Activating Transcription Factor-2 Mediates Transcriptional Regulation of Gluconeogenic Gene PEPCK by Retinoic Acid

A full-length rat ATF-2 cDNA was generated by RT-PCR and verified by sequencing. ATF-2 was subcloned into pCMX1 as BamHI (21) and BamHI/KpnI fragments, respectively. pCMX1 was a gift from Catherine Thompson (Carnegie Institution of Washington) and was used in coupled transcription-translation reactions. In vitro translation products were verified by [³⁵S]Met incorporation and SDS-PAGE analysis. The human p38β expression plasmid was provided by Jiahuai Han (The Scripps Research Institute) (22). The reporter plasmids PEPCK-275 and PEPCK-543 were constructed by PCR amplification of rat genomic DNA encompassing positions −275 or −543 through +73 of the PEPCK promoter. The cAMP response element (CRE)-1 mutant PEPCK-275 reporter was prepared by standard mutagenesis, changing CCGGCCCCTTACGTCAGAGGCG to CCGGCCCCTTTTTTCAGAGGCG.

HepG2 cells were harvested in Nonidet P-40 lysis buffer, and 30–50 μg nuclear protein was fractionated by SDS-PAGE. Proteins were electroblotted to Immobilon-P (Millipore), and membranes were blocked in Tris-buffered saline, 0.02% Tween 20, containing 5% nonfat milk. Specific antibody for ATF-2 was previously described (23), and antibodies for phospho-ATF-2, p38β, and phospho-p38β were obtained from Santa Cruz Biotechnology. Secondary detection utilized goat anti-rabbit-conjugated horseradish peroxidase (Amersham Pharmacia Biotech) visualized with enhanced chemiluminescence reagents (Amersham Pharmacia Biotech) essentially as described (23).

Nuclear extracts were prepared from HepG2 cells following stimulation with hormones or kinase inhibitors as indicated in the figure legends. Approximately 10 μg nuclear extract was immunoprecipitated with an ATF-2-specific antibody and incubated with a probe. A double-stranded oligonucleotide encoding the PEPCK CRE-1 sequence (promoter positions −99 to −76) was used for gel shift analysis: 5′ CCGGCCCCTTACGTCAGAGGCG (6). 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, 0.5× TBE (0.045 mol/l Tris, 0.045 mol/l boric acid, and 1.0 mmol/l EDTA [pH 8.0]) gels run at 200 V constant voltage (24).

HepG2 cells were transfected by the standard calcium phosphate method (24). Cells were incubated with DNA precipitates for 16 h, washed, and maintained in complete medium 48 h before harvest. Relative luciferase and β-galactosidase activities were determined as described (23). 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.

Cells were seeded with growth medium supplemented with 10% fetal bovine serum and 1% antibiotics and cotransfected with expression vectors encoding Gal4-DNA-binding domain fusions (pCMX/Gal4N-ATF-2N) and VP16-activation domain fusions (pCMX/VP16/-, pCMX/VP16-ATF-2C, or pCMX/VP16-ATF-2-full) as well as the previously described Gal4-tk-luc reporter plasmid. After 48 h, cells were harvested and the luciferase activity was normalized to the β-galactosidase internal standard. All the results represent the average of at least three independent experiments.

In a previous study, we showed that the HBx transactivator encoded by hepatitis B virus increased RXR-dependent transactivation of the PEPCK promoter in an RA-dependent fashion through direct protein-protein interaction (45). To extend these studies, we have examined the effect of additional nuclear receptor ligands (RA, 9-cis-RA, dexamethasone, and thyroid hormone [T3]) on transactivation activity of the PEPCK-275 and PEPCK-543 promoters. A schematic representation of these luciferase reporter constructs is shown at the bottom of Fig. 1. Region A contains the glucocorticoid response unit (GRU) composed of two glucocorticoid regulatory elements, three accessory factor-binding sites, and a CRE. This region provides several factor-binding sites for RAR, RXR, GR, TR, C/EBP, and HNF-3 (12). Region B contains a CRE (CRE-1) (−99 to −76) and is immediately adjacent to a NF-1-binding site (12).

The hepatoma cell line HepG2 was transfected with reporter plasmids PEPCK-275-luc and PEPCK-543-luc (Fig. 1A) or PEPCK-CRE-1-luc (Fig. 1B) with increasing concentrations of RA. RA treatment stimulated the activity of PEPCK-275 and PEPCK-543 reporters in a dose-dependent manner (Fig. 1A). It is noteworthy that the promoter region of PEPCK-275 does not have an RARE sequence for RAR/RXR binding. Interestingly, Scott et al. (44) also showed that RA increased transactivation of a PEPCK-306 promoter in H4IIE cells ∼3.3-fold.

To determine whether the CRE-1 sequence was responsible for RA-induced transactivation of the PEPCK-275 reporter, a mutant PEPCK −275 reporter was constructed by changing the CRE-1 sequence (research design and methods). The CRE-mutant PEPCK −275 promoter was not responsive to increasing concentrations of RA (Fig. 1B). However, transient transfection (Fig. 1B) shows that RA activated the PEPCK CRE-1 promoter essentially equivalently to activation of the PEPCK −275 reporter (Fig. 1A). These results indicate that RA can stimulate PEPCK gene transcription through the CRE-1 site in addition to the RARE-1 site.

To identify the transcription factor mediating RA responsiveness on the PEPCK-275 promoter, expression plasmids encoding ATF-2, C/EBPα, and c-Jun were cotransfected with the PEPCK-275 reporter in the presence or absence of RA. As shown in Fig. 2A, transfection of ATF-2 largely increased PEPCK promoter activity in a dose-dependent manner. In contrast, transfection of C/EBPα or c-Jun failed to show significant RA-dependent changes in transactivation. 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 approximately two- to fourfold when cells were cotransfected with plasmid encoding antisense ATF-2. The Western blot (Fig. 2B) shows nuclear extracts from a representative transfection experiment and demonstrates that ATF-2 protein levels decrease upon antisense ATF-2 expression (Fig. 2B, lanes 2–4) relative to steady-state ATF-2 levels (Fig. 2B, lane 1). These results are consistent with a role for ATF-2 in the regulation of PEPCK gene expression following RA treatment.

Transcriptional activation by ATF-2 is enhanced by phosphorylation of NH₂-terminal threonine residues through several mitogen-activated protein kinases. We tested whether ATF-2 phosphorylation is increased by RA treatment by Western blot analysis with an anti-phospho-ATF-2-specific antibody. Hepatoma cells were incubated with 1 μmol/l RA for varying times (3–48 h), lysed, and total cell lysates analyzed by SDS-PAGE immunoblots with an antibody specific for phosphorylated/activated ATF-2. As shown in Fig. 3, RA treatment potentiated strong phosphorylation of ATF-2 in a time-dependent manner, reaching a maximum at ∼24 h. Hyperphosphorylation is still observable, even after 48 h (Fig. 3). In contrast, the amount of total ATF-2 protein is similar in RA-stimulated and -unstimulated cells. These results show that ATF-2 phosphorylation is induced after RA treatment.

RA treatment of cells induces a time- and dose-dependent phosphorylation of p38 kinase, and such phosphorylation results in activation of its catalytic domain (25). To identify whether the increased phosphorylation of ATF-2 after RA treatment results from activation of the p38β kinase pathway, we cotransfected either a wild-type p38β kinase or an inactive p38β kinase mutant (T188A, Y190F), which cannot be phosphorylated by p38β kinase kinase (22), into cells. As shown in Fig. 4A, CRE-dependent transactivation after RA treatment increased with p38β kinase overexpression but not by JNK overexpression. In addition, we used a p38β-mutant (kinase dead) expression vector in the transient transfection assay, which shows that the RA effect was abrogated upon expression of the dominant interfering kinase (Fig. 4A). Finally, a p38 kinase selective chemical inhibitor, SB203580 (20 μmol/l), blocked the RA-dependent transactivation of −275 PEPCK, whereas an ERK inhibitor, PD98059, essentially showed no effect (Fig. 4A). To further confirm that p38β kinase activation is involved in the RA response, we assayed for the activation of p38β kinase itself by phosphorylation of the kinase after RA treatment. Total cell extracts were prepared at the times indicated and assayed for expression and phosphorylation of p38β kinase. As shown in Fig. 4B, although p38β kinase expression levels did not change, its phosphorylation state increased from 6 to 12 h after RA treatment. This time course correlates with the phosphorylation of ATF-2, as shown in Fig. 3. Hence, these data provide direct evidence that RA stimulation leads to p38β kinase activation and ATF-2 phosphorylation, potentiating the transactivation activity of ATF-2.

Previous studies showed that the basic-leucine zipper (bZIP) region of ATF-2 interacts with the amino-terminal region intramolecularly (26), inhibiting the transactivation activity of ATF-2. From this result we proposed that RA signaling might block this intramolecular interaction through phosphorylation of ATF-2. To test this possibility, we investigated the effect of RA on the intramolecular interaction between the amino- and carboxyl-terminal regions of ATF-2 by using the mammalian two-hybrid assay. Coexpression of Gal4-ATF-2N and VP16-ATF-2C or VP16-ATF-2-full enhanced Gal4-tk-luc-dependent transactivation, whereas RA treatment clearly inhibited the transactivation mediated by association of amino-terminal ATF-2 sequences (Gal4-ATF-2N) with carboxy-terminal ATF-2 sequences (VP16-ATF-2-full) (Fig. 5A). Given that p38β kinase modulates ATF-2-dependent transactivation of the PEPCK promoter after RA treatment (Fig. 4A), we tested the effect of constitutively active p38β kinase in the mammalian-2 hybrid assay. The assay shown in Fig. 5A was repeated with p38β kinase coexpression, the expression level of which is shown by p38β kinase-specific antibody (Fig. 5B, bottom). As shown in Fig. 5B, p38β kinase expression largely inhibited transactivation by ATF-2 in the mammalian-2 hybrid assay. These results strongly suggest that RA signaling inhibits the intramolecular interaction between the NH₂-terminal and COOH-terminal regions of ATF-2 by a mechanism involving the p38β kinase pathway.

The results shown in Fig. 5 suggest that the DNA binding domain of ATF-2 would be more exposed after RA treatment, possibly leading to more efficient binding to its cognate DNA element. To analyze whether increased DNA binding activity of ATF-2 may be involved in the RA-dependent stimulation of the PEPCK promoter activity, electromobility shift assay analysis was used. Nuclear extracts were prepared from control cells, RA-treated cells, or cells treated with phosphatase inhibitors and/or protein kinase inhibitors. Immunoprecipitates for ATF-2 from 10 μg of nuclear extract, as indicated, were incubated with a ³²P-labeled probe spanning the PEPCK CRE-1 promoter region between −99 and −76. As shown in Fig. 6, increasing concentrations of RA led to increased DNA binding activity by ATF-2. As would be predicted, ATF-2 binding was inhibited by the p38β kinase inhibitor SB203580, but not by the ERK inhibitor PD98059. This result is also consistent with induced DNA binding activity for ATF-2 through a p38β kinase pathway (Fig. 6). Similarly, the phosphatase inhibitor okadaic acid weakly potentiated CRE-ATF-2 complex formation. These results are consistent with a mechanism whereby the DNA binding activity of ATF-2 is induced through activation of a p38β kinase pathway downstream of RA treatment.

RA-transduced signals induce specific gene transcription by activating RAR binding to its cognate DNA element as a heterodimer with RXR. In addition, RA has been reported to mediate signal transduction through what has been termed a nongenomic mechanism (25). But, the mechanism of activation of Rac by RA is not clear and whether it involves a classical RARE-mediated genomic effect or occurs through some nongenomic action of RA will need to be determined in future studies. In nonsmall cell lung cancer, RARs and RXRs are expressed but are not transcriptionally activated by RA, such that RA does not increase the expression of genes typically activated by RXR-RAR heterodimers (20). Thus, these cells exhibit a transcriptional defect specific to retinoid nuclear receptors. In this study, we show that RA activates a p38β kinase pathway, subsequently phosphorylating ATF-2 for increased transcriptional regulation of the PEPCK promoter.

The transcription factors CREB (27), C/EBPα (28), C/EBPβ (29), AP1 (30), and d-site binding protein (31) have been reported to bind the PEPCK CRE-1 element. We previously noted that the CRE-1 site matches the consensus sequence reported for ATF-2 homodimers (8), rather than CREB or AP1 consensus elements previously noted (32). ATF-2, a bZIP transcription factor, is highly expressed in liver and brain, tissues central to glucose homeostatic mechanisms. Most bZIP transcription factors including Jun, CREB, C/EBPβ, ATF-2, and C/EBP homology protein are known substrates for mitogen-activated protein (MAP) kinases, which leads to induction of transcriptional activity. There are three main classes of MAP kinases, ERKs, stress-activated protein kinases, JNKs, and p38-MAPKs. ERKs are preferentially activated in response to growth factors, cytokines, and mitogens via the well-known Ras/Raf-1/MEK pathway. JNKs and p38-MAPKs are activated in response to a variety of cell stresses, including ultraviolet irradiation, proinflammatory cytokines, protein synthesis inhibitors, osmotic shock, and chemical stress, such as that induced by H₂O₂ and other reactive oxygen species. These kinases are activated in response to a variety of stimuli and mediate signals important for the generation of various biological responses (33). Two members of the MAP kinase family, ERK2 and JNK, have been previously shown to be involved in the generation of retinoid responses (34). ERK2 is activated in response to RA treatment in the HL-60 acute myelogenous leukemia cell line (20), where activation is required for the induction of RA-dependent cell differentiation and growth arrest. The engagement of the p38 signaling cascade by RA is of considerable interest, as this pathway is critical for the generation of signals required for important biological activities in response to stress and/or engagement of certain cytokine receptors. These include activation of transcription factors (35), transcriptional regulation (36), and induction of cytokine production (35). Therefore, our finding that the p38β kinase pathway is activated during RA treatment of cells led us to further studies to determine the functional role of p38β kinase and ATF-2 activation in the induction of RA responses for regulating PEPCK gene transcription.

Transcription of the PEPCK gene is enhanced by glucagon and glucocorticoids and inhibited by insulin in a dominant manner (37). These actions of insulin are phosphatidylinositol-3 kinase dependent and are not blocked by inhibition of the p70 S6 kinase or the ERK pathways (38–40). Interestingly, activation of the p38β kinase pathway led to increased PEPCK gene transcription after RA-dependent phosphorylation and activation of ATF-2 (Fig. 4). It is therefore important to determine whether the p38β pathway affects the insulin-dependent regulation of the PEPCK gene.

Gel shift analysis is commonly used to evaluate the phosphorylation status of ATF-2 (Fig. 6). It is important to note that RA induces a conformational change in the ATF-2 protein, such that homodimers form inefficiently, resulting in avid formation of heterodimers and binding a cognate DNA element, leading an efficient inducible transcription of PEPCK gene in a dominant manner. As shown in Fig. 5, RA signaling decreased the inactive ATF-2 conformation by inhibiting the intramolecular interaction. We already presented the formation of ATF-2:C/EBPα heterodimers, describing cross-family dimerization with ATF-2, a possible general property for C/EBP family proteins (23). The RA-induced inhibition of the intramolecular interaction of ATF-2 may allow the heterodimerization of ATF-2 with a variety of transcription factors and the functional interaction with coactivators. Heterodimer formation of ATF-2 with other transcription factors is appealing from a regulatory viewpoint because of the asymmetry that is generated. By selecting for asymmetric DNA elements, heterodimers bind their target in an orientation-dependent fashion, presenting distinct surfaces for interaction with proteins bound to adjacent DNA sites. Both the position and the orientation of protein-binding sites within some enhancer elements are important for the formation of what has been termed the “stereospecific” complex (41). Although the function of heterodimers comprising ATF-2 as a dimeric partner is not known, it has been proposed that cross-family heterodimers serve as a common target for the integration of signals arising from different extracellular stimuli. In this model, each subunit of the heterodimer would be modified by a unique protein kinase, p38β kinase, in response to one signal, RA, of the signals being transduced. Because each subunit of the heterodimer is independently modified in this scenario, signals from two pathways converge, leading to the appropriate changes in the pattern of PEPCK gene expression (42,43).

This research was supported by the Korea Science and Engineering Foundation Grant (2000-1-20900-005-1).

We thank Jon Shuman for comments on the manuscript and writing.