Results and Discussion
Combined over-expression of ZWF1 and SOL3 improves the NADPH supply and thus increasing α-farnesene production inP. pastorisX33-30*
The oxiPPP is the main inherent route for NADPH generation in P. pastoris , which catalyzed by glucose-6-phosphate dehydrogenase (ZWF1), 6-gluconolactonase (SOL3), 6-phosphogluconate dehydrogenase (GND2) and D-ribulose-5-phosphate 3-epimerase (RPE1).11, 18 In order to increase the NADPH availability for α-farnesene biosynthesis, the key enzymes in oxiPPP were optimized to overexpress in a α-farnesene high-producing strain P. pastoris X33-30*. Firstly, we respectively overexpressed the single gene zwf1 (encoding ZWF1),sol3 (encoding SOL3), gnd2 (encoding GND2) and rpe1(encoding RPE1) in strain X33-30*, resulted strains X33-30*Z, X33-30*S, X33-30*G and X33-30*R. Compared with the strain X33-30*, strains X33-30*Z and X33-30*S showed the increased NADPH concentration whereas the strains X33-30*G and X33-30*R showed no visible difference in NADPH concentration both at 24 h and at 72 h (Fig. 2a). Correspondingly, strains X33-30*Z and X33-30*S also showed the increased α-farnesene production, increased by about 6.5% and 12.0% as compared with strain X33-30* at 72 h, respectively (Fig. 2b). The similar results were also found in previous researches, in which overexpression of ZWF1 or SOL3 increased the foreign protein production in P. pastoris25-27 and the terpenoid production in S. cerevisiae 28-29 because of the increased intracellular NADPH level. ZWF1 and SOL3 catalyzed the first and the second steps of the oxiPPP and were inhibited by NADPH and ATP,11, 25 which catalyzed the rate-limiting steps in oxiPPP.26 In addition, the native expression level ofsol3 in P. pastoris was low.30Therefore, these may be why strain X33-30*Z with overexpression ZWF1 and strain X33-30*S with overexpression of SOL3 increased the NADPH availability and thus increasing α-farnesene production. It should be noted that overexpression of GND2 had no positive effects on increasing the NADPH availability and α-farnesene production (Fig. 2), which was similar with the results reported by Kim et al. 31 and Nocon et al. 26 but different to the results reported by Prabhu and Veeranki 25. Rebnegger et al.30 pointed out that gnd2 shows the high expression level while sol3 shows the low expression level inP. pastoris . Based on these, we speculated that GND2 was not a rate-limiting enzyme and only overexpression of GND2 did not enhance the carbon flux in oxiPPP.
To further analyze the synergetic effects of these key genes in oxiPPP on NADPH and α-farnesene production, we tried out different expression combination ways of these genes looking for the highest performing combos. As can be seen from Fig. 3a, the resulting strain X33-30*ZSGR (i.e., combined overexpression of ZWF1, SOL3, GND2 and RPE1) showed the highest intracellular NADPH level, followed by strain X33-30*ZSG (i.e., combined overexpression of ZWF1, SOL3 and GND2). Interestingly, combined overexpression of GND2 and RPE1 (i.e., strain X33-30*GR) had no obvious effect on increasing NADPH level (Fig. 3a), which was similar to single overexpression of GND2 or RPE1 (Fig. 2a). This may be down to the bottlenecks of the rate-limiting step, which catalyzed by ZWF1 and SOL3.26 However, it should be noted that the α-farnesene production was not increased with the increase of the intracellular NADPH level in the corresponding strain (Fig. 3b). Among the test strains, strain X33-30*ZS (i.e., strain X33-31) produced the highest α-farnesene (i.e., 2.54±0.21 g/L) in spite of the third highest NADPH level (Fig. 3). In contrast, although the strains X33-30*ZSGR and X33-30*ZSG showed the top two NADPH level (Fig. 3a), they showed the worst α-farnesene production, even lower than the original strain X33-30* (Fig 2b and Fig. 3b). Nocon et al. 26 and Prabhu and Veeranki 25 also found the similar results, in which combined overexpression of ZWF1, SOL3, RPE1 or/and GND2 would be detrimental to foreign protein production. It may be that excess overexpression of the key enzymes in oxiPPP may imbalance the PPP flux26 or disturb acetyl-CoA biosynthetic pathway31 and thus negatively impacting on product formation. Thus, the strain X33-31 was selected to further modify to increase α-farnesene production.
Inactivation of glucose-6-phosphate isomerase disturbs the cell growth and thus negatively affecting α-farnesene biosynthesis in P. pastorisX33-31
Previous research indicated that overexpression of the transcription factor STB5 increased cytosolic NADPH concentrations because STB5 upregulated the expression of most genes in the PPP and repressed the expression of glucose-6-phosphate isomerase-coding gene in glycolysis.31-32 Maybe since STB5 is a basal regulator of the PPP,33 overexpression of STB5 did not increase the protopanaxadiol production.31 PGI (encoded by pgi ) competes with ZWF1 for glucose-6-phosphate, which catalyzes glucose-6-phosphate to form frucose-6-phosphate (Fig. 1). In order to investigate the effect of PGI on α-farnesene production, the PGI was inactivated in strain X33-31, resulting in strain X33-31ΔP. As a control, strain X33-30*ΔP (i.e., deletion of pgi in strain X33-30*) was also done. Unfortunately, inactivation of PGI in strain X33-31 had negative effect on α-farnesene production, in which the resultant strain X33-31ΔP only produced about 5% of α-farnesene compared with the strain X33-31 (0.13±0.06 g/L vs. 2.54±0.21 g/L) (Fig. 4a). In addition, inactivation of PGI increased the intracellular NADPH level but repressed the cell growth and (Fig. 4b, c). In the past, Qin et al. 34also found that expression of pgi controlled by the ultra-low intensity promoter NAT2p in Saccharomyces cerevisiae decreased the cell growth and 3-hydroxypropionic acid production. The possible reasons could be that PGI plays an important role in the central carbon metabolism in yeast.35 In addition, Aguilera & Zimmermann 36 pointed out that inactivation of PGI inS. cerevisiae prevents growth on glucose. However, it should be noted that inactivation of PGI in strain X33-30* had little effect on increasing α-farnesene production in spite of the decrease of the cell growth (Fig. 4). There results indicated that inactivation of PGI enforced the carbon flux into PPP and thus increasing the NADPH availability for α-farnesene biosynthesis. Since the surplus NADPH cannot be re-oxidized, the PGI-deficient strain did not grow on glucose.37 Thus, we speculated the reason the strain X33-30*ΔP did not obviously decrease the cell growth is that more NADPH was used to biosynthesize α-farnesene. The previous results have reinforced this speculation. For example, Fiaux et al.38 restored the cell growth on glucose of PGI-deficient S. cerevisiae mutant by heterologous expression of transhydrogenase UdhA from E. coli . However, although the strain X33-30*ΔP showed the increased α-farnesene production as compared with the strain X33-30*, its final titer of α-farnesene was still lower than that of strain X33-31 (2.23±0.18 g/L vs. 2.54±0.21 g/L) (Fig. 4a), indicating that inactivation of glucose-6-phosphate isomerase is not the best strategy to increase α-farnesene production.
Low intensity expression of POS5 from S. cerevisiaebalances the NADPH/NADH ratio and thus promoting α-farnesene biosynthesis in P. pastoris
It is well known that the intracellular NADH level is higher than the intracellular NADPH level.39-41 Previous research indicated that heterologous expression of NADH kinase POS5 provides another source of NADPH except for the oxiPPP in yeast.11, 28 To further promote the α-farnesene production by optimizing the NADPH supply, we introduced the cPOS5 targeting in the cytosol from S. cerevisiae in strain X33-31. Interestingly, the resultant strain X33-32 with gene cPOS5controlled by promoter PGAP showed the bad cell growth and α-farnesene production, but it showed the increased productivity of NADPH and α-farnesene (Fig. 5). POS5 catalyzed the NADH to form NADPH, thereby reducing cell energy resources.11 In addition, excess NADPH in cell would repress the cell growth, glucose consumption and products production.40, 42 These comments may be an underlying cause of the decreased cell growth and α-farnesene production.
To try to dissolve this problem, we then decreased the expression level of POS5 by replacing PGAP with a series of weak promoters. Based on the previous reports, the relative intensity of the promoters PPISI, PGPM1, PMET3, and PPGK1 were 40%, 15~40%, 13%, and 0~10% as compared with that of the PGAP, respectively43-45. An increased α-farnesene production (i.e., 2.77±0.18 g/L) was obtained in strain X33-35 with gene cPOS5 controlled by PMET3, which increased by about 9.1% as compared with strain X33-31 (i.e., 2.54±0.21 g/L)(Fig. 5c). Correspondingly, strain X33-35 also exhibited the high intracellular NADPH level (Fig. 5b). In addition, decreasing the expression level of POS5 restored the cell growth as compared with the strain X33-32 (Fig. 5a), indicating that excess NADPH in cells would be detrimental to cell growth. These results indicated that low intensity expression of cPOS5 in strain X33-31 benefits to maintain the right amount of NADPH for cell growth and α-farnesene production. Although the strain X33-34 exhibited the high NADPH concentration and cell growth (Fig. 5a, b), it should be noted that strain X33-34 did not produce more α-farnesene than that of strain X33-31 (i.e., 2.56±0.26 g/L vs.2.54±0.21 g/L)(Fig. 5c), indicating that another limiting factor hampered the α-farnesene biosynthesis in strain X33-34. As can be seen from Fig. 1, 9 molecules of ATP are need to produce 1 molecule of α-farnesene. ATP is mainly produced by NADH oxidation via ETP under aerobic conditions.22 Therefore, we speculated that ATP availability is another limiting factor for further increasing α-farnesene production.
Overexpression of adenine phosphoribosyltransferase enhances the precursor AMP supply and thus increasing the ATP availability and α-farnesene production in P. pastoris
ATP can be synthesized either by substrate level phosphorylation (SLP) or by ETP in aerobic respiring bacteria, and the ETP is the main route for ATP generation using NADH as electron donor.22, 46In the process of ETP, AMP or/and ADP was used as the substrate for ATP biosynthesis.22 Therefore, increasing the AMP or ADP supply could increase the ATP availability in theory. To test this theory, the endogenous adenine phosphoribosyltransferase (APRT) was overexpressed in strain X33-35, resulting in strain X33-37. APRT catalyzes the formation of AMP from adenine and 5-phospho-α-Dribose-1-diphosphate (PRPP),47 and we found that the intracellular ATP level of strain X33-37 was 9.4% higher than that of strain X33-35 while the intracellular NADH level was slightly lower than that of strain X33-35 (Table 3). Similar results were also found in previous reports, in which the mutatedCorynebacterium glutamicum with inactivation of AMP nucleosidase showed the increased intracellular ATP level and the decreased intracellular NADH level.20 The most likely is that more NADH was used for ATP biosynthesis through ETC because of the abundant supply of AMP. Equally unsurprisingly, overexpression ofaprt gene promoted cell growth and α-farnesene production, the DCW and α-farnesene production of strain X33-37 reached 2.35±0.11 g/L and 2.94±0.25 g/L (Fig. 6), which were 10.3% and 6.1% higher than those of strain X33-35 (i.e., 2.13±0.16 g/L and 2.77±0.18 g/L, respectively), respectively. ATP is a key factor for cell growth and maintenance and controlling intracellular environment,48 and thus adequate ATP supply could increase biomass. In addition, the biosynthesis of 1 molecule of α-farnesene requires at least 9 molecules of ATP (Fig. 1), thus the increased intracellular ATP level effectively pulled more carbon flux into the α-farnesene biosynthetic pathway, resulting in higher α-farnesene production. These results indicated that overexpression of endogenous APRT is conducive to increase α-farnesene production because it facilitates the AMP biosynthesis and thus increasing the ATP supply.
Deletion of NADH-dependent dihydroxyacetone phosphate reductase elevates the intracellular NADH level and thus elevating the intracellular ATP level and α-farnesene production in P. pastoris
NADH plays an important role in maintaining the redox balance and energy generation NADH,18 it can used as precursor for the regeneration of NADPH and ATP. In order to maintain the abundant supply of NADH, we tried to decrease the NADH consumption in shunt pathway. To do this, we knocked out the NADH-dependent dihydroxyacetone phosphate reductase. (GPD1)-coding gene gpd1 in strain X33-37, resulting in strain X33-38. GPD1 catalyzes the biosynthesis of glycerol from dihydroxyacetone phosphate and used the NADH as reducing cofactor (Fig. 1).Previous research indicated that the glycerol biosynthetic pathway plays an important role in maintaining the intracellular NADH and NAD+ level.49 So obviously, the intracellular NADH level and NADH/NAD+ ratio in strain X33-38 increased by 11.6% and 28.6% as compared with strain X33-37, respectively (Table 3). Meanwhile, strain X33-38 had certain improvement in both the intracellular NADPH level and the ATP level (Table 3). The α-farnesene production of strain X33-38 reached 3.09±0.37 g/L after 72 h in shake-flask fermentation, which was 5.1% higher than that of strain X33-37 (i.e., 2.94±0.25 g/L) (Fig. 7a). The DCW of strain X33-38 was also slightly increased as compared with strain X33-37 (i.e., 2.41±0.23 g/L vs. 2.35±0.11 g/L)(Fig. 7b). It is worth noting that strain X33-38 did not accumulate glycerol throughout the fermentation process, which was different from the strain X33-37 (Fig. 7c). GPD1 is a key enzyme in glycerol biosynthesis 50 and He et al.49 also found the similar results, in which the strainS. cerevisiae DRY01 with silencing GPD1 dramatically decreased the glycerol accumulation. These results indicated that deletion of GPD1 not only improves NADH supply but also decreases the carbon flux in shunt pathway, and thus increasing the α-farnesene production.