Excited-state photophysics of hypericin and its hexamethoxy analog: Intramolecular proton transfer as a nonradiative process in hypericin

Article abstract of DOI:10.1021/ja962476h

The excited-state photophysics of the light induced antiviral agent, hypericin, are compared with those of its methylated analog, hexamethoxyhypericin. This comparison is instructive in understanding both the ground- and the excited-state properties of hy

Full text of DOI:10.1021/ja962476h

2980
J. Am. Chem. Soc. 1997, 119, 2980-2986
Excited-State Photophysics of Hypericin and Its Hexamethoxy
Analog: Intramolecular Proton Transfer as a Nonradiative
Process in Hypericin
D. S. English, W. Zhang, G. A. Kraus,* and J. W. Petrich*
Contribution from the Department of Chemistry, Iowa State UniVersity, Ames, Iowa 50011
ReceiVed July 17, 1996X
Abstract: The excited-state photophysics of the light induced antiviral agent, hypericin, are compared with those of
its methylated analog, hexamethoxyhypericin. This comparison is instructive in understanding both the ground- and
the excited-state properties of hypericin. That the hexamethoxy analog has no labile protons that can be transferred,
that it cannot protonate its own carbonyl groups, that it has a reduced fluorescence quantum yield and lifetime with
respect to hypericin, and that it exhibits no stimulated emission or, more specifically, rise time in stimulated emission
completely support our emerging model of the hypericin photophysics. The results are consistent with the presence
of intramolecular excited-state proton transfer in hypericin but not in its methylated analog.
Introduction
Interest in the polycyclic quinone, hypericin (Figure 1a) was
spawned by the discovery that it possesses extremely high
toxicity toward certain viruses, including HIV and that this
toxicity absolutely requires light.1-3 Hypericin is also very
similar in structure to the stentorin chromophore that confers
phototactic and photophobic responses to protozoan ciliates.4
The interaction of light with hypericin and hypericin-like
chromophores is clearly of fundamental biological importance.
In order to understand and eventually to exploit these properties
of hypericin, it is essential to elucidate its nonradiative excited-
state processes. We have undertaken this task using the tools
of ultrafast time-resolved absorption spectroscopy and have
presented our results in a series of articles.5-13
The argument for the presence of intramolecular excited-state
proton transfer in hypericin is as follows. The hypericin analog
* To whom correspondence should be addressed.
X Abstract published in AdVance ACS Abstracts, March 15, 1997.
(1) Carpenter, S.; Kraus, G. A. Photochem. Photobiol. 1991, 53, 169-
174.
(2) Meruelo, D.; Lavie, G.; Lavie, D. Proc. Natl. Acad. Sci. U.S.A. 1988,
85, 5230-5234. Degar, S.; Prince, A. M.; Pascual, D.; Lavie, G.; Levin,
B.; Mazur, Y.; Lavie, D.; Ehrlich, L. S.; Carter, C.; Meruelo, D. AIDS Res.
Hum. RetroViruses 1992, 8, 1929-1936. Lenard, J.; Rabson, A.; Vanderoef,
R. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 158-162. Meruelo, D.; Degar,
S.; Nuria, A.; Mazur, Y.; Lavie, D.; Levin B.; Lavie, G. In Natural Products
as AntiViral Agents; Chu, C. K., and Cutler, H. G., Eds. Plenum Press:
New York, 1992 pp 91-119, and references therein.
(3) Diwu, Z. Photochem. Photobiol. 1995, 61, 529-539.
(4) Dura´n, N.; Song, P.-S. Photochem. Photobiol. 1986, 43, 677-680.
Tao, N.; Orlando, M.; Hyon, J. S.; Gross, M. L.; Song, P.-S. J. Am. Chem.
Soc. 1993, 115, 2526.
(5) Gai, F.; Fehr, M. J.; Petrich, J. W. J. Am. Chem. Soc. 1993, 115,
3384-3385.
(6) Gai, F.; Fehr, M. J.; Petrich, J. W. J. Phys. Chem. 1994, 98, 5784-
5795.
(7) Gai, F.; Fehr, M. J.; Petrich, J. W. J. Phys. Chem. 1994, 98, 8352-
Figure 1. Two-dimensional structures of (a) hypericin, (b) hypericin
monotautomer, (c) hypericin bitautomer, (d) mesonaphthobianthrone,
(e) hexamethoxy hypericin, and (f) hypocrellin.
8358.
(8) Fehr, M. J.; Carpenter, S. L.; Petrich, J. W. Bioorg. Med. Chem.
Lett. 1994, 4, 1339-1344.
(9) Fehr, M. J.; McCloskey, M. A.; Petrich, J. W. J. Am. Chem. Soc.
1995, 117, 1833-1836.
lacking labile protons, mesonaphthobianthrone (Figure 1d), is
significantly fluorescent and has optical spectra that resemble
those of hypericin only when its carbonyl groups are proto-
nated.6,7 In hypericin, the fluorescent state grows in on a time
scale of several picoseconds, as measured by the rise time of
stimulated emission. Therefore, the combined observations of
the requirement of protonated carbonyls for strong hypericin-
(10) Carpenter, S.; Fehr, M. J.; Kraus, G. A.; Petrich, J. W. Proc. Natl.
Acad. Sci. U.S.A. 1994, 91, 12273-12277.
(11) Kraus, G. A.; Zhang, W.; Fehr, M. J.; Petrich, J. W.; Wannemuehler,
Y.; Carpenter, S. Chem. ReV. 1996, 96, 523-535.
(12) Fehr, M. J.; Carpenter, S. L.; Wannemuehler, Y.; Petrich, J. W.
Biochemistry 1995, 34, 15845-15848.
(13) Das, K.; English, D. S.; Fehr, M. J.; Smirnov. A. V.; Petrich, J. W.
J. Phys. Chem. 1996, 100, 18275-18281.
S0002-7863(96)02476-6 CCC: $14.00 © 1997 American Chemical Society

Hypericin and Its Hexamethoxy Analog
J. Am. Chem. Soc., Vol. 119, No. 13, 1997 2981
was added dropwise amide 2 (2.37 g, 10 mmol) in 15 mL of THF.
The mixture was stirred at -78 °C for 1 h, whereupon aldehyde 3
(1.66 g, 10 mmol) in 15 mL of THF was added dropwise over 5 min.
The resulting mixture was stirred at -78 °C for 1 h. Saturated
ammonium chloride solution (20 mL) was added. The solution was
diluted with 100 mL of ether and partitioned. The organic layer was
separated, washed with water and brine, and dried over MgSO4. After
concentration in vacuo, the residue was purified by sgc using 4:1 H:EA
to provide 3.90 g (97% yield) of 4 as a light yellow oil.
like fluorescence and the rise time of fluorescence in hypericin
was taken as evidence for intramolecular excited-state proton
transfer in hypericin.19 (In this article, we use the term “proton
transfer” very loosely. Our current knowledge of the excited
state transfer process is not detailed enough to specify whether
the process is a proton or an atom transfer.35)
This conclusion initially met with some resistance, despite
the precedent set by the large number of organic molecules
where the proximity of the enol and keto groups provides an
environment that is propitious for excited-state intramolecular
proton transfer or hydrogen atom transfer (in this article, we
use the two terms interchangeably): for example, malonalde-
hyde,14 salicylic acid,15 3-hydroxyflavone,16 benzothiazole,17 and
tropolone18 are but a few of the examples of a litany of species
that execute intramolecular excited-state proton transfer.
One argument proferred against intramolecular excited-state
proton transfer in hypericin is the observation of near mirror-
image symmetry between its absorption and emission spectra.
Such symmetry is typically taken as a signature of negligible
structural changes between the absorbing and the emitting
species. Intramolecular excited-state proton transfer usually
generates a broad structureless emission spectrum that bears little
resemblance to the absorbance spectrum. 3-Hydroxyflavone
provides a good example.
There are at least two ways to respond to this objection. It
is possible that the structural changes induced by proton transfer
do not significantly effect the electronic structure of the
tautomeric species (Figure 1a-c) in such a way as to destroy
the mirror-image symmetry. High-level quantum chemical
calculations will have much to offer in understanding this
problem. It is also possible, as we have argued elsewhere,6,7
that the ground-state of hypericin is already partially tautomer-
ized and that this ground-state heterogeneity yields the observed
mirror-image symmetry between absorption and emission
spectra.
Here we revisit the questions of ground-state heterogeneity
and nonradiative processes. We compare the excited-state
transients of hypericin obtained previously using an excitation
wavelength of 588 nm6,7 with results obtained from a regen-
eratively amplified and frequency-doubled Ti:sapphire laser
providing an excitation wavelength of 415 nm. Changes in the
kinetics using different excitation wavelengths is suggestive of
ground-state heterogeneity in hypericin. We also present the
synthesis of the hypericin analog with no labile protons,
hexamethoxyhypericin (Figure 1e, compound 1), and compare
its excited-state transients with those of hypericin. This analog
displays no rise time for stimulated emission, which we have
taken as the signature of intramolecular excited-state proton
transfer, and it displays no excited-state transients that can be
interpreted in terms of proton transfer.
4: NMR (CDCl3) d 6.61 (d, J ) 3 Hz, 1 H), 6.53 (d, J ) 3 Hz, 2
H), 6.50 (d, J ) 3 Hz, 2 H), 6.38-6.30 (m, 4 H), 6.26 (t, J ) 3 Hz,
1 H), 5.71 (s, 1 H), 5.53 (d, J ) 6 Hz, 1 H), 5.35 (d, J ) 9 Hz, 1 H),
3.83 (s, 3 H), 3.76 (s, 6 H), 3.74 (s, 3 H), 3.72 (s, 6 H), 3.68 (s, 3 H),
3.68-2.40 (m, 8 H), 1.3-0.75 (m, 12 H); IR (KBr) 3367, 2938, 1602,
1459, 1154, 840 cm-1; MS (EI) m/e 403, 330 (100), 315, 299, 271,
193; HRMS C22H29O6 calcd 403.19949; measured 403.19919.
ii. 5,7-Dimethoxy-3-(3′,5′-dimethoxyphenyl)isobenzofuran-1-one,
5. A solution of 4 (4.00 g, 10 mmol) and pTSA (100 mg) in 200 mL
of toluene was heated to reflux for 2 h. The organic solution was
washed with 2% sodium bicarbonate, water, and brine and dried.
Removal of the solvent by concentration in vacuo yielded 3.27 g (99%
yield) of 5 as white crystals (mp 182-184 °C).
1
5: H NMR (CDCl3) d 6.39 (m, 4 H), 6.28 (m, 1 H), 6.09 (brs, 1
H), 3.94 (s, 3 H), 3.79 (s, 3 H), 3.73 (s, 6 H); IR (KBr) 2843, 1771,
1600, 1113, 1007, 833 cm-1 13C NMR (CDCl3) d 55.52, 56.03, 56.15,
.
81.34, 98.38, 99.20, 100.91, 104.70, 106.16, 139.18, 154.68, 159.57,
161.25, 167.02, 168.45; MS (EI) m/e 330 (100), 300, 271, 257, 193,
165; HRMS C18H18O6 calcd 330.11034, measured 330.10952.
iii. 2,4-Dimethoxy-6-(3′,5′-dimethoxyphenylmethyl)benzoic acid,
6. A mixture of 5 (300 g, 90 mmol) and 10% Pd/C (500 mg) in 100
mL of ethyl acetate was stirred under a hydrogen atmosphere for 48 h
The suspension was carefully filtered, and the solvent was removed in
vacuo The resulting product was purified by recrystallization from
H:EA to provide 278 g (93% yield) of 6 as colorless crystals (mp 163-
164 °C).
6: 1H NMR (CDCl3) d 6.39 (d, J ) 3 Hz, 1 H), 6.36 (d, J ) 2 Hz,
1 H), 6.33 (d, J ) 2 Hz, 1 H), 6.28 (d, J ) 3 Hz, 1 H), 4.23 (br s, 2
H), 3.92 (s, 3 H), 3.75 (s, 3 H), 3.72 (s, 6 H); IR (KBr) 2930, 1694,
1606, 1153, 1081, 844; 13C NMR (CDCl3) d 170.32, 162.29, 160.79,
159.24, 144.30, 142.60, 144.30, 142.60, 113.57, 108.31, 107.32, 98.30,
96.82, 56.41, 55.47, 55.31, 40.07; MS (EI) m/e 332, 314, 299, 241,
213, 139; HRMS C18H20O6 calcd 332.12599, measured 332.12572.
iv. 1,3,6,8-Tetramethoxyanthracen-9-one, 7. To a solution of 6
(1.66 g, 5.0 mmol) in 100 mL of methylene chloride at 0 °C was added
trifluroacetic anhydride (1.05 g, 5.0 mmol) dropwise over 10 min. The
solution was stirred at 0 °C for 2 h. Twenty milliliters of methanol
was added, followed by 150 mL of 5% sodium bicarbonate. The
aqueous layer was extracted with 30 mL of methylene chloride. The
combined organic layers were washed with water and brine and dried.
After concentration in vacuo, the residue was purified by sgc using
5:1 H:EA to provide 1.41 g of 7 (90% yield) as light yellow crystals
(mp 207-208 °C).
7: 1H NMR (CDCl3) d 10.65 (s, 1 H), 7.43 (s, 1 H), 6.60 (d, J ) 3
Hz, 2 H), 6.29 (d, J ) 3 Hz, 2 H), 4.01 (s, 6 H), 3.89 (s, 6 H); IR
(KBr) 3319, 2996, 1630, 1578, 1562, 1370, 808; MS (EI) m/e 314
(100), 285, 271, 213, 185, 157; HRMS C18H18O5 314.11542, measured
314.11554.
v. 1,3,4,6,8,10,11,13-Octamethoxy phenanthro[1,10, 9, 8-opqra]-
perylene, 1. To a solution of anthrone 7 (1.30 g, 4.0 mmol) in 20 mL
of boiling EtOH was added a solution of ferric chloride hexahydrate
(1.48 g, 5.5 mmol) in 40 mL of EtOH over 10 min. The mixture was
heated at reflux for 30 min, cooled, and poured into 600 mL of 5%
HCl. The aqueous mixture was extracted three times with 100 mL of
EtOAc. The combined organic layers were washed with water and
brine and dried over magnesium sulfate.
After the solvent was concentrated in vacuo, the residue was mixed
with 5 g of KOH and 100 mL of EtOH and heated to reflux for 5 min.
The mixture was cooled to 0 °C, and K2S2O8 (1.35 g, 15.0 mmol) in
20 mL of water was added dropwise over 5 min at 0 °C. The mixture
was stirred at ambient temperature for 2 h. The reaction mixture was
then poured into 500 mL of 1% HCl and extracted three times with
100 mL of ethyl acetate. The combined organic layers were washed
Experimental Section
A. Synthesis. i. N,N-Diethyl 2,4-Dimethoxy-6-(hydroxy-3′,5′-
dimethoxyphenylmethyl)benzamide, 4. To a solution of TMEDA
(1.28 g, 11 mmol) and s-BuLi (11 mmol) in 50 mL of THF at -78 °C
(14) Carrington, T., Jr.; Miller, W. H. J. Phys. Chem. 1986, 84, 4364.
(15) Barbara, P. F.; Walsh, P. K.; Brus, L. E. J. Phys. Chem. 1989, 93,
29-34.
(16) (a) Brucker, G. A.; Swinney, T. C.; Kelley, D. F. J. Phys. Chem.
1991, 95, 3190-3195. (b) Strandjord, A. J. G.; Barbara, P. F. J. Phys. Chem.
1985, 89, 2355-2361. (c) McMorrow, D.; Kasha, M. J. Phys. Chem. 1984,
88, 2235-2243. (d) Brucker, G. A.; Kelley, D. F. J. Phys. Chem. 1987,
91, 2856-2861. (e) Schwartz, B. J.; Peteanu, L. A.; Harris, C. B. J. Phys.
Chem. 1992, 96, 3591-3598.
(17) (a) Frey, W.; Laermer, F.; Elsaesser, T. J. Phys. Chem. 1991, 95,
10391-10395. (b) Laermer, F.; Elsaesser, T.; Kaiser, W. Chem. Phys. Lett.
1988, 148, 119-124.
(18) Ensminger, F. A.; Plassard, J.; Zwier, T. S.; Hardinger, S. J. Chem.
Phys. 1993, 99, 8341-8344.

2982 J. Am. Chem. Soc., Vol. 119, No. 13, 1997
English et al.
Scheme 1
with water and brine and dried. After removing the MgSO4 by
filtration, the filtrate was irradiated in a Pyrex flask with a 500 W
tungsten lamp for 1 h. The solvent was concentrated in vacuo, and
the residue was purified by sgc using 30:1 methylene chloride:methanol
to afford 0.78 g (61% yield) of 1 as a red solid (mp 267 °C dec).
1: 1H NMR (CDCl3) d 6.98 (s, 4 H), 4.19 (s, 12 H), 4.13 (s, 12 H).
IR (KBr) 2930, 1637, 1540, 1239, 1141, 993 cm-1. MS (ESI) m/e
620 (100), 574, 412, 301.
B. Time-Resolved Measurements. Hypericin was used as received
from Carl Roth GmbH (distributed by Atomergic Chemetals Corp).
DMSO was dried over 4-Å molecular sieves. Care was taken to prevent
sieve dust contamination. All experiments were performed at ambient
temperature (22 °C) The time-correlated single-photon counting experi-
ments were performed with an apparatus described elsewhere.20 The
instrument response function is typically about 70 ps. Hexamethoxy
analog samples and hypericin samples were prepared for pump-probe
experiments with an optical density of 0.4-0.7 at the pump wavelength;
for time-correlated single-photon counting, samples were diluted by
10-fold or more. The time-resolved fluorescence data provide a
powerful assay of the purity of the hexamethoxy analog. In addition
to a 480-ps decay component, there is a small amount of an ∼2-ns
component whose contribution, possibly a result of oxidation, could
be diminished by column chromatography (Figure 7a). Purification,
however, did not affect the transient absorption data.
Steady-state fluorescent measurements were made using a Spex
Fluoromax with a 4-nm band-pass, and corrections were made for
detector response and excitation lamp spectrum. Steady state absor-
bance measurements were made using a Perkin Elmer Lambda 18
double beam UV-vis spectrophotometer with 1-nm bandpass. The
measurement of the fluorescence quantum yield of the hexamethoxy
analog was made relative to hypericin by exciting solutions of each
compound at several wavelengths where they had equal optical densities
and integrating the emission spectrum over the entire band on a
wavenumber scale.
Figure 2. Normalized steady-state absorption (s) and emission (- -)
spectra of hypericin (a) and hexamethoxyhypericin (b) in DMSO. The
excitation wavelength for (a) is 550 nm and for (b) 410 nm.
bandwidth. The output of the amplifier is incident on a beam splitter
which sends 56% of the light down a variable delay line and is
frequency doubled using a 1-mm BBO crystal. The remaining light
traverses a static delay line and is used to create a white light continuum
by focusing with a 20-cm lens into a 3-mm thick sapphire flat which
is translated to prevent damage. The white light is then passed through
the appropriate bandpass filters and split into a probe and reference
beam. The pump beam is chopped at 1/2 the repetition rate by a Palo
Alto Research Super Chopper 300, which is triggered by the third
boxcar averager in the signal processing discussed below. The beams
are focused with a 12.5-cm focal length lens and crossed at 36° in a
spinning sample holder33 that rotates at 3400 rpm ensuring a new sample
area for every shot. The reference and probe beams are directed into
an ISA HR-320 monochromator with a 1200 g/mm grating. The probe
and reference beams are detected with Hamamatsu S1336-5BQ
photodiodes, which are amplified using Analog Devices op467 precision
high-speed OP amps. The outputs are then sent to two Stanford
Research SRS250 gated boxcar averagers and subsequently to a
Stanford Research SR235 analog processor where the probe signal is
divided by the reference to provide a shot-to-shot normalization. The
natural log of the quotient is taken and sent to a third boxcar averager,
which is operated in toggle mode to provide an active baseline
subtraction. Coumarin 500 was probed at 500 nm and used as a
standard. Betaine was also used as a standard and compared to previous
data collected in a similar fashion.23 The lower limit of the sensitivity
of the detection system was estimated with one scan and without shot-
to-shot normalization as a change of 0.016 V in an initial signal of
0.750 V, which corresponds to a worst case limit of a 9.4 × 10-3 change
in optical density.
The pump probe experiments were carried out on a 2 kHz system
based on a Ti:sapphire chirped-pulse regenerative amplifier.21 Typical
pulse widths were 150 to 200 fs fwhm at 830 nm from autocorrelation
measurements. The regenerative amplifier seed pulse train is supplied
by a homemade Ti:sapphire oscillator of a Murnane-Kapetyn design.22
The rod is 4.75-mm, and typical pulse widths are sub 20 fs as
determined by measurement of the pulse autocorrelation and the spectral
(19) Of special relevance to the role of labile protons for light-induced
antiviral activity is the observation that hypericin retains its toxicity in the
absence of oxygen and that it acidifies its surroundings upon light
absorption.8,9,11,12 The retention of toxicity in the absence of oxygen excludes
unique assignment of antiviral activity to the trivial generation of singlet
oxygenseven though hypericin does generate triplets in high yield.25-28
(20) Gai, F.; Rich, R. L.; Petrich, J. W. J. Am. Chem. Soc. 1994, 116,
735.
An “artifact” appears frequently in the data and is most evident in
the kinetic traces displayed in Figure 3. The “spike” that appears at
zero time with either positive or negative sign is a result of the more
intense pump pulse modulating the phase of the less intense probe pulse.
This phenomenon of cross-phase modulation occurs with pulses of
different wavelength and does not require the pulses to be resonant
with an absorbing sample.31
(21) Salin, F.; Squier, J.; Vallincourt, G.; Mourou, G. Optics Lett. 1991,
16, 1964.
(22) Asaki, M. T.; Huang, C.-P.; Garvey, D.; Zhou, J.; Kapteyn, H. C.;
Murnane, M. M. Optics Lett. 1993, 18, 977.
(23) Johnson, A. E.; Levinger, N. E.; Jarzeba, W.; Schlief, Ralph E.;
Kliner, D. A. V.; Barbara, P. Chem. Phys. 1993, 176, 555-574.

Hypericin and Its Hexamethoxy Analog
J. Am. Chem. Soc., Vol. 119, No. 13, 1997 2983
Table 1. Global Fitting Parametersa
hypericin (nm) a1 τ1 (ps) a2 τ2 (ps) a3 τ3 (ps)
550a
0
11.6 1.0
∞
∞
∞
∞
∞
0
0
0
0
0
0
0
0
0
0
600a
620a
630a
640a
650b
660b
670a
680a
700a
0.30 11.6 -1.0
0.37 11.6 0.63
0.25 11.6 0.75
0.10 11.6 0.90
-0.61 11.6 -0.39 ∞
-1.0 11.6 0.27
0.74 11.6 0.26
0.30 11.6 0.70
0.32 11.6 0.68
∞
∞
∞
∞
hexamethoxy hypericin a1
τ1 (ps) a2
τ2 (ps) a3
τ3 (ps)
∞
∞
∞
∞
∞
∞
∞
∞
500c
510c
515b
520b
530a
540a
545d
580d
630d
0
2.5 0.68 480
0.32
0.10 2.5
0.17 2.5
0.10 2.5
0.72 480
-0.91 480
-1.2 480
-3.0 480
-1.2 480
0.39 480
0.35 480
0.30 480
0.18
0.83
0.90
1.0
0.89
0.61
0.50
0.70
0
2.5
0.11 2.5
2.5
0.15 2.5
2.5
0
0
∞
a The functional form used to fit the data is indicated by a superscript
in the leftmost column, which gives the probe wavelength. (a) ∆A(t)
) a1 exp(-t/τ1) + a2 exp(-t/τ2) + a3 exp(-t/τ3). (b) ∆A(t) ) a1[1 -
exp(-t/τ1)] + a2 exp(-t/τ2) + a3 exp(-t/τ3). (c) ∆A(t) ) a1 [1 - exp(-
t/τ1)] + a2[1 - exp(-t/τ2)] + a3 exp(-t/τ3). (d) ∆A(t) ) a1 exp(-t/τ1)
+ a2[1 - exp(-t/τ2)] + a3 exp(-t/τ3). The pre-exponential factors
are all normalized and should not be interpreted in terms of absolute
optical density changes owing to changes in pump intensities and gain
settings on data acquisition hardware.
our case were time constants, and local parameters were amplitudes or
pre-exponential factors. The time constants are specified, and each
curve is fit iteratively varying only the local parameters, that is the
amplitudes. A local ø2 is calculated for each local fit. After all curves
are fit, a global ø2 is calculated as shown below. The global parameters
are varied, and the whole process is repeated to calculate a new global
ø2. This process is repeated until a minimum is reached for ø2, which
is defined as:
Figure 3. Absorption transients of hypericin in DMSO, λpump ) 415
nm; probe wavelengths are indicated in the panels. The “spike” at zero
time that is most apparent in the data in this figure is a result of cross-
phase modulation of the pump pulse with the probe pulse.31 See
Experimental Section.
ø2local,i
n
∑
2
i)1
øbest,i
ø2 )
n
The instrument response of the 30-Hz and 2-kHz systems are
determined, respectively, by means of the ground-state bleach of nile
blue and the stimulated emission of coumarin 500 on a daily basis. On
a picosecond time scale, nile blue has an instantaneous bleach.6,7 On
a 100 femtosecond time scale, coumarin 500 in a 1:1 mixture of
methanol and water yields stimulated emission at 500 nm with no
detectable rise time. These traces are used to approximate the
instrument response (the convolution of the laser pulses in the medium)
by performing an iterative nonlinear least-squares fit in which a trial
instrument response is convoluted with the instantaneous molecular
response of the nile blue or the coumarin. Several trial instrument
response functions are tested until one is found that best fits the rising
edge of the data. This function is then used in subsequent analyses
until the experiment is either restarted or is known to have changed.
The trial functions used are either double-sided exponentials or
guassians, which have been convoluted with themselves in order to
simulate the temporal overlap of two similar pulses in the sample.
Sample preparation for the pump-probe experiments required bub-
bling argon through the hexamethoxy analog samples and filling the
spinning sample holder in a glove bag of argon to reduce oxygen
content. This was done after observing rapid degradation of sample
in the presence of oxygen. Deoxygenation sufficiently preserved the
sample to withstand pump incidence for several hours; without
deoxygenation the sample degrades in minutes. Hypericin is stable
for hours without deoxygenation and gives the same experimental results
in either case.
where ø2best,i is obtained for each curve letting all parameters local and
global (that is amplitudes and time constants) vary, and n is the number
of curves. The quality of the fit was determined by visual inspection
of the fit and its residuals. The time-resolved absorption/stimulated
emission data are displayed in Figures 3-6. The results of the global
fits, and the functional forms used to obtain them, are summarized in
Table 1.
Results and Discussion
A. Hypericin. i. Rise Time in Stimulated Emission.
Figures 3 and 4a present absorption transients of hypericin in
DMSO using excitation wavelengths of 415 and 588 nm,
respectively. The changes in signs of the traces in Figure 3
indicate the presence of isosbests for hypericin in the regions
550-600, 600-620, 640-650, and 660-670 nm. The decay
times of the absorption transients at probe wavelengths of 620-
640 and at 670-700 nm and the rise time of the stimulated
emission transients at probe wavelengths of 600 and 650-660
nm are all adequately represented, using a global fitting analysis,
by the same time constant of 11.2 ps. This result is consistent
with that of 9.2 ps obtained previously using ∼1-ps pulses at
588 nm at 30 Hz.6,7
ii. Ground-State Heterogeneity: Excitation Wavelength
Dependence. We call attention to the traces obtained using
Kinetic traces acquired at various wavelengths were fit using a global
fitting procedure found in Spectra Solve. The global parameters in

2984 J. Am. Chem. Soc., Vol. 119, No. 13, 1997
English et al.
Figure 4. Stimulated emission transients of hypericin, λpump ) 588
nm; probe wavelengths are indicated in the panels:7 (a) DMSO, ∆A(t)
) -0.30[1 - exp(-t/9.2 ps)] - 0.13 exp(-t/1.9 ps); and (b) ethylene
glycol, ∆A(t) ) -0.45[1 - exp(-t/6.4 ps)] - 0.53 exp(-t/2.4 ps).
probe wavelengths of 658-660 nm for excitation wavelengths
of 415 and 588 nm, respectively (Figures 3 and Figure 4). When
hypericin is excited at 588 nm, there is a component of
stimulated emission that appears instantaneously. This com-
ponent is absent for excitation at 415 nm. Changes in the
kinetics with excitation wavelength is taken to be suggestive
of ground-state heterogeneity of the hypericin population. The
origin of this apparent heterogeneity may be attributed to various
tautomers as well as to different twisted conformations of the
aromatic skeleton. See below. (We refer to the heterogeneity
as apparent because different laser systems were used to obtain
the two results.)
Figure 5. Absorption transients of hexamethoxy hypericin in DMSO,
λpump ) 415 nm; probe wavelengths are indicated in the panels. The
full scale is 40 ps.
transients observed at other probe wavelengths in Figures 5 and
6 clearly resolve a species with this time constant.
We do not attribute the 2.5-ps decays observed at 580 and
610 nm to the presence of weak stimulated emission, because
a genuine decay of absorbance on this time scale is already
consistent with the apparent rise time for the bleach at 540 nm,
as discussed above. Furthermore, Figure 5 indicates that in this
region there is a component of transient absorption appearing
with a 480-ps time constant. This is identical to the fluorescence
decay time of hexamethoxyhypericin in DMSO (Figure 7a) and
can be interpreted as the formation of triplet species. If the
fast transient at 580 and 610 in Figure 5 were due to the
superposition of a stimulated emission signal on a long-lived
absorption signal, it is reasonable to expect that we would also
observe simulated emission from the long-lived fluorescence
species as well, that is an apparent decay of absorbance of 480
ps. That we do not provides us with additional confidence to
conclude that our data provide no evidence of detectable
stimulated emission.
ii. Origin of the 2.5-ps Component in Hexamethoxy
hypericin. A possible explanation for the presence of the
component of transient absorption that decays in 2.5 ps is the
combination of internal conversion and dynamic solvation of
an excited state of the analog. As the solvent dynamically
readjusts itself to the change in the dipole of the solute upon
optical excitation, the solute is lowered in energy. This is
manifested by the decay of transient absorbance. The dynamic
B. Hexamethoxy Hypericin. i. Absence of Stimulated
Emission. Figures 5 and 6 present absorption transients of
hexamethoxyhypericin in DMSO using an excitation wavelength
of 415 nm. Isosbests are apparent at ∼515 nm and between
540 and 545 nm. The kinetics of hexamethoxyhypericin are
simplified with respect to those of hypericin. No stimulated
emission is detected for hexamethoxyhypericin in the region
550-710 nm. This is not particularly surprising given that we
measure its fluorescence quantum yield in DMSO to be only
19%32 that of hypericin. Any weak contribution from stimulated
emission is eclipsed by the absorption of excited-state species.
What is especially noteworthy, however, is that the transient
absorption in the spectral region where only emission is obserVed
in the steady state, >610 nm, occurs instantaneously and does
not decay on the time scale of the experiment. If there were a
component of stimulated emission that appeared with a finite
rise time, it would manifest itself as an apparent decay in the
transient absorption signal.
Examination of the steady-state spectra in Figure 2 indicates
that the transient at 540 nm arises from a bleach of the ground-
state absorption and not to a rise time in stimulated emission.
The apparent rise time in the bleach, whose time constant is
2.5 ps, must correspond to an excited-state species absorbing
at 540 nm and decaying with the same time constant. The rapid

Hypericin and Its Hexamethoxy Analog
J. Am. Chem. Soc., Vol. 119, No. 13, 1997 2985
Figure 7. Fluorescence decays obtained from time-correlated single-
photon counting at 22 °C. Residuals are displayed at the top of the
figure. The topmost set of residuals corresponds to hypericin. (a)
Hexamethoxyhypericin in DMSO: λex ) 300 nm, λem > 530 nm; F(t)
) 0.97 exp(-t/490 ps) + 0.03 exp(-t/2.1 ns), ø2 ) 1.15. The full
scale is 3 ns. (b) Hypericin in DMSO: λex ) 305 nm, λem > 555 nm;
F(t) ) 1.00 exp(-t/5.3 ns), ø2 ) 1.18. The full scale is 15 ns. To our
knowledge, the fluorescence decay of hypericin and its analogs does
not depend on excitation wavelength, visible and ultraviolet excitation
yielding similar results.6,34 This result should not be confused with the
excitation wavelength dependence of the picosecond transients in
hypericin indicated in Figures 3 and 4. These transients report on
primary processes that are not resolvable in the measurement of the
longer lived fluorescence decay.
Figure 6. Absorption transients of hexamethoxy hypericin in DMSO,
λpump ) 415 nm; probe wavelengths are indicated in the panels. The
full scale is 320 ps. The slowly rising transient apparent in each panel
has a time constant of 480 ps, which matches the fluorescence decay
of the analog in DMSO. This transient is attributed to the formation of
a triplet species.
solvation time of DMSO is known to be 3 ps.24 It may, of
course, be fortuitous that the transient is characterized by an
∼3-ps decay; and ideally, a correlation of decay time in various
solvents with the known dynamic solvation times would enable
us to make this assignment unambiguously. Unfortunately,
DMSO is the only solvent in which we are able to dissolve the
analog in sufficient quantities to perform these experiments and
at the same time avoid sample degradation.
Another possibility for the origin of this component is prompt
intersystem crossing to form triplets. We have already noted
that the fluorescence quantum yield of hexamethoxy hypericin
in DMSO is 19%32 that of hypericin and that hexamethoxyhy-
pericin degrades rapidly when optically excited in the presence
of oxygen. These results suggest that the reduced fluorescence
quantum yield of this analog may be the result of enhanced
intersystem crossing and that the high yield of triplet states
effectively generates singlet oxygen, which subsequently de-
stroys it.
for stimulated emission that we have attributed to intramolecular
proton transfer. One is led to conjecture that the 2.5-ps transient
observed at 580-610 nm in hexamethoxyhypericin and the 2-ps
transient in hypericin represent prompt intersystem crossing. (By
“prompt” intersystem crossing we mean merely to distinguish
this process from the slower process occurring with a 480-ps
time constant. Presumably these two intersystem crossing rates
arise from two different conformational or tautomeric species.)
If this is correct, the kinetic scheme postulated for hypericin in
Figures 12 and 3 of refs 6 and 7, respectively, needs to be
modified. There we proposed that the ∼2-ps decay of stimu-
lated emission in hypericin represents the production of a
monotautomer (Figure 1b) from the untautomerized species
(Figure 1a). The presence of a similar rapid transient in
hexamethoxyhypericin, which cannot tautomerize, considered
in the context of the above results, suggests that this fast
component in hypericin may correspond to ultrafast intersystem
crossing.
In a recent study of the perylene quinone, hypocrellin (Figure
1f), we observe absorption that is produced on the time scale
of the fastest kinetic event and have argued that it arises from
“prompt” triplet formation.13
iii. Rapid Transients in Hypericin and Its Analogs. The
stimulated emission kinetics of hypericin in ethylene glycol
(Figure 4b) indicate a component that appears instantaneously
and decays in ∼2 ps in addition to the characteristic rise time
Conclusions
A procedure has been presented for the synthesis of a crucial
analog in the investigation of the hypericin, hexamethoxyhy-
pericin (Figure 1e). Hexamethoxyhypericin has a fluorescence
(24) Barbara, P. F.; Jarzeba, W. AdV. Photochem. 1990, 15, 1.

2986 J. Am. Chem. Soc., Vol. 119, No. 13, 1997
English et al.
quantum yield that is only 19% that of hypericin in DMSO and
that exhibits no stimulated emission in DMSO. Based on our
previous work on hypericin, these results are consistent with
the inability of the hexamethoxy analog to execute an intramo-
lecular excited-state proton transfer.
chains protruding from the aromatic skeletons repel each other
and prevent the molecules from acquiring a planar conformation.
This raises the possibility that there are ground-state conforma-
tions of hypericin that cannot tautomerize and that, upon optical
excitation, undergo ultrafast deactivation by other processes such
as internal conversion or intersystem crossing. One can
speculate on how distortion of the skeleton in the hexamethoxy
derivative affects its spectroscopy: the absorption spectra of
the analog and hypericin are qualitatively very similar except
for the strong absorption at 400 nm in the analog (Figure 2).
Hexamethoxyhypericin possesses a rapid component in the
decay of its transient absorption (2.5 ps). We suggest that a
likely explanation for this component is a combination of
dynamic solvation and internal conversion of an excited singlet
state higher than S1. (Prompt intersystem crossing may also
be possible.) We further note that an assignment to intersystem
crossing implies that the rapid component (∼2 ps) of stimulated
emission, which is clearly evident in hypericin experiments
obtained with excitation wavelengths of 588 nm6,7 (Figure 5
part a and especially part b), may also have a similar origin.
An interesting question that requires further study is how does
the conformation of the hypericin skeleton influence its elec-
tronic structure and its photophysics. The X-ray structures of
hypericin30 and hypocrellin29 indicate that they are not planar.
(Similar conclusions can be drawn from semiempirical calcula-
tions or, more simply, from molecular models.) The large side
Acknowledgment. D.S.E. was supported by fellowships
from Amoco and GAANN. Part of this work was supported
by NSF Grants BIR-9413969 and CHE-9613962 to J.W.P. and
by a grant from the Center for Advanced Technology and
Development to G.A.K.
JA962476H
(31) Diels, J.-C.; Rudolph, W. Ultrashort Laser Pulse Phenomena;
Academic Press: San Diego, 1996; p 142.
(32) Because this number is obtained from a steady-state measurement,
which does not discriminate against the 2-ns impurity or degradation product
(Figure 7a), it overestimates the fluorescence quantum yield of the analog.
The decrease in fluorescence quantum yield of the analog is more clearly
apparent in the decay of its excited singlet state: in DMSO, it possesses a
fluorescence decay of 500 ps, as opposed to 5 ns for that of hypericin (Figure
7). Consequently, a large component of the triplet population in hexa-
methoxyhypericin is expected to appear with a 500 ps time constant, as is
observed in Figure 6.
(33) Savikhin, S.; Wells, T.; Song, P.-S.; Struve, W. S. Biochemistry
1993, 32, 7512.
(34) Yamazaki, T.; Ohta, N.; Yamazaki, I.; Song, P.-S. J. Phys. Chem.
1993, 97, 7870.
(25) Racinet, H.; Jardon, P.; Gautron, R. J. Chim. Phys. 1988, 85, 971-
977.
(26) Eloy, D.; Le Pellec, A.; Jardon, P. J. Chim. Phys. 1996, 93, 442-
457.
(27) Diwu, Z.; Lowen, J. W. Free Radical Biol. Medicine 1993, 14, 209-
215.
(28) Malkin, J.; Mazur, Y. Photochem. Photobiol. 1993, 57, 929-933.
(29) Wei-shin, C.; Yuan-teng, C.; Xiang-yi, W.; Friedrichs, E.; Puff, H.;
Breitmaier, E. Liebigs Ann. Chem. 1981, 1880-1885.
(30) (a) Etzlstorfer, C.; Falk, H.; Mu¨ller, N.; Schmitzberger, W.; Wagner,
U. G. Monatsh. Chem. 1993, 124, 751-761. (b) Falk, H. Personal
communication. (c) Freeman, D.; Frolow, F.; Kapinus, E.; Lavie, D.; Lavie,
G.; Meruelo, D.; Mazur, Y. J. Chem. Soc. Chem., Commun. 1994, 891-
892.
(35) Das, K.; English, D. S.; Petrich, J. W. J. Am. Chem. Soc. 1997, in
press.