Supramolecular aggregates of azobenzene phospholipids and related compounds in bilayer assemblies and other microheterogeneous media: Structure, properties, and photoreactivity

Article abstract of DOI:10.1021/ja971291n

A number of phosphatidyl choline derivatives containing trans-azobenzene units in the fatty ester backbone have been synthesized and studied in aqueous dispersions both pure and in the presence of saturated and unsaturated phospholipids. The structures of

Full text of DOI:10.1021/ja971291n

9144
J. Am. Chem. Soc. 1997, 119, 9144-9159
Supramolecular Aggregates of Azobenzene Phospholipids and
Related Compounds in Bilayer Assemblies and Other
Microheterogeneous Media: Structure, Properties, and
Photoreactivity1
Xuedong Song, Jerry Perlstein, and David G. Whitten*
Contribution from the Department of Chemistry and NSF Center for Photoinduced Charge
Transfer, UniVersity of Rochester, Rochester, New York 14627
ReceiVed April 23, 1997X
Abstract: A number of phosphatidyl choline derivatives containing trans-azobenzene units in the fatty ester backbone
have been synthesized and studied in aqueous dispersions both pure and in the presence of saturated and unsaturated
phospholipids. The structures of the assemblies formed have been investigated by microcalorimetry, dynamic light
scattering, cryo-transmission electron microscopy, and reagent entrapment. While many of the mixed phospholipid
dispersions give evidence for the formation of small unilamellar vesicles, the aqueous dispersions of pure azobenzene
phospholipids (APL’s) give evidence for several different structures, including relatively large plates in at least one
case. The azobenzenes show strong evidence of “H” aggregate formation both in the pure and mixed dispersions.
The aggregation number has been estimated for several of the APL’s and found to be typically 3 or a multiple
thereof. On the basis of simulations and studies with similar stilbene phospholipids as well as on the strong induced
circular dichroism signals observed for the aggregate, we infer a chiral “pinwheel” unit aggregate structure similar
to that found for several aromatics. The azobenzenes in the aqueous dispersions have been found to photoisomerize
to give cis-rich photostationary states; the cis-azobenzenes show no evidence for aggregation and no induced circular
dichroism. The cis-azobenzenes can be isomerized back to the trans either by irradiation or by thermal paths. Mixed
aqueous dispersions of trans-APL’s with saturated or unsaturated phospholipids can be prepared which entrap the
fluorescent dye carboxyfluorescein (CF) under conditions where the CF fluorescence is very low due to self-quenching.
By varying the APL/host phospholipid ratio the azobenzene can be aggregate, monomer, or dimer. In cases where
the azobenzene is monomer or dimer, irradiation produces complete isomerization but little “leakage” of CF from
the vesicle interior. In contrast, where the azobenzene is predominantly aggregate, irradiation results in both
photoisomerization and reagent release. That photoisomerization in the latter case can result in “catastrophic”
destruction of the vesicle can also be shown by cryo-transmission electron microscopy.
Introduction
for control of macroscopic properties such as volume,6 wetta-
bility,11 membrane permeability,12 solubility and viscosity in a
number of applications. The most likely explanation for the
differences between azobenzene and stilbene photoisomerization
is the occurrence of an “inversion” mechanism4,5 in the case of
the azobenzenes which is not available for the stilbenes and
alkenes.
In recent investigations we and others have found that
amphiphilic stilbene17 and azobenzene derivatives18 as well as
a number of other functionalized amphiphilies19,20 containing
an aromatic or dye chromophore tend to form relatively stable
aggregates when dispersed in aqueous media or incorporated
The thermal and photoisomerization reactions of azobenzenes
have been found to occur in a wide variety of media in a process
formally analogous to the much more widely studied (from a
mechanistic perspective) photoisomerization of the stilbenes and
related olefins.2-5 In contrast to the stilbenes, whose photoi-
somerization shows a high medium sensitivity, azobenzene
photoisomerization has been found to occur even in very viscous
solutions, liquid crystals, and rigid solids.6-16 This has led to
the use of azobenzene derivatives as a photoresponsive “trigger”
X Abstract published in AdVance ACS Abstracts, September 15, 1997.
(1) A portion of this research has appeared as a communication: Song,
X.; Perlstein, J.; Whitten, D. G. J. Am. Chem. Soc. 1995, 117, 7816.
(2) Saltiel, J.; Charlton, J. L. In Rearrangements in Ground and Excited
State; de Mayo, P., Ed.; Academic Press, New York, 1980; Vol. 3.
(3) Siapiringue, N.; Guyot, G.; Monti, S.; Bortulis, P. J. Photochem. 1987,
37, 185.
(12) Okahata, Y.; Fujita, S.; Iizuka, N. Angew. Chem., Intl. Ed. Engl.
1986, 25, 751.
(13) Bortolus, P.; Monti, S. J. Phys. Chem. 1987, 91, 5046.
(14) Shinkai, S.; Matsuo, K.; Harada, A.; Manabe, O. J. Chem. Soc.,
Perkin Trans. II 1982, 1261.
(15) Hentze, F. Z. Chem. 1977, 17, 294.
(4) Hammond, G.; Saltiel, J.; Lamola, A. A.; Turro, N. J.; Bradashaw,
G. S.; Cowan, D. O.; Counsell, R. S.; Vogt, V.; Dalton, C. J. Am. Chem.
Soc. 1964, 86, 3197.
(5) Jones, L. B.; Hammond, G. J. Am. Chem. Soc. 1965, 87, 4219.
(6) (a) Rau, H.; Liiddecke, E. J. Am. Chem. Soc. 1982, 104, 1616. (b)
Rau, J. Photochem. 1984, 26, 21.
(7) Monti, S.; Orlandi, G.; Palmieri, P. Chem. Phys. 1982, 71, 87.
(8) Eisenbach, C. D. Ber. Bunsen-Ges. Phys. Chem. 1980, 84, 680.
(9) Yamamoto, A. Macromolecules 1986, 19, 2472.
(10) Lamarre, L.; Sung, C. S. P. Macromolecules 1983, 16, 1729.
(11) (a) Ishihara, K.; Namada, N.; Kato, S.; Shinohara, I. J. Polym. Sci.,
Polym. Chem. Ed. 1984, 22, 121. (b) Ishihara, K.; Namada, N.; Kato, S.;
Shinohara, I. J. Polym. Sci., Polym. Chem. Ed. 1983, 21, 1551.
(16) Haberfield, P. J. Am. Chem. Soc. 1987, 109, 6177.
(17) (a) Whitten, D. G. Acc. Chem. Res. 1993, 26, 502. (b) Song, X.;
Geiger, C.; Furman, I.; Whitten, D. G. J. Am. Chem. Soc. 1994, 116, 4103.
(c) Song, X.; Geiger, C.; Leinhos, U.; Perlstein, J.; Whitten, D. G. J. Am.
Chem. Soc. 1994, 116, 10340.
(18) (a) Heesemann, J. J. Am. Chem. Soc. 1980, 102, 2167. (b) Fukuda,
K.; Nakahara, H. J. Colloid Interface Sci. 1980, 98, 555. (c) Kunitake, T.
Angew. Chem., Int. Ed. Engl. 1992, 31, 709.
(19) (a) Chen, H.; Farahat, M.; Law, K. Y.; Perlstein, J.; Whitten, D. G.
J. Am. Chem. Soc. 1996, 118, 2586. (b) Chen, H.; Law, K. Y.; Perlstein, J.;
Whitten, D. G. J. Am. Chem. Soc. 1995, 117, 7257.
(20) Farahat, C. W.; Penner, T. L.; Ulman, A.; Whitten, D. G. J. Phys.
Chem. 1996, 100, 12616
S0002-7863(97)01291-2 CCC: $14.00 © 1997 American Chemical Society

Aggregates of Azobenzene Phospholipids
J. Am. Chem. Soc., Vol. 119, No. 39, 1997 9145
Melting points were taken on an El-Temp II melting point apparatus
and are uncorrected. The pH values were measured with an Orien
Research digital ionalyzer/501 pH meter. Proton NMR spectra were
recorded on a General Electric QE300 MHz spectrometer using
deuterated solvent locks or on a 500 MHz Varian VXR-500S
spectrometer. FAB mass spectra were measured at the Midwest Center
for Mass Spectrometry. Absorption spectra were obtained on a Hewlett-
Packard 8452A diode array spectrophotometer. Fluorescence spectra
were recorded on a SPEX Fluorolog-2 spectrofluorometer and were
not corrected. The circular dichroism (CD) study was carried out on
a JASCO J-710 spectropolarimeter. Differential scanning calorimetry
(DSC) measurements were carried out on a Mc-2 Ultrasensitive
Scanning Calorimeter from MicroCal, Inc. Size extrusion experiments
for vesicles were performed with an extruder through CoStar Nucle-
opore Polycarbonate Filters. Dynamic light-scattering measurements
were carried out at the Eastman Kodak Research Laboratories.25 All
samples were routinely filtered through a 1.2 µm nylon syringe filter
before data acquisition. Cryo-transmission electron microscope mea-
surements (Cryo-TEM) were made either at the Eastman Kodak
Research Laboratories26 or at the laboratory of NSF Center for
Interfacial Engineering, Department of Chemical Engineering, Uni-
versity of Minnesota.26 A 200 W Mercury lamp (Oriel) was used for
irradiation. The 365 nm line was separated through an interference
filter (365 BP 10, T ) 217). An Oriel tungsten lamp (Model 6130)
was used to produce intense visible light.
into monolayer films at the air-water interface. Although it
was initially thought that these aggregates were formed as a
consequence of self-organization of the functionalized am-
phiphiles into a relatively ordered structure, more extensive
investigations have shown that the aggregates are stabilized by
strong noncovalent aromatic-aromatic interactions and tend to
form readily in the presence of water, even before major
amphiphile organization has occurred.21 For several aromatics
and dyes the key species in forming these aggregates is a “unit
aggregate” which is in many cases a trimer or tetramer, which
has been proposed to be a chiral “pinwheel” structure based
both on simulations and experimental evidence.1,17,21 Our
studies suggest that the extended aggregates in several as-
semblies are probably best described as a mosaic or lattice of
the small “unit” aggregates. Aggregation has been found
especially prominent when the functionalized amphiphiles are
incorporated into phospholipid bilayers, either as “guests”
solubilized within the bilayer generated by a “host” phospholipid
or when modified phospholipids containing the chromophore
are themselves dispersed in aqueous media. Since saturated
phospholipids and structurally related amphiphiles form small
unilamellar closed spherical vesicles upon dispersion and probe
sonication in aqueous solution which can entrap reagents, a
number of investigations22-24 have sought to prepare vesicles
containing a photoreactive component which can be used to
promote a controlled release of entrapped reagents in various
applications including drug delivery.
In the present paper we report a study of the assemblies
formed from several synthetic azobenzene derivatized phos-
pholipids (APL’s) in aqueous media, their structures, physical
properties, and photochemical behavior. Results of the study
of these compounds provide some insights into the importance
of aggregation in controlling both the microstructure as well as
the macroscopic properties of the assemblies. Some of the key
findings include the observation that the trans-APL’s do not
form closed spherical bilayer vesicles upon dispersion in water
but rather form a variety of different structures, including large
plates or sheets in some cases, presumably due to resistance to
cuvature in the extended bilayer. We find that closed vesicles
capable of entrapping reagents can be formed when mixtures
of trans-APL’s are codispersed with saturated and unsaturated
phospholipids and that, depending upon the state of the
incorporated azobenzene as well as the host, controlled release
of entrapped reagents may be promoted by photolysis.
The general methods used for preparing Langmuir-Blodgett films
and self-assemblies are based on techniques described by Kuhn et al.27
Monolayers of chromophore-derivatized fatty acids or phospholipids
were prepared by spreading a chloroform solution of material to be
studied onto an aqueous surface containing cadmium chloride (2.5 ×
10-4 M) and sodium bicarbonate (3 × 10-5 M, pH ) 6.6-6.8) on a
KSV 5000 automatic film balance at room temperature (23 °C). The
monolayers were then transferred onto quartz substrates. Reflectance
spectra for monolayers at the air-water interface were recorded with
a SD1000 fiber optics spectrometer (Ocean Optics, Inc.) equipped with
optical fibers, an LS-1 minature tungsten halogen lamp and a CCD
detector.
Bilayer vesicles were prepared according to established protocols.28
A Cell Disrupter W220F from Heat Systems Ultrasonics, Inc. (setting
6.5, 35 W) was used for probe sonication. The vesicles trapping CF
were prepared as following:29 Depending on the specific vesicle systems,
an appropriate amount of DPPC or POPC, and/or APL stock solution
and/or cholesterol was dissolved in a small amount of CHCl3. The
CHCl3 was removed by a nitrogen stream to form a thin film on the
wall of vial, which was then dried under vacuum overnight. An
appropriate amount (usually 1-3 mL) of 0.1 M CF aqueous solution
(pH ) 7.0) was added to the vial and the sample was incubated in a
water bath at 65 °C for 20 min. The mixture was vortexed for 20 min
followed by a 10-15 min sonication until a clear, transparent solution
was obtained. To remove nonentrapped CF, the resulted vesicles were
purified via gel filtration on a Sephedex G-50 column, using Milli-Q
water as the eluent. The vesicle fraction was collected and stored at 4
°C for DPPC vesicles and at room temperature for POPC vesicles.
Experimental Section
Materials and General Techniques. Synthetic reagents were
purchased from Aldrich Chemical Company and used as received unless
otherwise stated. Cholesterol (99%); R-, â-, and γ-cyclodextrins
(99+%); and Triton-X-100 were purchased from Aldrich. Rattlesnake
venom, L-R-glycero-3-phosphorylcholine as the cadmium chloride
complex, Sephedex-50G, L-R-dimyristoylphosphatidylcholine (DMPC,
99+%), D- and L-R-dipalmitoylphosphatidylcholine (DPPC, 99%), and
1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) were ob-
tained from Sigma. All solvents for spectroscopic studies were
spectroscopic grade from Fisher or Aldrich. Milli-Q water was obtained
by passing in-house distilled water through a Millipore -RO/UF water
purification system. Deuterated solvents were purchased from MSD
Isotopes or Cambridge Isotope Laboratories. 5(6)-Carboxyfluorescein
(CF) was obtained from Eastman Chemical.
The vesicles containing CF from gel filtration were diluted with water
to a specific concentration by monitoring the absorption spectra of the
samples. Release of CF was monitored by scanning spectrofluorometer
at 518 nm with excitation at 492 nm. The irradiation of the sample in
a quartz cuvette was made with a mercury lamp with a 365 nm filter.
Once the irradiation was complete, the sample was transferred quickly
(25) The authors thank Dr. Thomas Whitesides at the Eastman Kodak
Research Laboratories for light scattering measurements.
(26) The authors thank Dr. John Minter at the Eastman Kodak Research
Laboratories and Dr. Michael Bench in the Department of Chemical
Engineering, University of Minnesota, for the help with Cryo-TEM studies.
(27) Kuhn, H.; Mo¨bius, D.; Bu¨cher, H. In Physical Methods of Chemistry;
Weissberger, A., Rossiter, B. W., Eds.; Wiley: New York, 1972; Vol. 1,
P577.
(28) (a) Hope, M. J.; Bally, M. B.; Webb, G.; Cullis, P. R. Biophys.
Acta 1985, 55, 812. (b) Saunders, L.; Perrin, J.; Gammock, D. B. J. Pharm.
Pharmacol. 1962, 14, 567.
(29) (a) Weinstein, J. N.; Yoshikami, S.; Henkart, P.; Blumenthal, R.;
Hagins, W. A. Science 1977, 195, 489. (b) Liu, Y.; Regen, S. L. J. Am.
Chem. Soc. 1993, 115, 708. (c) Nagawa, Y.; Regen, S. L. J. Am. Chem.
Soc. 1992, 114, 1668.
(21) Chen, H.; Liang, K.; Song, X.; Samha, H.; Law, K. Y.; Perlstein,
J.; Whitten, D. G. In Micelles, Microemulsions and Monolayers: Science
and Technology; Shah, D., Ed.; Marcel Dekker, Inc: New York, 1995; in
press
(22) Pidgeon, C.; Hunt, C. A. Photochem. Photobiol. 1983, 37, 491.
(23) Morgan, C. G.; Thomas, E. W.; Sanhdu, S. S.; Yianni, Y. P.;
Mitchell, A. C. Biochim. Biophys. Acta 1987, 504.
(24) Kano, K.; Tanaka, Y.; Ogawa, T.; Shimomura, M.; Kunitake, T.
Photochem. Photobiol. 1981, 34, 323.

9146 J. Am. Chem. Soc., Vol. 119, No. 39, 1997
Song et al.
to the spectrofluorometer and the fluorescence intensity of CF as a
function of time was measured.
Chart 1
O
O
Aggregate Size Measurements. A series of samples containing
different concentration of APL and the same concentration of DMPC
(total volume of each sample was the same) was incubated in a water
bath with temperature at or above the Tc for APL vesicles overnight to
make sure that the samples were equilibrated. Another pure DMPC
sample of the same concentration without APL’s was also incubated
in the same water bath as a reference for absorption spectra. The ratio
of [DMPC]/[APL], which should be larger than 5 to meet the
requirement of DMPC as a medium, must be carefully selected so that
the equilibrated solutions contain significant concentration of both
aggregate and dimer to reduce the experimental error. The samples
were transferred as quickly as possible into a warm cuvette in a holder
maintained at constant temperature by the same water bath. Absorption
spectra were recorded immediately. The absorbance contributions of
both aggregates and dimers in equilibrated solutions were obtained by
simple spectral subtraction based on the reasonable assumption that
the pure APL vesicles have only aggregates and those with APL’s
highly diluted in DMPC ([DMPC]/[APL] larger than 80) contain only
dimer. The plot according to eq 6 allows determination of aggregate
sizes.
N
O
O
N
O
O
+
O
NMe3
N
O
P –
O
O
N
Ao8EPC
O
O
N
N
O
O
O
N
+
O
NMe3
O
P
O
O
O –
N
6Ao2EPC
O
O
N
O
N
O
O
+
O
NMe3
N
O
P
O
O
O –
N
Sample Preparation for Cryo-Transmission Electron Microscopy
(Cryo-TEM). The samples for Cyro-TEM were prepared in dim red
light to avoid possible photoisomerization. The vesicles were put into
a controlled environment vitrification system unit where the temperature
was controlled at just above room temperature and the humidity was
nearly 100%. A screened grid was spotted with the sample and blotted
slightly to form a thin layer of liquid containing the vesicle particles.
The grid was then plunged into liquid ethane to vitrify the liquid and
immobilize the vesicles. The sample was transferred under liquid
nitrogen to a Cryo-TEM specimen holder where the sample was inserted
into the TEM. The sample was held at a temperature of -172 °C.
Low dose electron images were taken of various areas of the grid where
there were thin layers of vitrified liquid across the holes in the grid.
Synthesis of Azobenzene Derivatives. The structures and acronyms
of the compounds used in this study are shown in Chart 1. Synthesis
of A4H and 4A4H is identical to the reported procedures.30 The general
synthetic scheme for APL’s is shown in Scheme 1.
Ao8,6Ao2EPC
O
N
N
N
O
O
+
O
O
NMe3
O
P
O
N
O –
4A4EPC
O
N
O
N
OH
OH
Ao8H
O
N
N
O
6Ao2H
4A4EPC. The following method is similar to other procedures
previously reported.30,31 Dry A4H (370 mg) was dissolved in 20 mL
4
of dry methylene chloride. Triethylamine (0.4 mL) was added via
syringe followed by trimethylacetyl chloride (1.2 mL) under a nitrogen
stream. The reaction mixture was stirred for approximately 20 h. The
reaction was monitored by TLC (solvent B: CHCl3/methanol/water )
65/25/4). After the reaction was complete, the solvent was removed
by rotary evaporation and the residue was dried under vacuum in a
dessicator for about 10 h to remove traces of unreacted chloride and
triethylamine.
A4EPC. The preparation procedure is identical to that for 4A4EPC.
Yield: 67%. 1H NMR (500 MHz, CDCl3), δ: 7.83 (d, J ) 8.1, 4 H),
7.87 (d, J ) 8.1, 4 H), 7.50 (m, 6 H), 7.36 (m, 4 H), 5.52 (m, 1 H),
4.40 (m, 3 H), 4.2 (m, 1 H), 4.05 (m, 2 H), 3.85 (s, 9 H), 2.70 (t, J )
7.0, 4 H), 2.40 (t, J ) 7.1, 4 H), 1.97 (m, 4 H). FAB: m/z calcd for
[M + 1]+)756.3, found 756.3.
6AoH: p-Hexylaniline (7 g) was dissolved in 1:1 aqueous acetone
(100 mL), and concentrated HCl (10 mL) was added. The mixture
was cooled in an ice bath, and sodium nitrite (2.9 g) dissolved in 50
mL of cold water was added with stirring. After addition, the solution
was allowed to stand for 15 min in an ice bath. The diazonium salt
solution obtained was slowly added to a cold aqueous solution (100
mL) of phenol (4 g), sodium hydroxide (1.7 g), and sodium carbonate
(7 g). The yellow precipitate which appeared was filtered out and
recrystallized from a mixed solvent of benzene and hexane. A total of
5.46 g of product was obtained. Yield: 48%. mp: 81-82 °C. 1H
NMR (300 MHz, CDCl3), δ: 7.92 (d, 2 H), 7.84 (d, 2 H), 7.30 (d, 2
H), 6.98 (d, 2 H), 2.70 (t, 2 H), 1.70 (m, 2 H), 1.38 (m, 6 H), 0.97 (t,
3 H).
Ao8H: AoH (3.2 g), Br(CH2)7COOH (3.3 g), and potassium
hydroxide (1.8 g) were added to a 200 mL round-bottomed flask with
ethanol (60 mL). The mixture was refluxed overnight. An orange
precipitate was collected by filtration and recrystallized from CH2Cl2.
A total of 1.5 g of orange crystals was obtained. Yield: 67%. mp:
150-151 °C. 1H NMR (300 MHz, CDCl3), δ: 7.90 (q, J ) 9.5, 4 H),
7.50 (m, 3 H), 7.04 (d, J ) 9.4, 2 H), 4.06 (t, J ) 7.1, 2 H), 2.40 (t,
J ) 7.1, 2 H), 1.85 (m, 2 H), 1.50 (m, 2 H), 1.25 (m, 4 H).
6Ao2H: 6AoH (3 g), ethyl bromoacetate (1.5 mL), and potassium
hydroxide (600 mg) was added to a round-bottomed flask containing
50 mL of ethanol. The mixture was refluxed for 2 h and cooled in an
ice bath. A precipitate formed and was collected and recrystalized from
pentane. A total of 1.9 g of yellow crystals (6Ao2M) was obtained.
The crude mixed anhydride obtained above was dissolved in 10 mL
of dry methylene chloride and added via syringe to a suspension of
L-R-glycero-3-phosphorylcholine as the cadmium chloride complex
(GPC‚CdCl2) (160 mg) and 4-(N,N-dimethylamino)pyridine (DMAP)
(90 mg) in CH2Cl2 (4 mL). The mixture was stirred under nitrogen
for about 24 h. TLC (solvent B) showed the completion of reaction
with only a trace of monosubstituted product. The solvent was removed
by rotary evaporation, and the residue was dissolved in solvent B. The
solution was passed through an ion exchange column, Rexyn-I 300 to
remove the CdCl2 and DMAP. The solvent was removed by rotary
evaporation. The crude product was purified by passing it through a
Sephedex LH-20 column eluting with chloroform/methanol (50/50)
followed by precipitation in ether three times. A total of 140 mg of a
dry white powder was obtained. Yield: 44%. 1H NMR (500 MHz,
CDCl3), δ: 7.81 (d, J ) 7.5, 8 H), 7.30 (d, J ) 8.0, 8H), 5.25 (m, 1
H), 4.42 (d, J ) 11.5, 1 H), 4.33 (m, 2 H), 4.17 (dd, J ) 9,5, 7.0, 1 H),
4.02 (m, 2 H), 3.80 (m, 2 H), 3.31 (s, 9 H), 2.68 (t, J ) 7.5, 8 H), 2.34
(q, J ) 8.0, 4 H), 1.95 (m, 4 H), 1.64 (5, J ) 7.5, 4 H), 1.39 (5, J )
7.0, 4 H), 0.94 (t, J ) 7.5, 6 H). FAB: m/z calcd for [M + 1]+
870.4, found 870.4.
)
(30) Sandhu, S. S.; Yianni, Y. P.; Morgan, C. G.; Taiyor, D. M.; Zaba,
B. Biochim. Biophys. Acta 1986, 860, 253.
(31) Furman, I.; Geiger, C. H.; Whitten, D. G.; Penner, T. L.; Ulman,
A. Langmuir 1994, 10, 837.

Aggregates of Azobenzene Phospholipids
J. Am. Chem. Soc., Vol. 119, No. 39, 1997 9147
Scheme 1
(1) NaNO2
(2) phenol
H(CH2)m
N
H(CH2)m
NH2
N
OH
mAoH
(1) Br(CH2)n–1COOR
(2) NaOH
H(CH2)m
N
(1) Me3CCOCl
(2) GPC/DMAP
N
O(CH2)n–1COOH
mAonH
H(CH2)m
H(CH2)m
N
O(CH2)n–1COO
O(CH2)n–1COO
N
+
O
NMe3
N
O
P
O
N
O –
mAonEPC
Ao8EPC
6Ao2H
rattlesnake
venom
Me3CCOCl
N
6Ao2 –OCCMe3
Ao8,6Ao2EPC
N
O(CH2)7COO
+
O
NMe3
O
P
O
HO
mono-Ao8EPC
O –
Yield: 52%. 1H NMR (300 MHz, CDCl3), δ: 7.95 (d, J ) 9.6, 2 H),
7.84 (d, J ) 9.6, 2 H), 7.34 (d, J ) 9.4, 2 H), 7.03 (d, J ) 9.4, 2 H),
4.74 (s, 2 H), 4.32 (q, J ) 5.4, 2 H), 2.70 (t, J ) 8.3, 2 H), 1.68 (t, J
) 6.0, 2 H), 1.32 (m, 8 H), 0.95 (m, 3 H).
6Ao2M (1.8 g) was dissolved in 30 mL of water/methanol (10/90)
and to the solution was added sodium hydroxide pellets (800 mg). The
mixture was heated to reflux for 15 min and a yellow solid formed.
The solid was collected and recrystallized from chloroform containing
2% glacial acetic acid. A total of 1.5 g of a yellow powder was
obtained. Yield: 87%. Mp: >260 °C, with decomposition. 1H NMR
(300 MHz, DMSO-d6), δ: 7.95 (d, J ) 9.6, 2 H), 7.84 (d, J ) 9.6, 2
H), 7.34 (d, J ) 9.4, 2 H), 7.03 (d, J ) 9.4, 2 H), 4.60 (s, 2 H), 2.64
(t, J ) 8.4, 2 H), 1.60 (m, 2 H), 1.30 (m, 8 H), 0.85 (m, 3 H).
6Ao2EPC: The synthesis is similar to that of 4A4EPC. Yield: 54%.
1H NMR (500 MHz, DMSO-d6), δ: 7.85 (t, J ) 9.3, 4 H), 7.73 (q, J
) 5.2, 4 H), 7.33 (m, 4 H), 7.12 (t, J ) 9.3, 4 H), 5.24 (m, 1 H), 5.90
(m, 4 H), 4.45 (d, J ) 10.3, 1 H), 4.33 (m, 1 H), 4.07 (m, 2 H), 3.87
(m, 2 H), 3.50 (m, 2 H), 3.12 (s, 9 H), 2.62 (m, 4 H), 2.50 (m, 4 H),
1.57 (m, 4 H), 1.28 (m, 8 H), 0.85 (t, 6 H). FAB: m/z calcd for [M
+ H]+ 902.4, found 902.4.
Ao8EPC: The synthesis is similar to that of 4A4EPC. Yield: 67%.
1H NMR (500 MHz, CDCl3), δ: 7.89 (q, J ) 8.5, 8 H), 7.50 (t, J )
7.5, 4 H), 7.43 (t, J ) 7.0, 2 H), 6.99 (d, J ) 7.5, 4 H), 5.24 (m, 1 H),
4.40 (m, 3 H), 4.15 (dd, J ) 7.0, 9.5, 1 H), 4.02 (m, 6 H), 3.91 (m, 2
H), 3.39 (s, 9 H), 2.33 (m, 4 H), 1.80 (m, 4 H), 1.63 (m, 4 H), 1.47 (m,
4 H), 1.37 (m, 8 H). FAB: m/z calcd for [M + H]+ 902.4, found
902.4.
the precipitation in the mixed solvent of ether and CHCl3. A total of
90 mg of mono-Ao8EPC product was obtained.
The mixed anhydride obtained from the reaction between Ao2H
6
(130 mg) and trimethylacetyl chloride (0.3 mL) was dissolved in dry
methylene chloride and added via a syringe to a suspension of mono-
Ao8EPC (45 mg) and 4-(N,N-dimethylamino)pyridine (DMAP) in 4
mL of CH2Cl2. The mixture was sealed and stirred in the dark under
nitrogen for about 48 h. TLC (solvent B) indicated the completion of
the reaction. The solvent was removed by rotary evaporation and the
residue was dissolved in solvent B. The crude product was purified
by chromotography (chromatotron) eluting with solvent B. A total of
24 mg of pure product was obtained. Yield: 33%. 1H NMR (500
MHz, CDCl3), δ 7.88 (m, 6 H), 7.78 (d, J ) 7.5, 2 H), 7.49 (t, J ) 7.6,
2 H), 7.43 (t, J ) 7.5, 1 H), 7.28 (d, J ) 9.0, 2 H), 6.96 (d, J ) 9.0,
2 H), 5.4 (m, 1 H), 4.84 (d, J ) 16.0, 1 H), 4.73 (d, J ) 16.0, 1 H),
4.42 (d, J ) 14, 1 H), 4.32 (m, 2 H), 4.16 (m, 1 H), 4.08 (m, 2 H),
3.97 (t, J ) 7.1, 2 H), 3.76 (m, 2 H), 3.27 (s, 9 H), 2.64 (t, J ) 8.0,
2 H), 2.27 (t, J ) 8.0, 2 H), 1.80 (m, 2 H), 1.60 (m, 2 H), 1.32 (m, 4
H), 0.88 (t, 3 H). FAB: m/z calcd for [M + H]+ 902.4, found: 902.4.
Results and Discussion
The structures of the azobenzene (trans form unless otherwise
stated) amphiphiles synthesized and used in these investigations
are shown in Chart 1 together with their acronyms. All of these
compounds exhibit good solubility in organic solvents such as
chloroform and methylene chloride to give solutions character-
ized by the typical azobenzene monomer absorption spectra with
a weak n-π* transition near 450 nm, an intense π-π* transition
(A band) at 348 nm, and a second π-π* transition (B band) at
Ao8, 6Ao2EPC: A8EPC was dissolved in 3 mL of CH2Cl2/methanol
(99/1, v/v) followed by addition of an aqueous solution containing 20
mM tris-HCl, pH 8.0, and 40 mM CaCl2, and 2 mg of rattlesnake venom
(Crotalus adamanteous) was added. The reaction vessel was sealed,
and the solution was vortexed for 3 d in the dark. When the reaction
was completed, the organic solvent was removed by a stream of nitrogen
and the lyso product was extracted by the Bligh-Dyer procedure.32
Some distilled water was added to the aqueous solution to a total volume
of 8 mL followed by addition of 10 mL of CHCl3 and 20 mL of
methanol to obtain a clear one-phase solution. Then to the clear mixture
was added 10 mL of water and 10 mL of CHCl3 to get a biphasic
mixture. The CHCl3 layer was separated, and 20 mL of CHCl3 was
added to the methanol-water phase in an additional extraction. The
CHCl3 layers were combined, and the solvent was removed by
evaporation. Most of the Ao8H byproduct was removed by repeating
248 nm (with exception of Ao2EPC in chloroform, a little
6
broader and 4 nm blue-shifted absorption spectrum relative to
the other APL’s is observed and assigned to an oblique dimer).
In no case have we been able to detect any fluorescence from
these compounds either in organic solvents or in other media
(see below). The phospholipid derivatives (APL’s) shown in
Chart 1 can be dispersed by sonication in water, giving clear
solutions which, however, obviously scatter light. The short-
chain APL, A4EPC, gives an absorption spectrum in water very
similar to those obtained for all of the azobenzene amphiphiles
in chloroform, suggesting that the chromophore is monomeric.
In contrast, all of the other APL’s exhibit absorption spectra in
the aqueous dispersions characterized by a blue-shifted A band
(32) Bligh, E. G.; Dyer, W. J. Can. J. Biochem. Physiol. 1959, 37, 911.

9148 J. Am. Chem. Soc., Vol. 119, No. 39, 1997
Song et al.
Table 1. Absorption Spectral Data of APL’s in Different Media
λ
max, nm (ꢀ)b
chloroform
water
DMPC
shift,c nm
FWHM,a nm
Bd
He
Ao8EPC
348 (3.6)
344 (4.3)
348 (3.8)
336 (4.8)
317 (3.8)
320 (3.4)
320 (4.3)
311 (3.3)
348 (5.4)
348 (5.2)
348 (5.3)
332 (4.0)
31
28
28
25
41
57
59
56
2.436
1.765
1.653
1.458
0.5659
0.5513
0.5257
0.6900
6Ao2EPC
Ao8,6Ao2EPC
4A4EPC
a Full width at half maximum. b Extinction coefficient × 10-4 at absorption maxima. c Absorption spectral maximum of aggregate relative to
monomer. d B ) ꢀ312/ꢀ332 for pure A4EPC vesicle; B ) ꢀ318/ꢀ348 for pure vesicles of other three APL’s. e H ) ꢀ312/ꢀ332 for A4EPC highly diluted
4
4
in DMPC vesicles; H ) ꢀ318/ꢀ348 for other three APL’s highly diluted in DMPC vesicles.
Figure 1. Absorption spectra of Ao8EPC in water, chloroform, and
DMPC; and Ao8H in γ-CDN: (s) water, (- - -) DMPC, (-9-) γ-CDN,
(-1-) chloroform.
(Table 1). A comparison between the spectra of Ao8EPC in
water, chloroform, and water with excess saturated fatty acid
phospholipid, dimyristoyl phosphatidyl choline (DMPC) co-
solubilized, is shown in Figure 1. By analogy with results
obtained for structurally similar amphiphiles such as the
isoelectronic trans-stilbene derivatives17 and tolans,20 we assign
the blue-shifted spectrum obtained in the aqueous dispersion
of pure APL to an “H” aggregate33 while the much less shifted
spectrum in the APL:DMPC dispersions in water are attributed
to a dimer. Evidently a polarity effect produces a red shift when
the enviroment is changed from the hydrophobic bilayer of the
vesicles to chloroform; in the case of dimer formation, the
consequent blue shift associated with formation of an “H” dimer
superimposed on the “solvent” red shift results in a net small
blue shift. Similar spectral shifts are associated with the change
in environment from monomer in an organic solvent to the
dimeric complex in â- or γ-cyclodextrin in aqueous solutions.
The azobenzophanes prepared in an earlier study which have a
face-to-face relationship imposed in a rigid structure show a
similar small blue shift in absorption.34
While the absorption spectra associated with the aggregate
are similar for the different APL’s, there are some significant
differences (Figure 2). For example, the spectrum of Ao8EPC
in water is sharper and slightly more blue shifted than those of
6Ao2EPC and Ao8,6Ao2EPC. Possible reasons (see below)
include either different aggregate sizes or a different packing
within the aggregate. As we have observed earlier with the
stilbene phospholipids17 and other aggregating aromatics,19,20
mixing of the “pure aggregate” solutions with aqueous solutions
of excess saturated phospholipid such as DMPC results in an
evolution of the aggregate spectrum to that associated with the
dimer. The spectral changes observed upon mixing aqueous
Figure 2. Absorption (upper) and ICD (lower) spectra of aqueous
dispersions of pure APL’s: (s) Ao8EPC, (- - -) 6Ao2EPC, (‚‚ׂ‚) Ao8,6
-
Ao2EPC, (‚‚9‚‚) A4EPC.
4
APL’s and aqueous DMPC (Figure 3, parts a and b show mixing
with 6Ao2EPC and Ao8EPC, respectively) illustrate some
additional differences. For Ao8EPC, it is clear from the mixing
experiment that the band corresponding to the monomer “A”
transition undergoes the characteristic blue shift associated with
H aggregation while the “B” band undergoes a concurrent red
shift during the aggregation (J shift).33,35 A similar result is
observed for 4A4EPC. In contrast, for 6Ao2EPC and Ao8,6Ao2-
EPC, there appears almost no change in the “B” band from the
H-aggregate solution of pure APL’s to the dimer solution in
DMPC vesicles. The H aggregation and J aggregation for A
band and B band, respectively, of APL aggregates are also
demonstrated by their strong excitonic ICD spectra (see below).
Physical Characterization of trans-Azobenzene Phospho-
lipid Dispersions in Water. The aqueous “solutions” of pure
APL’s or APL-saturated phospholipid mixtures can be charac-
(35) The transition dipole associated with the A band lies along the long
axis of the azobenzene while that for the B band is perpendicular to the
long axis. The head-to-head alignment of the dipoles for the A band in the
proposed trimeric structure corresponds to a head-to-tail arrangement of
the dipoles of the B band, which should lead to a red shift in the latter
transition.
(33) (a) Kasha, M.; Rawls, H. R.; El-Bayoumi, M. Pure Appl. Chem.
1965, 11, 371. (b) Czikkely, V.; Fo¨rsterling, H.; Kuhn, H. Chem. Phys.
Lett. 1970, 6, 11. (c) Kasha, M. Radiat. Res. 1963, 20, 55. (c) Czikklely,
V.; Fo¨rsterling, H. D.; Kuhn, H. Chem. Phys. Lett. 1970, 6, 207.
(34) Rau, H.; Luddecke, E. J. Am. Chem. Soc. 1982, 104, 1616.

Aggregates of Azobenzene Phospholipids
J. Am. Chem. Soc., Vol. 119, No. 39, 1997 9149
Figure 4. DSC of aqueous dispersions of pure APL’s (scan rate ) 60
°C/h): (s) Ao8EPC, (- - -) 6Ao2EPC, (‚‚[‚‚) Ao8,6Ao2EPC, (‚‚b‚‚) 4A4-
EPC.
Table 3. Dynamic Light-Scattering Data for APL’s (5 × 10-5 M)
size (diameter), nm
intensity, × 10-5
M
trans
355
86
cis
trans
cis
Ao8EPC
6Ao2EPC
Ao8,6Ao2EPC
4A4EPC
DMPC and DPPC
232
82
2.14
0.91
1.49
0.92
1.45
71
92
102
1.32
10-20a
a The size depends upon the preparation procedures; see ref 54.
Figure 3. Dilution of aqueous dispersions of Ao2EPC (upper trace)
6
and Ao8EPC (lower trace) with DMPC as a function of time.
the main transition than the APL with the azobenzene close to
the head group while the “mixed” APL has an even lower Tc
and a very small enthalpy of melting. 4A4EPC, which has the
azobenzene group in an intermediate position, has a Tc and
enthalpy above that of 6Ao2EPC but below that of Ao8EPC. As
will be discussed later, the behavior observed is consistent with
a somewhat different aggregate structure for the different APL’s.
For all four APL’s it is found that there is little change in the
absorption spectral maxima or band shape in aqueous dispersions
over the temperature range 10-80 °C with the exception of a
decrease in intensity. The persistence of photophysical proper-
ties over the range of temperatures where chain melting occurs
suggests that the small domains or structural units which
determine the photophysics persist and are relatively unperturbed
by this phase-transition process. Similar but broader DSC peaks
of DMPC and DPPC codispersed with APL’s in water compared
with pure DPPC or DMPC aqueous dispersions suggest the
APL’s are well solubilized in DPPC or DMPC bilayers.
The aqueous dispersions of the APL’s scatter light, and
dynamic light scattering can be used to estimate the size of the
microparticles that are formed. Table 3 compares sizes
(diameters, based on assumption of spherical particles) estimated
by dynamic light scattering for the four trans-APL’s and
compares these estimated diameters with those of DMPC and
DPPC. Values for the APL’s following irradiation to form a
cis-rich photostationary state are also listed in the Table 3. The
particle size for Ao8EPC was found to decrease while the other
APL’s showed little change on production of cis and no new
distribution of particle sizes. The estimation of particle sizes
on the order of or greater than 100 nm suggested that the APL
assemblies might not pass through a 100 nm pore filter. In
fact it was found that no trans-Ao8EPC dispersions in water
could be filtered through a 100 nm polycarbonate membrane.
About half of the APL’s in aqueous dispersions of trans-6Ao2-
EPC, Ao8, 6Ao2EPC, and 4A4EPC could be extruded through a
100 nm polycarbonate membrane but no trans-APL could be
Table 2. DSC Data of Aqueous Dispersions of Pure APL’s
T, °C (major) Tc, °C (minor) ∆H, kcal/mol (major)
Ao8EPC
74.5
40.1
35.2
45.5
11.1
7.8
2.1
6.2
6Ao2EPC
Ao8,6Ao2EPC
4A4EPC
36.5
39.2
terized by several different methods. In the present study we
have used dynamic light scattering, microfiltration, differential
scanning calorimetry (DSC), and cryo-transmission electron
microscopy (Cryo-TEM). Microcalorimetry on aqueous disper-
sions of saturated phospholipids such as DPPC or DMPC which
exist as bilayers (or even in multilamellar vesicles) shows
evidence of phase transitions at 42 and 23 °C, respectively.36
These are associated with chain melting or conversion through
heating of a solid or gel form to a liquid crystalline state. For
saturated phospholipids the phase-transition temperature is
nearly independent of the precise state of the dispersion
(unilamellar vesicle or multilamellar assembly) but highly
dependent on chain length. Addition of sites of unsaturation
either reduces the phase-transition temperature or eliminates the
phase transition. The APL’s are sufficiently soluble in water
to observe phase transitions and measure the corresponding heat
capacities; data in Table 2 and Figure 4 compare the phase-
transition behaviors of four APL suspensions in water. The
data obtained for 4A4EPC are consistent with the results reported
in literature.37 There is a clear trend among the three oxygenated
APL’s, which should all have about the same “hydrophobic”
chain length; the phospholipid with the azobenzene farthest from
the head group has a much higher Tc and a larger enthalpy for
(36) Kremer, J. M. H.; Kops-Werkhoven, M. M.; Pathmamanoharan, C.;
Gijzeman, O. L. J.; Wiersema, P. H. Biochim. Biophys. Acta 1977, 471,
177.
(37) Morgan, C. G.; Thomas, E. W.; Sanhdu, S. S.; Yianni, Y. P.;
Mitchell, A. C. Biochim. Biophys. Acta 1987, 504.

9150 J. Am. Chem. Soc., Vol. 119, No. 39, 1997
Song et al.
passed through a 20 nm polycarbonate membrane. This is
consistent with sizes near 100 nm diameter for these APL’s in
water as indicated by the light-scattering data. An indication
that the size of the APL particles in aqueous solution might be
dramatically decreased by conversion of the trans-azobenzene
to cis was found by filtration experiments following irradiation
to form a cis-rich photostationary state. For Ao8EPC it was
found that a dispersion of cis-Ao8EPC (as indicated by spectral
changes following irradiation of trans) could be extruded almost
completely through a 100 nm polycarbonate membrane but not
through a 20 nm polycarbonate membrane.38 The extruded
solution of cis could be irradiated with a sun lamp (visible
irradiation) or stored in the dark at room temperature for several
hours, in either case resulting in conversion back to trans; the
resulting dispersion failed to pass any trans upon extrusion
through the 100 nm membrane (as indicated by lack of any
absorption in the UV-visible following filtration). Similar
experiments were performed with the other APL’s; about half
blue-shifted absorption spectrum.39 The azobenzene H dimers
of APL’s highly diluted in DMPC or DPPC exhibit no CD
signals while a small blue shift in the absorption spectra of the
H dimers is observed. In contrast to the lack of or weak ICD
for the monomer and dimer, strong ICD spectra (Figure 2) were
observed for all pure APL vesicles. No linear dichroism signals
have been detected for APL vesicles. Characteristic of the ICD
spectra of Ao8EPC and A4EPC assemblies are triphasic ICD
4
signals for the blue-shifted A band (H aggregation)40 and
biphasic ICD signals for red-shifted B band (J aggregation) while
the ICD of 6Ao2EPC and Ao8,6Ao2EPC vesicles have biphasic
ICD signals from the H-aggregate A band and no significant
signal for the B band. ICD data of APL aggregates are
summarized in Table 4 (all vesicles were prepared by sonication
followed by fast cooling to room temperature). The J aggrega-
tion of the B band of Ao8EPC and A4EPC vesicles was also
4
observed in the absorption spectra which exhibit a small red
shift relative to the monomer. The absorption peaks of the A
bands and B bands of the aggregates lie at the wavelengths
similar to those of the zero cross points of the excitonic ICD
bands.40 No CD signals have been observed for the n-π*
transitions.
Although the blue-shifted absorption spectra of all four APL
vesicles are independent of conditions used for preparation of
the vesicles, their ICD spectra depend strongly upon the cooling
rate of the sonicated aqueous solutions following the vesicle
preparation. ICD spectra with reverse signs were obtained for
6Ao2EPC and Ao8,6Ao2EPC vesicles for the fast and slow
of cis-6Ao2EPC, Ao8,6Ao2EPC, and A4EPC obtained by ir-
4
radiation at 365 nm from the corresponding trans solutions could
be extruded through 20 nm membranes. As with the Ao8EPC,
these samples could be reconverted to trans and the resulting
solutions consisted of “particles” which could not pass through
the 20 nm membrane. Initially, we thought that APL’s dispersed
in aqueous solutions might have formed spherical closed vesicle
structures. They turned out not to be the case. We found that
pure aqueous dispersions of APL’s do not trap carboxyfluores-
cein (CF) as would be expected if closed bilayer vesicles are
formed. A Cryo-TEM study of trans-Ao8EPC aqueous disper-
sion (Figure 5a) shows very flat bilayer structures of extended
or folded sheets which are apparently open structures and show
little curvature while the Cryo-TEM image (Figure 5b) of a cis-
Ao8EPC sample prepared approximately a half hour after
irradiation of trans-Ao8EPC vesicles (a small percentage of the
cis-APL’s have thermally been converted back into trans-APL)
exhibits both faceted, flat sheet structures and smaller spherical
structures. A reasonable interpretation is that the aggregated
chromophores of trans-Ao8EPC resist the necessary curvature
that would be required to form small unilamellar, spherical
vesicles. The sizes (ranging from 100 to 300 nm in length and
width for trans-Ao8EPC and smaller structures for cis-Ao8EPC
rich sample) of microparticles from Cryo-TEM are consistent
with the results from light-scattering and extrusion experiments.
The TEM micrographs show no significant difference between
the original trans-Ao8EPC vesicles and trans-Ao8EPC vesicles
obtained from thermal and photoreisomerization of cis-Ao8EPC
solutions. trans-6Ao2EPC vesicles show totally different struc-
tures from the sheetlike structure of trans-Ao8EPC vesicles. The
Cryo-TEM of trans-6Ao2EPC vesicles shows fibrous structures
(Figure 5c). Some fibers are as long as 1000 nm with diameters
of a few nanometers.
cooling procedures (Figure 6), while the ICD signals of A4-
4
EPC and Ao8EPC vesicles are the same regardless of the cooling
rate. The identical absorption spectra and reverse ICD spectra
can be rationalized by the coexistence of two enantiomeric
aggregates which have different energies originating from the
chiral enviroment and their molecular structures. The measured
apparent ICD spectra are summations of the ICD spectra of the
populations of the two enantiomers in solution. For conven-
ience, we label the aggregates with negative CE at long
wavelengths and positive CE at short wavelengths of major A
band as “R aggregates” and the aggregates with opposite ICD
as “S aggregates”. The R aggregates of Ao2EPC and Ao8,6-
6
Ao2EPC are converted into the S aggregates simply by heating,
followed by slow cooling as indicated by the change in sign of
the ICD spectra associated with the A band. The aggregation
process is controlled by kinetics under the fast-cooling condition
so that the kinetically favored enantiomer dominates. The slow
conversion from the less stable enantiomer to the more stable
below Tc increases dramatically at temperatures above Tc in the
APL aqueous solutions. The R aggregates for Ao8EPC and 4A4-
EPC are both kinetically and thermodynamically more stable
while the R aggregates for Ao2EPC and Ao8,6Ao2EPC are
6
kinetically stable but thermodynamically less stable than the S
aggregates.
Another interesting observation for APL vesicles is that the
strong ICD signals disappear completely when the trans-APL
vesicles are photoisomerized into cis-APL’s by UV light. The
trans-to-cis photoisomerization of the aggregates in APL
Circular Dichroism (CD) Spectra of APL Vesicles. No
induced circular dichroism (ICD) signals are detected for APL’s
in chloroform where the monomer is formed, except for 6Ao2-
EPC which exhibits a weak biphasic CD spectrum with a
positive Cotton effect (CE) at 480 nm and a negative CE at
340 nm. The two CE’s are attributed to an oblique dimer of
azobenzene which is also supported by its broad and slightly
bilayers is much slower (∼30- and 130-fold slower for Ao2-
6
EPC and Ao8EPC, respectively) than the azobenzene monomers
of APL’s in chloroform. Figure 7 shows absoprtion and ICD
spectra of 4A4EPC vesicles at different stages of isomerization
from trans to cis. All the ICD spectra pass through the same
(38) In Table 3, the size of the aggregate from cis-Ao8EPC is estimated
from light scattering to be ∼200 nm. Although, as a reviewer has pointed
out, this may seem inconsistent with extrusion of cis through a 100 nm
membrane, the light-scattering data are generally somewhat skewed toward
larger sizes since big particles make larger contribution. Additionally, the
measurements in Table 3 were made at times following irradiaition during
which some thermal isomerization to trans had occurred.
(39) (a) Person, R. V.; Peterson, B. R.; Lighter, D. A. J. Am. Chem.
Soc. 1994, 116, 42. (b) Gargiulo, D.; Ikemoto, N.; Odingo, J.; Bozhkova,
N.; Iwashita, T.; Berova, N.; Nakanishi, K. J. Am. Chem. Soc. 1994, 116,
3760. (c) Polonski, T. J. Org. Chem. 1993, 58, 255.
(40) (a) Ebrey, T. G.; Becker, B.; Mao, B.; Kilbride, P. J. Mol. Biol.
1977, 112, 377. (b) Wu, S.; El-Sayed, M. A. Biophys. J. 1991, 60, 190. (c)
Cassim, J. Y. Biophys. J. 1992, 63, 1432.

Aggregates of Azobenzene Phospholipids
J. Am. Chem. Soc., Vol. 119, No. 39, 1997 9151
Figure 5. Cryo-TEM micrographs of aqueous dispersions of (a, upper left) trans-Ao8EPC, (b, upper right) cis-Ao8EPC, and (c, bottom) trans-6-
Ao2EPC.
Table 4. ICD Data of Aqueous Dispersions of Pure APL’s
λ
min, nm
θ
min,a × 10-5
λ
max, nm
θ
max,a × 10-5
λt,c nm
θt,a × 10-5
λθ)0, nm
Ao8EPC
324 (277)b
331 (243)
335 (246)
317 (258)
-23.8 (3.1)
-48.6 (9.4)
-42.3 (5.8)
-34.0 (0.9)
304 (256)
301
4.7 (4.4)
17.0
407
1.6
310 (256)
311
319
302 (252)
6Ao2EPC
Ao8,6Ao2EPC
4A4EPC
309
22.0
12.0 (1.1)
293 (247)
387
1.6
a Molecular ellipticity (mdeg cm2/dmol). b The data in parentheses correspond to B band. c Third CE for A band.
zero cross point and absorption spectra show two isosbestic
points. The persistence of the ICD of the same aggregate even
at relatively late stage of photoisomerization where most of
trans-APL is transformed into cis is consistent with a small
“unit” aggregate.1 Other APL’s show the same behavior. The
blue-shifted absorption spectra of the aggregates of trans-APL’s
recover from the cis-rich solution by visible light irradiation.
The ICD spectra of the newly formed trans-Ao8EPC and trans-
4A4EPC vesicles show little difference from the original trans
solutions except in intensity which is higher for Ao8EPC and

9152 J. Am. Chem. Soc., Vol. 119, No. 39, 1997
Song et al.
Figure 6. ICD spectra of aqueous dispersions of Ao2EPC prepared
6
by different cooling procedures: (s) fast cooling, (‚‚‚) slow cooling.
Figure 8. ICD spectra of Ao8EPC (upper) and 6Ao2EPC (lower)
aqueous dispersions through a reversible photoisomerizations cycle.
Figure 7. Absorption (upper) and ICD (lower) spectra of 4A4EPC
aqueous dispersions as a function of irradiation time at 365 nm.
lower for 4A4EPC compared with original trans solutions.
Interestingly, S aggregates dominate in the newly formed trans-
6Ao2EPC and trans-Ao8,6Ao2EPC aqueous solutions (Figure 8).
In these cases, the photoisomerization cycles of trans-APL
aqueous dispersions result in the transformation of one enan-
tiomer into the other.
APL’s diluted with DMPC in moderate ratios show almost
identical exciton ICD spectra to those of pure APL aggregates.
This result supports the mixing experiments and equilibration
between the aggregates and dimers which suggest no significant
presence of other intermediate aggregates. Figure 9 shows the
Figure 9. (a) ICD spectra of 6Ao2EPC in L-DPPC vesicles with various
DPPC/6Ao2EPC molar ratio. (b) The absorbance-normalized ICD
intensity of the mixed vesicle solution vs the DPPC/APL molar ratio:
(9) at 301 nm for Ao2EPC, (2) at 338 nm for Ao8,6Ao2EPC, (×) at
6
318 nm for A4EPC, (b) at 327 nm for Ao8EPC.
4
the intensity-normalized ICD signals on the dilution ratio clearly
demonstrates the coexistence of both enantiomeric H aggregates
which shift the equilibration between the two enantiomers due
to the enviromental change of the matrix. The relative stability
of the enantiomers depends upon both the positions of azoben-
zene in the fatty acid chains and the environment of the matrix.
ICD spectra of Ao2EPC at different dilutions with DPPC and
6
the intensity normalized ICD signals of APL in DPPC. All the
spectra in different dilutions have a common zero cross point
which indicates that the ICD signals are contributed from the
same aggregate (the dimers are CD silent). The dependence of
For Ao2EPC and Ao8,6Ao2EPC, the R-aggregate contribution
6

Aggregates of Azobenzene Phospholipids
J. Am. Chem. Soc., Vol. 119, No. 39, 1997 9153
increases when the dilution becomes high while the R-aggregate
contribution decreases for A4EPC. In contrast, the relative
4
contribution of both R and S aggregate does not change
significantly for Ao8EPC, as predicted by the relative large
aggregates which experience little enviromental change by
dilution of the matrix. The intensity-normalized ICD signal
levels off when the dilution reaches a certain level beyond which
further dilution does not cause significant environmental change
experienced by the aggregates.
Estimation of Aggregate Sizes. Similar to stilbene-,17
styrylthiophene-,41 and tolan20-derivatized phospholipids, the
evolution of absorption spectra of pure APL vesicles shows
either one isosbestic point (6Ao2EPC, Ao8,6Ao2EPC, and A4-
4
EPC) or two (Ao8EPC) when they are mixed with DPPC or
DMPC vesicles at or above Tc (Figure 3). The result suggests
that only two dominant species, an aggregate and a dimer, are
involved in the process. The aggregate-dimer conversion can
be accomplished either by changing the dilution ratio or the
solution temperature. The absorption spectra of the mixture of
APL and DMPC vesicles at different temperatures also display
one isosbestic point. This indicates that the aggregates and
dimers are in equilibrium with little contribution from other
species so that the equilibration between dimers and aggregates
can be expressed by eq 1. A modified Benisi-Hildebrand
Figure 10. Benesi-Hildebrand plot of ln Aagg vs ln Adi of APL’s in
DMPC vesicles: (9) Ao2EPC at 42 °C, (2) Ao8,6Ao2EPC at 50 °C,
6
(+) Ao8EPC at 75 °C, (b) A4EPC at 44 °C.
4
or
ln[Aggn] ) n ln[Di] + C′
or
Aggn (DMPC) ) nDimer (DMPC)
(1)
ln Aagg ) n lnAdi + C′′
(7)
approach42 is used to formulate the equilibrium, and eq 7 is
obtained in terms of the concentration and absorbance of
aggregates and dimers in equilibrium solution with moderate
dilution ratio. The plots according to eq 7 should be a straight
line whose slopes allow determination of aggregate sizes n
where n is the number of APL molecules for an aggregate.
Equilibration between aggregate and dimer occurs in the DMPC
matrix, which can be expressed by
Here, C, C′, and C′′ are constants and Aagg and Adi are
absorbances of aggregates and dimers in equilibrium solution,
respectively, which are easily obtained through eqs 8 and 9
where Atx and Aty are total absorbances at wavelengths of x and
y, respectively; B ) ꢀaggx/ꢀdiy, the ratio of extinction coefficient
at the peak wavelength (x) of aggregate, to that at the peak
wavelength (y) of dimer or monomer, for the pure aggregate
spectrum and H ) ꢀdix/ꢀdiy, the ratio of the extinction coeficient
at the same two wavelengths for pure dimer or monomer
absorption spectrum. The B and H for four APL’s are listed in
Table 1 along with the extinction coefficients of pure aggregates
and dimers.
K ) [Di]ndmpc/[Aggn]dmpc
(2)
It should be noted that the concentrations used here refer to
real concentrations in the DMPC microphase instead of the bulk
aqueous solution concentration. For convenience, we replace
the real concentrations in DMPC with concentrations in bulk
solution by a simple conversion:
Aaggx ) [H(Aty - Atx)]/[H/B - 1]
Adiy ) [B(Aty - Atx)]/[B - H]
(8)
(9)
[Aggn]aq ) [Aggn]dmpc (Vdmpc/Vaq)
[Di]aq ) [Di]dmpc (Vdmpc/Vaq)
Vdmpc ) MVm
(3)
(4)
(5)
The plot of ln Aagg vs ln Adi show reasonable straight lines for
all four APLs as shown in Figure 10. Aggregate sizes were
measured to be 42 (n ) 21 ( 4), 3, 3, and 3 (n ) 1.7 ( 0.2))
in term of azobenzene chromophore for Ao8EPC, Ao2EPC,
6
Ao8,6Ao2EPC, and 4A4EPC, respectively. Surprisingly, the
aggregate sizes of Ao2EPC, Ao8,6Ao2EPC, and A4EPC are
6
4
where Vdmpc is the volume of DMPC; Vaq, volume of aqueous
solution; M, number of moles; and Vm, partial specific volume
of DMPC which is considered to be constant.43 Combination
of eq 2-5 produces
much smaller than that of Ao8EPC although their molecular
structures are similar. The exceptionally large aggregate size
for Ao8EPC is consistent with its sharper and more blue-shifted
absorption spectrum as well as its higher Tc.
The aggregate-dimer conversion for all APL’s with the
treatment of excess DMPC vesicles show pseudo-first-order
kinetics for APL aggregates as observed for SPL’s. Rate
constants for the deaggregation processes of APL’s are listed
in Table 5 along with their equilibrium constants.
n-1
K ) [Di]aqn/{[Aggn]aq[DMPC]aqn-1 Vm
}
(6)
Equation 7 can be obtained from eq 6, if [DMPC] is kept
constant:
Monolayers and LB Films. Pure azobenzene derivatized
fatty acids (AFH’s) and APL’s form monolayer films when
spread from an evaporating chloroform solution over water at
pH 6.8. The surface pressure area isotherms are shown in Figure
K ) [Di]aqn/[Aggn]aqC
(41) Song, X.; Whitten, D. G. Manuscript in preparation.
(42) Benesi, H. A.; Hildebrand, J. H. J. Am. Chem. Soc. 1949, 71, 2703.
(43) (a) Gennis, R. B. Biomembranes, Molecular Structure and Function;
Springer-Verlarg: New York, 1989. (b) Fendler, J. H. Membrane Mimetic
Chemistry; John Wiley & Sons: New York, 1982.
11 for Ao8H and Ao2H along with the reflectance spectra of
6
the monolayers at the air-water interfaces and absorption
spectra of LB films on quartz. The isotherms are relatively

9154 J. Am. Chem. Soc., Vol. 119, No. 39, 1997
Song et al.
Table 5. Rate and Equilibrium Constants for Eq 1 for APL’s at
Different Temperatures
to that observed in APL vesicles and the monolayer of Ao8H at
the air-water interfaces while the absorption A band of the
LB film of Ao8H is further blue shifted to 295 nm. The intensity
for the A and B bands of the LB film of 6Ao2H are comparable.
In contrast, the intensity of the B band is almost twice as high
as A band of the LB film of Ao8H. These different blue shifts
of the aggregates for AFH LB films were also observed by other
groups and can be rationalized by the different orientation of
azobenzenes and aggregate sizes. In the LB films of Ao8H,
the aggregate is large and the major transition dipole moments
of A band in the aggregate may be more perpendicularly
oriented to the surface of the quartz than for the LB films of
6Ao2H. The transition dipole moments of the B band in the
LB film of Ao8H should be perpendicular to the transition dipole
of A band and nearly parallel to the plane of the quartz so that
the strong interaction between the dipole and light coming from
the perpendicular direction gives a strong absorption.
b
compounds
4A4EPC
T (°C)
Ka
k (min-1
)
40
45
75
70
50
45
50
65
0.26
1.0 × 10-2
1.1 × 10-1
2.0 × 10-3
6.1 × 10-2
Ao8EPC
4.6 × 10-32
6Ao2EPC
1.5
Ao8,6Ao2EPC
2.4
a K is the equilibrium constants for eq 1; units depend upon n. b k is
the pseudo-first-order rate constant for aggregate dissociation.
Photoregulatable CF Release From Vesicles. Much atten-
tion has been devoted to the potential of vesicles as a drug
delivery vehicle,44 mainly due to the compatibility of natural
phospholpid vesicles with cell membranes, minimal toxicity,
and easy control of drug release from vesicles by many
techniques22-24 such as hyperthermia,45 pH-sensitive polymer
surfactants,46 and photochemistry. As a drug delivery system,
it is very important that the release of drugs can be controlled
easily and the vehicle must produce minimal side effects.
Several studies have focused on photochemical perturbation of
vesicle bilayers22-24 which, to some extent, can promote the
release of water soluble reagents. Here we have investigated
photoisomerization of APL’s to control the reagent release from
vesicles as a function of the phospholipid structure, vesicle
composition and aggregation state of the APL’s. In this study,
carboxyfluorescein (CF) was used as a fluorescence indicator
to monitor change in pemeability of the vesicles.29 The
fluorescence intensity of CF in aqueous solution at very low
concentration is proportional to [CF] while it fluoresces weakly
due to self-quenching at high concentration (>100 mmol).
Vesicles with high concentration of CF inside and free of CF
outside can be prepared by gel filtration; the weak fluorescence
of CF will increase due to its dilution if any leakage of CF
occurs from the inside of vesicles to the bulk solution on the
outside. Thus leakage of the CF from the vesicles is easily
followed by monitoring the CF fluorescence. In these studies,
0.1 M CF aqueous solutions were used and ∼2000 CF molecules
are estimated entrapped in one vesicle of an average size47,48
(based on the assumption of 25 nm in diameter and 5 nm in
thickness of the bilayer, see Cryo-TEM study below).
Figure 11. (a) Surface pressure area isotherms of AFH’s monolayers.
(Inset) Reflectance spectra of Ao8H monolayer at different pressures.
(b) Absorption spectra of AFH monolayers on quartz: (s) Ao8H, (‚‚‚)
6Ao2H.
broad compared with those of fatty acids like arachidic acid.
The molecular area from the isotherm is siginifcantly smaller
than expected, probably due to some solubility of AFH’s in
water. The films at the air-water interface show almost
identical reflectance spectra at different pressures with a
dominant transition at 318 nm. The reflectance of the film is
significantly blue shifted relative to monomer absportion spectra
of AFH’s and APL’s in organic solvents such as chloroform
and is quite similar to the blue-shifted absorption spectra
obtained for APL vesicles or for LB film of Ao8EPC on quartz.
It is noteworthy that the blue-shifted spectrum is observed even
at very low surface pressure (0.02 mN/m) where little “forced”
packing of the chromophores is anticipated. The insensitivity
of the reflectance spectra to the surface pressure suggests that
the aggregate is a result of strong interaction between the
azobenzene units prior to compression rather than formation of
a forced assembly as a consequence of the compression and
organization of the monolayer.21
As expected from the open, flat, sheetlike structures observed
for Cryo-TEM study of pure APL aqueous dispersions, pure
APL vesicles do not entrap any CF. On the other hand, the
vesicles formed from mixtures of APL or AFH and DPPC,
where the azobenzene may exist as aggregates or dimer or
monomer, entrap CF without significant leakage for extended
periods of time (one week, for example) at room temperature
or at 4 °C. The leakage of CF from the inside to the ouside of
vesicles can be controlled by photoisomerization of azobenzene
aggregates or other photoreactive chromophore-derivatized
(44) (a) Philippot, J. R.; Schuber, F. Liposomes as Tools in Basic
Research and Industry; CRC Press: Ann Arbor, 1995. (b) Garber, B. P.;
Schnur, J. M.; Chapman, D. Biotechnological Application of Lipid Micro-
structures; Plenum Press: New York, 1988.
(45) Yatvin, M. B.; Weinstein, J. N.; Dennis, H. W.; Blumenthal, R.
Science 1978, 202, 1290.
(46) Thomas, T. L.; Tirrell, D. Acc. Chem. Res. 1992, 25, 336.
(47) Cullis, P. R.; Hope, M. J. In Biochemistry of Lipids and Membranes;
Vance, D. E., Vance, J. E., Eds.; Benjamin/Cummings: Menlo Park, CA,
1985; pp 25-72.
(48) Humphry-Baker, R.; Thompson, D. H.; Lei, Y.; Hope, M.; Hurst,
J. K. Langmuir 1991, 7, 2592.
The compressed monolayers of AFH’s at the air-water
interfaces can be readily transferred to solid supports such as
quartz to form LB films. The LB film of 6Ao2H on quartz gives
a blue-shifted absorption transition at 318 nm for A band similar

Aggregates of Azobenzene Phospholipids
J. Am. Chem. Soc., Vol. 119, No. 39, 1997 9155
solution of DPPC/6Ao2EPC (10/1) showed spherical vesicle
structures which are similar to that of pure DPPC vesicles, The
solutions obtained immediately after irradiation at 365nm
showed many needle-like structures at the expense of spherical
vesicles (Figure 13).
To clarify the effect of vesicle structure on the release
pathway induced by photoisomerization of APL aggregates, we
studied a series of vesicles containing different ratios of
1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine/cholesterol
(POPC/chol). The compactness of POPC vesicles (Tc ) -5
°C) is easily tuned by incorporating cholesterol into vesicles.
The vesicles made from POPC and POPC/chol exist in the fluid
phase at room temperature. The compactness of the mixed fluid
membrane, in general, tends to increase with increasing mole
fraction of cholesterol.29 In these studies, vesicle systems used
were made from pure POPC, POPC/chol ) 10/10, and pure
DPPC with 10/1 molar ratio of lipids/6Ao2EPC, which are in
increasing order of compactness. No significant leakage can
be detected at room temperature without irradiation for all three
systems. Figure 14 compares fluorescence intensity profiles of
CF upon irradiation for these samples. After irradiation at 365
nm for short times (1 min) which converts trans completely
into cis, for the DPPC/APL vesicles in the low temperature or
solid phase, 100% release of CF is obtained. For POPC/chol
) 10/10 in which the compactness of lipids is greater than POPC
but less than for DPPC, ∼30% CF is released by a “rupture”
pathway and 70% by a slower leakage pathway. For the POPC
vesicle, which is in fluid phase, a slow leakage pathway through
which CF is released dominates to nearly 100%. Apparently
the fluid state of vesicle favors a slow leakage pathway and the
solid or gel phase favors the rupture process. A reasonable
explanation is that for stiff DPPC vesicles, the integrity of the
vesicle is lost on irradiation and the gel-phase lipids are not
flexible enough to permit rapid healing. In contrast, we suspect,
for POPC vesicles, release is through small defects caused by
formation of cis-APL but that the vesicle structure remains
intact. These results demonstrate clearly that permeability of
vesicles can be tuned by choosing suitable lipid hosts and by
modulating external conditions such as temperature and irradia-
tion. The two different mechanisms are nicely supported by
Cryo-TEM microscopy as shown in Figure 13. For DPPC/APL
dispersions, all the spherical vesicle structures are converted
into needle-like structures upon photoisomerization while no
observable structural change occurs for POPC vesicles. In
contrast, both needle-like and spherical vesicle structutres are
observed for POPC/chol after irradiation.
Figure 12. Fractional change in fluorescence intensity of trapped CF
in vesicles uupon irradiation as a function of time. (Io, Ie, and I) Initial
intensity, final intensity after addition of 5% Triton-X-100 aqueous
solution (0.05 mL), and intensity at any time, respectively: (---)
[DPPC]/[6Ao2EPC] ) 20, (‚‚‚) [DPPC]/[Ao8H] ) 5, (s) [DPPC]/[6-
Ao2EPC] ) 10.
surfactants. Figure 12 compares the fluorescence profiles of
CF release upon irradiation at 365 nm for DPPC vesicles
containing different percentages of 6Ao2EPC and Ao8H. When
a 1:10 mixture of Ao2EPC/DPPC (under conditions where
6
∼30% of the APL is aggregated) is prepared with trapped CF,
irradiation of the vesicles with 365 nm light for only a short
time (40 s) results in total release of the CF concurrent with
trans-cis isomerization of the azobenzene.49 By controlling
the length and power of irradiation, partial or total release of
CF can be achieved. As a standard, total release can be achieved
by addition of Triton-X-100.29 In contrast, when vesicles
containing mostly dimeric azobenzene (a 1:20 mixture of 6Ao2-
EPC/DPPC with similar optical density, where less than 10%
APL is aggregated) or monomer (a 1:5 mixture of Ao8H/DPPC
where only monomer is expected) are irradiated, there is only
a small amount of leakage in each case, even when most of the
trans-azobenzene has been isomerized to cis. Irradiation with
365 nm light has almost no effect (as anticipated) upon the
release of CF from pure DPPC vesicles. The results indicate
that photoisomerization of the aggregates of the APL’s is much
more effective in promoting reagent release than an isolated
photoisomerization process in a nonaggregated azobenzene or
isolated dimer. Ao8,6Ao2EPC and A4EPC exhibit behavior
4
Aggregate Structures and Possible Origins of Stablization
for Aggregates. As has been done previously for monolayers
of stilbene,17c styrylthiophene fatty acids,41 and squaraine
derivatives,19 Monte Carlo cooling methods were applied to
determine the apparent global and nearby local minimum
structures of monolayer clusters of AFH’s. The detailed
procedures for the simulation have been reported elsewhere.50
The apparent global minimum and several nearby lying local
minima were found and the results are listed in Table 6 along
with some calculated parameters and experimental data. Exciton
shift spectra of the simulated structures were computed using
the extended dipole-extended dipole method of Kuhn and co-
workers.51 For this computation, a transition moment of 6.09
D was estimated from the oscillator strength of the azobenzene
monomer in solution at 348 nm, and the classical dipole length
similar to Ao2EPC. Consistent with the high tendency for
6
aggregation and larger H-aggregate size, a much lower percent-
age (less than 2%) of Ao8EPC in DPPC is required to promote
total CF release through photoisomerization.
An interesting feature of CF release induced by the photoi-
somerization of the aggregates of APL’s is that very rapid
release occurs only for the period of irradiation and the release
drops immediately after irradiation stops for all the cases. A
similar release pattern was observed for the interaction between
DPPC vesicles and Triton-X-100 micelles.29 This suggests that
the release of CF is due to the complete disruption of vesicle
structure. The formation of defects produced by the isomer-
ization of Ao8H monomer does not seem to promote any leakage
of CF. The small amount of release in DPPC/APL’s systems
with relative high ratio (for example, [DPPC]/[APL] ) 20) is
also caused by disruption of vesicles due to the trans-cis
isomerization of small number of aggregates. This disruption
mechanism is also supported by Cryo-TEM studies. Aqueous
(50) (a) Chen, H.; Law, K. Y.; Perlstein, J.; Whitten, D. G. J. Am. Chem.
Soc. 1995, 117, 7257. (b) Perlstein, J. J. Am. Chem. Soc. 1994, 116, 455.
(c) Perlstein, J. J. Am. Chem. Soc. 1994, 116, 11420.
(51) (a) Czikkely, V.; Fo¨rsterling, H.; Kuhn, H. Chem. Phys. Lett. 1970,
6, 11. (b) Kuhn, H.; Mob¨ius, D.; Bu¨cher, H. Techniques of Chemistry;
Weissberger, A., Rositer, B. W.; Eds.; Wiley: New York, 1973; Part B,
Vol. 1.
(49) No attempt has been made to determine a quantitative (e.g., quantum
yield) relationship between absorbed light intensity and CF release.

9156 J. Am. Chem. Soc., Vol. 119, No. 39, 1997
Song et al.
Figure 13. Cryo-TEM micrographs: mixed DPPC/trans-6Ao2EPC (10/1) vesicles before (a, upper left) and after (b, upper right) irradiation which
results in a total release of CF, and mixed POPC/chol/trans-6Ao2EPC (10/10/1) vesicles before (c, lower left) and after (d, lower right) irradiation
which results in partial release of trapped CF.
was estimated from the carbon-carbon end-end distance for
trans-azobenzene of 10.43 Å. The dielectric constant was
assumed to be 2.5. The computation was found to converge
for a layer structure of 14 × 14 unit cells. The glide layers
show a Davydov splitting whose oscillator strength ratio was
estimated from the resultant vector sum and difference of the
classical transition moments. The lowest energy structure which
shows closest agreement with the experimental result, is the
first local minimum glide layer for Ao8H. An overhead view
of this glide layer (showing only azobenzenes) is shown in
Figure 15 along with schematic representation of a trimer
isolated from the first local minimum structure of Ao8H
monolayer. Although we have not been able to calculate the
possible structures of APL bilayers, similarities in their absorp-

Aggregates of Azobenzene Phospholipids
J. Am. Chem. Soc., Vol. 119, No. 39, 1997 9157
in the aggregate structures caused by the different position of
azobenzene chromophores in the fatty acid chains.
Ao8,6Ao2EPC was originally designed to favor J-aggregate
formation because its two azobenzenes have six offset methylene
groups which presumably should favor the slipped stack
arrangement. Surprisingly, similar H aggregates are formed,
which implies that the H aggregates may be more stable than
the corresponding J aggregates as was also observed for the
mixture of squaraine derivatives at air-water interfaces. In
those studies,21 the J aggregates formed during the first
compression eventually completely transformed into H ag-
gregates after a few compression-decompression cycles. Many
factors such as the spontaneous formation of aggregates in the
AFH monolayers at the air-water interface without compression
and the persistence of the aggregates in the APL solutions
moderately diluted in saturated phospholipid vesicles suggest
the relative stability of the aggregates. The glide arrangement
of the azobenzenes in the aggregates is reasonable in terms of
the likely maximization of the weakly σ-π attractive interaction
and minimization of π-π repulsion54 between stilbene chro-
mophores. Such maximization of σ-π attraction and minimi-
zation of π-π repulsion may help stablize the aggregates.
Another interesting aspect about the small H aggregates of
azobenzene is that the energetically favored enantiomer for
APL’s with chromophores far away from head groups have
opposite chirality to the favored one for APL’s with chro-
mophores close to the head groups. The apparent ICD spectrum
reflects only the enantiomer with excess population. Their
relative stability for the two enantiomers of the aggregates are
possibly determined by the chirality of head group, the position
of chromophores, and environments in which they are formed.
Similar to the stilbene dimer,17 the azobenzene dimer also
does not exhibit significant shift in absorption while the trimer
gives a large blue shift. It may be attributed to some additional
interaction which is not possible for the dimer. Higher stability
of the unit aggregates over the dimer for stilbene and squaraine
also supports the possibility that other interaction (bonding)
besides hydrophobic and nonpolar interactions is involved. For
extended aggregates for some azobenzene fatty acids in LB
films, the observed largest shift in absorption spectra of H
aggregates is 53 nm. Several azobenzene fatty acids incorpo-
rated into LB films have been observed to have 30 nm shifts
which is the same as in vesicles. This seems reasonable that,
in the former case, the relatively extended aggregate is formed
while, in the later case, a mosaic of small aggregates (trimer)
is formed as in vesicle solutions.55,56
Figure 14. Fractional change in fluorescence intensity of trapped CF
in vesicles upon irradiation versus time. Io, Ie, and I have identical
meanings as in Figure 12: (s) [POPC]/6Ao2EPC] ) 10/1, (- - -)
[POPC]/[chol]/6Ao2EPC] ) 10/10/1, (- -s) [DPPC]/[6Ao2EPC] ) 10.
Table 6. Predicted Unit Cell, Surface Area/Molecule, and Spectral
Shifts for Ao8H Monolayer
layer
surface
areab λ1 λ2 ratiod energye
c
c
AFH type aa
ba
γa
Ao8H glide 4.70 10.89 90.0 25.59 331 349 0.01
glide 5.81 7.51 90.0 21.82 320 352 0.01
glide 4.98 9.47 90.0 23.58 328 350 0.01
glide 4.68 9.18 90.0 21.48 320 351 0.00
0.84
1.09
1.23
exptl
f
318
g
g
a Unit cell dimensions in Å; angle in deg. b Surface area/molecule
in Å2. c Exciton spectra peaks in nanometers; λ1 and λ2 for the glide
layer as a result of Davydov splitting. d Computed ratio of the oscillator
strength f(λ2)/f(λ1) for the Davydov splitting. e Energy above the
apparent global minimum in kcal. f Accurate surface area cannot be
obtained due to its relatively high solubility in water. g The peak is too
small to be identified.
tion spectra suggest that they may have similar aggregate
structures. On the basis of the measured aggregation numbers
and chiral nature, asymmetric trimers of an extended glide layer
are proposed as the smallest aggregate unit while the extended
structure should consist of a mosaic of trimers.52 This arrange-
ment is very similar to that concluded as the most likely structure
for spectrally blue-shifted aggregates observed for other aromatic
chromophores such as trans-stilbene, tolan, and squaraine
derivatives, and it may well be a general supramolecular
structure from which larger crystallites or aggregates may be
constructed.53 Excitonic interaction for the three degenerate
transition dipole moments within the nonsymmetric trimeric
“unit” aggregates produces excitonic states with three nonde-
generate energy levels.33,38 The relative intensities of the three
optical transitions will depend on the specific details of the
Summary
For all the amphiphilic azobenzene fatty acids and phospho-
lipids investigated in the present study, it is clear that azobenzene
may exist as several different species, depending upon its state
and the medium in which it is dispersed. Most of the AFH’s
and APL’s when dissolved in organic solvents are readily
identified as monomer from their absorption spectra. The
AFH’s in â- and γ-cyclodextrins and APL’s diluted in excess
DPPC or DMPC as well as 6Ao2EPC in organic solvents were
structural geometry.38 For Ao8EPC and A4EPC aggregates,
4
we observed two strong and one weak bands, whereas for -
6
Ao2EPC and Ao8,6Ao2EPC aggregates we observed only two
strong bands. The difference in ICD spectra of Ao8EPC and
4A4EPC aggregates compared to those for 6Ao2EPC and Ao8,6-
Ao2EPC aggregates indicates some perturbation and distortion
(54) (a) Arunan, E.; Gutowsky, H. S. J. Chem. Phys. 1993, 98, 4294.
(b) Hall, D.; Williams, D. F. Acta Crystallogr. 1975, A31, 56. (c) Schweitzer,
B. A.; Kool, E. T. J. Am. Chem. Soc. 1995, 117, 1863. (d) Winnik, F. M.
Chem. ReV. 1993, 93, 587. (e) Linse, P. J. Am. Chem. Soc. 1993, 115, 8793.
(f) Hunter, C. A.; Saunders, J. K. M. J. Am. Chem. Soc. 1990, 112, 5525.
(g) Jorgenson, W. L.; Severance, D. L. J. Am. Chem. Soc. 1990, 112, 4768.
(h) Shi, X.; Bartell, L. S. Phys. Chem. 1988, 92, 5667. (i) Pawliszyn, J.;
Szczesniak, M. M.; Scheiner, S. J. Phys. Chem. 1984, 88, 1726.
(55) Kawai, T.; Umemura, J.; Takenaka, T. Langmuir. 1989, 5, 1378.
(56) An alternative explanation is that the difference in absorption spectral
shift is attributed to the different orientation of azobenzene chromophores
in the aggregates.
(52) We infer that the extended arrays (mosaics) are regular glide or
herringbone structures, such as found in the simulation or in the crystals of
amphiphilic stilbenes,53 with similar intermolecular interaction to those in
the trimer “unit aggregate”. This, of course, does not explain the observation
that the limiting spectral shifts appear, in most cases, with the formation of
the unit aggregate. At present, we have no satisfactory explanation for this.
(53) Vaday, S.; Geiger, H. C.; Perlstein, J.; Whitten, D. G. J. Phys. Chem.
1997, 101, 321.

9158 J. Am. Chem. Soc., Vol. 119, No. 39, 1997
Song et al.

Aggregates of Azobenzene Phospholipids
J. Am. Chem. Soc., Vol. 119, No. 39, 1997 9159
found to form azobenzene dimers based on their small blue shift
in absorption similar to that of azobenzophanes and the weak
biphasic ICD signals observed in some cases. From the studies
of the monolayers, LB films of AFH’s and bilayers of APL’s
in aqueous solutions, it is seen that the formation of similarly
spectrally blue-shifted aggregates is a very general process as
observed for other aromatics such stilbene and styrylthiophene
as well as squaraine.21,57 It is observed that the position of the
azobenzene chromophore in the fatty acid chain of APL’s plays
important roles in controlling strength of the aggregation as
demonstrated by the unusually large and well-organized ag-
gregate and high Tc of Ao8EPC. The clear tendency of
azobenzene to form these aggregates reinforces the idea of a
relatively high stability for the “unit” H aggregate. The
proposed trimers for the unit aggregate are also supported by
the unusual triphasic ICD of A bands for 4A4EPC and Ao8EPC
vesicles.
Aggregation of chromophores results in some novel photo-
physical and photochemical properties such as the sharp, blue-
shifted absortpion spectrum and the less efficient trans-cis
photoisomerization compared with the isolated monomer. Pho-
toisomerization of the azobenzene aggregates provides a very
convenient way to enhance membrane permeability, which may
have some potential applications as useful photoregulatable
materials. We have also demonstrated the likely mechanistic
pathways for the reagent release from vesicles through photoi-
somerization of azobenzene aggregates and modulation of the
release mechanism by adjustment of the vesicle composition.
Acknowledgment. We are grateful to the U.S. National
Science Foundation (Grant CHE-9521048) for support of this
research.
(57) Chen, H.; Farahat, C. W.; Farahat, M. S.; Geiger, C. H.; Leinhos,
U.; Liang, K.; Song, X.; Penner, T. L.; Ulman, A.; Perlstein, J.; Law, K.
Y.; Whitten, D. G. Mater. Res. Soc. Bull: Organic Thin Film 1995, 20, 39.
JA971291N