Importance of cAMP-response Element-binding Protein in Regulation of Expression of the Murine Cyclic Nucleotide Phosphodiesterase 3B (Pde3b) Gene in Differentiating 3T3-L1 Preadipocytes*

Incubation of 3T3-L1 preadipocytes with isobutylmethylxanthine (IBMX), dexamethasone, and insulin, alone or in combination, demonstrated that IBMX, which increased cAMP-response element-binding protein (CREB) phosphorylation, was the predominant regulator of Pde3b expression. Real time PCR and immunoblotting indicated that in 3T3-L1 preadipocytes, IBMX-stimulated induction of Pde3b mRNA and protein was markedly inhibited by dominant-negative CREB proteins. By transfecting preadipocytes, differentiating preadipocytes, and HEK293A cells with luciferase reporter vectors containing different fragments of the 5′-flanking region of the Pde3b gene, we identified a distal promoter that contained canonical cis-acting cAMP-response elements (CRE) and a proximal, GC-rich promoter region, which contained atypical CRE. Mutation of the CRE sequences dramatically reduced distal promoter activity; H89 inhibited IBMX-stimulated CREB phosphorylation and proximal and distal promoter activities. Distal promoter activity was stimulated by IBMX and phorbol ester (PMA) in Raw264.7 monocytes, but only by IBMX in 3T3-L1 preadipocytes. Chromatin immunoprecipitation analyses with specific antibodies against CREB, phospho-CREB, and CBP/p300 (CREB-binding protein) showed that these proteins associated with both distal and proximal promoters and that interaction of phospho-CREB, the active form of CREB, with both Pde3b promoter regions was increased in IBMX-treated preadipocytes. These results indicate that CRE in distal and proximal promoter regions and activation of CREB proteins play a crucial role in transcriptional regulation of Pde3b expression during preadipocyte differentiation.

activation of membrane-associated PDE3B, most likely via phosphatidylinositol 3-kinase and protein kinase B, is an important component of the antilipolytic action of insulin (2,19). Previous studies from our laboratories indicated that RO 20-1724 (a specific inhibitor of PDE4 isozymes), but not cilostamide (a specific PDE3 inhibitor), could replace IBMX (a nonselective PDE inhibitor) during differentiation of 3T3-L1 adipocytes induced by IBMX, dexamethasone, and insulin (20). We and others (21,22) have also reported that specific inhibitors of PDE3, not PDE4, block the antilipolytic action of insulin in differentiated or mature adipocytes. Taken together, these observations suggest that PDE4 isoforms may regulate cAMP pools involved in initiation of differentiation, whereas PDE3 isoforms may regulate cAMP pools that control lipolysis in differentiated adipocytes. Experiments with PDE3 inhibitors have also indicated an important role for PDE3 in regulation of insulin secretion in pancreatic islets (9 -11, 23, 24) and, perhaps, oxygen consumption in human subjects (25).
Because the detailed molecular mechanisms that govern adipocyte Pde3b expression during adipogenesis are not known, and because terminal differentiation of 3T3-L1 adipocytes provides a well characterized model for the study of gene expression and regulation (26,27), we initiated studies of mechanisms underlying transcriptional control of Pde3b expression in these cells. We isolated the 5Ј-flanking region of the murine Pde3b gene, which included the putative first exon and an ϳ5-kb genomic fragment upstream of the translation start site, and which contained distal and proximal promoter regions and transcription initiation sites in the proximal promoter region (ϳ0.5 kb upstream of the translation start site), close to those described by Niiya et al. (28). In addition, our results suggest that cis-acting CRE (cAMP-response) elements in both promoter regions and activation of CREB (CRE-binding proteins) are important in the regulation of induction of Pde3b by cAMP during differentiation of 3T3-L1 adipocytes.

Materials
Tissue culture reagents (Dulbecco's modified Eagle's medium, fetal bovine serum, and penicillin/streptomycin), GeneRacer TM kit, and TOPO TA Cloning kit were from Invitrogen; [ 3 H]cAMP was from PerkinElmer Life Sciences; rolipram, cilostamide, bisindolylmaleimide I, and anti-CREB antibodies were from Calbiochem; isobutylmethylxanthine, dexamethasone, and insulin were from Sigma; Affi-Gel 601 was from Bio-Rad; the BCA protein assay kit and bovine serum albumin were from Pierce. Enzymes for DNA manipulation were purchased from Roche Applied Science; [␣-32 P]dATP or [␥-32 P]ATP was from Amersham Biosciences. Vectors (basic-pGL3, SV40 promoter-pGL3, and SV40 promoter/enhancer-pGL3 which contained the firefly luciferase gene, and TK-pRL which contained the thymidine kinase (TK) promoter and the Renilla luciferase gene), TRANSFAST TM reagent, the Dual Luciferase TM reporter assay system, and gel shift binding buffer were obtained from Promega (Madison, WI). SMART TM RACE cDNA amplification kits, AdvanTAge TM PCR cloning kits, Advantage-GC cDNA PCR kits, and Adeno-X kits were purchased from Clontech. Poly(A) ϩ Pure TM mRNA isolation kits were from Ambion (Austin, TX); plasmid DNA purification kits and QuantiTECT CYBR GREEN kits were from Qiagen (Studio City, CA); QuickChange multisite-directed mutagenesis kits were from Stratagene (Cedar Creek, TX); BigDye Terminator version 3.1 cycle sequencing kit was from Applied Biosystems (Foster City, CA); mouse 3T3-L1 preadipocytes, HEK293A cells, and RAW264.7 cells were from American Type Culture Collection (Manassas, VA).

cAMP Phosphodiesterase Assay
Portions of the supernatant fractions were assayed (30 min, 30°C) in 100 l containing 50 mM Tris-HCl (pH 7.4), 5 mM MgCl 2 , 0.1 mM EGTA, and 1 M [ 3 H]cAMP (50,000 dpm). Assays were terminated with 50 l of 0.5 M EDTA (pH 7.4). PDE3 activity was determined as that portion of total cAMP hydrolysis inhibited by cilostamide (a PDE3-selective inhibitor, 1 M). After diluting samples with 0.3 ml of buffer (0.1 M NaCl, 0.1 M HEPES (pH 8.5)), the reaction product, 5Ј-[ 3 H]AMP, was purified by polyacrylamide-boronate gel chromatography (Affi-Gel 601, 1-ml bed volume) and quantified by liquid scintillation counting. room temperature), and filtered (0.2-m filter) prior to use (29). Cells were fixed with 10 ml of Accustain formalin solution (Sigma) in PBS (1 h, 4°C) and stained with 10 ml of Oil Red O solution per dish (15 min, room temperature). Accumulation of neutral lipid (red color) was assessed using an Olympus IX51 microscope with a digital camera.

Adenovirus-mediated CREB Protein Expression in 3T3-L1 Preadipocytes
Recombinant adeno-CREB expression vectors, kindly supplied by J. Reusch and D. Klemm (Veterans Affairs Medical Center, Denver, CO), were constructed with replication-deficient adenovirus under control of the cytomegalovirus immediate early promoter that regulates the ectopic expression of cDNA encoding ␤-galactosidase, inactive MCREB or KCREB, or constitutively active VP16-CREB or DCREB. Adenovirus constructs were amplified in HEK293A cells and purified with Adeno-X kits (Clontech). Adenovirus particles were quantitated by measuring absorbance at A 260 nm , and infection efficiency was determined by immunostaining with anti-adenovirus hexon protein or anti-␤-galactosidase antibody. At ϳ80% confluence, preadipocytes were infected for 3 h with adenovirus constructs (500 virus particles per cell) in DMEM containing 1 g/ml polylysine. Cells were cultured in DMEM containing 2.5% FBS for 2 days, and after that, infected cells were incubated with or without 300 M IBMX for additional 5 days in DMEM containing 10% FBS. Expression of Pde3b mRNA and protein was assessed by real time RT-PCR, immunoblotting, and PDE3 activity assays.

RNA Ligase-mediated Rapid Amplification of 5 cDNA Ends (RLM-5-RACE) for Identification of Distal Transcripts of the Pde3b Gene
Poly(A)-mRNA was isolated from IBMX-treated 3T3-L1 preadipocytes using the Qiagen RNeasy mini kit and Oligotex mRNA mini kit. After first strand synthesis, cDNA was amplified by PCR using a GeneRacer 5Ј-primer and a gene-specific primer (5Ј-CAGCGGGCTCACGCAGCTCTTCACGTA-3Ј, reverse-complement 100 -126 bp downstream of ATG start codon). Nested PCR was performed on purified PCR products (Qiagen PCR purification kit), using a GeneRacer 5Ј-nested primer and a nested, gene-specific primer (5Ј-AGCTCTTCA-CGTAGCCGTTGCGCA-3Ј, reverse-complement 89 -112 bp downstream of ATG start codon). After TOPO TA cloning of PCR products, positive clones were identified by PCR and sequenced.

Transient Transfection with Transfast TM Reagent (Promega) and the Dual Luciferase Assay
Cells were cotransfected with pRL-TK and pGL3 reporter plasmid vectors that expressed Renilla and firefly luciferase activities, respectively. For the different pGL3 reporter plasmids, firefly luciferase expression was driven either by Pde3b 5Ј-flanking region genomic fragments or constructs (Pde3b-pGL3), by the SV40 promoter (SV40 (pro)-pGL3) or the SV40 enhancer and promoter (SV40 (con)-pGL3) as positive controls, or by promoter-free pGL3 (Basic-pGL3) as a negative control.
Transfection was performed using TransFAST TM reagent according to the manufacturer's instructions (Promega). On the day before transfection, 400 l of nuclease-free water was added to the TransFAST TM reagent, and the lipid film was dispersed by vigorous vortexing for 10 s. After 24 h, a mixture of plasmid DNA, TransFAST TM reagent (3 l), and serum-free growth medium (1 ml) was prepared, incubated (10 -15 min, room temperature), and used to transfect cells, after removal of growth medium from the cells. Each DNA sample contained the pRL-TK plasmid vector (0.02 g) and a pGL3 plasmid vector (1.0 g). 3T3-L1 preadipocytes were cotransfected in 6-well culture dishes at 60 -80% confluence, incubated for 1 h at 37°C, and then overlaid with prewarmed DMEM containing 10% FBS (2 ml) and maintained for 48 h at 37°C. For transfection of differentiating 3T3-L1 preadipocytes, preadipocytes were grown in DMEM containing 10% FBS to about 80% confluence (day 0), when differentiation was induced by addition of fresh DMEM containing 10% FBS and MDI (where MDI is IBMX, Dex, and insulin in differentiating preadipocytes). Three days later (day 3), differentiating 3T3-L1 preadipocytes were cotransfected by the procedure described above but were then overlaid with 2 ml of prewarmed DMEM containing 10% FBS and 1 g/ml insulin and maintained for 48 h at 37°C. After 48 h at 37°C, the cotransfected cells were then washed three times with PBS, disrupted in 250 l of Passive Lysis Buffer (PLB, Promega), and centrifuged (1,000 ϫ g, 5 min). Supernatant firefly and Renilla luciferase activities were measured with the dual luciferase reporter assay system (Promega) using a luminometer (Lumat LB 9501). Luciferase reagent substrate (100 l) was mixed with cell lysates (20 l), and firefly luciferase activity was measured (10 s, twice). To terminate this reaction and initiate the Renilla luciferase reaction, Stop & GloR Solution (100 l) was added to the mixture for 20 s, and Renilla luciferase activity was measured (10 s, twice). The firefly luciferase activity in lysates from cells transfected with Pde3b-pGL3 reporter constructs, and positive and negative controls, was normalized to Renilla luciferase activity values, and the ratio between firefly and Renilla luciferase activities was referred to as relative luciferase activity (RLA). RLA (mean Ϯ S.D., n ϭ 3) was assayed (in duplicate) in lysates from triplicate transfections as described above. Unless otherwise noted, experiments presented in the various figures are from single experiments, which were replicated once or twice.
For comparison of transcriptional activities in different cell types, HEK293A, 3T3-L1, and RAW264.7 cells were transfected with pGL3 plasmid DNAs containing various lengths of 5Ј-flanking regions of the Pde3b gene. After transfection, the cells were treated with IBMX (300 M) for 3 days or PMA (50 nM) for 24 h and assayed for luciferase activities.

Construction of Pde3b-luciferase Vectors and Reporter Plasmids
In general, Pde3b reporter plasmids were generated by insertion of Pde3b genomic fragments and constructs into pGL3-Basic vectors.
Distal-Proximal Region Fusion Constructs-The ϳ1.1-kb SalI-EcoRI portion (SE) from the SalI/XbaI distal genomic fragment (Figs. 4 and 5A), containing AP-2, Sp-1, and CRE cisacting elements, exhibited strong promoter activity when expressed as the SE-pGL3 reporter plasmid (Fig. 5A). SE was subcloned into SalI/EcoRI sites of pBluescript in between KpnI/ SmaI sites, and then excised with KpnI/SmaI. Pde3b pGL3-Basic plasmids containing proximal promoter regions ( Fig. 4) were digested with KpnI and MluI (restriction sites in the multicloning region of pGL3); the 5Ј-overhang resulting from MluI digestion was blunt-ended. The KpnI/SmaI fragment that contained the distal SE promoter region was ligated into the plasmids previously digested with KpnI/MluI (blunted) (Fig. 7). The resulting distal-proximal region fusion constructs contained a 13-bp (CGCGTGCTAGCCC) linker between the SE fragment of the distal promoter and the proximal promoter constructs (Fig. 7). One construct, SEEB, from which this linker was removed, was generated by digesting SE-Bfr-pGL3 with EcoRI to remove the Bfr-EcoRI portion of Bfr-pGL3, as well as the linker between distal and proximal promoter regions of the Pde3b gene, and by self-ligating at EcoRI sites (Fig. 7).

Chromatin Immunoprecipitation Analysis
ChIP analyses were performed using an assay kit (Upstate Biotechnology, Inc., Lake Placid, NY) according to the manufacturer's protocol. Confluent 3T3-L1 cells were cultured in DMEM containing 1% FBS and treated with or without 300 M IBMX (16 h), after which cells were cross-linked with 1% formaldehyde (30 min, room temperature), washed twice with ice-cold PBS containing protease inhibitors, and harvested. Cell pellets were resuspended in SDS lysis buffer (10 min, 4°C) and sonicated (Branson sonifier 450, power setting 5, duty cycle 50% for six times, each time for 25 s followed by 1 min on ice). After centrifugation, the supernatant was diluted in ChIP dilution buffer and incubated overnight at 4°C with rabbit IgG, anti-CREB, anti-phospho-CREB, anti-ATF1, anti-CREM1, or anti-CBP and anti-p300 antibodies (Santa Cruz Biotechnology (Santa Cruz, CA)). Immune complexes were incubated (2 h, 4°C, with rotation) with 60 l of a 50% slurry of protein A-agarose beads containing salmon sperm DNA. Agarose beads were pelleted by gentle centrifugation (720 ϫ g, 4°C), sequentially washed with low and high salt buffer, LiCl buffer, and twice with TE buffer. After washing, the immune complexes were eluted by incubation (15 min, 25°C) with 400 l of 1% SDS, 0.1 M NaHCO 3 . To reverse the cross-linking of DNA, 8 l of 5 M NaCl was added, and the mixture was incubated (4 h, 65°C). After treatment with proteinase K (1 h, 45°C), DNA was recovered by phenol/chloroform extraction and ethanol precipitation and resuspended in 50 l of TE buffer. Touch-down PCR (32 cycles, 3 l of DNA) amplified the segment (Ϫ4326/Ϫ4092) flanking the consensus CRE site in distal promoter region with forward primer 5Ј-ATGGGTGGCCTATTTGCTCTTA-3Ј and reverse primer 5Ј-GGAGAATTCTCCTTCGCCTCG-3Ј, and the segment (Ϫ698/Ϫ293) in the proximal promoter region containing putative CREB-binding sites with forward primer 5Ј-TGTGACT-TACCTCAAAGGGGACT-3Ј and reverse primer 5Ј-CAGTTA-CCTCACGGCCGCAGCCT-3Ј. PCR products were separated on 2% agarose gels.

Statistical Analysis
Data are presented as means Ϯ S.D. of at least three independent experiments. Within each experiment, values were means from three individual determinations for each experimental condition. Statistical analysis of ANOVA and Bonferroni-Dunn post hoc tests was performed to determine statistical differences among treatments with p Ͻ 0.05 considered significant. Fig.  1A (lane 2), immunoreactive PDE3B protein was significantly increased during incubation of preadipocytes with IBMX alone (p Ͻ 0.01 compared with control). Dex and/or insulin, alone (Fig. 1A, lanes 3 and 4) or in combination (lane 7), only slightly increased Pde3b expression but significantly potentiated the effect of IBMX (lane 8) (p Ͻ 0.01). As shown in Fig. 1B, cilostamide-inhibited PDE activity, which in adipocytes represents mainly PDE3B activity (2)(3)(4)6), was also increased in cells treated with IBMX, alone (lane 2) or in combination with Dex and/or insulin (lanes 5, 6, and 8). These results suggested that the PDE inhibitor IBMX, which increased (most likely via cAMP and PKA) phosphorylation of CREB in 3T3-L1 preadipocytes (cf. Fig. 8A), played a major role in the induction of Pde3b during differentiation and that insulin and dexamethasone (via unidentified signaling pathways) potentiated, or added to, the effects of cAMP. To assess effects of differentiation agents on neutral lipid accumulation, preadipocytes were incubated for 12 days in the presence of IBMX, Dex, and insulin, alone or in combination, and then stained with Oil Red O. As shown in Inhibitory Effects of Dominant-negative CREB on Induction of Pde3b by IBMX-CREB is apparently a crucial transcriptional activator of mitotic clonal expansion and the differentiation program in adipocytes, because of its role in regulating transcription of C/EBP␤ (14,15,30,31). To examine further the role of CREB on Pde3b expression, without triggering differentiation, 3T3-L1 preadipocytes were infected with adenovirus constructs expressing ␤-galactosidase (adeno-␤-galactosidase), dominant-negative forms of CREB (MCREB and KCREB), or active forms of CREB (V16-CREB, D-CREB) and treated with 300 M IBMX alone for 5 days. As seen in Fig. 2, IBMX increased expression of PDE3B immunoreactive protein ( Characterization of the 5Ј-Flanking Region of the Murine Pde3b Gene-A genomic SalI restriction fragment (ϳ14 kb), which contained 5Ј-flanking region (ϳ5.1 kb upstream from the ATG initiation codon) and putative exon 1 of the Pde3b gene (Fig. 3A), was cloned from murine 129/SvJ and Balb/c  The ϳ5.1-kb flanking region was sequenced (GenBank TM accession numbers AF547434 and AY159890). To determine putative transcription initiation sites, 5Ј-RACE analysis and RT-PCR were performed using mRNA isolated from mouse 3T3-L1 preadipocytes and differentiated adipocytes (data not shown (32)). As depicted in Fig. 3A and supplemental Fig. 1A, five different transcription initiation sites in the 5Ј-flanking region of the murine Pde3b gene were identified and located Ϫ346, Ϫ321, Ϫ293, Ϫ290, and Ϫ227 bp from the translation start site (ATG, ϩ1), respectively. Results from ribonuclease protection assays (RPA) were consistent with those from 5Ј-RACE (RT)-PCR. Several transcripts of ϳ320, ϳ300, ϳ280, and ϳ240 bp were markedly increased in differentiated adipocytes and protected by an antisense RNA probe corresponding to ϳ360 bp upstream of the translation start codon, consistent with the marked increase in Pde3b expression during differentiation (data not shown). By using primer extension assays, included SalI-XbaI (ϳ1. 7 kb) and XbaI-XbaI (ϳ6 kb) fragments, the latter of which contains the coding region of putative exon 1 (filled black bar). The two genomic fragments were subcloned into pBluescript and used to construct Pde3b-luciferase reporter vectors (Fig. 4). The 5Ј-flanking region of the Pde3b gene extends ϳ5.1 kb upstream from the ATG translation initiation codon. Putative proximal and distal promoter regions are identified with arrows and depicted separately, with several putative transcription factor-binding sites. The ATG start codon is boldface; putative transcription initiation sites are marked. B and C, promoter activities of luciferase reporter gene constructs containing various fragments of the 5Ј-flanking region. Luciferase reporter expression vectors were constructed with the whole or indicated portions of the ϳ5.1 kb SalI/BamHI fragment (Fig. 4). RLA (mean Ϯ S.D., n ϭ at least three separate experiments) was measured (duplicate assays) in lysates from triplicate transfections of 3T3-L1 adipocytes and differentiating preadipocytes (B) and of HEK293A cells incubated with or without IBMX (C), as described under "Experimental Procedures." RLA represents the fold increase in firefly luciferase activity normalized to Renilla luciferase activity. *, p Ͻ 0.05; **, p Ͻ 0.01, statistically different from control. Basic-pGL3 represents the promoter-and enhancer-free pGL3 vector.

Effects of Differentiation Agents on Pde3b Expression and Lipid Accumulation in 3T3-L1 Preadipocytes-As seen in
Niiya et al. (28) reported a transcription initiation site (TIS) ϳ195 bp upstream of the translation start site of Pde3b. In mouse EST databases, a unique EST was found in RIKEN (4931402H15) with a TIS at Ϫ322, and which spans the first ATG codon site in known putative exon 1 of Pde3b. In addition, the lengths of 5Ј-untranslated regions of human, mouse, rat, and chicken phosphodiesterase 3B mRNAs are reported in GenBank TM as 291 bp (AY459346), 270 bp (AF547435), 64 bp (Z22867), and 153 bp (AJ851613), respectively.
Promoter Activities in the 5Ј-Flanking Region of the Pde3b Gene-A series of reporter vectors, containing different portions of the 5Ј-flanking region of the Pde3b gene fused to the firefly luciferase gene (Fig. 4), were constructed and transfected into 3T3-L1 preadipocytes and differentiating preadipocytes (Fig. 3B). Direct effects of IBMX alone were examined in transfected, nondifferentiating HEK293A cells (Fig.  3C). As seen in Fig. 3, B and C, two regions in the 5Ј-flanking region of the Pde3b gene apparently were responsible for promoter activities. One region (the "distal" promoter region), which contained a canonical CRE, cis-acting sequence (supplemental Fig. S1B), was apparently located ϳ4 kb upstream of the TIS, in the distal SalI-XbaI (Ϫ5.1 to Ϫ3.4 kb) fragment (Fig. 3, B and C). Another, the proximal promoter region, which contained several atypical CRE cisacting elements (supplemental Fig. S1A), was located close to the identified putative TIS, Ͻ1 kb upstream from them. Promoter activity was enhanced during differentiation of 3T3-L1 preadipocytes (Fig. 3B) or in transfected HEK293A cells incubated with IBMX (Fig. 3C). The two regions seemed to be separated by a gene segment with reduced promoter activity or a negative regulatory region. Removal of the distal SalI-XbaI fragment (ϳ1.7 kb) from SXB-pGL3 produced a fragment, Xba-pGL3, with very low luciferase activity, and subsequent deletion of the downstream XbaI-ApaI fragment (ϳ2.6 kb) was required to allow expression of proximal promoter activity in the ApaI-PvuI region, i.e. with Apa-pGL3, Psh-pGL3, Not-pGL3, and Bgl-pGL3 reporter vectors (Fig. 3, B and C). Compared with Basic-pGL3, luciferase activity of all expression vectors was much greater in HEK293A cells (Fig. 3C) than 3T3-L1 preadipocytes (Fig. 3B), but the relative pattern and differences in the activities of the individual

Regulation of Murine Pde3b Gene Expression by CREB
reporter vectors were similar in the two cell lines (Fig. 3, B  and C).
The proposed locations of the distal and proximal promoter regions (upstream of the ATG translation start codon) are presented in Fig. 3A; their nucleotide sequences, which are homologous (ϳ99% identity) to the corresponding mouse genomic sequences deposited in GenBank TM , are presented in Fig. S1, A and B. Alignment of GenBank TM sequences indicated limited homology among the 5Ј-flanking regions of the mouse, rat, chicken, and human PDE3B gene; only the mouse contained the canonical CRE sequences in the distal promoter region. Atypical CREB-binding sites, however, were predicted in putative distal and proximal promoter regions in the 5Ј-flanking regions of the four species. Consistent with our experimental results, two promoter regions (beginning at position Ϫ4061 and another at Ϫ364) were predicted when 5105 bp of the 5Ј-flanking sequence of the murine Pde3b gene was queried online. On the other hand, analysis at another on-line site predicted one promoter (at Ϫ532).
Distal Promoter Activity-To characterize the distal promoter region, 3T3-L1 preadipocytes or HEK293A cells were transfected with Pde3b-pGL3 constructs generated from the ϳ1.7 kb SalI/XbaI fragment in the upstream or distal promoter region (Ϫ5.1 kb to Ϫ3.4 kb) of the 5Ј-flanking region of the Pde3b gene (Figs. 3A, 4, and 5, A and B). As depicted in Fig. 5, A and B, SX-pGL3, SB-pGL3, and SE-pGL3, but not SH-pGL3, exhibited strong promoter activity, indicating that active ciselements were located between HindIII and BamHI sites in the SalI/XbaI fragment. The promoter activity of this region, which contains a canonical CRE site (TGACGTCA) and Sp-1 and AP-2 cis-acting elements, as well as a putative TATA box (Fig.  3A, Fig. 5A, and Fig. S1B), was much higher than that of the SV40 promoter-pGL3 construct (Fig. 5, A and B) and was increased in differentiating preadipocytes (Fig. 5A), as well as in  3T3-L1 preadipocytes and HEK293A cells (Fig. 5B) incubated with IBMX. As also seen in Fig. 5A, in preadipocytes, the activities of the distal promoter fragments (SX-pGL3, SB-pGL3, and SE-pGL3) were orders of magnitude greater than Basic-pGL3. This suggested that activities of these distal promoter fragments were also much greater than the promoter activities (Fig.  3B) of the entire full-length 5Ј-flanking promoter region (SXB-pGL3) or proximal promoter fragments (Apa-, PshA-, Not-, and Bgl-pGL3), because, in preadipocytes, the luciferase activities of the latter constructs were only severalfold greater than Basic-pGL3 (note the differences in scale in Figs. 3B and 5A). Distal promoter activity was orientation-dependent because the SX*-pGL3 construct, with the same SalI-XbaI fragment as in SX-pGL3, but in reverse orientation, did not exhibit promoter activity (Fig. 5, A and B).
As seen in Fig. 5B, mutations of the CRE sites in SX-pGL3 caused a drastic reduction in basal as well as IBMX-stimulated promoter activities of the SX mutCRE-pGL3 reporter vector, demonstrating that the cis-acting CRE are critical for distal promoter function. Mutation of the putative TATA site exerted a much smaller effect on promoter activity (Fig. 5B). A series of luciferase-expression vectors containing the AP-2, Sp-1, and CRE sequences (alone and in combinations) were also gener-ated by PCR. As seen in Fig. 5C, in transfected HEK293A cells incubated with IBMX, the relative luciferase activity of AT1-pGL3 reporter constructs containing CRE alone (AT1-pGL3), or in combination with Sp-1 and AP-2 elements (AT3-, 6-, 7-pGL3), was much higher than that of the SV40 promoter-pGL3. AT2-, -4-, and -5-pGL3 vectors, luciferase constructs containing AP-2 and Sp-1 binding sequence(s) (alone or in combination), demonstrated little or no promoter activity (Fig.  5C). Promoter activity related to CRE was reduced by the presence of AP-2 and/or Sp-1 elements (AT3-, -6-, and -7-pGL3 constructs) (Fig. 5C). In addition, mAT1-pGL3, an expression luciferase vector containing mutated CRE sequences, exhibited no promoter activity. Similar results were demonstrated with these expression vectors in differentiating preadipocytes (data not shown).
Effects of PMA and IBMX on Promoter Activities and CREB Protein Phosphorylation in Raw264.7 Monocytes and 3T3-L1 Preadipocytes-As seen in Fig. 6, promoter activity of several distal region luciferase reporter vectors was stimulated by IBMX, but not phorbol ester (PMA) (50 nM), in transfected 3T3-L1 preadipocytes (Fig. 6A), whereas distal promoter activity was, under these conditions, stimulated by both IBMX and PMA in Raw264.7 monocytes (Fig. 6B). Mutation of CRE in the distal promoter SX mutCRE-pGL3 reporter vector significantly reduced the effect of PMA on luciferase reporter gene activities (Fig. 6C), suggesting that the CRE are important for the effects of PMA on Pde3b gene expression (Fig. 6B). In HEK293A cells transfected with SX-pGL3 (Fig. 6C) or with the NotI-pGL3 reporter vector (Fig. 6D) from the proximal promoter region (which contained an atypical CRE site), the effects of PMA on luciferase reporter gene activities were inhibited by bisindolylmaleimide I (a specific PKC inhibitor) and H89. The effects of IBMX were also blocked by H89 (Fig. 6D). PMA-induced CREB phosphorylation in monocytes (Fig. 6E) was more effectively inhibited by BisI than H89, suggesting that effects of PMA on CREB phosphorylation were mainly mediated by activation of PKC. (R p )-cAMPS and H89 blocked the effects of IBMX. In HL-60 cells, PMA-induced expression of ganglioside GM3 synthase is thought to be mediated via PKC/mitogen-activated protein kinase-dependent CREB phosphorylation (33).
Promoter Activity of Distal/Proximal Fusion Constructs-The results in Fig. 3, B and C, indicated that the negative regulatory region affected proximal promoter activities. To observe the effects of this gene segment (i.e. the XbaI-PshAI portion of the 5Ј-flanking region in Fig. 3, B and C) on distal promoter activity (cf. Fig. 5A), 3T3-L1 preadipocytes were transfected with reporter vectors in which the active SE fragment from the distal promoter region (ϳ1.7-kb SalI/XbaI fragment) was coupled with various downstream portions of the 5Ј-flanking region (XbaI-BamHI (Ϫ3.4 kb to Ϫ122 bp)) of the Pde3b gene (Fig. 7). As seen in Fig. 7, distal promoter activity of the SE Luciferase reporter expression vectors were constructed with indicated portions of the distal ϳ1.7-kb SalI/XbaI fragment, which contained the indicated restriction sites and AP-2, SP-1, and CRE cis-acting elements. RLA (mean Ϯ S.D., n ϭ 3 experiments) was measured (in duplicate) in lysates from triplicate transfections of 3T3-L1 preadipocytes and differentiating preadipocytes (A) and preadipocytes and HEK293A cells incubated without or with IBMX (B). SX*-pGL3 is orientation-reversed SX-pGL3 construct; mutCRE, mutTATA, and mutCRE/TATA are SX-pGL3 reporter vectors with mutations in CRE, TATA, and CRE and TATA sequences, respectively. C, HEK293A cells were transfected with luciferase reporter expression vectors that contained CRE, Sp-1, and AP-2 elements, alone and in combination, and that were generated by PCR from the ϳ1.7-kb SalI/XbaI distal fragment, using primers listed in Table 1. AT1-pGL3 and mAT1-pGL3 reporter vectors contained wild-type and mutant CRE sequences, respectively. Transfected cells were incubated with IBMX. RLA (mean Ϯ S.D., n ϭ at least three separate experiments) was measured (duplicate assays) in lysates from triplicate transfections. A-C, SV40 (Pro)-pGL3 refers to pGL3 expression vector driven by the SV40 promoter; SV40(Con)-pGL3, pGL3 expression vector driven by both the SV40 promoter and SV40 enhancer; Basic-pGL3, the pGL3-promoter-and enhancer-free vector. **, p Ͻ 0.01, statistically different from control.

Regulation of Murine Pde3b Gene Expression by CREB
fragment was markedly blocked by the portion of the downstream segment between the XbaI and PshAI sites (Ϫ3.4 kb to Ϫ0.5 kb). Activity of the SE fragment was observed only when it was coupled to certain portions of the proximal promoter region (between PshAI and PvuI restriction sites). Thus, these results further support the idea that the negative regulatory region between XbaI and PshAI sites could inhibit distal promoter activity, as well as proximal promoter activity (cf. Fig. 3, B  and C). Removal of the linker (ϳ13 bp, CGCGTGCTAGCCC) present between the distal and proximal promoters in each fusion construct did not apparently alter the effects of the negative region because the promoter activities of SEEB-pGL3 (without the linker) and SE-EcoR-pGL3 were virtually identical (Fig. 7).
ChIP Analyses of the Binding of CREB, ATF1, and CBP to the Proximal and Distal Promoter Regions in 3T3-L1 Preadipocytes-Treatment of preadipocytes with IBMX increased CREB phosphorylation (Fig. 8A). Antibodies to CREB, pCREB, ATF1, or CBP/p300 immunoprecipitated distal (Fig. 8B) and proximal (Fig. 8C) promoter fragments that contained CRE, whereas preimmune IgG and anti-CREM1 antibodies did not. Binding of anti-CREB and anti-phospho-CREB antibodies to the distal and proximal promoter sequences was greater in IBMX-treated preadipocytes than in untreated preadipocytes. Phosphorylation and activation of CREB increased its binding to both promoter regions. These results also suggest that IBMX enhances the interaction of CBP to the proximal promoter region to a greater extent than to the distal region. These findings are consistent with earlier reports of greater increases in activated/ phosphorylated CREB in differentiating 3T3-L1 adipocytes than preadipocytes (14,34).
Because the results in Figs. 5-8 demonstrated the importance of CRE cis-acting elements in the regulation of Pde3b expression, and because the distal promoter region contained a canonical CRE consensus sequence at Ϫ4166 to Ϫ4159 and a putative TATA box at Ϫ4093 to Ϫ4084 (Fig.  S1B), we speculated that Pde3b gene transcription might be driven by the distal promoter. Thus, as described in the Supplemental Material (cf. Fig. S1B), we attempted to use RT-PCR, RLM-5Ј-RACE, and RPA to identify transcripts that might be generated from the distal promoter region. However, multiple primer sets failed to amplify any Pde3b transcripts in the distal promoter region (data not shown). Also, RPA did not detect any protected RNA fragments following hybridization of adipocyte mRNA with antisense RNA probes designed based on sequence (nucleotides Ϫ4150 to Ϫ3867) from the distal promoter region of the Pde3b gene (data not shown). Although endogenous transcripts generated from the distal promoter region of the Pde3b gene were not identified, as seen in supplemental Fig. S1C, a fusion transcript generated in cells transfected with SX-pGL3 reporter vector (Fig. 4) indicated that there was transcription starting from SX and fusing to the firefly luciferase gene, which implied that the distal promoter could initiate transcription of an unidentified, noncoding exon or alternative exon1 from the murine Pde3b gene.

DISCUSSION
PDE3B appears or is markedly increased during differentiation of 3T3-L1 adipocytes in the presence of MDI (6,17,18). In adipocytes, PDE3B regulates cAMP pools important for lipolysis and the antilipolytic action of insulin but not for initiation of differentiation (20 -22). In addition, the presence of fat depots in PDE3B-deficient mice, recently generated in our laboratory (data not shown), also indicates that PDE3B is not required for differentiation. These data, however, do not rule out an important role(s) for Pde3b expression during differentiation or its impact on adipogenesis.
During differentiation in the presence of MDI, IBMX, which increased CREB phosphorylation, was the predominant regulator of Pde3b expression (Figs. 1 and 2). ChIP analysis indicated that phosphorylated CREB interacts with CRE cis-acting elements in the proximal and distal promoter regions of the 5Ј-flanking region of the Pde3b gene and that phosphorylated CREB interacts with and thereby recruits the multifunctional coactivator, CBP/p300 histone acetyltransferase, to Pde3b promoter regions. In keeping with the current hypotheses concerning coactivators and histone acetyltransferases (35,36), CBP/p300, by acetylating nucleosomal histone tails, could make the Pde3b promoter regions more accessible to other regulatory factors. CBP/p300 might also function as a scaffold for other transcription factors/coactivators or a bridge between Pde3b-specific factors and the general transcription factors and RNA polymerase II, and thus initiate transcription of Pde3b. In addition, CBP/p300 can catalyze covalent acetylation of proteins, including transcription factors and coactivators. Although CBP/p300 most likely is a critical coactivator for CREB-mediated regulation of Pde3b transcription, the relative importance of histone acetyltransferase activities, scaffold functions, and protein acetyltransferase activities in regulation of assembly of the transcriptional complex and Pde3b gene expression is not known. Although the precise role of cAMP signals in initiating adipocyte differentiation and adipogenesis in the intact rodent has not been defined, transgenic and knock-out models have identified components of cAMP-signaling pathways that are critical regulators of adipogenesis (37,38). For example, targeted disruption of the RII␤ gene of PKA produced mice with markedly reduced white adipose tissue depots that were resistant to dietinduced obesity and that exhibited increased basal lipolysis but reduced ␤-receptor stimulated lipolysis (37). cAMP has also been implicated in the expansion of brown adipose tissue in rodents, following activation of sympathetic innervation (response to cold temperature) or administration of ␤3-receptor agonists (39). Much of our current knowledge of mechanisms of adipogenesis has relied on use of model systems, including 3T3-L1 cells (26). Treatment of quiescent 3T3-L1 preadipocytes with dexamethasone, insulin, and an agent such as IBMX, which can increase cAMP, initiates the adipocyte differentiation program. Phosphorylation of CREB, mediated by cAMP and insulin (14,34,40,41), is crucial for initiation of differentiation, and phosphorylated CREB (presumably via recruitment of CBP) transcriptionally activates C/EBP␤ via cisacting CRE in its promoter. The activated preadipocytes reenter the cell cycle and undergo mitotic clonal expansion, during which time C/EBP␤ is phosphorylated, acquires DNA binding activity, and, along with C/EBP␦, transcriptionally activates the critical adipocyte transcription factors, C/EBP␣ and PPAR␥ (26). In addition, it has been suggested that not only does C/EBP␤ transcriptionally activate PPAR␥, but that C/EBP␤ and cAMP are involved in production of endogenous PPAR␥ ligands (42). cAMP may also be indirectly involved in transcriptional regulation of C/EBP␣, because the increase in cAMP produced by IBMX may down-regulate AP-2 and Sp-1 during differentiation. This may be FIGURE 7. Promoter activities of proximal/distal promoter fusion reporter vectors expressed in preadipocytes and differentiating preadipocytes. The SE fragment from the distal SalI/XbaI region (Figs. 4 and 5A) was coupled to the indicated fragments from the proximal promoter region (Figs. 3B and 4). Restriction sites used to generate the constructs are indicated. Promoter activity of the transfected proximal/distal promoter fusion constructs was measured as RLA (mean Ϯ S.D., n ϭ 3 experiments) as described under "Experimental Procedures." ࡗ, the 13-bp linker present between the SE fragment and the proximal promoter fragments. *, p Ͻ 0.05. critical because AP-2 and Sp-1 can inhibit expression of C/EBP␣ (26,31). The cAMP-induced decrease in AP-2 and Sp-1 may promote access of C/EBP␤ and/or C/EBP␦ to the C/EBP␣ regulatory elements and derepression of the C/EBP␣ gene that is required for 3T3-L1 differentiation. C/EBP␣ and PPAR␥, along with SREBP-1c and CREB (14,15), coordinately regulate genes responsible for acquiring and maintaining the adipocyte phenotype, perhaps including PDE3B (26,27).
Thus, IBMX-induced phosphorylation and activation of CREB, via cAMP and PKA (40,41), is important in regulation of transcription of some early genes that trigger the adipocyte differentiation program. Similarly, treatment of murine wild-type and KinϪ (cells that lack PKA) S49 lymphoma cells with the PKA-selective analog, 8-CPT-cAMP, indicated that cAMP, via activation of PKA, regulated transcription of a large number of genes and gene "networks" in these cells. Primary transcriptional networks that were regulated within 2 h of exposure to 8-CPT-cAMP were enriched in conserved CRE-cis-acting elements within ϳ5 kb of transcription initiation sites (43). In addition to its regulation of Pde3b expression, CREB is apparently also involved in induction of PDE4 (44) and PDE7 (45) gene transcription by cAMP. ChIP assays have suggested that there may be a selective increase in histone acetylation of the intronic promoter in PDE4D1/2 isoforms during their induction by cAMP in cultured vascular smooth muscle cells of the "synthetic/activated" phenotype but not the "contractile/quiescent" phenotype (46). It has also been suggested recently that critical signals in initiation of cardiomyocyte apoptosis involve a positive feedback loop leading to persistent down-regulation of PDE3A, i.e. reduction in PDE3A, activation of cAMP/PKA signaling, and subsequent induction/stabilization of inducible cAMP early repressor, which could compete with CREB binding to PDE3A promoter CRE sites (47,48).
Deletion analyses of promoter activity, utilizing luciferase reporter vectors (Figs. [5][6][7][8], demonstrated the presence of at least two distinct promoter regions in the 5Ј-flanking region of the Pde3b gene, separated by a negative regulatory gene segment. The proximal promoter region is located just upstream (within ϳ0.5-1 kb) of the TIS, determined by 5Ј-RACE and RPA to be ϳ300 bp upstream of the ATG initiation codon. It contains neither a typical TATA box nor a CAAT box but does contain atypical CRE cisacting elements and GC-rich sequences with multiple housekeeper promoter elements, including Sp-1, GC box, TATA-like box, and an initiator motif (49 -52). The distal promoter region, which contained a canonical CRE sequence, was mapped ϳ4 kb upstream of the ATG translation initiation codon. However, 5Ј-RACE (RT)-PCR, RLM-5Ј-RACE-PCR, and RPA analyses failed to detect transcripts in the distal promoter region or its immediate downstream region. After transfection of 3T3-L1 preadipocytes with a distal promoter reporter expression vector (SX-pGL3), however, a fusion transcript between the distal promoter region and the firefly luciferase gene was detected. This raised the possibility of the distal promoter generating Pde3b transcripts, still not identified, that might contain a small noncoding or alternative exon 1.
Canonical CRE cis-acting elements were clearly responsible for the strong activities of the distal promoter luciferase reporter vectors in transfected 3T3-L1 preadipocytes and differentiating adipocytes, as well as HEK293A cells. CREB-binding to target genes may be regulated in a cell-specific manner, and the ability of CREB to interact with specific CRE may be important in cell-specific regulation of gene expression (53). CRE may thus play a role in tissue-specific regulation of Pde3b gene expression, because PMA stimulated expression of several selected distal region reporter vectors (SE-pGL3, SX-pGL3, and SB-pGL3) in RAW monocytes, but not 3T3-L1 preadipocytes, whereas IBMX stimulated distal promoter activity in both cells. Mutation of the distal CRE sequences was associated with a marked reduction in distal promoter activities and loss of activation by IBMX in preadipocytes and HEK293A cells, and by MDI in differentiating adipocytes, as well as by PMA in HEK293A cells. IBMX-and PMA-stimulated expression of proximal promoter activity (NotI-pGL3 reporter vector) was inhibited by H-89 and BisI, respectively, further suggesting a role for CRE in regulation of Pde3b expression by both IBMX and PMA.
Because of the presence of the negative regulatory region and ambiguity regarding generation of transcripts from the distal promoter, the relative contributions of the distal and proximal regions in initiating transcription of the Pde3b gene have not been determined. Although the proximal promoter might drive transcription of the Pde3b gene, the canonical CRE in the distal promoter could function to generate alternate transcripts and/or to enhance regulation of Pde3b transcription through binding of transcription factors and coordination with the proximal promoter (35,36,41,54). Interaction of CREB with CBP/p300 could increase the interaction/association of CBP with multiple transcription factors and regulators and RNA polymerase II, and their recruitment to the proximal promoter region of the Pde3b gene (55,56), where they could collectively induce formation of a higher order complex, such as an enhanceosome (57). A loop formed between the distal and the proximal promoters could allow the distal promoter to interact/coordinate with the proximal promoter (54, 57). How- FIGURE 8. Effect of IBMX on chromatin immunoprecipitation of Pde3b promoter DNA sequences. A, 3T3-L1 preadipocytes, cultured in 1% FBS/ DMEM and treated with or without IBMX (300 M for 16 h, were harvested and homogenized. Samples (20 g of protein) were subjected to SDS-PAGE and immunoblotted with anti-CREB or anti-phospho-CREB antibodies. B and C, 3T3-L1 preadipocytes, treated with or without IBMX (300 M) for 16 h, were cross-lined with 1% formaldehyde. Chromatin-associated DNA was fragmented and immunoprecipitated with preimmune rabbit IgG or antibodies to CREB, phospho-CREB, ATF1, CREM1, and CBP/p300. Immunoprecipitated DNA was subjected to PCR amplification with specific primers flanking the CRE sites in the distal promoter region (B), and the putative CREB-binding sites in the proximal promoter region (C). In all cases, only the expected single ϳ244 bp (B) and single ϳ406 bp (C) PCR products were observed. Experiments in A-C were repeated, with similar results. ever, the identity of the transcription factors and coactivators, or protein complexes, which mediate CREB transactivation and Pde3b transcription, is not known.
Although insulin and dexamethasone alone were not as effective as IBMX in induction of Pde3b, both agents enhanced the effects of IBMX on Pde3b expression in differentiating preadipocytes. These agents might activate CREB via signaling pathways other than cAMP (40, 41, 56, 58 -60) or act through transcription factors other than CREB. There are several consensus motifs of insulin-response elements and peroxisomal proliferator response elements in the 5Ј-flanking region of the Pde3b gene. By activating transcription factors that bind to insulin-response elements and peroxisomal proliferator response elements, insulin and dexamethasone can regulate the expression of other adipocyte genes, including, perhaps, PPAR␥ and C/EBP␣ (61,62). Thus, C/EBP, insulin-response elements, and peroxisomal proliferator response element sites in the Pde3b gene might regulate Pde3b expression by acting coordinately with CRE. Taken together, our data suggest that cAMP-mediated CREB phosphorylation/activation plays a pivotal role in mediating Pde3b induction, whereas dexamethasone and insulin act as potentiators. Understanding the mechanisms of transcriptional control of Pde3b in adipose tissue, as well as pancreatic ␤-cells and liver tissues, could have important pathophysiological and therapeutic implications for obesity and noninsulin-dependent diabetes mellitus (63,64), and pharmacological implications for potential therapeutic actions of specific PDE3 inhibitors (64).