The Desymmetrization of Bicyclic β-Diketones by an Enzymatic Retro-Claisen Reaction

The enzyme 6-oxocamphor hydrolase, which catalyzes the desymmetrization of 6-oxocamphor to yield (2R,4S)-α-campholinic acid, has been purified with a factor of 35.7 from a wild type strain ofRhodococcus sp. NCIMB 9784 grown on (1R)-(+)-camphor as the sole carbon source. The enzyme has a subunit molecular mass of 28,488 Da by electrospray mass spectrometry and a native molecular mass of ∼83,000 Da indicating that the active protein is trimeric. The specific activity was determined to be 357.5 units mg− 1, and theK m was determined to be 0.05 mm for the natural substrate. The N-terminal amino acid sequence was obtained from the purified protein, and using this information, the gene encoding the enzyme was cloned. The translation of the gene was found to bear significant homology to the crotonase superfamily of enzymes. The gene is closely associated with an open reading frame encoding a ferredoxin reductase that may be involved in the initial step in the biodegradation of camphor. A mechanism for 6-oxocamphor hydrolase based on sequence homology and the known mechanism of the crotonase enzymes is proposed.

The desymmetrization of prochiral substrates in organic synthesis remains a powerful technique for the generation of chiral intermediates with, in principle, 100% yield with absolute optical purity. In addition to chemical processes that have been reviewed recently (1), enzyme-catalyzed methods have assumed an important role in desymmetrization (2) owing to the well documented advantages of biocatalysis in general. Such methods have usually exploited the regioselectivity of a hydrolytic enzyme, such as a lipase or nitrilase, to effect transformation of one of two identical functions in a molecule (3,34). Enzymatic desymmetrizations have for the most part been performed using carbon-heteroatom bond hydrolases of this type, although Taschner and Black (4) were successful in desymmetrizing a series of prochiral cyclic ketones using an enzymatic Baeyer-Villiger reaction.
The metabolism of (1R)-(ϩ)-camphor by Corynebacterium sp. T1 (now taxonomically reclassified and deposited as Rhodococ-cus sp. NCIMB 9784) was described in the 1960s by Gunsalus and co-workers (5). The pathway is distinct from that found in Pseudomonas putida (ATCC 17453; NCIMB 10007) in that initial hydroxylation occurs in the 6-endo position of the camphor skeleton (Fig. 1). 6-endo-Hydroxy camphor 2 is oxidized to a symmetrical diketone 3, which is then cleaved by a retro-Claisen reaction to yield a keto acid 4, which was reported by Gunsalus and co-workers (5) to have a negative optical rotation. This last enzymatic reaction is in fact a desymmetrization, and is interesting in that it apparently proceeds by an unusual enzyme-catalyzed retro-Claisen reaction. Enzymes that hydrolyze 1,3-diketo functionality, ␤-diketone hydrolases (6) are rare, and only three reports describe their purification to homogeneity and characterization (7,8,15,35). In the first two cases, the substrate specificity of the enzymes extends only to 3,5-diketo acids. This specificity has been recently explained by elucidation of the x-ray crystal structure of fumarylacetoacetate hydrolase (9), which shows a dependence on a divalent calcium ion for substrate recognition and enolate stabilization.
In view of the ongoing interest in the biocatalytic generation of chiral intermediates and desymmetrization in particular, we were interested in studying the potential of the enzyme, which we have named 6-oxocamphor hydrolase, for the desymmetrization of other cyclic and bicyclic ␤-diketones. In this paper, we present the purification and initial characterization of the enzyme from the wild type strain of Rhodococcus and the cloning of the gene encoding its activity. Our findings suggest that the enzyme does not share homology with other ␤-diketone hydrolases but rather is related to a different class of enzyme, the crotonase superfamily.

EXPERIMENTAL PROCEDURES
Enzymes and Chemicals-Restriction endonucleases were from New England Biolabs. T4 DNA ligase, T4 polynucleotide kinase, and calf intestinal alkaline phosphatase were from Roche Molecular Biochemicals. RNase was purchased from Sigma-Adrich (Poole, United Kingdom). Protein and DNA size markers were obtained from (Amersham, United Kingdom) Pharmacia Biotech. [␥-32 P]ATP (3000 Ci/mmol) was purchased from Amersham Pharmacia Biotech. Isopropyl-␤-D-thiogalactopyranoside was obtained from U. S. Biochemical Corp. All other chemicals were purchased from Sigma-Aldrich. 6-Oxocamphor 3 was synthesized by pyridinium chlorochromate oxidation (10) of 6-endohydroxy camphor derived from ethyl acetate extractions of the mother liquor of fermentations of Rhodococcus sp. NCIMB 9784 grown on (1R)-(ϩ)-camphor performed as described below.
Maintenance and Growth of Microorganisms-Escherichia coli XL1 Blue supercompetent cells were obtained from Stratagene (La Jolla, CA) and grown on 1% yeast extract, 1% Tryptone, and 0.5% NaCl at 37°C. Rhodococcus sp. NCIMB 9784 was obtained from the National Culture of Industrial and Marine Bacteria (Aberdeen, United Kingdom). The bacterium was maintained on nutrient agar slopes at room temperature. Ten 250-ml shake flasks containing 50 ml of basal salts medium supplemented with 35 mM sodium pyruvate were inoculated from slope and grown on an orbital shaker at 220 rpm at 30°C for 3 days. This combined inoculum was used to seed 10 liters of basal salts medium in a 12-liter fermentation vessel (BioFlo 1000 fermenter, New Brunswick Scientific) supplemented with 1 g liter Ϫ1 (R)-(ϩ)-camphor (Aldrich). Following 2 days of growth at 30°C with an impeller speed of 250 rpm and an air flow of 4 liters min Ϫ1 , a further 1 g liter Ϫ1 was added. After an additional 1 day of growth, the bacteria were harvested by centrifugation to yield a typical wet weight of 4 g liter Ϫ1 .
Enzyme Assays-Activity of 6-oxocamphor hydrolase was monitored using a Hewlett Packard 8453 UV-visible spectrophotometer. The assay was performed as follows: to a 3-ml stirred cuvette containing 2990 l of 50 mM Tris/HCl buffer, pH 7.0 (henceforth referred to as "buffer") and the appropriate concentration of substrate (for standard assays, 100 M) was added 10 l of enzyme solution, and the disappearance of substrate was measured using a decrease in absorption at 294 nm. Calculations were made using an ⑀ value of 258 mol dm Ϫ3 cm Ϫ3 for 6-oxocamphor. For inhibition/activation studies, 2990 l of buffer containing 3 units of 6-oxocamphor hydrolase and the appropriate concentration of additive were preincubated at 25°C prior to assay. The assay was initiated by addition of 10 l of a 30 mM ethanolic solution of the substrate.
Enzyme Purification-40 g of cell paste was suspended in 150 ml of buffer, and to this suspension was added 150 ml of 5-mm glass beads. The mixture was homogenized twice for 15 min each at 4500 rpm in a Type KDL Dynomill. The resulting homogenate was centrifuged at 11,000 rpm using a GSA rotor to yield a cell extract (320 ml) from which a 40 -80% ammonium sulfate cut was derived. After dissolution in buffer and dialysis against 6 liters of the same buffer (with one change), the protein solution (100 ml) was taken to a concentration of 1.7 M ammonium sulfate and loaded onto a 2.5 ϫ 10-cm phenyl Sepharose column that was eluted with a decreasing gradient of ammonium sulfate. Fractions exhibiting 6-oxocamphor hydrolase activity were pooled and precipitated by the addition of 80% ammonium sulfate. After overnight dialysis against 6 liters of buffer at 4°C, the protein solution (25 ml) was applied to a 2.5 ϫ 10-cm Fast Flow Q Sepharose column and eluted with an increasing gradient of potassium chloride (0 -0.5 M). Active fractions were pooled, precipitated, and dialyzed as above. The resulting protein solution (6 ml) was taken to a concentration of 1.7 M ammonium sulfate and divided into two aliquots, which were separately loaded onto a prepacked H/R 5/5 phenyl Superose column (1 ml) and eluted with a decreasing gradient of salt. Active fractions were identified, and their purity was assessed by SDS-polyacrylamide gel electrophoresis prior to being pooled. The enzyme was stored at Ϫ80°C for at least six weeks with no discernible loss of activity.
The isoelectric point of the purified protein was determined by isoelectric focusing using an Amersham Pharmacia Biotech Phastsystem. The native molecular weight was determined by calibrated gel filtration chromatography on a Superose 12 gel filtration column against a range of commercially available standards (Amersham Pharmacia Biotech). The void volume of the column was determined using blue dextran 2000. N-terminal sequencing was performed according to the method of Hayes et al. (11), and 18 residues were unambiguously assigned as Met-Lys-Gln-Leu-Ala-Thr-Pro-Phe-Gln-Glu-Tyr-Ser-Gln-Lys-Tyr-Glu-Asn-Ile.
Liquid chromatography-mass spectrometry of pure 6-oxocamphor hydrolase was performed on a Waters/Alliance 2690 HPLC system fitted with a Phenomenex Jupiter C18 300-Å, 250 mm ϫ 2 mm ϫ 5 m column. The flow rate was 200 l min Ϫ1 , and a gradient of 0 -95% water/acetonitrile was employed. The liquid chromatography apparatus was fitted to a Waters 286 UV detector and a Micromass Platform II single quadrupole mass spectrometer utilizing an electrospray ionization source controlled via the VG Mass-Lynx software (VG Biotechnology Ltd., Altrincham, Cheshire, United Kingdom). The source temperature was 140°C. Capillary voltage was 3.3 kV, and the cone voltage was ramped from 40 -90 V over a range of 500 -2000 m/z. The instrument was calibrated over this M r range with horse heart myoglobin (Sigma). Nucleic Acid Preparation and Cloning-Genomic DNA was prepared by a modified version of the method described by Kulakova et al. (12). Approximately 0.5 g of wet cell paste was washed twice in 5 ml of 10 mM EDTA, pH 8.0, and then resuspended in 5 ml of 75 mM NaCl, 25 mM EDTA, pH 8.0, 20 mM Tris/HCl, pH 8.0, containing 5 mg/ml lysozyme. The mixture was incubated for 2 h at 37°C before adding 50 l of proteinase K solution (20 mg/ml) and 300 l of 10% (w/v) SDS. This was incubated for a further 2 h at 55°C with occasional inversion. The solution was extracted once with phenol (equilibrated with Tris/HCl, pH 8.0) and then twice with chloroform. DNA was precipitated with isopropanol and spooled onto a glass rod. After rinsing with 70% (v/v) ethanol, the DNA was air-dried and dissolved in 0.5 ml of TE buffer (10 mM Tris HCl, pH8, 1 mM EDTA pH8).
A partially degenerate oligonucleotide (36-mer) designed against the N-terminal sequence (residues 7-18) of the 6-oxocamphor hydrolase activity was synthesized as a hybridization probe: 5Ј-CCSTTCCAG-GAGTACWSSCAGAAGTACGAGAACATC-3Ј (where S represents G or C, and W represents A or T). The oligonucleotide mix was radiolabeled by a kinase reaction using T4 polynucleotide kinase and [␥-32 P]ATP under standard conditions (13). Total DNA, digested to completion with different restriction endonucleases and blotted onto a Hybond-N membrane (Amersham Pharmacia Biotech) was hybridized to the radiolabeled oligonucleotide mix at 55°C for 48 h. The membrane was then washed twice at room temperature with 300 mM NaCl, 30 mM sodium citrate containing 0.1% (w/v) SDS for 15 min and twice at 55°C with 30 mM NaCl, 3 mM sodium citrate containing 0.1% (w/v) SDS for 15 min. Following autoradiography, a single cross-hybridizing band was detected in each lane. EcoRI, SacI, and SmaI digests of genomic DNA were separated on preparative gels; the regions spanning ϳ4.5, 3.6, and 2.0 kbp, 1 respectively, were excised; and the DNA was extracted and shotgun-cloned into pUC18 vector. Following transformation into E. coli XL1 Blue, positive clones were isolated by a colony lift procedure (13) using hybridization and washing conditions identical to those used for the Southern blot. Clones were verified using a dot blot procedure (13) prior to sequencing.
DNA Sequencing and Analysis-Double-stranded DNA sequencing (14) of plasmid DNA prepared from positive clones was carried out with an automated DNA sequencer (ABI PRISM 377, PerkinElmer Life Sciences). All sequencing was carried out on both strands. Computer-assisted sequence analysis was performed using the DNAStrider and MacVector software packages. Data base homology searches (SwissProt release 39 protein data base) were carried out using the NCBI BLAST server. The nucleotide sequence data reported in this paper has been deposited at EMBL and GenBank TM with the accession number AF323755. Table I shows that the purification of 6-oxocamphor hydrolase from crude cell extract proceeds with a yield of 5% and a factor of 35.7. SDSpolyacrylamide gel electrophoresis of the purified protein (Fig.  2) suggested a denatured molecular mass of ϳ35,000 Da, but electrospray mass spectrometry confirmed a smaller subunit mass of 28,488 Da. Analysis of the native protein by gel filtration chromatography revealed an apparent native molecular mass of 83,000 Da (average of two determinations). This suggests the protein exists in solution as a trimer (assuming the molecule is not highly elongated, which would give it a much larger than expected Stokes radius). An isoelectric point of 8.5 was recorded for the enzyme. 6-Oxocamphor Hydrolase from Rhodococcus sp. 6-Oxocamphor hydrolase was observed to have the same pI as penta-2,4-dione hydrolase and a comparable native molecular mass, the latter being a monomer of 75,000 Da (14). The pH optimum of the enzyme was ascertained to be 7.0, and the enzyme displayed 25% higher activity in 50 mM phosphate buffer than 50 mM Tris/HCl at the same pH.

Purification of 6-Oxocamphor Hydrolase-
Kinetic Properties-The specific activity of 6-oxocamphor hydrolase was determined to be 357.5 units mg Ϫ1 , the K m to be 0.05 mM, and the K cat to be 167 s Ϫ1 for 6-oxocamphor. The high specific activity of the enzyme is reflected in an inability to detect any 6-oxocamphor in fermentation extractions of Rhodococcus sp. NCIMB 9784, which yield a high proportion of 6-endo-hydroxycamphor. It is evident that the transformation of the hydroxycamphor to the diketone represents a bottleneck in the metabolism of camphor by this organism.
The effect of EDTA and various inhibitors of both thiol nucleophile-dependent hydrolases and the serine hydrolase inhibitor phenylmethylsulfonyl fluoride were tested. 6-oxocamphor hydrolase was inhibited to some degree by thiol active reagents, such as 1 mM N-ethylmaleimide (68% relative activity), but most notably 1 mM hydroxymercuribenzoate (14%); 1 mM EDTA had a slight activating effect (119%). Phenylmethylsulfonyl fluoride (1 mM) had almost no effect on activity.
Gene Cloning-The gene encoding the 6-oxocamphor hydrolase was cloned by hybridization with a mixture of oligonucleotides designed against the N-terminal sequence of the purified protein activity. Three overlapping fragments of DNA (2.0 kbp SmaI, 3.6 kbp SacI, and 4.3 kbp EcoRI) were isolated by this procedure (Fig. 3) and cloned into pUC18 (clones S2.0, Sa3.6, and E4.3, respectively). Sequence analysis of the clones revealed several potential open reading frames (ORFs) (Fig. 4).
All displayed the typical codon usage pattern found in Rhodococcus sp., with a strong bias toward GC-rich codons.
The deduced polypeptide translation of one such ORF (camK) matches the N-terminal sequence obtained from the isolated 6-oxocamphor hydrolase activity. camK encodes a protein of 257 amino acids, and the predicted ATG start codon is positioned 5 bp 3Ј of a purine-rich region that may act as a ribosome binding site (Fig. 5). Furthermore, the calculated mass of the polypeptide (28,482 Da) is very close to the experimentally determined mass of the purified protein by electrospray mass spectrometry (28,488 Da). Comparison of the translated sequence with the SwissProt data base revealed significant homology to the crotonase superfamily of enzymes from several sources. The best alignments were obtained with crotonase (enoyl-CoA hydratase) from Clostridium acetobutylicum (16) and E. coli (17), revealing 45 and 42% homology, respectively (data not shown). Significantly, close homology was also observed with 2-ketocyclohexanecarboxyl coenzyme A hydrolase from Rhodopseudomonas palustris (18) and 4-chlorobenzoyl-CoA dehalogenase from a Pseudomonas sp. (19). A sequence comparison of the translated sequence against representative members of the crotonase superfamily is given in Fig. 6.
The 3Ј end of the gene encoding 6-oxocamphor hydrolase (camK) has an overlap of 1 nucleotide, encompassing the TGA stop codon and a predicted GTG start codon of another open reading frame (ORF1). ORF1 appears to be translationally coupled to camK and encodes a protein of 167 amino acids. A BLAST search of the SwissProt data base indicates homology to maoC gene from Klebsiella aerogenes (20), which belongs to the aldehyde dehydrogenase family and to the short-chain dehydrogenase/reductase family of enzymes (e.g. 17␤-estradiol dehydrogenase from rat).
Downstream of ORF1 is an ORF2 encoding a polypeptide of 408 amino acids that displays similarity to nonspecific lipid transfer proteins from various species (e.g. 49% homology to chicken protein). The function of this open reading frame in relation to camphor metabolism is not known.
Upstream of camK, an ORF encoding a polypeptide of 206
Further upstream of the proposed transcriptional regulator is ORF4, transcribed in the same direction as camK. Two alternative potential ATG start codons in the same reading frame were identified at positions 492 and 498 in the nucleotide sequence (Fig. 5). Because both ATG codons are positioned just downstream (5 and 6 bp, respectively) of a potential ribosome binding site, it is not clear which one represents the start of the ORF. For clarity, we have assigned the more 5Ј ATG as being the start codon. The ORF encodes a polypeptide of 396 amino acids. Sequence analysis of the translated product revealed convincing homology to a number of ferredoxin reductase proteins in the data base. Conservation of sequence was particularly obvious in the regions involved in adenine nucleotide binding (data not shown). The best homology was found with the ferredoxin (rhodocoxin) reductase (ThcB) involved in the biodegradation of thiocarbamate from Rhodococcus sp. NI86/21 (21) (46% homology) and the putidaredoxin reductase involved in the hydroxylation of camphor by P. putida (22) (46% homology).
The activity of crotonase, or enoyl-CoA hydratase (ECH), has been the subject of intensive study over many years owing to its central role in the ␤-oxidation pathway. The essential activity of ECH in this regard has been the stereospecific reversible hydration of enoyl-CoA molecules of varying fatty acid length to yield ␤-hydroxy thioesters (25,26,30). The catalytic mechanism of ECH is dependent on stabilization of an enolate intermediate by hydrogen bonding to an oxyanion hole created by two peptidic NH groups in the active site of the enzyme, Ala-98 and Gly-141 (Fig. 8). In recent years, the comparison of genetic sequence information for a wide range of enzymatic activities has revealed that there exists a superfamily of crotonase-like proteins, each member of which catalyzes a reaction that is dependent on the same general stabilization of an enolate anion (17). Activities assigned to the crotonase family include double-bond isomerization (27), aromatic ring closure (28), 1,3dioxo cleavage (18), dehalogenation (19), and decarboxylation (29), in addition to double bond hydration (Fig. 9). Although the overall amino acid sequence is well conserved between these enzymes, crucial active site residues have been shown to be present in some members of the family and absent in others, suggesting a nonconserved mechanism of enol stabilization and water transfer. In enoyl-CoA hydratase from rat mitochondria, two glutamate residues at the active site are responsible for acid/base catalysis of double bond hydration: Glu-144 facilitates attack of nucleophilic water to the carbonyl, and Glu-164 donates a proton to the ␣-carbon in the final step to yield the hydrated product (Fig. 8).
Homology between ECH and 6-oxocamphor hydrolase is conserved throughout most of the length of the polypeptide chain. Identity is most pronounced in the central region of the protein, which constitutes the spiral domain of ECH, most especially in the A3 B3 ␤-strand region 129 PVIAAVNG, although the 140 GGG turn that is present in both ECH, 2-ketocyclohexanecarboxyl coenzyme A hydrolase (18) and 4-chlorobenzoyl-CoA dehalogenase (4CBD) (19) is notably absent in 6-oxocamphor hydrolase. Whereas homology persists throughout the so-called
It is notable that of the two active site glutamate residues in ECH, only Glu-144, which facilitates attack of water on the 3-carbon of the enoyl-CoA substrate, is conserved in 6-oxocam-phor hydrolase (Glu-124). This residue is absent in other members of the superfamily, including those of which the activity most closely resembles that of 6-oxocamphor hydrolase, e.g.

2-KCH (18) and 4-chlorobenzoyl-CoA dehalogenase, in which
Asp-145 has been identified as the crucial catalytic residue (19). Glu-144 is conserved between ECH and methylmalonyl-CoA decarboxylase (29), but in this case, it has been shown that this glutamate residue cannot be catalytic. Hence, it cannot be assumed with certainty that the homologous glutamate in 6-oxocamphor hydrolase is catalytic. The other catalytic residue of ECH, Glu-164, is not conserved in 6-oxocamphor hydrolase. There are several candidate residues with the required acid/base character capable of forming an acid-base couple with Glu-124 in 6-oxocamphor hydrolase, including Glu-136 (conserved with rat ECH and 2-KCH), Asp-142, and perhaps Asp-154, which is one residue distant from the active Asp-145 of 4CBD. Despite the absence of structural studies or site directed mutation experiments to determine the actual active site residue in 6-oxocamphor hydrolase responsible for water activation, it is nevertheless still possible to tentatively postulate a mechanism for asymmetric ␤-diketone hydrolysis based on the results of sequence comparison.
The activation of water in the active site of 6-oxocamphor hydrolase, possibly by Glu-124, in concert with another residue would facilitate nucleophilic attack of water at the pro-S carbonyl, yielding the (S)-enantiomer of keto acid 4 (Fig. 10). It is possible that tautomerization of the keto form is not an enzyme-catalyzed process because the same ratio of diastereomers (predominantly cis-) is also observed in acid-catalyzed hydrolysis of 3 to 4.
The desymmetrization of bicyclic ␤-diketones, and indeed the hydrolysis of 2,2-dialkylcyclohexanones, constitutes a novel reaction in the crotonase superfamily. The previously reported activity of a crotonase homologue that bears the closest resemblance is that of 2-KCH, which hydrolyzes a ␤-dicarbonyl species. Importantly, however, 6-oxocamphor hydrolase is the first crotonase homologue not to have an activity dependent on coenzyme A for substrate activation. Indeed, of those residues shown to be responsible for coenzyme A binding in ECH (e.g. Lys-92, Lys-101, and Lys-282, involved in forming salt bridges to the phosphates of ADP) and 4CBD (Arg-24 and Arg-67), none is conserved in 6-oxocamphor hydrolase.
Interestingly, all the compounds which act as substrates for 6-oxocamphor hydrolase have nonenolizable 1,3-diketones. This factor may explain why the retro-Claisen reaction mediated by 6-oxocamphor hydrolase is effected using a crotonase type mechanism. In contrast, similar reactions on enolizable diketones, such as fumarylacetoacetate (mediated by fumary-  6-Oxocamphor Hydrolase from Rhodococcus sp. lacetoacetate hydrolase (9)), utilize a serine hydrolytic triad type mechanism, in which the energy barrier to hydrolysis maybe greater. It remains to be seen whether other activities, such as cyclohexane-1,3-dione hydrolase (31) or polyvinylketone hydrolase (which is active against pentane-1,4-dione (15)), are of the fumarylacetoacetate hydrolase or crotonase type.
Sequence analysis of the clone encoding the 6-oxocamphor hydrolase activity revealed several proximal genes encoding proteins that may be involved in camphor metabolism. ORF1 is immediately downstream and apparently translationally coupled to camK, and it encodes a small protein that displays homology to dehydrogenase enzymes. This open reading frame encodes the sequence 40 SDISMFAGLTGD, which is somewhat similar to a sequence motif typical for FAD-binding enzymes, TXXXXhhhhGD (32) (where h denotes a hydrophobic amino acid). However, it does not contain the expected GXGXXG sequence typical of both FAD-requiring and NAD-dependent dehydrogenases (33). The function of this activity therefore remains uncertain at the present time. ORF3, encoding a regulatory protein, was found upstream of camK, along with ORF4, which encodes a ferredoxin reductase. The first step in the metabolism of (1R)-(ϩ)-camphor in Rhodococcus sp. NCIMB 9784 involves hydroxylation at the 6-endo position by a cytochrome P450 (Fig. 1). Because genes encoding enzymes involved in metabolic pathways are often linked, we speculate that the polypeptide encoded by ORF4 may be involved in furnishing the camphor hydroxylase with electrons.
In conclusion, we have demonstrated that the activity of 6-oxocamphor hydrolase, namely the retro-Claisen reaction of (bi)cyclic ␤-diketones, in the metabolism of camphor by Rhodococcus sp. NCIMB 9784 is attributable to a novel activity of the crotonase superfamily. The precise mechanism of the reaction awaits elucidation by site directed mutagenesis and detailed kinetic analysis of mutant activity. The wild type enzyme is already effective at performing the desymmetrization of bicyclic ␤-diketones with a high degree of enantioselectivity and hence constitutes an important addition to the array of biocatalysts that might be employed in the synthesis of fine chemical intermediates.