Steroid Binding Activity Is Retained in a 16-kDa Fragment of the Steroid Binding Domain of Rat Glucocorticoid Receptors*

The steroid binding domain of the rat glucocorticoid receptor is considered as extending from amino acids 550 to 795. However, such a synthetic protein (Le. amino acids 547-795; M, -31,000) has been reported to show very little affinity for the potent synthetic glucocorticoid dexamethasone. We now disclose that digestion of steroid-free rat glucocorticoid receptors with low concentrations of trypsin yields a single species, of M, = 16,000, that is specifically labeled by dexamethasone 2 1-mesylate. This 16-kDa fragment retains high affinity binding for [3H]dexamethasone that is only -23-fold lower than that seen with the intact 98-kDa receptor. Analysis of the protease digestion patterns obtained both with trypsin and with ly- sylendopeptidase C allowed us to deduce the proteolytic cleavage maps of the receptor with these enzymes. From these protease maps, the sequence of the 16-kDa fragment was identified as being threonine 537 to arginine 673. These results show that glucocorticoid receptor fragments smaller than 34 kDa do bind ste- roids and that the amino acids Thr637-Args73 constitute a core sequence for ligand binding within the larger steroid binding domain. The much slower kinetics in generating the 16-kDa fragment from affinity-labeled receptors suggests that steroid binding causes a con- formation change in the receptor near the cleavage sites,

The steroid binding domain of the rat glucocorticoid receptor is considered as extending from amino acids 550 to 795. However, such a synthetic protein (Le. amino acids 547-795; M, -31,000) has been reported to show very little affinity for the potent synthetic glucocorticoid dexamethasone. We now disclose that digestion of steroid-free rat glucocorticoid receptors with low concentrations of trypsin yields a single species, of M, = 16,000, that is specifically labeled by dexamethasone 2 1-mesylate. This 16-kDa fragment retains high affinity binding for [3H]dexamethasone that is only -23-fold lower than that seen with the intact 98-kDa receptor. Analysis of the protease digestion patterns obtained both with trypsin and with lysylendopeptidase C allowed us to deduce the proteolytic cleavage maps of the receptor with these enzymes. From these protease maps, the sequence of the 16-kDa fragment was identified as being threonine 537 to arginine 673. These results show that glucocorticoid receptor fragments smaller than 34 kDa do bind steroids and that the amino acids Thr637-Args73 constitute a core sequence for ligand binding within the larger steroid binding domain. The much slower kinetics in generating the 16-kDa fragment from affinity-labeled receptors suggests that steroid binding causes a conformation change in the receptor near the cleavage sites, The steroid binding domain of glucocorticoid receptors has been found by molecular biology techniques to comprise the carboxyl-terminal -250 amino acids of the protein (1-3). Nevertheless, the receptor fragment spanning this region ( i e . residues 547-795) has a very low affinity for the synthetic glucocorticoid dexamethasone (9a-fluoro-ll~,l7,21-trihydroxy-16a-methyIpregna-1,4-diene-3,20-dione); the K, of this fragment for dexamethasone is -350 times less than that of the intact receptor (3). The shortest reported segment possessing high affinity for dexamethasone is the 34-kDa fragment of 497-795, which binds dexamethasone with a KO that is five times less than that of the 98-kDa receptor (3). Units smaller than 547-795 (3), including the mero-receptor (4), have no observable affinity for dexamethasone. Furthermore, deletion of the carboxyl-terminal 29 amino acids appears to reduce the K, for dexamethasone by a factor of ~1 0 0 0 (3). All portions of the steroid binding domain appear to be necessary, since deletions and point mutations throughout this region * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
B2A-07, NIDDK/LAC, NIH, Bethesda, MD 20892. 1: To whom reprint requests should he addressed: Bldg. 8,Rm. have been found to eliminate steroid binding (1-3, 5 ) . The identification of methionine 622, cysteine 656, and cysteine 754 by electrophilic (6) or photoaffinity (7) labeling of the rat receptor afforded additional evidence for the involvement of this large region in steroid binding. However, in spite of these data, it seems unlikely that all of the carboxyl-terminal -250 amino acids are intimately involved in the construction of the steroid binding cavity, since other activities are also encoded in this domain. Thus, sequences involved in transcriptional activation (8,9) and the association of hsp90 (10) plus regions of homology with the other steroid receptors (10-12) are also found in this region. As part of the current efforts to localize the various activities in the carboxyl-terminal third of the receptor, we have used proteolytic enzymes as a means of peeling away the outer layers of the receptor to expose the core unit of the steroid binding domain. We now report the preparation of such a core unit that contains only half of the canonical steroid binding domain. Additional data argue that this region is defined by the sequence of threonine 537 to arginine 673 in rat receptors.

Steroid Binding Activity
Cells and Preparation and Labeling of Receptors-The growth of HTC cells in spinner and monolayer cultures of S77 medium supplemented with 5% fetal and 5% newborn bovine serum (Biofluids) and 0.03% glutamine has been describa (15). HTC cell cytosol containing the steroid-free receptors was prepared, stored in liquid N2, and labeled a t 0 "C with ['H]dexamethasone or ['H]dexamethasone 21mesylate & excess ['Hldexamethasone ? 20 mM Na2MoOI (14, 15). Free steroid was removed by adding a 10% dextran-coated charcoal suspension in pH 8.8 TAPS buffer (added volume = 20% of reaction volume for intact receptors; for 16-kDa fragments, added volume = 30% of reaction volume in order to reduce nonspecific binding). Nonspecific binding/labeling equaled that seen with excess ['HI dexamethasone. Covalent ["H]dexamethasone 21-mesylate-labeled receptors were quantitated on SDS-polyacrylamide gels (14, 15).
Polyacrylamide Gel Electrophoresis-The preparation of samples for gels and the procedures for electrophoresis are as described (17). Constant percentage acrylamide gels (10.5-11 or 15% with a 1:40 ratio of bisacrylamide to acrylamide) were run in a water-cooled (15 "C) Protean I1 slab gel apparatus (Bio-Rad) at 30 mA/gel (25 mA/gel for 15% gels). Gels were fixed and stained in 50% methanol, 7.5% acetic acid containing 0.01% Coomassie Blue R-250, destained in 10% methanol, 7.5% acetic acid, incubated for 1 h in Enhance (Du Pont-New England Nuclear) and 30-60 min in 10% Carbowax PEG 8000 (formerly PEG 6000; Fisher) with constant shaking at room temperature, dried on a Bio-Rad model 443 slab gel drier at 60 "C with a sheet of dialysis membrane backing (Bio-Rad) directly over the gel to prevent cracking, and fluorographed for 7-12 days with Kodak X-Omat XAR-5 film.

RESULTS AND DISCUSSION
Trypsin digestion of glucocorticoid receptors that were affinity labeled with ["Hldexamethasone 21-mesylate (18) at cysteine 656 (6) gives the normal pattern of labeled fragments on SDS gels (14,17,19), which includes the relatively trypsinresistant mero-receptors at 30.4 and 28.3 (obscured) kDa that yield the 16,100 f 150 (S.D., n = 7)-Da fragment at higher trypsin concentrations (Fig. U). Trypsin digestion of steroidfree receptors occurred more rapidly and demonstrated that mero-receptors can bind steroid and be covalently labeled (Fig. 1B). However, the predominant steroid-free fragment that was formed, even at low trypsin concentrations, was not mero-receptor but a species of M , = 16,000 f 200 (n = 10) that appears to be the same as the 16.1-kDa species of Fig.  1A. It should be noted that the 16-kDa fragment was the only specifically labeled material that was generated from steroidfree receptors with 14-28 pg/ml of trypsin (Fig. 1B). Digestions of steroid-free receptor were usually conducted in the presence of 20 mM NaPMo04, which stabilizes the 16-kDa species, and were stopped by the addition of a 10-fold excess (w/w) of soybean trypsin inhibitor (STI), which has no effect on ['H]dexamethasone binding or ['HJdexamethasone 21mesylate labeling of the 98 kDa receptor or any of its trypsin digest fragments (data not shown).
The 16-kDa species formed in trypsin digests of crude, steroid-free receptors was identified as being a fragment of the rat glucocorticoid receptor as follows. First, the covalent labeling by ['H]dexamethasone 21-mesylate is totally inhibited by a 100-fold excess of nonradioactive dexamethasone (Fig. 1). Second, under conditions where only the 16-kDa species is generated and specifically labeled by ['HJdexamethasone 21-mesylate, specific binding of ['HJdexarnethasone with high affinity is observed (Fig. 2)

Steroid Binding
Activity 14495 indicate that the dexamethasone binding of a trypsin-generated fragment of the rat glucocorticoid receptor is to the same 16-kDa species that is affinity labeled by dexamethasone 21mesylate. Scatchard analysis indicates that the amount of 16-kDa receptor fragment in the trypsin digest solutions is 49 f 18% (S.D., n = 4) of the initial 98-kDa receptors ( Fig. 2 and data not shown). This less than quantitative yield and the position in the receptor protein of the 16-kDa fragment both follow from the trypsin cleavage map of Fig. 3. This map was deduced  17) with the species predicted from the primary sequence (21). This map is supported by the report that glycine 518 is the amino terminus of the 27-kDa mero-receptor prepared from partially purified receptors (22). The observation of an -50% yield of the [3H]dexamethasone 21-mesylatelabeled, 1,600-Da trypsin limit digest fragment (amino acids 652-673) (17) indicates that digestion of the 18.4-kDa fragment is equally divided between the two paths of Fig. 3. The lack of fragments between 16 and 1.6 kDa (17) argues that the 16.1-kDa fragment is digested directly to the sequence of 652-673. Thus, a 50% yield of the 16.1-kDa fragment is the expected result.
The cleavage map of Fig. 3 was further supported by the fragments obtained from digestion of [3H]dexamethasone 21mesylate-labeled receptors with lysylendopeptidase C, which cleaves only after lysine and can cut at just four positions between amino acids 518 and 684. Thus, the predicted fragments a t -30.7, 28.2, and 17.2 kDa plus new species at 17.8 f 0.3 kDa (S.D., n = 4; corresponding to 500-662) and 14.2 k 0.6 kDa (S.D., n = 5; corresponding to 537-662) were observed; the trypsin digest fragments at 18.4 and 16.1 kDa, resulting from cleavage at an arginine residue, were missing (Figs. 3 and 4 and data not shown). Furthermore, lysylendopeptidase C limit digests are predicted to yield only two fragments of less than 10 kDa, i.e. 595-662 (= 8009 Da) and 598-662 (= 7624 Da), which are not resolved on our SDS gels (17). We observe just one lysylendopeptidase C limit digest band at 5.4 f 0.6 kDa (S.D., n = 10) with no decrease in recovered 3H-labeled disintegrations/min after 24 uersus 1 h of digestion at 20 "C (data not shown). This is lower than the predicted M,, as has usually been observed for [3H]dexamethasone 21-mesylate-labeled receptor fragments in this molecular weight range (6,17). In conclusion, the cleavage map of Fig. 3 was the only map that was internally consistent with the digestion patterns of affinity-labeled receptors that were obtained both with trypsin and with lysylendopeptidase C.
Using these protease digestion maps, we determined that trypsin digestion of both steroid-free and affinity-labeled receptors yields the same fragment of Thr537-Arg673 since the 16-kDa species derived from steroid-free receptors co-migrates with the major band seen in this region after trypsin digestion of partially purified ['H]dexamethasone 21-mesylate-labeled receptors ( Fig. 4 and Ref. 17). This co-migration occurs for separately applied (Fig. 4, lanes 5 versus 6 and 7) or mixed (lanes 10 and 11 ) solutions of receptor fragments. This conclusion is supported by the observation that ['HI dexamethasone 21-mesylate labeling of steroid-free receptors digested previously with different concentrations of lysylendopeptidase C yields no specifically labeled fragments of molecular weight below 28 kDa (data not shown). Since lysylendopeptidase C digestion of ['H]dexamethasone 21-mesylate prelabeled receptors does afford several species in the range of 17.8-14.1 kDa, but not the 16.1-kDa fragment generated by trypsin, we conclude that the 16.1-kDa fragment must have arginine as one of the trypsin cleavage sites and that the other potential fragments lack a sequence(s) critical for steroid binding. Collectively, these data argue that trypsin cleavage of steroid-free receptors a t lysine 536 and arginine 673 generates a 16.1-kDa fragment that binds dexamethasone and is labeled by dexamethasone 21-mesylate. To the best of our knowledge, this 16-kDa core unit is the smallest portion of the glucocorticoid receptor that still binds dexamethasone.
This assignment is consistent with the observation that molybdate stabilization of the 9 S form of glucocorticoid receptor-steroid complexes, and probably steroid binding to receptors, involves the region of 550-715 and most likely the sequences of 595-614 (10). Most, or all, of these amino acids lie within our 16-kDa fragment, to which the binding of ['HI dexamethasone is greatly stabilized by the presence of molybdate (see above).
The inhibition of trypsin cleavage of the 30/28-kDa meroreceptor to the 16-kDa fragment by bound steroid (Fig. 1) suggests a steroid-induced conformational change in the receptor. We cannot yet exclude the possibility that the covalent labeling of cysteine 656 sterically prevents the required cleavage a t both lysine 536 and arginine 673 to give the 16-kDa fragment (Fig. 3). Restricted cleavage at both sites must occur, because no fragments corresponding to proteolysis a t just one site (at lysine 536 to give a 25.9-kDa species or at arginine 673 to give an 18.4-kDa species) were observed ( Fig.  1 and Ref. 14) unless the affinity-labeled receptors were first denatured (data not shown and Ref. 17). Although inhibition of cleavage a t arginine 673 might be expected, due to the possible proximity of the covalently labeled cysteine 656, simultaneous steric hinderance of cleavage at the more remote lysine 536 by the covalently bound steroid is less palatable. We therefore believe that this inhibition is due to a conformational change which renders the steroid-bound receptor less susceptible to proteolysis with trypsin. A conformational change following ligand binding has been proposed previously for glucocorticoid receptors (4) and has been observed for estrogen (23)(24)(25) and CAMP (26) receptors. The more rapid digestion of steroid-free receptors may also explain why previous preparations of steroid-free mero-receptors did not bind dexamethasone (4) i.e. steroid-free receptors are further digested to nonbinding fragments at the higher trypsin concentrations needed to produce mero-receptors from pre-bound receptors (cf. 56 pg/ml trypsin in Fig. 1). We do not see any difference in the chymotrypsin digestion of steroid-free versus steroid-labeled receptors (data not shown).
Previous attempts to delinate the minimum glucocorticoid receptor sequence for steroid binding by construction of receptor point mutants and deletion mutants have been thwarted by the apparent absolute requirement of numerous amino acids throughout the carboxyl-terminal 250 amino acids (1-3, 5). Furthermore, this approach cannot discriminate between direct effects of mutations (or deletions) on steroid binding and indirect effects on the tertiary structure. Proteolytic enzymes offer an attractive complementary approach, since only those portions of the receptor that are exposed to solvent and accessible to macromolecules could be attacked and the hydrophobic core should remain undisturbed. Using this approach, we have now found a steroid binding core, located a t 537-673, which is about half the size of the steroid binding domain but which possesses a binding affinity for dexamethasone that is 215 times that of similarly located The sizes of the 98-, 42-, and 16-kDa receptor species were determined from SDS-polyacrylamide gels. The sizes of the other species were calculated from their amino acid sequence. The sequence of the 42-kDa species is based on the fact that cleavage of the 98-kDa receptor by chymotrypsin to give the 42-kDa species is reported to occur at positions 409 and 413 (22)