Structure-Guided Expansion of the Substrate Range of Methylmalonyl Coenzyme A Synthetase (MatB) of Rhodopseudomonas palustris

Abstract

Malonyl coenzyme A (malonyl-CoA) and methylmalonyl-CoA are two of the most commonly used extender units for polyketide biosynthesis and are utilized to synthesize a vast array of pharmaceutically relevant products with antibacterial, antiparasitic, anticholesterol, anticancer, antifungal, and immunosuppressive properties. Heterologous hosts used for polyketide production such as Escherichia coli often do not produce significant amounts of methylmalonyl-CoA, however, requiring the introduction of other pathways for the generation of this important building block. Recently, the bacterial malonyl-CoA synthetase class of enzymes has been utilized to generate malonyl-CoA and methylmalonyl-CoA directly from malonate and methylmalonate. We demonstrate that in the purple photosynthetic bacterium Rhodopseudomonas palustris, MatB (RpMatB) acts as a methylmalonyl-CoA synthetase and is required for growth on methylmalonate. We report the apo (1.7-Å resolution) and ATP-bound (2.0-Å resolution) structure and kinetic analysis of RpMatB, which shows similar activities for both malonate and methylmalonate, making it an ideal enzyme for heterologous polyketide biosynthesis. Additionally, rational, structure-based mutagenesis of the active site of RpMatB led to substantially higher activity with ethylmalonate and butylmalonate, demonstrating that this enzyme is a prime target for expanded substrate specificity.

INTRODUCTION

Malonyl-coenzyme A (malonyl-CoA) synthetase (MatB) belongs to the AMP-forming acyl-CoA synthetase protein family (PFAM 00501). These enzymes catalyze the conversion of organic acids to acyl-CoA thioesters via a ping-pong mechanism, in which ATP and the organic acid are first converted to acyl-AMP with the release of pyrophosphate, followed by coenzyme A binding, displacement of AMP, and release of the acyl-CoA product (Fig. 1). This class of enzymes activates a wide range of compounds, including short-, medium-, and long-chain fatty acids, dicarboxylic acids, and aromatic acids, which are then oxidized as a carbon and energy source or used in biosynthetic reactions. AMP-forming acyl-CoA synthetases also share sequence and structural homology with firefly luciferase and the adenylation domains of nonribosomal peptide synthetases (NRPSs), and all belong to the family of enzymes that include acyl-CoA synthetases, NRPS adenylation domains, and luciferase enzymes, also known as the ANL superfamily (5, 23, 43, 61).

Fig 1.

MatB converts malonate to malonyl-CoA in two steps via a malonyl-AMP intermediate.

The active site of acyl-CoA synthetases is located at the interface of the ∼400-amino-acid N-terminal domain and ∼100-amino-acid C-terminal domain. The C-terminal domain rotates ∼140° during the two-step reaction such that the enzyme exhibits one conformation (termed the adenylation conformation) during the reaction of ATP with the organic acid and a second conformation (thioester-forming conformation) during formation of the acyl-CoA product (25, 36, 53, 54; reviewed in reference 23). A short linker region connects the two domains, and most of the rotation during the conformational change occurs at a conserved hinge residue, typically either an aspartic acid or lysine (64).

MatB activates the C₃ dicarboxylic acid malonate to malonyl-CoA (Fig. 1) and has been isolated from a variety of soil bacteria and from mammalian mitochondria (27, 33, 34, 63). The best-studied member of this group is MatB from the nitrogen-fixing legume symbiont Rhizobium leguminosarum bv. trifolii (RlMatB). The gene encoding RlMatB, matB, is located in an operon with a malonyl-CoA decarboxylase gene, matA, and a putative dicarboxylate transporter gene, matC, suggesting that it is involved in malonate metabolism to acetyl-CoA (3).

It has been suggested that growth on malonate is important inside nitrogen-fixing nodules on plant roots. This idea is supported by the fact that a ΔmatB strain of R. leguminosarum displays impaired nodule formation (4). In contrast, matB from Sinorhizobium meliloti 1021 is required for growth on malonate in liquid medium, but a ΔmatB strain did not affect nodulation of alfalfa or nitrogen fixation (13). While matB homologues are found in a number of other nonsymbiotic soil-dwelling and marine bacteria, their functions in the cell have not been characterized.

MatB is of commercial interest for its utility in generating malonyl-CoA and methylmalonyl-CoA starter units for polyketide synthesis (for a review, see reference 12). Polyketides are pharmaceutically relevant natural products that are utilized as antibiotics, antiparasitics, antifungals, cancer treatments, and immunosuppressives, and malonyl-CoA and methylmalonyl-CoA are the two most common extender units in the synthesis of these products. Malonyl-CoA can be produced either from acetyl-CoA by acetyl-CoA carboxylase or directly from malonate by malonyl-CoA synthetase (MatB) (32, 60). Synthesis of polyketide antibiotics in heterologous hosts like Escherichia coli often requires introduction of a pathway for production of methylmalonyl-CoA, through either carboxylation of propionyl-CoA, conversion from succinyl-CoA by methylmalonyl-CoA mutase, or production directly from methylmalonate using MatB (20, 47). Total enzymatic synthesis of several polyketide natural products has also been reconstituted in vitro, and in these cases MatB serves as an affordable means of generating malonyl-CoA and methylmalonyl-CoA in situ (14, 41).

RlMatB is often utilized for polyketide synthesis, as it has been kinetically characterized and can use methylmalonate and other malonate derivatives. However, RlMatB activity with methylmalonate is only ∼20% of its activity with malonate (3, 41, 50). It was reported that Streptomyces coelicolor MatB (ScMatB) could synthesize 15 different polyketide extender units, encompassing all of the combinations of five different malonate derivatives and three different thiol acceptors (CoA, pantetheine, and N-acetylcysteamine), which could be used for in vitro polyketide synthesis (27), although the kinetics of this enzyme were not reported. This opened up the possibility that characterization of MatB homologues from different organisms may lead to expanded substrate specificity and improved activity for different malonate derivatives. Additionally, detailed kinetic and structural studies would provide a template for rational mutagenesis to further expand the substrate specificity of MatB.

In this study, we have characterized the MatB protein from the purple photosynthetic bacterium Rhodopseudomonas palustris (RpMatB), and we show that it is most likely a methylmalonyl-CoA synthetase in vivo. In support of its biological role, RpMatB can efficiently use methylmalonate and has significant activity with ethylmalonate. To further expand the substrate specificity of RpMatB, we determined both the apo and ATP-bound crystal structures of RpMatB and utilized these two structures combined with a thioester-forming conformation structure of ScMatB (27) as a homology model to identify residue T208 in the binding pocket that influences the binding affinity of malonate derivatives. Two RpMatB variants, RpMatB^T208A and RpMatB^T208G, had increased activity with the larger malonate derivatives ethylmalonate and butylmalonate. The data presented here suggest that RpMatB is a good template for future mutagenesis studies to generate unique polyketide extender units.

MATERIALS AND METHODS

Strains, culture media, and growth conditions.

Bacterial plasmids and strains used in this work are listed in Table 1. E. coli strains were grown at 37°C on LB medium (8, 9). Rhizobium leguminosarum bv. trifolii ATCC 14479 was obtained from the American Type Culture Collection and cultured at 30°C on TY medium (5 g liter⁻¹ tryptone peptone, 3 g liter⁻¹ yeast extract, 10 mM CaCl₂). Rhodopseudomonas palustris CGA009 was grown at 30°C on defined basal photosynthetic medium (30) supplemented with 150 nM vitamin B₁₂ and either succinate (10 mM), malonate (25 mM), or methylmalonate (10 mM). When necessary, growth media were supplemented with ampicillin (100 μg ml⁻¹) or kanamycin (50 μg ml⁻¹ for E. coli and 75 μg ml⁻¹ for R. palustris). For growth curves, R. palustris starter cultures were grown for 3 to 4 days photoheterotrophically on photosynthetic medium containing succinate and kanamycin in Balch tubes (7) flushed with O₂-free N₂. Strains were diluted 1:16 in fresh medium supplemented with the appropriate carbon source and kanamycin. Tubes were incubated at 30°C in light without shaking, and growth was monitored at 660 nm using a Spectronic Spec20D. Three biological replicates of each strain were monitored, and each experiment was repeated three times.

Table 1.

Strains and plasmids used in this work

Strain or plasmid	Genotype and description	Reference or sourcea
Strains
R. palustris CGA009		43
Derivatives of R. palustris
JE13366	ΔmatB
JE11529	WTb/pBBR1MCS-2
JE16618	WT/pBBR1MCS-5
JE13568	ΔmatB/pBBR1MCS-2
JE13569	ΔmatB/pRpMatB23
JE13911	ΔmatB/pRlMatB1
Rhizobium leguminosarum bv. trifolii ATCC 14479		ATCC
E. coli
JE8125	S17-1 recA pro hsdR RP4-2-Tc::Mu-kan+::Tn7	62
JE6090	C41λ(DE3)
JE9314	C41λ(DE3) pka12::kan+	Lab collection
Plasmids
pK18mobsacB	Deletion construction in R. palustris, kan+	60
pBBR1MCS-2	Broad host range complementation, kan+	40
pTEV5	Overproduction vector, N-terminal His6 tag	59
pRpMatB4	R. palustris matB deletion construct in pK18mobsacB
pRpMatB23	R. palustris matB in pBBR1MCS-2, kan+
pRpMatB1	R. palustris matB in pTEV5, bla+
pRpMatB39	R. palustris matB K488A in pTEV5, bla+
pRpMatB60	R. palustris matB T208A in pTEV5, bla+
pRpMatB61	R. palustris matB T208G in pTEV5, bla+
pRlMatB1	R. leguminosarum matB in pBBR1MCS-2, kan+
pRlMatB3	R. leguminosarum matB in pTEV5, bla+

^aStrains were constructed for this study unless otherwise indicated.

^bWT, wild type.

Genetic and recombinant DNA techniques.

An in-frame deletion of matB (locus tag rpa0221) was constructed using the method of Schäfer et al. (56), as described elsewhere (17). Primers used to amplify matB-flanking regions for cloning into pK18mobsacB are listed in Table 2.

Table 2.

Primer sequences used in this work

Primer name	Sequencea
MatB_delA_5′EcoRI	GAT GAA TTC CGA TTT CGT TAC TGC CGA AT
MatB_delB	TTA CTT GTA GAT GTC CTT GTA GGT GGC GAA CAG GTT GGC GTT CAT GTC
MatB_delC	GAC ATG AAC GCC AAC CTG TTC GCC ACC TAC AAG GAC ATC TAC AAG TAA
MatB_delD_3′HindIII	GTA AAG CTT AGC GGC ATT CGG TTA CTT C
MatB_5′HindIII	GTA AAG CTT ATG AAC GCC AAC CTG TTC GC
MatB_3′BamHI	CCT GGA TCC TTA CTT GTA GAT GTC CTT GTA GGT
MatB_5′NheI	GTA GAC GCT AGC ATG AAC GCC AAC CTG TTC GC
MatB_K488A_fwd	GCG CAA CAC CAT GGG TGC GGT CCA GAA GAA CGT C
MatB_K488A_rev	GAC GTT CTT CTG GAC CGC ACC CAT GGT GTT GCG C
MatB_T208A_fwd	GCC GAT CTA TCA CGC CCA TGG ATT GTT C
MatB_T208A_rev	GAA CAA TCC ATG GGC GTG ATA GAT CGG C
MatB_T208G_fwd	CTG CCG ATC TAT CAC GGC CAT GGA TTG TTC G
MatB_T208G_rev	CGA ACA ATC CAT GGC CGT GAT AGA TCG GCA G
Trifolii_MatB_5′NdeI	GTA CAT ATG AGC AAC CAT CTT TTC GA
Trifolii_MatB_5′HindIII	GTA AAG CTT ATG AGC AAC CAT CTT TTC GA
Trifolii_MatB_3′BamHI	CAT GGA TCC TTA CGT CCT GGT ATA AAG AT

^aUnderlined sequences identify restriction sites.

To construct complementation vectors, the matB gene was amplified from R. palustris genomic DNA using primers MatB_5′HindIII and MatB_3′BamHI (Table 2), cut with HindIII and BamHI (Fermentas), and ligated into plasmid pBBR1MCS-2 (37) to generate pRpMatB23. The R. leguminosarum matB gene was amplified with primers Trifolii_MatB_5′HindIII and Trifolii_MatB_3′BamHI (Table 2) from an R. leguminosarum colony and ligated into the HindIII and BamHI sites of pBBR1MCS-2. Gene sequences were confirmed using BigDye (ABI PRISM) protocols, and sequencing reactions were resolved and analyzed at the University of Wisconsin Biotechnology Center. The R. leguminosarum matB sequence matched the published sequence (40) deposited in the NCBI database under locus AF117694. Plasmids were introduced into R. palustris by electroporation or conjugation, using E. coli S17-1 as a donor strain (28, 58).

To construct overexpression vectors, the R. palustris matB gene was amplified from genomic DNA using primers MatB_5′NheI and MatB_3′BamHI (Table 2), cut with NheI and BamHI, and ligated into plasmid pTEV5 (55). The R. leguminosarum matB gene was amplified with primers Trifolii_MatB_5′NdeI and Trifolii_MatB_3′BamHI (Table 2) using pRlMatB1 as a template, and the PCR product was ligated into the NdeI and BamHI sites of pTEV5. Both plasmids directed the synthesis of RpMatB or RlMatB with an N-terminal His₆ tag that was removed using recombinant tobacco etch virus (rTEV) protease (10). Site-directed mutants were constructed using the QuikChange protocol (Stratagene) using primers listed in Table 2 and pRpMatB1 as a template.

Protein expression and purification.

Plasmids encoding wild-type and variant MatB proteins were transformed into strain JE9314, a Pka (formerly YfiQ)-deficient derivative of E. coli C41λ(DE3) (46). The resulting strains were grown to early stationary phase and subcultured 1:100 (vol/vol) in 2 liters of lysogenic broth supplemented with ampicillin (100 μg ml⁻¹). Cultures were grown with shaking to an optical density at 600 nm (OD₆₀₀) of ∼0.6, and gene expression was induced by the addition of isopropyl-1-thio-β-d-galactopyranoside (IPTG; 0.5 mM). Cultures were grown overnight at 30°C, cells were harvested by centrifugation at 8,000 × g for 12 min in a Beckman Coulter Avanti J-20 XOI refrigerated centrifuge with a JLA-8.1000 rotor, and cell pellets were frozen at −80°C until used.

To purify MatB, the pellet was resuspended in 30 ml of binding buffer (sodium phosphate buffer [50 mM, pH 8.0], NaCl [300 mM], imidazole [10 mM]) containing lysozyme (1 mg ml⁻¹). Cells were lysed using a sonic dismembrator (Fisher Scientific) at power level 9, with 2-s pulses separated by 5-s breaks, for 1.5 min on ice. Debris was removed by centrifugation at 39,000 × g for 30 min at 4°C, and the soluble fraction was passed through a 0.45-μm filter (Thermo Scientific). His₆-MatB was purified by Ni affinity purification using a 1.5-ml bed volume of Ni-nitrilotriacetic acid (Ni-NTA) Superflow resin (Qiagen). After binding to the column, the column was washed with wash buffer (sodium phosphate buffer [50 mM, pH 8.0], NaCl [300 mM], imidazole [20 mM]), and MatB was eluted with elution buffer (sodium phosphate buffer [50 mM, pH 8.0], NaCl [300 mM], imidazole [250 mM]).

Fractions containing His₆-MatB were pooled, and the tag was cleaved with rTEV protease (1:50 to 1:100 rTEV/H₆-MatB ratio) for 3 h at room temperature. The cleaved protein was dialyzed against buffer 1 (sodium phosphate buffer [50 mM, pH 8.0], NaCl [300 mM], EDTA [0.5 mM]), then buffer 2 (sodium phosphate buffer [50 mM, pH 8.0], NaCl [300 mM]), and finally binding buffer. The protein was passed over an Ni column again to remove the tag and His₇-tagged rTEV protease, using the buffers employed in the first purification step. Untagged proteins eluted in the flowthrough and were pooled and dialyzed against storage buffer 1 (Tris-HCl [50 mM, pH 7.5], NaCl [100 mM], EDTA [0.5 mM], glycerol [20%, vol/vol]) and then storage buffer 2 (Tris-HCl [50 mM, pH 7.5], NaCl [100 mM], glycerol [20%, vol/vol]) before being drop frozen into liquid N₂ and stored at −80°C until used.

Proteins were purified for crystallography in the same manner, except that prior to freezing they were dialyzed against storage buffer 3 (Tris-HCl [10 mM, pH 7.5], NaCl [100 mM], EDTA [0.5 mM]), then storage buffer 4 (Tris-HCl [10 mM, pH 7.5], NaCl [20 mM]), and finally storage buffer 5 [Tris-HCl (10 mM, pH 7.5), tris(2-carboxyethyl)phosphine (TCEP; 1 mM)]. Proteins were concentrated to 10 to 12 mg ml⁻¹ using Ultracel centrifugal filters with a molecular weight cutoff (MWCO) of 10,000 (Amicon Ultra) and drop frozen into liquid N₂. Protein concentrations were determined using a NanoDrop 1000 spectrophotometer (Fisher) and A₂₈₀ molar extinction coefficients for each protein, which were obtained from the Integrated Microbial Genomes (IMG) database (42).

Selenomethionine (Se-Met)-labeled MatB was prepared as follows. Plasmid pRpMatB1 was transformed into strain JE9314 [C41(λDE3) pka::kan], and a 60-ml culture of the resulting strain was grown overnight in LB medium. The culture was harvested by centrifugation and washed twice with M9 minimal medium. The washed cells were used to inoculate 2 liters of M9 glucose medium and grown to an OD₆₀₀ of ∼1. The culture was cooled on ice to ∼16°C for 10 min, and a defined amino acid mixture containing selenomethionine (50 mg liter⁻¹) was added to suppress methionine biosynthesis (62). The culture was grown at 30°C for 30 min before the addition of IPTG (1 mM) to induce matB expression. The culture was grown overnight at 30°C, and MatB protein was purified as described above.

Variant protein MatB^K488A was purified as described above and frozen at −80°C. A sample containing ∼40 mg of protein was thawed and dialyzed against HEPES buffer (50 mM, pH 7.5) to remove any traces of Tris-HCl buffer. The lysine side chains of MatB were reductively methylated using formaldehyde and dimethylamine borane complex as described previously (51). The reaction was quenched by addition of Tris-HCl buffer (pH 7.5) to a final concentration of 500 mM, and the methylated MatB was dialyzed overnight against 2 liters of Tris-HCl buffer (50 mM, pH 7.5) containing dithiothreitol (DTT; 2 mM), before dialysis into storage buffer 5.

Size exclusion chromatography was performed using an ÄKTA fast-protein liquid chromatography Explorer system equipped with a Superdex 200 column (GE Healthcare). The running buffer contained Tris-HCl (50 mM, pH 7.5) and NaCl (100 mM). RpMatB (0.6 to 2.3 mg) was applied onto the column, and the column was developed isocratically at a rate of 0.5 ml min⁻¹. A plot of the log of the molecular mass versus the retention time of thyroglobulin (670 kDa), gamma globulin (158 kDa), ovalbumin (44 kDa), and myoglobin (17 kDa) (R² = 0.999) was used to estimate the apparent oligomeric state of RpMatB.

RpMatB activity assays and kinetic parameters.

MatB specific activity was quantified using an NADH consumption assay (53). All chemicals were purchased from Sigma. Reaction mixtures contained HEPES buffer (50 mM, pH 7.5), TCEP (1 mM), ATP (2.5 mM), free CoA (CoASH; 0.5 mM), MgCl₂ (5 mM), phosphoenolpyruvate (3 mM), NADH (0.1 mM), pyruvate kinase (1 U), myokinase (5 U), lactate dehydrogenase (1.5 U), and organic acid substrate (2 mM). All reactions were started by the addition of MatB (3 pmol per 100-μl reaction mixture), and the change in the absorbance at 340 nm was monitored for 8 min in a 96-well plate format using a Spectramax Plus UV-visible spectrophotometer (Molecular Devices). For kinetic measurements, the same assay was used, except that the concentration of one substrate was varied while the others were held constant at a saturating concentration. Organic-acid concentrations ranged from 0.01 to 2 mM (malonate and methylmalonate), 0.01 to 10 mM (ethylmalonate), and 0.2 to 20 mM (butylmalonate), while ATP was held constant at 2.5 mM. Kinetic parameters for ATP were determined using a concentration range of 0.01 to 1 mM, while malonate (RpMatB and RlMatB) or ethylmalonate (RpMatB^T208A and RpMatB^T208G) was held constant at 3 mM. Pseudo-first order kinetic parameters were determined using Prism version 4.0 software (GraphPad Software), fitting the data to the Michaelis-Menten equation. Each point was measured in triplicate, and the entire experiment was repeated on a separate day. Representative results from one experiment are reported.

Isothermal titration calorimetry.

Proteins were dialyzed into Tris-HCl buffer (50 mM, pH 7.5, 22°C) containing MgCl₂ (10 mM). MatB was used at a concentration of 68 μM, and MatB^K488A was diluted to 21 μM (for ATP binding) or 81 μM (ADP binding). ATP (0.4 mM for RpMatB^K488A or 1 mM for RpMatB) and ADP (1 to 1.5 mM) were dissolved in spent dialysis buffer. Experiments were conducted at 22°C and consisted of 28 10-μl injections into a 1.45-ml sample cell containing RpMatB or RpMatB^K488A; the titration of ATP into RpMatB^K488A consisted of 34 5-μl injections. Injections were made over 20 s (10 s for ATP titration into RpMatB^K488A) and were separated by 3 min. The sample cell was stirred at 307 rpm. Data were acquired on a MicroCal VP-ITC microcalorimeter and analyzed using ORIGIN, version 7.0383, by fitting to the single-set-of-sites model. The change in free energy (ΔG) was calculated using the free-energy equation ΔG = −RTlnK_a, where R is the ideal gas constant, T is temperature (Kelvin), and K_a is the association constant. Each binding isotherm was measured in duplicate, and values for the number of binding sites, association constant, enthalpy, entropy, and free energy changes (N, K_a, ΔH, ΔS, and ΔG, respectively) were averaged.

X-ray crystallography.

Selenomethionine and native apo RpMatB crystals were grown by mixing 2 μl of 10-mg/ml protein in 10 mM Tris (pH 7.5), 1 mM TCEP, 4 mM CoA, 5 mM AMP, and 5 mM sodium methylmalonate with 2 μl of reservoir solution (100 mM bis-Tris [pH 6.5], 84 mM MgSO₄, 19.5% [wt/vol] monomethyl polyethylene glycol 2000) and hanging the resulting mixture over 500 μl of reservoir solution in Linbro plates at 25°C. After 24 h, crystals were nucleated by streak seeding with previously grown crystals and allowed to grow for an additional week. For freezing, crystals were transferred to 100 μl of reservoir solution with the addition of 1 mM CoA, 2.5 mM AMP, and 2.5 mM sodium methylmalonate and then transferred stepwise to a solution of 20% (wt/vol) polyethylene glycol 8000, 20% (wt/vol) glycerol, 100 mM bis-Tris (pH 6.5), 40 mM MgSO₄, 2 mM CoA, 2.5 mM AMP, and 2.5 mM sodium methylmalonate. Crystals were flash-frozen in liquid nitrogen.

Crystals of methylated RpMatB^K488A in complex with ATP crystals were grown by mixing 2 μl of 10-mg/ml protein in Tris buffer (10 mM, pH 7.5) containing TCEP (1 mM) and MgATP (10 mM) with 2 μl of reservoir solution (bis-Tris buffer [100 mM, pH 6.5]) containing trimethylammonium chloride (25 mM), polyethylene glycol 8000 (19.5%, wt/vol), and glycerol (5%, wt/vol) and hanging the resulting mixture over 500 μl of reservoir solution in Linbro plates at 25°C. After 24 h, crystals were nucleated by streak seeding with previously grown crystals and allowed to grow for an additional week. For freezing, crystals were transferred to 100 μl of reservoir solution with the addition of MgATP (10 mM) and then transferred stepwise to a solution of polyethylene glycol 8000 (25%, wt/vol), glycerol (20%, wt/vol), bis-Tris buffer (100 mM, pH 6.5), trimethylammonium chloride (100 mM), and MgATP (10 mM). Crystals were flash-frozen in liquid nitrogen.

X-ray diffraction data for all crystals were collected at the SBC 19-BM beam line (Advanced Photon Source, Argonne, IL). The data sets were integrated and scaled with the program HKL2000 (47a). X-ray data collection statistics are given in Table 3.

Table 3.

Crystal structure data collection and refinement statistics

Statistic(s)	MatB	MatB-ATP
Data collection
Space group	P21	P61
Cell dimensions
a,b,c (Å)	133.9, 58.8, 139.0	173.1, 173.1, 47.7
a, b, g (°)	90, 91.7, 90	90, 90, 120
Wavelength	0.98	0.98
Resolution (Å)a	25–1.7 (1.74–1.7)	25–2.0 (2.03–2.0)
Rmergea	7.6 (48.3)	4.7 (45.4)
I/σIa	16.8 (3.5)	31.4 (7.3)
Completeness (%)a	97.1 (95.5)	99.5 (100)
Redundancya	4.2 (3.8)	12.5 (10.8)
Refinement
Resolution (Å)a	25–1.7 (1.74–1.7)	25–2.0 (2.03–2.0)
No. of reflectionsb	225,397 (11,309)	53,454 (2,718)
Rwork/Rfreec	0.16/0.20	0.16/0.20
No. of atoms
Protein	15,413	3,927
Water	2,607	619
Average B-factor (Å2)	17.6	26.7
Ramachandran (%)
Most favored	98.9	99.2
Allowed	1.1	0.8
Disallowed	0	0
RMSDs
Bond lengths (Å)	0.006	0.007
Bond angles (°)	1.008	1.052
PDB accession no.	4FUQ	4FUT

^aData in parentheses represent highest resolution shell.

^bData in parentheses represent the number of reflections used during refinement.

^cR_factor is determined as R_factor = ∑|F_obs − F_calc|/∑|F_obs|, where R_work refers to the R_factor for the data utilized in the refinement, R_free refers to the R_factor for 5% of the data that were excluded from the refinement, F_obs refers to the observed structure factor amplitude, and F_calc refers to the calculated structure amplitude from the corresponding model.

The structure of apo MatB was solved by single-wavelength anomalous dispersion. The positions of the 52 selenium atoms were determined using the Hybrid Substructure Search submodule of the Phenix package (2, 22) and refined using the programs BP3 (48, 49) and Solomon (1). Parrot (16) was used for density modification, and these phases were used to build an initial model in Buccaneer (15). This was followed by iterative cycles of manual model building in Coot (19) and restrained and translation, libration, and screw motion (TLS) refinement in Refmac 5.6 and Phenix.refine:1.6_289 (2, 59). Data processing and refinement statistics are presented in Table 3.

The structure of methylated MatB^K488A with ATP was solved by molecular replacement using the program Phaser (11, 44) and the apo MatB structure presented here as the search model. This was followed by iterative cycles of manual model building in Coot (19) and restrained and refinement in Refmac 5.6 and Phenix.refine:1.6_289 (2, 59). Data processing and refinement statistics are presented in Table 3. All structural alignments were done with the program Superpose (38).

PDB accession numbers.

Structures have been deposited with the Protein Data Bank (PDB) under accession numbers 4FUQ and 4FUT for apo MatB and ATP-MatB, respectively.

RESULTS

matB is required for growth of R. palustris on methylmalonate.

In the soil alphaproteobacterium Rhizobium leguminosarum bv. trifolii, MatB catalyzes the conversion of malonate to malonyl-CoA, which is subsequently decarboxylated to yield acetyl-CoA (3). To test whether matB function was also required for growth of the purple photosynthetic alphaproteobacterium R. palustris on malonate, we constructed an in-frame deletion of matB. The ΔmatB strain showed a modest but reproducible growth defect on malonate (25 mM) (Fig. 2A, empty triangles, 214-h doubling time), which was corrected by ectopic expression of R. palustris matB⁺ (Fig. 2A, filled inverted triangles, 128-h doubling time) or R. leguminosarum matB (Fig. 2A, filled diamonds, 118-h doubling time); the doubling time of the wild-type strain carrying the empty expression vector on malonate was 116 h. Notably, the doubling time of the wild-type strain on methylmalonate was half of that on malonate (64 h), but the R. palustris ΔmatB strain struggled to grow on methylmalonate (426-h doubling time). Collectively, these data indicated that matB function was required for growth of R. palustris on methylmalonate but not on malonate. These data suggested that MatB most likely functions as a methylmalonyl-CoA synthetase in R. palustris.

Fig 2.

matB function is needed for R. palustris growth on methylmalonate. Wild-type and ΔmatB strains of R. palustris were grown photoheterotrophically on the indicated carbon sources: malonate (25 mM) (A), methylmalonate (10 mM) (B), and succinate (10 mM) (C). The symbols identify wild-type cells carrying the cloning vector (empty squares), the ΔmatB strain carrying the cloning vector (empty triangles), the vector containing R. palustris matB⁺ (filled inverted triangles), and the vector containing R. leguminosarum matB⁺ (filled diamonds). Absorbance was monitored at 660 nm.

The effect of the absence of MatB appeared to be limited to malonate and methylmalonate, since growth of the R. palustris wild type was indistinguishable from that of the ΔmatB strain on succinate (Fig. 2C, empty triangles and empty squares, 13 h each), and full density was already reached at 40 h of incubation.

Substrate specificity of RpMatB.

We purified RpMatB and used a coupled spectrophotometric assay to test its activity with a variety of short-chain mono- and dicarboxylic acids (Table 4). The highest specific activity was seen with malonate (17.7 ± 1.8 μmol AMP min⁻¹ mg⁻¹), although RpMatB also used methylmalonate and ethylmalonate well (specific activities, 11.3 ± 1.4 and 7.2 ± 0.5 μmol AMP min⁻¹ mg⁻¹, respectively). Since RpMatB could use malonate derivatives with one- and two-carbon substitutions in the C-2 position, we tested its ability to use butylmalonate, a malonyl moiety with a four-carbon chain at the C-2 position; surprisingly, it retained 4% of the activity measured with malonate (specific activity, 0.8 ± 0.1 μmol AMP min⁻¹ mg⁻¹). In contrast, RpMatB had very low activity with the C₄ dicarboxylic acid succinate and all of the monocarboxylic acids tested, indicating that the enzyme was specific for dicarboxylic acids with a three-carbon backbone. RpMatB also had low activity with monocarboxylic acids such as acetoacetate (C₄), β-hydroxybutyrate (C₄), and isovalerate (C₅).

Table 4.

R. palustris MatB specific activity with a variety of substrates

Enzyme	Substrate	Sp acta	% activity
RpMatB	Malonate	18 ± 2	100
	Methylmalonate	11 ± 1	61
	Ethylmalonate	7 ± 0.5	39
	Butylmalonate	0.8 ± 0.1	4.4
RpMatBK488A	Malonate	0.04 ± 0.01	0.2
	Methylmalonate	0.03 ± 0.01	<0.2

^aSpecific activity is reported as μmol AMP min⁻¹ mg⁻¹. Each compound was tested at a concentration of 2 mM. The specific activity of the enzyme when succinate, propionate, acetate, butyrate, acetoacetate, isobutyrate, isovalerate, β-hydroxybutyrate, or 2-methylsuccinate was used as substrate was <0.03 μmol AMP min⁻¹ mg⁻¹.

The conserved lysine residue K488 is known to be required for catalyzing the adenylation half-reaction in many acyl-CoA synthetases (26). Hence, we constructed an RpMatB^K488A variant to test whether this residue was important for RpMatB activity. As expected, activity of the RpMatB^K488A enzyme was affected strongly (0.04 ± 0.01 μmol AMP min⁻¹ mg⁻¹ for malonate and 0.03 ± 0.01 μmol AMP min⁻¹ mg⁻¹ for methylmalonate), representing an ∼400-fold reduction in activity compared to that of the wild-type enzyme (Table 4).

We investigated further the use of substituted derivatives of malonate by RpMatB. We determined kinetic parameters for malonate, methylmalonate, ethylmalonate, butylmalonate, and ATP (Table 5). RpMatB had similar K_m values for malonate and methylmalonate (110 ± 8 μM and 81 ± 6 μM, respectively), and although the k_cat for malonate was ∼50% higher (18 ± 0.3 s⁻¹ for malonate and 12 ± 0.2 s⁻¹ for methylmalonate), the catalytic efficiencies of the enzyme (k_cat/K_m) were similar for malonate and methylmalonate (1.64 × 10⁵ M⁻¹ s⁻¹ and 1.48 × 10⁵ M⁻¹ s⁻¹, respectively). Notably, the K_m values for ethylmalonate and butylmalonate were much higher, at 1,250 ± 55 μM and 19,460 ± 1,400 μM, respectively (Table 5). Unsurprisingly, the enzyme was very inefficient; i.e., the k_cat/K_m values were 5% and 0.2% (respectively) relative to the efficiency of the reaction when malonate was the substrate.

Table 5.

MatB kinetic parameters with different substrates

Enzyme	Substrate	Km (μM)	kcat (s−1)	kcat/Km (M−1 s−1)
RpMatB	Malonate	110 ± 8	18 ± 0.3	2 × 105
	Methylmalonate	81 ± 6	12 ± 0.2	1 × 105
	Ethylmalonate	1,250 ± 55	11 ± 0.2	9 × 103
	Butylmalonate	19,400 ± 1,400	7 ± 0.3	4 × 102
	ATP	242 ± 19	24 ± 0.7	1 × 105
RpMatBT208A	Malonate	391 ± 15	19 ± 0.3	5 × 104
	Methylmalonate	108 ± 6	18 ± 0.2	2 × 105
	Ethylmalonate	144 ± 7	15 ± 0.2	1 × 105
	Butylmalonate	349 ± 18	11 ± 0.2	3 × 104
	ATP	195 ± 14	20 ± 0.5	1 × 105
RpMatBT208G	Malonate	1,140 ± 30	14 ± 0.2	1 × 104
	Methylmalonate	425 ± 13	15 ± 0.2	4 × 104
	Ethylmalonate	370 ± 9	11 ± 0.1	3 × 104
	Butylmalonate	282 ± 15	11 ± 0.2	4 × 104
	ATP	93 ± 8	11 ± 0.3	1 × 105
RlMatB	Malonate	118 ± 12	14 ± 0.4	1 × 105
	Methylmalonate	464 ± 30	9 ± 0.2	2 × 104
	Ethlymalonate	5,600 ± 200	2 ± 0.04	3 × 102
	ATP	199 ± 15	21 ± 0.6	1 × 105

For comparison, we determined the kinetic parameters of RlMatB. The K_m (118 ± 12 μM) and k_cat (14 ± 0.4 s⁻¹) values for RlMatB when malonate was the substrate were similar to those for RpMatB, but the catalytic efficiencies of RlMatB with methylmalonate (1.98 × 10⁴ M⁻¹ s⁻¹) and ethylmalonate (3.21 × 10² M⁻¹ s⁻¹) were ∼7-fold and 27-fold lower than that of RpMatB. For RlMatB, the K_m values for methylmalonate (464 ± 30 μM) and ethylmalonate (5,600 ± 200 μM) were >5-fold higher than the corresponding K_m values of RpMatB (Table 5).

RpMatB substrate binding affinities.

We used isothermal titration calorimetry to measure the affinities of RpMatB and RpMatB^K488A for various substrates. No binding was observed when methylmalonate was titrated into a sample cell containing RpMatB alone. In contrast, ATP bound to RpMatB with a dissociation constant (K_d) of 5.6 ± 1.6 μM (Fig. 3; Table 6). This suggested to us that RpMatB likely binds ATP before binding its organic acid substrate. The K_d for ADP was 10.4 ± 2.0 μM, which is close to the K_d of ATP, suggesting that under some physiological conditions ADP could inhibit RpMatB activity. Surprisingly, the RpMatB^K488A protein bound ATP approximately 10-fold more tightly than wild-type RpMatB, with a K_d of 0.31 ± 0.09 μM (Fig. 3), and the affinity of the protein for ADP (K_d, 40.5 ± 17.5 μM) was 4-fold lower than that of the wild-type enzyme (Table 6). Under the conditions used, we did not observe methylmalonate binding to wild-type RpMatB preequilibrated with ADP or to RpMatB^K488A preequilibrated with ATP, suggesting that both the γ-phosphate of ATP and the K488 side chain may be involved in methylmalonate binding.

Fig 3.

Representative isothermal titration calorimetry binding curves. (A) Binding of ATP to RpMatB. ATP (1 mM) was used as the titrant. (B) Binding of ATP to RpMatB^K488A. ATP (400 μM) was used as the titrant. Top panels show the heat released after each injection, and bottom panels show the resultant binding curves versus the ratio of ATP to enzyme obtained by integrating the heats of injection in the top panels. Data were fit using a single-set-of-sites model, and average values of N, K_d, ΔH, ΔS, and ΔG can be found in Table 6.

Table 6.

Binding of ATP and ADP to MatB and MatB^K488A

Protein	Substrate	N	Kd (μM)	ΔH (kcal mol−1)	ΔS (cal mol−1 K−1)	ΔG (kcal mol−1)
MatB	ATP	0.96 ± 0.05	5.6 ± 1.6	−14.1 ± 0.62	−23.7 ± 2.69	−7.10 ± 0.17
MatB	ADP	1.06 ± 0.06	10.4 ± 2.0	−6.30 ± 0.60	1.49 ± 2.42	−6.74 ± 0.12
MatBK488A	ATP	0.90 ± 0.01	0.31 ± 0.09	−8.94 ± 1.12	−0.48 ± 3.19	−8.80 ± 0.18
MatBK488A	ADP	0.98 ± 0.11	40.5 ± 17.5	−6.17 ± 1.12	−0.71 ± 4.69	−5.96 ± 0.26

Crystal structure of RpMatB.

Crystals of native and Se-Met-labeled RpMatB were obtained, and although the two types of crystals showed the same space group and cell dimension, the Se-Met-labeled RpMatB crystals showed significantly greater order and diffraction, and therefore only the Se-Met-labeled structure is presented. The crystal structure of Se-Met-labeled RpMatB was determined to a 1.7-Å resolution using multiwavelength anomalous dispersion (MAD) phases and includes density for all residues except T166 (Fig. 4). Although AMP, CoA, and methylmalonate were added prior to crystallization, no electron density for any of these ligands was observed. There were four RpMatB monomers in the asymmetric unit that were essentially identical, with a root mean square deviation (RMSD) between monomers of <1.03 Å for 3,776 structurally equivalent atoms. These monomers were arranged as a dimer of dimers within the unit cell, and some other acyl-CoA synthetases have been reported to function as dimers in solution (6, 21, 24). However, RpMatB was eluted from a size exclusion column as a single peak with an apparent molecular mass of 49.0 kDa (expected molecular mass of 54.5 kDa), suggesting that under the solution conditions in this study, RpMatB functions as a monomer.

Fig 4.

RpMatB crystal structures. (A) Overlay of the RpMatB apo structure (the N-terminal domain is shown in light blue and the C-terminal domain in dark blue), with the structure of the RpMatB^K488A/ATP complex (N-terminal domain in light green, C-terminal domain in dark green, and ATP in red). The RpMatB^K488A/ATP structure is in the adenylation conformation, whereas in the RpMatB apo structure the C-terminal domain is rotated ∼20°. (B) Close-up view of the active site, highlighting the position of K488 in the apo structure, compared to the homologous residue A488 in the RpMatB^K488A/ATP structure.

Within the RpMatB structure without bound ligands, the C-terminal domain was rotated ∼20° away from the adenylation conformation, such that the catalytic lysine residue (K488) was not positioned in the active site (Fig. 4B). This conformation was seen in all four monomers within the unit cell, suggesting that it was not a result of crystal packing and might instead be representative of a stable apo conformation. Within the structure, each monomer contained one sulfate ion that interacted with residue K399, which is the conserved hinge residue (often an aspartate in other acyl-CoA synthetases) that rotates to produce the conformational change between the adenylation and thioesterification states. However, crystals grown without sulfate displayed unit cells and space groups identical to the presented structure, and lower-resolution structures with the same conformation (data not shown) did not contain sulfate ions within the unit cell, suggesting that although the sulfate ion may stabilize this conformation within the crystal lattice, it is not required.

The apo structure appears to represent a novel solution phase conformation of the C-terminal domain. Structures of other acyl-CoA synthetases lacking ligands are highly variable in the position of the C terminus, which is in direct contrast to the adenylation and thioester-forming conformations, which show high degrees of structural conservation. These conformations have been described as intermediates in the pathway, but we suggest that the MatB apo structure represents a stable conformation found before substrate binding, as the conformation of the C-terminal domain is outside the 140° sweep between the adenylation and thioester-forming conformations.

A structure of methylated RpMatB^K488A in complex with ATP was determined to 2.0 Å using molecular replacement with the N-terminal domain (residues 1 to 400) of the RpMatB apo structure as a search model (Fig. 4); density for all residues was observed. In order to obtain an ATP-bound structure of RpMatB, it was necessary to methylate the lysines within RpMatB (51). This is an established method to change the nature of the surface-exposed residues of a protein, which play a large role in the formation of the crystal lattice (18, 45), without changing the intrinsic structure of the protein, and it has been successfully utilized to crystallize numerous recalcitrant proteins (31, 35, 39, 52, 57). Additionally, the RpMatB^K488A mutant was utilized, as this protein binds ATP ∼10-fold more tightly than wild-type RpMatB (Table 6), and mutation of the active-site lysine to alanine prevented predicted complications due to methylated K488 within the active site.

The methylated RpMatB^K488A structure with ATP bound showed a high degree of similarity to the apo structure in which comparison of the N-terminal domains (residues 1 to 400) gave an RMSD of 0.92 Å for 1,579 structurally equivalent main-chain atoms and comparison of the C-terminal domains (residues 401 to 503) gave an RMSD of 0.47 Å for 406 structurally equivalent main-chain atoms. This suggested that methylation of RpMatB^K488A did not change the intrinsic structure of the protein and that very little structural change occurred within the domains between the apo and adenylation conformations. The C-terminal domain of the ATP-bound structure did show an ∼20° rotation relative to the apo structure (Fig. 4A), and this conformation is very similar to adenylation conformations seen with other acyl-CoA synthetases. The ATP-bound methylated RpMatB^K488A structure showed RMSDs of 1.95 Å for 428 structurally equivalent α-carbons and 2.10 Å for 468 structurally equivalent α-carbons compared to Homo sapiens acyl-CoA synthetase medium-chain family member 2A in the adenylation conformation (PDB accession number 3C5E) and Saccharomyces cerevisiae acetyl-CoA synthetase in the adenylation conformation (PDB accession number 1RY2), respectively (29, 36).

Attempts to crystallize RpMatB in the thioester-forming conformation were unsuccessful. However, RpMatB is 38% identical and 53% similar to MatB from Streptomyces coelicolor (ScMatB), which utilizes malonate and methylmalonate as substrates and whose structure was recently solved in the thioester-forming conformation (27). Since the thioester-forming conformation is highly conserved within the acyl-CoA synthetase family, the ScMatB structure is an ideal template for modeling the RpMatB thioester-forming conformation. As discussed above, comparison of the N- and C-terminal domains of RpMatB between the apo and adenylation conformations shows very little change within the domains. Therefore, the N- and C-terminal domains of apo RpMatB were separately aligned to the ScMatB structure using secondary-structure matching in Coot (19); this gave an RMSD of 1.66 Å for 414 structurally conserved α-carbons. Figure 5A shows an alignment of the N- and C-terminal domains of the RpMatB apo structure with the structure of ScMatB with methylmalonyl-CoA and AMP bound. Analysis of the active site shows a high degree of structural and sequence conservation between RpMatB and ScMatB. An alignment of RpMatB and related sequences showed a high degree of conservation in the methylmalonate-binding pocket (Fig. 6), further supporting our structural observations. Several RpMatB residues appear to be within 5 Å of the methyl group of methylmalonate, including T208, M302, T303, and M307. Of these, the closest was T208, whose side chain hydroxyl group was located within 3.7 Å of the methyl group of methylmalonate.

Fig 5.

Substrate-binding sites in RpMatB. (A) RpMatB (light green and dark green) modeled in the thioester-forming conformation overlaid onto ScMatB/methylmalonyl-CoA/AMP (PDB accession number 3NYQ, pink and purple); AMP is shown in yellow and methylmalonyl-CoA is shown in gray. Only the phosphopantetheine moiety of CoA is ordered and visible in the ScMatB structure. (B) Close-up view of the active site, with methylmalonyl-CoA shown in gray (AMP is not shown). The side chain of residue T208 of RpMatB is within 3.7 Å of the methyl group of methylmalonyl-CoA.

Fig 6.

Alignment of methylmalonate-binding pockets of MatB homologues. Residues indicated by diamonds are located within 5 Å of the methyl group of methylmalonate in R. palustris MatB. Sequences are from Rhizobium leguminosarum bv. trifolii, Methylobacterium extorquens AM1, Sinorhizobium meliloti 1021, Nitrobacter winogradskyi, Bradyrhizobium japonicum, Oligotropha carboxidovorans, Rhodopseudomonas palustris CGA009, Streptomyces coelicolor, and the Homo sapiens ACSF3 protein.

Single-amino-acid RpMatB variants with expanded range of substrates.

It was hypothesized that a reduction in the size of the side chain of T208 could increase the size of the binding pocket and thus expand the substrate range of RpMatB. Threonine 208 was mutated to alanine or glycine, and the activity of the resulting enzymes was assessed with a variety of substrates. Both enzymes utilized malonate, methylmalonate, ethylmalonate, and butylmalonate, and in some cases they were significantly more efficient than wild-type RpMatB (Table 5). However, like wild-type RpMatB, the variants had undetectable activity with succinate, 2-methylsuccinate, and the monocarboxylic acids acetate, propionate, butyrate, valerate, isobutyrate, acetoacetate, isovalerate, and β-hydroxybutyrate.

The RpMatB^T208A protein had a K_m of 144 ± 7 μM for ethylmalonate, which is ∼9-fold lower than that of wild-type RpMatB, and the K_m (349 ± 18 μM) for butylmalonate was >50-fold lower than that of the wild-type enzyme. While RpMatB exhibited the highest activities with malonate and methylmalonate, RpMatB^T208A was most active with methylmalonate and ethylmalonate, having catalytic efficiencies (k_cat/K_m) of 1.67 × 10⁵ M⁻¹ s⁻¹ and 1.04 × 10⁵ M⁻¹ s⁻¹, respectively. The RpMatB^T208G protein had an even lower K_m for butylmalonate (282 ± 15 μM), and in fact it had the highest catalytic efficiency with butylmalonate (3.90 × 10⁴ M⁻¹ s⁻¹). Interestingly, the catalytic efficiencies of the RpMatB^T208G enzyme with methylmalonate and ethylmalonate were similar (3.53 × 10⁴ M⁻¹ s⁻¹ and 2.97 × 10⁴ M⁻¹ s⁻¹, respectively) (Table 5). The RpMatB^T208A protein had a K_m (195 ± 14 μM) for ATP similar to that of wild-type RpMatB, whereas the RpMatB^T208G protein had a K_m for ATP (93 ± 8 μM) that was ∼2-fold lower (Table 5), demonstrating that ATP binding was not adversely affected by changing T208 to an alanine or glycine residue. These data demonstrated that a reduction in the size of the side chain at position 208 was sufficient to expand the range of substrates used by RpMatB while retaining high catalytic efficiency.

DISCUSSION

RpMatB is the first malonyl-CoA synthetase homologue that has been shown to function in methylmalonate metabolism in vivo. The genes encoding two other MatB homologues that have been studied within a physiological context (from R. leguminosarum and S. meliloti) are located within operons containing a malonyl-CoA decarboxylase and dicarboxylate transporter, suggesting that their primary roles are in malonate metabolism (3, 13). Although it was reported that an S. meliloti ΔmatB strain grows poorly on malonate, growth on methylmalonate was not assessed (13). In R. palustris, matB is independently transcribed. In the genome, the matB gene is located ∼1,500 nucleotides away from the sdhCDAB operon (encoding succinate dehydrogenase), with a divergently transcribed hypothetical gene between matB and sdhCDAB. While malonyl-CoA is commonly metabolized through decarboxylation to acetyl-CoA (malonyl-CoA decarboxylase, EC 4.1.1.9), we predict that methylmalonyl-CoA is converted to succinyl-CoA by methylmalonyl-CoA mutase (EC 5.4.99.2) and then enters the tricarboxylic acid cycle.

In addition to their biological roles, malonyl-CoA and methylmalonyl-CoA are the two most commonly utilized polyketide extender units, and malonyl-CoA synthetases are a common way to produce these extender units directly from malonate and methylmalonate. RlMatB is the most commonly used malonyl-CoA synthetase for production of malonyl-CoA and methylmalonyl-CoA and has been reported to use a variety of synthetic C-2 malonate derivatives, including isopropyl-, dimethyl-, cyclopropyl-, cyclobutyl-, and benzylmalonate, with catalytic efficiencies that ranged from <1 to 19% of those seen with malonate as the substrate (50). The RpMatB reported here uses malonate more efficiently than RlMatB (catalytic efficiency of 1.64 × 10⁵ M⁻¹ s⁻¹ compared to 1.16 × 10⁵ M⁻¹ s⁻¹) and methylmalonate significantly more efficiently, having a catalytic efficiency of 1.48 × 10⁵ M⁻¹ s⁻¹ compared to 1.98 × 10⁴ M⁻¹ s⁻¹, perhaps reflecting adaptation to different physiological roles.

Structural studies of RpMatB in the apo and adenylation conformations combined with a predicted model of RpMatB in the thioester-forming conformation based on a structure of ScMatB allowed for the identification of T208 in RpMatB, which appears to help define the size of the organic acid-binding pocket. By changing T208 to alanine and glycine, MatB variants were generated that efficiently used malonate derivatives with longer acyl chains at the C-2 position. RpMatB^T208A was more efficient at utilizing methylmalonate than wild-type RpMatB (catalytic efficiency of 1.67 × 10⁵ M⁻¹ s⁻¹) and was able to utilize ethylmalonate at 63% of the catalytic efficiency of malonate by wild-type RpMatB. RpMatB^T208G was able to utilize butylmalonate at 23% of the catalytic efficiency of malonate by wild-type RpMatB. Additionally, these enzymes had K_m values for ethylmalonate and butylmalonate of 144 ± 7 and 282 ± 15 μM, respectively, and these K_m values are low enough for these engineered enzymes to feasibly work under physiological conditions for the synthesis of polyketides in heterologous hosts like E. coli.

Interestingly, the four residues that are in close proximity to the methyl group of methylmalonate in RpMatB (T208, M302, T303, and M307) are conserved in most malonyl-CoA synthetases (Fig. 6), including RlMatB. Of the MatB sequences aligned in Fig. 6, only human ACSF3 and ScMatB have any changes in the methylmalonate binding pocket, with a relatively conservative difference of valine instead of threonine at position 208 (based on RpMatB numbering). While the structure of ScMatB is known, kinetic parameters for ScMatB have not been reported; thus, we do not yet know its relative activities with malonate and methylmalonate. A structure is not available for RlMatB, and as residues that contact methylmalonate are highly conserved compared to those in RpMatB, it is likely that subtle differences in structure outside the methylmalonate-binding site cause RlMatB to preferentially utilize malonate.