Transcriptomic analysis of formic acid stress response in Saccharomyces cerevisiae

Formic acid is a representative small molecule acid in lignocellulosic hydrolysate that can inhibit the growth of Saccharomyces cerevisiae cells during alcohol fermentation. However, the mechanism of formic acid cytotoxicity remains largely unknown. In this study, RNA-Seq technology was used to study the response of S. cerevisiae to formic acid stress at the transcriptional level. Scanning electron microscopy and Fourier transform infrared spectroscopy were conducted to observe the surface morphology of yeast cells. A total of 1504 genes were identified as being differentially expressed, with 797 upregulated and 707 downregulated genes. Transcriptomic analysis showed that most genes related to glycolysis, glycogen synthesis, protein degradation, the cell cycle, the MAPK signaling pathway, and redox regulation were significantly induced under formic acid stress and were involved in protein translation and synthesis amino acid synthesis genes were significantly suppressed. Formic acid stress can induce oxidative stress, inhibit protein biosynthesis, cause cells to undergo autophagy, and activate the intracellular metabolic pathways of energy production. The increase of glycogen and the decrease of energy consumption metabolism may be important in the adaptation of S. cerevisiae to formic acid. In addition, formic acid can also induce sexual reproduction and spore formation. This study through transcriptome analysis has preliminarily reveal the molecular response mechanism of S. cerevisiae to formic acid stress and has provided a basis for further research on methods used to improve the tolerance to cell inhibitors in lignocellulose hydrolysate.


Introduction
Lignocellulose is one of the most abundant clean and renewable energy resources in nature. Its hydrolysate can be fermented by Saccharomyces cerevisiae to produce cellulosic ethanol that can effectively alleviate the urgent energy crisis and environmental pollution problems (Auesukaree 2017;Cheng et al. 2019). However, toxic compounds are often introduced during the pretreatment and hydrolysis of lignocellulosic raw materials. These toxic compounds can be divided into three categories: weak acids (Martani et al. 2014;Matsushika and Sawayama 2012), furans and phenols (Fletcher and Baetz 2020;Yang et al. 2020). These toxins can reduce cell growth and negatively affect the performance of fermentation microorganisms. Specifically, formic acid and acetic acid are common weak acid inhibitors occurring in lignocellulosic hydrolysates. These acids are produced by the degradation of 5-hydroxymethylfurfural, and a part of formic acid is produced by the degradation of furfural under acidic conditions (Henriques et al. 2017;Palmqvist and Hahn-Hägerdal 2000). Acetic acid is produced in the process of deacetylation of hemicellulose, and its concentration is usually between 1 and 15 g/L (Dong et al. 2017). The concentration of formic acid is usually lower than that of acetic acid, but it is more toxic to S. cerevisiae than acetic acid (Martin et al. 2007). The concentrations of other toxic and weak acid hydrolysates have rarely been reported, and their concentrations are normally lower than that of formic acid. It has been suggested that the inhibitory effect of 1 3 34 Page 2 of 19 weak acids is due to uncoupling and intracellular anion accumulation (Palmqvist and Hahn-Hägerdal 2000). The undissociated weak acid molecules can pass through the cell membrane and enter the cell, and then dissociate into protons and acid radical ions, thereby causing intracellular acidification. This acidification may cause the intracellular pH to drop and inhibit cell growth and product formation, thus affecting cell metabolism (Lee et al. 2010;Guldfeldt and Arneborg 1998). It is believed that only the anion forms of weak acids can accumulate in the cell, thus causing the toxicity of weak acids and resulting in the degradation of vacuoles (Suzuki et al. 2012).
Yeast cells face various stress factorsduring ethanol fermentation, including high temperature, inhibitors in the cellulose hydrolysate, and low pH. Therefore, the response and tolerance mechanisms of yeast cells to stress are a focus for researchers (Wallacesafinas and Gorwagrauslund 2013;Kot et al. 2019). The development of transcriptomics has provided a foundation for an in-depth understanding of the mechanisms of various inhibitors (Li et al. 2020;Xia et al. 2019;Yang et al. 2021). Stanley et al. (2010) used transcriptomics to study the effect of ethanol accumulation during fermentation in an original yeast and an ethanol-tolerant yeast; the results suggested that a key component of the ethanol stress response is restoration of the NAD + /NADH redox balance, which increases glyceraldehyde-3-phosphate dehydrogenase activity, and higher glycolytic flux in the ethanol-stressed cell. Kieliszek et al. (2019) discussed the effect of selenium in aqueous solution on aspects of lipid and amino acid metabolism in the cell biomass of S. cerevisiae MY A-2200 and Candida utilis ATCC 9950 yeasts. The results indicated that selenium may induce lipid peroxidation that consequently causes a loss of integrity of the cytoplasmic membrane. Mira et al. (2010) found that multiple genes in yeast cells were involved in the toxic effects of acetic acid. These genes were involved in transcription, cell wall integrity, intracellular pH homeostasis, carbohydrate metabolism, and cell absorption of nutrients. Abbott et al. (2009) found that overexpression of CTT1 encoding cytoplasmic catalase increased the specific growth rate of yeast under acetic acid treatment at pH 3.
However, to our knowledge, the current tolerance mechanism of S. cerevisiae in response to formic acid stress has not been elucidated at the transcriptome level. Therefore, in this study scanning electron microscope (SEM) and Fourier transform infrared spectrum (FTIR) analysis combined with RNA-Seq technology were used to study can the formic acid stress response and tolerance mechanism of S. cerevisiae. This study can improve our understanding of the molecular mechanism of the response to formic acid stress in S. cerevisiae and provide a theoretical basis for exploring way to improve the tolerance of S. cerevisiae to lignocellulose hydrolysate inhibitors.

Strain culture
The Saccharomyces cerevisiae GGSF16 strain used in this study was preserved by the Microbiology Laboratory of Guangxi University of Science and Technology was gifted by Guangxi Jinglong Biochemical Co., Ltd. The strain is an industrial alcohol fermentation yeast. The yeast was stored on YPD medium plates (1% yeast extract, 2% peptone, 2% glucose, 2% agar powder). For the experiments, the yeast was activated and cultured for 14 h in a shaker at 30 °C and 150 rpm. The activated GGSF16 yeast cells were inoculated in a fresh YPD medium at 30 °C with an initial yeast cell number of 4.2 × 10 7 CFU/mL. After culturing for 6 h, an appropriate amount of formic acid was added to the culture medium to produce final concentrations of 1.2 g/L, 1.5 g/L, 1.8 g/L, and 2.1 g/L; the solution without formic acid was used as a control group. Then, yeasts were cultured for 24 h and monitored at 600 nm (OD 600 ) with an ultraviolet spectrophotometer. Three biological replicates were performed. We selected the 1.8 g/L formic acid treatment for subsequent experiments according to the growth state of S. cerevisiae GGSF16. The above culture steps were repeated using the group with 1.8 g/L formic acid as the treatment group (for1, for2, for3) and the non-treated yeast (ck1, ck2, ck3) as the control group. The yeast cells were collected for RNA extraction after continuous culture for 2 h.

Scanning electron microscope (SEM) analysis
After treatment with formic acid, yeast cells were centrifuged at 8000 × g, 4 °C for 10 min and washed with three times with phosphate buffer solution (PBS). Then, the cells were re-suspended in 2.5% glutaraldehyde overnight at 4 °C (Chen et al. 2018). Subsequently, the cells were eluted with a gradient concentration of ethanol (20%, 50%, 70%, 80% and 100%) for 10 min in each ethanol concentration. Finally, the yeast cells were dried, sprayed with gold, and were observed and photographed using a FehnPhenom SEM.

Fourier transform infrared spectroscopy (FTIR) analysis
The yeast cells were collected by centrifugation, washed three times with PBS, and freeze-dried. This experiment used the potassium bromide tableting method (Rogowska et al. 2018). The freeze-dried yeast cells and potassium bromide were combined in a mass ratio of 1:20 and ground to a uniform powder with an agate mortar. The scanning conditions were set as follows: the spectral range was 400-4000 cm −1 ; the resolution was 4 cm −1 ; potassium bromide was used as a blank control, and each sample was repeated three times under the same conditions.

Determination of malondialdehyde (MDA) content
MDA content was determined according to the instructions of a Lipid Peroxidation MDA Assay Kit (Solarbio, Beijing, China). The activated the yeast cells were inoculated in a shaker at 30 °C and 150 rpm with an initial OD 600 of 0.1. After culturing for 6 h, Formic acid was added to the culture medium to produce final concentrations of 1.8 g/L. The yeast cells were taken after 6 h of incubation. Cells were disrupted via an ultrasonic cell crusher and centrifuged at 8000 × g, 4 °C for 10 min. The supernatant was removed and incubated with pre-prepared MDA dye in boiling water at 100 °C for 60 min. The MDA content was determined by measuring the optical absorbance at 450 nm, 532 nm and 600 nm with a visible spectrophotometer. The experiment was performed in triplicate.

Determination of hexokinase (HK) enzymes activity
The yeast cells were taken after 1 h of incubation in a shaker at 30 °C and 150 rpm. Cells were disrupted via an ultrasonic cell crusher and centrifuged at 8000 × g, 4 °C for 10 min. The HK enzymes in the supernatants were determined at 4 °C. The enzyme activities were measured using a HK assay kit (Solarbio, Beijing, China). For the definition of enzyme activity units (U/10 4 cell), 1 nmol NADPH produced per minute per 10 4 cells was defined as one enzyme activity unit. The experiment was performed in triplicate.

Transcriptome data analysis
Samples containing yeast cells were centrifuged at 12,000 × g, 4 °C for 2 min. After the turbid supernatant was removed by decantation, yeast cells were recovered from the pellets and stored at − 80 °C. Total RNA was extracted using a yeast RNA extraction kit (Solarbio, Beijing, China). The extracted total RNA sample was subjected to simple quality tests before it was used for subsequent transcriptome analysis, including measurements of RNA integrity, concentration, and protein contamination. After the total RNA was subjected to quality control, the mRNA was enriched with magnetic beads with Oligo (dT), and fragments were added with an appropriate amount of interrupting reagent. The cDNA library was constructed using the mRNA as a template and was sequenced using the BGISEQ sequencing platform. The sequencing was performed by Shenzhen BGI Technology Services Co., Ltd.
Quality control was conducted on raw reads obtained by sequencing to obtain high-quality clead reads. First, the reads containing adapters, low-quality reads, and reads with a high number of unknown N bases were removed to obtain clead reads, which were then used as data for subsequent experimental studies. Sequence alignment analysis was performed between clean reads and the specified reference genome using HISAT, and differential expression analysis was performed between the two sample groups using the DESeq2 algorithm. The differentially expressed genes (DEGs) were screened using the criteria q-value < 0.05 and fold change ≥ 2.

Quantitative real-time PCR analysis
Twelve representative genes were selected to check the validity of RNA-Seq by real-time PCR. The Revertaid First Strand cDNA Synthesis Kit from Thermo Fisher Scientific was used to remove genomic DNA and reverse transcribe RNA to synthesize cDNA. qRT-PCR was performed using the Fluorescent Quantitative PCR Detection system Line Gene 4800 (FQD-48A, HANGZHOU BIO TECHNOLOGY CO., LTD). The PCR conditions were as follows: pre-denaturation at 95 °C for 2 min, denaturation at 95 °C for 15 s, annealing at 55 °C for 15 s, and extension at 72 °C for 1 min, with a total of 40 cycles. All reactions were performed in triplicate. The relative expression levels of target genes were measured by the 2 −△△CT method, and ACT1 was used as the reference gene (Cunha et al. 2015).

Effects of different concentrations of formic acid on the growth of S. cerevisiae GGSF16 cells
The cell growth of GGSF16 under different concentrations of formic acid stress is shown in Fig. 1. The degree of growth inhibition by formic acid on S. cerevisiae increased with the formic acid mass concentration. There was little effect on the growth of GGSF16 when exposed to 1.2 g/L formic acid. The growth of GGSF16 was inhibited under the stress of adding 1.5 g/L formic acid. Compared with the control group, the exponential period is extended by 2 h, but with time, the inhibition gradually weakened, and the final growth rate was only slightly lower than that of the formic acid-free treatment group. When the mass concentration of formic acid was increased from 1.5 to 1.8 g/L, the growth of yeast cells was significantly inhibited compared with the formic acid-free treatment group, and the stationary phase was extended to 12 h. When exposed to 2.1 g/L formic acid stress group, the growth of GGSF16 cells was slowed, and the final growth was also lower than that of the 1.8 g/L formic acid group. Considering the latter experiments, the GGSF16 cells exposed to 1.8 g/L formic acid stress were selected as the experimental group in this study.

Effects of formic acid on the morphology of yeast cells
The morphology of yeast cells in the control and formic acid treatment groups was significantly different (Fig. 2). In the control group (Fig. 2a, b), the cell surface was smooth and full, and the appearance was oval or round. However, under formic acid stress the surfaces of yeast cells were clearly deformed; the surface of the cell membrane became rough and pores, and the edges of cells wall collapsed under formic acid stress (Fig. 2c, d), indicating that formic acid clearly damaged the cell wall and cell membrane of S. cerevisiae. It can be assumed that the addition of formic acid may have influenced the composition and density of cell membranes, leading to the appearance of pores (Kieliszek and Dourou 2021). Thereby hindering the normal physiological metabolism of the cells due to the difficulty for nutrients to enter and exit the cells, thus accelerating cell damage and death.

FTIR analysis
FTIR is a structural analysis technology based on the vibrations of functional groups and polar bonds in compounds. It is characterized by ease of operation, rapidity, and high sensitivity. In recent years, it has been widely used in the structural analysis of macromolecular compounds and the analysis of the secondary structure of proteins. FTIR is a powerful tool for molecular structure information (Kosa et al. 2017;Szymanska-Chargot and Zdunek 2013). Studies have shown that the yeast cell wall is mainly composed of β-glucan, mannoglycoprotein, and small amount of chitin and lipids that play an important role in maintaining cell morphology and inter-cell recognition ). These substances can provide a large number of active groups, and the stretching, bending, and deformation vibrations generated by these groups can produce clear absorption peaks in the infrared spectrum. In the spectrogram of yeast, the most important band to characterize protein is the amide band, and the amide I band at 1650 cm −1 is formed by the stretching vibration of C=O and the bending vibration of N-H. The previous FTIR results indicate that the secondary structure of proteins in yeast is dominated by α helix (Piotrowska and Masek 2015). The amide II band at 1540 cm −1 is caused by the bending vibration of N-H and the stretching vibration of C-N. The amide III band at 1240 cm −1 may be caused by the bending vibration of C-N, the stretching vibration of C-O in -COOH, and the deformation vibration of P=O, representing the asymmetric stretching of the phosphate diester bond and being related to the phospholipid bilayer (Salman et al. 2019). The absorption peaks at about 1080 cm −1 and 915 cm −1 represent the carbohydrate and polysaccharide ring bonds, respectively, in the yeast RNA, DNA, or cell wall, respectively ). The absorption peak located at 3307 cm −1 is caused by the stretching vibration of -OH in chitin and -NH in secondary amines, while the absorption peak located at 3100-2800 cm −1 is due to the antisymmetric vibration of -CH groups in fatty acids, representing lipid functional groups (Shapaval et al. 2019).
Protein is the main component of the cell membrane and cell wall. The red shift of the amide I band (1651.3/1650.3 cm −1 ) and the blue shift of the amide side band (1542.9/1545.4 cm −1 ) were observed after formic acid treatment, indicating that oxygen and nitrogen atoms on protein peptide chains may have had corresponding changes, suggesting that formic acid stress may cause protein denaturation in yeast cells (Fig. 3). In addition, the peak of the amide III band (1243.3/1242.4 cm −1 ) Fig. 1 The growth status of GGSF16 cells under different concentrations 0 g/L, 1.2 g/L, 1.5 g/L, 1.8 g/L and 2.0 g/L of formic acid was added at the midexponential phase after cultivated for 6 h. Black line represents the growth curve of without formic acid the control group; red line represented the growth curve for 1.2 g/L formic acid treatment groups; green line represented the growth curve for 1.5 g/L formic acid treatment groups; blue line represented the growth curve for 1.8 g/L formic acid treatment groups; purple line represented the growth curve for 2.0 g/L formic acid treatment groups produced a red shift that which may have been caused by the phospholipid bilayer being damaged by formic acid treatment; the permeability of the cell membrane was altered, and some nucleic acids and proteins were released. The peak sites were significantly increased at 3307.7 cm −1 , 3002.6 cm −1 and 2928.6 cm −1 , indicating that formic acid altered the structure of chitin in the cell wall and damaged the lipids of the yeast cell membrane. The peak at 914.2 cm −1 was enhanced, indicating that formic acid changed the polysaccharide hydroxyl skeleton on the surface of S. cerevisiae, which in turn changed the structure of the yeast cell wall. Therefore, the results suggest that formic acid damages the yeast cell wall and membrane by altering the structures of proteins, lipids, and chitin, eventually leading to cell death.

Effects of formic acid on MDA content and HK activity
Malondialdehyde is the product of the peroxidation reaction between lipids and free radicals in organisms, and its content can reflect the degree of stress damage to yeast cells and the damage to the cell membrane to some extent. For the formic acid treatment group, there was an increase in malondialdehyde corresponding to a 42.4% increase compared to the control group at 2 h (p < 0.001) and a 78.5% increase at 6 h (p < 0.001). Therefore, it may be that the response to formic acid resulted in oxidative stress and increased lipid peroxidation of the cell membrane, and the cell membrane was destroyed. This was consistent with the SEM and FTIR results (Fig. 4a).
HK is the first key enzyme in the glycolysis process, and its activity is directly related to the rate of intracellular energy metabolism. HK plays an important role in the sugar metabolism of fungi . Interestingly, compared with the control group, we found that the activity of HK in yeast cells increased by 33.4% (p < 0.001) at 1 h under formic acid treatment (Fig. 4b). This shows that formic acid stress enhances the glycolytic pathway to a certain extent and promotes the synthesis of ATP, and this provides sufficient protection for yeast cells against formic acid damage.

Transcriptomic data for GGSF16 cells under formic acid stress
The original data were processed through a series of data processing programs to filter out impurities and obtain clean data. The average clean reads of each sample reached 6.41 Gb, and the average Q20 values of the two groups were 97.27% and 97.26%, with respective Q30 averages of 89.73% and 89.53%. High-quality reads were compared with the designated reference genome using HISAT software, and the results showed that more than 95.91% and 96.31% of the reads of the two groups covered the reference genome, and the quality of the original sequencing was sufficient to be used for subsequent data analysis.
To study the DEGs before and after the addition of formic acid, differential expression identification was conducted for all genes obtained after sequencing of the two groups, using the criteria fold Change > 2 and q-value < 0.05. A total of 1504 DEGs were identified, of which 797 were upregulated and 707 were downregulated (Fig. 5a, b).

GO enrichment analysis
GO enrichment analysis can provide three types of descriptions for gene products, namely biological process (BP), cellular component (CC) and molecular function (MF). The functions of genes can be studied through the GO enrichment analysis of DEGs (Fig. 6). Through GO enrichment, we found that 8 of 25 significantly enriched pathways were related to membrane transport function, including fructose transmembrane transport, mannose transmembrane transport, glucose transmembrane transport, and fructose transmembrane transporter activity. These DEGs had altered expression levels under formic acid stress, suggesting that formic acid affected the functions of the cell membrane and membrane transporters. In addition, most DEGs were Fig. 3 Infrared spectra of formic acid before and after treatment. a represents the formic acid treatment group, b represents the control group Fig. 4 Influence of formic acid on the MDA content and HK enzymes activities in GGSF16. a represents the change in MDA content, b represents the change in hexokinase activity, Compared with control groups, ***represented p < 0.001 were considered as statistically significant enriched in ribosome metabolic pathways, indicating that formic acid has a certain impact on ribosomes. It can be seen from the scanning electron micrographs that the surface morphology of the yeast cells was altered after formic acid treatment, indicating that the cell membrane and cell wall were damaged, and thus the cells lost the protection of the cell wall. The infrared analysis results showed that formic acid destroyed yeast cell walls and cell membranes by altering the structures of proteins, polysaccharides, lipids and chitin, consistent with the changes in cell morphology

KEGG pathway analysis
The KEGG enrichment analysis of DEGs between the control group and the formic acid treatment group is shown in Fig. 7. It is important to note that there were 115 DEGs are enriched in the ribosome pathways, and 27 DEGs were enriched in the ribosomal biogenesis of eukaryotes, indicating that ribosomes play an important role under the formic acid stress. Most genes related to cell body biosynthesis were downregulated, while genes related to mitochondrial ribosome biosynthesis were upregulated. In addition, 26 DEGs were enriched in glycolysis/gluconeogenesis, 20 DEGs were enriched in starch and sucrose metabolism; 13 DEGs were enriched in TCA cycle; 12 DEGs were enriched in galactose metabolism; 10 DEGs were enriched in fructose and mannose metabolism, 38 DEGs were enriched in meiosis of yeast, and 40 DEGs were enriched in the MAPK signaling pathway S. cerevisiae is thought to accelerate glucose metabolism through glucose transmembrane transport, thus synthesizing a large amount of ATP and providing sufficient energy for resisting formic acid stress; this is consistent with the GO enrichment analysis results. In addition, it is possible to initiate sexual reproduction and spore formation to improve offspring adaptation to the environment, thereby enhancing formic acid tolerance. Interestingly, we also found that the genes involved in cysteine and methionine metabolism and purine metabolism were also significantly differentially expressed, indicating that amino acid metabolism and purine metabolism also play a role in the formic acid stress response.

RNA-Seq expression validation by quantitative real-time PCR
To quantitatively determine the reliability of the transcriptome results, we determined the expression levels of Fig. 7 KEGG enrichment analysis of differentially expressed genes of GGSF16 cells in response to formic acid stress. The depth of the color represents the significance of enrichment 12 candidate DEGs by RT-qPCR. These candidate genes included 10 upregulated genes and 2 downregulated genes (Fig. 8). The RT-qPCR results showed that after the addition of formic acid, the expression levels of HXK2 and ENO2 involved in glycolysis, IMA3 involved in galactose metabolism, SOR1 and DSF1 involved in fructose and mannose metabolism, HSP30 involved in protein folding, GND2 is involved in the pentose phosphate pathway, and ATP14, ATP19 and COX6 involved in ATP biosynthesis were upregulated, whereas PRS3 and ADE5,7 are involved in purine metabolism and were downregulated. The RNA-Seq and RT-qPCR data were highly consistent, indicating validity of the RNA-Seq data for genes with distinct transcript levels.

Ribosome
The ribosome is an important organelle in eukaryotic cells, as it is involved in protein biosynthesis. Ribosomes consist of a 40S subunit, a 60S subunit, and four types of RNA (Horsey et al. 2004). The expression levels of genes related to ribosome synthesis were downregulated after the addition of formic acid (Table 1). In this study, we examined the large 60S ribosomal subunit and the small 40S ribosomal subunit. In yeast cells, ribosome biosynthesis is highly conserved, starting from the precursor rRNA in the cytoplasm and ending with the synthesis of mature 40S and 60S ribosomal subunits. In addition, the ribosomes synthesize cells through ribosomes (Horsey et al. 2004), and the required protein can consume about two-thirds of the available energy. The downregulation of these genes after the addition of formic acid indicated that the addition of formic acid inhibited protein synthesis in yeast cells. In addition, the expression levels of MRPS28, MRPL40, and MRPL6 genes were upregulated due to the addition of formic acid, and these genes encode the synthesis of mitochondrial ribosomal proteins. Mitochondrial ribosomes are essential for the translation of mitochondrial genes, and mitochondrial genes encode many basic mitochondrial functional proteins. Ribosomal protein is an important component of ribosomes. Studies have  found that ribosomal protein is not only involved in protein synthesis but also in cell differentiation, regulation of cell development, DNA repair, and other processes. Ribosomal protein plays a crucial role in the growth and development of eukaryotes (Tedesco et al. 2014). The expression levels of genes related to mitochondrial ribosomal proteins were upregulated, indicating that even if formic acid inhibits the biosynthesis of yeast cytoplasmic ribosomes, the cells can still use mitochondrial ribosomes to synthesize other proteins to resist the stress of formic acid. In addition, we also found that the expression levels of DEGs involved in ribosomal biosynthesis in eukaryotes were downregulated. UTP14, UTP13, UTP21, and UTP5 are components of complexes that have the activities of processing precursor rRNA and participating in the biosynthetic process. The synthesis of 18S rRNA in the nucleolus and the assembly process of small ribosomal subunits (Dragon et al. 2002) and the downregulation of its expression indicates that formic acid stress may inhibit the biosynthesis of yeast cell polypeptide chains. The addition of formic acid significantly inhibited the expression of genes involved in protein and RNA synthesis in yeast cells and affected the protein biosynthesis of yeast cells. The cells can rationally allocate the energy they generate and synthesize substances needed to relieve stress through reducing the synthesis of some ribosomes and degrading non-essential materials, thereby increasing the resistance to formic acid.

Protein degradation and autophagy
The endoplasmic reticulum is an important organelle in yeast cells; its functions include folding and transport of synthetic proteins. It is the center of various chaperone proteins and enzymes in eukaryotes that play important roles in response to formic acid stress. Chemical stress can destroy the conformation of a protein, leading to the unfolding and aggregation of the protein (Goldberg 2003). Small HSPs act as chaperones to help fold or refold newly synthesized or denatured proteins and enzymes to maintain their functional conformation (Rosenzweig et al. 2019). Interestingly, our results showed that most chaperones involved in protein folding and degradation such as SSA4, HSP30, and HSP26 were significantly upregulated under formic acid stress ( Table 2). Ma and Liu (2010) found that the deletion mutation of SSA4 exhibits a significantly longer lag period under HMF stress, indicating that SSA4 plays an important role in the adaptation and tolerance of HMF. Piper et al. (1997) found in a study of the ATPase activity of yeast cell membranes that HSP30 is important in maintaining the activity of cell ATPase. It can block the activity of Pma1H-ATPase and thereby ensure sufficient ATP in the cell. FES1, SSA3, SSA2, and SSA1 have been identified as chaperones of the Hsp70 family; these can aid in the correct folding of newly translated proteins, prevent the aggregation of denatured proteins, degrade misfolded proteins and play an anti-apoptotic role in cells (Gowda et al. 2013). Moreover, Hsp70s and Hsp90 act as ATP-hydrolyzing chaperones to provide enough energy for the protein processing in the ER (Li et al. 2013). EMP46 and DER1 are involved in the output of misfolded or unassembled proteins and the transport of the abnormal proteins to the degradation system. Thus, GGSF16 can function in the resistance to formic acid stress by improving the activity of molecular chaperones, degrading misfolded proteins, reducing misfolded proteins, providing raw materials for the synthesis of new proteins, and helping cells to restore normal physiological and metabolic functions. Autophagy is affected by a variety of stress factors, including energy stress (Meijer and Codogno 2011), oxidative stress (Yun et al. 2020), starvation (Hailey et al. 2010), and endoplasmic reticulum stress (Qin et al. 2010). It has been demonstrated that autophagy plays an important role in cell survival under adverse conditions via removal of degradation products and removal of macromolecules and organelles (Mizushima 2007). In this study, autophagyrelated genes were significantly upregulated under formic acid stress (Table 2). ATG9 encodes a transmembrane protein that participates in the formation of autophagy vesicles and cytoplasmic-cytoplasmic (Cvt) pathways. ATG7 and ATG13 encode autophagy-related proteins that participate in the formation of vesicles in the process of autophagy, and regulate the binding of ATG12p, ATG5p, ATG8p and phosphatidylethanolamine. ATG2 is involved in autophagy repair. The peripheral membrane protein encoded by ATG2 is involved in the Cvt pathway and the entry of autophagic vesicles. This process is a necessary condition for the formation of isolated cytoplasmic vesicles. VPS34p and VPS15p jointly encode the autophagy unit of the phosphatidylinositol 3-kinase complex that is necessary for the formation of autophagy vesicles. Studies have shown that autophagy is an important way for cells to ameliorate oxidative stress, and as such it forms the second level of self-protection against environmental oxidative stress (Jain et al. 2010). In the absence of autophagy, environmental stress induces the accumulation of intracellular reactive oxygen species (Tal et al. 2009). Interestingly, the expression levels of autophagy related genes were significantly increased under formic acid stress, suggesting that autophagy plays an important role in the response of S. cerevisiae to formic acid stress. Therefore, the results suggest that formic acid stress induces oxidative stress through the accumulation of reactive oxygen species in yeast cells, thus inducing autophagy in the yeast cells.

Central carbon and energy metabolism
The growth and development of yeast cells, including the synthesis of nucleic acids and proteins, the delivery of biofilms, and transport functions, all require energy consumption. Interestingly, we found that the overall expression levels of genes involved in central carbon and energy metabolism pathways were upregulated (Table 3). HXK1 and HXK2 encode HK, the enzyme that catalyzes glucose-phosphorus to glucose-6-phosphate. The expression levels of the enzymes were significantly upregulated under formic acid stress, indicating that the utilization of glucose by yeast was accelerated and corroborating the previous experimental results (Fig. 4a). TDH1, PGK1, and GPM1 are involved in the synthesis of ATP in the EMP pathway. ENO2 and ENO1 encode enolases that participate in the metabolic pathways of glycolysis and gluconeogenesis (Kornblatt et al. 2013). PYK2 encodes pyruvate kinase, an important rate-limiting enzyme in the glycolysis pathway, and its activity helps to accelerate the smooth progress of EMP (Mitsui et al. 2020). Gene PDC6 that regulates the pyruvate kinase activity was downregulated, leading to slower conversion of pyruvate to acetaldehyde. In addition, the expression of the gene ADH2 was downregulated. Studies have shown that knocking out the ADH2 gene in S. cerevisiae strain As2.4 can increase ethanol production by 52% (Ye et al. 2016). ALD4, which is related to aldehyde dehydrogenase, was downregulated, and the conversion of acetaldehyde was thus reduced. This may be because yeast gradually adapts to environmental pressures by regulating its metabolic activity and avoiding the decline of activity, thus affecting the fermentation environment (Zhang et al. 2012). PCK1 encodes phosphoenolpyruvate carboxykinase, and the downregulation of its transcription level promotes the metabolism of EMP pathway.
Interestingly, genes involved in starch and sucrose metabolism such as GSY1, GSY2, GLG1, GLG2, GPH1, and PGM2, genes involved in fructose and mannose metabolism such as PFK27, SOR1, DSF1, MAN2, and DAK2, genes involved in galactose metabolism such as IMA1, IMA2, IMA3, GAL7, GAL2, and MPH3 were significantly upregulated under formic acid stress. GSY1, GSY2, and GPH1 are involved in glycogen synthesis. When yeast cells are undernourished due to environmental stress, they synthesize glycogen to provide energy for metabolism and adjust the energy demand for biosynthesis. As a protective agent, glycogen plays a crucial role in the winemaking process. It serves as a main carbon source and energy reserve and enhances cell viability under glucose deprivation conditions (Cray et al. 2013;Pérez-Torrado et al. 2002). Therefore, it has been suggested that the yeast strains synthesize compatible solutes to resist the formation of formic acid in a nutrient-poor medium.
SOR1, DSF1 and MAN2 encode sorbitol dehydrogenase and mannitol dehydrogenase, respectively, providing fructose-6-phosphate for the EMP pathway and thereby accelerating ATP generation. IMA1, IMA2, IMA3, and GAL7 are involved in the decomposition of maltose, and GAL2, MPH3, MPH2, and MAL31 are involved in the transport of galactose and maltose. The results indicate that sugar transport and catabolism may play important roles in the response of S. cerevisiae to formic acid stress. Current research has identified three main superfamilies of sugar transporters: the main facilitator superfamily (MFS), the sodium-dependent glucose transporter family (SGLT), and the sweet protein family (SWEET) (Xuan et al. 2013;Deng 2016). In yeast, filamentous fungi, and mammals, sugar transporters that are members of the main promoter superfamily comprise the vast majority. In particular, the hexose transporters (HXTs) of S. cerevisiae have been studied in depth (Özcan and Johnston 1999;Nomura et al. 2015). The hexose transporter (HXT) family of Saccharomyces cerevisiae contains 20 members, including 18 hexose transporters (HXT1-HXT17, GAL2) and 2 glucose receptors (Rgt2, Snf3) (Özcan and Johnston 1999). Surprisingly, in our study, we found that the expression levels of HXT1, HXT2,  HXT3, HXT4, HXT6, HXT7, HXT15, GAL2, and SNF3 genes were significantly upregulated (Table 3). HXTI and HXT3 encode low affinity glucose transporters, whereas HXT2, HXT4, HXT6, and HXT7 encode glucose transporters with high affinity. Among these, HXTI was induced under high glucose concentration, while the expression of HXT6 and HXT7 was inhibited under high glucose concentration. HXT2 and HXT4 were induced under low glucose concentration. In addition, SNF3 does not directly encode sugar transporters but rather acts as an extracellular plasma membrane glucose sensor, and at the same time assists in the induction of expression of HXT1 and HXT3. Studies have shown that the increase of intracellular sugar transporter genes will help provide a broad substrate utilization capacity for response to a variety of environmental conditions (Lazar et al. 2017). Therefore, we speculate that yeast cells can improve tolerance to formic acid stress by regulating the expression of intracellular glucose transporter genes to reduce cell damage.
Mitochondria produce the bulk part of the cellular energy currency ATP that drives numerous energy requiring processes in the cell, and the expression levels of genes involved in this process were upregulated. COX17, COX10, COX8, and COX6 not only relate to mitochondrial respiratory chain complexes III, IV, mitochondrial electron transport, and oxidative phosphorylation but also encode cytochrome oxidases (COX), enzymes that catalyze electron transfer from cytochrome C to molecular oxygen and the proton pump across the mitochondrial inner membrane to produce large amounts of ATP. Interestingly, the expression levels of the energy-related genes ATP19, ATP14, ATP20, ATP4, and PPA2 that play important roles in ATP synthesis were significantly increased. Our study also found that the expression level of PMA2, a gene regulating plasma membrane H + -ATPase activity, was also significantly upregulated. This protein can pump protons out of cells and regulate cytoplasmic pH and the plasma membrane potential. Yeast cells under stress will enhance the expression of genes related to electron transport and ATP synthesis. Expression of PMA2 provides sufficient energy for H + and acid radical ion transport, maintains the stability of intracellular pH, and thus improves the tolerance of yeast cells to formic acid.

Amino acid metabolism
Amino acids are the main metabolites of cells. In the processes of cell metabolism, amino acids can not only participate in cell construction as important precursor substances but also are involved in intracellular biochemical reactions and metabolic regulation by forming catalytic enzymes. Transcriptome analysis revealed that the DEGs related to the amino acid metabolism pathways showed periodic changes ( Table 4). The most significant amino acids were those involved in methionine and methionine synthesis pathways (MET17,MET6,MET3,STR3,CYS4,CYS3,SAM2,SAH1,SPE3,SPE2,MDE1) and valine. The DEGs of leucine and isoleucine biosynthesis (ILV2, ILV3, ILV5, POT1, and Leu9) were significantly downregulated. Van et al. (2011) studied the response of Lactobacillus plantarum WCFS1 to the downregulation of amino acid metabolism-related genes after 8% ethanol treatment. In S. cerevisiae cells, the amino acid metabolism pathway is reduced; this may reduce energy requirements and enhance the viability of yeast cells. Weber et al. (2005) found that the DEGs related to cell growth and division and protein synthesis were significantly inhibited under external environmental stress, and the function was precisely to reduce energy loss and achieve self-protection. The reduction of amino acid metabolism in yeast cells may function to reduce energy requirements and rationally allocate energy to make them survive the stress of formic acid.
Glutathione is an important antioxidant substance that can reduce active oxygen free radicals to generate oxidized glutathione, thereby exerting its antioxidant function and protecting important organelles from damage (Mailloux et al. 2013). GSH provides a means to reduce the negative effects of copper on fermentation. GSH can be used by the yeasts as a cellular defense mechanism against heavy metal stress and oxidative damage (Zimdars et al. 2019). In this study, most of the genes involved in glutathione metabolism were upregulated under formic acid stress, including GSH2, GPX2, GND1, GND2, and SER3. GSH2 encodes glutathione synthetase that catalyzes the synthesis of glutathione from γ-glutamyl cysteine and glycine. GPX2 encodes glutathione peroxidase, an enzyme that protects cells from phospholipid hydroperoxides and non-phospholipid peroxides during oxidative stress (Lovato et al. 2017). GND1 and GND2 encode glucose-6 phosphate dehydrogenases, enzymes that catalyze the production of ribulose-5-phosphate from phosphogluconate to produce a large amount of NADPH in the pentose phosphate pathway. This process provides reducing power for various synthesis reactions of cells and maintains the level of redox in the cell. SER3 encodes 3-phosphoglycerate dehydrogenase, an enzyme that catalyzes the synthesis of serine and glycine and provides amino acids as precursor substances for glutathione biosynthesis. The above results indicate that formic acid stress can disturb the redox balance of yeast, induce oxidative stress, and cause cell damage, consistent with previous experimental results (Fig. 4a).

Meiosis and the cell cycle
As a special mode of cell proliferation, meiosis has a certain impact on cell metabolism. Compared with the control group, the expression levels of genes involved in meiosis in formic acid-treated yeast cells were all upregulated (Table 5). RIM15 is responsible for activation of meiotic genes. RAD53, DDC1, and MRC1 control the S-phase and G1/G2 DNA damage checkpoints. SWM1 and APC2 are essential genes that promote the activity of the complex/loop body (APC/C) in late meiosis. MCM7, MCM4, and MCM2 encode proteins for DNA replication as part of the MCM2-7 hexamer helicase complex that acts as the starting point of DNA replication in G1 and can promote the melting and extension of S-phase DNA. ESP1 is an isolating enzyme that cuts the meiosis-adhesin subunit Rec8p along the chromosome arm during meiosis I and the centromeric site during meiosis II, and its expression is inhibited by Pds1p. The subunit synthesis of the lectin complex (encoded by SMC4) increases, and the lectin recombines with chromosomes during mitosis and meiosis. Therefore, yeast cells can respond to formic acid stress through sexual reproduction, because sexual reproduction can improve the adaptability of offspring to the environment through gene recombination.
Cyclin is a type of protein that is expressed, accumulated, and decomposed in a cell cycle-specific or phased manner, and together with cyclin-dependent kinases affects the operation of the cell cycle. The cell cycle is divided into the early stage of DNA synthesis (G1 stage), the DNA synthesis stage (S stage), the late DNA synthesis stage (G2 stage), and the cell division stage (M stage). Most cell cyclerelated genes were upregulated (Table 5), including CLB1, CLB2, CLB6, CLN1, CLN2, CLN3, SPS4, PCL1, DUN1, and MBP1. PCL1 encodes the G1/S phase-specific cyclic protein involved in cell cycle regulation, interacts with the cyclin-dependent kinase Pho85p, and participates in the regulation of polarized growth and morphogenesis and progression during cell growth. The upregulated expression of PCL1 promotes the transition from the G1 phase to the S phase. The upregulated expression of cyclin CLN1, CLN2, and CLN3 genes indicates activation of Cdc28p kinase to promote the transition from the G1 phase to the S phase. MBP1 is involved in the regulation of the cell cycle from the G1 to S phases. In addition, CLB6 can activate Cdc28p to promote the initiation of DNA synthesis; it also plays a  Kelliher et al. 2018). It is worth noting that the upregulated expression of DUN1, as a cell cycle checkpoint protein gene, and Dun1p, as a signal transducer for cell cycle arrest and transcription response to damaged or unreplicated DNA, often indicates DNA damage. In addition, the gene SPS4 regulating spore formation was upregulated, and yeast showed meiosis under nitrogen stress, forming four haploid ascospores (Becker et al. 2015). Studies have shown that meiosis, spore formation, and the growth of pseudohyphae are the responses of S. cerevisiae to nitrogen starvation (Schröder et al. 2000). Therefore, we believe that formic acid stress may cause nitrogen starvation, and yeast cells can improve the adaptability of offspring to the environment through sexual reproduction and spore formation, thereby improving the tolerance to formic acid.

MAPK signaling pathway
The KEGG enrichment analysis indicated that formic acid could significantly change the MAPK signaling pathway in cells, i.e., under formic acid stress the cells were stimulated to respond by activating intracellular signal transduction pathways. The most important signal transduction pathway in yeast cells is the mitogen-activated protein kinase (MAPK) pathway (Levin 2005). At least four MAPK cascade reactions in S. cerevisiae cells are involved in yeast responses to different physiological stimuli: the cell wall integrity pathway, the pheromone response pathway, the high osmotic pressure glycerol pathway, and the filamentous or invasive growth pathway (Vandermeulen and Cullen 2020). In our study, the expression levels of MF(ALPHA)1, MF(ALPHA)2, STE3, STE2, and SST2 involved in the pheromone signaling pathway were all upregulated (Table 6). MF(ALPHA)1 and MF(ALPHA)2, as mating pheromone α factors, are composed of α cells and interact with mating type a cells, inducing cell cycle arrest and other mating reactions. STE3 and STE2, as receptors of α factor pheromone, cascade with MAP kinase, and are required for transcription in α cells and mating with α cells. The ligand-bound receptor is endocytosed and recycled to the plasma membrane. As the GTPase activator protein of Gpa1p, SST2 can regulate the desensitization of α-factor pheromone, and it also necessary to prevent receptor-independent signal transduction in the mating pathway. Bnilp is the key to the formation of linear actin filaments and participates in the cell processes that need to polarize actin clusters, such as budding and mitotic spindle orientation (Yu et al. 2008). Rholp can also activate Bnilp, and Rom1p and Rom2p as the guanine nucleotide exchange factors of Rho1p and Rho2p can activate Rholp (Krause et al. 2012). TUS1 can regulate Rho1p activity and participate in the interaction of cell wall integrity signaling pathway with Rgl1p. SSK2 encodes the MAP kinase of the mitogen-activated signaling pathway and interacts with Ssk1p to cause autophosphorylation and activation of Ssk2p, thereby leading to phosphorylation of Pbs2p, and also mediates the cytoskeleton actin recovery from osmotic stress. As a stress-induced dual-specific MAP kinase phosphatase, SDP1 negatively regulates Slt2p MAP kinase through direct dephosphorylation. SLN1, as a transmembrane histidine phosphate transfer kinase and osmotic sensor, regulates the cascade of MAP kinases and has a transmembrane protein with an intracellular kinase domain that signals to Ypd1p and Ssk1p. FLO11 is a gene regulating cell filamentous growth. FKS3 is involved in the assembly of spore wall proteins required for flocculation and biofilm formation. The upregulation of these genes indicates that formic acid has adverse effects on the growth of S. cerevisiae. The tolerance of S. cerevisiae to formic acid may be improved by regulating the signal transduction pathways and filamentous or invasive growth pathway of pheromones.

Conclusions
In this study, RNA-Seq technology was used to study the molecular mechanism of S. cerevisiae GGSF16 under formic acid stress. The results showed that the expression of genes involved in protein translation and synthesis and amino acid synthesis was downregulated, while the expression of genes related to central carbon metabolism, redox regulation, protein degradation, and autophagy were significantly upregulated. We speculate that formic acid stress can not only inhibit protein biosynthesis but also induce oxidative stress and cause cells to undergo autophagy. GGSF16 cells can also improve their tolerance to formic acid by altering their growth or modes of reproduction to include sexual reproduction or spore formation. In addition, formic acid stress can promote the synthesis of glycogen, and this has a protective effect on GGSF16 cells. This study provides comprehensive information at the transcriptional level for S. cerevisiae under formic acid stress. In the future, the results can be combined with metabonomics technology to conduct in-depth research that may help to further understand the acid resistance mechanism of GGSF16 strain and provide a theoretical basis for follow-up research on the tolerance to cell inhibitors in lignocellulose hydrolysate.