Improving Surfactin Production in Bacillus subtilis 168 by Metabolic Engineering

Surfactin is widely used in the petroleum extraction, cosmetics, biopharmaceuticals and agriculture industries. It possesses antibacterial and antiviral activities and can reduce interfacial tension. Bacillus are commonly used as production chassis, but wild-type Bacillus subtilis 168 cannot synthesise surfactin. In this study, the phosphopantetheinyl transferase (PPTase) gene sfp* (with a T base removed) was overexpressed and enzyme activity was restored, enabling B. subtilis 168 to synthesise surfactin with a yield of 747.5 ± 6.5 mg/L. Knocking out ppsD and yvkC did not enhance surfactin synthesis. Overexpression of predicted surfactin transporter gene yfiS increased its titre to 1060.7 ± 89.4 mg/L, while overexpression of yerP, ycxA and ycxA-efp had little or negative effects on surfactin synthesis, suggesting YfiS is involved in surfactin efflux. By replacing the native promoter of the srfA operon encoding surfactin synthase with three promoters, surfactin synthesis was significantly reduced. However, knockout of the global transcriptional regulator gene codY enhanced the surfactin titre to 1601.8 ± 91.9 mg/L. The highest surfactin titre reached 3.89 ± 0.07 g/L, with the yield of 0.63 ± 0.02 g/g DCW, after 36 h of fed-batch fermentation in 5 L fermenter. This study provides a reference for further understanding surfactin synthesis and constructing microbial cell factories.


Introduction
Surfactin is a lipopeptide biosurfactant with a cyclic structure composed of seven amino acid residues (both L-and D-amino acids) connected to a long β-hydroxyl fatty acid chain (usually C13−C15) by a lactone bond, making it an amphiphilic molecule [1][2][3].Due to its strong ability to reduce interfacial tension between oil and water, surfactin is used to reduce the emulsification and viscosity of crude oil for enhanced oil recovery in the petroleum extraction industry [4][5][6].Moreover, it can be effectively applied in cosmetics in the form of solvents, soaps and facial cleansers due to its emulsification, washing, foaming, solubilisation, wetting, penetration, dispersion and low toxicity characteristics [7,8].The long-chain fatty acid of surfactin can penetrate into the interior of cell membranes, causing dissolution and damage, which is the basis for its antibacterial activity [9,10].Surfactin can also break down viral lipid membranes, and it has antiviral properties [11].Therefore, it can be applied as a biological preservative to effectively control the growth of harmful microorganisms in food [12], and it may prevent fungal infections and skin diseases [13].In addition, it has broad application prospects in the agricultural field [14], such as controlling apple scab [15].
Currently, the main production methods for surfactin include chemical synthesis and microbial fermentation, with the latter favoured due to superior environmental friendliness and biocompatibility.However, low productivity of surfactin remains an important limiting factor in its industrial production [16,17].For instance, combinatorial metabolic engineering of Bacillus subtilis ATCC 21332, including enhanced nitrate reduction, fatty acid hydroxylation, rational transporter engineering and feeding, yielded 14.4 g/L of surfactin, with the productivity of 0.6 g/L/h, the highest reported productivity to date [18].Therefore, the development of microbial cell factories is crucial for its overproduction.Surfactin was first discovered in the fermentation broth of B. subtilis in 1968, with four isomers identified [19].Moreover, B. subtilis, a model Gram-positive bacterium, grows fast and is easy to cultivate.It has excellent expression systems with good genetic stability, excellent protein secretion ability, no strong codon preferences and mature gene modification technologies.It has been recognised by the USA Food and Drug Administration (FDA) as a 'Generally Recognised as Safe' (GRAS) strain [20].Therefore, B. subtilis is commonly used as a chassis for producing surfactin through metabolic engineering [18,21,22].
In B. subtilis, the biosynthesis of surfactin mainly involves the synthesis and structural assembly of precursors (fatty acids and amino acids).B. subtilis synthesises the corresponding precursors through primary metabolism, such as glycolysis, the tricarboxylic acid cycle, protein metabolism, lipid metabolism and other pathways (Figure 1).The assembly of surfactin is then catalysed by a non-ribosomal peptide synthase (NRPS) system, which includes initiation, peptide chain extension, cyclisation and release [23,24].The NRPS, also known as surfactin synthase, is a multi-domain protein composed of multiple modules, including the SrfAA, SrfAB, SrfAC and SrfAD subunits.They are encoded by srfAA, srfAB, srfAC and srfAD, respectively, together forming the 27 kb srfA operon [24].Transcription of the srfA operon is regulated by the quorum sensing system ComQXPA, and phosphorylated ComA binds to a specific region of its promoter, activating its transcription [25][26][27].In addition, downstream of the srfA operon, there is a phosphopantetheinyl transferase (PPTase) gene sfp that is essential for surfactin synthesis, which functions to activate surfactin synthase [28].However, sfp is an inactive pseudogene in wild-type B. subtilis 168, which cannot synthesise surfactin.Wu et al. (2019) integrated the sfp gene from Bacillus amyloliquefaciens MT45 into the genome of B. subtilis 168 at the same locus, resulting in a surfactin titre of 400 mg/L [21].Subsequently, biofilm formation-related genes were knocked out, potential self-resistance associated proteins were overexpressed, the branched-chain fatty acid biosynthesis pathway was engineered, comQXPA from B. amyloliquefaciens MT45 was overexpressed and the global transcriptional regulator gene codY was knocked out, resulting in a surfactin titre of 12.8 g/L [21].Wang et al. (2019) integrated sfp from B. amyloliquefaciens DSM7 into the ydeO (encoding an unknown protein) locus of the genome of B. subtilis 168, resulting in a surfactin titre of 450 mg/L [29].The clustered regularly interspersed short palindromic repeat interference (CRISPRi) technology was used to inhibit the expression of L-glutamate racemase genes yrpC and racE, resulting in a surfactin titre of 750 mg/L [29].Hu et al. (2020) integrated sfp from Bacillus velezensis BS-37 into the genome of B. subtilis 168 at the same locus, resulting in a surfactin titre of 982 ± 98 mg/L [30].Medium-chain acyl-acyl carrier protein (ACP) thioesterase (encoded by bte) and fatty acyl CoA ligase (encoded by yhfL) were overexpressed, and acyl-CoA dehydrogenase gene fadE was knocked out, resulting in a surfactin titre of 2203 mg/L, with xylose as the carbon source [30].However, there have been no studies on the overexpression of endogenous active sfp in B. subtilis 168.In addition, the efflux of surfactin is an important factor limiting its synthesis.Overexpression of three putative lipopeptide transporters genes: yerP (endogenous), ycxA and krsE (heterologous) in B. subtilis THY-7 resulted in a 145%, 89% and 52% increase in surfactin production, respectively [31], and overexpression of endogenous yerP in B. subtilis 168 resulted in a 35.3% increase in surfactin production [21], demonstrating that YerP acts as the major surfactin exporter.Research by Xia et al., (2024) showed that preprotein translocase SecA, signal recognition particle receptor FtsY and cell division ATP-binding protein FtsE are also involved in the transport of surfactin in B. subtilis ATCC 21332 [18].The present study used B. subtilis 168 as the parent strain.To facilitate gene knockout or integrated expression [32], strain BS168N integrated with Para-neo on the genome was employed.Firstly, the excess T base of endogenous sfp was removed, and the modified gene was integrated into the genome of BS168N for overexpression.The resulting recombinant strain was fermented and surfactin synthesis assessed.Next, plipastatin synthase gene ppsD and phosphotransferase gene yvkC were knocked out to reduce the consumption of amino acids, hydroxy fatty acids and ATP, and their effects on surfactin synthesis were investigated.Thirdly, endogenous yerP, ycxA and predicted surfactin transporter gene yfiS were integrated into the same locus of the genome for overexpression, and their effects on surfactin secretion were evaluated.Fourthly, the native promoter of the srfA operon was replaced in situ by constitutive promoters, and codY was knocked out, so their impacts on the transcriptional level of the srfA operon and surfactin synthesis were explored.Finally, the final recombinant strain was subjected to fed-batch fermentation, and strain growth and surfactin production were measured.

Microorganisms, Media and Culture Conditions
Bacillus subtilis 168 (BS168) and BS168N were provided by Tianjin University.All recombinant B. subtilis strains were constructed based on BS168N.The strains and plasmids used in this study are listed in Table 1.Luria-Bertani (LB) medium (10 g/L tryptone, 5 g/L The present study used B. subtilis 168 as the parent strain.To facilitate gene knockout or integrated expression [32], strain BS168N integrated with P ara -neo on the genome was employed.Firstly, the excess T base of endogenous sfp was removed, and the modified gene was integrated into the genome of BS168N for overexpression.The resulting recombinant strain was fermented and surfactin synthesis assessed.Next, plipastatin synthase gene ppsD and phosphotransferase gene yvkC were knocked out to reduce the consumption of amino acids, hydroxy fatty acids and ATP, and their effects on surfactin synthesis were investigated.Thirdly, endogenous yerP, ycxA and predicted surfactin transporter gene yfiS were integrated into the same locus of the genome for overexpression, and their effects on surfactin secretion were evaluated.Fourthly, the native promoter of the srfA operon was replaced in situ by constitutive promoters, and codY was knocked out, so their impacts on the transcriptional level of the srfA operon and surfactin synthesis were explored.Finally, the final recombinant strain was subjected to fed-batch fermentation, and strain growth and surfactin production were measured.

Microorganisms, Media and Culture Conditions
Bacillus subtilis 168 (BS168) and BS168N were provided by Tianjin University.All recombinant B. subtilis strains were constructed based on BS168N.The strains and plasmids used in this study are listed in Table 1.Luria-Bertani (LB) medium (10 g/L tryptone, 5 g/L yeast extract and 10 g/L NaCl) was used for strain cultivation.After adding the fragments to the competent cells, they were coated on LBC8 solid plates (added with 15 g/L of agar and 8 µg/mL of chloramphenicol to LB liquid medium) for preliminary screening of the transformants.After culturing the transformants in LB liquid medium for 4 h, they were coated on LBN16 solid plates (added with 15 g/L of agar and 16 µg/mL of neomycin to LB liquid medium) for secondary screening of the recombinant bacteria.A medium (50 mL, containing 25 g/L glucose, 10 g/L tryptone, 3 g/L beef extract, 1 g/L KH 2 PO 4 , 1 g/L K 2 HPO 4 , 1 g/L NaCl, 0.5 g/L (NH 4 ) 2 SO 4 , 0.2 g/L L-leucine, 0.05 g/L sodium glutamate, 0.05 g/L MgSO 4 •7H 2 O, 0.05 g/L MnSO 4 •H 2 O and 7.5 mg/L FeSO 4 •7H 2 O, pH 7.0) was used for the fermentation experiments.Each B. subtilis strain was transferred into a 250 mL flask with 30 mL LB liquid medium and incubated at 37 • C with shaking at 200 rpm for 12 h as the seed culture.Next, 2.5 mL of seed culture was inoculated into 250 mL flasks containing 50 mL fermentation medium and incubated at 37 • C with shaking at 220 rpm for 48 h.Amp R , ampicillin resistance; *, base T 496 in the sfp gene sequence was removed.

Construction of B. subtilis Knockout Mutants
For the knockout of yrpC, the 5039 bp UDCRG fragment was amplified from the genome of BSD1-∆yrpCm [33] using primer pair yrpC-U1/yrpC-G2 (Table 2).U refers to the upstream homologous arm, G refers to the downstream homologous arm, D refers to the homologous recombinant fragment, C refers to the fragment containing the chloramphenicol resistance gene cat and R refers to the fragment containing gene araR encoding the transcriptional repressor of the arabinose operon [32].AraR could inhibit the transcription of P ara -neo, resulting in the strain not exhibiting neomycin resistance but exhibiting chloramphenicol resistance [32].This UDCRG fragment was used to transform competent cells of BS168N as previously described [35], and the cells were cultured on chloramphenicol resistance plates.The yrpC knockout strain BSSF1 was obtained by a two-step screening process as previously described [32].
For the knockout of yvkC, the 5259 bp UDCRG fragment was amplified from the genome of BSD1-∆yvkCm [33] using primer pair yvkC-U1/yvkC-G2.This UDCRG fragment was used to transform competent cells of BSSF3.The resulting yrpC knockout strain BSSF4 was obtained by a two-step screening process.
For the knockout of codY, the 1116 bp U, 1092 bp D and 605 bp G fragments were amplified from the genome of BS168 using primer pairs codY-U1/codY-U2, codY-D1q/codY-D2 and codY-G1q/codY-G2, respectively.A 2189 bp CR fragment was amplified from BSD1-∆yrpCm using primer pair codY-CR1q/codY-CR2.These four fragments were spliced in the order of U-D-CR-G by two rounds of OE-PCR.Finally, this UDCRG fragment was used to transform competent cells of BSSF52.The resulting codY knockout strain BSSF64 was obtained by a two-step screening process.
For overexpression of yfiS, a 1399 bp fragment of UP, a 1332 bp fragment of yfiS containing the complete nucleotide sequence of yfiS and a 3954 bp fragment of DCRG were amplified from the genome of BSSF51m using primer pairs yvkC-U1/yvkC-P2, yfiS-1q/yfiS-2 and yvkC-yfiS-D1q/yvkC-yfiS-G2, respectively.These three fragments were spliced in the order of UP-yfiS-DCRG using primer pair yvkC-U1/yvkC-yfiS-G2. Finally, this fragment was used to transform competent cells of BSSF3.The resulting strain BSSF52 overexpressing yfiS was obtained by a two-step screening process.
For overexpression of ycxA, a 1399 bp fragment of UP, a 1489 bp fragment of ycxA containing the complete nucleotide sequence and terminator sequence of ycxA and a 3847 bp fragment of DCRG were amplified from the genome of BSSF51m using primer pairs yvkC-U1/yvkC-P2, ycxA-1q/ycxA-2 and yvkC-ycxA-D1q/yvkC-yfiS-G2, respectively.These three fragments were spliced in the order of UP-ycxA-DCRG using primer pair yvkC-U1/yvkC-yfiS-G2. Finally, this fragment was used to transform competent cells of BSSF3.The resulting strain BSSF53 overexpressing ycxA was obtained by a two-step screening process.
For overexpression of ycxA-efp, a 2631 bp fragment of UPA containing the upstream homologous arm U, the P 43 promoter the complete nucleotide sequence of ycxA, a 689 bp fragment of efp containing the nucleotide sequence and terminator sequence and a 3872 bp fragment of DCRG were amplified from the genome of BSSF53m using primer pairs yvkC-U1/2-ycxA-2, efp-1q/efp-2 and yvkC-efp-D1q/yvkC-yfiS-G2, respectively.These three fragments were spliced in the order of UPA-efp-DCRG using primer pair yvkC-U1/yvkC-yfiS-G2. Finally, this fragment was used to transform competent cells of BSSF3.The resulting strain BSSF54 overexpressing ycxA-efp was obtained by a two-step screening process.

In Situ Substitution of the Native Promoter of the srfA Operon
The native promoter of the srfA operon was replaced in situ by the P HpaII promoter.First, 1346 bp U, 850 bp C and 820 bp D fragments were amplified from the genome of BSD1-∆yrpCm using primer pairs srf-U1/srf-2, srf-C1q/srf-C2 and srf-D1q/srf-D2, respectively.A 274 bp P1 fragment containing the constitutive P HpaII promoter was amplified from plasmid pMA5 using primer pair srf-P1-1q/srf-P1-2.These four fragments were spliced in the order of U-C-P1-D using primer pair srf-U1/srf-D2.Finally, this fragment was used to transform competent cells of BSSF52.The resulting strain BSSF61 was obtained by a one-step screening process.
The native promoter of the srfA operon was replaced in situ by the P 43 promoter.First, 2196 bp UC and 820 bp D fragments were amplified from the genome of BSSF61 using primer pairs srf-U1/2-srf-C2 and 2-srf-D1q/srf-D2, respectively.A 267 bp P2 fragment containing the P 43 promoter was amplified from the genome of BS∆6-AD1m using primer pair srf-P2-1q/srf-P2-2.These three fragments were spliced in the order of UC-P2-D using primer pair srf-U1/srf-D2.Finally, this fragment was used to transform competent cells of BSSF52.The resulting strain BSSF62 was obtained by a one-step screening process.
The native promoter of the srfA operon was replaced in situ by the P SB expression cassette.First, 2196 bp UC and 820 bp D fragments were amplified from the genome of BSSF61 using primer pairs srf-U1/2-srf-C2 and 3-srf-D1q/srf-D2, respectively.A 128 bp P3 fragment containing the constitutive promoter P SB expression cassette [38] was amplified from plasmid pUC57-simple-PyrG E156K [33] using primer pair srf-P3-1q/srf-P3-2.These three fragments were spliced in the order of UC-P3-D using primer pair srf-U1/srf-D2.Finally, this fragment was used to transform competent cells of BSSF52.The resulting strain BSSF63 was obtained by a one-step screening process.

Extraction and Detection of Surfactin
The surfactin standard was purchased from Sigma-Aldrich (St. Louis, MO, USA).A 20 mL volume of fermentation broth was centrifuged at 8000 rpm for 10 min, and the pH of the supernatant was adjusted to 2.0 with HCl.After incubating overnight at 4 • C, centrifuging at 8000 rpm for 10 min and discarding the supernatant, the precipitate was dried at room temperature, 5 mL of methanol was added to dissolve the precipitate and the sample was filtered through a 0.22 µm organic filter.The concentration of surfactin was measured using an LC-2030 reversed-phase high-performance liquid chromatography (HPLC) system (Shimadzu, Kyoto, Japan) equipped with a ZORBAX SB-C18 column (250 mm × 4.6 mm, 5 µm; Agilent, Santa Clara, CA, USA) and a UV detector (Shimadzu) at 205 nm.The column temperature was set to 30 • C. The mobile phase consisted of methanol/water/trifluoroacetic acid solution (90:10:0.1,v/v/v) and the flow rate was 1.0 mL/min.

Quantitative Real-Time PCR (qRT-PCR)
A 1 mL volume of fermentation broth was collected after 24 h of fermentation, and the total RNA was extracted using a Bacterial Total RNA Rapid Extraction Kit (HLingene, Shanghai, China).Reverse-transcription was performed using a NG Script I cDNA Synthesis Kit (HLingene) to obtain the cDNA libraries.qRT-PCR was carried out using a LightCycler 480 instrument (Roche, Mannheim, Germany) using the 2× SYBR Green qPCR Mixture (HLingene).The relative transcriptional levels were calculated according to Pfaffl [39], with the carbon catabolite control protein A gene ccpA serving as an internal control.

Fed-Batch Fermentation
Fed-batch fermentation was conducted in a 5 L fermenter at the National Center of Bio-Engineering & Technology (Shanghai, China).A single colony of strain BSFF64 was inoculated into 5 mL LB medium and cultured at 37 • C, with shaking at 200 rpm overnight.The preculture was transferred into 100 mL fresh fermentation medium and cultured at 37 • C and 220 rpm for 12 h.Next, 100 mL seed culture was transferred to 1.9 L fermentation medium in a 5 L bioreactor, and fermentation was carried out at 37 • C with the pH maintained at 7.0 using NaOH (1 mol/L) and HCl (1 mol/L).The airflow was set at 2 vvm, with agitation at 300−500 rpm.The feed solution containing 240 g/L glucose, 30 g/L tryptone, 25 g/L beef extract, 5 g/L L-leucine and 75 mg/L FeSO4•7H2O was added automatically to maintain the dissolved oxygen (DO) at 40−50%.Duplicate samples were collected to determine the cell density and surfactin titre.

Statistical Analysis
Analysis of the response variables was performed using SPSS statistics 17.0 [40].One-way analysis of variance (ANOVA) and Tukey's test based on p < 0.05 were performed to determine the significance of the differences in response variables between strains.

Overexpression of Active PPTase to Endow B. subtilis 168 with the Ability to Synthesise Surfactin
PPTase encoded by sfp is responsible for connecting the long arm of coenzyme A (CoA), 4 ′ -phosphopantothenamine, to the conserved serine residue in the peptidyl carrier protein (PCP), which converts PCP from an inactive apo-form to active holo-form, thereby initiating surfactin synthesis [41].Due to the presence of an additional T 496 base in the nucleotide sequence of sfp in B. subtilis 168, its transcription is terminated prematurely (Figure S1).Therefore, sfp is an inactive pseudogene in B. subtilis 168.Indeed, no surfactin was detected in the supernatant of its fermentation broth (Figure 2A and Figure S2A).To make the PPTase active, we designed primers to amplify the preceding and following nucleotide sequences of the sfp gene, removing the T 496 base in the process.These two nucleotide sequences were spliced together through overlapping PCR, then spliced with the strong TP2 promoter [36] to integrate into the yrpC locus of the genome of strain BS168N, yielding recombinant strain BSSF2 (Table 1).In addition, we constructed yrpC gene knockout strain BSSF1 as a control for BSSF2.In B. subtilis, yrpC encodes glutamate racemase, which catalyses the racemisation of L-glutamate to produce D-glutamate [42].L-glutamate is one of the precursors for synthesising surfactin, so the knockout of yrpC should reduce the consumption of L-glutamate.Moreover, Wang et al. [29] inhibited the expression of yrpC using CRISPRi technology, which indeed promoted surfactin synthesis.

Effects of Plipastatin Synthetase and Phosphotransferase Deficiency on Surfactin Synthesis
In B. subtilis, the ppsABCDE operon encodes plipastatin synthase that catalyses the synthesis of plipastatin from ten amino acids, including L-glutamate, D-alanine and Ltyrosine, along with hydroxyl fatty acids [43].To increase the flow of amino acids and hydroxyl fatty acids towards surfactin synthesis, the ppsD gene in the ppsABCDE operon was knocked out, resulting in knockout strain BSSF3.The fermentation results showed that there was almost no change in surfactin production (Figure 3).Coutte et al. [44] inserted the spectacomycin resistance gene spc into ppsA to disrupt plipastatin synthesis in B. subtilis, resulting in a 80.7% increase in surfactin production after 24 h of fermentation in a Landy medium and a 2.9% decrease after 48 h of fermentation.The above results indicated that the effect of pps disruption on surfactin production was related to the fermentation medium and fermentation time.
In B. subtilis 168, YvkC may consume ATP and catalyse the synthesis of pyruvate to phosphoenolpyruvate.In addition, the yvkC gene encodes flavonoid phosphate synthetase, which catalyses the ATP-dependent phosphorylation of flavonoids to generate flavonoid monophosphates, AMP and orthophosphate [45].To block gluconeogenesis and flavonoid monophosphate synthesis and reduce ATP consumption, the yvkC gene in the genome of BSSF3 was knocked out, resulting in knockout strain BSSF4.The fermentation results indicated that this had no significant positive effect on the synthesis of surfactin (Figure 3).Although surfactin synthesis requires ATP, the weaker activity of YvkC may result in less ATP consumption by its catalytic reaction and its knockout having no significant positive effect on surfactin synthesis.After 48 h of fermentation, strain BSSF2 produced 747.5 ± 6.5 mg/L of surfactin (Figures 2A and S2B).There were four components of the surfactin standards, corresponding to peaks 1, 2, 3 and 4 (Figure S2C,D).The differences between them were the amino acid residues at positions 2, 4 and 7 of the circular peptide (leucine, isoleucine or valine) and the length of the carbon chain (C13 to C15) [31].The relative transcriptional level of sfp in BSSF2 was increased 17.8-fold relative to BSSF1 (Figure 2B).Compared to previous studies [21,29,30], overexpression of PPTase led to different levels of surfactin production, possibly due to the varying activities or expression levels of PPTases from different sources.The amino acid sequences of PPTase from B. amyloliquefaciens MT45 (GenBank: ASF27580.1),B. amyloliquefaciens DSM7 (GenBank: CBI41443.1),B. velezensis BS-37 (Gen-Bank: AWG39808.1)and the active PPTase from B. subtilis 168 are not completely identical, as shown in the alignment (Figure S3).The above results indicated that PPTase was the key enzyme for B. subtilis to synthesise surfactin, and the recovery of endogenous PPTase activity endowed B. subtilis 168 with the ability to produce surfactin.

Effects of Plipastatin Synthetase and Phosphotransferase Deficiency on Surfactin Synthesis
In B. subtilis, the ppsABCDE operon encodes plipastatin synthase that catalyses the synthesis of plipastatin from ten amino acids, including L-glutamate, D-alanine and L-tyrosine, along with hydroxyl fatty acids [43].To increase the flow of amino acids and hydroxyl fatty acids towards surfactin synthesis, the ppsD gene in the ppsABCDE operon was knocked out, resulting in knockout strain BSSF3.The fermentation results showed that there was almost no change in surfactin production (Figure 3).Coutte et al. [44] inserted the spectacomycin resistance gene spc into ppsA to disrupt plipastatin synthesis in B. subtilis, resulting in a 80.7% increase in surfactin production after 24 h of fermentation in a Landy medium and a 2.9% decrease after 48 h of fermentation.The above results indicated that the effect of pps disruption on surfactin production was related to the fermentation medium and fermentation time.

Overexpression and Identification of a Surfactin Transporter to Enhance Surfactin Synthesis
Since surfactin accumulates excessively in cells, it may disrupt the integrity of cell membranes [31], thereby affecting normal cell function and even leading to cell death, which is obviously disadvantageous for the sustained synthesis of surfactin.Therefore, promoting surfactin secretion may improve surfactin production.In B. subtilis 168, yerP (also known as swrC) encodes a protein similar to the acriflavin resistance protein that is involved in surfactin efflux [46], and the ycxA gene, located downstream of the srfA operon, encodes a protein homologous to members of the major facilitator superfamily (MFS) [47].In Bacillus thuringiensis, the krsE gene encodes an efflux protein involved in lipopeptide efflux [48].The amino acid sequence of KrsE (GenBank: AIG20548.1) of B. thuringiensis was BLAST in the B. subtilis genome database, and it was found that YfiS shares the highest homology with KrsE (Figure S4).In B. subtilis 168, yfiS encodes a protein similar to multidrug resistance proteins.
The endogenous genes yerP, yfiS and ycxA were separately integrated into the yvkC locus of the BSSF3 genome under the control of the P43 promoter [37], resulting in strains BSSF51, BSSF52 and BSSF53.After 48 h of fermentation, their surfactin production reached 809.3 ± 24.8 mg/L, 1060.7 ± 89.4 mg/L and 798.7 ± 61.4 mg/L, respectively, increased by 4.9%, 37.4% and 3.5%, respectively, compared to that of control strain BSSF4 (771.7 ± 17.3 mg/L; Figure 4A).Moreover, the growth of strain BSSF52 overexpressing yfiS was better than that of control strain BSSF4 and recombinant strains BSSF51 and BSSF53, during the first 24 h of fermentation (Figure S5C).The qRT-PCR results showed that adding one copy number to the genome resulted in a 0.70-fold (Figure 4B), 1.09-fold (Figure 4C) and 2.88fold (Figure 4D) increase in the relative transcriptional levels of mRNAs of genes yerP, yfiS and ycxA, respectively.Overexpressing ycxA resulted in the largest increase in the transcriptional level, but its promotion effect on surfactin synthesis was the weakest.Considering that the translation of ycxA is likely to require elongation factor P (Efp) due to the presence of several consecutive proline residues [49], we constructed artificial operon P43-ycxA-efp and integrated it into the yvkC locus of the BSSF3 genome to obtain strain BSSF54.Although the relative transcriptional level of efp was increased by 1.81-fold (Figure 4E), co-expression of ycxA and efp resulted in a 46.3% decrease in surfactin production compared to control strain BSSF53 overexpressing ycxA alone (Figure 4F) and a slight decrease in strain growth (Figure S5D).The above results indicated that YfiS was mainly responsible for surfactin efflux under the culture conditions of this study.In B. subtilis 168, YvkC may consume ATP and catalyse the synthesis of pyruvate to phosphoenolpyruvate.In addition, the yvkC gene encodes flavonoid phosphate synthetase, which catalyses the ATP-dependent phosphorylation of flavonoids to generate flavonoid monophosphates, AMP and orthophosphate [45].To block gluconeogenesis and flavonoid monophosphate synthesis and reduce ATP consumption, the yvkC gene in the genome of BSSF3 was knocked out, resulting in knockout strain BSSF4.The fermentation results indicated that this had no significant positive effect on the synthesis of surfactin (Figure 3).Although surfactin synthesis requires ATP, the weaker activity of YvkC may result in less ATP consumption by its catalytic reaction and its knockout having no significant positive effect on surfactin synthesis.

Overexpression and Identification of a Surfactin Transporter to Enhance Surfactin Synthesis
Since surfactin accumulates excessively in cells, it may disrupt the integrity of cell membranes [31], thereby affecting normal cell function and even leading to cell death, which is obviously disadvantageous for the sustained synthesis of surfactin.Therefore, promoting surfactin secretion may improve surfactin production.In B. subtilis 168, yerP (also known as swrC) encodes a protein similar to the acriflavin resistance protein that is involved in surfactin efflux [46], and the ycxA gene, located downstream of the srfA operon, encodes a protein homologous to members of the major facilitator superfamily (MFS) [47].In Bacillus thuringiensis, the krsE gene encodes an efflux protein involved in lipopeptide efflux [48].The amino acid sequence of KrsE (GenBank: AIG20548.1) of B. thuringiensis was BLAST in the B. subtilis genome database, and it was found that YfiS shares the highest homology with KrsE (Figure S4).In B. subtilis 168, yfiS encodes a protein similar to multidrug resistance proteins.
The endogenous genes yerP, yfiS and ycxA were separately integrated into the yvkC locus of the BSSF3 genome under the control of the P 43 promoter [37], resulting in strains BSSF51, BSSF52 and BSSF53.After 48 h of fermentation, their surfactin production reached 809.3 ± 24.8 mg/L, 1060.7 ± 89.4 mg/L and 798.7 ± 61.4 mg/L, respectively, increased by 4.9%, 37.4% and 3.5%, respectively, compared to that of control strain BSSF4 (771.7 ± 17.3 mg/L; Figure 4A).Moreover, the growth of strain BSSF52 overexpressing yfiS was better than that of control strain BSSF4 and recombinant strains BSSF51 and BSSF53, during the first 24 h of fermentation (Figure S5C).The qRT-PCR results showed that adding one copy number to the genome resulted in a 0.70-fold (Figure 4B), 1.09-fold (Figure 4C) and 2.88-fold (Figure 4D) increase in the relative transcriptional levels of mRNAs of genes yerP, yfiS and ycxA, respectively.Overexpressing ycxA resulted in the largest increase in the transcriptional level, but its promotion effect on surfactin synthesis was the weakest.Considering that the translation of ycxA is likely to require elongation factor P (Efp) due to the presence of several consecutive proline residues [49], we constructed artificial operon P 43 -ycxA-efp and integrated it into the yvkC locus of the BSSF3 genome to obtain strain BSSF54.Although the relative transcriptional level of efp was increased by 1.81-fold (Figure 4E), co-expression of ycxA and efp resulted in a 46.3% decrease in surfactin production compared to control strain BSSF53 overexpressing ycxA alone (Figure 4F) and a slight decrease in strain growth (Figure S5D).The above results indicated that YfiS was mainly responsible for surfactin efflux under the culture conditions of this study.(B) transcriptional level of yerP in strain BSSF51 relative to in control strain BSSF4 (defined as 1); (C) transcriptional level of yfiS in strain BSSF52 relative to that in strain BSSF4 (defined as 1); (D) transcriptional level of ycxA in strain BSSF53 relative to that in strain BSSF4 (defined as 1); (E) transcriptional level of efp in strain BSSF54 relative to that in control strain BSSF53 (defined as 1); (F) surfactin titre and growth of BSSF53 and BSSF54 after 48 h of fermentation.Significant differences were determined by SPSS statistics 17.0 based on p < 0.01 (**) and p < 0.001 (***).

Enhancing Transcription of the srfA Operon to Promote Surfactin Synthesis
The nucleotide sequence of the srfA operon is up to 27 kb, making it difficult to increase its transcriptional level through overexpression.However, the promoter can determine the (B) transcriptional level of yerP in strain BSSF51 relative to that in control strain BSSF4 (defined as 1); (C) transcriptional level of yfiS in strain BSSF52 relative to that in strain BSSF4 (defined as 1); (D) transcriptional level of ycxA in strain BSSF53 relative to that in strain BSSF4 (defined as 1); (E) transcriptional level of efp in strain BSSF54 relative to that in control strain BSSF53 (defined as 1); (F) surfactin titre and growth of BSSF53 and BSSF54 after 48 h of fermentation.Significant differences were determined by SPSS statistics 17.0 based on p < 0.01 (**) and p < 0.001 (***).

Enhancing Transcription of the srfA Operon to Promote Surfactin Synthesis
The nucleotide sequence of the srfA operon is up to 27 kb, making it difficult to increase its transcriptional level through overexpression.However, the promoter can determine the gene transcriptional level, so we attempted replacing the native promoter of the srfA operon.To this end, we used constitutive promoters P HpaII , P 43 and P SB [38] to replace the native promoter of the srfA operon in the genome of strain BSSF52, resulting in recombinant strains BSSF61, BSSF62 and BSSF63.The fermentation results showed that the surfactin production was only 16.5 ± 1.3 mg/L, 3.1 ± 0.2 mg/L and 7.1 ± 0.1 mg/L, respectively (Figure 5A), decreased by 98.4%, 99.7% and 99.3% compared to BSSF52.The qRT-PCR results showed that the relative transcriptional levels of srfAA and srfAB in the srfA operon decreased by varying degrees (Figure 5B).This was consistent with the previous research by Willenbacher et al. [50] and Jiao et al. [51].The surfactin production of B. subtilis DSM 10T was decreased to 0.04 g/L from 0.62 g/L when the P veg promoter replaced the promoter of the srfA operon [50].The ability of B. subtilis THY-7 to synthesise surfactin was lost when the P groE promoter induced by L-arabinose was used as a replacement [51].Therefore, the initial transcription effect of these three promoters was not as strong as that of the native promoter of the srfA operon.
gene transcriptional level, so we attempted replacing the native promoter of the srfA operon.To this end, we used constitutive promoters PHpaII, P43 and PSB [38] to replace the native promoter of the srfA operon in the genome of strain BSSF52, resulting in recombinant strains BSSF61, BSSF62 and BSSF63.The fermentation results showed that the surfactin production was only 16.5 ± 1.3 mg/L, 3.1 ± 0.2 mg/L and 7.1 ± 0.1 mg/L, respectively (Figure 5A), decreased by 98.4%, 99.7% and 99.3% compared to BSSF52.The qRT-PCR results showed that the relative transcriptional levels of srfAA and srfAB in the srfA operon decreased by varying degrees (Figure 5B).This was consistent with the previous research by Willenbacher et al. [50] and Jiao et al. [51].The surfactin production of B. subtilis DSM 10T was decreased to 0.04 g/L from 0.62 g/L when the Pveg promoter replaced the promoter of the srfA operon [50].The ability of B. subtilis THY-7 to synthesise surfactin was lost when the PgroE promoter induced by L-arabinose was used as a replacement [51].Therefore, the initial transcription effect of these three promoters was not as strong as that of the native promoter of the srfA operon.
The global regulatory factor CodY recognises GTP and branched-chain amino acids (BCAAs) as specific signals and affects the expression of >100 genes associated with metabolism [52].CodY binds to the promoter of the srfA operon, occupying the binding site between DNA and positive regulatory proteins, leading to inhibition of the transcription of this operon [53].We knocked out the codY gene of strain BSSF52 to generate knockout strain BSSF64.Its surfactin titre was 1601.8 ± 91.9 mg/L (Figure 5C), 51% higher than that of control strain BSSF52, as in accordance with a previous report, in which this knockout resulted in a 17.3% increase [21].The relative transcriptional levels of srfAA and srfAB increased by 1.31fold and 0.97-fold, respectively (Figure 5D).This indicated that the knockout of codY could indeed alleviate the partial regulation of the srfA operon, thereby increasing its transcription and promoting the expression of surfactin synthase, significantly improving surfactin synthesis.The global regulatory factor CodY recognises GTP and branched-chain amino acids (BCAAs) as specific signals and affects the expression of >100 genes associated with metabolism [52].CodY binds to the promoter of the srfA operon, occupying the binding site between DNA and positive regulatory proteins, leading to inhibition of the transcription of this operon [53].We knocked out the codY gene of strain BSSF52 to generate knockout strain BSSF64.Its surfactin titre was 1601.8 ± 91.9 mg/L (Figure 5C), 51% higher than that of control strain BSSF52, as in accordance with a previous report, in which this knockout resulted in a 17.3% increase [21].The relative transcriptional levels of srfAA and srfAB increased by 1.31-fold and 0.97-fold, respectively (Figure 5D).This indicated that the knockout of codY could indeed alleviate the partial regulation of the srfA operon, thereby increasing its transcription and promoting the expression of surfactin synthase, significantly improving surfactin synthesis.

Increasing Surfactin Production through Fed-Batch Fermentation
To improve the fermentation level of surfactin, we performed fed-batch fermentation of recombinant strain BSSF64 in a 5 L fermenter.After 6 h of fermentation, a large number of bubbles were evident, and the addition of soybean oil and polyether defoamer had no defoaming effect.Hence, fermentation could not proceed normally.A defoaming effect could only be achieved by adding organosilicon defoamer.After 36 h of fermentation, the maximum OD 600 value of BSSF64 was 11.54 ± 1.83 (Figure 6A).Although its growth was almost twice that of flask fermentation, it was far below the growth achieved when using engineered B. subtilis for fed-batch fermentation to produce other high value-added chemicals [34,54].The surfactin titres (yields) after fermentation for 12, 24, 36 and 48 h were 0.94 ± 0.15 g/L (0.22 ± 0.01 g/g DCW), 1.94 g/L (0.25 ± 0.01 g/g DCW), 3.89 ± 0.07 g/L (0.63 ± 0.02 g/g DCW) and 3.35 ± 0.17 g/L (0.52 ± 0.04 g/g DCW), respectively (Figure 6B).This might be due to inappropriate fermentation conditions leading to poor growth or to the toxicity of surfactin or organosilicon defoamer toward cells, which may have inhibited bacterial growth and surfactin synthesis.By optimising the fermentation medium and conditions alone, the OD 600 of B. subtilis ATCC 21332 reached 60, and surfactin production was increased from 335 mg/L to 5.3 g/L [18].Therefore, optimising the fermentation conditions, including carbon sources, nitrogen sources, pH, inoculum size and metal ions by adding Fe nanoparticles [55] or EDTA-Fe 2+ [56], may be crucial for improving surfactin production in the future.Regarding the foaming phenomenon, a two-compartment biofilm bioreactor [57] or a foam trap [58] could be designed to avoid the addition of organosilicon defoamer.

Increasing Surfactin Production through Fed-Batch Fermentation
To improve the fermentation level of surfactin, we performed fed-batch fermentati of recombinant strain BSSF64 in a 5 L fermenter.After 6 h of fermentation, a large numb of bubbles were evident, and the addition of soybean oil and polyether defoamer had defoaming effect.Hence, fermentation could not proceed normally.A defoaming effe could only be achieved by adding organosilicon defoamer.After 36 h of fermentation, t maximum OD600 value of BSSF64 was 11.54 ± 1.83 (Figure 6A).Although its growth w almost twice that of flask fermentation, it was far below the growth achieved when usi engineered B. subtilis for fed-batch fermentation to produce other high value-added chem icals [34,54].The surfactin titres (yields) after fermentation for 12, 24, 36 and 48 h were 0. ± 0.15 g/L (0.22 ± 0.01 g/g DCW), 1.94 g/L (0.25 ± 0.01 g/g DCW), 3.89 ± 0.07 g/L (0.63 ± 0. g/g DCW) and 3.35 ± 0.17 g/L (0.52 ± 0.04 g/g DCW), respectively (Figure 6B).This mig be due to inappropriate fermentation conditions leading to poor growth or to the toxici of surfactin or organosilicon defoamer toward cells, which may have inhibited bacter growth and surfactin synthesis.By optimising the fermentation medium and conditio alone, the OD600 of B. subtilis ATCC 21332 reached 60, and surfactin production was i creased from 335 mg/L to 5.3 g/L [18].Therefore, optimising the fermentation condition including carbon sources, nitrogen sources, pH, inoculum size and metal ions by addi Fe nanoparticles [55] or EDTA-Fe 2+ [56], may be crucial for improving surfactin producti in the future.Regarding the foaming phenomenon, a two-compartment biofilm bioreact [57] or a foam trap [58] could be designed to avoid the addition of organosilicon defoam

Conclusions
The wild-type B. subtilis 168 strain cannot synthesise surfactin due to the presence a frameshift mutation in the nucleotide sequence of sfp encoding PPTase, resulting in inactivation.This study first removed the T496 base and integrated the overexpression sfp*, which restored the activity of PPTase and increased the surfactin titre from zero 747.5 ± 6.5 mg/L.The knockout of ppsD and yvkC had a minimal effect on surfactin sy thesis.Overexpression of yerP, yfiS, ycxA and ycxA-efp indicated that YfiS might be t main surfactin transporter, and its overexpression increased surfactin production 37.4% compared to the control strain.Replacing the native promoter of the srfA oper with three constitutive promoters significantly reduced the ability to synthesise surfact with yields ranging from 3.1 ± 0.2 mg/L to 16.5 ± 1.3 mg/L.However, the knockout of cod

Conclusions
The wild-type B. subtilis 168 strain cannot synthesise surfactin due to the presence of a frameshift mutation in the nucleotide sequence of sfp encoding PPTase, resulting in its inactivation.This study first removed the T 496 base and integrated the overexpression of sfp*, which restored the activity of PPTase and increased the surfactin titre from zero to 747.5 ± 6.5 mg/L.The knockout of ppsD and yvkC had a minimal effect on surfactin synthesis.Overexpression of yerP, yfiS, ycxA and ycxA-efp indicated that YfiS might be the main surfactin transporter, and its overexpression increased surfactin production by 37.4% compared to the control strain.Replacing the native promoter of the srfA operon with three constitutive promoters significantly reduced the ability to synthesise surfactin, with yields ranging from 3.1 ± 0.2 mg/L to 16.5 ± 1.3 mg/L.However, the knockout of codY increased the surfactin titre by 51% to 1601.8 ± 91.9 mg/L.Finally, fed-batch fermentation was carried out in a 5 L fermenter without the need for antibiotics and inducers, and the highest titre of surfactin was 3.89 ± 0.07 g/L, with a yield of 0.63 ± 0.02 g/g DCW.Although the liquid production of surfactin was relatively low, production by dry weight was higher, indicating that engineered B. subtilis has the potential to become a cell factory for producing surfactin.

Figure 2 .
Figure 2. Surfactin titre and growth of wild-type Bacillus subtilis 168 (BS168), parent strain BS168N, yrpC gene knockout strain BSSF1 and sfp gene overexpression strain BSSF2, and the relative transcriptional levels of the sfp gene.(A) Surfactin titre and growth of the BS168, BS168N, BSSF1 and BSSF2 strains after 48 h of fermentation; (B) transcriptional level of sfp in strain BSSF2 relative to that in control strain BSSF1 (defined as 1).Significant differences were determined by SPSS statistics 17.0 based on p < 0.05 (*) and p < 0.001 (***).

Figure 2 .
Figure 2. Surfactin titre and growth of wild-type Bacillus subtilis 168 (BS168), parent strain BS168N, yrpC gene knockout strain BSSF1 and sfp gene overexpression strain BSSF2, and the relative transcriptional levels of the sfp gene.(A) Surfactin titre and growth of the BS168, BS168N, BSSF1 and BSSF2 strains after 48 h of fermentation; (B) transcriptional level of sfp in strain BSSF2 relative to that in control strain BSSF1 (defined as 1).Significant differences were determined by SPSS statistics 17.0 based on p < 0.05 (*) and p < 0.001 (***).

Figure 3 .
Figure 3. Surfactin titre and growth of control strain BSSF2, pssD gene knockout strain BSSF3 and yvkC gene knockout strain BSSF4 after 48 h of fermentation.

Figure 4 .
Figure 4. Surfactin titre, growth and relative transcriptional level of control strain BSSF4, and recombinant strains BSSF51 overexpressing yerP, BSSF52 overexpressing yfiS, BSSF53 overexpressing ycxA and BSSF54 overexpressing ycxA-efp.(A) Surfactin titre and growth of BSSF4, BSSF51, BSSF52 and BSSF53 after 48 h of fermentation;(B) transcriptional level of yerP in strain BSSF51 relative to in control strain BSSF4 (defined as 1); (C) transcriptional level of yfiS in strain BSSF52 relative to that in strain BSSF4 (defined as 1); (D) transcriptional level of ycxA in strain BSSF53 relative to that in strain BSSF4 (defined as 1); (E) transcriptional level of efp in strain BSSF54 relative to that in control strain BSSF53 (defined as 1); (F) surfactin titre and growth of BSSF53 and BSSF54 after 48 h of fermentation.Significant differences were determined by SPSS statistics 17.0 based on p < 0.01 (**) and p < 0.001 (***).

Figure 4 .
Figure 4. Surfactin titre, growth and relative transcriptional level of control strain BSSF4, and recombinant strains BSSF51 overexpressing yerP, BSSF52 overexpressing yfiS, BSSF53 overexpressing ycxA and BSSF54 overexpressing ycxA-efp.(A) Surfactin titre and growth of BSSF4, BSSF51, BSSF52 and BSSF53 after 48 h of fermentation;(B) transcriptional level of yerP in strain BSSF51 relative to that in control strain BSSF4 (defined as 1); (C) transcriptional level of yfiS in strain BSSF52 relative to that in strain BSSF4 (defined as 1); (D) transcriptional level of ycxA in strain BSSF53 relative to that in strain BSSF4 (defined as 1); (E) transcriptional level of efp in strain BSSF54 relative to that in control strain BSSF53 (defined as 1); (F) surfactin titre and growth of BSSF53 and BSSF54 after 48 h of fermentation.Significant differences were determined by SPSS statistics 17.0 based on p < 0.01 (**) and p < 0.001 (***).

Figure 5 .Figure 5 .
Figure 5. Surfactin titre, growth and relative transcriptional level of control strain BSSF52 and strains BSSF61, BSSF62 and BSSF63 obtained by replacing the native promoter of the srfA operon with promoters PHpaII, P43 and PSB, respectively, and codY knockout strain BSSF64.(A) Surfactin titre and growth of BSSF52, BSSF61, BSSF62 and BSSF63 after 48 h of fermentation; (B) transcriptional levels of srfAA Figure 5. Surfactin titre, growth and relative transcriptional level of control strain BSSF52 and strains BSSF61, BSSF62 and BSSF63 obtained by replacing the native promoter of the srfA operon with promoters P HpaII , P 43 and P SB , respectively, and codY knockout strain BSSF64.(A) Surfactin titre and growth of BSSF52, BSSF61, BSSF62 and BSSF63 after 48 h of fermentation; (B) transcriptional levels of srfAA and srfAB in strains BSSF61, BSSF62 and BSSF63 relative to those in BSSF52 (defined as 1); (C) surfactin titre and growth of BSSF52 and BSSF64 after 48 h of fermentation; (D) transcriptional levels of srfAA and srfAB in strain BSSF64 relative to those in strain BSSF52 (defined as 1).Significant differences were determined by SPSS statistics 17.0 based on p < 0.05 (*), p < 0.01 (**) and p < 0.001 (***).

Figure 6 .
Figure 6.Growth curve, surfactin titre and yield of the final recombinant strain BSSF64 during fe batch fermentation.(A) Growth curve of BSSF64 during fed-batch fermentations in a 5 L ferment (B) surfactin titre and yield of BSSF64 during fed-batch fermentations in a 5 L fermenter.

Figure 6 .
Figure 6.Growth curve, surfactin titre and yield of the final recombinant strain BSSF64 during fedbatch fermentation.(A) Growth curve of BSSF64 during fed-batch fermentations in a 5 L fermenter; (B) surfactin titre and yield of BSSF64 during fed-batch fermentations in a 5 L fermenter.

Table 1 .
Bacterial strains and plasmids used in this study.

Table 2 .
Sequences of the primers used in this study.