Functional interactions between posttranslationally modified amino acids of methyl-coenzyme M reductase in Methanosarcina acetivorans

The enzyme methyl-coenzyme M reductase (MCR) plays an important role in mediating global levels of methane by catalyzing a reversible reaction that leads to the production or consumption of this potent greenhouse gas in methanogenic and methanotrophic archaea. In methanogenic archaea, the alpha subunit of MCR (McrA) typically contains four to six posttranslationally modified amino acids near the active site. Recent studies have identified enzymes performing two of these modifications (thioglycine and 5-[S]-methylarginine), yet little is known about the formation and function of the remaining posttranslationally modified residues. Here, we provide in vivo evidence that a dedicated S-adenosylmethionine-dependent methyltransferase encoded by a gene we designated methylcysteine modification (mcmA) is responsible for formation of S-methylcysteine in Methanosarcina acetivorans McrA. Phenotypic analysis of mutants incapable of cysteine methylation suggests that the S-methylcysteine residue might play a role in adaption to mesophilic conditions. To examine the interactions between the S-methylcysteine residue and the previously characterized thioglycine, 5-(S)-methylarginine modifications, we generated M. acetivorans mutants lacking the three known modification genes in all possible combinations. Phenotypic analyses revealed complex, physiologically relevant interactions between the modified residues, which alter the thermal stability of MCR in a combinatorial fashion that is not readily predictable from the phenotypes of single mutants. High-resolution crystal structures of inactive MCR lacking the modified amino acids were indistinguishable from the fully modified enzyme, suggesting that interactions between the posttranslationally modified residues do not exert a major influence on the static structure of the enzyme but rather serve to fine-tune the activity and efficiency of MCR.

interactions between the modified residues, which alter the thermal stability of MCR in a 36 combinatorial fashion that is not readily predictable from the phenotypes of single 37 mutants. Surprisingly, high-resolution crystal structures of the various unmodified MCRs 38 were indistinguishable from the fully modified enzyme, suggesting that interactions 39 between the post-translationally modified residues do not exert a major influence on the 40 8 molecular replacement. Relevant data reduction and refinement statistics are provided in 134 Supplemental Table S1. 135 As expected from the high level of sequence conservation (90% identity), the 136 three-dimensional structure of MaMCR displays an architecture that is nearly identical to 137 that of MCR from Methanosarcina barkeri (MbMCR; PDB code: 1E6Y), with a root 138 mean square (RMS) deviation of all Cα atoms of 0.24 Å. The protein complex 139 crystallized as a α 2 β 2 γ 2 assembly with one complex in the crystallographic asymmetric 140 unit. Electron density for the affinity tag was not evident ( Figure 1A). Notably, the active 141 Desaturation of the Cα-Cβ linkage at Asp470 to yield a didehydroasparate cannot be 148 discerned at this resolution, but was confirmed by mass-spectrometry (see below). 149 Therefore, we have included the didehydroasparate modification in the structural model.  Significantly, phylogenetic profiling of the MA4545 locus revealed that homologs are 157 absent in two hyperthermophilic methanogens (Methanocaldococcus janaschiii and 158 Methanopyrus kandleri) that have been shown to lack S-methylcysteine within MCR 159 (24). The phylogenetic tree of MA4545 is incongruent with both the MCR phylogeny and 160 the reference phylogeny of archaea built with concatenated housekeeping genes ( Figure  161 2B) (9, 11). Accordingly, it seems likely that a gene loss event and subsequent HGT 162 event in the last common ancestor of the Methanosarcinaceae family is responsible for 163 the shared ancestry of this locus between members of the Methanosarcina genus and the 164 distantly related Methanobacteriales ( Figure 2B). 165 To test the hypothesis that MA4545 is responsible for S-methylation of Cys472 in The ∆mcmA mutant was viable on all growth substrates tested ( Figure 4A). 180 Relative to wild-type, the mutant grew 30% slower on dimethyl sulfide (DMS) (p = 0.002 181 for an unpaired t-test with the means of three biological replicates) ( Figure 4A). A 12% 182 decrease in growth yield (measured as the maximum optical density at 600 nm) was 183 observed on trimethylamine (TMA) (p = 0.018) ( Figure 4B). Thus, even though the S-184 methylation of Cys472 in McrA is dispensable in M. acetivorans, it is clearly important 185 for methanogenic growth on certain substrates. Curiously, the ∆mcmA mutant had better 186 growth rates and yields than the wild type on some substrates, with more pronounced 187 improvements at higher temperatures ( Figure 4). 188 Generation of combinatorial mutants of M. acetivorans lacking the thioglycine, 5-189

(S)-methylarginine, and S-methylcysteine modification in McrA 190
Since the α subunit of MCR contains multiple modified amino acids in spatial 191 proximity ( Figure 1B), epistasis between these residues may be important for optimal 192 enzyme function. To test this, we generated deletion mutants lacking the genes 193 responsible for installing S-methylcysteine, thioglycine and 5-(S)-methylarginine in all 194 possible combinations (Supplementary Figure S3A). These include MA4551, which we 195 have renamed mamA (methylarginine modification) to better reflect its function 196 (Supplementary Figure S4 and (27, 28)) and ycaO-tfuA (MA0165/MA0164), which 197 coverts glycine to thioglycine (26, 29). The phenotypic analyses described below were 198 carried out in mutants that encode wild-type MCR, while biochemical and structural 199 studies were conducted with MCR purified from a second set of mutants that encode a  Surprisingly, the triple deletion mutant grew faster than the ∆ycaO-tfuA/∆mcmA mutant, 235 indicating that the unmodified Arg285 residue alleviates the growth defect observed in 236 the absence of both the thioglycine and S-methylcysteine modifications ( Figure 4A and 237 4B). On substrates with high free energy yields (methanol and TMA), most of the 238 mutants lacking one or more of the modified residues grew as well as, or better than, the 239 wild-type strain ( Figure 4A). However, substantial growth rate defects, of varying 240 magnitude depending on the strain, were observed on substrates with low free energy 241 yields (DMS and acetate) ( Figure 4A). 242 In a previous study, we observed that the ∆ycaO-tfuA mutant has a severe growth 243 defect at elevated temperatures, which suggested that the thioglycine modification plays 244 in a role in stabilizing the active site of MCR (26). Therefore, we assayed the growth 245 phenotype of all mutants on 50 mM TMA at 29 °C, 36 °C, and 42 °C, where 36 °C 246 represents the optimal growth temperature. While all mutants grew at every temperature, 247 the growth phenotypes varied dramatically ( Figure 4C, Figure 4D, Supplementary Tables 248 S9-S13). Indeed, the mutant lacking all three modified residues grew 18% faster than 249 wild-type at 29 °C (p = 0.001) ( Figure 4C, Figure 4D, Supplementary Table 10). 250 Significantly, the mutant lacking only the S-methylcysteine modification had a 20% 251 increase in growth rate relative to wild-type at 42 °C (p = 0.018) ( Figure 4C, The global structures of MCR variants purified from the combinatorial mutants 282 remain unchanged relative to the wild-type, consistent with the lack of long-range 283 interactions between the modified residues. Surprisingly, the structural analyses show 284 that there are also no significant local changes to the active site pocket in any of the 285 variants ( Figure 6). Given the observation that the thioglycine and 5-(S)-methylarginine 286 modifications influence thermal stability, we had expected the corresponding structures 287 to reveal active site perturbations that would cause these differences. The similarity of the 288 structures is also surprising given the extensive side chain interactions that occur at the 289 site of each particular modification, which span the different subunits of MCR. We note, 290 however, that because the crystal structures are likely to capture the lowest energy 291 conformational state of each of the variant, dynamic movements that may occur at the 292 voids created by removal of the amino acid modifications are unlikely to be captured in 293 these static structures. 294

Discussion 295
All characterized MCRs contain a set of core and variable modified amino acids 296 near the active site (1, 9, 24). For the last two decades, researchers have speculated on the 297 role of these modifications vis-à-vis the function of MCR. These hypotheses have ranged 298 from certain modified residues being critical for activity to others playing minor roles (1). 299

Recent studies have identified genes installing thioglycine and 5-(S)-methylarginine and 300
shown that neither of these residues is essential for catalysis, although they might be 301 important for the structural integrity of the MCR complex (26, 27). Here, we build on our 302 previous study (26) by identifying a SAM-dependent methyltransferase involved in the 303 installation of S-methylcysteine, a variable modified amino acid, and also uncovering 304 evidence that epistasis between the three modified residues is also important for MCR in 305 vivo. 306 The data presented above show that a dedicated SAM-dependent 307 methyltransferase (renamed mcmA) is responsible for the S-methylation of a conserved 308 janaschii, as well as from psychrophilic methanogens like Methanococcoides burtonii or 315 methylcysteine modification is involved in the adaptation of MCR to mesophilic 317 environments ( Figure 2B). This idea is supported by our observation that the ∆mcmA 318 mutant grew faster than the parent strain at both elevated (42 °C) and reduced (29 °C) 319 temperatures ( Figure 4C). Despite enhanced growth at high temperatures, the thermal 320 stability of the MCR complex lacking the S-methylcysteine modification is 321 indistinguishable from wild-type ( Figure 5A). This indicates that the presence of S-322 methylcysteine does not improve the global stability of the enzyme complex. 323 Furthermore, the structure of MCR from this mutant is essentially identical to that of the 324 wild-type, including within the active site pocket near the modification site ( Figure 6). 325 Thus, we suspect that the temperature-dependent phenotypes of strains lacking S-326 methylcysteine are related to catalysis rather than structure. Curiously, homologs of 327 McmA are absent in a few mesophilic methanogens, like the gut-associated 328 Methanomassiliicoccus species ( Figure 2B). Whether the S-methylcysteine modification 329 is indeed absent in this organism, or whether another protein is capable of installing this 330 modification, remains to be determined. 331 Despite the fact that the modified residues in MCR are in spatial proximity, only 332 independent functions have been proposed for these unusual amino acids (i.e. they do not 333 exert any influence on each other (22, 24). Our in vivo growth data, as well as the in vitro 334 MCR thermal stability data, clearly show that each of the characterized modifications 335 influences the function of the others. Moreover, our data demonstrate that the function of 336 these modified amino acids is not reflected in static crystal structures. Thus, it seems 337 highly likely that they exert their effects during enzyme turnover. Whether these effects 338 are related to substrate binding and catalysis, or conformational/allosteric communication 339 between the two active sites, as in the proposed two-stroke catalytic mechanism (1, 36), 340 will await the in vitro characterization of active MCR from our mutants. Unfortunately, 341 given its sensitivity to oxidative inactivation, in vitro kinetic characterization of MCR is 342 especially problematic. The most widely used activation protocol relies hydrogenase 343 activity (16). Unfortunately, as M. acetivorans does not produce active hydrogenases, this 344 activation procedure is not directly transferable to our system. An alternate protocol 345 involving pretreatment of cells with sodium sulfide to generate the Ni(III) from of the  In summary, recent studies have changed our view of the modified amino acids in 401 MCR from biochemical novelties to evolutionary spandrels: features that are an offshoot 402 of adaptation rather than a direct product thereof. Despite this paradigm shift, it is still 403 perplexing as to why members of the ACR family contains between four to six unique 404 and rare post-translational modifications. If the role of these modifications is to fill voids 405 or contort the amino acid backbone, why didn't the twenty standard amino acids suffice? 406 It is entirely feasible that the functions of these modifications extend beyond the scope of 407 20 conditions that can be tested in a laboratory setting. As we sample diverse groups of 408 methanogenic archaea, ANMEs, and even anaerobic alkane oxidizing archaea and 409 identify the pattern of modifications in their ACRs, maybe the underlying reasons will 410 become apparent. In the near future, we expect that broader surveys of diverse ACRs 411 coupled with laboratory-based genetic experiments will enable the design of appropriate 412 experiments to tease apart the role of these unusual modified amino acids.  to control copy-number of oriV-based plasmids (41), was used as the host strain for all 453 plasmids generated in this study (Supplementary Table S16). Electrocompetent cells of 454 WM4489 were generated as described in (41)   The highest protein concentration (ca. 1 mg/mL) was obtained in fraction 2 therefore this 502 fraction was used to conduct the Thermofluor assay. To visualize the purified protein, 10 503 µL of fraction 2 was mixed with an equal volume of 2× Laemmli sample buffer (Bio-504 Rad, Hercules, CA) containing 5% β-mercaptoethanol, incubated in boiling water for 10 505 BioLabs, Ipswich, MA, USA) at 37 °C for 24 h. The resulting digested peptides were 520 desalted with C-18 zip-tips using acetonitrile with 0.1% formic acid prior to MS analysis. 521

HPLC purification of MCR tryptic fragments. The aforementioned tryptic 522
peptides were subjected to HPLC analysis using a C18 column (Macherey-Nagel, 4.6 × 523 250 mm, 5 µm particle size). Acetonitrile and 0.1% (v/v) formic acid were used as the 524 mobile phase. A linear gradient of 20 to 80% acetonitrile over 23 min at 1 mL/min was 525 used to separate the peptides. Fractions were collected at one-minute intervals and 526 analyzed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry 527 Bruker UltrafleXtreme MALDI TOF/TOF mass spectrometer was purchased in part with 590 a grant from the National Institutes of Health (S10 RR027109 A). 591

Didehydro-Asp470
S-methyl-Cys472 WT (1) All growth assays were performed at 36 °C. Heat maps depicting C) growth rate or D) 758 growth yield (measured as the maximum optical density at 600 nm) of mutants in 759 bicarbonate-buffered High Salt medium supplemented 50 mM TMA at three different 760 temperatures as indicated. Statistically significant differences in growth parameters (p < 761 0.05 or p<0.01) relative to the wild-type as determined by a two-sided t-test are indicated 762 with an * and ** respectively. The black box for the ∆mcmA∆ycaO-tfuA mutant in DMS 763 supplemented HS medium indicates that no measurable growth was detected after six 764 months of incubation. The primary data used to construct the heatmaps is presented in 765 Supplementary Tables 2-13.