Molecular Identification and Characterization of a and B Forms of the Glucocorticoid Receptor

Molecular Identification and Characterization of A and B Forms of the Glucocorticoid Receptor

Matthew R. Yudt and John A. Cidlowski

Laboratory of Signal Transduction National Institute of Environmental Health Sciences Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021 National Institutes of Health Research Triangle Park, North Carolina 27709

The human glucocorticoid receptor (hGR␣)isa responsive promoters reveal the shorter hGR-B to ligand-activated transcription factor that mediates be nearly twice as effective as the longer hGR-A the physiological effects of corticosteroid hor- species in gene transactivation, but not in transre- mones and is essential for life. Originally cloned in pression. (Molecular Endocrinology 15: 1093–1103, 1986, the transcriptionally active hGR␣ was re- 2001) ported to be a single protein species of 777 amino kDa). Biochemical 94 ؍ acids (molecular mass data, obtained using various mammalian tissues INTRODUCTION and cell lines, however, have consistently revealed an additional, slightly smaller, second hGR protein The glucocorticoid receptor (GR) mediates the physi- kDa) that is not recognized ological effects of corticosteroid hormones in species 91 ؍ molecular mass) by antibodies specific for the transcriptionally in- from fish to mammals. GR is a member of the nuclear active and dominant negative, non-hormone-bind- hormone receptor superfamily of ligand-activated ing hGR␤ isoform. We report here that when a transcription factors (1) and is essential for life (2). The single GR cDNA is transfected in COS-1 cells, or activity of the GR, as well as the progesterone (PR), transcribed and translated in vitro, two forms of androgen (AR), and mineralocorticoid receptors (MR) the receptor are observed, similar to those seen in is partially mediated through a palindromic response cells that contain endogenous GR. These data sug- element termed the glucocorticoid response element gest that two forms of the hGR␣ are produced by (GRE) located in the promoter regions of target genes alternative translation of the same gene and are (3). Unlike most members of the steroid receptor su- henceforth termed GR-A and GR-B. To test this perfamily, the GR is primarily cytosolic in the absence hypothesis, we have investigated the role of an of ligand. Activation and nuclear translocation of the internal ATG codon corresponding to methionine GR after ligand binding proceeds through a complex 27 (M27) as a potential alternative translation initi- mechanism involving a loss of energy-dependent pro- ation site for the GR. Mutagenesis of this ATG tein interactions with hsp90, hsp70, and several other codon to ACG in human, rat, and mouse GR cDNA proteins (4). Although the classical view of steroid results in generation of a single 94-kDa protein action involves an increase in gene transcription in species, GR-A. Moreover, mutagenesis of the ini- response to receptor activation, in fact several glu- tial ATG codon to ACG (Met 1 to Thr) also resulted cocorticoid target genes undergo a hormone-depen- in production of single, shorter protein species (91 dent repression (5). Furthermore, in addition to GRE- kDa), GR-B. Mutagenesis of the Kozak translation dependent processes, a growing body of literature initiation sequence strongly indicates that a leaky indicates that many glucocorticoid responses involve ribosomal scanning mechanism is responsible for protein interactions with other transcription factors generating the GR-A and -B isoforms. Western blot and likely proceed through a mutual inhibitory antag- analysis using peptide-specific antibodies show onism involving direct protein-protein interactions with both the A and B receptor forms are present in other transcription factors including, for example, nu- ␬ ␬ human cell lines. Both receptors exhibit similar clear factor- B (NF- B) and AP1 (6). subcellular localization and nuclear translocation Our understanding of the complexity of nuclear re- after ligand activation. Functional analyses of ceptor signaling mechanisms has advanced signifi- -hGR؊A and hGR؊B under various glucocorticoid- cantly in recent years. The discovery and character ization of receptor coactivators and corepressors 0888-8809/01/$3.00/0 bridge the gap between the DNA-bound receptors and Molecular Endocrinology 15(7): 1093–1103 the general transcription machinery (7–9). Similarly, Copyright © 2001 by The Endocrine Society Printed in U.S.A. our knowledge regarding the role of chromatin struc-

1093 MOL ENDO · 2001 Vol. 15 No. 7 1094

ture in steroid receptor signaling has been enhanced in protein, independent of the alternatively spliced GR␤ recent years (10, 11). The three-dimensional structure variant (21, 22). The human hGR␣ and -␤ variants differ of many nuclear receptor ligand binding domains has by only 35 amino acids in length at the extreme car- not only revealed a common protein fold and ligand boxy terminus. Although quantitative measurement of binding symmetry among superfamily members, but their coexpression in human tissues or cell lines re- exposed the subtle ligand interactions and associated mains difficult because of the relative abundance of conformational changes necessary for a mechanistic hGR␣ to ␤, the two isoforms can be discriminated understanding of steroid action (reviewed in Ref. 12). immunologically using specific antibodies (23, 24). In- Furthermore, examples of ligand-independent activa- terestingly, the recombinant hGR␣ when expressed tion mechanisms in nuclear receptor signaling con- alone, either in vitro with 35S methionine or in COS-1

tinue to multiply (13). cells, known to be void of detectable endogenous GR, Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021 An additional level of complexity of steroid hormone consistently appears as a doublet of approximately receptor action is the existence of multiple receptor equal intensities (Fig. 1A). subtypes and isoforms with unique biological roles There are several possible explanations for the ori- (14–16). For example, multiple genes encode different gin of the observed hGR␣ protein doublet. Although forms of the estrogen, retinoid, and thyroid hormone proteolysis could explain the appearance of such a receptors. Alternative splicing of progesterone, glucocorticoid, and retinoid receptor mRNA gives rise to multiple forms of these proteins. The progesterone receptor (PR) exists as a mixture of A and B forms, generated from the same gene by alternative translation initiation. Although both PR isoforms can arise from a single mRNA (17), it appears that specific promoters may also regulate mRNA production specific for each PR isoform (18). Both forms of PRs are well known to display distinct biochemical and physiological properties (19). This extensive multiplicity within the nuclear receptor superfamily suggests that the diversity of receptor expression may be an important component mediating the various physiological ac- tions of steroid hormones. We report here that the GR␣ gene is subject to alternative translation initiation from a downstream, in-frame ATG codon. Our data suggest that a leaky ribosomal scanning mechanism (20) produces two GR protein products, with the second initiating at an ATG codon corresponding to methionine 27 in the hGR. We term the longer protein, initiated from the first ATG codon (Met 1) as hGR-A, and the shorter protein (751 amino acids) as hGR-B. We have constructed a GR- A-specific antibody that, when used in conjunction with an antibody that recognizes both protein species, permits the discrimination of endogenous expression of the two hGR␣ isoforms. Interestingly, the shorter hGR-B is twice as effective as the longer hGR-A iso- Fig. 1. In Vitro and in Vivo Expression of Recombinant hGR␣ form in activating transcription from a GRE but has a A, The wild-type hGR␣ was prepared either by in vitro ␬ similar efficacy in repression of NF- B/p65 transacti- translation using 35S methionine and reticulocyte lysates (left vation. This discovery of an alternative initiation site panel) or by transient transfection of COS-1 cells (right panel). within the GR gene, and the functional divergence Approximately 25 ␮g of protein were electrophoresed on an observed, provides a new potential mechanism to ex- 8% polyacrylamide gel. The 35S-labeled receptor was de- plain the diversity of glucocorticoid responses in dif- tected by autoradiography of the dried gel, while the COS- ferent tissues. 1-expressed proteins were transferred to nitrocellulose and detected by Western blotting. The positions of the molecular mass markers in kilodaltons are indicated. Electrophoresis was carried out for an extended period to resolve the protein RESULTS doublet of approximately 94 and 91 kDa. B, Wild-type hGR (1–777) and two carboxy-terminal truncation mutants, Expression of Recombinant hGR hGR(1–742) and hGR(1–706), were expressed in vitro using reticulocyte lysates. Electrophoresis was carried out as in The cloning of the hGR␣ into mammalian and in vitro panel A to resolve the hGR protein doublet. Data shown are expression vectors has allowed a direct study of this representative of at least three different experiments. GR A and B 1095

doublet, inclusion of several protease inhibitors did not block production of the lower mol wt (Mr) product, arguing against degradation as the source of the doublet. Moreover, in vitro transcription and translation of two carboxy-terminal truncation mutants, hGR(1–742) and hGR(1–706), results in a similar doublet pattern (Fig. 1B), suggesting that carboxy-terminal degradation is not the source of the second band. Another possible source of the doublet is phosphorylation; however, phosphatase treatment of the reticulocyte or

COS-1 lysate containing hGR␣ does not affect the Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021 doublet pattern (data not shown). Fig. 2. Mutation Analyses of Potential hGR Start Sites Plasmids containing the hGR␣ cDNA used for in vitro Alternative Translation Initiation of the GR translation (T7 promoter) and for in vivo transient expression [cytomegalovirus (CMV) promoter] were independently sub- To investigate the possibility of alternative translation jected to site-directed mutagenesis, changing potential initiation of the GR as a source of the observed dou- translation initiation codons (ATG) 1 and 27 individually to blet, the hGR cDNA was examined for downstream threonine codons (ACG). The wild-type and start site mutants 35 ATG start codons. Only one in-frame ATG codon, cor- were prepared as both in vitro translated, S-labeled protein, responding to methionine 27, was found within the first as well as transiently expressed in COS-1 cells, and detected as described in Fig. 1. The synthesis of hGR␣ from methio- 300 nucleotides of the initial hGR ATG translation start nine 1 (in mutant M27T) is referred to as hGR-A while the site. Translation initiation from this internal ATG site protein synthesized from methionine 27 (in mutant M1T) is would yield a protein almost 3 kDa shorter (apparent referred to as hGR-B. The 83-kDa band detected in trans- molecular mass ϭ 91 kDa) than the full-length hGR fected COS-1 cells is a common GR degradation product. from residues 1–777 (apparent molecular mass ϭ 94 Western blots shown are representative of more than five kDa). To test the hypothesis of alternative initiation as separate experiments. a source of the protein doublet observed in GR expression systems, both the initial ATG start codon proteins (Fig. 3, C and E). This is a particularly impor- (methionine 1) and the internal ATG (methionine 27) tant observation considering neither mouse nor rat were mutated to ACG (a threonine codon). Mutagen- contain the ␤ form of the GR (Bofetiado, D. M., and J. ␣ esis of the individual ATG codons in the hGR cDNA in A. Cidlowski, unpublished observations). These re- both the in vitro expression vector and the mammalian sults suggest that the human, rat, and mouse GR expression vector resulted in the expression of a sin- genes can produce two proteins via alternative initia- ␣ gle hGR species (Fig. 2). We have termed the longer tion of the same gene transcript. GR, generated from the first ATG codon, GR-A. The shorter GR species, translated from the internal ATG Detection of Endogenous hGR-A Using an hGR- corresponding to methionine 27 (amino acid 28 in rat A-Specific Antibody and mouse GR), is designated as GR-B. A proteolytic fragment common to both GR forms of approximately To evaluate the endogenous expression pattern of the 83 kDa is consistently observed at higher levels in cells GR-A and -B isoforms, an antibody was generated expressing GR-B. that is specific to the longer, hGR-A isoform (Fig. 4A). In every mammalian species in which GR has been A dual Western blot analysis was then carried out on cloned and sequenced, except the guinea pig, an in- COS-1-expressed hGR wild type (wt) and the start site ternal ATG was found 27 or 28 codons from the initial mutants, hGR-A (M27T) and hGR-B (M1T), using both ATG (Fig. 3A). Interestingly, the guinea pig has been the hGR-A-specific antibody and Ab57, a polyclonal shown to be relatively glucocorticoid resistant in com- epitope-purified antibody generated against a peptide parison to other mammals (25). In addition, neither the sequence common to both GR-A and GR-B (26). As African frog nor rainbow trout GR contain a second seen in Fig. 4B, the hGR-A-specific antibody recog- potential translational start site near this position. nizes only the longer hGR-A species, while the Ab57 Since doublets are detected in both mouse (mGR) and recognizes both isoforms. A similar immunoreactive rat GR (rGR) expression systems, a similar start site evaluation was carried out using various human cell mutagenesis analysis was carried out on these recep- culture lines (HeLa, CEM-C7, and HEK-293) known to tor species. As observed for the hGR, mutagenesis of express endogenous hGR␣. As observed with the re- the potential start sites of the mGR and rGR resulted in combinant protein, the Ab57 detects the GR doublet expression of single protein species in vivo and in vitro while the hGR-A-specific antibody detects only the top (Fig. 3, B and D). To compare the native GR with the band (Fig. 4C). The 83-kDa band detected in HeLa cell recombinant forms, mouse and rat liver samples were extracts is a common GR degradation product ob- analyzed for GR expression. Both mouse and rat liver served in this cell line. These data suggest both hGR-A GR are expressed as doublets, directly corresponding and hGR-B are endogenously expressed in several to the A and B forms observed with the recombinant human cell lines. MOL ENDO · 2001 Vol. 15 No. 7 1096 Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021

Fig. 3. Species Comparison of GR Translation Initiation Sites A, Sequence alignments of residues 1–40 of GRs from various species. The indicated sequences were compiled from NCBI databases (SWISS-PROT accession numbers: human, P04150; squirrel monkey P79686; rat, P06536; mouse, P06537; xenopus, P49844; rainbow trout, P49843). A dot indicates a conserved residue, while a dash indicates a single residue gap. B, The mouse GR expression vector (pCMV-mGR) was mutated as in Fig. 2 for the hGR, changing potential translation initiation codons (ATG) 1 and 28 individually to threonine codons (ACG) to yield M1T and M28T, respectively. Protein expression in transfected COS-1 cells was measured by Western blot with Ab57. A similar band pattern was observed with in vitro translated wt and mutant mGR (not shown). C, Fifty micrograms of mouse liver protein extract were subjected to Western blot analysis with Ab57. The two major immunoreactive bands at 94 and 91 kDa correspond with the mGR-A and -B bands detected in panel B. D, The rat GR expression vector (pSG5-rGR) was also mutated at the potential rGR start sites (Met1 and Met 28) and subjected to in vitro expression using reticulocyte lysates and 35S-methionine followed by autoradiography. An identical pattern is observed when the same plasmids are expressed in COS-1 cells (not shown). E, Fifty micrograms of rat liver protein were analyzed for GR content by Western blotting with Ab57. The two major immunoreactive bands at 94 and 91 kDa correspond directly with the rGR-A and -B isoforms detected in panel D. All Western blots are representative of at least three separate experiments.

Unlike the majority of steroid hormone receptors, found in tandem (GRE2-luc), the maximal transactiva- the GR is primarily localized in the cytoplasm in the tion activity of hGR-B is enhanced to nearly 2-fold that absence of ligand (27). However, in response to hor- of hGR-A (Fig. 6B). Finally, this functional difference is mone signal, the functional GR undergoes nuclear also observed using a mouse mammary tumor virus translocation. To examine potential functional differ- (MMTV) promoter reporter gene, where the GR-B ences between the GR-A and -B isoforms, immuno- transactivation is at least 1.4-fold greater than in cytochemistry was performed to evaluate the nuclear hGR-A (Fig. 6C). These data suggest that GR-B, lack- and cytoplasmic localization of the two proteins in the ing the first 27 residues of GR-A, exhibits enhanced absence and presence of hormone. Using this transcriptional activity in a variety of promoter con- method, no difference is observed between the hGR-A texts and argue for a general mechanism, not strictly (M27T) and hGR-B (M1T) forms in nuclear transloca- dependent on promoter sequences. Although the tion in response to dexamethasone (Dex) (Fig. 5). EC50 values vary depending on the promoter used, none of our transactivation assays are sensitive Gene Activation by hGR-A and -B enough to detect a difference between GR-A and To compare the cellular functions of hGR-A and GR-B within 5 nM. hGR-B, transactivation studies were carried out on To further address the functional differences in both receptor forms using transient transfections of transactivation, experiments were carried out in which hGR-A and hGR-B in COS-1 cells. Utilizing three GRE- the expression vector concentrations were varied and driven reporter genes, a striking difference in transac- hormone levels were kept constant. When the GR-A tivation was observed between the two hGR␣ iso- and B levels were varied under saturating concentra- forms. As shown in Fig. 6A, the hGR-B form is more tion of Dex (100 nM), hGR-B was approximately twice than 1.5-fold as effective in transactivation from a as effective as the hGR-A in transactivating the GRE2- single GRE-driven reporter gene (GRE1-CAT) than the luc reporter (Fig. 6D). Similar results are observed longer hGR-A form (Fig. 6A). When two GREs are when using a luciferase reporter construct containing GR A and B 1097 Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021

Fig. 5. Immunocytochemical Analysis of hGR-A and hGR-B The nuclear translocation of the GR-A and -B isoforms in response to Dex was compared by immunocytochemistry. COS-1 cells were transfected with either hGR-A (top panels) or hGR-B (bottom panels) and plated on chamber slides the following day. Approximately 48 h after transfection, cells in chamber slides were treated for 2 h with 100 mM Dex (ϩ)or Fig. 4. Western Analysis of hGR-A and hGR-B vehicle (Ϫ) as indicated. Cells were fixed on slides and ana- A, hGR-A-specific antibody production. A polyclonal anti- lyzed by immunostaining as previously described (22). Both body was prepared from rabbits, using a synthetic peptide the Ab57 and GR-A-specific antibodies were used in these antigen corresponding to residues 3–22 of the hGR. The experiments. Data shown are representative of immuno- location of the Ab57 binding site (residues 346–357 of the stained cells from one experiment, which was reproduced mouse GR), which is present in both hGR-A and hGR-B, is three separate times. shown for comparison. B, The hGR-A-specific antibody recognizes the hGR initiated from the methionine 1 codon but not the GR-B form initiated from methionine 27 codon. COS-1 cells were transfected with the wt hGR␣ and the two the MMTV promoter. To test whether the functional start site mutants described in Fig. 2. As observed in Fig. 2, differences can be attributed to differences in protein Ab57 detects both forms of the hGR as illustrated by Western expression of hGR-A and hGR-B, equivalent amounts blotting (left panel). In contrast, the hGR-A specific antibody of the expression vectors were transfected in separate only detects the longer, hGR-A form (right panel). C, Endog- wells and subjected to Western blot analysis. Despite enous hGR production is a mixture of the A and B GR iso- the higher transactivation capacity observed for forms. Endogenously produced hGR from several cell lines hGR-B, Fig. 6E shows that transfection of equivalent (HeLa, HEK293, CEM-C7) were analyzed for hGR-A and -B amounts of expression vector consistently results in a production by Western blotting with the GR-A-specific anti- greater accumulation of expressed hGR-A protein. body. Approximately 50 ␮g of total protein extract from the These data suggest that the observed transactivation human cell lines indicated, were probed with the Ab57 (left panel). The same samples were probed with the GR-A- differences may be underestimates of the transcrip- specific antibody, in which case only the longer, GR-A, form tional potential of hGR-B when considering the actual was detected. The 83-kDa band detected in HeLa extracts is levels of expressed protein. Interestingly, the level of a commonly observed GR degradation product. Blots shown the 83-kDa degradation product is also significantly are representative of three or more experiments. higher when hGR-B vectors are used for receptor ex- MOL ENDO · 2001 Vol. 15 No. 7 1098 Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021

Fig. 6. Functional Comparison of hGR-A and hGR-B The transactivation function of hGR-A and -B in response to Dex was measured using three glucocorticoid responsive reporter genes containing single or multiple copies of GRE-binding sites. A schematic of each reporter gene is shown at the top of the data set in which it was used. A, COS-1 cells were transfected with a constant amount of expression vector (50 ng) for either hGR-A or hGR-B as well as a constant amount of the GRE1-CAT reporter (0.5 ␮g). After transfection, cells were treated with increasing concentrations of Dex ranging from 10 pM to 100 nM and harvested approximately 16–20 h later for CAT activity assay. Data shown are an average of three separate experiments in triplicate with the indicated SEM. B, Cells were transfected as in panel A, but with a constant amount of the GRE2-luc reporter (0.5 ␮g). Cells were treated with Dex as indicated and harvested approximately 16–20 h later for luciferase activity assay. Data shown are an average of five separate experiments with the indicated SEM. C, COS-1 cells transfected as in panels A and B, but with a constant amount of the MMTV-CAT reporter (0.5 ␮g). Cells were treated and harvested for CAT assay as in panel A. Data shown are an average of triplicate samples from a representative of three separate experiments, with the indicated SEM. D, To contrast the hormone titration experiments, cells were transfected with increasing amounts of either hGR-A or hGR-B together with the GRE2-luc reporter and treated with a constant amount of Dex (100 ␮M). Cells were harvested and analyzed for luciferase production as in panel B. Data shown are an average of two individual experiments with the indicated SEM. E, To measure relative expression levels of hGR-A and hGR-B, COS-1 cells were transfected with 2.5 ␮g of wt hGR as a control (lane 1), and hGR-A (lanes 2–4), or hGR-B (lanes 5–7) in 10-cm plates for 20 h. Cells were harvested in ice-cold RIPA buffer containing 5 mM DTT and protease inhibitors and immediately subjected to SDS-PAGE and Western blot analysis with Ab57. The approximate molecular masses of detected bands are shown to the right. The 83-kDa band, commonly observed in both transient and endogenous receptor expression systems, is likely a product of degradation. GR A and B 1099

pression, suggesting that hGR-B may be more sus- functions are contained within separate regions of the ceptible to proteolysis than is hGR-A (Fig. 6E). How- protein and suggest a role for the first 27 residues in ever, since we do not know the activity of the 83-kDa transactivation but not transrepression of NF-␬B. band, we cannot eliminate the possibility that it does contribute to the transactivation levels observed in Fig. Mechanism of the Alternative Translation 6, A–D. Initiation of GR

Gene Repression by hGR-A and -B The cause of alternative initiation of the GR transcript may lie within the sequence itself. Eukaryotic ribo- A growing body of data suggests that many GR- somes appear to select the start site for translation mediated effects occur independently of direct DNA initiation by a scanning mechanism (reviewed in Ref. Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021 (GRE) binding (28). These processes occur via cross- 30). An AUG start codon is classified as strong or weak talk with other signaling pathways and through protein depending on the adherence to a specific surrounding interactions independent of a GRE. One well studied consensus sequence (Kozak sequence). A purine at cross-talk pathway is the mutual repression observed position Ϫ3andaGatposition ϩ4 relative to the AUG between hormone-activated GR and the transcription (the A is considered as ϩ1) are considered strong factor NF-␬B (29). The ability of hGR-A and hGR-B to initiator sequences. However, leaky scanning, in which repress transactivation of the NF-␬B p65 subunit was the first AUG is bypassed in favor of a nearby down- evaluated. Interestingly, both receptor isoforms ap- stream AUG, is most predictably caused by deviations peared to antagonize p65 reporter activity to the same from the strong sequence context within the first AUG degree (Fig. 7) in contrast to the observed difference in (31). The sequence contexts of both AUG start sites of GRE-dependent transactivation. These data support human, mouse, and rat GR are shown in Fig. 8A. For the hypothesis that hGR activation and repression all three species (human, rat, and mouse), the first AUG does not contain a purine at position Ϫ3, indicating it is a weak initiator sequence. However, the second AUG in all three species of GR does contain a purine at this position, indicating it is actually a stron- ger translation initiation site, and could explain the high degree of leaky ribosomal scanning and the production of two GR proteins from the same mRNA. To test the hypothesis that leaky ribosomal scanning is the cause for the production of GR A and B isoforms, we mutated the Kozak sequence of the first AUG start site. When the “weak” consensus site of the first AUG (Met 1) is changed to a “strong” one by a point mutation at the Ϫ3 position (C to G), production of GR-B is completely lost (Fig. 8B). These data con- clusively demonstrate that leaky ribosomal scanning of a weak initiation sequence generates a second, B form, of the GR.

DISCUSSION

The data presented in this paper demonstrate that the GR is a product of alternative translation initiation, which results in the production of both a GR-A and Fig. 7. Inhibition of NF-␬B/p65 by GR-A and GR-B GR-B form. Both the A and B forms are generated in The ability of hGR-A and hGR-B to repress the transacti- approximately equivalent levels from a single cDNA vation function of the p65 subunit of NF-␬B was measured. A when expressed in vitro using reticulocyte lysates. constant amount of p65 expression vector (12.5 ng) and a However, the GR-B form appears to be more suscep- NF-␬B luciferase reporter construct (3XMHC-luc) were co- tible to degradation, at least when expressed in mam- transfected in COS-1 cells along with increasing amounts malian cells. Translation of an hGR␣ message from an (10, 50, and 250 ng) of expression vector for either hGR-A (A) internal AUG codon, corresponding to methionine 27, or hGR-B (B). In the presence of 100 nM Dex (ϩ), both hGR-A and hGR-B repressed p65 transactivation with approximately results in a protein with twice the amount of maximal the same efficiency over the entire range of hGR expression. transactivation as the longer protein initiated from the The x-axis labels shown are identical for both panels A and B. first AUG codon. Interestingly, deletion of the first Data shown are an average of three separate experiments 25–30 residues of hPR-B also results in more effective with the indicated SEM. transcriptional activation (Horwitz, K., and L. Tung, MOL ENDO · 2001 Vol. 15 No. 7 1100

The first 27 residues of hGR, however, do not appear to have significant homology with recognized protein- protein interaction sites or other known functions. The fact that the GR-B is a more effective transactivator on three separate glucocorticoid-responsive promoters argues for a general mechanism involving GR interactions with additional cellular factors. It remains to be seen whether both isoforms homo- and heterodimer- ize equivalently, bind DNA with the same affinities, and respond to different hormone signals with the same

relative potency. In addition, the tissue distribution of Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021 the two isoforms remains to be elucidated. It is now established that other steroid receptors produce N-terminal truncation variants. For example, it is well known that two PRs are derived from the same gene in humans and chickens (32). It was originally suggested that the origin of the PR isoforms was alternative translation initiation from the same message (17, 33). However, additional studies with human breast cancer cell lines and hPR or cPR cDNA suggest that the PR-A and -B isoforms may also be generated from distinct promoters (18, 34, 35). The data presented in this paper clearly show the production of two GR protein products from a single cDNA source. In addition, Northern blot analysis of hGR␣-transfected COS-1 mRNA shows only a single hGR-specific message, further supporting the one-message/two- protein hypothesis (36). The androgen receptor (AR) has also been shown to exist as multiple forms, dif- fering at the amino terminus, which are expressed in a tissue-specific manner (37). Alternative initiation has been implicated as the mechanism responsible for the Fig. 8. Translation Scanning of GR Transcripts generation of these two AR forms (38). A, Sequence analysis of GR cDNA surrounding the first The mechanism of alternative initiation of the GR is two ATG codons. The consensus Kozak sequence, required shown to be under the translational direction exerted for maximum efficiency of protein translation initiation, is in the sequence surrounding the ATG codons (31). As shown above the sequence surrounding the first two ATG would be expected, creation of a strong Kozak con- codons from human, mouse, and rat GR. The underlined ATG sensus site (Ϫ3 C to G) by a point mutation 3Ј of GR sequence represents bases ϩ1, ϩ2, and ϩ3 respectively. Met1 AUG completely abolished ribosomal read- The bases at position Ϫ3 and ϩ4 have been found to signal either weak or strong initiator codons, as indicated. The through and production of GR-B. This new GR expres- second ATG in human corresponds to Met 27, while in rat and sion construct contains no coding region mutations mouse it is Met 28. B, Mutagenesis of the Kozak sequences and functions similarly to the M27T hGR used in these surrounding the two hGR ATG start sites. The weak Kozak studies. That a point mutant in the noncoding, pro- sequence at the upstream (Met 1) ATG was changed to a moterless region of the GR cDNA had such a drastic strong Kozak consensus site (M1 Kozak mutant). COS-1 cells effect on protein expression and function is remark- were transfected with this mutant along with WT hGR, and able and unprecedented in the nuclear receptor field. the two ATG mutants described elsewhere (M1T and M27T). Recent interest in mRNA regulation mechanisms, such Cells were harvested and analyzed by Western blotting for as splicing, stability, and the role of structure, are likely hGR with Ab 57. to lead to increased analysis of translational control mechanisms as identified here. personal communication). These data suggest that the Alternative translation initiation produces two func- extreme amino termini of steroid receptors may be tionally distinct forms of the GR. The presence of both involved in regulating receptor function. forms in several human cell lines and rodent tissues One possible explanation for the increased activity suggests that the generation of these two protein spe- of the shorter GR-B is that the tertiary structure includ- cies may be a general phenomenon. However, the ing the first 27 residues masks the activation func- ultimate physiological significance of these receptor tion(s) associated with other regions of the receptor. A isoforms remains to be established. Although our data, second possibility is that this region of the GR is utilizing both in vitro translation and transient expres- essential for an important protein interaction respon- sion systems, suggest both the GR-A and -B are being sible for either transcriptional silencing or repression. expressed via leaky ribosomal scanning from a single GR A and B 1101

cDNA and corresponding mRNA, we cannot rule out Cell Culture, Transfections, Luciferase, and the existence of alternative promoters regulating ex- Chloramphenicol Acetyltransferase (CAT) Assays pression in vivo, as has been shown for the PR (35). COS-1 and HEK293 cells were maintained in DMEM with We have presented evidence suggesting that both GR high glucose containing 2 mM glutamine and 10% (vol/vol) products are generated in vivo and in a variety of mixture of heat-inactivated FCS/calf serum (1:1). For trans- mammalian cell lines and that a significant functional activation assays, cells were incubated for 1–2 days in media difference exists between the two. It is intriguing to containing dextran-coated charcoal-stripped sera to remove speculate whether differences exist between GR-A endogenous steroids. HeLa cells were maintained in Eagle’s MEM supplemented with glutamine and 10% FCS/calf se- and GR-B in tissue distribution and/or expression dur- rum. CEM-C7 cells were grown in suspension in RPMI 1640 ing development, aging, or cell death. Such studies, medium supplemented with 2 mM glutamine, 10% (vol/vol) however, will require currently unavailable antibodies heat-inactivated FCS, and 0.1 M HCl. All cell culture media Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021 that selectively recognize the shorter GR-B form in the contained 100 IU/ml penicillin and 100 mg/ml streptomycin. Cell cultures were maintained in a 5% CO2 humidified incu- presence of GR-A. For example, it is now known that bator at 37 C and passaged every 3–4 days. All transfections the PR-A and -B isoforms function in a tissue-specific were carried out with Mirus TransIT LT-1 reagent according fashion (39) and that both isoforms regulate a distinct to the manufacture’s protocol (Pan Vera, Madison WI). An subset of genes (40). Regulated expression of either appropriate amount of TransIT reagent (3 ␮l per ␮g of transfected plasmid) was added to OPTIMEM (Life Sciences, Inc., GR isoform in favor of the other would suggest a St. Petersburg, FL) for 5 min. Purified plasmid DNA was then physiological role of alternative translation initiation of added and allowed to complex for 30 min at room tempera- the GR. Indeed, there are several reports that present ture, before being added to cells with media containing evidence for physiological regulation of alternative stripped serum. Six to eight hours after transfection, the translation initiation of critical transcription factors and media were replaced with fresh serum-stripped media containing vehicle or Dex. cell cycle regulators (41, 42). The potential for differ- Transfected cells were incubated in the presence or ab- ential regulation of functionally distinct GR isoforms, at sence of the indicated amount of Dex for 18–24 h before the level of translation, is an area that clearly needs harvesting. Cells for luciferase assays were harvested in 1ϫ further inquiry. Reporter Lysis Buffer (Promega Corp.). Total protein was measured using the Bio-Rad Protein Assay reagent (Bio-Rad Laboratories, Inc., Hercules, CA) according to the manufacture’s protocol, and equivalent amounts of total protein were used for luciferase activity assays. The luciferase activity was MATERIALS AND METHODS measured using the 96-well plate format with an MLX auto- mated microtiter plate luminometer from Dynex. Materials, Antibody Production, and Plasmids CAT assays were carried out essentially as described previously (45). Approximately 0.1–0.2 mg of protein extracts

35 were incubated overnight at 37 C with 1 mM acetyl-coenzyme trans- S-label (1,108 Ci/mmol) was purchased from ICN Bio- ␮ 14 14 A and 0.1 Ci of [ C]chloramphenicol in Tris-EDTA (TE). chemicals, Inc. (Irvine CA). [ C]Chloramphenicol (40–60 Samples were extracted in mixed xylenes and then back mCi/mmol) was obtained from NEN Life Science Products extracted one time with TE, pH 8.0, before liquid scintillation (Boston, MA). Dex was supplied by Steraloids (Wilton, NH). counting. A standard curve was generated using commer- Acetyl-coenzyme A and protease inhibitors were purchased cially available, purified CAT as described by the manufac- from Roche Molecular Biochemicals (Indianapolis, IN). Oligo- turer (Promega Corp.). All experiments were conducted un- nucleotide primers for mutagenesis and PCR were synthe- der conditions in which substrate was in excess and the sized by Oligo’s Etc. (Bethel, ME). The hGR-A-specific anti- relationship of counts per min to CAT activity was linear. Data body was produced by Covance Laboratories, Inc.(Denver, are expressed as counts per min per microgram of total PA) using a synthetic peptide antigen corresponding to res- protein. idues 3–22 of the hGR synthesized at the University of North Carolina at Chapel Hill. The GR-A- specific antibody was purified on a peptide-linked sepharose affinity column as Animals previously described for antibody purification (24). Charac- terization of this antibody is described below. The peroxi- Male Sprague Dawley rats (2–3 months old) and C57BL mice dase-labeled secondary antibodies and enhanced chemilu- (6 months old) were used in all experiments. All animals were minescence (ECL) reagents were purchased from Amersham maintained under controlled conditions of temperature (25 C) Pharmacia Biotech (Piscataway, NJ). and lighting and allowed free access to food and saline. All Use of the hGR mammalian (pCMV-hGR␣) and in vitro experimental protocols were approved by the animal review (pT7-hGR␣) expression vectors was described in a previous committee at the institute and were performed in accordance publication (23). The mouse GR mammalian and in vitro ex- with the guidelines set forth in the NIH Guide for the Care and pression vectors were also used as previously described (43). Use of Laboratory Animals published by the USPHS. Rats The rat GR expression vector, pRSV-rGR, was a gift of Dr. were killed by decapitation and mice were asphyxiated under

Trevor Archer and was used for both in vitro and COS-1 CO2, before removal of liver tissue. Liver tissue fragments expression systems. The reporter plasmids GRE1- and were homogenized on ice for 30 sec at maximum speed with GRE2-CAT (44), and pGCMS-CAT (45) have been described a Tekmar Tissuemizer in a radioimmunoprecipitation assay elsewhere. All site-specific mutagenesis was done with the (RIPA) buffer (50 mM Tris-HCl, pH 8.0, 0.1% SDS, 1% Triton Quick Change Mutagenesis kit (Promega Corp., Madison WI), ϫ-100, 0.5% sodium deoxycholate, 2 mM EDTA, and 150 mM following their protocol for primer design. The carboxy- NaCl) containing 5 mM dithiothreitol (DTT) and protease in- terminal hGR␣ truncation mutants, hGR(1–742) and hGR(1– hibitors. After a brief low-speed centrifugation to remove 706), were created by mutating the codons at positions 743 tissue debris, extracts were incubated for an additional 20 and 707 to TGA stop codons. All mutants were verified by min on ice before centrifugation at 20,000 ϫ g for 20 min. The DNA sequencing. resulting supernatants were measured for total protein con- MOL ENDO · 2001 Vol. 15 No. 7 1102

centration (typically 10–20 mg/ml) and subjected to Western REFERENCES blotting as described below.

1. Evans RM 1988 The steroid and thyroid hormone recep- Immunocytochemistry and Western Blotting tor superfamily. Science 240:889–895 2. Cole TJ, Blendy JA, Monaghan AP, Krieglstein K, Schmid Immunocytochemistry was carried out essentially as previ- W, Aguzzi A, Fantuzzi G, Hummler E, Unsicker K, Schutz ously described (22). The day after transfection, cells were G 1995 Targeted disruption of the glucocorticoid recep- split on two-chamber glass slides. Approximately 48 h after tor gene blocks adrenergic chromaffin cell-development transfection, the cells were treated with 100 nM Dex or vehicle and severely retards lung maturation. Genes Dev for 2 h. The cells were fixed in 2% paraformaldehyde, washed 9:1608–1621 in PBS, and permeabilized with 0.2% Triton X-100. Cells were 3. Truss M, Beato M 1993 Steroid hormone receptors: in- again washed in PBS, treated with 2% normal goat serum, teraction with deoxyribonucleic acid and transcription washed in PBS, and incubated with epitope-purified Ab57 factors. Endocr Rev 14:459–479 Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021 (1:7500) for 20 h at 4 C. The cells were washed in PBS and 4. Pratt WB, Toft DO 1997 Steroid receptor interactions incubated with biotinylated goat antirabbit IgF (1:400) for 1 h with heat shock protein and immunophilin chaperones. at room temperature. Immunoreactivity was visualized by Endocr Rev 18:306–360 staining with avidin-biotin-peroxidase. 5. Stromstedt PE, Poellinger L, Gustafsson JA, Carlstedt- For Western blotting, cells were lysed for 20 min on ice in Duke J 1991 The glucocorticoid receptor binds to a se- RIPA buffer containing 150 mM NaCl, 5 mM DTT, and pro- quence overlapping the TATA box of the human osteo- ␮ ␮ tease inhibitors (0.1 mM Pefa Block, 1 M leupeptin, and 1 M calcin promoter: a potential mechanism for negative pepstatin). After a high-speed spin to remove cellular debris, regulation. Mol Cell Biol 11:3379–3383 total protein was measured using the Bio-Rad Laboratories, 6. Yangyen HF, Chambard JC, Sun YL, Smeal T, Schmidt ␮ Inc. detection kit. Unless indicated otherwise, 50 g of pro- TJ, Drouin J, Karin M 1990 Transcriptional interference tein extract were then separated on precast 8% Tris-glycine between C-Jun and the glucocorticoid receptor: mutual gels (Novex, San Diego CA) and transferred to nitrocellulose. inhibition of DNA-binding due to direct protein-protein The membranes were washed in TBST (Tris-buffered saline interaction. Cell 62:1205–1215 with 0.1% Tween-20) and blocked in TBST containing 5% 7. Horwitz KB, Jackson TA, Bain DL, Richer JK, Takimoto nonfat milk for a minimum of2hatroom temperature. Blots GS, Tung L 1996 Nuclear receptor coactivators and core- were next incubated in the same solution supplemented with pressors. Mol Endocrinol 10:1167–1177 affinity-purified primary antibodies, Ab57 (1:2,500) and GR-A 8. Glass CK, Rose DW, Rosenfeld MG 1997 Nuclear recep- specific (1:5,000), overnight at 4 C. After extensive rinsing tor coactivators. Curr Opin Cell Biol 9:222–232 and washing in TBST (three times, 10–15 min), the blots were 9. Robyr D, Wolffe AP, Wahli W 2000 Nuclear hormone probed with peroxidase-conjugated goat antirabbit second- receptor coregulators in action: diversity for shared ary antibody (1:10,000) for2hatroom temperature. Bands tasks. Mol Endocrinol 14:329–347 were visualized using ECL reagents (Amersham Pharmacia 10. Sheldon LA, Smith CL, Bodwell JE, Munck AU, Hager GL Biotech). 1999 A ligand binding domain mutation in the mouse glucocorticoid receptor functionally links chromatin re- Repression of NF-␬B/p65 modeling and transcription initiation. Mol Cell Biol 19: 8146–8157 GR-mediated repression of NF-␬B/p65 transactivation was 11. Fryer CJ, Archer TK 1998 Chromatin remodelling by the studied as reported previously (44). A constant amount (12.5 glucocorticoid receptor requires the BRG1 complex. Na- ng) of the NF-␬B-p65 subunit expression vector (pCMVp65) ture 393:88–91 was used, in conjunction with the NF-␬B-luciferase reporter, 12. Weatherman RV, Fletterick RJ, Scanlan TS 1999 Nucle- 3XMHC-luc (0.5 ␮g), to measure p65-mediated transactiva- ar-receptor ligands and ligand-binding domains. Annu tion. The repression of p65 activity by GRs was measured by Rev Biochem 68:559–581 cotransfecting increasing amounts (10, 50, and 250 ng) of the 13. Tremblay A, Tremblay GB, Labrie F, Giguere V 1999 hGR-A or hGR-B expression vectors. After transfections, Ligand-independent recruitment of SRC-1 to estrogen cells were treated without or with 100 nM Dex. Approximately receptor ␤ through phosphorylation of activation function 16–20 h after hormone treatment, cells were harvested and AF-1. Mol Cell 3:513–519 luciferase activity was determined as described above. 14. Kuiper G, Enmark E, PeltoHuikko M, Nilsson S, Gustafs- Equivalent amounts of total protein were assayed for lucif- son JA 1996 Cloning of a novel estrogen receptor ex- erase activity as described above, and data from individual pressed in rat prostate and ovary. Proc Natl Acad Sci experiments were averaged and presented along with USA 93:5925–5930 the SEM. 15. Lazar MA 1993 Thyroid hormone receptors: multiple forms, multiple possibilities. Endocr Rev 14:184–193 Acknowledgments 16. Orti E, Bodwell JE, Munck A 1992 Phosphorylation of steroid hormone receptors. Endocr Rev 13:105–128 17. Conneely OM, Kettelberger DM, Tsai MJ, Schrader WT, We wish to thank Dr. Trevor Archer for the rGR plasmid, Dr. O’Malley BW 1989 The chicken progesterone receptor A Turk E. Rhen for supplying the rat liver tissue, and Dr. Ester and B isoforms are products of an alternate translation Carballo-Jane for the mice. The authors would also like to initiation event. J Biol Chem 264:14062–14064 thank Chris Jewell for technical assistance and discussions, 18. Kastner P, Krust A, Turcotte B, Stropp U, Tora L, Grone- and Drs. J. Hall and H. Kinyamu for reviewing the manuscript. meyer H, Chambon P 1990 Two distinct estrogen-regulated promoters generate transcripts encoding the two functionally different human progesterone receptor Received November 22, 2000. Revision received March forms A and B. EMBO J 9:1603–1614 15, 2001. Accepted April 2, 2001. 19. Giangrande PH, McDonnell DP 1999 The A and B iso- Address requests for reprints to: Dr. John Cidlowski, Lab- forms of the human progesterone receptor: two function- oratory of Integrative Biology, National Institute of Environ- ally different transcription factors encoded by a single mental Health Sciences, National Institutes of Health, 111 TW gene. Recent Prog Horm Res 54:291–313 Alexander Drive NC F307, Research Triangle Park, North 20. Kozak M 1995 Adherence to the first-AUG rule when a Carolina 27709-2233. E-mail: [email protected]. second AUG codon follows closely upon the first [pub- GR A and B 1103

lished erratum appears in Proc Natl Acad Sci USA 1995 34. Wei LL, Gonzalez-Aller C, Wood WM, Miller LA, Horwitz Jul 18;92(15):7134]. Proc Natl Acad Sci USA 92: KB 1990 5Ј-Heterogeneity in human progesterone recep- 2662–2666 tor transcripts predicts a new amino-terminal truncated 21. Hollenberg SM, Weinberger C, Ong ES, Cerelli G, Oro A, “C”-receptor and unique A-receptor messages. Mol En- Lebo R, Thompson EB, Rosenfeld MG, Evans RM 1985 docrinol 4:1833–1840 Primary structure and expression of a functional human 35. Kastner P, Bocquel MT, Turcotte B, Garnier JM, Horwitz glucocorticoid receptor cDNA. Nature 318:635–641 KB, Chambon P, Gronemeyer H 1990 Transient expres- 22. Oakley RH, Sar M, Cidlowski JA 1996 The human glu- sion of human and chicken progesterone receptors does ␤ cocorticoid receptor isoform. Expression, biochemical not support alternative translational initiation from a sin- properties, and putative function. J Biol Chem 271: gle mRNA as the mechanism generating two receptor 9550–9559 isoforms. J Biol Chem 265:12163–12167 23. Oakley RH, Webster JC, Sar M, Parker Jr CR, Cidlowski 36. Bellingham DL, Sar M, Cidlowski JA 1992 Ligand-depen- JA 1997 Expression and subcellular distribution of the dent down-regulation of stably transfected human glu- Downloaded from https://academic.oup.com/mend/article/15/7/1093/2748001 by guest on 27 September 2021 ␤-isoform of the human glucocorticoid receptor. Endo- cocorticoid receptors is associated with the loss of func- crinology 138:5028–5038 24. Oakley RH, Webster JC, Jewell CM, Sar M, Cidlowski JA tional glucocorticoid responsiveness. Mol Endocrinol 1999 Immunocytochemical analysis of the glucocorticoid 6:2090–2102 receptor ␣ isoform (GR␣) using GR␣-specific antibody. 37. Wilson CM, McPhaul MJ 1996 A and B forms of the Steroids 64:742–751 androgen receptor are expressed in a variety of human 25. Keightley MC, Fuller PJ 1995 Cortisol resistance and the tissues. Mol Cell Endocrinol 120:51–57 guinea-pig glucocorticoid receptor. Steroids 60:87–92 38. Wilson CM, McPhaul MJ 1994 A and B forms of the 26. McGimsey WC, Cidlowski JA, Stumpf WE, Sar M 1991 androgen receptor are present in human genital skin Immunocytochemical localization of the glucocorticoid fibroblasts. Proc Natl Acad Sci USA 91:1234–1238 receptor in rat brain, pituitary, liver, and thymus with 2 39. Vegeto E, Shahbaz MM, Wen DX, Goldman ME, O’Malley new polyclonal antipeptide antibodies. Endocrinology BW, McDonnell DP 1993 Human progesterone receptor 129:3064–3072 A form is a cell- and promoter-specific repressor of hu- 27. Webster JC, Jewell CM, Sar M, Cidlowski JA 1994 The man progesterone receptor B function. Mol Endocrinol glucocorticoid receptor: maybe not all steroid-receptors 7:1244–1255 are nuclear. Endocrine 2:967–969 40. Mulac-Jericevic B, Mullinax RA, DeMayo FJ, Lydon LP, 28. Reichardt HM, Kaestner KH, Tuckermann J, Kretz O, Conneely OM 2000 Subgroup of reproductive functions Wessely O, Bock R, Gass P, Schmid W, Herrlich P, Angel of progesterone mediated by progesterone receptor-B P, Schutz G 1998 DNA binding of the glucocorticoid isoform. Science 289:1751–1754 receptor is not essential for survival. Cell 93:531–41 41. Liu Y, Garceau NY, Loros JJ, Dunlap JC 1997 Thermally 29. McKay LI, Cidlowski JA 1999 Molecular control of im- regulated translational control of FRQ mediates aspects mune/inflammatory responses: interactions between nu- of temperature responses in the Neurospora circadian ␬ clear factor- B and steroid receptor-signaling pathways. clock. Cell 89:477–486 Endocr Rev 20:435–59 42. Spotts GD, Patel SV, Xiao QR, Hann SR 1997 Identifica- 30. Kozak M 1999 Initiation of translation in prokaryotes and tion of downstream-initiated c-Myc proteins which are eukaryotes. Gene 234:187–208 dominant-negative inhibitors of transactivation by full- 31. Kozak M 1986 Point mutations define a sequence flank- ing the AUG initiator codon that modulates translation by length c-Myc proteins. Mol Cell Biol 17:1459–1468 eukaryotic ribosomes. Cell 44:283–292 43. Webster JC, Jewell CM, Bodwell JE, Munck A, Sar M, 32. Horwitz KB, Alexander PS 1983 In situ photolinked nu- Cidlowski JA 1997 Mouse glucocorticoid receptor phos- clear progesterone receptors of human breast cancer phorylation status influences multiple functions of the cells: subunit molecular weights after transformation and receptor protein. J Biol Chem 272:9287–9293 translocation. Endocrinology 113:2195–2201 44. McKay LI, Cidlowski JA 1998 Cross-talk between nuclear 33. Conneely OM, Maxwell BL, Toft DO, Schrader WT, factor-␬B and the steroid hormone receptors: mecha- O’Malley BW 1987 The A and B forms of the chicken nisms of mutual antagonism. Mol Endocrinol 12:45–56 progesterone receptor arise by alternate initiation of 45. Allgood VE, Powell-Oliver FE, Cidlowski JA 1990 Vitamin translation of a unique mRNA. Biochem Biophys Res B6 influences glucocorticoid receptor-dependent gene Commun 149:493–501 expression. J Biol Chem 265:12424–12433