In a previous study, we described affinity labeling of the lamb uterine estrogen receptor by 17 alpha-[(bromoacetoxy)alkyl/alkynyl]estradiols. However, the intrinsic receptor-alkylating activities of these compounds were probably very hampered by their poor hydrolytic stability in estrogen receptor-containing tissue extracts. Therefore, (i) to develop affinity labels of the receptor not susceptible to hydrolysis and (ii) to specify the structural requirements for 17 alpha-electrophilic estradiol derivatives to be potent affinity labels of the receptor, we prepared four 17 alpha-[(haloacetamido)alkyl]estradiols. Three were bromoacetamides differing at the alkyl substituent (methyl, ethyl, or propyl), and the last was an [(iodoacetamido)propyl]estradiol prepared under both nonradioactive and 3H-labeled forms. Although their affinities for the estrogen receptor were very low (from 0.008% to 0.02% that of estradiol), they appeared to be efficient affinity labels of the receptor due to their irreversible inhibition of [3H]estradiol specific binding in lamb uterine cytosol. The effect of the compounds was time-, pH-, and concentration-dependent, with > 50% and > 80% estrogen-binding sites inactivated at 0 degrees C and pH 8.5, for the less active and more active compounds, respectively; the corresponding IC50 values varied from approximately 20 nM to approximately 10 microM. The order of efficiency was [(bromoacetamido)methyl]estradiol < [(bromoacetamido)ethyl]estradiol << [(bromoacetamido)propyl]estradiol < [(iodoacetamido)propyl]estradiol. Affinity labeling was directly demonstrated by ethanol-resistant binding of [3H][(iodoacetamido)propyl]estradiol to the receptor. The irreversible inactivation of the hormone-binding site by the four haloacetamides was prevented by treatment of the cytosol with the thiol-specific reagent methyl methanethiosulfonate, suggesting that the target of these compounds was probably the -SH of cysteines. Negative results obtained with other 17 alpha-electrophilic estradiol derivatives suggested that affinity labeling of the receptor by such derivatives required a minimal distance, including at least four C-C or C-N bonds, between the steroid and the electrophilic carbon. We therefore concluded that target cysteines in the hormone-binding site were not in direct contact with the steroid but probably in the immediate neighborhood of the D ring of the bound steroid.

Mass spectrometry was used to identify the sites of covalent attachment of [(14)C]-17alpha-bromoacetamidopropylestradiol ([(14)C]17BAPE(2), an estradiol agonist) to the ligand-binding domain (LBD) of mouse estrogen receptor alpha (ERalpha). A glutathione S-transferase (GST)-LBD chimera protein was overexpressed in Escherichia coli, using a vector encoding GST fused with a C-terminal portion of mouse ERalpha (Ser(313)-Ile(599)), via a sequence enclosing a thrombin cleavage site (located 14 amino acids ahead of Ser313). [(14)C]17BAPE(2) covalent labeling experiments were carried out on the GST-LBD chimera immobilized on glutathione-Sepharose. After thrombin cleavage of the chimeric LBD, two major [(14)C]17BAPE(2)-labeled species of 34 ( approximately 75%) and 30 kDa ( approximately 25%) were detected by SDS-PAGE and autoradiography. Their identity was assessed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS): two main signals were consistent with the mass of the full-length (Ser(313)-Ile(599)) and truncated LBD (Ser(313)-Ala(573)), both comprising the extra 14 N-terminal amino acids and covalently bound [(14)C]17BAPE(2) (via HBr elimination). A purified (14)C-labeled LBD preparation was trypsinized to identify the covalent attachment sites of 17BAPE(2). HPLC of tryptic fragments only revealed two discrete and practically equivalent radioactive fractions. MALDI-TOF MS analysis of these two fractions showed only two signals which exactly matched the molecular masses of the [(14)C]17BAPE(2)-alkylated Cys(534)Lys(535) and Cys(421)-Arg(438) peptides, respectively. Hydrolysis of the second (14)C-labeled fraction by Staphylococcus aureus V8 Glu-C endoproteinase generated signals typical of alkylated the Cys(421)-Glu(423) tripeptide. We concluded that Cys421 and Cys534 were equivalent alternative covalent attachment sites of 17BAPE(2) on the LBD. These biochemical data were interpreted using the crystallographic structures of estradiol-LBD and raloxifene- or 4-hydroxytamoxifen-LBD complexes. The covalent attachment to Cys421, Cys534, or both could be interpreted according to the starting structure. Various hypotheses based on the biochemical results and molecular modeling simulations are discussed, with the likely involvement of dynamic interconversion between multiple conformational states of the LBD-17BAPE(2) complex.

Results obtained in a previous study suggested that cysteine residues in the estrogen receptor were covalent attachment sites for four 17 alpha-(haloacetamidoalkyl) estradiols (halo, bromo or iodo; alkyl, methyl, ethyl, or propyl). To identify the putative concerned cysteines, we expressed wild-type and various cysteine --> alanine mutants of the human estrogen receptor in COS cells and determined their ability to be alkylated by the four electrophiles. The quadruple mutant, in which all the cysteines (residues 381, 417, 447, and 530) of the hormone-binding site were changed to alanines, showed very little electrophile labeling, whereas the four single mutants (C381A, C417A, C447A, and C530A) were alkylated as efficiently as the wild-type receptor. These results (i) demonstrate that cysteine residues were covalent attachment sites of electrophiles and (ii) indicate that more than one cysteine residue could be alkylated. Analysis of three double mutants (C381A/C530A, C417A/C530A, and C447A/C530A) provided strong evidence that only C417 and C530 were sites for electrophile covalent attachment. Since C530 was also alkylated by tamoxifen aziridine, a nonsteroidal affinity-labeling agent, we propose a selective mode of superimposition of tamoxifen-class antiestrogens with estradiol, which could account for the relative positioning of the two types of ligands in the receptor hormone-binding pocket. According to the structure of the hormone-binding pocket of nuclear receptors, as inferred from crystallographic studies and general sequence alignment of hormone-binding domains, C417 and C530 appear to be (1) located at the extreme border or in structural elements involved in delineation of the hormone-binding pocket, (2) spatially in close proximity to each other, and (3) in positions highly homologous to those of glucocorticoid receptor sites alkylated by affinity- and photoaffinity-labeling agents, respectively.

Ten electrophilic estradiol 11beta-aryl derivatives were synthesized, with three different types of 11beta-substituent: (i) pOO(CH(2))(2)X (compounds: 6, X = OSO(2)CH(3); 7, X = I; 13, X = NHCOCH(2)Cl; 15, X = N(CH(3))COCH(2)Br; and 16, X = N(CH(3))COCH(2)Cl); (ii) pOO(CH(2))(5)X (compounds: 17, X = I; 20, X = NHCOCH(2)Br; and 22, X = N(CH(3))COCH(2)Br); and (iii) pOC(triple bond)CCH(2)X (compounds: 27, X = NHCOCH(2)Cl; and 29, X = N(CH(3))COCH(2)Cl). The range of their apparent affinity constants for binding the lamb uterine estrogen receptor alpha (ERalpha) was 3-40% that of estradiol. Six electrophiles, chloroacetamides 13, 16, 27, and 29, iodide 17, and bromoacetamide 20 (whose arm linking the electrophilic carbon to the 11beta-phenyl group includes at least six bonds), were able to irreversibly inhibit the binding of [(3)H]estradiol to ER (25-60% decrease in binding sites), with the following compound effectiveness order: 17 < 13 < 16 approximately 20 approximately 27 approximately 29. Mesylate 6, iodide 7 (whose linking arm includes only three bonds), and bromoacetamides 15 and 22 (which differ from 16 by the Cl to Br change and from 20 by the NH to NCH(3) change, respectively) were much less effective (<10% decrease in binding sites, if any). The fact that the inactivation of estradiol-binding sites by the six electrophiles was totally prevented by estradiol indicated that they were ER affinity labeling agents. When ER was modified by methyl methanethiosulfonate, an SH-specific reagent, the different compounds led to very contrasting results in ER affinity labeling. With modified ER, iodide 17 and chloroacetamides 27 and 29 were practically inactive, chloroacetamides 13 and 16 and bromoacetamide 20 were still active but less effective than on the native ER, whereas tertiary bromoacetamides 15 and 22, found to be practically inactive on native ER, became the most effective electrophiles ( approximately 45% and approximately 65% binding sites inactivated, respectively). The results indicate that in the steroid-filled hormone-binding pocket: (i) nucleophilic residues are localized on the beta-side but relatively remote from the steroid nucleus (distance from C-11 > "seven bonds"); (ii) relatively discrete changes in the electrophilic functionality, such as Cl to Br or NH to NCH(3) of haloacetamido compounds, can markedly modify the positioning of the electrophilic center which could no longer react with the nucleophilic residues; and (iii) cysteine residues (probably homologues of human ERalpha cysteine 381 and/or cysteine 530) are, at least partly, the covalent attachment sites of the electrophiles. Moreover, modification of cysteine residues by methyl methanethiosulfonate changes the structure of the hormone-binding pocket, whose labeling by the various electrophiles is profoundly altered.