Multigene disruption in undomesticated Bacillus subtilis ATCC 6051a using the CRISPR/Cas9 system

Bacillus subtilis ATCC 6051a is an undomesticated strain used in the industrial production of enzymes. Because it is poorly transformable, genetic manipulation in this strain requires a highly efficient genome editing method. In this study, a Streptococcus pyogenes CRISPR/Cas9 system consisting of an all-in-one knockout plasmid containing a target-specific guide RNA, cas9, and a homologous repair template was established for highly efficient gene disruption in B. subtilis ATCC 6051a. With an efficiency of 33% to 53%, this system was used to disrupt the srfC, spoIIAC, nprE, aprE and amyE genes of B. subtilis ATCC 6051a, which hamper its use in industrial fermentation. Compared with B. subtilis ATCC 6051a, the final mutant, BS5 (ΔsrfC, ΔspoIIAC, ΔnprE, ΔaprE, ΔamyE), produces much less foam during fermentation, displays greater resistant to spore formation, and secretes 2.5-fold more β-cyclodextrin glycosyltransferase into the fermentation medium. Thus, the CRISPR/Cas9 system proved to be a powerful tool for targeted genome editing in an industrially relevant, poorly transformable strain.

Cas9 system in a single knockout plasmid that could be used for highly efficiently genome editing in B. subtilis (Fig. 1a). The six knockout plasmids (pHYcas9dsrf1, pHYcas9dsrf2, pHYcas9dspo, pHYcas9dnpr, pHYcas9dapr and pHYcas9damy), which originated from plasmid pHY300PLK-β -CGTase, consist of the cas9 gene amplified from plasmid pwtcas9-bacterial, a sgRNA and its promoter P43, the temperature-sensitive replicon PE194, and a homologous repair template. The synthesis of Cas9 protein in vivo was driven by the α -amylase promoter (PamyQ), which originates from B. amyloliquefaciens. The P43 promoter is a constitutively expressed promoter 26 that can drive strong transcription of the sgRNA in B. subtilis. The homologous repair template was obtained through overlap extension PCR of regions upstream and downstream of the target locus. The length of the upstream and downstream regions ranged from 450 to 550 bp, and they introduced an Xho I site into the target locus. The knockout plasmids and plasmid pHY300PLK-β -CGTase encode ampicillin resistance in E. coli and tetracycline resistance in both E. coli and B. subtilis.
Disruption of the srfC gene using the CRISPR/Cas9 system. We used B. subtilis ATCC 6051a as the initial strain in which to perform genetic manipulations. Fermentation of B. subtilis ATCC 6051a in a 3 L fermenter in our laboratory produced a massive amount of foam that requires large quantities of antifoam agent (Fig. 2). Accumulation of the amphiphilic molecule surfactin promotes foam production, and the gene srfC is crucial to the regulation of surfactin production 5 . The CRISPR/Cas9 system developed in this study was first tested using srfC as the target (Fig. 1c). We used the knockout plasmid pHYcas9dsrf1, which produces both a sgRNA specific to srfC and the Cas9 protein, to transform the initial strain, B. subtilis ATCC 6051a. Because the temperature-sensitive knockout plasmid is maintained at low copy in transformants at 37 °C, the tetracycline-resistant transformants were confirmed by colony PCR of cas9. As for the commonly same genotype of transformation colonies, the target cleavage by Cas9 and homology-directed repair are commonly performed during the liquid incubation of transformation. The resulting mutants had an Xho I site within the repair locus, and this disruption genotype passed to the next generation through passaging in liquid media or LB plate. To test the effectiveness of the CRISPR/Cas9 system, the regions upstream and downstream of the repair locus were amplified by PCR, and then the PCR products were digested with Xho I (Fig. 3). There are 66 ± 14 transformants after transformation of knockout plasmid pHYcas9dsrf1. 30 transformants were screened, and 13 ± 2 colonies yielded a recombinant genotype with a disruption efficiency of 43% ± 6%. The results, combined with DNA sequencing of the homologous regions of the mutants ( Supplementary Fig. S1a) demonstrated that the editing system described above worked efficiently. Curing the knockout plasmid pHYcas9dsrf1 through overnight incubation at 51 °C produced the desired mutant, named BS1. Colonies of BS1 were used as the initial strain for next gene disruption. There are 20 ± 4 transformants after transformation of initial knockout plasmid pHYcas9d, and no srfC gene mutant was found among the transformants. Though DNA nonhomologous end joining exist in B. subtilis 27 , the low survival rate was probably due to the double-strand breaks in chromosome created by cas9 protein. To test foam production by B. subtilis ATCC 6051a and mutant BS1, cells were grown in a 3 L fermenter for 80 h, and the growth rate of B. subtilis ATCC 6051a and BS1 has no significant difference. During the whole fermentation process, B. subtilis ATCC 6051a produced much foam and required to add antifoam (490 ul) continuously; BS1 produced much less foam at a controllable level that needed 60 ul antifoam. The foam height of B. subtilis ATCC 6051a and BS1 was similar after 42 h, while the foam of BS1 shows great sensitive to antifoam (Fig. 2).
In addition to disruption described above, which deleted a relatively small section of the genome, we also designed and attempted a larger deletion of the srfC gene. An 1100 bp homologous repair template was designed to have a 500 bp upstream region and 600 bp downstream region flanking the Cas9 cleavage site. Using knockout plasmid pHYcas9dsrf2 that containing this repair template to delete a 284 bp region, which includes 44 bp upstream and 240 bp downstream of the PAM sequence ( Supplementary Fig. S2). The efficiency of the 284 bp deletion is 9.1% in B. subtilis ATCC 6051a, which is lower than the efficiency of disruption gene. The low efficiency may related to the short length of homologous repair template that result in low homologous recombination efficiency 28 .
Scientific RepoRts | 6:27943 | DOI: 10.1038/srep27943 Disruption of spoIIAC gene using the CRISPR/Cas9 system. In order to cope with limiting nutrient sources and high cell density, B. subtilis can form highly resistant spores 29 that will germinate and grow in favourable living conditions. These spores drastically hamper B. subtilis fermentations and limit their application in the food industry 30 . The SpoIIAC gene encodes sigma factor F, which permits cells to proceed through stage II of sporulation 6,31-33 . Using knockout plasmid pHYcas9dspo, we disrupted spoIIAC gene using the method described above. The resulting mutant, named BS2, was screened by amplifying the upstream and downstream regions and digesting the PCR products with Xho I (Fig. 3). The homologous regions of mutant BS2 were also subjected to DNA sequencing ( Supplementary Fig. S1b) to confirm the result. The disruption efficiency of SpoIIAC gene was 36% ± 3%. After being cultured at 40 °C for 48 h in a culture medium that favours spore formation, the sporulation efficiency of BS1 was 28.44% (262 colonies/921 colonies), while the sporulation efficiency of BS2 was 0% (0 colonies/182 colonies), which shows that mutant BS2 has great resistant to spore formation.
Disruption of the nprE and aprE genes using the CRISPR/Cas9 system. Strains that lack several extracellular protease genes generally show superior extracellular protein productivity 7,34 . The nprE and aprE genes encode alkaline and neutral extracellular proteases, respectively, in B. subtilis. We constructed the knockout plasmids pHYcas9dnpr and pHYcas9dapr and used them to sequentially disrupt nprE and aprE gene of mutant BS2 using the CRISPR/Cas9 method described above. Amplification of the upstream and downstream regions, and then digesting the PCR products with Xho I allowed us to select the appropriate mutant, named BS3 with disruption efficiency of 53% ± 6% (Fig. 3). The nprE and aprE double gene mutant, named BS4 with disruption efficiency of 33% ± 3%. The homologous regions of mutant BS3 and BS4 were subjected to DNA sequencing ( Supplementary Fig. S1c,d) to confirm the disruption. When cultured on specify agar plates containing 5% non-fat powdered milk, strain BS3 forms a protein clearance zone smaller than the one formed by its parent, mutant BS2, demonstrating that BS3 has reduced protease activity. And strain BS4 shows substantial inhibition of protein degradation on a 5% non-fat powdered milk plate (Fig. 4a).
Disruption of the amyE gene using the CRISPR/Cas9 system. B. subtilis releases many extracellular enzymes during the post-exponential growth phase, and α -amylase is one of the major proteins released 9,35 . These proteins hamper the purification of industrial products. We constructed the knockout plasmid pHYcas9damy and used it to disrupt the amyE gene of mutant BS4 as described above. After amplifying the sequences upstream and downstream of the target region, the PCR products were digested with Xho I to select the amyE deletion mutant, which was named BS5 with disruption of 53% ± 6% (Fig. 3). The homologous regions of mutant BS5 were also subjected to DNA sequencing ( Supplementary Fig. S1d) to confirm the disruption. The results of a starch-plate assay demonstrate that BS5 fails to release α -amylase activity (Fig. 4b,c).  Extracellular expression of β-CGTase using B. subtilis ATCC 6051a and mutant BS5. To evaluate the utility of the strain BS5, in which five genes (srfC, spoIIAC, nprE, aprE and amyE) have been disrupted, as a host for extracellular recombinant protein expression, the ability of mutant BS5 to produce β -CGTase was compared with that of B. subtilis ATCC 6051a. Both expression plasmid pHY300PLK-β -CGTase and empty expression plasmid pHY300PLK were transferred into B. subtilis ATCC 6051a and mutant BS5, and the resulting strains were used in an expression study. After 48 h of cultivation in TB medium, the culture supernatants of B. subtilis ATCC 6051a and BS5 that harbouring empty expression vector PHY300PLK show no β -CGTase activity. After 80 h of cultivation in 3 L fermenter, the highest β -CGTase activity of B. subtilis ATCC 6051a that harbours expression plasmid pHY300PLK-β -CGTase was 110.8 U/ml in 56 h, and the highest dry cell weight (DCW) was 68.8 g/L in 66 h; the highest β -CGTase activity of BS5 that harbours expression plasmid pHY-300PLK-β -CGTase was 277.8 U/ml in 70 h, which was 2.5 times of B. subtilis ATCC 6051a, and the highest DCW was 70.3 g/L in 75 h (Fig. 5).

Discussion
In this study, we established a CRISPR/Cas9 system that can disrupt target genes in B. subtilis ATCC 6051a (Fig. 1c), and then used this system to construct a mutant strain with improved fermentation characteristics. All of the elements required by the CRISPR system are present in a single plasmid that contains a constitutively expressed cas9, a strongly transcribed sgRNA, and a homologous repair template. The sgRNA recognizes a specific site on the B. subtilis genome (Supplementary Table S1) and guides the Cas9 protein to the target genome locus, where it creates a double-stranded break. This is followed by homology-directed repair that utilizes a homologous repair template provided by the knockout plasmid. Since the knockout plasmid contains the PE194 temperature-sensitive replicon, it can be easily cured after the mutation step by incubating the mutants at 51 °C overnight. The mutant colonies cured of the knockout plasmid can be used as the parent strain for additional  genetic modification. With an efficiency of 33% to 53%, the operate procedure of CRISPR system was simple and time-saving compared with the currently existing Bacillus subtilis genome editing methods ( Table 1).
The transformation efficiency in B. subtilis 168 and B. subtilis ATCC 6051a are 1553 ± 213 and 60 ± 13 transformants/μ g of the knockout plasmid, respectively. Its poor competence makes genetic manipulation of B. subtilis ATCC 6051a inconvenient. To increase the transformability of B. subtilis ATCC 6051a, we considered disrupting the gene encoding ComI. However, we were unable to find a target-specific sgRNA target within the 93 bp comI gene. The original knockout plasmid (pHYcas9d) did not contain a homologous repair template; therefore, a homologous repair template was transferred into B. subtilis ATCC 6051a in the form of PCR fragment, along with the knockout plasmid. This attempt did not meet the efficiency required for genetic editing, perhaps because B. subtilis ATCC 6051a was unable to simultaneously take up the knockout plasmid and the homologous repair PCR fragment.
The target recognition of the sgRNA mainly depends on the last 12 bp of the guide sequence; thus, the existence of highly homologous regions in chromosomal DNA may result in off-target effects 36 . These off-target effects have been reported in eukaryotic cells 37 , whereas little attention has been paid to this problem in prokaryotes. Although off-target effects may be less common in bacteria because of their relatively small genome size, bacterial genomes contain some high-homology clusters 38 . There are methods that can be used to reduce the off-target efficiency; for example, designing two sgRNAs to guide the Cas9 protein to cleave the genome at the adjacent sites, using a Cas9 nickase mutant, and making sure the last 12 bp of the guide sequence is highly specific 39 .
Like many Bacillus strains, B. subtilis produces surfactin, which contains a peptide moiety and a β -hydroxy fatty acid side chain 40 . Because surfactin is an amphiphilic molecule, its accumulation at gas-liquid interfaces can lead to foam production 40,41 . The biosynthesis of surfactin is controlled by a non-ribosomal peptide synthase enzyme (SrfC) 42 and a thioesterase/acyltransferase (SrfD) 43 . The internal thoiesterase domain of srfC, which is a one-module enzyme, controls the conversion of a linear lipoheptapeptide to its cyclic form, as well as the release of surfactin 41 . B. subtilis ATCC 6051a produces a large amount of foam during fermentation, which has an extreme affect on fermentation process control and may lead to contamination. Compared with its parent strain, the mutant BS1 produces much less foam and at the level of control. This shows that surfactin may be a major mediator of foam formation, and that there is more work to be done to thoroughly inhibit foam production.
Spore development in B. subtilis is governed by multiple RNA polymerase sigma factors 32 . Sigma F, encoded by SpoIIAC gene, controls the forespore, sigma E controls the early stage of sporulation, and then sigmas G and K control later stages 32,33 . SpoIIAC nonsense mutations can block the processing of sigma E precursor protein P31 to sigma E and prevent transcription of spoIID 31 . The transcription product, SpoIID, is a membrane-anchored enzyme essential for sporulation 6,44 . The observation that mutant BS2 shows great resistant to spore formation demonstrates that inactivation of sigma F can largely block the sporulation of B. subtilis.
Bacillus species produce many different types of extracellular protease to degrade heterologous extracellular proteins 45 , and many protease-deficient strains display favorable heterologous protein production 7,34 . B. subtilis WB600, which is deficient in six extracellular proteases due to knockouts of the nprB, nprE, aprE, mpr, bpr and epr genes. This strain displays a level of recombinant protein secretion higher than that of its parent strain 46 . Recently, a B. amyloliquefaciens strain lacking six extracellular protease genes was constructed. This strain displays improved production of levan and α -amylase 34 . B. subtilis releases multiple extracellular enzymes during the post-exponential period; α -amylase is among the major proteins released 35 . Secretion of a large amount of α -amylase by B. subtilis makes the isolation and purification of recombinant proteins difficult on an industrial scale. In addition, the massive expression of endogenous α -amylase increases secretion stress and influences the production of recombinant proteins 47 .
In summary, we established a CRISPR/Cas9 system in the poorly transformable strain B. subtilis ATCC 6051a, which is an undomesticated strain with favorable growth characteristics. To improve the usefulness of B. subtilis ATCC 6051a as an industrial expression host, we disrupted the srfC, spoIIAC, nprE, aprE and amyE genes with an efficiency of 33% to 53%. Compared with B. subtilis ATCC 6051a, the final mutant (BS5) forms less foam during fermentation, displays greater resistant to spore formation, and secretes 2.5 times more β -CGTase. Thus, the CRISPR system developed here can be used to modify industrially relevant strains with high efficiency, and mutant BS5 can be applied as a superior expression host. Delivery plasmid 58 Counter-selectable marker 59 Cre/loxP 28

Materials and Methods
Strains and plasmids. All bacterial strains and plasmids used in this study are described in Table 2.
Escherichia coli JM109 was used for plasmid construction. Plasmid pHY300PLK-β -CGTase was previously constructed in our laboratory by inserting the β -CGTase gene of Bacillus circulans 251 into the B. subtilis-E. coli shuttle expression vector pHY300PLK (Takara, Dalian, China) 48 . Plasmid pwtcas9-bacterial was purchased from Addgene (Addgene plasmid # 44250) 49 . B. subtilis ATCC 6051a was purchased from the American Type Culture Collection (ATCC). Fermentation Cultivate condition. Seed culture was obtained by inoculating 100 ul frozen glycerol stock (stored at −80 °C) into 100 ml LB medium and then incubating 12 h at 37 °C with shaking at 200 rpm. As for Shake-flask cultivation, a portion of 5% (v/v) seed culture (2.5 ml) was used to inoculate 50 ml of TB medium (Supplementary material and method)  0.86 L fermentation medium (Supplementary material and method), before a 10% (v/v) seed culture (100 ml) was inoculated, its pH was adjusted to 7.0 with 20% (v/v) H 3 PO 4 and NH 4 OH, its temperature was adjusted to 37 °C, and it was added with 30 ml 166.7 g/L glucose. After 2 h of inoculation, the inducer of 30 ml 333.3 g/L lactose was added into fermentation culture as needed. The feed solution (Supplementary material and method) was fed at rate 0 to 15 g glucose/h to confirm the glucose concentration was maintained 0.2 to 0.5 g/L. During the fermentation process, the batch cultivation was carried out at 37 °C, pH 7.0, and 30% dissolved oxygen that were maintained by automatically adjusting stirrer speed (300 rpm-900 rpm) and flow rate of air (1.5-4.0 L/min). Tetracycline (20 mg/L) was added as needed every 24 h. The foam height was measured as needed and antifoam was added manually. For both cultivation, the samples collected at certain time intervals were centrifuged at 12,000 × g for 10 min at 4 °C. For DCW determination, the pellet was resuspended with 0.9% (w/v) NaCl and centrifuged at 12000 × g for 10 min, and dried to a constant weight at 105 °C. As for the host strains that harbour expression plasmid pHY300PLK-β -CGTase, the culture supernatant contains β -CGTase.

Reagents and enzymes.
Plasmids construction. The sequences of all of the primers used in this study are listed in Table 3. The initial CRISPR system knockout plasmid pHYcas9d was assembled from four fragments (One, Two, Three and Four), each of which overlaps its two adjacent fragments by 15 bp, using the In-Fusion HD Cloning Plus kit 50 . Fragment One, which encodes the Cas9 protein, was amplified from plasmid pwtcas9-bacterial using the primer pair P03/P04. Fragment Two, which includes a P43 promoter for sgRNA expression, a target-specific 20-nt guide sequence specific for srfC fused with the sgRNA sequence, and a temperature-sensitive replicon PE194 for plasmid curing, was synthesized and ligated into the cloning vector pMD18-T. This fragment was amplified using the primer pair P07/P08. Fragment Three includes an E. coli replication origin (p15A ori), an ampicillin-resistance marker (ampR) and the α -amylase promoter (PamyQ) from B. amyloliquefaciens. This fragment was amplified from plasmid pHY300PLK-β -CGTase using primers P02 and P05. Fragment Four includes a tetracycline resistance marker (TcR) and a terminator from B. amyloliquefaciens. This fragment was amplified from plasmid pHY300PLK-β -CGTase using primers P01and P06. We amplified the upstream and downstream regions of the target locus using B. subtilis ATCC 6051A genomic DNA as a template and two primer pairs: P09/P10 and P11/P12. The amplified fragments were ligated together using overlap extension PCR, forming homologous repair template. To form the final knockout plasmid, pHYcas9dsrf1, the appropriate homologous repair template was inserted into the Xba I site of pHYcas9d (Fig. 1a). During double-strand repair, this repair template removes 6 bp of native sequence and inserts an Xho I restriction site and 5 bp of random sequence (Fig. 1b). The other five knockout plasmids (pHYcas9dsrf2, pHYcas9dspo, pHYcas9dnpr, pHYcas9dapr and pHYcas9damy) were constructed from pHYcas9dsrf1 by changing the 20-nt guide sequence (except pHYcas9dsrf2) and replacing the homologous repair template. The 20-nt guide sequence was changed using inverse PCR with the four primer pairs P17/P18 (for pHYcas9dspo), P23/P24 (for pHYcas9dnpr), P29/P30 (for pHYcas9dapr), and P35/ P36 (for pHYcas9damy), which hanging the new modified 20-nt guide sequence at the 5′ end. The homologous repair templates were created by overlap extension PCR of sequences upstream and downstream of the 20-nt guide sequence. These sequences were amplified from the B. subtilis ATCC 6051a genome using primer pairs P13/P14 and P15/P16 (for pHYcas9dsrf2), P19/P20 and P21/P22 (for pHYcas9dspo), P25/P26 and P27/P28 (for pHYcas9dnpr), P31/P32 and P33/P34 (for pHYcas9dapr), and P37/P38 and P39/P40 (for pHYcas9damy). Then the appropriate homologous repair template was inserted into the modified pHYcas9dsrf1 by digesting the PCR product and the modified plasmid with Xba I and ligating the appropriate fragments.
Genome editing. B. subtilis competent cells were made by the method of Anagnostopoulos and Spizizen 51 .
Tetracycline-resistant transformants were confirmed by colony PCR of the cas9 gene. Using transformant genomic DNA as the template, verification PCR reactions were carried out using specific primers (Supplementary Table S2) that anneal outside the homologous repair template. PCR products from disruption mutants were identified by digestion with Xho I, and the homologous repair regions of the mutants were subsequently verified by DNA sequencing.  Table 3. Primers used in this study. * The restriction enzyme sites are bold and underlined.