Identification of Regulatory Elements That Control PPARγ Expression in Adipocyte Progenitors

Adipose tissue renewal and obesity-driven expansion of fat cell number are dependent on proliferation and differentiation of adipose progenitors that reside in the vasculature that develops in coordination with adipose depots. The transcriptional events that regulate commitment of progenitors to the adipose lineage are poorly understood. Because expression of the nuclear receptor PPARγ defines the adipose lineage, isolation of elements that control PPARγ expression in adipose precursors may lead to discovery of transcriptional regulators of early adipocyte determination. Here, we describe the identification and validation in transgenic mice of 5 highly conserved non-coding sequences from the PPARγ locus that can drive expression of a reporter gene in a manner that recapitulates the tissue-specific pattern of PPARγ expression. Surprisingly, these 5 elements appear to control PPARγ expression in adipocyte precursors that are associated with the vasculature of adipose depots, but not in mature adipocytes. Characterization of these five PPARγ regulatory sequences may enable isolation of the transcription factors that bind these cis elements and provide insight into the molecular regulation of adipose tissue expansion in normal and pathological states.


Introduction
Obesity is a risk factor in multiple diseases, including type 2 diabetes, cardiovascular disease, and cancer [1]. The emergence of obesity as a grave public health problem has focused interest on adipose tissue and fat cell function. Adipose tissue is an important metabolic and endocrine organ that is critical for energy balance and insulin sensitivity [2]. White adipose tissue (WAT) serves as a storage site for excess energy, while brown adipose tissue (BAT) dissipates energy to generate heat. Adipocytes also secrete adipokines (e.g., leptin, adiponectin) that regulate multiple physiologic processes, including appetite and glucose homeostasis [3,4]. In obesity, the ability of adipocytes to store lipids, dispose of glucose, and secrete adipokines is compromised. Obesity-driven adipocyte dysfunction is intimately linked to the development of systemic insulin resistance and type 2 diabetes [5,6]. In response to a chronic energy imbalance, the number and the size of adipocytes increases to retain excess energy. Eventually, adipose tissue expansion is not sufficient to store surplus fatty acids and adipocyte-released lipids deposit in tissues such as liver and muscle where they dampen insulin action. A better understanding of how adipose tissue develops and expands is thus critical to devise new avenues to treat obesity and its associated complications.
Adipose tissue mass can expand throughout life [7]. Under normal circumstances, approximately 10% of human adipocytes are renewed each year [8]. Obesity can increase the rate of adipocyte proliferation and differentiation [9]. Because mature adipocytes are non-dividing, renewing or increasing the number of fat cells relies on the differentiation of proliferating adipose progenitors that are found in the stromal-vascular fraction of adipose depots [10]. Environmental stimulation (e.g., chronic high-fat feeding) induces adipose stem cells in this niche to commit to the preadipocyte lineage, which can then give rise to terminally differentiated adipocytes. While recent studies have identified cellsurface markers that allow isolation of progenitor cells with adipogenic potential [11][12][13], and lineage tracing analyses have shown that adipogenic precursors reside in the mural cell compartment of the adipose vasculature [13][14][15][16], little is known about the transcriptional events that prompt adipose progenitors to commit to the preadipocyte lineage (determination). Recent work has associated the zinc-finger protein Zfp423, its paralog Zfp521, and the factors Zfp467, Tcf7L1, and Ebf1 with preadipocyte determination, but the transcriptional regulation of early adipose commitment remains poorly understood [17][18][19][20][21].
In contrast, the major components of the transcriptional cascade that brings about preadipocyte to adipocyte terminal differentiation have been identified [22,23]. PPARc, a lipidregulated transcription factor of the nuclear receptor family, is the master regulator of adipocyte terminal differentiation. Expression of PPARc is required for fat cell formation [24][25][26]. Although PPARc expression was thought to be associated primarily with differentiated adipocytes, a recent lineage tracing analysis using PPARc-reporter strains has revealed the existence of immature PPARc-expressing cells that reside in the adipose vasculature [14]. This population of PPARc-expressing proliferating cells gives rise to the vast majority of adipocytes in the mature fat pad.
Because PPARc expression is the defining feature of the adipose lineage, greater understanding of the transcription factors that control PPARc expression in adipose progenitors may shed insight into the dynamics of adipose tissue expansion in normal and pathological states. In contrast to the attention that has been paid to pharmacologic activation of PPARc, much less is known about the regulation of PPARc expression, particularly during the early stages of adipose commitment. As a first step to discern the transcription factors that control the initial phases of adipocyte determination, we have carried out a comparative genomic analysis to identify conserved sequence elements in the 59-flanking region of the PPARc locus that may be responsible for its pattern Kb of sequence upstream of the PPARc2 transcriptional start site reveals 5 elements that are highly conserved across multiple mammalian species (indicated as CS1 to 5 in the UCSC genome browser schematic). (B) X-gal staining of subcutaneous (SubQ), visceral (Visc), and retroperitoneal (Retro) WAT, brown adipose tissue (BAT), and other organs from wild type, PPARc (+/2), and PPARc CS1-5_LacZ line 1 transgenic mice (6 weeks). Note that CS1 to 5 drive reporter expression in a similar tissue-specific pattern to that of LacZ expressed from the endogenous PPARc locus. (C) X-gal staining of wild type, PPARc (+/2), and PPARc CS1-5_LacZ line 1 and 7 embryos at E14.5. doi:10.1371/journal.pone.0072511.g001 of expression. We have isolated five elements that appear to be sufficient to recapitulate the tissue-specific pattern of PPARc expression in vivo. These 5 non-coding DNA sequences from the 59-flanking region of the PPARc locus can drive expression of a reporter in adipose progenitors localized in the vasculature of white and brown fat pads. Interestingly, the ability of these sequences to activate transcription decreases as adipocyte differentiation proceeds. These findings indicate that these 5 cis elements behave as enhancers that control PPARc expression at the earliest stages of adipocyte determination, but not during terminal differentiation.

Isolation of Conserved Genomic Regions that Regulate PPARc Expression
The tissue-specific pattern of expression of genes is thought to be primarily due to the action of enhancers, non-coding DNA sequences that are often located far away from the basal promoter of the gene whose transcription they control [27][28][29]. Comparison of sequence conservation across species can be useful to identify non-coding DNA sequences that behave as functional enhancers in vivo [30]. There are two major isoforms transcribed from the PPARc locus, PPARc1 and c2 [31]. Each isoform is transcribed from a different promoter and alternative exon usage gives rise to two proteins that differ in the N-terminus. The PPARc2 mRNA is expressed almost exclusively in adipose depots, while PPARc1 exhibits a broader pattern of expression. Since our intention was to identify genetic elements that regulate PPARc expression at the earliest stages of adipogenesis, we carried out a comparative genomic analysis of a 100 Kb genomic region upstream of the PPARc2 transcriptional start site (TSS) that includes the PPARc1 promoter. Five evolutionarily conserved sequences (CS1 to CS5), representing putative regulatory elements, were identified based on alignment of 30 mammalian species using the MULTIZ algorithm (Fig. 1A). These elements range in size from 357 to 991 bp and are .80% identical across mammals, similar conservation to that of PPARc exons, suggesting that they could contain the regulatory sequences that control PPARc expression. CS1, CS2, and CS3 are located between exon A2 and exon B (211 to 232 Kb from the PPARc2 TSS), while CS4 and CS5 are located upstream of the PPARc1 exon A1 and far from the PPARc2 TSS (,279 Kb) (exact genomic locations shown in Supplemental Table 1).
To evaluate the extent to which these sequences control PPARc expression in vivo (i.e. behave as enhancer elements that dictate tissue-specific PPARc expression), we cloned all 5 elements together into an Hsp68-LacZ reporter vector to generate PPARc CS1-5_Hsp68-LacZ transgenic mice (referred hereafter as PPARc CS1-5_LacZ; Supplemental Fig. 1). The Hsp68 minimal promoter was chosen because this is a widely used basal promoter for in vivo enhancer analysis [30]. To establish if these 5 conserved elements are sufficient to drive expression of the LacZ reporter in a pattern similar to that of endogenous PPARc, we analyzed LacZ expression by X-gal staining in tissues of 5 independently-derived PPARc CS1-5_LacZ transgenic lines. One line (line 1) showed very strong X-gal staining in brown fat and in all white adipose depots (Fig. 1B). To check the specificity of reporter expression, we analyzed LacZ expression in skeletal muscle, liver, spleen, and pancreas and found no X-gal staining in these organs ( Fig. 1B and Supplemental Fig. 5). The pattern of X-gal staining in this PPARc CS1-5_LacZ transgenic line mirrored that seen in PPARc (+/2) heterozygous null mice in which an allele of PPARc was targeted by an in-frame insertion of a neomycin-LacZ construct (b-geo) into exon 2 of PPARc [24]. Analysis of LacZ expression across tissues by RT-qPCR and Western Blot indicated that the PPARc CS1-5_LacZ transgene was expressed in a similar pattern to that of endogenous PPARc (Fig. 2), with greatest expression of mRNA and protein in fat depots, and lower levels in selected other organs. This adipose-enriched pattern of expression of the transgene suggested that these 5 conserved sequences contain most of the regulatory elements necessary for tissue-specific PPARc expression. Two additional PPARc CS1-5_LacZ transgenic lines (lines 6 and 7) showed an identical, but weaker, pattern of X-gal staining and LacZ mRNA expression, indicating that the pattern of transgene expression we observed is not the consequence of integration effects.
During mouse development, PPARc expression correlates with the appearance of the interscapular brown fat depot at embryonic day 14.5 (E14.5), and with the emergence of adipose progenitor cells that can be detected at postnatal day 1 and are associated with the vasculature of what becomes the white adipose tissue depots [14,24]. To examine the extent to which the 5 conserved PPARc sequences regulate PPARc expression during development, we evaluated expression of the PPARc CS1-5_LacZ transgene at E14.5 (Fig. 1C). X-gal staining in control PPARc (+/2) embryos showed that, as reported, PPARc expression at this stage is only evident in the brown fat depot. Line 7 PPARc CS1-5_LacZ transgenic embryos showed weak, but clearly detectable X-gal staining that was spatially restricted to the location of the BAT depot. Line 1 transgenic embryos showed a strong pattern of X-gal staining that encompassed the BAT depot, but broadened beyond the staining pattern in control PPARc (+/2) embryos. In this line, the one with highest transgene expression, the X-gal stain was additionally associated with what appeared to be the vascular network that underlies the epidermis, perhaps an indication that the transgene is active in cells that could form the basis of the subcutaneous fat layer that supports the dermis (Fig. 1C and Supplemental Fig. S2). It is probable that this additional X-gal stain is not detected in PPARc (+/2) embryos because these embryos express only one copy of the LacZ reporter, while line 1 embryos are likely to have multiple copies of the reporter transgene, as is often the case in transgenic lines. Together with our results in adult tissues, these data indicate that the 5 conserved sequences we have identified play an important role in the tissuespecific regulation of PPARc expression in vivo.

Transcription Driven by Conserved PPARc Sequences 1 to 5 Decreases during Adipocyte Differentiation
Adipocytes develop in coordination with the vasculature, which supplies oxygen, nutrients, and endocrine factors, and provides a niche for pericyte-derived adipocyte progenitors [16,32]. To explore the compartment(s) within adipose depots where CS1 to 5 PPARc sequences are transcriptionally active, we measured LacZ and PPARc mRNA expression after separation of the stromalvascular (SV) and adipocyte fractions of WAT and BAT depots of wild type, PPARc (+/2), and transgenic PPARc CS1-5_ LacZ mice. As expected, LacZ expression driven by the entire endogenous PPARc locus (as in PPARc [+/2] mice) was predominantly associated with the differentiated adipocyte compartment, particularly in WAT depots (Fig. 3A). In contrast, we found that the CS1-5 PPARc sequences activated LacZ mRNA . Endogenous PPARc mRNA was detected in the SV fraction, but was significantly enriched in the adipocyte compartment, with no differences among mice of different genotypes (Fig. 3B). The quality of our fractions was verified by measuring expression of adiponectin, a mature adipocyte marker that could only be detected in the adipocyte fraction (Fig. 3C). These results indicate that PPARc CS 1 to 5 are transcriptionally active only in the SV fraction that contains adipocyte progenitors and preadipocytes, as well as other cells that do not contribute to the adipose lineage. Interestingly, expression of transgenic LacZ, but not that derived from the endogenous locus (PPARc [+/2] mice), was consistently higher in BAT compared to WAT (Fig. 3A), perhaps a reflection of the larger vascular network that is present in BAT.
To evaluate in detail the behavior of the PPARc CS1-5_LacZ transgene during the course of adipocyte differentiation, we conditionally immortalized SV cells isolated from the subcutaneous white (inguinal) and brown fat depots of transgenic mice and measured LacZ expression at various time points after induction of differentiation. Cells derived from transgenic animals differentiated into adipocytes with normal frequency (Fig. 3D and Supplemental Fig. S4). Intriguingly, LacZ mRNA expression in cells isolated from transgenic mice decreased dramatically upon the induction of adipocyte differentiation and remained low in maturing adipocytes (Fig. 3E). In contrast, PPARc expression was highly induced during differentiation (Fig. 3F). The opposing pattern of CS1-5-driven LacZ expression relative to that of endogenous PPARc, and its association with the SV fraction rather than with the adipocyte compartment, indicated that these sequences could be responsible primarily for expression of PPARc in the progenitors that give rise to the adipocyte lineage.

PPARc Conserved Sequences 1 to 5 are Transcriptionally Active in Adipose Precursors that Line the Vasculature of White and Brown Adipose Tissue
Lineage tracing studies have taken advantage of the high stability of b-galactosidase protein to show that PPARc is expressed in proliferating cells that reside in the adipose vasculature and give rise to mature adipocytes [14]. To explore the possibility that the CS1-5 elements could be responsible for PPARc expression in adipocyte progenitors, we examined sections of X-gal stained WAT and BAT depots from PPARc CS1-5_LacZ and PPARc (+/2) mice. In PPARc (+/2) fat pads, the X-gal stain was associated with mature adipocytes in all depots (Fig. 4A-C), with a few LacZ positive cells along some capillaries. In contrast, in PPARc CS1-5_ LacZ transgenic fat pads the X-gal stain was detected in some mature adipocytes, but it was significantly more prominent along the vasculature of both WAT and BAT fat pads (Fig. 4D-F). The staining was particularly strong in sections of transgenic interscapular BAT, where the stain outlined many of the vessels present in this tissue (Fig. 4F). LacZ staining was present, not only in small capillaries, but also in a perivascular pattern in some larger size vessels. Images taken at higher magnification ( Fig. 4G-I) revealed the presence of LacZ positive cells in the mural cell compartment of the vasculature, where adipocyte progenitors reside. No LacZ positive cells were detected, in association with the vasculature or otherwise, in X-gal stained sections of other tissues such as liver, skeletal muscle, and spleen (Supplemental Fig. 5). Immunohistochemical analysis of X-gal stained PPARc CS1-5_ LacZ transgenic adipose tissue revealed that LacZ expression overlapped primarily with that of mural/endothelial/adipose progenitor cell markers (e.g., CD29 [10,11], Smooth Muscle Actin [14]), but not with perilipin, a marker of mature fat cells (Fig. 5). These cytochemical and X-gal staining patterns, together with our data showing that conserved PPARc sequences are active in the SV fraction but inactive in mature adipocytes, indicate that these elements play an important role in controlling PPARc expression in adipocyte precursors, but not in mature fat cells.
Here, we have shown that 5 small non-coding sequences from the 59-flanking region of the PPARc locus are sufficient to activate PPARc expression in adipocyte precursors of both, the white and the brown lineage. This is the first description of a transcriptional cassette that can direct gene expression to adipocyte progenitors. The extent to which all five elements are required to control PPARc expression in vivo is presently under investigation. Interestingly, these five elements do not appear to play a role in the dramatic induction of PPARc expression that accompanies adipocyte terminal differentiation. Members of the C/EBP family are thought to stimulate and sustain PPARc expression at this later stage [22,33]. Further characterization of these five PPARc regulatory sequences may enable isolation of the trans-acting factors that bind these cis elements. Identification of the transcription factors that induce PPARc expression through these elements in adipose progenitors will provide insight into the molecular regulation of normal adipose tissue turnover, its expansion in obesity, and perhaps its absence in lipodystrophies that remain to be associated with a molecular determinant.

Ethics Statement
Animal experiments in this work were limited to the harvest of tissues from humanely euthanized animals. The number of animals used was kept to the minimum necessary to insure data quality. The Scripps Research Institute's Institutional Animal Care and Use Committee approved all procedures.

Generation of Transgenic Mice
Conserved non-coding sequence elements from the PPARc locus were cloned by PCR. A fragment containing all 5 conserved elements (CS1-5) in the endogenous orientation was cloned into the Hsp68-LacZ vector [30] to generate the PPARc CS1-5_Hsp68-LacZ reporter. Transgenic mice were generated by pronuclear microinjection into C57BL/6 single cell embryos. Founders were identified by PCR (LacZ primers: LacZ-F: 59-TTTCCATGTTGCCACTCGC-39; LacZ-R: 59-AACGGCTTGCCGTTCAGCA-39) and bred to C57BL/6 mice to establish lines. F1 transgenics 5 to 7 weeks old were used for analysis unless otherwise indicated. All procedures were approved by the TSRI IACUC.

Tissue Fractionation
Minced WAT and BAT depots were digested in isolation buffer (123 mM NaCl, 5 mM KCl, 1.3 mM CaCl 2 , 5 mM glucose, 0.1 M HEPES, pH 7.4, 4% BSA) containing 1.5 mg/mL collagenase A at 37uC for 1 hr. Digested tissues were passed through a 100 mm mesh, and the flow-through separated into SV and adipocyte fractions by centrifugation.

Gene Expression and Protein Analysis
RNA was isolated using the NucleoSpin 96 RNA kit (Macherey-Nagel). Taqman-based real-time qPCR was performed using the Superscript III One-Step RT-PCR mix (Life Technologies).

Immunohistochemistry
Paraffin sections of X-gal stained subcutaneous WAT (10 mm) were deparaffinized and rehydrated, permeabilized with PBS-Triton X-100 0.5% for 15 min and antigen retrieval was performed with PBS-SDS 1% solution for 10 min at room temperature. Blocking was performed in 10% FBS-PBS/Triton X-100 0.1% for 1 hr at room temperature. Anti-Integrin b1 (CD29, BD Pharmingen #558741), anti-Perilipin (Cell Signaling #9349) or anti-Smooth Muscle Actin (SMA, Dako #M0851) antibodies were applied (1:200) overnight at 4uC in 5% FBS-PBS/ Triton X-100 0.1% followed by 3 washes in PBS-Triton X-100 0.1%. Incubation with secondary antibodies (1:500, AlexaFluor 488 donkey anti-rat IgG #A-21208, AlexaFluor 546 Donkey antirabbit IgG #A11035, AlexaFluor 546 donkey anti-mouse IgG #A10036) was performed in 5% FBS-PBS/Triton X-100 0.1% at room temperature for 1 hr; sections were then washed 3 times in PBS-Triton X-100 0.1%. Cell nuclei were stained with DAPI solution for 10 min at room temperature. Sections were then washed and mounted. Images were taken at 4X magnification and processed with Adobe Photoshop and ImageJ software. Figure S1 Schematic of the PPARc CS1-5_LacZ reporter transgene. Mouse PPARc conserved elements 1 to 5 (CS1 to CS5, white boxes) shown in their respective genomic positions (PPARc exons are numbered and shown in black) were cloned by PCR into a vector containing a minimal Hsp68 promoter upstream of the LacZ gene. The transgene shown in the right was excised from this vector and microinjected into C57BL/6 single-cell embryos to generate multiple lines of PPARc CS1-5_LacZ reporter mice. (TIF) Figure S2 Enlarged views of X-gal stained PPARc (+/2) and PPARc CS1-5_LacZ transgenic embryos at E14.5. Note that the stain in line 1 extends beyond the BAT depot to what appear to be capillaries in the dermis. A similar, but weaker, vasculature-like stain is also evident in line 7 embryos.