Role of PPARγ in Regulating a Cascade Expression of Cyclin-dependent Kinase Inhibitors, p18(INK4c) and p21(Waf1/Cip1), during Adipogenesis*

Molecular mechanisms coupling growth arrest and cell differentiation were examined during adipogenesis. Data are presented that document a cascade expression of members of two independent families of cyclin-dependent kinase inhibitors that define distinct states of growth arrest during 3T3-L1 preadipocyte differentiation. Exit from the cell cycle into a pre-differentiation state of post-mitotic growth arrest was characterized by significant increases in p21 and p27. During onset of irreversible growth arrest associated with terminal differentiation, the level of p21 declined with a concomitant, dramatic increase in p18 and a sustained level of p27. The expression of p18 and p21, regulated at the level of protein and mRNA accumulation, was directly coupled to differentiation. Stable cell lines were engineered to express adipogenic transcription factors to examine the active role of trans-acting elements in regulating these cell cycle inhibitors. Ectopic expression of peroxisome proliferator-activated receptor (PPAR) γ in non-precursor fibroblastic cell lines resulted in conversion to adipocytes and a coordinated increase in p18 and p21 mRNA and protein expression in a PPARγ ligand-associated manner. These data demonstrate a role for PPARγ in mediating the differentiation-dependent cascade expression of cyclin-dependent kinase inhibitors, thereby providing a molecular mechanism coupling growth arrest and adipocyte differentiation.

Adipocytes of white adipose tissue, as well as myocytes from heart and skeletal muscle, represent examples of terminal differentiation whereby the expression of a specialized phenotype is marked by cessation of cell proliferation and the accumulation of cells in the G 1 phase of the cell cycle. In mammalian cells, phase transition is regulated by the phosphorylated states of various substrates including the retinoblastoma family proteins which mediate S phase progression (1). These substrates are phosphorylated by a dimer complex comprising a regulatory "cyclin" subunit and a catalytic cyclin-dependent kinase (cdk). 1 Phosphorylating activity of cyclin/cdk complexes is further modulated by cyclin-dependent kinase inhibitors (CKIs), which are grouped into two distinct families based on sequence homology and targets of inhibition (2). To date, seven CKIs have been identified, including p15 INK4b , p16 INK4a , p18 INK4c , and p19 INK4d defining the INK4 family, and p21 Cip1 , p27 Kip1 , and p57 Kip2 , representing the CIP/KIP family. Recent reports have demonstrated that CKI expression is up-regulated during cell differentiation in vitro and in vivo (3,4), suggesting that these cell cycle inhibitors may play a universal role in exit from the cell cycle and/or maintenance of the irreversible growth arrest which defines terminal differentiation.
Adipocyte differentiation is largely controlled by two families of transcription factors: the CCAAT/enhancer-binding proteins (C/EBPs) and peroxisome proliferator-activated receptors (PPARs) (5)(6)(7). Members of the C/EBP family (C/EBP␣, C/EBP␤, and CEBP␦) form heterodimers and homodimers via a leucine zipper motif with dimers binding to regulatory elements within target genes via basic DNA binding domains. Ectopic expression of various C/EBPs has been shown to convert non-precursor fibroblastic cell lines into fully differentiated adipocytes (8 -10), whereas genetic knockouts in vitro and in vivo block adipocyte differentiation (11)(12)(13). The PPARs (PPAR␣, PPAR␦, and PPAR␥) define a family of ligand-activated nuclear hormone receptors that heterodimerize with the retinoid X receptor and bind to specific peroxisome proliferatorresponsive elements located within the promoters of target genes. Through utilization of different start sites and alternate splicing, the PPAR␥ gene gives rise to two isoforms, ␥1 and ␥2. Tissue distribution of PPAR␥2 is highly enriched in adipose tissue, and ectopic expression in various non-precursor cell lines also gives rise to adipocyte differentiation (14). Although the natural ligand for PPAR␥ is still under investigation, a synthetic class of specific ligands, called thiazolidinediones (TZDs), greatly enhance transcriptional activity (15).
Current models of the molecular process of adipogenesis involve a cascade expression of C/EBP␤ and C/EBP␦, followed by the expression of C/EBP␣ and PPAR␥, which precede and regulate the expression of many genes representative of the mature adipocyte. Much of our understanding of the interplay between these and other trans-acting elements that regulate adipocyte differentiation has been made possible with the establishment of preadipocyte cell lines (e.g. 3T3-L1 and 3T3-F442A) that differentiate from proliferative, fibroblastic-like cells into mature adipocytes exhibiting nearly identical morphological and biochemical properties of white adipose tissue (16). After reaching a state of density-induced growth arrest, preadipocyte cell lines can be induced to differentiate with exposure to a combination of mitogen and hormonal agents.
Immediately after exposure to these agents, the cells reenter the cell cycle for a limited period of cell proliferation, commonly referred to as clonal expansion. This is followed by the establishment of a unique state of post-mitotic growth arrest, referred to as G D , which has been reported to be permissive for subsequent differentiation (17). With the onset of differentiation, the cells enter a third state of growth arrest that is irreversible, where they are then considered to be terminally differentiated. Although numerous reports have recently emerged concerning the transcriptional regulation of adipocyte specific gene expression, little is known concerning the molecular events involved in the progression of clonal expansion and the establishment of distinct states of growth arrest that mark the progression toward terminal differentiation.
In this investigation, we demonstrate that induction of differentiation of 3T3-L1 preadipocytes results in gene expression representing classic cell cycle progression that switches to adipogenic gene expression concomitant with exit from the cell cycle. In addition, the data presented here document a cascade expression of members of two independent families of CKIs that define distinct states of growth arrest associated with adipogenesis. Moreover, the expression of p18 and p21 is shown to be regulated during the conversion of non-precursor fibroblasts into adipocytes by ectopic expression of the adipogenic transcription factor, PPAR␥, providing a molecular mechanism coupling growth arrest and adipocyte differentiation.

EXPERIMENTAL PROCEDURES
Stable Cell Lines-The NIH-3T3 cell line ectopically expressing C/EBP␤ and C/EBP␦ under control of a tetracycline operator was created and described previously (18). Stable cell lines expressing PPAR␥ were derived by retroviral infection as described previously (19). Briefly, packaging cells (BOSC23) were grown in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum (FBS). At approximately 80% confluence, the cells were transiently transfected with pBabe-derived PPAR␥2 expression vector by calcium phosphate coprecipitation with chloroquine as described (20). Viral supernatants were collected 48 h after transfection, filtered, and applied with 4 g/ml hexadimethrine bromide to proliferating Swiss and Balb/c fibroblasts for 24 -36 h. Medium was then changed to DMEM containing 10% calf serum. After 48 h, cells were passaged if necessary and exposed to 2 g/ml puromycin for selection. Resistant cells were propagated in puromycin until experimentation.
Cell Culture and Induction of Differentiation-Murine 3T3-L1 preadipocytes and fibroblast cell lines ectopically expressing adipogenic transcription factors were induced to differentiate into adipocytes as described previously (21). Briefly, cells were propagated in DMEM containing 10% calf serum (growth medium). At 2 days postconfluence, the medium was changed to DMEM containing 10% FBS supplemented with 0.5 mM 3-isobutyl-1-methylxanthine, 1 M dexamethasone, and 1.7 M insulin (MDI; differentiation medium). After 48 h, cells were maintained in DMEM containing 10% FBS and 0.4 M insulin throughout the remaining time course of experimentation. Maintenance medium was changed every 48 h until the cells were utilized for experimentation. Throughout the study, "time 0" refers to postconfluent cells immediately before chemical induction of differentiation with the addition of MDI to the culture medium. The term "post-MDI" refers to the time elapsed since the addition of MDI to the culture medium.
RNA Analysis-Total RNA was extracted from fibroblast cell lines with Trizol (Life Technologies, Inc.) according to manufacturer's instructions with modifications. Briefly, cultured cells were washed in ice-cold phosphate-buffered saline, lysed with Trizol reagent, passed through a 21-gauge needle, and gently mixed (5:1) with chloroform. Following centrifugation, the aqueous phase was mixed with an equal volume of isopropyl alcohol and centrifuged. The resulting pellet was dissolved in RNase-free water, mixed with an equal volume of chloroform/butanol (4:1), vortexed vigorously for 15 s, and centrifuged. The aqueous phase was collected and RNA precipitated with sodium acetate/ethanol. Following quantitation, 20 g of total RNA was denatured in formamide and electrophoresed through formaldehyde/agarose gels. The RNA was blotted to Hybond-N nylon (Amersham Pharmacia Biotech), cross-linked, hybridized, and washed. Probes were labeled by random priming using the Klenow fragment of DNA polymerase I (New England Biolabs Inc., Beverly, MA) and [␣-32 P]dCTP (NEN Life Science Products). Hybridization to the ribosomal 18 S subunit was used to quantitate equal loading.
Protein Analysis-Preparation and fractionation of isolated adipocytes from rat fat pads was performed as described previously (22). Cultured cells were washed with phosphate-buffered saline, lysed in Tris/SDS buffer containing Nonidet P-40 and protease inhibitors, vortexed, and centrifuged. Protein content of the supernatant was determined using a BCA kit (Pierce) according to manufacturer's instructions. Following quantitation, proteins were separated by electrophoresis through SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membrane (Bio-Rad). Following transfer, membranes were blocked with milk and probed with the following primary antibodies: p21 and proliferating cell nuclear antigen (Oncogene); p27 and cyclin D1 (Transduction Laboratories); p18, C/EBP␣, and PPAR␥ (Santa Cruz); and Glut4 (23). Results were visualized with horseradish peroxidase-conjugated secondary antibodies (Sigma) and enhanced chemiluminescence (Pierce) according to manufacturer's instructions.

Molecular Events Demonstrating a Switch between Adipocyte
Growth and Differentiation-Cultured 3T3-L1 preadipocytes induced to differentiate are documented to undergo an early phase of clonal expansion, which precedes the acquisition of a fat-laden phenotype. Heretofore, few investigations have explored potential molecular mechanisms that may play a role in exit from clonal expansion and/or maintenance of growth arrest associated with terminal differentiation. To assess the switch between growth and differentiation, we initially characterized changes in gene expression during cell cycle progression that follows the induction of differentiation. Cells were cultured to 2 days post-confluence and induced to differentiate as described under "Experimental Procedures." Total RNA was collected every 2 h for 30 h following a change from growth to differentiation medium and subjected to Northern analysis. As shown in Fig. 1A, 2-day post-confluent cells not exposed to chemical inducers (0 h) had entered a state of density-induced growth arrest, as indicated by the comparison of histone and cyclin gene expression to subconfluent, proliferating preadipocytes (PPA). Switching to differentiation medium consisting of DMEM supplement with 10% FBS and MDI resulted in cell cycle progression with sequential activation of ornithine decarboxylase (early G 1 ), cyclin D1 (mid G 1 ), cyclin E (late G 1 ), cyclin A (late G 1 /S), histone H2B (S phase), and cyclin B (G 2 /M) gene expression. The rapidity of early gene activation and the peak of histone expression, estimated at 18 -20 h, suggests that reentry of these density-arrested preadipocytes into the cell cycle occurred immediately following the change to differentiation medium. The kinetics of cyclin gene expression presented here are consistent with reported changes in Rb phosphorylation (24) and E2F-binding complexes (25) determined for differentiating 3T3-L1 preadipocytes. Based on the additional observation of immediate early (c-Myc), delayed early (ornithine decarboxylase), and S-phase (histone H2B) gene activation reported here and elsewhere (26,27), it appears that chemical induction of differentiation of these preadipocytes resulted in synchronous activation of cell cycle gene expression that begins in the very early phases of G 1 , possibly G 0 , and continues through to cell division. Fig. 1A also compares the kinetics of clonal expansion with adipogenic transcription factor gene expression. Chemical induction of differentiation resulted in immediate activation of C/EBP␤ and C/EBP␦, a process that has been shown to occur independent of protein synthesis and afforded by isobutylmethylxanthine and dexamethasone, respectively (28). Although it is still uncertain what role, if any, that C/EBP␤ and C/EBP␦ may play in the activation of post-confluent cell cycle progression, it is clear that cell proliferation kinetically preceded the expression of C/EBP␣ and PPAR␥. The virtual absence of cyclin and histone gene expression during late stages of differentiation (6 day; Fig. 1A) indicates that preadipocytes ceased proliferating at some point during the process of adipogenesis. To assess when the switch between growth and differentiation occurred, RNA was collected every 24 h for 9 days following the change to differentiation medium and subjected to Northern analysis. Data presented in Fig. 1B clearly illustrate the mutually exclusive nature of growth and differentiation with a clear switch in gene expression associated with these independent processes occurring approximately 3 days following induction of differentiation. Of particular interest, the onset of C/EBP␣ and PPAR␥ coincided with the switch in gene expression, suggesting that these transcription factors may play a role in coupling growth arrest and adipocyte differentiation. It is also important to note that the same chemical agents responsible for the induction of differentiation were also responsible for activation of clonal expansion. Thus, the decision to switch between growth and differentiation pathways, although continually in the presence of abundant mitogens, is made at the cellular level and not by the investigator. This is in contrast to other differentiating systems (e.g. skeletal muscle), where induction of differentiation typically requires technical manipulations necessary to ensure a prerequisite state of growth arrest. Considering this and the synchrony of clonal expansion, we propose, as have others (29), that differentiation of preadipocyte cell lines provides an excellent model for mechanistic studies concerning the coupling of growth arrest and cell differentiation.

A Cascade Expression of Cyclin-dependent Kinase Inhibitors That Define Distinct States of Growth Arrest during Adipocyte
Differentiation-It is well accepted that cell cycle progression is controlled by cyclin/cdk protein kinases where phosphorylating activity can be modulated by functionally and structurally distinct CKIs. To assess the involvement of CKIs in coupling of growth arrest and adipocyte differentiation, the gene expression of the seven known members of the INK4 and CIP/KIP families of CKIs was examined by Northern analysis, where it was determined that terminal differentiation was marked only by elevated levels p18 and p21 mRNA (data not shown). Based on this screen, p18 and p21 were further examined at the level of protein expression during the time course that entailed the switch between growth and differentiation. Albeit modestly regulated at the level of gene expression, p27 was also examined due to numerous reports linking this CKI to density arrest and the well documented post-transcriptional regulation of its protein expression (30,31). To carefully evaluate the kinetics of CKI expression during exit from clonal expansion and the onset of irreversible growth arrest, whole cell lysate proteins were harvested following a change to differentiation medium and subjected to Western analysis during two independent time courses. The first examined protein expression every 24 h for 6 days ( Fig. 2A) and the second every 6 h during the first 48 h and every 12 h thereafter for 96 h (Fig. 2B). For comparison, total RNA was isolated from the same experiment depicted in Fig.   FIG. 1. Pattern of gene expression that demonstrates the switch between 3T3-L1 preadipocyte growth and differentiation. A, cell cycle gene expression following induction of differentiation. Cultured preadipocytes were differentiated as described under "Experimental Procedures." Total RNA was collected every 2 h for 30 h following chemical induction of differentiation, and 20 g of RNA was examined by Northern analysis. RNA from subconfluent PPA and fully differentiated adipocytes (6 day) was included for reference to proliferation and differentiation, respectively. Phase change denoting cell cycle progression following chemical induction of differentiation was approximated and diagrammed above the illustrated data. B, switch between growth and differentiation gene expression. Total RNA was collected every 24 h for 9 days following chemical induction of differentiation, and 20 g of RNA was examined by Northern analysis. The switch between growth and differentiation was approximated and diagrammed above the illustrated data.
2B and subjected to Northern analysis for histone H2B mRNA expression as a marker of S-phase progression. Protein expression of C/EBP␣ and PPAR␥ was also examined to document the early onset of differentiation. As shown in Fig. 2, p27 protein was abundantly expressed in density-arrested preadipocytes (0 day), decreased transiently during the first 48 h of differentiation, returned to predifferentiation levels by day 3, and remained elevated throughout the course of differentiation. Interestingly, the transient increase in histone mRNA and the transient decrease in p27 protein correlated in a direct reciprocal fashion as cells entered and exited the S phase of clonal expansion. In direct contrast, the protein expression of p21 was significantly elevated in proliferating (PPA) but not densityarrested preadipocytes (day 0). Following a change to differentiation medium, p21 transiently increased during early stages of G 1 , decreased during S phase progression, increased again to abundant levels concomitant with the expression of C/EBP␣ and PPAR␥, and then decayed during later stages of differentiation (day 6). Although the first peak coincided with early (G 1 phase) cell cycle progression, the second peak of p21 protein accumulation (72-96 h) clearly occurred as cells exited the cell cycle as indicated by a significant decrease in the number of cells entering S phase (i.e. decreased histone expression). Interestingly, the protein expression of proliferating cell nuclear antigen, which began to accumulate during early G 1 , was maintained during this period, suggesting that the cells had entered a transient state of growth arrest that was unique from that observed during density arrest (day 0) or terminal differentiation (day 6). The expression of p18 appeared to kinetically succeed exit from the cell cycle and the early onset of adipocyte gene expression with significant protein levels accumulating only during later stages of terminal differentiation. Unlike p21 and p27, p18 protein was not significantly expressed in proliferating or density-arrested preadipocytes.
Numerous reports have indicated that CKI expression, in particular p21, can be directly modulated by various mitogens and hormonal agents. To confirm that the cascade regulation of CKI expression was linked to molecular processes of differentiation and not simply due to exposure to the chemical inducers, density-arrested preadipocytes were chemically induced to differentiate in the presence and absence of tumor necrosis factor ␣ (TNF␣). This cytokine has been shown to completely block the development of the fat-laden phenotype and associated gene expression when applied to cells during the induction of preadipocyte differentiation. Whole cell lysates were harvested on days 3 and 6 following chemical induction in the presence and absence of TNF␣ and subjected to Western analysis for CKI expression. Protein expression of Glut4 was also examined to confirm the state of adipocyte differentiation. As shown in Fig. 3A, the increase in p21 and p18 protein expression, observed on days 3 and 6, respectively, was prevented in the presence of TNF␣, suggesting that the expression of these CKIs was dependent upon adipocyte differentiation and not due to secondary effects of the mitogen and hormonal agents necessary to induce differentiation. This conclusion was further supported by the lack of effect of TNF␣ on p27 protein, suggesting that the inhibitory effect on differentiation was not due to toxicity of this potent cytokine.
To further support the pattern of CKI expression following terminal differentiation, protein expression was examined in native adipocytes isolated and purified from rat epididymal fats pads as described under "Experimental Procedures." The rat adipocytes were fractionated into cytosolic, nuclear, and whole cell lysates to ensure that protein expression in native adipocytes was not localized and therefore diluted by compartmental protein normalization. For comparison, whole cell lysates were isolated from 3T3-L1 cultured adipocytes at 0, 3, and 6 days of differentiation and subjected to Western analysis. As shown in Fig. 3B, both p18 and p27 proteins were expressed in rat whole cell lysates (W) to a value equivalent to that observed for terminally differentiated 3T3-L1 adipocytes (day 6). Similarly, p21 protein was not observed to any significant amount in fully differentiated 3T3-L1 adipocytes (day 6) or in any measured fraction of differentiated rat adipocytes.
A Role for Adipogenic Transcription Factors in Regulating CKI Expression-Considerable evidence implicates C/EBPs and PPAR␥ as major transcription factors responsible for development of the mature adipocyte. The next objective was to determine if these adipogenic transcription factors play a role in regulating the cascade expression of CKIs during adipogenesis. Initially, the mRNA accumulation of p18, p21, and p27 was kinetically compared with the expression of C/EBP␣ and PPAR␥ over the time course of 3T3-L1 preadipocyte differentiation. Total RNA was harvested every 24 h for 6 days following chemical induction and subjected to Northern analysis. As shown in Fig. 4, the mRNA expression of p27, highest in density-arrested preadipocytes (0 day), declined following chemical FIG. 2. Cascade expression of p18, p21, and p27 proteins during 3T3-L1 preadipocyte differentiation. A, cultured preadipocytes were differentiated as described under "Experimental Procedures." Whole cell lysates were collected every 24 h for 6 days following chemical induction of differentiation. One hundred g of protein was examined by Western analysis. Protein from subconfluent PPA was also included. B, whole cell lysates were harvested every 6 h during the first 48 h and every 12 h thereafter for 96 h, following a change to differentiation medium. Fifty g of protein was examined by Western analysis. The 42-kDa band of C/EBP␣ and the 48-and 50-kDa bands representing PPAR␥1 and PPAR␥2, respectively, were illustrated. For reference to S-phase progression, 20 g of total RNA, collected from the same experiment, was examined by Northern analysis for histone mRNA expression (*).
induction and remained at low levels throughout the course of differentiation. Although the precipitous decline in protein that immediately precedes S phase of clonal expansion was accompanied by a moderate fall in mRNA accumulation (compare Figs. 2 and 4), the return of p27 protein to predifferentiation levels following cell cycle progression appears to be predominantly independent of gene expression and likely to occur via post-transcriptionally controlled pathways. Steady state levels of p21 mRNA gradually increased kinetically with protein expression (compare Figs. 2 and 4) as cells entered the state of post-mitotic growth arrest following clonal expansion. Interestingly, the onset of terminal differentiation was marked by a decrease in p21 protein even though the mRNA remained elevated, suggesting both transcriptional and post-transcriptional processes were likely to be involved in p21 regulation. Northern blot analysis of p18 demonstrated a 2.4-and 1.2-kb transcript that began to accumulate coordinately with protein expression on day 3 and remained elevated throughout differentiation. Although the kinetics of both p18 mRNAs were similar, the onset of terminal differentiation was marked by dramatic changes in the 1.2-kb transcript. Consistent with this observation, others have reported two p18 mRNAs where changes occurred predominantly in the 1.2-kb transcript during cell differentiation (32,33). Collectively, these data indicate that differentiation-dependent increases in p18 and p21 protein expression were kinetically accompanied by coordinate changes in mRNA accumulation. Of particular interest, the accumulation of p18 and p21 mRNAs coincided with or succeeded the onset of C/EBP␣ and PPAR␥ gene expression.
To investigate the contribution of adipogenic transcription factors in the regulation of CKI gene expression, we utilized a previously described NIH3T3 fibroblastic cell line engineered to ectopically express C/EBP␤ and C/EBP␦ under the control of a tetracycline-responsive inducible expression system (18). These cells, designated "␤␦ cells," were propagated in the presence of tetracycline, which has been shown to repress the ectopic expression of both C/EBPs. Tetracycline was removed from the growth medium at near confluence, and at approximately 2 days post-confluence (day 0), the growth medium was replaced with differentiation medium supplemented with the TZD, ciglitazone. With the exception of the TZD supplement, conditions were maintained identical to those used for 3T3-L1 differentiation described under "Experimental Procedures." Total RNA was collected at 0 and 6 days of differentiation and subjected to Northern analysis. As depicted in Fig. 5A, ectopic expression of C/EBP␤ and C/EBP␦, in the presence of MDI and TZD, led to the expression of PPAR␥ and adipocyte-specific genes (e.g. adipsin). Concomitant with the onset of adipocyte differentiation was the accumulation of p18 and p21 mRNAs to values equivalent to those observed in fully differentiated 3T3-L1 adipocytes. Interestingly, CKI mRNA accumulation occurred in the absence of C/EBP␣, which has been shown to be repressed in NIH3T3 fibroblasts (9,10). It is important to note, however, that the role of C/EBP␣ as an adipogenic transcription factor may, in part, be played by the ectopic expression of C/EBP␤ and/or C/EBP␦. To continue to dissect the molecular mechanism responsible for CKI mRNA accumulation, these engineered fibroblasts were cultured under various diagnostic conditions to alter gene expression and exposure to chemical inducers and TZDs. Total RNA was harvested following 6 days of conditional treatment and the results of Northern analysis depicted in Fig. 5B. Exposing the NIH-3T3 fibroblasts, not expressing ectopic C/EBP␤ and C/EBP␦ (i.e. in the presence of tetracycline), to chemical inducers and TZD (lane 1) resulted in a modest increase in p21 mRNA that was not observed in cells ectopically expressing the C/EBPs in the absence of differentiation mixture (lane 2). Although chemical induction of cells expressing the C/EBPs resulted in PPAR␥ and adipocyte-specific (e.g. adipsin)  3. Dependence of p18, p21, and p27 protein expression on adipocyte differentiation. A, 3T3-L1 preadipocytes were differentiated as described under "Experimental Procedures" in the presence and absence 500 pM TNF␣. Whole cell lysates were collected at 3 and 6 days following chemical induction of differentiation. One hundred g of protein was examined by Western analysis. B, protein from cytosolic (C), nuclear (N), and whole cell lysates (W) was prepared from differentiated adipocytes isolated from epididymal fat pads collected from 200-g male Sprague-Dawley rats as described under "Experimental Procedures." Whole cell lysates were also isolated from 3T3-L1 cultured adipocytes at 0, 3, and 6 days of differentiation. Fifty g of protein was examined by Western analysis. The 42-kDa band of C/EBP␣ was illustrated. gene expression (lane 3), significant p18 and p21 mRNA accumulation and acquisition of the adipocyte phenotype was observed only when cells where cultured under conditions leading to PPAR␥ gene expression in the presence of an exogenously supplied ligand specific for PPAR␥ activation (lane 4).
As accumulation of both p18 and p21 mRNAs correlated with ligand-activated PPAR␥ and not C/EBP expression, we continued to the explore the relationship between this adipogenic transcription factor and CKI gene expression by utilizing a retroviral system to produce fibroblastic cell lines ectopically expressing the ␥2 isoform of PPAR␥. The pBabe-Puro expression vector containing the cDNA for PPAR␥ used in this investigation (kindly provided by B. M. Spiegelman) was previously characterized for its efficacy in producing functionally active protein in NIH-3T3 fibroblasts (34). Parental (V) and PPAR␥ (P␥) containing vectors were packaged into viruses that were used to infect Swiss and Balb/c fibroblasts as described under "Experimental Procedures." Following puromycin selection, the resulting stable cell lines were grown to confluence and induced to differentiate with MDI in the presence of the TZD, troglitazone. Northern analysis of total RNA collected from Swiss (SP␥) and Balb/c (BP␥) fibroblasts ectopically expressing PPAR␥ at 0 and 6 days of differentiation is depicted in Fig. 6A. Total RNA from differentiated 3T3-L1 adipocytes was included for comparison. The larger ectopically expressed transcript (arrow) was easily resolved from the endogenous PPAR␥, as illustrated in lanes 1-4 (Fig. 6A). Following the standard differentiation protocol including TZD supplement, both fibroblastic cell lines ectopically expressing PPAR␥ significantly displayed gene expression (i.e. Glut4) and morphology (Ն90% cells containing lipid droplets; data not shown) indicative of the mature adipocyte. Interestingly, both cells lines demonstrated a significant increase in p18 and p21 mRNAs concomitant with the acquisition of the adipocyte phenotype. It should be noted that, in contrast to NIH3T3 cells, C/EBP␣ was not repressed in these fibroblasts and was regulated in a coordinate fashion with adipocyte differentiation.
Swiss and Balb/c fibroblasts are often considered preadipocytes inasmuch as some conversion can be noted when normal, untransfected cells are exposed to chemical inducers of differentiation. To determine the contribution of PPAR␥ above this adipogenic background, fibroblasts ectopically expressing the PPAR␥ construct (P␥) or the parental vector (V) were differentiated for 6 days in the presence and absence of troglitazone. Total RNA was collected and the Northern analysis is depicted in Fig. 6B. Troglitazone supplement to the differentiation medium applied to Swiss P␥ cells resulted in a differential increase in Glut4, p18, and p21 mRNAs (compare lanes 1 and 2) that was not observed with TZD supplement to cells expressing the parental vector (compare lanes 3 and 4). It appeared, however, that exposing vector cells to the differentiation protocol increased the background expression of both p18 and p21 independent of TZD treatment. This level of expression was possibly due to a direct effect of the chemicals utilized to induce differentiation and/or the 10 -20% adipocyte conversion of Swiss vector cells that occurred following the differentiation protocol (data not shown). In contrast to Swiss, the Balb/c fibroblasts expressing PPAR␥ converted to adipocytes, as marked by Glut4 expression, independent of TZD supplement (compare lanes 6 and 7). Consistently, the mRNA for both p18 and p21 was significantly and equivalently enhanced in Balb/c P␥ cells in direct correlation with Glut4 expression and adipocyte differentiation. The increase in mRNA was most pronounced when comparing P␥ cells supplemented with troglitazone (lanes 7) and vector cells differentiated in the absence of any exogenous PPAR␥ ligand (lane 8) where greater than a 30-fold induction was observed for both p18 and p21 mRNAs. These data support the notion that the increase in CKI mRNA was not simply due to exposure to chemical inducers or changes in cell density as both P␥ and vector cells were exposed to an identical differentiation protocol. The increase in p18 and p21 mRNAs with troglitazone treatment of Balb/c vector cells (compare lanes 8 and 9) could be attributed to activation of endogenous PPAR␥, which increased under these conditions. Both transcriptional and post-transcriptional mechanisms were indicated in the regulation of p21 and p27 during 3T3-L1 adipocyte differentiation. To determine if adipocyte conversion by PPAR␥ could also lead to CKI protein expression, Balb/c fibroblasts expressing PPAR␥ (BP␥) and the empty vector (BV) were induced to differentiate with MDI in the presence and absence of troglitazone, respectively. Whole cell lysate proteins were collected over the course of differentiation, and the results of Western analysis are depicted in Fig. 6C. Interestingly, both magnitude and kinetics of p18 and p21 protein expression were equivalent to that observed for 3T3-L1 adipocytes with p21 preceding p18 expression by approximately 24 h. The pattern of p27 expression was identical in BP␥ cells, which converted to greater than 90% adipocyte morphology, and BV cells, which remained as fibroblasts demonstrating the independence of p27 on adipocyte differentiation for protein expression. It was also observed that cyclin D1 protein increased dramatically in BP␥ cells, which expressed abundant levels of p18 and p21, suggesting that adipocyte differentiation resulted in the accumulation of cells in the G 1 phase of the cell cycle. Although the specific

FIG. 5. Inducible ectopic expression of C/EBP␤ and C/EBP␦ in NIH-3T3 fibroblasts induces adipogenesis and accumulation of p18 and p21 mRNA in a PPAR␥ ligand-dependent manner.
A, NIH-3T3 fibroblasts, ectopically expressing C/EBP␤ and C/EBP␦ in a tetracycline-repressive manner (␤␦ cells) were differentiated as described under "Experimental Procedures." Total RNA was collected at 0 and 6 days following chemical induction of differentiation, and 20 g of RNA was examined by Northern analysis. The 1.2-kb transcript of p18 was illustrated. For comparison, RNA from 3T3-L1 adipocytes (L1) differentiated for 6 days was included. B, ␤␦ cells were differentiated for 6 days under various conditions depicted below the illustrated data. The absence of ectopic C/EBP␤ and C/EBP␦ expression (␤/␦) was accomplished by supplement of tetracycline (1 g/ml) to the differentiation medium. Cells expressing ␤/␦ only were cultured for an identical length of time as other conditions but in the absence of chemical inducers (MDI) and TZD supplement (ciglitazone; 10 M). Twenty g of total RNA was examined by Northern analysis.
functions of these CKIs during adipogenesis has yet to be determined, it would not be unexpected to find that one of their roles is to inhibit cell proliferation prior to S phase transition during the onset of terminal differentiation. DISCUSSION This investigation presents a molecular mechanism coupling growth arrest and adipocyte differentiation. First, we demonstrate a clear switch in gene expression mediating the processes of growth and differentiation and that growth arrest following clonal expansion correlates closely with the expression of adipogenic transcription factors, C/EBP␣ and PPAR␥. Second, data are presented documenting a cascade of CKI expression that mark distinct states of growth arrest associated with adipogenesis. Third, the differentiation-dependent up-regulation of p18 and p21 is regulated at the level of mRNA and protein expression when non-precursor fibroblasts are converted to adipocytes by the expression of PPAR␥ in a ligandassociated fashion. Collectively, these data demonstrate that transcription factors that mediate adipogenesis also regulate the expression of cell cycle inhibitors providing a molecular mechanism coupling these processes during exit from the cell cycle and ensuring the irreversible growth arrest of terminal differentiation.
Data presented in this investigation confirm and extend in greater detail a recent report (33) documenting a cascade expression of members of two independent families of CKIs during the course of adipogenesis. As summarized in Fig. 7A, the protein expression of p27, p21, and p18 defines three unique states of growth arrest associated with saturation density, post-mitotic growth arrest, and the onset of terminal differentiation, respectively. The complexity and timing of expression suggest that multiple CKIs may play specific and diverse roles in coupling growth arrest and cell differentiation. It is interesting to note the synchrony of the inverse relationship between p27 protein and histone gene expression during clonal expansion. Other reports have indicated that p27 protein accumulates under quiescent conditions of serum deprivation and density arrest, decays rapidly with the onset of cell cycle progression, remains low during subsequent cell cycles and returns to high levels concomitant with the onset of growth arrest (35). The high levels of p27 immediately juxtaposed to S-phase presented in this investigation attest to a potential function of this cell cycle inhibitor and to the synchrony of entry into and exit from the cell cycle associated with clonal expansion. Although the increase in p27 following clonal expansion is consistent with establishment of a new saturation density, the combined expression of p27 with p21 during exit from the cell cycle and p27 with p18 during terminal differentiation may represent a synergistic role for multiple CKIs during distinct states of growth arrest. In this regard, the expression of p21 and p27 simultaneously may attribute to the synchrony and rapidity of growth arrest following clonal expansion. The possibility also exists that combined up-regulation of p21 and p27 following clonal expansion may play a permissive and/or regulatory role for subsequent adipocyte differentiation. In support of this notion, the highest degree of myelomonocytic cell differentiation has been shown to occur independent of chemical inducers when p21 and p27 were ectopically expressed together, suggesting that multiple CKI expression may be required for complete cell differentiation (36). Moreover, the observation that ectopic expression of p21 or p27, but not of p16, leads to megakaryocytic differentiation suggests the possibility of CKI specificity in regulating cell differentiation independent of growth arrest (37). Therefore, the timing and overlapping nature of CKI expression during adipogenesis may impart synergistic and specific functions specific to distinct states of growth arrest and different stages of adipocyte differentiation.
Albeit absent in fully differentiated adipocytes in vitro and in vivo, it should be emphasized that the expression of p21 dramatically increases twice during the course of cultured adipocyte differentiation. The initial up-regulation of p21, which coincides with the G 1 phase of clonal expansion, is consistent with reported regulation and function of p21 during early phases of cell cycle progression (38). In contrast, the subsequent up-regulation of p21 correlates directly with "exit" from clonal expansion and is dependent on the differentiation program for expression. Thus, it appears that one mixture of mitogen and hormonal agents utilized in the induction of differentiation activates two kinetically independent peaks of p21 protein expression, suggesting the possibility of two independent regulatory mechanisms based on the progression of proliferation versus differentiation. The decay in p21 in the presence of other CKIs during advanced stages of adipogenesis is consistent with other reports indicating a transient expression of p21 during myocyte differentiation in vitro (39) and during cardiac development in vivo (40). The association of p18, p21, and p27 with other proteins mediating both growth and differentiation is currently being investigated to ascribe specific functions of these CKIs during adipogenesis.
This investigation also presents data, generated from three fibroblastic cell lines regulating the expression of PPAR␥ by two independent mechanisms and supplement with two pharmacologically different ligands for PPAR␥, that provide direct evidence for a role of adipogenic transcription factors in regulating CKIs at the level of mRNA and protein expression. As summarized in Fig. 7B, cell lines that were PPAR␥ ligand-dependent for adipogenic gene expression were also ligand-dependent for regulation of p18 and p21, suggesting a regulatory role for PPAR␥ at some point upstream during the course of adipogenesis. The proximity of PPAR␥ in the differentiation paradigm to the regulation of p18 and p21 has yet to be determined. It is predicted, however, that regulation of p18 during adipogenesis occurs primarily at the level of gene expression, as changes in protein expression correlated with equivalent changes in mRNA accumulation. In preliminary experiments, inhibition of protein synthesis by cycloheximide in cells expressing PPAR␥ prevented p18 mRNA accumulation (data not shown), suggesting that other proteins downstream of PPAR␥ are likely to be involved in mediating p18 gene expression. The notion of an intermediate transcription factor is consistent with the observed delay between the expression of PPAR␥ and p18. Preliminary cycloheximide studies with p21 were not interpretable, inasmuch as inhibition of protein synthesis, independent of PPAR␥ expression, led to a dramatic increase in p21 mRNA accumulation. However, based on a potential conserved consensus sequence in the promoter of p21 and the coordinate kinetics of p21 and PPAR␥ expression, it is possible that this CKI is directly regulated at the level of gene expression by PPAR␥. Regulation of p21 protein expression during adipogenesis, however, is likely to be complex with changes in the magnitude and kinetics of protein expression occurring without coordinate changes in mRNA accumulation. Although the data presented here demonstrate an upstream regulatory role for PPAR␥ in the regulation of p21 during adipocyte differentiation, it is important to note that p21 protein expression presented in this investigation correlated closely with the expression of C/EBP␣. Interestingly, a role for C/EBP␣ has recently been shown in the regulation of p21 in hepatocytes and fibrosarcoma cells at the level of protein stability (41,42). Thus, the possibility exists that post-transcriptional regulation of p21 protein may, in fact, be regulated by C/EBP␣ during adipogenesis. The observation that ectopic expression of PPAR␥ also results in the expression of C/EBP␣ shown here and elsewhere (24) suggests the possibility of a cascade effect of these transcription factors in the regulation of p21 at the level of gene expression and protein stability, respectively. As numerous reports have emerged demonstrating a synergy of these transcription factors in the regulation of many aspects of the mature adipocyte, it would not be surprising to find that the complex regulation of CKI expression during adipogenesis also involves the combined efforts of both PPAR␥ and C/EBP␣. Experiments addressing the direct and indirect mechanisms of CKI expression by these and other transcription factors during adipogenesis are currently under investigation.
Others have reported that both C/EBP␣ (43) and PPAR␥ (44), when ectopically expressed, suppress the growth of various subconfluent, proliferating fibroblastic cell lines. Although the data presented here suggest that growth arrest is coupled to adipocyte differentiation through the expression of CKIs, it should be noted that other growth arrest mechanisms independent of these cell cycle inhibitors may also be imparted by adipogenic transcription factors under conditions that may not support adipogenesis. For example, it has been reported that C/EBP␣ regulates the growth-arrest-associated gene, gadd45 (45), and that PPAR␥ can induce cell cycle withdrawal by inhibition of E2F binding activity via down-regulation of the protein phosphatase, PP2A (44). Thus, it is likely that various mediators of adipocyte gene expression may regulate independent and/or synergistic growth arrest mechanisms as a process to ensure terminal differentiation. Determining the function of multiple CKIs during adipogenesis and deciphering the complex interactions of numerous adipogenic transcription factors in regulating their expression will provide a better understanding of the physiological control of adipocyte proliferation through coupling of growth arrest and cell differentiation. FIG. 7. Schematic model coupling growth arrest with adipocyte differentiation. A, summary of p18, p21, and p27 protein expression during the switch between adipocyte growth and differentiation. Changes in p18 (green), p21 (red), and p27 (blue) protein expression during 3T3-L1 preadipocyte differentiation were summarized from densitometry measurements of data presented in panels A and B of Fig. 2. The kinetics of histone H2B mRNA expression (shaded gray) illustrated in Fig. 2B was included for reference to S-phase progression during clonal expansion. Segments representing three unique states of growth arrest were illustrated above the summarized data. B, adipocyte differentiation signals, initiated by chemical inducers (MDI), activate a sequential cascade of transcriptional events that regulate gene expression responsible for the development of the functional adipocyte. In addition, chemical inducers activate synchronous cell cycle progression referred to as clonal expansion and a cascade of CKIs that define distinct states of growth arrest characterized for adipocyte differentiation. The expression of these CKIs is shown to be regulated by adipocyte transcription factors, thereby providing a molecular mechanism coupling growth arrest with adipocyte differentiation.