Title: Interdomain Signaling in a Two-domain Fragment of the Human Glucocorticoid Receptor
Abstract: Studies of individual domains or subdomains of the proteins making up the nuclear receptor family have stressed their modular nature. Nevertheless, these receptors function as complete proteins. Studies of specific mutations suggest that in the holoreceptors, intramolecular domain-domain interactions are important for complete function, but there is little knowledge concerning these interactions. The important transcriptional transactivation function in the N-terminal part of the glucocorticoid receptor (GR) appears to have little inherent structure. To study its interactions with the DNA binding domain (DBD) of the GR, we have expressed the complete sequence from the N-terminal through the DBD of the human GR. Circular dichroism analyses of this highly purified, multidomain protein show that it has a considerable helical content. We hypothesized that binding of its DBD to the cognate glucocorticoid response element would confer additional structure upon the N-terminal domain. Circular dichroism and fluorescence emission studies suggest that additional helicity as well as tertiary structure occur in the two-domain protein upon DNA binding. In sum, our data suggest that interdomain interactions consequent to DNA binding imparts structure to the portion of the GR that contains a major transactivation domain. Studies of individual domains or subdomains of the proteins making up the nuclear receptor family have stressed their modular nature. Nevertheless, these receptors function as complete proteins. Studies of specific mutations suggest that in the holoreceptors, intramolecular domain-domain interactions are important for complete function, but there is little knowledge concerning these interactions. The important transcriptional transactivation function in the N-terminal part of the glucocorticoid receptor (GR) appears to have little inherent structure. To study its interactions with the DNA binding domain (DBD) of the GR, we have expressed the complete sequence from the N-terminal through the DBD of the human GR. Circular dichroism analyses of this highly purified, multidomain protein show that it has a considerable helical content. We hypothesized that binding of its DBD to the cognate glucocorticoid response element would confer additional structure upon the N-terminal domain. Circular dichroism and fluorescence emission studies suggest that additional helicity as well as tertiary structure occur in the two-domain protein upon DNA binding. In sum, our data suggest that interdomain interactions consequent to DNA binding imparts structure to the portion of the GR that contains a major transactivation domain. DNA binding domain glucocorticoid receptor glucocorticoid response element The major identified domains of the nuclear family of receptors are those for ligand binding, DNA binding (DBD),1 and transactivation, with other functional areas mapped throughout the molecule (1Evans R.M. Science. 1988; 240: 889-895Crossref PubMed Scopus (6276) Google Scholar, 2Giguere V. Hollenberg S.M. Rosenfeld M.G. Evans R.M. Cell. 1986; 46: 645-652Abstract Full Text PDF PubMed Scopus (672) Google Scholar, 3Simon S.S. Vitam. Horm. 1994; 49: 49-130Crossref PubMed Scopus (52) Google Scholar, 4Kumar R. Thompson E.B. Steroids. 1999; 64: 310-319Crossref PubMed Scopus (305) Google Scholar). Although the ligand and DNA binding domains have modular structures and, in “domain swapping” experiments, a remarkable ability to carry out their function within the context of other proteins, intramolecular signaling is also important for the proper natural functions of these receptors. Recent experiments studying the effect of mutations on function emphasize the importance of this intramolecular signaling (5Gandini O. Kohno O. Curtis S. Korach K. Steroids. 1997; 62: 508-515Crossref PubMed Scopus (12) Google Scholar, 6Shao D. Rangwala S.M. Bailey S.T. Krakow S.L. Reginato M.J. Lazar M.A. Nature. 1998; 396: 377-380Crossref PubMed Scopus (306) Google Scholar). How and when information is exchanged between domains is largely unknown. This is due in part to the fact that no structures are yet available for multidomain proteins from the nuclear receptor family. One intriguing problem is the structural basis for the major transcriptional transactivation function (AF1, tau1) that mutagenesis experiments have defined in the human GR (7Hollenberg S.M. Evans R.M. Cell. 1988; 55: 899-906Abstract Full Text PDF PubMed Scopus (545) Google Scholar). By molecular genetics, AF1 is defined by amino acids 77–262. It appears to function by evoking physical interactions with the basal transcription mechanism, including Ada2 and TATA-binding protein (TBP), possibly through intermediary adapter protein (8Henriksson A. Almlof T. Ford J. McEwan I.J. Gustafsson J.A. Wright A.P.H. Mol. Cell. Biol. 1997; 17: 3065-3073Crossref PubMed Scopus (65) Google Scholar, 9Ford J. McEwan I.J. Wright A.P.H. Gustafsson J.A. Mol. Endocrinol. 1997; 11: 1467-1475Crossref PubMed Scopus (70) Google Scholar). But unlike the ligand binding domain and DBD, AF1 does not appear to function well out of its protein context. Studies of recombinant peptides from the GR containing AF1 have shown it to have little or no structure in simple buffer solutions (10Dahlman-Wright K. Baumann H. McEwan I.J. Almlof T. Wright A.P.H. Gustafsson J.A. Hard T. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 1699-1703Crossref PubMed Scopus (137) Google Scholar). In the presence of the strong α-helix stabilizing agent trifluoroethanol, up to three α-helices could form at the C-terminal end of AF1, and functional mutagenesis has shown that the primary sequences at the C-terminal end of AF1 may be relevant to tau1 function in vivo (10Dahlman-Wright K. Baumann H. McEwan I.J. Almlof T. Wright A.P.H. Gustafsson J.A. Hard T. Proc. Natl. Acad. Sci. U. S. A. 1995; 92: 1699-1703Crossref PubMed Scopus (137) Google Scholar). We have recently shown that in the presence of the osmolyte trimethylamine N-oxide, a small molecule that enhances natural protein folding, AF1 can take on tertiary structure (11Baskakov I.V. Kumar R. Srinivasan G. Ji Y. Bolen D.W. Thompson E.B. J. Biol. Chem. 1999; 274: 10693-10696Abstract Full Text Full Text PDF PubMed Scopus (147) Google Scholar). Thus, data short of actual structural proof support the idea that conditional folding of the transactivation domain is an important requirement for its interaction with target factors and its subsequent role in gene regulation (9Ford J. McEwan I.J. Wright A.P.H. Gustafsson J.A. Mol. Endocrinol. 1997; 11: 1467-1475Crossref PubMed Scopus (70) Google Scholar, 11Baskakov I.V. Kumar R. Srinivasan G. Ji Y. Bolen D.W. Thompson E.B. J. Biol. Chem. 1999; 274: 10693-10696Abstract Full Text Full Text PDF PubMed Scopus (147) Google Scholar, 12Dahlman-Wright K. McEwan I.J. Biochemistry. 1996; 35: 1323-1327Crossref PubMed Scopus (39) Google Scholar, 13Gill G. Ptashne M. Cell. 1987; 51: 121-126Abstract Full Text PDF PubMed Scopus (179) Google Scholar, 14Tjian R. Maniatis T. Cell. 1994; 77: 5-8Abstract Full Text PDF PubMed Scopus (953) Google Scholar). It is unclear whether and under what conditions such interdomain communication happens. Equally unknown are the consequent conformational changes. In the present studies, we test the hypothesis that information signaled from the DBD can enhance the propensity of AF1 to become structured. We have studied the secondary/tertiary structures of the GR fragments 1–500 (N-terminal through the DBD) and 398–500 (DBD) in solution alone and when bound to the DNA of a consensus glucocorticoid response element (GRE). The data show greater structural content, largely in the N-terminal region of the two-domain fragment when it is bound to the GRE. We believe this to be the first study of its kind from any member of the nuclear receptor family involving two naturally contiguous domains. The expression vectors for human GR fragments 1–500 and 398–500 contained a frameshift mutant of the human GR coding for amino acids 1–500 and 398–500 plus codons for a unique 5-amino acid sequence followed by a stop codon, as described (15Chen H. Srinivasan G. Thompson E.B. J. Biol. Chem. 1997; 272: 25873-25880Abstract Full Text Full Text PDF PubMed Scopus (18) Google Scholar, 16Graham F.L. Van der Eb A.J. Virology. 1973; 52: 456-467Crossref PubMed Scopus (6464) Google Scholar). These GR fragments were expressed as glutathione S-transferase fusion proteins in Sf9 insect cells (17Summers M.D. Smith G.R. Tex. Agric. Exp. Stn. Bull. 1987; 1555: 1-56Google Scholar). The cytosolic fractions were prepared from the cell pellet (18Srinivasan G. Thompson E.B. Mol. Endocrinol. 1990; 4: 209-216Crossref PubMed Scopus (51) Google Scholar). The proteins were purified from cytosolic fractions by binding to a glutathione-Sepharose column and then cleaved from glutathione S-transferase by digesting with thrombin and analyzed as described (11Baskakov I.V. Kumar R. Srinivasan G. Ji Y. Bolen D.W. Thompson E.B. J. Biol. Chem. 1999; 274: 10693-10696Abstract Full Text Full Text PDF PubMed Scopus (147) Google Scholar). Fluorescence polarization measurements were performed with an SLM 8000 spectrometer equipped with a polarizer, at excitation and emission wavelengths of 295 and 480 nm, respectively. Deoxyribonucleotides containing consensus GREs 5′-CTAGGCTGTACAGGATGTTCTGCCTAG-3′ and 5′-CTAGGCAGAACATCCTGTACAGCCTAG-3′ were synthesized and annealed. Fluorescence polarization studies were done by labeling the DNA sequence containing GRE with fluorescent probe 7-ethylamino-3(4′-maleimidylphenyl)-4-methylcoumarin as described (19Heyduk T. Lee J.C. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 1744-1748Crossref PubMed Scopus (208) Google Scholar). The binding of fragment 1–500 to GRE was studied in a buffer containing 20 mm Tris, pH 7.9, 60 mm KCl, 5 mm MgCl2, and 1 mm dithiothreitol. Fluorescence measurements were made using 1.0-cm path rectangular cuvettes thermostated at 22 °C. All the CD spectra were recorded at 22 °C on an Aviv 62 spectropolarimeter, with the bandwidth of 1.0 nm and scan step of 0.5 nm. The far-UV CD spectra were recorded using a 0.1-cm quartz cell; a 1.0-cm quartz cell was used for near-UV CD spectra. All the spectra were recorded in 20 mm Tris, pH 7.9, 60 mm KCl, 5 mm MgCl2, and 1 mm dithiothreitol and were corrected for the contribution of respective buffers. Each spectrum is a result of five spectra accumulated, averaged, and smoothed. Fluorescence emission spectra of the GR fragment 1–500 in solution were monitored using a Spex Fluoro Max spectrofluorimeter (excitation 278 or 295 nm). All the measurements were made using 1.0-cm path rectangular cuvettes maintained at 22 °C, and all the data were corrected for the contribution of the respective solute concentrations. The buffer conditions were the same as in the CD experiments. The 1–500 (Fig. 1 A) and 398–500 GR fragments were expressed in a baculoviral system, purified to homogeneity, and examined initially by CD spectroscopy. As shown in Fig. 1 B, the far-UV CD spectra of fragment 1–500 are concentration-dependent. At the lowest concentration, the spectrum shows mostly random coil, with some helical structure. With increasing concentrations, the minimum at around 200 nm shifted toward higher wavelengths and decreased, indicating a decrease in the random coil content of the protein. The minimum around 220 nm in these spectra did not seem to change as greatly, suggesting a lesser change in the helical content of the protein with changing concentration. Apparently, these concentration-dependent changes reflect some conformational changes other than helix. There may be an unfolding transition in the protein, caused by association of the protein with itself or with the surrounding environment at the higher concentrations. Dissociation at the lower concentration could lead to unfolding. However, a series of equilibrium centrifugation studies at protein concentrations of 2–18 μm failed to show any evidence of dimers or larger complexes (data not shown). In contrast, the far-UV CD spectra of fragment 398–500 are the same at the various concentrations of the protein tested, even at molar concentrations greater than those used for fragment 1–500 (Fig. 1 C). This suggests that the concentration-dependent changes observed in fragment 1–500 require the N-terminal region. A comparison of the spectra of fragments 1–500 and 398–500 shows that generally the spectra of both fragments are similar in form. However, the helical content (as assessed by the minimum at 222 nm; Fig. 1, B and C) in fragment 1–500 is higher compared with the small piece, 398–500. This further suggests that the structure observed in fragment 1–500 was due not only to the contribution of the DBD, but also to structure in the N-terminal region. Because this region has little structure when expressed alone, it acquires greater structure when expressed with the DBD in a contiguous protein. Because fragment 1–500 appears to be a monomer in solution at these concentrations, these structural changes are primarily intramolecular interactions (of course, intermolecular interactions, such as an excluded volume effect exerted by the less structured N-terminal region, may also play some role). We studied whether binding of the GR fragment 1–500 to a GRE leads to further structural changes in the two-domain GR fragment. Using fluorescence polarization measurements, we first estimated the stoichiometry of binding of fragment 1–500 to the DNA sequence containing a consensus GRE, labeled at its 5′ end with the fluorescence probe 7-ethylamino-3(4′-maleimidylphenyl)-4-methylcoumarin (Fig.2). The labeled GRE was then titrated with varying concentrations of the protein, following the change in anisotropy. Assuming the protein binds to GRE as a dimer, curves were fitted to the data at concentrations of GRE ranging from 15 nm to 1 μm. All gave Kavalues in a close range. The best fit was obtained with 100 nm GRE, from which the value of Ka = 1.04 × 10−2 nm−1 (Fig.2 A) was obtained. The stoichiometry of binding was calculated from the binding data obtained at 1 μm GRE concentration. As shown in Fig. 2 B, the binding ratio of protein to GRE is in fact approximately 2:1, which suggests that fragment 1–500 binds to GRE as a dimer, consistent with published data for DBD alone (20Hard T. Dahlman K. Carlstedt-Duke J. Gustafsson J.A. Rigler R. Biochemistry. 1990; 29: 5358-5364Crossref PubMed Scopus (69) Google Scholar). Based on these observations, we were able to choose conditions, i.e. [GRE] of ∼100 × Kd, that forced most protein to be GRE-bound, for study of the secondary structure of the protein·DNA complexes. Fig.3 shows the far-UV CD data of fragments 1–500 and 398–500 at protein/GRE input ratios of 1 and 1.5. The CD spectrum of fragment 398–500 does not change when bound to the GRE, but fragment 1–500 consistently shows a blue shift and greater negative ellipticity at wavelengths higher than 210 nm at both protein concentrations. Similar results were given at a ratio of 0.5:1, protein/DNA (data not shown). It is evident from these observations that the secondary structure of fragment 1–500 is significantly changed following binding to GRE. The blue shifts of the whole spectra without significant changes in relative intensities of the 208 nm and 220 nm peaks imply an increase in β-sheet content, a change that would favor an intradomain interaction. Furthermore, the increase in negative ellipticity at wavelengths higher than 240 nm implies a perturbation of the environments of aromatic residues. To acquire direct evidence for tertiary structural changes occurring in the protein following its binding to GRE, we recorded the near-UV CD spectra of this protein when unbound and bound to GRE (Fig.4 A). It is evident that the tertiary structure of the protein is significantly changed following its binding to GRE. A comparison of the spectra of the protein in the absence and presence of the GRE shows that when bound to the DNA there are a maximum and a minimum around 290 nm and 280 nm, respectively, readings which reflect perturbation of Trp. In the protein there are two Trp residues, one is found in AF1, the other between AF1 and the DBD (Fig. 1 A). Therefore, spectral changes at 290 and 280 nm after binding to the GRE indicate that tertiary structure has developed in or near the AF1 contained in the N-terminal region. In Fig. 4 B are shown the fluorescence emission spectra of GR fragment 1–500, measured either upon excitation at 295 nm, to follow changes in the environment of Trp residues specifically, or upon excitation at 278 nm, in which emission arises from Tyr and Trp residues, as well as being a result of energy transfer from Tyr to Trp residues. Thus, the latter links Tyr probes distributed throughout the protein with fluorescence emission from two Trp residues. It is evident from Fig. 4 B that the protein fluorescence spectra are changed following their binding to GRE. Because a substantial amount of the fluorescence probes of the GR fragment 1–500 are located outside of the DBD (both Trp residues, Trp-213 and Trp-365, and seven of ten Tyr residues; Fig. 1 A), the intrinsic fluorescence reflects mainly changes involving the N-terminal region. On the basis of our far-UV CD data of the GR fragments 1–500 and 398–500 we suggest that there is “cross-talk” between the N-terminal region and the DBD, which leads to some structural changes in the N-terminal region. This suggests that the individual domains may not have exactly the same structure in the holoreceptor as they do alone. Molecular genetic studies have shown that at least in the case of peroxisome proliferator-activated receptor, there is intramolecular cross-talk between the N-terminal region and ligand binding domain (6Shao D. Rangwala S.M. Bailey S.T. Krakow S.L. Reginato M.J. Lazar M.A. Nature. 1998; 396: 377-380Crossref PubMed Scopus (306) Google Scholar). The data suggest that the N-terminal region of the peroxisome proliferator-activated receptor modulates ligand binding by altering the conformation of the unliganded receptor. These conformational changes have been correlated with the receptor's interaction with cofactors such as the silencing mediator of retinoic acid and thyroid hormone receptors (21Lavinsky R.M. Jepsen K. Heinzel T. Torchia J. Mullen T.-M. Schiff R. Del-Rio A.L. Ricote M. Ngo S. Gemsch J. Hilsenbeck S.G. Osborne C.K. Glass C.K. Rosenfeld M.G. Rose D.W. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 2920-2925Crossref PubMed Scopus (577) Google Scholar). Such data indicate that the situation in the holoreceptor may be more complex than the simplest modular model anticipates. We postulate that structural changes occurring in the N-terminal region due to domain-domain interactions are important for maintaining the GR in a conformation suited for interactions between AF1 and its cofactors and/or specific proteins in the basal transcription machinery. The published structures of the DBD alone in solution and as a dimeric complex bound to GRE look similar, with no major difference in folding: only the second zinc finger region appears to be less well defined in solution compared with its crystal structure, suggesting that this region is stabilized upon formation of the (DBD)2·GRE complex (22Luisi B.F. Xu W.X. Otwinowski Z. Freedman L.P. Yamamoto K.R. Sigler P.B. Nature. 1991; 352: 497-505Crossref PubMed Scopus (1210) Google Scholar, 23Hard T. Kellenbach E. Boelens R. Maler B.A. Dahlman K. Freedman L.P. Carlstedt-Duke J. Yamamoto K.R. Gustafsson J.A. Kaptein R. Science. 1990; 249: 157-160Crossref PubMed Scopus (449) Google Scholar, 24Baumann H. Paulsen K. Kovacs H. Berglund H. Wright A.P.H. Gustafsson J.A. Hard T. Biochemistry. 1993; 32: 13463-13471Crossref PubMed Scopus (93) Google Scholar). Published results found that in solution the DBD was monomeric, whereas it bound as a dimer at the GRE site. The increased secondary structural elements indicated by our CD data suggest that the structure of the fragment 1–500·GRE complex may be more compact than that of the protein alone. These observations further suggest that the conformational alterations taking place in the 1–500 fragment upon DNA binding are mainly occurring in the N-terminal region. This implies an intramolecular flow of information from the DBD to the N-terminal region. These conditional structural changes in the N-terminal region of the GR following DNA binding via the DBD to its cognate GRE may play an important role in triggering gene regulation. Assuming that these changes are taking place in AF1, it can be imagined that binding to GRE is required to bring the receptor's major transactivation domain into a conformation suited for its interaction with cofactors and/or the proteins of the transcriptional machinery. Of course, this model in no way rules out the possibility of further structural changes in AF1, or the entire GR, as a result of those protein-protein interactions. Many documented instances of negative effects of the GR on transcription involve interactions with other transcription factors without the GR bound to a proper GRE (25Schule R. Rangarajan P. Kliewer S. Ransone L.J. Bolado J. Yang N. Verma I.M. Evans R.M. Cell. 1990; 62: 1217-1226Abstract Full Text PDF PubMed Scopus (1033) Google Scholar). If part of the functional structure of the GR necessary for positive regulation of transcription depends on GR-GRE interactions, these would be lost at the negatively regulated sites where there is no proper GRE interaction. The near-UV spectrum of the protein·DNA complex resembles that of dimers of two identical chromophores (26Cantor C.R. Schimmel P.R. Biophysical Chemistry, Part II. W. H. Freeman and Co., San Diego, CA1980: 418-425Google Scholar). This spectrum implies an induced close interaction between the Trp residues upon binding of GRE. The specific changes occurring in our two-domain protein will require other methods for studying structure. It seems clear, however, that binding of the DBD to its GRE produces changes indicative of increased tertiary structure in the N-terminal domain. This is consistent with our hypothesis that structure develops in the N-terminal domain following binding of the DBD to the GRE. Trp-213 is especially interesting. It is a critical amino acid for the activation function of the GR, and mutation to either positively or negatively charged amino acids results in loss of function. Trp-213 seems to act in conjunction with a limited number of other amino acids in the AF1 domain (27Iniguez-Lluhi J.A. Lou D.Y. Yamamoto K.R. J. Biol. Chem. 1997; 272: 4149-4156Abstract Full Text Full Text PDF PubMed Scopus (109) Google Scholar). Taken together, our data suggest that there is domaindomain interaction between the N-terminal region and the DBD in the GR fragment 1–500 and that this interaction is enhanced when the DBD binds a GRE, leading to imposition of some structure in the N-terminal region. We cannot immediately specify whether these changes are taking place only in the AF1 part of the N-terminal region or whether they are general. The structural changes observed in the N-terminal domain following binding of the DBD to a GRE open the possibility that the receptor, upon interacting with GRE, adopts a conformation important for the receptor's activity in vivo. We thank Dr. S. H. Lin for her assistance in the DNA-protein study.