Regulation of the action of steroid/thyroid hormone receptors by medium-chain fatty acids.

Triiodothyronine (T3) causes a 30-fold increase in transcription of the malic enzyme gene in chick embryo hepatocytes; medium-chain fatty acids (MCFAs) inhibit this increase. T3 action is mediated by T3 receptors (TRs) that bind to T3 response elements (T3REs) in this gene's 5'-flanking DNA. In transiently transfected hepatocytes, fragments of 5'-flanking DNA of the malic enzyme gene or artificial T3REs that conferred T3 stimulation also conferred MCFA inhibition to linked reporter genes. Thus, MCFA inhibition may be mediated through cis-acting T3REs and trans-acting TRs, distinguishing MCFA action from that of other fatty acids which act through unique sequence elements. Using binding assays and overexpression of TR, we showed that MCFAs inhibited the transactivating but not the silencing function of TR and did not alter binding of T3 to TR or of TR to T3RE. The C-terminal ligand-binding domain of TR was sufficient to confer stimulation by T3, but not inhibition by MCFA. Inhibition of transactivation by MCFA was specific: ligand-stimulated transcription from T3 or estrogen response elements was inhibited, but that from glucocorticoid or cyclic AMP response elements was not. We propose that MCFAs or metabolites thereof influence the activity of a factor(s) that interacts with the T3 and estrogen receptors to inhibit ligand-stimulated transcription.


Triiodothyronine (T 3 ) causes a 30-fold increase in transcription of the malic enzyme gene in chick embryo hepatocytes; medium-chain fatty acids (MCFAs) inhibit this increase. T 3 action is mediated by T 3 receptors (TRs) that bind to T 3 response elements (T 3 REs) in this gene's 5-flanking DNA.
In transiently transfected hepatocytes, fragments of 5-flanking DNA of the malic enzyme gene or artificial T 3 REs that conferred T 3 stimulation also conferred MCFA inhibition to linked reporter genes. Thus, MCFA inhibition may be mediated through cisacting T 3 REs and trans-acting TRs, distinguishing MCFA action from that of other fatty acids which act through unique sequence elements. Using binding assays and overexpression of TR, we showed that MCFAs inhibited the transactivating but not the silencing function of TR and did not alter binding of T 3

to TR or of TR to T 3 RE. The C-terminal ligand-binding domain of TR was sufficient to confer stimulation by T 3, but not inhibition by MCFA. Inhibition of transactivation by MCFA was specific: ligand-stimulated transcription from T 3 or estrogen response elements was inhibited, but that from glucocorticoid or cyclic AMP response elements was not. We propose that MCFAs or metabolites thereof influence the activity of a factor(s) that interacts with the T 3 and estrogen receptors to inhibit ligand-stimulated transcription.
Malic enzyme (EC 1.1.1.40) catalyzes the oxidative decarboxylation of malate to pyruvate and CO 2 , simultaneously generating NADPH from NADP ϩ . In the livers of well fed birds, much of the NADPH generated by this reaction is used for de novo fatty acid synthesis. Malic enzyme activity, like that of other lipogenic enzymes, is regulated by nutritional state (1,2). When starved chickens are fed or fed chickens are starved, malic enzyme activity, malic enzyme mass, malic enzyme mRNA abundance, and transcription of the malic enzyme gene increase or decrease by similar relative amounts and with time courses consistent with transcription being the primary regulated process (3,4). These nutritionally induced changes in the expression of the chicken malic enzyme gene can be mimicked quantitatively in chick embryo hepatocytes in culture (5). In the presence of insulin and triiodothyronine (T 3 ), 1 transcription of the chicken malic enzyme gene is high, as it is in the livers of fed chicks (6). In the absence of T 3 and insulin or in the presence of insulin and T 3 plus glucagon or cAMP or mediumchain fatty acids (MCFAs), transcription is low, as in the livers of starved birds (6,7). Several T 3 response elements (T 3 REs) have been localized in the 5Ј-flanking DNA of the chicken malic enzyme gene (8 -10). Characterization of these T 3 REs indicates that each contributes differentially to the overall response of the gene to T 3 (9).
Hexanoate (C6:0) and octanoate (C8:0) inhibit the T 3 -induced increase in transcription of the malic enzyme gene in chick embryo hepatocytes within 30 min of their addition; inhibition is reversible upon removal of fatty acid. Inhibition by MCFAs is selective; they have no effect on total transcription in isolated nuclei or on transcription of the genes for glyceraldehyde-3-phosphate dehydrogenase or ␤-actin. Inhibition is also specific; 4-and 10-carbon fatty acids and several modified fatty acids have little or no effect (7). These results suggest that the MCFA effect may be relevant biologically. What might that biological relevance be?
Levels of MCFAs in chicken plasma that are high enough to inhibit transcription have not been reported. During starvation, a process that inhibits transcription of the malic enzyme gene, however, fatty acid oxidation is increased. Concomitantly, production of hydroxylated fatty acids similar in chain length to the inhibitory MCFAs also is increased (11)(12)(13). In hepatocytes in culture, MCFAs may be converted to hydroxylated fatty acids or metabolites similar thereto, and these may be the intracellular mediators of the inhibition of transcription.
How might MCFAs inhibit transcription of the malic enzyme gene? Hydroxylated long-chain fatty acids activate transcription of the fungal gene for cutinase (14) by stimulating phosphorylation of a trans-acting factor that binds to a specific cis-acting enhancer element (15,16). Polyunsaturated longchain fatty acids (PUFAs) stimulate transcription of the acyl-CoA oxidase gene by interacting with the trans-acting factor, peroxisomal proliferator-activated receptor (PPAR) (17), or its adipocyte counterpart, fatty acid-activated receptor (FAAR) (18). These factors bind to peroxisomal proliferator response elements in the acyl-CoA oxidase gene. PUFAs also appear to inhibit expression of the S14 gene via a response element, the PUFA response element; the mechanism for this is not known (19,20). Long-chain fatty acids inhibit binding of T 3 to nuclear TR in cells in culture (21,22). Both long-chain fatty acids and their acyl-CoA derivatives inhibit binding of T 3 to rat liver TR in vitro (23). In intact hepatocytes, however, MCFAs fail to inhibit binding of T 3 to TR (7), suggesting that they act by some mechanism other than displacement of T 3 .
We report here that inhibition of T 3 -stimulated transcription of the chicken malic enzyme gene caused by MCFAs is mediated by the receptor that binds to T 3 REs. This mechanism is distinct from that of fatty acids that act through unique sequence elements. Rather than displacing T 3 from TR or displacing TR from T 3 RE, MCFAs inhibit the transactivation function of TR. MCFAs also inhibit estrogen-stimulated transcription through an estrogen response element and the estrogen receptor (ER), but have no effect on ligand-stimulated transcription mediated by glucocorticoid or cAMP response elements.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes were obtained from Life Technologies, Inc., New England Biolabs Inc., U. S. Biochemical Corp., or Boehringer Mannheim. Other enzymes were obtained from the indicated sources: RQ-DNase I (Promega), T4 polynucleotide kinase and Klenow fragment of Escherichia coli DNA polymerase I (Boehringer Mannheim), DNA polymerase from Thermus aquaticus (Promega), and Bst polymerase (Bio-Rad). Nucleotides were purchased from Sigma, Amersham Pharmacia Biotech, or Life Technologies, Inc. Radiolabeled nucleotides, Hyperfilm-HP, ECL Western blotting kits, and donkey antirabbit IgG conjugated to horseradish peroxidase (secondary antibody) were obtained from Amersham Pharmacia Biotech. D-threo-[dichloroacetyl-1,2-14 C]Chloramphenicol was purchased from NEN Life Science Products. LipofectACE, Waymouth MD 705/1 medium, and E. coli cells of strain DH5␣ were obtained from Life Technologies, Inc. SeaKem LE agarose, NuSieve GTG agarose, and DNA isolation columns (SpinBind) were purchased from FMC Corp. Hormones, fatty acids (as sodium salts), and heparin columns were purchased from Sigma. Polyclonal antibody to chicken TR␣ was purchased from Santa Cruz Biotechnologies (cross-reacts with chicken TR␤). Monoclonal antibody to chicken RXR was provided by Pierre Chambon (Université Louis Pasteur/ IGBMC, Illkirch, France). Nitrocellulose membranes were purchased from Millipore Corp.
Plasmids-Expression plasmids containing 5Ј-flanking DNA of the chicken malic enzyme gene inserted upstream of the CAT gene in pKSCAT were constructed as described (10). Synthesis of DNA constructs containing fragments of chicken malic enzyme DNA inserted into the multiple cloning site 5Ј of the promoter of the herpes simplex virus thymidine kinase (TK) gene in pBLCAT2 (24) also has been described (10). Plasmid DR4ϫ5-TKCAT was made by inserting five copies of annealed oligonucleotides encoding a direct repeat element with a 4-bp spacer (AGGTCAnnnnAGGTCA) into the multiple cloning site of pBLCAT2. Structures of plasmid DNAs were confirmed by restriction enzyme mapping and partial sequence analysis.
Construction of pRSV-Luc was described (10). Plasmid CMV-␤GAL (25) was obtained from Richard Maurer (Oregon Health Sciences University). Plasmid [MEϪ3474/ϩ31]CAT was provided by F. Bradley Hillgartner (West Virginia University). Marc Montminy (Salk Institute) provided pRSV-CREB. Bruno Luckow and Gunter Schutz (German Cancer Research Center, Heidelberg, Germany) provided pBLCAT2. The expression vector for GAL4, pSG424, contained sequences encoding amino acids 1-147 of the GAL4 DNA-binding domain. Plasmid SG424 and the expression plasmid for the GAL4-binding element, pMC110, were obtained from Mark Ptashne (Harvard University). Herbert H. Samuels (New York University) provided the cDNA for chicken TR␣ (cloned into the pET8c expression vector) and pGAL4-cTR␣ (containing the sequences encoding amino acids 120 -408 of the ligand-binding domain of chicken TR␣ cloned into pSG424). Ronald Evans (Salk Institute) and Bert W. O'Malley (Baylor College of Medicine) gave us pTREpal-TKCAT and pERE-TKCAT, respectively. Expression vector for human ER (pHEO) was obtained from Geoffrey L. Greene (University of Chicago). Expression plasmids for the GRE linked to TKCAT DNA (p⌬GTCO) and rat GR (pVARO) were gifts from Keith R. Yamamoto (University of California, San Francisco). Ganes Sen (Cleveland Clinic Foundation) provided pCRE-TKCAT.
Cell Culture and Transient Transfection-Livers of 19-day-old chick embryos were removed, chopped, and treated with collagenase (26). Isolated hepatocytes were separated from red blood cells; resuspended in Waymouth medium MD 705/1 supplemented with penicillin (60 g/ml), streptomycin (100 g/ml), insulin (50 nM), and corticosterone (1 M); and incubated in 35-mm tissue culture dishes in an atmosphere of 5% CO 2 in air at 40°C. Twenty hours after the cells were plated, they were transiently transfected using 40 g of LipofectACE/well. Each 35-mm dish was transfected with 5 g of plasmid DNA: p[MEϪ5800/ ϩ31]CAT (10.7 kilobase pairs, 2.5 g) or a molar equivalent of other test constructs, pCMV-␤GAL or pRSV-LUC (0.5 g), and pBluescript KS ϩ (balance). Two 35-mm plates were used for each experimental condition. After 24 h in the transfection medium, the medium was removed by aspiration and replaced with Waymouth medium supplemented with or without 1.6 M T 3 and with or without 1 or 5 mM sodium hexanoate (26). Both of these hexanoate concentrations caused similar degrees of inhibition (data not shown).
CAT, ␤-Galactosidase, and Luciferase Assays-Forty-eight hours after adding T 3 , the cells were harvested; lysed by three cycles of freezing and thawing; and analyzed for soluble protein content (27) and ␤-galactosidase (28) or luciferase (29) and CAT (30) activities. ␤-Galactosidase activity was determined by measuring the absorbance at 420 nm following incubation of cell lysates with o-nitrophenyl ␤-galactopyranoside. Luciferase activity was determined by measuring the light units emitted following incubation of cell lysates with luciferin. For CAT assays, samples of the cell lysates were heat-treated for 30 min at 60°C; denatured protein was removed by centrifugation. CAT activity was determined by incubating a portion of cell lysate with acetyl-CoA and [ 14 C]chloramphenicol for 15 h at 37°C. Incubation mixtures were then extracted with ethyl acetate and subjected to thin-layer chromatography. Conversion to the acetylated product was detected by liquid scintillation spectrophotometry or direct autoradiography using the Packard InstantImager.
Gel Electrophoretic Mobility Shift Assay-Each oligonucleotide probe contained a 5Ј-extension and was labeled by a fill-in reaction catalyzed by the Klenow fragment of E. coli DNA polymerase I. Other procedures and the preparation of nuclear extracts were described previously (9).
DNase I Footprint Analysis-Nuclear extracts were prepared from chick embryo hepatocytes incubated with insulin plus corticosterone plus T 3 , with or without hexanoate (1 mM). Other procedures were as described previously (10).
Statistical Analysis-Statistical significances of differences between matched pairs were determined by the Wilcoxon matched-pairs, signedrank test (31). S.E. values are provided to indicate the degree of variability in the data.

Identification of the Cis-acting Element Involved in Mediat-
ing Inhibition by MCFA-DNA constructs containing deletions from the 5Ј-end of the 5Ј-flanking DNA of the chicken malic enzyme gene were transiently transfected into chick embryo hepatocytes in culture in an effort to localize the inhibitory effect of MCFAs (hexanoate). Cells transfected with the longest construct (p[MEϪ5800/ϩ31]CAT) responded to T 3 with a 14fold increase in CAT activity, and hexanoate inhibited T 3stimulated CAT activity by 83% (Fig. 1). We have previously reported that 5800 bp of 5Ј-flanking DNA of the malic enzyme gene contained the sequence element(s) required for stimulation of transcription by T 3 (10). The sequence element(s) necessary for inhibition by hexanoate were contained in the same DNA.
Deletions to Ϫ5200, Ϫ4135, Ϫ3845, and Ϫ3474 bp resulted in 16-, 17-, 9-, and 7-fold stimulations by T 3 , respectively. The degree of inhibition by hexanoate was the same for cells transfected with each of these constructs. Cells transfected with constructs containing 5Ј-deletions to Ϫ2715 bp or to shorter end points did not respond to either T 3 or hexanoate. These results confirmed the location of T 3 response units (T 3 RUs) between Ϫ4135 and Ϫ3845 bp and between Ϫ3845 and Ϫ2715 bp (9, 10). The upstream T 3 RU contains four functional direct repeat T 3 REs at Ϫ3883 to Ϫ3858, Ϫ3833 to Ϫ3808, Ϫ3809 to Ϫ3784, and Ϫ3794 to Ϫ3769 bp; the 5Ј-most of these T 3 REs conferred 90% of the responsiveness of the entire upstream T 3 RU. The downstream T 3 RU contains a single T 3 RE at Ϫ3081 to Ϫ3056 bp (9, 10). Short internal deletions were made in two constructs (p[MEϪ5800/ϩ31]CAT and p[MEϪ4135/ϩ31]CAT) that were responsive to both T 3 and hexanoate. Each deletion removed the major functional T 3 RE in the upstream T 3 RU; the minor T 3 REs in the upstream and downstream T 3 RUs were not deleted. T 3 responsiveness decreased by ϳ50% for both deletion constructs, but the MCFA responses were unaffected. Thus, any construct that conferred a T 3 response also conferred inhibition by hexanoate, consistent with the possibility that T 3 REs themselves might confer inhibition by MCFA.
Constructs that contain the upstream T 3 RU linked to TK-CAT (p[MEϪ3903/Ϫ3617]TKCAT and p[MEϪ3903/Ϫ3703]TK-CAT) conferred their responses to T 3 via one major and three weak T 3 REs (9, 10). In cells transfected with these constructs, CAT activity was stimulated by T 3 and inhibited by hexanoate (Fig. 2). Deletion of part of the strongest T 3 RE (p[MEϪ3868/ Ϫ3617]TKCAT or p[MEϪ3868/Ϫ3703]TKCAT) caused a large reduction in T 3 responsiveness, but did not affect responsiveness to hexanoate. Deletions from the 3Ј-end of p[MEϪ3903/ Ϫ3703]TKCAT to Ϫ3733 or Ϫ3769 bp did not affect T 3 or hexanoate responses. Deletions to Ϫ3799, Ϫ3823, and Ϫ3863 bp gradually reduced responsiveness to T 3 from Ͼ100-fold to 54-fold, consistent with the gradual loss of weak T 3 REs; inhibition by hexanoate was unaffected. In this set of experiments, the vector, pTKCAT, conferred a 2-fold stimulation by T 3 and 39% inhibition by hexanoate; responses of cells transfected with pTKCAT were detectable in some batches of hepatocytes, but not others (cf. Figs. 3 and 4) and may have been transduced through a cryptic T 3 RE in pTKCAT (32). In those experiments in which pTKCAT-transfected hepatocytes had small but statistically significant responses to T 3 and hexanoate, cells trans-fected with constructs containing the malic enzyme DNA inserted into pTKCAT were not judged to be responsive to T 3 or hexanoate unless the responses were significantly greater than those of cells transfected with the vector alone. In those experiments in which pTKCAT-transfected hepatocytes had no statistically significant response to T 3 , hexanoate had no inhibitory effect. The results of this set of experiments remain consistent with the hypothesis that inhibition by hexanoate is conferred by T 3 REs and that inhibition was not localized to a specific T 3 RE within this region.
Cells were transfected with constructs containing block or deletion mutations to verify that the T 3 and hexanoate effects were inseparable. In two sets of experiments, p[MEϪ3903/ Ϫ3703]TKCAT was transfected into hepatocytes; T 3 caused 65and 113-fold increases in CAT activity, and hexanoate reduced the T 3 -induced activity by 67 and 65%, respectively (Figs. 3 and 4). Mutation of the first half-site (pMUT1[MEϪ3903/ Ϫ3703]TKCAT) did not alter significantly either stimulation by T 3 or inhibition by hexanoate, compared with that of the nonmutated parent construct (Fig. 3). Cells transfected with constructs containing block mutations in sites 2, 3, 4, 5, or 6 (individually) exhibited significantly reduced responsiveness to T 3 , but no change in inhibition by hexanoate (Fig. 3). Block mutations in either the upstream (pMUTB[MEϪ3903/Ϫ3703]-TKCAT) or downstream (pMUTC[MEϪ3903/Ϫ3703]TKCAT) half-site of T 3 RE2 or both (pMUTD[MEϪ3903/Ϫ3703]TKCAT) significantly reduced responsiveness to T 3 , but did not significantly alter responsiveness to hexanoate (Fig. 4). Similarly, deletion of T 3 RE2 decreased the T 3 response by 95%, but inhibition by hexanoate was still 49% (Fig. 4). In all of the experiments reported thus far, the effects of T 3 and hexanoate were inseparable, suggesting that any T 3 RE was sufficient for the hexanoate effect. Cells transfected with a single copy of the strong T 3  , and pBluescript DNA (sufficient to balance DNA/plate to 5.0 g) and treated with or without T 3 and with or without hexanoate (C6 (5 mM), except that 1 mM C6 was used in two experiments). Each point represents the mean Ϯ S.E. of five to eight independent sets of hepatocytes, using at least two independently prepared batches of each plasmid. CAT and ␤-galactosidase activities of extracts from T 3 -treated hepatocytes transfected with p[MEϪ5200/ϩ31]CAT were 1.6 Ϯ 0.5 percentage of conversion/15 h/g of protein and (4.4 Ϯ 1.8) ϫ 10 Ϫ4 A 420 units/min/g of protein, respectively. Left, DNA constructs used in these experiments; middle, effects of T 3 , expressed as -fold change in CAT activity caused by T 3 (ϩT 3 /ϪT 3 ); right, effects of hexanoate. Relative CAT activity of cells treated with T 3 and hexanoate is expressed as a percentage of that in cells treated with T 3 alone. Statistical significance between means within a column is indicated as follows: a, versus T 3 and hexanoate (Figs. 4 and 5), indicating that T 3 RE2 by itself is sufficient to confer both stimulation by T 3 and inhibition by hexanoate.
We next tested constructs containing artificial T 3 REs, the malic enzyme gene T 3 RU, and T 3 RE2 to determine if the response to hexanoate was specific to natural T 3 REs of the malic enzyme gene or could be mediated by any element that responded to T 3 (Fig. 5). Constructs containing T 3 RE2 (p[MEϪ3883/Ϫ3858]-TKCAT) or T 3 RE2 within the entire T 3 RU (p[MEϪ3903/ Ϫ3703]TKCAT) bestowed 120-and 130-fold responses to T 3 , respectively. These T 3 -induced activities were decreased to 35 and 37%, respectively, by hexanoate. Transfection of constructs containing either two copies of the artificial palindromic T 3 RE or five copies of a consensus direct repeat T 3 RE linked to TKCAT DNA conferred robust responsiveness to T 3 and inhibition by hexanoate (Fig. 5). These results suggest that the inhibitory effect of hexanoate is not limited to sequences from the malic enzyme gene; inhibition by hexanoate is transduced by several kinds of T 3 REs. Moreover, when sequences containing the T 3 RUs were inserted into TKCAT in reverse orientation or were linked to the minimal promoter of the malic enzyme gene (Ϫ147 to ϩ31 bp) in either orientation and transfected into hepatocytes, T 3 stimulation and hexanoate inhibition were conferred (data not shown). The minimal promoter of the malic enzyme gene did not confer responsiveness to T 3 or MCFA. Therefore, the T 3 REs that act as cis-acting elements for the

FIG. 5. Effects of T 3 and hexanoate on CAT activity in hepatocytes transfected with constructs containing different kinds and numbers of T 3 REs.
Chick embryo hepatocytes were transiently transfected as described in the legend to Fig. 1 and treated with or without T 3 and with or without hexanoate (C6, 1 mM). Left, DNA constructs used in these experiments. First four columns, relative CAT activities are expressed as described in the legend to Fig. 1; each value is the mean Ϯ S.E. of six to nine independent experiments using at least two independently prepared batches of each plasmid. Relative CAT activities were calculated by setting the CAT activities for T 3 -treated hepatocytes transfected with pTKCAT to 1.0 and adjusting all other activities proportionately. CAT and ␤-galactosidase activities of extracts from T 3 -treated hepatocytes transfected with pTKCAT were 0.21 Ϯ 0.04 percentage of conversion/15 h/g of protein and (7.1 Ϯ 1.3) ϫ 10 Ϫ3 A 420 units/min/g of protein, respectively. Fifth column, effects of T 3 are expressed as -fold change in CAT activity caused by T 3 . Sixth column, the effect of hexanoate is expressed as CAT activity in cells treated with T 3 plus hexanoate divided by that in cells treated with T 3 alone ϫ 100. The boxed region A is wild-type T 3 RE2. TREpal contains two copies of a palindromic T 3 RE, and DR4 contains five copies of a DR4 T 3 RE, both linked to TKCAT. Statistical significance between means within a column (p Ͻ 0.05) is indicated as follows: a, versus pTKCAT; b, versus pDR4-TKCAT. effect of hexanoate can vary greatly in their distance from the start site of transcription, gene of origin, and orientation with respect to the direction of transcription.
Identification of the Trans-acting Factor(s) Involved in Mediating Inhibition by MCFA -TR is the trans-acting factor common to all the T 3 REs that conferred MCFA inhibition in transfected cells, suggesting that it also may be the transacting factor that mediates the effects of MCFAs. TR acts as a silencer of transcription in the absence of T 3 (33). Mutation or deletion of the major functional T 3 RE (T 3 RE2) from sequences linked to either a natural or a heterologous promoter causes an increase in basal activity in transfected cells (10). This increase is consistent with loss of the silencing action of TR. Nuclear receptors such as PPAR and chicken ovalbumin upstream promoter transcription factor inhibit T 3 -stimulated transactivation by T 3 ; they also inhibit TR-mediated silencing of basal activity (20, 34 -36). Three constructs, each of which contained the major functional T 3 RE of the malic enzyme gene, were tested for an effect of MCFA on basal activity (Fig. 6). Plasmid [MEϪ3903/Ϫ3703]TKCAT contains the entire T 3 RU; pMUT1-[MEϪ3903/Ϫ3703]TKCAT contains a block mutation in a nonfunctional direct repeat upstream of the major T 3 RE of the upstream T 3 RU; and p[MEϪ3883/Ϫ3858]TKCAT contains a single copy of the major T 3 RE. When transfected into hepatocytes, all three constructs conferred robust responses to T 3 and strong inhibitory responses to hexanoate (Fig. 6). None of these constructs nor the vector itself conferred responsiveness to hexanoate in the absence of T 3 . This result suggests that the factor that mediates responsiveness to hexanoate does not mediate the silencing action of unliganded TR.
To transactivate, liganded TR must bind to a T 3 RE. Previous work suggested that MCFAs do not inhibit transactivation by disrupting binding of T 3 to TR (7), suggesting that TR binding to the T 3 RE or transactivation may be disrupted by MCFAs. DNase I footprint and gel mobility shift analyses were used to test the effect of hexanoate on binding of TR to the T 3 REs in the upstream T 3 RU of the malic enzyme gene. Bacterially expressed (37) and partially purified (0.3 g) TR bound to regions of the T 3 RU corresponding to T 3 RE2 and an extension of T 3 RE-3 (10). Proteins present in nuclear extracts prepared from chick embryo hepatocytes in culture treated with T 3 and/or hexanoate for 24 h showed the same protection in these two regions (data not shown). Hexanoate added directly to binding reactions had no effect on binding of TR or other nuclear proteins, indicating that these fatty acids did not have a direct effect on binding of TR to the T 3 RE. Gel mobility shift analyses indicated that, in nuclear extracts from cells treated with T 3 , RXR/TR heterodimers bound to the major functional T 3 RE in a manner similar to that of those in nuclear extracts made from hepatocytes treated with T 3 plus hexanoate (data not shown). Thus, hexanoate does not disrupt binding of TR to the T 3 RE, nor does it disrupt the interaction between TR and its heterodimerization partner, RXR.
If inhibition by hexanoate were mediated by a factor that competed with TR or RXR/TR dimers for binding to the T 3 RE, inhibition by hexanoate should be decreased in cells that overexpress TR. We tested p[MEϪ3868/Ϫ3703]TKCAT and p[MEϪ3474/ϩ31]CAT for responsiveness to T 3 and hexanoate in the absence and presence of overexpressed TR␣ (Fig. 7). Cells transfected with these constructs have higher basal CAT activities than those transfected with constructs containing an intact T 3 RE2 or the entire upstream T 3 RU. Overexpression of chicken TR␣ caused 62-and 100-fold increases in T 3 responsiveness of cells transfected with p[MEϪ3868/Ϫ3703]TKCAT and p[MEϪ3474/ϩ31]CAT, respectively. Hexanoate inhibited the response to T 3 to a similar extent whether or not chicken TR␣ was overexpressed. As noted earlier, hexanoate had no effect on CAT activity in the absence of T 3 . Thus, the factor that mediates the MCFA response may not compete with TR for binding to the T 3 RE.
Many TR-associated proteins regulate transactivation by liganded TR (38 -40); MCFAs could inhibit transactivation by interacting with such TR-associated proteins. Some of these proteins also interact with ER, but do not interact with GR. To gain insight into the specificity of action by MCFA and to narrow the potential list of factors that might mediate inhibition by MCFA, we tested the effects of hexanoate on cells transfected with constructs that bestow responsiveness to estrogen and glucocorticoids. Single copies of elements that bind these receptors were linked to TKCAT DNA and transfected into hepatocytes with and without overexpression of the cognate receptor. Plasmid ERE-TKCAT conferred 30-and 37-fold responses to ␤-estradiol in the absence and presence of overexpressed ER, respectively (Fig. 8). Plasmid GRE-TKCAT conferred 220-and 160-fold responses to corticosterone without and with overexpression of GR, respectively. Therefore, endogenous forms of both ER and GR were functional on these cis-acting elements. Hexanoate inhibited function of the ERE, but not the GRE, with or without overexpression of the cognate receptors. A hormone-activated transcription factor that does FIG. 6. Effects of T 3 and hexanoate on CAT activity in hepatocytes transfected with constructs containing wild-type and mutant versions of a T 3 response region and wild-type T 3 RE2. Chick embryo hepatocytes were transiently transfected as described in the legend to Fig. 1 and treated with or without T 3 (1.6 M) and with or without hexanoate (C6, 1 mM). Left, DNA constructs used in these experiments. First four columns, relative CAT activities are expressed as described in the legend to Fig. 1; each value is the mean Ϯ S.E. of seven independent experiments using at least two independently prepared batches of each plasmid. Relative CAT activities were calculated by setting the CAT activities for T 3 -treated hepatocytes transfected with p[MEϪ3903/Ϫ3703]TKCAT to 1.0 and adjusting all other activities proportionately. CAT and ␤-galactosidase activities of extracts from T 3 -treated hepatocytes transfected with p[MEϪ3903/Ϫ3703]TKCAT were 11.1 Ϯ 3.3 percentage of conversion/15 h/g of protein and (9.5 Ϯ 3.9) ϫ 10 Ϫ3 A 420 units/min/g of protein, respectively. Fifth column, effects of T 3 are expressed as -fold change in CAT activity caused by T 3 . Sixth column, the effect of hexanoate on CAT activity is expressed as CAT activity in cells treated with T 3 plus hexanoate divided by that in cells treated with T 3 alone ϫ 100. The boxed region A is wild-type T 3 RE2; x indicates a mutation upstream of T 3 RE2 (MUT1; see Fig. 3) that has little or no effect on T 3 responsiveness. No statistically significant differences were observed between means for samples lacking T 3 , with or without hexanoate. not belong to the steroid/thyroid hormone receptor superfamily also was tested. The cis-acting element was a CRE linked to TKCAT DNA. Cells transfected with the CRE construct showed 10-and 23-fold responses to CPT-cAMP (a nonmetabolizable analog of cAMP) with and without overexpressed CRE-binding protein (CREB ), respectively; they did not respond to hexanoate. These results suggest that inhibition by hexanoate has specificity with respect to the involved transcription factor. The actions of TR and ER are inhibited; those of GR and CREB (or other CRE-binding protein) are not inhibited. This distinction may be due to the interaction of TR and ER with a factor that mediates the effect of MCFA; GR and CREB may not interact with that factor.
Members of the family of proteins that interact with TR and ER, but not GR, bind to the ligand-binding domains of these receptors (40). These proteins were identified in yeast two- Fifth column, effects of ligand are expressed as -fold change in CAT activity caused by ligand. Sixth column, the effect of hexanoate on CAT activity is expressed as CAT activity in cells treated with ligand plus hexanoate divided by that in cells treated with ligand alone ϫ 100. Constructs containing single copies of the estrogen, glucocorticoid, and cAMP response elements were obtained as described under "Experimental Procedures." Statistical significance between means within a row is indicated as follows: a, versus control without hexanoate (p Ͻ 0.05). hybrid screens using the ligand-binding domain (with no Nterminal or DNA-binding domains) of TR or ER as "bait." The ligand-binding domain of TR, fused to the DNA-binding domain of GAL4, was transfected into hepatocytes with a plasmid that encodes GAL4-binding sites (Fig. 9). In this model, T 3 -stimulated CAT activity is due to binding of T 3 to the hybrid GAL4-TR and not to endogenous TR (Fig. 9). We used p[MEϪ3903/Ϫ3703]TKCAT as a positive control; it conferred both T 3 and hexanoate responsiveness. In cells cotransfected with the negative control, pGAL4 (pSG424), and GAL4-CAT, CAT activity was unaffected by T 3 or hexanoate. Cotransfection of pGAL4-cTR␣, the ligand-binding domain of chicken TR␣ fused to pGAL4, with pGAL4-CAT resulted in a robust response to T 3 , indicating that the truncated TR was capable of binding T 3 and transactivating the linked promoter. Cells transfected with the GAL4-TR fusion did not respond to hexanoate. These results suggest that the N-terminal sequence of TR is required for inhibition by hexanoate. They also confirm that binding of T 3 to TR is not disrupted by hexanoate.

DISCUSSION
The cis-acting elements in the 5Ј-flanking DNA of the chicken malic enzyme gene that confer inhibition by MCFAs co-localized with the T 3 REs in this DNA. Furthermore, artificial palindromic and direct repeat T 3 REs also conferred inhibition by MCFA. MCFAs also inhibited transactivation from an ERE, but had no effect on basal activity in the absence of T 3 or estradiol. Thus, the trans-acting factor that binds to T 3 REs or EREs and participates in the inhibitory effect of MCFA is TR or ER, respectively. The MCFA-regulated factor did not compete with TR for binding to the T 3 RE, but probably interacted with TR to influence its ability to transactivate linked promoters. Sequences N-terminal to the ligand-binding domain of TR are required for the action of MCFA, presumably because they interact with a MCFA-regulated factor. T 3 REs are found in many nutritionally regulated genes; T 3 stimulates and MCFAs inhibit transcription of both malic enzyme and fatty acid synthase (7). Modulation of the transactivation function of TR bound to a T 3 RE may represent an effective mechanism for modulation of T 3 -stimulated transcription of lipogenic genes. MCFAs themselves are unlikely to be physiological regulators of the lipogenic genes in chicks because there is no evidence that plasma concentrations of MCFA in chicks ever reach the required levels. Nevertheless, the specificity of the response with respect to chemical structure, its selectivity with respect to the genes affected, and the rapid onset and reversal of its effects (Ref. 9 and data not shown) are consistent with physiological importance. MCFAs or a compound derived therefrom may be similar in structure to the true inhibitor, the production of which may signal the starved state.
Modulation of the transactivation function of TR without displacement of T 3 from TR or of TR from the T 3 RE has not been described previously as a mechanism by which fatty acids regulate transcription. Long-chain fatty acids, specifically polyunsaturated fatty acids, regulate transcription of some eucaryotic genes by binding to or modifying regulatory proteins, which, in turn, bind to unique DNA elements in the 5Ј-flanking regions of these genes (17)(18)(19)(20). Hydroxy fatty acids also regulate transcription through a unique cis-acting element (14 -16). By contrast, our results suggest that MCFAs differ from these other fatty acids in that their inhibitory actions are transduced through the T 3 RE⅐TR⅐T 3 complex, rather than by a complex unique to MCFAs.
Proximal promoter elements may play important roles in T 3 -regulated expression of some genes (41). This is unlikely to confound the interpretation of our results. First, T 3 REs linked to both TK and minimal malic enzyme promoters are inhibited by hexanoate; the two promoters have quite different structures (10). Second, both estrogen-and T 3 -stimulated transcription is inhibited by MCFA. Third, MCFAs have no effect on promoter activity in the basal state. Fourth, if hexanoate did regulate activity of a factor bound at a site common to both TK and malic enzyme promoters and that factor mediated both estrogen-and T 3 -regulated gene transcription, then one would expect fractional inhibition by hexanoate to decrease as T 3 responsiveness decreased; it did not.
The superfamily of steroid/thyroid hormone receptors is subdivided into two major classes. Class I includes GR and ER; these receptors are cytosolic until bound by ligand. Class II includes TR, RXR, and PPAR, receptors that are nuclear in the absence or presence of ligand (42). The interaction of estrogen with ER triggers the translocation of the complex to the nucleus, where estrogen-bound ER binds to an ERE present in the DNA (43). Because both TR and ER are affected by MCFA, it seems unlikely that MCFA acts cytosolically to disrupt the interaction of estrogen and ER, preventing ER from reaching the nucleus. The mechanisms by which MCFAs inhibit estrogen-and T 3 -induced transcription are probably the same.
Inhibition of the action of ER by MCFA is consistent with the hypothesis that RXR and PPAR, a known heterodimerization partner for TR at T 3 REs of the malic enzyme gene and a potential partner for TR, respectively, are not targets for MCFA because ER dimerizes with itself and not RXR or PPAR (42,43). Overexpression of TR should have increased the number of TR/TR dimers bound to T 3 REs. Overexpression of TR, however, failed to alter inhibition by MCFA, additional evidence that potential heterodimerization partners such as RXR FIG. 9. Effects of T 3 and hexanoate on CAT activity in hepatocytes transfected with chimeric GAL4-cTR␣ and a construct containing a GAL4-binding site. Chimeric GAL4-cTR␣ and the construct containing the GAL4-binding site were obtained as described under "Experimental Procedures." Chick embryo hepatocytes were transiently transfected as described in the legend to Fig. 1 and treated with or without T 3 (1.6 M) and with or without hexanoate (C6, 1 mM). pGAL4-DBD (pSG424) and pGAL4-DBD/TR-LBD were cotransfected (0.2 g each per plate) with the pGAL4-CAT (pMC110) reporter construct. Left, DNA constructs used in these experiments. First four columns, relative CAT activities are expressed as described in the legend to Fig. 1; each value is the mean Ϯ S.E. of six independent experiments using at least two independently prepared batches of each plasmid. Relative CAT activities were calculated by setting CAT activities for T 3 -treated hepatocytes transfected with p[MEϪ3903/Ϫ3703]TKCAT (T3RUTKCAT) to 100 and adjusting all other activities proportionately. CAT and ␤-galactosidase activities of extracts from T 3 -treated hepatocytes transfected with p[MEϪ3903/Ϫ3703]TKCAT were 22.7 Ϯ 2.3 percentage of conversion/15 h/g of protein and (6.5 Ϯ 1.3) ϫ 10 Ϫ3 A 420 units/min/g of protein, respectively. Fifth column, the effect of hexanoate on relative CAT activity is expressed as relative CAT activity in cells treated with T 3 plus hexanoate divided by that in cells treated with T 3 alone ϫ 100. and PPAR did not mediate the inhibitory action of MCFA. Furthermore, fatty acids with carbon chain lengths shorter than 10 are poor activators of PPAR (44). The ligand-binding domains of both ER and TR interact with TR-interacting proteins (TRIPs); TRIPs do not interact with GR (41). The specificity of the inhibitory action of MCFAs for ER or TR, but not GR, suggested that one or more of these TRIPs might be the target for MCFA; modification or displacement of a TRIP could decrease transactivation by ER or TR. TRIPs, however, bind to the ligand-binding domain of TR or ER, and this part of TR was not sufficient to confer inhibition by MCFA even though it was sufficient for T 3 responsiveness. TRIPs thus seem unlikely to be the MCFA-regulated factor that interacts with TR or ER.
T 3 responsiveness of hepatocytes transfected with constructs containing all or parts of the T 3 RU, mutant versions of T 3 REs, or artificial T 3 REs varied from Ͻ5-fold to Ͼ1500-fold. These differences in responsiveness are due to novel proteins that bind to the different T 3 REs and differentially influence responsiveness (9), differences in binding of RXR/TR to T 3 REs of different sequences (42), and artificial overexpression of TR. Despite the resulting broad range of promoter activities, MCFA always resulted in inhibitions of 50 -80%. Within individual sets of experiments, the range for inhibition was even narrower, despite a wide range in T 3 responsiveness (e.g. Fig. 2). Thus, MCFAs may modify function of a factor that has the same relative regulatory effect on RXR/TR heterodimers bound at any T 3 RE, whether the intrinsic ability of that T 3 RE to transactivate a linked promoter is large or small.
MCFA action appears to require that part of TR that is N-terminal to the ligand-binding domain because the N-terminal (A/B) and DNA-binding domains were absent from the GAL4-TR chimera that responded to T 3, but not MCFA. MCFA did not interfere with the T 3 RE-TR or RXR-TR interactions, suggesting that the DNA-binding domain and the N-and Cterminal regions required for heterodimerization were not targets for interaction with the factor that mediates the effect of MCFA. This leaves the N-terminal A/B domain of TR as a putative interaction site for the factor that mediates action of MCFA. A number of TR accessory factors have been identified and are required for full transactivation of the T 3 RE-bound receptors with which they interact. The TR accessory factors that have been characterized interact with activation function domains in the ligand-binding region of TR and other members of the superfamily of steroid/thyroid hormone receptors (45)(46)(47)(48)(49)(50). Because the MCFA effect does not appear to be mediated by this region of TR, the known TR accessory factors are not promising candidates for factors regulated by MCFA. Analogous factors may interact with the N terminus, and MCFAs may target one or more of these factors, thus eliciting its inhibitory response.