Introduction

l-valine is a hydrophobic and branched essential amino acid. It is used for infusion solutions, cosmetics, and as a precursor for the chemical synthesis of some herbicides and animal feed additives (Blombach et al. 2007; Park et al. 2007; Chen et al. 2015). l-valine is also pharmacologically relevant. For example, valsartan, whose main raw material is l-valine, is an excellent drug for treating high blood pressure (Senthil et al. 2009).

Fermentative production is becoming increasingly important due to its higher yield. The technique is used to produce l-valine, principally by Corynebacterium glutamicum, Serratia marcescens, Escherichia coli, Brevibacterium flavum, and Bacillus species (Kisumi et al. 1971; Chattopadhyay and Banerjee 1978; Park et al. 2007; Hou et al. 2012; Chen et al. 2015). The breeding of l-valine producing strains has been refined over many years. Methods of strain improvement range from random mutagenesis to rational metabolic engineering (Stephanopoulos 2002; Sang et al. 2005; Takors et al. 2007; Atsumi et al. 2008; Cobb et al. 2013; Oldiges et al. 2014). About 45 years ago, a mutant strain of S. marcescens No. 140 with alpha-aminobutyric (α-AB) resistance was obtained from S. marcescens No. 1 through ultraviolet light and N-methyl-N′-nitrosoguanidine (NTG) treatment (Kisumi et al. 1971). The authors reported the secretion of 8 g/L l-valine into the medium. Using C. glutamicum as a parent strain for mutagenesis, a mutant strain capable of producing 20 g/L of l-valine was obtained (Katsurada et al. 1993). However, the low and unstable l-valine yield in this mutant has hindered its use in industrial applications.

More recently, metabolic engineering methods have successfully generated mutant strains with a high l-valine yield (Park et al. 2007, 2011; Hasegawa et al. 2012, 2013; Hou et al. 2012). The biosynthetic pathway of l-valine is very complex. In corynebacteria, l-valine is synthesized from pyruvate in a pathway comprising four reactions (Fig. 1) catalyzed by acetohydroxy acid synthase (AHAS, encoded by ilvB and ilvN), acetohydroxy acid isomeroreductase (AHAIR, encoded by ilvC), dihydroxy acid dehydratase (DHAD, encoded by ilvD), and transaminase B (TA, encoded by ilvE) (Singh and Shaner 1995; Epelbaum et al. 1998; Hasegawa et al. 2012). Any gene abnormalities can affect the synthetic quantity of l-valine. 2-Ketoisovalerate is the most important precursor of l-valine. It has been used in l-leucine and pantothenate synthesis in a series of enzymatic reactions (Sahm and Eggeling 1999; Vogt et al. 2014).

Fig. 1
figure 1

Biosynthesis of l-valine in B. flavum. AHAS acetohydroxy acid synthase, AHAIR acetohydroxy acid isomeroreductase, DHAD dihydroxy acid dehydratase, TA transaminase B

Full size image

The complexity of l-valine synthesis spurred research into rational metabolic engineering as a means of re-engineering strains. As well as building on the first success in increasing the yield of l-tyrosine in Streptomyces fradiae (Zhang et al. 2002), the technique of genome shuffling has been successfully used to improve the production of metabolites (Gao et al. 2012; Luo et al. 2012; Lv et al. 2013; Wang et al. 2013; Zheng et al. 2013).

Mutation and selection are key components of evolution, which drives adaptation and the development of novel traits. Short generation times and a natural mutation frequency of 10−10 to 10−9 mutations per base pair in each replication cycle enable the selection of beneficial phenotype from high genetic diversity (Mahr et al. 2015). The success of any strain improvement depends on improvement in the required technologies and on the number and scope of mutants that should be screened after mutagenic treatment. It is imperative to establish a rapid high-throughput screening (HTS) method that has acceptable accuracy and repeatability, and is effective in evaluating a large number of isolates. Presently, we applied the protoplast mutation and genome shuffling breeding technique using B. flavum MDV1 to obtain mutant strains with improved l-valine yield, with the aim of identifying a strain suitable for industrial-scale fermentation (Fig. 2). We used an HTS technique termed 96 well plate-paper chromatography-high-performance liquid chromatography (96-PH), which is subsequently described in detail, as a simple, sensitive, and highly reproducible way to rapidly screen for mutants capable of l-valine production.

Fig. 2
figure 2

The procedure of l-valine breeding by protoplast mutagenesis and genome shuffling with 96-PH method

Full size image

Materials and methods

Chemicals

Lysozyme (> 22,800 U/mg), l-valine (> 98.5%), and polyethylene glycol 6000 (PEG6000) were purchased from Sangon Biotech (Shanghai, China). Binary ethylenimine (BEI, 98%) was purchased from Sigma-Aldrich (St. Louis, MO, USA). All other common reagents and solvents were analytical grade from commercial sources.

Strains and culture media

Brevibacterium flavum MDV1 was stored at the Department of Industrial Microbiology, Ministry of Education Engineering Research Center (Fujian Normal University). The storage solution and temperature were 25% (v/v) glycerol and − 80 °C, respectively. The complete medium (CM) used for B. flavum MDV1 storage contained (all g/L) glucose 5; peptone 10, beef extract 10, yeast extract 3, NaCl 5. The pH was adjusted to 7.0 with 4 mol/L NaOH.

Regeneration medium (RM) comprised CM supplemented with 0.5 M sucrose, 20 mM MgCl2·6H2O, and 20 mM sodium maleate. The pH was adjusted to 7.0 with 10% NaOH, and double-decked RM plates were prepared in which the lower RM contained 2% agar and the upper layer contained 0.8% agar.

Seed medium was comprised of (all g/L): glucose 25, corn steep liquor 5, yeast extract 10, soybean hydrolysate 1.5, (NH4)2SO4 25, MgSO4·7H2O 0.5, KH2PO4 0.75, K2HPO4 0.25, FeSO4·7H2O, CaCO3 10.

Fermentation medium comprised (all g/L): glucose 125, corn steep liquor 10, yeast extract 5, soybean hydrolysate 0.5, (NH4)2SO4 35, MgSO4·7H2O 0.5, KH2PO4 0.75, K2HPO4 0.25, FeSO4·7H2O 0.01, MnSO4·H2O 0.01, and CaCO3 35, as well as 0.1 mg/L each of vitamin B1, vitamin B5, and vitamin H.

The pH of the seed and fermentation media was adjusted to 7.0 with 4 mol/L NaOH for batch fermentation, and with 25% NH4OH for fed-batch fermentation.

Analytical procedure

Culture growth was monitored by measuring the absorbance at 562 nm (OD562nm) using a model UV-2000 ultraviolet spectrophotometer (Unico, Dayton, NJ, USA). Glucose concentration was measured using a model SBA-40D biosensor analyzer (Institute of Biology of Shandong Province Academy of Sciences, Shandong, China).

l-valine concentrations were determined by HPLC with a model SPD-10A UV detector (Shimadzu, Tokyo, Japan) according to the directions in the Venusil AA analysis kit (Chen et al. 2011). HPLC has been established as an accurate and preferred method to determine l-valine concentration. However, it is cumbersome to use for large-scale screening. The latter aim requires a system for determining l-valine concentration quickly and easily. Paper chromatography is a simple and rapid method for the qualitative and quantitative l-valine determination in fermentation samples. The technique relies on the determination of the Rf value, which is the distance travelled by a given component divided by the distance travelled by the solvent front. For a given system at a known temperature, the Rf value is characteristic of the component. The differing Rf values can be used to identify different amino acids in the sample. In addition, when different concentrations of l-valine are used for paper chromatography, greater concentrations will display a spot that is larger and more intensely colored than lower concentrations.

Based on the foregoing advantages, paper chromatography was used to determine the concentration of l-valine in broth, with a procedure that was slightly modified from that described previously (Li et al. 2009). Briefly, 1–2 µL of each sample was spotted onto the Xinhua Grade 1 chromatography paper (Xinhua Group, Hangzhou, China). The PC was developed at 25 °C with n-butanol–acetic acid–water (12:3:4, v/v/v) containing 0.5% ninhydrin. After development, the paper was directly dried for color yield in an 80 °C electric thermostatic drying oven for 10 min. The l-valine spots were cut from the paper and extracted with 10 mL of 75% v/v alcohol:0.1% w/v cupric sulfate (19:1, v/v) at 30 °C in a shaker operating at 220 rpm for 20 min. Absorption was determined using the aforementioned UV-2000 spectrophotometer at 510 nm. The calibration curve was linear over the concentration range tested (5.0–45.0 g/L). The resulting linear equation was y = 0.0091x + 0.1144 (R2 = 0.9991), where y is the absorption of the extracted solution at 510 nm and x is the concentration of l-valine.

Protoplast preparation, mutagenesis, and mutant isolation

Strains were pre-cultured at 30 °C for 9 h in 30 mL CM. Seed culture was transferred (600 µL) to a 250-mL flask containing 30 mL CW liquid and 1 mL 10% glycine, and growth was continued for 9 h. Cells were harvested by centrifugation at 4000 rpm for 10 min at 25 °C, washed twice with SMM buffer (0.5 M sucrose, 20.0 mM sodium maleate, and 20.0 mM MgCl2, pH 6.5), and suspended in 10 mL SMM buffer containing 1.5% lysozyme. After incubation for 6 h with gentle shaking (80 rpm) at 30 °C, protoplasts were obtained. The appearance of spherical cells as judged by light microscopy and was used as an indicator of protoplast formation. The protoplasts were collected by centrifugation at 3000 rpm for 5 min. The protoplast pellets were washed, re-suspended in SMM, and treated with UV (30 W) at an exposure distance of 20 cm in the dark for 0.5 min. This was followed by gradient dilution of 105 times with SMM. Aliquots (0.1 mL) of the diluted protoplasts were spread on RM containing 1.2 g/L BEI. After cultivation for 2–3 days at 30 °C, 60 colonies in every plate were transferred onto fresh slants. In the absence of a bacterial indicator of l-valine secretion, these strains were screened using 96-well plate analysis and flask fermentation to obtain a library of mutants with improved l-valine yield and preparation as starting strains for genome shuffling (Fig. 2). A diverse mutant population demonstrating improvement in the traits of interest, such as growth rate, yield, and impurity content, is required for a starting strain. Five rounds of random mutagenesis by UV and BEI were performed to achieve this end (Fig. 3).

Fig. 3
figure 3

Five rounds of protoplast mutagenesis and screening procedures were performed to prepare starting strains for genome shuffling

Full size image

Protoplast fusion and genome shuffling

Genome shuffling by recursive protoplast fusion was carried out as previously described, with some modifications (Patnaik et al. 2002; Zhang et al. 2002; Dai and Copley 2004). Candidate mutant strains from the protoplast mutagenesis library that displayed improved l-valine yield were used for the preparation of protoplasts. Equal numbers of protoplasts, from different parental protoplasts, were mixed and inactivated by 5 min of UV irradiation and then centrifuged at 4000 rpm for 10 min. PEG 6000 was added to 35% final concentration and supplemented with 0.02 mol/L of CaCl2. After 30 min treatment at 30 °C, the fused protoplasts were centrifuged, washed with SMM, and diluted in a gradient of SMM to the original concentration. Aliquots (0.1 mL) of the suspension were spread on RM, and the fusion frequency was calculated as the ratio of fusants from the protoplasts treated with PEG and CaCl2 to the number of the regenerating strains without inactivation treatment. After cultivation for about 2 days at 30 °C, the regenerated strains were screened using 96-well plate and shaking flask fermentation as described below. Two successive rounds of genome shuffling were carried out.

Screening and strain isolation

Efficient high-throughput screening can markedly improve the efficiency of breeding and shorten the screening cycle. All colonies from the protoplast regeneration plates were screened by 96-well plate fermentation first, and through shake flask fermentation in a second round of screening. Fermentation tests were carried out in batch culture to determine the maximum yield of l-valine produced by the obtained mutants. Seed cultures were grown for 16 h in 250-mL flasks containing 30 mL seed medium. Afterwards, 6.0 mL of these seed cultures were separately transferred to 250-mL shake flasks containing 24 mL of fermentation medium. The temperature was maintained at 30 °C and the incubation was carried out for 48 h. The final l-valine concentration produced by the mutants was determined at that time. The mutants with improved l-valine production were screened and preserved in 25% (v/v) glycerol at − 80 °C after isolation.

Fermentation process

l-valine fed-batch fermentations were performed in a 30 L automatic fermentor. A 2 mL bacterial suspension (1 × 108 bacteria per mL) of each test strain was inoculated into a 250-mL flask containing 30 mL of seed medium. The suspension was cultivated at 30 °C while shaking at 220 rpm for 16 h as a first seed culture.

One flask of 30 mL inoculums of this culture was aseptically added to a 5 L automatic fermentor (Zhenjiang East Biotech Equipment and Technology CO., Ltd, Jiangsu, China) containing 3 L seed medium, and cultivated at 30 °C for 16 h. Seed culture medium was inoculated into (20% inoculum) 12 L of production medium in a 30 L fermentor. The temperature was kept at 30 °C by a water circulation system, and the pH was maintained at about 7.0 with 25% NH4OH during the cultivation period. Dissolved oxygen (DO) was maintained between 25–35% by aeration at a rate of 0.9–1.2 m3/h and the agitation rate was kept between 250 and 600 rpm, with a straight-leaf disc turbine agitator with 6 blades. When the glucose level dropped below 10 g/L, sterilized glucose solution (70% w/v) was added to the fermentor to maintain a glucose concentration of 10–15 g/L in the fermentation broth. This glucose concentration is based on the pseudo-exponential batch-feeding strategy, and at the end of fermentation controlled glucose concentration is 1.0 g/L or less.

All experiments were conducted in triplicate, and the data were averaged and presented as the mean ± standard deviation.

Results

Protoplast mutagenesis and mutant screening

Genome shuffling requires a diverse population of mutants with demonstrable improvement in l-valine yield compared to that of the initial strain. In the present study, UV and BEI were used as the mutagenizing agents to improve the l-valine yield of the initial strain (B. flavum MDV1). MDV1 protoplasts were sensitive to UV radiation and BEI. Survival rate of the cells was reduced by increasing the exposure time to UV and BEI, with 15% survival at 0.5 min of UV irradiation or 1.2 g/L BEI. More than 1400 mutant clones were analyzed by UV and BEI mutagenesis of protoplasts in multiple rounds. Eight mutants displayed significantly enhanced l-valine production (MDV2-107, MDV3-39, MDV4-3, MDV4-69, MDV5-7, MDV5-15, MDV6-29, and MDV6-83) (Fig. 4). The eight mutants were picked out for the shake flask fermentation test. l-valine yields were 7.8, 8.2, 8.5, 9.2, 10.2, 10.3, 10.2, and 10.5 g/L for MDV2-107, MDV3-39, MDV4-3, MDV4-69, MDV5-7, MDV5-15, MDV6-29, and MDV6-83, respectively. These yields represent l-valine increases of 73.33–133.33% compared to the 4.5 g/L yield of the parental MDV1. Consequently, the eight mutants were used as the starting strains for genome shuffling. The mutagenesis family tree is presented in Fig. 4.

Fig. 4
figure 4

The mutagenesis family tree of some obtained mutants

Full size image

Genome shuffling

To obtain recombinants with an improved l-valine yield, two rounds of genome shuffling were carried out with the eight mutants serving as parental strains. The fusion moment was captured by the microscope after being stained with methylene blue (Fig. 5), and the fusion rates of the first and second genome shuffling reached 30.3 and 31.5% respectively. After each fusion, six colonies were selected from the first shuffled library according to their l-valine production for the next fusion. In the two rounds of genome shuffling, 360 colonies in each round were screened by 96-well plate fermentation and shaking flask fermentation. The l-valine production increased progressively. The highest l-valine yield in the first round of genome shuffling was 22.5 g/L after 48-h fermentation. The highest yield after the second round was 30.1 g/L in MDVR2-21. This yield was 568.8% higher than the yield of MDV1 (Fig. 6). These results indicate the success of genome shuffling in the marked improvement of l-valine yield.

Fig. 5
figure 5

Protoplast fusion micrograph for the starting strains

Full size image
Fig. 6
figure 6

Improvement of l-valine yield by protoplast mutation and genome shuffling

Full size image

As a quick and simple method to determine l-valine concentration in the broth, the PC assay was applied to the primary screening of the eight mutants during protoplast mutation and genome shuffling. Notably, the accuracy of PC as an amino acid analysis method is often limited by many factors such as sample composition, chromatography solvents, and ambient temperature. In our previous work, the accuracy and reliability of PC were examined. The results showed that the l-valine concentration in a sample can range from 5.0 to 45.0 g/L while maintaining a good linear relationship. Further, an l-valine standard solution with a concentration of 25.0 g/L was analyzed by the PC method five times and the test results were 25.32, 25.28, 25.21, 25.10 and 25.17 g/L with an average error of less than 2%. These data demonstrate that the accuracy of the PC method is excellent. In addition, the fermentation samples were also analyzed by PC and HPLC to further examine the accuracy of the PC method. The results show that the error rate between the two test results was within 5%, indicating very good reliability. A direct, obvious, and distinct change was evident (Fig. 7), indicating the utility of PC as a rapid and high-throughput method to identify mutants with a high l-valine yield.

Fig. 7
figure 7

The result of paper chromatography during protoplast mutation and genome shuffling. The sample volume placed onto the Xinhua Grade 1 chromatography paper varied, with 3 µL of initial strains added while 2 µL were added from the strains achieved by protoplast mutation and genome shuffling. The color yield from the initial strains to those from genome shuffling was rapidly enhanced, indicating that the mutant ability to secret the l-valine has greatly improved in turn

Full size image

Genetic stability of genome shuffling mutants

Genetic stability is an important parameter for a strain that will be used on an industrial-scale. To ensure the yield stability of the mutants screened by genome shuffling, six second-round genome shuffling strains were cultured for five generations and simultaneously fermented in shake flasks. The l-valine yields produced by subsequent generations of the six selected mutants are shown in Table 1. The MDVR2-16 l-valine yield decreased significantly after three culture generations. The properties of the other five high-yield strains were stable after five continual progeny tests. Mutant MDVR2-21, which exhibited the highest l-valine yield (average 30.2 g/L) in present work, was cultured for three generations, and then the biomass, l-valine production, and glucose conversion rate of each generation were measured. All the generations showed similar l-valine production, biomass, and glucose conversion rate as compared to the initial strain (Table 2), for which the range were 0.3 g/L, 1.7, and 1.1% respectively. These data suggest that mutant MDVR2-21 is genetically stable and suitable for further scale-up fermentation.

Table 1 Productivity stability tests of some genome shuffling mutants
Full size table
Table 2 The genetic stability of MDVR2-21
Full size table

Comparison of l-valine production, biomass, and glucose consumption of B. flavum MDV1 and MDVR2-21 in scale-up fermentation

To verify whether the mutant strain B. flavum MDVR2-21 is suitable for amplification fermentation, the fed-batch culture was carried out in 30 L and 120 m3 fermentors (Zhenjiang East Biotech Equipment and Technology CO., Ltd, Jiangsu, China). According to the fermentation process described above, samples from fermentors were taken every 2 or 4 h for analysis. The level of l-valine production was analyzed by HPLC after 24 h of fermentation, and sampled every 4 h. Meanwhile, both biomass and residual glucose were sampled every 2 h and determined by UV-2000 and SBA-40D, respectively. MDVR2-21 cultured fermentor for 64 h was particularly notable. MDVR2-21 displayed obvious differences to MDV1 regarding biomass and production of l-valine in the 30 L fermentor. The l-valine yield from MDVR2-21 reached a maximum of 70.1 g/L at 56 h, compared to the maximum yield of 26.3 g/L from MDV1. MDVR2-21 displayed a maximum l-valine production rate of 1.095 g/L/h after 64 h fermentation with a maximum glucose conversion rate of 38.9%. These represented increases of 266.5 and 126.3% when compared to MDV1, respectively. MDVR2-21 also showed better growth performance. From 0 to 12 h, the MDVR2-21 strain adapted to the fermentation environment and prepared for cell growth and product synthesis; this included an increase in cell density as well as an increase in glucose consumption within 12–24 h. This stage was considered the lag or adaptation phase of the MDVR2-21 strain and was significantly shorter (4 h shorter) than the that of the MDV1 strain. After 24 h of fermentation, MDVR2-21 strain cell density increased, and the rate of glucose consumption accelerated. The sterilization glucose solution (70%w/v) was added to fermentor started at 34 h, which was 10 h earlier than that of MDV1 strain (Fig. 8). With the increasing cell density, the MDVR2-21 biomass reached its maximum at 44 h with an OD562nm of 170.5 and a dry cell weight (DCW) per liter of 25.8 g/L, which were 5.9 and 23.3% higher than that of MDV1 strain respectively. After 44 h of fermentation, the biomass of MDVR2-21 stopped increasing and entered the stationary phase, which was also 4 h earlier than that for MDV1 (Fig. 8). l-valine accumulation and the biomass of both strains decreased between 60 and 64 h. Therefore, fermentation was terminated at 64 h.

Fig. 8
figure 8

Fed-batch fermentation of B. flavum strains MDV1 and MDVR2-21 in 30 L fermentor. The solid symbols represents the MDVR2-21 strain and the open symbols represents the MDV1 strain

Full size image

The reliable fed-batch fermentation performance of MDVR2-21 in the 30 L automatic fermentor prompted the exploration of its industrial-scale l-valine production in the 120 m3 fermentor. However, unlike the 30 L fermentor, a disk turbine agitator with six-arc blades is generally used in 120 m3 fermentor, and the dissolved oxygen level was controlled via agitation speed (100–200 rpm) adjustment to between 25 and 35% and an aeration rate of 2000–4000 m3/h was maintained. Other control parameters are the same as those for the 30 L fermentor. MDVR2-21 strain displayed similar growth curves in the 30 L automatic fermentor and 120 m3 industrial fermentor. Glucose consumption and cell growth were substantially synchronous (Figs. 8, 9). However, in the 120 m3 fermentor, the MDVR2-21 strain showed better environmental adaptation and growth characteristics than in the 30 L fermentor, and both the glucose consumption rate and the biomass growth rate were faster at 12–20 h than in the 30 L fermentor. After 20 h fermentation, cells entered the logarithmic phase with a sharp increase in the rate of glucose consumption which was 4 h earlier than that in the 30 L fermentor despite a lower biomass growth rate during this period. The cells reached the highest biomass at 44 h (OD562nm = 166.6) and the maximum DCW per liter was 25.2 g/L, which was 2.3 and 2.4% lower than that in the 30 L fermentor respectively. Subsequently, cell growth entered the stationary phase. In this phase, the concentration of l-valine in the fermentation broth also gradually increased (Fig. 9). After 64 h fermentation, the maximum l-valine production by MDVR2-21 was attained (66.8 g/L), with a glucose conversion rate of up to 36.5%. Both the l-valine yield and glucose conversion rate were lower than that obtained in the 30 L automatic fermentor. This was somewhat similar to our previous results (Huang et al. 2017). One reason could be that industrial fermentor microenvironments are unfavorable compared to those in the 30 L fermentor. For example, a relative lack of oxygen in the larger fermentor leads to unsynchronized cell growth and metabolism.

Fig. 9
figure 9

Fed-batch fermentation of B. flavum MDVR2-21 in 120 m3 fermentor

Full size image

Discussion

Improvement of strains that stably overproduce l-valine is important for their industrial-scale use. This goal is aided by knowledge of the l-valine biosynthetic pathway, including the key enzymes and their encoding genes. This knowledge has provided the groundwork for rational metabolic engineering approaches of strain improvement. However, the strict expectations of industrial strains in terms of high yields, high conversion rates, high specific growth rates, and low by-product concentrations means that metabolic engineering alone will not be successful for all strains. There are a variety of ways to improve l-valine yield from various strains. Of these, random mutagenesis is the most popular. l-valine production was greatly enhanced in B. flavum XQ-6 (Leul, Ilel, AHVr, α-ABhr, 2-TAhr) by mutagenesis. The highest l-valine production reached 67.7 g/L with a combinational regulation strategy (Liu and Zhang 2012). Another promising tool is genome shuffling. The best example of the effect of genome shuffling is the rapid improvement of tylosin production by Streptomyces fradiae (Zhang et al. 2002). After two rounds of genome shuffling in 1 year, the authors reported improvements that would have required 20 years using the recycle mutagenesis and selection approach. Moreover, the improvement was realized following the screening of 24,000 strains instead of the 1,000,000 strains required in the mutagenesis and selection approach. Even though genome shuffling is a relatively recent innovation, its prowess has been recognized in many fields (Patnaik et al. 2002; Stephanopoulos 2002; Zhang et al. 2002; Hida et al. 2007; Gong et al. 2009; Shi et al. 2009; Zhao et al. 2014). Presently, we successfully improved the l-valine production of strain B. flavum MDV1 by genome shuffling combined with mutagenesis. Compared to the traditional physical and chemical mutagenesis method, genome shuffling provides a safe way, and more efficacious, method of strain identification with significantly improved genotypes and phenotypes. This is achieved by eliminating the negative mutations in the process of setting up screening libraries, and overcomes the defect of classical mutagenesis.

In this study, the main challenge was to identify strains with improved l-valine yields from the regeneration medium. In the absence of an efficient bacterial biological indicator to identify cells capable of l-valine secretion, we needed to establish a new and efficient high-throughput screening method for l-valine. We took advantage of the recognized prowess of PC and HPLC to determine the yield of amino acids. PC is simple and we could easily use the technique to identify improved strains based on the size and color intensity of the resulting spots. For example, as shown in Fig. 7, the l-valine productivity of the mutant strains was obviously higher than that of other strains in the same generation. Additionally, 96-well plate were used to improve screening efficiency in this study. PC combined with 96-well plate fermentation is suitable for the screening of strains with enhanced l-valine production. It seems likely that this method will be useful for other amino acids.

In summary, we have demonstrated that genome shuffling enables rapid improvement of B. flavum strains concerning l-valine production. These results will be helpful for industrial l-valine production. Furthermore, to increase the yield of l-valine, improve industrial fermentation efficiency, and reduce cost, the use of comparative genomics and transcriptomics will help confirm that the increased expressions of these genes are correlated to l-valine biosynthesis, beneficial for further rationalization of metabolic engineering modifications, and also to provide additional targets for strain improvement.