Small Molecule Tunnels in Metalloenzymes Viewed as Extensions of the Active Site

Conspectus:

Rigorous substrate selectivity is a hallmark of enzyme catalysis. This selectivity is generally ascribed to a thermodynamically favorable process of substrate binding to the enzyme active site based upon complementary physiochemical characteristics, which allows both acquisition and orientation. However, this chemical selectivity is more difficult to rationalize for diminutive molecules that possess too narrow a range of physical characteristics to allow either precise positioning or discrimination between a substrate and an inhibitor. Foremost among these small molecules are the dissolved gases such as H₂, N₂, O₂, CO, CO₂, NO, N₂O, NH₃ and CH₄ so often encountered in metalloenzyme catalysis. Nevertheless, metalloenzymes have evolved to metabolize these small molecule substrates with high selectivity and efficiency.

The soluble methane monooxygenase enzyme (sMMO) acts upon two of these small molecules, O₂ and CH₄ to generate methanol as part of the C1 metabolic pathway of methanotrophic organisms. sMMO is capable of oxidizing many alternative hydrocarbon substrates. Yet remarkably, it will preferentially oxidize methane, the substrate with the fewest discriminating physical characteristics and the strongest C-H bond. Early studies led us to broadly attribute this specificity to formation of a ‘molecular sieve’ in which a methane- and oxygen-sized tunnel provides a size-selective route from bulk solvent to the completely buried sMMO active site. Indeed, recent cryogenic and serial femtosecond ambient temperature crystallographic studies have revealed such a route in sMMO. A detailed study of the sMMO tunnel considered here in the context of small molecule tunnels identified in other metalloenzymes, reveals three discrete characteristics that contribute to substrate selectivity and positioning beyond that which can be provided by the active site itself. Moreover, the dynamic nature of many tunnels allows exquisite coordination of substrate binding and reaction phases of the catalytic cycle. Here we differentiate between the highly selective molecular tunnel, which only allows one-dimensional transit of small molecules, and the larger, less selective, channels found in typical enzymes. Methods are described to identify and characterize tunnels as well as to differentiate them from channels. In metalloenzymes which metabolize dissolved gases, we posit that the contribution of tunnels is so great that they should be considered as extensions of the active site itself. A full understanding of catalysis by these enzymes requires an appreciation for the roles played by tunnels. Such understanding will also facilitate the use of the enzymes or their synthetic mimics in industrial or pharmaceutical applications.

Graphical Abstract

INTRODUCTION

The majority of the metalloenzymes that catalyze reductive and oxidative transformations of small molecule dissolved gases (e.g. H₂, N₂, O₂, CO, CO₂, NO, N₂O, NH₃ and CH₄) possess buried active sites. Often, these sites are accessible only via a molecular tunnel through the enzyme that connects the active site with bulk solvent, or in some cases, one active site to another. Here we use “tunnel” to define a passage approximating the diameter of the small molecule that allows movement in only one dimension. Most enzymes do not use tunnels, but rather establish a larger channel between bulk solvent and the active site that allows three-dimensional motion of the substrate as it seeks a binding position. The geometry of a molecular tunnel allows it to contribute significantly to the selectivity, specificity, and temporal progress of the reaction cycle beyond acting as a simple passageway.

Tunnels for Substrate and product transport in soluble methane monooxygenase (sMMO)

We became aware of many of the properties of tunnels that allow their unique contributions to catalysis through our study of the structure and mechanism of sMMO.^1–5 sMMO catalyzes conversion of two small molecule substrates, methane and oxygen, to methanol and water in the first step of the C1 metabolic pathway of methanotrophs.^5–8 Breaking the 105 kcal/mol C-H bond of methane requires a powerful oxidant. Indeed, sMMO generates Nature’s most powerful oxidant during its catalytic cycle: a unique oxygen-bridged diiron(IV) species termed Q.^9,10 It is expected that since Q can oxidize methane, it will also oxidize all other aliphatic hydrocarbons because they have weaker C-H bonds. This expectation has been borne out by experiments, which have shown that virtually any molecule with a C-H bond or a C-C double bond can be oxidized by Q if it can gain access to the active site.^11,12 The extreme substrate range of Q poses a problem for the methanotroph, which survives solely upon the oxidation of methane. It implies that sMMO must have a mechanism to selectively favor oxidation of methane, the most stable and physically nondescript substrate, over all others. The methanotroph oxidizes the sMMO product methanol utilizing a series of specialized enzymes that supply electrons or NADH for all aspects of bacterial metabolism. Thus, sMMO must avoid over-oxidation of methanol, a feat that has stymied development of small molecule catalysts for hydrocarbon oxidation.¹³ sMMO utilizes tunnels in addition to other strategies to address both of these fundamental problems.

No physical technique has shown evidence for methane binding to the diiron cluster of sMMO, and this observation has been supported by most (but not all) of the density functional theory and QM/MM calculations on the chemical mechanism of methane oxidation by Q.^5,7 Therefore, there is no selectivity for methane that can be afforded by coordinating the metal center. Also, transient-kinetic studies have shown that the diferrous form of sMMO reacts first with O₂ to start the process that forms Q several steps later. Only after Q forms does methane access the active site cavity from bulk solvent for reaction.⁹ Thus, the common mode of substrate selection and metalloenzyme regulation in which organic substrate binding to or near the metal potentiates O₂ binding and activation is not used by sMMO.^14,15

The key to understanding the regulation of the binding of O₂ and CH₄ to sMMO is to recognize the roles played by complex formation between the three proteins comprising the enzyme: the hydroxylase MMOH that contains the diiron active site, the small regulatory protein MMOB, and the reductase MMOR that transfers electrons from NADH to the MMOH metallocenter.⁵ The bulk of the regulation, especially pertaining to O₂ and CH₄ binding, is orchestrated by MMOB in complex with MMOH. Recently, we have been able to obtain high-resolution crystal structures of the MMOH:MMOB protein complex for the sMMO enzyme from Methylosinus trichosporium OB3b using two different methods: traditional cryogenic 100 K crystallography and serial femtosecond crystallography at room temperature at X-ray free electron laser (XFEL) sites.^1,2 These protein complex structures were solved for the resting diferric state of MMOH, both before and after turnover. Most importantly, the structure of the reduced diferrous state was solved, because this state is primed to bind O₂ and initiate catalysis.

One of the questions that these crystal structures have helped clarify is the manner by which MMOB binding to reduced MMOH engenders a 1000-fold increase in the rate constant of O₂ binding to the diiron cluster.¹⁶ Two MMOB binding-induced structural reorganization events within diferrous MMOH were found to be responsible for this enhanced O₂-reactivity. First, a conformational change removes a steric block caused by an iron ligand, enabling O₂ to bind in a cis-μ-1,2 fashion between the irons of the cluster.^1,17 Second, a tunnel leading from bulk solvent through the MMOH:MMOB protein interface and into the buried active site of MMOH is organized and opened.² This 50 Å-long tunnel, termed the W308-tunnel after a gating Trp residue at its entry into MMOH, maintains a diameter close to that of O₂ (Figure 1). A detailed description of the residues that line the tunnel and changes in the tunnel structure in various states of the sMMO system was recently published.² The relevance of the W308-tunnel was ascertained by making MMOB protein mutants that block the channel as it passes through the MMOH:MMOB interface.² These changes reduced the rate constant of O₂ binding to the diiron cluster by as much as 25,000-fold, and shifted the rate-limiting step to entry of O₂ from solution. Transient-kinetic studies using MMOB variants show that residues that profoundly influence the hydrocarbon substrate binding process are located in close proximity to the W308-tunnel.² Thus, the W308-tunnel may also traffic methane to the active site, probably after conformational changes in response to the progressive oxidation of the diiron cluster in the reaction cycle.

Figure 1.

The W308-tunnel (cyan surface) seen in the crystal structure of diferrous MMOH^red:MMOB (PDB:6YDI, panel A). The largely hydrophobic residues that line the tunnel are shown as yellow sticks. The O₂ and CH₄ molecules proposed to transit the tunnel are shown to scale in the upper left. The first major constriction in the tunnel (dashed box in Panel A) is at the point of entrance into the MMOH protein (enlarged in panel B) at the MMOH:MMOB interface. The tunnel is shown open but W308 and P215 rotate dramatically in the absence of MMOB to close the tunnel. Pairs such as A219 and T304 establish the tunnel width throughout its length. (Panel C): MMOB residues that affect the rate constants for catalysis are shown in juxtaposition to the tunnel. The “Quad” residues (three of which, S109, S110, and T111, are shown is space filling format) alter rate constants of P*, P, and Q formation and decay. Modification of MMOB-V41 to Arg causes a 25,000-fold decrease in the rate constant for O₂ binding, putatively by blocking the W-308 tunnel. This figure and other structure figures in this Account were produced using PyMol version 2.3.3 (Schrödinger).

The narrow aspect of the W308-tunnel along its entire length suggests that O₂ (and CH₄) can only move in one-dimension along the tunnel. A competing model for O₂ and CH₄ entry into the active site envisions passage through an interconnected chain of large interior protein cavities.^18,19 These MMOH cavities are structurally conserved in the broader family of bacterial multicomponent monooxygenases and have been experimentally shown to play an O₂ transport role in toluene/o-xylene monooxygenase.²⁰ These two models can be evaluated by the kinetic behavior of the Q reaction with methane. The large interior protein cavities in MMOH would be expected to pre-bind multiple CH₄ molecules as demonstrated by MMOH crystal structures soaked with alternative small molecule substrates.¹⁸ In this scenario, the rate constant for Q reaction with methane would display a saturation behavior, in contrast to the observed linear dependence. Linear dependence could come about either from rate-limiting binding of methane from solution or rapid binding followed by slow reaction where each binding event results in a substrate oxidation. As described below, the large deuterium KIE for methane oxidation is only consistent with the latter scenario at temperatures above 15 °C.^4,21 Efficient one-dimensional transit in the W308-tunnel would prevent pre-reaction substrate accumulation (See Supporting Information).

In this Account, we have compared the salient features of the W308-tunnel with the tunnels identified in other metalloenzymes with small molecule substrates (Figure 2). This comparative study highlights the manner in which small molecule tunnels select for substrates and engender catalytic regulation. The identification of small diameter tunnels in enzymes is a complex process that can easily lead to misidentification. The most commonly used methods and some cautions for their application are described in Supporting Information.

Figure 2.

Small molecule tunnels identified in enzyme and protein systems. The active site-containing protein is represented as a green cartoon while additional binding proteins are shown as a wheat colored cartoon. The enzymes along with the PDB identifiers and references to the methods used to identify the tunnels in the respective panels are: A: sMMO,^2,18,22 B: toluene 4-monooxygenase,^20,22 C: [NiFe]-hydrogenase,^23–27 D: nitrogenase,^28–31 E: nitric oxide reductase,^32,33 F: cytochrome ba3 oxidase,^33,34 G: lipoxygenase,^35–38 H: cytochrome P450 BM-3,^39,40 I: H-NOX protein,^41,42 J: α-ketoglutarate dependent-dioxygenase AlkB,^43,44 K: carbon monoxide dehydrogenase/acetyl-CoA synthase.^45–48 Tunnel identification methods are designated by numbers: 1=geometric methods, 2=Molecular dynamics simulations, 3=xenon pressurization, and 4=mutation of tunnel lining residues (see Supporting Information). P=Polar, NP=Nonpolar.

TUNNELS ENGENDER SUBSTRATE SELECTIVITY BASED UPON POLARITY

As illustrated in Figure 1 Panel A, the W308-tunnel of sMMO is predominately non-polar along its entire length as expected in order to selectively bind O₂ and probably also methane. Importantly, the entrance to the W308-tunnel from bulk solvent is increased in hydrophobicity by formation of the complex with MMOB. A dome of hydrophobic MMOB residues lines a new path passing along the MMOH:MMOB interface leading to the W308-tunnel entrance into the core of MMOH.² This theme of creating a tunnel with the appropriate polarity as the first determinant in substrate selection is maintained in other enzymes that utilize tunnels for substrate selection and access. Figure 3A summarizes the experimental values of dipole moment for various small molecule substrates and/or products for metalloenzymes. With a few exceptions, such as NH₃, superoxide, and CH₃OH, most of the small molecule substrates for metalloenzymes are non-polar in nature. Figure 3B depicts the distribution of amino acid residue type based upon polarity for tunnels that have been experimentally verified in a variety of metalloenzymes. Polar molecules such as NH₃ produced by the reduction of dinitrogen in the nitrogenase enzyme are believed to migrate within the tunnel by the formation and breakage of transient hydrogen-bonds with the polar and charged residues lining the tunnel.³¹ This “like dissolves like” model however, cannot account for the discrimination observed in [NiFe] and [FeFe] hydrogenase enzymes between the H₂ substrate and similarly non-polar O₂ and CO inhibitors, thus another level of selectivity is required as described next.²³

Figure 3.

A: Physical parameters for small molecule substrates and products.^49–51 B: Distribution of tunnel-lining residues based upon polarity for metalloenzymes that traffic polar and non-polar small molecules (listed above each column). The enzymes are: sMMO,² lipoxygenase (LOX),³⁷ cytochrome ba3 (Cyt ba3),³⁴ carbon monoxide dehydrogenase (CODH),⁴⁶ membrane-bound [NiFe] hydrogenase (MBH),²⁷ NAD⁺ synthetase (hsNadE),⁵² carbamoyl phosphate synthetase (CPS),⁵³ Class 1b ribonucleotide reductase β subunit (NrdF).⁵⁴ The small molecule tunnels were identified by CAVER using the literature parameters and cross-checked for accuracy. The residues lining the tunnel were identified by CAVER and then manually assessed by inspecting the crystal structure with the overlaid tunnel.

TUNNELS ENGENDER SUBSTRATE SELECTIVITY BASED UPON MOLECULAR SIZE

In practical terms, polarity of the small substrate tunnel can be viewed as nearly a binary effect, thereby limiting its ability to confer fine discrimination. This is not the case for selectivity on the basis of size. Figure 3A shows that there is a larger variation in the molecular diameter of the non-polar small molecules than there is in their dipole moment values, potentially allowing a tunnel of the appropriate diameter to function as a primary selective factor. This tunnel diameter can be uniform along its entire length or constricted at specific sites within the tunnel termed bottleneck regions. Using examples from the sMMO and the hydrogenase literature, we will show that the latter model is largely correct.

Based solely on bond dissociation energy, the 5 kcal/mole stronger C-H bond of methane over that of ethane predicts a 4000-fold difference in bond-breaking rate constant. The fact that the reactions with Q for these two substrates proceed with similar rate constants implies significant regulation within the sMMO system.^4,5 This ability of sMMO to show selectivity for methane over larger hydrocarbons was explained by a “molecular sieve” model based on studies revealing a different rate-limiting step for methane than for other substrates.³ For oxidation of methane, the chemical step of C-H bond cleavage by Q was found to be rate-limiting above 15 °C. A non-classical substrate deuterium-kinetic isotope effect (D-KIE) value of ~50, a ratio of Arrhenius factors A_H/A_D <<1, and a larger activation energy for CD₄ vs CH₄ were observed, characteristic of a quantum tunneling process.^21,55–57 For all hydrocarbon substrates larger than methane, no D-KIE was observed (Figure 4). It was proposed that the larger substrates are rate-limited by intramolecular diffusion into the active site.⁴ Thus, the ~4000-fold enhancement in sMMO methane oxidation over ethane oxidation arises from two phenomena. About 7-fold arises because the methane C-H bond breaking reaction by Q is accelerated by quantum tunneling (D-KIE of 50 vs. a classical limit of 7).⁵⁸ The remaining 700-fold arises from a size-restricted route for substrate entry that slows ethane to the point where its comparatively facile bond-breaking reaction is rate-limited by binding.^4,58 It is interesting to note that mutation of four MMOB residues in the “quad region” of the MMOH:MMOB interface appears to widen the tunnel through MMOH in the protein:protein complex.⁵⁹ The enlarged tunnel accelerates ethane binding so that C-H bond breaking becomes rate-limiting. However, the observed D-KIE of only 2 suggests a much smaller quantum tunneling contribution. For wild type MMOB at temperatures below 15 °C, the rate constants for ethane and methane binding are similar, but those for the slightly larger propane (Figure 4), propene, furan, or nitrobenzene⁴ are further slowed. The restrictive molecular tunnel allows methane to compete successfully in a mixed substrate environment by shifting the limiting factor from bond energy to molecular size.

Figure 4.

Eyring plots for the observed reaction of Q decay with methane, ethane and propane at a concentration of 200 μM.^4,58

The binding rate constant for methane can be approximated from the Eyring plot because the break observed at about 15 °C indicates similar binding and C-H bond cleaving constants (Figure 4).^4,58 The C-H cleavage rate constant by Q has been shown to be approximately 1.4×10⁴ M⁻¹s⁻¹ at 4 °C,⁴ which is at least three orders of magnitude below typical small molecule binding rate constants for enzymes.⁶⁰ Remarkably, the small decrease in size for O₂ (Figure 3A), allows it to bind at least two orders of magnitude faster than methane at the same concentration.^2,9,16 Thus, the W308-tunnel in sMMO allows the discrimination based on size that is key to both selectivity and timing of each substrate entry in the reaction cycle.

Another family of enzymes that has to overcome the challenge of discriminating between small, non-polar substrates are the [NiFe] and [FeFe] hydrogenases. The gas tunnels have to select for H₂ over CO (competitive inhibitor) and O₂ (irreversible inhibitor).²³ This selectivity was investigated by the study of a range of homologous members of the [NiFe] hydrogenase family that had varying sensitivity to O₂ inhibition. It was observed that this sensitivity correlated with the diameter of a constriction within the proposed small molecule tunnel. This bottleneck is present close to the active site and is known as the 74–122 motif based upon the amino acid residue numbering in the Desulfovibrio fructosovorans [NiFe]-hydrogenase.^23,25 A range of mutations was made at the two residues controlling the bottleneck in order to modulate the diameter of the constriction.²³ An analysis of the kinetic parameters of CO and O₂ inhibition and H₂ oxidation concluded that increasing the size of the side-chain group for the bottleneck residues led to a dramatic decrease in the rate of intramolecular diffusion of O₂ and CO and a less pronounced increase in the K_m for H₂. Comparing the experimental values of CO and O₂ diffusion across the range of hydrogenase mutants also demonstrated that the rates of intramolecular diffusion of O₂ and CO are similar within the gas tunnel. The similar molecular diameter of these two gases matches this experimental observation (Figure 3A). The relatively modest perturbation of the K_m for H₂ was shown to arise from two factors. First, the rates of diffusion of H₂ are much higher compared to those of CO and O₂. Indeed, the hydrogenase mutant with the most constricted bottleneck displayed a 30-fold higher rate of diffusion for H₂ over CO. Second, the K_m parameter is a composite of the rate constants, such that those of late unperturbed catalytic steps limit the modulation of the H₂ K_m value. A follow-up study employing multiscale MD simulations with Markov state modelling further clarified the understanding of gas transfer through this bottleneck in the mutant enzymes.²⁵ It demonstrated that the varying rates of gas diffusion did not simply correlate with the diameter of the tunnel at the bottleneck. Instead, there are two discrete constriction points at this bottleneck created by the juxtaposition of three residues that together create dynamic barriers to diffusion (Figure 5). Similar bottlenecks have also been identified in the [FeFe]-hydrogenases and verified by site-directed mutants that reduced the intramolecular diffusion rate of the CO inhibitor.²⁶

Figure 5.

Multiple small molecule tunnels present in the [NiFe]-hydrogenase from Desulfovibrio fructosovorans that coalesce into a single tunnel near the active site. The constriction in the tunnel lies adjacent to the [NiFe] active site cluster (panel B). We used MOLE2.5 with an interior threshold value of 1.25 Å and a bottleneck radius of 1.25 Å to verify the tunnels (see Supporting Information).

TUNNELS ENABLE TEMPORAL CONTROL OF SUBSTRATE DELIVERY

Perhaps the feature which most clearly differentiates enzyme catalysis from that of small molecule mimics is the ability to dynamically alter structure to control active site residue placement and substrate access or product release. Indeed, dynamic structural change is an important factor in the function of small molecule tunnels. This “gating” function can involve a single strategically placed amino acid residue sidechain, secondary structure conversion, or effects propagated through entire protein domains in response to allosteric binding. A detailed review of gating in enzymes has recently appeared.⁶¹ In studies probing the functional role of gates, the attention is typically focused upon the dramatic “fully open” and “fully closed” states of the gate, as these conformations can most readily be observed in static crystal structures. An effective gate in the context of a substrate access tunnel requires that it be synchronized to the temporal succession of chemical steps within the catalytic cycle. In this section, we describe how the regulated opening and closing of gates can engender such temporal control.

Our recent X-ray crystallography studies of the MMOH:MMOB complex have concisely illustrated how gating residues in a non-polar gas tunnel engender temporal control of substrate (oxygen) entry and putatively on product (methanol and water) egress. The W308-tunnel possesses two gates: (i) the eponymous W308 gate at the MMOH:MMOB protein interface and (ii) a gate in the MMOH interior in which a mobile Phe282 interacts with a static Phe212.² Both gates use the sterically bulky and planar aromatic side-chains to block the tunnel. Based upon a comparison of diferric and diferrous MMOH structures with and without MMOB bound, it was recognized that the W308-tunnel is dynamically re-configured during the catalytic cycle to allow O₂ delivery only in the MMOH^red:MMOB complex.² This state of the enzyme is primed to bind O₂ to the diiron cluster to start catalysis. Rotations and shifts of completely conserved and juxtaposed residues W308 and P215, and (orthogonally aligned) A219 and T304 cause the W308 gate to open upon diiron cluster reduction of MMOH (Figure 1, panel B and Figure 6), while similar motions cause the F212/F282 gate to open upon formation of the MMOH^red:MMOB complex (Figure 6). This type of dynamic reorganization illustrates the importance of searching for tunnels in the form of the enzyme most germane to substrate binding. It seems likely that just as the cluster Fe^III-Fe^III to Fe^II-Fe^II conversion initiates a conformational change to allow O₂ binding, the conversion to the Fe^IV-Fe^IV state in Q may cause a further change to allow methane binding at the right time in the reaction cycle.

Figure 6.

The catalytic cycle of sMMO is depicted through changes at the diiron cluster of MMOH. MMOB remains bound to MMOH in the steps between MMOH^red:MMOB (H^red:B) and the product complex T. Methanol release from T may require MMOB dissociation to open the “pore”. Subsequent binding of O₂ to begin the next turnover requires MMOR-mediated reduction of MMOH and MMOB binding. The MMOB-mediated closing of the pore and opening of the W308-tunnel are depicted in three structures at the top. The interior voids determined using PyMol within MMOH are shown as grey surfaces while the pore (magenta) and W308-tunnel (blue) determined using MOLE 2.5 are depicted by interconnected spheres. Gating residues are shown as sticks, and shifts in these residues occurring during transition to the next intermediate are indicated by orange arrows.

The crystal structures also depict the closure of a short polar tunnel called the “pore” into the MMOH active site upon diiron cluster reduction and MMOB binding (Figure 6, top). The reorganization of a strictly conserved Glu240 gate residue, and occlusion by MMOB residues at the interface, block this tunnel.^2,19 It is interesting to note that the conformational change that closes the pore is coordinated with the changes that open the adjacent W308-tunnel.^1,2 As the only polar route in and out of the active site cavity, the pore may enable methanol exit at the end of turnover. Accordingly, the pore region is highly conserved in the diiron oxygen-activating enzymes.²² In di-manganese form of E. coli class 1b ribonucleotide reductase NrdF, the analogous hydrophilic tunnel is proposed to perform the key role of transferring either superoxide or hydrogen peroxide to the metal site for activation.⁵⁴ In contrast, in the di-iron form of the same enzyme a different hydrophobic tunnel is propose to transport O₂ needed for the distinctly different activation process.

Another example of temporal control of the reaction effected by a small molecule tunnel is found in carbon monoxide dehydrogenase (CODH)/Acetyl-CoA synthase (ACS). CODH/ACS is a bifunctional enzyme that catalyzes CO₂ reduction to CO in the CODH protein and subsequently uses the CO to generate acetyl-CoA at the ACS protein.^46,48 A 70 Å-long tunnel that serves to channel CO between the active sites of the CODH and ACS proteins is observed in crystal structures of the multiprotein complex. The reactions occurring at the two active sites are synchronized as observed by the enhancement of both CO₂ reduction and acetyl-CoA synthesis by the binding of substrates to the partner active sites.^62,63 This phenomenon ensures a high coupling efficiency between the two reactions. The synchronization implies a temporal control of CO production/delivery to the ACS protein only when the cofactors required for acetyl-CoA synthesis are present in the ACS active site. Although a gate within the CO tunnel with open and closed states has been identified, its functional role in the temporal regulation of CO delivery to the ACS subunit has not yet been fully deciphered (Figure 7).^47,64,65 In addition to the CO tunnel, another tunnel has been observed to form as a result of protein dynamics revealed in MD simulations that connects the buried CODH active site to the protein surface.⁴⁵ This tunnel is proposed to serve as the entry route for CO₂. The MD simulation study, coupled with density functional calculations of CO₂ reduction at the C-cluster of the CODH active site, led to a proposed role for a gate in the CO₂-tunnel, which engenders directional control on CO₂ and CO diffusion. This gate is open at the start of the catalytic cycle, enabling CO₂ access to the C-cluster, while it is closed at the end of the CO₂ reduction reaction. This conformational change prevents CO from leaking out of the protein through the CO₂ tunnel and instead, directs it towards the CO-tunnel for diffusion towards the ACS subunit.

Figure 7.

The small molecule tunnel (interconnected blue spheres) trafficking CO from CODH to the ACS protein in the ACS closed state is observed to be blocked by a gating helix (red helix) in the ACS open state of the CODH/ACS complex crystal structure.

TUNNELS ENABLE SPATIAL CONTROL UPON SUBSTRATE DELIVERY

The previous sections have illustrated how small molecule tunnels select for native substrates and deliver them to the active site metallocenter in a temporally regulated manner. However, the role of tunnels does not end at merely binding substrates. Tunnels also direct substrate delivery to specific sites adjacent to the active site metallocenter. This functionality plays a critical role in catalysis as illustrated by the following examples.

The key role played by quantum hydrogen atom tunneling in the acceleration of the oxygenation of methane by Q implies strict donor/acceptor positioning.^55,56 Thus, methane must be delivered to a highly specific position within the active site cavity. The structural studies showed that this positioning can be achieved through MMOH:MMOB complex formation and alignment of the W308-tunnel. The active site cavity adjacent to one face of the diiron cluster is reduced in volume upon the binding of MMOB to MMOH through reorganization of three residues that line the cavity.¹ In addition, the W308-tunnel reorganized by MMOB binding opens into the active site close to the equatorial plane of the diiron cluster where the reactive oxygen species is generated in Q (Figure 8). Once bound, the non-polar methane molecule will remain close to this terminal position of the W308-tunnel due to the constrained volume of the active site and as a means to avoid the polar carboxylate moieties of the diiron cluster glutamate ligands. This position would put the methane molecule in a precise orientation relative to the reactive bridging Fe(IV)-μ-oxo or terminal Fe(IV)=O moiety of Q.⁵

Figure 8.

Modeled path of CH₄ and O₂ from the W308-tunnel into the active site (cyan). The structure of the MMOH^red:MMOB complex is shown in the model and CH₄ is computationally added. O₂ would bind and displace HOH1 to begin the reaction cycle. After formation of intermediate Q, a reactive oxygen species is proposed to be present in the position of HOH1. CH₄ would enter the active site at this stage and react by quantum tunneling of a hydrogen atom.

Analogous control of substrate spatial positioning by a small molecule tunnel is seen in the lipoxygenase family. Lipoxygenases are a family of non-heme mono-iron oxygenases that catalyze the regio- and stereo-specific hydroperoxidation of poly-unsaturated fatty acid substrates containing a pentadiene moiety.^35,36 The iron(III)-hydroxide active site metallocenter initiates the catalytic cycle by abstracting a hydrogen atom from a bis-allylic methylene carbon of the fatty acid substrate.³⁸ The resulting carbon radical is delocalized over the adjacent double bonds. O₂ subsequently attacks the carbon radical in a stereospecific manner at a position that is either +2 or −2 from the site of the original carbon radical. It is thus remarkable that these enzymes furnish highly regiospecific peroxidation products, given that three of the five positions of the delocalized radical have identical activation energy barriers to product formation. Two models have been shown to account for this regiospecificity, one of which involves a primary role for the small molecule tunnel delivering oxygen to the active site.³⁶ This model was explored for the soybean lipoxygenase-1 via mutations at residues that are present at the bottleneck of a putative O₂ tunnel. Two site-directed mutants with bulky side-chains specifically impacted the step of O₂ binding as measured by increased K_m values over the wild-type enzyme.³⁶ Additionally, there was a loss of regiospecificity of fatty acid peroxidation as judged by the increased product yield from oxidation at the alternative delocalized radical carbon positions. These observations were explained for one of the enzyme mutants as arising from a severely reduced rate of intramolecular oxygen diffusion that resulted in peroxidation occurring in bulk solution after dissociation. More interestingly, the altered regiospecificity with the other enzyme variant was proposed to originate from the opening of an alternative tunnel into the active site upon the occlusion of the original tunnel by the mutation. Thus, the specific path of the gas tunnel in lipoxygenases directs the O₂ substrate to one of the three delocalized radical carbon positions, resulting in a regiospecific product.^35,36

This model has been further supported by studies of how another motif, termed the Gly-Ala motif in lipoxygenases, engenders regiospecific product formation. The comparison of various lipoxygenases with altered regiospecificities has shown that the product outcome is dependent upon whether the key residue is glycine or alanine (420 in Figure 9). X-ray crystallographic, site-directed mutation and MD simulation studies probing the role of this residue have shown that the regiospecificity is altered because the point of O₂ delivery adjacent to the radical fatty acid intermediate is modulated by the mutations.^37,38

Figure 9.

The small molecule tunnel for O₂ delivery in the lipoxygenase enzyme from Pseudomonas aeruginosa is shown as a cyan surface (panel A). Panel B depicts the CAVER calculated tunnel (cyan spheres) threading its way through the interior voids (transparent grey surface) towards the C15 position of the modeled arachidonic acid substrate (pink sticks). The A420G mutation opens up an alternative O₂ route (magenta spheres) to the C11 position of arachidonic acid (panel C).³⁷

CONCLUSION

The importance of tunnels in the functions of metalloenzymes that catalyze reactions involving small molecules can be gauged from the abundance of these structural features (Figure 2). The tunnels play central roles in the substrate binding process, but perhaps more importantly, also in the substrate selection process. The tunnels adopt the common principles of selection based on polarity and electrostatics encountered in substrate binding in all enzyme active sites. However, the finely tuned radius of the small tunnels adds a new selection parameter based on the minimum substrate diameter. Moreover, the flexibility of the tunnels in response to factors such as regulatory protein binding or metal redox state, as well as the introduction of molecular gates, allow them to synchronize the introduction of substrates with the progress of the reaction cycle. It is also possible that the ability to dynamically constrict and expand allows the same tunnel in some cases to select more than one type of substrate as required during the catalytic cycle. Consideration of just the examples presented here reveals that enzyme tunnels can readily select between chemically similar substrates that differ by less than 1 Å in effective diameter. The impact of small molecule tunnels on catalysis goes far beyond the common ability of enzymes to select a substrate by binding in the active site. Indeed, an enzyme such as sMMO would not be able to fulfill its function in the absence of a highly specific tunnel. An open site constructed of hydrophobic residues would certainly bind methane, but it would also bind many other hydrophobic molecules, all of which would be far easier to oxidize. The W308-tunnel appears to allow sMMO to generate Q in a completely closed and protected environment before exposure to the only substrate the methanotroph can utilize for growth. It delivers the right substrate to the right location at the right time for specific oxidation of the strongest aliphatic hydrocarbon bond. Given the critical role played by the W308- and similar tunnels in metalloenzymes, we propose that they are best described as extensions of the active site. One implication to be drawn from this view is that in many cases the “buried active site” of an enzyme is not truly buried. Rather, it samples the surface via a highly evolved extension essential for its function.

Biographies

John D. Lipscomb was born in Wilmington, Delaware in 1947. The earned a B.A. from Amherst College in 1969. He received his Ph.D. in Biochemistry in 1974 from the University of Illinois, Champaign-Urbana for studies under the guidance of I. C. Gunsalus. After postdoctoral study at the Gray Freshwater Biological Institute with John Wood, he began his independent career at the University of Minnesota, Twin Cities in 1977. He is currently Professor in the Department of Biochemistry, Molecular Biology and Biophysics. His studies are focused on the structure, regulation and mechanism of iron-containing enzymes that catalyze aromatic ring and hydrocarbon oxygenation chemistry.

Rahul Banerjee was born in 1982 in India. He earned a B.Sc. in Chemistry from St. Stephen’s College, University of Delhi, New Delhi, India in 2003. He continued his studies for a M.Sc. in Biotechnology at the Indian Institute of Technology, Bombay, India under the guidance of Narayan S. Punekar. In 2013, he was awarded a Ph.D. from the University of Minnesota, Twin Cities for his studies of the structure and mechanistic intermediates of soluble methane monooxygenase (sMMO) with John D. Lipscomb. He is currently a research associate in Prof. Lipscomb’s laboratory carrying out biophysical studies of sMMO.