The structural basis of protein conformational switching revealed by experimental and AlphaFold2 analyses

Multidomain proteins that engage in conformational acrobatics in service of their cellular functions are typically challenging candidates for structure elucidation. Depending on the protein, metabolites, metals, or posttranslational modifications can potentially serve as ensemble conductors, shifting conformational equilibria to orchestrate function (1). Cobalamin (or B₁₂)-dependent methionine synthase is one such multimodular protein where the linear arrangement of four domains has conjured up “beads-on-a-string” imagery (Fig. 1A). In this protein, the two N-terminal catalytic domains alternately sample the (third) B₁₂ domain, while the (fourth) C-terminal regulatory domain is engaged, as needed, for cofactor repair. In a tour-de-force published in this issue (2), Ando et al have harnessed the combined power of experimental structure determination and plumbing the depths of the Alpha2Fold database (3), to describe an enzyme acrobat. Their study illuminates how methionine synthase switches between catalytic and repair conformations, and identifies methyl-tetrahydrofolate (CH₃-H₄F) as a key ensemble conductor, gating entry of the resting enzyme to the catalytic cycle (Fig. 1B).

Fig. 1.

The structural basis of catalytic and repair cycles of methionine synthase. (A) Beads-on-a-string domain arrangement of methionine synthase. (B) Advanced structural studies using a combination of experimental (SAXS, cryo-EM) and computational (AlphaFold2) approaches have furnished a view of the conformational juggling during the catalytic and repair cycles. The colors used to denote the domains are the same as in A. in the resting state; the AdoMet domain is disordered and is not shown. Hcy is homocysteine.

Ando et al have harnessed the combined power of experimental structure determination with plumbing the depths of the Alpha2Fold database to describe the movements of an enzyme acrobat.

Methionine synthase catalyzes the methyl group transfer from CH₃-H₄F to homocysteine, forming H₄F and methionine (4). While the enzyme is involved in de novo methionine synthesis in bacteria, it functions primarily to liberate H₄F from the circulating CH₃-H₄F form that enters cells, to support one-carbon metabolism in mammals (5). During catalysis, the cofactor toggles between the 4-coordinate cob(I)alamin and 6-coordinate methyl-cobalamin (MeCbl) states, as it alternately accepts a methyl group from CH₃-H₄F and donates it to homocysteine. The occasional oxidative interception of the supernucleophilic cob(I)alamin intermediate leads to cob(II)alamin and egress from the catalytic cycle. The inactive enzyme engages a repair system, comprising a methyl (S-adenosylmethionine, AdoMet) and an electron (flavodoxin in bacteria) donor, to regenerate MeCbl (6). Each of the substrates, homocysteine, CH₃-H₄F, and AdoMet, binds to a separate module (Fig. 1A), which must interact with the B₁₂ domain as methyl groups bounce onto and off cobalamin during the catalytic and repair reactions. Furthermore, the cofactor itself undergoes redox-linked coordination state changes during the catalytic cycle as noted above, and exists as 5-coordinate cob(II)alamin in inactive enzyme (7). A “divide and conquer” approach has been used to understand the structural underpinnings of this complicated juggling act, using single or didomain constructs of methionine synthase (8, 9), leaving gaps in our understanding of how module rotation is signaled. Now, using full-length protein and advanced structural approaches, the Ando study has filled many of these gaps and pieced together the choreography of the methyl transfer reaction catalyzed by methionine synthase (Fig. 1B).

B₁₂ binds to methionine synthase with its dimethylbenzimidazole tail tucked in a side pocket; the lower axial coordination position to cobalt is occupied by a histidine residue in the so-called “His-on” conformation (10). This histidine residue toggles to the “His-off” position in cob(II)alamin upon recruitment of flavodoxin (11), the repair protein for bacterial methionine synthase (12). In the current study, small-angle X-ray scattering (SAXS) analysis revealed that the conformational ensembles of full-length methionine synthase are similar for all three cobalamin oxidation states, and, as predicted by biochemical data, recruitment of flavodoxin by the cob(II)alamin-bound enzyme, stabilizes the His-off state. Importantly, SAXS analysis revealed that CH₃-H₄F exclusively elicits a conformational change in methionine synthase that is indifferent to the cobalamin oxidation state, while neither homocysteine nor AdoMet is capable of eliciting this change. Cryo-EM studies on methionine synthase from the thermophile, Thermus filiformis, then illuminated the nature of the CH₃-H₄F-induced conformational change and its functional significance for ushering the resting enzyme to the catalytic cycle.

The B₁₂ domain houses a four-helix cap that impedes solvent access and sequesters the cofactor from both the N-terminal substrate-binding domains in the resting state of methionine synthase (Fig. 1B). CH₃-H₄F enforces uncapping of the B₁₂ domain due to the competitive interaction between the substrate and the cap for a folate-sensitive loop. Thus, CH₃-H₄F binding drives the conformational transition from the resting enzyme in which the B₁₂ domain nestles between the two substrate domains but the cofactor is sequestered, to the active enzyme where the cofactor is exposed to substrates. Meanwhile, the C-terminal AdoMet domain, blurry from sampling a wide conformational space, is distant from the B₁₂ domain, making the methyl group of AdoMet unavailable during the catalytic cycle. The activation domain is in fact, encoded on a separate polypeptide in some bacteria (13), consistent with its motional independence from the other domains in the resting form of the T. filiformis enzyme. This first glimpse of full-length methionine synthase reveals that the catalytic core is organized as a rather compact structure and sets the stage for understanding the domain movements that are needed for progress through the catalytic cycle (Fig. 1B).

For the final piece of this study, the authors adroitly mined the depths of AlphaFold2 (3, 14) and found four distinct conformations of methionine synthase, including one corresponding to their experimentally derived resting state structure. This correspondence is remarkable in light of the fact that a full-length methionine synthase structure did not exist in the training dataset. The majority of the AlphaFold2 predicted structures represent the putative reactivation state, in which the cap is replaced by the AdoMet-domain, while a minority correspond to the cryo-EM-derived resting state conformation. The high similarity between the experimentally derived B₁₂-containing full-length structure and the predicted B₁₂-lacking model is interesting, and suggests that the cofactor domain is largely preorganized. Structural analysis of glutamate mutase has similarly revealed that the B₁₂ domain is largely preformed in this apo-enzyme (15). By assessing all 4,915 structures in the AlphaFold2 database, the authors found homocysteine methylation and CH₃-H₄F demethylation-ready conformations in which the respective substrate domains are adjacent to the uncapped B₁₂-binding Rossmann subdomain (Fig. 1B). The combination of structure determination and structure prediction analysis thus furnished models of four major conformers of methionine synthase.

As with any piece of scientific scholarship that is well done, this study helps tie together biochemical observations made over decades of study but also raises questions that are ripe for further investigation. For example, the study explains the functional significance of the preferred order of substrate binding (16), wherein CH₃-H₄F dislodges the cap, opening access to the B₁₂ cofactor, which alternately methylates homocysteine and demethylates CH₃-H₄F. It also explains the observations from a limited proteolysis study, which revealed that the AdoMet domain is the first to separate, and, that this cleavage is insensitive to the cobalamin oxidation state, i.e., occurs in the resting state of the enzyme (17). Some key questions that remain to be addressed include how the supernucleophile, cob(I)alamin, is sheltered during the large conformational rearrangement between the homocysteine methylation and CH₃-H₄F demethylation half-reactions, and the interactions between flavodoxin and the repair state of the protein (Fig. 1B). Finally, cofactor loading onto the highly homologous mammalian methionine synthase requires chaperones (7), and a structural perspective on how these additional proteins interact with and are accommodated by a multimodular enzyme will be invaluable for understanding the process, and its corruption by disease-causing variants.