Molecular umbrella-assisted transport of glutathione across a phospholipid membrane

Article abstract of DOI:10.1021/ja010124r

A di-walled molecular umbrella (1a) has been synthesized by acylation of the terminal amino groups of spermidine with cholic acid, followed by condensation with bis{3-O-[N-1,2,3-benzotriazin-4(3H)-one]y1}-5,5′-dithiobis-2- nitrobenzoate (BDTNB), and displacement with glutathione (γ-Glu-Cys-Gly, GSH). Replacement of the sterol hydroxyls with sulfate groups, prior to displacement with GSH, afforded a hexasulfate analogue 1b. Both conjugates have been found to enter large unilamellar vesicles (200 nm diameter, extrusion) of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), and to react with entrapped GSH to form oxidized glutathione (GSSG). Evidence for vesicular entry has come from the formation of oxidized glutathione (GSSG) within the interior of the vesicle, the appearance of the thiol form of the umbrella (USH), and the absence of release of GSH into the external aqueous phase. Results that have been obtained from monolayer experiments, together with the fact that the heavily sulfated conjugate is able to cross the phospholipid bilayer, have yielded strong inferential evidence for an "umbrella-like" action of these molecules as they cross the lipid bilayer.

Full text of DOI:10.1021/ja010124r

J. Am. Chem. Soc. 2001, 123, 5401-5406
5401
Molecular Umbrella-Assisted Transport of Glutathione Across a
Phospholipid Membrane
Vaclav Janout, Lan-hui Zhang, Irina V. Staina, Christophe Di Giorgio, and
Steven L. Regen*
Contribution from the Department of Chemistry and Zettlemoyer Center for Surface Studies,
Lehigh UniVersity, Bethlehem, PennsylVania 18015
ReceiVed January 16, 2001
Abstract: A di-walled molecular umbrella (1a) has been synthesized by acylation of the terminal amino groups
of spermidine with cholic acid, followed by condensation with bis{3-O-[N-1,2,3-benzotriazin-4(3H)-one]yl}-
5,5′-dithiobis-2-nitrobenzoate (BDTNB), and displacement with glutathione (γ-Glu-Cys-Gly, GSH). Replacement
of the sterol hydroxyls with sulfate groups, prior to displacement with GSH, afforded a hexasulfate analogue
1b. Both conjugates have been found to enter large unilamellar vesicles (200 nm diameter, extrusion) of
1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), and to react with entrapped GSH to form oxidized
glutathione (GSSG). Evidence for vesicular entry has come from the formation of oxidized glutathione (GSSG)
within the interior of the Vesicle, the appearance of the thiol form of the umbrella (USH), and the absence of
release of GSH into the external aqueous phase. Results that have been obtained from monolayer experiments,
together with the fact that the heavily sulfated conjugate is able to cross the phospholipid bilayer, have yielded
strong inferential evidence for an “umbrella-like” action of these molecules as they cross the lipid bilayer.
Introduction
Chart 1
One of the most significant challenges presently facing
medicinal chemists is to find ways of promoting the passive
transport of polar, biologically active agents across lipid
membranes. This challenge exists not only for large- and
intermediate-sized molecules (e.g., DNA and antisense oligo-
nucleotides, respectively), but also for relatively small molecules
(e.g., peptides).1-10 In an attempt to devise a general solution
to this problem, we have focused our attention on the synthesis
of amphiphilic molecules that mimic the structure and function
of umbrellas, i.e., “molecular umbrellas” that cover an attached
agent and shield it from an incompatible environment.11-14
The design principle upon which our efforts have been based
is illustrated in Chart 1. In brief, molecular umbrellas are
composed of two or more amphiphilic “walls” (rigid hydro-
carbon units that maintain a hydrophobic and a hydrophilic face),
which are covalently coupled to a central scaffold. Our working
hypothesis has been that such molecules will favor a shielded
conformation, when taken up into the hydrocarbon region of a
membrane, and that this conformation will mask the strong
hydrophilicity of the attached agent, thereby facilitating its
partitioning into, and diffusion across the membrane.
(1) Gennis, R. B. Biomembranes: Molecular Structure and Function;
Springer-Verlag: New York, 1989.
(2) Akhtar, S.; Juliano, R. L. Trends Cell Biol. 1992, 2, 139.
(3) Akhtar, S.; Basu, S.; Wickstrom, E.; Juliano, R. L. Nucleic Acids
Res. 1991, 19, 5551.
(4) Ramaswami, V.; Zhu, Z.; Romanowski, M.; Haaseth, R. C.; Misicka,
A.; Lipkowski, A. W.; Hruby, V. J.; O’Brien, D. F. Int. J. Peptide Protein
Res. 1996, 48, 87.
(5) De Mesmaeker, A.; Haner, R.; Martin, P.; Moser, H. E. Acc. Chem.
Res. 1995, 28, 366.
(6) Uhlmann, E.; Peyman, A. Chem. ReV. 1990, 90, 543.
(7) Lewis, J. G.; Lin, K. Y.; Kothavale, A.; Flanagan, W. M.; Matteucci,
M. D.; DePrince, R. B.; Mook, R. A., Jr.; Hendren, R. W.; Wagner, R. W.
Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 3176.
(8) Burton, P. S.; Conradi, R. A.; Hilgers, A. R. AdV. Drug DeliV. 1991,
7, 365.
(9) Kreimeyer, A.; Amdre. F.; Gouyette, C.; Huynh-Dinh, T. Angew.
Chem., Int. Ed. 1998, 37, 2853.
(10) Kreimeyer, A.; Andre, F.; Gouyette, C.; Huynh-Dinh, T. Angew.
Chem., Int. Ed. Engl. 1998, 37, 2853.
In previous studies, we have shown that molecular umbrellas
exhibit “molecular amphomorphism”, i.e., the ability to form a
shielded or exposed conformation in response to changes in the
hydrophilic/hydrophobic character of the microenvironment.11,12
We have also shown that certain di-walled molecular umbrellas,
as well as much larger tetra-walled analogues, can readily move
from one side of a lipid bilayer to the other.14 Whether or not
molecular umbrellas are able to promote the passive transport
of an attached polar agent across a membrane, however, has
been a key question that we have sought to answer.
(11) Janout, V.; Lanier, M.; Regen, S. L. J. Am. Chem. Soc. 1996, 118,
1573.
(12) Janout, V.; Lanier, M.; Regen, S. L. J. Am. Chem. Soc. 1997, 119,
640.
(13) Janout, V.; Lanier, M.; Deng, G. Bioconj. Chem. 1997, 8, 891.
(14) Shawaphun, S.; Janout, V.; Regen, S. L. J. Am. Chem. Soc. 1999,
121, 5860.
In this paper, we report the design and synthesis of two di-
walled molecular umbrellas which bear glutathione (Glu-Cys-
Gly) as the attached agent. We also report their transport
properties with respect to phospholipid bilayers made from
10.1021/ja010124r CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/18/2001

5402 J. Am. Chem. Soc., Vol. 123, No. 23, 2001
Janout et al.
1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC).15
Monolayer studies that have been carried out with one of these
conjugates in the absence, and in the presence, of POPC provide
insight into its likely conformation within a phospholipid bilayer.
TLC [silica, CHCl3/CH3OH/H2O, 103/27/2 (v/v/v)] to give 66.0 mg
(61%) of 3 having Rf 0.60 and H NMR (CD3OD) δ 8.22 (m, 2 H),
7.83 (m, 2 H), 7.56 (d, 2 H), 3.92 (s, 4 H), 3.78 (s, 4 H), 3.54 (m, 4
H), 3.35 (m, 4 H), 3.30 (m, 4 H), 3.10 (m, 4 H), 2.95 (m, 4 H), 0.90-
2.27 (m, 120 H), 0.90 (s, 12 H), 0.70 (s, 12 H).
1
Mixed Disulfide of GSH and N1,N3-Bis[dicholeamidospermidineyl]-
N2-(2-nitro-5-mercaptobenzamide) (1a). To a solution of 65.0 mg
(25.9 µmol) of 4 in 1.5 mL of methanol was added, dropwise, a solution
of 7.9 mg (25.9 µmol) of GSH in SHE buffer (8 mM NaCl, 2 mM
HEPES, 0.05 mM EDTA, pH 7.0) and the mixture stirred for 12 h at
room temperature. After concentration under reduced pressure (35 °C),
the product was dissolved in a minimum volume of CH3OH and purified
by preparative TLC [silica, CHCl3/CH3OH/H2O, 60/40/10 (v/v/v)] to
give 1a (53%) having Rf 0.37 and 1H NMR (CD3OD) δ 8.25 (m, 1 H),
7.84 (m, 1 H), 7.61 (d, 1 H), 4.69 (m, 1 H), 3.94 (s, 2 H), 3.78 (s, 2
H), 3.70 (m, 3 H), 3.35 (m, 2 H), 3.30 (m, 6 H), 3.15 (m, 2 H), 3.05
(m, 2 H), 2.50 (m, 2 H), 0.70-2.29 (m, 68 H), 0.69 (s, 6 H). HRMS
for C72H113N7O17S2 (MNa+): calcd 1434.7532, found 1434.7603.
Bis [N1,N3-Dicholeamidospermidineyl]-5,5′-dithiobis(2-nitro-
benzamide), Persulfate, Sodium Salt (5). To a solution of 41.0 mg
(18.5 µmol) of 4 in 1.20 mL of anhydrous DMF was added 143 mg
(0.9 mmol) of SO3‚Py and the mixture stirred at for 16 h at room
temperature. Removal of solvent under reduced pressure (55 °C, 10
Torr), followed by drying [23 °C, 2 h (1 Torr)], afforded a solid that
was redissolved in 2.50 mL of a saturated aqueous solution of NaHCO3.
The water was then removed under reduced pressure (10 Torr, 45°),
and the resulting solid extracted with 3 × 5 mL of CH3OH. The solution
was concentrated under reduced pressure to give 71 mg of crude
product, which was purified by preparative TLC (silica gel, CHCl3/
CH3OH/H2O, 34/28/8, v/v/v) to give 60 mg (94%) of 5 having Rf 0.60,
Experimental Section
General Methods. Unless stated otherwise, all reagents were
obtained from commercial sources and used without further purification.
1-Palmitoyl-2-oleoyl-sn-glycero-phosphocholine (POPC) was obtained
1
from Avanti Polar Lipids (Alabaster, AL). All H NMR spectra were
recorded on a Bruker 360 MHz instrument; chemical shifts are reported
in ppm and are referenced to residual solvent. The borate buffer that
was used in all experiments was composed of 0.1 M H3BO3, and 2
mM EDTA, where the pH was adjusted to 7.0 by use of 1 M NaOH.
N1,N3-Dicholeamidospermidine was prepared using procedures similar
to those previously described.12
Bis{3-O-[N-1,2,3-Benzotriazin-4(3H)-one]yl}-5,5′dithiobis-2-ni-
trobenzoate (BDTNB). To a suspension of 1.19 g (3.0 mmol) of 5,5′-
dithiobis(2-nitrobenzoic acid and 1.078 g (6.52 mmol) of 3-hydroxy-
1,2,3-benzotriazin-4H(3H)-one in 70 mL of CH2Cl2 was added 1.343
g (6.52 mmol) of dicyclohexylcarbodiimide. The heterogeneous mixture
was stirred at room temperature under a nitrogen atmosphere for 12 h,
and the solid separated by filtration and washed with CH2Cl2 (3 × 20
mL). The solid was dissolved in warm THF, and the solution then
concentrated to 30 mL under reduced pressure to allow for crystaliza-
tion, yielding 0.635 g (31%) of the desired diester having Rf 0.53
[CHCl3/acetone, 6/1 (v/v)], mp 213-214 °C dec, and 1H NMR (CDCl3)
δ 8.36 (dd, 1 H), 8.23 (m, 2 H), 8.18 (d, 1 H), 8.00 (dt, 1 H), 7.91 (dd,
1 H), 7.83 (dt, 1 H).
Mixed Disulfide Derived from 2-Nitro-5 mercapto-N,N-dimeth-
ylbenzamide and Glutathione (2). To a suspension of 54.0 mg (0.079
mmol) of BDTNB in 1 mL of anhydrous DMF was added 0.022 mL
of 40% dimethylamine/H2O (0.197 mmol) plus 50 µL of N,N-
diisopropyl-N-ethylamine (0.29 mmol). The reaction mixture was then
stirred for 3 h at room temperature and diluted with 20 mL of CH2Cl2.
Subsequent washing with 3 × 10 mL of saturated aqueous NaHCO3
and 3 × 20 mL of H2O, drying over Na2SO4, concentration under
reduced pressure, and purification by column chromatography (SiO2,
CHCl3/acetone, 6/1, v/v) afforded 35 mg (99%) of 5,5′-dithiobis(2-
nitro-N,N-dimethylbenzamide, having Rf 0.46 and 1H NMR (CDCl3) δ
8.12 (d, 2 H), 7.57 (d, 2 H), 7.40 (s, 2 H), 2.82-3.12 (d, 12 H).
To a solution of 20.0 mg of 5,5′-dithiobis(2-nitro-N,N-dimethyl-
benzamide) (44.4 µmol) in 1.4 mL of MeOH was slowly added a
solution of 13.6 mg of glutathione (44.3 µmol) in 0.4 mL of H2O and
the mixture then stirred under an argon atmosphere for 1 h at room
temperature. The mixture was then concentrated under reduced pressure
and the residue purified by flash chromatography (SiO2,CHCl3/MeOH/
H2O; 60/40/10; v/v/v) to give 8.1 mg (60%) of 2 having Rf 0.36 and
1H NMR (CD3OD) d 8.22 (d, 1 H), 7.82 (m, 1 H), 7.60 (s, 1 H), 4.66
(m, 1 H), 3.60-3.65 (m, 3 H), 2.80-3.11 (d, 6 H), 2.56 (m, 2 H), 2.14
(m, 2 H), 1.36 (m, 2 H).
N2,N2′-Bis[N1,N3-dicholeamidospermidineyl]-5,5′-dithiobis(2-ni-
trobenzamide) (4). To a stirred solution of 100 mg (0.108 mmol) of
N1,N3-dicholeamidospermidine (3) in 0.8 mL of anhydrous DMF and
0.1 mL of triethylamine was added 33.0 mg (0.048 mmol) of BDTNB.
The mixture was then stirred at room temperature under a nitrogen
atmosphere for 14 h, and the solvents removed under reduced pressure
(45 °C). The crude product was dissolved in 2 mL of methanol and
purified by precipitating into 35 mL of 10% NaHCO3. The solid was
separated, washed with water (3 × 20 mL), and purified by preparative
1
and H NMR (D2O) δ 8.28 (m, 2 H), 7.86 (m, 2 H), 7.46 (bs, 2 H),
4.64 (s, 4 H), 4.45 (s, 4 H), 4.18 (m, 4 H), 3.35 (m, 8 H), 3.16 (m, 8
H), 0.92-2.35 (m, 132 H), 0.73 (s, 12 H).
Mixed Disulfide of GSH and Persulfated N1,N3-Bis[dicholeamido-
spermidineyl]-N2-(2-nitro-5-mercaptobenzamide) (1b). To a stirred
solution of 50.0 mg (14.5 µmol) of 5 in 0.80 mL of MeOH was added
0.84 mL of an aqueous solution of 4.46 mg (14.5 µmol) of glutathione
(pH of which was adjusted to 7.16 by 1 N NaOH). After being stirred
for 60 min under an argon atmosphere, the solution was freeze-dried,
and the resulting solid was redisolved in 0.5 mL of MeOH/H2O (5/
2,v/v). The crude product was purified by preparative TLC (SiO2;
CHCl3/MeOH/H2O (34/28/8,v/v/v) under an argon atmosphere [solvents
were removed by a combination of rotatary evaporation (35 °C, 10
Torr) and freeze-drying] to give 21 mg (72%) of 1b with Rf 0.45 and
1H NMR (D2O) δ 8.26 (d, 1 H), 7.87 (d, 1 H), 7.53 (d, 1 H), 4.71 (m,
1 H), 4.68 (s, 2 H), 4.43 (s, 2 H), 3.91 (m, 2 H), 3.76 (t, 1 H), 3.50 (m,
2 H), 2.80-3.35 (m, 10 H), 2.42 (t, 2 H), 3.30-0.90 (m, 68 H), 0.71
(s, 6 H). Anal. Calcd for C72H107N7O35S8Na8‚10 H2O: C, 38.44; H,
5.60; N, 4.36. Found: C, 38.06; H, 5.52; N, 4.24.
Vesicles for Transport Experiments. Typically, 1.6 mL of a 25
mg/mL solution of POPC in chloroform was transferred to a Pyrex
test tube, and the solvent evaporated by rotating the tube under a stream
of nitrogen, resulting in a thin film of POPC. The last traces of solvent
were removed under reduced pressure (25°, 48 h, <0.2 Torr). To the
dried film was added 1.75 mL of a 1.20 mM solution of glutathione
(GSH) that was prepared in borate buffer, which had been adjusted to
pH 7.0 by addition of a few drops of 1 M NaOH. The mixture was
vortexed for 30 s, incubated at room temperature for 5 min, vortexed
for another 30 s, and incubated for 30 min, while being maintained
under an argon atmosphere. The dispersion was then subjected to 5
freeze/thaw cycles (-196 °C/30 °C), followed by extrusion through a
400 nm Nuclepore membrane (12 times) and a 200 nm membrane (15
times).
Gel Filtration Procedure. All gel filtrations were carried out via
minicolumn centrifugation using pre-packed Sephadex G-25, PD-10
(Pharmacia Biotech). Prior to use, the Sephadex G-25 was extensively
rinsed with borate buffer. Typically, 1.5 mL of the dispersion were gel
filtered (two times) using 4.0 g of Sephadex G-25 (fine), by expelling
the void volume of the column via centrifugation (4 min). To ensure
the complete removal of external GSH, the dispersion was dialyzed
using a bag made from SpectraPor #7 dialysis tubing (MWCO 50 000),
(15) A preliminary account of this work has previoiusly appeared: Janout,
V.; DiGiorgio, C.; Regen, S. L. J. Am. Chem. Soc. 2000, 122, 2671.

Transport of Glutathione Across a PhospholipidMembrane
J. Am. Chem. Soc., Vol. 123, No. 23, 2001 5403
where dialysis was made against 400 (1 h), 600 (20 h), and 500 mL (3
h) of borate buffer under a nitrogen atmosphere.
specifically designed as synthetic targets for this study. Our
choice of 1a was based on five considerations. First, glutathione
(GSH) was viewed as a reasonable drug model based on its
strong hydrophilicity. Thus, one would not expect GSH to
readily cross lipid bilayers. The fact that mammalian cells
Reaction of Vesicle-Entrapped GSH with 1a. In a typical
experiment, 200 µL of the vesicle dispersion was mixed with 300 µL
of a 20 µM solution of 1a in borate buffer. Mixing was carried out in
the “source” side of a 1.5 mL equilibrium dialysis cell containing an
equal volume of borate buffer in the “receiving” side (SpectraPor #7
dialysis tubing, MWCO 50 000), using a Burrell wrist-action shaker.
The extent of thiolate-disulfide interchange was then monitored in
the source side by the appearance of USH at λmax 427. Pseudo-first-
order rate constants, kobsd, were determined from linear plots of ln[1 -
(C/Co)]USH versus time. Analysis of the residual GSH in the source at
the end of the reaction was made by diluting a 30 µL aliquot of the
dispersion with 360 µL of a 10% buffer solution of sodium dodecyl
sulfate (SDS) and then adding 20 µL of a 2 mM buffer solution of
5,5′dithiobis(2-nitrobenzoic acid) (Ellman’s reagent). Measurement of
the UV spectrum (350-500 nm) was recorded after 5 min, and
compared with appropriate calibration curves. Note: Care must be taken
in subtracting the absorbance due to USH from the absorbance produced
upon reaction with Ellman’s reagent. Analysis of the receiving side of
the dialysis cell for GSH was made by withdrawing a 30 µL aliquot
and reacting it with Ellman’s reagent, in a manner that was similar to
that used for the source side.
Analysis of Oxidized Glutathione (GSSG). For analysis of GSSG
from the source side, 400 µL of the dispersion were diluted with 300
µL of ethanol in a 12-mL test tube. To this solution was added 60 µL
of a 1.0 mM buffer solution of tris(2-carboxyethyl)phosphine (TCEP),
followed by vortex mixing. After allowing this mixture to react for 24
h under an argon atmosphere, 50 µL of 6 M HCl was added, followed
by addition of 50 µL of an aqueous solution that was 4 mM in 4,4′-
dipyridyl disulfide (Aldrithiol-4) and 1 M in HCl. The UV spectrum
(300-380 nm) was then recorded after 5 min of reaction; 4-mercap-
topyridine has λmax 330 nm (ꢀ ) 1.819 × 103 M-1 cm-1, borate buffer).
A baseline that was used in this analysis was obtained from an
analogous sample, where the addition of TCEP was omitted. These
results were also compared with a blank determination of TCEP, in
which the vesicles were devoid of GSSG, i.e., vesicles containing
entrapped GSH, which were not subjected to reaction with 1a. Analysis
of GSSG in the receiving side (after 8 h of reaction with 1a) was made
by use of a standard fluorescamine assay.16,17
Affinity of 1a and 1b toward POPC Membranes. Multilamellar
vesicles of POPC were prepared using 40 mg of lipid (dried thin film)
plus 2 mL of borate buffer via vortex mixing. To this dispersion was
added 6 nmol of 1a or 1b in 0.3 mL of borate buffer. After incubation
for 120 min at room temperature, the vesicles were pelleted by
centrifugation. The percentage of the umbrella that remained in the
supernatant was then quantified by adding 0.6 µmol of TCEP in 120
µL of buffer and analyzing the solution for the reduced form of the
conjugate (USH) by UV at λmax 427 nm.
maintain millimolar concentrations of GSH within their cyto-
plasm, and that only micromolar levels of GSH can be detected
in the extracellular milieu, fully supports this premise. Second,
glutathione has the potential for being covalently attached to,
as well as released from, a molecular umbrella by taking
advantage of thiolate-disulfide interchange chemistry. Third,
the 5-thiol(2-nitrobenzoyl) “handle” provides a sensitive means
for monitoring the release of GSH through its strong UV
absorption. Fourth, previous studies from our laboratory have
shown that di-walled molecular umbrellas derived from cholic
acid and spermidine exhibit molecular amphomorphism, which
is a likely prerequisite for membrane transport.11,12 Fifth,
examination of 1a by CPK molecular models (not shown)
indicates that such an umbrella can provide significant coverage
for an attached glutathione molecule.15
Surface Pressure-Area Isotherms. Surface pressure-area iso-
therms were recorded by use of an MGW Lauda film balance, which
was equipped with a computerized data acquisition station. All
isotherms were measured at 25 °C. Water (ca. 1 L), which was used in
preparing a subphase, was purified via a Milli-Q filtration system and
purged with nitrogen for 15 min. Before addition to the film balance,
the surface of the aqueous solution was removed via aspiration to
eliminate surface-active contaminants. A chloroform solution of 1a was
spread onto the aqueous subphase having a surface area of 600 cm2,
using a gastight, 50 µL Hamilton syringe. Actual concentrations were
determined by direct weighing of aliquots after evaporation of solvent
using a Cahn 35 electrobalance. In all cases, spreading solvents were
allowed to evaporate for at least 30 min prior to compression under a
flow of nitrogen. Monolayers were compressed at the rate of 25 cm2/
min.
Our interest in the hexasulfate analogue of 1a (i.e., 1b) was
derived largely from mechanistic considerations. Specifically,
if 1a were to enhance the transport of glutathione across a POPC
bilayer, a question that could be asked is whether this enhanced
permeability relative to GSH was due to an “umbrella-like
action”, or simply due to a net increase in overall hydrophobic-
ity, resulting from the attached sterols. Since the permeability
coefficient of a permeant is directly proportional to its membrane/
water partition coefficient, an overall increase in hydrophobicity
due to conjugation would increase its membrane permeability.18-20
By replacing all of the hydroxyl groups of the cholic acid
moiety with sulfate groups, the likelihood of diffusion of 1b
across a lipid bilayer in an exposed or a fully exposed
conformation is remote. Thus, in addition to the strong hydro-
philicity that is expected for each of these sterol units, the
Results and Discussion
(18) Ramaswami, V.; Zhu, X.; Romanowski, M.; Haaseth, R. C.; Misicka,
A.; Lipkowski, A. W.; Hruby, V. J.; O’Brien, D. F. Int. J. Peptide Protein
Res. 1996, 48, 87.
(19) Stein, W. D. Transport and Diffusion Across Cell Membranes;
Academic Press: San Diego, CA, 1986; p 685.
Umbrella Design. Two molecular umbrella-glutathione con-
jugates (1a and 1b) and one nonumbrella analogue (2) were
(16) Udenfriend, S.; Stein, S.; Bohlen, P.; Dairman, W. Science 1972,
178, 871.
(17) Imai, K.; Watanabe, Y. Anal. Chim. Acta 1981, 130, 377.
(20) Diamond, J. M.; Katz, Y. J. Membr. Biol. 1974, 17, 121.

5404 J. Am. Chem. Soc., Vol. 123, No. 23, 2001
Janout et al.
Scheme 1
Scheme 2
Scheme 3
conditions used, 94% of 1a and 50% of 1b were removed from
solution by the pelleted vesicles, i.e., 1a exhibited significantly
greater affinity toward POPC bilayers relative to 1b.
Monolayer Properties of 1a at the Air-Water Interface.
To gain insight into the likely conformation of 1a, when bound
to a lipid membrane, we examined its monolayer behavior in
the absence and in the presence of POPC. This conjugate was
chosen for these experiments because of its lower water-
solubility, which decreases its tendency to dissolve in the
aqueous subphase.23
presence of six sulfate groups should strongly inhibit transbilayer
movement by anchoring the umbrella at the membrane-water
interface. It is noteworthy, in this regard, that the half-life for
the translocation of phosphatidylgycerols (bearing a single
negative charge in the headgroup) between the inner and outer
monolayer leaflet of a lipid bilayer is on the order of days to
weeks.21 Thus, the transport of an umbrella molecule such as
1b, bearing six ionized sulfate groups, is expected to be
considerably slower. If the molecule were to adopt a shielded
conformation, however, then much, if not all, of its hydro-
philicity could be masked, and movement across the bilayer
would be facilitated. The observation of movement of 1b across
a lipid bilayer, therefore, would constitute strong inferential
evidence for an umbrella-like action by the conjugate.
Finally, a nonumbrella analogue (2) was deemed worthy of
investigation, to serve as a frame of reference in which to judge
the effects of umbrella attachment on membrane transport.
Umbrella Synthesis. The synthetic strategy that was used
to prepare 1a is outlined in Scheme 1. Thus, cholic acid was
first activated by forming the corresponding N-hydroxy suc-
cinimide ester, and then condensed with spermidine to give the
di-walled molecular umbrella 3. Subsequent reaction with
bis{3-O-[N-1,2,3-benzotriazin-4(3H)-one]yl}-5,5′dithiobis-2-
nitrobenzoate (BDTNB) afforded the corresponding symmetrical
disulfide-based dimer (4), which was then cleaved with GSH
to give the requisite umbrella-glutathione conjugate, 1a.22
The hexasulfate analogue of 1a was obtained by treating 4
with Pyr‚SO3 to give 5, followed by nucleophilic displacement
with GSH (Scheme 2). Finally, the nonumbrella analogue (2)
was synthesized by reacting BDTNB with dimethylamine to
give the corresponding bisamide, and then subjecting it to
displacement by GSH (Scheme 3).
In Figure 1 is shown the surface pressure-area isotherm that
was obtained for 1a, spread over an aqueous solution that was
0.14 M in NaCl and 0.01 M in phosphate buffer (pH 7.0) at 25
°C. Here, three distinct regions can be seen: a highly compress-
ible region between 0 and 21 mN‚m-1, a very broad first-order
phase transition region, and a region of low compressibility.
Extrapolation of the condensed portion of the highly compress-
ible region to zero surface pressure yields a limiting area of
3.03 nm2‚molecule-1. This value is consistent with a conforma-
tion in which the hydrophilic face of each sterol and the attached
peptide are in intimate contact with water,; i.e., the entire
molecule lies flat on the aqueous subphase. The high compress-
ibility within this region can then be accounted for by a model
in which the umbrella “folds up” into a shielded conformation
as the area per molecule is reduced, i.e., both sterols lift up
into air, and encase the pendant GSH, the resulting structure
being anchored to the aqueous subphase via the C-3 hydroxyl
group of each sterol.
The appearance of the broad first-order phase transition,
together with the low surface area that 1a occupies at the end
of this transition (just before reaching the less compressible
region), implies that a radical change in umbrella conformation
has taken place. Such behavior is similar to that which we have
recently reported for a conjugate derived from spermine and
cholic acid.24 One plausible explanation, which could account
for this behavior, is that one of the sterol units has “flipped” up
into air or down into water.
(23) Approximate water solubilities were estimated by injecting a small
volume of a 34 mM methanolic solution of each conjugate into 0.750 mL
of borate buffer (pH 7) and determining the concentrations at which the
solutions became nontransparent at room temperature. In each case, the
methanol content in the final solution was <2.5% (v/v). For 1a, transparency
was lost when the concentration reached ca. 400 µM; for 1b, the solution
remained transparent at concentrations >34 mM, which was the highest
concentration tested. As expected, based on the greater hydrophilicity of
1b relative to 1a, the hexasulfate analogue exhibited substantially greater
water-solubility.
Relative Affinity toward POPC Bilayers. The relative
affinities of 1a and 1b toward bilayers of POPC were determined
by adding each conjugate to multilamellar vesicles (MLVs) of
the phospholipid under identical conditions. After a 120-min
incubation period, the MLVs were pelleted by centrifugation
and the supernatant analyzed for free conjugate. Under the
(21) Ganong, B. R.; Bell, R. M. Biochemistry 1984, 23, 4977.
(22) Janout, V.; Lanier, M.; Regen, S. L. Tetrahedron Lett. 1999, 40,
1107.
(24) Bandyopadhyay, P.; Janout, V.; Zhang, L. H.; Sawko, J. A.; Regen,
S. L. J. Am. Chem. Soc. 2000, 122, 12888.

Transport of Glutathione Across a PhospholipidMembrane
J. Am. Chem. Soc., Vol. 123, No. 23, 2001 5405
Scheme 4
chemical reaction is slow relative to permeation across the lipid
bilayer, then the observed half-life represents an upper limit
for the half-life of membrane transport.
Experimentally, large unilamellar vesicles of POPC (200 nm
diameter) were prepared in borate buffer (pH 7.0) containing
1.2 mM GSH via extrusion.26 Nonentrapped GSH was then
removed by gel filtering the dispersion. To ensure the complete
removal of nonentrapped GSH, the dispersion was then sub-
jected to dialysis. In a typical transport experiment, 200 µL of
a 26 mM dispersion of POPC liposomes was mixed with 300
µL of a 20 µM solution of 1a in borate buffer. Here, a
submicellar concentration of the conjugate was used to simplify
the interpretation of the results.27 Mixing was then carried out
in the “source” side of a 1.5-mL equilibrium dialysis cell, which
contained an equal volume of borate buffer in the “receiving
”side. [Note: Scheme 4 shows the chemistry that is expected
for this umbrella-assisted transport experiment. Although an
equilibrium dialysis cell was used to confirm that GSH did not
leak out of the vesicles, and also that all of the GSSG remained
entrapped in the vesicles, it has not been included in this scheme
for simplicity.] Agitation of the cell was made by use of a wrist-
action shaker. The extent of thiolate-disulfide interchange was
monitored in the source side by measuring the appearance of
the strongly UV absorbing, reduced form of the umbrella, USH
(λmax 427 nm; ꢀ 7.007 × 103 M-1 cm-1; borate buffer). An
analogous experiment that was performed using a higher
concentration of entrapped GSH (i.e., 2.0 mM) resulted in a
proportional increase in the observed pseudo-first-order rate
constant (Figure 2). These results indicate that the rate of
formation of GSH is determined by chemical reaction and not
by permeation across the POPC bilayer, meaning that diffusion
across the membrane is relatively fast (eq 1), i.e., the half-life
for membrane transport must be ,22 min.
Figure 1. Surface pressure-area isotherms for 1a and mixtures of 1a
plus POPC on the surface of 0.1 M NaCl plus 0.01 M Na2HPO4/
NaH2PO4 solution (pH 7.0) at 25 °C: (1) 1a; (2) POPC; (3) 5.0% 1a/
95% POPC; (3′) calculated isotherm for 5.0% 1a/95% POPC, assuming
ideal mixing and additivity between 1a and POPC; (4) 9.4% 1a/90.6%
POPC.
In sharp contrast, monolayers formed from POPC, and also
from two different mixtures of POPC/1a (i.e., 95/5 mol/mol
and 90.6/9.4 mol/mol), produced isotherms of low compress-
ibility with no discernible phase transition (Figure 1). If one
assumes that POPC and 1a mix ideally, then the calculated
isotherm for the POPC/1a (95/5, mol/mol) mixture (from 0 to
20 mN‚m-1, based on simple additivity) is in excellent agree-
ment with what is observed. This finding implies that 1a adopts
the same conformation as that found in monolayers made from
the conjugate itself, within this range of surface pressures. In
addition, the absence of a distinct phase transition implies that
this compressed structure, which we assign to a shielded
conformation, is stabilized in these POPC-rich films. Since the
internal surface pressure of fluid phospholipid bilayers is
generally thought to be ca. 30 mN‚m-1, these monolayer results
are fully consistent with a shielded umbrella conformation of
1a in POPC bilayers.25
Umbrella-Assisted Transport of Glutathione. One of the
more common experimental methods that has been used to
quantify transport rates of water-soluble permeants across
phospholipid bilayers has been to encapsulate the permeant
within liposomes, and to then measure their efflux rates.18
Because of the high affinity of 1a, and to a lesser extent 1b,
toward POPC membranes, we chose a different protocol for
this study. Specifically, we loaded the aqueous interior of POPC
liposomes with glutathione and then added 1a or 1b, externally,
to the bulk aqueous phase. In principle, if the conjugate crosses
the bilayer without disturbing it, one should then observe (i)
the formation of oxidized glutathione (GSSG) within the interior
of the Vesicle, (ii) the appearance of the thiol form of the
umbrella (USH), and (iii) the absence of release of GSH into
the external aqueous phase (Scheme 4).
USSG + GSH f USH + GSSG
(1)
To ensure that GSH did not leak out of the liposomes and
promote cleavage of the conjugate on the vesicle exterior, the
receiving side of the cell was analyzed for GSH (Ellman’s
reagent) after a 24 h period. This analysis showed that less than
0.1 µM was present. Since the half-life for permeation of GSH
across the dialysis membrane is 200 min, complete release of
GSH would have corresponded to a thiol concentration of ca.
28.5 µM in both the source and the receiving side of the cell
after 24 h. In addition, an analysis of the source side of the cell
for residual thiol content, after the liposomes were first destroyed
with excess sodium dodecyl sulfate, confirmed the presence of
22.6 nmol of GSH. Thus, the mass balance is excellent since
complete reaction of GSH with 1 would, theoretically, leave
23.5 nmol of the tripeptide within the vesicles.
An additional reason that we were interested in carrying out
membrane transport experiments in this manner was because it
simulates what could be a drug delivery process. Specifically,
one can envision taking advantage of intracellular glutathione
to cleave an umbrella-drug conjugate on the inner surface of
a cell membrane, thereby releasing the drug into the cytoplasm.
The umbrella-drug conjugate would then function as a prodrug.
One disadvantage of this approach, however, is that it relies on
a chemical reaction to detect membrane transport. If the rate of
(26) Hope, M. J.; Bally, M. B.; Cullis, P. R. Biochim. Biophys. Acta
1985, 812, 55.
(27) The critical micelle concentration of 1a in buffer is 60 µM (surface
tension method). For 1b, the cmc is .1000 µM, which was the highest
concentration tested; the cmc for 2 is 180 µM.
(25) Georgallas, A.; Hunter, D. L.; Lookman, T.; Zuckerman, M. J.; Pink,
D. A. Eur. Biophys. J. 1984, 11, 79.

5406 J. Am. Chem. Soc., Vol. 123, No. 23, 2001
Janout et al.
the POPC vesicles and reacts with the GSH to produce entrapped
GSSG.
Analogous transport experiments that were carried out with
the hexasulfate, 1b, using similar experimental conditions as
that employed for 1a, gave similar results (Figure 2). Again,
umbrella entry was confirmed by the appearance of USH in
the vesicle dispersion, the formation of entrapped GSSG, and
the absence of release of GSH into the external aqueous phase.
The only notable differences were that the rate of appearance
of USH was signficantly less than that observed with 1a, and
this rate was unchanged when the entrapped GSH concentration
was increased from 1.2 to 2.0 mM. The insensitivity of the rate
of formation of USH to the concentration of entrapped GSH
indicates that chemical reaction is not rate-limiting. At present,
we can only speculate on what may be rate-controlling. Two
possibilities that appear likely are partial dehydration of the
sulfate groups upon entry into the lipid membrane and diffusion
across the bilayer. Nonetheless, the ability of 1b to cross POPC
bilayers and to react with internalized GSH is signficant. It
provides inferential evidence that this conjugate is crossing the
bilayer in a shielded conformation, which minimizes exposure
of its strongly hydrophilic sulfate groups (vide ante).
Addition of 2 to liposomal dispersions of POPC that contained
entrapped GSH (1.2 mM), using experimental conditions that
were similar to those used for 1a and 1b, did not produce a
detectable reaction after 5 h. This result further demonstrates
the important role that the sterols in 1a and 1b have in promoting
the transport of the attached glutathione molecule across POPC
bilayers.
Figure 2. Rate of appearance of USH for reaction of 1a with 0.2-µm
POPC vesicles containing 1.2 mM GSH (9) and 2.0 mM GSH (b) at
23 °C; the inset shows the pseudo-first-order rate constant, kobsd, versus
the internal concentration of GSH. Analogous reactions carried out with
1b using 1.2 (O) and 2.0 mM GSH (×).
To obtain further evidence for umbrella entry, as is depicted
in Scheme 4, we analyzed both the source and receiving side
of the dialysis cell for GSSG after a 28 h reaction period. The
question here was whether all of the oxidized glutathione, which
was formed from the displacement reaction, was entrapped
within the vesicles. It should be noted that GSSG crosses the
dialysis membrane with a half-life of ca. 8 h. Analysis for GSSG
in the source side was performed by a combination of disulfide
reduction via excess tris(2-carboxyethyl)phosphine (TCEP) and
back-titration with 4,4′-dipyridyl disulfide (Aldrithiol-4), i.e.,
the excess of TCEP was quantified by reaction with Aldrithiol-
4. To prevent GSH, itself, from reacting with the Aldrithiol-4
reagent, strongly acidic conditions were used. Thus, when 12
µM (6 nmol) 1a and 57 mM (28.5 nmol) vesicle-entrapped GSH
were introduced into the source side, and the vesicles allowed
to react for 28 h, subsequent analysis of the source side revealed
the presence of 10.9 µM (5.45 nmol) GSSG and 45.2 µM (22.6
nmol) GSH. A separate analysis of the receiving side for GSSG,
using a more sensitive fluorescamine assay, indicated the
presence of less than 0.24 µM (0.12 nmol) peptide. As a final
control, GSSG was added, externally, to the vesicle dispersion
(corresponding to 10 µM of external GSSG in the source side),
and the receiving side analyzed for GSSG. After 8 h of dialysis,
the concentration of GSSG that appeared in the receiving side
was 2.2 µM. Thus, adsorption of GSSG to the lipid membrane
is negligible. Taken together, these results establish that 1a enters
Conclusions
The results reported herein provide clear evidence that
molecular umbrellas can transport polar peptides such as
glutathione across POPC bilayer membranes. Monolayer results
that have been obtained with 1a, in the absence and in the
presence of POPC, together with the fact that a hexasulfate
analogue (1b) can also cross POPC bilayer membranes,
constitute strong inferential evidence for an “umbrella-like”
action of these molecules during the transport process. In a
broader context, these findings provide a solid foundation (i.e.,
proof-of-principle) for the molecular umbrella concept for the
transport of polar molecules across lipid bilayers.
Efforts currently in progress are aimed at synthesizing
appropriate umbrella-peptide and umbrella-antisense oligo-
nucleotide conjugates for in vitro studies.
Acknowledgment. We are grateful to the National Institutes
of Health (PHS Grant GM51814) for support of this research.
JA010124R