Tissue-Specific 5′ Heterogeneity of PPARα Transcripts and Their Differential Regulation by Leptin

The genes encoding nuclear receptors comprise multiple 5′untranslated exons, which give rise to several transcripts encoding the same protein, allowing tissue-specific regulation of expression. Both human and mouse peroxisome proliferator activated receptor (PPAR) α genes have multiple promoters, although their function is unknown. Here we have characterised the rat PPARα promoter region and have identified three alternative PPARα transcripts, which have different transcription start sites owing to the utilisation of distinct first exons. Moreover these alternative PPARα transcripts were differentially expressed between adipose tissue and liver. We show that while the major adipose (P1) and liver (P2) transcripts were both induced by dexamethasone, they were differentially regulated by the PPARα agonist, clofibric acid, and leptin. Leptin had no effect on the adipose-specific P1 transcript, but induced liver-specific P2 promoter activity via a STAT3/Sp1 mechanism. Moreover in Wistar rats, leptin treatment between postnatal day 3–13 led to an increase in P2 but not P1 transcription in adipose tissue which was sustained into adulthood. This suggests that the expression of the alternative PPARα transcripts are in part programmed by early life exposure to leptin leading to persistent change in adipose tissue fatty acid metabolism through specific activation of a quiescent PPARα promoter. Such complexity in the regulation of PPARα may allow the expression of PPARα to be finely regulated in response to environmental factors.


Introduction
PPARa is a ligand-activated transcription factor, which belongs to the nuclear hormone receptor superfamily [1,2]. PPARa plays a major role in lipid homeostasis by regulating the transcription of genes that encode the rate limiting enzymes in b oxidation, namely carnitine palmitoyl transferase (CPT-1) and acyl CoA oxidase (AOX) [3][4][5]. Targeted disruption of the PPARa gene in mice leads to lipid accumulation in the liver, impaired insulin secretion during fasting [6], increased adipose tissue mass and an increased incidence of liver tumours [7,8]. Consistent with these findings, agonists of PPARa have been used as effective hypolipidemic drugs [9].
PPARa is mainly expressed in tissues with high rates of fatty acid b oxidation such as liver, skeletal muscle, brown fat, heart, and kidneys [10;11]. Its expression is known to be regulated through the action of glucocorticoids [12], by HNF4, a major regulator of gluconeogenesis, [13], and by PPARa itself [14,15]. Adenoviral induced hyperleptinemia, which causes a rapid loss of body fat without a rise in plasma FFA or ketone bodies, has also been shown to increase the expression of PPARa and its target genes in white adipose tissue, a tissue where PPARa is not normally expressed. Conversely, the expression of PPARc2 and its associated genes involved in lipogenesis were reduced. However, the effects of hyperleptinemia were transient and two months after the concentration of leptin returned to normal, levels of PPARa expression decreased in adipose tissue and fat levels were regained [16][17][18][19]. This transient transformation of adipocytes from fat storing cells into fat burning cells via the induction of PPARa expression might suggest a novel approach for the treatment of obesity and a potential target for weight reduction.
There is also evidence that PPARa gene transcription can be programmed by environmental factors in early life [20]. For example the expression of PPARa is increased in the liver of offspring born to dams fed a protein restricted (PR) diet during pregnancy. The increase in PPARa expression in the PR offspring is accompanied by the increased expression of its target gene acyl-CoA oxidase (AOX) and an increase in levels of fatty acid betaoxidation [20 21]. In contrast, a 70% global dietary restriction during pregnancy induces a persistent decrease in PPARa expression in the liver of the adult offspring suggesting that different nutritional challenges during pregnancy induce distinct long term effects on PPARa expression [22]. In addition, there is evidence in the rat that PPARa expression is programmed by neonatal leptin exposure [22,23]. Neonatal leptin administration which reverses the phenotypic effects of maternal under nutrition by slowing neonatal weight gain, normalizing caloric intake and locomotor activity, induced a persistent increase in hepatic PPARa expression [22]. This is in contrast to the effects of hyperleptinaemia on PPARa expression in adipose tissue of adult rats where the increase in PPARa expression was not sustained after leptin administration was discontinued, suggesting that the timing of leptin exposure may determine the longevity of the response.
To date, little is known about of the molecular factors that mediate the tissue-specific regulation of PPARa, or its regulation by perinatal factors or leptin. The genes encoding nuclear receptors frequently comprise multiple 59untranslated exons giving rise to transcripts, which are expressed differentially between tissues. The human [24], mouse [25], and rat [26] PPARa gene promoters have only been partially characterised. The human PPARa gene is composed of 12 exons and generates 7 mRNA variants with different 59UTR exons. The 59UTR of the human PPARa gene contains 7 exons (exons A, 1A, B, 1B, 2A, 2B and the 59end of exon 3), while the coding exons are derived from the remainder of exon 3 and exons 4-8 [27][28][29]. The mouse PPARa gene is composed of 9 exons and generates 3 transcripts. Four exons comprise the 59UTR (exon 1a, 1b, 2 and the 59end of exon 3), the coding exons, like the human orthologue, are composed of the 39end of exon 3 and exons 4-8 [25]. The rat PPARa gene is comprised of 8 exons with 1 transcript reported to date. Three exons comprise the 59UTR exons, exon 1, 2 and the 59 portion of exon 3 [26]. Because of the complexities of the human and mouse genes, we hypothesised that the rat PPARa gene might also exhibit heterogeneity in the 59UTR leading to the synthesis of multiple transcripts. We have, therefore, characterised the promoter structure of the rat PPARa gene, and determined the pattern of expression between tissues and the extent to which alternative transcripts are differentially regulated by leptin. We show that rat PPARa gene gives rise to three PPARa mRNA variants, which differ from each other at the 59 end owing to the presence of unique first exons. The alternative PPARa transcripts were differentially expressed between adipose tissue and liver. We show that while the major adipose (P1) and liver (P2) transcripts were both induced by dexamethasone, they were differentially regulated by clofibric acid and leptin. Leptin had no effect on the adipose specific P1 transcript, but induced the liver specific P2 promoter activity via a STAT3/Sp1 mechanism. Moreover, consistent with leptin regulation of transcription from the P2 promoter in vitro, neonatal leptin treatment led to a persistent increase in transcription from the P2 promoter and not the P1 promoter in adipose tissue. Such 59 heterogeneity and complexity of regulation of PPARa may provide additional layers of control by which PPARa expression can be intricately regulated in response to hormones and early life environment in a tissue specific manner.

Ethics Statement
All animal work was approved by the Animal Ethics Committee of the University of Auckland (Approval N856) and complied with the New Zealand Code of Ethical Conduct for the care and use of animals for scientific purposes (Animal Welfare Act, 1999).

Animal Methods
A detailed description of the study design has been published previously [23]. Briefly, virgin Wistar rats (age 10065 days) were time-mated and fed a standard rat chow fed ad libitum throughout gestation. Litter size was adjusted to 8 pups at birth to ensure standardised nutrition until weaning. At postnatal day 3, female pups were randomized to receive either saline or recombinant rat leptin (rat leptin from Protein Laboratories, Rehovot, Israel) (2.5 mg/g/day) for 10 days by subcutaneous injection (n = 16 per group). Dams were fed ad libitum until offspring were weaned on day 22. Saline or leptin-treated offspring were weaned onto standard rat chow. This produced 8 groups of female rats (n = 8 per group). On postnatal day 170, rats were fasted overnight, and killed by halothane anaesthesia followed by decapitation. Liver and retroperitoneal adipose tissue was removed immediately, frozen in liquid nitrogen and stored at 280uC.

Analysis of mRNA Expression
Total PPARa, P1 and P2 PPARa, AOX and carnitine palmitoyltransferase (CPT)-1 mRNA concentrations were determined by real time RTPCR [30,31]. Briefly, total RNA was isolated from cells using TRIZOL reagent (InVitrogen, Paisley, Scotland, UK), and 1 mg was used as a template to prepare cDNA using 100 U Moloney-Murine Leukemia Virus reverse transcriptase. cDNA was amplified using real time PCR primers specific to total PPARa, P1 and P2 transcripts, AOX and CPT-1 ( Table 1). The reaction was performed in a total volume of 25 ml with SYBRH Green Jumpstart Ready Mix (Sigma, Poole, Dorset, UK) as described by the manufacturer. Samples were analyzed in duplicate and Ct values were normalised to ribosomal 18S RNA using the DDCt method [30].

RNA Ligase Mediated Rapid Amplification of cDNA Ends (RLM RACE)
RNA was amplified using a 59 RNA Ligase Mediated Rapid Amplification of cDNA Ends (RLM RACE) kit (Ambion) according to manufacturer's instructions. Briefly, 10 mg of total RNA from liver and adipose tissue was incubated with calf intestinal alkaline phosphatase to remove free 59 phosphates, the 59 CAP structure was then removed from the RNA with tobacco acid pyrophosphatase (TAP) and the TAP treated RNA ligated to the 59 RACE adapter and reverse transcribed into cDNA using random hexamers. To amplify the 59 region of PPARa mRNA, generated from the 59RLM RACE, nested gene specific reverse primers for PPARa were designed (inner primer 59 TGACT-GAGGAGGGGCTGGAA 39; outer gene specific primer 59 AGCCTTCACATGCGTGGACT 3) and used with the nested forward 59 RACE adaptor primers provided by the kit. Cycling conditions for both outer and nested PCR were initial denaturation of 94uC 3 minutes, followed by 35 cycles of 94uC 30 s, anneal at 60uC 30 sec, and extension at 72uC 1 minute 30 sec. Resulting PCR products were purified, cloned into pGem T-Easy (Promega) and sequenced. The sequence of the transcripts P1-3 was confirmed by sequencing multiple clones from liver and adipose tissue.

DNA Cloning
100 ng rat genomic DNA was used as a template for PCR amplification with primers designed to amplify 1-1.5 Kb of upstream sequence from the transcription start site of the P1, P2 and P3 promoters (P1 and P2 forward primer: 59ATGAGCT-CAGCAGCGTCCTGAGGCGTT 39; P1 reverse primer 59 ATAAGCTTACCTGAGGCTGCGCTCCG 39; P2 reverse primer 59 ATAA. GCTTGTGCCCTTCCTAGCGTGT 39; P3 forward: 59 ATAAGCTTGGAGTCTTCCTTCTGGTT 39; P3 reverse 59 ATACTCGAGTCTGCGTGGGTGTCTAAT 39). All primers contained either a SacI or XhoI restriction site at the 59 end of the forward primer, and a HindIII restriction site at the 59 end of the reverse primer to allow cloning of PCR fragments into the pGL3 Basic reporter vector (Promega). Cycling conditions for PCR were as follows; initial denaturation of 94uC 2 minutes, followed by 40 cycles of 95uC 45 s, anneal at a 60uC 45 sec, and extension at 72uC 5 minutes. Resulting PCR products were gel extracted, cloned into the pGL3 Basic reporter vector (Promega) and sequenced to confirm the presence of the insert. The Quik change method of mutagenesis (Stratagene) was used to create a P2 promoter construct with a mutated Sp1 site. Two complementary primers were designed using the QuikChangeH Primer Design Program (Stratagene, Texas, USA) to change the Sp1 consensus sequence GGGCGG to an EcoR1 restriction site GAATTC (Forward primer 59 GCCTCAGGTGCCCAGGAATTC-GAGGGCACGCGCGAGG 39 and reverse complement 59 CCTCGCGCGTGCCCTCGAATTCCTGGGCACCT-GAGGC 39). The primers (0.4 mM) were then annealed to 30 ng P2-pGL3 DNA and extended using Pfu polymerase. Cycling conditions for PCR were as follows; initial denaturation of 95uC 2 minutes, followed by 18 cycles of 95uC 30 s, anneal at a 60uC 1 min, and extension at 72uC 4 minutes, followed by a final extension at 72uC for 5 minutes. Resulting PCR products were digested with the restriction enzyme DpnI at 37uC for 1 hour and the product transformed into DH5a E.Coli. Clones were cut with EcoR1 and sequenced to confirm the presence of the mutation.

Cell Culture and Transfections
The human hepatoma cell line HepG2 (ECACC-Sigma Aldrich) was cultured in DMEM supplemented with 10% fetal bovine serum, 2 mM glutamine, 10 u/ml penicillin and 100 ug/ ml streptomycin. HepG2 cells were transfected using calcium phosphate [31,32], clofibric acid, dexamethasone, the Stat3 inhibitor PpYLKTK-mts and leptin were added immediately after transfection at the stated concentrations and luciferase activity measured 24 hrs later using the Luciferase Assay System (Promega). All transfections were performed in triplicate and values are expressed as luciferase activity per mg of protein.

Statistical Analysis
Statistical comparisons of mRNA expression and methylation levels between treatments relative to the untreated control were carried out using a Students unpaired t test. Statistical comparisons of luciferase activity levels between promoter constructs and treatments were determined by ANOVA followed by Bonferoni's post hoc analysis. All values are plotted as the mean 6SEM.

59RACE Analysis Identifies Distinct Tissue Specific 59leader Sequences in Rat PPARa mRNA
To investigate tissue specific 59heterogeneity of rat PPARa mRNA, 59RLM RACE PCR analysis was conducted on total RNA from adipose and liver tissue from adult Wistar rats. 59RLM RACE revealed 2 different sized PCR products from the liver (723 bp and 463 bp) but only 1 from adipose tissue (465 bp) ( Figure. 1A). The PCR products were cloned into the pGEM Teasy vector and the resulting clones sequenced and compared to the published genomic sequence of rat PPARa (Ensemble gene ID ENSRNOG00000021463) to identify the genomic location of the exons. The 3 transcripts differed from each other at the 59 end, and comparisons to the genomic sequence showed that they had different transcription start sites owing to the presence of unique first exons. The PPARa transcripts were subsequently termed P1-3 and a schematic diagram indicating the genomic location of the PPARa 59UTR exons and the organisation of the alternative 59UTRs is shown in Figure 1b.
Five exons were found to encode the PPARa 59UTR covering a genomic region of 42010 bps. Exons were named according to their genomic location. Exons 1A and 2A were novel, one exon was a modified version of exon 1 (now termed 1B), and the remainder were exactly as identified by Ensembl (exons 2 and 3). The 59UTR of the adipose specific transcript (P1) is encoded by exons 1A, 2 and 3. The transcription start site for this transcript is 257 bp upstream of the Ensembl published start site (http://www. ensembl.org/index.html), and gave rise to a distinct unique first exon of 60 bp termed exon 1A. The liver specific transcripts were termed P2 and P3. P2 comprises exon 1B, 2 and 3. Exon 1B (previously exon 1) is 67 bp longer than the original exon 1 at the 59 end resulting in a 318 bp exon. The P3 transcript contains exon 2A which is a 145 bp novel exon located several Kb downstream from exon 2, and has been termed exon 2A. This exon is spliced directly onto Ensembl exon 3, and is the only transcript that does not contain exon 2 ( Figure.1c). All these exons conform to the GT:AG splice site rule ( Table 2). Comparison of the rat, mouse and human PPARa 59UTR's revealed that the previously unidentified exon 1A, found in the rat 59UTR was homologous to the mouse exon 1A and human exon A. The 59 end of this exon is longer in the human and mouse corresponding exons, but all share the same 39 exon boundary. Exon 1B in rat is longer at the 59end than previously reported making it more homologous with the mouse and human corresponding exons. Rat exon 2A identified by 59RACE had no corresponding exon in the mouse 59UTR but was present in the human 59UTR. (Figure 1d). The P2 Promoter has Highest Activity in Liver HepG2 Cells To investigate whether the alternative PPARa exons are associated with promoter activity, the 59 region immediately upstream of the P1, P2 and P3 transcription start sites (from approx 21.5 kb to +50 bp) were cloned into the reporter vector pGl3basic and transfected into HepG2 cells. This resulted in two partially overlapping promoter regions for P1 and P2 and a distinct downstream promoter for P3 ( Figure. 2a). Both P1 and P2 promoters were active in HepG2 cells with P2 having the highest promoter activity. The P3 promoter region showed very low levels of activity, similar to that of the promoter-less pGL3 Basic vector (Figure 2b).
To determine whether the individual PPARa promoters are differentially regulated, the PPARa promoter-pGL3 luciferase reporter constructs (P1, P2 and P3) were transfected into HepG2 cells and treated for 24 hrs with clofibric acid and dexamethasone at a range of concentrations which have previously been reported to induce PPARa expression [12,15]. Clofibric acid treatment repressed P1 promoter activity at 60 mM (2.8 fold p = 0.001), but induced P2 promoter activity at 80 mM (3.19 fold p = 0.001) and 100 mM (5.7 fold p = 0.001) (Figure. 2C). Both P1 and P2 promoter activity was increased in the presence of dexamethasone. A significant increase in P1 promoter activity at both 1 mM dexamethasone (2.2 fold p = 0.01) and 10 mM dexamethasone (3.22 fold p = 0.001) was observed. With P2 a significant rise in promoter activity was observed with 10 mM dexamethasone (2 fold p = 0.001) (Figure. 2D). There was no effect of either dexamethasone or clofibric acid on P3 promoter activity.
Leptin Treatment Induces Transcription from the P2 but not the P1 or P3 Promoter To determine whether leptin could induce transcription from the alternative PPARa promoters, HepG2 cells were transfected with the P1, P2 and P3 promoter constructs and treated with increasing concentrations of leptin for 24 hrs. We found that P2 promoter activity was significantly increased in the presence of 500 ng/ml and 1000 ng/ml leptin. In contrast, leptin had no effect on either P1 or P3 promoter activity ( Figure. 3a). As leptin has been shown to modulate transcription through a STAT3 signalling pathway [33] we used the highly selective STAT3 inhibitor (PpYLKTK-mts) [34] to test the role of STAT3 in the induction of P2 transcription by leptin. We found that while the STAT3 inhibitor had no effect on P1 promoter activity, the STAT3 inhibitor blocked leptin activation of P2 promoter activity at 10 nM ( Figure. 3b).
Leptin modulation of transcription via STAT3 has been shown to occur either directly through the binding of STAT3 to its response element in the promoter of a gene or indirectly through a STAT3/Sp1 co-operative mechanism [35]. As the PPARa P2 promoter lacks a STAT3 binding site but contains a Sp1 site which is located within the region unique to the P2 promoter, we next investigated whether this Sp1 site was essential for leptin induction of P2 transcription. The Sp1 site was mutated from GGCGGG to the EcoR1 recognition site GAATTC. The wild type and mutated promoter constructs were then transfected into HepG2 cells and treated with 1000 ng/ml of leptin for 24 hrs. We found that the P2 promoter containing the mutated Sp1 site no longer responded to leptin, suggesting that this Sp1 site is essential for leptin activation of PPARa transcription (Figure.3C).

Differential Regulation of Alternative PPARa Transcripts by Neonatal Leptin
As there is evidence that hepatic PPARa gene transcription can be programmed by environmental factors in early life including leptin, we next investigated whether neonatal leptin treatment induced a persistent increase in PPARa expression in adipose tissue and whether the P1 and P2 transcripts were differentially affected. The expression of PPARa together with its target genes AOX and CPT-1 were examined in adipose tissue from PN170 rats treated with either saline or leptin (2.5 mg/g/day) by subcutaneous injection for 10 days from PN3-13. This dosage of leptin was used, as similar levels have been reported previously to induce leptin receptor signaling, alter neuropeptide expression [36,37] and reverse the metabolic features induced by maternal under nutrition. To analyze PPARa expression, primers were designed to anneal to the coding region of PPARa, a region common to all isoforms of PPARa, in order to measure total PPARa transcript levels, and to the specific P1 and P2 transcripts. Neonatal leptin administration led to an increase in total PPARa, AOX and CPT-1 mRNA in adipose tissue from D170 old rats compared to saline treated controls. However there was no effect of leptin treatment on the expression of the P1 transcript ( Figure. 4b), while neonatal leptin treatment significantly increased P2 specific transcripts in adipose tissue.
To determine whether this persistent increase in transcription from the P2 promoter in response to neonatal leptin treatment was due to altered DNA methylation, sodium bisulfite pyrosequencing was performed using genomic DNA extracted from adipose tissue from neonatal saline and leptin treated adult female rats. The analysis of the region (2336 to 2117 bp) immediately upstream of the PPARa P2 transcription start site (TSS) which contains the Sp1 response element in the P2 unique region showed that all CpGs within this region had a methylation level of below 10% regardless of treatment ( Figure 5.), although differences in methylation were observed between the leptin and saline treated offspring at CpGs 7,11,12 and 17 (CpG 7 (2.6% to 1.2%,), CpG 11 (1.6% to 0.2% ), CpG 12 (2.85% to 0.73%) and CpG 17 (4.3% to 1.8%,).

Discussion
The genomic organisation of the human and mouse PPARa genes indicate that PPARa, like other nuclear receptors, contains multiple 59UTR variants and promoter regions [27]. Previously rat PPARa was not known to have any mRNA variants and only three exons (exons 1, 2 and 3) were found to encode the 59 UTR. In this study, 59 RACE analysis revealed three PPARa mRNA variants, with 2 transcripts identified in liver (P2 and P3) and one in adipose (P1). These transcripts differed from each other at the 59 end, and comparisons to the genomic sequence showed that they had different transcription start sites owing to the presence of unique first exons. Five exons were found to encode the PPARa 59UTR, Exons 1A and 2A were novel, one exon was a modified version of exon 1 (now termed 1B), and the remainder were exactly as identified previously [26]. The identification of the additional 59UTR first exons in the rat PPARa promoter increases the homology between the rat, mouse and human PPARa promoter structures. For example, the 59 extended exon 1B in the rat P2 transcript and the novel exon 1A in the rat P1 transcript are both present in the human and mouse 59UTR. In addition, the rat P1 and P2 PPARa transcripts are very similar with those identified in the mouse, the rat P2 transcript being equivalent to the mouse variant 1 and the newly identified rat P1 transcript being similar to the mouse variant 2.
Since the 59 UTR can modulate RNA stability [38][39][40], translation efficiencies [41,42] as well as subcellular localisation [43,44], the use of alternative promoters to regulate the expression of untranslated first exons may add a further layer of control to the regulation of PPARa expression. For example, the presence of a long 59UTR, high GC content, secondary structure, uATGs and uORFs are all associated with reduced translational efficiency of the main ORF [45]. The relatively high GC content of the 59UTR was fairly consistent between the three PPARa alternative transcripts, but the length of the 59UTR varied from 184 bp for P3, to 444 bp for P2. This difference in length was reflected by differences in the minimum free energies of the transcripts calculated using Zucker RNA mfold 2.3 software which ranged from 274.65 Kcal/mol for P1, 2169.75 Kcal/mol for P2 and 256.15 Kcal/mol for P3 [46]. Secondary structures within the 59UTR with values of less than 30 Kcal/mol can be melted by the ribosome during the normal scanning process [47]. However, all three PPARa transcripts contain hairpins with stabilities of greater than this, indicating these transcripts may impede ribosomal movement and may be subjected to translational regulation.
The P1 and P2 transcripts did not contain ATG initiation codons in the sequence upstream of the previously reported ATG codon suggesting that they possess the same open reading frame as the previously reported transcript. The P3 transcript, however, contains four uATGs, 3 have adequate Kozak consensus sequences [47], but are followed by a termination codon. However, one ATG is in frame with the downstream ATG, and thus has the potential to produce a protein with a 30aa extended N terminal. PPARc2 has an extended N terminal of 30aa and 28aa in mice and humans, respectively, compared to the predominant PPARc1 protein. Experiments have shown that the PPARc2 N terminal extension comprises an N terminal ligand independent activation domain that mediates 5-6 times the activation of PPARc1 under ligand depleted conditions [48]. However, further studies are needed to determine whether the PPARa P3 transcripts contain a functional extended protein.
The analysis of the sequence upstream of the transcription start sites of the P1, P2 and P3 sites revealed that the P1/P2 promoters possess the characteristics of typical GC rich promoters common to nuclear hormone receptors such as the absence of TATA elements and the presence of CpG islands containing multiple Sp1 response elements. The P3 promoter, in contrast, did not contain any CpG islands or Sp1 response elements. Moreover, unlike the sequences upstream of the P1 and P2 TSS, which gave rise to high levels of promoter activity in HepG2 cells, the sequence upstream of the P3 start site was not active in HepG2 cells. This low activity may indicate that important regulatory elements outside the region cloned in this study are required for P3 expression or that the promoter is inactive in the absence of stimulatory factors that are not present in liver cell line HepG2.
It has been reported previously that PPARa transcription is induced by CFA [15], dexamethasone [12] and leptin [49]. Interestingly, the response of the adipose specific (P1) and liver specific (P2) promoters to these treatments differed. Both P1 and P2 promoters were up-regulated by dexamethasone, suggesting that glucocorticoids modulate PPARa expression through a sequence shared by the P1 and P2 promoters. Previous experiments have shown that GR can directly regulate PPARa expression [50] although the precise sequence was not identified. Matinspector analysis (www.genomatix.de) of the promoter region of PPARa did not reveal any glucocorticoid response elements, but a putative NF-1 binding site was identified in the sequence shared by both P1 and P2. NF-1 is a transcription factor that has been shown to mediate GR responsiveness [51]. In contrast, P1 and P2 promoters were differentially regulated by clofibric acid, a PPARa agonist and leptin, suggesting that leptin and clofibric acid mediate their effects through a sequence(s) that are unique to the P2 promoter. Autoregulation of gene expression is commonly found in nuclear receptors and ligands of PPARa have previously been reported to activate PPARa expression at the transcriptional level by binding to either a PPRE or DR1 motif [24], the latter of which is present within the unique region of the P2 promoter. Leptin has been suggested to regulate gene expression through the activation of Stat3 via a JAK signalling pathway [52][53][54], or in a promoter which lack a STAT3 response element, through a Stat3-Sp1 co-operative mechanism whereby Stat3 phosphorylates Sp1 which, in turn, facilitates Sp1 binding to its response elements [55]. The mechanism of leptin induction of P2 transcription involved both Stat3 and an Sp1 response element present in the sequence that is unique to the P2 promoter.
Leptin plays a critical role in maintaining energy balance [56,57] and also has an emerging role in growth and development [58]. Adipogenesis is associated with a marked elevation in serum leptin concentration which occurs between 4 to 10 days after birth in mice [59], while in rats, peak leptin concentration occurs at about 10 days after birth [60]. This increase in leptin has been shown to play a crucial neurotrophic role in the development of projections from the arcuate nucleus of the hypothalamus, which regulates food intake and adiposity [36,61]. Several reports indicate that the neonatal leptin surge is disturbed in its timing and/or magnitude by maternal undernutrition [61,62]. Moreover neonatal leptin administration from PN3-13 has been shown to reverse many of the features of metabolic programming induced by maternal undernutrition by slowing neonatal weight gain, normalizing caloric intake, locomotor activity, body weight and fat mass in adult offspring of undernourished mothers fed a HF diet [23]. The mechanism by which neonatal leptin treatment induces persistent changes in the regulation of fat mass was presumed to involve an effect on hypothalamic neurogenesis. Our findings however show that neonatal leptin administration induces a persistent increase in PPARa mRNA expression and its target genes AOX and CPT-1 in adipose tissue, suggesting that leptin may also have peripheral metabolic effects on adipose tissue. The increase in PPARa mRNA is consistent with the previous reports that hyperleptineamia induces PPARa expression in adipose tissue. Although interestingly while a transient increase in PPARa expression was observed in response to hyperleptinemia which disappeared when levels of leptin returned to normal, neonatal leptin treatment induced a stable increase in PPARa expression in adipose tissue which persisted into adulthood. This difference in the duration of the response to leptin may reflect the timing of the exposure during the life-course, as there is now growing evidence that early life exposures can induce long term effects on the metabolism and physiology of the offspring via epigenetic processes [63]. Consistent with the induction of P2, but not P1, promoter activity in cell culture experiments, neonatal leptin treatment differentially-regulated transcription from the PPARa P1 and P2 promoters in adipose tissue by persistently inducing P2, but not P1, transcription. A decrease in methylation at four CpGs within the P2 promoter was seen in the neonatal leptin treated compared to the saline treated animals. However given the low levels of methylation in this region, it is unclear whether such differences in methylation would mediate a switch in promoter usage although there is precedent for differences in methylation at low levels altering gene function [64].
In conclusion we show that the rat PPARa gene has multiple transcripts, which are expressed in a tissue specific manner and are differentially regulated by leptin. Moreover neonatal leptin exposure induces a persistent change in adipose tissue gene expression through specific activation of an otherwise quiescent PPARa promoter. It is possible that leptin may have similar effects in other species. For example, human PPARa also has multiple promoters and hence it is possible that long term modulation of PPARa activity, and hence lipid metabolism, by promoter switching as a result of leptin exposure in early life may also occur in humans. If the changes to gene expression in adipose tissue reported in the present study were to occur in humans, it may be expected that those who experience higher leptin exposure Figure 3. Leptin activates the PPARa P2 Promoter in HepG2 Cells P1, P2 and P3 promoter constructs were transfected into HepG2 cells and treated for 24 hrs with a vehicle control or an increasing concentration of leptin (0,50, 500, 1000 ng/ml). B) P1 and P2 promoter constructs were transfected into HepG2 cells and treated for 24 hrs with leptin (1000 ng/ml) and an increasing concentration of the STAT3 inhibitor PpYLKTK-mts (0,1,10 nM). C) Mutation of the Sp1 site blocks leptin activation of P2 promoter activity. P2 (P2-pGL3) and P2 promoter construct containing the mutated Sp1 response element (SP1M EcoRI-pGL3) was transfected into HepG2 cells and treated with leptin (1000 ng/ml) for 24 hrs. All values represent the mean of 6 independent experiments 6SEM. Statistical comparisons of luciferase activity between treatments relative to the untreated control were determined by ANOVA followed by Bonferroni post hoc analysis. (* p,0.05, ** p,0.001, *** p,0.001). doi:10.1371/journal.pone.0067483.g003 Figure 4. Neonatal leptin treatment leads to a persistent increase in PPARa P2 transcription. A) Neonatal leptin treatment induces a persistent increase in total PPARa transcripts, AOX and CPT-1 mRNA expression in retroperitoneal adipose. Values represent the mean 6 SEM relative to the saline treated control group. Statistical comparisons between the control (Con) and neonatal leptin treated (Leptin) groups were made using a Students unpaired t test. B) The expression of the P2 but not the P1 PPARa transcript is significantly altered by neonatal leptin treatment. Values represent the mean 6S EM relative to the saline treated control group. Statistical comparisons between the control (Con) and leptin treated (Leptin) groups were made using Students unpaired t test. (* p,0.05, ** p,0.001, *** p,0.001). doi:10.1371/journal.pone.0067483.g004 in early life would have greater capacity to meet the metabolic challenge of a high calorie diet after weaning, while a lower leptin exposure would result in a reduced capacity to regulate fatty acid deposition in adipose tissue. This may explain, at least in part, the observation that low umbilical cord blood leptin levels are associated with rapid postnatal weight gain [65]. One implication is that leptin exposure during specific periods in development may influence future risk of obesity.