Chemical synthesis of six novel 17β-estradiol and estrone dimers and study of their formation catalyzed by human cytochrome P450 isoforms

Article abstract of DOI:10.1021/jm0707323

Our earlier studies have shown that over 20 nonpolar 17β-estradiol metabolite peaks were detected following incubations of radioactive 17β-estradiol with human liver microsomes or recombinant human cytochrome P450 isoforms in the presence of NADPH as a co

Full text of DOI:10.1021/jm0707323

5372
J. Med. Chem. 2007, 50, 5372-5381
Chemical Synthesis of Six Novel 17â-Estradiol and Estrone Dimers and Study of Their
Formation Catalyzed by Human Cytochrome P450 Isoforms
Aaron Yun Chen, Anthony J. Lee,† Xiang-Rong Jiang, and Bao Ting Zhu*
Department of Pharmacology, Toxicology and Therapeutics, School of Medicine, UniVersity of Kansas Medical Center,
2146 West 39th Street, Kansas City, Kansas 66160
ReceiVed June 21, 2007
Our earlier studies have shown that over 20 nonpolar 17â-estradiol metabolite peaks were detected following
incubations of radioactive 17â-estradiol with human liver microsomes or recombinant human cytochrome
P450 isoforms in the presence of NADPH as a cofactor. The structures of two representative nonpolar
metabolites were identified earlier as dimers of 17â-estradiol linked through a diaryl ether bond between
the C-3 phenolic oxygen of one molecule and the C-2 or C-4 aromatic carbon of another. Six additional
putative dimers between estrone and 17â-estradiol with structures similar to the two identified ones were
synthesized in this study. Using these newly synthesized estrogen dimers as reference standards, we
demonstrated that incubations of human liver microsomes or various human cytochrome P450 isoforms
with estrone or 17â-estradiol alone or two estrogens in combination in the presence of NADPH as a cofactor
resulted in the formation of all eight estrogen dimers in varying quantities.
1. Introduction
E2 dimers.7 They are linked together through a diaryl ether bond
between the C-3 phenolic oxygen of one E2 and the C-2 or C-4
aromatic carbon of another E2.
The endogenous estrogens, such as 17â-estradiol (estradiol
or E2 a) and estrone (E1), are important female gonadal hormones
that have very diverse physiological and pathophysiological
actions.1 Metabolically, E2 and E1 undergo extensive metabolism
in vivo, including oxidation and conjugation (such as glucu-
ronidation, sulfonation, and O-methylation). It is known that
members of the cytochrome P450 (CYP) family are the major
enzymes catalyzing the NADPH-dependent oxidative metabo-
lism of endogenous steroidal estrogens to multiple hydroxylated
or keto metabolites.2,3 There is considerable experimental
evidence suggesting that some of the endogenous estrogen
metabolites, such as 4-hydroxy-E2, 15R-hydroxy-E2, 16R-
hydroxy-E1, 16-epiestriol (16â-hydroxy-E2), 17R-estradiol, and
2-methoxy-E2 (an O-methylated product of 2-hydroxy-E2 formed
by catechol-O-methyltransferase), may have unique biological
functions that are not associated with their parent hormones E2
and E1.4,5,12 These multiple metabolic pathways not only
determine the pharmacokinetic features of the endogenous
estrogens in various target tissues in the body but also diversify
the biological actions of endogenous estrogens in certain target
sites through metabolic formation of biologically active estrogen
derivatives.
On the basis of the structural information derived from these
two E2 dimers, we believed that several other structurally related
dimers between two E1 molecules or between one E2 and one
E1 could also be formed under the same in vitro metabolic
conditions. To test this hypothesis, we first chemically synthe-
sized all six additional putative dimers of E2 and E1 (compounds
15-20, structures shown in Figure 1). The synthesis used E2
and E1 as starting materials and followed a four-step procedure
with the Ullmann condensation reaction as a key step.8-10 Using
these newly synthesized estrogen dimers as reference standards,
we then demonstrated that incubations of human liver mi-
crosomes or recombinant human CYP isoforms with E2 and E1
as substrates (singly or in combination) in the presence of
NADPH as cofactor resulted in the formation of all eight
estrogen dimers in varying quantities.
2. Results and Discussion
2.1. Chemical Synthesis of Six Estrogen-Estrogen Dimers.
Using E1 and/or E2 as the starting material, we have synthesized
six new estrogen dimers that are covalently linked together
through a diaryl ether bond between a phenolic oxygen atom
of one E1 or E2 molecule and the C-2 or C-4 aromatic carbon
of another E1 or E2. The synthesis scheme is depicted in Figure
2. The synthetic method included four steps: (1) bromination
of E1 or E2 to yield 4-bromo-E1/E2 and 2-bromo-E1/E2 (com-
pounds 1-4, structures shown in Figure 1); (2) protection of
the C-3 phenolic hydroxyl group of the step 1 product to yield
the 3-O-benzyl ethers of 2- or 4-bromo-E1/E2 (compounds 5-8,
structures shown in Figure 1); (3) the Ullmann condensation
reaction to link C-2 or C-4 bromo-E1/E2 to the 3-O position of
another E1/E2 to yield the 3-O-benzyl ethers of the dimers
(compounds 9-14, structures shown in Figure 1);18 (4) removal
of the protective C-3 benzyl group by hydrogenation catalyzed
by Pd-C to yield the final estrogen dimers (compounds 15-
20, structures shown in Figure 1). In this synthetic procedure,
the diaryl ether linkage (step 3), which is also commonly known
as the Ullmann condensation reaction,3,8,9 is a key step. The
Using [3H]E2 as substrate, we recently demonstrated, for the
first time, the formation of over 20 nonpolar radioactive
metabolites of E2 under the commonly used NADPH-dependent
conditions for human CYP enzyme-mediated oxidative metabo-
lism of [3H]E2.5 Some of the nonpolar metabolite peaks appeared
to be preferentially formed in large quantities with certain human
CYP isoforms, most notably CYP3A4 and CYP3A5.6 Two of
the major metabolites have been successfully confirmed to be
* To whom correspondence should be addressed. Phone: 913-588-9842.
Fax: 913-588-8356. E-mail: BTZhu@kumc.edu.
† Present address: Department of Drug Metabolism, Abbott Laboratories,
Abbott Park, IL 60064.
a Abbreviations: E2, 17â-estradiol; E1, estrone; CYP, cytochrome P450;
GC-MS, gas chromatography-mass spectrometry; NADPH, nicotinamide
dinucleotide phosphate (reduced form); HPLC, high-pressure liquid chro-
matography; NMR, nuclear magnetic resonance; HRMS, high-resolution
mass spectrometry.
10.1021/jm0707323 CCC: $37.00 © 2007 American Chemical Society
Published on Web 10/09/2007

Enzymatic Formation of NoVel Estrogen Dimers
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5373
Figure 1. Chemical structures of the estrone (E1) and 17â-estradiol (E2) dimers. Compounds 15-20 are the E1 or E2 dimers that were chemically
synthesized in the present study. The other two compounds (21 and 22) are the two E2 dimers that were synthesized in our laboratory before.6
yields of steps 1, 2, and 4 were above 90% or even 95%, while
the yields of step 3 was only about 50%. Thus, the overall yields
for the synthesis of these estrogen dimers were slightly above
40%. For convenience, some relevant information concerning
the synthetic procedures is given below.
In the synthesis of 2 and 4-bromo-E1 (step 1) we initially
suspected that the C-17 keto group of E1 might not be stable in
the reaction. Thus, we tried to use a 1,3-dioxolane group to
protect the C-17 keto group before bromination.14,15 We later
found that using E1 directly (without the protection) for the
synthesis of dimers worked even better. No appreciable amounts
of the C-17-modified side products were detected when the C-17
keto group of E1 was not protected.
It is also of note that the use of finer granules of the substrates
(E1/E2 and N-bromosuccinimide) in the synthesis of 2- or
4-bromo-E1/E2 (step 1) would markedly facilitate the reaction.
If small clumps of N-bromosuccinimide (NBS) were used, brief
sonication of the reaction mixture would be helpful for
accelerating the reaction. Also, the flask should be continuously
shaken when N-bromosuccinimide was slowly added. Otherwise,
the yields of the dibromo byproducts may increase considerably.
The purification of 2-bromo-E1 from the mixture of 2- and
4-bromo-E1 turned out to be a rather difficult step. We tried to
use recrystallization, which was very successful in yielding
2-bromo-E2. However, 2-bromo-E1 just would not come out of
the solvents alone without the company of 4-bromo-E1. We tried
to use a variety of organic solutions for this purpose, including
some of the quite nonpolar solvents (e.g., chloroform and ethyl
acetate) as well as some of the polar protic solvents (such as
acetone, acetonitrile, acetic acid, ethanol, and methanol). Dif-
ferent combinations of those solvents were tested for crystal-
lization. Unfortunately, none was effective enough for this
purpose. Eventually, we settled on using a semipreparative
HPLC method to obtain high-purity 2-bromo-E1 (>99%) with
a mobile phase consisting of methanol, water, and acetonitrile
(v:v:v ) 2:1:2).
such as dehalogenated arene and homocoupled diaryls. In our
synthesis, in order to increase the stability of substrates and
also to avoid the formation of too many side products, we chose
to use relatively moderate temperatures (155-160 °C) with a
longer reaction time (72 h). It turned out that the reaction
proceeded slowly during the first 48 h, and the reaction appeared
to further slow down between 48 and 72 h. Between 72 and 96
h, the reaction proceeded very slowly, and thus, this reaction
was terminated at 72 h after it had begun, with a yield of slightly
over 50%.
Although the products from the Ullmann condensation
reaction could be used directly without purification for the last-
step reactions, we found that a rough separation of the products
using silica gel column chromatography was advantageous for
simplifying the procedures in the last step. This separation step
made the crystallization of the estrogen dimers in the last step
easier and faster, and it also yielded estrogen dimers with much
higher purity. In addition, the Pd-C catalyst that was used in
the last step would have a much longer recycling life when most
of the polymers and other catalysts were removed.
2.2. Formation of Nonpolar E1 Metabolites by Human
CYP Isoforms. As part of our continuing effort to expand the
knowledge on the human CYP isoform-mediated formation of
nonpolar estrogen metabolites, we determined in this study the
nonpolar E1 metabolites formed by 15 recombinant human CYP
isoforms (CYPs 1A1, 1A2, 1B1, 2A6, 2B6, 2C8, 2C9*1
(Arg144), 2C18, 2C19, 2D6, 2E1, 3A4, 3A5, 3A7, and 4A11)
using [3H]E1 as substrate and NADPH as cofactor. The same
shortened HPLC gradient condition (method B in Table 1),
which was optimized in our early study for the separation of
most nonpolar E2 metabolites,5 was used for quantifying the
various nonpolar [3H]E1 metabolites.
Representative HPLC chromatographs showing the detection
of nonpolar [3H]E1 metabolite peaks formed by representative
human CYP isoforms are shown in Figure 3. A total of more
than 20 radioactive nonpolar metabolite peaks were detected
after incubations of [3H]E1 with various human CYP isoforms.
For convenience of description, the major nonpolar E1 metabo-
lite peaks formed by human CYP isoforms were denoted as
It was noted earlier that the Ullmann condensation reaction
usually requires quite high temperatures (up to 300 °C), a long
reaction time (up to 3 days), and strong polar solvents.16,17 These
harsh conditions often also produce competitive side products

5374 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
Chen et al.
Figure 2. Overall scheme for the chemical synthesis of all estrogen dimers described in the present study.
N1 through N14 according to their elution order. The rates of
their formation by different CYP isoforms are summarized in
Figure 4.
recombinant human CYP isoforms was assayed. This observa-
tion suggested that N8 likely was formed by other CYP isoform-
(s) or, more likely, it was formed by non-CYP enzyme(s) or
formed nonenzymatically.
Among the nonpolar estrogen metabolites detected, five of
them, namely, N2, N3, N5, N6, and N7, were metabolically
formed by most human CYP isoforms at varying rates (Figure
4). In comparison, N1, N4, and N8-N14 were only formed
with a few of the CYP isoforms. CYP3A5 had the highest
catalytic activity for the formation of N11 (65 (pmol/nmol
P450)/min), followed by CYP3A4. CYPs 1A1, 1B1, 2B6, and
2C9*1 (Arg144) showed weak but detectable catalytic activity
for the formation of this nonpolar metabolite. All other CYP
isoforms did not have appreciable catalytic activity for the
formation of N11. A comparison with the retention times of
the eight synthesized estrogen dimers suggested that peak N11
contained a mixture of two E1 dimers (compounds 15 and 18,
structures shown in Figure 1). Information concerning their
structural identification was described in section 2.3.
When the formation of N11 is used as an example, it appeared
that its formation by 15 human CYP isoforms did not correlate
with their overall catalytic activity for the oxidative metabolism
of estrogens. For instance, CYP1A2 had a high catalytic activity
for the formation of 2-hydroxylated metabolites from E2 and
E1.11 However, no appreciable activity for the formation of N11
by this CYP isoform was detected. This observation suggested
that the overall catalytic activity of different human CYP
isoforms for the oxidative metabolism of estrogens may not
necessarily correlate with their ability to form certain nonpolar
estrogen metabolites.
In summary, a total of over 20 nonpolar estrogen metabolite
peaks were detected following incubations of [3H]E1 with human
CYP isoforms using NADPH as cofactor. The N11 was only
formed with a few human CYP isoforms (mainly CYP3A5),
and its formation was not correlated with the overall catalytic
activity of various human CYP isoforms for the oxidative
metabolism of E1.
Notably, the metabolite peak N8, which was detected as a
quantitatively major peak when [3H]E1 was incubated with
human liver microsomes (data not presented), was only a
quantitatively minor nonpolar peak when any of the 15

Enzymatic Formation of NoVel Estrogen Dimers
Table 1. HPLC Mobile Phase Gradients
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5375
NADPH as cofactor are shown in Figures 6 and 7. All eight
nonpolar estrogen dimer peaks were clearly shown, and their
retention times matched exactly the synthetic reference stan-
dards. The first and third peaks have already been unequivocally
confirmed in our early study to be the 2′-E2-3′-E2 dimer and
the 4′-E2-3′-E2 dimer.5 The other six peaks matched with the
six additional estrogen dimers that were newly synthesized in
the present study. Moreover, when the extracts containing the
metabolically formed nonpolar estrogens were analyzed with a
different HPLC method that used the Ultracarb column, both
UV and radioactivity detections also simultaneously detected
these nonpolar peaks (although not separated well from
each other) and they matched all reference standards (Figures
6 and 7).
Notably, when E2 was used as the single substrate, 2′-E2-
3′-E2 and 4′-E2-3′-E2 dimers (compounds 22 and 21, structures
shown in Figure 1) were the quantitatively predominant dimers
formed by CYP3A4 + b5 in the presence of NADPH as cofactor
(parts B and E of Figure 6). The peaks of the other six dimers
could also be detected, albeit not very clearly because of their
small quantities. The rates of formation of all these dimers
formed metabolically were summarized in Table 2 for com-
parison.
When E1 was used as the single substrate, the first six
nonpolar estrogen dimer peaks were generally much smaller
compared to those seen when E2 was used as substrate (parts E
and F of Figure 6). As expected, the rates of formation of 2′-
E1-3′-E1 and 4′-E1-3′-E1 (compounds 18 and 15, structures
shown in Figure 1) were relatively higher than the rates of
formation of the other six dimers. Moreover, the rate of
formation of 2′-E1-3′-E1 (compound 18, structure shown in
Figure 1) was higher than that of 4′-E1-3′-E1 (compound 15,
structure shown in Figure 1).
time
(min)
solvent A
(%)
solvent B
(%)
solvent C
(%)
curve
no.
Method A
0
20
30
40
50
60
70
40
40
30
25
20
10
5
30
30
35
35
30
30
30
30
30
35
40
50
60
65
1
3
3
3
3
3
Method B
0
8
15
28
38
68
68
64
59
57
40
30
30
0
16
16
18
21
21
30
35
35
50
16
16
16
18
20
22
30
35
35
50
16
1
9
8
2
6
6
6
1
1
70
105
165
175
190
68
Method Ca
0
8
15
28
38
68
68
64
59
57
40
30
30
0
16
16
18
21
21
30
28
28
50
16
16
16
18
20
22
30
42
42
50
16
1
9
8
2
6
6
6
1
1
70
105
150
165
180
68
a Solvent A: double-distilled water. Solvent B: pure methanol. Solvent
C: pure acetonitrile. In the mobile phase conditions A and B, a Phenomenex
Ultracarb column was used, but in method C, a Varian Microsorb-MV 100A
column was used.
In the NADPH-dependent metabolism of E1 + E2 mediated
by CYP3A4 + b5, the rate of formation of four heterodimers,
i.e., compounds 16, 17, 19, and 20 (structures shown in Figure
1), was markedly increased. This was as expected because when
both E1 and E2 were present at higher concentrations in the in
vitro metabolism system, the chances for these two substrates
to interact with the enzyme molecules would increase, and
consequently, product formation would also increase.
Similarly, when the human liver microsomes were incubated
under the same in vitro conditions where E1, E2, or E1 + E2
were used as substrates, all eight estrogen dimers were also
detected (Figure 7). Overall, the formation of all eight dimers
by human liver microsomes had patterns similar to those formed
by human CYP3A4 + b5. These observations suggest that
CYP3A4 is among the CYP isoforms present in human liver
that are largely responsible for the formation of nonpolar
estrogen metabolites (including the eight dimers).
2.3. Confirmation of the E1/E2 Dimers Formed in Vitro
by Human CYP3A4 and Human Liver Microsomes. In our
recent studies, two dimers of E2 (compounds 21 and 22,
structures shown in Figure 1) were confirmed to be metaboli-
cally formed in vitro following incubations of [3H]E2 with
human CYP3A4 in the presence of NADPH as cofactor.5
Because E1 and E2 have the same structural frame and their
only difference resides in the C-17 position, we speculated that
E1 most likely can also be metabolically activated to form the
same types of dimers that have been detected for E2. Moreover,
in addition to the formation of homodimers between two E2 or
E1 molecules, we suspected that the heterodimers between E1
and E2 can also be formed metabolically under the same in vitro
conditions. In an effort to provide a definitive answer to these
questions, we chemically synthesized all eight possible estrogen
dimers (described under sections 3.3-3.6), which were used
as reference standards here to help confirm the metabolic
formation of these dimers in vitro.
Using two HPLC separation conditions (methods B and C,
Table 1), which were modified from an HPLC condition used
in our early study,5 we demonstrated that all eight estrogen
dimers could be formed in vitro following incubations of human
CYP3A4 + b5 or human liver microsomes with E1, E2, or E1 +
E2 as substrate(s) in the presence of NADPH as cofactor. We
chose to use the human CYP3A4 + b5 as a representative CYP
isoform for additional study because it had the highest overall
catalytic activity for the formation of most of the E1 and E2
nonpolar metabolites in vitro, as shown in Figure 4 and
previously.5,7 Representative HPLC traces (with UV detection)
for the nonpolar estrogen metabolites formed by human
CYP3A4 + b5 using E1, E2, or E1 + E2 as substrate(s) and
It is of note that in recent years there is an increased interest
in studying the metabolic formation and biological functions
of certain nonpolar estrogen metabolites such as 2-methoxy-
4,18
E2
and estrogen-17â-fatty acid esters.19-22 Studies of the
naturally occurring estrogen-17â-fatty acid esters have suggested
that these endogenous estrogen derivatives have preferential
mitogenic and tumorigenic effects in the fat-rich mammary
tissue compared the other target sites, such as the uterus and
pituitary in the female rats.21,22 In comparison, E2 had a
significantly stronger effect than E2-17â-fatty acid esters in
stimulating the growth and/or tumorigenesis of the pituitary and
uterus in these animals.22 The nonpolar estrogen dimers represent
a new class of nonpolar estrogen metabolites. When the relative
rates for the in vitro formation of various nonpolar estrogen

5376 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
Chen et al.
Figure 3. Representative HPLC traces of metabolites of E1 by six representative CYP isoforms using NADPH as a cofactor. The separation was
carried out by using the mobile phase method A on a Phenomenex Ultracarb column (radioactivity detection). E1 was applied as substrate, and
different CYP isoforms were used as indicated in each figure. All the polar metabolites were eluted within the first 8 min and indicated by a brace
of each trace. The peak N11 with an asterisk was the mixture of two dimers of E1. The quantitatively predominant peaks were labeled by numbers,
and their peak areas were calculated for comparison (data are summarized in Figure 4).
(Na2SO4), and 10% palladium on carbon (Pd-C) were of ACS
grades and purchased from ACROS (through Fisher Scientific,
Atlanta, GA). Acetonitrile, chloroform-d (isotope D, 99.6%),
dimethyl sulfoxide-d6 (isotope D, 99.9%), N,O-bis(trimethylsilyl)-
trifluoroacetamide (BSTFA) with 1% trimethylchlorosilane (TMCS),
methanol-d4 (isotope D, 99.96%), scintillation cocktail (ScintiVerse
LC), and other solvents (HPLC grade or better) were purchased
from Fisher Scientific (Atlanta, GA).
The human liver microsomes (HLM HG03) and the 15 selec-
tively expressed human CYP isoforms (1A1, 1A2, 1B1, 2A6, 2B6,
2C8, 2C9, 2C18, 2C19, 2D6, 2E1, 3A4, 3A5, 3A7, and 4A11) were
obtained from BD Gentest (Woburn, MA). These human CYP
isoforms were expressed in insect cells that were selectively
transfected with a baculovirus expression system containing the
cDNA for each of the desired human CYP isoforms. The total
microsomal protein concentration, CYP content, CYP reductase
activity, cytochrome b5 content, and the specific catalytic activity
for the marker substrate(s) for each expressed CYP isoform were
listed in our recent study.11
3.2. Equipment. High-resolution mass spectra (HRMS) were
recorded by using a VG 70S analytical mass spectrometer. Methanol
solution or powder samples of the synthesized chemicals were used
for direct-probe HRMS. Gas chromatography-mass spectrometry
(GC-MS) spectra were obtained by using the Agilent Technology
model 6890N GC system, with a model 5973 mass selective
detector. BSTFA with 1% TMCS was used as a derivatization
reagent for GC-MS analysis. The nuclear magnetic resonance
(NMR) spectra were recorded on a Varian Inova 300 or 400
spectrometer. Chemical shifts were given as δ values with reference
to deuterium solvent.
metabolites were compared with the polar estrogen metabo-
lites,2,3,5,6,11 we were surprised to realize that the overall rates
of formation of the former metabolites by human liver mi-
crosomes or CYP isoforms were comparable to the overall rates
of formation of the latter metabolites. Given the high stability
and lipophilicity of these metabolically formed nonploar
estrogen metabolites, they are expected to have very long half-
lives in vivo and they may also have a preferential biological
activity in the fat-rich mammary tissues, in a similar way as
known for the endogenous estrogen-17â-fatty acid esters.21,22
It has been known for years that covalently attaching a bulky
nonpolar side chain to the C-7R or C-11â positions of an E2
molecule is an effective strategy to produce high-affinity
antiestrogens for therapeutic use in humans.15,23 It was reported
recently that different types of nonpolar estrogen dimers, which
were chemically linked together and have different structures
compared with the biologically formed ones described in the
present study, have varying degrees of binding affinity for the
estrogen receptors and also have cytotoxicity in a murine skin
cancer cell line.24 Given these intriguing earlier studies, it is
tempting to suggest that some of the nonpolar estrogen
metabolites formed by human CYP isoforms may be able to
function as endogenously formed “selective estrogen receptor
modulators” (SERMs). Certainly, more studies are needed to
determine whether these nonpolar estrogen dimers are formed
in vivo and also to determine what are their potential biological
functions.
The HPLC system used for separation of various estrogen
metabolites consisted of a Waters 2690 separation module (Waters,
Milford, MA), a radioactivity detector (â-RAM; IN/US Systems,
Inc., Tampa, FL), a Waters UV detector (model 484), and an
Ultracarb 5 ODS column (150 mm × 4.60 mm; Phenomenex,
Torrance, CA) or a Microsorb-MV 100A column (300 mm × 3.9
mm; Varian, Palo Alto, CA). The detailed solvent gradients used
for the simultaneous separation of both polar and nonpolar estrogen
metabolites on an Ultracarb column and those used for the selective
separation of the eight estrogen dimers on a Microsorb-MV 100A
column are described later in section 2.7.
3. Materials and Methods
3.1. Chemicals and Reagents. E1 and E2 were purchased from
Steraloids (Newport, RI). NADPH and ascorbic acid were obtained
from Sigma-Aldrich (St. Louis, MO). [2,4,6,7 -3H(N)]E1 (specific
activity of 65.5 Ci/mmol) and [2,4,6,7,16,17-3H]E2 (specific activity
of 123.0 Ci/mmol) were obtained from Perkin-Elmer Life and
Analytical Sciences (Boston, MA). NBS, potassium carbonate (K2-
CO3, anhydrous), benzyl bromide, anhydrous 4-picoline (4-meth-
ylpyridine), copper(II) oxide (CuO), anhydrous sodium sulfate

Enzymatic Formation of NoVel Estrogen Dimers
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5377
Figure 4. Rate of formation of nonpolar E1 metabolites catalyzed by 16 human CYP isoforms. The unit is (pmol/nmol CYP)/min. Peak N11 was
a mixture containing two E1 dimers (compounds 15 and 18, structures shown in Figure 1).
Figure 5. HPLC separation of the reference standards of the eight estrogen dimers using two different columns and two different mobile phase
conditions. Both HPLC traces were obtained by using a UV detector: (A) HPLC trace showing the partial separation of eight estrogen dimers by
using method B coupled with a Phenomenex Ultracarb column; (B) HPLC trace showing a better separation of the eight estrogen dimers using the
HPLC method C on a Varian Microsorb-MV 100A column.
3.3. Chemical Synthesis of 2-Bromo-E1/E2 and 4-Bromo-E1/
E2 (Compounds 1-4). E1 (5 mmol) was suspended evenly in
chloroform (50 mL), into which fine granules of NBS (5 mmol)
were then added as the bromine donor. The mixture was shaken
until it appeared to be a colorless transparent solution. The reaction
usually took only less than 5 min if the NBS is in fine granules.
The solution was then dried in a rotary evaporator to yield solids.
Upon crystallization of the dried crude products that were redis-
solved in pure ethanol, 4-bromo-E1 (compound 1, structure shown
in Figure 1) was first separated as white powder. 2-Bromo-E1
(compound 3, structure shown in Figure 1) was obtained by using
a semipreparative HPLC. This step of the reaction was rather
complete, and the yields for compounds 1 and 3 were between 45%
and 48%. GC-MS analysis of its trimethylsilylated (TMS) product
showed two molecular ions with masses (m/z) of 420 and 422
(purity >95%).

5378 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
Chen et al.
Figure 6. Representative HPLC traces showing the formation of nonpolar estrogen metabolites in which CYP3A4 + b5 was used as the enzyme.
E1 + E2 (A), E2 (B), or E1 (C) was used as the substrate. (A-C) Radioactivity detection was used. CYP3A4 at 140, 70, and 70 pmol/mL was used
to produce panels A, B and C, respectively. The separation was carried out by using the mobile phase method B on a Phenomenex Ultracarb
column (radioactivity detection). Panels D-F are the results from UV detection of the same metabolite extracts shown in panels A-C, respectively.
Here, the mobile phase method C and a Varian Microsorb column was applied. Note that larger-scale in vitro reactions (e.g., 2 mL in total volume)
were used in order to produce clearer HPLC traces (with UV detection) of all eight estrogen dimer peaks.
1
Compound 1: H NMR (400 Hz, CDCl3) 7.189 (1H, d, J )
8.4), 6.872 (1H, d, J ) 8.4), 5.553 (1H, s), 2.968 (2H, dd, J )
18.0, 6.4 Hz), 2.376 (2H, dd, J ) 6.4, 4.0 Hz), 0.901 (3H, s).
HRMS: 348.0725 (calculated) and 348.0723 (observed).
by crystallization from acetonitrile. The yields of this step were
90-92%. If pure products are not necessary, the whole reaction
mixture can be dried in a rotary evaporator and then in a vacuum
desiccator. The resulting powder can be used directly for the next
synthetic step.
1
Compound 3: H NMR (400 Hz, CDCl3) 7.342 (1H, s), 6.760
Compound 5: 1H NMR (400 Hz, CDCl3) 7.483 (2H, d, J ) 6.4
Hz), 7.386 (2H, t, J ) 8.0), 7.313 (1H, m), 7.190(1H, d, J ) 8.8
Hz), 6.781(1H, d, J ) 8.4 Hz), 5.152 (2H, s), 0.898 (3H, s), 3.065
(2H, dd, J ) 18.4, 6.0 Hz), 2.514 (2H, dd, J ) 19.2, 9.6 Hz).
HRMS: 438.1194 (calculated) and 438.1181 (observed).
Compound 7: 1H NMR (400 Hz, CDCl3) 7.469 (2H, t, J ) 7.6
Hz), 7.390 (2H, t, J ) 8.0 Hz), 7.258 (1H, s), 7.676 (1H, s), 5.114
(1H, s), 0.906 (3H, s), 2.829 (2H, dd, J ) 9.2, 4.4 Hz), 2.504 (2H,
dd, J ) 19.2, 9.2 Hz). HRMS: 438.1194 (calculated) and 438.1193
(observed).
GC-MS analysis of the 3-O-benzyl ethers of 4- or 2-bromo-E2
(compounds 6 and 8, structures shown in Figure 1) showed that
the molecular masses matched the calculated mass values (double
molecular m/z) 512 and 514) for these compounds and their purity
was >95%.
(1H, s), 5.293 (1H, s), 2.838 (2H, m), 2.507 (2H, dd, J ) 14.8, 8.4
Hz), 0.909 (3H, s). HRMS: 348.0725 (calculated) and 348.0713
(observed).
The same method was used to produce 4-bromo-E2 (compound
2, structure shown in Figure 1) with E2 as the starting material.
After 4-bromo-E2 was separated from the reaction mixture, the
solution was evaporated to dryness and then redissolved in
acetonitrile with mild sonication. 2-Bromo-E2 (compound 4,
structure shown in Figure 1) was then crystallized from the ethanol
solution. The structure and purity of both compounds were
confirmed by using GC-MS analysis of their TMS products
(double molecular m/z ) 494 and 496), and each had a purity of
>95%.
3.4. Synthesis of 3-O-Benzyl Ether of 2-Bromo-E1/E2 or
4-Bromo-E1/E2 (Compounds 5-8). The sample of 2- or 4-bromo-
E1/E2 (1.0 mmol) was first dissolved in acetonitrile (150 mL),
followed by addition of anhydrous potassium carbonate (10 mmol)
and benzyl bromide (M:M ) 1:1 with bromo-E1 or bromo-E2).
Under nitrogen gas protection, the reaction mixture was refluxed
with stirring for 2 h. After filtration of the mixture, the colorless
3-O-benzyl ethers of the starting materials could be readily obtained
3.5. Synthesis of 3-O-Benzyl Ethers of the Estrogen Dimers
(Compounds 9-14).13 Each of the products (compounds 5- 8, at
1.0 mmol) from the last synthetic step was dissolved in small
amount of 4-picoline (10 mL). E1 or E2 (1.0 mmol) was then added
as cosubstrate depending on which target estrogen dimers would
be synthesized. Cupric oxide (40 mg, 0.5 mM) was added as

Enzymatic Formation of NoVel Estrogen Dimers
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5379
Figure 7. Representative HPLC traces showing the formation of nonpolar estrogen metabolites. Human liver microsomes were used as the enzyme.
E1 + E2 (A), E2 (B), or E1 (C) was used as the substrate. (A-C) Radioactivity detection was used. Human liver microsomes at 4, 2, and 2 mg/mL
were used to produce panels A, B, and C, respectively. The separation was carried out by using the mobile phase method C on a Phenomenex
Ultracarb column (radioactivity detection). Panels D-F are the results from UV detection of the same metabolite extracts shown in panels A-C,
respectively. Here, the mobile phase method C and a Varian Microsorb column were used. Note that larger-scale in vitro reactions (e.g., 2 mL in
total volume) were used in order to produce clearer HPLC traces for UV detection of all eight estrogen dimer peaks.
Table 2. Rates of Metabolic Formation of Eight Nonpolar Estrogen Dimers Catalyzed by CYP3A4 + b5 and Human Liver Microsomes
enzyme: CYP3A4 + b5
((pmol/nmol of CYP)/min)
enzyme: human liver microsomes
((pmol/mg protein)/min)
dimer
substrate E1 + E2
substrate E2
substrate E1
substrate E1 + E2
substrate E2
substrate E1
15
16
17
18
19
20
21
22
12.3
21.9
22.6
19.8
13.7
26.1
16.8
28.8
0.8
3.1
2.1
5.6
3.7
52.3
4.8
49.2
2.4
0.4
0.4
7.1
0.6
1.4
0.5
1.6
8.5
11.3
12.2
11.5
7.8
10.8
9.3
14.0
1.2
4.0
4.5
4.3
3.3
26.9
4.5
31.3
3.9
0.6
0.9
9.0
0.5
0.8
0.3
1.2
catalyst. The whole reaction mixture was heated to 155-160 °C
with constant stirring under nitrogen for 72 h.
representative 3-O-benzyl ether of the estrogen dimers) was purified
by using an HPLC system with an isocratic mobile phase containing
methanol, water, and acetonitrile (v:v:v ) 2:1:2) and a C18
preparation column (Alltech, Apollo C18 (5 µm), 250 mm × 10
mm). NMR and HRMS analyses of this product were performed
to confirm its structure. The yield of this step was 50-55%.
Compound 9: 1H NMR (400 Hz, CDCl3) 7.255-7.220 (2H, m),
7.174-7.150 (1H, m), 7.089-7.050 (2H, m), 6.883 (1H, d, J )
8.8 Hz), 5.229 (1H, s), 5.016 (2H. s), 2.542-2.475 (2H, m), 0.920
(6H, s). HRMS: 628.3563 (calculated) and 628.3563 (observed).
3.6. Synthesis of Various Estrogen Dimers (Compounds 15-
20). Each of the products from the preceding steps (i.e., compounds
9-14, structures shown in Figure 2) was redissolved in ethanol in
After the mixture was cooled to nearly room temperature, ethyl
acetate (200 mL) was added to dilute the reaction mixture, which
was then filtered. The resulting filtrate was washed with 1 N HCl
(50 mL) three times to remove 4-picoline until the aqueous washing
phase appeared to be light-brown. The organic layer was dried with
anhydrous Na2SO4 and concentrated in a rotary evaporator. The
product mixture appeared to be a brown oily liquid. The product
was then roughly purified by going through a silica gel column
with a mixture of hexane and ethyl acetate (v:v ) 2:1) as the mobile
phase to remove most of the polymer impurities. The eluent was
dried and then placed in a vacuum desiccator. Compound 9 (a

5380 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
Chen et al.
a Parr flask in the presence of Pd-C (500 mg) as catalyst. The
mixture was kept shaken under hydrogen at 40 psi for 4 h. Then
most of the catalyst was removed after filtering through four layers
of the regular filter papers. The filtered catalyst was ready for reuse
after washing with pure ethanol. The solution was then concentrated
in a rotary evaporator to a final volume of approximately 20 mL.
The residue was then filtered using a 0.22 µm filtration membrane
to remove the remaining catalyst. The filtrate was collected into a
glass bottle and placed in a hood for slow crystallization to yield
the white powdered estrogen dimers. To obtain a small amount of
pure product, an HPLC system with an isocratic mobile phase
containing methanol, water, and acetonitrile (v:v:v ) 2:1:2) and a
C18 preparation column (Alltech, Apollo C18 (5 µm), 250 mm ×
10 mm) was employed to purify each dimer from the mixture. The
yield of this step was 95-98%.
pure methanol (solvent B), and pure acetonitrile (solvent C). In
our early studies, we have developed method A to quickly elute
all polar metabolites in the first 10 min as clusters (without much
separation) and then selectively separate most of the nonpolar
metabolites within the next 60 min. This method was also used in
the present study for the elution of various nonpolar metabolites
formed with E1 as enzymatic reaction substrate. In addition, we
have optimized a reported HPLC method,5 which was also
developed in our lab, to reach the best separation of the eight
nonpolar dimers of interest (method B in Table 1).
It is of note that the HPLC methods A and B could not fully
separate all eight chemically synthesized nonpolar estrogen dimers
on an Ultracarb column (Phenomenex, 15 cm × 4.6 mm, 5 µm,
ODS). Therefore, we further modified the separation conditions
by using a different mobile phase condition (method C in Table 1)
on a different HPLC column (Microsorb-MV 100A column, Varian
Analytical Instruments). With this modified HPLC elution condition,
all eight synthetic dimers could be better resolved in a single HPLC
run (a representative HPLC trace is shown in Figure 5). This method
enabled us to simultaneously quantify the formation of all eight
estrogen dimers formed by human liver microsomes or CYP
isoforms. Last, it is also of note that while method C in combination
with a Microsorb-MV 100 A column produced a better separation
of the eight synthetic nonpolar estrogen dimers, it had a rather low
efficacy in separating various polar estrogen metabolites (in the
first 60 min of elution).
1
Compound 15: H NMR (400 Hz, DMSO-d6) 7.102 (1H, d, J
) 8.8 Hz), 6.947 (1H, d, J ) 8.8 Hz), 6.840 (1H, d, J ) 8.0 Hz),
6.516 (1H, s), 6.423 (1H, s), 2.710-2.611 (2H, m), 0.799 (6H, s).
HRMS: 538.3083 (calculated) and 538.3077 (observed).
1
Compound 16: H NMR (400 Hz, MEOH-d4) 7.141 (1H, d, J
) 8.4 Hz), 7.052 (1H, d, J ) 8.4 Hz), 6.749 (1H, d, J ) 8.4 Hz),
6.529 (1H, d, J ) 8.8 Hz), 6.464 (1H, d, J ) 4.0 Hz), 3.650 (1H,
d, J ) 8.8 Hz), 2.817-2.743 (3H, m), 2.503-2.434 (3H, m), 0.867
(3H, s), 0.770 (3H, s). HRMS: 540.3240 (calculated) and 540.3255
(observed).
1
Compound 17: H NMR (400 Hz, CDCl3) 7.160 (1H, d, J )
8.4 Hz), 7.087 (1H, d, J ) 8.7 Hz), 6.846 (1H, d, J ) 8.4 Hz),
6.618 (1H, m), 5.116 (1H, s), 3.660 (2H, m), 2.713 (2H, dd, J )
14.8, 8.4 Hz), 0.887 (3H, s), 0.757 (3H, s). HRMS: 540.3239
(calculated) and 540.3228 (observed).
Acknowledgment. This study was supported by a grant from
the American Cancer Society (Grant No. RSG-02-143-01-CNE).
Part of the work described in this paper was completed when
the authors were at the University of South Carolina, Columbia,
SC.
Compound 18: 1H NMR (400 Hz, DMSO-d6) 7.346-7.226 (2H,
m), 6.886 (1H, s), 6.735 (1H, s), 6.618 (2H, dd, J ) 8.8, 3.2 Hz),
6.568 (1H, d, J ) 2.4 Hz), 4,146 (1H, s), 2.830 (1H, m), 0.891
(6H, s). HRMS: 538.3083 (calculated) and 538.3092 (observed).
1
Compound 19: H NMR (400 Hz, MEOH-d4) 7.306 (1H, m),
Supporting Information Available: The GC-MS separation
conditions25 and representative chromatographs of compounds 1-4,
6 and 8 are provided. This material is available free of charge via
the Internet at http://pubs.acs.org.
7.198 (1H, m), 7.076 (1H, d, J ) 8.8 Hz), 6.686 (1H, s), 6.546
(2H, s), 6.469 (1H, s), 3.557 (1H, t, J ) 8.0 Hz), 2.380 (2H, dd, J
) 19.2, 9.2 Hz), 0.805 (3H, s), 0.667 (3H, s). HRMS: 540.3240
(calculated) and 540.3258 (observed).
Compound 20: 1H NMR (400 Hz, CDCl3) 7.216 (1H, s), 7.187
(1H, s), 6.847 (1H, s), 6.732 (1H, s), 6.732 (4H, m), 5.269 (1H, s),
3.656 (1H, t, J ) 8.7 Hz), 2.459 (2H, dd, J ) 15.6, 10.4 Hz), 0.898
(3H, s), 0.749 (3H, s). HRMS: 540.3239 (calculated) and 540.3236
(observed).
References
(1) Cavalieri, E.; Frenkel, K.; Liehr, J. G.; Rogan, E.; Roy, D. Estrogens
as endogenous genotoxic agentssDNA adducts and mutations. J.
Natl. Cancer Inst., Monogr. 2000, 27, 75-93.
(2) Lee, A. J.; Kosh, J. W.; Conney, A. H.; Zhu, B. T. Characterization
of the NADPH-dependent metabolism of 17â-estradiol to multiple
metabolites by human liver microsomes and selectively expressed
human cytochrome P450 3A4 and 3A5. J. Pharmacol. Exp. Ther.
2001, 298, 420-432.
(3) Lee, A. J.; Mills, L. H.; Kosh, J. W.; Conney, A. H.; Zhu, B. T.
NADPH-dependent metabolism of estrone by human liver mi-
crosomes. J. Pharmacol. Exp. Ther. 2002, 300, 838-849.
(4) Zhu, B. T.; Conney, A. H. Functional role of estrogen metabolism
in target cells: review and perspectives. Carcinogenesis 1998, 19,
1-27.
(5) Lee, A. J.; Zhu, B. T. NADPH-dependent formation of polar and
nonpolar estrogen metabolites following incubations of 17â-estradiol
with human liver microsomes. Drug Metab. Dispos. 2004, 32, 876-
883.
(6) Lee, A. J.; Sowell, J. W.; Cotham, W. E.; Zhu, B. T. Chemical
synthesis of two novel diaryl ether dimers of estradiol-17â. Steroids
2004, 69, 61-65.
(7) Zhu, B. T.; Lee, A. J. NADPH-dependent metabolism of 17â-estradiol
and estrone to polar and nonpolar metabolites by human tissues and
cytochrome P450 isoforms. Steroids 2005, 70, 225-244.
(8) Weingarten, H. Mechanism of the Ullmann condensation. J. Org.
Chem. 1964, 29, 3624-3626.
(9) Williams, A. L.; Kinney, R. E.; Bridger, R. F. Solvent-assisted
Ullmannether synthesis. Reactions of dihydric phenols. J. Org. Chem.
1967, 32, 2501-2505.
(10) Ullmann, F. Another representation method of phenyl acid. Chem.
Ber. 1904, 37, 853-854.
3.7. Methods for Studying the Estrogen Dimers Formed in
Vitro by Human Liver Microsomes and CYP Isoforms. 3.7.1.
In Vitro Metabolism Conditions. The conditions used for the in
vitro metabolic formation of nonpolar [3H]E1 and [3H]E2 metabolites
were the same as those used in the past for studying the NADPH-
dependent oxidative metabolism of estrogens.6,7,11 Specifically, the
reaction mixtures consisted of the desired amount of human liver
microsomes (at a final concentration of 1 mg of protein/mL) or
human CYP isoforms (containing 70 or 140 pmol of CYP/mL), 50
µM E1 and E2 as substrate (singularly or in combination), 2 µCi of
radioactive E1 and/or E2, 2 mM NADPH, and 5 mM ascorbic acid
in a final volume of 500 µL of 0.1 M Tris-HCl/0.05 M HEPES
buffer, pH 7.4. The enzymatic reaction was initiated by addition
of human liver microsomes or a CYP isoform, and the incubation
was carried out at 37 °C for 20 min with periodic mild shaking.
The reaction was then arrested by placing the reaction tubes in ice-
cold water, followed immediately by extraction with 5 mL of ethyl
acetate. The supernatant was transferred to another set of test tubes
and dried under a stream of nitrogen. The resulting residue was
redissolved in 100 µL of methanol, and an aliquot (50 µL) was
injected into an HPLC system coupled with radioactivity and UV
detection for analysis of estrogen metabolite composition.
3.7.2. HPLC Analytical Conditions. In order to separate various
estrogen metabolites formed in vitro for quantification, different
mobile phase conditions had to be used. The HPLC methods
combined the use of three solvents in complex gradients, and the
three basic solvents included doubly distilled water (solvent A),
(11) Lee, A. J.; Cai, M. X.; Thomas, P. E.; Conney, A. H.; Zhu, B. T.
Characterization of the oxidative metabolites of 17â-estradiol and
estrone formed by 15 selectively expressed human cytochrome p450
isoforms. Endocrinology 2003, 144, 3382-3398.

Enzymatic Formation of NoVel Estrogen Dimers
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5381
(12) Kabat, G. C.; O’Leary, E. S.; Gammon, M. D.; Sepkovic, D. W.;
Teitelbaum, S. L.; Britton, J. A.; Terry, M. B.; Neugut, A. I.; Bradlow,
H. L. Estrogen metabolism and breast cancer. Epidemiology 2006,
17, 80-88.
(20) Hochberg, R. B. Biological esterification of steroids. Endocr. ReV.
1998, 19, 331-338.
(21) Mills, L. H.; Lee, A. J.; Parlow, A. F.; Zhu, B. T. Preferential growth
stimulation of mammary glands over uterine endometrium in female
rats by a naturally occurring estradiol-17â-fatty acid ester. Cancer
Res. 2001, 61, 5764-5770.
(22) Mills, L. H.; Lee, A. J.; Zhu, B. T. Naturally-occurring estradiol-
17â-fatty acid ester, but not estradiol-17â, preferentially induces the
development of mammary tumor in female ACI rats. Proc. Am. Assoc.
Cancer Res 2003, 44, 835-836.
(23) Zhang, J. X.; Labaree, D. C.; Hochberg, R. B. Nonpolar and short
side chain groups at C-11beta of estradiol result in antiestrogens. J.
Med. Chem. 2005, 48, 1428-1447.
(24) Berube, G.; Rabouin, D.; Perron, V.; N’Zemba, B.; Gaudreault, R.
C.; Parent, S.; Asselin, E. Synthesis of unique 17â-estradiol homo-
dimers, estrogen receptors binding affinity evaluation and cytocidal
activity on breast, intestinal and skin cancer cell lines. Steroids 2006,
71, 911-921.
(25) Adlercreutz, H.; Kiuru, P.; Rasku, S.; Wahala, K.; Fotsis, T. An
isotope dilution gas chromatographic-mass spectrometric method
for the simultaneous assay of estrogens and phytoestrogens in urine.
J. Steroid Biochem. Mol. Biol. 2004, 92, 399-411.
(13) Harrington, P. M.; Singh, B. K.; Szamosi, I. T.; Birk, J. H. Synthesis
and herbicidal activity of cyperin. J. Agric. Food Chem. 1995, 43,
804-808.
(14) Leese, M. P.; Hejaz, H. A.; Mahon, M. F.; Newman, S. P.; Purohit,
A.; Reed, M. J.; Potter, B. V. A-ring-substituted estrogen-3-O-
sulfamates: potent multitargeted anticancer agents. J. Med. Chem.
2005, 48, 5243-5256.
(15) Jiang, X. R.; Sowell, J. W.; Zhu, B. T. Synthesis of 7R-substituted
derivatives of 17â-estradiol. Steroids 2006, 71, 334-342.
(16) Smith, K.; Jones, D. A superior synthesis of diaryl ethers by the use
of Ultrasound in the Ullmann reaction. J. Chem. Soc., Perkin Trans.
1 1992, 407-408.
(17) Palomo, C.; Palomo, C.; Oiarbide, M.; Lopez, R.; Gomez-Bengoa,
E. Phosphazene P-4-Bu-t base for the Ullmann biaryl ether synthesis.
Chem. Commun. 1998, 19, 2091-2092.
(18) Zhu, B. T.; Conney, A. H. Is 2-methoxyestradiol an endogenous
estrogen metabolite that inhibits mammary carcinogenesis? Cancer
Res. 1998, 58, 2269-2277.
(19) Hochberg, R. B.; Pahuja, S. L.; Zielinski, J. E.; Larner, J. M. Steroidal
fatty acid esters. J. Steroid Biochem. Mol. Biol. 1991, 4, 577-585.
JM0707323