Role of Krüppel-like Factor 15 (KLF15) in Transcriptional Regulation of Adipogenesis*

Krüppel-like zinc finger transcription factors (KLFs) play diverse roles during cell differentiation and development in mammals. We have now shown by microarray analysis that expression of the KLF15 gene is markedly up-regulated during the differentiation of 3T3-L1 preadipocytes into adipocytes. Inhibition of the function of KLF15, either by expression of a dominant negative mutant or by RNA interference, both reduced the expression of peroxisome proliferator-activated receptor γ (PPARγ) and blocked adipogenesis in 3T3-L1 preadipocytes exposed to inducers of adipocyte differentiation. However, the dominant negative mutant of KLF15 did not affect the expression of CCAAT/enhancer-binding protein β (C/EBPβ) elicited by inducers of differentiation in 3T3-L1 preadipocytes. In addition, ectopic expression of KLF15 in NIH 3T3 or C2C12 cells triggered both lipid accumulation and the expression of PPARγ in the presence of inducers of adipocyte differentiation. Ectopic expression of C/EBPβ, C/EBPδ, or C/EBPα in NIH 3T3 cells also elicited the expression of KLF15 in the presence of inducers of adipocyte differentiation. Moreover, KLF15 and C/EBPα acted synergistically to increase the activity of the PPARγ2 gene promoter in 3T3-L1 adipocytes. Our observations thus demonstrate that KLF15 plays an essential role in adipogenesis in 3T3-L1 cells through its regulation of PPAR γ expression.

cell number, or both (2). The size of adipocytes varies markedly and reflects largely the amount of stored triglyceride, whereas the number of adipocytes is thought to increase as a result of the proliferation of preadipocytes and their subsequent differentiation into mature adipocytes.
Murine preadipose cell lines, such as 3T3-L1 and 3T3-F442A, have been studied extensively to elucidate the mechanisms of growth and differentiation of preadipocytes (2,3). In response to exposure to appropriate hormonal inducers (such as agents that increase the intracellular concentration of cyclic AMP, agonists of the insulin-like growth factor-1 receptor, glucocorticoids, and fetal bovine serum), these cells first undergo several rounds of mitosis, known as clonal expansion, and then become quiescent again, express adipocyte-specific proteins, and acquire biochemical and morphological characteristics of mature adipocytes (2,3).
Both the proliferation and differentiation of preadipocytes are characterized by marked changes in the pattern of gene expression that are achieved by the sequential induction of transcription factors. Preadipocytes exposed to inducers of differentiation thus manifest an early and transient increase in the expression of the transcription factors CCAAT/enhancerbinding protein ␤ (C/EBP␤) 1 and C/EBP␦, which in turn appear to contribute to cell proliferation as well as to the subsequent increase in the expression of C/EBP␣ and peroxisome proliferator-activated receptor ␥ (PPAR␥) (4,5). The latter two proteins are thought to act synergistically in the transcriptional activation of a variety of adipocyte-specific genes, with each also reciprocally activating the expression of the other (2)(3)(4).
Krü ppel-like zinc finger transcription factors (KLFs) are DNA-binding transcriptional regulators that contain the C 2 H 2 zinc finger motif and play diverse roles in the regulation of cell proliferation, cell differentiation, and development (6,7). All 16 members of the KLF family identified to date bind to GC-rich sequences including GC boxes and GT boxes (also known as CACCC boxes) (7,8). Certain KLF proteins have been implicated in adipogenesis. KLF2 has thus been shown to negatively regulate adipogenesis through inhibition of PPAR␥ gene expression (9). In addition, expression of KLF6 is transiently induced during adipogenesis in 3T3-L1 preadipocytes (10). Fur- thermore, overexpression of KLF15 induces adipocyte maturation and GLUT4 expression (11), although the physiological significance of KLF15 in the induction or maintenance of terminal differentiation has remained unclear. We now provide evidence that KLF15 promotes maintenance of the biochemical and morphological characteristics of mature adipocytes through direct induction of PPAR␥ expression in cooperation with C/EBP␣.
Cell Culture and Staining-3T3-L1 preadipocytes were maintained as described previously (15). Their differentiation into adipocytes was induced by treatment of confluent cells first for 2 days with insulin (5 g/ml), 0.25 M dexamethasone, and 0.5 mM isobutylmethylxanthine (IBMX) in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and then for 2 days with insulin (5 g/ml) alone in the same medium. The cells were then returned to the basal medium, which was replenished every other day.
NIH 3T3 or C2C12 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and cultured to confluence. Adipogenesis in transfected cells was induced by treatment for 6 days with insulin (5 g/ml), 0.25 M dexamethasone, 0.5 mM IBMX, and 10 M troglitazone in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum. The cells were then returned to the basal medium, which was replenished every other day.
Mouse embryonic fibroblasts (MEFs) obtained from PPAR␥ knockout mice or their wild-type littermates were immortalized by the 3T3 protocol (16) and exposed to inducers of adipocyte differentiation as described above for NIH 3T3 cells. Cells were stained with oil red O as described (17).
Oligonucleotide Microarray Analysis-Total RNA was isolated from 3T3-L1 preadipocytes or fully differentiated 3T3-L1 adipocytes (8 days after the induction of differentiation) and was processed with RNeasy columns (Qiagen, Hilden, Germany). Portions (10 g) of the RNA were then used for synthesis of biotin-labeled cRNA, which in turn was used to probe Genechip Mouse MU74 microarrays (Affymetrix, Santa Clara, CA). After washing and staining, the arrays were scanned with a Hewlett Packard confocal laser scanner and visualized with Affymetrix Genechip 3.1 software. Finally, the results were analyzed with Genechip Analysis Suite software version 4.0 (Affymetrix), and the -fold differences in hybridization intensity between samples from undifferentiated 3T3-L1 preadipocytes and fully differentiated 3T3-L1 adipocytes were determined.
Expression Plasmids-A mammalian expression vector for KLF15 (pcDNA3.1/KLF15) was constructed by inserting the products of reverse transcription (RT) and the PCR obtained from rat adipose tissue RNA into pcDNA3.1 (Invitrogen). To construct an expression vector for a deletion mutant of KLF15 that lacks the NH 2 -terminal 318 amino acids (⌬318), we amplified a DNA fragment encoding ⌬318 by PCR with the sense primer 5Ј-CTGCCATGCACAAATGCACTTTC-3Ј and the antisense primer 5Ј-CTTCAGTTGATGGCGCGTAC-3Ј and then inserted the PCR product into the PT7-blue T vector (Takara, Tokyo, Japan). The sequence of the inserted fragment was verified and then subcloned into pcDNA3.1. Mammalian expression vectors for C/EBP␣ or C/EBP␤ were constructed by subcloning the corresponding mouse cDNAs (kindly provided by S. Akira, Osaka University, Osaka, Japan) into pcDNA3.1.
NIH 3T3 cells stably expressing KLF15 (NIH/KLF15 cells) were generated with the retroviral vector pWZL/KLF15 as described (18), with the following modifications: Plat-E packaging cells were transfected with the retroviral vector with the use of FuGENE 6 (Roche Applied Science, Indianapolis, IN). After 24 h, the cells were incubated for an additional 24 h in fresh medium to obtain retrovirus-containing supernatants. NIH 3T3 cells (50 to 60% confluence) were then incubated with these supernatants in the presence of hexadimethrine bromide (4 g/ml) for 12 h. Cells expressing KLF15 were selected on the basis of their resistance to blasticidine S (10 g/ml) or hygromycin B (500 g/ml).
NIH 3T3 cells stably expressing other ectopic proteins with the exception of PPAR␥2; 3T3-L1 cells stably expressing GFP (3T3-L1/GFP cells), KLF15, or ⌬318; and C2C12 cells stably expressing GFP (C2C12/ FIG. 1. Microarray and Northern blot analyses of the expression of KLF genes during differentiation of 3T3-L1 cells. A, cells harvested before or 8 days after exposure to inducers of differentiation were subjected to DNA microarray analysis. Data for the indicated KLF genes are presented as the -fold change in adipocyte/preadipocyte ratio of mRNA abundance. B, cells harvested at the indicated times after exposure to inducers of differentiation were subjected to Northern blot analysis of the indicated mRNAs. The 28 S rRNA bands on the ethidium bromide-stained gel are also shown. All data are representative of at least three independent experiments. GFP cells) or KLF15 were generated as described above for NIH/KLF15 cells. NIH 3T3 cells stably expressing PPAR␥2 (NIH/PPAR␥2 cells) were similarly generated with pMXs/PPAR␥2 and were selected on the basis of their resistance to puromycin (2 g/ml).
Northern and Immunoblot Analyses-Total RNA (15 g) extracted from cells was subjected to Northern blot analysis with a full-length mouse aP2 cDNA, a fragment of mouse PPAR␥ cDNA, full-length mouse C/EBP␣ cDNA, full-length mouse C/EBP␤ cDNA, full-length mouse C/EBP␦ cDNA, full-length mouse KLF6 cDNA, full-length mouse KLF9 cDNA, full-length mouse KLF15 cDNA (RT-PCR product of mouse adipose tissue RNA), or a fragment of human GLUT4 cDNA as probes. Cell lysates (ϳ100 g of protein) were subjected to SDS-polyacrylamide gel electrophoresis and immunoblot analysis as described (15).
Electroporation-3T3-L1 preadipocytes were transfected with plasmids by electroporation as described previously (19). For transfection of 3T3-L1 adipocytes, the cells were detached from culture dishes 5 days after exposure to inducers of differentiation by treatment with 2.5% trypsin and collagenase (0.5 mg/ml); they were then washed with phosphate-buffered saline before resuspension with plasmid DNA in the FIG. 2. Effect of a dominant negative mutant of KLF15 on adipocytic differentiation of 3T3-L1 preadipocytes. A, NIH 3T3 cells were transfected with a GLUT4 promoter-luciferase gene reporter plasmid (or pGL3-basic as a control), a ␤-galactosidase expression plasmid, and expression vectors for KLF15, the ⌬318 mutant, or both KLF15 and ⌬318 (or the pcDNA3.1 empty vector). The cells were subsequently assayed for luciferase and ␤-galactosidase activities, and the former was normalized by the latter. Data represent normalized luciferase activity expressed relative to the value for cells transfected with pGL3-basic and pcDNA3.1; data are mean Ϯ S.D. of triplicates from a representative experiment. B, 3T3-L1 preadipocytes stably expressing GFP, KLF15, or ⌬318, as indicated, were subjected to Northern blot analysis of KLF15 mRNA. Control 3T3-L1 cells before or 8 days after exposure to inducers of differentiation were similarly analyzed. C, 3T3-L1 preadipocytes stably expressing GFP, KLF15, ⌬318, or both ⌬318 and PPAR␥2 were stained with oil red O 8 days after exposure to inducers of differentiation. Macroscopic (upper panel) and microscopic (ϫ100, lower panel) views are shown. D, 3T3-L1 preadipocytes stably expressing GFP, KLF15, ⌬318, or both ⌬318 and PPAR␥2 were subjected to immunoblot analysis with antibodies to PPAR␥, C/EBP␣, aP2, or CREB (control) 8 days after exposure to inducers of differentiation. E and F, 3T3-L1 adipocytes 8 days after exposure to inducers of differentiation were infected with adenovirus vector AxCA-lacZ or AxCA-Klf15 at the indicted multiplicity of infection. 48 h after adenovirus infection (MOI), the cells were subjected to Northern blot analysis of KLF15 mRNA (E) or subjected to immunoblot analysis with antibodies to PPAR␥, C/EBP␣, or CREB (control) (F). Data in B-F are representative of at least three independent experiments.

FIG. 3. Effect of a dominant negative mutant of KLF15 on the expression of C/EBP␤ or PPAR␥ during adipocyte differentiation in 3T3-L1
preadipocytes. 3T3-L1 preadipocytes stably expressing GFP, KLF15, or ⌬318 were exposed to inducers of differentiation for the indicated times and then subjected to immunoblot analysis with antibodies to C/EBP␤, PPAR␥, or CREB. Data are representative of at least three independent experiments.
solution provided with Nucleofector Kit V (Amaxa, Cologne, Germany). The plasmid DNA was then introduced into the cells by electroporation with the use of a Nucleofector instrument (program U-28). After electroporation, the cells were resuspended in culture medium and replated on culture dishes.
Luciferase Reporter Assays-For GLUT4 gene promoter assays, NIH 3T3 cells plated in 24-well dishes (50 to 60% confluence) were transiently transfected with 500 ng of the reporter plasmid pGL3-basic (Promega, Madison, WI) containing the human GLUT4 gene promoter (nucleotides Ϫ894 to ϩ84 relative to the transcriptional start site), 500 ng of the indicated expression plasmids, and 100 ng of pCMV encoding ␤-galactosidase (pCMV/␤-gal) with the use of the Lipofectamine reagent (Invitrogen). After incubation for 48 to 72 h, the cells were lysed in 100 l of 1 times passive lysis buffer (Promega), and portions of the lysate were subjected to assays for firefly luciferase (Promega) and ␤-galactosidase (Clontech, Palo Alto, CA) activities. Promoter activity was determined as the ratio of luciferase to ␤-galactosidase activities. For PPAR␥2 gene promoter assays, 3T3-L1 adipocytes (ϳ3 ϫ 10 6 ) were transfected with 500 ng of pGL3-basic containing the mouse PPAR␥2 gene promoter (nucleotides Ϫ626 to ϩ60), 500 ng of the indicated expression plasmids, and 100 ng of pCMV/␤-gal by electroporation and were subsequently assayed for firefly luciferase and ␤-galactosidase activities as described for the GLUT4 gene promoter assay.
RNA Interference-The target sequences of the KLF15 gene selected for RNA interference (RNAi) comprised nucleotides 300 to 319 and nucleotides 1112 to 1131 relative to the transcriptional start site. To construct mammalian expression vectors for each of the corresponding small interfering RNAs (siRNAs), designated si300 and si1112, respectively, we amplified by PCR a DNA fragment including the mouse U6 (mU6) promoter and encoding the sense sequence of the siRNA, a short hairpin sequence, the antisense sequence of the siRNA, and a poly(A) terminator. PCR was performed for 35 cycles of 95°C for 30 s, 45°C for 30 s, and 72°C for 30 s with specific primers, the pMXs/mU6 plasmid (kindly provided by K. I. Nakayama, Kyushu University, Fukuoka, Japan) as the template, and Ex-taq polymerase (Takara). The PCR primers were 5Ј-AGCGC-CCGACCGAAAGGAG-3Ј (sense) and 5Ј-ACTAAGGTCGACAAAAAGCC-TTCTGTTCCTGATACATCTCTTGAATGTAGCAGGAACAGAAGGCAA-ACAAGGCTTTTCTCCAAG-3Ј (antisense) for si300 and 5Ј-AGCGCCCG-ACCGAAAGGAG-3Ј (sense) and 5Ј-ACTAAGGTCGACAAAAAGATGTA-CACCAAGAGCAGCTCTCTTGAAGCTGCTCTTGGTGTACATCAAACA-AGGCTTTTCTCCAAG-3Ј (antisense) for si1112. The PCR products were subcloned into PT7-blue T and their sequences were verified with the CUGA sequencing system (Nippongene, Tokyo, Japan). They were then subcloned into the NotI and SalI sites of pcDNA3.1, from which the cytomegalovirus promoter was subsequently deleted.

Up-regulation of KLF15 Gene Expression during Adipocyte
Differentiation in 3T3-L1 Cells-We first profiled genes whose level of expression changed in association with the differentiation of 3T3-L1 preadipocytes induced by IBMX, dexametha- . Cells stably expressing KLF6, KLF9, or KLF15 were also stained with oil red O at 10 days after exposure to inducers of adipocyte differentiation; microscopic views (ϫ100) of the stained cells are shown (right panel). B, before or 10 days after exposure to inducers of adipocyte differentiation, NIH 3T3 cells stably expressing KLF6, KLF9, or KLF15 were subjected to Northern blot analysis of PPAR␥, aP2, C/EBP␣, and GLUT4 mRNAs. 3T3-L1 cells before or 8 days after exposure to differentiation inducers were similarly analyzed for comparison. C, C2C12 cells infected with retroviral vectors for GFP or KLF15 were subjected to Northern blot analysis of KLF15 mRNA (upper panel). They were also stained with oil red O at 10 days after exposure to inducers of adipocyte differentiation; microscopic views (ϫ100) are shown (lower panel). D, C2C12 cells stably expressing GFP or KLF15 were subjected to Northern blot analysis of PPAR␥ and aP2 mRNAs at 10 days after exposure to inducers of adipocyte differentiation. All data are representative of at least three independent experiments. sone, and insulin. Total RNA isolated from 3T3-L1 preadipocytes and adipocytes was analyzed with mouse oligonucleotide microarrays, revealing that the differentiation of these cells was accompanied by an ϳ80-fold increase in the abundance of KLF15 mRNA (Fig. 1A). The amounts of KLF9, KLF10, and KLF13 mRNAs were increased to much smaller extents, whereas those of KLF6 mRNA and of several other KLF gene transcripts were decreased, in association with 3T3-L1 cell differentiation. Consistent with these results, Northern blot analysis showed that the abundance of KLF15 and KLF9 mRNAs increased during the differentiation of 3T3-L1 cells, whereas that of KLF6 mRNA decreased (Fig. 1B). KLF15 mRNA was first detected 4 days after the induction of differentiation, 2 days later than the initial increase in the amounts of PPAR␥ and aP2 mRNAs, and its abundance was increased maximally at 6 and 8 days. The time course of the up-regulation of KLF15 mRNA during adipocyte differentiation was similar to that apparent for C/EBP␣ and GLUT4 mRNAs. The amounts of C/EBP␤ and C/EBP␦ mRNAs were already maximal at 2 days and had returned to basal levels by 4 to 6 days after the induction of differentiation.
Inhibition of KLF15 Function by a Dominant Negative Mutant in 3T3-L1 Cells-To determine whether KLF15 is required for adipocyte differentiation in 3T3-L1 cells, we constructed an expression vector for a putative dominant negative mutant (⌬318) that lacks the NH 2 -terminal 318 amino acids of KLF15, which contain the transactivation domain, leaving the DNA binding domain intact. To verify that the ⌬318 mutant was able to inhibit KLF15 function, we examined the effect of its expres-sion on the activity of the GLUT4 gene promoter, a target of KLF15 action (11). We transiently cotransfected NIH 3T3 cells with a GLUT4 promoter-luciferase gene reporter construct together with an expression plasmid either for full-length KLF15 or for the ⌬318 mutant. The activity of the GLUT4 gene promoter was increased by ϳ50% by expression of wild-type KLF15 and was reduced by ϳ50% by expression of ⌬318 ( Fig.  2A). In addition, ⌬318 prevented the increase in GLUT4 gene promoter activity induced by the full-length protein, confirming that ⌬318 acts in a dominant negative manner.
We next constructed retroviral vectors encoding ⌬318, wildtype KLF15, and GFP for expression of these proteins in 3T3-L1 preadipocytes (Fig. 2B). Expression of ⌬318 in these cells greatly inhibited both the lipid accumulation (Fig. 2C) and the induction of PPAR␥, C/EBP␣, and aP2 (Fig. 2D) normally apparent 8 days after exposure of the cells to inducers of differentiation. Expression of GFP or of wild-type KLF15 did not affect these aspects of adipocyte differentiation. The amount of exogenous KLF15 mRNA by a retrovirus vector was one-half of that of endogenous KLF15 in 3T3-L1 adipocytes (Fig. 2B). We have therefore investigated the effect of overexpression of KLF15 by an adenovirus vector on the induction of PPAR␥ and C/EBP␣. Although the abundance of KLF15 mRNA in 3T3-L1 adipocytes infected with an adenovirus vector for this protein at an multiplicity of infection of 30 plaque-forming units (pfu)/cell was eight times to that of endogenous KLF15 mRNA in 3T3-L1 adipocytes (Fig. 2E), this overexpression of KLF15 did not affect the amount of PPAR␥ and C/EBP␣ protein (Fig. 2F) in 3T3-L1 adipocytes. To examine whether this

FIG. 6. Effects of overexpression of KLF15 on adipocyte differentiation in immortalized PPAR␥؉/؉ or PPAR␥؊/؊ MEFs as well as ectopic expression of PPAR␥2 on the abundance of KLF15 mRNA in NIH 3T3 cells.
A, PPAR␥ϩ/ϩ or PPAR␥Ϫ/Ϫ MEFs were infected with retroviral vectors for GFP, KLF15, or PPAR␥2 and subsequently subjected to Northern blot analysis of PPAR␥ and KLF15 mRNAs. B and C, MEFs stably expressing GFP, KLF15, or PPAR␥2 were exposed to inducers of adipocyte differentiation and, after 10 days, either stained with oil red O (microscopic views are shown; ϫ100) (B) or subjected to Northern blot analysis of aP2 mRNA (C). D, NIH 3T3 cells infected with a retroviral vector for PPAR␥2 (or with the empty vector) were subjected to Northern blot analysis of PPAR␥ mRNA. 3T3-L1 cells before or 8 days after exposure to inducers of adipocyte differentiation were similarly analyzed for comparison. E, NIH 3T3 cells stably expressing PPAR␥2 were exposed to inducers of adipocyte differentiation and, after 10 days, stained with oil red O (microscopic views are shown; ϫ100). F, NIH 3T3 cells stably expressing PPAR␥2 before or 10 days after exposure to inducers of adipocyte differentiation were subjected to Northern blot analysis of KLF15 mRNA. 3T3-L1 cells before or 8 days after exposure to inducers of adipocyte differentiation were similarly analyzed for comparison. All data are representative of at least three independent experiments. inhibitory effect of ⌬318 on adipogenesis was prevented by coexpression of PPAR␥2, we infected 3T3-L1 cells both with a ⌬318-encoding retroviral vector containing a blasticidine resistance gene and with a PPAR␥2-encoding retroviral vector containing a puromycin resistance gene. The infected cells were then subjected to selection in the presence of both blasticidine S and puromycin. The inhibitory effect of ⌬318 on adipogenesis, as monitored by both lipid accumulation (Fig. 2C) and the expression of C/EBP␣ and aP2 (Fig. 2D), was indeed prevented by coexpression of PPAR␥2, indicating that it was attributable to the prevention of PPAR␥ induction.
Lack of Effect of ⌬318 on the Expression of C/EBP␤ in 3T3-L1 Cells-The increase in the amount of KLF15 mRNA during adipocytic differentiation of 3T3-L1 cells occurred slightly later than did that in the amount of PPAR␥ mRNA (Fig. 1B). PPAR␥ mRNA was thus detected 2 days after exposure of cells to inducers of differentiation, whereas KLF15 mRNA was not apparent at this time. The transient increase in the abundance of the mRNAs for C/ebp␤ and C/ebp␦, both of which directly induce the expression of PPAR␥ (20,21), was maximal 2 days after induction of differentiation. Furthermore, MEFs that lack C/EBP␤ and C/EBP␦ are unable to differentiate into adipocytes (22). These two proteins thus likely induce the expression of PPAR␥ before the induction of C/EBP␣ or KLF15 during adipogenesis in 3T3-L1 cells. We next investigated whether the expression of the dominant negative mutant of KLF15 (⌬318) affected the induction of C/EBP␤ or PPAR␥ during the 4 days of exposure of 3T3-L1 cells to inducers of differentiation, a period when the levels of C/EBP␤ and C/EBP␦ are increased. Immunoblot analysis revealed that expression of ⌬318 in 3T3-L1 cells inhibited the expression of PPAR␥ on day 4 (Fig. 3) as it did on day 8 (Fig. 2D); it did not affect PPAR␥ abundance on day 2 (Fig. 3), however, a time before induction of KLF15 gene expression is apparent. The up-regulation of C/EBP␤ expression, which was maximal 1 and 2 days after exposure to the inducers of differentiation, was not inhibited by ⌬318 (Fig. 3). Moreover, overexpression of wild-type KLF15 slightly increased the abundance of PPAR␥ at 1 or 2 days after the induction of differentiation (Fig. 3), consistent with the previous observation that KLF15 promoted adipocyte differentiation (9).
Inhibition of Adipogenesis by RNAi-mediated Depletion of KLF15 in 3T3-L1 Cells-It was possible that expression of a dominant negative mutant (⌬318) of KLF15 that contains the DNA binding domain of the protein affected transactivation by other KLF isoforms. To exclude this possibility, we designed plasmid constructs that encode two different siRNAs (siRNA300 and siRNA1112) and are controlled by the mU6 promoter (23) to deplete KLF15 from 3T3-L1 cells. Cotransfection with the mU6/siRNA1112 plasmid resulted in a marked reduction in the amount of KLF15 mRNA in NIH 3T3 or 293 cells also transfected with an expression vector for KLF15, whereas cotransfection with the mU6/siRNA300 plasmid had no such effect (data not shown). We therefore transfected 3T3-L1 preadipocytes with the mU6/siRNA1112 plasmid, a plasmid containing only the mU6 promoter, or the corresponding empty plasmid by electroporation. Similar electroporation with an expression vector for ␤-galactosidase revealed that Ͼ70% of 3T3-L1 preadipocytes expressed the exogenous gene (19), indicative of a high efficiency of transfection. Transfection of 3T3-L1 cells with the mU6/siRNA1112 plasmid resulted in marked inhibition both of lipid accumulation (Fig. 4A) and of the expression of KLF15, PPAR␥, C/EBP␣, aP2, and GLUT4 at the mRNA (Fig. 4B) or protein (Fig. 4C) levels apparent 8 days after exposure of the cells to inducers of differentiation. Transfection with either the mU6 plasmid or the empty plasmid had no such effects. These data thus confirmed that induction of KLF15, rather than that of other KLFs, is essential for adipocytic differentiation of 3T3-L1 cells.
Induction of Adipogenesis by Ectopic Expression of KLF15 in NIH 3T3 and C2C12 Cells-To examine whether KLF15 is able to induce adipogenesis, we next determined the effect of ectopic expression of this protein on adipocyte differentiation in NIH 3T3 cells, which, unlike 3T3-L1 preadipocytes, are not committed to the adipocyte lineage and do not express PPAR␥ or C/EBP␣ in response to treatment with IBMX, dexamethasone, insulin, and synthetic PPAR␥ ligands such as ciglitazone (24). NIH 3T3 cells were infected with retroviral vectors for KLF15 or for KLF6 or KLF9, changes in the expression of which also accompanied adipocyte differentiation in 3T3-L1 cells (Fig. 1). The infected cells were then exposed to IBMX, dexamethasone, insulin, and the synthetic PPAR␥ ligand troglitazone. Such treatment did not induce lipid accumulation (Fig. 5A) or the expression of PPAR␥, aP2, C/EBP␣, or GLUT4 genes (Fig. 5B) in cells infected with the KLF6 or KLF9 vectors. In contrast, the cells infected with the KLF15 vector exhibited marked lipid accumulation (Fig. 5A) and expression of PPAR␥ and aP2 genes (Fig. 5B). Induction of the C/EBP␣ gene was not apparent in the KLF15-overexpressing cells, however, suggesting that adipogenesis mediated by KLF15 was dependent largely on the induction of PPAR␥. Induction of GLUT4 gene expression was induced by KLF15 even in the absence of differentiation inducers, consistent with the previous observation that KLF15 binds to and transactivates the GLUT4 gene promoter (11). These effects of KLF15 on adipocyte differentiation were also observed in the multipotent cell line C2C12, which possesses the ability to differentiate into myocytes (25) or osteoblastic cells (26). Treatment with IBMX, dexamethasone, insulin, and tro-

FIG. 7. Effect of ectopic expression of C/EBP␤ or C/EBP␦ on the induction of KLF15 in NIH 3T3 cells.
A, retrovirus-mediated expression of C/EBP␤ or C/EBP␦ in NIH 3T3 cells was examined by immunoblot analysis. 3T3-L1 cells either before or 2 days or 6 h after exposure to inducers of adipocyte differentiation were similarly analyzed for comparison. B, NIH 3T3 cells stably expressing C/EBP␤ or C/EBP␦ were subjected to Northern blot analysis of PPAR␥ and KLF15 mRNAs before or 10 days after exposure to inducers of adipocyte differentiation. 3T3-L1 cells before or 8 days after exposure to differentiation inducers were similarly analyzed for comparison. All data are representative of at least three independent experiments. glitazone thus induced both lipid accumulation (Fig. 5C) and PPAR␥ and aP2 gene expression (Fig. 5D) in C2C12 cells infected with a retroviral vector for KLF15 but not in those infected with a control vector for GFP. These data indicated that KLF15 possesses the ability to execute adipogenesis through the induction of PPAR␥ in NIH 3T3 and C2C12 cells, neither of which are programmed to express PPAR␥.
We next tested whether PPAR␥ was able to induce the expression of KLF15 in NIH 3T3 cells. The abundance of PPAR␥2 mRNA in NIH 3T3 cells infected with a retroviral vector for this protein was similar to that of endogenous PPAR␥ mRNA in 3T3-L1 adipocytes (Fig. 6D). Overexpression of PPAR␥2 elicited adipogenesis in NIH 3T3 cells exposed to inducers of adipocyte differentiation (Fig. 6E), but it did not affect the amount of KLF15 mRNA in these cells (Fig. 6F). These results suggested that PPAR␥ is a downstream effector in the induction of adipocyte differentiation by KLF15.
Induction of KLF15 by Ectopic Expression of C/EBP␤ or C/EBP␦ in NIH 3T3 Cells-C/EBP␤ and C/EBP␦ are early participants in the adipocytic differentiation of 3T3-L1 cells (Fig.  1B), and ectopic expression of these proteins induces the differentiation of NIH 3T3 cells into adipocytes (20,21). We therefore examined whether C/EBP␤ or C/EBP␦ were able to induce the expression of KLF15 in NIH 3T3 cells. Immunoblot analysis revealed that the abundance of C/EBP␤ or C/EBP␦ in NIH 3T3 cells infected with retroviral vectors for these proteins was similar to that of the endogenous proteins in 3T3-L1 adipocytes at 2 days or 6 h, respectively, after exposure to inducers of differentiation (Fig. 7A). Overexpression of C/EBP␤ or C/EBP␦ induced FIG. 8.
Cooperative effect of C/EBP␣ and KLF15 on adipocyte differentiation. A, retrovirus-mediated expression of C/EBP␣ in NIH 3T3 cells was examined by immunoblot analysis. 3T3-L1 cells either before or 8 days after exposure to inducers of adipocyte differentiation were similarly analyzed for comparison. B, NIH 3T3 cells stably expressing C/EBP␣ were subjected to Northern blot analysis of PPAR␥ and KLF15 mRNAs before or 10 days after exposure to inducers of adipocyte differentiation. 3T3-L1 cells before or 8 days after exposure to differentiation inducers were similarly analyzed for comparison. C, retrovirus-mediated expression of C/EBP␣ or KLF15 in NIH 3T3 cells was examined by Northern blot analysis. D and E, NIH 3T3 cells stably expressing C/EBP␣, KLF15, or both C/EBP␣ and KLF15 were exposed to inducers of differentiation and, after 10 days, either stained with oil red O (microscopic views are shown; ϫ100) (D) or subjected to Northern blot analysis of PPAR␥, aP2, and GLUT4 mRNAs (E). Data in A-E are representative of at least three independent experiments. F, 3T3-L1 adipocytes at 5 days after exposure to inducers of differentiation were transfected with a PPAR␥2 promoter-luciferase gene reporter plasmid, ␤-galactosidase expression plasmid, and expression vectors for C/EBP␣, C/EBP␤, or KLF15, as indicated. The cells were subsequently assayed for luciferase and ␤-galactosidase activities, and the former was normalized by the latter. Data represent normalized luciferase activity expressed relative to the value for cells transfected with the empty expression vector (pcDNA3.1); they are mean Ϯ S.D. of triplicates from a representative experiment. lipid accumulation (data not shown) and the up-regulation of both PPAR␥ and KLF15 mRNAs (Fig. 7B) in NIH 3T3 cells exposed to inducers of adipocyte differentiation, suggesting that KLF15 is a downstream effector of C/EBP␤ or C/EBP␦ in the induction of PPAR␥ expression during adipocyte differentiation.
Synergistic Induction of Adipocyte Differentiation by KLF15 and C/EBP␣-The time course of the induction of KLF15 gene expression during adipocytic differentiation of 3T3-L1 cells was similar to that of the up-regulation of C/EBP␣ mRNA (Fig. 1B). We therefore examined whether C/EBP␣ was able to induce the expression of KLF15 in NIH 3T3 cells. Immunoblot analysis revealed that the abundance of C/EBP␣ in NIH 3T3 cells infected with retroviral vectors for these proteins was similar to that of the endogenous proteins in 3T3-L1 adipocytes at 8 days, after exposure to inducers of differentiation (Fig. 8A). Overexpression of C/EBP␣ induced lipid accumulation (Fig. 8D) and the upregulation of both PPAR␥ and KLF15 mRNAs (Fig. 8B) in NIH 3T3 cells exposed to inducers of adipocyte differentiation, suggesting that KLF15 is also a downstream effector of C/EBP␣ in the induction of PPAR␥ expression during adipocyte differentiation.
We next investigated whether KLF15 promotes adipocyte differentiation in cooperation with C/EBP␣. We infected NIH 3T3 cells both with a C/EBP␣ retroviral vector that contained a blasticidine resistance gene and with a KLF15 retroviral vector that contained a hygromycin resistance gene, and then subjected the cells to selection with both blasticidine S and hygromycin B (Fig. 8C). Whereas overexpression of KLF15 or C/EBP␣ alone each induced lipid accumulation (Fig. 8D) and up-regulation of PPAR␥ and aP2 mRNAs (Fig. 8E) in NIH 3T3 cells exposed to inducers of adipocyte differentiation, the coexpression of these proteins had an even more pronounced effect on adipocyte differentiation. The amount of GLUT4 mRNA was increased in cells overexpressing KLF15 but not in those overexpressing C/EBP␣ alone (Fig. 8E). Finally, we examined the effect of overexpression of C/EBP␣, C/EBP␤, or KLF15 on the activity of a PPAR␥2 promoter-luciferase gene reporter construct in 3T3-L1 adipocytes. The activity of the PPAR␥2 gene promoter was increased ϳ5and 3.5-fold by overexpression of C/EBP␣ or KLF15, respectively, but was not affected by that of C/EBP␤ (Fig. 8F). In addition, coexpression of KLF15 and C/EBP␣ induced a ϳ14-fold increase in the activity of the PPAR␥2 gene promoter (Fig. 8F), consistent with the notion that KLF15 and C/EBP␣ act synergistically to promote adipocyte differentiation. DISCUSSION We have shown here that up-regulation of transcription factor KLF15 plays an essential role in the differentiation of 3T3-L1 preadipocytes into adipocytes, and that this action of KLF15 is mediated, at least in part, through induction of PPAR␥ expression. Ectopic expression of KLF15 thus induced the expression of PPAR␥ in both NIH 3T3 and C2C12 cells. In addition, PPAR␥ expression was found to be required for KLF15-induced adipocyte differentiation in MEFs. These data indicate that KLF15 possesses the ability to mediate adipogenesis through the induction of PPAR␥ expression. KLF15 was originally identified on the basis of its ability to bind to a GA element in the promoter of the CLC-K1 gene, which encodes a kidney-specific member of the CLC family of Cl Ϫ channels (28). The gene for the insulin-sensitive glucose transporter GLUT4 in adipose or muscle tissue and the acetyl-CoA synthetase 2 gene in muscle are also targets of KLF15, which binds to CACCC elements in the promoters of these genes (11,29). The PPAR␥2 gene contains two tandem KLF binding sites (shown in bold) that bind KLF2 (9) within the sequence 5Ј-CCCACCTCTCCCA-3Ј (nucleotides Ϫ82 to Ϫ70 relative to the transcriptional start site). However, the activity of the PPAR␥2 gene promoter was increased 3.5-fold by overexpression of KLF15 alone (Fig. 8F), but was not affected by the deletion of the CACCC elements of PPAR␥2 gene promoter. 2 Furthermore, we could not find that KLF15 bound to the CACCC ele-  ments in the promoter of the PPAR␥2 by using chromatin immunoprecipitation assay. 2 These data suggest that KLF15 binds to different sites from these CACCC elements and thereby increases the promoter activity of the PPAR␥2 gene, but they do not exclude the possibility that KLF15 may function through interaction with other transcription factors. The exact nature of KLF15-mediated activation of PPAR␥2 transcription warrants further investigation. Inhibition of KLF15 function by expression of a dominant negative mutant (⌬318) or by RNAi during exposure of 3T3-L1 preadipocytes to inducers of differentiation resulted in inhibition of the expression of PPAR␥ measured on day 8, but ⌬318 did not inhibit the expression of PPAR␥ measured on day 2 (before the onset of KLF15 expression). The abundance of C/EBP␤ mRNA and protein increased transiently during adipocytic differentiation of 3T3-L1 cells, reaching a maximum within 2 days and returning to basal levels on day 4, at which time KLF15 mRNA was detected. The initial induction of PPAR␥ expression is thus likely mediated by C/EBP␤, rather than by KLF15, during adipogenesis in 3T3-L1 cells, and this effect of C/EBP␤ is likely the direct result of its binding to C/EBP recognition sites in the PPAR␥ gene promoter (30,31). We also demonstrated that C/EBP␤ and C/EBP␦ were each able to induce the expression of KLF15 as well as that of PPAR␥ in NIH 3T3 cells. The expression of KLF15 induced by C/EBP␤ and C/EBP␦ during adipocyte differentiation in 3T3-L1 cells is thus likely responsible for that of PPAR␥ after the abundance of C/EBP␤ and C/EBP␦ has decreased on day 4. In addition, we found that KLF15 and C/EBP␣ acted synergistically to activate transcription of the PPAR␥2 gene in 3T3-L1 adipocytes. KLF15 might thus function to maintain the differentiated state by mediating the persistent expression of PPAR␥ in cooperation with C/EBP␣.
We showed that ectopic expression of KLF15 in NIH 3T3 cells induced lipid accumulation and the expression of PPAR␥ after exposure of the cells to IBMX, dexamethasone, insulin, and the synthetic PPAR␥ ligand troglitazone. The omission of troglitazone resulted in a similar extent of PPAR␥ induction but largely prevented lipid accumulation (data not shown). These results suggest that production of the cellular ligand for PPAR␥ is not regulated by KLF15. Previous studies have suggested a role for C/EBP␤ (32) or ADD1 (SREBP-1) (33) in the transient production of the cellular PPAR␥ ligand during an early phase of adipocyte differentiation in 3T3-L1 cells (34).
In conclusion, our study has demonstrated an essential role for KLF15 in the complex regulation of gene transcription during adipogenesis. A model for the transcriptional control of adipogenesis in 3T3-L1 cells that takes into account both our present data and those of previous studies (2,20,21,35,36) is shown in Fig. 9. A transient increase in the expression of C/EBP␤ and C/EBP␦ in response to inducers of differentiation results in up-regulation of PPAR␥ expression (20,21,37). C/EBP␤ and C/EBP␦ also activate transcription of the KLF15 and C/EBP␣ genes, the proteins encoded by which then act synergistically to maintain the increase in PPAR␥ expression after the levels of C/EBP␤ and C/EBP␦ have decreased. PPAR␥ is also able to increase expression of C/EBP␣, thereby establishing a positive feedback loop between PPAR␥ and C/EBP␣ (36,38,39). In addition, KLF15 directly induces the expression of adipocyte-specific genes such as that for GLUT4 (11). The cooperative effects of KLF15, PPAR␥, and C/EBP␣ are thus responsible for the full range of adipocytespecific gene expression.