Water Oxidation by a Cytochrome P450: Mechanism and Function of the Reaction

P450cam (CYP101A1) is a bacterial monooxygenase that is known to catalyze the oxidation of camphor, the first committed step in camphor degradation, with simultaneous reduction of oxygen (O2). We report that P450cam catalysis is controlled by oxygen levels: at high O2 concentration, P450cam catalyzes the known oxidation reaction, whereas at low O2 concentration the enzyme catalyzes the reduction of camphor to borneol. We confirmed, using 17O and 2H NMR, that the hydrogen atom added to camphor comes from water, which is oxidized to hydrogen peroxide (H2O2). This is the first time a cytochrome P450 has been observed to catalyze oxidation of water to H2O2, a difficult reaction to catalyze due to its high barrier. The reduction of camphor and simultaneous oxidation of water are likely catalyzed by the iron-oxo intermediate of P450cam, and we present a plausible mechanism that accounts for the 1∶1 borneol:H2O2 stoichiometry we observed. This reaction has an adaptive value to bacteria that express this camphor catabolism pathway, which requires O2, for two reasons: 1) the borneol and H2O2 mixture generated is toxic to other bacteria and 2) borneol down-regulates the expression of P450cam and its electron transfer partners. Since the reaction described here only occurs under low O2 conditions, the down-regulation only occurs when O2 is scarce.


Introduction
Cytochrome P450 enzymes (P450s or CYPs) belong to a family of heme-thiolate enzymes that couple the reduction of oxygen to the oxidation of non-activated hydrocarbons [1]. The catalytic cycle of cytochrome P450 cam [2] (Fig. 1a) starts with binding of camphor to the resting enzyme 1 and expulsion of the axial water molecule to form 2. Enzyme-substrate complex 2 accepts two electrons from the nicotinamide cofactor (NADH) via two redox partner proteins: an iron-sulfur protein, putidaredoxin (PdX), and a flavoprotein, putidaredoxin reductase (PdR) [3]. P450 utilizes the two electrons to reduce oxygen, O 2 , in a stepwise manner, via intermediates 3 and 4 [4,5]. This leads to the formation of peroxo complex 5, which is protonated to give hydroperoxo complex 6. Protonation of the distal oxygen of 6 and elimination of water gives rise to a high valent iron-oxo complex 7 known as compound I (Cpd I) [6] (Fig. 1a). The oxygen from 7 is then inserted into a C-H bond of the substrate, giving an alcohol product complexed to the iron, 8. The catalytic cycle is complete when water displaces the product.
P450 cam (CYP101A1) enables a strain of Pseudomonas putida (a soil bacterium) to use (1R)-(+)-camphor 9 (Fig. 1) as a carbon source, and it oxidises camphor at the 5 th position to give 5-exohydroxycamphor 10 and 5-ketocamphor 11 ( Fig. 1) [12]. Here we describe how P450 cam can oxidize water to H 2 O 2 and simultaneously reduce camphor to borneol 12 (Fig. 1b) under low O 2 conditions, and how borneol regulates the expression of the P450 cam system. Catalytic water oxidation is difficult to achieve, because the reaction is endothermic and has a large barrier. [13,14,15] To our knowledge, this is the first description of a cytochrome P450 oxidizing water.
We have observed that at low oxygen concentration, regardless of whether Cpd I forms via reduction of O 2 or by shunting with oxidants, P450 cam not only produces the oxidation products 10 or 11, but can also reduce camphor to borneol (Fig. 1b) [16]. We have interpreted this reaction to give P. putida an ecological advantage over other non-camphor metabolising bacteria because borneol is bactericidal to non-P450 containing bacteria, but not to P. putida [16]. In this paper, we present the mechanism of the camphor reduction reaction and the regulatory effect of borneol on the expression of P450 cam .

II) Methods
Deuterium ( 2 H) NMR spectra were recorded on a Bruker AVANCE II 600 MHz spectrometer (operating at 92.124 MHz). A Bruker 5 mm TCI cryoprobe was used with samples maintained at a temperature of 298 K. 2 H field-locking and field sweep were turned off. Samples were contained in 3 mm diameter MATCH nmr tubes filled to 40 mm (volume ca. 185 mL). Acquisition details: 10,240 transients summed, spectral width 15 ppm, transmitter offset 6.5 ppm, 11054 complex points acquired, 15 degree pulse with recycle delay of 1 s between transients, no decoupling of 1 H during FID acquisition. Acquisition time was 14.2 h per spectrum.
The 17 O NMR spectra were run on a Bruker AVANCE III 500 MHz NMR spectrometer (operating at 67.808 MHz) equipped with a Bruker 5 mm TBO probe and samples were maintained at a temperature of 298 K. Samples were contained in 5 mm diameter nmr tubes filled to 50 mm (volume ca. 600 mL). Acquisition details: 1,000 or 25,000 transients summed, spectral width 503 ppm, 17 O transmitter offset 50 ppm, 1 H transmitter offset 4.78 ppm, 32768 complex points acquired, 90 degree pulse with recycle delay of 1 s between transients, and inverse-gated Figure 1. The catalytic cycle of P450 cam and the formation of the products, 10, 11 and 12. a) R-H represents the substrate, camphor. i, ii, iii and iv represent the peroxide shunt reaction, four-electron uncoupling, two-electron uncoupling, and the loss of superoxide. b) Under highly oxygenated conditions, P450 cam hydroxylates camphor 9 to 5-exo-hydroxy camphor 10 and further to 5-ketocamphor 11, whereas under low oxygen conditions, P450 cam reduces camphor to borneol 12. doi:10.1371/journal.pone.0061897.g001 WALTZ-16 composite pulse decoupling of 1 H during FID acquisition. Acquisition time was 12 min per spectrum or 300 min (when catalase was present). The chemical shifts (d) for all compounds are listed in parts per million using the NMR solvent as an internal reference (0 ppm for 17 O or 4.78 ppm for 2 H).
The lysis steps of P450 and PdR remained the same as described for P. putida [10] except that 1 mM camphor was added to the buffer in which P450 culture was lysed. The dialysed lysate of P450, or PdR was individually subjected to a 20% ammonium sulphate cut to remove the cell debris. The 20% supernatant was then carried forward to 45% ammonium sulphate saturation to isolate the protein. The 20-45% pellet was resuspended in T-100 buffer (50 mM Tris, 100 mM KCl, pH 7.4), camphor (1 mM) was added in the case of P450 and purified by DE-52 (anion exchange column) using a linear gradient with buffer T-100 to T-400, 1 mM camphor and 1 mM b-mercapto ethanol (P450 only) at 1 mL min 21 . The fractions with high absorbances at l 392 (in the case of P450), l 454 (in the case of PdR) were checked with SDS-PAGE. The collected fractions were pooled and concentrated using an Amicon ultrafiltration cell equipped with a YM-10 membrane and the concentrated protein was individually loaded onto a S-100 column, eluted with T-100 buffer, 1 mM sucrose, 1 mM camphor (P450 only) at 1 mL min 21 . SDS-polyacrylamide gel electrophoresis showed a single band for P450 and PdR.
In the case of PdX, cells from 2 L of culture were lysed in lysis buffer (0.25 M NaCl, 20 mM Tris/HCl, pH 8.0). Lysozyme (10 mg mL 21 ), DNase (2 mg, Sigma), and RNase (10 mg, Sigma) were added and the solution was stirred for 30 min at 4uC. The lysate was sonicated with 50% duty cycle for 10 minutes, stirred for 10 min at 4uC, and homogenized with a pestle. The homogenized cells were then harvested by centrifugation (105006g, 30 min) and dialysed with frequent changes of lysis buffer followed by further purification by ammonium sulphate precipitation. The dialysed lysate was subjected to a 20% ammonium sulphate cut to remove the cell debris. The 20% supernatant was then carried forward to 90% ammonium sulphate saturation overnight to isolate the protein. The 20-90% pellet was resuspended in 5 mL of rinse buffer (20 mM Tris HCl, pH 8.5), dialysed against this buffer for 3 h and harvested at 5000 rpm for 5 min. The dialysed supernatant was loaded on a ,5 cm Ni 2+ -His bind column and eluted with strip buffer (10 mL63), low imidazole buffer (10 mL62), high imidazole buffer (10 mL64). The procedures for the enzymatic assays with the recombinant proteins, as well as the superposition and docking procedures, using Molecular Operating Environment (MOE, Montréal, Canada) are included in Material S1.

I) Reaction Conditions Leading to Formation of Borneol
We have observed that borneol forms as a major product of P450 cam when camphor is present and the O 2 concentration is low (O 2 #2 mg/L, #63 mM). In vivo, this occurs when cultures are poorly aerated [16] and, in vitro, this occurs when the buffer is sparged with argon in an open vial. In contrast, the known oxidation products 10 and 11 form at high O 2 concentrations (,9 mg/L = 284 mM). In vivo, this occurs when cultures are well aerated [16] and, in vitro, this occurs when pure O 2 is bubbled into the buffer (Fig. 1b). To map the mechanism of the reduction, we have performed experiments with the recombinant proteins (P450 cam , PdX, and PdR), isolated from expression in E. coli (Table 1). Assays were carried out in phosphate buffer (50 mM phosphate, 150 mM K + , pH 7.4), with NADH and camphor. Our extinction coefficient values were used for the calculation of the enzyme concentration (Table S1). Under high oxygenation (with pure O 2 bubbled into the buffer), we observed 5-exo-hydroxy camphor as a major product ( Table 1, entry 1). Similar experiments under mid-range oxygenated conditions (with airtreated buffer) yielded borneol as the only product (Table 1 entry 2). The formation of borneol under these conditions was 34-fold less compared to 5-exo-hydroxy camphor that formed under highly oxygenated conditions, and this could be because of the slower formation of iron-oxo species (Compound I).
Under poor buffer oxygenation, in the absence of NADH, P450 cam shunted with m-CPBA (Fig. 1a, pathway ''i'') reduced camphor to borneol (Table 1 entries 3 and 4). The observation that borneol formed in the absence of NADH indicates that NADH is not the source of electrons for the reduction reaction. Furthermore, shunted P450 cam under high buffer oxygenation gave more 5-ketocamphor than borneol (Table 1

II) Source of the 2-H in Borneol
Because NADH is not the source of electrons for the reduction of camphor, the source of the hydrogen attached to C-2 of borneol was further investigated in assays using deuterated phosphate buffer (50 mM phosphate in D 2 O, 150 mM K + , pD 7.4). Using recombinant proteins (P450 cam , PdR, and PdX), under mid-range oxygenated conditions (with air), we detected the enzymatic conversion of camphor to 2-D-borneol 12D (Fig. 2a, Table S2) using 2 H NMR. We also detected 5-ketocamphor, as well as the depletion of NADH (Table S2). Similar experiments using NADD (deuterated nicotinamide cofactor) in non-labeled phosphate buffer did not yield 2-D-borneol [18]. Enzymatic assays in deuterated buffer (with recombinant P450 cam , shunted with m-CPBA, in the absence of NADH) also yielded borneol that was deuterated at C-2 (H exo ) (Fig. 2a, Table S2). All these experiments lead to the conclusion that water is the source of H exo attached to C-2 in borneol formed by P450 cam .

III) 17 O NMR of H 2 O 2
If camphor is reduced to borneol by electrons from water, then water should be oxidized to hydrogen peroxide. We observed H 2 O 2 along with borneol, approximately in a 1:1 stoichiometric ratio when P450 cam was shunted with m-CPBA (Table 1, entries 3- The reaction mixture contained recombinant P450 cam , PdR and PdX and NADH. Oxygen (99%) was bubbled into the buffer for 60 seconds before the assay. The 4e 2 uncoupling was calculated by taking the difference between the total NADH required and observed. 2 The reaction mixture contained recombinant P450 cam , PdR, PdX, and NADH. Air (charcoal filtered) was bubbled into the buffer before the assay. 3 The assay was performed using recombinant P450 cam and m-CPBA as a shunt agent. 4 The buffer was sparged with argon (99%). 5 The buffer was treated with oxygen (99% pure, Sigma Aldrich) and assays were performed using camphor. doi:10.1371/journal.pone.0061897.t001   (Fig. 2b(i)) which matched the chemical shift of H 2 17 O 2 reported in the literature [20] and of our prepared standard [19]. The effect of pH on the chemical shift of hydrogen peroxide was also checked (Fig. S1). Controls (in the absence of m-CPBA, enzyme or substrate) were run simultaneously, and this resonance was not detected (Figs. 2b(ii), 2b(iii) and 2b(iv)), which led us to conclude that the new peak could not have come from the hydrolysis of m-CPBA. When catalase (an enzyme that disproportionates H 2 O 2 to water and O 2 ) was added to the reaction mixture, the resonance at 178 ppm disappeared (Fig. S2 b), confirming that the 178 ppm resonance is due to H 2 17 O 2 .

IV) Kinetic Isotope Effects (KIE)
The reaction catalyzed by P450 cam , shunted with m-CPBA in D 2 O, gave 2-D-borneol at a much slower rate than the same reaction performed in normal water. The magnitude and temperature independence of the 1 H/ 2 H kinetic isotope effect (KIE) of ,50 (Fig. 3a, Table S3) suggests that hydrogen transfer through tunnelling could occur at the rate-determining step in the reduction of camphor to borneol [21,22,23]. In contrast, the KIE ( 1 H/ 2 H) for hydrogen peroxide formation are much smaller, suggesting that this product does not form at the rate-limiting step (Fig. 3b, Table S4).

V) Reduction Mechanism
Borneol formation under shunt conditions is saturable, with a K M = 699688 mM and k cat = 426620 min 21 for camphor (Fig. 3c). Similarly, ketocamphor formation under oxygenated shunt conditions is saturable with a K M = 83610 mM and k cat = 461614 min 21 for camphor (Fig. 3d). In D 2 O buffers, the formation of D-borneol was saturable with a K M = 8026107 mM and k cat = 960.4 min 21 for camphor (Fig. 3). Ketocamphor formation under oxygenated shunt conditions is saturable with K M = 11866 mM and a similar k cat = 46566 min 21 for camphor (Fig. 3). From control experiments we know that reducing P450 cam and camphor with dithionite does not yield any borneol (Fig. S3). Therefore, borneol formation requires oxidation of P450 cam , either through shunting or through intermediates 2 to 7 of the catalytic cycle (Fig. 1a). Therefore, Cpd I must be involved in both borneol and ketocamphor formation (Fig. 1a).
We propose that water reduces and protonates Cpd I as a first step in the borneol cycle, giving protonated Cpd II 13 and a hydroxyl radical (OH N ) (Fig. 4). The formation of OH N in water has been estimated from electrochemical data [24], and the formation of species 13 from Cpd I has been estimated at DGu = -410 kJ/mol [25]. Therefore, the first step of the proposed reduction mechanism (Steps I and II, Fig. 4) involves the abstraction of a hydrogen atom from water by Cpd I to form the Fe(IV)-OH complex 13 (Fig. 4), which is favourable (DH,2160 kJ/mol) (Fig. 4). Three water molecules are known to be poised above the Fe-porphyrin and are held in place by hydrogen bonds to Thr 252, Asp 251 and Glu 366 [26], so it is plausible that Cpd I could attack water instead of camphor. Next, we propose that the hydroxyl radical combines with the water molecule to yield hydrogen peroxide and a hydrogen atom (Step III). By our estimate, this step is highly unfavourable (DHu > 570 kJ/mol, Material S1, section 2.9). Simultaneous transfer of the hydrogen atom from step III to the carbonyl group of camphor forms a borneol radical (Step IV, Fig. 4). A non-strained ketone such as acetone reacting with a hydrogen atom has a potential of approximately 22 V (DGu is +173 kJ/mol) [27]. However, because camphor is quite a strained ketone, and that strain is relieved by the reduction, we have estimated this reaction to be slightly favourable (DHu,27968 kJ/mol, Material S1, section 2.9). Finally, the transfer of a hydrogen atom from protonated Cpd II to the borneol radical forms borneol and Cpd I (Steps V and VI, Fig. 4) (DHu > 13 kJ/mol), completing the ''borneol cycle''. The net reaction is endothermic, with DHu > 30568 kJ/mol (Material S1, section 2.9, Fig. S4a).
The involvement of OH-radicals in H 2 O 2 formation has been proposed previously for electrolytic catalysts that oxidize water to O 2 (via an intermediate peroxide) [13] and also for a recently discovered water oxidation catalyst that produces H 2 O 2 during electrolysis [14,15]. Interestingly, the latter MnO x catalyst stops the water oxidation process at H 2 O 2 , because the peroxide is solvated and stabilized by hydrogen bonding to ethylamine and/or water in the electrolyte. [14,15] Analogously, here we propose that hydrogen bonding within the water cluster in the hydrophobic P450 cam active site is essential for stabilization of the various reactive intermediates and of the H 2 O 2 formed. The turnover numbers with regard to H 2 O 2 formation we have observed are ,7, whereas the electrocatalytic systems give turnover numbers of 20-1500 for complete water oxidation to O 2 . [13] This difference arises because P450 cam only has access to thermal energy to perform this ''uphill'' reaction, whereas the electrocatalytic systems are run at overpotentials. [13,14,15].
The proposed mechanism accounts for our observation that borneol and hydrogen peroxide form in a 1:1 stoichiometric ratio, provided 2-electron uncoupling is negligible ( Table 1, Table S5). Given that Cpd I appears to be involved in the borneol cycle, our previous [16] and current data also suggest that Cpd I might be regulated by O 2 levels: under high oxygenation, Cpd I sequentially hydroxylates camphor to 10, or 10 to 11 (Fig. 1b). Under poor oxygenation, Cpd I enters the borneol cycle that couples the oxidation of water to H 2 O 2 to the simultaneous reduction of camphor to borneol (Fig. 1b). The borneol cycle is independent of how Cpd I is generated: through the reduction of O 2 or the shunt pathway (Fig. 1a). Borneol formation was seen with all the shunt agents tested (m-CPBA, cumene hydroperoxide, periodate and bleach; Table S5).
In assays with CYP3A4 (a human cytochrome P450) under shunt conditions, 5-exo-hydroxy camphor formed as a major product. There were no detectable amounts of borneol, suggesting that the reduction cycle is specific to P450 cam (Material S1, section 2.5). A BLAST search against the P450 cam sequence revealed many other bacterial cytochromes P450 that show sequence identities for the three residues that hold a set of water molecules above the porphyrin (Asp 251, Thr 252 and Glu 366 in P450 cam , Fig. S5), as well as for the hydrophobic residues that are involved in O 2 binding (see below and Material S1, section 2.10). Superposition of P450 cam (1DZ4, [28] on CYP3A4 (1TZN, [29] reveals that the active site of CYP3A4 is much larger and more polar than that of P450 cam . In P450 cam , camphor is surrounded by closely packed hydrophobic residues, which could form a cage around the reactive intermediates. (Fig. 5) The only water in the active site of the camphor-bound structure is in the water channel between Glu 366 and Thr 252, whereas the CYP3A4 active site can hold numerous water molecules in the absence of a ligand (Fig. 6). Docking of camphor into the active site of CYP 3A4 reveals the camphor bound near the porphyrin, capped by five phenylalanine residues and surrounded by Arg 212, Ser 119, Ile 120, Ile 301 and H-bonded to Arg 105 (Fig. 6). This more open arrangement may not provide the necessary stabilization for water oxidation to occur. Furthermore, the different positioning of the camphor within the active site may also preclude its utilization as an electron acceptor during the water oxidation and, therefore, the reaction was not observed in CYP3A4.
Amunom et. al. have stated that mammalian P450s can reduce 4-hydroxynonenal to 1,4-dihydroxynonenal under low oxygen conditions, [30] similar to our results presented in this paper. However, differences in the reaction mechanisms can be associated with the different reacting species of the P450. They proposed that the reduction they observed occurs via the ferrous (Fe (II)) species of P450, where the electron source is from NADPH through the NADPH-P450 reductase. We have shown that, in our case, the ferrous species is not involved (Fig. 4) We therefore propose that the reduction of camphor to borneol involves the iron-oxo species where the source of electrons is from water, and not from NADH. Kaspera et. al. have stated that P450BM-3 from Bacillus megaterium can reduce p-methoxy-benzaldehyde to methoxy-benzalcohol [31]. Electrons for this reaction are provided by a direct hydride transfer from NADPH to the aldehyde, or by NADPH reduction of the flavin mononucleotide in the reductase, which then reduces the substrate. In comparison, we found that the source of electrons in our case is clearly from water, and not from a direct hydride transfer.

VI) Control Experiments with Reactive O 2 Species/ quenchers
In vitro assays with P450 under shunt conditions were performed with a free radical quencher (BHT), a free metal chelator (EDTA), catalase and superoxide dismutase, to determine whether free reactive oxygen species are involved in borneol formation ( Table 2). Under shunt conditions using m-CPBA, in Ar-sparged buffer, the enzyme formed more borneol than 5-ketocamphor (     Table 2, entry 4). To check if the radicals proposed in the mechanism of the reduction (Fig. 4) can diffuse out of the P450's active site, we experimented with BHT, and noticed no significant effect (Table 2, entry 5). This suggests that any radical species involved in the borneol cycle do not exist long enough to diffuse out of the active site of P450 cam . To test if a metal impurity plays a role in our assays under shunt conditions, experiments were performed with EDTA, and we detected no effect on borneol formation (Table 2, entry 6). To check if free iron (outside of the active site) plays a role in reduction reaction, experiments were performed with ferrous sulphate and m-CPBA, in the absence of P450 cam , and we did not detect borneol or 5-ketocamphor (Table 2, entry 7). These experiments suggest that the reduction of camphor to borneol is catalysed by P450 cam alone, does not involve any adventitious metal species outside of the P450 active site and does not involve the diffusion of reactive oxygen species, other than the product H 2 O 2 , out of the active site.

VII) Role of Oxygen in the Borneol Cycle
The reaction path taken by P450 cam (camphor oxidation vs. borneol cycle, Fig. 1b) is controlled by oxygen concentration. Oxygen could exert its effect in two ways: 1) by affecting the interaction of P450 with its redox partners, or 2) by directly interacting with P450. Our results demonstrate that the former cannot be the case, because the effect was seen in the absence of PdX and PdR (Table 1). Therefore, P450 cam must bind O 2 not only for catalysis, but also for allosteric regulation.
Recently, cytochrome P450 2E1 has been shown to form endoperoxide rearrangement products when reacted with 1,1,2,2tetramethylcyclopropane [32]. This suggests that there must be O 2 bound in that enzyme near the active site, which reacts with the rearranged radical formed by H-atom abstraction from 1,1,2,2tetramethylcyclopropane. Cytochromes P450 are known to be allosterically regulated by their substrates or co-substrates [33]. Studies with other O 2 -utilizing enzymes, such as diiron monooxygenases [34], laccase [35] and amine oxidase [36] have revealed that O 2 can be bound in hydrophobic tunnels that are separated from the access channel for the other substrates of these enzymes. In P450 cam , a hydrophobic O 2 entry channel and two O 2 binding cavities have been identified in Xe-treated crystals [37]. Two Xe atoms are bound near the porphyrin edge in a hydrophobic pocket lined by F163, A167, heme allyl, I220, A219, C242 and L245. The other two Xe atoms appear bound in a crevasse lined by L371, T370, L257, M261, water and S260 (first Xe) and I275, K372, T376, L375, L371, P278 and I281 (second Xe). This O 2 binding site in P450 cam is located near the edge of the porphyrin, near the water channel ( Fig. 6 and 1 A. and B).
We have found a hydrophobic tunnel in P450 cam that includes the Xe binding sites, using MOLEonline 2.0 [38] on the structure believed to represent the P450 cam oxo complex (1DZ9). The binding sites are good candidates for O 2 binding because they are hydrophobic and distinct from the substrate access route [11,37]. Also, the sites are good candidates for allosteric regulation of P450 because they are near the plane of the porphyrin. It is plausible Figure 6. Sites in P450 cam and in CYP3A4 with camphor docked. a) Oxygen binding site in P450 cam (residues shown in red), with superimposed residues in CYP3A4 shown in pink. The porphyrin of P450 cam is gray, and the one for CYP3A4 is brown. b) Water channel in P450 cam (residues shown in red), with superimposed residues in CYP 3A4 shown in pink. The view in a) and b) are from a similar angle, to emphasize the closeness of the O 2 binding site and the water channel in P450 cam . c) and d) Camphor docked into the active site of CYP3A4 (orthogonal views). The H-bond from Arg 105 to the camphor ketone can be seen in the lower right portion of d). doi:10.1371/journal.pone.0061897.g006 that an O 2 molecule bound near the heme could affect the reactivity of Cpd I.
The O 2 binding site in P450 cam is closer to the porphyrin than the equivalent site in CYP3A4, and the O 2 binding site is lined by different residues (Fig. 6). Furthermore, the O 2 binding site in P450 cam is close to the water channel, the only source of water in the camphor-filled active site of P450 cam . It is reasonable to hypothesize that the O 2 site, the porphyrin, the water channel and the tightly held camphor, all of which are near each other, could affect each other by allosteric effects in P450 cam .
It is interesting to note that the K M for ketocamphor formation under high O 2 concentration is 9-fold lower (see above) than that for borneol formation under low O 2 concentration. This suggests that camphor binding and possibly positioning might be affected by O 2 concentrations. Surprisingly, the k cat is the same for both reactions, even though there appears to be a larger barrier in the borneol cycle than in the normal oxidation reaction. This largerthan-expected k cat suggests that, consistent with the observed KIE, H-atom tunneling is occurring in the borneol cycle. Under high O 2 concentrations using D 2 O as the solvent, 5-ketocamphor (Table S2) was detected as the only product suggesting that deuterium atoms from the solvent do not participate in that reaction. Steady-state kinetic assays for ketocamphor formation in D 2 O buffers resulted in similar k cat as in H 2 O buffers. In contrast, a 60-fold decrease in k cat (with a similar K M ) was detected for borneol formation (Fig. 3). This illustrates that the solvent molecules participate only in the borneol formation, but not in ketocamphor formation.
There are two ways the cycle could end. Cpd I might oxidize a nearby enzymatic residue or, alternatively, the borneol radical might abstract a H-atom from water, giving borneol and OH N , and the hydroxyl radical could rebind with the OH N bound in Cpd II-H, to give a second H 2 O 2 and the ferric enzyme (Fig. S4 b).

VIII) Adaptive Advantage of Borneol and H 2 O 2 to P. putida
Previously we have determined the effect of borneol and camphor on the growth of P. putida and E. coli [16]. To determine the effects of hydrogen peroxide, we have tested the toxicity of H 2 O 2 and a 1:1 stoichiometric mixture of borneol and H 2 O 2 on both P. putida and E. coli, a bacterium that lacks cytochrome P450 [39] (Figs. S6 and S7). The borneol/H 2 O 2 mixture was lethal to E. coli and slightly toxic to P. putida (Fig. S7). The latter observation prompted us to explore whether borneol affects the expression of the P450 cam system.
The camphor metabolism pathway, of which P450 cam catalyzes the first step, is encoded on the Cam plasmid under the control of the Cam repressor. This repressor dissociates from the upstream control region of the Cam operon upon binding of camphor, ensuring that the entire operon is expressed when camphor is present [40]. To study this induction, we cultured P. putida in the absence of camphor for seven generations, then divided the culture and treated the sub-cultures as shown in Fig. 7a (with camphor, borneol or vehicle, dimethyl sulfoxide (DMSO)). We detected a steep increase in the characteristic absorption bands of P450 cam , PdR, and PdX only in the culture induced with camphor, about 80 min after initial induction. Absorptions plummeted approximately 60 min after the addition of borneol to camphor-induced culture(s) (Fig. 7b, Figs. S8 and S9). This decrease in P450, PdR, and PdX expression must be due to the borneol addition, because the camphor-induced cultures that did not receive borneol expressed significantly higher levels of P450, PdR and PdX/ CFU/mL than the borneol-treated cultures.
The borneol down-regulation of P450 cam , PdX, and PdR might be advantageous to P. putida during periods of low soil aeration. Because the camphor degradation pathway requires four O 2 / camphor (to reach 5-hydroxy-3,4,4-trimethyl-2-heptenedioic acidd-lactone), and the P450 cam -catalyzed oxidation is the first committed step [41], it is advantageous to regulate camphor metabolism at the first step. When aeration increases, low levels of P450 cam convert borneol back to camphor [16], and this frees the Cam operon from borneol down-regulation.

Conclusions
We describe the borneol cycle of P450 cam , a cycle that occurs at low O 2 concentration. The cycle connects to the known catalytic The assay was performed using recombinant P450 cam and m-CPBA as a shunt agent. 2 The assay was performed using recombinant P450 cam , m-CPBA and catalase. 3 The assay was performed using recombinant P450 cam , m-CPBA and glucose/glucose oxidase. 4 The assay was performed using recombinant P450 cam , m-CPBA and superoxide dismutase. 5 The assay was performed using recombinant P450 cam , m-CPBA and butylated hydroxytoluene. 6 The assay was performed using recombinant P450 cam , m-CPBA and EDTA. 5 The assay was performed using recombinant P450 cam , m-CPBA and butylated hydroxytoluene. 6 The assay was performed using recombinant P450 cam , m-CPBA and EDTA. 7 The assay was performed using m-CPBA and ferrous sulphate.  (Fig. 4) is independent of the redox partner proteins (PdX and PdR) and of how Cpd I forms (O 2 reduction or shunt). The reaction occurs both in vitro (this paper) and in vivo [16]. We show here that P450 cam couples the oxidation of water to H 2 O 2 and the reduction of camphor to borneol. We have presented evidence that: i) water is the source of the 2-H in the borneol; ii) water is oxidized to form H 2 O 2 when camphor is reduced; and iii) the transfer of an H atom from water to C-2 of camphor occurs at a rate-limiting step in the borneol cycle. We propose that the reactivity of Cpd I is regulated by O 2 concentration, and we have located a potential access channel where O 2 might bind to P450 cam to exert its allosteric control.
The borneol and H 2 O 2 formed serve several ecological functions. First, borneol and H 2 O 2 are not very toxic to P. putida, whereas the combination is lethal to bacteria such as E. coli which do not contain any P450 [16], and this may give P. putida an advantage in bacterial communities. Secondly camphor induces the expression of P450, PdX, and PdR, whereas borneol decreases the expression of these gene products in P. putida. These features of the P450 cam may protect the bacteria from excessive exposure to borneol and reactive oxygen species during prolonged periods of low oxygen concentration.