Characterization of CCAAT/Enhancer-binding Protein α as a Cyclic AMP-responsive Nuclear Regulator*

The α isoform of CCAAT/enhancer-binding protein (C/EBPα) is a transcription factor that regulates expression of genes linked to adipose differentiation and hepatic nutrient metabolism. Recently, our laboratory has characterized a role for C/EBPα in mediating hormonal responsiveness. For example, the cAMP responsiveness of the phosphoenolpyruvate carboxykinase gene promoter in liver requires synergism among the cAMP response element-binding protein (CREB), C/EBPα, and activator protein-1. In the present study, we show that C/EBPα can functionally substitute for CREB in this cAMP response unit, i.e. cAMP responsiveness can occur in the absence of CREB. This observation is physiologically relevant since both CREB and C/EBPα have been shown to bind with high affinity to the cAMP response element in this particular promoter. Structure/function analysis of C/EBPα identified specific mutations that differentially affected its constitutive and protein kinase A-inducible activities. This finding suggests that the mechanism whereby C/EBPα mediates constitutive transactivation is distinct from that whereby it mediates cAMP responsiveness. These data support the hypothesis that C/EBPα plays a critical role in metabolism, in part by participating in the hormonal regulation of expression of metabolically important genes.

The ␣ isoform of CCAAT/enhancer-binding protein (C/ EBP␣) is a transcription factor that regulates expression of genes linked to adipose differentiation and hepatic nutrient metabolism. Recently, our laboratory has characterized a role for C/EBP␣ in mediating hormonal responsiveness. For example, the cAMP responsiveness of the phosphoenolpyruvate carboxykinase gene promoter in liver requires synergism among the cAMP response element-binding protein (CREB), C/EBP␣, and activator protein-1. In the present study, we show that C/EBP␣ can functionally substitute for CREB in this cAMP response unit, i.e. cAMP responsiveness can occur in the absence of CREB. This observation is physiologically relevant since both CREB and C/EBP␣ have been shown to bind with high affinity to the cAMP response element in this particular promoter. Structure/function analysis of C/EBP␣ identified specific mutations that differentially affected its constitutive and protein kinase A-inducible activities. This finding suggests that the mechanism whereby C/EBP␣ mediates constitutive transactivation is distinct from that whereby it mediates cAMP responsiveness. These data support the hypothesis that C/EBP␣ plays a critical role in metabolism, in part by participating in the hormonal regulation of expression of metabolically important genes.
The CCAAT/enhancer-binding proteins are a family of transcription factors that all possess a nearly identical basic region leucine zipper DNA-binding motif, but differ significantly in their amino-terminal transactivation domain (1,2). The first member of this protein family to be cloned, C/EBP␣, 1 displays a tissue-specific pattern of expression, with the highest levels of expression occurring in liver, adipose tissue, and lung (3). A number of studies indicate that C/EBP␣ is necessary and sufficient to induce adipocyte differentiation (4 -7), and it transactivates a number of adipocyte-specific genes, such as 422(aP2) and SCD1 (8 -10). Antisense inhibition of C/EBP␣ expression in 3T3-L1 preadipocytes was shown to prevent accumulation of triacylglycerol (7), and adipocytes from C/EBP␣ knockout mice are devoid of lipid (11). Besides this important role in adipose differentiation and lipid metabolism, C/EBP␣ also appears to be a critical player in hepatic nutrient metabolism. It transactivates genes such as phosphoenolpyruvate carboxykinase (PEPCK) (12), serum albumin (13), and alcohol dehydrogenase (14), and C/EBP␣ knockout mice display reduced expression of glycogen synthase and low levels of liver glycogen as well as delayed expression of PEPCK and glucose-6-phosphatase genes (11). Consequently, the mice die shortly after birth from hypoglycemia and other complications.
Most of the studies performed to date characterize C/EBP␣ as a constitutive transactivator (15)(16)(17). Structure/function analyses of C/EBP␣ have demonstrated that the amino-terminal region contains motifs that allow for protein-protein interactions with TATA-binding protein and TFIIB, with mutational analysis indicating that these interactions are important for the constitutive activity of C/EBP␣ (18). Recently, in studies examining the regulatory properties of the PEPCK promoter, our laboratory uncovered a role for C/EBP␣ in mediating hormonal responsiveness via mechanisms that appear to be distinct from those utilized to confer constitutive transactivation. The PEPCK gene, which is expressed at the highest levels in liver and kidney, is transcriptionally responsive to cAMP, but shows tissue-specific response patterns; in liver, the gene is robustly activated by cAMP, whereas in kidney, the response is weak (19). The explanation for this difference appears to reside in the presence of C/EBP-binding sites in the PEPCK promoter. The cAMP response unit in this promoter, which is functional only in liver-derived cells, consists of a typical cAMP response element (CRE), an AP-1-binding site, and three C/EBP-binding sites (20). All five of these cis-elements are required for maximal responsiveness to cAMP. We have shown that the ␣ isoform, but not the ␤ isoform, of C/EBP functions in this cAMP response unit (21).
Another critical player in this cAMP response unit is the CRE-binding protein (CREB), which participates through its binding to the CRE (22). However, the CRE in the PEPCK promoter is rather unique in that it also binds C/EBP proteins with high affinity (23)(24)(25). In this report, we show that C/EBP␣ can substitute for CREB in the cAMP response unit, i.e. that cAMP responsiveness can occur in the absence of CREB.

EXPERIMENTAL PROCEDURES
Materials-DNA-modifying enzymes were purchased from Promega and New England Biolabs Inc. [acetyl-3 H]CoA (10 Ci/mmol) was purchased from NEN Life Science Products. Tissue culture supplies were from Life Technologies, Inc. HepG2 cells were acquired from American Type Culture Collection.
Transfection Experiments and Plasmids-HepG2 cells were cultured and transfected as described previously (21). An expression vector for ␤-galactosidase was cotransfected to monitor transfection efficiency. Descriptions of the reporter gene plasmids and expression plasmids have been previously reported (20 -22, 24), with the exceptions provided below. The expression vector for the C/EBP␣ mutant Y67A,F77A,L78A, driven by the cytomegalovirus promoter, was described previously (18).
The chloramphenicol acetyltransferase (CAT) reporter genes Ϫ68G3 and Ϫ68G4 consist of a minimal PEPCK promoter (26) to which were ligated three or four copies, respectively, of a double-stranded oligonucleotide (27) that contains the recognition site for the yeast transcription factor Gal4. The CAT reporter gene Ϫ68G4A1 was created by linking the A-site oligonucleotide (20) to the Ϫ68G4 vector. The Ϫ68C4 reporter gene consists of four copies of the C-site oligonucleotide (20) linked to a minimal PEPCK promoter. The Ϫ68C4A1 reporter gene was created by linking the A-site oligonucleotide to the Ϫ68C4 reporter gene.
The Gal4-C/EBP␣ deletion mutants were based on the parent vector G␣2 (21), which had the C/EBP␣ transactivation domain (amino acid residues 6 -217) fused in-frame to the Gal4 DNA-binding domain as an EcoRI/PstI fragment. The general strategy was to introduce BamHI restriction sites by site-directed mutagenesis (28) at selected positions within the coding region of the transactivation domain. In every case, the BamHI site was introduced such that it encompassed two intact codons, thereby allowing in-frame fusions between various regions of C/EBP␣ coding regions to create the internal deletion mutants. The EcoRI/BamHI fragments generated were cloned into the Gal4 vector expression plasmid pMI (29) to generate the carboxyl-terminal deletion mutants of C/EBP␣. The internal deletion mutants were created by restricting the appropriate carboxyl-terminal deletion mutant with BamHI and PstI and ligating to it the appropriate carboxyl-terminal BamHI/PstI fragment generated by the site-directed mutagenesis strategy.
The expression plasmid for the Gal4 derivative of Y67A,F77A,L78A was created by introducing an EcoRI site over codons 5 and 6 of the mutant transactivation domain (18) in a similar fashion as described previously (21). The resulting EcoRI/PstI fragment, which contained codons 6 -217 along with the appropriate mutations, was ligated into pM1.
Verification that similar levels of Gal4 fusion proteins of the appropriate length were expressed in HepG2 cells was determined in lysates from transiently transfected cells by Western blot analysis (30) using an antibody specific for the DNA-binding domain of Gal4 (Upstate Biotechnology, Inc.). It should be noted that in the experiments where transcription factors, either wild type or Gal4 fusions, were overexpressed, the amount of expression plasmid used for each protein was different and was based upon the amount that provided the optimal -fold induction. However, we have performed numerous experiments with a wide concentration range of each plasmid. While the absolute values were, in some cases, slightly different from those presented, the general conclusions were not affected.

C/EBP␣ Can Participate in Mediating cAMP
Responsiveness in the Context of the PEPCK Promoter-Our laboratory previously demonstrated that five cis-elements make up the cAMP response unit of the PEPCK promoter and consist of a CRE, a binding site for AP-1, and three binding sites for C/EBP (see Fig. 2A). Furthermore, using synthetic promoters that reconstitute the cAMP response unit, we observed that no rigid architectural requirements were required for the activity of this hormone response unit (20,24). Finally, we provided evidence that supported the hypothesis that the ␣ isoform, but not the ␤ isoform, of C/EBP participated in this cAMP response unit (21). This latter conclusion was based on studies using Gal4-C/EBP hybrid proteins that were tested on an artificial promoter that reconstituted the cAMP response unit, i.e. contained a binding site for CREB, AP-1, and three Gal4-binding sites. However, because these observations were made on an artificial promoter, we decided to test the ability of the Gal4-C/EBP hybrid proteins to reconstitute cAMP response within the context of the PEPCK promoter. A PEPCK promoter-CAT reporter plasmid (Ϫ490P3G4) was used for this study, which has the highest affinity C/EBP-binding site replaced, within the context of the intact promoter, by a Gal4-binding site (27). In Table I, it can be observed that the replacement of this C/EBP-binding site by a Gal4 site resulted in a significant decrease in PKA responsiveness (compare Ϫ490wt with Ϫ490P3G4), consistent with previous studies showing the importance of this cis-element for maximum hormone response.
Expression of the Gal4-C/EBP␣ hybrid protein G␣2 (21) resulted in reconstitution of PKA responsiveness, whereas the Gal4-C/EBP␤ fusion protein G␤2 (21) had no effect on PKA responsiveness. These findings, obtained within the context of the PEPCK promoter, lend further support to our hypothesis that C/EBP␣ participates in mediating the cAMP response.
To further test this hypothesis, we capitalized on the observation that C/EBP␣ levels in HepG2 cells are ϳ5% of those present in hepatocytes (13) and therefore provide a suitable cell background in which to overexpress C/EBP proteins to examine their transcriptional effects. If our hypothesis is correct about the relative roles of C/EBP isoforms in the cAMP responsiveness of the PEPCK promoter, overexpression of C/EBP␣ should enhance the transcriptional response of the promoter to PKA. Conversely, overexpression of C/EBP␤ should either have no effect or, due to its ability to compete with C/EBP␣ for binding to the promoter (25), even act in a dominant-negative fashion. The results of these experiments are shown in Fig. 1. The PEPCK promoter-CAT plasmid was induced strongly (25-fold) by overexpression of the catalytic subunit of PKA. Overexpres-  within the context of the PEPCK promoter HepG2 cells were transfected with 7 g of CAT reporter plasmid, 10 ng of Gal4 expression vector, 2 g of PKA expression vector, and 2 g of ␤-galactosidase plasmid (transfection control) per plate. Ϫ490PCK-CAT contains 490 base pairs of 5Ј-flanking sequences of the PEPCK promoter; Ϫ490P3G4 contains the same sequences, except that a C/EBPbinding site is replaced by a Gal4-binding site. -Fold activation was calculated as the ratio of the CAT activity measured in the presence and absence of PKA catalytic subunit expression vector cotransfection. The values shown are the means Ϯ S.E. of at least three experiments. Ϫ490PCK sion of both C/EBP␣ and C/EBP␤ had a relatively small 2-3fold effect on the basal activity of the PEPCK promoter. Overexpression of C/EBP␣ and the catalytic subunit of PKA acted synergistically to increase PEPCK promoter activity 40-fold, whereas overexpression of C/EBP␤ diminished the inducibility of the promoter in response to PKA. C/EBP␣ Can Mediate cAMP Responsiveness in the Absence of CREB-In a previous study (22), we defined a role for CREB in the cAMP response unit of PEPCK via its ability to bind to the near-consensus CRE in the promoter ( Fig. 2A). In that study, we showed that when the CRE was replaced by a Gal4binding site, expression of a Gal4-CREB fusion protein could reconstitute the cAMP response, confirming the ability of CREB to participate in this cAMP response unit. However, the CRE in the PEPCK promoter is rather unique in that C/EBP proteins bind to this sequence with a similar affinity as CREB (25), allowing the speculation that C/EBP may also be a regulatory protein working through this site (see "Discussion"). We tested this hypothesis using our Gal4 system. In Fig. 2B, we show that G4-PEPCK, a PEPCK promoter derivative that has the CRE replaced by a Gal4-binding site (Ref. 31; see Fig. 2A), not only had its PKA responsiveness reconstituted by expres-sion of a Gal4-CREB fusion protein in HepG2 cells, but a significant restoration of PKA responsiveness was also obtained by overexpression of G␣2 (Fig. 2B). It should be noted that the wild type PEPCK promoter typically shows a 25-60fold induction by PKA in transfection assays in HepG2 cells (Table I) (20,23,24). Overexpression of G␤2 had no effect on PKA responsiveness (Fig. 2B). Thus, it appears that cAMP responsiveness can occur in the complete absence of bound CREB and that C/EBP␣ can functionally substitute for CREB at the CRE.
To test this hypothesis further without the use of Gal4 fusion proteins, we constructed a synthetic promoter that contained four native C/EBP-binding sites and an AP-1 site (linked to a minimal promoter), rather than the three C/EBP ϩ AP-1 ϩ CRE site combination making up the cAMP response unit of the PEPCK promoter. This promoter, called Ϫ68C4A1, showed a 23-fold induction in response to coexpression of the catalytic subunit of PKA in HepG2 cells (Fig. 3). It should be noted that the parent promoter, Ϫ68PCK-CAT, is not responsive to PKA (20). We also examined the role of the AP-1 site by assessing the responsiveness of a synthetic promoter containing only the four C/EBP-binding sites. This promoter, called Ϫ68C4, also displayed a significant (9-fold) induction by PKA, albeit weaker than Ϫ68C4A1 (Fig. 3). These results suggest that C/EBP can independently mediate cAMP responsiveness and can synergize with AP-1 to mediate a larger response. This latter observation is consistent with our previous data, which showed that the AP-1 site in the PEPCK promoter had no intrinsic ability to mediate cAMP responsiveness, but did synergize with other cis-elements in the promoter to provide a robust response to cAMP (20).
Because the experiments shown in Fig. 3 could not distinguish which C/EBP isoform may be binding to the C/EBP sites in the test promoters, we performed additional experiments using our Gal4 system. Two promoters similar to those shown in Fig. 3 were constructed that had the C/EBP sites replaced with the Gal4 sites (Fig. 4A). Using these promoters, we demonstrated that PKA responsiveness was negligible unless Gal4-CREB or G␣2 was coexpressed (Fig. 4, B and C). However, there was a notable difference between these two Gal4 fusion proteins. Gal4-CREB showed no requirement for the AP-1 site in order to mediate PKA induction, whereas the inducible activity displayed by G␣2 was significantly enhanced by the presence of the AP-1 site (Fig. 4, compare B and C). Gal4-CREB also possessed a lower degree of intrinsic constitutive transactivation activity compared with G␣2, but had greater PKAinducible activity. It should also be noted that the Gal4-C/ EBP␤ fusion protein, G␤2, had no significant ability to mediate PKA responsiveness (Fig. 4, B and C).
Structure/Function Analysis of the cAMP-inducible Transactivation Properties of C/EBP␣-Several previous studies have characterized the domains of C/EBP␣ that mediate its constitutive transcriptional activity. We were interested to see whether the domains in C/EBP␣ that mediate cAMP-inducible transcription were similar or distinct from those involved in its constitutive transcriptional activity, as has been observed for CREB (31). Various carboxyl-and amino-terminal deletion mutants of the C/EBP␣ transactivation domain, as well as those containing internal deletions, were fused to the DNA-binding domain of Gal4 and tested both for their basal transcriptional activity and for their cAMP-inducible activity. All of the Gal4-C/EBP␣ fusion proteins shown in Fig. 5 were expressed at comparable levels in HepG2 cells as assessed by Western blot analysis using an antibody specific for Gal4 (data not shown). The PKA-inducible activities of these fusion proteins were tested using Ϫ109G3A1, a synthetic promoter that reconstitutes the cAMP response unit of the PEPCK promoter used previously to assess the activity of Gal4-C/EBP fusion proteins (21). However, because this promoter has a CRE and an AP-1 site, we tested the basal transcriptional activities of the Gal4-C/EBP␣ fusion proteins using Ϫ68G3, which has three Gal4binding sites linked to a minimal promoter, so the transcriptional activity measured was derived exclusively from the action of the Gal4-C/EBP␣ fusion proteins.
Our analysis first examined the activity of carboxyl-terminal deletions of C/EBP␣, and we identified three domains that contributed to its constitutive activity: amino acids 176 -217, 97-124, and 51-96 (Fig. 5). A putative attenuation domain, residing within amino acids 136 -175, was also identified, based upon the observation that the constitutive activity of N135 was 3-fold greater than that of the intact transactivation domain. It should be mentioned that examination of aminoterminal deletions was unsuccessful since all fusion proteins lacking the amino-terminal 50 amino acids showed no activity. 2 Internal deletion mutants demonstrated a critical role for amino acid residues 51-95 (Fig. 5). All of these findings are in 2 P. J. McFie and W. J. Roesler, unpublished observations.

FIG. 4. A Gal4-C/EBP␣ fusion protein can mediate PKA responsiveness and is augmented by AP-1.
A schematic of the promoters used in this experiment is shown (A). These reporter genes, Ϫ68G4A1 (B) and Ϫ68G4 (C), were constructed as described in the legend to Fig.  3, except that an oligonucleotide containing a Gal4-binding site was used in place of a C/EBP oligonucleotide. The CAT reporter plasmids (7 g) were cotransfected into HepG2 cells along with a PKA expression vector and, where indicted, an expression vector for Gal4-CREB (1 g), G␣2 (50 ng), or G␤2 (1 g). The CAT activity measured in the presence of the Gal4 fusion proteins, but in the absence of PKA expression (solid bars), is expressed relative to the CAT activity obtained with transfection of the CAT reporter plasmid alone, which was set arbitrarily at 1.0. The CAT activity achieved by overexpression by PKA is depicted by hatched bars. The values shown are the means Ϯ S.E. of at least three experiments.
The PKA-inducible activity of these fusion proteins was next examined. In general, the carboxyl-terminal deletion mutants that displayed constitutive activity also demonstrated PKAinducible activity. However, it was also evident that the carboxyl-terminal deletions had differential effects on the constitutive and inducible activities of C/EBP␣. Deletion of amino acids 176 -217 resulted in a loss of 65% of the constitutive activity of C/EBP␣ without any measurable loss of its ability to mediate PKA responsiveness (Fig. 5). Further deletion to amino acid 135 resulted in a large increase in constitutive activity without affecting its inducible transcriptional activity, and deletion of an additional 11 amino acids (N124) resulted in a decrease in constitutive activity without any effect on inducible activity. Furthermore, comparison of the properties of N175 with those of the internal deletion mutant ⌬113-134 show that the latter fusion protein has a higher constitutive activity than N175, but significantly lower PKA-inducible activity. Finally, comparison of N124 with the parent compound G␣2 demonstrated that the constitutive activity of C/EBP␣ can be compromised without necessarily affecting its PKA-inducible activity. It should be noted that the residual PKA responsiveness that was detected even when the Gal4 DNA-binding domain alone was expressed originates from the inherent activity of the test promoter that, due to the presence of the CRE, does confer some PKA inducibility (21,23).
A consistent observation made in the experiments shown in Fig. 5 was that all C/EBP␣ fusion proteins that were devoid of constitutive activity (N50, ⌬51-95, and ⌬51-111) also pos-sessed no measurable PKA-inducible transcriptional activity. The common region missing in all of these mutant proteins was the domain encompassing amino acids 51-95. This region contains the only conserved motif within the transactivation domain among C/EBP family members, consisting of two conserved boxes (A and B) separated by a spacer region ranging in length from 5 to 34 amino acids (18). The B box contains a homology box 2 core, which was previously identified as a transactivation domain residing within c-Fos, c-Jun, and c-Myc (32,33). This conserved motif in C/EBP␣, spanning amino acids 55-86, has been shown to mediate protein-protein interactions with the TATA-binding protein and with TFIIB, and mutation of conserved amino acids within this motif significantly reduced the constitutive activity of C/EBP␣ (18).
Because point mutations represent a less severe approach to protein structure/function analysis compared with deletion analysis, we decided to test a triple-point mutant for its PKAinducible activity. This C/EBP␣ mutant, termed Y67A,F77A, L78A, contains alanine substitutions at tyrosine 67, phenylalanine 77, and leucine 78 and was previously shown to possess constitutive activity that was ϳ10% of the wild type (18). We fused the transactivation domain (amino acids 6 -203) containing these mutations to the DNA-binding domain of Gal4 and tested its activity on the reporter plasmid Ϫ109G3A1. As shown in Table II ingly, G␣Y67A,F77A,L78A retained full capacity to enhance promoter activity in response to PKA (Table II), providing strong evidence that the constitutive and PKA-inducible transactivation activities of C/EBP␣ are mediated by distinct mechanisms.

DISCUSSION
While the PEPCK promoter is regulated by a number of hormones (reviewed in Ref. 34), the induction by cAMP has been intensively investigated due to the strong and rapid nature of the response and its tissue-specific characteristics (19,35). The CRE in this promoter was the first of its kind to be identified (26), and from it arose the discovery that many promoters contain this cis-element (36). A transcription factor that bound to the CRE, termed CREB, was subsequently identified, and it contains a serine residue that is phosphorylated by PKA, resulting in its activation (37). This initial simplified model of how cAMP responsiveness occurs has become significantly more complex with the discovery of a coactivator for CREB (38) and the finding that CREB is but one member of a large protein family (reviewed in Ref. 39). However, in the case of the PEPCK promoter, an additional layer of potential complexity was uncovered when it was shown that members of the C/EBP protein family can also bind to this CRE with high affinity (12,25), although their affinity for perfect consensus CREs in other promoters is relatively weak (24). Not only can C/EBP bind to the CRE in the PEPCK promoter, but it also produces transactivation, thereby creating a controversy as to whether CREB or a C/EBP protein is the physiologically relevant transcription factor at this site.
Several previous studies addressing this issue have produced conflicting data. For example, in support of a role for CREB at this site, our laboratory showed that (i) mutation of the CRE to a perfect consensus sequence, which resulted in a significant decrease in C/EBP binding affinity but had no effect on CREB binding affinity (24), did not alter the cAMP responsiveness of the PEPCK promoter; and (ii) a Gal4-CREB fusion protein, when tethered to the PEPCK promoter by replacement of the CRE with a Gal4-binding site, was able to reconstitute the cAMP responsiveness of the promoter (22). Conversely, in support of a role for a C/EBP protein, it was shown that (i) mutation of the CRE to a C/EBP-binding site, which reduced CREB binding affinity without affecting C/EBP binding affinity, did not reduce the cAMP responsiveness of the promoter (23); and (ii) mice that contain a targeted inactivation of the CREB gene have no detectable alterations in PEPCK gene expression or any apparent abnormalities in glucose homeostasis (40). In this paper, we present data that suggest that the binding of either CREB or C/EBP␣ to the CRE allows the promoter to retain its cAMP responsiveness (Fig. 2B), i.e. that C/EBP␣ can functionally substitute for CREB, which provides an explanation for the apparent contradictory data summarized above. The fact that C/EBP␤ can also bind to this CRE, but with a detrimental effect on cAMP responsiveness (Fig. 2B), suggests that the prevailing responsiveness of this promoter to cAMP can be exquisitely regulated depending on the in vivo concentrations of CREB, C/EBP␣, and C/EBP␤. An additional level of regulation has been suggested by the observations that both C/EBP␣ and C/EBP␤ can have their DNA-binding activity modulated by phosphorylation of specific serine residues lying within the basic region leucine zipper domain (41,42).
We also presented experimental data to support the hypothesis that C/EBP␣ may function independently as a modest cAMP-inducible regulator. One criticism of these data could be that we demonstrated cAMP-inducible activity for C/EBP␣ only on promoters containing multiple binding sites for either the intact protein or the Gal4 fusion counterpart. Indeed, we were able to detect only a residual amount of PKA responsiveness when a single C/EBP site was linked to a minimal promoter. 2 However, it should be noted that similar observations have been made with promoters containing a single CREB molecule tethered to the promoter. Our laboratory has previously shown that the PEPCK promoter CRE, when tested independently, is essentially devoid of PKA-inducible activity, as is a single molecule of a Gal4-CREB molecule when recruited to a promoter (22)(23)(24). In fact, it takes the multimerization of three CRE cis-elements before detectable activity of the CRE manifests itself (24). As reviewed recently (43), it is in fact a common observation that cAMP responsiveness (and responses to other hormones) is often mediated by the synergistic interactions of several cis-elements within a promoter, even those that contain typical CREs. Employing several different cis-elements in the hormonal responsiveness of a promoter may offer unique regulatory opportunities, including an expansion of the range of responses and the coordination of information from several signaling pathways into an integrated transcriptional response.
If C/EBP␣ is a cAMP-activated nuclear regulator, what is the molecular mechanism whereby it mediates this activity? While we do not have an answer to this question at present, it is known that C/EBP␣ is not phosphorylated by PKA to any significant extent (41). It is possible that some as yet unidentified coactivator for C/EBP␣, which itself can be phosphorylated and activated by PKA, allows C/EBP␣ to display this hormone-inducible activity. Recently, the CREB-binding protein (CBP), which is a coactivator for CREB (38), was implicated as a possible coactivator for C/EBP␣ (44), although no direct protein-protein interaction between CBP and C/EBP␣ has been reported. Interestingly, the transactivation potential of CBP can apparently be induced by PKA (38). Thus, CBP may fit the role of this putative C/EBP␣ coactivator that mediates the PKA-inducible activity. Arguing against this hypothesis, however, is that C/EBP␤, which does not function as a cAMPinducible transcription factor (see Figs. 1, 2B, and 4 (B and C) and Table I), also appears to utilize CBP as a coactivator (38). How CBP could confer PKA inducibility to one isoform and not the other is not readily obvious, although one could speculate that the nature of the protein-protein interaction could differ, in one case allowing the expression of the inducible activity and in the other instance masking it. Answers to these questions should be forthcoming as the coactivator role of CBP for these two C/EBP isoforms is solidified and the molecular nature of the protein-protein interactions is characterized.
Any model developed to define the mechanism whereby C/EBP␣ is able to mediate cAMP responsiveness must incor- porate the observation that the constitutive and PKA-inducible transcriptional activities of this factor appear to be exerted via at least partially independent mechanisms. The first evidence for this came from experiments where we observed that G␣2, when expressed in the choriocarcinoma JEG-3 cell line (i.e. non-hepatoma cells), exhibited PKA-inducible activity, but little if any constitutive activity (21). In the present study, we have presented additional data in support of this hypothesis by showing that the C/EBP␣ mutant Y67A,F77A,L78A remained fully competent as a cAMP-dependent transactivator even though its constitutive transcriptional activity was significantly inhibited (Table II). Deletion analysis (Fig. 5) also provided some indication of the distinct nature of the constitutive and inducible activities of this factor. While our analysis cannot identify a precise "domain" that confers the inducible activity, our data do allow us to conclude that this activity is fully contained within amino acids 6 -124 since N124 remains fully functional with respect to this activity. Since N96 lost all inducible activity, it was interesting to speculate that region 96 -124 was the PKA-inducible domain. However, our attempts to analyze this region in isolation were unsuccessful since the presence of amino acids 6 -50 was absolutely required to detect any activity of the Gal4 fusion proteins. We also recognize that the lack of activity of N96 does not necessarily indicate the lack of involvement of any of the remaining amino acids in the PKA-inducible activity since the loss of activity could be due solely to alterations in tertiary structure as a result of the deletion. Another conclusion that we can make is that the sole conserved transactivation motif among C/EBP family members, which lies between amino acids 55 and 86 (18), is likely not involved in the PKA-inducible activity of C/EBP␣. The triple mutant that we tested contained mutations in three conserved amino acids found in all C/EBP isoforms, and the amino acids at positions 77 and 78 are part of the homology box 2 core (18). Nerlov and Ziff (18) demonstrated that these residues are involved in the constitutive activity of C/EBP␣, apparently exerting their effects by participating in protein-protein interactions with the TATA-binding protein and TFIIB. Since C/EBP␤ also contains these conserved amino acids but does not act as a cAMP-inducible factor, one would predict that this conserved motif, which is found in all C/EBP isoforms, does not participate in the cAMP-inducible activity of C/EBP␣. Indeed, our data support this hypothesis, allowing us to conclude that cAMP-inducible activity is not mediated via interactions with the TATA-binding protein or TFIIB, but via interactions with an associated coactivator and/or another target within the preinitiation complex. There is ample evidence to indicate that C/EBP␣ has several important biological roles, beyond being simply a constitutive transactivator. As evidenced in this paper, C/EBP␣ can function as an effector molecule in the cAMP signaling pathway, leading to the alteration of expression of target genes. Additionally, we have recently shown that C/EBP␣ can participate in mediating triiodothyronine responsiveness (45). Given its tissue-limited expression pattern (3), it also contributes to the tissue-specific pattern of gene expression and tissue differentiation, particularly adipose. Since the expression of C/EBP␣ is temporally regulated, with no accumulation occurring until just prior to birth (3), it also appears to function as a developmental transcription factor. Its involvement in the activation of the PEPCK gene at parturition (46) plays a particularly critical role in providing the newborn with the capacity for glucose synthesis, as this gene codes for the enzyme generally thought to be the rate-limiting enzyme of this pathway. Many of these important biological roles were highlighted by examination of mice that contained a targeted deletion of the C/EBP␣ gene or of adult mice with the gene conditionally disrupted. The knockout mice died within hours after birth of hypoglycemia; displayed no ability to store liver glycogen or adipocyte lipid; and had reduced expression of genes for glycogen synthase, PEPCK, glucose-6-phosphatase, and uncoupling protein of brown adipose tissue (11). Similar alterations in gene expression were observed in the conditional knockout mice (47). The findings of the present paper emphasize an additional role of this transcription factor, that of a hormone-responsive activator, and provide further evidence of the important role it plays as a central regulator of energy homeostasis (48).