Saccharomyces Cerevisiae Strains Selected from Nature Signicantly Increased the Production of 2-Phenylethanol Through Protoplast Fusion

Background: 2-Phenylethanol (2-PE) is an aromatic alcohol with rose fragrance, which is widely used as an additive in food, tobacco and daily chemical industries. Yeast is the main microorganism producing natural 2-PE, but it is limited by low production and weak tolerance. Nature and fermented products is a resource treasury of yeasts with excellent traits. Screening strains with good phenotypic traits and conducting breeding by cell fusion for genetic pyramiding is an effective way to improve strains. Results: In this study, 25 strains of 2-PE-producing yeasts were isolated from Chinese brewed samples. Three Saccharomyces cerevisiae strains with good traits in tolerance and 2-PE titre were screened out. The strain LSC-1 produces 2-PE of 3.41 g/L with an increase of 9.3% compared to the industrial strain CWY132. The strain NGER shows good tolerance to 2-PE at the concentration of 3.60 g/L in agar plate, and the thermotolerant strain S.C-1 shows growth ability at 41 ℃ . Two rounds of protoplast fusion were performed with these three parent strains for pyramiding of traits. A fusant strain RH2-16 with high 2-PE titre and increased tolerance was obtained. Using 5g/L L-phenylalanine as the precursor substrate, the maximum titre of 2-PE produced by the RH2-16 strain through fermentation and transformation is 4.31 g/L, and the average titre is 4.04 g/L. The molar conversion rate of L-Phe reached 115% in 36 h. Compared to the parental strain LSC-1 and the industrial strain CWY132, 2-PE titre in RH2-16 increased by 26.4% and 38.1%, respectively. Conclusion: Diversied S. cerevisiae strains with different traits can be isolated from the brewing related samples. Protoplast fusion technology can effectively pyramid excellent genetic traits and breed yeast strains with signicantly improved tolerance and 2-PE titre. Our research provided a breeding strategy for S. cerevisiae and a strain for industrial production of 2-PE.

Background 2-Phenylethanol is an aromatic compound with fragrance widely used in the food and cosmetics industries in the world [1][2][3]. At present the global annual output of 2-PE is nearly 10,000 tons. 2-PE is mainly produced by chemical synthesis, but this method produces toxic by-products such as biphenyl [4,5]. Another production method is the natural extraction from rose essential oil, but this method is ine cient and costly to carry out large-scale industrial production that affected by plant raw materials [6]. Recently the method of microbial conversion for natural 2-PE production has received extensive attention [1,[7][8][9]. The microorganisms that can produce 2-PE mainly include yeast, fungus and bacteria.
Nowadays, the most researched are yeasts, such as S. cerevisiae [10,11], Zygosaccharomyces rouxii [10,11], Pichia pastoris [12], Kluyveromyces lactis [13] and K. marxianus [14]. It is reported that the fungus Aspergillus oryzae can synthesize 2-PE [15]; Enterobacter sp. CGMCC5087 produces 0.1 g/L 2-PE after 24 h of fermentation [7]. In the research and application of 2-PE producing yeasts, S. cerevisiae is the most commonly used one. Compared with other microorganisms, S. cerevisiae is a model strain with clear genetic background, GRAS(generally regarded as safe) and other signi cant advantages. It also shows tolerance to many adverse environmental stresses and a high degree of adaptability to industrial environments. The large-scale industrial fermentation production technology of S. cerevisiae is relatively mature.
S. cerevisiae synthesizes 2-PE by de novo Shikimate pathway or Ehrlich pathway. When L-phenylalanine (L-Phe) is used as the sole nitrogen source in the medium, 2-PE is mainly synthesized through the Ehrlich pathway. Production of 2-PE from glucose and a non-amino acid nitrogen source through de novo pathway in S.cerevisiae has been discussed and studied [1,16,17]. Improved 2-PE production by strengthening the shikimate pathway in S.cerevisiae was reported recently. The engineered strain JM26 achieved 643 mg/L of 2-PE, which is 6.1-fold higher than the control strain with glucose as the sole carbon source [18]. Hassing et al reported that the geneticly manipulated strains were capable of producing 13 mM of 2-PE at a titre of 0.113 mol mol − 1 , represents the highest titre for de novo production 2PE in S. cerevisiae and other yeast strain ever achieved [19]. However, synthesis of 2-PE via de novo pathway in high concentration remains a metabolic engineering challenge [19].For the industrial production of 2-PE, it is mainly synthesized through exploration and utilization of the Ehrlich pathway with L-Phe as the only nitrogen source [1].L-Phe is catalyzed by aromatic amino acid aminotransferase I/II to generate phenylpyruvate, and then decarboxylated by phenylpyruvate decarboxylase to generate phenylacetaldehyde. Phenylacetaldehyde is reduced to 2-PE under the catalysis of alcohol dehydrogenase. Based on the Ehrlich pathway, how to increase the production of 2-PE has been extensively studied, including genetic engineering and optimization of bioconversion conditions. When two genes ARO9 and ARO10 in Ehrlich pathway and the global transcription factor gene ARO80 were overexpressed and acetaldehyde dehydrogenase gene ALD3 in branch pathway was knocked out simultaneously, the 2-PE titre increased by nearly 5 times in S. cerevisiae strain W303-1B [20].
Studies have shown that the tolerance of yeast to 2-PE is a key factor affecting titre. In addition, S. cerevisiae will produce more by-product ethanol when converting to produce 2-PE, and both will exert product inhibition effect simultaneously, which becomes the biggest obstacle to further increase the titre of 2-PE [21,22]. By using continuous culture and In-Situ Product Recovery (ISPR) for product separation [2,[23][24][25], relieving the inhibitory effect of 2-PE on the growth of yeast, the output of 2-PE increased from 2.28 g/L to 5.14 g/L at the level of shake ask fermentation [26]. These studies indicated that the release of product inhibition or the improvement of the tolerance of S. cerevisiae to 2-PE is the bottleneck for the breakthrough of yield.
High concentrations of 2-PE affect many aspects of cells, including changes in morphology [27,28], metabolic activity, gradual decrease in cell membrane permeability, intracellular ROS production and mitochondria damage [29]. The cell's tolerance to alcohols is complex and controlled by many factors, but the molecular mechanism is still unclear [30]. Using genetic engineering strategies to accurately manipulate the synthesis pathway of 2-PE might effectively increase the titre of 2-PE, but the improvement of tolerance is limited. In addition, in terms of natural 2-PE production requirements, the safety and naturalness of genetically engineered bacteria also cause a certain degree of concern. Therefore, it is necessary to explore new stable and e cient breeding strategies to improve the tolerance of S. cerevisiae to 2-PE and increase the production accordingly.
Yeast and its brewed food have accompanied human civilization for thousands of years. China's brewing industry has been developed since ancient times, and the diversity of yeast strains is very rich. For example, when preparing Moutai avored wine, S. cerevisiae exists in an environment with multiple stress factors such as high temperature, high acid and high ethanol concentration. During the process of making koji, the temperature can reach up to 60℃, the pH is around 2.0, and the ethanol can reach over 15% [31]. The microbial ora that grows in this special brewing environment has unique physical and chemical characteristics. Therefore, yeasts with various excellent traits can be obtained through selective screening. For example, a thermotolerant strain obtained from multiple stress showed increased tolerance to 2-PE of 2.5 g/L, signi cantly higher than the nonthermotolerant strain of 1 g/L. Correspondingly, the 2-PE titre of the heat-resistant strain reached 4.5 g/L, which was 57% higher than that of the non-heatresistant strain [32].
After obtaining multiple yeast strains with different excellent traits, it is necessary to integrate these different traits into one yeast strain. This can be obtained through genetic pyramiding, namely, combining several important genes in one strain. Protoplast fusion has greater potential for e cient gene pyramiding, through which a stable recombinant with parental genetic characteristics can be obtained through recombination of large fragments of genomic DNA. Protoplast fusion offers an opportunity for circumventing barriers to sexual reproduction and enriching the gene pool of nal strains. Morover, in the process of cell fusion, new genetic variability derived from random combinations of genes of the parental genomes could be e ciently utilized. Therefore the purpose of improving traits can be effectively achieved without understanding the speci c regulation mechanism of a trait [33][34][35].
In the process of protoplast fusion, parental inactivation is an effective and commonly used method for screening fusants, which means that the parents are each inactivated with different factor, and only the complementary regenerated fusants can grow. This method has been applied to improve the ethanol production performance of S. cerevisiae. For example, the fusion strain Q/L-F2 has both the fermentation performance and ethanol tolerance of the parents, and the tolerance of ethanol has reached 14% (v/v) [36]. In addition, for yeast strains of different species, protoplast fusion still has a signi cant effect. The fusion strain SP2-18 of S. cerevisiae and P.pastoris can assimilate xylose, and the ethanol titre is increased to 74.65 g/L compared to the parent's 33.12 g/L and 65.44 g/L [37]. The fusion strain R6 of S. cerevisiae and Candida ethanolica displayed superior performance and desirable properties from both parental strains and was described with great potential for the production of high-quality, low-alcohol cider [38].These studies demonstrated that the protoplast fusion is an effective strategy for strain improvement in S. cerevisiae.
After obtaining better strains through cell fusion, genetic engineering technology including gene editing can be used to implement more precise gene regulation and strain construction to further increase the ability of 2-PE synthesis. CRISPR-Cas9 technology has been successfully applied to the genetic manipulation of S. cerevisiae [39].
Our previous research had shown that the S. cerevisiae industrial strain CWY132 produced 2-PE of 3.52 g/L under the optimized culture conditions [40,41]. At present, the main bottleneck faced in breeding for strain to further increase 2-PE titre is to improve the tolerance. In this study, we excavated natural S. cerevisiae resources from China to screen strains with high 2-PE titre and tolerance, and then used protoplast fusion technology to carry out trait pyramiding to further improve 2-PE production.

Results And Discussion
Isolation and identi cation of yeast strains A total of 25 yeast strains were obtained from the samples collected in China (Table 1).S. cerevisiae is the main yeast among the isolated strains. Observation of yeast colonies showed that the colony morphology was generally round ridges, milky white, opaque, neat edges, and no wrinkles on the surface.
The Wickerhamomyces anomalus colony has irregular edges and a large number of wrinkles on the surface. Through colony morphology analysis, microscopic cell observation and molecular identi cation, 19 strains of S. cerevisiae, 2 strains of W. anomalus, 2 strains of P.pastoris, 1 strain of Tolerula spores, 1 strain of Candida anglica were identi ed. According to ITS sequence analysis, those with a similarity greater than 95% belong to the same genus, and the genetic relationship of each strain was shown by drawing a phylogenetic tree. The S. cerevisiae CWY132 stored in our laboratory shows a close genetic relationship with the SC001, SC003 and SC009 isolated from Sichuan brewed samples. They were more closely related to S. cerevisiae JYC2558 in GenBank. LSC-1 from CICC and SC008 isolated from Gansu homemade sourdough starters has high homology, while S.C-1 isolated from Zhejiang Quzhou rice wine is not closely related to S.C-2 from Zhejiang Shaoxing wine, the same province ( Fig. 1).

Evaluation of the Performance of Isolated Strains
The titre of 2-PE, tolerance to 2-PE and high temperature of the 25 isolated yeast strains were tested and evaluated. The industrial S. cerevisiae strain CWY132 in our laboratory was used as a reference. The results showed that the ability of different strains to convert L-Phe and synthesize 2-PE was signi cantly different. As shown in Table 2, there are 7 yeast strains with 2-PE titres between 1~2 g/L, and 4 yeast strains with titres below 1 g/L. The 2-PE synthesis capacity of S. cerevisiae is greater than other yeast species. Among them, the titres of 2-PE of the ve S. cerevisiae strains LSC-1, SC002, SC003, SC004 and SC005 were stable above 3.4 g/L, which were at least 10% higher than the industrial strain CWY132 ( Table 2). This result proved that S. cerevisiae has a natural advantage in the synthesis of 2-PE, compared with other unconventional yeasts. These results further con rmed that the production of 2-PE by yeasts depends on the yeast strains as described previously [13,42].
With CWY132 as a control, the ve high-titre S. cerevisiae strains LSC-1, SC002, SC003, SC004 and SC005 were tested to evaluate their growth performance. The results showed that there were no signi cant differences in the growth rates among these ve strains, but their growth rate were all signi cantly higher than CWY132, and the maximum growth was also higher (Fig.2). These results demonstrated that these strains with high titre of 2-PE also had good growth performance. As the growth rate of LSC-1 was higher than the other four strains, this strain was selected as the parental candidate strain for protoplast fusion.
Through gradient dilution and drop plate method, the strains tolerance to 2-PE was evaluated. Three strains of S. cerevisiae NGER, S.C-1 and LSC-1 were screened out, which showed good tolerance phenotype to 2-PE. When the concentration of 2-PE was lower than 3.00 g/L, the growth of these three yeasts and CWY132 were not signi cantly inhibited. But when the concentration was increased to 3.40 g/L, these three yeast strains grew signi cantly better than CWY132. When the 2-PE concentration in plate reached 3.60 g/L, only strain NGER showed growth at cell concentration of 10 7 /mL (Fig. 3a), indicating this strain has good traits underlying the ability to 2-PE tolerance. Our results are similar to those previously reported, that is, 4.0 g/L 2-PE can completely inhibit the growth of tested yeast cells [22].
The performance of these strains with good 2-PE tolerance at high temperature was also tested further.
The results showed that at 40°C the growth of CWY132 was signi cantly inhibited, whereas the other three yeast strains still showed growth. When the temperature was increased to 41°C, only strain S.C-1 grew after incubation for 48h ( Fig. 3b), indicating that S.C-1 resist the stress of high temperature very well.
Studies have shown that thermo-tolerance is a good indicator of other stress tolerance. For example, strains with higher temperature tolerance also have stronger resistance to oxidative and osmotic stress [43]. It was found that a thermo-resistant strain Ye9-612 had a constant growth rate at 30°C to 40°C, but the growth rate of the non-heat-resistant strain Ye9-596 (the offspring of the same parent) was severely affected [32]. The multiple stress resistance phenotype also exists in fungi and bacteria [44,45].
In conclusion, using the 2-PE industrial strain CWY132 as a control, we obtained the LSC-1 strain with stronger 2-PE synthesis ability, the NGER strain with signi cantly improved tolerance to 2-PE, and the S.C-1 with better high temperature tolerance. These three strains served as parent strains for subsequent cell fusion.

Optimization of Conditions for Protoplast Preparation
In protoplast preparation by enzymatic hydrolysis, the effect of enzyme concentration and treatment time on the preparation rate and regeneration rate of protoplasts was explored. We set the snailase concentration of 20, 30, 40, 50, 70, 100 and 140 mg/mL, respectively. The results showed that when treatment by snailase was 70 mg/mL for 100 min, the preparation rate was close to 75%~85%, and the regeneration rate was close to 10~18% ( Fig. 4a; Fig. 4b). Further, we improved the protoplast preparation by addition of driselase. At a concentration of 2%, the protoplast preparation rate of LSC-1 was close to 95% after 30 minutes of treatment with driselase (Fig. 4c).
Highly tolerant fusants produced by the rst round of protoplast fusion S. cerevisiae NGER and S.C-1 were used as the parental strains to construct the rst round protoplast fusion through inactivating parental protoplasts to obtain the strains capable of enhancing 2-PE related characters. The inactivation rate was close to 100% at the condition of that NGER protoplasts were exposed to a water bath at 60 ℃ for 2 min and S.C-1 protoplasts were irradiated at a distance of 20 cm under a 25W UV lamp for 15 min (Fig.4d, Fig.4e).There are 20 fusants were obtained. One fusant RH1-4 showed similar growth performance to NGER but with increased 2-PE tolerance at a concentration of 3.60 g/L in drop plate (Fig. 5a). Next, we compared the growth performances of RH1-4, NGER, S.C-1 and CWY132 incubated at 40°C. The results showed that RH1-4 has the same thermal resistance as S.C-1 ( Fig.5b), demonstrating that the fusion strain RH1-4 has obtained both the traits of high temperature and 2-PE stress tolerance from parent NGER and S.C-1.
RH1-4 also showed increased 2-PE titre of 2.11 g/L, increased by 11%~24% compared with two parental strains (Fig.5c). The results showed that the stress resistance of the fusant RH1-4 was improved, thereby increasing the titre of 2-PE. The high tolerance of RH1-4 is very valuable for further improving the 2-PE synthesis ability in S. cerevisiae.
Construction of 2-PE high-titre fusants through the second round of protoplast fusion S. cerevisiae RH1-4 and LSC-1 were used as parent strains for the second round of protoplast fusion to obtain a strain with high tolerance to 2-PE and increased titre. There are about 100 fusant strains were obtained and classi ed into grades A to C according to 2-PE tolerance, as follows: A: close to the level of LSC-1; B: between LSC-1 and RH1-4; C: close to the level of RH1-4 (Fig. 6a). Contrary to the expected result, only a small number of fusants in the C grade titreed 2-PE close to RH1-4 of 2.11 g/L, and most of them were lower than RH1-4, which was close to the titre of rst round parent SC-1 of 1.67g/L (Fig. 6b). Although the tolerance of grade B fusants did not reach the expected level as RH1-4, their titres were relatively higher. Among them, one strain RH2-16 produced 2-PE of 4.05 g/L, increased by 18.8% compared with the parent LSC-1.The other selected strain RH2-26, also showed a high 2-PE titre of 3.96 g/L, an increase of 16.5% (Fig. 6b). The results of transferring the excellent tolerance traits of RH1-4 to LSC-1 through the second round of fusion show that this method is very effective in further improving 2-PE titre. These results demonstrates that direct screening of yeast strains with multiple traits from nature and pyramid of traits through protoplast fusion is a practical method for yeast breeding. This method is more simple and e cient, compared with genome shu ing which needs mutant library construction and recursive fusion [46]. This approach would have a good application prospect in microbial breeding.

2-PE titer of yeast strains at different fermentation temperature
Given that the fusant strain has acquired the heat tolerance of the parent strain S.C-1, does the 2-PE titre of the fusant strain increase correspondingly under high temperature conditions? Therefore, the 2-PE production of strains CWY132, NGER, S.C-1, LSC-1, RH1-4 and RH2-16 at 30°C, 37°C, and 40°C were measured. The results showed that as the temperature increased from 30°C to 37°C, the 2-PE titres of the six strains all showed a signi cant decline, with an average decrease of 12.8%. The decline in titre was most signi cant at 40°C, with an average decline of 42.7%. Among the six strains, the titre of RH2-16 decreased the most, from 4.18 g/L (30°C) to 3.24 g/L (37°C) and 1.13 g/L (40°C), with the largest decrease of 73.0%. Although S.C-1, which has a lower 2-PE titre (1.88 g/L, 30°C), has a certain growth ability at 41°C, the 2-PE titre at 40°C still drops by 23.4% (Table 3). This result indicates that the 2-PE synthesis of thermotolerant strains of S. cerevisiae will still drop signi cantly under high-temperature stress. The reason for this phenomenon is that S. cerevisiae is under the dual stress of high temperature and 2-PE when it is fermented and transformed to produce 2-PE under high temperature conditions. Therefore, cells must achieve a compromise and balance between maintaining growth, tolerance to stress and synthesis of 2-PE. Studies have shown that heat-resistant strains can help reduce the toxic effects of 2-PE [47].Considering that the high temperature conditions are not suitable for the actual conditions of industrial production, combined with this results, we believe that a higher production of 2-PE can be obtained by fermentation at 30°C, and the high temperature resistance of the strain can be fully utilized to overcome the stress of 2-PE.

Identi cation of fusants and determination of their genetic stability
To test the genetic stability as well as to eliminate the effect of reverse mutation, fusants of interest were transferred at least 5 generations and the 2-PE and ethanol production capacity was detected and compared. The result demonstrated that the fusants showed good genetic stability. In order to con rm the fusants at the level of cell biology, the DNA content of the fusants and parent cells was determined. The results indicated that the DNA content of NGER, SC-1 and LSC-1 cells were 0.71, 0.67 and 1.08 μg per 10 7 cells, respectively. The DNA content of fusants RH1-4 and RH2-16 was higher, reaching 1.05 and 1.49 μg per 10 7 cells (Table 4). The DNA content of the fusants was greater than the DNA content of two parent strains, but was less than its sum, which was consistent with the characteristics of heterozygoma, con rming the preparation of the fusion strain. Fusants of interest were subcultured and the synthesis ability of 2-PE was found to remain stable, the titre of RH2-16 reached the highest 4.31 g/L at the third generation, the average titre of the ve generation strains is 4.04 g/L ( Table 5). The comprehensive comparison of 2-PE titre, tolerance and other traits of parent and fusant strains are shown in Table 6. The minimum number of cells needed to grow at 40 ℃ can be used to characterize the high temperature tolerance of different strains Table 6 .
Inokuma et al [48] performed transcriptome analysis on the diploid yeast strain with good stress tolerance and xylose utilization obtained by rearranging the spore genome, and found genes that affect the strain's tolerance to multiple stresses. It helped to understand the multiple tolerance mechanism of yeast to high temperature and acid. Therefore, the RH2-16 in this study with good traits should be an excellent candidate strain for analysis of character and research on mechanisms underlying 2-PE metabolic pathway regulation.

Conclusion
In order to obtain 2-PE high-titre strains, a strategy of screening yeasts from natural and brewed foods and breeding through cell fusion was selected. Three excellent yeast strains, LSC-1 (good performance in both growth and 2-PE synthesis), NGER (tolerance to high content of 2-PE) and SC-1 (thermal-resistant and grows at 41°C) were screened out. After two rounds of protoplasts fusion, the traits underlying 2-PE titre enhance and tolerance have been successfully pyramided into a fusant strain RH2-16 with the highest 2-PE titre of 4.31 g/L. This strain shows improved traits, which is better than industrial strain CWY132 in both the growth and 2-PE production performances. This strain also shows good tolerance to 2-PE and high temperature, which is very suitable for the environmental conditions of large-scale industrial fermentation production. In short, we suggest that screening of S. cerevisiae strains with different traits from nature and breeding by protoplast fusion is an e cient strategy for yeast improvement.

Materials And Methods
Strains and media S. cerevisiae strains CWY132 and NGER are stored in our laboratory. LSC-1, LSC-2, and LSC-3 were purchased from the China Center of Industrial Culture Collection (CICC). The remaining strains were isolated in this study. YPD medium (20 g/L glucose, 20 g/L peptone, 10 g/L yeast extract, 20 g/L agar) was used for strain activation and isolation. Single colonies were picked into YPD liquid medium and cultured in a 30°C shaker at 200 rpm for 12 h; YPDS medium (protoplast regeneration medium): 182.17 g/L sorbitol is added to YPD medium; 2-PE fermentation medium: 30 g/L glucose, 0.5 g/L MgSO 4 , 5 g/L K 2 PO 4 , 1.5 g/L yeast extract, add 5 g/L L-Phe as precursor. All media were sterilized at 121°C for 15 minutes before use. Isolation and identi cation of yeast strains Samples were collected from breweries including Zhejiang Shaoxing wine, Zhejiang Quzhou rice wine, and Sichuan white wine, commercial fermentated broth, homemade sourdough starters in Gansu province, and Sichuan soy sauce and vinegar factory preparation residues.
Yeast strains were isolated as follows. 1.0 g of the sample was dissolved in 9 mL of sterilized water and mixed well. After gradient dilution, sample was spread on YPD medium plate and incubated at 30°C for 48 h. The yeast colonies characterized by spherical, milky white and opaque were further streaked to obtain single colony and con rmed by microscopic observation. The isolated strains were further subjected to molecular identi cation.
The genomic DNA of the yeast strain was extracted with Rapid Yeast Genomic DNA Isolation Kit (Sangon Biotech, Shanghai, China). The full sequences of ITS were ampli ed with primer pair Seq-F (5 , -CGCCAGGGTTTTCCCAGTCACGAC-3 , ) and Seq-R (5 , -GAGCGGATAACAATTTCACACAGC-3 , ). PCR reaction program: pre-denaturation at 94℃ for 5 min; denaturation at 94℃ for 1 min, annealing at 55℃ for 1 min, extension at 72℃ for 1 min, 30 cycles; nal extension at 72℃ for 5 min. TransTaq DNA polymerase High Fidelity (Transgen) was used in PCR. PCR products were puri ed for sequencing after agarose gel electrophoresis. The full ITS sequences obtained by PCR were Blasted against GenBank nucleic acid database for similarity analysis. The species of the strains were determined according to whether the sequences have high similarity, and MEGA7.0 software was used to perform phylogenetic tree analysis.

Determination of growth curve
The growth curve of all strains was made by measuring the cell density with spectrophotometer. The activated single colony was picked into YPD liquid medium and cultured in a shaker at 30°C at 200 rpm for 12 h. Then the OD 600 of the cultured yeast solution was measured and adjusted to a density of 1×10 7 cells/mL. The cell suspension was inoculated into the fresh YPD medium at a ratio of 10% (v/v), and the OD 600 of the culture medium was measured every 2 h. Using the OD 600 as ordinate and culture time as abscissa, growth curves were drawn.

Determination of tolerance to 2-PE and high temperature
The tolerance of 2-PE was analyzed by dilution gradient plate method. The yeast concentration was adjusted to 10 7 cells/mL, and then diluted with sterile water to 10 6 , 10 5 , 10 4 , 10 3 cells/mL. 5 μL of the serious diluted yeast solution was taken and dropped on the YPD plate with series concentrations of 2-PE (2.6, 2.8, 3.0, 3.2, 3.4, 3.6 g/L). After being placed to dry, put it in a 30℃ incubator for 2-3 days. In the high temperature tolerance test, the plates were cultured at 37°C, 40°C and 41°C for 3 days to observe the growth of the strains.

Protoplast preparation, fusion, regeneration
The yeasts were picked into 5 mL fresh YPD liquid medium, and cultured in a 30°C shaker at 200 rpm for 12-14 h until logarithmic growth phase. 4 mL of the yeast suspension was taken and centrifuged at 6000 rpm for 5 min. The pellet was washed twice with PB buffer and 4 mL of pretreatment agent was added and shaken slowly at 60 rpm for 30 min in a shaker at 30°C. Cells were collected by centrifugation at 6000 rpm for 5 min and mixed with 4 mL enzyme cocktail consisting of 7% snailase(Sangon Biotech Shanghai, China) and 2% driselase(Sigma-Aldrich,USA) in PBS solution. The enzymatic hydrolysis was carried out for 2 h at 30℃ with shaking at 60 rpm. Then the digested cell solution was centrifuged at 3000 rpm for 10 min to remove the enzyme solution. The protoplast was washed twice and resuspended in the PBS. The methods for determining the protoplast formation rate and regeneration rate are as follows. The yeast suspension before enzymolysis was took out and diluted with sterile water, then spread on the YPD plate, incubated at 30°C for 2 d, and the number of colonies were recorded (A); The yeast suspension after lytic enzyme treatment was diluted with sterile water and spread on the YPD plate, incubated at 30°C for 2 d, and the number of unbroken colonies were recorded (B); The obtained protoplast suspension was spread on the YPDS plate and incubated at 30℃ for 2 d, and the number of regenerated colonies were recorded (C); The formation rate and regeneration rate of protoplasts were calculated based on the values of A, B, and C. Protoplast formation rate=(A-B)/A×100%. Regeneration rate of protoplast= (C-B)/(A-B)×100%.
The method of parental inactivation is used for cell fusion and selection. Reliable fusants could be obtained only when the inactivation e ciency reached 95% or more.
We observed and counted the prepared protoplast suspension on a hemocytometer, diluted it to 10 5 with PBS, add 4 mL of the protoplast suspension to a sterile plate. The plate suspension was placed under a 25W preheated ultraviolet lamp and irradiated at 20 cm for different time, then diluted with PBS solution and coated on the YPDS plate. The protoplast suspension without ultraviolet treatment was used as a control. The plates were incubated statically at 30°C and calculated the survival rate of protoplasts to determine the best UV mutagenesis time. Similarly, the protoplast suspension with a concentration of 10 5 was treated in a 60℃ water bath for different time. The survival rate of protoplasts was calculated with no heat treatment as a control.
The concentration of the two parent protoplast suspensions was adjusted to 10 7 /mL, 0.5 mL of each was mixed in a 1.5 mL EP tube, centrifuged at 3000 rpm for 10 min, the supernatant was discarded, and 1 mL of 40% PEG6000 was added to suspend the suspension and incubated at 30°C for 25 min. Protoplast suspensions were centrifugated at 3000 rpm for 10 min and collected cells were diluted with PBS solution to 10 5 /mL. 100 μL was taken and spread on YPDS plate, and incubated at 30°C for 3 d. Through the screening method of parent inactivation, the colonies grown on the YPDS plate were preliminarily determined as fusants.

Fusants identi cation
Fusants were also characterized by measuring the genomic DNA content. It was believed that the fusion cell's genomic DNA content should be greater than that of any parent, but should be close to or less than the sum of the DNA content of the two parents. The following formula was used to quantify the DNA concentration: where C was the DNA concentration (g/mL), K i was the dilution factor, and E was the DNA concentration where 1 UV absorption unit was equal to 50μg/mL [38]. C=OD 260/280 ×K i ×E/1000 2-PE fermentation production and titre determination A single colony was inoculated into 50 mL YPD liquid medium and cultured at required temperature with shaking at 200 rpm for 12-14 h to the logarithmic growth phase. The yeast cell concentration was adjusted to 10 7 cells/mL and the resulting cell suspension was added to 50 mL fermentation medium containing 5 g/L L-Phe at the inoculation ratio of 10%(v/v). The fermentation was carried out for 36 h by incubation at 30°C, 200 rpm. Fermentation broth samples were taken to determine the concentration of 2-PE. The experiment was conducted three times independently, with at least three repetitions for each treatment.  Values are mean±standard error of three assays of 2-PE and ethanol. a(vs CWY132; one-way betweensubjects ANOVA) Values are mean±standard error of three assays for 2-PE and ethanol and OD 600 . Values are mean±standard error of three assays  Figure 1 The phylogenetic tree of yeast strains was drawn using MEGA7.0 software. An unrooted phylogenetic tree was constructed by the neighbour-joining method, together with 1000 bootstrap replicates based on the  Growth curve of yeast strains. Yeast cell growth was monitored by the increase in the optical density of the cultures at 600 nm (OD600). The data were calculated from three biological replicates. for 48 h. CWY132 was used as a control.
after water bath heat treatment at 60℃; e The lethality rate of S.C-1 after UV treatment. The data represents the average of three independent experiments.

Figure 5
Screening strains with high 2-PE titre and strong tolerance from the rst round of protoplast fusants. a Five fusants among 20 were cultured on YPD plates for 48 h with the addition of 3.6 g/L 2-PE with parent NGER as a control. b High temperature tolerance test of RH1-4 at 40°C. c The 2-PE synthetic capability of Page 28/28 RH1-4. The data represents the average of three independent experiments, and the standard error is between 5% and 10%. The asterisk indicate that compared with S.C-1, the 2-PE titre of each strain is signi cantly different at the level of p<0.05.

Figure 6
Screening of 2-PE high-titre fusants in the second round of protoplast fusion. a Fusants strains and RH1-4 were tested under heat stress at 40℃. b The 2-PE production of fusants strains were measured, and strains with different 2-PE titres were classi ed into grades A to C according to 2-PE tolerance.