Re-expression of GATA2 Cooperates with Peroxisome Proliferator-activated Receptor-γ Depletion to Revert the Adipocyte Phenotype*

Nuclear peroxisome proliferator-activated receptor-γ (PPARγ) is required for adipocyte differentiation, but its role in mature adipocytes is less clear. Here, we report that knockdown of PPARγ expression in 3T3-L1 adipocytes returned the expression of most adipocyte genes to preadipocyte levels. Consistently, down-regulated but not up-regulated genes showed strong enrichment of PPARγ binding. Surprisingly, not all adipocyte genes were reversed, and the adipocyte morphology was maintained for an extended period after PPARγ depletion. To explain this, we focused on transcriptional regulators whose adipogenic regulation was not reversed upon PPARγ depletion. We identified GATA2, a transcription factor whose down-regulation early in adipogenesis is required for preadipocyte differentiation and whose levels remain low after PPARγ knockdown. Forced expression of GATA2 in mature adipocytes complemented PPARγ depletion and impaired adipocyte functionality with a more preadipocyte-like gene expression profile. Ectopic expression of GATA2 in adipose tissue in vivo had a similar effect on adipogenic gene expression. These results suggest that PPARγ-independent down-regulation of GATA2 prevents reversion of mature adipocytes after PPARγ depletion.

Peroxisome proliferator-activated receptors (PPARs) 2 are ligand-activated transcription factors that bind to PPAR response elements as heterodimers with the retinoid X receptor ␣ (1). Among all PPARs, the expression of PPAR␥ exhibits the greatest specificity for adipose tissue (2,3), and the antidiabetic thiazolidinedione drugs are high affinity PPAR␥ ligands that promote adipogenesis (4). One of the most common models used to study adipocyte differentiation is the mouse 3T3-L1 cell line. After hormonal stimulation of growth-arrested preadipocytes, cells undergo a clonal expansion phase before they permanently exit the cell cycle for the final adipocyte commitment (5). During this process, PPAR␥ is induced and stimulates the expression of many adipocyte-specific genes. PPAR␥ is both necessary (6) and sufficient (7) for the differentiation of murine fibroblasts into adipocytes, and no factor is known to stimulate adipocyte differentiation in the absence of PPAR␥.
PPAR␥ itself is activated mainly by hormonally induced changes in the expression of transcriptional activators and repressors. Upstream of the induction of PPAR␥ are, for instance, EGR2 (early growth response 2) and members of the Krüppel-like factors and C/EBP (CCAAT/enhancer binding protein) families. C/EBP␤ has been shown to directly or indirectly induce PPAR␥ expression (5). On the other hand, several inhibiting factors upstream of PPAR␥ are regulators of alternative cell fates, including some members of the WNT family (8) as well as factors highly expressed in preadipocytes and downregulated during differentiation, including PREF-1 and GATA2/3 (9 -11).
In contrast to the thoroughly studied role of PPAR␥ during adipocyte differentiation, our knowledge about the role of PPAR␥ in maintaining the adipocyte phenotype in mature adipocytes is limited. PPAR␥ activation has been shown to regulate a variety of genes involved in glucose and lipid metabolism as well as many secreted adipokines (12), suggesting a central position of PPAR␥ in the transcriptional control of adipocyte gene expression. Indeed, recent genome-wide analyses identified thousands of binding sites of PPAR␥ in mature 3T3-L1 adipocytes (13,14).
In vivo, complete depletion of PPAR␥ in mature adipocytes resulted in cell death and replacement with newly differentiated PPAR␥-expressing cells (15). However, lentivirus-mediated knockdown of PPAR␥ in 3T3-L1 adipocytes altered glucose uptake and inflammatory responses but did not lead to dedifferentiation of committed adipocytes (16). Here, we confirmed that adipocyte characteristics were maintained after knock-down of PPAR␥. Although most adipocyte-specific genes returned to preadipocyte levels after PPAR␥ depletion, we identified a subset of genes whose regulation during adipogenesis was not reversed by PPAR␥ knockdown. Among these genes was Gata2, which was down-regulated during adipocyte differentiation and remained low after prolonged PPAR␥ depletion. Forced expression of GATA2 in 3T3-L1 adipocytes as well as adipose tissue in vivo impaired adipocyte functionality and, in combination with PPAR␥ depletion, led to a more preadipocyte-like gene expression profile. These results suggest that PPAR␥-independent regulation of Gata2 prevents reversion of mature adipocytes after PPAR␥ depletion.

EXPERIMENTAL PROCEDURES
Cell Culture and Differentiation-Reagents were obtained from Invitrogen unless otherwise noted. 293A and murine 3T3-L1 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (U. S. Bio-Technologies Inc., Parkerford, PA), 100 units/ml penicillin, and 100 g/ml streptomycin. 3T3 cells were grown to confluence and induced to differentiate 2 days after confluence with medium containing 1 M dexamethasone, 10 g/ml bovine insulin, and 0.5 mM 3-isobutyl-1-methylxanthine (all from Sigma) for 2 days and for an additional 2 days in insulin only. Adipocytes were considered mature after 8 days with at least 95% conversion into the adipocyte morphology.
Transfection and Knockdown-3T3-L1 cells at the indicated stages of differentiation were transfected by electroporation (Nucleofector II, Amaxa). Cells were detached from culture dishes with 0.25% trypsin and 0.5 mg/ml collagenase in phosphate-buffered saline, washed twice, resuspended in electroporation buffer (solution V, Amaxa), mixed with 2 or 3 nmol of control nontargeting oligonucleotide (Dharmacon) or siRNA oligonucleotides (supplemental Table S1), and seeded into 12-well plates after electroporation.
Immunoblot Analysis-Proteins were isolated and separated in 4 -20% SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membrane. After incubation with the primary antibodies for PPAR␥ (sc-7273, Santa Cruz Biotechnology), GATA2 (sc-9008, Santa Cruz Biotechnology, or 4595, Cell Signaling), or the ubiquitously expressed GTPase RAN (BD Biosciences), a secondary horseradish-conjugated antibody (Thermo Scientific) was added, and an enhanced chemiluminescent substrate kit (Thermo Scientific) was used for detection.
mRNA Isolation and Quantitative Polymerase Chain Reaction (qPCR)-RNA was purified with the RNeasy mini kit (Qiagen GmbH, Hilden, Germany). cDNA was generated using the Sprint PowerScript system (Clontech). All primers and probes are listed in supplemental Table S1. All PCRs were carried out using either Taqman Universal Polymerase Master Mix or SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) and the PRISM 7900 instrument (Applied Biosystems) and evaluated according to the standard curve method. All mRNA expression data were normalized to 36B4 gene expression.
Gene Expression Profiling and Analysis-Gene expression values were determined using Affymetrix mouse4302 Gene-Chip arrays (Affymetrix, Santa Clara, CA) by hybridizing trip-licate samples of preadipocytes and control or PPAR␥ siRNAtreated adipocytes (day 14 after initiation of differentiation). The University of Pennsylvania Microarray Core Facility performed cRNA labeling, GeneChip hybridization, and biotinylated cRNA detection according to standard Affymetrix protocols. Hybridized arrays were scanned using an Affymetrix GeneChip Scanner 3000. The image files were analyzed for probe intensities and converted to tabular formats using Microarray Suite Expression Analysis software from Affymetrix. Probe set intensity summarization and normalization were calculated using RMA (17) implemented using the Bioconductor project (18) in the software package R (19). Prior to statistical analysis for differential probe set expression, the data were filtered to exclude probe sets with low variability by selecting those probe sets whose expression values had an inner quartile range Ͼ0.5. To determine differential probe set expression between the treatment conditions, we used the linear modeling functions from the BioC limma package (20). Initially, a linear model was fit to a group-means parameterization design matrix defining each treatment condition. A contrast matrix defining all of the pairwise comparisons was subsequently fit, which utilized an empirical Bayes method to moderate the S.E. of the estimated log -fold changes as described (20). Controlling the false discovery rate was used to correct for multiple testing (21). Both processed and raw array data have been deposited with the Gene Expression Omnibus at NCBI under accession number GSE14004.
Heat Maps, Gene Ontology (GO) Analysis, and Association with PPAR␥ Binding-Heat maps were generated with the software package R using probe sets with a -fold change Ն7 and a q-value Ͻ0.01 (see Fig. 2A) or with a -fold change Ն2 of at least two conditions and limited to up-or down-regulation during both differentiation and knockdown (see Fig. 3A). Detection of overrepresented GO terms among genes presented in the heat maps was calculated using the hyperGTest function in R to calculate hypergeometric p values for GO term overrepresentation using a universe defined by the mouse4302 GeneChip. Overrepresentation was considered significant at p Ͻ 0.05. Genomic coordinates of genes regulated by Ͼ2-fold after PPAR␥ knockdown were assigned to a previously generated data set for whole genome PPAR␥ binding (14). Relative binding to up-or down-regulated genes was expressed in 200-bp bins relative to transcriptional start sites.
Adenoviral Overexpression of GATA2 in 3T3-L1 Cells-Adenoviral overexpression of GATA2 in 3T3 cells was carried out by using Adeno-X Expression System 2 (Clontech). In short, the coding sequence of Gata2 was cloned into the pDNR-CMV entry vector. After verification of the insert by sequencing, a recombination with the adenoviral acceptor vector pLP-Adeno-X-CMV was performed. PacI-digested positive recombined vectors were used to infect 293A cells. Produced adenoviruses were purified using the Adeno-X virus purification kit (Clontech), and titers were determined using the Adeno-X rapid titer kit (Clontech). Equal titers of Gata2-or green fluorescent protein (GFP)-containing adenoviruses were used to infect preadipocytes. Mature adipocytes were infected by the addition of 0.5 g/ml polylysine (Sigma) to the virus-containing solution. Adipocytes were incubated with the virus overnight in 0.5% bovine serum albumin/Dulbecco's modified Eagle's medium. Overexpression of GATA2 was confirmed by qPCR and protein expression and compared with the GFP control.
Measurement of Lipolysis-Lipolysis was assessed by measuring glycerol release into the culture medium. Adipocytes were serum-starved for 2 h in Dulbecco's modified Eagle's medium containing 0.2% bovine serum albumin, washed with Krebs-Ringer phosphate buffer, and incubated for 45 min in Krebs-Ringer phosphate buffer containing 4% bovine serum albumin and 1 M isoproterenol. Glycerol concentrations were determined by a glycerol reagent (Sigma) following the manufacturer's instructions and correlated to total protein concentration.
Mouse Studies-Experimental procedures were in accordance with institutional guidelines and regulations. Male C57BL/6J wild-type mice were fed a high fat diet (60 kcal % fat diet) from Research Diets for 20 weeks starting at 6 weeks of age. After high fat chow loading, body weight-matched mice were anesthetized using isoflurane prior to dissection of the skin and body wall. The adenoviral preparation (ϳ0.5 ϫ 10 10 plaque-forming units in a volume of 100 l) was injected into three points each on both epididymal fat pads of five (GFP) and six (GATA2) mice per group. 2 weeks later, animals were sacrificed, and total epididymal fat pads were isolated and processed for protein and mRNA.
Statistical Analysis-For 3T3-L1 gene expression experiments, representative results of at least three independent experiments are shown. Results are expressed as mean Ϯ S.D. of triplicates. Murine expression data are shown as mean Ϯ S.E. Statistical significance was determined using either the twotailed Student's t test or analysis of variance, as appropriate, and p Ͻ 0.05 was deemed significant (*, p Ͻ 0.05).

PPAR␥ Depletion Minimally Affects Appearance of Mature
Adipocytes-Differentiating day 4 adipocytes, which already express PPAR␥ at significant levels and are close to acquiring the typical adipocyte morphology, were electroporated with four different siRNA oligonucleotides (supplemental Table S1), and PPAR␥2 expression and adipocyte conversion were compared 48 h after reseeding the cells. All oligonucleotides efficiently reduced PPAR␥2 mRNA expression and blocked further adipocyte conversion compared with the control siRNA oligonucleotide as assessed by phase-contrast microscopy, and the most efficient oligonucleotide was used for all subsequent knockdown experiments (supplemental Fig. S1A). We next compared the impact of PPAR␥ knockdown at different stages of differentiation. 3T3-L1 adipocytes are considered as terminally differentiated 6 -8 days after the addition of hormonal inducers. As shown in Fig. 1A, PPAR␥ knockdown inhibited adipocyte conversion only during the early stages (day 4) of differentiation when investigated 48 h after electroporation by phase-contrast microscopy (Fig. 1A). However, PPAR␥ seems less important for the maintenance of mature adipocytes because no apparent physical changes were observed in adipocytes electroporated at day 6 or 14 (Fig. 1A), which is in agreement with the recent results of Liao et al. (16). Furthermore, even after an extended time of PPAR␥ knockdown (electroporated at day 10 and grown for an additional 10 days), there was no major change in adipocyte appearance (data not shown).
PPAR␥ Depletion in Mature Adipocytes Partially Reverses the Majority of Adipogenic Gene Expression Changes-To explore why PPAR␥ depletion does not reverse adipogenesis, we compared global gene expression of preadipocytes with control and PPAR␥ siRNA-treated adipocytes. To avoid interfering with the differentiation process, mature day 14 adipocytes were studied. 48 h after electroporation, PPAR␥ protein levels of adipocytes treated with the corresponding siRNA oligonucleotide were comparable with those in preadipocytes (supplemental Fig. S1B). Gene expression in PPAR␥ knockdown, control siRNA-treated adipocytes, and preadipocytes was analyzed by Affymetrix expression microarrays. 759 genes were found to be changed by Ͼ7-fold during differentiation and PPAR␥ knockdown. qPCR was used to validate these changes in several genes (supplemental Fig. S1C). Heat map analysis of the differentially regulated genes showed that there were two main categories of reciprocal regulation; genes up-regulated during differentiation were down-regulated by PPAR␥ knockdown, and conversely, genes downregulated during differentiation were up-regulated after depletion of PPAR␥ ( Fig. 2A). Interestingly, in both groups, PPAR␥ knockdown led to only a partial rescue of preadipocyte gene expression (compare color spectra in Fig. 2A). In accordance with this finding, qPCR analysis of adipocytespecific aP2 and preadipocyte-specific Mmp14 showed a higher degree of regulation during differentiation than after PPAR␥ knockdown (Fig. 2B).
Adipocyte Genes Down-regulated by PPAR␥ Knockdown Regulate Lipid Metabolism-Consistent with the known role of PPAR␥ in lipid metabolism, genes up-regulated during differentiation and down-regulated by PPAR␥ knockdown were highly enriched for lipid metabolic pathways as determined by GO analysis (Fig. 2C, top). Most adipocyte-specific genes can be found in this group, including several genes known to contain functional PPAR response elements (supplemental Table S2). Indeed, the genome-wide analysis of PPAR␥ binding in adipocytes (14) demonstrated an enrichment of PPAR␥-binding sites in genes down-regulated after PPAR␥ knockdown, especially those close to the transcriptional start site (Fig. 2D), suggesting that these genes are direct targets of PPAR␥. By contrast, GO analysis of the genes down-regulated during adipogenesis and up-regulated in mature adipocytes depleted of PPAR␥ demonstrated enrichment for genes important for fibroblast growth (Fig. 2C, bottom). These genes were not enriched for PPAR␥ binding in the genome-wide analysis (Fig. 2D), suggesting that although their reduced adipocyte expression is dependent on PPAR␥, this regulation is indirect.
Subgroup of Genes Is Resistant to Reversal after PPAR␥ Knockdown in Mature Adipocytes-Because the findings thus far indicate that PPAR␥ regulates numerous adipocyte genes involved in lipid metabolism, the explanation for a lack of altered adipocyte appearance after PPAR␥ depletion remained unclear. We hypothesized that the residual expression of adipocyte-specific genes is above a certain threshold necessary for the maintenance of the differentiated state. The reason for that could be an incomplete depletion of PPAR␥ or the existence of additional factors that are not regulated by PPAR␥ contributing to gene expression and maintenance of the differentiated state of mature adipocytes.
To address this, we focused on genes whose expression changed in a manner that was not reversed by PPAR␥ knockdown. By applying the 7-fold cutoff used earlier, we identified only a small number of genes. To increase this number, we investigated genes that changed by Ͼ2-fold. We found 1,315 genes exhibiting these characteristics, along with a small subset of genes that were not regulated during differentiation but altered after PPAR␥ knockdown (Fig. 3A). GO analysis of all upor down-regulated genes that were not reversed after PPAR␥ knockdown showed enrichment for functional pathways not related to the specific functions of the mature adipocyte (Fig.  3B). Genes up-regulated during differentiation but not downregulated by PPAR knockdown exhibited modest enrichment for lipid metabolic processes. In contrast, genes down-regulated during differentiation but not up-regulated by PPAR knockdown were enriched for cell cycle functions (Fig. 3B). This suggested that the expression of some genes involved in lipid metabolism and cell cycle regulation is not reversed after PPAR␥ knockdown and therefore could be responsible for the maintained adipocyte morphology.
Special Role for GATA2 among Upstream Factors after PPAR␥ Depletion-We next ordered regulated genes not reversed after PPAR␥ knockdown according to their -fold change during differentiation (see top 10 lists in supplemental Table S2). High on this list was GATA2, a transcription factor important for various types of cell differentiation (22). In adipocytes, GATA2 inhibits differentiation and is downregulated early after hormonal stimulation of preadipocytes (10). GATA2 has been described to inhibit adipocyte differentiation just upstream of PPAR␥ by direct binding to the PPAR␥ promoter and by interfering with C/EBP␣-and C/EBP␤-mediated transcription (23). We therefore investigated whether PPAR␥ knockdown only affects genes that are regulated downstream of PPAR␥ during adipocyte differentiation. Surprisingly, transcription factors implicated in the differentiation process upstream of PPAR␥ activation were increased after PPAR␥ knockdown, with the exception of GATA2 (Fig. 3C). qPCR confirmed that Gata2 gene expression was reduced in adipocytes whether or not PPAR␥ was depleted (Fig. 3D); likewise, GATA2 protein expression was also decreased during adipocyte differentiation and did not increase after PPAR␥ depletion (Fig. 3E). We therefore hypothesized that continued low Gata2 expression could contribute to the maintenance of residual adipocyte-specific gene expression and adipocyte morphology after depletion of PPAR␥.
Forced GATA2 Expression in Mature Adipocytes Complements PPAR␥ Depletion-3T3-L1 preadipocytes infected with a GATA2-expressing adenovirus and grown to confluency (supplemental Fig. S2A, left panel) showed markedly decreased adipocyte conversion in comparison with the GFP control, confirming that GATA2 down-regulation is required for adipogenesis (supplemental Fig. S2A, right panel). Adenoviral gene transduction was next used to restore Gata2 expression in mature adipocytes, with substantial expression of GFP and GATA2 noted after 48 h (Fig. 4A) and maintained for at least 7 days (data not shown). Forced expression of GATA2 in control siRNA-treated adipocytes led to a reduction of PPAR␥ and its target gene aP2 (Fig. 4B). Interestingly, forced expression of GATA2 also decreased the expression of Ces3 (carboxylesterase 3), a major lipase in mature 3T3-L1 adipocytes (24 -26) that, like Gata2, was high on the list of regulated genes not reversed after PPAR␥ knockdown (supplemental Table S2). Combining GATA2 expression with PPAR␥ depletion resulted in an even stronger reduction of aP2 (Fig. 4B). Furthermore, simultaneous FIGURE 2. PPAR␥ depletion in differentiated adipocytes reverses most adipogenic gene expression. A, heat map of genes with differential expression in preadipocytes (Preads) and adipocytes electroporated with control or PPAR␥ siRNA oligonucleotides (genes with a -fold change Ն7 between all conditions) determined by microarray. B, representative expression of adipocyte-and preadipocyte-specific genes measured by qPCR. C, GO analysis of these genes divided into two groups according to indicated patterns of regulation. D, localization of PPAR␥ binding to up-and down-regulated genes after PPAR␥ depletion in relation to distance to their transcriptional start sites (TSS; Pos. 0). Diff, differentiation; Kd, knockdown.
PPAR␥ knockdown and GATA2 expression changed adipocyte morphology more than either manipulation alone (Fig. 4C). In addition, the adipocyte-specific function of isoproterenol-stimulated lipolysis was reduced in cells with combined PPAR␥ knockdown and GATA2 expression but not with either treatment alone (Fig. 5).
Overexpression of GATA2 in White Adipose Tissue in Vivo Reduces Adipocyte-specific Gene Expression-Gata2 expression is reduced in several rodent models of obesity (10). Direct injection of adenovirus into the epididymal fat pad led to selective expression in 50 -70% of the whole fat pad, as shown for GFP (supplemental Fig. S2B) for up to 3 weeks (data not shown). Isolated overexpression of GATA2 in epididymal white adipose tissue using this method was confirmed by immunoblot analysis (Fig. 6A). Remarkably, ectopic expression of GATA2 in adult adipose tissue in vivo revealed that GATA2 exhibited markedly reduced mRNA expression of PPAR␥2, aP2, as well as Ces3 (Fig. 6B).

DISCUSSION
We have found that depletion of adipocyte PPAR␥ to levels comparable with undifferentiated cells leads to a partial reversal of the vast majority of gene expression changes occurring during adipogenesis. In agreement with a previous study (16), this did not result in a dedifferentiated appearance of the cells. To explain this finding, we focused on genes that are highly regulated during differentiation but not reversed after PPAR␥ knockdown. The antiadipogenic transcription factor Gata2 ranked high among these genes, and its forced expression in mature adipocytes and adipose tissue in vivo resulted in the down-regulation of adipocyte-specific gene expression and partial dedifferentiation of the cells. Thus, lack of Gata2 expression appears to be a factor that contributes to maintenance of the adipocytes independently of PPAR␥.
A large number of the genes whose high level of expression in adipocytes was reduced by depletion of PPAR␥ play a role in the cellular lipid and glucose metabolism, which are key specialized properties of adipocytes. Many of these genes have nearby binding sites for PPAR␥ as determined by genomewide analysis of PPAR␥ binding in adipocytes (13,14). Thus, these genes are likely to be direct targets of PPAR␥. We also identified a large number of genes whose expression was reduced during adipocyte differentiation and up-regulated by PPAR␥ knockdown. The fact that the expression of these genes is induced by PPAR␥ knockdown implies a role for PPAR␥ in their regulation. However, unlike the genes whose high level of expression depended upon PPAR␥, we found far fewer PPAR␥binding sites in the vicinity of these genes, consistent with what has been observed for the entire set of genes that are downregulated during the more global analysis of genes during adipocyte differentiation (14). Thus, although the repressed expression of many of these genes depends upon the presence of PPAR␥, this appears to be an indirect effect.
Although most metabolic genes induced during adipogenesis were dependent upon PPAR␥ for expression in adipocytes, a notable exception was the adipocyte lipase Ces3. Ces3 is involved in maintaining the metabolic properties of adipocytes but, like Gata2, was unaffected by PPAR␥ knockdown. Forced expression of GATA2 in mature adipocytes repressed Ces3, suggesting that Ces3 is likely a GATA2 target whose regulation is independent of PPAR␥. Consistent with this, recent genomewide analysis of PPAR␥ binding in adipocytes did not detect PPAR␥ binding within 65 kb of the transcriptional start siteW. In addition, a recent study showed that thiazolidinedione treatment does not regulate Ces3 expression in adipocytes (24), providing further evidence of a PPAR␥-independent mechanism. FIGURE 3. PPAR␥ depletion in differentiated adipocytes does not entirely reverse the adipogenic gene expression. A, heat map of genes with differential expression in preadipocytes (Preads) and adipocytes electroporated with control or PPAR␥ siRNA oligonucleotides (with a -fold change Ն2 with the same orientation between differentiation and knockdown or with a -fold change Ն2 between at least two conditions) determined by microarray. B, GO analysis of these genes divided into two groups according to indicated patterns of regulation. C, heat map of transcription factors expressed in preadipocytes and known to control adipocyte differentiation. D and E, mRNA and protein expression of Gata2 in preadipocytes (Pre), control, or PPAR␥ siRNA-treated adipocytes measured by qPCR and immunoblotting (IB).
The combination of PPAR␥ knockdown and GATA2 overexpression synergistically reversed the adipocyte-specific pattern of gene expression, and indeed, only the combined treatments markedly decreased phenotypic characteristics such as lipolytic capacity of mature adipocytes. Thus, in addition to high levels of PPAR␥, low Gata2 expression is critical for the maintenance of adipocyte characteristics. Forced expression of GATA2 repressed PPAR␥ mRNA, consistent with the notion that GATA2 is upstream of PPAR␥ (23), although GATA2 also functions as a PPAR␥-independent regulator of adipocyte genes, as exemplified by Ces3. Whether GATA2 represses Ces3 directly or via another factor remains to be elucidated. GATA2 has been shown to mediate the inhibitory effects of SFPI1 and NR2F2 (27, 28) on adipocyte differentiation, which could also be part of this pathway. However, it is important to note that even the combined knockdown of PPAR␥ and reintroduction of GATA2 did not completely dedifferentiate mature adipocytes. Although potentially due to the experimental limitations of our cell model, this finding may also point to the existence of additional factors critical for maintaining the molecular and metabolic properties of differentiated adipocytes.