In Vitro Characterization of a Recombinant Blh Protein from an Uncultured Marine Bacterium as a β-Carotene 15,15′-Dioxygenase

Codon optimization was used to synthesize the blh gene from the uncultured marine bacterium 66A03 for expression in Escherichia coli. The expressed enzyme cleaved β-carotene at its central double bond (15,15′) to yield two molecules of all-trans-retinal. The molecular mass of the native purified enzyme was ∼64 kDa as a dimer of 32-kDa subunits. The Km, kcat, and kcat/Km values for β-carotene as substrate were 37 μm, 3.6 min−1, and 97 mm−1 min−1, respectively. The enzyme exhibited the highest activity for β-carotene, followed by β-cryptoxanthin, β-apo-4′-carotenal, α-carotene, and γ-carotene in decreasing order, but not for β-apo-8′-carotenal, β-apo-12′-carotenal, lutein, zeaxanthin, or lycopene, suggesting that the presence of one unsubstituted β-ionone ring in a substrate with a molecular weight greater than C35 seems to be essential for enzyme activity. The oxygen atom of retinal originated not from water but from molecular oxygen, suggesting that the enzyme was a β-carotene 15,15′-dioxygenase. Although the Blh protein and β-carotene 15,15′-monooxygenases catalyzed the same biochemical reaction, the Blh protein was unrelated to the mammalian β-carotene 15,15′-monooxygenases as assessed by their different properties, including DNA and amino acid sequences, molecular weight, form of association, reaction mechanism, kinetic properties, and substrate specificity. This is the first report of in vitro characterization of a bacterial β-carotene-cleaving enzyme.


EXPERIMENTAL PROCEDURES
Reagents-The expression vector pET-24a(ϩ) was purchased from Novagen (San Diego). The expression host, E. coli ER 2566, and all restriction enzymes were purchased from New England Biolabs (Hertfordshire, UK). Luria-Bertani (LB) medium was purchased from BD Biosciences. ␤-Carotene and the other carotenoid substrates were purchased from Sigma and Carotenature (Lupsingen, Switzerland), respectively. The pre-stained ladder for SDS-PAGE and the gel filtration calibration kit were purchased from MBI Fermentas (Hanover, MD) and Amersham Biosciences, respectively. Western blot detection system and His-tagged monoclonal antibody were purchased from Intron Biotechnology (Seongnam, Gyeonggi, Korea) and Bio-Rad, respectively. Isotopically labeled water (H 2 18 O) and molecular oxygen ( 18 O 2 ) were purchased from Sigma and Cambridge Isotope Laboratories (Andover, MA), respectively. All other regents were purchased from Sigma.
DNA Cloning and Site-directed Mutagenesis-The amino acid sequence used for codon optimization was obtained from the blh gene of the uncultured marine bacterium 66A03 (Gen-Bank TM accession number AAY68319). Codons of the blh gene sequence were optimized by selection based on probabilities obtained from the codon usage table without consideration of internal mRNA secondary structures or DNA repeats (31). The entire gene was synthesized (Genofocus, Daejeon, Korea), cloned into the pET-24a(ϩ) expression vector using the restriction enzymes EcoRI and XhoI, and transformed into E. coli ER2566 as an expression host. Mutations of the conserved four histidine residues in the Blh protein were generated by sitedirected mutagenesis using the QuickChange kit and protocol (Stratagene, Beverly, MA). DNA sequencing was performed at the Macrogen facility (Seoul, Korea).
Expression and Purification-The recombinant E. coli for expression of the wild-type and mutant enzymes was cultivated in 500 ml of LB medium (1.0% tryptone, 0.5% yeast extract, and 1.0% sodium chloride) in a 2,000-ml flask containing 20 g/ml kanamycin at 37°C with shaking at 200 rpm. When the absorbance of the culture reached 0.5 at 600 nm, isopropyl ␤-D-thiogalactopyranoside was added to a final concentration of 0.1 mM to induce expression of the recombinant enzyme, and the culture was incubated at 16°C for 16 h. The cells were harvested from the culture broth by centrifugation at 6,000 ϫ g for 30 min at 4°C, washed twice with 0.85% NaCl, and then resuspended in a lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, pH 8.0) containing 1 mg/ml lysozyme. The resuspended cells were disrupted using a sonicator. The cell debris was removed by centrifugation at 13,000 ϫ g for 20 min at 4°C, and the supernatant was filtered through a 0.45-m filter. The filtrate applied to His-Trap HP affinity chromatography (Amersham Biosciences) equilibrated with 50 mM sodium phosphate buffer containing 300 mM NaCl. The column was washed extensively with the same buffer, and the bound protein was eluted with a linear gradient between 10 and 200 mM imidazole at a flow rate of 1 ml/min. The purification step using chromatography was carried out in a cold room at 4°C with a fast protein liquid chromatography system (Bio-Rad). The active frac-tions were collected and dialyzed against 100 mM Tricine 2 -KOH buffer (pH 8.0). After dialysis, the resulting solution was used as the purified enzyme.
Amino Acid Sequencing-Partial amino acid sequences of the Blh protein were investigated. After separation via SDS-PAGE and staining of the sample, each chosen band was isolated, destained, and washed. In-gel digestion was then performed with 500 ng of sequencing-grade chymotrypsin in 100 mM Tris-HCl buffer (pH 8.0) with 10 mM CaCl 2 at 37°C for 16 h. The digested peptides were extracted with 5% formic acid in acetonitrile and cleared by centrifugation at high speed for 5 min. The supernatant was dried in a SpeedVac for mass analysis. After desalting with Zip-Tip filter (Millipore, Billerica, MA), the digested peptides were loaded onto a fused silica microcapillary C18 column (75 m ϫ 150 mm).
Liquid chromatography separation was conducted using linear gradient elution with 0.1% formic acid in H 2 O (solvent A) and formic acid in acetonitrile (solvent B). The elution program was as follows: a solvent composition of 97:3 (A:B) from 0 to 5 min; 60:40 from 5 to 72 min; 10:90 from 72 to 87 min; and 97:3 from 87 to 120 min. The flow rate was a constant 200 nl/min. The separated peptides were subsequently analyzed on a model LTQ linear ion-trap mass spectrometer (ThermoFinnigan, San Jose, CA). The electrospray voltage was set to 2.0 kV, and the threshold for switching from MS to MS/MS was 250. The normalized collision energy for MS/MS was 35% of the main R F amplitude, and the duration of activation was 30 ms. All spectra were acquired in data-dependent mode. Each MS scan was followed by MS/MS scans of the three most intense peaks from the full MS scan. The repeat count of peak for dynamic exclusion was 1, and its repeat duration was 30 s. The dynamic exclusion duration was set to 180 s, and the exclusion mass width was Ϯ1.5 Da. The list size of dynamic exclusion was 50.
Data Base Analysis-Data base searches were performed for all MS/MS spectra using an E. coli protein data base with the addition of the amino acid sequence of the Blh protein.
SEQUEST was used as the peptide-searching program, which included the dynamic modifications of oxidized methionine (ϩ16 Da) and carboxyamidomethylated cysteine (ϩ57 Da). SEQUEST criteria for peptide selection were based on the values of XCorr, which requires values greater than 1.8, 2.3, and 3.5 for ϩ1, ϩ2, and ϩ3 charge state peptides, respectively, with a ⌬Cn above 0.1. The criterion for protein selection was a consensus score above 10.1.
Western Blot Analysis-The soluble fraction of recombinant E. coli lysate was subjected to Western blot analysis. The fraction was separated by SDS-PAGE on 12% gels and then transferred to a polyvinylidene difluoride membrane using a Mini-Protean II transfer apparatus (Bio-Rad) according to the manufacturer's instructions. For immunodetection, the membranes were first blocked with 5% (w/v) skim milk in phosphate-buffered saline containing Tween 20 (PBST) and then incubated with His-tagged monoclonal antibody for 1 h. The membrane was visualized using a WEST-one TM Western blot detection system. The membranes were then placed between two overhead transparency films and exposed to Kodak film.
Determination of Molecular Weight-The subunit molecular weight of the Blh protein from uncultured marine bacterium 66A03 was examined by SDS-PAGE under denaturing conditions, using the proteins of a prestained ladder as reference proteins. All protein bands were stained with Coomassie Blue for visualization.
The molecular weight of the native enzyme was determined using Sephacryl S-300 HR gel filtration chromatography (Amersham Biosciences). The purified enzyme solution was applied to the chromatography and eluted with 100 mM Tricine-KOH buffer (pH 8.0) containing 0.1 M NaCl at a flow rate of 0.3 ml/min. The column was calibrated with aldolase (158 kDa), conalbumin (75 kDa), ovalbumin (43 kDa), carbonic anhydrase (29 kDa), and RNase A (13.7 kDa) as reference proteins, and the native enzyme was calculated by comparing with migration length of reference proteins.
Enzyme Assay-␤-Carotene of 200 M was dispersed in 2 ml of toluene containing 1.2% (v/v) Tween 80 to form detergent micelles, and the solution was homogenized at 10,000 rpm for 10 s using a homogenizer (IKA, Kuala Lumpur, Malaysia). The toluene was evaporated by N 2 gas at room temperature, and then 2 ml water was added. The resulting solution was used as the substrate solution. The substrate and enzyme solutions were mixed with the ratio of 1:3 (v/v) and then the mixture was used as the reaction solution. The reaction solution contained 50 M ␤-carotene, 0.3% (w/v) Tween 80, 125 mM sodium chloride, 10 M Fe 2 SO 4 , 5 mM Tris(2-carboxyethyl)phosphine hydrochloride, and 1.0% (w/v) 1-S-octyl-␤-D-thioglucopyranoside. The enzyme reaction was performed in 100 mM Tricine-KOH buffer (pH 8.0) containing 50 M ␤-carotene and 0.04 unit/ml enzyme at 40°C for 60 min. After incubation, the reaction was stopped by 3.7% (v/v) formaldehyde, and additional incubation was done at 40°C for 10 min (32). The recovery percentage of retinal was varied under the various conditions as temperature was affecting the recovery. One unit of enzyme activity was defined as the amount of enzyme required to produce 1 mole of retinal per min at 40°C and pH 8.0.
Effects of Metal Ions, pH, and Temperature on Enzyme Activity-The Blh protein was obtained after incubating at 20°C with 1 mM phenanthroline as an iron-chelating agent for 1 h and following by overnight dialysis at 4°C against 100 mM Tricine-KOH buffer (pH 8.0), and then the effects of various metal ions on its activity were investigated in the presence of 10 M Fe 2ϩ , Fe 3ϩ , Co 2ϩ , Ca 2ϩ , Cu 2ϩ , Mn 2ϩ , or Ba 2ϩ . The effect of Fe 2ϩ was evaluated with and without phenanthroline treatment. To examine the effect of pH on the enzyme activity, the pH was varied between 6.0 and 9.0 using 100 mM potassium phosphate buffer (pH 6.0 -7.0) and 100 mM Tricine-KOH buffer (pH 7.0 -9.0). The reactions were performed with 50 M ␤-carotene and 0.04 unit/ml enzyme at 40°C for 60 min. To investigate the effect of temperature on the enzyme activity, temperature was varied from 25 to 50°C. The reactions were performed in 100 mM Tricine-KOH buffer (pH 8.0) containing 50 M ␤-carotene and 0.04 unit/ml enzyme for 60 min.
The influence of temperature on enzyme stability for retinal formation was monitored in 100 mM Tricine-KOH buffer (pH 8.0) for 18 h at temperatures from 35 to 55°C. A sample was withdrawn at each time interval and was assayed for the residual relative activity in 100 mM Tricine-KOH buffer (pH 8.0) containing 50 M ␤-carotene and 0.04 unit/ml enzyme at 40°C for 60 min. The experimental data for thermal deactivation of enzyme were fitted to a first-order curve, and the half-lives of the enzyme were calculated using Sigma-plot software (version 9.0, 2004).
Determination of Kinetic Parameters for Various Carotenoid Substrates-␣-Carotene, ␤-carotene, ␥-carotene, ␤-cryptoxanthin, zeaxanthin, lutein, ␤-apo-4Ј-carotenal, ␤-apo-8Ј-carotenal, ␤-apo-12Ј-carotenal, and lycopene were used to determine the kinetic parameters of the enzyme. The reaction was performed in 100 mM Tricine-KOH buffer (pH 8.0) at 40°C for 30 min using various concentrations of substrates (5- 18 O 2 on ice for 5 min. The enzyme reaction was performed in 100 mM Tricine-KOH buffer (pH 8.0) containing 50 M ␤-carotene and 0.04 unit/ml enzyme at 40°C for 60 min. At the end of the incubation, an equal volume of acetonitrile was added to the reaction solution. The resulting solution was centrifuged at 10,000 ϫ g for 10 min at 4°C, and the supernatant was evaporated. The samples were resuspended in acetonitrile and analyzed by high resolution liquid chromatography-mass spectrometry (HR-MS).
Computational Analysis-Theoretical weights and pI values for proteins were calculated using the Compute pI/Mw tool at the ExPASy web site. Prediction of transmembrane helices in the protein was performed using the TMHMM program (33).
Molecular Modeling of the Blh Protein-The Blh protein had no homologous sequences. Thus, we predicted the secondary structure of the Blh protein using the protein structure prediction server (PSIPRED) (34) and compared it with other secondary structures in the Protein Data Bank. As a result, we selected 10 secondary structure candidates from among those analyzed and averaged the candidates. Based on the selected secondary structure, a simulated backbone was constructed, and side chains were added using the side chain prediction program SCWRL (35). The three-dimensional structure of the Blh protein from the uncultured marine bacterium 66A03 was generated using the Accelrys Discovery Studio Modeler (Accelrys, San Diego) (36). The generated structure was checked by PROCHCK (37), and then structure minimization was conducted using Chemistry at Harvard Molecular Mechanics (38). After energy minimization, molecular dynamics modeling was performed at 300 K, 1 atm for 500 ps with 1 fs each step. All simulation experiments were carried out on an HP XW6200 Workstation with dual Intel Xeon 3.2-GHz processors.
␤-Carotene, ␤-cryptoxanthin, and lutein were docked in the Blh models using the Surflex X docking program (Tripos, St. Louis, MO). Each docking run consisted of 100 independent docks with 1,000 iteration cycles. A random start was used to generate the substrate position within the docking box. The substrate orientation giving the lowest interaction energy was chosen for additional rounds of docking.
Analytical Methods-High pressure liquid chromatography (HPLC) analyses of carotenoids and retinoids were performed based on a previously reported method (10). The same volume of acetonitrile was added to the reaction solution, and the resultant solution was mixed and kept on ice for 5 min. After centrifugation at 10,000 ϫ g for 10 min at 4°C, substrates and products of the supernatant were analyzed by an HPLC system (Agilent 1200 series, Santa Clara, CA) equipped with a UV detector. The substrates ␤-carotene and lycopene were detected at 460 and 445 nm, respectively, and lutein and zeaxanthin were determined at 450 nm using a Zorbaxsil column (250 ϫ 4.6 mm, Agilent). The column was eluted with the mixture of hexane and tertbutyl methyl ether with 97:3 (v/v) as the mobile phase with a flow rate of 1 ml/min. The products (3R)-3-hydroxy-retinal, retinal, and ␣-retinal were detected at 370 nm with retention times of 1.3, 1.8, and 2.2 min, respectively, and the product acycloretinal was detected at 400 nm using an YMC-ODS A column (50 ϫ 2.0 mm, YMC, Kyoto, Japan). The column was eluted with a 90:10 (v/v) mixture of acetonitrile and water as the mobile phase at a flow rate of 0.4 ml/min. The retinal isomers 13-cisretinal, 9-cis-retinal, and all-transretinal were detected at 370 nm using an YMC-ODS A column (250 ϫ 4.6 mm, YMC) with retention times of 26.1, 26.3, and 27.3 min, respectively. The column was eluted with 80% acetonitrile as the mobile phase at a flow rate of 1.0 ml/min.
The HR-MS data were obtained using a JMS-SX102A spectrometer (Jeol, Tokyo, Japan) operated at an accelerating voltage of 10 kV. The electron impact ionization mass spectra were collected in the positive ion mode. The HR-MS was performed under the acquisition conditions as follows: ion source temperature 230°C, ionization energy 70 eV, and ionization current 300 A.

RESULTS
DNA Sequence of the Codon-optimized blh Gene-Codons of the blh gene encoding the ␤-carotene-cleaving enzyme from the uncultured marine bacterium 66A03 were optimized, and the entire gene was synthesized. Among the 828 bp of the blh gene, including a stop codon, 213 bp (25.8%) were changed by codon optimization (supplemental Fig. 1). The synthesized gene was cloned into the pET-24a(ϩ) vector and expressed in E. coli. The expressed Blh protein consisted of 275 amino acids. An alignment of the Blh protein with other Blh and Brp-like proteins, including membrane spanning domains, is shown in Fig. 1. The proposed metal-binding residues, His-21, His-78, His-188, and His-192, based on molecular modeling were absolutely conserved across all Blh and Brp-like proteins.
Expression, Molecular Weight, and Identification of the Blh Protein-The crude extract of soluble protein obtained from harvested cells was purified by His-Trap HP affinity chromatography. The expressed enzyme was purified, without detergent, as a single band in SDS-PAGE, with a final purification of 7.3-fold, a yield of 32%, and a specific activity of 45 nmol mg Ϫ1 min Ϫ1 . The molecular weight of the purified protein as analyzed by SDS-PAGE was about 32 kDa ( Fig. 2A). The protein was purified again with detergent, including 0.05% 1-S-octyl-␤-D-thioglucopyranoside, 0.5% 1-S-octyl-␤-D-thioglucopyranoside, 0.5% Tween 80, 6 M urea, 1% Triton X-100, or 1% cocoamidopropyl hydroxysultaine. This further purification, however, did not change the purification yield. The immunoblot was performed with a His-tagged monoclonal antibody against the His-tagged Blh protein.
A band obtained from the soluble fraction of the cell extract confirmed the isolation of a highly purified recombinant protein. The Blh protein was expressed as a soluble form in E. coli, and the protein was purified without detergent even though it is a member of the bacteriorhodopsin family. Its soluble form may result from the codon optimization, which is known to increase the expression and solubility of enzymes (39).
The molecular weight of native enzyme, based on the weights of reference proteins, was estimated using gel filtration chromatography. The native Blh protein had a mass of 64 kDa as a dimer of 32-kDa subunits (Fig. 2B). A chromatogram for the gel filtration of the Blh protein confirms its purity and molecular weight (Fig. 2C).
The Blh protein on a SDS-PAGE was cut out and subjected to trypsinization. The digested peptides were recovered and characterized on a nano liquid chromatography-MS/MS to obtain amino acid sequence data. The following high confidence peptide sequences were obtained: YNIAFELIG, SRRHFS-FVWKQL, and FIGLPHGALD (supplemental Fig. 2). A comparison between these data and the Blh sequence showed that all three peptides were present in the sequence (Fig. 1).
Determination of the Retinal Product of the Blh Protein-The enzymatic conversion of ␤-carotene into retinal was examined by varying the reaction time and concentrations of enzyme and substrate, based on a standard reaction performed in 100 mM Tricine-KOH buffer (pH 8.0) containing 50 M ␤-carotene and 0.04 unit/ml enzyme at 40°C for 60 min (Fig. 3, A-C). The products formed at different reaction times and concentrations of enzyme and substrate were analyzed by an HPLC system using an YMC-ODS A column and showed the same retention time as an all-trans-retinal standard. However, other retinal isomers including 13-cis-retinal and 9-cis-retinal showed different retention times, indicating the enzymatic product of ␤-carotene is all-trans-retinal. With increasing reaction time from 10 to 60 min, the substrate ␤-carotene decreased whereas the product all-transretinal increased (Fig. 3D).
The enzyme cleaved ␤-carotene at its central double bond (15,15Ј) to yield two molecules of retinal. The formation of ret- inal as a reaction product increased linearly up to ϳ0.2 unit/ml enzyme, 20 M ␤-carotene, and 30 min of reaction time, respectively. However, above these parameters, the rates of product formation were not linear; increasing the concentrations of enzyme and substrate and the reaction time subsequently developed sigmoidal curves. We performed enzyme assays in the absence of substrate and in the presence of heat-inactivated Blh as controls, which exhibited no activity.
Effects of Metal Ions, pH, and Temperature on the Activity of the Blh Protein-The purified enzyme obtained after the removal of metal ions by phenanthroline exhibited 2% activity relative to that without phenanthroline treatment. Among the metal ions tested, Fe 2ϩ had the strongest effect on the activity of the Blh protein, whereas Ba 2ϩ had the least effective. The effects of other ions decreased in the order of Fe 3ϩ , Co 2ϩ , Ca 2ϩ , Cu 2ϩ , and Mn 2ϩ (Fig. 4A). The addition of 10 M Fe 2ϩ after the removal of all metal ions induced a 54% recovery of the active form of the enzyme. The optimal concentration of Fe 2ϩ was 10 M, regardless of phenanthroline treatment (Fig. 4, B and C). Thus, all subsequent experiments were performed in the presence of 10 M Fe 2ϩ .
The enzymatic conversion of ␤-carotene into all-trans-retinal was examined at pH values ranging from 6.0 to 9.0. The maximum enzyme activity was observed at pH 8.0 (Fig. 5A). The effect of temperature on the enzyme activity is shown in Fig. 5B. The maximum activity was recorded at 40°C. Above this temperature, the enzyme activity decreased significantly, exhibiting 66% of the maximum activity at 50°C. Below 40°C, the enzyme activity decreased with decreasing temperature, exhibiting 15% of the maximum activity at 25°C.
The thermostability of the Blh protein was measured at five incubation temperatures. The activity of the enzyme was very stable at below 45°C but significantly decreased at above 50°C with increasing reaction time (Fig. 5C). The enzyme followed the first-order kinetics of thermal inactivation, and the halflives of the enzyme at 35, 40, 45, 50, and 55°C were 17.6, 15.0, 12.5, 6.1, and 1.5 h, respectively.

Characterization of Bacterial ␤-Carotene 15,15-Dioxygenase
Origin of the Oxygen Atom of Retinal in ␤-Carotene Cleavage by the Blh Protein-The molecular weight of retinal as a product was determined after the enzyme was incubated with ␤-carotene in the presence of H 2 18 O or 18 O 2 . For labeling experiments with H 2 18 O, freeze-dried enzyme and ␤-carotene were suspended in the H 2 18 O solution. The mass spectrum of retinal formed under these conditions was the same as that using H 2 O, and the molecular weight of retinal was determined as 284 m/z (Fig. 6A), indicating that the oxygen atom of retinal did not originate from water. However, after the enzyme reaction was performed in an 18 O 2 atmosphere, the formed retinal was labeled with the 18 O atom, and the molecular mass was 286 m/z (Fig. 6B). The fragment molecular masses of labeled retinal containing isotope 18 O such as 213 and 241 m/z were 2 m/z greater than those of unlabeled retinal containing normal 18O such as 211 and 239 m/z. Thus, the oxygen atom of retinal as a reaction product of the Blh protein originated from molecular oxygen rather than from water.

DISCUSSION
Although most recent critical debates about ␤-carotenecleaving enzymes have tended to center around mammalian ␤-carotene 15,15Ј-monooxygenases, we became interested in ␤-carotene-cleaving enzymes from other sources, especially bacteria. Hence, we used codon optimization to synthesize the blh gene from the uncultured marine bacterium 66A03 for The experiments for effects of pH, temperature, and thermostability were compensated with controls that were performed with the same reactions without enzyme. A, two different buffers for pH experiment were used as follows: 100 mM potassium phosphate buffer for pH 6.0 -7.0 (E) and 100 mM Tricine-KOH buffer for pH 7.0 -9.0 (F). The reactions were performed with 50 M ␤-carotene and 0.04 unit/ml enzyme at 40°C for 60 min. B, reactions were performed 100 mM Tricine-KOH buffer (pH 8.0) containing 50 M ␤-carotene and 0.04 unit/ml enzyme for 60 min in temperature range of 25-50°C. C, thermostability of the Blh protein for retinal production was measured at 35 (F), 40 (Ⅺ), 45 (OE), 50 (E), and 55°C (f). A sample was withdrawn at each time interval, and the residual relative activity was measured after the reaction was performed with 100 mM Tricine-KOH buffer (pH 8.0) containing 50 M ␤-carotene and 0.04 unit/ml enzyme at 40°C for 60 min. expression in E. coli. The expressed enzyme converts ␤-carotene into all-trans-retinal. Characterization of the bacterial Blh protein as a new member of the ␤-carotene cleavage enzymes is a worthwhile undertaking.
The Blh protein is a putative membrane protein closely related to bacteriorhodopsin (40). Bacteriorhodopsin is a membrane protein containing seven transmembrane ␣-helices and a covalently bound molecule of retinal. The Blh protein is predicted by the TMHMM program to have seven transmembrane segments, whereas human ␤-carotene 15,15Ј-monooxygenase has no transmembrane segments. These results suggest that the Blh protein is a membrane protein The theoretical pI values of the Blh proteins from marine bacteria (8.89 -9.56) were much higher than those of the mammalian ␤-carotene 15,15Ј-monooxygenases (5.49 -6.30), indicating that the two proteins are different (Table 2). There were 48 -56% hydrophobic amino acids in the bacterial Blh proteins and 33-34% in mammalian ␤-carotene 15,15Ј-monooxygenases.

TABLE 1 Kinetic parameters of the Blh protein for various carotenoids as substrates
The reaction was performed in 100 mM Tricine-KOH buffer (pH 8.0) at 40°C for 30 min using the amounts of substrates ranging from 5 to 800 M. Data represent the mean of three experiments, and Ϯ values represent S.D. ND means kinetic parameters were not detected by the analytical methods used in this study.
The subunit molecular weight of the Blh protein was calculated as 32 kDa based on the 281-residue amino acid structure plus a hexahistidine tag at the carboxyl terminus. The native protein has a total molecular mass of about 64 kDa as a dimer (Fig. 2). ␤-Carotene 15,15Ј-monooxygenases from humans, mice, chickens, and rats consisted of 547, 566, 526, and 566 amino acid residues, respectively, and their molecular masses were ϳ60 -64 kDa ( Table 3). The molecular mass of the native human ␤-carotene 15,15Ј-monooxygenase as a tetramer is 250 kDa (41).
Mammalian ␤-carotene 15,15Ј-monooxygenases are reported to be Fe 2ϩ -dependent protoporphyrin (non-heme iron) enzymes and are inhibited by iron-chelating agents (4,42,43). Phenanthroline behaves as an iron-chelating agent and strongly inhibited the conversion of ␤-carotene into retinal by the Blh protein. Furthermore, among the metal ions tested, Fe 2ϩ had the strongest effect on the activity of the Blh protein. Fe 2ϩ is bound by four conserved histidine residues in the active sites of mammalian ␤-carotene monooxygenase and apocarotenoid oxygenase (43,44) and in the molecular model of the Blh protein (Fig. 7A). In the Blh protein, Fe 2ϩ is coordinated with His-21, His-78, His-188, and His-192, which are absolutely conserved across all Blh and Brp-like proteins (Fig. 1). The four conserved histidine residues were replaced with alanine to produce alanine-substituted mutants. Enzyme activities were determined for the mutants and were compared with that of wild-type enzyme. The relative activities of the H21A, H78A, H188A, and H192A mutants were 3, 0, 0, and 5%, respectively. Thus, we can propose that the four histidine residues in the Blh protein are metal-binding residues. The enzyme activity in the presence of 10 M Fe 2ϩ before dialysis was almost the same as after dialysis. These results imply that Fe 2ϩ is tightly bound to the Blh protein.
To interpret the substrate specificity of the Blh protein on carotenoid substrates, a simulated backbone of the enzyme structure was constructed, and side chains were added based on averaged similar secondary structures in the protein data base. Simulated molecular docking was performed with ␤-carotene, ␤-cryptoxanthin, and lutein substrates using the Surflex X docking program. When ␤-carotene was bound in the active site, the para position of one ␤-ionone ring of ␤-carotene, located in the interior of the active site, and the para position of the other unhydroxylated ␤-ionone ring interacted with the oxygen atom of the carbonyl group of Thr-179, at a distance of 3.01 Å (Fig. 7B). When ␤-cryptoxanthin was bound in the active site, the distance between the hydroxyl group of the (3R)-3hydroxy-␤-ionone ring of ␤-cryptoxanthin and the oxygen atom of the carbonyl group of Thr-179 decreased to 2.31 Å (Fig.  7C). The shorter distance may explain the observed higher affinity of the Blh protein for ␤-cryptoxanthin than for ␤-carotene. When lutein (or zeaxanthin) was bound in the active site, two hydroxyl groups of two (3R)-3-hydroxy ␤-ionone rings of lutein (or zeaxanthin) were noneffective in anchoring to the active site because the distance to the oxygen atom of Thr-179 was too long (Fig. 7D). As a result, the Blh protein exhibited no activity for lutein or zeaxanthin. The Blh protein showed activity for ␤-apo-4Ј-carotenal but not for ␤-apo-8Ј-carotenal. In the molecular modeling studies, ␤-apo-4Ј-carotenal was large enough to interact with Thr-179, but ␤-apo-8Јcarotenal was too small to interact (data not shown). However, it is necessary to obtain the actual structure of the enzyme complexed with these substrates to provide further evidence for this conclusion. To confirm our molecular modeling and structural analyses, determination of the crystal structure of the Blh protein and mutational analysis of the active site residues must be performed.
Carotenoid oxygenases can be divided into mono-and dioxygenases based on two different reaction mechanisms (Fig.  8). The oxygen atom of the product from the monooxygenase is provided by molecular oxygen and water via an epoxide intermediate, whereas that from the dioxygenase is provided by molecular oxygen rather than from water via a dioxetane inter-mediate (51,52). The oxygenation mechanism of chicken ␤-carotene 15,15Ј-oxygenase was investigated with isotopic molecular oxygen and water. The oxygen atom in the terminal aldehyde group of retinal as a product was provided by molecular oxygen and water (51), indicating that the enzyme catalyzes the oxidative cleavage using a monooxygenase rather than a dioxygenase mechanism. The other mammalian ␤-carotene 15,15Ј-oxygenases were postulated to use the same mechanism. However, some controversies on the reaction mechanism was suggested because oxygen between retinal and water was exchanged within 5% (51). In contrast, the oxygen atom of the retinal formed by the reaction of the Blh protein originated from molecular oxygen rather than from water without oxygen exchange between retinal and water, indicating that the Blh protein used a dioxygenase mechanism unlike the mammalian ␤-carotene 15,15Ј-monooxygenases. The Blh protein cleaved ␤-carotenoid substrates at its central double bond (15,15Ј) and showed the highest activity for ␤-carotene among the substrates tested. Thus, the Blh protein can be identified as a ␤-carotene 15,15Ј-dioxygenase.
In the molecular model, the active site of the Blh protein exhibited coordination of Fe 2ϩ with His-21, His-78, His-188, and His-192 (Fig. 7A). According to the cleaving mechanism of apocarotenoid oxygenase (52), the O 2 molecule in the active site of the Blh protein binds to the coordination shell of Fe 2ϩ in a side-on fashion. The side-on complex of Fe-O 2 then attacks and cleaves between the C-15 and C-15Ј atoms when bound to the Blh protein.
Most enzymes that catalyze the same reaction sequence are structurally similar. However, structurally and mechanistically unrelated enzymes that catalyze the same biochemical reactions do occur. In many cases, one of these analogous enzymes is found in the bacteria and the other in eukaryotes (53). Although the bacterial Blh protein and the mammalian ␤-carotene 15,15Ј-monooxygenases could both convert ␤-carotene into retinal, the properties of the bacterial Blh protein such as DNA and amino acid sequences, molecular weight, form of association, reaction mechanism, kinetic properties, and substrate specificity were different from the mammalian ␤-carotene 15,15Ј-monooxygenases. Thus, the Blh protein is a bacterial analog to the mammalian ␤-carotene 15,15Ј-monooxygenases. This is the first report of in vitro characterization of a bacterial ␤-carotene-cleaving enzyme. This study contributes to the understanding of bacterial ␤-carotene-cleaving enzymes and provides a stepping stone for further studies. Moreover, this study will be useful in evolutionary studies investigating the relationships between bacterial and mammalian enzymes and industrial applications of retinal biosynthesis.