Expression of a bacterial PDH complex to increase the acetyl-CoA supply
A FFA producing strain YJZ08 with four gene deletions (faa1Δ faa4Δ pox1Δ hfd1Δ) was used as the background strain for cytosolic PDH evaluation [6]. The three subunits of PDH complex, encoded by pdhA, pdhB, aceF and lpd, and lipoate ligases from E. faecalis, encoded by lplA and lplA2 [16], were integrated into the two genome sites Chr XII-5 and Chr XI-3 [28] of, yielding strain PDH1. As shown in Figure 2, YJZ08 with deregulated FA synthesis produced FFA at 458.9 mg/L in 72 h in minimal medium with 20 g/L glucose, which was consistent with previous studies [6, 9]. Meanwhile, the constructed strain PDH1 produced 512.7 mg/L FFA at 72 h (Figure 2A). From the growth data, we found that E. faecalis PDH reduced the maximum specific growth rate from 0.33 h-1 to 0.28 h-1 but didn’t result in much difference in the final biomass. PDH1 accumulated 1.90 g/L glycerol, compared with 1.90 g/L glycerol with YJZ08 (Figure 2A), suggesting that cytosolic PDH did not cause an NADH burden in YJZ08.
Theoretically, a combination of the PDH route with the phosphoketolase (PK) pathway can result in a higher maximum FFA yield [15], since the PDH route could produce acetyl-CoA with low energy cost and the PK pathway could produce acetyl-CoA with low carbon loss. Moreover, the PK pathway utilizes xylulose-5-phosphate from the pentose phosphate (PP) pathway as the substrate, and it may thermodynamically help to pull the carbon flux towards the PP pathway and increase NADPH generation. However, we tested the integration of xylulose-5-phosphate phosphoketolase from Leuconostoc mesenteroides [29] and phosphotransacetylase from Clostridium kluyveri [30] into both YJZ08 and PDH1, and did not find significant improvement in cell growth and FFA synthesis (Supplementary Figure 1), probably because of the strict control of carbon flux through the PP pathway [31] or inefficient phosphoketolase enzymes [29].
Moreover, the yeast ZS01 strain with simplified lipid metabolic networks and redirected flux towards FFA synthesis was constructed by deletion of PAH1, DPP1, LPP1 and ARE1 in YJZ08. Then, the genes encoding the PDH complex, composing of pdhA, pdhB, aceF, lpd, lplA and lplA2, were integrated into the chromosome of ZS01, yielding strain PDH2. As shown in Figure 2, when cultivated in minimal medium with 20 g/L glucose, ZS01 accumulated 497.9 mg/L FFAs at 72 h with a high productivity of 133.7 mg/L/OD, which was consistent with previous studies [8, 9]. After 72 h, ZS01 accumulated more FFAs since there was ethanol remaining and the FFA titer reached 734.2 mg/L at 144 h when ethanol was exhausted (Figure 2A). Interestingly, PDH2 produced 660.2 mg/L FFAs at 72 h and 756.3 mg/L FFAs at 96 h when ethanol was completely exhausted. In PDH2, the faster FFA production rate accompanied with a faster ethanol consumption rate and lower glycerol level, indicating that cytosolic PDH may alleviate NADH burden and speed up cell growth and FFA production (Figure 2B), probably due to the lower energy cost.
To address the possible mismatch between acetyl-CoA supply, NADH production and NADPH consumption for FFA synthesis, we evaluated whether PDH from E. faecalis might utilize NADP+ as well as NAD+ as redox cofactors in a zwf1Δ ald6Δ mutant, which hardly grew on glucose medium due to lack of ability to produce cytosolic NADPH [32]. The results showed that cytosolic PDH expression could not rescue cell growth of the zwf1Δ ald6Δ mutant on glucose (Figure 2C). A previous study showed that cofactor specificity of E. coli PDH could be converted from NAD+ to NADP+ via seven amino acid mutations in E3 [33, 34]. The alignment result of E3 proteins from E. faecalis and E. coli suggested that the mutated amino acids are highly conserved, as shown in Supplementary Figure S2A. Expression of the mutated PDH recued the growth on glucose and disruption of E3 encoding gene lpd suspended the growth recovery, suggesting the mutated PDH might utilize NADP+ as redox cofactor. Growth assay on minimal glucose medium didn’t show similar results, which might be caused by its poor activity (Figure 2C). However, these E3 mutations didn’t increase FFA titer, but reduced the titer to 439.5 mg/L, comparable with YJZ08 (Supplementary Figure S2B). The FFA results indicated that the mutations resulted in enzymatic activity lost, probably due to unsuccessful protein assembly. To improve its enzymatic activity towards NADP+, rational design based on PDH structures or directed evolution will be required for further investigation. As E. faecalis PDH has been demonstrated to be a functional enzyme in E. coli and yeast at both aerobic and anaerobic conditions [16, 35], a novel PDH complex with NADP+ preference will clearly be valuable for bioproduction of acetyl-CoA derived chemicals.
It was also observed that, PDH expression altered the distribution of saturated and unsaturated FFAs (C16 and C18) in PDH2, but not in PDH1 (Figure 2D). The double bonds in unsaturated FFAs were formed by Delta-9 fatty acid desaturase, encoded by OLE1, through an oxygen-dependent mechanism that requires reducing equivalents from NADH. Thus, the increase in unsaturated FFA ratio might also indicate that the functional PDH resulted in cytosolic redox changes without imposing an increased NADH burden, due to its insensitive feature to high NADH/NAD+ ratios [35-37].
Balance of cytosolic redox factors for FFA production
NAD and NADP redox couples are crucial to maintain cellular redox hemostasis, with lower cytosolic NADH/NAD+ ratios (range from 0.001 to 0.01) and higher NADPH/NADP+ ratios (range from 15 to 60) due to their different functions [38]. In FFA production strains, over-production of NADH resulted in glycerol accumulation, while NADPH as the cofactor required for fatty acid elongation was limited. Thus, the balance between NAD and NADP redox couples would be important for FFA production.
To balance the redox state of the cell and further improve FFA synthesis, NADP+-dependent glyceraldehyde-3-phosphate dehydrogenase (GapN) from Streptococcus mutans [39] was expressed on a multi-copy plasmid pGapN in the FFA production strain YJZ08, yielding strain ZS02. As shown in Figure 3, GapN expression resulted in reduced glycerol accumulation and increased FFA production.
We further balanced cytosolic redox conditions through interruption of glycerol synthesis via glycerol-3-phosphate dehydrogenase (GPD) deletions. Since GPD1 was regulated by osmotic stress [23], GPD2 was first deleted in YJZ08, and the resulting strain ZS03 showed decreased glycerol accumulation, and increased FFA titer and productivity (Figure 3A). When both GPD1 and GPD2 were deleted, the specific growth rate of the resulted strain ZS04 decreased dramatically, which was consistent with previous studies [23]. The final FFA titer of ZS03 was 556.0 mg/L, 23.6% higher than YZJ08. When GapN was introduced into ZS04, the impaired growth was improved, and FFA production at 72 h improved significantly. The final FFA titer of ZS05 reached 578.7 mg/L at 96 h, which was 28.6% higher than YJZ08 and the productivity was comparable with ZS04.
Similarly, the distributions of saturated and unsaturated FFA also varied in the constructed strains (Figure 3B). GPD deletion did not alter the distributions, whereas GapN expression increased the ratios of saturated FFAs. With GapN expressed in YJZ08, the percentage of saturated FFAs increased slightly from 50% to 53%, with the increases in the titers of C16:0 and C18:0 but no increases of C16:1 and C18:1. When GapN expressed in gpd1Δ gpd2Δ mutant, in ZS05 the percentage of saturated FFAs further increased to 59%, indicating the possible redox changes between NAD and NADP redox couples.
Increased provision of acetyl-CoA and NADPH improved FFA production
Plasmid pGapN was transformed into PDH1 and PDH2, yielding PDH3 and PDH6, respectively. As shown in Figure 4, the FFA production in PDH3 and PDH6 increased to 534.0 mg/L and 797.6 mg/L, respectively. Also, glycerol accumulation levels in PDH3 and PDH6 reduced to 0.93 g/L and 0.58 g/L, respectively. The decrease in glycerol level and increase in FFA titer in both PDH3 and PDH6 were similar with those observed in ZS02.
Similarly, when GPD1 and GPD2 were deleted in PDH1 and PDH2, the resulting strains PDH4 and PDH7 exhibited decreased growth rates, but increased FFA production to 664.0 mg/L and 840.5 mg/L, respectively (Figure 4). Finally, when GapN was introduced into PDH4 and PDH7, yielding PDH5 and PDH8, the impaired cell growth improved, whereas the FFA production decreased to 614.0 mg/L and 715.3 mg/L, respectively (Figure 4). The improved cell growth and decreased FFA titer upon GapN expression indicated a possible competition for the supply of the precursors and redox factors. Previous studies with heterologous acetyl-CoA pathways suggested that the ACL pathway is promising for production of fatty acid and their derivatives [6, 7, 10, 11, 14]. ACL expression with enhanced fatty acid synthesis resulted in 81% higher FFAs than YJZ08 [6, 7], and in this study the titer of PDH7 was 83.2% higher compared to YJZ08, similar with ACL expression combined with other engineering efforts.
Regarding the distribution of saturated and unsaturated FFAs (C16 and C18) of engineered PDH strains, it was found that PDH expression significantly increased the ratios of unsaturated FFAs (Figure 5A), even with efforts for redox cofactor rebalance. These changes might be because of the PDH complex with excess NADH produced and its consistent activity under high NADH/NAD+ ratios. The PDH complex may be a potential target to increase the unsaturation of the FFAs and their derivatives, besides manipulations on Delta-9 fatty acid desaturase OLE1 reported to control the fatty acid saturation in previous studies [40, 41].
The intracellular NADH/NAD+ ratios of the engineered PDH strains were lower than that of ZS01 (Figure 5B), suggesting the PDH complex alleviated NADH burden, as we conferred from cell growth and FFA production. GapN expression could further relieve the NADH burden, as the NADH/NAD+ ratio of PDH6 and PDH8 was lower than that of PDH2, and the ratio of PDH8 lower than PDH7.