Activating Transcription Factor-2 Regulates Phosphoenolpyruvate Carboxykinase Transcription through a Stress-inducible Mitogen-activated Protein Kinase Pathway*

Several protein-nucleic acid complexes are observed when nuclear extracts from hepatoma cells are assayed for binding to the cAMP response element found in the phosphoenolpyruvate carboxykinase-cytosolic (PEPCK-C) promoter. Although cAMP response element-binding protein and CCAAT/enhancer binding proteins α and β have been identified as liver factors that bind this motif, an uncharacterized, slower migrating complex was also observed. We identify activating transcription factor-2 (ATF-2) as the factor in this complex and show that ATF-2 stimulates expression from the PEPCK-C promoter. ATF-2 is a basic-leucine zipper transcription factor and a target for stress-activated protein kinases. We demonstrate that p38β mitogen-activated protein (MAP) kinase augments ATF-2 transactivation activity on the PEPCK-C promoter, which is consistent with the interpretation that PEPCK-C promoter activity is maintained under stress through a p38 MAP kinase dependent pathway. In this regard, we show that treatment with sodium arsenite, a known activator of p38 MAP kinases, also stimulates expression from the PEPCK promoter. These results show that ATF-2 can stimulate transcription of the PEPCK-C promoter and support a role for stress inducible kinases in the maintenance of PEPCK-C expression.

The minimal sequence of the PEPCK-C promoter sufficient for reduplication of hormonal regulation in liver has been mapped to the region encompassing Ϫ460 to ϩ73 (7)(8)(9)(10), and many of the transcription factors that bind elements in this region have been identified (2,11). For example, glucagon secretion leads to an increase in cAMP levels, which exerts effects by inducing factors that bind to the element denoted cyclic AMP response element I (CRE-1) and may also induce factors that bind to the element denoted P3. Prima facie, CREB might be anticipated as the primary factor impacted by increased cAMP levels. However, PEPCK-C expression was reported to be normal in CREB knockout mice (12). Binding to CRE-1 has also been reported for C/EBP␣ (13), C/EBP␤ (14), AP1 (15), and D-site binding protein (16). In fact, evidence from gene deletion experiments supports a role for C/EBPs ␣ and ␤ in the regulation of PEPCK-C expression (17).
The PEPCK-C gene is essential in humans (18), and its expression is not only maintained but, in fact, induced following partial hepatectomy (19). We uncovered the PEPCK-C CRE-1 motif TTACGTCA during a search of promoter sequences as a perfect match to the consensus reported for ATF-2 homodimers (20). ATF-2, a basic-leucine zipper transcription factor, is expressed in liver tissue (21) and exhibits increased DNA binding and transcriptional activation activities following phosphorylation of specific residues by p38 MAP kinases (22) or c-Jun N-terminal MAP kinases (JNK) (23,24). Because PEPCK-C transcription increases after partial hepatectomy (19), a stress condition, we considered that ATF-2 may contribute to regulation through the CRE-1 element. We show here that ATF-2 binds the PEPCK-C CRE-1 element, that sense ATF-2 expression correlates with increased promoter activity, and that antisense ATF-2 expression correlates with decreased promoter activity. Furthermore, expression of p38␤ MAP kinase, a known modifier of ATF-2, correlates with increased reporter activity, while cotransfection of p38␤ MAP kinase and ATF-2 shows augmented transcriptional activation. These results support a role for ATF-2 in the maintenance of PEPCK-C expression in the liver that may have special relevance under stress conditions.

EXPERIMENTAL PROCEDURES
Cell Lines, Culture Conditions, and Plasmid Constructions-Fao cells are a hepatoma derivative that exhibit many of the characteristics of differentiated hepatocytes and were kindly provided by Dr. Mary Weiss (Institut Pasteur) (25). Fao cells were maintained in Coon's modified, Ham's F-12 medium (Life Technologies, Inc.) supplemented with 5% fetal calf serum, penicillin, streptomycin, and glutamine. The ATF-2 expression plasmid has been described (21). The human p38␤ expression plasmid was provided by Jiahuai Han (The Scripps Research Institute) (26). The Reporter plasmid PEPCK Ϫ275 was prepared by polymerase chain reaction of rat genomic DNA using 5Ј primer GG-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
TACC ACA GTC AGCG GTC AAA GTT TAG TCA ATC and 3Ј primer GCTCGA GAG ATC TCA GAG CGT CTC GCC GG, which encompasses positions Ϫ275 through ϩ73 of the PEPCK-C promoter using numbering according to Roesler et al. (27).
For sodium arsenite treatment, 1 ϫ 10 6 Fao cells were plated on a 10-cm dish, transfected by standard calcium phosphate treatment, washed, and fed complete medium for 48 h. Medium was supplemented with 50 M sodium arsenite (Sigma) for 6 h prior to cell harvest.
Gel Shift Analysis-Full-length C/EBP␣ and ATF-2 proteins were prepared by in vitro transcription-coupled translation (TNT, Promega, Madison, WI.) under conditions described by the manufacturer. A double-stranded oligonucleotide encoding the PEPCK-C CRE-1 sequence (promoter positions Ϫ99 to Ϫ76) was used for gel shift analysis: 5Ј CCGGCCCC TTACGTCA GAGGCG. Binding reactions were assembled without probe and held 5 min on ice followed by 5 min at room temperature. Probe was added with further room temperature incubation for 30 min. Samples were separated in 4% acrylamide, 0.5ϫ TBE (0.045 M Tris, 0.045 M boric acid, 1.0 mM EDTA (pH 8.0) gels run at 200 V constant voltage (28).
Transient Transfection and Luciferase Assays-Fao cells were transfected by the standard calcium phosphate method (28). Cells were incubated with DNA precipitates for 16 h, washed, and maintained in complete medium 48 h prior to harvest. Relative luciferase and ␤-galactosidase activities were determined as described (21). Basal promoter activity is reported as the activity observed after transfection of the reporter plus an appropriate amount of empty expression vector. In all cases, transfection data represent the mean of three independent experiments, where error bars indicate the S.D. of the mean.

RESULTS AND DISCUSSION
ATF-2 Binds the PEPCK-C CRE-1 Sequence-The transcription factors CREB (29), C/EBP␣ (13), C/EBP␤ (14), AP1 (15), and D-site-binding protein (16) have been reported to bind the PEPCK-C CRE-1 element (Fig. 1A) (13). Notably, the CRE-1 site matches the consensus sequence reported for ATF-2 homodimers ( Fig. 1A) (20) rather than the CREB or AP1 consensus sequences. Because ATF-2 expression was recently demonstrated in rat liver (21), we wondered if it bound the CRE-1 site. As shown in Fig. 1B, the slowest migrating shift produced with hepatoma nuclear extracts (lane 1) is supershifted when anti-ATF-2 serum is included in the binding reaction (lane 3). By comparison, C/EBP␣, a factor reported to bind this site, also produces a supershift complex (lane 2) upon inclusion of specific antiserum. However, the C/EBP␣ supershift complex is not abundant and the band from which the supershift complex derives is not obvious. To demonstrate the expected migration positions for these protein-nucleic acid complexes, shift reactions were formed with full-length C/EBP␣ and ATF-2 prepared by in vitro transcription-coupled translation (lanes 4 and 5, respectively). These results demonstrate that ATF-2, like C/EBP␣, binds the CRE-1 element in the PEPCK-C promoter.
Modulation of ATF-2 Expression Levels Correlates with Reporter Gene Activity-To determine what effect, if any, ATF-2 might have on PEPCK-C promoter activity, we transfected Fao hepatoma cells with a luciferase reporter construct encompassing Ϫ275 to ϩ73 of the PEPCK promoter ( Fig. 1A) (27). This region contains the CRE-1 motif as well as the P3(I) motif, both reported binding sites for C/EBP␣. As shown in Fig. 2A, luciferase activity increased up to 4-fold with increasing amounts of transfected sense ATF-2 expression plasmid.
To extend these observations, we cotransfected the reporter vector with increasing amounts of antisense ATF-2 expression plasmid. As shown in Fig. 2B, basal reporter activity decreased about 2-fold when cells were cotransfected with plasmid encoding antisense ATF-2. The Western blot (Fig. 2C) shows nuclear extracts from a representative transfection experiment and demonstrates that ATF-2 protein levels decrease upon antisense ATF-2 expression (lane 2) and increase upon sense ATF-2 expression (lane 3) relative to steady state ATF-2 levels (lane 1). These results are consistent with a role for ATF-2 in the regulation of PEPCK-C gene expression.
p38␤ MAP Kinase Augments ATF-2-dependent Transcriptional Activation-ATF-2 is a known substrate for p38 MAP kinases (26), a subfamily of stress-activated protein kinases in the MAP kinase family (30 -33). Therefore, we tested the effect of p38␤ expression on a reporter construct containing the minimal thymidine kinase promoter modified by proximal insertion of two ATF-2 consensus sites (2ϫ CRE-1-Luc) (21). As shown in Fig. 3A, p38␤ MAP kinase expression correlated with an 8-fold increase in 2ϫ CRE-1-Luc reporter activity. This is comparable with expression of ATF-2 alone, which stimulates activity approximately 10-fold. When ATF-2 and p38␤ MAP kinase are cotransfected, reporter activity increased about 24fold. In contrast, expression of a transfected p38␤ MAP kinase "dead" mutant showed activity that was comparable with basal levels (data not shown). These results suggest that p38␤ MAP kinase augments ATF-2 activity in Fao hepatoma cells.
We next tested the PEPCK Ϫ275 reporter, which shows much higher basal levels of transcription despite the fact that cultures were propagated in the presence of complete medium (serum and glucose), repressing conditions for PEPCK-C expression. As shown in Fig. 3B, p38␤ MAP kinase alone stimulates transcription about 2-fold. This result is comparable with the 2-fold effect observed upon expression of ATF-2 alone. Upon cotransfection, ATF-2 and p38␤ MAP kinase stimulate PEPCK Ϫ275 reporter activity about 5-fold. These results are analogous to those obtained using the minimal promoter/reporter construct, although they show lower levels of transcriptional activation. This is likely due to endogenous factors that contribute to regulation of the PEPCK-C promoter. Given the level of induction observed, the results are consistent with the notion that ATF-2 functions in maintenance of PEPCK-C transcription, likely during conditions that activate a stress-inducible MAP kinase pathway(s).
As a specificity control for the effects we observed with p38␤ MAP kinase, we cotransfected protein kinase A or p42 ERK-1 with ATF-2. As shown in Fig. 4, expression of protein kinase A alone showed reporter activity twice basal. This is not surprising, as PEPCK-C is a known target for cAMP mediated effects. When ATF-2 was cotransfected with PKA, activity was 2.7 times basal, which is equivalent to ATF-2 alone (2.5-fold). Similar results were obtained with p42 ERK-1, which shows activity that is equivalent to basal (1.5 times basal). When ATF-2 is cotransfected, reporter activity increases 2.6-fold, again, equivalent to ATF-2 alone. These observations argue that the transcriptional augmentation we observed with ATF-2 and p38␤ MAP kinase is not generalizable to any protein kinase, but is restricted to stress activated protein kinases.
An Oxidative Stress That Activates p38 MAP Kinases Also Stimulates PEPCK Reporter Activity-Because sodium arsenite activates p38 MAP kinases (32), a finding we confirmed in Fao cells using antipeptide sera specific for phosphorylated p38 MAP kinase (data not shown), we wondered if PEPCK-C pro-  moter activity would be impacted. As a first test, Fao cells transfected with the PEPCK Ϫ275 reporter gene were cultured 48 h in complete medium followed by treatment with 50 M sodium arsenite for 6 h. As shown in Fig. 5, promoter activity after arsenite treatment was comparable with that observed following transfection of ATF-2. However, when ATF-2 transfectants were subjected to treatment with sodium arsenite, a larger increase (4.4-fold) in luciferase activity was observed. Importantly, transcriptional effects observed after arsenite treatment were blocked in cells transfected with an antisense ATF-2 expression plasmid. These results are consistent with the interpretation that stressors, which activate the p38 MAP kinase pathway, contribute to maintenance of PEPCK-C transcription.
The CRE-1 site located between Ϫ100 and Ϫ80 of the PEPCK-C promoter is a perfect match to the consensus binding sequence reported for ATF-2 homodimers (20) and ATF-2 is one of the transcription factors in hepatoma cell nuclear extracts that binds this sequence in electrophoretic mobility shift assays. Consistent with this, sense and antisense ATF-2 transfections increase and decrease, respectively, reporter gene activity driven by the PEPCK-C promoter. Preserving glucose homeostasis is crucial to organismal survival, as PEPCK-C catalyzes a step regulating entry of metabolites into the gluconeogenic pathway. This reaction is a liver-specific function that is maintained during stress, for example, PEPCK-C mRNA increased 5-fold following two-thirds resection of the liver (19). Despite a previous report concluding that oxidative stress induced by sodium arsenite and/or hydrogen peroxide selectively repressed PEPCK-C expression in hepatoma cells (34), our observations are consistent with a function for ATF-2 in maintaining PEPCK-C expression during stress conditions. Because of grossly differing choices of methodology it is difficult to reconcile the opposing results, but should be noted that the previous report initiated assays of PEPCK-C expression from conditions that are maximally inducing, whereas our studies initiated assays from culture conditions that repress promoter activity to basal levels.
ATF-2 is expressed in most tissues, is a substrate for stressinducible MAP kinases (both the p38 and the stress-activated protein/JNK subfamilies), and exhibits increased transcription activating and DNA binding activities in the phosphorylated form (23,24). The p38 MAP kinases and the JNKs modify the same residues in ATF-2, Thr 69/71 and Ser 90 (35). Although the studies presented here were limited to p38 MAP kinases, we expect that JNK MAP kinases would result in a similar augmentation of ATF-2-dependent transcription. Starting from conditions that are repressing for endogenous PEPCK-C activity, serum and high glucose, we observed stimulation of PEPCK-C promoter activity by transfection of ATF-2 alone or p38␤ MAP kinase alone. When ATF-2 and p38␤ MAP kinase were cotransfected, augmented transcriptional activation was observed. Failure to observe augmented transcription with PKA and ATF-2 or with ERK-1 and ATF-2 indicates that the transcriptional increase was specific. In addition, treatment of Fao cells (in the presence of serum) with sodium arsenite, an activator of p38 MAP kinases, resulted in a similar increase over basal promoter activity. The fact that this effect was eliminated upon expression of antisense ATF-2 again indicates specificity. These results support a role for ATF-2 in the maintenance of PEPCK-C expression in hepatocytes in response to extracellular signals that activate stress-inducible MAP kinases.
Our studies are consistent with available results from gene deletion experiments. Homozygous deletion of the genes encoding transcription factors CREB, CREM, C/EBP␤, and D-sitebinding protein show no obvious defect in glucose homeostasis (2). In contrast, deletion of the gene for C/EBP␣ results in hypoglycemia, failure to store liver glycogen, failure to develop fat tissue, and lethality. (17). However, C/EBP␣ is neither an acute phase-reactive protein (36), nor does it's expression increase following partial hepatectomy (19). In fact, C/EBP␣ decreases under both of these conditions, consistent with observations regarding its antiproliferative effects in cell lines (37)(38)(39). Although C/EBP␤ expression increases after partial hepatectomy (19), it is not a known substrate for stress-induced MAP kinases, suggesting that C/EBP proteins and CREB/ CREM proteins are not likely activators of PEPCK-C expression during conditions of stress. Although transcriptional effects observed in cell culture systems do not always faithfully reflect regulation as it occurs in vivo (2), our studies support a role for ATF-2 as a regulator of the hepatocyte PEPCK-C promoter during conditions of stress.