Overexpression of the a-Thyroid Hormone Receptor in Avian Cell Lines EFFECTS ON EXPRESSION OF THE MALIC ENZYME GENE ARE SELECTIVE AND CELL-SPECIFIC*

The role of the a-thyroid hormone receptor (TRa) in regulation of transcription of the gene for chicken malic enzyme was analyzed in fibroblast cell lines normally unresponsive to triiodothyronine (T3). The gene for this transcription factor was introduced stably and overexpressed using a replication-competent retro- viral vector. In chick embryo fibroblasts (CEF), overexpression of TRa decreased malic enzyme activity by 90% in the absence of T3. Addition of T3 almost com- pletely restored malic enzyme activity to the level of similarly treated control CEF infected with virus lack- ing TRa. These TRa-induced changes in malic enzyme activity were mediated by alterations in transcription of the malic enzyme gene. Similar results were ob- tained when transcriptional activity of TRa was analyzed using a transient co-transfection system. Thus, the unliganded TRa is a transcriptional repressor of the malic enzyme gene; binding of T3 to the receptor abolishes this repression. In contrast, stable overexpression of TRa in QT6 cells had no effect on malic enzyme expression in the absence or presence of TB. Nuclear T3 binding was equally high in CEF and QT6 cells overexpressing TRa. These findings suggest that cell-specific factors control the ability of TRa to regulate the malic enzyme gene. Overexpression of TRa in CEF had no effect on the expression of fatty acid synthase and acetyl-coA carboxylase, lipogenic enzymes that are stimulated by T3 in hepatocytes in culture. Thus,

The role of the a-thyroid hormone receptor (TRa) in regulation of transcription of the gene for chicken malic enzyme was analyzed in fibroblast cell lines normally unresponsive to triiodothyronine (T3). The gene for this transcription factor was introduced stably and overexpressed using a replication-competent retroviral vector. In chick embryo fibroblasts (CEF), overexpression of TRa decreased malic enzyme activity by 90% in the absence of T3. Addition of T3 almost completely restored malic enzyme activity to the level of similarly treated control CEF infected with virus lacking TRa. These TRa-induced changes in malic enzyme activity were mediated by alterations in transcription of the malic enzyme gene. Similar results were obtained when transcriptional activity of TRa was analyzed using a transient co-transfection system. Thus, the unliganded TRa is a transcriptional repressor of the malic enzyme gene; binding of T3 to the receptor abolishes this repression.
In contrast, stable overexpression of TRa in QT6 cells had no effect on malic enzyme expression in the absence or presence of TB. Nuclear T3 binding was equally high in CEF and QT6 cells overexpressing TRa. These findings suggest that cell-specific factors control the ability of TRa to regulate the malic enzyme gene. Overexpression of TRa in CEF had no effect on the expression of fatty acid synthase and acetyl-coA carboxylase, lipogenic enzymes that are stimulated by T3 in hepatocytes in culture. Thus, gene-specific factors also may control the transcriptional activity of TRa.
Malic enzyme catalyzes the oxidative decarboxylation of malate to pyruvate and Con, simultaneously generating NADPH from NADP. This reaction is the primary source of reducing equivalents for de nouo fatty acid biosynthesis in chicken hepatocytes (1). Malic enzyme is subject to nutritional and hormonal regulation. In newly hatched unfed chicks, malic enzyme activity is low; feeding a high-carbohydrate mash diet stimulates enzyme activity about 70-fold (2). The diet-induced increase in malic enzyme activity can be mimicked in chick embryo hepatocytes in culture. Triiodothy-ronine (T3)l stimulates a 40-fold or greater increase in malic enzyme activity in this culture system (3). Insulin has no effect by itself, but amplifies the effect of T3 by 3-fold. Glucagon, acting through cyclic AMP, almost completely blocks the effects of T3 or insulin plus T3. All of these dietand hormone-induced changes in enzyme activity are due to alterations in the rate of synthesis of the enzyme which, in turn, are correlated with changes in mRNA abundance for malic enzyme, indicating that regulation is pretranslational (4). Alterations in the rate of transcription initiation, as measured by nuclear run-on assay, account for most of the changes in abundance of malic enzyme mRNA (5, 6). Transcription of the chicken malic enzyme gene is not responsive to diet or hormones in non-hepatic tissues indicating that tissue-specific factors are involved in the regulation of this gene (5).
Determination of the mechanisms responsible for the nutritional, hormonal, and tissue-specific regulation of malic enzyme transcription initially requires the identification of the trans-acting factors that interact with cis-acting regulatory elements. High-affinity nuclear T3 receptors synthesized in reticulocyte lysates bind specifically to a thyroid hormone response element (T3RE) in the 5'-flanking DNA of the rat malic enzyme gene in vitro (7). In addition, the T3-induced stimulation of transcription of malic enzyme in chick embryo hepatocytes is both rapid (11 h) and insensitive to inhibitors of protein synthesis (6). Together, these findings indicate that a nuclear T3 receptor(s) directly interacts with the malic enzyme gene to modulate its transcription rate.
The function of individual factors involved in the hormonal regulation of transcription can be analyzed via transfection analyses. One approach is to introduce and overexpress genes for transcription factors in cells that are unresponsive to hormones. The activity of the factor is determined by monitoring the expression of an endogenous regulated gene if permanent transfection is employed or a "reporter" gene ligated to the promoter/regulatory region of a regulated gene if transient co-transfection is used. This strategy assumes that the diminished hormonal responsiveness of the cells employed in the assay is caused, in part, by a lack of activity of the transcription factor being tested. By employing multiple cell types, this approach can be used to assess the role of cellspecific factors in modulating the activity of a transcription factor.
In the present study, we have analyzed the regulation of the gene for malic enzyme by the a-subtype of the chicken nuclear T3 receptor (TRa), the predominant form of the The abbreviations used are: TS, 3,5,3'-triiodo-~-thyronine; TRa, nuclear T, receptor, a subtype; CEF, chick embryo fibroblasts; CAT, chloramphenicol acetyltransferase; TsRE, thyroid hormone response element; kb, kilobase(s). 12299 receptor expressed chick embryo hepatocytes and fibroblasts (8,9). Both permanent transfection and transient co-transfection methods have been used. Retroviral vector-mediated gene transduction was employed to introduce stably and overexpress TRa in avian cell lines in which malic enzyme is normally unresponsive to hormones. Three cell types were employed to analyze the role of cell-specific factors in modulating TRa activity. The effect of exogenous TRa on expression of other T3-responsive lipogenic enzymes was examined to assess the role of gene-specific factors in regulation of nuclear T3 receptor function. Finally, the gene regulatory activity of TRa was compared with that of v-erbA, a mutated version of the receptor which comprises one of the two oncogenic loci in avian erythroblastosis virus (10).

EXPERIMENTAL PROCEDURES
Cell Culture-Chick embryo fibroblasts (CEF) were obtained from SPAFAS, Inc. (Norwich, CT). Quail QT6 cells were kindly provided by C. Moscovici (Gainesville, FL) (11). Both cell lines were routinely cultured in DMEM/M199 (Dulbecco's modified Eagle's medium (25 mM glucose) and Medium 199 (GIBCO/Bethesda Research Laboratories (BRL)) in a 1:l ratio (v/v), containing 10,000 units/liter penicillin G, 10 mg/liter streptomycin sulfate, and 25 pg/liter amphotericin B) supplemented with 5% fetal bovine serum. Primary cultures of chick embryo hepatocytes were prepared as described previously (12) and maintained in serum-free Waymouth medium MD705/1 containing 50 nM insulin (gift from Eli Lilly Corp.). Incubation of all cell types was at 40 "C in a humidified atmosphere of 5% COZ and 95% air. Where indicated, thyroid hormone was removed from fetal bovine serum by treatment with AG-1X-8 ion exchange resin (13).
Construction of a Retroviral Vector Expressing the Gene for Chicken TRa-A Hind111 restriction fragment (1250 base pairs) containing the coding sequence of the chicken TRa (c-erbAa)' gene was excised from the plasmid, pFlA (kindly provided by B. Vennstrom) (14), and subcloned into the adaptor plasmid, , to form CLA-12-c-erbAa. c-erbAa and polylinker sequences in this plasmid are flanked by ClaI sites. The ClaI restriction fragment of CLA-12-c-erbAa was then subcloned into the replication-competent avian retroviral vector, RCAS (subgroup A)(15), at the ChI site to form RCAS-c-erbAa. Orientation of the insert was determined by digestion with restriction endonucleases. The supercoiled forms of the plasmids, RCAS and RCAS-c-erbAa were purified twice by CsCl gradient centrifugation (16) and used to transfect cells.
Stable Introduction of TRa and v-erbA into Avian Cell Lines-CEF and QT6 cells were transfected with RCAS-c-erbAa using lipofectin (17). Based on reverse transcriptase activity (18) in the medium, the extent of infection of cells by recombinant retrovirus was maximal after 8-12 days of culture. Mass cultures of infected CEF were employed in experiments. Individual clones of QT6 cells infected with RCAS-c-erbAa virus were isolated via dilution on 96-well plates. QT6 cell lines employed in this study were derived from two rounds of clonal isolation.
The replication-incompetent retroviral vector, XJ12 (kindly provided by J. Samarut)(21), was used to introduce and express v-erbA in CEF. Mass cultures of XJ12-infected CEF were produced by cotransfection with pRAV-2 helper virus DNA. Control cells were infected with TXN3' which is similar to XJ12 except that it lacks v-erbA sequence. XJ12-and TXN3"infected CEFs were selected in regular growth medium containing 200 pg/ml G418 (GIBCO/BRL).
Malic enzyme activity (19) and protein (20) were measured by the indicated methods.
Isolation of RNA and Quantitation of mRNA Leuek-RNA was extracted from cells by the guanidinium thiocyanate/phenol/chloroform method (22). Total RNA was treated with formaldehyde and subjected to electrophoresis in 0.9% agarose gels. The separated RNAs were transferred to "Genescreen" (Du Pont-New England Nuclear) and hybridized with 32P-labeled DNA probes labeled by "random priming" (Amersham Multiprime DNA Labeling Kit) according to the manufacturer's instructions. Membranes were hybridized and washed as described (23). Washed filters were subjected to autoradiography at -70 "C with Kodak XAR-5 film and intensifying screens.
The proto-oncogene, c-erbAa, encodes the same protein as the gene for chicken TRa. The former designation is used in naming plasmids that contain this gene.
Exposed films were scanned at 633 nm in an LKB Ultrascan XL densitometer.
Quantitation of Nuclear T3 Receptor Concentration-Nuclear binding of 1Z51-T3 was measured in intact cells as described by Samuels et al. (24) with minor modifications. CEF and QT6 cells were incubated 16 h with DMEM/M199 supplemented with 5% fetal bovine serum depleted of thyroid hormone. The medium was removed, and the monolayers were washed three times with serum-free DMEM/M199. The cells were incubated with 1.2 nM 1251-T3 (Du Pont-New England Nuclear) in serum-free medium for 4 h. The assay measures total T3 receptor levels because this concentration of T3 saturates more than 90% of the receptors. After the incubation nuclei were isolated as described (25). Nonspecific binding was determined by incubating cells with a 1000-fold excess of nonradioactive T3. Nonspecific binding was subtracted from total binding to obtain specific binding. Radioactivity was measured in a y spectrometer. DNA was assayed by the method of Labarca and Paigen (26).
Nuclear T3 binding was measured in chick embryo hepatocytes 24 h after the cells were placed into culture. The same procedure described for CEF and QT6 cells was used except that incubation prior to the binding assay was in serum-free medium. The medium was changed prior to addition of 1Z51-T3.
Nuclear Run-on Assay of Transcription Rates-Nuclei were isolated from CEF by the method of Milstead et al. (27) with the modification that the nuclear storage buffer was 50% glycerol, 50 mM HEPES, pH 7.9, 75 mM NaC1, 0.1 mM EDTA, 5 mM dithiothreitol, and 0.125 mM phenylmethylsulfonyl fluoride. The nuclear run-on assay has been described previously (28).
DNA Probes-A near full-length cDNA for duck malic enzyme (pDME1) has been described (29). Chicken cDNAs for fatty acid synthase (30), glyceraldehyde-3-phosphate dehydrogenase (31), 0actin (32), and TRa (14)  . The cDNA probe for chicken acetyl-CoA carboxylase was generated by amplification of sequences 6012-6964 (33) by the polymerase chain reaction using a cDNA copied from total chicken liver RNA with avian reverse transcriptase. The genomic DNA probes for chicken malic enzyme used in nuclear runon assays were subclones of genomic DNA that had been isolated from bacteriophage X libraries 95, 6). M.E.-4.8-5' (4.8 kb in pUC 19) is an EcoRI fragment from X clone 20B and is derived from the most 5' intron of the malic enzyme gene. M.E.-4.8-3' (4.8 kb in pUC 19) is an EcoRI fragment of X clone 2B and contains exon 8 from the middle of the mRNA as well as surrounding intronic DNA. M.E.-2.3 (2.3 kb in M13 mp18) is a HindIII-EcoRI fragment of A clone 1 and contains intron DNA and exon 12 from the 3' third of the mRNA.
Transient Transfection Experiments-CEF and QT6 cells were seeded on T-75 flasks and grown in DMEM/M199 containing 5% fetal bovine serum until 70% confluent. Twenty-four hours before transfection, medium was changed to DMEM/M199 containing 5% fetal bovine serum depleted of thyroid hormone. The cells were incubated in this medium throughout the experiment. Cells were transfected with plasmid DNA using the calcium phosphate method (34). Transfected DNAs were pME5.8-CAT (test/reporter plasmid), pCMV-@galactosidase (internal transfection standard) (35), pRSVchicken-c-erbAa (thyroid hormone receptor expression vector provided by H. Samuels) (36) and pRS-v-erbA (v-erbA expression vector provided by K. Damm and R. Evans) (37). pUC-19 DNA was added as necessary to maintain a constant amount of DNA in each transfection. Exposure to the calcium phosphate/DNA precipitate was for 5 h followed by a 30 s glycerol shock. Twelve hours later, the transfected cells were trypsinized and distributed to 100 x 20-mm tissue culture plates. Hormones were added to the medium at this time. After 48 h of incubation, cells were harvested and extracts were prepared for CAT (38, 39) and @-galactosidase (16) assays. CAT activity was expressed relative to P-galactosidase activity to correct for differences in transfection efficiency between samples.

Development of Avian Cell Lines That Stably Overexpress
Chicken TRa-The replication-competent retroviral vector, RCAS-c-erbAor, was used to introduce stably and overexpress the chicken TRa gene in CEF and QT6 cells. To determine Regulation of the Malic Enzyme Gene by the T3 Receptor 12301 whether CEF and QT6 cells integrated the intact provirus, total RNA from cells was isolated and subjected to Northern analysis. Transcription of the integrated recombinant provirus should produce three RNA species; all originate in the 5'long terminal repeat, which contains a promoter/enhancer, and terminate in the 3'-long terminal repeat, which provides a polyadenylation signal. The full-length genomic RNA transcript is co-linear with the integrated provirus and serves as the mRNA for the gag and pol genes. Alternative processing results in separate subgenomic transcripts which serve as the mRNAs for the enu and c-erbAa genes. Hybridization of RNA to a c-erbAa cDNA probe detected three transcripts of sizes similar to those expected for the genomic RNA (8.5 kb), env mRNA (4.3), and c-erbAa mRNA (2.0 kb) expressed from the intact provirus (data not shown).
RNA was also hybridized to a gag cDNA probe that should detect only genomic RNA transcripts derived from the provirus. In CEF infected with RCAS-c-erbAa, gag cDNA hybridized to a single transcript (8.5 kb) identical in size to the genomic RNA transcript detected by the c-erbAa probe, suggesting that most of the proviruses in these cells did not contain major insertions or deletions (data not shown). In RNA from QT6 cells infected with RCAS-c-erbAa, however, gag hybridized to two transcripts (8.5 and 7.2 kb). The larger transcript was identical in size to the genomic RNA transcript that hybridized to c-erbAa and probably was transcribed from the intact provirus. The smaller transcript did not hybridize to c-erbAa and probably originated from a provirus in which the entire c-erbAa insert had been deleted.
To obtain separate sets of QT6 cells that integrated only intact provirus, or provirus lacking the c-erbAa insert, individual cells from the mixed population of RCAS-c-erbAainfected QT6 cells were isolated and cloned. Two QT6 cell lines (clones 3 and 4) expressed a predominant viral genomic RNA transcript that hybridized to both gag and c-erbAa. The sizes of the genomic transcript and the two subgenomic mRNAs that hybridized to c-erbAa were similar to those expected from expression of the intact retroviral transcription unit (data not shown). QT6 cell clones 3 and 4, therefore, contained intact undeleted proviruses and were named RCASc-erbAa-QT6-3 and RCAS-c-erbAa-QT6-4, respectively. Two other QT6 cell lines (clones 1 and 2) did not express exogenous c-erbAa sequences. In these cells, gag hybridized to a single transcript (7.2 kb) that was similar in size to the genomic RNA transcript observed in CEF infected with the parent vector RCAS (data not shown). Thus, QT6 cell clones 1 and 2 contained proviruses from which the c-erbAa insert had been deleted and were named RCAS-QT6-1 and RCAS-QT6-2, respectively.
Effects of Overexpression of TRa on Malic Enzyme Activity-CEF infected with RCAS or RCAS-c-erbAa were incubated in serum-free medium for 2 days to deplete endogenous thyroid hormone. This was followed by a 4-day incubation with or without added hormones in medium supplemented with thyroid hormone-depleted fetal bovine serum (5%). In the absence of hormones, malic enzyme activity in RCAS-c-erbAa-CEF was only 9% of that in similarly treated RCAS-CEF ( Table 11). Incubation of either cell type with insuIin (50 nM) had no effect on malic enzyme activity. Addition of T3 (1.5 phi) caused an 11-fold increase in enzyme activity in RCAS-c-erbAa-CEF. This contrasts with a 1.4-fold stimulation of enzyme activity by T3 in RCAS-CEF. In both cell types, addition of insulin plus T, had the same effect on enzyme activity as treatment with Ta alone. Malic enzyme activity in uninfected CEF subjected to the above hormonal manipulations was similar to that in RCAS-infected CEF (data not shown).
In a separate experiment, we determined the kinetics of the increase in malic enzyme activity caused by T3. In both cell types, activity appeared to approach a new steady state about 4 days after T, was added (data not shown). After 6 days with TB, malic enzyme activity in RCAS-c-erbAa-CEF was 78% of that in similarly treated RCAS-CEF. The concentration of TB that resulted in maximum malic enzyme activity (5 nM) was the same in both RCAS-and RCAS-c-erbAa-CEF (data not shown).
In the preceding experiment, an effect of insulin on malic enzyme activity, either by itself or in combination with T,, may have been masked by the presence of insulin or a factor(s) with insulin-like activity in fetal bovine serum depleted of thyroid hormone. The effects of insulin on malic enzyme activity in RCAS-or RCAS-c-erbAa-CEF was also explored in cells incubated with or without added hormones in serumfree medium. Incubation of RCAS-erbAa-CEF with insulin plus TB caused a significantly greater increase (11-fold) in malic enzyme activity than treatment with TS alone (7-fold) ( Table 11). Thus, under serum-free conditions, insulin amplified the effect of T3 on malic enzyme activity in RCAS-erbAa-CEF but had little effect on malic enzyme activity by itself. Serum depleted of thyroid hormones must contain a factorb)

TABLE I1
Effects of overexpression of TRa on malic enzyme activity in chick embryo fibroblasts incubated in the absence or presence of hormones Chick embryo fibroblasts infected with RCAS or RCAS-c-erbAa were incubated in serum-free medium for 2 days to deplete endogenous thyroid hormone. This was followed by a 4-day incubation with or without hormones in serum-free medium or medium supplemented with thyroid hormone-depleted fetal bovine serum (5%). The media were changed to ones of the same composition 2 days after the start of hormone treatment. Cells were harvested and malic enzyme activity (19)

I11
Effects of over-expression of the c-erbAafnuclear T3 receptor on malic enzyme activity in QT6 cells incubated in the absence or presence of hormones Individual clones of retrovirus-infected QT6 cells and a mixed population of uninfected QT6 cells were incubated in serum-free medium for 2 days. The medium was changed to one of the same composition supplemented with thyroid hormone-depleted fetal bovine serum (5%) and hormones were added. After 2 days of incubation, cells were harvested and malic enzyme activity was measured. The experiment presented in this table is representative of three that were performed. Values are means f S.E. ( n = 3). The concentrations of the hormones are the same as in Table 11. Effects of overexpression of TRa on malic enzyme activity in QT6 cells were investigated using the experimental design that utilized serum depleted of thyroid hormone (Table 111).

RCAS-c-erbAn
Malic enzyme activity in RCAS-c-erbAa-QT6-3 and RCASc-erbAa-QT6-4 incubated in the absence of added hormones was increased 31 and 37%, respectively, relative to RCAS-QT6-1 or RCAS-QT6-2. These small differences in enzyme activity between the various cloned cell lines could be due to overexpression of TRa or clonal variation. When comparing malic enzyme activity (no hormones) of a mixed population of uninfected QT6 cells with that of RCAS-c-erbAa-QT6-3 or RCAS-c-erbAa-QT6-4, no differences were observed, suggesting that overexpression of TRa in QT6 cells did not alter enzyme activity. Malic enzyme activity was unresponsive to Ts treatment in both RCAS and RCAS-c-erbAa QT6 cell lines (Table 111).
mRNA Levels-RCAS-and RCAS-c-erbAa-CEF were preincubated for 2 days in serum-free media followed by incubation with insulin or insulin plus TS in medium containing thyroid hormone-depleted serum (5%). The abundance of malic enzyme mRNA in RCAS-c-erbAa-CEF, incubated with insulin for 24 h, was decreased by 90% relative to similarly treated RCAS-infected CEF (Fig. 1). Addition of T3 for 24 h stimulated 1.3-and 5-fold increases in malic enzyme mRNA in RCAS-and RCAS-c-erbAa-CEF, respectively, compared dance of the mRNAs for malic enzyme, fatty acid synthase, acetyl-coA carboxylase, glyceraldehyde-3-phosphate dehydrogenase, and &actin in chick embryo fibroblasts incubated in the absence or presence of Ts. Chick embryo fibroblasts infected with RCAS or RCAS-c-erbAa were incubated for 2 days in serum-free medium. This was followed by incubation with insulin (50 nM) or insulin plus T, (1.5 p~) in medium supplemented with thyroid hormone-depleted fetal bovine serum (5%). After 24 and 48 h of incubation, total RNA was isolated and subjected to Northern analysis as described under "Experimental Procedures." RNA (20 pg/lane) was hybridized to "'P-labeled cDNAs for malic enzyme, fatty acid synthase, acetyl-coA carboxylase, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 0-actin, and c-erbAa. This experiment was conducted a total of three times with similar results. The sizes (in kilobases) of the mRNAs were: malic enzyme, 2.1; fatty acid synthase, 10.8 and 12.2; acetyl-coA carboxylase, 12.0; &actin, 2.0; glyceraldehyde-3-phosphate dehydrogenase, 1.3; and c-erbAa, 2.0.
with corresponding cells treated with insulin alone. Similar results were obtained after 48 and 96 h of hormone treatment (data not shown for 96 h). Levels of malic enzyme mRNA in RCAS-c-erbAa-CEF after 24 h with insulin plus Ts were about 50% of those in similarly treated RCAS-CEF. Differences in malic enzyme mRNA levels between different CEF cell lines (Fig. 1) were correlated with similar differences in malic enzyme activity (Table 11), indicating that the effects of TRa Regulation of the Malic Enzyme Gene by the T3 Receptor overexpression on malic enzyme activity in CEF were pretranslational.
The abundance of c-erbAa mRNA transcribed from RCASc-erbAa was unaffected by T3 treatment for 24, 48, and 96 h ( Fig. 1, data not shown for 96 h). Thus, the T3-induced increase in malic enzyme expression in RCAS-c-erbAa-CEF was not due to alterations in TRa expression. Malic enzyme mRNA levels in RCAS-c-erbAa-QT6 cell lines incubated in the absence of hormones were about 1.5fold higher than those of similarly-treated RCAS cell lines (data not shown). Concentrations of malic enzyme mRNA in cells not treated with hormones were similar among RCASc-erbAa-QT6 cell lines and a mixed population of uninfected QT6 cells. Treatment with insulin plus TI had no effect on the abundance of malic enzyme mRNA in any of the QT6 cell lines. These results are consistent with malic enzyme activity measurements (Table 111).

Specificity of the Effects of Overexpression of TRa in CEF-
In chick embryo hepatocytes, T3 stimulates the accumulation of mRNAs for acetyl-coA carboxylase and fatty acid synthase without affecting the abundance of mRNAs for glyceraldehyde-3-phosphate dehydrogenase and @-actin (40, 41, 42). Incubation of RCAS-or RCAS-c-erbAa-CEF with insulin plus Ts for 24 or 48 h had no effect on the mRNA levels of acetyl-CoA carboxylase, fatty acid synthase, glyceraldehyde-3-phosphate dehydrogenase, and B-actin relative to corresponding cells treated with insulin alone (Fig. 1). The concentrations of these mRNAs also were similar between CEF cell types. Thus, effects of overexpression of TRa are selective for malic enzyme.
In RCAS-QT6 cell lines, T3 treatment had no effect on fatty acid synthase, acetyl-coA carboxylase, glyceraldehyde-3-phosphate dehydrogenase, and @-actin mRNA levels (data not shown). Similar results were observed for RCAS-c-erbAa-QT6 cell lines with the exception that T3 treatment caused a 2-fold stimulation in acetyl-coA carboxylase mRNA abundance (data not shown).
Transcription-To determine the mechanism by which overexpression of TRa controlled malic enzyme mRNA levels, transcriptional activity was measured using the nuclear runon assay and a DNA probe from 3' region on the malic enzyme gene (M.E.-2.3). Transcription of the malic enzyme gene in RCAS-c-erbAa-CEF incubated with insulin for 1 or 24 h was 20 or 15%, respectively, of that of similarly treated RCAS-CEF (Fig. 2). Treatment of RCAS-c-erbAa-CEF with insulin plus TI for 1 or 24 h stimulated malic enzyme transcription by 3or 4-fold, respectively, relative to cells of the same genotype treated with insulin. Transcription of the malic enzyme gene in RCAS-CEF was not affected by T3. In RCASc-erbAa-CEF incubated with insulin plus TI for 1 or 24 h, rates of transcription of the malic enzyme gene were about 80 or 60%) respectively, of those in similarly-treated RCAS-CEF. Results obtained from run-on assays using genomic DNA probes spanning the 5' (M.E.-4.8-5') and middle (M.E.-4.8-3') regions of the malic enzyme gene were similar to those described above using M.E.-2.3 (data not shown). Rates of transcription of the fatty acid synthase, P-actin, and glyceraldehyde-3-phosphate dehydrogenase genes were similar in the different cell types and were not altered by T3.
To confirm that the effects of exogenous TRa on malic enzyme expression were mediated by changes in transcription and to examine further the mechanisms of cell type-specific expression of the malic enzyme gene, we measured the promoter/regulatory activity of 5'-flanking DNA from the malic enzyme gene in a transient transfection system. pME5.8-CAT is a chimeric DNA comprised of a continuous piece of DNA of genes for malic enzyme, fatty acid synthase, @-actin, and glyceraldehyde-3-phosphate dehydrogenase in chick embyro fibroblasts incubated in the absence or presence of Ts. Chick embryo fibroblasts infected with RCAS or RCAS-c-erbAn were incubated for 2 days in serum-free medium. This was followed by incubation with insulin (50 nM) or insulin plus Ta (1.5 p M ) in medium supplemented with thyroid hormone-depleted fetal bovine serum (5%). After 1 and 24 h of incubation, cells were harvested, nuclei isolated, and transcription run-on assays performed as described under "Experimental Procedures." Independent experiments were performed a t each time of hormone treatment. Identical strips were hybridized with equal amounts of "P-labeled nascent RNA (35 X IO6 and 25 X 1 0 ' cpm for 1 and 24 h, respectively) from nuclei of fibroblasts treated with or without To. The probe for malic enzyme ( M E ) was a genomic DNA. The probes for fatty acid synthase (FAS), p-actin, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were cDNAs. X DNA (X) was used as a negative control. Autoradiography for malic acid and X was for 21 days at 1 h of hormone treatmentJ4 days a t 24 h of hormone treatment. Autoradiography for FAS, @-actin, and GAPDH was for 3 days at 1 h of hormone treatment, 4 days a t 24 h of hormone treatment. Experiments were carried out three times with similar results. from the malic enzyme promoter/regulatory region (-5800 to +32) coupled to the gene for chloramphenicol acetyltransferase (CAT). T3 stimulates a 30-fold increase in expression of transiently transfected pME5.8-CAT in chick embryo hepa-tocyte~.~ The extent of this regulation is similar to T3-induced changes in the transcription of the endogenous malic enzyme gene in these cells (6). In RCAS-CEF transfected with pME5.8-CAT, incubation with insulin plus T3 for 48 h had no effect on CAT activity relative to cells of the same genotype treated with insulin alone (Fig. 3A). In contrast, insulin plus T3 for 48 h stimulated a 2.7-fold increase in CAT activity in RCAS-c-erbAa-CEF. The magnitude of the effects of stably expressed TRa on pME5.8-CAT activity in CEF were not as great as the changes detected by the nuclear run-on transcription assay. This may have been caused by a depletion of transcription factors due to the uptake of multiple copies of pME5.8-CAT per cell. To test this hypothesis, pME5.8-CAT was co-transfected with pRSV-chicken-c-erbAa, an expression plasmid for TRa (37). When pME5.8-CAT was cotransfected with 10 pg of pRSV-chicken-c-erbAa, the T3induced stimulation in CAT activity in RCAS-CEF increased to 50-fold (Fig. 3A). Similar results were observed for RCASc-erbAa-CEF. In a dose-response experiment, transfection of 210 pg of pRSV-chicken-c-erbAa produced maximal T3-induced changes in CAT activity in both cell lines (data not shown). Increases in the T3-induced stimulation of CAT activity caused by co-transfection with pRSV-chicken-c-erbAa were due to a reduction in basal CAT activity in cells incubated in the absence of T3. The results from these transient expression experiments are consistent with those of the nuclear run-on transcription assay and confirm that TRa-induced changes in malic enzyme mRNA levels in CEF primarily result from alterations in transcription of the malic enzyme gene.
In RCAS-QT6-1 and RCAS-c-erbAa-QT6-4 transfected S. A. Klautky and A. G. Goodridge, unpublished results. Twenty-four hours before transfection, subconfluent cultures of CEF (A and C) and QT6 cells ( B ) infected with RCAS or RCAS-c-erbAa were incubated in medium supplemented with thyroid hormone-depleted fetal bovine serum (5%). Cells were transfected with pME5.8-CAT (20 pg), pCMV-0-galactosidase (20 pg), and pRSVchicken-c-erbAa (0 or 10 pg) using calcium phosphate co-precipitation (35); for details see "Experimental Procedures." In C, RCAS-infected CEF were transfected with pRS-v-erbA (0 and 10 pg) instead of pRSV-chicken-c-erbAa. After transfection, cells were incubated with insulin (50 nM) or insulin plus T3 (1.5 pM) for 48 h in medium supplemented with thyroid hormone-depleted serum (5%). Cells were then harvested, extracts prepared, and CAT (39,40) and 0-galactosidase (16) assays performed. CAT activity was expressed relative to p-galactosidase activity to correct for differences in transfection efficiency. Within a given cell line, CAT activity was normalized relative to that of cells transfected with 0 pg of pRSV-chicken-c-erbAa (A and B ) or pRS-v-erbA (C) and incubated with insulin alone. Actual values for RCAS-CEF (6.4 f 1.9) and RCAS-c-erbAa-CEF (6.9 f 1.9) in A, RCAS-QT6-1 (1.7 ? 0.4) and RCAS-c-erbAa-QT6-4 (2.6 f 0.5) in B, and RCAS-CEF (1.3 f 0.1) in C transfected with 0 pg of expression plasmid and incubated with insulin are indicated in the parentheses. In A and B, values are means f S.E. of three to six experiments. In C, values are means f S.D. ( n = 2). with pME5.8-CAT, incubation with insulin plus T3 for 48 h had no effect on CAT activity relative to cells of the same genotype treated with insulin alone (Fig. 3B). These results are consistent with the data for malic enzyme activity (Table  111) and mRNA levels (data not shown), indicating a lack of effect of stable overexpression of TRa on malic enzyme expression in QT6 cells. When pME5.8-CAT was co-transfected with 10 pg of pRSV-chicken-c-erbAa, incubation with insulin plus T3 caused a 5-and 2.5-fold increase in CAT activity in RCAS-QT6-1 and RCAS-c-erbAa-QT6-4, respec-tively, relative to corresponding cells treated with insulin alone. Transfection of 210 pg pRSV-chicken-c-erbAa produced the maximal TB-induced increase in CAT activity in both cell lines (data not shown). Thus, transient expression of T R a is able to confer T3 responsiveness on a co-transfected malic enzyme promoter/reporter gene in QT6 cells. The magnitude of this effect, however, was only 5-10% of that observed in similar co-transfection experiments using CEF (Fig. 3A).
Effects of v-erbA on Malic Enzyme Expression-The v-erbA gene is a mutated version of the chicken T R a gene that participates in the neoplastic transformation of cells by the avian erythroblastosis virus (10). The v-erbA protein lacks the ability to bind TS but retains the wild-type receptor's ability to bind to specific DNA sequences (43,44). Unlike the wild-type receptor, the v-erbA protein is reported to be a constitutive repressor of genes induced by thyroid hormones (37,(45)(46)(47). We asked whether expression of v-erbA constitutively inhibited malic enzyme expression in CEF. The gene for v-erbA was stably introduced and expressed in CEF using the retroviral vector XJ12. Northern analysis confirmed the expression of v-erbA mRNA in XJ12-CEF, the abundance of which was similar to that of the provirally derived c-erbAa transcript in RCAS-c-erbAa-CEF (data not shown). When XJ12-CEF were incubated in serum-free medium for 2 days followed by 4 days of incubation with insulin (50 nM) in medium supplemented with thyroid hormone-depleted fetal bovine serum, malic enzyme activity was unchanged relative to that of similarly treated control cells infected with virus lacking v-erbA sequence (42 k 2 versus 38 f 2 milliunits/mg protein, mean * S.E., n = 3, respectively). Treatment with T3 caused similar increases in enzyme activity in XJ12-CEF (29%) and control CEF (33%). Thus, stable expression of v-erbA in XJ12-CEF had no effect on malic enzyme expression.
We next determined whether transient expression of v-erbA was able to modulate the activity of the malic enzyme promoter, since transient expression of T R a was more effective in regulating malic enzyme promoter activity than stable expression of this transcription factor. The v-erbA expression plasmid, pRS-v-erbA, was transiently transfected into RCAS-CEF and effects on pME5.8-CAT activity were determined. In cells transfected with 10 kg of pRS-v-erbA and incubated with or without T3 for 48 h, pME5.8-CAT activity was reduced 50% relative to similarly treated cells not transfected with v-erbA expression plasmid (Fig. 3C). Transfection with 20 pg of pRS-v-erbA caused a 75% decrease in pME5.8-CAT activity in the absence or presence of TB ( n = 1, data not shown).
Thus, transient expression of v-erbA repressed malic enzyme promoter activity in a hormone-independent manner. This contrasts with the transcriptional repressor activity of the wild-type receptor which is dependent on the absence of hormone.

DISCUSSION
We have developed a retroviral vector-based stable expression system to analyze the mechanism by which the TRa regulates malic enzyme expression. This system has several advantages over a transient transfection system in the functional analysis of trans-acting factors. First, due to the high efficiency of infection, the activity of the stably overexpressed factor can be monitored by simply measuring the activity of a responsive endogenous gene. In contrast, due to the low efficiency of transfection, the activity of a transiently expressed factor must be assessed by measuring the activity of a co-transfected reporter gene coupled to a responsive promoter. Expression of an endogenous gene is a more physiological indicator of factor activity than expression of a pro-moter in a reporter plasmid because the former is in a normal chromatin context whereas the latter is not. The high efficiency of retroviral infection also permits the measurement of the average level of expression of the exogenous transacting factor per infected cell without the need to correct for efficiency of gene transfer. Such a correction is necessary in transient transfection assays in order to determine the average level of expression per transfected cell. Last, integration of recombinant provirus into the host genome is limited to one to three copies/cell(15), whereas in transient transfection assays uptake and expression of foreign DNA may involve 100 copies/cell or more. Thus, the lower level of overexpression (per cell) achieved using a retroviral vector-based stable expression system decreases the probability of assaying nonphysiological effects of a trans-acting factor.
Using this stable expression system in CEF, we demonstrated that TRa had two different activities that contributed to the T3-mediated regulation of malic enzyme expression. First, in the absence of T3, overexpression of TRa repressed basal malic enzyme activity. Second, T3 partially relieved this inhibition resulting in an elevation of malic enzyme expression. Both repression and T3-dependent de-repression/activation were mediated by changes in transcription. The T3induced stimulation (de-repression) of malic enzyme transcription was rapid (11 h), indicating that this response was mediated by the direct interaction of the ligand-occupied TRa with the malic enzyme gene. These findings are consistent with the mechanism of action of the nuclear T3 receptor proposed by . In their model, the nonliganded nuclear T3 receptor interacts constitutively with a T3RE and functions as a repressor. Binding of T3 to the DNA-bound receptor triggers a conformational change that relieves the repressor activity and reveals an activation function. The foregoing model is based on experimental data from transient transfection assays. Our results provide the first evidence that this model applies to endogenous T3-regulated genes.
Comparison of basal and T3-stimulated malic enzyme activities in CEF with those in chick embryo hepatocytes indicates that impaired T,-responsiveness in the former cell type is caused by a decrease in both the transcriptional repressor and T3-dependent de-repressor/activation activities of the endogenous nuclear T3 receptor (Table 11). Malic enzyme activity in the absence of hormones is 6-fold higher in RCAS-CEF than in similarly treated chick embryo hepatocytes. T3induced stimulation of malic enzyme activity in RCAS-CEF is about 3% of that in chick embryo hepatocytes. Overexpression of TRa in CEF restores basal malic enzyme activity to the same low level observed in chick embryo hepatocytes and partially unmasks transcriptional activation by T3. Diminished activity of the endogenous nuclear T3 receptor in CEF is not due to decreased nuclear T3 receptor concentration, because RCAS-CEF exhibit similar levels of nuclear T3 binding as chick embryo hepatocytes (Table I). Thus, it is the specific transcriptional activity per unit T3 binding activity of the endogenous nuclear T3 receptor that is decreased in CEF.
The above observations suggest that cell-specific factors are involved in regulating the transcriptional activity of the nuclear TI receptor. This hypothesis is supported by the finding that stable overexpression of TRa had different effects on malic enzyme expression in CEF and QT6 cells. Stable overexpression of TRa in QT6 cells did not confer T3 regulation on malic enzyme activity and had little or no effect on the basal expression of this enzyme. This contrasts with the dramatic effects of enhanced expression of TRa on these variables in CEF. Because CEF and QT6 cells infected with RCAS-c-erbAa expressed TRa at similar levels, the specific transcriptional activity of TRa must be decreased in QT6 cells relative to CEF. Results from transient transfection analyses support this conclusion. The responsiveness of pME5.8-CAT to T3 was increased 2-fold in RCAS-c-erbAa-CEF relative to RCAS-CEF, whereas T3-mediated regulation of this chimeric gene was not altered in RCAS-c-erbAa-QT6-4 relative to RCAS-QT6-1 (Figs. 3, A and B ) . Transient expression of pRSV-chicken-c-erbAa increased the T3-induced stimulation of pME5.8-CAT expression to a maximum of about 50-fold in CEF. In contrast, the maximum T3-induced stimulation of pME5.8-CAT activity in QT6 cells transfected with pRSV-chicken-c-erbAa was 5-fold. The increased ability of transiently versus stably expressed TRa to modulate malic enzyme promoter activity in CEF and QT6 cells is probably due to a higher level of expression of the transcription factor per cell in the former system. In summary, the ability of the nuclear T3 receptor to regulate the malic enzyme gene is highest in chick embryo hepatocytes, followed by CEF and QT6 cells in the order of decreasing activity. Experiments are in progress aimed at determining the mechanism(s) responsible for cell-specific differences in TRa activity.
In the rat hepatoma cell line, FAO, stable expression of TRa caused a 3-5-fold increase in the ability of T3 to stimulate malic enzyme mRNA levels (48). It is unclear whether this effect was mediated by changes in basal and/or T3-induced malic enzyme expression. In another study, transient expression of TRa in NIH 3T3 cells conferred T3 regulation to a cotransfected chimeric gene comprised of a continuous piece of DNA (882 base pairs) from the rat malic enzyme promoter/ regulatory region linked to CAT (49). This effect was due entirely to an increase in the activity of the malic enzyme promoter in the presence of T3. Overexpression of TRa had no effect on promoter activity in the absence of T3. Transient expression of TRa in RCAS-QT6-1 also stimulated malic enzyme promoter activity in the presence of T3 with only a small repression in its absence (Fig. 3B). In contrast, overexpression of TRa in CEF mainly affected malic enzyme transcription in the absence of T3 (Table 11, Figs. 2 and 3A). These differences in the effects of TRa on malic enzyme promoter activity in CEF uersus RCAS-QT6-1 and NIH 3T3 cells suggest that distinct mechanisms may be involved in the cell-specific regulation of repressor and T3-dependent derepressor/activation functions of TRa.
Expression of the genes for malic enzyme, fatty acid synthase, and acetyl-coA carboxylase are coordinately regulated under a wide variety of nutritional and hormonal conditions (50). In hepatocytes in culture, for example, T3 stimulates expression of all three of these genes (6, 28, 41). The observation that stable overexpression of TRa in CEF conferred T3 regulation on malic enzyme but not fatty acid synthase suggests that different mechanisms may be involved in the regulation of these genes (Figs. 1 and 2). The T3-induced stimulation of transcription of the malic enzyme gene in chick embryo hepatocytes does not require ongoing protein synthesis (6). In contrast, stimulation of fatty acid synthase by T3 is inhibited when protein synthesis is inhibited. These results are consistent with the hypothesis that the different mechanisms for T3-mediated regulation of malic enzyme and fatty acid synthase involve requirements for different proteins. Expression of acetyl-coA carboxylase also was unresponsive to enhanced TRa expression in CEF, suggesting that the mechanism by which this gene is regulated by T3 is also different from that of malic enzyme.
Based on the transient expression experiments, the v-erbA

Regulation of the Malic Enzyme
Gene by the T, Receptor protein can constitutively repress activity of the malic enzyme promoter in CEF. Interestingly, stable expression of v-erbA in XJIP-CEF had no effect on malic enzyme activity. Differences in stable and transient expression systems may due to differences in the level of expression of v-erbA per cell. XJ12-CEF exhibit an enhanced growth potential (51), a phenomenon confirmed in our experiments. This observation, together with the fact that malic enzyme activity in XJ12-CEF is not diminished, suggests that the malic enzyme gene is not a target of v-erbA action in neoplastic transformation. In summary, TRa possesses multiple activities with respect to the malic enzyme gene. Cell-specific as well as gene-specific factors are involved in controlling the activities of this transcription factor. The differential responsiveness of chick embryo hepatocytes, CEF and QT6 cells to T3 should facilitate identification and characterization of these factors. The retroviral vector-based stable expression system developed in the present study will be useful in future work aimed at analyzing the function of factors that regulate nuclear T3 receptor activity in avian cells.