Complementary analyses of transcriptome and proteome revealed the formation mechanism of ethyl acetate, ethanol and organic acids in Kluyveromyces marxianus L1-1 in Chinese fermented acid rice soup

Background: Recently, more chemical and biotechnological applications have been found in Kluyveromyces marxianus than Saccharomyces cerevisiae in the food eld because they show advantageous metabolism features in the production of avor components of interest. However, most of study demonstrated Kluyveromyces marxianus involved in ethanol synthesis in the dairy products in food elds. Our study aims to clarify the formation mechanism of ethyl acetate and organic acids in acid rice soup inoculated with Kluyveromyces marxianus. Results: The higher concentration of ethyl acetate than ethanol and organic acids in fermented acid rice soup inoculated with Kluyveromyces marxianus. Up-regulated genes/proteins, including ADH1, ADH2, ADH6, ATF1, ACCT, and TES1, and down-regulated ALD family involved in glycolysis/gluconeogenesis and pyruvate metabolism played the crucial roles in the formation of ethyl acetate and other esters. In addition, up-regulated genes/proteins involved in starch and sucrose metabolism, amino sugar and nucleotide sugar metabolism, glycolysis/gluconeogenesis, TCA cycle, and pyruvate metabolism played the important roles in the formation of organic acids, ethanol and esters. Conclusion: Our results reveals the formation mechanism of ethyl acetate and organic acids in acid rice soup inoculated with K. marxianus L1-1. This study provides the basis for improving aroma and taste of fermented foods and reveals the formation mechanisms of avors in no-dairy products. malate dehydrogenase. key KEGG pathways related to the formation mechanism of the avor based on the DEGs are recombination, mismatch autophagy-yeast, MAPK signaling pathway-yeast, phenylalanine metabolism, base excision repair, nucleotide excision repair, other types of O-glycan biosynthesis, tryptophan metabolism, cell cycle-yeast, penicillin and cephalosporin biosynthesis and


Background
Recently, more chemical and biotechnological applications have been found in non-saccharomyces yeasts than Saccharomyces cerevisiae in the food eld because they show advantageous metabolism features in the production of avor components of interest. Kluyveromyces marxianus (K. marxianus), as a non-saccharomyces yeast, has various advantages over mesophilic yeasts, such as fast growth rate, wide spectrum of substrates and reduced cooling cost [1]. It is a haploid, homothallic, thermotolerant, and hemiascomycetous yeast and is closely related to Kluyveromyces lactis. Unlike S. cerevisiae, K. marxianus has the assimilating capability of lactose, glucose and xylose. Notably, compared to S. cerevisiae and K. lactis, K. marxianus has the more signi cant ability concentrate on the intrinsic fermentation capability of various sugars at high temperatures [2]. K. marxianus grows in a wide temperature range from 4 to 52 °C, indicating it is thermotolerant and can be applied in the processes of low temperatures and high temperatures, which can also prevent the growth of microorganisms sensitive to heat [3]. Furthermore, the probiotic properties of K. marxianus have been extensively explored [4].
Interestingly, K. marxianus shows a great potential in the production of esters, which are key aromatic compounds in the food industry [5]. Ethyl acetate and other short-chain volatile esters are used as industrial solvents and perfume ingredients. It was reported that the global market demand of ethyl acetate was more than 1.7 million tons per year [6]. Therefore, it is signi cant to produce ethyl acetate in the food production industry. Three synthesis ways of ethyl acetate have been reported [7]: hemiacetal oxidation (spontaneous formation of hemiacetal from acetaldehyde and ethanol under the action of enzymatic oxidization), condensation of ethanol and acetyl-CoA, and esteri cation of acetate and ethanol (the reverse synthesis of ethyl acetate from ethanol and acetate). However, it is di cult to produce ethyl acetate through esteri cation of ethanol and acetate because the ester-hydrolyzing activity of esterase is much higher than its ester-synthesizing activity [5]. The previous report demonstrated that ethyl acetate synthesis was characterized by the direct utilization of ethanol as a substrate or the hemiacetal reaction between sugar and acetaldehyde [5]. Ethyl acetate as an aroma component plays an increasingly important role in foods and other elds. However, the synthesis mechanism of ethyl acetate in K. marxianus in rice-acid is unknown. Most studies on K. marxianus focused on its role in alcoholic fermentation in the fuel industry and food eld [8], however, the production of ethyl acetate or organic acids with K. marxianus was seldom reported. The metabolism of K. marxianus is less well understood than that of S. cerevisiae. The formation mechanisms of ethyl acetate, other esters or avor compounds in fermented foods inoculated with K. marxianus are not yet completely understood.
More multi-omics analysis technologies have been used to explore the synthesis, metabolism and accumulation of nutrients and avor components in foods. RNA Sequencing is a novel high-throughput sequencing technology with many advantages, such as much information, low data redundancy and accurate analysis. In addition, it does not require the background of genomics, but it can analyze the transcriptional expression of multiple materials [9]. Proteomics can enhance the understanding of the biochemical processes of avor development in fermented foods [10]. However, due to non-coding RNA regulation, protein degradation, protein secretion, the quantity of differently expressed proteins (DEPs) are often less than differently expressed genes (DEGs). In addition, the detectable protein content, protease hydrolysis and operational errors in proteomics detection also limit the application of proteomic analysis [11]. Many factors lead to the difference in the analysis results of transcritptomics and proteomics. Notably, a previous report proved that sequence features contributed to 15.2-26.2% of total variations of mRNA and proteins [12]. Therefore, the combination of transcriptomics and proteomics may reveal the avor formation mechanism in K. marxianus L1-1 in acid rice soup (rice-acid).
In order to clarify the formation mechanism of ethyl acetate and organic acids in rice-acid inoculated with Kluyveromyces marxianus L1-1, volatile compounds and organic acids in rice-acid were measured in this study. In addition, we analyzed the differently expressed genes and proteins of K. marxianus L1-1 in riceacid in the key fermentation days (the rst and third days) through the complementary analysis of mRNA sequencing and proteomics. Through GO enrichment analysis and KEGG pathway enrichment analysis, the formation mechanisms of ethyl acetate and organic acids in K. marxianus L1-1 in rice-acid were explored in this study.

Results
Variations of the quantities of K. marxianus L1-1 in the fermentation process of rice-acid K. marxianus, which is able to utilize various sugars, may be a suitable microbe for lignocellulose hydrolysis and grain matrix at 30 °C [13]. In this study, the fermentation temperature was determined as 30 °C based on our previous study. The number of K. marxianus L1-1 changed signi cantly in the fermentation process. The number of K. marxianus L1-1 gained the most signi cant increase rate from 0 d to 1 d (Fig. 1a) and decreased from 1d to 2 d. Interestingly, it increased from 2 d to 3 d. The above variations may be related to the oxygen content in the fermentation tank. The limited supply of oxygen (the terminal electron acceptor) also initiated the synthesis of some esters, but it primarily forced ethanol production during the growth of K. marxianus DSM 5422 [14]. We explored the formation of ethyl acetate, other esters and organic acids based on the growth of K. marxianus L1-1 in this study. In the study, we investigated the key fermentation process of rice-acid with K. marxianus L1-1 in 1 d and 3 d and analyzed the formation mechanism of avors in K. marxianus L1-1.
Variations of key volatile compounds in the fermentation process of rice-acid The variations of avor compounds in 1 d and 3 d were explored. The basic conditions for the formation of ethyl acetate are acetic acid, ethanol and some key enzymes which were discussed in this study. In the obtained volatile compounds, 5 key acids, 13 key alcohols and 12 key esters were found (Table 1). From 1 d to 3 d, ethyl acetate content increased from 162.98 ± 5.02 to 241.37 ± 6.20 g/kg; ethanol content increased from 36.11 ± 4.54 to 52.68 ± 14.45 g/kg; acetic acid content increased from 0.21 ± 0.06 to 32.67 ± 1.57 g/kg. Acetic acid, 2-phenylethyl ester, 2-methyl-propanoic acid, ethyl ester and 9 other esters were also found. Interestingly, ethyl acetate, ethanol and acetic acid are important volatile compounds because of their high contents and low odor thresholds [15]. Ethyl acetate made a signi cant contribution to the formation of the fruity avor and promoted the overall avor balance in rice-acid. Moreover, ethyl acetate exhibits probiotic properties such as being closely linked to the antioxidant function in the fruit [16]. The formation of esters in the alcoholization stage was closely related to the enzyme activity of yeasts. Therefore, it is necessary to explore the formation mechanism of ethyl acetate and other esters. Table 1 The key volatile compounds (mg/L) and organic acids (mg/L) in fermented rice-acid inoculated with K. marxianus L1-1 at 1 d and 3 d.  Variations of organic acids in the fermentation process of rice-acid Seven organic acids were found in rice-acid, including L-lactic acid, acetic acid, malic acid, succinic acid, citric acid, oxalic acid and tartaric acid ( Table 1). Among 7 organic acids, L-lactic acid had the highest content. The content of L-lactic acid increased from 3.01 ± 0.61 g/kg in Day 1 to 6.02 ± 1.67 g/kg in 3 d.
The contents of the other 6 organic acids did not show the signi cant increase during the fermentation process. However, although the contents of the 6 other organic acids were low, they interacted with each other to promote the formation of the sourness and taste of rice-acid. In our study, both volatile components and organic acids affected the formation of the avor of rice-acid. Lactic acid exists in two isomeric forms which include L-(+) and D-(−)-Lactic acid. It is produced by microbial fermentation and chemical synthesis and used in food, cosmetic, pharmaceutical, and chemical industries [17]. In the study, we mainly focused on the increase in L-(+) -Lactic acid caused by microbial fermentation. L-(+)-Lactic acid with a high enantiomeric purity is required in many industries, especially in medical, pharmaceutical and food industries, since D-(−)-Lactic acid is harmful to humans and can cause decalci cation or acidosis [18]. L-(+)-Lactic acid not only promoted the formation of the avor in rice-acid, but also had an important effect on the health. We will further explore the genes, proteins and enzymes associated with the formation of organic acids.  Table 2). All of high-quality clean reads were used for gene comparison and more than 88% of total reads of clean reads were mapped to the database by software Bowtie 2. By using |log2FC| > 1.5 and FDR < 0.05, we identi ed 1390 DEGs (788 up-regulated and 602 down-regulated) between Y 1 d and Y 3 d (Fig. 2a). GO analysis of the DEGs showed the enrichment of three major cellular components, biological processes and molecular functions (Fig. 3). In terms of cellular components, most of the DEGs were enriched in nucleolus, small-subunit processome and preribosome, and large-subunit precursor. In terms of biological processes, most of the DEGs were enriched in endonucleolytic cleavage in ITS1 to separate SSU-rRNA from 5.8S rRNA and LSU-rRNA from tricistronic rRNA transcript (SSU-rRNA, 5.8S rRNA, and LSU-rRNA). In terms of molecular functions, these DEGs were enriched in structural constituent of ribosome, snoRNA binding and rRNA binding. All DEGs were subjected to KEGG pathway enrichment analysis. KEGG analysis assigned the DEGs of Y1-d and Y3-d to metabolic pathways. At least 4231 genes were identi ed, including 1390 DEGs annotated to 279 KEGG pathways (Table S1). The signi cantly enriched pathways (p-value < 0.05 and q-value < 0.05) were ribosome, cytosolic DNA-sensing pathway, RNA polymerase, DNA replication, ribosome biogenesis in eukaryotes, pyrimidine metabolism and purine metabolism. Importantly, the crucial KEGG pathway related to the ethyl acetate and organic acids included amino sugar and nucleotide sugar metabolism, starch and sucrose metabolism, glycolysis/gluconeogenesis, pyruvate metabolism and TCA cycle. These pathways showed the different roles of K. marxianus L1-1 in the formation of avor and taste of rice-acid.
Ethyl acetate and organic acids promoted the maturity of avor and taste of rice-acid inoculated with K. marxianus L1-1. Key genes were found to be involved in the ethyl acetate metabolism process in glycolysis, including GLK1 (K00844), GPD2 (K00134), 3 genes of ALD family (K00129 and 2 K00128), 6 genes of ADH family (K13953, K13953 and 4 genes of K13953) and ATF1 protein (BAO42650, BAO42650 and BAO42650) ( Table 3). The genes of GLK1, GPD2, ALD family, ADH family and ATF1 encode glucokinase-1, glyceraldehyde 3-phosphate dehydrogenase, aldehyde dehydrogenase, alcohol dehydrogenase and alcohol O-acetyltransferase, respectively. In addition, the gene ERG10 encoding acetyl-CoA C-acetyltransferase was found to be involved in the ethyl acetate metabolism process in pyruvate metabolism. The key genes related to organic acids found in pyruvate metabolism included CYB2 (K00101), DLD1 (K00102), 2 genes of MDH family (2 K00026), and FUM1 (K01679) respectively encoding L-lactate dehydrogenase, D-lactate dehydrogenase, malate dehydrogenase and fumarate hydratase. The key genes related to organic acids found in the citrate cycle included 2 genes of CIT family (2 K01647), 3 genes of SDH family (K00234, K00236 and K00237) and 2 genes of MDH2 (K00026) respectively encoding citrate synthase succinate dehydrogenase (ubiquinone) avoprotein subunit and malate dehydrogenase. The key KEGG pathways related to the formation mechanism of the avor based on the DEGs are discussed below. Note: FC value represented the differential expression multiple of mRNA and protein.

Proteomics Characterization
The total proteins were extracted from the Y1-d and the Y3-d at lling stage and subjected to 4D label-free proteomics analysis to complement the transcriptome analysis. According to the abundance levels of proteins, 610 proteins were identi ed as DEPs at p-value < 0.05, including 135 proteins with increased abundance levels and 475 proteins with decreased abundance levels (Fig. 2b), and the difference ratio reached > 1.5. The number of up-regulated proteins was smaller than that of down-regulated proteins at the lling stage since the growth of K. marxianus L1-1 was inhibited due to the acid environment in the later fermentation stage of rice-acid.
To obtain a global diagram of proteomic changes, at least 2937 proteins were identi ed and 187 DEPs were annotated with GO analysis and KEGG analysis (Table S2). In GO functional analysis, 187 proteins were annotated to 59 GO terms. The results of GO analysis showed that the distributions of DEPs in functional classi cation were consistent with the distributions of transcription levels of DEGs (Fig. 4). In terms of cellular components, most of the up-regulated DEPs were enriched in mitochondrion, tricarboxylic acid cycle enzyme complex and mitochondrial matrix. In terms of the biological process, most of the up-regulated DEPs were enriched in citrate metabolic process, tricarboxylic acid metabolic process and galactose catabolic process. In terms of molecular functions, these up-regulated DEPs were enriched in oxidoreductase activity, L-malate dehydrogenase activity, malate dehydrogenase activity and alcohol dehydrogenase activity. The 6 GO terms involved in alcohol dehydrogenase activity had the smallest p-value (p-value < 0.01) and were related to the formation of ethyl acetate. The 187 DEPs were annotated to 30 KEGG pathways (Table S2). Most up-regulated KEGG pathways analyzed by proteomics were related the formation of ethyl acetate and organic acids, which explained reasonably the formation of avor and taste in rice-acid inoculated with K. marxianus L1-1. Meanwhile, the down-regulated KEGG pathways obtained by proteomics analysis were most related to the growth of K. marxianus L1-1 (Fig. 5b), verifying that it was reasonable to select the third day as the ending of rice-acid fermentation.
According to the pathway analysis (Fig. 5a), we could conclude that many proteins took part in various metabolic pathways including amino sugar and nucleotide sugar metabolism, starch and sucrose metabolism, glycolysis/gluconeogenesis, pyruvate metabolism and citrate cycle, which might affect many aspects of the metabolism of K. marxianus L1-1 during the key fermentation period of rice-acid.
However, the different KEGG pathways obtained by proteomics analysis played different roles in the formation of avor and taste in rice-acid.
Ethyl acetate is the most important volatile compound in rice-acid inoculated with K. marxianus L1-1.
Four key proteins were found to be involved in the ethyl acetate metabolism process in glycolysis, including GLK1(BAO37673), GAP1 (BAO40242), 2 ADH1 (BAO40648 and BAO40126) and ADH6 (BAO42650) ( Table 2). GLK1, GAP1, ADH1 and ADH6 encode glucokinase-1, glyceraldehyde-3-phosphate dehydrogenase 1, alcohol dehydrogenase and NADP-dependent alcohol dehydrogenase, respectively. In addition, four key proteins were found to be involved in organic acids metabolic process in pyruvate metabolism, including 2 MDH family (BAO41458 and BAO40079), FUM1 (BAO42339) and LYS21 (BAO38393). MDH, FUM1 and LYS21 encode malate dehydrogenase, fumarate hydratase and homocitrate synthase, respectively. Four key proteins were found to be involved in the metabolism process of organic acids in citrate cycle, including CIT1 (BAO38563), SDH1 (BAO38924), MDH2 (BAO40415) and FUM1 (BAO42339), which respectively encode citrate synthase, succinate dehydrogenase (ubiquinone) avoprotein subunit, malate dehydrogenase and fumarate hydratase. The proteins related to the metabolism of organic acids were consistent with the genes, indicating that the combination of transcriptomics and proteomics were useful tools to analyze the formation of ethyl acetate and organic acids during the key fermentation period of rice-acid inoculated with K. marxianus L1-1.

Correlation Analysis Of Transcriptome And Proteome Data
Transcriptomic and proteomic analysis results are shown in Fig. 1b-1d. At least 4231 genes were identi ed and 1390 DEGs were annotated to 279 KEGG pathways by using transcriptomic analysis. In addition, 2937 proteins were identi ed and 610 DEPs were annotated to 30 KEGG pathways by using proteomics analysis. Pearson correlation coe cient was 0.3761 and the results of the two analysis methods were signi cantly different. Therefore, the combination of transcriptome and proteome data could be an effective way to reveal the formation mechanism of the avor in rice-acid inoculated with K. marxianus L1-1. The 5 KEGG pathways related to the synthesis of ethyl acetate and organic acids are shown in Table 3.

Discussion
Rice-acid, as a cereal-based fermented food used for seasoning, is famous in China. However, the traditional rice-acid process requires two times of fermentation and the long-term fermentation time may lead to the unstable and non-persistent avor. In this study, we adopted a novel inoculation strain (K. marxianus L1-1), the inoculation of K. marxianus L1-1 promoted the fermented rice-acid has the unique avor and shorten the fermentation period of rice-acid from 40 d to 4 d. Our previous study proved that this fermentation method could realize the high-quality avor. However, the formation mechanism of the avor in rice-acid inoculated with K. marxianus L1-1 is not clear. In this study, the RNA-seq and 4D labelfree technologies were used to explore the genes and proteins in the formation mechanism of ethyl acetate and organic acids in rice-acid inoculated with K. marxianus L1-1. DEGs and DEPs were identi ed and annotated to key KEGG metabolism pathways, including starch and sucrose metabolism, amino sugar and nucleotide sugar metabolism, glycolysis/gluconeogenesis, pyruvate metabolism and TCA cycle. Furthermore, the results of transcriptome and proteome were combined to reveal the formation mechanism of ethyl acetate and organic acids. We provided a comprehensive interpretation and exact measurements of genes and protein expressions involved in the changes of the avor of rice-acid for the rst time.
Up-regulated proteins and genes involved in starch and sucrose metabolism, amino sugar and nucleotide sugar metabolism provide the energy for the formation of acids, ethanol and esters Starch and sucrose metabolism provides an important transient pool in the sugar accumulation pathways. The genes and proteins could provide the energy for the formation of the avor and taste in rice-acid, including GLK1 encoding glucokinase-1 and KLMA_10051 encoding hexokinase in the KEGG pathways of starch and sucrose metabolism and amino sugar and nucleotide sugar metabolism (Table 3). Protein GLK1 was a glycolysis-initiating enzyme [19] and showed the 13.384-Log2FC upregulation. The gene of KLMA_10051 encodes a hexokinase, which showed the 1.83-Log2FC upregulation. Up-regulated GLK1 indicated an increase in NADPH amount and more energy generated in K. marxianus L1-1. Lane et al. [20] also reported that the catabolite repression could be reduced by modulating the expression of glucose-phosphorylating enzymes, such as GLK1 and hexokinase (HK). In addition, SCW4 (KLMA_30608) encoding glucan 1,3-beta-glucosidase showed the upregulation of 6.986-Log2FC, as indicated by proteomic analysis, and the upregulation of 1.53-Log2FC, as analyzed by transcriptomic analysis. SCW4 (KLMA_30608) encoding glucan 1,3-beta-glucosidase might have the hydrolytic activity and provide the energy to promote the formation of the avor and taste in rice-acid. In a previous report, it was also indicated that a 1,3-β-glucosidase BGL1 puri ed from the pilei showed the hydrolytic activity toward laminarin, laminarioligosaccharides including laminaribiose, and p-nitrophenylβ-d-glucopyranoside (pNPG) [21]. In addition, another study also showed that 1,3-β-glucosidase had a certain hydrolytic activity towards gentiobiose, cellobiose, and related polysaccharides [22]. Therefore, the hydrolysis of sugar compounds may provide substrates and energy for orderly metabolism. Eventually, with hydrolytic enzymes, carbohydrates were easily converted into glucose in K. marxianus L1-1. Our fermentation belonged to a static fermentation method. K. marxianus seemed to enhance glucose metabolism and shift to fermentation, implying the connection between oxygen and glucose-sensing pathways [13]. In future study we will explore the content on oxygen in the fermenter and its correlation with avor in rice-acid. However, genes and proteins were differentially expressed in the different fermentation days (the rst and third days) of rice-acid with the inoculation of K. marxianus L1-1. The correlation between transcriptomics and proteomics data was not perfect and there are some differences in the results of the two methods [23]. Protein expressions may be affected by various factors in the translational stage [12]. The e ciency of protein biosynthesis and accumulation depends on various factors in the biological regulation process. Therefore, we adopted the complementary analysis method of transcriptomics and proteomics to reveal formation mechanism of ethyl acetate, ethanol and organic acids in Chinese rice-acid inoculated with K. marxianus.
Up-regulated proteins and genes involved in glycolysis/gluconeogenesis and pyruvate metabolism played an important role in the formation of ethyl acetate and other esters Volatile esters are secondary metabolites produced by yeasts and fungi during fermentation of fermented foods [24]. Interestingly, our study showed compared to alcohol, more esters existed in rice-acid inoculated with K. marxianus L1-1. This difference contributed to the avor formation in rice-acid. Ethyl acetate was one of most important volatile esters in rice-acid. Both DEGs and DEPs in glycolysis/gluconeogenesis and pyruvate metabolism played an important role in the formation of ethyl acetate and other esters (Fig. 7). Glycolysis is the cytosolic pathway that converts glucose to pyruvate. ADH1, ADH2, ADH3 (alcohol dehydrogenase) and ADH6 (NADP-dependent alcohol dehydrogenase 6) involved in glycolysis/gluconeogenesis respectively encode the proteins of BAO40648, BAO40126 and BAO42650 (Table 3 and Fig. 6a). This type of esteri cation is carried out from primary and secondary alcohols, aldehydes or ketones [25]. ADH enzymes catalyze the synthesis of ethyl acetate through the oxidation of hemiacetal. ADH3 was only found in the transcriptomic analysis in this study. A previous study proved that ADH2 was constitutively expressed in aerobic growth with glucose as a carbon source, whereas ADH3 expression increased as cells reached the stationary phase. These results were in agreement with the previous analyses of K. marxianus transcriptome [26]. Our study proved that the ADH family was important in the complementary analysis of transcriptomics and proteomics. ADH family had a signi cant effect on the formation of ethyl acetate and ethanol, and the up-regulated genes and proteins suggested that ADH1, ADH2 and ADH6 were the dominant enzymes in ethanol production when glucose was used as a carbon source. In the identi ed up-regulated genes and proteins, ADH2 and ADH6 were critical in the reduction of acetaldehyde to ethanol (a precursor to ethyl acetate). This result was consistent with the previous study [26]. Another report proved that the alcohol acetyltransferase was associated with intracellular lipid particles in cytosol [27]. We analyzed the up-regulated genes and proteins of ADH family and found that they promoted the formation of ethanol and ethyl acetate. Herein, one possible formation pathway was the oxidation of hemiacetal (the spontaneous product of ethanol and acetaldehyde) under the catalysis action of ADH activity. Although the two methods of transcriptomics and proteomics showed some differences, all the up-regulated ADH genes and proteins indicated that the alternative biosynthetic routes of ethyl acetate existed in K. marxianus L1-1.
In addition, the formation of ethyl acetate in rice-acid fermentation inoculated with K. marxianus L1-1 was mainly catalyzed by two enzymes named ATF1 (alcohol O-acetyltransferase) and TES1 (acylcoenzyme A) in this study (Table 3 and Fig. 6b), which possessed an acyl-coenzyme A: ethanol Oacyltransferase (AEATase) activity as well as the esterase activity. We found that ATF1 could use ACCT (acetyl-CoA C-acetyltransferase) to synthesize acetate esters, including ethyl acetate, acetic acid, 2phenylethyl ester and isobutyl acetate (Fig. 7). The previous analysis of the ATF1p also found that acetyl-CoA was used to synthesize acetate esters [28,29]. It was demonstrated that acetyl-CoA taking a crucial role to produce the generation of more NADH and more ATP in glucose [13]. The previous study postulated that ester biosynthesis in K. marxianus may also occur through homologs to the mediumchain acyltransferases from S. cerevisiae, the isoamyl acetate-hydrolyzing esterase, the Nacetyltransferase Sli1 and/or the alcohol-O-acetyltransferase [30]. Interestingly, we also found that the another important ester family including propanoic acid, 2-methyl-propanoic acid, ethyl ester, 2-propenoic acid, ethenyl ester and propanoic acid, 2-methyl-2-phenylethyl ester, and the increase in propanoic acid, ethyl ester was closely related to propanoate metabolism. The propanoate is expected to be converted to acetyl-CoA or pyruvate, as suggested by examination of likely propanoate metabolism. It was also demonstrated that propanoate was converted to acetyl-CoA in three classes of mycolate [31]. However, the differentially expressed gene ATF1 was only found under the transcriptomics analysis. The combination method of transcriptomics and proteomics could provide the more reasonable explanation for the formation of ethyl acetate and propanoic acid, ethyl ester.
Furthermore, some genes and proteins related to the formation of ethyl acetate and other esters included GLK1 (glucokinase-1), KLMA_10051 (hexokinase), GPD2 (glyceraldehyde 3-phosphate dehydrogenase) and ALD family (aldehyde dehydrogenase) ( Table 3 and Fig. 7). Protein GLK1 was a glycolysis-initiating enzyme, which promoted the formation of ethyl acetate and other esters and also played an important role in glycolysis/gluconeogenesis and pyruvate metabolism. Glyceraldehydes-3-P and pyruvate were the intermediate products in the glycolysis process and provided the carbon skeleton for volatile compound biosynthesis in rice-acid. A recent transcriptomic study suggested that a β-glucosidase homolog in K. marxianus may be responsible for cellobiose degradation [13]. Interestingly, the genes of ALD family (aldehyde dehydrogenase showed the downregulation, indicating that more energy was used in ADH family encoding alcohol dehydrogenase. In the further study, we will explore the change of aldehyde dehydrogenase. Some different genes were involved in the formation of ethyl acetate in K. marxianus and S. cerevisiae, although the metabolism of ethyl acetate in K. marxianus was seldom reported. The previous study suggested that the biosynthesis of acetate ester could be interpreted with the antagonistic activity of esterase IAH1 [32], but we did not nd a reverse esterase playing a role in the formation of ethyl acetate and other esters in our study (Fig. 6c). A previous study also reported that there was no esterase involved in the biosynthesis ethyl acetate or other esters in K. marxianus CBS 6556 [26].
Up-regulated genes and proteins involved in TCA cycle and pyruvate metabolism played important roles in the formation of organic acids The mitochondrial TCA cycle, also known as Krebs cycle, is one of the major pathways of carbon metabolism in higher organisms that provides electrons during oxidative phosphorylation within the inner mitochondrial membrane. TCA cycle is crucial in mitochondrial membranes for respiration. Both TCA cycle and pyruvate metabolism played important roles in the formation of organic acids (Table 3 and Fig. 7). It was also reported that organic acids were closely related to TCA cycle in rice [33]. The upregulated gene PYC2 (KLMA_10253) encoding pyruvate carboxylase played the crucial role in TCA cycle and pyruvate metabolism. It was demonstrated that pyruvate carboxylase as an anaplerotic enzyme had a special effect and played an essential role in various cellular metabolic pathways including gluconeogenesis, glucose-induced insulin secretion, de novo fatty acid synthesis and amino acid synthesis [34]. The DEGs and DEPs related to TCA cycle and pyruvate metabolism reasonably interpreted the formation of organic acids during the key fermentation period of rice-acid inoculated with K. marxianus L1-1.
The up-regulated gene LYS21 (KLMA_10771) encoding homocitrate synthase, has been identi ed by the combined analysis of transcriptomics and proteomics in this study. The previous study proved that homocitrate synthase was responsible for the rst important step of the pathway and played the crucial role in pyruvate metabolism [35]. Notably, homocitrate synthase LYS21 was linked to the key process of DNA damage repair in a nucleus and TCA cycle in the cytoplasm [13,36]. The up-regulated gene CIT1 (KLMA_20105) encoding citrate synthase reasonably interpreted the increase in citric acid in the fermentation process of rice-acid. CIT1 acts as a quantitative marker for healthy mitochondrion and is encoded by the nuclear DNA [37]. Interestingly, lactic acid has optical isomers: L-lactic acid and D-lactic acid, which can be produced by chemical synthesis (DL-lactic acid) or microbial fermentation (L-lactic acid, D-lactic acid, or DL-lactic acid). Compared to chemical synthesis processes, microbial fermentation processes present more advantages since they make use of renewable substrates from lactic acid bacteria [38]. Consistently, our study demonstrated that fermented rice-acid inoculated with K. marxianus L1-1 could produce more concentration of L-lactic acid in 3 day than 1 day, and the L-lactic acid has some advantages for healthy. The genes CYB2 (KLMA_10621 and KLMA_40341) encoding L-lactate dehydrogenase were up-regulated, whereas another CYB2 (KLMA_30013) was down-regulated. DLD1 reported proved that fumarate hydratase (FUM1) and succinate dehydrogenase (SDH) were tumour suppressors [39]. Indicating K. marxianus L1-1 could have potential probiotic characteristics. Therefore, the enhanced activities of proteins and enzymes in TCA cycle and pyruvate metabolism indicated the increased organic acids in rice-acid inoculated with K. marxianus L1-1.

Down-regulated proteins and genes indicated the stable formation of the avor
Interestingly, most of the genes and proteins involved in the 5 KEGG pathways includes starch and sucrose metabolism, amino sugar and nucleotide sugar metabolism, glycolysis/gluconeogenesis, pyruvate metabolism and TCA cycle were up-regulated except the genes CTS1, ALD family and DLD1.
The reason had been discussed above. The up-regulated genes and proteins played the active role in the formation of avor during rice-acid fermentation, whereas the down-regulated genes and proteins played the important role in maintaining the stable key avor. Many reports focused on the up-regulated genes or proteins in the role of promoting the avor maturity of fermented foods, the down-regulated genes or proteins were seldom reported. In our study, as seen from Fig. 5B, the down-regulated genes and proteins were involved in the KEGG pathways, including DNA replication, meiosis-yeast, homologous recombination, mismatch repair, autophagy-yeast, MAPK signaling pathway-yeast, phenylalanine metabolism, base excision repair, nucleotide excision repair, other types of O-glycan biosynthesis, tryptophan metabolism, cell cycle-yeast, penicillin and cephalosporin biosynthesis and D-Arginine and Dornithine metabolism. Most of the down-regulated pathways were related to the growth of K. marxianus L1-1 and this result provided a reasonable explanation of the decrease in the quantity of K. marxianus L1-1 in rice-acid during the key fermentation period (Day 3). The down-regulated proteins in DNA replication had the small Log2FC value (data is not displayed), indicating that related DEPs had small effects on the growth of K. marxianus L1-1. Therefore, the third day was the suitable ending of the fermentation process. In addition, the down-regulated KEGG pathways were related to the decomposition and utilization of substrates by K. marxianus L1-1. Consistently, a previous report demonstrated that downregulated proteins related to advanced glycation end products were implicated in the aging process [40].
In the future, we will focus on the in uences of the content of substrate of K. marxianus L1-1 on the avor formation of rice-acid.

Conclusion
The transcriptome and proteome of K. marxianus L1-1 in rice-acid were determined in the study. The differentially expressed genes and proteins related to the formation of ethyl acetate and organic acids were determined. DEGs and DEPs were identi ed and found to be enriched in the key KEGG metabolism pathways, including starch and sucrose metabolism, amino sugar and nucleotide sugar metabolism, glycolysis/gluconeogenesis, pyruvate metabolism and TCA cycle. With the complementary analyses of the transcriptome and proteome, we revealed the formation mechanism of ethyl acetate and organic acids in Chinese rice-acid inoculated with K. marxianus L1-1. This study provides the basis for improving aroma and taste of fermented foods and reveals the formation mechanisms of avors.

Methods
Strain culture and growth determination The strain of K. marxianus L1-1 was previously screened and isolated in the traditional fermented riceacid and could produce high concentration of aroma compounds, was used in the fermentation experiments.

Determination Of Volatile Compounds
Through SPME-GC-MS analysis, volatile compounds were determined according with the method of Molyneux and Schieberle [41]. Retention times and mass spectral data were used to identify each compound. The retention times of the volatile compounds were determined using a C6-C26 alkane standard. The concentrations of volatile components were calculated with the peak areas of the internal standard (10 µL of 2-methyl-3-heptanone, 10 mg/L). The mass spectra and retention indices were determined on at least two different GC columns that have stationary phases of different polarities and results were compared to spectra and retention indices.

Determination Of Organic Acids
After the settlement, the rice-acid samples inoculated with K. marxianus L1-1 were ltered with doublelayer lter paper. The obtained ltrate was ltered through a 0.

Proteomic Sequencing And Data Analysis
Similarly, three independent biological replicates of proteomic sequencing of K. marxianus L1-1were used in the samples Y 1 d and Y 3 d. The preparation for proteomic analysis of K. marxianus L1-1 cell, liquid chromatography and mass spectrometry, peptide and protein identi cation and quanti cation according to the method by Xu et al [42]. All data with a 95% con dence and false discovery rate (FDR) less than 1% were considered to result in false positive results. According to the protein abundance level, the difference of more than 1.5-fold change (FC) and the statistical test of the p-value less than 0.05 were deemed to be differentially expressed proteins (DEPs) between Y 3 d and Y 1 d. All of DEPs were analyzed by GO and KEGG. The FASTA protein sequences of DEPs were blasted against KEGG database to retrieve their KEGG Orthologies (KOs) and were subsequently mapped to the pathways in KEGG. The corresponding KEGG pathways were extracted.Correlation Analysis Between Proteomic And Transcriptomic Results The DEGs and the DEPs were separately counted, and the Venn diagrams were plotted according to the counted results. Correlation analysis (Pearson correlation coe cient) was performed by Origin Pro 2018, and the four-quadrant maps were drawn based on changes in the transcriptome and proteome analysis.

Statistical analysis
All experiments were conducted in triplicate. Data were represented as the means ± standard deviation.
Duncan's multiple range test and t-test were carried out to analyze signi cant differences in SPSS version 20.0 (SPSS Inc., Chicago, IL, USA). whereby P 0.05 or P 0.01 were considered to be statistically signi cant.   The volcano map of differentially expressed genes (a) and differentially expressed proteins (b) in K. marxianus L1-1when inoculated rice-acid (Y 3 d vs Y 1 d).

Figure 3
Statistical diagram of the second node annotation (a) and the most enriched GO Terms (b) of the differentially expressed genes in K. marxianus L1-1 when inoculated rice-acid (Y 3 d vs Y 1 d).
Page 28/32 The GO analysis (molecular function, cellular component and biological process) of differentially expressed proteins in K. marxianus L1-1 when inoculated rice-acid (Y 3 d vs Y 1 d). Rich factor, the more signi cant the enrichment was. The Q-value was the corrected p-value after multiple hypotheses testing, which was ranged from 0 to 1. The closer to zero, the more signi cant the enrichment was (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this study).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.