The Human Liver-specific Homolog of Very Long-chain Acyl-CoA Synthetase Is Cholate:CoA Ligase*

Unconjugated bile acids must be activated to their CoA thioesters before conjugation to taurine or glycine can occur. A human homolog of very long-chain acyl-CoA synthetase, hVLCS-H2, has two requisite properties of a bile acid:CoA ligase, liver specificity and an endoplasmic reticulum subcellular localization. We investigated the ability of this enzyme to activate the primary bile acid, cholic acid, to its CoA derivative. When expressed in COS-1 cells, hVLCS-H2 exhibited cholate:CoA ligase (choloyl-CoA synthetase) activity with both non-isotopic and radioactive assays. Other long- and very long-chain acyl-CoA synthetases were incapable of activating cholate. Endogenous choloyl-CoA synthetase activity was also detected in liver-derived HepG2 cells but not in kidney-derived COS-1 cells. Our results are consistent with a role for hVLCS-H2 in the re-activation and re-conjugation of bile acids entering liver from the enterohepatic circulation rather than in de novo bile acid synthesis.

Synthesis of bile acids from cholesterol and their conversion to bile salts (conjugated bile acids) takes place in the liver (reviewed in Refs. 1 and 2). The primary bile acids synthesized in humans are cholic acid and chenodeoxycholic acid. These compounds are conjugated to the amino acids glycine or taurine, which are then secreted into the bile canaliculi. For conjugation, bile acids must first be activated to their coenzyme A (CoA) thioesters. This activation is catalyzed by an acyl-CoA synthetase (bile acid:CoA ligase) that is distinct from the enzyme(s) that activate long-chain fatty acids (3,4); however, the bile acid ligase reaction mechanism is analogous to that of other acyl-CoA synthetases (5). Whereas long-chain fatty acid activation occurs in essentially all tissues, bile acid activation takes place only in liver (3). Subcellular fractionation studies revealed that activation occurred in the endoplasmic reticulum of liver (3, 6 -8).
Our laboratory recently identified a family of human proteins homologous to very long-chain acyl-CoA synthetase (VLCS) 1 (9 -11), an enzyme originally purified from rat liver peroxisomes (12). Unlike long-chain acyl-CoA synthetase (LCS), VLCSs from both rat (rVLCS) and human (hVLCS) are capable of activating fatty acids containing more than 22 carbons (11,12). One protein, which we previously designated human VLCS homolog 2 (hVLCS-H2), possessed the two criteria necessary for a bile acyl-CoA synthetase, a liver-specific tissue distribution and an endoplasmic reticulum subcellular localization (10). Therefore, we investigated the ability of this enzyme to activate cholic acid.  (13). Choloyl-CoA was synthesized by the mixed anhydride method as described by Schulz (14). Protein was determined by the method of Lowry et al. (15).

Materials and General
Acyl-CoA Synthetase cDNAs, Transfection of COS-1 Cells, and Assay of Fatty Acyl-CoA Synthetase Activity-Full-length cDNA encoding human VLCS (hVLCS), hVLCS-H2, and a rat LCS (rACS1) were obtained and cloned into the mammalian expression vector pcDNA3 (Invitrogen) as described previously (10,11,16). Details of hVLCS homolog 1 (hV-LCS-H1) will be published elsewhere. 2 COS-1 cells were transiently transfected with full-length cDNA encoding each acyl-CoA synthetase or with the pcDNA3 vector alone by electroporation at 250 V as described previously (11). Three days post-transfection, cells were harvested by trypsinization, washed with phosphate-buffered saline, and stored at Ϫ80°C prior to assay. Assay of activation of [1-14 C]fatty acids was performed as described previously (17).
Choloyl-CoA Synthetase Assays-Two assays of choloyl-CoA synthetase were used. Non-isotopic assays contained 40 mM Tris, pH 7.5, 10 mM ATP, 5 mM MgCl 2 , 0.2 mM CoA, 20 M cholic acid (sodium salt), and COS-1 cells transiently expressing acyl-CoA synthetases (40 g protein) in a total volume of 0.25 ml. After 20 min at 37°C, duplicate 100-l aliquots were removed and the reaction was stopped by addition of 10 l of 2 N HCl and 12.5 l of 0.6 M MES in 2 N KOH. Samples were stored at Ϫ20°C. Just prior to analysis, samples were thawed and 47 l of acetonitrile was added. HPLC analysis was performed using a modification of the method of Kurosawa et al. (18). Samples were injected onto a Phenomenex 15-cm, 3-m C-18 reverse-phase HPLC column. Bile acyl-CoAs were eluted using a gradient of acetonitrile in 16.9 mM sodium phosphate, pH 6.9 (25-37% over 20 min). Absorbance at 254 nm was monitored. For the radioactive assay, 20 M [1-14 C]cholic acid (20,000 dpm/nmol) replaced the unlabeled compound. Labeled choloyl-CoA was separated from the substrate by a modification of the method of Kelley and Vessey (5). The reaction was terminated by the addition of 0.05 ml of 0.1 M EDTA, pH 7.5, and 0.2 ml of 60 mM succinic acid. Unreacted cholate was extracted with two 1.0-ml portions of watersaturated butanol. Radioactivity in the aqueous phase containing choloyl-CoA was determined. * This work was supported by National Institutes of Health Grants NS10533, NS 37355, and HD10981. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

RESULTS AND DISCUSSION
hVLCS-H2 Fulfills the Criteria Necessary for a Bile Acyl-CoA Synthetase-It was hypothesized more than 20 years ago that the enzyme that activates bile acids to their CoA derivatives is distinct from LCS, the enzyme that activates the most abundant endogenous and dietary fatty acids (3). This hypothesis was based primarily on the observation that whereas microsomes from nearly all tissues activated 16-and 18-carbon fatty acids, only those from liver were capable of activating cholic acid (3). Furthermore, activation of fatty acids takes place in mitochondria and peroxisomes, as well as in microsomes (19), whereas cholate activation is almost exclusively microsomal (3, 6 -8). Thus, the two criteria that a bile acyl-CoA synthetase must fulfill are liver specificity and a microsomal subcellular location.
Our investigations of a new family of proteins that includes VLCS and fatty acid transport protein (FATP) revealed that a member of this family, hVLCS-H2, possessed these two attributes (Table I) (10). We recently reported cloning of fulllength hVLCS-H2 cDNA and initial characterization of its gene product (10). When an hVLCS-H2-specific probe derived from this clone was hybridized to a human multiple tissue Northern blot, a transcript was detected only in liver (10). Furthermore, hVLCS-H2 containing the c-myc epitope at its N terminus was found localized to the endoplasmic reticulum in either COS-1 cells or HepG2 cells transiently expressing the protein (10). When the acyl-CoA synthetase activity of COS-1 cells expressing hVLCS-H2 was assessed, the enzyme weakly activated fatty acids with chain length acids from 18 to 26 carbons. However, hVLCS-H2 was not nearly as robust at activating either long-or very long-chain fatty acids as hVLCS, an enzyme that is also present in liver endoplasmic reticulum (11). Therefore, we questioned whether the endogenous substrate for hV-LCS-H2 might be a liver-specific fatty acyl moiety such as cholic acid.
Amino acid sequence comparisons also supported hVLCS-H2 as a candidate for choloyl-CoA synthetase. Mallonee et al. (20) previously reported that the baiB gene from Eubacterium sp. encoded a bile acid:CoA ligase. The deduced amino acid sequence of hVLCS-H2 exhibited 26% amino acid identity and 48% similarity to the baiB gene product. Furthermore, the bacterial ligase contains a motif that we previously identified as unique to the VLCS family ( Fig. 1) (9).
hVLCS-H2 Activates Cholic Acid-To determine whether hVLCS-H2 could activate cholic acid to its CoA derivative, COS-1 cells transiently expressing this enzyme were incubated with cholic acid in the presence of ATP/Mg 2ϩ and CoA. Reaction products were separated by reverse-phase HPLC as described under "Experimental Procedures." A single peak eluting at 16.2 min was observed (Fig. 2). This peak was not formed when reaction mixtures lacked ATP, CoA, or cholate, suggesting that the peak is choloyl-CoA (data not shown). The retention time of this peak was similar to that reported for authentic choloyl-CoA chromatographed under similar conditions (18). To verify that the reaction product was truly choloyl-CoA, we added a known quantity of authentic choloyl-CoA to hVLCS-H2-containing reaction mixtures prior to injection onto the HPLC column and observed a single peak on chromatograms (Fig. 3).
COS-1 cells transfected with the pcDNA3 vector alone showed little capacity to activate cholate (Fig. 2). This result is consistent with a lack of endogenous choloyl-CoA synthetase in these cells that were originally derived from monkey kidney.   1. Bacterial bile acid:CoA ligase and hVLCS-H2 share a conserved motif. We previously described a motif unique to the VLCS/ FATP protein family (9). 17 of 24 residues in this region are identical in the mammalian proteins (bold). This motif was also present in the bacterial protein encoded by the baiB gene, a bile acid:CoA ligase. 13 of 24 amino acids of BaiB protein and hVLCS-H2 are identical (shaded). mFATP, murine FATP. Furthermore, hVLCS, the heart-specific hVLCS-H1, and rat LCS (rACS1) were all incapable of activating cholic acid when expressed in COS-1 cells (Fig. 2). In contrast, each of these expressed proteins showed enzymatic activity with their appropriate long-or very long-chain fatty acyl substrates (Table II).
Preference of hVLCS-H2 for Cholic Acid-To assess more thoroughly the substrate preference of hVLCS-H2 for cholic acid, acyl-CoA synthetase assays using [1-14 C]cholic acid as substrate were performed. As shown in Table II, COS-1 cells expressing hVLCS-H2 activated cholic acid to its CoA derivative. Cells expressing vector alone, hVLCS, hVLCS-H1, or rACS1 had essentially no detectable choloyl-CoA synthetase activity, as predicted from the non-isotopic assay. To compare increases in acyl-CoA synthetase activities due to expression of the different enzymes, the endogenous activity of COS-1 cells (cells transfected with pcDNA3 alone) was subtracted. The hVLCS-H2-specific increase in choloyl-CoA synthetase was more than 2-fold higher than the increase in C24:0 synthetase (1.97 versus 0.80 nmol/hr/mg of protein), suggesting that cholic acid, rather than very long-chain fatty acid, was the preferred substrate of hVLCS-H2.
Choloyl-CoA Synthetase Activity in HepG2 Cells-We previously showed by reverse-transcription PCR that hVLCS-H2 message was present in the human hepatoma-derived cell line, HepG2, whereas no message was detected in skin fibroblasts, heart or brain (10). To verify that HepG2 cells were capable of activating cholic acid, these cells were incubated with [1-14 C]cholic acid. Endogenous cholate synthetase activity of HepG2 cells was 7-fold higher than that of COS-1 cells, whereas activation of C16:0 and C24:0 were comparable in the two cell lines (Table III).
Role of Choloyl-CoA Synthetase In Bile Acid Metabolism-Conversion of the C27 sterol, cholesterol, to the C24 bile acids is a multistep process (reviewed in Refs. 1 and 2). Initial reactions include a series of ring modifications including hydroxylations, saturation of the ring system, and epimerization of the 3-␤-hydroxyl group. One of the two terminal methyl groups of the side chain is oxidized to a carboxylic acid, yielding trihydroxycholestanoic acid (THCA) and dihydroxycholestanoic acid (DHCA), the primary intermediates in bile acid synthesis. These synthetic reactions are catalyzed by enzymes in the endoplasmic reticulum, cytosol, and mitochondria. Subsequently, the side chains of THCA and DHCA must be activated to their CoA derivatives, followed by one cycle of ␤-oxidation in peroxisomes (21). The products of THCA and DHCA ␤-oxidation are choloyl-CoA and chenodeoxycholoyl-CoA, respectively; propionyl-CoA is released in this process. Conjugation to glycine or taurine, catalyzed by bile acid CoA:amino acid N-acyltransferase, has also been reported to occur in peroxisomes (22). Glyco-and tauro-conjugates exit the peroxisome by an unknown mechanism and are secreted into bile canaliculi.
If taurocholate and glycocholate are synthesized de novo within peroxisomes, what then is the role of the endoplasmic reticulum protein choloyl-CoA synthetase/hVLCS-H2? About 90% of the bile acids secreted into the small intestine are reabsorbed throughout the small bowel, primarily at the distal ileum (1,23). Bacterial flora deconjugate and subsequently dehydroxylate a considerable portion of cholic and chenodeoxycholic acids (primary bile acids), yielding the secondary bile acids, deoxycholic and lithocholic acids, respectively (1,23). Any unconjugated bile acids returning to the liver must be re-activated and re-conjugated. Whereas the highest specific activity of bile acid CoA:amino acid N-acyltransferase has been found in peroxisomes, activity is also found in cytosol and endoplasmic reticulum (6,7,22). Thus, based on the subcellular localization for hVLCS-H2 and other bile acid-metabolizing enzymes, we hypothesize that the primary role of hVLCS-H2 may be in the re-activation pathway and not the de novo biosynthetic pathway. Studies to evaluate this hypothesis are underway. Furthermore, Wheeler et al. (24) recently purified a bile acid:CoA ligase from rat liver. It will be interesting to determine whether their enzyme is the rat ortholog of hVLCS-H2.   [1-14 C]fatty acids or [1-14 C]cholate by COS-1 cells transiently expressing acyl-CoA synthetases COS-1 cells were transiently transfected with the indicated plasmid. Three days post-transfection, cells were harvested, frozen at Ϫ80°C, and subsequently assayed for their ability to activate palmitic acid (C16:0), lignoceric acid (C24:0), or cholic acid as described under "Experimental Procedures." Results are the average of three independent transfection experiments Ϯ S.D. Compared with endogenous (pcDNA3-transfected COS-1 cell) activity, hVLCS activated both C16:0 and C24:0. hVLCS-H2, weakly activated C24:0, whereas rACS1 primarily activated C16:0. These activities are consistent with our previous observations (10,11