A novel non prophage(-like) gene-intervening element within gerE that is reconstituted during sporulation in Bacillus cereus ATCC10987

Gene rearrangement is a widely-shared phenomenon in spore forming bacteria, in which prophage(-like) elements interrupting sporulation-specific genes are excised from the host genome to reconstitute the intact gene. Here, we report a novel class of gene-intervening elements, named gin, inserted in the 225 bp gerE-coding region of the B. cereus ATCC10987 genome, which generates a sporulation-specific rearrangement. gin has no phage-related genes and possesses three site-specific recombinase genes; girA, girB, and girC. We demonstrated that the gerE rearrangement occurs at the middle stage of sporulation, in which site-specific DNA recombination took place within the 9 bp consensus sequence flanking the disrupted gerE segments. Deletion analysis of gin uncovered that GirC and an additional factor, GirX, are responsible for gerE reconstitution. Involvement of GirC and GirX in DNA recombination was confirmed by an in vitro recombination assay. These results broaden the definition of the sporulation-specific gene rearrangement phenomenon: gene-intervening elements are not limited to phage DNA but may include non-viral genetic elements that carry a developmentally-regulated site-specific recombination system.

Bacterial endospore formation is an adaptive response to environmental changes such as nutrient exhaustion 1 . At the beginning of sporulation, asymmetric cell division produces two distinct cell types: the forespore and mother cell. The mother cell nurtures the forespore to maturity during spore morphogenesis, with a number of sporulation-related genes controlled at the transcriptional, translational, and posttranslational level 2, 3 . In such genetic regulatory systems, the gene rearrangement phenomenon represents different aspects, in terms of participation of exogenous genetic elements, of host gene expression control. In the process, an exogenous genetic element interrupting a sporulation-related gene is precisely excised from the host chromosome in a sporulation-specific manner, leading to the reconstitution of the intact and functional host gene [4][5][6][7][8][9][10][11][12][13][14] . To date, various gene-intervening elements have been found within sporulation-related genes in Bacilli 11,12,14 and Clostridia 10,13 , suggesting that gene rearrangement is a common and widely used phenomenon in spore forming bacteria. In particular, sigK-intervening elements (skin) are distributed among these bacterial strains 11 .
Gene-intervening elements that were previously reported to contain phage-related genes [9][10][11]15 are considered to be defective prophages that have lost the ability to produce infectious phage particles. This is supported by our recent report that both infectious and defective SPβ prophages in Bacillus subtilis and B. amyloliquefaciens can induce the rearrangement of spsM, whose gene product is involved in the production of spore surface polysaccharides 12,14 . During gene rearrangement, the serine-type DNA recombinase encoded in the element recombines the disrupted gene segments through a site-specific recombination reaction [6][7][8][12][13][14] . The reaction is most likely identical to phage excision, where the DNA recombinase recognizes and recombines the approximately 50 bp specific DNA motifs located upstream and downstream of the gene segments, which are called attachment sites (attL and attR), comprising the short direct and inverted repeat sequences 14,16,17 . The phage-encoded serine recombinases is composed of 225 nucleotides, which encodes a member of the LuxR-FixJ family transcriptional factors (Fig. 1a). The coding region is divided into two segments by an intervening element; the 5′ half (5′-gerE; formerly BCE4626; Fig. 1b, green) and 3′ half (gerE-3′; formerly BCE4594; Fig. 1b, purple) of gerE Bc encode the 66 and 42 aa proteins corresponding to the N-and C-terminal regions of the entire GerE Bc , respectively. An overlapping sequence, 5′-CATCTCAAA-3′, is found at the end of 5′-gerE and at the beginning of gerE-3′ (Fig. 1b, box), which is expected to be the integration site of the intervening element. The intervening element, named gin (gerE-intervening), is an exogenous genomic island with a lower GC content compared with the B. cereus ATCC10987 genome 24 . The gin element consists of 31 genes (Fig. 2a), most of which have no significant homologies with previously characterized proteins except the restriction/modification system (BCE4604 and BCE4605) 25 and a recJ homologue (BCE4610). While the B. cereus ATCC10987 genome contains three prophage elements, the gin element is unlikely to be derived from prophages due to the lack of bacteriophage-related proteins ( Fig. S1; Table S1). Another notable feature of the gin element is that it contains three serine recombinase genes unlike the prophage(-like) gene-intervening elements, which possess a single site-specific recombinase gene for gene rearrangement 11 . girA and girB (gin-encoded recombinase) form an operon, while girC is distant from the operon. The sequence homology for all three recombinases was low (approximately 22-26% identity; alignments of their amino acids sequences are shown in Supplementary Figure S2). The gin element is found in some B. cereus and B. toyonensis strains; the gin element in B. cereus FT9 is identical to that in ATCC10987 at the nucleotide sequence level (Fig. 2b). All of the gin elements identified possess all three recombinase genes even though their gene contents are different. In addition, recJ is also conserved in the B. cereus strains (ATCC10987, FT9, AH820, and Q1), suggesting that three recombinases and recJ may be required for maintenance or mobilization of the elements in B. cereus. The composite sigK and gerE genes after the rearrangement encode a sporulation-specific sigma factor, σ K , and a transcriptional factor, GerE, respectively, which govern mother cell-specific gene expression at the late stage of sporulation. RNAP, RNA polymerase. (b) Whole nucleotide sequence of gerE from Bacillus cereus ATCC10987. Deduced amino acid sequence is shown under the nucleotide sequence. The gin element is inserted at position 138-146 nt (indicated by the red box). Nucleotides shaded green and purple correspond to the coding regions of 5′-gerE and gerE-3′, respectively. START, start codon; STOP, stop codon. Sporulation-specific gerE rearrangement in B. cereus ATCC10987. Initially, we examined whether the gerE rearrangement occurred during sporulation in B. cereus ATCC10987. The 28.6 kb gin element was expected to be excised during sporulation as shown in Fig. 3a, given that the rearrangement created the intact gerE gene. B. cereus ATCC10987 cells were cultured at 37 °C in sporulation medium (DSM) to induce sporulation. The sporulation process in B. cereus ATCC10987 was very similar to that in B. subtilis. Engulfment began 3-4 hours after the onset of the sporulation (Fig. S3, T 3 , and T 4 ), and bright forespores appeared in the sporangia at T 6 and during the later stages. Consequently, free spores were released at T 12 . Chromosomal DNA was extracted from the vegetative (T -1 ) and sporulating (T 6 ) cells ( Fig. 3b; Fig. S3). The primers specific for 5′-gerE and gerE-3′ (Fig. 3a, arrows) were designed to amplify the DNA fragment containing the junction site of the composite gerE and the excised gin (Fig. 3a, attB and attG). The 925 bp PCR product was successfully generated only during the sporulation phase, but not during the vegetative phase (Fig. 3c, attB; Fig. S7a). Using the gin-specific primer set, the junction site of the excised gin element (attG; 690 bp) was also amplified only from chromosomal DNA in sporulating cells (Fig. 3c, attG, spo; Fig. S7a). However, the excision of gin was not induced when mitomycin C, a phage inducer, was added (Fig. 3d, MMC+; Fig. S7a), which is in contrast to the SPβ prophage-mediated spsM rearrangement in B. subtilis 12 . These results suggested that the B. cereus gerE rearrangement is a sporulation-specific event. We then addressed the timing of the rearrangement during sporulation. Because the PCR-based assay was too sensitive to small differences in the sporulation state, we performed Southern blotting to determine the precise time point when the rearrangement commenced. Southern blotting using the gerE probe (Fig. 3a, thick line, gerE probe) detected a single 2.8 kb band, indicative of the 5′-gerE, starting during the vegetative stage (T -1 ) and up until three hours (T 3 ) after the onset of the sporulation (Fig. 3e, top panel, T -1 -T 3 ; Fig. S7a). By contrast, at T 4 , a second 4.5 kb band began to appear and remained during later time points (T 4 -T 8 ), indicating the rearrangement. At the same time, the excised gin element was detected (Fig. 3e, bottom panel, T 4 -T 8 , 2.1 kb band). These results indicate that the gerE rearrangement commenced at the middle stage of sporulation. Consistent with this, the B. subtilis sigK and spsM rearrangements have been reported to take place at the mid-phase of sporulation, after chromosome segregation and compartmentalization is complete 4,5,12 .
To investigate whether the disrupted gerE segments were combined precisely, the four junction sites before and after to the DNA recombination were analyzed using a DNA sequencer (Fig. 4a). The sequencing data demonstrated that the nucleotide sequences of 5′-gerE and gerE-3′ were exchanged within the 9 bp consensus nucleotide sequence (5′-CATCTCAAA-3′; Fig. 4a, box), which encodes Ile-Ser-Asn at position 47-49 of the entire GerE Bc (Fig. 4b). This suggests that DNA recombination takes place within the consensus sequence. The 27 bp imperfect inverted repeat sequence was found around the consensus sequence (Fig. 4b, arrows), which is likely to contain the recombinase binding sites. Alignments of GerE proteins showed that the B. cereus composite GerE (GerE Bc ; 74 aa) was almost identical to B. subtilis GerE (GerE Bs ) with only three substitutions at the 5 th , 6 th , and 34 th residues (Fig. 4c, asterisks), implying that the intact GerE Bc was functional.
DNA rearrangement produces a functional gerE. In B. subtilis, GerE is required for the transcription of genes encoding spore outer-coat and germination receptor proteins, and depletion of the GerE activity results in the production of lysozyme-sensitive and germinant-insensitive spores 26 . To evaluate the activity of truncated and composite GerE in vivo, we conducted a gerE complementation test using the B. subtilis gerE-deletion mutant strain, GEd, and its derivatives harboring the truncated (5′-gerE; GEd-5) and composite gerE (gerE Bc ; GEd-C). A P cotG -lacZ construct was introduced into the strains, generating 168 (wild type), GEd (ΔgerE Bs ), GEd-5 (ΔgerE Bs , 5′-gerE + ), and GEd-C (ΔgerE Bs , gerE Bc + ) strains to monitor the cotG promoter activity, which is positively regulated by GerE 23,27 . During sporulation, cotG transcription began at T 5 and reached its maximum at T 7 in the wild-type strain (Fig. 5a, filled circle), while depletion of GerE Bs failed to activate cotG transcription (Fig. 5a, open circle). This result was consistent with a previous report 27 . The composite gerE Bc was able to compensate for the gerE Bs -deletion with a LacZ activity of 83.5% relative to that of the wild-type strain (Fig. 5a, open triangle), while 5′-GerE was not functional (Fig. 5a, filled triangle). A substitution in the 5 th residue of B. subtilis GerE has been reported to decrease the transcription rates from the GerE-dependent cotC promoter 22 . Therefore, the slight reduction in the cotG transcription level in GEd-C may have been caused by the differences in amino acid residues between GerE Bs and GerE Bc (Fig. 4c, asterisks). Next, we examined the germination rates and lysozyme sensitivities of the spores produced from wild type, GEd, GEd-5, and GEd-C strains. Addition of a germinant, L-alanine (1 mM), decreased the OD 600 values of the wild-type and GEd-C spores (Fig. 5b, filled circle and open triangle, respectively), which was a result of the reduced spore refractivity as germination was initiated. The wild type spore exhibited a 39% reduction in the OD value 120 min after the addition of L-alanine, whereas the GEd and GEd-5 spores showed only approximately 18% reductions (open circle and filled triangle, respectively), which indicated the insensitivity of the GEd and GEd-5 spores to the germinant. In addition, the GEd and GE-5 spores showed a significant loss of resistance to lysozyme (Fig. 5c, open circle and filled triangle, respectively). Taken together, these results indicate that the B. cereus composite GerE Bc is functional, while the truncated 5′-GerE prior to DNA rearrangement has no activity as a transcriptional factor. GirC and GirX are required for gerE rearrangement. Previous studies described DNA recombination during sporulation-specific gene rearrangement events as catalyzed by serine recombinases encoded in the gene-intervening elements [5][6][7][8][11][12][13] . Because gin carries three serine recombinase genes, we addressed which recombinase was responsible for the gerE rearrangement. Due to the difficulty in genetically manipulating B. cereus ATCC10987, the B. cereus gerE encompassing the entire gin element was introduced into the amyE locus of B. subtilis 168 (168Gin; Fig. 6a, top), which enabled the introduction of mutations into the gin element. In 168Gin, the gin element was capable of gerE reconstitution during sporulation (Fig. 6b, 168Gin, spo; Fig. S7b). Using 168Gin, we created the girAB-and girC-deletion strains (Fig. 6a, GABd and GCd, respectively). The gerE rearrangement was not observed in the girC-deletion strain (Fig. 6b, GCd, spo; Fig. S7b), while girAB was dispensable for the rearrangement (GABd, spo), indicating that only GirC was required for the rearrangement.
Overexpression of girC at the vegetative stage did not result in gerE rearrangement in a girC-inducible strain (Fig. S4, GC-i). This suggested that additional factor(s) for gerE rearrangement are controlled in a sporulation-specific manner. To localize the factor(s), GR1-3 strains, carrying partial deletions of gin elements, were constructed (Fig. 6a) and examined to determine their ability to induce the gerE rearrangement. As a result, GR1 and GR3 successfully generated the composite gerE Bc (Fig. 6b, GR1 and GR3, spo; Fig. S7b), while GR2 showed no PCR product from the composite gerE Bc (GR2, spo), indicating that the additional factor was encoded in a region from BCE4620 to BCE4625 (Fig. 6a, shaded area). Out of six genes in the BCE4620-BCE4625 region, we found that BCE4620 was conserved in the gin elements from B. cereus ATCC10987, AH820, Q1, and B. toyonensis BCT-7112 (Fig. 6c); hence, we designated BCE4620 as girX. As shown in Fig. 6d, girX encodes a 42 aa protein with no conserved motif, but showed high similarity with the gin-encoded proteins from B. cereus AH820 (BCAH4598), Q1 (the corresponding ORF is not annotated in the public databases, but located at the intergenic region of BCQ4304-BCQ4305), and B. toyonensis BCT-7112 (located at the intergenic region of Btoyo1736-Btoyo1737). Introduction of B. cereus ATCC10987 girX into the thrC locus of GR2 recovered the rearrangement ability (Fig. 6e, GR2X; Fig. S7b). These results suggest that GirC and GirX are required for gerE rearrangement.    sporulation, and girC, girX, and gerE cDNA was generated by reverse transcription using gene specific primers (Fig. 7a). PCR products from the girC and girX cDNA were detected at T 3 -T 5 (Fig. 7b; Fig. S7c), and their signal intensities peaked at T 4 , which was consistent with the timing of gerE rearrangement (Fig. 3e). This indicates that the accumulation of sufficient amounts of GirC and GirX is required for the gerE rearrangement although the transcription begins at T 3 . The RT-PCR data suggests that the gerE rearrangement is a sporulation-specific reaction catalyzed by GirC and GirX. Transcription of the composite gerE Bc occurred at T 5 and T 6 , following the rearrangement (gerE Bc , T 5 and T 6 ). By contrast, girAB expression was not detected at any time between the vegetative state and sporulation phase (girAB, T −1 -T 6 ); hence, girAB expression may be induced under specific conditions other than sporulation.
Establishment of the in vitro gerE rearrangement system. To verify the GirC/GirX-mediated gerE rearrangement, an in vitro recombination system was established. As shown in Fig. 8a, we constructed two plasmids harboring either the attB-tet-attG (pMDTIn) or attL-tet-attR (pMDTEx) sites to use as DNA substrates. The site-specific DNA recombination reaction between attB and attG in pMDTIn (or attL and attR in pMDTEx) was expected to produce two circular DNA molecules of 1,871 bp containing attR (or attG) and 2,947 bp containing attL (or attB). Recombinant GirC and GirX fused with a his 6 -tag at their C-termini were expressed in E. coli and purified by affinity chromatography. The purified GirC and GirX were detected as single 58 kDa and 6.5 kDa bands, respectively ( Fig. 8b; Fig. S7d). pMDTIn (200 ng) was reacted with various amounts of GirC (0-1.5 μM) at 37 °C for 1 hr, and separated by agarose gel electrophoresis. Signals of the integrative recombination products, the circular DNA containing attL and attR, were detected in a GirC-dose dependent manner, and peaked at the GirC concentration of 1 μM (Fig. 8c; Fig. S7d). Excessive amounts of GirC decreased the yields of the recombination products (Fig. 8c, 1.25 and 1.5 μM GirC), which was consistent with previous reports on phage-encoded serine recombinases 14,28 . In the excision reaction, pMD-TEx (200 ng) was reacted with various amounts of GirX (0-4 μM) in the presence of GirC (1 μM) at 37 °C for 1 hr. The recombination products were detected when both GirC and GirX were added to the reactions (Fig. 8d, 2-4 μM GirX; Fig. S7d). The addition of 4 μM GirX generated the recombination product most efficiently when mixed with 1 μM GirC (lane 8). By contrast, neither of them showed excision activity when alone (lanes 2 and 9). This supports the results shown in Fig. 6. Unlike the excision reaction, the integration reaction using pMDTIn was only catalyzed by GirC (Fig. 8e, lane 2; Fig. S7d); the addition of GirX blocked the integrative recombination (lane 4). These results are consistent with phage-medicated site-specific recombination, whereby phage integrases catalyze the recombination between phage and host DNA and the recombination directionality is controlled by an additional recombination directionality factor (RDF) 16,18 . We therefore concluded that GirC is the integrase that catalyzes the recombination between the gin element and the host gerE locus, and that GirX is the RDF for GirC. Furthermore, we found that GirC catalyzed the integration reaction between attG and B. subtilis gerE (attB Bs ) with approximately 40% efficiency compared with the recombination with B. cereus gerE (Fig. 8f, lane 4; Fig. S7d). Surprisingly, the integration site was identical to that of gerE Bc , although differences in nucleotides were found around the recombination site (Fig. S5, red font). The in vitro study raised the possibility that the gin element is potentially mobile between B. cereus and B. subtilis.

Discussion
In the present study, we demonstrated gin-mediated gerE rearrangement in B. cereus. Sporulation-specific expression of girC and girX (Fig. 7) induced site-specific DNA recombination (Figs 3 and 4) to produce the functional gerE (Fig. 5). Our results indicate that the gene rearrangement system encoded in the exogenous element is successfully incorporated into the gene expression program during sporulation in B. cereus.
We found that the att sites for GirC contain a 27 bp imperfect inverted repeat and the 9 bp consensus sequence (Fig. 4b, arrows and box). Generally, attP sites for phage-encoded serine recombinases comprise a symmetric sequence of the inverted repeat motif with respect to the central dinucleotides 16,17 . The inverted repeat motif is the recognition site of the recombinase and DNA cleavage occurs at the central dinucleotides during DNA strand exchange. In the case of gin, the DNA cleavage site is predicted to be the 3′-end diadenine nucleotides within the 9 bp consensus sequence (Fig. 4b, attG, 5′-CATCTCAAA-3′; underlined) at the middle of the attG site inverted sequence.
In gerE rearrangement, an additional factor, GirX, was required in combination with GirC (Fig. 6), which controlled the directionality of the GirC-mediated recombination reaction (Fig. 8). These data demonstrated that gerE rearrangement occurs in a similar manner to prophage-mediated gene rearrangements, such as B. subtilis SPβ prophage 12 and C. difficile prophage-like skin Cd 13 , in which serine recombinases play the central role in the reaction. However, there are differences in regulation; the serine recombinases are expressed constitutively in B. subtilis SPβ 12 and C. difficile skin Cd 13 , while being expressed specifically at the sporulation phase for the B. subtilis skin 8 , B. weihenstephanensis vfbin 11 , and B. cereus gin elements. In the cases of B. subtilis SPβ and C. difficile skin Cd , their RDFs are under the control of sporulation-specific sigma factors. The B. cereus gin element proves that non-viral mobile elements, not only prophage-like elements, can be gene-intervening elements if they harbor the site-specific recombination system and are integrated into sporulation genes. As an interesting example, a prophage can be found integrated into the gerE locus in B. glycinifermentas BGLY (GenBank accession number, LT603683). If the rearrangement is verified in the strain, it will strongly support the idea that both mobile elements and prophages have the potential to induce gene rearrangements. A question of why those mobile elements targeted the gerE-coding region, which consists of only 225 nucleotides, is an important issue of interest. In the previous study on spsM rearrangement, we argued that spsM was targeted by SPβ phage because a unique symmetric sequence within the spsM-coding region was the preferred integration site of the phage integrase, SprA 14 . Unlike spsM, no symmetric sequence was found at the attB site (Fig. 4b); however, the attG site was probably selected as the recombination site because of the inverted repeat encompassing the 9 nucleotide sequence shared with the gerE-coding region. Functionally, GerE activates the late mother cell-specific genes, including spsM, while it represses σ K expression 20,23 . Therefore, gerE disruption by mobile elements would be beneficial for the host to block expression leakage at earlier stages during sporulation, which could disturb the gene expression program. Such physiological benefits for the host might have been given priority over the nucleotide preference of mobile element-encoded recombinases, followed by adaptation of the recombinases to the precise site recognition in the evolutionary history.
The gin element in B. cereus ATCC10987 is identical at the nucleotide level to the element in the thermotolerant B. cereus strain, FT9 29,30 . This implies that the gin element was transferred across the strains very recently; however, the transfer mechanism is still unknown. The non-viral element, gin, may have spread among the strains by horizontal gene transfer, such as conjugation and natural competence. In B. cereus ATCC10987, the gin element itself has no structural genes for conjugation, while the host plasmid, pBc10987, carries a series of genes for conjugative transfer-like proteins 29 . The gin element, therefore, might be horizontally transferred in association with the plasmid-borne conjugation system. From the results of the in vitro recombination assay (Fig. 8), it is plausible that the gin element is integrated following transfer into the new host genome via the GirC-catalyzed DNA recombination. As shown in Fig. 7b, weak expression of girC was observed at T 1 , while the girX expression was not. This suggests that girC may be controlled by multiple regulation mechanisms, which would be necessary for the gin element to integrate into new host genome. Control of the transfer is also a significant issue. In B. subtilis, an integrative and conjugative element, ICEBs1, employs ImmR and ImmA to control excision and transfer 31,32 . ImmR is a transcriptional factor that represses the transfer, while ImmA is a metalloproteinase that specifically degrades ImmR. In response to DNA damage and cell-to-cell signaling, ImmA is activated to lyse ImmR, thereby causing derepression of ICEBs1 excision, and consequently, transfer. Homologues of ImmR and ImmA are found in many mobile elements 31 . In the gin element, BCE4623 and BCE4624 encode a putative transcriptional factor and a metalloproteinase, respectively, and therefore are considered as candidates for the regulators of excision, or may be remnants of the conjugation system derived from the ancestor of the gin element.
Because girAB was not expressed under our conditions and was dispensable for the gerE rearrangement, the role of girAB remains to be clarified. The GirA and GirB from B. toyonensis show 83-86% identity with the serine recombinases from the B. thuringiensis MC28 plasmid, pMC319 33 , while GirC has 52% identity with the integrase encoded by the B. glycinifermentas prophage inserted in the gerE locus (Fig. S6). The gin element is likely to be a composite of mobile elements derived from the pMC319 plasmid and the prophage. In addition, GirC and GirX are highly conserved in B. cereus and B. toyonensis strains (81-85% identity), whereas GirA and GirB from B. cereus strains have lower homologies with those from B. toyonensis (approximately 30% identity), indicating that B. cereus girAB was derived from mobile elements other than pMC319. Well-known examples of composite mobile elements include Staphylococcal chromosomal cassettes (SCCs) 34, 35 , which are clinically significant genetic elements because they convey methicillin resistance genes (mec complex) among Staphylococcal strains. The serine recombinase genes encoded in SCCs are called chromosomal cassette recombinase (ccr). The SCCs are classified based on variations in the ccr and mec complex; however, composites of different SCCs are also found 34, 36 . ccrA and ccrB are always found together as a two-gene operon in SCCs. Their gene products can catalyze both integration and excision reactions without the RDF, and can also recombine canonical pairs of att sites (e.g., attB × attS CC and attL × attR) and non-canonical pairs (e.g., attS CC × attS CC and attR × attR), which is thought to allow the fusion of multiple SCCs 37 . From this viewpoint, girAB may have contributed to composite gin elements. Alternatively, because girAB are encoded in pMC319, they might be a multimer resolution system for circular DNA. Further functional analysis of girAB will be required to understand the mechanism of horizontal gene transfer and co-evolution of the gin element and the host.

Methods
Bacterial strains and plasmids. The bacterial strains and plasmids used in this study are listed in Table S2.
Primers used in this study are shown in Table S3. Bacillus cereus and B. subtilis strains were routinely incubated at 37 °C in Luria-Bertani (LB) medium with shaking. Transformation of B. subtilis was performed through natural competence. Detailed description of B. subtilis strain construction is presented in Supplementary Methods. Plasmids were constructed using the E. coli strain DH5α. For culture of E. coli strains harboring plasmids, 50 or 100 μg/ml ampicillin was added to the LB medium. Fluorescence microscopy. B. cereus cells were cultured in DSM containing FM4-64 (2.5 μg/ml) and SYTO16 (66 nM) to stain the cell membrane and chromosomal DNA, respectively. Microscopic observation was performed using an Olympus BX50 microscope with a 100× Uplan Apo objective lens. Images were captured and processed, using Metamorph software version 7.6.5 (Metamorph Inc., TN, USA).

Sporulation of
Detection of the gerE rearrangement. Chromosomal DNA was extracted from the B. cereus vegetative and sporulating cells, according to the method described previously 11 . The junction sequences at the attB (composite gerE) and attG (the excised gin element) sites were amplified from 100 ng of the chromosomal DNA with the primer sets PA246/PA249 and PA247/PA248, respectively. PCR products were separated by agarose gel electrophoresis, and stained using EZ-Vision DNA Dye (AMRESCO, OH, USA). Gel images were cropped, using Paintgraphic2 (SOURCENEXT, Tokyo, Japan). DNA sequencing was performed, using an ABI 3500 DNA SCieNTiFiC REPoRTS | 7: 11426 | DOI:10.1038/s41598-017-11796-8 analyzer (Thermo Fisher Scientific, WI, USA) with the PA246 (for the 5′-and composite gerE) or PA248 (for 3′-gerE and the excised gin element) primers and the BigDye ® Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific).

Southern blotting.
To generate the digoxigenin (DIG)-labeled probes, DNA fragments corresponding to parts of the gerE and gin element adjacent of the junction sites were amplified from the B. cereus chromosomal DNA with the primer sets PA246/PA306 and PA309/PA310, respectively. PCR products were gel-purified and labeled by the incorporation of DIG-11-dUDP using the DIG High Prime kit (Roche, Mannheim, Germany). Five micrograms of chromosomal DNA from B. cereus was digested by HindIII overnight, separated by 1.2% agarose gel electrophoresis, and transferred onto a Hybond-N + membrane (GE Healthcare, NJ, USA) by the capillary method using 10 × SSC. Hybridization and detection were performed using the DIG-labeled probes, anti-DIG antibody conjugated to alkaline phosphatase (Roche), and a nitro-blue tetrazolium/5-bromo-4-chloro-3-indolyl-phosphate solution (NBT/BCIP; Roche). Blot images were cropped and converted to gray scale, using Paintgraphic2 (SOURCENEXT).
β-galactosidase assay. The 168-Z, GEd-Z, GEd-5Z and GEd-CZ strains were sporulated at 37 °C in liquid DSM. The cultures were collected at various time points once the exponential phase of growth had ended. β-galactosidase activity was measured using the method described previously 38 .
Lysozyme sensitivity assay. Spores from B. subtilis 168, GEd, GEd-5, and GEd-C were purified as described previously 12 , resuspended in distilled and deionized water (DDW), and adjusted to 0.5 optical density at 600 nm (OD 600 ). The spore resuspensions were incubated at 37 °C in the presence of lysozyme at a final concentration of 250 μg/ml. Lysozyme sensitivity was determined by a decrease in the OD 600 .
Germination assay. The spore resuspensions (OD 600 = 0.5) were heated at 70 °C for 30 min, 1 mM L-alanine was then the added and further incubated at 37 °C. Germination rates were determined by measuring the decrease in the OD 600 .

RT-PCR.
Total RNA was extracted from B. cereus sporulating cells using glass beads as described previously 39 .
Preparation of DNA substrates for in vitro recombination. The DNA substrates for an in vitro recombination assay were generated by inserting the attL-tetracycline resistant gene (tet) -attR or attB-tet-attG constructs into the pMD20 T-vector (Takara Bio, Kyoto, Japan) to produce pMDTEx and pMDTIn, respectively. A DNA fragment containing tet was amplified from pHY300 PLK (Takara Bio) with the PA541/PA542 primer set and digested with EcoRI. A Taq polymerase reaction at 72 °C for 10 min without primers was performed to fill the over-hanging end produced by EcoRI and to add an adenine at the 3′-end. The resulting DNA fragment was ligated with the pMD20 T-vector using the TA-cloning method to produce a pMDT plasmid. DNA fragments containing attL, attR, attB, and attG were obtained by PCR using the primer sets PA770/PA771, PA772/ PA773, PA770/PA773, and PA772/PA771, respectively. The attR and attG fragments were digested with EcoRI and BamHI, and inserted into the EcoRI-BamHI site of the pMDT, while the attL and attB fragments were digested with HindIII and NdeI and inserted into the HindIII-NdeI site of the pMDT to create the pMDTEx and pMDTIn plasmids, respectively. To construct pMDTIn-Bs, the attB region of pMDTIn (attB Bc ) was replaced by HindIII/NdeI digestion following insertion of a DNA fragment containing the gerE-coding regions from B. subtilis 168 (attB Bs ), which was amplified from the chromosomal DNA using the PA805/PA806 primer set. The plasmids were propagated in E. coli DH5α. The purified plasmids were used as substrates in the in vitro recombination assay.
Preparation of GirC and GirX proteins. girC and girX were amplified by PCR with the primer sets PA705/ PA706 and PA702/PA703, respectively. PCR products were digested with NdeI and XhoI, and cloned into the NdeI-XhoI site of the pET22b(+) vector (Merck Millipore, MW, USA) to obtain the pET-girC and pET-girX expression vectors. E. coli cells harboring the expression vectors were grown at 30 °C in LB medium containing 100 μg/ml ampicillin to an OD 600 of 0.5. The recombinant GirC and GirX proteins with a his 6 tag fused to the C-termini were induced by addition of 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 30 °C for 3 hrs. The cells were then harvested by brief centrifugation and resuspended in a solution containing 50 mM sodium phosphate (pH 8.0), 0.3 M NaCl, 1× FastBreak reagent (Promega, WI, USA), 100 μg/ml lysozyme, and 100 μg/ml DNase I. To remove cell debris, the cell lysate was centrifuged at 20,400 g for 10 min and the supernatants were collected. The E. coli lysate containing his 6 -tagged GirC was loaded onto a HisTrap HP column (GE Healthcare), washed with 50 mM sodium phosphate (pH 7.4), 1 M NaCl, and 10 mM imidazole, and then bound proteins were eluted with 50 mM sodium phosphate (pH 7.4), 0.3 M NaCl, and 500 mM imidazole. The eluted fraction was directly loaded to a Heparin HP column (GE Healthcare), washed with 50 mM sodium phosphate (pH 7.4) and 0.5 M NaCl, and GirC was eluted with 1 M NaCl. The E. coli lysate containing the his 6 -tagged GirX was loaded onto a TALON column (Takara Bio), washed with a solution containing 50 mM sodium phosphate (pH 8.0) and 0.3 M NaCl, and eluted with 250 mM imidazole. Protein concentrations were measured using the Bradford quantification kit (BioRad, CA, USA) with BSA as the standard.