Metabolic pathway engineering for production of 1,2-propanediol and 1-propanol by Corynebacterium glutamicum

Background Production of the versatile bulk chemical 1,2-propanediol and the potential biofuel 1-propanol is still dependent on petroleum, but some approaches to establish bio-based production from renewable feed stocks and to avoid toxic intermediates have been described. The biotechnological workhorse Corynebacterium glutamicum has also been shown to be able to overproduce 1,2-propanediol by metabolic engineering. Additionally, C. glutamicum has previously been engineered for production of the biofuels ethanol and isobutanol but not for 1-propanol. Results In this study, the improved production of 1,2-propanediol by C. glutamicum is presented. The product yield of a C. glutamicum strain expressing the heterologous genes gldA and mgsA from Escherichia coli that encode methylglyoxal synthase gene and glycerol dehydrogenase, respectively, was improved by additional expression of alcohol dehydrogenase gene yqhD from E. coli leading to a yield of 0.131 mol/mol glucose. Deletion of the endogenous genes hdpA and ldh encoding dihydroxyacetone phosphate phosphatase and lactate dehydrogenase, respectively, prevented formation of glycerol and lactate as by-products and improved the yield to 0.343 mol/mol glucose. To construct a 1-propanol producer, the operon ppdABC from Klebsiella oxytoca encoding diol dehydratase was expressed in the improved 1,2-propanediol producing strain ending up with 12 mM 1-propanol and up to 60 mM unconverted 1,2-propanediol. Thus, B12-dependent diol dehydratase activity may be limiting 1-propanol production. Conclusions Production of 1,2-propanediol by C. glutamicum was improved by metabolic engineering targeting endogenous enzymes. Furthermore, to the best of our knowledge, production of 1-propanol by recombinant C. glutamicum was demonstrated for the first time.


Background
The usage of 1,2-propanediol ranges from building blocks in plastics industry, in de-icing and anti-freeze fluids, and as additive in cosmetics, nutrition, medicines, dyes, and liquid detergents [1]. Due to the very broad spectrum of applications of the bulk chemical 1,2-propanediol, also known as propylene glycol, annually over 1 billion pounds of 1,2-propanediol are sold in the United States and at least 1.2 million tons are consumed worldwide [2]. To date, most of this demand is accommodated by petrochemistry. In the main route, the steam cracking product propylene [3] is converted to propylene oxide [4,5], which is further hydrolyzed to 1,2-propanediol [6]. The occurrence of toxic intermediates and sideproducts initiated efforts to find more sustainable and less toxic routes, e.g., by fermentation of renewable carbon sources by microorganisms. Various microorganisms showing potential to produce 1,2-propanediol from renewable feed stocks have been described, e.g., Clostridium thermosaccharolyticum [7], Saccharomyces cerevisiae [8,9], Escherichia coli [1,10], Synechoccus elongates [11], and Corynebacterium glutamicum [12].
The Gram-positive and generally-recognized-as-safe rod-shaped soil bacterium Corynebacterium glutamicum [13] is the main source of the worldwide production of the amino acids glutamate and lysine in a scale of over 5 million tons per year [14]. A wealth of information on C. glutamicum exists [14][15][16][17][18] including sequencing its genome [19] and creating a genome-streamlined chassis organism [20]. Metabolic engineering aimed at the production of not only many other amino acids [14,21] but also for example at monomers of bioplastics (e.g., cadaverine [22,23] and putrescine [23]), organic acids [24], carotenoids [25], and biofuels. C. glutamicum was engineered for isobutanol production and shown to exhibit less toxicity to isobutanol than E. coli [26,27]. The isobutanol yield by recombinant C. glutamicum was competitive with E. coli [28]. In particular, overproduction of the biofuel ethanol under oxygen deprivation conditions is well-described for C. glutamicum and shown to be efficient [29][30][31]. Importantly, under these conditions C. glutamicum showed high tolerance to organic acid, furan, and phenolic inhibitors present in lignocellulose hydrolysates [30]. Thus, C. glutamicum is a promising alternative biofuel production host. To enable sustainable production from several alternative carbon sources, the substrate spectrum of C. glutamicum was widened by metabolic engineering [32]. Since 1,2-propanediol production by C. glutamicum has been shown [12] in principle, this study aimed at improving 1,2-propanediol production and at producing 1-propanol as derived compound. This primary alcohol, also named npropanol, finds application in the solvent, cosmetic, and pharmaceutical industries, in antiseptic solutions, as precursor for diesel fuels and in the plastics industry and finally as biofuel [33][34][35]. C. glutamicum has previously been engineered for production of the biofuels ethanol [31] and isobutanol [26][27][28] but not for 1-propanol. Natural microorganisms are not known to secrete significant amounts of 1-propanol. However, Propionibacterium freudenreichii has been engineered for the direct conversion of propionyl-CoA to 1-propanol [34]. Engineered E. coli strains either convert 2-ketobutyrate to 1-propanol by variants of the threonine and citramalate pathways [36,37] or by extending succinate dissimilation [35]. Finally, 1,2-propanediol can be converted in a two-step conversion to 1-propanol by diol dehydratase from Klebsiella oxytoca [33]. The latter pathway was chosen in this study for construction of a C. glutamicum 1-propanol-producing strain.

Results
Co-overexpression of yqhD from E. coli increased 1,2-propanediol production C. glutamicum has previously been engineered for 1,2propanediol production by expressing the heterologous genes mgsA and gldA encoding methylglyoxal synthase gene and glycerol dehydrogenase from E. coli [12]. Expression of these genes as artificial operon from the plasmid pEKEx3-mgsA-gldA in C. glutamicum WT yielded 19 ± 1 mM 1,2-propanediol within 51 h (Fig. 2) when using modified CGXII minimal medium with a decreased nitrogen content (5 g/L ammonium sulfate) and 184 ± 1 mM glucose as sole carbon source. Thus, the base strain produced 1,2-propanediol with a yield of 0.103 mol/mol glucose.
Deleting ldh prevented transient L-lactate accumulation and led to faster and higher 1,2-propanediol production The deletion of hdpA prevented formation of about 40 mM glycerol but increased 1,2-propanediol accumulation by about 22 mM only (Fig. 3). Since 1,2-propanediol  is more reduced than glycerol and since it is known that C. glutamicum utilizes excess NADH to reduce pyruvate to L-lactate, lactate formation may compete with 1,2-propanediol formation for NADH. In C. glutamicum, L-lactate is formed by fermentative, NADH-dependent lactate dehydrogenase LdhA under oxygen deprivation conditions [44] but transiently also during aerobic cultivation [45]. Re-uptake and re-utilization of lactate does not generate NADH but menaquinol, because both L-and D-lactate dehydrogenases LldD and Dld oxidize lactate to pyruvate in menaquinone-dependent reactions [45,46]. Thus, ldh was deleted and the resulting strain C. glutamicum ΔhdpAΔldh(pEKEx3-mgsA-yqhD-gldA) was compared to strain ΔhdpA(pEKEx3-mgsA-yqhD-gldA) in batch cultivations. As consequence of introducing the ldh deletion, 1,2-propanediol production increased by about 38 %. C. glutamicum strain ΔhdpAΔldh(pEKEx3-mgsA-yqhD-gldA) accumulated 63 ± 4 mM 1,2-propanediol ( Fig. 4b), which corresponds to a product yield of 0.343 mol/mol. Moreover, the ldh deletion strain utilized glucose faster and accumulated 1,2-propanediol faster than the parental strain, while the growth rates of both strains were comparable (Fig. 4a). Neither DHA nor glycerol accumulated to significant concentrations (<5 mM), but more acetol (15 mM as compared to 7 mM) was produced by the ldh deletion strain (Fig. 4b). Lactate formation by the ldh deletion strain was not detectable (<1 mM), while the parental strains and all other strains mentioned in Figs. 2, 3, and 4 accumulated lactate to low concentrations (between 1 and 4 mM) over the whole fermentation process. Taken together, ldh deletion improved 1,2-propanediol production considerably.

Discussion
In this study, production of 1,2-propanediol by C. glutamicum was improved and production of the biofuel molecule 1-propanol by C. glutamicum was shown for the first time. It has been shown previously that expression of the heterologous methylglyoxal synthase gene mgsA from E. coli was required for 1,2-propanediol and had to be coupled with glycerol dehydrogenase either encoded by heterologous gene gldA from E. coli or endogenous cgR_2242 [12]. Within 96 h, up to 25 mM 1,2-propanediol and 44 mM acetol were produced from 333 mM glucose as a sole carbon source [12]. Using a comparable strain but the cultivation setup employed in this study, it was possible to produce 19 mM 1,2-propanediol in 51 h from 184 mM glucose by overexpression of mgsA and gldA from E. coli in C. glutamicum WT (Fig. 2). Notably, accumulation of 1,2-propanediol and side products started after the cells entered the stationary phase, thus, production was not coupled to growth (Fig. 2).
Alcohol dehydrogenase YqhD proved beneficial for 1,2-propanediol production (increased by 27 % to a yield of 0.131 mol/mol glucose, Fig. 2), presumably because conversion of methylglyoxal to acetol and 1,2-propanediol was improved by YqhD. This enzyme has the following characteristics: a reductase activity for at least 12 aldehydes and thus increasing tolerance to aldehydes as aldehyde scavenger; preferring aldehydes over alcohols as substrates; a better conversion of alcohols longer than three carbon atoms; dependence of NADPH/NADP and divalent cations (e.g., zinc) as cofactors [48]. Notably, YqhD is NADPHdependent [48] as compared to the NADH-dependent GldA, thus, YqhD is coupled to anabolic metabolism, which is driven by NADPH. Overexpression of yqhD proved beneficial for production of, e.g., 3-hydroxypropionic acid by E. coli [49], poly(3-hydroxypropionate) from glycerol using engineered Klebsiella pneumoniae [50], short-chain alcohols by E. coli [51], or acetol by E. coli [52].
Heterologous expression of gldA and yqhD from E. coli resulted in production of the side-product glycerol since these aldehyde reductases reduced DHA to glycerol [40]. Two possible enzymes were considered to be involved in the reduction of DHA metabolism, namely cg1497 and hdpA [42,43]. Only the deletion of hdpA prevented glycerol formation and improved 1,2-propanediol production increasing the yield by about 90 % up to 0.249 mol/mol glucose (Fig. 3). The strain lacking endogenous hdpA showed improved 1,2-propanediol production due to two possible advantages. First of all, DHAP is not converted to DHA and, thus, supply of DHAP for the MgsA reaction to methylglyoxal was improved. Secondly, preventing reduction of DHA to glycerol increased provision of the redox cofactor NADH for the reactions converting methylglyoxal to 1,2-propanediol. Formation of glycerol as sideproduct of C. glutamicum strains expressing heterologous gldA and/or yqhD is distinct from glycerol production of C. glutamicum WT. In C. glutamicum WT, glycerol is formed from glycerol 3-phosphate by glycerol 3-phosphate phosphatase Gpp [38]. Since C. glutamicum WT secretes DHA under certain condition [41,42], it is devoid of an enzyme catalyzing reduction of DHA to glycerol as efficient as observed in recombinants expressing heterologous gldA and/or yqhD from E. coli.
With the additional deletion of the gene ldh, it was possible to further increase the 1,2-propanediol production by about 38 % resulting in a yield of 0.343 mol/mol (Fig. 4). Deletion of ldh is a common strategy to improve production of organic acids under oxygen deprivation conditions [53,54] since L-lactate is secreted by C. glutamicum under conditions of excess NADH. Two factors may have led to improved 1,2-propanediol production as result of ldh deletion. Firstly, provision of NADH for reduction of methylglyoxal to acetol and 1,2-propanediol is increased since pyruvate is not reduced to L-lactate. Secondly, pyruvate and possibly also other intermediates of glycolysis may accumulate as consequence of ldh deletion. This accumulation is plausible since deletion of pyruvate kinase Pyk led to accumulation of pyruvate and other glycolytic intermediates [55,56]. In E. coli, methylglyoxal reacts spontaneously with glutathione to form a hemithioacetal, followed by detoxification by the glycoxalase system leading to the production of D-lactate [57]. C. glutamicum lacks glutathione but possesses mycothiol as its primary low molecular weight thiol [58]. A number of mycothiol-dependent reactions have been described for C. glutamicum including formaldehyde oxidation to formate [59,60]. Although the reaction between mycothiol and methylglyoxal is currently not known in C. glutamicum, the overexpression of mshAencoding mycothiol glycosyltransferase led to an increased robustness towards methylglyoxal [61].
Provision of NAD(P)H for reduction of acetol to 1,2propanediol may still be limiting since even strain C. glutamicum ΔhdpAΔldh produced up to 15 mM acetol (Fig. 4). Notably, the accumulation of acetol increased after glucose was depleted while the 1,2-propanediol concentration decreased. Thus, 1,2-propanediol may be taken up again and oxidized to acetol to generate NADH, which may provide the cells with ATP in oxidative phosphorylation. Currently, it is not known whether oxidation of 1,2propanediol occurs via the heterologous GldA from E. coli or by an endogenous enzyme. Interestingly, in a recombinant cyanobacterium producing 1,2-propanediol, alternative NADPH-alcohol dehydrogenases led to higher 1,2-propanediol titers, while acetol was not produced as side-product [11].
Additionally, the production of 1-propanol by C. glutamicum is reported for the first time in this study. Heterologous expression of the operon ppdABC from K. oxytoca encoding diol dehydratase in a 1,2-propanediol producing C. glutamicum strain was required for 1-propanol production of up to 12 mM (Fig. 5). Diol dehydratase PpdABC has the following characteristics: consisting of three subunits (α, β, and γ) with two units of a heterotrimer building the quaternary structure; indicated that the αand γ-subunit promote the correct folding of each subunit; substrates are 1,2-propanediol, glycerol and 1,2-ethanediol with Km values of 0.08 μM, 0.73 mM, and 0.56 mM, respectively; lack of stereospecificity accepting (R)-and (S)-1,2-propanediol; dependent of adenosylcobalamin and divalent cations (e.g., potassium) as cofactors [62][63][64]. The observation that 1,2-propanediol was still the major product (up to 62 mM; Fig. 5) indicated that 1,2-propanediol to is not converted efficiently to 1-propanol by B 12dependent diol dehydratase PpdABC and YqhD. However, vitamin B 12 may be limiting since it is not known if C. glutamicum can synthesize vitamin B 12 . In addition, provision of the cofactor NADPH may be a bottleneck.
There is potential for improving 1-propanol production with C. glutamicum as exemplified for E. coli [33,47]. Overexpression of ppdABC in E. coli BW25113 for conversion of DHAP to 1,2-propanediol yielded 0.036 mol/ mol 1-propanol from glucose [33], which is comparable to the yield of 0.032 mol/mol reported here (Fig. 5). The yield with C. glutamicum doubled as consequence of deleting ldh and hdpA (Fig. 5). Jain et al. (2014) optimized 1-propanol production by E. coli further [47]. The improvements included co-cultivation of one strain converting glucose to 1,2-propanediol and a second strain converting 1,2-propanediol to 1-propanol [47]. The first strain was improved by overexpressing an optimized gene set for conversion of DHAP to 1,2-propanediol and by deleting four genes to improve NADH provision [47]. Furthermore, heterologous expression of a gene coding for formate dehydrogenase and feeding the additional carbon source sodium formate and yeast extract improved the redox balance [47]. The second strain harbored a synthetic diol dehydratase gene cluster with optimized gene order (ppdA-C-B) and separation by linker sequences [47]. These metabolic engineering and medium optimization approaches may be helpful for improving 1-propanol production by the C. glutamicum strains described in this study. A number of engineering strategies to improve NADPH provision in C. glutamicum have been developed and include, e.g., transmembrane transhydrogenase PntAB [65], phosphoglucose isomerase mutants [66], NADPH-dependent glyceraldehyde-3phosphate dehydrogenase [67], or NAD kinase [68]. Thus, production of 1-propanol may be increased further over the proof-of-concept established in this study.

Conclusions
Metabolic engineering improved 1,2-propanediol production by C. glutamicum. Deletion of the endogenous genes hdpA and ldh combined with overexpression of the E. coli genes mgsA, gldA, and yqhD resulted in strain producing 1,2-propanediol from glucose in mineral salt medium with a product yield of 0.343 mol/mol. Further strain engineering led to strain capable of producing 1-propanol. This is the first report of 1-propanol production by recombinant C. glutamicum.

Microorganisms, media, and cultivation conditions
In Table 1, all C. glutamicum strains and plasmids which were used for this study are presented. The E. coli strain DH5α [69] was used for the plasmid construction and was cultured in lysogeny broth complex medium (LB) [70]. Precultivation of C. glutamicum was performed in LB with 2 % glucose by inoculation from LB plates. For the main cultures of C. glutamicum, the cells of an overnight preculture were harvested by centrifugation (10 min; 3220 × g) and transferring the appropriate volume for an optical density (λ = 600 nm) (OD 600 ) of 1 in 50-mL cultures. These cells were washed with CGXII minimal medium [71] without carbon source and without urea and ammonium sulfate. The cells were again centrifuged and resuspended with the same CGXII. As sole nitrogen source 5 g/L ammonium sulfate were added and as sole carbon source, glucose was used in the measured concentration given in the results. All cultivations of C. glutamicum were carried out in a volume of 50 mL in 500-mL baffled flasks at 30°C and 120 rpm. The gene expression was induced by adding 1 mM isopropylβ-D-thiogalactopyranoside (IPTG) at inoculation of the main culture. When appropriate, the medium was supplemented with 25 μg/mL kanamycin and 100 μg/mL spectinomycin. For 1-propanol production, it was necessary to add 10 μM of vitamin B 12 to the medium. Growth was observed by measuring the OD 600 using the V-1200 spectrophotometer (VWR International, Darmstadt, Germany) by diluting the samples into an OD 600 range of 0.05-0.25. Additionally, 1-mL samples were taken at the time points given in the results and centrifuged (10 min; 16.000 × g), and the resulting supernatants were stored at −20°C until further analysis.

Recombinant DNA work
All oligonucleotides used in this study were obtained from Eurofins MWG Operon (Ebersberg, Germany) or metabion international AG (Planegg, Germany) ( Table 2). The plasmid construction was carried out with PCR fragments (KOD, Novagen, Darmstadt, Germany) generated with genomic DNA of C. glutamicum WT, E. coli DH5α (DNA preparation described by [72]), or K. oxytoca DSM4798 (DSMZ, Braunschweig, Germany) as template DNA. These Δcg1497 In-frame deletion of cg1497 in C. glutamicum WT This work

ΔhdpA
In-frame deletion of hdpA (cg2474) in C. glutamicum WT This work

Δcg1497ΔhdpA
In-frame deletion of hdpA (cg2474) in C. glutamicum Δcg1497 This work

Construction of C. glutamicum deletion strains
To delete the genes cg1497 and hdpA new plasmids were constructed by using the suicide vector pK19mobsacB [75]. For the deletion of cg1497, genomic regions flanking this gene were amplified via PCR from genomic DNA of C. glutamicum using the primer pairs cg1497_upstrm_ fw_pK19/cg1497_upstrm_rv and cg1497_dwnstrm_fw/ cg1497_dwnstrm_rv_pK19 ( Table 2). The resulting PCR fragments were purified and cloned via Gibson-Assembly into the linearized vector pK19mobsacB resulting in the plasmid pK19mobsacB-Δcg1497 (Table 1). The deletion of the gene cg1497 was carried out with this plasmid by a two-step homologous recombination procedure described before [71]. For the verification of the correct in-frame deletion of the gene cg1497, a PCR (Taq DNA polymerase with ThermoPol® Buffer, NEB, Frankfurt am Main, Germany) was performed using the primer pair Dcg1497_fw_v3/Dcg1497_rv_v3 (Table 2). Accordingly the deletion of hdpA (cg2474) was realized, using the primer pairs nagD_upstrm_fw_pK19/nagD_upstrm_rv and nagD_dwnstrm_fw/nagD_dwnstrm_rv_pK19 (Table 2) for the cloning procedure of the plasmid pK19mobsacB-ΔhdpA (Table 1) and the primer pair DnagD_fw/DnagD_rv (Table 2) for the verification of the in-frame deletion via PCR. The plasmid pK19mobsacB-Δldh (Table 1) was already available [28]. Thus, the primer pair ldhfow/ldhrev ( Table 2) was used to verify the successful in-frame deletion of ldh after the two-step homologous recombination.

GC-MS measurements
The supernatants of the samples taken in the cultivation were analyzed using a TRACE GC ULTRA connected to an AS 3000 Auto-sampler and to an ISQ Single Quadrupole Mass Spectrometer using a TG-WAXMS (length: 30 m; I.D.: 0.25 mm; film: 0.25 μm) (Thermo Scientific, Dreieich, Germany). The thawed supernatants were directly diluted 1:10 with methanol (HPLC gradient grade; VWR Chemicals, Fontenay-sous-Bois, France) or after an additional 1:10 dilution step with water (Milli-Q grade). Prior to injection, the diluted samples were centrifuged (10 min; 16,000 × g) and the resulting supernatant was used for analysis. The operating setup was the following: the temperature of the MS transfer line and the ion source were hold at 230°C; the injector temperature was set to 220°C and a gradient was used for the oven (holding 40°C for 1 min, increasing the temperature with a rate of 12°C/min up to 230°C and holding this for 5 min); in the constant flow mode, the flow rate of the carrier gas helium was 1 mL/min using the splitless mode of the injector (split flow: 10 mL/min; splitless time: 1.5 min; focus liner: 5 × 8 × 105 mm, splitless for 50-mm needle with glass wool); the electron impact ionization energy was 70 eV. The compounds 1,2-propanediol and acetol were measured with this method by creating a calibration curve with an external standard. The peaks were identified by retention time and were quantified using the intensity of one specific m/z value (1,2-propanediol: m/z = 45; acetol: m/z = 43). For the computational quantification, the program Xcalibur 2.1 (2.1.0 SP1.1160, Thermo Scientific, Dreieich, Germany) was employed.

HPLC measurements
The compounds glucose, glycerol, DHA, lactate, propanal, and 1-propanol were quantified with a HPLC system (1200 series, Agilent Technologies, Böblingen, Germany). As a immobile phase, an organic acid resin column (300 × 8 mm) with the appropriate pre-column (40 × 8 mm) (Chromatographie-Service GmbH, Langerwehe, Germany) was installed and heated up to 60°C while the mobile phase was 5 mM sulfuric acid in water (Milli-Q grade) with a flow rate of 0.8 mL/min or 1 mL/min. The signals were acquired with a refractive index detector (glucose, glycerol, propanal, and 1-propanol) and a diode array detector at a signal wavelength of 210 nm and a reference wavelength of 360 nm (DHA, lactate). For the calibration curve, external standards for every compound were prepared and the supernatants of the samples from the cultivations were measured undiluted after thawing.