Characterization of Cupriavidus metallidurans CYP116B1 – A thiocarbamate herbicide oxygenating P450–phthalate dioxygenase reductase fusion protein


Kirsty J. McLean or Andrew W. Munro, Faculty of Life Sciences, Manchester Interdisciplinary Biocentre, University of Manchester, 131 Princess Street, Manchester M1 7DN, UK
Fax: 0044 161 306 8918
Tel: 0044 161 306 4194/2715
Tel: 0044 161 306 5151
Fax: 0044 161 306 8918


The novel cytochrome P450/redox partner fusion enzyme CYP116B1 from Cupriavidus metallidurans was expressed in and purified from Escherichia coli. Isolated CYP116B1 exhibited a characteristic Fe(II)CO complex with Soret maximum at 449 nm. EPR and resonance Raman analyses indicated low-spin, cysteinate-coordinated ferric haem iron at both 10 K and ambient temperature, respectively, for oxidized CYP116B1. The EPR of reduced CYP116B1 demonstrated stoichiometric binding of a 2Fe-2S cluster in the reductase domain. FMN binding in the reductase domain was confirmed by flavin fluorescence studies. Steady-state reduction of cytochrome c and ferricyanide were supported by both NADPH/NADH, with NADPH used more efficiently (Km[NADPH] = 0.9 ± 0.5 μm and Km[NADH] = 399.1 ± 52.1 μm). Stopped-flow studies of NAD(P)H-dependent electron transfer to the reductase confirmed the preference for NADPH. The reduction potential of the P450 haem iron was -301 ± 7 mV, with retention of haem thiolate ligation in the ferrous enzyme. Redox potentials for the 2Fe-2S and FMN cofactors were more positive than that of the haem iron. Multi-angle laser light scattering demonstrated CYP116B1 to be monomeric. Type I (substrate-like) binding of selected unsaturated fatty acids (myristoleic, palmitoleic and arachidonic acids) was shown, but these substrates were not oxidized by CYP116B1. However, CYP116B1 catalysed hydroxylation (on propyl chains) of the herbicides S-ethyl dipropylthiocarbamate (EPTC) and S-propyl dipropylthiocarbamate (vernolate), and the subsequent N-dealkylation of vernolate. CYP116B1 thus has similar thiocarbamate-oxidizing catalytic properties to Rhodoccocus erythropolis CYP116A1, a P450 involved in the oxidative degradation of EPTC.


carbon monoxide


cytochrome P450 reductase


Pseudomonas putida camphor hydroxylase P450cam or CYP102A1


CYP116B1, cytochrome P450 (116B1) from Cupriavidus metallidurans CH34


distilled, deionized water


S-ethyl dipropylthiocarbamate






imidazolyl decanoic acid


imidazolyl undecanoic acid


imidazolyl dodecanoic acid


isopropyl thio-β-d-galactoside


liquid chromatography


multi-angle laser light scattering


normal hydrogen electrode


nitric oxide


cytochrome P450 or CYP

P450 BM3

cytochrome P450 BM3 from Bacillus megaterium or CYP102A1


phthalate dioxygenase


Burkholderia cepacia phthalate dioxygenase reductase


size-exclusion chromatography


S-propyl dipropylthiocarbamate


The cytochromes P450 (P450s) are a superfamily of cysteine thiolate-coordinated haem b-binding enzymes widespread in nature [1]. The vast majority of P450s catalyse the reductive scission of molecular oxygen bound to the haem iron, resulting in monooxygenation of a substrate molecule bound close to the haem [2]. A highly reactive iron-oxo species (compound I) is considered to be the major oxidant in P450-mediated substrate oxidations. Only recently has the first compelling evidence been presented for the formation and the reactivity of compound I. Using the P450 CYP119 from the thermophilic bacterium Sulfolobus acidocaldarius, Rittle and Green demonstrated (using EPR, Mössbauer and UV-visible absorption spectroscopy) characteristic spectra near-identical to those observed for compound I species in peroxidase hemoproteins, and an apparent second-order rate constant for oxidation of unactivated hydrocarbons of 1.1 × 107 m−1·s−1 for the CYP119 compound I species [3]. However, other iron-oxo species in the P450 catalytic cycle are also potential substrate oxidants in selected P450 transformations. For instance, the ferric hydroperoxo species (compound 0) that precedes compound I is also a potential oxidant (e.g. for epoxidation across carbon–carbon double bonds, or for sulfoxidation), albeit with conflicting evidence for its general catalytic relevance emerging from experimental and computational studies [4–6]. There is a broad spectrum of substrate specificity seen across the P450 enzyme superfamily, and a wide array of oxidative transformations are catalysed (e.g. hydroxylation, epoxidation, demethylation, S- and N-oxidation and C-C bond cleavage) [1,2,6]. Various P450 enzymes can also catalyse reductive chemistry, in addition to isomerizations and dehydrations [e.g. 2,7,8].

P450-mediated substrate oxygenation requires the delivery of two consecutive electrons to a substrate-bound, ferric enzyme. These events achieve reduction of the iron, first to the ferrous form (which binds dioxygen) and then to the ferric peroxy state [1,2]. Successive protonations of the latter intermediate produce first compound 0 and then (with water loss) the transient and highly reactive compound I [3]. Electrons are typically derived from NAD(P)H and are delivered to the P450 by iron-sulfur and/or flavoproteins. Historically, two major classes of P450 redox partners were thought to exist. The first (class 1) is typified by the Pseudomonas putida camphor hydroxylase P450cam (CYP101A1), to which electrons are shuttled from NADH through the FAD-binding putidaredoxin reductase and the iron–sulfur (2Fe-2S) cluster-binding putidaredoxin [9,10]. The second system (class 2) is exemplified by the mammalian hepatic P450 enzymes (that are pivotal to drug and xenobiotic metabolism) and involves membrane-bound P450s and a diflavin reductase [cytochrome P450 reductase (CPR)] that sources its electrons from NADPH [11]. However, more recent studies of diverse P450 systems, particularly from microorganisms, have demonstrated that the repertoire of P450 redox partners is much broader than was first considered [12] and that certain P450s have disposed altogether with redox partner proteins and interact directly with NADH or hydrogen peroxide to facilitate catalysis [13,14]. Also, other P450s exist as natural fusions to both CPR-like and other forms of redox partners [12,15].

The first major class of P450–redox partner fusion enzymes characterized were the CYP102A family, headed by the well characterized Bacillus megaterium P450 BM3 (BM3, CYP102A1) [16]. This is a cytoplasmic enzyme in which a fatty acid hydroxylase P450 (N terminal) is fused to an NADPH-dependent CPR [17]. P450 BM3 has the highest reported rate of substrate oxidation for any P450 oxygenase (∼ 17 000 min−1 with arachidonate), facilitated by the efficient electron transport system operating both within its CPR domain and between the CPR and the P450 [18,19]. Members of the CYP102A family are widespread in Bacillus and in other bacteria, and a similar type of P450–CPR fatty acid hydroxylase fusion is also found in lower eukaryotes (e.g. Fusarium oxysporum P450foxy, CYP505A1) [20,21]. P450 BM3 has been the subject of intensive studies aimed at understanding its structure and mechanism, with recent studies demonstrating that the dimeric form of the enzyme is catalytically functional in fatty acid oxidation, with electron transfer between monomers probably supporting its catalytic function [22–24]. Moreover, as a consequence of its catalytic efficiency and the availability of good crystal structure data, P450 BM3 has been extensively engineered in order to explore its structure–function relationships and to enable oxidation of novel substrates such as short-chain fatty acids, alkanes and steroids [25–28].

A distinctive type of P450–redox partner fusion system was identified from genome analysis of pathogenic Burkholderia species, the heavy metal-tolerant bacterium Cupriavidus metallidurans CH34 and Rhodococcus sp. NCIMB 9784 [29]. In this system, P450s of undefined structure and substrate selectivity (classified in the CYP116B P450 family) are fused to a reductase module with an amino-acid sequence resembling that of Burkholderia cepacia phthalate dioxygenase reductase (PDOR), an enzyme that provides electrons to phthalate dioxygenase (PDO) to enable this enzyme to oxygenate phthalate in a pathway for its degradation and exploitation as a carbon source for growth [30,31]. As with P450 BM3, the P450 module in the novel CYP116–PDOR fusion enzymes is at the N terminus of the fusion protein (Fig. 1). The Rhodococcus enzyme from this family (CYP116B2) was expressed and purified, and shown to exhibit P450-like spectral characteristics. Genetic dissection of the enzyme enabled characterization of FMN and 2Fe-2S centres in the expressed PDOR module [32,33]. CYP116B2 was also shown to have weak activity with a prototypical fluorescent substrate for the P450 superfamily (7-ethoxycoumarin). However, no physiologically relevant substrate was identified [32].

Figure 1.

 Schematic showing domain organization and electron flow in CYP116B and CYP102A (P450 BM3) enzymes. (A) The genetic arrangement of the individual domains (indicated by cofactor content: haem [P450] in the P450 domain, and FMN and 2Fe-2S [Fe-S] in the reductase domain) in CYP116B1 is shown at the left. The diagram at the right shows the direction of electron flow from NADPH through FMN, iron-sulfur and onto the P450 haem within a CYP116B1 monomer. (B) A similar diagram to (A) of genetic arrangement for P450 BM3 (CYP102A1) is shown, with FMN and FAD cofactors in the reductase domain. The diagram on the right illustrates electron flow within the P450 BM3 dimer. Electron transfer occurs from NADPH to FAD and then to FMN in monomer 1. Thereafter, electrons are transferred from FMN of monomer 1 to the haem in monomer 2 [22]. In the electron-transfer diagrams, the flow of electrons from NADPH is indicated by large blue arrows. The haem domains are represented by pink circles with the haem shown in red; the iron-sulfur domain is in a coral circle with the iron-sulfur cluster in red and yellow spacefill; and the FMN and FAD domains are in orange and yellow circles, respectively, with the isoalloxazine ring portions shown in the centre. CYP116B1 is represented as a monomer, while one monomer of the P450 BM3 dimer is shown coloured as described above, while the second monomer is shown in greyscale.

In this manuscript we present the first report of the cloning, expression and detailed kinetic, spectroscopic, hydrodynamic and thermodynamic analysis of the cofactor-replete C. metallidurans P450–PDOR fusion enzyme (CYP116B1). The enzyme exhibits characteristic haem Soret spectral properties for the P450 class, but displays some notable differences to features observed for CYP116B2. In this work, light scattering studies showed CYP116B1 to be monomeric and to bind polyunsaturated lipids with an induced shift in haem iron spin-state equilibrium. We showed that CYP116B1 catalyses NADPH-dependent oxygenation of thiocarbamate herbicides, an activity consistent with its evolutionary relationship with a distinct (non-partner fused) P450 enzyme (CYP116A1, also known as ThcB) from Rhodococcus erythropolis NI86/21 that oxidatively degrades the thiocarbamate herbicide S-ethyl dipropylthiocarbamate (EPTC) [34].


Expression, purification and UV-visible absorption features of CYP116B1

The gene encoding the C. metallidurans CH34 CYP116B1 P450 (gene 4932, UniProt Q1LDI2) was cloned by PCR from bacterial genomic DNA and expressed in Escherichia coli (Origami (DE3)) using a pET11a construct. CYP116B1 was purified using a combination of ion-exchange and hydrophobic affinity chromatography, producing a single major band on SDS/PAGE (Fig. 2A, inset), consistent with the predicted 86.4-kDa molecular mass of the P450–PDOR fusion protein. The UV-visible absorption properties of the pure CYP116B1 in its oxidized, sodium dithionite-reduced and reduced/carbon monoxide (CO)-bound forms are illustrated in Fig. 2A. For oxidized CYP116B1, the spectral features of a haem cofactor are clearly evident. The main Soret band has its absorption maximum at 418 nm, with longer-wavelength alpha/beta bands positioned at 566 and 532 nm, respectively. A pyridine haemochromagen assay to determine the haem concentration of CYP116B1 was performed in triplicate and a CYP116B1-specific extinction coefficient was determined to be ε418 = 121 mm−1·cm−1 at the Soret peak of the oxidized enzyme. This value includes the spectral contributions from iron-sulfur and flavin centres in the cofactor-replete CYP116B1, and was used to determine the CYP116B1 concentration in all further experiments.

Figure 2.

 UV-visible absorption spectroscopy of CYP116B1. (A) UV-visible absorption spectra for CYP116B1 (2.9 μm) showing the purified oxidized enzyme (thick solid line) with Soret maximum at 418 nm, the dithionite-reduced enzyme (dashed line) and the reduced/CO-bound adduct (thin solid line) showing the Fe(II)CO peak at 449 nm with a minor peak at 420 nm from the (protonated) cysteine thiol-bound form. Inset: SDS/PAGE gel (10%) of CYP116B1 showing the purified enzyme at 86.4 kDa at three different concentrations of 1, 0.5 and 0.1 μg (left to right), sizes of marker bands indicated. (B) Complexes of CYP116B1 with small-molecule inhibitors. Spectra shown are for oxidized CYP116B1 (2.3 μm, thick solid line); NO-bound CYP116B1 (thin solid line), showing a Soret intensity decrease and shift to 433 nm with Q-bands at 574 and 545 nm; cyanide (CN)-bound (50 mm KCN) CYP116B1 (dotted line), showing a Soret intensity decrease and shift to 430 nm; and imidazole (2 mm)-bound CYP116B1 (dashed line), exhibiting a Soret shift to 424 nm with a small decrease in intensity. Inset: CYP116B1-imidazole binding curve generated from difference spectral data (as described in the Materials and methods), yielding a Kd value of 646 ± 13 μm.

On aerobic reduction with sodium dithionite, the Soret band decreased in intensity and was slightly blue-shifted to 417 nm. The less intense haem alpha and beta bands remained at ∼ 566 and 532 nm, respectively. The extent of Soret absorbance shift was less than that observed for other P450s (e.g. the P450 BM3 Soret shifted to ∼ 409 nm on full reduction to the ferrous, thiolate-coordinated state) [35,36]. The less-extensive Soret shift for CYP116B1 is a consequence of only partial reduction of the haem iron by dithionite under aerobic conditions, which, in turn, is a result of the rapid reaction of its ferrous haem iron with dioxygen and its subsequent re-oxidation with production of superoxide. Anaerobic titration of CYP116B1 with dithionite led to complete reduction of the haem iron and a more extensive Soret shift (see the section ‘Examination of the thermodynamic properties of CYP116B1’). However, the simultaneous drop in absorption intensity in the region beyond ∼ 425 nm suggests that the major cause of the apparent Soret absorption decrease is actually the reduction of, and concomitant changes in the spectral contributions of, the other cofactors bound to the protein. The reductase (PDOR) domain (amino acids ∼ 445–780 of the 780-residue CYP116B1) is expected to bind both flavin (FMN) and iron–sulfur cofactors, by homology with both the B. cepacia PDOR and the reductase domain of Rhodococcus sp. NCIMB9784 CYP116B2 [33]. Reductive bleaching of FMN, and 2Fe-2S cofactor absorption by dithionite thus underlies the decreased Soret intensity and the absorption bleaching observed beyond ∼ 425 nm.

In view of the differing coenzyme selectivity of the B. cepacia PDOR (which favours NADH) and the Rhodococcus CYP116B2 (which favours NADPH) [30,32], we also analyzed spectral changes on aerobic mixing of oxidized CYP116B1 with NADH and NADPH (50 μm). Interestingly, spectral changes similar to those observed with dithionite were seen using NADPH as reductant, but NADH only effected partial reduction of the FMN/iron–sulfur centres (data not shown). In view of our subsequent work (vide infra), which shows that both NADPH and NADH support CYP116B1 catalytic activity, it is likely that this reflects faster enzyme reoxidation (and/or slower reduction) for NADH-reduced CYP116B1 compared with the NADPH-reduced form. This conclusion is consistent with stopped-flow studies reported later, under ‘Steady-state and stopped-flow kinetic studies on CYP116B1’.

Addition of CO to the dithionite-reduced CYP116B1 resulted in a shift of the Soret peak to 449 nm, with a much smaller feature at ∼ 424 nm. The former band is the characteristic signature for the cysteine thiolate-ligated P450 haems in their Fe(II)CO complexes. The minor feature at ∼ 424 nm probably reflects a small proportion of the (protonated) cysteine thiol-coordinated P420 form. Protonation of the proximal ligand can arise as a consequence of either haem reduction or the binding of CO to ferrous haem iron, or both [37,38]. Other spectroscopic studies indicate that the cysteine thiolate-coordinated state is predominant in the oxidized enzyme (see ‘Characterization of CYP116B1 haem and iron–sulfur clusters by EPR’). In the Fe(II)CO complex of CYP116B1, there is a single absorption Q-band peak in the visible region at ∼ 549 nm, indicative of complete conversion of the haem iron to the ferrous state (Fig. 2A).

To further define the spectral features of CYP116B1, we generated ligand complexes using the small-molecule inhibitors imidazole, cyanide and nitric oxide (NO) (Fig. 2B). All three molecules induced red (type II) shifts in the Soret band, which is typical for haem iron-coordinating P450 inhibitors. The imidazole complex had its Soret band shifted to 423.5 nm with a small decrease in intensity. An optical binding titration with imidazole revealed a hyperbolic dependence of induced absorption change on ligand concentration and a Kd of 646 ± 13 μm (Fig. 2B, inset). The Soret shift from ∼ 418 nm to 423.5 nm on complexation of imidazole and triazole inhibitors is typical of behaviour seen in other P450s (e.g. 38,39). However, it was reported that purified Rhodococcus CYP116B2 has its Soret maximum at ∼ 424 nm in the ligand-free state [32]. Given that this enzyme was purified by imidazole elution from nickel resin, it is probably the case that the CYP116B2 form purified was the imidazole complex. The CYP116B1 complex with (sodium) cyanide resulted in a Soret shift to 431 nm and a substantial decrease in peak intensity (as also observed, for example, in the Mycobacterium tuberculosis sterol demethylase CYP51B1) [38]. Optical titration indicated a Kd value of 9.7 ± 0.6 mm for cyanide. Finally, binding of the gaseous ligand NO led to a Soret shift to 433 nm, with a distinctive sharpening of features in the visible region (Q-bands at 574 and 545 nm), and a spectrum comparable to the NO-bound form of P450 BM3 (Fig. 2B) [40].

The binding of P450 substrate molecules often induces spectral shifts in the Soret region. Blue shifts (towards ∼ 390 nm) are observed when substrates cause displacement of an aqua ligand bound weakly to the sixth coordination position on the haem iron, as seen, for example, for camphor binding to P450cam (CYP101A1) and for fatty acid binding to P450 BM3 [41,42]. To identify potential substrate molecules for CYP116B1, we undertook extensive binding trials with a diverse range of prototypical P450 substrates, including fatty acids (e.g. lauric acid and myristic acid), steroids and other cyclic molecules (e.g. testosterone, hydrocortisone, 7-ethoxycoumarin and 6-deoxyerythronolide B). While there were no significant optical changes observed with the saturated fatty acids tested (lauric acid [C12], myristic acid [C14], pentadecanoic acid [C15] and palmitic acid [C16]), small type-I spectral shifts (typically 2–3 nm, spin-state change of ∼ 4–10%) were seen with a number of monounsaturated fatty acids (myristoleic acid [C14], palmitoleic acid [C16] and oleic acid [C18]), and with the polyunsaturated C20 arachidonic acid. The low solubility of oleic acid in aqueous buffer did not allow for affinity to be determined by optical titration. However, this was feasible for the three other lipids, and data for induced optical change versus fatty acid concentration were fitted using a hyperbolic function to generate Kd values of 4.8 ± 1.7 μm (myristoleic acid), 33.5 ± 6.3 μm (palmitoleic acid) and 5.6 ± 1.3 μm (arachidonic acid). Data are shown for myristoleic acid in Fig. 3A. Given the successful demonstration of binding of fatty acid molecules to CYP116B1, we also investigated whether fatty acid-linked imidazole inhibitors (imidazolyl decanoic acid [ImC10], imidazolyl undecanoic acid [ImC11] and imidazolyl dodecanoic acid [ImC12]) could act as potent inhibitors of P450. In previous studies we demonstrated tight binding of ImC10, ImC11 and ImC12 to the P450 BM3 fatty acid oxygenase [43]. These molecules also bound tightly to CYP116B1 and induced type-II spectral changes (Soret shifts to 424 nm), consistent with haem iron coordination. The Kd values obtained from fitting optical-titration data using a quadratic equation for tight-binding ligands were 3.5 ± 0.4 μm (ImC10), 0.59 ± 0.12 μm (ImC11) and 0.52 ± 0.11 μm (ImC12) [44]. Data for the binding of ImC12 to CYP116B1 are shown in Fig. 3B. These Kd values compare favourably to those determined for binding of ImC10, ImC11 and ImC12 to P450 BM3 (0.9, 7.5 and 1.35 μm, respectively), and with data for binding to a rat CYP4A1–CPR fatty acid hydroxylase fusion protein (12.4, 0.37 and 0.35 μm, respectively) [43,45].

Figure 3.

 Substrate- and inhibitor-type binding in CYP116B1. (A) Binding curve for myristoleic acid-induced spectral change in CYP116B1 (2.0 μm). Data were fitted using a quadratic function [44], generating a Kd value of 4.8 ± 1.7 μm for myristoleic acid. Inset: difference spectra showing the progressive substrate-like Soret absorbance shifts to shorter wavelength (blue shifts) upon addition of myristoleic acid at 7.96, 15.92, 23.88, 47.76 and 79.60 μm. (B) Binding curve for ImC12-induced spectral change in CYP116B1. Data were fitted as described for (A), producing a Kd value of 0.52 ± 0.11 μm. Inset: UV-visible absorbance spectra for CYP116B1 in ligand-free form (2.1 μm, solid line) and with ImC12 at a near-saturating concentration (5.13 μm, dashed line). There is a shift in the Soret peak maximum from 418 to 424 nm on binding ImC12.

In view of the close relationship (52% amino-acid sequence identity) of the CYP116B1 haem domain with the R. erythropolis CYP116A1 (also known as ThcB, a non-fused P450 shown to catalyse the N-dealkylation of EPTC [34]), we also analysed the binding of EPTC and the related thiocarbamate herbicide S-propyl dipropylthiocarbamate (vernolate) to CYP116B1. Very small type-I haem perturbations were observed, but these were not large enough to enable accurate determination of Kd from optical binding. However, we were able to show that CYP116B1 catalyses the oxidation of both vernolate and EPTC (see ‘Identification of CYP116B1-catalysed reaction products of thiocarbamate herbicides’).

Multi-angle laser light scattering analysis of CYP116B1

The P450 the BM3 P450-CPR fusion enzyme was shown to be catalytically functional in its dimeric state, with electron transfer between the FMN of monomer 1 and the haem of monomer 2 probably supporting its fatty acid hydroxylase activity (Fig. 1) [22,24]. In order to analyze the aggregation state of the P450–PDOR fusion enzyme (CYP116B1) in solution, we performed size-exclusion chromatography (SEC) coupled to multi-angle laser light scattering (MALLS). In contrast to P450 BM3, CYP116B1 was found to be predominantly monomeric in solution, eluting from a size-exclusion column in a single band with an apparent average molecular mass of 86.1 ± 1.7 kDa, very close to that predicted for the intact enzyme based on its protein sequence (86.4 kDa) (Fig. 4). Steady-state kinetic assays (see ‘Steady-state and stopped-flow kinetic studies on CYP116B1’) revealed no discernible CYP116B1 concentration dependence on the specific rate of vernolate-induced NADPH oxidase activity. This is consistent with a single (monomeric) state of CYP116B1 and contrasts with the decreased specific activity of P450 BM3 at low enzyme concentrations that occurs as a consequence of dissociation of the dimer into inactive monomers [22]. It thus appears likely that CYP116B1 (and probably also CYP116B2 and related CYP116B family enzymes) operate by intramolecular electron transfer within a monomer, as opposed to intermolecular electron transfer observed for P450 BM3 (and the eukaryotic l-arginine oxidase nitric oxide synthase flavocytochromes) [22,46].

Figure 4.

 SEC-MALLS analysis of CYP116B1. The MALLS profile is shown for a pure CYP116B1 sample resolved by SEC on a Superdex 200 gel-filtration column before passing directly through MALLS and refractive index (RI) detectors (the RI trace is shown as a solid line). The number average molar mass (Mn) across the single peak resolved (filled blue circles) was determined as 86.1 ± 1.7 kDa. These data are consistent with CYP116B1 being a monomeric enzyme.

Characterization of CYP116B1 haem and iron–sulfur clusters by EPR

EPR spectra for the oxidized and sodium dithionite-reduced forms of pure CYP116B1 were collected as described in the Materials and methods. For oxidized CYP116B1, a rhombic spectrum consistent with a ferric cysteinate-coordinated haem iron was observed, with g-values of 2.42 (gz), 2.25 (gy) and 1.93 (gx) (Fig. 5). These g-values lie in the range typical for other low-spin P450 enzymes [e.g. 38,47,48], and are consistent with an essentially completely low-spin P450 enzyme. There were no significant signals in the region associated with the high-spin P450 haem iron (e.g. g = 8.18/7.85, 3.44/3.97 and 1.66/1.78 for P450 BM3/P450cam in substrate-bound forms [48,49]), suggesting that ferric CYP116B1 is hexacoordinated with a water ligand trans to the cysteine thiolate. Similar sets of g-values were obtained for the low-spin forms of other bacterial P450s. These include P450 BM3 and P450 BioI from Bacillus subtilis (CYP107H1), whose EPR spectra yield g-values of 2.42, 2.26 and 1.92, and 2.41, 2.25 and 1.92, respectively [47,48].

Figure 5.

 X-Band EPR spectra for CYP116B1. Upper spectrum: oxidized CYP116B1 (184 μm) showing a rhombic signal with g-values typical for a low-spin, six-coordinate ferric haem iron at gz = 2.42, gy = 2.25 and gx = 1.93. Lower spectrum: CYP116B1 (184 μm), reduced anaerobically by dithionite, showing almost complete loss of the signal from the ferric haem species (as haem is reduced to the EPR-silent ferrous state) and the emergence of bands at gz = 2.06, gy = 1.97 and gx = 1.88, typical of a plant-like 2Fe-2S cluster in the EPR active [2Fe-2S]+ form. No significant signal is observed for a flavin semiquinone.

For oxidized CYP116B1, there were no apparent EPR signals that could be ascribed to either iron–sulfur or flavin centres that are likely to be present in the PDOR domain of the enzyme. Oxidized flavins are EPR silent, and the 2Fe-2S cluster in its oxidized form (i.e. [2Fe-2S]2+) should contain two ferric iron atoms, in which the magnetic moments of the unpaired electrons are spin-coupled, giving S = 0. However, reduced 2Fe-2S clusters are EPR active in the [2Fe-2S]+ form, with one ferric and one ferrous iron atom [e.g. 50,51]. An EPR spectrum was collected for the dithionite-reduced form of CYP116B1 and is shown in Fig. 5. The spectrum is more compact than that for the oxidized enzyme with the major features appearing between 3250 and 3750 Gauss. The reduced CYP116B1 spectrum has a rhombic shape and g-values of gz = 2.06, gy = 1.97 and gx = 1.88 that fall within the typical ranges reported by More et al. [52], clearly identifying the presence of a reduced 2Fe-2S cluster. There is no EPR signal that could be clearly assigned to a flavin semiquinone in the reduced CYP116B1. The EPR spectrum for the 2Fe-2S cofactor-reconstituted PDOR (2Fe-2S- and FMN-binding) domain of CYP116B2 from Rhodococcus sp. NCIMB 9784 produces an identical set of g-values [33]. The CYP116B1 2Fe-2S EPR spectrum is also highly similar to that for the homologous B. cepacia PDOR (g = 2.04, 1.95 and 1.90), and to that of the Arabidopsis photosynthetic ferredoxin Fd2 (g = 2.05, 1.95, 1.89) [53,54]. Integration of the EPR signals for the oxidized haem and for the reduced iron–sulfur cofactors in CYP116B1 indicate that the 2Fe-2S centre is present at ∼ 98% of the concentration of the haem, thus suggesting near-complete incorporation of the iron-sulfur cluster in CYP116B1.

To identify the flavin bound to CYP116B1, we isolated the bound cofactor from a purified enzyme sample, as described in the Materials and methods. Thereafter, flavin fluorescence was determined at both pH 7.7 and pH 2.0, as described previously [55], and compared with FAD and FMN standards treated in the same way. Decreased flavin fluorescence was observed for both the CYP116B1 cofactor extract and for FMN standards on acidification, while enhanced fluorescence was seen for FAD (owing to hydrolysis of its pyrophosphoryl bond and release of ADP and the more highly fluorescent FMN). FMN content was estimated at ∼55% by comparison with the quantity of bound haem, consistent with the previous report of variable FMN content in CYP116B2 [33]. The inclusion of free FMN (1 mm) in buffers used for CYP116B1 purification enabled increased FMN content in the purified enzyme (typically to > 80%).

Steady-state and stopped-flow kinetic studies on CYP116B1

Analysis of the ability of CYP116B1 to reduce electron acceptor molecules (potassium ferricyanide, cytochrome c) in a NAD(P)H-dependent manner and to perform P450 substrate-dependent oxidation of NAD(P)H was examined in steady-state reactions (Table 1). There was a strong preference for NADPH over NADH, with a difference of ∼ 440-fold in Km values for NADPH (0.9 μm) and NADH (399.1 μm) determined in cytochrome c-reduction studies. The kcat values for both ferricyanide and cytochrome c reduction were also ∼three-fold greater with NADPH as the coenzyme than with NADH (e.g. 833.1 min−1 versus 286.8 min−1 for ferricyanide), resulting in a 99-fold difference in catalytic efficiency for ferricyanide reduction (measured as the ratio of kcat/Km[NADPH] to kcat/Km[NADH]) and a 1174-fold difference for cytochrome c reduction in favour of NADPH (Table 1). CYP116B2 also demonstrated preference for NADPH (Km = 3.0 μm) over NADH (102.1 μm) in steady-state ferricyanide-reduction assays [32]. In contrast, the B. cepacia PDOR flavoprotein has strong preference for NADH (Km ∼ 10 μm) over NADPH, and NADPH did not support measurable activity with either cytochrome c or the partner protein PDO, pointing to an important evolutionary divergence between the stand-alone PDOR enzyme and the fused PDOR domains in the CYP116B enzymes.

Table 1.   Steady-state kinetic properties of CYP116B1. Km and kcat values for CYP116B1 were determined from steady-state kinetic assays with NADH and NADPH as reductants, and using the electron acceptors potassium ferricyanide (FeCN) and cytochrome c, and the P450 substrates vernolate and EPTC. Km values shown are for NAD(P)H in the experiments using cytochrome c and FeCN, and for vernolate/EPTC in the experiments using the thiocarbamates. In experiments using the thiocarbamate substrates, NADH (1.5 mm) and NADPH (200 μm) were used at near-saturating concentrations. All assays were performed on a dual-beam spectrophotometer (Jasco) in buffer A at 25 °C using CYP116B1 at a final concentration of 75–100 nm
Acceptor or substrateReductant
Cytochrome c
 Kmm)0.9 ± 0.5399.1 ± 52.1
 kcat (min−1)151.1 ± 8.057.1 ± 2.7
 Kmm)3.0 ± 1.0102.1 ± 12.2
 kcat (min−1)833.1 ± 17.8286.8 ± 7.5
 Kmm)210.2 ± 30.7270.4 ± 58.8
 kcat (min−1)57.3 ± 1.934.0 ± 2.2
 Kmm)50.3 ± 9.6110.4 ± 10.0
 kcat (min−1)30.7 ± 0.517.0 ± 0.4

In view of the preceding data indicating substrate-like (type I) Soret spectral shifts induced on binding both unsaturated fatty acids and (to a lesser extent) the thiocarbamate herbicides EPTC and vernolate, we analysed steady-state oxidation of NADPH by CYP116B1 in the presence of these potential substrates. Despite favourable haem optical changes induced by myristoleic, palmitoleic, oleic and arachidonic acids, none of these molecules stimulated CYP116B1-dependent oxidation of either NADH/NADPH to any significant levels. By contrast, substantial stimulation of oxidation of both coenzymes was observed with EPTC and vernolate as substrates. A hyperbolic dependence of coenzyme oxidation rate on EPTC concentration was obtained using both NADH and NADPH, and data were fitted using the Michaelis function to produce values of kcat = 30.7 min−1 and Km EPTC = 50.3 μm (with NADPH as the donor) and kcat = 17.0 min−1 and Km EPTC = 110.4 μm (with NADH as the donor). For vernolate, the comparable values were kcat = 57.3 min−1 and Km vernolate = 210.2 μm (using NADPH as the donor) and kcat = 34.0 min−1 and Km vernolate = 270.4 μm (using NADH as the donor) (Table 1). Thus, EPTC (a substrate that is dealkylated in whole cells of R. erythropolis that express CYP116A1) clearly stimulates NAD(P)H oxidation in CYP116B1, and the related dithiocarbamate vernolate enhances NAD(P)H oxidation to an even greater extent. In the absence of vernolate, NADPH-dependent reduction of the CYP116B1 haem iron is slow in a CO-saturated solution (∼ 0.2 min−1) and results mainly in the P420 form of the enzyme. In the presence of saturating vernolate, there is some stimulation of the rate of haem iron reduction, but data analysis is complicated by a greater tendency of the enzyme to aggregate with development of turbidity in the solution. Despite the progressive aggregation of vernolate-bound CYP116B1 in this experiment, it is notable that the predominant species is now the P450 form, indicating substrate-dependent stabilization of the thiolate ligand, as observed previously in other systems [37,38,56]. Given the limited solubility of vernolate, its marginal effect on haem iron spin-state equilibrium and its influence on aggregation of CYP116B1, it is difficult to establish the extent to which electron transfer to the CYP116B1 haem iron is thermodynamically gated in the presence of this substrate and a kinetic gate to haem reduction in CYP116B1 cannot be ruled out [57]. Consistent with the monomeric form of CYP116B1 being the catalytically relevant species, specific rates of vernolate-dependent NADPH oxidation, determined across a range of CYP116B1 concentrations (10–500 nm), showed no significant variation. In contrast, similar studies with P450 BM3 and the substrate lauric acid showed diminished specific activity at low enzyme concentration as a result of disassociation of the catalytically active dimer [22].

To further characterize the reaction of CYP116B1 with NAD(P)H, stopped-flow absorption spectroscopy was used to analyse the kinetics of reduction of CYP116B1 by NADH and NADPH. Absorption transients were collected at 465 nm (i.e. near the expected absorption maximum for the oxidized FMN cofactor [33]) after mixing CYP116B1 with different concentrations of NAD(P)H. Reaction transients were biphasic in all cases and were fitted accurately using a double exponential function. Approximately 80% (NADH) and 90% (NADPH) of the overall absorption change observed on coenzyme-dependent reduction of the FMN/2Fe-2S centres occurred in the fast phase of the reaction transients. The CYP116B1 haem was not reduced by NAD(P)H to any extent in its substrate-free form. Figure 6A shows a plot of the individual rate constants for the fast (kobs1) and slow (kobs2) phases of the transients against NADH concentration, while Fig. 6B shows the same data set with NADPH as reductant. For NADH-dependent reduction, kobs1 shows a hyperbolic dependence on the NADH concentration, and data were fitted to yield a limiting rate constant (klim) of 21.9 ± 0.2 s−1 and an apparent NADH Kd of 21.7 ± 0.8 μm. The kobs2 values for NADH are much slower (0.47 ± 0.03 s−1) and show no dependence on the NADH concentration (Fig. 6A). For NADPH-dependent reduction of CYP116B1, the reaction rate constants are higher, but there is no dependence of either kobs1 (72 ± 6 s−1) or kobs2 (5.5 ± 3.0 s−1) on the NADPH concentration (Fig. 6B).

Figure 6.

 Stopped-flow kinetic analysis of NADH- and NADPH-dependent CYP116B1 reduction. Stopped-flow absorption kinetic studies of CYP1161B1 (3.3 μm final concentration) were performed by monitoring the reduction of FMN/2Fe-2S centres at 465 nm (the wavelength of maximal overall change between oxidized and reduced CYP116B1) using both NADH (A) and NADPH (B) reducing coenzymes at a range of concentrations up to 500 μm. Open circles represent individual rate constants derived from the fast phase (kobs1) of double-exponential fits to the ΔA465 versus time transients, while open squares represent kobs2 rate constants for the slow phase of the reaction transients. There is no apparent dependence of either kobs1 (72 ± 6 s−1) or kobs2 (5.5 ± 3.0 s−1) on [NADPH]. With NADH as reductant, there is a hyperbolic dependence of kobs1 on [NADH], giving a limiting rate constant (klim) of 21.9 ± 0.2 s−1. However, there is no apparent dependence of kobs2 on [NADH] (0.47 ± 0.03 s−1).

Examination of the thermodynamic properties of CYP116B1

To analyse the redox properties of CYP116B1, a spectroelectrochemical redox titration was performed, using methods described previously [58,59]. For the intact CYP116B1, the overlapping spectral contributions of the individual haem, FMN and 2Fe-2S cofactors resulted in difficulty in deconvoluting potentials for the three centres in the multicofactor enzyme. However, it is clear that the reduction process occurs in two phases. In the first phase (black spectra in Fig. 7A, completed in the range of 0 to −200 mV versus the normal hydrogen electrode [NHE]), spectral bleaching of the contributions of the iron–sulfur and FMN cofactors occurs at ∼ 400–600 nm (as seen for the dithionite-reduced spectrum in Fig. 2A). There is no obvious development of a long wavelength signal (at ∼ 600 nm) during this reductive phase (Fig. 7A), suggesting that there is negligible formation of the FMN neutral semiquinone, as also observed in the dithionite-reduced CYP116B1 EPR spectrum (Fig. 5). This may indicate that the reduction potential for the FMN semiquinone/hydroquinone couple (E2) is more positive than that for the oxidized/semiquinone couple (E1). With the assumption that absorption changes at 485 nm (in the potential range where there is negligible haem contribution) reflect primarily the transition from FMN quinone to hydroquinone (E12) (with some absorption change also resulting from the reduction of the 2Fe-2S cluster), the midpoint potential for the two-electron FMN reduction was estimated as −122 ± 9 mV (Fig. 7B). Data fitted at 409 nm (an isosbestic point for the haem iron Fe(III) to Fe(II) transition) produced a similar value (−134 ± 8 mV) for the combined reduction of the FMN/2Fe-2S centres. Although differing proportions of absorption contribution probably come from the FMN and iron–sulfur cluster at the two wavelengths, the midpoint potential values are within error and it is clear that complete reduction of both cofactors occurs in this potential range.

Figure 7.

 Spectroelectrochemical titration of CYP116B1. (A) UV-visible spectra recorded during a redox titration of CYP116B1 (6.3 μm). The arrows show directions of absorbance changes occurring at different wavelengths during the reductive phase of the titration. Absorbance spectra demonstrating the reduction of the FMN and 2Fe-2S clusters at more positive potentials are shown as black lines, while spectra showing mainly reduction of the CYP116B1 haem iron from ferric to ferrous (at more negative potentials) are shown as red lines. (B) A plot of absorbance data at 485 nm versus applied potential in the range from ∼ +25 to −200 mV (versus NHE). Data are fitted using the Nernst function to define a midpoint potential of −122 ± 9 mV for the reduction of both the FMN and 2Fe-2S centres in CYP116B1. (C) A plot of absorbance data at 418 nm versus applied potential in the range from ∼ +25 to −450 mV (versus NHE). The plot shows two absorbance transitions, and data were fitted using a two-electron Nernst function to define midpoint potentials of 160 ± 9 mV (encompassing both FMN and 2Fe-2S centres) and −301 ± 7 mV (for the CYP116B1 haem iron Fe(III)/Fe(II) couple).

In the second stage of CYP116B1 reduction that occurs in the range from −200 to −500 mV versus NHE, the haem iron is converted from the ferric to the ferrous state. Haem iron reduction occurs with a shift in the Soret peak from 418 to 410 nm (red spectra in Fig. 7A), consistent with the retention of a cysteinate-ligated CYP116B1 haem iron and similar to the spectral maximum observed for the reduced form of P450 BM3 [58]. Data fitted at 393 nm, where there are negligible absorbance changes from the early part of the reductive (FMN and iron-sulfur) titration, show a single transition with a midpoint potential of −291 ± 9 mV (data not shown). Absorption versus potential data plots at the oxidized Soret maximum (418 nm) show a clear two-stage transition (Fig. 7C). The decrease in A418 intensity in the first stage of the titration is attributed to the reduction of the FMN and iron-sulfur cofactors. This accounts for approximately half of the overall absorption change and is indicative of high levels of incorporation of these cofactors in CYP116B1. The midpoint potentials for the two transitions were determined using a two-electron Nernst function [58], yielding values of (a) −160 ± 9 mV to encompass the FMN E12 and iron–sulfur cluster [2Fe-2S]2+/[2Fe-2S]+ couples and (b) −301 ± 7 mV for the haem iron Fe(III)/Fe(II) couple. These midpoint potential values are consistent with the data fits at other wavelengths. Variations in the relative optical contributions of the FMN and 2Fe-2S cofactors may explain the small differences in redox-potential estimates for their combined reduction at different analysis wavelengths.

Identification of CYP116B1-catalysed reaction products of thiocarbamate herbicides

Following the fatty acid/thiocarbamate binding and steady-state kinetic studies reported above, we undertook experiments to establish whether CYP116B1 catalysed the oxidation of unsaturated fatty acids or of the thiocarbamates vernolate and EPTC. No formation of oxidized products was observed for NAD(P)H-driven turnover of CYP116B1 with monounsaturated fatty acids. However, preliminary studies indicated that products were formed from both vernolate and EPTC, with superior levels of product formation seen using NADPH as the reducing coenzyme (consistent with the steady-state kinetic data in Table 1). A detailed characterization of the products from the NADPH-dependent turnover of CYP116B1 with these thiocarbamates was thus undertaken by liquid chromatography (LC)-MS, as described in the Materials and methods.

HPLC was used to analyse products from CYP116B1-catalysed oxidation of vernolate and revealed major features in the chromatogram at 34–39 min after sample loading, in addition to the remaining parent molecule that eluted between 50 and 52 min (Fig. 8A). The new features consisted of two major product peaks, with a further small peak distinguishable between these major peaks. These product features are absent from vernolate-only or enzyme-only controls. The shorter retention time of these peaks indicates that they arise from compounds more polar in nature than the parent vernolate. The mass spectra of the major bands at 37.1 and 38.4 min in the chromatogram are nearly identical, with two major peaks observed with m/z values at 220 and 144 (Fig. 8B). The larger peak, at m/z 220, corresponds to the mass (M+H) of vernolate with a single hydroxylation. Further analysis of this peak performed by collision-induced disassociation (MS2) identified a single MS2 fragment with m/z ratio 144. Analysis of the vernolate structure revealed this MS2 fragment to be a hydroxylated form of vernolate lacking the S-propyl group and probably arising from cleavage of the S-C bond (Fig. 8C). MS2 analysis of the secondary HPLC-derived peak at m/z 144 revealed a single feature with an m/z ratio of 116, which again is a hydroxylated product of a vernolate fragment lacking both the S-propyl group and the carbonyl group (Fig. 8D). This fragmentation pattern that shows loss of the S-propyl group is consistent with that observed in the vernolate-only controls (Fig. S2A,B) and confirms the position(s) of CYP116B1-catalysed hydroxylation to be on one of the N-propyl chains. The fact that we observed two major peaks in the HPLC chromatogram with an almost identical mass spectrum may also be indicative of hydroxylation occurring at different positions on the N-propyl chain, although we cannot define their exact position(s) from these data. The mass spectrum of the third small species observed at 37.7 min in the chromatogram (indicated by an arrow in Fig. 8A) gave a distinct single peak with an m/z of 162 (Fig. 8E). This m/z ratio is consistent with dealkylation (loss of a propyl group) from the parent vernolate. This is most likely to be an N-dealkylation reaction given the different m/z ratio to that observed for the loss of the S-propyl group upon fragmentation. However, this dealkylated product is a minor species compared with the hydroxylated form(s) of vernolate, as seen from the difference in sizes of the respective peaks in the HPLC chromatogram. Similar hydroxylated products on the N-propyl chains are also observed for the thiocarbamate EPTC, which differs by a CH2 group in the S-ethyl chain (S-propyl in vernolate) (Fig. S3), although these are found in lower amounts compared with the vernolate products. There is no evidence of CYP116B1-mediated N-dealkylation with EPTC, although we cannot rule out small amounts of such a product. Given the greater proportion of hydroxylation and N-dealkylation seen with vernolate over EPTC, it appears likely that CYP116B1 binds more favourably to the S-propyl group in vernolate than to the S-ethyl group in EPTC to enable oxidative catalysis.

Figure 8.

 Analysis of products of CYP116B1-catalysed vernolate oxidation by liquid chromatography (LC)-MS. (A) HPLC chromatogram showing elution profiles of vernolate substrate (upper blue line) and CYP116B1-catalysed oxidation products of vernolate (lower black line). Vernolate elutes as a single peak between 50 and 52 min. Two additional peaks are observed following turnover with CYP116B1. These elute between 34 and 39 min and were determined to be hydroxylated forms of vernolate. A further sharp peak of lower intensity is also identified within this region, eluting at 37.7 min and indicated by the arrow. This feature is assigned to an N-dealkylated product of vernolate. (B) The mass spectrum of the chromatogram peaks at 37.1 and 38.4 min (both found to contain almost identical masses) showing two major species with m/z values of 220 and 144, and a smaller m/z feature at 162. The m/z 220 species results from hydroxylation of vernolate (m/z = 204 + 16). The structure of a potential hydroxylated product of vernolate is shown as an inset. (C) MS2 analysis of the m/z 220 species with a resultant m/z at 144. This 144 m/z species is assigned to a hydroxylated fragment of vernolate, with the S-propyl group removed. A representative structure of the hydroxylated vernolate fragment structure is shown as an inset. (D) MS2 analysis of the m/z 144 species with a resultant m/z at 116. This is assigned to a hydroxylated fragment of vernolate, with the S-propyl and adjacent carbonyl groups removed. These data are further confirmatory that the hydroxylation occurs on one of the N-propyl chains, but do not define an exact position. A representative structure of this hydroxylated fragment is shown as an inset. (E) Mass spectrum of the minor peak at 37.7 min in the ‘CYP116B1 + vernolate’ chromatogram (A), showing a predominant m/z peak at 162. This is consistent with dealkylation (loss of a propyl group) from vernolate. While the dealkylation could theoretically occur with removal of either an N- (shown inset) or an S-propyl chain, the likely product is the N-dealkylated form (as seen for the reaction of Rhodococcus erythropolis ThcB with EPTC), occurring by CYP116B1-catalysed C-N bond cleavage following successive hydroxylations at adjacent carbons on the N-propyl chain [34].


We report here the first biochemical characterization of CYP116B1, a representative of a family of cytochrome P450–PDOR fusion enzymes with biotechnological potential. Like P450 BM3 and other related P450–P450 reductase fusion enzymes, this makes CYP116B1 a catalytically self-sufficient P450 enzyme, requiring only substrates [NAD(P)H coenzyme and thiocarbamate herbicide] for turnover (2,12,32).

The C. metallidurans CYP116B1 was expressed in E. coli and purified to homogeneity using a combination of anion-exchange and hydrophobic affinity chromatography. Heterologous expression of cofactor-binding proteins in E. coli can result in incomplete incorporation of non-covalently bound cofactors. Indeed, in the case of the Rhodococcus sp. CYP116B2 enzyme, deficiencies in both FMN and 2Fe-2S cofactors were noted [33]. In our study of the C. metallidurans CYP116B1, EPR quantification revealed essentially stoichiometric incorporation of the 2Fe-2S cluster and the P450 haem (Fig. 5). The FMN cofactor was more weakly bound, but inclusion of FMN in purification buffers enabled FMN incorporation to > 80%. UV-visible, EPR and resonance Raman spectroscopy (see the Supporting Information) confirmed that that enzyme was isolated in a predominantly low-spin ferric state, with cysteinate and water as the fifth and sixth (axial) ligands to the haem iron (Figs 2A and 5, Fig. S1). The CYP116B1 Soret maximum at 418 nm is typical of such coordination and spin-state in P450s, whereas the previous report of the related CYP116B2 enzyme with Soret peak at 424 nm is instead indicative of an enzyme in which imidazole (used to elute the enzyme from nickel affinity resin) has displaced water as the sixth ligand to produce an inhibited form of the enzyme [32]. Previous studies of intact CYP116B2 and its PDOR domain indicated a low content of the 2Fe-2S cluster (in the range of 4–36%), necessitating iron–sulfur cofactor reconstitution of the enzyme under anaerobic conditions [33]. Given the essentially stoichiometric incorporation of the 2Fe-2S cluster reported here for CYP116B1 (purified without using a nickel affinity step), it appears possible that the nickel Sepharose resin used for purification of CYP116B2 resulted in stripping of iron from the 2Fe-2S centre in this enzyme.

There is no EPR signal attributable to bound flavin in its EPR active (semiquinone) form in the dithionite-reduced form of CYP116B1 (Fig. 5), and no obvious optical signals typical of a flavosemiquinone are observed for either dithionite- or NADPH-reduced CYP116B1 (Fig. 2A). We thus infer that the reduction potential for the CYP116 FMN oxidized/semiquinone couple (E1) is more negative than that for the semiquinone/hydroquinone couple (E2), meaning that little FMN semiquinone accumulates during flavin reduction. Bound flavin is converted from quinone to the two-electron hydroquinone form on reduction with dithionite, and thus remains EPR silent. Redox potentiometry of CYP116B1 (Fig. 7) demonstrates clearly that the reduction of the non-haem centres in CYP116B1 takes place at a more positive potential than does the reduction of the P450 haem iron. Bleaching of absorption occurs in the flavin (and iron–sulfur) region between ∼ 420 and 520 nm before there is any notable shift in the position of the haem Soret band. At more negative potentials, the Soret shifts from 418 to ∼ 410 nm, consistent with retention of cysteine thiolate as the haem iron fifth ligand in the reduced (ferrous) state of CYP116B1 [37,38]. The CYP116B1 haem iron potential (−301 ± 7 mV versus NHE from data analysis at 418 nm) is more positive than that previously reported for the isolated haem (P450) domain of the related CYP116B2 enzyme (−380 ± 4 mV), but is also more negative than the reduction potential of the 2Fe-2S and FMN centres in CYP116B1, whose reduction is completed by ∼ −200 mV. This suggests that binding of substrate to the CYP116B1 P450 active site may make the haem iron potential more positive (through displacement of the sixth water ligand to the ferric iron and concomitant shift in haem iron spin-state towards high-spin), as is the case for several other P450s [e.g. 41,58]. The limited solubility of vernolate/EPTC, their marginal effects on CYP116B1 haem spin-state equilibrium and their promotion of CYP116B1 aggregation prevents us from establishing whether the binding of these dithiocarbamates is associated with a large elevation in haem iron potential. This makes it difficult to establish the extent to which substrate-dependent haem iron spin-state modulation is important for control of enzyme catalysis in CYP116B1, or whether an alternative mode of regulation may be involved [e.g. 57,58].

In other work reported here, CYP116B1 was shown to bind to unsaturated fatty acids with chain lengths from C14 to C20, producing substrate-like type I spectral changes with accumulation of HS haem iron (e.g. Fig. 3A). However, no evidence for their oxidation by CYP116B1 could be obtained either through measurements of fatty acid-dependent stimulation of CYP116B1 NAD(P)H oxidation or isolation of oxidized fatty acid products. In contrast, the thiocarbamate herbicides vernolate and EPTC, despite showing rather small P450 type I (HS) spectral changes, not only stimulated CYP116B1-dependent NAD(P)H oxidation (Table 1), but were also shown to be substrates for CYP116B1, and to be hydroxylated (EPTC and vernolate) and N-dealkylated (vernolate) in an NADPH-dependent manner. The modest type I shifts induced by these herbicides possibly indicates that they can occupy different positions in the CYP116B1 active-site pocket, with transient occupancy of the HS position probably linked to oxidative catalysis by the P450. Steady-state kinetic analysis indicates a preference for vernolate over EPTC in CYP116B1, and this is confirmed by the greater amounts of hydroxylated products obtained with vernolate, as well as there being clear evidence for N-dealkylation only in the case of vernolate. Vernolate differs from EPTC in the length of its S-alkyl side chain (S-propyl for vernolate and S-ethyl for EPTC). The vernolate S-propyl chain may confer greater specificity for CYP116B1 than does the respective S-ethyl chain in EPTC. Although CYP116B1 catalyses N- rather than S-dealkylation of vernolate, it is conceivable that the S-propyl chain portion of the thiocarbamate makes stabilizing interactions in the active site that are important for orientating one of the N-propyl chains of vernolate towards the CYP116B1 haem to enable its oxidation and subsequent N-dealkylation. At this stage we cannot entirely rule out the absence of CYP116B1-mediated S-dealkylation, although products are not detected under the experimental conditions used. Other N-dealkylation reactions are known in P450 chemistry, and such oxidations are common in human enzymes. For instance, several such reactions are catalysed by the major human drug-metabolizing enzymes CYP2D6 and CYP3A4, including N-dealkylation of the opioid addiction drug buprenorphine [e.g. 60,61]. In the case of the vernolate N-dealkylation reaction catalysed by CYP116B1, there are clear functional similarities between CYP116B1 and the non-partner fused P450 ThcB in R. erythropolis NI86/21 [34]. ThcB (CYP116A1 in the P450 superfamily) was shown to oxidatively dealkylate EPTC. This is probably the first of a series of reactions enabling thiocarbamate catabolism in the Rhodococcus strain. CYP116A1 and the haem domain of CYP116B1 show 52% amino-acid sequence identity and are clearly orthologues. In R. erythropolis, genes encoding a 2Fe-2S ferredoxin (rhodocoxin) and a ferredoxin reductase (rhodocoxin reductase) are located immediately downstream of the P450, forming a classical bacterial class I P450 redox system [2,34]. The divergent evolution of the CYP116B family has clearly involved the genetic fusion of the CYP116-type P450 with a distinctive redox partner system, presumably as a result of advantages gained in catalytic efficiency. While it is uncertain whether thiocarbamates such as EPTC and vernolate are physiologically relevant substrates of CYP116B1, it is quite possibly the case that C. metallidurans exploits this P450 as part of its armoury of enzymes for metabolism of xenobiotics. C. metallidurans is a bacterium able to thrive in stressful pollutant/heavy metal-contaminated and nutrient-limited environments, and oxidases of diverse substrate selectivity are probably important components of its survival strategy [62].

In conclusion, we present the first report of the biochemical, spectroscopic, thermodynamic and catalytic properties of C. metallidurans CYP116B1, a novel, catalytically self-sufficient P450–PDOR fusion enzyme with oxidative activity towards thiocarbamate herbicides. From a biotechnological viewpoint, there is growing interest in the exploitation of such single-component P450–redox partner fusion enzymes. This comes both from the perspective of using protein engineering on high-activity fusion enzymes, such as P450 BM3, to enable industrially useful oxidation chemistry [e.g. 25–28], and also with respect to mimicking the efficiency of the naturally fused P450 enzymes by generating artificial fusions of exogenous P450s to redox partner systems. In the case of P450 BM3, fusions of exogenous P450s to the BM3 reductase have met with limited success, probably as a consequence of its dimeric nature and the difficulty in orientating reductase and P450 domains to achieve efficient intermolecular electron transfer. However, CYP116B1, and probably the other CYP116B family members, are monomeric enzymes, and thus fusions of exogenous P450s to their PDOR modules could provide a more robust route to the generation of novel self-sufficient P450 oxidases for biotechnological applications [63].

Materials and methods

Cloning, expression and purification of CYP116B1

The CYP116B1 gene (Rmet_4932) was cloned from genomic DNA prepared from a culture of C. metallidurans CH34 supplied by Professor Nigel Brown of the University of Birmingham, UK. To generate the megaplasmid DNA template for Rmet_4932 cloning, single colonies of C. metallidurans were resuspended in 1 mL of distilled, deionized water (ddH2O), and 50 μL of the sample was heated at 99 °C for 9 min in a PHC-2 thermal cycler (Techne, Burlington, NJ, USA). The PCR reaction to amplify the CYP116B1 gene involved 30 cycles of (a) denaturation at 95 °C for 45 s, (b) annealing at 59 °C for 1 min and (c) polymerization at 72 °C for 5 min, followed by a final polymerization step of 72 °C for 10 min. The forward and reverse primers used for this reaction were based upon the published genome sequence of C. metallidurans (, and were as follows: CYP116B1F, GGACTAATCTCGCTGGACTGATTAATGCCGC; and CYP116B1R, CGCTCAGCATCCGTGATGCCGTGCTGAG (Nucleotides in bold are sites for restriction enzymes AseI and BlpI, respectively, which were incorporated to facilitate the subcloning of CYP116B1. Underlined nucleotides indicate the initiation (ATG) and termination (TGA) codons.) The PCR (∼ 2.4 kbp) product was purified using a Min-Elute gel-purification kit (Qiagen, Crawley, UK) using the manufacturer’s protocol. The PCR fragment was then A-tailed and cloned into the plasmid vector pGEM-T (Promega, Southampton, UK). CYP116B1 clones were verified by blue/white screening following the manufacturer’s instructions, and subsequently by DNA sequencing of the entire genes (PNACL, University of Leicester). The CYP116B1 gene was transferred from the pGEM-T cloning vector into the expression plasmid, pET11a (Merck, Nottingham, UK). CYP116B1 was excised from the pGEM-T clone by restriction digestion with AseI and BlpI and the resultant gene fragment (2423 bp) was resolved by electrophoresis through a 1% agarose gel and purified as above. The pET11a vector was digested using BlpI and NdeI (the latter enzyme produces ends cohesive with those generated by AseI) and similarly purified. Resultant colonies from ligation reactions were identified by diagnostic restriction enzyme digestions and verified by DNA sequencing. Validated CYP116B1–pET11a constructs were used for CYP116B1 production.

Expression trials were performed with CYP116B1–pET11a under a variety of conditions (varying growth temperature, culture time, growth medium and isopropyl thio-β-d-galactoside (IPTG) concentration) and in various DE3 lysogen E. coli strains (Rosetta, Origami and HMS174, all from Merck). Optimal expression of CYP116B1 was obtained in the Origami(DE3)/CYP116B1–pET11a transformant with growth at 37 °C in Luria–Bertani (LB) medium until the culture reached an optical density at 600 nm (D600) of 0.6. Thereafter, the culture temperature was reduced to 18 °C and growth continued until an D600 of 0.8 was reached. At this point, IPTG was added to a final concentration of 100 μm and growth was continued overnight (typically 12–16 h). For protein preparation, a total culture volume of ∼ 6 L was used. Cells were harvested by centrifugation (30 min at 10 000 g in a Beckman JLA 8.1000 rotor). Cell pellets were resuspended in buffer A (50 mm Tris/HCl, pH 7.2, containing 1 mm EDTA) and then re-pelleted as before. Cells were resuspended in the same buffer (∼ 50 mL) and kept on ice. A CompleteTM protease inhibitor tablet (Roche, West Sussex, UK) was added at this stage, along with the protease inhibitors phenylmethanesulfonyl fluoride (1 mm) and benzamidine hydrochloride (1 mm). These protease inhibitors were added to all buffers from this stage in the purification. The cells were disrupted by passage (twice) through a French Press (ThermoFisher, Loughborough, UK) at 950 psi, followed by ultrasonication using a Bandelin Sonopuls GM2600 sonicator (6 × 20-s bursts at 50% full power, with 5 min cooling time on ice allowed between bursts). The lysate was centrifuged at 46 000 g for 30 min at 4 °C and then the supernatant was transferred to a DEAE column pre-equilibrated in buffer A. CYP116B1-containing fractions were eluted using a gradient (500 mL) of buffer A to buffer A + 500 mm KCl. CYP116B1 eluted at ∼ 350 mm KCl. Fractions were pooled and transferred to a phenyl Sepharose column pre-equilibrated with buffer B (buffer A containing 0.75 m ammonium sulphate solution). The column was washed extensively in buffer B and CYP116B1 was eluted using a gradient of buffer B to buffer A (500 mL). Fractions were pooled, concentrated by ultrafiltration (using Centriprep™ concentrators; Millipore, Watford, UK) and dialysed extensively against buffer A to desalt the protein. The protein solution was then transferred to a Q-Sepharose column and CYP116B1 was eluted as described for the DEAE step above. CYP116B1 was pure at this stage and was concentrated by ultrafiltration (as already described) to a final concentration of ∼ 500 μm, and then dialysed into buffer A, containing 50% (v/v) sterile glycerol, before storage at −80 °C.

UV-visible spectroscopic analysis and cofactor identification

UV-visible spectra of CYP116B1 were collected using a Cary UV-50 scanning spectrophotometer (Varian) with a 1 cm-pathlength quartz cuvette. Spectra were recorded for oxidized, sodium dithionite-reduced and various substrate-(fatty acid and thiocarbamate) and ligand- (CO, NO, cyanide, imidazole and substituted imidazole) bound forms, as described previously [35,38]. Reduction of CYP116B1 and binding of CO was performed as described previously [64]. The haem concentration of CYP116B1 was determined using the pyridine haemochromagen method [65].

SEC-MALLS studies of CYP116B1

MALLS analysis of CYP116B1 was coupled to SEC and performed on a Supredex-200 24/30 gel-filtration column (GE Healthcare, Chalfont St Giles, UK) run in 50 mm potassium phosphate (pH 8.0), containing 200 mm KCl, on a Dionex BioLC HPLC at 0.71 mL·min−1. Protein was passed directly through a Wyatt EOS 18-angle laser photometer coupled to a Wyatt Optilab rEX refractive index detector. The molecular weight moments and concentrations of the resulting peaks were analysed using the Astra v.5.3.2 software (Wyatt Technology, Haverhill, UK). Samples were passed through the column twice to ensure that no aggregating species were included in the mass calculations.


EPR spectra were recorded at 10 K on a Bruker ER-300D series electromagnet and microwave source, interfaced to a Bruker EMX control unit and fitted with a dual microwave cavity from the same supplier (model ER-4116DM). Spectra were recorded at 2 mW microwave power at a modulation amplitude of 10 G. Spectra were recorded for both oxidized and sodium dithionite-reduced forms of CYP116B1.

Determination of CYP116B1 flavin content

Flavin fluorescence was recorded with sample excitation at 450 nm, and emission data were collected at 500–600 nm using a Cary Eclipse fluorescence spectrometer in a 2 ml quartz fluorescence cuvette of 1 cm pathlength. To remove flavin cofactor from CYP116B1, enzyme samples (150 nm–3 μm) were heated to 95 °C for 3 min and aggregated protein was separated by centrifugation at 14 000 g for 10 min at 4 °C. The supernatant was retained and analysed. Fluorescence data were collected for the protein in 100 mm potassium phosphate (pH 7.7) and following acidification with perchloric acid at pH 2.0, according to the method of Knight and Hardy [66]. Fluorescence data at the different pH values were compared with those for standard FAD and FMN solutions for identification and quantification of the CYP116B1-bound flavin, as described in our earlier review [55].

Substrate- and ligand-binding studies

Analysis of interactions of CYP116B1 with various ligands and candidate substrates was performed by spectral titration in buffer A, monitoring ligand-induced absorption changes across the spectral region from 300 to 750 nm. Difference spectra were created at each point in the titrations by subtraction of the spectrum for the ligand-free enzyme from that of the relevant ligand-bound form. Titrations were continued until no further spectral change could be discerned. Maximal induced absorption change data were computed as the difference in absorption at the peak and trough in the difference spectra, and these values were plotted versus the relevant ligand concentration applied. To determine Kd values, these data were fitted to either a standard hyperbolic function or to a quadratic function (for very tight-binding ligands), as described previously [39]. Titrations were performed at 25 °C, with ligands added either in buffer A or (for fatty acids and thiocarbamate herbicides) in ethanol or dimethylsulfoxide. The total volume of ligand addition was < 1% of the final volume in the assay. Volume changes as a result of ligand additions were accounted for by spectral correction before data analysis.

Steady-state and stopped-flow analysis of CYP116B1

Steady-state kinetic analysis was carried out with CYP116B1 in buffer A at 25 °C on a dual-beam spectrophotometer (Jasco). Assays for reduction of electron acceptor substrates were performed with cytochrome c (horse heart) and potassium ferricyanide. Reactions typically used 75–100 nm CYP116B1, and the Km values for cytochrome c (29.9 ± 1.7 μm) and ferricyanide (103.6 ± 20.5 μm) were established in preliminary assays using NADPH at a saturating concentration (200 μm). Thereafter, the kcat and Km values for NADPH and NADH were established in assays in which the coenzyme concentration was varied for both NADH (0–2 mm) and NADPH (0–200 μm), and the electron acceptor concentration was maintained at a constant near-saturating concentration (300 μm for cytochrome c and 1 mm for ferricyanide). Rates were determined at 550 nm for cytochrome c using Δε550 = 22640 m−1·cm−1 [18] and at 420 nm for ferricyanide using Δε420 = 1020 m−1 cm−1. Similar assays for P450 substrate-dependent NAD(P)H oxidation (Δε340 = 6210 m−1 cm−1) were carried out using various fatty acids (myristoleic acid, palmitoleic acid and arachidonic acid, typically 0–500 μm) and using the thiocarbamate herbicides EPTC and vernolate, at concentrations up to 2 mm, using previously described methods [48]. Vernolate-dependent NADPH oxidation was also measured across a range of CYP116B1 concentrations (10–500 nm) in order to evaluate the effects on the specific catalytic rate. CO-trapping experiments (to examine the influence of dithiocarbamate substrate on the reduction of CYP116B1 haem) were performed anaerobically in the absence and presence of vernolate, and in saturated CO buffer with 3 μm CYP116B1. Electron transfer was initiated by the addition of 10-fold excess NAD(P)H, and spectral data were recorded to monitor the formation of (reduced-CO) P450-adducts.

Stopped-flow studies of CYP116B1 were carried out under the same conditions as the steady-state assays. Assays were performed using both NADH and NADPH as reductants and at 465 nm (the wavelength of maximal overall change between oxidized and PDOR domain-reduced CYP116B1). Data were recorded using an SX18MV UV-visible stopped-flow instrument (Applied Photophysics, Leatherhead, UK). Enzyme (6.6 μm) was mixed with NAD(P)H (up to 500 μm) in the stopped-flow instrument and single-wavelength data were recorded for up to 0.5 s. Data were fitted accurately using a double exponential function and the applied photophysics software.

Thermodynamic characterization

A spectroelectrochemical redox titration of CYP116B1 was performed, as described previously [58,59], with CYP116B1 at ∼ 8 μm. Spectral data were fitted using Nernst functions at wavelengths reflecting transitions from the oxidized to the reduced forms of the bound cofactors, in order to derive midpoint reduction potentials for the CYP116B1 haem iron Fe3+/Fe2+ transition and to establish the potential range in which the 2Fe-2S and FMN cofactors were reduced.

Product analysis

In order to identify the products of vernolate and EPTC oxidation by CYP116B1, turnover reactions were performed according to the methods reported by Lawson et al. [47]. The 10-ml reactions contained CYP116B1 at a final concentration of 6 μm, 1 mm NADPH, 20 mm glucose and 8.25 units of glucose dehydrogenase (an NADPH regeneration system), and 0.33 mm of either vernolate or EPTC. The reaction constituents were all solubilized in buffer A, with the exception of vernolate and EPTC, which were prepared in ethanol. The reactions were performed in 30-mL glass vials that had been wrapped in foil to exclude light and washed with methanol, dichloromethane and ddH2O to remove potential contaminants. Control reactions lacking either CYP116B1 or substrate were also performed. Reaction mixtures were incubated at 25 °C and stirred continuously for 16 h. Reactions were terminated by the addition of 0.5 mL of 1.0 m HCl. Reaction products were extracted by adding 3 mL of HPLC-grade dichloromethane to each reaction. Organic phases were retained and aqueous phases were subjected to a second extraction with dichloromethane to maximize product recovery. Organic phases from each reaction were pooled and the residual aqueous phase was removed by the addition of anhydrous magnesium sulphate. Samples were filtered to remove any particulates before being evaporated to dryness under vacuum. The resultant pellets were resuspended in a minimal volume of HPLC-grade methanol (∼ 50 μL) for LC-MS analysis. LC-MS was performed using an Agilent 1100 series HPLC-MSD (Agilent Technologies, Waldbronn, Germany). Samples were resolved using a Vydac C8 column (21 × 150 mm) run at a flow rate of 1 mL·min−1 and maintained at 30 °C. Following sample loading, an isocratic gradient of the mobile phase (66% solvent A [100% HPLC-grade acetonitrile] and 34% solvent B [0.2% acetic acid in ddH2O]) was run for 28 min. Subsequently, the concentration of solvent B was raised to 100% and the column was run for an additional 32 min. Detection of reaction products from the resolved samples was performed using the ion trap mass spectrometer in the positive electrospray mode.


The authors thank the Biotechnology and Biological Sciences Research Council UK (BBSRC) for funding (PhD studentship to Ashley Warman). The authors also thank Dr Harriet Seward (University of Leicester) for assistance with EPR data collection, Dr Tom Jowitt and Mrs Marj Howard for assistance with MALLS collection and analysis and Mrs Marina Golovanova (University of Manchester) for excellent technical support.