A carbohydrate-linked hypericinic photosensitizing agent

Article abstract of DOI:10.1007/s00706-008-0928-y

With respect to an enhanced solubility under physiological conditions, a carbohydrate-containing hypericin-based second-generation photosensitizer was prepared. Its photochemical properties were tested by means of the light-sensitized destruction of bilir

Full text of DOI:10.1007/s00706-008-0928-y

Monatsh Chem 139, 1387–1390 (2008)
DOI 10.1007/s00706-008-0928-y
Printed in The Netherlands
A carbohydrate-linked hypericinic photosensitizing agent
¨
Joachim Zuschrader, Wolfgang Schofberger, Heinz Falk
Institute of Organic Chemistry, Johannes Kepler University, Linz, Austria
Received 20 February 2008; Accepted 13 March 2008; Published online 2 June 2008
# Springer-Verlag 2008
Abstract With respect to an enhanced solubility under
physiological conditions, a carbohydrate-containing
hypericin-based second-generation photosensitizer
was prepared. Its photochemical properties were test-
ed by means of the light-sensitized destruction of bil-
irubin IXꢀ to be even better than those of the parent
compound hypericin. Investigations on binding-inter-
actions with DNA showed promising results as well.
interactions with DNA combined with an enhanced
solubility under aqueous conditions.
Keywords Hypericin; Tetraacetyl-D-glucosamine; Calf-
thymus DNA; Photodynamic agent; Reactive oxygen species.
Introduction
The compatibility of photosensitizing agents with
respect to physiological issues concerning their ap-
plication in the photodynamic therapy of cancer is an
important facet for developing second-generation
derivatives of hypericin (1) [1]. Having in mind the
demands of a ‘‘perfect’’ photosensitizer (solubility,
bathochromically shifted absorption maxima, and
high quantum yields of photosensitizing singlet oxy-
gen and=or reactive oxygen species), we focused on
an enhanced solubility in protic solvents, in particu-
lar in water. Although hypericin (1) itself and other
similar phenanthroperylenequinone structures as
well are known for their interaction behavior with
cell ingredients [2], no representive derivatives of
hypericin have been reported so far aiming at an
increased cell toxicity based on specific binding-
Results and discussion
With respect to the glycosidic occurrence of the
‘‘monomeric’’ precursor emodin in the bark of the
breaking buckthorn (Cortex frangulae) we consid-
ered this as a lead in targeting of second-generation
hypericinic photosensitizers [1]. However, unlike to
the glycosidic natural compounds, which are prone
to hydrolysis under physiological conditions, we de-
cided not to substitute anomeric or other hydroxylic
positions of the sugar moiety to guarantee hydrolytic
stability. Otherwise, phenolic positions of the hyper-
icin chromophore would also represent a rather unfor-
tunate choice because investigations have shown the
need of intact phenolic functionalities for the photo-
chemical properties of interest, i.e. the sensitization of
reactive oxygen species [1]. Therefore, linker chem-
istry on the !,!0-methyl positions of the chromophore
of 1 seemed to be the more desirable method.
Correspondence: Heinz Falk, Institute of Organic Chemistry,
Johannes Kepler University, Altenbergerstraße 69, 4040 Linz,
Austria, Europe. E-mail: heinz.falk@jku.at

1388
J. Zuschrader et al.
Scheme 1
Synthesis
Properties
Starting from emodin aldehyde (Scheme 1) via a
Wittig-strategy and reduction to the anthrone with
subsequent hydrogenation of the double bond, and
finally dimerization according to Ref. [3], the pro-
pionic acid double-elongated intermediate 2 intend-
ed to serve as the starting point for the hypericin
moiety was obtained. For the next step, we envisaged
a N,N0-dicyclohexylcarbodiimide (DCCDI) mediated
formal substitution of the acid 2 by D-glucosamine
using triethylamine for deprotonation and tetrahy-
drofuran as the solvent. However, these attempts
resulted in the formation of the corresponding
O-acylurea derivative, which then underwent a 1,3-
rearrangement to the N-acylurea derivative as de-
scribed recently [4]. This might be explained by the
decreased reactivity of the chain-elongated hypericin
system on the one side, and the low solubility of the
carbohydrate in the organic solvent on the other side.
From a series of such preliminary experiments it
followed that we had to use the DCCDI-activated
acid 2 together with the acetyl-protected sugar de-
rivative tetraacetyl-ꢁ-D-glucosamine 3 (prepared in
analogy to Ref. [5]) to achieve the linking. Under
these conditions the hypericin derivative substituted
at positions !,!0 with the acetyl-protected glucos-
amine moiety 4 was obtained in a reasonable yield
of 66%. Its structure was proven using 1H NMR and
mass spectroscopy. Experiments to hydrolyze the
acetyl groups of 4 using classical methods failed,
but however, it turned out that the solubility of 4
in aqueous systems was high enough for the envis-
aged purpose.
Beside the demands on a second-generation photo-
sensitizer, namely an increased solubility under phy-
siological conditions and a shift of the absorption
maximum in the regions of medical lasers, the possi-
bility to create singlet oxygen and=or reactive oxygen
species was investigated in this context. The second-
generation derivative 4 proved to be highly effective in
the hypericin-sensitized destruction of bilirubin IXꢀ
measured according to Ref. [6] (Fig. 1). In compari-
son to the naturally occurring 1, the glyco-derivative
Fig. 1 Hypericin derivative sensitized photooxidation of
bilirubin IXꢀ: normalized absorption (A=A0) vs. time curves
of solutions of disodium bilirubinate IXꢀ together with the
sodium salts of hypericin (1, — ) and glyco-derivative 4 (– – –)

A carbohydrate-linked hypericinic photosensitizing agent
1389
4 displayed an even better activity to generate singlet
oxygen and=or reactive oxygen species.
generation hypericin photosensitizer and thus, a
potential candidate for photodynamic therapy con-
cerning its interaction or intercalation with polynu-
cleotides, in particular with DNA.
In addition to its high sensitizing activity, we ex-
pected 4 to show interactions with cell ingredients, in
particular DNA. We prepared a 10ꢂM solution of 4 in
water (10% dimethyl sulfoxide) whereby we obtained
its characteristic absorption spectrum. This absorp-
tion spectrum of 4 dissolved in the aqueous solvent
system was indicative of strong intermolecular homo-
interactions of the pigment (cf. the spectroscopic con-
sequences of homoassociation of hypericin (1) in
form of stacked H-aggregates [7]). Then this solution
was titrated with increasing molar ratios of hypericin
derivative and calf-thymus DNA leaving the concen-
tration of 4 constant. Higher molar concentrations up
to 400 ꢂM of calf-thymus DNA were observed to go
hand in hand with a diminished stacking of the phe-
nanthroperylene quinone skeleton of 4 due to the
fact that more and more molecules of 4 became het-
eroassociated with the double helix, whereby an in-
creased slightly hypsochromically shifted absorption
maximum with a significant disproportionality of the
two vibrational bands typical of monomeric hypericin
chromophores was obtained (cf. Fig. 2 for starting
and end point of this titration). So we concluded that
the effect of lessened stacking is dominant in com-
parison to the expected decrease of absorption as a
result of reduced concentration of non-bonded photo-
sensitizer molecules. It should be mentioned, that a
similar behavior of the absorption spectra has been
observed for the binding of 1 to serum albumine [7].
In conclusion, we present the highly water-soluble
and most efficiently photosensitizing 4 as a second-
Experimental
The characterizations of the products were established by
means of m.p. (Kofler microscope, Reichert), NMR (Bruker
DPX 200 MHz or Bruker 500 MHz spectrometer using a TXI
cryoprobe with z-gradient coil using standard pulse sequences
as provided by the manufacturer. Typical 90ꢀ hard pulse dura-
tions were 9.6 ꢂs (1H) and 18.4ꢂs (1H), for a 180ꢀ pulse, 90ꢀ
pulses in decoupling experiments were set to 80ꢂs. HSQC
and HMBC experiments were optimized for decoupling con-
stants of 145 Hz for single quantum correlations and 7 Hz for
multi-bond correlations. The NOESY mixing time was set to
400 ms), IR (Bruker Tensor 27, KBr), MS (Thermofinnigan
LCQ Deca XP Plus), fluorescence (Varian Carey Eclipse fluo-
rescence instrument), and UV-Vis data (Varian Cary 100 Bio).
Fluorescence quantum yields were calculated according to the
comparative method of Williams et al. [8] using hypericin (1) as
standard sample. The production of singlet oxygen=oxidizing
species by 1 and 4 was monitored by bilirubin-IXꢀ-degradation
according to Ref. [6]. The long-wavelength absorption band
intensities of the two compounds were made equal by ad-
justing concentrations to provide comparable light absorp-
tions for the two compounds. N,N0-Dicyclohexylcarbodiimide
was obtained from Merck and used as obtained. Compound 2
was prepared according to Ref. [3]. The per-acetylated D-guco-
samine hydrochloride was synthesized in analogy to Ref. [5]
using benzaldehyde as the imide-protecting reagent during the
acetylation process followed by the subsequent hydrolysis to
the tetra-O-acetyl-protected ꢁ-D-glucosamine hydrochloride 3.
All experiments involving calf-thymus DNA (Sigma) were
performed with MilliQ (18 Mꢀ) water. According to Refs. [9,
10], calf-thymus DNA solutions were prepared by dispersing
the desired amount of DNA in water (MilliQ 18Mꢀ) with
stirring over night at temperatures below 4ꢀC. The con-
centration was expressed as the concentration of nucleotides
and was calculated by using an average molecular weight
of 338 for a nucleotide and an extinction coefficient of
6600Mꢁ1 cmꢁ1 at 260 nm.
N-Benzylidene-ꢁ-D-glucosamine (C13H17NO5)
To a stirred solution of 2.0g ꢁ-D-glucosamine hydrochloride
(9.28mmol) and 1.0 g (11.90 mmol) NaHCO3 in 12 cm3 distilled
water 1.1 cm3 benzaldehyde, diluted with 10 cm3 methanol,
were added. The mixture was kept at room temperature for 3 h.
After evaporation of the solvent, the crude product was sus-
pended in CH2Cl2, washed with a small amount of water and
50 cm3 cold diethyl ether. The Schiff base was obtained in a
40% yield. ESI-MS (positive ion mode): m=z ¼ 268 ([M þ H]þ).
Fig. 2 UV-Vis spectra showing the interaction behavior of 4
with calf thymus DNA: a) 4 (10 ꢂM) in aqueous solution
containing 10% dimethyl sulfoxide (–); b) 4 (10 ꢂM) in
aqueous solution containing 10% dimethyl sulfoxide and calf
thymus DNA (400 ꢂM) (. . .)
N-Benzylidene-O1,O3,O4,O6-tetraacetyl-ꢁ-D-glucosamine
(C21H25NO9)
After dissolving 1.0 g of the above Schiff base (3.74 mmol) in
5 cm3 pyridine, the mixture was cooled to 0ꢀC and 4 cm3 acetic

1390
J. Zuschrader et al.
anhydride (40mmol) were added. The solution was stirred for
5 h and for further 10h at room temperature, then poured onto
ice=water, and filtered to yield 1.13 g (69%) of the product as a
(CHCl3:MeOH¼ 3:1); 1H NMR (500 MHz, DMSO-d6, 30ꢀC):
ꢃ ¼ 18.48 (s, 3-OH, 4-OH), 14.72 (s, 1-OH, 6-OH), 14.08 (s, 8-
OH, 13-OH), 7.79 (s, –CONH–), 7.77 (s, –CONH–), 7.45 (s,
ar-H9, ar-H12), 6.58 (s, ar-H2, ar-H5), 5.48 (m, H20), 3.54 (m,
ar-CH2–CH2–), 2.26 (m, ar–CH2–CH2–), 2.05 (s, 10-OAc
or 30-OAc) ppm; 13C NMR data were partially derived by
2D-experiments; a total characterization of the sugar moiety
was not possible; 13C NMR (125MHz, DMSO-d6, 30ꢀC): ꢃ ¼
183.8 (C14, C15), 174.1 (C3, C4), 169.1 (–CONH–), 168.5
(C1, C6), 162.3 (C8, C13), 119.7 (C3b, C4b), 119.0 (C11b,
C10b), 117.6 (C9, C12), 108.2 (C6a, C14a), 104.9 (C2a, C5a),
101.9 (C3a, C4a), 51.7 (ar–CH2–CH2–), 49.6 (C20), 36.4 (1-
COCH3 or 3-COCH3), 31.6 (ar–CH2–CH2–) ppm; NOESY
(DMSO-d6): ar-H9 and ar-H12 $ ar-CH2–CH2–, ar–CH2–
CH2–, and 1-COCH3 or 3-COCH3, –CONH– $ H20 and ar–
CH2–CH2–, ar–CH2–CH2– $ ar–CH2–CH2–, ar-H9, ar-H12,
and 1-COCH3 or 3-COCH3, (–CONH–) $ (m, ar–CH2–
CH2–); HMBC (DMSO-d6): ar–CH2–CH2– $ C7, C14, C8,
C13, C11b, C10b, C6a, C14a, and ar–CH2–CH2–, ar-H2,
ar-H5 $ C3, C4, C1, C6, C3b, C4b, C2a, C5a, and C3a,
C4a, –CONH– $ C20 and ar–CH2–CH2–; HSQC data were
according to structure; IR (KBr): ꢄꢁ¼ 3420, 2927, 2854,
1718, 1559, 1465, 1244, 1114, 1035, 847, 669 cmꢁ1; UV-
Vis (80% EtOH, c ¼ 2.0 ꢂ 10ꢁ6 mol ꢂ dmꢁ3): lmax (") ¼ 555
(10000), 600 (60000) nm (dm3 ꢂ molꢁ1 ꢂ cmꢁ1); UV-Vis
(DMSO, c ¼ 2.0 ꢂ 10ꢁ5 mol ꢂ dmꢁ3): lmax (") ¼ 555 (10850),
600 (2600) nm (dm3 ꢂ molꢁ1 ꢂ cmꢁ1); fluorescence (80% EtOH,
c ¼ 1.0 ꢂ 10ꢁ7 mol ꢂ dmꢁ3, lex ¼ 550 nm): lem (rel. int.) ¼ 597
(100), 644 (10) nm; F ¼ 0.17; fluorescence (DMSO, c ¼
1.0 ꢂ 10ꢁ7 mol ꢂ dmꢁ3, lex ¼ 550 nm):lem (rel. int.) ¼ 604 (100),
656 (3) nm, F ¼ 0.11.
1
white solid. H NMR (500 MHz, DMSO-d6, 30ꢀC): ꢃ ¼ 8.39
(s, –N¼CH–), 7.71 (m, ar-H20 and ar-H60), 7.47 (m, ar-H30,
ar-H40, and ar-H50), 6.10 (d, J ¼ 7 Hz, 1-CH), 5.48 (t, J ¼ 10 Hz,
3-CH), 4.99 (t, J ¼ 10Hz, 5-CH), 4.25 (m, 6-CH2), 4.03 (d,
J ¼ 12Hz, 4-CH), 3.52 (t, J ¼ 10Hz, 2-CH), 2.03 (s, 6-OAc),
1.99 (s, 1-OAc and 4-OAc), 1.83 (s, 3-OAc) ppm; 13C NMR
(125MHz, DMSO-d6, 30ꢀC): ꢃ ¼ 170.6 (6-CH3C¼O), 170.0
(1-CH3C¼O or 4-CH3C¼O), 169.5 (4-CH3C¼O or 1-
CH3C¼O), 169.1 (3-CH3C¼O), 165.9 (–N¼CH–), 136.0
(C10), 132.0 (C30, C40, or C50), 129.3 (C30, C40, or C50), 128.8
(C30, C40, or C50), 93.0 (C1), 72.8 (C3), 72.1 (C2), 68.4 (C5),
62.2 (C6), 21.1 (6-COCH3), 21.0 (4-COCH3 or 1-COCH3), 20.8
(3-COCH3) ppm; HMBC (DMSO-d6): –N¼CH– $ –N¼CH–,
C30, C20, C60, C10, and C2, 1-CH– $ C1 and C2, 2-CH– $
C1, C3, and –N¼CH–, 6-COCH3 $ 6-COCH3, and C6, 3-
COCH3 $ 3-COCH3 and C3; HSQC data were according to
structure; ESI-MS (positive ion mode): m=z ¼ 436 ([Mþ H]þ).
O1,O3,O4,O6,-Tetraacetyl-ꢁ-D-glucosamine hydrochloride
(3, C14H22ClNO9)
To a refluxing solution of 500 mg (1.87 mmol) of the above
product in 2.5 cm3 acetone 265 mm3 conc. HCl were added.
The product precipitated spontaneously whereupon the mix-
ture was cooled to 0ꢀC, filtered, and washed with diethyl ether
yielding 73%. 1H NMR (500 MHz, DMSO-d6, 30ꢀC): ꢃ ¼ 8.69
(s, –NH2), 5.90 (d, J ¼ 9 Hz, 1-CH), 5.35 (t, J ¼ 10 Hz, 3-CH),
4.94 (t, J ¼ 10 Hz, 4-CH), 4.19 (m, 6-CH2), 4.07 (m, 5-CH),
4.01 (m, 6-CH2), 3.58 (t, J ¼ 10Hz, 2-CH), 2.17 (s, 1-OAc),
2.04 (s, 1-OAc), 2.00 (s, 6-OAc), 1.98 (s, 4-OAc) ppm; 13C
NMR (125 MHz, DMSO-d6, 30ꢀC): ꢃ ¼ 171.0 (3-CH3C¼O),
170.8 (4-CH3C¼O and 6-CH3C¼O), 170.3 (1-CH3C¼O),
91.5 (C1), 73.7 (C5), 71.8 (C3), 69.3 (C4), 63.3 (C6), 53.9
(C2), 22.3 (1-COCH3 and 3-COCH3), 21.8 (4-COCH3 and 6-
COCH3) ppm; HMBC (DMSO-d6): C1 $ 2-CH, 1-CH, and
1-COCH3, C2 $ 3-CH and 1-CH, C3 $ 1-CH, 4-CH, and
3-COCH3, C6 $ 6-CH2 and 5-CH, C4 $ 5-CH and 3-CH;
HSQC data were according to structure; ESI-MS (positive
ion mode): m=z ¼ 385 ([M þ H]þ).
Acknowledgements
W.S. acknowledges the support through a Standalone Project
P18384-B10 funded by the Austrian Science Fund (FWF).
References
1. For reviews see: a) Falk H (1999) Angew Chem
111:3306, Angew Chem Int Ed Engl 38:3116; b) Waser
M, Falk H (2007) Curr Org Chem 11:547
2. Zeng Z, Qiao R, Zhou J, Xia A, Zhang Y, Liu Y, Chen J,
Wang X, Zhang B (2007) J Phys Chem B 11:3742
3. Waser M, Popova Y, Etzlstorfer C, Huber W, Falk H
(2005) Monatsh Chem 136:1221
10,11-Bis(N-(D-tetrahydro-2,4,5-triacetoxy-6-(acetoxymethyl)-
2H-pyran-3-yl)carbamoylethyl)-1,3,4,6,8,13-hexahydroxy-
phenanthro[1,10,9,8-opqra]perylene-7,14-dione
(4, C62H58N2O28)
4. Zuschrader J, Reiter G, Falk H (2008) Monatsh Chem,
doi: 10.1007=s00706-008-0877-5
5. Appeldoorn CCM, Joosten JAF, Maate F, Dobrindt U,
Hacker J, Liskamp RMJ, Khan AS, Pieters RJ (2005)
Tetrahedron Asymm 16:361
6. Hagenbuchner K, Falk H (1999) Monatsh Chem 130:1075
7. Falk H, Meyer J (1994) Monatsh Chem 125:753
8. Williams ATR, Winfield SA, Miller JN (1983) Analyst
108:1067
9. Zeng Z, Qiao R, Zhou J, Xia S, Zhang Y, Liu Y, Chen J,
Wang X, Zhang B (2007) J Phys Chem B 111:3742
10. Pasternack RF, Brigandi RA, Abrams MJ, Williams AP,
Gibbs EJ (1990) Inorg Chem 29:4483
An argon-flushed mixture of 42mg 2 (0.12 mmol) and
750 mm3 triethyl-amine in 50cm3 dry THF was heated to
40ꢀC under stirring. After the addition of 40 mg 3 (0.06 mmol),
the suspension was stirred for further 30 min. An amount of
26mg N,N0-dicyclohexylcarbodiimide (0.12 mmol) was dis-
solved in 5 cm3 dry THF. The reaction mixture and the acti-
vating diimide solution were combined and stirred for 23h
under Ar. Upon subsequent evaporation of the solvent under
reduced pressure and extraction with ethyl acetate=water a
black solid was obtained. Purification by means of column
chromatography with chloroform=methanol¼ 3=1 yielded
51mg (66%) of the product. Mp >350ꢀC; TLC: Rf ¼ 0.56