Crystal Structure of Methyl-Coenzyme M Reductase: The Key Enzyme of Biological Methane Formation

See allHide authors and affiliations

Science  21 Nov 1997:
Vol. 278, Issue 5342, pp. 1457-1462
DOI: 10.1126/science.278.5342.1457


Methyl–coenzyme M reductase (MCR), the enzyme responsible for the microbial formation of methane, is a 300-kilodalton protein organized as a hexamer in an α2β2γ2 arrangement. The crystal structure of the enzyme from Methanobacterium thermoautotrophicum, determined at 1.45 angstrom resolution for the inactive enzyme state MCRox1-silent, reveals that two molecules of the nickel porphinoid coenzyme F430 are embedded between the subunits α, α′, β, and γ and α′, α, β′, and γ′, forming two identical active sites. Each site is accessible for the substrate methyl–coenzyme M through a narrow channel locked after binding of the second substrate coenzyme B. Together with a second structurally characterized enzyme state (MCRsilent) containing the heterodisulfide of coenzymes M and B, a reaction mechanism is proposed that uses a radical intermediate and a nickel organic compound.

Methyl–coenzyme M reductase is the key enzyme of methane formation in methanogenic Archaea. It catalyzes the reduction of methyl–coenzyme M (methyl-CoM) [CH3-S-CoM, 2-(methylthio)ethanesulfonate] with coenzyme B (CoB) (CoB-S-H, 7-thioheptanoyl-threoninephosphate) to methane and the heterodisulfide of CoM (CoM-S-H, 2-thioethane sulfonate) and CoB under strictly anaerobic conditions (1, 2).

About 109 tons of CH4 are produced per year by the reaction in Scheme 1. Part of it escapes to the atmosphere and acts as a potent greenhouse gas (3). Methyl-CoM reductase was first characterized by Ellefson and Wolfe (4) as a yellow protein of an apparent molecular mass of 300 kD composed of three different subunits arranged in an α2β2γ2 configuration. The hexameric protein contains two molecules of the tightly but not covalently bound coenzyme F430 (4), which is a Ni porphinoid (5).

Spectroscopic investigations of methyl-CoM reductase have revealed several Ni electron paramagnetic resonance (EPR) active and inactive states of the enzyme (6). After harvest of H2-CO2 grown cells, the enzyme is present in an inactive EPR silent state designated as MCRsilent. In this state, methyl-CoM reductase contains bound CoM (7) and CoB (8) and can only be partially reactivated by enzymatic reduction (9). When cells are gassed with H2before harvesting, the enzyme is present in an active MCRred1 state whose characteristic Ni(I) F430 EPR spectrum, designated red 1, can be correlated with the enzymatic activity in the enzyme (10). Even under strictly anaerobic conditions, the activity of the enzyme is completely lost within a few hours, and the enzyme enters an inactive EPR-silent Ni(II) state denoted as MCRred1-silent. When cells are gassed with CO2-N2 before being harvested, the enzyme enters into the MCRox1 state, which exhibits a Ni EPR spectrum, designated ox1, substantially different from that of the MCRred1 state. The MCRox1state has only very low activity but can be activated in vitro by reduction with Ti(III) citrate (11) into the MCRred1 state. Preparations in the MCRox1 state slowly turn into an inactive EPR silent state, referred to as MCRox1-silent.

Methyl-CoM reductase (isoenzyme I) was aerobically crystallized in the enzymatically inactive enzyme states MCRox1-silent and MCRsilent, as described by Shima et al. (12). The crystal structure was solved in the MCRox1-silent state at 2.9 Å resolution by the method of multiple isomorphous replacement (Table1) and refined to currentR cryst and R free values (Table 1) of 19.5 and 22.0%, respectively, in the resolution range 1.45 to 10.0 Å. The structure of the MCRsilent state was refined to current R cryst andR free values of 18.0 and 22.4%, respectively, in the resolution range 2.0 to 10.0 Å with the use of the MCRox1-silent structure for initial phase determination. The model of the MCRox1-silent state is mainly distinguished from that of MCRsilent by binding of CoM and CoB in the reduced form instead of the oxidized heterodisulfide form of CoM and CoB. The two enzyme states exhibit a nearly identical overall structure. In both structures, the electron density map reveals five modified amino acids located in subunits α and α′ at or very near the active site region. The observed 1-N-methyl-Hisα 257, 4-methyl-Argα 271, 2-methyl-Glnα 400, andS-methyl-Cysα 452, as well as Glyα 445, where the carbonyl oxygen appears to be substituted by sulfur, have to be confirmed by biochemical analysis.

Table 1

X-ray structure determination. Crystals of methyl-CoM reductase from M. thermoautotrophicum grown with 25% polyethylene glycol 400 as precipitant have the space group P21 and unit cell dimensions of a = 83.1 Å, b = 120.2 Å,c = 123.1 Å, and β = 91.70 with one hexamer per asymmetric unit (12). Native data (Nat1, Nat2, and Nat3) and eight heavy-atom derivative data sets [Hg1, thimerosal; Hg2, mersalylic acid; Hg3, C2H5HgPO4; Hg4,CH3HgCl; Pt1, K2PtCl4; Pt2, di-μ-iodobis(ethylenediamine)-di-platinum nitrate; Pt3, K2PtCl2(NH2)2; and Au, KAuCl4] were collected with a Rigaku rotating anode x-ray generator producing CuKα radiation and an image plate detector (Mar-Research, Hamburg, Germany). Reflections were processed with MOSFLM (30), DENZO (31), and the CCP4 program suite (32). The Patterson map of Hg1 was interpreted with SHELXS (33). Heavy-atom positions were refined and phases were computed with MLPHARE (34). Phases were improved by solvent flattening with DM (35) and molecular averaging with RAVE (36). The quality of the resulting electron density maps to a resolution of 2.9 Å was sufficient to incorporate nearly the complete polypeptide model with the known primary structure (37). Model building was done within the program O (38). The coordinates were refined by means of the simulated annealing and positional refinement protocol of X-PLOR (39) with data set Nat2 collected at the Max-Planck beam line at the Deutsches Elektronensynchrotron (Hamburg) for the MCRox1-silent state and data set Nat3 measured in house for the MCRsilent state. After refinement, the root-mean-square deviation from ideal stereochemical parameters is 0.009 Å for bond lengths and 1.30° for bond angles. The average temperature factor is 9 Å2. The model of the MCRox1-silent state comprises residues α3 to α549, β2 to β443, γ2 to γ248, α′3 to α′549, β′2 to β′443, and γ′2 to γ′248; two sets of coenzymes F430, B, and M; one Zn2+ molecule; two Mg2+ molecules; and 1320 water molecules.

View this table:

The overall structure of methyl-CoM reductase is characterized by a series of α helices arranged in a compact form with an ellipsoidal shape (Fig. 1) of about 120 by 85 by 80 Å. The subunits are mutually tightly associated, as indicated by extended interface areas, particularly between subunits α and α′ and subunits β and β′, and by the fact that, except for subunits γ and γ′, each subunit contacts all other subunits of the multisubunit complex (Fig. 1). The fold of the subunits is described in the legend of Fig. 1.

Figure 1

Structure of methyl-CoM reductase. Ribbon diagram of the heterohexamer. Secondary structure segments were defined according to DSSP (40). Subunits α and α′ are depicted in red and orange, subunits β and β′ in dark green and light green, and subunits γ and γ′ in dark blue and light blue. Subunit α can be subdivided into an NH2-terminal region (α3 to α101), an α + β domain (α102 to α276), an α helix domain (α277 to α506), and a COOH-terminal region (α506 to α549). The α + β domain consists of a four-stranded antiparallel β sheet flanked by extended helical regions. This fold is reminiscent of the α,β sandwich motif of the βαββαβ class (41) found, for example, in a formyltransferase (42), which is another protein of the methanogenic pathway. The α-helix domain of subunit α is composed of eight α helices of variable length, building up a helix sandwich structure in a way similar (43) to that observed for the transmembrane domain of diphtheria toxin (44). The general architecture of subunit β and the fold of the α + β domain and the α-helix domain resemble those of subunit α, suggesting a common ancestor. Subunit γ is primarily built up by an α,β sandwich folding unit (γ81 to γ160) that corresponds again to the subclass of doubly intertwined βαββαβ folding motifs. As observed for subunits α and β, subunit γ has also extended NH2-terminal (γ2 to γ80) and COOH-terminal (γ202 to γ248) regions serving as contact arms to other subunits. The positions of the coenzymes F430 (displayed in yellow) indicate the approximate location of the two active sites. Figures 1 and 4 were produced with MOLSCRIPT (45) and RASTER3D (46).

The binding sites of two sets of coenzymes F430, M, and B are roughly 50 Å apart (Fig. 1), forming two separated structurally identical active sites. The MCRox1-silent and MCRsilent structures reveal that coenzymes F430, M, and B and the heterodisulfide CoM-S-S-CoB, respectively, are embedded inside a narrow channel, 30 Å long, extending from the protein surface deeply into the interior of the protein complex (Fig. 2, A and B). The channel and the coenzyme binding sites are formed by residues of subunits α, α′, β, and γ (and equivalently α′, α, β′, and γ′), indicating that one trimer is not sufficient to accomplish the enzymatic reaction. The overall structure looks as if the protein has been developed from an α2β2 tetramer with vestigial channels at the appropriate interfaces.

Figure 2

(A) Molecular surface representation of methyl-CoM reductase that shows the entrance of one of the channels (indicated by a white arrow). The two channels are formed by subunits α, α′, β, and γ and α′, α, β′, and γ′, respectively. The colors of the subunits are as in Fig. 1. (B) Molecular surface of methyl-CoM reductase without subunit α′ allows a view into the interior of the channels, showing the binding place of the coenzymes. Coenzymes F430 are yellow, and the heterodisulfides of coenzymes M and B are white. The entrances of both channels are indicated by a white arrow. This figure was generated with GRASP (47).

Coenzyme F430 fits neatly into a pocket at the apex of the channel inside the protein (13). Its tetrapyrrole plane is oriented such that its front face (reduced pyrrole rings A, B, C, and D clockwise, Fig. 3) points roughly toward the mouth of the channel, whereas its rear face points to the channel apex.

Figure 3

Structure of coenzyme F430 (viewed toward the front face) and the final 2|F obs| − |F calc| electron density at 1.45 Å resolution contoured at the 2σ level. The quality of the electron density is sufficiently high to determine the exact configuration of the coenzyme that is in agreement with previous results (5). The propionate substituents of rings A, B, and C are perpendicular to the tetrapyrrole plane pointing toward the apex, whereas the lactam ring is directed toward the mouth of the channel (Fig. 2B). The six-membered carbocyclic ring joined with ring D, and the protruding acetate and acetamide substituents lie approximately in the tetrapyrrole ring plane. Figures 3 and 5 were produced with SETOR (48).

Because of the excellent quality of the electron density map (Fig. 3), the structure of coenzyme F430 could be confirmed, particularly the absolute configuration of the pyrrole substituents derived previously by nuclear magnetic resonance, circular dichroism, and x-ray studies (5). The tetrapyrrole ring of coenzyme F430 is bound in a rather flat conformation to the enzyme, as predicted for the free coenzyme F430 in the hexagonally coordinated Ni(II) state (5).

When analyzing the noncovalent interactions between the Ni porphinoid and the polypeptide chain, two features are particularly striking. First, the five negative charges of the deeply buried F430 carboxylate groups are neutralized by ionic interactions only to a minor extent, but they are delocalized by a series of hydrogen bonds. The presence of the carboxylate groups in the charged state is likely because of their exclusive contacts with hydrogen donors. Second, coenzyme F430 not only has a large number of interactions, but it is also apparently rigidly bound to the protein. Fifteen of the 21 hydrogen bonds between coenzyme and protein are formed to peptide amide nitrogens.

The Ni atom present as Ni(II) sits almost exactly in the tetrapyrrole plane and is coordinated to six ligands arranged in a nearly optimal octahedral configuration (Fig.4, A and B). The four equatorially located nitrogen atoms of the tetrapyrrole ring have distances to the nickel of 2.14 Å for ring A, 2.11 Å for ring B, 2.10 Å for ring C, and 1.99 Å for ring D, compared with 2.09 Å as derived from x-ray absorption spectroscopic studies (14). As the fifth ligand, the side chain oxygen of Glnα 147 protrudes from a long loop between helix 5 and strand D of the α + β domain to the rear face of F430 and approaches the Ni atom to 2.3 Å. The 1.45 Å electron density map demonstrates that the oxygen and not the nitrogen of the side chain of Glnα 147is the axial ligand. The sixth coordination site of nickel located in front of the tetrapyrrole ring plane of F430 is occupied by the thiol group of CoM in the MCRox1-silent structure and by a sulfonate oxygen of the heterodisulfide in the MCRsilent structure (Fig. 4, A and B). The distance between nickel and sulfur is 2.4 Å; between nickel and oxygen, it is 2.1 Å.

Figure 4

(A) The active site region of the MCRox1-silent structure. The binding positions of the coenzymes suggest the active site between the nickel of coenzyme F430 and the sulfur atom of CoB. The active site is coated mostly by nonpolar and aromatic residues. Five mutually contacting phenylalanine and tyrosine side chains are arranged as ring forming a tunnel. (B) The active site region of the MCRsilent structure. Compared with the MCRox1-silent structure, CoM has moved through the tunnel to form with CoB a heterodisulfide, the oxidation product of the reaction. The sulfonate moiety of CoM lost its interactions to the protein matrix and is coordinated to the Ni atom.

The analysis of the MCRox1-silent structure reveals that CoM is positioned almost parallel to the tetrapyrrole plane in contact with its front face (Fig. 4A). The thiol group binds axially to the nickel and interacts with the hydroxyl groups of Tyrα 333 and Tyrβ 367and a water molecule that bridges CoM and CoB. The ethyl moiety is embedded between the lactam ring of the Ni porphinoid and the phenyl ring of Pheα 443. Coenzyme M is anchored to the polypeptide chain by its negatively charged sulfonate group, forming a salt bridge to the guanidinium group of Argγ 120, a hydrogen bond to the peptide nitrogen of Tyrα 444, and a hydrogen bond to a water molecule connected to the peptide oxygen of Hisβ 364 (Fig. 4A).

With its elongated conformation, CoB fits accurately into the most narrow segment of the channel formed by residues of subunits α, α′, and β (Fig. 2). Coenzyme B is anchored to the protein mainly by salt bridges between the negatively charged threoninephosphate moiety and five positively charged amino acids (Fig.5). The heptanoyl arm is in van der Waals contact with several hydrophobic residues. The thiol group of CoB interacts with the side chain nitrogen of Asnα 481, the main chain peptide nitrogen of Valα 482, and the bridging water molecule mentioned above (Fig. 5). Asnα 481 is within hydrogen bond distance of the sulfur that is presumed to replace the backbone carbonyl oxygen of the modified Glyα 445. A functional role of this modification is therefore possible.

Figure 5

The CoB binding in the MCRox1-silent structure. Coenzyme B binds in an elongated conformation into the most narrow segment of the channel built up of residues of subunits α (red), α′ (orange), and β (green). The enzymatically relevant thiol group is directed to the apex, whereas the threoninephosphate moiety points toward the mouth of the channel. Only the phosphate and the carboxylate groups are accessible to bulk solvent. The threoninephosphate moiety is strongly linked to the protein mainly by salt bridges between both the phosphate and carboxylate group and five basic amino acid residues. Argα 270, Hisβ 379, Lysα ′256, and methyl-Hisα ′257 interact with the phosphate group and Argα ′225 and Lysα ′256 with the carboxylate group. Argα 271 neighboring Argα 270 is methylated.

Although CoM and CoB bind separately in the MCRox1-silentstate (Fig. 4A), they are covalently linked as a heterodisulfide in the MCRsilent state (Fig. 4B). A superposition of the structures reveals that the reduced CoB and the CoB moiety of the heterodisulfide align perfectly except that the sulfur is turned slightly toward CoM. In contrast to CoB, CoM has moved more than 4 Å away from its position in the MCRox1-silent state. The thiol group is shifted perpendicular and the sulfonate group parallel to the tetrapyrrole plane of F430, resulting in a 90° rotation of CoM. In this position, one oxygen atom of the sulfonate is axially coordinated with the nickel and contacts the hydroxyl group of Tyrα 333. The second oxygen atom is hydrogen-bonded to the lactam ring of F430 and to the hydroxyl group of Tyrβ 367 and the third to a water molecules located at the former binding site of the sulfonate (Fig. 4B).

Both the MCRox1-silent and the MCRsilent structures display inactive states of methyl-CoM reductase with coenzyme F430 bound in the Ni(II) oxidation state. Nevertheless, the arrangement of the coenzymes and their protein environment, combined with the available biochemical and spectroscopical data, allow for conclusions about the active site and a proposal for the catalytic mechanism outlined in Fig.6.

Figure 6

Cartoon 1 showing the proposed steps in methane formation from methyl-CoM and CoB. The reaction cycle starts (step 1) with a nucleophilic attack of Ni(I) on the methyl group of CoM, forming a methyl-Ni(III) organic compound. Model reactions indicate that the methyl transfer reaction is facilitated when the leaving group is protonated (19). Perhaps the proton originates from CoB. In step 2, the strongly oxidative Ni(III) (19) withdraws an electron from the protonated CoM, generating a CoM thiyl radical that is a strong acid and therefore dissociates. In step 3, the methyl-Ni(II) generated in step 2 is protonolyzed in a spontaneous reaction (19,49). Almost simultaneously, the CoM thiyl radical couples with the thiolate group of CoB, and the surplus electron of the generated disulfide anion radical is returned to Ni(II). In step 4, methane and CoM-S-S-CoB are released, and the next cycle can be started after renewed binding of methyl-CoM and CoB. The two Ni(I) intermediates (lower and upper left of the figure) are related to the MCRred1 state of the enzyme described in the text. The MCRsilent state is probably derived from the MCRred1 state (lower left) by one electron oxidation. The MCRox1 state could be formed by reaction of the CoM thiyl radical with Ni(II) after protonolysis of the methyl-Ni(II) intermediate, yielding a Ni(III) thiolate adduct. MCRox1-silent would then be derived from MCRox1by one electron reduction.

Scheme 1

First of all, the two structurally identical active sites separated from each other by roughly 50 Å (Fig. 1) appear to exclude all mechanisms requiring two molecules of coenzyme F430 in methane formation from methyl-CoM and CoB.

The relative arrangement of the three coenzymes (Fig. 4) suggests that the catalytic reaction takes place at the front side of coenzyme F430 in the channel between the nickel and the thiol groups of CoB (Figs. 2 and 4). Each active site is lined up by an annular arrangement of Pheα 330, Tyrα 333, Pheα 443, Pheβ 361, and Tyrβ 367 flanked by further hydrophobic and aromatic residues (Fig. 4). These amino acids are completely conserved in all methyl-CoM reductases (15). The active site region is accessible only through one channel and only for small molecules up to a diameter of 6 Å (16). This channel is completely locked when CoB is bound (17), shielding the active site from bulk solvent (Figs. 2 and 5). The sole water molecule found in the active site region of MCRox1-silent between CoM and CoB should be displaced after the binding of the more bulky methyl-CoM. Methane formation from methyl-CoM and CoB thus takes place in a hydrophobic aromatic environment and probably does not involve water molecules. These conditions would allow for radical intermediates, which have been proposed (18-20) and accounted for in the proposed reaction mechanism outlined in Fig. 6. Solvent-inaccessible active sites coated by nonpolar aromatic residues and attainable by a channel have also been observed in several radical-based enzymes, such as galactose oxidase (21), prostaglandin-H2 synthase–1 (22), and methylmalonyl–coenzyme A mutase (23).

During the catalytic reaction of methane formation, CoB releases and the activated methyl group accepts a hydrogen atom. The MCRox1-silent structure reveals an interaction between the thiol group of CoB and two hydrogen donors, the amide and peptide nitrogens of Asnα 481 and Valα 482 (Fig. 5), which would facilitate the cleavage of a proton and permit the presence of a thiolate anion (Fig.6). However, the CoB sulfur is 6 Å from the tentatively modeled Ni-CH3, which is probably too far away for direct hydrogen transfer. Therefore, the participation of CoM as hydrogen mediator between CoB and the activated methyl group (Fig. 6), perhaps through Tyrα 333 or Tyrβ 367(Fig. 4), is an attractive possibility. The distances of their phenolate oxygens to the nickel are 4.4 and 4.3 Å and to the modeled methyl group are around 3.1 and 3.3 Å, respectively.

In the MCRox1-silent structure (Fig. 4A), CoM probably mimics the binding position of methyl-CoM with respect to the binding mode of the sulfonate moiety but presumably not with respect to the binding mode of the thiol group. A Ni-S-CoM intermediate in the catalytic cycle is not attractive because of the large distance of 6.2 Å between the sulfurs of CoM and CoB. Model building studies indicate, however, that the two sulfurs of CoM and CoB come in van der Waals contact when the methyl group of methyl-CoM is placed in van der Waals distance of the potentially attacking nickel. Therefore, a Ni-CH3 intermediate proposed from free coenzyme F430 studies (19, 24, 25) appears to be compatible with the steric requirements of the active site (see Fig. 6). The proximity of the sulfurs of CoM and CoB in the model allows for heterodisulfide formation that could induce a shift of CoM toward CoB and thus loosen the interactions between the sulfonate moiety of CoM and the protein matrix.

In the structure of MCRsilent (Fig. 4B), the bound heterodisulfide CoM-S-S-CoB cannot leave the enzyme because a sulfonate oxygen of CoM binds to the Ni(II) atom of F430. However, both protein and coenzyme conformations might be very close to an intermediate of the enzymatic reaction after product formation (Fig. 6, lower left), assuming that a coordination of the sulfonate oxygen to nickel is prevented when, under reducing conditions, nickel is present as nucleophilic Ni(I). The repulsion between Ni(I) and the sulfonate oxygen molecules might even be used as a driving force to push the heterodisulfide out of the channel. It is not evident at present where and how the methane gets out.


View Abstract

Stay Connected to Science

Navigate This Article