Photoregulated transmembrane charge separation by linked spiropyran-anthraquinone molecules

Article abstract of DOI:10.1021/ja0545620

Amide-linked spiropyran-anthraquinone (SP-AQ) conjugates were shown to mediate ZnTPPS^4--photosensitized transmembrane reduction of occluded Co(bpy)₃³⁺ within unilamellar phosphatidylcholine vesicles by external EDTA. Overa

Full text of DOI:10.1021/ja0545620

Published on Web 12/23/2005
Photoregulated Transmembrane Charge Separation by Linked
Spiropyran-Anthraquinone Molecules
Linyong Zhu,† Rafail F. Khairutdinov,† Jonathan L. Cape,‡ and James K. Hurst*,†
Contribution from the Department of Chemistry and Institute of Biological Chemistry,
Washington State UniVersity, Pullman, Washington 99164-4630
Received July 9, 2005; E-mail: hurst@wsu.edu
Abstract: Amide-linked spiropyran-anthraquinone (SP-AQ) conjugates were shown to mediate ZnTPPS4-
-
3+
photosensitized transmembrane reduction of occluded Co(bpy)3 within unilamellar phosphatidylcholine
vesicles by external EDTA. Overall quantum yields for these reactions were dependent upon the isomeric
state of the dye; specifically, 30-35% photoconversion of the closed-ring spiropyran (SP) moiety to the
open-ring merocyanine (MC) form caused the quantum yield to decrease by 6-fold in the simple conjugate
and 3-fold for an analogue containing a lipophilic 4-dodecylphenoxy substituent on the anthraquinone moiety.
Transient spectroscopic and fluorescence quenching measurements revealed that two factors contributed
to these photoisomerization-induced changes in quantum yields: increased efficiencies of fluorescence
quenching of 1ZnTPPS4- by the merocyanine group and lowered transmembrane diffusion rates of the
merocyanine-containing redox carriers. Transient spectrophotometry also revealed the sequential formation
and decay of two reaction intermediates, identified as 3ZnTPPS4- and a species with the optical properties
of a semiquinone radical. Kinetic profiles for Co(bpy)33+ reduction under continuous photolysis in the presence
and absence of added ionophores indicated that transmembrane redox mediated by SP-AQ was
electroneutral, but reaction by the other quinone-containing mediators was electrogenic. The minimal reaction
3
mechanism suggested from the combined studies is oxidative quenching of vesicle-bound ZnTPPS4- by
the anthraquinone unit, followed by either H+/e- cotransport by transmembrane diffusion of SP-AQH• or,
for the other redox mediators, semiquinone anion-quinone electron exchange leading to net transmembrane
electron transfer, with subsequent one-electron reduction of the internal Co(bpy)33+. Thermal one-electron
reduction of Co(bpy)33+ by EDTA is energetically unfavorable; the photosensitized reaction therefore occurs
with partial conversion of photonic energy to chemical and transmembrane electrochemical potentials.
Introduction
nized systems for directed chemical syntheses that utilize solar
energy. Although this area of research is often called “bio-
Early work from several laboratories has established that
lipophilic quinones could mediate redox reactions between
compartmented reactants that are phase-separated by bilayer
membranes or bulk liquid membranes.1-5 Interest in these
reactions has been rekindled by recent studies from the Gust,
Moore, and Moore group, who have demonstrated that as-
semblies containing a vectorially organized donor-sensitizer-
acceptor electron transport chain and quinone pool are capable
of generating a proton electrochemical gradient of sufficient
magnitude to drive ATP synthesis by a membrane-incorporated
FoF1-ATP synthase.6,7 This latter demonstration suggests the
possibility of developing energy-transducing membrane-orga-
mimetic” or “bioinspired”, it is important to recognize that the
electrogenic transmembrane redox mechanisms functioning in
these systems are totally unlike the more elaborately constructed
biological Q-cycles that involve transmembrane translocation
of protons by membrane-localized quinones. The biological
systems contain quinone-binding sites at alternate ends of
membrane-spanning electron transport chains at which the
quinones undergo two sequential one-electron reductions to the
corresponding hydroquinones, or vice-versa.8,9 One consequence
of this organization is that protons are vectorially transported
via the Q-QH2 couple, thereby avoiding accumulation of O2-
reactive semiquinones. In contrast, the artificial systems lack
this capability and must generate free semiquinone in photo-
initiated one-electron reactions.
† Department of Chemistry.
‡ Institute of Biological Chemistry.
(1) Anderson, S. S.; Lyle, I. G.; Paterson, R. Nature 1976, 259, 147-148.
(2) Grimaldi, J. J.; Boileau, S.; Lehn, J.-M. Nature 1977, 265, 229-230.
(3) Futami, A.; Hurt, E.; Hauska, G. Biochim. Biophys. Acta 1979, 547, 583-
596.
Mechanisms by which the quinone pool mediates transmem-
brane redox in these artificial systems are not well-characterized.
To gain insight into these processes, we have coupled a
photoisomerizable spiropyran molecule to anthraquinone, the
(4) Joshi, N. B.; Lopez, J. R.; Tien, H. T.; Wang, C. B.; Lin, Y. Q. J.
Photochem. 1982, 20, 139-151.
(5) Yablonskaya, E. E.; Klabunovskii, E. I.; Shafirovich, V. Y.; Shilov, A. E.
Dokl. Akad. Nauk SSSR 1986, 286, 150-152.
(6) Steinberg-Yfrach, G.; Liddell, P. A.; Hung, S. C.; Moore, A. L.; Gust, D.;
Moore, T. A. Nature 1997, 385, 239-241.
(8) Okamura, M. Y.; Paddock, M. L.; Graige, M. S.; Feher, G. Biochim.
Biophys. Acta 2000, 1458, 148-163.
(9) Darrouzet, E.; Moser, C. C.; Dutton, P. L.; Daldal, F. Trends Biochem.
Sci. 2001, 26, 445-451.
(7) Steinberg-Yfrach, G.; Rigaud, J.-L.; Durantini, E. N.; Moore, A. L.; Gust,
D. A.; Moore, T. A. Nature 1998, 392, 479-482.
9
10.1021/ja0545620 CCC: $33.50 © 2006 American Chemical Society
J. AM. CHEM. SOC. 2006, 128, 825-835
825

A R T I C L E S
Zhu et al.
Scheme 1. Spiro(SP)-Mero(MC) Interconversion in a Spiropyran
Scheme 2. Synthetic Scheme for Preparing Spiropyran-Anthraquinone Dyads
intent being to identify and achieve control of the anthraquinone
location within the membrane, as well as the capability to alter
its bilayer permeability. Specifically, as illustrated in Scheme
1, UV illumination induces heterolytic photocleavage of the
C(spiro)-O bond, forming the more polar merocyanine form
of the dye. As a consequence of its decreased lipophilicity, one
expects that transmembrane diffusion of anthraquinone when
appended to the mero form will be slower than when appended
to the spiro form, and that the mero form will locate within a
more polar microphase region of the membrane, i.e., closer to
the aqueous-organic interface.10,11 The MC forms of spiropy-
rans undergo thermal ring-closing at rates that can vary from
several seconds to several hours, depending upon the medium
and structural composition of the dye.12,13 For the PC-solub-
lilized MC-AQ conjugates used in this study, t1/2 ≈ 2 h for
ring-closing to the SP-AQ form. Ultraviolet illumination
immediately prior to undertaking a kinetics experiment therefore
allows one to regulate the fractional composition of the MC
form in the membrane. The MC form possesses a visible
absorption band, whereas the SP form does not, so that they
are readily distinguishable spectroscopically. Because this band
is solvatochromic,14 the intramembrane location of MC can be
estimated; because photoisomerization occurs more rapidly than
molecular diffusion,14,15 transient spectrophotometry can be used
to determine the intramembrane location of SP as well, i.e., from
the MC spectrum obtained immediately following flash excita-
tion. Thus, appending spiropyrans to redox carriers allows one
to manipulate the location of reactants within the microphase
and the dynamics of individual reaction steps in a manner that
facilitates mechanistic analysis of charge transport mechanisms.
Experimental Section
Syntheses. All chemicals were obtained from commercial suppliers
as reagent grade materials and were used as received. The compound
listed as 4-dodecylphenol (Aldrich) was reported upon inquiry to be a
mixture of >90% 4-isomer and <8% 2-isomer. In-house deionized
water was further purified using a Milli-Q ion exchange/reverse osmosis
system. Linked spiropyran-anthraquinone dyads were synthesized
according to the sequences summarized in Scheme 2. To prepare
compound 7, the alkylphenol was first attached to the anthraquinone
via a diaryl ether bond, after which the isolated product was amide-
linked to a Fischer’s base containing an amino substituent. Condensation
of activated anthraquinones with 5-nitrosalicylaldehyde gave the final
products. Experimental details and analyses are given below.
Compound 1 (1-(4-Dodecylphenoxy)-5-chloroanthracence-9,10-
dione): 2.77 g (10 mmol) of 1,5-dichloroanthraquinone and 2.8 g (11
mmol) of 4-dodecylphenol were dissolved in 30 mL of DMF, 6.9 g of
K2CO3 were added, and the mixture was stirred overnight at 110 °C
under N2. The cooled mixture was filtered, and solvent was removed
(10) Khairutdinov, R. F.; Giertz, K.; Hurst, J. K.; Voloshina, E. N.; Voloshin,
N. A.; Minkin, V. I. J. Am. Chem. Soc. 1998, 120, 12707-12713.
(11) Khairutdinov, R. F.; Hurst, J. K. Langmuir 2001, 17, 6881-6886.
(12) Bertelson, R. C. In Organic Photochromic and Thermochromic Compounds;
Crano, J. C., Guglielmetti, R. J., Eds.; Plenum Press: New York, 1999;
Vol. 1, pp 11-83.
(15) Wilkinson, F.; Worrall, D. R.; Hobley, J.; Jansen, L.; Williams, S. L.;
Langley, A. J.; Matousek, P. J. Chem. Soc., Faraday Trans. 1996, 92,
1331-1336.
(13) Go¨rner, H. Phys. Chem. Chem. Phys. 2001, 3, 416-423.
(14) Minkin, V. I. Chem. ReV. 2004, 104, 2751-2776.
9
826 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Spiropyran-Modulated Quinone-Mediated Transmembrane Redox
A R T I C L E S
from the filtrate by vacuum rotary evaporation. The residue was
dissolved in CH2Cl2 and purified by flash column chromatography on
silica gel to give 3.2 g (64%) of compound 1 as a yellow viscous fluid.
1H NMR (300 MHz, CDCl3): δ 0.4-1.3 (m, 25H, C12H25), 6.6 (m,
2H, 2′′′,6′′′-C6H6), 6.76 (m, 1H, 6-AQ), 6.92 (m, 2H, 3′′′,5′′′-C6H6),
7.2 (m, 2H, 3,7-AQ), 7.3 (m, 1H, 8-AQ), 7.56 (d, 1H, 2-AQ), 7.79 (m,
1H, 4-AQ) (labeling as defined for compound 7).
Compound 6 (9,10-Dihydro-N-(1,3-dihydro-1,3,3-trimethyl-6-
nitrospiro[2H-1-benzopyran-2,2′-(2H)-indole])-9,10-dioxoanthracene-
1-carboxamide) (SP-AQ): A solution of 1.08 g (2.6 mmol) of
compound 4 and 0.43 g (2.6 mmol) of 2-hydroxy-5-nitrobenzaldehyde
in 200 mL of MeOH was refluxed for 4 h. After removal of solvent by
rotoevaporation, the residue was dissolved in 100:3 (v/v) CH2Cl2/CH3-
OH and purified by flash chromatography on silica gel to give 1.02 g
(67%) of compound 6 (SP-AQ) (mp > 200 °C). MS m/z 571.5. Calcd
Compound 2 (5-(4-Dodecylphenoxy)-9,10-dihydro-9,10-dioxoan-
thracence-1-carbonitrile): 3.2 g of compound 1 (6.4 mmol) and 0.85
g of cuprous cyanide (9.5 mmol) were stirred at reflux in 50 mL of
dimethylacetamide for 4 h under N2. The hot solution was poured into
700 mL of vigorously stirred water to give the copper(I) complex of
the product. The crude complex was decomposed with 200 mL of
boiling 4 N nitric acid for 4 h to yield 2.9 g (92%) of compound 2 as
a yellow solid. 1H NMR (300 MHz, CDCl3): δ 0.4-1.3 (m, 25H,
C12H25), 6.65 (m, 2H, 2′′′,6′′′-C6H6), 6.82-7.2 (m, 3H, 6-AQ + 3′′′,5′′′-
C6H6), 7.3 (m, 2H, 3,7-AQ), 7.6 (t, 1H, 8-AQ), 7.78 (d, 1H, 4-AQ),
8.2 (d, 1H, 2-AQ) (labeling as defined for compound 7).
for C34H25N3O6: C, 71.44; H, 4.41; N, 7.35. Found: C, 71.28; H, 4.84;
N, 7.14. 1H NMR (300 MHz, d6-DMSO): δ 1.15 (s, 3H, 8′′-CH3-
indoline), 1.23 (s, 3H, 8′′-CH3-indoline), 2.67 (s, 3H, 1′′-CH3-indoline),
6.01 (d, 1H, 2′-oxazine), 6.64 (d, 1H, 3′′-indoline), 6.9 (d, 1H,
3′-oxazine), 7.35 (d, 1H, 8′-oxazine), 7.58 (m, 2H, 4′′,6′′-indoline), 7.9-
8.05 (m, 3H, 3,6,7-AQ), 8.2-8.5 (m, 5H, 2,4,5,8-AQ; 7′-oxazine), 8.78
(s, 1H, 5′-oxazine), 10.55(s, 1H, NH).
Compound 7 (5-(4-Dodecylphenoxy)-9,10-dihydro-N-(1,3-dihy-
dro-1,3,3-trimethyl-6-nitrospiro[2H-1-benzopyran-2,2′-(2H)-indole])-
9,10-dioxoanthracene-1-carboxamide) (SP-AQ-C12): A solution of
0.205 g (0.3 mmol) of compound 5 and 0.05 g (0.3 mmol) of 2-hydroxy-
5-nitrobenzaldehyde in 20 mL of MeOH was refluxed for 4 h. After
removal of solvent by rotoevaporation, the residue was dissolved in
100:2 (v/v) CH2Cl2/CH3OH and purified by flash chromatography on
silica gel to give 0.1 g (40%) of compound 7 (SP-AQ-C12) (mp 112-
114.5 °C). MS m/z 831.9. Calcd for C52H53N3O7: C, 75.07; H, 6.42;
Compound 3 (5-(4-Dodecylphenoxy)-9,10-dihydro-9,10-dioxoan-
thracence-1-carboxylic Acid): 2.8 g of compound 2 (5.7 mmol) were
added to 200 mL of 70% sulfuric acid, and the mixture was stirred at
reflux for 1.5 h, after which the hot solution was poured onto ice. The
precipitate was collected by filtration, washed with water, and
resuspended in 200 mL of CH2Cl2. This slurry was filtered, and the
filtrate was dried over anhydrous sodium sulfate. The crude product
was isolated by rotoevaporation, dissolved in 100:5 (v/v) CH2Cl2/CH3-
OH, and purified by flash chromatography on silica gel to give 1.3 g
1
(45%) of compound 3. H NMR (300 MHz, CDCl3): δ 0.4-1.3 (m,
25H, C12H25), 6.65 (m, 2H, 2′′′,6′′′-C6H6), 6.83 (s, 1H, 6-AQ), 6.92
(m, 2H, 3′′′,5′′′-C6H6), 7.15-7.4 (m, 3H, 3,7,8-AQ), 7.65 (s, 1H, 4-AQ),
7.95 (s, 1H, 2-AQ) (labeling as defined for compound 7).
Compound 4 (9,10-Dihydro-N-(1,3,3-trimethyl-2-methylenein-
dolin-5-yl)-9,10-dioxoanthracene-1-carboxamide): A mixture of 1.73
g (9.2 mmol) of 5-amino-1,3,3-trimethyl-2-methyleneindoline, 1.98 g
(9.2 mmol) of Proton-Sponge (1,8-bis(dimethylamino)naphthalene), and
2.5 g (9.2 mmol) of anthraquinone-1-carbonyl chloride in 150 mL of
dry THF was stirred overnight at room temperature under N2. Following
filtration, the crude product was isolated from the filtrate by rotoevapo-
ration and purified after dissolution in 100:3 (v/v) CH2Cl2/CH3OH and
flash column chromatography on silica gel to give 1.9 g (52%) of
compound 4. 1H NMR (300 MHz, CDCl3): δ 1.32 (s, 6H, 8′′,8′′-CH3-
indoline), 3.03 (s, 3H, 1′′-CH3-indoline), 3.85 (s, 2H, 9′′-CH2-indoline),
6.43 (d, 1H, 3′′-indoline), 7.36 (dd, 1H, 4′′-indoline), 7.51 (d, 1H, 6′′-
indoline), 7.76 (m, 2H, 6,7-AQ), 8.2 (m, 2H, 5,8-AQ), 8.28 (m, 2H,
3,4-AQ), 8.42 (s, 1H, NH), 8.59 (d, 1H, 2-AQ) (labeling as defined
for compound 6).
Compound 5 (5-(4-Dodecylphenoxy)-9,10-dihydro-N-(1,3,3-tri-
methyl-2-methyleneindolin-5-yl)-9,10-dioxoanthracene-1-carbox-
amide): A solution of 6 g (11.7 mmol) of compound 3 and 2.22 g
(12.6 mmol) of 2-chloro-4,6-dimethoxy-1,3,5-triazine in 150 mL of
dry THF was stirred in an ice bath under N2. 1.35 mL (12.6 mmol) of
N-methylmorpholine was then slowly syringe-injected at 4 °C. The
filtrate from this reaction was then added to 1.47 g (7.8 mmol) of
5-amino-1,3,3-trimethyl-2-methyleneindoline and 1.35 mL (12.6 mmol)
of N-methylmorpholine in 50 mL of THF. This solution was stirred at
4 °C for 2 h and then warmed to room temperature for 24 h. After
removal of solvent, the residue was dissolved in 100:3 (v/v) CH2Cl2/
CH3OH and purified by flash column chromatography on silica gel to
N, 5.05. Found: C, 74.89; H, 6.40; N, 5.20. 1H NMR (300 MHz,
CDCl3): δ 0.3-1.5 (m, 31H, C12H25 + 8′′,8′′-CH3-indoline), 2.72 (s,
3H, 1′′-CH3-indoline), 5.85 (d, 1H, 2′-oxazine), 6.48 (d, 1H, 3′′-
indoline), 6.8 (dd, 2H, 2′′′,6′′′- C6H6), 7.02 (d, 2H, 3′′′,5′′′-C6H6), 7.11
(m, 1H, 3′-oxazine), 7.3 (m, 2H, 8′-oxazine + 4′′-indoline), 7.4-7.68
(m, 5H, 6′′-indoline + 3,6,7,8-AQ), 7.9-8.12 (m, 4H, 2,4,-AQ + NH
+ 7′-oxazine), 8.42 (s, 1H, 5′-oxazine).
13C NMR spectra for all of the synthesized compounds are given in
the Supporting Information. 9,10-Anthraquinone was reduced to the
corresponding diol by H2O2-induced reduction with sodium borohydride,
as described in the literature.16
1
Vesicle Preparation. Phosphatidylcholine (PC) was extracted from
fresh hen’s egg yolk and purified by column chromatography following
literature procedures.17 Aqueous suspensions of PC and the dopants
(SP, AQ, SP-AQ, or SP-AQ-C12) in 40 mM Tris buffer, pH 8.0,
give 2.8 g (52.5%) of compound 5. H NMR (300 MHz, CDCl3): δ
0.3-1.4 (m, 31H, C12H25 + 8′′,8′′-CH3-indoline), 3.03 (s, 3H, 1′′-CH3-
indoline), 3.85 (s, 2H, 9′′-CH2-indoline), 6.42 (d, 1H, 3′′-indoline), 6.96
(m, 2H, 2′′′,6′′′-C6H6), 7.07 (s, 1H, 6-AQ), 7.2-7.4 (m, 4H, 3′′′,5′′′-
C6H6 + 4′′,6′′-indoline), 7.4-7.68 (m, 3H, 3,7,8-AQ), 7.9 (m, 1H,
4-AQ), 7.98 (s, 1H, 2-AQ), 8.04 (s, 1H, NH) (labeling as defined for
compound 7).
(16) Panson, G. S.; Weill, C. E. J. Am. Chem. Soc. 1955, 22, 120-121.
(17) Singleton, W. S.; Gray, M. S.; Brown, M. L.; White, J. L. J. Am. Oil Chem.
Soc. 1965, 42, 53-56.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 827

A R T I C L E S
Zhu et al.
were formed by first mixing the components in CHCl3 and rotoevapo-
rating the solution to form a surfactant surface film, which was then
detached from the flask and hydrated by introduction of the aqueous
buffer and repeated freeze-thaw cycling. Unilamellar vesicles with
average diameters of 100 nm were prepared by high-pressure extru-
sion;18 dopant molecule/PC vesicle ratios used in the various experi-
ments ranged from ∼25-3000.
of ZnTPPS4- in the filtrate and retentate, measured spectrophotometri-
cally at the Soret band maximum (ꢀ421 ) 6.8 × 105 M-1 cm-1).20
Computational Methods. All calculations were performed in the
Gaussian 03 suite of programs.20 Ab initio determination of the redox
potential for the anthraquinone moiety of compound 7 was determined
using the method of Fu and co-workers.21 Geometry optimizations and
frequency calculations of compound 7 truncated with methyl ether and
dimethyl amino groups at positions 5 and 5′′, respectively, were carried
out at the B3LYP/6-31G(d, p) level. Single-point energy calculations
using the PCM solvation model were carried out at the B3LYP/6-311G-
(d, p) level using dielectric constants of 78 (water), 46 (dimethyl
sulfoxide), 36.4 (acetonitrile), 20.7 (acetone), 8.9 (methylene chloride),
4.9 (chloroform), and 2 (cyclohexane). All redox potentials are
referenced the Fc+/Fc couple extrapolated to the dielectric of DMSO
(E° ) 0.570 V vs NHE).22
Optical Measurements. Continuous photolysis experiments were
performed using filtered light from a 1.5-kW xenon lamp. For
experiments involving spiro h mero photoisomerization, aqueous
CuSO4 and either Schott UG11 or Corning OG590 filters were used to
provide broadband illumination in the UV (260 nm < λ < 390 nm) or
visible (540 nm < λ < 650 nm) regions, respectively. For quantum
yield measurements a narrow band-pass interference filter with
transmission at 420 nm was used to selectively excite the Soret band
of ZnTPPS4-. The filtered light was then passed via an optical fiber
bundle to a reaction cuvette that was mounted in a Hewlett-Packard
8452 diode array spectrophotometer interfaced to a ChemStation data
acquisition/analysis system. Light intensities were measured with a
calibrated PowerMax 500D laser power meter maintained at room
temperature; with the UG11 filter in place, the values obtained were
(0.5-5) × 10-8 einstein/(cm2 s).
Results and Discussion
Photochromic Behavior of SP-AQ Dyads in PC Vesicles.
In earlier studies, we had found that simple amphiphilic
spiropyrans and structurally similar spirooxazines underwent
reversible photoinitiated ring-opening-closing reactions, as well
as thermal ring-closing reactions, that were similar to those
occurring in homogeneous solution.10,11 However, one distin-
guishing feature of the reactions in vesicles is that rapid ring-
opening14,15 was often followed by a slower reaction on
millisecond time scales. This reaction was not observed for the
corresponding reactions when vesicles were absent and was
attributed to intramembrane structural reorganization following
generation of the more polar MC form of the dye. From the
solvatochromic properties of the MC visible band, we inferred
that this step involved an increase in the medium polarity
surrounding the dye; this could arise from either physical
disruption of the membrane bilayer structure, allowing greater
access by solvent, or relocation of the dye itself from the
nonpolar membrane interior toward the membrane-aqueous
interface. Rates of K+ efflux from SP-containing DHP vesicles
were found to decrease upon photoisomerization to the MC
form, consistent with the latter, but not the former, interpreta-
tion.11
SP-AQ and SP-AQ-C12 are water insoluble and are
strongly incorporated into PC vesicles during their formation.
The photoresponses of both these compounds in the vesicles to
continuous illumination were comparable to those observed for
simple PC-bound spiropyrans.11 The nature of the optical
changes that occurred is illustrated for SP-AQ in Figure 1,
where the spectra obtained under ultraviolet and visible il-
lumination are compared. These transformations were fully
reversible over several alternating cycles of UV and visible light,
and conversion rates were proportional to the intensities of
illuminating light. Slow thermal ring-closing also occurred,
which in PC liposomes followed first-order decay with an
apparent rate constant of k ) 9 × 10-5 s-1 at 23 °C. The MC
visible absorption maxima were used to estimate the microphase
location of the chromophores within the membrane. The
positions of these bands in homogeneous solutions exhibited
an approximately linear dependence upon the solvent dielectric
constant (ꢀ) over the range ꢀ ) 2-32, i.e., λmax ) a1 + a2 ×
Absorption spectra and decay kinetics of photoinduced intermediates
were measured by laser flash photolysis using a transient spectropho-
tometer whose general characteristics have been described.19 The third
harmonic (355 nm) from a Continuum Surelite III Nd:YAG laser was
used as the excitation source; optical absorbances of samples at 355
nm were maintained below 0.1. Fluorescence measurements were made
using a SPEX Fluorolog-3 spectrofluorometer.
Other Measurements. Mass spectra were determined by MALDI-
TOF in 2,5-dihydroxybenzoic acid using an Applied Biosystems
Voyager DE RP instrument.
Cyclic voltammograms were obtained in dry dimethyl sulfoxide
solutions containing 0.1 M tetrabutylammonium tetrafluoroborate using
a Princeton Applied Research model 273 potentiostat/galvanostat.
Measurements were made in a standard two-compartment, three-
electrode cell consisting of a glassy carbon working electrode, a coiled
platinum wire auxiliary electrode, and a Ag/0.1 M AgNO3 reference
electrode in DMSO, which was isolated from the reaction chamber by
a Vycor frit. Half-cell potentials were measured at scan rates of 50-
200 mV s-1; ferrocene was added as an internal redox standard.
The extent of binding of ZnTPPS4- to the PC vesicles was
determined by ultrafiltration using Centricon-10 centrifugal microcon-
centrators.19 This method relies upon forcing solution through a
semipermeable membrane (in this case, one with a 10 000 molecular-
weight cutoff), which permits unrestricted passage of unbound ZnTPPS4-
but prevents permeation by the vesicles and any ZnTPPS4- bound to
them. Comparison of ZnTPP4- concentrations in the exudate and
retentate allows direct measurement of the fraction that is bound. In a
typical experiment, 2.5 mL of a vesicle suspension containing
ZnTPPS4- was spun for 2.5 min at 4000 rpm in a Sorvall RC-2B
refrigerated centrifuge fitted with an SS-34 rotor (∼5800 g). Under
these conditions, about 0.1 mL of solution containing unbound
ZnTPPS4- passed through the membrane and was collected. Because
less than 5% of the solution volume was allowed to pass through the
membrane during sample collection, the vesicle concentration in the
retentate was essentially unchanged; consequently, the concentration
of ZnTPPS4- in the exudate could be taken as equal to that of the
unbound reagent in the retentate. The portion of PC-bound ZnTPPS4-
in retentate was then calculated as the difference in total concentrations
(20) Kalyanasundaram, K.; Neumann-Spallart, M. J. Phys. Chem. 1982, 86,
5163-5169.
(21) R. C., Frisch, M. J.; et al. Gaussian 03; Gaussian, Inc.: Wallingford CT,
2004 (complete citation given in the Supporting Information).
(22) Fu, Y.; Liu, L.; Yu, H. Z.; Wang, Y. M.; Guo, Q. X. J. Am. Chem. Soc.
2005, 127, 7227-7234.
(18) Mayer, L. D.; Hope, M. J.; Cullis, P. R. Biochim. Biophys. Acta 1986,
858, 161-168.
(19) Lymar, S. V.; Khairutdinov, R. F.; Soldatenkova, V. A.; Hurst, J. K. J.
Phys. Chem. B 1998, 102, 2811-2819.
9
828 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Spiropyran-Modulated Quinone-Mediated Transmembrane Redox
A R T I C L E S
relative to their equilibrium values, although virtually no
temporal change was observed in the initially formed MC-
AQ (Figure 3).
The kinetic response following flash excitation is illustrated
in Figure 4 for SP-containing PC vesicles. The initial rapid rise
in absorbance, which occurred within the rise time of our
instrument, can be attributed to initial ring-opening steps and
subsequent equilibration of isomeric merocyanine forms. In
homogeneous solution, these reactions are complete within the
first millisecond following flash excitation.13-15 The slow step,
which we attribute to environmental relaxation within the
vesicle, exhibited first-order decay with relaxation times at 23
°C of τ ) 0.3 ms for MC and τ ) 1.2 ms for MC-AQ-C12.
The relatively slow relaxation time and smaller spectroscopic
shift of MC-AQ-C12, as well as the absence of a detectable
relaxation for MC-AQ, may reflect “anchoring” effects of the
lipophilic substituents, which would thermodynamically oppose
movement of the molecule toward the more polar microphase
regions of the bilayer.
Figure 1. Absorption spectra of SP-AQ in PC vesicles under continuous
illumination with filtered UV (red) and visible light (black). The merocya-
nine-spyropyran difference absorption spectrum is given in green. Condi-
tions: 0.23 mM SP-AQ, ∼80 nM PC (5 mg/mL) in 40 mM Tris/Cl, pH
8.0.
Fluorescence Quenching of ZnTPPS4-. Although ZnTPPS4-
is nominally water-soluble, ultrafiltration studies indicated that,
under the conditions of these studies, 80 ( 5% of the zinc
porphyrin was bound to the formally neutral PC vesicles. This
behavior is consistent with previous measurements reported for
PC vesicle suspensions27 but differs markedly from that of
similar suspensions of anionic dihexadecyl phosphate (DHP)
vesicles, where electrostatic repulsion between ZnTPPS4- and
the negatively charged membrane interface precludes substantive
adsorption of the metalloporphyrin.28 Addition of SP caused
no change in the intrinsic ZnTPPS4- fluorescence from either
methanolic solutions or aqueous suspensions of PC vesicles.
However, as illustrated in Figure 5a, partial conversion to MC
by preillumination with UV light did cause significant quenching
of the porphyrin fluorescence. The energy of the first excited
singlet state of PC vesicle-bound ZnTPPS4- was estimated to
be 2.07 eV from the average of frequencies of the Q(0,0)
absorption and emission bands. Similar calculations based upon
absorption and emission maxima gave first-excited singlet
energies of 2.25, 2.13, and 1.98 eV for vesicle-bound MC, MC-
AQ, and MC-AQ-C12, respectively. Some uncertainty is
introduced in the latter numbers because the distribution of
isomeric merocyanine forms14,15,29 is probably different for the
absorbing and fluorescing species.
Nonetheless, these data suggest that fluorescence quenching
of 1ZnTPPS4- by MC is not efficient, since it would be
endergonic by ∼200 mV in PC vesicles. This conclusion differs
from that reached in a similar study of a zinc porphyrin dyad
containing a structurally analogous spiropyran unit, where
quenching was attributed to porphyrin f merocyanine energy
transfer.29
To provide an estimate of the electrochemical potentials for
the various redox-active components in vesicles, half-wave
potentials were measured in DMSO by cyclic voltammetry.
Irreversible one-electron reduction and oxidation waves assign-
able to the spiropyran moiety were clearly evident in the
voltammograms, consistent with earlier reports;30 reduction
Figure 2. Dependence of λmax for MC (red circles), MC-AQ (black closed
circles), and MC-AQ-C12 (black open circles) upon the solvent dielectric
constant (ꢀ). The solvents used were as follows: cyclohexane (2.03),
chloroform (4.8), tetrahydrofuran (7.6), dichloromethane (9.08), pyridine
(12.4), 2-propanol (18.3), ethanol (24.3), and methanol (32.6).
ꢀ.10,11 This relationship is illustrated in Figure 2, for which best
fit values to the experimental data were obtained with a1 )
599 nm and a2 ) -2.5 nm for SP and a1 ) 613 nm and a2 )
-2.6 nm for SP-AQ and SP-AQ-C12. The positions of the
MC visible band maxima in PC vesicles are ∼500 nm for SP,
∼550 nm for SP-AQ, and ∼600 nm for SP-AQ-C12. These
values were used to estimate apparent dielectric constants of
40, 25, and 6, respectively, of the chromophore environments
within the bilayer (e.g., the horizontal line in Figure 2 represent-
ing λmax for SP-AQ intersects the upper line at ꢀ ) 25).
Modeling studies suggest that the dielectric strength drops
precipitously across the headgroup region of PC bilayers from
a value of ꢀ ) 78 in bulk H2O to ꢀ ≈ 6 within the acylcarbonyl
groups of the glycerol backbone, a lateral distance of ∼0.5
nm.24-26 Within the ∼3.0 nm thick interior hydrocarbon portion
of the bilayer, the dielectric strength is relatively constant with
ꢀ < 5. The positions of the MC bands are therefore sensitive
indicators of relatively small changes in microenvironments
within the interfacial region of the membrane. The data suggest
that attachment of the anthraquinone and dodecylphenoxy units
to SP cause progressive displacement of SP-AQ and SP-AQ-
C
12 toward the membrane interior, which is consistent with the
hydrophobic nature of these substituents.
Transient spectroscopy revealed that the visible bands of MC
and MC-AQ-C12 in PC vesicles immediately following laser
flash-induced SP f MC photoisomerization were red-shifted
(23) Zuman, P.; Meites, L. Electrochemical Data; Vol. A.; John Wiley & Sons:
New York, 1974.
(27) Kuhn, E. R.; Hurst, J. K. J. Phys. Chem. 1993, 97, 1712-1721.
(28) Hurst, J. K.; Lee, L. Y.-C.; Gra¨tzel, M. J. Am. Chem. Soc. 1983, 105, 7048-
7056.
(29) Bahr, J. L.; Kodis, G.; de la Garza, L.; Lin, S.; Moore, A. L.; Moore, T.
A.; Gust, D. J. Am. Chem. Soc. 2001, 123, 7124-7133.
(24) Mazeres, S.; Schramm, V.; Tocanne, J.; Lopez, A. Biophys. J. 1996, 71,
327-335.
(25) Wiener, M. C.; White, S. H. Biophys. J. 1992, 61, 434-447.
(26) Sanders, C. R., II.; Schwonek, J. P. Biophys. J. 1993, 65, 1207-1218.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 829

A R T I C L E S
Zhu et al.
Figure 3. Normalized absorption spectra of PC vesicles containing (left panel) MC-AQ (black) or MC-AQ-C12 (red) or (right panel) MC. Solid circles
are normalized absorbance changes measured at 10-6 s after applying a 5 mJ 355 nm laser pulse; solid lines are normalized absorption spectra obtained after
100 s of continuous UV illumination of the same suspensions. Conditions: [SP], [SP-AQ], or [SP-AQ-C12] ) 0.23 mM, [PC] ≈ 80 nM (5 mg/mL) in
40 mM Tris/Cl, pH 8.0.
shift the SP0/- potential to lower values as a consequence of
relocation to a less polar environment within the microphase.
However, these explanations cannot account for the differing
quenching capabilities of MC and SP in methanol. A Stern-
1
Volmer analysis was made of the quenching of ZnTPPS4-
fluorescence intensities by MC in methanol, i.e., using the
equation Io/I ) 1 + kq[MC]τ, where the fluorescence lifetime
(τ) is 1.6 ns.20 These analyses gave calculated bimolecular
quenching rate constants (kq > 1 × 1013 M-1 s-1) that were
several orders of magnitude above the diffusional limit of ∼2
× 1010 M-1 s-1, excluding dynamic quenching and implying a
mechanism involving static quenching within associated species.
Aggregation of bound reactants is a common feature of vesicle-
Figure 4. Normalized transient absorbance decay kinetics at 480, 500, and
580 nm of SP-doped PC vesicles. Data are averages of 8 individual
organized systems28,32-35 that could also complicate thermody-
namic analyses of these reactions. Thus, for all these reasons,
experiments. The solid lines are calculated curves using ∆Abs ) 1 + b(1
1
the conclusion that fluorescence quenching of ZnTPPS4- by
- exp(-t/τ)), where b is the wavelength-dependent value of the total
amplitudinal change. Conditions: 0.23 mM SP, ∼80 nM PC (5 mg/mL) in
MC and SP is primarily oxidative must be regarded as tentative.
When the spiropyran is covalently linked to an anthraquinone,
both ring-open and ring-closed forms of the dye are capable of
40 mM Tris/Cl, pH 8.0, 23 °C.
waves for the AQ unit were reversible, indicating a simple one-
electron reaction at the electrode to the AQ semiquinone anion.
Half-wave potentials, referenced against the Fc+/0 couple, are
listed in Figure 6. No difference was observed in either oxidation
or reduction waves when SP solutions were ∼70% photo-
isomerized to MC or when SP-AQ solutions were ∼40%
photoisomerized to MC-AQ, suggesting that the potentials for
the ring-open and ring-closed forms were nearly identical; this
observation is consistent with previous reports on the redox
properties of similar spiropyrans.29,31 The appended AQ unit
had no effect upon the apparent E1/2 for SP or MC reduction;
however, it did cause the SP(MC) oxidation wave to shift
cathodically by almost 200 mV (Figure 6). Based upon these
1
quenching ZnTPPS4-; from the data presented in Figure 5b,
the relative quenching efficiencies in PC vesicles are estimated
as MC-AQ > MC-AQ-C12 ≈ SP-AQ > SP-AQ-C12. The
acquired capacity for quenching by the ring-closed SP com-
pounds may reflect one-electron transfer directly from 1ZnTPPS4-
to the appended AQ moiety. Based upon the potentials given
in Figure 6, the reaction
1ZnTPPS4- + SP-AQ f ZnTPPS3- + SP-AQ•- (2)
should be exothermic by ∼0.5 V. In contrast, singlet-singlet
energy transfer to either the SP or AQ units is unfavorable by
at least 1 V, based upon the position of the lowest-energy
absorption band in PC vesicles (∼400 nm (Figure 1), for which
E ≈ 3.1 eV). In support of reaction 2, ∼15% quenching of
1ZnTPPS4- fluorescence was also observed when an equivalent
amount of AQ was incorporated into the vesicles (data not
1
potentials, it appears that oxidative quenching of ZnTPPS4-
by MC (and SP), i.e.,
1ZnTPPS4- + MC (SP) f ZnTPPS3- + MC•- (SP•-) (1)
is thermodynamically feasible (∆E ≈ 0.07 V) but that reductive
quenching is not (∆E ≈ -0.30 V). The absence of 1ZnTPPS4-
quenching by vesicle-bound SP, which should also be thermo-
dynamically allowed, might be attributed to greater penetration
of the bilayer by SP, which could increase the effective electron-
transfer distance, thereby causing the redox reaction to become
noncompetitive with normal deactivation pathways, or could
1
shown). Singlet-singlet energy transfer from ZnTPPS4- is
considerably less endergonic for MC-AQ than MC and is
exergonic for MC-AQ-C12. In principle, then, all three
mechanisms (energy transfer to MC and oxidative quenching
(32) Cellarius, R. A.; Mauzerall, D. Biochim. Biophys. Acta 1966, 112, 235-
255.
(33) Kunitake, T.; Ihara, H.; Okahata, Y. J. Am. Chem. Soc. 1983, 105, 6070-
6078.
(30) Alberti, A.; Barberis, C.; Campredon, M.; Gronchi, G.; Guerra, M. J. Phys.
Chem. 1995, 99, 15779-15784.
(31) Zhi, J. F.; Baba, R.; Hashimoto, K.; Fujishima, A. J. Photochem. Photobiol.
A 1995, 92, 91-97.
(34) Nakashima, N.; Morimitsu, K.; Kunitake, T. Bull. Chem. Soc. Jpn. 1984,
57, 3252-3257.
(35) Yamazaki, I.; Tamai, N.; Yamazaki, T. J. Phys. Chem. 1990, 94, 516-
525.
9
830 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Spiropyran-Modulated Quinone-Mediated Transmembrane Redox
A R T I C L E S
Figure 5. Normalized emission spectra of ZnTPPS4- in PC vesicle solutions. (a) (black) no quencher; (red) [SP] ) 20 µM; (pink) [SP] ) 6 µM, [MC] )
14 µM. (b) (black) no quencher; (blue) [SP-AQ-C12] ) 20 µM; (light blue) [SP-AQ-C12] ) 14 µM, [MC-AQ-C12] ) 6 µM; (red) [SP-AQ] ) 20 µM;
(pink) [SP-AQ] ) 13 µM, [MC-AQ] ) 7 µM. All experiments were in 40 mM Tris/Cl, pH 8.0, with [PC] ≈ 80 nM (5 mg/mL).
Figure 7. Normalized decay kinetics at 460 nm of the transient absorption
following photoexcitation of ZnTPPS4- in suspensions of PC liposomes
(black) and SP-containing PC liposomes (blue). The red line is a biexpo-
nential fit of the experimental data with τ1 ) 0.5 ms and τ2 ) 1.4 ms.
Conditions: [PC] ≈ 80 nM (5 mg/mL), [SP] ) 0.23 mM (when present),
in 40 mM Tris/Cl, pH 8.0, 23 °C.
Figure 6. Half-cell potentials for redox components in DMSO. aE1/2(Fc+/0
)
b
) +0.13 V vs Ag/0.1M AgNO3(DMSO); estimated from the ZnTPPS4-
first excited singlet (2.07 V (PC vesicles)) and triplet state (1.62 V20)
energies.
quenching processes. Indeed, the rate constant calculated for
the triplet quenching reaction in methanol at 23 °C was only kq
) 5 × 107 M-1 s-1, which is compatible with an activation
barrier of several hundred millivolts. For this reaction, then,
the most likely quenching mechanism is oxidative, i.e.:
by both MC and AQ) could contribute to fluorescence quenching
in these systems.
3
Quenching of ZnTPPS4- by SP. The intrinsic lifetime of
3ZnTPPS4- + SP f ZnTPPS3- + SP•-
(3)
3ZnTPPS4- was determined by transient spectroscopy to be τ
) 2.1 ms in both methanol and PC vesicle suspensions; under
the prevailing experimental conditions, decay of 3ZnTPPS4- at
its absorption maximum of 460 nm was single-exponential,
indicating negligible contribution from bimolecular pathways36
or vesicle-bound aggregated porphyrins.28
3
The ZnTPPS4- decay kinetics in SP-doped vesicle suspen-
sions obeyed the equation: A460(t) ) A1 exp(-t/τ1) + A2 exp-
(-t/τ2) (Figure 7); best-fit values of the parameters obtained
were A1/(A1 + A2) ) 0.85, τ1 ) 0.5 ms, and τ2 ) 1.4 ms.
3
Quenching of ZnTPPS4- by the vesicle-bound SP could, in
Addition of SP to methanolic solutions caused the 3ZnTPPS4-
decay rate to increase, although the waveform remained single-
exponential; addition of SP to vesicles also increased the decay
rate, with the waveform now appearing markedly biexponential
(Figure 7). The energy of the first-excited triplet state of SP
principle, involve either or both vesicle-bound and aqueous
fractions of the zinc porphyrin. Specifically, the calculated
relaxation time for a diffusion-limited bimolecular reaction
between ZnTPPS4- and the vesicles under the experimental
conditions38 is τ ≈ 30 µs, which is an order of magnitude larger
than the intrinsic 3ZnTPPS4- lifetime. However, the rate constant
3
(∼2.1 eV)37 is considerably above that of ZnTPPS4- (∼1.6
eV),19 so that triplet-triplet energy transfer is unlikely to be
the quenching mechanism. However, based upon the data in
Figure 6, both oxidative quenching (∆E ≈ -0.38 V) and
reductive quenching (∆E ≈ -0.75 V) are also energetically
3
for reaction between ZnTPPS4- and SP in methanol is 104-
fold smaller than the calculated diffusion-limited constant for
(38) The diffusion-limited bimolecular quenching constant (kd) can be estimated
from the Debye equation, kd ≈ (4 × 10-3)π × Rv × D × No, where Rv is
3
unfavorable. Apparently, the greater reactivity of ZnTPPS4-
3
1
the vesicle radius, D is the ZnTPPS4- aqueous diffusion coefficient, and
than ZnTPPS4- toward SP can be attributed to the 106-fold
No is Avogadro’s number. When Rv ) 5 × 10-6 cm and D ) 2 × 105
cm2/s, kd ) 7.5 × 1011 M-1 s-1. The corresponding relaxation time, defined
as τd ) (kd × [PC])-1, is 33 µs when the effective concentration of vesicles
([PC]) is 8 × 10-8 M. Assuming that the average molecular weight of the
PC surfactant monomers is 730 amu and that their cross-sectional areas in
the outer and inner bilayer leaflets are 0.74 nm2 and 0.61 nm2, respectively
[Huang, C.; Mason, J. T. Proc. Natl. Acad. Sci. U.S.A. 1978, 75, 308-
310], this concentration corresponds to 5 mg/mL PC.
greater lifetime of the triplet-excited state, whose very slow
normal decay pathways allow expression of relatively inefficient
(36) Pekkarinen, L.; Linschitz, H. J. Am. Chem. Soc. 1960, 82, 2407-2411.
(37) Lashkov, G. I.; Ermolaev, V. L.; Shanlya, A. V. Opt. Spektrosk. 1966, 21,
546-549.
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 831

A R T I C L E S
Zhu et al.
quenching reaction was insensitive to the isomeric state of the
dye. Thus, in these assemblies as well, the fast step corresponds
3
to reactions of PC-bound ZnTPPS4-, and the slow step, to
decay of the unbound fraction of the zinc porphyrin.
3
The decay of ZnTPPS4- at 460 nm was accompanied by a
transient increase in absorbance at wavelengths above 600 nm;
because this absorption was relatively weak, it was not possible
to determine from the transient spectrum the identity of the
intermediate, which could be either ZnTPPS3- or ZnTPPS5- 39
.
However, two lines of evidence strongly suggest that this species
is ZnTPPS3-, formed by oxidative quenching by the an-
thraquinone moiety, e.g.,
3ZnTPPS4- + SP-AQ f ZnTPPS3- + SP-AQ•- (5)
Figure 8. Decay of transient absorption at 460 nm following photoexci-
tation of ZnTPPS4- in a (SP-AQ)-containing PC vesicle solution. The initial
phase of the decay curve is shown in the insert. The red line is a
biexponential fit with τ1 ) 20 µs and τ2 ) 1.7 ms. Conditions are the same
as those given in Figure 7.
First, the increase in decay rate was observed in all cases where
the AQ group was present, implying the presence of an
additional decay channel, i.e., reaction 5. Given the large rate
enhancement, this must be the most important pathway under
the experimental conditions. Consistent with the relatively large
quenching rate, reaction 5 is predicted from the thermodynamic
potentials given in Figure 6 to be slightly exothermic (∆E )
40 mV). Second, disappearance of the transient was accelerated
in the presence of EDTA by a reaction that was first-order in
the electron donor concentration. The reported porphyrin
potential E°(ZnTPPS3-/4-) ) 0.87 V20) is very similar to
potentials for EDTA estimated from electrochemical data,23,40
so that the reaction
the reaction with vesicles (kd ) 8 × 1011 M-1 s-1).38 One might
therefore reasonably assume that electron transfer from the
vesicle-bound SP is also relatively slow and consequently that
3ZnTPPS4- quenching only occurs for the vesicle-bound frac-
tion. This assumption is supported by the rate behavior; the
amplitude of the fast-decaying component (85%) corresponds
to the fraction of ZnTPPS4- that is vesicle-bound (80 ( 5%),
and the relaxation time for the slowly decaying component
corresponds to the intrinsic decay rate of 3ZnTPPS4-. No
changes were detected in the transient or ground-state absorption
spectra or intermediate decay rates after exposure to more than
100 laser flashes, indicating that, if oxidative quenching is the
mechanism, it was followed quantitatively by the reaction:
ZnTPPS3- + EDTA f ZnTPPS4- + EDTA+
(6)
should be nearly isoenergetic. Under the experimental conditions
used (5 mM EDTA, 2 µM ZnTPPS4-, 5 mg/mL PC, 20-100
ZnTPPS3- + SP•- f ZnTPPS4- + SP
The extinction coefficient of the ZnTPPS3- π-cation radical at
(4)
3
mJ/pulse), reductive quenching of ZnTPPS4- was negligible.
The bimolecular rate constant calculated for reaction 6 from
the enhanced rates of transient decay at 650 nm was k6 ≈ 105
460 nm is about one-third of the value for 3ZnTPPS4-
;
M-1 s-1
.
nonetheless, there was no evidence for accumulation of ZnTPPS3-
during 3ZnTPPS4- decay. Thus, either quenching is not oxidative
or, consistent with the thermodynamic potentials, back-electron
transfer (reaction 4) is more rapid than the oxidative quenching
step (reaction 3).
In the absence of exogenous electron donors, ZnTPPS3-
underwent first-order decay to regenerate the original absorption
spectrum; these relaxation times were moderately dependent
upon the identity of the vesicle-bound AQ-containing com-
pounds (τ ≈ 100 µs (SP-AQ-C12 or 30% MC-AQ-C12), τ
≈ 200 µs (30% MC-AQ), τ ≈ 300 µs (SP-AQ), τ ≈ 50 µs
(AQ)), presumably reflecting differences in back-electron-
transfer rates between products of the oxidative quenching
reaction, e.g.,
Quenching of 3ZnTPPS4- by AQ, SP-AQ, and SP-AQ-
C12. Biexponential decay of 3ZnTPPS4- was also observed when
AQ or its SP-linked analogues were incorporated into PC
vesicles. In these systems, the relaxation time of the fast
decaying component (τ1) was ∼10-fold smaller than that for
SP at equivalent loadings (Figure 8). As before, the magnitude
of τ2 corresponded to the intrinsic relaxation time of 3ZnTPPS4-
and, for all reactions, A1/(A1 + A2) ) 0.7-0.9. In contrast, the
relaxation rate for the fast component was dependent upon the
identity of the quinone, with τ1 ) 30 µs (AQ), 20 µs (SP-
AQ), and 150 µs (SP-AQ-C12) under conditions equivalent
to those given in Figure 8. The relatively slow quenching rate
constant for SP-AQ-C12 is consistent with conclusions drawn
from transient absorption spectra that this compound is bound
relatively deeply within the hydrocarbon bilayer (Figure 3).
Biphasic kinetics with nearly identical rate parameters were also
observed when the SP-AQ doped PC vesicles were preillumi-
nated with UV light to convert 30-35% of the dye to the mero
form, i.e., τ1 ) 25 µs and 180 µs, respectively, for MC-AQ
and MC-AQ-C12-containing vesicles, indicating that the
ZnTPPS3- + SP-AQ•- f ZnTPPS4- + SP-AQ (7)
The reaction of SP-AQ with 3ZnTPPS4- in methanol exhibited
qualitatively similar dynamical behavior to the reactions of
vesicle-bound SP-AQ. Specifically, an EDTA-reactive transient
intermediate was detected, whose spectrum was consistent with
3ZnTPPS4-; from the second-order decay rate of the intermedi-
ate, measured at either 460 or 650 nm, a back-electron-transfer
rate constant of k7 ) 6 × 109 M-1 s-1 was calculated. Rate
parameters for the various reactions studied are collected in
Table 1.
Photosensitized Transmembrane Oxidation-Reduction.
3+
One-electron reduction of Co(bpy)3 by EDTA, i.e.,
EDTA + Co(bpy)33+ f EDTA+ + Co(bpy)3
(8)
2+
9
832 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Spiropyran-Modulated Quinone-Mediated Transmembrane Redox
A R T I C L E S
Table 1. Apparent Redox Rate Constants for PC-Bound Reactants Measured by Transient Spectrophotometry
reaction
ka
3ZnTPPS4- f ZnTPPS4-
5 × 102 s-1
e2 × 103 s-1
3 × 104 s-1
5 × 104 s-1
4 × 104 s-1
7 × 103 s-1
6 × 103 s-1
3ZnTPPS4- + SP f ZnTPPS3- + SP•-
3ZnTPPS4- + AQ f ZnTPPS3- + AQ•-
3ZnTPPS4- + SP-AQ f ZnTPPS3- + SP-AQ•-
3ZnTPPS4- + SP(MC)-AQ f ZnTPPS3- + SP(MC)-AQ•- b
3ZnTPPS4- + SP-AQ-C12 f ZnTPPS3- + SP-AQ•--C12
b
3ZnTPPS4- + SP(MC)-AQ-C12 f ZnTPPS3- + SP(MC)-AQ•--C12
ZnTPPS3- + EDTA f ZnTPPS4- + EDTA•+
1 × 105 M-1 s-1 c
1.5 × 102 s-1
7 × 102 s-1
AQ•- + {Co(bpy)33+}i f AQ + {Co(bpy)3
}
i
2+ d,e
SP-AQH•+ {Co(bpy)33+}i f SP-AQ + {Co(bpy)32+}i + H+
d,f
i
SP(MC)-AQ-• + {Co(bpy)33+}i f SP(MC)-AQ + {Co(bpy)3
}
2 × 102 s-1
2+ b,d,e
i
SP-AQ•--C12 + {Co(bpy)33+}i f SP-AQ-C12 + {Co(bpy)3
SP(MC)-AQ•--C12 + {Co(bpy)33+}i f SP(MC)-AQ-C12 + {Co(bpy)3
}
2 × 102 s-1
2+ d,f
i
2+ b,d,e
}
5 × 101 s-1
i
a At 23 °C, determined from relaxation times under the prevailing conditions as described in the text. b Vesicles contained 30-35% of the merocyanine
form. c Apparent value, determined under the conditions given in Figure 8. d Rate-limited by semiquinone-mediated transmembrane redox. e As written,
electrogenic. f As written, electroneutral.
Figure 9. Co(bpy)32+ accumulation monitored at 320 nm under continuous illumination of the ZnTPPS4- Soret band. Conditions: [Co(bpy)33+] ) 25 µM,
[ZnTPPS4-] ) 2 µM, [EDTA] ) 5 mM, [PC] ≈ 80 nM (5 mg/mL), and [quinone] ) 2 µM in 40 mM Tris/Cl, pH 8.0. (a) SP-AQ (black squares) plus 20
µM CCCP (red circles) or 40 µM CCCP (green triangles); (b) SP-AQ-C12 (black squares) plus 20 µM CCCP (red circles) or 40 µM CCCP (green
triangles).
is substantially enderogonic, based upon the reported potential
of E°(Co(bpy)33+/2+) ) 0.35 V41 and an estimated potential of
≈ 1V for one-electron oxidation of EDTA obtained from
electrochemical data.23,40. Consistent with these potentials,
3+
EDTA and Co(bpy)3 are unreactive when mixed in aqueous
solutions. However, net ZnTPPS4--photosensitized one-electron
transfer was observed under certain conditions between EDTA
contained in the bulk medium and Co(bpy)33+ occluded within
the inner aqueous compartments of PC vesicles. Specifically,
this reaction required the presence of AQ or an AQ-containing
compound within the bilayer; substitution of SP or SP/MC
mixtures or omission of any of the reaction components
completely blocked Co(bpy)32+ formation. Representative results
Figure 10. Formation and decay of transient absorption at 590 nm following
photoexcitation of ZnTPPS4- in (SP-AQ)-containing PC vesicles without
obtained under continuous illumination of the ZnTPPS4- Soret
band are displayed in Figure 9, which illustrates the dependence
of the overall reaction rate upon the identity of the quinone-
containing molecule.
3+
occluded Co(bpy)3 (blue line) or with [Co(bpy)33+] ) 25 mM (black).
The red line is the best fit of experimental data using the equation and
parameters given in the text [EDTA] ) 6 mM; other conditions are the
same as those given in Figure 7.
Laser-flash spectrophotometry made in the presence of EDTA
revealed the appearance of weakly absorbing long-lived transient
species at 590 nm, whose decay was markedly accelerated when
to the following equation: A590(t) ) A1(1 - exp(-t/τ1) + A2
exp(-t/τ3), where the first and second terms are taken to be
the formation and decay of AQ•- species, respectively. The best-
fit value for τ1 (Figure 10) is 20 µs, corresponding to the
relaxation time measured for 3ZnTPPS4- decay. Best-fit values
obtained for τ3 were 1.5 ms and ∼130 ms in the presence and
absence of Co(bpy)33+, respectively; analyses of analogous
3+
the vesicles contained occluded Co(bpy)3 (Figure 10). This
wavelength is nearly isosbestic for ZnTPPS4- and either
ZnTPPS3- or ZnTPPS5- 39 but is a region where the semi-
quinone absorbs strongly.42 These data were analyzed according
(39) Neta, P. J. Phys. Chem. 1981, 85, 3678-3684.
experiments with AQ and SP-AQ-C12 gave τ3 values of 6.8
(40) Sˇtul´ık, K.; Vydra, F. J. Electroanal. Chem. 1968, 16, 385-395.
(41) Farina, R.; Wilkins, R. G. Inorg. Chem. 1968, 7, 514-518.
(42) Buschel, M.; Stadler, C.; Lambert, C.; Beck, M.; Daub, J. J. Electroanal.
Chem. 2000, 484, 24-32.
3+
and 6.4 ms, respectively, in the presence of Co(bpy)3
.
Remarkably, 30-35% conversion to the mero forms caused τ3
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 833

A R T I C L E S
Zhu et al.
2+
to increase severalfold, i.e., to 5.0 ms for MC-AQ and 20 ms
for MC-AQ-C12 in the presence of Co(bpy)3 (Table 1).
rates of formation and yields of Co(bpy)3 in the latter case
(Figure 9b) but had no effect upon transmembrane redox when
SP-AQ was the carrier (Figure 9a). Very similar results were
obtained when CCCP was replaced by the K+-selective ion
transporter valinomycin in K+-buffered media; however, addi-
tion of valinomycin to Na+-buffered media did not affect the
3+
Based upon the AQ/AQ•- reduction potential measured in
DMSO, electron transfer to Co(bpy)33+ is highly exergonic, with
∆E ) 1.5 V.43 Given that τ3 was 20-90-fold smaller when
Co(bpy)33+ was occluded within the vesicles, τ3 must be a good
approximation under these conditions to the relaxation time for
transmembrane reduction by an externally generated AQ semi-
quinone radical, i.e.,
3+
time course of Co(bpy)3 reduction, consistent with a mech-
anism whereby valinomycin collapses the developing electro-
chemical potential by selective transmembrane transport of K+.46
Reactions using vesicles that contained AQ or MC-AQ as redox
mediators also exhibited dynamical properties under continuous
illumination associated with electrogenic charge transport, i.e.,
rate acceleration in the latter stages of the reaction in the
presence of CCCP or K+-valinomycin. In contrast, although the
SP-AQ•- (or SP-AQH•)o + (Co(bpy)33+)i f
SP-AQ + (H+)i + (Co(bpy)32+)i (9)
3+
Because exergonic one-electron reduction of Co(bpy)3 in
3+
rate profile and extent of Co(bpy)3 reduction in MC-AQ-
homogeneous solution is very rapid,44 it also follows that the
relatively slow redox reaction measured as τ3 in these assemblies
is rate-limited by the transmembrane charge separation step.
The redox mechanism(s) could involve any combination of
electroneutral processes such as transmembrane diffusion of
AQH-containing carriers following uptake of a proton from the
external aqueous-organic interface and electrogenic processes
such as transmembrane diffusion of AQ•--containing carriers
or AQ/AQ•- electron exchange, leading to net translocation of
the electron across the bilayer. In closed bilayer membranes,
electrogenic processes can often be distinguished from electro-
neutral processes by the self-impeding nature of the former
reactions.44,45 Specifically, electron transport that is uncompen-
sated electrostatically by movement of other charges across the
bilayer will generate a transmembrane potential that opposes
further movement of charge as the reaction proceeds; this
developing potential leads to marked diminution in rate in the
latter stages of the reaction. Providing a pathway for movement
of charge-compensating ions, e.g., by adding an ionophoric
compound, will collapse the potential and accelerate transmem-
brane charge transport. However, if the reaction is electroneutral,
no transmembrane potential will develop and addition of an
ionophore will have negligible effect on the transport process.
Examples of both types of behavior are illustrated in Figure 9.
C12-containing vesicles were consistent with electrogenic elec-
tron transport, the reaction rates were not appreciably accelerated
by adding CCCP. It is possible that, at the very low dopant
loadings used in these studies, a significant side reaction of this
relatively immobile semiquinone depleted the carrier concentra-
tion and thereby lowered transport fluxes. Disproportionation
•-
of the semiquinone radical, i.e., 2 MC-AQ-C12 + 2H+
f
MC-AQ-C12 + MC-AQH2-C12, may contribute to loss of
the functioning carrier; the hydroquinone, anthracene-9,10-diol
(AQH2), when solubilized in PC vesicles, reacted only very
slowly with Co(bpy)33+ present in the external medium (t1/2
≈
1-2 × 104 s, [AQH2] ) 10-20 µM, [Co(bpy)33+] ) 20-30
µM, 23 °C), consistent with a high energetic barrier for one-
electron oxidation to the semiquinone.
The relaxation times for decay of the 590 nm transients
(Figure 10) are very similar to τ-values previously measured
for the corresponding reactions of uncharged N-alkyl-4-cyan-
opyridinium radicals,19 whereas diffusion of lipophilic ions such
as N,N-dialkyl-4,4,-bipyridinium monocation radicals is gener-
ally several orders of magnitude slower.44 Thus, the transmem-
brane redox rates are compatible with SP-AQ functioning as
a neutral cotransporter of electrons and protons, i.e., as SP-
AQH, but not with the semiquinone ions as mobile electron
transporters, i.e., as AQ•- and SP-AQ•--C12. More likely,
electronic charge in these cases is transported by AQ-AQ•-
exchange, or site-to-site electron hopping, across the bilayer.
The physical basis for the apparent difference in mechanisms
between SP-AQ and AQ or SP-AQ-C12 may reside in limited
access of the latter quinones to the membrane aqueous-organic
interface which, for AQ, is a consequence of its extreme
lipophilicity and, for SP-AQ-C12, reflects the immobilizing
effect of its dodecanyl “anchor”. We have noted very similar
effects of appending long alkyl chains to mobile viologens that
can function as electron transporters across bilayer membranes,
i.e., conversion from an electroneutral to an electrogenic
mechanism with addition of lipophilic ions being required to
achieve complete reaction.47 The influence of membrane mi-
cropolarity might also account for the ∼3-fold decrease in the
transmembrane redox rate constant that accompanies partial
conversion of SP to MC in the SP-AQ compounds (Table 1).
Specifically, the open-ring merocyanine form is more polar than
the closed-ring spiropyran form (Scheme 1), so that the
hydrocarbon interior of the membrane presents a greater
3+
In panel a, internal Co(bpy)3 was relatively rapidly and
completely reduced under continuous illumination of the
ZnTPPS4- photosensitizer when SP-AQ was the membrane
charge carrier, whereas the corresponding reaction using SP-
AQ-C12 (panel b) was only ∼50% complete over the same
illumination period and was characterized by a rapidly attenuat-
ing rate of reaction. Addition of the proton transporter carbonyl
cyanide m-chlorophenylhydrazone (CCCP) accelerated both
(43) Although the membrane interior presents a complex microheterogeneous
environment, the measured one-electron quinone potential in DMSO should
be a good approximation to the actual potential in PC membranes.
Specifically, polarographically determined values for the AQ/AQ•- potential
in various aprotic solvents were relatively independent of solvent dielectric
constant (ꢀ), varying by less than 200 mV over a range of ꢀ ) 13-47,
with a reported value of E1/2 ) -1.28 V (vs Fc+/0) in DMSO [Jaworski, J.
S.; Lesniewska, E.; Kaliowski, M. K. J. Electroanal. Chem. 1979, 105,
329-334]. Moreover, ab initio calculations made in the course of these
studies gave values for the potential that were virtually independent of
dielectric strength over the range ꢀ ) 78-21 (E° ) -1.36 to -1.42 V (vs
Fc+/0)), corresponding to the aqueous-organic interfacial region of the
membrane. At lower ꢀ, the calculated reduction potential became progres-
sively more negative, reaching E° ) -1.78 V when ꢀ ) 4.9, corresponding
to the hydrophobic membrane interior. Results of the computations are
appended as Supporting Information.
(44) Lymar, S. V.; Hurst, J. K. J. Phys. Chem. 1994, 98, 989-996.
(45) Hurst, J. K. In Kinetics and Catalysis in Microheterogeneous Media; Gratzel,
M., Kalyanasundaram, K., Eds.; Surfactant Science Series, Vol. 38; Marcell
Dekker: New York, 1991; pp 183-226.
(46) Nicholls, D. G.; Ferguson, S. J. Bioenergetics 3; Academic Press: New
York, 2002.
(47) Patterson, B. C.; Hurst, J. K. J. Phys. Chem. 1993, 97, 454-465.
9
834 J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006

Spiropyran-Modulated Quinone-Mediated Transmembrane Redox
A R T I C L E S
bound 1ZnTPPS4- that does not undergo static quenching, and
Φt is the quantum yield for singlet-triplet intersystem crossing
of 1ZnTPPS4-. For these calculations, Φt was assigned a value
of 0.84,20 and Φs was estimated from fluorescence quenching
data measured under the prevailing conditions after correcting
Table 2. Kinetic Parameters for Quinone-Mediated
Transmembrane Redoxa
τ7
(ms)
τ8
(ms)
1
compound
æb τ5-1/(τ5-1 + τ-
)
Xc
Φcalcd
Φexptl
AQ
0.73
0.99
0.99
0.99
0.93
0.92
0.05 0.8 0.26 0.06
0.0345
0.071
SP-AQ
0.48
0.24
0.74
0.3
0.2
0.1
1.5 0.26 0.12
5.0 0.12 0.0285 0.012
6.4 0.065 0.045 0.034
1
for the fluorescence arising from unbound ZnTPPS4- (Figure
MC-AQd
SP-AQ-C12
5). For this correction, the relative fluorescence quantum yield
of 1ZnTPPS4- in buffer was determined to be ∼46% of that of
the PC-bound zinc porphyrin. Calculated values of Φ are listed
in Table 2, along with the parameters used in the calculations.
Given the large number of parameters involved, whose experi-
mental uncertainties are (10-40%, the correspondence between
directly measured and calculated overall quantum yields is good;
in particular, experimentally determined trends involving indi-
vidual redox carriers are accurately reproduced.
e
MC-AQ-C12 0.58
0.1 20
0.052 0.028 0.013
a Conditions: [ZnTPPS4-] ) 2 µM, [EDTA] ) 5 mM, [quinone] ) 2
µM, [Co(bpy)33+]i ) 25 µM, [PC] ≈ 80 nM (5 mg/mL) in 40 mM Tris/Cl,
pH 8.0, 23 °C. b Corrected for fluorescence from unbound fraction (20%)
of 1ZnTPPS4- as indicated in the text. c X ) (k6[EDTA] + τ9-1)/(k6[EDTA]
-1
+ τ7 + τ9-1). d 0.7 µM MC-AQ + 1.3 µM SP-AQ. e 0.6 µM MC-
AQ-C12 + 1.4 µM SP-AQ-C12.
energetic barrier to its transmembrane diffusion. Similarly,
electron transfer from MC-AQ•--C12 to SP-AQ-C12 could
be energetically unfavorable within the membrane because it
involves transfer of charge into a less polar microenvironment.
This latter effect would not be evident from comparisons of
potentials in homogeneous solution.
Mechanistic Conclusions. The quantum yields for trans-
membrane charge transfer in these assemblies are relatively low
but consistent with earlier studies where it was found that
photosensitized transmembrane redox was efficient only when
the systems were organized so that oxidative quenching occurred
in the bulk aqueous phase or across the organic-aqueous
interface of the membrane, i.e., from free 3ZnTPPS4- to vesicle-
bound quencher.19,28,48 In the present case, examination of the
individual reaction steps revealed that the low quantum yields
was a consequence primarily of two factors, unproductive
quenching of 1ZnTPPS4- by the SP and MC groups on the dye
and relatively efficient back-electron transfer between the
semiquinone and ZnTPPS3- formed by oxidative quenching of
3ZnTPPS4- by AQ, SP-AQ, or SP-AQ-C12 (Table 2). In
contrast, oxidative quenching of 3ZnTPPS4- by the AQ moiety
occurred with near-unitary efficiency under the experimental
conditions. The 6-fold decrease in overall quantum yield for
transmembrane redox observed when SP-AQ was partially
converted to the MC form is also the product of two factors:
2-fold reduction in æ, indicating more efficient quenching of
1ZnTPPS4- by MC-AQ, and a 3-fold slower transmembrane
redox step (Table 1). The lower overall quantum yield for SP-
AQ-C12 compared to SP-AQ is attributable to its much lower
transmembrane charge-transfer rate, since, in this case, fluo-
rescence quenching was less extensive. The (2-3)-fold reduction
in overall quantum yield observed upon partial conversion to
MC-AQ-C12 can also be attributed to a combination of
increased 1ZnTPPS4- quenching and lower transmembrane
charge-transfer rate.
Quantum Yields for Transmembrane Redox. Overall
quantum yields for transmembrane oxidation-reduction were
determined from the initial rates of Co(bpy)33+ disappearance,
as described in the Experimental Section. To examine the effect
of partial conversion of the dyes to their mero forms, photo-
excitation of the ZnTPPS4- sensitizer at 420 nm was preceded
by 10-15 min of broadband UV illumination (260-390 nm).
Under these conditions, 30-35% of the bound dye was
converted to the mero form, as determined from the steady-
state optical absorption spectra. Quantum yields measured under
an equivalent set of conditions are listed in Table 2; the values
decrease upon partial conversion to the mero form and upon
addition of an alkyl chain to the molecule. The quantum yields
were analyzed quantitatively by considering the individual
reactions investigated in this study, including fluorescence
1
quenching of ZnTPPS4- (Figure 5) and reactions 5-7 and 9.
Reactions 3 and 4 were not explicitly considered because it was
evident from the transient decay profiles (cf., Figures 7 and 8)
3
that quenching of ZnTPPS4- by the AQ moiety was far more
efficient than either SP or MC. With the further assumptions
that only the vesicle-bound fraction of the photoexcited ZnTPPS4-
reacted with the dye and that semiquinone capable of functioning
as a redox mediator was formed only by oxidative quenching
of 3ZnTPPS,4- the quantum yield for Co(bpy)33+ reduction (Φ)
is given by:
Acknowledgment. The authors gratefully acknowledge help-
ful discussions with Drs. David M. Kramer (WSU), Scot
Wherland (WSU), and C. Michael Elliott (Colorado State)
concerning determination of nonaqueous potentials. Financial
support was provided by the Division of Chemical Sciences,
Office of Basic Energy Sciences, U.S. Department of Energy,
under Grant DE-FG03-99ER14943.
Φ ) æ × k5[SP-AQ]/(k5[SP-AQ] + τ-1) ×
(k6[EDTA] + τ9-1)/(k6[EDTA] + τ7-1 + τ9-1) (10)
3
where æ is the quantum yield for ZnTPPS4- formation and τ
3
) 2.1 ms is the intrinsic lifetime of ZnTPPS4-. The second
3
right-hand term in eq 10 is the fraction of ZnTPPS4- that
Supporting Information Available: Tabulations of 13C NMR
peaks for the synthesized compounds and the calculated one-
electron reduction potentials for an SP-AQ analogue. This
material is available free of charge via the Internet at
http://pubs.acs.org.
undergoes oxidative quenching and the third term is the fraction
of the semiquinone formed that escapes recombination with
ZnTPPS3- to reduce Co(bpy)33+. The value of æ was calculated
from the equation: æ ) Φs × Φt, where Φs is the fraction of
(48) Khairutdinov, R. F.; Hurst, J. K. J. Am. Chem. Soc. 2001, 123, 7352-
7359.
JA0545620
9
J. AM. CHEM. SOC. VOL. 128, NO. 3, 2006 835