Physiology of M. thermophila WT and JG207
This experiment was performed in a 300 mL bioreactor system, in which sampling and evaporation will affect the concentration of biomass and product. So we focused on the total amount in the reactor rather than concentration to calculate kinetic parameters. The two strains exhibited different growth characteristics in exponential phase (Fig. 1, Table 1). Compared with WT strain, the glucose uptake rate of JG207 increased ~ 36%, but similar specific growth rate of the two strains were observed. The additional uptake glucose for JG207 was mainly used to produce malic acid and by-product succinic acid, the yields of which were 18.6% and 5.2% (Cmol/Cmol), respectively. Accordingly the biomass yield was decreased by around 30% for JG207. In addition, the oxygen uptake rate (OUR) of JG207 was lower than WT but carbon dioxide evolution rate (CER) was higher.
Table 1
Kinetic parameters and carbon recovery in exponential phase of M. thermophila WT and JG207.
| WT | JG207 |
Specific rate | | |
µ(h-1) | 0.26 ± 0.02 | 0.26 ± 0.03 |
qs(mmol/gDCW·h-1) | 3.03 ± 0.26 | 4.13 ± 0.41 |
qmal(mmol/gDCW·h-1) | | 1.15 ± 0.31 |
qsuc(mmol/gDCW·h-1) | | 0.35 ± 0.13 |
qCO2(mmol/gDCW·h-1) | 5.21 ± 0.17 | 6.16 ± 0.03 |
Yield(Cmol/Cmol) | | |
YX/S | 0.627 | 0.425 |
Ymal/S | | 0.186 |
Ysuc/S | | 0.052 |
YCO2/S | 0.287 | 0.248 |
Reconstruction Of Biomass Synthesis Reaction
Table 2
Proteinogenic amino acid content in dry biomass.
Name | Value(mmol/gDCW) | SD |
Ala | 0.347 | 0.005 |
Ser | 0.226 | 0.020 |
Asp | 0.179 | 0.017 |
Asn | 0.179 | 0.017 |
Thr | 0.219 | 0.018 |
Gly | 0.266 | 0.021 |
Glu | 0.369 | 0.013 |
Gln | 0.369 | 0.013 |
Val | 0.179 | 0.014 |
Leu | 0.332 | 0.033 |
Ile | 0.168 | 0.012 |
Lys | 0.270 | 0.026 |
His | 0.061 | 0.006 |
Arg | 0.226 | 0.015 |
Tyr | 0.020 | 0.003 |
Phe | 0.144 | 0.014 |
Cys | 0.004 | 0.001 |
Met | 0.008 | 0.000 |
Pro | 0.194 | 0.018 |
Biomass synthesis reaction determines how much of each precursor is used for growth, so it has great influence on biomass yield to substrate thus impacts the accuracy of the calculated flux distribution. Accurate precursor composition of biomass is important to model validation and flux calculation(18, 19). In order to make the results of metabolic flux analysis more reliable, we reconstructed the biomass synthesis reaction formula (Additional file 5). We measured the content of each amino acids in dry biomass (Table 2) to determine the contribution of each amino acid to biomass synthesis. 17 amino acids were detected using the amino acid analyzer. After hydrolysis with 6M HCl, tryptophan was completely destroyed, asparagine was hydrolyzed to aspartic acid, and glutamine was hydrolyzed to glutamic acid. Here we referred to the treatment of Ye et al. (19) and assumed that these two groups of amino acids were evenly distributed.
Metabolic Flux Distribution Of The Central Carbon Metabolism Pathways
In order to better understand the metabolic differences of the two strains, we used 13C-MFA to estimate the intracellular flux distribution. Samples were taken at the mid-exponential phase during the batch cultivation, when the specific growth rate reaches its maximum and keeps constant, i.e. the metabolism is under steady state (20). The isotopmers information of amino acids (Additional file 1) indicated that samples of both strains were at isotopic steady state, which satisfied the assumption for 13C-MFA (20). It is reported that proper combinations of different 13C labeled tracers have big impact on the accuracy of estimated fluxes for different pathways(21, 22). Therefore, the tracer used in this experiment was the combination of 50% 12C-Glucose, 30% [U-13C]-Glucose and 20% [1-13C]-Glucose (isotopic enrichment 98–99%, Cambridge Isotope Laboratories, Inc.) in order to get the best estimation of metabolic fluxes. The estimated flux distributions of the two strains are mapped to the central carbon metabolism pathway as shown in Fig. 2. The measured and EMU model simulated MDVs of amino acids are listed (Additional file 2) with flux confidence intervals (Additional file 3), and χ2 tests showed that the flux results are statistically acceptable.
The results of 13C-MFA (Fig. 2) showed that the flux distribution of the two strains were distinct at several key nodes, such as at branch metabolites like G6P, PYR. Compared with WT, the EMP flux of JG207 increased significantly, while the PPP flux decreased. This is different from the metabolic flux results of Knuf et al. (23) for malate high-production A. oryzae strain. Their research showed that the PPP flux of high malate-producing A. oryzae strain was higher than wild type. The reason is that the biomass yield of high malate-producing A. oryzae strain also increased, which needs more NADPH and nuclear acid precursors supply. Then our results show that E4P and X5P produced by PPP will regenerate F6P and GAP, which then consumed in EMP. The difference of EMP between the two strains is mainly reflected in the activity of glucose-6-phosphate isomerase, which catalyzes the reaction from G6P to F6P. In addition, the pyruvate carboxylation flux of JG207 was significantly improved, while the flux to TCA was not significantly improved. More accumulated oxaloacetate was instead converted to malate via rTCA in the cytoplasm rather than transport into the mitochondria. Therefore, the additional EMP flux of JG207 are all directed to pyruvate carboxylation for malate synthesis. As the main source of intracellular NADH, the flux of TCA start step in JG207 is not much different from WT, but flux in the downstream TCA steps are larger than WT to provide enough NADH for maintaining high malate synthesis and enough ATP supply for cell growth. Increased downstream TCA flux also lead to increased CO2 production rate in JG207, indicated by higher CER. According to the 13C-MFA results, the yields of malic acid and succinic acid over glucose were calculated to be 18.2% and 5.6%, respectively, which were basically consistent with the results of direct experimental measurements.
Cofactors Metabolism Of The Two Strains
To analyze the effect of cofactors on malate synthesis, we calculated the intracellular NADH flux, including both synthesis and consumption fluxes. Corresponding to the higher substrate uptake rate, absolute net flux of intracellular NADH in JG207 was much higher than WT (Fig. 3a). For both strains, EMP and TCA were the main sources for NADH production, and the NADH produced by TCA accounted for about 60% of the total production of NADH under the condition test. The major consumption of intracellular NADH is the electron transport chain, where oxidative phosphorylation is performed for synthesizing ATP to provide energy for cell growth. NADH is also consumed in cytoplasm at rTCA, so the high production of malate by JG207 requires enough supply of NADH in the cytoplasm. This study (Fig. 3) shows that the additional NADH supplied for JG207 rTCA mainly rewired from the NADH those was consumed by oxidative phosphorylation before in the WT strain. NADH for oxidative phosphorylation consumption comes from TCA and EMP. It is not well documented that NADH can be transported out of mitochondria. So it is possible that reduced NADH flux which transfer into mitochondria compensates for the consumption of rTCA in JG207, and this part of NADH provided by EMP.
Intracellular Metabolite Pool Analysis Between The Two Strains
We measured the intracellular metabolite pool of the two studied strains by LC-MS and Coenzyme Assay Kit (Solarbio). The isotope dilution mass spectrometry (24) is mainly used to eliminate the ion suppression effect caused by the electrospray ionization in mass spectrometry measurement. The 13C isotope internal standard required for the experiment was obtained by culturing Pichia Pastoris with U-13C Glucose. Combined with the results of 13C-MFA and metabolomics, the mechanism behind the high malate production in M. thermophila was further elucidated.
The heatmap of intracellular metabolites (Fig. 4) showed that the up-regulated metabolites of JG207, including mainly MAL, FUM, SUC, cofactors and so on. The reason for this phenomenon was mainly related to the increase of pyruvate carboxylation flux, and more carbon flux directed to rTCA, which caused the accumulation of corresponding metabolites in the pathway. JG207 has a higher reduced state (Fig. 5: the ratio of NADH/NAD + has significant difference) and a lower energy state, due to decreased oxidative phosphorylation flux, which made NADH consumption and ATP synthesis decreased. Higher transhydrogenation flux (Fig. 3; from NADH to NADPH) may be the reason for higher NADPH pool in JG207. Furthermore, the amino acid pool size is related to the precursors for biomass synthesis. For example, the pool of pyruvate-derived amino acids in JG207 is lower than WT, which may be caused by increased pyruvate carboxylation flux of JG207, resulting in decreased corresponding amino acids synthesis flux. The reason for the decreased pool of oxaloacetate derived amino acids is that more oxaloacetate was directed to the production of malic acid and succinic acid. The reason for the smaller pool of α-ketoglutarate derived amino acids is that α-ketoglutarate has higher decarboxylation flux in JG207.
We then performed principal component analysis (PCA) on the 2 groups of metabolomic data, trying to identify key nodes that contribute more for malic acid production. The PCA score plot (Fig. 6a) shows that the two strains were clearly separated, indicating that JG207 had obvious metabolic changes from WT. Then, metabolites that contribute to strain separation are displayed in the corresponding loading plots (Fig. 6b). The variance was larger along the first primary component PC1(71.49%), and it is obvious that load for NADH and NAD + had largest absolute value, which indicating their high influence for the higher malate production. There were other metabolites which had larger absolute load value, such as metabolites in the rTCA and so on.