The human enzyme that converts dietary provitamin A carotenoids to vitamin A is a dioxygenase

Background: The human enzyme β-carotene 15-15′-oxygenase (BCO1) has been thought to be a monooxygenase. Results: Incubation of BCO1 and β-carotene in H 218 O- 16 O 2 or H 216 O- 18 O 2 medium yields 2 retinals both of which contain oxygen atoms originating solely from O 2 gas. Conclusion: BCO1 is a dioxygenase. Significance: It is important to clearly establish an enzyme’s reaction mechanism especially when the name reflects the mechanism. ABSTRACT β-Carotene 15-15′-oxygenase (BCO1) catalyzes the oxidative cleavage of dietary provitamin A carotenoids to retinal (vitamin A aldehyde). Aldehydes readily exchange their carbonyl oxygen with water, making oxygen labeling experiments challenging. BCO1 has been thought to be a monooxygenase, incorporating oxygen from O 2 and H 2 O into its cleavage products. This was based on a study that used conditions that favored oxygen exchange with water. We incubated

O medium for 15 minutes at 37°C, and the relative amounts of 18 O-retinal and 16 O-retinal were measured by liquid chromatography-tandem mass spectrometry (LC-MS/MS). At least 79% of the retinal produced by the reaction has the same oxygen isotope as the O 2 gas used. Together with the data from 18 O-retinal-H 2 16 O and 16 O-retinal-H 2 18 O incubations to account for non-enzymatic oxygen exchange, our results show that BCO1 incorporates only oxygen from O 2 into retinal. Thus, BCO1 is a dioxygenase.
Vitamin A deficiency is the most common vitamin deficiency in the world and affects an estimated 190 million preschool-age children and 19.1 million pregnant women worldwide (1). In areas of endemic vitamin A deficiency, people obtain vitamin A almost exclusively as provitamin A carotenoids found in foods of plant origin (2). Provitamin A carotenoids are enzymatically converted to retinal (vitamin A aldehyde) ( Figure  1A) by the enzyme β-carotene 15-15′-oxygenase (BCO1) (3). Hence, understanding the mechanism and regulation of this enzyme is important.
The reaction mechanism, and consequently, the nomenclature of BCO1 and other carotenoid cleavage oxygenases (CCO's) have been controversial (4)(5)(6). The first report of a CCO was made in 1965 by Goodman and Huang (7), who showed that β-carotene was converted to retinal by cell-free rat intestinal homogenates in the presence of O 2 . The following year, the same group then showed using 3 H labels that the hydrogens of the 15-15′ double bond of β-carotene (the site of oxidative cleavage) are retained during the enzymatic oxidation reaction, and proposed that the reaction most likely has a dioxygenase mechanism (8). However, the label "dioxygenase" should only be used when oxygen labeling experiments have clearly established that only oxygen from O 2 is incorporated by the enzyme into its oxidative cleavage products. BCO1 was given the Enzyme Commission (EC) number 1. 13.11.21 in 1972, designating a dioxygenase (9), 29 years before the first report of an oxygen labeling experiment. A monooxygenase mechanism was proposed for BCO1 in 2001 (10). In that study, α-carotene, purified chicken BCO1 and horse liver alcohol dehydrogenase (HLADH) were incubated in an 85% 17 O 2 -95% H 2 18 O environment. HLADH was used to form retinols from the aldehydes, which readily exchange their carbonyl oxygen with water (11). The resulting products (retinol and α-retinol) were purified by high-performance liquid chromatography and silylated. Using gas chromatography-mass spectrometry (GC-MS), the authors found virtually equal enrichment of 17 O and 18 O in both silylated retinols, suggesting a monooxygenase mechanism (Figure 1 B). However, it is possible that the long reaction time (7.5 hours) and extensive processing favored oxygen exchange between the initial aldehyde products and the aqueous medium. Also, the HLADH reaction is reversible, and the enzyme displays dismutase activity (interconverting the aldehyde into alcohol and carboxylic acid) (6,12). This means that the aldehydes were never completely eliminated during the 7.5 hour incubation, and a significant amount of oxygen exchange with water may have occurred. Despite the inconclusiveness of this study, the enzyme's EC number was changed to 1.14.99.36, classifying it as a monooxygenase (13), and subsequent literature has referred to the animal orthologs of the enzyme as β-carotene 15-15′-monooxygenase (BCMO1). Indeed, the National Center for Biotechnology Information named the gene BCMO1 (14).
To elucidate the reaction mechanism of human BCO1, we conducted multiple oxygen labeling experiments with minimal reaction and processing times to minimize oxygen exchange between retinal and water. Our results demonstrate that BCO1 is not a monooxygenase, but a dioxygenase.

Synthesis of
Freeze-drying of purified recombinant human BCO1-Purified recombinant human BCO1 was prepared according to our previously published method (16). The purified enzyme preparation catalyzed the oxidative cleavage of β-carotene with a V max = 197.2 nmol retinal/mg BCO1 × h, K m = 17.2 μm and catalytic efficiency k cat /K m = 6098 M −1 min −1 . Ten μg of purified recombinant human BCO1 and 40 μL of 5x reaction buffer (500 mM Tricine-KOH, pH 8.0 at 37°C, 2.5 mM dithiothreitol, 20 mM sodium cholate, 75 mM nicotinamide) (16) were combined in a 10-mL amber headspace vial, and the vial was capped and flash-frozen in liquid nitrogen. The headspace vials were stored in dry ice for 30 minutes during transport to the freeze-dryer. The caps of the headspace vials were then fitted with individual syringe needles for venting, and the vials were placed in the jar of the manifold freeze dryer (Labconco). Freeze-drying was done for 16 hours at 0.14 mBar. The syringe needles were removed, and the headspace vials were stored at -80°C until use. Each vial of freeze-dried enzyme produces about 60 pmol of retinal from 4 nmol of βcarotene with the in vitro BCO1 activity assay system described in the following section.
In vitro BCO1 activity assay in 16 18 O-The in vitro enzyme assay using purified recombinant human BCO1 was based on our previously published method (16). The freeze dried enzymereaction buffer mixture in the headspace vial (described in previous section) was dissolved in H 2 18 O to a final volume of 160 μL and placed in a 37°C shaking water bath. The reaction was initiated by adding 40 μL of β-carotene substrate solution (containing 4 nmol β-carotene, 0.3 μL Tween-40 and 20 nmol α-tocopherol) prepared in H 2 18 O. The reaction was allowed to proceed in the water bath with gentle shaking and the vial exposed to air (which contains oxygen as 99.8% 16 O 2 (17)) for 15 minutes. The quenching with 37% formaldehyde in the original method had to be omitted since the latter contains H 2 16 O. Instead, the reactions were quenched with 300 μL of acetonitrile, and the lipophilic compounds were extracted with 3x1 mL of hexanes under red lights. The combined hexane extracts were dried under N 2 , re-dissolved in 100 μL 3:1 (v/v) acetonitrile-H 2 18 O, filtered through a 0.22 μm syringe-driven filter, and injected into the HPLC. The whole process from the start of the reaction to the elution of the retinal peak in the HPLC takes about 50-60 minutes.
In vitro BCO1 activity assay in 18 O 2 -H 2 16 O-The enzyme-reaction buffer solution (10 μg of purified recombinant human BCO1, 40 μL of 5x reaction buffer and water to a total volume of 160 μL) was placed in a headspace vial and degassed by exposure to water aspirator vacuum for 2 minutes. The headspace vial was purged with nitrogen gas, then connected to the 18 O 2 gas cylinder and placed in a 37°C water bath. Forty μL of β-carotene substrate solution, prepared in degassed water (H 2 16 O), was then injected into the vial using a syringe. The reaction was allowed to proceed in the water bath with gentle shaking for 15 minutes. The reaction was quenched by injecting 300 μL of acetonitrile into the vial before the 18 O 2 gas flow was turned off. The reaction mixture was then extracted and processed as in the previous section, except that the extract residue was re-dissolved in acetonitrile-H 2 16 O.  18 Oretinal (m/z=287.226) and found to optimize at a collision energy of 7.5 eV. Source and CID gas was high purity (>98%) nitrogen. Calibration was performed using ESI-L tuning mix (Agilent Technologies G1969-85000) and within-run reference compound was hexakis (1H, 1H, 3Htetrafluoropropoxy) phosphazine, m/z 922.010 (Agilent Technologies HP-0921).

Quantification of retinal oxygen isotopologues-
The fragmentation patterns of 18 O-retinal and 16 Oretinal are virtually the same (Figure 3). The parent retinals were not used for quantification to minimize errors arising from other naturally occurring isobaric species, which constitute about 2.2% based on the natural abundance of 13 C (17). MS/MS was used to discriminate between the retinal analytes from these isobaric species, which will give different fragmentation patterns.

RESULTS
For the BCO1-β-carotene reaction in 16  O medium suggests that BCO1 is a dioxygenase ( Figure 1B). If the enzyme is a dioxygenase, then theoretically, it should produce only 16 O-retinal, and the 18 O-retinal we observed was due to oxygen exchange with water. To verify this, we incubated 60 pmol of 16 O-retinal with BCO1 in H 2 18 O under the same conditions. The % 18 O-retinal formed was similar (5-13%) to that produced in the reaction of BCO1 and β-carotene ( Figure 1C). This confirms that the 18 O-retinal we were detecting was coming from the oxygen exchange of retinal with water, and not from the enzyme incorporating oxygen from water during the oxidative cleavage reaction.
We then conducted the BCO1-β-carotene reaction in 18 O 2 -H 2 16 O medium. Consistent with our previous experiments, the majority of the retinal product obtained contains the same oxygen isotope as that of O 2 (79-85% 18 O-retinal). As in the previous section, this range reflects what we obtained from experiments done on different days. A sample LC-MS chromatogram is shown in Figure 2B, and the MS/MS traces for m/z= 285.218 and 287.226 (corresponding to 16 O-retinal and 18 O-retinal, respectively) are shown in Figure  3. To verify that the 16 O-retinal (15-21%) we observed was due to oxygen exchange with water, we also incubated 18 O-retinal (91% 18 O-retinal) with BCO1 in H 2 16 O under the same conditions. We observed 67-84% 18 O-retinal, corresponding to a 7-24% net exchange ( Figure 1C). Consistent with the previous section, these values strongly suggest that BCO1 reacts with β-carotene in an 18 O 2 -H 2 16 O to form only 18 O-retinal, and the small relative amount of 16 O-retinal is due to oxygen exchange with water.
We also performed the BCO1β-carotene incubation in 16  O experiments strongly suggest that BCO1 incorporates only oxygens from O 2 into retinal formed from the oxidative cleavage of β-carotene, and the minor amount of retinal with the same oxygen isotope as water is formed by non-enzymatic oxygen exchange. Thus, BCO1 is a dioxygenase and not a monooxygenase as had been previously thought.
If the parent retinals are used for quantification, the values differ by only 0-6% from the MS/MS calculation (Table 1), and the data still lead to the same conclusion that BCO1 incorporates only oxygen from O 2 into retinal formed from oxidative cleavage of β-carotene.

DISCUSSION
At this point, there is a very limited amount of literature on other CCO's with which to compare our results. Most of the functionally characterized CCO's are from plants, and these enzymes have been called "dioxygenases" despite the lack of conclusive oxygen labeling experiments (6,18). This error can be traced back to the lignostilbene "dioxygenases." As of 1997, these enzymes were called as such even though no oxygen labeling experiments were carried out (6,(19)(20)(21)(22)(23). At best, these studies showed that these enzymes require O 2 . This error in naming was propagated into the CCO's in 1997, when the first CCO to be cloned and characterized, maize Viviparous 14, was called a dioxygenase based on its sequence similarity to lignostilbene "dioxygenase" and not on oxygen labeling experiments (24)(25). Even if the lignostilbene oxygenases were truly established to be dioxygenases back then, a sequence similarity is not necessarily a substitute for oxygen labeling experiments. Interestingly, the first report of an oxygen labeling experiment for a stilbene oxygenase (which was also identified because of sequence similarity to the plant CCO's) showed a monooxygenase reaction mechanism (26).
Of the more than 200 putative CCO's to be found in sequence databases (5), there are only four other oxygen labeling experiments done apart from the aforementioned 2001 BCO1 study. The oxygen labeling experiments on water-stressed leaves of Xanthium strumarium in 1984 (27) looked at only one cleavage product, and the Arabidopsis thaliana study in 2006 (24) was deemed inconclusive because of the failure to show a consistent labeling pattern for the two cleavage products (Kloer and Schulz give a detailed critique of these two studies (5)). An oxygen labeling experiment was done with Microcystis PCC 7806 cells, which generate βcyclocitral and crocetindial from oxidative cleavage of β-carotene (28 O incubations were contradictory, and the authors acknowledge that the longer processing time for crocetindial may have favored oxygen exchange. Another oxygen labeling study done with a purified recombinant marine bacterial CCO that also cleaves β-carotene to retinal also shows a dioxygenase mechanism (29), consistent with our results.
Unlike other enzyme names such as "isomerase" or "lyase", the names "dioxygenase" and "monooxygenase" both indicate a specific reaction mechanism. Thus, the mechanism should be elucidated first before the name of an oxygenase is assigned. For oxygenases that yield aldehydes, oxygen exchange with water should be minimized and accounted for. BCO1 was called a dioxygenase in 1972 without an oxygen labeling experiment, and a monooxygenase in 2001 despite an inconclusive study. Our results demonstrate that BCO1 is a dioxygenase.