Bioinformatic and phylogenetic analyses

The genomes of T73 (NCBI SRA ref. SRX4791864) and BMV58 (NCBI SRA ref. SRX4086831) have previously been sequenced, assembled and annotated in our laboratory (Macías et al., 2021; Morard et al., 2019). From them, the sequences of the putative genes of this study were obtained. Y. lipolytica ER‐encoding gene sequences were obtained from the National Center for Biotechnology Information (NCBI) (http://blast.ncbi.nlm.nih.gov/Blast) and their homologous genes in Saccharomyces were searched with Basic Local Alignment Search Tool (BLAST) using ‘The BLAST Sequence Analysis Tool’. Other yeast AKR sequences were obtained from the AKR superfamily homepage (Perelman School of Medicine, University of Pennsylvania, www.med.upenn.edu/akr), maintaining the same references used in this database. Sequences were translated into amino acids and aligned with MUSCLE (Edgar, 2004) implemented in the MEGA v. 11.0.13 program (Tamura et al., 2021). The best evolutionary protein models based on empirical frequencies of amino acid replacements were selected with ProtTest 3.4.2 (Darriba et al., 2011), and the best‐fitting model was the LG model (Le & Gascuel, 2008) with gamma‐distributed rates with an α shape parameter of 1.57 and the observed amino acid frequencies. Maximum‐Likelihood phylogenies were also obtained by using the MEGA v. 11.0.13 program (Tamura et al., 2021).

Knockout production by CRISPR‐Cas9

CRISPR‐Cas9 was used to knock out the putative genes in S. uvarum BMV58 following the protocols described for Saccharomyces (Dicarlo et al., 2013). The gRNA for each gene was designed using Geneious software to reduce the off‐target probability and maximize correct Cas9 nuclease guidance. This gRNA was inserted into pRCC‐K Cas9 plasmid (Generoso et al., 2016) by PCR using the appropriate primers (Table S1) and Phusion™ High‐Fidelity DNA Polymerase (Thermo Fisher) followed by DpnI treatment (Thermo Fisher) to remove the original E. coli plasmid. Donor ssDNA was designed to be used as a template during homologous recombination, this sequence contains homologous regions upstream and downstream of the ORF of each gene and includes the stop codon (Stovicek et al., 2017). Forward and reverse ssDNAs (Table S1) were incubated and dsDNA was obtained. The S. uvarum strain was transformed using the lithium acetate method for DNA extraction (Gietz & Woods, 2006) with the plasmid containing each gRNA and with each donor DNA. Mutants were grown on GPY plates containing G418 antibiotics for the selection because pRCC‐K has a neomycin‐resistance marker. After 3days, the obtained colonies were tested to verify the deletion with each specific primer set for each gene (Table S1) and with genomic DNA extracted using the lithium acetate method (Lõoke et al., 2011). Colonies that were confirmed as positive knockouts were plated on GPY without G418 to get rid of the plasmid. Finally, verified mutants were stored in 15% glycerol. When double (or multiple) knockouts were done, the single (or multiple) mutants were used as the initial strain, and the same procedure was followed.

Fermentations

To determine if the putative ERs were responsible for the erythrose reduction to erythritol, single and multiple knockout strains of each gene were used in fermentation at 25°C, and the amount of erythritol produced was measured. Fermentations were performed in 100‐mL bottles with 70mL media of synthetic must (Rossignol et al., 2003) with different concentrations of sugar (glucose and fructose) ranging from a total of 200–260g/L and two YAN (Yeast Assimilable Nitrogen) concentrations, 200 or 300mg/L. Overnight precultures of each strain (wild type and mutants) were used to inoculate in triplicate the synthetic grape must at 2×10⁶C/mL using a Guava® Muse® Cell Analyser (Merck) and bottles were incubated at 25°C with continuous orbital shaking (100rpm). Weight loss of the bottles was measured to follow the fermentation phases and samples were taken at the end of the fermentation to measure the metabolites.

Extracellular metabolite quantification

The quantification of the sugars and fermentative by‐products (glycerol, ethanol, 2,3 butanediol, erythritol, succinic, acetic and lactic acids) was done by High‐Performance Liquid Chromatography (HPLC) (Thermo Fisher Scientific) that has a refraction index detector and UV/VIS (210nm) detector equipped with a HyperREZTM XP Carbohydrate H⁺ 8mm column (Thermo Fisher Scientific) and HyperREZTM XP Carbohydrate Guard (Thermo Fisher Scientific). Regarding the analysis conditions, the eluent was 1.5mM of sulfuric acid at a flux of 0.6mL/min and the oven temperature was 50°C. The samples were diluted, filtered through a 0.22‐μm nylon filter (Symta) and injected. To quantify the metabolites of interest, the retention times of the eluted peaks were compared to those produced by commercial analytical standards with known concentrations (Sigma‐Aldrich).

Gene expression analysis

To correlate if the putative ERs were expressed at a higher level when erythritol production increased, the relative mRNA levels of these genes were determined. The fermentation was performed at 25°C using 500mL bottles with 400mL of synthetic must (Rossignol et al., 2003) with 240g/L glucose + fructose and 300mg/L YAN.

1×10⁸ cells were immediately harvested after 0, 1, 4, 24, 48, 72, 96 and 192h by centrifugation, then cells were washed and frozen in liquid nitrogen_. RNA from the frozen cells was extracted using RNeasy Mini Kit (Qiagen). This RNA was converted to cDNA by adding 0.8mM of the four dNTPs, 80pmol Oligo (dT) to 2μg of RNA in 13μL. This was heated at 65°C for 5min and incubated on ice for 1min. Then in a 20‐μL reaction, 5mM dithiothreitol (DTT), 50U of RNase inhibitor (Invitrogen), First Strand Buffer (Invitrogen), and 200U Superscript III (Invitrogen) were added and incubated at 50°C for 60min and then, inactivated at 70°C during 15min. The desired genes (GCY1, YPR1, ARA1, GRE3, YJR096 and GPD1) were quantified by qRT‐PCR (quantitative real‐time PCR) by using gene‐specific primer sets (Table S2) and the Light Cycler FastStart DNA MasterPLUS SYBR green (Roche Applied Science) in a LightCycler® 2.0 System (Roche Applied Science). The actin ACT1, and 18S ribosomal RNA RDN18‐1, genes were used as constitutive genes to normalize the amount of mRNA and guarantee accuracy, repeatability and reproducibility (Starovoytova et al., 2013).

In parallel, the metabolites produced during fermentation at the same sampling times were measured by HPLC as previously described.

Osmotic and cold stress assays

T73, BMV58 and the selected knockouts were cultured in 50mL of GPY at 25°C. Overnight cultures were centrifuged, washed and resuspended in water. The initial cell concentration of the preculture was measured and adjusted using a Guava® Muse® Cell Analyzer (Merck) so that 50μL were inoculated (in triplicates) into 950μL media, on a 24‐well plate, to reach 1×10⁶ cells/mL.

For the determination of tolerance to sorbitol, experiments were carried out in minimal medium YNB with 20g/L glucose as a control condition and different sorbitol concentrations as test conditions: 1, 2 and 3M. The microplates were incubated inside a camera that maintains 75% humidity at 25°C. The settings of the microplates were shaking in an orbital mode for 20s each half an hour at 300rpm. In the case of cold stress, the same settings were used although the temperature was 12°C. 1mM erythritol was added to each stress condition (cold or different sorbitol concentrations), to know if its addition was useful for yeast growth. Moreover, for cold stress conditions, additions of 5 and 10mM erythritol were also tested. Besides, 10mM glycerol was used as a control due to its known properties to protect against stress. Yeast growth in each different medium was followed by measuring the OD₆₀₀ in Stacker Microplate Handling System attached to plate readers SPECTROstar Omega (BMG LABTECH). The OD₆₀₀ results were analysed by the Growth Curve Analysis Tool (GCAT) (Bukhman et al., 2015) using the sigmoid model option GCAT selected between Richards, Gompertz and logistic sigmoid curve models. Then, the reparametrized formulas (Zwietering et al., 1990) were used to represent growth in terms of essential growth curve characteristics: lag time, growth rate and asymptotic growth value, and these were analysed.

Knockout mutant generation

To determine the role of the Saccharomyces genes with potential ER activity, we decided to obtain knockout mutants of these genes in BMV58, as this is the strain with the highest erythritol production reported in our laboratory (Minebois et al., 2020a, 2020b). Single, double and triple knockouts were confirmed by PCR with the corresponding primers (Table S1) followed by electrophoresis on an agarose gel. When each gene was present in the WT, there was a band of GCY1 (1495bps), YPR1 (1544bps), GRE3 (1190bps), ARA1 (1180bps) and YJR096W (1151bps). In the knockout mutants, these amplification bands were smaller GCY1 (559bps), YPR1 (608bps), GRE3 (209bps), ARA1 (148bps) and YJR096W (305bps), indicating that the genes were successfully deleted. The efficiency of the transformation ranged between 40% and 60%. The successful knockouts (Table S4) were plated on GPY with G418 to ensure that they were not able to grow, confirming they lost the pRCC‐K plasmid.

Erythritol synthesis during fermentation

To test if the selected genes are involved in the synthesis of erythritol, the mutants were used in fermentation in synthetic must. Wild‐type (WT) S. cerevisiae T73 and S. uvarum BMV58 strains were also inoculated in the same conditions as controls, and erythritol was measured at the end of fermentation. The production of erythritol has been reported to increase at high sugar concentrations in Y. lipolytica (Janek et al., 2017). Therefore, we performed fermentations at 25°C using different sugar and nitrogen concentrations to know which condition led to the highest erythritol yield. When using 200g/L sugars and 200mg/L YAN, we obtained a relatively low amount of erythritol, and the knockouts did not significantly reduce its production (Figure S2). However, when using 240g/L sugars and 300mg/L YAN, the production of erythritol was the highest, being (0.72±0.02g/L) (Table S5). Thus, this medium was selected to test the effect of the deletion of the putative ER genes. Strikingly, we observed that the WT strain of S. uvarum produced more than twice the amount of erythritol than S. cerevisiae in this medium (Figure 2).

Open in a separate window

FIGURE 2

Concentrations of erythritol produced by Saccharomyces cerevisiae T73 (dark blue), S. uvarum BMV58 (red) as WT, the single mutants (green), double mutants (light blue), triple mutants (purple), quadruple mutant (orange) and quintuple mutant, with all selected genes deleted (black). This concentration was measured after fermentation on SM with 240g/L glucose + fructose and 300mg/L YAN. Standard deviation is shown, and the asterisk represents significant differences at the level of 0.05 in comparison with the WT (BMV58).

When using 240g/L sugars and 300mg/L YAN, ΔGRE3 was the single mutant showing the largest reduction in erythritol synthesis (0.4±0.01g/L). The double ΔARA1ΔGRE3, ΔYJR096WΔGRE3 and triple ΔYJR096WΔGRE3ΔARA1 knockout mutants significantly produced less erythritol than the WT, but the same amount as the single mutant ΔGRE3. Therefore, the combination of the deleted genes YJR096W and ARA1 did not decrease the production more than ΔGRE3. The double deletant ΔARA1ΔYJR096W significantly reduced erythritol synthesis compared to the WT, although less than ΔGRE3. In the case of the paralogous GCY1 and YPR1, neither the single mutants, ΔYPR1 and ΔGCY1, nor the double mutant ΔYPR1ΔGCY1 decreased erythritol production. The double deletion of GCY1 or YPR1 with GRE3, ΔYPR1ΔGRE3 and ΔGCY1ΔGRE3, decreased erythritol synthesis to a similar level to ΔGRE3. Nevertheless, the combined deletion ΔYPR1ΔGCY1ΔGRE3 is the triple knockout exhibiting the highest decrease in erythritol production (0.21±0.03g/L), even more than the single ΔGRE3 mutant does. Therefore, YPR1, GCY1 and GRE3 could be the main genes involved in erythritol production. Finally, the combined deletion of the five AKR genes, called Δ5K, produced almost a 3‐fold reduction of the erythritol yield compared to the WT, although there was some production yet (Figure 2).

Analysis of the possible role of erythritol in osmotic and cold stress tolerances

The higher erythritol production observed in Saccharomyces yeast grown in media with high sugar concentrations indicates that erythritol synthesis could be related to a response to osmotic stress. To determine if erythritol could play a role in osmotic stress tolerance, strains T73, BMV58 (WT) and the BMV58 knockouts ΔGRE3, ΔYPR1ΔGCY1ΔGRE3 (from now called Δ3K) and ΔYPR1ΔGCY1ΔARA1ΔYJR096W (called Δ4K) were grown under different sorbitol concentrations (1, 2 and 3M). As a control, the addition of 10mM glycerol was also tested because it is well‐known that glycerol is produced and accumulated in yeasts to counterbalance external osmotic pressure (Blomberg & Adler, 1989).

A simple way to summarize the effect of the osmotic stress on the growth of the yeast is to observe significant changes in the maximum specific growth rate (μ _max), which is a key parameter calculated as the first derivative at the inflection point of the growth curve (Zwietering et al., 1990). Our results showed the effect of osmotic stress due to increasing concentrations of sorbitol in the growth media. At 1M sorbitol, no effect was observed in the μ _max of T73, but the μ _max was decreased in strain BMV58 and all its knockout mutants (Figure 3A). At 2M sorbitol, all the strains were significantly affected, especially the quadruple knockout Δ4K that was unable to grow (Figure S3). Finally, in the presence of 3M sorbitol, no growth was observed in any of the strains (data not shown). Focusing on 1M sorbitol on BMV58 it was observed that this stress significantly reduced the μ _max from 0.52 to 0.35. Nevertheless, with the addition of erythritol and glycerol, growth was recovered, being at the same level as without sorbitol stress with erythritol addition (0.45) and not significantly different with sorbitol when adding glycerol (0.4). In the case of ΔGRE3, the presence of 1M sorbitol significantly reduced growth. Although μ _max was recovered by adding erythritol and glycerol at the same level as without sorbitol stress. The same effect happened on Δ4K, although μ _max was lower, being 0.12 with 1M sorbitol stress and doubling this value without stress or with stress but adding erythritol or glycerol. In the case of Δ3K, the effect of the addition of erythritol with 1M sorbitol stress, significantly increases μ _max (being 0.48, even higher than without any stress that was 0.33). Also, the addition of glycerol recovered growth until 0.41.

Open in a separate window

FIGURE 3

(A) μ _max (log. OD/h) of Saccharomyces cerevisiae (T73), S. uvarum (BMV58), BMV58 mutants ΔGRE3, ΔYPR1ΔGCY1ΔGRE3 (Δ3K) and ΔYPR1ΔGCY1ΔARA1ΔYJR096W (Δ4K) in YNB at 25°C without sorbitol and with different concentrations of sorbitol, as well as sorbitol and the addition of erythritol or glycerol. Values represent media of biological triplicates and the standard deviation is shown. (B) μ _max log. OD/h of S. cerevisiae (T73), S. uvarum (BMV58), BMV58 mutants ΔGRE3, ΔYPR1ΔGCY1ΔGRE3 (Δ3K) and ΔYPR1ΔGCY1ΔARA1ΔYJR096W (Δ4K) in YNB at 25 and 12°C as cold stress as well as 12°C with the addition of erythritol and glycerol. Values represent media of biological triplicates and the standard deviation is shown. The statistics were performed by anova and Tukey as a post‐hoc range test.

To study the role of erythritol in cold stress, we performed fermentation at low temperature (12°C) with the selected knockout strains ΔGRE3, Δ3K and Δ4K, and T73, BMV58 as WT, in the presence of erythritol at different concentrations: 1mM, a low concentration similar to the amount that S. cerevisiae produces, 5mM as a medium concentration, similar to the highest concentration produced by S. uvarum, and 10mM as a high concentration for comparative purposes because fermentations at low temperature was also performed in the presence of 10mM glycerol, a concentration produced and accumulated in Saccharomyces species as a cryoprotectant (Oliveira et al., 2014).

As controls, fermentation in the absence of erythritol were performed at low (12°C) and medium (25°C) temperatures. A temperature of 12°C is stressing for yeast growth, which significantly decreased in all the strains (Figure 3B). The addition of glycerol and erythritol improved growth under cold stress in all strains, although μ _max values were not as high as the ones at 25°C. There were no significant differences between the addition of glycerol and erythritol except in T73, where it was observed that the effect of 10mM glycerol is the same as 1 and 5mM erythritol, but the μ _max was higher when using 10mM erythritol. In the case of Δ4K, the μ _max is lower than the rest, being (0.27±0.03) at 25°C and (0.04±0.004) at 12°C. In this strain, the addition of erythritol did not recover growth significantly, but glycerol did. Δ4K lacks GCY1, YPR1, ARA1 and YJR096W but GRE3 is present; therefore, this could be due to different responses to cold stress. To sum up, these results suggest that erythritol contributes to growth recovery when cold and osmotic stresses are present (Figure 3).

Selected mutant performance during fermentation and analysis of the expression of the putative genes involved in erythritol synthesis

Fermentations at 25°C were conducted by the selected knockout strains ΔGRE3, Δ3K and Δ4K, and the WT strains BMV58 and T73 to study their performance. The production of important fermentative by‐products (ethanol, glycerol, succinate, etc.) was also quantified to assess the impact of the deleted genes on the central carbon metabolism during fermentation. Samples were taken at 0, 1, 4, 24, 48, 72, 96, 192h, and at the end of the fermentation, to measure for each knockout mutant both the amount of metabolites produced as well as gene expression levels, described in the next section.

The differences in extracellular concentrations of the main products of fermentation are depicted in Figure 4 and Table S6. The gene expressions among the different knockout mutants and reference strains are shown in Figure 5. As expected, the fastest growth is exhibited by the S. cerevisiae T73 strain, followed by S. uvarum BMV58 and its mutant derivatives, which behave in the same manner, and, although deletant Δ4K shows a delayed growth, it finished at the same time (Figure S5).

Open in a separate window

FIGURE 4

Metabolic pathways of Saccharomyces including glycolysis and pentose phosphate pathway, which are involved in erythritol metabolism, are the main cofactors and the enzymes that could have an impact on erythritol formation. The graphics show the extracellular concentrations of the main products of fermentation on SM 240g/L sugars and 300YAN measured by HPLC produced by S. cerevisiae T73 (blue), S. uvarum BMV58 (red) and BMV58 mutants ΔGRE3 (green), ΔYPR1ΔGCY1ΔGRE3 (Δ3K, purple) and ΔYPR1ΔGCY1ΔARA1ΔYJR096W (Δ4K, orange). The values are presented as the mean and SDs from triplicate experiments. For more detail, the statistics of all metabolites can be found in Table S6.

Open in a separate window

FIGURE 5

Expression of the selected genes represented by the mRNA levels of the genes relative to ACT1 and 18S ribosomal RNA gene at 0, 1, 4, 24, 48, 72, 96 and 192h of fermentation by Saccharomyces cerevisiae (T73), S. uvarum (BMV58) and BMV58 mutants ΔYPR1ΔGCY1ΔGRE3 (Δ3K) and ΔYPR1ΔGCY1ΔARA1ΔYJR096W (Δ4K). The values represent the median and standard deviations of biological and technical triplicates.

Concerning erythritol (Figure 4), the highest yield was observed in BMV58 (0.72±0.02g/L), which started to produce extracellular erythritol after 48h of fermentation increasing at a constant rate until the end of fermentation. Mutant Δ4K, in which GRE3 is not deleted, exhibited a similar but delayed pattern in which a slightly lower production started after 72h of fermentation but increased at a similar constant rate as BMV58 until the end of fermentation. In the case of T73 and ΔGRE3, they showed a similar pattern of erythritol production, but slightly higher in ΔGRE3 (0.4±0.01g/L). In them, production started after 72h at a similar level that in BMV58, but along the rest of fermentation only increased slightly at a very low rate. Finally, mutant Δ3K started to produce erythritol after 96h of fermentation but after that, the concentration remained constant until the end of fermentation. At the end of the fermentation, strain BMV58 produced the highest erythritol contents; followed by mutant Δ4K (0.54±0.02g/L). Mutant Δ3K produced significantly less erythritol than the rest of the strains (0.21±0.03g/L).

The effect of the genes knocked out in BMV58 also had an impact, in addition to erythritol, on other metabolites, such as ethanol, glycerol, 2,3‐butanediol, succinic acid and acetic acid.

The production of ethanol (Figure 4) follows a similar pattern, indicating that the kinetics and yield of ethanol in the knockouts ΔGRE3 and Δ3K were not affected compared to BMV58 WT (11.86%±0.68%), while the kinetics was delayed in Δ4K, ending up to 12.71±0.22%. In the case of glycerol production, as expected S. uvarum BMV58 is the highest producer (8.08±0.65g/L) but the three knockout mutants produced very similar yields. In terms of acetic acid, the highest production was observed in Δ4K (8.39±0.16g/L), then BMV58 (0.62±0.04g/L), the rest of the mutants, and T73 presented the lowest production of acetic acid (0.46±0.02). BMV58 produced 1.22±0.05g/L of 2–3 butanediol, ΔGRE3 produced 1.15±0.07g/L, and Δ3K ended up with 1.12±0.05g/L, values significantly higher than the 0.47±0.07g/L produced by T73. Mutant Δ4K produced less 2–3 butanediol (0.9±0.04g/L) than the rest. Δ4K produced less ethanol, 2–3 butanediol and succinic acid, but more acetic acid (Figure 4), compounds that have in common an acetaldehyde precursor. Ethanol and 2–3 butanediol produce NAD+, meanwhile during succinate and acetate production, NAD(P)+ is consumed. Therefore, this is a case of different flux pathways to balance the cofactors needed in other processes.

In global, the results of the production of metabolites in all the mutants developed in the present study are presented in a heat map where the strains were grouped by hierarchical clustering (Figure S6). BMV58 exhibits a pattern similar to those observed for ΔYJR096W, ΔARA1, ΔGRE3, ΔYPR1 and ΔGCY1 (single mutants), characterized by higher production of erythritol, succinic acid and glycerol but lower lactic and acetic acids. On the contrary, T73 metabolite production was more similar to that exhibited by multiple deletants ΔYPR1ΔGCY1ΔARA, ΔYPR1ΔGCY1ΔYJR096W, Δ4K and Δ5K, which synthesize lower amounts of erythritol, 2,3‐butanediol, succinic and lactic acids, but higher of acetic acid and ethanol. All the mutants with a deletion of gene GRE3 decreased erythritol production and showed higher levels of succinic and lactic acids.

The authors want to thank the entire research team of the Systems Biology of Yeasts of Biotechnological Interest group of the IATA‐CSIC. We also thank Dr. Eva Balsa‐Canto for her comments and suggestions to improve this article. SA was supported by an FPI contract from Ministerio de Ciencia, Innovación y Universidades (ref. PRE2019‐088621). This project has received funding from the Spanish Government and EU ERDF‐FEDER projects PID2021‐126380OB‐C31 and PID2021‐126380OB‐C33 to AQ and EB, respectively. Finally, IATA‐CSIC received funding from the Spanish Government, ref. MCIN/AEI/10.13039/501100011033, as a ‘Severo Ochoa’ Center of Excellence (CEX2021‐001189‐S), with AQ as PI.