Study of pharmacokinetics of prenylflavonoids and dynamics of estrogen action in sera following ingestion of epimedium using validated, ultra sensitive cell based bioassays 2

CHAPTER 3 RESULTS 3.1 Measurement of combinatorial estrogenic activity using ultra-sensitive, cell-based ER-driven reporter gene assays 85 3.1.1 Responses of ERα- and β-driven reporte

CHAPTER 3

RESULTS

3.1 Measurement of combinatorial estrogenic activity using

ultra-sensitive, cell-based ER-driven reporter gene assays

85

3.1.1 Responses of ERα- and β-driven reporter gene assays to

estradiol and its metabolites

85

3.1.2 Specificity of ER-driven reporter gene assays 88 3.1.3 Validation of ERα and ERβ cell-based bioassays 90 3.1.4 Isoform selectivity of DPN, PPT and genistein 93

3.1.5 Estrogenicity of common flavonoids and Epimedium

prenylflavonoids

95

3.1.6 Anti-estrogenic activities of SERMs and ER-antagonist in

3.1.7 Combinatorial effects of binary mixtures of estradiol with

genistein and Epimedium prenylflavonoids

3.2.3 Validation of MCF-7 breast cancer cell proliferation assay 109

3.2.4 Induction of MCF-7 breast cancer cell growth by Epimedium

prenylflavonoids and genistein

111

3.2.5 Effects on cell proliferation of estrogen-induced of MCF-7

breast cancer cells by Epimedium prenylflavonoids, genistein

and 4-hydroxytamoxifen

113

3.3 Traditionally prepared aqueous Epimedium decoction does not

exert significant estrogencity in humans revealed using a panel of

sensitive, cell-based estrogen-responsive bioassays

116

3.3.1 ERα bioactivity and concentration of estrogenic aglycones in

a traditionally prepared Epimedium decoction

116

3.3.2 Estrogenicity of serum before and following ingestion of

estradiol valerate and Epimedium pubescens decoction

118

3.3.3 Correlations of estrogen-responsive bioassays with estrone

and estradiol levels in human sera

123

3.4 Lack of estrogenicity of a traditionally prepared Epimedium

decoction was due to low bioavailability of estrogenic Epimedium

3.4.2 Concentrations of icaritin and desmethylicarition in human

sera obtained after oral ingestion of a traditionally prepared

Epimedium decoction

134

3.5 Vast majority of Epimedium prenylflavonoids were conjugated

and rendered non-estrogenic following oral ingestion of a

prenylflavonoid-enriched, standardized Epimedium extract in

3.5.2 Unconjugated flavonoid content in rat sera following

3.6 Convergence of global gene expression profiles of MCF-7 breast

cancer cells treated with estradiol, genistein, Epimedium extract

and its compounds

160

3.6.1 Global gene expression profiles of estradiol, genistein,

4-hydroxytamoxifen, Epimedium extract and its compounds in

human MCF-7 breast cancer cells

160

3.6.2 Comparison of gene expression profiles of genistein,

Epimedium extract, icariside I, icaritin, desmethylicaritin to

estradiol

166

3.6.3 Gene functions specific to genistein, Epimedium extract,

icariside I, icaritin and desmethylicaritin

168

3.6.4 Effects of estradiol, genistein, 4-hydroxytamoxifen,

Epimedium extract and its compounds on GREB1 expression

in human MCF-7 breast cancer cells

179

3.1 Measurement of combinatorial estrogenic activity using ultra-sensitive,

cell-based ER-driven reporter gene assays

3.1.1 Responses of ER α- and β-driven reporter gene assays to estradiol and its metabolites

To examine their response to natural steroidal estrogens, ERα and ERβ stable

cells that the bioassays were based on were exposed separately to increasing doses of estradiol and its metabolites, namely, estrone and estriol, which were dissolved in Eagle’s minimum essential medium supplemented with 10% charcoal-stripped fetal calf serum for 24 h All three estrogenic ligands exhibited a sigmoidal dose-response behavior in both cell lines (Fig 10) Estrone and estriol were full agonists towards both ER isoforms as they elicited the same maximum activation level of the luciferase reporter gene as estradiol

Dose-response experiments with the standard estrogen, estradiol, indicated that the ERα stable cell line had an EC50 value of 85 pM for this compound (Table 4) Estradiol was a less potent activator of the ERβ cell line and its EC50 value was about 3-fold higher than that observed with ERα (232 and 85 pM for ERβ and ERα

respectively) (Table 4)

Estriol and estrone were less potent agonists for both ERα and ERβ compared

to estradiol (Table 4) The rank order in terms of descending estrogenic potencies for these three natural steroidal estrogenic ligands in both cell lines is estradiol > estriol >

Figure 10: Dose-dependent activation of the luciferase reporter gene in ER α and ERβ cells by estradiol and its metabolites

ERα (A) or ERβ (B) cells were exposed to increasing doses of estrone (─■─),

estradiol (─○─) and estriol (─×─) dissolved in Eagle’s minimum essential medium

supplemented with 10% charcoal-stripped fetal bovine serum for 24 h Results are expressed as a percentage relative to saturation dose of estradiol and presented as mean ± SEM

0 20 40 60 80 100 120

0 20 40 60 80 100 120

log[ligand]

Estrogenic Activity (% of 10 nM E2)

(A) ERα

(B) ERβ

Table 4: Estrogenic effects of natural estrogens, ER-isoform selective agents,

common flavonoids and Epimedium prenylflavonoids measured in-vitro with

ER-driven luciferase reporter genes in HeLa cells stably expressing ERα or ERβ

EC50: Half-maximal activity extrapolated from dose-response curves; numbers are

given in presented as the mean ± SEM from three determinations

Emax: Maximum estrogenic effect extrapolated from dose-response curves

NSA: No significant activity

3.1.2 Specificity of ER-driven reporter gene assays

In addition to estrogens, there are other endogeneous steroids present in the

blood in-vivo, for example, testosterone, cortisol, dihydrotestosterone and

progesterone Hence, it is necessary to test for cross reactivity between these endogeneous steroids with estrogens To do this, ERα and ERβ stable cells were

incubated with increasing doses of steroidal hormones in the presence of 50 pM estradiol and their effects on estradiol’s estrogenicity

Addition of dihydrotestosterone did not change the estrogenicity of estradiol as measured by ERα and ERβ bioassays (Fig 11) Similar addition of cortisol or progesterone also did not change the estrogenicity of the added estradiol However, dose-dependent increases in serum estrogenicity were detected when cells were treated with increasing doses of testosterone in the presence of estradiol (Fig 11) This increase was due to the conversion of testosterone to estradiol by aromatase in the cells, which was prevented by the addition of 50 µM of DL-aminoglutethimide, an aromatase inhibitor, to all samples during the assay DL-aminoglutethimide also

inhibits the transformation of androstenedione to estrone (Mak et al., 1999)

Figure 11: Effects of endogenous steroids on the estrogenicity of 50 pM of

estradiol in ER-driven reporter gene assays

ERα (A, B) and ERβ (C, D) stable cells were incubated with increasing doses of

progesterone (―■―), dihydrotestosterone (―□―), hydroxycortisone (―•―) and

testosterone (―○―) in the presence of 50 pM estradiol Dose-response curves were

obtained in the absence (A, C) and presence of 50 µM of DL-aminoglutethimide, AG

(B, D) Results are expressed as fold induction relative to vehicle treated control and

presented as the mean ± SEM, (* p<0.01)

3.1.3 Validation of ERα and ERβ cell-based bioassays

The inter- and intra-assay variations of both ERα and ERβ bioassays, were determined by using 0.1 nM and 0.5 nM of estradiol, respectively Inter-assay variation is defined as the relative SD from the mean of determinations done on 10 different occasions Intra-assay variation is the relative SD of the mean of 8 determinations in a single assay The sensitivity towards the standard estrogen, estradiol, for both assays was ascertained by determining the detection limit, which is the concentration of estradiol that induced a luciferase activity equivalent to the mean

plus three SD of vehicle (Legler et al., 1999)

For ERα, dose-response experiments with estradiol indicated that the detection limit was 8.45 pM Intra- and inter-assay variations near the EC50 value (0.1 nM) were about 6% and 14%, respectively As for ERβ, the detection limit, intra- and inter-assay variations were 13.1 pM, 8.1% and 16% respectively, which are relatively similar to that for ERα

To quantify the estrogenicity of serum samples obtained from volunteers who have ingested an estrogenic preparation during a clinical trial, calibration curves need

to be constructed alongside these samples Estrogenicity of these serum samples can

be expressed as estradiol equivalents, which can be derived by interpolation from the calibration curve Calibration curves using estradiol were constructed by adding increasing doses of estradiol to charcoal-treated serum The aromatase inhibitor, DL-aminoglutethimide, needs to be added to all samples at a final concentration 50 µM,

so as to minimize the conversion of endogenous testosterone to estradiol Resulting

best fits the points within the range of the test samples that maximizes the correlation coefficient using Graphpad Prism software

A representative ERα calibration curve generated by exposing ERα cells to increasing doses of estradiol is shown in Fig 12A The curve was best fitted with a power regression curve within the clinical range (R2 > 0.95) A representative ERβ calibration curve within the clinical range was best fitted with a exponential curve with a correlation coefficient of R2 > 0.95 (Fig 12B)

Figure 12: Representative calibration curves of ER-driven reporter gene

bioassays for the quantification of estrogenicity of ex-vivo serum samples

Calibration curves were constructed by adding increasing doses of estradiol to dextran-coated, charcoal-treated serum and then equilibrated for 4 h at 37ºC All serum samples were diluted to 20% with serum-free media before exposure to ERα and ERβ cells in duplicate The aromatase inhibitor DL-aminoglutethimide at a final concentration of 50 µM was added to every sample Calibration curves were fitted with a suitable regression method which maximizes the correlation coefficient within the range of the test samples Luminescence is given as relative light units (RLU)

0 25 50 75 100

0 250 500 750 1000

[E2], pM

y = 320 - [39185/(1- exp(314-x)/37]

r2 = 0.98

3.1.4 Isoform selectivity of DPN, PPT and genistein

To assess the ER-isoform selectivity of ERα and ERβ bioassays, compounds such as, DPN, PPT and genistein, which are non-steroidal compounds that have been reported to show ER isoform selectivity were tested

PPT was reported to be an ERα-selective ligand (Escande et al., 2006) As

shown in Fig 13A, PPT exhibited strong ERα selectivity, with EC50 value of 0.163

nM (Table 4) It showed negligible transactivation of the luciferase reporter gene in the ERβ cell line in the range of concentrations tested PPT showed a full ERα agonistic activity as its maximal activation of the luciferase reporter gene was similar

to that obtained for estradiol

DPN transactivated both cell lines with EC50 values of 101 nM and 2.16 nM for ERα and ERβ, respectively (Fig 13B) It is a ERβ-selective ligand that showed

~50 fold higher relative potency for the ERβ isoform DPN is a full ERα agonist for both ERα and ERβ cell lines

Genistein is a phytoestrogen present in soy that has been reported to be a weak

ERβ selective phytoestrogen (Escande et al., 2006) As shown in Fig 13C, genistein

showed ~7 fold higher potency for the ERβ isoform (EC50 values for ERα was 921

nM and ER β was 137 nM, Table 4) In contrast with PPT and DPN, genistein is a superagonist in both ERα and ERβ cell lines, where the maximal activation of the luciferase reporter gene was ~5-fold higher than that obtained for estradiol

Figure 13: Isoform selectivity of PPT, DPN and genistein

ERα (―•―) and ERβ (―○―) stable cells were exposed to increasing doses of (A) PPT; (B) DPN and (C) genistein (GEN), for 24 h Results are expressed as a percentage relative to saturation dose of estradiol and presented as the mean ± SEM

0 100 200 300 400 500

ERα ERβ

PPT

Estrogenic Activity (% of 10 nM E2)

0 20 40 60 80 100 120

3.1.5 Estrogenicity of common flavonoids and Epimedium prenylflavonoids

Estrogenic effects of four common flavonoids were examined and compared to estradiol using ERα and ERβ bioassays Flavonoids studied were the flavones (apigenin, luteolin) and flavonols (kaempferol, quercetin) commonly present in the diet These compounds were paradigms for phytoestrogens with 3 hydroxyl (apigenin),

4 hydroxyl (luteolin and kaempferol) and 5 hydroxyl (quercetin) groups, allowing the study of relative estrogenic activity contributed by these hydroxyl groups (Table 1)

The ERα effects exerted by the four flavonoids are ~ 5 to 6 orders of magnitude less potent than estradiol (Fig 14A) Relative bioactivities, in terms of

EC50 values and peak activities were - apigenin>kaempferol=luteolin>quercetin - corresponding to increasing number of hydroxyl groups Apigenin, kaempferol and luteolin induced superagonist effects, with relative peak activities that were 1.7- to 3.2-fold higher than that observed with maximal doses of estradiol Peak activity of quercetin was only half of that observed for estradiol

Apigenin, luteolin, kaempferol and quercetin were relatively more potent stimulators of ERβ than ERα, with EC50 value that was 2 to 3 times smaller than ERα (Table 4, Fig 14B) Apigenin with 3 hydroxyl groups and quercetin with 5 hydroxyl groups exhibited most and least relative ERβ potencies respectively Kaempferol and luteolin had potencies and peak activities that were intermediate between apigenin and quercetin Kaempferol was more potent than luteolin Superagonism was observed for apigenin, but not kaempferol and luteolin Quercetin was a weak stimulator of ERβ

not exhibit either ERα or ERβ activity Desmethylicaritin, icaritin, icariside I, and icariside II are partial ERα agonists which induced ERα activity that was about 45%

of the peak activity observed for estradiol (Fig 15A) The potency of the compounds were in the order desmethylicaritin > icaritin > icariside I > icariside II with their EC50values being 0.07, 0.84, 5.5 and 13 µM respectively (Table 4) Icaritin and icariside I were ERα-selective as these compounds did not stimulate ERβ activity (Fig 15B) Icariside II and desmethylicaritin exhibited slight ERβ activity corresponding to 20%

of that observed with estradiol

Figure 14: Estrogenic activity of common flavonoids in-vitro

(A) ERα or (B) ERβ cells were exposed to increasing doses of apigenin (―×―), kaempferol (―■―), luteolin (―∆―) and quercetin (―○―) Apigenin showed superagonist effects in both ERα and ERβ stable cells that are above the maximal activation observed for the ER cognate ligand, estradiol Kaempferol and luteolin are superagonists in ERα but not ERβ cells Quercetin is a partial agonist in both systems Results are expressed as a percentage relative to saturation dose of estradiol and presented as mean ± SEM

0 50 100 150 200

K L Q

Estrogenic Activity (% of 10 nM E2)

Figure 15: Estrogenic activity of Epimedium prenylflavonoids in-vitro

HeLa cells stably expressing ERα (A) and ERβ (B) proteins and an ER-driven luciferase reporter gene, pERE4-Luchygro were exposed to increasing concentrations of estradiol (□), icariin (×), icariside II (○), icariside I (●), icaritin (▲) and desmethylicaritin (Δ) Epimedium prenylflavonoids preferentially

activate the ERα isoform and icariside I and icaritin are both ERα-selective agents Those that activated the ER showed partial agonistic activity Estrogenic activity of ligands is expressed as a percentage of maximal effect observed with 10 nM estradiol

0 20 40 60 80 100 120

DICTICT

ICAICAR IICAR IIE2

ERα

Estrogenic Activity (% of 10 nM E2)

-14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 0

20 40 60 80 100

3.1.6 Anti-estrogenic activities of SERMs and ER- antagonist in ERα and ERβ cell lines

The ability of ERα and ERβ bioassays to detect anti-estrogenic activity was studied by exposing cells to 4-hydroxytamoxifen, raloxifene and ICI182,780 in the presence of EC50 concentration of estradiol (Fig 16)

Both 4-hydroxytamoxifen and ICI 182,780 showed relatively similar antagonistic activities towards the two stable cell lines (Fig 16A, B) They were also more antagonistic towards the ERβ isoform where ICI 182,780 and 4-hydroxytamoxifen showed ~3 and ~2 fold more potent estrogen antagonistic activity towards ERβ over ERα isoform

On the other hand, raloxifene was a more potent ERα antagonist with ~500 times more selectivity for ERα over ERβ Interestingly, raloxifene was the weakest ERβ but the most potent ERα antagonist among the three compounds tested

Figure 16: Anti-estrogenic activities of SERMs and ER-antagonist in the presence of EC 50 concentration of estradiol

ERα (A) or ERβ (B) cells were exposed to increasing doses of ICI 182,780 (―○―), 4-hydroxytamoxifen (―•―) and raloxifene (―×―) in 10% serum in the presence of 0.1 nM (ERα) and 0.5 nM (ERβ) of estradiol, respectively Results are expressed as a percentage relative to saturation dose of estradiol and presented as the mean ± SEM Both 4-hydroxytamoxifen and ICI 182,780 showed relatively similar antagonistic activities towards the two stable cell lines and were more antagonistic towards the ERβ isoform Raloxifene was a more potent ERα antagonist

0 10 20 30 40 50 60

ERβ ERα ICI 182,780

Estrogenic Activity (% of 10 nM E2)

(A)

10 20 30 40 50

60

Raloxifene

Estrogenic Activity (% of 10 nM E2)

(C) 0 10 20 30 40 50

3.1.7 Combinatorial effects of binary mixtures of estradiol with genistein and

Epimedium prenylflavonoids

To study combinatorial effects of flavonoids and estradiol, binary mixtures

containing increasing doses of genistein and Epimedium prenylflavonoids in the

presence of an EC50 dose of estradiol were separately tested in ERα and ERβ bioassays

Genistein was the only compound tested that showed agonistic effects in the presence of estradiol Luciferase induction in treatments where genistein was incubated with estradiol in ERα and ERβ cell lines was higher than the maximal level elicited by 10 nM of estradiol (Fig 17A & 18A)

Of the five Epimedium prenylflavonoids, icariin, which is a diglycoside, did

not show any agonistic or antagonistic activity towards estrogen-induced luciferase activity in the concentration range tested (Fig 17B & 18B)

The remaining four Epimedium prenylflavonoids, namely, monoglycosides,

icariside I and icariside II and aglycones, icaritin and desmethylicaritin, did not exhibit significant agonistic activity in the presence of estradiol in both cell lines (Fig 17C-D, Fig 18C-D) These four compounds elicited reduced estrogen-induced luciferase activity only at higher doses (> 10 µM), which was more apparent in the ERβ than ERα cell line

Figure 17 : ERα combinatorial effects of binary mixtures of estradiol with

genistein and Epimedium prenylflavonoids

ERα stable cells were incubated with increasing doses of genistein (A), icariin (B), icariside I (C), icariside II (D), icaritin (E) and desmethylicaritin (F) in the presence of 0.1 nM of estradiol for 24 h Genistein, in combination with estradiol, showed

superagonist effects All Epimedium prenylflavonoids, except icariin, showed antagonism towards estrogen-induced activity at doses > 10 µM Results are

expressed as a percentage relative to saturation dose of estradiol and presented as the mean ± SEM, (* p<0.01)

*

log[ligand], M

(F)

Figure 18 : ERβ combinatorial effects of binary mixtures of estradiol with

genistein and Epimedium prenylflavonoids

ERβ stable cells were incubated with increasing doses of genistein (A), icariin (B), icariside I (C), icariside II (D), icaritin (E) and desmethylicaritin (F) in the presence of 0.5 nM of estradiol for 24 h Genistein, in combination with estradiol, showed

superagonist effects All Epimedium prenylflavonoids, except icariin, showed

antagonism towards estrogen-induced activity at doses > 10 µM Results are expressed as a percentage relative to saturation dose of estradiol and presented as the mean ± SEM, (* p<0.01)

25 50 75 100

*

*

(F) 0

100(B)

0 25 50 75 100

3.2 Combinatorial cell proliferative effects on estrogen-responsive MCF-7

breast cancer cells

3.2.1 MCF-7 breast cancer cell proliferative effects of estradiol and its

metabolites

A hallmark physiological response to estrogenic stimuli is the proliferation of

cells in-vivo This can be studied in-vitro by using breast cancer cells such as MCF-7

cells which express mainly endogenous ERα and are known to respond to estrogens and antiestrogens to modulate cell proliferation The MCF-7 cell proliferation assay, also known as the ‘E-screen’, is therefore used to study the physiological endpoint of estrogen action and also discrimination of agonists from antagonists Relative cell numbers were determined by Invitrogen’s CyQuant system and relative cell proliferation were expressed as fold-increase in fluorescence of test compound-treated cells compared to that of vehicle treated cells, which is expressed as 1-fold

The cell proliferative effects of estradiol were examined together with two of its metabolites, namely, estrone and estriol Estradiol was found to dose-dependently induce ERα-positive MCF-7 cell proliferation, reaching a peak level at 100 pM, with

an EC50 value of about 4 pM (Fig 19), indicating the sensitivity of this assay towards stimulation of estrogenic compounds Estrone and estriol also induced cell proliferation and their maximal cell proliferation levels were similar to what was observed for estradiol, suggesting that they are full estrogen agonists Based on their

EC50 values presented in Table 5, the potency of these three endogeneous estrogens were in the order estradiol > estriol > estrone

Figure 19: MCF-7 breast cancer cell proliferative effects of estradiol and its metabolites

MCF-7 breast cancer cells were treated with increasing doses of estrone (―•―), estradiol (―ο―) and estriol (―■―) dissolved in Eagle’s minimum essential medium supplemented with 5% charcoal-stripped fetal calf serum for 6 days Relative cell proliferation is expressed as fold induction over the vehicle control and presented as the mean ± SEM

-14 -13 -12 -11 -10 -9 -8 -7 -6 -50.5

1.0 1.5 2.0 2.5

Table 5: Effects on the proliferation of MCF-7 breast cancer cells by natural

estrogens, common flavonoids and Epimedium prenylflavonoids measured vitro

in-EC50: Half-maximal activity extrapolated from dose-response curves

Emax: Maximum estrogenic effect extrapolated from dose-response curves

NSA: No significant activity

Compound

EC50 (nM)

Emax (fold induction over vehicle) Natural estrogens

3.2.2 Effects on estrogen-induced breast cancer cell growth by endogenous

Similar to what was observed for ERα and ERβ bioassays reported in Fig 11, the addition of dihydrotestosterone, progesterone and hydroxycortisone in three increasing doses did not change the estrogenicity of estradiol in this assay (Fig 20)

Similarly, dose-dependent increases in serum estrogenicity were detected when cells were incubated with increasing doses of testosterone in the presence of estradiol (Fig 20) This increase was due to the conversion of testosterone to estradiol

by aromatase in the cells, which was prevented by the addition of 50 µM of aminoglutethimide, an aromatase inhibitor, to all samples during the assay

DL-Figure 20: Effects on estrogen-induced MCF-7 breast cancer cell growth by endogenous steroids

MCF-7 breast cancer cells were incubated with increasing doses of progesterone (―■―), dihydrotestosterone (―□―), hydroxycortisone (―•―) and testosterone (―○―) for 6 days in the presence of 50 pM estradiol Dose-response curves are obtained in the absence (A) and presence of 50 µM of DL-aminoglutethimide (B) Results are expressed as pM estradiol equivalents and presented as the mean ± SEM, (* p<0.01)

0 20 40 60 80

PGDHTHCT

0 50 50 50 50 pM E2

0 0 10 100 1000 pM steroid

0 20 40 60 80

(A) − AG

(B) + AG

3.2.3 Validation of MCF-7 breast cancer cell proliferation assay

The inter- and intra-assay variations of MCF-7 breast cancer cell proliferation assay were determined by using 10 pM of estradiol Inter-assay variation was defined

as the relative SD from the mean of determinations done on 10 different occasions Intra-assay variation was defined as the relative SD of the mean of 8 determinations in

a single assay The sensitivity towards estrogenic ligands of the assay was ascertained

by determining the detection limit, which was the concentration of the ligand that

induced a luciferase activity equivalent to the mean plus three SD of vehicle (Legler et

al., 1999)

Dose-response experiments with estradiol indicated that MCF-7 breast cancer cells had a detection limit of 0.112 pM Intra- and inter-assay variations determined using a dose of 10 pM estradiol were about 3% and 8%, respectively

The calibration curve for the 7 assay was constructed by exposing

MCF-7 breast cancer cells to increasing doses of estradiol encountered physiologically The calibration curve was fitted using the Boltzmann sigmoidal model in the dynamic range of estradiol concentrations tested (8 to 258 pM estradiol) (Fig 21) Higher concentrations of estradiol showed saturation of estrogenicity

Figure 21: Representative calibration curve of MCF-7 cell proliferation assay for

the quantification of estrogenicity of ex-vivo serum samples

A calibration curve is constructed by adding increasing doses of estradiol to treated pooled human male serum and allowed to equilibrate for 4 h at 37ºC before being diluted to 20% with serum-free media before exposure to MCF-7 breast cancer cells in duplicate DL-aminoglutethimide, an aromatase inhibitor is added to every sample to give a final concentration of 50 µM

y = 3204 + 69.18x -0.1054x2

r2 = 0.98

3.2.4 Induction of MCF-7 breast cancer cell growth by Epimedium

prenylflavonoids and genistein

Effects on MCF-7 breast cancer cell growth by genistein, a soy-derived

phytoestrogen and prenylflavonoids found in Epimedium, namely, icariin, icarisides I and II, icaritin and desmethylicartin, were examined and comparisons were made

against estradiol

From Fig 22, genistein induced MCF-7 breast cancer cell growth in a dependent manner It was the most potent phytoestrogen tested but was a weaker estrogen than estradiol Genistein is a full agonist as its maximal cell proliferating effect is similar to what was observed for estradiol

dose-The two aglycones found in Epimedium, icaritin and desmethylicaritin, were

weaker estrogens compared to estradiol and genistein Desmethylicaritin is a more potent estrogenic aglycone than icaritin and both compounds are also full estrogen agonists like estradiol and genistein in this assay system

Icariin, icarisides I and II are glycosides found in Epimedium Icariin did not

induce any appreciable cell growth in the range of concentrations tested Icarisides I and II, on the other hand, were weakly estrogenic These two compounds are partial agonists as their maximal cell proliferative effects were lower than what was observed for estradiol

Figure 22: Induction of MCF-7 breast cancer cell growth by estradiol, genistein,

4-hydroxytamoxifen and Epimedium prenylflavonoids

MCF-7 breast cancer cells were treated with increasing doses of estradiol (―■―), genisten (―□―), desmethylicaritin (―•―), icaritin (―○―), icariside I (―▲―), icariside II (―∆―), icariin (―◊―) and 4-hydroxytamoxifen (―×―) which were dissolved in Eagle’s minimum essential medium supplemented with 5% charcoal-stripped fetal calf serum for 6 days Relative cell proliferation is expressed as fold induction over the vehicle control and presented as the mean ± SEM

3.2.5 Effects on cell proliferation of estrogen-induced of MCF-7 breast cancer

cells by Epimedium prenylflavonoids, genistein and 4-hydroxytamoxifen

The ability of MCF-7 breast cancer cells to detect anti-estrogenic activity was accessed by exposing cells to 4-hydroxytamoxifen and genistein in the presence of 50

Of the five Epimedium prenylflavonoids, icariin, a diglycoside, did not show

any significant agonistic or antagonistic activity towards estrogen-induced cell proliferation of MCF-7 breast cancer cells (Fig 24)

Similar to genistein, the remaining four Epimedium prenylflavonoids, namely,

monoglycosides, icariside I and icariside II and aglycones, icaritin and desmethylicaritin, did not exhibit appreciable additive estrogenic effects at lower

doses (< 1 µM) and a reduction in relative cell numbers for four Epimedium

prenylflavonoids was observed only at higher doses (> 1 µM) (Fig 24)

Figure 23: Growth inhibition of estrogen-induced of MCF-7 breast cancer cells

by 4-hydroxytamoxifen and genistein

MCF-7 breast cancer cells are treated with increasing doses of 4-hydroxytamoxifen (A) and genistein (B) in the presence of 50 pM of estradiol for 6 days Relative cell proliferation is expressed as fold induction over vehicle and presented as the mean ± SEM, (* p<0.01)

0.0 0.5 1.0 1.5 2.0

0.0 0.5 1.0 1.5

(A)

(B)

Figure 24: Effects on cell proliferation of estrogen-induced of MCF-7 breast

cancer cells by Epimedium prenylflavonoids

MCF-7 breast cancer cells are incubated with increasing doses of icariin (A), icariside

I (B), icariside II (C), icaritin (D) and desmethylicaritin (E) in the presence of 50 pM

of estradiol for 6 days Relative cell proliferation is expressed as fold induction over vehicle and presented as the mean ± SEM, (* p<0.01)

0.0 0.5 1.0 1.5 2.0

log[ICA], M

0.0 0.5 1.0 1.5 2.0

1.0 1.5 2.0

*

3.3 Traditionally prepared aqueous Epimedium decoction does not exert

significant estrogencity in humans revealed using a panel of sensitive, cell-based

estrogen-responsive bioassays

3.3.1 ERα bioactivity and concentration of estrogenic aglycones in a traditionally

prepared Epimedium decoction

A standardized water decoction of Epimedium pubescens was prepared by

soaking and boiling grounded leaves in water To assess its bioactivity and concentration of active constituents, the decoction was evaporated and the residue was dissolved in DMSO Serial dilution was done before performing LC/MS and bioassay measurements

The concentrations of major flavonoid glycosides and the two active aglycones

were determined using methods previously described (Shen et al., 2007) The dried

water decoction contained 2.00 mg/g icariin, 119 µg/g icaritin and 31 µg/g desmethylicaritin

The extract was found to be estrogenic as shown by the dose-dependent increase in estrogenicity when ERα cells are incubated with increasing concentrations

of the reconstituted extract (Fig 25) The extract showed a maximum ERα bioactivity

up to 50% of the maximum observed with a saturating dose of estradiol Higher doses

of the extract were not tested due to insolubility and precipitation in cell culture media

Figure 25: ERα bioactivity of a traditionally prepared Epimedium pubescens

decoction

A standardized water decoction of Epimedium pubescens was prepared by grinding

dry leaves into powder and soaked in cold water for 30 min After soaking, the mixture was then boiled for 30 min Herb residue was removed by filtration and the volume of decoction reduced by further boiling The water extract was aliquotted into glass bottles and stored at 4°C overnight The decoction was evaporated and the residue was dissolved in DMSO and then serially diluted before adding to Eagle’s minimum essential medium supplemented with 10% charcoal-stripped fetal bovine serum Results are expressed as a percentage relative to saturation dose of estradiol and presented as the mean ± SEM

0 20 40 60 80 100 120

µg/mL

3.3.2 Estrogenicity of serum before and following ingestion of estradiol valerate

and Epimedium pubescens decoction

Eight male subjects were enrolled and completed the study The subjects, of Chinese (n=6) and Indian (n=2) ethnic origins, were aged 34±3.4 yr and had mean body mass index of 24±0.97 kg/m2 One subject developed diarrhea after

administration of estradiol valerate and Epimedium decoction; this adverse event was judged not to be related to the interventions All serum samples were assayed ex-vivo

for levels of estrone and estradiol by LC-MS/MS, for ERα and ERβ bioactivity and MCF-7 breast cancer cell proliferative effects

Male subjects were used in this study so as to reduce interference from endogenously produced estrogens, which may mask the effects of administered estrogenic drugs Post-menopausal women with low basal estrogen production may be

a more physiological model considering that one possible utility of the Epimedium

decoction is for hormone replacement therapy However, these women are likely to be elderly, frail and have co-morbidities As such, they may not meet the study’s inclusion criteria which require healthy volunteers able to donate sufficient amounts

of blood for the bioassays Since one use for the Epimedium decoction is hormone

replacement therapy in postmenopausal women, it is possible that differences in metabolism and plasma transport of steroids between men and women may affect applicability of the bioassays However, this possibility has been reduced by the use

of the aromatase inhibitor DL-aminoglutethimide to prevent conversion of endogenous testosterone to estradiol by aromatase

metabolism of the prodrug to bioactive estrogens (Fig 26A, B) As expected, the levels of estrone (Fig 26B) rose to over 3 times that observed with estradiol (Fig 26A) Due to first pass metabolism, estradiol is rapidly metabolized to estrone in the liver

Examination of serum samples, obtained in the baseline period before drug treatments, showed a 15 to 19% dip in estrogenic activity of the two samples taken at evening and night compared to the three day time samples This dip in estrogenic activity was observed for ERα (-15%), ERβ (-15%) and MCF-7 (-19%) assays, corresponding to the night dip in estradiol (-10%) and estrone (-14%) levels In terms

of estradiol equivalents, the daytime baseline estrogenicity was 55.3±2.5 pM for ERα, 28.5±2.8 pM for ERβ and 119.1±9.3 pM for MCF-7, compared to 86.3±5.6 pM estradiol when measured by mass spectrometry The secretion of testosterone and estradiol follows an intrinsic diurnal rhythm with a clear nocturnal and early morning rise This is driven by similar diurnal patterns in GnRH pulse generator and LH secretion from the pituitary gland

Ingestion of estradiol valerate resulted in highly significant increases (p<0.001)

in the ratio of adjusted GMR of AUC over baseline of 67%, 67% and 23% for the ERα, ERβ, and MCF-7 bioassays respectively (Table 6) After adjustment for endogenous diurnal variation in estrogenicity, remarkably similar estrogenicity profiles were observed with estradiol valerate for the different bioassays (Fig 27) The activity profiles obtained with ERα, ERβ, or MCF-7 bioassays closely paralleled

Administration of Epimedium decoction led to a 6% increase in ERα adjusted GMR of AUC over baseline (p<0.05) (Table 6) The significance of this small increase is uncertain because this increase was due to the relatively large relative contribution of the 24 h timepoint to the AUC calculations There was a trend, not reaching significance (P=0.08), for reduction in MCF-7 growth after decoction No changes in estrone/estradiol concentrations, and ERβ activities were observed

following Epimedium administration

Androgen receptor bioassays were performed to assess the effect of drug treatment on endogenous androgen production Androgen bioactivity in patient sera,

as measured by AR-driven reporter genes, remained stable following drug

administration of indicating that estradiol valerate or Epimedium do not significantly

change the bioactivity of androgenic compounds in sera (Fig 26F)

Figure 26: Estrogenicity in serum following ingestion of estradiol valerate or

Epimedium decoction

Following baseline blood sampling (—▲—), subjects (n=8) were administered estradiol valerate (—●—) and a standardized water decoction of Epimedium pubescens (—o—) in a randomized sequence Serum samples were sampled at

indicated time points (h) following interventions and assayed for (A) estradiol and (B) estrone (E1) concentrations via LC-MS/MS; and (C) ERα, (D) ERβ, (E) breast cancer cell (MCF-7) and (F) androgen receptor (AR) bioactivities Data points were the means±SEM and expressed as pM estradiol equivalents