Investigation of Mannheimia haemolytica bacteriophages relative to host diversity



Tim A. McAllister, Agriculture and Agri-Food Canada Research Centre, 5403 1st Ave. South, Lethbridge, Alberta T1J 4B1, Canada. E-mail:



This study aimed to characterize the impact of lytic and temperate bacteriophages on the genetic and phenotypic diversity of Mannheimia haemolytica from feedlot cattle.

Methods and Results

Strictly lytic phages were not detected from bovine nasopharyngeal (n = 689) or water trough (n = 30) samples, but Myoviridae- or Siphoviridae-like phages were induced from 54 of 72 M. haemolytica strains by mitomycin C, occasionally from the same strain. Phages with similar restriction fragment length polymorphism profiles (RFLP ≥70% relatedness) shared common host serotypes 1 or 2 (< 0·0001). Likewise, phages with similar RFLP tended to occur in genetically related host bacteria (70–79% similarity). Host range assays showed that seven phages from host serotypes 1, 2 and 6 lysed representative strains of serotypes 1, 2 or 8. The genome of vB_MhM_1152AP from serotype 6 was found to be collinear with P2-like phage φMhaA1-PHL101.


Prophages are a significant component of the genome of M. haemolytica and contribute significantly to host diversity. Further characterization of the role of prophage in virulence and persistence of M. haemolytica in cattle could provide insight into approaches to control this potential respiratory pathogen.

Significance and Impact of the Study

This study demonstrated that prophages are widespread within the genome of M. haemolytica isolates and emphasized the challenge of isolating lytic phage as a therapeutic against this pathogen.


Mannheimia haemolytica, previously known as Pasteurella haemolytica, is a Gram-negative bacterium and a principal agent of bovine respiratory disease (BRD; Zecchinon et al. 2005; Griffin et al. 2010). Mortalities as a result of BRD and the treatment of morbidities that exhibit impaired growth and inferior carcass quality have a significant economic impact on the beef industry (Duff and Galyean 2007). Mannheimia haemolytica exists as a commensal bacterium in the upper respiratory tract of healthy cattle, but in some circumstances, pathogenic populations predominate (Rice et al. 2007). Of the 12 capsular serotypes, 2 is most often isolated from healthy cattle (Frank and Smith 1983), while 1 and 6 are more commonly detected in cattle with BRD (Zecchinon et al. 2005; Griffin et al. 2010). The shift from a commensal to pathogenic population is a multifactorial response to altering host conditions (Rice et al. 2007) and is likely influenced by the ecology of the microbial community, including the prevalence and nature of bacteriophages.

As agents of lateral gene transfer, prophages contribute significantly to the variability and evolution of bacterial genomes, constituting as much as 20% of the genome (Casjens 2003). Prophage sequences contribute to interstrain heterogeneity, environmental adaptation and gene regulation within bacterial hosts (Canchaya et al. 2003a). This evolutionary process can also introduce genes coding for virulence factors that confer pathogenicity. Indeed, horizontal DNA transfer has previously been implicated in the acquisition of lktA and ompA genes by M. haemolytica and in strain divergence (Davies et al. 2001, 2002). A previous study in our lab revealed considerable genomic diversity among M. haemolytica isolates endemic to Western Canadian feedlots (Klima et al. 2011), which may be explained in part by phage-mediated lateral gene transfer.

Phages were first isolated from M. haemolytica in the 1950s (Saxena and Hoerlein 1959). It was later proposed that all strains of M. haemolytica biotype A, serotype 1 (A1), harbour the temperate phage φPhaA1 (Richards et al. 1985). Froshauer et al. (1996) examined 14 strains of M. haemolytica serotype 1 isolated from cattle with shipping fever and found all strains carried ~40 kb prophages. Highlander et al. (2006) revealed that M. haemolytica strains A1, A5, A6, A9 and A12 contained sequences of the P2-like phage, φMhaA1-PHL101. Davies and Lee (2006) examined 15 bovine and 17 ovine M. haemolytica isolates and showed that they contained prophages with genomes ranging from 22 to 45 kb. These studies suggest that prophages are widespread in M. haemolytica, but no work has been undertaken to specifically compare the presence of inducible phages with host diversity.

Contrary to the reputation of phages in promoting virulence of hosts, they may also augment conventional antimicrobial therapy to control M. haemolytica populations in cattle. It has previously been shown that danofloxacin, a fluoroquinolone with broad-spectrum antibacterial activity, inhibits bacterial DNA gyrase, resulting in DNA damage during replication, inducing the prophage lytic cycle in M. haemolytica and causing cell lysis (Froshauer et al. 1996). Although BRD is primarily controlled through the use of antimicrobials, therapeutic administration of lytic phages could also be an attractive alternative treatment of M. haemolytica infections. Phage-mediated lysis is effective in treating multidrug-resistant bacterial infections and can be host-specific, leaving commensal bacterial flora unharmed (Kutateladze and Adamia 2010). Moreover, phages can replicate in vivo, reducing the need for repeated treatment with antimicrobials (Kutateladze and Adamia 2010).

The objective of this study was to determine the association of the presence of bacteriophages with genetic diversity, enrofloxacin sensitivity and serotype of endemic M. haemolytica from feedlot cattle. An extensive collection of bovine nasopharyngeal swabs and water trough samples were screened for strictly lytic bacteriophages, and prophages from M. haemolytica isolates were induced by exposure to mitomycin C.

Materials and methods

Bacterial strains

A total of 80 M. haemolytica isolates (Table 1), 10 reference strains from the American Type Culture Collection (ATCC, Manassas, VA, USA) and University of Guelph culture collection (UGCC, Guelph, ON, Canada) and 70 field isolates from bovine nasopharyngeal swabs collected at four feedlots in Southern Alberta, Canada, were used in this study. Identification of M. haemolytica isolates was confirmed by multiplex polymerase chain reaction (Alexander et al. 2008) and subsequently genotyped by PFGE (Klima et al. 2011). Isolates were stored at −80°C in brain heart infusion broth (BHIB; BD Canada, Inc., Mississauga, ON, Canada) containing 20% (v/v) glycerol. Starter cultures were prepared by inoculating 10 ml of BHIB with freshly cultured single colonies of each isolate and incubated overnight at 37°C in a shaking incubator.

Table 1. Properties of Mannheimia haemolytica strains used in the study
StrainSerotypeSource and feedlot ID (sampling event)Enrofloxacin susceptibilityaPhage morphologyb
  1. NA, not applicable; NT, not typeable.

  2. a

    Enrofloxacin zone diameters: 21–26 mm = Low; 35–45 mm = High.

  3. b

    Myoviridae/Siphoviridae: Both Myoviridae and Siphoviridae morphologies were detected.

Reference strains
BAA-4101ATCCNA Myoviridae/Siphoviridae
G88GuelphNA Myoviridae/Siphoviridae
Nasopharyngeal sample isolates
140A221 (Arrival)High Siphoviridae
143A221 (Arrival)Low Siphoviridae
251B221 (Arrival)High Siphoviridae
254A221 (Arrival)High Siphoviridae
309A221 (Arrival)Low Siphoviridae
407A221 (Arrival)Low Siphoviridae
410A25 (Arrival)Low Siphoviridae
413A25 (Arrival)High Myoviridae
501A15 (Arrival)High Siphoviridae
535A121 (Arrival)Low Siphoviridae
587A221 (Exit)High Myoviridae
611A15 (Exit)Low Siphoviridae
613A221 (Exit)High Myoviridae/Siphoviridae
622A221 (Exit)High Myoviridae/Siphoviridae
755A25 (Arrival)High Myoviridae
842A221 (Arrival)High Myoviridae
946A25 (Exit)High Siphoviridae
964A621 (Arrival)High Myoviridae
1040A221 (Exit)High Myoviridae
1059A25 (Exit)Low Myoviridae
1098A221 (Exit)High Siphoviridae
1127A221 (Exit)Low Myoviridae
1152A621 (Exit)Low Myoviridae
1312A25 (Exit)Low Myoviridae
1492A25 (Exit)High Myoviridae
1498A25 (Exit)High Myoviridae
1719A221 (Exit)Low Myoviridae
1745A25 (Exit)Low Myoviridae
1795A221 (Exit)Low Myoviridae/Siphoviridae
1829A25 (Exit)Low Siphoviridae
2024A221 (Exit)Low Myoviridae/Siphoviridae
2237A15 (Arrival)Low Myoviridae
2256A15 (Arrival)Low Siphoviridae
2657A25 (Arrival)High Siphoviridae
2673B25 (Arrival)High Siphoviridae
2677A25 (Arrival)High Myoviridae
2679A25 (Arrival)High Myoviridae
2707A25 (Arrival)High Myoviridae
2713ANT5 (Arrival)Low Siphoviridae
2791A121 (Exit)Low Myoviridae
3083A65 (Exit)Low Myoviridae
3089A65 (Exit)Low Siphoviridae
3721A221 (Exit)High Myoviridae
3724A221 (Exit)High Myoviridae
3911A25 (Exit)High Myoviridae
3927A65 (Exit)Low Myoviridae
4386A221 (Exit)High Siphoviridae
4387A221 (Exit)High Myoviridae
4645A221 (Exit)High Siphoviridae
4653A221 (Exit)High Siphoviridae
4664A221 (Exit)High Myoviridae
5039A25 (Exit)High Myoviridae

Enrofloxacin susceptibility

Field isolates were selected from a collection of 409 M. haemolytica which were screened for enrofloxacin susceptibility by the disc diffusion assay (data not shown), in accordance with Clinical and Laboratory Standards Institute document M31-A3 (CLSI 2008). Of the isolates considered susceptible (≥21 mm), 8% was selected from both the high (35–45 mm, = 33) and low (21–26 mm, = 33) zone-diameter limits, to represent isolates more and less sensitive to enrofloxacin, respectively. These groups of isolates were compared to determine whether there was a correlation between enrofloxacin sensitivity of M. haemolytica isolates and the presence of inducible phages.

Lytic bacteriophages

A total of 689 bovine nasopharyngeal swabs were obtained from four feedlots in Southern Alberta, Canada, with equal representation from animals recently arrived or >60 days on feed at each feedlot. Swabs were submerged and vortexed in BHIB containing 20% (v/v) glycerol. Either individual or pooled nasopharyngeal swab suspensions were filtered through 0·8-/0·2-μm-pore-size low-protein-binding syringe filters (Pall Canada Ltd., Mississauga, ON, Canada). In addition, 30 pooled water samples (~500 ml) were obtained from multiple water troughs at a fifth feedlot in Southern Alberta, Canada, and filtered through 0·2-μm-pore-size bottle-top filter (Nalgene; Fisher Scientific, Ltd, Nepean, ON, Canada). Water filtrates (15 ml) were concentrated by Centriprep YM-30 centrifugal filter units (Millipore, Bedford, MA, USA) at 1500 g for 15 min at room temperature, to a final volume of 3–5 ml. Swab filtrates and water concentrates were stored at 4°C until they were further processed.

Cocktails of M. haemolytica serotypes A1, A2 and A6 including three reference strains (BAA410, 33396 and 33370) or six field strains (413A, 501A, 535A, 964A, 1098A and 1498A) were selected as genetically diverse hosts for isolation of lytic phages. Cocktails of early log-phase M. haemolytica liquid cultures (~5 × 108 CFU ml−1) were combined with 1 ml of the swab filtrate or water filtrate supplemented with 10 mmol l−1 MgSO4. Different ratios (1 : 1, 5 : 1, 10 : 1) of M. haemolytica cocktail to sample filtrate were tested. The mixture was incubated for 18–20 h at 37°C in a shaking incubator and filtered through 0·8-/0·2-μm-pore-size filter. The double-agar overlay method was used to detect lytic phages. Indicator strains were identical to the M. haemolytica strains used during enrichment. Each enriched filtrate (500 μl) was mixed with either individual or cocktails of M. haemolytica liquid cultures (500 μl) at early log phase (~5 × 108 CFU ml−1). The mixture was incubated for 15–20 min at room temperature to allow phage adsorption. Molten top agar was added to the mixture and poured onto agar plates (bottom agar). Agar plates were incubated at 37°C for 18–20 h and examined for plaque formation. Various types and concentrations of agar and divalent cations were used in the double-agar overlay procedure. This included modified nutrient agar, BHI agar and modified BHI agar as the bottom agar. UltraPure™ agarose (Invitrogen, Burlington, ON, Canada), Difco™ Agar (BD Canada, Inc.) and BHI agar were prepared with agar concentrations ranging from 0·4 to 0·7% to serve as top agar. Either top agar or bottom agar was supplemented with 5 mmol l−1 CaCl2 or 10 mmol l−1 MgSO4.

Temperate bacteriophages

Bacteriophages were induced from 72 M. haemolytica including two reference strains and 70 field isolates (Tables 1 and 2) using a modification of the procedure by Davies and Lee (2006). Briefly, 30 ml of BHIB was inoculated with an appropriate volume (0·3–1·0 ml) of overnight liquid starter culture of each isolate and incubated at 37°C in a shaking incubator. Each culture was incubated for 3–6 h to obtain isolates in early log phase (~5 × 108 CFU ml−1). Phage induction was initiated by adding freshly prepared mitomycin C (Sigma-Aldrich Canada Ltd., Oakville, ON, Canada) to each culture to a final concentration of 0·2 μg ml−1. Optical density at 600 nm of the induced culture was monitored for 7–8 h. Induced M. haemolytica cultures were centrifuged at 3000 g for 20 min at 4°C and filtered through 0·22-μm Steriflip-GP filter units (Millipore, Billerica, MA, USA) to remove bacterial debris. Phage filtrates were stored at 4°C until they were characterized.

Table 2. Host range of eight inducible phage preparations against Mannheimia haemolytica strains
Mannheimia haemolytica indicator strains by serotype (n per serotype)Incidence of plaque formation for phages (host serotype)
535AP (Sero 1)2256AP (Sero1)587AP (Sero 2)622AP (Sero 2)1127AP (Sero2)1152AP (Sero 6)3927AP (Sero 6)G8P (Sero 8)
1 (n = 12)11011900
2 (n = 56)814507271110
6 (n = 6)00000000
8 (n = 1)11111110
5, 7, 9, 12, 14 (n = 1)00000000

Transmission electron microscopy

Morphological examination of phage using transmission electron microscopy (TEM) was performed as described by Carlson (2005). Briefly, 30 ml of phage filtrate was centrifuged in a fixed-angle rotor at 25 000 g for 90 min to sediment phage particles. Supernatants were carefully removed and phage pellets were resuspended in sterile 0·1 mol l−1 ammonium acetate buffer (pH 7·0). This washing process was repeated twice to remove proteins, sugars and salt. Final phage pellets were resuspended in 250 μl of 0·1 mol l−1 ammonium acetate buffer. One drop of the phage suspension was absorbed onto a carbon-coated 200-mesh copper grid (Canemco, Inc., Canton de Gore, QC, Canada). Phage samples were negatively stained with 5% uranyl acetate solution (pH 4·0) for approximately 1 min. Excess stain was removed with filter paper and grids were examined by TEM (Hitachi H7100: Hitachi High-Technologies Canada, Inc., Toronto, ON, Canada) at 75·0 kV at a magnification range of 20 000–100 000×.

Host range assay

Eight M. haemolytica temperate phages (Table 2) were selected to represent a cross-section of the phages detected in this study for host range analysis. Phages were chosen to represent different virion morphologies, restriction fragment length polymorphism (RFLP) profiles and host attributes, including: PFGE-banding patterns, serotype and origin of the sample. Lawns of M. haemolytica indicator strains (consisting of the 10 reference strains and 70 field strains, Table 1) were prepared by mixing 1 ml of overnight liquid culture with 2 ml of molten agar (0·7% agarose in BHIB) with 2 mmol l−1 CaCl2 and poured onto BHI agar plates. After the top agar solidified, 10 μl of each phage filtrate was spotted and absorbed into prepared agar. Plates were incubated upright overnight at 37°C and examined for clearing zones. To confirm clearing zones were not due to bacteriocins, 10 μl of M. haemolytica cultures of the identical strain used for phage induction was spotted onto lawns of M. haemolytica indicator strains and examined for growth inhibition.

Genome sizing estimation

Bacterial nucleotides were removed from the eight filtered phage lysates using DNase 1 (Sigma-Aldrich, Oakville, ON, Canada) and RNase A (Sigma-Aldrich), and the phage lysates were concentrated by use of polyethylene glycol (PEG) 8000 (Sambrook and Russell 2001). Concentrated phage lysates were subjected to PFGE for determination of genome size (Lingohr et al. 2009). Phage DNA embedded in 1% SeaKem Gold (SKG) agarose was subjected to electrophoresis in 0·5× Tris–borate–EDTA buffer at 14°C for 18 h using a Chef DR-II Mapper electrophoresis system (Bio-Rad, Mississauga, ON, Canada), with pulse times of 2·2–54·2 s, 6 V cm−1. The markers were prepared using Salmonella branderup H9812 digested by XbaI. Banding patterns were viewed with UV illumination and photographed using the Speedlight Platinum Gel Documentation System (Bio-Rad).

Restriction fragment length polymorphism

Genomic phage DNA was extracted from concentrated phage suspensions using a Phage DNA Isolation Kit (Norgen Biotek Corp., Thorold, ON, Canada) according to the manufacturer's instructions, with proteinase K (Qiagen, Toronto, ON, Canada). Extracted DNA was quantified fluorometrically using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen) with the NanoDrop 3300 fluorospectrometer (Fisher Scientific Ltd.). Isolated phage genomic DNA was digested using the restriction enzymes HindIII and ClaI (New England Biolabs Inc., Pickering, ON, Canada) as previously described (Davies and Lee 2006). Digested phage DNA fragments were separated by agarose gel (0·7% w/v) electrophoresis and stained with ethidium bromide. As necessary, gels were soaked in 1 mmol l−1 MgSO4 solution to remove background fluorescence.

Genome sequencing and annotation

Genomic DNA was prepared from a single phage preparation, vB_MhM_1152AP (1152AP), selected on the basis of host attributes (pathogenic serotype), host infectivity assay (broadly lytic against indicator strains) and prevalence. Phage DNA was extracted and DNA concentration was determined as described previously. Subsequent DNA quality control assurance and amplification was conducted by Eurofins MWG Operon (Huntsville, AL, USA) prior to sequencing by GS FLX Titanium series chemistry (Roche 454). Whole genome sequencing yielded coverage of 100×. Sequencing data were assembled by Celera Assembler (version 5.3) and Staden gap4 and critical gaps were identified and closed by conventional Sanger sequencing. Initial genome annotation was completed using myRAST (Aziz et al. 2008). SeqBuilder application (DNASTAR, Inc., Madison, WI, USA) was used to visually scan the sequence for potential genes. All translated proteins were scanned for homologues using BLASTP and PSI-BLAST (Altschul et al. 1997). Ribosomal binding sites (RBS) were identified using RBS finder online and rho-independent terminators were identified using MFold (Zuker 2003) and TransTerm (Ermolaeva et al. 2000). Promoters were identified by neural network promoter prediction (Reese 2001) with visual inspection. Transfer RNA (tRNA) genes were screened using Aragorn (Laslett and Canback 2004) and tRNAScan (Lowe and Eddy 1997). Sequence manipulation and graphical representation of phage genome were made using the Geneious v5.4 program (Biomatters Ltd., Auckland, New Zealand). The GenBank accession number for the 1152AP consensus sequence is JN255163.

Statistical analysis

Correlation with enrofloxacin susceptibility was determined by a chi-square value generated by the FREQ of SAS program (version 9.1; SAS Institute Inc., Cary, NC, USA). Differences in the number of phages induced from high- vs low-enrofloxacin-susceptibility groups (Table 1) were considered statistically significant at P ≤ 0·1. BioNumerics software 6.5 (Applied Maths, Inc., Austin, TX, USA) was used to analyse RFLP and PFGE profiles. Dendrograms of phages' RFLP profiles were generated with UPGMA clustering of Dice coefficient values with 1·0% optimization and 1·5% position tolerance settings. Within- and between-group similarities (genetic relatedness) of host PFGE and phage RFLP profiles were tested using the Dimensioning Techniques package of BioNumerics. For these analyses, binary character profiles using the band-matching option were created after clustering using Dice coefficients. Group similarity was then evaluated using bootstrapping analyses (n = 1000), with significant differences reported at P ≤ 0·1.


Bacteriophages isolation and characterization

Despite considerable effort, strictly lytic phages infecting M. haemolytica were undetectable in either bovine nasopharyngeal swabs or water samples.

Preliminary studies on phage induction were performed with reference strain BAA-410. The published genome of M. haemolytica BAA-410 has been shown to contain a complete, inducible prophage (Highlander et al. 2006), and the induction of temperate phages from BAA-410 was confirmed by TEM. This strain was subsequently used as a control in all phage induction studies. Overall, 74% of M. haemolytica strains (52/70) were inducible (Table 1). Each phage was designated as vB (bacterial virus), followed by Mh (M. haemolytica), M or S (Myoviridae or Siphoviridae), then the specific isolate designation of the bacterial host with the suffix P (phage). For example, a phage induced from isolate 1152A is designated vB_MhM_1152AP, with 1152AP as the short form. Structurally, all phages identified in this study were tailed and identified as members of either Myoviridae or Siphoviridae (Table 1, Fig. 1), in the order of Caudovirales. A total of 27 M. haemolytica isolates produced phages displaying Myoviridae morphology, an icosahedral head and a contractile tail consisting of a central tube with a sheath located below the capsid (Fig. 1a). A further 21 M. haemolytica isolates produced Siphoviridae-like phages that exhibited a long, noncontractile tail with no sheath (Fig. 1b). Both morphologies were simultaneously identified in preparations of two reference strains (BAA-410 and G8) and four field isolates (613A, 622A, 1795A and 2024A).

Figure 1.

Representative transmission electron microscopy micrographs of virions negatively stained with uranyl acetate. 755AP, Myoviridae-like phage (a); 1098AP, Siphoviridae-like phage (b).

Of the eight representative phages selected, seven induced from M. haemolytica serotypes 1, 2 or 6 were capable of lysing indicator serotypes 1, 2 and 8 (Table 2). Phage 587AP displayed the broadest host range against M. haemolytica serotype 2, lysing 50 strains (89%), but it did not lyse any serotype 1 or 6 strains. Phage 1152AP showed the broadest lytic spectrum against M. haemolytica serotype 1, lysing nine strains (82%). Only phage G8P was unable to form plaques against any strain tested, whereas phage 587AP was the only one able to infect its lysogenic strain. No phage preparation was able to generate plaques against any of the six indicator strains from M. haemolytica serotype 6.

Genome size of the eight phages preparation ranged from 35 to 73 kb as estimated by PFGE (data not shown). The phage 1152AP genome consisted of linear double-stranded DNA of approximately 35 kb. The sequence was assembled into 34 719 bp (41·6% G+C) in length. A total of 54 open reading frames (ORFs) were identified. A summary of identified genes with predicted protein function is provided in Table S1. Furthermore, 11 rho-dependent terminators and 14 promoters recognized by host RNA polymerase were identified, but no tRNA was discovered. Pairwise genomic analysis revealed 1152AP was very collinear with 99·1% nucleotide similarity to P2-like phage φMhaA1-PHL101 (Fig. 2). Computational analysis of CoreGenes showed that this phage shares 48 (98%) homologues with phage φMhaA1-PHL101. BLASTN alignment showed that 1152AP sequence shared various degree of similarity with whole genome sequences of M. haemolytica serotype A1 PHL213 (GenBank accession #: AASA00000000), M. haemolytica serotype A2 str. OVINE (GenBank accession #: ACZX00000000) and M. haemolytica serotype A2 str. BOVINE (GenBank accession #: ACZY00000000). Contigs 38, 39, 129 and 159 of the M. haemolytica PHL213 genome contained sequences with 99–100% identity to different regions of the 1152AP consensus sequence and together they nearly encompassed the 1152AP genome. Contigs 00021, 00030, 00073 and 00134 of M. haemolytica serotype A2 str. OVINE genome contained regions that were 93–99% identical to regions of 1152AP, but several gaps were still evident. Regions of contig 00033 of M. haemolytica serotype A2 str. BOVINE genome exhibited 97% identical sequences and covered 75% of 1152AP sequence.

Figure 2.

Whole genome comparisons of phage 1152AP (a) and φMhaA1-PHL101 (b) using a progressive MAUVE alignment. The degree of sequence similarity is indicated by the intensity of the shaded region. The contiguous black boxes under the shaded region represent the position of the genes.

Comparison of phage RFLP profiles and enrofloxacin susceptibility with Mannheimia haemolytica serotype and genetic relatedness

The presence of temperate phages in M. haemolytica isolates were used to examine the correlation of enrofloxacin susceptibility levels with the induction of phages. High- and low-enrofloxacin-susceptible isolates (Table 1) were designated as either phage inducible or noninducible, but there was no evidence of a correlation (P = 0·4) between enrofloxacin sensitivity and the presence of inducible phages.

Of 54 induced phages, 21 were able to yield distinguishable RFLP profiles. Considerable variability was observed for fragment banding patterns between individual profiles. A dendrogram was generated with nine distinct clusters based on relatedness of ≥70% (Fig. 3). The clusters largely coincided with the observed phage morphologies, with some exceptions (143AP, 2237AP, 2256AP). Host characteristics including feedlot origin, the host animal's duration at the feedlot, serotype and enrofloxacin susceptibility were considered in parallel with this analysis to detect correlations between phage RFLP type and host traits. Generally, there were no apparent trends, except a degree of serotype specificity. Notably, all phages induced from M. haemolytica serotype 1 were clustered in a single RFLP type 8 and genetically related (81% similarity, < 0·0001, Table 3 and Fig. 2) as compared to those from serotypes 2 and 6. In addition, all phages induced from serotype 2 had highly similar RFLP profiles (< 0·0001, Table 3) as compared to those from serotypes 1 and 6, even though a variety of RFLP types were assigned within these phages (Fig. 3). These four groups of inducible phages (i.e. 587AP and 1127AP; 1040AP and 3911AP; 1492AP, 1745AP and 1059AP; 2237AP and 2256AP) were determined to have phage RFLP profiles showing 90–100% similarity and were associated with a common host serotype within each group (Fig. 3). In contrast, there were few occurrences of multiple host serotypes within any phage RFLP cluster (at ≥70% relatedness), with RFLP type 1 being the exception. Corresponding host PFGE profiles (Fig. S1) were then compared to phage RFLP profiles to identify common relationships between host genotypes and phage genotypes (Table 4). Host strains inducing phages of RFLP type 8 exhibited higher genetic similarity (77%, P = 0·006) than those inducing other RFLP-type phages. With the exception of RFLP type 1, phages tended to be more closely related if they originated from the same serotype (genetic similarities 70–79%), although the genetic relatedness of phage was not statistically different among groups formed on the basis of a common host serotype.

Table 3. Similarity of restriction fragment length polymorphism (RFLP) profiles of induced phages, grouped by serotype of Mannheimia haemolytica host strains
Serotype GroupNo. of isolatesSimilarity of phage RFLP profiles (%)aP valueb
Within groupBetween groups
  1. a

    RFLP profiles were produced from HindIII- and ClaI-digested phage DNA.

  2. b

    Significance of difference in relatedness of the induced phages within and between host serotypes.

Table 4. Restriction fragment length polymorphism (RFLP) types of phages induced from Mannheimia haemolytica strains and similarity of PFGE profiles among their host
RFLP typeNo. of isolatesSimilarity of PFGE profiles (%)aP valueb
Within RFLPBetween RFLP
  1. a

    PFGE profiles were produced from XbaI-digested host genomic DNA.

  2. b

    Significance of difference in relatedness of M. haemolytica strains within and between RFLP types.

Figure 3.

Dendrogram of restriction fragment length polymorphism (RFLP) profiles of 21 Mannheimia haemolytica phages. Isolates origin (feedlot information). Phages morphology was based on transmission electron microscopy observations. Serotype is described for the phage host, and phages RFLP group is arbitrarily assigned 1–9, from top-bottom, based on clusters ≥70% relatedness.


A comprehensive screening process was undertaken in an attempt to isolate strictly lytic bacteriophages against M. haemolytica. Nasopharyngeal swabs were the main focus in this study, because M. haemolytica typically exists as a commensal in the upper respiratory tract of healthy ruminants (Highlander et al. 2006). An enrichment step was included to facilitate the isolation of virulent, broad-spectrum phages that may be present at low levels in samples (Carlson 2005; Gill and Hyman 2010). Cocktails containing pathogenic and nonpathogenic serotypes as well as local field strains and reference strains of M. haemolytica were prepared in an effort to maximize the likelihood of isolating lytic phage. Pooling of multiple samples also presumably increased the potential of isolating lytic phages (Gill and Hyman 2010).

In spite of rigorous screening and extensive adjustments to enrichment protocols, growth media and phage detection assays, no lytic phages against M. haemolytica were detected. The reason for our inability to isolate lytic phages against M. haemolytica is unclear. Our laboratory has successfully used similar approaches to isolate numerous lytic phages with activity against Escherichia coli O157:H7 from cattle and feedlot environments (Niu et al. 2009). Several potential difficulties may have led to our inability to isolate M. haemolytica-infecting lytic phage in this study. The number of M. haemolytica-specific lytic phage may be rare in the environment sampled, and transportation and processing of raw samples may have further reduced phage infectivity. Naturally, bacteria utilize various innate phage-resistance mechanisms including DNA modification, abortive infection and phage adsorption resistance (Sturino and Klaenhammer 2006; Hyman and Abedon 2010). Various studies suggest clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated sequence (CAS) proteins are linked to a novel mechanism of acquired resistance against phages (Sorek et al. 2008). The CRISPR system is widespread in the genomes of many bacteria and may provide protection against phages carrying similar sequences to existing prophages (Sorek et al. 2008). A recent study identified CRISPR/CAS regions in the genomes of three sequenced strains of M. haemolytica (Gioia et al. 2006; Lawrence et al. 2010). This suggests that the CRISPR systems may be common in M. haemolytica and provide this bacterium with a degree of immunity to lytic phages and lysogenization by prophages.

Traditionally, TEM has been a valuable tool for comparative virology and phage classification (Ackermann 2009). The temperate phage morphologies observed here were diverse within the tailed phage group, but consistent with members of Myoviridae and Siphoviridae, previously described in M. haemolytica (Davies and Lee 2006) and the respiratory pathogen, Pasteurella multocida (Ackermann and Karaivanov 1984). Previous studies have shown that phage particles were widely distributed in M. haemolytica with 24 of 32 M. haemolytica isolates examined containing inducible phages (Davies and Lee 2006). The prevalence of phages was similar in this study, where 52 of 70 M. haemolytica field isolates contained inducible phages.

Uniquely, in this study, Myoviridae- and Siphoviridae-like phages were simultaneously induced from single M. haemolytica isolates, indicating multiple prophages may be harboured within a single host. This is a first report that phages belonging to multiple families can be induced from a single M. haemolytica host, but multiple prophages have been identified in the genomes of E. coli O157:H7, Streptococcus pyogenes and Bacillus subtilis (Canchaya et al. 2003b; Casjens 2003). The genome of M. haemolytica strain BAA-410 contains at least two Mu-like and one P2-like-prophages as well as λ-like prophage elements (Gioia et al. 2006). Typically, P2-like and Mu-like phages are myoviruses, while λ-like phages are siphoviruses (Guttman et al. 2005). The combination of myovirus and siphovirus sequences in BAA-410 is in agreement with our observation that multiple phage morphologies can be induced from a single host.

Phage RFLP analysis can provide a rapid assessment of genetic diversity of induced phages (Carlson 2005). In this study, substantial genetic variation was evident from the patterns generated with double enzyme digestions. Common host serotypes among genetically related lysogenic phages may indicate that these agents target common lipopolysacchride (LPS) receptors on the bacterial cell surface. Interestingly, genetically related strains of M. haemolytica did tend to produce phages with similar RFLP profiles, a factor that likely reflects the prominence and persistence of the prophages within the host genome. Others have shown that prophage or prophage-like sequences are important contributors to diversity in M. haemolytica (Lawrence et al. 2010), possibly promoting the exchange of genetic information (Brüssow et al. 2004).

Host range analysis was used to test the efficiency of a phage's ability to infect and develop in a particular host. This approach may be used as a discriminatory tool that compliments RFLP analysis, as host range analysis of the eight selected phages showed distinct lysis patterns against a collection of potential host strains. Overall, phages exhibited a limited host range that was serotype specific as plaques were formed only with serotypes 1, 2 and 8. Davies and Lee (2006) also found that all prophages induced from bovine serotype A1 and A6 were incapable of lysing serotype A6 indicator strains. Presumably, compared to M. haemolytica serotype A1 and A2, serotype A6 might exhibit divergent receptor structure of LPS and outer membrane for phage recognition. Prophages can also confer common immunity by way of a common repressor protein (Guttman et al. 2005). As the majority of our indicator strains are known to harbour prophages, their presence might have precluded infection of the host by genetically related phages. None of the phages induced were able to form the clear plaques that would be indicative of a strictly lytic phage. This was probably due to lysogenization of some bacteria within the plaque, analogous to typical turbid plaques formed by λ phage and temperate phage in general (Guttman et al. 2005; Gill and Hyman 2010).

A variety of antimicrobials are used to treat cattle infected with BRD (Fulton 2009). Previous studies demonstrated that the presence of prophages increased the sensitivity of M. haemolytica to danofloxacin, and exposure to this antimicrobial induced prophage production (Froshauer et al. 1996). We selected 66 M. haemolytica isolates with high and low levels of enrofloxacin sensitivity based on disc diffusion assays and found no correlation between presence of inducible phages and the sensitivity of M. haemolytica to this antibiotic. Both danofloxacin and enrofloxacin inhibit bacterial DNA gyrase (Froshauer et al. 1996), resulting in DNA damage during replication. Many bacteria possess an SOS-response system which is an inducible DNA repair system that allows bacteria to overcome DNA damage (Foster 2007). If the SOS-response system is activated, RecA protein is produced which reduces the repression of the phage lytic pathway and promotes induction (Sauer et al. 1982; Kuzminov 1999).

Prophages are found in a wide range of bacterial species and can contribute important biological properties to their host. Genome mosaicism is characterized by patches of high sequence similarity separated by nonhomologous regions (Casjens 2003; Skurnik et al. 2007), a characteristic that was evident in our phage sequence which closely aligned with phage φMhaA1-PHL101 and with contigs in three M. haemolytica sequenced genomes (Gioia et al. 2006; Lawrence et al. 2010). Although 1152AP showed regions of very high sequence identity to these M. haemolytica genomes, alignments were not always found on the same contig. This likely reflects the occurrence of large-scale inversions and rearrangements within the M. haemolytica genome (Lawrence et al. 2010). Overall, the prevalence of sequences with similar identity to the 1152AP genome suggests widespread horizontal transfer of P2-like phages among M. haemolytica strains.

Mannheimia haemolytica isolated from beef cattle harbour lysogenic phages with genetic and phenotypic diversity. Based on RFLP profiles and host plaque assays, we suspect that the predominant M. haemolytica serotypes (1, 2 and 6) may each be affiliated with a unique prophage. Prophages clearly play a role in the genotypic diversity among M. haemolytica strains. Continued efforts to sequence the genomes of M. haemolytica and characterization of their associated prophages will provide greater insight into the role of these mobile elements in the evolution of this important veterinary pathogen.


This work was supported by Canada Matching Investment Initiative and Alberta Agricultural Research Institute. We would like to thank Byron Lee, Lorna Selinger and Meng Samuel Qi, for their technical expertise, members of Feedlot Health Management Services and Fred VanHerk for sample collection.