Molecular Cloning and Expression of cDNA Encoding 3α,7α,12α-Trihydroxy-5β-cholestanoyl-CoA Oxidase from Rabbit Liver*

The steroid side chain cleavage in bile acid formation is catalyzed by liver peroxisomal enzymes (Pedersen, J. I. and Gustafsson, J. (1980) FEBS Lett. 121, 345–348; Kase, F., Björkhem, I., and Pedersen, J. I. (1983) J. Lipid Res. 24, 1560–1567). We here describe the cloning and sequencing of a cDNA coding the first of these enzymes, a 3α,7α,12α-trihydroxy-5β-cholestanoyl-CoA oxidase (THCA-CoA oxidase) from rabbit liver peroxisomes. After tryptic digestion of purified protein in a polyacrylamide gel, five peptides were isolated and sequenced. Using two oligonucleotides deduced from the amino acid sequence data, two overlappping clones were isolated from a rabbit liver cDNA library, which together made up a unique cDNA sequence of 2139 base pairs. It contained an open reading frame of 2046 base pairs encoding a protein of 681 amino acids with a molecular mass of 76,209 daltons. All five peptides could be localized within the sequence. Transfection of COS cells with the coding part of the cDNA resulted in a significant expression of THCA-CoA oxidase activity. We were not able to demonstrate 3α,7α-dihydroxy-5β-cholestanoyl-CoA oxidase activity under the same conditions. The obtained sequence showed 73.6% similarity with a proposed rat THCA-CoA oxidase and 81% similarity with a recently reported human branched chain acyl-CoA oxidase, indicating that these three proteins represent the same enzyme. The similarity with rat palmitoyl-CoA oxidase was 41.8%. The C-terminal tripeptide of the protein was SNL, a previously undescribed variant of the main class of peroxisomal targeting signals. Northern blot analysis revealed that the gene is transcribed in liver and kidney, and the major mRNA fraction had a size of approximately 2.6 kilobase pairs.

Formation of bile acids from cholesterol takes place exclusively in the liver. The sequence of reactions involves a large number of enzymes localized to several compartments of the cell (1). The final steps start with activation to CoA esters of 3␣,7␣,12␣-trihydroxy-5␤-cholestanoic acid (THCA) 1 and of 3␣,7␣-dihydroxy-5␤-cholestanoic acid (DHCA) in the endoplas-mic reticulum (2,3) followed by oxidative cleavage of the steroid side chain in peroxisomes with formation of cholic acid and chenodeoxycholic acid, respectively. The reaction mechanism of this oxidative cleavage is similar to that of ␤-oxidation of fatty acids (4 -6), but a different oxidase seems to be involved (7,8).
Recently, it was shown that rat liver peroxisomes contain a noninducible THCA-CoA oxidase that could be separated from palmitoyl-CoA oxidase inducible by clofibrate and another noninducible acyl-CoA oxidase (9). This last enzyme was found to have the highest activity toward branched chain fatty acids such as the synthetic 2-methylpalmitic acid and the natural pristanic acid and has therefore been termed pristanoyl-CoA oxidase (10). In human liver, only two acyl-CoA oxidases have been detected, palmitoyl-CoA oxidase and a branched chain acyl-CoA oxidase active also on bile acid intermediates (11,12).
THCA-CoA oxidase has been difficult to purify both because of the small amounts of protein available and because of the instability of the enzyme. Recently, Van Veldhoven et al. reported on the successful purification (13) and (during preparation of this manuscript) also on the cloning of this enzyme from rat liver (14) as well as of a similar enzyme from human liver (15). We have noted that a THCA-CoA oxidase from rabbit liver was more stable than its rat counterpart, and this source also provided more material. In the present work we report on the partial purification of THCA-CoA oxidase from rabbit liver, cloning and sequencing of a cDNA coding for the enzyme, and expression of enzyme activity in COS cells after transfection with this cDNA.

EXPERIMENTAL PROCEDURES
Materials-3␣,7␣,12␣-Trihydroxy-5␤- [7-3 H]cholestan-27-oic acid as well as 3␣,7␣,-dihydroxy-5␤-cholestan-27-oic acid were prepared as described (16) using bile from Alligator mississipiensis as starting material. The natural form of THCA in bile of A. mississipiensis has been identified as the 25R-diastereoisomer (17), but the strong alkaline hydrolysis generally used during isolation causes isomerization at C-25 and the appearance of the 25S-isomer (18). The isolated material was separated by HPLC on a 5-m C-18 Nucleosil column (0.5 ϫ 25 cm) with 24% 30 mM trifluoroacetic acid (pH 2.9 with triethylamine) in methanol as eluting solvent. The 25R-and the 25S-isomers of THCA were almost completely separated in this system (19). The ratio between the 25R and the 25S isomers in the material primarily isolated was about 7:3. The combined fractions of tritium-labeled (25R)-and (25S)-THCA as well as unlabeled DHCA were converted into CoA esters using the same method as used for the synthesis of choloyl-CoA (20). Purified palmitoyl-CoA oxidase was a gift from Prof. T. Hashimoto (Shinshu University, Matsumoto, Japan). 2Ј,7Ј-Dichlorofluorescein diacetate was from Eastman Kodak Co. (Rochester, NY). The sources of special reagents and kits are given below. Other chemicals were commercial high purity material. * This work was supported by Swedish Medical Research Council Grants 03X-3141 and B96 -13V-11540 -01 S1404. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  Enzyme Preparation-Liver peroxisomes were prepared from a male rabbit fasted for 24 h essentially as described for rat liver (21). The light mitochondrial fraction generated by centrifugation of the postmitochondrial supernatant at 16,200 rpm (24,000 ϫ g av ) for 10 min in the Kontron T324 centrifuge (A8.24 rotor) was layered on a linear Nycodenz gradient ranging from 15% (w/v) in 0.25 M sucrose, 1 mM EDTA, and 1 mM Hepes at pH 7.4 to 45% in 1 mM EDTA, 1 mM Hepes at pH 7.4. The gradient tubes contained a prelayered 2-ml Maxidenz cushion and were centrifuged at 30.000 rpm for 30 min at 4°C in the Sorvall TV 850 vertical rotor. Fractions of 1.7 ml were collected and assayed for catalase activity (22,23). The five fractions with the highest activity were combined, diluted with 0.25 M sucrose, and centrifuged at 33,000 rpm (100,000 ϫ g) for 45 min. The pellet (14 mg of protein from two gradients) was resuspended in 10 ml of solubilization buffer consisting of 10 mM sodium pyrophosphate, pH 9, 10% ethylene glycol, 0.1% Triton X-100, 10 M FAD, 0.5 mM dithiothreitol. After stirring on ice for 1 h and centrifugation at 33,000 rpm for 45 min, the supernatant was brought to pH 7.4 and applied to a hydroxylapatite (Bio-Gel HTP, Bio-Rad) column (volume, 5 ml; diameter, 1 cm). The column was equilibrated in 10 mM phosphate buffer pH 7.4, 10 M FAD, 0.5 mM dithiothreitol, 10% ethylene glycol and eluted with the same buffer to remove unbound protein. The flow rate was 0.35 ml/min, and 1.75-ml fractions were collected. Protein containing THCA-CoA oxidase (and devoid of palmitoyl-CoA oxidase activity) was eluted with 60 mM potassium phosphate buffer, pH 7.4, 10 M FAD, 0.5 mM dithiothreitol, and 10% ethylene glycol. The active fractions (total volume, 9.5 ml) were concentrated to 2.2 ml in a dialysis bag (Spectra-Por 2 membrane) covered with polyethylene glycol (molecular weight, 15,000 -20,000; Sigma). Aliquots (0.45 ml) of the concentrated protein solution were subjected to gel filtration on a Superdex 200 HR 10/30 column connected to a Smart chromatography system (Pharmacia Biotech, Inc., Uppsala, Sweden). Elution was performed with phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 10 mM KP i buffer, pH 7.4) containing 0.5 mM dithiothreitol and 20% ethylene glycol at a flow rate of 0.25 ml/min, and fractions of 0.25 ml were collected.
Generation of Peptides by Digestion in Gel-Protein in the peak fractions from the gel filtration above was concentrated by precipitation (24), and the dried sample was dissolved in 25 l of sample buffer containing 8 mM dithiothreitol. After heating to 95°C for 4 min, the reduced sample was alkylated by incubation at room temperature for 20 min with 16 mM iodoacetamide. The sample was subjected to polyacrylamide gel electrophoresis in SDS (25) on a 10% gel of 1-mm thickness using the Mini-Protean II Slab Cell (Bio-Rad). After Coomassie Blue staining, the band corresponding to a molecular mass of 72 kDa, using the 72-kDa band of the palmitoyl-CoA oxidase as a reference (26), was excised, transferred to an Eppendorf tube, and subjected to in-gel digestion essentially according to Hellman et al. (27). In brief, the gel pieces were washed with 0.5 M Tris-Cl, pH 9.2, and 50% acetonitrile and then completely dried. To introduce the protease into the gel piece, 10 l of 0.1 M Tris-Cl, pH 9.2, containing 0.5 g of endo-LysC (protease source Achromobacter lyticus, WAKO Chemicals GmbH, Neuss, Germany) was added. Rehydration was continued by adding 0.1 M Tris-Cl, pH 9.2, in small aliquots until the gel piece was immersed. After overnight incubation at 30°C, the supernatant was saved and combined with extractions from the gel piece. Generated peptides were isolated by reversed phase liquid chromatography on a RPC C2/C18 SC 2.1/10 column in a SMART TM chromatography system. Peptides were sequenced on a Perkin-Elmer Applied Biosystems model 494 instrument, following the manufacturer's instructions.
cDNA Cloning-Oligonucleotides were synthesized on a 380B DNA synthesizer (Applied Biosystems, CA), using the ␤-cyanoethylphosphoamidite chemistry. A cDNA library made from New Zealand White rabbit liver in the phage gt10 was obtained from CLONTECH (Palo Alto, CA). The library (5Ј-STRETCH PLUS cDNA library) was constructed with oligo(dT) and random priming utilizing EcoRI linkers and was found to contain 2.5 ϫ 10 10 plaque-forming units/ml. The library was propagated in Escherichia coli C600 Hfl according to the manufacturer's recommendations. Screening of the library was performed with oligonucleotide probes, end-labeled with 32 P. Positive clones were isolated, and the DNA was purified with the Wizard TM Lambda Preps DNA purification system (Promega, Madison, WI).The inserts were excised from the vector by digestion with EcoRI, extracted from the agarose gel (GenElute TM agarose spin columns, Supelco, Bellefonte, PA), and subcloned into pUC18 using the Ready-To-Go TM DNA Ligase (Pharmacia). Plasmid DNA was purified with the Jetstar Mini Plasmid Purification system (Genomed, Research Triangle Park, NC). Sequencing of inserts were done automatically on a 373 DNA sequencer stretch (Applied Biosystems), using the ABI PRISM TM dye terminator cycle sequencing kit with AmpliTaq R polymerase FS (Perkin-Elmer).
Transfection of COS-M6 Cells-Stock cultures of the cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and 100 units/ml penicillin, 100 g/ml streptomycin. Transfection was carried out by electroporation in 0.4-cm cuvettes (Gene Pulser II without the pulse controller unit; Bio-Rad), using a single pulse of 0.40 kV and 25 microfarads. After transfection, the cells were grown for 48 h, harvested by low speed centrifugation, and homogenized (sonication for 6 -10 s) in the incubation medium.
THCA-CoA racemase activity was assayed in the same incubation buffer by following the conversion of (25R)-THCA-CoA into (25S)-THCA after hydrolysis and extraction as above. Protein was determined by the method of Bradford (28) using bovine serum albumin as a standard. Protein in the eluent from the gel filtration column was estimated based on the UV absorbance in the eluent compared with that of a fixed amount of injected bovine serum albumin.

RESULTS
Partial Purification of THCA-CoA Oxidase-During our attempts to purify the enzyme, we experienced, as did others (13), considerable problems with the stability of the enzyme. During the different purification steps, we noticed, however, that activity was always associated with a band on SDS-polyacrylamide gel electrophoresis of about 72 kDa and a migration distance identical to that of the 72-kDa fragment of purified palmitoyl-CoA oxidase (26). Even if it was possible after several purification steps to obtain a preparation that showed only this band on SDS-polyacrylamide gel electrophoresis, the total amount of protein was too small for further characterization (29). We therefore decided to reduce the number of purification steps and base our cloning strategy on in-gel digestion and peptide analysis of this protein band. The steps used are shown in Table I. Solubilization of the peroxisomes in hypotonic pyrophosphate buffer in the presence of detergent on ice was found to be efficient. Heat treatment was avoided, since it was found that, unlike palmitoyl-CoA oxidase (26), THCA-CoA oxidase was very heat-labile. The acyl-CoA oxidase activity and the THCA-CoA activity were separated on a hydroxylapatite column. THCA-CoA oxidase eluted at 60 mM phosphate buffer, while acyl-CoA oxidase was eluted at 200 mM phosphate buffer. No acyl-CoA oxidase was detected in the 60 mM fraction eluate, while a small amount of THCA-CoA oxidase eluted in the 200 mM fraction. The 60 mM fraction eluate was concentrated by dialysis against polyethylene glycol and further fractionated by gel filtration (Fig. 1). Maximum THCA-CoA oxidase activity was detected in fraction 14 that eluted between 13.00 and 13.25 ml, which was immediately in front of the elution volume corresponding to bovine serum albumin (13.5 ml), suggesting a molecular mass of the oxidase of about 70 kDa. Fractions between 12.75 and 13.50 ml were used as a source of the oxidase. Activity measurement in the active fractions was complicated by the presence of a THCA-CoA racemase activity (interconversion of (25R)-and (25S)-THCA-CoA) (19) that started to elute immediately after the peak fraction of THCA-CoA oxidase activity. Maximum THCA-CoA racemase was found to be in a fraction between 14.75 and 15.00 ml, corresponding to a molecular mass of about 40 kDa. This compares with the values given by Schmitz et al. of 45 kDa for a similar rat liver mitochondrial enzyme (30) and 47 kDa for a human liver peroxisomal enzyme (31). Fractions between 14.5 and 15.5 ml (fractions 20 -23) were pooled and used as a source of racemase in incubations when required (see legend to Fig. 1). The racemase preparation was found to be contaminated with some THCA-CoA oxidase activity, corresponding to about 10% of that in the peak fraction of the oxidase. The substrate used for the THCA-CoA oxidase assay contained the 25R-and 25S-isomers in a ratio of 7:3, and since the S-isomer appears to be the physiological one (19), the amount of this substrate was not at saturation level. The presence of the racemase activity therefore resulted in an erroneously high activity in some of the fractions on the low molecular side of the eluate. For this reason a fixed amount of racemase was included in all incubations of fractions from the gel filtration step to be assayed for the oxidase activity.
Isolation and Sequencing of Tryptic Peptides and Preparation of Oligonucleotides for Screening-Polyacrylamide gel electrophoresis of the active fractions from the gel filtration showed the appearance of a band at aout 72 kDa that varied in intensity corresponding to the activity (Fig. 2). The migration of this band was identical to that of the 72-kDa band of palmitoyl-CoA oxidase. In-gel digestion of the protein contained in this band and separation by HPLC generated a large number of peptides (Fig. 3), five of which were sequenced and found to contain 6 -17 amino acids (not including the initial lysine required by the protease endo-LysC for cleavage). The sequence of peptide 3 was used to design two oligonucleotides intended for screening for the THCA-CoA oxidase cDNA (Fig. 4). In accordance with the cleavage pattern of the enzyme endo-LysC, it was assumed that a lysine residue would precede the aminoterminal aspartic residue of the peptide. Two oligonucleotides were synthesized, denoted Ox-1 and Ox-2, which differed only in the codon for serine, where different triplets were utilized (Fig. 4). The most prevailing codons for the amino acids found in the rabbit genome were used (32), and in two positions with a high degree of degeneracy inosine was inserted (Fig. 4).
Isolation of Rabbit THCA-CoA Oxidase Clones and Sequence of a Full-length Clone-Approximately 500,000 plaque-forming units were screened, and only those plaque-forming units were picked that generated a signal with both oligonucleotides. In this way, several cDNA-clones were isolated. One of these, denoted 10, numbered 2092 bp and contained an open reading frame of 2046 bp (Fig. 5). Another shorter clone (69) overlapped with 10 in the 5Ј-end with 47 bp, making the total  1. Gel filtration chromatography of the concentrated 60 mM KP i fraction eluate from the hydroxylapatite column. Approximately 0.25 mg of protein in 0.45 ml was subjected to gel filtration as described under "Experimental Procedures." Fractions of 0.25 ml were collected. THCA-CoA oxidase (open circles) was assayed as described under "Experimental Procedures" except for the addition of 10 l of a racemase fraction (combined fractions 20 -23) to all fractions to be assayed (see "Results"). The specific activity of the peak fraction is given in Table I Fig. 1 and 0.5 g of purified palmitoyl-CoA oxidase (PO) were applied to the gel. The protein bands were visualized by silver staining using the Bio-Rad Silver Stain kit. The most intensely colored band with a migration distance identical to that of the 72-kDa band of palmitoyl-CoA oxidase was that of fraction 14, which also corresponded to the one with maximum THCA-CoA oxidase activity. cDNA sequence obtained 2139 bp. Since the cDNA sequence contained an EcoRI site (in position 1173 of the complete sequence), the insert of 10 was cleaved in two fragments, which were subcloned separately in plasmid vectors. The sequence around the EcoRI site was verified by sequence analysis of a polymerase chain reaction fragment from the phage clone, spanning this region. All of the five peptides could be localized within the sequence (Fig. 6). The shortest of the peptides (peptide 2) represented the C-terminal end of the polypeptide chain. However, this peptide partly overlapped with another of the peptides (peptide 4), and therefore the enzyme cleavage must have occurred after a glutamine and not, as expected, after a lysine residue. Whether this indicates the presence of microheterogeneity in this position or possibly some protease contaminant in the endo-LysC preparation is presently not known.
The open reading frame encodes a protein of 681 amino acids with a molecular mass of 76,209 daltons and a theoretical isoelectric point of 7.74.
The obtained peptide sequence showed a homology with rat THCA-CoA oxidase of 73.6% and a homology with rat acyl-CoA oxidase of 41.8% (Fig. 7). Interestingly, there are two additional methionine residues in the amino-terminal part of the deduced sequence (in positions 18 and 22). The most proximal of these corresponds to the initiation methionine residue of rat acyl-CoA oxidase, the sequence of which is 17 amino acid residues shorter in the amino-terminal end than the other two enzymes. Very recently, the sequence for a branched chain acyl-CoA from human liver was published (15) that has a homology of 81% with our enzyme and contains the same number of amino acid residues.
The C-terminal tripeptide in our protein was found to be SNL (Ser-Asn-Leu), which is different from previously described peroxisomal targeting signals (33). In particular, the sequence is different from HKM (His-Lys-Met) found in rat THCA-CoA oxidase (14) and SKL (Ser-Lys-Leu) found in human branched chain acyl-CoA oxidase (15).
Expression Pattern of THCA-CoA Oxidase-To visualize the mRNA for THCA-CoA oxidase, a cDNA probe of 1135 bp (corresponding to positions 47-1172 in the cDNA sequence) was used. A strong signal was obtained with rabbit liver total RNA corresponding to a band at 2.6 kilobase pairs (Fig. 8). A much weaker signal was obtained from rabbit kidney, while no signal was detected from other organs tested (not shown). Signals were also obtained with mouse and rat liver (Fig. 8). Fasting had no significant effect on mRNA level in any of the three species tested.
Expression of THCA-CoA Oxidase Activity in COS Cells-To express THCA-CoA activity in eukaryotic cells, a plasmid vector, pCAGGS, carrying a modified chicken ␤-actin promoter (35) was used. A cDNA fragment including the coding sequence of the oxidase was prepared from the clone 10 by polymerase chain reaction (Gene Amp XL polymerase chain reaction kit; Perkin-Elmer). The primers contained a restriction site for XbaI, and the fragment obtained numbered 2081 bp, corresponding to positions 46 -2126 in the cDNA. This fragment was ligated into the XbaI site of the expression vector pCAGGS, and the construct was propagated in Epicurian Coli XLI-Blue MR-FЈKan supercompetent cells (Stratagene, La Jolla, CA). The insert could be cleaved out of the vector by digestion with XbaI, and the orientation was established by DNA sequencing. An estimated quantity of 20 g of plasmid DNA with the cDNA inserted the correct way was used to transfect approximately 2 ϫ 10 6 cells. In a control experiment, a similar quantity of plasmid DNA with the cDNA inserted in the opposite direction was allowed to transfect the same number of cells.
When incubating a homogenate of the transfected cells with tritium-labeled THCA-CoA as described under "Experimental Procedures," a low but significant THCA-CoA oxidase activity was detected with the assay based on radio-HPLC (Fig. 9). Two independent experiments resulted in 3 and 2.3% total conversion of (25R)-and (25S)-THCA-CoA into ⌬ 24 -THCA. A homogenate of cells transfected with a construct with the coding region inversely inserted had no detectable activity. It should be noted FIG. 3. HPLC chromatogram of peptides obtained after in-gel digestion of protein contained in the 72-kDa band (Fig. 2). The peptides sequenced were as follows: peptide 1, (K)RYHLQK; peptide 2, (K)SWRSNL; peptide 3, (K)D-NYFMSQNELYEAATRK; peptide 4, (K)-YIQPLLQSWRSN; peptide 5, (K)NILXX-DFTLLPE. The first K of each peptide is within parenthesis to indicate that the endopeptidase used, endo-LysC, always cleaves after a lysine. In peptide 5, XX denotes two uncertain amino acids.

FIG. 4. Amino acid sequence of peptide 3 obtained after in-gel
digestion of the 72-kDa band and the sequence of corresponding oligonucleotides used for screening. Note that the amino-terminal Lys was not found in the peptide but rather, as described in the legend to Fig. 3, deduced from the specificity of endo-LysC.
that it is only (25S)-THCA-CoA that is a substrate of the enzyme (19). It could be calculated that the conversion of this stereoisomer into ⌬ 24 -THCA was 13 and 16%, respectively, in the above two experiments with cells transfected with the cDNA inserted the correct way. (cf. Fig. 9).
Combined gas chromatography-mass spectrometry allows a considerably more sensitive and accurate assay than the above HPLC method. Fig. 10 shows analysis by this method of hydro-  (14), and rat palmitoyl-CoA oxidase (rACox) (33). Residues matching the whole peptide sequence are shaded. 3␣,7␣,12␣-Trihydroxy-5␤-cholestanoyl-CoA Oxidase lyzed extracts of incubations of (25R)-and (25S)-THCA-CoA with the same amount of cells transfected by either a complete or a truncated insert. The ion at m/z 500, corresponding to the molecular ion of the methyl ester trimethylsilyl derivative of THCA, and the ion at m/z 498, corresponding to the molecular ion of the same derivative of ⌬ 24 -THCA, were followed through the gas chromatography. There was no separation of the R-and S-stereoisomers of THCA on the column used. In the analysis of the incubation with cells transfected with the complete insert (Fig. 10A), a conversion of 1.2% of the substrate (25R)-and (25S)-THCA-CoA, was obtained. In addition, there was another product peak, corresponding to a conversion of 0.3%, occurring in the tracing of the ion at m/z 498 (Fig. 10A). This peak had a retention time identical to that of the methyl ester trimethylsilyl ether of 24-OH-THCA. This compound has a prominent ion at m/z 498 in its mass spectrum, corresponding to the M-90 ion. Formation of 24-OH-THCA is most probably due to the presence of hydratase activity in the COS cells, converting the newly synthesized ⌬ 24 -THCA-CoA into the corresponding 24hydroxylated product. A cDNA fragment devoid of the N-terminal sequence corresponding to the region up to the second methionine (in position 18) was also used in a separate expres- sion experiment. A very small product peak calculated to represent a conversion of about 0.3% of the substrate was observed (Fig. 10B). Most probably, this represents a small endogenous THCA-CoA oxidase activity in the COS cells.
When using unlabeled DHCA-CoA as a substrate and analyzing the product by gas chromatography-mass spectrometry as above, no conversion into ⌬ 24 -DHCA could be detected in incubations with the transfected cells. In these analyses, the ion at m/z 412, corresponding to the molecular ion of the methyl ester trimethylsilyl derivative of DHCA, and the ion at m/z 410, corresponding to the molecular ion of the same derivative of ⌬ 24 -DHCA, were used. A significant conversion was obtained, however, in incubations with either purified rabbit liver peroxisomes or with the crude enzyme from the gel filtration step. The rates of these reactions were about half of those obtained with THCA-CoA as substrate. DISCUSSION In the present work, we have cloned a THCA-CoA oxidase from rabbit liver peroxisomes. That the cloned protein truly represents this enzyme was demonstrated by the expression of THCA-CoA oxidase activity after transfection in COS cells. Even if the activity was low, it exceeded severalfold that in mock-transfected cells. Furthermore, the product of the oxidase reaction was verified both by HPLC and by gas chromatography-mass spectrometry. During the preparation of this manuscript, the cloning of two proteins closely related to the one described here was published, a rat peroxisomal THCA-CoA oxidase (14) and a human peroxisomal branched chain acyl-CoA oxidase (15). Expression of the branched chain acyl-CoA oxidase cDNA was demonstrated by the use of antibody against the protein. For neither of the two has enzyme activity been demonstrated in an expression system, however. Formal proof of the correct protein being cloned was thus not presented. The high degree of homology between the protein described here and both rat peroxisomal THCA-CoA oxidase and human peroxisomal branched chain acyl-CoA oxidase (73.6 and 81%, respectively) makes it clear that these three proteins represent the same enzyme from the three different species.
When compared with peroxisomal rat acyl-CoA oxidase, the similarity was only 41.8%. Of special interest was the presence of two methionines at the N-terminal part of rabbit THCA-CoA oxidase, the second of which (at position 18) corresponds to the start of rat acyl-CoA oxidase. Obviously, the initial 17 amino acid residues are essential for catalytic activity, since a clone lacking this portion was unable to elicit a significant THCA-CoA activity when transfected into COS cells. The more specific function of this part of the protein is unknown; possibly, it is required for correct binding of the substrate.
The C-terminal tripeptide of the protein was found to be SNL (Ser-Asn-Leu). This is different from the previously reported C-terminal tripeptides considered to be peroxisomal targeting signals. From site-directed mutagenesis experiments it is known that certain variations of the tripeptide motif are tolerated for it to function as a peroxisomal targeting signal (36). A substitution of Asn in SNL for Lys in SKL (Ser-Lys-Leu), the most common peroxisomal targeting signal (33), is compatible with the rule of a basic amino acid in the second position of the tripeptide (36). It thus appears evident that SNL represents another variant of the major class of peroxisomal targeting signals found in eukaryotic cells (33). It should be noted that SNL differs from the C-terminal tripeptide of human peroxisomal branched chain acyl-CoA oxidase only by Asn being altered to Lys, while it differs from that of rat THCA-CoA oxidase (HKM) in all three positions. From this point of view, rabbit THCA-CoA oxidase is closer to the human protein than to its rat counterpart. Baumgart et al. (14) could not confirm the interaction of the C terminus of rat THCA-CoA oxidase with the human peroxisomal targeting signal receptor, in contrast to the situation with rat palmitoyl-CoA oxidase and pristanoyl-CoA oxidase, and suggested that THCA-CoA oxidase was targeted to peroxisomes by a hitherto unknown mechanism. In view of the finding that SKL and SNL are the C termini of the counterparts of this enzyme in human and rabbit liver, respectively, this suggestion does not appear to be true for THCA-CoA oxidases of other species.
It has generally been considered that both THCA and DHCA are chain-shortened by the same peroxisomal enzymes. The crude enzyme preparation that eluted from the gel filtration column (Fig. 1) converted both THCA-CoA and DHCA-CoA to the respective unsaturated derivatives, although DHCA-CoA was converted at a considerably lower rate than THCA-CoA. This may not be surprising in view of the fact that chenodeoxycholic acid is normally found only in trace amounts in rabbit bile (37). After transfection of COS cells with the appropriate clone, we were not able to detect any activity toward DHCA-CoA. Whether this is due to the activity being too low for detection by the method used or to the existence of a separate oxidase for DHCA-CoA can only be speculated on.
As expected, the enzyme was expressed in livers from both rat and mouse in addition to rabbit. As previously demonstrated (14), kidney also gave a weak signal. Formation of bile acids is considered to take place only in the liver. The function of the enzyme in the kidney is not known. The possibility cannot be excluded that small amounts of cholesterol hydroxylated in the 27-position in other organs may eventually be further oxidized and chain-shortened in the kidneys.