Extracellular acidity and oxygen availability conjointly control eukaryotic cell growth via modulation of cytoplasmic translation

Data compilation and the experimental background for the datasets

The data was collected from duplicate batch cultivations of homozygous hoΔ/hoΔ strain of S. cerevisiae BY4743 (MATa/MATα his3Δ1/his3Δ1 leu2Δ0/leu2Δ0 lys2Δ0/+ met15Δ0/+ ura3Δ0/ura3Δ0) grown in synthetic medium (Table 1) [33]. The dissolved oxygen (dO₂) saturation was maintained at above 90% in air+ cultures by sparging the cultures with air at a rate of 0.1 vvm, and was monitored in air− cultures. pH was controlled at 5.5 in pH+ cultures, and was only monitored in pH− cultures. The samples for transcriptome and metabolite analyses were collected at OD₆₀₀ range of 0.6–0.8. Biomass density was determined gravimetrically. Microarray analysis was carried out by as described in the GeneChip®Expression Analysis Technical Manual (relevant kits, chips, and instrumentation: Affymetrix Inc., USA). The tricarboxylic acid (TCA) cycle intermediate and amino acid profiles were acquired from the Agilent 6430 Triple Quadrupole LC-MS system (USA). The glucose, glycerol, ethanol and ammonium content of the supernatant was determined enzymatically. The unprocessed raw data files for transcriptome analysis were accessed from the European Bioinformatics Institute ArrayExpress Data Repository [34] under the Accession number E-MTAB-6628. The submitted data was in compliance with the MIAME [35] standards, with fully accessible experimental details. The growth characteristics, endometabolic and exometabolic amino acid and TCA cycle intermediate profiles were accessed from [32].

Microarray data preprocessing

The raw image files (.cel extension) were processed and normalised with their quality assessed by the dChip software [36]. The baseline array for normalization was selected as the array having the best call rate, which was 59.7% for this dataset. The expression values were obtained using the model-based expression modelling method and conducting mismatch probe background subtraction calculated from the difference between the perfect match and mismatch values. The smoothing method was running median. All single, array and probe outliers were checked and datasets were ensured to have <1% array outliers and <0.5% single outliers. The expression values were log₂ transformed prior to data analysis.

Data analysis

The significance of physiological characteristics and exometabolic metabolite measurements was evaluated employing 2-way ANalysis Of VAriance (ANOVA) for data with replicates followed by Tukey’s Honest Significant Difference (HSD) test for multiple testing correction. For the transcriptome analysis and endometabolic measurements Student’s t-test was employed for statistical analysis whenever sample size ≥3, followed by the Benjamini-Hochberg test for multiple testing correction of the false discovery rate (FDR). Two-way ANOVA for data with replicates was employed as required for datasets where samples size <3, followed by Tukey HSD test as post-hoc analysis for multiple comparisons. Significance was evaluated at the confidence level of α=0.05. Hierarchical clustering of samples was carried out using Hierarchical Clustering Explorer (HCE) 3.5 [36], with Pearson correlation employed as the distance metric and average linkage employed as the linkage metric. Princeton GO tools Generic GO Term Finder was employed for Gene Ontology Enrichment Analysis (http://go.princeton.edu/cgi-bin/GOTermFinder) accessed on (02/2018) [37]. Bonferroni correction was applied for p-values and FDR was included in the evaluation. Gene ontology associations inferred from electronic annotation (IEA) were excluded from analysis.

Extracellular pH and oxygen availability jointly control cellular growth

The physiological characteristics of the batch cultures where (i) the cells were grown in a pH− controlled and aerated environment (air+ pH+), (ii) in environments where neither of these parameters were controlled (air− pH−), or (iii) and (iv) only one parameter was controlled (air+ pH−, air− pH+), were investigated. Growth rate, mid-exponential phase and steady state biomass accumulation, uptake of main carbon (glucose) and nitrogen (ammonium) sources, and the secretion of main by-products (glycerol and ethanol), amino acids, and TCA cycle intermediates were analysed to shed light on how cellular physiology was affected by synergistic effects between oxygen availability and extracellular pH.

The pH of the controlled cultures was maintained in the range of 5.44–5.51 at all times, whereas the extracellular pH was observed to drop down to the range of 3.56–3.84 during the exponential phase of growth in pH− cultures. The pH decreased further down to the range of 2.81–2.96 as those cultures reached stationary phase. The dissolved oxygen content of the air+ cultures remained always above 90%. Conversely, the dissolved oxygen content was in the range of 45–63% during the mid-exponential phase of the cultures that were fully saturated with oxygen at the time of inoculation, and were later operated as micro-aerated systems. The dissolved oxygen content of these cultures were observed to drop down to below 5% as they reached stationary phase. The overhead oxygen content of all cultures was measured as 20.34±1.03% for the duration of the cultivation indicating that the positive pressure in the vessels did not yield to oxygen escape from dissolved to gas state; the overhead oxygen content did not vary significantly with air supplementation regime, except for marginally low overhead oxygen measured during the stationary phase of air− cultures (Table 2).

The specific growth rate (μ_s) at exponential phase of cultivation varied by 26% under different conditions, and it was the lowest (μ_s=0.31 h⁻¹) in air− pH− cultures. The time for the cultures to reach mid-exponential phase significantly increased with high oxygen availability (p-value=0.032). The optical density (OD) of the cultures at mid-exponential phase was significantly affected by the interactive effect of extracellular pH and oxygen availability (p-value=0.043), whereas the OD measured at stationary phase (prior to the diauxic shift) was significantly affected by the environmental pH only (p-value=0.002). In both instances, the OD measurements were lower in pH− cultures. The biomass densities of the cultures were significantly affected by the interactive effect of culture pH and oxygenation both at the mid-exponential phase and at stationary phase (p values=6.43×10⁻¹² and 7.96×10⁻⁵, respectively) (Table 3).

The uptake of the macronutrients acting as the primary carbon and nitrogen sources was significantly affected by an interaction between the two environmental factors under investigation (p-values=0.008 and 0.047 for glucose and ammonium, respectively). The amount of glucose consumed by the yeast populations varied by 25%, whereas the amount of ammonium consumed by the cultures varied only by 0.5%, although the differences between the cultures under different conditions were apparent, and significant in both cases. The ethanol and glycerol secretion of yeast cultures were significantly different (p-values=2.60×10⁻⁴ and 1.60×10⁻² for ethanol and glycerol, respectively), and extracellular pH and oxygen availability at mid-exponential phase exhibited an interaction effect on the amount of by-product secreted. Both by-products were detected at lower concentration in the pH− cultures than in pH+ cultures (Table 3).

The uptake of histidine or leucine, to which the strain employed in the study was auxotrophic, was not significantly different between cultures. Leucine and histidine supplementation in the medium was in excess of the cellular requirements by design, thus their uptake only created a marginal and insignificant difference. On the other hand, the extracellular concentration of seven amino acids (alanine, arginine, asparagine, aspartate, cysteine, methionine, and threonine) as well as five TCA cycle intermediates (citrate, isocitrate, malate, cis-acotinate, and pyruvate) were significantly different between cultivation conditions. Since the cultures were not supplemented with either of these small molecules, all were considered as metabolites; the cellular by-products metabolically produced and secreted in differing ratios in response to different cultivation conditions here explored. The differences in arginine, cysteine, methionine, threonine, citrate, and malate concentrations were significantly affected by the interactive effect between the two environmental factors under investigation. Extracellular alanine and isocitrate concentrations were significantly responsive to the dO₂ level, whereas asparagine, aspartate, cis-acotinate, and pyruvate concentrations were significantly responsive to extracellular pH only (Table 4). The differences in the extracellular amino acid and TCA cycle intermediate metabolite concentrations revealed the directionality of material flow through metabolic fluxes of these populations grown under different environmental conditions, and highlighted the preferential production and secretion of different families of amino acids, some of which employ TCA cycle intermediates as precursors for biosynthesis (Figure 1). Aspartate synthesized from the TCA cycle intermediate oxaloacetate acted as the major precursor amino acid for the amino acids, whose exometabolic profiles displayed significant differences in response to different environmental conditions investigated in this study.

Figure 1:

Extracellular distribution of amino acids from TCA cycle precursors.

The amino acids and their TCA cycle intermediate precursors are depicted in rounded rectangles connected either by solid lines indicating a single metabolic step or by dashed lines indicating multiple metabolic steps. The reversibility of the steps is indicated by the direction of the arrows. The red dashed box groups two metabolites produced from aspartate in the same metabolic process. The blue banners denote metabolite concentrations that are significantly high at dO₂>90% due to aeration effect, at low acidity due to pH effect, or at low acidity due to an interaction effect between culture pH and aeration levels. The orange banners denote metabolite concentrations that are significantly low at low acidity due to pH effect, or due to an interaction effect between culture pH and aeration levels.

Differences in growth characteristics were traceable in endometabolic and transcriptomic profiles of yeast populations

The differences observed in the growth characteristics and the physiology of the yeast populations in response to extracellular pH and dissolved oxygen availability led to an in-depth investigation of the accompanying intracellular changes of these cell populations in order to gain insight on the underlying molecular mechanisms, which are responsible for inducing these differences in the culture characteristics.

For this purpose, we first investigated how the intracellular TCA cycle intermediate and amino acid concentrations varied in these cultures. Significant differences were observed around the amino acid metabolism with the intracellular concentration of 14 of the 20 amino acids measured (alanine, arginine, glutamate, glutamine, glycine, histidine, isoleucine, leucine, lysine, methionine, proline, serine, threonine, and valine) as well as of four TCA cycle intermediates (α-ketoglutarate, malate, cis-acotinate, and pyruvate) being altered between cultivation conditions. The significant differences in alanine, arginine, glutamate, histidine, leucine, methionine, proline, serine, and valine concentrations, as well as in α-ketoglutarate, malate, and pyruvate concentrations were caused by an interactive effect of both culture parameters investigated as factors; extracellular acidity and oxygen availability. Furthermore, the differences in cis-aconitate concentration was only responsive to differences in oxygen availability and the differences in glycine, lysine and threonine concentrations were only responsive to differences in extracellular acidity. Interestingly, the differences in glutamine and isoleucine concentrations were responsive to differences in either extracellular acidity or oxygen availability alone, but these two factors did not have an interactive effect on the concentration of these intracellular metabolites (Table 5). Glutamate, which employed α-ketoglutarate as its TCA cycle precursor was identified as the key amino acid in repurposing the intracellular amino acid pools available within yeast (Figure 2).

Figure 2:

Intracellular distribution of amino acids from TCA cycle precursors.

The amino acids and their TCA cycle intermediate precursors are depicted in rounded rectangles connected either by solid lines indicating a single metabolic step or by dashed lines indicating multiple metabolic steps. The reversibility of the steps is indicated by the direction of the arrows. The blue banners denote metabolite concentrations that are significantly high at dO₂>90% due to aeration effect, at low acidity due to pH effect, or at low acidity due to an interaction effect between culture pH and aeration levels. The orange banners denote metabolite concentrations that are significantly low at low acidity due to pH effect, or due to an interaction effect between culture pH and aeration levels. Metabolic pathways are represented in ovals. The block arrows represent the uptake of glucose and ammonium as the main carbon and nitrogen source macronutrients, respectively. The banners indicate, using the same colour codes, the significant differences in the amount of metabolite consumed in response to differences in the control of aeration and extracellular acidity.

The transcriptional landscape of the yeast populations growing in different pH and dO₂ controlling and monitoring regimes were investigated in order to identify the changes in gene expression events accompanying or underlying the differences observed in the exometabolic and the endometabolic profiles of these cultures. The effect of controlling dissolved oxygen or the pH of the culture was investigated by pooling the cultures such that the effect of either factor was investigated irrespective of the other, or by conducting an analysis of variance to identify the unique effect of each factor on the transcriptional profiles as well as the interaction (i.e. the synergy) between the factors.

Pooling experiments by a single factor, irrespective of the other factor proved to be a very stringent criterion to identify differential expression at a threshold of p-value <0.05 post FDR correction (see Materials and methods for details) between air+ and air−, or between pH+ and pH− cultures. In fact, this was the primary indicator of the fact that the synergistic effects between the two factors had an important contribution to explaining the transcriptional landscape underlying these observations at the phenotypic and metabolic levels. The expression of only three genes were significantly affected as a response to dissolved oxygen availability, and these genes were significantly associated with establishing protein localization to membrane (p-value=0.0118). The expression levels of 433 genes were responsive to controlling extracellular pH, although these genes were not significantly associated with a specific molecular function or biological process. However, the set was enriched for genes that are located in membrane-bound organelles (p-value=4.48×10⁻⁶), and a subset of those genes were significantly associated with membrane rafts (p-value=0.0168).

Analysis of variance showed that the expression level of 156 genes was significantly affected only by the level of aeration of the culture, and of 286 genes only by the extracellular pH of the cultivation environment. The transcription of 50 more genes were differentially affected by the interaction between the extracellular pH and dissolved oxygen availability. The 156 genes whose expression levels were affected only by the level of dissolved oxygen of the culture were not significantly associated with any biological process or molecular function, nor with any specific subcellular compartment of the cell, as classified by the Gene Ontology. On the other hand, a subset of the 286 genes, whose transcription levels were affected by extracellular pH, was significantly associated with actomyosin contractile ring organization (p-value=5.48×10⁻³), a cytokinesis-associated phenomenon. The subset of genes whose transcription was significantly altered in response to an interaction effect created by both oxygen availability and extracellular acidity were involved in growth-associated cellular mechanisms. These genes were significantly associated with cytoplasmic translation (p-value=0.0155); more specifically, they were genes encoding proteins that constitute the cytoplasmic ribosome structure (p-value=2.20×10⁻⁴), and unsurprisingly, the gene products were significantly associated with the ribosome (p-value=4.00×10⁻⁴).