John D. Baxter)), and Norman L. EberhardtSWI. With the technical assistance of Cheri Mueske$. From the $Endocrine Research Unit, Departments of Medicine ...
THEJOURNALOF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Bioehemistry and Molecular Biology, Inc.
Vol. 267. No. 2. Issue of January 15, pp. 913-921,1992 Printed in U.S . A
Thyroid HormoneResponsiveness in Human Growth Hormone-related Genes POSSIBLECORRELATIONWITHRECEPTOR-INDUCEDDNACONFORMATIONALCHANGES* (Received for publication, June 6, 1991)
Fritz Leidig$$, Allan R. ShepardST, Wengang Zhangll**, Adele Stelterl, Peter A. Cattini$$$$, John D. Baxter)),and Norman L. EberhardtSWI With thetechnical assistance of Cheri Mueske$ From the $Endocrine Research Unit, Departments of Medicine and BiochemistrylMolecularBiology, Mayo Clinic, Rochester, Minnesota 55905, the 11Metabolic Research Unit, Department of Medicine, University of California, San Francisco, California 94143, and the $$Departmentof Physiology, University of Manitoba, Winnipeg, Manitoba R3E OW3, Canada
Triiodothyronine (T3) induces the transcription of the human chorionic somatomammotropin (hCS) promoter transfected into rat pituitary (GC) cells, but does not stimulate the homologous human growth hormone (hGH) promoter. As demonstrated by forward and reverse mutagenesis, this differential T3 responsiveness is due to subtle structural differences in a T3 response element located between nucleotides -64 and -44 of the 5’-flanking DNA of the hGH and hCS promoters. Synthetic hCS(-70/-40) DNA binds thyroid hormone receptors with a 4-fold higher affinitythan the corresponding hGH Ts response element, indicating that small differences in receptor binding properties are reflected bymajor differences in T3 responsiveness. Analysis of circular permutation fragments containing the native hGH and hCS or mutated hCS(-70/-40) sequences demonstrates that the thyroid hormone receptor induces DNA bending. The extent ofbending shows a possible correlation with the function of these sequences, suggesting that the receptor-induced changes in DNA conformation may be required for thyroid hormone receptor action.
Triiodothyronine (T3)’regulates gene expression
through
* This work was supported in part by National Institutes of Health Grants DK41206 (to N. L. E.), DK41842 (to J. D. B.), and a grant from the Medical Research Council of Canada (to P. A. C.). F. L. and A. R. S. made equal contributions to this work and thereforeare designated as co-equal first authors. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 9 Supported by a stipend from the Deutsche Forschungsgemeinschaft. T Partially fulfilled the requirements for the degree of Doctor of Philosophy, Mayo Graduate School with these studies. ** Supported by a University of California, San Francisco Cheng Scholarship. Current address: Hormone Research Inst., Box 0543/ HSW 1088, University of California, San Francisco, CA 94143. $$ Recipient of a Medical Research Council of Canada Scholarship. TW To whom correspondence should be addressed: 4-407 Alfred, Mayo Clinic, Rochester, MN 55905. GH, The abbreviations used are: TB, 3’,3,5-triiodo-~-thyronine; growth hormone; hGH, human growth hormone;rGH, rat growth hormone; CS, chorionic somatomammotropin; hCS, human chorionic somatomammotropin; TRE, thyroidhormoneresponseelement; CAT, chloramphenicol acetyltransferase; 5’-FR, 5”flanking region; FBS, fetal bovine serum; PBS, phosphate-buffered saline; nt(s), nucleotide(s); bp, base pair(s).
specific nuclear receptors. The genes encodingthese receptors are membersof a superfamily of genes, including those encoding the steroid hormone, retinoic acid, vitamin D3 (Evans, classes of receptors 1988; Carson-Jurica et al., 1990), and other (Oro et al., 1988; Wang et al., 1989; Mlodzik et al., 1990). Thyroid hormone receptors (Tg receptors) bind to specific DNA sequences termed thyroid hormone response elements (TREs). The most extensively characterized T R E is on the 5”flanking DNA (5’-FR) of the rat growth hormone (rGH) gene thatparticipatesinT3-mediatedinduction of rGH mRNA (Flug et al., 1986; Glass et al., 1987; Koenig et al., 1987; Lavin et al., 1988; Brent et al., 1989; Norman et al., 1989). This TRE can activate a heterologous promoter in a T3- and receptor-dependent fashion, suggesting that it is a hormone-responsive enhancer element (Ye et al., 1988). The full structural requirements for TREs, the primary mechanisms by which T3receptors act to enhance transcription and their interrelationships with other DNA elements on a T3responsive promoter, remain to be clarified. It is generally assumed that basic elements of the mechanisms of action of the steroid/thyroid hormone receptorsuperfamily are similar and that protein-protein interactionsbetween receptors and other elementsof the transcriptional apparatus participate in receptor action. For this and other classes of eukaryotic transcription factors, very little attention hasbeen focused on potentialreceptor influences on theDNA. Protein-induced DNA conformationalchangeshave beenshown to be important for the function of certain prokaryotic transcription factors (reviewed in Travers, 1990; Hagerman, 1990), endonucleases (Frederick et al., 1984), and histones (Richmond et al., 1984). Thus, the cataboliteactivatorprotein when complexed with CAMP binds to many Escherichia coli promoters and induces DNA bending (Kolbet al., 1983; Wu and Crothers,1984), and curved DNA sequences lacking catabolite activator protein-binding sitescanfunctionally replace cataboliteactivatorproteinbinding sites in theE . coli gal promoter (Bracco et al., 1989). Integration of bacteriophage h DNA into the E. coli chromosome requires the DNA-bending protein IHF (Robertson and Nash, 1988); IHF canbe functionally replaced by insertion of DNA sequences that bind catabolite activator protein but not by DNA sites that bind proteins that do not induce DNA bending (Goodman and Nash,1989). Several eukaryotictranscription factors, including the canonical zinc finger protein TFIIIA, which is required for 5 S ribosomal genetranscription (Schroth et al., 1989), a Drosophila heat shock transcription factor (Shuey and Parker, 1986), a factor that binds to the c-
913
914
T3Responsiveness and DNA Bending
fos and a-actin promoters (Gustafson et al.., 1989), and NF-
KB(Schreck et al.,1990) all induce DNA bending upon binding to DNA. However, it is unknown in these cases whether the protein-induced bending is important for their function. For these reasons, it is critical to determine whether eukaryotic regulatory proteins such as those of the steroid/thyroid hormone receptor family can also induce DNA bending and whether such bending can be correlated with the ability of these proteins to regulate promoter activity. That the site that binds T3 receptors may have more importance than simply an anchor was suggested by our earlier studies with the rGHpromoter. We found that a mutated,T3unresponsive rGH promoter still retained its ability to bind T3 receptors with high, albeit somewhat reduced, affinity (Norman et aZ., 1989). It therefore seemed important to address whether TI responsiveness of other promoters also lacked a strict correlation of receptor binding to DNA and whether this could be related to some other property of the receptor-DNA interaction. We found previously that the transfected human growth hormone (hGH) and chorionic somatomammotropin (hCS, also designated placental lactogen) genes differ in their T3 responsiveness in rat pituitary tumor (GC) cells (Cattini et ab, 1986). These two genes have nearly identical nucleotide sequences (-95%) over much of their 5'- and 3'-flanking DNA, exons, and introns (Miller and Eberhardt, 1983; Selby et al., 1984; Hirt et al., 1987; Chen et al., 1989), and their expression in GC cells depends on sequences in the proximal promoters (Nachtigal et al., 1989; Nickel et.al.,1990) that bind GHF-l/Pit-l, a factor that mediates pituitary-specific GH gene expression (Lefevre et al.,1987; Bodner et al., 1988; Dolle et al., 1990; Li et al., 1990). Whereas the intact hCS gene and the endogenous rGH gene or the rGH and hCS 5'flanking DNA linked to the chloramphenicol acetyltransferase (CAT) gene are positively regulated by TI (Cattini and Eberhardt, 1987), the intact hGHgene is negatively regulated by T3 (Cattini et al.,1986), and a hybrid gene containing the hGH 5'-FR fused to the CAT gene isnot T3-responsive (Cattini and Eberhardt, 1987; Cattini et al., 1988; Nickel et al.,1990) in transfected GC cells; sequences mediating negative TScontrol of the hGH gene are localized to the 3'-FR (Zhang et al., 1990). Since the hGH and hCS 5"FR DNAs are nearly identical, subtle structural differences must account for their differential TBresponsiveness. This regulatory behavior appears to mimic the physiological regulation of the hGH and hCS genes in the pituitary and placenta, respectively. The hGH gene does not appear to be regulated by T3 in acromegalic pituitary tumor cells maintained in primary culture (Isaacs et al., 1987); whereas the hCS gene is positively regulated by T3 in human choriocarcinoma cells (BeWo) (Nickel and Cattini, 1991). We report here that sequences responsible for T3 regulation of the hCS promoter are localized to at least two distinct regions in theproximal promoter between nts -124/-87 and -87/+2. The upstream element binds T3 receptors and can be replaced by the equivalent sequence from the hGH promoter. By contrast, the downstream element of the hGH gene is not functional and its T3nonresponsiveness is localized to at least two sequence differences from the hCS promoter in the -7O/-40 region. T3 receptors bind to the hCS(-70/-40) region with high affinity and induce DNA bending as evidenced by analysis of circularly permuted DNA fragments. Whereas the T3 nonresponsive wild-type hGH and mutated hCS(-70/-40)-binding sites also bind the T3 receptors with high affinity, their affinity is somewhat reduced, andthe A. Stelter and N. L. Eberhardt, unpublished results.
degree of T3receptor-induced conformational change is much less. These correlations between T3 receptor-induced DNA bending and thefunctional response suggest that T3 receptorinduced DNA conformational changes may participatein receptor action. EXPERIMENTAL PROCEDURES
Materials-Plasmids containing Bal 31 deletion mutants of the hGH and hCS 5"FR were constructed as described by Cattini et al. (1986). Hybrid promoter constructs containing selected regions of the hGH and hCS 5'-FR, as well as the deletion/insertion mutants (nts -87/-83) have been previously described by Nachtigal et al. (1989). Oligonucleotides were obtained from California Biotechnology Inc. (Mountainview, CA) and the Molecular Biology Core Facility, Mayo Clinic. Site-specific mutagenesis of the hCS 5"FR cloned into M13 was performed with an Amersham mutagenesis kit according to the manufacturer's specifications. The hGH 5'-flanking mutant, hGH(1,3X)p.CAT, was generated by inverse polymerase chain reaction-mediated mutagenesis (Hemsley et al., 1989). Briefly, 5 ng of a plasmid containing the hGH 5'-flanking DNA (nts -492/SalI to +2/ BarnHI) in pUC8 was incubated with 100 pmol of each primer in a 100-pl reaction containing 10 mM Tris.HC1 (pH 8.3), 50 mM KCl, 2.5 mM MgC12, 200PM each dNTP, 0.01% (w/v) gelatin, and 2.0 units of Taq DNA polymerase (Perkin Elmer-Cetus Instruments) and subjected to 25 polymerase chain reaction cycles consisting of 1 min X 94 "C denaturation, 1 min X 50 "C annealing, and 8 min x 72 "C elongation steps. The primers were 5'-TCCCTGCTTGACCCCACC3' (hCS lower strand; nts -47/-64) and 5"GAGAGAAGGGGCCAG3' (hGH upper strand; nts -46/-33). Themutant product was isolated as described by Hemsley et al. (1989), and the hGH 5"FR was removed by SalIlBarnHI digestion and cloned into SalIlBglII digested pCrS vector (Cattini et al., 1986). The end points of the individual deletion mutants and the sequences of the site-specific mutants were verified by dideoxy chain termination sequencing (U. S. Biochemical Corp.). Y-[~'P]ATP(>6,000 Ci/mmol), ~ ' - ( Y [ ~ ~ S ] ~ A T P (>1,000 Ci/mmol), and ['2SI]triiodothyronine(>1,000Ci/mmol) were purchased from Amersham Corp. Poly(dI.dC) was purchased from Pharmacia LKB Biotechnology Inc. Cell Culture and DNA Transfections-Monolayers of rat GC cells were maintainedin Dulbecco'smodifiedEagle'smedium supplemented with 5% fetal bovine serum (FBS), 5% calf serum. Cells for DNA transfections were deinduced in Dulbecco's modified Eagle's medium supplemented with 4% hormone-depleted FBS (Samuels et al., 1979) for 40-48 h. DNA tranfections were performed by electroporation (Bio-Rad). Deinduced cells were harvested with trypsinEDTA and washed twice with 4% hormone-depleted FBS. For each transfection, 40 X lo6 cells were resuspended in 0.4 ml of Dulbecco's phosphate buffered saline (PBS), 0.1% glucose containing 15 fig of DNA. The suspended cells were subjected to 350 V at 960 microfarads at 4 "C in 4-mm cuvettes (Bio-Rad), allowed to stand at 4 "C for 7 min, and subsequently suspended in PBS supplemented with 4% hormone-depleted FBS. Suspended cells were incubated a t room temperature for an additional 10 min, diluted with Dulbecco's modified Eagle's medium containing 4% hormone-depleted FBS, and plated onto 4 X 10-cm dishes. T3was added to a final concentration of 10 nM to two of the dishes, and thecells were maintained for 3648h. In all cases, the data represent the average T3 stimulation observed in at least three separate transfectionsperformed at different times. CAT Assays-Transiently transfected GC cells were washed with PBS lacking Ca2+and M $ + (PBS-CMF) and harvested with 1 mM EDTA-PBS-CMF (pH 7.5). The cells were centrifuged (500 X g for 2 min), and the cell pellet was resuspended in 0.5 ml of 100 mM TrisHCl (pH 7.8), 0.1% Triton X-100. After incubation at 4 "C for 15 min, the suspension was centrifuged at 14,000 X g for 15 min at 4 "C, and the supernatantwas removed, assayed for protein by Coomassie binding (Pierce) and stored at -70 "C. Aliquots of cell extracts (1550 pg) were assayed using the two-phase liquid scintillation assay of Neumann et al. (1987), except that 0.5 pCi of [3H]acetyl-CoAwas used without added unlabeled acetyl-coA (Eastman, 1987).The samples were counted a t room temperature for 1or 2 min for at least five cycles and a total incubation time of300-500 min. CAT activity represents the linear regression slope of the data (plotted as cpm uersus time) normalized per mg of protein (cpm.min".mg"). The CAT expression data were normalized to total cellular protein. In separate experiments, T, was found to depress the amount of total
T3Responsiveness and cellular protein/cell to 77% of control cells under the conditionsused for deinduction andinduction of the cells. The normalized CAT expression data was not corrected for this effect. Normalization of the data to DNA instead of protein did not alter the relative differences in the various constructs. In some cases, 3 pg of the luciferase expression plasmid pA,RSVp.LUC (Wood et al., 1989) was co-transfected to control for differences in transfection efficiency. Inclusion of this normalization did not alter any of the relative values or the significance of the data as determined by statistical analysis. Despite wide variation in normalized expression (75-13,700 cpm. min” .mg”) observed inthedatain Fig. 1,the observedregulation of the -496hCSp.CAT gene by TS(4.5 & 1.2 S.D.) was independent of basal expression. We were unable to employ RSVp.PGa1 as a cotransfection control, because it was found to be negatively regulated 2.5-fold by T, in GC cells when normalized to totalcellular proteins. In addition, this plasmid has been shown to selectively interfere with the expression of different deletion mutants in two promoter systems (Gorman et al., 1985; Flug et al., 1987). Thyroid Hormone Receptor DNA Binding Assays-The partially purified (-1% homogeneity) rat liver thyroid hormone receptor was purified accordingto the method of Apriletti et al. (1988). The binding assays were performed as previously described (Lavin et al., 1988; Apriletti et al., 1988; Norman et al., 1989) using restriction enzymegeneratedhGHandhCS5”FRfragments or synthetic doublestranded oligonucleotides. DNA fragments andoligonucleotides were purified by polyacrylamide gel electrophoresis. The partiallypurified T, receptor was incubated with [“sI]TS at 4 “C for 2 h prior to the addition of DNA. Binding assays (15 pl) contained 10 ng (126-505 fmol) of the DNA fragment or oligonucleotide, 3 fig of poly(dI.dC), and 10 fmol of T, receptor in 20 mM sodium phosphate (pH 7.6), 1 mM MgCl,, 0.5 mM EDTA, 80 mM NaC1,0.1% monothioglycerol containing 5 pg/ml each of pepstatin A, leupeptin, chymostatin, and antipain. After incubation a t room temperature for 10 min, the reaction mixturewas loaded onto a 5% polyacrylamide/bisacrylamide (30:l) gel and electrophoresed in 6.7 mM Tris acetate (pH7.5), 1 mM EDTA, 3.2 mM sodium acetate at 240 V for 30 min to 2 h at 4 “C. All polyacrylamide gels were pre-electrophoresed a t 200 V for 90 min at 4 “C. Circular PermutationlDNA Bending Assays-T3-induced DNA bending was analyzed by circular permutation analysis, aspreviously described by (WuandCrothers, 1984; ShueyandParker, 1986; Robertson and Nash, 1988; Leidig et al., 1990). Oligonucleotides (30 bp) corresponding to hGH(-70/-40), wild-type hCS(-70/-40), and mutated hCS(-70/-40)-1X and hCS(-70/-40)-3X sequences were subcloned into the SmaI siteof a derivative pUC19 vector, pAUC19. The vector pAUC19 was modified by inverse polymerase chain reaction-mediated mutagenesis(Hemsley et al., 1989) utilizing 5’ACATCCCCCTTTCG-3’ and 5‘-CCCAGTCACGACG-3’ as primers t o remove a thyroid hormone receptor-binding sitebetween the FokI and BstNI sites within the lac Z region (nts 334/355) (Leidig et al., 1990). In all cases, single inserts with the syn orientation relative to the Amp’ gene were selected. In each case, a SauIIIA/BarnHI fragment (nts 276/417) containing the insertwas inserted into its parent pAUC19 vector at the BamHI site (nt 417), and clones were selected that contained the duplicated inserts in the same orientation yield to the circular permutation plasmids pCPGH, pCPCS, pCPCS-lX, and pCPCS-3X. These plasmids were subsequently digested with SauIIIA, PuuII, FokI, EcoRI and Asp7181 and purified by polyacrylamide gel electrophoresis to yield a series of identically sized fragments that differ only in the position of the hGH or hCS sequences relative to the ends of the DNAs. Each circular permutation fragment (10 ng, 108 fmol) was incubated with 10 fmol of T8 receptor that had been pre-equilibrated with 30 fmol of [‘251]T3(762 mCi/pmol, Du PontNew England Nuclear) in 10 mM sodium phosphate (pH 7.6), 67 mM NaC1,0.25 mM EDTA, 0.5 mM MgCl,, 5% glycerol, and 3 pg of poly(dI.dC) in a total volume of 15 p1 for 15 min a t 4 “C. Ten pl of this reaction mixture was electrophoresed at 200 V for 4 h at 4 “C in 7.5% polyacrylamide, 0.26% bisacrylamide gels in TBAE buffer (pH 7.50) (107 mM Tris base, 89 mM boric acid, 2 mM EDTA adjusted to p H 7.50 with glacial acetic acid). The gels were always pre-electrophoresed at 200 V for 2 h a t 4 “C prior to the application of the samples. RESULTS
Responsiveness of the hGH and hCS Promoters-confirming earlier studies (Cattini and Eberhardt, 1987), a hybrid gene containing thehCS5“FR fused tothe CAT gene
DNA Bending
915
(hCSp.CAT), but not the hGHp.CAT gene, was positively regulated 4.5-fold by T3 in transiently transfected GC cells (Fig. 1A). The TSresponsiveness of the hCSp.CAT gene was lost upon deletion to nt -83 (Fig. lA). These data suggested that elements required for T3 responsiveness of the hCS 5’FR were localized between nts -496 and -83 and that the hGH 5”FR lacked such elements. A hybrid promoter with thedistalhGH5”FR(nts -4431-83) linked tothe -83hCSp.CAT gene was regulated 3.9-fold by T3. By contrast, an analogous construct containing the distal hCS5’-FR (nts -496/-83) linked to the -83hGHp.CAT gene did notrespond to T3 (Fig. 1B). Taken together, these data demonstrate that sequences in both the distal andproximal hCS promoter are required for fullT3responsiveness and thatlack of T3responsiveness of the hGH 5”FR is due to a structural difference from the hCS proximal promoter (nts -83/+2). To define the 5”boundary of the T3-responsive domain,we examined5”deletion mutants of the hCSp.CAT gene. No significant decrease in T3 responsiveness was observed until deletion to nt-94, where about 40% of the T3responsiveness is lost (Fig. IC). Further deletion to nts -83 (Fig. L4) or -61 (Fig. 1C) abolished T3 induction. Thus, at least part of the T3-responsive domain of the hCS promoter is localized beMODIFIED hGHp.CAT AND hCSp.CAT GENES
T3 REGULATION (FOLD S.E.)
*
C
D
p -871-83
E.
-4 96
FIG. 1. Thyroid hormone regulation of CAT activity (cpm min” X mg” of protein) from hGHp.CAT and hCSp.CAT genes in transfected GC cells. A, effect of deletion to nt -83 on X
T3 induction
of wild-type hGHp.CAT and hCSp.CAT genes. B, promoter-hybrid constructions containing distal hGH and hCS 5”flanking DNA fused to the proximal (nts -83/+2) hCS and hGH promoters. C, effect of 5”deletion on T, induction of hCSp.CAT activity. D, effect of 5”deletion on the hGH/hCSp.CATpromoterhybrid gene. E , effect of an internal 4-bp deletion (-87/43) on the T3 induction of the hGHp.CAT andhCSp.CAT genes. F, effect of insertion of two BamHIlinkers (solid pyramid) between nts -87/-83 onthe T, induction of the hGHp.CAT and hCSp.CAT genes. Also shown are the effects of placing the distal 5”flankingDNA a t the 3’-end of the CAT gene. Errors represent standard errors; the number of trials is indicated inparentheses; the numberof transfections isgenerally half the number of trials.
916
T3Responsiveness and
tween nts -124 and -83 with apartially active domain localized between nts -94 and -83. Deletion of the hybrid promoter hGH/hCSp.CAT gene (Fig. 1D) to nt-135 did not significantly affect Ts responsiveness. Thus, thesequences in the hGH 5”FR that confer Tnresponsiveness onto the hCS proximal promoter are localized within nts -135/-83, suggesting that this portion of the response domain is similar in the hGH and hCS 5’-FRs. The proximal binding site for the cell-specific factor GHF1 spans nts -96 to -62 (Lefevre et al., 1987). To determine if this site is involved in mediating the Ts response, a 4-bp deletion between nts -87 and -83 was introduced into the hCSp.CAT and hGHp.CAT genes. Fig. 1E demonstrates that this manipulation had little or no effect on the T3responsiveness of the hCSp.CAT gene. Similarly, inclusion of a 20-bp RamHI linker into the-87/-83 site of hCSp.CAT and hGH/ hCSp.CAT promoter hybrid genes resulted in a modest decrease (24-36%) in thyroid hormoneresponsiveness (Fig. 1F). By contrast, these deletion/insertionmutants only exhibit 38% of the basal activity of the wild-type hGH and hCS promoters (data are not shown but they confirm the reports of Lefevre et al. (1987), Nachtigal et al. (1989), and Nickel et al. (1990)), and this loss of activity is correlated with the inability of these mutated sequences to bind GHF-1 (Lefevre et al., 1987;Nachtigal et al., 1989). Since TI responsiveness is only marginally affected by these mutations, it appears that GHF-1 is not required and at least two different elements localized upstream and downstream of nts -87/-83 constitute the Ts response domain. However, these elements appear to act only in tandem and do not appear to act as classical enhancers, because Taresponsiveness is lost completely when the two regions are separated by large distances (Fig. 1F). Binding of Thyroid Hormone Receptors to hGH and hCS 5’FR-To further localize and define elements in thehGH and hCS promoters that interactwith the thyroid hormonereceptors, thebinding of unlabeled hCS and hGH 5“FR DNAs by [ “‘I]T3-labeled thyroid hormone receptors partially purified from rat liver were examined by gel retardation assays (Apriletti et al., 1988; Lavin et al., 1988; Norman et al., 1989). Since the functional data (Fig. 1, C and D) indicate that thedifferencein Ts responsiveness between the two promoters is localized to the -87/+2 region andthatboth promoters require element(s) between nts -135 and -87,we analyzed receptor binding to a series of overlapping 30-bp oligonucleotides spanning the hCS 5”FR from nts -150 to +6 (Fig. 2). Three general regions of binding activity are centered near nucleotides -90 (lanes f and g),-60 (lanes i and j ) , and -20 (lanes 1-n). Receptor binding tothe previously defined rGH(-27/-6) sequence (Norman et al., 1989) is shown as a positive control (Fig. 2, lane a). Fig. 3 demonstratesthat hGH 5”FR DNA fragments corresponding in location to those of the hCS 5”FR that bind receptors also bind receptors. The binding of the -loo/-70 fragment corresponds to the upstream element defined by the functional analysis (Fig. 1D). Similar to thehCS oligonucleotides, the highest affinity binding occurs with the hGH(-70/-40) oligonucleotides and the weakest binding activity is seen in the more proximal 5”FR (nts -4O/-1). Functional Analysis of the hCS Proximal Promoter-The functional analyses (Fig. 1) suggest that the difference in Tn responsiveness of the hGH and hCSpromoters is due to structural differences in the proximal promoter. Since the most proximal promoter (nts -55/+2) contains seven of the eight sequence differences present in the first 150 nts of the hGH and hCS 5”FR and three of these differences occur within the highest affinity Ts receptor binding site (nts-701
DNA Bending hCS 5’-Flanking /
/
!J
Receptor . T3’
] Receptor a
b
c
d
e
f
g
h
*
.
T3* DNA
i j k l m n o
FIG.2. Binding of T3receptors to a series of overlapping 30-bp oligonucleotides spanning the hCS 5‘-flanking DNA from nts -150 to -1 (lanes b o ) . Each adjacent oligonucleotide shares 20 bp of sequence with its neighbor. The number designations at thetop depict the end points of the DNA relative to the transcription initiation site. The receptor.T3*complex remains near theorigin and the DNA.receptor.T3* complexes migrate into the gel. Lane a contains a positive control rGH(-27/-6) oligonucleotide (Norman et al., 1989).
FIG. 3. Binding of T3receptors to selected 30-bp oligonucleotides spanningthe hGH 5“flanking DNA between nts -1 10 and -1. The number designations at the top of the figure indicate the end points of the DNA sequence relative to the transcription initiation siteof the hGHgene. The pairsof selected oligonucleotides correspondto thosefor the hCS oligonucleotides that bind the T3receptors shown in Fig. 2.
-40), we made site-specific mutants that converted the hCS 5”FR to the corresponding hGH sequence in this region. As shown in TableI, deletion of a Ga t n-54 t or the simultaneous conversion of G + T and A + G at nts -50 and -48, respectively, abolished TSresponsiveness. We also created an hGH 5’-FR mutant thatconverted the -7O/-40 region to the hCS sequence (hGH(1,3X)p.CAT). As shown in Table I, the thyroid hormone responsiveness of the hGH(1,3X)p.CAT gene is restored to levels that are not significantly different from wild-type hCS levels. These data demonstrate that the high affinity Ts receptor binding site localized near nt -60 is required for Ts responsiveness and that the lack of Ts responsiveness of the hGH 5’-FR is explained by the sequence differences occurring at nts-54, -50, and -48. Comparative Receptor Binding to Proximal hGH and hCS Promoter-The receptor binding in the -87/+2 region of the two genes was examined in greaterdetail. When a “P-labeled hCS(-7O/-40) oligonucleotide was employed with receptorT3complex, an unlabeled hCS(-70/-40) oligonucleotide competed with the 32P-labeledDNA about 4-fold more avidly than the corresponding hGH(-70/-40) oligonucleotide (Fig. 4). Also, Scatchard analysis(data notshown) of receptor binding to varying amounts of DNA indicated 3.4-fold tighter binding to the hCS(-70/-40) site ( K d = 6.3 & 3.1nM, n = 2) than
T3Responsiveness and DNA Bending TABLE I Mutation of the hCS and hGH(-54/-48) region defines a functional TRE containing direct repeats The represented sequences are hGH (nts- 64/-39) and hCS (nts-65/-39). Basal activity is expressed in cpm X min-l X mg of protein-' X lo-'. The activity of the control CATvector lacking a promoter varied between 0.06 and 0.16 cpm X min" X mg" of protein" X lo-? and never exceeded 5% of the expression of the -496hCSp.CAT DNA and/or to theexpression of a luciferase control (pA3RSVLUC; (wild-type) gene. Normalization of the data to Experimental series 2) to normalize differences in transfection efficiency did not significantly change the relative values or the statisticaloutcomes of the experiments. Construct Sequence Basal activity" TIregulation"
917
n
x 10-3
Experimental series 1 -443hGH(w.t.) 2.95 f 1.25 (0.999) 0.6 f 0.1 (0.029) (-65)GGTGGGGTCAA:CAGtGgGAGAGAAgg 2.6 f 0.3 (-64)GGTGGGGTCAAgCAGgGaGAGAGAAct -496hCS(w.t.) 4.45 f 0.88 0.7 f 0.1 (0.000) .................... T.G ......... 6.36 f 1.40 (0.251) -496hCS(lX) 29.17 f 5.21 (0.001) 0.7 f 0.1 (0.001) ................................ -496hCS(3X) Experimental series 2 2.5 f 0.1 (-65)GGTGGGGTCAAgCAGgGaGAGAGAAct 13.18 f 2.14 -496hCS(w.t.) 1.0 f 0.1 (0.002) .................... T.G ......... 35.85 8.54 (0.010) -496hCS(lX) 0.7 f 0.0 (0.000) 12.71 f 4.57 (0.888) -443hGH(w.t.) (-64)GGTGGGGTCAA:CAGtGgGAGAGAAgg ................ G ...G . A ......... 1.9 f 0.1 (0.101) 21.62 2.90 (0.007) -443hGH(1,3X) Probability (inparentheses) was calculatedaccording to a Wilcoxon nonparametric one-wayanalysis of variance due to unequal variances in someof the compared samples. All probability values are derived from paired data and compared with the values for -496hCSp.CAT (wild-type) expression.
* *
4 9 9 8 26 4 12 4
hGH(-70/-40)
20
40 60 80 Competitor (nM)
100
FIG.4. Relative affinityof thyroid hormone receptor bind- FIG.5. Differentialmigration of the hCS(-87/+2) and ing to hGH(-70/-40) and hCS(-70/-40) oligonucleotides. The hGH(-87/+2) DNA. T3 receptor complexes during polyacrylamide hCS (-70/-40) oligonucleotide was "2P-end-labeled with [y-:"P]ATP and combined with varying concentrations of unlabeled hGH(-70/ -40) or hCS(-70/-40) oligonucleotide. The insets show autoradiograms of the data;numerical data were derived by excising the bands and counting thesamples.
gel electrophoresis. Left panel, the hCS and hGH DNA fragments were "'P-end-labeled and electrophoresed in the absence of protein. Rightpanel, the unlabeled hGHandhCS DNA fragments were incubated with TSreceptor. ["'1]T3 complex and electrophoresed.
Wu and Crothers(1984) have shown that bentDNA migrates more slowly than linear DNA in an electric field. Thus differthe hGH(-70/-40) site ( K d= 21.6 2.8 nM, n = 2). Surprisingly, we found differences in the migration of the ential migration of the circularly permuted DNA fragments -87/+2 hGH and hCS DNA fragments when complexed with reflects a measure of the T3 receptor-induced DNA conforreceptor were observed in gel retardation experiments (Fig. mational change. As shown in Fig. 6, for all of the binding 5). These two DNA fragments that differ by 7 nucleotides are sites, the dataclearly indicate that the mobility of the DNA identical insize and exhibit identical migration in the absence is decreased as the binding site moved is toward the centerof of receptor (Fig. 5). Although the binding of additional pro- the DNA fragment. Control fragments (Fig. 6A) lacking either teins to thehCS(-87/+2) DNA might accountfor this differ- hGH or hCS(-70/-40) regions did not bind receptor, indiential migration, this isconsidered highly unlikely due to the cating that the observed binding anddifferential DNA-recepminor difference inmigration (Fig. 5). Alternatively,itis tor complex mobility are only dependent upon the inserted possible that theT3receptor may be inducing a change in the DNA. Thus, these data fulfill the criterionof Wu and Crothers (1984) that DNA bending is induced by Ts receptor binding. DNA conformation (Norman etal., 1989). Analysis of Receptor-induced DNA Bending-To ascertain T o determine the relative degree of TS receptor-induced whether DNA bending could account for the difference in conformational change, the circular permutation experiments were repeated several times, and the relative migration data electrophoretic mobility of the hGH and hCS(-87/+2) fragments, DNA bending experiments (Wu and Crothers, 1984) were plotted (Fig. 7). All of the binding sitesexhibited signifwereperformed with circularly permuted DNA fragments icant differences in their relativedegree of migration. Imporcontaining the -7O/-40 region of the hGH, hCS, hCS-1X tantly, the magnitude of the change in mobility decreases in and hCS-3X genes. In these experiments, a series of identi- the order CS > CS-3X > CS-1X > GH. These data indicate cally sized DNA fragments are generated in which the Ts that the degree of receptor-induced DNA bending is related receptor-binding site is situated at different positions relative to the structureof the DNA-binding site and that this differto theDNA ends. In accordance with electrophoresis theory, ence is related directly to the functional activities that are
*
T3Responsiveness and DNA Bending
918 FIG. 6. Analysis of T3 receptor binding to circularly permuted DNA fragments ( W u and Crothers, 1984) containing the hGH (pCPGH), wildtype hCS (pCPCS), and mutated hCS-3X (pCPCS-3X) and hCS-1X (pCPCS-1X) -7O/-40 TS receptorbinding sites. Free ["'1]T3receptor complex migrates diffusely at the topof the gels, and the ["'1]T3receptor DNA complexes migrate into the gel as defined bands. A, comparison of circularly permuted pAUC19 control fragments (lanes f - j ) with pCPGH fragments (lanes a-e). H , comparison of circularly permuted pCPCS (lanes a-e) with pCPGH (lanes f-j) DNA fragments. C, comparison of circularly permuted pCPCS (lanes a-e) with pCPCS-1X (lanes f-j) DNA fragments. D,comparison of circularly permuted pCPCS (lanes a-e) with pCPCS3X (lanes f-j) D N A fragments. Restriction enzymes used to generate the individualfragmentsare Sau3Al(lanesa and f ) ,PuuII (lanes b andg), FokI (lanes c and h ) , EcoRI (lanes d and i) and Asp7181 (lanes e and;). In all cases, the comparisons between the pCPCS DNA fragmentsandtheother T, receptorbindingsites were performedonthe same gel to avoid intergel variation.
1
I'
L a
CPCsl x
(-ma) 1
J b
c
d
e
f
g
B.I cpcs(-~140)
h
i
j
a
b
D.
CPGH (-70/-40) I
c
d
e
f
CPCS (-70/-40)
g
h
i
j
(-ma)
CPCS~X
I
1
c" a
b
c
d
e
expressed by these T3 receptor-binding sites. It is also of interest that thetwo individually mutated CS sequences exhibit DNA bending characteristics that are intermediate between the GH and CS sequences. Since the hGH sequence contains all of the sequence differences represented by the CS-1X- and CS-3X-binding sites, the data suggest that each of these structural differences contributes about equally to the overall conformational change that is induced upon receptor binding.
f
g
h
i
j
a
b
c
d f eg
h
i
j
had little (insertion of aBamHIlinker at nt -83) or no (deletion of nts -87/-83) effect on TSresponses but markedly decreased basal expression (Nachtigal et al., 1989). However, spatial constraints determine the T3 responsiveness of this region, because its placement downstream of the -83hCSp.CAT gene does not complement the promoter (Fig. 1F).Both the hGH and hCS upstream regions (nts -110/ -80) bind Ts receptors with high affinity. The strong correlation between the requirement of these DNA sequences for Ta responsiveness and their ability to bind the TS receptor DISCUSSION suggest that T3 receptor binding to this upstreamsite is These studies addressthe differential TSresponsiveness of involved in mediating the hormone effect. The decreased TSresponsiveness of the hGH promoterwas the nearly identical hGH and hCS 5'-FR. Confirming earlier found to be due to structural differences in the proximal results (CattiniandEberhardt, 1987), we found that the promoter for the hCS, but not the hGH, gene is positively region (nts -83/+2), since this DNA could not restore TS regulated by T S . We defined two separate regions of the hCS responsiveness when substituted for the hCS proximal pro5"FR that are required for T3 responsiveness and specific moter (Fig. 1B). Studies using oligonucleotides demonstrated sequence differences in the hGH promoter that account for two specific receptor-binding regions in both promoters cenits lack of T3 responsiveness. The two regions required for Ts tered approximately at nts-60 and -30. The -54/-48 region responsiveness bind the T g receptor and are localized up- is essential for mediating the TSresponse, since the deletion stream and downstream of the proximal cell-specific control of a G residue at nt -54 of the hCS promoter (absent in the element (nts -96/-62; Lefevre et al., 1987; Bodner and Karin, hGH promoter) or simultaneous conversion of G -P T at nt 1987; Bodner et al., 1988) that binds the transcription factor -50 and A + G at nt-48 of the hCS promoter to generate a more hGH-like promoter destroyed TSresponsiveness of the GHF-1. The upstream region required for T3responsiveness of the hCS promoter. Likewise, conversion of the hGH(-54/-48) hCS promoter was localized to nts -124/-87 (Fig. 1, A , C, region to thehCS sequence restores thyroidhormone responand D). The importantregion appears to extend just upstream siveness to the hGH promoter, indicating that no additional of nt -94, since the responsiveness of the -94hCSp.CAT gene structural differences between the hGH and hCS proximal is diminished, as compared with constructs extending further promoters contribute to theirdifferential TSresponsiveness. The functional and receptor binding data suggest that the upstream of nt -94. The hCS(-124/-87) region can be retwo separate receptor-binding sites the in -loo/-80 and -70/ placed by the equivalent -135/-87 region from the hGH -50 regions contribute to T3responsiveness of the hCS progene. This region contains the distal GHF-1/Pit-1-binding site (nts -128/-106) that may participate in the expression moter. These binding sites, which are separated by about 24 of the hGH and hCS genes in GC cells (Lefevre et al., 1987; bp (2.3 helical turns) do not require an absolute phase relaBodner et al., 1988). The T3-responsive element appears tobe tionship, since insertion/deletion mutations at nts -87/-83 separate from the proximal and functionally more important that introduce net changes of +16 and -4 bp between the two GHF-1-binding site (nts -96/-62), because mutations that structures do not abolish responsiveness. However, these two destroy this site (Lefevre et al., 1987; Nachtigal et al., 1989) elements must be maintained in close proximity to one an-
T3Responsiveness and
DNA Bending
919
known whether this reduced T3 responsiveness is due to disruption of the T3 receptor-binding site or to loss of the distal GHF-1 site. Thus, T3 responsiveness of the hCS gene is not dependent onGHF-1, although this factor may contribute to the degree of responsiveness. Sequences that aresimilar to the Spl consensus and that bind Spl or Spl-like factors are present in both the hGH andhCS genes (nts -144/-128, Lemaigre et al., 1989);these are not required for TBregulation, since this site is deleted in the -124hCSp.CAT gene, which retains full TSresponsiveness. Fig. 8 shows the alignment of T3 receptor-binding sites of the hGH, hCS, and several other thyroid hormone-responsive genes. TREs are proposed to contain about 20 nucleotides consisting of two hexanucleotide half-sites. The upstream consensus corresponds to thesequence (G/A)GGTCA and the downstream, more degenerate structure, corresponds to (G/ A)GG(A/C)CN, where N is any nucleotide. These consensus sequences are similar to the GGG(T/A)C proposed by Lavin et al. (1989) or AGGTCA proposed by Glass et al. (1987) and Koenig et al. (1987). However, it is noted thatthe best alignment of these naturally occurring sequences is a direct repeat, unlike the palindromic consensus proposed earlier (Glass et al., 1988). The two half-sites are separated by a spacer of 3-5 nucleotides, resulting in anet separation of one helical turn so that they lie on the same face of the DNA. Additional evidence that functional TREs arerepresented by direct repeat structures separated by a variable spacer has been presented by Umesono et al. (1991) and Naar et al., (1991). These studies also provide evidence that the variable spacer may allowdiscrimination by thyroid hormone, retinoic acid, and vitamin D3 receptors (Umesono et al., 1991) or thyroid hormone, retinoic acid, and estrogen receptors (Naar et al., 1991). As illustrated in Fig. 8, the hGH and hCS genes TRE POSITION (bp)
SEQUENCE
FIG. 7. Combined comparisons of several circular permutation analyses of the pCPCS,pCPGH, pCPCS-lX, and pCPCS-3X DNA fragments. A , pCPCS and pCPGH. B, pCPCS and pCPCS-1X. C, pCPCS and pCPCS-3X. The difference between the migration of the Suu3Al ( S )and the FokI ( F ) fragments of the pCPCS plasmid was set at 100% The migration of other fragments was measured relative to this andplotted as a percentof the maximal migration difference. The lower axis depicts the distance of the Ttreceptor binding site measured at themidpoint of the 30-bp inserted oligonucleotide from the Suu3Al site. The various DNA fragments are 140 bp in length. In the absence of receptor, the migration of the different DNA fragments was identical (data not shown). For all trials ( n = 4 or 51,the various comparisons (A-C) were always made on the same gel to avoid intergel variations. The individual DNA fragments were generated by cutting the circular permuted plasmids FokI ( F ) ,EcoRI ( E ) ,and Asp7181 ( A ) . with Suu3Al ( S ) ,PuuII (P),
other, since they do not function when separated by large distances. Our data provide some information about the potential participation of the cell-specific factor, GHF-1, and of Spllike factors in mediating TSresponsiveness of the hCS promoter. Ye et al. (1988) suggested that GHF-1 is required for TBstimulation of rGH promoteractivity. The proximal GHF1 binding site is not necessary for T3stimulation of the hCS gene, since mutationsthat abolish GHF-1 binding (Lefevre et al., 1987) at theproximal site (nts-96/-62) and that reduce basal promoter activity greater than 93% (Lefevre et al., 1987; Nachtigal et al., 1989) only diminish T3 responsiveness to a small extent. Furthermore,the distal GHF-1-bindingsite (nts -128/-110) in the hCS promoter is deleted in the -94hCSp.CAT gene, which is regulated by T3,albeit less well than other constructs containing more 5'-FR DNA. It is not
'HALF-SITE SPACING
GENE
11
hCS(UI-44):
" ATQQQQCCACTQACQQQCT " ATQQQQCCACTQACQQQCT " QTQQQQTCA : AGCAQQQAQAQA
hQH(43I-U):
" QTQQQQTCAACAQTQQQAQAQA
9,11
:O )sl-U I-h C S (' 'hQH(-UI-lW)
hCS(-%/-14): h~~(-~al-14):
-
"
CCAQQQTATAAAAAQQQCCCACA " CCAQQQTATAAAAAQQQCCCACA
11
10
11 11
ffiH(-lW-lC8):
AAXQATCAQcorcbGQC
11
ffiH(-W-16):
A-AAAAAQGZZQ
11
~ O H - W T C ~ ~ ~ W ~ ) :I T Q ~ CX ~Q CA Q :: A
e
+rm~~c(-ls1/-157):
~ ~ c o T o i ~ : ~ ~ ~ ~ c o a c10i ~ ~ A o c c c ~
fmMHC(-l42/-lW:
" QAQGTQAC:AQQAQGACA
rm(-ml*1b
CAI Q T A Q ~T :Q ~ ~ P ~ ~ C A " %oMLVLlR(S1)3p): CAGQGTCA: TTTCAQGTCCrr CONSENSUS :
:QQTCA
10 10 10
N,-N, ~ Q Q ~ C N
FIG. 8. Comparison of various nucleotide sequences from several genes that bind the TB receptor and contribute to thyroid hormone responsiveness. The locations of the sequences are designated by the numbers in parentheses. Half-site spacing refers to the distance between adjacent half-sites that form direct repeat structures. The sequences are aligned with respect to their left (upstream) half-sites, and colons are introduced to show the differences in half-site phasing. The weighted consensus sequence is shown at the bottom of the figure; N refers to any nucleotide. The abbreviations are: rGH (Norman et el., 1989), rat growth hormone; rGH-INT C (Sap et el., 1990) rGH intron C; raMHC (Izumo and Mahdavi, 1988) and haMHC(Flink and Morkin, 1990), rat andhuman myosin heavy chain ( a subunit), respectively; ra (Burnside et al., 1989), rat glycoprotein a subunit; MoMLV LTR (Sap et ul., 1989), murine Maloney leukemia virus long terminal repeat. *, direct repeats only; $, sequence of the lower strand.
920
T3 Responsiveness and DNA Bending
contain structures in the -108/-87, -65/-44, and -4O/-14 regions that specifically bind theT3receptor and arehomologous to these consensus TREs. Consistent with the functional interchangeability of the upstream hGH and hCS 5’FR (nts -135 to -87) is the fact that the two genes contain identical receptor binding regions in thissegment of DNA. The -64/-44 regions of the hGH andhCS promoters that account for their differential T3responsiveness have several sequence differences. If the sequences are aligned such that the spacer region of hGH contains 3 bp, then thespacer region of the hCS(-64/-44) region contains 4 bp due to an extraG and the downstream hGH hexanucleotide structure contains a T instead of a G at -50 and anA instead of a G at -48. As discussed above, deletion of a Gin thehCS spacer or changing sequence these hCS sequences at nts-50 and -48 to the hGH abolishes TI responsiveness of the hCS promoter. Since the G deletion is in the spacer and the-5O/-48 mutations are in the downstream hexanucleotide half-site, both of these regions of this hCS TRE must be important. Alternatively, the hGH and mutated hCS (CS-1X) structures in the -64/-44 region can be aligned with a 5-bp spacer separating the two hexanucleotide half-sites. This results in a perfect alignment of the GGGAGA half-sites, which are phase-shifted 11 and 12 nucleotides for the T3-unresponsive hGH and mutated hCS sequences, respectively, as compared with 10 nucleotides for the nativehCSstructure. Inthis alignment, theT3unresponsive hexanucleotide structures would lie on more opposed sides of the DNA helix than thenative hCS structure. This could result in impaired binding affinity or binding of the receptor in a nonfunctional manner. Forexample, if each hexanucleotide structure represents a Ta receptor monomerbinding site(Brent et al., 1989), then thealtered phasingmay impair important receptor-receptor interactions or interactions with other factors. The -64/-44 regions of the hGH andhCS promoters bind T3receptors, although the hCS sequence binds with a 4-fold higher affinity than the hGH sequence. Generally, such a small reduction in affinity for the receptor binding to the hGH sequence might be expected to reduce but not abolish the T, responsiveness. However, such small quantitative differences might reflect qualitatively important processes such as critical interactions between receptors and otherreceptors, other factors, or DNA. The notion that qualitative differences in receptor-DNA interactions of the two promoters may be important is supported by the finding that hGH(-87/+2)receptor complexes migrated faster in gel retardation assays than hCS(-87/+2)-receptor complexes (Fig. 5). These results could be explained by either receptor-induced differences in the conformation of the DNA or by the binding of another factor to thehCS promoter. To distinguish between these possibilities, we examined the effect of placement of the receptor-binding DNA elements within a larger piece of DNA on the mobility of the T1receptor DNA complexes (circular permutation analysis; Wu and Crothers (1984)). These studies demonstrated that the migration is a function of their placement within a DNA fragment. With the hCS(-70/-40) region, the retardationis greatest when the receptor-binding structure is in the center of the DNA fragment. It is intermediate when the sequence is off center, and it is least when the structure is on either end. Because the primary structures of these DNAs are identical except for the relative placement of the receptor-binding sequences, it is unlikely that these fragmentsbind to different factors. Insteadthe dataindicate that receptor-induced bending of the DNA is involved. The TI receptors also bend the Ta-unresponsive -7O/-40
region of the hGH 5’-FR, but much less than the equivalent hCS region. The observed bending of the two T3-unresponsive, mutated hCS(-70/-40) regions (CS-1X and CS-3X) is intermediate to the native hCS or hGH region, suggesting that the structural changes as reflected by the relative bending elicited by these mutations may be additive. These correlations suggest that the DNA bending is a requisite for T3 responsiveness of this promoter. Also, since the CS-1X and CS-3X mutations destroy thyroid hormone responsiveness but exhibit greater bending behavior than thehGH(-70/-40) region, it ispossible that the receptor-induced DNA bending must exceed some threshold value to achieve thyroid hormone responsiveness. It should be mentioned that these studies do not address the exact form of the receptor that is involved in the DNA binding and bending phenomena. These studies were performed with partiallypurified rat liver receptors that are now known to be associated with another protein of unknown function that enhances the receptors’ ability to bind to DNA, but does not appear to bind to DNA in the absence of the receptors or to affect the sequence-specificity for receptor binding to DNA (Ribeiro and Lavin, 1990). Whether this factor can influence the receptor-induced bending of DNA is not known; however, the requirement of the receptor for binding of this factor to DNA indicates that the receptor is required for DNA bending. Our finding of thyroidhormone receptor-induced DNA bending suggests that other members of the steroid/thyroid hormone family of receptors may share this property. The data to date do not provide information on this point, although Carballo and Beato (1990) recently reported that the glucocorticoid receptor induces a topological change in plasmids containing a murine mammary tumor virus GRE. The mechanism of this structural change was not addressed, and the results were not correlated with glucocorticoid receptormediated functions. The datacited in theintroduction suggest that DNA bending is induced relatively commonly by proteins that bind to DNA. These data plus the functional correlations in the current study and the previously cited studies that emphasize the functional importance of DNA bending with prokaryotic systems (Bracco et al., 1989; Goodman and Nash, 1989) indicate that furtherstudies of whether protein-induced DNA bending has functional significance are warranted. The mechanisms involved in DNA bending-associated activation of transcription are unknown. Shuey and Parker (1986) proposed thatbent DNA representsa recognition signal for DNA-binding proteins that slide along the DNA (one-dimensional diffusion) to find their sites of action (von Hippel et al., 1984; Park et al., 1982. In thismodel, bent DNA might cause RNA polymerase I1 to slow, thereby facilitating the formation of stable preinitiation complexes, possibly by promoting strand separation. DNA bending may promote protein-protein interactions that are required for trans-activation functions to become expressed. Theseinteractions might involve receptors and other transcription factors or RNA polymerase such as proposed for the phage repressor protein (Hochschild et al., 1983; Hawley and McClure, 1983) or could involve interactions among transcription factor subunits such as those of the X repressor (Hochschild and Ptashne, 1986). In this lattermodel, DNA deformability that is determined by sequences between two DNA-binding sites (Hogan and Austin, 1987) may play a significant role. Any of these models could accommodate DNA bending in the mechanism of steroid/thyroid hormone action. For example, receptor-induced conformation changes in the DNA could influence protein-protein interactions between receptor
T3 Responsiveness and DNA Bending subunits or between receptors and other proteins that are crucial for the expression of the tram-activationfunctions of the receptor. Such interactions could result in more stable receptor-DNA complexes. In fact, this may explain the 4-fold higher affinity of the hCS (-70/-40) region for the T S receptor compared with the hGH region. An additional possible consequence of such DNA deformation might be acquisition of a receptor structure thatis essential for the functioning of receptor-activating domains. In support of this concept, Tan and Richmond (1990) have recently shown that DNA binding induces a conformational change of PRTF, a yeasttranscriptional activator. Acknowledgments-We wish to thank Drs. Jim Apriletti and Tom Lavin for providing guidance and direction in the isolation of the T, receptor preparation and for many helpful discussions during the course of these studies.We are indebted toMary Craddock and Hayley Cranston-Shepard for preparation and editing of the manuscript. REFERENCES Apriletti, J. W., Baxter, J . D., and Lavin, T. N. (1988) J. Biol. Chem. 263,9409-9417 Bodner, M., and Karin, M. (1987) Cell 50,267-275 Bodner, M., Castrillo, J.-L., Theill, L. E., Deerinck, T., Ellisman, M., and Karin, M. (1988) Cell 55,505-518 Bracco, L., Kotlarz, D., Kolb, A., Dickinson, S., and Buc, H. (1989) EMBO J. 8,4289-4296 Brent, G.A., Harvey, J. W., Chen, Y., Warne, R. L., Moore, D. D., and Larsen, P. R. (1989) Mol. Endocrinol. 3, 1996-2004 Burnside, J., Darling, D. S., Carr, F. E., and Chin, W. W. (1989) J. Biol. Chem. 264,6886-6891 Carballo, M., and Beato, M. (1990) DNA Cell Bwl. 9, 519-525 Carson-Jurica, M. A., Schrader, W. T., and O’Malley, B. W. (1990) Endocr. Rev. 1 1 , 201-219 Cattini, P. A., and Eberhardt, N. L. (1987) Nucleic Acids Res. 15, 1297-1309 Cattini, P. A., Anderson, T. R., Baxter, J. D., Mellon, P., and Eberhardt, N. L. (1986) J. Biol. Chem. 261, 13367-13372 Cattini, P. A., Klassen, M., and Nachtigal, M. (1988) Mol.Cell. Endocrinol. 60, 217-224 Chen,E. Y., Liao, Y.-C., Smith, D.H., Barrera-Saldana, H. A., Gelinas, R., and Seeburg, P.H.(1989) Genomics 4,479-497 Dolle, P., Castrillo, J.-L., Theill, L. E., Deerinck, T., Ellisman, M., and Karin, M. (1990) Cell 60,809-820 Eastman, A. (1987) Biotechniques 5, 731-732 Evans, R. M. (1988) Science 240,889-895 Flink, I. L., and Morkin, E. (1990) J. Biol. Chem. 265, 11233-11237 Flug, F., Copp, R. P., Casanova, J., Horowitz, Z. D., Janocko, L., Plotnick, M., and Samuels, H. H. (1987) J. Biol. Chem. 262,63736382 Frederick, C.A., Grable, J., Melia, M., Samudzi, C., Jen-Jacobsen, L., Wang, B.-C., Greene, P., Boyer, H. W., and Rosenberg, J. M. (1984) Nature 309,327-331 Glass, C. K., Franco, R., Weinberger, C., Albert, V. R., Evans, R. M., and Rosenfeld, M. G. (1987) Nature 3 2 9 , 738-741 Glass, C. K., Holloway, J. M., Devary, 0. V., and Rosenfeld, M. G. (1988) Cell 54,313-323 Goodman, S. D., and Nash, H.A. (1989) Nature 341,251-254 Gorman, C. M., Rigby, P. W. J., and Lane, D. P. (1985) Cell 42,519526 Gustafson, T. A., Taylor, A., and Kedes, L. (1989) Proc. Natl. Acad. Sci. U. S. A. 8 6 , 2162-2166 Hagerman, P. J. (1990) Annu. Reu. Biochem. 59, 755-781 Hawley, D. K., and McClure, W. R. (1983) Cell 32, 327-333 Hemsley, A., Arnheim, N., Toney, M. D., Cortopassi, G., and Galas, D. J. (1989) Nucleic Acids Res. 1 7 , 6545-6551 Hirt, H., Kimelman, J., Birnbaum, J. J., Chen, E. Y., Seeburg, P. H., , Eberhardt, N. L., and Barta, A. (1987) DNA (N. Y . ) 659-70 Hochschild, A., and Ptashne, M. (1986) Cell 44, 681-687 Hochschild, A., Irwin, N., and Ptashne, M. (1983) Cell 32, 319-325 Hogan, M. D., and Austin R. H. (1987) Nature 329, 263-266
92 1
Isaacs, R. E., Findell, P. R., Mellon, P., Wilson, C. B., and Baxter, J. D. (1987) Mol. Endocrinol. 1, 569-576 Izumo, S., and Mahdavi, V. (1988) Nature 334,539-542 Koenig, R. J., Brent, G. A., Warne, R. L., Larsen, P. R., and Moore, D. D. (1987) Proc. Natl. Acad. Sci. U. S. A. 84, 5670-5674 Kolb, A., Spassky, A., Chapon, C., Blazy, B., and Buc, H. (1983) Nucleic Acids Res. 11, 7833-7852 Lavin, T. N., Baxter, J. D., and Horita, S. (1988) J. Bwl. Chem. 263, 9418-9426 Lavin, T. N., Norman, M. F., Eberhardt, N., and Baxter, J. D. (1989) in The Steroid/Thyroicl Hormone Receptor Family and Gene Regulation (Gustafsson, J. A., Eriksson, H., and Carlstedt-Duke, J., eds) pp. 69-81, Birkhaeuser Verlag, Basel Lefevre, C., Imagawa, M., Dana, S., Grindlay, J., Bodner, M., and Karin, M. (1987) EMBO J. 6, 971-981 Leidig, F., Baxter, J. D., and Eberhardt, N. L. (1990) Trans. Am. ASSOC.Phys. CIII, 154-162 Lemaigre, F. P., Courtois, S. J., Lafontaine, D. A., and Rousseau, G. G. (1989) Eur. J. Biochem. 181,555-561 Li, S., Crenshaw, E. B., 111, Rawson, E. J., Simmons, D. M., Swanson, L. W., and Rosenfeld, M. G. (1990) Nature 347, 528-533 Miller, W. L., and Eberhardt, N. L. (1983) Endocr. Rev. 4, 97-130 Mlodzik, M., Hiromi, Y., Weber, U., Goodman, C. S., and Rubin, G. M. (1990) Cell 60, 211-224 Naar, A. M., Boutin, J.-M., Lipkin, S. M., Yu, V. C., Holloway, J. M., Glass, C, K., and Rosenfeld, M. G. (1991) Cell 65, 1267-1279 Nachtigal, M. W., Nickel, B. E., Klassen, M. E., Zhang, W., Eberhardt, N. L., and Cattini,P. A. (1989) Nucleic Acids Res. 17,43274337 Neumann, J. R., Morency, C. A., and Russian, K. 0.(1987) Biotechniques 5,444-447 Nickel, B. E., and Cattini, P. A. (1991) Endocrinobgy 128, 23532359 Nickel, B. E., Kardami, E., and Cattini, P. A. (1990) Biochem. J . 267,653-658 Norman, M. F., Lavin, T. N., Baxter, J . D., and West, B. L. (1989) J. Biol. Chem. 264,12063-12073 Oro, A. E., Ong, E. S., Margolis, J . S., Posakony, J. W., McKeown, M., and Evans, R. M. (1988) Nature 336,493-496 Park, C. S., Wu, F. Y., and Wu, C. W. (1982) J. Biol. Chem. 257, 6950-6956 Ribeiro, R.C. J., and Lavin, T. N. (1990) Prog. & Abs. Endo. SOC. Ann. Mtg. 72,357 Richmond, T. J., Finch, J. T., Rushton, B., Rhodes, D., and Klug, A. (1984) Nature 311, 532-537 Robertson, C. A., and Nash, H. A. (1988) J. Biol. Chem. 263,35543557 Samuels, H., Stanley, F., and Casanova, J. (1979) Endocrinology 105, 80-85 Sap, J., Munoz, A., Schmitt, J., Stunnenberg, H., and Vennstroem, B. (1989) Nature 340, 242-245 Sap, J., de Magistris, L., Stunnenberg, H., and Vennstroem, B. (1990) EMBO J. 9,887-896 Schreck, R., Zorbas, H., Winnacker, E.-L., and Baeuerle, P. A. (1990) Nucleic Acids Res. 18,6497-6502 Schroth, G. P., Cook, G. R., Bradburg, E. M., and Gottesfeld, J. M. (1989) Nature 340, 487-488 Selby, M. J., Barta, A., Baxter, J. D., Bell, G. I., and Eberhardt, N. L. (1984) J. Biol. Chem. 259, 13131-13138 Shuey, D. J., and Parker, C. S. (1986) Nature 323, 459-461 Tan, S., and Richmond, T. J. (1990) Cell 62,367-377 Travers, A. A. (1990) Cell 60, 177-180 Umesono, K., Murakami, K. K., Thompson, C. C., and Evans, R. M. (1991) Cell 65,1255-1266 von Hippel, P. H., Bear, D. G., Morgan, W. D., and McSwiggen, J. A. (1984) Annu. Reu. Biochem. 53, 389-446 Wang, L.-H., Tsai, S. Y., Cook, R. G., Beattie, W. G., Tsai, M.-J., and O’Malley, B. W. (1989) Nature 3 4 0 , 163-166 Wood, W. M., Kao, M. Y., Gordon, D. F., and Ridgway, E. C. (1989) J. Biol. Chem. 264, 14840-14847 Wu, H.-M, and Crothers, D. M. (1984) Nature 308, 509-513 Ye, Z.-S., Forman, B. M., Aranda, A., Pascual, A., Park, H.-Y., Casanova, J., andSamuels, H. H. (1988) J. Biol. Chem. 263,78217829 Zhang, W. G., Brooks, R., Baxter, J. D., and Eberhardt, N. L. (1990) Clin. Res. 38, 410A
