Metabolic engineering of Bacillus amyloliquefaciens for poly-gamma-glutamic acid (γ-PGA) overproduction

We constructed a metabolically engineered glutamate-independent Bacillus amyloliquefaciens strain with considerable γ-PGA production. It was carried out by double-deletion of the cwlO gene and epsA-O cluster, as well as insertion of the vgb gene in the bacteria chromosome. The final generated strain NK-PV elicited the highest production of γ-PGA (5.12 g l−1), which was 63.2% higher than that of the wild-type NK-1 strain (3.14 g l−1). The γ-PGA purity also improved in the NK-PV strain of 80.4% compared with 76.8% for the control. Experiments on bacterial biofilm formation experiment showed that NK-1 and NK-c (ΔcwlO) strains can form biofilm; the epsA-O deletion NK-7 and NK-PV strains could only form an incomplete biofilm.


Introduction
Poly-γ-glutamic acid (γ-PGA) is an unusual homopolymer of D/L-glutamate acid polymerized through γ-glutamyl bonds (Ashiuchi and Misono, 2002). γ-PGA is biodegradable, edible, water-soluble and non-toxic toward humans and the environment. Therefore, it has attracted wide interests in a broad range of fields: food, medicine and water treatment (Shih and Van, 2001;Ashiuchi, 2013).
γ-PGA-producing strains are divided into two categories based on their requirement for exogenous glutamic acid: glutamate-dependent strains and glutamate-independent strains (Shih and Van, 2001). Compared with the glutamate-dependent strains, the independent strains are more preferable for the industrial production of γ-PGA with their low cost and simplified fermentation process (Cao et al., 2011). However, their lower γ-PGA productivity compared with glutamate-dependent strains limits their industrial application. Construction of a metabolically engineered glutamate-independent strain with high γ-PGA productivity is required.
Bacillus amyloliquefaciens NK-1 is a derivative of B. amyloliquefaciens LL3 (Cao et al., 2011). It is a glutamate-independent strain that produces γ-PGA comprising units of L-glutamic acid and D-glutamic acid. The endogenous plasmid pMC1 and upp gene of the NK-1 strain has been deleted, so it is suitable for genetic manipulation such as markerless deletion or insertion of genes.
The approaches employed for the enhancement of γ-PGA production has been mainly focused on two aspects: (i) optimization of the fermentation conditions (Bajaj and Singhal, 2009;Huang et al., 2011) and (ii) characterization of the genes involved in γ-PGA production and use of bioengineering methods to improve the γ-PGA yield (Ashiuchi et al., 2006;Scoffone et al., 2013;Zhang et al., 2013). Increasing numbers of studies have focused on the second approach to construct metabolically engineered microorganisms with high γ-PGA productivity. Heterologous expression of the γ-PGA synthetase complex (pgsBCA) is another strategy for γ-PGA production improvement, and it has been carried out in coryneform bacteria and Escherichia coli (Sung et al., 2005;Cao et al., 2013).
Exopolysaccharides (EPS) are synthesized primarily in microbial cells and then secreted into the extracellular environment as biofilms and capsules (Branda et al., 2001;Marvasi et al., 2010;Donot et al., 2012). Some Bacillus strains can produce abundant EPS and are essential for the formation of biofilms (Donot et al., 2012). The epsA-O cluster has been reported to be associated mainly with EPS production (Branda et al., 2004;, which are the main by-products of some γ-PGA-producing strains. Thus, deletion of epsA-O can depress EPS production, and metabolic flux could be used to enhance γ-PGA productivity. Vitreoscilla hemoglobin (VHb) is one of the most extensively studied bacterial hemoglobin molecules synthesized by the Gram-negative bacterium Vitreoscilla (Wakabayashi et al., 1986). It has been confirmed that VHb can interact directly with the terminal respiratory oxidase by delivering oxygen to it, in order to enhance oxidative phosphorylation, and thus the production of adenosine triphosphate (Park et al., 2002). It has been demonstrated that VHb expression in heterologous bacterial hosts can enhance cell density and metabolic production, especially under oxygen-limiting conditions (Kang et al., 2002;Urgun-Demirtas et al., 2003). Heterologous expression of the vgb gene (which encodes VHb) has been conducted in various microorganisms as well as plant and mammal cells (Zhang et al., 2007). During the fermentation process of γ-PGA, broth viscosity increased and dissolved oxygen in the broth reached 0% after 12-18 h of incubation, which decreased oxygen transfer to cells, inhibited cell growth and γ-PGA production. Expression of the vgb gene in a γ-PGAproducing strain will alleviate oxygen limitation at the later stage of fermentation to enhance cell density and γ-PGA production.
In the present study, we aimed to construct a highproductivity γ-PGA-producing strain based on the glutamic acid-independent B. amyloliquefaciens NK-1.
We double-deleted the cwlO gene and epsA-O cluster and simultaneously inserted the vgb gene in the bacterial chromosome. The generated engineered strain NK-PV showed higher production of γ-PGA and γ-PGA purity compared with the wild-type NK-1 strain.

Identification of target gene knockouts or insertion
To construct a metabolically engineered, γ-PGA highproducing strain, double-deletion of the cwlO gene and epsA-O cluster and insertion of a vgb gene were carried out in the B. amyloliquefaciens NK-1 strain. A gene marker-less knockout method was used to construct the gene deletion and insertion mutant, which was based on an upp cassette and the 5-FU selection (Keller et al., 2009).
The primers cwlO-SS/XX, AO-SS/XX and Amy-SS/XX were used to verify the gene mutants ( Fig. 1). DNA sequencing was used to confirm construction of the engineered strain.

Confirmation of expression of the vgb gene and its effect on cell growth
It has been reported that VHb expression enhances the cell density and yield of metabolite production (Olano et al., 2008). Here, we inserted the vgb gene in the AmyA locus of the NK-7 chromosome. To identify expression of the vgb gene, RT-PCR and carbon monoxide (CO)difference spectra were used. The vgb gene transcriptional levels of the NK-PV strain were tested: It was transcribed successfully ( Fig. 2A). VHb has been reported to bind with CO and to show a characteristic absorption peak at 420 nm (Liu and Webster, 1974). CO-difference Lane M, DNA marker III; Lane L, DNA ladder (1 kb); Lane 1, 2, gene deletion or insertion strains amplified with relative primers; Lane NK-1, wild-type NK-1 strains amplified with relative primers. spectra of the NK-PV strain showed an absorption peak at approximately 420 nm after 36 h of induction, but this peak was absent in the NK-1 strain (Fig. 2B). These results suggested that the mutant NK-PV strain bearing the vgb gene could express functional VHb.
To confirm the effect of VHb on cell growth, the growth curve of B. amyloliquefaciens NK-1 and mutant strains were monitored by measuring OD600. The growth rate of NK-PV was higher than that of other strains without the vgb gene (Fig. 3A). The dry cell weight of the NK-PV strain was also higher than that of the other mutant strains. These results suggested that VHb expression increased cell growth and dry cell weight in our strain.
Comparison of the results of γ-PGA fermentation among B. amyloliquefaciens NK-1 and mutant strains by flask culture To evaluate accumulation of gene-targeted B. amyloliquefaciens mutants on γ-PGA production, flask culture of B. amyloliquefaciens NK-1 and gene mutant strains NK-c, NK-7 and NK-PV was undertaken under identical conditions. The results of γ-PGA fermentation among B. amyloliquefaciens NK-1 and the other mutant strains are shown in Fig. 3B and Table 1. The γ-PGA yield from NK-c was approximately 26.2% higher than that of the NK-1 strain. The molecular weight was also higher than that of the NK-1 strain. These results were in accordance with a study carried out in the glutamate-dependent strain, in which deletion of cwlO gene increased not only the γ-PGA yield but also its molecular weight (Mitsui et al., 2011). γ-PGA yields of the NK-7 strain increased by 41.5% compared with that of the NK-1 strain, and by 12.1% compared with that of the NK-c strain. The NK-PV strain showed the highest production of γ-PGA, leading to a yield of 5.12 g l −1 , which was 63.2% higher than that of the NK-1 strain (3.14 g l −1 ). The molecular weight of γ-PGA in NK-7 and NK-PV strains was also higher than that of the NK-1 strain ( Table 1).
The viscosity of the culture broth of the NK-1 strain and other mutant strains was 10.05, 16.70, 19.40 and 31.57 (cP) respectively (Table 1). Our previous studies have shown that a higher concentration or higher molecular weight of γ-PGA leads to a more viscous broth. In the present study, the molecular weight of γ-PGA in gene mutant strains was similar whereas the γ-PGA yield changed considerably. The strain with the higher concentration of γ-PGA made the broth more viscous. Thus, even though several factors affect broth viscosity, the viscosity of the culture broth in these fours strains was mainly determined by its γ-PGA component.
We further determined the fermentation results between NK-1, NK-c, NK-PV, NK-c-HB and NK-PV-HB strains. γ-PGA yield and molecular weight of NK-1 strain were 3.4 g l −1 and 391.5 K respectively. The γ-PGA yields of cwlO gene complementary strains NK-c-HB (3.32 g l −1 ) and NK-PV-HB (3.73 g l −1 ) are all lower than their corresponding gene deletion strains NK-c (4.35 g l −1 ) and NK-PV (5.33 g l −1 ). Moreover, molecular weights of NK-c-HB and NK-PV-HB strains are 379.7 k and 401.2 k, which were all lower than NK-c (455.6 k) and NK-PV (440.1 k). A. RT-PCR results between NK-1 and NK-PV strain. Lane M, DNA marker III; Lane 1-4, amplification results using primers VG-F and VG-R with RNA from NK-1 strain, RNA from NK-PV strain, cDNA from NK-1 strain and cDNA from NK-PV strain as template respectively. B. CO-spectra difference analysis of NK-1 and NK-PV strain.

Effect of epsA-O deletion on biofilm formation
Microbes can construct structurally complex biofilms through production of an extracellular matrix. Studies have shown that a wild-type strain of the Gram-positive bacterium B. subtilis can build such a matrix. Some genetic, biochemical and cytological evidence have suggested that the matrix is composed predominantly of a protein component (TasA) and an EPS component, and that absence of TasA or the EPS results in a residual matrix, and that absence of both components leads to complete failure of biofilm formation (Branda et al., 2004;. The epsA-O operon has been reported to encode biosynthetic machinery to produce EPS (Branda et al., 2004;Kearns et al., 2005). Once the epsA-O cluster is disrupted, Bacillus strains cannot secrete EPS, which greatly affects biofilm formation. In this work, the epsA-O cluster was deleted and its affect on biofilm formation determined. The NK-1 strain and NK-c strain could form a complete pellicle (Fig. 4). The NK-7 strain and NK-PV strain could only form a deficient incomplete pellicle.

Effect of epsA-O deletion on γ-PGA purity and colony morphology
Some EPS can also be deposited by cold ethanol and, upon mixing with γ-PGA, decrease the purity of the γ-PGA product. To increase the purity of γ-PGA, we knocked out  the epsA-O cluster. The purities of γ-PGA in the epsA-O cluster-deleted strains NK-7 and NK-PV were 81.8% and 80.4% respectively, which were higher than those of NK-1 and NK-c strains (76.8% and 75.7% respectively). The colony morphologies of NK-1 and gene mutant strains are shown in Fig. 5. The colonies of different strains were similar, only colonies of the NK-PV strain showed a slightly flatter and larger appearance than the other strains.

Discussion
Most γ-PGA-producing Bacillus strains are unsuitable for genetic manipulation because of their low or absent genetic competence (Sung et al., 2005). Thus, few metabolic engineering strategies have been reported to improve γ-PGA production. The temperature-sensitive shuttle plasmid pKSV7 and upp counter selection-based marker-less deletion method has been used successfully for the manipulation of multiple genes (Smith and Youngman, 1992;Keller et al., 2009). In this work, we deleted the cwlO and epsA-O clusters and inserted a vgb gene in the NK-1 chromosome based on this method. The final metabolically engineered B. amyloliquefaciens NK-PV strain showed increased production of γ-PGA as well as increased purity. Certain studies have focused on improving γ-PGA production by deletion of its degrading enzyme genes, but the ggt and pgdS gene deletion mutant strain produced similar amounts of γ-PGA to the wild-type strain (Kimura et al., 2004;Mitsui et al., 2011). This may be because GGT and PgdS are regulated appropriately in vivo and they function only if nitrogen in the medium is limited. Mitsui and colleagues (2011) investigated the effects of the cell wall lytic enzymes, LytE, LytF, CwlO and CwlS, on γ-PGA production in B. subtilis (natto). They found that Fig. 4. Cell pellicle formation between NK-1 and mutant strains. Cells were cultured at 30°C for 80 h in MSgg broth contained within a 6-well microtiter dish. disruption of the lytE, lytF and cwlS genes had little effect on γ-PGA production, whereas γ-PGA levels in the cwlO mutant were approximately twofold higher than that of the wild-type. Furthermore, they found that the cwlO deletion strain showed a higher molecular weight of γ-PGA. cwlO deletion stops γ-PGA being hydrolysed by CwlO, and its absence results in a higher yield and molecular weight of γ-PGA. Domínguez-Cuevas and colleagues (2013) reported that the cwlO gene deletion strain exhibits a shorter and wider cell size, resulting in a higher specific surface area that could increase the γ-PGA transport and thereafter the production of γ-PGA. In the present study, the cwlO gene deletion strain NK-c showed a higher yield and molecular weight of γ-PGA compared with the NK-1 strain, which was consistent with our previous work (Feng et al., 2014) and that of Mitsui and colleagues (2011); moreover, the cwlO gene complementary NK-c-HB and NK-PV-HB strains exhibited lower yield and molecular weight of γ-PGA. These results indicated that CwlO indeed related to the γ-PGA hydrolyse. No report has demonstrated that deletion of the epsA-O cluster or expression of the vgb gene increases the molecular weight of γ-PGA, so the higher molecular weights of γ-PGA in NK-c, NK-7 and NK-PV strains must be related mainly to deletion of the cwlO gene.
Bacteria can produce a wide variety of EPS such as curdlan (Shih et al., 2009), alginate (Celik et al., 2008), xanthan (Kalogiannis et al., 2003) and levan (Shih et al., 2010). The biosynthesis of EPS involves consumption of a large amount of the carbon source, which decreases the metabolic flux used for γ-PGA production. Moreover, EPS are secreted into the medium and interact with γ-PGA, thereby influencing the purity of γ-PGA. Liu and colleagues (2011) depressed EPS production in B. amyloliquefaciens C06, which caused overproduction of its γ-PGA. We deleted the epsA-O cluster based on the NK-c strain: The generated NK-7 strain showed a higher yield and purity of γ-PGA. Disruption of the epsA-O cluster blocks the pathway of EPS synthesis, which reduce the production of extracellular products as well as enables the metabolic flux to be used more efficiently for γ-PGA production. Moreover, defects in EPS production would reduce spatial competition and cell membrane permeability pressure, which would enable intracellular γ-PGA to be transported more efficiently and to accumulate outside cells. Thus, the epsA-O deletion strain showed good properties for γ-PGA production.
In Bacillus strains, EPS and TasA are two major components of the extracellular matrix that hold differentiated cell chains together to form a highly organized colony and pellicle architecture. Branda and colleagues (2006) demonstrated that single deletion of the eps cluster can result in incomplete formation of biofilms, and that doubledeletion of eps and tasA ensure that the strain fails to make a pellicle. In the present study, the wild-type strain and cwlO deletion strain could form complete pellicles. In accordance with the previous study, the epsA-O deletion strains NK-7 and NK-PV could form only an incomplete pellicle (Branda et al., 2006). However, biofilm formed by NK-PV strain seems even more incomplete than NK-7 strain (Fig. 4). To test whether the expression of VHb influences bacteria biofilm formation, we determined the biofilm formation in two vgb gene expression strains NK-1-vgb and NK-c-vgb. Form Fig. 4, we can find that the two strains can all form complete pellicle like NK-1 and NK-c strains. These results indicated that VHb expression has little effect on bacteria bioflim formation, and it is not efficient to affect biofilm formation in the wild-type strains.
VHb has been used widely for the enhancement of cell density and target products, especially in highly viscous media, in which oxygen availability is limited (Frey and Kallio, 2003;Su et al., 2010). γ-PGA synthesis results in extreme viscosity of the culture medium, which seriously limit its oxygen content. p43 is a strong and wellcharacterized promoter in B. subtilis (Wang and Dio, 1984), which is assumed to express genes efficiently in other Bacillus genus stains. We expressed the vgb gene under the p43 promoter in the present study. However, γ-PGA in the NK-PV strain increased by only 15.3% compared with the NK-7 strain. This finding could be due to several reasons. VHb elicits its effects under oxygenlimited conditions (Zhang et al., 2007;Su et al., 2010). However, the oxygen-limited condition in NK-PV fermentation appeared after 12-18 h of incubation, which reduced the time in which VHb could elicit its effects. It has been reported that VHb is more likely to elicit its effects, for example, in a greater medium volume or at a lower rotation speed. However, the γ-PGA yields under those conditions are lower than those under optimal conditions (Zhang et al., 2007;Su et al., 2010). Therefore, under our cultivation conditions, the limitation of oxygen transfer may be less severe than that in other strains.
In summary, a metabolically engineered high γ-PGA productivity strain was created in a glutamateindependent γ-PGA-producing strain. We double-deleted the cwlO gene and epsA-O cluster and inserted the vgb gene in the NK-1 strain chromosome. The final generated NK-PV strain showed the highest γ-PGA production, which was 63.2% higher than that of the wild-type NK-1 strain; moreover, the γ-PGA purity in NK-PV strain were also higher than that of the NK-1 strain.

Strains, plasmids and growth conditions
All the strains and plasmids used in this work are listed in Table 2. Escherichia coli DH5α was used for the propagation and transformation of plasmids. Demethylation of plasmids Construct a strain with enhanced γ-PGA production 451 was carried out in E. coli GM2163. The p43-vgb gene fragment was isolated from the pKSPVK plasmid . The oligonucleotide primers used in this study are listed in Table 3.

DNA manipulation and construction of mutant strains
The temperature-sensitive plasmid p-upp was used to construct the epsA-O deletion vector. The upstream and downstream fragments of epsA-O were amplified by PrimeSTAR HS DNA Polymerase (Takara Bio, Tokyo, Japan) using primers AO-SF/AO-SR and AO-XF/AO-XR respectively. The two regions of DNA were joined by overlap-polymerase chain reaction (PCR). The generated fragment was restricted by SalI and BamHI, and cloned into the p-upp vector using the same enzymes to create the deletion plasmid p-upp-ΔAO. The vgb gene combined with the p43 promoter was inserted into the amylase gene AmyA region of the chromosome. To construct the vgb insertion vector, the p43-vgb fragment was obtained from the plasmid pKSPVK  by primers PV-F/PV-R, and the upstream and downstream regions were amplified using primers Amy-SF/Amy-SR and Amy-XF/Amy-XR respectively. These three regions of DNA fragments were joined by overlap-PCR, and the generated fragment cloned into the pMD19-T simple vector for sequencing. The generated plasmid T-SPVX was digested with BamHI and SalI, and the fragment cloned into the p-upp vector to generate the insertion vector p-upp-spvx.
Gene knockout mutant strains were constructed by a gene marker-less deletion method based on the upp cassette (Keller et al., 2009). The cwlO deletion strain NK-c has been constructed in our previous work (Feng et al., 2014). p-upp-ΔAO and p-upp-spvx plasmids were first introduced into Shuttle vector, temperature-sensitive (ts) replication origin, Amp r (gram-negative)and Cm r (gram-positive) (Smith and Youngman, 1992) pMD19-T T easy vector for gene cloning; Amp r Takara pKSPVK pKSV7 derivative consists of LL3 16S rRNA gene, promoter p43, vgb, and Kan r for selection  p-upp pKSV7-derivation with upp gene This lab p-upp-ΔAO pKSV7-derivation with upp gene and deletion fragment of epsA-O This work p-upp-ΔcwlO pKSV7-derivation with upp gene and the deletion of fragment cwlO (Feng et al., 2014) p-upp-spvx pKSV7-derivation with upp gene and insertion fragment p43-vgb This work pWHV pWH1520 derivative, vgb expression vector coding for VHb  pWH1520-cwlO pWH1520 derivative carrying the structural gene cwlO (Feng et al., 2014)  E. coli GM2163 for demethylation. Plasmids isolated from E. coli GM2163 were treated with BamHI methyltransferase before being transformed into target strains . Cells were incubated on LB agar plates with 5 μg ml −1 chloramphenicol. Single colonies were picked, and primers AO-SS/AO-ZZ and Amy-SS/Amy-XX used to determine single-cross clones by PCR. Selected single-cross clones were then incubated in LB medium, supplemented with 100 μg ml −1 5-FU, for 24 h at 42°C. Cells were diluted 10 5 times and spread on LB agar plates with 5-FU. Single colonies were picked, and primers AO-SS/AO-XX and Amy-SS/ Amy-XX used to determine the gene mutant clones by PCR.

Reverse transcription-polymerase chain reaction (RT-PCR) analyses
RT-PCR was undertaken to test expression of the vgb gene between NK-1 and NK-PV strains. Cells were harvested for RNA extraction after 16 h of cultivation in LB medium. The commercial RNApure Bacteria kit (Cwbio, Beijing, China) was used to extract total RNA. cDNA was obtained by reverse transcriptional synthesis with a HiFi-MMLV cDNA kit (Cwbio). Primers VG-F and VG-R were used to determine gene transcription.

Carbon monoxide-difference spectral analysis
VHb activity was detected by CO-difference spectra. Hemoglobin can react with CO. The complex comprising CO and reduced hemoglobin has a characteristic peak at 420 nm (Choi et al., 2003;Zhang et al., 2013). B. amyloliquefaciens NK-1 and B. amyloliquefaciens NK-PV were harvested by centrifugation at 8000 r.p.m. for 20 min at 4°C after 36-h cultivation in γ-PGA fermentation medium. Cell pellets were washed with 0.1 M potassium phosphate buffer (pH 7.2) thrice. Cells were re-suspended in 20 ml of the same buffer, and were broken with a sonicator on ice (600 W for 20 min with cycles of sonication of 30 s each and 30 s pause). The crude extract was centrifuged at 8000 r.p.m. for 20 min at 4°C to remove cell debris. Samples were divided into two aliquots. One was exposed to CO for 2 min and the other to air. Then, hemoglobin levels were obtained by CO-difference spectra using a UV-1800 spectrophotometer (Shimadzu, Kyoto, Japan).

Analyses of formation of colonies and biofilms
Bacteria were grown on LB agar for 24 h and then incubated in LB medium to an optical density at 600 nm (OD600) of 1.0. For analyses of colony architecture, 2 μl of starting culture were spotted onto LB agar plates for 24 h at 37°C. For analyses of pellicle formation, 10 μl of starting culture were added to 10 ml of MSgg broth (Branda et al., 2006) contained within a six-well microtiter dish, and the dish incubated for 80 h at 30°C without agitation. For inducible expression of vgb gene, 0.5% xylose was added to MSgg medium. All colonies and pellicles were photographed using a digital camera equipped with a close-up lens (Canon, Tokyo, Japan).
Analytical procedures γ-PGA was purified by a method described previously (Kubota et al., 1993). The OD of the fermentation broth was measured by the UV-1800 spectrophotometer (Shimadzu). The molecular weight of γ-PGA was measured by a gel permeation chromatography system (Cao et al., 2010). Relative viscosity was measured by a DV-I Digital Rheometer (Brookfield, Middleborough, MA, USA) fitted with a spindle S00 code at the shear rate of 10 r.p.m. (25°C). Whole polysaccharide content (%) in the γ-PGA product was measured by the phenol-sulfuric acid method (DuBois et al., 1951). The purity of γ-PGA (%) was defined as 1-polysaccharide content (%).