Improving the Yield of Xenocoumacin 1 by P BAD Promoter Replacement in Xenorhabdus nematophila CB6

: Xenocoumacin 1 (Xcn1), which is produced by Xenorhabdus nematophila CB6, exhibits strong inhibition activity against plant pathogens, especially fungi and oomycetes. Therefore, it has attracted interest in developing it into a novel biofungicide applicable for plant protection. However, its low yield with concomitant high cost during the fermentation process limits its widespread application. In this study, we replaced the native promoter of xcnA with the arabinose-inducible araBAD promoter ( P BAD ), a well-known and widely used promoter for expressing heterologous genes, to evaluate its effects on Xcn1 yield and antimicrobial activity. Compared with wildtype strain, the fermentation yield of Xcn1 was improved from 68.5 mg/L to 249.7 mg/L (3.6-fold) and 234.9 mg/L (3.4-fold) at 0.5% and 1.0% L-arabinose concentration, respectively. We further explored the transcription level of the biosynthesis related genes of Xcn1 and found that their upregulation resulted in the yield improvement of Xcn1. Moreover, the antimicrobial activity of Xcn1 against Bacillus subtilis and Phytophthora capsici was determined by agar diffusion plate and growth inhibition assay, as expected, it was also found to be enhanced. The promoter-replacement strategy utilized here improves the yield of Xcn1 efﬁciently, which provides a basis for the industrial production of Xcn1.


Introduction
Xenorhabdus nematophila, forming a mutual association with the specific soil-dwelling nematode Steinernema carpocapsae, is a type of Gram-negative bacteria of the Morganellaceae family [1][2][3][4][5]. This type of bacteria colonizes in a specialized intestinal receptacle of their cognate nematode partners when the nematodes are in the infective juvenile (IJ) form. As an entomopathogenic complex, they can invade an insect's body cavity (hemocoel) [2,[6][7][8]. After entering the hemocoel, the nematode releases X. nematophila; then, the bacteria secrete different compounds, including immunosuppressive compounds and toxins, to overcome the host immune system and kill it [9][10][11][12][13][14]. With the quick proliferation of X. nematophila, they also secrete a large amount of exoenzymes for degrading insect tissue, and both X. nematophila and the digested macromolecules derived from the insect host could be utilized by the nematode as nutrient resources for proliferation [15][16][17][18][19][20][21]. Finally, the nematodes develop into the IJ stage, and, together with X. nematophila, they emerge from the insect cadaver and search for a new insect host [3,4,6,22].
Xcn1 exhibits a strong antifungal activity against Alternaria alternata, Botrytis cinerea, Rhizoctonia solani, and Phytophthora species, etc. [26,36,39,40]. Xcn2, however, shows significantly decreased bioactivities [26,37,40]. Phytophthora infestans is one of the most important plant pathogens, which can cause potato late blight disease, the most devastating disease during potato production [26]. It is reported that Xcn1 exhibited strong inhibition effects against the mycelial growth of P. infestans, reaching 100% inhibition at 1.5 µg/mL. Moreover, it also effectively inhibited other species of Phytophthora, with EC 50 values below 4.2 µg/mL [26,39,40]. Therefore, Xcn1 has great potential to be used as a new biofungicide for plant protection. However, the low fermentation yield leading to high cost hampers the practical application of Xcn1 in the agricultural field.
While it is a challenging task, different strategies have been employed to improve the production of Xcn1. Yang et al. tried to improve the yield of Xcn1 by physical and chemical mutagenesis methods, the classical strategy for strain improvement, and several positive mutants were obtained, with inhibition activity enhanced in the range of 60-70% [41,42]. Fermentation optimization is another traditional strategy for increasing the titer of microbial secondary metabolites. The effects of medium components, parameters of fermentation process, such as initial pH, inoculum size, temperature, speed of rotation, etc., and feeding precursor on the yield of Xcn1 were systematically investigated, the highest yield reaching 173.99 µg/mL under the optimal fermentation conditions [35,[43][44][45]. In situ separation technology (ISPR), which could remove Xcn1 from the fermentation broth, therefore preventing the conversion of Xcn1 to Xcn2, was reported to improve the yield of Xcn1 from 42.5 to 73.8 µg/mL in three days [46]. Furthermore, regulators, particularly the global regulators, related to the production of Xcn1 were studied. It was established that the global regulators OmpR, LeuO, and CpxR were negative regulators of Xcn1, negatively regulating the transcription of xcnA-L, while Lrp and FliZ positively regulated the production of Xcn1 [4,28,47].
As described above, despite multiple efforts were attempted to increase the yield of Xcn1, it still cannot meet the requirement of industrial production. Therefore, new strategies for engineering the production strain, which might complement previous endeavors, should be attempted. Replacement of the natural promoter of a gene cluster of interest with a strong constitutive promoter or an inducible promoter whose regulation mechanism is well understood could be an effective strategy to improve the yield of the aimed secondary metabolite [48,49]. It has been reported that the well-known P BAD promoter was used successfully to improve the production of secondary metabolites in the entomopathogenic bacteria Photorhabdus luminescens and Xenorhabdus doucetiae [49]. Hence, we speculate that replacement of the original promoter of xcnA with the P BAD promoter could be an useful strategy to improve the yield of Xcn1. Therefore, in this study, we constructed a ∆P BAD -xcnA mutant based on X. nematophila CB6 and studied its effects on Xcn1 production and antimicrobial activity.

Materials and Methods
2.1. Bacterial Strains, Media, and Growth Conditions X. nematophila CB6 was isolated from the entomopathogenic nematode Steinernema sp. screened from a soil sample in Beijing, China. Details of the strains and plasmids used in this study are provided in Table 1. X. nematophila CB6 and E. coli were cultured with Luria-Bertani (LB) broth (Bacto tryptone 10 g/L, yeast extract 5 g/L, and NaCl 10 g/L) or LB agar media (15 g/L agar). X. nematophila CB6 was cultured at 28 • C, and E. coli was grown at 37 • C. Unless noted otherwise, strains were initially grown overnight in 3 mL of LB medium, then inoculated into 100 mL of fresh LB medium in a 250 mL conical flask and incubated for the desired period. When screening the X. nematophila CB6 mutants of single crossover homologous recombination, ampicillin and gentamicin were added to the medium with a final concentration of 100 and 50 µg/mL, respectively. An LB agar plate (no NaCl) containing 50 µg/mL of kanamycin and 5% sucrose was applied for screening the second crossover homologous recombination mutants. Bacillus subtilis was grown at 37 • C on an LB medium, and Phytophthora capsici was grown at 25 • C on potato dextrose agar (PDA).

Construction of ∆P BAD -xcnA Mutant Strain
To construct the ∆P BAD -xcnA mutant strain by homologous recombination, upstream (859 bp) and downstream (853 bp, beginning with the start codon ATG of xcnA gene) fragments were amplified from CB6 chromosomal DNA with primer pairs XcnA-5DT/XcnA-P1 and XcnA-3DT/XcnA-P2 using FastPfu fly DNA polymerase (TransGen Biotech, China) according to the manufacturer's instructions. The PCR fragment containing a kanamycinresistant cassette (1075 bp) was amplified from pYBA-1132 using primer pairs Km-F/Km-R. The P BAD promoter fragment was amplified using primer pairs P BAD -F/P BAD -R from pBAD30 plasmid. pJQ200SK plasmid was linearized with infusion-F and infusion-R primers. All these fragments were connected by a seamless cloning and assembly kit (Beijing Zoman Biotechnology Co., Ltd, Beijing, China) following the user manual. The resulting plasmid pJQP BAD -xcnA was transformed into E. coli S17-l λpir and transferred into wild type X. nematophila CB6 strain by conjugation. The detailed steps are as follows: E. coli S17-1 λpir and X. nematophila CB6 were cultured in an LB medium at 37 • C and 28 • C, respectively. When liquid cultures were grown to OD 600 0.6-0.8, 1 mL cultures were harvested and washed twice using fresh LB medium. The cells were resuspended in 50 µL LB. The E. coli S17-1 λpir and X. nematophila were mixed and spotted onto LB agar plate without antibiotics. The plate was grown at 28 • C for 18 h. Then, the bacterial colony was resuspended in the liquid LB medium and spread on an LB agar plate containing ampicillin and gentamicin. The single crossover homologous recombination bacteria were screened by selecting colonies and confirmed by PCR [4]. Thereafter, we cultured the colonies for 12 h in LB (without NaCl) and spread them on an LB plate containing 50 µg/mL of kanamycin and 5% sucrose (no NaCl) [50]. Sucrose resistant colonies were picked into LB supplemented with kanamycin and incubated at 28 • C. The replacement generated via the second crossover was confirmed by sensitivity to gentamicin as well by PCR [50].

Monitoring the Growth of ∆P BAD -xcnA Mutant Strain
Wildtype CB6 and ∆P BAD -xcnA strains were grown overnight in an LB medium at 28 • C by shaking at 200 rpm. Then, the cells were transferred to a fresh LB medium with an initial OD 600 value reaching 0.05. Thereafter, L-arabinose (0.5-2%) was added at 8 h, and the cultivation was continued for 48 h. For measuring cell density, the optical density of wildtype CB6 and ∆P BAD -xcnA strains at 600 nm [OD 600 ] was monitored at predetermined time points. Each experiment was performed in triplicate.

Analysis of the Yield of Xcns
Analysis of the yield of Xcns was performed according to a previously reported method [4,25,26,46] by high-performance liquid chromatography (HPLC). The equal volume of MeOH was added into the cell-free fermentation broth for HPLC analysis. The HPLC analysis condition was used according to previously described methods [4,46]. The yields of Xcn1 and Xcn2 were calculated based on comparison with standards we obtained previously [25,39].

Quantitative Real-Time PCR (qRT-PCR)
The total ribonucleic acid (RNA) of the wildtype CB6 strain and the ∆P BAD -xcnA mutant strain was extracted using a bacterial RNA extraction kit (TransGen Biotech, ER501-01). RNA reverse transcription to cDNA was performed using One-Step gDNA Removal and a cDNA Synthesis SuperMix kit (TransGen Biotech, AE311-04). qRT-PCR was carried out using SYBR Green qPCR SuperMix (GenStar, A304-10) according to the manufacturer's instructions. The reaction was performed as follows: 95 • C denaturation step for 30 s, 40 cycles of 95 • C for 5 s, and 60 • C for 30 s. The recA gene was chosen as the internal reference gene to normalize other genes' expressions [4,46]. The relative transcript levels were determined using the 2 −∆∆C T method with three independent experiments [51]. All the primers for qRT-PCR are listed in Table 2.

Antimicrobial Activity Assay
To assess the antimicrobial activity of the wildtype X. nematophila CB6 and the ∆P BAD -xcnA mutant, we performed an agar diffusion plate assay and a growth inhibition assay against Bacillus subtilis and Phytophthora capsici, which were sensitive to Xcn1 in our previous studies [26,46,52]. Concerning the agar diffusion plate assay, sample wells were prepared on a two-layer agar diffusion plate, in which 20 mL of medium containing 0.75% agar and 10 7 −10 8 cells of B. subtilis were layered above 10 mL of 2% sterilized agar. Then the tested sample was added into the wells, and the plates were cultured at 28 • C for 24 h to measure the diameter of the inhibition zone. To test the antifungal activity of X. nematophila, the P. capsici strain that is sensitive to Xcn1 was chosen as a model fungus. Briefly, the tested sample was mixed with a PDA medium with a final concentration of 0.5 µg/mL of Xcn1 and then poured onto a Petri dish. The mycelia PDA block was cut from the edges of a five-day colony of P. capsici and then was placed at the center of the plate for culturing at 25 • C. After seven days, the diameter of P. capsici was measured, and the growth inhibition rate was calculated using the following equation: growth inhibition rate (%) = 100 × [(Diameter of control fungal colony-Diameter of test fungal colony)/Diameter of control fungal colony] [53]. Each treatment was repeated three times.

Construction of ∆P BAD -xcnA Strain
The yield of Xcn1 was related with the transcription level of its biosynthetic gene cluster, thus, we speculated that promoting the transcription level of Xcn1 biosynthetic gene cluster by replacement of the original promoter of xcnA with the P BAD promoter could boost the yield of Xcn1. To test this hypothesis, we first performed a homologous recombination experiment to construct the ∆P BAD -xcnA mutant. Then the resulted mutants were confirmed by PCR and sequencing using external primers XcnA-Pro-TF and XcnA-Pro-TR ( Figure 1). The sizes of the PCR product were consistent with the expected results, which were 2300 bp and 4193 bp when using the genomes of wildtype and mutant strains as templates, respectively. Furthermore, the sequencing results were identical with the aimed replacement. Therefore, it proved that the xcnA original promoter in X. nematophila CB6 was successfully replaced with the P BAD promoter, correspondingly, the xcnA original promoter was knocked out from X. nematophila CB6.

Growth Profile of ΔPBAD-xcnA Strain
Once obtained the mutant, we were rather curious whether the genetic manipulation performed in the CB6 strain could affect its growth, which might subsequently influence the yield of Xcn1. In the whole fermentation process, the ΔPBAD-xcnA mutant displayed a similar growth curve with the wildtype CB6 strain. In addition, the production of Xcn1 in the ΔPBAD-xcnA mutant strain must be coupled with the addition of arabinose utilized for activating the inducible promoter PBAD. So, it was also necessary to assess the influence of arabinose on the growth of ΔPBAD-xcnA mutant strain. When 0.5% L-arabinose was added to the LB medium, the growth curve of the ΔPBAD-xcnA mutant was decreased relative to the wildtype CB6 strain. With the increase in the added L-arabinose, a greater impact on

Growth Profile of ∆P BAD -xcnA Strain
Once obtained the mutant, we were rather curious whether the genetic manipulation performed in the CB6 strain could affect its growth, which might subsequently influence the yield of Xcn1. In the whole fermentation process, the ∆P BAD -xcnA mutant displayed a similar growth curve with the wildtype CB6 strain. In addition, the production of Xcn1 in the ∆P BAD -xcnA mutant strain must be coupled with the addition of arabinose utilized for activating the inducible promoter P BAD . So, it was also necessary to assess the influence of arabinose on the growth of ∆P BAD -xcnA mutant strain. When 0.5% L-arabinose was added to the LB medium, the growth curve of the ∆P BAD -xcnA mutant was decreased relative to the wildtype CB6 strain. With the increase in the added L-arabinose, a greater impact on the growth was observed (Figure 2). performed in the CB6 strain could affect its growth, which might subseque the yield of Xcn1. In the whole fermentation process, the ΔPBAD-xcnA mutan similar growth curve with the wildtype CB6 strain. In addition, the product the ΔPBAD-xcnA mutant strain must be coupled with the addition of arabino activating the inducible promoter PBAD. So, it was also necessary to assess th arabinose on the growth of ΔPBAD-xcnA mutant strain. When 0.5% L-arabino to the LB medium, the growth curve of the ΔPBAD-xcnA mutant was decrea the wildtype CB6 strain. With the increase in the added L-arabinose, a grea the growth was observed (Figure 2).

Effects of PBAD Promoter Replacement on the Yield of Xcn1
As a type of tight control promoters, PBAD in the ΔPBAD-xcnA mutant c as a master switch for modulating Xcn1 and Xcn2 production. Indeed, X detected in the fermentation broth of ΔPBAD-xcnA mutant without the additi nose ( Figure 3). After adding arabinose, a significant amount of Xcns wa HPLC. The Xcn1 yields, however, did not improve with the increase in L-a centration, and it kept a similar level between 0.5 and 1.5% L-arabinose in th dia. When the L-arabinose concentration reached 0.5% and 1%, relative to strain the fermentation yield of Xcn1 of the ΔPBAD-xcnA mutant at 48 h was im 68.5 mg/L to 249.7 mg/L (3.6-fold) and 234.9 mg/L (3.4-fold), respectively (Fi 2% L-arabinose was added to the medium, the yield of Xcn1 reduced sig hough it was still higher than the wildtype CB6 strain. The yield of Xcn2, the highest, about 4.4-fold more than the wildtype CB6 strain, in the mediu

Effects of P BAD Promoter Replacement on the Yield of Xcn1
As a type of tight control promoters, P BAD in the ∆P BAD -xcnA mutant could function as a master switch for modulating Xcn1 and Xcn2 production. Indeed, Xcns cannot be detected in the fermentation broth of ∆P BAD -xcnA mutant without the addition of Larabinose ( Figure 3). After adding arabinose, a significant amount of Xcns was detected by HPLC. The Xcn1 yields, however, did not improve with the increase in L-arabinose concentration, and it kept a similar level between 0.5 and 1.5% L-arabinose in the culture media. When the L-arabinose concentration reached 0.5% and 1%, relative to the wildtype strain the fermentation yield of Xcn1 of the ∆P BAD -xcnA mutant at 48 h was improved from 68.5 mg/L to 249.7 mg/L (3.6-fold) and 234.9 mg/L (3.4-fold), respectively ( Figure 3). When 2% L-arabinose was added to the medium, the yield of Xcn1 reduced significantly, although it was still higher than the wildtype CB6 strain. The yield of Xcn2, however, was the highest, about 4.4-fold more than the wildtype CB6 strain, in the medium containing 2% L-arabinose. The total yield of Xcns, namely the sum of Xcn1 and Xcn2, was also investigated, and as shown in Figure 3, relative to the WT CB6 strain remarkably higher yields were found in the ∆P BAD -xcnA mutants induced by L-arabinose of different concentrations.

Transcriptional Analysis of the Xcn Genes
To evaluate the xcn genes' transcription levels of the ∆P BAD -xcnA mutant in different L-arabinose concentrations, qRT-PCR was employed to monitor the four critical genes (xcnA, xcnF, xcnG, and xcnM) during the fermentation process ( Figure 4). Consistent with our expectations, induced by L-arabinose, all of the four genes in the ∆P BAD -xcnA mutant showed higher transcription levels relative to the wildtype CB6 strain at both 24 h and 36 h. It is worth noting that the expression levels of xcnA, encoding the important NRPS enzyme responsible for Xcn1 biosynthesis, in the ∆P BAD -xcnA mutant were enhanced 7-fold, 9-fold, 8-fold, and 6-fold at 36 h with addition of 0.5%, 1.0%, 1.5%, and 2% Larabinose, respectively. Another crucial gene xcnF, encoding a PKS enzyme involved in Xcn1 biosynthesis, was also highly upregulated with 2.7-fold, 3.2-fold, 2.3-fold, and 2.9-fold increase at 0.5%, 1.0%, 1.5%, and 2% L-arabinose, respectively. The transcription levels of xcnG, encoding a specific peptidase that can cleave the inactive prexenocoumacins into the active compound Xcn1, were increased at 36 h by 2.4-fold, 3.0-fold, 2.0-fold, and 1.9-fold, respectively. The over-transcription of xcnA, xcnF, and xcnG, involved in Xcn1 biosynthesis, was in concert with the observed higher Xcn1 yield produced by the ∆P BAD -xcnA mutant. The transcription level of the xcnM gene was increased by 2.5-fold, 2.3-fold, 2.0-fold, and 2.2-fold at 36 h with the addition of 0.5%, 1.0%, 1.5%, and 2% L-arabinose, respectively. It led to a higher conversion rate from Xcn1 to Xcn2, therefore, increased Xcn2 yield.
2% L-arabinose. The total yield of Xcns, namely the sum of Xcn1 and Xcn2, was also investigated, and as shown in Figure 3, relative to the WT CB6 strain remarkably higher yields were found in the ΔPBAD-xcnA mutants induced by L-arabinose of different concentrations.  tive compound Xcn1, were increased at 36 h by 2.4-fold, 3.0-fold, 2.0-fold, and 1.9-fol respectively. The over-transcription of xcnA, xcnF, and xcnG, involved in Xcn1 biosynth sis, was in concert with the observed higher Xcn1 yield produced by the ΔPBAD-xcnA m tant. The transcription level of the xcnM gene was increased by 2.5-fold, 2.3-fold, 2.0-fol and 2.2-fold at 36 h with the addition of 0.5%, 1.0%, 1.5%, and 2% L-arabinose, respe tively. It led to a higher conversion rate from Xcn1 to Xcn2, therefore, increased Xcn2 yiel

Antimicrobial Activity of ΔPBAD-xcnA Mutant Strain
We utilized an agar diffusion plate assay and growth inhibition assay against B. su tilis and P. capsici, respectively, to test whether the higher yield of Xcn1 gave a correspon ing greater antimicrobial activity. As expected, in comparison with the wildtype CB strain, the fermentation broth of ΔPBAD-xcnA mutant exhibited higher antimicrobial acti ity, the zone of inhibition against B. subtilis significantly increased from 11.83 mm to 16. mm ( Figure 5A, Table 3). Similarly, better inhibition activity against P. capsici was detect for the fermentation broth of the ΔPBAD-xcnA mutant strain as well. As shown in Figure 5 and Table 3, similar inhibition effects were achieved using the fermentation broth of t ΔPBAD-xcnA mutant strain and the wildtype CB6 strain in 500-fold and 137-fold dilutio the diameter of P. capsici being 18.50 mm and 17.64 mm, (Figure 5B), and correspondin inhibition rates being 77.80% and 78.80% (Table 3), respectively. The results indicated th the antimicrobial effects of the ΔPBAD-xcnA mutant were significantly improved due to t enhanced yield of Xcn1 and possible contribution from increased Xcn2. Relative transcription levels of the biosynthesis and conversion related genes of Xcn1 (xcnA, xcnF, xcnG, and xcnM) were measured by qRT-PCR; error bars represent the standard deviation of three independent replicates. Data were subjected to the Duncan ANOVA test, columns labelled with different lowercase letters were significantly different at p ≤ 0.05.

Antimicrobial Activity of ∆P BAD -xcnA Mutant Strain
We utilized an agar diffusion plate assay and growth inhibition assay against B. subtilis and P. capsici, respectively, to test whether the higher yield of Xcn1 gave a corresponding greater antimicrobial activity. As expected, in comparison with the wildtype CB6 strain, the fermentation broth of ∆P BAD -xcnA mutant exhibited higher antimicrobial activity, the zone of inhibition against B. subtilis significantly increased from 11.83 mm to 16.34 mm ( Figure 5A, Table 3). Similarly, better inhibition activity against P. capsici was detected for the fermentation broth of the ∆P BAD -xcnA mutant strain as well. As shown in Figure 5B and Table 3, similar inhibition effects were achieved using the fermentation broth of the ∆P BAD -xcnA mutant strain and the wildtype CB6 strain in 500-fold and 137-fold dilution, the diameter of P. capsici being 18.50 mm and 17.64 mm, (Figure 5B), and corresponding inhibition rates being 77.80% and 78.80% (Table 3), respectively. The results indicated that the antimicrobial effects of the ∆P BAD -xcnA mutant were significantly improved due to the enhanced yield of Xcn1 and possible contribution from increased Xcn2.

Discussion
Xcn1 has great potential to be applied in the agricultural field to protect crops the infection of many pathogens due to its wide range of antimicrobial [25,26,28,39,40]. However, the present fermentation yield of Xcn1 is still low, whic pers its large-scale production. Numerous well-known strategies have been repo promote the yield of secondary metabolites in microorganisms, for example, inc the precursor supply, deleting the competing pathways, engineering the crucial e of the biosynthetic pathway, engineering regulatory networks, and controlling t scription of the biosynthetic gene cluster [49,[54][55][56][57][58][59][60][61][62][63][64]. In our research, we found biosynthetic gene cluster of Xcn1 remained at a low transcriptional level during mentation process (unpublished data), so we supposed that a promoter replaceme egy could be effective to increase its transcription, thereby improving the yield of The PBAD promoter, which possesses the advantage to accurately control th

Discussion
Xcn1 has great potential to be applied in the agricultural field to protect crops against the infection of many pathogens due to its wide range of antimicrobial activity [25,26,28,39,40]. However, the present fermentation yield of Xcn1 is still low, which hampers its large-scale production. Numerous well-known strategies have been reported to promote the yield of secondary metabolites in microorganisms, for example, increasing the precursor supply, deleting the competing pathways, engineering the crucial enzymes of the biosynthetic pathway, engineering regulatory networks, and controlling the transcription of the biosynthetic gene cluster [49,[54][55][56][57][58][59][60][61][62][63][64]. In our research, we found that the biosynthetic gene cluster of Xcn1 remained at a low transcriptional level during the fermentation process (unpublished data), so we supposed that a promoter replacement strategy could be effective to increase its transcription, thereby improving the yield of Xcn1.
The P BAD promoter, which possesses the advantage to accurately control the transcription level of target genes with varying concentrations of inducer, has been used extensively for modulation of gene expression [49,65,66]. It can play an important role in natural products isolation and structure elucidation through strictly regulating overproduction (with inducer) or loss (without inducer) of the desired compound in fermentation process [49]. This strategy is also widely applicable to activate gene clusters that are "silent" under common conditions and will enrich our knowledge of natural products and their biological functions [49].
Previous research had reported that the mutant strains generated by genetic manipulation were often companied by the changes in phenotype and physiology, for instance morphology, metabolites, growth, etc. [4,8,28,64]. Here, the growth curves of the ∆P BAD -xcnA mutant strain and wildtype CB6 strain in LB were similar when no L-arabinose was added, indicating the genetic manipulation caused no obvious influence on its growth. When we added L-arabinose to the LB broth, the P BAD promoter was activated, thus, the production of Xcn1 increased remarkably. These results are in agreement with previous research where Bode et al. applied arabinose-inducible promoters in entomopathogenic bacteria to overproduce the desired natural products for their structure elucidation [49]. As Xcn2 was converted from Xcn1, we also paid attention to the production changes in Xcn2. Obviously, the yield of Xcn2 was also increased, coupled with the increased yield of Xcn1. Correspondingly, relative to the wildtype strain, the higher transcription level of xcnA, xcnF, xcnG, and xcnM were observed in the ∆P BAD -xcnA mutant strain with the induction of L-arabinose, which revealed the underlying mechanism for the increased yield of Xcn1 and Xcn2. However, the induction strength for the P BAD promoter could be increased within a certain range of L-arabinose concentration, and once exceeded, the transcription levels of target genes might be reduced [67]. Indeed, 2% L-arabinose resulted in lower transcription levels of xcnA, xcnF, and xcnG at 36 h compared with 0.5-1% L-arabinose. As we observed in the growth curves, a high concentration of L-arabinose inhibited the growth of the mutant strain, this is another reason responsible for the decreased yield of Xcn1 in the fermentation medium containing 2% L-arabinose. Moreover, a large amount of Xcn2, which was converted from Xcn1 mediated by both the XcnM and XcnN enzymes, was found in the fermentation broth. So, in future research, the addition of an adsorber resin to the culture during the fermentation of the ∆P BAD -xcnA strain may be a practicable way to in situ collect the produced Xcn1 to reduce its conversion [46].
In previous research [4,26,37], Xcn1 was regarded as the main antimicrobial compound in X. nematophila. In other words, the higher concentration of Xcn1 corresponds to higher antimicrobial activity. In this study, we improved the yield of Xcn1 remarkably in the ∆P BAD -xcnA mutant, so more obvious antimicrobial activity should be identified in the corresponding fermentation broth. As anticipated, higher inhibition activity against model strains B. subtilis and P. capsici was detected. The study presented here provides an effective strategy to engineer the CB6 strain to increase the yield of Xcn1; together with other complementary means, for instance, increasing the precursor supply, deleting the competing pathways, optimizing the fermentation process, controlling the regulators, and enhancing the resistance of the CB6 strain to Xcn1 et al., the greater improvement of the yield of Xcn1 could be realized in the near future, paving the way for its industrial production.

Conclusions
In summary, we achieved the remarkable improvement of the yield of Xcn1 by promoter replacement strategy, the fermentation yield of Xcn1 in the ∆P BAD -xcnA mutant strain was improved from 68.5 mg/L to 249.7 mg/L helped by the induction of 0.5% L-arabinose, and it was companied with increased antimicrobial activity. Further study revealed that upregulation of the biosynthesis related genes of Xcn1 contributed to the yield improvement. We hope that the promoter replacement strategy employed here could be used as a reference for other Xenorhabdus/Photorhabdus species, which might promote the commercial development of more nature products from nematode associated bacteria.