The Critical Role of Substrate-Protein Hydrogen Bonding in the Control of Regioselective Hydroxylation in P450cin

Abstract

Cytochrome P450cin (CYP176A1) is a bacterial P450 isolated from Citrobacter braakii that catalyzes the hydroxylation of cineole to (S)-6β-hydroxycineole. This initiates the biodegradation of cineole, enabling C. braakii to live on cineole as its sole source of carbon and energy. P450cin lacks the almost universally conserved threonine residue believed to be involved in dioxygen activation and instead contains an asparagine at this position (Asn-242). To investigate the role of Asn-242 in P450cin catalysis, it was converted to alanine, and the resultant mutant was characterized. The characteristic CO-bound spectrum and spectrally determined K_D for substrate binding were unchanged in the mutant. The x-ray crystal structures of the substrate-free and -bound N242A mutant were determined and show that the only significant change is in a reorientation of the substrate such that (R)-6α-hydroxycineole should be a major product. Molecular dynamics simulations of both wild type and mutant are consistent with the change in regio- and stereoselectivity predicted from the crystal structure. The mutation has only a modest effect on enzyme activity and on the diversion of the NADPH-reducing equivalent toward unproductive peroxide formation. Product profile analysis shows that (R)-6α-hydroxycineole is the main product, which is consistent with the crystal structure. These results demonstrate that Asn-242 is not a functional replacement for the conserved threonine in other P450s but, rather, is critical in controlling regioselective substrate oxidation.

Members of the superfamily of cytochromes P450 (P450s) catalyze an impressive array of oxidative transformations and are key enzymes in processes as varied as human xenobiotic metabolism, steroid hormone biosynthesis, and bacterial biodegradation. P450s utilize molecular oxygen as the ultimate oxidant in the reaction R-H + NAD(P)H + O₂ + H⁺ → R-OH + NAD(P)⁺ + H₂O.

The enzyme forms a hydroperoxy, Fe(III)-OOH, intermediate (1) during the catalytic cycle. Proper protonation of the distal oxygen results in heterolytic cleavage of the peroxide O-O bond, thus generating the active ferryl species, Fe(IV) = O, responsible for substrate oxidation, whereas protonation of the proximal oxygen directly linked to the iron atom would give rise to hydrogen peroxide. A threonine that is almost universally conserved across all P450s has been implicated in the control of this process (2, 3) by assisting in the proper delivery of solvent protons to the iron-linked dioxygen and possibly stabilizing the hydroperoxy intermediate (3). Whatever the exact mechanistic role of the threonine, it is generally required for efficient P450-catalyzed oxidation of a substrate. For example, mutation of the conserved threonine to alanine in P450cam, a widely studied model enzyme isolated from Pseudomonas putida, results in a majority of NADH-reducing equivalents being funneled into hydrogen peroxide formation rather than hydroxylated product (4, 5). Such non-productive use of electrons is generally termed uncoupling and may result in the formation of water or superoxide in addition to hydrogen peroxide.

A small number of P450s exist that naturally lack the conserved threonine. The majority of these do not catalyze the activation of molecular oxygen but, rather, the rearrangement of a hydroperoxide already present in the substrate. Examples include thromboxane synthase, prostacyclin synthase (6), or members of CYP74 subfamilies (7–10). P450EryF is a rare example of a P450 that lacks the conserved threonine but which can efficiently catalyze hydroxylation of a non-activated carbon. In this case, a hydroxyl group present within the substrate is essential for correct protonation of the ferric hydroperoxy intermediate and efficient catalysis (11, 12). It was originally proposed that the substrate hydroxyl facilitates the transfer of solvent protons to the iron-linked dioxygen site (13). However, recent x-ray crystallographic studies show that water is expelled from the active site upon oxygen binding to the P450eryF-substrate complex, leaving the substrate hydroxyl as the most likely candidate for delivering the required proton (14).

P450cin provides a second example of a P450 that catalyzes a hydroxylation reaction but lacks the conserved threonine (see Fig. 1). P450cin was isolated from Citrobacter braakii and is thought to initiate a cascade of biodegradative reactions that allows the bacterium to live on cineole as its sole carbon and energy source. Sequence alignments have shown that the conserved threonine is absent and is replaced by an asparagine residue (Asn-242). The crystal structure of P450cin shows that Asn-242 donates an H-bond to the substrate ethereal oxygen (15) (Fig. 1). In this paper we address the following question. Is Asn-242 a functional replacement for the conserved threonine in P450cin? Clearly it would be capable of H-bonding and stabilizing the water network believed to be necessary for normal catalytic function and, thus, may be essential for substrate hydroxylation. Secondly, does Asn-242 play an important role in the high affinity of P450cin for cineole, and/or does it control the regio- and stereochemistry of substrate oxidation? Here we report the enzymatic properties and crystal structures of the N242A P450cin mutant in the presence and absence of cineole.

FIGURE 1.

A, overall structure of P450cin. Helix I, which contains Asn-242, is highlighted in cyan and labeled. Also labeled are the F and G helices. In some P450s these two helices and the loop connecting the F and G helices undergo a large open/close motion to allow substrate in and product out. B, a close-up view showing the interactions between Asn-242 and the substrate. C, the same active site region in P450cam. D, the nomenclature of the hydroxycineoles can be complex. Cineole is an achiral, meso compound, and hydroxylation at either methylene adjacent to the C1 bridgehead carbon leads to enantiomeric products and the creation of three new stereogenic centers. The secondary alcohol carbon may be either R/S depending on whether the hydroxyl moiety is a (opposite side to oxybridge) or b (same side as oxy bridge). Therefore, we have defined the methylene that upon oxidation leads to the R-C1 isomer as the pro-R carbon and the methylene that leads to the S-C1 isomer the pro-S carbon.

MATERIALS AND METHODS

Construction of N242A Mutant—The N242A mutant was generated in two PCR steps using forward (CP-1, 5′-CCGAATTCCATATGACTGCGACAGTCGC-3′) and reverse (CP-2, 5′-CCGAATTCTAGAGGAGCTTGCTCATTCCG-3′) flanking primers (16) and forward (N242A-f, 5′-CTCGGCGGCATCGACGCCACCGCACGCTTCCTC-3′) and reverse (N242A-r, 3′-CGCCGTAGCTGCGGTGGCGTGCGAAGGAGTCG-5′) mutagenic primers, pCW-P450cin as a template, and Vent® DNA polymerase. In the first step two separate reactions were carried out, one with CP-1 and N242A-r and one with CP-2 and N242A-f. Both contained pCW-P450cin as template DNA and were performed under the following conditions: 95 °C for 100 s, 30 cycles of 94 °C for 30 s, 59 °C for 60s, and 72 °C for 60s. Extension was concluded with incubation at 72 °C for 5 min. The two partially overlapping DNA fragments obtained were purified using a QIAquick® PCR purification kit according to the manufacturer's instructions.

The two partially overlapping PCR fragments produced above were then used as both primer and template to direct amplification of a complete gene construct. A PCR experiment containing the two PCR fragments was performed under the following conditions: 10 cycles of 94 °C for 30 s, 61 °C for 60 s, and 72 °C for 60 s. CP-1 and CP-2 were then added followed by another 25 cycles using the same conditions given above. The reaction was concluded with incubation at 72 °C for 5 min. The 1.2-kilobase PCR product was cloned into pUC18 using the SureClone® Ligation kit (Amersham Biosciences), and the mutation was confirmed by DNA sequencing. Finally, a ∼0.8-kilobase fragment containing the mutation was excised from the pUC18 construct using NcoI/pPUM1 and ligated into similarly digested pCW-P450cin.

Expression and Purification of the P450cin N242A Mutant—The P450cin N242A mutant was expressed and purified as reported in the literature for wild type P450cin (16). This yielded ∼90 nmol of purified protein/liter of original culture.

Substrate Binding—The dissociation constant (K_D) for the enzyme-substrate complex was calculated using a standard procedure (16).

NADPH Consumption—The standard reaction mixture (Tris-HCl 50 mm, pH 7.4, total volume of 800 μl) contained a P450 (0.5 μm), cindoxin (10 μm), Escherichia coli reductase (5 μm), cineole (1 μl), and NADPH (200 μm). NADPH was added last to initiate the reaction. The rate of NADPH consumption was calculated by monitoring the disappearance of absorbance at 340 nm.

Determination of Coupling—To calculate the percentage coupling (NADPH consumed:product formed), a solution (800 μl, 50 mm Tris-HCl, pH 7.4) containing a P450 (0.5 μm), cindoxin (4 μm), E. coli reductase (1 μm), catalase (1 μm), cineole (6 mm), and a known amount of NADPH (∼1–2 mm) was incubated at 25 °C for 1.5 h. α-Terpineol (400 μl, 1.03 mm) was added as an internal standard, and the reaction mixture was extracted into ethyl acetate (0.5 ml) and the extract dried (MgSO₄). The samples were analyzed by gas chromatography on a DB-WAX polar column under the following conditions: 80 °C for 2 min, 10 °C/min to 200 °C, hold 20 min. The amount of hydroxycineole produced was determined via a standard curve generated from stock solutions of α-terpineol and hydroxycineole. This was then compared with the amount of NADPH added to calculate the percentage coupling.

H₂O₂ Production—The amount of H₂O₂ formed was measured using an iron(II) thiocyanate assay as previously described (17). The standard reaction (50 mm Tris-HCl, pH 7.4) contained P450 (0.5 μm), cindoxin (4 μm), E. coli reductase (1 μm), cineole (6 mm), and a known concentration of NADPH (∼1–2 mm).

Product Stereochemistry—A standard reaction was performed as described above for determination of H₂O₂ production. The reaction was extracted into ethyl acetate (0.5 ml), and this was then evaporated under a gentle stream of N₂. The sample was redissolved in dichloromethane (400 μl), trifluoroacetic anhydride (150 μl) was added, and the mixture left to stand at room temperature for 30 min. The dichloromethane was removed under a gentle stream of N₂, and the sample was redissolved in ethyl acetate. Analysis of the esterified hydroxycineoles was performed on a gas chromatograph fitted with a cyclodextrin B enantioselective column under the following conditions: 120 °C for 2 min, 1 °C/min to 200 °C. Retention times were compared with synthetic standards (18) that had been identically derivatized.

Crystallization—Crystallization of cytochrome P450cin N242A mutant was carried out using sitting drop vapor diffusion at room temperature. Drops of 4 μl were prepared by adding 2 μl of 42 mg/ml protein and 2 μl of well solution for the substrate-free samples and 2 μl of 21 mg/ml protein and 2 μl of well solution for the substrate (cineole)-bound protein. These were equilibrated against a well solution of 0.8 m sodium malonate, pH 8.0.

A single plate crystal that was separated from a cluster of plates was used for data collection. Before freezing crystals in liquid nitrogen for cryogenic data collection, crystals were transferred to a solution containing 2.5 m sodium malonate as the cryoprotectant. Two data sets were collected for the N242A mutant in the presence and absence of substrate. Both the substrate-free and cineole-bound structures of P450cin N242A mutant belong to the monoclinic space group P2₁ with two molecules per asymmetric unit. However, they showed different cell dimensions (Table 1).

TABLE 1.

Data collection and refinement statistics

	N242A without cineole	N242A with cineole
Cell dimensions (Å)	a = 63.33, b = 68.28, c = 103.60	a = 59.40, b = 128.7, c = 69.0
Space group P21	β = 95.50	β = 97.8
Data resolution (Å)	2.0	3.05
Total observations	115,432	109,815
Unique reflections	46,409	17,907
Rsyma	0.071 (0.364)b	0.149 (0.491)b
〈I/σ〉	8.6 (2.3)b	5.5 (2.1)b
Completeness	0.78 (0.74)b	0.911 (0.801)b
Redundancy	2.5	6.1
R factorc	0.181	0.237
R-freed	0.227	0.313
No. of protein atoms	6,942	6,510
No. of hetero atoms	114	108
No. of waters	584	114

^aRsym = Σ|I – 〈I〉|/ΣI, where I is the observed intensity, and 〈I〉 is the averaged intensity of multiple symmetry related observations of the reflection

^bThe values in parentheses were obtained in the outermost resolution shell

^cR factor = Σ||Fo|| – |Fc||/Σ|Fo|, Fo and Fc are the observed and calculated structure factors, respectively

^dR-free was calculated with the 5% of reflections set aside randomly throughout the refinement

Data Collection and Structure Determination—X-ray diffraction data were collected on beam line 5.0.2 at ALS (Berkeley, CA) using an ADSCQ315, 3 × 3 CCD array detector. Optimization of data collection was guided by the STRATEGY function of MOSFLM (19). All data were reduced using DENZO and SCALEPACK, and rejections were performed with SCALEPACK (20). The substrate-free and -bound structures of cytochrome P450cin N242A mutant were solved using molecular replacement as implemented in AMORE in the CCP4 suite (21) and the wild type cytochrome P450cin structure complexed with cineole (Protein Data Bank code 1T2B) (15) as the search model.

Starting with the wild type model, side chains of cytochrome P450cin N242A mutant were fit into the electron density map using the graphical model building program O (22). However, because of the absence of the substrate (in the substrate free structure) compared with those in the wild type, it was necessary to manually adjust stretches of backbone. Many rounds of model building and simulated annealing refinements with CNS (23) structure generated a well improved electron density map that enabled the construction of some difficult regions, including the N terminus and F/G-loop region.

In the 2.0-Å substrate-free structure it was anticipated that water would coordinate the heme iron in the absence of substrate, but it was clear from the earliest Fo - Fc maps that something much larger interacts with iron. The shape and size of the electron density envelopes indicated that malonate binds to the heme iron in multiple orientations. The cineole bound structure on the other hand diffracted to only 3.05 Å. Although the resolution is limited, it was possible to clearly see the substrate, and the orientation of the substrate was deduced by performing several minimization rounds using CNS and different fittings using O.

Waters were added using the WATERPICK routine during several additional rounds of refinement and model building. The final cineole-bound model contains 6732 atoms, residues Met-8 —Glu-404, heme, 1,8-cineole, and 114 water molecules, whereas the substrate-free structure had a total of 7640 atoms, residues Met-8 —Glu-404, heme, malonate, and 548 water molecules.

The refined structure has a R_free values of 18.1 and 22.7% and R values of 22.6 and 31.8% for the substrate-free and -bound structures. Backbone geometry was checked in PROCHECK (24). All the residues fall into the allowed region in the Ramachandran plot.

The atomic coordinates have been deposited to the Protein Data Bank with the Protein Data Bank codes 3BDZ and 3BE0, respectively, for the substrate-free and -bound structures. Data collection and refinement statistics are provided in Table 1. Figures were prepared with PYMOL.

Molecular Dynamics—Molecular dynamics simulations were carried out with Amber 9.0. Charges and optimal geometry for cineole were obtained using density functional theory and the 6–31G^* basis set as implemented in JAGUAR, whereas heme parameters were provided by Dr. Dan Harris (25). The crystal structures were stripped of crystallographically defined water molecules except for one water molecule near the substrate in the N242A mutant. This new water molecule found in the mutant was retrained to its crystallographic position using a force constant of 10 kcal/mol/Å. The protein was solvated with TIP3 waters within a 30-Å radius sphere around the substrate. The entire solvated protein was energy minimized as follows. First, water and H atoms were allowed to move in a short 10-ps low temperature (50 K) MD4 simulation followed by 500 cycles of energy minimization for both protein and solvent. Second, both protein and solvent were allowed to move in a 10-ps 50 K MD simulation followed by 500 cycles of energy minimization. Finally, a 2-ns MD simulation was carried out at 300 K. Residues within a 20-Å sphere surrounding the substrate were allowed to move. Coordinates sets were saved every 10 ps, giving a total 200 coordinate sets. The molecular mechanics Poisson-Boltzmann surface accessibility (26) method was used for estimating the free energy of cineole binding. In this method several energetic calculations are carried out for the enzyme alone, the enzyme-substrate complex, and the substrate alone using the last 150 structures generated during the 2-ns MD simulation. This enables the ΔG of binding to be estimated, whereas ΔΔG = ΔG_binding(wild type) - ΔG_binding(mutant).

RESULTS AND DISCUSSION

Construction and Expression of N242A Mutant—The gene encoding the N242A P450cin mutant was produced by PCR utilizing a “mega-primer” method. Mutagenic primers were used in which the AAC codon for Asn-242 was replaced with the GCC codon for alanine. The identity of the mutant was verified by sequencing after initial cloning of the PCR construct into pUC18. Expression and purification of the encoded protein was carried out under previously reported conditions to give a homogeneous sample by SDS-PAGE analysis. The level of expression (90 nmol/liter original culture) was estimated using the extinction coefficient reported for wild type P450cin and was significantly lower than that observed for the wild type enzyme (500–2000 nmol/liter). The purity of the N242A mutant was followed using the ratio of the absorbance at 417 nm to that at 280 nm (Reinheitzahl ratio), and the purified protein had an A₄₁₇/A₂₈₀ ratio greater than 1.3, as did wild type P450cin.

Characterization of the N242A Mutant—The UV-visible absorption spectra of the wild type and N242A P450cin ferric forms both have a maximum at 417 nm. This maximum shifts to 392 nm upon the addition of cineole, but the percentage spin change observed in the mutant (30%) is smaller than that observed when wild type enzyme binds cineole. This spectral change is typical of a bacterial P450 when it binds substrate and results from a shift in the heme iron from low to high spin as a result of the expulsion of water from the active site, particularly the axial aqua ligand. The purified N242A protein displayed the CO-difference spectra characteristic of P450s (λ_max ∼ 450 nm), with no absorption observed at 420 nm due to an inactive form of the P450 (Fig. 2).

FIGURE 2.

UV-visible spectral data for N242A P450cin. A, low spin, ferric form of mutant λmax = 416 nm. B, Difference spectrum of CO complex of ferrous enzyme versus ferrous enzyme alone λmax = 449 nm. C, difference spectrum of ferric form of N242A P450cin with increasing concentrations of cineole versus N242A P450cin alone. AU, absorbance units.

The binding affinity of the N242A mutant for cineole was determined by UV-visible difference spectroscopy utilizing the spin state change and consequent spectral shift in the Soret that occurs when substrate binds. The dissociation constant of cineole from N242A P450cin (0.3 μm) was comparable with that observed for wild type P450cin (0.7 μm), with both binding cineole tightly. Similar unchanged (or increased) affinities have been reported for the analogous threonine to alanine mutants in P450cam and P450BM3 with their cognate substrates (27). The x-ray structure of P450cin with substrate bound shows an H-bond from Asn-242 to the ethereal oxygen of the cineole (Fig. 1). This must clearly be absent in the mutant, but its loss is not reflected in a change in the affinity of the protein for cineole. This is in line with the result observed for the Y96F P450cam mutant in which a H-bond from the protein to the substrate is lost without noticeable change in affinity (28, 29). Taken as a whole, these results suggest that the architecture of the active site and, in particular, the heme environment is maintained in the N242A mutant, a conclusion borne out by subsequent structural characterization to be described next.

Substrate-free Crystal Structure—As shown in Fig. 1, Asn-242 in the I helix in P450cin donates an H-bond to the substrate and, thus, provides the only polar interaction between protein and substrate. The N242A mutant structure is the same as the wild type structure with no significant differences other than the site of mutation. An additional benefit of the N242A mutant was access to the structure of the substrate-free protein since many attempts to crystallize the wild type substrate-free enzyme have been unsuccessful. There now are a growing number of examples of P450s where the enzyme adopts an “open” conformation in the absence of substrate and a “closed” conformation with substrate bound. The range of motions can be very large as in P4502B4, where some groups move as much as 18 Å (30, 31). We, therefore, were hoping to obtain a P450cin structure in the open conformation. Instead we found that the N242A substrate-free structure is very similar to the substrate-bound wild type structure, both being in a closed conformation. The only significant difference between the substrate-free and bound structures is in the F-G loop region (Fig. 1). This is similar to P450cam where the substrate-free and -bound structures were found to be the same with the only difference being higher thermal B factors in the F/G loop region in the substrate-free structure (32). This is perhaps not too surprising since the F and G helices as well as the F/G loop are known to undergo large changes in those P450 structures where open and closed structures have been determined (30, 31, 33). We also had anticipated that water would take the place of the substrate and one molecule would coordinate the heme iron as has been found in a number of other bacterial P450s. Instead we found a lobe of electron density consistent with a molecule much larger than water coordinated to the iron atom. Because the crystals had been soaked in 2.5 m malonate as a cryoprotectant, it seemed reasonable that the extra density is due to a malonate carboxyl oxygen coordinated to the heme iron. The size and shape of the electron density envelope is consistent with malonate but not in a single orientation. It, thus, appears that malonate coordinates with the heme iron but is able to do so in more than one orientation. The best model consistent with the electron density maps is shown in Fig. 3. Because a ligand is bound, it could be argued that P450cin adopts the closed conformation. However, malonate is small, charged, and does not occupy the substrate binding pocket. In addition there are no nearby Lys, Arg, or His residues that can provide electrostatic stabilization to malonate, so there appears to be little energetic incentive for the protein to close down around the iron-coordinated malonate. Thus, P450cin is similar to P450cam where only the closed conformation is found in the crystal structures of substrate-free and -bound complexes. The initial indication of how substrate binds to P450cam is that the crystallographic thermal parameters are much higher near the entry of the substrate entry channel (32). The lack of an open structure for either P450cam or P450cin does not prove that neither P450 adopts an open conformation but does suggest the equilibrium between the open and closed forms favors the closed conformation. Moreover, the substrates for P450cam and P450cin are small and spherical so large open/close motions are not required. This contrasts with P450BM3, which binds much larger long chain fatty acids, and the open and closed conformations have been well defined crystallographically (33, 34).

FIGURE 3.

2Fo - Fc omit electron density map of the substrate-free structure. The model that best correlates with the electron density is a malonate molecule in two possible orientations coordinated to the heme iron.

Substrate-bound Crystal Structure—Wild type P450cin hydroxylates cineole exclusively at the pro-R carbon to give the (S)-6β-hydroxycineole 1 (Fig. 1) (see below). The wild type structure at 1.7 Å shows clearly that the substrate adopts one orientation with the pro-R methylene about 5 Å from the iron. However, the resolution of the cineole-bound N242A structure is only 3.05 Å. Although the electron density of the substrate is clearly visible, there is some ambiguity in how to position the substrate. As a control, the wild type electron density was truncated to 3.05 from 1.7 Å (Fig. 4, A and B, left and right, respectively) to provide a more direct comparison with the N242A structure. As shown in Fig. 4A, even at 3.05 Å the substrate orientation in wild type P450cin is clear. However, in the N242A mutant the wild type substrate orientation is clearly inconsistent with the wild type orientation (Figs. 4, C–F). We first made the assumption that the substrate binds in the same orientation as in the wild type enzyme. The resulting fits after refinement are shown in Fig. 4C for molecule A in the asymmetric unit and Fig. 4E for molecule B. Neither fit is very good. We next optimized the fit to model-unbiased difference maps followed by refinement. The resulting fits for molecules A and B are shown in Figs. 4, D and F, respectively. Two things are clear from this exercise. First, the substrate does not bind in the same orientation as in the wild type enzyme. Second, the fit in molecule A is clearer and less ambiguous than in molecule B because of better resolved electron density for the substrate. One additional reason for favoring the molecule A orientation is illustrated in Fig. 5. In both molecules A and B there is a well ordered water molecule situated between the substrate and Asp-235, although the electron density for Asp-235 is too weak in molecule A to unambiguously determine its orientation. This water is about 3.7 Å from the substrate in molecule A and 3.1 Å from the substrate in molecule B. Orientation of the substrate such that the ethereal oxygen is close to this water molecule should be energetically more favorable in molecule A. The structure, thus, predicts that hydroxylation of the pro-S methylene should occur in the N242A mutant. Therefore, in the absence of the restraining H-bond between the substrate and Asn-242, the substrate is free to adopt an alternate conformation which places the ethereal oxygen in an optimal position for polar interactions with solvent. In addition to the missing H-bond in the mutant, another subtle structural change caused by the mutation is a slight bulge of the I helix near Gly-246 that places the I helix closer to the substrate binding site. This slight tightening around the substrate pocket may also disfavor the wild type substrate orientation.

FIGURE 4.

A series of omit electron density maps contoured at 1.0σ showing the fit of the substrate. A and B show the wild type structure at 3.05 and 1.7 Å, respectively. C and E show the maps for N242A molecules A and B with the substrate in the same orientation as in the wild type. D and F show the best fit of the substrate after refinement to the N242A molecules A and B. Even at 3.05 Å it is clear that the substrate does not adopt the wild type orientation in either molecules A or B of the mutant.

FIGURE 5.

2Fo - Fc electron density map contoured at 1.0σ showing the ordered water (Wat) molecule situated between the substrate and Asp-235 in both molecules A and B.

Catalytic Activity of N242A—A catalytically active system was reconstituted in vitro with P450cin or N242A P450cin, cindoxin (the P450cin FMN-containing redox partner), and E. coli flavodoxin reductase. Previous studies (35) had shown that flavodoxin reductase was able to mediate the transfer of electrons from NADPH and support catalytic turnover. The rate of NADPH consumption for the N242A mutant, determined by monitoring the disappearance of NADPH at 340 nm, is significantly reduced in comparison with the wild type enzyme. The mutant consumes NADPH at ∼25% (235 μm/min/μm P450) of the rate of the native enzyme. The effect of mutation of the highly conserved threonine varies in other bacterial enzymes. In P450cam, conversion of the Thr to Ala leads to NADH consumption at ∼85% of the wild type enzyme, whereas in P450BM3 the same mutant consumes NADPH at 20–100% of the wild type enzyme, depending upon the substrate. Despite the drop in the rate of NADPH consumption observed in the N242A mutant of P450cin, the rate is still relatively fast when compared with other P450 enzymes.

Catalytic Efficiency—Simply measuring NADPH consumption is, however, not an accurate measure of substrate oxidation as reducing equivalents can be shuttled into unproductive pathways, such as the production of hydrogen peroxide. Such processes are collectively termed uncoupling. There are three uncoupling products possible in the P450 catalytic cycle in the form of superoxide, hydrogen peroxide, and water production. The hydroxycineole produced was quantified using GC analysis, and the amount of hydrogen peroxide produced was determined via an iron thiocyanate assay (36). The percentage of uncoupling to hydrogen peroxide in P450cin and its N242A mutants is shown in Table 2. Preliminary analysis of the N242T mutant shows that that the degree of uncoupling is about the same as the N242A mutant.

TABLE 2.

Distribution of reducing equivalents

Enzyme	Product per NADPH consumed
	% Coupling	% Uncoupling to hydrogen peroxide	% Uncoupling to watera
Wild type	80 ± 2	12 ± 1	8 ± 3
N242A	49 ± 3	52 ± 7	0 ± 10

^aCalculated by subtraction of hydrogen peroxide and organic product formed from NADPH consumed

The native enzyme exhibits a strong monooxygenase activity converting ∼80% of the NADPH consumed to hydroxylated product while funneling only 12% of reducing equivalents into the formation of hydrogen peroxide. This agrees with that determined previously (35) and is comparable with other bacterial P450s, such as P450cam and P450BM3, which are ∼95% coupled under optimal conditions with optimal substrates (4, 37, 38). Although the N242A mutant is still able to catalyze the oxidation of cineole, it displayed a decrease in coupling with ∼49% of reducing equivalents going toward cineole oxidation and ∼52% into hydrogen peroxide production. The fact that the N242A mutant is not completely uncoupled is an interesting and unexpected outcome since the T252A mutant in P450cam results in 83% uncoupling (4, 5). If Asn-242 in P450cin plays an analogous role to the conserved threonine in other P450s, which stabilizes a water network in the active site and possibly stabilizes the hydroperoxy intermediate, then the mutant would be expected to be highly uncoupled.

One explanation for the unexpectedly high level of coupling is that the N242A mutant can use hydrogen peroxide in the so-called “peroxide shunt pathway.” In this “shunt” pathway, hydrogen peroxide is utilized directly by the Fe(III) enzyme to form the Fe(IV)=O species responsible for hydroxylation. In this scenario the N242A is significantly uncoupled, but the excess hydrogen peroxide generated is cycled back into the system via the shunt pathway, thus generating product. To assess this possibility, both P450cin and N242A mutant were combined with cineole and hydrogen peroxide and incubated at room temperature. GC/MS analysis indicated that no hydroxycineole was produced by either the native enzyme or the mutant, suggesting that P450cin is unable to utilize hydrogen peroxide to catalyze the hydroxylation of cineole.

These results imply that the N242A mutant is considerably less uncoupled than the corresponding threonine mutants in P450cam. This further suggests that Asn-242 in P450cin is not a functional replacement for the conserved threonine seen in most P450s and that its role in oxygen activation and hydroxylation is far less important than is the threonine in other P450s.

Product Profiles—Initial GC/MS analysis of the products formed from cineole oxidation by the N242A mutant revealed that little of the product 1 (Fig. 6) formed by the wild type enzyme was produced by the mutant. Moreover, the N242A mutant also produced a new hydroxycineole as the major product (Fig. 6A). It was clear by inspection of the mass spectral fragmentation pattern (Fig. 6B) that another hydroxycineole was formed, and literature comparison suggested that one of the 6-hydroxycineoles 3 or 4 (Fig. 7) was produced. Four different hydroxycineole isomers are possible (Fig. 7) with 1/2 and 3/4 having an enantiomeric relationship. The initial GC/MS analysis was unable to differentiate enantiomeric compounds, and it was, therefore, necessary to synthesize all four possible isomers of an enantio-enriched form and compare them with the turnover products of the P450cin mutants to determine the regio- and stereoselectivity of the reaction using enantioselective GC. The four isomers 1–4 were prepared from enantiopure α-terpineol according to the method of Carman and Fletcher (18) with the stereochemistry of the starting α-terpineol determining which enantiomer is formed (18). The four hydroxycineoles were then esterified with trifluoroacetic anhydride to yield the trifluoroacetate esters as these gave sharper, better-resolved peaks when analyzed by GC (Fig. 7).

FIGURE 6.

A, GCMS profile of products formed from the oxidation of cineole by P450cin and N242A mutant. B, MS fragmentation of (S)-6β-hydroxycineole 1 (right panel) and (R)-6α-hydroxycineole 3 (left panel).

FIGURE 7.

Enantioselective GC analysis of trifluoroacetate-derivatized products formed from the oxidation of cineole by P450cin and N242A mutant as well as synthetic standards of 1–4.

GC comparison of the products of enzyme turnover with the synthetic standards clearly identifies 1 as the only product of P450cin catalyzed oxidation (35). In contrast, the N242A mutant converts cineole to 1, 3, and 4, although 1, the natural product, only accounts for ∼5% of the products (Fig. 7). Surprisingly, 4 is the major product, accounting for ∼90% of the products, whereas isomer 3 makes up the remaining 5%. The molecule A orientation of the substrate found in the crystal structure also is consistent with 4 as a major product. Such a result is unprecedented in P450s when the conserved threonine is mutated to an alanine. Hydrocarbon hydroxylation greatly decreases in two well studied cases, P450cam and P450BM3, but the regio- and stereochemical profile of the products formed by the mutant enzyme are identical to the wild type enzyme (4, 39). This is consistent with a decreased ability to form the active ferryl species. The results observed with the N242A mutant of P450cin are more similar to those observed with Y96F mutant of P450cam (28). Tyr-96 provides a hydrogen bond to the carbonyl of d-camphor, the natural substrate for P450cam, assisting in regiochemical control of the oxidation. In the Y96F mutant of P450cam, removal of this hydrogen bond results in the formation of a number of new products. Even so, the effects are not as dramatic as seen with N242A mutant of P450cin since the major product (92%) from the P450cam mutant is the same as produced by the wild type enzyme (28).

Molecular Dynamics—To complement the crystallographic work which has limited resolution for the substrate complex, we next carried out a series of 2-ns MD simulations to explore the dynamic behavior of the substrate in both wild type P450cin and the N242A mutant. Structures were saved every 10 ps giving a total of 200 sets of coordinates. Fig. 8 shows the beginning crystal structures together with relevant distances, whereas Fig. 9 shows the starting orientation of thesubstrate based on the crystal structures (yellow) superimposed on the MDaverage structure (green). In the wild type structure (Fig. 9A) the substrate does not reorient, and pro-R remains closest to the iron to give the observed product 1. In molecule A of the N242A mutant MD also does not result in a major reorientation of the substrate, yet the MD average structure places the pro-S methylene slightly closer to the iron (Fig. 9B). In molecule B of the mutant the substrate undergoes a substantial reorientation to closely match the orientation in molecule A. In addition, the pro-S methylene remains closer to the iron over the course of the MD simulation. These results indicate that the molecule A orientation is favored, which is consistent with the crystallography since, as noted earlier, the electron density for molecule A is better defined than in molecule B. Both the MD and crystallography also are consistent, with 4 being a major product, consistent with the analytic data (Fig. 7).

FIGURE 8.

The refined structures of wild type (A) compared with the mutant molecules A and B, panels B and C, respectively. The distance between the iron and the pro-R and pro-S carbons are indicated.

FIGURE 9.

Superimposition of the crystal structures (yellow) on the MD averaged structures (green). Panels A, B, and C show the wild type, mutant molecule A, and mutant molecule B, respectively. Also indicated are the MD averaged pro-R methylene-iron and pro-S methylene-iron distances.

Another question we wished to address computationally was whether Asn-242 contributed significantly to the overall free energy of binding. The 2-ns MD trajectories and the MM-PBSA method were used to estimate the ΔΔG (ΔG_bindingwild type - ΔG_bindingmutant) of cineole binding. The computed ΔΔG values for molecule A and B orientation are small, 0.5 and 0.1 kcal/mol, respectively, which is consistent with the only a small change in K_D estimated by spectral titration data. This might at first suggest that Asn-242 contributes very little to the binding free energy. However, in the mutant the substrate reorients to position the ethereal oxygen in a more polar environment. Water-substrate interactions, thus, replace the Asn-242-substrate H-bond, resulting in no net change in the free energy of binding.

These types of comparisons, however, do not directly address the contribution of the Asn-242-substrate H-bond in the wild type substrate orientation. To make this comparison we carried out MM-PBSA calculations on single energy-minimized structures of wild type and the in silico-generated N242A mutant with the substrate held in the crystallographically determined wild type orientation using a 10-kcal/Å restraint. The mutant-wild type ΔΔG using this protocol is 2.4 kcal/mol, indicating that Asn-242 does contribute favorably to substrate binding in the wild type substrate orientation.

Conclusions—Our results show that Asn-242 in P450cin is not a functional mimic of the highly conserved Thr found in many P450s. Mutating Thr-252 in P450cam results in uncoupling such that the majority of NADH-reducing equivalents are not funneled into productive substrate hydroxylation (4, 5). This is generally thought to be due to the disruption of the proper H-bonded network that ensures proper protonation of the iron-linked dioxygen and, possibly, stabilization of the Fe(III)-OOH intermediate (3). Because the N242A mutant is highly coupled, we can conclude that Asn-242 is not a critical part of the oxygen activation machinery in P450cin. There is one small caveat to this argument. Water could conceivably take the place of the missing Asn-242 side chain in the oxy complex, which might then provide the proper H-bonding links to the Asn-242 side chain.

Although there is some ambiguity in precisely how the substrate should be oriented in the x-ray crystal structure of the N242A mutant, the orientation in molecule A is preferable since the electron density is better defined than in molecule B, and the MD simulations favor the molecule A orientation. In addition, the molecule A orientation would give the correct product, 4 (Fig. 1). The reason the substrate reorients in the N242A mutant is to place the substrate ethereal oxygen atom in a more polar milieu where it can interact with solvent. It, thus, appears that the primary role of Asn-242 is to control regio- and enantio-selective hydroxylation of the substrate via H-bonding to the ethereal oxygen of the substrate. In this regard Asn-242 is analogous to Tyr-96 in P450cam. When Tyr-96 is removed and replaced with phenylalanine and the orienting H-bond is lost, regioselective hydroxylation also is lost, resulting in the production of several hydroxylated products (28).