Development of the β-lactam type molecular scaffold for selective estrogen receptor α modulator action: synthesis and cytotoxic effects in MCF-7 breast cancer cells

Abstract The estrogen receptors (ERα and ERβ) which are ligand inducible nuclear receptors are recognized as pharmaceutical targets for diseases such as osteoporosis and breast cancer. There is an increasing interest in the discovery of subtype Selective Estrogen Receptor Modulators (SERMs). A series of novel β-lactam compounds with estrogen receptor modulator properties have been synthesized. The antiproliferative effects of these compounds on human MCF-7 breast tumor cells are reported, together with binding affinity for the ERα and ERβ receptors. The most potent compound 15g demonstrated antiproliferative effects on MCF-7 breast tumor cells (IC50 = 186 nM) and ERα binding (IC50 = 4.3 nM) with 75-fold ERα/β receptor binding selectivity. The effect of positioning of the characteristic amine containing substituted aryl ring (on C-4 or N-1 of the β-lactam scaffold) on the antiproliferative activity and ER-binding properties of the β-lactam compounds is rationalized in a molecular modeling study.


Introduction
Breast cancer is by far the most frequent cancer among women globally, with an estimated 1.67 million new cancer cases diagnosed in 2012 ranking second overall (25% of all female cancers), and accounting for 521 817 deaths 1,2 . Incidence rates vary from 27 per 100 000 women in Middle Africa and Eastern Asia to 96 per 100 000 women in Western Europe 3 . In Europe, breast cancer is the most frequent cancer in women with 464 000 new diagnoses in 2012, which accounts for 29% of all new female cancers in Europe. One in three people in Ireland will develop cancer during their lifetime. Irish statistics note that breast cancer now accounts for 32% of all cancers in women in Ireland, with 2942 new diagnoses in 2013 4 .
Breast cancers are classified as estrogen receptor (ER) positive or negative with 70-80% of all primary breast tumors being ER positive, which is a less aggressive type than ER negative breast cancer 5 . The first antiestrogen to show positive clinical results was tamoxifen (1), (Figure 1), a synthetic non-steroidal antiestrogen, which was approved by the Food and Drug Administration in 1977 for the treatment of women with advanced breast cancer and several years later for adjuvant treatment of primary cancer 6,7 . Tamoxifen is extensively metabolized by the human hepatic cytochrome P450 enzyme system into several metabolites including 4-hydroxytamoxifen (2) and 4-hydroxy-N-desmethyltamoxifen (endoxifen), (3), Figure 1 8,9 . They are $100-fold more potent as antiestrogens than tamoxifen and are most likely contributors to the base antiproliferative activity observed with tamoxifen 10 . The metabolite norendoxifen was shown to have dual aromatase inhibitory and estrogen receptor modulatory activities 11 . Novel tamoxifen analogs that avoid CYPD6 metabolism have been recently reported 12 . The SERM raloxifene (4, Figure 1), a 2,3-disubstitued benzothiophene containing compound, is a potent antiestrogen that binds to the ER with an affinity higher than that of tamoxifen or 4-hydroxytamoxifen and equal to that of estradiol 13,14 . Raloxifene was approved for the treatment of osteoporosis in 1997 and has shown modest activity in ERpositive breast cancer while lacking the increased risk for endometrial cancer associated with the use of tamoxifen 15 . Aromatase inhibitors such as the non-steroidal agent's letrozole and anastrozole are reported to be more efficacious than tamoxifen as a first-line therapy, and are useful for second-line therapy and against tamoxifen-resistant tumors 16,17 . The steroid fulvestrant (5), Figure 1, is also effective as second line therapy against advanced breast cancer in patients who develop resistance to tamoxifen 18 . Due to its pure antiestrogen activity, it is devoid of endometrial stimulation and therefore the risk of endometrial cancer; similarly it does not possess the positive side effects on the skeletal and cardiovascular systems of SERMs.
There are two types of ER assigned as ERa and ERb. A comparison of the amino acid sequence of ERa and ERb shows that they share the same functional domain architecture, and the full length residues are $50% identical. The high homology between the DNA binding domain (96%) suggests that ERa and ERb are expected to bind various estrogen response elements (EREs) with similar specificity and affinity and therefore interact with, and activate the same genes. However, because there is only 58% similarity between the ligand-binding domain, various estrogens may differentially bind to the two ERs 19,20 . ER subtypes bind some ligands with different affinity; ligands may also demonstrate different agonist or antagonist character mediated by the two receptors 21 . SERMs such as 4-hydroxytamoxifen (2) and raloxifene (4) act as partial agonists on Era, but they exert exclusively antagonistic activity on Erb 22 .
Due to the increased risk of endometrial cancer with the use of tamoxifen many different fixed ring structures have been developed as potential SERMs to prevent E/Z isomerization of the triarylethylene structure 23 . Some of these are illustrated in Figure 1 and include the tetrahydronaphthalene lasofoxifene (6) 24 , spiroindene (7) 25 , pyrazole (8) 26 , quinoline (9) 27 , benzopyran (10) 28 , benzoxathiin (11) 29 and benzoxepin ring structures (12) 30 . The recently reported benzopyranobenzoxepanes were identified as potent SERMs for the treatment of postmenopausal symptoms 31 .
We are specifically interested in the development of novel heterocyclic ring scaffolds as ER antagonists 32 . Natural and synthetic azetidinone derivatives occupy a central place among medicinally important compounds due to their diverse and interesting biological activities 33 . Their importance is no longer exclusive due to the extensive clinical use of the b-lactam antibiotics but also because of their potential as intermediates in the synthesis of other types of compounds of biological interest. The antitumour properties of b-lactams have previously been reported 32,[34][35][36] together with inhibitory activity against serine proteases such as prostate-specific antigen (PSA) 37 , and other serine proteases such as tryptase 38 and human leukocyte elastase (HLE) 39 .
In a study to discover subtype selective ER scaffolds, we have identified a novel estrogen receptor modulator scaffold structure containing the b-lactam ring as a potential scaffold for SERMs, and have reported the antiproliferative effects and ER-binding properties of a series of 1,4-diarylsubstituted azetidin-2-ones, where the required basic amine function was positioned on the benzylic substituent at the C-3 position 32,40 . We now report the further investigation of this novel heterocyclic core scaffold structure as ER subtype selective ligands and the subsequent synthesis of a number of structurally varied b-lactam compounds, which are substituted at N-1, C-3 and C-4 with the required aryl rings. We have evaluated the antiproliferative activity of these products in ER-positive MCF-7 human breast tumor cells and also in ER-negative MDA-MB-231 cells and have determined their relative binding affinities for ERa and ERb. The two main b-lactam structural types now reported contain the important basic side-chain substituent positioned on the phenyl ring at the C4-position (type I) and at the N1-position (type II) as these were the optimal positions for substitution indicated from initial molecular modeling and docking studies. The general features of the b-lactam target scaffold structures selected for synthesis are illustrated in Figure 2.

Chemistry
All reagents were commercially available and were used without further purification unless otherwise indicated. Tetrahydrofuran (THF) was distilled immediately prior to use from Na/Benzophenone under a slight positive pressure of nitrogen, toluene was dried by distillation from sodium and stored on H-NMR spectra were assigned relative to the TMS peak at 0.00 and 13 C-NMR spectra were assigned relative to the middle CDCl 3 triplet at 77.00 ppm. For CD 3 OD, 1 H and 13 C-NMR spectra were assigned relative to the center peaks of the CD 3 OD multiplets at 3.30 d and 49.00 ppm, respectively. 19 F-NMR spectra were not calibrated. Electrospray ionization mass spectrometry (ESI-MS) was performed in the positive ion mode on a liquid chromatography time-of-flight (TOF) mass spectrometer (Micromass LCT, Waters Ltd., Manchester, UK) equipped with electrospray ionization (ES) interface operated in the positive ion mode at the High Resolution Mass Spectrometry Laboratory by Dr. Martin Feeney in the School of Chemistry, Trinity College. Mass measurement accuracies of5±5 ppm were obtained. Low resolution mass spectra (LRMS) were acquired on a Hewlett-Packard 5973 MSD GC-MS system (Hewlett-Packard, Palo Alto, CA) in electron impact (EI) mode. R f values are quoted for thin layer chromatography (TLC) on silica gel Merck F-254 plates, unless otherwise stated. Flash column chromatography was carried out on Merck Kieselgel 60 (particle size 0.040-0.063 mm), Aldrich aluminum oxide, (activated, neutral, Brockmann I, 50 mesh) or Aldrich aluminum oxide, (activated, acidic, Brockmann I, 50 mesh). All products isolated were homogenous on TLC. Analytical high-performance liquid chromatography (HPLC) to determine the purity of the final compounds was performed using a Waters 2487 Dual Wavelength Absorbance detector, a Waters 1525 binary HPLC pump, a Waters In-Line Degasser AF and a Waters 717plus Autosampler (Waters Corporation, Milford, MA). The column used was a Varian Pursuit XRs C18 reverse phase 150 Â 4.6 mm chromatography column (Agilent, Santa Clara, CA). Samples were detected using a wavelength of 254 nm. All samples were analyzed using acetonitrile (70%): water (30%) over 10 min and a flow rate of 1 mL/min. Schiff bases 13a and 13d were prepared as previously reported 41 .
General preparation of Schiff bases 13c, 13e, 13 h. A solution of the appropriately substituted aryl aldehyde (0.1 mol) and the appropriately substituted aryl amine (0.1 mol) in ethanol (50 mL) was heated to reflux for 3 h. The reaction mixture was reduced to 25 mL under vacuum, the Schiff base product crystallized from the solution. The crude product was then recrystallized from ethanol to afford the purified product.

MTT assay procedure
All assays were performed in triplicate for the determination of mean values reported. Compounds were assayed as the free bases isolated from reaction. The human breast tumor cell line MCF-7 was cultured in Eagles Minimum Essential (MEM) medium in a 95% air/5% CO 2 atmosphere supplemented with 10% fetal bovine serum, 2 mM L-glutamine and 100 mg/mL penicillin/streptomycin The medium was further supplemented with 1% non-essential amino acids. MDA-MB-231 cells are human breast adenocarcinoma cells, and representative of ER-negative breast cancer. They were grown as monolayer cultures at 37 C, under a humidified atmosphere of air supplemented with 5% CO 2 . The cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM), supplemented with 10% (v/v) Fetal Bovine Serum (FBS), 2 mM L-glutamine and 100 mg/mL penicillin/streptomycin. This medium contained phenol red, which is suspected to have estrogen-effects. However, as this assay was only concerned with anti-proliferative effects of the compounds, and these were compared to control cells grown in the same media, it was not deemed necessary to remove the phenol red prior to the assay being conducted. Similarly, the serum was not stripped with charcoal so estrogen levels were not reduced in this assay.
Cells were trypsinized and seeded at a density of 0.5 Â 10 4 cells/well into a 96-well plate and incubated for 24 h. After this time, they were treated with 2 ml volumes of test compound which had been pre-prepared as stock solutions in ethanol to furnish the final concentration range of study, 1 nM-100 mM, and reincubated for a further 72 h. Control wells contained the equivalent volume of the vehicle ethanol (1% v/v). This vehicle had no adverse effect on the cells. The culture medium was then removed and the cells were washed with 100 mL phosphate buffer saline (PBS) and 50 mL of 1 mg/ml MTT solution was added. Cells were incubated for 2 h in darkness at 37 C. At this point, solubilization was begun through the addition of 200 mL DMSO and the cells maintained at room temperature in darkness for 20 min to ensure thorough color diffusion before reading the absorbance at 595 nm. The absorbance value of control cells (vehicle treated) was set to 100% cell viability and from this graphs of absorbance versus cell density per well were prepared to assess cell viability and from these, graphs of percentage cell viability versus concentration of subject compound were drawn. Effect of compounds on MCF-7 cells treated with estradiol: MCF-7 cells were trypsinised and seeded at a density of 0.5 Â 10 4 cells/well into a 96-well plate as described above. The cells were then treated with 1 ml of test compound and 1 ml of estradiol with (starting concentration is 1 in 200 dilution), which had been preprepared as stock solutions in ethanol to furnish the final concentration range of study, 1 nM-100 mM, and re-incubated for a further 72 h. Control wells contained the equivalent volume of the vehicle ethanol (1% v/v). The antiproliferative assay is then completed in the same manner as described above.

Cytotoxicity studies
Human MCF-7 breast cancer cells or human MDA-MB-231 breast cancer cells were plated at a density of 0.5 Â 10 4 cells per well into a 96-well plate (200 mL per well) and incubated at 37 C in air supplemented with 5% CO 2 for 24 h. Cells were treated with 2 mL volumes of test compound which had been pre-prepared as stock solutions in ethanol to furnish the final concentration range of study, 1 nM-100 mM, and re-incubated for a further 72 h. Control wells contained the equivalent volume of the vehicle ethanol (1% v/v). Following incubation, 20 mL of lysis solution was added to one row of wells to act as a 100% lysis control. After 30 min, 50 mL aliquots of medium were removed from all wells and placed in a clean 96-well plate. Cytotoxicity was determined using the LDH assay kit 42 obtained from Promega (Madison, WI), following the manufacturer's instructions for use. A 50 mL per well LDH substrate mixture was added and the plate left in darkness at room temperature for equilibrium. Stop solution (50 mL) was added to all the wells and absorbance read at 490 nM. Data were presented following calculation of percentage cell lysis versus concentration of subject compound.

Estrogen receptor binding assay
ERa and ERb fluorescence polarization based competitor assay kits were obtained from Panvera at Invitrogen Life Technologies, Carlsbad, CA. The recombinant ER (insect expressed, full length, untagged human ER obtained from recombinant baculovirusinfected insect cells) and the fluorescent estrogen ligand were removed from the À80 C freezer and thawed on ice for 1 h prior to use. The fluorescent estrogen ligand (2 nM) was added to the ER (40 nM for ERa and 30 nM for ERb) and screening buffer (100 mM potassium phosphate (pH 7.4), 100 mg/ml BGG, 0.02% NaN 3 was added to make up to a final volume that was dependent on the number of tubes used (number of tubes (e.g. 50) Â volume of complex in each tube (50 mL) ¼ total volume (e.g. 2500 mL). Test compound (1 mL, concentration range 100 nM-1 mM) was added to 49 mL screening buffer in each borosilicate tube (6 mm diameter). To this, 50 mL of the fluorescent estrogen/ER complex was added to make up a final volume of 100 mL.

Estrogenic activity: alkaline phosphatase assay
Following the procedure of Littlefield et al. 43 , human Ishikawa cells were maintained in Eagle's Minimum Essential Medium (MEM containing 10% v/v fetal bovine serum (FBS) and supplemented with 100 U/mL penicillin and 10 mg/mL streptomycin, 2 mM glutamine and 1 mM sodium pyruvate. Twenty four hours before the start of the experiment, near confluent cells were changed to an estrogen-free basal medium (EFBM), A 1:1 mixture of phenol-free Ham's F-12 and Dulbecco's Modified Eagles Medium, together with the supplements listed above, and 5% calf serum, stripped of endogenous estrogens with dextrancoated charcoal. On the day of the experiment, cells were harvested with 0.25% trypsin and plated in 96-well flat bottomed microtitre plates in EFBM at a density of 2.5 Â 10 4 cells/well. Test compounds were dissolved in ethanol at 10 À 3 M, diluted with EFBM (final concentration of ethanol 0.1%) and filter sterilized. After addition of the test compounds, (plated in 50 mL, added estradiol in 50 mL and blank medium to give a final volume 150 mL) the cells were incubated at 37 C in a humidified atmosphere containing 95%air/5% CO 2 for 72 h. All experimental values were obtained in triplicate. The microtitre plates were then inverted and the growth medium removed. The plates were then rinsed by gentle immersion and swirling in 2 L of PBS (0.15M NaCl, 10 mM sodium phosphate, pH 7.4). The plates were removed from the container, the residual saline in the plate was not removed, and the wash was repeated. The buffered saline was then shaken out, and the plate blotted on paper towel. The covers were replaced and the plates were placed at À80 C for at least 15 min, and then thawed at room temperature for 5-10 min. The plates were then placed on ice and 50 mL ice cold solution containing 50 mM p-nitrophenyl phosphate, 0.24 mM MgCl 2 and 1M diethanolamine (pH 9.8) was added. The plates were warmed to room temperature (time zero), and the yellow color from the production of p-nitrophenol was allowed to develop. The plates were monitored at 405 nm until maximum stimulation of the cells showed an absorbance of $1.2.
Computational procedures: protein preparation PDB entry 3ERT 44 (4-hydroxytamoxifen co-crystallised with ERa) was downloaded from the Protein Data Bank (PDB) was utilized for all ERa dockings and also as a template for creation of a homology model of ERb. The model was constructed in an automated fashion using EsyPred3D. 45 Hydrogens were added to both structures using MOE 46  (Openeye Scientific Software, Santa Fe, NM) retaining all default settings and employing the Chemgauss3 scoring function 50 for prioritizing correct binding orientations. The active site of top docked poses of each compound was minimized and refined using LigX (Chemical Computing Group) 46 to allow side-chain repositioning.

Chemistry
The synthesis of the Type I series of b-lactam compounds which contain the basic pyrrolidinylethoxy substituent located on the C-4 aryl ring is illustrated in Scheme 1. The Staudinger reaction requiring a cycloaddition reaction of a ketene with an imine is the most commonly used method for synthesis of 1,3,4-trisubstituted b-lactams 34,51 . The initial target compound 15a, containing a methyl ether on the N-1 aryl ring (which is important for interaction with His524 in the ligand binding domain of the estrogen receptor), was synthesized using this reaction. The required Schiff base 13a was prepared in high yield (88%) by condensation of 4-hydroxybenzaldehyde with 4-methoxyaniline. To avoid difficulties in the following cycloaddition reaction, the phenol was protected as a tert-butyldimethylsilyl ether using tertbutyldimethylsilyl chloride and the base DBU to afford the protected Schiff base 13b. The b-lactam product 14a was obtained by subsequent Staudinger reaction of imine 13b with phenylacetyl chloride in the presence of triethylamine. To optimize the yield of the b-lactam product, a number of reaction conditions were investigated. b-Lactam 14a was initially obtained successfully using DMF as the solvent 52 . This method resulted in exclusive synthesis of the trans product in a moderate yield (45%), (one enantiomer only shown for the products 14a-c and 15a-g in Scheme 1). However, the reaction required over 24-h reflux so an alternative method was investigated. When toluene was used as the solvent and the acid chloride was added dropwise to a refluxing solution of appropriate imine and triethylamine 53 , an improved yield (65%) was obtained with exclusive isolation of the trans product. Finally, reaction conditions were optimized using dichloromethane as the solvent and with the dropwise addition of the acid chloride to a mixture of the imine 13b and triethylamine, initially at À20 C 34 . The trans product 14a was again isolated exclusively and an improved yield of 78% obtained. This method was then employed for all subsequent Staudinger reactions. As indicated, all three methods of reaction above yielded the trans b-lactam product, with the stereochemistry of the b-lactam deduced from the coupling constants of the C-3 and C-4 protons and were found to be in the region 1-3 Hz. The cis-b-lactams have larger coupling constants (5-6 Hz) than the trans-b-lactams which are usually 1-3 Hz 34,52 . Stereoselectivity in the Staudinger reaction that depends on a number of experimental factors including structure of the imine and acid chloride, sequence of reagent addition, solvent, temperature and organic amine base 34 . In general, when the acyl chloride is added dropwise, preferably at low temperature, to a solution of imine and a tertiary amine such as triethylamine, the cis cycloadduct is the major or exclusive stereoisomer detected. In contrast, when the tertiary base is added to a mixture of imine and acetyl chloride, mixtures of cis and trans cycloadducts are obtained, in which the trans is the major or exclusive product 34,51 . However in our synthesis, using all the above conditions with reaction temperatures varied from À20 to 150 C, the trans isomer was always exclusively formed. We also observed a similar stereochemical outcome with 4-methoxyphenylacetyl chloride and 4-benzyloxyphenylacetyl chloride in this reaction indicating that the specific acid chloride is an important variable in determining the stereochemical selectivity. Alonso et al reported synthesis of a mixture of cis and trans isomers when tetrahydrofuroyl chloride was the acid component 53 . Phthalimidoacetyl chloride and crotonyl chloride are other reported acid chlorides, which form exclusively trans b-lactams.
Removal of the silyl protecting group from 14a using tetrabutylammonium fluoride (TBAF) resulted in the isolation of the phenolic 14c (50%), which was confirmed in the IR spectrum that shows a carbonyl absorption at 1735.6 cm À 1 and a broad absorption corresponding to the hydroxy group at 3300.0 cm À 1 . In the 1 H NMR spectrum, coupled doublet signals at 4.22 and 4.98 (J ¼ 2.26 Hz) are assigned to H-3 and H-4 respectively, indicating the trans b-lactam stereochemistry.
The alternative method used for the preparation of 14c involves the use of a benzyl ether as the protecting group in place of the silyl ether (Scheme 1), which can be easily removed using catalytic hydrogenolysis. Catalytic hydrogenation has been reported under ambient pressure of hydrogen at 50 C in methanol with Pd/C as the catalyst to remove the benzyloxy group without any effect on the b-lactam ring 34,54,55 . However, care must be taken to avoid over-hydrogenation as C-4-N-1 bond cleavage has been reported to proceed by palladium catalyzed hydrogenolysis when an aryl substituent is located at the C-4 position i.e. a benzylic carbon. 56,57 The Schiff base 13c (obtained following the condensation of 4-methoxyaniline with 4-benzyloxybenzaldehyde) was reacted with phenylacetyl chloride under Staudinger conditions to afford the benxyloxy protected product 14b, (78%). The IR spectrum contained the characteristic b-lactam carbonyl absorption band at 1735 cm À 1 while the 1 H-NMR spectrum confirmed the trans b-lactam isomer with H-3 and H-4 observed as coupled doublets at 4.19 and 4.81, respectively, (J ¼ 2.26 Hz). Careful removal of the benzyl protecting group from 14b by hydrogenation yielded 14c (70%). Alkylation of the phenolic 14c with 1-(2-chloroethyl)pyrrolidine affords the required product 15a (30%).
Fluoro-substituted tamoxifen and cyclofenol derivatives have been investigated as ER-imaging agents for breast cancer 58,59 and we have previously reported the potent ER-binding properties of fluorine-containing benzoxepine type ER antagonists 30 . We now wished to examine the effect of the inclusion of the lipophilic fluorine substituent on the ER activity of the b-lactam ER antagonist compounds 15b and 15c. The Schiff base 13d, obtained in 58% by the standard method (Scheme 1), was directly alkylated with 1-(2-chloroethyl)pyrrolidine hydrochloride to yield 13f which was used without further purification in the subsequent reactions. Treatment of 13f with phenylacetyl chloride under the usual Staudinger reaction conditions afforded the trans b-lactam product 15b, (Scheme 1). The 1 H-NMR spectrum indicated two coupled doublets at 4.27 and 4.87 (J 3,4 ¼ 2.04 Hz) assigned to H-3 and H-4, respectively. 15c was similarly prepared from 13f and 4-methoxyphenylacetyl chloride The requirement for a phenolic substituent in many ER antagonists such as 4-hydroxytamoxifen 4 and raloxifene is significant for successful binding to the ER as shown by interactions with Glu353 and Arg394. Therefore, in the present work it was critical to include a phenolic substituent group on the N-1 aryl ring of the Type-1 b-lactam products to provide a 4hydroxytamoxifen analog containing the b-lactam ring. Synthesis of 15f by direct demethylation of 15a proved unsuccessful using several different reagents including ethanethiol and boron trifluoride-methyl sulfide; in both cases resulting in degradation of the b-lactam ring. An alternative method of synthesis of 15f and 15g was pursued which involved the preparation of a Schiff base 13g with both the required benzyl protected phenol and the basic side-chain are in position. This protected Schiff base was then treated with the relevant acid chloride to form the b-lactam which could then be deprotected by hydrogenation to yield the free phenol as required on the N-1 aryl ring. Schiff base 13e was prepared in 96% yield by condensation of 4-hydroxybenzaldehyde with 4-benzyloxyaniline and was then alkylated with 1-(2-chloroethyl)pyrrolidine hydrochloride to afford 13g (60%). Subsequent treatment of 13g with phenylacetyl chloride under the usual Staudinger conditions resulted in the isolation of the protected b-lactam 15d (61%). The 1 H-NMR spectrum of 15d showed characteristic b-lactam doublet signals for H-3 at 4.18 and H-4 at 4.85, (J 3,4 ¼ 2.52 Hz). The benzyloxy group was carefully removed by hydrogenation yielding 15f (50%); the 1 H-NMR spectrum confirmed the presence of the trans b-lactam ring with H-3 at 4.20 and H-4 at 4.84, (J 3,4 ¼ 2.24 Hz). Compound 15g was synthesized in a similar reaction sequence. The protected Schiff base 13g was treated with 4-methoxyphenylacetyl chloride resulting in the isolation of 15e (33%). Subsequent removal of the benzyl protecting group from 15e by careful hydrogenation yields the phenolic b-lactam product 15g in 78% yield, with no evidence of ring hydrogenolysis (Scheme 1).
The synthetic route used for the Type II b-lactams (containing the basic side chain substituent on N1 aryl ring) is shown in Scheme 2, which again employs the Staudinger reaction for b-lactam formation. The phenolic Schiff base 13h (obtained in 96% yield on reaction of 4-benzyloxybenzaldehyde with 4aminophenol) was treated with 1-(2-chloroethyl)pyrrolidine hydrochloride yielding the alkylated product 13i (60%). Reaction of 13i with phenylacetyl chloride and 4-methoxyphenylacetyl chloride under the usual Staudinger reaction conditions afforded the trans b-lactams 15h (15%) and 15i (33%) as confirmed by 1 H NMR spectra: 15h (J 3,4 ¼ 2.5 Hz) and 15i (J 3,4 ¼ 2.0 Hz). Removal of the benzyl protecting group from 15h and 15i by hydrogenation yielded the required phenolic products 15j and 15k respectively, (one enantiomer shown for compounds 15h-15k, Scheme 2).

Antiproliferative activity in MCF-7 and MDA-MB-231 breast cancer cells
The b-lactam compounds prepared above were evaluated in a series of in vitro assays which determined their antiproliferative activity in ER positive MCF-7 and ER negative MDA-MB-231 breast cancer cell lines and also their affinity for the estrogen receptor and estrogenic effects in Ishikawa cells.
Compounds 15a-15c, 15f-g and 15j-k were initially screened for their antiproliferative activity using the ER expressing (ERdependent) MCF-7 human breast cancer cell line and the ERindependent MDA-MB-231 human breast cancer cell line. Table 1 shows the antiproliferative effects of type I and type II b-lactams in both MCF-7 and MDA-MB-231 cell lines. The majority of the compounds show anti-proliferative effects at concentrations similar to that of the positive control tamoxifen 60,61 . Many of the compounds (e. g. 15a, 15f, 15g and 15j) have IC 50 ¼ 0.185 mM), representative of the Type-I structure, containing the phenolic substitution in Ring C, which would be required for interaction with the Asp351 of the estrogen receptor LBD 14 . This indicates that the possible mechanism of action of the compound is mediated through binding to the estrogen receptor. Tamoxifen shows some antiproliferative effects in MDA-MB-231 ER-negative cell lines at much higher concentrations (approx. 20 mM) than in MCF-7 cells 62,63 . The cytotoxic effect of Table 1. Antiproliferative and cytotoxic effects for compounds 15a-c, 15f-g and 15j-k in MCF-7 and MDA-MB-231 cells; ERa and ERb binding affinities for compounds 15a-c, 15f-g and 15j-k.   63,64 . xLactate dehydrogenase assay: following treatment of the cells, the amount of LDH was determined using LDH assay kit from Promega. Data are presented as % cell lysis at compound concentration of 10 mM 42 . ||Values are an average of at least nine replicate experiments, for ERa with typical standard errors below 15%, and six replicate experiments for ERb, with typical standard errors below 15%. #The ER binding values obtained are in agreement with the reported ER IC 50 binding data for tamoxifen (ERa 60.9 nM ERb188 nM, Panvera/ Invitrogen). **Work by Seeger et al. 61 . these b-lactam compounds in MCF-7 cells as determined in the lactose dehydrogenase (LDH) assay (Table 1) is also lower than that of tamoxifen 64 with the exception of compound 15c, which resulted in 26% and 25% toxicity in MCF-7 and MDA-MB-231 cell lines, respectively. This cytotoxic action may explain the equal antiproliferative effects in both MCF-7 and MDA-MB-231 cell lines observed for compound 15c.
The antiproliferative effect of 15g (Type I) in MCF-7 cells was significantly reduced with the addition of estradiol (50 nM), with IC 50 value increased from 0.186 mM to 9.17 mM, ( Figure 3A). A similar result is also obtained for compound 15f (IC 50 values increased from 0.519 to 27.85 mM). Compound 15k (Type II) demonstrated a small increase in IC 50 value on addition of estradiol (50 nM) from 3.30 to 8.29 mM). These results support the indication that the antiproliferative effects for compounds 15f and 15g are a result of interaction of the compounds with the ER and therefore preventing estrogen mediated proliferation. Reversal of antiestrogen-mediated cell growth antagonism by estradiol has been suggested to indicate the degree to which antagonism is mediated through ER 65 . Tamoxifen and 4-hydroxytamoxifen have been shown to have an estradiol reversible and estradiol irreversible components to their inhibition of cell proliferation in vitro, with the former being highly correlated with affinity for ER. The antiproliferative effect appears to be estradiol reversible with the cytotoxic effect being irreversible 62 . This effect is seen at the higher ER antagonist concentrations, and is demonstrated for tamoxifen where estradiol is shown to reverse the antagonist effect at lower concentrations of tamoxifen, but not at higher concentrations; (IC 50 value ¼ 4.22 mM for tamoxifen in the presence of estradiol (50 nM), Figure 3B). Compounds 15g and 15f show estradiol-reversible effects at concentrations higher than tamoxifen, indicating that they have less estradiol-irreversible effects than tamoxifen, and suggesting that the main mechanism of action of these b-lactam compounds is through affinity for the ER.

Estrogen receptor binding studies
The affinity for the ER of the b-lactam compounds 15a-15c, 15fg and 15j-k is confirmed through estrogen receptor binding studies. Table 1 shows the results of the competitive ER binding assay for b-lactam SERM Types I and II with both ERa and ERb. While the level of affinity to the ER varies depending on the aryl substituents present, compounds of structural Type I appear to have better binding ability than compounds of structural type II. Of the Type I compounds, 15g established the most potent binding to the ER (IC 50 ERa ¼ 4.3 nM, ERb ¼ 322 nM) and significantly more potent than both tamoxifen (IC 50 ERa ¼ 70 nM) and 4hydroxytamoxifen (IC 50 ERa ¼ 40 nM) for ERa. This result also correlated with the antiproliferative activity of 15g in the MCF-7 cell line (IC 50 ¼185 nM). The fluoro-substituted b-lactam 15c indicated some improved ERa and ERb interaction when compared to the methoxy substituted product 15a, ( Table 1).
The IC 50 value for ERa binding of the Type II compound 15k (3.04 mM) is much greater than tamoxifen or that of the type I b-lactam compounds 15f and 15g. However, the moderate antiproliferative effect in MCF-7 cells is very similar to 15c with an IC 50     Hydrogen bonding between the B-ring 3-OH substituent and Glu353, Arg394, HOH is depicted. A salt bridge between the basic side chain and Asp351 is also formed. Figure 5. Key anti-estrogenic interactions are observed in the docking of 15g in ERa. Hydrogen bonding between the C-ring 3-OH substituent and Glu353, Arg394, HOH is depicted. A salt bridge between the basic side chain and Asp351 is also formed.
anti-proliferation does not seem to be mediated through the ER.
(Addition of estradiol (50 nM) with compound 15k in MCF-7 cells produced only a small increase in the IC 50 value). This indicates that the antiproliferative effect is not estradiol reversible and therefore compound 15k does not appear to act solely as an ER antagonist, but may also have an alternative mode of action. Although, the main mechanism of action of tamoxifen is primarily mediated through the ER, studies have shown non-ER mediated mechanisms are present including the involvement of certain signaling proteins, proto-oncogenes and transforming growth factor b in tamoxifen-mediated apoptosis 66 . All of the Type I b-lactam compounds examined also show a preference for ERa binding over ERb. In the case of the most potent example 15 g, the a:b binding selectivity ratio is 75:1 and for 15b no binding to ERb was observed at concentrations of up to 100 mM. Tamoxifen demonstrates almost equal binding affinity for ERa and ERb. Type II compounds 15j and 15k also shows competitive binding to ERa in excess of ERb, while compound 15k shows no binding to ERb up to 50 mM (see Supplementary Information: Figure S1: Effect of compounds 15a, 15f , 15g, 15k, 15j on the inhibition of proliferation of MCF-7 breast cancer cells; Figure S2: Effect of compounds 15a, 15f, 15g, 15k, 15j on the inhibition of proliferation of MDA-MB-231 breast cancer cells; Figure S3: Estrogen receptor a binding affinities for compounds 15a, 15f, 15g, 15k, 15j; Figure S4: Estrogen receptor b binding affinities for compounds 15a, 15f, 15g, 15k, 15j).

Antiestrogenic activity in ishikawa cells
The estrogen stimulation and antagonistic properties of the most active ER compound 15g was determined in an estrogen bioassay which is based on the stimulation of alkaline phosphatase (AP) in the Ishikawa human endometrial adenocarcinoma cell line 43 . Enmark et al have reported that Ishikawa cells contain both ERa and ERb receptors but with ERa being far in excess 67 . Compound 15g was examined as an estrogen antagonist by its effect on the inhibition of estradiol stimulation in the Ishikawa cells in a dose-dependent manner, (Figure 4). The estrogenic stimulatory property of this compound was also monitored in Ishikawa cells by measuring the stimulation of alkaline phosphatase (AP) in these cells in the absence of estradiol. 15g showed little ability to inhibit the effect of estradiol in Ishikawa cells at concentrations of up to 1 mM, however, at the same time 15g itself shows a higher level of estrogen stimulation (21% at 1 mM concentration) when compared to tamoxifen, (10% stimulation at 1 mM concentration). This may indicate that the stimulatory effect is due to both estradiol and 15g and is not indicative of the inability of estradiol to be displaced by the b-lactam compound. Other known anti-estrogens also show estrogen stimulation in Ishikawa cells with reports having shown that 4-hydroxytamoxifen stimulates AP activity to a level 47% of that of estradiol and the SERM lasofoxifene increases AP activity by 18% 68 .

Molecular modeling studies of novel b-lactam compounds
To rationalize the observed ERa/b affinity of Type I and Type II compounds, a semi-flexible ligand receptor docking study of compounds 15g and 15k was undertaken. The PDB entry 3ERT was used (ERa co-crystallized with 4-hydroxytamoxifen) to examine the ERa binding mode of these compounds 44 . Figures  5 and 6 illustrate the docked binding poses of compounds 15g and 15k respectively in the binding site of ERa. It is immediately evident that both compounds adopt a typical antiestrogenic orientation in the ligand binding site. Key hydrogen bonding interactions between the phenolic group of the ligands and Glu353, Arg394 and bridging water are observed. A salt bridge is also formed between the basic side chain nitrogens and Asp351 of Helix-12 (15g -2.8 Å , 15k -2.6 Å ). As there is currently no antiestrogen co-crystallized in both isoforms of ERa and ERb, it was deemed necessary to construct a homology model of ERb using PDB entry 3ERT as a template 45 . This process ensured that bias of ligand induced residue motion would be reduced and the process would provide more realistic dockings when used in conjunction with our semi-flexible docking approach. Figure 7 illustrates the docked positions of both compounds 15g and 15k in ERa/b superimposed by backbone. It is firstly apparent that the b-lactam ring of 15g lies in close proximity to the ERa Leu384/ ERb Met336 residue mutation and favorable interaction with Leu384 could account for some of the ERa selectivity observed (&75-fold). For the case of 15k (Type II structure), a different binding pose occurs compared with 15g whereby the b-lactam ring carbonyl oxygen of 15k is positioned towards ERa residue Met343 (ERb Met295) making a favorable interaction, whilst Met295 seems to reposition itself so as not to cause any steric hindrance. It would appear for the most part that the significant ERa selectivity observed is mainly a result of the tighter packing of 15g in the active site of ERa by surrounding residue side chains.

Conclusion
There is currently much interest in the discovery of novel molecular scaffolds with SERM profile properties which could be suitable for development of new therapies for the treatment of breast cancer, osteoporosis, and related hormone-dependent conditions. Both raloxifene and tamoxifen are good preventive choices for treatment of postmenopausal women with elevated risk for breast cancer. Because of the known importance of ERa as a pharmaceutical target and also the potential importance of ERb, molecules that act as agonists or antagonists selectively a or b ER subtypes are currently being investigated for their therapeutic potential. ERa predominates in the breast and in reproductive tissues such as the uterus, whereas ERb is the principal subtype in the ovary and certain regions of the brain. We have synthesized a number of novel b-lactam compounds designed as potential estrogen receptor ligands, which demonstrate antiproliferative activity against the MCF-7 human breast cancer cell line. The compounds also demonstrate good affinity for the estrogen receptor and selectivity for ERa. The most potent antiproliferative compound 15g having Type I structural scaffold, demonstrated ERa binding with IC 50 ¼ 4.3 nM and relative binding affinity ERa/ERb of 75:1. Further biochemical studies will determine the effects of these novel analogs on ERE transcription and ERa stability in MCF-7 cells, and will determine the mechanistic differences between their activity and that of tamoxifen.