General patterns in the photochemistry of pregna-1,4-dien-3,20-diones

Article abstract of DOI:10.1021/jo034070a

The photochemistry of six pregna-1,4-dien-3,20-diones has been compared and found to involve both the cyclohexadienone moiety in ring A and the isolated ketone at C-20. The two reactions take place proportionally to the fraction of light absorbed by each chromophore. The cross-conjugated ketone absorbs predominantly or exclusively at both 254 and 366 nm and undergoes the lumi rearrangement to bicyclo[3.1.0]hex-3-en-2-one. The quantum yield of the reaction diminished somewhat with increasing λ_exc, e.g., for prednisolone Φ_{254 nm} = 0.42, Φ_{366 nm} = 0.3. A much stronger lowering is caused by halogen substitution in position 9 (by a factor of 3 for F, > 50 for Cl), apparently due to a shortened triplet lifetime caused by heavy atom effect. At 310 nm, both chromophores absorb to a comparable degree and both may react. The reaction at C₂₀ ketone involves either quite efficient α-cleavage (C₁₇-C₂₀) for compounds bearing an acetal or hydroxyl function at C₁₇ or less effective (by a factor of ca. 10) hydrogen abstraction from the 18-methyl group in the other cases (finally resulting in Norrish II fragmentation or Yang cyclization). The results allow generalizing how the substitution pattern surrounding each chromophore affects the photoreactivity at that site and the competition between the two modes, allowing predicting the photochemistry of this family of antiinflammatory drugs.

Full text of DOI:10.1021/jo034070a

Gen er a l P a tter n s in th e P h otoch em istr y of
P r egn a -1,4-d ien -3,20-d ion es
Andrea Ricci, Elisa Fasani, Mariella Mella, and Angelo Albini*
Dep. Organic Chemistry, University of Pavia, Via Taramelli 10, 27100 Pavia, Italy
albini@chifis.unipv.it.
Received J anuary 21, 2003
The photochemistry of six pregna-1,4-dien-3,20-diones has been compared and found to involve
both the cyclohexadienone moiety in ring A and the isolated ketone at C-20. The two reactions
take place proportionally to the fraction of light absorbed by each chromophore. The cross-conjugated
ketone absorbs predominantly or exclusively at both 254 and 366 nm and undergoes the “lumi”
rearrangement to bicyclo[3.1.0]hex-3-en-2-one. The quantum yield of the reaction diminished
somewhat with increasing λexc, e.g., for prednisolone Φ254 nm ) 0.42, Φ366 nm ) 0.3. A much stronger
lowering is caused by halogen substitution in position 9 (by a factor of 3 for F, >50 for Cl), apparently
due to a shortened triplet lifetime caused by heavy atom effect. At 310 nm, both chromophores
absorb to a comparable degree and both may react. The reaction at C20 ketone involves either quite
efficient R-cleavage (C17-C20) for compounds bearing an acetal or hydroxyl function at C17 or less
effective (by a factor of ca. 10) hydrogen abstraction from the 18-methyl group in the other cases
(finally resulting in Norrish II fragmentation or Yang cyclization). The results allow generalizing
how the substitution pattern surrounding each chromophore affects the photoreactivity at that
site and the competition between the two modes, allowing predicting the photochemistry of this
family of antiinflammatory drugs.
In tr od u ction
254 nm of prednisolone (11â,17R,21-trihydroxy-pregna-
1,4-dien-3,20-dione, 1, and similarly of prednisone, with
a third ketone functionality at C11 replacing the C11-â-
hydroxy group, as well of their 21-acetates) in dioxane
solution was the “lumiketone” rearrangement6 of the
cyclohexadienone moiety to yield products such as 5,
characterized by the bicyclo[3.1.0]hex-3-en-2-one struc-
ture (Scheme 1).
Pregna-1,4-diene-3,20-diones are largely used as topic
antiinflammatory drugs, e.g., against solar light induced
erythema.1 However, these compounds may also be
phototoxic.2 From the photochemical point of view, these
steroid derivatives are interesting because they bear two
spatially separated chromophores, viz., the cross-conju-
gated cyclohexadienone moiety in ring A and an isolated
ketone functionality in C20. Photochemical work on these
compounds includes a number of studies in the solid
state3 and in solution.4,5 Williams and co-workers5 found
that the only photoprocess occurring upon irradiation at
In a recent study on some fluorinated pregna-1,4-diene-
3,20-diones (2-4, Scheme 2), we found,7 however, a 2-fold
photochemistry. Irradiation at 254 or 366 nm, where the
cross-conjugated hexadienone moiety absorbs predomi-
nantly or exclusively, the incident light caused the
“lumirearrangement” of ring A. On the contrary, irradia-
tion at 310 nm, where the absorption by the two chro-
mophores is similar, caused the Norrish I photocleavage
of the C20 ketone. Apparently, no intramolecular energy
transfer occurred between the two chromophores.
* To whom correspondence should be addressed. Phone: +39-0382-
507316. Fax: +39-0382-507323.
(1) Ortel, B.; Sivayathorn, A.; Honigsmann, H. Dermatology (Basel)
1989, 178, 39.
(2) (a) Suzuki, T.; Kato, T.; Kitagachi, T.; Ono, M.; Shirawaka, K.;
Nagata, M.; Konishi, R. J . Toxicol. Sci. 1996, 21, 475. (b) Keknes, A.;
J ahn, P.; Lange, L. J . Am. Acad. Dermatol. 1993, 28, 786. (c)Uchiyama,
H.; Tanaka, T.; Uehara, N.; Nakamura, M.; Tsuji, M.; Shinomiya, M.;
Tanaka, H. J . Toxicol. Sci. 1996, 21, 475.
(3) (a) Reisch, J .; Enkel, G.; Ekiz-Guecer, N.; Nolte, G. Liebigs Ann.
Chem. 1992, 63. (b) Reisch, J .; Topaloglu, Y.; Henkel, G. Acta Pharm.
Tech. 1986, 32, 115. (c) Taka`cs, M.; Ekiz-Guecer, N.; Reisch, J .;
Gergely-Zobin, A. Pharm. Acta Helv. 1991, 66, 137. (d) Ekiz-Guecer,
N.; Reisch, J .; Nolte, G. J . Pharm. Biopharm. 1991, 37, 234.
(4) (a) Arakawa, Y.; Fukaja, C.; Yamanouchi, K.; Yokoyama, K.
Yakagaku Zasshi 1985, 105, 1029. (b) Hidaka, T.; Huruumi, S.;
Tamaki, S.; Shiraishi, M.; Minato, H. Yakagaku Zasshi 1980, 100, 72.
(c) Thoma, K.; Kerker, R.; Weissbach, C. Pharm. Ind. 1987, 49, 961.
(d) Ogata, M.; Noro, Y.; Yamada M.; Tahara, T.; Nishimura, T. J .
Pharm. Sci. 1998, 87, 91.
Establishing whether such a wavelength-dependent
photobehavior is common to all of the pregna-1,4-diene-
3,20-diones and to what extent it could be directed by
means of a judicious choice of the substituents is a target
(6) (a) Schuster, D. I. Acc. Chem. Res. 1978, 11, 65. (b) Shaffner,
K.; Demuth, M. In Rearrangements in the Ground and Excited State;
De Mayo, P., Ed., Academic Press: New York, 1980; Vol. 3, p 281. (c)
Barton, D. H. R. Helv. Chim. Acta 1959, 42, 1604. (d) Caine, D. In
CRC Handbook of Photochemistry and Photobiology; Horspool, W. M.;
Song, P. S., Eds.; CRC Press: Boca Raton, FL, 1995; p 701. (e) Schultz,
A. G. In CRC Handbook of Photochemistry and Photobiology; Horspool,
W. M.; Song, P. S., Eds.; CRC Press: Boca Raton, FL, 1995; p 716.
(7) Ricci, A.; Fasani, E.; Mella, M.; Albini, A. J . Org. Chem. 2001,
66, 8086.
(5) (a) Williams, J . R.; Moore, R. H.; Blount, J . F. J . Am. Chem. Soc.
1979, 15, 5019. (b) Williams, J . R.; Moore, R. H.; Li, R.; Weeks, C. M.
J . Org. Chem. 1980, 45, 2324.
10.1021/jo034070a CCC: $25.00 © 2003 American Chemical Society
Published on Web 05/03/2003
J . Org. Chem. 2003, 68, 4361-4366
4361

Ricci et al.
TABLE 1. Qu a n tu m Yield for th e P h otoch em ica l
Rea ction s of P r egn a d ien d ion es
SCHEME 1
overall reaction at the
reaction at
irradiation quantum cross-conjugated C20-ketone
steroid wavelength yield (Φr)
dienone (Φ)
(Φ)
1
1
1
2
2
2
3
3
3
254 nm
310 nm
366 nm
254 nm
310 nm
366 nm
254 nm
310 nm
366 nm
254 nm
310 nm
254 nm
310 nm
366 nm
254 nm
310 nm
366 nm
0.423
0.305
0.302
0.181
0.450
0.060
0.051
0.310
0.031
0.052
0.110
0.008
0.083
e0.0001
0.052
0.073
0.029
5 (0.296)
5 (0.183)
5 (0.181)
6 (0.135)
6 (0.049)
6 (0.030)
7 (0.043)
7 (0.015)
7 (0.015)
10 (0.041)
10 (0.011)
8 (0.005)
8 (0.293)
9 (0.005)
9 (0.263)
4
4
11 (0.005)
11 (0.071)
12
12
12
13
13
13
14a , b (0.001)
14a , b (0.042)
SCHEME 2
15 (0.039)
15 (0.015)
15 (0.014)
16+17 (0.033)
SCHEME 3
photoprocess, though with a somewhat lower quantum
yield (Φ 0.3) and isolated yield in the preparative
irradiations (60%).
The quantum yields at the three λirr are reported in
Table 1, where these are compared with the data for the
previously considered fluorinated glucocorticosteroids, 2
(triamcinolone acetonide), 3 (fluocinolone acetonide), and
4 (flumethasone) with the characteristic wavelength-
dependent photochemistry.
We then explored the effect of a different halogen by
studying the photochemistry of 9-chloro-11â,17R,21-tri-
hydroxy-16R-methyl-pregna-1,4-dien-3,20-dione-17,21-
dipropionate (beclomethasone 17,21-dipropionate 12),
that bore a chlorine atom (rather than a fluorine as in
compounds 2-4) in position 9 and further had the
hydroxyl groups in C17 and C21 protected as the propi-
onate ester. The photodegradation of compound 12 at 254
or 310 nm occurred with different quantum yields (0.008
and 0.08 respectively) but afforded the same mixture of
two photoproducts (in 1:4 ratio at both wavelengths,
Scheme 3). Spectroscopic analysis, in particular NMR
data, indicated that the steroid skeleton was unaffected,
but both the C20 ketone and the 18-methyl signals were
lacking. These data, along with the appearance of a new
methylene carbon (see the Experimental Section for
details and further information) supported the structure
not only for basic research but also for practical applica-
tion in view of predicting and controlling the photolability
and phototoxicity of such semisynthetic drugs on the
basis of their structure. Therefore, we presently report
a product and quantum yield study on prednisolone 1 and
some of its derivatives with strategic changes in the
substitution pattern adjacent to both chromophores, the
cross-conjugated ketone in ring A, and the isolated ketone
at position C20.
Resu lts
As mentioned in the Introduction, the photochemistry
of prednisolone 1 has already been reported5 to form the
lumiderivative 5 at 254 nm, although the quantum yield
of the reaction has not been measured. We repeated the
irradiation in argon-flushed acetonitrile and confirmed
that compound 5 is essentially the only product (70%
isolated yield). Separate experiments at low (e20%)
conversion, to avoid secondary photolysis, allowed mea-
suring a quantum yield of 0.42. When the irradiation was
carried out at 310 and 366 nm, representative of the
UV-B and, respectively, the UV-A component of UV
light, rearrangement to 5 was again essentially the only
4362 J . Org. Chem., Vol. 68, No. 11, 2003

Photochemistry of Pregna-1,4-dien-3,20-diones
SCHEME 4
Discu ssion
In our previous study7 of pregnadiendiones 2-4, we
showed that both of the separated chromophores on these
derivatives, the cross conjugated dienone in ring A and
the ketone at C20, react independently from the respec-
tive triplet states, and intramolecular T-T energy trans-
fer does not compete with fast local reaction, as expected
from theory. As a result, it is the fraction of light absorbed
by each chromophore at each wavelength that determines
the site of the reaction. For derivative 2, it has been
measured7 that the fraction of light absorbed by the
cyclohexadienone is 0.992 at 254 and 1 at 366 nm,
whereas at 310 nm, it is only 0.57. A substantial amount
of the incident light is thus absorbed by the isolated
ketone at this wavelength (0.43). This essentially holds
also for the present derivatives and one would expect the
rearrangement at 254 and 366, with some degree of
competition by reaction at the ketone functionality at C20
at 310 nm. The aim of the present investigation was to
explore the structural effects on the efficiency of the
photoreactions at each chromophore and to rationalize
the overall chemical path followed as a result of their
competition. This would also allow arriving at predicting
the photoreactivity (and indirectly the phototoxicity) or
these family of compounds, all of which are used as drugs.
“Lu m ik eton e” Rea r r a n gem en t. This rearrangement
and its role in the photochemistry of steroids has been
largely investigated in the previous literature.6 On mono-
cyclic cyclohexa-2,5-dienones, it occurs with quantum
yields close to unitary (0.75-1),8 although it has been
reported that some substituents affect the efficiency of
the photoprocess, a fact mainly attributed to a change
in the stability of the suggested zwitterionic intermedi-
ates of the rearrangement.8-10 In polycyclic derivatives,
and in particular with steroid cyclohexadienones, the
quantum yield is slightly lower. A value of 0.58 has been
reported for ∆1-dehydrotestosterone,11 a compound strictly
related to the presently studied ketones.
In keeping with this result, we measured a quantum
yield value of 0.42 for the reaction of prednisolone 1 in
argon flushed acetonitrile at 254 nm, and the only process
occurring was the “lumi” rearrangement. A conspicuous
substituent effect was observed. Thus, under the same
conditions, the quantum yield of this rearrangement
decreased to 0.18 with 2 (a 9-monofluoro derivative) and
to ca. 0.05 with 3, 4, and 13 (6R,9-difluoro derivatives).
With the 9-choro derivative 12, the overall quantum yield
of reaction was only 0.008 at 254 nm. The main contribu-
tion to the reaction, however, was fragmentation of the
ketone in position 20, as shown by the formation of
products 14a ,b even at this wavelength, and rearrange-
ment to the nondetected lumiketone must have a still
lower Φ value.
of the diastereoisomeric 20-hydroxy-18,20-cyclopregna-
1,4-dien-3-ones 14a and 14b for these products. The
(cumulative) isolated yield in preparative experiments
was 50% at 310 nm and 10% at 254 nm. Surprisingly,
no trace of the expected “lumiproduct” was detected at
either 254 or 310 nm, or indeed at 366 nm, where
compound 12 showed to be virtually photostable (Φ < 1
× 10-3).
A further variation was explored with 6R,9-difluoro-
21-hydroxy-16R-methyl-pregna-1,4-diene-3,20-dione-21-
valerate (13, diflucortolone 21-valerate). This pregna-
diendione has the same fluorination pattern as compounds
3 and 4 but differed by lacking an oxygenated functional-
ity at C17, while maintaining the one at the C21 (hydroxyl
group esterified as valerate). This compound showed
again a wavelength-dependent photoreactivity (Scheme
4). Thus, irradiation at 254 nm in argon-flushed aceto-
nitrile (Φ ) 0.05) afforded in high yield a single product
isomeric with the starting material, which was recognized
as the lumiderivative 15 (75%, see the Experimental
Section for the structural attribution). When the irradia-
tion was effected at 366 nm, the quantum yield of the
reaction was somewhat lower (0.03) and 15 was again
the main product (70%), although it underwent further
photoreaction at a longer irradiation time.
On the other hand, irradiation of 13 at 310 nm (Φ )
0.07) gave, besides a certain amount of “lumiketone” 15
(30%), two new compounds, which were separated and
characterized by analytical and spectroscopic methods.
Key evidence from the NMR spectra was the fact that
rings A-C and the valerate chain were unaffected,
whereas the 18-methyl signal was missing in both of
them. The major one (30%) maintained the ketone at C20
unaffected and showed an olefin methylene signal,
whereas the minor one (5%) lacked the carbonyl C20
signal and showed the presence of an aliphatic methylene
group. These data, along with further evidence (see the
Experimental Section for details) allowed the assignment
of the structure 16 to the major product and 17 to the
minor one.
The simplest rationalization of this remarkable de-
crease induced by halogens is a heavy atom effect that
(8) Schuster, D. I.; Patel, D. I. J . Am. Chem. Soc. 1968, 90, 5145.
(9) (a) Schuster, D. I.; Abraitis, Y. Y. Chem. Commun. 1969, 419.
(b) Schuster, D. I.; Patel, D. I. J . Am. Chem. Soc. 1968, 90, 5137.
(10) (a) Caine, D.; Brake, P. F.; de Bartdeleben, J . F.; Dawson, J . B.
J . Org. Chem. 1973, 38, 967. (b) Broca, C. A. J . Org. Chem. 1988, 53,
575. (c) Stille, C. J . K.; Retting, T. A.; Kneunerle, E. W. J . Org. Chem.
1976, 41, 2950. (d) Zimmerman, H. E.; Pasteris, R. J . J . Org. Chem.
1980, 45, 4864.
The quantum yields data for derivative 12 and 13,
obtained at low conversion at the three wavelengths, are
likewise reported in Table 1.
(11) Schuster, D. I.; Barringer, W. C. J . Am. Chem. Soc. 1971, 93,
73.
J . Org. Chem, Vol. 68, No. 11, 2003 4363

Ricci et al.
SCHEME 5
less rigid structure diminishes the optimal alignment of
the C-O bond with the fragmentable C-C bond. It
should be noted that a hydroxyl group at freely rotating
C21 is present in most of these compounds, but the
alternative C20-C21 Norrish I cleavage is never observed,
a fact that can be similarly attributed to the lack of the
favorable alignment.
The determining role of a rigid acetal or, to a lower
degree, of a hydroxyl or alkoxy function in promoting the
Norrish type I cleavage at C17 can be appreciated by the
fact that such a fragmentation does not occur with 12
(which has a propionyloxy group, apparently unable to
reach the correct configuration and/or a poorer electron-
donor) and with 13 which has no oxygenated functions
at C17. Different reactions of the nπ* ketone take place
with this compounds, viz., hydrogen abstraction from the
close-lying 18-methyl group followed by cyclization (Yang
reaction, path b in Scheme 5) with the first, and to a
lesser degree, with the latter ketone, or by fragmentation
of the C13-C18 bond (path c, main process with the latter
ketone). The initial abstraction is facilitated in both
compounds by the fact that C20 ketone and 18-methyl
groups, both with â orientation, are held in a suitable
conformation by the rigid skeleton. Cyclization of the
biradical to hydroxycyclobutane has been reported by
Shaffner et al.13 for some pregnan-20-ones lacking further
photoreactive moieties in the structure and bearing a
hydrogen atom in position 17R. Alternatively, â-scission
of the 1,4-biradical (Norrish type II photoreaction) leads
to the open chain alkene 16, the major process from
ketone 13 (also in this case, there is a precedent for a
simpler steroid).14
Summing up, a hydroxyl or acetal function in 17 (but
not in 21) exerts a bias toward R-fragmentation. In the
other cases, intramolecular hydrogen abstraction from
the γ-position leading to Norrish type II fragmentation
or cyclization results. The first process is more efficient
and occurs with a quantum yield of ∼0.3 (for acetals 2
and 3) and ∼0.07 (for alcohol 4) at 310 nm, which, taking
into account that the absorption by the ketone at this
wavelength is around 40%, indicates an intrinsic ef-
ficiency of cleavage of 0.8 and 0.2, respectively. On the
contrary, hydrogen abstraction is characterized by a
quantum yield of ∼0.04 at this wavelength and, thus,
an intrinsic efficiency of ∼0.1
shortens the triplet lifetime by increasing the rate of ISC
to S0 (by a factor of 3 with a 9-F, 9 with 6R,9-F2, of >50
with 9-Cl). Inspection of MM2 models shows that the
C-X bond in 9 is collinear with the π system of the
cyclohexadienone and ideally suited for exerting such an
effect. Dramatic effects of halogens on T1-S0 ISC rate
have been well documented with aromatics,12 and al-
though there is no close precedent for dienones, it appears
reasonable that the favorable geometric position makes
halogen induced kISC competitive with kr (for 4,4-
dimethylcyclohexa-2,5-dienone the rate of rearrangement
has been estimated kr ) 7 × 108 mol-1 s-1 by quenching
experiments).8 The further decrease in quantum yield
caused by introducing a second fluorine in position 6 may
be due to the same reason or to a destabilization of the
zwitterionic intermediate caused by the electron-with-
drawing ability of the halogen.
The second trend that has been noted is a decrease by
a factor of 2 from 254 to 366 nm in the rearrangement
quantum yield upon increasing λirr. That occurs with all
of the steroids examined. We are not aware of previous
systematic investigations of the effect of adsorbed radia-
tion on the lumirearrangement, but the present data
suggest that either ISC (Sn-T1) changes significantly for
higher-lying singlets or that rearrangement from T1
encounters a significant energy barrier, at least when the
cyclohexadienone moiety is incorporated in a rigid skel-
eton as in the present case.
Com p etition betw een th e Tw o P h otop r ocesses.
The competition between the two photoprocesses is
determined by the absorption by each chromophore and
the efficiency of the reaction at each site. The cyclohexa-
dienone moiety absorbs all of the light at 366 and >99%
of it at 254, and 50% or more at 310 nm. Thus, with a
halogen-free pregnadiendione such as 1, the lumirear-
rangement occurs efficiently with no measurable com-
petition by reaction at the isolated ketone (here not
particularly favored, because there is an OH group, not
an acetal at C17). A wavelength-independent photochem-
istry thus results, with some decrease in the quantum
yield with decreasing excitation energy.
Rea ction a t th e Isola ted Keton e. Reaction at C20
ketone occurs for most of the examined pregnadiendiones
upon irradiation at 310 nm and also in this case there is
a strong dependence on the substitution close to the
moiety involved. As previously found with compounds 2
and 3, both of which bear an acetalic function at C16-
C17, the main process at 310 nm is the Norrish type I
cleavage of the C17-C20 bond (path a in Scheme 5) with
Φ ) 0.41 and 0.3, respectively. With 4, bearing a hydroxyl
group at C17, Φ is lower (0.11), reasonably because the
(13) (a) Buchschacher, P.; Cereghetti, M., Wehrli, H.; Schaffner, K.;
J eger, O. Helv. Chim. Acta 1959, 42, 2122. (b) Cereghetti, M.; Wehrli,
H.; Schaffner, K.; J eger, O. Helv. Chim. Acta 1960, 43, 354. (c) Wehrli,
H.; Cereghetti, M.; Schaffner, K.; Urech, J .; Vischer, E. Helv. Chim.
Acta 1961, 44, 1927.
(14) Suginome, H.; Nakayama, Y. J . Chem. Soc., Perkin Trans. 2
1992, 1843.
(12) (a) Ermolaev, V. L.; Svitashev, K. K. Opt. Spectr. 1959, 7, 399.
(b) Castro, G.; Hochstrasser, R. M. J . Chem. Phys. 1966, 44, 412. (c)
McGlynn, S. P. Chem. Rev. 1958, 58, 1113. (d) McGlynn, S. P.; Azumi,
T.; Kinoshita, M. Molecular Spectroscopy of the Triplet State; Prentice
Hall: Engelwood Cliffs, New J ersey, 1969. Birks, J . B. Photophysics
of Aromatic Molecules, Wiley-Interscience: London, 1970.
4364 J . Org. Chem., Vol. 68, No. 11, 2003

Photochemistry of Pregna-1,4-dien-3,20-diones
Compound 12 represent the opposite case. In this
instance, the 9-chloro atom so effectively suppresses the
lumirearrangement that the photochemistry is again
wavelength-independent, but this time it is the one of
the isolated ketone. Consistently with the proposed
mechanism, the quantum yield is much higher at 310
nm than at 254 nm, because the reactive chromophore
absorbs a much larger fraction of light at the former
wavelength (and indeed, no reaction takes place at 366
nm where the isolated ketone is transparent). In all of
the other cases (2-4 and 13) a wavelength-dependent
photochemistry takes place, with reaction at both of the
photolabile moieties proportional to the relative absorp-
tions by each chromophore.
factors when planning the improvement of pharmacologi-
cal effect and the lessening of undesirable side effects in
semisynthetic glucocorticosteroids. Thus, a halogen atom
in position 9, found to increase the antiinflammatory
activity,16 also reduces the possible phototoxic effects due
to the cyclohexadienone photoreactivity, as it partially
quenches the rearrangement of this moiety. On the
contrary, introduction of an oxygenated function in
position 17, found to reduce mineralcorticoid activity, an
undesired side effect, unfortunately enhances the C20
ketone photoreaction and, thus, the formation of the
radical, which may induce a specific phototoxicity (though
relevant only upon UV-B irradiation).
For the studied pregandiendiones, the overall quantum
yield at 254 nm is consistently larger than Φ366, because
at both of the wavelengths it is the cross conjugated
ketone that absorbs almost exclusively the light and the
efficiency decreases by shifting λirr to the red (except for
ketone 12 that does not react at this moiety). On the
contrary, Φ310 may be lower or higher than Φ254 according
to the efficiency of the reaction of the isolated ketone that
contributes significantly in this case.
Exp er im en ta l Section
Diflucortolone-21-valerate (13) and beclomethasone-17,21-
dipropionate (12) were kindly supplied by Farmabios, Gropello
Cairoli, Italy. Their purity was checked by HPLC and NMR
spectroscopy. Prednisolone (1) was a commercial sample.
Spectroscopic grade solvents were used for irradiations.
P r ep a r a tive P h otoch em ica l Rea ction s. Irradiation at
254 nm was performed in an immersion well apparatus fitted
with a quartz filtered 20 W low-pressure mercury arc lamp.
Before irradiation, purified argon was flushed under stirring
for 2 h, and a low gas flux was maintained during the reaction.
For the irradiations at 310 and 366 nm, the solution was
distributed in a series of serum-capped quartz tubes, nitrogen
flushed, and then externally irradiated in a multilamp ap-
paratus fitted with eight 15 W phosphor coated lamps with
the required emission band.
The reactions were monitored by means of HPLC (Hypersil
ODS-2, 4,6 × 250 mm, 5 µm column, eluting with acetonitrile/
water mixtures) and TLC (eluting with cyclohexane-ethyl
acetate mixtures).
When the desired conversion was reached, the solvent was
rotary evaporated and the products were purified by recrys-
tallization or by preparative chromatographic separation with
a 0.04-0.063 silica gel column eluting with the above solvent
mixtures.
Con clu sion s
The main finding in this work is the dramatic decrease
of the quantum yield of the lumiketone rearrangement
by the effect of suitably situated fluorine or chlorine
atoms through heavy atom effect (dropping to <1/50 with
chlorine in position 9). This removes from the pregna-
diendiones most part of the photoreactivity. As for the
isolate ketone at C20, this is protected from photoreaction
by the strong absorption by the cross-conjugated chro-
mophore through an inner filter effect, because it has a
much lower absorption over a large part of the spectrum.
When it shows up (irradiation at 310 nm), reaction at
this site is again substituent-dependent. An acetal or
hydroxyl function in position 17 induces R-cleavage
(intrinsic efficiency 0.2-0.8), whereas hydrogen abstrac-
tion from the methyl in position 18 occurs with a lower
intrinsic efficiency, ∼0.1, when no oxygenated function
is present and assists the previous reaction.
The conclusions presented here allow to predict the
type and efficiency of the photoreaction(s) occurring in a
steroid of this family on the basis of the above-discussed
key structural features in the vicinity of the two chro-
mophores. This should allow us to propose hypotheses
on the mechanism of the phototoxic behavior of these
drugs. When a concerted process such as the lumiketone
rearrangement takes place, toxicity, if any, must be
related to the formation of toxic end-photoproducts. When
a process involving homolytic fragmentation such as the
Norrish type I cleavage or hydrogen atom abstraction
occurs, the phototoxicity may be related not only to the
formation of toxic photoproducts but also to the role of
alkyl radicals formed as transients or of hydroperoxyl
radicals formed from them by oxygen addition (such
radicals have been demonstrate to damage DNA).15
The substituent effect on the photobehavior of the
chromophores should be considered along with other
The characterization of the new compounds was based on
analytical and spectroscopic techniques. [RD] were measured
at 20 °C in MeOH (5 mg/mL), and NMR spectra were recorded
on a 300 MHz spectrometer.
P h otolysis of P r ed n isolon e (1) a t 254 n m . A solution of
compound 1 (0.4 g, 1.1 mmol) was irradiated in acetonitrile
(220 mL) for 1 h and a 90% conversion was reached. Recrys-
tallization from ethyl acetate of the residue gave 250 mg of
the previously reported5 11â,17R,21-trihydroxy-1,5-ciclopregn-
3-ene-2,20-dione 5. Colorless crystals; mp 233-234 °C (AcOEt);
[R]D ) -73,2°; IR: 3461, 2929, 2846, 1710, 1670, 1100 cm-1
.
1H NMR (CDCl3): δ 0.9 (s, 3H, Me-18), 1.28 (s, 3H, Me-19),
2.3 (bs, 1H, H-1), 1.0-2.8 (m, 15H); 4.46 (m, 1H, H-11), 4.32
and 4.67 (AB system, J ) 20 Hz, 2H, H-21), 5.9 (d, J ) 6 Hz,
1H, H-3), 7.28 (d, J ) 6 Hz, H-4).
P h otolysis of Beclom eth a son e-17,21-d ip r op ion a te (12)
a t 310 n m . A solution of compound 12 (1 g, 1.9 mmol) was
irradiated at 310 nm in acetonitrile (384 mL) for 10 h until a
50% conversion was reached. Chromatographic separation
gave a mixture of two isomers (247 mg, 50%). These were
identified as the two isomeric cyclobutanols 14a and 14b,
mainly on the basis of NMR (no carbonyl group, presence of a
hydroxyl group a new methylene replacing a methyl group).
Signals superimposition discouraged investigating the C-20
configuration in this case, and the stereochemistry was
assigned by comparing the NMR with the analoguous deriva-
tive 18 (see below) which was isolated as pure compound. IR:
3300, 1730, 1650 cm-1
.
(15) Adam, W.; Arnold, M. A.; Saha-Mo¨ller, C. R. J . Org. Chem.
2001, 66, 597.
(16) Foye, W. O.; Lemke, L.; Williams, T. Principles of Medicinal
Chemistry; Lea&Febiger Books: Philadelphia, PA, 1995.
J . Org. Chem, Vol. 68, No. 11, 2003 4365

Ricci et al.
9-Ch lor o-16r-m eth yl-11â,17r,20,21-tetr a h yd r oxy-18,20-
cyclop r egn a -1,4-d ien -3-on e 17,21-d ip r op ion a te (14a ). 1H
NMR(CDCl3): δ 1.15 (t, J ) 7, 3H, Me), 1.2 (t, J ) 7, 3H, CH3),
1.25 (d, J ) 7, 3H, CH3), 1.6 (s, 3H, Me), 2.3-2.4 (m, 4H, CH2),
1.2-2.8 (m, 15H), 4.55 (m, 1H, H-11), 4.5 and 4.7 (m, 2H, AB
system, J ) 18, CH2-21), 6.05 (bt, J ) 1.5, 1H, H-4), 6.4 (dd,
J ) 10, 1.5, 1H, H-2), 7.2 (d, J ) 10, 1H, H-1). 13C NMR
(CDCl3): δ 8.9 (CH3), 9.1 (CH3), 18.9 (CH3), 24.2 (CH2), 24.2
(CH3), 26.6 (CH2), 27.5 (CH2), 30.4 (CH2), 31.1 (CH2), 34.3 (CH),
37.7 (CH2), 39.2 (CH2), 40.3 (CH), 46.5, 48.4 (CH), 49.6 (C-
10), 67.9 (CH2-21), 75.3 (CH-11), 77.7 (C-9), 82.2 (C-20), 94.4
(C-17), 124.8 (CH-4), 129.3 (CH-2), 151.7 (CH-1), 165.5 (C-5),
174.7 (COOR), 174.9 (COOR), 186.3 (C-3).
9-Ch lor o-16r-m eth yl-11â,17r,20,21-tetr a h yd r oxy-18,20-
cyclop r egn a -1,4-d ien -3-on e 17,21-d ip r op ion a te (14b). 1H
NMR (CDCl3): δ 1.18 (t, J ) 7, 3H, Me), 1.22 (s, 3H, CH3),
1.25 (d, J ) 7, 3H, Me), 1.65 (s, 3H, Me), 2.3-2.4 (m, 4H, CH2),
1.2-2.8 (m, 15H), 4.4 (m, 1H, H-11), 4.3 and 5.1 (m, 2H, AB
system, J ) 18, CH2-21), 6.05 (bt, J ) 1.5, 1H, H-4), 6.42 (dd,
J ) 10, 1.5, 1H, H-2), 7.22 (d, J ) 10, 1H, H-1). 13C NMR
(CDCl3): δ 8.5 (CH3), 8,7 (CH3), 18.6 (CH3), 24.2 (CH3), 27.2
(CH2), 27.3 (CH2), 27.4 (CH2), 28.2 (CH2), 31.2 (CH2), 34.3 (CH),
37.4 (CH2), 39.2 (CH2), 40.3 (CH), 45.2, 48.0 (CH), 49.8, 63.6
(CH2-21), 75.6 (CH-11), 77.9 (C-9), 86.3 (C-20), 87.0 (C-17),
124.8 (CH-4), 129.3 (CH-2), 151.9 (CH-1), 165.6 (C-5), 173.3
(COOR), 174.7 (COOR), 186.3 (C-3).
1H NMR (CDCl3): δ 0.93 (t, 3H, J ) 7 Hz, CH3), 0.98 (d, 3H,
J ) 7 Hz, CH3), 1.45 (s, 3H, CH3, Me-19), 1.4 (m, 2H), 1.65
(m, 2H, J ) 7 Hz), 2.46 (t, 2H, J ) 7 Hz), 4.3 (m, 1H, H-11),
4.62 (AB system, 2H, H-21), 5.0 and 5.05 (bs, 2H, CH2-18),
5.3 (dddd, J ) 50, 13, 8, 2 Hz, 1H, H-6), 6.4 (d, J ) 10 Hz, 1H,
H-2), 6.45 (bs, 1H, H-4), 7.14 (d, J ) 10 Hz, 1H, H-1), 1.3-2.8
(m, 11H + OH). 13C NMR (CDCl3): δ 13.4 (CH3), 20.9 (CH3),
21.9 (CH2), 23.05 (d, J ) 5 Hz, CH3 Me-19), 26.7 (CH2), 27.0
(CH), 33.3 (CH2), 33.4 (d, J ) 23 Hz, CH2-7), 35.1 (CH2), 39.4
(dd, J ) 18, 8 Hz, CH-8), 40.5 (CH2), 40.8 (CH), 44.6 (CH2),
47.3 (d, J ) 10 Hz, C-10), 68.0 (CH2), 68.9 (d, J ) 35 Hz, CH-
11), 86.3 (d, J ) 175 Hz, CH-6), 98.0 (d, J ) 175 Hz, C-9),
112.3 (CH2-18), 120.9 (d, J ) 10 Hz, CH-4), 130.0 (CH-2), 149.7
(C-13), 149.7 (CH-1), 160.2 (C-5), 172.9 (COOR), 184.9 (C-3),
203.1 (CO).
6r,9-Diflu or o-16r-m et h yl-11â,20,21-t r ih yd r oxy-18,20-
cyclopr egn a-1,4-dien -3-on e 21-valer ate (17). Colorless crys-
tals; mp 105 °C; [R]D +42°. IR: 3250, 2690, 1660, 1620, 900
cm-1. Anal. Cald for C27H36F2O5: C, 67.76; H, 7.58. Found: C,
676.72; H, 7.63. 1H NMR (CDCl3): δ 0.92 (t, J ) 7 Hz, 3H),
0.94 (d, J ) 7 Hz, 3H), 1.37 (ses, 2H, butyl methylene), 1.48
(s, 3H, Me-19), 1.65 (qui, 2H, butyl methylene), 1.9 (d, J ) 4
Hz, 1H, H-17), 2.1 (exch, 2H, OH), 2.0 (m, 1H, H-16), 2.0 and
2.4 (m, 2H, H-12), 2.35 (t, J ) 7 Hz, 2H, buthyl methylene),
1.5-2.2 (m, 8H), 4.0 (AB system, 2H), 4.4 (m, 1H), 5.4 (dddd,
J ) 50, 13, 8, 2 Hz, 1H), 6.35 (dd, J ) 10, 2 Hz, 1H), 6.43 (bs,
J ) 2 Hz, 1H), 7.1 (d, J ) 10 Hz, 1H). The C-20 configuration
was attributed on the basis of NOESY experiment: a cross-
peak between H-21 (4.0 ppm) and the hydrogen at 2.02 ppm
(attributed to H-18) was apparent, whereas no correlations
with H-17 (1.9 ppm) and with the methyl group in position 16
were found. A cross-peak between H-17 and the methyl in 16
position was apparent, confirming their spatial closeness.
Thus, in the cyclobutane ring the hydroxy group was cis to
H-17, whereas the chain was at the opposite side.
13C NMR (CDCl3): δ 13.5 (CH3), 22.0 (CH2), 22.5 (CH3), 23.0
(d, J C-F ) 5 Hz, CH3), 26.9 (CH2), 32.3 (dd, J C-F ) 18, 8 Hz,
CH-8), 32.8 (CH2), 33.8 (CH), 33.9 (d, J C-F ) 23 Hz, CH2-7),
33.9 (CH2), 34.2 (CH2), 35.8 (CH2), 41.2 (CH2), 43.7, 43.9 (CH),
48.1 (d, J C-F ) 27 Hz, C-10), 60.1 (CH-17), 66.9 (CH2), 72.5
(C-20), 72.5 (d, J C-F ) 36 Hz, CH-11), 86.4 (d, J C-F ) 175 Hz,
CH-6), 99.1 (d, J C-F ) 175 Hz, C-9), 120.9 (d, J C-F ) 13 Hz,
CH-4), 129.9 (CH-2), 150.7 (CH-1), 161.3 (d, J C-F ) 10 Hz, C-5),
174.0 (COOR), 185.5 (C-3).
Qu a n tu m Yield s Mea su r em en ts. A total of 3 mL of an
acetonitrile solution (2.5 × 10-3 M for all of the steroids, but
2 × 10-4 M for beclomethasone base) of the pregnane deriva-
tives were irradiated in spectrophotometric sealed cuvettes.
The light source was a low-pressure mercury arc lamp for 254
nm and a focalized high-pressure mercury arc lamp with the
appropriate interference filter for 313 and 366 nm. The fraction
of light absorbed was measured by means of a photon counter
placed behind the cuvette. The steroids conversion was deter-
mined by HPLC analyses. The light flux was measured by
ferrioxalate actinometry.
P h ot olysis of Diflu cor t olon e-21-va ler a t e (13) a t 254
n m . A solution of compound 13 (1 g, 2.1 mmol) in acetonitrile
(420 mL) was irradiated for 7 h, and 60% conversion was
reached. Chromatographic separation of the residue (eluting
with a cyclohexane-ethyl acetate, 80/20 mixture) gave 300 mg
(50%) of compound 15.
6r,9-Diflu or o-16r-m eth yl-11â,21-d ih yd r oxy-1,5-cyclo-
p r egn a -3-en -2,20-d ion e 21-va ler a te (15). Colorless crystals;
mp 220 °C (AcOEt); [R]D +8°. IR: 3490, 2954, 1721, 1685, 1172,
968 cm-1. Anal. Calcd for C27H36F2O5 C, 67.76; H, 7.58.
1
Found: C, 67.88; H, 7.31. H NMR (CDCl3): δ 0.9 (s, 3H, Me-
18), 0.93 (d, J ) 7, 3H), 0.98 (t, J ) 7, 3H), 1.3 (s, 3H, Me-19),
1.6 (m, 2H, CH2), 2.45 (t, J ) 7, 2H, CH2-CO), 2.75 (m, 1H,
H-16), 4.33 (m, 1H, H-11), 4.45 and 4.7 (AB system, J ) 17,
2H, H-21), 5.2 (dt, J ) 53, 7.5, 1H, H-6), 6.1 (d, J ) 5.5, 1H,
H-3), 7.67 (d, J ) 5.5, 1H, H-4). 13C NMR (CDCl3): δ 7.89
(CH3), 13.4 (CH3), 15.5 (CH3), 21.6 (CH3), 21.9 (CH2), 26.6
(CH2), 27.6 (d, J ) 23, CH2-7), 29.1 (dd, J ) 18.8, CH-8), 30.9
(CH), 32.6 (CH2), 33.2 (CH2), 36.0 (t, J ) 7, CH-1), 41.6 (CH2),
44.8, 46.8 (CH), 52.3 (d, J ) 27, C-10), 68.8 (d, J ) 35, CH-
11), 69.1 (CH2-21), 86.5 (d, J ) 175, CH-6), 94.8 (d, J ) 175,
C-9), 133.6 (CH-3), 159.8 (d, J ) 10, CH-4), 172.9 (COOR),
202.8 (C-2), 203.6 (C-20).
P h otolysis of Diflu cor tolon e-21-va ler a te a t 310 n m . A
solution of compound 13 (1 g, 2.1 mmol) in acetonitrile (420
mL) was irradiated for 4 h, when 50% conversion was reached.
Column chromatography (eluting with cyclohexane/ethyl ac-
etate 80/20) afforded the following photoproducts: 15, 20 mg
(20%); 16, 125 mg (30%) and a cyclobutanol derivative identi-
fied as 17, 20 mg (5%).
6r,9-Diflu or o-16r-m eth yl-11â,21-d ih yd r oxy-13-m eth yl-
en e-18-n or -13,17-seco-p r egn a -1,4-d ien -3,20-d ion e 21-va l-
er a te (16). Colorless crystals; mp 190 °C (AcOEt); [R]D +10°.
IR: 3300, 2960, 1725, 1660, 1620, 1075, 900 cm-1. Anal. Calcd
for C27H36F2O5: C, 67.76; H, 7.58. Found: C, 67, 8; H, 7.61.
Ack n ow led gm en t. Partial support of this work by
MIUR, Rome, and ISS, Rome, is gratefully aknowledged.
J O034070A
4366 J . Org. Chem., Vol. 68, No. 11, 2003