In this study we have identified the target genes of sterol regulatory element binding protein (SREBP)-1a and SREBP-1c in primary cultures of human skeletal muscle cells, using adenoviral vectors expressing the mature nuclear form of human SREBP-1a or SREBP-1c combined with oligonucleotide microarrays. Overexpression of SREBP-1a led to significant changes in the expression of 1,315 genes (655 upregulated and 660 downregulated), whereas overexpression of SREBP-1c modified the mRNA level of 514 genes (310 upregulated and 204 downregulated). Gene ontology analysis indicated that in human muscle cells SREBP-1a and -1c are involved in the regulation of a large number of genes that are at the crossroads of different functional pathways, several of which are not directly connected with cholesterol and lipid metabolism. Six hundred fifty-two of all genes identified to be differentially regulated on SREBP overexpression had a sterol regulatory element (SRE) motif in their promoter sequences. Among these, 429 were specifically regulated by SREBP-1a, 69 by SREBP-1c, and 154 by both 1a and 1c. Because both isoforms recognize the same binding motif, we determined whether some of these functional differences could depend on the environment of the SRE motifs in the promoters. Results from promoter analysis showed that different combinations of transcription factor binding sites around the SRE binding motifs may determine regulatory networks of transcription that could explain the superposition of lipid and cholesterol metabolism with various other pathways involved in adaptive responses to stress like hypoxia and heat shock, or involvement in the immune response.
- transcription factor
- DNA binding motif
- skeletal muscle
sterol regulatory element binding proteins (SREBPs) belong to the basic-helix-loop-helix leucine zipper (bHLHLZ) family of DNA-binding proteins that play fundamental roles in both cholesterol and fatty acid metabolism (36). Three isoforms have been identified in mammalian tissues that vary in structure, regulation, and function. SREBP-1a and SREBP-1c (originally cloned as ADD1; Ref. 47) are protein products of alternative promoter usage of the srebf1 gene, whereas SREBP-2 is transcribed from a different gene.
These unique members of the bHLHLZ family of transcriptional regulatory proteins can be distinguished from other family members by two characteristics. First, they are synthesized as precursors. The amino-terminal half of the precursor is clipped out of the membrane in two steps responding to regulatory cues that signal the need for increased cellular cholesterol (2). The released amino-terminal protein activates the expression of SREBP target genes. The second distinguishing feature of the SREBPs is that they have an atypical tyrosine residue in the conserved basic DNA binding domain (24), which allows binding to the direct repeat sterol regulatory element (SRE) 5′-TCACNCCAC-3′ (50). SREBP-1a and SREBP-1c differ only in the first exon, which encodes a longer acidic transactivation domain for SREBP-1a, thus conferring a stronger activation potential compared with SREBP-1c (48).
Most of what is known regarding the function of SREBP-1 has come from studies in tissues of high lipogenic capacity, such as liver and adipose tissue. It has been demonstrated that the ratios of the relative levels of SREBP-1a and SREBP-1c mRNA vary in these tissues. In liver and adipose tissue, SREBP-1c mRNA is nine- and threefold more abundant, respectively, than SREBP-1a (44). Because SREBP-1c mRNA is the predominant isoform in adult liver and adipose tissue, it is likely to be the key protein involved in SREBP-1-dependent processes in these tissues. Indeed, in the liver insulin upregulates SREBP-1c expression, and this isoform has been proposed to mediate the transcriptional effects of insulin on lipogenic and glycolytic enzymes (13).
We (12, 16) and others (17) have demonstrated that SREBP-1a and -1c are also expressed in skeletal muscle. In this tissue, SREBP-1c is also regulated by insulin (8, 17). Unlike adipose tissue and liver, skeletal muscle is not regarded as being highly lipogenic. Nonetheless, muscle triglyceride stores are substantial and vary under a number of metabolic conditions (1, 32, 45). Skeletal muscle is also a major site for fatty acid oxidation (38). Until now, the role of SREBP-1a and -1c in skeletal muscle has remained unclear, because few of their target genes have been identified.
While identification of new SREBP target genes dramatically increased with microarray analysis of livers from transgenic mice overexpressing SREBP-1a or SREBP-2 (20), of isolated rat islets transfected with SREBP-1c (9), or of fibroblasts transfected with SREBP-1a (23), no information on the targets for SREBP-1a and/or SREBP-1c in human muscle cells is presently available at the transcriptome level. Interestingly, in these previous studies, the overexpression of SREBP isoforms led to changes in the expression of a large number of genes not directly related to fatty acid metabolism and lipogenesis. Thus we suspected that SREBP-1 might also be involved in the regulation of important cellular functions in skeletal muscle. Moreover, because two SREBP-1 isoforms exist with different potencies, we postulated that some of their target genes might be different. To validate this hypothesis, and to define the molecular function of SREBP-1 in skeletal muscle, we performed microarray analysis on human primary myotubes overexpressing either SREBP-1a or SREBP-1c and identified clusters of genes regulated by SREBP-1a and/or SREBP-1c.
MATERIALS AND METHODS
Primary cultures of human skeletal muscle cells.
Differentiated myotubes were prepared from three different skeletal muscle biopsies according to a procedure previously described in detail (7, 39). Briefly, skeletal muscle tissue samples were obtained from three healthy lean volunteers who participated in a global study on insulin action on gene expression. The experimental protocol was approved by the Ethical Committees of the Hospices Civils de Lyon (agreement number 2003/039/125A). After differentiation, cells showed polynucleated status and expressed specific markers of human skeletal muscle such as creatine kinase, sarcomeric α-actin, and myosin. There were no apparent differences among cultured skeletal muscle cells from the three control subjects (2 men and 1 woman).
Generation of recombinant adenoviruses encoding mature forms of human SREBP-1a and SREBP-1c.
The construction of the expression vectors encoding the mature nuclear forms of human SREBP-1a (pCMV-hSREBP-1a) and SREBP-1c (pCMV-hSREBP-1c) was described previously (8). Recombinant adenoviral genomes carrying the mature forms of human SREBP-1a and SREBP-1c were generated by homologous recombination and amplified as described previously (6). Skeletal muscle myotubes were grown in six-well plates. After 7 days of differentiation, myotubes were infected for 48 h with recombinant adenovirus expressing Renilla (control), nuclear SREBP-1a, or nuclear SREBP-1c. We used 5 × 107 infection units of recombinant adenovirus for a 3-cm well of differentiated cells.
Total RNA preparation and amplification.
Total RNA was extracted from infected human myotubes with TRIzol reagent (Invitrogen Life Technologies). RNA integrity was determined with the Agilent 2100 Bioanalyzer and RNA 6000 labChip Kit (Agilent Technologies, Massy, France). Five hundred nanograms of total RNA was amplified with the Amino Allyl MessageAmp II aRNA kit (Ambion, Austin, TX) according to the manufacturer's instructions.
Probe labeling and microarray hybridization.
Fluorescent probes were synthesized by chemical coupling of 5 μg of aminoallyl aRNA with cyanine (Cy)3 or Cy5 dyes (GE Healthcare Biosciences, Orsay, France). After purification with an RNeasy Mini Kit (Qiagen, Courtaboeuf, France), probes were fragmented with 25× RNA Fragmentation Reagents (Agilent Technologies) and hybridized with 2× Agilent Hybridization Buffer (Agilent Technologies) to RNG/MRC human pangenomic microarrays (26) in an Agilent oven at 62°C for 16 h. Microarrays were washed and scanned with a Genepix 4000B scanner (Axon Instruments, Foster City, CA).
The oligonucleotide microarrays produced by the French Genopole Network (RNG) consisted of 25,342 oligonucleotides of 50-mers printed on glass slides. TIFF images were analyzed with Genepix Pro 4.0 software (Axon Instruments). Signal intensities were log-transformed, and normalization was performed by the intensity-dependent Lowess method. To compare results from the different experiments, data from each slide were normalized in log space to have a mean of zero with Cluster 3.0 software. Only spots with recorded data on the six slides (3 for SREBP-1a and 3 for SREBP-1c) were selected for further analysis. With these selection criteria, 12,825 spots were retrieved. Data were analyzed with the one-class significance analysis of microarrays (SAM) procedure (49).
Microarray data are available in the GEO database under number GSE10918 with the link http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=xbwnjumqwscyknu&acc=GSE10918.
Quantitation of mRNAs with real-time RT-PCR.
First-strand cDNAs were synthesized from 500 ng of total RNA in the presence of 100 U of Superscript II (Invitrogen, Eragny, France) and a mixture of random hexamers and oligo(dT) primers (Promega, Charbonnières, France). Real-time PCR assays were performed with a LightCycler (Roche Diagnostics, Meylan, France). A list of the PCR primers is available on request ().
Protein-DNA complexes were fixed by formaldehyde and treated as described previously (35). Sonicated chromatin was divided into three aliquots of 50 μg. One aliquot was incubated overnight at 4°C with 8 μg of SREBP-1 antibody [Santa-Cruz Biotechnology (H-160)]. As negative control for immunoprecipitation (mock), 50 μg of sonicated chromatin was incubated overnight with Tris-EDTA (TE) buffer (10 mM Tris·HCl, pH 8.0, 1 mM EDTA) at 4°C. The remaining 50 μg of chromatin was used as PCR positive control (input). Protein A Sepharose beads (Pharmacia Biotech) were used to collect the antibody-chromatin complexes for 3 h at 4°C. After washing, the beads were successively treated with DNase-free RNase A (50 μg/ml, Qbiogen) for 30 min at 37°C and proteinase K (20 mg/ml, Qbiogen) at 37°C overnight. DNA was extracted and resuspended in 80 μl of water for quantification. PCR amplification of target sequences was performed in reactions containing 4 μl of either immunoprecipitates or input. After amplification, PCR products were analyzed on 3% agarose gel. A list of primers is available on request ( ).
Analysis of gene promoter sequences.
The promoter sequences (1,000 bp upstream of the transcription starting site) of the genes with significant changes in mRNA levels following the overexpression of SREBP-1a or SREBP-1c were retrieved from TRASER (http://genome-www6.stanford.edu/cgi-bin/Traser/traser). We then used MatInspector from the Genomatix software package (Genomatix Suite release 3.0, Munich, Germany) to obtain the list of all the transcription factors that have putative binding sites on these promoter sequences (5).
Measurement of glucose oxidation.
Myotubes in six-well plates were incubated for 3 h with 5 mM glucose-DMEM supplemented with d-[U-14C]glucose (1 μCi/ml) (Perkin Elmer, Courtaboeuf, France). The incubation media were then transferred to new vials. After acidification with 5 N HCl, 14CO2 was trapped in Carbosorb (Packard Instruments) and then measured by scintillation counting in UltimaGold (Packard Instruments). Cells were washed twice in ice-cold PBS and scraped in PBS supplemented with 0.1% sodium dodecyl sulfate (SDS). Protein concentrations were measured with the Bio-Rad assay (Bio-Rad, Marnes-la-Coquette, France). The glucose oxidation rate was calculated as the amount of produced CO2 per hour and milligram of protein.
Measurement of glycogen synthesis.
Myotubes in six-well plates were incubated for 3 h with 5 mM glucose-DMEM supplemented with 12.5 mmol/l HEPES and containing 37 kBq/ml d-[U-14C]glucose (1 μCi/ml; Perkin Elmer). The incubation media were then transferred to new vials. After incubation cells were washed twice with PBS and scraped in PBS supplemented with 0.1% SDS. Glycogen was extracted as described previously (7), and the amount of [14C]glucose incorporated into glycogen was determined by scintillation counting.
Protein analysis by Western blotting.
Myotubes in six-well plates were lysed at 4°C in (mM) 200 NaF, 20 NaH2PO4, 150 NaCl, 50 HEPES, 4 NaVO4, 10 EDTA, and 2 PMSF, with 1% Triton X and 10% glycerol. After quantification by the Bradford assay (Bio-Rad), proteins were separated by SDS-10% PAGE and transferred to polyvinylidene difluoride (PVDF) membrane. Membranes were incubated overnight with primary antibody. The signal was detected with a horseradish peroxidase-conjugated secondary antibody and revealed with an enhanced chemiluminescence system (Pierce, Rockford, IL). Primary antibodies used were SREBP-1 (H160), FASN (H300), Insig-1 (N19), SIRT-1 (H300), and LaminB1 (H90) from Santa Cruz Biotechnology and ACC (3662) from Cell Signaling.
Microarray data analysis.
In this study we used adenoviral vectors expressing the mature nuclear forms of human SREBP-1a or SREBP-1c combined with oligonucleotide microarrays to determine the target genes of SREBP-1a and SREBP-1c in primary cultures of human skeletal muscle cells. We have previously verified (8) that with adenovirus infection, both SREBP-1 isoforms are overexpressed in human muscle cells at the mRNA level by quantitative RT-PCR (qRT-PCR) and at the protein level by Western blotting. Importantly, the 10- to 20-fold increase in SREBP-1a or -1c protein levels observed after adenoviral infection is similar to the increase in SREBP-1 after insulin stimulation in vivo (15, 16, 19) (Fig. 1). Our experimental conditions are thus close to physiological conditions.
For microarray analysis, only spots with recorded data on all slides were selected for further analysis, which led to the analysis of a subset of 12,825 spots among the 25,342 spots initially spotted on the arrays. Using a false discovery rate (FDR) of 1%, the SAM procedure sorted out a list of 1,315 genes significantly regulated by SREBP-1a (655 up- and 660 downregulated) and 514 genes significantly regulated by SREBP-1c (310 up- and 204 downregulated) (Supplemental Table S1).1 Because this complex pattern of gene expression was the result of a combination of posttranscriptional and/or posttranslational mechanisms, mRNA levels could not be directly linked to metabolic pathways. Thus we focused this study on analysis of the functional pathways affected by the overexpression of SREBP-1 rather than on the analysis of specific genes, and we performed a global gene promoter analysis rather than a study of cell metabolism.
Gene Ontology annotations SOURCE and KEGG were used to assign the regulated genes into functional categories. As shown in Supplemental Table S1, genes identified to associate with metabolism were mainly related to lipid (45 genes), cholesterol (19 genes), and carbohydrate (28 genes) metabolism. Although these functions are coherent with the role of SREBPs described in other tissues (43), they represented only 6% of the genes regulated by SREBP-1a and/or -1c in the human myotubes. In fact, the majority of the regulated genes encoded proteins involved in transcriptional and translational regulation (20%). One hundred ten genes encoded transcription factors regulated by one or both isoforms. This result indicated that besides the regulation of the lipid pathway, SREBP-1a and/or -1c are at the origin of a complex network of gene regulation in human skeletal muscle. The other functional categories affected by SREBP overexpression were composed of genes related to intracellular signaling (10%), cytoskeleton and vesicle trafficking (7%), receptors, carriers, and transporters (5%), mitochondrial respiration and oxidoreduction (4%), immune response and inflammation (3%), regulation of cell cycle and proliferation (3%), and apoptosis (2%).
To validate the microarray results, changes in mRNA expression of genes implicated in different cellular pathways not directly connected with the lipid metabolism were verified by qRT-PCR (Table 1). Quantitative PCR results were in accordance with the microarray data and confirmed their regulation by both SREBP-1 isoforms.
Direct target genes of SREBP-1a and SREBP-1c in human myotubes.
To identify the genes that may be direct targets of SREBP-1 as opposed to genes whose expression results from an indirect effect of the forced expression of SREBP-1a and 1c, we used the Genomatix software package to search for SRE motifs in the promoters of the regulated genes. We found 652 genes with a SRE motif in their proximal promoters (Supplemental Table S1), of which 583 were regulated by SREBP-1a (318 up- and 265 downregulated) and 223 by SREBP-1c (143 up- and 80 downregulated). Thus ∼44% of the 1,490 SREBP-1a- and/or SREBP-1c-regulated genes in human myotube cells were potential direct targets of these transcription factors.
We performed chromatin immunoprecipitation (ChIP) assays to confirm the in vivo DNA binding of SREBP-1a and/or -1c to the promoter sequences of nine genes with SRE motifs. Because there were no specific antibodies commercially available for SREBP-1a and -1c separately, chromatin was immunoprecipitated with a SREBP-1 antibody that recognized both isoforms. The results indicate that SREBP-1 binds elements in the promoters of the ARF4, SPOP, FEM1b, VSP29, HIGD1A, PGRMC2, SDC1, and SF1 genes (Fig. 2) and support the results of the microarray analysis.
Among the 652 potential target genes of SREBP-1 in human skeletal muscle identified to contain a SRE, 429 genes were specifically regulated by SREBP-1a, 69 by SREBP-1c, and 154 by both isoforms. Thus SREBP-1a regulated almost sixfold the number of genes regulated by SREBP-1c. In addition, the mRNA levels of the 154 genes regulated by both isoforms were globally higher when cells were infected by SREBP-1a than when transfected by SREBP-1c, either for the up- or downregulated genes. (Fig. 3). This result is in agreement with the fact that the two isoforms differ in their ability to stimulate transcription and that SREBP-1c is a less efficient transcriptional activator than SREBP-1a. Moreover, it was demonstrated previously that SREBP-1a and SREBP-1c proteins can bind the SREBP-1c promoter of the human srebf-1 gene. Conversely, SREBP-1a expression is not regulated by SREBP- 1a or -1c. Thus, in our experiment, it seems likely that the overexpression of SREBP-1a led to the activation of SREBP-1c promoter, and that both endogenous and adenovirus-derived isoforms regulated gene expression. Together, these results help to explain the higher fold changes observed for the genes regulated by SREBP-1a.
We then determined which biological terms displayed differential distribution between the 652 genes with a SRE motif and the 838 genes with no SRE by using FatiGO+ (http://babelomics.bioinfo.cipf.es) and KEGG (http://www.genome.jp/). The results showed that the biological pathways involved in inflammatory response, cellular morphogenesis and adhesion, and carbohydrate and sterol metabolism were significantly enriched in genes containing a SRE motif (Table 2).
Glucose oxidation and glycogen synthesis in myotubes overexpressing SREBP-1a and/or -1c.
To confirm the impact of the overexpression of SREBP-1a and -1c on carbohydrate metabolism, we investigated glucose oxidation and glycogen synthesis in infected human myotubes overexpressing SREBP-1. As shown in Fig. 4, we found that rates of glucose oxidation were significantly increased by fold changes of 1.79 ± 0.21 (P < 0.05) and 1.64 ± 0.15 (P < 0.05) in cells overexpressing SREBP-1a or -1c, respectively. At the same time, glycogen synthesis was also increased by both SREBP-1a (fold change of 1.42 ± 0.14) and SREBP-1c (fold change of 1.36 ± 0.22). Because overexpression of SREBP-1a and -1c also led to an increase of GLUT4 mRNA (Table 1), glucose uptake is increased and metabolized through both oxidation and glycogen storage.
Protein levels in myotubes overexpressing SREBP-1a and/or -1c.
To demonstrate that the microarray data are consistent with changes in protein levels in SREBP-1 overexpressing myotubes, the protein levels of several target genes were analyzed by Western blotting (Fig. 5). We first tested two proteins involved in fatty acid metabolism and showed that protein levels of fatty acid synthase (FAS) and acetyl-CoA carboxylase (ACC) are upregulated by SREBP-1, with a major effect with SREBP-1a. As previously demonstrated (8), insulin-induced gene-1 (INSIG-1) protein level is also increased on SREBP-1 overexpression. The relevance of the microarray data was also tested with another independent protein, the histone deacetylase sirtuin-1 (SIRT-1). Microarray data showed SIRT-1 mRNA increase only on SREBP-1c overexpression, but Western blot analysis indicated that SIRT-1 protein levels are increased in both SREBP-1a- and SREBP-1c-overexpressing myotubes. Together, these results indicate that although protein levels are mostly consistent with microarray data, caution should be taken when stating the specificity of SREBP-1a versus -1c on target gene expression.
Because it has been demonstrated that SREBPs are weaker transcriptional activators in isolation than when they function in combination with coregulatory proteins that bind to neighboring sites (4, 33, 34, 46, 52, 53), we wished to determine whether other transcription factors (TFs) were specifically associated with SREBP-1a and -1c in the promoter sequences of the 652 genes with SRE motifs by comparison with the genes without SRE motifs, but also regulated during the overexpression (i.e., 838 genes). z scores were calculated for each of the TFs identified. Table 3⇓ showed that 31 TF matrixes had significantly different occurrences (z score > 1.96) in the set of genes with a SRE motif.
It appeared that SRE motifs were always associated with the V$SP1F matrix that binds TFs related to the TF Sp1. In addition, all the SRE motifs were associated with RXR heterodimer binding sites (V$RXRF) for nuclear receptors like liver X receptors, retinoic acid receptors, thyroid hormone receptors, and vitamin D receptor, known to be involved in lipid metabolism. Besides these two TF matrixes, almost all the SRE binding motifs were also associated with DNA binding motifs for TFs implicated in muscle development and differentiation (V$EGRF and V$E2FF) or that mediated cAMP signaling (V$CREB). Finally, SRE binding motifs were associated to matrixes related to hypoxia (V$HIFF) or that bind TFs regulated by hypoxia (V$HESF, V$ZF5F).
SRE motif environment in promoters of genes regulated by SREBP-1a or SREBP-1c.
We then further analyzed the specificity of SREBP-1a and SREBP-1c in terms of their target genes. Because both isoforms recognized the same binding motif, we postulated that some of these functional differences would rely on the environment of the SRE motif in the promoters. To validate this hypothesis we searched for TF matrixes that were enriched in the promoters of the 429 SREBP-1a-regulated genes and in the promoters of the 69 SREBP-1c-regulated genes. Table 4A shows that two TF matrixes, V$HEAT, the DNA motif for heat shock factors, and V$GZF1, recognized by ZNF336 involved in development and morphogenesis, displayed significantly different rates of occurrence (z score > 1.96) in the promoter data set of 429 SREBP-1a genes versus the genes regulated by both isoforms. Table 4B shows the TF matrixes with significantly different rates of occurrence in the promoter data set of the 69 SREBP-1c genes versus the genes regulated by both isoforms. The majority of the 69 promoters had DNA motifs for CCAAT binding factors (V$CAAT), which are ubiquitous TFs involved in positive regulation of transcription from RNA polymerase II promoters, and for p53 tumor suppressors (V$P53F) involved in apoptosis. The other matrixes bind TFs involved in cellular defense response, inflammatory response (V$AP1F, V$OCTP, v$AIRE), and TGF-β signaling (V$FAST).
In the present study we have deciphered, for the first time, the pattern of target genes regulated by SREBP-1 in a model of human skeletal muscle cells with the use of pangenomic microarray combined with the overexpression of the mature nuclear forms of SREBP-1a and SREBP-1c. Moreover, we have also analyzed the specific transcriptional role of SREBP-1a and SREBP-1c individually, in order to discriminate between the isoforms in terms of target genes and related functions.
As observed in previous analyses of livers from transgenic mice overexpressing SREBP-1a or SREBP2 (20), isolated rat islets transfected with SREBP-1c (9), or fibroblasts transfected with SREBP-1a (23), we found that SREBP-1a and/or SREBP-1c led to changes in the expression of a large number of genes in human muscle cells (i.e., 1,490 genes representing 12% of all genes analyzed on the microarrays).
Functional analyses of the 1,490 genes regulated by one or both isoforms indicated that a large number of genes regulated on infection with SREBP-1a and/or -1c were not directly connected to lipid metabolism. Interestingly, the cellular functions affected during the treatment were grossly similar to those described previously as being regulated by insulin in vivo in human skeletal muscle (41, 51). These data are in agreement with the concept that SREBP-1 is one of the key regulators of the transcriptional action of insulin in its target tissues (36). SREBP-1 has been described as a mediator of insulin action and increases the expression of several insulin-regulated target genes like FASN, ACC, and HK2 (15). Considering the insulin signaling pathway, we confirmed in this study that SREBP-1 triggers an increase in PIK3R3 expression. Because this phosphatidylinositol 3-kinase regulatory p55 γ-subunit was recently described as a direct target gene of SREBP-1c (23), it would be interesting to further examine whether SREBP-1 can also modulate the insulin signaling cascade.
In previous studies we have demonstrated (41, 42) a marked impact of insulin on the expression of genes of the ubiquitin-proteasome system (i.e., 7% of the 800 genes regulated by insulin in human skeletal muscle), but the microarray data obtained in this study indicated that this pathway is not particularly affected by the forced expression of either SREBP-1a or -1c in human myotubes. This suggests that the effect of insulin on the expression of the ubiquitin-proteasome system is not mediated by SREBP-1.
Because both SREBP-1 isoforms are master regulators of lipid metabolism, their overexpression would induce important changes in fatty acid synthesis, transport, and catabolism. In agreement with this, we found a marked impact of the overexpression of SREBP-1a and/or -1c on the regulation of 55 genes encoding mitochondrial proteins and proteins involved in mitochondrial respiration and oxidoreduction. The importance of SREBP-1 in regulating mitochondrial function is supported by a recent study describing that in a human liver cell line (HepG2) overexpression of active SREBP-1a significantly modified the abundance of >100 mitochondrial proteins (27). More recently, it was demonstrated that SREBP-1 regulates the transcription of the mitochondrial citrate carrier (CIC), which exports citrate from the mitochondria to the cytosol (22). Interestingly, 30 genes among the 55 genes encoding mitochondrial proteins regulated in this study have SRE motifs in their promoter sequences. Altogether, these results suggest that SREBP-1a and -1c might be involved in the regulation of mitochondrial metabolism.
It seems likely that many of the observed mRNA changes were the result of nondirect effects of SREBP-1 overexpression, because only half of the 1,490 regulated genes had a DNA binding motif for SREBP in their promoter sequences. Gene ontology analysis indicated that the direct target genes of SREBP-1a and/or -1c (i.e., those promoters in which a SRE motif was identified) were significantly enriched in genes involved in the immune response. In support of this, it was demonstrated previously that changes in cholesterol levels modulated mediators of inflammation in endothelial cells and could be reversed when the level of SREBP was decreased (53). In other tissues such as the kidney (3, 37) or liver (25), increased levels of intracellular fatty acid and cholesterol triggered a proinflammatory situation. Our data indicate that SREBP-1 can be involved in the regulation of both the immune response and lipid and cholesterol metabolism pathways simultaneously.
SREBP transcription factors are known to induce expression of genes involved in cholesterol and fatty acid synthesis. Our microarray data and immunoblotting experiments confirmed that both of these pathways are activated by SREBP-1a and -1c. Interestingly, we also found 15 potential direct target genes for SREBP-1s involved in the carbohydrate metabolism. These potential SREBP-1 target genes are involved in membrane glucose transport, glycolysis, pentose phosphate pathway, and the pyruvate-citrate cycle. Furthermore, we have demonstrated that the overexpression of both SREBP isoforms leads to an increase of glucose oxidation rates as well as glycogen synthesis. Altogether, these results indicate that SREBP-1 not only mediates insulin action on glycogen storage but also triggers carbohydrate use to produce acetyl-coA and NADPH, which are necessary for cholesterol and fatty acid synthesis.
It is recognized that SREBPs are weaker transcriptional activators in isolation than when they function in combination with coregulatory proteins. For example, synergistic activation of transcription by Sp1 and SREBP-1a has been described for the regulation of low-density lipoprotein receptor (54), ACC (34), and FAS (52) gene expression. In addition, the connections of the CCAAT-binding factor/nuclear factor-Y (CBF/NF-Y) with SREBP are interesting, because an increasing number of genes involved in cholesterol metabolism appear to be regulated by SREBPs through cooperation between SRE and NF-Y binding sites (4, 33, 46). In agreement, we have found putative SP1 and NF-Y binding motifs located near the SRE motif in the hexokinase 2 (HK2) promoter (16), which reinforced the idea that these three transcription factors may have cooperative effects. As expected from these previous studies, it appeared that all the genes with a SRE motif in their promoter sequences also had a DNA binding motif for members of the Sp1 family. However, we did not find significant enrichment for DNA motifs associated to CBF/NF-Y, indicating that in skeletal muscle the global transcription action of SREBP-1a/1c does not necessarily involve the cooperation between SREBP and NF-Y binding sites. Finally, 97% of the genes with a SRE motif had also DNA binding motifs for cAMP-responsive element (CRE)-binding proteins. Cooperation between SREBP and CRE-binding proteins has been demonstrated for the regulation of lanosterol 14α-demethylase (CYP51) (18) and 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) synthase (11), both involved in the cholesterol biosynthesis pathway. In human skeletal muscle, the presence of a DNA motif for CRE-binding proteins in the neighborhood of SRE motifs is not restricted to the genes involved in the lipid pathway and might be necessary for the global transcriptional action of SREBP-1a and -1c in this tissue.
Besides these three TFs, we identified DNA motifs for TFs involved in the adaptative response to hypoxia. First described in Schizosaccharomyces pombe, in which homologs of SREBP-1 stimulated the transcription of genes required for adaptation to hypoxia (21), the relationship between SREBP and the response to hypoxia was further demonstrated in murine models. Intermittent hypoxia induces hyperlipidemia in the liver of lean (10, 30) and obese (28) mice, suggesting a link between hepatic lipid metabolism and hypoxia. More recently, it was demonstrated that hyperlipidemia in response to hypoxia was mediated by SREBP-1 (29). In our study, half of the genes with a SRE motif also had DNA binding motifs for hypoxia-inducible factors. Our results indicate that, as in the liver, hypoxia and SREBP pathways are connected in human skeletal muscle.
The physiological relevance of the coexpression of both SREBP-1a and -1c in mammalian tissues is not clear. It has been postulated that because SREBP-1a interacts with coactivators and also activates gene expression more robustly than SREBP-1c, SREBP-1a would be expressed in tissues and conditions in which increased lipid synthesis is required and SREBP-1c would be preferentially expressed when low but regulated levels of lipid synthesis are important (48). Here we have shown that, as in other tissues, SREBP-1a and SREBP-1c isoforms differentially activated gene transcription (i.e., number of genes and fold changes). Unexpectedly, we also found that SREBP-1a and SREBP-1c have specific target genes in human muscle cells. To explain part of the molecular basis of this difference, we determined whether some DNA binding motifs for TFs were enriched in the promoter sequence of the gene regulated either by SREBP-1a or by SREBP-1c. Our data indicated that SRE motifs were statistically associated with DNA binding motifs for TFs involved in the immune response, in the promoter sequences of the genes specifically regulated by SREBP-1c. This function, already discussed as being one of the important functions of the target genes of SREBP-1 in human skeletal muscle, would be more dependent on the transcriptional action of SREBP-1c. Interestingly, the immune response was also one of the functional pathways described as regulated by insulin in vivo in human muscle (41, 51). Because in skeletal muscle insulin regulates SREBP-1c transcription, we can speculate that SREBP-1c would mediate the action of insulin on the regulation of genes involved in the immune signaling pathway.
We found that SRE motifs in the promoter sequences of the genes regulated specifically by SREBP-1a were statistically associated with DNA binding motifs for heat shock factors. A link between the heat shock response and the SREBP pathway was described recently in adipocytes overexpressing SREBP2 (40). In these cells, several genes encoding HSP proteins were upregulated. Heat shock proteins (Hsps) are molecular chaperones that aid in protein synthesis and trafficking and have been shown to protect cells/tissues from various protein-damaging stressors. In skeletal muscle, Hsps are upregulated during muscle hypertrophy (14, 31) or in aged skeletal muscle (31). In our study we found that four HSP proteins are specifically regulated by SREBP-1a. Furthermore, it has also been found that adipocytes overexpressing DnaJA4 (a HSP40/DnaJ protein family member) exhibited a specific increase in the rate of cholesterol synthesis. In our study, four proteins belonging to the same HSP40/DnaJ protein family were upregulated during the overexpression of SREBP-1a. These data suggested that at least part of the heat shock response could be mediated by SREBP-1a in human skeletal muscle.
Together, the results described in this study indicate that in human muscle cells SREBP-1a and -1c are involved in the regulation of a large number of genes that are at the crossroads of different functional pathways. Our in silico promoter analysis showed that different combinations of TFs that have binding sites around the SRE binding motifs may determine regulatory networks of transcription that could explain the superposition of lipid and cholesterol metabolism with other various pathways involved in adaptive responses to stress like hypoxia and heat shock, or in the immune response.
This work was supported in part by a research grant from the French National Program of Research on Diabetes (PNRD, ROSIH project).
The authors acknowledge Jérémy Besson, Florine Jordano, and Erwin Guet for technical help in microarray data analysis, Corine Picat for cell culture, and Kevin Hogeveen for English corrections.
↵* S. Rome and V. Lecomte contributed equally to this work.
↵1 The online version of this article contains supplemental material.
Address for reprint requests and other correspondence: S. Rome, UMR INRA 1235/INSERM 870, Régulations Métaboliques, Nutrition et Diabètes, Faculté de Médecine Lyon Sud-BP 12, 165 chemin du Grand Revoyet, F-69921 Oullins Cedex, France (e-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society