Sulfuryl Transfer: The Catalytic Mechanism of Human Estrogen Sulfotransferase*

Estrogen sulfotransferase (EST) catalyzes the transfer of the sulfuryl group from 3′-phosphoadenosine 5′-phosphosulfate (PAPS) to 17β-estradiol (E2). The sulfation of E2 prevents it from binding to, and thereby activating, the estrogen receptor. The regulation of EST appears to be causally linked to tumorigenesis in the breast and endometrium. In this study, recombinant human EST is characterized, and the catalytic mechanism of the transfer reaction is investigated in ligand binding and initial rate experiments. The native enzyme is a dimer of 35-kDa subunits. The apparent equilibrium constant for transfer to E2 is (4.5 ± 0.2) × 103 at pH 6.3 andT = 25 ± 2 °C. Initial rate studies provide the kinetic constants for the reaction and suggest a sequential mechanism. E2 is a partial substrate inhibitor (K i = 80 ± 5 nm). The binding of two E2 per EST subunit suggests that the partial inhibition occurs through binding at an allosteric site. In addition to providing the dissociation constants for the ligand-enzyme complexes, binding studies demonstrate that each substrate binds independently to the enzyme and that both the E·PAP·E2S andE·PAP·E2 dead-end complexes form. These results strongly suggest a Random Bi Bi mechanism with two dead-end complexes.

ϳ Ϫ19 kcal/mol) (7). Thus, the sulfuryl moiety is well positioned energetically for facile and favorable transfer to its metabolic recipients. The chemical and regulatory parallels between sulfation/desulfation and phosphorylation/dephosphorylation are quite strong, yet relatively little is know about the chemistry and enzymol-ogy of sulfuryl transfer.
The transcriptional activity of the estrogen receptor (ER) is regulated by sulfation/desulfation at the 3-hydroxyl position of 17␤-estradiol (E 2 ). The binding of E 2 to the ER, located in the nuclear membrane, elicits a complex cellular response that is rooted in the transcriptional activity of the ER⅐E 2 complex. E 2 binds tightly to the ER (K d ϳ 1 nM) (8,9); E 2 S, on the other hand, binds weakly, if at all (10). Thus, the ER-binding activity of estrogen (and, in turn, ER activation) is regulated by E 2 sulfation. The sulfation of E 2 (Reaction 1), is catalyzed by the enzyme estrogen sulfotransferase (3Ј-phosphoadenylylsulfate:estrone 3-sulfotransferase, EC 2.8.2.4). Recent studies indicate that the expression of this enzyme is causally linked to the estrogen-dependent growth response that is believed to underlie the genesis of epithelial breast tumors (11,12).
Human liver estrogen sulfotransferase (EST) has been cloned and expressed in Escherichia coli K1 (13). In this paper, recombinant EST purified to homogeneity from E. coli is physically characterized. The equilibrium constant for Reaction 1 is determined, and the catalytic mechanism of EST is evaluated in initial rate and ligand binding studies.

EXPERIMENTAL PROCEDURES
Materials-The buffers, salts, enzymes, and reagents, unless specified otherwise, were of the highest grades available from Sigma. [2,4,6, H]Estradiol (85 Ci/mmol) was purchased from NEN Life Science Products. Adenosine 3Ј,5Ј-[5Ј-32 P]bisphosphate (3000 Ci/mmol) was purchased from ICN Pharmaceuticals. PAPS was purchased from Professor S. Singer (University of Dayton, Dayton, OH). Factor Xa protease was purchased from Enzyme Research Labs. The Bradford assay reagents were purchased from Bio-Rad. The Superdex 200 column was purchased from Amersham Pharmacia Biotech. Amylose resin was obtained from New England Biolabs Inc.
Purification of Estrogen Sulfotransferase-Competent XL1-Blue (14) cells were transformed with the estrogen sulfotransferase expression vector pMBP-hEST-1 (13) and immediately used to inoculate a 100-ml culture of LB medium (15) containing ampicillin at 50 g/ml. The pMBP-hEST-1 vector directs the synthesis of a maltose-binding protein-human EST-1 fusion protein, which can be purified by amylose affinity chromatography (13). The culture was incubated overnight at 37°C and then used to inoculate 10 liters of LB medium to A 600 ϭ 0.06. At A 600 ϭ 0.6, isopropyl-␤-D-thiogalactopyranoside was added to 350 M. 3.5 h later, the cells were pelleted at 3500 ϫ g for 25 min, suspended in 500 ml of buffer containing 20 mM Tris-Cl (pH 7.4), 0.20 M KCl, and 5 mM ␤-mercaptoethanol, and frozen at Ϫ70°C. All of the purification steps were performed at 4°C. The cells were thawed, and 1 liter of 4°C lysis buffer (115 mM Tris-Cl (pH 8.0), 0.375 M sucrose, 0.375 mM EDTA, and 0.03 mg/ml lysozyme) was added. 20 min later, the cells were pelleted at 3500 ϫ g for 25 min. The cell pellet was suspended in sonication buffer (5.0 mM KPO 4 (pH 7.4), 1.5 mM DTT, and 57 M phenylmethylsulfonyl fluoride) and sonicated. The cellular debris was pelleted at 27,000 ϫ g for 60 min, and the concentration of protein, determined by the method of Bradford (16), was adjusted to 3.0 mg/ml using 5.0 mM KPO 4 (pH 7.4).
The protein solution was loaded onto a 100-ml bed (3.6 ϫ 17 cm) of amylose resin and equilibrated with 5.0 mM KPO 4 (pH 7.4). The column was then washed with 200 ml of wash buffer (10 mM NaH 2 PO 4 , 0.20 mM NaCl, 1.0 mM DTT, and 1.0 mM EDTA-HCl (pH 6.3)). The maltosebinding protein-EST fusion protein was then eluted from the column with wash buffer containing 10 mM maltose. 280 A 280 units of pure fusion protein was obtained. Glycerol was immediately added to the protein solution to 10% (v/v). Tris-HCl (1.0 M, pH 7.5) was then added to the protein solution to a final concentration of 25 mM. Factor Xa protease (0.66 mg/ml) was added to a solution of maltose-binding protein-EST fusion protein (A 280 ϭ 7.0) to a final concentration of 20 g/ml. The Factor Xa cut site has been engineered into the fusion protein such that the proteolytically produced EST has at its N terminus its initiator methionine (13). The proteolysis was allowed to proceed for 4 h at 4°C; the reaction reached ϳ90% completion. The proteolyzed solution was then purified by size-exclusion chromatography using a Superdex 200 (XK 27/70) column equilibrated with 10 mM NaH 2 PO 4 and 1.0 mM DTT. The fractions containing EST were pooled, concentrated using Amicon Centriprep-10 concentrators, and rerun over the Superdex column to remove the remaining traces of fusion protein and maltose-binding protein. The EST was pooled, and the solution was made 10% (v/v) in glycerol. The sample was concentrated to A 280 ϭ 1.6. The enzyme was aliquoted, frozen in a dry ice/EtOH bath, and stored at Ϫ70°C. Under these conditions, the enzyme showed no significant loss of activity over the ensuing 3-4 weeks.
Native Molecular Mass-The native molecular mass of EST was determined by size-exclusion chromatography using a Superdex 200 (XK 27/70) column. The column was equilibrated and run in 50 mM K ϩ -Hepes (pH 8.0) at 4°C. Bio-Rad gel filtration standards were used to calibrate the column. The apparent native molecular mass of EST was 62 Ϯ 2.3 kDa.
Extinction Coefficient-The extinction coefficient of EST was determined gravimetrically at 280 nm, the max for the enzyme. EST was dialyzed at 4°C for 8 h and then overnight against 10 mM NaPO 4 (pH 6.3), 1.0 mM KCl, and 0.10 mM DTT. The absorbance of the dialyzed EST was determined in triplicate at 280 nm in a masked 1-cm path length cuvette. Dialyzed enzyme or dialysis buffer (200 Ϯ 1.6 l) was added to an aluminum weigh boat and dried under vacuum and over P 2 O 5 to a constant weight (Ͻ40 h). Triplicate samples of the dialyzed enzyme and buffer were each weighed three times. The absorbance at 280 nm was divided by the concentration of enzyme to calculate the extinction coefficient: ⑀ 280 * ϭ 1.7 Ϯ 0.1 A 280 ϫ (mg/ml) Ϫ1 ϫ cm Ϫ1 . Equilibrium Constant-The equilibrium constant for the EST reaction was determined at pH 6.3 and T ϭ 22 Ϯ 2°C. The measurements were performed in duplicate at each of the following three sets of initial concentrations of E 2 S, [5Ј-32 P]PAP, and EST, respectively: 100 M, 240 nM, and 8.0 nM; 200 M, 240 nM, and 8.0 nM; and 500 M, 630 nM, and 32 nM. The progress curve for each reaction was determined, and the equilibrium constant was calculated from the reactant concentrations in the stationary phase of the progress curve. The buffer was the same as that used in the initial rate experiments. The concentration of product formed in each reaction was at least 10 times the enzyme active-site concentration. The equilibrium constant was (4.5 Ϯ 0.2) ϫ 10 3 .
Divalent Cation Activation-The initial rate of the forward reaction was studied as a function of MgCl 2 concentration. The assays were performed as described below under "Initial Rate Studies of the Forward Reaction." The conditions of these experiments were as follows: ranged from 0.2 to 5 times its K m . The concentrations of E 2 (2.0, 2.8, 4.8, and 15.5 nM) ranged from 0.4 to 3 times its K m . Partial substrate inhibition by E 2 is negligible at these E 2 concentrations. The velocities were measured under initial rate conditions since the consumption of the concentration-limiting substrate was Ͻ7% of its end point in all cases. The data were statistically fit to the sequen model using the program developed by Cleland (17).
Initial Rate Studies of the Reverse Reaction-5.0 l of EST (4.0 nM) was added to 15 l of solution containing PAP at varying concentrations and E 2 S at 250 M. The reactions were quenched by the addition of 4.0 l of 0.10 M KOH (final pH 10.0). 9.0 l of solution was spotted onto polyethyleneimine-fluorescamine TLC plates. The radiolabeled reactants were separated using a LiCl (0.50 M) and HCO 2 H (2.0 M) mobile phase and quantitated using an AMBIS two-dimensional radioactivity detector. The experiments were performed at 25 Ϯ 2°C. The data were statistically fit to the hyper model using the weighted least-squares program of Cleland (17).
Substrate Inhibition by E 2 -These initial rate assays were performed as described above for the forward reaction. The concentrations of EST and PAPS were 1.0 and 600 nM, respectively. To determine whether inhibition was caused by changes in k cat or the K m for PAPS, the rates at the highest concentrations of E 2 (i.e. the plateau region of the curve shown in Fig. 2) were performed at a 10 times higher concentration of PAPS (6.0 M). Each velocity was determined in duplicate. The data were fit using a partial substrate inhibition model (see "Results and Discussion") with the program Kaleidograph, which uses the Marquardt-Levenberg minimization algorithm.
Fluorescence Titrations-The EST fluorescence excitation and emission wavelength maxima are 275 and 340 nm, respectively. To avoid possible inner filter effects caused by the absorption of PAP and PAPS, a 285-nm excitation wavelength was used in all of the experiments. The highest nucleotide concentration used in the titrations (5.5 M) corresponds to an absorbance of 4.7 ϫ 10 Ϫ4 . This absorbance is ϳ100-fold below the threshold where inner filtering begins to influence intensity measurements (18). The emitted light was detected at 340 nm in all experiments except those involving E 2 , which is weakly fluorescent when excited at 285 nm. In experiments involving E 2 , the emission wavelength was set at 360 nm. At this wavelength, the emission from E 2 is negligibly low (Ͻ0.2% of the EST emission at an equivalent concentration). The solutions used in the titrations were thermally equilibrated and maintained at 25 Ϯ 2°C during the experiment. The buffer (50 mM KPO 4 (pH 6.3), 1.0 mM DTT, 7.0 mM MgCl 2 , and 10% (v/v) glycerol) was filtered using Gelman 0.2-m Acrodiscs. The fluorometer used in these studies was a Perkin-Elmer Model LS-5B; the entrance and exit slit widths were set at 10 nm. The titration data were fit to a single-site binding model using the Sigma Plot program, which employs the Marquardt-Levenberg fitting algorithm. The single-site binding model is described by a second order polynomial. The data were fit to the appropriate root of this polynomial to obtain the best fit binding constants.
The protocol of the titration experiments designed to determine K d differed from those intended to determine stoichiometry. The titrant in the K d experiments was a solution of EST and ligand in which the ligand concentration was ϳ10 times the highest ligand concentration reached in the titration. The titrant was added to a concentrationmatched solution that did not contain ligand. In the stoichiometry experiments, an EST solution that did not contain ligand was added to a concentration-matched solution that did.

Native Molecular Mass and Extinction Coefficient of EST-
The apparent native molecular mass of EST, determined by size-exclusion chromatography (see "Experimental Procedures"), is 62 (Ϯ2.3) ϫ 10 3 Da. The subunit molecular mass, predicted from the DNA sequence of the EST coding region, is 35,123 kDa (13). Thus, the native enzyme appears to be a dimer. The extinction coefficient of EST, determined gravimetrically, is 1.7 Ϯ 0.1 A 280 ϫ (mg/ml) Ϫ1 ϫ cm Ϫ1 (see "Experimental Procedures"). The experimentally determined ⑀ 280 * is identical to that calculated for EST from its amino acid composition and the ⑀ 280 * of Trp and Tyr (19). Stabilizing EST Activity-In the absence of glycerol, the activity (i.e. the initial rate of PAP and E 2 S synthesis) of EST is stable for several days at Ϫ70°C, and the half-life of the activity is ϳ2 h at 25°C. Glycerol (10%, v/v) prevents detecta-ble deterioration of the activity over 3 h at 25°C. The addition of E 2 to EST in glycerol causes a rapid loss of activity (t1 ⁄2 ϳ 30 min). This E 2 -induced inactivation was prevented by the addition of DTT. At 1.0 mM DTT, the activity was not affected at a saturating concentration of E 2 over 4 h at 25°C. Thus, it appears that the binding of E 2 potentiates an inactivating oxidation reaction that is suppressed by DTT.
Optimizing Turnover-Most enzyme-catalyzed transfer reactions involving nucleotides require divalent cations. In these cases, the cations are often directly bound to the polyphosphate chain of the nucleotide. The mechanism of this activation appears to be due predominantly to the entropy reduction associated with the positioning of functional groups for reaction (20). It is interesting that while sulfotransferases catalyze transfer reactions that, in many ways, resemble phosphoryl transfer reactions, they do not require divalent cations for activity. They are, however, activated by divalent cations. A plot of the initial rate of the forward EST reaction versus [MgCl 2 ] is bell-shaped with a maximum at 7.0 mM MgCl 2 (see "Experimental Procedures"). The initial rate at zero MgCl 2 and 2.0 mM EDTA is 0.18 times the initial rate at 7.0 mM MgCl 2 . The buffers used in the mechanism studies described in this paper contained MgCl 2 at 7.0 mM.
To determine an optimum pH for the EST mechanism studies, the initial rate of the forward reaction was studied as a function of pH at subsaturating E 2 and saturating PAPS concentrations. This condition maximizes turnover with respect to both the K m for E 2 and k cat . The assay protocol was as described under "Initial Rate Studies of the Forward Reaction", except for the following changes. EST at 0.50 nM and PAPS at 6.0 M were used, and the pH of the buffer (50 mM KPO 4 ) was varied in 0.4 pH unit increments between 5.4 and 7.4, inclusive, by mixing dibasic and monobasic solutions of KPO 4 . The pH rate profile was bell-shaped with a maximum initial rate at pH 6.3. The recent structure of mouse testis estrogen sulfotransferase implicates His-108 as a general base that abstracts a proton from the 3-hydroxyl group of E 2 , thereby activating it for attack (21). Further pH rate studies will help to determine whether this residue contributes to the pH rate dependence of the ESTcatalyzed reaction.
Equilibrium Constant-To evaluate the energetics associated with transferring the sulfuryl group between PAPS and E 2 and to aid in the design of initial rate experiments, the equilibrium constant for the EST reaction was determined at pH 6.3 and T ϭ 25 Ϯ 2°C (the conditions of the initial rate studies). The equilibrium constants were calculated from reactant concentrations in the stationary phases of reaction progress curves constructed in duplicate at three different sets of E 2 S and PAP concentrations (see "Experimental Procedures"). Controls were run to ensure that the EST activity did not change during the experiments. The equilibrium constant is (4.5 Ϯ 0.2) ϫ 10 3 . The ⌬G 0 associated with this K eq is Ϫ5.0 kcal/mol. It should be emphasized this apparent equilibrium constant does not explicitly include the proton and divalent cation dependences.
Initial Rate Study of the Forward Reaction-To determine the kinetic constants for the forward reaction and to obtain a preliminary assessment of the order of substrate binding, a classical initial rate study of the forward reaction was performed. The initial rate was determined as a function of both E 2 and PAPS concentrations (see "Experimental Procedures"). The results of this study are shown in Fig. 1. The pattern of the data is indicative of a sequential mechanism (one in which both substrates must bind to the enzyme before product is released). However, it does not rule out a ping-pong mechanism with an unstable enzyme intermediate. An equilibrium ordered mechanism is ruled out by the fact that the lines through the data of the 1/V versus 1/[E 2 ] and 1/V versus 1/[PAPS] (not shown) plots do not intersect on the 1/V axis (22). The kinetic constants obtained from this study are compiled in Table I. k cat (1.3 Ϯ 0.08 s Ϫ1 ) and the K m for PAPS (59 Ϯ 13 nM) are similar to those measured for other sulfotransferases (23,24). The K m for E 2 is comparable to the in vivo concentration of E 2 , ϳ1 nM (25), suggesting that the enzyme is optimized to perform at the physiological concentration of E 2 .
Initial Rate Study of the Reverse Reaction-The K m for PAP and k cat for the reverse reaction were determined in an initial rate study at a saturating (920 ϫ K d ) concentration of E 2 S (250 M). Controls were run to ensure that E 2 S did not inhibit the velocity at this concentration. The K m for PAP was 38 Ϯ 0.8 nM, and k cat was 0.16 Ϯ 0.0013 min Ϫ1 (Table I). The experimental protocol is described under "Experimental Procedures." Given the technical obstacles associated with the unfavorable equilibrium constant for the reverse reaction and the relatively high K m for E 2 S, the order of substrate addition for the reverse reaction was determined with the equilibrium binding studies described below.
Partial Substrate Inhibition by E 2 -The initial rate data shown in Fig. 2 demonstrate that E 2 inhibits the forward reaction. The fact that the velocity decreases to a plateau, rather than to zero, means that one or more of the kinetic parameters for the reaction are being titrated from one value to another as E 2 adds to the enzyme. The inhibition experiment (Fig. 2, q) was performed at a fixed near-saturating concentration of PAPS (600 nM, 10 ϫ K d ). If the inhibition were due solely to an increase in the K m for PAPS, causing the concentration of the reactive form(s) of the enzyme to decrease, increasing the concentration of PAPS would drive the initial rates, at inhibitory FIG. 1. An initial rate study of the synthesis of PAP and E 2 . The initial rate of PAP and E 2 S synthesis is shown as a function of PAPS and E 2 concentrations. The PAPS concentration was varied between 0.20 and 5.0 times its K m (59 nM). The E 2 concentration was varied between 0.4 and 3 times its K m (5.2 nM). The E 2 concentration range was sufficiently low that inhibition by E 2 was negligible. Each point represents the average of two independent determinations. The lines through the points represent the best fits to the data. The experiments were performed at 25 Ϯ 2°C. For the experimental protocol, see "Experimental Procedures." concentrations of E 2 , back to the uninhibited levels. This, in fact, does not occur. Increasing the concentration of PAPS 10-fold (Fig. 2, Ⅺ) had no significant effect on the initial rates at high E 2 concentrations. Thus, it is k cat that is affected by the binding of E 2 in these experiments. The K i for E 2 was evaluated by fitting the data shown in Fig.  2 to the algebra that describes the kinetic behavior of the model shown in Fig. 3. In the model, all of the enzyme forms are saturated with PAPS. The kinetic constants V 1 and K m for E 2 were obtained from the initial rate study of the forward reaction (Table I). V 2 was set at 0.18 nM/min, which is slightly below the plateau shown in Fig. 2. The data were fit to the following equation (26). This equation assumes that substrate binding is at equilibrium, which is plausible given the very low turnover of the enzyme. The value of K i that provides the best fit to the data is 80 Ϯ 5 nM. It should be mentioned that the inhibition model was used to select the E 2 concentrations used in the initial rate studies (Fig. 1) such that inhibition by E 2 was negligible.
Equilibrium Binding Studies-The excitation and emission wavelength maxima for EST are 275 and 340 nm, respectively. The fine structure and max of the emission spectrum do not change significantly when substrates bind to EST; however, the intensity decreases 30 -50% depending on the ligand. These ligand-dependent decreases in the quantum yield of EST provide excellent experimental handles to determine both the equilibrium constants and stoichiometry of the enzyme-ligand interactions.
The stoichiometry of the enzyme-ligand interactions was determined in fluorescence titration experiments in which the concentration of enzyme active sites was Ͼ15 ϫ K d . At these enzyme concentrations, the binding isotherms have two linear regions. They are linear in the substrate concentration range 0ϳ0.  Table I; V 1 was calculated from k cat (Table I), and V 2 was set at 0.18 min Ϫ1 , slightly below the initial rate in the plateau of Fig. 2. The algebra that describes the kinetic behavior of this model was used to obtain a best fit value for K i . 2:1. The binding of two E 2 to each EST subunit strongly suggests that one of the binding sites is the catalytic site, whereas the other is the allosteric site that regulates the turnover of the enzyme (Fig.  3). Notwithstanding the possibility that these fluorescence experiments monitor the formation of nonproductive complexes, the independent binding of each of the EST substrates demonstrates that the mechanism of the enzyme is random sequential.
The equilibrium constants for the formation of enzyme-substrate complexes were determined by fluorescence titrations in which the enzyme concentration was held fixed between 0.4 and 5 ϫ K d . The binary complex binding constants were determined by fitting the data shown in Fig. 5 (A, C, and D) (Table II). The data associated with the binding of E 2 (Fig. 5B) are equally well fit using either a single-site model (K d ϭ 26 Ϯ 2 nM; shown in Fig. 5B) or a two-site model that assumes no interaction energy (27) in which the dissociation constants are within an order of magnitude of one another and symmetrically disposed about the best fit K d predicted by the single-site model.
The formation of the ternary dead-end complexes was also investigated using fluorescence titrations (Fig. 5, E and F). The binding of PAPS to the E⅐E 2 S complex and of PAP to the E⅐E 2 complex was monitored at a saturating concentration of E 2 S or E 2 . The data clearly demonstrate that both dead-end complexes form. The nucleotide dissociation constants for the E⅐PAPS⅐E 2 S and E⅐PAP⅐E 2 complexes are 20 Ϯ 2.8 and 22 Ϯ 1.7 nM, respectively. Comparison of the binary and dead-end dissociation constants for PAPS and PAP reveals a slight binding synergism between the ligands in the ternary complex (Table II). These results corroborate the dead-end complexes implicated by earlier product inhibition studies with arylsulfotransferase (24) and strongly suggest that the mechanism of EST is Random Bi Bi with two dead-end complexes.
Conclusions-Initial rate and ligand binding experiments have been used to investigate the catalytic mechanism of EST. The kinetic parameters for the mechanism were determined from the initial rate studies, which also suggested that the mechanism is sequential. Ligand binding studies were used to determine the equilibrium constants and stoichiometries of the enzyme-substrate interactions. The binding studies demonstrated that each of the substrates can bind independently to the enzyme and that two dead-end complexes can form. These results strongly suggest a Random Bi Bi mechanism with two dead-end complexes. The initial rate experiments revealed that E 2 is a partial substrate inhibitor of the reaction with a K i of 80 Ϯ 5 nM. The mechanism of the inhibition is partially delineated by the stoichiometry studies, which show that the enzyme contains two E 2 -binding sites/catalytic subunit, suggesting that the enzyme harbors an allosteric E 2 -binding site.
FIG. 5. Binary and ternary (dead-end) complex formation. The affinities of enzyme-ligand interactions were determined by monitoring changes in the intrinsic fluorescence of EST as a function of ligand concentration. I/I 0 is the ratio of the fluorescence intensity of the enzyme at a given ligand concentration to that in the absence of ligand. The curves through the points represent the binding isotherms predicted by the best fit parameters obtained by fitting the experimental data to a single-site binding model. Each point represents the average of two to three independently determined values. The binary complexes were as follows: A, PAPS binding to EST; B, E 2 binding to EST; C, PAP binding to EST; D, E 2 S binding to EST. The ternary (dead-end) complexes were as follows: E, PAPS binding to EST⅐E 2 S; F, PAP binding to EST⅐E 2 . The EST concentrations associated with A-F were 75, 100, 75, 117, 150, and 100 nM, respectively. The concentrations of E 2 S and E 2 used in the ternary complex experiments were 6.0 M (22 ϫ K d ) and 1.5 M (Ͼ30 ϫ K d ), respectively.