Applied Microbiology and Biotechnology

, Volume 98, Issue 10, pp 4533–4544

Thioesterases for ethylmalonyl–CoA pathway derived dicarboxylic acid production in Methylobacterium extorquens AM1

Authors

  • Frank Sonntag
    • DECHEMA Forschungsinstitut
  • Markus Buchhaupt
    • DECHEMA Forschungsinstitut
    • DECHEMA Forschungsinstitut
Biotechnologically relevant enzymes and proteins

DOI: 10.1007/s00253-013-5456-y

Cite this article as:
Sonntag, F., Buchhaupt, M. & Schrader, J. Appl Microbiol Biotechnol (2014) 98: 4533. doi:10.1007/s00253-013-5456-y

Abstract

The ethylmalonyl–coenzyme A pathway (EMCP) is a recently discovered pathway present in diverse α-proteobacteria such as the well studied methylotroph Methylobacterium extorquens AM1. Its glyoxylate regeneration function is obligatory during growth on C1 carbon sources like methanol. The EMCP contains special CoA esters, of which dicarboxylic acid derivatives are of high interest as building blocks for chemical industry. The possible production of dicarboxylic acids out of the alternative, non-food competing C-source methanol could lead to sustainable and economic processes. In this work we present a testing of functional thioesterases being active towards the EMCP CoA esters including in vitro enzymatic assays and in vivo acid production. Five thioesterases including TesB from Escherichia coli and M. extorquens, YciA from E. coli, Bch from Bacillus subtilis and Acot4 from Mus musculus showed activity towards EMCP CoA esters in vitro at which YciA was most active. Expressing yciA in M. extorquens AM1 led to release of 70 mg/l mesaconic and 60 mg/l methylsuccinic acid into culture supernatant during exponential growth phase. Our data demonstrates the biotechnological applicability of the thioesterase YciA and the possibility of EMCP dicarboxylic acid production from methanol using M. extorquens AM1.

Keywords

Methylobacterium extorquensThioesterasesEthylmalonyl–CoA pathwayYciADicarboxylic acidCoenzyme A

Introduction

Dicarboxylic acids are important building blocks for chemical and pharma industry and can also be used as chelators or food additives (Lee et al. 2002; Werpy and Petersen 2004). Many production processes are still based on chemical synthesis mainly due to economical reasons (Kircher 2006). Nevertheless, substantial progress has been made in the last years for bio-based production of dicarboxylic acids focused on succinic acid (Sauer et al. 2008; Thakker et al. 2012) and adipic acid (Polen et al. 2013). The allocation of new and uncommon dicarboxylic acids by bio-based production processes would provide a variety of sustainable monomers for chemical industry (e.g., novel bioplastics).

The recently elucidated ethylmalonyl–CoA pathway (EMCP) (Erb et al. 2007; Erb et al. 2009; Peyraud et al. 2009) contains several enantiomeric, branched and satured or unsatured carboxylated C4– and C5-acyl-CoA esters (see Fig. 1). Their corresponding dicarboxylic acids ethylmalonic acid, (2S)-methylsuccinic acid, mesaconic acid, (2R,3S)-methylmalic acid and methylmalonic acid are especially interesting as novel synthons being commercially unavailable in bulk quantities to date (Alber 2011). Ethylmalonic acid as well as (2S)-methylsuccinic acid are particular interesting because of the unique existence of their corresponding CoA esters in the EMCP. Ethylmalonic acid can be used as co-crystallization additive (Aitipamula et al. 2010) and it has been patented, as well as methylsuccinic acid, as monomer components for various polymers in application for coatings and solvents for cosmetics (Hu and Bailey 1999; Loos et al. 2012; Muller and Richard 2012). Especially the application of 2-methylsuccinate as monomer for cosmetic solvents (Muller and Richard 2012) turns it to an attractive target for biotechnological production of a compound labeled as natural. Mesaconic acid, which also occurs in the mesaconate pathway for l-glutamate fermentation (Kato and Asano 1997), is used as fire retardant (Di Giulio and Bauer 1982).
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-013-5456-y/MediaObjects/253_2013_5456_Fig1_HTML.gif
Fig. 1

Selected dicarboxyl-CoA esters of the ethylmalonyl–CoA pathway. Their corresponding dicarboxylic esters after cleavage of the coenzyme A are displayed underneath

The EMCP is present in some bacteria such as the methylotrophic model organism Methylobacterium extorquens AM1 (Peyraud et al. 2009) offering the possibility to synthesize the novel acids from methanol as a sustainably producible, non-food competitive substrate (Schrader et al. 2009). The fact of low activity of most EMCP enzymes during growth on C3 or C4 carbon sources (Skovran et al. 2010; Šmejkalová et al. 2010) emphasizes the use of methanol as a substrate.

The presence of the dicarboxylic acids as CoA esters (see Fig. 1) requires a hydrolyzation of their thiol groups, which can be performed by CoA-transferases or thioesterases. The latter do not need an additional donor/acceptor molecule which lowers the chance of adding metabolic imbalances. However, to our knowledge thioesterases being specifically active for EMCP CoA esters have not been described to date. The number of characterized thioesterases is steadily increasing and their classification was excellently reviewed recently (Cantu et al. 2011; Cantu et al. 2010). Unfortunately, most members of the families TE1–TE13, which are acyl-CoA hydrolases, have a rather broad substrate spectrum or are only active on C10– or longer acyl-CoAs (Cantu et al. 2010). Furthermore, the HotDog-fold acyl-CoA thioesterases of mentioned families lack conserved catalytic residues and binding pockets (Zhuang et al. 2008) making an in silico prediction of enzyme activity for the EMCP CoA esters unfeasible.

In this article, we present the identification of thioesterase candidates probably active on EMCP-CoA esters by a comprehensive literature research followed by in vitro enzyme activity analysis of the selected enzymes. M. extorquens AM1 expressing suitable candidates were tested for the release of EMCP-related dicarboxylic acids. For the first time, in vivo production of dicarboxylic acids derived from the EMCP with a methylotrophic bacterium growing on methanol is described.

Materials and methods

Chemicals, bacterial strains and growth conditions

M. extorquens AM1 (Peel and Quayle 1961) was grown at 30 °C in minimal medium with increased cobalt concentration of 12.5 μM as described before (Kiefer et al. 2009) containing 123 mM methanol or 31 mM succinate. Utilization of TCA and EMCP-related dicarboxylic acids as sole carbon source was tested in equivalent medium containing 2 mM of a single acid instead of methanol. Escherichia coli strains DH5α (Gibco-BRL, Rockville, MD, USA) and BL21(DE3) (Novagen, Madison, WI, USA) were grown in lysogeny broth (LB) medium (Bertani 1951) at 37 °C. Kanamycin was used at concentrations of 30 μg/ml for E. coli and 50 μg/ml for M. extorquens. Cumate (4-isopropylbenzoic acid) was used as inducer at a final concentration of 5 μg/ml diluted from 1 mg/ml 50:50 methanol–water stock solution.

The dicarboxylic acids ethylmalonic acid, methylmalonic acid, racemic (rac-) 2-methylsuccinic acid and mesaconic acid as well as palmitoyl-, decanoyl-, rac-methylmalonyl-CoA and succinyl-CoA and cumate were purchased from Sigma-Aldrich (Steinheim, Germany) in their highest available degree of purity. Rac-beta-methylmalic acid was synthesized by Otava (Kiev, Ukraine). Coenzyme A was purchased by Chemos (Regenstauf, Germany).

Genetic manipulations and plasmid construction

Polymerase chain reactions (PCR) were performed using High-Fidelity PCR Master and amplified DNA was purified by PCR purification Kit (both from Roche, Basel, Switzerland). Restriction enzymes and T4 ligase were purchased from New England Biolabs (Beverly, USA). Plasmid DNA was purified from E. coli by Roti Prep Plasmid MINI Kit (Carl-Roth, Karlsruhe, Germany). Primers were purchased from Sigma-Aldrich. Confirmation of nucleotide sequences was performed by SRD (Bad Homburg, Germany).

Genomic DNA from E. coli DH5α and M. extorquens AM1 (DSM1338) was purified using GenElute Bacterial Genomic DNA Kit from Sigma-Aldrich. Genomic DNA of Azoarcus evansii (DSM6898) and Bacillus cereus (DSM31) was purchased from DSMZ (Braunschweig, Germany). Mus musculus acyl-CoA thioesterase 4 (acot4) cDNA (Unigene ID Mm.482217) was ordered from Source Bioscience (Nottingham, UK).

All standard cloning techniques were carried out as described before (Sambrook and Russell 2001). Transformation of plasmids into M. extorquens was performed as described by Toyama et al. (1998).

Genes encoding the acyl-CoA thioesterase B from E. coli DH5α (tesB; NP_414986) and M. extorquens AM1 (YP002963160), acyl-CoA thioesterase 4 from Mus musculus (acot4; NP599008), paaI from Azoarcus evansii (AAG28967), acyl-CoA thioester hydrolase from E. coli (yciA; YP_002270185) and 3-hydroxyisobutyryl-CoA hydrolase from Bacillus cereus (bch; NP_832055) were amplified from their genomic DNA using the corresponding primers listed in Table 1. Resulting PCR products were digested with NdeI + EcoRI and cloned into pET28a(+) for expression in E. coli or digested with KpnI + EcoRI and cloned into pCM160 for expression in M. extorquens (Table 1).
Table 1

Primers, plasmids and strains used in this work. Underlined sequences (all 5′–3′) indicate restriction enzyme recognition sites

Name

Description

Reference

Primers

 tesBext_pET_fw

ACTCATATGATGCCCGACCCCGTCGATGC

this study

 tesBcoli_pET_fw

ACTCATATGATGAGTCAGGCGCTAAAAAA

this study

 ACOT4_pET_fw

ACTCATATGATGGCAGCGACACTGAGCG

this study

 PaaI_pet_fw

ACTCATATGATGACTGAGGCGGGCTATCG

this study

 YciA_pET_fw

ACTCATATGATGTCTACAACACATAACGTC

this study

 bch_pET28_fw

ACTCATATGATGACTGAACAAGTTTTATTTTC

this study

 tesBext_rev

ACGTGAATTCTCAGCTGCGCCGGGAGCGG

this study

 tesBcoli_rev

ACGTGAATTCTTAATTGTGATTACGCATCAC

this study

 ACOT4_rev

ACGTGAATTCCTACAGTCTACAGGAGGCTC

this study

 PaaI_rev

ACGTGAATTCTCACAACTCGACGACCGCC

this study

 YCiA_rev

ACGTGAATTCTTACTCAACAGGTAAGGCG

this study

 bch_rev

ACGTGAATTCTTATGCATTAAGTAAGTTAAAG

this study

 tesBext_pCM_fw

ACTGGTACCATGCCCGACCCCGTCGATGC

this study

 tesBcoli_pCM_fw

ACTGGTACCATGAGTCAGGCGCTAAAAAA

this study

 ACOT4_pCM_fw

ACTGGTACCATGGCAGCGACACTGAGCG

this study

 PaaI_pCM_fw

ACTGGTACCATGACTGAGGCGGGCTATCG

this study

 YciA_pCM_fw

ACTGGTACCATGTCTACAACACATAACGTC

this study

 bch_pCM160_fw

ACTGGTACCATGACTGAACAAGTTTTATTTTC

this study

 RBS_pCM160-fw

CGCGACGGTCTCGTAAAAAGGAAGGAGGTATTAAGGTAC

this study

 RBS_pCM160-rev

CTTAATACCTCCTTCCTTTTTACGAGACCGTCGCGCATG

this study

Plasmids

 pET28a(+)

Inducible expression vector for E. coli; pT7, KanR, pBR322ori, His-Tag

Novagen

 pCM160

constitutive expression vector for Methylobacterium extorquens; KanR, pmxaF, oriT, pBR322ori

Marx and Lidstrom 2001

 pLC290

Cumate inducible expression vector for M. extorquens; ; KanR, oriT, pBR322ori, ttrnB

Chubiz 2013

 pET28-tesBext

pET28 containing thioesterase B gene from M. extorquens

this study

 pET28-tesBec

pET28 containing thioesterase B gene from E. coli

this study

 pET28-yciA

pET28 containing acyl-CoA thioesterase yciA gene from E. coli

this study

 pET28-acot4

pET28 containing succinyl-CoA thioesterase gene from Mus musculus

this study

 pET28-paaI

pET28 containing paaI gene from Azoarcus evansii

this study

 pET28-bch

pET28 containing 3-hydroxyisobutyryl-CoA hydrolase gene from Bacillus cereus

this study

 pCM160-tesBext

pCM160 containing thioesterase B gene from M. extorquens

this study

 pCM160-tesBec

pCM160 containing thioesterase B gene from E. coli

this study

 pCM160-yciA

pCM160 containing acyl-CoA thioesterase gene from E. coli

this study

 pCM160-acot4

pCM160 containing succinyl-CoA thioesterase gene from M. musculus

this study

 pCM160-bch

pET28 containing 3-hydroxyisobutyryl-CoA hydrolase gene from B. cereus

this study

 pCM160-RBS-yciA

pCM160-yciA containing an optimized RBS

this study

 pLC290-tesBec

pLC290 containing thioesterase B from E. coli

this study

Strains

E. coli DH5alpha

F, Φ80dlacZΔM15, Δ(lacZYA-argF)U169, deoR, recA1, endA1, hsdR17(rK mK+), phoA, supE44, λ, thi-1

ATCC

E. coli BL21(DE3)

F ompT gal dcm lon hsdSB(rB mB) λ(DE3 [lacI lacUV5-T7 gene 1 ind1 sam7 nin5])

Novagen

M. extorquens AM1

Facultative methylotrophic, obligate aerobic, gram-negative, pink pigmented α-proteobacterium, CmR

Peel and Quayle 1961 DSM1338

The ribosome binding site of pCM160-thioesterase constructs was optimized using the RBS calculator (Salis 2011). The optimized sequence was introduced by ligation of previously annealed primers RBS_pCM160-fw and RBS-pCM160-rev (Table 1) into SphI + KpnI restricted pCM160-thioesterase vectors.

For inducible expression of thioesterase B from E. coli in M. extorquens, tesB gene from pCM160-tesBec was cut out with SphI + EcoRI and ligated into equally digested pLC290 (Chubiz et al. 2013).

Protein purification and quantification

TesB from E. coli (TesBec) and M. extorquens (TesBext), Acot4, PaaI, YciA and Bch were produced heterologously by transformation of each pET28a(+)-thioesterase construct (Table 1) into E. coli BL21. Cells were grown in 100 ml LB-medium with 30 μg/ml kanamycin to an OD600 of 0.5–0.7. IPTG induction was performed for 16 h at the following concentrations and temperatures: 0.5 mM at 16 °C for TesBec and Bch, 0.4 mM at 16 °C for TesBext and YciA, 0.6 mM at 22 °C for PaaI and 0.1 mM at 22 °C for Acot4. Cells were harvested by centrifugation, washed twice with 50 mM phosphate buffer pH 7.5 and disrupted by sonification (0.5 s pulse, 1 s break for 12 min by an amplitude of 20 %). Cell lysate was centrifuged for 20 min at 15,000 × g and 4 °C. The supernatant was applied to a Ni-sepharose matrix in a Poly-Prep column (Bio-Rad) that had been equilibrated with 10 bed volumes of buffer A containing 50 mM NaH2PO4, 300 mM NaCl and 10 mM imidazole at a pH of 8. The column was washed twice with 4 bed volumes of buffer A containing 20 mM imidazole. Finally, thioesterase was eluted with buffer A containing 250 mM imidazole.

Imidazole was removed and protein concentrated by application of Amicon spin columns (Millipore) at appropriate MWCOs. Enzyme was stored at 4 °C overnight.

Protein was analyzed via sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) for purity with the help of PageRuler Plus Prestrained Protein Ladder (Fermentas). Enzyme concentration was measured with Pierce BCA Protein Assay Kit (Thermo Scientific).

Synthesis of CoA esters

CoA esters of the EMCP were synthesized from their corresponding acid (see chemicals) via its anhydride as described by Stadtman (1957). The resulting CoA esters have the following conformation: ethylmalonyl-CoA as enantiomeric mixture of (2R)- and (2S)-ethylmalonyl-CoA (Erb et al. 2007), methylsuccinyl-CoA as a mixture of (2R)-(2S)-(3R)- and (3S)-methylsuccinyl-CoA (Erb et al. 2009) and mesaconyl-CoA as mixture of structural isomers mesacony-C1-CoA and mesaconyl-C4-CoA. Due to lack of references, methylmalyl-CoA was assumed to be a diastereo-/enantiomeric mixture of (2R/3S)(2S/3R)(2R/3R)(2S/3S)-beta-methylmalyl-CoA and (2R/3S)(2S/3R)(2R/3R)(2S/3S)-gamma-methylmalyl-CoA. Product formation of the CoA esters was confirmed by HPLC analysis (Erb et al. 2009). Full conversion was checked by undetectable DTNB absorption indicating fully converted free coenzyme A.

DTNB assay for CoA-ester cleavage

DTNB assay for thioesterase activity measurements was based on a modified protocol of previously described assays from Zheng et al. (2004), Cho and Cronan (1993) and Zhuang et al. (2008). Reaction mixture contained final concentrations of 10 mM HEPES pH 7.5 (25 °C), 150 mM KCL, 10 mM KH2PO4, 0.1 mM DTNB, 10–200 μM CoA ester and 1.5 μM protein. The reaction was monitored for 5 min at 25 °C and 412 nm in 96-well microtiter plates using a microplate reader (TECAN, Switzerland).

Absorption slope was an average of three measurements. Specific enzyme activity was calculated by the division of the volumetric enzyme activity by the protein concentration determined as described above.

Analysis and quantification of dicarboxylic acids

First, 1 ml of culture broth was centrifuged for 3 min at 16,000 × g for sedimentation of cells. Then, supernatant was passed through a 0.22-μm filter and analyzed on HPLC.

Samples were analyzed on SLC10-A system equipped with a SPD20A UV–Vis detector and a SPD10Avp PDA detector (all from Shimadzu) together with an Aminex HPX-87H column (length: 300 mm, diameter: 7.8 mm; Bio-Rad). Separation was performed isocratically for 30 min with a mobile phase consisting of 2 mM H2SO4 in MilliQ water. Column temperature was maintained at 30 °C. Elution was monitored continuously at 205 nm with the UV–Vis detector for quantification and at a range of 190–600 nm with PDA-detector for absorption spectra scanning. Ethylmalonic acid, methylmalonic acid, rac-2-methylsuccinic acid, rac-b-methylmalic acid and mesaconic acid were quantified with the help of calibration curves prepared by the use of their corresponding standards.

Results

Identification of thioesterase candidates acting on EMCP intermediates

Acyl-CoA thioesterases (families TE1–TE13) (Cantu et al. 2010) are most promising for investigation of their activity towards dicarboxyl-CoAs of the EMCP. Most described members of this group cleave off CoA of a broad range of acyl-CoAs with increasing activity by increasing C length. Only a few specific short-chain acyl-CoA thioesterases have been investigated but so far not for their activity towards the EMCP CoA esters. Two species of acyl-CoA thioesterases were chosen for in vitro analysis: "broad range" thioesterase cleaving CoA off acyl-chains from four to 16 or more carbon atoms and specific short-chain thioesterases preferring substrates with a related structure to the EMCP-CoA esters, i.e., C4–C5 (di)carboxyl-CoA esters.

Thioesterase II/B (TesB) of E. coli is one of the longest known and investigated broad-range thioesterase originally described in 1970 by Barnes and co-workers. It cleaves CoA off acyl moieties of 4–18 carbon atoms with highest activity for palmityl-CoA (Barnes et al. 1970; Naggert et al. 1991). In addition, it has already been used for biotechnological production of several other acids such as hydroxyvalerates, polyketides, 3-hydroxybutyric acid or 3-hydroxydecanoic acid (Chung et al. 2009; Liu et al. 2008; Martin and Prather 2009; Tseng and Prather 2012) and is therefore a promising candidate for cleaving CoA off the EMCP-CoA esters. A TesB homolog also exists in M. extorquens (Vuilleumier et al. 2009), which makes it the second candidate for EMCP-CoA activity investigation.

YciA is another broad range thioesterase which is active on acyl-CoAs with C2 (acetyl-CoA) to C18 moieties (oleyl-CoA) including substrates of the EMCP such as methylmalonyl-CoA and structural related compounds like isobutyryl-CoA or b-methylcrotonyl-CoA (Zhuang et al. 2008). Therefore, YciA of E. coli was chosen as another candidate.

The PaaI of Azoarcus evansii (Song et al. 2006) is mainly active towards aromatic CoA esters but also on short-chain acyl-CoAs such as methylmalonyl-CoA or isobutyryl-CoA and is therefore an additional candidate.

Succinyl-CoA thioesterase Acot4 from Mus musculus and 3-hydroxyisobutyryl-CoA hydrolase Bch from Bacillus cereus were chosen as candidates of the specific short-chain acyl-CoA thioesterase group. Both enzymes are active on substrates which show similarity to the EMCP-CoA esters. Acot4 (Westin et al. 2005) shows activity for succinyl- (C4) and glutaryl-CoA (C5) but not on acyl-CoAs with a shorter or longer carbon chain. Bch has been successfully used by Lee and co-workers (2008) for cleavage of the C4-carboxyl CoA-ester 3-hydroxyisobutyryl-CoA in E. coli.

The six selected thioesterases provide a starting point for in vitro enzyme analysis to identify possible EMCP CoA-ester hydrolyzation abilities.

Protein purification

The genes coding for TesB from E. coli and M. extorquens, YciA, Acot4, PaaI and Bch were cloned into pET28a(+) vector and resulting constructs transformed in E. coli BL21(DE3). Enzymes were heterologously expressed by IPTG induction and purified by the use of their N-terminal HIS-Tag. As shown in Fig. 2, expected bands of ∼32 kDa for thioesterase B of E. coli and M. extorquens, ∼15 kDa for YciA, ∼50 kDa for Acot4, ∼14 kDa for PaaI and ∼36 kDa for Bch are visible for every enzyme after the final purification step. Strong bands at the expected sizes in all samples and only minor impurities in the TesBext, YciA and Acot4 sample allow enzymatic assay for substrate specificity and activity of the thioesterases in vitro.
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-013-5456-y/MediaObjects/253_2013_5456_Fig2_HTML.gif
Fig. 2

Sections of several SDS-PAGE gels showing protein bands of thioesterases after final HIS-Tag purification and Amicon concentration. Lane 1: marker; lane 2: thioesterase B from E. coli (TesBec); lane 3: thioesterase B homologue from M. extorquens (TesBext); lane 4: bacterial acyl-CoA thioesterase from E. coli (YciA); lane 5: succinyl-CoA hydrolase from M. musculus (Acot4); lane 6: PaaI from A. evansii; lane 7: 3-hydroxyisobutyryl-CoA hydrolase from B. cereus (Bch)

Testing selected thioesterases for their activity towards EMCP-CoA esters by in vitro DTNB assay

Low hydrolyzation activities for short-chain acyl-CoAs (C6 or lower), mostly tested by CoA-cleavage of substrates having unbranched carbonyl chains like succinyl-CoA or hexanoyl-CoA, has been shown in the literature for all selected thioesterases except TesBext. The carbon chains of all EMCP CoA esters are branched with an additional unsaturated bond for mesaconyl-CoA or hydroxyl group for (2R,3S)-methylmalyl-CoA (see Fig. 1). This might have strong influence on acceptance by the thioesterases. Due to the only recent elucidation of the EMCP and the current unpurchasability of most of its CoA esters (exception: rac-methylmalonyl-CoA), none of them has been tested as a substrate for thioesterases yet. Therefore, the enzyme activities of the six purified thioesterases were measured for rac-methylmalonyl-, rac-ethylmalonyl-, rac-methylsuccinyl-, rac-methylmalyl-, mesaconyl- and succinyl-CoA as a control (HPLC data of synthesized CoA esters not shown).

Table 2 summarizes the determined specific activity and Km values of all purified thioesterases for the different EMCP CoA esters. Although PaaI is active for rac-methylmalonyl-, rac-ethylmalonyl- and mesaconyl-CoA, its high Km values make it unattractive for in vivo application. The acyl-CoA thioesterase YciA of E. coli shows the highest specific activities for the EMCP CoA esters (especially for mesaconyl-CoA). Thioesterase B of M. extorquens seems to have similar substrate preferences of C10 or longer acyl-CoAs as its E. coli homologue. In contrast, TesBext has distinct higher Km values for the EMCP-CoA esters than TesBec at comparable activities. Although (3S)-isobutyryl-CoA hydrolase of B. cereus (Bch) shows activity for all EMCP CoA esters, the corresponding Km values are very high with the exception for rac-ethylmalonyl-CoA.
Table 2

Specific activities (U/mg) and Km values (μM) of different thioesterases for succinyl-CoA, decanoyl-CoA, palmitoyl-CoA and various CoA-ester intermediates of the ethylmalonyl-CoA pathway in their following conformation: methyl- and ethylmalonyl-CoA as (2R/2S)-enantiomeric mixtures; methylsuccinyl-CoA as mixture of (2R)-(2S)-(3R)- and (3S)-methylsuccinyl-CoA; mesaconyl-CoA as mixture of structural isomers mesaconyl-C1– and mesaconyl-C4-CoA and methylmalyl-CoA as diastereo-/enatiomeric mixture of (2R/3S)(2S/3R)(2R/3R)(2S/3S)-beta-methylmalyl-CoA and (2R/3S)(2S/3R)(2R/3R)(2S/3S)-gamma-methylmalyl-CoA

Substrate

TesBec

TesBext

Acot4

PaaI

YciA

Bch

Specific activity [U/mg]

 Succinyl-CoA

n.d.

3.5 (±0.13)

1.8 (±0.28) [3.98b]

n.d.

31.2 (±0.3)

n.d.

 Methylmalonyl-CoA

8.8 (±1.23)

2.7 (±0.33)

0.5 (±0.08)

5.4 (±1.5) [2.9c]

10.6 (±3.5)

3.5 (±0.09)

 Ethylmalonyl-CoA

3.6 (±0.54)

2.3 (±0.51)

0.2 (±0.02)

3.6 (±2.4)

6.2 (±0.1)

1 (±0.42)

 Methylsuccinyl-CoA

0.5 (±0.24)

1.4 (±0.04)

0.3 (±0.04)

n.d.

7.2 (±0.48)

2.6 (±0.56)

 Mesaconyl-CoA

2.2 (±0.43)

0.4 (±0.07)

0.2 (±0.01)

2.1 (±0.39)

12.3(±0.46)

0.6 (±0.2)

 Methylmalyl-CoA

2.2 (±0.49)

1.9 (±0.29)

0.7 (±0.04)

n.d.

3.5 (±1.3)

1.3 (±0.03)

 Decanoyl-CoA

n.m.

94.3 (±13)

    

 Palmitoyl-CoA

181 (±32) [93.8a]

112 (±2.4)

    

Km [μM]

 Succinyl-CoA

n.d.

118 (±4)

18 (±3) [13.3b]

n.d.

17 (±0.2)

n.d.

 Methylmalonyl-CoA

98 (±13)

54 (±6)

23 (±4)

771 (±189) [1100c]

145 (±43)

199 (±5)

 Ethylmalonyl-CoA

99 (±15)

112 (±22)

26 (±3)

1269(±724)

43 (±1)

43 (±12)

 Methylsuccinyl-CoA

71 (±26)

411 (±12)

45 (±5)

n.d.

50 (±3)

4580 (±864)

 Mesaconyl-CoA

24 (±5)

115 (±20)

8 (±1)

254 (±46)

168 (±6)

n.m.

 Methylmalyl-CoA

51 (±12)

75 (±10)

35 (±2)

n.d.

13 (±4)

3566 (±82)

Values were determined at pH 7.5 and 25 °C using the DTNB spectrophotometric assay (see Materials and methods). All measurements were performed in triplicates. Corresponding standard deviations are given in parentheses. If available, values described in literature are shown in square brackets behind the determined values

TesBec thioesterase B from E. coli, TesBext thioesterase B homologue from M. extorquens, YciA bacterial acyl-CoA thioesterase from E. coli, Acot4 succinyl-CoA hydrolase from M. musculus, PaaI phenylacetate thioesterase from A. evansii, Bch 3-hydroxyisobutyryl-CoA hydrolase from B. cereus, n.d. not detectable (detection limits: ≤0.1 U/mg; Km ≥5,000 μM), n.m. not measured

aNaggert (1991)

bWestin (2005)

cSong (2006)

We identified thioesterase YciA as an effective enzyme for EMCP CoA-ester hydrolyzation with vmax values up to 12.3 U/mg protein in vitro, making it a promising candidate for release of EMCP derived acids in vivo. Thioesterase B, Acot4 and Bch might also be suitable for the release of some EMCP related acids, especially methyl- and ethylmalonate.

Release of dicarboxylic acids by M. extorquens AM1 expressing thioesterases

Enzyme properties determined in vitro may vary from its in vivo characteristics due to protein folding, different reaction conditions, inhibitors and so on. To verify whether the identified thioesterase candidates are suitable for cleaving the EMCP CoA esters in vivo, their genes were cloned into pCM160 vector and transformed in M. extorquens cells, thereafter plated on methanol agar.

Colonies were obtained for M. extorquens transformed with pCM160-tesBext, -yciA, -bch and -acot4 after 4 days but not for pCM160-tesBec. The strong constitutive, methanol inducible pmxaF promoter of pCM160 probably cause high TesBec production which seems to be lethal for M. extorquens growing on methanol. Therefore, pCM160-tesBec was transformed again and cells plated on minimal medium containing succinate as sole carbon source. On these plates colonies appeared after 4 days. Additionally, tesBec was cloned into pLC290 vector containing a cumate inducible promotor. Transformants containing the respective expression plasmid were able to grow on methanol medium.

Production of EMCP derived dicarboxylic acids and growth of tesBext-, ycia-, acot4- and bch-expressing M. extorquens was investigated by cultivation of cells harboring the pCM160-thioesterase constructs and the empty vector control in shake flasks, measuring OD600 and analyzing supernatants for dicarboxylic acids after certain time points. Growth characteristics of strains harboring pCM160-acot4, -bch and -yciA were similar to the empty vector control with μmax values of about 0.12–0.14 per hour. AM1_pCM160-tesBext showed a reduced maximal growth rate of 0.09 per hour.

Profile of EMCP derived dicarboxylic acids in culture supernatant is not altered by overexpression of bch and tesBext compared to empty vector control considering the detection limit of 0.05 mg/l for mesaconate and 2 mg/l for the remaining acids. Mesaconate is released in small amounts by cells containing pCM160-acot4 accumulating up to 1.5 mg/l after 30 h of cultivation, whereas other EMCP derived dicarboxylic acids were not detectable. In contrast, the supernatant of AM1_pCM160-yciA showed concentrations of up to 10 mg/l mesaconate and 2-methylsuccinate compared to AM1 harboring the empty vector where no 2-methylsuccinate and maximal amounts of 0.8 mg/l mesaconate were detectable (data not shown). To further raise acid release, the ribosome binding site of the pCM160_yciA construct was optimized for M. extorquens and the yciA gene. Growth characteristics of cells containing pCM160-RBS-yciA and their mesaconate and 2-methylsuccinate supernatant accumulation are shown in Fig. 3.
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-013-5456-y/MediaObjects/253_2013_5456_Fig3_HTML.gif
Fig. 3

Time-dependent release of mesaconate and 2-methylsuccinate into culture supernatant and growth characteristics of M. extorquens AM1 carrying pCM160-RBS-yciA compared to the empty vector control (a). All measurements were performed from three independent cultures. b Absorption spectra of mesaconate standard (grey line) and mesaconate peak of pCM160-RBS-yciA (black dotted line). c Absorption spectra of 2-methylsuccinate standard (grey line) and 2-methylsuccinate peak of pCM160-RBS-yciA (black dotted line). Maximum absorbance of each substance is shown above the curves

The optimized ribosome binding site of the new pCM160_RBS-yciA construct further increased the amount of mesaconate and 2-methylsuccinate released into the culture supernatant of M. extorquens as shown in Fig. 3a. The identity of mesaconate and 2-methylsuccinate peak was confirmed by retention time and identical absorption spectra shown in Fig. 3b and c. As 2-methylsuccinyl-CoA is present in its (2S) conformation in the EMCP, a release of (2S)-methylsuccinate can be expected although it was not directly confirmed. Up to 70 mg/l mesaconate and 58 mg/l 2-methylsuccinate accumulated after 30 h. That point of time correlates to the maximum optical density of the AM1_pCM160-RBS-yciA culture. During the stationary growth phase (30–48 h), the concentration of both acids in supernatant is reduced by more than 50 % from 70 to 38 mg/l mesaconate and 58 to 30 mg/l 2-methylsuccinate.

As already mentioned, colonies for AM1_pCM160-tesBec could not be obtained on minimal medium containing methanol. Therefore two approaches were followed to determine a possible EMCP derived acid production by TesBec. In a first approach, the preculture of AM1_pCM160-tesBec was cultivated in succinate minimal medium, washed with sterile water and further used to inoculate a main culture growing on methanol. The growth profile of the main culture showed a significantly prolonged lag phase of 24 h compared to 8–10 h of the other thioesterase expressing strains or the empty vector control. However, maximum growth rate during exponential growth phase was not altered compared to the empty vector control as well as the reached maximum optical density of 3–3.5 and the dicarboxylic acid profile in culture supernatant (data not shown). To ensure that the prolonged lag phase and succinate precultivation does not lead to, e.g., plasmid alteration, pLC290-tesBec construct harboring a cumate inducible promotor was used as a second approach. AM1_pLC290-tesBec grew in methanol minimal medium and was induced with cumate after reaching middle exponential growth phase. Compared to the empty vector control and the non-induced culture, induction of tesBec did not lead to reduced growth or altered acid accumulations.

We identified thioesterase YciA as an appropriate enzyme for cleavage of the EMCP CoA-esters 2-methylsuccinyl-CoA and mesaconyl-CoA leading to release of 2-methylsuccinic- and mesaconic acid into culture supernatant of M. extorquens AM1. During stationary growth phase a significant reduction of the previously released acids was observed.

Utilization of (EMCP)-dicarboxylic acids by AM1

Dicarboxylic acids like mesaconate and 2-methylsuccinate are very stable when dissolved in water which strongly suggests an uptake of the acids during stationary growth phase. This consequently leads to the question whether M. extorquens is able to utilize the EMCP related dicarboxylic acids as energy and carbon source. M. extorquens AM1 is able to use several dicarboxylic acids as sole carbon source which was already shown by Peel and Quayle (1961) in the paper of AM1’s first characterization. Therefore, ethylmalonate, 2-methylsuccinate, mesaconate and methylmalonate were tested as sole carbon source for growth. The TCA-cycle intermediate α-ketoglutarate was used as a comparable C source as the chosen EMCP derived dicarboxylic acids, except methylmalonate, also consist of 5C atoms.

As shown in Table 3, M. extorquens AM1 is able to use every tested EMCP related dicarboxylic acid as sole carbon source at which growth on methylmalonate (C4) was faster than on the C5-dicarboxylic acids ethylmalonate, 2-methylsuccinate or mesaconate. Growth on the C5 acid α-ketoglutarate as a part of the TCA was also observable.
Table 3

Ability of Methylobacterium extorquens AM1 to grow on various dicarboxylic acids of the TCA and EMCP (derived from CoA esters) cycle

Dicarboxylic acid

Pathway

Growth

Reference

Succinate

TCA

++

Peel and Quayle 1961

Fumarate

TCA

++

Peel and Quayle 1961

Malate

TCA

++

Peel and Quayle 1961

α-Ketoglutarate

TCA

++

This study

Ethylmalonate

EMCP

+

This study

2-Methylsuccinate

EMCP

+

This study

Mesaconate

EMCP

(+)

This study

Methylmalonate

EMCP

++

This study

++ growth similar to succinate, + growth slower than on succinate, (+) minimal growth, − no growth

The utilization of the EMCP derived dicarboxylic acids as sole carbon source and the observed uptake of mesaconate and 2-methylsuccinate by AM1_pCM160-yciA implicates a functional transport system for these acids in M. extorquens AM1. Protein BLAST analysis was performed to identify possible dicarboxylate carriers responsible for transport of the EMCP related acids in M. extorquens AM1. Results are summarized in Table 4. Several homologues of the DctA dicarboxylic acid transport system in the genome of AM1 were identified.
Table 4

Proteins of M. extorquens AM1 showing highest homology to different dicarboxylic acid transporters identified by protein BLAST analysis

 

Homologues in M. extorquens AM1

Transporter

Protein ID and gene locus taq

Identitiy

Similarity

DctAa (succinate importer)

YP_002964290.1

63 %

80 %

MexAM1_META1p3271

 

YP_002964292.1

55 %

73 %

MexAM1_META1p2326

 

YP_002963386.1

48 %

69 %

MexAM1_META1p4339

DctBa (transport sensor protein)

YP_002962215.1

32 %

49 %

MexAM1_META1p1044

DctDa (transcriptional regulator of dctA)

YP_002964292.1

39 %

53 %

MexAM1_META1p4339

KgpTb (α-ketoglutarate permease)

YP_002966015.1

61 %

75 %

MexAM1_META1p5134

aHomology to Rhodobacter capsulatus SB1003

bHomology to Escherichia coli K12 proteins

Discussion

Applications of acyl-CoA thioesterases in biotechnology are steadily increasing as several publications for the production of hydroxylated fatty acids (Zheng et al. 2004) and carboxylic acids (Chung et al. 2009; Liu et al. 2007; Martin et al. 2013; Martin and Prather 2009) by microorganisms illustrate. Many production systems were established with the help of thioesterase B/II (TesB), which is beside thioesterase A/I (TesA) one of the most thoroughly investigated thioesterases including a 3D X-ray structure (Li et al. 2000). Its broad substrate spectrum (Barnes et al. 1970; Naggert et al. 1991) turns TesB to an universally applicable enzyme which, however, might also be often used due to lacking alternatives especially for short-chain acyl-CoAs. Our results show that TesB does not necessarily have to be the thioesterase of choice for cleaving all types of acyl-CoAs. Bacterial thioesterase YciA showed higher CoA-hydrolyzation activities towards five tested EMCP CoA esters than TesB from E. coli (see Table 2). This might be due to the branched and unsaturated carbon chains of the CoA esters as most of the substrates investigated for TesB (Barnes et al. 1970; Martin and Prather 2009; Naggert et al. 1991; Zheng et al. 2004) consist of linear hydroxylated carbon chains such as hydroxyvalerate, hydroxydecanoate or hydroxybutyrate. The unobservable EMCP derived dicarboxylic acid release of AM1_pCM160-tesBec underline the low activities of TesB from E. coli measured in vitro.

Annotated TesB of M. extorquens AM1 (TesBext (Vuilleumier et al. 2009)) shares 45 % identity and 60 % positives to its E. coli homologue. TesBext prefers longer acyl-CoAs as substrates as well as TesBec does, revealed by in vitro enzymatic analysis (Table 2). Enzyme properties for EMCP CoA esters are in the similar orders of magnitude as well. In contrast, constitutive expression of tesBext does not have a lethal effect as tesBec, which is probably a result of a disrupted metabolism as the broad range thioesterase TesB might act on various CoA esters pools of M. extorquens AM1. The preferred codon usage of tesBext in contrast to tesBec makes a malfunction of expression highly improbable although TesBext protein occurrence in AM1_pCM160-tesBext was not directly assured.

The selected specific short-chain acyl-CoA hydrolases Acot4 and Bch showed low activities towards all EMCP CoA esters and very high Km values in case of Bch (see Table 2). Although small amounts of mesaconate were detectable in AM1_pCM160-acot4 supernatant, Acot4 was not further applied due to the higher yield of the same product by AM1_pCM160-yciA. Although the measured Km values of 3-hydroxyisobutyryl-CoA hydrolase (Bch) might suggest it as a candidate for specific rac-ethylmalonyl- or rac-methylmalonyl-CoA cleavage, the supernatant dicarboxylic acid profile of AM1_pCM160-bch was not altered. The only described application of Bch which is also the first publication of this enzyme so far does not provide any further information on other substrates than 3-hydroxyisobutyryl-CoA (Lee et al. 2008). Homologues from Rattus norvegicus or Homo sapiens show only very limited activities towards structural related compounds like methylmalonyl-CoA or 3-hydroxybutyryl-CoA or are even inhibited by these CoA esters (Shimomura et al. 1994; Shimomura et al. 2000), which might also be the case for Bch.

There might be additional promising candidates for EMCP CoA-ester hydrolyzation, but most characterized acyl-CoA thioesterases which could be an alternative show too low or no activity on acyl-CoA esters with six or less C atoms, e.g., TesA (Bonner and Bloch 1972; Cho and Cronan 1993), EntH (Chen et al. 2009), YbgC (Zhuang et al. 2002) or thioesterase III (Nie et al. 2008). Other candidates being active on short-chain acyl-CoAs like Acot12 seem to be difficult to express in bacteria (Suematsu et al. 2002).

The release of 2-methylsuccinate and mesaconate into culture supernatant of M. extorquens AM1 expressing yciA confirms that the bacterial thioesterase is also active in vivo. The cleavage of mesaconyl-CoA is consistent with the measured in vitro activity towards this substrate which was the highest among the EMCP CoA esters (Table 2). Note that an isomeric mixture of mesaconyl-C1– and mesaconyl-C4-CoA was used for in vitro enzyme assays whereas only mesaconyl-C1-CoA is present in vivo. Though not all values measured in vitro fit exactly to the in vivo observations as the in vitro activities towards succinyl- and rac-methylmalonyl-CoA are higher than those for rac-methylsuccinyl-CoA, but a release of succinate or methylmalonate was not detectable. This might be due to various reasons: the intracellular environment can be very different to the in vitro assay conditions such as pH, salt concentrations and additional inhibitor or activator substances shifting the enzyme activities and substrate preferences. Moreover, the pool sizes of the EMCP CoA esters in M. extorquens AM1 growing on methanol are rather low — between 16 and 95 μM (Kiefer 2009) — additionally weighting any shifted Km values and activities resulting from in vivo conditions.

The drop of mesaconate and 2-methylsuccinate concentrations after the end of the exponential growth phase in culture supernatant of M. extorquens expressing yciA (Fig. 3) demonstrates the uptake of these acids. M. extorquens AM1 is able to use several C4– dicarboxylic acids of the TCA and methylmalonate of the EMCP as sole carbon source as well as α-ketoglutarate and ethylmalonate, 2-methylsuccinate and mesaconate which consist of 5C atoms (Table 3). Cellular acid uptake may be triggered upon transition from exponential to stationary growth phase. Another possibility might be a permanent acid uptake which is lower than the release during exponential growth phase but higher during stationary growth phase where the metabolic flux towards the CoA-ester substrates ceases. As M. extorquens AM1 is able to co-consume methanol and succinate (Peyraud et al. 2012) it might also co-consume other acids with methanol. Because of the CoA ester forms of all EMCP intermediates, the acids may be condensed to a CoA molecule by CoA-ligases or -transferases before they are further metabolized.

Several dicarboxylic acid uptake systems exist in bacteria such as the DctA family mainly occurring in aerobic bacteria, the CitT family, SdcS described for Staphylococcus aureus, the DcuAB family present in anaerobic bacteria or the tripartite ATP-independent periplasmic (TRAP) transport carriers (Janausch et al. 2002). An α-ketoglutarate permease was identified in E. coli by Seol and Shatkin (1991). M. extorquens AM1 very likely harbors the DctA dicarboxylic acid transporter including its regulators DctB and DctD (see Table 4), which are located downstream of annotated dctA gene MexAM1_META1p3271 in the AM1 genome suggesting YP_002964290.1 as most likely DctA homologue. An α-ketoglutarate permease KgpT also seems to be existent in AM1. Proteins with high homology to CitT, CitA, DcuC/AB/SR, DctQM or SdcS were not found.

All of these transporters might be able to take up the EMCP derived acids, and it can be speculated that the succinate importer DctA might be responsible for the uptake of the C4 dicarboxylic acid methylmalonate whereas KgpT transports the other EMCP related dicarboxylic acids ethylmalonate, methylsuccinate and mesaconate which consist of 5C atoms such as α-ketoglutarate. Investigations of Van Dien and coworkers (2003) show that an M. extorquens AM1 deletion mutant for dctA is not able to grow on succinate or pyruvate anymore. Therefore application of this mutant — alone or in combination with a kgpT knockout — might be interesting to test towards their EMCP acid uptake.

In summary, we identified bacterial thioesterase YciA which has not been used to date for biotechnological applications as the first thioesterase shown to cleave EMCP CoA esters (2S)-methylsuccinyl-CoA and mesaconyl-C1-CoA. Consequently, we also report the production of the corresponding acids 2-methylsuccinate and mesaconate by expression of yciA in EMCP harboring M. extorquens AM1 for the first time. Increasing CoA-ester concentrations de novo by metabolic engineering will lead to further production increase of mesaconate or 2-methylsuccinate and might also reveal other thioesterases as being able to release other CoA-ester related acids.

Acknowledgments

This work was funded by the European Union in the context of PROMYSE research project (FP7-KBBE.2011.3.6-04). We thank Dr. Tobias Jürgen Erb (Swiss Federal Institute of Technology Zurich, Institute of Microbiology) for his very useful advices on the synthesis and analytics of CoA esters.

Copyright information

© Springer-Verlag Berlin Heidelberg 2014