Title: The Factor Binding to the Glucocorticoid Modulatory Element of the Tyrosine Aminotransferase Gene Is a Novel and Ubiquitous Heteromeric Complex
Abstract: Glucocorticoid induction of the tyrosine aminotransferase gene deviates from that of many glucocorticoid-responsive genes by having a lower EC5 and displaying more agonist activity with a given antiglucocorticoid. A cis-acting element, located 3646 base pairs upstream of the start of tyrosine aminotransferase gene transcription, has been found to be sufficient to reproduce these variations with heterologous genes and promoters (Oshima, H., and Simons, S. S., Jr.(1992) Mol. Endocrinol. 6, 416-428). This element has been called a glucocorticoid modulatory element, or GME. Others have called this sequence a cyclic AMP-responsive element (CRE) due to the binding of the cyclic AMP response element binding protein (CREB). We now report the partial purification and characterization of two new proteins (GMEB1 and −2) of 88 and 67 kDa that bind to the GME/CRE as a heteromeric complex. This purification was followed by the formation of a previously characterized, biologically relevant band in gel shift assays. By several biochemical criteria, the GMEBs differed from many of the previously described CREB/CREM/ATF family members. Partial peptide sequencing revealed that the sequences of these two proteins have not yet been described. Size exclusion chromatography and molecular weight measurements of the gel-shifted band demonstrated that the GMEBs bound to the GME as a macromolecular complex of about 550 kDa that could be dissociated by deoxycholate. Similar experiments showed that CREB bound to the GME as heteromeric complexes of about 310 and 360 kDa. As determined from gel shift assays, GMEB1 and −2 are not restricted to rat liver cells but appear to be ubiquitous. Thus, these novel GMEBs may participate in a similar modulation of other glucocorticoid-inducible genes in a variety of cells. Glucocorticoid induction of the tyrosine aminotransferase gene deviates from that of many glucocorticoid-responsive genes by having a lower EC5 and displaying more agonist activity with a given antiglucocorticoid. A cis-acting element, located 3646 base pairs upstream of the start of tyrosine aminotransferase gene transcription, has been found to be sufficient to reproduce these variations with heterologous genes and promoters (Oshima, H., and Simons, S. S., Jr.(1992) Mol. Endocrinol. 6, 416-428). This element has been called a glucocorticoid modulatory element, or GME. Others have called this sequence a cyclic AMP-responsive element (CRE) due to the binding of the cyclic AMP response element binding protein (CREB). We now report the partial purification and characterization of two new proteins (GMEB1 and −2) of 88 and 67 kDa that bind to the GME/CRE as a heteromeric complex. This purification was followed by the formation of a previously characterized, biologically relevant band in gel shift assays. By several biochemical criteria, the GMEBs differed from many of the previously described CREB/CREM/ATF family members. Partial peptide sequencing revealed that the sequences of these two proteins have not yet been described. Size exclusion chromatography and molecular weight measurements of the gel-shifted band demonstrated that the GMEBs bound to the GME as a macromolecular complex of about 550 kDa that could be dissociated by deoxycholate. Similar experiments showed that CREB bound to the GME as heteromeric complexes of about 310 and 360 kDa. As determined from gel shift assays, GMEB1 and −2 are not restricted to rat liver cells but appear to be ubiquitous. Thus, these novel GMEBs may participate in a similar modulation of other glucocorticoid-inducible genes in a variety of cells. INTRODUCTIONFor many years, the accepted model of steroid hormone action predicted that the responses of all regulated genes were a property of the steroid used. Thus, a gene is induced, or repressed, by agonists, and the action of agonists is prevented by antisteroids. Furthermore, the concentration of steroid required for half-maximal induction by an agonist and the amount of agonist activity possessed by a given antisteroid should be constant for each steroid and independent of the gene examined (reviewed in (1Simons Jr., S.S. Mercier L. Miller N.R. Miller P.A. Oshima H. Sistare F.D. Thompson E.B. Wasner G. Yen P.M. Cancer Res. 1989; 49: 2244-2252PubMed Google Scholar) and (2Simons Jr., S.S. Oshima H. Szapary D. Mol. Endocrinol. 1992; 6: 995-1002PubMed Google Scholar)).Recently, this model has had to be modified as exceptions were defined. Thus, jun•fos heterodimers (AP-1), and AP-1 inducers such as phorbol esters block steroid induction (3Jonat C. Rahmsdorf H.J. Park K-K. Cato A.C.B. Gebel S. Ponta H. Herrlich P. Cell. 1990; 62: 1189-1204Abstract Full Text PDF PubMed Scopus (1363) Google Scholar, 4Diamond M.I. Miner J.N. Yoshinaga S.K. Yamamoto K.R. Science. 1990; 249: 1266-1272Crossref PubMed Scopus (1066) Google Scholar) in what can be a cell-specific manner(5Cho H. Katzenellenbogen B.S. Mol. Endocrinol. 1993; 7: 441-452Crossref PubMed Scopus (187) Google Scholar, 6Maroder M. Farina A.R. Vacca A. Felli M.P. Meco D. Screpanti I. Frati L. Gulino A. Mol. Endocrinol. 1993; 7: 570-584Crossref PubMed Scopus (55) Google Scholar), while jun•jun homodimers augment glucocorticoid induction(4Diamond M.I. Miner J.N. Yoshinaga S.K. Yamamoto K.R. Science. 1990; 249: 1266-1272Crossref PubMed Scopus (1066) Google Scholar). Cyclic AMP, via protein kinase A, can often (but not always(7Wasner G. Simons Jr., S.S. Mol. Endocrinol. 1987; 1: 109-120Crossref PubMed Scopus (9) Google Scholar, 8Kazmi S.M.I. Visconti V. Plante R.K. Ishaque A. Lau C. Endocrinology. 1993; 133: 1230-1238Crossref PubMed Scopus (38) Google Scholar, 9Nordeen S.K. Bona B.J. Moyer M.L. Mol. Endocrinol. 1993; 7: 731-742Crossref PubMed Scopus (103) Google Scholar)) cause greater induction by agonists (10Rangarajan P.N. Umesono K. Evans R.M. Mol. Endocrinol. 1992; 6: 1451-1457Crossref PubMed Scopus (143) Google Scholar, 11Sartorius C.A. Tung L. Takimoto G.S. Horwitz K.B. J. Biol. Chem. 1993; 268: 9262-9266Abstract Full Text PDF PubMed Google Scholar) and increased percentages of agonist activity for antisteroids(11Sartorius C.A. Tung L. Takimoto G.S. Horwitz K.B. J. Biol. Chem. 1993; 268: 9262-9266Abstract Full Text PDF PubMed Google Scholar, 12Beck C.A. Weigel N.L. Moyer M.L. Nordeen S.K. Edwards D.P. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 4441-4445Crossref PubMed Scopus (193) Google Scholar). Heat shock, or chemical shock, afforded a synergistic increase in glucocorticoid inducibility(13Sanchez E.R. Hu J-L. Zhong S. Shen P. Greene M.J. Housley P.R. Mol. Endocrinol. 1994; 8: 408-421Crossref PubMed Scopus (67) Google Scholar), while the immunosuppressive agent FK506 augmented the activity of subsaturating concentrations of glucocorticoids(14Ning Y-M. Sanchez E.R. J. Biol. Chem. 1993; 268: 6073-6076Abstract Full Text PDF PubMed Google Scholar). Finally, dopamine can cause ligand-independent gene activation of some receptors(15Power R.F. Mani S.K. Codina J. Conneely O.M. O'Malley B.W. Science. 1991; 254: 1636-1639Crossref PubMed Scopus (496) Google Scholar). None of the above agents effected any shift in the dose-response curve for agonists except for FK506, which was postulated to increase the nuclear binding of activated complexes(14Ning Y-M. Sanchez E.R. J. Biol. Chem. 1993; 268: 6073-6076Abstract Full Text PDF PubMed Google Scholar).Other observations that did not appear to fit with the conventional model of steroid hormone action originated from studies on glucocorticoid induction of the tyrosine aminotransferase (TAT) 1The abbreviations used are: TATtyrosine aminotransferasebpbase pair(s)CREBcyclic AMP response element binding proteinGMEglucocorticoid modulatory elementHTChepatoma tissue cultureCREcyclic AMP-responsive elementFPLCfast protein liquid chromatographyCBPCREB binding proteinGREglucocorticoid response elementCATchloramphenicol acetyltransferasePAGEpolyacrylamide gel electrophoresisHPLChigh pressure liquid chromatography. gene in rat hepatoma tissue culture (HTC) cells, which had become a paradigm for steroid-inducible genes. We found that the dose-response curve for dexamethasone induction of TAT gene expression in the related Fu5-5 rat hepatoma cell line was left shifted compared to that in HTC cells(16Mercier L. Thompson E.B. Simons Jr., S.S. Endocrinology. 1983; 112: 601-609Crossref PubMed Scopus (80) Google Scholar). Similarly, TAT enzyme activity was induced at lower cAMP concentrations in Fu5-5 cells than in HTC cells(7Wasner G. Simons Jr., S.S. Mol. Endocrinol. 1987; 1: 109-120Crossref PubMed Scopus (9) Google Scholar). Furthermore, all antiglucocorticoids examined displayed a higher percentage of agonist activity for TAT gene expression in Fu5-5 than in HTC cells(16Mercier L. Thompson E.B. Simons Jr., S.S. Endocrinology. 1983; 112: 601-609Crossref PubMed Scopus (80) Google Scholar, 17Mercier L. Miller P.A. Simons Jr., S.S. J. Steroid Biochem. 1986; 25: 11-20Crossref PubMed Scopus (41) Google Scholar, 18Simons Jr., S.S. Yen P.M. Spelsberg T.C. Kumar R. Steroid and Sterol Hormone Action. M. Nijhoff, Boston1987: 251-268Crossref Google Scholar). This left shift in the TAT dose-response curve, and increased agonist activity with antisteroids, was not a general response of all glucocorticoid-inducible genes in Fu5-5 cells (19Wasner G. Oshima H. Thompson E.B. Simons Jr., S.S. Mol. Endocrinol. 1988; 2: 1009-1017Crossref PubMed Scopus (23) Google Scholar) and occurred at the level of correctly initiated transcripts(7Wasner G. Simons Jr., S.S. Mol. Endocrinol. 1987; 1: 109-120Crossref PubMed Scopus (9) Google Scholar, 19Wasner G. Oshima H. Thompson E.B. Simons Jr., S.S. Mol. Endocrinol. 1988; 2: 1009-1017Crossref PubMed Scopus (23) Google Scholar). Surprisingly, the magnitudes both of the left shift in the dose-response curve and of the increased amount of agonist activity were not constant but varied slowly over time (17Mercier L. Miller P.A. Simons Jr., S.S. J. Steroid Biochem. 1986; 25: 11-20Crossref PubMed Scopus (41) Google Scholar, 20Simons Jr., S.S. Miller P.A. Wasner G. Miller N.R. Mercier L. J. Steroid Biochem. 1988; 31: 1-7Crossref PubMed Scopus (11) Google Scholar) in a manner that was eventually found to be related to the density of the cells in culture(21Oshima H. Simons Jr., S.S. Endocrinology. 1992; 130: 2106-2112PubMed Google Scholar). Therefore, it appeared that some event downstream of steroid binding to the glucocorticoid receptor selectively modulated the properties of TAT gene induction by glucocorticoid agonists and antagonists.We previously proposed that this modulation of TAT gene induction in rat hepatoma cells occurred via the binding of a trans-acting factor to a cis-acting element of the TAT gene(1Simons Jr., S.S. Mercier L. Miller N.R. Miller P.A. Oshima H. Sistare F.D. Thompson E.B. Wasner G. Yen P.M. Cancer Res. 1989; 49: 2244-2252PubMed Google Scholar). Stable (22Szapary D. Oshima H. Simons Jr., S.S. Endocrinology. 1992; 130: 3492-3502Crossref PubMed Scopus (23) Google Scholar) and transient (23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar) transfection assays succeeded in identifying such a cis-acting element, at about −3646 bp of the rat TAT gene, that conveyed all of the glucocorticoid induction properties of the endogenous TAT gene to heterologous genes and promoters. This cis-acting element was called a glucocorticoid modulatory element (GME) and was found to bind a trans-acting factor(s)(23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar). The mechanism of action of the GME, unlike that of the commonly discussed transcription factor binding sites, does not involve synergism with the glucocorticoid response element, or GRE(24Oshima H. Simons Jr., S.S. J. Biol. Chem. 1993; 268: 26858-26865Abstract Full Text PDF PubMed Google Scholar). This suggests that the GME-bound factor(s) (GMEB) might be a novel protein.The binding site of the GMEB has also been identified as a cyclic AMP-responsive element (CRE)(25Boshart M. Weih F. Schimdt A. Fournier R.E.K. Schutz G. Cell. 1990; 61: 905-916Abstract Full Text PDF PubMed Scopus (115) Google Scholar, 26Nichols M. Weih F. Schmid W. DeVack C. Kowenz-Leutz E. Luckow B. Boshart M. Schutz G. EMBO J. 1992; 11: 3337-3346Crossref PubMed Scopus (274) Google Scholar, 27Nitsch D. Boshart M. Schutz G. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 5479-5483Crossref PubMed Scopus (161) Google Scholar), but several lines of evidence indicate that two different sets of proteins are responsible for GME and CRE activity. First, the biological activities mediated by GMEB and the CRE binding protein (CREB) are quite dissimilar ((21Oshima H. Simons Jr., S.S. Endocrinology. 1992; 130: 2106-2112PubMed Google Scholar, 22Szapary D. Oshima H. Simons Jr., S.S. Endocrinology. 1992; 130: 3492-3502Crossref PubMed Scopus (23) Google Scholar, 23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar) versus 25, 28). Second, no additional element is needed for GME biological activity, while a functional CRE requires a second TAT gene sequence, initially called BIII (25Boshart M. Weih F. Schimdt A. Fournier R.E.K. Schutz G. Cell. 1990; 61: 905-916Abstract Full Text PDF PubMed Scopus (115) Google Scholar) and more recently found to bind HNF-4(27Nitsch D. Boshart M. Schutz G. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 5479-5483Crossref PubMed Scopus (161) Google Scholar). Third, both the GME (23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar) and the CRE (28Weih F. Stewart A.F. Boshart M. Nitsch D. Schutz G. Genes & Dev. 1990; 4: 1437-1449Crossref PubMed Scopus (66) Google Scholar) give a closely spaced, three-band pattern in gel shift assays. However, the GMEB is responsible for slowest migrating of the three bands (23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar), which has been shown not to contain CREB(26Nichols M. Weih F. Schmid W. DeVack C. Kowenz-Leutz E. Luckow B. Boshart M. Schutz G. EMBO J. 1992; 11: 3337-3346Crossref PubMed Scopus (274) Google Scholar).The GME contains the sequence CGTCA, which is a common CRE element that binds homo- and heterodimers of CREB/ATF along with other family members or unrelated proteins (29Shepard A.R. Zhang W. Eberhardt N.L. J. Biol. Chem. 1994; 269: 1804-1814Abstract Full Text PDF PubMed Google Scholar) such as AP-1(30Ray A. Sassone-Corsi P. Sehgal P.B. Mol. Cell. Biol. 1989; 9: 5537-5547Crossref PubMed Scopus (162) Google Scholar). Thus, many known and unknown trans-acting factors could bind to the GME/CRE site at −3646 bp of the TAT gene. The purpose of this paper, therefore, was to characterize and purify the binding protein(s) proposed to be responsible for the GME activity of modulating glucocorticoid receptor function.DISCUSSIONWe have called the element at −3.6 kilobases of the rat TAT gene a glucocorticoid modulatory element, or GME, because it modulates the induction properties of both subsaturating concentrations of agonists and saturating concentrations of antagonists(23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar). The binding of a protein(s) to the GME was observed that was directly related to the biological activity of the GME oligonucleotide(23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar, 24Oshima H. Simons Jr., S.S. J. Biol. Chem. 1993; 268: 26858-26865Abstract Full Text PDF PubMed Google Scholar), which was different from synergism(24Oshima H. Simons Jr., S.S. J. Biol. Chem. 1993; 268: 26858-26865Abstract Full Text PDF PubMed Google Scholar). We now report that the GMEB appears to be a heterooligomer of two previously unsequenced proteins, GMEB1 and GMEB2, of apparent molecular masses of 88 and 67 kDa, respectively. However, conclusive identification must await the cloning of both proteins and a demonstration of biological activity with the cloned proteins in cells lacking GMEBs.Several properties of the GMEBs emerged during their purification that pertain to the mechanism of GME action. First, although the DNA sequence to which the GMEBs bind is very similar to that for the CREB/CREM/ATF and the Jun/Fos/AP-1 superfamilies, and CREB even binds to a non-consensus CRE at the same position as the GME at −3646 bp of the TAT gene(25Boshart M. Weih F. Schimdt A. Fournier R.E.K. Schutz G. Cell. 1990; 61: 905-916Abstract Full Text PDF PubMed Scopus (115) Google Scholar, 26Nichols M. Weih F. Schmid W. DeVack C. Kowenz-Leutz E. Luckow B. Boshart M. Schutz G. EMBO J. 1992; 11: 3337-3346Crossref PubMed Scopus (274) Google Scholar, 27Nitsch D. Boshart M. Schutz G. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 5479-5483Crossref PubMed Scopus (161) Google Scholar), there is little similarity between the GMEBs and these other proteins. The GMEBs are not related to CREB by the criteria of size, biological activity(23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar, 24Oshima H. Simons Jr., S.S. J. Biol. Chem. 1993; 268: 26858-26865Abstract Full Text PDF PubMed Google Scholar, 38Lalli E. Sassone-Corsi P. J. Biol. Chem. 1994; 269: 17359-17362Abstract Full Text PDF PubMed Google Scholar), antibody reactivity, methylation interference patterns for protein binding (Fig. 2A versus Fig. 7 of (28Weih F. Stewart A.F. Boshart M. Nitsch D. Schutz G. Genes & Dev. 1990; 4: 1437-1449Crossref PubMed Scopus (66) Google Scholar)), or amino acid sequence (Table 3). Some AP-1 sites contain the CGTC of the GME, and AP-1 may bind to the GME/CRE, as it has recently been reported that 12-O-tetradecanoylphorbol-13-acetate both inhibited glucocorticoid (and cAMP) induction of TAT and caused a decreased protein occupancy of the CRE at −3646 bp(45Reik A. Stewart A.F. Schutz G. Mol. Endocrinol. 1994; 8: 490-497PubMed Google Scholar). However, again there was no similarity between the peptide sequences of the GMEBs and AP-1, an anti-AP-1 antibody did not cause a supershift of GMEB-GME complexes, and 12-O-tetradecanoylphorbol-13-acetate alone did not elicit any response from GME-containing constructs (data not shown). Thus, there is little physical or biological similarity between the GMEBs and the other factors binding to the same DNA sequence. This, then, is an additional example of different proteins that bind to the same DNA region(46Steitz T.A. Q. Rev. Biophys. 1990; 23: 205-280Crossref PubMed Scopus (460) Google Scholar, 47Harrison S.C. Nature. 1991; 353: 715-719Crossref PubMed Scopus (499) Google Scholar).Second, we do not know if GMEB and CREB can both bind to the GME/CRE at the same time. However, it seems that CREB is unable to block GMEB action. The low levels of the protein kinase A regulatory subunit (Tse-1) in liver cells are thought to result in high amounts of active CREB that would bind to the GME/CRE(48Jones K.W. Shapero M.H. Chevrette M. Fournier R.E. Cell. 1991; 66: 861-872Abstract Full Text PDF PubMed Scopus (107) Google Scholar). Nevertheless, reporter constructs containing either a single GME (GMEGREtkCAT) or other elements that are needed for CRE activity, such as multiple tandem repeats of the GME or the GME plus the downstream BIII sequence(25Boshart M. Weih F. Schimdt A. Fournier R.E.K. Schutz G. Cell. 1990; 61: 905-916Abstract Full Text PDF PubMed Scopus (115) Google Scholar), show full GME activity in Fu5-5 rat hepatoma cells(23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar). Furthermore, conditions that elevate protein kinase A activity, such as forskolin treatment, did not inhibit GME activity. Thus, while CREB binds to the same DNA sequence as GMEB, CREB does not appear to competitively inhibit GMEB binding in intact cells.Third, GME activity is not limited to rat liver cells (Table 1), and the GMEBs are not tissue-specific proteins (Fig. 1A). Furthermore, the fact that the GME was active with synthetic GREs and a variety of promoters, including a minimum thymidine kinase promoter (23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar), suggests that no tissue-specific DNA binding factors are required for GME activity.Fourth, the GMEBs are clearly of nuclear origin but can be readily extracted from nuclei under conditions where other factors, such as CREB, stay in the nucleus. This is reminiscent of several other nuclear proteins(49Guiochon-Mantel A. Delabre K. Lescop P. Milgrom E. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 7179-7183Crossref PubMed Scopus (78) Google Scholar), including the progesterone and estrogen receptors, which are predominantly nuclear but appear in most cytosolic preparations. The cytosolic appearance of the GMEBs could be indicative of a dynamic equilibrium between the two cellular compartments, as established for the progesterone receptors(49Guiochon-Mantel A. Delabre K. Lescop P. Milgrom E. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 7179-7183Crossref PubMed Scopus (78) Google Scholar, 50Guiochon-Mantel A. Lescop P. Christin-Martre S. Loosfelt H. Perrot-Applanat M. Milgrom E. EMBO J. 1991; 10: 3851-3859Crossref PubMed Scopus (246) Google Scholar, 51Chandran U.R. DeFranco D.B. Mol. Endocrinol. 1992; 6: 837-844PubMed Google Scholar), or may simply reflect a repartitioning of the GMEBs in the lysis buffer.Fifth, the mass and stability of the GMEB complex are notable. The 550-600-kDa size of the protein complex seen in both gel shift assays and size exclusion chromatography (Fig. 3, A and B) argue against a nonspecific aggregate. Involvement of the 265-kDa protein CBP that binds phosphorylated CREB (52Chrivia J.C. Kwok R.P.S. Lamb N. Hagiwara M. Montminy M.R. Goodman R.H. Nature. 1993; 365: 855-859Crossref PubMed Scopus (1758) Google Scholar) was discounted by the observed sizes of GMEB1 and −2 and their lack of immunoreactivity with anti-CBP antibody. The most purified preparation of GMEB appeared to contain about equal amounts of GMEB1 and −2 (Fig. 5A), which would require three or four molecules of each protein in the final complex to achieve a molecular mass of 550-600 kDa. Such a massive complex is probably not too large to be extracted intact from HTC cell nuclei because identically prepared nuclei were found to be permeable to molecules as large as the 240-kDa protein complex of phycoerythrin(53Miyashita Y. Miller M. Yen P.M. Harmon J.M. Hanover J.A. Simons Jr., S.S. J. Steroid Biochem. Mol. Biol. 1993; 46: 309-320Crossref PubMed Scopus (5) Google Scholar). However, the GMEB complex must be quite stable to retain specific binding to the GME after extraction from the nucleus (Fig. 1), even in the presence of up to 2.7 M guanidinium hydrochloride and after various degrees of purification (Table 2). Despite the stability of the GMEB complex with regard to dissociation, the rate of reassociation of the separated components was relatively slow (Fig. 4).Finally, from the yield of purified GMEB1 and −2 in Table 3, it can be calculated that there are about 40,000 molecules of each GMEB per HTC cell. This is similar to the approximately 80,000 molecules of glucocorticoid receptor that are present in an HTC cell(16Mercier L. Thompson E.B. Simons Jr., S.S. Endocrinology. 1983; 112: 601-609Crossref PubMed Scopus (80) Google Scholar). Considering the fact that most glucocorticoid-responsive genes contain two GREs, each of which binds a dimer of the receptor, the ratio of GME-bound GMEB complexes to GRE-bound receptors is about 1:2. Given the facts that the GMEBs are not limited to rat liver cells and that GME-like modulation has been observed with several other glucocorticoid regulated genes(2Simons Jr., S.S. Oshima H. Szapary D. Mol. Endocrinol. 1992; 6: 995-1002PubMed Google Scholar), it will be interesting to pursue the possible role of a GME and its heteromeric binding complex in the transcription of genes other than TAT.While CREB has been identified as a component that also binds to the DNA sequence of the GME/CRE(26Nichols M. Weih F. Schmid W. DeVack C. Kowenz-Leutz E. Luckow B. Boshart M. Schutz G. EMBO J. 1992; 11: 3337-3346Crossref PubMed Scopus (274) Google Scholar), it is not known if it is the only protein in the complex. Several lines of evidence argue that the CREB-containing complexes bound to the GME are also multimeric. Most obvious is the size of the CREB-containing complexes, which were ~310-360 kDa in gel shift assays and 400 kDa on size exclusion columns (Fig. 3A and data not shown). Deoxycholate blocked the formation of the CREB-containing bands in gel shift assays, just as was observed for GMEB (Fig. 3B). Given the fact that CREB is relatively small (~42 kDa), it would seem that the CREB complexes must contain either multiple copies of CREB or other proteins. CREB will bind to the GME-containing oligonucleotide in Southwestern blots (see “Results”) and will afford gel-shifted bands with a CRE-containing oligonucleotide (26Nichols M. Weih F. Schmid W. DeVack C. Kowenz-Leutz E. Luckow B. Boshart M. Schutz G. EMBO J. 1992; 11: 3337-3346Crossref PubMed Scopus (274) Google Scholar) so that homooligomeric complexes of CREB may form. However, the gel-shifted band that the somatostatin CRE formed with purified CREB exhibited a faster migration than that with crude nuclear extracts(26Nichols M. Weih F. Schmid W. DeVack C. Kowenz-Leutz E. Luckow B. Boshart M. Schutz G. EMBO J. 1992; 11: 3337-3346Crossref PubMed Scopus (274) Google Scholar). Thus, the CREB-containing complex from nuclear extracts probably is not the same as that formed just from CREB and would contain some other protein(s). Further experiments are required to determine whether the suspected additional proteins are CBP (52Chrivia J.C. Kwok R.P.S. Lamb N. Hagiwara M. Montminy M.R. Goodman R.H. Nature. 1993; 365: 855-859Crossref PubMed Scopus (1758) Google Scholar), other members of the CREB/CREM/ATF superfamily that can heterodimerize with CREB(39Hummler E. Cole T.J. Blendy J.A. Ganss R. Aguzzi A. Schmid W. Beermann F. Schutz G. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 5647-5651Crossref PubMed Scopus (334) Google Scholar), or even GMEB1 or −2.In summary, we have found that a heteromeric complex of two potentially new proteins binds to a cis-acting element of the TAT gene. These two proteins, GMEB1 and GMEB2, are associated with changes in the transcriptional activity of antiglucocorticoids and low concentrations of glucocorticoids. These are phenomena that have not been previously described for steroid receptors and thus are of considerable mechanistic interest. It remains to be seen whether the GMEBs interact with glucocorticoid receptors and the transcriptional machinery in the manner that we have proposed(2Simons Jr., S.S. Oshima H. Szapary D. Mol. Endocrinol. 1992; 6: 995-1002PubMed Google Scholar, 23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar). The cloning of GMEB1 and GMEB2 and the production of specific antibodies will be of major assistance in understanding the mechanistic details of this interesting system. INTRODUCTIONFor many years, the accepted model of steroid hormone action predicted that the responses of all regulated genes were a property of the steroid used. Thus, a gene is induced, or repressed, by agonists, and the action of agonists is prevented by antisteroids. Furthermore, the concentration of steroid required for half-maximal induction by an agonist and the amount of agonist activity possessed by a given antisteroid should be constant for each steroid and independent of the gene examined (reviewed in (1Simons Jr., S.S. Mercier L. Miller N.R. Miller P.A. Oshima H. Sistare F.D. Thompson E.B. Wasner G. Yen P.M. Cancer Res. 1989; 49: 2244-2252PubMed Google Scholar) and (2Simons Jr., S.S. Oshima H. Szapary D. Mol. Endocrinol. 1992; 6: 995-1002PubMed Google Scholar)).Recently, this model has had to be modified as exceptions were defined. Thus, jun•fos heterodimers (AP-1), and AP-1 inducers such as phorbol esters block steroid induction (3Jonat C. Rahmsdorf H.J. Park K-K. Cato A.C.B. Gebel S. Ponta H. Herrlich P. Cell. 1990; 62: 1189-1204Abstract Full Text PDF PubMed Scopus (1363) Google Scholar, 4Diamond M.I. Miner J.N. Yoshinaga S.K. Yamamoto K.R. Science. 1990; 249: 1266-1272Crossref PubMed Scopus (1066) Google Scholar) in what can be a cell-specific manner(5Cho H. Katzenellenbogen B.S. Mol. Endocrinol. 1993; 7: 441-452Crossref PubMed Scopus (187) Google Scholar, 6Maroder M. Farina A.R. Vacca A. Felli M.P. Meco D. Screpanti I. Frati L. Gulino A. Mol. Endocrinol. 1993; 7: 570-584Crossref PubMed Scopus (55) Google Scholar), while jun•jun homodimers augment glucocorticoid induction(4Diamond M.I. Miner J.N. Yoshinaga S.K. Yamamoto K.R. Science. 1990; 249: 1266-1272Crossref PubMed Scopus (1066) Google Scholar). Cyclic AMP, via protein kinase A, can often (but not always(7Wasner G. Simons Jr., S.S. Mol. Endocrinol. 1987; 1: 109-120Crossref PubMed Scopus (9) Google Scholar, 8Kazmi S.M.I. Visconti V. Plante R.K. Ishaque A. Lau C. Endocrinology. 1993; 133: 1230-1238Crossref PubMed Scopus (38) Google Scholar, 9Nordeen S.K. Bona B.J. Moyer M.L. Mol. Endocrinol. 1993; 7: 731-742Crossref PubMed Scopus (103) Google Scholar)) cause greater induction by agonists (10Rangarajan P.N. Umesono K. Evans R.M. Mol. Endocrinol. 1992; 6: 1451-1457Crossref PubMed Scopus (143) Google Scholar, 11Sartorius C.A. Tung L. Takimoto G.S. Horwitz K.B. J. Biol. Chem. 1993; 268: 9262-9266Abstract Full Text PDF PubMed Google Scholar) and increased percentages of agonist activity for antisteroids(11Sartorius C.A. Tung L. Takimoto G.S. Horwitz K.B. J. Biol. Chem. 1993; 268: 9262-9266Abstract Full Text PDF PubMed Google Scholar, 12Beck C.A. Weigel N.L. Moyer M.L. Nordeen S.K. Edwards D.P. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 4441-4445Crossref PubMed Scopus (193) Google Scholar). Heat shock, or chemical shock, afforded a synergistic increase in glucocorticoid inducibility(13Sanchez E.R. Hu J-L. Zhong S. Shen P. Greene M.J. Housley P.R. Mol. Endocrinol. 1994; 8: 408-421Crossref PubMed Scopus (67) Google Scholar), while the immunosuppressive agent FK506 augmented the activity of subsaturating concentrations of glucocorticoids(14Ning Y-M. Sanchez E.R. J. Biol. Chem. 1993; 268: 6073-6076Abstract Full Text PDF PubMed Google Scholar). Finally, dopamine can cause ligand-independent gene activation of some receptors(15Power R.F. Mani S.K. Codina J. Conneely O.M. O'Malley B.W. Science. 1991; 254: 1636-1639Crossref PubMed Scopus (496) Google Scholar). None of the above agents effected any shift in the dose-response curve for agonists except for FK506, which was postulated to increase the nuclear binding of activated complexes(14Ning Y-M. Sanchez E.R. J. Biol. Chem. 1993; 268: 6073-6076Abstract Full Text PDF PubMed Google Scholar).Other observations that did not appear to fit with the conventional model of steroid hormone action originated from studies on glucocorticoid induction of the tyrosine aminotransferase (TAT) 1The abbreviations used are: TATtyrosine aminotransferasebpbase pair(s)CREBcyclic AMP response element binding proteinGMEglucocorticoid modulatory elementHTChepatoma tissue cultureCREcyclic AMP-responsive elementFPLCfast protein liquid chromatographyCBPCREB binding proteinGREglucocorticoid response elementCATchloramphenicol acetyltransferasePAGEpolyacrylamide gel electrophoresisHPLChigh pressure liquid chromatography. gene in rat hepatoma tissue culture (HTC) cells, which had become a paradigm for steroid-inducible genes. We found that the dose-response curve for dexamethasone induction of TAT gene expression in the related Fu5-5 rat hepatoma cell line was left shifted compared to that in HTC cells(16Mercier L. Thompson E.B. Simons Jr., S.S. Endocrinology. 1983; 112: 601-609Crossref PubMed Scopus (80) Google Scholar). Similarly, TAT enzyme activity was induced at lower cAMP concentrations in Fu5-5 cells than in HTC cells(7Wasner G. Simons Jr., S.S. Mol. Endocrinol. 1987; 1: 109-120Crossref PubMed Scopus (9) Google Scholar). Furthermore, all antiglucocorticoids examined displayed a higher percentage of agonist activity for TAT gene expression in Fu5-5 than in HTC cells(16Mercier L. Thompson E.B. Simons Jr., S.S. Endocrinology. 1983; 112: 601-609Crossref PubMed Scopus (80) Google Scholar, 17Mercier L. Miller P.A. Simons Jr., S.S. J. Steroid Biochem. 1986; 25: 11-20Crossref PubMed Scopus (41) Google Scholar, 18Simons Jr., S.S. Yen P.M. Spelsberg T.C. Kumar R. Steroid and Sterol Hormone Action. M. Nijhoff, Boston1987: 251-268Crossref Google Scholar). This left shift in the TAT dose-response curve, and increased agonist activity with antisteroids, was not a general response of all glucocorticoid-inducible genes in Fu5-5 cells (19Wasner G. Oshima H. Thompson E.B. Simons Jr., S.S. Mol. Endocrinol. 1988; 2: 1009-1017Crossref PubMed Scopus (23) Google Scholar) and occurred at the level of correctly initiated transcripts(7Wasner G. Simons Jr., S.S. Mol. Endocrinol. 1987; 1: 109-120Crossref PubMed Scopus (9) Google Scholar, 19Wasner G. Oshima H. Thompson E.B. Simons Jr., S.S. Mol. Endocrinol. 1988; 2: 1009-1017Crossref PubMed Scopus (23) Google Scholar). Surprisingly, the magnitudes both of the left shift in the dose-response curve and of the increased amount of agonist activity were not constant but varied slowly over time (17Mercier L. Miller P.A. Simons Jr., S.S. J. Steroid Biochem. 1986; 25: 11-20Crossref PubMed Scopus (41) Google Scholar, 20Simons Jr., S.S. Miller P.A. Wasner G. Miller N.R. Mercier L. J. Steroid Biochem. 1988; 31: 1-7Crossref PubMed Scopus (11) Google Scholar) in a manner that was eventually found to be related to the density of the cells in culture(21Oshima H. Simons Jr., S.S. Endocrinology. 1992; 130: 2106-2112PubMed Google Scholar). Therefore, it appeared that some event downstream of steroid binding to the glucocorticoid receptor selectively modulated the properties of TAT gene induction by glucocorticoid agonists and antagonists.We previously proposed that this modulation of TAT gene induction in rat hepatoma cells occurred via the binding of a trans-acting factor to a cis-acting element of the TAT gene(1Simons Jr., S.S. Mercier L. Miller N.R. Miller P.A. Oshima H. Sistare F.D. Thompson E.B. Wasner G. Yen P.M. Cancer Res. 1989; 49: 2244-2252PubMed Google Scholar). Stable (22Szapary D. Oshima H. Simons Jr., S.S. Endocrinology. 1992; 130: 3492-3502Crossref PubMed Scopus (23) Google Scholar) and transient (23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar) transfection assays succeeded in identifying such a cis-acting element, at about −3646 bp of the rat TAT gene, that conveyed all of the glucocorticoid induction properties of the endogenous TAT gene to heterologous genes and promoters. This cis-acting element was called a glucocorticoid modulatory element (GME) and was found to bind a trans-acting factor(s)(23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar). The mechanism of action of the GME, unlike that of the commonly discussed transcription factor binding sites, does not involve synergism with the glucocorticoid response element, or GRE(24Oshima H. Simons Jr., S.S. J. Biol. Chem. 1993; 268: 26858-26865Abstract Full Text PDF PubMed Google Scholar). This suggests that the GME-bound factor(s) (GMEB) might be a novel protein.The binding site of the GMEB has also been identified as a cyclic AMP-responsive element (CRE)(25Boshart M. Weih F. Schimdt A. Fournier R.E.K. Schutz G. Cell. 1990; 61: 905-916Abstract Full Text PDF PubMed Scopus (115) Google Scholar, 26Nichols M. Weih F. Schmid W. DeVack C. Kowenz-Leutz E. Luckow B. Boshart M. Schutz G. EMBO J. 1992; 11: 3337-3346Crossref PubMed Scopus (274) Google Scholar, 27Nitsch D. Boshart M. Schutz G. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 5479-5483Crossref PubMed Scopus (161) Google Scholar), but several lines of evidence indicate that two different sets of proteins are responsible for GME and CRE activity. First, the biological activities mediated by GMEB and the CRE binding protein (CREB) are quite dissimilar ((21Oshima H. Simons Jr., S.S. Endocrinology. 1992; 130: 2106-2112PubMed Google Scholar, 22Szapary D. Oshima H. Simons Jr., S.S. Endocrinology. 1992; 130: 3492-3502Crossref PubMed Scopus (23) Google Scholar, 23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar) versus 25, 28). Second, no additional element is needed for GME biological activity, while a functional CRE requires a second TAT gene sequence, initially called BIII (25Boshart M. Weih F. Schimdt A. Fournier R.E.K. Schutz G. Cell. 1990; 61: 905-916Abstract Full Text PDF PubMed Scopus (115) Google Scholar) and more recently found to bind HNF-4(27Nitsch D. Boshart M. Schutz G. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 5479-5483Crossref PubMed Scopus (161) Google Scholar). Third, both the GME (23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar) and the CRE (28Weih F. Stewart A.F. Boshart M. Nitsch D. Schutz G. Genes & Dev. 1990; 4: 1437-1449Crossref PubMed Scopus (66) Google Scholar) give a closely spaced, three-band pattern in gel shift assays. However, the GMEB is responsible for slowest migrating of the three bands (23Oshima H. Simons Jr., S.S. Mol. Endocrinol. 1992; 6: 416-428PubMed Google Scholar), which has been shown not to contain CREB(26Nichols M. Weih F. Schmid W. DeVack C. Kowenz-Leutz E. Luckow B. Boshart M. Schutz G. EMBO J. 1992; 11: 3337-3346Crossref PubMed Scopus (274) Google Scholar).The GME contains the sequence CGTCA, which is a common CRE element that binds homo- and heterodimers of CREB/ATF along with other family members or unrelated proteins (29Shepard A.R. Zhang W. Eberhardt N.L. J. Biol. Chem. 1994; 269: 1804-1814Abstract Full Text PDF PubMed Google Scholar) such as AP-1(30Ray A. Sassone-Corsi P. Sehgal P.B. Mol. Cell. Biol. 1989; 9: 5537-5547Crossref PubMed Scopus (162) Google Scholar). Thus, many known and unknown trans-acting factors could bind to the GME/CRE site at −3646 bp of the TAT gene. The purpose of this paper, therefore, was to characterize and purify the binding protein(s) proposed to be responsible for the GME activity of modulating glucocorticoid receptor function.