Human Glycolipid Transfer Protein Gene (GLTP) Expression Is Regulated by Sp1 and Sp3

Glycolipid transfer protein (GLTP) accelerates glycolipid intermembrane transfer via a unique lipid transfer/binding fold (GLTP fold) that defines the GLTP superfamily and is the prototype for functional GLTP-like domains in larger proteins, i.e. FAPP2. Human GLTP is encoded by the single-copy GLTP gene on chromosome 12 (12q24.11 locus), but regulation of GLTP gene expression remains completely unexplored. Herein, the ability of glycosphingolipids (and their sphingolipid metabolites) to regulate the transcriptional expression of GLTP via its promoter has been evaluated. Using luciferase and GFP reporters in concert with deletion mutants, the constitutive and basal (225 bp; ∼78% G+C) human GLTP promoters have been defined along with adjacent regulatory elements. Despite high G+C content, translational regulation was not evident by the mammalian target of rapamycin pathway. Four GC-boxes were shown to be functional Sp1/Sp3 transcription factor binding sites. Mutation of one GC-box was particularly detrimental to GLTP transcriptional activity. Sp1/Sp3 RNA silencing and mithramycin A treatment significantly inhibited GLTP promoter activity. Among tested sphingolipid analogs of glucosylceramide, sulfatide, ganglioside GM1, ceramide 1-phosphate, sphingosine 1-phosphate, dihydroceramide, sphingosine, only ceramide, a nonglycosylated precursor metabolite unable to bind to GLTP protein, induced GLTP promoter activity and raised transcript levels in vivo. Ceramide treatment partially blocked promoter activity decreases induced by Sp1/Sp3 knockdown. Ceramide treatment also altered the in vivo binding affinity of Sp1 and Sp3 for the GLTP promoter and decreased Sp3 acetylation. This study represents the first characterization of any Gltp gene promoter and links human GLTP expression to sphingolipid homeostasis through ceramide.

Despite the significance of the GLTP fold, the in vivo function(s) of GLTP remains unsettled. GLTP resides in the cytoplasm (17,18), a favorable location for interaction with newly synthesized glucosylceramide (GlcCer) generated by GlcCer synthase on the cytoplasmic face of the Golgi (19,20). However, GlcCer destined for higher GSL synthesis is transferred through the Golgi by FAPP2, which contains a C-terminal GLTP-like domain, rather than by GLTP (13). RNAi knockdown of GLTP in the presence of the vesicle trafficking inhibitor, brefeldin A, suggests a role in GlcCer trafficking to the plasma membrane (21). Yet, GLTP docking with vesicle-associated membrane protein-associated proteins of the endoplasmic reticulum also appears possible as well as action as an intracellular glycolipid sensor involved in GSL homeostasis (1,17,18).
In the present study, our goal was to evaluate GLTP gene expression within the context of GSL metabolic homeostasis by determining if alterations in key sphingolipid metabolites trigger changes in GLTP transcription, as regulated by its previously uncharacterized GLTP gene promoter. Recently, we characterized human GLTP, a single-copy gene on chromosome 12 (12q24.11) (12). GLTP mRNA matures via classic cis-splicing into 5-exon transcripts, a highly conserved organizational pattern in therian mammals and other vertebrates (12). The discovery of an unusually GϩC-rich CpG island in the 5Ј-UTR of GLTP indicated possible regulation by transcriptional factors that bind to GC boxes, e.g. Sp1 (specific protein-1)/Sp3 (22,23). The present study provides the first insights into human GLTP transcriptional regulation, including characterization of constitutive and basal GLTP promoter (GenBank TM accession no. GU971358) and adjacent regulatory regions. Promoter analyses using luciferase and GFP reporters as well as in vivo analyses by real-time PCR and other approaches show GLTP regulation via mechanistic participa-tion of Sp1/Sp3 transcription factors in a manner influenced by ceramide but not by related sphingolipid metabolites.

EXPERIMENTAL PROCEDURES
Cell Culture-HEK 293T, HeLa, and T47D cells (American Type Culture Collection, Rockville, MD) were cultured at 37°C under 5% CO 2 in DMEM (Mediatech Inc, Herndon VA) supplemented with 10% heat-inactivated fetal bovine serum (Innovative Research, Inc., Novi, MI). To assess human GLTP promoter regulation in response to increasing endogenous ceramide, HeLa cells were transfected with pGL3(Ϫ1150/ ϩ19) and then treated with vehicle (0.1% DMSO) or with GlcCer synthase inhibitor, D-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP; Sigma-Aldrich) at 10 M for 24 h before measuring luciferase activity. To reduce endogenous ceramide and assess the effect on human GLTP promoter activity, HeLa cells were transfected with pGL3(Ϫ1150/ϩ19) for 8 h before replenishing with fresh DMEM medium and treating with (dihydro)-ceramide synthase inhibitor, fumonisin B1 (FB1; Sigma-Aldrich) for 40 h at 25 M. To elevate endogenous ceramide levels, cells were treated with PDMP for 24 h at 10 M. To analyze the effect of C 6 -ceramide treatment on endogenous ceramide levels, cells were grown to ϳ60% confluency and then treated with 10 M C 6 -ceramide for 24 h. Endogenous ceramide levels were assessed by HPLC mass spectrometry (Lipidomics Core, Medical University of South Carolina, Charleston, SC).
5Ј-Rapid Amplification of cDNA Ends Assay (RACE)-Total RNA was isolated from HeLa cells using TRIzol reagent (Invitrogen). Transcriptional start sites were identified by First-Choice RNA ligase-mediated rapid amplification of cDNA ends (RLM-RACE, Ambion, Inc., Austin, TX). Master-Amp TM Tth DNA polymerase and PCR enhancer (Epicenter, Madison, WI) were used for 70°C reverse transcription. Herculase II fusion DNA polymerase (Stratagene, La Jolla CA) supplemented with betaine (Sigma-Aldrich) was used for standard PCR amplification. Primer Ra1 and Ra2 were used for first and second round PCR amplifications (supplemental Table S1). Reaction products were separated by 1.2% agarose gel electrophoresis before cloning in pGEM-T (Promega, Madison, WI) and sequencing (Genewiz, South Plainfield, NJ).
EMSA-5Ј-Biotin-labeled, single-strand probes were synthesized by Sigma. Double-stranded oligonucleotide probes were generated by incubating equimolar amounts of complementary oligonucleotides in STE annealing buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM EDTA) for 3 min at 95°C, slow cooling to room temperature, and storing (Ϫ20°C). EMSA reaction mixtures were incubated in 5ϫ binding buffer (Promega) on ice for 10 min with or without unlabeled competitor, prior to adding end-labeled oligonucleotides for 20 min on ice. For supershift assays, HeLa cell nuclear extracts (Promega) were incubated on ice for 10 min with anti-Sp1 (PEP-2) and anti-Sp3 (D-20) antibody (Santa Cruz Biotechnology) prior to addition of end-labeled oligonucleotides for 30 min on ice and electrophoresis on a 5% nondenaturing polyacrylamide gels in Tris borate-EDTA buffer (0.5ϫ) at 70 V for 30 min at 4°C and then at 120 V for ϳ60 min. Binding reactions were analyzed by transferring to Biodyne B precut modified nylon membrane (Pierce) at 350 mA for 60 min, cross-linking for 5 min using UV light (312 nm bulb), and fixing at 80°C for 60 min before detection with biotin-labeled DNA using light-shift electrophoretic mobility shift reagent (Pierce). Antibody-protein complexes were observed as supershifted or immunodepleted complexes.
ChIP-Chromatin isolated from HeLa cells was used in ChIP assays performed according to the manufacturer's instructions (Upstate Biotech., Lake Placid, NY). For amplification of the GLTP promoter, primer pairs CH-1/CH-2 (first round) and CH-3/CH-4 (second round) were used for nested PCR. Primer pair Ne-1/Ne-2 (first and second PCR rounds), designed to amplify ϩ489/ϩ707 of human GLTP exon 5, served as control lacking Sp1/Sp3 binding sites. Herculase polymerase plus betaine, was used for PCR amplification. Cycling conditions were: 2 min at 98°C; 25 cycles of 98°C for 20 s, 63°C for 20 s, and 72°C for 30 s; and final extension at 72°C for 3 min.
RNA Interference-eIF4E translation was silenced by transfecting HeLa cells with eIF4E siRNA (Cell Signaling; 25 nM) using TransIT-siQUEST TM (Mirus, Madison, WI) and analyzing for exogenous protein expression level after 48 h. Sp1/Sp3 siRNA (Santa Cruz Biotechnology) were used to knockdown Sp1 or Sp3. Nonspecific siRNA served as control.
Lipid Effects on GLTP Promoter Activity-HeLa cells were transfected with different plasmid constructs for 24 h and then replenished with fresh medium. C 6 -ceramide, C 6 -dihydroceramide, C 8 -ceramide 1-phospate, C 18 -sphingosine, C 18sphingosine-1-phosphate, C 8 -glucosylceramide, ganglioside GM 1 or C 12 -3-sulfo-galactosylceramide (Avanti Polar Lipids, Alabaster, AL), dissolved in DMSO, were added to the medium to final concentrations of 5 and 10 M. Luciferase activity was measured using the Dual-Luciferase reporter assay system 24 h after lipid treatment. To assess the regulatory role of acetylation, HeLa cells were transfected with pGL3(Ϫ1150/ ϩ19) and, after 24 h, treated with trichostatin A (TSA; dissolved in DMSO and added to medium to 100 ng/ml final concentration), or with C 6 -ceramide (10 M) and TSA for 24 h. Luciferase activity was measured using the dual luciferase reporter assay system.
Real-time RT-PCR-Total RNA, isolated from HeLa cells using RNeasy Plus minikits (Qiagen, Valencia, CA) and TRIzol reagent, was reverse-transcribed with Superscript III (Invitrogen). Real-time RT-PCR was performed using TaqMan Gene Expression assays (ID Hs00829505_g1 for human GLTP gene and ID Hs99999903_m1 for the ␤-actin gene; Applied Biosystems, Foster City, CA).
Statistical Analysis-Three to six independent experiments were always performed, and the Student's t test was used to compare mean values and generate standard errors using Excel (Microsoft, Redmond, WA).

RESULTS
Human GLTP Transcriptional Start Sites-RLM-RACE is designed to utilize only full-length, capped mRNA during amplification of cDNA, facilitating accurate identification of transcriptional start sites (TSS) (25,26). To identify the TSSs of human GLTP, we performed RLM-RACE PCR using mRNA from HeLa cells and GLTP-specific primers (supplemental Table S1). A single cDNA band of ϳ400 bases was obtained and cloned into T vector for DNA sequencing (supplemental Fig. S1). Ten randomly selected clones revealed TSSs located 26 bases (TSS1, seven clones) and 24 bases (TSS2, three clones) from the translation initiating ATG codon of human GLTP (Fig. 1). Similar results were obtained with HEK 293T cells (data not shown), indicating that human GLTP can be transcribed from more than one start sites but that TSS1 represents the major start site.
Human GLTP Promoter-To define the human GLTP promoter, the Ϫ1150/ϩ19 region (relative to major transcriptional start site) was inserted into promoterless phrGFP. Fig.  2A shows that cells transfected with phrGFP(Ϫ1150/ϩ19) became fluorescent after 48 h but not when mock-transfected with empty phrGFP. Similar results were obtained in HeLa and HEK 293T cells, indicating a functional promoter in the Ϫ1150/ϩ19 region of the human GLTP gene.
GLTP Expression Is Not Regulated by mTOR Signaling Pathway-A notable characteristic of human GLTP promoter is the extremely high GϩC content, which is 79.54% for Ϫ416/ϩ19 and 76.13% for the entire CpG island. Human GLTP also contains an extremely long first intron and the 5Ј and 3Ј regions of the first exon are highly GϩC-rich. These features raised the issue of whether human GLTP falls into a gene class regulated at the translational level (27). With such mRNAs, the highly structured 5Ј-UTRs require eIF4E binding to the mRNA cap structure to mediate initiation of translation, thus facilitating efficient scanning and start codon recognition and enhancing translation of these mRNAs (27,28). As shown in Fig. 3, A and B, neither RNAi suppression of eIF4E  by 3-to 4-fold nor rapamycin treatment decreased GLTP levels despite phospho-Ser 65 4E-BP1 decrease, which enhances eIF4E and 4E-BP1 binding, preventing assembly into eIF4F complex and inhibiting cap-dependent translation. The results show that GLTP gene expression is not regulated by the mTOR (mammalian target of rapamycin) signaling pathway.
Sp1 and Sp3 Bind to Multiple Sites in Human GLTP Promoter and Silencing Reduces Activity-Bioinformatics analyses indicated that the 225-bp region (Ϫ350/Ϫ126) lacked canonical CCAAT and TATA boxes but contained consensus binding sites for various other transcription factors including Sp1/Sp3. To determine whether GC-boxes functioned as viable binding sites for Sp1 and Sp3, nuclear extracts from HeLa cells were analyzed by EMSA. As shown in Fig. 4, bound biotin-labeled probe showed several labeled bands. Binding specificity was confirmed by the significantly decreased intensity resulting from incubation with 100-fold excess of unlabeled probe and the lack of intensity change by incubation with a 100-fold excess of unlabeled mutated probes. A major, low mobility Sp1 complex was observed, which supershifted upon addition of Sp1-specific antibody (Fig. 4). Furthermore, in the presence of Sp3-specific antibody, the supershifted band became very faint, the intensity of the major complex decreased, and the lower Sp3 complex disappeared, indicating competitive binding by Sp3. When both Sp1 and Sp3 antibodies were added, the intensity of the major complex was dramatically decreased, and the lower Sp3 complex again disappeared (Fig. 4). The binding affinity hierarchy for Sp1/Sp3 was as follows: Ϫ191/Ϫ186 Ͼ Ϫ270/Ϫ265 Ͼ Ϫ219/Ϫ214 Ͼ Ϫ154/Ϫ149 based on staining density. Collectively, the EMSA analyses demonstrate that the GC-boxes located on the GLTP proximal promoter do serve as binding sites for Sp1 and Sp3.
The in vivo binding status of Sp1 and Sp3 was assessed by ChIP. As shown in Fig. 5A, the human GLTP promoter was immunoprecipitated by either Sp1 or Sp3 antibody, but not by ␤-actin antibody. Plasmid controls as well as sheared and cross-linked input DNA served as templates for positive control bands (Fig. 5A). No signal was observed using a control primer pair specific for the GLTP exon 5 region (Fig. 5A). The results clearly show in vivo binding of Sp1/Sp3 to the proximal region of the GLTP promoter.
To directly assess the effect of Sp1/Sp3 on GLTP promoter activity, siRNA was used to knock down Sp1 and Sp3 expression (Fig. 5B). Down-regulation of Sp1/Sp3 expression resulted in a 27ϳ35% reduction in GLTP promoter activity (Fig. 5C). Treatment with mithramycin A, a drug that binds GC-rich regions of DNA and blocks Sp1 binding (29), also reduced GLTP promoter activity (Fig. 5C). Taken together, the data show regulation of human GLTP gene expression by Sp1/Sp3. Sp1/Sp3 Binding Site Mutation at Ϫ219/Ϫ214 of GLTP Promoter Inhibits Transcriptional Activity-The contribution of representative Sp1/Sp3 binding sites were defined by sitedirected mutagenesis. Because previous EMSA indicated that the mutated oligonucleotide duplexes could not compete for transcription factor binding, we concluded these mutant probes could not bind transcription factors Sp1/Sp3. As shown in Fig. 6A, site mutation of Ϫ219/Ϫ214 had a strong negative effect on GLTP promoter activity; whereas mutations at Ϫ270/Ϫ265, Ϫ191/Ϫ186, and Ϫ154/Ϫ149 had only mild effects on promoter activity. Interestingly, in T47D cells, sig- nificantly decreased GLTP promoter activity resulted from mutation at Ϫ191/Ϫ186, but not in HeLa cells and HEK 293T cells (Fig. 6A). BLAST searches of the GLTP 5Ј-flanking sequence of H. sapiens and Macaca mulatta showed conservation of all four Sp1/Sp3 binding sites except for a single base substitution at the (Ϫ154/Ϫ149) binding site ( Fig. 6B; M4). Thus, the Ϫ219/Ϫ214 Sp1/Sp3 binding site, but not the Ϫ270/Ϫ265, Ϫ191/Ϫ186, and Ϫ154/Ϫ149 sites, is a key regulator of human GLTP promoter activity.
Sp1 and Sp3 Involvement in Ceramide-induced Up-regulation of GLTP-Truncations of the GLTP promoter enabled mapping of the ceramide-response region. As shown in Fig.  8A, ceramide treatment increased the activity of pGL3(Ϫ284/ ϩ19) by ϳ20%, but had no effect on pGL3(Ϫ213/ϩ19) or pGL3(Ϫ126/ϩ19), suggesting that the Ϫ350/Ϫ213 region contains a ceramide response element. To gain further insights in vivo, the effect of ceramide treatment on expression levels of endogenous Sp1 and Sp3 was assessed, but no significant changes were observed (Fig. 8B). However, ChIP analyses revealed that ceramide treatment alters the in vivo binding affinity of Sp1 and Sp3 for GLTP, resulting in lower Sp1 binding but higher Sp3 binding (Fig. 8C). Because acetylation of Sp3 and Sp1 is known to regulate their binding (30 -32), and ceramide treatment can alter Sp acetylation status (33,34), we used TSA to increase Sp acetylation levels and found GLTP promoter activity to be diminished by ϳ60 -65% (Fig. 8D). TSA alters Sp3/Sp1 acetylation by inhibiting lysine deacety-lases, originally referred to as histone deacetylases (31,35,36). Ceramide treatment attenuated the TSA-induced loss of GLTP promoter activity (Fig. 8C) and significantly reduced in vivo levels of acetylated-Sp3 (Fig. 8E) without affecting acetylation status of Sp1 (supplemental Fig. S3). Collectively, the data show regulation of GLTP expression via Sp1/Sp3 by a complex mechanism that responds to elevated ceramide levels.

DISCUSSION
This investigation represents the first characterization of any Gltp gene promoter and provides the first insights into the regulation of human GLTP gene expression. BLAST searches show the GLTP promoter to be highly conserved in primates, but less so in Carnivora, Cetartiodactyla, and rodents, suggesting relatively recent and potentially important evolutionary developments within the human GLTP promoter. Our earlier study of GLTP gene organization, tissue transcript levels, and phylogenetic/evolutionary relationships revealed orthologs in all vertebrate genomes and encoded within a highly conserved five-exon/four-intron mRNA organizational pattern (12). In humans, two single-copy genes occur. The 12q24.11 gene (GLTP) accounts for all detectable GLTP transcript (12). In contrast, a complete GLTP ORF (94% homology) on human chromosome 11 (11p15.1) and present only in primates, is a transcriptionally silent pseudogene (GLTPP1) based on methylation analyses of CpG islands and well controlled PCR analyses. An unexpected outcome was the discovery of a 5Ј-UTR in GLTP with a CpG island ChIP assays were performed using DNA from HeLa cells as sources of the human GLTP promoter and specific antibodies for Sp1, Sp3, or ␤-actin. GLTP promoter region (Ϫ350/Ϫ115) containing putative Sp1/Sp3 binding sites was amplified by nested PCR as described under "Experimental Procedures." The human GLTP exon 5 fragment (ϩ489/ϩ707), amplified using primer pair Ne-1/Ne-2, served as control lacking Sp1/Sp3 binding sites. Amplification controls were as follows: cross-linked, sheared DNA prior to immunoprecipitation (input); plasmid carrying (Ϫ350/Ϫ115) or (ϩ489/ϩ707) was used as template (plasmid). B and C, down-regulation of Sp1/Sp3 by siRNA knockdown reduces GLTP promoter activity. Cells transfected with pGL3(Ϫ1150/ϩ19) were grown for 24 h, transfected with Sp1/Sp3 siRNA (25 nM), and grown for an additional 24 h before Western blot analysis and measurement of luciferase activity. D, mithramycin A treatment down-regulates Sp1/Sp3 expression and decreases GLTP promoter activity. Cells transfected with pGL3 (Ϫ1150/ϩ19) were grown 24 h, treated with mithramycin A (0, 100, 200, or 500 nM) for 24 h, and analyzed for luciferase activity (normalized to Renilla luciferase activity). Vehicle, 0.1% DMSO. Bars show the means Ϯ S.E. of three to six determinations in HeLa cells. *, p Ͻ 0.05; **, p Ͻ 0.01. unusually GϩC-rich. The high number of GC boxes revealed by on-line Transcription Element Search System analysis of the GLTP promoter prompted our focus on regulation by Sp1/Sp3.
The very high GϩC content (ϳ76% for entire CpG island, including exon 1 ORF; ϳ80% for Ϫ416/ϩ19) also raised the issue of whether GLTP might belong to a class of genes that are very efficiently transcribed (37) and translationally regu-lated in mammalian cells. In such genes, the GϩC-rich 5Ј-UTRs are highly structured and require eIF4E binding to the mRNA cap to initiate translation (27,28). However, GLTP levels were unaffected by RNAi knockdown of eIF4E or by rapamycin treatment, indicating a lack of significant regulation by the mTOR signaling pathway. The outcome could reflect the rather short 5Ј-UTR length of GLTP transcript. RLM-RACE PCR revealed more than one transcription start FIGURE 7. Ceramide increases human GLTP promoter activity. Elevated endogenous ceramide level increases GLTP expression. A, comparison of the effects of various sphingolipids on GLTP promoter activity. HeLa cells were transfected with pGL3(Ϫ1150/ϩ19) and then treated with different lipids. Luciferase activity was measured 24 h after lipid treatment and was normalized to Renilla luciferase activity. B, real-time PCR analyses of ceramide-induced GLTP transcript level up-regulation in vivo. Cells were grown to ϳ60% confluency and then were treated with 0, 5, or 10 M C 6 -ceramide for 3, 6, or 12 h before harvesting for real-time RT-PCR analyses. GTLP mRNA levels were normalized to ␤-actin mRNA levels. C, effect of C 6 -ceramide on endogenous ceramide levels. Cells were grown to ϳ60% confluency, and then cells were treated with 10 M C 6 -ceramide for 24 h and analyzed for their ceramide species content by HPLC-MS as described under "Experimental Procedures." Results were normalized to cell lipid phosphate. GTLP mRNA levels were normalized to ␤-actin mRNA levels. Cer, ceramide. Bars represent the means Ϯ S.E. of three to six determinations, *, p Ͻ 0.05 compared with cells treated with vehicle (0.1% DMSO), respectively. **, p Ͻ 0.01. site, consistent with the observed lack of canonical CCAAT and TATA boxes (38,39). Nonetheless, the major start site (TSS1) in 7 of 10 clones was only 26 bp upstream of the mRNA start codon. Such a short 5Ј-UTR length could help avoid highly structured conformation(s) and the need for eIF4E involvement in mediating initiation of GLTP translation.
By luciferase reporter analyses, we demonstrated a 1169-bp (Ϫ1150/ϩ19) region relative to TSS1 to be transcriptionally active in HeLa, HEK 293T, and T47D cells. 5Ј and 3Ј mutational deletion analyses suggested a negative regulatory region upstream of Ϫ350, a positive regulatory region (ϩ19/ϩ200), and a basal GLTP promoter (Ϫ350/Ϫ126). The 225-bp core promoter is active in HeLa, HEK 293T, and T47D cells. However, among these cells, differential regulation involving regions upstream of the core promoter may occur (Figs. 2 and 6). Future studies will be needed to elucidate the molecular basis of the differences as well as the role of other predicted transcription factor sites (including other Sp1/ Sp3 sites) in the regulation of GLTP promoter in a tissuespecific context.
What is clear is that the GLTP promoter can be regulated by Sp1/Sp3 as shown by the 27ϳ35% reduction in activity after Sp1/Sp3 knockdown by RNAi or mithramycin A treatment. Sp1 and Sp3 interaction with GLTP GC-boxes is evident in vitro and in vivo as assessed by EMSA and ChIP, respectively. Sp1 binding to GC-box elements is known to occur via zinc finger domains and enables RNA polymerase II binding to the transcription initiation site in TATA-boxless promoters (40 -43) such as occurs in GLTP. More than 20 potential Sp1/Sp3 binding sites are predicted by the on-line Transcription Element Search System program within the Ϫ350/Ϫ213 region that also could contribute to the ceramide response. Our EMSA analyses establish the following hierarchy of binding affinity among the four sampled GC-boxes: (Ϫ191/Ϫ186 Ͼ Ϫ270/Ϫ265 Ͼ Ϫ219/Ϫ214 Ͼ Ϫ154/Ϫ149). However, GC-box mutation indicated that site Ϫ219/Ϫ214 is essential for maximal GLTP promoter activity. Luciferase activity was measured 24 h after lipid treatment. *, p Ͻ 0.05 compared with cells treated with vehicle (0.1% DMSO), respectively. Luciferase activity was normalized to Renilla luciferase activity. B, ceramide treatment does not alter the endogenous levels of Sp1 or Sp3. HeLa cells were grown with or without C 6 -ceramide (10 M) for 24 h. Sp1, Sp3, and ␤-actin levels in cell extracts were analyzed by Western blot analysis. C, ceramide treatment alters the Sp1 or Sp3 binding affinity to the GLTP promoter. ChIP assays were performed using DNA from HeLa cells and treated with vehicle (DMSO) or C 6 -ceramide (10 M) for 24 h. Immunoprecipitation of DNA-Sp1 or DNA-Sp3 complexes were performed using antibodies specific for Sp1 and Sp3. Amplification of GLTP promoter region(Ϫ350/Ϫ115) was performed as described under "Experimental Procedures." D, TSA-induced decrease in GLTP promoter activity is partially blocked by ceramide. HeLa cells were transfected with pGL3(Ϫ1150/ϩ19) for 24 h and then treated with TSA (100 ng/ml) and C 6 -ceramide (10 M). After 24 h, luciferase activity was measured. *, p Ͻ 0.05. E, ceramide decreases acetylated Sp3 levels. HeLa cells were grown with or without C 6 -ceramide (10 M) for 24 h. After immunoprecipitation with Sp3 antibody, acetylated Sp3 levels were determined by Western blot analysis (acetylated lysine antibody). Specific binding is indicated by preclearing with beads containing no antibody. Vehicle, 0.1% DMSO; Cer, ceramide. Bars represent the mean Ϯ S.E. of three to six determinations.
Because Sp transcription factors regulate many housekeeping, tissue-specific, viral, and inducible genes, involvement of Sp1/Sp3 in GLTP promoter regulation is hardly surprising given its high GϩC content. Although Sp3 can either activate or repress, Sp1 more often activates gene promoters. For the GLTP promoter, Sp1 and Sp3 both serve as activators, as indicated by RNAi knockdown and mithramycin A treatment. Adding to the complexity is the potential role of Sp1-like KLF proteins (44) and co-regulatory transcription activator factors, able to bind to the gluatamine and serine/threonine-rich regions of Sp1 proteins, and be recruited to the multiprotein preinitiation complex during interaction with gene promoters. Transcription activator factors are known to confer cell type-specific promoter selectivity, but their involvement in the regulation of the GLTP promoter remains to be determined (45,46). It is clear that the involvement of these other factor(s) in GLTP transcriptional regulation cannot be excluded.
It is noteworthy that elevated ceramide not only up-regulates GLTP promoter activity but also mitigates decreases in promoter activity induced by Sp1/Sp3 knockdown. Interestingly, ceramide treatment does not alter endogenous levels of Sp1 and Sp3 but rather, their binding affinity for the GLTP promoter. Furthermore, in the case of Sp3, the altered binding affinity can be linked to ceramide-induced changes in acetylated Sp3 levels in vivo. Thus, our results suggest that ceramide treatment can affect the GLTP promoter in two ways: (i) by altering Sp1/Sp3 binding affinity and (ii) by altering Sp3 acetylation status. Our findings are supported by recent work showing that ceramide regulates transcriptional activity of human telomerase reverse transcriptase in A549 human lung adenocarcinoma cells by changes in Sp3 acetylation status (34). The ability of ceramide treatment to alter gene expression by promoter regulation has also been noted for human glucosylceramide synthase, c-Myc, surfactant protein B, and matrix metallopeptidase 2 (e.g. Refs. [55][56][57][58][59][60]. From the standpoint of GSL metabolism, it is noteworthy that the promoter activity of the GlcCer synthase gene increases in response to C 6 -ceramide treatment by a mechanism involving Sp1 (52,54). A ceramide-stimulated increase in GlcCer synthase, involving coordinate elevation of GLTP gene expression, could provide an effective means for adjusting intracellular ceramide levels while maintaining cellular GSL homeostasis. Increased GLTP would be expected to compete with FAPP2, which normally transfers newly synthesized GlcCer from the cis-medial Golgi using its pleckstrin homology domain to target the trans-Golgi, the site of complex GSL synthesis (13). The lack of a pleckstrin homology domain in GLTP minimizes specific targeting to the trans-Golgi. Thus, up-regulation of GLTP could enable GlcCer to be siphoned away from the trans-Golgi, preventing elevation of downstream GSL levels in the biosynthetic pathway, from lactosylceramide to complex gangliosides. In such a way, elevated GLTP levels could help to maintain complex GSL homeostasis during periods of elevated ceramide levels.
The link that we observe between sphingolipid homeostasis and GLTP expression via ceramide-responsive, Sp1/Sp3-mediated, transcriptional regulation of GLTP is an intriguing development, given the lack of effect elicited by other powerful sphingolipid signaling metabolites, i.e. ceramide 1-phosphate, sphingosine 1-phosphate, or sphingosine. The characterization of the human GLTP promoter and the fundamental insights into human GLTP transcriptional regulation undoubtedly will aid future elucidation of normal and pathological conditions involving GLTP expression.