Single-molecular artificial transmembrane water channels

Article abstract of DOI:10.1021/ja302292c

Hydrazide-appended pillar[5]arene derivatives have been synthesized. X-ray crystal structure analysis and ¹H NMR studies revealed that the molecules adopt unique tubular conformations. Inserting the molecules into the lipid membranes of vesicles leads to the transport of water through the channels produced by single molecules, as supported by dynamic light scattering and cryo-SEM experiments. The channels exhibit the transport activity at a very low channel to lipid ratio (0.027 mol %), and a water permeability of 8.6 × 10^-10 cm s^-1 is realized. In addition, like natural water channel proteins, the artificial systems also block the transport of protons.

Full text of DOI:10.1021/ja302292c

Communication
pubs.acs.org/JACS
Single-Molecular Artificial Transmembrane Water Channels
Xiao-Bo Hu, Zhenxia Chen, Gangfeng Tang, Jun-Li Hou,* and Zhan-Ting Li
Department of Chemistry, Fudan University, 220 Handan Road, Shanghai 200433, China
S
* Supporting Information
a
Chart 1. Structures of Compounds 1−4
ABSTRACT: Hydrazide-appended pillar[5]arene deriva-
tives have been synthesized. X-ray crystal structure analysis
1
and H NMR studies revealed that the molecules adopt
unique tubular conformations. Inserting the molecules into
the lipid membranes of vesicles leads to the transport of
water through the channels produced by single molecules,
as supported by dynamic light scattering and cryo-SEM
experiments. The channels exhibit the transport activity at
a very low channel to lipid ratio (0.027 mol %), and a
water permeability of 8.6 × 10−10 cm s−1 is realized. In
addition, like natural water channel proteins, the artificial
systems also block the transport of protons.
ransmembrane water transport is crucially important for
the regulation of the osmotic pressure of cells. Nature
T
utilizes water channel proteins, called aquaporins,1 to realize
this purpose.2 The development of synthetic systems displaying
powerful water transport capability not only is fundamentally
important but also may lead to the generation of new
medicinal3 and environmental materials, for example, for
water purification.4 Currently, a large number of artificial
systems for transmembrane transport of ions and neutral
organic solutes have been reported.5,6 In contrast, although
theoretical calculations have suggested that many structures can
be used for transmembrane water transport,7,8 the creation of
mimicking systems has been a challenge because of the lack of
synthetic architectures that can form long and narrow channels
for water.9 We were inspired to design artificial water channels
by mimicking the single pore feature of natural channel
proteins.10 Herein we describe the first example of artificial
single-molecular transmembrane water channels formed from
hydrazide-appended pillar[5]arenes, which display transport
activity at very low concentration.
a
The hydrazide units of 2a, 2b, and 3 form pentameric cylinders
through intramolecular hydrogen bonding to induce the formation of
tubular structures.
We recently reported that pillar[5]arene derivative 1 (Chart
1) induces the formation of water wires,11 which could be
utilized to mediate transmembrane proton transport.12,13 To
exploit further the potential of such tubular systems for water
transport, we prepared 2a, 2b, and 3 by attaching 10 hydrazide-
incorporated side chains to the central pillar[5]arene scaffold
(Chart 1). It was envisioned that the hydrazide units in the side
chains might form pentameric cylinders through intramolecular
hydrogen bonding to induce the whole molecules to produce
tubular structures.14
Diluting their solutions from 2 mM to 0.2 mM did not cause an
observable change in the chemical shift of the N−H signals.
The spectra recorded in deuterated dimethyl sulfoxide
(DMSO-d6) had high resolution [see the Supporting
Information (SI)]. All of these results suggested that the
hydrazide units of 2a, 2b, and 3 were involved in intramolecular
hydrogen bonding, which should induce the whole molecule to
form a tubular conformation. Single crystals of 2b suitable for
X-ray crystallographic analysis were obtained by slow
evaporation of its solution in 1:1 (v/v) CHCl3/MeOH. The
crystal structure revealed that eight of the 10 hydrazide groups
1
The H NMR spectrum of control molecule 4 in CDCl3
(Figure 1a) displayed one set of sharp signals, and the N−H
signals of the hydrazide units appeared at 9.28 and 8.67 ppm. In
contrast, the spectra of 2a, 2b, and 3 all had lower resolution,
and their N−H signals were shifted downfield (Figure 1b−d).
Received: March 8, 2012
Published: May 10, 2012
© 2012 American Chemical Society
8384
dx.doi.org/10.1021/ja302292c | J. Am. Chem. Soc. 2012, 134, 8384−8387

Journal of the American Chemical Society
Communication
Figure 3. (a) Changes in light scattering and (b) mean diameter of the
vesicles with time after addition of a DMSO solution of 1, 2a, or 3 (0.3
mol % relative to lipid) or pure DMSO. The vesicles were loaded with
NaCl (100 mM) and suspended in distilled water.
Figure 1. Partial 1H NMR spectra of (a) 4, (b) 2a, (c) 2b, and (d) 3 in
CDCl3 at 298 K. The concentration was 2.0 mM.
and driven by the osmotic pressure difference between the
outside and inside of the vesicles (Figure 4). The larger light
scattering decrease caused by 3 suggests that it was more
efficient than 1 and 2a in transporting water. After 125 min, the
light scattering decrease continued for 3 but showed larger
fluctuations, probably due to precipitation of the very large
vesicles.18
Figure 2. Crystal structure (stick model) of 2b: (left) side view;
(right) top view. The O atoms of four entrapped water molecules are
highlighted using CPK models. The tert-butyl groups and H atoms on
the C atoms have been omitted for clarity.
form eight intramolecular hydrogen bonds in a head-to-tail
manner (Figure 2, left). Another two form intermolecular
hydrogen bonds. However, the whole molecule produces a
tubular cavity (Figure 2, right). In view of the dynamic nature
of hydrogen bonding in solution, it is reasonable to assume that
this series of pillar[5]arene derivatives should also form similar
tubular conformations. The tubular cavity of 2b also
encapsulates four water molecules that are hydrogen-bonded
with carboxylic groups (Figure 2, left). This observation
prompted us to investigate the capacity of these molecules in
transporting water across membranes.
The water transport property of the tubular structures across
lipid bilayers was investigated using dynamic light scattering
(DLS).15 Large unilamellar vesicles in buffer containing NaCl
(100 mM) were first prepared from egg yolk L-α-
phosphatidylcholine (EYPC)16 and then suspended in pure
water to produce different osmotic pressures inside and outside
the vesicles. Aliquots of DMSO solutions of 1, 2a, and 3 (0.3%
molar ratio relative to lipid) were injected into the suspensions
of the vesicles. The light scattering and vesicle diameters were
then monitored for 400 min (Figure 3). For the systems
containing 1 and 2a, the normalized light scattering slowly
decreased by 5 and 10%, respectively, after 400 min, while for
the system containing 3, the light scattering decreased by 25%
after 125 min (Figure 3a). The decrease in light scattering is a
result of swelling of the vesicles.9 It is established that lipid
membranes themselves allow only for slow transport of water.17
A control experiment also showed that in the absence of the
tubular compounds, a very slight light scattering decrease (2%)
was observed (Figure 3a, black). Thus, the above light
scattering decrease can be attributed to water transport from
the outside to the inside of the vesicles that was realized
through insertion of the tubular molecules into the membranes
Figure 4. Schematic representation of the increase in vesicle size
caused by 3-mediated outside-to-inside water transport.
By fitting the above light scattering versus time plots with a
linear model,9 the water-transport activity rate constants (k) for
1, 2a, and 3 were calculated to be (3.6 0.4) × 10−7, (9.4
0.5) × 10−7, and (3.58 0.2) × 10−5 s−1, respectively. The fact
that 3a displayed the highest transport activity may be
rationalized by considering that its length (5.1 nm as
determined using a CPK model) matches well the width of
the lipid bilayer (5.0 nm).5b CPK modeling showed that tubular
2a has a length of 3.5 nm. Thus, it might also transport water in
a single-molecular manner that is less efficient because of
possible blockage at the ends by lipid molecules. The even
shorter compound 1 could mediate this process only through
stacking of two molecules in the membrane12 and thus was also
less efficient.
Figure 3b shows that adding the solution of 1 or 2a in
DMSO caused a small increase in the mean diameter of the
vesicles, while pure DMSO did not impose a detectable
influence. In contrast, adding 3 caused the mean vesicle
diameter to increase from ca. 150 to ca. 370 nm within 125
min. After that, the vesicle diameter decreased gradually, most
likely because of the precipitation of large vesicles. The size of
the vesicles increased 1.5-fold before 125 min, which exceeded
the limitation of simple vesicle swelling (10%).19 To get a deep
8385
dx.doi.org/10.1021/ja302292c | J. Am. Chem. Soc. 2012, 134, 8384−8387

Journal of the American Chemical Society
Communication
insight into the vesicle swelling process, the vesicle samples
were checked by cryogenic scanning electron microscopy
(cryo-SEM) (Figure 5), which indicated that the size of the
mechanism. The osmotic water permeability (Pf)21 of 3 at 0.3
mol % was calculated to be 8.6 × 10−10 cm s−1 (see the SI) by
assuming that two vesicles fused to produce one big vesicle and
that all of the added channel molecules were inserted into lipid
bilayers. The actual value should be higher than this estimated
one, since it is impossible for all of the molecules to be inserted
into the membranes.
The selectivity of the channels for OH− and H+ was also
studied by imposing a pH gradient across the membranes of the
vesicles. A suspension of vesicles entrapping the pH-sensitive
dye 8-hydroxypyrene-1,3,6-trisulfonate (HPTS)22 and Na+
buffer (pH 7.0 inside the vesicles) was prepared and then
added to a K+ buffer having a pH of 7.6 to produce a higher K+
and OH− concentration outside the vesicles. The OH− flux
through the membranes was assessed by monitoring the
fluorescence intensity of HPTS. Adding 1, 2a, and 3 (0.3 mol %
relative to lipid) to the above vesicle suspension caused a small
change (<5%) in the fluorescence intensity (Figure 7a).
Figure 5. Cryo-SEM images of the sample recorded (a) 0, (b) 60, (c)
150, (d) 350 min after addition of 3 (0.3 mol % relative to lipid),
highlighting the size increase of vesicles with time as well as vesicle
fusion (b). The bright particles are ice crystals.
vesicles increased with time. Figure 5b also showed the fusion
of vesicles, which may account for the formation of large
vesicles. When the vesicles were suspended in buffer containing
NaCl (100 mM), the osmotic pressure was the same inside and
outside the vesicles. Adding 3 to the suspension did not cause
the light scattering and size of the vesicles to change within 200
min (Figure S20 in the SI), implying that 3 did not promote
the fusion of vesicles. Thus, the above fusion should be a result
of vesicle swelling that occurred through the outside-to-inside
water transport (Figure 4).20
The water transport activities of 1, 2a and 3 at different
channel to lipid ratios (mol %) was further investigated. It was
found that a high channel/lipid ratio (>0.9 mol %) caused the
vesicles to break, affording precipitates in a few minutes. When
the molar ratio was <0.09%, the transport constants of 1 and 2a
could not be obtained because of their weak transport ability.
The transport activity of 3 was found to be linearly dependent
on the molar ratio over the range 0.027−0.45% (Figure 6). This
linear relationship supports the above single-molecular
Figure 7. (a) Change in fluorescence intensity (λex = 460 nm, λem
=
510 nm) of vesicles with time after addition of 1, 2a, or 3 (0.3 mol %
relative to lipid) and/or valinomycin (0.002 mol % relative to lipid).
The vesicles were loaded with NaCl (100 mM) and HPTS (0.1 mM)
buffered at pH 7.0 with HEPES (10 mM) and suspended in KCl (100
mM) solution (buffered to pH 7.6). (b) Schematic representation of
(top) OH− transport and (bottom) H+ blockage in the channels.
However, adding their mixtures with valinomycin (0.002 mol
% relative to lipid), an efficient K+ carrier,23 caused the intensity
to increase by 28, 42, and 51%, respectively, in 10 min. In
contrast, the same amount of valinomycin itself resulted in only
a 13% increase in the fluorescence intensity. These results
support that the conclusion that 1, 2a, and 3 can also transport
OH− (Figure 7b, top). Again, 3 exhibited the highest transport
ability, which can also be attributed to the fact that its length fits
best with the width of the lipid bilayer. Using a similar strategy,
we prepared a suspension of vesicles entrapping HPTS and K+
buffer (pH 7.0 inside the vesicles) and then added it to a Na+
buffer having a pH of 5.5 to produce higher H+ concentration
outside the vesicles. In the presence or absence of valinomycin,
1, 2a, and 3 (0.3 mol % relative to lipid) did not cause any
change in the fluorescence intensity of HPTS (Figure S21),
suggesting that they did not transport H+ across the membrane
(Figure 7b, bottom). This was not unexpected, as a flux of H+
has to be realized through a continuous hydrogen-bonding
network formed by water or hydroxyl groups.12,24 This result
features one of the key functions of natural water channels,
Figure 6. (a) Changes in light scattering with time in the presence of 3
⧫
●
○
△
▽
[0.027% ( ), 0.075% ( ), 0.15% ( ), 0.23% ( ), 0.3% ( ), and
■
0.45% ( ) relative to lipid]. (b) Plot of the transport rate constant k
for 3 against the molar ratio relative to lipid.
8386
dx.doi.org/10.1021/ja302292c | J. Am. Chem. Soc. 2012, 134, 8384−8387

Journal of the American Chemical Society
Communication
(11) Si, W.; Hu, X.-B.; Liu, X.-H.; Fan, R.; Chen, Z.; Weng, L.; Hou,
J.-L. Tetrahedron Lett. 2011, 52, 2484.
(12) Si, W.; Chen, L.; Hu, X.-B.; Tang, G.; Chen, Z.; Hou, J.-L.; Li,
which do not form the hydrogen- bonded water wires and thus
block H+ flux.25
In conclusion, we have developed a class of artificial
transmembrane water channels based on hydrazide-incorpo-
rated pillar[5]arenes. The pillar[5]arene derivatives are induced
to form tubular conformations by intramolecular hydrogen
bonding and thus can function as single-molecular channels to
transport water across the lipid membrane at very low
concentration. Although the transport rate is not as high as
for natural channels, the results described here were obtained
using unoptimized molecules. Currently we are modifying the
backbones to achieve more efficient systems.
Z.-T. Angew. Chem., Int. Ed. 2011, 50, 12564.
(13) For the application of pillar[5]arene derivatives in molecular
recognition, see: (a) Ogoshi, T.; Kanai, S.; Fujinami, S.; Yamagishi, T.-
A.; Nakamoto, Y. J. Am. Chem. Soc. 2008, 130, 5022. (b) Li, C.; Xu, Q.;
Li, J.; Yao, F.; Jia, X. Org. Biomol. Chem. 2010, 8, 1568. (c) Dong, S.;
Luo, Y.; Yan, X.; Zheng, B.; Ding, X.; Yu, Y.; Ma, Z.; Zhao, Q.; Huang,
F. Angew. Chem., Int. Ed. 2011, 50, 1397. (d) Hu, X.-B.; Chen, L.; Si,
W.; Yu, Y.; Hou, J. Chem. Commun. 2011, 47, 4694. (e) Zhang, H.;
Strutt, N. L.; Stoll, R. S.; Li, H.; Zhu, Z.; Stoddart, J. F. Chem.
Commun. 2011, 47, 11420. (f) Cragg, P. J.; Sharma, K. Chem. Soc. Rev.
2012, 41, 597.
(14) Zhao, X.; Wang, X.-Z.; Jiang, X.-K.; Li, Z.-T.; Chen, G.-J. J. Am.
Chem. Soc. 2003, 125, 15128.
ASSOCIATED CONTENT
* Supporting Information
Synthetic procedures and characterization data for 1−4;
detailed procedures for water, OH−, and H+ transport
experiments; and a CIF file for 2b. This material is available
free of charge via the Internet at http://pubs.acs.org.
■
S
(15) (a) Hallett, F. R.; Marsh, J.; Nickel, B. G.; Wood, J. M. Biophys.
J. 1993, 64, 435. (b) Borgnia, M. J.; Kozano, D.; Calamita, G.;
Maloney, P. C.; Agre, P. J. Mol. Biol. 1999, 291, 1169.
(16) Jeon, Y. J.; Kim, H.; Jon, S.; Selvapalam, N.; Oh, D. H.; Seo, I.;
Park, C.; Jung, S. R.; Koh, D.; Kim, K. J. Am. Chem. Soc. 2004, 126,
15944.
(17) (a) Finkelstein, A.; Cass, A. J. Gen. Physiol. 1967, 50, 1765.
(b) Andrasko, J.; Forsen, S. Biochem. Biophys. Res. Commun. 1974, 60,
813. (c) Lipschitz-Farber, C.; Degani, H. Biochim. Biophys. Acta 1980,
600, 291. (d) Huster, D.; Jin, A. J.; Arnold, K.; Gawrisch, K. Biophys. J.
1997, 73, 855.
AUTHOR INFORMATION
Corresponding Author
houjl@fudan.edu.cn
■
Notes
(18) Sakai, N.; Matile, S. Chirality 2003, 15, 766.
(19) Haines, T. H.; Li, W.; Green, M.; Cummins, H. Z. Biochemistry
The authors declare no competing financial interest.
1987, 26, 5439.
ACKNOWLEDGMENTS
■
(20) Cohen, F. S.; Akabas, M. H.; Finkelstein, A. Science 1982, 217,
458.
(21) Zhang, R.; Logee, K. A.; Verkman, K. S. J. Biol. Chem. 1990, 265,
15375.
(22) Kano, K.; Fendler, J. H. Biochim. Biophys. Acta 1978, 509, 289.
(23) Dobler, M. Ionophores and Their Structures; Wiley: New York,
1981.
(24) Weiss, L. A.; Sakai, N.; Ghebremariam, B.; Ni, C.; Matile, S. J.
Am. Chem. Soc. 1997, 119, 12142.
This work was supported by NSFC (20902012 and 91027008),
FANEDD (200930), STCSM (10PJ1401200), the Innovation
Program of SHMEC (10ZZ01), and Shanghai Key Laboratory
of Molecular Catalysis and Functional Materials.
REFERENCES
■
(1) Preston, G. M.; Carroll, T. P.; Guggino, W. B.; Agre, P. Science
1992, 256, 385.
(2) de Groot, B. L.; Frigato, T.; Helms, V.; Grubmuller, H. J. Mol.
(25) (a) Burykin, A.; Warshel, A. Biophys. J. 2003, 85, 3696.
(b) Jensen, M.; Rothlisberger, U.; Rovira, C. Biophys. J. 2005, 89, 1744.
Biol. 2003, 333, 279.
(3) Agre, P.; King, L. S.; Yasui, M.; Guggino, W. B.; Ottersen, O. P.;
Fujiyoshi, Y.; Engel, A.; Nielsen, S. J. Physiol. 2002, 542, 3.
(4) (a) Kumar, M.; Grzelakowski, M.; Zilles, J.; Clark, M.; Meier, W.
Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 20719. (b) Kaufman, Y.;
Berman, A.; Freger, V. Langmuir 2010, 26, 7388.
(5) (a) Akerfeldt, K. S.; Lear, J. D.; Wasserman, Z. R.; Chung, L. A.;
DeGrado, W. F. Acc. Chem. Res. 1993, 26, 191. (b) Gokel, G. W.;
Mukhopadhyay, A. Chem. Soc. Rev. 2001, 30, 274. (c) Fyles, T. M.
Chem. Soc. Rev. 2007, 36, 335. (d) Davis, A. P.; Sheppard, D. N.;
Smith, B. D. Chem. Soc. Rev. 2007, 36, 348. (e) Matile, S.; Vargas
Jentzsch, A.; Montenegro, J.; Fin, A. Chem. Soc. Rev. 2011, 40, 2453.
(f) Gale, P. A. Acc. Chem. Res. 2011, 44, 216.
(6) (a) Cho, H.; Widanapathirana, L.; Zhao, Y. J. Am. Chem. Soc.
2011, 133, 141. (b) Cho, H.; Zhao, Y. Langmuir 2011, 27, 4936.
(c) Zhang, S.; Zhao, Y. Chem.Eur. J. 2011, 17, 12444.
(7) (a) Hummer, G.; Rasaiah, J. C.; Noworyta, J. P. Nature 2001,
414, 188. (b) Zhu, F.; Schulten, K. Biophys. J. 2003, 85, 236.
(c) Striolo, A. Nano Lett. 2006, 6, 633.
(8) (a) Engels, M.; Bashford, D.; Ghadiri, M. R. J. Am. Chem. Soc.
1995, 117, 9151. (b) Portella, G.; de Groot, B. L. Biophys. J. 2009, 96,
925. (c) Titov, A. V.; Wang, B.; Sint, K.; Kral, P. J. Phys. Chem. B 2010,
114, 1174. (d) Liu, J.; Fan, J.; Tang, M.; Cen, M.; Yan, J.; Liu, Z.;
Zhou, W. J. Phys. Chem. B 2010, 114, 12183.
(9) An example of a self-assembled imidazole-quartet water channel
has been reported. See: Duc, Y. L.; Michau, M.; Gilles, A.; Gence, V.;
Legrand, Y.-M.; Lee, A.; Tingry, S.; Barboiu, M. Angew. Chem., Int. Ed.
2011, 50, 11366.
(10) Verkman, A. S. J. Cell Sci. 2011, 124, 2107.
8387
dx.doi.org/10.1021/ja302292c | J. Am. Chem. Soc. 2012, 134, 8384−8387