Crystal structures of Streptomyces coelicolor methylmalonyl-CoA epi- merase with substrate or transition state analog contradicts a simple general acid-base catalytic mechanism Lee M. Stunkard, Aaron B. Benjamin, James B. Bower, Tyler J. Huth and Jeremy R. Lohman* Department of Biochemistry, Purdue University Center for Cancer Research, Purdue University, West Lafayette, Indiana 47907, United States Supporting Information Placeholder ABSTRACT: Isomerases that flip carbon stereocenter chirality intermediary metabolism, but also for the production of polyketides often use straightforward general acid base catalysis where a proton in certain heterologous hosts. is abstracted from one face of a central carbon and a proton is de- Another source of methylmalonyl-CoA for polyketide biosyn- posited on the opposite. That mechanism has been ascribed to thesis is through the activity of malonyl-CoA synthetases such as methylmalonyl-CoA epimerase (MMCE). The proposed mecha- MatB from S. coelicolor or Rhizobium trifolii although they gener- nism is use of concerted general acid-base catalysis similar to the ate (2R)-methylmalonyl-CoA thus also requiring MMCE to sup- proline racemase, based on the exchange of the α-carbon proton port efficient DEBS-like polyketide biosynthesis.5-6 MatB enzymes with solvent. Tritium kinetic isotope effect experiments lead to the also accept substrates other than malonate and methylmalonate hypothesis of a “protected base” in catalysis. Structures of Propi- with poor efficiency,7 although mutants have been generated with onibacterium shermanii MMCE with catalytic Co2+ bound revealed expanded substrate specificity.8 Therefore, an attractive path to that two glutamates were positioned on opposite sides of the active polyketide derivatives is through the engineering of MatB to pro- site, that could fulfil the roles of the general acids. However, in duce malonyl-CoA with novel 2-substitutents combined with acyl- preliminary modeling of the substrate, we noticed the distances and transferases engineered to accept the novel malonyl-CoAs.9-14 geometry were not typical suggesting conformational changes upon However, it is likely the engineered enzymes generate substrates substrate binding. In order to further understand the puzzling rela- that are not directly accepted by the acyltransferases, requiring tionships between the catalytic mechanism, substrate preference spontaneous epimerization which has a half-life of ~90 hours for and structure, we solved MMCE structures with the substrate and a methylmalonyl-CoA at ~ pH 7 and 0 °C.15 Therefore, MMCE en- transition state analog bound. Our structures reveal that there are zymes are likely a necessary addition in the engineered production no acids or bases near the α-carbon for simultaneous deprotona- of polyketides through MatB mutants, supporting further charac- tion/reprotonation. Therefore, we propose two alternatives for terization of MMCE structure-function relationships. MMCE catalysis, one where conformational changes and water in- vasion overcome the long C2-catalytic acid distances observed in to or from O O from branched chain TCA cycle O amino or fatty acid the crystal structures. Alternatively, a typically disallowed 1,3-sig- CoA S 1 CoA S 1 2 2 acid catabolism matrophic hydrogen shift results in enolization, whereby the thioe- O ster ketone or carboxylate acts as the “protected base”. Both mech- carboxy-biotin carboxyl- anisms are compatible with other 2-substituted malonyl-CoAs, as mutase such our structures provide a platform to design mutations for ex- transferase biotin panding substrate scope to support combinatorial biosynthesis. O O O O (R) epimerase (S) 1 1 CoA S 2 O CoA S 2 O Methylmalonyl-CoA epimerase (MMCE) or racemase is central AMP + PP DEBS-like MatB i PKS to linking TCA cycle intermediates with other metabolic pathways ATP in various organisms, including humans.1 Typically, MMCEs are + CoA found with methylmalonyl-CoA mutases, where the mutase revers- O O polyketides or specialized fatty acids ibly converts succinyl-CoA to (2R)-methylmalonyl-CoA (L-config- O O uration) and the epimerase generates (2S)-methylmalonyl-CoA (D- configuration). Some organisms like Escherichia coli and Pseudo- monas putida are lacking a MMCE.2 In order to generate E. coli Figure 1. MMCE is central to the metabolism of methylmalonyl- capable of supporting the biosynthesis of polyketides like erythro- CoA by linking enzymes with different stereochemical preference mycin, Streptomyces coelicolor MMCE (ScMMCE) was intro- or outcomes. duced to link the production of (2R)-methylmalonyl-CoA from the The catalytic mechanism of Propionibacterium shermanii activity of the native methylmalonyl-CoA mutase (Sbm) to the MMCE (PsMMCE) was probed using hydrogen-tritium exchange (2S)-methylmalonyl-CoA specific 6-deoxyerythronolide B syn- assays.15-16 In both the tritium exchange assays and an NMR 3-4 thase (DEBS). This establishes MMCE as important not only in study,17 PsMMCE catalyzes exchange of the C2 hydrogen-isotope with solvent-isotopes, which was the first piece of evidence sug- on opposing sides of the active site.18 Based on modelling of the gesting a general acid-base catalysis mechanism, Figure 2. In the (2R)-epimer, it was predicted that the glutamates acted as the cata- tritium exchange assays, in which only the (2R)→(2S) direction lytic acid-base pair. In order to clarify the roles of the proposed was followed, it was determined the enzyme essentially always ex- catalytic residues and gain insight into enzyme-substrate interac- changes the C2 hydrogen-isotope with solvent during catalysis and tions for expanding substrate scope, we solved the structure of the 2-hydrogen isotope is not exchanged with solvent or “pro- ScMMCE with methylmalonyl-CoA and a putative transition state tected” until product release. However, there was a low kinetic iso- analog, 2-nitropropionyl-CoA, Figure 3. We previously synthe- tope effect reported suggesting other factors rather than chemistry sized the nitro bearing methylmalonyl-CoA analog for examining are rate limiting, such as conformational changes or product re- the structure-function activities of various enzymes. In the active lease. Substrate bound structures of the enzyme could provide in- site of E. coli methylmalonyl-CoA decarboxylase and a bifunc- sight into those other factors. tional acyltransferase/decarboxylase, LnmK, the nitro analog binds 19-20 Structures of PsMMCE in the apo and Co2+ bound holoforms, as a nitronate with a deprotonated C2. We expected 2-nitro- revealed an active site with partially conserved glutamate residues propionyl-CoA to be an ideal transition state analog to reveal inter- actions with the catalytic acids. A M M M M M substrate solvent H2O OH2 H H binding O O deprotonate O O protonate O O exchange O O H H O O CoA S O CoA S O CoA S O 2x H O 2 H CH H 3 CH3 3C H CoA S O A HA AH HA AH A AH A H CH3 B M M M H O H substrate 2 solvent O O O O H2O OH2 binding O O enolize tautomerization O O H O O CoA S O CoA S O CoA S O CoA S O 2x H O 2x H2O 2 H CH CH CH H C H CoA S O 3 3 3 3 H CH3 Figure 2. Two possible catalytic mechanisms for MMCE. A) Previously proposed general acid-base catalysis. B) Enolization via 1,3-H shift with solvent tautomerization. His84 His7 His84 His7 His84 His7 N N N N N N H H H H H H HN NH Gln60 HN NH Gln60 HN NH Gln60 N N N N N N N N N HN H HN H HN H Glu134 O Glu134 O Glu134 O O O O Co Ser115 Co Ser115 Co Ser115 H2N HN H2N HN H2N HN O O O O O O O O ? O 4.0Å 4.1Å O O N HO HO O O H CoA S O CoA S O H H NH O CH O CH O Gln39 2 O O N 3 N 3 HN O O 4.4Å O NH O 4.1Å O NH Glu43 Gln39 Glu43 Gln39 Glu43 N NH2 NH2 H Figure 3. Active sites of ScMMCE as the holoenzyme and ligand bound states with flattened schematics. Hydrogens are shown as diminu- tive ball and sticks. Black dashes represent potential hydrogen bonds or close interactions. Red dashes represent distances to predicted catalytic acids. A) Holoenzyme, bold arrows represent conformational changes necessary for ligand binding. Hydrogens are not shown on the active site waters shown as red “O” in the flattened schematic. B) 2-nitronate-propionyl-CoA (NO2Pr-CoA) is shown as gray sticks. C) Methylmalonyl-CoA is shown as black sticks. Figure 4. Active site of holo-ScMMCE with substrate bound. Hydrogens are shown as diminutive ball and sticks. A) Electron density for the methylmalonyl-thioester portion of methylmalonyl-CoA in the ScMMCE active site displayed as sigma-A weighted 2mFo-DFc maps displayed 1.0 σ (blue mesh) and 2.0 σ (cyan mesh). B) Overlay of methylmalonate-Co2+ complex (CSD JERMAU) on the methylmalonyl- thioester-Co2+ of holo-ScMMCE representing the (2R) configuration. The structure JERMAU has two overlapping configurations, one where the methyl-group and non-Co2+ coordinating oxygens are coplanar designated in cyan, and the other where the methyl group is equatorial designated in magenta. C) Similar to panel B but representing the (2S) configuration. D) Overlay of holo-ScMMCE without substrate and with the methylmalonyl-CoA enol/enolate bound, coloring same as Figure 3. The water indicated by a star from holo-MMCE without substrate bound makes a close contact with C2 of the enol/enolate intermediate and would be activated by Glu134, indicated by red dashes. The conformation of Glu43 from holo-MMCE without substrate configuration also makes a close contact with the substrate C2 shown as red dashes. We solved the structure of holo-ScMMCE in order to compare ScMMCE and the asymmetric unit only has a monomer.