Search

We have advanced the view that estrogens activate the estrogen receptor (ER) alpha complex differently. A group of planar (estradiol, genistein, and coumestrol) and nonplanar (methoxychlor and its mono- and didemethylated phenolic metabolites) environmental estrogens, which are all full estrogens in MCF-7 breast cancer cell proliferation assays, was shown to segregate discretely into planar and nonplanar groups. These groups were delineated using a novel assay of mutant ER cDNAs stably transfected into MDA-MB-231 cells and the activation of the transforming growth factor alpha target gene in situ that putatively describes the external shape of the ER complex. Planar compounds activate estrogen action through the two traditional activation functions (AFs), AF1 and AF2, in the ER. In contrast, nonplanar compounds can activate estrogen action through AF1 and the amino acids Asp-351 and Asp-538, which are exposed when helix 12 silences AF2. The observation that class I (planar) and class II (nonplanar) compounds have different mechanisms of estrogen action may have important implications for tissue selective modulation of the ER.

Amino acid Asp-351 in the ligand binding domain of estrogen receptor alpha (ERalpha) plays an important role in regulating the estrogen-like activity of selective estrogen receptor modulator-ERalpha complexes. 4-Hydroxytamoxifen is a full agonist at a transforming growth factor alpha target gene in situ in MDA-MB-231 human breast cancer cells stably transfected with the wild-type ERalpha. In contrast, raloxifene (Ral), which is also a selective estrogen receptor modulator, is a complete antiestrogen in this system. Because D351G ERalpha allosterically silences activation function-1 activity in the 4-hydroxytamoxifen-ERalpha complex with the complete loss of estrogen-like activity, we examined the converse interaction of amino acid 351 and the piperidine ring of the antiestrogen side chain of raloxifene to enhance estrogen-like action. MDA-MB-231 cells were either transiently or stably transfected with Asp-351 (the wild type), D351E, D351Y, or D351F ERalpha expression vectors. Profound differences in the agonist and antagonist actions of Ralcenter dotERalpha complexes were noted only in stable transfectants. The agonist activity of the Ralcenter dotERalpha complex was enhanced with D351E and D351Y ERalpha, but raloxifene lost its agonist activity with D351F ERalpha. The distance between the piperidine nitrogen of raloxifene and the negative charge of amino acid 351 was critical for estrogen-like actions. The role of the piperidine ring in neutralizing Asp-351 was addressed using compound R1h, a raloxifene derivative replacing the nitrogen on its piperidine ring with a carbon to form cyclohexane. The derivative was a potent agonist with wild type ERalpha. These results support the concept that the side chain of raloxifene shields and neutralizes the Asp-351 to produce an antiestrogenic ERalpha complex. Alteration of either the side chain or its relationship with the negative charge at amino acid 351 controls the estrogen-like action at activating function 2b of the selective estrogen receptor modulator ERalpha complex.

Estrogens are involved in a multiplicity of programmed events in target tissues e.g.: uterus, breast, and pituitary gland, and hormone-responsive tumors occur at these target sites. We have addressed the possibility that all of the estrogens do not produce the same conformation of estrogen receptor alpha (ER). A novel assay in vitro was used to activate the transforming growth factor alpha (TGF-alpha) gene in situ in MDA-MB-231 cells stably transfected with cDNA for D351 ER or D351G ER. Three estrogen types were used: estradiol, diethylstilbestrol, and a triphenylethylene (TPE) derivative of tamoxifen without the antiestrogenic side chain. Computer molecular modeling was used to interpret data. A flat estrogen such as estradiol or diethylstilbestrol can induce TGF-alpha through a correctly positioned activating function 2 (AF2) and bind SRC-1. The TPE did not activate AF2 but activated the TGF-alpha gene through AF2b. This was demonstrated because D351 but not D351G ER activated the TGF-alpha gene with the TPE. We propose two classes of estrogens with different ER complexes that may incorporate different coactivators to function. Phytoestrogens and environmental xenoestrogens will fall into different classes based on structure and may exhibit selective actions and carcinogenic potential based on different ER conformations.

4-Hydroxytamoxifen (4-OHT), a selective estrogen receptor modulator, is an agonist at a transforming growth factor-alpha (TGF-alpha) target gene in situ in MDA-MB-231 human breast cancer cells stably transfected with wild-type human ERalpha. In contrast, raloxifene (Ral) is a complete antiestrogen silencing activation function (AF) 1 and AF2 in this system. A natural mutation D351YERalpha enhances 4-OHT agonist activity and changes Ral-like compounds from antagonists to partial agonists. We reasoned that: either the conformation of the Ral-D351YERalpha is altered, thereby reactivating AF2 in the ligand binding domain, or the change at amino acid 351 allosterically reactivates AF1 in the Ral-D351YERalpha complex. Unlike the estradiol-ERalpha complex, agonist activity of 4-OHT and raloxifene through ERalpha and D351YERalpha were not attributed to coactivator (such as SRC-1, AIB1) binding to the ligand binding domain. We conclude that the classic AF2 is not responsible for the agonist activities of 4-OHT-ERalpha, 4-OHT-D351YERalpha, and Ral-D351YERalpha. To address the role of AF1, stable transfectants of ERalpha or D351YERalpha with an AF1 deletion (D351deltaAF1, D351YdeltaAF1) were generated in MDA-MB-231 cells. Additionally, D538A/E542A/D545A triple mutations within helix 12 (D351-3m, D351Y3m) or the COOH-terminal 537 deletion (D351delta537) were tested. The agonist activities of 4-OHT and raloxifene were lost in these stable transfectants, but antiestrogenic action was retained. The reactivation of an estrogen-like property of the Ral-ERalpha complex through AF1 with the D351Y mutation illustrates a novel allosteric mechanism for the selective estrogen receptor modulator ERalpha complex.

The active metabolite of tamoxifen, 4-hydroxytamoxifen (4-OHT), is used in the laboratory for mechanistic studies of antiestrogen action. This compound binds to the estrogen receptor alpha (ER) and silences activating function 2 (AF2) in the ligand binding domain, but activating function 1 (AF1) at the other end of the ER remains constitutive and is considered to be ligand independent. Amino acid D351 in the ligand binding domain appears to be critical for interactions with the antiestrogenic side chain of antiestrogens. We have devised an assay to evaluate the biological activity of 351 mutant ERs and antiestrogens at the transforming growth factor alpha (TGFalpha) gene in situ (J. I. MacGregor Schafer et al., Cancer Res., 59: 4308-4313, 1999). The substitution of glycine for aspartate at position 351 results in the conversion of the 4-OHT:ER complex from estrogen-like to completely antiestrogenic. In cells stably expressing D351G ER, the ER retains responsiveness to estradiol (E2) and also retains antiestrogenic responsiveness to both raloxifene and ICI 182,780. The relative binding affinity of E2 for D351G ER (0.77 +/- 0.17 x 10(-9) M) is comparable with wild-type ER (0.42 +/- 0.08 x 10(-9) M). In addition, the D351G ER retains the ability to bind SRC-1 in the presence of E2, thus D351G ER AF2 activity has not been compromised. We also used a cell line stably expressing an ER with a triple mutation in helix 12 (D538A, E542A, and D545A) that ablated AF2 activity, which resulted in decreased effects of E2, suggesting that both AF1 and AF2 activity are required for maximal estrogen activity in MDA-MB-231 cells. Interestingly, the triple mutation also completely reduced the estrogen-like actions of 4-OHT. We propose that a specific mutation at amino acid 351 can allosterically silence AF1 in the 4-OHT:ER complex by either preventing the binding of coactivators or encouraging the binding of a corepressor molecule. We suggest that the 4-OHT-specific site responsible for estrogen-like actions can be referred to as AF2b. This binding site would consist of at least four carboxylic acids at amino acids 351 and 538, 542 and 545 in helix 12 to permit coactivator docking for gene activation. The AF2b site is distinct from AF2 for E2 action. Further studies will provide insight into the estrogen-like actions of tamoxifen in select tissues and breast tumors and identify a significant mechanism of drug resistance to tamoxifen.

A mutant human thymidylate synthase (TS) has been created in which a glutamine residue at position 214 has been replaced by glutamate. Glutamine at position 214 is postulated to be involved in maintaining the enzyme in a conformation that facilitates the binding of the substrate dUMP. Although the kcat/Km of the mutant protein for the substrate, dUMP, is 10(3) lower than that of wild-type TS, the mutant TS confers thymidine prototrophy on a TS-deficient bacterial strain when expressed at high levels. In the present investigation, a TS-deficient Chinese hamster lung cell line was transfected with DNA encoding the defective protein. Thymidine prototrophs were isolated that expressed the defective protein at levels that were physiologically relevant. The activities of the enzymes expressed endogenously in representative prototrophs were consistent with the activities observed for the purified proteins. At similar levels of TS expression, thymidine prototrophs expressing Glu214 TS were 8-fold more resistant to 5-fluoro-2'-deoxyuridine (FdUrd) cytotoxicity than are prototrophs expressing Gln214 TS. FdUrd is a prodrug of the tight-binding TS inhibitor, 5-fluoro-2'-deoxyuridine-5'-monophosphate (FdUMP). The resistance to FdUrd was associated with a significant decrease in the binding of FdUMP to the purified mutant enzyme. The data are consistent with the interpretation that TSs that are highly defective are capable of sufficient dTMP production for cell survival and optimal growth, yet may confer resistance to TS-directed inhibitors.

NMR has been employed for structural and dynamic studies of the bacterial response regulator, Spo0F. This 124-residue protein is an essential component of the sporulation phosphorelay signal transduction pathway in Bacillus subtilis. Three-dimensional 1H, 15N, and 13C experiments have been used to obtain full side chain assignments and the 1511 distance, 121 dihedral angle, and 80 hydrogen bonding restraints required for generating a family of structures (14 restraints per residue). The structures give a well-defined (alpha/beta)5 fold for residues 4-120 with average rms deviations of 0.59 A for backbone heavy atoms and 1.02 A for all heavy atoms. Analyses of backbone 15N relaxation measurements demonstrate relative rigidity in most regions of regular secondary structure with a generalized order parameter (S2) of 0.9 +/- 0.05 and a rotational correlation time (taum) of 7.0 +/- 0.5 ns. Loop regions near the site of phosphorylation have higher than average rms deviation values and T1/T2 ratios suggesting significant internal motion or chemical exchange at these sites. Additionally, multiple conformers are observed for the beta4-alpha4 loop and beta-strand 5 region. These conformers may be related to structural changes associated with phosphorylation and also indicative of the propensity this recognition surface has for differential protein interactions. Comparison of Spo0F structural features to those of other response regulators reveals subtle differences in the orientations of secondary structure in the putative recognition surfaces and the relative charge distribution of residues surrounding the site of phosphorylation. These may be important in providing specificity for protein-protein interactions and for determining the lifetimes of the phosphorylated state.

Transient phosphorylation at an aspartate residue on the Spo0F protein is a central step in the phosphorelay signal transduction pathway controlling sporulation in Bacilli. The response regulator Spo0F-P is stable to hydrolysis (t1/2 > 24 h at 23 degrees C in the absence of Mg2+), allowing the use of nondenaturing PAGE to separate the phosphorylated and non-phosphorylated forms of Spo0F. Using this novel assay, phosphoramidate containing compounds were found to specifically phosphorylate Spo0F, a reaction that requires the presence of a divalent metal, but mixed phosphate-carboxylate compounds did not act as phospho donors. Rapid hydrolysis of Spo0F-P generated with phosphoramidate by proteins downstream in the phosphorelay (Spo0B and Spo0A) is consistent with phosphorylation at the active site of Spo0F. The initial rate of Spo0F-P formation from phosphoramidate displays Michaelis-Menten kinetics, providing evidence for the proposal that response regulators, such as Spo0F, function as phosphoryl transferase enzymes (McCleary et al., 1993). The results establish that Spo0F functions as a phosphoryl transferase that uses exclusively a phosphoramidate rather than an acyl phosphate as substrate during autophosphorylation.

Spo0F, sporulation stage 0 F protein, a 124-residue protein responsible, in part, for regulating the transition of Bacillus subtilis from a vegetative state to a dormant endospore, has been studied by high-resolution NMR. The 1H, 15N, and 13C chemical shift assignments for the backbone residues have been determined from analyses of 3D spectra, 15N TOCSY-HSQC, 15N NOESY-HSQC, HNCA, and HN(CO)CA. Assignments for many sidechain proton resonances are also reported. The secondary structure, inferred from short- and medium-range NOEs, 3JHN alpha coupling constants, and hydrogen exchange patterns, define a topology consistent with a doubly wound (alpha/beta)5 fold. Interestingly, comparison of the secondary structure of Spo0F to the structure of the Escherichia coli response regulator, chemotaxis Y protein (CheY) (Volz K, Matsumura P, 1991, J Biol Chem 266:15511-15519; Bruix M et al., 1993, Eur J Biochem 215:573-585), show differences in the relative length of secondary structure elements that map onto a single face of the tertiary structure of CheY. This surface may define a region of binding specificity for response regulators. Magnesium titration of Spo0F, followed by amide chemical shift changes, gives an equilibrium dissociation constant of 20 +/- 5 mM. Amide resonances most perturbed by magnesium binding are near the putative site of phosphorylation, Asp 54.

The recent determination of the crystal structure of Escherichia coli thymidylate synthase (TS) [Matthews et al. (1989) J. Mol. Biol. 205, 449-454] has implicated the glutamic acid residue at position 58 in a mechanistic role which could involve the interaction of its gamma-carboxyl side chain with the nucleotide substrate and/or the folate cofactor. The site-specific mutagenesis of Glu-58 to Gln-58 in E. coli TS provided the opportunity to explore its functional role in activity and binding. When profiled by the spectrophotometric and tritium release assays, the 370- and 760-fold decreases, respectively, in kcat and the elevated Km values for the Gln-58 mutant enzyme indicated a significant involvement of Glu-58 in substrate binding and turnover. The apparent dissociation constant for the covalent FdUMP-enzyme binary complex was 30 microM, which is 5-fold higher than that found for the wild-type enzyme, while the inhibitory ternary complex apparent dissociation constants for FdUMP and CH2H4folate for the Gln-58 enzyme were 10- and 60-fold higher, respectively, than those for the wild-type enzyme under saturating conditions. The extent of covalent FdUMP binding to the Gln-58 enzyme was reduced from 1.5 to 0.7 per dimer in the inhibitory ternary complex but only from 0.7 to 0.5 per dimer in the binary complex of the Gln-58 enzyme. The usual 2.1-fold enhancement of FdUMP binding to wild-type TS in the presence of CH2H4folate was not observed for the Gln-58 enzyme.(ABSTRACT TRUNCATED AT 250 WORDS)

The binding of 5-fluorodeoxyuridylate (FdUMP) to carboxypeptidase-inactivated thymidylate synthase obtained from methotrexate-resistant Lactobacillus casei was investigated using [3H]FdUMP in a trichloroacetic acid precipitation assay and by 19F nuclear magnetic resonance spectroscopy. The cleavage of 1 valine residue from the carboxyl terminus of one of the identical subunits of the enzyme dimer correlates with complete loss of thymidylate synthesis (Aull, J. L., Loeble, R. B., and Dunlap, R. B. (1974) J. Biol. Chem. 249, 1167-1172). We have further investigated the phenomenon of carboxypeptidase A-dependent inactivation of thymidylate synthase by employing immobilized carboxypeptidase A in order to facilitate the isolation and characterization of the inactivated enzyme. The time course of carboxypeptidase treatment of thymidylate synthase has been profiled by the spectrophotometric assay, tritium release assay, trichloroacetic acid precipitation assay (covalent adduct analysis), 19F nuclear magnetic resonance spectroscopy, and amino acid analysis. The techniques utilized in this study yielded results which showed that the completely inactivated enzyme (failure to catalyze thymidylate formation) continued to catalyze both covalent FdUMP-enzyme interactions and the formation of the covalent inhibitory ternary complex with the cofactor, 5,1O-methylenetetrahydrofolate, although to a reduced extent, thus effectively uncoupling these processes from thymidylate synthesis activity.

Recombinant mouse thymidylate synthase (TS) expressed at high levels in Escherichia coli was purified to homogeneity in greater than 70% yield by a rapid three-step procedure. Both 0.1% Triton X-100 and 10% glycerol were required to stabilize the enzyme whose activity remained unchanged after 1 month when stored at -20 degrees C. Thermal inactivation of the enzyme was a first-order process at 37 degrees C, with t1/2 values of 6.9, 15.6 and 3.0 min at pH 5.5, 7.0 and 8.5, respectively. The presence of saturating levels of dUMP at pH 8.5 increased the t1/2 of inactivation of 38 min. The pH profile for enzyme activity showed a narrow optimum region centered at pH 7.0, which was mirrored by the shape of the Km, dUMP/Vmax plot. The pH dependence of Kd for the covalent inhibitory ternary complex of enzyme, 5-fluoro-2'-deoxyuridylate and 5,10-methylenetetrahydrofolate exhibited a broad minimum between pH 5.5 and 8.5, and ranged between 3.1, 0.8 and 1.1 nM at pH 5.5, 7.0 and 8.5, respectively. The UV/VIS spectrum of the native enzyme exhibited a maximum at 280 nm (epsilon = 98,200 M-1 cm-1), while that of the inhibitory ternary complex showed an additional maximum at 320 nm. The 19F-NMR spectrum of the mouse enzyme:FdUMP binary complex revealed two new resonances at -2.8 and -34.8 ppm. The most deshielded resonance represented the noncovalent binary complex while the other resonance was assigned to the nucleotide covalently bound to the enzyme. The alteration of nucleotide binding equilibria produced by addition of H4 folate was exemplified by both an increase in intensity and a 5 ppm deshielding of the resonance attributed to the covalent FdUMP-enzyme complex. Addition of formaldehyde to the latter mixture produced the covalent ternary complex which resulted in the collapse of the resonances at -2.8 and -39.5 ppm and the appearance of a new resonance at -12.4 ppm.

The results presented herein describe the development of a new method for the purification of thymidylate synthase from amethopterin-resistant L. casei. This method includes three chromatographic steps: batchwise CM-Sephadex, Q-Sepharose, and 10-formylfolate-Sepharose affinity steps, and the whole procedure can be performed in three days. Additionally, the procedure has consistently produced enzyme with a specific activity of 3.0 to 3.2. The first step, the batchwise purification of L. casei cell free extracts on CM-Sephadex, resulted in a 120-fold purification with 115% recovery of the activity applied to the resin. The Q-Sepharose purification step yielded a 1.5-fold purification producing enzyme with a specific activity of 2.4 units/mg. As the final step in the sequence, 10-formylfolate-Sepharose affinity chromatography yielded enzyme with a specific activity of 3.0-3.2 units/mg. The entire procedure provided a 230-fold purification with 67% recovery of the activity originally present in the cell free extract. However, the purified enzyme required extensive dialysis to achieve maximal activity. The data presented also clearly demonstrate the resolution of thymidylate synthase on the 10-formylfolate affinity column into forms which have different specific activity values and FdUMP binding ratios. Generally, enzyme with low activity leaches off this affinity matrix first, while enzyme with high activity elutes off last. These results are in accord with the sensitivity of the dimeric thymidylate synthase active site cysteines to oxidization. Consequently, when one of the two active site cysteines is oxidized, the enzyme loses the ability to bind FdUMP at this active site and its specific activity is proportionally lower. Thus, pure thymidylate synthase protein exists as a mixture of heterogeneous activity states. This condition is best examined by performing native PAGE on enzyme that has been incubated with excesses of FdUMP and CH2H4folate.