Introduction

White adipose tissue plays an important role in energy storage and organismal homeostasis through secretion of molecules with endocrine activities. The study of adipose differentiation has been highly facilitated by cell culture models, such as the fibroblastic 3T3-L1 and the 3T3-F442A cell lines. Both sister cell lines undergo adipose differentiation following stimulation with adipogenic serum proteins or growth hormones1,2,3,4. 3T3-F442A cells have the ability to differentiate into fat pads in vivo, suggesting that they respond to physiological signals in the body5, whereas 3T3-L1 cells do not6. 3T3-F442A cells can also differentiate into an osteoblast phenotype7, demonstrating the plasticity of this cell line. Thus, the 3T3-F442A cell line is suitable for the study of adipose differentiation, adipogenesis-regulating molecules and the expression of very early genes involved in this pathway.

Adipogenesis has previously been studied in culture conditions using highly adipogenic serum that is supplemented without or with molecules such as methyl isobutyl xanthine (MIX) and dexamethasone (Dex) that enhance adipose conversion8. However, none of these molecules have been shown to induce commitment in the absence of adipogenic serum. The use of small molecules to assist in adipocyte 3T3-F442A cells conversion provides the opportunity to study some of the mechanisms involved in the early stages of differentiation. Such an approach also facilitates the examination of signaling and transcriptional cascades involved in adipogenesis and discriminates between other stimulated pathways that could be unnecessarily and simultaneously activated by serum or growth hormones.

Recently, we demonstrated the ability of staurosporine (St) to induce adipose differentiation in the absence of adipogenic proteins9 and the ability of dexamethasone (Dex) to enhance differentiation of 3T3-F442A cells induced with St10. This model offers the following two advantages: 1) St/Dex can promote a maximum differentiation in 4 h without adipogenic serum (incubation with the serum alone requires approximately 48 h for the cells to reach the same level of adipogenesis9); and 2) St induces two well-defined stages for differentiation before clonal expansion that then permit the identification and analysis of early gene regulation. The first stage consists of 4 h of induction, where St induces progenitor cells to differentiate and the second stage consists of a subsequent 44 h of stabilization, where differentiation continues in the absence of the inducer but can still be reversed by anti-adipogenic substances or cytokines9. It is important to note, that this model of 3T3-F442A cells, allows an early response to adipogenesis, since only 4 hours of treatment with St is sufficient to trigger the adipose program, whereas adipogenic serum and MIX/Dex require, at least, 48 hours for induction in 3T3-L1 cells8. All the studies published with the 3T3-L1 cells are carried out in medium supplemented with adipogenic serum during growth phase of the culture and MIX/Dex are added later at confluency to enhance adipose conversion. The 3T3-L1 cells are able to differentiate with adipogenic serum without the addition of MIX/Dex, which later were found to enhance, but not initiate adipose conversion2,8 . These data, shows that the 3T3-F442A model we are using, is suitable to carry out kinetic analysis of gene expression, mainly during the early stages of induction.

Studies using cell lines have shown that a complex cascade of transcription factors is engaged in the generation of fat cells11. The peroxisome proliferator activated receptor gamma (PPARγ, pparg), particularly isoform 2 and CCAAT enhancer binding protein alpha (C/EBPα, cebpa) are considered the orchestrators of adipogenesis and their expression patterns determine adipose differentiation12,13,14. Also, it has been suggested that in 3T3-L1 cells, expression of PPARγ to promote adipogenesis depends on glucocorticoids, cAMP and serum mitogens via C/EBPβ11.

The principal transcription factor involved in the early steps of adipogenesis is C/EBPβ (gene: cebpb). Its expression precedes the expression of pparg2 and cebpa by approximately 30 h15, raising the question of whether other genes that are expressed such a long time after cebpb expression can regulate the expression of pparg2 and cebpa.

The Sterol Responsive Element Binding Proteins (SREBPs) are transcription factors encoded by the srebf1 and srebf2 genes. Recently, the importance of the SREBPs has been highlighted; these proteins function as the central hubs in lipid metabolism16. Using genome-wide expression analysis, we identified srebf1 as an early gene induced during adipogenesis of 3T3-F442A cells. The srebf1 gene encodes two proteins, SREBP1a and SREBP1c, with isoform -1c being the predominant isoform in most tissues analyzed17. However, the roles of SREBP1a and -1c during the early events of adipogenesis remain unknown. In this study, we found that srebf1a was expressed very early in induction and preceded the expression of all other adipogenic genes, including pparg2, cebpa, srebf1c and fabp4. Loss-of-function experiments revealed that expression of srebf1a was necessary for adipogenesis. We also showed that expression of srebf1a depended on the activity of GSK3β probably via the phosphorylation of C/EBPβ and that GSK3β activity played an essential regulatory role in the very early stages of induction of adipogenesis and maintenance of the adipose cell phenotype. Our results indicate the importance of the srebf1a gene during early adipogenesis and demonstrate that srebf1a acts as a link between the GSK3β-dependent signaling pathway and the beginning of the transcriptional cascade that results in the expression of pparg2.

Results

Adipogenesis involves lineage commitment, clonal amplification of committed cells and expression of the adipose cell phenotype. In 3T3-F442A cells, replication of DNA and clonal amplification precede the expression of the adipose phenotype when cells are induced to differentiate upon treatment with adipogenic serum18. Clonal amplification, which marks the beginning of adipose phenotype expression, has been shown to occur approximately 68 h after adipogenic serum supplementation in non-differentiating resting cultures and DNA synthesis takes place about 12 h earlier18. Because adipogenesis of 3T3-F442A cells has been shown to occur more rapidly following induction with St/Dex compared to adipogenic serum9, we attempted to identify the time course of clonal amplification under these conditions. Two-day post-confluent cultures of 3T3-F442A cells were incubated in DMEM (Dulbecco's Modified Eagle Medium) containing non-adipogenic serum with St/Dex for 4 h to induce cells to differentiate. At various time points, cell numbers were assessed in each culture. We found that the total cell number increased approximately 56 h following treatment, which corresponded to clonal amplification (Figure 1A). Based on our previous data, these results suggested that DNA synthesis likely occurred about 44 h into induction. In combination with previously described studies9, our data demonstrate the following two early stages for for differentiation: a 4-hour induction stage initiated by St/Dex, followed by a stabilization stage that lasts up to 44 h, which is followed by DNA synthesis and clonal amplification at 56 h post-treatment with St/Dex (Figure 1B). Diaz-Velasquez et al. (2008) found that at about 14 h after clonal amplification, the activity of the lipogenic enzyme, glycerol phosphate dehydrogenase, an early marker of phenotypic expression, began to increase in a time-dependent manner. These results are essential to more precisely identify the molecular events that regulate each adipogenesis stage.

Figure 1
figure 1

Expression of srebf1a and adipogenic genes during adipose differentiation of 3T3-F442A cells.

(A) Clonal expansion of 3T3-F442A cells induced to differentiate with staurosporine and dexamethasone (St/Dex). Post-confluent 3T3-F442A cells were induced to differentiate into adipose cells with St/Dex for 4 h followed by incubation in non-adipogenic conditions (N-Ad). Cells were harvested and counted at the indicated times. (B) Schematic representation of the time course of the adipogenic program induced by St/Dex. The main stages are shown at the indicated time frames. (C) Adipose conversion in post-confluent cultures induced with fetal bovine serum (FBS) for 48 h or St/Dex for 4 h in non-adipogenic medium. After induction, cultures were switched to non-adipogenic medium and grown for 144 h. Adipose clusters were assessed by staining with Oil Red O. (D) and (E) Expression of srebf1a and other adipogenic genes. Post-confluent cultures under non-adipogenic conditions (N-Ad) or induced with St/Dex were harvested at different time points and gene expression was evaluated by qRT-PCR. (F) The calculated time for the increase in the expression of adipogenic genes based on the gene expression kinetics. Briefly, we calculated the fit curve at a maximal R2 and determined the inflection points from the second derivative. Data are presented as the mean ± s.d.

Full size image

To quantify differentiation into adipocytes, we counted the number of adipocyte clusters in cultures at the conclusion of the experiment. Cells induced to differentiate into adipocytes form adipose clusters9 and as each cluster arises from the selective multiplication of one parental cell19, the number of adipose clusters corresponds to the number of induced cells. When we cultured 3T3-F442A cells in non-adipogenic conditions and then induced them with St/Dex for 4 h, differentiation to adipocytes was 10-fold higher compared to non-induced cultures and the number of adipocyte clusters was in agreement with the number of adipose clusters observed in cultures incubated with fetal bovine serum, which is highly adipogenic (Figure 1C).

The genes coding for the SREBPs are early and highly expressed during differentiation of 3T3-F442A cells

We performed a global screening using genome-wide expression microarrays to identify genes that may be involved in the various stages of adipogenesis. We compared 3T3-F442A cells under three distinct culture conditions: 1) cells in the induction stage with St/Dex (4 h); 2) cells in the stabilization stage (30 h from induction, prior to clonal amplification); and 3) cells in the phenotype expression stage (144 h), which consisted primarily of terminally differentiated mature adipocytes. We observed an early increase in the expression of the srebf1 gene. This gene differentially expresses two isoforms, -1a and -1c; however, microarray analyses did not reveal isoform-specific expression changes (data not shown).

Although the participation of the srebf1 gene in lipid metabolism has been described, the roles of the -1a and -1c isoforms have not been documented. Our results demonstrated that compared to non-induced cells, expression levels of both isoforms of the srebf1 gene were increased several fold during induction and stabilization stages and these levels continued to increase (Figure 1D). The Srebf1a isoform showed a rapid and sharp increase in expression after 4 h of induction and by the stabilization stage (8 h), the expression of srebf1a reached a maximum level at about 15-fold, which was maintained thereafter. During the first 3 h of the induction stage, srebf1a expression remained at levels similar to non-induced cells (Figure 1E). Srebf1c expression increased significantly at the end of the stabilization stage (48 h), reaching a maximum of 350-fold well into the phenotypic expression stage (72 h) and remained at those levels during terminal differentiation, forming mature adipocytes (Figure 1D). It is remarkable that during commitment, expression of srebf1a was higher than srebf1c and it took about 60 h longer for srebf1c to reach its maximum level of expression compared with srebf1a (Figure 1D). Srebf2, the third isoform in this family, which regulates cholesterol metabolism, did not undergo any significant change in expression in either of the culture conditions (Figure 1D), suggesting that cholesterol metabolism was not significantly stimulated during adipose conversion. In agreement with this, other studies have shown that total cholesterol synthesis is not altered during 3T3-L1 fat cell differentiation1.

The increase in srebf1a expression preceded the changes in the expression of other genes that have been described as characteristic regulators of 3T3 cell adipogenesis. We found that expression of pparg2, the regarded master regulator of adipogenesis14, did not change during induction but began to increase after 8 h and was 50-fold higher by the end of stabilization (48 h); at the phenotypic expression and terminal differentiation stage, pparg2 expression was 140-fold higher (Figure 1D, E). The cebpa gene did not show any change in expression during induction, but its expression was increased (12-fold) at the end of the stabilization stage, reaching its maximum (160-fold) after clonal expansion and at the beginning of phenotype expression (72 h) (Figure 1D).

The fabp4 gene, which encodes aP2, a lipid carrier expressed during adipose differentiation20, had similar expression kinetics as cebpa; fabp4 reached its maximum level of expression of 108-fold in terminally differentiated adipocytes (Figure 1D). The expression of cebpb transiently increased (6-fold) during the 4 h induction stage but returned quickly to the basal levels found in non-induced cells by 8 h (Figure 1D, E). The early and transient increase in expression was accompanied by changes in the phosphorylation state of the C/EBPβ protein.

The expression kinetics for each gene was different. To more precisely identify the time at which expression of each gene began to increase, we determined the mathematical equation for each kinetic plot. We used the second derivative of each function and calculated the inflection point of the derivative that corresponded to the time point at which expression of each gene began to increase. We observed the following time course of gene expression (Figure 1F): srebf1a was the first expressed gene, with expression beginning at less than 1 h of the induction stage. The other genes began to be expressed well into the stabilization stage of commitment. Increasing expression levels of pparg2 followed the expression of srebf1a and then cebpa, srebf1c and fabp4 expression followed. This analysis demonstrated that these genes began to be expressed during induction and stabilization and likely before clonal expansion. It also suggests that SREBP1a may be the transcription factor that orchestrates the adipogenic program, regulating even pparg2, the well-known adipogenic master regulator.

With the available data on cebpb expression, we could not calculate the second derivative, but there was a significant increase in cebpb expression that clearly occurred in the first hour of induction (Figure 1E). Thus, cebpb and srebf1a were the two genes that changed their expression in the very early stage of induction.

Loss of function of srebf1a inhibits adipose differentiation of 3T3-F442A cells

To study the necessity of the expression of the srebf1a gene, we knocked down its expression by shRNA targeting. 3T3-F442A cells were transfected with the pSingle-shSRE1a vector, which expresses an shRNA targeting srebf1a upon treatment with doxycycline. After transfection, cells were seeded and induced to differentiate with St/Dex. Silencing of the srebf1a gene by pSingle-shSRE1a led to about a 70% inhibition in srebf1a mRNA expression compared to control cultures not treated with doxycycline (Figure 2A). As expected, parallel cultures transfected with a vector containing a non-related sequence did not show silencing of the gene (results not shown). The expression of the adipogenic genes that are characteristic of adipose differentiation, including pparg2, cebpa and srebf1c, also decreased by an amount similar to srebf1a (Figure 2A). These results demonstrate that SREBP1a activity is upstream of pparg2, cebpa and srebf1c and that SREBP1a is required for the regulation of these adipogenic genes.

Figure 2
figure 2

Srebf1a is necessary for adipogenesis of 3T3-F442A cells.

(A) Knockdown of srebf1a expression by specific shRNA blocks the adipogenic program. 3T3-F442A cells were transfected with pSingle-shSRE1a containing a specific shRNA against srebf1a. Transfected cells were then induced to the adipogenic program with St/Dex for 4 h. Expression of shRNA was induced by doxycycline (Dox) and gene expression was evaluated by qRT-PCR at 48 h after induction. (B) St/Dex induced early phosphorylation of C/EBPβ at Thr188 during adipocyte differentiation. Cultures of 3T3-F442A cells were induced with St/Dex and total protein was extracted and immunoblotted at the indicated times. (C) St/Dex induction stimulates GSK3β phosphorylation at Tyr216. Protein extracts from 3T3-F442A cells at time zero (T0) in non-adipogenic conditions at 4 h (N-Ad) or induced with St/Dex for 4 h were immunoblotted with specific antibodies. (D) Inhibition of GSK3β activity inhibits phosphorylation of C/EBPβ at Thr188. Cells were induced with St/Dex for 4 h and cultures were then switched to fresh medium supplemented with 50 µM SB415286 (SB), a specific inhibitor of GSK3β activity, for 20 h. Protein extracts were obtained at time zero (T0) or in the presence (SB) or absence (St/Dex) of the inhibitor and analyzed by immunoblotting. (E) GSK3β activity is necessary for the expression of the adipogenic genes and subsequent adipocyte differentiation. Cells were treated as in D. Cultures in the presence (SB) or absence (St/Dex) of SB415286 were followed for 144 h to evaluate adipose conversion. Gene expression was evaluated at the indicated times by qRT-PCR. Data are presented as the mean ± s.d.

Full size image

GSK3β regulates the expression of srebf1a through C/EBPβ phosphorylation

The C/EBPβ protein is a transcription factor that targets distinct genes in several tissues. Phosphorylation of C/EBPβ is required for DNA binding and determines the specificity of target genes transactivated by C/EBPβ. For the adipogenic program, phosphorylation of C/EBPβ at Thr188 is required. Phosphorylation of Thr188 could be mediated by ERK or GSK3β, as Thr188 is part of the ERK/GSK3 phosphorylation consensus site21. An antibody against phospho-C/EBPβ Thr188 showed that phosphorylation of the protein began to increase within 0.5 h of the induction stage, reaching a maximum at 2 h and then it began to decrease to basal levels after 4 h, corresponding with the end of the induction stage (Figure 2B). Our results also showed that this phosphorylation event preceded the expression of various adipogenic genes, including srebf1a, the earliest expressed gene in the adipogenic pathway (compare Figure 2B with Figure 1D and E). We have previously demonstrated that GSK3β activity is essential for the induction of the adipogenic program9. The activity of GSK3β is regulated by posttranslational modifications. Phosphorylation of GSK3β at Tyr216 activates the kinase, but phosphorylation at Ser9 inactivates the kinase even in the presence of Tyr216 phosphorylation22. In cultures induced with St/Dex, we found that GSK3β phosphorylated at Tyr216 was the more predominant form than GSK3β phosphorylated at Ser9 (Figure 2C). By densitometric analysis of the immunoblots, GSK3β-Tyr216 was approximately 80% more abundant than the inactive form of GSK3β, GSK3β-Ser9 (Figure 2C). We also inhibited the activity of GSK3β using its selective inhibitor, SB415286. Phosphorylation of C/EBPβ at Thr188 was completely abolished in cultures treated with SB415286 (Figure 2D). These results strongly suggest that the activity of GSK3β regulates the phosphorylation of C/EBPβ.

The GSK3β inhibitor, SB415286, also blocked the expression of the adipogenic genes tested and thus prevented adipose differentiation (Figure 2E). We next analyzed the promoter sequence of the srebf1 gene. Murine srebf1 has two separate promoters that give rise to isoforms -1a and -1c; these forms mutually exclude each other17. Sequence analysis of the srebf1a promoter using the TFSearch algorithm23 revealed the existence of at least 5 putative binding sites for C/EBPβ between positions −1075 to −3527. Additionally, there was a consensus site for C/EBP at position −841. These results show that GSK3β is involved in regulating phosphorylation and activation of C/EBPβ and that this transcription factor is required to transactivate srebf1a expression, leading to the early steps of adipogenesis.

PPARγ does not transactivate srebf1a expression

We showed that srebf1a is one of the earliest genes expressed during the adipogenic program, even prior to pparg2 expression (Figure 1D, E). Srebf1c and srebf1a share a high level of sequence homology and differ in the first exon; here, SREBP1a is 24 amino acid residues longer. Sequence analysis of the srebf1c promoter revealed a putative binding site for PPARγ that was about −6900 bp from the transcription start site. Similar analyses of the srebf1a promoter did not reveal any recognition site for PPARγ. To date, there has been no experimental evidence that PPARγ transactivates the expression of srebf1a or -1c. Nevertheless, due to the high homology between srebf1a and -1c and the putative binding site for PPARγ in the srebf1c promoter, it is conceivable that basal levels of PPARγ in pre-adipocytes could be sufficient to transactivate, either directly or indirectly, the expression of srebf1a or -1c.

To exert its transcriptional activity, PPARγ requires an endogenous lipid ligand14, but rosiglitazone (Ros), a synthetic ligand, can also activate the transcriptional function of PPARγ24. Therefore, we attempted to identify by functional experiments with Ros whether srebf1a or -1c expression relied on PPARγ transcriptional activity. Early treatment with Ros during induction should activate the basal levels of PPARγ and thus transactivate PPARγ target genes. This process would be evident by an early shift in the expression curve of those genes. We induced post-confluent cultures of 3T3-F442A cells with St/Dex and then added Ros to the cultures during induction (0–4 h). Cultures were maintained for up to 144 h to assess adipose conversion. Gene expression was evaluated by qRT-PCR at various time points.

Ros accelerated the expression of pparg2, cebpa, srebf1c and fabp4, suggesting that Ros activation of basal PPARγ leads to increased self-activation and the transactivation of cebpa, srebf1c and fabp4. More importantly, srebf1a expression did not increase upon addition of Ros (Figure 3A), demonstrating that the expression of this gene is not under the regulation of PPARγ or other transcription factors that are responsive to this drug.

Figure 3
figure 3

Srebf1a expression does not depend on PPARγ transcriptional activity.

(A) Rosiglitazone (Ros), a potent PPARγ activator, accelerates the expression of adipogenic genes, but not of srebf1a. Post-confluent cultures were induced with St/Dex supplemented with 1 µM Ros during induction. Gene expression was determined by qRT-PCR. (B) Adipose differentiation induced with St/Dex with and without Ros. Data are presented as the mean ± s.d. (C) Representative cultures differentiated in the presence or absence of Ros; note that the adipose clusters treated with Ros are larger in size and more intensely stained with Oil Red O. N-Ad, non-adipogenic condition. (D) Schematic representation of the proposed signaling pathway and transcriptional cascade during adipogenesis. The data show that the early expression of srebf1a depends on GSK3β activity and the expression of srebf1a is necessary for the expression of the adipogenic genes, pparg, cebpa and srebf1c.

Full size image

In parallel cultures, we determined the number of adipose clusters at the end of the experiment. We found that the number of adipose clusters was similar in all culture conditions either with or without Ros (Figure 3B). This indicated that the accelerated gene expression did not increase induction into differentiation of the adipose precursors in the cultures. However, the cluster size was larger in the cells treated with Ros (Figure 3C), suggesting that the early activation of the genes might lead to higher adipose conversion through larger adipose clusters and/or higher lipid accumulation.

Discussion

Adipose differentiation of 3T3 cells is regulated by networks of transcriptional factors11. Cebpb is one of the earliest transcription factors expressed during adipose differentiation and it precedes pparg2 and cebpa expression by 30 h15. Recently, SREBPs have been regarded as important central hubs in lipid metabolism16.

The ratio of the SREBP1c protein to the -1a protein varies widely in different tissues. The -1c isoform is more abundant in adipose tissue and liver than the -1a isoform17, although -1a shows higher transcriptional activity25. Deletion of srebf1a is lethal to the embryo, suggesting an unknown important role of the protein during embryonic development26. Overexpression of srebf1a in transgenic mice produced adipocyte hypertrophy and fatty liver27. These and other data suggest that isoforms -1a and -1c may be controlled independently and the two may respond differentially to organ-specific and metabolic factors17. Despite the key function of these transcription factors in lipid homeostasis, a role for SREBP1a in adipose differentiation has not been reported.

To study the adipogenesis program, we took advantage of the two-stage model in 3T3-F442A cells, which includes an induction stage of 4 h followed by stabilization for 44 h, a time in which the adipogenic inducer is no longer required9. We also used St as a specific small molecule to induce differentiation without any adipogenic proteins9, circumventing the ERK/MAPK signaling pathway that is activated by serum adipogenic proteins or growth hormone to promote adipogenesis28. It has been described that St directly activates GSK3β29, a molecular switch in the Wnt pathway. Here, we report that GSK3β induces C/EBPβ phosphorylation at Thr188, which then promotes the expression of the srebf1a transcription factor. These events precede the expression of pparg2 and cebpa and regulate the early stages of adipogenesis. As deletion of srebf1a leads to embryonic lethality30, expression studies of this gene in 3T3-F442A cells offer a suitable experimental model to elucidate the role of this transcription factor during adipogenesis. The early expression of srebf1a and the loss-of-function experiments showed that the expression of srebf1a was essential for adipogenesis through regulation of pparg2, cebpa and srebf1c. Srebf1a expression increased at least 10-fold in the 4 h of the induction stage and increased prior to augmentation of pparg2 expression, which occurred after the induction stage. After the inducer (St/Dex) was removed, cells were at the stage of stabilization well after srebf1a mRNA had reached its maximum level. Therefore, the well-known, adipogenic transcriptional cascade involving PPARγ is activated after the cells have begun to express srebf1a and no longer require an inducing agent.

ADD1/SREBP131, which was later identified to be the -1c isoform, does not bind to the pparg2 promoter32, but it can activate PPARγ through the production of a fatty acid derivative, which acts as an endogenous ligand33. Analysis of the pparg2 sequence with the TFSearch algorithm did not reveal any recognition site in its promoter region for binding of SREBP1a or -1c. However, the same analysis of the srebf1c promoter revealed a recognition site for PPARγ. In combination with previous work, our data demonstrating the up-regulation of srebf1c after pparg2 and the accelerated expression of srebf1c in the presence of Ros (an exogenous ligand for PPARγ activation) support the idea that PPARγ may directly regulate srebf1c expression. Additionally, in the phenotype expression stage, a mutual regulation of PPARγ and SREBP1c may take place to regulate adipogenesis and lipid metabolism through a feedback loop, where PPARγ transactivates srebf1c expression and SREBP1c produces an internal ligand to activate PPARγ (Figure 3D). Due to the lack of a recognition site for SREBP1a in the pparg2 promoter sequence, we suggest that SREBP1a activity, similar to SREBP1c, may produce an intermediary that transactivates pparg2 and upregulates its expression, warranting further investigation into a possible pathway of pparg2 regulation by SREBP1a (Figure 3D).

SREBP1a appears to be the strongest activator of Sterol Responsive Element (SRE), SRE-like and E-box sequences25. Previous work has shown that SREBP1a and -2, but not -1c, transactivate SRE binding sites found in cholesterogenic genes25. SREBP1a and -1c transactivate targets with E-box sequences in their promoters to activate the lipogenic pathway25. Interestingly, SREBP1c has an SRE binding site in its promoter34, suggesting that SREBP1a could transactivate it. Analysis of the promoter sequence of srebf1a did not show any SRE or E-box binding sites. Srebf1a appears to have a constitutively low expression pattern in liver, but its overexpression in transgenic mice leads to liver steatosis35, which is consistent with a report indicating that the activation of the lipogenic program could also occur through SREBP1a25. Based on our data, we propose that an SRE binding site in srebf1c could be occupied by SREBP1a to trigger srebf1c expression, which then contributes to the activation of the lipogenic program. It is conceivable that SREBP1a activates expression of srebf1c and because it is required for expression of pparg2 either directly or indirectly, it would contribute to a feedback loop among these three factors and trigger the rest of the transcriptional adipogenic cascade (Figure 3D). The feedback loops between PPARγ and CEBPα36 and between SREBP1a, SREBP1c and PPARγ described here raise the possibility that they not only regulate induction to adipogenesis but also have a regulatory function in lipid metabolism, as their genes remain highly expressed in mature adipocytes.

One of the early steps for adipose conversion is the inhibition of the Wnt pathway37. The Wnt pathway maintains pre-adipocytes in an undifferentiated state37 and direct activation of GSK3β inhibits intracellular Wnt signaling38. Staurosporine activates GSK3β by maintaining its Tyr216 phosphorylation, which is the active state of the enzyme29. Our results show that staurosporine induces adipose conversion by preserving GSK3β phosphorylation at Tyr216, which is subsequently required to induce C/EBPβ phosphorylation. This is followed by the expression of srebf1a during the first 4 h of induction. After this point, the cells enter an intrinsic and autonomous pathway of commitment without the involvement of St or other adipogenic factors. The persistent GSK3β activity has been found to be required not only for induction but also for its subsequent stabilization and clonal amplification of adipogenic cells9. Some of the possible mechanisms of progression into intrinsic signaling during differentiation include persistent protein kinase activation, transcriptional autoactivation and late stabilization of gene expression39. Because GSK3β remains active for progression into intrinsic signaling during commitment, terminal differentiation of adipose cells and lipid metabolism, this kinase should have a crucial role during adipogenesis. The persistent activation of GSK3β is due to its autophosphorylation on Tyr216 that is initially mediated by chaperone proteins, but after maturation of the kinase, the chaperone proteins are no longer required40.

More recently, it has been shown that two additional transcription factor genes, klf4 and klf5, are expressed early during 3T3-L1 cell adipogenesis. KLF4 binds to the cebpb promoter41. Klf5 expression is induced by C/EBPβ and C/EBPδ and it has been suggested that KLF5 directly affects the expression of pparg2 and cebpa, probably by forming an enhanceosome with C/EBPβ/δ on the pparg2 promoter42. Despite the possible differences between the principal adipogenesis models, the 3T3-L1 and 3T3-F442A cells, it is clear that the expression of klf4, klf5 and srebf1a occurs very early. Based on data from this and other studies, we propose that induction of adipocyte commitment involves the following (see Figure 3D): 1) inhibition of the Wnt pathway through GSK3β activation; 2) KLF4 binding to the cebpb promoter, activation of CEBPβ through induction of phosphorylation by GSK3β and expression of srebf1a via phosphorylated CEBPβ; and 3) following activation of cebpb by KLF4, one pathway could involve klf5 and another srebf1a, with both pathways converging on the expression of pparg2. It is equally likely that the expression of srebf1a and klf5 are coordinated and not independently regulated. This could be explored by determining the expression of klf5 in cells in which srebf1a is knocked down. Finally, after the expression of klf4, klf5 and srebf1a during the early stage of adipogenesis and without the requirement of the inducer, cells continue through the stabilization stage with the expression of pparg2, cebpa and srebf1c. The progression into these stages during points to a regulatory transcriptional cascade for adipogenesis that is more complex than previously thought. The necessity of srebf1a expression and the presence of feedback loops between SREPB1a/1c and PPARγ and between CEBPα and PPARγ (Figure 3D) suggest the existence of finely tuned regulatory pathways involving the SREBP1 transcription factors in adipose biology. These possible alternate pathways for adipogenesis warrants further study into these new lines of investigation. Several crucial points of adipogenesis regulation emerge from our data and offer new insights on obesity and its related pathologies. Our studies point to the activity of GSK3β and SREBP1a/1c as new control points for adipogenesis and, thus, new targets for therapeutic intervention in obesity and diabetes. It is important to note from our studies and others that adipogenesis induction or stimulation by various substances (St/Dex, adipogenic proteins, growth hormones or MIX/Dex) all require GSK3β activation. Therefore, this kinase appears to have a crucial function as the molecular switch to initiate a differentiation program. Our data clearly demonstrate the requirement for persistent activation of GSK3β during all stages of adipose differentiation and phenotype expression and its close relationship with SREBP1a, indicating the importance of these molecules as regulatory factors for lipid homeostasis.

Methods

Materials

Eagle's medium modified by Dulbecco and Vögt (DMEM) and TRIzol® solution were purchased from Invitrogen-Life Technologies (Carlsbad, CA). Adult bovine serum was from M.A. Bioproducts (Walkersville, MD). We obtained adult cat serum by bleeding domestic adult cats in accordance with the NIH guidelines for the welfare of research animals. Epidermal growth factor was from IMCERA Bioproducts Inc. (Terre Haute, IN). Insulin, d-biotin, human transferrin, triiodothyronine, staurosporine, dexamethasone, β-mercaptoethanol, Oil Red O, dimethyl sulfoxide, 3-[(3-Chloro-4-hydroxyphenyl)amino]-4-(2-nitrophenyl)-1H-pyrrol-2,5-dione (SB415286) and rosiglitazone were all from Sigma-Aldrich Corp. (Saint Louis, MO). All other reagents were of analytical grade.

Cell culture and induction of adipose differentiation

The 3T3-F442A cells were seeded in culture dishes at 1.25×103 cell/cm2 in non-adipogenic growing medium consisting of DMEM supplemented with 4% adult cat serum, 5 µg/ml insulin and 1 µM d-biotin3,10. For experiments, one- to two-day post-confluent cultures were grown in non-adipogenic medium containing DMEM supplemented with 2% adult cat serum, 0.2% adult bovine serum, 5 µg/ml insulin, 5 µg/ml transferrin, 1 µM d-biotin, 2 nM triiodothyronine, 40 µM β-mercaptoethanol and 0.01 ng/ml epidermal growth factor4. Adipogenesis was induced in post-confluent cultures with staurosporine (St) and dexamethasone (Dex) for 4 h in the non-adipogenic medium as previously described9,10. Cultures were incubated for up to 144 h and adipose conversion was evaluated by counting adipose clusters stained with Oil Red O43. All cultures were incubated at 37°C in a 10% CO2 atmosphere.

Short hairpin RNA cloning and cell transfection

A short hairpin RNA (shRNA; Supplementary Table S1) against srebf1a mRNA was cloned into the pSingle-tTS-shRNA plasmid (Clontech Laboratories; Mountain View, CA) at the XhoI and HindIII cloning sites. The resulting construct was termed pSingle-shSRE1a. Expression of shRNA was driven by a tetracycline-responsive Pol III hybrid promoter. For experiments, purified endotoxin-free plasmid extracts were transfected into 3T3-F44 cells using Amaxa's Nucleofector™ (Lonza; Basel, Switzerland) with a Cell Line Nucleofector Kit V (Amaxa; Cologne, Germany). The 3T3-F442A cells were then harvested and resuspended in Nucleofection™ solution at 3.0×106 cells/100 µl. After the addition of 2.5 µg shRNA expression vectors, the cells were transfected using Nucleofector and seeded onto 35-mm plates at a near-confluent density of 4×105 in the absence or presence of 1 µg/ml doxycycline. Transfected cells were induced to differentiate with St/Dex 2 days after transfection. Differentiation experiments were carried out using the same media and culture conditions as described earlier and the knockdown efficiency was verified by relative quantitative RT-PCR assays.

Relative-quantitative RT-PCR

Cellular RNA was isolated with TRIzol® from cultured cells. Two micrograms of total RNA was transcribed into cDNA using reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Relative-quantitative PCR (qRT-PCR) was carried out using the FastStart Universal Sybr Green Master (Rox) kit (Roche Applied Science; Mannheim, Germany) and amplification was performed on a 7500 real-time thermal cycler (Applied Biosystems; Carlsbad, CA). Specific primers for each gene are listed in Supplementary Table S2. Expression values were determined by the 2−ΔΔCT formula using the expression of the ribosomal phosphoprotein large P0 (rplp0) gene as the normalizing factor.

Immunoblotting

At the times indicated, cells were lysed with ProteoJet Mammalian Cell Lysis Reagent (Fermentas Inc., Glen Burnie, MA) supplemented with Complete® protease inhibitor (Roche Applied Science) and phosphatase inhibitors. Proteins were separated by SDS/PAGE and immunoblotted with antibodies against human phospho-C/EBPβ Thr235 (the equivalent of murine Thr188) from Cell Signaling Technology (Berkeley, CA), phospho-GSK3β Ser9 (Abcam; Cambridge, UK), C/EBPβ and phospho-GSK3β Tyr216 from Santa Cruz Biotechnology (Santa Cruz, CA) and with Rab Guanine Nucleotide Dissociation Inhibitor (GDI, Invitrogen).

Data management and statistical analysis

Data are presented as the mean plus/minus standard deviation of the mean. Data were analyzed by the two-tailed Student's t-test for means analysis when comparing two data groups, or analysis of variance (ANOVA) when comparing three or more data groups. Statistical significance was identified if the P value was lower than 0.05.