0013-7227/90/1274-1770$02.00/0 Endocrinology Copyright © 1990 by The Endocrine Society
Vol. 127, No. 4 Printed in U.S.A.
Two Glucocorticoid Binding Sites on the Human Glucocorticoid Receptor* DEEPAK SRIVASTAVA AND E. BRAD THOMPSON Department of Human Biological Chemistry and Genetics, University of Texas Medical Branch, Galveston, Texas 77550
ABSTRACT. Glucocorticoids are known to have a lytic effect in leukemic cells via interactions with the glucocorticoid receptor (GR). Cortisol and various synthetic glucocorticoids bind to the GR with one-site kinetics. Cortivazol (CVZ) is a unique, high potency synthetic glucocorticoid, which has a phenylpyrazol fused to the A-ring of the steroid nucleus and displays binding consistent with two or more sites in the cytosol from CEM C7 cells (a human acute lymphoblastic T-cell line). It has previously been shown that the lower affinity class of sites are similar in affinity and site molarity to those recognized by dexamethasone. The higher affinity sites bind CVZ with 20- to 50-fold greater affinity, consistent with CVZ's enhanced biological effects. In mutant leukemic cells resistant to the lytic effects of dexameth-
F
ROM NUMEROUS observations regarding the binding characteristics of steroid hormones with their receptor protein, generally it has been concluded that steroid hormone receptors contain a single binding site for their respective ligands. In particular, Scatchard plots of steroidireceptor binding usually have resulted in monophasic, linear curves indicative of a homogeneous class of binding sites (1). Recently, however, some data have led to speculation that more complex receptorligand interactions exist modulated by yet poorly characterized hydrophobic zones (2, 3) as well as by the state of phosphorylation of the receptor protein (4, 5). The glucocorticoid receptor (GR), a predominantly cytosolic protein when not bound to ligand (6-8), is composed of several functional domains (9). The amino terminal region appears to be strongly immunogenic, and also includes a sector that appears to increase the protein's ability to activate transcription. The DNA-binding domain is a highly conserved sequence of about 65 amino acids (421-486 in the human GR) required for interactions with the hormone response element, specific cis-
Received April 13, 1990. Address requests for reprints to: Dr. E. Brad Thompson, Department of Human Biological Chemistry and Genetics, Route F-45, University of Texas Medical Branch, Galveston, Texas 77550. * This work was conducted in conjunction with the Walls Medical ^Research Foundation and supported by NIH Grant CA-41407 to E. Brad Thompson.
asone, CVZ both lyses the cells and recognizes a single class of sites similar to the high affinity site in CEM C7 cells. We have carried out experiments to define the nature of the higher affinity CVZ binding site. We now show that: 1) CVZ has more than one binding site in a second, independent, B-cell line, IM-9; 2) the antiglucocorticoid RU 38486 is able to block both CVZ's higher and lower affinity sites; 3) all of CVZ's binding sites are on a protein immunologically indistinguishable from the human GR; and 4) freshly isolated clones of CVZ-resistant cells have lost all binding sites for CVZ. These data indicate that CVZ is recognizing two glucocorticoid binding sites on the human GR or a protein very similar to it. (Endocrinology 127:1770-1778,1990)
acting DNA sequences to which the receptor binds to alter transcription (10). The carboxy terminal amino acids (528-777 in the human GR) comprise the steroidbinding domain, and in the absence of ligand appear to somehow maintain the receptor in a repressed state (11). When glucocorticoids bind to this region of the protein, the receptor, in vitro, is able to undergo a temperaturedependent activation, which results in exposure of the DNA-binding domain and thus in modulation of transcriptional events (12). Although this steroid-binding domain kinetically appears to have a single class of binding site for cortisol and most other standard natural and synthetic glucocorticoids, it has been shown to embody two distinct sequences, each of which can maintain the receptor in an unactivated state (13). In certain leukemic cells, glucocorticoids are known to have a lytic effect and thus are used for the treatment of human leukemias. Interestingly, a unique synthetic glucocorticoid (lla, 17/3, 21-trihydroxy-6-16o!-dimethyl-2/phenyl-2'-H-pregna-2,4,6-trieno [3,2-c] pyrazol-20-one 21 acetate) known as cortivazol (CVZ) (14) has been shown to be 20- to 50-fold more effective in lysing human leukemic cells than dexamethasone (DEX) (15), 40-fold more potent than DEX in inducing tyrosine aminotransferase in hepatoma cells (16), and effective in lysing even those cells that are resistant to the lytic effects of DEX (17). CVZ has the unique feature of a phenylpyrazol
1770
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 November 2015. at 01:02 For personal use only. No other uses without permission. . All rights reserved.
TWO GLUCOCORTICOID RECEPTOR AGONIST SITES
moiety fused to the A-ring of the cyclophenanthrene nucleus, which confers increased hydrophobicity to the molecule. When the binding of 3H-CVZ to cytosols from the human leukemic lymphoid cell line, CEM C7, was analyzed by the method of Scatchard, it was found to be consistent with at least two binding sites (18). For one of these sites, CVZ had an affinity similar to that seen between DEX and the GR, whereas for the other, CVZ bound with approximately 50-fold higher affinity. In the DEX-resistant, GR-defective leukemic cell clone, ICR27, derived from CEM C7, the cytosol retained a higher affinity binding site for CVZ, but showed no evidence for the lower affinity site (19). Because CVZ exhibits unusual receptor-ligand interactions as well as enhanced biological effects, we have carried out studies to explore further this novel glucocorticoid's binding behavior. Previously it has been shown that its lower affinity site in CEM C7 cells and the single site in the ICR-27 mutant are on the GR (17, 18). We here investigate whether the high affinity site for CVZ is also on a GR, as suggested by our earlier preliminary results (20) and whether that site is unique to CEM C7 cells. We have used antiglucocorticoids, immunological studies, and newly isolated CVZ-resistant cell lines. Materials and Methods Cells Cells of the lymphoid T-cell line CEM cloned and characterized previously were used. The properties of these clones have been described (21, 22). They are CEM C7 (wild type for GR) and ICR-27 (receptor mutants of CEM C7 with drastically reduced DEX binding sites, but only 50% reduction in GR protein) (23). The cell line ICCV described in this article is a CVZ-resistant subclone of ICR-27 cells. The lymphoid B-cell line, IM-9 (24), was also used. Cells were grown at 37 C in suspension culture in Roswell Park Memorial Institute (RPMI) tissue culture medium 1640, pH 7.4, containing 5-10% fetal calf serum, in a humidified atmosphere of 95% air and 5% CO2. All cells were harvested in the late logarithmic phase of growth at a density of ~1 x 106 cells/ml as determined by hemacytometer counts using trypan blue exclusion to determine viable cells (> 90%). Cytosol Cells were harvested and washed two times with phosphatebuffered isotonic saline, pH 7.4, at 4 C. Cells were resuspended in an equal volume of HEPES buffer (10 mM HEPES, 10 mM EDTA, 100 mM NaCl, 20 mM sodium molybdate, and 10% glycerol, pH 7.8) and immediately frozen at —70 C. Two to 4 h after freezing, cells were thawed at 4 C and found to be lysed. The lysate was centrifuged at 100,000 x g for 90 min in a Beckman L8-70M ultracentrifuge (Beckman, Palo Alto, CA), and the supernatant fraction (cytosol) was harvested. Its pro-
1771
tein content was estimated using a dye-binding assay (Bio-Rad Protein assay) with BSA as a standard (25). The cytosol was diluted to a protein concentration of 2.5-3.5 mg/ml for most binding assays and 10 mg/ml for immunoprecipitation reactions. Steroids DEX and [1,2,4-3H]DEX (47 Ci/mmol) were obtained from Sigma (St. Louis, MO) and New England Nuclear (Boston, MA), respectively. Aldosterone and [l,2,6,7-3H]aldosterone (77 Ci/mmol) were obtained from the same sources. CVZ was supplied through the kind offices of J. P. Raynaud, RousselUCLAF, Paris, France. 3H-CVZ (32 Ci/mmol) was custom prepared by Amersham (Arlington Heights, IL). Both were analyzed by HPLC and were shown to be >99% chromatographically pure. RU 38486 was synthesized at the RousselUCLAF Research Center and tested for purity (>98%) by TLC. Binding assays From cytosol preparations, 100-400-^1 aliquot fractions were incubated in duplicate overnight at 4 C with tritiated steroid with (competed) and without (uncompeted) 100- to 500-fold excess nonradioactive steroid. The larger volumes of cytosol were used when studying the binding of steroids at very low concentrations. Greater excess of nonradioactive steroid was used when cytosol with higher protein content was assayed with low concentrations of labeled steroid. No more than 1% ethanol carrier was added to any sample. The dextran-coated charcoal from 50 jil 5% suspension was collected by centrifugation at 10,000 X g for 2 min and aspiration of the supernatant. Cytosol was added to the charcoal pellet with gentle mixing. After 30 sec, the suspension was centrifuged at 10,000 X g for 2.5 min, and the supernatant transferred to 5 ml Beckman Ready Safe scintillation cocktail. The radioactivity in the sample was estimated by a Beckman LS 5801 Scintillation Counter. Specific binding was considered to be the radioactivity in uncompeted-competed fractions. All values expressed here were specifically bound amounts, unless stated otherwise. Immunoprecipitation reactions Labeled receptor. Cytosol in 200-^1 aliquots at 10 mg protein/ ml were incubated with high (10 nM) and low (0.5 nM) concentrations of [3H]steroid with or without 100-fold excess nonradioactive steroid. Nonspecific binding of the free ligand to antibody or staphalococcus aureus was estimated by use of a control containing only the radioactive steroid in HEPES buffer. A polyclonal anti-GR antiserum was raised in a rabbit against a synthetic peptide sequence from the immunogenic domain of the human GR (hGR) (amino acids 150-175). The peptide was synthesized by the University of Texas Medical Branch oligonucleotide/peptide synthesis lab on an Applied Biosystems 470A Protein Sequencer and purified by column chromatography. After taking a preimmune serum, a rabbit was injected with 100 mg peptide mixed in complete Freund's adjuvant in divided aliquots at multiple sc sites. Boosts in incomplete Freund's adjuvant were given after approximately 2 and 6 weeks and at longer intervals thereafter over the course
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 November 2015. at 01:02 For personal use only. No other uses without permission. . All rights reserved.
1772
TWO GLUCOCORTICOID RECEPTOR AGONIST SITES
Endo • 1990 Vol 127 • No 4
erably higher GR content than CEM C7 cells (27, 28). IM-9 cells have been shown to contain a typical GR but no receptors for other steroid hormones (29). Concentrations of [3H]CVZ in the specific binding assays were varied over a range of 0.03-52 nM. Figure 1A shows the total, the nonspecific, and the specifically bound ligand calculated from the difference between the total and nonspecific curves. Nonspecific binding was linear over the concentration range tested, indicating that the specifically bound counts were suitable for Scatchard analysis (30). The Scatchard plot of the data shown in Fig. IB reveals a concave curve similar to that seen in CEM C7 cytosol and indicative of two or more binding sites. This type of curve of the Scatchard plot could also be seen with negative cooperativity between ligand sites. However, when the data were displayed in a Hill plot (Fig. 1C), a Hill coefficient (31) of 1.0 was obtained, suggesting that negative cooperativity was not occurring. This was consistent with a previous report of the independence of CVZ sites in CEM C7 cells as determined by the dissoUnlabeled receptor. The GR in cytosol at 10 mg protein/ml was ciation method of DeMeyts (18). The mixture of horimmunoprecipitated as just described, except a monoclonal mone-binding sites was resolved by the method of Roantibody against purified holoreptor (GR) from CEM C7 cells senthal (32) for correction of curvilinear Scatchard plots. (26) was used, and the supernatant cytosol tested for CVZThis correction yielded the dashed line for the lower binding activity. Binding activity was compared with that in affinity site and the solid straight line for the higher supernatants from a control containing cytosol and S. aureus affinity site, assuming that there are two classes of sites. without antibody and an untreated control. Both the higher and lower affinity sites had similar dissociation constants (Kd) relative to those seen in CEM Isolation of CVZ-resistant clones C7 cells (7.5 X 10"10 M and 1.1 X 10"8 M, respectively). 6 Thus, the presence of more than one cytosolic binding ICR-27 cells were cloned by serial 1:10 dilutions from 10 to site for CVZ was established in a second, independent 10° cells in 1 ml medium in the wells of multiwell tissue culture plates. They were fed RPMI 1640 medium with 10% fetal calf human cell line. All subsequent experiments described serum and 10"6 M CVZ. Cells that grew out at the greatest in this report were performed with cytosol from CEM C7 dilution were grown to 105 cells/ml and then recloned by the or CEM C7 derived cell lines because CVZ's properties same process two additional times to obtain resistant clones. have been more thoroughly studied in these cells. 3
of 1 year. The animal seroconverted after the third set of injections, and the titer of antibody continued to improve with the subsequent injections. The antiserum obtained was used for immunoprecipitation reactions. After cytosols had been incubated with steroid for 3 h at 4 C, the anti-GR antiserum was added to all fractions at dilutions of 1:20 to 1:40, and the mixture incubated overnight. A 10% vol/vol suspension of formalin-fixed S. aureus (100 n\) obtained from Sigma was washed and resuspended in HEPES buffer (100 n\) and added to the cytosol. The cytosol was incubated for 30 min at 4 C with frequent mixing. The immunoprecipitate was then isolated by centrifuging at 10,000 x g for 10 min and washed five times with HEPES buffer + 0.1% Triton X-100. By the fifth wash, no further radioactivity was observed by liquid scintillation counting of the washes. The pellets were then resuspended in 500 ^1 HEPES buffer and the suspension assayed for radioactivity. Radioactivity in the competed precipitates was identical to that in the control precipitate containing no cytosol. Therefore, specific counts were determined by subtracting radioactivity in competed samples from that in uncompeted samples, both treated to form immunoprecipitates.
Clones arose in wells containing 10 cells in the first cycle, and in wells with 102 cells in cycles 2 and 3 of cloning. This is consistent with our experience of the cloning efficiency of these cells in the absence of a feeder layer. Cells were then grown in large volumes in 10~6 M CVZ. Four clones, from independent wells of the first cloning cycle, were thus isolated. Six days before assay, the CVZ-containing medium was replaced with fresh growth medium in a protocol aimed at removing the steroid. For details, see text.
Results Binding of CVZ in IM-9 cytosol We have previously reported that CVZ binds in CEM C7 cytosol with kinetics consistent with a two-binding site model (18). However, the possibility existed that the presence of two binding sites for CVZ was unique to these clonal, leukemic T cells. Therefore, we performed a similar analysis with cytosolic preparations of IM-9 cells, an independent lymphoid B-cell line with consid-
Antiglucocorticoids help resolve CVZ's binding sites RU 38486 is a molecule that has been documented as being a powerful antiglucocorticoid and as displaying high affinity binding for the human GR (hGR), including the hGR of lymphoid cells (33, 34). It is also known to bind strongly to progesterone receptors; however, the progesterone receptor is not expressed in CEM C7 cells. Because of its high affinity and specificity for hGR, we used RU 38486 to help resolve whether CVZ was binding to the classic GR or to another protein. Cytosol from CEM C7 cells was incubated either with 0.5 nM tritiated CVZ to bind predominantly its high affinity site, with 20 nM CVZ to occupy total CVZ sites, or with 20 nM DEX to occupy its single site. In parallel tubes, 100-fold excess nonradioactive RU 38486 was added as well. Results are displayed in Fig. 2 and show that all specific CVZ-binding at both the high affinity and low affinity sites was
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 November 2015. at 01:02 For personal use only. No other uses without permission. . All rights reserved.
TWO GLUCOCORTICOID RECEPTOR AGONIST SITES
1
100
3.0
1773
•—•TOTAL O—ONONSPeanc
C33H SUreld UnoempaUd E3 3 H Steroid + 1H Steroid E3 3 H Steroid + 'H RU 38486
A—Asnanc 75
50
25
20nM DEX
10
20
30
40
50
60
[ 3 H]C0RTIVAZ0L(nM)
20nMCVZ
.5nMCVZ
FiG. 2. RU 38486 competition for CVZs high and low affinity binding sites. CEM C7 cytosol was incubated with 20 nM 3H DEX, 20 nM 3HCVZ (both sites), or 0.5 nM 3H-CVZ (high affinity site). 3H-Steroid bound was considered 100% and shown as clear bars. The 100% values, left to right, were 424, 456, and 46 fmol/mg protein. H indicates each ^-steroid bound in the presence of 100-fold excess of its respective nonradioactive steroid. M indicates amount of 3H-steroid bound in the presence of excess RU 38486. Means for 5 different incubations are shown and standard deviation is given.
blocked by the addition of unlabeled RU 38486. RU 38486 also blocked binding of DEX to its single site. Thus, it appears that the high affinity CVZ-binding site, as is its lower affinity site, is localized to proteins that also bind RU 38486, a property of the well characterized hGR. 0.1
-2.250 -2.250
-1.500
OJ 03 0.4 BOUND (pmol/mg PROTEIN)
-0.750
0.000
0.750
1.500
0.5
2.250
3
LOG1O [ H]CORTIVAZOL(nM) 3
FIG. 1. Binding of H-CVZ to IM-9 cytosol. Upper panel, Aliquots of cytosol were incubated with increasing concentrations of 3H-CVZ (0.03 nM-52 nM) in the absence (total binding) or presence (nonspecific) of excess nonradioactive CVZ. Specific binding (A • ) was determined by subtracting nonspecific binding (O O) from total binding (• • ) Middle panel, Scatchard analysis of binding of [3H]CVZ to IM-9 cytosol. Data from two separate experiments are shown to better define the curvilinear nature, but the results from this combination of data do not significantly differ from the Kds or site molarities obtained in the experiments separately. The straight lines were derived by the method of Rosenthal (32). Kds obtained from the reciprocal slopes, and site molarities (N) calculated from the x intercepts. Values were as follows: High affinity Kd = 5 X 10"10 M, n = 0.025 pmol/mg; low affinity Kd = 1.2 x 10~8 M, n = 0.425 pmol/mg. Lower panel, Hill plot of the
Use of anti-hGR antibodies to resolve CVZ'-binding sites
To determine if CVZ at its high affinity site was binding to the GR, we used antibodies (Ab) raised against the hGR or a synthetic fragment thereof. Because CEM C7 cytosol contains two classes of CVZ-binding sites, we used a high (10 nM) and a low (0.5 nM) concentration of 3 H-CVZ to label all the binding proteins responsible for the total CVZ sites or predominantly those providing the high affinity sites, respectively. Immunoprecipitation was carried out as described in Materials and Methods, and the S. aureus-Ab-GR-3H-CVZ pellet counted for the presence of labeled GR. At 10 nM CVZ, both CVZ sites should have been occupied, and, not surprisingly, CVZlabeled GR was found in the precipitate. However, because the concentration of the low affinity site is much greater than the high affinity site, we could not distinguish whether the high affinity site was precipitated or not. To solve this, we used a concentration of 0.5 nM CVZ at which, according to simple Michaelis-Menton kinetics, 90% of the CVZ specifically bound would be at the high affinity site and 10% at the low affinity site. A concentrated solution of cytosol was used (protein = 10 mg/ml) to provide sufficient high affinity sites for labeling. Under these conditions, the anti-GR antibodies were binding data. Data from the experiment in Fig. 1A are shown and reveal a Hill coefficient of 1.0; inclusion of the additional data points from a second experiment used in Fig. IB does not alter this result.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 November 2015. at 01:02 For personal use only. No other uses without permission. . All rights reserved.
TWO GLUCOCORTICOID RECEPTOR AGONIST SITES
1774
TABLE 1. Immunoprecipitation of 3H CVZ-labeled cytosol with antiGR antibodies Steroid + cytosol 10 nM CVZ + CEM C7 0.5 nM CVZ + CEM C7 2 nM CVZ + ICR-27 10 nM CVZ + buffer (control)
Specific [3H]CVZ Specific [3H]CVZ bound immunoprecipitated (fmol/mg) (fmol/mg) 125.2 ± 4.3 20.5 ± 1.2 21.5 ± 0.9 —
40.0 ± 6.4 ± 6.2 ± 0.7 ±
2.3 0.3 0.4 0.1
CEM C7 cytosol was incubated with 10 nM CVZ and 0.5 nM CVZ (to predominantly label the high affinity site); ICR-27 cytosol was incubated with 2 nM CVZ to label its single site. Separate fractions were competed with 100-fold excess unlabeled CVZ. Specific binding activity of the cytosol preparation used for each condition was determined in parallel incubations and is shown in the middle column. AntiGR antiserum was added to all fractions overnight, and immune complexes precipitated with S. aureus as described in Materials and Methods. Radioactivity in competed precipitates were subtracted from that in uncompeted precipitates to obtain specific amounts of 3H-CVZ precipitated. A control precipitate of buffer + 3H CVZ + Ab + S. aureus yielded similar counts relative to the competed fractions and is shown above. The percent of bound CVZ immunoprecipitated was calculated by dividing the second column by the last column. Results are means from duplicate samples with ranges shown.
still able to precipitate the protein to which CVZ was bound (Table 1). Thus, CVZ's high affinity site is indeed on a protein recognized by the antiserum prepared against a specific synthetic peptide from the immunogenic domain of the GR. To confirm this further, cytosol from ICR-27 cells was used. This cell line, which was selected from CEM C7 cells for resistance to DEX, has lost the ability to bind DEX in standard assays, yet still expresses normal basal amounts of GR messenger RNA and ~50% GR protein relative to CEM C7 (22). CVZ binds to a single class of sites in these cells with a high affinity, and based on physical properties, this site is believed to be on a protein very similar to the GR (19). Thus, we used 3H-CVZ (2 nM) to label the CVZ binding protein and then precipitated the hGR with anti-GR antisera. Again, we were able to precipitate 3H CVZ, indicating that the single site in ICR-27's was indeed on a protein immunologically indistinguishable from the traditional GR. To use a slightly different approach for resolving whether all of CVZ's binding sites are on the GR or not, we first immunoprecipitated the unlabeled hGR with a monoclonal antibody to holo hGR, then tested the remaining hGR-free cytosol for CVZ and DEX binding activity. Results in Table 2 show that at both low and high concentrations of CVZ and DEX, there was no residual binding in the supernatant cytosol over the immunoprecipitate. The possibilities of alterations in the cytosol secondary to S. aureus or other extrinsic factors were excluded by appropriate parallel controls in which all immunoprecipitation steps were performed without the presence of Ab. The control supernatant cytosol
Endo • 1990 Vol 127 • No 4
TABLE 2. CVZ, DEX, and aldosterone binding in GR-free CEM C7 cytosol Cytosol treatment
Ligand tested for binding
No immune reagents +S. aureus
20 nM CVZ 20 nM CVZ 20 nM aldosterone +Anti-GR + S. aureus 20 nM aldosterone 20 nM CVZ 0.5 nM CVZ 20 nM DEX
Spec. 3H ligand bound (pmol/mg) 0.35 ± 0.35 ± 0.26 ± 0.25 ± 0.00 ± 0.00 ± 0.00 ±
0.3 0.2 0.2 0.1 0.0 0.0 0.0
Unlabeled CEM C7 cytosol was immunoprecipitated using hybridoma-conditioned medium containing a monoclonal anti-GR antibody and S. aureus, as described in Materials and Methods. The supernatant cytosol was tested for binding activity with CVZ, DEX, and aldosterone. Controls included cytosol carried through all incubations but without treatment with immune reagents and cytosol treated with S. aureus only. Results are averages from duplicate samples with ranges given.
retained CVZ-binding after addition of S. aureus compared with cytosol untreated with
Vol. 127, No. 4 Printed in U.S.A.
Two Glucocorticoid Binding Sites on the Human Glucocorticoid Receptor* DEEPAK SRIVASTAVA AND E. BRAD THOMPSON Department of Human Biological Chemistry and Genetics, University of Texas Medical Branch, Galveston, Texas 77550
ABSTRACT. Glucocorticoids are known to have a lytic effect in leukemic cells via interactions with the glucocorticoid receptor (GR). Cortisol and various synthetic glucocorticoids bind to the GR with one-site kinetics. Cortivazol (CVZ) is a unique, high potency synthetic glucocorticoid, which has a phenylpyrazol fused to the A-ring of the steroid nucleus and displays binding consistent with two or more sites in the cytosol from CEM C7 cells (a human acute lymphoblastic T-cell line). It has previously been shown that the lower affinity class of sites are similar in affinity and site molarity to those recognized by dexamethasone. The higher affinity sites bind CVZ with 20- to 50-fold greater affinity, consistent with CVZ's enhanced biological effects. In mutant leukemic cells resistant to the lytic effects of dexameth-
F
ROM NUMEROUS observations regarding the binding characteristics of steroid hormones with their receptor protein, generally it has been concluded that steroid hormone receptors contain a single binding site for their respective ligands. In particular, Scatchard plots of steroidireceptor binding usually have resulted in monophasic, linear curves indicative of a homogeneous class of binding sites (1). Recently, however, some data have led to speculation that more complex receptorligand interactions exist modulated by yet poorly characterized hydrophobic zones (2, 3) as well as by the state of phosphorylation of the receptor protein (4, 5). The glucocorticoid receptor (GR), a predominantly cytosolic protein when not bound to ligand (6-8), is composed of several functional domains (9). The amino terminal region appears to be strongly immunogenic, and also includes a sector that appears to increase the protein's ability to activate transcription. The DNA-binding domain is a highly conserved sequence of about 65 amino acids (421-486 in the human GR) required for interactions with the hormone response element, specific cis-
Received April 13, 1990. Address requests for reprints to: Dr. E. Brad Thompson, Department of Human Biological Chemistry and Genetics, Route F-45, University of Texas Medical Branch, Galveston, Texas 77550. * This work was conducted in conjunction with the Walls Medical ^Research Foundation and supported by NIH Grant CA-41407 to E. Brad Thompson.
asone, CVZ both lyses the cells and recognizes a single class of sites similar to the high affinity site in CEM C7 cells. We have carried out experiments to define the nature of the higher affinity CVZ binding site. We now show that: 1) CVZ has more than one binding site in a second, independent, B-cell line, IM-9; 2) the antiglucocorticoid RU 38486 is able to block both CVZ's higher and lower affinity sites; 3) all of CVZ's binding sites are on a protein immunologically indistinguishable from the human GR; and 4) freshly isolated clones of CVZ-resistant cells have lost all binding sites for CVZ. These data indicate that CVZ is recognizing two glucocorticoid binding sites on the human GR or a protein very similar to it. (Endocrinology 127:1770-1778,1990)
acting DNA sequences to which the receptor binds to alter transcription (10). The carboxy terminal amino acids (528-777 in the human GR) comprise the steroidbinding domain, and in the absence of ligand appear to somehow maintain the receptor in a repressed state (11). When glucocorticoids bind to this region of the protein, the receptor, in vitro, is able to undergo a temperaturedependent activation, which results in exposure of the DNA-binding domain and thus in modulation of transcriptional events (12). Although this steroid-binding domain kinetically appears to have a single class of binding site for cortisol and most other standard natural and synthetic glucocorticoids, it has been shown to embody two distinct sequences, each of which can maintain the receptor in an unactivated state (13). In certain leukemic cells, glucocorticoids are known to have a lytic effect and thus are used for the treatment of human leukemias. Interestingly, a unique synthetic glucocorticoid (lla, 17/3, 21-trihydroxy-6-16o!-dimethyl-2/phenyl-2'-H-pregna-2,4,6-trieno [3,2-c] pyrazol-20-one 21 acetate) known as cortivazol (CVZ) (14) has been shown to be 20- to 50-fold more effective in lysing human leukemic cells than dexamethasone (DEX) (15), 40-fold more potent than DEX in inducing tyrosine aminotransferase in hepatoma cells (16), and effective in lysing even those cells that are resistant to the lytic effects of DEX (17). CVZ has the unique feature of a phenylpyrazol
1770
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 November 2015. at 01:02 For personal use only. No other uses without permission. . All rights reserved.
TWO GLUCOCORTICOID RECEPTOR AGONIST SITES
moiety fused to the A-ring of the cyclophenanthrene nucleus, which confers increased hydrophobicity to the molecule. When the binding of 3H-CVZ to cytosols from the human leukemic lymphoid cell line, CEM C7, was analyzed by the method of Scatchard, it was found to be consistent with at least two binding sites (18). For one of these sites, CVZ had an affinity similar to that seen between DEX and the GR, whereas for the other, CVZ bound with approximately 50-fold higher affinity. In the DEX-resistant, GR-defective leukemic cell clone, ICR27, derived from CEM C7, the cytosol retained a higher affinity binding site for CVZ, but showed no evidence for the lower affinity site (19). Because CVZ exhibits unusual receptor-ligand interactions as well as enhanced biological effects, we have carried out studies to explore further this novel glucocorticoid's binding behavior. Previously it has been shown that its lower affinity site in CEM C7 cells and the single site in the ICR-27 mutant are on the GR (17, 18). We here investigate whether the high affinity site for CVZ is also on a GR, as suggested by our earlier preliminary results (20) and whether that site is unique to CEM C7 cells. We have used antiglucocorticoids, immunological studies, and newly isolated CVZ-resistant cell lines. Materials and Methods Cells Cells of the lymphoid T-cell line CEM cloned and characterized previously were used. The properties of these clones have been described (21, 22). They are CEM C7 (wild type for GR) and ICR-27 (receptor mutants of CEM C7 with drastically reduced DEX binding sites, but only 50% reduction in GR protein) (23). The cell line ICCV described in this article is a CVZ-resistant subclone of ICR-27 cells. The lymphoid B-cell line, IM-9 (24), was also used. Cells were grown at 37 C in suspension culture in Roswell Park Memorial Institute (RPMI) tissue culture medium 1640, pH 7.4, containing 5-10% fetal calf serum, in a humidified atmosphere of 95% air and 5% CO2. All cells were harvested in the late logarithmic phase of growth at a density of ~1 x 106 cells/ml as determined by hemacytometer counts using trypan blue exclusion to determine viable cells (> 90%). Cytosol Cells were harvested and washed two times with phosphatebuffered isotonic saline, pH 7.4, at 4 C. Cells were resuspended in an equal volume of HEPES buffer (10 mM HEPES, 10 mM EDTA, 100 mM NaCl, 20 mM sodium molybdate, and 10% glycerol, pH 7.8) and immediately frozen at —70 C. Two to 4 h after freezing, cells were thawed at 4 C and found to be lysed. The lysate was centrifuged at 100,000 x g for 90 min in a Beckman L8-70M ultracentrifuge (Beckman, Palo Alto, CA), and the supernatant fraction (cytosol) was harvested. Its pro-
1771
tein content was estimated using a dye-binding assay (Bio-Rad Protein assay) with BSA as a standard (25). The cytosol was diluted to a protein concentration of 2.5-3.5 mg/ml for most binding assays and 10 mg/ml for immunoprecipitation reactions. Steroids DEX and [1,2,4-3H]DEX (47 Ci/mmol) were obtained from Sigma (St. Louis, MO) and New England Nuclear (Boston, MA), respectively. Aldosterone and [l,2,6,7-3H]aldosterone (77 Ci/mmol) were obtained from the same sources. CVZ was supplied through the kind offices of J. P. Raynaud, RousselUCLAF, Paris, France. 3H-CVZ (32 Ci/mmol) was custom prepared by Amersham (Arlington Heights, IL). Both were analyzed by HPLC and were shown to be >99% chromatographically pure. RU 38486 was synthesized at the RousselUCLAF Research Center and tested for purity (>98%) by TLC. Binding assays From cytosol preparations, 100-400-^1 aliquot fractions were incubated in duplicate overnight at 4 C with tritiated steroid with (competed) and without (uncompeted) 100- to 500-fold excess nonradioactive steroid. The larger volumes of cytosol were used when studying the binding of steroids at very low concentrations. Greater excess of nonradioactive steroid was used when cytosol with higher protein content was assayed with low concentrations of labeled steroid. No more than 1% ethanol carrier was added to any sample. The dextran-coated charcoal from 50 jil 5% suspension was collected by centrifugation at 10,000 X g for 2 min and aspiration of the supernatant. Cytosol was added to the charcoal pellet with gentle mixing. After 30 sec, the suspension was centrifuged at 10,000 X g for 2.5 min, and the supernatant transferred to 5 ml Beckman Ready Safe scintillation cocktail. The radioactivity in the sample was estimated by a Beckman LS 5801 Scintillation Counter. Specific binding was considered to be the radioactivity in uncompeted-competed fractions. All values expressed here were specifically bound amounts, unless stated otherwise. Immunoprecipitation reactions Labeled receptor. Cytosol in 200-^1 aliquots at 10 mg protein/ ml were incubated with high (10 nM) and low (0.5 nM) concentrations of [3H]steroid with or without 100-fold excess nonradioactive steroid. Nonspecific binding of the free ligand to antibody or staphalococcus aureus was estimated by use of a control containing only the radioactive steroid in HEPES buffer. A polyclonal anti-GR antiserum was raised in a rabbit against a synthetic peptide sequence from the immunogenic domain of the human GR (hGR) (amino acids 150-175). The peptide was synthesized by the University of Texas Medical Branch oligonucleotide/peptide synthesis lab on an Applied Biosystems 470A Protein Sequencer and purified by column chromatography. After taking a preimmune serum, a rabbit was injected with 100 mg peptide mixed in complete Freund's adjuvant in divided aliquots at multiple sc sites. Boosts in incomplete Freund's adjuvant were given after approximately 2 and 6 weeks and at longer intervals thereafter over the course
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 November 2015. at 01:02 For personal use only. No other uses without permission. . All rights reserved.
1772
TWO GLUCOCORTICOID RECEPTOR AGONIST SITES
Endo • 1990 Vol 127 • No 4
erably higher GR content than CEM C7 cells (27, 28). IM-9 cells have been shown to contain a typical GR but no receptors for other steroid hormones (29). Concentrations of [3H]CVZ in the specific binding assays were varied over a range of 0.03-52 nM. Figure 1A shows the total, the nonspecific, and the specifically bound ligand calculated from the difference between the total and nonspecific curves. Nonspecific binding was linear over the concentration range tested, indicating that the specifically bound counts were suitable for Scatchard analysis (30). The Scatchard plot of the data shown in Fig. IB reveals a concave curve similar to that seen in CEM C7 cytosol and indicative of two or more binding sites. This type of curve of the Scatchard plot could also be seen with negative cooperativity between ligand sites. However, when the data were displayed in a Hill plot (Fig. 1C), a Hill coefficient (31) of 1.0 was obtained, suggesting that negative cooperativity was not occurring. This was consistent with a previous report of the independence of CVZ sites in CEM C7 cells as determined by the dissoUnlabeled receptor. The GR in cytosol at 10 mg protein/ml was ciation method of DeMeyts (18). The mixture of horimmunoprecipitated as just described, except a monoclonal mone-binding sites was resolved by the method of Roantibody against purified holoreptor (GR) from CEM C7 cells senthal (32) for correction of curvilinear Scatchard plots. (26) was used, and the supernatant cytosol tested for CVZThis correction yielded the dashed line for the lower binding activity. Binding activity was compared with that in affinity site and the solid straight line for the higher supernatants from a control containing cytosol and S. aureus affinity site, assuming that there are two classes of sites. without antibody and an untreated control. Both the higher and lower affinity sites had similar dissociation constants (Kd) relative to those seen in CEM Isolation of CVZ-resistant clones C7 cells (7.5 X 10"10 M and 1.1 X 10"8 M, respectively). 6 Thus, the presence of more than one cytosolic binding ICR-27 cells were cloned by serial 1:10 dilutions from 10 to site for CVZ was established in a second, independent 10° cells in 1 ml medium in the wells of multiwell tissue culture plates. They were fed RPMI 1640 medium with 10% fetal calf human cell line. All subsequent experiments described serum and 10"6 M CVZ. Cells that grew out at the greatest in this report were performed with cytosol from CEM C7 dilution were grown to 105 cells/ml and then recloned by the or CEM C7 derived cell lines because CVZ's properties same process two additional times to obtain resistant clones. have been more thoroughly studied in these cells. 3
of 1 year. The animal seroconverted after the third set of injections, and the titer of antibody continued to improve with the subsequent injections. The antiserum obtained was used for immunoprecipitation reactions. After cytosols had been incubated with steroid for 3 h at 4 C, the anti-GR antiserum was added to all fractions at dilutions of 1:20 to 1:40, and the mixture incubated overnight. A 10% vol/vol suspension of formalin-fixed S. aureus (100 n\) obtained from Sigma was washed and resuspended in HEPES buffer (100 n\) and added to the cytosol. The cytosol was incubated for 30 min at 4 C with frequent mixing. The immunoprecipitate was then isolated by centrifuging at 10,000 x g for 10 min and washed five times with HEPES buffer + 0.1% Triton X-100. By the fifth wash, no further radioactivity was observed by liquid scintillation counting of the washes. The pellets were then resuspended in 500 ^1 HEPES buffer and the suspension assayed for radioactivity. Radioactivity in the competed precipitates was identical to that in the control precipitate containing no cytosol. Therefore, specific counts were determined by subtracting radioactivity in competed samples from that in uncompeted samples, both treated to form immunoprecipitates.
Clones arose in wells containing 10 cells in the first cycle, and in wells with 102 cells in cycles 2 and 3 of cloning. This is consistent with our experience of the cloning efficiency of these cells in the absence of a feeder layer. Cells were then grown in large volumes in 10~6 M CVZ. Four clones, from independent wells of the first cloning cycle, were thus isolated. Six days before assay, the CVZ-containing medium was replaced with fresh growth medium in a protocol aimed at removing the steroid. For details, see text.
Results Binding of CVZ in IM-9 cytosol We have previously reported that CVZ binds in CEM C7 cytosol with kinetics consistent with a two-binding site model (18). However, the possibility existed that the presence of two binding sites for CVZ was unique to these clonal, leukemic T cells. Therefore, we performed a similar analysis with cytosolic preparations of IM-9 cells, an independent lymphoid B-cell line with consid-
Antiglucocorticoids help resolve CVZ's binding sites RU 38486 is a molecule that has been documented as being a powerful antiglucocorticoid and as displaying high affinity binding for the human GR (hGR), including the hGR of lymphoid cells (33, 34). It is also known to bind strongly to progesterone receptors; however, the progesterone receptor is not expressed in CEM C7 cells. Because of its high affinity and specificity for hGR, we used RU 38486 to help resolve whether CVZ was binding to the classic GR or to another protein. Cytosol from CEM C7 cells was incubated either with 0.5 nM tritiated CVZ to bind predominantly its high affinity site, with 20 nM CVZ to occupy total CVZ sites, or with 20 nM DEX to occupy its single site. In parallel tubes, 100-fold excess nonradioactive RU 38486 was added as well. Results are displayed in Fig. 2 and show that all specific CVZ-binding at both the high affinity and low affinity sites was
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 November 2015. at 01:02 For personal use only. No other uses without permission. . All rights reserved.
TWO GLUCOCORTICOID RECEPTOR AGONIST SITES
1
100
3.0
1773
•—•TOTAL O—ONONSPeanc
C33H SUreld UnoempaUd E3 3 H Steroid + 1H Steroid E3 3 H Steroid + 'H RU 38486
A—Asnanc 75
50
25
20nM DEX
10
20
30
40
50
60
[ 3 H]C0RTIVAZ0L(nM)
20nMCVZ
.5nMCVZ
FiG. 2. RU 38486 competition for CVZs high and low affinity binding sites. CEM C7 cytosol was incubated with 20 nM 3H DEX, 20 nM 3HCVZ (both sites), or 0.5 nM 3H-CVZ (high affinity site). 3H-Steroid bound was considered 100% and shown as clear bars. The 100% values, left to right, were 424, 456, and 46 fmol/mg protein. H indicates each ^-steroid bound in the presence of 100-fold excess of its respective nonradioactive steroid. M indicates amount of 3H-steroid bound in the presence of excess RU 38486. Means for 5 different incubations are shown and standard deviation is given.
blocked by the addition of unlabeled RU 38486. RU 38486 also blocked binding of DEX to its single site. Thus, it appears that the high affinity CVZ-binding site, as is its lower affinity site, is localized to proteins that also bind RU 38486, a property of the well characterized hGR. 0.1
-2.250 -2.250
-1.500
OJ 03 0.4 BOUND (pmol/mg PROTEIN)
-0.750
0.000
0.750
1.500
0.5
2.250
3
LOG1O [ H]CORTIVAZOL(nM) 3
FIG. 1. Binding of H-CVZ to IM-9 cytosol. Upper panel, Aliquots of cytosol were incubated with increasing concentrations of 3H-CVZ (0.03 nM-52 nM) in the absence (total binding) or presence (nonspecific) of excess nonradioactive CVZ. Specific binding (A • ) was determined by subtracting nonspecific binding (O O) from total binding (• • ) Middle panel, Scatchard analysis of binding of [3H]CVZ to IM-9 cytosol. Data from two separate experiments are shown to better define the curvilinear nature, but the results from this combination of data do not significantly differ from the Kds or site molarities obtained in the experiments separately. The straight lines were derived by the method of Rosenthal (32). Kds obtained from the reciprocal slopes, and site molarities (N) calculated from the x intercepts. Values were as follows: High affinity Kd = 5 X 10"10 M, n = 0.025 pmol/mg; low affinity Kd = 1.2 x 10~8 M, n = 0.425 pmol/mg. Lower panel, Hill plot of the
Use of anti-hGR antibodies to resolve CVZ'-binding sites
To determine if CVZ at its high affinity site was binding to the GR, we used antibodies (Ab) raised against the hGR or a synthetic fragment thereof. Because CEM C7 cytosol contains two classes of CVZ-binding sites, we used a high (10 nM) and a low (0.5 nM) concentration of 3 H-CVZ to label all the binding proteins responsible for the total CVZ sites or predominantly those providing the high affinity sites, respectively. Immunoprecipitation was carried out as described in Materials and Methods, and the S. aureus-Ab-GR-3H-CVZ pellet counted for the presence of labeled GR. At 10 nM CVZ, both CVZ sites should have been occupied, and, not surprisingly, CVZlabeled GR was found in the precipitate. However, because the concentration of the low affinity site is much greater than the high affinity site, we could not distinguish whether the high affinity site was precipitated or not. To solve this, we used a concentration of 0.5 nM CVZ at which, according to simple Michaelis-Menton kinetics, 90% of the CVZ specifically bound would be at the high affinity site and 10% at the low affinity site. A concentrated solution of cytosol was used (protein = 10 mg/ml) to provide sufficient high affinity sites for labeling. Under these conditions, the anti-GR antibodies were binding data. Data from the experiment in Fig. 1A are shown and reveal a Hill coefficient of 1.0; inclusion of the additional data points from a second experiment used in Fig. IB does not alter this result.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 27 November 2015. at 01:02 For personal use only. No other uses without permission. . All rights reserved.
TWO GLUCOCORTICOID RECEPTOR AGONIST SITES
1774
TABLE 1. Immunoprecipitation of 3H CVZ-labeled cytosol with antiGR antibodies Steroid + cytosol 10 nM CVZ + CEM C7 0.5 nM CVZ + CEM C7 2 nM CVZ + ICR-27 10 nM CVZ + buffer (control)
Specific [3H]CVZ Specific [3H]CVZ bound immunoprecipitated (fmol/mg) (fmol/mg) 125.2 ± 4.3 20.5 ± 1.2 21.5 ± 0.9 —
40.0 ± 6.4 ± 6.2 ± 0.7 ±
2.3 0.3 0.4 0.1
CEM C7 cytosol was incubated with 10 nM CVZ and 0.5 nM CVZ (to predominantly label the high affinity site); ICR-27 cytosol was incubated with 2 nM CVZ to label its single site. Separate fractions were competed with 100-fold excess unlabeled CVZ. Specific binding activity of the cytosol preparation used for each condition was determined in parallel incubations and is shown in the middle column. AntiGR antiserum was added to all fractions overnight, and immune complexes precipitated with S. aureus as described in Materials and Methods. Radioactivity in competed precipitates were subtracted from that in uncompeted precipitates to obtain specific amounts of 3H-CVZ precipitated. A control precipitate of buffer + 3H CVZ + Ab + S. aureus yielded similar counts relative to the competed fractions and is shown above. The percent of bound CVZ immunoprecipitated was calculated by dividing the second column by the last column. Results are means from duplicate samples with ranges shown.
still able to precipitate the protein to which CVZ was bound (Table 1). Thus, CVZ's high affinity site is indeed on a protein recognized by the antiserum prepared against a specific synthetic peptide from the immunogenic domain of the GR. To confirm this further, cytosol from ICR-27 cells was used. This cell line, which was selected from CEM C7 cells for resistance to DEX, has lost the ability to bind DEX in standard assays, yet still expresses normal basal amounts of GR messenger RNA and ~50% GR protein relative to CEM C7 (22). CVZ binds to a single class of sites in these cells with a high affinity, and based on physical properties, this site is believed to be on a protein very similar to the GR (19). Thus, we used 3H-CVZ (2 nM) to label the CVZ binding protein and then precipitated the hGR with anti-GR antisera. Again, we were able to precipitate 3H CVZ, indicating that the single site in ICR-27's was indeed on a protein immunologically indistinguishable from the traditional GR. To use a slightly different approach for resolving whether all of CVZ's binding sites are on the GR or not, we first immunoprecipitated the unlabeled hGR with a monoclonal antibody to holo hGR, then tested the remaining hGR-free cytosol for CVZ and DEX binding activity. Results in Table 2 show that at both low and high concentrations of CVZ and DEX, there was no residual binding in the supernatant cytosol over the immunoprecipitate. The possibilities of alterations in the cytosol secondary to S. aureus or other extrinsic factors were excluded by appropriate parallel controls in which all immunoprecipitation steps were performed without the presence of Ab. The control supernatant cytosol
Endo • 1990 Vol 127 • No 4
TABLE 2. CVZ, DEX, and aldosterone binding in GR-free CEM C7 cytosol Cytosol treatment
Ligand tested for binding
No immune reagents +S. aureus
20 nM CVZ 20 nM CVZ 20 nM aldosterone +Anti-GR + S. aureus 20 nM aldosterone 20 nM CVZ 0.5 nM CVZ 20 nM DEX
Spec. 3H ligand bound (pmol/mg) 0.35 ± 0.35 ± 0.26 ± 0.25 ± 0.00 ± 0.00 ± 0.00 ±
0.3 0.2 0.2 0.1 0.0 0.0 0.0
Unlabeled CEM C7 cytosol was immunoprecipitated using hybridoma-conditioned medium containing a monoclonal anti-GR antibody and S. aureus, as described in Materials and Methods. The supernatant cytosol was tested for binding activity with CVZ, DEX, and aldosterone. Controls included cytosol carried through all incubations but without treatment with immune reagents and cytosol treated with S. aureus only. Results are averages from duplicate samples with ranges given.
retained CVZ-binding after addition of S. aureus compared with cytosol untreated with
