Stable conditional expression and effect of C/EBPβ-LIP in adipocytes using the pSLIK system

Murine 3T3-L1 adipocytes are widely used as a cellular model of obesity. However, whereas transfection of 3T3-L1 preadipocytes is straightforward, ectopic gene expression in mature 3T3-L1 adipocytes has proved challenging. Here, we used the pSLIK vector system to generate stable doxycycline-inducible expression of the liver-enriched inhibitor protein isoform of CCAAT/enhancer binding protein β (CEPB (C/EBPβ-LIP)) in fully differentiated 3T3-L1 adipocytes. Because overexpression of C/EBPβ-LIP impairs adipocyte differentiation, the C/EBPβ-LIP construct was first integrated in 3T3-L1 preadipocytes but expression was induced only when adipocytes were fully differentiated. Increased C/EBPβ-LIP in mature adipocytes down-regulated C/EBPβ target genes including 11β-hydroxysteroid dehydrogenase type 1, phosphoenolpyruvate carboxykinase and fatty acid binding protein 4 but had no effect on asparagine synthetase, demonstrating that transcriptional down-regulation by C/EBPβ-LIP in 3T3-L1 adipocytes is not a general effect. Importantly, these genes were modulated in a similar manner in adipose tissue of mice with genetically increased C/EBPβ-LIP levels. The use of the pSLIK system to conditionally express transgenes in 3T3-L1 cells could be a valuable tool to dissect adipocyte physiology.


Introduction
First developed almost 40 years ago from Swiss mouse embryo cells, 3T3-L1 adipocytes (Otto & Lane 2005, Rosen & MacDougald 2006 have proved to be an invaluable tool, advancing our understanding of cellular mechanisms underlying obesity-related disorders. However, introduction of nucleic acids and, consequently, the manipulation of genes in fully differentiated 3T3-L1 adipocytes are difficult to achieve. pSLIK plasmids, encoding a tetracycline-regulated transcriptional unit in a lentivirus-based vector backbone, were originally designed to facilitate stable transduction of microRNA-like short hairpin RNAs into RAW264.7 cells (Shin et al. 2006) but have not been tested in adipocytes.
The transcription factor CCAAT/enhancer binding protein b (CEBPB (C/EBPb)) is produced as three major isoforms resulting from differential use of in-frame translation start sites or from proteolysis (Ossipow et al. 1993, Welm et al. 1999, Calkhoven et al. 2000; the 38 and 35 kDa liver-enriched activator protein isoforms (C/EBPb-LAP* and C/EBPb-LAP respectively) and the 20 kDa liver-enriched inhibitor protein (LIP; C/EBPb-LIP). C/EBPb-LIP contains the carboxyl-terminal dimerisation and DNA binding domains but lacks the amino-terminal activation domain of C/EBPb-LAP (Ramji & Foka 2002). Consequently, C/EBPb-LIP represses transcription by forming either homodimers or heterodimers with the C/EBPb-LAP isoforms or other C/EBPs (Descombes et al. 1990, Calkhoven et al. 2000 and manipulation of C/EBPb-LIP levels may change C/EBPb-LIP:LAP ratio. However, the majority of reports do not differentiate between C/EBPb isoforms, and genetic disruption of the encoding gene eliminates production of all isoforms. While C/EBPb is well recognised as a major player in adipocyte differentiation and metabolism, the role of the specific isoforms, and especially the C/EBPb-LIP isoform, is less clear. We have recently shown that genetic manipulation of levels of C/EBPb-LIP in vivo changes adipose tissue expression of the C/EBPb target gene that encodes the glucocorticoid amplifying enzyme 11b-hydroxysteroid dehydrogenase type 1 (Hsd11b1 (11b-HSD1); Seckl et al. 2004, Esteves et al. 2012. However, exploration of mechanisms in 3T3-L1 adipocytes is hindered by the impairment of preadipocyte differentiation by C/EBPb-LIP (Yeh et al. 1995, Payne et al. 2007, Abdou et al. 2011. Here, we have used the pSLIK system to generate stable inducible expression of C/EBPb-LIP in 3T3-L1 adipocytes. Increased levels of C/EBPb-LIP in 3T3-L1 adipocytes decreased the expression of the metabolically relevant C/EBPb target genes encoding phosphoenolpyruvate carboxykinase (Pck1 (PEPCK)), fatty acid binding protein 4 (Fabp4 (FABP4)) and asparagine synthetase in a manner similar to manipulation of C/EBPb-LIP in vivo.

Ethics statement
All animal experiments were conducted in strict accordance with internationally accepted standards of humane animal care following prior approval by the Institutional Animal Care and Use Committee in Berlin, Germany.

Cells and animals
Murine 3T3-L1 preadipocytes (American Type Culture Collection (ATCC)) were maintained in DMEM (Lonza, Cambridge, UK) supplemented with newborn calf serum (10%; Lonza), unless otherwise stated. Adult male C/EBPb mutant mice, heterozygous C/EBPb (C/L) , homozygous C/EBPb DuORF and respective wild-type (WT) littermate control mice (nZ6-7/group) were generated as described previously (Smink et al. 2009, Wethmar et al. 2010. Stable transfection of 3T3-L1 preadipocytes C/EBPb-LIP DNA sequence was excised from pMSV-C/EBPb (a gift from S L McKnight and W-C Yeh), ligated to an internal ribosome entry site (IRES) to allow for downstream GFP expression and the resulting C/EBPb-LIP-IRES construct was cloned into pMBA-266. C/EBPb-LIP and control (empty) pSLIK vectors (Shin et al. 2006) were generated by gateway recombination between a pEN (MBA-266; ATCC) entry vector harbouring C/EBPb-LIP or empty vector and the pSLIK-Neo destination vector (MBA-235; ATCC). 3T3-L1 preadipocytes were transfected with control or C/EBPb-LIP-expressing pSLIK vectors using Lipofectamine 2000 (Invitrogen), followed by R15-day selection of cells resistant to 1 mg/ml geneticin (Invitrogen). Transfection efficiency was high (data not shown) and a total pool of transfected preadipocytes was obtained. Both control and C/EBPb-LIP transfectants expressed GFP in the presence of doxycycline (DOX; 1 mg/ml; Sigma) and were further selected by cell sorting in the presence of DOX (FACSVantage, BD Biosciences, San Jose, CA, USA). Preadipocytes were maintained throughout in geneticin.

Differentiation of 3T3-L1 preadipocytes into adipocytes
3T3-L1 preadipocytes, cultured in the absence of DOX, harbouring C/EBPb-LIP or control empty vector, were induced to differentiate into mature adipocytes as described previously (Napolitano et al. 1998). Cells were grown to confluence, and 2 days later, they were induced to differentiate in DMEM supplemented with 10% FCS, 0.5 mM dexamethasone, 500 mM 3-isobutyl-1-methylxanthine and 5 mg/ml insulin for 2 days. Thereafter, 3T3-L1 cell differentiation continued in medium supplemented with 5 mg/ml insulin alone. Experiments were performed 8-12 days following induction of differentiation, unless otherwise stated. DOX (1 mg/ml) was added to the medium in the experiments designed to induce C/EBPb-LIP.

Oil Red O staining
3T3-L1 adipocytes were stained by Oil Red O as described previously (Lillie & Ashburn 1943) with modifications.  Cells were immersed in isopropanol (60%), followed by the staining solution (0.04% Oil Red O in 60% isopropanol) and ammonia (1%) and then visualised using a Zeiss Axioskop microscope. Quantification of Oil Red O staining density was performed with ImageJ (NIH, Bethesda, MA, USA).

RNA extraction and real-time PCR
Cells were harvested in TRIzol (Invitrogen) and RNA was extracted according to the manufacturer's protocol. Mouse subcutaneous adipose tissue was dissected and frozen (K70 8C) prior to RNA extraction using Qiagen RNeasy Lipid Tissue Mini kit (Qiagen). RNA (1 mg) was reverse transcribed using SuperScript III (Invitrogen), and specific mRNAs were quantified by real-time PCR using a LightCycler 480 (Roche) as described (Sai et al. 2008). Primer-probe sets were purchased from Applied Biosystems: C/EBPb (Mm00843434_s1), 11b-HSD1 (Mm00476182_m1), PEPCK (Mm01247058_m1), FABP4 (Mm00445878_m1) and asparagine synthetase (Mm 00803785_m1). TATA binding protein (TBP; Mm 00446973_m1) was used as internal control and did not change with treatment or between groups.

Statistical analysis
All data were analysed by Student's t-test or one-way ANOVA followed by post hoc least square difference test using SigmaStat 2.03 Statistical Software. Significance was set at P%0.05.

Conditional expression of C/EBPb-LIP in stably transfected 3T3-L1 preadipocytes and differentiated adipocytes
To generate a vector to conditionally express C/EBPb-LIP, the encoding cDNA was cloned and recombined in the MBA-266 and the pSLIK-Neo MBA-235 vectors, respectively, producing the pSLIK-C/EBPb-LIP vector in which a tetracycline response element drives expression of C/EBPb-LIP from a minimal CMV promoter in the presence of DOX (a Tet-ON system) and a contiguous IRES allows for GFP expression. Constitutive expression of the reverse Tet transactivator (rtTA3) and the neomycin resistance gene (Neo) is driven by the ubiquitin-c (Ubi-c) promoter (Fig. 1A).
In order to express C/EBPb-LIP in mature adipocytes, we first generated stably transfected 3T3-L1 preadipocytes that, in contrast to fully differentiated adipocytes, could be readily transfected with pSLIK-C/EBPb-LIP or the 'empty' vector. This was highly efficient allowing selection of transfectants with geneticin as a total pool of cells, avoiding clonal differences due to plasmid integration site. To ensure that only transfected cells were used in the subsequent experiments, preadipocytes were sorted by flow cytometry for GFP expression 24 h after incubation with DOX, with yields of 26 and 56% for pSLIK-C/EBPb-LIP and pSLIK-transfectants respectively (Fig. 1B). In the absence of DOX, only 0.9% of pSLIK-C/EBPb-LIP and 4.3% of pSLIK preadipocytes were GFP positive, demonstrating low spontaneous expression in the transfected preadipocytes. pSLIK-C/EBPb-LIP preadipocytes were then differentiated into mature adipocytes in DOX-free medium (to avoid the inhibitory effects of C/EBPb-LIP upon adipocyte differentiation) and subsequent DOX treatment robustly increased cellular C/EBPb-LIP levels in the mature adipocytes ( Fig. 2A and B). Following removal of DOX from the cell medium, C/EBPb-LIP decreased as expected, showing that it can be transiently induced (Fig. 2B). Importantly, in the absence of DOX, pSLIK-C/EBPb-LIP adipocytes showed comparable low endogenous levels of C/EBPb-LIP to control adipocytes harbouring the pSLIK empty vector ( Fig. 2A).

3T3-L1 preadipocytes stably transfected with pSLIK-C/EBPb-LIP differentiate normally into adipocytes
Consistent with previous data (Yeh et al. 1995, Payne et al. 2007, Abdou et al. 2011, adipocyte differentiation (demonstrated by Oil Red O staining of lipid) was markedly attenuated if C/EBPb-LIP expression was induced in 3T3-L1 pSLIK-C/EBPb-LIP preadipocytes prior to and during differentiation by DOX (Fig. 2C). In contrast, when pSLIK-C/EBPb-LIP preadipocytes were differentiated in the absence of DOX, both pSLIK control and pSLIK-C/EBPb-LIP preadipocytes differentiated normally into mature adipocytes. Induction of C/EBPb-LIP in fully differentiated adipocytes had no effect on their differentiation status as shown by quantification of Oil Red O staining (Fig. 2D).
Increased C/EBPb-LIP reduces expression of 11b-HSD1, PEPCK and FABP4 but not asparagine synthetase in 3T3-L1 mature adipocytes The transcriptional impact of increased C/EBPb-LIP in adipocytes was examined following induction with DOX only once 3T3-L1 adipocytes were fully differentiated. Concurrent with the increase in C/EBPb-LIP expression, and consequently C/EBPb-LIP:LAP ratio, DOX reduced 11b-HSD1 and PEPCK mRNA levels in 3T3-L1 pSLIK-C/EBPb-LIP adipocytes compared with control cells (Fig. 3A and B). FABP4 mRNA levels were decreased following 4 days of DOX treatment but asparagine synthetase mRNA levels were unaffected ( Fig. 3C and D).

Modulation of C/EBPb-LIP levels in adipose tissue in vivo changes the expression of PEPCK and FABP4
To test the relevance of the 3T3-L1 pSLIK-C/EBPb-LIP adipocyte model to the in vivo situation, we measured the levels of mRNA encoding PEPCK, FABP4 and asparagine synthetase in the adipose tissue of mice genetically engineered to produce altered levels of C/EBPb-LIP, which we have previously shown to regulate adipose tissue 11b-HSD1 mRNA levels (Esteves et al. 2012). C/EBPb (C/L) mice, with a 'knock-in' of C/EBPb-LIP (L allele) replacing the normal C/EBPb gene, have increased C/EBPb-LIP levels and therefore increased C/EBPb-LIP:LAP ratio (Smink et al. 2009). In contrast, in C/EBPb DuORF mice, deletion of the upstream open reading frame (DuORF allele) prevents translation of C/EBPb-LIP (Wethmar et al. 2010), resulting in decreased C/EBPb-LIP:LAP ratio (Esteves et al. 2012). In agreement with the results obtained in the 3T3-L1 pSLIK-C/EBPb-LIP adipocytes, levels of mRNA encoding PEPCK were lower in the adipose tissue of C/EBPb (C/L) mice compared with C/EBPb DuORF (Fig. 4A). However, there was no significant decrease in PEPCK mRNA levels in C/EBPb (C/L) mice compared with WT controls, suggesting a larger effect of C/EBPb-LIP Elevated C/EBPb-LIP represses expression of PEPCK and FABP4, but not asparagine synthetase in adipose tissue in vivo. Real-time PCR measurement of levels of mRNA encoding (A) PEPCK, (B) FABP4 and (C) asparagine synthetase in adipose tissue of wild-type control mice (Con, white bars), C/EBPb (C/L) (C/L, black bars) and C/EBPb DuORF (DuORF, grey bars). C/EBPb (C/L) mice are heterozygous for an allele of C/EBPb in which the normal gene has been replaced by C/EBPb-LIP (a 'knock-in'; Smink et al. 2009) and C/EBPb DuORF mice have a homozygous disruption of the upstream ORF (Wethmar et al. 2010). Adipose tissue mRNA levels, normalised to TBP, are expressed relative to levels in control mice (arbitrarily set to 100%) and are meanGS.E.M.; nZ6-9/group. *P%0.05. in vitro than in vivo or that compensation has occurred in vivo (e.g. through tissues other than adipose). FABP4 levels were lower in the adipose of C/EBPb (C/L) mice compared with control and C/EBPb DuORF mice, while asparagine synthetase was unchanged ( Fig. 4B and C). Taken together, these results show that the 3T3-L1 pSLIK-C/EBPb-LIP adipocytes mimic the in vivo regulation.

Discussion
The use of the pSLIK system to conditionally express C/EBPb-LIP in fully differentiated 3T3-L1 mature adipocytes allowed investigating its impact on the transcriptional levels of 11b-HSD1, PEPCK, FABP4 and asparagine synthetase in these cells. Although transfection of adipocytes with small RNAs (siRNA and shRNA) has recently been achieved using nanoparticles of virusderived amphipathic peptides (Li et al. 2007, Esteves et al. 2012, transfection of vectors harbouring expression cassettes remains a challenge. Electroporation, used in some studies (Thurmond et al. 1998), causes high levels of adipocyte death, is inefficient, and requires large amounts of DNA (or RNA). Adenovirus transduction allows expression of exogenous genes in 3T3-L1 adipocytes (Gnudi et al. 1997, Boizard et al. 1998, Duong et al. 2002, Carlotti et al. 2004) but does not produce stable expression and can induce immune responses. Many genes expressed in adipocytes, including 11b-HSD1 (Chapman et al. 2009), are immunoresponsive (Shoelson et al. 2006). Lentiviral vectors can be used to stably transduce adipocytes but require production of high titres of virus (Carlotti et al. 2004, Liao et al. 2006. The pSLIK vector used in the current study was originally designed as an intermediate in lentiviral construction (Shin et al. 2006) and combines all components required for expression (Tet-ON) and selection (Neo, GFP) in a single plasmid vector. This allows for the transfection of the pSLIK-C/EBPb-LIP vector into 3T3-L1 preadipocytes without the need for lentivirus and proves a highly efficient process. Here, we used DOX to restrict expression of C/EBPb-LIP to fully differentiated 3T3-L1 adipocytes, but ectopic transgene expression can be initiated at any stage of 3T3-L1 adipocyte differentiation or even induced transiently. Moreover, it may prove useful for expression of shRNA in differentiated adipocytes, to conditionally knockdown gene expression. Contrary to standard lentiviral transduction of adipocytes, here, all adipocytes harbour the gene of interest and expression is achieved soon after induction with DOX.
Elevation of C/EBPb-LIP in mature differentiated 3T3-L1 adipocytes concomitantly decreased PEPCK and 11b-HSD1 mRNA levels, suggesting direct regulation of 11b-HSD1 by C/EBPb-LIP in adipocytes. Indeed, both genes are bound by C/EBPb and this has been shown for 11b-HSD1 in several cell types (Arai et al. 2007, Sai et al. 2008, Yang et al. 2009), including adipocytes (Esteves et al. 2012), and for PEPCK in hepatoma cells (Park et al. 1993, Choudhury et al. 2011. PEPCK mRNA levels declined faster than other genes, possibly reflecting the short half-life of PEPCK (Tilghman et al. 1974) compared with 11b-HSD1 mRNA, for example (Balachandran et al. 2008). FABP4 mRNA levels were only affected after more prolonged elevation of C/EBPb-LIP, suggesting that either the mRNA is very stable or regulation is indirect. However, C/EBPb-LIP expression did not depress global transcription in adipocytes, as asparagine synthetase mRNA levels were unaffected by elevated adipocyte C/EBPb-LIP both in vitro and in vivo. This contrasts with the HepG2 hepatoma cells where asparagine synthase is regulated by the C/EBPb-LIP:LAP ratio (Siu et al. 2001) and might reflect differences in epigenetic regulation.
Finally, we note that this system may be of broader application to explore the role of other adipocyte relevant genes in fully differentiated adipocytes, providing insights into their effect on adipocyte physiology.

Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding
This work was supported by the Wellcome Trust Programme Grant (grant number WT083184 to J R S and K E C).