Concerning the absorption and photochemical properties of an ω-4-dimethylaminobenzal hypericin derivative

Article abstract of DOI:10.1007/s007060170009

A hypericin derivative containing ω,ω′ -4-dimethylaminobenzal residues was shown to undergo an intramolecular [2 + 2] cycloaddition upon irradiation leading to a cyclobutane derivative whose main absorption band is hardly shifted as compared to hypericin. The corresponding ω-substituted derivative displayed a 34 nm bathochromic shift and a strongly reduced fluorescence quantum yield rendering it a nice candidate for a photodynamic therapy agent. Unfortunately, however, it produced virtually no photosensitized active oxygen species, making it thus unsuited for this purpose.

Full text of DOI:10.1007/s007060170009

Monatshefte fuÈr Chemie 132, 1519±1526 (2001)
Concerning the Absorption and Photochemical
Properties of an x-4-Dimethylaminobenzal
Hypericin Derivative
Ã
Roland A. ObermuÈller and Heinz Falk

Institut fur Chemie, Johannes Kepler Universitat, A-4040 Linz, Austria
Summary. A hypericin derivative containing !,! 0-4-dimethylaminobenzal residues was shown to
undergo an intramolecular [2 ‡ 2] cycloaddition upon irradiation leading to a cyclobutane derivative
whose main absorption band is hardly shifted as compared to hypericin. The corresponding
!-substituted derivative displayed a 34 nm bathochromic shift and a strongly reduced ¯uorescence
quantum yield rendering it a nice candidate for a photodynamic therapy agent. Unfortunately, how-
ever, it produced virtually no photosensitized active oxygen species, making it thus unsuited for this
purpose.
Keywords. Photodynamic therapy; Bilirubin photooxidation; Bathochromic shift; Fluorescence;
Intramolecular cycloaddition.
Introduction
Hypericin (1) is a photosensitizing pigment isolated from the weed St. John's
wort (Hypericum perforatum L.). It has been shown to constitute a promising
photodynamic agent for the therapy of certain viral infections and tumors [1].
However, a serious drawback for its use is its main absorption band situated
slightly below 600 nm, which is just outside of the 640±660 nm wavelength of
commercial medicinal lasers. Therefore, efforts have been undertaken to shift the
absorption band of 1 to longer wavelengths, but concomitantly retaining its main
functional out®t. Thus, restricting the derivatization of 1 to its carbon atoms,
derivatives like 2,5-diiodo-hypericin or the enhanced conjugation derivative 2 have
been synthesized [2]. However, upon irradiation with visible light 2 rapidly under-
went an intramolecular [2 ‡ 2] cycloaddition to 3, thereby destroying the 36 nm
bathochromic shift of its long-wavelength absorption band as compared to 1.
To overcome this problem we investigated if substitution of the benzene rings
of 2 with an electron donating auxochromic group would perhaps suppress the
[2 ‡ 2] cycloaddition on the one hand; on the other hand, it seemed of interest if
such a substitution would suf®ciently shift the long-wavelength band if only one
methyl group of the hypericin moiety is substituted with a benzal moiety.
Ã
Corresponding author. E-mail: hfalk@soft.uni-linz.ac.at


R. A. Obermuller and H. Falk
1520
Results and Discussions
Following the well established route to functionalize the methyl group of hypericin
[2, 3] the emodin phosphonium salt 4 was reacted with 4-dimethylamino-benz-
aldehyde to yield the (E )-anthraquinone derivative 5. One-pot deprotection and
reduction with SnCl2/ HBr in acetic acid provided the anthrone derivative 6 in high
yield. Dimerization of the latter in the conventional way [4] gave the proto-
hypericin derivative 7. Upon irradiation of 7, the desired hypericin derivative 8
could only be traced by its bathochromically shifted long-wavelength absorption
band which showed up at 655 nm. However, after being produced it immediately
underwent a light-induced intramolecular [2 ‡ 2] cycloaddition reaction to give the
cyclobutane derivative 9 ± even more ef®ciently than its bis-des-dimethylamino
parent 2 [2]. Accordingly, the electron donating auxochromic group did not prevent
the cycloaddition reaction, and therefore this dimerization strategy was not pursued
any further.
Following this result, it was investigated if only half of the bathochromically
shifting dimethylaminobenzal moieties of 8 would be suf®cient to move the long-
wavelength absorption band of 1 into the medicinal laser region. The synthesis of
such a derivative was achieved by reacting the dimethylamino derivative 6 with
emodin anthrone 10 [4]. The resulting protohypericin derivative 11 together with
the corresponding symmetrical dimers, i.e. protohypericin and 7, were photo-
cyclyzed, and the desired product 12 was eventually separated in rather low yield
from the symmetrical dimers 1 and 8/9.
The constitution of 12 was immediately evident from its spectroscopic prop-
erties (see Experimental). In addition, these data together with appropriate experi-
ments revealed detailed information on its deprotonation and photochemical
behaviour, which are relevant to its potential photodynamic properties. Thus, a
spectrophotometric titration showed that a three-species deprotonation system
prevails within the pH range from 1 to 12. The absorption spectra of the three
species 12, 12 , and 122 characterized by pKa values of 3.4 and 10.7 are
displayed in Fig. 1.

Properties of an !-4-Dimethylaminobenzal Hypericin Derivative
1521
No experiments were available to discriminate between 12 and its zwitterion
12Æ; however, we inferred that the equilibrium would be shifted in favour of
12 because of the basicity of the nitrogen atom of 12 (vinylogous amide). This


R. A. Obermuller and H. Falk
1522
Fig. 1. Absorption spectra of the three species 12, 12 , and 122 in 80% ethanol
could be substantiated by AM1 calculations, which showed 12 to be more stable
1
by 276 kJ Á mol than its corresponding zwitterion 12Æ. These calculations also
‡
revealed that protonation of 12 to yield 12 Á H would be favoured at the carbonyl
1
group in position 14 by about 20 kJ Á mol over the protonation at the nitrogen
atom, stressing again the vinylogous amide substructure of 12. These calculations
were also in accordance with the observation that 12 could not be extracted into
1 M hydrochloric acid for isolation and puri®cation purposes. The assignment of
the phenolate ion 12 was based upon an electrospray mass spectroscopic

Properties of an !-4-Dimethylaminobenzal Hypericin Derivative
1523
experiment in analogy to 1 [5], which showed that at pH 6.1 the [M±H] ion is
the only one that can be detected. Moreover, in DMSO-d6 only one H NMR OH
1
signal at 18.3 ppm was observed, which is characteristic [6] of a hydroxyl proton
in position 3/4 strongly bonded intramolecularly to the 4/3 phenolate ion.
(3
Accordingly, at physiological pH values
)12 will be the predominating species
(AM1 calculations revealed that 3- and 4-deprotonation is of virtually identical
energy). Further deprotonation yields 122 ± based on AM1 calculations the
(3 ,6
second deprotonation site 6 (yielding
)12) was found to be more stable than
1
the most stable of the other possible species by about 30 kJ Á mol . Comparison of
the absorption spectra of 12 in several solvents pointed also to a prevalence of the
)12 ion in such solutions. Thus, the bathochromic shift of more than 30 nm to
(3
about 630 nm compared to hypericin (1) together with a moderately high extinction
coef®cient would make 12 a proper candidate to match the medicinal laser
wavelength. It should also be mentioned that the ¯uorescence spectra of the three
species were quite different from each other. Thus, in 80% ethanol and at pH ˆ 0.6
it displayed an emission band at 621 nm (FF ˆ 0.07), at pH ˆ 6.9 at 644 nm
(FF ˆ 0.03), and at pH > 12 at 718 nm (FF ˆ 0.02).
The tautomeric situation of 12 was investigated in some detail by means of
AM1 calculations. Accordingly, the 7,14-dioxo tautomer proved to be much more
stable than all other conceivable tautomers, with a difference in their enthalpies of
1
formation of 79 kJ Á mol between the 7,14-dioxo tautomer and the next one in
the stability series, the 1,6-dioxo tautomer. In addition, in analogy to the results
obtained for 1 and (3 )1 [8] it could also be con®rmed that for (3 )12 the 7,14-dioxo
tautomer shown in the formulae predominates.
(3
With respect to the photophysical and photochemical behaviour of
)12,
which is of outmost importance for its intended use, its ¯uorescence spectrum
displayed an emission which was Stokes-shifted by nearly 20 nm. This is quite
(3
large compared with the only 5 nm of
)1 [9]. Together with the rather small
¯uorescence quantum yield FF of only about one tenth of that of hypericin ((3 )1)
[9] this pointed to excited state processes different from those of (3 )1. At this point
the use of 12 as a photodynamic agent seemed even more promising since a low
¯uorescence quantum yield could point to an enhanced intersystem crossing, which
would be necessary to sensitize active oxygen species. However, hypericin sen-
sitized photooxidation of bilirubin, which has been shown to provide a sensitive
measure of the generation of sensitized active oxygen species ± in peculiar of
singlet oxygen and superoxide radicals [10] ± revealed that (3 )12 did not produce
the sensitized singlet oxygen and superoxide species thought to be the most active
agents in hypericin photodynamic therapy [1] (Fig. 2). Accordingly, a non-radiative
channel other than intersystem crossing (or a suppressed capacity of the triplet
state to sensitize singlet oxygen) is operating in 12. Perhaps an intramolecular
quenching process similar to the intermolecular quenching of the triplet state of 1
by amines [11] might be operative.
In conclusion, 12 as a monomethyl conjugated and amine functionalized
molecule showed a promising bathochromic long-wavelength shift together with a
reduced ¯uorescence yield. However, the auxochromic dimethylamino substituent
of the benzal residue led to a dramatic change in the photochemical behaviour
of 12 and, in particular, its phenolate ion as compared to 1, making it virtually


R. A. Obermuller and H. Falk
1524
Fig. 2. Hypericin derivative sensitized photooxidation of bilirubin: normalized absorption changes
(A/A0) with time of solutions of disodium bilirubinate together with (3 )1 or (3 )12 in aereated 80%
ethanol upon irradiation at l > 570 nm
non-reacting with respect to the generation of active oxygen species and thus
unsuited for its use as a photodynamic therapy agent. Nevertheless, the behaviour
of this compound added a valuable facet to the chemistry and photochemistry of
hypericin.
Experimental
1
Solvents were of p.a. quality. H and 13C NMR, UV/ Vis, ¯uorescence, and mass spectra were
recorded using Bruker DRX 500 and DPX 200, Hewlett Packard 8453 UV/ Vis, Hitachi 4010F, and
Hewlett Packard 59987 quadrupole instruments. HSQC NMR experiments for signal assignments
were carried out using standard parameters. The ¯uorescence quantum yield was determined as
described previously [9]. The emodin phosphonium salt 4 was prepared according to Ref. [3].
Emodin anthrone 10 was synthesized in the usual way [6]. Hypericin sensitized photooxidation of
bilirubinate was executed as described in Ref. [10]. AM1 calculations [12] were performed at the
SGi Origin 2000 of the LIZENS using the MOPAC package [13, 14].
(E)-6-(2-(4-Dimethylaminophenyl)-ethenyl)-1,3,8-trimethoxy-anthraquinone (5; C27H25NO5)
A mixture of 5 g 60% 4 (4.6 mmol), 2.17 g dry and pulverized K2CO3 (15.73 mmol), 1.45 g 18-
crown-6 (5.5 mmol), and 293 cm3 dry CH2Cl2 was re¯uxed for 15 min. To this dark blue ylide
solution, a solution of 11.47 g commercial (Fluka) 4-dimethylamino-benzaldehyde (76.9 mmol) in
293 cm3 dry CH2Cl2 was added in three portions, re¯uxing the mixture between the additions for
40 min. After re¯uxing for additional 30 min the cooled reaction mixture was diluted with 400 cm3
CH2Cl2, ®ltered, and extracted with brine. The brine phase was extracted with 3 Â 100 cm3 CH2Cl2,
and the organic phase was dried over Na2SO4, ®ltered, and evaporated. The residue was chromato-
graphed on silica using CHCl3:EtOAc ˆ 1:1 as the eluent.
Yield: 0.86 g (42%); m.p.: 237±240ꢀC; 1H NMR (DMSO-d6, ꢀ, 200 MHz): 7.80 7.30 (m,
2HPh ‡ 4Har ‡ HCꢁCH), 7.15 (m, Har-4), 6.19 (m, Har-2), 6.74 (d, J ˆ 8.4 Hz, XX0 part of an AA0XX0
system, HPh-3,5), 3.95 (s, OCH3), 3.89 (s, OCH3), 2.96 (s, N(CH3)2) ppm; 13C NMR (DMSO-d6, ꢀ,
50 MHz): 183.4 (CˆO), 179.6 (CˆO), 163.3 (Car±OCH3), 161.0 (Car±OCH3), 159.4 (Car±OCH3),
143.5, 135.6, 133.7, 131.7, 131.4, 128.9, 128.6, 128.3, 121.9, 121.2, 117.7, 115.7, 114.9, 105.2
(13Car ‡ CˆC), 56.32 (OCH3), 56.26 (OCH3), 55.9 (OCH3) ppm (the N(CH3)2 signal was masked

Properties of an !-4-Dimethylaminobenzal Hypericin Derivative
1525
by the solvent peak); UV/ Vis (DMSO; c ˆ 1.10 5 mol Á dm 3): lmax(") ˆ 479 (12630), 368 (24570),
275 (27070) nm.
(E)-9,10-Dihydro-6-(2-(4-dimethylaminophenyl)-ethenyl)-1,3,8-trihydroxy-anthracene-9-one
(6; C24H21NO4)
To 1.57 g 5 (3.57 mmol) dissolved in 100 cm3 glacial acetic acid, 7.0 g SnCl2 Á 2H2O (31 mmol) and
20 cm3 47% HBr were added, and the reaction mixture was re¯uxed for 2 h. Upon cooling to 15ꢀC
for 2 h the product was ®ltered, washed with a few drops of glacial acetic acid, and dried in high
vacuum.
1
Yield: 1.33 g (97%); m.p.: > 320ꢀC; H NMR (DMSO-d6, ꢀ, 200 MHz): 12.49 (s, OH-1), 12.37
(s, OH-8), 10.79 (s, OH-3), 7.52 (d, J ˆ 8.4 Hz, AA0 part of an AA0XX0 system, HPh-2,6), 7.39 (d,
J ˆ 16.3 Hz, HCˆC), 7.23 (d, J ˆ 16.3 Hz, CˆCH), 7.14 (s, Har-5), 7.00 (s, Har-7), 6.83 (d, J ˆ 8.4
Hz, XX0 part of an AA0XX0 system, HPh-3,5), 6.44 (s, Har-4), 6.24 (m, Har-2), 4.36 (s, CH2), 2.99 (s,
N(CH3)2) ppm; 13C NMR (DMSO-d6, ꢀ, 50 MHz): 190.5 (CˆO), 164.9 (Car±OH), 164.5 (Car±OH),
162.0 (Car±OH), 145.4, 144.9, 142.3, 133.1, 128.4, 125.4, 122.6, 122.5, 116.6, 113.5, 113.0, 11.3,
108.5, 107.4, 101.0 (13Car ‡ CˆC), 32.4 (CH2) ppm (the N(CH3)2 signal was masked by the solvent
peak); UV/ Vis (DMSO, c ˆ 1.10 5 mol Á dm 3): lmax(") ˆ 443 (21500), 298 (8660), 258 (11670) nm.
Dimerization of 6
This reaction sequence was executed 0.1 mmolar in analogy to Ref. [2], but in a qualitative way
using UV/ Vis spectroscopy as the monitor and without isolation and further characterization of
intermediates and product. Thus, the protohypericin derivative was characterized by absorption
bands at lmax(DMSO) ˆ 589 (33), 386 (65), and 269 (100) nm (in parentheses: relative intensities).
The desired product 8 exhibited a long-wavelength band at 655 nm, whereas the ®nal product 9
displayed a hypericin-like UV/ Vis spectrum similar to its parent compound 3 [2] with a long-
wavelength band at 589 nm.
(E)-10-(2-(4-Dimethylaminophenyl)-ethenyl)-1,3,4,6,8,13-hexahydroxy-11-methyl-phenanthro
[1,10,9,8-opqra]perylene-7,14-dione (12; C39H25NO8)
To a solution of 113 mg 6 (0.29 mmol) and 450 mg 10 (1.76 mmol) in 10.5 cm3 absolute pyridine and
1.5 cm3 absolute piperidine, 1.05 g pyridine-N-oxide (11.04 mmol) and 28.5 mg FeSO4 Á 7H2O ( p.a.;
0.10 mmol) were added, and the reaction mixture was stirred for 1 h at 100±110ꢀC under Ar and
exclusion of light. After cooling to room temperature, the violet coloured reaction mixture was
poured into 73 cm3 of 2 M HCl. After standing for 30 min it was centrifuged, and the residue was
washed twice with 3% HCl, three times with H2O, and dried over silica. The residue (500 mg)
consisting of a mixture of the protohypericines including 11 was dissolved in 4 dm3 acetone and
irradiated with a 700 W Hg high pressure lamp for 1.5 h under admission of air. The resulting wine
red solution was evaporated, washed with 20 cm3 MeOH, and dried over silica. The residue con-
taining hypericin (1) and its derivatives 8/9 and 12 was chromatographed on a silica plate (20 Â
20  0.2 cm) using THF:glacial acetic acid ˆ 10:1. The band containing 12 was triturated three times
with 50 cm3 THF:methanol ˆ 1:1 and ®ltered. After evaporation, the residue was chromato-
graphed twice with MeOH on a Sephadex LH 20 column.
Yield: 8 mg (5%); 1H NMR (DMSO-d6, ꢀ, 500 MHz): 18.33 (s, OH-3), 14.73 (br s, OH-1,6,8,13),
7.63 (d, J ˆ 16.3 Hz, HCˆC), 7.43 (d, J ˆ 8.6 Hz, AA0 part of an AA0XX0 system, HPh-2,6), 7.41 (s,
Har-9,12), 7.03 (d, J ˆ 16.3 Hz, CˆCH), 6.78 (d, J ˆ 8.6 Hz, XX' part of an AA0XX0 system, HPh-3,5),
6.58 (s, Har-2,5), 2.97 (s, N(CH3)2), 2.74 (s, CH3) ppm; 13C NMR (DMSO-d6, ꢀ, 125 MHz): 183.4,
182.8 (2 CˆO), 174.6, 174.3 (Car±O-3,4), 168.14, 168.13 (Car±O-1,6), 161.8, 150.6 (Car±O-8,13),

1526
Properties of an !-4-Dimethylaminobenzal Hypericin Derivative
144.1, 142.9 (Car-10,11), 142.7 (CPh-4), 138.0 (Cˆ ), 132.1 (CPh-2,6), 128.2 (CPh-3,5), 127.1, 127.0
(Car-3a,3b), 126.7, 126.2 (Car-6b,14b), 124.5 (ˆC), 124.2 (CPh-1), 121.6, 120.3 (Car-7c,14c), 121.0,
120.0 (Car-10a,10b), 119.5, 119.4 (Car-7b,13b), 112.3, 112.2 (Car-9,12), 108.8, 108,6 (Car-6a,14a),
105.64, 105.63 (Car-2,5), 102.3, 102.1 (Car-7a,13a), 40.5 (N(CH3)3), 24.4 (CH3) ppm (signal
assignment by 1H /13C HSQC experiments); UV/ Vis (DMSO, c ˆ 1.10 5 mol Á dm 3): lmax(") ˆ 633
(26800), 625 (24600), 603 (22770), 464 (16660), 392 (19570) 338 (28700), 260 (41230) nm; UV/ Vis
(80% EtOH, c ˆ 1.10 5 mol Á dm 3): lmax(") ˆ 623 (26700), 597 (23720), 463 (17210), 457 (17240),
384 (19570) nm; UV/ Vis (THF): lmax(rel. int.) ˆ 630 (76), 601 (45), 582 (45), 464 (29), 382 (40), 337
(63), 289 (68), 243 (100) nm; UV/ Vis (acetone): lmax(rel. int.) ˆ 628 (64), 581 (41), 464 (27), 383
(37), 336 (59), 210 (100) nm; UV/ Vis (CH3CN): lmax(rel. int.) ˆ 623 (34), 581 (22), 459 (15), 383
(20), 334 (32), 273 (57), 252 (59), 200 (100) nm; UV/ Vis (MeOH): lmax(rel. int.) ˆ 618 (40), 593 (30),
581 (26), 458 (19), 383 (22), 331 (39), 284 (42), 203 (100) nm; UV/ Vis (H2O): lmax ˆ 640 (sh),
600 nm; ¯uorescence (DMSO, c ˆ 1 Á 10 7 mol Á dm 3, lex ˆ 550 nm): lem(rel. int.) ˆ 653 (45), 603
(100) nm, FF ꢂ 0.03; ¯uorescence (80% EtOH, c ˆ 1 Á 10 7 mol Á dm 3, pH ˆ 0.6, lex ˆ 550 nm):
l
em(rel. int.) ˆ 621 (100), 597 (70; sh) nm, FF ꢂ 0.07; ¯uorescence (80% EtOH, c ˆ
1 Á 10 7 mol Á dm 3, pH ˆ 6.9, lex ˆ 550 nm): lem(rel. int.) ˆ 641 (34), 596 (100) nm, FF ꢂ 0.03;
¯uorescence (80% EtOH, c ˆ 1 Á 10 7 mol Á dm 3, pH > 12, lex ˆ 550 nm): lem(rel. int.) ˆ 718
(100) nm, FF ꢂ 0.02; electrospray MS (i-propanol/NH4OAc buffer, pH ˆ 6.1): m/z ˆ 634 [M±H] .
Acknowledgements
We gratefully acknowledge the assistance of DI K. Hohenthanner in recording the 2D NMR spec-
tra and of Dr. C. Klamp¯ in recording the mass spectra. Thanks are due to DI M. Sonnleitner and
Dr. G. J. Schutz for technical assistance.
References
[1] For a recent overview see: Falk H (1999) Angew Chem 111: 3306; Angew Chem Int Ed 38: 3116

[2] Obermuller RA, Hohenthanner K, Falk H (2001) Photochem Photobiol 74: 211
[3] Falk H, Tran TNH (1996) Monatsh Chem 127: 717
[4] Falk H, Meyer J, Oberreiter M (1993) Monatsh Chem 124: 339
[5] Ahrer W, Falk H, Tran TNH (1998) Monatsh Chem 129: 634
[6] Falk H, Schmitzberger W (1992) Monatsh Chem 123: 731
[7] Burel L, Jardon P (1996) J Chim Phys 93: 300
[8] Dax TG, Falk H, Kapinus EI (1999) Monatsh Chem 130: 827
[9] Falk H, Meyer J (1994) Monatsh Chem 125: 753
[10] Hagenbuchner K, Falk H (1999) Monatsh Chem 130: 1075
[11] Darmanyan AP, Jenks WS, Eloy D, Jardon P (1999) J Phys Chem 103: 3323
[12] Dewar MJS, Zoebisch EG., Healy EF, Stewart JJP (1985) J Am Chem Soc 107: 3902; Dewar
MJS, Dieter KM (1986) J Am Chem Soc 108: 8075; Stewart JJP (1990) J Comp Aided Mol
Design 4: 1
[13] MOPAC 6.0 DEC-3100 Ed 1990, FJ Seiler Res Lab, USAF Acad CO80840
[14] For AM1 calculations on the geometrical and energetic realms of phenanthroperylene quinones
see: Gutman I, Markovic Z, Solujic S, Sukdolak S (1998) Monatsh Chem 129: 481; Etzlstorfer C,
Falk H (1998) Monatsh Chem 129: 855; Gutman I, Markovic Z (1998) Monatsh Chem 129: 1019;
Etzlstorfer C, Gutman I, Falk H (1999) Monatsh Chem 130: 1333
Received July 11, 2001. Accepted July 18, 2001