Anthralin: Primary products of its redox reactions

Article abstract of DOI:10.1021/jo060622o

One-electron reduction significantly enhances the ability of anthralin, 1, to act as a hydrogen atom donor. On annealing of an MTHF glass in which the radical anion of anthralin, 1^.-, is generated radiolytically, this species decays mainly by loss of H^. to give the anthralyl anion, 2. On the other hand, radicals formed on radiolysis of matrices that are suitable for the generation of radical anions or cations are capable to abstract H ^. from anthralin to give the anthralyl radical, 2^.. Both 2^- and 2^. are obtained simultaneously by mesolytic cleavage of the radical anion of the anthralin dimer. Contrary to general assumptions, the anthralyl radical is found to be much more reactive toward oxygen than the anion. All intermediates are characterized spectroscopically and by reference to quantum chemical calculations. Attempts to generate the radical cation of anthralin by X-irradiation of an Ar matrix containing anthralin led also to significant formation of its radical anion, i.e., anthralin acts apparently as an efficient electron trap in such experiments.

Full text of DOI:10.1021/jo060622o

Anthralin: Primary Products of Its Redox Reactions
Malgorzata Czerwinska, Adam Sikora, Piotr Szajerski, Jacek Zielonka, Jan Adamus, and
Andrzej Marcinek*
Institute of Applied Radiation Chemistry, Technical UniVersity, 90-924 Lodz, Poland
Krzysztof Piech, Pawel Bednarek, and Thomas Bally*
Department of Chemistry, UniVersity of Fribourg, CH-1700 Fribourg, Switzerland
marcinek@p.lodz.pl; thomas.bally@unifr.ch
ReceiVed March 21, 2006
One-electron reduction significantly enhances the ability of anthralin, 1, to act as a hydrogen atom donor.
On annealing of an MTHF glass in which the radical anion of anthralin, 1•-, is generated radiolytically,
this species decays mainly by loss of H• to give the anthralyl anion, 2-. On the other hand, radicals
formed on radiolysis of matrices that are suitable for the generation of radical anions or cations are
capable to abstract H• from anthralin to give the anthralyl radical, 2•. Both 2- and 2• are obtained
simultaneously by mesolytic cleavage of the radical anion of the anthralin dimer. Contrary to general
assumptions, the anthralyl radical is found to be much more reactive toward oxygen than the anion. All
intermediates are characterized spectroscopically and by reference to quantum chemical calculations.
Attempts to generate the radical cation of anthralin by X-irradiation of an Ar matrix containing anthralin
led also to significant formation of its radical anion, i.e., anthralin acts apparently as an efficient electron
trap in such experiments.
1. Introduction
destruction of membrane lipids.1-7 There is growing evidence
that these biological effects are related to redox reactions of
anthralin resulting in the formation of anthralyl radicals8-15 and
the production of reactive oxygen species (ROS).4,5,16,17 These
species may be also responsible for undesired side effects of
anthralin, such as skin irritation and staining.6
Anthralin (1) (1,8-dihydro-9-anthrone or 1,8-dihydro-9,10H-
anthracenone) is one of the most effective drugs for topical
treatment of psoriasis, a common skin disease. This synthetic
substitute of the natural product, chrysarobin, targets both the
inflammatory and hyperproliferative aspects of the disease.
Despite the fact that this drug has been successfully used for
over 70 years, the accurate mechanism of the action of anthralin
has not been elucidated.1-5
(6) Mu¨ller, K.; Gu¨rster, D. Biochem. Pharmacol. 1993, 46, 1695.
(7) Wiegrebe, W.; Mu¨ller, K. Skin Pharmacol. 1995, 8, 1.
(8) Martinmaa, J.; Vanhala, L.; Mustakallio, K. K. Experientia 1978,
34, 872.
(9) Davies, A. G.; Hawari, J. A. A.; Whitefield, M. Tetrahedron Lett.
1983, 24, 4465.
The therapeutic action of anthralin is related to its interaction
with DNA and the inhibition of enzymes associated with cell
proliferation and inflammation or lipid peroxidation and the
(10) Mustakallio, K. K.; Martinmaa, J.; Vilvara, R.; Hamekoski, J. Med.
Biol. 1984, 62, 155.
(1) Ashton, O. I.; Andre, P.; Lowe, N. J.; Whitefield, M. J. Am. Acad.
Dermatol. 1983, 9, 173.
(2) Kemeny, E.; Ruzicka, T.; Braun-Falco, O. Skin Pharmacol. 1990, 3,
1.
(3) van de Kerkhof, P. M. Eur. J. Dermatol. 1991, 1, 79.
(4) Mu¨ller, K. Biochem. Pharmacol. 1997, 53, 1215.
(5) Mu¨ller, K. Gen. Pharmac. 1996, 27, 1325.
(11) D’Ischia, M.; Prota, G. Gazz. Chim. Ital. 1985, 115, 511.
(12) Shroot, B.; Brown, C. Arzneim. Forsch. 1986, 36, 1253.
(13) Fuchs, J.; Packer, L. InVest. Dermatol. 1989, 92, 677.
(14) Lambelet, P.; Ducret, F.; Lo¨liger, J.; Maignan, J.; Reichert, U.;
Shroot, B. Free Radic. Biol. Med. 1990, 9, 183.
(15) Hayden, P. J.; Free, K. E.; Chignell, C. F. Mol. Pharmacol. 1994,
46, 186.
10.1021/jo060622o CCC: $33.50 © 2006 American Chemical Society
Published on Web 06/09/2006
5312
J. Org. Chem. 2006, 71, 5312-5319

Anthralin: Primary Products of Its Redox Reactions
SCHEME 1. Compounds Potentially Involved in the Redox
Reactions of Anthralina
SCHEME 2. Reductive Cleavage of the Dehydro Dimer of
Anthralin
intermediates may be involved in the mode of action of the
drug, among them the primary products of electron-transfer
reactions, i.e., radical ions or neutral radicals which we
characterize in this work.
2. Results and Discussion
Anthralin Radical Anion. Anthralin (1) is poorly soluble
in aqueous solutions, alcohols, and in the organic solvents used
commonly to form low-temperature glasses. In experiments
where radical ions are generated by radiolytic reduction or
oxidation in solid matrices, the concentration of solute molecules
must be at least 10-3 M to ensure efficient scavenging of the
initially generated electrons or holes. Poor solubility of the
solute, even if it has a low ionization potential or a high electron
affinity, may therefore represent a significant obstacle.
On the other hand, in alkaline aqueous or alcoholic solutions,
or in polar solvents such as acetonitrile, anthralin is deproto-
nated, at least partially, to 2-. The pKa value of anthralin was
found by titration in aqueous buffer solutions to be 9.5.19,20
Although the concentration of 2- must therefore be very limited
under physiological conditions, there is conclusive evidence for
the formation of superoxide during the auto-oxidation of
anthralin anion.17,21 In contrast to 2- or 1(enol), both of which
are susceptible to auto-oxidation, the keto form of anthralin is
not; hence, the transient products of redox reactions of 1 in its
keto form have not been identified to date. We start out by
presenting the characteristics of the radical anion of 1.
The optical spectrum obtained on pulse radiolysis of a solution
of anthralin in a glass of deoxygenated 2-methyltetrahydrofuran
(MTHF) at 77 K22 (red trace in Figure 1A) is composed of a
strong band peaking at 740 nm (followed by two members of
a vibrational progression of ca. 980 cm-1), accompanied by
some weaker features at shorter wavelengths. As 1 exists
exclusively in its keto form (which shows a strong structureless
band at 356 nm) in MTHF, the above spectrum must be ascribed
to 1•-. According to our calculations, the slightly bent keto form
of 1•- is more stable than its planar enol form by nearly 33
kcal/mol; i.e., the order of keto/enol stability of 1 is not inverted
on reduction. Hydrogen atom transfer from one of the hydroxyl
groups to the central carbonyl group, as it had been observed
to occur in the excited state of 1,23,24 is also unfavorable in 1•-
a Species 2 lacks a hydrogen atom compared to species 1.
Studies of the redox reactions of anthralin may be helpful in
elucidating its therapeutic action. In the present study, the
transient species generated upon one-electron reduction and
oxidation of anthralin are identified and characterized for the
first time. These studies reveal that upon reduction anthralin
spontaneously loses a proton to yield the anthralyl anion, 2-
(the enolate of 1,8,9-trihydroxyanthracene), while the corre-
sponding radical, 2• (1,8-dihydro-9-anthron-10-yl), is formed
directly from the neutral anthralin by hydrogen atom abstraction
rather than by oxidation of the anthralyl anion (cf. Scheme 1).
It has been claimed that anthralin reacts with oxygen to yield
the superoxide radical anion, but there is strong evidence that
auto-oxidation of anthralin can only occur in its enol form (1-
enol) or in the anthralyl anion 2- (Scheme 1).4,5,17 Therefore, it
was proposed that these two species react with oxygen,
producing superoxide and the radical cation of 1-enol or the
anthralyl radical, 2•, respectively. The role of 2• in the therapeutic
process is not known, but it is often questioned because 2• is a
particularly stable radical, as shown by its low reactivity toward
oxygen or vitamin E.18
In this paper, we explore a different pathway for the
generation of 2•: we have found that one-electron reduction of
1 or its 10,10′-dehydrodimer (2)2 (1,8,1′,8′-tetrahydroxy-10,-
10′-bisanthrone) leads to concomitant formation of 2- and 2•
under the same experimental conditions (Scheme 2) and thus
allows for the direct comparison of their reactivity toward
oxygen.
The photochemistry of anthralin is also of considerable
interest since UV radiation is often used as an adjunct to
anthralin therapy, although exposure of skin treated with the
drug to light causes irritation and erythema. This sensitivity to
light and oxygen indicates that some photochemically generated
(19) Mu¨ller, K.; Kanner, R. C.; Foote, C. S. Photochem. Photobiol. 1990,
52, 445.
(20) Sa e Melo, T.; Dubertret, L.; Prognon, P.; Gond, A.; Mahuzier, G.;
Santus, R. J. InVest. Dermatol. 1983, 80, 1.
(21) Mu¨ller, K.; Wiegrebe, W.; Younes, M. Arch. Pharm. (Weinheim)
1987, 320, 59.
(16) Mu¨ller, K.; Kappus, H. Biochem. Pharmacol. 1988, 37, 4277.
(17) Bruce, J. M.; Kerr, C. W.; Dodd, N.; J. F. J. Chem. Soc., Faraday
Trans. 1 1987, 83, 85.
(18) Bruce, J. M.; Dodd, N. J. F.; Gorman, A. A.; Hamblett, I.; Kerrr,
C. W.; Lambert, C.; McNeeney, S. P. Photochem. Photobiol. 1990, 52,
345.
(22) Shida, T. Electronic Absorption Spectra of Radical Ions; Elsevier:
Amsterdam, 1988.
(23) Andersen, K. B.; Spanget-Larsen, J. Spectrochim. Acta 1997, 53 A,
2615.
(24) Moller, S.; Andersen, K. B.; Spanget-Larsen, J.; Waluk, J. Chem.
Phys. Lett 1998, 291, 51.
J. Org. Chem, Vol. 71, No. 14, 2006 5313

Czerwinska et al.
TABLE 2. Excited States and Electronic Transitions of 2- by
TD-B3LYP/6-31+G* Calculations
state
λ
max/nm
f × 102
excitationsa
11A1
21A1
11B1
11A2
11B2
21B1
31B1
21A2
21B2
(6b1)2
433
387
369
359
336
311
310
308
7.36
<0.01
0
25.15
0.09
0.34
0
6b1 f 7b1
6b1 f 28a1
6b1 f 23b2
6b1 f 5a2
6b1 f 29a1
6b1 f 30a1
6b1 f 24b2
6b1 f 6a2
6.54
a In terms of the π-MOs shown in Figure S2 of the Supporting
Information (the a1 and b2 σ*-MOs are very similar to those shown in Figure
S1 of the Supporting Information); the listed excited configurations have
coefficients >0.9 in the CIS-type wave function.
TABLE 3. Excited States and Electronic Transitions of 2• by
TD-B3LYP/6-31G* Calculations
state
λ
max/nm
f × 102
excitationsa
12B1
12A2
22B1
22A2
12B2
32B1
(6b1)1
676
549
501
468
463
8.86
<0.01
0.58
0
4a2 f 6b1
5b1 f 6b1
3a2 f 6b1
22b2 f 6b1
6b1 f 7b1
FIGURE 1. (A) Red trace: electronic absorption spectrum obtained
on radiolysis of anthralin 1 embedded in an MTHF glass at 77 K (0.02
M solution, radiation dose 2.5 kGy, optical path 2 mm). Gray and black
traces: decay of 1•- and buildup of 2- (bands at 400-500 nm) on
annealing at 90-95 K. (B) Spectral changes observed on further
annealing of an MTHF glass already containing 2- (green spectrum)
at 95-100 K. The final blue spectrum is asssigned to 2• (see text).
The red bars in (A) and the blue bars in (B) represent the results of the
TD-DFT calculations for 1•- (Table 1) and 2• (Table 3), respectively.
1.56
a In terms of the MOs shown in Figure S2 in the Supporting Information;
the listed excited configurations have coefficients >0.9 in the CIS-type wave
function. No intense transitions are predicted betweeen 450 and 350 nm.
radiation dose) showed that the yield of 1•- is independent of
the concentration of 1 and that no traces of unscavenged,
solvated electrons (λmax) 1200 nm) remain in that concentration
range.
TABLE 1. Excited States and Electronic Transitions of 1•- by
TD-B3LYP/6-31+G* Calculations
The inhibitory effect of the rigid matrix on intermolecular
processes gradually withers on annealing a MTHF matrix
slightly above its glass transition temperature, ca. 90 K. If this
is done, a new system of bands with twin peaks centerd at 430,
455, and 485 nm arises at the expense of the bands of 1•- (see
Figure 1A). By comparison with spectra of 1 taken in DMF or
in alkaline MeOH solution,25,26 as well as with the TD-DFT
calculations in Table 2, we assign this new spectrum to 2-.
Unfortunately, the strongest band of this anion (observed at
392 nm in DMF25 and predicted at 359 nm in vacuo) is not
visible under our experimental conditions because it is obscured
by strong absorptions of 1. The observed band system appears
to be due to a single electronic transition (HOMO f LUMO)
showing vibrational progressions of ca. 1330 and 310 cm-1
which had not been revealed by the previous solution spectra.25,26
The absorptions of 2- grow until the bands of 1•- have
completely decayed.
Prolonged annealing of the matrix at elevated temperatures
(above 100 K) leads to the decay of 2- and the appearance of
a new, different band system at 600-800 nm (λmax ) 730 nm,
Figure 1B) which we assign to the neutral radical 2• by
comparison to the known spectrum of this species in solution18,27
and the results of the TD-DFT calculations in Table 3. These
show again that the observed band system must be due to a
single electronic transition (which is predicted ca. 0.14 eV too
state
λ
max/nm
f × 102
excitationsa
12B1
22B1
12A2
12A1
12B2
22A1
22A2
(7b1)1
695
686
669
611
532
516
1.01
11.91
0.28
0
0
3.41
7b1 f 8b1
7b1 f 5a2
7b1 f 28a1
7b1 f 23b2
7b1 f 29a1
7b1 f 6a2
a In terms of the MOs shown in Figure S1 of the Supporting Information;
the listed excited configurations have coefficients >0.9 in the CIS-type wave
function. No intense transitions were predicted between 500 and 350 nm.
(∆H ) +6.4 kcal/mol). The observed spectrum is in accord
with the results of TD-DFT calculations (Table 1) which predict
a strong absorption with λmax)686 nm, preceded by a 10 times
weaker one at 695 nm (which could be responsible for the weak
long-wavelength shoulder).
The next possibly palpable transition is predicted at 516 nm,
i.e., where the experimental spectrum is empty, but it is unclear
whether gas-phase TD-DFT calculations are capable of predict-
ing the positions of higher lying excited states of radical anions
(which are usually dissociative in vacuo) in condensed phase
with any accuracy, especially if these states involve σ* MOs.
It is, however, interesting to note that all electronic transitions
of 1•- in the visible range involve excitation from the singly
occupied MO (SOMO, 7b1) to higher lying virtual orbitals
(including a1 and b2 σ*-MOs) and that excitations into the
SOMO do not contribute to the spectrum of 1•- predicted by
the TD-DFT method.
(25) Sellmer, A.; Terpetschnig, E.; Wiegrebe, W.; Wolfbeis, O. S. J.
Photochem. Photobiol. A: Chem. 1998, 116, 39.
(26) Dabestani, R.; Hall, R. D.; Sik, R. H.; Chignell, C. F. Photochem.
Photobiol. 1990, 52, 961.
(27) Bruce, J. M.; Gorman, A. A.; Hamblett, I.; Kerr, C. W.; Lambert,
C.; McNeeney, S. P. Photochem. Photobiol. 1989, 49, 439.
Experiments conducted with initial concentrations of 1
ranging from 5 × 10-3 to 7.5 × 10-2 M (but with the same
5314 J. Org. Chem., Vol. 71, No. 14, 2006

Anthralin: Primary Products of Its Redox Reactions
estimated. As the yield of 2- and the rate for its decay do not
depend on the initial concentration of 1, the parallel decay of
the anion and the growth of the radical absorption (Figure 1B)
should be assigned to two independent intermolecular reactions.
Although MTHF is an aprotic solvent, the positive ions,
C5H11O+, that are generated on irradiation of the matrix (reaction
5 in the Experimental Section) may act as proton donors. A
good example of this is the well-documented formation of ketyl
radicals on radiolysis and subsequent annealing of MTHF
matrices containing benzophenone and acetophenone.22,29 The
same could happen to 2- which would then re-form 1 (reaction
1).
2- + C5H11O+ f 1 + C5H10O
1 + C5H9O• f 2• + C5H10O
(1)
(2)
FIGURE 2. Kinetic traces of the changes in absorbance at 680 nm of
the transient species generated by pulse radiolysis of (a) 0.005 M, (b)
0.01 M, and (c) 0.05 M solutions of 1 in deareated MTHF at 25 °C
(radiation dose 65 Gy, optical path 10 mm).
On the other hand, because of the high stability of the radical
2•, anthralin itself may act as a hydrogen atom donor in reactions
with less stable radicals, a process which has been amply
documented.16,18,27,30 In the present case, C5H9O•, the other
product that is formed in irradiation of solid MTHF (cf. reaction
5 in the Experimental Section), may play this role and react
with 1 to form 2• (reaction 2). This process would explain the
dependence of the rate of formation of 2• on the concentration
of 1 (see Figure 2). On the other hand, the yield of 2• increases
also with the irradiation dose, i.e., when more C5H9O• radicals
are formed. According to our calculations, reaction 2 is
exothermic by 21.1 kcal/mol. This “repair” reaction of anthralin
with free radicals may be responsible for the antioxidant
properties of this compound.
If reaction 2 is responsible for the formation of 2• in the
cryogenic experiments, it is interesting to compare the hydrogen
atom donor properties of anthralin itself and its radical anion.
Addition of one electron to the molecule appears to significantly
increase the propensity for hydrogen atom detachment, as seen
by the difference in the rates for both processes under matrix
conditions (hydrogen atom transfer from 1•- precedes that from
1). According to DFT calculations, a good part of the spin in
high in energy), this time from the highest doubly occupied
MO to the SOMO.
TD-DFT predicts weak transitions at ca. 500 and 460 nm
which could give rise to the little bumps in that region, but no
further intense transitions down to 350 nm. Thus, the band at
440 nm, which later continues to grow at the expense of the
bands of 2•, is not due to 2•. Bruce et al. have previously
suggested18 that this band could be due to a tautomeric form of
the dimer (2)2, formed upon radical dimerization, but it could
also be due to secondary radicals.13-15 As will be discussed in
the following section, we also found that, upon reduction, (2)2
undergoes spontaneous fragmentation to 2• and 2-, which
therefore constitutes an independent route for the formation of
these species in an MTHF glass.
Formation of 2- from 1•- requires the detachment of a
hydrogen atom from the C-10 position,25 which may be a facile
process if a suitable acceptor is available. Possible candidates
are neutral solvent or solute molecules or radicals formed upon
matrix irradiation. As the yield of 2- depends only on the initial
yield of 1•- which in turn does not depend on the concentration
of 1 (see above), we assume that 1 does not serve as a hydrogen
atom acceptor. Thus, hydrogen atoms must be transferred to
the solvent molecules or onto the byproducts of radiolysis, the
concentration of which depends only on the irradiation dose.
The fact that the formation of 2• occurs at the expense of 2-
seems to indicate that the process is a simple auto-oxidation of
2-, as it has been shown to occur in reactions with stable radicals
such as nitroxides.17,21,28 It could be that byproducts of the
radiolysis act as electron acceptors in frozen MTHF. However,
experiments conducted with different initial concentrations of
1, both under matrix conditions and in solution (see Figure 2),
show an increase of the yield of 2• and the rate of its formation
with the concentration of 1, in contrast to 2- (see above).
As both 1•- and 2• absorb around 700 nm, the initial decay
seen in the kinetic traces in Figure 2 reflects the decay of 1•-,
but on a longer time scale the formation of 2• dominates the
absorption at 680 nm. From this part of the trace, pseudo-first-
order rate constants for the formation of 2• can be obtained from
which second-order rate constants of ca. 108 M-1 s-1 can be
1•-
is located at C10 (Mulliken spin population ) 0.28), which
facilitates abstraction of one of the hydrogen atoms attached to
this carbon atom. Hence, 1•- may react with MTHF, whereas 1
requires species of higher hydrogen atom affinity, such as
C5H9O•.
Anthralin Dimer Radical Anion. We have shown above
that adding an electron to anthralin facilitates cleavage of the
C(10)-H bond. This finding led us to expect that reduction of
the anthralin dimer, (2)2, will lead to its spontaneous cleavage.
Such so-called mesolytic fragmentation processes31 are more
commonly activated by remoVal of an electron, thus providing
a useful method of breaking bonds that are quite strong in neutral
substrates,31,32 but there are also some examples of fragmenta-
tions following attachment of an electron.33 According to DFT
calculations, mesolytic fragmentation of the radical anion of
(2)2 to 2- and 2• is exothermic by 5.7 kcal/mol in the gas phase
(29) Hayon, E.; Ibata, T.; Lichtin, N. N.; Simic, M. J. Phys. Chem. 1972,
76, 2072.
(30) Mu¨ller, K.; Leukel, P.; Ziereis, K.; Gawlik, I. J. Med. Chem. 1994,
37, 1660.
(31) Mazlak, P. Top. Curr. Chem. 1993, 168, 1.
(32) Rogowski, J.; Zielonka, J.; Marcinek, A.; Gebicki, J.; Bednarek, P.
J. Phys. Chem. A. 2003, 107, 810.
(28) Pecar, S.; Schara, M.; Muller, K.; Wiegrebe, W. Free Rad. Biol.
Med. 1995, 18, 459.
J. Org. Chem, Vol. 71, No. 14, 2006 5315

Czerwinska et al.
anion 2- formed on its fragmentation, whereas 2• is effectively
quenched (Figure 3B). This contrasts with the general assump-
tion that, due to its stability and inertness toward oxygen, 2•
should be discounted as an important intermediate in the mode
of action of anthralin as an antipsoriatic drug,4,18 in contrast to
2-, the formation of which was considered as a first step toward
generation of reactive oxygen species. From the above experi-
ments, it is, however, clear that, in actual fact, the anion is even
more inert toward oxygen than the radical. Therefore, the direct
autoxidation of 2- through reaction with O2 (to form O2•-) is
probably too slow to be effective in the absence of catalysts
such as iron or copper.4,5,35
If anthralin itself is reduced in O2-saturated MTHF matrices
the radical absorption is also absent from the spectra at any
stage of the annealing process (in contrast to the spectra
presented in Figure 1B) although the decay of the anion 2- is
not affected. As dimerization of radicals under these experi-
mental conditions is impeded, reaction with oxygen takes over,
leading probably to the formation of the anthralyl peroxy radical.
Attempts To Generate the Anthralin Radical Cation.
Radicals such as 2• may also be formed by way of radical
cations which often undergo facile deprotonation.36 Therefore,
we decided to look also at the oxidation of 1. We were able to
dissolve enough 1 in the Freon mixture to get a strong absorption
at <300 nm. However, γ-irradiation of the glass formed from
this solution only led to a broad band peaking at ca. 600 nm
(dashed line a in Figure 4) which usually occurs when no species
that are more easily oxidizable than the Freons are available to
trap the positive charges. This band can be readily bleached
(Figure 4b), but still no absorptions that could be attributed to
1•+ are revealed. In contrast, a solution of similar concentration
of 1 in an MTHF glass exposed to the same conditions of
irradiation resulted in spectrum c in Figure 4, which is almost
identical to that obtained after pulse radiolysis in the same
solvent (Figure 1A, red spectrum).
Because we were not sure whether our failure to observe 1•+
was due to an insufficient concentration of 1 in the Freon
mixture, we resorted to the Ar matrix technique for further
experiments. Thus, we fed 1 into a stream of Ar/N2 (10:1) doped
with 0.1 mol % CH2Cl2 (which usually acts as an efficient
electron scavenger) and deposited it on a CsI window held at
20 K. The resulting matrix was exposed to 90 min of
X-irradiation, a procedure that usually results in the conversion
of 10-15% of the neutral substrate to its radical cation (or
secondary products thereof).37
FIGURE 3. (A) Red spectrum: Electronic absorption spectrum
obtained after radiolysis of a frozen MTHF saturated with anthralin
dimer, (2)2 (<0.005 M, deoxygenated solution, radiation dose 4.1 kGy,
sample thickness 2 mm) at 77 K. Black and gray traces: spectral
changes observed upon annealing of the above matrix at 90-95 K.
(B) Same as A, but starting with a solution that was saturated with O2
before freezing (note that the initial spectrum is not affected by the
presence of O2).
(and probably more so in MTHF because of stronger solvation
of the more localized monomer anion).
The electronic absorption spectrum obtained upon reduction
of (2)2 in MTHF at 77 K is presented in Figure 3 (red spectrum).
It is composed of a broad absorption band peaking at 650 nm
and a more intense one at 445 nm with a weak shoulder at 470
•-
nm. We assign this spectrum to the primary radical anion, (2)2
,
although substantiating this assignment by calculations is not a
trivial matter because it is unclear whether solvation favors a
situation where charge and spin are localized in one of the two
anthralin moieties or whether they are delocalized over both,
even in MTHF. DFT calculations invariably predict the latter
because it is impossible to enforce a localization of spin and
charge with such methods.34 It is also unclear whether the C-C
σ-bond linking the two halves of (2)2•- is still intact or whether
the red spectrum in Figure 3 is due to a π-complex between 2-
and 2•. In any event, this species is stable at 77 K because its
dissociation is inhibited by the rigidity of the MTHF matrix.
However, slight softening of the matrix by annealing it to 90 K
leads to the disappearance of the red spectrum, a decay which
is accompanied by the rise of the sharp bands at 440-500 nm
of 2- and the structured band system at 500-800 nm of 2• (gray
and black traces in Figure 3A).
To our surprise, the spectrum which we obtained by this
procedure (Figure 4d, dashed line) looked, however, almost
indistinguishable from that recorded in MTHF and attributed
to the radical anion, 1•-. We had never before obtained a radical
anion under such conditions, although previous work had
indicated that it is possible to generate radical anions of species
with high electron affinities in Ar matrices by X-irradiation if
amines are added (as hole scavengers).38 In an experiment where
we doped the Ar matrix with DABCO instead of CH2Cl2, a
very similar spectrum was obtained (the yield of the radical
anion was slightly enhanced), except that the absorption rising
Both species are formed simultaneously and with similar rates.
Their decay upon further annealing of the matrix is also similar.
If 2- were oxidized to 2•, some differences in the decay kinetics
of both species should occur. Their parallel disappearance
indicates, however, independent decay paths which are triggered
under the same conditions of softening the matrix.
(35) Miller, D. M.; Buettner, G. R.; Aust, S. D. Free Rad. Biol. Med.
1990, 8, 95.
(36) Roth, H. D.; Weng, H.; Zhou, D.; Lakkaraju, P. S. Acta Chim. Scand.
1997, 51, 626.
Saturating the MTHF solution with oxygen before freezing
does not affect the yield of the primary radical anion or of the
(37) Bally, T. In ReactiVe Intermediate Chemistry; John Wiley & Sons:
(33) Chanon, M.; Rajzmann, M.; Chanon, F. Tetrahedron 1990, 46, 6193.
(34) Bally, T.; Sastry, G. N. J. Phys. Chem. A. 1997, 101, 7923.
Hoboken, NJ, 2004; p 797.
(38) Gebicki, J.; Michl, J. J. Phys. Chem. 1988, 92, 6452.
5316 J. Org. Chem., Vol. 71, No. 14, 2006

Anthralin: Primary Products of Its Redox Reactions
SCHEME 3. Tautomerization of the Radical Cation of
Anthralin
TABLE 4. Excited States and Electronic Transitions of 1′•+ by
TD-B3LYP/6-31G* Calculations
state
λ
max/nm f × 102
leading excitationsa
12A′′
22A′′
22A′
(10a′′)1
1405
813
675
560
495
431
411
0.31
<0.01
2.95
0.06
0.08
9a′′ f 10a′′
49a′ f 10a′′
8a′′ f 10a′′
0.78 (10a′′ f 11a′′) -0.67 (9a′′ f 11a′′)
32A′′
42A′′
52A′′
62A′′
72A′′
7a′′ a` 10a′′
3.70
8.25
0.58 (10a′′ f 11a′′) + 0.67 (9a′′ f 11a′′)
0.78 (9a′′ f 11a′′)
FIGURE 4. Difference spectra obtained on (a) 60Co γ-radiolysis of a
saturated solution of anthralin 1 in a Freon mixture (see the Experi-
mental Section) at 77 K; (b) photolysis of the sample giving spectrum
(a) at >600 nm; (c) 60Co γ-radiolysis of a 10-3 M solution of 1 in
MTHF at 77 K; (d, e) X-irradiation of a 12 K Ar matrix containing ca.
1 mol % of 1 and (d): a similar amount of CH2Cl2 (an efficient electron
scavenger) or (e) a similar amount of DABCO (an efficient hole
scavenger).
a In terms of the MOs shown in Figure S3 in the Supporting Information;
unless indicated, the listed excited configurations have coefficients >0.9
in the CIS-type wave function (note that the coefficients are not normalized
in the G03 output).
these experimental conditions. Second, TD-DFT calculations
for 1(enol)•+ predict an intense band at 705 nm (f ) 0.107) but
no bands of similar intensity down to 300 nm, a prediction that
is clearly incompatible with our observations.
On the other hand, it turns out that in the radical cation the
quinoid tautomeric form 1′ (which is disfavored by ca. 26 kcal/
mol in the neutral, due to loss of an aromatic ring, and by 6.4
kcal/mol in 1•-) is only 0.8 kcal/mol less stable than 1•+ by
B3LYP/ 6-31G* and that the H-shift that leads to 1′•+ is nearly
barrierless (Scheme 3). Single-point calculations at the CCSD-
(T) level predict 1′•+ to be more stable than 1•+ by 4.6 kcal/
mol, so we believe that upon ionization, 1 tautomerizes
spontaneously to the quinoid form 1′.
below 500 nm was missing (Figure 4e). At the same time, the
formation of DABCO•+ (which has only a weak visible
absorption) could be seen very clearly in the IR spectrum. Thus,
we are forced to conclude that 1 is indeed a more efficient
electron scavenger than CH2Cl2, i.e., that we have witnessed
the efficient formation of a radical anion in an Ar matrix.39,40
The question remains whether, in the presence of CH2Cl2,
1•+ is perhaps also formed, but does not manifest itself in the
UV/Vis spectrum (unfortunately, the changes in the IR spectra
are too weak to be diagnostically useful). To examine this
possibility, we carried out some B3LYP/6-31G* calculations
on 1•+. The first thing to note from these results is that, similar
to many other cases,41,42 but in contrast to 1•-, the order of
stability of the keto and enol forms is inverted in 1•+: While 1
is more stable than 1(enol) by ca. 19 kcal/mol the radical cation
of 1(enol) is more stable than 1•+ by 13.4 kcal/mol (this
inversion of stabilities may explain why the enol form is
susceptible to autoxidation, whereas the keto form is not).
However, two reasons speak against the formation of 1(enol)•+
under matrix conditions: first, the lowest energy pathway for
the conversion of 1•+ to its enol form (a sequence of five 1,2-H
shifts, see Figure S4 in the Supporting Information) involves a
barrier of over 40 kcal/mol which cannot be surmounted under
According to TD-DFT calculations, 1′•+ has two very weak
NIR transitions, the first of which involves an electron transfer
from the aromatic to the quinoid part of 1′•+. Two bands of
similar (medium) intensity are predicted at 675 and 431 nm,
followed by a much stronger one at 411 nm (see Table 4). If
the 675 nm band is hidden under the much more intense one of
the (simultaneously formed) radical anion, then the latter two
bands may account for the steep rise below 500 nm which
disappears when the electron scavenger (CH2Cl2) is replaced
by a hole scavenger (DABCO) in the Ar matrix experiments.
In conclusion, if 1 is oxidized under the conditions which
usually lead to radical cations of substrates added to Ar matrices,
then it will probably relax spontaneously to 1′•+. However, the
unmistakable presence of 1•- under these conditions suggests
that 1 is a much better trap of the electrons that are liberated
on X-irradiation of solid Ar than of the holes which are created
at the same time. In fact, its electron-scavenging ability exceeds
that of CH2Cl2.
(39) Note that many radical anions were generated in Ar matrices and
observed by the very sensitive technique of ESR spectroscopy (see, e.g.:
Kasai, P. H. Acc. Chem. Res. 1971, 4, 329). However, to obtain transmission
spectra, much higher concentrations of these species are required.
(40) To convince ourselves that such efficient formation of radical anions
in Ar matrices doped with CH2Cl2 is indeed possible we carried out a similar
set of experiments with p-benzoquinone. Indeed, upon X irradiation of such
an Ar matrix containing this compound we obtained the characteristic
spectrum of the corresponding radical anion.22,38 In the IR spectrum, we
witnessed the simultaneous formation of the radical cation of p-benzo-
quinone, a species which, to the best of our knowledge, has hitherto not
been observed in condensed phase (Piech, K., Bally, T. Unpublished results).
(41) Gebicki, J.; Marcinek, A. In General Aspects of Free Radical
Chemistry; Alfassi, Z. B., Ed.; Wiley: Chichester, 1999.
As mentioned above, the solubility of anthralin in solvents
such as 2-chlorobutane is too low for generation of any traces
of the radical cation in this glassy matrix. However, upon
softening of the irradiated 2-chlorobutane matrix containing
soluble amounts of anthralin the formation of small amounts
of radical 2• can be observed. Much better yields of this product
are obtained if the experiment is done in ionic liquid glasses.
(42) Gebicki, J.; Bally, T. Acc. Chem. Res. 1997, 30, 477.
J. Org. Chem, Vol. 71, No. 14, 2006 5317

Czerwinska et al.
Tables 3 and 4), the strong increase of the absorption at this
wavelength indicates that 2• cannot possibly be formed only at
the expense of (i.e., by deprotonation of) 1′•+. Thus, we presume
that 2• arises through reaction of 1 with radical byproducts of
the radiolysis (analogous to reaction 2 in MTHF). At still higher
temperature, the spectrum of the radical decays, too, leaving
behind a species with an absorption at 440 nm.
3. Conclusions
Almost a century has elapsed since anthralin 1 was introduced
as an antipsoriatic drug, but the mechanism by which 1 exerts
its action on biological targets and the exact nature of the
chemical species which are pharmacologically active has not
yet been elucidated. In this paper, we presented for the first
time the primary products of redox reactions involving 1, i.e.,
its radical ions.
Addition of an electron to the 1 significantly increases its
ability to act as a hydrogen atom donor. Although anthralin itself
can lose a hydrogen atom in reactions with, e.g., peroxyl
radicals, the extra electron diminishes the C(10)-H or C(10)-
C(10′) bond dissociation energy in anthralin or its dimer,
respectively.
A suprising finding of this study is that the anthralyl radical,
2•, is much more reactive toward oxygen than the corresponding
(closed shell) anion, 2-. Therefore, the direct auto-oxidation of
2- cannot be responsible for the production of reactive oxygen
species. Conversely, 2• cannot be regarded as inactive toward
oxygen.
Another surprising finding is that under conditions that usually
lead to the formation of radical cations (X-irradiation in Ar
doped with CH2Cl2), anthralin beats CH2Cl2 in scavenging the
electrons that are liberated in the process and forms the radical
anion. We plan to explore the scope of this method which allows
in principle to record also the vibrational spectra of radical
anions, something that, to the best of our knowledge, has not
been tried to date.
FIGURE 5. (a) Red trace: absorption spectrum obtained on radiolysis
of 1 in an ionic liquid/CHCl3 glass at 77 K (radiation dose 10 kGy,
sample thickness 2 mm). Blue traces: spectral changes observed upon
subsequent annealing of the matrix (trace b: final spectrum; c: spectrum
of 2• (reproduced from Figure 1); d: spectrum of 2• obtained by
depositing (2)2 into an Ar matrix.
We have shown previously that these novel solvents, many of
which form transparent glasses at 77 K,23 can be used for
generating radical ions by radiolysis and studying them by
transmission spectroscopy. Recently, we found that the glass
quality does not change upon addition of an organic component
to the ionic liquid. For example, 1:1 mixtures of 1-butyl-3-
methylimidazolium hexafluorophosphate and CHCl3 form a
transparent glass. Not only does CHCl3 improve the solubility
of many precursors, but it leads to a higher yield of radical
cations on radiolysis due to its ability to scavenge electrons by
•
dissociative attachment (CHCl3 + e- f CH2Cl + Cl-).
The initial spectrum of ionized anthralin embedded in the
above mixture shows only a very weak absorption at 700 nm
(Figure 5, red trace). However, annealing of the matrix (up to
180 K) causes a significant growth of the anthralyl radical
absorption (cf. blue lines). Juxtaposition of this spectrum to that
obtained in MTHF (blue dashed line) and that obtained by
depositing (2)2 into an Ar matrix51 indicates that the low energy
shoulders at 740-800 nm are much weaker in the spectrum in
the ionic liquids, a phenomenon for which we have no
explanation. Although according to TD-DFT the absorptivity
of 2• at 700 nm is about 3 times greater than that of 1′•+ (cf.
4. Experimental Section
4.1. Materials. Anthralin, 2-chlorobutane, 2-methyltetrahydro-
furan (MTHF), and chloroform were obtained from commercial
sources. The anthralin dimer (1,8,1′,8′-tetrahydroxy-10,10′-bisan-
throne) was prepared according to the procedure described previ-
ously.43
Preparation of 1-Butyl-3-methylimidazolium Hexafluoro-
phosphate. A mixture of 1-methylimidazole (41 g, 0.5 mole) and
1-chlorobutane (47 g, 0.51 mole) was vigorously stirred and heated
at 75 °C for 75 h. The resulting solid, 1-butyl-3-methylimidazolium
chloride, was thoroughly washed with ethyl acetate (g4 × 100 mL),
and the remaining volatile compounds were removed by heating
to 50 °C under 0.1 mmHg of pressure. To the chloride solution in
300 mL of water, cooled with an ice-water mixture and vigorously
stirred, was added dropwise 160 g (0.65 mole) of 60% hexafluo-
rophosphoric acid solution. The lower layer was separated and
washed with deionized water until the washings were no longer
acidic. The resulting yellowish viscous oil was dissolved in 150
mL of dichloromethane and purified on a chromatographic column
packed with silica gel (15 g, lower) and charcoal (10 g, upper)
layers. After the dichloromethane was removed on a rotavap, the
remaining colorless oil was pumped at 0.1 mmHg/50 °C for 2 h to
give 125 g (88%) of 1-butyl-3-methylimidazolium hexafluorophos-
phate.
(43) Auterhoff, H.; Scherff, F. C. Arch. Pharm. (Weinheim) 1960, 293,
918.
(44) Gebicki, J.; Marcinek, A.; Rogowski, J. Radiat. Phys. Chem. 1992,
39, 41.
(45) Karolczak, S.; Hodyr, K.; Lubis, R.; Kroh, J. J. Radioanal. Nucl.
Chem. 1986, 101, 177.
(46) Marcinek, A.; Zielonka, J.; Gebicki, J.; Gordon, C. M.; Dunkin, I.
R. J. Phys. Chem. A 2001, 105, 9305.
(47) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(48) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785.
(49) Frisch, M. J.; et al. Gaussian 03, Rev B.01, Gaussian, Inc.,
Pittsburgh, 2003 (full reference given in Supporting Information).
(50) Casida, M. E. Time-Dependent Density Functional Response Theory
for Molecules. In Recent AdVances in Density Functional Methods, Part I;
Chong, D. P., Ed.; World Scientific: Singapore, 1995; p 155.
(51) Ca. 135 °C are required to volatilize (2)2. Under these conditions,
the dimer appears to undergo partial dissociation before it is trapped in the
Ar matrix.
4.2. Pulse Radiolysis. Pulse radiolysis experiments were carried
out with high energy (6 MeV) 17 ns electron pulses generated from
a linear electron accelerator. The dose absorbed per pulse was
5318 J. Org. Chem., Vol. 71, No. 14, 2006

Anthralin: Primary Products of Its Redox Reactions
determined with an N2O-saturated aqueous solution of KSCN (0.01
M), assuming G((SCN)2•-) ) 6.0 and ꢀ((SCN)2•-) ) 7600 M-1
cm-1 (G represents the yield of radicals per 100 eV of energy
absorbed and ꢀ is a molar extinction coefficient at 475 nm). The
dose delivered per pulse was 5-80 Gy. The concentration of the
solute was kept in the range of (0.5-10) × 10-3 M. Details of the
pulse radiolysis system are given elsewhere.44,45
The radiolysis of 2-methyltetrahydrofuran leads to the formation
of solvated electrons (es-) and the radical cations of solvent
molecules (reaction 3).
placed in a optical dewar with liquid nitrogen or in a liquid nitrogen-
cooled cryostat where any temperature between 77 and 150 K can
be maintained by controlled heating. Ionization was effected by
irradiation with 4 µs electron pulses from the linear accelerator
(Lodz) or by irradiation in a 60Co γ-source (Fribourg).
In the rare gas matrix isolation experiments37 ground crystals of
1 or (2)2 were placed in heatable tube attached to the cryostat. From
there they were evaporated very slowly at 65 (1) or 135 °C,
respectively, for the dimer and co-deposited onto a CsI window
held at 19 K with a 8:1 mixture of Ar and N2 containing 1 mol %
of CH2Cl2 (which acts as an electron scavenger). After taking
reference spectra, the samples were exposed to 90 min of X-
irradiation from a tungsten anode tube operated at 40 kV/40 mA.
4.4. Quantum Chemical Calculations. The geometries of all
species were optimized by the B3LYP density functional method47,48
as implemented in the Gaussian 03 suite of programs49 using the
6-31G* basis set for neutral and cationic species, 6-31+G* for
(radical) anions. Relative energies, including ZPE corrections, were
calculated at the same levels.
C5H10O f C5H10O•+ + es-
(3)
The solvated electrons react with the solute molecules (A) to
give the corresponding radical anions (reaction 4), while the ionized
solvent molecules easily transfer a proton onto neighboring neutral
molecules (reaction 5).22
es- + A f A•-
(4)
(5)
Electronic transitions were predicted by density functional-based
time-dependent response theory.50 We used the formulation of TD-
DFT implemented in the Gaussian 03 program,49 together with the
same functional and Gaussian basis sets as mentioned above.
C5H10O•+ + C5H10O f •C5H9O + C5H11O+
The products of this latter process can act as acids or as hydrogen
atom donors/acceptors. Under matrix conditions these bimolecular
reactions are, however, largely suppressed. They can be observed
only upon softening of a matrix on its annealing.
Acknowledgment. This work was supported by grant from
the Ministry of Science and Informatization (No. 4/T09A/020/
24) and the Swiss National Science Foundation (project 200020-
105217). We thank Prof. Jerzy Gebicki (Technical University
of Lodz) for participation in the early stages of this project and
for helpful discussions.
For the attempted generation of radical cations, solutions of the
anthralin in alkyl halides (2-chlorobutane or a 1:1 mixture of CFCl3
and BrF2C-CF2Br) were used as these solvents dissociatively
capture electrons ejected from the solvent molecules, while the
positive charge is transferred to solute molecules of lower ionization
potential.22 Glassy matrices of 1-butyl-3-methylimidazolium hexaflu-
orophosphate were used for the same purpose as this ionic liquid
has proven suitable for the generation of radical cations.46
4.3. Cryogenic Measurements. Glassy samples of the MTHF,
2-chlorobutane, the above Freon mixture, or 1-butyl-3-methylimi-
dazolium hexafluorophosphate/chloroform were prepared by im-
mersing room-temperature solutions in liquid nitrogen in special
cryogenic cuvettes. The samples were 0.5-3 mm thick and were
Supporting Information Available: Pictures of the orbitals
involved in the electronic transitions of 1•-, 2-, and 2• listed in
Tables 1-3 and lowest energy pathway for the enolization of 1•+.
Cartesian coordinates and absolute energies, including thermal
corrections, for all species discussed in this work from B3LYP
calculations. This material is available free of charge via the Internet
at http://pubs.acs.org.
JO060622O
J. Org. Chem, Vol. 71, No. 14, 2006 5319