Diverse Temperate Bacteriophage Carriage in Clostridium difficile 027 Strains

Background The hypervirulent Clostridium difficile ribotype 027 can be classified into subtypes, but it unknown if these differ in terms of severity of C. difficile infection (CDI). Genomic studies of C. difficile 027 strains have established that they are rich in mobile genetic elements including prophages. This study combined physiological studies, electron microscopy analysis and molecular biology to determine the potential role of temperate bacteriophages in disease and diversity of C. difficile 027. Methodology/Principal Findings We induced prophages from 91 clinical C. difficile 027 isolates and used transmission electron microscopy and pulsed-field gel electrophoresis to characterise the bacteriophages present. We established a correlation between phage morphology and subtype. Morphologically distinct tailed bacteriophages belonging to Myoviridae and Siphoviridae were identified in 63 and three isolates, respectively. Dual phage carriage was observed in four isolates. In addition, there were inducible phage tail-like particles (PT-LPs) in all isolates. The capacity of two antibiotics mitomycin C and norfloxacin to induce prophages was compared and it was shown that they induced specific prophages from C. difficile isolates. A PCR assay targeting the capsid gene of the myoviruses was designed to examine molecular diversity of C. difficile myoviruses. Phylogenetic analysis of the capsid gene sequences from eight ribotypes showed that all sequences found in the ribotype 027 isolates were identical and distinct from other C. difficile ribotypes and other bacteria species. Conclusion/Significance A diverse set of temperate bacteriophages are associated with C. difficile 027. The observed correlation between phage carriage and the subtypes suggests that temperate bacteriophages contribute to the diversity of C. difficile 027 and may play a role in severity of disease associated with this ribotype. The capsid gene can be used as a tool to identify C. difficile myoviruses present within bacterial genomes.


Introduction
Clostridium difficile is a nosocomial pathogen responsible for gut inflammation, with resultant diarrhoea or pseudomembranous colitis. Infection caused by hypervirulent strains such as C. difficile ribotype 027 is a serious global challenge [1,2]. A deletion in the tcdC gene of C. difficile ribotype 027 is thought to result in uncontrolled toxin production [3]. Furthermore, 027 strains are often fluoroquinolone resistant and sporulate readily, both of which may contribute to the success of the pathogen [4,5]. A recent report showed that C. difficile 027 isolates could be divided into 23 and 5 subgroups using multiple-locus variable-number tandem repeat analysis (MLVA) and pulsed-field gel electrophoresis (PFGE), respectively [6]. These subtypes were associated with variable disease severity. It is important to establish factors that may contribute to the diversity and success of this hypervirulent ribotype and consequently to the epidemic potential of new strains.
Temperate bacteriophages have been shown to play several key roles in the evolution of pathogenic bacteria either by supplementing or modifying toxin production [7]. Studies have shown that phage infection of toxigenic C. difficile strains can cause an increase in toxin production [8,9,10]. Similarly, reports on Clostridium perfringens have shown that there may be a link between phage carriage and sporulation [11]. Several temperate bacteriophages have been isolated from C. difficile using mitomycin C as an inducing agent [12,13,14]. Alternative antibiotics such as norfloxacin have been shown to be highly effective as a prophage inducing agent in E. coli [15]. Although few studies have compared the efficacy of multiple agents to induce prophages in bacteria, clear differences in E. coli phage induction were seen using different inducing agents [16].
Five temperate C. difficile phages have been fully sequenced and annotated. Three belong to the Myoviridae (phiC2, phiCD119, and phiCD27) and two to the Siphoviridae (phiCD6356 and phiCD38-2) [9,17,18,19,20]. In addition, prophages have been identified in most C. difficile strains sequenced to date [17,21]. Although temperate myovirus genes such as those that encode for the capsid, portal and holin proteins have been shown to share nucleotide homology, they have not previously been used in order to determine phylogenetic relationships within C. difficile bacteriophages.
Recent genomic analysis has identified mobile genetic elements including prophages in 22 of 25 sequenced C. difficile 027 strains, but the bacteriophage content was not characterised further [21]. In another study, when a C. difficile 027 outbreak strain was investigated, only single bacteriophage morphology was observed [14]. No previous studies have examined bacteriophage carriage within C. difficile ribotype 027 subgroups. To determine if bacteriophages are contributing to the diversity within C. difficile 027, the range of temperate bacteriophage types associated with ribotype subgroups must first be established. We aimed to isolate and characterise temperate bacteriophages from clinical C. difficile 027 subtype isolates, to determine if phage carriage correlates with subtype, and to design a molecular maker for the identification of phages.

C. difficile isolates used in this study
The 91 Clostridium difficile 027 isolates used in this study were collated by Prof. Mark Wilcox, University of Leeds, UK. They were isolated from C. difficile toxin positive faecal samples that were submitted as part of routine diarrhoea surveillance in 9 hospitals in England. The isolates have previously been divided into 23 MLVA and 5 pulsovar types using MLVA and PFGE, respectively [6].
Other C. difficile ribotypes used in this study were 001, 002, 005, 014, 015 and 020. These were isolated from C. difficile toxin A and B positive faecal samples collected from University Hospitals of Leicester NHS Trust, Leicester, England. Faecal samples were added to a 50% solution of industrial methylated spirit in water, mixed thoroughly and left to stand at room temperature for 1 h. This material was then used to inoculate Braziers CCEY agar. C. difficile colonies were identified by their morphology, horse manure smell, yellow/green fluorescence under long wave uv light and agglutination using a latex agglutination test (Oxoid, Hampshire, UK).

PCR ribotyping
Identified C. difficile colonies were subcultured on Brain Heart Infusion (BHI) (Oxoid, Hampshire, UK) medium supplemented with 7% defibrinated horse blood for 48 h. Bacterial DNA was extracted using Chelex 100 Molecular Biology Grade Resin (BioRad Laboratories, Carlifornia, USA) as previously described [6,22]. PCR ribotyping was performed as described by Stubbs et al. with modifications [23]. C. difficile 16S-23S intergenic spacer regions were amplified using cdiffFP3 59-CTGGGGTGAAGTCGTAACAAGG-39 (forward) and cdiffRP4 59-GCGCCCTTTGTAGCTTGACC-39 (reverse) oligonucleotide primers. The reaction mixture was reduced in volume to ,20 ml on a heating block at 75uC. PCR products were separated in 3% (w/v) RESponse Regular PCR Agarose gel (Bioplastics, The Netherlands) prepared in 16 Tris-acetate-EDTA (TAE, pH 8), containing 0.1 ml/ml GelRed (Biotium, Hayward, Carlifornia, USA). A 100-bp molecular standard (Fermentas, York, United Kingdom) was used and the gel was run in TAE buffer for 4 h (150 mA). C. difficile ribotypes were determined by visualizing on agarose gels and comparing resultant PCR ribotype profiles with those of known ribotypes (obtained from C. difficile Ribotyping Network for England and Northern Ireland reference laboratory at Leeds).

Prophage induction
About 10 ml overnight BHI broth cultures of the 91 C. difficile 027 isolates were exposed to mitomycin C (Fisher Scientific, Loughborough, UK) or norfloxacin (Sigma-Aldrich, Dorset, UK) at a final concentration of 0.3, 1, 3, 6 and 9 mg/ml with shaking at 100 rpm for 24 h. Cultures were then centrifuged at 34006 g for 10 min. The resultant supernatants were passed through a 0.22 mm filters and stored at 4uC until further analysis.

Transmission electron microscopy (TEM)
A 5 ml volume of each induced prophage suspension was placed on individual glow discharged pioloform/carbon coated copper grids which were allowed to stand for 5 min for bacteriophage to bind. The grids were blotted with Whatman filter paper, rinsed with 10 ml of double distilled water, blotted and stained with 10 ml of 1% w/v uranyl acetate. The stain was added one drop at a time, and allowed to stand for 5-10 s. Excess stain was removed by blotting leaving a thin film of suspension on the surface of the grids. The grids were allowed to dry for approximately 5 min and examined using JEOL 1220 electron microscope with an accelerating voltage of 80 kV. Digital images were captured using SIS Megaview III Digital camera with associated analysis software.

Phage purification
C. difficile 027 broth cultures were induced and centrifuged in a 45 ml volume as previously described for prophage induction. The supernatants were decanted into sterile 50 ml centrifuge tubes and NaCl was added to a concentration of 1 M. The cultures were mixed and incubated on ice for 1 h. A 10% (w/v) of polyethylene glycol (PEG) 8000 (Fisher Scientific, New Jersey, USA) was slowly added with continuous stirring at 4uC until completely dissolved. The treated supernatants were stored at 4uC overnight and then centrifuged at 14, 3346 g for 10 min at 4uC. The resultant pellets were resuspended in 1 ml of SM buffer (10 mM NaCl, 8 mM MgSO 4 .7H 2 O, 50 mM Tris-Cl, pH 7.5), washed with an equal volume of chloroform and centrifuged at 3, 3986 g for 15 min. The upper aqueous phases were collected and filtered through 0.22 mm filters. The filtrates were used for PFGE analysis and DNA extraction.

Pulsed-field gel electrophoresis (PFGE)
Agarose plugs were made by mixing 40 ml of PEG purified phage lysates and 60 ml 2% (w/v) Seaplaque CTG agarose (Cambrex Bio Science Wokingham, Berkshire, UK) in 0.56 Tris-Borate-EDTA (TBE, pH 8). This was dispensed into 120 ml capacity PFGE molds and solidified at 4uC for 30 min. The plugs were then removed and incubated in lysis buffer containing 100 mM EDTA, 100 mM Tris-Cl pH 9.0, 1% SDS, and 0.5 mg/ ml proteinase K at 55uC overnight. After three washes in 16 TE (10 mM Tris-Cl pH 7.5, 1 mM EDTA pH 8.0), the plugs were placed in a 1% (w/v) Pulsed-field certified megabase agarose (BioRad Laboratories, Carlifornia, USA). The products were separated in 0.56 TBE buffer using a Bio-Rad CHEF-DR-II Pulsed-Field Electrophoresis System (Bio-Rad, Richmond, CA) at 6 V/cm for 15 h with a pulse ramp from 5-13 s at 14uC. The low range PFG marker (New England Biolabs, Herts, UK) was used as a molecular ladder. Gels were stained with GelRed (Biotium, Hayward, Carlifornia, USA). A digital image of the gel was captured using a SynGene camera assembly and GeneSnap software.
Phage DNA extraction, primer design, PCR, cloning and sequencing Phenol chloroform extraction and isopropanol precipitation was used for phage and genomic DNA extraction [24]. Four C. difficile phage sequences were used to design degenerate primers for the myovirus major capsid gene ( Figure S1). Three of the four C. difficile gene sequences CD630, phiC2 and phiCD119 were obtained from NCBI and the fourth (phi12) was a partial DNA sequence of a C. difficile temperate myovirus kindly donated by Katherine Hargreaves (University of Leicester). A BLAST search for each of these sequences was performed to determine the specificity of the gene. The sequences were aligned using ClustalW and candidate sequences for the forward and reverse primers were manually selected between 11-29 nucleotide and 345-365 nucleotide regions respectively ( Figure S1). The forward and reverse primers selected for targeting the capsid-encoding sequences were CDMCapF1 59-CACTARMKTAYGSA-MAAGWW-39 and CDMCapR1 59-CWRTAAGCAT-CYATCTCTGG -39 respectively. Polymerase chain reactions were performed in a total volume of 50 ml, containing 0.25 mM dNTPs, 3 mM MgCl 2 , 2 mM forward and reverse primers, 16 PCR buffer, ,50 ng of template DNA and 0.5 U of Taq polymerase (Bioline, London, UK). Amplification conditions were: 94uC for 2 min, 30 cycles of 94uC for 45 sec, 48uC for 45 sec, 72uC for 1 min, with a final extension of 10 min at 72uC. PCR products were gel-purified using NucleoSpin Extract II (Fisher Scientific, Leicestershire, UK), and subjected to TOPO TA cloning (Invitrogen, Paisley, UK). Plasmid extraction was performed using the Sigma GeneElute plasmid miniprep kit (Sigma-Aldrich, Dorset, UK). Plasmids were digested with EcoR1 (New England Biolabs, Herts, UK) to confirm the inserts and sequencing was carried out by GATC Biotech using M13 primers. Sequencing results were edited using Chromas 2.33 and translated into amino acids using Transeq sequence analysis in EMBL-EBI. The nucleotide and protein sequences were analysed using nucleotide and protein BLASTs respectively (NCBI) and phylogenetic analysis was carried out using MEGA 5.

Results
Morphological diversity of temperate C. difficile 027 bacteriophages TEM analysis following induction revealed a diverse set of phage morphologies associated with C. difficile ribotype 027, with a clear bias towards myoviruses, which were induced from 63 isolates ( Figure 1, Table S1). Intact myoviruses (A) were only induced from two isolates and all had hexagonal capsids of ,70 nm in diameter which were joined to the tail by a short portal protein ( Figure 1). The tails were ,200 nm by 20 nm with horizontal striations on the sheaths which end in 5-6 ,50 nm long tail fibres. The two isolates yielding these intact myoviruses also yielded defective myoviruses (Morphologies B and C). These 'defective myoviruses' were the only viral particle morphology induced from a further 59 isolates (Table S1). These typically had contracted tail sheaths with the wider part of the sheath at various positions along the remaining tail tube (Morphologies B and C) ( Figure 1). Some of the defective myoviruses had very long tail fibres such as those of morphology D ( Figure 1). The remaining one isolate that yielded a defective myovirus also yielded a siphovirus. A novel C. difficile myovirus (E) was isolated from one isolate. This particle had a small hexagonal capsid of ,40 nm diameter and tail of approximately 20 nm in diameter and 220 nm long ( Figure 1). The capsid was joined to its tail by a thin portal protein like structure of ,10 nm in length and width. The tubular tail had an average of 55 striations 63 (based on an average of six) and the sheath ended with around 6-7 tail fibres.
Intact siphoviruses typified by F were released from three of the isolates ( Figure 1, Table S1). These all had a hexagonal capsid ,70 nm in diameter with a flexible ,230-350 nm tail. Dual phage carriage was observed in four isolates; one isolate had a defective myovirus and a siphovirus (66L), two had a defective myovirus and an intact myovirus (96L and 91L) and finally one had two morphotypes of siphoviruses 48LM (Table S1). The two morphotypes of siphoviruses had the same capsid diameters but differed in their tail length. One had a tail length of ,230 nm and the other a tail length of ,350 nm (Table S1).
Twenty-five isolates produced phage tail-like particles only. There were two discrete categories of phage tail-like particles identified. One category (G) had particles which looked like genuine phage tails and were ,160 nm by 20 nm terminating in horizontal protrusions (Figure 1). These phage tail-like particles were produced by all isolates (in addition to other morphologies) (Table S1). Two isolates harboured particles with morphology belonging to the second category (H) (Figure 1). This particle type had tightly coiled spirals approximately 20 nm diameter and between 80-350 nm long. Only one isolate contained no prophage particles.

Norfloxacin and mitomycin C prophage induction
To optimize prophage induction, five concentrations (0.3, 1, 3, 6 and 9 mg/ml) of norfloxacin and mitomycin C were used on ten of our samples. All of the induced cultures released prophages with 0.3, 1 and 3 mg/ml of antibiotic and there was an increase in phage abundance within this antibiotic concentration range. Although most cultures were poorly induced at 6 mg/ml, one sample (16L) produced the highest phage abundance at this concentration. All cultures were killed after 24 h of induction at 9 mg/ml and no phages were observed in these culture filtrates. We therefore established that 3 mg/ml concentration of either antibiotic is optimal for C. difficile phage induction. We also observed that morphologies of the induced phage particles did not vary with antibiotic concentration.
The growth responses of the 91 induced cultures and the efficacy of the two antibiotics (at 3 mg/ml) to cause prophage induction were compared ( Figure S2, Table S1). Intact phages were induced from approximately 61% of the isolates with both antibiotics. However, 8% and 31% of the isolates yielded intact phages with mitomycin C and norfloxacin, respectively. In four isolates, mitomycin C and norfloxacin induced different prophages from their hosts. Isolate 48L yielded two siphoviruses morphotypes following mitomycin C induction, but only one morphotype following norfloxacin induction. Similarly, isolate 66L yielded two different phage morphologies (a siphovirus and a defective myovirus) with norfloxacin but only a defective myovirus with mitomycin C. Isolate 96L yielded a myovirus and a defective myovirus when induced with norfloxacin and phage tail-like particles with mitomycin C. Finally, isolates 91L yielded a myovirus with mitomycin C but induction with norfloxacin yielded a myovirus and a defective myovirus (Table S1).

Correlation of phage carriage with ribotype 027 subtypes
A correlation between phage carriage and the MLVA types was observed in this study (Table 1). MLVA 1-15 yielded defective myoviruses apart from three isolates (96L, 66L and 91L) which in addition to defective myoviruses contained a myovirus (96L, 91L) or a siphovirus (66L). MLVA 16,17,19,20,21,22 and 23 were found only to contain phage tail-like particles with one exception each in MLVA 16 (isolate 16L contained a novel myovirus E) and MLVA 22 (isolate 36L which had no phage particles under the induction conditions above). There were three isolates classified as MLVA 18 and two had inducible siphoviruses and one, a defective myovirus ( Table 1, Table S1). Pulsovar I (containing 14 different MLVA types, 53 isolates) was strongly associated with the production of defective myoviruses. All contained defective myoviruses except isolate 71L which contained only phage taillike particles. Pulsovar II was only represented by one isolate and contained a defective myovirus. Two isolates from Pulsovar III were examined and both contained siphoviruses ( Table 2, Table  S1). Thirty-three isolates from Pulsovar IV were examined and they all yielded phage tail-like particles. In addition, ten isolates also yielded defective myoviruses. Pulsovar V consists of two isolates; one contained phage tail-like particles and the other (36L) contained no phage.
Genetic diversity of C. difficile 027 temperate bacteriophages PFGE analysis revealed that phage genome size ranged from ,15 to 50 kb ( Figure 2). The smallest genomes were seen in the defective myoviruses B and C from isolates 96LN2, 84LN, 91LN, and 66LM (Lanes IV, V, VI and VII respectively) and the novel myovirus E (lane VIII). (Suffixes N and M symbolise isolates induced with norfloxacin and mitomycin C, respectively). Typical myoviruses A induced from isolates 96LN1 and 91LM (Lanes IV and IX respectively) had a genome size of ,30 kb as well as the siphoviruses isolated from 37LM (Lane II), 48LM2 (Lane III) and 37LN (Lane X). Only the siphovirus isolated from 48LM1 and one defective myovirus isolated from 82LN had a large genome size of ,50 kb (Lanes III and XI respectively). Two isolates (48LM and 96LN) shown to previously have dual phage carriage exhibited genomes of two distinct sizes, 50 and 30 kb for 48LM1 and 48LM2 (Lane III) and 30 and 15 kb for 96LN1 and 96LN2 (lane IV) respectively. This provides good evidence of their dual phage status. On the other hand, morphologically distinct phages that were induced from the same isolate but using different antibiotics showed different genome sizes (91LM with ,30 kb genome size and 91LN ,15 kb). Furthermore, in all cases examined, the siphoviruses which were morphologically identical under TEM also showed identical genome sizes (for example 48LM2, 37LN and 37LM having genome size of ,30 kb).
Phylogenetic analysis of C. difficile phages PCR was performed on a range of C. difficile clinical isolates including ten C. difficile 027 and six other ribotypes (014, 005, 002, 020, 015 and 001) to amplify the capsid gene of the phage, followed by cloning and sequencing of the inserts. In all the isolates examined the portion of the capsid gene sequence was 358 nucleotides which covers 40% of the gene. The degenerate primers were also used to amplify genomes of three other C. difficile 027 strains (CD196, QCD-66c26 and R20291), ribotype 012 (CD630) and phiC2 using in-silico PCR (http://insilico.ehu.es/ PCR/). Results indicated that the primers were specific to C. difficile myoviruses or lysogenic hosts harbouring these prophages. Nucleotide BLAST (blastn) combined with protein BLAST (blastp) confirmed the capsid sequences. Phylogenetic analysis of both the nucleotide and amino acid sequences showed that all C. difficile isolates cluster together, forming the C. difficile clade. Within this, all C. difficile 027 strains (including CD196, QCD-66c26 and R20291 from in-silico PCR) have identical sequences and therefore form a single subclade (Subclade I) (Figure 3). Two other distinct subclades can be defined from the analysis of the capsid gene, subclade II and III. Subclade II is comprised of ribotypes 005, 002, 020 and 015 and Subclade III is comprised of C. difficile 001, C. difficile 012 (CD630) and C. difficile phage phiC2 [17,21]. Within Subclade III C. difficile 012 (CD630) and phiC2 formed a much tighter clade [17]. The ribotype 014 sequence falls outside the other C. difficile ribotypes. Regarding the outgroup taxa; the E.coli phage, C. perfringens, Lactobacillus and Bacillus clades are distinct from each other and from the C. difficile clade (Figure 3).

Discussion
Given the excess morbidity and mortality associated with C. difficile ribotype 027, we investigated prophage content as a factor that may be contributing to its diversity and virulence. Bacteriophage carriage within a single C. difficile ribotype has not been extensively examined. Previous work examined six isolates from different C. difficile ribotypes and found that each ribotype harboured a morphologically and genetically different prophage and suggested a correlation between the type of prophage and ribotype [14]. However, the examination of 91 ribotype 027 isolates show that prophage carriage within C. difficile 027 is morphologically and genetically variable and therefore may contribute to diversity.
The morphology of the putative myovirus A identified in our study is similar to the temperate phage induced from a clinical isolates of C. difficile 027 and to phages found in unknown ribotypes [13,25]. To our knowledge, this is the only temperate bacteriophage morphology previously reported to be associated with ribotype 027. However, despite the capsid and tail diameters being similar to the previously described phage, the viruses described here have longer tail lengths suggesting that they may be different [12,14]. The defective myoviruses B, C and D isolated from the majority of our lysates have also previously been shown to be associated with mitomycin C induced lysates which also contained typical myoviruses [14]. However the majority of lysates examined here only had these defective myoviruses. Although it is assumed that phage tail sheaths contract when infecting a host and ejecting DNA, our TEM analysis did not confirm this. Our data therefore support previous observations that C. difficile carries two morphologies of myoviruses [12].
The novel C. difficile myovirus E identified from one isolate is similar to 'killer particles' found in Bacillus species, Acetobacter species and Clostridium botulinum [26,27]. These particles are thought to package host DNA instead of bacteriophage DNA and are able to kill sensitive hosts but not replicate within them [26,28]. We have made several attempts to propagate this phage through plaque assays and spot tests but unfortunately, no host was found despite testing them on 63 different C. difficile hosts (including clinical and environmental isolates). However, myovirus E was able to clear lawn of one host (CD630) but scrapped zones of killing were not able to further propagate, and TEM observation on these lysates could not identify whole phages ( Figure S3). Therefore work is on-going to further characterise this phage and its possible impact on the pathogenicity of its host. Other temperate bacteriophage morphology identified here are the siphoviruses typified by F. Although, siphoviruses with dimensions similar to those isolated here have previously been reported on other ribotypes of C. difficile, this report is the first to show the association of this phage morphology to ribotype 027 [9,13,14,25]. Interestingly, two of the three siphoviruses we isolated were induced from isolates within a pulsovar subtype associated with very high severity of CDI (death within the first thirty days of infection) [6]. This suggests their possible role in the pathogenicity of 027 strains. Furthermore, a previous report has shown that infection of C. difficile ribotype 027 by phiCD38-2 (a temperate C. difficile siphovirus) resulted in the increase expression of toxin A and B [9]. Although bioinformatics analysis of phiCD38-2 has failed to identify toxins or virulence factors in the viral genome, the phage has been shown to interact with the C. difficile pathogenicity locus (PaLoc) thus having the potential to influence toxin production. Therefore, phage detection may represent a useful marker to identify strains that are associated with increased disease severity, and this early detection could aid clinical practice.
Dual phage carriage has previously been reported in CD630, CD8 and CD38 in which two morphologically and genetically distinct phages were induced from a single C. difficile strain [14,17]. However, CD630, CD8 and CD38 strains were of different ribotypes. Therefore, this study is the first to show two inducible and morphologically different phages with different genome sizes from ribotype 027 isolates. The correlation of phage carriage with MLVA types of ribotype 027 suggests that it is stable within this ribotype and does contribute to its diversity and severity of CDI [29].
The presence of phage tail-like particles, such as those represented by G in all our induced lysates, has previously been reported to be found in several induced lysates of C. difficile [14,25]. However, the lengths of the observed phage tail-like particles in this study were longer than those in previous reports (an average size of 160 nm as opposed to 130 nm). These phage tail-like particles have morphology similar to bacteriocin of Budvicia aquatica and Pragia fontium and phage tail-like particles of Vibrio spp [30,31]. Another phage-like particles typified by H in two mitomycin C induced samples, have a morphology that has not previously been observed to be associated with C. difficile but they do closely resemble a bacteriocin induced from Pseudomonas aeruginosa p28 [32,33,34]. Some of the bacteriocins have been shown to have bactericidal activities and provide competitive advantage to their hosts [34,35]. Although the phage tail-like particles G have been isolated from C. difficile isolates, their biological functions have not yet been determined. However, we found them present in all our induced isolates (in addition to other phage morphologies). This suggests their potential role in the survival of these isolates as earlier suggested [34,35]. Since their tails are non-contractile and their capsid DNA sequences show significant homology with myovirus sequences during in-silico PCR, our report concurs with the previous study that these particles may share some evolutionary characteristics with the myoviruses [36].
It is evident from the results of this study that the temperate phages isolated also have diverse genome sizes. The DNA size of ,15 kb (96LN2, 84LN, 91LN and 66LM) observed here are amongst the smaller sizes for phages isolated from Clostridium species [37]. The genome sizes of temperate bacteriophages from isolates 91LM, 96LN1 (myoviruses) and 82LN (defective myovirus) are in agreement with the genome sizes of temperate bacteriophages of C. difficile belonging to Myoviridae, having genomes varying in size from 29-160 kb [17,18,37]. The genome sizes of the siphoviruses isolated from 37L and 48L are in agreement with previous reports [9,20]. We tested ten restriction endonucleases BamH1, Nde1, BstB1, HindIII, EcoR1, Sty1, Sma1, Sau3A1, Mbo1 and Dpn1 on the DNA of these phages. However, none of these restriction enzymes could digest the DNA from these phages, or if they digest the DNA, it was only partial so they did not result in suitable banding patterns to allow genomic comparisons. This is likely to be attributed to DNA methylases which are found in the genomes of C. difficile strains and which provide immunity to the restriction enzymes [18,38].
The characterisation of phages by their morphology, genome size and inducibility is vital to understanding phage dynamics but these methods are labour intensive and dependent on induction antibiotic choice. Molecular markers to establish bacteriophage diversity and carriage would be of clear interest in terms of C. difficile host-phage dynamic characterisation. Such information would greatly enhance morphological studies.
Since prophages integrate into their host genomes, the use of these primers on genomic DNA will help to detect phages in isolates, thus direct strain choice before the time consuming phage induction procedures and TEM analysis. Phylogenetic analysis was done both at the nucleotide and the amino acid level and both trees gave the same supported tree topology within C. difficile phages. The amino acid sequences however allowed us to align the sequences from the C. difficile phages with phages which infect several other taxa and thus the analysis could be contextualised with respect to other phages. The ability of all the 027 isolates used in this study to form a strong grouping both at the nucleotide and amino acid level showed the much conserved nature of the genes of this ribotype [21]. This result is consistent with a previous report where DNA microarrays and Bayesian-based algorithms were used to phylogenetically analyse whole genomes of 75 C. difficile . Evolutionary relationship of Clostridium difficile based on the myovirus capsid gene. The evolutionary history was inferred using the Neighbor-Joining method. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches. The evolutionary distances were computed using the p-distance method based on their amino acid sequences. All positions containing gaps and missing data were eliminated. The analysis involved 29 amino acid sequences of 10 (84L, 16L, 96L, 82L, 52L, 68L, 91L, 90L, 80L and 73L) representative isolates of the ribotype 027 subclades and seven other ribotypes (ribotypes 014, 005, 002, 020, 015 and 001). Five other sequences including CD196, CDR20291 and QCD-66c26 (ribotype 027), CD630 (ribotype 012) and phiC2 were obtained from in-silico PCR. Other sequences were obtained from NCBI searches. All sequences with 75% similarities were assigned into a subclade. Evolutionary analyses were conducted in MEGA5. doi:10.1371/journal.pone.0037263.g003 strains. They observed that the nineteen 027 strains studied appeared to form a tight group which was distinct from the other 56 strains [39]. This is in agreement with previous work in which it was found that C. difficile 027 strains have very conserved genes, are descended from a single ancestor and acquire diverse external genetic element such as prophages, plasmids, antibiotic resistant genes and mutations via horizontal gene transfer during evolution [21]. These factors are likely to contribute to their pathogenicity [38]. The C. difficile phage group which is distinct from other members of the Clostridial family and other bacterial species suggests that the capsid is a good molecular marker to indicate phage presence and study phylogenetic relationships in bacterial species.
Clearly, C. difficile 027 isolates contain morphologically and genetically diverse sets of bacteriophages which are inducible with specific antibiotics. The carriage of different bacteriophages by C. difficile 027 isolates shows that there is significant variation within the mobile genetic elements associated with this ribotype. The correlation of phage carriage with MLVA and pulsovar types within C. difficile 027 also suggests that these bacteriophages contribute to its diversity, and in turn contribute to the success of particular strains in terms of disease severity. Future work will aim to determine how bacteriophages contribute to the spectrum of disease severity observed with different C. difficile ribotypes. Table S1 Table showing diverse phage morphologies isolated from mitomycin C and norfloxacin inductions of 91 C. difficile 027 isolates belonging to different subclades. Morphologies were identified using TEM. Bar ,70 nm based on the measurement of six phages in each sample. (DOC) Figure S1 Multiple alignment of four C. difficile phage sequences used in the design of the capsid primers. DNA sequences of PhiC2, CD630 and phiCD119 were obtained from NCBI searches and phi12 was a partial DNA sequences provided by Katherine Hargreaves, University of Leicester. The sequences were aligned using ClustalW. Regions for the forward and reverse primers were manually selected and indicated using arrows. (TIF) Figure S2 Graph showing different patterns of growth responses of C. difficile cultures during prophage inductions. Overnight broth cultures of C. difficile 027 isolates in BHI at OD 550 ,1.2 were induced with norfloxacin or mitomycin C at final concentration of 3 mg/ml for 24 h. Final OD 550 were taken at the end of 24 h. The effect of antibiotics on the growth of bacterial cultures was determined by measuring the OD 550 values before and after induction. The growth responses all fitted into one of four patterns (A-D) and this was compared to phage release. I. For pattern A, there was a drop in OD 550 with both mitomycin C and norfloxacin inductions. Although this pattern has previously been reported in C. difficile inductions, only 35 of the 91 isolates used in this study showed this profile. Furthermore, from these 35 isolates, 33 were found to harbour intact phages with the remaining two harbouring phage tail-like particles following induction by either antibiotic. II. For pattern B, the OD 550 dropped with mitomycin C induction but increased following norfloxacin induction. Twenty-eight of isolates showed this pattern. The TEM analysis showed that there was phage release from 16 of the isolates following mitomycin C and 11 with norfloxacin induction. One isolate was found to contain no phages. III. Pattern C was the opposite of pattern B with OD 550 increasing following mitomycin C induction but dropping following norfloxacin induction. Only five isolates showed this pattern. There was phage release from two isolates with mitomycin C and three with norfloxacin induction as confirmed by TEM analysis. IV. For pattern D, the OD 550 remained relatively constant following either norfloxacin or mitomycin C induction of the cultures. This pattern was observed in 23 isolates. Only phage tail-like particles were observed with these isolates. In