Thermodynamic analysis of glycerol catabolic pathway
Gibbs free energy (ΔrG) and equilibrium constant (Keq) of reactions and pathways related to glycerol catabolism were estimated with the help of eQuilibrator, and results are given in Table 1.
DhaD catalyses the conversion of glycerol to DHA, and NAD is used as the cofactor (No. 1 in Table 1). The estimated ΔrG of this reaction was positive, and Keq was lower than 1. It indicated that this reaction prefers glycerol formation rather than DHA formation. At reaction equilibrium condition, the level of DHA would be much lower than that of glycerol.
DHA kinase I that is encoded by dhaK catalyses DHA phosphate (DHA-P) formation and this reaction use ATP as the phosphate donor (No. 2 in Table 1). The estimated ΔrG of this reaction was -13.7 ± 4.3 kJ/mol. Combining this reaction with the reaction of DHA formation from glycerol that is catalysed by DhaD, the ΔrG of this pathway (No. 1+2 in Table 1) was 10.9 kJ/mol. Consider the levels of ATP and ADP in K. pneumoniae were nearly equal . Thus, the pathway of glycerol catabolism that is catalysed by the corresponding two enzymes was not preferred from the thermodynamic aspect. In other words, the reverse pathway of glycerol formation from DHA phosphate was possible.
DHA kinase II catalyses DHA phosphate formation with phosphoenolpyruvic acid (PEP) as the phosphate donor (No. 3 in Table 1). The estimated ΔrG of this reaction was -41.4 ± 4.4 kJ/mol, which was more negative than the reaction that is catalysed by DHA kinase I. Accordingly, the ΔrG of the pathway that catalysed by DHA kinase II and DhaD (No. 1+3 in Table 1) was -16.8 kJ/mol. This glycerol catabolism pathway was preferred from a thermodynamic aspect.
DHA phosphate dephosphorylase (hdpA) of C. glutamicum catalyses the reaction of DHA formation from DHA phosphate (No. 4 in Table 1). DHA phosphate was hydrolyzed to release free phosphate and DHA. This reaction was not linked with ATP or PEP formation. The ΔrG of this reaction was -12.7 ± 4.3 kJ/mol. Combining this reaction with the DhaD catalysed reaction to set up a reverse glycerol catabolism pathway (No. 4-1 in Table 1), and the estimated ΔrG of this pathway was -37.3 kJ/mol. Then, DHA and glycerol formation from DHA phosphate through this pathway was feasible from a thermodynamic aspect.
DHA production from DHA phosphate
The glycerol catabolism pathway that formed by DhaD and Dhak (No. 1+2 in Table 1) has a positive ΔrG. It indicated glycerol might be formed from DHA phosphate, which is an intermediate of the glycolysis. However, there are no reports of glycerol formation from glucose or other sugars by wild type K. pneumoniae.Kp ΔtpiA was constructed to block the conversion of DHA phosphate to glyceraldehyde 3-phosphate. This strain and the wild type of K. pneumoniae were cultured in flasks with glucose as the carbon source and results are shown in Fig. 2.
23 g/L glucose was consumed by the wild-type K. pneumoniae after 9 hours of cultivation. Cell growth was coincided with glucose consumption and 3.2±0.2 OD units of cell density was achieved after glucose exhausted. The main metabolites of the process were 2,3-butanediol and acetoin, and their final titers were 3.6±0.5 and 2.6±0.2 g/L, respectively. Low level of acetic acid and lactic acid were produced in the process (data not shown). Cell growth of Kp ΔtpiA was much slower compared with that of the wild-type strain, and the final cell density was only 1.0±0.1 OD unit. This strain needed 15 hours to consume all the glucose. 2,3-Butanediol and acetoin were still the main metabolites of this strain. But their titers were reduced to 2.1±0.5 and 1.4±0.1 g/L, respectively. Neither DHA nor glycerol were detected in the fermentation broth. Thus, reversing the native glycerol catabolic pathway failed for DHA or glycerol production.
As the DHA phosphate hydrolysis reaction has a high negative ΔrG (No. 4 in Table 1). hdpA of C. glutamicum was heterologously expressed in Kp ΔtpiA to construct Kp ΔtpiA-hdpA. This strain was cultured in flasks and results are shown in Fig. 2.
Cell growth of Kp ΔtpiA-hdpA was slower than that of Kp ΔtpiA. The highest cell density was only 0.5±0.04 OD units. Glucose was exhausted and 2.0±0.1 g/L of 2,3-butanediol and 1.5±0.1 g/L of acetoin were produced after 21 hours of cultivation. DHA was produced in the process, and its concentration was continuously increasing, and a final tier of 7.0±0.3 g/L was obtained. 2.5 g/L of glycerol was also generated in the process. It should be pointed out that glycerol synthesis started later than DHA synthesis. Glycerol synthesis commenced after 15 hours of cultivation, at this time the DHA concentration reached 6.7 g/L. Kp hdpA, wild type strain with overexpression of hdpA, was also cultured. But cell growth and metabolite production of this strain were similar to those of the wild type strain. No DHA nor glycerol was produced by Kp hdpA (data not shown). In summary, a reverse glycerol catabolism pathway was set up based on the knock-out of tpiA and overexpression of hdpA, and DHA and glycerol were produced from glucose through this pathway.
Blocking by-products synthesis in DHA production
Metabolites of Kp ΔtpiA-hdpA include DHA, glycerol, 2,3-butanediol and acetoin. If we consider DHA as the target product, then other metabolites are all by-products. dhaD and gldA encoding isoenzymes of glycerol dehydrogenase that catalyse the conversion between DHA and glycerol. budA encoding acetolactate synthase, which is a key enzyme of 2,3-butanediol and acetoin synthesis pathway. Besides, DHA can be converted to methylglyoxal and further be converted to pyruvate (Fig. 1). Methylglyoxal synthase is encoded by mgsA. These key genes were knocked out individually or combined and pDK6-hdpA was transformed to obtain Kp ΔtpiA-ΔmgsA-hdpA, Kp ΔtpiA-ΔdhaD-hdpA, Kp ΔtpiA-ΔmgsA-ΔgldA-hdpA, Kp ΔtpiA-ΔmgsA-ΔgldA-ΔdhaD-hdpA, and Kp ΔtpiA-ΔbudA-hdpA. These strains and the Kp ΔtpiA-hdpA were cultured in flasks and results are shown in Fig. 3.
Cell growth, glucose consumption, 2,3-butanediol and acetoin production of Kp ΔtpiA-ΔmgsA-hdpA were similar to those of Kp ΔtpiA-hdpA. 5.6 ± 0.1 g/L of DHA and 2.5 ± 0.1 g/L of glycerol were produced by this strain. While, 6.4 ± 0.5 g/L of DHA and 2.2 ± 0.2 g/L of glycerol were produced by Kp ΔtpiA-hdpA. It indicated that blocking the pathway of DHA to methylglyoxal had no positive effect on DHA production.
Glycerol production by Kp ΔtpiA-ΔdhaD-hdpA was reduced, with the titer of 1.8 ± 0.1 g/L. Whereas, no glycerol was produced by Kp ΔtpiA-ΔmgsA-ΔgldA-hdpA. Kp ΔtpiA-ΔmgsA-ΔgldA-ΔdhaD-hdpA also produced no glycerol. However, DHA production by these three strains were all reduced remarkably, rather than increased. Cell growths of Kp ΔtpiA-ΔmgsA-ΔgldA-hdpA and Kp ΔtpiA-ΔmgsA-ΔgldA-ΔdhaD-hdpA were enhanced, with final cell densities of 2.3 ± 0.2 and 2.1 ± 0.1 OD units, which were about four times that of Kp ΔtpiA-hdpA.
Neither 2,3-butanediol nor acetoin were synthesized by Kp ΔtpiA-ΔbudA-hdpA. Glucose consumption and cell growth of this strain became slow, and 6.5 g/L of glucose was still unused in the broth after 24 hours of cultivation. Glycerol produced by this strain was similar to that of Kp ΔtpiA-hdpA. However, DHA titer of this strain was lower than that of strain Kp ΔtpiA-hdpA. Thus, blocking the 2,3-butanediol synthesis pathway had no positive effect on DHA and glycerol production.
These efforts to reduce glycerol or other by-products production all failed to enhance the level of DHA.
Blocking the reaction of glyceraldehyde 3-phosphate to 1,3-bisphospho-glycerate for DHA production
The knock-out of tpiA prevented further catabolism of DHA phosphate in the glycolysis and resulted in its conversion to DHA. Glyceraldehyde-3-phosphate dehydrogenase catalyses the conversion of glyceraldehyde 3-phosphate to 1,3-bisphospho-glycerate. It was suspected, that blocking further catabolism of glyceraldehyde 3-phosphate might also result in DHA phosphate flow to DHA. Glyceraldehyde-3-phosphate dehydrogenase has three isoenzymes as noted in the genome of Klebsiella variicola 342, and they are encoded by gap, gapA and gapC, respectively . The genome of this strain was highly homologous to that of K.pneumoniae CGMCC 1.6366 used in this study. However, only gapA and gapC were found in the genome of K.pneumoniae CGMCC 1.6366. The two genes were knocked out individually and combined to get Kp ΔgapA, Kp ΔgapC and Kp ΔgapA-ΔgapC. pDK6-hdpA was transformed into these genes knock-out strains to get the corresponding strains. These strains were cultured in flasks and results were shown in Fig. 4.
Physiological characteristics of Kp ΔgapA were comparable to that of wild-type strain (data shown in Fig. 2). 23 g/L of glucose was exhausted by these strains after 12 hours of cultivation. 2,3-Butanediol and acetoin were the main products of this strain, with the titer of 3.3 ± 0.1 and 2.5 ± 0.3 g/L, respectively. Overexpression of hdpA in Kp ΔgapA failed for DHA or glycerol production. 3.3 ± 0.1 g/L of 2,3-butanediol and 1.5 ± 0.1 g/L of acetoin were produced by Kp ΔgapA-hdpA, which were similar to those of wild-type strain.
Cell growth of Kp ΔgapC was slightly slower compared to that of the wild-type strain, and glucose was exhausted after 15 hours of cultivation. 2,3-Butanediol and acetoin produced by this strain were similar to those of wild-type stain. Physiological characteristics of Kp ΔgapA-ΔgapC were nearly the same as those of Kp ΔgapC.
Over-expression of hdpA in Kp ΔgapC has a distinct effect on the host cell. Cell growth of Kp ΔgapC-hdpA was very slow. After 33 hours of cultivation, 3 g/L of glucose was still unused in the broth, and the cell density was lower than 1 OD unit during most periods of the process.
2,3-Butanediol and acetoin produced by Kp ΔgapC-hdpA were reduced to 1.4 ± 0.1 and 1.0 ± 0.1 g/L, respectively. These metabolites were mainly synthesized in the beginning 6 hours of cultivation. 2.3 ± 0.2 g/L of DHA was produced after 9 hours of cultivation, and after that its concentration had no distinct change. Glycerol was produced in the process, but its level was very low. 0.56 g/L of glycerol was detected after 24 hours of cultivation. Metabolites of Kp ΔgapA-ΔgapC-hdpA were very likely that of Kp ΔgapC-hdpA. But the growth of this strain was very weak, and the highest cell density was only 0.2 OD unit.
Blocking the conversion of glyceraldehyde 3-phosphate to 1,3-bisphospho-glycerate and over-expression of the hpdA lead to DHA and glycerol production. However, the levels of DHA and glycerol produced by Kp ΔgapC-hdpA or Kp ΔgapA-ΔgapC-hdpA were lower than those produced by Kp ΔtpiA-hdpA. This indicated that blocking further catabolism of DHA phosphate in the cell by knocking out of tpiA or gapC were both effective for DHA and glycerol production through the reverse glycerol catabolism pathway.
The effect of disruption of DHA kinases on DHA production
DHA kinases and DHA phosphate dephosphorylase catalyse the reaction of DHA phosphorylated and DHA phosphate dephosphorylated, respectively. If they both working at the same time, they form an futile cycle in the cell. To erase this reaction cycle, the subunits of the two kinases were disrupted individually and hdpA was expressed to obtain the following strains: Kp ΔtpiA-ΔdhaK-hdpA, Kp ΔtpiA-ΔdhaK1-hdpA, Kp ΔtpiA-ΔdhaK2-hdpA, Kp ΔtpiA-ΔdhaK3-hdpA, and Kp ΔtpiA-ΔDHAK-hdpA. Kp ΔtpiA-ΔDHAK-hdpA was a strain where dhaK, dhaK1, dhaK2 and dhaK3 were all knocked out. These strains were cultured in flasks and results are shown in Fig 5
Cell growth of Kp ΔptiA-ΔdhaK-hdpA was slow compared with that of Kp ΔptiA-hdpA. The final cell density of this strain was 0.3±0.1 OD units. Accordingly, glucose consumption and metabolites productivity were all in low rates. 0.5±0.1 g/L of 2,3-butanediol, 0.8±0.1 g/L of acetoin and 3.6±0.5 g/L of DHA were produced by this strain, while glycerol was not detected in the process.
Cell growth of Kp ΔptiA-ΔdhaK1-hdpA, Kp ΔptiA-ΔdhaK2-hdpA and Kp ΔptiA-ΔDHAK-hdpA were faster than that of Kp ΔptiA-hdpA. The highest cell densities of these strains were 0.8±0.1, 0.6±0.1 and 0.7±0.1 OD units, respectively. Glucose consumption rate of these strains were faster than that of Kp ΔptiA-hdpA. 2,3-Butanediol titers of these strains were 2.5±0.1, 2.4±0.2 and 2.3±0.1 g/L, respectively. Acetoin titer of these strains were similar to that of Kp ΔptiA-hdpA. DHA produced by these strains were 7.2±0.4, 7.0±0.2 and 7.8±0.3 g/L, respectively. Kp ΔptiA-ΔDHAK-hdpA had the highest DHA titer among these strains. Glycerol titers of these strains were similar to that of Kp ΔptiA-hdpA, but productivities were higher than that of Kp ΔptiA-hdpA. Cell growth and glucose consumption of Kp ΔptiA-ΔdhaK3-hdpA were similar to that of Kp ΔptiA-hdpA. Glycerol produced by this strain was higher than Kp ΔptiA-hdpA.
The DHA and glycerol production by these strains are summarized in Table 2.
Except Kp ΔtpiA-Δdhak-hdpA, the conversion ratio of glucose to DHA and glycerol in these strains were all improved compared with that of Kp ΔtpiA-hdpA. The total conversion ratio of glucose to DHA and glycerol was 0.95 in Kp ΔtpiA-ΔDHAK-hdpA, which was nearly the maximum theoretical conversion ratio. This strain also had the highest DHA and glycerol productivity and was selected for further investigation.
DHA production by Kp ΔtpiA-ΔDHAK-hdpA using different carbon sources
Kp ΔtpiA-ΔDHAK-hdpA was cultured in flasks with glucose, xylose, sucrose and fucose as the main carbon source, and results are shown in Fig. S1. All these carbon sources can be used by the cell for DHA and glycerol production. The conversion ratios of glucose, sucrose and fucose to DHA and glycerol were similar. The conversion ratio of xylose to DHA and glycerol was lower than others, with the value of 0.38 mol/mol. These results were reasonable. Sucrose was hydrolysed to form glucose and fructose, and the two monosaccharoses were all catabolised through the glycolysis pathway. Fucose is a monosaccharose that is rich in marine algae. In K. pneumoniae, fucose was converted to fuculose and further to fuculose-phosphate. DHA-phosphate and lactaldehyde was formed from fuculose-phosphate with the catalysis of an aldolase. Xylose was catabolised through the Pentose phosphate pathway, only part of carbon was converted to DHA-phosphate and resulted a low conversion ratio to DHA and glycerol. While, more carbon was used for cell growth, and a high cell density was obtained with xylose as the carbon source. It can be concluded that any carbon source that can be catabolised to form DHA-phosphate is suitable for DHA and glycerol production by Kp ΔtpiA-ΔDHAK-hdpA.
Fermentation parameters optimization
Culture pH optimization.
Kp ΔtpiA-ΔDHAK-hdpA was cultured in 5 L bioreactors with fermentation medium, where the culture pH was stabilized at 5.5, 6.0, 6.5, and 7.0. The air flow rate was set at 2 L/min and the stirring rate of the bioreactor was set at 250 rpm. Fermentation results are presented in Fig. 6.
Cell growth was positively related to the culture pH. The lowest and the highest cell densities were obtained at the culture pH 5.5 and 7.0, respectively. All cell densities obtained were higher than that in flask culture (shown in Fig. 5). Glucose consumption rates were similar in the range of pH 5.5-6.5. Glucose consumption rate in culture pH 7.0 was low compared with other conditions, and it took 24 hours to use all the glucose supplied, which was slower than the consumption in flask culture.
2,3-Butanediol and acetoin production was inversely related to the culture pH. 2.7 g/L of 2,3-butanediol and 1.7 g/L of acetoin were produced in culture pH 5.5, while only 0.9 g/L of 2,3-butanediol was produced in culture pH 7.0, and no acetoin was detected at this condition.
DHA was produced from the beginning of cultivation, and quickly reached its high levels at 9 hours of cultivation. Glycerol production started at 9 hours, and its levels increased towards the end of the process. 6.1 g/L of DHA was produced at culture pH 6.0. Lower or higher culture pH all resulted in lower DHA titers. The lowest DHA titer was obtained at culture pH 7.0, reaching 4.1 g/L. Glycerol titers were in agreement with DHA titers. The highest glycerol titer was 3.1 g/L, which was obtained at culture pH 6.0.
DHA and glycerol titers and the substrate conversion ratios of these experiments are summarized in Table 3. The conversion ratio of glucose to DHA and glycerol both had the highest values at culture pH 6.0, and the total value was 0.90. Productivities of DHA and glycerol in culture pH 6.0 were the fastest among all experimental conditions. Thus pH 6.0 was selected as the optimized culture pH.
Kp ΔtpiA-ΔDHAK-hdpA was cultured in 5L bioreactors. The air flow rate was set at 2L/min and the stirring rate of the bioreactor was set at 50, 150, 250, 250 and 450 rpm to obtain different aerobic conditions. Culture pH were set at 6.0. Fermentation results are presented in Fig. 7.
Cells growth and glucose consumption had a positive relationship with stirring rates. The higher the stirring rate the higher the cell density and the glucose consumption rate. 3.6 OD unit was obtained at a stirring rate of 450 rpm. The lowest cell density was 2.3 OD unit at stirring rate of 50 rpm. Glucose was exhausted after 13 hours of cultivation at a stirring rate of 450 rpm, followed by stirring rate of 350 and 250 rpm. Whilst 3.5 and 3.6 g/L of glucose were still unused at stirring rate of 50 and 150 rpm after 24 hours of cultivation. 2,3-Butanediol and acetoin productivities were improved with the increase of the stirring rate. Stirring rate of 50 and 150 rpm had the lowest titers of 2,3-butanediol and acetoin.
The productivity of DHA at stirring rate of 50 and 150 rpm conditions was similar, and these were the lowest value among all experimental conditions. However, the final titers of DHA were the highest, with the value of 7.1 g/L and 7.4 g/L, respectively. At the same time, 3.3 g/L and 3.1 g/L of glycerol were produced, which were also the highest value among all agitation conditions.
The final titers of DHA and glycerol and substrate conversion ratios at different oxygen supplementation are summarized in Table 4. The total conversion ratio of glucose to DHA and glycerol was 0.97 and 0.98 at stirring rate of 50 and 150 rpm conditions. It means nearly all DHA phosphate were transferred to DHA and glycerol in these conditions. In contrast, the total conversion ratio of glucose to DHA and glycerol was 0.58 at stirring rate of 550 rpm. Thus, lower oxygen supplementation favored DHA and glycerol production.
Fed batch fermentation
High culture pH and high oxygen supplementation favor for cell growth, but the conversion ratio of glucose to DHA and glycerol were low. Whereas low culture pH and low oxygen supplementation favor a high substrate conversion ratio, with low cell growth rate. To solve this inverse relationship, a two- phase culture strategy was used. In the beginning 6 hours of culture, the culture pH and stirring rate were set at 7.0 and 450 rpm, which were favorable for cell growth. Then, the culture pH and stirring rate were switched to 6.0 and 150 rpm, since this condition favored DHA and glycerol production. Fed batch fermentations were conducted to obtain a superior final product level. Fermentation results are presented in Fig. 8.
In fed batch fermentation, cells were rapidly growing during the first 16 hours of cultivation and the cell density reached to 2.1 OD units. After about 30 hours cells entered a stationary phase and the highest cell density was reached at 39 hours of culture. The glucose concentration in the broth dropped to 2.4 g/L after 26 hours. After addition of highly concentrated glucose solution due to this initial drop, the glucose concentration increased to 27.2 g/L. At 49 hours, another glucose feed had to be performed. DHA was produced in the broth and its concentration continued to increase. The highest DHA titer of 23.9 g/L was achieved after 91 hours of cultivation. Glycerol production followed production of DHA and its titer was 10.8 g/L after 91 hours of cultivation. The total conversion ratio of DHA and glycerol from glucose was 0.97 mol/mol. The main by-products of this process were 2,3-butanediol and acetoin, reaching a titer of 9.2 and 3.5 g/L, respectively.
DHA is a toxic chemical to the cell and high levels of DHA are known to inhibit the activity of cells . The inhibitory effect of DHA on the wild-type K. pneumoniae was determined and results are shown in Fig. S2. As expected cell growth was inhibited by DHA. With the concentration of DHA in the broth increasing, the inhibitory effect became more obvious. Cell growth was totally stopped at 22 /L of DHA. The 23.9 g/L of DHA obtained in fed batch fermentation exceeded the highest level of DHA that cell can tolerate. Thus, cells might have lost metabolic activities in this condition; preventing the DHA titer to increase further.