|Home | About | Journals | Submit | Contact Us | Français|
Bacterial coenzyme B12-dependent 2-hydroxyisobutyryl-CoA mutase (HCM) is a radical enzyme catalyzing the stereospecific interconversion of (S)-3-hydroxybutyryl- and 2-hydroxyisobutyryl-CoA. It consists of two subunits, HcmA and HcmB. To characterize the determinants of substrate specificity, we have analyzed the crystal structure of HCM from Aquincola tertiaricarbonis in complex with coenzyme B12 and the substrates (S)-3-hydroxybutyryl- and 2-hydroxyisobutyryl-CoA in alternative binding. When compared with the well studied structure of bacterial and mitochondrial B12-dependent methylmalonyl-CoA mutase (MCM), HCM has a highly conserved domain architecture. However, inspection of the substrate binding site identified amino acid residues not present in MCM, namely HcmA IleA90 and AspA117. AspA117 determines the orientation of the hydroxyl group of the acyl-CoA esters by H-bond formation, thus determining stereospecificity of catalysis. Accordingly, HcmA D117A and D117V mutations resulted in significantly increased activity toward (R)-3-hydroxybutyryl-CoA. Besides interconversion of hydroxylated acyl-CoA esters, wild-type HCM as well as HcmA I90V and I90A mutant enzymes could also isomerize pivalyl- and isovaleryl-CoA, albeit at >10 times lower rates than the favorite substrate (S)-3-hydroxybutyryl-CoA. The nonconservative mutation HcmA D117V, however, resulted in an enzyme showing high activity toward pivalyl-CoA. Structural requirements for binding and isomerization of highly branched acyl-CoA substrates such as 2-hydroxyisobutyryl- and pivalyl-CoA, possessing tertiary and quaternary carbon atoms, respectively, are discussed.
Coenzyme B12-dependent acyl-CoA mutases are a relatively small family of enzymes catalyzing carbon skeleton rearrangements through a unique radical mechanism (1). Thus far, methylmalonyl-CoA mutase (MCM),5 ethylmalonyl-CoA mutase (ECM), and isobutyryl-CoA mutase (ICM) have been characterized (2,–4). In addition, a variant of ICM (IcmF) has been described as a fusion of ICM with the G-protein chaperon MeaI (5), a paralog to the MCM-associated MeaB protein. Recently, we have characterized 2-hydroxyisobutyryl-CoA mutase (HCM) as the newest member of the acyl-CoA mutase family (6). All these radical enzymes catalyze reversible 1,2-rearrangements of the carboxylic acid skeleton of their CoA ester substrates (Fig. 1), e.g. the MCM-catalyzed interconversion of methylmalonyl- and succinyl-CoA, with very narrow substrate specificity. Mutases play a central role in primary metabolism, such as branched-chain amino acid catabolism in mammals (MCM) and the ethylmalonyl-CoA pathway for acetate assimilation in prokaryotes (ethylmalonyl-CoA mutase). In contrast, bacterial ICM seems to be mainly involved in synthesis of secondary metabolites, such as macrolide and polyether antibiotics (7, 8). However, a role of the ICM-like IcmF in bacterial catabolism of the prodrug component pivalic acid has recently been suggested (9). Likewise, HCM is employed in some bacteria for dissimilation of 2-hydroxyisobutyric acid (6). This unusual short-chain carboxylic acid is formed during degradation of xenobiotic fuel oxygenates (10) but is also found in humans with lactic acidosis (11), and lysine 2-hydroxyisobutyrylation has recently been identified as a widely distributed histone mark (12). Besides these characterized enzymes, several not yet identified B12-dependent mutases have been postulated to be involved in microbial degradation pathways for the anaerobic mineralization of alkanes and ethylbenzene (8), indicating that different subfamilies of acyl-CoA mutases may exist as key enzymes in microbial degradation pathways of natural compounds as well as xenobiotics.
The unusual radical mechanism exchanging a hydrogen with a thioester group bonded on vicinal carbon atoms (Fig. 1) has been thoroughly studied in MCM from the Gram-positive bacterium Propionibacterium freudenreichii subsp. shermanii (PfMCM) (13, 14). These studies showed that binding of the acyl-CoA substrate results in a large conformational change that causes the formation of an initial radical by homolysis of the adenosylcobalamin-cobalt bond in the coenzyme B12. This 5′-deoxyadenosyl radical abstracts a hydrogen atom from the substrate (Fig. 1). Then, the substrate radical undergoes 1,2-rearrangement by migration of the thioester group and re-abstracts the hydrogen from the adenosine to form the acyl-CoA product. A striking feature of all acyl-CoA mutases is that the coenzyme B12 and the CoA substrate are completely buried within the enzyme. The amino acids interacting with B12 and the CoA moiety of the substrate are highly conserved in all acyl-CoA mutases. In contrast, only very few residues seem to be characteristic for each mutase subfamily for specifically interacting with the acyl moiety of the substrate, namely Tyr89 and Arg207 in PfMCM (15, 16). These residues form several H-bonds with the free carboxyl group of the substrate and radical intermediate. In addition, the Tyr determines strict specificity toward the (R)-enantiomer of methylmalonyl-CoA (15). Based on sequence comparisons, residues homologous to Tyr89 and Arg207 of PfMCM have been identified as PheA80 and GlnA198 in ICM subunit A (IcmA) of Streptomyces cinnamonensis (4) and IleA90 and GlnA208 in HCM subunit A (HcmA) of Aquincola tertiaricarbonis (6), suggesting that a switch in substrate specificity could be achieved by a few mutations in these enzymes. However, corresponding mutations, e.g. Y89F and R207Q in PfMCM, always led to inactive enzymes (17), indicating that other residues might additionally be involved in determining substrate specificity.
Thus far, crystal structures for studying active site architecture and catalytic mechanism have been revealed for PfMCM (13, 14) and human (mitochondrial) MCM (HsMCM) (18), which have an overall sequence identity of 42 and 39% with HCM, respectively. Structures of other mutases have been deduced by homology modeling based on these MCM structures (3, 4). However, the mutation experiments for changing substrate specificity (17) clearly showed that the current structural models for other mutases are not accurate enough. In particular, specific substrate-enzyme interaction might be controlled by different interactions than those mediated by Tyr89 and Arg207 in PfMCM, e.g. the two methyl groups of the ICM substrate isobutyryl-CoA cannot form H-bonds with the corresponding active site amino acids PheA80 and GlnA198 or any other residue as the free substrate carboxyl group does in MCM. Likewise, the preference for (S)-3-hydroxybutyryl-CoA found in HCM cannot be explained without a reliable structure model (6). In addition, differences in subunit organization (Fig. 2) might also lead to deviations from the thus far studied MCM structures. PfMCM and HsMCM are structurally closely related as coenzyme B12 and acyl-CoA binding sites are located on the same catalytically active subunit. PfMCM is heterodimeric, consisting of one catalytic subunit and a homologous but only regulatory subunit of similar size lacking essential binding domain residues (e.g. Tyr89 and Arg207 for acyl-CoA binding). HsMCM is homodimeric. In contrast to both PfMCM and HsMCM, however, in bacterial ICM and HCM as well as in recently described archaeal MCMs (19, 20), acyl-CoA and coenzyme B12 binding sites are distributed on two subunits of significantly different size (i.e. large and small subunits A and B, respectively).
Here, we present the crystal structure to 2.5 Å resolution of the complex between the two subunits of recombinant HCM (HcmA and HcmB), coenzyme B12, and the substrates (S)-3-hydroxybutyryl- and 2-hydroxyisobutyryl-CoA. In comparison with the structure of MCM, decisive differences in the active site could be identified, enabling the specific interconversion of (S)-3-hydroxybutyryl- and 2-hydroxyisobutyryl-CoA.
2-Hydroxyisobutyric acid (>98%) was purchased from Merck Schuchardt (Hohenbrunn, Germany). Anhydride of isovaleric acid (99%) was from ABCR GmbH & Co. KG (Karlsruhe, Germany). Anhydride of pivalic acid (≥98%), sodium salts of (R)-3-hydroxybutyric (96%) and (S)-3-hydroxybutyric (≥97%) acid, coenzyme B12 (≥97%), and CoA (≥93%) were purchased from Sigma-Aldrich (Steinheim, Germany). Isovaleryl- and pivalyl-CoA were prepared from their anhydrides (21). 2-Hydroxyisobutyryl-CoA, (R)-3-hydroxybutyryl-CoA, and (S)-3-hydroxybutyryl-CoA were synthesized from the free acid forms via thiophenyl esters (22).
Genes encoding for the large and small subunits of HCM from A. tertiaricarbonis L108 (hcmA and hcmB) were cloned into pASG-IBA43 vectors (IBA GmbH, Goettingen, Germany) and chemically transformed into Escherichia coli TOP10. After growth in Luria-Bertani medium containing 100 mg liter−1 ampicillin to an optical density at 550 nm of 0.5, induction was performed with 200 μg liter−1 anhydrotetracycline for 3 h at 30 °C. Cells were harvested by centrifugation and suspended in 100 ml of wash buffer (100 mm Tris, 150 mm NaCl, pH 8.0) for further analysis. Likewise, hcmA mutant genes cloned into pASG-IBA43 were transformed into E. coli TOP10 and expressed as described above. Site-directed mutagenesis of the wild-type hcmA (GeneCust Europe) resulted in HcmA I90V, I90L, and I90A mutants (with the point mutations a268g, a268c, and a268g plus t269c, respectively) and in HcmA D117A and D117V mutants (with the point mutations a350c and a350t, respectively).
For the purification of recombinant mutase subunits used for testing enzyme activities, crude extracts of induced E. coli cells were prepared by mechanical disruption using a mixer mill (MM 400, Retsch GmbH, Haan, Germany) with glass beads (212–300 μm, Sigma) at 30 s−1 for 30 min. The recombinant HcmA and HcmB subunits were purified with the help of their His- and Strep-tags, respectively, as described previously (6). For protein purification of A. tertiaricarbonis HcmA and HcmB used for crystallization, the E. coli cells were resuspended in lysis buffer (20 mm NaH2PO4, 500 mm NaCl, 30 mm imidazole, 5% glycerol, pH 7.5), homogenized, and ultracentrifuged at 48,000 × g for 1 h. The supernatants were loaded onto a HisTrap (GE Healthcare) nickel affinity chromatography column and eluted with 250 mm imidazole. The pooled fractions were applied to a HiLoad 16/60 Superdex 200 prep grade (GE Healthcare) size-exclusion chromatography column using 20 mm Tris, 150 mm NaCl, 2 mm DTT, 5% glycerol, pH 7.5, as buffer and concentrated to 20 mg ml−1.
Crystals were grown in 16% PEG 3350, 0.1 m Tris, pH 7.5, by the hanging-drop vapor diffusion method at 4 °C after mixing 0.7 μl of protein solution with 0.3 μl of microseeds (of previously grown HCM crystals) and 1 μl of reservoir buffer. The protein solution was prepared by mixing the two individual proteins HcmA and HcmB in equimolar ratio to 12 mg ml−1 followed by the addition of 2 mm coenzyme B12 and 6 mm (S)-3-hydroxybutyryl-CoA. Crystals, which appeared after 2 days at 4 °C, were directly frozen into liquid nitrogen after opening the cover glass with the crystallization drop. No cryoprotectant was added, but we noted that crystals from drops that were in contact with air for a prolonged time (~30–60 s between reservoir opening and shock cooling) during crystal harvesting resulted in better cryoprotection and diffraction, most likely due to partial water evaporation. Diffraction data were collected to a resolution of 2.5 Å using synchrotron radiation (beamline BL14.1 at Helmholtz-Zentrum Berlin (HZB) BESSY II in Berlin, Germany). Data processing was accomplished using XDS (23) for integration and SCALA (24) for scaling within the CCP4 program package. With unit cell dimensions a = 69.0 Å, b = 119.6 Å, and c = 173.9 Å, the crystal belongs to space group C2 and contains four protein chains per asymmetric unit (two HcmA chains and two HcmB chains). Details of data collection and refinement are summarized in Table 1. The structure was solved by molecular replacement using the program Molrep (25) and chain A of Protein Data Bank (PDB) entry 4REQ (14) as the search model. Model building was performed using Coot (26), and the structure was refined with BUSTER (27). The programs MolProbity (28) and Procheck (29) were used to evaluate the final model, and PyMOL (36) was used for the visualization of the protein structure. Amino acids AsnA169 and AlaA199 from both peptide chains within the asymmetric unit lie in the generously allowed region of the Ramachandran plot, but both residues exhibit well defined electron density.
Isomerization activity was assayed by directly measuring product formation employing an HPLC ion-pair chromatography system as described previously (6). Briefly, equimolar amounts (3 μm) of recombinant HcmA and HcmB were incubated in 50 mm potassium phosphate buffer (pH 6.6) amended with 833 μm coenzyme B12, 10 mm MgCl2, and 10% glycerol at 30 °C in the dark. After a 5-min preincubation, the reaction was started by adding acyl-CoA substrate. For stopping the reaction, samples were mixed with an equal volume of 100 mm acetate buffer (pH 3.5) and heated at 60 °C for 5 min prior to HPLC analysis. Kinetic parameters were calculated by nonlinear regression analysis applying the Michaelis-Menten equation (OriginPro 9).
The structure of HCM in complex with coenzyme B12 and the substrates (S)-3-hydroxybutyryl- and 2-hydroxyisobutyryl-CoA has been determined to 2.5 Å resolution (Fig. 3). In contrast to the structure of HsMCM and PfMCM in which two domains are connected by a long linker, HCM consists of two different subunits representing the MCM domains, HcmA for the N-terminal TIM barrel domain and HcmB for the C-terminal Rossmann fold domain (13,–15, 18). Furthermore, HCM forms an α2β2-heterotetramer, in which the interface between the two protomers is formed entirely by HcmA (Fig. 3A). HCM and MCM share similar protein architectures with root mean square deviation values of 1.34/0.98 and 1.52/0.85 Å for the two HCM subunits to catalytically active subunits of PfMCM and HsMCM, respectively. In agreement, the active monomer from the heterodimeric PfMCM (Fig. 3B) and a monomer of the homodimeric HsMCM (Fig. 3C) fit well to the αβ-subunit assembly in HCM. In HCM, both αβ-protomers are functional (concerning the binding of cofactors and substrates in a catalytically competent arrangement), and the only difference between HCM and the MCM structures is the absence of the interdomain regions. A sequence alignment of HCM with PfMCM and HsMCM shows that they share sequence identities of 44 and 41% with HcmA and 43 and 42% with HcmB, respectively (see also sequence alignment of HCM with PfMCM and HsMCM including secondary structure elements in Fig. 4).
The active site cavity of HcmA is formed by the β-strands of the TIM barrel domain and is closed by HcmB with the bound cobalt-corrin ring (Fig. 3A). For co-crystallization, (S)-3-hydroxybutyryl-CoA was added to the enzyme, but (S)-3-hydroxybutyryl-CoA as well as the product 2-hydroxyisobutyryl-CoA were included in the final model because the electron density indicates alternative binding of the two substrates in the active site in approximately equal occupancy (Fig. 5A). A similar situation of a mixture of both substrates (or substrate/product pair) bound to the active site was observed in the PfMCM structure (15). The carbonyl groups of both compounds form H-bonds to HisA245, which is homologous to His244 in PfMCM (Fig. 5B). In PfMCM, Tyr89 plays a central role in the reaction mechanism by forming an H-bond to the carboxyl groups of the ligands, and it has been suggested to be involved in the generation of the 5′-deoxyadenosyl radical by displacement of 5′-deoxyadenosine from the cobalt atom (14, 15, 18). In HCM, Tyr89 is exchanged by IleA90, which is smaller than the tyrosine side chain and is not capable of forming H-bonds to the ligands. Instead, AspA117 replaces the free space and forms H-bonds with the hydroxyl group of both HCM substrates. AspA117 and IleA90 most likely displace the 5′-deoxyadenosyl from cobalt together with the bound substrate after HcmA/HcmB assembly. Similar to the situation in MCM, these residues would clash with the adenosyl group if the cofactor would be bound as in the open substrate-free enzyme form (14) (Fig. 5, B and C). The homologous residue in PfMCM is Ala116, which is unable to form a similar interaction. In PfMCM, the carboxyl groups of methylmalonyl- and succinyl-CoA form salt bridges with Arg207 (Fig. 5C) (15). In HCM, this residue is replaced by GlnA208, which is too short for H-bond interactions with (S)-3-hydroxybutyryl- or 2-hydroxyisobutyryl-CoA (Fig. 5C).
In the free coenzyme B12, the dimethylbenzimidazole is coordinated to the cobalt ion. As in the complex structures of PfMCM (13), this group is flipped into a deep pocket of HcmB and HisB18 is coordinated to the cobalt ion at a distance of 2.5 Å. This long coordinative bond is believed to weaken the cobalt-C bond of the other distal ligand (13), and it is stabilized by the protein environment; HisB18 makes H-bonding interactions with AspB16, whose charge is compensated by LysB12 as in MCM. The 5′-deoxyadenosyl ligand is completely displaced from the cobalt atom, and it is bound with an occupancy of 0.5.
As described above, AspA117 of HCM fixes the conformation of the substrates 2-hydroxyisobutyryl- and (S)-3-hydroxybutyryl-CoA by forming H-bonds to their hydroxyl groups. Via this interaction, AspA117 is probably also important for the stereospecificity favoring the interconversion of 2-hydroxyisobutyryl- and the (S)-enantiomer of 3-hydroxybutyryl-CoA. In the observed substrate binding mode, the hydrogen atoms that need to be abstracted for catalytic turnover are well positioned for abstraction. The hydrogens are at 2.3–2.5 Å distance to the C5′ of 5′-deoxyadenosyl, which bears the unpaired electron after dissociation from the cobalt ion (Fig. 5, B and C). In the R-isoform of 3-hydroxybutyryl-CoA, the methyl group would change position with the hydrogen atom if the hydroxyl group maintains the favorable interaction with AspA117, and the hydrogen would not be positioned for abstraction.
Residues IleA90 and AspA117 are likely the most important amino acids for substrate specificity of HCM when compared with MCM. Mutating IleA90 to Tyr or Phe results in a complete loss of HCM activity (6). The HCM crystal structure suggests that these larger side chains would clash with the catalytically competent conformation of AspA117 (Fig. 5C). Surprisingly, also the substitution of IleA90 against the similar amino acid Val dramatically decreased the reaction rate and substrate affinity (6). In line with this, the mutations HcmA I90L and I90A resulted in a significant reduction in conversion rates and 6- to >300-fold diminution of the catalytic efficiencies (Table 2) for the main substrates of the wild-type enzyme, (S)-3-hydroxybutyryl- and 2-hydroxyisobutyryl-CoA. Interestingly, reduction of activity was smallest for the isomerization of 2-hydroxyisobutyryl-CoA with the substitution of the IleA90 by the relatively small Ala, retaining about 50% of the wild-type Vmax. In addition, with (R)-3-hydroxybutyryl-CoA, this mutant showed a 6-fold higher catalytic efficiency than the wild-type enzyme, although the Vmax of 2.6 nmol min−1 mg−1 was still quite low. As predicted from the active site architecture, the nonconservative mutation HcmA D117V resulted in reversion of stereospecificity, now clearly favoring (R)-3-hydroxybutyryl-CoA at a Vmax of 125 nmol min−1 mg−1 that is close to the wild-type rate previously determined for the (S)-enantiomer (Table 2). With the latter substrate, about 3-fold reduced activities were observed, indicating that a catalytically competent orientation is more uncommon but still possible, i.e. positioning of the hydroxyl group of the S-isoform in the neighborhood to ValA117. The HcmA D117A enzyme, on the other hand, showed similar Vmax values of about 20–30 nmol min−1 mg−1 for all three hydroxyacyl-CoA substrates. However, due to the quite low Km value obtained for (R)-3-hydroxybutyryl-CoA, the catalytic efficiency exceeded 200 mm−1 min−1, which is the highest value among all HCM variants and substrates tested (Tables 2 and and33).
As pivalyl-CoA is another potential substrate for coenzyme B12-dependent mutases possessing a similar branching complexity as the HCM substrate 2-hydroxyisobutyryl-CoA, interconversion of this acyl-CoA and isovaleryl-CoA was tested (Table 3). The wild-type HCM could isomerize these substrates at 6–8% of the Vmax value of 140 nmol min−1 mg−1 previously obtained for its most favorite substrate (S)-3-hydroxybutyryl-CoA. In addition, the HcmA I90V and I90A mutant enzymes also isomerized pivalyl- and isovaleryl-CoA at rates close to the wild-type Vmax values. Due to lower Km values, the HcmA I90A mutant even possesses slightly higher catalytic efficiencies than the wild-type enzyme. The HcmA I90L mutant, however, did not show any significant activities with pivalyl- and isovaleryl-CoA. In contrast, conversion rates of pivalyl-CoA by the HcmA D117V enzyme were much higher than observed with the wild-type and HcmA I90 mutant enzymes, showing a Vmax value of about 100 nmol min−1 mg−1.
In HCM, the coenzyme B12- and substrate-binding domains form two separate protein chains in contrast to PfMCM and HsMCM. However, the missing linkage between the two domains does not influence the protein assembly. The substrate-bound structure of HCM reveals significant differences in substrate binding when compared with MCM.
The most significant structural difference between MCM, ICM, and HCM substrates is the type of the substituents attached to the thioester-linked carbon atom (Fig. 1). Aside from the thioester carbonyl carbon, the linked carbon atom forms covalent bonds with methyl and carboxyl carbons in methylmalonyl-CoA, but the fourth substituent is a hydrogen atom. A similar situation is found with isobutyryl-CoA where only the carboxyl group of MCM substrates is replaced by a methyl residue. In contrast, only in 2-hydroxyisobutyryl-CoA the thioester-linked carbon is attached to four non-hydrogen substituents. The structure of HCM reveals that the orientation of the hydroxyl group of HCM substrates corresponds to the hydrogen attached to the thioester-linked carbon of MCM substrates (Fig. 5C). A consequence of this bulkier structure of the HCM substrates is that the active site amino acid Tyr89 of MCM is replaced by the smaller IleA90 in HCM, which is in close neighborhood to the hydroxyl group of the substrate. In line with this, MCM is absolutely specific to (R)-methylmalonyl-CoA, as the methyl group of (S)-methylmalonyl-CoA would have a similar orientation as the hydroxyl residue in the HCM substrates and would clash with the aromatic ring of Tyr89 (15). The demand of additional space for the acyl residue in HCM substrates could explain the complete loss of activity previously observed with HcmA I90Y and I90F mutant forms (6), in addition to the displacement of AspA117, which is important for substrate binding. Surprisingly, even the conservative mutation HcmA I90V caused a dramatic decrease in HCM activity (6). Now, we have demonstrated that HcmA I90A and I90L mutations also resulted in significant reduction in catalytic efficiency for the HCM main substrates 2-hydroxyisobutyryl- and (S)-3-hydroxybutyryl-CoA (Table 2). Possible explanations are steric clashes of the introduced LeuA90 side chain with neighboring HcmA residues resulting in a repositioning of some protein residues important for the catalytic cycle and substrate binding and, on the other hand, fewer stabilizing interactions in the case of the smaller side chains of ValA90 and AlaA90. The larger side chain of IleA90 is involved in van der Waals interactions with the bound substrates, and its absence may cause more flexibility within the acyl group of the CoA ester substrates.
In PfMCM, H-bond formation of the substrate carboxyl group with Tyr89 and Arg207 plays a central role in substrate binding, radical formation, and intermediate stabilization (15, 16). By analogy, a similar role can be expected from the hydroxyl group of 2-hydroxyisobutyryl- and 3-hydroxybutyryl-CoA. In fact, AspA117 has been identified in the HCM structure as the important active site residue specifically interacting with the acyl moiety by forming an H-bond with the hydroxyl group of HCM substrates (Fig. 5B). Obviously, this interaction also determines the stereospecificity of HCM. Only in the (S)-enantiomer the hydrogen atom is positioned properly toward the C5′ of 5′-deoxyadenosyl for its abstraction. If the (R)-enantiomer is modeled in a catalytically competent binding mode with the hydrogen atom facing the C5′ of 5′-deoxyadenosyl, the methyl group would form close contacts with AspA117. If the hydroxyl group of the (R)-enantiomer is positioned in a favorable interaction with AspA117, the H-atom is not properly positioned for hydrogen abstraction. Thus, a catalytically competent low energy binding mode is only available for the (S)-enantiomer. Consequently, the R-isoform is converted with a nearly 1000-fold lower catalytic efficiency than (S)-3-hydroxybutyryl-CoA (6). With the mutant HcmA I90A, catalytic efficiency toward the (R)-enantiomer is slightly improved, likely due to increased freedom for the AspA117 conformation to avoid unfavorable close contacts to the methyl group of the substrate. Residue GlnA208 does not seem to play as essential a role in substrate binding as the homologous residue Arg207 in PfMCM. The MCM substrates (R)-methylmalonyl- and succinyl-CoA are not turned over by HCM as Arg207 and Tyr89 are not available for salt bridge and H-bonding interactions with the carboxyl groups (Fig. 5C). The homologous side chain of GlnA208 in HCM is probably too short to replace Arg207 of MCM at least for a direct H-bonding interaction. Consequently, HCM does not catalyze rearrangement of these substrates (6). As expected, the mutation HcmA D117V resulted in reversion of stereospecificity likely due to a now favored orientation of the methyl group of (R)-3-hydroxybutyryl-CoA to ValA117 when compared with a similar positioning of the hydroxyl group of the S-isoform. Nevertheless, the high conversion rates of the mutant enzyme obtained for the R-isoform are somewhat surprising. Considering the orientation of (R)-3-hydroxybutyryl-CoA in the HcmA D117V mutant, GlnA208 would be the only candidate for H-bond interaction, and it might interact with the hydroxyl group via a water molecule. In addition, hydrophobic interactions with IleA90 and ValA117 support substrate binding. In summary, active site amino acids IleA90 and AspA117 determine substrate specificity and catalytic efficiency in HCM. Consequently, both residues are conserved in all HCMs identified thus far, whereas mutases of the other subfamilies possess different amino acids at these positions (Fig. 6).
For the bacterial dissimilation of pivalic acid, another coenzyme B12-dependent mutase has been postulated isomerizing pivalyl-CoA into isovaleryl-CoA (32). Like 2-hydroxyisobutyryl-CoA, pivalyl-CoA possesses four non-hydrogen atoms attached to the thioester-linked carbon. However, in contrast to the former acyl-CoA ester, both R residues depicted in Fig. 1 are methyl groups in the case of pivalyl-CoA. In the HCM structure, placing pivalyl-CoA analogous to 2-hydroxyisobutyryl-CoA would result in unfavorable close interactions between the substrate methyl group and the carboxyl group of AspA117. In line with this, isomerization activity with pivalyl- and isovaleryl-CoA turned out to be low for HCM, reaching only less than 10% of the (S)-3-hydroxybutyryl-CoA conversion rates. However, this observed PCM activity of HCM is quite high when compared with the isovaleryl-CoA isomerization rate recently determined for two IcmF enzymes. Conversion of the PCM substrate isovaleryl-CoA is 60–900-fold lower by IcmF than the ICM activity with isobutyryl-CoA (9). In IcmF, the additional methyl group in pivalyl-CoA likely clashes with Phe589 (numbering as in the IcmF of Geobacillus kaustophilus HTA426, see also Fig. 2), which is homologous to Tyr89 of PfMCM. Consequently, IcmF is a rather specific ICM enzyme and displays only minor activity toward the PCM substrates. From the HCM structure, it could be deduced that a mutase with specific PCM activity would have an HCM-like IleA90 or similar small aliphatic residue but would lack a polar residue homologous to AspA117 of HCM in the substrate binding cavity. Accordingly, the HcmA D117V mutation resulted in a 10-fold increased PCM activity of HCM.
Analysis of the structure of HCM and comparison with the architecture previously found in the MCM subfamily (14, 18) has shed new light on active site architecture and determinants of substrate specificity in this fascinating group of radical enzymes. In the future, tailor-made mutases might be developed for the rearrangement of carboxylic acids relevant for industrial and pharmaceutical production, in particular for the generation or identification of enzymes with efficient HCM, PCM, and ICM activity (7, 33,–35). Detailed structural information of the substrate interaction helps in the rational enzyme design of these industrially relevant biocatalysts.
We thank C. Dilssner (UFZ) and M. Neytschev (UFZ) for excellent technical assistance. In addition, we thank Ulrike Krug (University of Leipzig) for helpful discussions. We thank the Joint Berlin MX Laboratory at the Helmholtz Zentrum Berlin (Bessy II) for beam time and assistance during synchrotron data collection, as well as for traveling support.
5The abbreviations used are: