Intermediates and gene expression associated with central carbon metabolism
Cells are able to assess the magnitude of metabolic flux by tuning the rate of metabolic reactions, consequently regulating cellular physiology under various stressful conditions (Litsios, Ortega, Wit, & Heinemann, 2018). During CVEF with glucose as the sole carbon source and major products such as ethanol, CO2, biomass, and glycerol, the central carbon metabolism pathways of S. cerevisiae cells mainly constitute the Embden-Meyerhof-Parnas pathway (EMP), the pentose phosphate pathway (PPP) and the tricarboxylic acid cycle (TCA) (Figure S3). Thus, metabolomics and transcriptomic datasets of yeast were integrated and interpreted via a pathway-based method, referring to 26 metabolites and 90 genes related to central carbon metabolism, whose responses in an oscillation period were visualized in Figure 3.
Glucose firstly fluxes into EMP to provide ATP and precursors for cell growth. EMP can be divided into an energy-consuming phase (–2 mol ATP/mol glucose from GLCin to GAP) and an energy-producing phase (4 mol ATP/mol glucose from GAP to PYR). Most of the pyruvate produced is converted into ethanol in the condition of low dissolved oxygen. The PPP as the source of NADPH and pentose contains an oxidative pathway (from G6P to Ri5P) and a non-oxidative pathway (from Ri5P to F6P and GAP). The TCA is a series of enzyme-catalyzed chemical reactions that form a key part of aerobic respiration in cells. As to the changing tendency of carbon metabolism during the oscillation (Figure 3), it was weakened during the P2 to P3 phase (especially the EMP and TCA processes) while being enhanced during the P3 to P5 with the decreasing residual glucose and increasing ethanol concentration.
The significant decline of most metabolites with associated down-regulated gene expression was observed in P3/P2, but the opposite phenomenon happened in P5/P4. The reason may be found in Figure 1 where biomass increased 3.3 g/L from P2 to P3, but decreased 3.32 g/L from P4 to P5 was observed. In addition, more changed metabolites and gene expression appeared in P3/P2 and P5/P4 than that observed for P2/P1 and P4/P3 and this may be due to the biomass difference between the adjacent samples. From Figure 1, the biomass changes between P2 and P3, P4 and P5 were about two-fold of that between P1 and P2, P3 and P4. The pool of cofactors and ATP was enhanced in P2/P1 and P5/P4, but decreased from P2 to P4, which reflected the activity of the energy metabolic reactions. In spite of an expected sinusoidal oscillation found in Figure 3, there are many additional details that can be elucidated by analyzing the metabolites and genes one by one (Figure 4).
Figure 4A showed the specific concentration and changing trend of each metabolite, most of which are involved in Figure 3, except adenylate energy charge (EC). Throughout the oscillation cycle, ATP was the dominant intracellular adenine nucleotide and the adenylate energy charge (EC = ([ATP] + 0.5*[ADP])/([ATP] + [ADP] + [AMP])) remained high (>0.9), which indicates that the cell has a strong potential for phosphoryl transfer and works as the ‘normally metabolizing’ cell during the CVEF. In addition, we also found that the changing tendency of EC is identical with biomass (Figure 1), which seems to indicate that a higher EC value is benifitial for cell proliferation. As described in Figure S1, most profiles of genes and metabolites are able to be divided into four patterns of sin function: 0π, π/2, π, 3π/2 phases. As shown in Figure 4B, the changing tendencies of 16 out of 22 genes related to EMP pathway presented a π/2 phase, while the rest five genes (HXK1 , HXT1,2,4 and PFK1 ) showed a 3π/2 phase, 0π phase and π phase, respectively. This inferred that the processes of glucose transport, glucose phosphorylation, and the catalysis of PFK could be the nodes for the oscillation regulation of intracellular metabolites. Genes PFK1 and PFK2 encoded the 4α- and 4β-subunits which compose the heterooctameric enzyme(PFK)in yeast and both types of subunits contributing equally to catalysis instead of that one of the subunits functioning preferentially in allosteric regulation (Heinisch, Boles, & Timpel, 1996). It’s interesting that PFK2 was not induced more drastically between P2 and P3/4 as PFK1 was (Figure 4B), which seems to indicate thatPFK1 is upregulated more rapidly and variably than PFK2 in response to the increased substrate concentration. In addition, PFK activity is subject to allosteric control mechanism, with inhibition by ATP and activation by AMP (Heinisch et al., 1996). Here, the change trend of ATP (0π phase) is contrary to PFK1 (π phase), which reflects its inhibitory effect on PFK allosteric control. But the change trend of activator AMP is not consistent with PFKs , probably because its absolute oscillation change values are much lower than that of ATP, making its effect was not shown intuitively in the overall trend of PFKs . Two distinct changing tendencies for the upper and down neighboring metabolites of PFK, G6P and F6P are in π/2 phase while FBP and GAP are in 0π phase. A margin of π/2 phase between their profiles was assumed to be induced by PFK activity to catalyze F6P to produce FBP, indicating that PFK is the key rate-limiting enzyme in the energy-consuming phase of the glycolytic pathway. This description is similar to the mechanism proposed for the metabolic oscillation in continuous culture or yeast extracts (Boiteux, Goldbeter, & Hess, 1975; Papagiannakis, Niebel, Wit, & Heinemann, 2017; Thoke et al., 2018).
In ethanol production after the EMP pathway, the RPKM value ofADH4 was always over 2000, peaking at P3 with a value of 4300, while PDC1 and ADH1 presented a π/2 phase, consistent with most genes of the EMP pathway, contrary with PDC5,6 andADH4 (Figure 4B). The expression profiles of ADH4 and PDC5 are π/2 phase earlier than ethanol, and the overall expression level ofADH4 is much higher than that of PDC5 , suggesting that Adh4p might be responsible for ethanol formation during CVEF. The geneADH4 encodes the zinc-dependent alcohol dehydrogenase isoenzyme, and its transcription is induced in response to zinc deficiency, which was consistent with the changing tendency of the expression level of zinc deficiency related gene ZPS1 (Figure S2). The expressions ofADH1 and PDC1 were regulated by the same regulatory factors, while that of ADH4 and PDC5 and PDC6 by a different kind. It could be assumed that the regulation of the expression of gene ADH1 and PDC1 were closely related to the flux changes of catabolism, while the regulation of the expression of gene ADH4 and PDC5 and PDC6 were closely related to the flux changes of anabolism and zinc level.
In addition, yeast cells produced 10.2-13.7 g/L glycerol during CVEF, and the changing tendency of the extracellular glycerol level showed a π phase pattern, which was similar to PFK1 but contrary to theGPD1, HOR2 and the ORP value (Figure S4), while those ofGPD2 and RHR2 showed π/2 phase pattern (Figure 4B). The expression of GPD1 and GPD2 are regulated by osmotic stress and redox homeostasis, and studies have shown that the HOG pathway where the GPD1 and GPD2 are located can respond to ethanol stress (Klein, Swinnen, Thevelein, & Nevoigt, 2017; Udom, Chansongkrow, Charoensawan, & Auesukaree, 2019), suggesting that the glycerol production during VHG oscillation was induced by osmotic stress and the redox environment.
The expression level of genes related to PPP was far below that in the EMP (Figure 4B), suggesting that cellular anabolism might be repressed during CVEF. The oxidation stage of the PPP pathway is from G6P to 6PG, and finally Ri5P. Both steps generate NADPH, which provides reducing power and precursor metabolites for the cell’s biosynthesis pathways. The changing tendency of 6PG showed π phase pattern, similar to biomass (Figure 1), indicating that yeast cell growth is closely related to NADPH synthesis. Furthermore, the expression level of ALD4, ALD6 ,MAE1 (Figure S5), and ADH4 exhibited a 3π/2 phase pattern, ahead of NADH and NADPH. The products of ALD4, ALD6 andMAE1 are involved in catalyzing cofactor NAD(P)H generation steps, suggesting that cofactor NAD(P)H generation might be the driving force of PPP flux and biomass accumulation.
As to the TCA cycle, in this study, CVEF was conducted with aeration of only 0.05 vvm, therefore, the metabolites fluxed into the TCA cycle might be at low concentration, which was confirmed by transcriptional analysis result that showed the expression levels of genes related to TCA cycle were much lower than these related to EMP pathway.