Choline transporter-like (CTL) proteins of the CTL1 family are novel transmembrane proteins implicated in choline transport for phospholipid synthesis. In this study, we characterized the 5′-flanking region of the human (h)CTL1 gene and examined some of the possible mechanisms of its regulation, including promoter activity, splicing, and expression. The transcription start site of the hCTL1 gene was mapped by 5′-rapid amplification of cDNA ends (RACE), and the presence of two splice variants, hCTL1a and hCTL1b, was investigated using isoform-specific PCR and 3′-RACE. The hCTL1 promoter region of ∼900 bp was isolated from MCF-7 human breast cancer cells. The promoter was TATA-less and driven by a long stretch of GC-rich sequence in accordance with widespread expression of hCTL1 at both mRNA and protein levels. Deletion analyses demonstrated that a very strong promoter is contained within 500 bp of the transcription start site, and more upstream regions did not increase its activity. The core promoter that conferred the minimal transcription is within the −188/+27-bp region, and its activity varied in human breast cancer and mouse skeletal muscle cells. Multiple motifs within the promoter regulatory region bound nuclear factors from both cultured cells and normal human skeletal muscle. The motifs within the three regions [S1 (−92/−61 bp), S2 (−174/−145 bp), and S3 (−289/−260 bp)] contained overlapping binding sites for hematopoietic transcription factors and ubiquitous transcription factors, in line with the expected gene function. Genomic analyses demonstrated a high conservation of hCTL1 and mouse CTL1 proximal promoters. Accordingly, mRNA profiles demonstrated that human splice variants were expressed ubiquitously, as demonstrated for the mouse transcripts; however, they differed from the profiles of rat CTL1 transcripts, which were more restricted to neurons and intestinal tissues. The shorter hCTL1b variant contained the cytosolic COOH-terminal motif L651KKR654 for endoplasmic reticulum retrieval/retention. This retention signal was conserved in hCTL1b and rat and mouse CTL1b and is typical for transmembrane proteins of type 1 topology.
- choline transporter-like protein 1 promoters
- gene expression
the majority of choline is incorporated into phosphatidylcholine (PC) or proceeds through mitochondrial oxidation to betaine, thus entering the one-carbon methylation pool (74, 62), or is used in the production of the neurotransmitter acetylcholine (45). Interest in the factors that influence choline transport and metabolism has grown because of choline’s association with neural tube defects (49, 51), memory development (75), Alzheimer’s disease (53), homocysteine production (50), and inherited disorders of neuromuscular transmission (40, 78). The phosphorylation of intracellular choline by choline kinase and the subsequent production of CDP-choline, the main precursor of PC, by CTP:phosphocholine cytidylytransferase (Pcyt1) are the principal pathways for choline utilization and have been extensively studied and are reasonably well understood (23, 63, 69). Oddly enough, the regulation of choline transport, although critical for CDP-choline and PC production, has not been well characterized.
Choline is a positively charged molecule (9) unable to freely pass through the cell membrane and thus requires the assistance of transmembrane proteins to gain access into cells. The long search for a specific choline transporter recently concluded with the discovery of the first choline-specific transporter, CHT1 (37–39), which is responsible for the supply of choline destined for acetylcholine biosynthesis in neurons.
Importantly, choline transporter-like (CTL) protein 1, a member of a second family of transporters distinct from CHT1, has been concurrently identified in the eel [Torpedo (t)CTL1 (40)], rat [rCTL1 (40)], mouse [mCTL1 (73)], and human [CDw92/hCTL1 (71)]. The physiological function of CTL proteins is less evident, and they have been suggested as candidate transporters for the provision of choline that is used in membrane lipid biosynthesis (40). mCTL1 protein is predominantly expressed in skeletal muscle and is barely detectable in other tissues (73). rCTL1 mRNA is found solely in the colon, ileum, brain, and spinal cord (40, 60), whereas hCTL1 protein, also named CDw92, is initially detected as a cell surface antigen in monocytic and dendritic cells (72).
The regulatory promoter and expression of hCTL1 at both the transcriptional and translational levels have not been characterized, and we describe here the first study in this area. The purpose of our work was to establish the links between hCTL1 gene structure, promoter activity, and expression of two COOH-terminal splice variants. We demonstrated that hCTL1 is a “housekeeping” gene that has a strong TATA-less promoter, with consensus cis-regulatory elements for ubiquitous transcription factors such as Sp1, and protein expression in multiple human tissues, including normal and pathological human skeletal muscle.
MATERIALS AND METHODS
Cell culture and human tissues.
The MCF-7 human breast cancer cell line was maintained in minimum essential medium (MEM; HyClone) supplemented with 10% FBS (Sigma). Mouse C2C12 cells were maintained in high-glucose DMEM (Sigma) supplemented with 10% FBS (Sigma). Cells were grown in 60- or 100-mm tissue culture dishes at 37°C in a humidified incubator with 5% CO2.
Human muscle (vastis lateralis) biopsy samples were obtained at the time of a diagnostic muscle biopsy from subjects (34–62 yr old) who were referred for the evaluation of possible myopathy. After a histological (light, immunohistology, and electron miscroscopy), electromyographic, genetic, and serological assessment, these participants were classified as having no neuromuscular disease [healthy (n = 6)] or having a myopathy [nemaline rod congenital myopathy (n = 1), inflammatory myopathy (n = 2), limb girdle muscular dystrophy (n = 3), and mitochondrial myopathy (n = 1)]. All participants provided written informed consent for the biopsy and for “biological waste” samples to be used in future research in accordance with the guidelines set forth by the McMaster University Medical Ethics Committee. These samples were kept at −86°C and remained after the completion of a previous study (59) evaluating protein and mRNA content in human skeletal muscle.
5′-Rapid amplification of cDNA end determination of the transcriptional start site.
To locate the transcriptional start site(s) of the hCTL1 gene, 5′-rapid amplification of cDNA ends (RACE) was carried out using a 5′-RACE kit (GIBCO). Approximately 11 μg of human hepatoma HepG2 (Geneka) mRNA were reverse transcribed using murine leukemia virus reverse transcriptase and an oligo-dT primer. After first-strand DNA synthesis, 2 units of RNase H were added to degrade mRNA, and cDNA was tailed and amplified by PCR using an abridged anchor primer provided in the 5′-RACE kit and combined with the hCTL1 gene-specific primer CTL1RP2 (5′-TCACAGGAGCACAGCGATG-3′). PCR was performed under the following conditions: 3-min denaturation at 94°C followed by 35 cycles at 94°C for 45 s, 56°C for 30 s, and 72°C for 1 min, and the reaction was completed with a final extension for 10 min at 72°C. To ensure a higher specificity for hCTL1, a nested PCR amplification was performed using the same abridged anchor primer and the nested gene-specific primer RP1.2MD (5′-TTGGACACGCTGCTACACAC-3′). The PCR product was subcloned into the pCR2.1 vector (Invitrogen), and multiple clones were confirmed by sequencing. The hCTL1 transcription initiation (start) site was determined from the longest 5′-end product after a comparison with human expressed sequence tag (EST) sequences.
3′-RACE identification of the hCTL1 3′-end.
To identify the 3′-end of hCTL1, 3′-RACE was performed using a 3′-RACE kit (GIBCO). To assure a bigger transcript pool, total mRNA was isolated from HepG2 human hepatoma cells or normal human blood. Five micrograms of total mRNA were reverse transcribed using murine leukemia virus reverse transcriptase and a 3′-adapter primer (AP) provided by the manufacturer. cDNA was amplified by PCR using the abridged universal anchor primer AUAP (GIBCO) and hCTL1 primer FP1 (5′-AGTGCTGATGGAGTTTGTGGA-3′). A nested PCR was performed using the hCTL1-specific nested primer FP2 (5′-GAGAATGCTTTGCGAGTGG-3′) as the forward primer and AUAP as the reverse primer. PCR was conducted under the following conditions: 3-min denaturation at 94°C followed by 35 cycles at 94°C for 45 s, 55°C for 30 s, and 72°C for 1 min, and the reaction was completed with a final extension for 10 min at 72°C. The final PCR products were subcloned into the pCR2.1 vector (Invitrogen), and multiple clones were subjected to sequencing. The 3′-RACE reaction was repeated at least five times with similar results.
Cloning of the hCTL1 promoter.
To study the regulatory promoter, a 920-bp 5′-flanking region of the hCTL1 gene was amplified from MCF-7 genomic DNA using primers based on the location of the transcription start site determined by 5′-RACE and the hCTL1 genomic structure. Genomic DNA was isolated using TRIzol (Invitrogen) and amplified with the forward primer hCTLF2 (5′-GAGCTCCAGTATGGCTTTGACTTGCGAGCC-3′), which was derived from the hCTL1 genomic sequence (NT008470) and contained an additional SacI site at its 5′-end (underlined), and the reverse primer hCTLR (5′-AAGCTTATGGCGGGGCCGAAGGAGCC-3′), which was derived from the 5′-RACE data and contained an additional HindIII site at its 5′-end (underlined). The PCR was performed on 1 μg of genomic DNA, and the resulting fragment was ligated into a PCR (pDrive, Invitrogen) vector for sequencing and further cloning. The CTL1 promoter insert was released from the pDrive vector and ligated into the SacI/HindIII cloning site of the promoterless pGL3-Basic vector (Promega) containing luciferase cDNA as a reporter gene. This first promoter-luciferase reporter construct was designated as LUC.CTL1 (−897/+27 bp).
To further characterize the hCTL1 regulatory promoter, 5′ truncations and internal deletions of the LUC.CTL1 promoter-luciferase reporter were performed as shown in ⇓⇓Fig. 3. To delete 401 bp of the 5′-upstream region, the LUC.CTL1 plasmid was digested with XhoI and HindIII, and the resulted fragment was subcloned into the pGL3-Basic vector to produce LUC.CTL2 (−497/+27 bp). For internal deletions, both LUC.CTL1 and LUC.CTL2 plasmids were digested with SmaI (New England Biolabs), and the larger fragments were religated using T4 DNA ligase (Invitrogen). The resulting deletion clones, LUC.CTL1Δ and LUC.CTL2Δ, were of the same type as the original clones but without the internal −56 to −318-bp region. Finally, one more 5′-deletion was prepared from LUC.CTL1 after digestion with SacI and MluI, blunt ending, and religation to produce LUC.CTL3 (−188/+27 bp)
To test for promoter activity, the CTL1 promoter-reporter constructs were transiently transfected into MCF-7 human breast cancer cells and C2C12 mouse muscle cells, along with the promoterless pGL3-Basic vector as a negative control. Cells were plated in 60-mm dishes, grown until 80% confluent, and cotransfected with 5 μg of the appropriate plasmid and 25 ng of pCMV-Renilla luciferase plasmid (Promega), which served as an internal control for transfection efficiency. Cells were transfected with plasmid DNA using DOTAP lipid reagent (Avanti Polar Lipids), and, 4 h after transfection, the serum-free medium was replaced with serum-containing medium. After 48 h, transfected cells were harvested in passive lysis buffer (Promega), and their luciferase activity was analyzed by luminometry (Luminometer TD-20/20, Turner Designs) using Promega reagents. Each experiment was performed in triplicate and repeated at least four times under identical conditions.
Preparation of nuclear extracts and gel-shift assays.
MCF-7 cells were grown in MEM supplemented with 10% FBS as described in Cell culture and human tissues, and ∼107 cells/10-cm dish were washed with 5 ml PBS (160 mM NaCl, 3 mM KCl, 20 mM Na2PO4, and 3 mM KH2PO4), wrapped with parafilm, and kept for 1 h at −80°C. The frozen cells or 100 mg of human skeletal muscle were lysed on ice for 3 min with 0.5 ml of lysis buffer [10 mM HEPES-KOH (pH 7.9), 10 mM KCl, 1.5 mM MgCl2, 250 mM dl-dithiothreithol, 10 μg/ml each of leupetin, aprotinin, and soybean trypsin inhibitor, 0.5 mM PMSF, 1 μg/ml pepstatin A, and 0.01% Nonidet P-40]. The lysate was centrifuged (850 g for 5 min at 4°C), the pellet was resuspended in the same buffer, and the centrifugation was repeated. The pellet was then dissolved in the same buffer and centrifuged once more at 6,000 g, the supernatant was discarded, 40 μl of extraction buffer [20 mM HEPES (pH 7.9), 420 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 25% glycerol, 250 mM dl-dithiothreithol, 10 μg/ml each of leupetin, aprotinin, and soybean trypsin inhibitor, 0.5 mM PMSF, and 1 μg/ml pepstatin A] were added, and the pellet was kept on ice for 10 min with occasional vortexing. Subsequently, the lysate was centrifuged (9,300 g for 5 min at 4°C), and the final supernatant (nuclear extract) diluted with 60 μl of dilution buffer [20 mM HEPES (pH 7.9), 50 mM KCl, 0.2 mM EDTA, 20% glycerol, 250 mM dl-dithiothreithol, 10 μg/ml each of leupetin, aprotinin, and soybean trypsin inhibitor, 0.5 mM PMSF, and 1 μg/ml pepstatin A]. The C2C12 myoblasts were maintained in DMEM supplemented with 10% FBS and were differentiated into myotubes in DMEM containing 2% FBS. Myoblast and myotube nuclear proteins were extracted as described above. The protein concentration of the nuclear extracts was determined by bicinchoninic protein assay (Pierce), and nuclear extracts were aliquoted and kept at −80°C before being used.
On the basis of the CTL1 promoter deletion analysis and structure, we synthesized double-stranded oligonucleotides from four promoter regions: S1 (−92/−61 bp), S2 (−174/−145 bp), S3 (−289/−260 bp), and S4 (−442/−410 bp). To identify potential transcription factors, based on the sequence analysis, Sp1 and E2F consensus oligonucleotides (Sp1: forward 5′-TCCCAGCCCCGCCCCGTCTGC-3′ and reverse 5′-GCAGACGGGGCGGGGCTGGGA-3′; E2F: forward 5′-AGGTGTTTTTGGCGGGAGTTGGGGG-3′ and reverse: 5′-CCCCCAACTCCCGCCAAAAACACCT-3′) were synthesized and used as specific competitors. Gel-shift analysis was performed using a nonradioactive “LightShift” chemiluminescent assay (Pierce). Single-stranded oligonucleotides were 3′-end labeled using a biotin 3′-end DNA labeling kit, and complementary strands were annealed as described previously (4). The binding between labeled probe and nuclear proteins was carried out for 20 min at room temperature in 20 μl of shift buffer (1× binding buffer, 2.5% glycerol, 0.05% Nonidet P-40, and 500 mM KCl, supplemented with 1 μg poly-dIdC). The reaction contained 20 fmol of 3′-end-labeled probe and 12 μg of nuclear protein. For the competition assays, 20-fold of unlabeled DNA probe was added to the same reaction mixture. All reactions were loaded on a nondenaturing 6% polyacrylamide gel, and the electrophoresis performed in 0.5× Tris-borate-EDTA buffer at 100 V at 4°C for 4 h. Reaction products were transferred to a Nylon+ membrane (Roche) at 380 mA at 4°C for 30 min, membranes were cross linked for 15 min on a transilluminator at 312 nm, and biotin-labeled DNA was detected by chemiluminescence (Pierce).
To identify some of the hCTL1 promoter-binding protein(s), chromatin immunoprecipitation (ChIP) was carried out using a QuikChip ChIP Kit (IMGENEX). Human breast cancer MCF-7 cells and mouse C2C12 myoblast cells were fixed in 1% formaldehyde solution at 37°C for 10 min. The fixed cells were harvested with 1 ml SDS lysis buffer (provided by the kit) and sonicated for the preparation of sheared chromatin. Promoter immunoprecipitation was performed using anti-Sp1 polyclonal antibody (Santa Cruz Biotechnology) or an unrelated Pcyt2 polyclonal antibody as a negative control. Precipitates were reverse cross linked for DNA isolation and PCR analysis. The PCR primers flanking the potential Sp1-binding site S3 (from −289 to −260 bp) were designed as the forward P1 primer 5′-GAGTCGATGGGAGACGCGC-3′ (−363/−343 bp) and the reverse P2 primer 5′-CAGCTGCTGCAGCAGGAGCA-3′ (−218/−199 bp) to amplify a 169-bp fragment of the hCTL1 promoter region.
Expression of isoform-specific mRNAs.
The expression of hCTL1 splice variants was assessed by RT-PCR using 3′-end-specific primers. Total RNA from several human cell lines [THP-1 (monocytes), THP-1 + PMA (differentiated macrophages), U937 (monocytes), U937 + PMA (differentiated macrophages), Du145 (prostate), HepG2 (hepatoma), and Kelly (neurons)] was purchased from Genika Biotechnology. Total RNA from normal human skeletal muscle and skeletal muscle from subjects with inflammatory myopathy, muscular dystrophy, and mitochondrial myopathy was isolated using TRIzol reagent (Invitrogen). Total RNA from the aforementioned cell lines and human skeletal muscle was reverse transcribed using Superscript II reverse transcriptase (Invitrogen) and an oligo-dT primer. Total cDNA of a human tissue panel (brain, heart, liver, kidney, and placenta) was obtained from Ambion (Austin, TX). To amplify a CTL1a-specific fragment, PCR was carried out using the isoform-specific primers FP2 (5′-GGGTTGGACCTTTAAATTGC-3′) and TP2 (5′-GCAATGTTGGTTTCCCAGAT-3′) and to amplify a CTL1b-specific fragment using primers FP1 (5′-TGCTGATGGAGTTTGTGGAA-3′) and TP1 (5′-CTGGTGAGTGGGAGAACCTC-3′) under the following conditions: 3 min of initial denaturation at 94°C followed by 35 cycles of 45 s at 94°C, 30 s at 53°C, and 90 s at 72°C, with a single final extension at 72°C for 10 min. The amplified fragments of ubiquitously expressed β-actin (forward primer 5′-CCCAAGGCCAACCGCGAGAAGAT-3′ vs. reverse primer 5′-GTCCCGGCCAGCCAGGTCCAG-3′) or GAPDH (forward primer 5′-TCCACCACCCTGTTGCTGTA-3′ vs. reverse primer 5′-ACCACAGTCCATGCCATCAC-3′) were used to standardize the quality of mRNA preparations and expression by densitometry (NIH Image). Each experiment was repeated a minimum of three times.
Protein expression in human tissues.
To investigate the tissue distribution of the hCTL1 protein, the multiple-tissue protein panel (INSTA-Blot, Imgenex) and normal and pathological human skeletal muscle were examined by Western blot analysis using the CTL1 monoclonal antibody VIM15 (Research Diagnostics). The human samples collected by muscle biopsies were homogenized in lysis buffer [50 mM Tris·HCl (pH 7.4), 1% Nonidet P-40, 0.25% sodium deoxycholate, 10% glycerol, and 150 mM NaCl] supplemented with protease and phosphotase inhibitor cocktails (Sigma). Tissue lysates of 30 μg of total protein were denatured, resolved on 10% SDS-PAGE, and transferred to polyvinylidene difluoride membranes (Millipore). The membranes, including the human multiple-tissue INSTA-Blots, were blocked with 5% nonfat dry milk in Tris-buffered saline [TBS; 25 mM Tris·HCl (pH 7.4) and 0.15 M NaCl] for 1 h at room temperature and then incubated with VIM15 antibody (1:1,000 dilution) at 4°C overnight, as previously described (73). Afterward, membranes were extensively washed with TBS-Tween 20 (0.1%), incubated with horseradish peroxidase-conjugated goat anti-mouse antibody (1:10,000 dilution, Promega), and visualized by enhanced chemiluminescence (Sigma).
To obtain the hCTL1 gene structure, the hCTL1 genomic (Accession No.NT008470) and cDNA sequence (Accession Nos. NM_080546 and AJ245620) alignments were performed using the BLAST database at the National Center for Biotechnology Information (NCBI). Similarity searches for CTL1a and CTL1b mRNAs and their derived protein sequences were performed using Procite and BLAST. Searches for potential transcription factor-binding sites within the hCTL1 promoter region were performed using MatInspector 2.2 (46). To compare regulatory promoters among hCTL1, mCTL1, rCTL1, and chimpanzee CTL1 genes, genomes of those species were aligned with the hCTL1 promoter using the VISTA server (10, 12).
Gene structure and splicing.
To our knowledge, no detailed analysis of the hCTL1 gene organization, promoter activity, and expression was available before this work. We first reviewed human ESTs and confirmed the presence of two full-length hCTL1 cDNAs, with the longer form initially designated as hCTL1a (Accession No. NM_080546, 45) and the shorter form as hCTL1b (Accession No. AJ245620). The hCTL1a cDNA is 4,301 bp in length, including 42 bp of the 5′-untranslated (UTR) region, 1,974 bp of the coding region, and 2,285 bp of the 3′-UTR (Fig. 1). The hCTL1b cDNA is shorter (2,224 bp), has an identical 5′-UTR, a 1,965-bp coding region (9 bp less than CTL1a), and a shorter (217 bp) and distinct 3′-UTR. CTL1a and CTL1b share most of the coding region (1,951 bp) and have identical 5′-UTRs, suggesting that they are transcribed from the same promoter but represent 3′-end splice variants. The last 21 bp (1,994–2,015 bp) of CTL1a encode the COOH-terminus peptide A651SGASSA657, whereas the last 12 bp of CTL1b encode L651KKR654 (Fig. 1) (60, 72).
To better understand the hCTL1 gene organization that directs the production of the two gene products, we examined the hCTL1 genomic contingent at chromosome 9q31.2 (Accession No. NT_008470) (Fig. 1). The hCTL1 gene is considerably large and spans over 194 kb. It consists of 17 exons and 16 introns, of which the last 2 exons could be alternatively spliced. In the entire gene, the exon sizes range from 75 bp (exon 1) to 2,221 bp (exon 17, only present in CTL1b). The intron sizes vary from the shortest intron, intron 10 (919 bp), to the largest intron, intron 1 (54 kb). In addition, the first intron, shared by both isoforms, and the last intron of the CTL1b isoform are particularly large, over 100 kb in total, comprising more than one-half of the entire gene size.
hCTL1a is made of exons 1–16, contains a long 3′-UTR (∼2 kb), and could potentially have an alternative stop codon within exon 16, producing yet another splice variant, named hCTL1c (60). On the other hand, hCTL1b uses exons 1–15 and exon 17 instead of exon 16. Exon 17 is rather distant (∼47 kb) from exon 16. Because of differential splicing, the CTL1b transcript has a smaller and structurally distinct 3′-UTR from the CTL1a transcript, which allowed us to distinguish their specific expression, as described below. Because it had not been performed before, we attempted to verify the mRNA ends for hCTL1a and hCTL1b variants by using 3′-RACE (Fig. 2B). Even after multiple attempts, we could not obtain any 3′-RACE product for hCTL1a from HepG2 cells. However, we were able to monitor the expression of this transcript in various human tissue preparations, as shown later, using the set of specific hCTL1a primers. To test for the hCTL1b 3′-end, and the utilization of distant exon 17, we successfully amplified HepG2 mRNA by 3′-RACE and established that the CTL1b 3′-end and the 3′-UTR (Fig. 2B) were as anticipated (Accession No. AJ245620). Taken together, our RACE data agreed with the expression data for the rat forms, confirming that the 3′-end splicing is conserved between hCTL1 and rCTL1 genes (60).
Isolation of the regulatory promoter.
To determine the 5′-end and the transcriptional initiation site of the hCTL1 gene, we performed 5′-RACE on HepG2 mRNA as for 3′-RACE. 5′-RACE extended the hCTL1 cDNA up to 27 bp upstream of the translation start codon ATG (Fig. 2A). After comparing this with the hCTL1 genomic sequence, we concluded that the regulatory promoter lies immediately upstream of the first exon, and its structure is shown in Fig. 2C. On the basis of this information, we amplified the 5′-flanking (promoter) region of the hCTL1 gene from MCF-7 cells (Fig. 2C). A 925-bp promoter fragment was isolated, cloned into a promoterless luciferase vector (pGL3-Basic), and then tested for promoter activity (LUC.CTL1 −898/+27 bp in Fig. 2C). The hCTL1promoter was very active and able to drive luciferase expression in both MCF-7 and C2C12 cells (Fig. 3, B and C). Although not shown, we also established that the hCTL1 gene has a very strong promoter similar in activity to the simian virus 40 promoter of the pGL3-control vector (Promega) and the mouse promoter of Pcyt1 [previously studied extensively by us (4)].
To identify the minimal (core) promoter required for transcription, we generated LUC.CTL3 (−188/+27 bp) from LUC.CTL1, in which the 710-bp upstream sequence was deleted (Fig. 3A). In breast cancer cells (MCF-7), this deletion significantly diminished (by 50%) luciferase activity relative to the full-length promoter LUC.CTL1 (Fig. 3A). On the other hand, in muscle cells (C2C12), the same deletion increased (1.6-fold) luciferase activity (Fig. 3B). We concluded that the promoter region of LUC.CTL3 is sufficient to support basal transcription; however, the upstream region (−497/−188 bp deleted from LUC.CTL2) had an opposite effect on the promoter activity in human cancer cells relative to its activity in mouse muscle cells. To characterize the distal promoter regions, we generated three additional constructs, as shown in Fig. 3A. The 5′-region from −898 to −497 bp was deleted in LUC.CTL1 to produce LUC.CTL2 (−497/+27 bp). Also, the internal region from −318 to −56 bp was deleted in both LUC.CTL1 and LUC.CTL2 to generate the mutants LUC.CTL1Δ and LUC.CTL2Δ. Luciferase assays showed that the 5′-deletion of ∼400 bp did not significantly change promoter activity (LUC.CTL1 vs. LUC.CTL2) in either cell line (Fig. 3, B and C). This suggests that the deleted (−898/−497 bp) upstream region is not critical and that the major promoter function is enclosed within the LUC.CTL2 reporter. On the other hand, the internal deletions from −318 to −56 bp in both longer (LUC.CTL1Δ) and shorter (LUC.CTL2Δ) constructs completely abolished promoter activity in both cell types (Fig. 3, B and C), strongly implying that this region contains regulatory elements essential for promoter function and hCTL1 gene transcription.
The longer promoter (−898/+27 bp, LUC.CTL1) and the putative cis-acting elements within the essential ∼500-bp region (LUC.CTL2) are shown in Fig. 2C. The region is rich in consensus-binding elements and possibly contains overlapping sites for multiple transcription factors: Sp1/GATA1/myeloid zinc finger (MZF)1, Ikaros (IK2)/MZF1/MyoD, MZF1/GATA1/2/IK2, acute myeloid leukemia (AML)1a/neurofibromin 1 (NF1), and Sp1/MZF1. Single and double binding sites were also found for IK2, Sp1, GC, and E2F/NF-κB transcription factors. No canonical TATA or CCAAT boxes were present in the vicinity of the transcription start site ggcgC+1cgcct (Fig. 2C and determined by 5′-RACE), suggesting that the nearby GC-rich region (up to the −68 bp; Fig. 2C) contributes to the initiation of transcription.
On the basis of the reporter gene assays (Fig. 3, B and C) and the type of consensus cis-elements residing within the promoter sequence (Fig. 2C), we also designed in vitro electromobility gel-shift assays to probe for interactive DNA-regulatory protein complexes. The results are shown in Figs. 4 and 5. Double-stranded oligonucelotide probes, designated as S1–S4, were prepared from four promoter regions, and the sequences and locations of these are shown in Fig. 2C. The potential binding factors and the tested probes were: AML1a/NF1 (S1; −92/−61 bp), MZF1/GATA1/2/IK2 (S2; −174/−145 bp), Sp1/IK2 (S3; −289/−260 bp), and E2F/NF-κB (S4; −442/−410bp). The data shown in Fig. 4 are gel shifts obtained using human breast cancer cells (A), normal human muscle (B), and C2C12 cells (myoblasts and myotubes; C). The data shown in Fig. 4, A and B, demonstrates that all four oligonucleotide probes (S1–S4) produced specific DNA-protein bands. The S3 probe (Sp1/IK2) produced two bands when incubated with cancer cell proteins and a single band when incubated with human muscle proteins. In addition, the S3 (Sp1/IK2) and S4 (E2F/NF-κB) probes also had very different mobilities and intensities depending on whether they were incubated with breast cancer or human muscle proteins, suggesting that the complexes could belong to different regulatory factors in those two human protein preparations.
We focused more on the hCTL1 promoter region from −497 to −188 bp (deleted from LUC.CTL2 to make LUC.CTL3), which had an inhibitory effect in C2C12 myoblasts and a stimulatory effect in MCF-7 cancer cells (compare Fig. 3, B with C). Because the S3 and S4 probes belonged to this region, the binding was tested in more detail using competitive gel-shift and supershift analyses and ChIP (Figs. 4 and 5). The S4 region was tested for E2F factors, as shown in Fig. 4C. Because of the limited amount of human tissue samples, binding assays were mostly performed with C2C12 cells. E2F1/2 was abundantly expressed in C2C12 cells (data not shown); however, the E2F1/2 antibodies did not supershift the S4-bound proteins. Multiple testing produced partial competition between labeled S4 and an excess of unlabeled S4 or E2F consensus probes (Fig. 4C). Finally, when the E2F site was mutated (CGCGCC to GCatGG), the mutation increased the CTL1 promoter activity in C2C12 cells (Z. Yuan and M. Bakovic, unpublished observations). Altogether, these data suggest that the E2F site within the S4 region could contribute to the promoter activity, but the contributing factors are likely not related to E2F1/2 proteins. It could be that other E2F family members, like E2F4 or E2F5, might be involved; however, that needs to be further investigated.
The S3 probe was tested for the transcription factor Sp1. As shown in Fig. 5A, C2C12 myoblasts and human skeletal muscle contain Sp1-like proteins that strongly bind to this region. The binding could be partially (C2C12) or completely (human muscle) competed with an Sp1 consensus oligonucleotide. Myotubes apparently did not contain a significant amount of Sp-related proteins because only a very weak band was detected with the same probe (Fig. 5A). When myoblast proteins were competed with Sp1 antibody, a supershift of the lower band was obtained, whereas the upper band did not change significantly, suggesting that factor(s) additional to Sp1 could be bound to the S3 probe (Fig. 5B). To confirm that the Sp1 transcription factor belongs to at least one of the protein complexes, we performed in vivo ChIP assays using the same antibody and promoter primers P1 and P2 (shown in Fig. 2C), which encompass the Sp1-binding site within the S3 probe. As shown in Fig. 5C, the CTL1 promoter region was amplified as expected with the actual size of the PCR fragment of ∼170 bp. The same-size product was amplified from myoblasts as was from MCF-7 cells, but no promoter DNA was amplified from the myotube chromatin, which could be predicted given that myotube proteins did not bind to the S3 probe (Fig. 5A). Taken together, the above data strongly suggests that Sp1 binds at the S3 site of the promoter region. Sp1 likely interacts with other factors (regions) depending on the cell type, which explains why the −497- to −188-bp region inhibited luciferase reporter activity in myoblasts and stimulated the activity in cancer cells (shown in Fig. 3B). In addition, the S3 site did not bind Sp1 from myotubes, a feature that could be important for the regulation of CTL1 gene transcription in myoblasts (undifferentiated cells) relative to myotubes (differentiated cells). As we describe later, the microarray data (Table 1) as well as our unpublished data both suggest that myotubes actually express more CTL1 mRNA than myoblasts and have a stronger promoter activity, essentially when Sp1 is absent from the S3 site.
To relate the regulation of the hCTL1 gene with other CTL1 genes, the analysis of the 5′-flanking regions from several mammalian species was performed using the VISTA multiple-alignment server, which revealed high structural conservation among CTL1 promoters, as shown in Fig. 5D. mCTL1 and chimpanzee CTL1 promoters were strongly conserved from −898 to −223 bp from 99% to 93% compared with humans (Fig. 5D,a), and the same region was less conserved in the rat promoter (not shown). On the other hand, the proximal promoter regions (−256 to + 26 bp) were well conserved among human, mouse (95.7%), and rat (94.7%) genes (Fig. 5D,b) but less conserved in the chimpanzee gene (not shown). Considering those similarities among mammalian CTL1 promoters, one might anticipate more similar regulation of hCTL1 and mCTL1 transcription because of a strong similarity (93%) for the entire promoters of up to ∼900 bp. To some extent, divergent regulation of the rat gene, which is different structurally from human and mouse genes in its distal promoter regions, could be expected. Essentially, this is in agreement with a more restricted expression of the rat gene in neuronal and intestinal tissues (60), as opposing to the broad expression of the mCTL1 and hCTL1 genes, at least at the mRNA level (73). No data are available for chimpanzee gene, and mRNA and protein expression data for the human form are described below, demonstrating a close relationship with the mouse gene. Altogether, the relevant information so far regarding hCTL1 promoter activity is the importance of the overall length (−494 and +27 bp), the nature of the internal sequence (−318 to −56 bp), and the basal elements (−188 to +27 bp) of the core promoter, which is highly conserved across the human and rodent species. All promoters are lacking TATA or CAAT-like motifs but contain GC-rich sites in the vicinity of the initiation start site (designated as +1 in Fig. 2), which could initiate transcription as typical for many ubiquitous promoters, for example, the cytomegalovirus promoter and the Pcyt1 promoter (4, 5). All CTL1 promoters also contain conserved response elements for potential transcription factors common for housekeeping genes (Sp1 binding demonstrated for mouse and human muscle in Fig. 5) and with possible additional responses to a large variety of stimuli and physiological conditions, which will be more apparent when the promoters from other species become available in the future.
Distribution of splice variants.
To establish how the hCTL1a and hCTL1b transcripts are distributed in human tissues, we analyzed the multiple-tissue panel of brain, heart, liver, kidney, and placenta cDNAs. The expression data relative to the GAPDH control revealed that hCTL1a is abundant in the placenta and brain, mildly expressed in the kidney, and significantly less present in the human heart and liver (Fig. 6A,a). On the other hand, hCTL1b mRNA was dominantly expressed in the liver and evenly, but significantly less detectible, in other organs (Fig. 6A,b). A comparison of the relative transcript ratios (Fig. 6B) clearly revealed some organ specificity in transcript distribution. The human liver dominantly expressed hCTL1b, whereas the brain and kidney expressed more hCTL1a than hCTL1b. No significant difference in their distribution was detected in the placenta (both are high) or the heart (both are low). Further testing of hCTL1 splicing was performed on multiple human cell lines, including monocytic cells (THP-1 and U-937), prostate cancer cells (DU145), neuroblastoma (Kelly), and hepatoma (HepG2) cells (Fig. 7). Overall, the results demonstrated that the splicing mechanism is preserved in those human cell lines, and both CTL1a and CTL1b transcripts were consistently detected. Variations in CTL1a mRNA could reach 2- to 3-fold (Fig. 7A), whereas the expression of hCTL1b mRNA was more constant, showing a 1.5- to 2-fold variation among different cells (Fig. 7B). In addition, the human blood cells (THP-1 and U397) maintained a similar ratio of the two variants after differentiation with phorbol esters to macrophages (Fig. 7, A vs. B). Interestingly, human hepatoma cells (HepG2) expressed CTL1a and CTL1b in equal proportions, differently from the normal human liver, where CTL1b was the overly dominant form (compare Figs. 6 and 7).
Protein expression in human tissues.
Although hCTL1 protein was initially detected as a surface protein in human monocytic cells (72), little is known about its distribution in human tissues. We performed Western blot analysis using the hCTL1 monoclonal antibody VIM-15 (72) and multiple human tissues (INSTA-Blot protein panel), as shown in Fig. 8A. hCTL1 protein was detected in all tissues (Fig. 8A) as a major ∼50-kDa band. The band size was similar to those obtained with mouse tissues and human cell lines THP-1 and MCF-7 (73). The major difference is that the hCTL1 protein was uniformly abundant (Fig. 8A), whereas the mCTL1 protein was abundant only in skeletal muscle (73). The protein molecular mass calculated of the human and mouse cDNAs was similar and ∼73 kDa (73). The apparent molecular masses of the mature ∼50-kDa band and the additional 23-kDa band summed to that of the theoretical protein, suggesting that some type of proteolytic degradation of the initially translated protein is required to produce the mature 50-kDa form (73). The appearance of multiple bands for other membrane proteins including neuronal CHT1 is frequently attributed to similar proteolytic processing (28, 44).
Previously, we hypothesized that CTL1 is more abundantly expressed in mouse muscle relative to other tissues because it might have a specialized functional role in the muscle (73). We tested for hCTL1 protein expression in healthy human muscle and different muscle pathologies. We found out that the ∼50-kDa hCTL1 band in the samples of congenital myopathy (Fig. 8B, lane 1), muscular dystrophy (Fig. 8B, lanes 2–4), inflammatory myopathy (Fig. 8B, lanes 5 and 6), and mitochondrial myopathy (Fig. 7B, lane 7) was significantly stronger than in healthy human muscle (Fig. 8B, a, lane 8). In particular, a very robust ∼50-kDa doublet, with an appearance of an additional band at ∼31 kDa, was apparent in inflammatory myopathies. Quantitatively, when samples were pooled together, the higher band was approximately threefold stronger for the pathological muscle samples than in normal muscle (Fig. 8D). When we investigated the mRNA expression profiles in healthy muscle and different human muscle pathologies (Fig. 8C), we detected only slight increases (20–40%) or no change in the hCTL1 mRNA levels among the pathologic muscle samples. Collectively, these data suggest that an increased hCTL1 protein (band intensity in Fig. 8B) mostly manifests posttranslational modifications (band shifts and doublets) indicative of an activity in pathological states not observed in healthy human muscle.
The CTL1 family of transporters consists of newly discovered transmembrane proteins (40) proposed to supply choline for phospholipid biosynthesis (40, 60, 73) and possibly having an additional yet unidentified function in immune cells (72). The functionality of rCTL1 and tCTL1 cDNAs as choline transporters was established after their expression in a choline auxotrophic mutant ctrl-ise (hnm1-ise) (40) and Xenopus oocyces (41) and the expression of mCTL1 cDNA in mammalian cells (73). The function of CTL1 proteins as CHTs was disputed after tCTL1 expression in a pns1p-deficient yeast mutant did not affect choline transport (79); however, it has been recently reaffirmed when the expression of both rCTL1a and rCTL1b variants in mouse neuroblastoma (N18) cells, known to lack CTL1, stimulated choline transport, as was recently found (60).
The expression of rat CTL1 is strongly tissue specific, with rCTL1a expressed only in the ileum and colon and rCTL1b exclusively expressed in the brain (60). Mouse mRNA is broadly detected; however, mCTL1 protein is robustly detected in skeletal muscle, suggesting a complex regulation of mCTL1 expression. To better understand the tissue-specific expression and role of alternative splicing of the CTL1 family of proteins, we first analyzed the structure of the hCTL1 gene and isolated its regulatory promoter. The hCTL1 gene is located at chromosome 9q.31.2, whereas the rat and mouse homologs are located at chromosomes 5q.24 and 4.B2, respectively. Both mCTL1 and rCTL1 genes are positioned in regions syntenic to human chromosome 9, where the order of the neighboring genes is highly conserved (Fig. 9). For example, the olfactory receptor and ABC1 form of ATP-binding cassette transporters are upstream of the CTL1 gene, and the Fukuyama type congenital muscular dystrophy homolog (fucutin) and T cell acute lymphocytic leukemia 2 genes are downstream from the CTL1 gene in all three species. CTL1 is a transmembrane protein like the two upstream genes. Interestingly, CTL1 is upregulated in myopathies (protein data in Fig. 8B) and leukemias (microarray data in Table 1).
The hCTL1 promoter is within the 5′-flanking region immediately upstream of the first exon. It is a TATA-less and GC-rich promoter, with multiple binding sites for ubiquitous transcription factors of the Sp1 family, strongly conserved among human and rodent promoters. The reporter assays with human MCF-7 and mouse C2C12 cells revealed that the promoter is differently activated in those two cell lines. Because putative protein binding is predicted to be identical in specific promoter regions (Fig. 5), the variation in transcription factors may be responsible for the observed differences between mouse and human cells (Fig. 3). When the factors from different human tissues were compared (breast cancer and skeletal muscle), two promoter regions (S3 and S4; Fig. 4) formed complexes having different strength and mobility, which suggests binding of different regulatory factors from normal and cancer human tissues. Furthermore, the factor binding to an additional region, probably binding blood cell-specific transcription factors (S1 and S2; Fig. 4), highlights the importance of multiple regions for the promoter activity that need to be further investigated.
The first exon of the hCTL1 gene contains the ATG translation codon at position +27. The exon is, however, separated from the second exon by ∼50 kb. We tried to determine whether intron 1 contains a second promoter using 5′-RACE. We obtained two 5′-end products containing the second exon; however, after they were tested for activity, no intronic promoter could be identified (data not shown). We concluded that hCTL1 is predominantly transcribed from the main upstream promoter that is responsible for the majority of hCTL1 transcription.
This study is the first to address the transcriptional regulation of the hCTL1 gene, and there is no published information available with which to compare our results. To understand some of the above results, we examined the CTL1 microarray expression profiles deposited in the NCBI Gene Expression Omnibus (GEO) database (Table 1). On the basis of mouse transcriptome data, mCTL1 mRNA is ubiquitously expressed (GEO No. GDS592), as we initially reported (73). Also, hCTL1 transcription correlated with the type of blood cells investigated and the type of reagent used for activation and differentiation (GEO Nos. GDS601, GDS755, GDS275, and GDS892); the data lean toward the gene’s role in immune cell function (72), which is in agreement with the detection of the hematopoietic transcription factor consensus binding sites within the hCTL1 promoter (Fig. 2). These could include any of the GATA1/2, IK2, MZF1, and AML1 transcription factors. Most of their binding sites are overlapping or in close proximity to each other, and they could interact for the regulation of the hCTL1 promoter. The hematopoietic factors GATA1 and GATA2 are involved in the development of red blood cell lineage (56). Several recent reports have suggested that GATA1 plays multiple roles in survival, proliferation, and differentiation of erythroid cells (42), and GATA2 is required for proliferation/survival of early hematopoietic cells and mast cell formation (61). IK2 and AML1 are oncogenes expressed in human leukemias (21, 58) and regulate the transcription of many important hematopoietic genes (14, 31, 69). Transcription regulator MZF1 is a myeloid zinc finger gene that controls cell proliferation and tumorigenesis (47). It recognizes DNA consensus sequences with a common G-rich core, similarly to Sp1 factors (20). Within the CTL1 promoter, MZF1 frequently has overlapping binding sites with Sp1, GATA1/2, and IK2 (Figs. 2 and 5).
Human mammary epithelial cells (HMECs) express more CTL1 than most breast cancers (GDS90), and the basal-like subtype of invasive breast carcinoma (associated with poor outcomes) expresses more CTL1 than normal tissues (GDS850). Breast cancer cell lines and cancers in general fluctuate broadly in CTL1 mRNA levels (GDS823 and GDS89), suggesting the widespread importance of this gene in cell survival and cancers. Our previous work (73) showed that MCF-7 cells express high levels of hCTL1 (73), and here we demonstrate that MCF-7 cells have the capacity to drive expression from the hCTL1 promoter. Finally, the microarray data suggest that CTL1 is more expressed in differentiated than undifferentiated skeletal muscle cells (C2C12; GEO No. GDS587), mouse smooth muscle cells (P19; GEO No. GDS799), human skeletal myoblasts (KM109; GEO No. GDS942), and differentiated embryonic hepatoblasts (HBC-3; GEO No. GDS970). The transcription factor MyoD is involved in myogenesis and has a binding site (overlapping with IK2 and MZF1) within the hCTL1 promoter also conserved in mCTL1 and rCTL1 promoters (Fig. 5D). The mCTL1 gene is strongly expressed in mouse skeletal muscle (46) and, as shown in this study, in normal and degenerated human muscle (Fig. 8). MyoD is involved in maintaining fast muscle fiber phenotypes; however, it also appears to have other regulatory functions (66). Other myogenic factors, such as the myogenic factor (Myf), the myogenic regulatory factor (MRF), or the myocyte enhancer factor (MEF), could also regulate gene expression in healthy and myopathic human muscles (66). Importantly, skeletal muscle has a remarkable capacity for regeneration largely depending on the number of available satellite cells and their proliferative capacity (34). Injured or diseased muscle also demonstrates increased macrophage cell accumulation; the extent of macrophage accumulation correlates with both the clearance of protein after injury and efficiency of regeneration (13). The increased CTL1 protein in multiple human myopathies (Fig. 8, B and D) is a sign that this gene is responsive to the processes common to most degenerated muscle states, and it could represent a therapeutic target for improving muscle function.
Both hCTL1 and rCTL1 are spliced (60, 72) into longer CTL1a and shorter CTL1b transcripts that differ in their mRNA 3′-UTR structures and COOH-terminal ends (Fig. 1). Variation in the 3′-UTR secondary structure is a known regulatory point of mRNA turnover and stability (1, 54, 72). On the basis of our data (Figs. 6 and 7), hCTL1 splicing is ubiquitous in human tissues and preserved in cultured cells. However, the proportion of transcripts could vary in some tissues. For example, the hCTL1b form was predominant in the liver, whereas the brain and kidney expressed the hCTL1a form in higher amounts. The placenta and heart contained an equal ratio of the two variants as most human cell lines. rCTL1 splice variants were very tissue restricted, with rCTL1a dominating in the ileum and colon and rCTL1b in the brain (60). These data suggest that, in addition to the transcriptional differences discussed earlier, the mRNA turnover could be an additional element for differential expression of rCTL1 and hCTL1 forms.
We believe that the most exciting possibility for the regulation of CTL1 by 3′-end splicing is to derive proteins with distinctive functional properties. Splicing produces the hCTL1a protein with the A651SGASSA657 peptide and hCTL1b with the L651KKR654 peptide at the COOH-terminus. The identical COOH-terminus of hCTL1a is present in rat (CAD12728), Bos taurus (XP_615457), and Pan troglodytes (XP_520165.1) proteins; Xenopus leavis instead has ASGTSTA (AAH82837), and the sequence has not been found in other species. It remains to be established whether this peptide has any role in CTL1b protein localization, transport, or processing. On the other hand, the COOH-terminal L651KKR654 of the CTL1b protein has long been recognized as a classical dilysine (KKXX) motif for endoplasmic reticulum (ER) retrieval of type I transmembrane proteins (27). The KKXX motif is cytosolic and interacts directly with the coat protein I (COPI) coat during retrograde protein transport (7, 29, 65). It is present in the tails of numerous type I transmembrane proteins including ion channels, transporters, and receptors (32, 43, 52). The matching signal in rCTL1b (NP_445944) and mCTL1b (NP_598652) is 649LRKR653, an RXR motif that similarly directs ER localization of signal-bearing proteins (52, 65). Soluble luminal proteins of the ER are retained by a different retention signal, KDEL (35). Thus the CTL1b protein contains the information in its primary sequence for determining its subcellular location at the ER. The COOH-terminal cytoplasmic domain may act as a retrieval signal for returning the CTL1b protein (from the Golgi apparatus) that has left the compartment in which it resides (ER) to ensure its correct localization along the exocytic pathway. This structural information, although firmly supported by evidences for other similar proteins, needs to be further substantiated when more information becomes available in the future.
What precise role the ER resident choline transporter might have is not apparent, despite the fact that the final stage of PC synthesis from CDP-choline and diacylglycerols occurs at the ER membrane. The production of CDP-choline with membrane-activated Pcyt1 is preceded by two steps, choline membrane transport and then choline phosphorylation with choline kinase, which is known to be a cytosolic protein (5, 15, 22, 25). Traditional channeling of intermediates for PC synthesis from the plasma membrane to ER (8, 19) assumes a unidirectional choline supply by a cell surface transporter, and such a role could be perhaps attributed to the CTL1a form of this family of transporters. The ER resident form CTL1b might, however, have a more specialized role, such as to distribute choline to different intracellular compartments, to mitochondria for oxidation to betaine (mostly in the human liver, where hCTL1b was abundantly expressed; see Fig. 6B), and perhaps to the nucleus for choline utilization in the nuclear PC cycle (26). It is well established that the Pcyt1 α-isoform resides in (68) or could be translocated to the nucleus (36), where it is required for the synthesis of CDP-choline, presumably for nuclear membrane production (19, 25, 26).
Determining the physiological role of the CTL1 gene and each of its splice variants may answer significant questions about choline metabolism. We also put forth the possibility that regulation of the expression of this gene could be of great interest in diseases characterized by muscle dysfunction and that further assessment of choline transport for muscle membrane synthesis and perhaps choline recycling from acetylcholine at neuromuscular junctions may provide clues as to whether or not certain diseases of the muscle are related to abnormal hCTL1 transport function.
This study was supported by an operating grant from the Natural Sciences and Engineering Research Council of Canada (to M. Bakovic) and by an Ontario Government Premier Research Excellence Award (to M. Bakovic).
We are grateful to Adeel Safdar for preparing human skeletal muscle samples.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: M. Bakovic, Univ. of Guelph, Animal Science and Nutrition Bldg., Rm. 346, Guelph, Ontario, Canada N1G 2W1 (e-mail:)
- Copyright © 2006 the American Physiological Society