Electron transfer through the hydrogen-bonded interface of a β-turn- forming depsipeptide

Article abstract of DOI:10.1021/ja981384k

Hydrogen-bonding networks are believed to play an important role in electron-transfer pathways in a protein medium. A porphyrin-quinone donor- acceptor compound with a depsipeptide bridge which forms a β-turn has been synthesized to study hydrogen bond-mediated electron transfer. The placement of the donor and acceptor has been chosen to favor electron transfer through the hydrogen bond interface of the β-turn. Use of ester linkages also allows control of the hydrogen-bonding pattern within the β-turn-forming depsipeptide. Infrared spectroscopy in the amide A (NH stretch) and amide I (carbonyl stretch) regions indicates that the β-turn conformation is about 85% populated in dichloromethane and essentially completely disrupted in dimethyl sulfoxide at 296 K. The electron-transfer rate constant, k(et), was evaluated using the singlet excited-state lifetimes of the porphyrin in the presence and absence of an electron acceptor. The lifetimes were obtained using time-correlated single-photon-counting fluorescence spectroscopy. Very fast electron transfer (k(et) = (1.1 ± 0.1) x 10⁹ s^-1) was observed in the presence of the β-turn conformation. When the β-turn structure was disrupted using the solvent DMSO, electron transfer was no longer competitive with the intrinsic fluorescence emission. Analysis of the data in terms of Marcus theory and the pathway model for electronic coupling yielded a value for the hydrogen bond coupling decay factor, ε(hb), of 0.8 ± 0.4, which is of the same order of magnitude as the theoretically predicted value of 0.36.

Full text of DOI:10.1021/ja981384k

10902
J. Am. Chem. Soc. 1998, 120, 10902-10911
Electron Transfer through the Hydrogen-Bonded Interface of a
â-Turn-Forming Depsipeptide
David A. Williamson† and Bruce E. Bowler*
Contribution from the Department of Chemistry and Biochemistry, 2190 East Iliff AVenue,
UniVersity of DenVer, DenVer, Colorado 80208-2436
ReceiVed April 22, 1998
Abstract: Hydrogen-bonding networks are believed to play an important role in electron-transfer pathways in
a protein medium. A porphyrin-quinone donor-acceptor compound with a depsipeptide bridge which forms
a â-turn has been synthesized to study hydrogen bond-mediated electron transfer. The placement of the donor
and acceptor has been chosen to favor electron transfer through the hydrogen bond interface of the â-turn.
Use of ester linkages also allows control of the hydrogen-bonding pattern within the â-turn-forming depsipeptide.
Infrared spectroscopy in the amide A (NH stretch) and amide I (carbonyl stretch) regions indicates that the
â-turn conformation is about 85% populated in dichloromethane and essentially completely disrupted in dimethyl
sulfoxide at 296 K. The electron-transfer rate constant, ket, was evaluated using the singlet excited-state lifetimes
of the porphyrin in the presence and absence of an electron acceptor. The lifetimes were obtained using
time-correlated single-photon-counting fluorescence spectroscopy. Very fast electron transfer (ket ) (1.1 (
0.1) × 109 s-1) was observed in the presence of the â-turn conformation. When the â-turn structure was
disrupted using the solvent DMSO, electron transfer was no longer competitive with the intrinsic fluorescence
emission. Analysis of the data in terms of Marcus theory and the pathway model for electronic coupling
yielded a value for the hydrogen bond coupling decay factor, ꢀhb, of 0.8 ( 0.4, which is of the same order of
magnitude as the theoretically predicted value of 0.36.
Introduction
electronic coupling medium (see eq 2).7 Dutton and co-workers
Long-range electron transfer has been studied extensively in
proteins,1 protein-protein complexes,2 models for photosyn-
thetic reaction centers,3 and peptides,4 including recent work
with peptides that mimic â-sheet structure.5 Outer-sphere
electron transfer is a function of nuclear position, electronic
coupling, driving force, medium, and electron donor and electron
acceptor spatial arrangement. The Marcus equation (see eq 1)6
relates these factors to the rate constant for electron transfer,
ket. Some studies indicate that electronic coupling between the
electron donor and electron acceptor decays exponentially,
scaled by a constant factor, â, indicative of a homogeneous
1/2
2
π
[Hab]2 e-[(∆G°+λ) /4λk T]
(1)
B
ket )
2
(
)
p λkBT
ket [Hab]o2 e-(âR)
(2)
report â ) 1.4 Å-1 for proteins,7 in line with the theoretical
prediction of Hopfield.8 In the pathway model, developed by
Beratan and Onuchic,9 electron transfer between the donor and
acceptor is expected to have structure-dependent inhomogene-
ities and thus will not depend smoothly on through-space
distance. In this model, the electronic coupling matrix element,
Hab, has contributions from different types of orbital interactionss
through bond, through space, and through hydrogen bond. Each
type of orbital interaction is represented by a decay constant
(ꢀc, ꢀts, and ꢀhb, respectively), which reflects the magnitude of
the decrease of electronic coupling across that type of orbital
interaction. Electronic coupling through hydrogen bonds is
important to understand in the case of protein electron transfer
due to the prevalence of hydrogen bond networks in proteins.
Several studies have demonstrated that hydrogen bonds can
mediate electron transfer efficiently,10 although the exact nature
† Present address: Department of Chemistry, University of Kansas,
Lawrence, KS 66045.
(1) (a) Wuttke, D. S.; Bjerrum, M. J.; Winkler, J. R.; Gray, H. B. Science
1992, 256, 1007-1009. (b) Mayo, S. L.; Ellis, W. R., Jr.; Crutchley, R. J.;
Gray, H. B. Science 1986, 233, 948-952. (c) Bowler, B. E.; Raphael, A.
L.; Gray, H. B. Prog. Inorg. Chem. 1990, 38, 259-322. (d) Langen, R.;
Chang, I.-J.; Germanas, J. P.; Richards, J. H.; Winkler, J. R.; Gray, H. B.
Science 1995, 268, 1733-1735. (e) Gray H. B.; Winkler, J. R. Annu. ReV.
Biochem. 1996, 65, 537-561.
(2) Nocek, J. M.; Zhou, J. S.; De Forest, S.; Priyadarshy, S.; Beratan,
D. N.; Onuchic, J. N.; Hoffman, B. M. Chem. ReV. 1996, 96, 2459-2489.
(3) (a) Moore, T.; Gust, D. AdV. Photochem. 1991, 16, 1-64. (b) Meyer,
T. J. Acc. Chem. Res. 1989, 22, 163-170.
(4) (a) Isied, S. In Electron Transfer in Inorganic, Organic, and
Biological Systems; Bolton, J., Mataga, N., McLendon, G., Eds.; ACS
Advances in Chemistry Series 228; American Chemical Society: Wash-
ington, DC, 1991; pp 229-245. (b) Isied, S. In Electron-Transfer Reac-
tions: Inorganic, Organometallic, and Biological Applications; Isied S.,
Ed.; ACS Advances Chemistry Series 253; American Chemical Society:
Washington, DC, 1997; pp 331-347.
(7) Moser, C. C.; Keske, J. M.; Warncke, K.; Farid, R. S.; Dutton, L. P.
Science 1992, 355, 796-802.
(8) Hopfield, J. J. Proc. Nat. Acad. Sci. U.S.A. 1974, 71, 3640-3644.
(9) Onuchic, J. N.; Beratan, D. N.; Winkler, J. R.; Gray, H. B. Annu.
ReV. Biophys. Biomol. Struct. 1992, 21, 349-377.
(10) (a) de Rege, P. J. F.; Williams, S. A.; Therien, M. J. Science 1995,
269, 1409-1413. (b) Turro, C.; Chang, C. K.; Leroi, G. E.; Cukier, R. I.;
Nocera, D. G. J. Am. Chem. Soc. 1992, 114, 4013-4015. (c) Kirby, J. P.;
Roberts, J. A.; Nocera, D. G. J. Am. Chem. Soc. 1997, 119, 9230-9236.
(d) Roberts, J. A.; Kirby, J. P.; Wall, S. T.; Nocera, D. G. Inorg. Chim.
Acta 1997, 263, 395-405.
(5) (a) Gretchikhine, A. B.; Ogawa, M. Y. J. Am. Chem. Soc. 1996, 118,
1543-1544. (b) Fernando, S. R. L.; Kozlov, G. V.; Ogawa, M. Y. Inorg.
Chem. 1998, 37, 1900-1905.
(6) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985, 811, 266-
273.
10.1021/ja981384k CCC: $15.00 © 1998 American Chemical Society
Published on Web 10/09/1998

Electron Transfer through a H-Bonded â-Turn Interface
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10903
Por/PL/Q are identical. Thus, comparison of the electron-
transfer properties of Por/PL/Q and PAQ should allow the effect
of a single hydrogen bond on electronic coupling to be evaluated
in a straightforward way.
The organic solvent of choice for a â-turn-forming peptide
must be a non-hydrogen-bonding solvent so that hydrogen bond
formation is favored. Electron transfer has been studied in
hydrogen bond-promoting and hydrogen bond-disrupting sol-
vents. Thus, electron transfer in the presence and absence of
the hydrogen bond interface shown in Figure 1 for Por/PL/Q
has been evaluated.
Experimental Section
tBoc-L-Proline-L-lactate-N-2,5-dimethoxybenzylamide (tBoc/PL/
DMB), porphyrin-L-proline-L-lactate-N-2,5-dimethoxybenzylamide (Por/
PL/DMB), the p-benzoquinone form of Por/PL/DMB, Por/PL/Q (Figure
1), and the non-hydrogen-bonding IR standard, N-methyl-D-5-oxoproline
methyl amide (NPMA), were synthesized according to Williamson and
Bowler.14 Tetrahydrofuran (THF) was dried on molecular sieves and
then distilled from sodium benzophenone ketyl and stored on molecular
sieves. N,N-Dimethylformamide (DMF) was stored on anhydrous
MgSO4 before distillation under reduced pressure. Triethylamine was
distilled from sodium benzophenone ketyl and stored on KOH pellets
to maintain dryness. All other chemicals were reagent grade or better.
Two other IR standards were prepared as follows.
Methyl-1-acetyl-2-L-pyrrolidinecarboxylate (MAPC). L-Proline
(3.00 g, 26.1 mmol) was dissolved in dry methanol (15 mL) and stirred
at -20 °C (ice/salt bath). Thionyl chloride (7 mL) was added to the
slurry over a 1-h period with an addition funnel. The reaction was
stirred for an additional 6 h. After all the solids were dissolved, the
reaction was concentrated to dryness. The resulting yellow oil was
dissolved in toluene (50 mL) and concentrated to dryness by rotary
evaporation to remove residual thionyl chloride and methanol. The
oil was dried overnight under vacuum. It was then dissolved in dry
DMF (10 mL) containing triethylamine (4 mL) and stirred on an ice
bath. Acetyl chloride (2.3 mL, 32 mmol) was dissolved in dry THF
(50 mL) and added dropwise with an addition funnel over a period of
1 h. The reaction mixture was stirred overnight under a CaCl2 drying
tube. The mixture was diluted with dichloromethane (250 mL) and
was extracted with NaHSO4 (3 × 250 mL), saturated sodium bicarbon-
ate (250 mL), and brine (250 mL). The organic layer was dried with
anhydrous sodium sulfate and concentrated by rotary evaporation,
yielding 3.77 g (84.5%) of a brown oil. 1H NMR (CDCl3): δ 1.97-
2.20 (4 H, m), 2.01 and 2.10 (3 H, two singlets), 3.5-3.75 (2 H, m)
3.75 (3 H, d), 4.44 (1 H, dd).
Methyl-1-benzoyl-2-L-pyrrolidinecarboxylate (MBPC). Procedure
was identical to the synthesis of methyl-1-acetyl-2-L-pyrrolidine car-
boxylate, except benzoyl chloride (3.2 mL, 27 mmol) was used instead
of acetyl chloride. Yield of a light yellow solid was 4.29 g (68.6%),
mp 83 °C. 1H NMR (CDCl3): δ 2.00-2.30 (4 H, m), 3.57-3.65 (2
H, m), 3.75 (3 H, s), 4.70 (1 H, t), 7.42-7.57 (5H, m).
Temperature-Dependent Amide NH NMR Spectra. Temperature-
dependent NMR studies were carried out in 99.9% dichloromethane-
d2 (Isotec, Miamisburg, OH). All chemical shifts are reported in ppm
and are referenced to the residual solvent peak (CH2Cl2 ) 5.32 ppm).
Carbon tetrachloride was spectroscopic grade purchased from Mallinck-
rodt and was distilled from Al2O3 before use. All chemical shifts in
carbon tetrachloride are reported in ppm and are referenced to
tetramethylsilane (TMS ) 0.0 ppm). All 1H NMR data were obtained
using a Chemagnetics 200-MHz FT-NMR operating at 199.167 MHz.
Temperature was controlled with a Chemagnetics REX-C1000 temper-
ature controller. The temperature was calibrated with methanol
according to Van Geet.15 The methanol shifts reported by Van Geet
and the methanol shifts observed on the Chemagnetics 200-MHz NMR
spectrometer were coincident as a function of temperature. Therefore,
temperature was read directly from the Chemagnetics temperature
Figure 1. Structures of porphyrin-quinone donor-acceptor com-
pounds. Por/PL/Q and Por/P/Qaa-PMA11 both form â-turn structures.
PAQ12,13 connects the donor and acceptor used in Por/PL/Q directly
via an amide bond. The through-bond electron-transfer pathways
between the edge of the tritolyl-p-carboxyphenylporphyrin donor and
the edge of the quinone acceptor are indicated on each structure. A
shared hydrogen between two heteroatoms is counted as two bonds.
of the hydrogen-bonded interface can alter this efficiency
dramatically.10b,d
In the work described here, a donor-acceptor compound with
a â-turn-forming depsipeptide bridge (Por/PL/Q in Figure 1)
has been developed to study hydrogen bond-mediated electron
transfer through a single amide NH to amide carbonyl (NH- - -
OdC) peptide hydrogen bond. Tamiaki and Muruyama have
previously reported electron-transfer studies with a donor-
acceptor compound bridged by a â-turn-forming peptide (Por/
P/Qaa-PMA in Figure 1).11 An advantage of the depsipeptide
bridge used in Por/PL/Q is that the central amide linkage has
been replaced with an ester, ensuring â-turn folding (C10 turn)
and eliminating the possibility of γ-turn folding (C7 turn). The
shortest “pathway” that results passes through the hydrogen bond
and has fewer bonds than the pathways possible for Por/P/Qaa-
PMA. Por/PL/Q can also be compared to electron transfer in
the compound PAQ (see Figure 1), reported on previously by
Bolton and co-workers.12,13 The donor porphyrin and acceptor
quinone are identical to those used in Por/PL/Q. Relative to
PAQ, the shortest through-bond pathway of Por/PL/Q has had
the NH- - -OdC hydrogen bond inserted into it. Also, three of
the four covalent bonds in the shortest pathways of PAQ and
(11) Tamiaki, H.; Maruyama, K. Chem. Lett. 1993, 1499-1502.
(12) (a) Archer, M. D.; Gadzepko, V. P. Y.; Bolton, J. R.; Schmidt, J.
A.; Weedon, A. C. J. Chem. Soc., Faraday Trans. 2 1986, 82, 2305-2313.
(b) Schmidt, J. A.; Liu, J.-Y.; Bolton, J. R.; Archer, M. D.; Gadzekpo, V.
P. Y. J. Chem. Soc., Faraday Trans. 1 1989, 85, 1027-1041.
(13) Lui, J.-Y.; Schmidt, J. A.; Bolton, J. R. J. Phys. Chem. 1991, 95,
6924-6927.
(14) Williamson, D. A.; Bowler, B. E. Tetrahedron 1996, 52, 12357-
12372.
(15) Van Geet, A. L. Anal. Chem. 1970, 42, 679-680.

10904 J. Am. Chem. Soc., Vol. 120, No. 42, 1998
Williamson and Bowler
controller without correction. Before sample preparation, all samples
were dried on a drying pistol using KOH pellets as a drying agent and
refluxing methanol for heat. All samples were prepared as ap-
proximately 1.0 mM solutions and were stored in the dark. The
chemical shift of the amide NH as a function of temperature was fit to
a linear equation to obtain the rate of change of the amide NH chemical
shift with respect to temperature, ∆δNH/∆T. The correlation coef-
ficient, r2, was greater than 0.99 in all cases.
NMR tube was then sealed with an O2 torch. During spectral
acquisition, the sample temperature was controlled at 5.0 °C. The
spectrum was obtained with τ1 ) 300 µs (150 increments, 96 scans/
increment), τ2 ) 100 µs (4096 points), τm ) 0.25 s with a spin lock
pulse of ∼2 kHz, a delay time of 1.2 s, and a spectral width of 12
kHz. Data were analyzed with Felix (ver. 95.0) software.
Fluorescence Lifetime Measurements. The time-correlated single-
photon-counting (TCSPC) experiments17 were performed at the Beck-
man Institute’s Laser Resource Center (California Institute of Tech-
nology, Pasadena, CA). TCSPC fluorescence decay curves were
produced by the buildup of a histogram of counts versus time. A mode-
locked, synchronously pumped, cavity-dumped dye laser was used to
photoexcite the sample at 600 nm (60 mW, 3.8 MHz, and 12 ps pulse
width). The fluorescence emission was passed through a polarizer set
at the magic angle of 54.7°. The light then entered a monochromator,
which was set at 650 nm. The histograms obtained were fit to single-
or multiple-exponential decays using curve-fitting software developed
at Caltech and based upon least-squares methods. The fit was believed
Amide A and Amide I Infrared Spectra. FTIR studies were
carried out in anhydrous dichloromethane (DCM) or dimethyl sulfoxide
(DMSO), which were purchased from Aldrich at 99.98% purity in sure
seal bottles. DCM was treated with anhydrous K2CO3 to remove any
residual water or acid. Infrared spectral data were obtained with a
Nicolet Prote´ge´ 460 FT-IR spectrometer at 2-cm-1 resolution and are
reported in cm-1. The spectrometer was purged with dry N2 for 15
min before spectra were recorded. Samples were prepared between
CaF2 plates with a 1.0-mm path length spacer. All spectra were
processed using OMNIC (ver. 2.1) software by subtraction of the
solvent background (using the optically clear region near 2000 cm-1),
subtraction of residual water vapor, and baseline correction. Curve-
fitting was carried out using Grams/386 (ver. 3.02) software. Curve-
fitting routines were stopped when a minimum ø2 value was obtained
for the anticipated number of stretches, and the curve fit resembled
the actual spectrum by visual inspection. Residual line correlations
(R2) were better than 0.995. If the curve fit for the anticipated number
of peaks did not resemble the peak profile, additional peaks were added
until the fit resembled the peak profile. Normally, a mixed Gaussian
and Lorentzian curve shape was used to represent the IR absorption
bands in the curve-fitting procedure. If the fit for a particular curve
was either 100% Lorentzian or Gaussian, the peak was fixed as such,
and curve-fitting for the other curves was allowed to continue using
mixed functions. All compounds were dried on a drying pistol as
described above for NMR samples. DCM and DMSO samples were
prepared at a concentration of ∼1 mM in a drybox flushed with dry
N2.
Temperature-Dependent Amide A IR Measurements. For tem-
perature-dependent IR studies in DCM, A Nicolet 5DXC FTIR
spectrometer was used at a resolution of 4 cm-1. A cold head with
KBr windows (RMC Cryosystems, model 22, Tucson, AZ), equipped
with a He compressor and a gold-plated liquid sample holder, was used.
Samples were placed in a 1-cm path length IR quartz cell (International
Crystal Labs, Inc., Garfield, NJ), which was sealed with a Teflon lid
and Parafilm to prevent solvent evaporation. A Palm Beach Cyro-
physics, Inc. series 4000 cryogenic thermometer/controller was used
to control and monitor the temperature within the cold head. The
temperature was allowed to equilibrate for 20 min before each spectrum
was acquired. Spectra were processed as above for room-temperature
data. The non-hydrogen-bonding standard, NPMA, was used to
evaluate the extinction coefficient for the non-hydrogen-bonded second-
ary amide NH stretching vibration at 3445 cm-1 as a function of
temperature. This extinction coefficient was then used to evaluate the
equilibrium constant for â-turn formation at each temperature, Kâ-turn(T),
for Por/PL/DMB, using eq 3, where [Por/PL/DMB]total is the total
2
to be appropriate for the decay when øν ≈ 1.0, where øν2 is chi-squared
normalized to the number of degrees of freedom. The samples for the
TCSPC experiments were dried overnight in a drying pistol as for the
temperature-dependent NMR experiments. The samples were prepared
at a concentration of 2-10 µM in a 1-cm quartz fluorescence cell,
which was attached to a 10-mL Pyrex volumetric flask and a Teflon
stopcock. After the dried samples were dissolved into anhydrous
dichloromethane or dimethyl sulfoxide, the samples were freeze-
pump-thawed in the attached Pyrex flask a minimum of 10 cycles to
remove dioxygen. The degree of oxidation of the quinone moiety of
Por/PL/Q was estimated from the increase in the extinction coefficient
at 248 nm (ꢀ248 ) 20 300 M-1 cm-1 for methyl-p-benzoquinone in CH2-
Cl2)18 relative to Por/PL/DMB.
Calculation of Rate Constants, Driving Force, and Reorganiza-
tion Energies. Electron-transfer rate constants, ket, were calculated
using eq 4,19 where τo is the fluorescence lifetime for Por/PL/DMB
1
1
ket )
-
(4)
τ1 τ0
and τ1 is the lifetime for Por/PL/Q. It is assumed that electron transfer
is the only additional excited-state decay pathway introduced in the
conversion from Por/PL/DMB to Por/PL/Q. Energy transfer is not
expected to compete since there is no spectral overlap between the
porphyrin emission near 650 nm and quinone absorbance bands.
The driving force (-∆G°) for a free base Por/PL/Q molecule in
dichloromethane was calculated using eq 5. This equation has been
e2
∆G° ) e(EP° - EQ°) -
- ∆GES
(5)
4πꢀoꢀsaPQ
found to successfully correlate electron-transfer rate constants for
porphyrin-quinone donor-acceptor compounds in a variety of sol-
vents.12 The values for EP° and EQ°, the electrochemical potentials
for one-electron oxidation of free base porphyrin and the one-electron
reduction of quinone, respectively, are taken from previously published
electrochemical data for the porphyrin and quinone moieties used in
Por/PL/Q.12 Since EP° and EQ° were not reported for DMSO in the
previous work,12 the CH3CN values were used, since ꢀs(CH3CN) is
similar to ꢀs(DMSO). EP°-EQ° values in a wide range of solvents
deviate from the value of 1.41 eV in CH3CN by (0.1 eV. We have
used this deviation to define the error in the evaluation of -∆G° for
DMSO and propagated this error into subsequent calculations. The
second term in eq 5 is a Coulombic correction to account for
stabilization of the charge-transfer state produced after electron transfer.
In the second term, ꢀo is the permittivity of free space (8.854 × 10-12
C2 N-1 m-2). The static dielectric constants used were ꢀs(DCM) )
Kâ-turn(T) ) {[Por/PL/DMB]total
(AbsNHB-Por/PL/DMB(T)/ꢀNPMA(T))}/
(AbsNHB-Por/PL/DMB(T)/ꢀNPMA(T)) (3)
-
concentration of Por/PL/DMB in the sample, AbsNHB-Por/PL/DMB(T) is
the absorbance at the non-hydrogen-bonding NH stretching frequency
(∼3445 cm-1) of Por/PL/DMB at temperature T, and ꢀNPMA(T) is the
extinction coefficient of the stretching vibration of the secondary amide
NH of NPMA, the non-hydrogen-bonding standard, at temperature T.
ROESY 1H NMR Experiments. The 2D ROESY experiment16 on
Por/PL/DMB was performed on a Varian 500-MHz NMR spectrometer
in the Department of Chemistry and Biochemistry at the University of
Colorado at Boulder. Por/PL/DMB was dissolved in CD2Cl2, freeze-
pump-thawed (five cycles) to remove O2, and refilled with N2. The
(17) Small, E. W. In Topics in Fluorescence Spectroscopy, Vol. 1:
Techniques; Lakowicz, J. R., Ed.; Plenum Press: New York, 1991; pp 97-
182.
(18) McIntosh, A. R.; Siemiarczuk, A.; Bolton, J. R.; Stillman, M. J.;
Ho, T.-F.; Weedon, A. C. J. Am. Chem. Soc. 1983, 105, 7125-7223.
(19) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Plenum
Press: New York, 1983; pp 261-262.
(16) Bothner-By, A. A.; Stephen, R. L.; Lee, J.; Warren, C. D.; Jeanloz,
R. W. J. Am. Chem. Soc. 1984, 160, 811-813.

Electron Transfer through a H-Bonded â-Turn Interface
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10905
8.93 and ꢀs(DMSO) ) 46.68 (refs 20a,b, respectively). The distance
between the donor and acceptor, aPQ, was estimated using the molecular
modeling program HyperChem (ver. 3.0). To obtain aPQ, a confor-
mational search was carried out about the dihedral angles on either
side of the methylene group connecting the quinone to the â-turn. A
type I â-turn was assumed on the basis of IR data (see Discussion
section; for Φ,Ψ conformational angles, see ref 21). The tolyl and
p-carboxyphenyl groups were assumed to be perpendicular to the
porphyrin ring, and the carboxy group that connects the porphyrin
system to the proline was assumed to be perpendicular to the benzene
ring it is attached to. The center-to-center distance from the porphyrin
to the quinone for each conformation within 5 kcal/mol of the lowest
energy conformation was used to calculate a weighted average value
of aPQ. The weighting for each distance was based on the Boltzmann
population of each conformation expected from its energy relative to
the lowest energy conformation. Energies were calculated with the
mm+ force field22 provided with HyperChem. We have assumed that
the porphyrin and quinone moieties do not significantly perturb the
depsipeptide â-turn backbone angles. For the â-turn formed by Por/
PL/Q, a value of aPQ ) 9.8 Å was obtained. The range of distances
within a Boltzmann weighting g0.1 is 8.7-10.7 Å. The third term in
eq 5, ∆GES, is the free energy of the excited state and was calculated
using eq 6,23 where E0 is the energy of the fluorescence maximum (657
nm) of Por/PL/DMB and ø′ is the reorganization energy due to solvent
and low-frequency vibrational modes. ø′ is given by eq 7, where ∆νj0,1/2
Figure 2. Temperature dependence of the 1H NMR shifts of the amide
NH for â-turn-forming compounds in chloromethane solvents: 1.0 mM
tBoc/PL/DMB in CDCl3 (∇, solid line); 1.36 mM Por/PL/DMB in CD2-
Cl2 (b, dashed line); 1.35 mM tBoc/PL/DMB in CCl4 (4, solid line);
1.36 mM Por/PL/DMB in CCl4 ([, dashed line).
Table 1. Temperature Dependence of Amide NH NMR
Resonances for â-Turn-Forming Compounds
∆GES ) E0 + ø′
(6)
compound
solvent
∆δNH/∆T (ppb/K)
2
(∆νj0,1/2
)
tBoc/PL/DMB
Por/PL/DMB
tBoc/PL/DMB
Ac/PL/NHMea
Por/PL/DMB
CCl4
CCl4
CDCl3
CD2Cl2
CD2Cl2
-4.33
-4.78
-5.30
-3.51
-4.79
ø′ )
(7)
(16kBT ln 2)
is the full width of the emission band at half-maximum. Both E0 and
ø′ are enthalpic quantities. The use of this equation to represent the
free energy of the excited state is justified by the negligible pressure-
volume work23b expected in solution and the small contribution expected
from the electronic entropic change.23a
a Data from ref 24a.
systems. Thus, we have carried out structural studies on Por/
PL/DMB and tBoc/PL/DMB in chloromethanes and dimethyl
sulfoxide using NMR and infrared spectroscopic methods. The
data are compared with data for standard compounds as well
as N-acetyl-L-proline-L-lactate-N-methylamide (Ac/PL/NHMe).24
These comparisons have allowed determination of the impor-
tance of the capping groups on the conformation of the proline-
lactate depsipeptide. The presence of a single amide NH in
these compounds and carbonyl groups with very distinct
stretching frequencies also has allowed detailed conformational
analysis using these spectroscopic methods.
Structural and Thermodynamic Properties of Por/PL/
DMB in Chloromethanes. Temperature-dependent NMR
studies were carried out to assess the presence of the â-turn
(C10) conformation. The amide NH chemical shift was moni-
tored from -25 to 60 °C in CCl4, -30 to 60 °C in CDCl3, and
-45 to 25 °C in CD2Cl2. Por/PL/DMB was examined in CCl4
and CD2Cl2, and tBoc/PL/DMB was examined in CCl4 and
CDCl3 (see Figure 2). All four temperature-dependent experi-
ments were compared to temperature-dependent NMR experi-
ments on Ac/PL/NHMe24a in CD2Cl2 (see Table 1). The amide
NH chemical shift could not be observed for the Por/PL/DMB
in CD2Cl2 from -5 to -20 °C since the amide NH resonance
was hidden by porphyrin resonances. The temperature ranges
selected were limited by the melting and boiling points of the
solvent.
The reorganization energy (λ ) λs + λi) was calculated from the
solvent reorganization energy (λs) using eq 86 and an internal reorga-
nizational energy (λi) of 0.2 eV.12,13 Values for ꢀop were obtained from
(∆e)2
1
1
1
1
1
λs )
+
-
-
(8)
(
)(
)
4πꢀo 2aP 2aQ aPQ ꢀop ꢀs
the same sources cited above for ꢀs. The radii of the donor, aP, and
acceptor, aQ, were estimated using HyperChem (ver. 3.0). The radius
of the quinone was evaluated as the average of the three cross-ring
distances (oxygen-oxygen, hydrogen-hydrogen, and hydrogen-me-
thylene carbon) and gave a value of aQ ) 3.2 ( 0.2 Å. The radius of
the porphyrin was taken as an average of the radius out to the edge of
the pyrrole ring and the radius out to the tolyl methyl groups. A value
of aP ) 8 ( 2.5 Å was obtained. For all measurements, the atomic
radii of the outer atoms were taken into account (atomic radii used:
H, 0.37 Å; C, 0.77 Å; O, 0.73 Å). Given ∆G° and λ, the Franck-
Condon factor can be determined. Thus, using the experimentally
determined rate constants for the â-turn, the electronic coupling matrix,
Hab, was calculated from eq 1. Similar methods were used to evaluate
∆G°, λ, and Hab for the previously reported 5-(4-carboxyphenyl)-10,
15,20-tri(p-tolyl)porphyrin-N-quinonyl-2-methylamide (PAQ) (aPQ
12.6 Å, assuming trans stereochemistry about the amide bond; range
)
with Boltzmann weighting g0.1 is 12.5-12.9 Å).
Results
Structural characterization is of prime importance for inter-
pretation of electron-transfer data in nonrigid donor-acceptor
(20) (a) Riddick, J. A.; Bunger, W. B. Techniques of Chemistry II,
Organic SolVents, 4th ed.; Wiley-Interscience: New York, 1970; pp 348-
349; (b) pp 466-467.
A small value of the rate of change of the amide NH chemical
shift with respect to temperature, ∆δNH/∆T, is typically
associated with an amide NH which does not change hydrogen-
bonding state significantly as a function of temperature. Gener-
(21) Wuthrich, K. NMR of Proteins and Nucleic Acids; Wiley-Inter-
science: New York, 1986.
(22) Allinger, N. L. J. Am. Chem. Soc. 1977, 99, 8127-8134.
(23) (a) Opperman, K. A.; Mecklenburg, S. L.; Meyer, T. J. Inorg. Chem.
1994, 33, 5295-5301. (b) Hupp, J. T.; Neyhart, G. A.; Meyer, T. J.; Kober,
E. M. J. Phys. Chem. 1992, 96, 10820-10830.
(24) (a) Liang, G.-B.; Rito, C. R.; Gellman, S. H. J. Am. Chem. Soc.
1992, 114, 4440-4442. (b) Boussard, G.; Marraud, M.; Neel, J. Biopolymers
1977, 16, 1033-1052.

10906 J. Am. Chem. Soc., Vol. 120, No. 42, 1998
Williamson and Bowler
ally, the lower the magnitude of ∆δNH/∆T, the more complete
hydrogen bonding is considered to be relative to that of other
similar compounds in a given solvent.25 As is evident in Figure
2 and Table 1, the data for tBoc/PL/DMB and Por/PL/DMB in
chloromethane solvents show that the hydrogen-bonding state
is not changing with a change in temperature, and that the
addition of the porphyrin moiety does not strongly affect ∆δNH/
∆T. Similarly, the folding of the tBoc/PL/DMB â-turn in
chloromethanes is not significantly affected by the addition of
the dimethoxybenzene moiety when compared to the ∆δNH/
∆T results from the depsipeptide, Ac/PL/NHMe.24a Ac/PL/
NHMe is known to be essentially completely in a â-turn
conformation in DCM.24 The somewhat larger magnitude of
∆δNH/∆T for Por/PL/DMB in DCM may indicate a smaller
equilibrium constant for â-turn formation.
Although the temperature dependence of the amide NH NMR
resonance is suggestive, it is not definitive proof of hydrogen
bonding, and thus the conformation was also assessed by
infrared spectroscopy. In the amide A region (amide NH stretch,
3300-3500 cm-1), the stretching frequency for the amide NH
group depends on its environment. If the amide NH is involved
in a hydrogen bond to a peptide carbonyl, then the stretch
appears in the range 3300-3370 cm-1 in chloromethanes.24,26
The non-hydrogen-bonding amide typically occurs around 3450
cm-1 24,26
In previously reported work, Por/PL/DMB was
.
examined at room temperature to determine the relative amount
of â-turn formed versus the non-hydrogen-bonded state.14 The
equilibrium constant for â-turn formation, Kâ-turn, was found
to be 5.7 (86% turn formation). Given that Ac/PL/NHMe is
considered to be completely in a â-turn conformation under
these conditions, the observation of 86% â-turn formation for
Por/PL/DMB is consistent with the somewhat larger magnitude
of ∆δNH/∆T observed above.
Figure 3. (A) Infrared spectra in the amide A (NH stretch) region for
Por/PL/DMB at 210 (solid line) and 240 K (dashed line) at a
concentration of 1.04 mM in CH2Cl2. The low broad peak near 3445
cm-1 is due to amide NH which is not hydrogen bonded. The sharp
peak near 3325 cm-1 is due to amide NH involved in a hydrogen bond.
(B) Van’t Hoff plot using Kâ-turn values (eq 3, Experimental Section)
extracted from amide A data.
In the work reported here, the infrared spectrum of Por/PL/
DMB was examined as a function of temperature to determine
the temperature dependence of folding. The amide stretch of
the non-hydrogen-bonding standard, NPMA, showed a slight
temperature dependence of the molar absorptivity, ꢀ. The molar
absorptivity changes linearly, with a temperature coefficient of
-0.28 M-1 cm-1 K-1 (from linear regression analysis, ꢀ(T) )
-0.28 M-1 cm-1 K-1(T) + 205 M-1 cm-1; r2 ) 0.982 at 3445
cm-1). Similar temperature dependences for the infrared
absorptivity of non-hydrogen-bonded amides have been ob-
served previously for O-acetyl-L-lactate-N-methylamide (Ac/
L/NHMe)27 and N-methylacetamide.28 A concentration of 1.04
mM was used for NPMA to avoid intermolecular hydrogen
bonding, which was evidenced by the absence of a hydrogen-
bonded amide peak in the range of 3300-3370 cm-1. The only
peak present was the expected non-hydrogen-bonded amide NH
near 3445 cm-1. Lack of intermolecular hydrogen bonding at
this concentration is typical for compounds of this type.26c,27
Thus, Por/PL/DMB was examined over the same temperature
range and at a similar concentration as for NPMA. At all
temperatures, both hydrogen-bonded and non-hydrogen-bonded
amide A IR absorbances were present (Figure 3A). The
absorbance of the non-hydrogen-bonded amide was much lower
than that for the hydrogen-bonded amide. Calculation of Kâ-turn
(eq 3, Experimental Section) throughout the temperature range
gave values from 6.3 to 5.7. Thus, the folding for the
depsipeptide â-turn in Por/PL/DMB is almost temperature
independent.
A van’t Hoff plot of the temperature-dependent infrared
folding data can give insight into the driving force for the folding
process (see Figure 3B). The van’t Hoff analysis yields a linear
plot within error, where the slope represents -∆H° and the
y-intercept is ∆S° for the folding process. A line where the
slope approaches zero indicates that the folding process is driven
not by enthalpy, but by entropy. For the â-turn-folding
equilibrium of Por/PL/DMB in CH2Cl2, ∆H° ) -0.04 ( 0.05
kcal/mol and ∆S° ) 3.4 ( 0.2 eu. Thus, â-turn formation is
entropically driven. Similar results have been observed for other
â-turn-forming peptides.24a
A ROESY NMR experiment16 was also carried out on Por/
PL/DMB in CD2Cl2 in an attempt to obtain confirmatory
structural data on the â-turn conformation. Since this is a
depsipeptide, the amide NH between the n + 1 and n + 2
residues of the turn is replaced with an ester linkage. This NH
normally has short through-space distances to the CRH of the n
+ 1 (proline) and the amide NH of the n + 3 residue, which
are characteristic of a â-turn.21 Sometimes the CRH (n + 1) to
(25) Dyson, H. J.; Rance, M.; Houghten, R. A.; Lerner, R. A.; Wright,
P. E. J. Mol. Biol. 1988, 201, 161-200.
(26) (a) Haque, T. S.; Little, J. C.; Gellman, S. H. J. Am. Chem. Soc.
1994, 116, 4105-4106. (b) Haque, T. S.; Little, J. C.; Gellman, S. H. J.
Am. Chem. Soc. 1996, 118, 6975-6985. (c) Liang, G.-B.; Rito, C. R.;
Gellman, S. H. Biopolymers 1992, 32, 293-301.
(27) Gallo, E. A.; Gellman, S. H. J. Am. Chem. Soc. 1993, 115, 9774-
9788.
(28) Gellman, S. H.; Dado, G. P.; Liang, G.-B.; Adams, B. R. J. Am.
Chem. Soc. 1991, 113, 1164-1173.

Electron Transfer through a H-Bonded â-Turn Interface
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10907
amide NH (n + 3) cross-peak can be observed,29 although the
distance (3.3-3.6 Å) is at the edge of what can normally be
observed in a ROESY NMR experiment, and it is often not
detected.21 This latter interaction is the only one characteristic
of a â-turn which might be observable for Por/PL/DMB. It
was not observed. The only cross-peaks observed for the amide
NH of Por/PL/DMB were with the methylene hydrogens next
to the amide NH and the closest aromatic ring hydrogen (very
weak) on the dimethoxybenzene ring. Notably, no cross-peaks
between the dimethoxybenzene ring and the prophyrin ring were
observed. This is consistent with previously reported visible
absorption data,14 which indicate no electronic interaction
between the porphyrin ring and the dimethoxybenzene ring in
Por/PL/DMB or between the porphyrin ring and the quinone
ring for Por/PL/Q. Thus, the ROESY NMR data confirm that
there are no significantly populated conformations where the
porphyrin and the dimethoxybenzene ring come in contact with
each other.
Effects of Dimethyl Sulfoxide on the Structural Properties
of Por/PL/DMB. The ability of Por/PL/DMB and tBoc/PL/
DMB to form a â-turn was examined in the hydrogen bond-
disrupting solvent DMSO as well. The NMR spectra of tBoc/
PL/DMB and Por/PL/DMB were examined in DMSO-d6 versus
CDCl3 and CD2Cl2 to determine the conformation in hydrogen-
bonding versus non-hydrogen-bonding solvents. NMR spectra
of both of these compounds in CDCl3 and CD2Cl2 show no
evidence of conformational heterogeneity, consistent with the
predominance of the â-turn conformation. Clear evidence for
disruption of hydrogen bonding can be seen for Por/PL/DMB
and tBoc/PL/DMB in DMSO-d6. Several proton resonances in
both compounds are doubled, indicating two distinct conforma-
tions interconverting slowly on the NMR time scale. Notably,
the tBoc methyl groups in tBoc/PL/DMB show one sharp singlet
at 1.39 ppm in CDCl3. In DMSO-d6, the tBoc methyl groups
are split into two peaks of approximately equal intensity,
reflecting the presence of cis and trans isomers about the tertiary
proline amide. The trans proline isomer is no longer locked in
by â-turn formation.
To further characterize the hydrogen bonding of Por/PL/DMB
in DMSO versus chloromethanes, infrared amide I spectroscopy
was used. The amide I region of the infrared spectrum is due
predominately to the CdO stretching vibration.30 Since Por/
PL/DMB has three different types of carbonyls (secondary
amide, ester, and primary amide), which resonate at different
frequencies, interpretation of the spectra is relatively straight-
forward. Figure 4A shows the IR spectrum of Por/PL/DMB in
CH2Cl2. Four distinct absorbances can be seen at 1743.5
(proline ester), 1674.4 (secondary amide), 1619.5 (tertiary
benzoyl amide), and 1607.5 cm-1 (aromatic ring vibrations).
The assignments are based on studies of model compounds (see
Table 2). Aromatic ring vibrations near 1605 cm-1 are observed
for 2,5-dimethoxytoluene and for 5,10,15-tri(p-tolyl)-20-(p-
methylbenzoate)porphyrin in both DCM and DMSO (data not
shown). Interestingly, the tertiary benzoyl amide is shifted ∼12
cm-1 to lower energy compared to that of the standard
compound, methyl-1-benzoyl-2-L-pyrrolidinecarboxylate (MBPC)
(see Table 2). This shift is of the magnitude expected for the
participation of this carbonyl in the hydrogen bond of a â-turn.24b
Figure 4B shows the IR spectrum of Por/PL/DMB in DMSO.
Clear differences are evident. The aromatic ring vibration peak
(1604.9 cm-1) is farther from the tertiary benzoyl amide peak
Figure 4. Amide I (CdO stretch) data for Por/PL/DMB. (A) Data
acquired in DCM at a concentration of 1.89 mM and a temperature of
298 K. Curve-fitting analysis (see Experimental Section) yielded the
following peak components: ester, 1743.5 cm-1; 2° amide, 1674.4
cm-1; 3° amide 1619.5 cm-1; aromatic ring, 1607.5 cm-1. (B) Data
acquired in DMSO at a concentration of 0.67 mM and T ) 298 K.
Curve-fitting analysis yielded the following components: ester, 1747.5,
1731.3 cm-1; 2° amide, 1688.0, 1667.0 cm-1; 3° amide, 1629.9 cm-1
aromatic ring, 1604.9 cm-1
;
.
Table 2. Amide I Infrared Vibrational Frequencies (in cm-1
)
DCM
DMSO
3° amide ester 2° amide 3° amide
ester
2
2° amide
compound
1
2
3
1
3
Por/PL/DMBa 1619.5 1743.5 1674.4 1629.9 1747.5
1667
1731.3 (sh) 1688 (sh)
1645.0 1742.0
1630.7 1742.9
MAPCb
MBPCc
1646.0 1743.9
1631.5 1744.0
a R ) 5-(p-phenyl)-10,15,20-tri(p-tolyl)prophyrin moiety. b R )
methyl, ester 2 is a methyl ester. c R ) phenyl, ester 2 is a methyl
ester.
(1629.9 cm-1), such that it now appears as a low shoulder on
this peak. The secondary amide peak is more complex, and
the prolyl ester peak is broadened and less intense. Of note,
the tertiary benzoyl amide is no longer shifted to lower energy,
(29) Imperiali, B.; Fisher, S. L.; Moats, R. A.; Prins, T. J. J. Am. Chem.
Soc. 1992, 114, 3182-3188.
(30) Krimm, S.; Bandekar, J. AdV. Protein Chem. 1986, 38, 181-364.

10908 J. Am. Chem. Soc., Vol. 120, No. 42, 1998
Williamson and Bowler
Table 3. Fluorescence Lifetimes (in ns) for Por/PL/DMB and
Por/PL/Q in DCM and DMSOa
compound
solvent
DCM
Por/PL/DMB
8.95
Por/PL/Q
8.76 (24%)b
0.85 (76%)b
11.47 (94%)b
2.27 (6%)b
DMSO
11.56
a Errors in the lifetimes are (0.1 ns. b Percent amplitude for
fluorescence emission decays that are not monoexponential.
Table 4. Electron-Transfer Rate Constants and Calculated Values
for Driving Force, Solvent Reorganization Energy, and Electronic
Coupling Matrix Element
compound,
solvent
Por/PL/Q, DCM Por/PL/Q, DMSO
PAQ,a DCM
b
k
et (s-1
)
(1.1 ( 0.1) × 109 (3.5 ( 0.2) × 108 (8.0 ( 0.4) × 108
-∆G° (eV)c 0.67 ( 0.02
0.5 ( 0.1
0.73 ( 0.07
3 ( 2
0.63 ( 0.01
0.77 ( 0.01
3.1 ( 0.2
λs (eV)d
ab (cm-1
0.64 ( 0.06
2.4 ( 0.5
e
H
)
b
a ket for PAQ is from ref 13. ket is calculated from eq 4 in the
Experimental Section using the lifetime for Por/PL/DMB for τo and
the fast lifetime for Por/PL/Q for τ1. c Calculated using eq 5 in the
Experimental Section. The error reflects the range of -∆G° values
due to the uncertainty in aPQ.
d Calculated using eq 8 in the Experimental
Section. The error reflects the range of λs values due to the uncertainty
in aPQ
e Hab was calculated using eq 1 and the values of -∆G°, λs,
and ket from this table. The error in Hab reflects the range of Hab values
due to the errors in -∆G° and λs resulting from the uncertainty in aPQ
.
.
Figure 5. TCSPC data at 25 °C for Por/PL/DMB (A) and Por/PL/Q
(B) in DCM. [Por/PL/DMB] ) 8.9 µM. [Por/PL/Q] ) 1.8 µM. The
y-axis fluorescence intensity represents the number of counts per
channel. Each channel has been converted to time in nanoseconds on
the x-axis. Solid lines through the data points represent the fit to the
data. The data in part A was fit to a single-exponential decay. The
data in part B was fit to two exponential decays. Increasing the number
of exponential decays in the fit for either data set did not significantly
improve the fit. The residuals of each fit to the data are shown below
the data.
The values of aP and aQ cause errors in -∆G° and λs which will affect
the evaluation of Hab as well. However, for comparison between PAQ
and Por/PL/Q, the values of aP and aQ must be the same for each
comparison. The most extreme effects occur when aP and aQ are both
at the high or the low end of their error ranges. For aP ) 10.7 and aQ
) 3.4, the values of Hab for the DCM data are as follow: Por/PL/Q,
1.9 ( 0.2 cm-1; PAQ, 2.1 ( 0.1 cm-1. For aP ) 5.6 and aQ ) 3.0, the
values of Hab for the DCM data are as follow: Por/PL/Q, 4 ( 1 cm-1
;
PAQ, 6.2 ( 0.5 cm-1
.
DMB is that electron transfer is not competitive with intrinsic
fluorescence decay for Por/PL/Q in non-â-turn conformations.
as would be expected if it were hydrogen-bonded in a â-turn
conformation. Its frequency is shifted only 1 cm-1 to lower
energy relative to the non-hydrogen-bonding standard MBPC
in DMSO (Table 2), indicating almost complete disruption of
the â-turn hydrogen bond.
To test this hypothesis, fluorescence lifetime measurements
were also made in dimethyl sulfoxide. IR data presented above
indicate that dimethyl sulfoxide nearly completely disrupts the
â-turn hydrogen bond. Thus, excited-state quenching due to
electron transfer will have to occur through the depsipeptide
backbone (10 covalent bonds). The Por/PL/Q compound in
DMSO gave a biexponential decay (Table 3). The major
lifetime (94% of the amplitude) is essentially identical to the
lifetime of Por/PL/DMB in DMSO (11.56 ns), indicating that
electron transfer does not compete with fluorescence decay for
the majority of Por/PL/Q conformations in DMSO. A minor
fast component in the fluorescence decay is also resolved,
suggesting that a small amount of â-turn conformation still
persists in DMSO, consistent with the amide I IR data.
Electron-transfer rate constants, ket, for Por/PL/Q in DCM
and DSMO are collected in Table 4. Data for PAQ12,13 are
included for comparison. The large magnitude of the rate
constant for Por/PL/Q suggests that electron transfer through
the hydrogen bond interface of this compound is very efficient.
Direct comparison of ket, however, cannot be done between
different solvents and different compounds to assess the
efficiency of electronic coupling through the hydrogen bond
interface. Variations in average donor-acceptor separation in
different compounds and the solvent dielectric in different
solvents will affect both the driving force and the reorganization
energy for the electron-transfer reaction. To determine elec-
tronic coupling more accurately, changes in the Franck-Condon
Fluorescence Lifetime Measurements. The photoinduced
electron transfer between the excited state of the porphyrin and
the quinone was monitored in dichloromethane by TCSPC
fluorescence spectroscopy. In previous work, a reduction in
the steady-state fluorescence of Por/PL/Q relative to Por/PL/
DMB indicated electron transfer was competitive with the
intrinsic emission lifetime.14 Figure 5 compares the fluorescence
decays for the excited states of Por/PL/DMB and Por/PL/Q in
DCM. Clearly, the emission lifetime of Por/PL/Q is signifi-
cantly shorter than that of Por/PL/DMB, consistent with a
competitive electron-transfer process. The lifetimes extracted
from curve-fitting analysis are reported in Table 3. From visual
inspection of Figure 5B, it is clear that the fluorescence decay
of Por/PL/Q is not monoexponential. The lifetime for the slower
process (Table 3) is within error of that for Por/PL/DMB.
Spectroscopic evaluation of the degree of oxidation of Por/PL/Q
indicated that it was 91% oxidized (9% in the hydroquinone
form, Por/PL/QH2) for the lifetime data shown. This small
quantity of Por/PL/QH2, which is not competent for electron
transfer, explains part of the amplitude with a lifetime within
error of that for Por/PL/DMB. Since ∼15% of Por/PL/Q is
not in the â-turn conformation, a possible explanation of the
additional amplitude with a lifetime similar to that of Por/PL/

Electron Transfer through a H-Bonded â-Turn Interface
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10909
factor for different electron-transfer reactions must be assessed.
The driving force, -∆G°, the solvent reorganization, λs, and
the electronic coupling matrix element, Hab, calculated as
described in the Experimental Section, are presented in Table
4. Qualitatively, this evaluation of Hab is consistent with
expectation. PAQ, with the shorter through-bond pathway (four
bonds), has the larger value of Hab. Por/PL/Q in DCM has the
smaller value of Hab, consistent with the longer through-bond
pathway (six bonds).
indicates that the steric bulk of the capping groups in proline-
lactate depsipeptides has only a small impact on the â-turn
propensity.
Conformational searches using HyperChem (see Experimental
Section) indicate that the porphyrin and quinone moieties of
Por/PL/Q cannot directly interact with each other in the â-turn
conformation. Previously reported UV-vis spectroscopic stud-
ies of Por/PL/Q and Por/PL/DMB indicate no electronic
interaction between the donor and acceptor chromophores.14 The
ROESY NMR data reported here confirm this previous observa-
tion. Porphyrin-quinone donor-acceptor compounds with
flexible alkane spacers31 have shown considerable electronic
interaction between the donor and acceptor chromophores.
Clearly, the relatively rigid conformation of the â-turn prevents
such interactions. Thus, the fast electron-transfer event observed
here for Por/PL/Q is not due to direct chromophore interaction,
but results from the shorter electronic coupling pathway
provided by the hydrogen bond of the â-turn.
Discussion
Structural Analysis of the Donor-Acceptor Compound
Bridged by a Depsipeptide. Por/PL/Q is based on a well-
studied depsipeptide, Ac/PL/NHMe, which is known to adopt
essentially completely a â-turn conformation in chloromethanes.24
Since the capping groups on the depsipeptide (PL) are signifi-
cantly altered in Por/PL/DMB(Q), we carried out spectroscopic
studies on Por/PL/DMB and its simpler analogue, tBoc/PL/
DMB, to determine their structural propensities. We have used
Por/PL/DMB rather than Por/PL/Q for structural studies because
it is available in larger quantities and is more stable than Por/
PL/Q. Modeling studies indicate that steric constraints prevent
the oxygen atoms of either the quinone or dimethoxybenzene
moieties from hydrogen bonding to the amide NH. Thus, Por/
PL/DMB and Por/PL/Q should have similar structural properties.
Our data show that DMSO profoundly affects the conforma-
tion of Por/PL/DMB. Previous work indicates that DMSO
disrupts hydrogen bonds in proteins and peptides.26b,32 Both
NMR and IR data show evidence of conformational heterogene-
ity. The IR data also indicate that the â-turn hydrogen bond
has been nearly completely disrupted. The tertiary benzoyl-
proline amide carbonyl stretch is now centered at ∼1630 cm-1
,
only slightly less than the value observed for the model
compound, MBPC (see Table 2).
Our data provide strong support for the presence of the â-turn
in DCM. In the amide A region, the amide NH stretch is almost
entirely shifted to the lower wavenumber region (3325 cm-1),
and only a small absorbance band remains at 3445 cm-1 in the
region expected for a non-hydrogen-bonded amide (see Figure
3). In the amide I region, a shift of ∼12 cm-1 is observed for
the 3° amide carbonyl of the benzoyl-proline linkage relative
to the model compound, MBPC (see Table 2). A shift of this
magnitude is typical for a 3° amide carbonyl involved in a
hydrogen bond to an amide NH.24b No shift in the position of
the prolyl ester amide I absorption of Por/PL/DMB (ester 2, in
Table 2) relative to the model compounds MAPC and MBPC
is observed, indicating the absence of a γ-turn. In support of
this observation, O-Acetyl-L-lactyl-N-methylamide24b has only
one amide A absorbance centered at 3469 cm-1, indicating that
the ester carbonyl to amide NH hydrogen bond needed to form
a γ-turn in Por/PL/DMB is not very favorable. Thus, the â-turn
structure involving the benzoylproline tertiary amide, within the
limits of detection, is the only hydrogen-bonded conformation
formed by Por/PL/DMB. The 120-cm-1 shift of the amide A
band in the hydrogen-bonded state relative to the non-hydrogen-
bonded state indicates that Por/PL/DMB forms a type I â-turn
in DCM. For depsipetides, typically the type II â-turn results
in an 80-90-cm-1 shift in the amide A frequency, whereas the
type I â-turn produces a >100 cm-1 shift in the amide A
frequency due to the shorter hydrogen bond distance (2.89 versus
2.97 Å O-to-N distance) in the type I depsipeptide â-turn.24b
The hydrogen bond distance of Por/PL/Q of ∼2.89 Å is thus
very close to the optimal distance of 2.8 Å used in the pathway
model.9
Electron Transfer through a Hydrogen-Bonded â-Turn
Interface. Por/PL/Q was designed such that the shortest
through-bond electronic coupling pathway would require elec-
tronic coupling through a hydrogen bond. A number of donor-
acceptor systems involving electron transfer across peptide
bridges have been studied. Some variability in the decrease in
electron-transfer rate per peptide unit has been reported. For
short proline-bridged systems, the decrease in ket tends to be
on the order of 15-30-fold per proline unit.4,33 For short proline
bridges, a random conformation is expected. An electron-
transfer rate of 8.0 × 108 s-1 in DCM has been reported for
direct linkage via an amide bond of the donor and acceptor used
in this study (PAQ, Figure 1 and Table 4). Thus, insertion of
two bridging amino acid units would be expected to decrease
ket by a factor of ∼200-1000 for a random coil conformation.
Electron transfer would thus not be expected to compete well
with the intrinsic excited-state decay rate of the porphyrin (k )
1/τDMB ) 1.1 × 108 s-1) for Por/PL/Q in a random coil
conformation. The presence of secondary structure, however,
has been shown to significantly enhance the electronic coupling
in longer polyproline-bridged systems.4 In the case of Por/PL/
Q, the formation of a â-turn is expected to provide a shorter
through-hydrogen-bond electronic coupling pathway which
could significantly enhance the electron-transfer rate (four
covalent bonds + one hydrogen bond versus 10 covalent bonds
in the absence of the hydrogen bond of the â-turn). Clearly,
(31) (a) Siemiarczuk, A.; McIntosh, A. R.; Ho, T.-F.; Stillman, M. J.;
Roach, K. J.; Weedon, A. C.; Bolton, J. R.; Connolly, J. S. J. Am. Chem.
Soc. 1983, 105, 7224-7230. (b) Ho, T.-F.; McIntosh, A. R.; Weedon, A.
C. Can. J. Chem. 1984, 62, 967-974.
(32) (a) Hollosi, M.; Majer, Z.; Ronai, A. Z.; Magyar, A.; Medzihradszky,
K.; Holly, S.; Perczel, A.; Fasman, G. D. Biopolymers 1994, 34, 177-184.
(b) Jackson, M.; Mantsch, H. H. Biochim. Biophys. Acta 1991, 1078, 231-
235.
(33) Cabana, L. A.; Schanze, K. S. In Electron Transfer in Biology and
the Solid State; Johnson, M. K., King, R. B., Kurtz, D. M., Jr., Kutal, C.,
Norton, M. L., Scott, R. A., Eds.; ACS Advances in Chemistry Series 226;
American Chemical Society: Washington, DC, 1990; pp 101-124.
From a thermodynamic standpoint, the type I â-turn formed
by Por/PL/DMB is only weakly favored (Kâ-turn ) 6.0 and
∆G°â-turn ) -0.97 kcal/mol at 0 °C). This corresponds to a
â-turn population of about 85%, which is adequate for our
purposes. Kâ-turn is essentially temperature independent, which
is typical for â-turn-forming peptides and depsipeptides in
chloromethanes.24a The similarity of the thermodynamic prop-
erties of Por/PL/DMB, tBoc/PL/DMB, and Ac-PL-NHMe24

10910 J. Am. Chem. Soc., Vol. 120, No. 42, 1998
Williamson and Bowler
electron transfer competes with excited-state decay, indicating
that the presence of the â-turn conformation has the expected
effect.
of each exponential decay will be proportional to the fractional
occupancy of that conformation. For a two-state equilibrium,
such as the â-turn/random coil equilibrium, the time dependence
of the porphyrin excited state would be given by eq 9, where
A recent study on electron transfer with pyrazolate-bridged
iridium dimers34 has shown that the electronic coupling can vary
significantly as a function of conformationsas is clearly the
case heressuch that electron transfer occurs exclusively from
a given conformation. However, as these authors point out,
other factors, such as the reorganization energy, λ, change with
conformation, also modulating ket. Such effects are also
expected for the random coil conformation relative to the â-turn
conformation of Por/PL/Q. Electron transfer in the random coil
conformation is not competitive with that from the â-turn
conformation, not only because Hab is expected to be smaller
but also because ∆G° decreases in magnitude and λ increases
in magnitude due to the longer average donor-acceptor
separation expected in the random coil state of Por/PL/Q. Both
these effects are expected to decrease the magnitude of the
Franck-Condon factor in eq 1 and hence the magnitude of ket.
Thus, the noncompetitiveness of electron transfer from the
random coil conformation relative to the â-turn conformation
is not solely attributable to electronic coupling.
In any system where a conformational equilibrium exists
between more than one state, the potential effect of the
conformational equilibrium on the electron-transfer rate data
must be considered.1e If the rate constants for the equilibration
process are faster than the rate of electron transfer, then the
apparent rate of electron transfer will be a conformationally
weighted average of the electron-transfer rates from the various
conformations.35 Based on our IR data, the conformational
equilibrium for Por/PL/DMB can be approximated as a two-
state equilibrium involving a hydrogen-bonded type I â-turn
and a non-hydrogen-bonded random coil structure. Recent
studies36 on the rates of formation of ordered secondary structure
from random coil peptides indicate that â-hairpin structures form
with time constants on the order of 1 µs and R-helices form
with time constants of 100-200 ns. Since formation of an
R-helix is expected to be limited by the rate of nucleation of
the first turn of the helix, the time constant for formation of an
isolated â-turn should be no faster than the rate of helix
formation. In fact, the â-turn folding rate should be somewhat
slower than the R-helix folding rate since helix-forming peptides
have multiple possible nucleation sites, whereas an isolated
â-turn has only one way to form.
Kâ-turn
k
et,â-turn
)t
[Por*]total(t) ) [Por*]o,total
exp-(k +
+
f
((
)
1 + Kâ-turn
1
et,random)t
exp-(k + k
(9)
f
)
(
)
1 + Kâ-turn
[Por*]o,total is the total porphyrin emission intensity at t ) 0, kf
is the intrinsic fluorescence decay rate constant, ket,â-turn and
ket,random are the electron-transfer rate constants for the â-turn
and random conformations, respectively, and Kâ-turn is the
equilibrium constant for â-turn formation. Thus, for Por/PL/
Q, two exponential decays are possible. If not all of the quinone
is in the oxidized state, a third exponential due to Por/PL/QH2
would be expected, too. Spectroscopic data indicated that the
degree of oxidation of Por/PL/Q was 91% for the reported
lifetime measurements. Thus, ∼9% of the amplitude should
have a lifetime close to that of Por/PL/DMB. Since 85% of
Por/PL/Q is in the â-turn conformation, ∼77% of Por/PL/Q is
both oxidized and in the â-turn conformation in our sample.
This is very close to the amplitude (76%) of the fast lifetime of
Por/PL/Q (Table 3). The remaining ∼13% of Por/PL/Q is both
oxidized and in the non-hydrogen-bonded conformation. The
combined population of the unoxidized and non-hydrogen-
bonded states (∼22%) is close to the amplitude (24%) of the
slow lifetime of Por/PL/Q (see Table 3). This analysis indicates
that ket for the non-hydrogen-bonded conformation is not
competitive with the intrinsic fluorescence decay of the por-
phyrin, and thus its lifetime is indistinguishable from the decay
due to Por/PL/QH2.
To confirm this amplitude analysis, we monitored the
fluorescence decay of Por/PL/Q in DMSO. In this case, almost
all of the amplitude has a lifetime coincident with that of Por/
PL/DMB. Infrared studies in the amide I region indicate that
the â-turn is nearly completely disrupted (Figure 4, Table 2).
Thus, it would appear that electron transfer in the absence of
the â-turn conformation does not compete with the intrinsic
fluorescence decay. This confirms the assertion that the faster
lifetime observed in DCM is due to electron transfer in the
â-turn conformation and that electron transfer from non-â-turn
conformations does not compete with the intrinsic fluorescence
decay, resulting in a lifetime which is indistinguishable from
that due to Por/PL/QH2. The small residual fast lifetime in
DMSO (6% of total amplitude) appears to be due to a small
amount of residual â-turn conformation, since the electronic
coupling matrix element, Hab, calculated from ket extracted from
this lifetime is similar to that obtained for the â-turn conforma-
tion in DCM (see Table 4).
Electronic Coupling through a Hydrogen Bond versus a
Covalent Bond. One model which has had some success in
predicting electronic coupling in proteins in instances where
the homogeneous barrier model7,8 has proven inadequate is the
pathway model of Beratan and Onuchic.1a,d,e,9 In this model, it
is assumed that the electronic coupling matrix element, Hab, is
proportional to the product of electronic coupling across three
different types of orbital interactionsscovalent bonds, hydrogen
bonds, and through-space interactions (eq 10). Habo represents
Intrinsic fluorescence and electron transfer for Por/PL/Q occur
in the 1-10-ns time range, a time scale that is 1-2 orders of
magnitude faster than the time scale expected for the confor-
mational equilibrium between the â-turn and random coil states
of Por/PL/Q. When the conformational equilibrium between
states is slow relative to intrinsic fluorescence and electron-
transfer rates, each conformation is expected to produce a
separate emission decay. If the intrinsic fluorescence lifetimes
are assumed to be independent of conformation, the observed
emission lifetimes for each state will be controlled by the
electron-transfer rate from that state. The relative amplitudes
(34) Kurnikov, I. V.; Zusman, L. D.; Kurnikova, M. G.; Farid, R. S.;
Beratan, D. N. J. Am. Chem. Soc. 1997, 119, 5690-5700.
(35) (a) Hoffman, B. M.; Ratner, M. A.; Wallin, S. A. In Electron
Transfer in Biology and the Solid State; Johnson, M. K., King, R. B., Kurtz,
D. M., Jr., Kutal, C., Norton, M. L., Scott, R. A., Eds.; ACS Advances in
Chemistry Series 226; American Chemical Society: Washington, DC, 1990;
pp 125-146. (b) Hoffman, B. M.; Ratner, M. A. J. Am. Chem. Soc. 1987,
109, 6237-6243.
(36) (a) Eaton, W. A.; Munoz, V.; Thompson, P. A.; Chan, C.-K.;
Hofrichter, J. Curr. Opin. Struct. Biol. 1997, 7, 10-14. (b) Munoz, V.;
Thompson, P. A.; Hofrichter, J.; Eaton, W. A. Nature 1997, 390, 196-
199.
o
Hab ) Hab
ꢀ
ꢀ
ꢀ
(10)
∏ c∏ hb∏ ts
i
j
k
coupling of the donor and acceptor orbitals to the orbitals of

Electron Transfer through a H-Bonded â-Turn Interface
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10911
the molecular bridge, and ꢀc, ꢀhb and ꢀts represent electronic
coupling decay factors for covalent bonds, hydrogen bonds, and
through-space interactions, respectively.1e,9 The value of ꢀc has
been estimated to be 0.6, and ꢀhb has been estimated to be
equivalent to coupling across two covalent bonds for an optimal
proteins.1d,e,38,39 Results on the distance dependence of electron
transfer in peptide â-sheet mimics have also recently been
reported.5b To a first approximation, ket from the protein and
peptide systems all show the same distance dependence for
electron transfer (â ∼ 1.1 Å-1),1d,e,5b,38b indicating that electronic
coupling is similar in all of these systems and particular to the
secondary structure involved.1d,e,38b The surprising result from
work with the blue copper protein, azurin,1d,e,38,39 is that the
log(ket) versus distance dependence data for electron transfer
along different â-strands all fall on the same line.1e,38b However,
the â-strand extending out from the copper ligand, Cys 112, is
expected to be more strongly coupled to the copper center due
to the stronger electronic interaction of this ligand with the
copper center.1e,38 The explanation provided for the observed
data was that hydrogen bonds in â-sheets provide electronic
(2.8 Å) hydrogen bond (ꢀhb ) ꢀc ) 0.36).1e,9
2
To facilitate estimation of ꢀhb, we have calculated Hab from
our data (Table 4). The value of ꢀhb can be estimated by
comparison to previously reported data13 for the compound PAQ
(see Table 4). This compound has the donor and acceptor of
Por/PL/Q directly connected via an amide linkage (see Figure
1). The covalent pathway between the donor and acceptor has
four covalent bonds. Thus, the only difference in the covalent
pathways of PAQ and Por/PL/Q is the insertion of a hydrogen
bond. The value of ꢀhb can be calculated from eq 11 without
having to assign a value to ꢀc, since the ratio of the two
2
coupling similar to that of a covalent bond (ꢀhb ) ꢀc ), allowing
electron transfer from all â-strands to couple through Cys
112.1e,38 The data from our â-turn model system indicate that
this is the case for peptide hydrogen bonds and supports the
notion that hydrogen bonds of protein secondary structure are
crucial for efficient electronic coupling in proteins.
Hab(Por/PL/Q,â-turn) (ꢀc)4ꢀhb
)
(11)
(ꢀc)4
Hab(PAQ)
electronic coupling matrix elements is equal to ꢀhb. Since three
of the four covalent bonds in the dominant through-bond
pathways for Por/PL/Q and PAQ (see Figure 1) are identical,
the assumption that ꢀc cancels exactly is not unreasonable. The
emission lifetime data for Por/PL/Q and PAQ were obtained in
the same solvents, and the donor porphyrin and acceptor quinone
are the same, also making this an appealing comparison. We
obtain a value for ꢀhb of 0.77 using eq 11. When the error in
Hab resulting from the uncertainties in aPQ, aP, and aQ (see Table
4, footnote e) is accounted for the range of values obtained for
ꢀhb is 0.44-1.04. Thus, 0.8 ( 0.4 provides a reasonable
experimental estimate for ꢀhb. There will be additional error in
our evaluation of ꢀhb since we have ignored possible interference
effects37 from the completely covalent 10-bond pathway along
the backbone of Por/PL/Q (see Figure 1). We expect this effect
to be small relative to other sources of error. Our experimental
value for ꢀhb compares well to that of 0.51 reported by Therien
and co-workers for electron transfer across a carboxylic acid
hydrogen bond interface.10a
The amide NH- - -OdC hydrogen bond is the most prevalent
hydrogen bond in proteins. Significantly, the value of ꢀhb we
report here for this type of hydrogen bond is of the order of
magnitude predicted theoretically and provides direct evidence
that the hydrogen bond networks provided by protein secondary
structure can efficiently mediate electronic coupling. A key
component of the efficiency of electronic coupling across the
amide NH- - -OdC hydrogen bond likely results from the very
weak acidity of the donor NH and the very weak basicity of
the acceptor CdO. The Franck-Condon barrier due to proton
motion during electron transfer is expected to be minimal in
this case.10c,d
Conclusions
We have evaluated the electronic coupling across the
hydrogen bond of a â-turn using a porphyrin-quinone donor-
acceptor compound bridged by a depsipeptide. Qualitatively,
the data demonstrate that the hydrogen bond of a â-turn can
provide efficient electronic coupling since a very large rate
constant for electron transfer (ket ) 1.1 × 109 s-1) is observed
in the â-turn conformation. A quantitative assessment of the
coupling efficiency yielded a value of ꢀhb ) 0.8 ( 0.4, indicating
that peptide hydrogen bonds are efficient electron-transfer
mediators. This value for ꢀhb is of the same order of magnitude
as the theoretically predicted value of 0.36,9 it agrees well with
a previous estimate of ꢀhb ) 0.51 for a carboxylic acid hydrogen
bond interface,10a and it is consistent with other results indicating
efficient electron transfer across hydrogen-bonded interfaces.10b,c,d
Acknowledgment. We thank Mike Machczynski and Jay
Winkler at the Beckman Institute’s Laser Resource Center
(California Institute of Technology; Pasadena, CA) for their
willingness to help with the TCSPC fluorescence lifetime work.
We also thank Art Pardi at the Department of Chemistry and
Biochemistry at the University of Colorado at Boulder for help
with ROESY NMR experiments and Julanna Gilbert at the
University of Denver for use of FT-IR instrumentation. Ac-
knowledgment is made to the donors of the Petroleum Research
Fund, administered by the American Chemical Society, for
support of this research.
JA981384K
(38) (a) Regan, J. J.; Di Bilio, A. J.; Langen, R.; Skov, L. K.; Gray, H.
B.; Onuchic, J. N. Chem. Biol. 1995, 2, 489-496. (b) Langen, R.; Colon,
J. L.; Casimiro, D. R.; Karpishin, T. B.; Winkler, J. R.; Gray, H. B. J. Biol.
Inorg. Chem. 1996, 1, 221-225.
(39) (a) Farver, O.; Pecht, I. Biophys. Chem. 1994, 50, 203-216. (b)
Farver, O.; Skov, L. K.; Gilardi, G.; van Pouderoyen, G.; Canters, G. W.;
Wherland, S.; Pecht, I. Chem. Phys. 1996, 204, 271-277. (c) Farver, O.;
Bonander, N.; Skov, L. K.; Pecht, I. Inorg. Chim. Acta 1996, 243, 127-
133.
Comparison with Protein Systems and â-Sheet Mimics.
A significant body of work exists on electron transfer in â-sheet
(37) (a) Skourtis, S. S.; Onuchic, J. N.; Beratan, D. N. Inorg. Chim. Acta
1996, 243, 167-175. (b) Skourtis, S. S.; Beratan, D. N. J. Phys. Chem. B
1997, 101, 1215-1234.