Coenzyme M biosynthesis in bacteria involves phosphate elimination by a functionally distinct member of the aspartase/fumarase superfamily

For nearly 30 years, coenzyme M (CoM) was assumed to be present solely in methanogenic archaea. In the late 1990s, CoM was reported to play a role in bacterial propene metabolism, but no biosynthetic pathway for CoM has yet been identified in bacteria. Here, using bioinformatics and proteomic approaches in the metabolically versatile bacterium Xanthobacter autotrophicus Py2, we identified four putative CoM biosynthetic enzymes encoded by the xcbB1, C1, D1, and E1 genes. Only XcbB1 was homologous to a known CoM biosynthetic enzyme (ComA), indicating that CoM biosynthesis in bacteria involves enzymes different from those in archaea. We verified that the ComA homolog produces phosphosulfolactate from phosphoenolpyruvate (PEP), demonstrating that bacterial CoM biosynthesis is initiated similarly as the phosphoenolpyruvate-dependent methanogenic archaeal pathway. The bioinformatics analysis revealed that XcbC1 and D1 are members of the aspartase/fumarase superfamily (AFS) and that XcbE1 is a pyridoxal 5′-phosphate–containing enzyme with homology to d-cysteine desulfhydrases. Known AFS members catalyze β-elimination reactions of succinyl-containing substrates, yielding fumarate as the common unsaturated elimination product. Unexpectedly, we found that XcbC1 catalyzes β-elimination on phosphosulfolactate, yielding inorganic phosphate and a novel metabolite, sulfoacrylic acid. Phosphate-releasing β-elimination reactions are unprecedented among the AFS, indicating that XcbC1 is an unusual phosphatase. Direct demonstration of phosphosulfolactate synthase activity for XcbB1 and phosphate β-elimination activity for XcbC1 strengthened their hypothetical assignment to a CoM biosynthetic pathway and suggested functions also for XcbD1 and E1. Our results represent a critical first step toward elucidating the CoM pathway in bacteria.

Coenzyme M (2-mercaptoethanesulfonate, CoM 2 ) was once thought to be exclusive to methanogenesis in archaea, where it functions as a C1 carrier and plays a key role in the biosynthesis of methane gas (1)(2)(3)(4)(5). In the late 1990s, CoM was discovered to serve as a C3 carrier in a bacterial pathway for alkene metabolism in the proteobacterium Xanthobacter autotrophicus Py2 (6). Since then, other bacterial species from the Actinobacteria phylum have also been shown to use CoM in the metabolism of alkenes such as ethylene (6 -11). In X. autotrophicus Py2, propylene is converted to acetoacetate, which is subsequently funneled into the tricarboxylic acid cycle in the form of two molecules of acetyl-CoA, where it serves as a carbon source for growth. The pathway begins with the epoxidation of propylene by an NADH-dependent alkene monooxygenase, forming propylene oxide. This is followed by nucleophilic attack by CoM to break the epoxide ring and form an enantiomeric mixture of R-and S-hydroxypropyl-CoM, catalyzed by epoxyalkane: CoM transferase (12)(13)(14)(15)(16)(17). CoM then functions as a carrier, orienting R-and S-hydroxypropyl groups as oxidation substrates for a pair of stereoselective short-chain dehydrogenases, yielding 2-ketopropyl-CoM (16 -18). In the final step of the pathway, CoM again serves to orient the 2-ketopropyl group for reductive cleavage and carboxylation, forming acetoacetate and free CoM (16, 17, 19 -22). CoM, in essence, serves an analogous role in propylene metabolism as it does in methanogenesis, in promoting the proper orientation of these small organic substrates (16).
Methanogens have two known pathways for CoM synthesis, in which the carbon backbone for CoM is derived either from phosphoenolpyruvate (PEP) or L-phosphoserine (Scheme 1). The PEP-dependent pathway is initiated by a phosphosulfolactate synthase (ComA), which catalyzes the nucleophilic addition of sulfite to PEP (23,24). The phosphosulfolactate product subsequently undergoes oxidative dephosphorylation to yield sulfopyruvate (25)(26)(27)(28). Decarboxylation of sulfopyruvate, yielding sulfoacetaldehyde, is presumably followed by reduction and thiol addition to generate CoM (29,30). The L-phosphoserine-dependent pathway begins with the concerted elimination of phosphate and addition of sulfite to generate L-cysteate. L-Cysteate is subsequently transaminated to form a common intermediate with the PEP dependent pathway, sulfopyruvate (31), and after that stage, the pathways are presumed to follow the same chemical steps as CoM.
Although it is essential for alkene metabolism, a bacterial pathway for CoM synthesis has never been described. The genes responsible for alkene metabolism in the bacterium X. autotrophicus Py2 are located on a 320-kb linear megaplasmid (pXAUT01) (Fig. 1) (32,33). Proteomic experiments showed that proteins encoded in a gene cluster adjacent to the alkene-metabolizing genes were co-expressed with enzymes involved in propylene metabolism under conditions of propylene-dependent growth (14,32). These experiments also revealed additional copies of the same genes. Only the cluster directly adjacent to the alkene-metabolizing genes is considered in this work, and these are designated xcbB1, C1, D1, and E1 (XAUT_RS24680, RS24685, RS24690, and RS24695, respectively).
One of the genes in the cluster (xcbB1) is a homolog of comA, suggesting that the bacterial pathway for CoM biosynthesis is PEP-dependent. Additional genes within the cluster, however, are not similar to those from either methanoarchaeal pathways, suggesting that the bacterial CoM is synthesized via a chemically distinct series of reactions. The adjacent xcbC1 gene encodes a member of the aspartase/fumarase superfamily (AFS). Enzyme families within this superfamily include the argininosuccinate lyases/␦2-crystallins, with which the XcbC1 gene product has the closest similarity, as well as class II fuma-rases, aspartases, and adenylosuccinate lyases. All of these AFS members catalyze ␤-elimination reactions that result in the formation of an unsaturated organic product (34).
In this work, we demonstrate that the ComA homolog XcbB1 catalyzes the conversion of PEP to phosphosulfolactate (Scheme 2). We also show that XcbC1 then catalyzes ␤-elimination using phosphosulfolactate as the substrate, releasing phosphate and sulfoacrylic acid as the analogous unsaturated product. This is a new activity for a member of the AFS and a highly unusual mechanism for biological dephosphorylation. Based on the confirmation of these two activities, bioinformatic analyses of the sequences for XcbD1 and E1, and partial biochemical characterization of XcbE1, we can now propose a complete biosynthetic pathway for CoM biosynthesis in bacteria.

Sequence analyses identify gene families and suggest possible roles for putative CoM biosynthetic genes
The four putative CoM biosynthetic genes under study have deduced amino acid sequence similarities to characterized enzymes, providing clues regarding the likely chemical steps associated with bacterial CoM biosynthesis. The xcbB1 gene is homologous to comA, which encodes the phosphosulfolactate synthase that initiates the PEP-dependent CoM biosynthetic pathway in methanogens. The pathway in X. autotrophicus Py2 is consequently presumed to be PEP-dependent. The adjacent xcbC1 and xcbD1 genes encode members of the AFS. Enzyme families within this superfamily catalyze reversible ␤-elimina-Scheme 1. The PEP-dependent and L-phosphoserine-dependent pathways to CoM in methanoarchaea. Pathway I and pathway II are depicted. Both pathways culminate in production of sulfopyruvate, which undergoes decarboxylation and reduction plus thiol addition to yield CoM.
Using these preliminary sequence analyses, hypothetical roles for the enzymes in CoM biosynthesis have been proposed (Scheme 2). Homology between XcbB1 and ComA suggested that the pathway may begin with the addition of sulfite to PEP to form phosphosulfolactate. Conversion of this pathway intermediate to CoM requires net dephosphorylation, decarboxylation, and thiolation steps. The annotation of XcbC1 and XcbD1 as members of the AFS provides clear clues about the most logical trajectory by which these steps might occur. Within the AFS, XcbC1 has the strongest homology to the arginosuccinate lyases ( Fig. 2). Enzymes in this family catalyze the reversible ␤-elimination of argininosuccinate through general base proton abstraction from the C ␤ of the succinate moiety, yielding arginine and fumarate (35). Phosphosulfolactate is a loose structural analog of arginosuccinate, containing a potential proton abstraction site at the C ␤ position relative to the phosphoryl group. A proton abstraction analogous to that catalyzed by argininosuccinate lyases would lead to elimination of phosphate and formation of a carbon-double bond, yielding sulfoacrylic acid. This product, in turn, is a structural analog of fumarate, the coproduct of the ␤-elimination of arginine.
The subsequent decarboxylation and thiolation steps are less clear, although the annotations are again enlightening of at least likely elements of the next steps. XcbD1 groups most closely with members of the adenylosuccinate lyase family within the AFS (Fig. 2), which catalyze the reversible elimination of AMP from adenylosuccinate to form fumarate (52). We propose that the most likely role for XcbD1 is therefore in catalyzing an analogous reaction in the addition direction, in which a substrate is added across the double bond. Attempts to use AMP as a substrate have not resulted in formation of adenylated product (data not shown). Addition of H ϩ and an as yet undetermined co-substrate across the sulfoacrylic acid double bond (Scheme 2) would yield the putative substrate for XcbE1, the final enzyme in the pathway.
XcbE1 is homologous to pyridoxal phosphate (PLP)-dependent D-cysteine desulfhydrases that catalyze the ␣,␤-elimination of D-Cys to yield H 2 S, pyruvate, and ammonia (54). We tested XcbE1 for desulfhydrase activity via H 2 S formation assays and found that both D-and L-Cys isomers were effective substrates, where XcbE1-specific activity was 27 Ϯ 2 nmol H 2 S/ min Ϫ1 mg-XcbE1 Ϫ1 for L-Cys and 5 Ϯ 2 nmol H 2 S/min Ϫ1 mg XcbE1 Ϫ1 for D-Cys (Fig. S1). It is attractive to propose, therefore, that XcbE1 supplies cysteine-derived sulfur that is ultimately incorporated into CoM. However, without knowing the product of the XcbD1 reaction, the cosubstrate for XcbE1 the mechanism of thiolation cannot yet be determined. In addition, the substrate of XbcE1 will also require decarboxylation to arrive at the CoM product. PLP-dependent enzymes catalyze a number of reactions, including decarboxylation, transamina- Figure 1. The 320-kb linear megaplasmid of X. autotrophicus Py2 contains the genes for the putative CoM biosynthetic pathway (purple) immediately downstream of the genes that encode the enzymes responsible for propylene metabolism. Alkene monooxygenase subunits are shown in yellow, and the remaining four enzymes involved in transforming propylene oxide to acetoacetate are shown in green. Alkene-related functions for the open reading frames shown in gray have not been assigned. The locus tags shown are truncated for clarity and contain the prefix "xaut_RS2" in the pXAUT01 plasmid.

Scheme 2. Proposed bacterial pathway for CoM biosynthesis.
Steps shown in blue are supported by data reported in this study. Steps shown in red are proposed based on bioinformatics analyses. Cysteine desulfhydrase activity has additionally been demonstrated for XcbE1 in the presence of either Lor D-Cys.

A new pathway for coenzyme M biosynthesis
tion, racemization, and elimination/replacement reactions of a variety of predominantly amino acid substrates (55,56). Many PLP-dependent enzyme reactions, moreover, share common intermediates, and several enzymes have been shown to be bifunctional, catalyzing combinations of reactions that utilize the basic PLP reaction chemistry (55). Examples include the decarboxylation and transamination catalyzed by dialkylglycine decarboxylase and the ␥-elimination/␤-replacement catalyzed by threonine synthase (57,58). It is therefore conceivable that XcbE1 could be a bifunctional PLP enzyme, catalzying a more complicated final step coupling for example thiolation and decarboxylation in the production of CoM (Scheme 2).

XcbB1 catalyzes the conversion of phosphoenolpyruvate to phosphosulfolactate
To begin to test these proposed steps of the pathway, we first examined the ComA homolog XcbB1. ComA catalyzes the addition of sulfite to PEP to yield phosphosulfolactate, an unusual metabolite that, according to currently annotated Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway assignments, is unique to the methanogen PEP-dependent CoM biosynthetic pathway. An active site Mg 2ϩ in ComA coordinates the enol form of PEP, facilitating nucleophilic addition of sulfite and protonation of the adduct by a conserved activesite lysine to form phosphosulfolactate (24). The predicted phosphosulfolactate-producing activity of the ComA homolog XcbB1 was initially examined via sulfite consumption assays using monobromobimane (mBBr) as a fluorescent label for free sulfite (Fig. 3A). Sulfite consumption increased specifically when XcbB1 was incubated with the substrates PEP and sulfite, with a specific activity of 487 Ϯ 45 nmol sulfite min Ϫ1 mg Ϫ1 . This specific activity is ϳ20% of the activity reported for ComA from Methanocaldococcus jannaschii, supporting the conclusion that XcbB1 and ComA share the same physiological function (23). Q-TOF MS was subsequently used to identify phosphosulfolactate as the product (Fig. 3B). The predicted m/z for phosphosulfolactate in negative ion mode ((M-H)-species) was 248.94, which matched the m/z of the emergent signal in XcbB1-catalyzed reactions. Additionally, the isotope distribution matched the predicted isotope distributions within 5% of their predicted values. Ϫ ) and XcbB1 (Fig. S4A). PEP consumption was temporally coupled to phosphosulfolactate production, detected via the increase of a characteristic doublet signal (3.36/3.35 ppm) upfield. The phosphosulfolactate triplet signal was not visible, likely because of the suppression of the water peak, plus the additional Tris and glycerol peaks in the 3.5-4.8 ppm region. The reaction went to completion within 30 min, consistent with the general time course for sulfite consumption expected from the specific activity measured in Fig. 3. We therefore assign XcbB1 as a PEP-dependent phosphosulfolactate synthase (EC:4.4.1.19).

XcbC1 catalyzes the ␤-elimination of phosphate from phosphosulfolactate to form sulfoacrylic acid
The proposed substrate for XcbC1 is the phosphosulfolactate generated by the upstream enzyme XcbB1. In keeping with its sequence-based assignment to the arginosuccinate lyase family, the proposed ␤-elimination to yield phosphate was monitored using a coupled assay involving XcbB1, XcbC1, and molybdenum blue as a phosphate indicator (Fig. 4A). When equimolar amounts of XcbB1 and XcbC1 were present in solution with a large excess of PEP and sulfite, a robust level of phosphate production was observed (specific activity ϭ 30 Ϯ 8 nmol PO 4 3Ϫ min Ϫ1 mg XcbC1 Ϫ1 ). We therefore concluded that XcbC1 catalyzes the dephosphorylation of phosphosulfolactate, which was in turn generated by the upstream catalyst XcbB1.
Canonical alkaline/acid phosphatases are typically metal-dependent enzymes that catalyze the hydrolysis of phosphomonoesters using metal-activated water, resulting in inorganic phosphate and alcohol products (59 -65). A classic phosphohydrolase-type reaction would be expected to yield sulfolactate from phosphosulfolactate (Fig. S3A). However, Q-TOF MS analysis of the reaction products indicated that sulfoacrylic acid was the organic product ( Fig. 4B), based on its measured m/z (150.967, (M-H)-species, negative ion mode) and isotope pattern. These matched predicted values for sulfoacrylic acid (m/z ϭ 150.97) rather than sulfolactate (m/z ϭ 168.98). This suggested that XcbC1 does not catalyze a simple hydrolytic reaction on the phosphosulfolactate substrate; rather, the reaction requires concomitant release of phosphate and a proton to form an unsaturated product, consistent with ␤-elimination (see below).
Time-resolved 1 H NMR of the XcbC1 reaction directly demonstrated that disappearance of the doublet signal attributed to phosphosulfolactate was coupled with the appearance of a new pair of doublets further downfield (6.96/6.93 and 6.53/6.50 ppm, integrated to 1) (Fig. S4B). These doublets match the predicted 1 H NMR spectrum for sulfoacrylic acid (Fig. 4C and Fig.  S3B). The measured spectrum furthermore bears no overlap with the 1 H NMR spectrum observed for a sulfolactate standard (Fig. S2B), confirming the production of a new double bond. The large j-values for the pair of doublets (16.08) indicate significant separation of the olefinic hydrogens relative to each Figure 2. Bioinformatics analysis of XcbC1 (WP_011992986) and XcbD1 (WP_011992987) sequences identifies the AFS families to which each belongs is shown. The maximum likelihood tree (100 bootstrap reps) was constructed for XcbC1 and appears to show two subbranches for argininosuccinate lyasetype enzymes. Bootstrap values are shown only for nodes of relevance to this work, as indicated by circles (Ն90, black; 80 -89, gray). The canonical argininosuccinate lyase/␦2-crystallin enzymes appear to occupy a separate clade from the XcbC-type enzymes that may be involved in a pathway for CoM biosynthesis, and XcbD appears to form an additional subgroup in the adenylosuccinate lyase clade.

A new pathway for coenzyme M biosynthesis
other, meaning that the XcbC1 product exists specifically in the trans conformation. The conversion of phosphosulfolactate to sulfoacrylic acid proceeded to completion within ϳ12-14 min, again consistent with the reaction time scale indicated by the phosphate release data in Fig. 4A. We therefore conclude that XcbC1 is an unusual AFS-type phosphatase that catalyzes a ␤-elimination reaction using phosphosulfolactate as the substrate.

Modeling XcbC1 active site reactivity
The sequence of XcbC1 was analyzed in light of canonical AFS structures and mechanisms to understand the features that permit its unique reactivity. AFS members share a signature GSSXXPXKXN sequence, tertiary/quaternary fold, and active site structure and a general acid-base catalytic strategy, even while their sequence identities may be comparatively low (20 -30%) (Fig. S5) (40). Canonical AFS enzymes have been proposed to use a general base for proton abstraction from the C␤ atom of the substrate, followed by collapse of the carbanion intermediate and cleavage of the substrate. Product release may be facilitated by donation of a proton from an active-site acid to the leaving group (34). Mutagenesis and structural studies point toward the strictly conserved first serine of the GSSXX-PXKXN motif as the base (35,50,66). Interactions with backbone amides or the ␤-carboxylate group of the substrate may stabilize the catalytic Ser in its oxyanion form (41,48). The catalytic acid has been proposed to be an incompletely con-served histidine residue in argininosuccinate lyase/␦2-crystallin, adenylosuccinate lyases, and fumarate lyases, with various replacements in aspartases and members of the pCMLE (3-carboxy-cis,cis-muconate lactonizing enzyme) family (35,40,50). Finally, a strictly conserved lysine has been shown to interact with and position the ␣-carboxylate group of the substrate (41,66). The substrate-binding cavity otherwise appears to be variable, with various residues stabilizing substrates through an extensive hydrogen bonding network (50,66).
A homology-modeled structure for XcbC1 was generated using the crystal structure for a closely related ␦2-crystallin in its argininosuccinate-bound form (Fig. 5A) (40). Residues from the crystal structure that were proposed to interact with the substrate are superimposed with the corresponding residues from the XcbC1 homology model. XcbC1 appears to retain the catalytic base (Ser-285B, XcbC1 numbering) and strictly conserved lysine (Lys-291B) but very little else. Apart from the residues important for catalysis, the other residues highlighted for ␦2-crystallin form a stabilizing hydrogen bonding network around argininosuccinate (40,67). Notably, the catalytic His is absent in the XcbC1 model, with Tyr-164A-Ala-298B replacing His-162A-Glu-296B. Previous mutagenesis studies of aspartase from Bacillus sp. YM55-1 showed that the histidine is not absolutely required for activity and, when considered with the incomplete conservation of the residue, may indicate that the protonation of the substrate leaving group varies among the

A new pathway for coenzyme M biosynthesis
superfamily (34,43). Using the homology model and known AFS chemistry, it is possible to propose a catalytic mechanism for XcbC1 (Fig. 5B). Like the canonical AFS enzymes, the conserved Ser-285B could feasibly initiate the general base-catalyzed reaction through abstraction of the phosphosulfolactate C␤ proton, which we expect to have a relatively high pKa, with stabilization from Lys-291B and putative H-bond donors present in the binding site. If the phosphoryl group of phosphosulfolactate is already singly protonated, it is possible that acid catalysis is not needed to facilitate phosphate release (phosphate pKa values ϭ 2.12 and 7.21). Hence, it is plausible that, distinct from the proposed aspartase/fumarase mechanism (34), proton abstraction and phosphate release from phosphosulfolactate occur in a concerted step. Interestingly, a similar ␤-elimination resulting in the release of phosphate was recently described for the OspF family of bacterial enzymes (68,69). These catalyze the removal of phosphate from phosphothreonine, generating dehydrobutyrine (an alkene) using a conserved Lys and His as a base and acid, respectively. This so-called "eliminylation" reaction bears intriguing similarities to the proposed reaction catalyzed by XcbC1, although the OspF family does not appear to be evolutionarily related to the aspartate/fumarate lyases, and XcbC1/OspF enzymes share little homology.

Conclusions
Investigating the putative gene products of xcbB1 and xcbC1 by informatics, biochemical, and spectroscopic means provided a critical first step toward elucidating a PEP-dependent pathway for bacterial CoM biosynthesis that is distinct from the PEP-dependent pathway in methanogens. Of the four enzymes possibly involved in CoM biosynthesis, only XcbB1 is homologous to the pathway enzyme ComA, which is known to encode the first step in the PEP-dependent CoM biosynthesis in methanogens. XcbB1 catalyzes the addition of sulfite to PEP to yield phosphosulfolactate, which is delivered as the substrate for the subsequent XcbC1 reaction. The ␤-elimination of phosphate catalyzed by XcbC1 yields sulfoacrylic acid and inorganic phosphate. To our knowledge, this reaction has not been observed before in AFS enzymes, marking a novel activity for an enzyme from a large, well-characterized family as well as a novel pathway intermediate. Although the activity of XcbD1 remains unidentified, we have implicated an important hypothetical role for PLP-dependent XcbE1 in providing the source of the CoM thiol derived from Cys. This work will serve as the framework for future studies aimed at uncovering the final stages of the biosynthetic pathway. By elucidating the XcbB1 and XcbC1 reactions, we have made significant strides toward understanding bacterial CoM biosynthesis, which has evaded characterization in previous years.

Growth of X. autotrophicus Py2
Cells were grown as described previously in phosphate buffer, assorted nutrients, and trace minerals at 30°C with . The XcbC1 reaction was established through biochemical and spectroscopic means. A, the production of phosphate was monitored from a coupled assay containing 1 mM PEP, 1 mM sulfite, and one, both, or neither enzyme (XcbB1 and XcbC1). Inorganic phosphate, measured via absorbance using the molybdenum blue assay, was produced only in the presence of both XcbB1 and C1, suggesting that the latter removes phosphate from the XcbB1 product, phosphosulfolactate. B, the products of the coupled assay initiated by 5 mM PEP and 5 mM sulfite were analyzed by MS. The inset shows the isotope distribution of sulfoacrylic acid along with the corresponding ratios. The predicted m/z 150.970 and the respective predicted isotope distribution are in agreement with the results. C, time-resolved 1 H NMR showed conversion of PSL to sulfoacrylic acid (SAA) over time by the XcbB1/C1-coupled reaction, initiated by 2 mM PEP and 1 mM sulfite. For clarity, only three time points are shown (0 -2 min, black; 8 -10 min, red; 28 -30 min, blue). XcbC1 appears to fully convert PSL before the 30 min end point. For all experiments, samples included 0.1 mg of XcbB1 in pH 8 buffer and were incubated at 30°C.

A new pathway for coenzyme M biosynthesis
shaking (180 rpm) in Erlenmeyer flasks sealed with rubber septum stoppers (70, 71). The cells in liquid medium were sparged with compressed air for ϳ15 min, and propene gas was injected into the headspace (10% volume) every 12 h. The cultures were allowed to reach an optical density at 600 nm (A 600 ) of ϳ1-1.5 before harvesting by centrifugation at 6000 ϫ g for 10 min. Aliquots of 5-10 ml were reserved for genomic DNA extraction using the DNeasy blood and tissue protocol for Gram-negative bacteria.

Amplification of genes for putative CoM biosynthesis
Sequences for the putative biosynthetic operon were obtained from the NCBI database file for the pXAUT01 megaplasmid. Primers (Integrated DNA Technologies, San Diego, CA) were designed for xcbB1 (XAUT_RS24680), xcbC1 (XAUT_ RS24685), and xcbE1 (XAUT_RS24695) using the respective locus tags given in parentheses (Table S1). Restriction sites were added to clone each gene with an added N-terminal His tag into a Duet expression system (Novagen). The forward and reverse restriction sites for each ORF were as follows: XcbB1, Sac1/Nde1; XcbC1, BamHI/BglII; XcbE1, PstI/EcoRV. Each amplicon was initially cloned into a pGEM-T vector and transformed into JM109-competent cells for propagation before being cloned into a Duet vector. XcbB1 and XcbE1 were inserted into multiple cloning site 1 (MCS1) of individual pETDuet-1 (Amp R ) vectors, whereas XcbC1 was cloned under MCS1 of pACYCDuet-1 (Cm R ). Sequences were verified via Davis Sequencing (Davis, CA).

Expression and purification of putative CoM biosynthesis gene products
BL21(DE3)-competent cells were transformed with each construct, XcbB1-pETDuet-1, XcbC1-pACYCDuet-1, and XcbE1-pETDuet-1, and cells were grown on lysogeny broth agar plates supplemented with ampicillin (Amp) (0.1 mg/ml) or chloramphenicol (Cm) (0.034 mg/ml), as specified for each construct. A single colony was used to inoculate an overnight culture in liquid lysogeny broth ϩAmp or Cm medium. Expression was initiated with a 10% volume of the overnight culture (XcbE1 expression included 40 M pyridoxine) and incubated at 37°C with shaking at 250 rpm until the culture reached an A 600 of 0.6 -0.8. Protein expression was induced with 1 mM isopropyl 1-thio-␤-D-galactopyranoside, and the cells were moved to a 30°C incubator with shaking at 180 rpm. Protein expression proceeded for 3 h before the cells were harvested by centrifugation (7500 ϫ g, 10 min), followed by flash-freezing of the pellets in liquid nitrogen.

A new pathway for coenzyme M biosynthesis
For purification, frozen cell pellets were resuspended in lysis buffer (20 mM Tris, 500 mM NaCl, 5 mM imidazole (pH 8)) and homogenized with the addition of lysozyme, DNase, and phenylmethylsulfonyl fluoride. Cell pellets from cultures larger than 1 liter were lysed using a microfluidizer (M-110L Microfluidics Corp., Newton, MA). Lysates were cleared by centrifugation: 45,000 ϫ g for 30 min when using a gravity flow column or 105,000 ϫ g for 1 h prior to FPLC (Bio-Rad). The nickel-nitrilotriacetic acid columns were equilibrated to the lysis buffer prior to loading the cleared lysate. For FPLC elution, a 75-ml gradient from 0 -100% elution buffer was used. The elution buffer contained 20 mM Tris, 500 mM NaCl, 300 mM imidazole, and 20% glycerol. SDS-PAGE and Western blot anti-His tag antibodies (alkaline phosphatase-conjugated monoclonal immunoglobulin from hybridoma clone His-1, Sigma, catalog no. A5588, lot no. 096M4841V) were used to determine the purity of the protein as well as the integrity of the His tag. Buffer exchange of the pure protein into imidazole-free pH 8 buffer (20 mM Tris, 100 mM NaCl) was carried out via centrifuge filtration as a final step prior to storage at Ϫ80°C in 10% glycerol.

Determining sulfite uptake by the XcbB1-catalyzed reaction
An assay using a fluorescent sulfite indicator was adapted from prior methods (23). Reaction mixtures contained 100 mM Tris (pH 8), 100 mM NaCl, 5 mM MgCl 2 , 1 mM NaHSO 3 , and 1 mM PEP in 50 l. The assay mixture was incubated at 30°C for 5 min before addition of 0.1 mg XcbB1 enzyme in 50 mM Tris, 50 mM NaCl, and 20% glycerol (pH 8). The reaction was incubated at 30°C for a further 5 min prior to addition of 5 l of terminating solution (0.5 M arginine, 0.1 M EDTA adjusted to pH 12.8 with NaOH). Post-termination, 3 l of 50 mM mBBr dissolved in acetonitrile was added to the assay. The reaction was then incubated in the dark for 15 min at room temperature and diluted to 1 ml using 50 mM glycine and 10 mM EDTA (pH 10). Fluorescence of the sulfite-mBBr adducts was measured on a Cary Eclipse fluorescence spectrophotometer: excitation wavelength 410 nm, emission 480 nm, 350 V photomultiplier tube. Standard curves were generated using a gradient from 0 -1 mM NaHSO 3 in the reaction buffers. Data fit to linear equations (Kaleidagraph) were used to calculate concentrations of sulfite in enzymatic reactions.

Measuring inorganic phosphate production by the XcbC1catalyzed reaction
Phosphosulfolactate produced by XcbB1 was used as the substrate for XcbC1 in a coupled reaction, and the resulting inorganic phosphate was quantified. Samples containing 50 mM Tris (pH 8), 50 mM NaCl, 5 mM MgCl 2 , 0.1 mg XcbB1 enzyme, 1 mM sulfite, 1 mM PEP, and distilled H 2 O in a final volume of 500 l were incubated at 30°C for 30 min, followed by incubation at 95°C for 10 min to stop XcbB1 activity. The samples were centrifuged for 10 min at 14,000 ϫ g, and the supernatant was used for the subsequent reaction. To the supernatant, 0.1 mg XcbC1 was added to begin consumption of phosphosulfolactate. The reaction was allowed to continue for an additional 30 min before termination as with XcbB1. The supernatant was then used for phosphate determination.
Formation of inorganic orthophosphate can be detected using ammonium molybdate to form colored molybdenum blue complexes (25,72). A procedure was adapted from prior methods (73,74). Briefly, to a 500 l sample, 100 l of 2.5 M H 2 SO 4 was added, followed by 100 l of 2.5% ammonium molybdate. 10 l of reducing solution (0.2 g of 1-amino-2naphthol-4-sulfonic acid, 1.2 g of sodium bisulfite, and 1.2 g of sodium sulfite in 100 ml) was added, followed by distilled H 2 O to bring the volume to 1 ml. The samples were thoroughly mixed and then incubated at 50°C for 15 min. Absorbance was monitored at 700 nm using a Thermo Spectronic Biomate 3. Concentrations of phosphate were calculated from a standard curve generated using KH 2 PO 4 (0 -1 mM).

Determination of XcbE1 activity with an assay for H 2 S formation
The production of H 2 S can be detected using a method that converts H 2 S to methylene blue with the addition of NЈ,NЈdimethyl-p-phenylenediamine dihydrochloride (20 mM) in 7.2 M HCl and 30 mM FeCl 3 in 1.2 M HCl (75). Enzymatic assays were conducted in sealed crimp vials containing 250 l of reaction buffer (50 mM Tris and 50 mM NaCl (pH 8)), 0.1 mg of XcbE1, and distilled H 2 O to 450 l. 50 l of 10 mM L-cys, D-cys, or L-Ala was injected through the septum to initiate the reaction. Reactions were incubated for 1 h at 30°C, followed by quenching and derivatization with 100 l each of NЈ,NЈ-dimethyl-p-phenylenediamine dihydrochloride and acidified FeCl 3 . Color development proceeded for 30 min at room temperature. Absorption was measured at 670 nm and referenced to an Na 2 S standard (0 -150 M) generated under identical conditions.

Mass spectrometric analysis of reaction products
Sample buffers were prepared at 1 mM Tris and 1 mM MgCl 2 (pH 8) to avoid ion suppression. XcbB1 (0.1 mg) was incubated with 5 mM PEP and 5 mM sulfite for 45 min at 30°C prior to molecular weight cutoff filtration to remove enzyme. To determine whether phosphosulfolactate was consumed by XcbC1, 0.1 mg XcbC1 was added to the quenched XcbB1 reactions and allowed to react for 20 or 45 min. Samples were quenched by molecular weight cutoff filtration and analyzed by Q-TOF MS without additional derivatization or extractions.

Q-TOF MS
Analytes were detected with a 6530 series Q-TOF MS equipped with an electrospray ionization source (operated in negative polarity mode), and data were analyzed with Mass-Hunter workstation software version B.03.01 (Agilent Technologies). Samples were introduced via direct injection.

A new pathway for coenzyme M biosynthesis
to determine baseline peaks. The substrates PEP and bisulfite were then added to the reaction at varying concentrations (minimum 2 mM PEP for ease of detection) and monitored over the course of 30 min with NMR experiments every 2 min. 1D 1 H NMR spectra were acquired using the Bruker supplied noesygppr1d pulse sequence with 32 scans and a spectral width of 12 ppm, and data were collected into 32,000 data points. Spectral processing and analysis were performed using the Topspin TM software. Approximately 0.1 mg of XcbC1 was added to the reaction upon completion of the XcbB1 time course. The reaction was again monitored over 30 min with scans every 2 min as above.

Phylogeny and homology modeling
27 diverse members of the AFS, including six argininosuccinate lyase family members, were used to construct the maximum likelihood tree on MEGA 6.06 with 100 bootstrap replications. The resulting tree was prepared for publication using FigTree v1.4.2. Homology models for XcbC1 were constructed with SWISSMODEL, using the crystal structure of T161D variant duck ␦2-crystallin with bound argininosuccinate (PDB code 1TJW, 76% query coverage, 28% identity).