Communication
[
8]
cal to those previously reported, within experimental error,
and showed no trace of 14.
of benzene oxide 13 by CÀO bond cleavage is very unfavora-
ble. A pathway for CÀO bond cleavage to give the zwitterionic
intermediate proposed by Wong et al. (Scheme 4) could not be
located on the potential energy surface, and only became fea-
sible when the oxygen was protonated. Thus we believe the
mechanism of decomposition of both oxepins involves conver-
sion to the protonated benzene oxide, followed by CÀO bond
cleavage (and further 1,2 alkyl shifts in the case of 14). (Sup-
porting Information, Scheme S1). Surprisingly, however, experi-
ments with 6 and synthetic 14, revealed significant differences
in the stabilities of the two oxepins towards decomposition.
Whereas tert-butyloxepin 6 decomposed with a t1/2 of 13 min
in aqueous buffer (pH 7.4 50 mm Tris), the decomposition of
dimethyloxepin 14 under similar conditions (or even in unbuf-
fered water) was complete in <1 min by GCMS analysis, giving
a mixture of cyclohexadienone 12 and 2,6-dimethylphenol
(15). The ratio of these two products was essentially the same
as observed in the enzyme-mediated dearomatization of 11. In
contrast, 14 did not decompose noticeably in a less-polar solu-
tion (ethyl acetate, methanol) at room temperature over
a period of hours, or even in ethyl acetate/acetic acid solution.
The reason for the more rapid decomposition of 14 (compared
to 6) in aqueous solution is not immediately obvious from the
energetics shown in Scheme 5. We speculate that 13/14 is
more sensitive to protonation than 2a/6; the more readily
One of the primary difficulties in identifying arene oxides as
intermediates is their rapid rearrangement to more stable
[
6]
compounds (e.g., phenols) under physiological conditions. If
arene oxides are produced in the oxidations of both 1 and 11,
then the failure to detect oxepin 14 could in principle be due
to either a much slower valence bond isomerisation of arene
[8]
oxide 13 to 14, as proposed by Wong et al., or to a lower sta-
bility of oxepin 14 under the experimental conditions. We thus
investigated the differences between oxepins 6 and 14 experi-
mentally and computationally. An authentic, synthetic sample
[16a,20,21]
of 14 was prepared using Vogel’s reported route.
High-accuracy CBS-QB3 calculations (see the Supporting
Information) were performed to evaluate the stabilities of
[22]
oxepins 6 and 14 (Scheme 5). The isomerization of tert-butyl
benzene oxide 2a to oxepin 6, and of dimethyl benzene oxide
3 to oxepin 14 (Scheme 5a), are both computed to have
1
°
À1
small activation barriers (DG ꢀ5 kcalmol ), indicating that
equilibrium will be rapidly attained for both pairs. The values
À1
of DG (À1.5 and À2.5 kcalmol , respectively) predict that in
both cases the oxepin will be the dominant species at equilib-
rium and that, in fact, the arene oxide/oxepin equilibrium for
1
3/14 lies more strongly towards the oxepin than that of 2a/6
although the equilibrium position in polar solvents may con-
(
[16a]
+
+
tain somewhat more benzene oxide).
These computational
formed cations 13H /14H rearrange with almost negligible
+
+
data agree well with our experimental observations, where
NMR and GCMS analysis suggested that both 6 and 14 were
stable to rearrangement at room temperature in organic sol-
vents and at high temperature in the gas phase and that the
equilibrium between 6 and 2a lay largely toward 6. Vogel had
previously reported that the 13/14 equilibrium similarly
barrier to give 12H and 15H .
Considered together, the relatively slow decomposition of
14 in non-polar solvents (which may approximate the polarity
of a P450 active site), compared with its rapid decomposition
in aqueous solution, and the identical product ratios observed
from the enzyme catalyzed oxidation of 11 and the rearrange-
ment of 14 in aqueous solution together make it likely that
13/14 are indeed direct products of the P450BM-3 mediated oxi-
dation of 11. The experimentally and theoretically determined
properties of 14 are consistent with the idea that arene oxide
13 would have been formed in the P450BM-3-catalyzed oxida-
[16a]
strongly favored 14 even in polar solvents.
Furthermore, calculations predict that, under acidic condi-
tions, the conversion of both oxepin 6 to phenol 4a, and of
oxepin 14 to cyclohexadienone 12 (Scheme 5b), are strongly
À1
exergonic reactions, with low barriers (7–8 kcalmol ). Compu-
[
8]
tations suggest that, in the absence of acid, the ring opening
tion of ortho-xylene, as proposed by Wong et al. However, in
the relatively nonpolar enzyme active site, as seen with 2a/6,
13 would have converted rapidly to give predominantly
oxepin 14. This (14), upon release from the active site into
aqueous solution, would then rapidly rearrange to the
observed 12 and 15, thus eluding direct detection.
Despite a sustained interest in P450-catalyzed oxidation of
aromatic compounds, our detection of 6 among the P450-cata-
lyzed oxidation products of 1 is the first report of a discrete
oxepin product from a reconstituted in vitro system, and
indeed from a bacterial P450. The observation of oxepin 6 as
a product from P450 - and P450cam-mediated oxidation of
cin
1
but not of 14 from P450BM-3 catalyzed oxidation of 11 ap-
pears significant. Quantum chemical calculations reveal that
the rates of equilibration and relative stabilities of the 2a/6
and 13/14 pairs are comparable. As none of the enzyme/sub-
strate pairs (P450 /1, P450cam/1, P450BM-3/11) are natural pair-
cin
À1
Scheme 5. a) Computed activation barriers and energetics (kcalmol ) of
ings, it is unlikely that active site residues specifically inhibit
the valence bond tautomerisation of an intermediate benzene
oxide in one, but not another P450, or that the steric con-
a ring opening of benzene oxides to oxepins, and b) conversion of oxepin 6
to phenol 4a and of oxepin 14 to cyclohexadienone 12 under acidic
conditions, calculated with CBS-QB3.
Chem. Eur. J. 2016, 22, 4408 – 4412
4411
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim