Cyclic AMP-mediated inhibition of transcription of the malic enzyme gene in chick embryo hepatocytes in culture. Characterization of a cis-acting element far upstream of the promoter.

Glucagon, acting via cAMP, inhibits transcription of the malic enzyme gene in chick embryo hepatocytes. In transiently transfected hepatocytes, fragments from the 5'-flanking DNA of the malic enzyme gene confer cAMP responsiveness to linked reporter genes. The major inhibitory cAMP response element at -3180/-3174 base pairs (bp) is similar to the consensus binding site for AP1. DNA fragments from -3134/-3115, -1713/-944, and -413/-147 bp also contain inhibitory cAMP response elements. The negative action of cAMP is mimicked by overexpression of the catalytic subunit of protein kinase A, inhibited by overexpression of a specific inhibitor of protein kinase A, and inhibited by overexpression of the T3 receptor; these results indicate involvement of the classical eukaryotic pathway for cAMP action and suggest interaction between the T3 and cAMP pathways. Sequence-specific complexes form between nuclear proteins and a DNA fragment containing -3192/-3158 bp of 5'-flanking DNA. In nuclear extracts prepared from cells treated with chlorophenylthio-cyclic AMP and T3, the complexes have different masses than those formed with extracts from cells treated with T3 alone. Antibodies to c-Fos or ATF-2 inhibit formation of the complex formed by proteins from cells treated with chlorophenylthio-cyclic AMP and T3 but not by those from cells treated with T3 alone. These results suggest an important role for c-Fos and ATF-2 in glucagon-mediated inhibition of transcription of the malic enzyme gene.

Malic enzyme (ME) 1 (EC 1.1.1.40) catalyzes the oxidative decarboxylation of malate to pyruvate and CO 2 , simultaneously generating NADPH from NADP ϩ . In avian liver, most of the NADPH used in the de novo synthesis of long chain fatty acids is generated by malic enzyme (1). Malic enzyme is a typical lipogenic enzyme; its activity in avian liver increases about 70-fold when newly hatched chicks are fed a diet high in carbohydrate (1) and decreases dramatically when animals are starved (2). In chicken embryo hepatocytes in culture, insulin plus T3 causes about a 50-fold increase in malic enzyme activity and abundance of its mRNA; glucagon or cAMP blocks these effects (3,4). Within 1 h after adding T3 to chick embryo hepatocytes, transcription of the malic enzyme gene increases by 30 -40-fold; cAMP completely inhibits this increase (4). The T3-dependent increase in transcription of the malic enzyme gene is mediated by several T3 response elements (T3REs), with the major one between Ϫ3883 and Ϫ3858 bp upstream of the start site for transcription (5).
Several positive acting cAMP response elements have been described (6). Cyclic AMP stimulates gene expression by activating protein kinase A (PKA), which, in turn, phosphorylates members of the CREB/ATF family of transcription factors, thereby increasing their transactivation potential (7). The CREB/ATF family includes polypeptides encoded by at least seven distinct genes (6) that share the ability to bind, with different affinities, to CREs in the 5Ј-flanking DNA of genes activated by cAMP. The CREB/ATF proteins also dimerize with AP1 proteins, members of another family of leucine zipper proteins (8) for which the binding site differs from a consensus CRE by only one bp (9).
Much less is known about negative-acting cAMP-response elements. In the gene for L-type pyruvate kinase, the L4 element, Ϫ168 to Ϫ144 bp, binds major late transcription factor; it is the glucose/insulin response element and is required for inhibition by cAMP. Inhibition by cAMP also requires the contiguous L3 element, an element that binds hepatic nuclear factor 4 (10). Cyclic AMP also inhibits transcription of the genes for IL-2 and IL-2R in EL4 cells; the inhibition requires an AP1 site. In this case, cAMP increases the binding of Jun/Fos heterodimers to the AP1 site and alters the composition of Jun proteins that participate in the AP1 complex (11). A third example of inhibition of transcription by cAMP involves the hepatic gene for fatty acid synthase; insulin-induced transcription of this gene is inhibited by cAMP (12). The cis-acting element required for the inhibitory effect is an inverted CAAT box (13). The proteins that bind to this element have not been identified.
In this study, we have found that the 5Ј-flanking DNA of the gene for malic enzyme contains at least four cis-acting DNA sequences that are involved in responsiveness to the inhibitory action of cAMP. We have examined the function of the probable major inhibitory element and identified some of the nuclear proteins that bind to it.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes were obtained from New England Biolabs (Beverly, MA) or Boehringer Mannheim. Other enzymes were obtained from the indicated sources: Taq DNA polymerase (Perkin-Elmer), T4 DNA ligase (Pharmacia Biotech Inc.), Klenow fragment of Escherichia coli DNA polymerase I and calf intestinal phosphatase (Boehringer Mannheim). CPT-cAMP, 3,5,3Ј-L-triiodothyronine, and corticosterone were purchased from Sigma. Crystalline bovine insulin was a gift from Lilly. LipofectAce TM and Waymouth medium MD 705/1 were obtained from Life Technologies, Inc. [␣-32 P]dCTP (800 Ci/mmol) was purchased from Amersham Corp., and D-threo-[dichloroacetyl-1-2- 14 C]chloramphenicol was from NEN Life Science Products. D-Luciferin, potassium salt, was obtained from Analytical Luminescence Laboratory (San Diego, CA). Antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA); all were raised against human protein except TR␣, for which the antigen was chicken in origin. All other chemicals were of reagent grade or of the highest purity commercially available.
Plasmid Constructions-Plasmid RSV-PKAc and pRSV-PKi were from R. A. Maurer (Oregon Health Sciences University) (14). Plasmid RSV-TR␣ was provided by H. H. Samuels (New York University) (15). The luciferase reporter plasmid, pXP1 (16), was from S. K. Nordeen (University of Colorado Health Sciences Center). Plasmid RSV-LUC was constructed as described previously (5). B. Luckow and G. Schutz (Heidelberg, Germany) provided pBLCAT2 (Ref. 17; pTKCAT). Plasmid KSCAT (promoterless plasmid), p[MEϪ5800/ϩ31]CAT, and the 5Ј-deletions thereof were constructed as described previously (5). For the TK constructs, we inserted various fragments of ME 5Ј-flanking DNA into the multiple cloning site 5Ј of the HSV TK promoter in pBLCAT2 (Ref. 17 Cell Culture and Transient Transfection-Isolated hepatocytes were prepared from the livers of 19-day-old chick embryos (18). The cells were incubated in 35-mm plates with Waymouth medium MD 705/1 and streptomycin (100 g/ml), penicillin G (60 g/ml), insulin (50 nM), and corticosterone (1 M). After 16 -24 h, the cells were transfected with p[MEϪ5800/ϩ31]CAT (2.5 g) or an equimolar amount of another reporter plasmid, pRSV-LUC (0.5 g) and sufficient pBluescript to bring the total amount of transfected DNA to 5 g/plate. The cells were incubated with the DNA/lipofectACE TM mixture for 16 -24 h; thereafter, the medium was replaced with fresh medium with or without T3 (1.6 M) and with or without CPT-cAMP (500 M). After an additional 48 h of incubation, the hepatocytes were harvested, and extracts were prepared (5). A detailed description of our transfection procedures has been published (19).
Analysis of Cell Extracts and Statistical Analysis-Cell extracts were analyzed for protein quantity (20), luciferase activity (21), and CAT activity (22). The results were expressed initially as percentage of substrate converted to acetylated product per milligram of unheated soluble protein and then normalized as described in the figure legends. The ratio of average relative CAT activities with and without T3 or cAMP is not always the same as the corresponding average -fold changes in CAT activities or percentages of control activities, respectively, because the latter are the averages of the individual -fold changes or percentages of control activities for each experiment. The statistical significance of differences between pairs of means was determined by the Wilcoxon matched pairs, signed rank test (23). Standard errors of the mean are provided to indicate the degree of variability in the data.
Gel Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared (24) from chick embryo hepatocytes incubated for 48 h with insulin, corticosterone, and T3 with or without CPT-cAMP. The extraction buffer contained 0.6 M KCl and the protease inhibitors leupeptin, benzamidine, aprotinin, and phenylmethylsulfonyl fluoride. Doublestranded oligonucleotides were labeled by a fill-in reaction with [␣-32 P]dCTP catalyzed by the Klenow fragment of DNA polymerase. A volume of nuclear extract containing 6 g of protein was mixed with 12 l of binding buffer containing 20,000 cpm of 32 P-labeled probe, 2 g of poly(dI-dC), 0.01% Nonidet P-40, 0.8 g of bovine serum albumin, 5% (v/v) glycerol, and 5 g of salmon sperm DNA with or without a 100-fold molar excess of competitor oligonucleotide. The reaction was incubated for 15 min at room temperature. Antibody experiments used the same incubation conditions except that 1 l of IgG (1 g) was incubated with the reaction mixture for an additional 15 min at room temperature. The reaction mixture was then subjected to electrophoresis on 5% polyacrylamide gels at 150 V in 25 mM Tris-HCl, 0.19 M glycine, 1 mM EDTA at 4°C (24). Gels were dried and subjected to autoradiography.

Cis-acting Elements that Confer Responsiveness to cAMP Are
Located within the 5.8 kb of DNA Upstream of the Start Site of the Malic Enzyme Gene-When chick embryo hepatocytes were transfected with 5.8 kb of 5Ј-flanking DNA of the malic enzyme gene linked to the CAT gene (p[MEϪ5800/ϩ31]CAT), CAT activity was increased more than 40-fold by T3 and inhibited by 96% by CPT-cAMP (Fig. 1). CPT-cAMP did not have a statistically significant effect on luciferase activity in cells transfected with RSV-LUC. Thus, 5.8 kb of 5Ј-flanking DNA contains both positive acting T3REs, as previously reported (5), and one or more inhibitory cAMP response elements (ICREs). The magnitude of the cAMP effect is similar to that for the inhibition of transcription of this gene caused by cAMP (4), suggesting that all of the elements necessary for the negative effect of cAMP may be located in this fragment of DNA.
We next tested a series of 5Ј-deletions of p[MEϪ5800/ ϩ31]CAT to localize the ICRE(s) (Fig. 1). Deletion of the DNA from Ϫ5800 to Ϫ3845 bp caused an increase in responsiveness to cAMP. This deletion removes part of the T3 response region (5) and may contain sequences that dampen the ability of cAMP to inhibit transcription. When the DNA from Ϫ3845 to Ϫ3474 bp was deleted, responsiveness to cAMP increased slightly, but the increase was not statistically significant. Deletion of the region from Ϫ3474 to Ϫ2715 bp decreased cAMPmediated inhibition from 99 to 92%. This represents an 8-fold difference in sensitivity to cAMP between sets of cells transfected with these two plasmids. An additional decrease in responsiveness occurred when DNA from Ϫ2715 to Ϫ944 bp was removed ( Fig. 1). Both p[MEϪ944/ϩ31]CAT and p[MEϪ413/ ϩ31]CAT conferred statistically significant 64 and 45% inhibitions by cAMP, respectively. When cells were transfected with p[MEϪ147/ϩ31]CAT, there was measurable basal activity but no inhibition by cAMP (results not shown). These results suggest that 3474 bp of 5Ј-flanking DNA contains three or more ICREs that are distinct from the major T3RE and several minor T3REs that are located between Ϫ3903 and Ϫ3703 bp (5).
Cyclic AMP-induced Inhibition of Promoter Activity Involves the Classical PKA Signaling Pathway-An expression vector containing the coding sequence for the catalytic subunit of PKA (pRSV-PKAc) (14) was cotransfected with p[MEϪ5800/ ϩ31]CAT. Overexpression of catalytic subunit inhibited T3induced CAT activity in a dose-dependent manner ( Fig. 2A). At 0.1 g/plate, overexpression of the catalytic subunit inhibited T3-induced CAT activity by 80%. Even in cells treated with cAMP, overexpression of the catalytic subunit decreased CAT activity. These results suggest that the catalytic subunit of PKA itself is sufficient to inhibit promoter activity of the malic enzyme gene. The basal level of CAT activity (no T3) also was inhibited by cAMP, suggesting that inhibition by cAMP may be independent of T3.
We also cotransfected a construct that expresses a specific inhibitor of the catalytic subunit (PKi or Walsh inhibitor) (25). When 1 or 2 g of pRSV-PKi was cotransfected with p[MEϪ5800/ϩ31]CAT, inhibition by exogenously added cAMP was completely blocked; in fact, activity was higher than that in the absence of cAMP. In addition, overexpression of pRSV-PKi in the absence of cAMP, with or without T3, stimulated CAT activity. This suggests that these cells contain a significant level of free catalytic subunit of PKA in the absence of added cAMP. We conclude that the negative effect of cyclic AMP on transcription of the malic enzyme gene is mediated by the classical eukaryotic signaling pathway that involves PKAmediated phosphorylation of target proteins.
T3 and cAMP Can Function Independently but Have Interacting Effects on Promoter Activity-In the absence of T3, cyclic AMP caused 53 and 75% decreases in CAT activity in cells transfected by p[MEϪ5800/ϩ31]CAT and p[MEϪ3474/ ϩ31]CAT, respectively (Fig. 3). This result is consistent with the inhibition of promoter activity caused by overexpression of the free catalytic subunit of PKA and suggests that cAMPmediated inhibition of transcription of the malic enzyme gene does not inhibit TR function per se.
The degree of inhibition by cAMP was much greater when promoter activity of the malic enzyme gene was induced by T3 than in the absence of T3 (Fig. 3); this may be due to greater sensitivity to cAMP in T3-treated cells. Alternatively, the lower cAMP responsiveness of the malic enzyme gene in cells that were not treated with T3 may be due to a combination of the lower level of promoter activity in the absence of T3 and a Left, constructs used in this experiment. The number at the left of each construct indicates the 5Ј-end of that fragment in nucleotides relative to the major start site for transcription. For all constructs, the 3Ј-end was at ϩ31 bp. Columns 1-3, CAT activity normalized by luciferase activity. Each result was expressed as a percentage of [ 14 C]chloramphenicol converted to acetylated chloramphenicol per milligram of soluble protein and then corrected for differences in transfection efficiency by dividing by luciferase activity of the same extract. Relative CAT activities were then calculated by setting the corrected CAT activity for T3-treated hepatocytes transfected with p[ME-4135/ϩ31]CAT DNA to 100 and adjusting all other activities proportionately. We normalized to activity in cells transfected with this plasmid because it was the only construct present in every set of experiments that was performed. The results are the means Ϯ S.E. of 5-9 experiments (n), each one using an independent batch of hepatocytes. Two independently prepared batches of each plasmid were used. CAT and luciferase activities of extracts from T3-treated hepatocytes transfected with p[MEϪ4135/ϩ31]CAT were 6.4 Ϯ 1% (mean Ϯ S.E., n ϭ 9) conversion/15 h/mg of protein and 8.1 Ϯ 1.6 ϫ 10 6 (mean Ϯ S.E., n ϭ 9) light units/mg of protein, respectively. Column 4, relative CAT activity in cells incubated with insulin, corticosterone, and T3 divided by that in control cells incubated with insulin and corticosterone. Column 5, relative CAT activity in cells incubated with insulin, corticosterone, T3, and CPT-cAMP divided by that in cells incubated with insulin, corticosterone, and T3, times 100. Column 6, number of experiments. Statistical significance of comparisons made within a column is as follows: a, p Ͻ 0.05 versus p[MEϪ5800/ϩ31]CAT. I, insulin; C, corticosterone; T3, triiodothyronine; cA, CPT-cAMP.
cAMP-mediated Inhibition of Transcription of the ME Gene constitutive basal level of activity that is independent of T3 or cAMP.
The conclusion that cAMP functions independently of T3 is supported by the results of a series of experiments in which different artificial and natural T3REs were linked to TKCAT and tested for responsiveness to T3 and cAMP. One construct contained five copies of a consensus T3RE (5Ј-AGGT-CANNNAGGTCA-3Ј) linked to TKCAT (26); a second construct contained a palindromic T3RE linked to TKCAT (27). Hepatocytes transfected with each of these constructs gave robust responses to T3 but did not respond to cAMP (results not shown). Similarly, cells transfected with T3RE2 of the malic cAMP-mediated Inhibition of Transcription of the ME Gene enzyme gene linked directly to TKCAT also failed to respond to cAMP (Fig. 4). Thus, a T3 response by itself is not sufficient to make transcription of the malic enzyme gene responsive to cAMP.
Plasmid [MEϪ5800/ϩ31]CAT and an expression vector for chicken TR␣ were cotransfected into hepatocytes. At many T3REs, TR is a repressor in the absence of T3 (15). Overexpression of TR␣ caused the expected decrease in basal activity (IC-treated cells) and an increase in T3-induced activity (Fig.  5); inhibition by cAMP decreased from ϳ95 to ϳ40%.
Functional Inhibitory cAMP Response Elements: Localization of Several and Identification of One-Three regions appear to contain ICREs (Fig. 1). DNA fragments containing the putative ICREs were subcloned upstream of the TK promoter in pTKCAT. A 35-bp oligonucleotide containing T3RE2 of the malic enzyme gene (5) was inserted upstream of each of the ICRE-containing fragments that lacked a T3RE. This ensured a high level of promoter activity in T3-treated cells (Fig. 4). When hepatocytes were transfected with the p[MEϪ3474/ Ϫ2715]TKCAT, cAMP inhibited T3-induced CAT activity by 98% (Fig. 4A). In cells transfected with p[ME(T3RE2)Ϫ1713/ Ϫ944]TKCAT and p[ME(T3RE2)Ϫ413/Ϫ147]TKCAT, cAMP caused 77 and 72% inhibition of CAT activity, respectively (Fig.  4B). Cyclic AMP had no effect on promoter activity in control cells transfected with p[ME-T3RE2]TKCAT or pTKCAT. These results and those in Fig. 1 are consistent with there being at least three ICREs in the 5Ј-flanking DNA of the malic enzyme gene.
For further localization of the 5Ј-most ICRE(s), we constructed and tested a series of deletions of p[MEϪ3474/ Ϫ2715]TKCAT (Fig. 4A). Deletion of the 5Ј-end to Ϫ3260 bp or the 3Ј-end to Ϫ2929 bp did not affect either T3 or cAMP responsiveness of hepatocytes transfected with these constructs. When the DNA from Ϫ3259 to Ϫ3115 bp was deleted from the parent plasmid, T3 responsiveness was essentially unchanged, but inhibition by cAMP decreased from 98 to 22%, indicating that an ICRE was located in this DNA fragment. When the DNA from Ϫ3114 to Ϫ2930 bp was deleted, T3 responsiveness decreased to 2-fold, and inhibition by cAMP decreased to 23%. This result suggests that a T3RE is located in this region. The decrease in responsiveness to cAMP could have been due to 1) the presence of a second ICRE in this region, 2) an ICRE that

FIG. 4. Effects of T3 and cAMP on cells transfected with constructs containing fragments of three different regions of malic enzyme 5-flanking DNA linked to TKCAT.
Chick embryo hepatocytes were isolated, incubated in culture, and transfected as described in the legend to Fig. 1 and under "Experimental Procedures." Left, constructs used in this experiment. TK, thymidine kinase promoter; T3RE2, major T3 response element located between Ϫ3883 and Ϫ3858 bp in the 5Ј-flanking DNA of the malic enzyme gene. Numbers in large type represent the extremities of the tested fragments; numbers in small type represent the ends of internal deletions. Columns 1-3, CAT activity was normalized by luciferase activity. The results were calculated and are presented as described in the legend to Fig. 1. Column 4, relative CAT activity in cells incubated with insulin, corticosterone, and T3 divided by that in control cells incubated with insulin and corticosterone. Column 5, relative CAT activity in cells incubated with insulin, corticosterone, T3, and CPT-cAMP divided by that in cells incubated with insulin, corticosterone, and T3, multiplied by 100. Column 6, number of experiments. A, the relative CAT activity for T3-treated hepatocytes transfected with p[MEϪ3474/ Ϫ2715]TKCAT was fixed at 100, and the other values were adjusted proportionately. The results, calculated as described in the legend to Fig. 1, are the means Ϯ S.E. of 6 -10 experiments (n). CAT and luciferase activities of extracts from T3-treated hepatocytes transfected with p[MEϪ3474/ Ϫ2715]TKCAT were 8.7 Ϯ 2.6% (mean Ϯ S.E., n ϭ 10) conversion/15 h/mg of protein and 7.6 Ϯ 1.9 ϫ 10 6 (mean Ϯ S.E., n ϭ 10) light units/mg of protein, respectively. Statistical significance of comparisons made within a column is as follows: a, p Ͻ 0.01 versus TKCAT; b, p Ͻ 0.05 versus TKCAT; c, p Ͻ 0.02 versus TKCAT. B, the relative CAT activity for T3-treated hepatocytes transfected with p[ME(T3RE2)Ϫ1713/Ϫ944]TKCAT was fixed to 100, and the other values were adjusted proportionately. The results, calculated as described above, are the means Ϯ S.E. of five or six experiments (n). CAT and luciferase activities of extracts from T3-treated hepatocytes transfected with p[ME(T3RE2)Ϫ1713/Ϫ944]TKCAT were 2.2 Ϯ 0.7% (mean Ϯ S.E., n ϭ 6) conversion/15 h/mg of protein and 12.3 Ϯ 2.5 ϫ 10 6 (mean Ϯ S.E., n ϭ 6) light units/mg of protein, respectively. Statistical significance of comparisons made within a column is as follows: d, p Ͻ 0.05 versus p[MET3RE2]TKCAT. Abbreviations are as defined in the legend to Fig. 1. cAMP-mediated Inhibition of Transcription of the ME Gene overlaps both of these deletions, or 3) the deletion of the T3RE localized between Ϫ3114 and Ϫ2930 bp. T3-induced activity in the absence of a T3RE is little different from basal activity and may be too low to permit detection of a larger response to cAMP.
Each of the fragments containing a potential ICRE was subcloned upstream of the TK promoter in pTKCAT and transfected into hepatocytes (Fig. 6). The major T3RE of the malic enzyme gene was inserted upstream of the Ϫ3259-/Ϫ3115-bp fragment. This was not necessary for the Ϫ3114-/Ϫ2930-bp fragment, because it contains a T3RE. In cells transfected with p[ME(T3RE2)Ϫ3259/Ϫ3115]TKCAT, CAT activity was strongly suppressed by cAMP (98%). In contrast, when the cells were transfected with p[MEϪ3114/Ϫ2930]TKCAT, there was no change in CAT activity in response to cAMP despite the fact that CAT activity was stimulated 12-fold by T3. These results indicate that at least one ICRE is localized between Ϫ3259 and Ϫ3115 bp in the 5Ј-flanking DNA of the gene for malic enzyme. They also confirm the presence of a T3RE between Ϫ3114 and Ϫ2930 bp and support the suggestion that the loss of cAMP responsiveness that occurred when this fragment was deleted from a plasmid containing the Ϫ3474 to Ϫ2715 bp fragment was due to loss of this T3RE.
To localize the ICRE more precisely, we prepared a series of deletions of p[ME(T3RE2)Ϫ3259/Ϫ3115]TKCAT (Fig. 7). Deletion from Ϫ3259 to Ϫ3192 bp had no effect on responsiveness to cAMP. Deletion from Ϫ3192 to Ϫ3161 bp decreased inhibition by cAMP from more than 98 to 76%, a 15-fold change in responsiveness to cAMP. Further deletion to Ϫ3134 bp had no effect. These results suggest that there are at least two ICREs in this fragment, one between Ϫ3192 and Ϫ3161 bp and one between Ϫ3134 and Ϫ3115 bp.
We next focused on the ICRE between Ϫ3192 and Ϫ3161 bp. This 34-bp fragment contains a sequence that is very similar (one mismatch) to those of AP1 sites that are required for cAMP-mediated inhibition of transcription the IL-2 and IL-2R genes (11). p[ME(T3RE2)Ϫ3192/Ϫ3158]TKCAT was constructed in a wild-type form or with a block mutation in the putative AP1 site (complementary sequence) (Fig. 8). In cells transfected with the wild-type construct, CAT activity was stimulated by T3 and inhibited by cAMP (94%). When the mutant construct was introduced into the cells, stimulation by T3 was preserved, albeit at a lower level, and inhibition by cAMP was lost. This AP1-like site thus appears to be necessary for at least part of the cAMP-mediated inhibition of T3-induced transcription.
The AP1-like Site Binds Nuclear Proteins-Proteins from the nuclei of T3-treated hepatocytes bound to a double-stranded 32 P-labeled oligonucleotide (Ϫ3192 to Ϫ3158 bp, wild type) and produced one major complex (Fig. 9, a). One major complex with lower mobility (b) was observed when the experiment was performed with nuclear extracts from hepatocytes treated with T3 plus cAMP. This result suggests that additional or different nuclear proteins bind to this site when cAMP is added to hepatocytes. A 100-fold molar excess of the unlabeled form of the probe displaced the labeled retarded probe (Fig. 9). Neither the mutant version of the probe containing the block mutation tested in intact cells nor a similarly sized probe containing the major T3RE of the malic enzyme gene displaced the retarded probe (Fig. 9). Thus, the a and b complexes are sequencespecific, and the mutated bases within the AP1-like site are necessary for the protein-DNA interaction. An apparently specific complex of lower molecular weight may represent degradation products or complexes missing a component of complexes a or b.
To identify proteins that bind to this DNA fragment, we used antibodies (IgG) raised against mammalian forms of several candidate proteins (Fig. 10). AP1 is a Fos/Jun heterodimer. The Fos antibody used in this experiment (k-25) reacts with all members of the Fos gene family: c-Fos, FosB, Fra-1, and Fra-2. The antibody used for Jun is specific of the c-Jun isoform and does not cross-react with JunB or JunD. AP1 also can contain heterodimers that include nuclear factors of the CREB/ATF family (28,29). The ATF-1 antibody (25c10g) cross-reacts with ATF-1, CREB, and CREM. The antibody for ATF-2 is specific for ATF-2. We also tested an antibody raised against CREBbinding protein (CBP), a potential CREB co-activator (30). TR␣ binds to the major T3RE of the malic enzyme gene (5). We tested for TR binding to this element because the T3 and cAMP signaling systems appear to interact (Fig. 4). Formation of complex a with nuclear proteins extracted from T3-treated cells was not inhibited by any of the antibodies (Fig. 10). When the analysis was performed with nuclear proteins from cells treated with T3 plus cAMP, formation of complex b was inhibited completely by IgG raised against c-Fos and ATF-2 but not by any of the other IgGs (Fig. 10). These results suggest that cAMP-mediated Inhibition of Transcription of the ME Gene different proteins are present in the a and b complexes. Complex b contains both c-Fos and ATF-2, probably as a heterodimer. Proteins in complex a remain unidentified; we cannot exclude a member of the Jun family, because our antibody is specific for c-Jun. Assuming that the chicken homologs of these proteins cross-react with antibodies to the human forms, CREB, CREM, ATF-1, and CBP in these hepatocyte nuclear extracts do not appear to bind to this AP1-like site, with or without cAMP in the medium. In addition, the interaction of the T3 and cAMP signaling pathways does not appear to involve direct interaction of TR␣ with proteins in the AP1-like complex.

DISCUSSION
The chicken malic enzyme gene is expressed primarily in hepatocytes and is subject to regulation by various hormones and nutritional states (2). Agents that increase intracellular levels of cAMP markedly decrease transcription of the gene for malic enzyme in chicken embryo hepatocytes in culture (4); four different regions in the 5Ј-flanking DNA of the malic enzyme gene appear to be required for the full cAMP inhibition endowed by 5.8 kb of 5Ј-flanking DNA. In this report, we have focused on a specific, AP1-like site at Ϫ3180 to Ϫ3174 bp of the malic enzyme gene. What do we know about the other ICREs? The region from Ϫ3134 to Ϫ3115 bp (63% inhibition by cAMP) contains a CRE-like element (2 mismatches with respect to the consensus element, TGACGTCA) (6) between Ϫ3127 and Ϫ3120 bp. Both the Ϫ1713to Ϫ944-bp and Ϫ413to Ϫ147-bp fragments endowed about 75% inhibition by cAMP. Each of these fragments contains one or more AP1 consensus sites (TGA(G/C)TCA) (9). The functional roles of these downstream AP1-like sites remains to be determined.
The AP1-like site (TAAATCA) centered at Ϫ3177 bp contains one mismatch with respect to a site (TAAGTCA) that is involved in cAMP-mediated inhibition of transcription of the IL-2 and IL-2R genes in EL4 cells (11). The AP1 sites of the IL-2 and IL-2R genes bind Fos/Jun heterodimers; cAMP increases their binding activity and alters the type of Jun protein in the AP1 complex. AP1 consists of a collection of structurally related transcription factors that belong to the Fos and Jun families. These proteins form a variety of homo-and heterodimers and constitute an important group of signal-regulated transcription factors, BZip proteins. The consensus sequence for an AP1 binding site (TGA(C/G)TCA) is similar to that for a CRE consensus sequence (TGACGTCA). CREs bind other members of the BZip protein family: CREB and ATF (31). Factors that bind to the AP1 site also form heterodimers with the members of the CREB/ATF family (28,29). Of the various members of these two gene families for which we assayed, only c-Fos and ATF2 were found in the complexes that bound to the AP1-like site implicated in responsiveness to cAMP; they were present in the complex only when nuclear extracts were from cAMP-treated hepatocytes. It has been reported that c-Fos cannot heterodimerize with ATF-2 (31). However, others have reported the formation of a c-Fos/ATF-2 complex in vitro but did not characterize its DNA binding site (32).
To our knowledge, we are the first to report the formation of a complex containing c-Fos and ATF-2 in intact cells. This unusual complex may explain why the AP1-like sequence char- acterized in this report does not match perfectly with the consensus site for the binding of Jun/Fos heterodimers. The two thymidines and the cytidine in an AP1 site (TGAGTCA) are important for the binding of the Jun/Fos heterodimer (33). In our sequence, these three bases are conserved. However, replacement of the first guanosine with an adenosine (as in our sequence) abolishes the binding of the Jun/Fos complex.
Our results confirm that cAMP inhibits transcription of the malic enzyme gene via the classical PKA signaling pathway. In response to the binding of cAMP to regulatory subunits of PKA, the catalytic subunits are released and catalyze phosphorylation of target proteins (34,35). In the nucleus, the phosphorylation states of transcription factors and related proteins appear to modulate function and expression of cAMP-responsive genes directly (6).  Fig. 1.   FIG. 8. The effect of a block mutation in the ؊3192 to ؊3158 bp fragment. Chick embryo hepatocytes were isolated, incubated in culture, and transfected as described in the legend to Fig. 1 and under "Experimental Procedures." Left, constructs that were used in this experiment. The wild-type (WT) plasmid contains the natural sequence of this putative ICRE, a consensus AP1 binding site. The mutant (MUT) plasmid contains a block mutation in the AP1 site (complementary sequence, as shown in the figure). Columns 1-3, the relative CAT activity for T3-treated hepatocytes transfected with the wild-type, p[ME(T3RE2)Ϫ3192/Ϫ3158]TKCAT, was fixed to 100, and the other values were adjusted proportionately. The results are means Ϯ S.E. of 3-6 experiments (n). CAT and luciferase activities of extracts from T3-treated hepatocytes transfected with the wild-type, p[ME(T3RE2)Ϫ3192/Ϫ3158]TKCAT, were 8.6 Ϯ 2.8% (mean Ϯ S.E., n ϭ 6) conversion/15 h/mg of protein and 9.2 Ϯ 2.5 ϫ 10 6 (mean Ϯ S.E., n ϭ 6) light units/mg of protein, respectively. Columns 4 -6, presentation of the results is similar to that described in the legend to cAMP-mediated Inhibition of Transcription of the ME Gene enzyme gene? c-Fos is phosphorylated after activation of protein kinase C, but phosphorylation by PKA has not been reported (36). ATF-2 also is phosphorylated in intact cells (37), phosphorylation increases its DNA binding activity and appears to be a primary determinant for formation of c-Jun/ ATF-2 heterodimers. A stress-activated protein kinase appears to be involved. To our knowledge, phosphorylation of ATF-2 by PKA has not been reported. We speculate that the rapid cAMPmediated inhibition of transcription of the malic enzyme gene may involve a cascade of kinases and/or phosphatases. Inhibition of transcription of the malic enzyme gene by glucagon and cAMP persists for days in hepatocytes in culture (4). A cAMPinduced increase in transcription of the c-Fos gene (38,39) may account for the long term effect.
Overexpression of TR␣ in chicken embryo hepatocytes decreased cAMP responsiveness, suggesting an interaction between the T3 and cAMP signaling pathways. TR␣ and AP1 can antagonize one another in vivo and in vitro by a mechanism that does not involve competition for the same binding site or interference from adjacent binding sites (40,41). The ability of these factors to interfere with one another's regulation of gene expression appears to involve a TR-AP1 interaction rather than a TR-DNA interaction. The mechanism by which cAMP inhibits the T3-induced transcription of the malic enzyme gene is different because cAMP or overexpression of the catalytic subunit of PKA inhibited promoter activity whether T3 was present or not. Furthermore, cAMP had no effect on CAT activity in cells transfected with p[MET3RE2]TKCAT, a construct that contains only the major T3RE of the malic enzyme gene, or in cells transfected with idealized artificial T3REs. The gel mobility shift experiment using a specific antiserum for the ␣-form of TR also suggests that there is no direct interaction between TR␣ and proteins in the complex that binds to the Ϫ3192/Ϫ3158 bp fragment. So, how can we explain the inhibition of T3-induced transcription of the malic enzyme gene by cAMP and the abolition of this inhibition when we overexpress TR␣? Both TR␣ and the bound c-Fos/ATF-2 complex may interact with a bridging factor that regulates transcription of the malic enzyme gene. Interaction of this factor with TR␣ may lead to the stimulation of transcription; interaction of the DNA-bound, cAMP-activated form of c-Fos/ATF2 may lead to the inhibition of transcription. The DNA-bound, cAMP-activated form of c-Fos/ATF2 may have a higher affinity for the bridging factor than TR␣; factors bound to the AP1-like site in the absence of cAMP may have a lower affinity for the bridging factor than TR␣. Alternatively, the T3 and cAMP pathways may interact through the downstream ICREs. We do not know whether TR interacts directly with complexes at these sites or not.
In summary, the 5Ј-flanking DNA of the chicken malic enzyme gene contains at least four sequences involved in cAMP responsiveness of this gene. An ICRE centered at Ϫ3177 contains an AP1-like sequence that appears to bind a c-Fos/ATF-2 heterodimer in a cAMP-dependent manner. This mechanism may explain the inhibition of transcription of the malic enzyme gene caused by starvation because starved animals have elevated concentrations of glucagon in the blood. FIG. 9. Gel electrophoretic mobility shift assay using the ؊3192to ؊3158-bp fragment of the malic enzyme gene as probe. Chick embryo hepatocytes were isolated, incubated in culture, and transfected as described in the legend to Fig. 1 and "Experimental Procedures." Nuclear extracts were prepared from cells incubated with T3 and with (lanes 6 -9) or without (lanes 2-5) CPT-cAMP. 32 P-labeled double-stranded Ϫ3192to Ϫ3158-bp fragment (wild-type) was incubated with 6 g of hepatic nuclear protein in the presence or absence of a 100-fold molar excess of unlabeled competitor oligonucleotide. The competitor DNAs were wild-type fragment (WT, lanes 3 and 7), mutated Ϫ3192to Ϫ3158-bp fragment (MUT, lanes 4 and 8), and T3RE2 from the malic enzyme gene (T3RE, lanes 5 and 9). The major complexes are designated a and b. Apparent complexes of lower molecular weight may represent degradation products or complexes missing a component of complexes a or b. Lane 1, no nuclear extract. Abbreviations are as defined in the legend to Fig. 1.   FIG. 10. The effects of various antibodies on formation of complexes between nuclear proteins and the ؊3192to ؊3158-bp fragment of the malic enzyme gene. Chick embryo hepatocytes were isolated, incubated in culture, and transfected as described in the legend to Fig. 1 and under "Experimental Procedures." Nuclear extracts (6 g of protein) were prepared from cells incubated with T3 and with (lanes 9 -15) or without (lanes 2-8) CPT-cAMP. The labeled probe was the 32 P-labeled double-stranded Ϫ3192to Ϫ3158-bp fragment of the malic enzyme gene. IgGs (1 g each) were added after mixing labeled probe and nuclear extract; they were anti-c-Fos (lanes 3 and 10), antic-Jun (lanes 4 and 11), anti-ATF-1 ( lanes 5 and 12), anti-ATF-2 (lanes 6 and 13), anti-CBP (lanes 7 and 14), and anti-TR␣ (lanes 8 and 15). Lanes 2 and 9, no IgG. Lane 1, no nuclear extract. The major complexes are designated a and b. Apparent complexes of lower molecular weight may represent degradation products or complexes missing a component of complex a or b. Abbreviations are as defined in the legend to Fig. 1. cAMP-mediated Inhibition of Transcription of the ME Gene