Enhancement of Sphingolipid Synthesis Improves Osmotic Tolerance of Saccharomyces cerevisiae

This study demonstrated a novel strategy for the manipulation of membrane complex sphingolipids to enhance S. cerevisiae tolerance to osmotic stress. Elo2, a sphingolipid acyl chain elongase, was related to osmotic tolerance through transcriptome analysis of the wild-type strain and an osmosis-tolerant strain generated from ALE. Overexpression of ELO2 increased the content of complex sphingolipid with longer acyl chain; thus, membrane integrity and osmotic tolerance improved.

were downregulated. In mutant XCG001, the expression levels of 609 genes displayed differential expression, where 416 genes were upregulated and 193 genes were downregulated. Additionally, 180 upregulated and 68 downregulated genes were common to both strains. Gene Ontology (GO) analysis indicated that the commonly upregulated genes were involved in glycolysis/gluconeogenesis, pyruvate metabolism, lipid metabolism, signaling transduction, fructose, and mannose metabolism. On the other hand, 68 downregulated genes were involved in the ribosome and amino acid metabolism (see Data Sets S1 and S2 in the supplemental material).
The significantly expressed genes in mutant XCG001, relative to those in the wild-type strain, then were analyzed at both 0 M and 1.5 M NaCl (Fig. 1F). At 0 M NaCl, the expression levels of 212 genes were upregulated and 271 genes were downregulated. At 1.5 M NaCl, 131 genes were upregulated and 235 genes were downregulated. These 131 upregulated genes include 13 genes that were commonly upregulated at 0 M and 1.5 M NaCl, and 118 genes were significantly upregulated only at 1.5 M NaCl. Based on the GO analysis, these 118 genes were involved in the steroid biosynthesis process, pentose-phosphate shunt, translation, regulation of transcription, phosphate ion transport, and response to stress. Moreover, 13 commonly upregulated genes were involved in transport, pyrimidine metabolism, and lipid metabolism, whereas 28 commonly downregulated genes were involved in pyruvate metabolism and transport (Data Sets S3 and S4). These results suggested that mutant XCG001 strengthened transport, pyrimidine metabolism, and lipid metabolism, which contribute to osmotic tolerance.
Overexpression of ELO2 enhanced very long fatty acid contents. The fatty acid contents of the strains XCG016, XCG017, XCG010, and XCG002 were analyzed using gas chromatography. It was found that the contents of membrane fatty acids in strains XCG016, XCG017, and XCG010 were altered, especially that of C 22:0 ( Fig. 3A and B). At 0 M NaCl, the contents of C 20:0 , C 22:0 , and C 24:0 in strain XCG010 increased by 52.3%, 94.1%, and 14.4%, respectively, compared with the corresponding values of the control strain XCG002, whereas the contents of C 16:0 , C 16:1 , C 18:0 , and C 18:1 remained the same. At 1.0 M NaCl, the contents of C 20:0 , C 22:0 , and C 24:0 in strain XCG010 increased by 33.1%, 106.4%, and 31.5%, respectively, while the contents of C 16:0 , C 16:1 , C 18:0 , and C 18:1 remained unchanged. All of the fatty acid contents in strains XCG016 and XCG017 were similar to those in strain XCG010 at 0 M or 1.0 M NaCl ( Fig. 3A and B).
The average fatty acid length in strains XCG016, XCG017, and XCG010 was equal to that of the control strain XCG002 at 0 M or 1.0 M NaCl, suggesting that the membrane "thickness" was not affected (Fig. 3C). The reason for unchanged membrane thickness may be that the proportion of C 20:0 and C 22:0 contents to the total fatty acid contents was only approximately 0.5%. In addition, the fatty acid unsaturation/saturation ratio did not increase in strain XCG016, XCG017, or XCG010 at 0 M or 1.0 M NaCl (Fig. 3D).
Overexpression of ELO2 altered complex sphingolipid contents. The effect of overexpression of ELO2 on the contents of phospholipids and complex sphingolipids in strain XCG010 was analyzed ( Fig. 4A to G). It was found that overexpression of ELO2 can change contents of the phospholipid (Fig. 4A to F) and complex sphingolipid ( Complex sphingolipids improve osmotic tolerance. To validate whether the increase in complex sphingolipid contents enhanced osmotic tolerance, the genetic details of strain XCG010 were investigated. The mRNA expression level of the complex sphingolipid biosynthesis genes in strains XCG010 and XCG002 was compared at 0 M and 1.0 M NaCl. At 0 M NaCl, the mRNA levels of AUR1, CSG2, IPT1, LAG1, and LAC1 in strain XCG010 increased by 1.4-Ϯ 0.12-, 1.7-Ϯ 0.13-, 1.3-Ϯ 0.08-, 1.5-Ϯ 0.15-, and 1.8-Ϯ 0.17-fold, respectively, compared to the corresponding values of the control strain XCG002 (Fig. 5A). At 1.0 M NaCl, the mRNA levels of AUR1, CSG2, IPT1, LAG1, and LAC1 in strain XCG010 increased by 1.5-Ϯ 0.12-, 2.8-Ϯ 0.13-, 1.5-Ϯ 0.11-, 2.1-Ϯ 0.10-, and 2.5-Ϯ 0.24-fold, respectively, compared to the corresponding values of the control strain XCG002 (Fig. 5B). These results are consistent with the high content of complex sphingolipids in strain XCG010. However, the mRNA level of the complex sphingolipid biosynthesis genes in strain XCG010 was different from the corresponding values in mutant XCG001 (Table S2). The reason for this may be that the comparison objects (strain XCG010 to XCG002 and mutant XCG001 to the wild-type strain) and conditions (under 1.0 M NaCl and 1.5 M NaCl) were different.
To evaluate whether the inhibition of complex sphingolipid biosynthesis affects the growth of strain XCG010, LAC1, which is involved in the synthesis of ceramide, was deleted to generate strain XCG018. The content of the complex sphingolipid in strain XCG018 was tested. At 0 M NaCl, the contents of IPC (t18:0/26:0), MIPC [t18:0/22: The spot results indicated that at 1.0 M NaCl, the growth of strain XCG018 was better than that of control strain XCG002 but worse than that of strain XCG010 (Fig. 5D). Moreover, at 1.0 M NaCl the survival rate of strain XCG018 (65.6% Ϯ 2.1%) increased by 10.2% Ϯ 0.2% compared with that of the control strain XCG002 (59.5% Ϯ 1.1%) (Fig. 5E). These results suggested that an increase in complex sphingolipids is crucial for S. cerevisiae osmotic tolerance.
Increased content of complex sphingolipid improved membrane integrity. To investigate the effect of complex sphingolipids on membrane integrity, strains XCG002, XCG010, and XCG018 were treated with 0 M or 1.0 M NaCl for 4 h and subjected to SYTOX green and FM4-64 uptake analysis. As illustrated in Fig. 6, at 0 M NaCl, the fluorescence microscope showed that almost all the cells of strains XCG002, XCG010, and XCG018 exhibited an integral membrane (Fig. 6A), whereas at 1.0 M NaCl, the number of cells with an integral membrane for strain XCG018 was more than that for the control strain XCG002 but less than that for strain XCG010 (Fig. 6B). The cells of strains XCG002, XCG010, and XCG018 were further analyzed using flow cytometry. At 0 M NaCl, the percentage of cells with an integral membrane of strains XCG010 and XCG018 was similar to that of the control strain XCG002, whereas at 1.0 M NaCl, the percentage of cells with an integral membrane for strain XCG010 (85.7% Ϯ 3.2%) and XCG018 (76.6% Ϯ 2.9%) increased by 24.4% Ϯ 1.0% and 11.2% Ϯ 0.4%, respectively, compared with that of the control strain XCG002 (68.9% Ϯ 2.9%). These results suggested that an increase in complex sphingolipid content improved membrane integrity.

DISCUSSION
S. cerevisiae is a well-established microorganism that is widely used for the industrial production of fine chemicals, such as organic acids and amino acids, which cause the low pH of the fermentation broth (1). To modulate a suitable pH for a medium, some alkaline reagents need to be added, which leads to osmotic stress (23). To elucidate the physiological mechanism of the cell membrane in osmotic resistance, RNA-seq analysis of the osmosis-tolerant mutant XCG001, obtained through ALE, found ELO2 was associated with osmotic tolerance. Overexpression of ELO2 enhanced the content of complex sphin- golipids. The increased lipid content, as mentioned above, contributed to an improvement of membrane integrity, and, as a result, osmotic resistance increased.
ALE is a very efficient way to improve the phenotype of an industrial strain (24). For example, ALE has been used to increase the specific growth rate for the deletion of some genes in S. cerevisiae or genome-reduced E. coli with glucose as the energy source (25,26) or enhance Schizochytrium species tolerance to high-salinity stress (27). After securing the ALE strains, an important objective was to further identify the targets for genetic modification. Three omics tools were applied for this purpose: genomics, transcriptomics, and metabolomics (25,27,28). For example, the growth of S. cerevisiae on glycerol was increased via ALE, and the transcriptome data revealed that the genes that were related to the tricarboxylic acid cycle and oxidative phosphorylation contributed to the increased growth (28). In this study, the evolutionary effect on the expression of the osmotic stress-related genes was divided into two aspects: (i) upregulated pathways, i.e., glycerol metabolism and ion transport were upregulated in mutant XCG001 under 1.5 M NaCl, as the expression levels of genes encoding glycerol-3-phosphate dehydrogenase (GPD1) and Na ϩ -exporting P-type ATPase (ENA1) were upregulated 1.3-and 0.9-fold in mutant XCG001, respectively, compared with the corresponding values of the wild-type strain at 1.5 M NaCl; and (ii) without influencing the pathway, i.e., trehalose metabolism had no influence in mutant XCG001 under 1.5 M NaCl, as the expression level of the gene encoding trehalose-phosphatase (TPS2) was upregulated 0.1-fold (see Table S3 in the supplemental material). Glycerol as a major

Complex Sphingolipids Regulate Osmotic Tolerance
Applied and Environmental Microbiology osmolyte can improve intracellular osmotic pressure, and, as a result, osmotic tolerance is improved (29). Na ϩ -exporting P-type ATPase is one of the most important ion transports that can pump Na ϩ out of the cell to maintain osmotic pressure (30). However, trehalose synthesis was also unchanged in a yeast-like fungus under osmotic stress (31). Therefore, upregulated glycerol metabolism and ion transport may account for part of the increased osmotic resistance of mutant XCG001. Moreover, RNA-seq analysis of mutant XCG001 and the wild-type strain suggested that ELO2 of 13 commonly upregulated genes was associated with osmotic tolerance. Elo2 is a fatty acid elongase that catalyzes C 16 -carbon fatty acids to C 22 , and its mutations have regulatory effects on 1,3-beta-glucan synthase, vacuolar ATPase, and the secretory pathway (32). ELO2 overexpression may be caused by the changes of the sequence of ELO2 promoter and transcription factors (TFs) due to evolution under osmotic conditions. ELO2 overexpression could be controlled by downregulated negative TFs or upregulated positive TFs. For example, when the expression of YY1 (a negative TF of ELO2) decreased due to the change of YY1 sequence, less YY1 can bind to the promoter of ELO2, which could lead to ELO2 overexpression (33). The overexpression of ELO2 changed the lipid composition, including that of fatty acids, phospholipids, and complex sphingolipids. Sphingolipids play an important role in physiological functions by regulating cell growth and responding to environmental stress (34). The effect of sphingolipid synthesis on environmental stress can be divided into two aspects: (i) the overexpression or knockout of the sphingolipid synthesis gene (17,35,36), for example, when SUR1 was deleted in S. cerevisiae, mutant sur1 was sensitive to Ca 2ϩ (35), and (ii) the addition of a certain sphingolipid, for instance, the addition of phytosphingosine or glucosylceramide or sphingolipid long-chain bases enhanced tolerance to environmental stress (34,37,38). Furthermore, lipid composition and content may undergo changes because of the metabolic pathway genes, harsh environmental conditions, and transcription factors (11,39). The manipulation of lipid biosynthesis genes can change lipid content. ELO3, an ELO2 paralog, is related to biosynthesis of sphingolipid with a C 26 acyl chain (32). Environmental or chemical stresses affect lipid metabolism, which plays a role in maintaining membrane homeostasis and cell growth. A case in point is the membrane unsaturated fatty acids to saturated fatty acids ratio being increased under high-pressure homogenization stress, which enables the strain to avoid damage (39). Transcription factors, such as Mga2, that enable changes in the expression of lipid biosynthesis genes may change the lipid composition indirectly (40).
In this study, the enhancement of the complex sphingolipid content increased the membrane integrity and osmotic tolerance of S. cerevisiae. Membrane integrity could be enhanced by engineering membrane components, including (i) transporter proteins, for example, when the sugar and ion transporter OmpF was deleted and the long-chain fatty acid transporter, FadL, was overexpressed in E. coli, the membrane integrity was enhanced and the fatty acid titer improved (41); (ii) phospholipids, i.e., membrane integrity can also be altered by modifying the distribution of phospholipid head groups, by adjusting phospholipid saturation, and by altering the phospholipid acyl chain length (11); (iii) sterols can modulate the membrane integrity to resist stress (42); and (iv) sphingolipids, i.e., when the sphingolipid biosynthesis genes were deleted in S. cerevisiae, the resultant strains exhibited resistance to amphipathic peptidomimetic, which decreased membrane integrity (43). The increased complex sphingolipids may change the raft structure to help osmotic tolerance in three ways: lipid-lipid interaction, lipid-protein interaction, and membrane fusion. Lipid-lipid interaction requires sterols and sphingolipids acting as functional pairs to help nanodomain formation on the membrane, and as a result, membrane stability increases (11). Lipid-protein interaction could help enhance the function and stability of GPI-anchored and transmembrane domains of proteins (44). Due to an intrinsic property of the very long acyl chain of sphingolipids, membrane fusion may get stimulated (45). Furthermore, Na ϩ may bind to sphingolipids to enhance osmotic tolerance through the calcium channel being activated directly or indirectly. The Ca 2ϩ influx channel is directly activated by Na ϩ binding to sphingolipids, which activates Ca 2ϩ -binding proteins and upregulates the Na ϩ /H ϩ antiporter to pump Na ϩ out of the cell (46). The calcium channel is activated indirectly through Na ϩ binding to sphingolipids to drive the formation of a microdomain on the membrane, which triggers the calcium signals and leads to osmosistolerant responses (47).
In summary, ALE was used to obtain an osmosis-tolerant strain, XCG001, and RNA-seq analysis of mutant XCG001 and the wild-type strain was used to identify a key gene, ELO2, associated with osmotic tolerance. Furthermore, overexpression of ELO2 increased the content of complex sphingolipids with longer acyl chains. As a result, membrane integrity increased, and the osmotic resistance enhanced. This study provides a novel strategy to manipulate membrane complex sphingolipids to increase membrane integrity and osmotic tolerance.

MATERIALS AND METHODS
Strains and media. All S. cerevisiae strains and plasmids used in this study are listed in Table 1. Plasmids pY131 and pY132 were constructed by replacing the promoter P TEF of pY13 with P ADH1 and P TDH3 , respectively. Overexpression strains were constructed using pY13, pY131, and pY132 plasmids carrying the target genes. All plasmids were transformed into yeast cells using the lithium acetate transformation method (48). Homologous recombination was used for lac1 gene deletion. For the LEU2 marker, the upstream and downstream regions of the target gene open reading frame were fused by fusion-PCR, and the PCR products were transformed into yeast cells using the lithium acetate transfor- Replace pY13 promoter P TEF with promoter P ADH1 This study pY132 Replace pY13 promoter P TEF with promoter P TDH3 This study pY131-ELO2 2m, Amp, HIS1, P ADH1 -ELO2 This study pY132-ELO2 2m, Amp, HIS1, P TDH3 -ELO2 This study mation method. All primers used in this study are listed in Table 2. Yeast was cultivated in yeast extract peptone dextrose (YPD) medium and yeast nitrogen base (YNB) medium at 30°C with shaking at 200 rpm. Adaptive laboratory evolution. S. cerevisiae BY4741 was cultivated in 25 ml of YNB medium with histidine, leucine, methionine, uracil, and increasing salt concentrations in a 100-ml flask (0.5 M NaCl, 0.75 M NaCl, 1.0 M NaCl, 1.2 M NaCl, and 1.5 M NaCl). When the optical density at 600 nm (OD 600 ) reached around 4, the strain was transferred to a new salt medium with an initial OD 600 of approximately 0.1. The concentration of salt was increased when the maximum specific growth rate reached around 0.3.
Spot assay. Yeast cells were cultivated in the logarithmic phase and diluted to an OD 600 of 1.0. Aliquots (4 l) of 10-fold serial dilutions were spotted onto YNB agar plates with or without the indicated concentration of NaCl. Growth was assessed after incubation for 2 to 4 days at 30°C. IC 50 , growth curve, and survival rate. Maximum exponential growth rates of yeast were determined in YNB supplemented with increasing salt concentrations. The half-maximal inhibitory

Primer function and name
Sequence (5=-3=) CATCTTCATTCACCGTAT IPT1-F2 TTATTGCCATTGCTGTTA concentration (IC 50 ) was calculated by fitting a Hill-type model to the data. Data points and error bars represent means and standard deviations (SD) from three biological replicates. To test the growth curve of S. cerevisiae at different concentrations of NaCl, cells were cultivated in log phase and diluted into fresh YNB medium at an OD 600 of 0.1 with different concentrations of NaCl. The OD 600 values were recorded by taking determining curves at regular time intervals. Cell survival rates were assessed by log-phase cells treated with various concentrations of NaCl for 1 h at 30°C with shaking at 200 rpm. Cells next were diluted and plated on YNB agar plates with various concentrations of NaCl. After incubation for 2 to 4 days at 30°C, the surviving colonies were counted. The survival rates are expressed relative to that of untreated cells of the corresponding strain. The treatment level of NaCl was chosen according to the standard that the cell density of a strain growing to the stationary phase is similar to that of the wild-type strain at 0 NaCl. At 1.5 M NaCl, mutant XCG001 can grow well, and its cell density at stationary phase was close to that of the wild-type strain at 0 M NaCl (Fig. 1D). At 1.0 M NaCl, mutant XCG010 can grow well, and its cell density at stationary phase was close to that of the wild-type strain at 0 M NaCl (Fig. 2D). At 1.5 M NaCl, the cell density at the stationary phase in mutant XCG010 decreased by 12.7% compared with that of the wild-type strain at 0 M NaCl (see Fig. S3 in the supplemental material). Therefore, mutant XCG001 was treated with 1.5 M NaCl and mutant XCG010 was treated with 1.0 M NaCl. Transcriptome analysis. The wild-type strain and mutant XCG001 were cultured in log phase at 0 M and 1.5 M NaCl. The collected strains were frozen at Ϫ80°C and sent to the Genewiz Institute for RNA extraction and global gene analysis.
qRT-PCR analysis. Total RNA was extracted using a MiniBEST universal RNA extraction kit, and 1 g was taken to synthesize cDNA using the PrimeScript II first-strand cDNA synthesis kit (TaKaRa, Japan). The cDNA mixture was diluted to about 100 ng/l and used as the template for gene expression level analysis by qRT-PCR. qRT-PCR was performed with TB green premix Ex Taq (TaKaRa Bio) using an iQ5 continuous fluorescence detector system (Bio-Rad, Hercules, CA). Data were normalized to that of the ␤-actin gene ACT1. The primer sequences for qRT-PCR are listed in Table 2.
Fatty acid analysis. Fatty acids of yeast were extracted using a NaOH-methanol-distilled water solution (3:10:10, wt/vol/vol) and freeze-dried. The dried sample then was treated with 2 ml boron trifluoride (BF 3 )-methanol (12:88, vol/vol) to produce fatty acid methyl esters, as described previously (49). Finally, samples were analyzed by gas chromatography (GC) with a polyethylene glycol capillary column, eluted at a flow rate of 29.6 ml/min and a column pressure of 63.4 kPa. Data analysis was based on the Supelco 37 standard (47885-U; Sigma). Fatty acid was determined according to GC analysis with fatty acid standards (Supelco 37) (Fig. S4).
Phospholipid measure. Phospholipids were extracted from the freeze-dried samples using chloroform-methanol as described previously (50). Dried phospholipids were obtained under a nitrogen stream and reconstituted in chloroform-methanol (1:1, vol/vol). Samples were analyzed by ultrahighperformance liquid chromatography-tandem mass spectrometry (UPLC-MS; Waters, USA) with a CORTECS UPLC hydrophilic interaction liquid chromatography (HILIC) column (2.1 by 150 mm; inner diameter, 1.6 m) with gradient elution at 45°C and a rate of infusion of 0.3 ml · min Ϫ1 .
Complex sphingolipid measure. Strains were cultured in YNB medium with or without 1.0 M NaCl for 6 to 8 h and washed with phosphate-buffered saline (PBS). The cell pellets were lysed in PBS by bead-beating mechanical disruption at 4°C. The supernatants then were extracted with chloroformmethanol (2:1, vol/vol) at a final ratio of 20% (vol/vol). Centrifugation using a refrigerated centrifuge at 4°C was performed to obtain the supernatant. The extracts were evaporated to dryness under nitrogen at room temperature and stored at Ϫ80°C. The dried samples were sent to the Profleader Institute for complex sphingolipid analysis and solubilized in dichloromethane-methanol (2:1, vol/vol) before analysis by UHPLC quantitative time of flight mass spectrometry (Agilent) analysis (Fig. S5).
Cell membrane integrity analysis. Cell membrane integrity was analyzed by microscopy and flow cytometry. For microscopy analysis of cell membrane integrity, the log-phase cells were treated with 0 M and 1.0 M NaCl for 4 h and washed with PBS twice. The samples then were subjected to SYTOX green and FM4-64 uptake for 20 min, placed on a microscope slide, and covered with a coverslip (51)(52)(53). Images were acquired using a Nikon ECLIPSE 80i microscope equipped with a Nikon DS-Ri1 camera. For flow cytometry of cell membrane integrity, 10,000 counts of stained cells were recorded using a 0.5-ml s Ϫ1 flow rate. All data were exported in FCS3 format and processed using FlowJo software (FlowJo, LLC).
Statistical analysis. Experimental data are shown as the means Ϯ standard errors of the means (SEM). All quantitative data were analyzed using Student's t test or one-way analysis of variance (ANOVA). Each experiment was repeated at least three times.
Accession number(s). The RNA-seq raw reads were submitted to NCBI under BioProject number PRJNA568205, and the Sequence Read Archive (SRA) entries are SRR10150286, SRR10150285, SRR10150284, and SRR10150283.