Structural basis for organohalide respiration

See allHide authors and affiliations

Science  24 Oct 2014:
Vol. 346, Issue 6208, pp. 455-458
DOI: 10.1126/science.1258118

How bacteria break down organohalides

Anaerobic bacteria can break down a range of organohalide pollutants. To do so, they use unusual reductive dehalogenase enzymes that remove the halogen ion from the molecule, making the pollutants less toxic. Bommer et al. describe x-ray crystal structures of one such enzyme from Sulfurospirillum multivorans (see the Perspective by Edwards). Vitamin B12, present near the substrate binding site, catalyzes the reduction of trichloroethylene in concert with two iron-sulfur clusters. The structures provide mechanistic clues for how to engineer enzymes to recognize other pollutants.

Science, this issue p. 455; see also p. 424


Organohalide-respiring microorganisms can use a variety of persistent pollutants, including trichloroethene (TCE), as terminal electron acceptors. The final two-electron transfer step in organohalide respiration is catalyzed by reductive dehalogenases. Here we report the x-ray crystal structure of PceA, an archetypal dehalogenase from Sulfurospirillum multivorans, as well as structures of PceA in complex with TCE and product analogs. The active site harbors a deeply buried norpseudo-B12 cofactor within a nitroreductase fold, also found in a mammalian B12 chaperone. The structures of PceA reveal how a cobalamin supports a reductive haloelimination exploiting a conserved B12-binding scaffold capped by a highly variable substrate-capturing region.

Anaerobic microorganisms use alternative terminal electron acceptors during respiration, such as nitrate, sulfate, iron(III), or even organohalides. The accumulation of polluting organohalides such as perchloroethylene (PCE, also tetrachloroethene) and trichloroethene (TCE) in the environment, which have been heavily used for dry cleaning and degreasing, is problematic because of their toxicity; however, microbial organohalide respiration can transform these compounds into less toxic forms. Organohalide respiration requires reductive dehalogenases (RDases) for the central reduction step (1). In contrast to terminal reductases that contain prosthetic heme groups, molybdopterin, or flavins as cofactors, RDases harbor a corrinoid cofactor and two Fe/S clusters (2). RDases are able to convert some of the most noxious environmental pollutants, including halogenated phenols, dioxins, biphenyls, and aliphatic hydrocarbons. RDase genes were identified in distantly related bacterial genera belonging to the chloroflexi; firmicutes; gamma-, delta-, and epsilonproteobacteria; and even archaea (3).

Several hundred RDase gene sequences deposited in databases await testing for functionality and determination of the substrate spectrum. Low growth yields and the oxygen sensitivity of the RDases have hindered large-scale purification and biochemical characterization of RDases (1). Genetic manipulation of the bacterial isolates has thus far been difficult, and only recently was functional heterologous production of RDases reported, but it remains challenging (4). Structural data will help to resolve how RDases evolved, function, and specifically select the many different substrates, some of which have been present in the biosphere for less than a century.

Here we report the crystal structure of a reductive dehalogenase, PceA (5), of the microaerophilic epsilonproteobacterium Sulfurospirillum multivorans (formerly Dehalospirillum multivorans) (6) (i) in an empty state; (ii) in the presence of TCE; (iii) in the presence of the cis-dichloroethene (cis-DCE) (product) analog cis-dibromoethene (cis-DBE); and (iv) in the presence of iodide, a substitute for the leaving chloride, at a maximum resolution of 1.6 Å (7). S. multivorans was isolated in the mid-1990s from activated sludge and possesses a flexible catabolism integrating numerous terminal reductases encoded in its genome (8). S. multivorans is able to couple the reductive dechlorination of PCE, TCE, or dibromoethene (DBE) to growth (9, 10) through its prototypical RDase PceA.

The 464 amino acids of PceA are structured in a compact α/β fold domain (Fig. 1A). The structure can be divided into an N-terminal unit (residues 1 to 138), a norpseudo-B12 binding core (residues 139 to 163 and 216 to 323), an insertion unit (residues 164 to 215), an iron-sulfur cluster binding unit (residues 324 to 394) and a C-terminal unit (residues 395 to 464) (fig. S1). Two protomers in the P41 asymmetric unit interact tightly to form a dimer with a twofold noncrystallographic symmetry. Two α helices of the norpseudo-B12 binding core and one α helix of the N-terminal unit form a helical bundle with their symmetry mates. Along with extensive loop regions in the N- and C-terminal units, these bundles create the dimer interface. The interface covers 20% of the accessible surface area of the protomer, supporting a compact and stable dimeric arrangement. Using gel filtration, an apparent molecular mass of 89 kD (fig. S2) was determined for PceA purified from the membrane fraction, which agrees with a dimeric rather than a monomeric structure as reported previously for the soluble wild-type enzyme (5).

Fig. 1 Crystal structure of PceA.

(A and B) Overall structure of dimeric PceA. One protomer is shown in gray. The other protomer is colored as a conserved region, which binds the cofactors (green, residues 139 to 163 and 216 to 462), and a variable region, which contributes most of the substrate binding and selection function (blue, residues 1 to 138 and 164 to 215). Each PceA protomer harbors two cubane-type [4Fe-4S] clusters (proximal and distal, spheres) and norpseudo-B12 (purple sticks). Supposed electron transfer pathways are indicated in (B), along with distances between the closest Fe atoms of the [4Fe-4S] clusters and Co of norpseudo-B12. (C) Active-site cavity of PceA. Conserved polar residues in proximity to the axial position of Co are depicted as white sticks, hydrophobic residues of the second ligand sphere as blue sticks. A 12 Å–long channel was identified as the substrate entry site to the active-site dome.

RDases lack obvious sequence similarities to other enzyme families, and the fold of PceA is unlike that of known corrinoid-dependent methyltransferases (11) or mutases (12). The most similar protein with clear homology found was methylmalonic aciduria cblC type with homocysteinuria (MMACHC) (13) (fig. S3). MMACHC is a B12-trafficking chaperone essential for the formation of adenosyl- or methylcobalamin in humans by catalyzing the reductive removal of the upper axial ligands from cyanocobalamin and alkylcobalamins. Structural homology is limited to the B12/norpseudo-B12–binding core, which resembles the nitroreductase family fold (14) (fig. S3). Consequently, RDases and MMACHC most likely evolved from a common ancestral B12-binding protein.

In addition to the norpseudo-B12 cofactor, PceA also harbors two [4Fe-4S] clusters. Short distances between the two [4Fe-4S] clusters and the proximal [4Fe-4S] cluster and the Co bound to the corrin ring are expected to allow for a rapid electron transfer within a protein monomer (15) (Fig. 1B). The proximal [4Fe-4S] cluster is in van der Waals (vdW) contact distance to the C83 carboxamide side chain and C8 of the corrin ring [for atom numbering, see (2)], with the carboxamide N84 being in hydrogen bond distance to a μ3-sulfido ligand of the [4Fe-4S] cluster. The two active sites of the PceA dimer are at a Co-Co distance of 42 Å without cofactors between them, indicating two independent catalytic units per dimer (Fig. 1A).

Both [4Fe-4S] clusters are within 6 Å of the enzyme surface, whereas norpseudo-B12 is deeply buried in the structure. Most interactions between the cobalamin cofactor's ring system and the protein matrix are due to hydrogen bonds between the cofactor and the norpseudo-B12–binding core of PceA (Fig. 2). Binding of the phosphate and parts of the adenine group of norpseudo-B12 is mediated by a loop in the iron-sulfur cluster binding unit. Co is not coordinated by adenine (base off), and the conformation of the linker is curled rather than extended as is typical for base-off B12 (14, 16). Compared to the base-on conformation of isolated norpseudovitamin B12 (2), several conformational changes, including a 90° rotation around the bond between phosphate and the C3′ of ribose and flip of the ribose, are needed to convert the base-on conformation into the base-off conformation found in PceA (Fig. 2). In comparison to the dimethylbenzimidazole moiety of B12, the adenine moiety of norpseudo-B12 has several hydrogen-bond acceptor and donor functions, which in addition to anchoring the cofactor allows solvent accessibility (Fig. 2, inset). C176 of the nucleotide loop packs closely against the adjacent β sheet, and the methyl group found in B12 at this position would force the loop to adopt a different conformation, which might explain the preference for norpseudo-B12.

Fig. 2 Norpseudo-B12 binding.

Potential hydrogen-bonding residues are colored according to their position within PceA: blue, N-terminal (residues 1 to 138); purple, norpseudo-B12–binding core (residues 139 to 163 and 216 to 323); green, insertion unit (residues 164 to 215); yellow, [4Fe-4S]-cluster coordinating loop (residues 324 to 394). The 1σ 2mFo-DFc electron density map for norpseudo-B12 and water molecules is shown as mesh. The inset shows the solvent-accessible adenine moiety. A comparison of corrinoid linker geometries is shown on the right: free norpseudovitamin B12 (2), base on (Cambridge Crystallographic Data Centre: 217274, top), norpseudo-B12 bound to PceA (middle), and base-off B12 in MMACHC (14) (Protein Data Bank: 3SOM, bottom). The β ligand is omitted for clarity.

PceA contains CoII in the as-isolated state (2), for which an axial ligand is typically observed. Additional density is found above the β face of Co, with a Co-X distance of 2.5 Å. The density has been tentatively assigned to a water molecule. Reduction to the catalytically relevant superreduced CoI is probably facilitated by the weak axial ligation, making Co effectively tetracoordinated in the protein, and agrees with the elevated midpoint potential of the CoII/CoI transition of –380 mV (pH 7.5, versus a standard hydrogen electrode) (2).

Substrate access to the active site is restricted by a selection filter with the shape of a 3 × 5.5 Å “letterbox,” a gap made up of side chains from the N-terminal and insertion units (fig. S4). After the filter, PCE and TCE have to pass through a 12 Å–long hydrophobic channel isolating the active site inside the core of the protein (Fig. 1C, arrow). The channel then expands to create the amphiphilic active site pocket at the β face of the corrin ring, where access to the cofactor itself is further restricted by a ring fence of side chains (fig. S4A). The cavity is predominantly lined by tryptophan and tyrosine residues, which define its shape and volume. Three adjacent polar residues, namely Tyr246, Arg305, and Asn272 from the B12-binding core, are highly conserved (Figs. 2 and 3C). Of these Tyr246 is invariant, whereas some RDases have functionally conserved replacements with a lysine residue in place of Arg305 (fig. S5). Asn272 is conserved in a subset of RDases. The residues are close to the corrin ring and point with their side chains into the active site pocket, positioning the phenolic hydroxyl group of Tyr246 in hydrogen-bonding distance to the guanidinium group of Arg305 (Fig. 3C).

Fig. 3 Substrate/product–analog binding.

(A and C) TCE, overlayed in a high- and low-occupancy (50 and 20% as modeled in the structure) orientation, as indicated in the chemical sketch. Electron density maps are shown for TCE, the β ligand, and water: 1σ 2mFo-DFc (blue) and 4σ mFo-DFc (green). (B) Overlay of the active site with TCE, cis-DBE, iodide, and the coordinated water (empty structure). Side chains forming the hydrophobic pocket are shown as sticks and are colored according to their position in PceA using the same scheme as in Fig. 2. Hydrogen bonding distances from Tyr246: 2.8 Å (Asn272), 2.9 Å (Arg305), and 2.5 Å (TCE).

To further analyze how substrate specificity and regioselectivity are determined by the protein structure, we analyzed the structure of PceA with its substrate TCE and the tribromoethene dehalogenation product cis-DBE. After soaking with TCE, additional density matching the shape of TCE was found in the active site pocket above the corrin ring (Fig. 3), concomitant with a reduction in the occupancy of the close-by water ligand. TCE binds with the dichlorinated carbon (C1) facing the corrin ring. Judged by the shape of the electron density, TCE bound in two orientations with the lone chloride in cis and trans to the Co-facing chloride (Fig. 3A). The PCE/TCE binding pocket is restricted by vdW contacts to the surrounding aromatic side chains (fig. S4). Tyr246 is within hydrogen-bonding distance to the Cl substituent of TCE nearest to Co.

Reductive dehalogenation of TCE produces cis-DCE, whose binding site and orientation we revealed using cis-DBE. cis-DBE was used because the stronger anomalous scattering of Br as compared to Cl allowed identification of cis-DBE also at low occupancy. Two strong patches of density were observed in the same place as two Cl atoms in the minor orientation of TCE, marking the position of cis-DBE. The chloride atom of TCE nearest to Co and Tyr246 is unoccupied in the product complex (Fig. 3B and fig. S6).

The side chains in the substrate-binding pocket are tightly packed with little conformational freedom and probably disfavor the binding of molecules significantly larger than PCE or TCE by steric exclusion. Thus, the binding pocket provides a second gate for substrate selection. In contrast to the variable residues forming the channel entrance, aromatic residues within the pocket are partially conserved between different RDases but are dispersed in sequence (fig. S5). Two layers of control over substrate selection (the “letterbox” and the active site) could achieve the necessary selectivity within the wide range of organohalides converted by distinct RDases, which often exist within the same organism (3).

PceA is attached to the periplasmic side of the cytoplasmic membrane in S. multivorans cells (17). For both PceA monomers to function independently, the twofold noncrystallographic symmetry axis of the dimer should be perpendicular to the membrane plane in the PceA-PceB complex. This would agree with two principal orientations of our PceA structure on the membrane. Amino acid residues 411 to 431 (Fig. 4, red ribbon) show high flexibility in all crystals and could only be modeled in one out of six protomers in an alternative (P21) crystal form, where α-helix 15 (fig. S1) is stabilized by a crystal contact. The disordered helices in both monomers are on the same face of the structure, and we speculate that this is the site of interaction with the PceB membrane anchor that is destabilized after complex dissociation in the purification process. Our proposed arrangement would locate the two electron entrance ports close to the membrane and the substrate channel pointing toward the periplasmic space (Fig. 4).

Fig. 4 Putative electron transfer pathways in PceA.

(Left) Model for PceA attachment to the periplasmic face of the cytoplasmic membrane. The red helix/loop represents structure only observed (helix) or well defined (β sheet/loop) in the presence of a crystal contact in the P21 crystal form. (Right) Distances (in angstroms) for putative electron (dashed lines) and proton (arrow) transfers onto the substrate. The figure shows the geometry observed in the TCE bound crystal.

RDase catalysis involves a transfer of two electrons and a proton, while accommodating the dissociation of a chloride ion from the substrate. The initial step in dechlorination of PCE probably involves a dissociative electron transfer from CoI to PCE, resulting in the formation of a trichlorovinyl radical (18, 19) by chloride elimination, while the cofactor is returned to the CoII state (18, 20, 21). The distance for this electron transfer between Co and C1 of TCE is 5.8 Å in our structure (Fig. 4).

The intermittent water (Fig. 3A) is depopulated in the TCE-bound active site only, suggesting that the β-ligand water, 2.2 Å from the proximal chloride, is displaced by TCE. The same position was identified as a weak halide-binding site by a strong anomalous signal for iodide at 0.5 M concentration (fig. S6D) but the absence of such a signal at lower chloride concentrations. This would qualify the proximal chloride at C1 over the tightly enclosed distal chloride as the leaving group, ultimately yielding cis-DCE from the minor orientation of TCE displayed in Fig. 3A, consistent with the data obtained with the cis-DBE product analog.

Recombination of the trichlorovinyl radical with CoII, forming an organometallic intermediate as proposed for free B12 (22), is an attractive next step but would be disfavored in PceA because of the dense packing above the β face of the corrin ring (fig. S4A). The same steric constraints would disable an initial nucleophilic attack of CoI on PCE. Instead, the short substrate-cofactor distances would allow the second electron transfer to occur either directly from the proximal [4Fe-4S] cluster or via the Co ion (Fig. 4). The strictly conserved Tyr246 is pointing with its phenolic hydroxyl group toward C1 and could donate the required proton to neutralize the carbanion (20). Deprotonation of Tyr246 could be stabilized by the neighboring positive charge of Arg305. Equally, a role of Tyr246 in a radical route (18) cannot be excluded.

Supplementary Materials

Materials and Methods

Figs. S1 to S6

Table S1

References (2335)

References and Notes

  1. Materials and methods are available on Science Online
  2. ACKNOWLEDGMENTS: H.D. acknowledges support by the Cluster of Excellence “Unifying Concepts in Catalysis (UniCat).” G.D. and T.S. were supported by the Deutsche Forschungsgemeinschaft (DFG) (Research Unit FOR 1530). M.B. was funded by the DFG through grant SFB 1078-A5. We are grateful for the funding of C.K. by the Ernst Abbe Foundation; in addition, we thank the staff at the Helmholtz-Zentrum Berlin (HZB) MX beamlines and DESY PETRA III beamline P11 for their assistance. We acknowledge access to beamlines of the BESSY II storage ring (Berlin, Germany) via the Joint Berlin MX-Laboratory. The authors thank O. Einsle for critical reading of the manuscript. Structure factors and models have been deposited in the Protein Data Bank under accession numbers 4UQU (empty), 4UR0 (TCE bound), 4UR1 (cis-DBE bound), 4UR2 (iodide bound), and 4UR3 (P21 crystal form).
View Abstract

Navigate This Article