Methanogenesis marker protein 10 (Mmp10) from Methanosarcina acetivorans is a radical S-adenosylmethionine methylase that unexpectedly requires cobalamin

Methyl coenzyme M reductase (MCR) catalyzes the last step in the biological production of methane by methanogenic archaea, as well as the first step in the anaerobic oxidation of methane to methanol by methanotrophic archaea. MCR contains a number of unique post-translational modifications in its α subunit, including thioglycine, 1-N-methylhistidine, S-methylcysteine, 5-C-(S)-methylarginine, and 2-C-(S)-methylglutamine. Recently, genes responsible for the thioglycine and methylarginine modifications have been identified in bioinformatics studies and in vivo complementation of select mutants; however, none of these reactions has been verified in vitro. Herein, we purified and biochemically characterized the radical S-adenosylmethionine (SAM) protein MaMmp10, the product of the methanogenesis marker protein 10 gene in the methane-producing archaea Methanosarcina acetivorans. Using an array of approaches, including kinetic assays, LC-MS–based quantification, and MALDI TOF–TOF MS analyses, we found that MaMmp10 catalyzes the methylation of the equivalent of Arg285 in a peptide substrate surrogate, but only in the presence of cobalamin. We noted that the methyl group derives from SAM, with cobalamin acting as an intermediate carrier, and that MaMmp10 contains a C-terminal cobalamin-binding domain. Given that Mmp10 has not been annotated as a cobalamin-binding protein, these findings suggest that cobalamin-dependent radical SAM proteins are more prevalent than previously thought.

Recently, genes that encode proteins responsible for two of these modifications have been identified by bioinformatics methods and complementation of select knockouts in genetically tractable organisms. Mass spectrometric characterization of McrA from WT Methanosarcina acetivorans and mutants thereof showed that the thioglycine modification requires the tfuA and/or ycaO gene products (13). Similar studies showed that the 5-C-(S)-methylarginine modification in McrAs from M. acetivorans (14) and Methanococcus maripaludis (15) requires the mmp10 gene, which encodes methanogenesis marker protein 10 (Mmp10).
Metabolic feeding studies indicate that the methyl groups of the MeHis, MeCys, MeArg, and MeGln modifications are all derived from S-adenosylmethionine (SAM) (16). For the MeHis and MeCys modifications, a typical SAM-dependent nucleophilic displacement is assumed, wherein the target nucleophilic atom attacks the activated methyl group of SAM in a polar S n 2 process. Contrarily, the methylation of C5 of Arg or C2 of Gln is less intuitive mechanistically, because the protons on these carbon centers are not sufficiently acidic to generate a carbon nucleophile. Nevertheless, in the absence of additional clues to these reactions, two mechanisms for the formation of MeArg and MeHis were proposed (16); however, they were not particularly satisfying with respect to mechanistic precedent.
The discovery of mmp10's role in MeArg formation is of interest, because Mmp10 is a member of the radical SAM (RS) superfamily of enzymes, which catalyze an amazing array of reactions via substrate-derived radical intermediates (17)(18)(19). RS enzymes require a [4Fe-4S] 1ϩ cluster cofactor coordinated by three conserved cysteine residues in the RS domain-typically a full or partial triose-phosphate isomerase barrel-with SAM serving as the fourth ligand to the unique iron site. SAM is then cleaved reductively to methionine and a 5Ј-deoxyadenosyl 5Ј-radical (5Ј-dA ⅐ ), although one exception to this paradigm has been recently noted (20,21). Typically, 5Ј-dA ⅐ abstracts a substrate hydrogen atom, which initiates substrate-based catalysis.
One major subgroup of RS enzymes is the methylases, currently categorized into five classes (A, B, C, D, and E), with members of class B perhaps performing the most diverse chemistry (22,23 (Fig. 2). As expected, the products of the reaction are 5Ј-dAH, methionine (Met) S-adenosylhomocysteine (SAH), and the methylated peptide.
In the presence of the required low-potential reductant, Ti(III) citrate, MaMmp10 exhibits a k cat of 1.87 min Ϫ1 , which is comparable with that of other class B methylases that catalyze reactions via a 5Ј-dA ⅐ intermediate (24).

Overexpression, purification, and characterization of as-isolated MaMmp10
The M. acetivorans mmp10 gene was optimized for expression in Escherichia coli and chemically synthesized. The gene was cloned into pET-26b such that the encoded protein would contain a hexahistidine tag separated from its native C-terminal amino acid by a linker of two amino acids. The resulting plasmid, termed pMa4551, was used to transform E. coli harboring pDB1282, a plasmid that houses genes from Azotobacter vinelandii that are important in iron-sulfur (FeS) cluster biosynthesis (25,26). The transformed bacteria were grown in M9 minimal media before inducing gene expression with isopropyl ␤-D-thiogalactopyranoside (IPTG) at a final concentration of 200 M. The protein was purified by immobilized metal affinity chromatography in an anoxic environment due to the known oxygen sensitivity of iron-sulfur (FeS) clusters in RS enzymes. From 16 liters of E. coli cell culture, ϳ150 mg of Ͼ90% pure protein is obtained (Fig. 3A).
Analysis of as-isolated MaMmp10 by UV-visible spectroscopy shows a spectrum that is characteristic of an FeS clustercontaining protein, exhibiting a broad feature centered at 410 nm (Fig. 3B, red trace). Upon chemical reconstitution of the FeS cluster and subsequent purification of MaMmp10 by size-exclusion chromatography on an ÄKTA LC system housed in an EDITORS' PICK: Mmp10 is Cbl-dependent radical SAM methylase anaerobic chamber, a more prominent feature at 410 nm is observed (Fig. 3B, blue trace). Iron analysis of this reconstituted (Rc) protein reveals it to contain 5.09 Ϯ 0.06 Fe ions per polypeptide after multiplying by a correction factor of 0.9328obtained by amino acid analysis-for the Bradford method for protein quantification. This form of protein, in which only the FeS cluster has been reconstituted, is termed "as-purified."

MaMmp10 requires cobalamin for its function
A 13-amino acid peptide (EMLPARRARGPNE) spanning residues 279 -291 of the M. acetivorans McrA subunit (site 285 of methylation is shown in bold type) was synthesized by Think Peptides (www.thinkpeptides.com) and was used to assess the ability of as-purified MaMmp10 to methylate Arg 285 . The concentration of a stock solution of the peptide was determined by amino acid analysis. The calculated mass of the peptide is 1497.7 [M ϩ H] ϩ ; however, the unmodified peptide is best detected in its ϩ3 charge state, exhibiting an m/z ϭ 499.9 [M ϩ 3H] 3ϩ . In the presence of 0.5 mM SAM, 55 M peptide substrate, and 0.5 mM Ti(III) citrate, 40 M MaMmp10 displays very poor activity, generating less than 1 eq of SAH, 5Ј-dAH, and MeArg-containing peptide (m/z ϭ 504.6 [M ϩ 3H] 3ϩ ) over a period of 90 min (Fig. 3C). Although 5Ј-dAH concentrations can sometimes be higher than the theoretical yield because of abortive cleavage of SAM (25,27,28), our studies of other RS methylases indicate that SAH and the methylated product are usually strongly coupled (24, 29 -32) The extremely-low activity of MaMmp10 suggested that a necessary component may be missing from the reaction mixtures. Given that cobalamin-dependent RS enzymes are wellknown to methylate unactivated sp 3 -hybridized carbon centers, both MeCbl and hydroxocobalamin (OHCbl) were added to assess their effect on the MaMmp10 reaction, even though MaMmp10 is not annotated as a cobalamin-binding protein.
As shown in Fig. 4, the addition of either MeCbl or OHCbl results in a dramatic increase in enzyme activity, as evidenced by the substantial time-dependent increase in SAH, 5Ј-dAH, and MeArg concentrations. The similar effects of MeCbl and OHCbl on the MaMmp10 reaction suggest that the enzyme readily converts OHCbl to MeCbl, an event that has been observed in other cobalamin-dependent RS enzymes (21,24,30).
Given the effect of cobalamin on the MaMmp10 reaction, as well as the ability of as-purified MaMmp10 to catalyze low amounts of turnover in the absence of added cobalamin, the as-purified MaMmp10 was treated with acid and then analyzed by HPLC with analysis by multiple-reaction monitoring MS (MRM LC-MS) for its cobalamin content. As shown in Fig. S1 MaMmp10 was then reconstituted with its cobalamin cofactor by treating the protein with a 4-fold excess of OHCbl for 2 h before removing unbound molecules by gel-filtration chromatography using a buffer lacking cobalamin. This protein is referred to as cobalamin-reconstituted (Cbl-Rc) MaMmp10. A fraction of the eluted protein was heated with potassium cyanide to convert any bound cobalamin to dicyanocobalamin, which was subsequently quantified by UV-visible spectroscopy (⑀ 367 nm ϭ 30,800 M Ϫ1 cm Ϫ1 ). After gel filtration, MaMmp10 retains ϳ0.8 cobalamins per polypeptide (Fig. 5A). Comparison of UV-visible spectra shows a new, broad feature between 300 and 600 nm that is indicative of cobalamin (Fig. 5B, black trace) (33).
MaMmp10 activity was re-assessed by incubating 50 M Cbl-Rc or as-purified MaMmp10 with 1 mM SAM and 0.28 mM peptide under the reaction conditions described above, but in the absence of added cobalamin. As shown in Fig. S2 (blue squares), the reaction is completed within 50 -60 min of incubation time. By contrast, as-purified MaMmp10 catalyzes formation of only ϳ20 M product (less than one turnover) after 120 min (Fig. S2, red squares).

Kinetic analysis of Cbl-Rc MaMmp10
The effect of cobalamin on MaMmp10 activity was substantiated by performing reactions in triplicate using lower enzyme concentrations (10 M) and saturating concentrations of SAM (1 mM) and peptide substrate (450 M) to allow the linear phase of the reaction to be captured. Under these conditions, Cbl-Rc MaMmp10 catalyzes the near stoichiometric formation of MeArg, SAH, and 5Ј-dAH with an apparent k cat value of ϳ1.87 min Ϫ1 (Fig. 6A). Reactions were also conducted under conditions in which SAM at natural abundance was replaced with S-adenosyl-[methyl-2 H 3 ]methionine (d 3 -SAM) (Fig. 7A). In this instance, the mass of the methylated product increases by 3 mass units, consistent with a C 2 H 3 group incorporated into the peptide with m/z ϭ 505.6 ([M ϩ 3H] 3ϩ ) rather than 504.6 ([M ϩ 3H] 3ϩ ) for the peptide containing a CH 3 group at natural abundance. This result is consistent with the transfer of an intact methyl moiety from d 3 -MeCbl and is inconsistent with mechanisms that involve hydrogen atom abstraction from the methyl moiety of SAM, as has been observed in the class A and class C methylases (29,34,35). Fig. S3 shows the formation of 5Ј-dAH (red squares), SAH (blue squares), and MeArg-containing peptide (black squares) in the absence of SAM (Fig. S3A), in the absence of Ti(III) citrate (Fig. S3B), or in the absence of MaMmp10 (Fig. S3C), indicating that each is required for turnover. In addition, turnover is severely diminished when Ti(III) citrate is replaced with dithionite (DDT) as the required lowpotential reductant (Fig. 8).

Analysis of bound OHCbl, MeCbl, and d 3 -MeCbl under turnover conditions
To assess whether the methyl group from SAM is first transferred to cobalamin before the peptide substrate, the samples from reactions displayed in Figs. 6A and 7A were re-analyzed by LC-MS for the formation of MeCbl or d 3 -MeCbl, respectively.
Given that MaMmp10 was reconstituted with OHCbl in this experiment, the enzyme should release OHCbl into solution upon denaturation with acid. Upon addition of SAM and reductant, OHCbl should be converted into cob(I)alamin, which can receive a methyl group from SAM to afford MeCbl. Therefore, a reduction in OHCbl and an increase in MeCbl should be observed. Fig. 6B shows m/z (ϩ2 charge state) transitions and elution profiles that are characteristic of OHCbl (664.5 3 635, [M ϩ 2H] 2ϩ ; 3.77 min; solid lines) and MeCbl (673.0 3 665, [M ϩ 2H] 2ϩ ; 4.91 min; dotted lines), the two major species present during the reaction. At t ϭ 0, cobalamin exists exclusively in its hydroxylated form. After 30 s of incubation time, there is a rapid decrease in the amount of OHCbl and a corresponding increase in the amount of MeCbl. Theoretically, the amount of OHCbl should continue to decrease until all cobalamin exists as MeCbl; however, the lack of complete decay of OHCbl suggests that some of the cofactor does not participate in catalysis.
Samples from Fig. 7A were similarly re-examined to provide evidence that the methyl group is transferred intact from SAM to cobalamin. The samples were monitored for ions with a mass transition of 674.5 3 665.0 ([M ϩ 2H] 2ϩ ), which would be consistent with a ϩ3.0 mass increase indicative of d 3 -MeCbl. As detailed above, natural abundance MeCbl exhibits a mass transition of (673.0 3 665, [M ϩ 2H] 2ϩ ). Fig. 7B shows elution profiles and formation/decay patterns for d 3 -MeCbl and OHCbl, respectively. These patterns are similar to those

Methylcobalamin is an intermediate methyl acceptor in the MaMmp10 reaction
To generate MaMmp10 containing bound MeCbl, the protein was incubated with 5-fold excess SAM and Ti(III) citrate for 30 min and then exchanged into buffer that did not contain SAM or Ti(III) citrate by gel-filtration chromatography. A 40 M sample of "pre-methylated" MaMmp10 was combined with Ti(III) citrate and the McrA peptide substrate before initiating the reaction with either SAM at natural abundance (NA SAM) or d 3 -SAM. Fig. 9 shows the results of MaMmp10 turnover using NA SAM (Fig. 9, A and C) or d 3 -SAM (Fig. 9, B and D) as the co-substrate. Approximately 7.5 M (ϳ25%) of bound cobalamin exists as OHCbl (Fig. 9, A and B, crosses), and this percentage does not change significantly throughout the reaction, suggesting that this fraction of MaMmp10 is unreactive or poorly active. Fig. 9A shows that ϳ30 M MaMmp10 is in the pre-methylated state at t ϭ 0 (closed squares) and that the concentration of this form of the enzyme stays relatively consistent throughout the reaction. As-expected, no d 3 -MeCbl (Fig. 9A, open squares) is observed, given that the experiment was conducted with NA SAM. Accordingly, when the exact same reaction is monitored for product formation, only the MeArg-containing peptide at natural abundance (Fig. 9C, closed circles) is observed during the 10-min incubation time.
When the pre-methylated MaMmp10 was incubated with d 3 -SAM (Fig. 9, B and D), rapid formation of 22 M d 3 -MeCbl (Fig. 9B, open squares) takes place concomitant with a nearequivalent concentration of NA MeCbl lost over 10 min. Approximately 8 M cobalamin is not converted into MeCbl, which again may reflect unreactive or poorly-reactive enzyme. When formation of the MeArg-containing peptide is monitored, a burst of the species containing a methyl group at natu-ral abundance is observed, consistent with the form of premethylated MaMmp10 used in the reaction (Fig. 9D). The magnitude of the burst is similar in concentration to the starting concentration of pre-methylated MaMmp10 (ϳ30 M). This burst is accompanied by an observed lag in d 3 -MeArg formation, which occurs because NA MeCbl must be consumed before d 3 -SAM can donate a methyl group to cobalamin to generate d 3 -MeCbl. d 3 -MeCbl is then used to generated the d 3 -MeArg-containing peptide product in all subsequent turnovers. Taken together, Fig. 9 supports the use of cobalamin as an intermediate methyl acceptor in the formation of the MeArg-containing McrA peptide.

Verification of Arg 285 as the site of methylation of the peptide substrate analog
To verify that the methyl group is indeed transferred to Arg 285 of the peptide substrate, an MaMmp10 reaction was analyzed by high-resolution MS on a Brüker Ultraflextreme MALDI TOF-TOF instrument. Shown in Fig. 10 are the b and y ion series for the unmodified peptide substrate (Fig. 10A) and the methylated peptide product (Fig. 10B)  ϩ ion of the methylated peptide matches that of its theoretical mass based on its sequence, whereas the y 7 ϩ ion and those beyond are ϳ14.00 mass units greater than they should be based on their sequences. Similarly, the b 6 ϩ ion of the methylated peptide matches that of its theoretical mass based on its sequence, whereas the b 7 ϩ and b 8 ϩ ions and those beyond are ϳ14.00 mass units greater than they should be based on their EDITORS' PICK: Mmp10 is Cbl-dependent radical SAM methylase sequences. Taken together, the data show unambiguously that the methyl group is appended to Arg 285 .

Discussion
MaMmp10 catalyzes the methylation of Arg 285 in the ␣ subunit of MCR. Although the absence of MeArg does not drastically influence the activity of MCR, it has been found to increase the stability of the enzyme at elevated temperatures (13). An analysis of the primary structure of MaMmp10 strongly supports its membership in the RS superfamily of enzymes, and in vitro characterization of the isolated enzyme, reported herein, is consistent with this assignment. However, recombinantly expressed and anaerobically purified MaMmp10 methylates a 13-amino acid peptide mimic of McrA poorly. Less than 0.1% of various cobalamin species are bound to MaMmp10, most likely deriving from the E. coli expression host scavenging residual cobalamin from the growth vessels or from the LB starter culture, given that E. coli does not synthesize cobalamin de novo. Because of this poor activity, OHCbl or MeCbl was added to MaMmp10 reactions. When incubated with cobalamin and then purified by size-exclusion chromatography, MaMmp10 retains 0.8 cobalamins per polypeptide. Partial cofactor incorporation could be due to the use of hydroxocobalamin in the reconstitution procedure instead of the native cofactor, Factor EDITORS' PICK: Mmp10 is Cbl-dependent radical SAM methylase III, which contains a 5-hydroxybenzimidazolyl group instead of the typical dimethylbenzimidazole group (Fig. 11) (36, 37). This form of cobalamin is unique to methanogenic archaea (38). Nonetheless, in the presence of cobalamin or with MaMmp10 reconstituted with cobalamin, the enzyme can perform multiple turnovers, exhibiting a k cat value of 1.87 min Ϫ1 , which is similar to that observed for Fom3 using Ti(III) citrate as a reductant (0.78 min Ϫ1 ) (24). Experiments using d 3 -SAM support the use of cobalamin as an intermediate methyl carrier in the reaction, accepting an intact methyl group from SAM before transferring it intact to C5 of Arg 285 .
Although MaMmp10 is not annotated as a cobalamin-binding protein, the use of this cofactor as a methyl donor in RS reactions is well-precedented (21, 24, 39 -44). In-depth bioin-formatic analysis of Mmp10s from a variety of organisms does not reveal significant homology to any cluster of characterized RS enzymes and does not suggest any connection with known cobalamin-dependent RS enzymes outside of the homologs that likely perform the same function. In fact, all other cobalamin-dependent RS enzymes have a cobalamin-binding motif N-terminal to the RS motif, whereas MaMmp10's RS motif is at the extreme N terminus of the protein. A sequence similarity network (SSN) of Mmp10s was made through a BLAST analysis of the MaMmp10 amino acid sequence using the enzyme similarity tool on the Enzyme Function Initiative webserver (45). The sequences fell into four major clusters, with several proteins falling outside of these clusters. Although they have been shown in vivo to catalyze the same reaction,
A multiple-sequence alignment of MaMmp10 with other known cobalamin-dependent RS enzymes (Fig. S4) does not show significant homology, exhibiting only 15% sequence identity to OxsB, a known cobalamin-dependent RS enzyme that has recently had its X-ray crystal structure determined (46). The alignment shows the known and predicted cobalamin domains of several annotated cobalamin-dependent enzymes, but the domain is clearly not present in MaMmp10. Even the RS motif (CX 3 CX 2 C) is not conserved among them (Fig. S4). This analysis suggests that the cobalamin-binding domain in MaMmp10 is located C-terminal to the RS motif, which is a significant finding.
In validating the annotation of MaMmp10 across multiple databases, we discovered one annotation from Integrated Microbial Genomes and Microbiomes (IMG/M) as containing a DUF512 domain (47). DUF512 is a domain of unknown function (DUF) that is typically located C-terminal to an RS domain. However, although the DUF512 domain is annotated in the IMG/M database, MaMmp10 is not present in the proteins that have been grouped into the DUF512-containing family by the InterPro or Pfam databases (48,49). Additionally, none of the DUF512 proteins have been characterized, so the function of this domain is unknown.
An alignment of MaMmp10 with the proteins curated into the DUF512 domain by InterPro (48) does show a region of homology that is located C-terminal to the RS domain (Fig. S5). The results of the analysis do not confirm that the DUF512 domain is present in Mmp10, but it could represent a possible domain for cobalamin binding, which would expand the subfamily of cobalamin-dependent RS enzymes.
In conclusion, we show that MaMmp10 is a new member of class B RS methylases, using MeCbl to methylate Arg 285 of McrA. Initially, the methyl group derives from SAM, with cobalamin acting as an intermediate carrier. Furthermore, we show that MaMmp10 is the first reported cobalamin-dependent RS enzyme that contains a C-terminal cobalamin-binding domain. This new subclass of class B RS methylases is currently made up of proteins only from methanogenic archaea and can be further differentiated into several subclusters (Fig. 12). However, it is likely that they are all involved in the methylation of the arginine residue of MCR. This discovery highlights that Nature's use of cobalamin by RS enzymes is more abundant than previously believed.

Chemicals
Enzymes and other reagents for cloning and nucleic acid manipulation were obtained from New England Biolabs (Ipswich, MA). Sodium sulfide (Na 2 S), SAH, 5Ј-dAH, lysozyme, ␤-mercaptoethanol, OHCbl, MeCbl, and AdoCbl were purchased from Sigma. Ferric chloride was obtained from EMD Biosciences (Gibbstown, NJ). Bradford reagent for protein concentration determination as well as the bovine serum albumin (BSA) standard was purchased from Pierce, Thermo Fisher Scientific (Rockford, IL). Dithiothreitol (DTT), L-(ϩ)-arabinose, IPTG, ampicillin, and kanamycin were purchased from Gold Biotechnology (St. Louis, MO). SAM was synthesized from ATP and L-methionine and purified as described previously (50). Ti(III) citrate was prepared as described previously (51). All other materials and reagents have been previously reported (21,24,30) or were of the highest available quality.

General methods and instrumentation
DNA sequencing was performed at the Huck Life Sciences Genomics Core Facility at Pennsylvania State University, University Park. Amino acid analysis was performed at the University of California Davis Proteomics Core Facility. UV-visible spectra were recorded on a Cary 50 spectrometer from Varian (Walnut Creek, CA) using the associated WinUV software package. All anaerobic manipulations were conducted in a Coy anaerobic chamber (Grass Lakes, MI) filled with 95% N 2 and 5% H 2 gases and containing palladium catalysts to maintain the level of oxygen at Ͻ1 ppm. HPLC with detection by MS (LC-MS) was conducted using an Agilent Technologies (Santa Clara, CA) 1200 system coupled to an Agilent Technologies 6410 QQQ mass spectrometer. Data collection and analysis were performed using the associated MassHunter software. Size-exclusion chromatography was performed on an ÄKTA system (GE Healthcare) housed in a Coy anaerobic chamber and equipped with a HiPrep 26/60 Sephacryl HR S-200 column (GE Healthcare). SDS-PAGE was performed using a mini-vertical electrophoresis unit from Hoefer (Holliston, MA).

Construction of pMa4551
The gene encoding M. acetivorans Mmp10 was optimized for expression in E. coli and chemically synthesized by Invitrogen GeneArt Gene synthesis (Fig. S6). Mmp10 was designed to EDITORS' PICK: Mmp10 is Cbl-dependent radical SAM methylase contain NdeI and XhoI restriction sites at the 5Ј and 3Ј termini, respectively. The gene was removed from the GeneArt pMA-T vector using the appropriate restriction enzymes and was ligated into linearized pET26b plasmid using T4 DNA ligase. The resulting plasmid was designated pMa4551. The correct sequence of pMa4551 was verified at the Pennsylvania State Genomics Core Facility.

Overexpression and purification of MaMmp10
pMa4551 was used to transform an E. coli BL-21 (DE3) strain harboring plasmid pDB1282, which contains genes for ironsulfur (FeS) cluster assembly and trafficking from A. vinelandii (26). Expression of MaMmp10 was conducted using previously described protocols (52), and the resulting protein was purified in a Coy Laboratories anaerobic chamber under a 95% N 2 and 5% H 2 atmosphere as described previously (52).
A single colony of E. coli BL-21 (DE3) harboring pMa4551 and pDB1282 was used to inoculate 5 ml of LB media starter culture. After 12 h of agitation at 37°C and 250 rpm, 400 l of culture was added individually to 4 flasks of 4 liters of M9 minimal media, pH 7.4, supplemented with 50 g/ml kanamycin and 100 g/ml ampicillin. At A 600 ϭ 0.3, 25 M FeCl 3 was added, and expression of the isc operon was induced with 0.2% arabinose. At A 600 ϭ 0.6, 25 M FeCl 3 was added to the media, and the flasks were placed in an ice bath for 40 min. The expression of mmp10 was then induced upon addition of a final concentration of 200 M isopropyl ␤-D-1-thiogalactopyranoside. Cells were harvested in a Beckman Coulter Avanti J-26 XP centrifuge at 7000 ϫ g. Approximately 40 g of cells was obtained, which were flash-frozen in liquid nitrogen and stored in a Ϫ80°C ultralow freezer until further use.
Frozen cells were brought into a Coy Laboratories anaerobic chamber and resuspended in "lysis buffer" containing 50 mM HEPES, pH 7.5, 300 mM KCl, 10% w/v glycerol, 10 mM ␤-mercaptoethanol (BME), and 4 mM imidazole, pH 8.0. After 15 min of stirring, 1 mg/ml lysozyme, 0.1% Triton X-100, 1 protease EDITORS' PICK: Mmp10 is Cbl-dependent radical SAM methylase inhibitor tablet (Sigma), and 0.1 mg/ml DNase were added to the buffer. The solution was allowed to stir for another 20 min or until the cells had completely resuspended. Cell lysis by sonic disruption was preformed using a Fisher Scientific Model 550 sonic dismembrator at 30 -40% output. The lysate was subjected to six 45-s bursts and then distributed into Nalgene high-speed centrifuge tubes. The tubes were capped, further sealed with vinyl tape, and brought out of the glovebox. The lysate was clarified by centrifugation in a Beckman Coulter Avanti J-30I centrifuge for 1 h at 4°C. Clarified lysate was brought back into the anaerobic chamber and loaded onto a 10-ml cobalt-Talon resin (Clontech) column. The lysate was placed on ice while loading it onto the resin. After loading, the resin was washed with 10 col-umn volumes of cold lysis buffer. The purified MaMmp10 was eluted with buffer containing 50 mM HEPES, pH 7.5, 300 mM KCl, 10% glycerol, and 10 mM BME, and ϳ20 ml of protein was collected. The eluate was loaded into Ultra-15 centrifugal units (Merck Millipore Ltd.) with a 10-kDa molecular mass cutoff, and the units were capped, sealed with vinyl tape, and brought out of the anaerobic chamber to concentrate. Upon reducing the protein volume to less than 2.5 ml, the centrifugal units were brought back into the chamber and loaded onto a PD-10 desalting column (GE Biosciences) pre-equilibrated with 15 ml of "storage buffer" containing 50 mM HEPES, pH 7.5, 500 mM KCl, 10% glycerol, and 5 mM DTT. The protein was eluted with storage buffer, aliquoted, and flash-frozen in liquid nitrogen.

Chemical reconstitution of [4Fe-4S] cluster and cobalamin
All steps were performed in an anaerobic chamber unless noted otherwise. Solid DTT was added to a specific volume of storage buffer to a final concentration of 2.5 mM. 250 M FeCl 3 was added to the buffer solution, which was incubated on ice for 5 min before adding MaMmp10 to a final concentration of 50 M. The protein solution was gently stirred for 20 min on ice. After 20 min, 250 M Na 2 S was added incrementally over the span of 2 h. The protein solution was left to incubate on ice overnight in the anaerobic chamber. The following morning, the MaMmp10 was loaded into Ultra-15 centrifugal units, sealed using vinyl tape, and brought out of the chamber to concentrate. The concentrated protein was brought back into the chamber and purified further using an ÄKTA 10 system equipped with a Sephadex S-200 size-exclusion column. The migration of MaMmp10 on the column was monitored using a variable wavelength detector at wavelengths 280 and 410 nm. Fractions corresponding to a peak with significant 410 nm absorbance, excluding aggregated protein (eluting at or near the void-volume of the column), were pooled, concentrated outside of the glovebox using appropriate centrifugal units, brought back into the chamber, and flash-frozen in liquid nitrogen. The resulting protein is labeled as-purified.
As-purified MaMmp10 was incubated with ϳ2or 4-fold excess hydroxocobalamin for 2 h on ice to reconstitute it with its cobalamin cofactor. The protein solution was loaded onto a PD-10 desalting column pre-equilibrated with 15 ml of storage buffer lacking any cobalamin. The band corresponding to MaMmp10 was collected, concentrated, and either immediately assayed or flash-frozen in nitrogen and stored in liquid nitrogen.

Determination of MaMmp10 correction factor and McrA 13-mer peptide concentration
The preparation of samples for amino acid analysis has already been described (52). Lyophilized MaMmp10 and McrA 13-mer peptides were sent to the Molecular Structure Facility at the University of California Davis, where amino acid analysis was conducted on a fee for service basis.

MaMmp10 activity assays
Standard reactions were performed in triplicate and contained 50 mM HEPES, pH 7.5, 200 mM KCl, varying concentrations of MaMmp10, Ti(III) citrate, and peptide substrate. Solutions were heated at 30°C for 10 min before initiating the reaction by addition of SAM to a final concentration of 0.5 or 1 mM. Reactions were allowed to proceed before being quenched in a solution of 100 mM H 2 SO 4 and 100 M tryptophan, which was used as an internal standard.

LC-MS-based quantification and analysis
An Agilent Technologies 1200 Series HPLC system coupled to a 6410 QQQ electrospray-ionization mass spectrometer was used to detect and quantify the predicted reaction products, SAH, 5Ј-dAH, and the MeArg-containing peptide, as well as the internal standard tryptophan. Tryptophan, SAH, and 5Ј-dAH were detected by MRM in positive mode. Analyte concentrations were calculated using Agilent MassHunter Quantitative Software Version B.7.00. Calibration curves were generated using known concentrations of SAH and 5Ј-dAH. To quantify the MeArg-containing peptide, a calibration curve of known concentrations of unmethylated peptide was created and used in the detection of m/z ϭ 504.6 ([M ϩ 3H] 3ϩ ) ions, with the assumption that the methylated and nonmethylated peptides fly similarly in the mass spectrometer.

Determination of the site of methylation by MALDI TOF-TOF MS
To determine the precise site of methylation on the peptide substrate mimic, reaction mixtures were analyzed on a Bruker (Billerica, MA) Ultraflextreme MALDI TOF-TOF instrument The results demonstrate that there are four major clusters out of seven, with several proteins not belonging to a specific cluster (light blue). The red cluster represents those sequences that are closely related to MaMmp10 (indicated by a black star), and the green cluster represents the sequences that are related to MmMmp10 (indicated by a yellow star).
EDITORS' PICK: Mmp10 is Cbl-dependent radical SAM methylase equipped with a 355-nm frequency-tripled NdYAG smartbeam-II laser. The mass spectra were acquired using a factoryconfigured instrument method for the reflector positive-ion detection over the 700 -3500 m/z range. Laser power attenuation and pulsed ion extraction times were optimized to achieve the best signal-to-noise ratio. The instrument was calibrated with a BSA tryptic peptide mixture (Protea Biosciences, part no. PS-204-1). Mass spectra were opened in FlexAnalysis (Bruker), smoothed (SavitzkyGolay, 0.2 m/z, 1 cycle), and baseline-subtracted (TopHat), and the mass lists were generated using a Snap peak detection algorithm with the signal-to-noise threshold set at six and using the Averagine SNAP average composition.
The peaks at m/z 1496.8 ([M ϩ H] 1ϩ ) and 1510.8 ([M ϩ H] 1ϩ ) were selected for TOF-TOF analysis to establish the position of modification. The resulting MS 2 data were processed using the default parameters and interpreted manually. Fragment mass lists were generated using Protein Prospector (http://prospector.ucsf.edu/prospector/cgi-bin/ msform.cgi?formϭmsproduct), 3 and the fragment ions were identified based on the best match between the experimental and predicted values.

Bioinformatic analysis of MaMmp10
The MaMmp10 sequence similarity network (SSN) was prepared by a BLAST (53) analysis of the MaMmp10 amino acid sequence (UniProt accession number Q8THG6) on the Enzyme Function Initiative (EFI) website (https://enzymefunction. org) using the enzyme similarity tool (Fig. 10) (45). An e-value of Ϫ150 was chosen and resubmitted to the servers. Next, a sequence identity of 70% was chosen for generation of the colored SSN (45). Both SSNs were visualized and manipulated in Cytoscape version 3.5.1 (54).
All of the alignment figures were generated in the following manner. The FASTA sequences were taken from Uniprot (55) and aligned using Clustal Omega (56). The aligned sequences were copied into Microsoft Word where they were colored manually. The alignment of known cobalamin-dependent proteins (Fig. S4) was generated by including RS enzymes that have been confirmed to use cobalamin. OxsB was used as a reference for the alignment, because it is the only cobalamindependent RS enzyme to have its three-dimensional structure determined (46). The alignment of MaMmp10 with representative DUF512-containing proteins (Fig. S5) was constructed by aligning the MaMmp10 and MmMmp10 amino acid sequences against 20 randomly selected DUF512 family of proteins annotated by InterPro, out of a total of 3027 DUF512-containing proteins (48). The highlighting of the PDZ, aldolase-type TIM barrel, and DUF512 domains was approximated based on the predicted ranges from InterPro annotations (48). The percent conservation to decide highlighting was based on all 22 out of 22 sequences containing the particular amino acid for 100% conservation; greater than 19 out of 22 for greater than 85% conservation, and 6 out of 22 for greater than 25% conservation. Hydrophobic, charged, and serine/threonine amino acids were allowed to substitute for each other when conservation less than 100% was being considered.