Characterization of diverse natural variants of CYP102A1 found within a species of Bacillus megaterium

Materials

Isopropyl-β-D-thiogalactopyranoside (IPTG), glucose-6-phosphate, glucose-6-phosphate dehydrogenase, δ-aminolevulinic acid (δ-ALA), reduced β-nicotinamide adenine dinucleotide phosphate (NADPH), fatty acids, N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA), ferricyanide, phenacetin, acetaminophen, chlorzoxazone, coumarin, 7-ethoxycoumarin, and cytochrome c were obtained from Sigma-Aldrich (St. Louis, MO).

Bacterial strains

PCR and cloning of CYP102A1 natural variants

For DNA preparations, cells were grown in nutrient broth. After overnight growth at 37°C, the cells were centrifuged, washed, lysed, and enzymatically treated to remove RNA and protein. The DNA preparation was then treated with phenol-chloroform (50:50) and ethanol-precipitated. The purity was evaluated by measuring UV absorbance. The variant genes from B. megaterium were amplified by polymerase chain reaction (PCR) using oligonucleotide primers and B. megaterium chromosomal DNA template. First, PCR was carried out in a 50 μl reaction mixture containing template plasmid, forward primer BamHI-F (5'- AGCGGATCCATGACAATTAAAGAAATGCCTC-3') and reverse primer SacI-R (5'-ATCGAGCTCGTAGTTTGTAT-3'), dNTPs, and pfu polymerase. The PCR was carried out for 30 cycles consisting of 45 s of denaturation at 94°C, 45 s of annealing at 52°C, and 90 s of extension at 72°C. Next, PCR was carried out in a similar way by use of forward primer SacI-F (5'-ATACAAACTACGAGCTCGAT-3') and reverse primer XhoI-R (5'-ATCCTCGAGTTACCCAGCCCACACGTC-3'). The PCR product was digested with BamHI and SacI, and ligated into the pCW ori expression vector that had previously digested with the same restriction enzymes (Farinas et al. 2001). The amplified genes were subsequently cloned into the pCWBM3 BamHI/SacI vector at the BamHI/SacI restriction sites.

Because PCR amplification could lead to the introduction of random mutations and cloning of PCR products can fortuitously select the mutated sequences, all genes of CYP102A1 variants were PCR amplified a second time from genomic DNA and the sequences were directly determined without prior cloning. Exactly the same variations as those shown in Table 1 were again found, indicating that they were not artificially introduced during the PCR amplification.

Sequencing and phylogenetic analysis of 16S rRNA and ITS between 16s and 23s rRNA

The amplification of partial 16S rRNA genes was carried out using the primers 9F (5'-GAGTTTGATCCTGGCTCAG-3') and 1512R (5'-ACGGCTACCTTGTTACGACTT-3') (Ni et al. 2008). The amplification reaction (25 μl) contained 50 ng DNA, 0.50 μM of each primer, 250 μM dNTPs, 1.5 mM MgCl₂, and 1.25 U pfu polymerase in the buffer supplied by the manufacturer. The PCR was carried out for initial denaturation at 95°C for 5 min, followed by 30 cycles consisting of 95°C for 45 s, 55°C for 45 s, and 72°C for 90 s and final extension at 72°C for 10 min. Amplification products (10 μl) were electrophoresed in a 2% agarose gel and visualized under UV light after staining with ethidium bromide. Direct sequencing of the PCR products was performed with an ABI BigDye terminator v3.1 sequencing Ready Reaction kit.

One ITS region was amplified with primers 16S-F (5'- AAGTCGGTGGAGTAACCGT-3') and 23S-R (5'- TGTTAGTCCCGTCCTTCAT-3'). PCR reactions (25 μl) contained 50 ng DNA, 0.5 μM of each primer, 250 μM dNTPs, and 2.5 U Taq DNA polymerase in the buffer supplied by the manufacturer. The PCR was carried out for initial denaturation at 95°C for 15 min, followed by 35 cycles consisting of 95°C for 20 s, 52°C for 30 s, and 72°C for 60 s and final extension at 72°C for 3 min.

All sequencing procedures were repeated at least twice for each strain. The 16S rRNA gene sequences and the 16S-23S rRNA intergenic spacers were compared to sequences in the GenBank database using BLAST (Altschul et al. 1990). The sequences were aligned by using the CLUSTAL W program (Thompson et al. 1997).

Expression and purification of CYP102A1 natural variants

Plasmids were transformed into E. coli DH5αF'-IQ cell. Overnight cultures (20 ml) grown in Luria-Bertani broth with ampicillin (100 μg/ml) selection at 37°C were used to inoculate a 250 ml culture of Terrific broth containing 100 μg/ml ampicillin, 1.0 mM thiamine, trace elements, 50 μM FeCl₃, 1.0 mM MgCl₂, and 2.5 mM (NH₄)₂SO. Cells were grown at 37°C and 250 rpm to an OD₆₀₀ of between 0.6-0.8. Protein expression was induced by adding 1.0 mM IPTG and 1.5 mM δ-ALA, and cultures were grown at 28°C and 200 rpm for 50 h. The cells were harvested by centrifugation (15 min, 5,000 g, 4°C). The cell pellet was resuspended in TES buffer [100 mM Tris-HCl (pH 7.6), 500 mM sucrose, 0.5 mM EDTA] and lysed by sonication (Sonicator, Heat Systems - Ultrasonic, Inc.). After the lysate was centrifuged at 100,000 g (90 min, 4°C), the soluble cytosolic fraction was collected and used for the activity assay. The cytosolic fraction was dialyzed against 50 mM potassium phosphate buffer (pH 7.4) and stored at -80°C until use. The P450 concentration was determined by Fe²⁺-CO versus Fe²⁺ difference spectra (Omura and Sato 1964).

Binding affinity of fatty acids to CYP102A1 variants

To determine dissociation constants (K _d values) of fatty acids to the CYP102A1 variants, spectral binding titration was performed for enzymes with saturated fatty acids (lauric acid, myristic acid, and palmitic acid). The K _d values of substrates to the CYP102A1 variants were determined (at 23°C) by titrating 2.0 μM enzyme with the ligand, in a total volume of 1.0 ml of 100 mM potassium phosphate buffer (pH 7.4). The ligands were dissolved in dimethylsulfoxide and final dimethylsulfoxide concentrations were <1% (v/v). Absorbance increases at 390 nm and decreases at 420 nm as the substrate concentration increases (Lentz et al. 2001). The absorption difference between 390 nm and 420 nm was plotted against the substrate concentration (up to 1.0 mM) (Kim et al. 2008a, b). The K _d values were determined from plots of induced absorption changes versu s ligand concentration. The data were fitted using a standard hyperbolic function or (where the K _d value was within 5-fold of the P450 concentration) a quadratic function for tight-binding ligands, as described elsewhere (Girvan et al., 2010).

Assay of fatty acid hydroxylation by natural variants and distribution of hydroxylated products

Metabolites were generated by incubation of 1.0 mM fatty acids and P450 enzyme (100 pmol) in a 1.0 ml volume of 100 mM potassium phosphate (pH 7.4) for 20 min at 37°C (Gustafsson et al. 2004). An aliquot of a NADPH-generating system was used to initiate reactions; final concentrations were 10 mM glucose 6-phosphate, 0.5 mM NADP⁺, and 1 IU/ml yeast glucose 6-phosphate dehydrogenase. The reactions were terminated with a 2-fold excess of ice-cold dichloromethane. After centrifugation of the reaction mixture, the organic solvent was removed under a gentle stream of nitrogen and the residue was dissolved in BSTFA (50 μl) containing trimethylchorosilane (1%, v/v). The solution was transferred to a glass vial and incubated at 75°C for 20 min to yield trimethylsilylated products. To determine the regioselectivity of hydroxylated products of fatty acids at the ω-1, ω-2, and ω-3 positions, GC/MS analysis was carried out on a Shimadzu QP2010 (column length, 30 m; internal diameter, 0.25 mm; film thickness, 0.1 μm), with electron-impact ionization. The GC oven temperature was programmed for 1 min at 70°C followed by an increase to 170°C at 25°C/min, to 200°C at 5°C/min, and to 280°C at 20°C/min. The oven was finally held at 280°C for 5 min. The MS source and interface were maintained at 250 and 280°C, respectively, and a solvent delay of 4 min was used. The mass spectra were collected using electron ionization at 70 eV. The products were identified by their characteristic mass fragmentation patterns (Lentz et al. 2001). Turnover numbers of the hydroxylation of fatty acids (lauric acid, myristic acid, palmitic acid) by the variants of CYP102A1 were determined by a GC-FID detector (Shimadzu GC2010 with FID detector). Essentially the same procedure was used for the regioselectivity of the hydroxylated products of fatty acid oxidation. The distribution of products was based on the relative peak area of the chromatogram of GC using hydroxylated products at ω position as standards.

NADPH oxidation activities supported by natural variants

Reaction mixtures contained 1.0 mM fatty acid and P450 enzyme (25 nM) in a 1 ml volume of 100 mM potassium phosphate (pH 7.4). Initial rates of fatty acid-induced NADPH oxidation were measured by monitoring the absorption change at 340 nm (ε ₃₄₀ = 6,220 M^-1cm^-1) after NADPH was added at a concentration of 200 μM. Rates of change in A ₃₄₀ absorbance were converted into activity units (moles of NADPH oxidized per minute per mole of enzyme) (Noble et al. 1999).

Enzymatic activities of reductase domains of natural variants

For the reductase assay, two different types of reductase substrates were used. One was a chemical substrate, ferricyanide, and the other was cytochrome c, which is a protein substrate, as described previously (Gustafsson et al. 2004). Assays for reductase domain-dependent electron transfer to exogenous electron acceptors (ferricyanide or cytochrome c) were also performed at 37°C in potassium phosphate (pH 7.4), with 2.5 nM enzyme, 200 μM NADPH, and electron acceptors (500 μM ferricyanide; 100 μM cytochrome c). Ferricyanide reduction was measured at 420 nm (ε ₄₂₀ = 1.02 mM^-1cm^-1 for the ferricyanide reduction product) and cytochrome c reduction was measured at 550 nm (ε ₅₅₀ = 21.0 mM^-1cm^-1 for the reduced cytochrome c).

Thermal stability

To analyze enzyme stability, enzymes (2.0 μM) were incubated at different temperatures between 25 and 70°C for 20 min with subsequent cooling to 4°C in a PCR thermocycler (Eiben et al. 2007). The stability of the heme domain was calculated from heat-inactivation curves of CO-binding difference spectra (Omura and Sato 1964). The stability of the reductase domain was calculated from the reduction of ferricyanide catalyzed by reductase activity, as described above.

Catalytic activity assays towards human P450 substrates

Purified natural variants of CYP102A1 were characterized for human P450 enzyme activities using specific substrates as summarized elsewhere (Yun et al. 2006): phenacetin O-deethylation for human P450 1A2; 7-ethoxycoumarin (7-EC) O-deethylation for human P450s 1A2, 2A6, and 2E1; 7-ethoxy-4-trifluoromethylcoumarin (7-EFC) O-deethylation for P450s 1A2 and 2B6; chlorzoxazone 6β-hydroxylation for P450 2E1; coumarin 7-hydroxylation for P450 2A6.

Sequence analysis

DNA sequences of CYP102A1 variants, 16S rRNA sequences, and the ITS alleles between 16S and 23S rRNA genes obtained in this study were deposited in GenBank. The accession numbers are provided at Table 1. Genetic information of B. megaterium QM B1551 (ATCC 12872) regarding the CYP102A1 variant, 16S rRNA, and ITS was obtained from the homepage of Whole Genome Sequencing of B. megaterium http://www.bios.niu.edu/b_megaterium/.

The sequences were aligned using the MEGA 3.1 program (Molecular Evolutionary Genetic Analysis) (http://www.megasoftware.net/mega_dos.html). The size of CYP102A1 variants was 1,049 amino acids (Additional file 1). ITS (338 nucleotides) between 16S and 23S rRNA genes of B. megaterium was analyzed in this study. Phylogenetic trees were conducted by the neighbor-joining method using the MEGA 3.1 program. Bootstrap analysis of the neighbor-joining data, using 1,000 resamplings, was carried out to evaluate the validity and reliability of the tree topology.

Nucleotide sequence accession numbers

The nucleotide sequences determined in this study have been deposited in the GenBank database (Table 1): FJ859036, FJ899078, FJ899080 to FJ899082, FJ899084, FJ899085, FJ899091, and FJ899092 for CYP102A1 variants; FJ917385, FJ969746, FJ969749, FJ969751, and FJ969753 to FJ969764 for 16S rRNA genes of B. megaterium; FJ969765 to FJ969769, FJ969772, FJ969774, FJ969781, FJ969783 to FJ969787, FJ969792, FJ969794, FJ969795 for ITS of 16S-23S rRNA genes of B. megaterium.

Natural variants of CYP102A1 within a species of B. megaterium

Among 16 different strains of B. megaterium, 12 strains have natural genetic variants of CYP102A1 (Table 1). As some of them shared exactly the same DNA sequences, there were ultimately nine different types of CYP102A1 natural variants (Figure 1a, Table 1 and 2), including four previously reported variants (CYP102A1.1) (Ruettinger et al. 1989). Amino acid sequences of the CYP102A1 variants showed more than 96% identity with CYP102A1.1 (Table 2 and Additional file 1). The amino acid differences among the variants included 20 residues (CYP102A1.3, 20/1049, 1.9%) to 33 residues (CYP102A1.7, CYP102A1.8, CYP102A1.9; 33/1049, 3.1%) among a total of 1,049 amino acids (Table 2). Phylogenetic analyses of the amino acid sequences of CYP102A1 variants showed that three variants are closely related to CYP102A1.1 and five variants are distinct from it (Figure 1a). Among the total 55 mutated amino acid residues, those located in the reductase domains (residues 474-1049) (45 of 55, 82%) occurred at a much higher frequency than in heme domain (residues 1-473) (10 of 55, 18%) (Table 2). Interestingly, no substitutions in the amino acid residues of the active site or substrate channel (Ravichandran et al. 1993; Li and Poulos 1997) were seen among the 55 substitutions.

Phylogenic analysis of bacterial strains and natural variants

The 16S rRNA gene has been the molecular standard in studying evolutionary relationships among bacteria (Woese et al. 1990). Although DNA sequences of the 16S rRNA genes of 16 B. megaterium strains are well conserved (2 nucleotides are variable among a total of 1,394 nucleotides, 99.9% identity) (Figure 2a), the intergenic sequence (ITS_ alleles between 16S and 23S rRNA genes, which reflect the evolution of the bacterial strains (Gürtler 1999), showed 7 nucleotide variations among a total of 338 nucleotides (98.8% identity) (Figure 2b). Interestingly, the phylogenetic tree of ITS alleles was quite different from that of CYP102A1 natural variants. RNA analyses showed that the evolutionary profile of CYP102A1 variants is different from that of host strains (Figure 1).

Biochemical characterization of the natural variants

The biochemical properties of the variants were examined. All CYP102A1 variants could bind to saturated fatty acids in the range of 12-16 carbons with a general preference for long fatty acids (Figure 3a). The affinity of the variants to the fatty acids was quite different from that of CYP102A1.1 in the range of >50-fold for palmitic acid. However, the variations were less than 5-fold for lauric acid and myristic acid.

Although there were no apparent variations in hydroxylation activity towards myristic acid (C₁₄) and palmitic acid (C₁₆), the oxidation rates of lauric acid (C₁₂) by the variants varied in the range of >25-fold (Figure 3b). However, most of them did not show apparent changes in regioselectivity towards fatty acids (Additional file 2). For all fatty acids (C_12, C₁₄, C₁₆) tested here, there were no apparent variations of regioselectivity among a set of CYP102A1 variants. CYP102A1 variants showed a preference for hydroxylation at the ω-1 position of lauric acid, and myristic acid, and at the ω-2 position for palmitic acid. Fatty acid-dependent NADPH oxidation rates by the variants were also determined in the presence of lauric, myristic, and palmitic acids (Kitazume et al. 2007) (Figure 3c). We could not find a direct correlation between NADPH oxidation and product formation of hydroxylated fatty acids.

The reductase activity towards ferricyanide was quite dependent on the type of CYP102A1 variant (Additional file 3). Variant CYP102A1.3 showed a 3-fold higher activity than that of CYP102A1.1. In the case of cytochrome c, variant CYP102A1.2 had the highest activity, which was 3-fold higher than that of CYP102A1.1. These variations seem to be related to the variations in amino acid sequence.

Thermal stability of heme and reductase domains in the natural variants

The thermal stability of the heme and reductase domains was examined. The T ₅₀ value of the CYP102A1.1 heme domain was 51°C and the variants showed similar T ₅₀ values in the range of 51-55°C (Figure 4). The T ₅₀ value of the CYP102A1.1 reductase domain was 45°C and the T ₅₀ values of the variants' reductase domains were in the range of 40-48°C. CYP102A1.5 (T ₅₀, 48°C) showed the highest thermal stability among CYP102A1 variants. The thermal stabilities of the reductase domains were much lower than those of the heme domains of the CYP102A1 variants.

Catalytic promiscuity of the natural variants towards non-natural substrates

It is known that wild-type and several mutants of CYP102A1 could oxidize several human P450 substrates, including pharmaceuticals (Yun et al. 2007). We examined the catalytic promiscuity of the CYP102A1 variants towards non-natural substrates. They showed quite distinct catalytic activities towards typical human P450 substrates including drugs (Figure 5). CYP102A1.7 could oxidize all human P450 substrates tested here. Although the oxidation rates of the variants for all tested human P450 substrates were fairly low (< 0.4 min^-1), we detected potential evidence for the evolvability of P450 catalytic activities. Low catalytic activity is an intrinsic property of human P450 enzymes (Guengerich 2005). This result indicates that the variants show catalytic promiscuity towards non-natural substrates.