Modulation of the Murine Peroxisome Proliferator-Activated Receptor γ 2 Promoter Activity by CCAAT/Enhancer Binding Proteins

PPAR γ and C/EBPs are transcriptional regulators essential for adipocyte differentiation and function. Previous findings indicate that PPAR γ 2 transcription is regulated by members of the C/EBP family. We demonstrate here that C/EBP α and C/EBP δ, but not C/EBP β, induce the activity of the PPAR γ 2 promoter in transiently transfected 3T3-L1 preadipocytes and bind to two juxtaposed-low affinity C/EBP binding sites. Results obtained with chimeras containing interchanged C/EBP α -C/EBP β N-terminal-transactivation domain and C-terminal DNA binding-dimerization domain indicate that the N-terminal part of C/EBP β prevents it from binding to the PPAR γ 2 promoter. Indeed, deletion mutants of C/EBP β lacking the N-terminal part of the molecule are able to bind to the PPAR γ 2 promoter. We further demonstrate that deletion of a region located between amino acids 184-212, upstream of the DNA binding domain, permits C/EBP β binding to the PPAR γ 2 promoter, implicating an inhibitory region in C/EBP β for modulating DNA binding specificity to the PPAR γ 2 promoter. In summary, this study indicates that C/EBP β but not C/EBP α or C/EBP δ is unable to bind to C/EBP binding sites in the mouse PPAR γ 2 promoter. The lack of binding is due to a region N-terminal of C/EBP β DNA binding domain. Our findings illustrate a mechanism by which C/EBP isoforms differentially modulate the transactivation of the PPAR γ 2 promoter. the addition of C/EBP which in a supershift. The results show that unlike C/EBP β (number 2 ∆ 5) and ∆ 6) were able to bind to the PPAR γ 2 promoter. Two specific

and atherosclerosis (1). It results from an excessive accumulation of white adipose tissue, composed of adipocytes, which play a central role in energy storage and release of lipid metabolites. Among the factors orchestrating adipocyte differentiation, the nuclear receptor Peroxisome Proliferator Activated Receptor γ (PPARγ) 1 and the CCAAT/ Enhancer Binding Proteins (C/EBPs) are two key transcription factors (2,3). PPARγ has received considerable attention due to the fact that PPARγ synthetic ligands, such as thiazoladinedione, are potent insulin-sensitizing drugs administrated to type II diabetic patients (2). Two PPARγ isoforms, PPARγ1 and PPARγ2 are expressed in different tissues; the latter is reported to be restricted to the adipose tissue and the mammary gland (4)(5)(6). Compared to PPARγ1, PPARγ2 contains an additional N-terminal region composed of 30 amino acids and distinct promoters regulate the expression of these two isoforms (5).
C/EBPs are expressed in a number of tissues and are involved in the regulation of several biological processes such as acute-phase response, inflammatory and immune response, cell proliferation and differentiation and control of energy metabolism (7)(8)(9). C/EBP family members display highly similar C-terminal basic DNA binding domains and leucine zipper dimerization domains but exhibit different N-terminal regions containing the activation domains. Consequently, the various C/EBP proteins form both homodimers and heterodimers and bind to a common DNA consensus sequence (7). C/EBPα and C/EBPβ expression are regulated at the translational level, via a leaky 4 ribosome scanning mechanism. The C/EBPα mRNA is translated to 42 KDa and 30 KDa proteins both of which are activators that differ in their transcriptional potencies (10).
Translation of the C/EBPβ mRNA generates three different products; two transactivatorproteins of 35 KDa and 32 KDa called LAP1 and LAP2 (Liver-enriched transcriptional Activator Protein 1 and 2) and a dominant negative form of 20 KDa called LIP (Liverenriched transcriptional Inhibitory Protein). The inhibitory activity results from the deletion of the transactivation domain in the 151 amino acids of the N-terminal region which is truncated in LIP. Homodimers or heterodimers containing LIP bind to C/EBP binding sites but are transcriptionally inactive (11). In addition, some C/EBP family members act as inhibitors; C/EBPζ (also called CHOP and GADD153), harbors a dimerization domain but not a functional DNA binding domain. Consequently, homodimers and heterodimers containing C/EBPζ fail to bind C/EBP binding sites (12).
The cell type specificity of C/EBP-regulated gene expression is thought to result from the tissue-restricted and temporal expression of a family member and combinatorial interactions with other transcription factors or coactivators (13)(14)(15)(16)(17)(18). This combination results in transcriptional activation, but also in some cases in inhibition of promoter PPARγ during terminal differentiation, which act cooperatively to complete adipogenesis. (2,3,9). C/EBPα and C/EBPδ transactivate the mouse PPARγ2 promoter, via sites located at position -340 bp and -327 bp relative to the transcriptional start site (24,25) but no data has been shown to support a transactivation effect of C/EBPβ on the mouse PPARγ2 promoter. We investigate here the transcriptional activity of various C/EBP isoforms on the mouse PPARγ2 promoter and demonstrate that in contrast to C/EBPα and C/EBPδ, C/EBPβ is unable to bind to C/EBP binding sequences and to stimulate PPARγ2 promoter activity.

EXPERIMENTAL PROCEDURES
Plasmid constructs. Cloning of the mouse PPARγ2 promoter into the p19 Luciferase vector and different C/EBPs into pEFbos vector have been described previously (24). The human p21 (WAF1/CIP1) gene promoter linked to the luciferase reporter vector was provided by Dr. M. Liu (26). Full-length murine coding sequences from the following vectors were excised with the restriction enzymes indicated: C/EBPα (EcoRI/HindIII fragment from MSV/C/EBP), C/EBPβ (ΕcoRI/BamHI from MSV/C/EBPβ), C/EBPδ (EcoRI/BamHI from MSV/C/EBPδ), vectors were provided by 191, ∆116-191) were originally generated as published, as the second translation form of C/EBP (27) and were provided by Dr. P.F. Johnson. A NcoI/PstI fragment of the coding sequences containing the internal deletion were inserted into pKS C/EBPβ, excised with NcoI/PstI. The numbering system used hereafter refers to the full-length C/EBPβ isoform (∆184-212C/EBPβ , ∆ 137-212C/EBPβ ). A Kozak sequence (referred as KOZ) was generated around the first ATG of the coding sequence by replacing the EcoRI/SphI fragments of pKS C/EBPβ, pKs ∆184-212C/EBPβ and pKs ∆137-212C/EBPβ with a double-stranded oligonucleotide flanked by EcoRI/SphI restriction sites (The sense primer: 5'-AATTCCACCATGGACCGCCTGCTGGCCTGGGACGCAGCATG-3'). The coding sequences were then isolated as EcoRI/XbaI fragments and inserted into pSVSPORT1.
N152C/EBPβ was generated by replacing an Asp718/NcoI fragment from pKS C/EBPβ and replacing it with a double-stranded oligonucleotide flanked by Asp718/NcoI sites (The sense primer: 5'-GTACCGAATTCCAC-3'). The coding sequence isolated as an Asp718/XbaI fragment was inserted into pSVSPORT1. KOZ-N208 and KOZ-N213 C/EBPβ (the latter containing two additional amino acids A and K at the N-terminal) were generated by deletion of an EcoRI/XcmI fragment in pKS C/EBPβ and replacing these fragments with double-stranded oligonucleotides flanked by restriction sites EcoRI/X c m I .
T h e s e n s e o l i g o n u c l e o t i d e s w e r e 5'- well plates) were transiently transfected with 100 ng PPARγ2 or p21 promoter/luciferase reporter constructs, 50 ng of each expression construct or empty vector as a control. The transfection assays, using an adenovirus system, and the luciferase assay were performed as previously described (28).

Differential effect of various C/EBP isoforms on PPARγ2 promoter activity.
The transcriptional activity of C/EBPα, C/EBPβ and C/EBPδ on the mouse PPARγ2 promoter linked to a luciferase reporter gene was determined by transient transfection in 3T3-L1 preadipocytes. Fig. 1 shows that C/EBPα and C/EBPδ transactivate the PPARγ2 promoter by 4 fold and 7 fold respectively. In contrast, C/EBPβ inhibits the basal promoter activity by approximatively 70%. As a control for C/EBPβ inhibitory action, the cells were cotransfected with p21 promoter/luciferase gene, a known target gene promoter for both C/EBPα and C/EBPβ (23,30). Cotransfection of C/EBPα, C/EBPβ or C/EBPδ in 3T3-L1 preadipocytes stimulated the activity of the p21 promoter (Fig. 1).
These experiments demonstrate that C/EBPβ is unable to activate the PPARγ2 promoter.
We further investigated the modulating function of different C/EBPβ isoforms on PPARγ2 promoter activity. The expression of the LAP1 isoform of C/EBPβ was enhanced by the insertion of a Kozak sequence around the first start codon as previously described (11), while LIP was generated by deletion of the coding sequence for the 151 N-terminal amino acids. The different C/EBPβ constructs were cotransfected with the using a C/EBPβ antibody. Fig

Binding analysis of various C/EBPs to C/EBP binding sites on the PPARγ2
promoter. To further examine the mechanism involved in the differential activation of the PPARγ2 promoter by C/EBP, we analyzed C/EBP binding to previously identified binding sites on the PPARγ2 promoter (24). These binding sites are composed of a core of two C/EBP half sites (GCAAT). As a control the sequence on the PPARγ2 promoter was mutated to form two adjacent C/EBP consensus binding sequences (TTGCGCAAT), created by changing the 5' flanking sequences adjacent to the core C/EBP recognition element (31). C/EBP proteins were produced in COS-1 cells by transfection of various C/EBP constructs. The expression of the different C/EBPs in the cell extracts was analyzed by Western blot and detected using C/EBPα, C/EBPβ or C/EBPδ antibodies ( Fig. 3A). C/EBP expression was not detectable in the control extract, consistent with the specific overexpression of different C/EBPs in the transfected cells. Binding analysis by a gel shift assay using C/EBP binding sites on the PPARγ2 promoter as a probe showed that C/EBPα and C/EBPδ, but not C/EBPβ, were able to bind to the C/EBP binding sites.
As a positive control, we demonstrate that all three C/EBPs bind to the C/EBP consensus binding sequence with high efficacy. Addition of different C/EBP antibodies selective for each of the C/EBPs resulted in a supershift (Fig. 3B). These results indicates that C/EBP family members exhibits differential binding ability to the non-consensus C/EBP binding sites on the PPARγ2 promoter and subsequently they demonstrate different transactivation activity.
We indicate that the C/EBP binding sites on the PPARγ2 promoter display a much lower binding affinity, as compared to the consensus sequence.

Effect of the coexpression of different C/EBPs on binding and transactivation of the PPARγ2 promoter
Since C/EBPβ fails to bind to PPARγ2 promoter binding sites and inhibits the basal activity of the PPARγ2 promoter, we assumed that C/EBPβ competed for endogenous C/EBPα or C/EBPδ homo/heterodimers binding by producing inactive C/EBPα/β or C/EBPβ/δ heterodimers. supershifted with C/EBPα antibody, consistent with the fact C/EBPα was expressed in relative excess, forming a homodimer. In contrast, heterodimers containing C/EBPβ did not bind to the non-consensus sequence as illustrated by the fact that C/EBPβ antibody was unable to supershift the binding of C/EBPs on the non-consensus sequence (Fig. 5C).
These results demonstrate that C/EBPβ homodimers and heterodimers do not bind to the C/EBP non-consensus binding sequence present on the PPARγ2 promoter and consequently C/EBPβ heterodimers do not stimulate PPARγ2 promoter activity.

Identification of the region of C/EBPβ that inhibits binding to C/EBP
binding sites on the PPARγ2 promoter. To investigate the structural basis involved in the differential binding of C/EBPβ versus C/EBPα and C/EBPδ, two hybrid molecules were constructed. C/EBPα-β, harbored the N-terminus part of C/EBPα (amino acids (a.a) 1-273) and the C-terminus of C/EBPβ (a.a. 213−296); its counterpart C/EBPβ-α was composed of the N-terminal part of C/EBPβ (a.a. 1-212) and the C-terminus of C/EBPα (a.a.274-359). The N-terminal part of the molecules included the transactivation domain and the C-terminus contained the DNA binding and dimerization domains (Fig.6A). Fig.   6B and Fig. 6C show that C/EBPα-β was able to bind and transactivate the PPARγ2 promoter to a similar extent as C/EBPα. In contrast, C/EBPβ-α failed to bind and activate the PPARγ2 promoter. As a control, C/EBPα-β and C/EBPα or C/EBPβ-α and

DISCUSSION
In this study, the results obtained using the luciferase reporter assay indicate that the mouse PPARγ2 promoter is differentially regulated by different C/EBPs. C/EBPα and C/EBPδ induce but C/EBPβ inhibits the activity of the promoter. We show that unlike C/EBPα and C/EBPδ, which display low binding affinity, C/EBPβ does not bind to two juxtaposed non-consensus sequences in the PPARγ2 promoter. The negative effect of C/EBPβ on PPARγ2 promoter activity is possibly due to the formation of heterodimers containing exogenous expressed C/EBPβ and other endogenous C/EBP family members which are unable to bind C/EBP binding sites in the PPARγ2 promoter. Similarly, it has been reported previously that C/EBPδ but not C/EBPβ transactivates the IGF-I promoter via a core C/EBP half-site, GCAAT (32). A number of other studies show differential action of C/EBPα and C/EBPβ in activating various promoters (13,18,33,34). Our observations establish a link between the DNA sequence of C/EBP binding sites and the differential binding of various C/EBP members.
To understand the structural differences between C/EBPα and C/EBPβ that mediate the differential biological effect on the PPARγ2 promoter, we created two hybrid molecules. The activity of these molecules generated by the exchange of the regulatory domains of C/EBPα and C/EBPβ demonstrate that the N-terminal domain but not the DNA binding or leucine zipper domains of C/EBPβ is responsible for the differential Previously, it was reported that the binding activity of C/EBPβ is modulated by its interaction with other transcription factors and in response to cytokines, which enhanced C/EBPβ binding to cognate DNA sequences (13,18,35). This region of C/EBPβ contains a repressor domain called RD2 (repressor domain2) which was characterized for its ability to inhibit partially C/EBPβ binding to the albumin promoter in a cell specific natural promoter but are fully active on an artificial promoter bearing a high affinity C/EBP binding site (36). Current models for C/EBPβ transactivation involved a C/EBPβ inactive state that is switched to be active, leading to the C/EBPβ-induced gene transcription (22,37). The N-terminal transactivation domain interacts with C-terminal part of the molecule and prevents its interaction with basic transcription machinery (22,27). Therefore a range of signaling pathways, effector molecules and protein-protein corresponding to residues in HFH-1 (HNF3/Forkhead Homologues) alters HNF-3β binding (38). Similarly, two other transcription factors Ets1 and Ets2 contain inhibitory regions located adjacent to their DNA binding domains that affect DNA binding activity (39). Our results clearly demonstrate that the N-terminal part of C/EBPβ modulates its binding to and function on the mouse PPARγ2 promoter. C/EBPβ is a highly regulated transcription factor and our observation may be relevant as part of a mechanism by which C/EBPβ regulates PPARγ2 expression. In this context, the human PPARγ2 promoter is activated by C/EBPβ via a binding site located at -56 base pairs from the start codon (40). This site is conserved in mouse PPARγ2 promoter (located at -120 base pairs) but the two other binding sites characterized in this study are different in the human and mouse promoters (5). These differences in C/EBP binding sites may reflect differences in the expression of the human PPARγ2, which seems to be expressed at a lower level than its murine homologue (4,5).
It appears that the role of C/EBPs in the transcriptional regulation of PPARγ promoters is certainly complex and can not be fully understood from results obtained in a unique experimental system. Based on experiments utilizing ectopic expression of C/EBPs, C/EBPα or C/EBPβ but not C/EBPδ induce PPARγ gene expression during the conversion of multipotential mesenchymal stem cells to adipocytes (41)(42)(43). Consistent with these results, the level of expression of PPARγ during primary embryonic fibroblast differentiation is drastically reduced in cells derived from C/EBPβ knocked-out mice compared to cells from wild type animals (44). In contrast, targeted deletion of C/EBPβ in mice does not alter PPARγ expression in adipose tissue but impairs adipogenesis (44).
Studies on the PPARγ promoters further reveal the complexity of the regulation of PPARγ expression by C/EBPs. C/EBPs are able to activate the human PPARγ2 promoter but not the human PPARγ1 promoter (40). Glucocorticoid-induced adipocyte differentiation from bone marrow stromal cells mediated C/EBPδ gene transcription within hours whereas PPARγ2 gene transcription is activated within days (25). Our study and others (24,25) show that C/EBPδ is a potent transactivator of the mouse PPARγ2 promoter but the ectopic expression of this transcription factor does not induce PPARγ expression (43). The inability of C/EBPβ to bind and to stimulate the PPARγ2 promoter activity in our experimental system implies structural changes, mediated by the flanking region of C/EBPβ DNA binding domain. Also, tissues expressing high level of C/EBPα and C/EBPβ, such as liver or lung, do not contain detectable amount of PPARγ2. It is possible that multiple C/EBP binding sites on the PPARγ2 promoter mediate C/EBP response. Mutation of these two C/EBP binding sites at -340 bp and -327 bp relative to the transcriptional start site reduced C/EBPα and C/EBPδ activation of the PPARγ2 promoter by approximately 50% (24), indicating these sites contribute to PPARγ2 promoter activity. In addition, deletion of the promoter at -320 bp which did not include the tandem repeat of the C/EBP binding sites resulted in partial loss of C/EBPδ inducible activity, suggesting other C/EBP binding sites might be involved in PPARγ2 promoter activation (25). These results together support a C/EBPβ mechanism of action involving context-specific effects due to promoter composition or/and signal-dependent regulatory pathways.
The differential expression of various C/EBPs may play a central role in the transcriptional regulation of a number of adipocytic genes including PPARγ2. During the process of adipogenesis in 3T3-L1 preadipocytes, the expression of C/EBPβ and C/EBPδ is elevated during the early phase. While C/EBPδ expression declines abruptly, the level of C/EBPβ decreases at a slower rate to a basal level and, in parallel, the expression of C/EBPα is induced (9). Therefore, the expression of C/EBPβ is accompanied by coexpression of C/EBPα and C/EBPδ. High level of expression of C/EBPs is likely necessary for the subsequent activation of the PPARγ2 promoter, considering the promoter's low affinity C/EBP binding sites. The proximal promoter of the C/EBPα gene contains a C/EBP regulatory element but it appears that a delay in transcription activation of C/EBPα by C/EBPβ and C/EBPδ occurs. This phenomenon is likely due to a delay in the acquisition of binding activity by these transcription factors as suggested by the phosphorylation which C/EBPβ undergoes concomitantly with the acquisition of DNA binding activity (37). Similarly, PPARγ2 transcription activation occurs with a delay of 18 hours to 3 days as compared to the C/EBPβ and C/EBPδ expression (25,37,40). To analyze whether the differential activity of various C/EBPs controls PPARγ2 transcription, further dissection of the molecular pathways involving C/EBPs in PPARγ2 expression is required. It may provide an insight into the finely tuned regulatory mechanisms necessary for adipocyte function.