Mouse very long-chain Acyl-CoA synthetase 3/fatty acid transport protein 3 catalyzes fatty acid activation but not fatty acid transport in MA-10 cells.

The family of proteins that includes very long-chain acyl-CoA synthetases (ACSVL) consists of six members. These enzymes have also been designated fatty acid transport proteins. We cloned full-length mouse Acsvl3 cDNA and characterized its protein product ACSVL3/fatty acid transport protein 3. The predicted amino acid sequence contains two highly conserved motifs characteristic of acyl-CoA synthetases. Northern blot analysis revealed that the mouse Acsvl3 mRNA is highly expressed in adrenal gland, testis, and ovary, with lower expression in the brain of adult mice. A developmental Northern blot revealed that Acsvl3 mRNA levels were significantly higher in embryonic mouse brain (embryonic days 12-14) than in newborn or adult mice, suggesting a possible role in nervous system development. Immunohistochemistry revealed high ACSVL3 expression in adrenal cortical cells, spermatocytes and interstitial cells of the testis, theca cells of the ovary, cerebral cortical neurons, and cerebellar Purkinje cells. Endogenous ACSVL3 was found primarily in mitochondria of MA-10 and Neuro2a cells by both Western blot analysis of subcellular fractions and immunofluorescence analysis. In MA-10 cells, loss-of-function studies using RNA interference confirmed that endogenous ACSVL3 is an acyl-CoA synthetase capable of activating both long-chain (C16:0) and very long-chain (C24:0) fatty acids. However, despite decreased acyl-CoA synthetase activity, initial rates of fatty acid uptake were unaffected by knockdown of Acsvl3 expression in MA-10 cells. These studies cast doubt on the designation of ACSVL3 as a fatty acid transport protein.

The transport of fatty acids into cells and their subsequent "activation" by thioesterification to CoA are fundamental processes required for entry of fatty acids into the metabolic stream (1). The mechanism of fatty acids entry into cells remains controversial. Some investigators argue that specific proteins are required to transport the fatty acid across the plasma membrane (2)(3)(4)(5)(6). Others have provided evidence that proteins are not necessary for translocation of fatty acids through the lipid bilayer (7,8). One group of proteins proposed to mediate fatty acid entry into cells are the fatty acid transport proteins (FATPs) 1 (4). The mammalian FATP family consists of six homologous proteins (FATP1-6) that share 35-58% amino acid identity. 2 Studies with cultured cells overexpressing FATP1-6 have demonstrated increased rates of accretion of fluorescent or radiolabeled fatty acids (3,9). However, interpretation of fatty acid transport studies is hampered by the fact that, once inside cells, fatty acids are rapidly metabolized. Metabolism will decrease the intracellular concentration of the unesterified fatty acid, shifting the concentration gradient across the plasma membrane to promote entry of additional fatty acids into the cell. The design of most transport studies does not distinguish between transport mechanisms that can occur in a protein-free phospholipid bilayer and transport plus metabolism.
Independent of studies on FATP and its potential role in fatty acid transport, Hashimoto and co-workers (10) used classical protein purification methods to isolate the first enzyme capable of activating very long-chain fatty acids (VLCFA; fatty acids containing more than 22 carbons). This enzyme with very long-chain acyl-CoA synthetase (ACSVL) 3 activity was purified to homogeneity from rat liver peroxisomes, a known source of ACSVL activity. Sequencing of peptide fragments obtained from proteolytic digestion of this protein led to cloning of fulllength Acsvl1 cDNA (11). Translation of the cDNA sequence revealed that the protein with the highest amino acid identity to ACSVL1 was not one of the five long-chain acyl-CoA synthetases but rather the protein now known as FATP1. ACSVL1 contains two highly conserved amino acid sequence motifs (12,13). Motif 1 is an AMP-binding domain common to all known acyl-CoA synthetases and related enzymes whose reaction mechanism involves formation of an adenylated intermediate (14). The second motif overlaps the long-chain acyl-CoA synthetase "signature motif" described by Black and co-workers (15); this motif is thought to be critical for binding of the fatty acid substrate.
We and others identified five proteins that are highly homologous to ACSVL1 in the regions containing motifs 1 and 2 (9,12,16,17). The six members of the ACSVL family are identical to the six members of the FATP family (Table I). A fundamental unanswered question is whether these proteins are acyl-CoA synthetases, fatty acid transporters, or bifunctional proteins with both activities. Several laboratories have now demonstrated that at least four ACSVL/FATP proteins have acyl-CoA synthetase activity (18 -21). We now report that a fifth member of the murine protein family, ACSVL3 or FATP3, is also an acyl-CoA synthetase capable of activating both long-and very long-chain fatty acid substrates. We further report that endogenous ACSVL3 does not exhibit fatty acid transport activity. In addition, the expression pattern of this protein suggests that it may play an important role in brain development.

EXPERIMENTAL PROCEDURES
Materials and General Methods- [1-14 C]Palmitic acid (C16:0) and [1-14 C]lignoceric acid (C24:0) were obtained from Moravek, Inc. [9,10-3 H]Palmitic acid was from American Radiolabeled Chemicals. Unlabeled fatty acids were from Sigma. Protein was measured by the method of Lowry et al. (22). COS-7 cells and Neuro2a cells were maintained at 37°C in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with penicillin/streptomycin and 10% fetal bovine serum (Invitrogen) in a 5% CO 2 atmosphere. MA-10 cells (23) were kindly provided by Dr. Mario Ascoli (University of Iowa) and were maintained in Waymouth's medium containing 15% horse serum. DNA sequencing was performed either at MWG-Biotech (Ebersberg, Germany) or at the Johns Hopkins University Department of Biological Chemistry Biosynthesis and Sequencing Facility using the fluorescent dideoxy terminator method of cycle sequencing on an Applied Biosystems Inc. 377 automated DNA sequencer, following ABI protocols. General conditions for PCR were as reported (19,24). COS-7 cells were transfected by electroporation as described previously (24). Confocal microscopy was performed at the Microscope Facility, The Johns Hopkins University School of Medicine. Statistical significance was calculated using Student's t test. Subcellular fractionation of MA-10 cells was performed as previously described (25).
Cloning of the Mouse Acsvl3 cDNA-The expressed sequence tag (EST) division of GenBank TM (26) was queried using the murine cDNA sequences for two members of the Acsvl family, GenBank TM accession numbers AJ223958 and AJ223959, as probes. Several EST clones representing homologous but distinct cDNAs, including Acsvl3, were identified. The contig formed when Acsvl3 ESTs were aligned and joined was incomplete. To obtain additional 5Ј and 3Ј cDNA sequence, RACE protocols were carried out. For 3Ј-RACE, oligonucleotide primers 5Ј-G-AGCCAGTGCCGGGGTACCTCTCTGCC-3Ј and 5Ј-CATCTTCACCTC-TGGCACTACTGGCCTGC-3Ј were used as nested primers in combination with adaptor primers (AP-1 and AP-2; Clontech) and a murine brain RACE library as template. A combination of 5Ј-RACE and genome primer walking were used to obtain cDNA sequence containing the putative start codon and 5Ј-untranslated region. For genome walking, a murine genomic DNA library (Clontech) and nested gene-specific prim-ers 5Ј-GTCCTGCTGGTGGACTCCACACAGATG-3Ј and 5Ј-AGCACGA-GCGCACTCGCACCGCAGGCT-3Ј were used in combination with AP-1 and AP-2. For 5Ј-RACE, nested primers 5Ј-CCAGCGCAGCTGAGGCC-AGACGTCTA-3Ј and 5Ј-GAGGGCCGCCATGGTGCCCTTTTCCTC-3Ј, along with AP-1 and AP-2, were used with a murine embryonal RACE library (Clontech) as template. For expression studies, the Acsvl3 cDNA open reading frame was amplified from a murine liver cDNA library by PCR using forward primer 5Ј-CTAGAGGAAAAGGGCACCATGGCGG-C-3Ј and reverse primer 5Ј-GAGCCCCTCCCTCAAGTGGAAGGATT-3Ј and was T/A cloned into pCR2.1 (Invitrogen) and then subcloned into the HindIII and XhoI sites of pcDNA3.1ϩ (Invitrogen). The Acsvl3 cDNA sequence has been deposited in the EMBL nucleotide sequence data base (accession number AJ577573). A partial amino acid sequence of ACSVL3 was previously published (3,9). Genomic sequence information was partially derived by genomic primer walking as well as by in silico analysis of the mouse genome probed with the Acsvl3 cDNA sequence using either the BLAST algorithm, available on the World Wide Web at www.ncbi.nlm.nih.gov/genome/seq/MmBlast.html, or the BLAT algorithm, available on the World Wide Web at genome.ucsc.edu/ cgi-bin/hgBlat. The official gene symbol for Acsvl3 is Slc27a3 (solute carrier family 27, member 3).
Antibody Production and Affinity Purification-To produce polyclonal antiserum to ACSVL3, a glutathione S-transferase/ACSVL3 fusion protein was produced. An EST (GenBank TM accession number AA458592) encoding the carboxyl-terminal portion of human ACSVL3 was obtained from the I.M.A.G.E. consortium and digested with EcoRI and NotI. The resulting 683-bp fragment, which encodes the C-terminal 175 amino acids of the human ortholog of ACSVL3, was cloned into the EcoRI and NotI sites of the bacterial expression vector pGEX5X1 (Amersham Biosciences). The resulting construct was transfected into Escherichia coli strain BL21-DE3, and protein expression was induced for 5 h at 37°C following the addition of 1 mM isopropyl-␤-D-thiogalactopyranoside. Bacterial cells were lysed by sonication and the fusion protein solubilized with 1% Triton X-100 was purified using glutathione-Sepharose beads (Amersham Biosciences) according to the manufacturer's instructions. Immunization and bleeding of rabbits was done by a commercial firm (Cocalico Biologicals, Reamstown, PA). For affinity purification of the antibody, the fusion protein was subjected to preparative SDS-PAGE on a 10% gel. The fusion protein in sample buffer was loaded into a single large well and electrophoresed for 5 min, after which a second aliquot of protein was loaded. This process was repeated three more times before allowing electrophoresis to proceed to completion. After transfer to nitrocellulose, proteins were identified by staining with Ponceau-S, and a membrane strip containing the five protein bands was excised and stored at 4°C in phosphate-buffered saline (PBS). Crude antiserum was diluted 5-fold with PBS and incubated with the membrane overnight at 4°C. After washing with PBS, bound antibody was eluted with 0.1 M glycine, pH 2.5. The eluate was immediately neutralized by the addition of 1 M Tris-HCl, pH 8.0, and the buffer was exchanged back to PBS using a Centricon 30 (Millipore Corp.). Affinity-purified antibody was stored at Ϫ80°C in the presence of 1% bovine serum albumin (Sigma). The specificity of the purified antibody for ACSVL3 was determined by Western blot analysis of COS-1 cells overexpressing each of the six ACSVL/FATP family members and two acyl-CoA synthetases belonging to two other families, FIG. 1. Specificity of the ACSVL3 antibody. Full-length cDNA for human (h) ACSVL3, mouse (m) Acsvl3, ACSVL1, ACSVL2, ACSB, FATP1, FATP4, BG1, and ACSL6 were transfected into COS-1 cells by electroporation. Three days post-transfection, cells were harvested for analysis by Western blot and indirect immunofluorescence. The ACSVL2, FATP1, FATP4, and ACSL6 constructs contained the c-Myc epitope at their amino termini. Expression of each protein was verified by immunofluorescence using either specific antibody or anti-Myc antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The number of COS-1 cells expressing each protein was generally greater than 50% and ranged from 30 to 80%. For Western blot analysis, ϳ30 g of cellular protein (ϳ10 g for hACSVL3) was loaded onto a 10% gel. After transfer to nitrocellulose membrane, reactivity of the ACSVL3 antibody was assessed. Anti-actin antibody (Sigma) was used as a loading control.  3 The enzyme previously described as VLCS-H2/VLACSR/BACS has been shown to preferentially activate bile acids to their CoA derivatives (28,45) and is thus designated ACSB. The FATP nomenclature follows that of Stahl et al. (46). ACS Very Long-chain Acyl-CoA Synthetase 3 ACSL6 and BG1. Only mouse or human ACSVL3 was detected by the purified antibody (Fig. 1).
Northern Analysis-For tissue expression of Acsvl3, total RNA was prepared using the Trizol reagent (Invitrogen) from freshly harvested tissues from adult (ϳ3-month-old) mice. Twenty micrograms of RNA was electrophoresed on a 1% agarose gel at 4 V/cm for 2.5 h and then transferred overnight to a Hybond-Nϩ membrane (Amersham Biosciences). A cDNA probe for detection of Acsvl3 mRNA was prepared by excision of a 1196-bp fragment corresponding to bp 405-1601 using Eco47III and BspeI. For control, a 528-bp glyceraldehyde-3-phosphate dehydrogenase probe was prepared by PCR amplification using 5Ј-AC-CACCATGGAGAAGGCTGG-3Ј and 5Ј-CTCAGTGTAGCCCAGGATG-C-3Ј as forward and reverse primers, respectively, and mouse brain cDNA as template. Conditions for probe labeling, hybridization, and detection were as previously described (20). For developmental analysis of Acsvl3 expression, a Northern blot was prepared from 1-2 g of poly(A) ϩ RNA as previously described (24). For this blot, the probe consisted of 1315 nucleotides corresponding to bp 807-2121 of Acsvl3 cDNA. As a control for loading and transfer, the blot was probed with a mouse cDNA of the ubiquitously expressed protein cyclophilin.
RT-PCR Analysis-Approximately 2 g of total cellular RNA (extracted as described above) was reverse transcribed and PCR-amplified using the PerkinElmer Life Sciences RT-PCR system with forward primer 5Ј-AATGCCCAGGGGCACTGCATGACCACAT-3Ј (within exon 6; see Table II) and reverse primer 5Ј-ATTCGAAGGTCTCCAGACAG-GAGGGCA-3Ј (within exon 10). The primer combination was selected such that a 634-bp Acsvl3-specific product can be amplified from cDNA but not from genomic DNA, potentially contaminating the RNA extracts. RT-PCR was performed under conditions not allowing quantitative analysis. As a positive control, RT-PCR of a 680-bp fragment of the ubiquitously expressed glyceraldehyde-3-phosphate dehydrogenase mRNA using forward primer 5Ј-ACTGGCGTCTTCACCACCAT-3Ј and reverse primer 5Ј-TCCACCACCCTGTTGCTGTA-3Ј was performed on the same cDNA preparations.
Animals and Their Care-Wild type 129SvEv mice were obtained from Taconic, Inc. (Germantown, NY) and housed in facilities of The Johns Hopkins University School of Medicine. Animals were housed under controlled conditions, between 22 and 27°C, on a 12-h light/dark cycle, with food and water ad libitum. Procedures involving animals and their care were conducted in conformity with institutional guidelines that are in compliance with national and international laws and policies (47).
Antibodies, Indirect Immunofluorescence, Direct Immunofluorescence, and Immunohistochemistry-Polyclonal antiserum to the peroxisomal protein PMP70 was a gift from Dr. S. Gould. Monoclonal antibody to the endoplasmic reticulum marker, protein-disulfide isomerase, and polyclonal antibody to the mitochondrial marker, Mn-superoxide dismutase, were from Stressgen, and polyclonal antibody to a plasma membrane marker, E-cadherin, was from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Rabbit antibody to BG1 was affinity-purified as previously described (25). For indirect immunofluorescence analysis, cells were fixed in 4% formaldehyde in PBS and permeabilized with 1.0% Triton X-100 prior to incubation with primary and secondary antibodies as described previously (27). In some experiments, cells were incubated following the manufacturer's instructions with MitoTracker red (Molecular Probes, Inc., Eugene, OR) before fixation. Because affinity-purified polyclonal antibodies to ACSVL3 and BG1 were both raised in rabbits, double-labeling studies done by sequential application of antibodies resulted in an unacceptable level of nonspecific binding of fluorescence-labeled secondary antibodies. Therefore, these antibodies were also directly labeled using Zenon TM Rabbit IgG Labeling reagents (Molecular Probes). Alexa Fluor 555 and Alexa Fluor 488 reagents were used to fluorescently label ACSVL3 and BG1 antibodies, respectively, following the manufacturer's protocol. For direct immunofluorescence, cells were incubated concurrently with both antibodies, washed extensively with PBS, fixed again with 4% formaldehyde, washed, and mounted. Control experiments containing labeling reagents prepared without either ACSVL3 or BG1 antibody resulted in no nonspecific labeling when used in conjunction with the opposite directly labeled antibody. For immunohistochemistry, tissues from 3-month-old mice were harvested, quickly frozen in liquid nitrogen, and stored at Ϫ80°C. Tissue sections were cut using a cryostat and fixed with 4% paraformaldehyde as described previously (28). Brain sections were 20 m thick; sections of adrenal gland, testis, and ovary were 5-8 m thick. After fixation, sections were incubated for 30 min with 0.6% H 2 O 2 in methanol followed by 20 min with 5% normal goat serum. Sections were then incubated sequentially with Avidin D and biotin, 15 min each (Avidin/Biotin Blocking Kit; Vector Laboratories). Incubation with primary rabbit antibody (affinity-purified anti-hACSVL3 antibody; 1:100) was for 1 h at 37°C or overnight at 4°C. Peroxidasebased detection was done using a Vectastain ABC kit (Vector Laboratories). After counterstaining with hematoxylin-Harris stain for 30 s, sections were dehydrated and mounted with DPX mounting solution (Fluka Biochemika).
RNA Interference-Four pairs of cDNA oligonucleotides containing complementary sense and antisense Acsvl3 sequences were obtained from Integrated DNA Technologies and used to synthesize small interfering RNA 21-mers (siRNA) using the Silencer kit (Ambion) as previously described (25). Targeted sequences of siRNA constructs 1-4 corresponded to bp 55-75, 695-715, 1243-1263, and 1855-1875, respectively, of the Acsvl3 coding region. MA-10 cells were transfected with the individual siRNA constructs or a mixture of all four, essentially as described (25), except that cells were incubated with siRNA for only 12 h prior to the addition of the usual culture medium.
Acyl-CoA Synthetase and Fatty Acid Uptake Assays-Acyl-CoA synthetase assays utilizing radiolabeled palmitic acid (C16:0) and lignoceric acid (C24:0) were performed as previously described (29,30). Fatty acid uptake assays were similar to those described by Trotter et al. (31). Labeling solution was prepared by adding unlabeled plus [9,10-3 H]palmitic acid in ethanol to a solution of 100 M fatty acid-poor bovine serum albumin in PBS such that the final fatty acid concentration was 100 M (5 Ci/ml) and the final ethanol concentration was 0.5%. MA-10 cells in 60-mm culture dishes were grown to confluence in Waymouth's medium containing 15% horse serum. Two hours before the assay, the medium was replaced with warm medium without serum. All subsequent procedures were carried out in a 37°C room. After discarding the medium, cells were washed three times with warm PBS prior to the addition of 1 ml of warm labeling solution. Cells were incubated for the indicated time, at which point the labeling medium was aspirated and simultaneously replaced with ice-cold PBS. Dishes were placed on ice; cells were washed three times with ice-cold PBS and scraped into 0.5 ml of 0.25 M sucrose, 10 mM Tris (Cl Ϫ ), pH 8.0, 1 mM EDTA. Bovine serum albumin, often used by other investigators to remove fatty acid nonspecifically bound to the cell surface, was intentionally absent from the cold PBS used to quench uptake and wash cells following uptake, since it has been demonstrated that fatty acid-poor albumin can extract fatty acids already transported into cells by promoting reversal of the flip-flop of unionized fatty acids across the plasma membrane (32). After removal of an aliquot of cell suspension Very Long-chain Acyl-CoA Synthetase 3 for protein determination, radioactivity was determined by liquid scintillation counting.

RESULTS
Cloning of Full-length Mouse Acsvl3 cDNA and Initial Characterization-Early evidence for the existence of ACSVL3 came from homology probing of the expressed sequence tag data base using the amino acid sequence of ACSVL1 as a probe. The highly conserved region now referred to as Motif 2 (12, 13) was found in numerous ESTs that could be placed into five groups of candidate ACSVL1 homologs, including ACSVL3. Fulllength mouse Acsvl3 cDNA was cloned into the mammalian expression vector pcDNA3 as described under "Experimental Procedures." The nucleotide sequence shown in Fig. 2 contains 137 bp of 5Ј-untranslated region, a Kozak consensus sequence flanking the likely start codon (underlined), an open reading frame encoding 667 amino acids, and 117 bp of 3Ј-untranslated region containing a polyadenylation signal (boldface type). The Acsvl3 gene is found on mouse chromosome 3. The predicted amino acid sequence encodes a protein with a calculated molecular mass of 72,968 daltons and a pI of 7.59. Motif 1 (underlined italic type) and Motif 2 (underlined and boldface type) are also indicated in Fig. 2.
The predicted ACSVL3 amino acid sequence was found to share 35-47% identity and 54 -78% overall similarity with the five other members of the murine ACSVL/FATP protein family. Examination of the amino acid sequence using both neural networks and hidden Markov models trained on eukaryotes revealed the presence of a signal peptide sequence with the most likely cleavage site between residues 19 and 20 (33). A Kyte-Doolittle hydropathy plot obtained using the TopPred II program (34) (available on the World Wide Web at bioweb. pasteur.fr/seqanal/interfaces/toppred.html) suggested that ACSVL3 contains one or two potential membrane-spanning regions. The genomic organization of Acsvl3 (Table II) was determined by genomic primer walking, in combination with in silico analysis. The intron/exon structure of the mouse gene is similar to that of its human ortholog. 4 Tissue Expression of Acsvl3-Before examining the biological function of ACSVL3, it was first necessary to determine the tissues and the cell types within these tissues in which the protein was expressed. In addition, it was necessary to assess the subcellular location of the endogenous protein, since a true cellular "fatty acid transport protein" should be found in the plasma membrane. The tissue expression pattern of Acsvl3 was assessed by RT-PCR and Northern blot analysis. RT-PCR studies suggested that the mRNA was well expressed in lung, kidney, testis, ovary, brain, brain stem, and adrenal gland, less abundant in liver, and barely detectable in heart, skeletal muscle, and spleen (Fig. 3A). To expand upon the results of the nonquantitative RT-PCR, a Northern blot containing total RNA from adult mouse tissues was prepared. Northern blot analysis revealed that a ϳ2.6-kb ACSVL3 transcript was highly expressed in adrenal gland, ovary, and testis, with a lower expression level in brain, spinal cord, and lung (Fig. 3B).
Because of the quantitative importance of VLCFA in brain lipids, we also examined the developmental expression pattern of Acsvl3 in mouse brain by Northern blot analysis. Acsvl3 mRNA levels were significantly higher in embryonic mouse brain (embryonic days 12-14) than in newborn or adult mice (Fig. 3C). A continuous decrease in expression was seen between embryonic day 14 and postnatal day 15. This observation suggests that ACSVL3 might play an important role in normal mouse brain development.
Cell Type Specificity of Acsvl3 Expression-Because the tissues expressing Acsvl3 contained diverse populations of cells, immunohistochemical analysis of mouse tissues was used to determine which specific cell types expressed the protein. In the adult mouse brain, ACSVL3 protein was primarily observed in neurons (Fig. 4). Cortical neurons, particularly hippocampal neurons, showed high levels of expression. Most, but not all, cerebellar Purkinje cells showed moderately high levels of expression (Fig. 4). Adrenal gland ACSVL3 expression was confined to the cortex and was particularly high in the outer zona glomerulosa and inner zona reticularis but was detectable in the middle zona fasciculata as well (Fig. 4); no ACSVL3 was detected in the adrenal medulla. In the testis, ACSVL3 was present at high levels in primary spermatocytes and was also found in secondary spermatocytes and interstitial Leydig cells (Fig. 4). Ovarian ACSVL3 expression was mainly found in internal and external theca cells, but some follicle cells also contained the protein (Fig. 4).
Subcellular Localization of ACSVL3-We next investigated the subcellular localization of ACSVL3 in cell lines predicted to express the protein by the histochemical analyses. The mouse testis Leydig cell line MA-10 (23) and the mouse neuroblastoma cell line Neuro2a were found to express Acsvl3 both by RT-PCR and Western blot (data not shown). Subcellular fractionation by differential centrifugation revealed that ACSVL3, which migrates as an 84-kDa band on Western blots, was found primarily in the M fraction, which is enriched in mitochondria, of both MA-10 cells (Fig. 5A) and Neuro2a cells (data not shown). An additional band of ϳ58 kDa was occasionally seen in the M fraction and, to a lesser extent, in the peroxisome-enriched L fraction; based on RNA interference studies (below), this band appears to be nonspecific and unrelated to ACSVL3. No 84-kDa ACSVL3 band was seen in the N (nuclear), L (light mitochondrial/peroxisomal), P (microsomal), or S (cytosolic) fractions.
Indirect immunofluorescence studies were also done to examine the subcellular compartment(s) containing ACSVL3. Both COS-7 cells overexpressing the protein and MA-10 and Neuro2a cells that endogenously express ACSVL3 were examined. A fine, reticular immunostaining pattern was observed in COS-7 cells transfected with full-length Acsvl3 cDNA (Fig. 6A) that exhibited significant colocalization with the endoplasmic reticulum marker, protein-disulfide isomerase (Fig. 6, B and   FIG. 3. Tissue expression of Acsvl3. A, RT-PCR analysis of murine tissues. Mouse tissue RNA was obtained, reverse transcribed, and used as template for amplification of a fragment of Acsvl3 as described under "Experimental Procedures." Amplification of a glyceraldehyde-3-phosphate dehydrogenase fragment as a control is shown below. B, Northern blot analysis of Acsvl3 mRNA expression in adult murine tissues. Total RNA (20 g) was hybridized with a 32 P-labeled Acsvl3 cDNA probe. A control hybridization with a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe is also shown. C, Northern blot analysis of Acsvl3 mRNA during mouse brain development. Poly(A) ϩ RNA (1-2 g/lane) was extracted from dissected brain or whole head (lanes E12 and E14) at the indicated developmental stages and hybridized with 32 P-labeled cDNA probes. A cyclophilin (cyph) probe was used as loading control for this blot. The lanes during brain development are labeled as follows: E12-E18, 12.5-18.5 days of gestation; P2-P30, days of postnatal development.
FIG. 4. Immunohistochemical localization of ACSVL3 in mouse tissues. Tissues were processed for immunohistochemical localization using affinity-purified anti-ACSVL3 antibody as described under "Experimental Procedures." Cells containing ACSVL3 protein have a brownish appearance against the blue hematoxylin counterstain. In the brain sections (cerebral cortex, hippocampus, and cerebellum), neuronal expression is evident. In the adrenal cortex, cells of the outer zona glomerulosa (zg) and the inner zona reticularis (zr) showed greater expression than did the middle zona fasciculata (zf); the adrenal medulla (med) had no detectable ACSVL3. In testis, primary spermatocytes showed high expression, with lower levels observed in secondary spermatocytes and interstitial Leydig cells. Both internal and external theca cells of the ovary showed prominent ACSVL3 expression; some labeling of follicle cells was also evident. C). In contrast, the endogenous ACSVL3 protein in MA-10 cells (Fig. 6, D, G, J, and N) and Neuro2a cells (not shown) exhibited a more punctate fluorescence pattern when observed by confocal microscopy. ACSVL3 immunostaining of these punctate structures did not colocalize with protein-disulfide isomerase staining in MA-10 cells (Fig. 6, K and L); nor did it colocalize with peroxisomes (Fig. 6, H and I). ACSVL3 immunofluorescence partially colocalized with mitochondria visualized using the vital stain, MitoTracker red (Fig. 6, E and F). This finding is in agreement with the results of subcellular fractionation (Fig. 5A). Because colocalization of ACSVL3 and mitochondria was not complete, we speculated that ACSVL3 might be found in the mitochondria-associated membrane fraction, an endoplasmic reticulum-derived compartment that sediments with mitochondria during differential centrifugation (35). Therefore, we prepared a crude mitochondrial/peroxisomal (ML) pellet and subfractionated it into a mitochondria-associated membrane (MAM) fraction and purified mitochondrial fraction. Western blot analysis revealed that ACSVL3 was mainly in the purified mitochondrial fraction (Fig. 5B).
Like ACSVL3, the acyl-CoA synthetase BG1 (also called lipidosin) was previously detected in purified mitochondrial fractions by Western blot and in punctate vesicles that partially colocalized with mitochondria by immunofluorescence analysis (25). To determine whether both ACSVL3 and BG1 were located in the same vesicular compartment, we directly labeled affinity-purified antibodies and performed double-labeling experiments. As shown in Fig. 6, M-O, it appears that ACSVL3 and BG1 are found in different vesicle populations. The precise nature of ACSVL3-containing vesicles remains under investigation.
Knockdown of ACSVL3 Expression in MA-10 Cells by RNA Interference-To study the enzymatic activity and fatty acid transport activity of endogenous ACSVL3, RNA interference using siRNA was used to decrease the levels of this protein in MA-10 cells. Four double-stranded siRNA molecules were tested for their ability to decrease ACSVL3 protein levels in MA-10 cells. By Western blot analysis, all four constructs decreased the quantity of the 84-kDa ACSVL3 protein (Fig. 7A,   FIG. 5. ACSVL3 localization in MA-10 cell subcellular fractions. A, Western blot of fractions prepared by differential centrifugation. Mouse MA-10 (Leydig) cells grown to 80% confluence were harvested and fractionated by differential centrifugation as described previously (25). ACSVL3 protein was detected only in the M fraction, which is enriched in mitochondria as indicated by the presence of the marker enzyme Mn-superoxide dismutase (MnSOD). No ACSVL3 was found in the nuclear (N), light mitochondrial (L; enriched in peroxisomes), particulate (P; enriched in microsomes), or soluble (S; or cytosolic) fractions. The plasma membrane marker, E-cadherin, was detected in the N, L, and P fractions, but not in the M fraction. Plasma membrane sheets typically sediment in the N fraction; whereas plasma membrane fragments are typically found in the P fraction, they can also be present in other fractions (44). B, Western blot of MAM and purified mitochondria. A combined mitochondria/peroxisome-enriched (ML) fraction of MA-10 cells was subfractionated into purified mitochondria (Mito) and MAM fraction as described previously (25). MnSOD is found in purified mitochondria but not in MAM. Because the MAM marker phosphatidylethanolamine N-methyltransferase 2 (PEMT) cannot be detected in MA-10 cells, an ML fraction of mouse liver was subjected to the identical MAM isolation protocol; phosphatidylethanolamine N-methyltransferase 2 was found in MAM but not purified mitochondria. lanes 1-4) as compared with control siRNA (Fig. 7A, lane con). A combination of all four siRNAs consistently decreased ACSVL3 protein to nearly undetectable levels (Fig. 7A, mix), and this combination was used in subsequent experiments. Knockdown of ACSVL3 using siRNA was confirmed by immunofluorescence studies (Fig. 7, B-D).
Acyl-CoA Synthetase Activity of Endogenous ACSVL3 in MA-10 Cells-To determine whether ACSVL3 catalyzed fatty acid activation, loss-of-function studies were conducted. Total acyl-CoA synthetase activity was measured in MA-10 cells in which ACSVL3 protein levels were decreased by siRNA treatment. Compared with control siRNA-treated cells, MA-10 cells lacking ACSVL3 had reduced ability to activate the long-chain fatty acid, palmitate (C16:0), and the very long-chain fatty acid, lignocerate (C24:0), to their respective CoA derivatives (Fig. 8).
Reductions of 30% in the rate of C16:0 activation and 26% in the rate of C24:0 activation were observed that were statistically significant (p Ͻ 0.01 and p Ͻ 0.02, respectively).
Fatty Acid Transport Activity of Endogenous ACSVL3 in MA-10 Cells-Having established that ACSVL3 has acyl-CoA synthetase activity, the ability of this protein to affect fatty acid transport was assessed. MA-10 cells in which ACSVL3 protein levels were decreased to near zero (Fig. 7) and in which activation of both C16:0 and C24:0 were reduced (Fig. 8) by siRNA treatment were incubated with radiolabeled C16:0, and the rate of cellular uptake was determined over a 30-s period. There was no difference in initial rates of C16:0 uptake between untreated MA-10 cells, cells treated with ACSVL3-specific siRNA, or cells treated with nonspecific siRNA (Fig. 9). This absence of apparent fatty acid transport activity in MA-10 cells is consistent with the observation that ACSVL3 was not found in plasma membrane either by immunofluorescence (Fig.  6) or by Western blot (Fig. 5). DISCUSSION Acyl-CoA synthetases play a central role in fatty acid metabolism by converting a relatively inert fatty acid molecule into an activated form that can be utilized by numerous metabolic pathways (1). Fatty acyl-CoAs can be degraded for energy production or incorporated into triacylglycerol for storage. They are substrates for the synthesis of phospholipids, sphingolipids, and glycolipids and can acylate proteins or regulate gene expression. Based on this metabolic diversity, along with the wide range of fatty acid chain lengths in nature, it is not surprising that mammalian genomes encode about 25 distinct acyl-CoA synthetases. 5 The acyl-CoA synthetase reaction mechanism involves formation of an acyl-adenylate intermediate and the release of pyrophosphate, with subsequent displacement of AMP by CoA-SH. A similar reaction mechanism is employed by other enzymes, such as firefly luciferase and bacterial gramicidin-S synthetase, and all of these proteins contain a characteristic AMP-binding domain (14). This domain is present in the predicted amino acid sequence of ACSVL3 when analyzed by the Protein Families (PFAM) data base (available on the World Wide Web at www.sanger.ac.uk/Software/Pfam/). Within the AMP-binding domain is a highly conserved sequence of 10 amino acids referred to as Motif 1 (Fig. 2) that is nearly identical in all members of the ACSVL family, including ACSVL3: (F/Y)TSGTTGLPK. Black et al. (15) described a "signature motif" that was conserved among members of the long-chain acyl-CoA synthetase family and found evidence that this region played a role in acyl chain-length substrate specificity of the enzymes. We described a highly conserved region in the ACSVL family that overlapped with the long-chain enzyme signature 5 P. A. Watkins, manuscript in preparation.  1-4) siRNA constructs. Cells were also treated with a mixture (mix) of all four siRNAs. After 3 days, cells were harvested and examined for ACSVL3 expression by Western blot. A decrease in intensity of the 84-kDa ACSVL3 band was observed, but no change in the level of the ϳ58-kDa band was noted, indicating that it is not related to ACSVL3. The control siRNA was specific for the unrelated protein, mouse serum albumin. B-D, immunofluorescence of siRNA-treated MA-10 cells. Cells treated with either control siRNA (B) or a mixture of the four ACSVL3-specific siRNA constructs (C and D) were examined 3 days post-transfection using the anti-ACSVL3 antibody. motif, which we referred to as Motif 2 (12,13). Within Motif 2 is a sequence of 11 amino acids that are identical in all six members of the ACSVL family: GDTFRWKGENV. It should be noted that this family is designated "very long-chain" for two reasons: first, because at least five of the six members are capable of activating VLCFA, and second, to distinguish it from the long-chain (ACSL), medium-chain (ACSM), and other acyl-CoA synthetase families. Whereas both ACSL and ACSVL family enzymes can activate long-chain fatty acid substrates, only the latter are capable of VLCFA activation.
Several lines of evidence support the conclusion that ACSVL3 is indeed an acyl-CoA synthetase. First, the ACSVL3 amino acid sequence contains the two motifs thought to be critical for enzyme activity and also shares 35-42% overall amino acid identity with ACSVL1, ACSB, FATP1, and FATP4, proteins previously shown to have acyl-CoA synthetase activity (10, 18 -21). Second, RNA interference studies indicated that disruption of ACSVL3 led to decreased cellular activation of both C16:0 and C24:0. Third, overexpression of the human ortholog of ACSVL3 in COS-7 cells led to an overall increase in cellular activation of both C16:0 and C24:0. 4 Similarly, results presented here suggest that ACSVL3 does not function as a plasma membrane transporter of long-chain fatty acids, at least in MA-10 cells and Neuro2a cells. On Western blots, there was no detectable ACSVL3 protein in subcellular fractions containing plasma membrane. Confocal microscopy of cells immunolabeled using anti-ACSVL3 antibody failed to show any plasma membrane fluorescence. Under conditions where ACSVL3 protein levels and acyl-CoA synthetase activity were decreased using RNA interference, there was no decrease in initial rates of long-chain fatty acid uptake by cells. Thus, in the cell lines examined, we found evidence for enzyme activity but no evidence for a fatty acid transport function of ACSVL3. In further support of this conclusion, DiRusso et al. 6 found that ACSVL3 only weakly promoted cellular uptake of a fluorescent fatty acid when expressed in a yeast mutant lacking endogenous proteins necessary for fatty acid uptake; in these experiments, ACSVL3 was clearly present in plasma membrane fractions.
Interest in acyl-CoA synthetases that are capable of activating VLCFA derives in part from the human peroxisomal disorders. In the disorders of peroxisome biogenesis (e.g. Zellweger syndrome), X-linked adrenoleukodystrophy (X-ALD), acyl-CoA oxidase deficiency, and D-bifunctional protein deficiency, tissue VLCFA levels are elevated (36 -38). In the mouse, both ACSVL3 and BG1 ("bubblegum"; lipidosin (25,39,40)), an acyl-CoA synthetase belonging to a different enzyme family (13), are expressed in cells of the central nervous system, adrenal cortex, and testis (Figs. 3 and 4) (24,25,40). These are the primary tissues that exhibit pathology in X-ALD (37). The defective X-chromosomal gene (ABCD1) in this disease encodes a peroxisomal membrane transporter protein with no known role in VLCFA metabolism (37,41). We have hypothesized that a nonperoxisomal enzyme with ACSVL activity contributes to the biochemical pathology in X-ALD (42). The tissue expression pattern of both ACSVL3 and BG1, along with their ability to activate VLCFA substrates, makes these enzymes candidates for further investigation in this disease. However, no compelling evidence for a role for BG1 in X-ALD has been reported (24,42). Future studies will examine the potential role of ACSVL3 in this disease.
Observations reported here indicate that ACSVL3 may play an important role in biological processes. The expression pattern of this protein in the brain suggests that it is important in early stages of nervous system development. Very high mRNA levels were detected at embryonic days 12-14, but the adult levels were barely detectable. Despite the significant decrease in ACSVL3 mRNA levels as mice age, histochemical studies clearly showed ACSVL3 protein in cerebral cortical neurons and hippocampal neurons. Expression in cerebellar Purkinje cells was also observed, but cell-to-cell variation was evident. In contrast, BG1 expression was weaker in cortical neurons but higher and more consistent in Purkinje cells of adult mice (25). The expression of mBG1 during brain development, as assessed by Northern blot analysis (24), was precisely the opposite of that for ACSVL3. This observation suggests that perhaps the two enzymes perform similar functions. A switch from expression of ACSVL3 to BG1 could provide a critical modification of substrate specificity necessary for normal brain development. Histochemical evaluation of the brain at late embryonic and early postnatal stages may yield further insight into the possible roles of ACSVL3 and BG1 in brain development.
In addition to expression in neurons of the cerebral cortex and cerebellum, both ACSVL3 and BG1 are also found in the same steroidogenic tissues: adrenal gland, ovary, and testis (25). Whereas there is some overlap in the specific types of cells expressing both proteins (e.g. ACSVL3 and BG1 are both expressed in the testicular Leydig cell line MA-10), there are some differences. In testis, ACSVL3 expression was higher in spermatocytes than in Leydig cells, whereas BG1 expression was found almost exclusively in Leydig cells. In ovary, both BG1 and ACSVL3 were found in theca cells, but ACSVL3 expression was also observed in follicle cells. Whereas BG1 was found primarily in the zona fasciculata of the adrenal cortex, ACSVL3 was most prominent in the zona glomerulosa and the zona reticularis. Whereas adrenal ACSVL3 could participate in the synthesis of stored cholesterol esters and/or triacylglycerol, the relationship between fatty acid activation and synthesis and secretion of steroid hormones is unclear. No obvious step in the conversion of cholesterol to steroid hormone is known to require activation of an acyl moiety. However, the presence of two distinct acyl-CoA synthetases, ACSVL3 and BG1, in steroidogenic tissues suggests that they perform a vital metabolic function.