Mutagenesis and redox partners analysis of the P450 fatty acid decarboxylase OleTJE

The cytochrome P450 enzyme OleTJE from Jeotgalicoccus sp. ATCC 8456 is capable of converting free long-chain fatty acids into α-alkenes via one-step oxidative decarboxylation in presence of H2O2 as cofactor or using redox partner systems. This enzyme has attracted much attention due to its intriguing but unclear catalytic mechanism and potential application in biofuel production. Here, we investigated the functionality of a select group of residues (Arg245, Cys365, His85, and Ile170) in the active site of OleTJE through extensive mutagenesis analysis. The key roles of these residues for catalytic activity and reaction type selectivity were identified. In addition, a range of heterologous redox partners were found to be able to efficiently support the decarboxylation activity of OleTJE. The best combination turned out to be SeFdx-6 (ferredoxin) from Synechococcus elongatus PCC 7942 and CgFdR-2 (ferredoxin reductase) from Corynebacterium glutamicum ATCC 13032, which gave the highest myristic acid conversion rate of 94.4%. Moreover, Michaelis-Menton kinetic parameters of OleTJE towards myristic acid were determined.

Scientific RepoRts | 7:44258 | DOI: 10.1038/srep44258 forming a stable Fe 4+ -OH species (compound II), which provides a rationale for the final carbon-carbon scission reaction 15,20 . Moreover, the crystal structure of OleT JE in complex with eicosanoic acid (C 20 ) strongly suggested an essential role of the active site residues Arg245 and His85 for catalysis and reaction type selectivity, respectively 9 .
These studies have significantly advanced the understanding on the structural basis and catalytic mechanism of OleT JE . However, there remain a number of unsolved problems: What are the catalytic residues of OleT JE ? What are the key amino acids determining whether decarboxylation or hydroxylation would occur? Is it possible to further improve the decarboxylation activity of OleT JE for practical application? Attempting to address these questions, in this work, we performed systematic mutagenesis analysis of four key residues including Arg245, Cys365, His85, and Ile170 to elucidate their functionality. Furthermore, a select group of redox partner proteins were screened in order to identify an optimal decarboxylation system.

Results and Discussion
Mutagenesis analysis of OleT JE . In the CYP152 family, an arginine residue has been found to be absolutely conserved ( Supplementary Fig. S1). In the crystal structures of OleT JE , P450 BSβ , and P450 SPα 9,10,21 , the fixation of fatty acid substrate in active site all relies on the salt bridges between the guanidyl group of this arginine (Arg241 in P450 SPα , Arg242 in P450 BSβ , and Arg245 in OleT JE ) and the terminal carboxyl group of fatty acids. Moreover, this arginine located near the heme iron reaction center is thought to be responsible for activating the iron(III)-bound H 2 O 2 via acid-base catalysis to synthesise Compound I 10 .
Specifically in OleT JE , the guanidyl group of Arg245 was observed to be only 2.8 Å away from the carboxyl group of arachidic acid (Fig. 2B). To test if this residue is required for OleT JE activity, a series of single point mutants including R245A, R245Q, R245H, R245L, R245E and R245K were constructed. In the presence of 220 μ M H 2 O 2 and 200 μ M myristic acid (C 14 ) as substrate, the activities of all Arg245 mutants (2 μ M) were compared in vitro using the wild type enzyme as positive control. As expected, except for R245K retaining marginal hydroxylation activity perhaps due to the similar chemical property of Lys and Arg, all other mutants were completely inactive (Table 1). These results clearly indicate an essential role of this arginine residue in OleT JE catalysis.
All P450 enzymes including CYP152 peroxygenases unanimously use a cysteine residue to coordinate with the heme iron. By contrast, almost all enzymes (except for chloroperoxidase) in non-P450 peroxygenase superfamily  (A) Structural superimposition of OleT JE (in purple, PDB ID code 4L40) and P450 BSβ (in grey, PDB ID code 1IZO); (B) Comparison of substrate binding pockets between OleT JE and P450 BSβ . Red: heme iron; yellow: eicosanoic acid for OleT JE ; green: palmitic acid for P450 BSβ ; purple: major active site residues in OleT JE ; grey: major different amino acids in P450 BSβ . employ a histidine as the iron-coordinating ligand 3,7 . In consideration of the key role in O-O scission of the His ligand for peroxygenases 22 , we replaced Cys365 of OleT JE with His to investigate the impact on its peroxygenase activity. As a result, no activity was detected for the OleT JE C365H mutant (Table 1), which indicates this Cys ligand is critical for OleT JE and perhaps other P450 peroxygenases.
Among all CYP152 members that have been biochemically characterized so far, OleT JE is the only one that predominantly catalyzes fatty acid decarboxylation with α -and β -hydroxylation as side reactions. Other CYP152 enzymes including P450 SPα 12 , P450 BSβ 23 , and CYP-MP 24 were identified as fatty acid hydroxylases primarily. P450 SPα only hydroxylates fatty acids at α -position 12 . P450 BSβ generates α -hydroxy and β -hydroxy fatty acids as major products, and a small amount of α -alkenes 23 . CYP-MP is able to introduce the hydroxyl group at α -, β -, γ -, δ -, and ε -position of C 12 -C 18 fatty acids, while it only displayed minor decarboxylation activity against myristic acid (C 14 ) and palmitic acid (C 16 ) 24 . Taken together, it is important to identify the key amino acid residues, structural elements or any other factors responsible for the reaction type selectivity of P450 fatty acid decarboxylase/ hydroxylase.
Comparatively, OleT JE and P450 BSβ show 41%/60% amino acid sequence identity/similarity. Their overall structures and active site compositions are highly similar to each other (Fig. 2). The only two obviously different residues in their active sites are His85 (Gln85 in P450 BSβ ) and Ile170 (Val170 in P450 BSβ ), which reside at the two sides of the carboxyl group of substrate with a distance of 5.1 Å and 3.4 Å, respectively, in the crystal structure of OleT JE (Fig. 2). Thus, we hypothesize that these two amino acids might be related to the decarboxylation/hydroxylation selectivity.
To test this hypothesis, saturation mutagenesis of His85 and Ile170 in OleT JE was performed, and the in vitro activities of H85X and I170X variants towards myristic acid were evaluated (Fig. 3). As results, 11 out of 19 H85X variants were completely dead mutants. The rest 8 mutants unanimously lost their decarboxylation activities, while retaining varying hydroxylation activities (Fig. 3A). Notably, the two substitutions with an amide side chain (H85Q and H85N) retained most hydroxylation activity for unknown reasons.
Previously, Rude et al. proposed the importance of His85 for the decarboxylation activity of OleT JE based on the result that the Q85H mutation of P450 BSβ enhanced its decarboxylation activity towards palmitic acid for about 50% 13 , which led to an increase in the ratio of decarcarboxylation to hydroxylation from 0. 19   P450 BSβ to 0.30 in the P450 BSβ Q85H mutant. In this work, we confirmed the necessity of His85 for the decarboxylation activity of OleT JE . This is well consistent with the mechanism proposed by Belcher et al., in which His85 could act as a proton donor to compound I. This proton donation resulting in the protonated compound II was thought to be required for generation of the carboxylate radical for homolytic scission of the substrate C-C α bond, thereby forming the terminal alkene. In the absence of proton donor, hydroxylation would be the only reaction 9,20 . Similarly, none of the I170X mutants showed any decarboxylation activity against myristic acid. However, 11 out of 19 variants were able to catalyze the α -and/or β -hydroxylation to different extents (Fig. 3B). Thus, the mutagenesis analysis of His85 and Ile170 clearly indicates that these two residues adjacent to the carboxyl end of substrate are key factors for fatty acid decarboxylation.
Together with Arg245, these three amino acids are likely involved in the exact substrate positioning required for decarboxylation, explaining why conserved substitutions (e.g. I170V or I170L) also abolished the decarboxylation activity. In this regard, it might be highly challenging to rationally design a better version of OleT JE that is more selective for decarboxylation than hydroxylation without compromising the total activity, at least based on the current knowledge on the structure-function relationship of P450 fatty acid decarboxylases. A recent study of site-directed mutagenesis of OleT JE at selected sites lining the substrate binding pocket also proved difficulty in improving OleT JE activities towards structurally different aromatic carboxylic acid substrates. Only meager improvements (less than 1-fold) were observed in the few positively responded mutants (F79L and F294A) 17 . To overcome these rational design challenges, random gene mutagenesis or DNA shuffling coupled to high-throughput screening could be a more feasible strategy.
In vitro decarboxylation activity of OleT JE supported by different redox partners. When OleT JE was first identified to be a P450 fatty acid decarboxylase with potential application in the field of biofuels, it was thought to be an obligate peroxygenase as P450 SPα and P450 BSβ . However, our laboratory recently revealed the H 2 O 2 -independent activity of OleT JE (i.e. the activity depending on O 2 /redox partner(s)/NAD(P)H). This discovery has initiated the development of different olefin producing systems based on OleT JE and alternative redox partner protein(s). For instance, we have shown that the flavodoxin/flavodoxin reductase from E. coli and the RhFRED reductase from Rhodococcus sp. NCIMB 9784 are capable of supporting the OleT JE activity both in vitro and in vivo 8 . Dennig et al. employed putidaredoxin and putidaredoxin reductase from Pseudomonas putida to achieve the decarboxylation of short-chain fatty acids (C 4 -C 9 ) into corresponding α -alkenes in vitro 14 (Fig. 1). More importantly, by taking advantage of heterologous P450 redox partners, the engineered E. coli 8 and Saccharomyces cerevisiae 25 cells with OleT JE expression were able to produce 97.6 mg/L and 3.7 mg/L total α -alkenes, respectively.
To identify the H 2 O 2 -independent activity of OleT JE with different redox partners, we in vitro screened a series of ferredoxins (Fdx) and ferredoxin reductases (FdR) derived from the cyanobacterial strain Synechococcus elongatus PCC 7942 and the Gram-positive bacterium Corynebacterium glutamicum ATCC 13032 (Supplementary DNA sequences of redox partners). Specifically, three FdRs (SeFdR-1 from S. elongates, and CgFdR-1 and CgFdR-2 from C. glutamicum) were individually coupled with ten Fdxs (SeFdx-1-7 from S. elongatus and CgFdx-1-3 from C. glutamicum), and 30 different combinations of redox partner proteins were mixed with OleT JE , myristic acid, and NADPH, respectively. The supportive activities of all redox partner combinations were compared to that of RhFRED and H 2 O 2 (Fig. 4). Interestingly, all tested hybrid redox systems were able to support the in vitro decarboxylation activity of OleT JE to some extent, indicating the low selectivity of redox partners by this P450 fatty acid decarboxylase. The best combination turned out to be CgFdR-2 and SeFdx-6, which gave the highest conversion rate of 94.4% (Fig. 4). Using these two optimal redox partner proteins to mediate the electron transfer from NADPH, the steady-state kinetic parameters of OleT JE towards myristic acid were determined to be K m = 5.0 ± 2.4 μ M, k cat = 2.2 ± 0.2 min −1 , and k cat /K m = 0.4 μ M −1 min −1 (Supplementary Fig. S2A). Comparatively, the values of K m and k cat were 24.2 ± 8.7 μ M and 71.0 ± 8.4 min −1 (Supplementary Fig. S2B), respectively, when H 2 O 2 was employed as the sole oxygen and electron donor. The k cat /K m value of 2.9 μ M −1 min −1 greater than 0.4 μ M −1 min −1 seemed inconsistent with the qualitative results that the CgFdR-2/SeFdx-6/NADPH redox system showed higher fatty acid to α -alkene conversion rate (94.4%) than that supported by H 2 O 2 (49.5%). We reason this contradiction might be due to inactivation of OleT JE by H 2 O 2 during prolonged incubation. Taken together, these results demonstrated efficient monooxygenase-like property of OleT JE to use the O 2 /redox partner(s)/ NAD(P)H system, which is critical for the future investigation of the unique mechanism and better application of this enzyme.

Conclusion
We have systematically investigated the functions of three active site residues of OleT JE including Arg245, His85, and Ile170 by site-directed mutagenesis. It was found that they are all required for the decarboxylation activity of OleT JE , presumably by forming a salt-bridge with the substrate carboxyl group (Arg245), by acting as a proton donor (His85), and by precisely coordinating substrate positioning in the active site (Ile170). We also studied the H 2 O 2 -independent activity of OleT JE and revealed a series of heterologous redox partners capable of supporting its decarboxylation activity efficiently in vitro. These results not only further our understanding on the unique decarboxylative mechanism of OleT JE , but also serve as a guide for further bioengineering of this P450 system and the future industrial application.
Strains, plasmids and media. Escherichia coli DH5α cells were used for plasmid transformation and mutant screening. Escherichia coli BL21(DE3) was used for protein overexpression. The plasmid pET28b was used for gene cloning. E. coli cells were grown in Terrific Broth medium composed of 1.2% tryptone, 0.5% glycerol, 2.4% yeast extract, 0.23% KH 2 PO 4 and 1.25% K 2 HPO 4 , supplemented with the selective antibiotic (50 μ g/mL kanamycin), thiamine (1 mM) and rare salt solution for protein expression 26 .

Molecular cloning.
With pET28b-oleT JE as template, the oleT JE mutants were constructed using the Quikchange mutagenesis method and cloned into pET28b vector. Mutagenesis primers are listed in Supplementary Table S1.
Overexpression and purification of proteins. The recombinant expression plasmids were transformed into E. coli BL21(DE3). After cultivation in LB medium containing 50 μ g/mL kanamycin at 37 °C, 220 rpm overnight, 1% volume of the seed culture was used to inoculate 1 L Teffific Broth medium containing 50 μ g/mL kanamycin, 1 mM thiamine and rare salt solution. When the OD 600 reached 0.6-0.8, IPTG was added to a final concentration of 0.2 mM for induction, followed by shaking incubation at 18 °C for 20 h.
After harvesting by centrifugation at 4 °C, 6000 rpm for 10 min, cells were stored at − 80 °C for 30 min. Then cells were thawed and re-suspended in 40 mL lysis buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10% glycerol and 10 mM imidazole, pH 8.0) and sonicated on a JY92-IIDN ultra-sonicator for 30 min with a 5 s on and 5 s off pulse. The whole cell lysates were centrifuged at 12,000 rpm for 30 min at 4 °C. The clarified cell lysates were collected and incubated with 1 mL Ni-NTA resin under gentle shaking at 4 °C for 1-2 h. The mixture was then loaded onto an empty column and washed with approximately 100 mL wash buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10% glycerol and 20 mM imidazole, pH 8.0) until no proteins were detected in the flow-through. The bound protein was eluted by 10 mL elution buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10% glycerol and 250 mM imidazole, pH 8.0) and then concentrated using a Millipore Ultra-filter (30 K). The concentrated protein solution was loaded onto a PD-10 column, which was used for removal of imidazole and buffer exchange into desalting buffer (50 mM NaH 2 PO 4 and 10% glycerol, pH 7.4). The eluent was flash-frozen by liquid nitrogen and stored at − 80 °C for later use.
Determination of enzyme concentration. The ferric-CO complex of P450 enzyme was prepared by slow bubbling of carbon monoxide gas into a solution of purified ferric P450 for 1 min. Then its UV-visible absorption spectrum from 300 nm to 600 nm was recorded on a UV-visible spectrophotometer DU800 (Beckman Coulter, Fullerton, CA, USA). Following reduction by sodium dithionite, the corresponding spectrum of reduced ferrous-CO adducts was recorded. The functional concentration of P450 was calculated from the CO-bound reduced differential spectrum using a molar extinction coefficient (ε 450-490 ) of 91 mM −1 . cm −1 27 .
The concentration of purified ferredoxin and ferredoxin reductase was determined by monitoring the absorbance at 325 nm and 454 nm, respectively, and using their corresponding molar extinction coefficient 15.4 mM −1 . cm −1 (ε 325 ) and 11.3 mM −1 . cm −1 (ε 454 ) 28 . Analytical method. Qualitative and quantitative analysis of the products were performed by GC-MS 29 . To detect α -alkenes, an Agilent 7890A gas chromatography equipped with HP-INNOWAX (Agilent Technologies, Inc., cross-linked polyethylene glycerol, Santa Clara, CA, USA; 0.25 μm film thickness, 30 m by 0.25 mm) column was adopted. The column heating program was as follows: the initial temperature of oven was set to 40 °C for 4 min, then increased at a rate of 10 °C/min to 250 °C, and hold for 15 min. The α -and β -hydroxy fatty acids detection was carried out using the Agilent J&W DB-5 MS column (0.25 μ m film thickness, 30 m by 0.25 mm). Furthermore, an Agilent 5975C MSD quadrupole mass spectrometer with a scan range from 50 to 500 m/z under electron ionization condition (70 eV) was coupled to the GC. The oven temperature was 50 °C initially and ramped up to 300 °C at the above mentioned rate, then 300 °C for 5 min. Quantification was performed using the corresponding authentic standard compounds and heptadecanoic acid as the internal standard.