid bilayers and decrease their fluidity.12
Yet, the enhanced hydrophobicity caused by cholesterol facilitated
the pore formation of the oligocholate macrocycles and increased
the permeability of glucose across the membranes. Larger hydro-
philic molecules normally have difficulty moving across a hydro-
phobic barrier. The cyclic cholate tetramer, however, was more
effective at permeating maltotriose than glucose.
philic inner pore. The polar solvent phase-separates from the
bulk into the hydrophilic pore and efficiently solvate the
introverted polar groups of the oligocholate (Scheme 1, left).
Since the folded helix has three monomer units per turn,17a
cyclic tricholate 1 essentially represents the cross section of the
folded helix. According to the CPK model, the molecule has a
triangular hydrophilic cavity about 1 nm on the side (the N-N
distance is ∼1.3 nm). Its exterior is completely hydrophobic
and fully compatible with lipid membranes. Its rigidity, resulting
from both the triangular geometry and the fused steroid
backbone, is expected to prevent the inner cavity from collaps-
ing. Note that, although other strategies (e.g., internal charge
repulsion)18 have been used with success, rigidity of the building
blocks is a key factor in keeping synthetic nanopores from
collapsing.3-9
Results and Discussion
Molecular Design. With a cholesterol-like backbone and facial
amphiphilicity, cholic acid derivatives have been used in
membrane-related applications including as ion channels9,13 and
molecule14 or anion transporters.15 In an effort to prepare
conformationally controllable foldamers,16 we synthesized linear
oligomers of facially amphiphilic cholates.17 In nonpolar
solvents (e.g., CCl4 or hexane/ethyl acetate) containing a small
percentage of a polar solvent (e.g., DMSO or MeOH), the
oligocholate folds into a helix with a nanometer-sized hydro-
(11) Nezil, F. A.; Bloom, M. Biophys. J. 1992, 61, 1176–1183.
(12) Holthuis, J. C. M.; van Meer, G.; Huitema, K. Mol. Membr. Biol.
2003, 20, 231–241.
(13) (a) Bandyopadhyay, P.; Janout, V.; Zhang, L. H.; Sawko, J. A.; Regen,
S. L. J. Am. Chem. Soc. 2000, 122, 12888–12889. (b) Goto, C.;
Yamamura, M.; Satake, A.; Kobuke, Y. J. Am. Chem. Soc. 2001, 123,
12152–12159. (c) Maulucci, N.; De Riccardis, F.; Botta, C. B.;
Casapullo, A.; Cressina, E.; Fregonese, M.; Tecilla, P.; Izzo, I. Chem.
Commun. 2005, 1354–1356. (d) Wilson, C. P.; Webb, S. J. Chem.
Commun. 2008, 4007–4009.
(14) (a) Janout, V.; Di Giorgio, C.; Regen, S. L. J. Am. Chem. Soc. 2000,
122, 2671–2672. (b) Janout, V.; Staina, I. V.; Bandyopadhyay, P.;
Regen, S. L. J. Am. Chem. Soc. 2001, 123, 9926–9927. (c) Janout,
V.; Jing, B. W.; Staina, I. V.; Regen, S. L. J. Am. Chem. Soc. 2003,
125, 4436–4437. (d) Janout, V.; Regen, S. L. J. Am. Chem. Soc. 2005,
127, 22–23.
We hypothesized that the same solvophobic driving force in
the folding of the linear oligocholates17 would prompt 1 to stack
in the z-direction (Scheme 1, right). In a largely nonpolar solvent
mixture, the polar solvent molecules should phase-separate into
the middle of the macrocycle and solvate the inward-facing polar
groups. Aggregation also allows these polar solvents to move
up and down the polar channelsan entropically favorable
process.
(15) (a) Koulov, A. V.; Lambert, T. N.; Shukla, R.; Jain, M.; Boon, J. M.;
Smith, B. D.; Li, H. Y.; Sheppard, D. N.; Joos, J. B.; Clare, J. P.;
Davis, A. P. Angew. Chem., Int. Ed. 2003, 42, 4931–4933. (b)
Whitmarsh, S. D.; Redmond, A. P.; Sgarlata, V.; Davis, A. P. Chem.
Commun. 2008, 3669–3671.
(16) For several representative reviews, see: (a) Gellman, S. H. Acc. Chem.
Res. 1998, 31, 173–180. (b) Hill, D. J.; Mio, M. J.; Prince, R. B.;
Hughes, T. S.; Moore, J. S. Chem. ReV. 2001, 101, 3893–4012. (c)
Hecht, S.; Huc, I. Foldamers: structure, properties, and applications;
Wiley-VCH: Weinheim, 2007. (d) Cubberley, M. S.; Iverson, B. L.
Curr. Opin. Chem. Biol. 2001, 5, 650–653. (e) Kirshenbaum, K.;
Zuckermann, R. N.; Dill, K. A. Curr. Opin. Struct. Biol. 1999, 9, 530–
535. (f) Goodman, C. M.; Choi, S.; Shandler, S.; DeGrado, W. F.
Nat. Chem. Biol. 2007, 3, 252–262. (g) Bautista, A. D.; Craig, C. J.;
Harker, E. A.; Schepartz, A. Curr. Opin. Chem. Biol. 2007, 11, 685–
692.
In addition to 1, cyclic tetramer 2 and linear trimer 3 were
prepared as control compounds. Compound 4 has a pyrene group
on the side chain, allowing us to use fluorescence to probe the
self-assembly. Linear oligocholates such as 3 were synthesized
through standard amide-coupling reactions.17a Cyclization of the
corresponding amino/carboxyl-terminated oligocholates yielded
compounds 1 and 2 (see the Supporting Information for details).
For compound 4, we took advantage of the azide group at the
end of the oligocholates and used the click reaction19 for the
(17) (a) Zhao, Y.; Zhong, Z. J. Am. Chem. Soc. 2005, 127, 17894–17901.
(b) Zhao, Y.; Zhong, Z.; Ryu, E. H. J. Am. Chem. Soc. 2007, 129,
218–225. (c) Cho, H.; Zhao, Y. J. Am. Chem. Soc. 2010, 132, 9890–
9899. (d) Zhao, Y. J. Org. Chem. 2009, 74, 834–843. (e) Pan, X.;
Zhao, Y. Org. Lett. 2009, 11, 69–72.
(18) Som, A.; Matile, S. Chem. BiodiV. 2005, 2, 717–729.
9
142 J. AM. CHEM. SOC. VOL. 133, NO. 1, 2011

Water-Templated Transmembrane Nanopores
A R T I C L E S
Scheme 2. Synthetic Route for Pyrene-Labeled Macrocycle 4
cyclization (Scheme 2). The pyrene group was introduced at
the side chain of the L-ornithine inserted in between two
cholates.
macrocycle, but another 2-fold increase in the transporter
concentration caused a dramatic increase in leakagesover 90%
of glucose leaked out after 60 min.
Glucose Leakage from POPC/POPG LUVs. An ideal system
to test the stacking is the lipid bilayer. The lipid hydrocarbon
tails essentially are the nonpolar solvent in Scheme 1 and the
assembly of 1 in the z-direction would create a transmembrane
nanopore (Figure 1, C). Because the nanopore is open to bulk
water on both ends, the water molecules inside the pore can
readily exchange with bulk water. This is very important if the
pore-formation is to occur. Because the entropic cost for trapping
a single water molecule can be as high as 2 kcal/mol,20 any
partial pore-formation (as in A or B) sould be strongly
disfavored.
Many macrocycles of bile acids have been reported in the
literature.15b,21 A similar cyclic trimer of a cholate derivative
was found to bind monosaccharides.21a,b We thus employed the
well-established glucose leakage assay to test the pore-formation.
Briefly, glucose (300 mM) was first encapsulated within POPC/
POPG large unilamellar vesicles (LUVs) and the external
glucose was removed by gel filtration. When different amounts
of 1 were added to the liposomal solution,22 the glucose that
leaked out was converted by extravesically added hexokinase
and ATP to glucose-6-phosphate, which was oxidized by
glucose-6-phosphate dehydrogenase while NADP was reduced
to NADPH. Because of the fast enzymatic kinetics, the
formation of NADPH at 340 nm correlates directly with the
rate of glucose efflux.23
Because 1 cannot turn its hydrophilic inside out, we did not
consider “toroidal pores”, in which amphiphilic molecules
(typically surfactants or amphipathic peptides) cause local phase
changes in the lipids and produce transient openings in the
membrane.3 We performed the lipid mixing assay and tested
the possibility of membrane fusion as a potential cause of
leakage,24 but <10% lipid mixing occurred even at the highest
[oligocholate]/[lipid] ratio used in the glucose leakage assay
(Figure 1S).25
After lipid fusion was excluded as a main cause for leakage,
we considered three other possible mechanisms for the glucose
efflux, with either the monomer or an aggregate of 1 as the
main transporting species. In a carrier- or ferry-based mecha-
nism, a glucose-bound macrocycle (either in the form of A or
B shown in Figure 1) migrates from the inner to the outer leaflet
of the bilayer, where the guest is released.26 In a relay
mechanism, the guest still has to be bound by A or B but the
binding is only transient and the guest hops from one station to
another before exiting the bilayer. The third possibility is the
hypothesized nanopore, represented by (the idealized) C in
Figure 1.
A carrier-based mechanism typically gives a linear relation-
ship between the leakage rate and the monomer concentration27
but the strong dependence of leakage on the concentration of 1
suggests that its aggregate was the active transporter (vide
infra).3,27 Either B or C would fit such a scenario. To distinguish
between the latter two mechanisms, we studied cyclic tetramer
2 and linear trimer 3 in the transport. The difference between 1
and 2 is not just in size but, more importantly, in their rigidity.
A triangle cannot change its shape as long as the sides are rigid,
To our delight, tricholate 1 was highly effective at transporting
glucose across lipid membranes (Figure 1S, Supporting Infor-
mation). The leakage was strongly dependent on its concentra-
tion. Glucose efflux was negligible below 0.125 µM of 1. The
leakage showed a noticeable increase at 0.25 µM of the
Figure 1. Possible arrangements of cyclic tricholate 1 in a lipid bilayer.
9
J. AM. CHEM. SOC. VOL. 133, NO. 1, 2011 143

A R T I C L E S
Cho et al.
Figure 2. (a) Percent leakage of glucose at 30 min from POPC/POPG LUVs as a function of oligocholate concentration for 1 (0), 2 (4), and 3 (×) at
ambient temperature. Total concentration of phospholipids was 107 µM. These leakage experiments were typically done in duplicates and the error within
the two <10%. (b) Nonlinear least-squares fitting of the leakage data to the Hill equation for 1 (0) and 2 (4). The fraction activity (Y) is the percent glucose
leakage of the LUVs at 30 min after addition of the oligomers.
but a quadrilateral can bend and twist even if the sides are
completely rigid. Thus, stacking and transmembrane pore-
formation should be more difficult with 2 than with 1. Linear
trimer 3 should be even less competent, as it has to fold before
it can stack to form the pore (assuming the same pore-formation
mechanism is involved).
monomer that produces 50% activity. The Hill coefficient (n)
indicates both the stability of the self-assembly and the number
of monomers in the supramolecule responsible for the activity.
A Hill coefficient of n > 1 means that the monomer is the
dominant species and yet the supramolecule is responsible for
the observed activity.3b,29
Figure 2a shows the glucose leakage of LUVs at 30 min in
the presence of the different oligocholates. The topology of the
oligocholates impacted the transport strongly. It took 4-5 times
as much 2 as 1, for example, to leak the same amount of glucose
from the LUVs. The general facial amphiphilicity of the cholate
clearly was not the determining factor, as tetramer 2 contained
more cholates than trimer 1 and yet was less effective. The
conclusion was further supported by linear trimer 3, which
displayed leakage slightly above the background (6-10%) even
at the highest tested concentration. Once the ring structure was
removed, the oligocholate completely lost its ability of transport.
The leakage data in Figure 2a suggests high cooperativity
among the macrocycles.3 A common way to analyze the
cooperativity of a supramolecular system is through the Hill
equation, Y ) Ylow + (Yhigh - Ylow)/[1 + (EC50/c)n], in which
the fractional activity (Y) of a supramolecule is related to the
monomer concentration (c).28 EC50 is the concentration of the
Significantly, the leakage data fit well to the Hill equation
(Figure 2b), yielding a Hill coefficient of n ) 4.0 ( 0.3 for
cyclic trimer 1 and 4.4 ( 0.5 for tetramer 2.30 Thus, for both
macrocycles, the active transporting supramolecule seems to
consist of four macrocycles. POPC bilayer is about 2.6 nm in
the hydrophobic thickness.11,31 The cholate is about 0.6-0.7
nm on the side. To the extent that the Hill analysis reflects
accurately the assembly process, the active transporting species
was most likely a transmembrane pore assembled from four
macrocycles.
Comparison of the three possible arrangements of the
macrocycle in Figure 1 reveals the reasons for the pore-
formation.32 When a cyclic oligocholate enters the membrane,
the introverted hydroxyl/amide groups demand the internal
cavity to be filled with water/glucose instead of lipid hydro-
carbon. If the macrocycle exists as a monomer (Figure 1, A),
the entrapped water/glucose would face hydrocarbon at least
on one end of the cavity or on both ends if the molecule
penetrates deep into the bilayer. The unfavorable hydrophilic-
hydrophobic contact can be reduced if the molecules stack on
top of one another (B), but can be eliminated only if a
transmembrane pore (C) is formed. In the last case, not only is
water-hydrocarbon contact eliminated on both ends of the cavity
for all the macrocycles involved, but also the water molecules
(19) (a) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B.
Angew. Chem., Int. Ed. 2002, 41, 2596–2599. (b) Kolb, H. C.; Finn,
M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2001, 40, 2004–2021.
(20) Dunitz, J. D. Science 1994, 264, 670.
(21) (a) Bonar Law, R. P.; Davis, A. P.; Murray, B. A. Angew. Chem., Int.
Ed. Engl. 1990, 29, 1407–1408. (b) Davis, A. P.; Walsh, J. J. Chem.
Commun. 1996, 449–451. (c) Brotherhood, P. R.; Davis, A. P. Chem.
Soc. ReV. 2010, 39, 3633–3647. (d) Brady, P. A.; Bonar Law, R. P.;
Rowan, S. J.; Suckling, C. J.; Sanders, J. K. M. Chem. Commun. 1996,
319–320. (e) Ghosh, S.; Choudhury, A. R.; Row, T. N. G.; Maitra, U.
Org. Lett. 2005, 7, 1441–1444.
(27) (a) Deng, G.; Merritt, M.; Yamashita, K.; Janout, V.; Sadownik, A.;
Regen, S. L. J. Am. Chem. Soc. 1996, 118, 3307–3308. (b) Merritt,
M.; Lanier, M.; Deng, G.; Regen, S. L. J. Am. Chem. Soc. 1998, 120,
8494–8501.
(22) In our leakage assay, we added 1 to the liposomal solution as a solution
of DMSO (10-20 µL). The amount of DMSO was tested and found
to have negligible effects on the LUVs.
(28) Hille, B. Ionic channels of excitable membranes; 2nd ed.; Sinauer
Associates: Sunderland, MA, 1992.
(23) Kinsky, S. C.; Haxby, J. A.; Zopf, D. A.; Alving, C. R.; Kinsky, C. B.
Biochemistry 1969, 8, 4149–4158.
(29) Litvinchuk, S.; Bollot, G.; Mareda, J.; Som, A.; Ronan, D.; Shah,
M. R.; Perrottet, P.; Sakai, N.; Matile, S. J. Am. Chem. Soc. 2004,
126, 10067–10075.
(24) Struck, D. K.; Hoekstra, D.; Pagano, R. E. Biochemistry 1981, 20,
4093–4099.
(25) The 100% end point in a fusion assay can be measured either after a
surfactant such as Triton X-100 is added to destroy the liposomes or
through a “mock” fusion product (i.e., liposomes whose probe density
corresponds to that of completely fused liposomes). Although the
Triton method gave similar results, we mainly used the second
approach for our fusion experiments because Triton X-100 impacts
the quantum yield of NBD. For a discussion on different lipid-mixing
fusion assays, see: Hoesktra, D.; Du¨zgu¨nes¸, N. Methods Enzymol.
1993, 220, 15–32.
(30) Leakage at 60 min gave similar results, but the curve fitting to the
Hill equation was less certain because the data points clustered at either
very high (Y > 90%) or very low (Y < 30%) leakage.
(31) Lewis, B. A.; Engelman, D. M. J. Mol. Biol. 1983, 166, 203–210.
(32) Hydrogen bonds among the amides along the pore axis are unlikely
to contribute significantly because the amides are exposed to water
molecules inside the pore. Also, the steroid backbone is large on
the amino end and small near the carboxy tail. The geometry of the
macrocycle thus prohibits close contact of the amide bonds in the
z-direction.
(26) Smith, B. D.; Lambert, T. N. Chem. Commun. 2003, 2261–2268.
9
144 J. AM. CHEM. SOC. VOL. 133, NO. 1, 2011

Water-Templated Transmembrane Nanopores
A R T I C L E S
Figure 3. Percent leakage of glucose at 30 min from POPC/POPG LUVs
as a function of oligocholate concentration for 1 (0), 2 (4), and 3 (×) at
ambient temperature. The data points connected by solid lines are for the
LUVs containing 30 mol % of cholesterol, whereas those connected by
dotted lines are for the LUVs without cholesterol, taken from Figure 2.
Total concentration of phospholipids was 107 µM.
Figure 4. Percent leakage of maltotriose at 30 min from POPC/POPG
LUVs as a function of oligocholate concentration for 1 (0), 2 (4), and 3
(×) at ambient temperature. The data points connected by solid lines are
for maltotriose, whereas those connected by dotted lines are for glucose,
taken from Figure 2. Total concentration of phospholipids was 107 µM.
inside the macrocycle are no longer confined in a nanospace.
The water molecules within the pore can solvate all the polar
groups on the inner wall and yet still exchange with the bulk
water rapidly. The exchange of water clearly will be more
difficult if one or both ends of the cavity are capped by
hydrocarbon, as in A or B.
Effect of Cholesterol on the Oligocholate-Induced Leakage.
To gain additional evidence for the pore-formation, we included
30 mol % of cholesterol in the lipid formulation. Cholesterol is
known to increase the hydrophobic thickness11 of POPC bilayer
and decrease its fluidity.12 Cholesterol-containing bilayers are
much less permeable to hydrophilic molecules, including glucose
and glycerol.33
Figure 5. Two views of space-filling molecular models of compound 1
with an included maltotriose. The molecular models were generated by
Chem3D and optimized with the MM2 force field.
Notably, glucose leakage became significantly faster when
the bilayers contained 30 mol % of cholesterol (Figure 3). The
data points connected by solid lines represent leakage from the
cholesterol-containing LUVs and those by dotted lines are from
the cholesterol-free ones. Both the trimer and the tetramer clearly
benefited from cholesterol. The concentration of the transporter
that causes 50% leakage at 30 min (i.e., EC50) for the trimer
went from ∼0.5 to ∼0.1 µM upon cholesterol inclusion; that
for the tetramer decreased from ∼2.4 to ∼0.5 µM. The cyclic
topology remained critical to the transport, as the linear trimer
(×) was completely unaffected by the cholesterol added.
These results strongly support the pore-formation mechanism.
Because cholesterol makes the membrane more hydrophobic,33
the hydrophobic driving force mentioned above is higher for
the stacking of the macrocycles. Cholesterol is able to induce
lateral heterogeneity in lipid membranes.34 If cholesterol-rich
and -deficient domains exist, pore formation should be easier
if the macrocycle phase-separates into one domain. Given that
cholic acid is a metabolite of cholesterol, 1 and 2 are highly
likely to fall into cholesterol-rich domains.
increases. Because it is more difficult for a larger hydrophilic
guest to hop from one station to another in a hydrophobic
membrane, relay will become less efficient as well. Leakage
through a nanopore, on the other hand, should not be affected
very much as long as the cross section of the guest is smaller
than the pore diameter.
We thus studied the permeation of maltotriose by the cholate
macrocycles (Figure 4). Although the trisaccharide is much
longer than glucose, its cross section remains the same.
Consistent with the pore-formation mechanism, the increase of
the sugar size did not slow down the leakage. The EC50 for the
trimer (0) was almost the same for maltotriose (∼0.6 µM) and
glucose (∼0.5 µM). Remarkably, tetramer 2 was considerably
more effective at leaking the longer sugar; the EC50 of the
tetramer (4) went from 2.4 µM for glucose to ∼0.7 µM for
maltotriose.
Why did the tetramer transport a longer sugar better than
a shorter one? The result is highly unusual and contrary to
what the diffusion of the guest (inside the pore) would
predict. As shown by the molecular models (Figure 5), the
trisaccharide is too long to fit within one macrocycle. Hence,
as the sugar enters the membrane, it will thread through the
macrocycles and template the pore formation. If one assumes
that a longer sugar diffuses more slowly than a shorter one
inside the pore, the fact that 2 benefited more than 1 in the
trisaccharide transport suggests that the template effect was
stronger in the tetramer. The conclusion is supported by both
the rigidity consideration and the earlier glucose leakage data,
which suggest that pore-formation is more difficult with the
tetramer. In general, an effect is manifested most strongly
where it is most needed. Such a trend is frequently observed
Effect of Guest Size on the Oligocholate-Induced Leakage.
Different transport mechanisms are expected to respond very
differently to an increase in the guest size. Diffusion of the
carrier-guest complex in the membrane slows down as its size
(33) (a) Demel, R. A.; Bruckdorfer, K. R.; Van Deenen, L. L. Biochim.
Biophys. Acta 1972, 255, 321–330. (b) Papahadjopoulos, D.; Nir, S.;
Ohki, S. Biochim. Biophys. Acta 1972, 266, 561–583.
(34) (a) London, E. Curr. Opin. Struct. Biol. 2002, 12, 480–486. (b) Edidin,
M. Annu. ReV. Biophys. Biomol. Struct. 2003, 32, 257–283. (c) Silvius,
J. R. Biochim. Biophys. Acta 2003, 1610, 174–183. (d) Zhao, J.; Wu,
J.; Heberle, F. A.; Mills, T. T.; Klawitter, P.; Huang, G.; Costanza,
G.; Feigenson, G. W. Biochim. Biophys. Acta 2007, 1768, 2764–2776.
9
J. AM. CHEM. SOC. VOL. 133, NO. 1, 2011 145

A R T I C L E S
Cho et al.
in physical organic chemistrysthe extent of neighboring
group participation, for example, increases with an increase
of the electron demand at the reactive center.35
Another reason why the tetramer benefited more from the
template effect than the trimer could be its larger pore size.
According to the molecular model (Figure 5), a sugar fits quite
snugly within the inner cavity of 1 and forms multiple hydrogen
bonds with the inner wall of the pore. As the sugar gets longer,
the number of hydrogen bonds involved would increase, making
the passage of the guest more difficult. Hence, the template
effect of maltotriose was probably offset by the small pore size
of the trimer. For the tetramer, the larger gap between the guest
and the pore might have geometrically prohibited the formation
of some hydrogen bonds, allowing the trisaccharide to go
through the pore with less hindrance.
Figure 6. The excimer/monomer ratio as a function of [4]/[lipid] ratio in
liposomes made of DLPC (4), POPC/POPG (0), and POPC/POPG with
30 mol % cholesterol (]). The theoretical curves are nonlinear least-squares
fitting of the fluorescence data to the Hill equation.
It should be mentioned that the leakage data with the
cholesterol-containing LUVs (Figure 3) or with maltotriose as
the guest (Figure 4) did not give the high cooperativity shown
in Figure 2. The Hill coefficient (n) was only 1-2 in all the
cases. The Hill coefficients have been reported to change
significantly with minor structural modification of a given
system.29,36 Similar observations were made in biology including
in well-established systems such as the hemoglobin-oxygen
binding. The reason for the change was not always clear. The
Hill equation is known to operate best when extreme positive
cooperativity exists between the binding of the first and second
molecule.37 Such a condition may not be met in the glucose
transport across the cholesterol-containing hydrophobic mem-
brane or in the maltotriose transport where a template effect is
operating.
Figure 6 shows the excimer/monomer ratio for compound
4 as a function of the [4]/[lipid] ratio in three different
bilayers. Two trends are immediately noticeable. First, the
pyrene excimer became stronger as the membrane became
more hydrophobic. Among the three bilayers, the C12 DLPC
membrane (4) was the least hydrophobic and the cholesterol-
containing C16-18 POPC/POPG (]) the most hydrophobic.
With an increase in the membrane hydrophobicity, the
inflection point in the excimer/monomer curves, correspond-
ing to the critical aggregation concentration (CAC) of the
macrocycle, decreased steadily from 0.02 to 0.01 to 0.007.
The trend shows that the formation of pyrene excimer is
promoted by membrane hydrophobicity and tracks well with
the leakage data. Second, all the excimer/monomer curves
were sigmoidal in shapesa hallmark of cooperativity be-
havior.41 In fact, when the fluorescence data were fit to the
Hill equation, the Hill coefficient (n) was ∼1.5 for the DLPC
membrane, ∼3 for POPC/POPG, and ∼4 for the cholesterol-
containing POPC/POPG. The results showed that the number
of the macrocycles in the aggregates correlate with the
membrane thickness. The result is in full agreement with our
pore-forming mechanism and the leakage data. Because the
aggregation is driven by hydrophobic interactions and the
pore needs to span the bilayer (to allow water molecules
inside the pore to exchange with bulk water), the pore length
should not exceed the hydrophobic thickness of the membrane.
Aggregation of Cyclic Oligocholates Studied by Fluorescence.
Due to its long fluorescence lifetime, pyrene can form excimers
quite readily even at relative low concentrations.38 Aggregation
of 4 within a lipid bilayer brings the pyrene groups within
proximity and should promote the excimer formation (emitting
at 470 nm). Instead of the membrane extrusion method39 used
to prepare the LUVs for the leakage assay, we incorporated the
oligocholates into lipid bilayers for dilauroylphosphatidylcholine
(DLPC) and POPC/POPG liposomes by the detergent dialysis
method. This procedure is often employed to reconstitute
membrane proteins into liposomes40 and has the benefit of
generating the most homogeneously mixed lipids. Because
cholesterol was found to interfere with the detergent dialysis
(possibly because of its strong interactions with the Bio-Beads
used in the procedure), we added 4 to preformed LUVs prepared
with the membrane extrusion method.
Conclusions
Classical hydrophobic effect drives the aggregation of
hydrophobic molecules in water. By pulling water into lipid
bilayers with the amphiphilic cholate macrocycles, we “acti-
vated” the water molecules and used them to assemble the
macrocycles into transmembrane nanopores.42 Aggregation and
pore-formation seem to be quite efficient for the cyclic tricholate,
as it only took 1 macrocycle out of 200 lipid molecules to leak
50% of 300 mM glucose from the LUVs in 30 min. Leakage
from the cholesterol-containing LUVs was even more efficients
the same leakage only required 1 macrocycle out of 1000 lipid
molecules.
(35) (a) Lambert, J. B.; Mark, H. W. J. Am. Chem. Soc. 1978, 100, 2501–
2505. (b) Lambert, J. B.; Larson, E. G. J. Am. Chem. Soc. 1985, 107,
7546–7550.
(36) (a) Talukdar, P.; Sakai, N.; Sorde´, N.; Gerard, D.; Cardona, V. M. F.;
Matile, S. Bioorg. Med. Chem. 2004, 12, 1325–1336. (b) Bhosale, S.;
Matile, S. Chirality 2006, 18, 849–856. (c) Shank, L. P.; Broughman,
J. R.; Takeguchi, W.; Cook, G.; Robbins, A. S.; Hahn, L.; Radke, G.;
Iwamoto, T.; Schultz, B. D.; Tomich, J. M. Biophys. J. 2006, 90, 2138–
2150. (d) Ferdani, R.; Pajewski, R.; Djedovicˇ, N.; Pajewska, J.;
Schlesinger, P. H.; Gokel, G. W. New J. Chem. 2005, 29, 673–680.
(e) Chen, W. H.; Shao, X. B.; Regen, S. L. J. Am. Chem. Soc. 2005,
127, 12727–12735.
(37) Weiss, J. N. FASEB J. 1997, 11, 835–841.
(38) Birks, J. B.; Munro, I. H.; Dyson, D. J. Proc. R. Soc. London Ser. A
1963, 275, 575–588.
(41) Chan, H. S.; Bromberg, S.; Dill, K. A. Philos. Trans. R. Soc. London
B 1995, 348, 61–70.
(39) Olson, F.; Hunt, C. A.; Szoka, F. C.; Vail, W. J.; Papahadjopoulos,
D. Biochim. Biophys. Acta 1979, 557, 9–23.
(42) For an example of water-promoted assembly in the solid state, see:
Carrasco, H.; Foces-Foces, C.; Pe´rez, C.; Rodr´ıguez, M. L.; Mart´ın,
J. D. J. Am. Chem. Soc. 2001, 123, 11970–11981.
(40) Smith, S. A.; Morrissey, J. H. J. Thromb. Haemost. 2004, 2, 1155–
1162.
9
146 J. AM. CHEM. SOC. VOL. 133, NO. 1, 2011

Water-Templated Transmembrane Nanopores
A R T I C L E S
The hydrophobically driven pore-forming mechanism yielded
some quite unusual transport properties. Contrary to conven-
tional expectations, permeation of hydrophilic guests occurs
more readily as the membrane becomes more hydrophobic and
longer sugars passed through the membranes more readily than
shorter ones. Transmembrane movement of sugars is ac-
complished by complex protein transporters in nature,43 but our
oligocholates can be synthesized in a few steps from the cholate
monomer.44 Given the unique pore-forming mechanism, the easy
synthesis of the oligocholates, the biocompatibility of cholic
acid, and the numerous uses of nanopores,1-9 these compounds
may find many applications in biology and chemistry in the
future.
Acknowledgment. We thank Dr. Xingang Pan for providing
compound 1 and performing some initial leakage experiments. We
thank NSF (CHE-0748616 and DMR-1005515) for financial
support.
Supporting Information Available: Experimental Section,
including general experimental details, synthesis and charac-
terization of the compounds, leakage assays, and fluorescence
experiments. This material is available free of charge via the
Internet at http://pubs.acs.org.
(43) Saier, M. H., Jr. In The Structure of biological membranes; Yeagle,
P., Ed; CRC Press: Boca Raton, 1992; Chapter 18.
(44) For boronic acid-based sugar transporters, see: (a) Westmark, P. R.;
Smith, B. D. J. Am. Chem. Soc. 1994, 116, 9343–9344. (b) Westmark,
P. R.; Gardiner, S. J.; Smith, B. D. J. Am. Chem. Soc. 1996, 118,
11093–11100.
JA109036Z
9
J. AM. CHEM. SOC. VOL. 133, NO. 1, 2011 147