Development of an inducible T7 expression system in Bacillus subtilis ATCC 6051a for production of α-L-arabinofuranosidase CURRENT STATUS: UNDER REVIEW

Background: Xylan is the second most abundant polysaccharide biomass on the earth, the polymers have a backbone of β-1,4-linked xylose residues with various side-chain substitutions, such as arabinose, acetic acid, glucuronic acid, and other esterified groups, thus, the removal of arabinose side groups by α-L-arabinofuranosidases is helpful in various industrial processes involving xylan treatment. Bacillus subtilis ATCC 6051a is known for its excellent capacity of secretory production of recombinant peptides, however, poor experience in genetic manipulation and lack of universal expression elements impede this strain for wider application. Results: Xylose inducible comK was integrated into the genome of B. subtilis ATCC 6051a, and the transformation efficiency of the engineered strain B. subtilis 164S was increased by more than 1000 folds. B. subtilis 164S was further modified to generate B. subtilis 164T7P which incorporates a D-xylose inducible T7 RNA polymerase. The recombinant GFP expressed by 164T7P is more than thirteen times that achieved by P43 promoter, representing the most efficient expression system that had been ever reported in B. subtilis . Subsequently, abfA1 , encoding a glycoside hydrolase (GH) family 51 enzyme was cloned and overexpressed in 164T7P. The activity of recombinant α-L-arabinofuranosidase (AbfA1) reached 90.6±2.0 U mL -1 in the fermentation broth. Using p NPA as a substrate, kinetic parameters of the crude enzyme were Km of 1.4±0.1 mM and kcat of 139.4 s −1 . The optimum temperature and pH of the recombinant AbfA1 towards p NPA were observed to be 45 °C and pH 6.5, respectively. Conclusion: With enhanced cellular competence and introduction of T7 RNA polymerase, B. subtilis ATCC 6051a was engineered as a versatile cell tool for recombinant production of heterologous peptides employing T7 promoter. The novel expression kit demonstrated a

3 system were demonstrated by high-level production of a bacterial type α-Larabinofuranosidase.

Background
Given that their biological nature of nonpathogenic GRAS (generally recognized as safe), Bacillus subtilis species serve as one type of the most widely used cell factories, especially for the production of recombinant enzymes applicable in food-processing industry. B. subtilis strains are thus the most characterized gram-positive bacteria, which is intensively studied in the fields of genetics, biochemistry, physiology and fermentation technologies [1][2][3]. Due to the native transformability, B. subtilis strain 168 and its protease deficient subspecies are commonly used for lab-scale production of proteins.
Nevertheless, the less domesticated strain B. subtilis ATCC 6051a exhibited much better secretory capability and superior growth properties in complex mediums than B. subtilis 168 [4][5][6], however, engineering in B. subtilis ATCC 6051a was greatly limited or retarded by its poor transformation performance.
The genome sequence of B. subtilis ATCC 6051a was published and compared side by side with B. subtilis 168 [6], which revealed that the severely reduced competence of B.
subtilis ATCC 6051a was likely caused by a frameshift mutation in comP that encodes a two-component sensor kinase, which activates an operon related to competence forming in B. subtilis by turning on expression of its downstream ComK, a decisive regulator for natural competence development [7,8]. It has been reported that the overexpression of ComK was capable of upturning the transformation efficiency for nearly 1000 times [9][10][11].
However, the attempt of raising the ComK level in B.subtilis ATCC 6051a had never been reported previously. On the other side, the repertoire of efficient promoters in B. subtilis is really limited. The most known promoter applied in B. subtilis is P43 [12], which is an artificial element widely utilized. Some other novel promoters include Pgrac [13], PyxiE 4 [14], Pglv [15], dual promoter PgsiB-PHpaII [16] and pShuttle-09 [17], but none of them is comparable to phage derived T7 promoter in transcriptional strength. The T7 expression system employing the T7 promoter was first introduced in Escherichia coli and displayed high efficiency and specificity in heterologous expression of target peptides [18,19].
Studies have shown that the engineered T7 expression system was able to improve the expression level of heterologous peptides in Bacillus [20][21][22], although it has not tested in undomesticated B. subtilis strain, such as in B.subtilis ATCC 6051a. Another important feature of a promoter is its controllability. Since it has been confirmed in E. coli, IPTG inducible element was also introduced in Gram-positive strains, in promoter such as Pspac. However, it is known that the control on the leaky expression run by Pspac or other lacO containing promoters was not desirable as expected to be [23]. Thus, the efforts to seek a stable, tightly controllable and highly-active promoter are vital and urgent for the further expanding of the applicability of B. subtilis in the production of recombinant proteins.
In this study, comk cassette was integrated into the chromosome of B. subtilis ATCC 6051a, creating an easy-to-manipulate strain B. subtilis 164S, thereafter, a revised cloning strategy based on 'Simple Cloning' [24] was invented, enabling convenient seamless cloning of shuttle vectors. To create a host cell utilizing T7 promoter, DNA cassette encoding T7 RNA polymerase under P xylA , a xylose inducible promoter, was integrated into the genome B. subtilis 164S, generating B. subtilis 164T7P. The efficiency of the T7 promoter in B. subtilis was demonstrated by the expression of GFP and an α-Larabinofuranosidase encoded by abfA1 from B. licheniformis.

Results And Discussions
Engineering B. subtilis ATCC 6051a to raise its natural competence by expressing recombinant comK Compared with traditional genetic manipulation through circular plasmid mediated homologous recombination, double crossover by homologous fragments exchange through linear DNA eases the handling of genome engineering in prokaryotic cells in terms of efficiency and labor intensity. However, comparing with circular plasmid transformation, homologous integration through linear DNA requires much higher transformation efficiency. Therefore, it was an intimidating task to perform one step gene replacement in B. subtilis ATCC 6051a, a genetically indocile, however industrial favored strains. The alignment analysis using the genomic sequences of B. subtilis ATCC 6051a and B. subtilis 168 revealed a frameshift mutation from AA to A that leads to functional disability of ComP in B. subtilis ATCC 6051a, subsequently reducing the cellular level of ComK [6], one of the key activators modulating natural competence in Bacillus. Overexpression of ComK was proposed and confirmed to improve the transformation efficiency even earlier [9-11, 24, 25]. Thus, we proposed to perform homologous expression of ComK in 6051a by cloning a copy of the native comk gene under a xylose inducible promoter, P xylA from B. megaterium [9], then inserting the cassette into 6051a's chromosome via doublecrossover recombination. However, that could not be directly accomplished duo to host cells' low transformation efficiency. Therefore, we constructed pMK4-comk, a selfreplicating plasmid containing the expression cassette, and transformed the shuttle vector into B. 6051a (as shown in Fig. 1a, more details are shown in Additional file 1: Fig S1).
With the lifted transformation efficiency gained through the plasmid mediated expression of ComK, linear comk cassette together with kanamycin resistance marker (kan) was integrated into the genome and replaced native nprE in 6051a (Fig. 1b). Finally, a plasmid expressing Cre integrase was transformed into the bacillus mutant (Fig. S1), creating a markerless strain, which was named as B. subtilis 164S.

Enhanced transformation efficiency upon induction of ComK
The transformation efficiency of B. subtilis 164S was determined by transforming 1.0 μg of pMK4 DNA (circular DNA of 5585 bp in length) to properly prepared competence cells. As the result of induced expression of ComK by culturing cells in medium containing Dxylose, the transformation efficiency reached 3×10 4 transformants per μg DNA, which increased more than one thousand times over 6051a (data not shown).
To determine the recombination efficiency through double crossover, linear DNAs with varying sizes of homologous fragments flanking four different open reading frames (encoding spoII AC, aprE, amyE and srfAC respectively, Table 1) were prepared by fusion PCR [26,27]. The numbers of transformants obtained after transformation into 164S or 6051a using each linear DNA were counted. As seen in Table 1, 164S transformants' colony numbers in all experiments fell into a range of 70~160, in contrast, no transformant was obtained from 6051a using the same linear DNA (data not shown).
Previous studies postulated that the minimal length of DNA fragments required for homologous recombination was around 400~500 bp [27,28], it has never been confirmed in industrial bacillus strain such as 6051a. To determine the relationship between the length of the homologous flanking region and the double crossover efficiency, we prepared linear DNAs for in-frame deletion of spoII AC with different sizes of homologous arms (200 to 1000 bp, respectively, Fig. 2a), used them for gene replacement in 164S. As presented in Fig. 2b, linear DNA mediated transformations in 164S were DNA size dependent, longer DNA arms evoked much higher recombination efficiency in 164S, and the minimal length to initiate an efficient DNA exchange event was about 400 bp, that is in agreement with the previous report [27].

Seamless cloning method for one step construction of Bacillus expression vector
Seamless cloning is advantageous in preparing faithful DNA constructs. The method takes advantage of restriction-independent cloning strategies, such as overlap PCR, DNA recombination or DNA ligation using type IIs restriction enzymes. Available commercial kits for seamless cloning include Golden Gate cloning technology, Gibson assembly, or Infusion Cloning. Additionally, an enzyme-free cloning method known as "Simple Cloning" was developed and proved to have high transformation efficiency in B. subtilis SCK6 [24].
According to the report, ~10 7 transformants per μg of DNA multimer can be achieved.
Here we slightly modified the method and developed it as one step seamless method for the construction of desired plasmids. As demonstrated in Fig. 3a, the backbone plasmid was linearized by restriction enzyme, it joined with the insert through overlap PCR using primers without restriction site sequences that originally existed on the plasmid DNA; during PCR program, the unmatched sequences would be automatically trimmed by 3'-5' exonuclease of high-fidelity DNA polymerase; after transformation into B. subtilis cell, linear DNA multimers generated by PCR trigger efficient in vivo splicing that leads to 8 formation of circular DNA [24].
Construction of pMK4-T7-abfA1 demonstrated the feasibility of the novel cloning strategy.
First, pMK4-T7 was digested with BamHI to give linear double strand DNAs that were used in PCR reaction for preparation DNA multimers, the PCR system also included PCR prepared abfA1 fragment containing homologous ends to the vector DNA, however without BamHI residue site that originally existed immediate downstream of T7 promoter on pMK4-T7; PCR generated multimers were subsequently transformed into 164S, that promptly produced transformants harboring circularized pMK4-T7-abfA1. As shown in Fig. 3b, the DNA sequencing confirmed BamHI site was eliminated on generated pMK4-T7-abfA1. The transformation efficiency can be over 1.0×10 6 transformants per μg of DNA multimer, which was lower than the report [24] but sufficient for most strain engineering jobs.

Construction of inducible T7 expression system in B. subtilis
T7 RNA polymerase over-expressing strains, such as BL21 (DE3) or JM109 (DE3), were required for T7 promoter guided robust expression in E. coli, for the same reason, T7 RNA polymerase has to be expressed in B. subtilis if T7 promoter is recruited for transcription of target genes [20][21][22]. However, most bacterial systems utilize IPTG as the expression switch, however, IPTG is costly and unhealthy if not removed from the products, thereby limiting the system's application in the food or pharmaceutical field [29]. In this study, a D-xylose inducible T7 expression system was constructed. As demonstrated in Fig. 4a, we first assembled an artificial cassette encoding T7 RNA polymerase under the inducible promoter P xylA , delivery of the linear DNA into 164S knocked out aprE in 164S and replaced it with the DNA cassette expressing T7 RNA polymerase. A copy of xylR was included in the cassette to stop the "leaky expression" of the polymerase when D-xylose is not available. 164T7P was tested in the expression of GFP (cloned in plasmid pMK4-T7-gfp, as seen in Additional file 2: Fig S2). Since P43 was recognized as a strong and constitutive promoter in B. subtilis [12,30], a 164T7P control strain harboring the plasmid pMK4-P43-gfp (Additional file 2: Fig S2) was made for comparison between P xylA and P43.
To examine GFP expression levels, the average fluorescence intensity in cells grown upon the presence of different concentrations of D-xylose was determined in a microplate reader. As shown in Fig. 4b, the level of heterologous expressed GFP in 164T7P correlated with the concentration of D-xylose supplemented in mediums ( Fig. 4b and Fig. 4c). It showed that in medium with 1.0% D-xylose, the fluorescence intensity of GFP expressed by T7 promoter reached 2.83×10 7 au, which is more than 13 times of that achieved by P43 promoter (2. analyzed by SDS-PAGE. As seen in Fig. 5a, a high level of α-L-arabinofuranosidase was secreted out of host cells, with an estimated mass of 56.9 kDa. The total protein concentration in the supernatant was measured to be 0.8 g L -1 in Bradford assay, and the recombinant AbfA1 represented more than 90% of the total protein in the supernatant (Fig.   5a). The activity and the kinetic parameters of AbfA1 were determined using pNPA (4nitrophenyl α-L-arabinofuranoside) as the substrate. The maximum activity during fermentation was measured to be 90.6±2.0 U mL -1 at 52 h (Fig. 5b), with a Km value of 1.4±0.1 mM and the kcat value of 139.4 s − 1 . The enzyme remained to be active in the pH range of 3.5-9.0 with an optimal pH at 6.5 (Fig. 5c). Its optimal temperature was determined to be 45 ºC while working between 25-55 ºC (Fig. 5d). This is by far the first report on the heterologous secretion expression of α-L-arabinofuranosidase in B. subtilis.
It is worth mentioning that the protein sequence of AbfA1 showed neither Sec-nor Tat-

Conclusion
We developed a xylose-inducible expression system in B. subtilis ATCC 6051a. First, we improved industrial B. subtilis' natural competence through the inducible expression of Comk that was placed under the control of a P xylA promoter; then, we integrated T7 RNA polymerase behind another P xylA promoter on 6051a's genome. The constructed T7 RNAP expression system exhibited high efficiency of T7 promoter-specific protein's expression in engineered 6051a strain, that included GFP and a bacterial arabinofuranosidase, the latter was secreted out of host cells unexpectedly, apparently through a non-classical secretion pathway since it lacks any known secretive signal peptide. Also, we modified a seamless cloning strategy, enabling one-step, rapid and precise construction of expression vectors in B. subtilis. These efforts not only established an attractive expression platform with high efficiency but also provided benefits for DNA or genome manipulation in previously imbecile B. subtilis ATCC 6051a.

Strains, plasmids, reagents, and cultivation conditions
Bacteria strains and plasmids used in this study are listed in Table 1 [27]. First, a DNA cassette containing cre behind Pspac was prepared using pDGC as the template with the primer pair Cre-F/Cre-R, the PCR product was digested with EcoRI and ligated into pMK4 to generate pMK4-cre; then the vector was transformed into 164K, that elicited the homologous recombination of loxP sites in 164K upon the expression of Cre recombinase and resulted in elimination of kan from the chromosome. Next, the vector cure procedure was performed to remove pMK4cre from the transformant, which finally generated B. subtilis 164S.

Generation of linear DNAs for gene replacement in B. subtilis
Linear DNAs used for gene replacement contain an erythromycin resistance gene as the selective marker which is surrounded by homologous DNAs [26,27]. For example, aprerm-apr was used for apr (apr encodes alkaline protease in the strain) deletion in 6051a.
To prepare apr-erm-apr, ermC cassette DNA was first amplified using vector pUCK-syn-sigf as the template with the primer pairs Erm-F/Erm-R, then the primer pairs ApU-F/ApU-R and ApD-F/ApD-R were used to prepare flanking regions of apr. Above mentioned three fragments contains ending sequences homologous to each other and were put into PCR tube for overlap PCR that was done using 2 × Phanta ® Master Mix. The obtained PCR products were treated with DpnI, and followed by purification with the Axygen DNA purification kit. With the same strategy, linear DNAs for gene replacement of spoII AC, amyE or srfAC were also prepared; and the primers used were listed in Additional file 4: Table S1.

Quantitative transformation efficiency assay for B. subtilis
The transformation of B. subtlilis 164S or B. subtlilis 164T7P was performed using the method described in the previous study with minor modification [9]. Briefly, 0.5 mL of a fresh culture grown in LB was inoculated into 4.5 ml of pre-warmed LB containing 1% (m/v) D-xylose, followed by incubation at 37 o C on a rotary shaker for 1.5 h, or until OD 600 reached ~1.0. 100 μL of fresh culture was immediately used for transformation by mixing with ~100 ng of linear or circular DNA. The DNA-cells mixture was incubated at 37 o C with rotary shaking for 2 h before spreading onto LB agar plates with an appropriate antibiotic.
The linear DNAs prepared for gene replacement (apr, amyE , spoIIAC or srfAC ) were comprised of a DNA cassette encoding drug resistance and two flanking fragments for homologous recombination.

Construction of B. subtlilis 164T7P
A DNA cassette encoding T7 RNA polymerase was prepared by PCR using the genome DNA of E. coli BL21 (DE3) as the template with the primer pair F77-F/F77-R. A DNA fragment containing P xylA promoter and xylR was amplified from vector pMK4-comk using the primer pair Fxy-F/Fxy-R. Overlap PCR was performed to obtain a fused DNA fragment, which was then digested with XmaI/KpnI before ligating with the same enzymes treated pMD19T-aea.
The generated plasmid (pMD19T-aea-T7P) was linearized by ApaLI before transforming into B. subtilis 164S. Colonies formed on the LB agar plate containing erythromycin were verified for DNA insertion of a cassette encoding T7 RNA polymerase on the locus of aprE.
The loss of the drug resistance (ermC) was achieved by Cre recombinase using the procedure described early. Finally, the markerless strain derived from 164S was named as 164T7P since it harbors a DNA insert on its chromosome for expression of T7 RNA polymerase.

Construction of B. subtilis strains expressing GFP or α-L-arabinofuranosidase
pMK4-T7-gfp and pMK4-T7-abfA1 were constructed for heterologous expression of GFP and the arabinofuranosidase respectively. First, we cloned a DNA fragment that contains a T7 promoter, T7 terminator and some restriction cloning sites, into pMK4 to produce pMK4-T7. Next, pMK4-T7 was linearized with BamHI and EcoRI, which generated linear DNAs for the preparation of the multimer DNAs, using the overlap PCR based on the 'Simple Cloning' method described earlier, and, the insert DNA encoding gfp or abfA1 was required for the overlap PCR, they were also prepared by PCR. The previously constructed pMK4-P43-gfp was used as the template for PCR preparation of the gfp insert, the template DNA to prepare abfA1 insert was chemically synthesized using the sequence derived from B.
licheniformis ATCC 14580 (accession number NZ_CP034569) with the primer pair ArbF-F/Arbf-R. Finally, the generated multimer DNAs were transformed into B. subtilis 164T7P, creating the strain for expression of GFP or AbfA1.

Fluorescence measurements
Culture