Host range and lytic capability of four bacteriophages against bovine and clinical human isolates of Shiga toxin-producing Escherichia coli O157:H7

Authors


T.A. McAllister, Agriculture and Agri-Food Canada, Lethbridge Research Centre, Lethbridge, Alberta, Canada T1J 4B1.
E-mail: mcallistert@agr.gc.ca

Abstract

Aims:  To evaluate host range and lytic capability of four bacteriophages (rV5, wV7, wV8 and wV11) against Escherichia coli O157:H7 (STEC O157:H7) from cattle and humans.

Methods and Results:  Four hundred and twenty-two STEC O157:H7 isolates (297 bovine; 125 human) were obtained in Alberta, Canada. The four phages were serially diluted and incubated for 5 h with overnight cultures of STEC O157:H7 to estimate their multiplicity of infection (MOI). All bovine STEC O157:H7 were subjected to pulsed-field gel electrophoresis (PFGE) and phage typing (PT). Phage wV7 lysed all human and bovine isolates irrespective of PFGE genotype or PT phenotype and exhibited the lowest MOI (0·004–0·006, P < 0·0001) of all phages. Phages rV5 and wV11 exhibited a lower MOI (0·002–0·04, P < 0·0001) than did phage wV8 (25–29) and they had a narrower host range than wV7 or wV8. Phages rV5, wV11 and wV8 lysed 342 (81·0%), 321 (76·1%) and 407 (96·4%), respectively, of the 422 isolates. Susceptibility of bovine STEC O157:H7 to rV5, w11 and wV8 was influenced by PFGE genotype and/or PT phenotype.

Conclusions:  Phages exhibited activity against the majority of bovine and human STEC O157:H7 isolates. PFGE genotype and/or PT phenotype of the host-target influenced their vulnerability to phage attack. Susceptibility of bovine STEC O157:H7 to phage may also differ among farms. Both lytic capability and host range should be considered in the selection of therapeutic phage for on-farm control of STEC O157:H7.

Significance and Impact of the Study:  The present work indicates that a four-phage cocktail should be equally effective at mitigating STEC O157:H7 isolates both of bovine and of human origin. Given that some STEC O157:H7 exhibited resistance to some but not all phages, a phage cocktail is the logical approach to efficacious on-farm therapy.

Introduction

Shiga toxin-producing Escherichia coli O157:H7 (STEC O157:H7) is a significant public health concern, causing severe, sometimes life-threatening, human illness (Tarr et al. 2005). Cattle are recognized as primary reservoirs of STEC O157:H7 (Elder et al. 2000) and contaminated foods of bovine origin (Rangel et al. 2005) or vegetable products contaminated with bovine faeces (Serna and Boedeker 2008) are sources of human infection. Occurrence and diversity of STEC O157:H7 in beef and dairy cattle has been investigated in a number of countries (Stanford et al. 2005; Nastasijevic et al. 2007; Oporto et al. 2008), but much of the ecology of this organism remains unclear.

Interest in lytic (virulent) bacteriophages (phages) as antibacterial agents is increasing, owing to their ability to self replicate, their high specificity against target bacteria and their ubiquity within the environment (Sulakvelidze et al. 2001; Merril et al. 2006). Phages are being reassessed for their ability to control infections in humans, livestock and plants and to decontaminate processed foods and agricultural products (Brüssow 2005; Greer 2005; Jones et al. 2007). Numerous lytic phages have recently been shown to reduce populations of E. coli O157:H7 in vivo and in vitro (Abuladze et al. 2008; Callaway et al. 2008; Niu et al. 2008; Viscardi et al. 2008).

Phages that exhibit a broad host range and strong lytic activity tend to be more effective therapeutically than those that exhibit a narrow host range and limited bacteriolytic activity (Skurnik et al. 2007). In earlier studies, a mixture of six phages eliminated E. coli O157:H7 from experimentally inoculated pre-ruminant calves (Waddell et al. 2000). Subsequently, a mixture of four of those six phages (rV5, wV7, wV8 and wV11) reduced populations of five nalidixic acid-resistant strains of STEC O157:H7 in experimentally inoculated feeder cattle (Niu et al. 2008). The four phages were selected because their combined susceptible host range of STEC O157:H7 included the 12 most common O157:H7 phage types (PT) in North American isolates (95%) obtained between 2000 and 2002 (R.P. Johnson, unpublished data).

Evaluating the susceptibility of naturally occurring STEC O157:H7 isolates to phages is an important step in defining effective phage therapies. Characterization of genotypic and phenotypic properties that influence the susceptibility of STEC O157:H7 to phage may also aid in predicting the effectiveness of phage therapy. Thus, this study was conducted to evaluate the host range and lytic capabilities of phages rV5, wV7, wV8 and wV11 against a collection of STEC O157:H7 isolates from cattle and humans. These characteristics were determined using a microplate phage virulence assay that estimated the lowest ratio of phage to bacteria (i.e. the multiplicity of infection (MOI)) that caused complete lysis of cultures of individual isolates. As well, the bovine isolates were phenotyped by PT and genotyped using pulsed-field gel electrophoresis (PFGE) to determine if relationships existed between PT phenotype, PFGE genotype and susceptibility to phage.

Materials and methods

Collection and identification of STEC O157:H7 isolates

Over a 12-month period from October 2002 to September 2003, pooled faecal pats and samples of oral microflora were collected at approximately monthly intervals from cattle in four commercial feedlots (F1–F4) and four commercial dairies (D5–D8) in southern and central Alberta, Canada (Stanford et al. 2005). All samples were tested for O157 antigen by enrichment culture and immunomagnetic separation (IMS) using Dynabeads anti-E. coli O157 (Dynal, Lake Success, NY, USA) according to the manufacturer’s instructions. The presence of O157 antigen in suspect E. coli O157 isolates was determined using the E. coli O157 latex kit (Oxoid, Nepean, ON, Canada). A total of 297 isolates were confirmed as STEC O157:H7 based on the presence of vt, eaeA and fliC (H) genes in multiplex PCR assays (Gannon et al. 1997).

One hundred and twenty-five isolates collected from multiple separate human clinical cases in southern Alberta between 2002 and 2004 were identified as STEC O157:H7 in the Provincial Laboratory for Public Health in Calgary, Alberta. Briefly, presence of O157-antigen and H7-antigen were confirmed by slide agglutination tests with O157 antiserum (Difco) and the tube agglutination method using H7 antiserum (Difco). The 4-methylumbelliferyl-β-d-glucuronide (MUG) test was performed to confirm the presence of β-glucuronidase enzyme.

Phage typing of STEC O157:H7 bovine isolates

Bovine isolates were phage-typed (PT) at the E. coli Reference Laboratory of the Laboratory for Foodborne Zoonoses, Guelph, Ontario, using previously described procedures (Ahmed et al. 1987; Khakhria et al. 1990) and 16 phages (numbered 1–16) that differentiated 88 PT.

Genetic characterization of STEC O157:H7 bovine isolates

PFGE was conducted on XbaI-digested genomic DNA from bovine isolates according to the standard method of the United States Centers for Disease Control and Prevention as described by Bach et al. (2005). The human clinical isolates were not characterized further, because patient confidentiality agreements precluded access to their PT and PFGE profiles. Banding patterns of the bovine isolates were viewed with UV illumination, and photographed using the Speedlight Platinum Gel Documentation System (Bio-Rad, Mississauga, ON, Canada). Banding patterns in the digital images were compared using Dice coefficients with unweighted pair group methods arithmetic average algorithm, 1% position tolerance and 0·5% optimization (Bionumerics 4·6, Applied Maths BVBA, Sin-Martens-Latem, Belgium).

Preparation of bacteria and phages for the microplate phage virulence assay

STEC O157:H7 bovine or human isolates were grown separately in tryptic soy broth (TSB) overnight at 37°C. One millilitre of the overnight culture was diluted with 9 ml of TSB for the microplate assay, and the titres of these cultures were determined by standard plate counts of serial 10-fold dilutions on MacConkey Agar (Dalynn Biologicals, Calgary, AB, Canada).

Phages (rV5, wV7, wV8 and wV11) were provided from the collection of lytic E. coli O157:H7 phages held by the Laboratory for Foodborne Zoonoses of the Public Health Agency of Canada (Guelph, ON, Canada). Three of these lytic phages were originally recovered from river water near a sewage treatment facility in Ontario (Ahmed et al. 1987). Phage rV5 was recovered from an E. coli O157:H7-inoculated calf that had been treated with a panel of six phages containing wV5, wV7, wV8 and wV11 (Waddell et al. 2000). Stock phages for the microplate assay were prepared for each phage by combining 100 μl of phage suspension (containing 108–109 PFU ml−1) with 1 ml of an early-log-phase culture of E. coli O157:H7 R508 (OD600 = 0·2–0·3) and incubating for 15–20 min at 37°C to allow attachment of phage to the host. This was followed by addition of 250 ml of TSB amended with MgSO4 at 10 mmol l−1 (mTSB), and further incubation at 37°C for 18–20 h with shaking (150 rev min−1). The overnight mixture was then centrifuged at 6000 g for 10 min and filtered through a 0·2-μm SFCA serum filter (Nalgene, Rochester, NY, USA). Titres of phages in the stock filtrates were then determined by a plaque assay using the soft agar overlay technique (Sambrook and Russell 2001). Stock preparations of rV5, wV7, wV8 and wV11 contained 9·16 × 109, 9·00 × 108, 5·5 × 1010 and 8·0 × 108 PFU ml−1, respectively, and were maintained at 4°C.

Microplate phage virulence assay

Susceptibility of STEC isolates to a given phage was assessed using a microplate phage virulence assay. Twenty-microlitre volumes of each phage were serially diluted 10-fold in 180-μl volumes of mTSB in 96-well microplates. Duplicate wells of each phage dilution were then inoculated with 20 μl of overnight cultures of each STEC O157:H7 isolate diluted 1 : 10 in TSB and the plates were incubated at 37°C for 5 h. Control wells were prepared in each microplate that contained mTSB inoculated with bacteria (E. coli O157:H7 R508) only as a negative control for lysis, wells with mTSB only as a blank, and wells with phage and a known susceptible bacterial strain (E. coli O157:H7 R508) as a positive control for complete lysis. After incubation, the wells were examined visually for turbidity due to bacterial growth. For each STEC isolate-phage combination, the greatest dilution of phage that resulted in complete lysis (no discernable turbidity) of bacteria was recorded. The MOI for each phage-host assay was calculated by dividing the initial number of phages in the greatest-dilution wells by the initial number of bacteria added, as determined from plate counts of serially diluted bacterial cultures. Isolates were then defined as sensitive (0·000001 ≤ MOI < 100) or resistant (MOI ≥ 100 as complete lysis did not occur) to the phages tested. Sensitivity to phage was further categorized as follows: extremely susceptible: (0·000001 ≤ MOI < 0·01); highly susceptible: (0·01 ≤ MOI < 1); moderately susceptible: (1 ≤ MOI < 10) and minimally susceptible: (10 ≤ MOI < 100).

Statistical analyses

The MOIs were analysed using mixed procedure and STEC O157:H7 susceptibility to the phages was evaluated by χ2 tests within the genmod procedure. Correlations of lytic capability among the four phages were assessed using the Spearman rank-order correlation coefficient (r). These analyses were conducted using SAS System for Windows, ver. 9.1 (SAS Institute, Cary, NC, USA). Relationships between the isolates’ genetic similarity, PT and phage susceptibility were assessed by χ2 tests using grouped PFGE data in BioNumerics.

Results

Phages wV7, rV5, wV11 and wV8 lysed 422, 342, 321 and 407 (i.e. 100%, 81·0%, 76·1% and 96·4%) of the 422 STEC O157:H7 isolates screened (297 bovine + 125 human), respectively (Table 1). Phages wV7, rV5 and wV11 had high lytic capabilities against the isolates screened (i.e. 0·01 ≤ MOI < 1).

Table 1.   Degrees of susceptibility* of human and bovine STEC O157:H7 to four E. coli O157:H7-infecting bacteriophages
Source and site†No. of isolatesPhage wV7Phage rV5Phage wV11Phage wV8
++++++±++++++±++++++±++++++±
  1. *Susceptibilities to the phages are grouped on the basis of multiplicity of infection (MOI, the lowest ratio of phage to bacteria that resulted in complete lysis of an overnight bacterial culture during 5 h of incubation with serial dilutions of the phage). +++, extremely susceptible (0·000001 ≤ MOI < 0·01); ++, highly susceptible (0·01 ≤ MOI < 1); +, moderately susceptible (1 ≤ MOI < 10); ± minimally susceptible (10 ≤ MOI < 100); –, non-susceptible (i.e. resistant: MOI ≥ 100).

  2. †Sites F1–F4 denote four beef feedlots in southern Alberta, Canada; D5–D8 denote four bovine dairies in southern Alberta, Canada.

Bovine isolates
 F1858230002152057201006401161121
 F28987200081700156292025029505
 F3515010004230063671071516272
 F4990000701107000210260
 D522220000210001210001201190
 D61212000012000010200010542
 D722220000220000211000002200
 D8770000700003400050002
Total2972916000213153165174443076151611613812
Human isolates
Total1251232000100541151000002516423793

Bovine STEC O157:H7 susceptibility to phages

Among the four phages, phage wV7 exhibited the overall lowest mean MOI (0·004 ± 0·002, P < 0·0001). Mean MOIs of phages rV5 (0·2 ± 0·08) and wV11 (0·04 ± 0·01) were similar to each other, but lower (< 0·0001) than that of phage wV8 (25 ± 2).

Phage susceptibilities, grouped by MOI, of the STEC O157:H7 isolates from the eight farm sites are shown Table 1. Phage wV7 lysed all 297 of the bovine isolates, with 291 classed as extremely susceptible and six as highly susceptible. This phage yielded the highest (P < 0·001) number of extremely susceptible isolates, followed by rV5, wV11 and wV8.

Phage rV5 showed slightly greater virulence for the bovine isolates compared with phage wV11, although mean MOI of these two phages did not differ (P > 0·1). STEC isolates ranking as extremely susceptible were more numerous (213 vs 174; P < 0·05) for phage rV5 than for phage wV11, and fewer isolates were found to be resistant to rV5 as compared to wV11 (65 vs 76). For both phages, ‘extremely susceptible’ was the most frequently determined (< 0·0001) isolate sensitivity rating, followed (< 0·001) by ‘highly susceptible’ (15 and 44 isolates for rV5 and wV11, respectively). For both of these phages, however, resistant isolates were more numerous (< 0·01) than for the fourth phage, wV8.

Phage wV8 was less virulent than the other three phages, returning higher proportions (P < 0·0001) of moderately- (39·1%) or minimally susceptible (46·5%) STEC isolates compared with extremely- or highly susceptible rankings. However, only 12 isolates (4%) were completely resistant to phage wV8.

Lytic capability of phages within feedlots and dairies

Site of origin differences were observed in the sensitivities of STEC O157:H7 isolates. Pooled across the four phages tested, mean MOI ranked as: ((D8, D6, F1) < F3 < (F2, D5, F4) < D7). Specifically, mean MOI for STEC O157:H7 isolates from dairies D8 and D6 (0·003 ± 0·0009 and 4 ± 2, respectively) and feedlot F1 (4 ± 0·7) were lower (P < 0·05) than MOI for isolates from the other five sites. Among those, MOI was lower (< 0·01) for isolates from site F3 (6 ± 1) than from F2, F4 or D5 (9 ± 1; 9 ± 2 and 10 ± 3, respectively), all of which had lower (< 0·05) MOI than did isolates from site D7 (11 ± 2).

The virulence (mean MOI) of phages rV5, wV7 and wV11 against the isolates did not differ (P > 0·1) whether compared across feedlots (F1–F4), dairies (D5–D8) or across all eight sites of origin. Phage wV8, however, exhibited a lower (< 0·0001) MOI against isolates from site D8 (0·0006 ± 0·0001) than from the other dairies, and against isolates from F1 (9 ± 2) than from the other feedlots.

The STEC isolates from dairy D8 exhibited similar susceptibilities (P > 0·05) to each of the phages, whereas from the seven other sites, the isolates were less susceptible (P < 0·0001) to wV8 than to the other three phages.

Human STEC O157:H7 susceptibility to phages

As observed with bovine STEC O157:H7 isolates, phage wV7 had the lowest MOI (0·006 ± 0·006) of the four phages against the human STEC O157:H7 isolates, and wV8 had the highest (29 ± 2; P < 0·0001). Mean MOIs of phages wV11 and rV5 against the clinical isolates were intermediate, at 0·002 ± 0·0003 and 0·4 ± 0·2, respectively.

All 125 clinical isolates were sensitive to phage wV7, with 123 of these (98·4%) classified as extremely susceptible (Table 1). One hundred isolates tested as extremely susceptible to each of phages rV5 and wV11, and 15 and 25 isolates, respectively, were resistant to rV5 and wV11. Phage wV8 had relatively lower virulence, with 79 of the clinical STEC O157:H7 isolates classifying as minimally susceptible to this phage, and three displaying resistance.

Lytic capability of phages bovine vs human STEC O157:H7 isolates

Overall, the four phages were more efficient (P < 0·001) at lysing the bovine STEC isolates than the human isolates. This difference was attributable to the relatively greater sensitivity (< 0·0001) of the bovine isolates to phage wV8, compared with the human isolates. Excluding phage wV8, isolates from the susceptibilities of human and bovine isolates to the other three phages were similar (P > 0·1).

Lytic capability of phages

A strong positive correlation was observed in the lytic capabilities of phages rV5 and wV11 against bovine (r = 0·85; < 0·0001) and human isolates (r = 0·99; < 0·0001). Sixty-five bovine and 15 human isolates that were resistant to phage rV5 were also resistant to phage wV11. Further, 173 bovine and 100 human isolates classifying as extremely susceptible to phage rV5 were also extremely susceptible to wV11. There was a slight correlation between the lytic capabilities of these phages to that of phage wV7 among the bovine isolates (r = 0·17–0·18; < 0·01) and among the clinical isolates (r = 0·24–0·26; < 0·01). For phage wV8, however, the correlations with phages rV5 and wV11 in lytic capability against bovine isolates were negative (r = −0·27 and r = −0·31, respectively; both < 0·0001), and there was no correlation (= −0·07; > 0·1) of lytic capability against human isolates between wV8 and any of the other three phages.

Gentic relatedness and STEC O157:H7 susceptibility to phages

Susceptibilities of bovine STEC isolates to phages rV5 and wV11 were correlated to their genetic relatedness (Table 2). Isolates that were sensitive to phage rV5 or phage wV11 differed genetically (P < 0·0001) from those that were resistant to those same phages. The STEC O157:H7 isolates sensitive to phage rV5 and those sensitive to phage wV11 shared approximately 80% genetic similarity, whereas those resistant to phage rV5 or phage wV11 shared genetic similarities of 75·0% and 77·0%, respectively. In contrast, a high degree of relatedness was observed only among certain sub-categories of isolates sensitive to phage wV8 (77·5% among those extremely susceptible; 79·4% among those classed as minimally susceptible). The 15 STEC isolates in the extremely susceptible category, and the 138 isolates in the minimally susceptible category both differed genetically (P < 0·05) from those exhibiting other degrees of susceptibility to phage wV8 (highly susceptible; moderately susceptible).

Table 2.   Genetic relatedness (PFGE similarities) and phage types observed among bovine STEC O157:H7 grouped as susceptible (S) or resistant (R) to four E. coli O157:H7-infecting bacteriophages
 Phage wV7Phage rV5Phage wV11Phage wV8
SRSRSRSusceptibility to wV8*R
++++++±
  1. *Susceptibility ratings are based on multiplicity of infection (MOI, the lowest ratio of phage to bacteria resulting in complete lysis of overnight bacterial culture during 5 h of incubation). +++, extremely susceptible (0·000001 ≤ MOI < 0·01); ++, highly susceptible (0·01 ≤ MOI < 1); +, moderately susceptible (1 ≤ MOI < 10); ± : minimally susceptible (10 ≤ MOI < 100); –, non-susceptible (i.e. resistant: MOI ≥ 100).

  2. †As determined from PFGE profiles produced from XbaI-digested genomic DNA. NS, not significant (P > 0·05). NA, not applicable.

  3. ‡Significance of differences in relatedness of STEC isolates within and across susceptible/resistant status.

  4. NA, Not applicable; AT, atypical.

No. of isolates29702326522176151611613812
PFGE similarity (%)†
Within S/R statusNANA79·877·080·175·077·575·473·679·477·0
Across S/R statusNA 69·8 70·8 75·776·075·176·076·4
P value‡NA <0·0001 <0·001 <0·05NSNS<0·0001NS
Phage types observed           
I1, 4, 8, 14, 14a, 14cNA1, 4, 8, 14, 14a, 14c14a1, 4, 8, 14, 14a, 14c1, 14a14, 14a14a1, 4, 8, 14, 14a1, 4, 8, 14, 14a, 14c1, 8, 14a
II21, 32 21, 32  21,32 3221, 32  
III23,54  23, 54 23, 54 2323, 5423, 54 
IV34,45 3434, 45 34, 45 4534, 454534, 45
V63, 67, 74 6367, 746367, 7463, 74 67  
VI47, 87 47, 87 47, 87 87 47  
OtherAT AT AT AT  ATAT

Bovine STEC O157:H7 susceptibility to phages and PT

Two hundred and ninety two of the STEC O157:H7 bovine isolates were classified into 17 different PT: PT1, 4, 8, 14, 14a, 14c, 21, 23, 32, 34, 45, 47, 54, 63, 67, 74 and 87 (Tables 2 and 3). Only 5 of the 297 isolates classed as atypical. On the basis of similarities of reactions during typing (Frost et al. 1989; Khakhria et al. 1990), the 17 PTs could be grouped into main profile groups, named I–VI (Table 3). Over 87% of the isolates were accounted for in six of the 17 PT. These were PT 1, 8, 14 and 14a, from main profile group I, PT 23 (profile group III) and PT 45 (profile group IV). Nearly all (92·3–100%) of the isolates belonging to these six PTs were sensitive to the four phages tested with the exception of the PT23 and PT45 isolates, all of which were resistant to phages rV5 and wV11 (Table 4).

Table 3.   Phage type characteristics of bovine STEC O157:H7 isolates grouped by similarity of their reactions with typing phages
Main profile groupPhage type classificationNo of isolatesIdentities of the typing phages to which the isolates reactions were distinctive*
  1. *Within a phage-typing main profile group, the STEC isolates shared similar reactions to the 16 typing phages except those specifically noted (Ahmed et al. 1987; Frost et al. 1989; Khakhria et al. 1990). Five of the 297 isolates returned atypical results from phage-typing.

I (n = 210)1
4
8
14
14a
14c
18
6
39
21
124
2
6, 9, 10, 13
II (n = 8)21
32
3
5
6, 13
III (n = 29)23
54
26
3
6, 13
IV (n = 33)34
45
2
31
5, 6, 13, 14
V (n = 4)63
67
74
1
2
1
3, 7, 9, 10
VI (n = 8)47
87
1
7
2, 6, 13, 14
Table 4.   Prevalence of sensitivity to four bacteriophages and similarity of PFGE profiles among STEC O157:H7 isolates displaying the most commonly observed phage-typing phenotypes
Phage type (PT)No. of isolatesSensitivity to phage (% of isolates)Similarity of PFGE profiles (%)*
wV7rV5wV11wV8Within PTBetween PTP value†
  1. *PFGE profiles were produced from XbaI-digested genomic DNA.

  2. †Significance of difference in relatedness of STEC isolates calculated within and across phage types.

11810010094·494·484·374·8<0·0001
83910010010092·382·773·0<0·0001
142110010010010092·579·1<0·0001
14a12410010098·410083·474·9<0·0001
23261000010082·167·5<0·0001
45311000096·880·570·5<0·0001

Most of the isolates sensitive to phage rV5 belonged to profile groups I, II and VI, whereas those sensitive to phage wV11 mainly belonged to profile groups I and VI. Isolates belonging to profile groups III (PT23, PT54), IV (PT45) and V (PT67 and PT74) were resistant to phage rV5, accounting for 97% of the 65 resistant isolates. All isolates belonging to profile groups II (PT21, PT32), III (PT23, PT54), IV (PT34, PT45) and V (PT67, PT74) were resistant to phage wV11, accounting for 96% of 76 resistant isolates. The isolates extremely susceptible to phage wV8 belonged to profile groups I (PT14, n = 1; PT14a, n = 4), V (PT63, n = 1; PT74, n = 1), VI (PT87, n = 7) or were atypical (n = 1). One hundred and thirty-eight isolates minimally-susceptible to phage wV8 were mainly from profile group I (PT1, n = 12; PT8, n = 21; PT14, n = 19; PT14a, n = 67 and PT14c, = 2). The isolates resistant to phage wV8 were classed as PT1 (n = 1), PT8 (n =5), PT14a (n = 2), PT34 (n = 1), PT45 (n = 1) and atypical (n = 2).

Susceptibility of isolates to three of the phages was strongly associated with their PT (Table 2). With the exception of PT34 (which was represented by 2 isolates only), 100% of the isolates in main profile groups II, III, IV, V and VI, and 98·6% of the isolates in profile group I (i.e. all except one PT1 and two PT14a isolates) were observed to be exclusively susceptible or resistant to phages wV7, rV5 and wV11. This segregation was also true for reaction to phage wV8, with the exception of seven isolates in main profile group I (1/18 PT1, 5/39 PT8 and 2/124 PT14a) and a single isolate (1 of 31 PT45) in main profile group IV. These PT-specific reactions were further reflected in distinctive patterns in terms of their susceptibility to the four phages that we investigated. For profile groups I to VI, overall patterns of reaction to phages wV7, rV5, wV11 and wV8 (where S = sensitive, or s = slightly sensitive re: wV8; R = resistant; and M = mixed, i.e. split by PT) were: SSSM, SSRS, SRRs, SRRM, SMMS and SSSS, respectively (Table 2).

Phage types and genetic relatedness of STEC O157:H7

Genetic similarities among the bovine STEC O157:H7 isolates within the six most commonly observed PT (Table 4) ranged from 80·5% to 92·5%. These similarities were greater (P < 0·0001) than those that existed between PT, indicating that isolates within a PT differed genetically (P < 0·0001) from those in other PT.

With the exception dairy D8, the STEC isolates collected from each farm site (genetic similarities 73·6–94·0%, Table 5), were found to be have greater (P < 0·0001) genetic similarity (i.e. relatedness) than isolates compared across sites of origin. Seven of the eight farms yielded more than one profile group and six sites had ≥2 profile groups. Profile group I was predominant across feedlots and dairies.

Table 5.   Phage type distribution of bovine STEC O157:H7 isolates from eight southern Alberta sites, and analysis of their relatedness as determined from PFGE profiles
Site of origin‡nProfile group and phage type*Profile group(s) identifiedPFGE similarity (%)†
IIIIIIIVVVIATWithin siteBetween sitesP value§
1481414a14c2132235434456367744787
  1. *Determined using 16 phages that differentiated 88 types (Ahmed et al. 1987; Khakhria et al. 1990). AT, atypical.

  2. †PFGE profiles were produced from XbaI-digested genomic DNA.

  3. ‡Sites F1–F4 denote four beef feedlots; D5–D8 denote four bovine dairies.

  4. §Significance of differences between relatedness of isolates calculated within and across sites of origin.

F1852 1 18 1524 130 2 1  I to VI73·672·3<0·0001
F289 634 412      1 1 22I, V, VI80·375·6<0·0001
F35112  131   231      1I, III, IV81·276·8<0·0001
F49   25 2           I, II84·878·1<0·0001
D5223   18      1      I, IV90·877·9<0·0001
D6121 4 7             I83·477·6<0·0001
D722   184             I94·079·6<0·0001
D87                52VI75·171·5NS
Total2971863921124235263231121175    

Discussion

This study was the first to investigate lytic capability and host range of phages to endemic STEC O157:H7 isolated from cattle and humans and to compare these lytic properties to PT phenotype, and PFGE genotype of the isolates. Given that the phage investigated lysed naturally occurring STEC O157:H7 isolates of bovine and human origin, this cocktail may have the potential to mitigate STEC O157:H7 within feedlots. The fact that the cocktail also possessed activity against clinical isolates from humans is particularly promising. Phages wV7, rV5, wV11 and wV8 are members of the Myoviridae family which exhibits a broad host range (Wichels et al. 1998). Of the four phages examined, wV7 exhibited the highest lytic activity and the broadest host range for STEC O157:H7. Phages rV5 and wV11 exhibited a low MOI, but possessed a narrower host range. In contrast, phage wV8 had a broad host range, but had the highest mean MOI, indicating that its lytic activity was limited. These results emphasize the importance of considering both MOI and host range when selecting phage for therapeutic purposes.

Phage infection is initiated by specific adsorption of phage onto the bacterial cell surface. In most cases, failure of coliphage to attack a given strain of E. coli has been shown to be due to lack of adsorption (Weinbauer 2004). The host range is controlled by the interaction of the phage with receptors on the host cell surface. Phage wV7 was capable of lysing all 422 STEC O157:H7 isolates, irrespective of their PT phenotype or PFGE genotype. Presumably, all of these STEC O157:H7 isolates share a conserved or similar surface component that was recognized as a receptor site by phage wV7. Inoue et al. (1995a,b) demonstrated that an outer membrane protein common to Vibrio species served as the receptor for a broad-host-range vibriophage belonging to the Myoviridae. Tail fibre proteins are also used by phage to identify and attach to bacterial hosts (Scholl et al. 2001), thus those phages that possess a broader host range may possess more than one specific tail fibre protein.

All of the STEC O157:H7 isolates screened, irrespective of their origin, were sensitive to at least one of four phages. This sensitivity may have arisen from the isolates displaying a common lipopolysaccharide (LPS) O-antigen, one of major cell receptors recognized by phages (Heller and Bryniok 1984; Sharma et al. 2008). LPS in the outer membrane of Gram-negative bacteria is composed of lipid A, core oligosaccharide and repeating O-antigen subunits. Like many enteric bacteria, E. coli O157 produces LPS that contains an extensive O-antigen (Perry et al. 1986). The relatively narrower host range of phages rV5 and wV11 (81·0% and 76·1% of isolates screened) as compared to wV7 and wV8 (100% and 96·4% of isolates screened) might be attributable to the impact of structure and/or length of LPS O-chain on adherence of the phage to the host cell (Iguchi et al. 2007). This possibility is supported by the observations of Kudva et al. (1999) who found that E. coli O157 strains with a truncated LPS were resistant to O157-specific phages.

Outer membrane proteins, alone or in conjunction with LPS, can also serve as receptors for recognition of O157-lysing phages (Morita et al. 2002). Thus, the presence of STEC O157:H7 isolates resistant to the phages may indicate that in some instances, recognition of multiple receptors may be required to initiate a successful attack on the host cell (Heller 1992). In contrast, isolates sensitive to phages rV5 or wV11 shared similar PFGE patterns, and were classed into a series of related PTs, indicating that certain PFGE genotypes and PT phenotypes of STEC O157:H7 have a common degree of susceptibility to specific phages.

Adsorption rate, latent period and burst size are proposed as factors in determining the proliferation rate of a given phage (Adams 1959). These biological parameters depend on PT, host physiology and nutritional conditions (Weinbauer 2004). Theoretically, a high phage proliferation rate would accelerate lysis of the host cell and reinfection of new cells, thereby increasing lytic capability of the phage. In this study, the majority of STEC O157:H7 isolates were extremely- or highly susceptible to phages wV7, rV5 and wV11 irrespective of PT phenotype or PFGE genotype. This suggests that these three phages are likely capable of generating progeny rapidly (Arisaka 2005) using a broad range of hosts, properties that would make them effective biocontrol agents. Selection of phages with characteristics such as these could decrease the dosage needed to make phage therapy a viable mitigation strategy for STEC O157:H7 in cattle.

In contrast to wV7, rV5 and wV11, the lytic capability of phage wV8 depended on PT phenotype and PFGE genotype of the isolates. The isolates that were extremely- or minimally susceptible to phage wV8 were genetically closely related within susceptibility group. The extremely susceptible isolates were predominantly those of main profile group VI, whereas those minimally susceptible to wV8 were mainly of profile group I. Moreover, correlations of lytic capabilities between phage wV8 and the other three phages varied with the host isolates’ sites of origin, also indicating subtypes of STEC O157:H7 differ in their susceptibility to this phage. This would suggest that the relative efficacy of a simpler phage cocktail that consisted of only wV8 would be farm-dependent.

The broad host range and strong lytic capability of phage wV7 both in bovine and in human isolates across PT phenotypes and PFGE genotypes makes this phage a strong candidate for inclusion in a phage cocktail. Observations of wV7 also suggest the possible effectiveness of using a single phage to combat diverse STEC O157:H7 isolates prior to or independent of identifying their subtype. In vivo studies have demonstrated that treatment with a single phage allowed mice to recover from infections caused by E. coli (Smith and Huggins 1982), Pseudomonas aeruginosa (Watanabe et al. 2007), methicillin-resistent Staphylococcus aureus (Capparelli et al. 2007) and vancomycin-resistent Enterococcus faecium (Biswas et al. 2002). Watanabe et al. (2007) reported that treatment with multiple P. aeruginosa-infecting phages was not more efficacious than treatment with a single phage at eliminating this bacterium from the gastrointestinal tract of mice.

Using a single phage as an antimicrobial agent would permit a more precise definition of pharmacokinetics, an approach that would be expected to simplify the regulatory approval of this therapeutic approach (Carlton et al. 2005). However, it is probable that STEC O157:H7 isolates may also rapidly develop resistance to a single phage, as has been observed in in vitro (Kudva et al. 1999; Tanji et al. 2005) and in vivo (Rozema et al. 2009). A cocktail of rV5 and wV11 along with wV7 may prevent or delay the emergence of resistant bacterial strains. Moreover, use of a multi-phage cocktail may maintain therapeutic efficacy due to phage-resistant STEC O157:H7 mutants remaining sensitive to a different to phage in the cocktail (Yoichi et al. 2004), to evolution of phage mutants (Mizoguchi et al. 2003) or to possible loss of virulence factors in the host (Santander and Robeson 2007). Further investigation would be required to identify whether or not phage wV8, with its broad host range but limited lytic ability, would enhance the therapeutic effectiveness of a phage mixture.

The strong correlation of lytic capabilities of phages rV5 and wV11 against bovine and human isolates, with host ranges differing only with regard to main profile group II (Table 2), suggests that they are analogous, an observation that is corroborated by other work that has shown that these two phages have similar genome size and tail fibres (R.P. Johnson, unpublished data). Phage rV5 was isolated during a challenge experiment in which another phage, wV5 was administered to dairy calves (Waddell et al. 2000). The slightly broader host range and greater virulence of rV5 compared with wV11 could be attributable to virulent evolution of rV5 as a result of within-host competition and adaptation from its parent wV5. Capparelli et al. (2006) demonstrated that phage ΦD, a derivative of phage ΦW recovered from the circulatory system of mice inoculated with E. coli O157:H7 after phage ΦW treatment, cleared STEC O157:H7 in the mice more efficiently than did phage ΦW.

PT of STEC O157:H7 differ geographically among continents (Mora et al. 2007) and at least 88 PT have been documented for STEC O157:H7 (Ahmed et al. 2001). Five (PT1, PT8, PT14, PT14a and PT23) of the six predominant subtypes observed in our study, along with sporadic subtypes (PT4, PT21, PT32, PT34, PT54 and PT87) have previously been described in Canada (Woodward et al. 2002), Australia (Leotta et al. 2008), Europe (Mora et al. 2004; Carroll et al. 2005), South America (Leotta et al. 2008) and Japan (Nishikawa et al. 2001). In the present study, phage wV7 was able to lyse all of the PT21, PT23 and PT34 bovine STEC O157:H7 isolates examined, and some of those isolates were sensitive to rV5, wV11 and wV8 as well.

It is noteworthy that 7·1% and 41·8% of the bovine STEC O157:H7 isolates studied belonged to PT14 and PT14a, which have been demonstrated to be associated with numerous sporadic cases and outbreaks of disease in Canada (Woodward et al. 2002). Specifically, PT14a was the predominant PT of human and non-human STEC O157:H7 isolates collected across Canada in 2004 (R. Ahmed and W. Demczuk, personal communication). The four phages evaluated were each able to lyse virtually all of the isolates within PT14 and PT14a, suggesting that they are effective against those STEC O157:H7 that are most likely to cause disease in Canadians. In this study, genetic similarity was confirmed for the bovine isolates within each of the six most common PT groups, which agrees with other findings that a relationship exists between PT and genotype in STEC O157:H7 (Nishikawa et al. 2001; Guth et al. 2003).

The host range and lytic capability of phages fundamentally determine their potential for therapeutic application. Those with a broad host range and strong lytic capability would be superior candidates for phage therapy. In the present study, selection of phages as candidates to control STEC O157:H7 was based on consistently low MOI as determined by microplate phage virulence assay. The MOI is a simple and rapid quantitative parameter that can be employed as a criterion by which to select therapeutic phages.

In conclusion, the four phages selected for study were capable of lysing STEC O157:H7 isolates endemic to northern, southern and central Alberta collected from bovine (feedlots and dairies) and human (clinical case) sources. Among these phages, wV7 exhibited the broadest host range and strongest lytic capability against all the STEC O157:H7 isolates. In contrast to phages rV5, wV7 and wV11, phage wV8 showed more limited potential for phage therapy due to lower activity even though it did exhibit a broad host range. With the exception of phage wV7, host range and/or lytic capability of phages rV5, wV11 and wV8 did depend on PT and PFGE genotype of the targeted cells. Both MOI and host range should be considered in the selection of therapeutic phage. The present work suggests that our phage cocktail should be equally effective at mitigating STEC O157:H7 isolates of bovine and human origin. Susceptibility of bovine STEC O157:H7 to an individual phage (e.g. wV8) may be farm-dependent. Given the potential for some STEC O157:H7 isolates to be phage-resistant, as well as for others to develop resistance upon phage exposure, a cocktail of phages may be the most efficacious approach to controlling this pathogen in feedlots.

Acknowledgements

Financial support for this research project was provided by the Food Safety Initiative of Alberta Agriculture and Rural Development and by the Beef Cattle Research Council. We gratefully acknowledge G. L. Wallins, H. Zahiroddini, C. Agopsowicz and A. Mazzocco (Public Health Agency of Canada, Guelph, ON) for technical assistance and support. Review of the manuscript by K. Jakober is also greatly appreciated. The provision of E. coli O157:H7 R508 host by Public Health Agency of Canada, Guelph, ON is also duly recognised.

Ancillary