3.1. Constructing the biosynthetic pathway to produce tyrosine from ethanol
For constructing the biosynthesis of tyrosine from ethanol, ethanol needs to be assimilated to form acetyl-CoA (Figure 1), which can be transformed through TCA cycle, glyoxylate shunt gluconeogenesis and pentose phosphate pathway (PPP) into two key node intermediates, phosphoenolpyruvate (PEP) and erythrose 4-phosphate (E4P). In the presence of D-arabino-heptulosonate-7-phosphate (DAHP) synthase (encoded by aroF, aroG or aroH) 6,13, PEP and E4P can be condensed into DAHP (Figure 1). DAHP then enters the shikimate pathway to form chorismite, a precursor to all three aromatic amino acids: L-phenylalanine 14, L-tyrosine and L-tryptophan 15.
In a previous study 16, we constructed a series of strains for producing tyrosine from glucose. Two genes (tyrR and pheA) were deleted to interrupt the regulation of tyrosine biosynthesis and to prevent accumulation of phenylalanine as a by-product. The strains were engineered to overexpress a feedback-resistant DAHP synthase (encoded by aroG* [Asp146 to Asn], 17 and a feedback-resistant chorismate mutase/prephenate dehydrogenase (encoded by tyrA* [Met53 to Ile and Ala354 to Val]). The strains carried plasmids with different replication origins (p5, p15, pMB1, and pUC18 and promoters (PT7, 19 to express aroG* and tyrA* in various orders, resulting in E. coli TYR01, TYR02, TYR03 and TYR04 (The strain information is summarized in Table 1).
We first examined the possibility of coupling the ethanol assimilation module with the tyrosine production module (Figure 2a) in E. coli. Previously, we had constructed 2 the ethanol utilization pathway (EUP), in which ada and adh2 were overexpressed by using promoter PgyrA and replication origin pMB1 (Plasmid: pEUP02). To balance the ethanol utilization module and tyrosine production module, we further constructed two plasmids with different replication origins (p15, pRSF), resulting in new plasmids pEUP01 and pEUP03. The three plasmids, pEUP01, pEUP02 and pEUP03 was introduced into TYR01, TYR02, TYR03 and TYR04, resulting in 10 new strains (from ETY01 to EYT10). The ability of producing tyrosine was investigated in a chemically defined medium containing 10 g/L ethanol and 1 g/L CSM. All the strains can grow on ethanol and consumed all the ethanol by 72h (Figure 2b). Nine out of the ten strains produced tyrosine, ranging from 0.2 to 1.0 g/L (Figure 2b). The best strain (ETY08) carried a high copy number ethanol-utilizing plasmid and a medium copy number tyrosine-producing plasmid.
3.2. Optimizing expression of the enzymes in tyrosine biosynthesis and simplify the fermentation process
We next systematically optimized gene expression involved in tyrosine biosynthesis since the abundance of enzymes inside the cell determines the intracellular metabolic fluxes. The two genes involved in tyrosine synthesis were arranged in different orders (aroG*-tyrA* or tyrA*-aroG*) with two different promoters (PT7 and PLac) and medium copy replication origin (p15) (Figure 3a). The new strains (ETY11, ETY12 and ETY13) should have expressed the two genes at different levels. The best strain (ETY11) improved the tyrosine titer by 30% (to 1.3 g/L by 72 h) and used the PLac promoter to drive the aroG*-tyrA* operon (Figure 3b). IPTG was added to the final concentration of 0.001 mM during the early exponential growth phase 20.
We next evaluated different concentrations of IPTG in the range of 0-0.5 mM and found that the highest tyrosine titer (1.6 g/L) was achieved when 0.1 mM IPTG was added during the growth phase (Figure 3c). We further simplified the fermentation by adding 0.1 mM IPTG upon inoculation and found that similar cell growth, ethanol consumption, and tyrosine production profiles were obtained. This condition was used in the following experiments.
3.3. Transcriptome analysis
When ETY11 was grown on 10 g/L ethanol and 1 g/L CSM, 1.6 g/L L-tyrosine was produced (Supplementary figure 4.1) which is higher than while 10 g/L of glucose was added. The first possible reason is that no by-products were detected when ethanol was used as the carbon source, whereas 4 g/L of acetate was produced when glucose was used. We hypothesized that ethanol may have triggered transcriptional regulations 21 that favor tyrosine production. To gain the depth insights into the mechanisms, we performed a transcriptome analysis.
We extracted RNA from cells that were cultured in a chemically defined medium containing 10 g/L of ethanol, or 10 g/L of glucose. Both cultures contained 1 g/L CSM at the start of the fermentation. The RNA extraction 22 was conducted during the early exponential growth phase (OD600 is around 0.5). The samples were sequenced using the Illumina PE150 technology.
The glyoxylate shunt 23 was upregulated when the cells were cultured in ethanol (Figure 4). An isocitrate lyase (encoded by aceA) 24 and malate synthase (encoded by aceB) were upregulated by 74-fold and 60-fold respectively, when glucose was replaced with ethanol. AceA splits isocitrate into succinate and glyoxylate (Figure 4). Succinate may be used to return the oxaloacetate borrowed at the first step of the TCA cycle, and the glyoxylate can be condensed by AceB with another molecule of acetyl-CoA to form malate.
Malate can be transformed into PEP 25 through two pathways (Figure 4). PEP is an important building block of L-tyrosine (making one L-tyrosine needs two PEP and one E4P). The first pathway oxidatively decarboxylates malate into pyruvate through malic enzymes and the pyruvate is activated into PEP using ATP with PEP synthase (encoded by ppsA) as the catalyst. The second pathway oxidizes malate into OAA (Figure 4) through the malate dehydrogenase (encoded by mdh), and OAA is converted into PEP by Pck (PEP carboxykinase). Genes ppsA, mdh and pck (Figure 4) were upregulated by 1600%, 160% and 100% respectively when ethanol was replaced with glucose, suggesting that both pathways may have been used to produce PEP when ethanol was used. Considering that taking up ethanol does not consume PEP, we could hypothesize that the PEP availability was better in the ethanol medium than the glucose medium, which may be a key reason why L-tyrosine yield was higher when ethanol was used.
We attempted to further activate the glyoxylate cycle by overexpressing aceA using plasmids with different replication origins (p5, p15 and pMB1). However, the tyrosine yield was not improved (Data not shown).