Purification, cloning, and bacterial expression of retinol dehydratase from Spodoptera frugiperda.

Anhydroretinol and 14-hydroxy-4,14-retro-retinol, retro-retinoids endogenous to both mammals and insects, act as agonist and antagonist, respectively, in controlling proliferation in lymphoblasts and other retinol-dependent cells. We describe here the identification, purification, cloning, and bacterial expression of the enzyme retinol dehydratase, which converts retinol to anhydroretinol in Spodoptera frugiperda. Retinol dehydratase has nanomolar affinity for its substrate and is, therefore, the first enzyme characterized able to utilize free retinol at physiological intracellular concentrations. The enzyme shows sequence homology to the sulfotransferases and requires 3′-phosphoadenosine 5′-phosphosulfate for activity.

Retinoids and Other Reagents-All-trans-retinol, PAPS, and other chemicals were purchased from Sigma. [ 3 H]Retinol was purchased from DuPont NEN. Synthetic AR was synthesized by acid-catalyzed dehydration of all-trans-retinol, followed by purification on HPLC as described (11).
Retinol Dehydratase Assay-S. frugiperda cells (10 7 ) were incubated with 1 Ci of [ 3 H]retinol in Grace's insect cell medium for 6 h and delipidated according to the procedure by McClean et al. (17). Radiolabeled retinoid metabolites were analyzed by on-line liquid scintillation counting after separation by HPLC gradient elution (20 mM Tris⅐HCl, pH 7.4, methanol, chloroform) from a C 18 reverse phase 201TP54 (Vydac) column, as described previously (10,11). Retinol dehydratase activity in fractionated Sf-21/Sf-9 cell sonicates was assayed by incubation of 10 7 cell eq. in 200 l of assay buffer (20 mM Tris⅐HCl, pH 7.5, 150 mM NaCl, and 1 mM MgCl 2 ), 1 Ci of [ 3 H]retinol, and as indicated, supplemented with either 50 l of Sf-9 cytosolic supernatant or 2 M PAPS for 60 min at 24°C, followed by HPLC analysis.
Cloning and Expression of Retinol Dehydratase-Purified retinol dehydratase was electroblotted onto nitrocellulose membrane. The protein sequencing facility of Sloan-Kettering Institute was used to subject the protein to trypsin digestion. Peptides were separated by reversephase HPLC. Three peptides were sequenced, and the predicted masses were confirmed by mass spectroscopy. Degenerate oligonucleotides were synthesized to the peptides, and a 200-bp probe was generated from Sf-21 cDNA by PCR. The 32 P-labeled 200-bp DNA probe was used to screen a Sf-21 cDNA pSPORT-1 expression library (Life Technologies, Inc.). The cDNA sequence was determined from both strands. The coding region of a full-length clone (#61) was subcloned into the bacterial HisTag expression vector pET-15b (Novagen, Madison, WI). Recombinant enzyme was purified using a Ni 2ϩ affinity column.
Kinetic Analysis-Kinetic studies were performed with native retinol dehydratase (0.1 g/assay) and [ 3 H]retinol in the presence of constant substrate carrier protein concentrations, either delipidated BSA or recombinant CRBP at 24°C. The initial rate of AR synthesis at different substrate concentrations was determined by on-line liquid scintillation counting of [ 3 H]AR after HPLC separation. Conversion of * This work was supported by National Institutes of Health Grant DK48022 and American Cancer Society Grant BE-227. 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 /EBI Data Bank with accession number(s) U28654.

RESULTS AND DISCUSSION
AR Is Enzymatically Produced in Sf-21 Cells-The all-trans and cis isomers of AR (10,11) were identified as the predominant metabolites of [ 3 H]retinol in the S. frugiperda cell line Sf-21 (Fig. 1A); the closely related cell line Sf-9, however, did not synthesize detectable levels of AR (Fig. 1B). AR isomers were identified by coelution with synthetic standards and their characteristic vibronic fine structure in the UV-visible absorption spectrum: maxima at 270, 348, 368, and 390 nm. AR synthesis in Sf-21 cells was time dependent and could be disrupted by 2% glutaraldehyde fixation, 0.1% sodium azide treatment, or heat inactivation at 55°C for 15 min, implying that synthesis of AR is energy and protein dependent. Enzymatic production of AR had also been reported previously in spontaneously transformed mouse fibroblasts (22).
Purification, Cloning, and Bacterial Expression of Retinol Dehydratase-Subcellular fractionation of Sf-21 cells revealed retinol dehydratase activity solely in the 100,000 ϫ g supernatant fraction (Fig. 1, C and D). Enzyme activity was lost following dialysis (Fig. 1E) but could be restored by supplementation with cytosol from either Sf-9 cells (Fig. 1F) or heatinactivated Sf-21 cells (data not shown), indicating a requirement for a small dialyzable cofactor.
The enzyme activity was purified by sequential column chromatography. Activity was monitored by quantitation of [ 3 H]AR production in the presence of Sf-9 cytosolic factor. The activity eluted in one peak, corresponding to an apparent size of 35-45 kDa from a gel filtration column, at 0.7-0.4 M ammonium sulfate from a methyl hydrophobic interaction column, at 0.2-0.3 M NaCl from a Mono Q anion exchange column, and at 0.125 M NaCl from a DEAE-5PW anion exchange column. The final enrichment of the enzyme activity was 11,700-fold and correlated with a 41-kDa protein ( Fig. 2A). Amino acid sequence information on three tryptic peptides (peptide sequences bracketed in Fig. 3) was used to generate a specific cDNA PCR probe for the screening of a Sf-21 cDNA plasmid library. Four independent full-length clones predicted the same 352-amino acid residue protein (41.5 kDa). Each clone exhibited retinol dehydratase activity when assayed in crude bacterial lysates (Fig.  2B). Recombinant retinol dehydratase was subsequently expressed and purified using a HisTag fusion system (Fig. 2C).
Retinol Dehydratase Is a Sulfotransferase-The predicted amino acid sequence (Fig. 3) is homologous to that of sulfotransferases (overall 20 -26% amino acid homology; 35% for a contiguous 200-amino acid C-terminal region). Sulfotransferases transfer sulfonate (SO 3 Ϫ ) groups from the universal active sulfate donor to acceptor alcohol or amine functional groups. Using purified recombinant retinol dehydratase, PAPS was necessary to restore activity and was sufficient in replacing the Sf-9 cytosol supplement (k m ϭ 0.26 Ϯ 0.05 M; n ϭ 3), whereas adenosine 5Ј-phosphosulfate, the biosynthetic precursor of PAPS, was inactive. The reaction mechanism for retinol dehydratase probably proceeds via the sulfated intermediate, retinyl sulfate.
Kinetic Analyses-The k m for retinol was determined by kinetic analyses (Fig. 4A), performed with purified natural retinol dehydratase and confirmed with recombinant enzyme. Carrier protein for retinol, either delipidated BSA or recombinant human CRBP-1, was added to reduce nonspecific binding and micelle formation. All assays gave a V max of 490 Ϯ 50 pmol AR min Ϫ1 mg Ϫ1 enzyme (n ϭ 5), but the k m for retinol varied between assays containing different carrier proteins when the total retinol present in the system was used as the nominal substrate concentration (k m was 4 ϫ 10 Ϫ7 M with 10 Ϫ3 M BSA, 5 ϫ 10 Ϫ8 M with 10 Ϫ4 M BSA, and 2.5 ϫ 10 Ϫ8 M with 2 ϫ 10 Ϫ7 M CRBP-1). However, recalculating the k m values in reference to the free retinol concentration present in the reaction mixture (obtained from the respective binding affinities of carrier proteins, i.e. BSA K d ϭ 1 ϫ 10 Ϫ6 M for retinol (23); CRBP-1 K d ϭ 1.2 ϫ 10 Ϫ8 M for retinol (24) gave a consistent average k m for retinol of 1.0 ϫ 10 Ϫ9 M (range, 0.6 -2.0 ϫ 10 Ϫ9 M, n ϭ 5).
Binding of Retinol to Retinol Dehydratase-The extremely low k m value for retinol suggests a high affinity interaction between enzyme and substrate. Direct binding experiments performed with recombinant enzyme by fluorescence titrations in the absence of cofactor PAPS, i.e. under conditions that do not allow catalysis, indicated an equilibrium dissociation constant (K d ) of 2.7 nM Ϯ 0.8 nM (n ϭ 7) (Fig. 4B). These data also indicate that high affinity binding of retinol to the enzyme does not require the presence of the cofactor PAPS. Due to its high . Bracketed residues represent tryptic peptides T34 (A-K, residues 26 -37), T66 (S-K, residues 74 -96), and T56 (Y-P, residues 103-125), for which amino acid sequence data were obtained. Highly conserved sulfotransferase sequence motifs (29,30), implicated for a PAPS binding site, are underlined (GenBank accession number, U28654). affinity, retinol dehydratase would not be rate limited at physiological concentrations of free retinol.
Distinct sulfotransferases are responsible for the sulfation of steroid hormones, thyroid hormones, monoamine neurotransmitters, and alcohols (25); sulfation of these bioactive signaling molecules modulates their activities (26). Sulfotransferases are also responsible for detoxification of xenobiotics by sulfation (25,26). However, retinol dehydratase does not appear to act as part of a xenobiotic elimination pathway. The low K d and K m of retinol dehydratase for retinol (10 Ϫ9 M) is indicative of a highly specific interaction similar to endogenous steroid substrates and their cognate sulfotransferases. In contrast, xenobiotic substrates and the aryl sulfotransferase family members generally display 10 Ϫ3 -10 Ϫ5 M affinities (27,28). Furthermore, insects utilize retinol metabolites (retinal and 3-OH-retinal) as chromophores in vision, demonstrating a requirement for retinol-metabolizing enzymes. In addition to AR, insect cell lines metabolize retinol to 3-OH-retinol, retinoic acid, 14-HRR, and retinylesters. 2 We speculate that in mammals the synthesis of the antagonist AR, from the prohormone agonist retinol via the retinol dehydratase (sulfotransferase) pathway, may help to locally inhibit the proliferation of fibroblasts and suppress immune system activation. This may be of benefit in tissues continually subjected to environmental stress and insult, e.g. liver and lung, where chronic nonspecific activation needs to be downregulated; both tissues are sites of high AR biosynthetic activity. 3 The enzymatic control of ligand synthesis, in addition to putative retro-retinoid receptor proteins, is, therefore, likely to be an important regulatory element in this novel signal transduction pathway. Retinol dehydratase represents the prototypic experimental system for studying how a cytosolic retinolutilizing enzyme may regulate local levels of "vitamin A activity" in target cells and tissues.