Identification and genomic comparison of temperate bacteriophages derived from emetic Bacillus cereus

Cereulide-producing Bacillus cereus isolates can cause serious emetic (vomiting) syndrome and even acute lethality. As mobile genetic elements, the exploration of prophages derived from emetic B. cereus isolates will help in our understanding of the genetic diversity and evolution of these pathogens. In this study, five temperate phages derived from cereulide-producing B. cereus strains were induced, with four of them undergoing genomic sequencing. Sequencing revealed that they all belong to the Siphoviridae family, but presented in different forms in their hosts. PfNC7401 and PfIS075 have typical icosahedral heads, probably existing alone as phagemids in the host with self-replicating capability in the lysogenic state. PfEFR-4, PfEFR-5, and PfATCC7953 have elongated heads, with the genomes of the former two identified as linear dsDNA, which could be integrated into the host genome during the lysogenic state. Genomic comparison of the four phages with others also derived from emetic B. cereus isolates showed similar genome structures and core genes, thus displaying host spectrum specificity. In addition, phylogenic analysis based on the complete genome and conserved tail fiber proteins of 36 Bacillus species-derived phages confirmed that the phages derived from emetic B. cereus strains were highly similar. Furthermore, one endolysin LysPfEFR-4 was cloned and showed lytic activity against all tested emetic B. cereus strains and cross-lytic activity against some other pathogenic bacteria, implying a potential to control bacterial contamination in the food supply.


Introduction
Bacillus cereus is a common opportunistic pathogen causing clinical cases of food poisoning, i.e., diarrhea and emesis. The latter is associated to cereulide, a single heat-stable cyclic dodecadepsipeptide toxin [1]. As cereulide is toxic to mitochondria due to its action as a potassium ionophore and killer of human natural killer cells, it represents one of the most serious food safety risks of B. cereus [2]. The ecology and evolution of emetic B. cereus has, therefore, raised many concerns. Most cereulide-producing B. cereus isolates, even from different ecological PLOS  niches, are relatively conserved at the chromosomal level, with previous study confining them to a single and homologous evolutionary lineage [3]. In addition, although rare, cereulide-producing Bacillus weihenstephanensis isolates with psychrotolerant features, which similarly belong to the B. cereus group, have also been identified [4]. The genomic variation at the chromosomal level between emetic B. weihenstephanensis and B. cereus is greater than that within the emetic B. cereus lineage. Furthermore, they both show flexible plasmid content [5], as well as different genomic locations (on chromosomes or plasmids) of cereulide biosynthesis gene clusters and their relatedness to mobile genetic elements (MGEs) [6]. In addition to plasmids, prophages represent another extrachromosomal material bearing adaptive genetic determinants and horizontal genetic transfer potential. The exploration of the prophages in emetic B. cereus strains will be helpful for better understanding the evolution and expansion of their ecological niches. Prophages are ubiquitous in many kinds of bacteria, and result from the lysogenic integration of the genomes of temperate phages into their host bacterium's chromosomes or plasmids. They can replicate themselves along with bacterial genomes or exist as an extrachromosomal plasmid. If harboring genes enough to produce a mature virion, the prophage can be excised from the bacterial genome and access the lytic cycle when exposed to UV light or certain chemical agents [7][8][9]. As a type of MGE, prophages significantly contribute to bacterial genetic diversity and endow the host with special characteristics, such as pathogenicity and distinct phenotypic traits through transduction, lysogenic conversion, or active lysogeny driving host evolution [10][11][12][13]. For instance, prophage f11 can lead Staphylococcus aureus to acquire antibiotic resistance through autotransduction [14]. In addition, prophages Gifsy-1 and Gifsy-2 play important roles in the infection ability of Salmonella enterica serovar Typhimurium in mice by carrying virulence or immune-related genes that help their bacterial hosts defend against macrophages [15]. The A118-like prophage regulates Listeria monocytogenes to escape from mammalian cells phagosomes by reversible active lysogeny via insertion into or excision from the competence system master regulator comK gene [12]. Cyanobacterial cells differentiate into nitrogen-fixing cells along with the non-reversible excision of prophage elements from three nitrogen fixation process-related genes under nitrogen-starving conditions [12]. Based on their ubiquitous and unique features, prophages have been used for bacterial typing in Escherichia coli, Streptococcus pneumonia, and Bacillus anthracis [16][17][18]. However, despite a number of B. cereus group phages being isolated and sequenced, only limited information on emetic B. cereus phages and genomic data have been reported (e.g., Tp250, vB_BceS-IEBH, and 11143) [8,19,20], hindering further research on the coevolution and interaction of cereulide-producing B. cereus and its related phages.
In this study, we describe the general features of temperate bacteriophages originating from emetic B. cereus isolates and genomic insights resulting from comparative and functional gene analysis. Five bacteriophages were isolated and characterized. The whole genomes of four bacteriophages were sequenced and compared with those derived from Bacillus species isolates. One endolysin was cloned and characterized. This study will help in furthering our understanding of the temperate phages from emetic B. cereus isolates as well as provide a possible strategy to control this pathogen.

Bacteriophage induction and morphological observation
With the application of mitomycin C, bacterial lysis was observed during the growth of emetic B. cereus strains NC7401, IS075, EFR-4, EFR-5, and ATCC7953, indicating the induction of their prophages (S1 Fig). Furthermore, five phages, named PfNC7401, PfIS075, PfEFR-4, PfEFR-5, and PfATCC7953, were observed by transmission electron microscopy (TEM) from their concentrated lysates (Fig 1). All belonged to the Siphoviridae family in Caudovirales. Both PfNC7401 and PfIS075 had typical icosahedral heads (55.11−57.90 nm) and long non-contractile tails (2.9−3.5 nm × 170.83−70.95 nm), whereas PfEFR-4, PfEFR-5, and PfATCC7953 had elongated heads (44.32−57.66 nm × 88.61−100.23 nm) and long non-contractile tails (5.88−8.82 nm × 241.43−269.11 nm). The major structural protein profiles of PfNC7401 and PfIS075 were identical, as were those of PfEFR-4, PfEFR-5, and PfATCC7953 (data not shown). Therefore, the major structural proteins of PfIS075 and PfEFR-4 were identified through LC-MS/MS, and were found to contain a capsid protein, tail protein, tail fiber protein, DNA packaging protein such as portal protein, and some uncharacterized proteins, as listed in Fig 2. Host range and plaque formation The host range was evaluated by the lysis zone formed from a droplet of the purified phages (titer > 10 5 PFU/mL) onto the bacterial lawns. The tested strains sensitive to PfNC7401, PfIS075, PfEFR-4, PfEFR-5, and PfATCC7953 all belonged to emetic B. cereus, except ATCC10987 and Schrouff (Table 1), indicating both species and strain specificity. The strain displaying the strongest sensitivity was used as the propagation host for the corresponding phage. PfNC7401, PfIS075, and PfATCC7953 formed small and clear plaques, whereas PfEFR-4 and PfEFR-5 formed large and clear plaques on their propagation host strains, respectively (Fig 3). 81 and 79 putative open reading frames (ORFs), respectively. Alignments of the two phages and their host genomes, which were sequenced previously, were performed. The genome of PfNC7401 demonstrated a complete match to the sequence of pNC1 (Accession No. NC_ 016772.1) of B. cereus NC7401, which was originally annotated as a plasmid, and PfIS075 also matched accurately to one scaffold (Accession No. AHCH02000044.1) of the gapped genome of IS075 (data not shown). PfEFR-4 and PfEFR-5 had a genome of 43,223 and 43,773 bp, both with ca. 35.4% GC content, encoding 67 and 69 putative ORFs, respectively (S1 Table).
Restriction profiles (S2 Fig) and PCR results using the primers listed in S2 Table revealed that the genomic DNA of PfIS075 and PfNC7401 were circular, whereas PfEFR-4 and PfEFR-5 exhibited linear dsDNA (S3 Fig). All four phages had mini-satellite DNA, mainly related to DNA replication proteins (data not shown). No tRNA, rRNA, or sRNA were identified in the genomes. The genomes of PfEFR-4 and PfEFR-5 were almost identical, except that PfEFR-4 had only one integrase, whereas PfEFR-5 had one more (ORF31) upstream of the lysogeny control module. PfIS075 and PfNC7401 also shared great similarity, though differences were observed within two hypothetical proteins, of which PfIS075-ORF33 and PfNC7401-ORF44 had similar N-terminal sequences belonging to the DUF2479 superfamily encoding a putative long tail fiber, and PfIS075-ORF34 had no match in PfNC7401.
b Strains were used as the propagation host for the corresponding phage.
No. indicates GenBank accession number of phages. ND means that the information was not determined.
The host lysis regions of the five phages were analyzed. The endolysin of PfEFR-4 shared 95% similarity to that of Tp250 and vB_BceS-IEBH, with an amidase 02_C cell wall binding domain (CBD) and a peptidoglycan recognition protein (PGRP) N-terminal catalytic domain (EBD). The endolysin of PfIS075 contained a SH3_3 CBD and PGRP EBD, and the endolysin of 11143 (ORF20) had a putative GH25_PlyB-like EBD and amidase 02_C homologous CBD.
In addition to the common DNA replication proteins of the phages, there was a plasmidlike region, including a potential replication protein and plasmid segregation protein (ParM) in PfIS075 (ORF44-45), with the latter as an actin homologue driving plasmid partition and DNA segregation by polymerizing to filaments [27, 28]. The highly similar plasmid-like region was also presented in Tp250 and vB_BceS-IEBH, which has self-replicating capability in vivo [19,20,29]. Their existence within the hosts was evidence for these phages as phagemids.

Phylogeny of 36 Bacillus phages
The genomes of 36 phages derived from Bacillus spp., listed in S3 Table with their GenBank accession numbers and basic phage properties, were aligned and compared using Gegenees [30]. The heat-plot based on the nucleotide alignment revealed three clusters (A-C) and four singletons (Fig 5)  Furthermore, comparative phylogenetic analysis using tail fiber proteins (Fig 6), which are a crucial part of phages initiating absorption and infection to the host [31-36], displayed similar classification to that based on the genome sequences of the derived phages. Remarkably, phages derived from the emetic isolates shared conserved tail fiber proteins and similar genome structures, corresponding to their host spectrum specificity. Cloning and characterization of lysins in vitro The predicted lysins PfEFR-4-ORF24 (named LysEFR-4) and PfNC7401-ORF49 (named LysNC7401) were cloned and expressed heterologously. The purified recombinant protein LysEFR-4 was visualized as a clear protein band of ca. 32 kDa on SDS-PAGE gel, in line with the predicated size (data not shown). Bioinformatics analysis showed that LysEFR-4 carried a PGRP superfamily EBD domain and a putative amidase 02_C CBD domain. LysEFR-4 could significantly lyse the derived emetic bacterial cells, causing a decrease in absorbance at 600 nm due to 60% and 80% cell death (Fig 7A and Fig 7B). In addition, LysEFR-4 could lyse all 15 tested emetic B. cereus and B. weihenstephanensis isolates, and could also cross-lyse some other Bacillus species strains (Fig 7C).
Although putative lysNC7401 was cloned into four positive-selection vectors (i.e., pET28a, pQE30, pGEX-6p-1, and pMAL-C2x), after transformation into E. coli only a few colonies were able to grow. Sequencing these recombinants showed that they all harbored one or several random mutations, which occurred in the active center of the predicted functional domain

Discussion
The B. cereus group members, including B. cereus, B. thuringiensis, B. anthracis, B. weihenstephanensis, Bacillus mycoides, Bacillus pseudomycoides, and Bacillus cytotoxicus, share similar chromosomal backbones and biochemical characters [38][39][40][41][42][43][44][45]. Their classification is mainly based on distinctive pathotypes and ecotypes, which are determined by the accessory genes exclusively owned by each species. The MGEs (e.g., large plasmids, transposons, insertion sequences, and phages) play important roles in the acquisition of highly specialized accessory genes, and thus some are used for differentiating or typing species. A typical example is B. anthracis, whose fatal anthrax is genetically determined by its pXO1 and pXO2 plasmids harboring pathogenicity islands with the specific anthrax virulence genes flanked by insertion sequences [46]. Furthermore, as well as the pXO1 and pXO2 plasmids, the B. anthracis genome contains four putative lambdoid prophages, which can be used for differentiating B. anthracis from other B. cereus members [18]. Previous study has shown flexible plasmid content of the emetic B. cereus group strains [5], and relatedness between the "ces" cluster and transposons [6]. However, phages have received less attention in terms of their potential contribution to distinctive ecotypes and pathotypes. In this study, we focused on the characterization of the phages preying on emetic B. cereus isolates. This study will enrich our knowledge on bacterial genetic variability and the evolution of emetic B. cereus group isolates. The phages induced from the five emetic strains all belonged to the Siphoviridae family, but presented in different forms in their hosts. PfNC7401 and PfIS075 had typical icosahedral heads and their genetic materials were circular dsDNA containing both phage and plasmid replication genes. Due to their high similarity with vB_BceS-IEBH, whose genome had a plasmid-like region and phagemid state, PfNC7401 and PfIS075 should exist alone in plasmid form in the host, with self-replicating capability in the lysogenic state, and produce active phages through induction. Indeed, the alignment of PfNC7401 to the host genome makes a complete match to one replicon (plasmid) of B. cereus NC7401. Two other intact prophages (ca. 49−51 kbp) were predicted within the chromosome of B. cereus NC7401 using PHAST (data not shown), which showed no obvious similarity with PfNC7401 and no induction to active phages in this study. Although no complete genome was available for IS075, the current alignment results indicated that PfIS075 might match to one plasmid replicon of IS075, but not to the other predicted prophages (ca. 48 −53 kbp) located in the chromosome (data not shown). PfEFR-4, PfEFR-5, and PfATCC7953 had elongated heads, with the genomes of PfEFR-4 and PfEFR-5 being linear dsDNA, whereas that of PfATCC7953 was undetermined. The replication region of PfEFR-4 and PfEFR-5, containing only DnaB and DnaC and lacking other replication factors, suggests it is not autoreplicated, but integrated into the host genome during the lysogenic state.
Although displaying diverse sizes and DNA forms, the phages derived from emetic B. cereus isolates showed similar genome structures and core genes. Phylogenic analysis based on the complete genomes and conserved tail fiber proteins of 36 Bacillus phages confirmed that the phages derived from emetic B. cereus strains were highly similar, which corresponded to their host spectrum specificity. This suggests there is a coevolutionary relationship among the prophages and the hosts. Previous study has reported that prophages of tectivirus GIL01 and GIL16 greatly contribute to the adaptation of the host to the environment, and the cholera toxin, which is produced by filamentous phage CTXf, endows the host pathogen with cholera virulence [47,48]. Whether the Siphoviridae prophages explored in this study are associated with the hosts' pathotypes and ecotypes remains unclear, and the host-phage coevolution needs to be further studied.
At the end of their life cycle, the phage endolysin degrades the host cell wall with the help of the membrane protein holin, releasing mature virions [49]. Thus, the lysis spectrum of a phage is related with that of the endolysin, though not exactly. The process under which phages infect bacteria and produce progeny viruses is influenced by the invaders as well as the hosts, which own a series of defense systems such as CRISPER-Cas, restriction-modification, and resistant mutations [50]. However, lysins can directly lyse Gram-positive bacteria in vitro due to exposing the peptidoglycan, which is the outermost cell wall structure. Therefore, it is reasonable that the antimicrobial spectrum of endolysin is usually wider than the host range of the phage. The host of PfEFR-4 was restricted to emetic B. cereus, whereas LysEFR-4 could not only lyse all the tested emetic B. cereus group isolates, but also showed cross-bactericidal activity against other Bacillus strains and Gram-negative pathogen Salmonella enteric var. Typhimurium, implying a potential of endolysin as an additional antimicrobial agent combined with other additives to control food contamination. An exhaustive attempt was carried out to clone ORF49 encoding endolysin in PfNC7401 in E. coli. However, almost all clones screened contained mutations (ca. 95.7% mutation rate) in the predicted active center of the enzyme. Furthermore, after addition of IPTG to the only plasmid without mutation, the E. coli BL21 became clear. This suggests that the endolysin of PfNC7401 might be toxic or have some lysis activity against E. coli, and therefore, only the clones with mutations could survive. Even if one correct recombinant plasmid existed by chance, the recombinant bacteria containing the plasmid could not express abundant proteins in the cell [37].

Bacterial strains and growth conditions
The bacterial strains used for phage induction and host spectrum analysis in this study consisted of 24 B. cereus group isolates isolated from food, animal, clinical, and soil samples, and four from other species. Their characteristics, origins, genome accession numbers, and sources/references are listed in Table 1. The Bacillus strains in this study were grown in standard Luria-Bertani (LB) medium at 30˚C, and the other strains, including Staphylococcus aureus and S. Typhimurium, were grown at 37˚C.

Induction of bacteriophages and host range assay
200 μl of overnight bacterial culture was transferred into 50 mL of fresh LB medium to subculture to the logarithmic phase (OD 600 = 0.2−0.3), after which mitomycin C (Sigma, USA) with a final concentration of 1 μg/mL was added. Bacterial lysis was observed after 1 h of incubation at 30˚C. Until the OD 600 of the bacterial culture did not decrease, the cell lysates were centrifuged for 15 min at 10,000 × g and at 4˚C. The supernatant was then passed through a 0.22-μm pore size syringe filter (Millipore, USA) and stored as crude extract of the phage at -80˚C and 4˚C until use. The host range of the phages were tested with the drop method on 26 B. cereus group strains and four strains from other species, including B. subtilis 168, Lysinibacillus sphaericus C3-41, S. aureus RN4220, and S. Typhimurium O901. The lysis zones were observed after 10 μL of phage lysates were plotted on soft agar (0.5%) with bacterial lawns and incubated at 30˚C overnight.

Propagation and purification of the phages
Propagation and purification of the phages were performed as per previous methodology [51], with some modifications. The B. cereus strain most sensitive to infection, as per the spot test under visual observation, was used as the propagation host of the corresponding phage. Accordingly, EFR-4 was used to propagate PfEFR-4, PfEFR-5, and PfATCC7953, EFR-5 was used to propagate PfIS075, and LH001 was used to propagate PfNC7401. The propagation strain with the phage was incubated on double layer plates until plaques formed. One plaque was picked and suspended in 100 μL of SM buffer (50 mM Tris-Cl, pH 7.5; 100 mM NaCl; 10 mM MgSO 4 Á7H 2 O), then mixed with 200 μL of propagation strain culture (OD 600 % 1.0) and incubated for 30 min at 30˚C. Subsequently, the mixtures were added to 5 mL of molten semisolid medium with soft agar (0.5%) kept at 47˚C and poured onto solid medium (1.5%) after thorough mixing, then incubated at 30˚C overnight. For purification, the phages were passaged at least five times until homogeneous plaques were formed. Then, the purified temperate bacteriophages were propagated using the method described above, except that bacteriophage suspensions with a titer above 10 5 PFU/mL were used. After plaques were formed, 5 mL of SM buffer was added onto the plate and left at 4˚C for 4 h with moderate rotation. The suspension was then recovered and centrifuged at 4˚C for 15 min at 10,000 × g. The supernatant was filtered through a 0.22-μm sterile filter, and the phage preparations were stored at -80˚C and 4˚C until further use.

Morphological observation of the phages under TEM
The phage preparation was centrifuged at 118,000 × g for 2 h at 4˚C (Beckman), and the pellet was resuspended in SM buffer. Then, 20 μL of the phage suspension (10 10−11 PFU/mL) was immediately deposited onto copper grids with carbon-coated Formvar films for 3−5 min. The grids were then stained with 2% phosphotungstic acid solution (pH = 7) for 3 min and examined with a H-7000FA TEM (Hitachi, Tokyo, Japan) at an acceleration voltage of 75 kV.

Bacteriophage DNA extraction
Genomic DNA from the phages was extracted according to previous methods [7], with minor modifications. The phage suspension was treated with DNase I (7 U/mL) and RNase A (20 μg/ mL) at 37˚C for 1 h to remove the bacterial DNA and RNA, respectively. Then, 20 μL of ZnCl 2 solution (2 M) was added and the solution was incubated at 37˚C for 5 min. After centrifugation at room temperature for 1 min at 10,000 × g, the supernatant was removed and the pellet was suspended in 500 μL of TES buffer (0.1 M Tris-Cl, pH 8.0; 0.1 M EDTA; 0.3% SDS) and incubated at 60˚C for 15 min for lysis of the phage particles. After the addition of 20 μL of proteinase K (20 mg/mL) and incubation at 37˚C for 90 min, 60 μL of 3 M potassium acetate solution (pH 5.2) was completely mixed with the suspension and then kept on ice for 15 min. To remove proteins from the phage suspension, the mixture was treated with phenol/chloroform/ isoamyl alcohol (25:24:1, v/v) twice, and then with chloroform/isoamyl alcohol (24:1, v/v) once. Subsequently, phage DNA was precipitated with an equal volume of isopropanol and washed with 70% ethanol, and finally dissolved in 10−20 μL of distilled water. The genomic DNA isolated from the phages was examined by 0.6% agarose-gel electrophoresis.

DNA restriction and structural protein analysis
One microgram of phage DNA was digested for 3 h at 37˚C with EcoRI, BamHI, and PstI (Takara), respectively, in appropriate restriction buffer. The digested products were then analyzed by 0.6% agarose-gel electrophoresis. The structural proteins of the five purified phages were preliminarily visualized by 15% SDS-PAGE. To better illustrate the proteomics of the phages, PfIS075 and PfEFR-4 were boiled for 10 min and then underwent 10% SDS-PAGE. After the samples were fully condensed and the bromophenol blue had migrated about 1 cm in the separating gel at 80 V for 1.5 h, the gel containing all the proteins was cut, reduced, alkylated, and trypsin-digested for peptide LC-MS/MS identification, as reported previously [20].
The lytic activity of LysEFR-4 was tested by turbid reduction and plate lysis assays [64][65][66]. In brief, B. cereus EFR-4 was grown to the logarithmic phase at 30˚C in LB medium, and then harvested by centrifugation at 10,000 × g for 1 min at room temperature and resuspended in 20 mM Tris-Cl (pH 9.0) to adjust the OD 600 to 0.8−1.0. We then added 0.6 μM LysEFR-4 to the bacterial suspension, with an equal volume of dialysis buffer added to the control. Lytic activity measured using the turbid reduction assay was then monitored by the decrease in absorbance at OD 600 at 37˚C for 30 min with a multimode reader (Bio-Tek Synergy HT, Winooski, VT, USA), whereas the lytic activity using the plate lysis assay was measured via the relative bacteria survival rate at 37˚C for 1 h by counting the growing colonies (CFU/mL) of the mixture with bacterial cells and endolysin divided by that of the control without endolysin. To determine the antimicrobial spectrum of the endolysin, we measured lytic activity monitored by turbid reduction on 28 bacteria, including B. cereus group isolates (covering B. cereus, B. thuringiensis, B. anthracis, B. weihenstephanensis, and B. mycoides), B. subtilis, L. sphaericus, S. aureus, and S. Typhimurium. When treating the Gram-negative bacterium, 0.1 M EDTA was added to penetrate the outer membrane to access the peptidoglycan of the cell wall.
Supporting information S1