Catalytic strategy for carbon−carbon bond scission bythe cytochrome P450 OleTJob L. Granta, Megan E. Mitchella, and Thomas Michael Makrisa,1

aDepartment of Chemistry and Biochemistry, University of South Carolina, Columbia, SC 29208

Edited by Michael Green, University of California, Irvine, CA, and accepted by Editorial Board Member Harry B. Gray July 15, 2016 (received for review April19, 2016)

OleT is a cytochrome P450 that catalyzes the hydrogen peroxide-dependent metabolism of Cn chain-length fatty acids to synthesizeCn-1 1-alkenes. The decarboxylation reaction provides a route for theproduction of drop-in hydrocarbon fuels from a renewable andabundant natural resource. This transformation is highly unusualfor a P450, which typically uses an Fe4+−oxo intermediate knownas compound I for the insertion of oxygen into organic substrates.OleT, previously shown to form compound I, catalyzes a differentreaction. A large substrate kinetic isotope effect (≥8) for OleT com-pound I decay confirms that, like monooxygenation, alkene forma-tion is initiated by substrate C−H bond abstraction. Rather thanfinalizing the reaction through rapid oxygen rebound, alkene syn-thesis proceeds through the formation of a reaction cycle interme-diate with kinetics, optical properties, and reactivity indicative of anFe4+−OH species, compound II. The direct observation of this inter-mediate, normally fleeting in hydroxylases, provides a rationale forthe carbon−carbon scission reaction catalyzed by OleT.

oxygen activation | metal−oxo | cytochrome P450 | hydrocarbon |compound II

Cytochrome P450 (CYP) enzymes catalyze an extraordinarybreadth of physiologically important oxidations for xenobiotic

detoxification and specialized biosynthetic pathways (1, 2). Themetabolic diversity of CYP enzymes originates from a sophisti-cated interplay of substrate molecular recognition with precisetuning of metal oxygen species formed at the enzyme active site.CYPs use a thiolate-ligated heme iron cofactor to activate mo-lecular oxygen and produce short-lived ferric superoxo (3–5),ferric (hydro)peroxo (6–8), and ferryl (9, 10) intermediates. Co-ordinated efforts over several decades have resulted in isolation ofeach intermediate, including recent characterization of the prin-cipal oxidant thought to be responsible for the vast majority ofP450 oxidations, the Fe4+−oxo pi−cation radical species com-monly referred to as compound I (or P450-I) (10).The archetypal P450 hydroxylation involves C−H bond ab-

straction by P450-I and ensuing rapid oxygen insertion through arecombination process termed “oxygen rebound” (11, 12). Thehydrogen abstraction/rebound mechanism describes the strategyused for most P450 transformations, and is thought to describecatalysis by numerous metal-dependent oxygenases and syntheticbio-inspired metal−oxo complexes (e.g., refs. 13–17). Despite theubiquity of this mechanism, in some metalloenzymes, a metal−oxospecies is formed that is not ultimately destined for incorporationinto a substrate. Some characterized examples include the non-heme dinuclear iron ribonucleotide reductase (RNR R2) (18, 19)and mononuclear iron halogenase SyrB2 (20–22). The mecha-nisms for RNR and SyrB2 highlight the importance of quaternarystructural elements, an extensive 35-Å proton-coupled electrontransfer pathway in RNR (23), and extremely subtle (subangstrom)substrate positional tuning (SyrB2) (21, 22), to enable efficientcircumvention from the monooxygenation reaction coordinate.P450-I has also been linked to a number of important trans-

formations in which an oxygen rebound step is not readily ob-served, including the desaturation of pharmaceuticals by liverP450s to afford hepatotoxic metabolites (24) and the C−C lyase

activity of human P450 aromatase that is critical for estrogenbiosynthesis (25), among others (26, 27). However, unambiguousdetermination of the oxidant responsible and operant mechanismfor these aberrant P450 transformations has met significant chal-lenges. Compound I has not been isolatable from an O2-dependentreaction due to rate-limiting electron transfer processes, and theproducts resulting from these atypical reactions can sometimesrepresent only minor channels of the enzyme/substrate(s) involved.Attempts to resolve the mechanisms for these reactions havetherefore largely relied on indirect methods: isotopic labelingstrategies [oxygen incorporation (25, 28), steady-state substrate(24, 27), and solvent (29) kinetic isotope effects (KIE)], inferencefrom the spectroscopic characterization of preceding reaction cycleintermediates (30), or computational methods (31, 32). AlthoughP450-I species have been widely hypothesized to carry out manyreactions in addition to monooxygenation, direct visualization hasremained elusive due to the fleeting nature of this intermediate.OleT is a recently discovered bacterial P450 that catalyzes an

unusual carbon−carbon scission reaction, converting a Cn chain-length fatty acid to a Cn-1 1-alkene (33, 34) and carbon dioxidecoproduct (35) (Fig. 1). In addition to the peculiar chemical natureof this reaction, OleT catalysis has garnered significant bio-technological interest, as it involves the conversion of a bioavailableand abundant feedstock into a petrochemical and valuable syn-thetic precursor (36–38). OleT differs from most P450s in twofundamental aspects. Belonging to the CYP152 family (39, 40),catalysis is efficiently initiated by hydrogen peroxide, rather thanO2 and reducing equivalents delivered through a redox chain.Moreover, single (35) and multiple turnover (36, 38) studies ofwhat is generally believed to approximate the chain length of thephysiological substrate, eicosanoic acid (EA), have indicated thatoxidative decarboxylation, affording nonadecene, is the largelydominant reaction route (34). Combined 13C substrate and H2

18O2isotopic labeling by our laboratory has shown that cleavage of theterminal fatty acid carboxylate does not lead to recovery of 18O in

Significance

The biocatalytic production of hydrocarbons from abundantnatural resources provides a means to expand the current fuelinventory. The recently identified cytochrome P450, OleT, syn-thesizes 1-alkenes from fatty acids through a carbon−carbonscission reaction that is highly atypical for the enzyme super-family. Rapid kinetics studies reveal two critical reaction cycleintermediates that define the catalytic strategy used by OleT andhighlight the functional versatility of P450 enzymes.

Author contributions: J.L.G., M.E.M., and T.M.M. designed research; J.L.G., M.E.M., andT.M.M. performed research; J.L.G., M.E.M., and T.M.M. analyzed data; and T.M.M.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. M.G. is a Guest Editor invited by the EditorialBoard.1To whom correspondence should be addressed. Email: makrist@mailbox.sc.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1606294113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1606294113 PNAS | September 6, 2016 | vol. 113 | no. 36 | 10049–10054

BIOCH

EMISTR

YCH

EMISTR

Y

the 13CO2 product, ruling out a multistep oxygenolytic mechanisminvolving successive hydroxylations.Benefitting from the high chemoselectivity of OleT and a facile

means to rapidly trigger catalysis, we recently reported that ahighly accumulating compound I intermediate (Ole-I) could beisolated in reactions of OleT using the native H2O2 terminal oxi-dant and a bound deuterated substrate analog (35). The observa-tion of a compound I species in OleT, optically indistinguishablefrom those that promote hydroxylations (10, 41), intimated thatthe same P450 intermediate can be retuned to catalyze a funda-mentally different reaction. In this work, we directly demonstratethat P450 compound I species can promote reactions that do notinvolve oxygen rebound. A mechanistic basis for the functionaldivergence of OleT is provided by identification of the subsequentintermediate in the decarboxylation reaction sequence. Single-turnover kinetic studies show that alkene formation is initiated bysubstrate C−H bond abstraction by Ole-I, forming an Fe4+−OHspecies, compound II. The remarkable stability of this intermediateprovides direct insight into how metal−oxo reprogramming isachieved by P450 cytochromes.

ResultsAlkene Formation Proceeds by Fatty Acid C−H Abstraction by CompoundI. Stopped-flow absorption studies have shown that first detectableintermediate in the single-turnover reaction of perdeuterated EA(D39-EA) bound high-spin ferric OleT and H2O2 is Ole-I (35),which has absorption maxima at 370 and 690 nm, as reported forother thiolate-ligated compound I species (9, 10, 41). Ole-I is achemically competent species as its decay rate constant, determineddirectly below, exceeds the turnover number of OleT [∼0.2 s−1 at25 °C (34)]. Single-turnover reactions of EA−OleT and H2O2produce nearly one equivalent of alkene per enzyme (35). Fittingthe time course for Ole-I decay at 370 nm using regression methods(Fig. 2, red trace) required a two-summed exponential expressionwith reciprocal relaxation times (RRTs) 1/τ1 = 77 ± 2 s−1 and 1/τ2 =8.1 ± 0.4 s−1 (Table S1 and Fig. S1A). The complex biphasic kineticbehavior for Ole-I decay may be attributed to multiple populationsof the intermediate that react at different rates, or may alternativelyhint toward the presence of an additional intermediate that absorbsat this wavelength. To delineate between these possibilities, thedecay of Ole-I was alternatively monitored at 690 nm (Fig. S2). Thedata, which could be accurately fit with a single exponential ex-pression, indicate a homogeneously reactive Ole-I species. Con-sequently, the measured RRT measured at this wavelength (1/τ =83 ± 2 s−1) correlates well with the fast phase from the 370-nm data.We (35) and others (34) have postulated that alkene formation

may proceed via initial substrate H atom abstraction based on theaccepted mechanism for aliphatic hydroxylations by P450-I, lossof a hydrogen from the Cβ position during conversion of a Cnfatty acid to a Cn-1 alkene, and structural similarity of OleT (33)with fatty acid hydroxylases (40, 42) that exhibit significantly largesubstrate 2H steady-state KIEs (39). Appreciable accumulation ofOle-I was not previously detected in the OleT single-turnover re-action with protiated EA (H39-EA), suggesting that the rate ofreaction was too fast to be captured in our earlier photodiode array(PDA) studies (35). To further probe the reactivity of Ole-I, theH39-EA single-turnover system was monitored at 370 nm with a

photomultiplier tube (PMT), providing greater sensitivity. Biphasiccharacteristics were again observed, although the initial decayphase was significantly more rapid than for D39-EA, reachingcompletion within 10 ms (Fig. 2, blue trace, Fig. S1B). Despite thisrapid disappearance, relatively accurate RRTs (within 10% error)were obtained from a two-summed exponential fit of the data,with 1/τ1 = 630 ± 60 s−1 and 1/τ2 = 6.9 ± 0.4 s−1. Only the fastRRT demonstrated appreciable isotopic sensitivity. This indicatesthat, if the slower phase indeed derives from a second in-termediate species that contributes to the absorbance at 370 nm, itdoes not directly abstract an H atom. An observed 2H KIE forOle-I decay at this H2O2 concentration, which represents a lowerlimit for the unmasked value, is provided by a ratio of the fastRRTs, [Hkapp/

Dkapp = RRT1 (H39-EA)/RRT1 (D39-EA)], giving aKIE ≥ 8.1 ± 1.1. The large KIE, which has not been previouslymeasured for a P450-I generated with a bound substrate, is inaccord with unmasked KIEs for steady-state P450 hydroxylations(12), rapid mixing studies of P450-I (10), and similar thiolate-ligated heme iron−oxo species (41), and theoretical considerations(43). This confirms that the distinctive C−Cα cleavage catalyzedby OleT most likely commences in a very similar way to P450monooxygenations.

A Second Intermediate in the Decarboxylation Reaction. Kinetic evi-dence for an additional chromophoric species in OleT de-carboxylation prompted a close examination of the PDA data forthe EA single-turnover reaction. In spectra obtained from mixingD39-EA–bound OleT with a large excess of H2O2, the loss of Ole-I(Fig. 3A, green spectrum) over 40 ms is accompanied by the for-mation of a new transient species (Fig. 3 A and B, orange spec-trum). We refer to this second OleT intermediate as Int-2. Int-2has a Soret maximum that appeared to be red-shifted and signif-icantly less intense than the ferric water ligated low-spin (λmax =417 nm) product state of the enzyme, which formed more slowlyover the course of 500 ms (Fig. 3B, blue spectrum). The opticalfeatures of this presumptive new intermediate are partiallyrevealed by subtracting the final Fe3+−H2O species from the 40-msspectrum (Fig. 3B, Inset), which shows a prominent positive featureat 440 nm and a secondary peak at 370 nm. A similar species wasdetected in parallel studies of OleT−H39-EA (Fig. 3C), althoughits kinetics are clearly altered, instead maximizing within 10 ms. Asa result, the difference spectrum (10 ms to 500 ms) shares the samepositive optical features at 370 and 440 nm, but is approximatelytwofold more intense (Fig. 3C, Inset). This kinetic behavior impliesthat one or both rate constants for Int-2 formation and decay arealtered upon substrate isotopic substitution.

COH

OH

H

H

HH3C

H

H H

H

H3C

H

HH

α + CO2β

H O2 2

OleT

16 16

αβ

Fig. 1. The decarboxylation reaction catalyzed by OleT. Using a hydrogenperoxide cosubstrate, OleT metabolizes a Cn chain-length fatty acid to pro-duce a Cn-1 alkene and carbon dioxide coproduct.

0.0 0.1 0.2 0.3 0.4 0.50.50

0.60

0.70

0.80

0.56

0.58

0.60

Abs

(370

nm

)

0.01 0.03 0.05

time (s)

H39-EA

D39-EA

H39-EA

Fig. 2. Stopped-flow absorption evidence for a substrate 2H KIE in the OleTreaction. Representative single-wavelength time course for the reaction of20 μM OleT−D39-EA (red) or OleT−H39-EA (blue, and Inset) with 10 mM H2O2

monitored at 370 nm. The superimposed white traces represent two-summedexponential fits to the data.

10050 | www.pnas.org/cgi/doi/10.1073/pnas.1606294113 Grant et al.

Int-2 Is the Decay Product of Compound I. The kinetics of Int-2 wasmonitored at 440 nm using single-wavelength stopped flow. Thetime courses for both substrates reveal rapid development of thechromophore associated with Int-2 and a subsequent slower decayphase upon conversion to the final ferric low-spin product form(Fe3+−OH2) of the enzyme (Fig. 4A and Fig. S1 C andD). Each ofthe single-wavelength time courses could be appropriately fit toa sum of two exponential functions with well-resolved RRTs(Table S1). Intriguingly, the fast-phase RRT exhibited a large

dependence on the isotopic composition of the substrate and, forboth types of substrate, matched the decay rate constant for Ole-I.Both kinetic features suggest that Int-2 results from the de-composition of Ole-I, and therefore represents the next in-termediate in the reaction sequence.To delineate the kinetics of Int-2 more fully, the dependence of

both RRTs at various H2O2 concentrations was examined. Forboth H39-EA and D39-EA substrates, the slow RRT associated withInt-2 decay was independent of H2O2 concentration, and demon-strated only minor isotopic sensitivity (Fig. S3). This is consistentwith an assignment of Int-2 as an intervening intermediate betweenOle-I and Fe3+−OH2. For the D39-EA plot (Fig. 4B and Inset), thefastest RRT shows a hyperbolic dependence on H2O2 concentra-tion, indicating that there are at least two reaction steps that lead tothe generation of Int-2. Provided that this phase corresponds to anirreversible step in the reaction sequence, then the measured RRTcorresponds to the rate constant for Int-2 formation. Results froma hyperbolic fit of the H2O2 concentration dependence show thatthis is the case. The y intercept of the plot, zero, corresponds to thereverse rate constant for the Int-2 formation step (k−2 = 0). Theasymptote of the plot (equal to k2 + k−2) represents the formationrate constant for Int-2 (k2 = 86 ± 4 s−1) and gives an apparentKH2O2D ≈ 100 μM. These results support a scheme in which Int-2 is

produced as a result of an irreversible process, which would be

0.4

0.8

1.2

Abs

orba

nce

300 400 500 600 7000.0

0.4

0.8

1.2

wavelength (nm)

Abs

orba

nce

300 400 500 6000.0

wavelength (nm)440 nm

417 nm

300 400 500-0.3

0.0

0.3

300 400 500

-0.15

0.00

0.15

40 ms

500 ms 440 nm

417 nm

A

B

C

10 ms

500 ms

300 400 500 6000.0

0.4

0.8

1.2

370 nm

370 nm

Abs

orba

nce

wavelength (nm)

Difference

Difference

2 ms

40 ms

Fig. 3. Identification of an additional intermediate (Int-2) that forms con-comitant with OleT compound I (Ole-I) decay. PDA spectra of a single-turnover reaction of 20 μM OleT−D39-EA with 10 mM H2O2 in (A) 2- to 40-msand (B) 40- to 500-ms timeframes. (C) The OleT−H39-EA + H2O2 reaction,from 2 ms to 500 ms. Insets represent kinetic difference spectra, 500 ms to40 ms (in B) and 500 ms to 10 ms (in C).

0.0 0.1 0.2 0.3 0.4 0.50.3

0.4

0.5

0.6

0.7A

bs (4

40 n

m)

time (s)

H39-EA

D39-EA

A

0 500 1000 1500 2000 25000

100

200

300

400

500

600

-1k

(s

) ob

s

H O (μM) 2 2

H39-EA

D39-EA

-1k

(s

) ob

s

H O (μM) 2 2

B

440 nm370 nm

0 1000 20000

20

40

60

80

D39-EA

Fig. 4. Transient kinetics of Int-2 formation and decay. (A) Representativesingle-wavelength time course for the reaction of 20 μM OleT−D39-EA (red) orOleT−H39-EA (blue) with 10mMH2O2 monitored at 440 nm. The superimposedwhite traces represent two-summed exponential fits to the data. (B) Plots ofthe dependence of the larger RRT on H2O2 concentration in respective colors.The solid lines represent nonlinear fitting to a hyperbolic expression (for theD39-EA reaction) and linear fitting (H39-EA). The Inset shows the decay rate ofOle-I at 370 nm and formation rate of Int-2 for the OleT–D39-EA reaction.

Grant et al. PNAS | September 6, 2016 | vol. 113 | no. 36 | 10051

BIOCH

EMISTR

YCH

EMISTR

Y

expected for substrate C−D bond cleavage, and that its rate offormation is identical to the decay rate constant of Ole-I.The Ole-I→ Int-2→Fe3+−OH2 catalytic sequence is further

verified by examination of the fast RRT versus H2O2 for the H39-EA reaction (Fig. 4B). In this case, the plot reveals a linear de-pendence of the RRT and H2O2, with no evidence of saturationwithin the accessible range of H2O2 concentrations in which ratescould be reliably measured. This altered kinetic behavior can beinterpreted as arising from more facile cleavage of the target sub-strate C−H bond (relative to C−D), such that Int-2 formation isnow partially dominated by the rate constant for H2O2 binding.Accordingly, the forward and reverse rate constants for the peroxidebinding step are provided by the slope (k1 = 5.4 × 105 M−1·s−1), andintercept of the plot (k−1 = 27 ± 3 s−1), respectively. A kinetic modelthat accounts for all of the observed kinetic data is shown in Fig. 5.

Assignment of Int-2 as an Iron(IV)−Hydroxide Species. If Int-2 derivesdirectly from hydrogen atom abstraction by compound I, with nointervening reaction steps, it should resemble an Fe4+−OH speciessimilar to compound II. Rapid radical recombination by oxygenrebound prohibits the observation of compound II in P450 hy-droxylations. However, the progressive decrease in decay rateconstants for Ole-I and Int-2 allows for the latter to appreciablyaccumulate in OleT, most significantly for the H39-EA reaction(Fig. S4). Using global analysis methods, we extracted the pureoptical spectra for Ole-I and Int-2. A comparison of the opticalspectra of these species with ferric low-spin Fe3+−OH2 OleT isshown in Fig. 6A. The optical spectrum of Int-2 has characteristicsthat are highly similar to those reported for protonated thiolateFe4+ hydroxide complexes prepared in P450 (44) and the thiolate-ligated heme peroxygenase APO (45). Notably, Int-2 is character-ized by a Soret maximum at λmax = 426 nm that is clearly red-shifted and has a lower molar extinction coefficient relative to theFe3+−OH2 form of the enzyme. Int-2 also exhibits a split Soretband, with an additional absorption maximum at 370 nm. Thisfeature contributes to the slower, isotopically insensitive kineticphase described earlier. To rule out the possibility that Int-2 couldbe more appropriately assigned as a deprotonated ferric hydroxidespecies (Fe3+−OH), which, in principle, could arise from hydrogenabstraction and ensuing rapid electron transfer from the substrateor a neighboring redox active amino acid, this form of the enzymewas generated for direct comparison. The addition of strong baseto the substrate free Fe3+−OH2 enzyme results in generation of atransiently stable Fe3+−OH (Fig. S5). Notably, the Fe3+−OHspecies did not exhibit any of the principle optical spectroscopicfeatures (Soret maximum, extinction coefficient, hyperporphyrin)that clearly define the spectrum of Int-2.A model-independent approach was also used to verify the

optical features of Int-2 obtained from global analysis. FollowingOle-I depletion, the reaction consists of only two principle ab-sorbing species, Int-2 and Fe3+−H2O. As a result, the spectrum ofInt-2 can be simply derived at any timeframe from a linear

combination of two spectra, one known. No additional input ofkinetic parameters is necessary. This subtraction method consis-tently reproduced the same Int-2 spectrum determined from globalanalysis (Fig. S6). Moreover, the relative composition of eachspecies is in excellent agreement with those computed using rateconstants from the kinetic model.Additional support for the assignment of Int-2 as having an oxi-

dation state higher than Fe3+ is provided through an examination ofits reactivity in double-mixing studies. Thiolate-ligated compound IIspecies, including that of chloroperoxidase (CPO-II) (46) and aro-matic peroxygenase (APO-II) (45), are able to oxidize phenols, al-beit with variable proficiency. Int-2 was first prepared by mixing theEA-H39–bound enzyme with H2O2 to favor the rapid depletion ofOle-I and formation of Int-2. After a 20-ms aging time, variousconcentrations of pH buffered phenols were added in a second push.The decay of Int-2 was measured at 440 nm and fit to a single-exponential decay process (kobs) (Fig. S7). The addition of phe-nols significantly accelerated the decay of Int-2 in a concentration-dependent manner. This, along with the optical characterization ofInt-2, rules out assignment as an Fe3+−OH species, which would beinert to phenols. Representative plots for the reactivity of phenol and3-Cl phenol with Int-2 are shown in Fig. 6B with apparent second-order rate constants, calculated from the slopes, in Fig. 6B. Addi-tional data for a series of substituted phenols is provided in Table S2.

Fe3+

S

Fe4+

S

H O2 2O

+

5 -1 -17 x 10 M s

-130 s

EA-

-1k 700 sH

Int-2nonadecene

+CO

Fe3+

S

HOH

2

-110 sEA(D)H EA- (D)H

-1k = 80 sD

Fig. 5. Summary of the kinetic parameters for OleT compound I formation,and subsequent decay to Int-2 and the low-spin ferric enzyme. The decayrate constants for Ole-I with D39-EA and H39-EA are indicated as kD and kH,respectively. The formation of nonadecene and carbon dioxide was dem-onstrated in a previous study (35).

300 400 500 600 7000.0

0.4

0.8

1.2

370 nm

417 nm

426 nm

690 nmAbs

orba

nce

wavelength (nm)

0 10 20 30 40 500

10

20

30

40

50

60

phenol, mM

Int-2

dec

ay -1

rate

con

stan

t (s

)A

B

-1k = 1.0 mM s-1

-1k = 4.7 mM s-1

Fig. 6. Optical spectroscopic features and reactivity of Int-2. (A) The purecomponent spectra of OleT Int-2 (orange) and compound I (green) as obtainedfrom global fitting analysis of PDA data, and compared with the ferric H2Olow-spin enzyme (blue). (B) Plot of the apparent Int-2 decay rate constant withphenol (circle) and 3-chlorophenol (square). OleT−H39-EA (40 μM) was mixedwith 2 mM H2O2, aged 20 ms, and then mixed 1:1 with substituted phenols.The concentrations shown are after mixing. Error bars represent 1 SD.

10052 | www.pnas.org/cgi/doi/10.1073/pnas.1606294113 Grant et al.

DiscussionTransient kinetics directly demonstrate that a P450 compound Ioxidant can carry out reactions that do not get finalized by an ox-ygen rebound step. Although alternative reactions for CYP-I spe-cies have been proposed for decades, prompting considerablediscussion, the highly fleeting nature of compound I prevented priorobservation of such a reaction taking place. The ability to rapidlytrigger catalysis with the native oxidant used by the enzyme in thepresence of a tightly bound substrate, and fortuitous rate constantsfor intermediate interconversions, provides direct insight into theorigins of the impressive metabolic versatility of CYP oxidations. Asa result of the visualization of two ferryl intermediates during OleTcatalysis, C−C bond cleavage can be added to the diverse repertoireof transformations catalyzed by metal−oxo species.The mechanistic strategy adopted by OleT contrasts with those

hypothesized for other metalloenzymes involved in hydrocarbonbiosynthesis. In alkane formation by dinuclear iron aldehydedeformylating oxygenases, cleavage of the terminal aldehyde carbon,which is eventually lost as formate (47), is thought to be mediated byformation of a nucleophilic iron peroxide intermediate (48). Thisspecies is different from the highly oxidizing dinuclear Fe4+ clustermore commonly used by structurally related monooxygenases (13).Recently, a family of unrelated nonheme Fe2+-dependent enzymes(UndA) has also been shown to catalyze the conversion of medium-chain-length Cn fatty acids to Cn-1 alkenes (49). Although the activeoxidant involved in this O2-dependent reaction has yet to be clari-fied, an Fe3+ superoxide intermediate has been proposed on thebasis of the enzyme having no strict requirement for a reductant tofacilitate O−O heterolysis. The 2H substrate KIE for Ole-I decaydemonstrates that alkene synthesis is not accommodated by the useof an alternative active-site oxidant that precedes compound I, norfrom the rearrangement of an oxidized product. Instead, the OleTreaction coordinate diverges after C−H abstraction, and resultsfrom an apparent suppression of radical recombination.The remarkable stability of Ole-II provides a framework to

compare OleT catalysis with hallmark CYP monooxygenations.Although thiolate-ligated compound II species have recently beenprepared in both CYP (44) and APO (45) enzymes, these speciesare produced from the deleterious oxidation of nearby aromaticamino acids from the protein framework (CYP), or by the addi-tion of suitable poised reductants to rapidly quench compound I(APO), rendering them quasi-stable. The catalytic efficiency ofthe H2O2 single-turnover system (35) and kinetic behavior of theFe4+−OH species observed in OleT demonstrates that Ole-II is acompetent reaction intermediate, and that it is directly producedfrom substrate C−H abstraction. To our knowledge, an analogous“rebound” intermediate has not been observed in any metal-containing oxygenase to date. Although the optical data presentedhere do not permit a finite measure of the lifetime of the substrateradical, nor the precise localization of this species, a comparison ofthe rate of Ole-II decomposition to rates of radical recombinationestimated for CYPs using radical clock substrates is informative.Although the rates computed from these approaches can vary withthe CYP and substrate tested, typical radical lifetimes generallyreside at the nanosecond to picosecond time scale (26). Ole-II, bycomparison, persists for hundreds of milliseconds and does notelicit oxygen rebound.

A number of factors may contribute to the recalcitrance ofOle-II •OH rebound. Intriguingly, the reactivity of Ole-II towardphenols more closely resembles that of chloroperoxidase CPO-IIthan the high proficiency observed for APO-II. However, the limitedreactivity of Ole-II toward these substrates may not necessarily signala change in electronic structure but is more likely to stem from therestricted access of molecules to the heme iron when a fatty acid isbound. Consequently, the pseudo-first-order decay rate constantsexhibit a poor correlation with OH bond strength (Table S2).The predisposition of OleT to generate a substantial proportion

of fatty alcohol products with shorter, nonphysiological substrates(36, 38) highlights that the enzyme is able to simultaneouslyfunction as an oxygenase. We propose, based on analogy tomechanisms purported for desaturases, which can also exhibit dualfunctionality (reviewed in ref. 50), that bifurcation of the twopathways results from a competition of •OH rebound and ab-straction of an additional substrate electron by Ole-II. The latterpathway, coupled to the recruitment of a proton to restoreFe3+−OH2 (Fig. S8), would lead to generation of a carbocationpoised for C−Cα cleavage to liberate CO2. Although the intrinsicreactivity of Ole-II is most likely undervalued here, largely due tolimitations that stem from its method of preparation, the estimatedreduction potential (51) and oxidative prowess of APO-II (46)serve as a useful guide. As olefin and alcohol products result fromthe metabolism of chemically similar substrates, both rigid andprecise positioning of a carbon-centered radical would seem nec-essary to inhibit oxygen rebound. Extensive contact with hydro-phobic aromatic amino acids (e.g., Phe291, Phe79) may serve toanchor EA and provide a source of radical stabilization. Althoughregiospecific Cβ−H atom abstraction would most certainly need tobe primarily satisfied, it does not solely ensure subsequent over-oxidation to furnish a hydrocarbon by OleT and related CYP152orthologs (40, 52). Interesting comparisons are found with thefunctional divergence observed for nonheme iron enzymes andsynthetic complexes, which effectively negotiate •OH bond in-sertion, rebound of alternative radical species (22), and substratedesaturation (50, 53). We anticipate that further elucidation of thefactors that promote alkene formation will provide importantparallels with these systems in a different structural framework, andprovide opportunities to better leverage OleT activity with a broadersubstrate scope.

Materials and MethodsStopped-flow experiments were performed using an Applied Photophysics Ltd.(APP) SX.20 stopped-flow spectrophotometer with PMT or PDA detection asindicated. EA-boundOleTwas prepared by incubation of H2O2 pretreated anddesalted substrate-free enzyme (typically 10 μM to 40 μM) in 200 mMK2HPO4 (pH 7.4) with a threefold molar excess of EA (H39-EA or D39-EA),prepared as a 10-mM stock in 70% (vol/vol) ethanol:30% (vol/vol) Triton X-100,for 15 h at 4 °C. Undissolved fatty acid was removed by centrifugation for10 min at 3,000 × g before loading into a stopped-flow syringe. The enzymewas mixed at 4 °C with H2O2, similarly prepared in a 200-mM K2HPO4 (pH 7.4)buffer. See Supporting Information for fitting methods.

ACKNOWLEDGMENTS. We thank John H. Dawson and Michael T. Green forvaluable discussions. This work was supported byMagellan Scholar grants (to J.L.G.and M.E.M.) through the University of South Carolina (USC) Office of Undergrad-uate Research, an ASPIRE grant from the USC Vice President of Research (toT.M.M.), and National Science Foundation CAREER Grant 1555066 (to T.M.M.).

1. Ortiz de Montellano PR (2005) Cytochrome P450: Structure, Mechanism, and

Biochemistry (Kluwer Academic, New York), 3rd Ed.2. Denisov IG, Makris TM, Sligar SG, Schlichting I (2005) Structure and chemistry of cy-

tochrome P450. Chem Rev 105(6):2253–2277.3. Bangcharoenpaurpong O, Rizos AK, Champion PM, Jollie D, Sligar SG (1986) Reso-

nance Raman detection of bound dioxygen in cytochrome P-450cam. J Biol Chem

261(18):8089–8092.4. Hu S, Schneider AJ, Kincaid JR (1991) Resonance Raman studies of oxycytochrome P450cam:

Effect of substrate structure on ν(O−O) and ν(Fe−O2). J Am Chem Soc 113(13):4815–4822.5. Sharrock M, et al. (1976) Cytochrome P450cam and its complexes. Mössbauer pa-

rameters of the heme iron. Biochim Biophys Acta 420(1):8–26.

6. Davydov R, Macdonald IDG, Makris TM, Sligar SG, Hoffman BM (1999) EPR and ENDOR ofcatalytic intermediates in cryoreduced native and mutant oxy-cytochromes P450cam:

Mutation-induced changes in the proton delivery system. J AmChemSoc 121(45):10654–10655.7. Davydov R, et al. (2001) Hydroxylation of camphor by reduced oxy-cytochrome

P450cam: Mechanistic implications of EPR and ENDOR studies of catalytic interme-

diates in native and mutant enzymes. J Am Chem Soc 123(7):1403–1415.8. Mak PJ, et al. (2007) Resonance Raman detection of the hydroperoxo intermediate in

the cytochrome P450 enzymatic cycle. J Am Chem Soc 129(20):6382–6383.9. Kellner DG, Hung SC, Weiss KE, Sligar SG (2002) Kinetic characterization of compound

I formation in the thermostable cytochrome P450 CYP119. J Biol Chem 277(12):9641–9644.

Grant et al. PNAS | September 6, 2016 | vol. 113 | no. 36 | 10053

BIOCH

EMISTR

YCH

EMISTR

Y

10. Rittle J, Green MT (2010) Cytochrome P450 compound I: Capture, characterization,and C-H bond activation kinetics. Science 330(6006):933–937.

11. Groves JT, McClusky GA (1976) Aliphatic hydroxylation via oxygen rebound. Oxygentransfer catalyzed by iron. J Am Chem Soc 98(3):859–861.

12. Groves JT, McClusky GA (1978) Aliphatic hydroxylation by highly purified liver mi-crosomal cytochrome P-450. Evidence for a carbon radical intermediate. BiochemBiophys Res Commun 81(1):154–160.

13. Banerjee R, Proshlyakov Y, Lipscomb JD, Proshlyakov DA (2015) Structure of the keyspecies in the enzymatic oxidation of methane to methanol. Nature 518(7539):431–434.

14. Bukowski MR, et al. (2005) A thiolate-ligated nonheme oxoiron(IV) complex relevantto cytochrome P450. Science 310(5750):1000–1002.

15. Price JC, Barr EW, Glass TE, Krebs C, Bollinger JM, Jr (2003) Evidence for hydrogenabstraction from C1 of taurine by the high-spin Fe(IV) intermediate detected duringoxygen activation by taurine:α-ketoglutarate dioxygenase (TauD). J Am Chem Soc125(43):13008–13009.

16. Sastri CV, et al. (2007) Axial ligand tuning of a nonheme iron(IV)−oxo unit for hy-drogen atom abstraction. Proc Natl Acad Sci USA 104(49):19181–19186.

17. Valentine AM, Stahl SS, Lippard SJ (1999) Mechanistic studies of the reaction of re-duced methane monooxygenase hydroxylase with dioxygen and substrates. J AmChem Soc 121(16):3876–3887.

18. Bollinger JM, Jr, et al. (1991) Mechanism of assembly of the tyrosyl radical-dinucleariron cluster cofactor of ribonucleotide reductase. Science 253(5017):292–298.

19. Dassama LMK, et al. (2013) A 2.8 Å Fe−Fe separation in the Fe2III/IV intermediate, X,

from Escherichia coli ribonucleotide reductase. J Am Chem Soc 135(45):16758–16761.20. Galoni�c DP, Barr EW, Walsh CT, Bollinger JM, Jr, Krebs C (2007) Two interconverting

Fe(IV) intermediates in aliphatic chlorination by the halogenase CytC3. Nat Chem Biol3(2):113–116.

21. Martinie RJ, et al. (2015) Experimental correlation of substrate position with reactionoutcome in the aliphatic halogenase, SyrB2. J Am Chem Soc 137(21):6912–6919.

22. Matthews ML, et al. (2009) Substrate positioning controls the partition betweenhalogenation and hydroxylation in the aliphatic halogenase, SyrB2. Proc Natl Acad SciUSA 106(42):17723–17728.

23. Minnihan EC, Nocera DG, Stubbe J (2013) Reversible, long-range radical transfer inE. coli class Ia ribonucleotide reductase. Acc Chem Res 46(11):2524–2535.

24. Rettie AE, Boberg M, Rettenmeier AW, Baillie TA (1988) Cytochrome P-450-catalyzeddesaturation of valproic acid in vitro. Species differences, induction effects, andmechanistic studies. J Biol Chem 263(27):13733–13738.

25. Yoshimoto FK, Guengerich FP (2014) Mechanism of the third oxidative step in theconversion of androgens to estrogens by cytochrome P450 19A1 steroid aromatase.J Am Chem Soc 136(42):15016–15025.

26. Auclair K, Hu Z, Little DM, Ortiz De Montellano PR, Groves JT (2002) Revisiting themechanism of P450 enzymes with the radical clocks norcarane and spiro[2,5]octane.J Am Chem Soc 124(21):6020–6027.

27. Fukuda H, et al. (1994) Reconstitution of the isobutene-forming reaction catalyzed bycytochrome P450 and P450 reductase from Rhodotorula minuta: Decarboxylationwith the formation of isobutene. Biochem Biophys Res Commun 201(2):516–522.

28. Akhtar M, Calder MR, Corina DL, Wright JN (1982) Mechanistic studies on C-19 de-methylation in oestrogen biosynthesis. Biochem J 201(3):569–580.

29. Khatri Y, Luthra A, Duggal R, Sligar SG (2014) Kinetic solvent isotope effect in steady-state turnover by CYP19A1 suggests involvement of Compound 1 for both hydrox-ylation and aromatization steps. FEBS Lett 588(17):3117–3122.

30. Mak PJ, Luthra A, Sligar SG, Kincaid JR (2014) Resonance Raman spectroscopy of theoxygenated intermediates of human CYP19A1 implicates a compound I intermediatein the final lyase step. J Am Chem Soc 136(13):4825–4828.

31. Ji L, et al. (2015) Drug metabolism by cytochrome p450 enzymes: What distinguishesthe pathways leading to substrate hydroxylation over desaturation? Chemistry21(25):9083–9092.

32. Hackett JC, Brueggemeier RW, Hadad CM (2005) The final catalytic step of cyto-chrome p450 aromatase: A density functional theory study. J Am Chem Soc 127(14):5224–5237.

33. Belcher J, et al. (2014) Structure and biochemical properties of the alkene producingcytochrome P450 OleTJE (CYP152L1) from the Jeotgalicoccus sp. 8456 bacterium.J Biol Chem 289(10):6535–6550.

34. Rude MA, et al. (2011) Terminal olefin (1-alkene) biosynthesis by a novel p450 fattyacid decarboxylase from Jeotgalicoccus species. Appl Environ Microbiol 77(5):1718–1727.

35. Grant JL, Hsieh CH, Makris TM (2015) Decarboxylation of fatty acids to terminal al-kenes by cytochrome P450 compound I. J Am Chem Soc 137(15):4940–4943.

36. Dennig A, et al. (2015) Oxidative decarboxylation of short-chain fatty acids to 1-alkenes.Angew Chem Int Ed Engl 54(30):8819–8822.

37. Liu Y, et al. (2014) Hydrogen peroxide-independent production of α-alkenes byOleTJE P450 fatty acid decarboxylase. Biotechnol Biofuels 7(1):28.

38. Zachos I, et al. (2015) Photobiocatalytic decarboxylation for olefin synthesis. ChemCommun (Camb) 51(10):1918–1921.

39. Matsunaga I, et al. (2002) Enzymatic reaction of hydrogen peroxide-dependent per-oxygenase cytochrome P450s: Kinetic deuterium isotope effects and analyses byresonance Raman spectroscopy. Biochemistry 41(6):1886–1892.

40. Lee DS, et al. (2003) Substrate recognition and molecular mechanism of fatty acidhydroxylation by cytochrome P450 from Bacillus subtilis. Crystallographic, spectro-scopic, and mutational studies. J Biol Chem 278(11):9761–9767.

41. Wang X, Peter S, Kinne M, Hofrichter M, Groves JT (2012) Detection and kineticcharacterization of a highly reactive heme-thiolate peroxygenase compound I. J AmChem Soc 134(31):12897–12900.

42. Fujishiro T, et al. (2011) Crystal structure of H2O2-dependent cytochrome P450SPalphawith its bound fatty acid substrate: Insight into the regioselective hydroxylation offatty acids at the α position. J Biol Chem 286(34):29941–29950.

43. de Visser SP, Ogliaro F, Sharma PK, Shaik S (2002) What factors affect the re-gioselectivity of oxidation by cytochrome p450? A DFT study of allylic hydroxylationand double bond epoxidation in a model reaction. J Am Chem Soc 124(39):11809–11826.

44. Yosca TH, et al. (2013) Iron(IV)hydroxide pK(a) and the role of thiolate ligation in C-Hbond activation by cytochrome P450. Science 342(6160):825–829.

45. Wang X, Ullrich R, Hofrichter M, Groves JT (2015) Heme-thiolate ferryl of aromaticperoxygenase is basic and reactive. Proc Natl Acad Sci USA 112(12):3686–3691.

46. Lambeir AM, Dunford HB, Pickard MA (1987) Kinetics of the oxidation of ascorbicacid, ferrocyanide and p-phenolsulfonic acid by chloroperoxidase compounds I and II.Eur J Biochem 163(1):123–127.

47. Warui DM, et al. (2011) Detection of formate, rather than carbon monoxide, as thestoichiometric coproduct in conversion of fatty aldehydes to alkanes by a cyano-bacterial aldehyde decarbonylase. J Am Chem Soc 133(10):3316–3319.

48. Pandelia ME, et al. (2013) Substrate-triggered addition of dioxygen to the diferrouscofactor of aldehyde-deformylating oxygenase to form a diferric-peroxide in-termediate. J Am Chem Soc 135(42):15801–15812.

49. Rui Z, et al. (2014) Microbial biosynthesis of medium-chain 1-alkenes by a nonhemeiron oxidase. Proc Natl Acad Sci USA 111(51):18237–18242.

50. Shanklin J, Guy JE, Mishra G, Lindqvist Y (2009) Desaturases: Emerging models forunderstanding functional diversification of diiron-containing enzymes. J Biol Chem284(28):18559–18563.

51. Wang X, Peter S, Ullrich R, Hofrichter M, Groves JT (2013) Driving force for oxygen-atom transfer by heme-thiolate enzymes. Angew Chem Int Ed Engl 52(35):9238–9241.

52. Amaya JA, Rutland CD, Makris TM (2016) Mixed regiospecificity compromises alkenesynthesis by a cytochrome P450 peroxygenase from Methylobacterium populi. J InorgBiochem 158:11–16.

53. Bigi MA, Reed SA, White MC (2011) Diverting non-haem iron catalysed aliphatic C-Hhydroxylations towards desaturations. Nat Chem 3(3):216–222.

10054 | www.pnas.org/cgi/doi/10.1073/pnas.1606294113 Grant et al.