Nature uses carboxylate-bridged diiron centres at the active sites of enzymes that catalyse the selective hydroxylation of hydrocarbons to alcohols. The resting diiron(III) state of the hydroxylase component of soluble methane monooxygenase enzyme is converted by two-electron transfer from an NADH-requiring reductase into the active diiron(II) form, which subsequently reacts with O2 to generate a high-valent diiron(IV) oxo species (Q) that converts CH4 into CH3OH. In this step, C–H bond activation is achieved through a transition state having a linear C⋯H⋯O unit involving a bound methyl radical. Kinetic studies of the reaction of Q with substrates CH3X, where X=H, D, CH3, NO2, CN or OH, reveal two classes of reactivity depending upon whether binding to the enzyme or C–H bond activation is rate-limiting. Access of substrates to the carboxylate-bridged diiron active site in the hydroxylase (MMOH) occurs through a series of hydrophobic pockets. In the hydroxylase component of the closely related enzyme toluene/o-xylene monooxygenase (ToMOH), substrates enter through a wide channel in the α-subunit of the protein that tracks a course identical to that found in the structurally homologous MMOH. Synthetic models for the carboxylate-bridged diiron centres in MMOH and ToMOH have been prepared that reproduce the stoichiometry and key geometric and physical properties of the reduced and oxidized forms of the proteins. Reactions of the diiron(II) model complexes with dioxygen similarly generate reactive intermediates, including high-valent species capable not only of hydroxylating pendant C–H bonds but also of oxidizing phosphine and sulphide groups.
Bacterial multicomponent monooxygenases (BMMs; table 1) inhabit the Earth's surface at the aerobic/anaerobic interface where they consume hydrocarbons as their sole source of carbon and energy (Hanson & Hanson 1996; Leahy et al. 2003). BMMs are soluble systems that use a carboxylate-bridged diiron centre in their hydroxylase component to activate dioxygen for formal insertion into a C–H bond of the hydrocarbon substrate. The product alcohol is subsequently converted into biomass by the organism. Evolving from a common ancestor, four classes of BMMs have been identified (Leahy et al. 2003; Notomista et al. 2003). Each contains several proteins (figure 1; Leahy et al. 2003), including a hydroxylase, a reductase and a small regulatory component required for efficient coupling of electron consumption with hydrocarbon oxidation. Extensive experimental and theoretical studies conducted in our laboratory (Valentine et al. 1999; Ambundo et al. 2002; Baik et al. 2002; Guallar et al. 2002) and elsewhere (Pulver et al. 1993; Brazeau & Lipscomb 2000; Brazeau et al. 2001a,b; Siegbahn 2001; Baik et al. 2003) have provided much insight into the nature of the dioxygen and hydrocarbon activation steps in the catalytic mechanism of soluble methane monooxygenase (MMO), the flagship member of the BMM family. Consequently, we have gained considerable insight into the workings of this intriguing catalytic machine employed by nature to convert methane selectively to methanol, a difficult transformation to achieve under ambient conditions. In this article, we summarize the key findings of this work. In addition, we present the results of recent structural investigations of a related toluene/o-xylene monooxygenase (ToMO), which provide additional clues about the catalytic mechanism. Finally, we describe a selection of synthetic model complexes that provide insight into the physical and chemical properties of the carboxylate-bridged diiron cores in the enzymes.
2. Soluble methane monooxygenase
(a) Component proteins and the catalytic reaction cycle
Figure 2 displays the structures of the hydroxylase (MMOH), reductase (MMOR) and regulatory protein (MMOB) components of methane monooxygenase from Methylococcus capsulatus (Bath; Rosenzweig et al. 1993; Rosenzweig et al. 1995; Walters et al. 1999; Whittington & Lippard 2001; Whittington et al. 2001b; Karlsson et al. 2002; Müller et al. 2002; Chatwood et al. 2004). The carboxylate-bridged diiron cores of MMOH in its resting diiron(III) and reduced diiron(II) states, Hox and Hred, respectively, are depicted in figure 3. Spectroscopic studies and density functional theory (DFT) have revealed much about the nature of the catalytic cycle of the enzyme (Lee et al. 1993; Valentine et al. 1999; Siegbahn 2001; Baik et al. 2002). Hred reacts with O2 rapidly to form early, thus far uncharacterized, transient species (Liu et al. 1995a–c) that convert to Hperoxo, the first observable intermediate with well-defined Mössbauer and optical spectroscopic properties (Liu et al. 1995a,b; Valentine et al. 1999). As indicated in figure 4, in the absence of substrates, Hperoxo converts to Q, a diiron(IV) oxo intermediate with a short Fe⋯Fe distance of ∼2.5 Å (Shu et al. 1997) that has also been spectroscopically characterized (Lee et al. 1993; Liu et al. 1995a,b; Valentine et al. 1999). The Hperoxo and Q intermediate structures shown in figure 4 have been determined by DFT and remain to be confirmed experimentally. MMOB must be present for efficient generation of MMOH intermediates in the reaction cycle, and this component appears to gate the electron transfer and O2 binding steps of the hydroxylase in a manner that is not well understood (Liu et al. 1995c; Gassner & Lippard 1999). Various substrate probes have been used to define the nature of the C–H bond activating step in the reaction of hydrocarbon substrates with MMOH. In steady-state turnover experiments, with all three components present, substrates such as trans-phenylmethylcyclohexane and chiral ethane (CH3CHDT) convert to produce alcohols in such a manner that suggests that the transition state involves a bound radical species with a sub-picosecond lifetime (Priestley et al. 1992; Liu et al. 1993; Valentine et al. 1997; Choi et al. 1999). One substrate with results that are inconsistent with this conclusion is norcarane, for which a longer-lived radical species has been suggested (Brazeau et al. 2001a,b; Newcomb et al. 2002). Detailed discussion of these apparent inconsistencies can be found in Baik et al. (2003).
(b) Reactions of CH3X substrates with intermediate Q
Double-mixing stopped-flow investigations allow for the kinetic isolation of Q at various temperatures, after which substrates can be rapidly introduced. Comparison of the reactions of CH4 with CD4 reveals a kinetic isotope effect (KIE), kH/kD of 26, consistent with proton tunnelling in the transition state. DFT studies indicate a transition state energy of approximately 18 kcal mol−1 with a bound methyl radical having a linear C⋯H⋯O geometry, the formation of which is accompanied by electron transfer to the Fe2 site (figure 5), which is now in the Fe(III) oxidation state (Baik et al. 2002). Stereoelectronic details of the second electron transfer are provided in figure 5. Rotation of the bridging H–O group upward directs the lone pair of electrons on the oxygen atom toward the remaining electron on the bound methyl radical. If an electron of alpha spin transfers to Fe2 in forming the transition state, then the remaining electron on carbon has beta spin. Because it is known that, in the mixed-valent Fe(III)Fe(IV) species QX, generated from a frozen solution of Q by cryoreduction (Valentine et al. 1998), the iron atoms are antiferromagnetically coupled, the LUMO on Fe1 must have beta symmetry. As the H–O bond rotates upward, transfer of a beta spin electron from the lone pair on the bridging O-atom leaves an alpha spin in the p-orbital, which facilitates C–O bond formation with the beta spin on the bound methyl radical (figure 5). The stereoelectronic details of the methane to methanol conversion are satisfying, at least at the level of DFT (Baik et al. 2003).
Extension of these kinetic methods to substrates other than methane and its isotopomers, however, revealed interesting complexities. An investigation of the reaction of Q with C2H6 and C2D6 indicates that, despite the fact that the C–H bond in ethane is approximately 5.6 kcal mol−1 weaker than that in methane, the rate constant for reaction with ethane is the same as that for methane, and surprisingly, there is no appreciable deuterium isotope effect (Brazeau & Lipscomb 2000; Ambundo et al. 2002). This result suggests that for ethane there must be a rate-limiting step that precedes C–H bond cleavage, most probably binding to the active site pocket of MMOH (Brazeau et al. 2001a,b; Gherman et al. 2004). Although such a process had long been postulated for MMOH (Valentine et al. 1997; Brazeau & Lipscomb 2000), the first experimental proof was provided by the reaction of CH3NO2 with intermediate Q (Ambundo et al. 2002; Muthusamy et al. 2003). Because the solubility of CH3NO2 in aqueous media is greater than that of methane or ethane, it is possible to attain sufficiently high concentrations such that, in a plot of kobs versus CH3NO2 concentration, saturation behaviour is observed. Figure 6 illustrates this effect as well as the kinetic mechanism to account for the data and results for CD3NO2, from which a KIE of 8.1 could be determined (Ambundo et al. 2002). A stopped-flow infrared spectroscopic analysis of the reaction of Q with CD3NO2, monitoring the N–O stretching band at 1548 cm−1, yielded the same kobs as determined by optical spectroscopy for the reaction. This result provides conclusive evidence that the diiron(IV) intermediate is the active species for the conversion of methane and its derivatives to the corresponding alcohols (Muthusamy et al. 2003). Analysis of the DFT energy profiles for reactions of Q with additional CH3X substrates acetonitrile, methanol and methyl fluoride, as well as experimental data for the first two of these molecules, further delineated two classes of reactivity (Gherman et al. 2005). Class I substrates are defined as those in which the C–H bond breaking step is rate determining, whereas, in class II, substrate binding to the protein is the slow step in the mechanism.
3. Toluene/o-xylene monooxygenase hydroxylase
The inability to clone and express active MMOH in a heterologous host organism, such as Escherichia coli, in sufficient quantities to perform detailed mechanistic and structural studies led us to expand the scope of our activities to include other MMOs for which such expressed protein was more readily available. Accordingly, in collaboration with the Di Donato laboratory in Naples, we have been investigating the protein components of the ToMO and phenol hydroxylase (PH) families. Recently, we determined the structure of the ToMO hydroxylase enzyme (ToMOH), in its oxidized form and containing bound azide or 4-bromophenol (Sazinsky et al. 2004). Both ToMOH and MMOH are (αβγ)2 dimers having a twofold symmetry axis relating the individual (αβγ) protomers. As indicated in figure 7, the overall topologies of MMOH and ToMOH are quite similar, especially for the two major subunits α and β. The γ subunits have completely different folds, however, and are positioned differently with respect to α and β. The reason for this difference is presently unknown. The active sites of ToMOH and MMOH in the resting diiron(III) state are also comparable (figure 8). The major differences are: (i) the use of an ε N-atom of the histidine side chain bound to Fe2 in the former and (ii) the presence of thioglycolate from the purification buffer in a position bridging the two iron atoms rather than hydroxide ion, water or , as found in MMOH (Sazinsky et al. 2004). The most striking feature differentiating the structure of ToMOH from that of MMOH is the presence of a surface-accessible channel passing from the carboxylate-bridged diiron centre through the α subunit and forking at the exit into the protein exterior through two clearly visible openings at the protein surface (figure 9). This channel is 35–40 Å in length and 6–10 Å wide, which is sufficient to permit the passage of aromatic substrates or products. In crystals soaked with 4-bromophenol, three of these product analogues are clearly delineated in the electron density maps that reside in the channel. Comparison of the pathway traced by the channel with the positions of three hydrophobic pockets in MMOH, through which methane and dioxygen had previously been postulated to travel (Rosenzweig et al. 1993; Whittington et al. 2001a), reveals that the two trace very similar paths. This finding strongly implies the existence of universal pathways for small molecule access to and egress from the diiron cores of bacterial MMOs. Binding of substrates in the ToMOH channel, and along the pathway defined by the cavities in MMOH, can be either rate-limiting or not, according to their classification as defined in the preceding section. Future work on ToMOH and the other components of the ToMO system will include site-directed mutagenesis. Of particular interest is to test the hypothesis (Sazinsky et al. 2004) that the inability to detect intermediates in the ToMOH catalytic cycle, analogous to those in MMOH, is the result of their quenching by buffer components that have much better access to reactive species like Q owing to unimpeded access through the channel.
4. Synthetic models for the MMOH and ToMOH diiron cores
(a) Diiron(II) and resting state diiron(III) core models
The preparation and study of structural, spectroscopic and functional models for the active sites in metalloproteins can sharpen and extend our understanding of their metal-containing cores. In the case of the MMO hydroxylase, the ultimate goal is to obtain carboxylate-bridged diiron units with the stoichiometry and geometry of the enzyme that can react with dioxygen, generating intermediates analogous to Hperoxo and Q, as well as hydroxylate C–H bonds. Obtaining the requisite complexes is challenging given the kinetic lability of high-spin Fe(II) and Fe(III) complexes and their propensity to convert to unreactive, thermodynamically stable species of little relevance to the metalloprotein cores (Lippard 1988, 2002). Recently, we (Lee & Lippard 1998) and others (Tolman & Que 2002) discovered that with the use of sterically hindered carboxylate ligands such as 2,6-di(p-tolyl)benzoate (−O2CArTol), it is possible to prepare by self-assembly both diiron(II) and diiron(III) complexes having the [Fe2(O2CR)4(N-donor)2] composition of the cores of MMOH in its reduced and oxidized states, respectively (Lee & Lippard 1998; Yoon & Lippard 2004). Synthetic routes to these complexes are depicted in figure 10. A significant challenge in this chemistry has been to position the N-donor ligands on the same side (syn) of the Fe–Fe vector, as in the hydroxylase enzymes. This goal was accomplished by introducing the N-donors as quinoline moieties provided by the 1,2-diethynylbenzene-based ligand Et2BCQEBEt, or 1,2-bis(3-ethynyl-8-carboxylatequinoline)benzene ethyl ester. Reaction of this ligand with an iron(II) precursor under the conditions in figure 10 afforded [Fe2(Et2BCQEBEt)(μ-O2CArTol)3]+, in which the N-donors are in the syn positions (figure 11; Kuzelka et al. 2003). Although this complex provides proof of principle for the approach, significantly more work must be carried out to achieve a stoichiometric and geometric match to the reduced MMO hydroxylase cores.
(b) Dioxygen reactions to afford iron(IV) transient species
Introduction of dioxygen into CH2Cl2 solutions of [Fe2(O2CArTol)4(N-donor)2], where the N-donor is pyridine or a derivative thereof, at −78 °C leads to the formation of a deep green colour characteristic of the formation of the mixed-valent Fe(II)Fe(III) oxidized product (Lee et al. 2002). Analysis of the solution by EPR and Mössbauer spectroscopy revealed the concomitant formation of a mixed-valent Fe(III)Fe(IV) complex that is capable of oxidizing phenolic substrates in a manner analogous to the oxidation of tyrosine present at the active site of the R2 subunit of class I ribonucleotide reductase (Baldwin et al. 2001). In this enzyme, which has a carboxylate-bridged diiron active site similar to those in the BMM hydroxylases, an Fe(III)Fe(IV) intermediate is formed by reaction of the reduced diiron(II) form with O2 for the express purpose of forming a functional tyrosyl radical. The failure of our model chemistry to generate transient species that resemble Hperoxo or Q has been attributed to their being rapidly quenched by the unoxidized diiron(II) precursor. As illustrated in figure 12, such a reaction concomitantly produces both Fe(II)Fe(III) and Fe(III)Fe(IV) species (Lee et al. 2002). In order to circumvent this problem, a series of complexes were prepared, in which the substrates are positioned to be in close proximity to the putative high-valent diiron core. The chemistry of these compounds with dioxygen is discussed in the following section.
(c) Oxidation of tethered substrates
When [Fe2(μ-O2CArTol)2(O2CArTol)2(NH2CH2CH2N(CH2Ph)2)2] is allowed to react with dioxygen, oxidative N-dealkylation of a tethered benzyl substituent occurs, as illustrated in figure 13 (Lee & Lippard 2001). An 18O-labelling experiment confirmed that >90% of the oxygen in the benzaldehyde product comes from dioxygen. Possible mechanisms for the reaction are shown in figure 14, one of which is analogous to related chemistry in cytochrome P450, and studies are currently in progress to distinguish between them. Diiron(II) compounds in the [Fe2(O2CArTol)4(N-donor)2] class are in equilibrium between tetra(μ-carboxylato)diiron(II) and di(μ-carboxylato)dicarboxylatodiiron(II) forms (figure 15; Lee & Lippard 2002), and preliminary evidence favours the latter as the one that reacts with dioxygen.
The analogous benzylamine (BA) and 4-methoxybenzylamine (BAp-OMe) complexes [Fe2(μ-O2CArTol)4L2], where L=BA or BAp-OMe, also react with O2, forming benzaldehyde after putative hydroxylation of the benzylic C–H bond followed by N-dealkylation of the resulting aminoalcohol. A KIE, kH/kD∼3, for oxidation of [Fe2(μ-O2CArTol)4(α-d1-BAp-OMe)2] at −78 °C in CH2Cl2 confirms that the reaction mechanism involves hydroxylation of the benzylic C–H bond in the product-determining step (Yoon & Lippard 2003). The proximity of this bond to a putative high-valent iron oxo intermediate presumably facilitates the reaction.
The terphenyl-based carboxylate-bridged diiron platform has also been employed to study the oxidation of additional tethered substrates apart from alkyl groups. Reaction of [Fe2(μ-O2CArTol)3(O2CArTol)(N-donor)] complexes in which the N-donor contains an appended phosphine or sulphide moiety with dioxygen leads to efficient conversion to the corresponding phosphine oxide or sulfoxide, respectively (Carson & Lippard 2004). The efficiency of the conversion depends upon the relative proximity of the appended substrate to the oxidized form of the carboxylate-bridged diiron centre responsible for its oxidation. In a recent study, reaction of [Fe2(μ-O2CAr4-FPh)3(O2CAr4-FPh)(2-Ph2Ppy)] with dioxygen in CH2Cl2 followed by crystallization of the final product of the oxidation led to the identification of [Fe2(μ-OH)2(μ-O2CAr4-FPh)(OH2) (2-Ph2P(O)py)(O2CAr4-FPh)3] as the end product (figure 16; E. C. Carson & S. T. Lippard 2004, unpublished work). This species has many features in common with the oxidized form of MMOH, including the terminal water molecule, one of the iron atoms. Investigations to determine the mechanism of phosphine oxidation are currently in progress.
5. Toward the future
Although much has been learned about how nature uses multicomponent monooxygenases to convert hydrocarbons selectively to alcohols, significant gaps in our knowledge remain. More work must be carried out to understand the factors that differentiate the reactivity of class I and class II substrates with intermediate Q of MMOH. The role of the coupling protein MMOB in modulating the reactions of MMOH will only begin to be elucidated following determination of the crystal structure of this component in complex with the hydroxylase. Expression of soluble MMOH in a heterologous host, such as E. coli, would facilitate site-directed mutagenesis experiments to test further aspects of the catalytic mechanism. The finding that olefins react with the peroxo intermediate in MMOH (Beauvais & Lippard in press; Valentine et al. 1999) has opened another line of investigation for future work. Bringing our knowledge of the other BMMs up to the same level as that for methane monooxygenase will require much additional study. In the synthetic models area, the chief goals for the near future are to obtain systems with syn-disposed N-atoms that have the stoichiometry of the protein diiron cores. The reactivity of these complexes with tethered and free hydrocarbon substrates and achieving catalysis are high priorities.
This work was supported by grants from the NIGMS and NSF and was performed by many graduate students, postdoctoral associates and collaborators, the latter including Professors R. Friesner at Columbia University, R. Thorneley at the John Innes Center in Norwich, UK and A. Di Donato at the Universita' di Napoli Federico II in Naples, Italy. No attempt was made to provide comprehensive coverage of the literature. More discussion and many additional references may be found in several recent review articles from our laboratory (Du Bois et al. 2000; Baik et al. 2003; Lee & Lippard 2003; Tshuva & Lippard 2004). I thank Leslie Murray for technical assistance in the preparation of this manuscript.
One contribution of 19 to a Discussion Meeting ‘Catalysis in chemistry and biochemistry’.
- © 2005 The Royal Society