A Cytochrome P450 Involved in the Metabolism of Abietane Diterpenoids by Pseudomonas abietaniphila BKME-9

Abstract

Diterpenoids are naturally occurring plant compounds which have pharmaceutical properties. We have sequenced a 10.4-kbp extension of the dit cluster in Pseudomonas abietaniphila BKME-9, containing genes involved in abietane diterpenoid biodegradation. The ditQ gene was found to encode a cytochrome P450 monooxygenase, P450_dit, and to be homologous to the tdtD gene of Pseudomonas diterpeniphila A19-6a. Knocking out ditQ had little effect on growth of BKME-9 on abietic acid but severely impaired growth on dehydroabietic acid (DhA) and palustric acid (PaA), increasing doubling times from 3.8 to 15 h on DhA and from 5.6 to 18.5 h on PaA. A xylE transcriptional fusion showed that transcription of ditQ was induced by a range of diterpenoids. Substrate binding assays of P450_dit expressed in Escherichia coli revealed that DhA binds to the enzyme and yields a type I binding spectrum with a K_d of 0.4 μM. These results indicate that P450_dit is involved in the metabolism of DhA and PaA and are consistent with its putative role in converting DhA to 7-hydroxy-DhA. Finally, an amino acid sequence identity of greater than 72% and conserved gene arrangement support the hypothesis that the dit gene cluster of P. abietaniphila BKME-9 and the tdt cluster of P. diterpeniphila A19-6a contain functional homologues.

Resin is produced by most species of coniferous trees such as the Grand fir (Abies grandis) and Norway spruce (Picea abies). Resin consists of a volatile turpentine fraction, containing monoterpenes and sesquiterpenes, plus a nonvolatile rosin fraction. The rosin fraction consists mostly of diterpenoids, or resin acids, which can constitute up to a few percent of the total biomass of these trees. Resin acids are C-20 carboxylic acid-containing compounds produced in conifers through the cyclization of geranylgeranyl diphosphate by diterpene synthase and subsequent oxidations involving two cytochromes P450 (7, 13). This report focuses on bacterial metabolism of the abietane diterpenoids, palustric acid (PaA), dehydroabietic acid (DhA), abietic acid (AbA), and 7-oxo-dehydroabietic acid (7-oxo-DhA) (Fig. 1).

FIG. 1.

Proposed pathway for abietane degradation by P. abietaniphila BKME-9. Chemical designations: I, PaA; II, DhA; III, 7-hydroxy-DhA; IV, 7-oxo-DhA; V, 7-oxo-11,12-dihydroxy-8-13-abietadien acid; VI, 7-oxo-11,12-dihydroxydehydroabietic acid; VII, AbA.

The potential use of diterpenoids in pharmaceutical applications is under investigation. AbA and its derivatives have recently been evaluated for their ability to function as inhibitors of fungi (5), tumors, mutagenesis, viruses, nitric oxide production (9, 12), inflammation (6), and lipoxygenase activity (26). Therefore, enzymes capable of transforming diterpenoids may have potential applications in the synthesis of useful compounds.

A bacterium isolated from bleach kraft pulp mill effluent, Pseudomonas abietaniphila BKME-9, grows on the abietane diterpenoids, AbA, DhA, PaA, and 7-oxo-DhA, as sole organic substrates (3). The dit gene cluster of BKME-9 encodes enzymes required for the catabolism of these compounds (14, 15). Several genes of the dit cluster were previously sequenced and characterized, and a convergent pathway for abietane diterpenoid metabolism was proposed. The enzyme catalyzing the formation of a catecholic intermediate in the proposed pathway is the ring-hydroxylating dioxygenase, DitA. DitA catalyzes the hydroxylation of the aromatic ring to form a cis-dihydrodiol intermediate (15). Evidence for a convergent pathway came from studies of a ditA1 knockout mutant (14). 7-Oxo-DhA accumulated in cell suspensions of the ditA1 mutant strain incubated with AbA or DhA, while DhA and 7-oxo-DhA accumulated in cell suspensions incubated with PaA. The aromatization of the C ring of both AbA and PaA suggested a convergent pathway with DhA serving as an intermediate.

The initial steps in the biodegradation pathway have not been elucidated, but some evidence suggests that a P450 monooxygenase is involved. P450s are heme-thiolate proteins that function as the catalytic component of an electron transport chain generally including a ferredoxin and a ferredoxin reductase in bacterial systems (19). A putative P450, encoded by the tdtD gene, that may function in abietane diterpenoid degradation was recently identified in Pseudomonas diterpeniphila A19-6a, another resin acid-degrading bacterium closely related to BKME-9 (17, 18). Morgan and Wyndham (17) reported that a tdtD knockout mutant of A19-6a was retarded in its removal of DhA or AbA from its growth medium when compared to that of the wild type. The mutant retained the ability to grow on DhA and AbA as sole organic substrates; however, any effects of the mutation on growth rate were not reported. These results suggest involvement of the tdtD gene in diterpenoid metabolism but give no conclusive evidence for a functional P450 gene product or the role of such an enzyme in resin acid metabolism. Morgan and Wyndham also provided evidence for a homologue of the tdtD gene in BKME-9 but were unable to conclude whether this gene was linked to the previously described dit cluster.

In light of the above evidence, a BKME-9 genomic DNA fragment contiguous to the dit cluster was cloned and characterized. Here we provide previously missing evidence for the existence of a P450 enzyme and demonstrate that it is involved in diterpenoid metabolism. We also show an apparent difference in strains BKME-9 and A19-6a, as the former does not require the P450 for normal metabolism of AbA. We characterized the P450 by using a gene fusion transcriptional reporter and by determining CO and substrate binding spectra of the protein expressed in Escherichia coli. We also show that in BKME-9 the P450 gene is linked to the dit genes, and we provide additional evidence for the homology of diterpenoid degradation genes in both BKME-9 and A19-6a.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. E. coli was cultured on Luria-Bertani medium and P. abietaniphila BKME-9 was cultured on tryptic soy broth or mineral medium supplemented with diterpenoids as previously described (16). Diterpenoids were supplied by Helix Biotechnologies, Richmond, Canada. The AbA used was approximately 96% pure, with another diterpenoid, most likely DhA, comprising the remainder. The PaA used was approximately 90% PaA, 7% DhA, and 3% AbA. The 7-oxo-DhA used was greater than 97% pure with trace amounts of several undetermined diterpenoids. The DhA and isoprimanc acid used in the study were greater than 99% pure.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Genotype or description	Reference or source
Strains
P. abietaniphila
BKME-9	Wild type; grows on abietane diterpenoids	3
P450KO	ditQ::xylE-Gmr	This study
P. diterpeniphila
A19-6a	Wild type; grows on DhA	17
E. coli
DH5α	endA1 hsdR17(rK− mK−) supE44 thi-1 recA1 gyrA (Nalr) relA1 Δ(lacIZYA-argF)U169 deoR (φ80dlacΔ(lacZ)M15)	Gibco BRL
S17-1	recA pro thi hsdR with integrated RP4-2-TcMu::Kna::Tn7; Tra+ Trr Smr	23
Plasmids
pUC19	Cloning vector, Apr	29
pEX100T	sacB conjugable plasmid for gene replacement; Apr	22
pX1918G	xy/E-Gmr fusion cassette-containing plasmid; Apr Gmr	22
pDS1	5.1-kb EcoRI fragment of pLC162 cloned into the EcoRI site of pUC19	This study
pEXP450	1,482-bp EcoRV-SmaI of pDS1 cloned into the SmaI site of pEX100T	This study
pEXP450KO	PstI xylE-Gmr cassette of pX1918G cloned into the PstI site of pEXP450	This study
pLC48	SuperCos1 cosmid library clone containing DhA degradation genes	15
pLC162	SuperCos1 cosmid library clone containing DhA degradation genes	15

DNA manipulation.

Plasmid DNA was isolated by the standard alkali lysis (2) or with a QIAprep Spin Miniprep kit (QIAGEN, Santa Clarita, Calif.). Restriction endonuclease (New England Biolabs, Beverly, Mass., or Gibco BRL, Gaithersburg, Md.) digestions were performed by standard procedures. DNA fragments were purified from agarose gels with a QIAquick gel extraction kit (QIAGEN). Plasmid pUC19 and E. coli DH5α were used to clone the 5.1-kbp EcoRI fragment from pLC162 to generate pDS1. Standard techniques of Southern hybridization analysis were followed as previously described (2). Briefly, SS maximum strength Nytran Plus was used for blotting (Schleicher and Schuell, Keene, N.H.). The immobilized DNA was hybridized to tdtD (kindly provided by R. C. Wyndham, Institute of Biology, Carleton University) labeled with [α-³²P]dCTP by using the Nick translation system from Gibco BRL. Hybridization was analyzed by using standard phosphorimager scanning and autoradiography techniques.

Sequencing and sequence analysis.

Successive unidirectional deletions of pDS1 DNA were prepared for sequencing using the double-stranded nested deletion system from Pharmacia Biotech (Uppsala, Sweden). Support protocol provided with the system was followed as per the manufacturer's recommendations. Standard M13 primers for sequencing the deletion clones were supplied by the Nucleic Acid and Protein Service at the University of British Columbia. Primers used for “primer walking” of cosmid library clones were supplied by AlphaDNA (http://www.alphadna.com). AlphaDNA also supplied primers used for colony PCR of P450KO and sequencing of the PCR product. DNA sequences were determined by the Nucleic Acid and Protein Service. A consensus nucleic acid sequence was prepared using Bioedit (version 5.0.9), available at http://www.mbio.ncsu.edu/RNaseP/info/programs/BIOEDIT/bioedit.html. ORF finder software at http://www.ncbi.nlm.nih.gov/gorf/gorf.html was used to determine open reading frames (ORFs) and to conduct sequence similarity searches using the BLASTP software (2.2.6) from the National Center for Biotechnology Information website. The ClustalW multiple alignment program included with the Bioedit software was used to align and analyze protein sequences by using the default setting.

Knockout mutants.

A knockout of ditQ was created by gene replacement to yield strain P450KO. Plasmid pEXP450 was constructed by ligating a 1,482-bp EcoRV-SmaI blunt-end fragment from pDS1 into the dephosphorylated unique SmaI site of pEX100T (22), containing the sacB counterselectable marker and transforming E. coli DH5α. Next, a PstI-digested xylE-Gm^r transcriptional fusion antibiotic cassette of pX1918G (22) was ligated into the dephosphorylated unique PstI site of pEXP450, which disrupted the P450 gene, and the product was used to transform E. coli S17-1 to create pEXP450KO. Homologous recombination of the mutated allele into strain BKME-9 was accomplished by diparental conjugation (8) followed by a two-step selection method, as previously described (15). Successful gene replacement was verified by colony PCR (30) with primers targeted to the P450 gene (P450-404left, 5′-GCGGACCTTGAAGGTAGCGA-3′; and P450-3567right, 5′-GCAACTTCATGGCAGGCCTT-3′) at an annealing temperature of 61°C and a 4-min extension time. In order to confirm insertion into the gene of interest, the 3,163-bp amplicon from the above PCR was then used in two sequencing reactions with P450-404left and P450-3567right as primers.

Growth and cell suspension assays.

Cultures of BKME-9 and P450KO were grown overnight at 28°C on mineral medium supplemented with 90 mg of DhA/liter or with 1 g of sodium pyruvate/liter supplemented with 4 mg of gentamicin/liter. These overnight cultures were then transferred to mineral medium supplemented with 1 g of sodium pyruvate/liter. After overnight growth, the cells were collected by centrifugation, washed, and suspended in sterile saline at an optical density at 600 nm (OD₆₀₀) of 0.6. These cell suspensions were then used to inoculate (0.1%) 2-ml cultures in solvent-washed tubes with mineral medium supplemented with either 1 g of sodium pyruvate/liter, approximately 90 mg of AbA/liter, 90 mg of DhA/liter, 90 mg of PaA/liter, or 95 mg of 7-oxo-DhA/liter. All cultures were then incubated on a rotary shaker at 28°C. At selected time intervals, two to four repicates of 2-ml cultures of each strain were removed from the incubator. Half of the cultures were acidified with two drops of 1 M HCl and immediately frozen at −20°C for later analysis of abietanes by gas chromatography (GC)-flame ionization detector, as previously described (16). From the other half of the cultures, 1 ml was centrifuged, the pellet was washed with 0.9% sterile saline, the suspension was centrifuged again, and the pellet was frozen at −20°C. These samples were later used to determine protein concentration using the microbicinchoninic acid protein assay kit (Sigma) and bovine serum albumin as the standard (24). Bicinchoninic acid protein quantification was used to monitor growth as opposed to OD, because resin acids precipitated in the medium prevented accurate measurement of OD.

Cell suspension assays were conducted as previously described (16). GC electron impact mass spectrometry (MS) of methyl ester derivative was conducted as previously described (15) using an Agilent Technologies 6890N network GC system equipped with an Agilent 5973 mass selective detector. National Institute of Standards and Technology MS Search (2.0) was used to analyze mass spectral data.

C23O assays.

Previously, it was reported that wild-type BKME-9 shows no endogenous catechol-2,3-dioxygenase (C23O) activity and that activity served as an adequate reporter for gene induction studies (14). For C23O assays, strain P450KO was grown on mineral medium supplemented with 1 g of sodium pyruvate/liter to an OD₆₀₀ between 0.15 and 0.3 and then spiked with a potential inducer, 150 mg of DhA/liter, 150 mg of AbA/liter, 158 mg of 7-oxo-DhA/liter, 150 mg of isopimaric acid/liter, 37 mg of 12,14-dichlorodehydroabietic acid/liter, 15.4 mg of biphenyl/liter, 12.0 mg of naphthalene/liter, or 17.8 mg of phenanthrene/liter. These cultures were incubated until they reached an OD₆₀₀ between 0.6 and 0.7. Cultures were then harvested, washed in 10 mM KPO₄ buffer (pH 7.5) at 4°C, and suspended in the buffer at an OD₆₀₀ of 6.0. Triplicate enzyme assays were performed on whole cells suspended at an OD₆₀₀ of 0.1 in 1 ml of the buffer containing 500 μM catechol. C23O activity was assayed spectrophotometrically at 30°C as the formation of 2-hydroxy semialdehyde at 375 nM (ɛ = 44 mM⁻¹ cm⁻¹) for 3.5 min.

Spectrophotometric assays.

A 2-liter flask containing 1 liter of Luria-Bertani medium with 50 μg of ampicillin/ml was inoculated with 5 ml of an overnight culture of E. coli harboring pEXP450 or pEX100T. The culture was incubated with shaking until the OD₆₀₀ reached approximately 0.6. Expression of ditQ was induced by addition of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) and further incubation for 18 to 24 h. Cells were harvested by centrifugation at 8,275 × g in a Sorvall SLA 3000 rotor for 15 min. The pellet was washed with 1 liter of Tris-Cl, pH 7.4, and centrifuged as above. The pellet was then suspended in 5 ml of the buffer plus 1 mM dithiothreitol and 1 mM phenylmethylsulfonyl fluoride. The suspension was passed through a French pressure cell two times, and the crude lysate was centrifuged for 30 min at 25,000 × g in a Sorvall SS-34 rotor. The supernatant was removed and the crude extract was used for spectrophotometric analysis, using a Cary 1E spectrophotometer and Cary UVWin scan application version 2.00 software.

The reduced CO binding spectrum was obtained with 200 μl of crude extract added to 1.80 ml of the same buffer as above plus a few crystals of sodium dithionite to reduce the sample. The sample was then equally divided in two 1-ml, optically matched cuvettes. One sample was treated by bubbling carbon monoxide through the cuvette slowly for 30 s. The second sample was used as the reference in difference spectroscopy with the carbon monoxide-treated sample.

Substrate binding assays were performed in two optically matched, 3-ml cuvettes, each with 300 μl of the above crude extract plus 2.70 ml of buffer A. Increasing concentrations of substrate were added to the sample cuvette, and difference spectra were determined from 350 to 500 nm. The binding constant, K_d, was determined using the following nonlinear fitting equation: ΔA = ΔA_M{([L_T] + [E_T] + K_d) − ({[L_T] + [E_T] + K_d}² − 4[L_T][E_T])^0.5}/(2[E_T]), where ΔA is the difference in absorbance between 387 and 425 nm, ΔA_M is the maximum change in absorbance, [L_T] is the total ligand concentration, and [E_T] is the total enzyme concentration (4).

Nucleotide sequence accession number.

The nucleotide sequences reported in this study have been submitted to GenBank under accession no. AF119621.

RESULTS

Sequencing of 10.4-kbp region containing a putative P450 gene.

To determine if a gene cluster corresponding to the tdt cluster of A19-6a is located adjacent to the dit cluster of BKME-9, we analyzed by Southern blotting nine EcoRI-digested BKME-9 cosmid library clones containing ditA1 by using the tdtD gene as a probe. The lanes containing fragments from cosmids pLC48 and pLC162 had single bands with high intensity of approximately 6 and 5.1 kbp, respectively. The 5.1-kbp fragment of pLC162 hybridizing to the tdtD probe was cloned and sequenced. This 5.1-kbp fragment is located approximately 2.8 kbp downstream from ORF1 of the previously characterized dit cluster (Fig. 2). The 2.8-kbp gap was sequenced from cosmid library clone pLC162 by primer walking. We also used primer walking to sequence 2.5 kbp beyond the end of the 5.1-kbp fragment opposite to the gap, by using the pLC48 cosmid as a template. Thus, a total region 10.4 kbp adjacent to the dit cluster was sequenced.

FIG. 2.

Physical map of the dit gene cluster of BKME-9, the tdt gene cluster of A19-6a, and putative homologues from LB400. Genes are represented by arrows with patterns corresponding to putative functional groups. Numbers below each gene represent the percent amino acid identity corresponding to the deduced protein sequence of the BKME-9 gene above. Double vertical lines indicate gaps in the genome sequence of unspecified length, which may contain additional ORFs. Horizontal lines refer to the cosmid library clone used for subcloning and sequencing of BKME-9 DNA and the dit cluster.

The region of the above 5.1-kbp fragment, presumably hybridizing to the tdtD probe, corresponds to an ORF designated ditQ. A similarity search of the nonredundant GenBank database using BLASTP (1) with the deduced amino acid sequence of ditQ indicated that ditQ codes for a putative P450 (Table 2). Comparison of the inferred amino acid sequence of ditQ to the Cluster of Orthologous Groups of protein (COG) and Protein families database of alignments and HMMs showed similarity to P450s (Table 3). Alignment of the P450_dit deduced protein with the well-characterized P450_cam showed conservation of functional residues including the highly conserved heme-binding region with consensus sequence FG(F/H)G(P/S)H(M/L)C.

TABLE 2.

Amino acid sequence comparison of deduced DitQ (P450_dit) to proteins in the nonredundant database obtained by BLASTP search

Proteins with similar sequence (protein function)	E-value	% identity (no. of residue)	Organism	Reference	Accession no.
TdtD (cytochrome P450)	e-178	84 (424)	P. diterpeniphila	17	AAK95585
ERYK cytochrome P450 113A1 (erythromycin B/D C-12 hydroxylase)	7e-22	26 (397)	Saccharopolyspora erythraea	25	P48635
CYP108 cytochrome P450terp (α-terpineol oxidation)	1e-18	23 (428)	Pseudomonas sp.	21	P33006
CamC P450cam (cytochrome P450; camphor 5-monooxygenase)	0.023	17 (415)	Pseudomonas putida	27	P00183

TABLE 3.

Conserved domain search and COG comparison

Gene	Deduced no. of residues	Functional assignment based on COG comparison	Pfam database search resultsa
			Conserved domain(s)	E-value	Residues
ORF3	398	COG1960 (Acyl-CoA dehydrogenases)	Acyl-CoA dehydrogenase, N-terminal domain (pfam02771)	2.4e-6	27-145
			Acyl-CoA dehydrogenase, middle domain (pfam02770)	3.4e-16	147-248
ORF4	435	No hits	Amidohydrolase family (pfam04909)	3.4e-08	41-373
ditQ	424	COG2124 (cytochrome P450)	Cytochrome P450 (pfam00067)	2.3e-24	39-421
ditP	140	COG2128 (uncharacterized ancient conserved region)	No hits
ditO	391	COG0183 (acetyl-CoA acetyltransferase)	Thiolase, N-terminal domain (pfam00108)	1.6e-86	1-260
			Thiolase, C-terminal domain (pfam0280)	1.6e-63	265-389
ditN	301	COG1250 (3-hydroxyacyl-CoA dehydrogenase)	3-Hydroxyacyl-CoA dehydrogenase, NAD binding domain (pfam02737)	2.3e-53	3-188
			3-Hydroxyacyl-CoA dehydrogenase, C-terminal domain (pfam00725)	5.8e-41	190-285
ditM	289	COG0179 (2-keto-4-pentenoate hydratase/2-oxohepta-3-ene-1,7-dioic acid hydratase [catechol pathway])	Fumarylacetoacetate hydrolase (pfam01557)	1e-45	86-254
ditL	359	No hits	Amidohydrolase (pfam04909)	3.6e-32	1-344
ditK	249	No hits	Bacterial regulatory proteins, tetR family (pfam00440)	2e-12	67-113
ditJ	546	COG0318 (acyl-CoA synthetases [AMP-forming]/AMP-acid ligases II)	AMP-binding enzyme (pfam00501)	7.5e-105	43-459

^aPfam, Protein families database of alignments and HMMs.

Sequence analysis of 10.4-kbp dit cluster extension.

The nine complete ORFs and one partial ORF in the newly determined 10.4-kbp dit cluster extension are similar to genes encoding two dehydrogenases, a thiolase, a hydrolase, an ancient uncharacterized conserved region, a cytochrome P450, a regulator, a coenzyme A (CoA) ligase, and two unknown hypothetical proteins (Fig. 2). The CoA ligase sequence is contiguous with the previously identified ORF1 (14) of the dit cluster and completes this gene sequence, now designated ditJ (Fig. 2; Tables 2 and 3). A region of the 10.4-kbp extension corresponds very closely to the tdt cluster (17), having the same ORF arrangement and greater than 72% identity of deduced amino acid sequences. ditP is a short coding region that was identified by COG analysis as an ancient conserved region common to two or more phylogenetic branches (Table 3). Morgan and Wyndham (17) did not identify this coding region on the tdt cluster. Further analysis of the tdt sequence did not reveal an ORF corresponding to ditP.

Growth of ditQ mutant on abietanes.

Growth curves of P450KO revealed that ditQ is required for a growth phenotype similar to the wild-type strain on DhA and PaA but not on AbA or 7-oxo-DhA. Doubling times of BKME-9 and P450KO were similar on either 7-oxo-DhA (data not shown) or AbA (Fig. 3A), and both strains reached approximately the same final protein concentrations on either substrate. The growth rates and yields of strain P450KO were substantially lower than those of BKME-9 on DhA (Fig. 3B) or PaA (data not shown). With DhA as a carbon source, BKME-9 had a doubling time of 3.8 h and a final protein concentration of approximately 17 μg/ml, whereas P450KO had a doubling time of 15 h and a final protein concentration of approximately 8.4 μg/ml. Similarly, on PaA, BKME-9 had a doubling time of 5.6 h and a final protein concentration of approximately 19 μg/ml, whereas P450KO had a doubling time of 18.5 h and final protein concentration of 13.5 μg/ml. These results suggest that a P450 encoded by ditQ plays an important role in metabolism of DhA and PaA but not metabolism of AbA or 7-oxo-DhA. We cannot exclude the possibility that the xylE-Gm^r insertion cassette used to create P450KO may have a polar effect on transcription of ORFs downstream of ditQ. But, this is unlikely given that the cassette does not contain a transcription terminator and thus allows transcription of downstream sequences from the aacC1 promoter. Additionally, there is a classic rho-independent terminator sequence located 30 bp downstream of ditQ. The terminator mRNA sequence forming the stem-loop is 5′-ACCCGUGCCU-GAGA-AGGCGCGGGUUUUUU-3′ (with underlined bases indicating the stems and hyphens indicating the loop). The 3′ end of the mRNA has a poly(U) tail that is required for termination. The hairpin structure has a free energy of −20.5 kcal/mol as predicted by Kinefold (http://kinefold.u-strasbg.fr/) or −19.53 kcal/mol as predicted by RNAfold (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi).

FIG. 3.

Growth and substrate removal of BKME-9 and P450KO on DhA and AbA. (A) Protein growth curves of the wild-type, BKME-9, and ditQ mutant strain, P450KO, on AbA showing AbA removal. (B) Protein growth curves of the wild-type and the ditQ mutant strain on DhA showing DhA removal. Error bars indicate standard deviations of the average of three samples.

Abietane removal by P450KO.

Specific rates of removal (μmol h⁻¹ mg of protein⁻¹) of all four abietane diterpenoids were slightly higher for BKME-9 than for P450KO (Fig. 3). Absolute removal rates (μmol h⁻¹) of DhA and PaA were much higher for BKME-9 than for P450KO, but this difference was mainly due to the slower growth and consequent lower biomass of the latter strain. The highest rates of removal occurred during the highest rates of growth in all cases.

Along with the removal of PaA by P450KO, accumulation of DhA was also observed (data not shown), suggesting that PaA is transformed to DhA by the strain. The PaA reagent was 90% pure, also containing 7% DhA and 3% AbA. In cultures of P450KO on PaA, the DhA concentration increased to a maximum at 80 h and was reduced to an undetectable level by 100 h. The AbA concentration did not increase and was also undetectable by 100 h. The increase in DhA was not observed when BKME-9 grew on PaA, and in those cultures, the trace amounts of both DhA and AbA associated with the PaA reagent were removed by 50 h. When P450KO grew on DhA, two putative metabolites accumulated. The same two metabolites were also found at lower concentrations when P450KO was grown on PaA. These metabolites did not accumulate when the wild-type strain was grown on any resin acid tested. Mass spectral analysis by GC-MS was not sufficient to determine the structure of the metabolites.

Specific induction of P450_dit by abietane diterpenoids.

The xylE transcriptional fusion of the ditQ knockout strain encodes catechol-2,3 ring cleavage dioxygenase, which allowed for analysis of induction of ditQ transcription by spectrophotometrically monitoring the production of cleaved catechol-suspended cells incubated with various substrates. The ditQ gene was induced by all four abietane diterpenoids tested (Fig. 4). In fact, a pimerane diterpenoid, isopimaric acid, and a chlorinated diterpenoid, 12,14-dichlorodehydroabietic acid, also induced ditQ, despite these two compounds not being growth substrates for BKME-9. However, this is not surprising considering that the same inducers were identified for ditA1 and ditA3, the genes encoding the α subunit and ferredoxin of the ring-hydroxylating dioxygenase, respectively (14). As previously seen with the dioxygenase components, nonditerpenoid compounds did not induce expression of ditQ above the level of the pyruvate control.

FIG. 4.

Expression of ditQ gene fusion product, C23O, in response to various diterpenoids and aromatic compounds. C23O activity was assayed spectrophotometrically at 30°C as the formation of the yellow catechol cleavage product, 2-hydroxy semialdehyde, at 375 nM (ɛ = 44 mM⁻¹ cm⁻¹) for 3.5 min. Activity values are means ± standard deviations (n = 3) of enzyme assays.

P450_dit reduced carbon monoxide and substrate binding spectra.

P450s are identified by a characteristic Soret maximum produced at 450 nm by binding of carbon monoxide to the reduced enzyme. The difference spectrum of reduced P450_dit expressed in E. coli with and without carbon monoxide produced a Soret maximum at 450 nm (data not shown). This result confirms that ditQ codes for a cytochrome P450.

DhA substrate binding experiments with the crude lysate of E. coli expressing P450_dit yielded a type I substrate binding spectrum, which is a strong indicator that DhA is a substrate for P450_dit. Titration of P450_dit with DhA yielded a type I substrate binding spectrum with a minimum at 387 and a maximum at 425 nm (Fig. 5). Type I curves result from the conversion of low spin hexacoordinated ferric heme with a Soret peak at around 417 nm to a high spin pentacoordinated ferric heme with the displacement of the distal water ligand after substrate binding (11). This results in a decrease in the Soret peak at 417 and an increase of a Soret peak at 387 nm. A plot of the difference in absorbance between 387 and 425 nm versus substrate concentration fitted to the binding curve equation gave an estimated K_d of 0.4 μM with a standard deviation of ±0.03.

FIG. 5.

Binding spectrum for P450_dit with DhA. Data points represent the difference in absorbance between 387 and 425 nm caused by increasing DhA concentration. The curve represents a best fit of the data to the binding equation in which K_d = 0.4 μM and ΔA_max = 0.051. Inset: UV-visible difference spectra of P450_dit with increasing concentration of DhA.

Neither AbA nor PaA produced typical P450 substrate binding spectra, and thus are likely not substrates for P450_dit (data not shown). Although both seemed to cause a perturbation of the heme environment, resulting in a shift of the Soret maximum, the curves produced were ambiguous, and further study is required for definitive analysis of binding on these compounds. It should be noted that impurities of the AbA and PaA solutions, particularly DhA, might have affected the results. 7-Oxo-DhA is also a growth substrate for BKME-9 but clearly did not bind to P450_dit and is therefore not likely a substrate for P450_dit. Isopimaric acid, which is not a growth substrate for BKME-9, may bind to P450_dit weakly but did not yield a typical binding spectrum, and therefore is not likely a substrate of P450_dit.

DISCUSSION

In this study we demonstrated the involvement of a newly identified cytochrome P450 in the metabolism of abietane diterpenoids by BKME-9. A gene knockout of ditQ, coding for P450_dit, indicates that this gene is involved in the degradation of DhA and PaA but not AbA or 7-oxo-DhA. The knockout increased the doubling times and lowered the final protein concentrations of the mutant growing on either DhA or PaA in comparison to those of the wild type (Fig. 3). The P450_dit mutant retained the ability to grow slowly on DhA relative to the wild type. This is an indication that an alternate degradation pathway, not involving P450_dit, is able to productively metabolize DhA and PaA. Perhaps it is this same alternate pathway that is responsible for the metabolism of AbA and 7-oxo-DhA. In addition, growth of strain P450KO on PaA or DhA produced the same putative metabolites, suggesting disruption in the metabolism of these compounds at the same point in a convergent pathway. Induction analysis indicates that ditQ expression is inducible by a range of diterpenoids (Fig. 4), while a substrate binding assay showed that DhA is a likely substrate for this enzyme with a relatively low K_d of 0.4 μM (Fig. 5).

The results of this study are not in complete agreement with those of Morgan and Wyndham (17), who reported that a tdtD mutant of Pseudomonas sp. A19-6a retained the ability to grow on DhA and AbA but exhibited similar decreases in removal rates for both substrates. However, since abietanes were not extracted from the cells in that study, abietanes sorbed to cells but not necessarily degraded would have been considered removed. It is also possible that A19-6a differs from BKME-9 in its complement of diterpenoid degradation enzymes in a way that does not allow for metabolism of AbA in the A19-6a tdtD mutant.

We hypothesize that the function of P450_dit is to hydroxylate DhA at C-7 (Fig. 1). In a previous study on abietane degradation by BKME-9, Martin and Mohn (14) showed that a ring-hydroxylating dioxygenase mutant, BKME-41, accumulated 7-oxo-DhA in cell suspension assays on DhA, PaA, or AbA. They also showed that the substrate for the ring-hydroxylating dioxygenase, DitA, required a ketone group at C-7, as DhA was not a substrate for the dioxygenase. The bacterial degradation of several natural plant products involves P450 monooxygenases that catalyze ring hydroxylation followed by oxidation of the hydroxyl group to a carbonyl. Some examples of this mechanism include the degradation of camphor involving P450_cam (20), the degradation of limonene involving the P450, limonene-6-hydroxylase, (28), and the recently reported degradation of cineole involving P450_cin (10). The metabolism of abietane diterpenoids appears to follow the same pattern, with P450_dit catalyzing the hydroxylation of DhA to 7-hydroxy-DhA before a further oxidation to 7-oxo-DhA.

Since 7-oxo-DhA is a metabolic intermediate of AbA and substrate binding assays indicate that AbA is not a substrate of P450_dit, how then is AbA transformed to 7-oxo-DhA? Possibly another pathway is used for AbA metabolism, involving another P450 which functions to hydroxylate AbA or one of its derivatives. The existence of an additional P450 that can partially complement P450_dit could also explain how the P450_dit mutant strain was able to grow, albeit slowly, on DhA and PaA. Further, the possibility of a second P450 is also consistent with sequence analysis of Burkholderia sp. LB400 (see below).

The results of this study and a previous one (14) are consistent with a mechanism of PaA degradation involving DhA as an intermediate that is subsequently hydroxylated at C-7 by P450_dit (Fig. 1). Martin and Mohn (14) showed an accumulation of DhA along with 7-oxo-DhA from PaA in a cell suspension assay of the ditA1 mutant. In this study, we observed an increase in DhA concentration during growth of the P450_dit mutant on PaA. The dit cluster contains several putative dehydrogenase genes, which could function in the formation of DhA from PaA. Substrate binding data strongly suggest that DhA is the better substrate for P450_dit, while PaA did not produce a typical substrate binding spectrum and does not appear to be a good substrate for this enzyme. Additional work, with a more pure PaA reagent, would lead to greater insight regarding this potential substrate.

This study confirms the relationship between the newly described 10.4-kbp extension of the dit cluster in BKME-9 and the tdt cluster of P. diterpeniphila A19-6a. These two sequences encode highly similar proteins and share the same gene arrangement (Fig. 2). Sequence alignment of the deduced amino acid sequences from the tdt cluster with the corresponding putative homologues in the dit cluster showed 72% or greater amino acid identity. We hypothesize that the P450 and the putative thiolase, dehydrogenase, hydrolase, hypothetical, regulator, and CoA ligase genes of the two organisms are functional homologues. Based on deduced amino acid sequence identity between tdtD and P450 _dit, the latter would constitute a second member of the new P450 superfamily proposed by Morgan and Wyndham (17).

Sequence comparison of the dit cluster with the recently sequenced Burkholderia sp. LB400 genome suggests that LB400 also contains homologues of dit cluster genes. With the exception of ditE, coding for a putative permease of the major facilitator superfamily, every protein encoded by the dit cluster (including the 10.4-kbp extension) has a putative homologue in a 60-kbp region of the LB400 genome (Fig. 2). Further, most of the genes are in small groups that have the same gene order as their putative homologues in BKME-9. An alignment of the deduced amino acid sequence shows high sequence identity between these deduced proteins of BKME-9 and LB400. Preliminary results indicate that LB400 can grow on DhA as a sole organic substrate (unpublished data). We are currently testing the hypothesis that this 60-kbp region of the LB400 genome codes for proteins that are required for diterpenoid degradation.

Interestingly, the genome sequence of LB400 provides additional evidence for the involvement of two P450s in diterpenoid metabolism. The above 60-kbp region in the LB400 genome includes two genes coding for putative cytochromes P450, Bcep5906, and Bcep5938 whose deduced protein products both have a high percent identity to P450_dit, relative to other P450 homologues in the databases (Fig. 2). Possibly, one of the two genes codes for a P450_dit homologue responsible for DhA/PaA degradation while a second codes for a second P450 responsible for AbA degradation. Other genes of interest in the 60-kbp region include (i) Bcep5887, with high sequence identity to ferredoxin reductase genes, (ii) Bcep5908, a ferredoxin gene homologue with similarity to those of P450 ferredoxins, and (iii) Bcep5919, a gene putatively coding for a methyl-accepting chemotaxis protein. Additionally, the most highly conserved genes shared between the dit cluster and the 60-kbp region of LB400 are the two encoding hypothetical proteins. High sequence conservation suggests that the gene products may perform an essential unknown function. Mutations are currently being generated in LB400 to investigate the functions of selected genes.

Figure 1 shows a proposed pathway for abietane diterpenoid metabolism in BKME-9. In this convergent scheme, PaA is transformed to DhA followed by hydroxylation at C-7 and further oxidation to form 7-oxo-DhA. This agrees with previous reports on resin acid degradation (14, 15), which showed the requirement of a carbonyl group at C-7 for DitA dioxygenase activity and showed the accumulation of 7-oxo-DhA during growth of a ditA1 knockout mutant on AbA, DhA, or PaA. In accordance with the results of this study, AbA is transformed to 7-oxo-DhA without the formation of DhA. Possibly a P450, other than P450_dit, is involved in this transformation, as suggested by the LB400 genome analysis. We are confident that DhA is the substrate for P450_dit; however, at this time we have not characterized the product of this reaction. We hypothesize that the product is 7-hydroxy-DhA.