Fra1 overexpression directly blocks adipogenesis

To confirm that the inhibition of adipocyte differentiation is caused by the increased Fra1 level and is not a consequence of the transgene itself, we retrovirally stably transfected an adiopogenic cell line with a vector expressing Fra1. By Western analysis, we verified that Fra1 is overexpressed in this cell line (Fig. 3.25A). Consistently with the experiments using Fra1tg primary osteoblasts, we observed a reduced differentiation to mature adipocytes by Oil Red O staining at day 21 of adipogenic differentiation (Fig. 3.25B). In addition, the expression of marker genes for mature adipoytes, Glut4 and Ap2 was not induced in cells overexpressing Fra1 as it was using cells infected with an empty vector (Fig. 3.25C and D).

We could show that the expression of transcription factors regulating the early differentiation to adipocytes, C/ebpβ ^{and C/ebp}δ were unchanged (Fig. 3.25E). However, in agreement with the results obtained from primary osteoblasts, the expression of C/ebpα and Pparγ^{2 was} significantly reduced in the Fra1 overexpressing cell line (Fig. 3.25E). The inhibition of adipocyte differentiation as well as the decreased expression of C/ebpα was confirmed by differentiation experiments using two other independently infected cell lines (data not shown) as well as on protein level (Fig. 3.25F).

A

Figure 3.25: Fra1 overexpression in adipogenic cell line blocks adipogenesis by inhibiting C/ebpα expression. (A) Western blot analysis of Fra1 expression in the adipogenic cell line infected with Fra1 encoding virus (pBabe-Fra1) or empty vector (vector); β-actin is used as loading control. (B) Oil Red O staining of the infected cells cultured for 21 days in adipogenic condition. Quantitative PCR analysis of Glut4 (C) and Ap2 (D) expression during the course of differentiation of cells infected with the vector or pBabe-Fra1. (−) control cell culture; (+) Ins+IBMX+Dex treated cell culture. Data are the mean ± s.e.m. of at least four different experiments. (E) Quantitative PCR analysis of the expression of markers for adipogenesis, the data represent the ratio of the level measured in cells overexpressing Fra1 reported to the level in cells infected with the empty vector at day 21 of differentiation. (F) Western blot analysis of C/EBPα expression in Fra1 overexpressing cells and cells infected with the empty vector; β-actin is used as loading control. (Fig. 3.25A, B and F are part of the PhD thesis of Frank Driessler)

3.4.4 No detectable regulation of C/ebpαααα expression via binding of Fra1 to C/EBPββββ

Similarly to the effect of ∆FosB on adipogenesis (Kveiborg et al., 2004), reduced levels of C/ebpα could be the result of the binding of Fra1 to C/EBPβ inhibiting transcriptional activation of C/ebpα. To examine protein interactions we performed co-immunoprecipitation experiments using protein extracts of a cell line overexpressing Fra1 without stimulation or stimulated with a cocktail of insulin, IBMX and dexamethasone. We could not detect binding of the flag-tagged Fra1 protein to C/EBPβ by precipitating with an anti-FLAG antibody (IP Flag, Fig. 3.26A). Nevertheless, Fra1 protein was clearly present in the protein extracts and dimer formation of Fra1 with c-Jun could be detected. Conversely, we were not able to

immunoprecipitate the flag-tagged Fra1 protein by performing the experiment with an anti-C/EBPβ antibody (IP anti-C/EBPβ, Fig 3.26B), despite an increased C/ebpβ expression after adipogenic stimulation. Therefore, the reduced level of C/ebpα is most probably not a result of the inhibition of C/EBPβ activity.

input

IP Flag

negative control isotypecontrol

C/EBPββββ (30kD, 20kD)

c-Jun (40kD)

IP C/EBPββββ

Flag

C/EBPββββ (30kD, 20kD)

Flag

withIns/IBMX/Dex

withoutIns/IBMX/Dex negative control isotypecontrol

withIns/IBMX/Dex

withoutIns/IBMX/Dex

input

A B

Figure 3.26: No detectable binding of Fra1 to C/EBPβ. Western blot analysis of the immunoprecipitation of Fra1-flag using an anti-FLAG antibody (IP Flag) or an isotype control (A) and of C/EBPβ (IP C/EBPβ) (B).

Total extract of stimulated cells was used as control (input). Experiments were performed with antibodies directed against C/EBPβ, FLAG or Jun.

3.4.5 Fra1 directly inhibits the transcription of C/ebpα

The decreased level of C/EBPα observed in the white adipose tissue of Fra1tg mice as well as in the adipogenic cells overexpressing Fra1 strongly suggest that Fra1 was inhibiting adipogenesis by directly blocking C/ebpα expression. We performed experiments using a POB-derived adipogenic cell line stably overexpressing a vector containing Fra1 fused to the ligand binding domain of the human estrogen receptor. These cells can be induced by Estradiol leading to an accumulation of the fusion protein in the nucleus. Consistent with experiments using cells transfected with the vector containing Fra1, it has been shown that addition of Estradiol to the adipogenic cocktail leads to a reduced adipogenic potential of cells overexpressing the fusion protein compared to cells transfected with an empty vector. By using the cell line expressing the inducible ER-Fra1 we defined the short term effect of Fra1

induced nuclear Fra1 accumulation by Estradiol. We detected a reduced C/ebpα expression level already 12 hours after addition of Estradiol in cells expressing ER-Fra1 (Fig. 3.27A) suggesting a direct repression of C/ebpα transcription by Fra1.

By sequence analysis using TFSearch, we identified five potential AP-1 binding sites in the sequence of the C/ebpα promoter (Fig. 3.27B). Thus, we analyzed if overexpression of Fra1 mediates the transcriptional repression of C/ebpα . We therefore cloned 3359 base pairs of the C/ebpα promoter into the pGL3-basic luciferase reporter vector. Transient co-transfection experiments with an adipogenic cell line using the plasmid containing the promoter construct as well as a Fra1 coding plasmid were performed. With increasing amounts of the Fra1 expressing vector, we observed a dose dependent decrease in luciferase activity (Fig. 3.27C).

To determine which AP-1 binding site mediates C/ebpα repression, we generated two other constructs containing the two or one proximal AP-1 binding sites. Deletion of the three distal AP-1 binding sites did not abolish the repressive effect of Fra1 on the C/ebpα promoter activity (Fig. 3.27D), nor did the deletion of the four distal AP-1 binding sites (Fig. 3.27E).

To analyze if this AP-1 binding site most proximal to the transcription start is responsive to Fra1, we performed chromatin immunoprecipitation experiments using extract isolated of a cell line overexpressing Fra1 as well as a control cell line. We could show that Fra1 indeed binds to this sequence (Fig. 3.27F). Thus, these data demonstrated that Fra-1 regulates adipogenesis by directly inhibiting C/ebpα expression.

B (ER-Fra1) cultured in the absence of estradiol (-E2) or stimulated for the indicated time with estradiol (+E2) 24 hours after adding the adipogenic cocktail. Data are the mean ± s.e.m. of three independent experiments. (B) Schematic representation of constructs of the C/ebpα promoter cloned in front of the luciferase gene into pGL3-basic. Grey boxes represent potential AP-1 binding sites; arrows show localization of primers for ChIP experiments. (C) Luciferase activity in an adipogenic cell line transfected with vector containing the full-length construct and increasing concentrations of pBabe-Fra1 relative to cells transfected with the empty vector. Luciferase activity in cells transfected with the promoter constructs containing the two proximal (D) or the first (E) proximal AP-1 binding site and 0, 0.5 and 1µg of pBabe-Fra1. (F) Chromatin immoprecipitation (ChIP) for AP-1 binding site 1.

PCR products are shown in the input, after immunoprecipitation with Fra-1 antibody (IP-Fra1) or with isotype

4. Discussion

Adipocyte and osteoblast differentiation are known to be both systemically and locally co-regulated (Rosen, 2008). Several components of the AP-1 transcription factor are involved in osteoblast function. This is particularly true for Fra1 whose gain of function increases and loss of function decreases osteoblast activity (Eferl et al., 2004; Jochum et al., 2000). However, a function for AP-1 members in adipogenesis was until now only described for ∆FosB (Kveiborg et al., 2004; Rowe et al., 2009; Sabatakos et al., 2000). In this thesis, we investigated the role of the Fos family member Fra1 in adipogenesis. We showed that mice overexpressing Fra1 that develop a progressive osteosclerosis display a concomitant progressive loss of white adipose tissue accompanied by a decreased size of adipocytes and that this defect is most probably caused by a cell autonomous direct inhibition of C/ebpα transcription by Fra1.

4.1 Phenotype of the Fra1 transgenic and the Fra1 knockout mouse

AP-1 acts as a sensor for extracellular signals regulating a variety of cellular processes.

Several mouse models have been generated to investigate the effects of overexpression or deletion of the single AP-1 family members. In this study, we analyzed the function of Fra1 in adipogenesis. We therefore used the Fra1 transgenic mice for gain of function as well as the Ap2-cre Fra1^f/f mice for loss of function. The Fra1 transgenic mice express the murine Fra1 gene under the control of the H2-K^b-promoter, leading to expression of the transgene in most tissues. Young Fra1 transgenic mice do not have any overt phenotype. However, starting at the age of 1 month, a reduced growth rate was described (Jochum et al., 2000). Mice overexpressing Fra1 develop an osteosclerotic phenotype caused by an enhanced bone formation as a consequence of an increased number of mature osteoblasts, a phenotype similar to the osteosclerosis described for ∆FosB transgenic mice (Sabatakos et al., 2000). In vitro experiments using primary osteoblasts derived from Fra1tg mice showed that the increased number was not caused by an increased proliferation rate, but due to a cell autonomous accelerated osteoblast differentiation. As a result of the increased bone formation, splenomegaly due to extramedullary hematopoiesis was observed. In addition, Fra1tg mice can develop lung tumors and liver cirrhosis (Jochum et al., 2000; Kireva et al., 2011).

Before analyzing mice overexpressing Fra1, we tried to determine the effect of Fra1 deficiency in adipogenesis. As Fra1 knockout mice are embryonic lethal due to defects in the placenta (Schreiber et al., 2000), Fra1^f/f mice have been generated and crossed with MORE-cre mice, leading to a deletion of Fra1 in the entire embryo except in the extraembryonal tissue. MORE-cre Fra1^f/f mice were described to be osteopenic due to a decrease in bone matrix formation with a reduced level of osteocalcin, collagen1a2 and matrix Gla protein. No other overt alterations were found in the MORE-cre Fra1^f/f mice (Eferl et al., 2004). For our studies, we crossed the Fra1^f/f mice with the Ap2-cre mice to obtain a specific deletion of Fra1 in mature adipocytes. We could not observe a clear adipocyte phenotype in Ap2-cre Fra1^f/f mice. However, this could be due to the inefficient cre-mediated recombination in the adipose tissue. Thus, we investigated the effect of gain-of-function of Fra1 by analyzing Fra1 transgenic mice.

4.2 Fra1 overexpression decreases adipose tissue mass We found that Fra1 transgenic mice show an age-dependant loss of body weight due to a reduced amount of adipose tissue mass. While the weight of the visceral fat pad is similar in young Fra1 transgenic mice compared to wild-type littermates, the transgenic mice completely loose their visceral fat tissue when aging. To investigate if the cause of the observed lipodystrophy is based on a systemic defect, we analyzed the glucose and lipid metabolism of the Fra1 transgenic mice and compared our results to the lipodystrophic phenotype characterized for the ∆FosB transgenic mice. Indeed, ENO2-∆FosB mice, expressing ∆FosB under the neuron-specific enolase promoter, show a decrease in adipose mass (Sabatakos et al., 2000). However, no effect on adipose tissue mass could be observed by adipocyte- or osteoblast-targeted expression of ∆FosB, indicating that the fat phenotype is not cell-autonomous or caused by increased bone mass, respectively (Kveiborg et al., 2004;

Rowe et al., 2009). Analysis of endocrine parameters of ENO2-∆FosB mice revealed no change in triglyceride levels, but reduced levels of insulin, free fatty acids and fed and fasting blood glucose levels. Additionally, increased glucose tolerance and enhanced insulin sensitivity were observed. Thus, the decrease in white adipose tissue observed in

ENO2-∆FosB mice is most likely caused by systemic metabolic alterations due to an increase in energy expenditure associated with an elevated fatty-acid oxidation in the muscle (Rowe et al., 2009).

In contrast, metabolic studies of Fra1 transgenic mice revealed only few marginal alterations.

As expected for a reduced amount of adipose tissue mass, Fra1 transgenic mice show a reduced level of leptin in the serum while the increased osteocalcin serum levels areconsistent with the increased bone mass. The adipocyte-derived leptin acts via the hypothalamus to regulate food intake and energy expenditure (Yadav et al., 2011). However, the reduced serum level of leptin could not alter the food intake of the Fra1tg mice. Furthermore, the effect of leptin deficiency to increase bone formation (Ducy et al., 2000) was shown to not cause the osteosclerotic phenotype of ∆FosB transgenic mice (Kveiborg et al., 2002).

As it was recently reported, that osteocalcin regulates insulin signaling by increasing pancreatic β-cell proliferation (Lee et al., 2007), insulin secretion and insulin sensitivity (Clemens and Karsenty, 2011), we expected that the increased osteocalcin levels would lead to alterations in the insulin signaling of Fra1tg mice. However, we observed normal levels of β-cell proliferation and of circulating insulin in the Fra1 transgenic mice, probably because of the increased Esp levels described to reduced bioavailability of osteocalcin (Ferron et al., 2010b). This observation of normal insulin levels and unchanged number and volume of pancreatic β-cell islets despite an increased osteocalcin level was also described for ∆^FosBtg mice (Rowe et al., 2009). Furthermore, fasting blood glucose levels in the Fra1tg mice were decreased, probably because of their increased sensitivity to insulin.

The only major metabolic alterations in Fra1tg mice are increased levels of circulating triglycerides and of non-esterified fatty acids. Further features typically associated with severe forms of lipodystrophy as described in typical mouse models for lipodystrophy like in A-ZIP/F transgenic mice (Moitra et al., 1998) or mice overexpressing a constitutively active form of SREBP-1c (Shimomura et al., 1998) are type 2 diabetes, insulin resistance as well as hepatic steatosis. This phenotype was not observed in Fra1 transgenic mice. It thus appears unlikely that the lipodystrophy of the Fra1 transgenic mice is due to systemic metabolic alterations.

4.3 Fra1 regulates adipocyte differentiation in a cell autonomous manner

In addition to systemic alterations, cell-autonomous defects can cause a reduction in adipose tissue mass. Adipogenic differentiation in vitro can be induced by stimulation with an adipogenic cocktail consisting of insulin, IBMX and dexamethasone. We could show in studies using a cell line overexpressing Fra1 and primary calvarial osteoblasts of Fra1tg mice that increased levels of Fra1 result in a reduced amount of mature adipocytes as shown by Oil

Red O staining and confirmed by the decreased expression of late adipogenic markers as C/ebpα^{, Ppar}γ2 and their transcriptional targets Ap2 and Glut4. The finding of a cell-autonomous defect is further supported by the similar results obtained in studies using an estradiol-inducible Fra1 in a cell culture system.

A shift to an increased osteoblastogenesis at the expense of adipogenesis was described in osteoblast differentiation experiments using cells overexpressing Fra2 (Bozec et al., 2010).

However, in cells overexpressing Fra1 as well as in the adipose tissue of Fra1tg mice, markers for lineage commitment (Pref1, Runx2, Osx1, Sox9 and MyoD) are not changed, nor were the markers for early adipocyte differentiation (C/ebpβ ^{and C/ebp}δ), supposing that the differentiation to mature adipocytes rather than adipocyte commitment is regulated by Fra1.

Consistent with an adipocyte maturation defect is the observation of a significant decreased adipocyte size in the visceral fat depots that develop in Fra1 transgenic mice indicating an increased number of immature adipocytes. Thus, we could demonstrate that Fra1 cell-autonomously regulates adipocyte maturation.

4.4 Fra1 directly regulates adipocyte differentiation by reducing C/ebpαααα expression While the expression of genes marking early steps of adipocyte development, as well as several factors involved in later differentiation stages are not affected in vivo, the level of C/ebpα, a key determinant of adipogenesis required and sufficient for terminal adipocyte differentiation of 3T3-L1 preadipocytes (Cao et al., 1991; Lin and Lane, 1994; Wu et al., 1998) as well as for the maintenance of the differentiated phenotype (Mandrup and Lane, 1997) is significantly reduced. Adipocytes lacking C/ebpα display a decreased insulin-regulated glucose uptake as C/EBPα regulates the insulin receptor and the glucose transporter Glut4 leading to reduced lipid accumulation. We could not detect any alterations in the insulin response in vitro by analyzing the phosphorylation of ERK and AKT. However, in differentiation experiments, we observed a reduced accumulation of lipid droplets and a decreased expression of Glut4 in the cells overexpressing Fra1. Consistent with a downregulation of C/ebpα are the experiments using the estradiol-inducible Fra1 cell line where activation of Fra1 blocks the early induction of C/ebpα by the adipogenic cocktail. We also found C/ebpα downregulated in other tissues with increased Fra1 expression, namely the liver, the muscle and the bone, further supporting a key role for C/ebpα repression by Fra1.

Another in vivo model for lipoatrophic diabetes is a transgenic mouse expressing the dominant-negative protein, A-ZIP/F under the control of the Fabp4 (Ap2) promoter. A-ZIP/F forms heterodimers with b-ZIP transcription factors via its leucine zipper and therefore inhibits the binding of C/EBP and AP-1 family members to DNA. Inhibition in activity of these transcription factors exclusively in adipose tissue leads to the ablation of white adipose tissue throughout development (Moitra et al., 1998). Studies overexpressing another dominant negative protein (A-C/EBP) in mice that only inhibits the C/EBP binding to promoter sequences leads to a lack of white adipose tissue in young and a reduced amount of WAT in older mice (Chatterjee et al., 2011). Due to the less severe phenotype of A-C/EBP compared to A-ZIP/F mice, it is proposed that both, AP-1 and C/EBP family members are important for growth and differentiation of white adipose tissue.

In agreement, some AP-1 members such as c-Fos, its viral homologue v-Fos as well as ∆FosB where shown to alter adipocyte differentiation cell-autonomously by regulating transcription factors of the C/EBP family. In particular, c-Fos was described to influence the activity of C/EBP family members in vitro as it can directly bind to C/EBPβ (Hsu et al., 1994). We could also show in initial experiments that overexpression of c-Fos in an adipogenic cell line reduced adipocyte differentiation and decreased the expression level of C/ebpα. However, overexpression of Fra2 in this cell line did not result in reduced adipocyte differentiation or in an altered expression of C/ebpα. Although the expression of adipogenic transcription factors in adipose tissue of ENO2-∆FosB mice was not changed (Rowe et al., 2009), the reduced adipogenic differentiation of primary osteoblast cultures overexpressing ∆FosB was reported to be caused by an interaction of ∆FosB with C/EBPβ resulting in the inhibition of transcriptional activation of C/ebpα (Kveiborg et al., 2004). In contrast, the viral homologue of c-Fos, v-Fos that inhibits adipocyte differentiation in vitro results in a decreased level of transcription as well as inhibited activity of C/EBPα while it does not affect the expression or activity of C/EBPβ (Abbott and Holt, 1997). Therefore, the regulatory role of v-Fos is not mediated by binding to C/EBPβ or C/EBPα. Performing co-immunoprecipitations, we were not able to demonstrate an interaction of Fra1 with C/EBPβ. In contrast, we could show by promoter studies and chromatin immunoprecipitation experiments that the reduced C/ebpα expression level in Fra1 transgenic mice is caused by the inhibition of transcription due to the binding of Fra1 to a putative binding site in the C/ebpα promoter.

4.5 Influence of the reduced amount of mature adipocytes on metabolism

Indeed, a decreased level of C/ebpα can explain the lipodystrophic phenotype observed in Fra1 transgenic animals as consistent observations were reported in C/ebpα knockout strains.

Although C/ebpα deficient mice die due to liver failure (Wang et al., 1995), studies using a C/ebpα knockout mouse with a transgenic expression of C/ebpα under the control of the albumin promoter in order to rescue the liver phenotype(Linhart et al., 2001) as well as a poly IC-inducible C/ebpα knockout mouse model where the postnatal deletion can be obtained in various especially C/EBPα-rich tissues (Yang et al., 2005) demonstrated, that a reduced level

Although C/ebpα deficient mice die due to liver failure (Wang et al., 1995), studies using a C/ebpα knockout mouse with a transgenic expression of C/ebpα under the control of the albumin promoter in order to rescue the liver phenotype(Linhart et al., 2001) as well as a poly IC-inducible C/ebpα knockout mouse model where the postnatal deletion can be obtained in various especially C/EBPα-rich tissues (Yang et al., 2005) demonstrated, that a reduced level