The tamoxifen cation reacts to give indene products

Article abstract of DOI:10.1139/v00-122

The tamoxifen carbocation (Ph(Ar)C=CPh-CH⁺-CH₃, Ar = 4-Me₂NCH₂CH₂OC₆H₄) is generated from acetate and sulfate precursors by S(N)1 ionization in water. The cation exists in (E) and (Z) forms which equilibrate before reaction. The major products are the α-hydroxytamoxifens Ph(Ar)C=CPh-CHOH-CH₃, both (E) 64% and (Z) 29%, with the ratio independent of the configuration of the starting ester. Two minor products with a total yield of 7% account for the rest of the products. These have been characterized as indenes derived from intramolecular cyclization, a 4.5% yield of the indene derived from cyclization into the Ar ring with 2.5% due to cyclization into the phenyl ring. Experiments in acid solutions (0.01-0.1 M HCl) starting with pure (E)- or (Z)-α-hydroxytamoxifen reveal that the two alcohols equilibrate. This occurs by H⁺-catalyzed formation of the carbocation followed by water capture. Occurring about 10-fold slower than this isomerization is an irreversible process resulting in the two indenes. This cyclization will result in the destruction of the α-hydroxytamoxifens upon exposure to acidic conditions and also makes the direct observation of the tamoxifen carbocation under super-acid conditions difficult, if not impossible. The indenes do form in low yield whenever the tamoxifen carbocation is generated from an S(N)1 precursor. Thus these products could serve as markers for the formation of the tamoxifen carbocation in cellular systems or in in vivo experiments.

Full text of DOI:10.1139/v00-122

1186
The tamoxifen cation reacts to give indene
products
Cristina Sanchez and Robert A. McClelland
Abstract: The tamoxifen carbocation (Ph(Ar)C=CPh-CH+-CH3, Ar = 4-Me2NCH2CH2OC6H4) is generated from acetate
and sulfate precursors by SN1 ionization in water. The cation exists in (E) and (Z) forms which equilibrate before reac-
tion. The major products are the α-hydroxytamoxifens Ph(Ar)C=CPh-CHOH-CH3, both (E) 64% and (Z) 29%, with the
ratio independent of the configuration of the starting ester. Two minor products with a total yield of 7% account for
the rest of the products. These have been characterized as indenes derived from intramolecular cyclization, a 4.5%
yield of the indene derived from cyclization into the Ar ring with 2.5% due to cyclization into the phenyl ring. Experi-
ments in acid solutions (0.01–0.1 M HCl) starting with pure (E)- or (Z)-α-hydroxytamoxifen reveal that the two alco-
hols equilibrate. This occurs by H+-catalyzed formation of the carbocation followed by water capture. Occurring about
10-fold slower than this isomerization is an irreversible process resulting in the two indenes. This cyclization will re-
sult in the destruction of the α-hydroxytamoxifens upon exposure to acidic conditions and also makes the direct obser-
vation of the tamoxifen carbocation under super-acid conditions difficult, if not impossible. The indenes do form in low
yield whenever the tamoxifen carbocation is generated from an SN1 precursor. Thus these products could serve as
markers for the formation of the tamoxifen carbocation in cellular systems or in in vivo experiments.
Key words: carbocation, indene, allylic cation, isomerization.
Résumé : On a généré le carbocation tamoxifène, [Ph(Ar)C=CPHCH+CH3, Ar = 4-Me2NCH2CH2OC6H4] à partir de
précurseurs acétate et sulfate, par le biais de réactions SN1, dans l’eau. Le cation existe dans les formes (E) et (Z) qui
se mettent en équilibre avant de réagir. Les produits principaux sont les α-hydroxytamoxifènes Ph(Ar)C=CPH-CHOH-
CH3, E, 64 % et Z, 29 %, dont le rapport est indépendant de la configuration de l’ester de départ. Deux produits
mineurs, qui correspondent à un rendement total de 7 %, forment le reste des produits. Il s’agit d’indènes qui
découlent d’une cyclisation intramoléculaire, dont 4,5 % de l’indène dérivé d’une cyclisation dans le noyau aromatique
et 2,5 % dû à une cyclisation dans le noyau phényle. Des expériences faites en milieu acide (0,01 à 0,1 M, HCl) faites
à partir des (E)- ou (Z)-α-hydroxytamoxifènes purs révèlent que les deux alcools se mettent en équilibre. Ceci résulte
d’une formation acidocatalysée du carbocation, suivie par une fixation d’eau. La vitesse de réaction du processus
irréversible conduisant aux deux indènes est dix fois plus faible. Cette cyclisation conduit à la destruction des α-
hydroxytamoxifènes lorsqu’ils sont exposés à des conditions acides et elle rend pratiquement impossible toute observa-
tion directe du carbocation tamoxifène dans des conditions superacides. Les indènes se forment avec de faibles
rendements chaque fois que le carbocation tamoxifène est généré par le biais d’une réaction SN1. Ces produits
pourraient donc servir de marqueurs pour la formation du carbocation tamoxifène dans des systèmes cellulaires ou dans
des expériences in vitro.
Mots clés : carbocation, indène, cation allylique, isomérisation. Sanchez and McClelland 1193
We have recently reported the direct observation of the
volves allylic hydroxylation of the parent drug (Tam) to
form E-α-hydroxytamoxifen E-TOH, a known metabolite.
This is followed by sulfotransferase-mediated conversion to
tamoxifen carbocation in aqueous solution through the use
of the technique of laser flash photolysis (LFP) (1). This cat-
ion appears to be responsible for the formation of DNA ad-
ducts observed in both animal models (2, 3) and in women
treated with the drug tamoxifen (4–7). Such adducts could
explain the hepatocarcinogenicity observed in rats (8–10),
and the small number of endometrial cancers in women (11–13).
The metabolic pathway of Scheme 1 (14–20) accounts for
the formation of the cation. The initial metabolic event in-
–
the sulfate E-TOSO3 (21–24). This ester is quite labile in
water, undergoing relatively rapid SN1 ionization, for exam-
ple, with a half-life around 10 s in water at 20°C (1).
In comparison with other carbocations (25), the tamoxifen
intermediate is relatively long-lived. Limiting lifetimes
(1/k(decay)) are 160 µs (side-chain amine as free base) and
40 µs (amine protonated).2 An interesting feature is that the
Received November 30, 1999. Published on the NRC Research Press website on August 30, 2000.
C. Sanchez and R.A. McClelland.1 Department of Chemistry, University of Toronto, 80 St. George St., Toronto, ON M5S 3H6,
Canada.
1Author to whom correspondence may be addressed. Telephone: (416) 978-3592. Fax: (416) 978-3592.
e-mail: rmcclell@alchemy.chem.utoronto.ca
2These numbers are based on measurements in wholly aqueous solutions (E. Sauer, C. Sanchez, and R.A. McClelland, unpublished
data). Our previously reported lifetimes (1) of 125 µs and 22 µs refer to a 60:40 water:acetonitrile mixture.
Can. J. Chem. 78: 1186–1193 (2000)
© 2000 NRC Canada

Sanchez and McClelland
1187
Scheme 1.
Fig. 1. Spectra obtained with 2 × 10–4 M (E)-α-
hydroxytamoxifen (E-TOH) in 0.1 M HCl (20°C). Spectra were
recorded every 20 min. The absorbance increase at 310 nm fits
to a single exponential with a rate constant of 1.1 × 10–4 s–1.
intermediate appears to be a rapidly equilibrating mixture
of the two isomers E-T+ and Z-T+. Thus starting from
–
E-TOSO3 , a mixture of the E-TOH and Z-TOH is ob-
with λmax at 310 nm. The increase at 310 nm follows good
single exponential kinetics and there are reasonable
isosbestic points at 260 and 270 nm. This suggests a first-
order conversion of the starting alcohol to product (or prod-
ucts). However, HPLC chromatograms recorded on the same
solution show that the behavior is more complicated.
Starting with E-TOH as in Fig. 2, this is found to convert to
an approximately 1:1 ratio with its (Z)-isomer Z-TOH,
whose HPLC peak appears at a slightly longer retention time
(see labels in Fig. 2B). The same interconversion to an ~1:1
E:Z ratio is obtained with Z-TOH as the starting alcohol. At
the same time as the isomerization, two new peaks at longer
retention time are appearing. This process is considerably
slower, and, as will be apparent later, it accounts for the
spectral changes in the UV spectra. Ultimately, the two alco-
hols completely disappear, and the chromatogram contains
only the two peaks for the new products.
served (1, 18). The same ratio of these two alcohols is ob-
tained with the (Z)-sulfate as the starting material (1).
Mixtures of (E):(Z) products are also observed when the cat-
ion reacts with azide ion (1) and with the NH2 group of gua-
nine (18, 19, 26).
The rapid interconversion can be explained by noting that
the cation is better viewed as an α-vinyl substituted
diarylmethyl cation such as E-T+’, i.e., where there is little
double bond character in the central C—C bond. In fact
AM1 calculations show that steric congestion forces the vi-
nyl group to rotate 60–70 degrees out of the plane of the C+-
center shown in structure (1). Despite this rotation however
we have been unable to detect any products derived from ad-
dition of nucleophiles to the C+-center α to the the two aryl
groups. In searching for these, we did note two minor peaks
in the HPLC chromatogram eluting at considerably longer
retention time than the alcohols E-TOH and Z-TOH. Exper-
iments under acidic conditions where the two alcohols
interconvert then revealed that with time an irreversible reac-
tion occurs to produce these same peaks, ultimately as the
only peaks in the chromatogram. This provided conditions
for isolation of the products responsible. In this paper we
provide the details of these experiments, including the char-
acterization of the new products. They turn out to be derived
from intramolecular cyclization in the tamoxifen cation. A
kinetic analysis shows that the cyclization competes with
water addition. Thus, the cyclic products can be taken as an
indicator of the formation of the tamoxifen cation.
Identification of Indenes
The new products were identified by working on a larger
scale in 1 M HCl, and waiting a sufficient time so that there
were none of the α-hydroxytamoxifens remaining. An HPLC
chromatogram of this mixture was very similar to that ob-
tained at long times on the smaller scales, with the peak at
shorter retention time having twice the area of that at the
longer retention time. The products were then isolated after
neutralization and extraction. Column chromatography un-
fortunately failed to provide any separation whatsoever, so
that identification had to be carried out on the mixture. The
mass spectrum showed a molecular ion at m/z 369 MS, with
high resolution MS providing a molecular formula of
C26H27NO. This indicates that the products are derived from
loss of water from the starting material. Considering the
chemistry, i.e., that these products arise under acidic condi-
tions where the tamoxifen cations Z-T+ and E-T+ are being
formed, the conclusion is that they are isomeric indenes
Z-In and E-In arising from intramolecular cyclization in the
Results and discussion
Reaction of α -hydroxytamoxifens in acid
The reactions of E-TOH and Z-TOH in acidic aqueous
solutions were first seen in absorption spectra recorded as a
function of time in such solutions. As illustrated by the ex-
ample in Fig. 1, these show the slow growth of a new peak
© 2000 NRC Canada

1188
Can. J. Chem. Vol. 78, 2000
Fig. 2. HPLC chromatograms (detector at 260 nm) obtained with 2 × 10–4 M (E)-α-hydroxytamoxifen (2) in 0.1 M HCl (20 ± 1°C).
Solutions were injected at 0.8 (A), 23.8 (B), 95.3 (C), and 164.3 (D) min after preparation of the solution.
cations. This of course is well established chemistry of
phenyl-substituted allylic cations (28–34). The surprising as-
pect of the present study is that the intramolecular reaction
competes with the nucleophilic addition of water. Previous
studies of the cyclization have been carried out under strongly
acidic conditions where the competing reagents were weakly
nucleophilic.
Z-In. The indicator here is the aromatic protons next to the
alkoxy group, which can be distinguished from the complex
multiplet associated with the other aromatic protons. In the
indene Z-In these protons are nonequivalent. The proton la-
belled Ha in the representation above appears as a doublet of
doublets at 6.87 ppm, with coupling constants of 2.2 and
8.2 Hz representing coupling with Hb and Hc, respectively.
The proton Hb is a doublet at 7.18 ppm, with a coupling con-
stant of 2.2 Hz. In Z-In, these protons (Hd) are equivalent,
and appear as a doublet at 6.95 ppm with a coupling con-
stant of 8.6 ppm. The integration of the doublet of doublets
at 6.87 (Ha) vs. the doublet at 6.95 is ~1:1. Since two pro-
tons provide the latter doublet, the ratio of Z-InE-In is ~2:1,
so that Z-In is the major product eluting at the earlier reten-
tion time in the HPLC chromatograms.
Quantitative analysis
The HPLC chromatograms of Fig. 2 show that there are
four species present during the reaction that is occurring in
acids. The isosbestic point at 260 nm must therefore mean
that these four have essentially identical extinction coeffi-
cients at this wavelength. This in turn suggests that by oper-
ating at 260 nm in the HPLC, the sensitivities of the four
peaks are the same, or, in other works, relative peak areas
can be converted directly to relative molar amounts. This
was verified for E-TOH and Z-TOH by injecting known
amounts of authentic alcohols. It was further observed that
the sum of the areas of the four peaks was constant
throughout the reaction, within an experimental error of
±3%. This of course means that the products E-In and Z-In
must have the same sensitivity as the alcohols.
1
The H NMR spectrum of the mixture is consistent with
this interpretation. This also shows that the major product at
shorter retention time is obtained upon cyclization into the
alkoxy-substituted aromatic ring; i.e., the major product is
Working on the basis of the HPLC sensitivities being the
same at 260 nm, Figs. 3A and 3C were constructed. These
© 2000 NRC Canada

Sanchez and McClelland
1189
Fig. 3. HPLC analysis of solutions of (Z)-α-hydroxytamoxifen (Z-TOH) (Figs. 3A, 3B) and (E)-α-hydroxytamoxifen (E-TOH)
(Figs. 3C, 3D) in 0.02 M HCl, 0.08 M NaCl (20°C). Legend for Figs. 3A, 3C: (᭿) — [Z-TOH]:[Total], (ٗ) — [E-TOH]:[Total], (᭺)
— [Z-In]:[Total], (᭹) — [E-In]:[Total], where [Total] = [E-TOH] + [Z-TOH] + [E-In] + [Z-In]. Legend for Figs. 3B, 3D: (᭝) —
[Z-TOH]/([E-TOH] + [Z-TOH]), (᭡) — [E-TOH]/([E-TOH] + [Z-TOH]), (9)) — ([E-TOH] + [Z-TOH]):[Total]. The points are
experimental. The lines are based upon fitting the data to a double exponential or single exponential equation (see Table 1).
plot the fractions of each of the four species as a function of
time for experiments starting with Z-TOH (Fig. 3A) and E-
TOH (Fig. 3C) in the same acid solution. It can be seen that
the two alcohols interconvert, reaching what appears to be
equilibrium position with the ratio [E-TOH]:[Z-TOH]
slightly greater than unity regardless of the starting alcohol.
Superimposed on this is the slower process in which the al-
cohols convert to the indenes. The ratio [Z-In]:[E-In] is
constant throughout each of the runs, and, moreover is the
same whether E-TOH or Z-TOH is the starting material.
The results of kinetic analyses of the data in Figs. 3A and
3C are provided in Table 1. The data for the two alcohols
clearly require fitting to an equation that contains two expo-
nential terms, and indeed satisfactory fits are so obtained.
The two rate constants determined in such fits differ by a
factor of 10–12. The data for the increases in the two
indenes, on the other hand, are satisfied by a single exponen-
tial. The rate constants so determined are the same as those
associated with the slower process of the double exponential
fits. Within experimental uncertainty, the same rate constants
are obtained whether Z-TOH or E-TOH is the starting ma-
terial. Experiments were also performed in solutions with
different HCl concentrations with the ionic strength main-
tained constant at 0.1 M with NaCl. These show that both
the fast process and the slow process are first-order in H
final observation is that the slower rate constant obtained
from the HPLC data is, within experimental error, the same
as the single rate constant obtained from the exponential rise
in the UV spectra (obviously for experiments at the same
acid concentration). The explanation for this behavior is that
the spectra of the two alcohols E-TOH and Z-TOH are so
similar that their interconversion is invisible in the UV.
Thus, the spectral change that is observed represents the irre-
versible process forming the indene products.
+. A
Kinetic analysis
Our proposed kinetic model is shown in Scheme 2. This
contains the two intermediate carbocations E-T+ and Z-T+.
Each reacts with water to form the corresponding alcohol,
and, in competition, cyclizes to an indene. Associated with
these reactions are four first-order rate constants kw(E),
kcyc(E), kw(Z), and kcyc(Z). The carbocations can also form
© 2000 NRC Canada

1190
Can. J. Chem. Vol. 78, 2000
Table 1. Kinetic analyses of HPLC chromatograms of Fig. 3 obtained with Z-OH and E-TOH in 0.02 M HCl, 0.08 M NaCl at 20°C.
Constant
Start with Z-OH
Start with E-TOH
Double exponential decay
k1 2.57 × 104 s–1
k2 1.89 × 105 s–1
k1 2.61 × 104 s–1
k2 1.87 × 105 s–1
1.88 × 105 s–1
1.87 × 105 s–1
1.86 ± 0.05
2.33 × 104 s–1
2.01 × 105 s–1
2.40 × 104 s–1
2.01 × 105 s–1
2.11 × 10–5 s–1
2.07 × 10–5 s–1
1.83 ± 0.04
2.13 × 10–4 s–1
0.537 ± 0.004
1.89 × 10–5 s–1
of starting alcohola
Double exponential rise
and fall of isomerb
Single exponential rise of major indene productc
Single exponential rise of minor indene productc
Ratio of major indene to minor indened
Single exponential approach to equilibrium — k(isom)e [E-TOH]/([Z-TOH] + [E-TOH])
[E-TOH]([Z-TOH]+[E-TOH] at equilibriumf
Single exponential decay of ([E-TOH] + [Z-TOH]):[total] – k(cyc)g
2.37 × 10–4 s–1
0.535 ± 0.02
2.12 × 10–5 s–1
aFit to A1·Exp(–k1t) + A2·Exp(–k2t) with A1 + A2 = 1. (Forced to begin at one and end at zero).
bFit to A1·{Exp(–k1t) – Exp(–k2t)}. (Forced to begin and end at zero).
cFit to (–A1)·Exp(–kt) + A1. (Forced to begin at zero).
dAveraged over all the points.
eFit to eq. [12]. (Forced to begin at zero when Z-TOH was starting material and to begin at one when E-TOH was starting material).
f(k2:(k1 + k2)) of eq. [12].
gFit to eq. [13]. (Forced to begin at one and end at zero).
from the appropriate alcohols in H+-catalyzed processes. The
Scheme 2.
second order rate-constants for these pathways are defined
as kH(E) and kH(Z); in a solution of a particular acid concen-
tration, the cations will form in pseudo-first-order reactions
with rate constants kH(E)[H+] and kZ(E)[H+].
The same carbocations are also the intermediates of the
solvolysis of sulfate and acetate precursors. These reactions
can be carried out under conditions of kinetic control, i.e., in
solutions of very low H+-concentration such that the alcohol
products are stable on the time scale required for complete
solvolysis of the ester. An important observation from such
experiments is that regardless of whether the ester is (E) or
(Z), E-TOH and Z-TOH are produced in the same ratio (1),
a value of slightly greater than 2:1 in favor of E-TOH. The
implication is that E-T+ and Z-T+ achieve equilibrium
within their lifetime. These lifetimes (of the equilibrating
cations) are known from the LFP experiments, i.e., 40 µs in
aq acid. Thus, the (E)-(Z) interconversion in the cations must
be very rapid.3
We define the equilibrium constant [Z-T+]:[E-T+] as K+.
For the solvolysis reactions under conditions of kinetic con-
trol, eqs. [1]–[4] can be written for the fraction of each prod-
uct that is obtained.
[1]
Fr(E-TOH) =
kw(E)
kw(E) + kw(Z)K+ + kcyc(Z)K+ + kcyc(E)
= 0.640 ± 0.005
[3]
Fr(E-In) =
[2]
Fr(Z-TOH) =
kw(Z)K+
kcyc(E)
kw(E) + kw(Z)K+ + kcyc(Z)K+ + kcyc(E)
kw(E) + kw(Z)K+ + kcyc(Z)K+ + kcyc(E)
= 0.290 ± 0.003
= 0.025 ± 0.002
3 We have recent evidence that the cations can be trapped before rotation at high concentrations of azide ion, a nucleophile which is a very ef-
fective quencher (1).
© 2000 NRC Canada

Sanchez and McClelland
1191
Fig. 4. HPLC chromatogram obtained after solvolysis of 1 × 10–4
M
Scheme 3.
(E)-α-acetoxytamoxifen (E-TOAc) in 0.0005 M HCl. Note that
the gradient employed to obtain this chromatogram is different
from that used in Fig. 2 (see Experimental section). The small
peak between the peaks for Z-TOH and Z-In is due to a small
amount of unreacted E-TOAc.
where
[8]
[9]
k1 = Fr(Z-ROH)·kH(E)[H+]
k2 = Fr(E-ROH)·kH(Z)[H+]
[10] k3 = Fr(In)kH(E)[H+]
[11] k4 = Fr(In)kH(Z)[H+]
These equations are identical to ones that pertain to the ki-
netic system of Scheme 3, a model where E-TOH and Z-
TOH interconvert with rate constants k1 and k2, and react ir-
reversibly to form the indenes with rate constants k3 and k4.
The HPLC data of Fig. 3 show that the alcohols equilibrate
about 10 times faster than they cyclize to form the indenes.
Thus, we make the approximation that k1,k2 > k3,k4. This is
equivalent to treating the kinetics in two distinct phases — a
rapid phase where the alcohols achieve equilibrium and a
slow phase where they react irreversibly to form the indenes.
This approximation results in eq. [12] for the equilibration
and eq. [13] for the cyclization.
[4]
Fr(Z-In) =
[E-TOH]
k2
= Ae−k(isom)t
+
kcyc(Z)K+
kw(E) + kw(Z)K+ + kcyc(Z)K+ + kcyc(E)
= 0.045 ± 0.002
[12]
[E-TOH] + [Z-TOH]
k1 + k2
where k(isom) = (k1 + k2), and A = k1:(k1 + k2) or – k2:(k1 + k2)
for the cases where E-TOH and Z-TOH, respectively are
the starting alcohols.
Figure 4 shows a chromatogram obtained in such an ex-
periment, the almost complete solvolysis of (E)-α-
acetoxytamoxifen (E-TOAc) in a very dilute acid solution.
The acid was added here to ensure that the products reflect
reactions of carbocations where the side-chain amine is
protonated. The major products are E-TOH and Z-TOH, but
the peaks for the indenes Z-In and E-In are present in small
amounts. These experiments provide numerical values of the
fractions in eqs. [1]–[4]; the ones given above are averages
of several experiments. It can be noted that the ratio [Z-
In]:[E-In] is 1.75 ± 0.2, within experimental error the same
as the value obtained when these two indenes are formed
from alcohol starting materials in more acidic solutions.
To analyze the experiments with one of the alcohols as
starting material, we calculate the steady state concentration
of the two carbocations with the assumption that they are in
equilibrium. For further simplification we treat the two
indene isomers as one, by setting [In] = [E-In] + [Z-In] and
Fr(In) = (Fr(E-In) + Fr(Z-In)). This results in the three dif-
ferential equations of eqs. [5]–[7].
[E-TOH] + [Z-TOH]
[13]
= e – k(cyc)t
[E-TOH] + [Z-TOH] + [ln]
where k(cyc) = (k2k3 + k1k4)/(k1 + k2).
As shown in Figs. 3B and 3D, [E-TOH]/([E-TOH] + [Z-
TOH]) and ([E-TOH] + [Z-TOH])/([E-TOH] + [Z-TOH]
+ [In]) satisfy a single exponential, as required by eq. [12]
and eq. [13] respectively. As seen in Table 1, the rate con-
stant k(isom) from the fit to eq. [12] is essentially the same
as the larger of the rate constants obtained by fitting the data
for the individual alcohols to equations with two exponen-
tials. The rate constant k(cyc) is the same as the smaller of
the rate constants and is also identical to the value obtained
from the appearance of the indene products.
From experiments at different acid concentrations, both
k(isom) and k(cyc) were found to be proportional to [H+].
Average values of the second-order rate constants obtained
by dividing by H+ are given in eqs. [14] and [18] below. In
the case of the isomerization process, the following relations
apply:
d[E-TOH]
[5]
[6]
[7]
= k2[Z-TOH] – (k1 + k3)[E-TOH]
= k1[E-TOH] – (k2 + k4)[Z-TOH]
dt
k(isom)
[H+ ]
k1 + k2
[H+]
[14]
=
d[Z-TOH]
dt
= (1.16 ± 0.06) × 10–2 M–1 s–1
d[ln]
= k3[E-TOH] + k4[Z-TOH]
dt
= Fr(Z-TOH)·kH(E) + Fr(E-TOH)·kH(Z)
© 2000 NRC Canada

1192
Can. J. Chem. Vol. 78, 2000
The fit to eq. [12] also provides the fraction of E-TOH
that is present when the E-TOH: Z-TOH equilibrium is
established:
Small amounts of the indenes also form at neutral pH
when the carbocations are generated in SN1 solvolysis reac-
tions of acetate and sulfate precursors. Observation of these
indenes could be used as an indicator of the formation of the
tamoxifen carbocations in cellular systems and even in vivo.
To date such experiments have relied upon the formation of
the DNA adducts to indicate the formation of the cation
(26). The indenes with their λmax at 310 nm and their rela-
tively long retention time should be easily detected by re-
versed-phase HPLC. Their formation would be unambiguous
proof of the formation of the tamoxifen carbocation (and the
SN1 precursor to this cation).
[E-TOH]eq
k2
[15]
= 0.535 =
k1 + k2
[E-TOH]eq + [Z-TOH]eq
[E-TOH]eq
=
•
•
Fr[Z-TOH] kH(E) + Fr[E-TOH] kH(Z)
With some algebraic manipulation, inputting the values of
Fr(Z-TOH) and Fr(E-TOH) from eqs. [1] and [2] gives the
second-order rate constants:
Experimental section
[16] kH(E) = 2.14 × 10–2 M–1 s–1
[17] kH(Z) = 8.42 × 10–3 M–1 s–1
(E)-α -Hydroxytamoxifen and (Z)-α -hydroxytamoxifen
((E)- and (Z)-1-[4-[2-(dimethylamino)ethoxy]phenyl]-1,2-diphenyl-
1-buten-3-ol) were prepared as a mixture and separated by
column chromatography following a literature procedure
(27). (E)-Acetoxytamoxifen was prepared from the (E)-alco-
hol by treatment with acetic anhydride in pyridine, followed
by chromatography as described previously (26, 35).
It is interesting to note that E-TOH forms from the
equilibrating cations with a slightly greater than 2:1 prefer-
ence over Z-TOH, but it also undergoes H+-catalyzed ion-
ization over two-fold faster. The consequence is that the
ratio of the two alcohols at equilibrium is close to unity.
For the cyclization process,
Z-In (6-(2-dimethylaminoethoxy)-1-methyl-2,3-diphenyl-
1H-indene) and E-In (3-[4-(2-dimethylaminoethoxy)phenyl]-
1-methyl-2-phenyl-1H-indene) were obtained as a mixture
by adding 50 mg of (E)-α-hydroxytamoxifen to 100 mL of
1 M HCl. After stirring at ambient temperature for 14 h, the
pH was adjusted to 10–11 by the addition of sodium hydrox-
ide, and the indenes extracted with dichloromethane. After
drying (MgSO4) and filtration, the solvent was removed to
yield a viscous oil. This was purified by chromatography us-
ing 8:11:1 ether:hexanes:triethylamine. This failed to pro-
vide any separation of the two isomers, as was also the case
of TLC experiments using various other solvent combina-
tions. MS (EI): m/z 369 (100%), 324 (5%), 298 (50%), 265
(75%), 233 (45%), 202 (30%). HRMS m/z 369.2097;
C26H27NO requires 369.2093. The 1H NMR (400 MHz,
CDCl3) of the mixture exhibited peaks that could be as-
signed to the major isomer (Z-In) and minor isomer (E-In):
1.315 (major) and 1.320 (minor) (3H, d, J = 7.5 Hz) (CH3),
2.384 (major and minor) (6H, s) (N(CH3)2), 2.790 (major)
and 2.780 (minor) (2H, t, J = 5.6 Hz) (NCH2), 4.025 (major)
and 4.044 (minor) (1H, q, J = 7.6 Hz) (CH), 4.152 (major)
and 4.118 (minor) (2H, t, J = 5.6 Hz), 6.87 (major) (1H, dd,
J = 2.2 Hz and 8.2 Hz) (5-H of indene ring of Z-In), 6.95
k(cyc)
[H+ ]
[18]
= (1.01 ± 0.04) × 10–3 M–1 s–1
k2
k1
= Fr(In) kH(E)
+ kH(Z)
k1 + k2
k1 + k2
where the last term is obtained by substituting the various
relations for k(cyc) from eq. [13], and k3 and k4 from
eqs. [10] and [11]. Numerical values for all the quantities
except Fr(In) are known from the isomerization kinetics (see
eqs. [15]–[17]). Substitution gives FR(In) = 0.066. The
overall consistency in the analysis is seen in the excellent
agreement with the value of 0.070 obtained as the sum of
Fr(E-In) and Fr(Z-In) obtained from the product analysis in
the acetate solvolysis.
Summary
These experiments show that the equilibrating tamoxifen
carbocations undergo intramolecular cyclization in competi-
tion with their reaction with aqueous solvent. The
cyclizations represent 7% of the fate of the cations under
conditions of kinetic control. Thus they occur with a rate
constant of 1.8 × 103 s–1, i.e., 7% of the rate constant of 2.5
× 104 s–1 obtained by LFP for cation decay (where these rate
constants refer to acidic conditions where the side chain
amino group is protonated).
We also show that acid conditions must be avoided when
performing quantitative analysis of α-hydroxytamoxifens.
The H+-catalyzed formation of the tamoxifen carbocations
will occur under these conditions, and the cyclizations repre-
sent irreversible processes that result in removal of the alco-
hols. We had also hoped that we would be able to generate
the tamoxifen carbocations under weakly nucleophilic super-
acid conditions. The occurrence of the rapid intramolecular
cyclization however has made this impossible to date, even
working at lower temperatures.
(minor) (2H, d,
J
=
8.2 Hz) (protons ortho to
OCH2CH2N(CH3)2 group in E-In), 7.16–7.14 (multiplet) (all
the other phenyl protons, H7 of the indene ring of Z-In
could be seen as a doublet at 7.18 Hz with J = 2.2 Hz).
HPLC experiments were conducted with a Waters 600E
system, a Waters 486 Tunable Absorbance Detector set at
260 nm, and Waters 746 Data Module. The column was a
Water Symmetry C18 µm column of dimensions 4.6 mm ×
150 mm. The experiments with the α-hydroxytamoxifens as
starting material were performed with a flow of 1 mL/min
using a programmed eluting system of 0–2 min — isocratic
60% buffer:40% acetonitrile; 2–8 min — linear gradient
ending at 15% buffer:85% acetonitrile; and 8–18 min —
isocratic 15% buffer:85% acetonitrile, where buffer =
0.025 M sodium acetate:0.025 M acetic acid. Retention
times were 4.6 min (E-TOH), 5.9 min (Z-TOH), 14.2 min
© 2000 NRC Canada

Sanchez and McClelland
1193
13. F.E. van Leeuwen, J. Benraadt, J.W.W. Coebergh, L.A.L.
Kiemeney, F.W. Diepenhorst, A.W. van den Belt-Dusebout,
and H. van Tinteren. Lancet, 343, 448 (1994).
14. G.A. Potter, R. McCague, and M. Jarman. Carcinogenesis, 15,
439 (1994).
15. D.H. Phillips, G.A. Potter, M.N. Horton, A. Hewer, C.
Crofton-Sleigh, M. Jarman, and S. Venitt. Carcinogenesis, 15,
1487 (1994).
(Z-In), and 15.4 min (E-In). Experiments starting with
α-acetoxytamoxifen used a program of 0–1 min — isocratic
70% buffer:30% acetonitrile; 1–11 min — linear gradient
ending at 15% buffer:85% acetonitrile; and 8–18 min —
isocratic 15% buffer:85% acetonitrile, where buffer = as
above. Retention times were 11.3 min (E-TOH), 12.1 min
(Z-TOH), 13.7 min (E-TOAc), 14.8 min (Z-In), and
15.3 min (E-In). Solutions contained 1 × 10–4 M of the sub-
strate with 60 µL being directly injected into the HPLC.
16. D.H. Phillips, P.L. Carmichael, A. Hewer, K.J. Cole, and G.K.
Poon. Cancer Res. 54, 5518 (1994).
17. D.H. Phillips, P.L. Carmichael, A. Hewer, K.J. Cole, I.R.
Hardcastle, G.K. Poon, A. Keogh and A.J. Strain.
Carcinogenesis, 17, 89 (1994).
18. M.R. Osborne, A. Hewer, I.R. Hardcastle, P.L. Carmichael,
and D.H. Phillips. Cancer. Res. 56, 66 (1996).
19. M.R. Osborne, I.R. Hardcastle, and D.H. Phillips.
Carcinogenesis, 18, 539 (1997).
20. G.K. Poon, B. Walter, P.E. Lonning, M.N. Horton, and R.
McCague. Drug. Metabol. Dispos. 23, 377 (1995).
21. S. Shibutani, L. Dasaradhi, I. Terashima, E. Banoglu, and
M.W. Duffel. Cancer Res. 58, 647 (1998).
22. S. Shibutani, P.M. Shaw, N. Suzuki, L. Dasaradhi, M.W. Duf-
fel, and I. Terashima. Carcinogenesis, 19, 2007 (1998).
23. W. Davis, S. Venitt, and D.H. Phillips. Carcinogenesis, 19, 861
(1998).
Acknowledgement
Financial support from the Natural Sciences and Engi-
neering Research Council of Canada (NSERC) is gratefully
acknowledged.
References
1. C. Sanchez, S. Shibutani, L. Dasaradhi, J.L. Bolton, P.W. Fan,
and R.A. McClelland. J. Am. Chem. Soc. 120, 13513 (1999).
2. X. Han and J.G. Liehr. Cancer Res. 52, 1360 (1992).
3. I.N.H. White, F. de Matteis, A. Davies, L.L. Smith, C.
Crofton-Sleigh, S. Venitt, A. Hewer, and D.H. Phillips.
Carcinogenesis, 13, 2197 (1992).
24. K. Randerath, B. Moorthy, N. Mabon, and P. Sriram.
Carcinogenesis, 15, 2087 (1994).
25. R.A. McClelland. Tetrahedron, 52, 6823 (1996).
26. L. Dasaradhi and S. Shibutani. Chem. Res. Toxicol. 10, 189
(1997).
4. K. Hemminki, H. Rajaniemi, B. Lindahl, and B. Moberger.
Cancer Res. 56, 4374 (1996).
5. K. Hemminki, H. Rajaniemi, M. Koskinen, and J. Hansson.
Carcinogenesis, 18, 9 (1997).
27. A.B. Foster, M. Jarman, O.-T. Leung, R. McCague, G.
Leclerq, and N. Devleeschouwer. J. Med. Chem. 28, 1491
(1985).
28. Chemistry of Carbon Compounds. Vol. III, Edited by E.H.
Rodd. Elsevier, Amsterdam, 1954.
29. D.Y. Curtin, H.W. Johnston, and E.G. Steiner. 77, 4566 (1955).
30. D.H.R. Barton and D.W. Jones. J. Chem. Soc. xx, 3563
(1965).
31. M.G. Miller and C.V. Pittman. J. Org. Chem. 39, 1955 (1974).
32. R. Knorr, E. Lattke, F. Ruf, and H.-U. Reissig. Chem. Ber.
114, 1592 (1981).
33. R.R. Kostikov, G.S. Varakin, and K.A. Ogloblin. J. Org.
Chem. USSR (Engl. Transl.) 19, 1438 (1983).
34. N.J. Head, G.A. Olah, and G.K.S. Prakash. J. Am. Chem. Soc.
117, 11205 (1995).
35. M.R. Osborne, A. Hewer, I.R. Hardcastle, P.L. Carmichael,
and D.H. Phillips. Cancer Res. 56, 66 (1996).
6. S. Shibutani, L. Dasaradhi, S. Sugarman, A.P. Grollman, and
M. Pearl. Proc. Am. Assoc. Cancer Res. 39, 636 (1998).
7. S. Shibutani. Environ. Mutagen Res. 20, 201 (1998).
8. G.M. Williams, M.J. Iatropoulos, M.V. Djordjevic, and O.P.
Kaltenberg. Carcinogenesis, 14, 315 (1993).
9. P. Greaves, R. Groonetilleke, G. Nunn, J. Topham, and T.
Orton. Cancer Res. 53, 3919 (1993).
10. G.C. Hard, M.J. Iatropoulos, K. Jordan, L. Radi, O.P.
Kaltenberg, A.R. Imondi, and T. Orton. Cancer Res. 53, 4534
(1993).
11. M.A.-F. Seoud, J. Johnson, and J.C. Weed. Obstet. Gynecol.
82, 165 (1993).
12. B. Fischer, J.P. Costantino, C.K. Redmond, E.R. Fisher, D.L.
Wickerham, and W.M. Cronin. J. Natl. Cancer Inst. 86, 527
(1994).
© 2000 NRC Canada