L-carnitine and PPARα-agonist fenofibrate are involved in the regulation of Carnitine Acetyltransferase (CrAT) mRNA levels in murine liver cells

Background The carnitine acetyltransferase (CrAT) is a mitochondrial matrix protein that directly influences intramitochondrial acetyl-CoA pools. Murine CrAT is encoded by a single gene located in the opposite orientation head to head to the PPP2R4 gene, sharing a very condensed bi-directional promoter. Since decreased CrAT expression is correlated with metabolic inflexibility and subsequent pathological consequences, our aim was to reveal and define possible activators of CrAT transcription in the normal embryonic murine liver cell line BNL CL. 2 and via which nuclear factors based on key metabolites mainly regulate hepatic expression of CrAT. Here we describe a functional characterization of the CrAT promoter region under conditions of L-carnitine deficiency and supplementation as well as fenofibrate induction in cell culture cells. Results The murine CrAT promoter displays some characteristics of a housekeeping gene: it lacks a TATA-box, is very GC-rich and harbors two Sp1 binding sites. Analysis of the promoter activity of CrAT by luciferase assays uncovered a L-carnitine sensitive region within −342 bp of the transcription start. Electrophoretic mobility shift and supershift assays proved the sequence element (−228/-222) to be an L-carnitine sensitive RXRα binding site, which also showed sensitivity to application of anti-PPARα and anti-PPARbp antibodies. In addition we analysed this specific RXRα/PPARα site by Southwestern Blotting technique and could pin down three protein factors binding to this promoter element. By qPCR we could quantify the nutrigenomic effect of L-carnitine itself and fenofibrate. Conclusions Our results indicate a cooperative interplay of L-carnitine and PPARα in transcriptional regulation of murine CrAT, which is of nutrigenomical relevance. We created experimental proof that the muCrAT gene clearly is a PPARα target. Both L-carnitine and fenofibrate are inducers of CrAT transcripts, but the important hyperlipidemic drug fenofibrate being a more potent one, as a consequence of its pharmacological interaction. Electronic supplementary material The online version of this article (doi:10.1186/1471-2164-15-514) contains supplementary material, which is available to authorized users.

Background L-carnitine (L-3-hydroxy-4-Ntrimethylaminobutyrate) is a low molecular ammonium compound, which is synthesized from the two essential amino acids lysine and methionine mostly in liver and kidney [1][2][3]. It is an important cofactor for the transport of long chain fatty acids across the mitochondrial membranes into the matrix where they can be broken down by ß-oxidation to acetyl-CoA to obtain energy via the citric cycle [4,5]. Due to its regulative functions on the acetyl-CoA pools and its transporting features of acyl groups, Lcarnitine covers also a key role in glucose metabolism and assists in fuel sensing [6]. An abnormal increase of intramitochondrial acetyl-CoA concentration in liver cells leads to increased gluconeogenesis, one aspect of diabetes [7]. High levels of acetyl-CoA have also been associated with abnormalities in skeletal muscles of diabetic patients, where insulin seems to be unable to mediate the switch from lipid to glucose metabolism leading to decreased glucose utilization [8].
Carnitine Acetyltransferase (CrAT) is a mitochondrial matrix enzyme, which transfers short acyl groups from acyl-CoA to L-carnitine resulting in an acetylcarnitine-ester [9]. Therefore it defines the equilibrium of acetyl-CoA (+free L-carnitine) and acetylcarnitine (+free CoA) [10]. Highest expression levels of CrAT are reached in muscle cells and testis [9]. Although expression levels are low in liver it exerts a significant metabolic function by regulating acetyl-CoA pools at the crossroads of anabolic and catabolic pathways [11]. Carnitine supplementation promotes CrAT-mediated acetylcarnitine efflux and improves metabolic outcome in obese rodents [12,13]. Furthermore the importance of CrAT has been recently underlined by the fact that muscle-specific knock-out mice showed increased metabolic inflexibility, meaning that they failed to adjust appropriately to mitochondrial fuel selection in response to nutritional cues [14]. Another very supportive fact for our study was that NIDDM patients showed decreased levels of CrAT mRNA levels [14]. These findings render CrAT to be an interesting pharmacological target for treatment of NIDDM. L-carnitine itself could be a pharmacological tool since supplementation after artificially induced L-carnitine deficiency induces CrAT expression in a moderate way [15].
Murine CrAT gene (GeneID 12908) had been mapped on chromosome 2 next to the protein phosphatase 2A, regulatory subunit B (PR 53) gene (PPP2R4). No detailed promoter studies have been published so far, but PPARα plays a key role in CrAT transcription control [16,17]. So we hypothesized that a cooperative interplay between L-carnitine and PPARα positively influences CrAT expression.

L-carnitine and PPARα induce murine CrAT expression
The L-carnitine levels in the FCS after dialysis against phosphate buffered saline for 48 h dropped significantly from 36 μM to 16 μM (see Table 1). In a direct consequence the intracellular L-carnitine levels after cultivation in dialyzed FCS were reduced more than 70% and after L-carnitine supplementation could be restored to more than 70% of the original level of 973 μM L-carnitine (see Table 2). Under these cell culture conditions we performed our L-carnitine deprivation/supplementation studies.
Our first experiments examined the regulation of the murine CrAT gene by L-carnitine and fenofibrate. CrAT mRNA levels were increased by L-carnitine supplementation (for 4 h) after artificially induced Lcarnitine deficiency ( Figure 1A). Additionally we tested the influence of 80 μM L-carnitine on muCrAT mRNA levels up to 48 hours after supplementation. After 4 hours the first peak of CrAT mRNA levels was detected (1.65 fold increase). A second application rate of 80 μM L-carnitine was added after 15 h and lead to even higher CrAT mRNA levels. (2.23 fold after 18 h) ( Figure 1B). Also the PPARα-agonist fenofibrate induced muCrAT up to 11-fold (with 40 μM fenofibrate) ( Figure 1C).

L-carnitine raises PPARα presence in the nucleus
PPARα protein levels in the nucleus increase constantly after L-carnitine supplementation ( Figure 2A). Already 4 hours after L-carnitine supplementation slightly increased levels of PPARα levels are detectable in the nucleus (1.2 fold) and gets more distinct after longer supplementation periods (1.5 fold after 15 hours and 1.7 fold after 24 hours. The western blot presented in Figure 2B shows the almost even induction pattern of CrAT protein levels in TIB-73 cells cultivated in fetal calf serum, dialyzed fetal calf serum and after L-carnitine (80 μM) and/or fenofibrate (40 μM) supplementation. The cultivation in dialyzed fetal calf serum slightly reduces the TIB73 CrAT protein levels and subsequent L-carnitine does not really influence the steady state CrAT protein. Only fenofibrate is able to increase it distinctly.

L-carnitine increases CrAT promoter activity
Three luciferase constructs (mCrAT-1, mCrAT-2 and mCrAT-3) were designed containing different regions of the 5′flanking region of the murine CrAT gene ( Figure 3). All three constructs were transfected separately into TIB-73 murine liver cells and promoter activity was measured after 4 h ( Figure 4). mCrAT-2 shows higher promoter activity than mCrAT-3 at all 3 different L-carnitine supplementation levels. Within this construct a RXRα, two Sp1, a CAC-binding protein and a PPARα binding site were identified using the Transfac database via the online-tools TESS and PATCH.

Analysis of the murine CrAT 5′ flanking sequence
The murine CrAT promoter and its putative cisregulatory elements for nuclear factors are depicted in Figure 3. Analysis of the 5′-flanking region of exon 1 revealed several putative transcription factor-binding sites. Most importantly a RXRα element was found at −228 bp to −222 bp. The CrAT promoter is a TATA-less one with two SP1 binding sites. Further on the 5′ promoter region harbors binding sites for HES-1, CREB, CAC-binding protein and at last a PPARα site within the first exon.
L-carnitine raises binding affinity of nuclear extracts to RXRα probeanti-PPARα antibody abrogates this DNA-protein formation To gain more insight into the binding affinities of nuclear extracts to different probes representing the putative transcription factor binding sites we performed several EMSAs. Nuclear extracts of TIB-73 cells were prepared after 4 h of L-carnitine supplementation at different concentrations. EMSA analysis of the DNA probe representing the RXRα binding site revealed increased affinity of nuclear extracts to this ODN with rising L-carnitine concentrations. Additional administration of anti-PPARα antibody (sc-9000X) directly to the EMSA reaction mix lead to weakening of the band shift signal at 40 μM L-carnitine and nearly to extinction of the signal at 80 μM L-carnitine ( Figure 5A, B). Based on this result we concluded that Lcarnitine influences transcriptional activation via PPARα. Further on we performed Southwestern Blotting to gain more information about the complexity of the protein factors binding to this specific RXRα promoter element. We received three distinct signals at 51 kDa, 70 kDa and 145 kDa. A search in the Transfac database via molwsearch (www.gene-regulation.com) presented LXRα and PPARα as candidate factors for the 51 kDa signal, c-Myb and cMyc as putative factors for the 70 kDa band and Evi-1 for the 145 kDa signal ( Figure 6A, B).

Discussion
No detailed promoter study of the murine CrAT promoter has been published so far. Basically the CrAT gene   exhibits the typical characteristics of a housekeeping promoter: it harbours no TATA box, is GC rich and has two Sp1 binding sites. The distance to the transcriptional start of the opposing PPP2R4, which is encoded on the complementary strand, is only 586 bp. This leads to the reasonable postulation that this promoter very likely is a bi-directional one. For the human PPP2R4 promoter it could be shown that Yin-yang 1 (YY-1) is essential for core promoter activity and that it is a p53 target gene [18,19]. By applying TESS we could find three YY-elements, two at positions −52 to −44 and +27 to +35 relative to the PPP2R4 transcription start and a third one already in the first exon of the PPP2R4 gene. These putative YY-binding sites very likely represent the murine equivalents to the human promoter ( Figure 3). One aim of our work was to identify inducers of transcriptional activation of CrAT. As depicted in Figure 1 L-carnitine and fenofibrate are such transcriptional activators. L-carnitine induces CrAT and other members of the "acylcarnitine shuttle system" like CPT1a and b, as well as CPT2 transcription levels in the human system similar to mice [15]. Beyond that, in a parallel chip-screen study performed by our lab, we observed that hundred of genes throughout the whole genome  are transcriptionally in-or decreased by L-carnitine, underlining the importance of this metabolite [20]. In case of the murine CrAT we observed a rather moderate increase of mRNA levels (up to 1.8 fold) after 4 h of L-carnitine supplementation following artificially induced L-carnitine deficiency. The above-mentioned opposing PPP2R4 gene is transcriptionally induced by L-carnitine and fenofibrate very similar to the CrAT gene as shown in the human liver cell line Hep G2 (see Additional file 1: Figure S1), which is another argument for the bi-directionality of the promoter.
The important hyperlipidemic drug fenofibrate is a much more potent inducer of CrAT transcription levels (up to 11-fold after 3 hours of fenofibrate treatment). But no firm indications exist for a PPRE element in CrAT promoter from bioinformatical analysis. This was also  assumed for CPT1a and CPT2 [10], but for the latter PPREs could finally be defined [21]. CPT2 hosts a special PPRE, namely only one half proportion with perfect consensus sequence (TGACCT) [22]. Our results undoubtedly prove that muCrAT is a PPARα target, as it has also been indicated in experiments with PPARα knock-out mice [16,17]. By reportergene assays we were able to define an Lcarnitine sensitive region within 342 nt upstream the transcription start ( Figure 3B and 4). Within this sequence many different putative binding sites for nuclear factors were predicted in silico. Our band shift experiments clearly revealed one RXRα element to be sensitive to Lcarnitine supplementation.
Based on our data we propose the sequence TGACC CCGTGACGG at −238 upstream the transcription start to be a possible L-carnitine sensitive RXRα/PPARα binding site. Band shift assays performed with an oligo desoxynucleotide carrying only the 5′-half of this sequence element showed sensitivity to L-carnitine supplementation. At least the sequence box TGACC is present in the CrAT L-carnitine binding site, therefore we propose in analogy to the CPT2 gene this sequence element being a further imperfect but functional PPRE site. Increasing Lcarnitine levels lead to enhanced binding affinity of nuclear extracts. Interestingly the application of anti-PPARα to the band shift reaction mix lead to a mitigation of the band shift signal. With increasing L-carnitine supplementation levels the band shift signal almost disappeared, indicating the interference of anti-PPARα with the DNA-protein complex. Such fading out effects have been observed before and were accepted as an experimental proof for a supershift [23]. To rule out that this explicit attenuation effect was an unspecific one, caused by the use of antibodies in general, Supershift assays were performed with antibodies against several candidates of nuclear receptors (see Additional file 1: Figure S2). These validation experiments did not result in any significant changes of the DNA-protein formation except for PPARα. Therefore this effect had to be qualified as a specific interaction.
In addition to this effect we were able to reveal an additional close connection between the PPAR-system and Lcarnitine. Transcription levels of PPAR-binding protein, PPARbp, are also inducible by L-carnitine. Chip screen experiments as well as quantitative PCR also provided experimental proof for this effect [20]. Li JL et al. found experimental proof for an interaction of PPARα and Lcarnitine as a protective response to oxidative stress [24]. This finding provides additional support to our observations that L-carnitine shows cooperativeness with the PPAR-system. This supposition is strengthened by the results of the Western blot depicted in Figure 2A: Lcarnitine supplementation leads to increased levels of PPARα in the nucleus. Obviously increased L-carnitine levels foster the translocation of PPARα into the nucleus. It is of fundamental nutrigenomic importance that RXRα/ PPARα heterodimers are positively regulating transcription rates of fatty acid degradation genes as counterparts to LXRα and SREBP1-c, factors which mostly induce anabolic acting fatty acid synthesis genes [25,26]. Carnitine Acetyltransferase is a central regulator of intramitochondrial acetyl-CoA pools. The latter holds a prominent position in intermediary metabolism as the universal end product of fatty acid, glucose and amino acid oxidation [12,14]. Acetyl-CoA excess leads to blockage of the TCA cycle and subsequently to accumulation of short carbon compounds, which are further on used for gluconeogenesis in liver [7,27]. This is besides insulin resistance another factor of the pathogenesis of NIDDM [28].
Recently the pivotal role of CrAT in acetyl-CoA metabolism was confirmed since experiments with CrAT M−/− knock-out mice showed that absence of CrAT leads to accumulation of medium and long chain acylcarnitines in muscles and subsequently to glucose intolerance via overloaded mitochondria [14]. These findings indicate a possible positive effect of L-carnitine and fenofibrate in the regulation of glucose homeostasis via direct transcriptional activation of CrAT, the first due to a metabolic interaction and the latter based on its pharmacology. Singular and combined administration of these two compounds should be evaluated in controlled clinical trials in order to verify our findings in vivo.
The interaction of L-carnitine and PPARα as we could show is of transcriptional importance. Therefore the growth condition where this collaborative interplay is able to trigger the genome deserves deeper investigation. L-carnitine could already be identified as a nutritional modulator of glucocorticoid receptor functions [29]. In our electrophoretic mobility assays we also observed a slight induction at one of the glucocorticoid responsive elements (GRE) at position −421 to −396 of the CrAT promoter, after Lcarnitine supplementation, but to a much lesser extend as at the RXRα/PPARα binding site (see Additional file 1: Figure S3). Thus possible interactions of L-carnitine with other nuclear receptors like PPARα and GRE and its molecular basis are worthwhile being investigated in the future.

Conclusions
Based on our investigations we could present a complete study of the murine CrAT promoter and provide strong evidence for a cooperative interplay of L-carnitine and PPARα for its transcriptional regulation, which undoubtedly is of nutrigenomical importance. Reportergene and electrophoretic mobility assays located a L-carnitine sensitive region within 342 nt upstream the transcription start, that contains one RXRα/PPARα element which is directly responding to L-carnitine supplementation. Super shift assays performed with a polyclonal antibody demonstrated that the muCrAT gene is a PPARα target. These evidences are strengthened by our findings that transcription levels of the PPAR-binding protein (PPARbp) are also inducible by L-carnitine. A direct comparison of L-carnitine with the drug fenofibrate revealed, both are inducers of CrAT transcripts, with one implication that the hyperlipidemic drug fenofibrate exerted a more pronounced effect, based on its pharmacological interaction. Within the investigated CrAT promoter sequence a variety of different putative binding sites for nuclear factors were predicted in silico and verified by experimental approaches. Thus possible interactions of L-carnitine with other nuclear receptors like PPARα and GRE and its molecular basis are worthwhile being investigated in the future. We also could append facts for the bidirectional function of the CrAT promoter in conjunction with the opposing PPP2R4 gene.

Cell culture
The normal embryonic murine liver cell line BNL CL. 2 (ATCC® TIB-73™) was grown at 37°C, 7.5% CO 2 in DMEM supplemented with 10% fetal calf serum (Sigma Aldrich) and 1% antibiotics (30 mg/l penicillin, 50 mg/l streptomycin sulphate). For the experiments cells were either kept in DMEM containing 10%FCS or dialyzed 10%FCS (dialysis was performed against 1xPBS for 48 h with five buffer changes). L-carnitine (LONZA) and fenofibrate (Lannacher Heilmittel) were added to the culture medium to obtain defined final concentrations as indicated in the experiments.
DNA (0,4 μg pCMV-ßgal and 2 μg of the respective luciferase reporter construct) was mixed with 150 mM NaCl in a total volume of 100 μl for each well. Polyethylenimine (PEI Sigma, 18,5 μl for each well) was mixed with 150 mM NaCl in a total volume of 100 μl for each well. The solution was slowly added to the DNA solution to ensure proper formation of DNA/ PEI-complexes, mixed and left at room temperature for 20 min. 200 μl DNA/PEI mix was added to each well and the cells were kept at serum free conditions for~4 hours.
Afterwards, the cells were washed with DMEM and were allowed to grow in DMEM supplemented with 10% dialyzed FCS and antibiotics for 24 hours. After an additional medium change and cultivation for further 4 hours the cells were treated with various concentrations of Lcarnitine (10 μM, 40 μM, 80 μM) and incubated with and without L-carnitine for additional 4 or 24 hours.

L-carnitine assay
To determine the intracellular L-carnitine levels induced by our cell culture conditions as well as the L-carnitine concentrations in the fetal calf serum (dialyzed/normal) employed in the cell culture the L-carnitine colorimetric/ fluorometric assay kit (BioVision) was used. TIB-73 were grown under the conditions described above and 10 6 cells were homogenized in 100 μl assay buffer exactly as described in the user manual provided by the vendor. In the case of the fetal calf serum 2 μl and 10 μl aliquots were subjected the same assay procedure. Since the enzyme based assay is very sensitive to contaminating protein levels all samples were treated with 4 M HClO 4 and neutralyzed with 2 M KOH. After centrifugation to remove the precipitated proteins samples were drawn and correlated with an L-carnitine standard curve.
After the given time points (described above) the cells were harvested, transferred to a microfuge tube, followed by two rounds of freezethaw circles with liquid nitrogen versus 37°C to ensure complete lysis. The cell lysate was centrifuged at 15.000 rpm, 4°C for 3 min and the supernatant was transferred to a fresh microfuge tube. 200 μl cell extract was mixed with 16 μl luciferin (Applichem), 4 μl ATP (0,1 M) and 1 μl DTT (1 M), vortexed and the luciferase activity in each lysate was measured by a Berthold LB953 luminometer.
In silico analysis of mouse CrAT promoter and 5′UTR for putative nuclear factor binding sites To identify putative binding sites of transcriptional active nuclear factors in the mouse CrAT promoter 850 kb of the 5′ flanking region of mouse CrAT from −700 to +150 relative to the transcription start were analysed using the online tools TESS (http://www.cbil.upenn.edu/cgi-bin/ tess/tess) and PATCH (http://www.gene-regulation.com/ cgi-bin/pub/programs/patch/bin/patch.cgi).

Western blot analysis
PPARα protein levels from nuclear extracts were analysed by Western blot. Nuclear extracts were prepared after an established protocol as described above. Aliquots containing 25 μg of nuclear extract were loaded on to a 10% SDS polyacrylamid gel. After electrophoresis proteins were transferred to a nitrocellulose membrane. The membrane was probed with antibodies against PPARα (sc-9000, Santa Cruz Biotechnology) and GAPDH (sc-47724, Santa Cruz Biotechnology). In the case of the CrAT western blot 10 μg whole cell protein extracts were separated on 12% SDS polyacrylamid gels and probed with CrAT antibody (Abcam ab91478) and GAPDH (sc-47724, Santa Cruz Biotechnology). The membranes were then processed with HRP conjugated secondary antibodies specific for the appropriate species. Proteins were visualized with Western Lightning ECL kit (Perkin Elmer). The statistical analyses were carried by SPSS software (IBM) using the t-test subroutine for independent samples (student's T test) to calculate the p-values given in the figure legends. In the case of Figure 4 we performed a Kruskal-Wallis test with GraphPad Prism (GraphPad Software Inc.).

Southwestern analysis
For southwestern analysis nuclear extracts and oligonucleotides were prepared as described above. Protein samples were separated by SDS-PAGE and then transferred to nitrocellulose membrane at 110 V for 90 min at 4°C. Blotted proteins were renatured in 1xTNED buffer (10 mM TRIS pH 7.5, 50 mM NaCl, 0,1 mM EDTA, 1 mM DTT) with 5% skim milk for 24 h. The membrane was incubated with γ-32 P-ATP marked oligonucleotides in 1xTNED with 2.5% skim milk for another 24 h. Blot was exposed to chemiluminescent sensitive films visualized via autoradiography. Protein size was determined and further on analyzed via molwsearch using Transfac database (http://www.gene-regulation.com/cgi-bin/pub/ programs/molwsearch/molwsearch.cgi).

Additional file
Additional file 1: Figure S1. qPCR of human PPP2R4 from the human liver cell line HepG2: (A) cells were treated 24 h with dialyzed FCS and supplemented afterwards for 4 hours with L-carnitine (80 μM). Values show mean SD, n = 4, ***p < 0.001 vs. DMEM + 10% FCS. (B) Cells were grown in DMEM + 10%FCS for 24 hours and afterwards treated with fenofibrate (10-40 μM) for four hours. Values represent means ± SD (n = 4). Supplemented cultures were compared to physiological control (DMEM + 10% FCS) ***p < 0.001. Methods: Human liver cell line HepG2 was treated as described in the methods section for TIB-73 cells. For quantitative PCR given protocols as described in the methods sections were followed. Following primers were used: PPP2R4 Ps: 5′CAAGAGTGAAAGGCGAGACG3′, Pas:5′CCATGTCTGGAACTGTGTGG′; ß-actin Ps: 5′GATGAGTATGCCTGCCGT GTG3′, Pas: 5′TCAACTGGTCTCAA-GTCAGTG3′. Figure S2. Electrophoretic mobility super shift assay with anti-RXRα and anti-PPARγ: Nuclear extracts from TIB-73 cells supplemented with increasing concentrations of Lcarnitine were incubated with -32P-labeled oligonucleotides representing the RXRα-binding site with anti-RXRα and anti-PPAR as indicated. No mitigation effect was observable as seen with anti-PPARα antibodies in Figure 5. Figure S3. Electrophoretic mobility shift assay of one of the CrAT promoter GR-binding sites: Nuclear extracts from TIB-73 cells supplemented with increasing concentrations of L-carnitine were incubated with a -32Plabeled oligonucleotide representing the GR-binding site sense: 5′ GTCA ACAGTTGTGTTCTCCTGCCATTC3′.