Efficient Production of 2,3-Butanediol from whey Powder by Metabolic Engineered Klebsiella Oxytoca

Backgrounds: Whey is the major pollution source from the dairy industry. Exploring new outlets for whey utilization is urgently needed to decline its environmental pollution. In this study, we explored the possibility of using whey powder to produce 2,3-butanediol (2,3-BD), an important platform chemical. Results: A Klebsiella oxytoca strain PDL-0 was selected from five 2,3-BD producing strains based on its ability to efficiently produce 2,3-BD from lactose, the major fermentable sugar in whey. Five genes including pox , pta , frdA , ldhD , and pflB were knocked out in K. oxytoca PDL-0 to decrease the production of byproducts like acetate, succinate, lactate, and formate. Using fed-batch fermentation of K. oxytoca PDL-0 Δ pox , 74.9 g/L 2,3-BD was produced with a productivity of 2.27 g/L/h and a yield of 0.43 g/g from lactose. In addition, when whey powder was used as the substrate, 65.5 g/L 2,3-BD was produced within 24 h with a productivity of 2.73 g/L/h and a yield of 0.44 g/g. Conclusion: This study proved the efficiency of K. oxytoca PDL-0 to metabolize whey for 2,3-BD production. Due to its characteristics of non-pathogenicity and efficient lactose utilization, K. oxytoca PDL-0 might also be used in the production of other important chemicals using whey as the substrate.


Background
Whey is a liquid byproduct in cheese making process which contains most of the water-soluble components in milk [1,2]. Despite its annual production of 145 million tons worldwide, only a little over one-half of the whey produced is utilized [3]. Whey is regarded as a serious pollutant because of its high biochemical oxygen demand (30,000-50,000 mg/L) and chemical oxygen demand (60,000-80,000 mg/L) [3]. Economic disposal of whey becomes a world-wide problem of the dairy industry. Lactose, a utilizable disaccharide of many microbial strains, is the major contributor to biochemical oxygen demand and chemical oxygen demand in whey [4,5]. Using the lactose in whey as a substrate for microbial fermentation may transform a potential pollutant into a value-added product and deserves an intensive study.
2,3-Butanediol (2,3-BD) is an important platform chemical which can be applied in many industrial fields [6 − 8]. It is estimated that the derivatives of 2,3-BD have a potential global market of around 32 million tons per year. Nowadays, the common method for 2,3-BD synthesis is a chemical route conducted under a harsh condition (160-220 °C, 50 bar) with a C 4 hydrocarbon fraction of crack gases as the substrate [9,10]. Due to the shortage of fossil fuels and increasing global environmental concerns, the green production of 2,3-BD through microbial fermentation using renewable resources has received great attentions [11 − 16]. Several 2,3-BD producing microorganisms can use fermentable sugars including glucose, xylose, fructose, and lactose as the sole carbon source for growth [17 − 20]. However, these strains exhibited unsatisfactory fermentative performances of 2,3-BD production when using lactose as the carbon source. For example, Klebsiella oxytoca NRRL-B199 can use the mixture of glucose and galactose as substrate for growth and produce 2,3-BD as its main product. Nevertheless, 2,3-BD was present in a low concentration and the strain produced acetate as the major product in the fermentation broth with lactose [21,22]. Thus, it is of vital necessity to select suitable microbial strains with potential of efficient 2,3-BD production from lactose.
In this study, five strains including Klebsiella pneumonia ATCC 15380, Enterobacter cloacae SDM, Bacillus licheniformis DSM13, K. oxytoca PDL-0 and Escherichia coli BL21-pETRABC were cultured in fermentation broth with lactose as the carbon source. K. oxytoca PDL-0 was found to possess the best performance in lactose utilization and 2,3-BD production. Then, byproduct-producing genes including pox, pta, frdA, ldhD, and pflB in K. oxytoca PDL-0 were knocked out to improve the efficiency of 2,3-BD production from lactose. Finally, high production of 2,3-BD from whey powder was achieved through fed-batch fermentation using the recombinant strain K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB (Fig. 1).
All of the five strains can grow in M9 medium supply with 5 g/L yeast extract and about 40 g/L lactose. B. licheniformis DSM13 is the only strain that can not consume lactose ( Fig. 2a and Fig. 2b). E. cloacae SDM and E. coli BL21-pETRABC could efficiently utilize lactose (about 30 g/L) but these two strains only accumulated about 2 g/L 2,3-BD ( Fig. 2b and Fig. 2c). K. pneumonia ATCC 15380 and K. oxytoca PDL-0 can produce 2,3-BD from lactose, with a yield of 0.21 g/g and 0.30 g/g lactose, respectively. Considering the fact that K. oxytoca PDL-0 belongs to the Risk Group 1 [15] and produces 2,3-BD with a higher yield from lactose, this strain was selected for further study in successive experiments.
In K. oxytoca PDL-0, the formation of acetate, succinate, lactate, and formate is catalyzed by pox and pta, frdA, ldhD, and pflB, respectively [24]. To achieve higher yield of 2,3-BD, these genes were successively deleted in strain K. oxytoca PDL-0 (Additional file 1: Fig. S1). Effects of these genes deletion on growth, lactose consumption, by-product accumulation, and 2,3-BD production were studied in M9 medium supply with 5 g/L yeast extract and about 40 g/L lactose. As shown in Fig. 3a and Fig. 3b, deletion of all these by-product pathways in K. oxytoca PDL-0 had no effect on lactose consumption but slightly increased its growth. Accumulation of byproducts including acetate, succinate, lactate, and formate reduced remarkably due to deletion of pox, pta, frdA, ldhD, and pflB (Fig. 3c). The final strain K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB exhibited higher concentration and yield of 2,3-BD ( Fig. 3d and Fig. 3e) and lower byproducts production ( Fig. 3c) than other recombinant strains.

Performance of recombinant strain in 1-L batch fermentation
Then, the effects of inactivation of by-product pathways on 2,3-BD production were further studied through batch fermentation in a 1-L fermenter. The strains K. oxytoca PDL-0 and K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB were cultured in a fermentation medium containing corn steep liquor powder as a cheap nitrogen source and about 40 g/L lactose as carbon source. As shown in Fig. 4a and 4b, K. oxytoca PDL-0 consumed 42.75 g/L lactose and produced 15.26 g/L 2,3-BD with a yield of 0.36 g/g at 12 h, while K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB consumed 39.29 g/L lactose and produced 17.65 g/L 2,3-BD with a yield of 0.45 g/g. Thus, the recombinant strain K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB possesses advantages over wild type in both concentration and yield of 2,3-BD.

Utilization of lactose for 2,3-BD production in fed-batch fermentation
To achieve a higher product concentration, fed-batch fermentation using strain K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB with an initial lactose concentration of 100 g/L was conducted. Fermentation medium containing corn steep liquor was used in a 7.5-L fermenter. As shown in Fig. 5a, 173.2 g/L lactose was consumed and 74.9 g/L 2,3-BD was produced within 33 h. The productivity was 2.27 g/L/h, and the yield was 0.43 g/g lactose. The concentration of acetate, which was included in the medium, was 0.59 g/L at the end of the fermentation. The concentration of lactate, which was also included in the medium, decreased to 0.13 g/L at 33 h. The final concentration of succinate was 0.82 g/L and there was no formate production throughout the fermentation process (Additional file 1: Fig. S2a).

Utilization of whey powder for 2,3-BD production in fed-batch fermentation
Fed-batch fermentation using whey powder as the carbon source by strain K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB was also carried out. After 24 h of fermentation, 65.5 g/L 2,3-BD was obtained from 148.3 g/L lactose (Fig. 5b). The productivity and yield of 2,3-BD were 2.73 g/L/h and 0.44 g/g, respectively. The major by-products in final fermentation broth were acetate and lactate, which were found at concentrations of 3.24 g/L and 0.38 g/L, respectively (Additional file 1: Fig. S2b).
Several microbial strains have been screened to produce 2,3-BD from whey or lactose. However, as shown in Table 1, the final concentration and yield of 2,3-BD produced by wild type isolates were relatively low. For example, Vishwakarma tried to use strain K. oxytoca NRRL-13-199 for 2,3-BD production from whey. After the addition of 50 mM acetate, a concentration of 8.4 g/L 2,3-BD was acquied with a yield of 0.365 g/g lactose [25]. Barrett et al studied production of 2,3-BD from whey by K. pneumoniae ATCC 13882 [20]. After 60 h of fermentation, 19.3 g/L 2,3-BD was produced from whey with a productivity of 0.32 g/L/h. Ramachandran et al got a concentration of 32.49 g/L 2,3-BD from lactose by using K. oxytoca (formerly known as Aerobacter aerogenes or K. pneumoniae ATCC 8724), however, the yield (0.207 g/g lactose) and productivity (0.861 g/L/h) of 2,3-BD were still unsatisfactory [26]. In a previous work, Lactococcus lactis MG1363 was metabolic engineered to produce 2,3-BD from residual whey permeate and the final titer of 51 g/L was acquired [27]. Exogenous antibiotics was needed for the maintenance of two plasmids pJM001 and pLP712, which respectively carries the genes needed for 2,3-BD production and metabolism of lactose. To make bio-based 2,3-BD production from whey more economically efficient and environment-friendly, 2,3-BD production without antibiotic addition in the fermentation system for the maintenance of plasmid should be initiated. In this work, K. oxytoca PDL-0 was metabolic engineered to efficiently produce 2,3-BD from lactose through deleting pox, pta, frdA, ldhD, and pflB. Using whey powder as the carbon source, the recombinant strain can produce 65.5 g/L 2,3-BD (Table 1). Compared with other strains used for 2,3-BD production from whey, the engineered strain has significant production advantages such as high product concentration (65.5 g/L), high productivity (2.73 g/L/h), and unnecessary exogenous antibiotics. Table 1 Comparison of 2,3-BD production using whey/lactose as substrate by different microorganisms. Recently, lactose or whey have been used to produce various biochemicals, e.g., ethanol [28], butanol [29], lactic acid [30], citric acid [31], poly(3-hydroxybutyrate) (PHB) [32], and gluconic acid [33], through endogenous or exogenous biosynthetic pathways (Table 2). However, because of the low utilization efficiency of lactose in these chassis cell, it is difficult to produce the target chemicals with high productivity and high yield [29,31]. Ahn et al constructed a fermentation strategy with cell recycle membrane system for the production of PHB from whey [32]. High consumption rate of lactose (7.67 g/L/h) was acquired using this complicated fermentation strategy. In this work, the engineered K. oxytoca PDL-0 was confirmed to have the ability to efficiently transform lactose into 2,3-BD with relatively high yield (0.44 g/g) and high consumption rate of lactose (6.18 g/L/h) ( Table 1 and Table 2). Considering its excellent characteristics of non-pathogenicity (Risk Group 1) and efficient lactose utilization, K. oxytoca PDL-0 might be a promising chassis for production of various chemicals from whey through metabolic engineering.

Conclusions
In this study, the ability of K. oxytoca PDL-0 to metabolize lactose and produce 2,3-BD was firstly identified. Then, by-product pathways encoding genes in K. oxytoca PDL-0 was knockout to improve the yield of 2,3-BD. The engineered strain K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB could utilize whey powder as the substrate for high production of 2,3-BD. The process developed here may be a promising alternative for both biotechnological production of 2,3-BD and whey utilization.

Enzymes and chemicals
FastPfu DNA polymerase was purchased from TransGen Biotech (Beijing, China) and T4 DNA ligase from Thermo Scientific (Lithuania). Restriction enzymes were purchased from TaKaRa Bio Inc. (Dalian, China). Polymerase chain reaction (PCR) primers were provided by Tsingke Biology Co., Ltd (QingDao, China). Racemic acetoin (AC) and 2,3-BD was purchased from Apple Flavor & Fragrance Group (Shanghai, China) and ACROS (The Kingdom of Belgium), respectively. Whey powder was purchased from KuoQuan Biotech (Shandong, China). All other chemicals were of analytical grade and commercially available.

Bacterial strains, plasmids and culture medium
The strains and plasmids used in this study are listed in Table 3. All engineered strains used in this work are based on K. oxytoca PDL-0 and its derivatives. E. coli S17-1 was used to hold and amplify plasmids as well as for conjugation with K. oxytoca. The plasmid pKR6K Cm was used for gene knockout in K. oxytoca [24]. Table 3 Strains and plasmids used in this study.
Strain or plasmid Characteristic(s) Reference or source

Strain
Escherichia coli S17-1 recA, pro, thi, conjugative strain able to host λ-pir-dependent plasmids [37] Enterobacter cloacae SDM Wild-type [12] E Luria-Bertani (LB) medium was used for the cultivation of all the strains used. The M9 minimal medium [34] supplemented with 5 g/L yeast extract and 40 g/L lactose was used in shake flasks experiments for selection of the efficient 2,3-BD producing strain. The selection medium for single exchange strains of K. oxytoca was M9 minimal medium supplemented with 20 g/L sodium citrate and 40 µg/mL chloramphenicol. The selection medium for double exchange strains of K. oxytoca was solid LB medium supplemented with 15% sucrose.

Knockout the genes of K. oxytoca PDL-0
The primers used for knockout of byproduct-producing genes in K. oxytoca PDL-0 are listed in Additional file 1: Table S1. Vector isolation, restriction enzyme digestion, agarose gel electrophoresis, and other DNA manipulations were carried out using standard protocols [35]. Knockout mutants of K. oxytoca PDL-0 were generated via allele exchange using the suicide plasmid pKR6K Cm [24]. The left and right flanking sequences were amplified from K. oxytoca PDL-0 and then ligated through PCR to get Δpox fragment using primer pairs PΔpox.f (EcoRI)/PΔpox.r (overlap) and PΔpox.f (overlap)/PΔpox.r (BamHI), respectively. The gel-purified Δpox fragments were ligated to the pKR6K Cm digested with EcoRI and BamHI. The resulting plasmid was designated pKDΔpox and introduced into E. coli S17-1. Then, a three-step deletion procedure was applied to select the Δpox mutant after conjugating the pKDΔpox in K. oxytoca PDL-0 as described previously [24]. The pta, frdA, ldhD, and pflB mutants of strain K. oxytoca PDL-0 were generated by using the same procedure and primers listed in Additional file 1: Table S1.

Batch and fed-batch fermentations
Batch fermentations were conducted in a 1-L bioreactor (Multifors 2, Infors AG, Switzerland) with 0.8 L of  Effects of by-product pathway genes knockout when using lactose as the carbon source. Biomass (a), consumption of lactose (b), by-products (c), concentration (d) and yield (e) of 2,3-BD by K. oxytoca PDL-0 and its derivatives were assayed. The experiments were conducted in a 300-mL flask containing 50 mL of M9 minimal medium supplemented with 5 g/L yeast extract and 40 g/L lactose with shaking at 180 rpm for 24 h. The culture temperature was 37 °C. Error bars indicate the standard deviations from three independent cultures.  Fed-batch fermentation using lactose (a) and whey powder (b) as the carbon source. Biomass, consumption of lactose, concentration of 2,3-BD and acetoin (AC) by K. oxytoca PDL-0 ΔpoxΔptaΔfrdAΔldhDΔpflB were assayed. The experiments were conducted in a 7.5-L fermenter containing 5 L of medium with an initial lactose concentration of 100 g/L approximately.