Modular design of metabolic network for robust production of n-butanol from galactose–glucose mixtures

Design of galactose utilization and n-butanol production modules in E. coli

The galactose metabolizing pathway, also known as the “Leloir pathway”, involves several enzymes that catalyze galactose catabolism in addition to glucose metabolism. As shown in Fig. 1, galactose is imported by galactose transporters (encoded by galP, and mglBAC) and is subsequently converted into glucose-1-phosphate through multiple enzymes encoded by the galactose operon (galETKM). Finally, phosphoglucomutase (encoded by pgm) converts glucose-1-phosphate to glucose-6-phosphate to enter glycolysis. In the presence of glucose, however, galactose metabolism is regulated by several factors, including a Gal repressor (GalR), Gal iso-repressor (GalS), and intracellular cyclic AMP, which induces CCR, enabling the preferential utilization of glucose. Although multiple enzymatic steps and the regulatory hierarchy in the pathway limit the rate of galactose utilization and the simultaneous fermentation of many carbohydrates [21], the reconstruction of galactose pathway on chromosome enhanced the galactose utilization rate and the simultaneous assimilation of galactose and glucose [10].

The assembly of both modules for synthetic galactose utilization and n-butanol production was rationally designed to enhance n-butanol production from increased galactose utilization. As the platform strain to introduce each module, we used previously reported E. coli strain JHL59 whose by-product pathways (acetate, lactate, ethanol, succinate) were blocked to amplify flux toward acetoacetyl-CoA [18]. For the galactose utilization module, the native galactose metabolism was engineered by deleting the Gal repressor (encoded by galR) and replacing the nascent regulatory elements on the “Leloir pathway”, including the enzymes encoded by galP, galETKM, and pgm, with predictable regulatory elements such as a synthetic constitutive strong promoter (BBa_J23100 from the Registry of Standard Biological Parts, http://partsregistry.org), and an optimized 5′-UTR designed by UTR Designer (http://sbi.postech.ac.kr/utr_designer) to amplify gene expression and avoid the host regulatory system [22]. Consequently, optimized galactose utilization module could be constructed (E. coli GAL_059).

In the case of the n-butanol production module, n-butanol, non-native metabolite in E. coli, is mainly produced by the introduction of four heterologous enzymes from Clostridium acetobutylicum and Treponema denticola with endogenous acetoacetyl-CoA as a starting material. Acetoacetyl-CoA (catalyzed by AtoB in E. coli) is firstly converted to 3-hydroxybutyryl-CoA by 3-hydroxybutyryl-CoA dehydrogenase (encoded by hbd from C. acetobutylicum). The crotonyl-CoA is then produced by crotonase (encoded by crt from C. acetobutylicum) from 3-hydroxybutyryl-CoA. Crotonyl-CoA is further modified to butyryl-CoA by trans-enoyl-CoA reductase (encoded by ter from T. denticola). Subsequently, n-butanol is finally produced with bifunctional aldehyde/alcohol dehydrogenase (encoded by adhE2 from C. acetobutylicum) (Fig. 1). Accompanying deletion of by-product forming pathway, the overexpression of these heterologous enzymes with synthetic promoter and UTRs designed in the previous study [18] was utilized. Accordingly, the strain with both optimized modules could be obtained by transforming the pCDF-BuOH harboring maximized n-butanol production module [18] into E. coli GAL_059 (E. coli GAL_061).

Enhanced n-butanol production of engineered strain with galactose

The E. coli GAL_059 strain was aerobically cultivated to evaluate its metabolic capacity when grown in M9 minimal medium containing galactose as a sole carbon source (Fig. 2a). As expected, the engineered strain showed a 37.2 % higher specific growth rate and a 66.4 % improvement in specific sugar uptake rate (Fig. 2b, c). The higher sugar uptake rate by the engineered strain resulted in the depletion of the galactose supplement in the medium within 14 h. In contrast, the parental strain JHL61 consumed about 60 % of the initial galactose.

Furthermore, the n-butanol production capacity of our engineered strain GAL_061 was evaluated after anaerobic fermentation in galactose-supplemented medium for 48 h. After fermentation, this strain showed a dramatically increased final n-butanol titer of 4.5 g/L, about 2.5-fold higher than that produced by the parental strain. Moreover, the amount of galactose consumed was notably greater for this strain (19 g/L) than for the parental strain JHL61 (7.4 g/L), indicating that intracellular carbon flux was enhanced in the engineered strain (Fig. 3).

In addition, butyrate, which can be catalyzed by endogenous tesB (encoding acyl-CoA thioesterases) [19], was produced abnormally as a by-product in the engineered strain. This can be accounted for accumulation of butyryl-CoA (an intermediate of both n-butanol and butyrate) with increased carbon flux using galactose, probably due to insufficient activity of AdhE2 or the stoichiometric imbalance of reducing equivalent, NADH [23].

Redox re-balancing by tuning the NADH supplementation module

Generally, increasing the availability of intracellular NADH is important to produce reduced metabolites, such as ethanol [24], succinate [25], and n-butanol [26, 27] and may be attainable by expressing yeast NAD⁺-dependent formate dehydrogenase (FDH1), an enzyme that converts formate to NADH and CO₂ to supply additional NADH [25–27]. Our previous work elucidated that the optimal supply of NADH determined the efficiency of n-butanol production and was significantly different in galactose and glucose supplemented medium depending on the assimilation rate of sugars [18]. Moreover, the optimization of redox state was successfully accomplished through fine-tuning of FDH1 based on a predictive method of controlling translation using a biophysical thermodynamic model of the translation initiation process (UTR Designer) [22].

In this manner, the commitment of expression cassettes for various levels of FDH1, termed as NADH supplementation module, was expected to effectively coordinate galactose-utilizing and n-butanol producing modules by optimization of intracellular redox state. To find the optimal redox state for n-butanol production equipped with galactose module, fdh1 variants showing different levels of expression [18] were used to transform the GAL_061 strain, finally resulting in the GAL_080, 081, 082, 083, and 084 strains (Additional file 1: Table S1). After fermentation of the variants for 48 h, the relationship between n-butanol and fdh1 expression level titer exhibited a concave curve (Fig. 4a). The carbon flux towards n-butanol was enhanced up to 22 % depending on the enzymatic activity of FDH1, indicating that the GAL_083 strain was redox balanced.

Optimizing the production of n-butanol increased the consumption of galactose and decreased the production of oxidized metabolites, such as pyruvate and butyrate, consistent with previous findings (Additional file 1: Table S3). Moreover, the optimal redox state in galactose-supplemented medium was markedly higher than that of the parental strain, indicating that increased carbon flux elevated the availability of NADH in our engineered strain [18]. Ultimately, the redox balanced strain GAL_083 produced 6.2 g/L of n-butanol after 48 h of fermentation (Fig. 4b). Collectively, the redox balancing module successfully optimized the catalytic amounts of reducing equivalents depending on carbon flux, resulting in maximum production of n-butanol from galactose.

Simultaneous utilization of galactose and glucose for n-butanol production

Typically, biomass consists of a variety of polysaccharides with the composition depending on species, cultivation conditions, and methods of pretreatment [28–31]. Also, galactose is always obtainable as a sugar mixture of glucose from hydrolysate of seaweed, dairy waste [8, 9]. Therefore, the efficient conversion of cost-effective feedstock into valuable products requires the optimal, robust, and simultaneous consumption of galactose and glucose, regardless of their composition [11]. However, most microorganisms have evolved to utilize a preferred sugar in the presence of other sugars, enabling them to grow quickly under nutrient limited conditions [32]. The galactose metabolic pathway is repressed, both at the transcriptional and translational levels by glucose metabolism [33, 34].

We previously observed that deregulation and constitutive expression of transcripts whose 5′-UTR were designed for maximized expression allowed galactose to be utilized in the presence of glucose [10]. Accordingly, our engineered strain GAL_083 was expected to produce n-butanol through co-fermentation of galactose and glucose. To confirm this, GAL_083 and its parental strain were cultivated in simulated synthetic medium containing the same amounts of galactose and glucose that are generally present in seaweed and lactose [35]. Additionally, the ratios of these two sugars were varied from 2:1 to 1:2, reflecting the possible changes in the medium [29, 30, 36].

Under these conditions, the engineered strain utilized galactose and glucose simultaneously, while the parental strain showed no galactose consumption until glucose was depleted (Fig. 5). Moreover, n-butanol production by the engineered strain was consistently above 5 g/L regardless of the composition of the medium, while the yield of metabolites produced by the parental strain varied. Taken together, these results indicate that this strain would be promising as a platform strain for the optimal and robust production of valuable bio-chemicals from various galactose–glucose mixtures without regard to possible changes in the composition.

Reagents, bacterial strains, and plasmids

The E. coli strains, plasmids, and primers used in this study are listed in Additional file 1: Table S1. Phusion DNA polymerase was purchased from New England Biolabs (Beverly, MA, USA). Oligonucleotides were synthesized by Genotech (Daejeon, Korea) and listed in Additional file 1: Table S2. Amplified DNA fragments were purified by GeneAll^R Expin^TM Gel SV kit (GeneAll Biotechnology, Seoul, Korea). Plasmids were prepared by AccuPrep^R Nano-Plus Plasmid Mini Extraction Kit (Bioneer, Daejeon, Korea). The reagents for cell culture were purchased from BD Biosciences (Sparks, MD, USA). Other reagents were obtained from Sigma (St. Louis, MO, USA).

All chromosomal manipulations were performed using the Red recombination with pKD46 and pCP20 [37, 38]. The synthetic galactose utilization pathway was constructed by replacing the nascent pathway of the JHL59 strain with synthetic regulatory elements, as described [10, 39]. To construct the GAL_059 strain, galR was deleted by insertion of a FRT-Kan ^R -FRT fragment amplified using D-galR-F/D-galR-R primers. The native galETKM operon was deleted by insertion of a FRT-Kan ^R -FRT fragment amplified using D-galETKM-F/D-galETKM-R primers and the resultant synthetic operon was introduced by PCR amplifying of pACYC_galO as a template with the O-galETKM-F/O-galETKM-R primers. To overexpress galP and pgm, FRT-Kan ^R -FRT fragments were amplified by O-galP-F/O-galP-R1/O-galP-R2 and O-pgm-F/O-pgm-R, respectively. The GAL_059 strain, which had deleted competing pathways (adhE, ldhA, paaFGH, frdABCD, pta) and overexpressed the producing pathway (atoB, lpd-G1060A, aceEF), was transformed with pCDF-BuOH containing heterologous n-butanol producing enzymes (crt, hbd, ter, adhE2) to yield the GAL_061 strain. GAL_080~084 strains harboring fdh1 variants for redox optimization were prepared by electrical transformation of the GAL_061 strain with the plasmids designated in Additional file 1: Table S1.

Media and growth conditions

To evaluate galactose assimilation rates, E. coli strains were cultivated in M9 minimal medium supplemented with galactose (4 g/L), M9 salt solutions, 5 mM MgSO₄, and 0.1 mM CaCl₂. The cells were subsequently cultivated in modified Terrific Broth, consisting of 12 g tryptone, 24 g yeast extract, 2.31 g KH₂PO₄, and 12.54 g K₂HPO₄ per liter, supplemented with a 25 g/L carbon source for n-butanol production, but without added glycerol. Multiple plasmids were maintained by including 25 μg/mL of streptomycin and 15 μg/mL of kanamycin in the media; single plasmid was maintained in media containing 50 μg/mL alone. The bacteria were cultured anaerobically in rubber-sealed, 60 mL serum bottles and an anaerobic chamber (Coy Laboratories, Ann Arbor, MI, USA) containing 97.5 % nitrogen and 2.5 % hydrogen gas. Oxygen was continuously removed by reaction with hydrogen in the presence of a palladium catalyst. The cells were cultured at 37 °C with shaking (250 rpm), and bacterial density was determined by measuring optical density at 600 nm (OD₆₀₀) using a UV-1700 spectrophotometer (Shimadzu, Kyoto, Japan).

Seed cultures were aerobically prepared by inoculating colonies from an LB plate into 3 mL of galactose-containing M9 minimal medium. This initial seed culture was again inoculated under the same conditions with an OD₆₀₀ of 0.05. When the OD₆₀₀ of second-round seed cultures reached 1.0, the cultures were individually inoculated into M9 or modified TB. Unless otherwise stated, all bacteria were cultivated anaerobically. The culture pH was adjusted to around 7.2 using 10 M NaOH at every 6 h. Samples were frozen at −80 °C until analysis. All cell culture experiments were performed in duplicate.

Analytical methods

The concentrations of sugars and metabolites were determined by UltiMate^TM 3000 analytical HPLC systems (Dionex, Sunnyvale, CA, USA) using an HPX-87H column (Bio-Rad Laboratories, Richmond, CA, USA) with a flow rate of 0.6 mL/min at 14 °C using 5 mM H₂SO₄ as the mobile phase. Signals were monitored with a Shodex RI-101 refractive index detector (Shodex, Klokkerfaldet, Denmark) and a UV–Vis diode array detector at 210 nm.