Two bacteriophages, isolated from faeces, were assayed as biocontrol agents of pathogenic Escherichia coli during milk fermentation. Phage DT1 was tested on the strain E. coli DH5α, one enteropathogenic E. coli (EPEC) strain and one Shiga toxigenic E. coli O157:H7 (STEC) strain. Phage DT6 was tested on two STEC strains (O157:H7 and non-O157). One additional assay was performed by using a cocktail of both phages against the O157:H7 STEC strain. Streptococcus thermophilus 10-C, the strain used as lactic starter, reached 109 CFU ml−1 after 4 h, while pH values fell to 4·5 after 8 h, regardless of the presence of E. coli strains and/or phages. In absence of phages, E. coli strains reached 4–6 log CFU ml−1 at 5–6 h. Escherichia coli DH5α and O157:H7 STEC strains were rapidly and completely inactivated by phage DT1 and phage cocktail, respectively, while O157:H7 STEC was completely inactivated either by DT1 or by DT6, after 8 h. The EPEC strain was not detected at 1 h (<10 CFU ml−1) but grew afterwards, though at lower rates than without phage. For non-O157:H7 STEC, reductions lower than 1 log CFU ml−1 were observed for all sampling times. Phages DT1 and DT6, either individually or as a cocktail, effectively reduce O157:H7 STEC counts during milk fermentation, without compromising the starter culture performance.
Significance and Impact of the Study
Coliphages DT1 and DT6, isolated from faeces and selected on the basis of their host range, showed to be valuable tools for the control of pathogenic Escherichia coli during milk fermentation, without compromising the starter culture performance. Both phages, either individually or as a cocktail, may function as an extra safety barrier beyond traditional pasteurization, effectively reducing O157:H7 Shiga toxin-producing Escherichia coli (STEC) counts during early growth, thus avoiding Shiga toxin production and accumulation.
The incidence of foodborne pathogens such as Salmonella spp., Shigella spp. and Escherichia coli continues to increase considerably in many countries (Rivas et al. 2008; Farrokh et al. 2012). Particularly, Shiga toxin-producing Escherichia coli (STEC) are human pathogens that can cause diarrhoea, as well as severe clinical diseases including haemorrhagic enterocolitis, haemolytic uraemic syndrome (HUS) and thrombotic thrombocytopenic purpura (Su and Brandt 1995; Griffin et al. 2002). Moreover, the enterocyte attaching-and-effacing lesion gene (eaeA) present in enteropathogenic strains (EPEC) could contribute to the virulence of STEC (Frankel et al. 1998). Recent epidemiological studies showed that there is a sustained global increase in the isolation of emerging non-O157 STEC serogroups responsible of infection: E. coli O26, O103, O111, O121, O45 and O145 (Mathusa et al. 2010). In Argentina, HUS is endemic, with an annual rate incidence of 13·9 cases per 100 000 children under 5 years of age, as informed by Hospital Nephrology Units (NCASP 1995; Roldán et al. 2007; Governmental Agency of Control 2012).
Ground beef is still the most frequent source of E. coli outbreaks (Vugia et al. 2009), but dairy products can be directly contaminated by cattle faeces during milking process as well (Fremaux et al. 2008). Epidemiological studies on this matter led to the isolation of E. coli O157:H7 from milk handling pipes and bottling machines in dairy plants, indicating that inadequate pasteurization or postpasteurization contamination may have caused the outbreak. As a consequence, many studies have emphasized the design and application of several complementary strategies to reduce the incidence of these foodborne diseases related to dairy products (Viazis and Diez-Gonzalez 2011). Due to the high heat sensitivity of these pathogens, thermal treatments are the most widely applied in order to inactivate them when present in raw milk. However, the problem can not be solved if postpasteurization contamination occurs, since the survival of E. coli O157:H7 after 28 days in milk at 5°C (Wang et al. 1997) and after 21 days in cheese whey at 4, 10 or 15°C (Marek et al. 2004), was demonstrated. Consequences might be even worse if we consider that pasteurization temperatures were not validated for free Shiga toxin (Rasooly and Do 2010), which requires 5 min at 100°C for inactivation. Thus, the possible persistence of the heat-resistant toxin reinforces the need for a method to control the pathogen during early growth in order to block its production. Antibiotics are the most common alternative, but are restricted for application in foods. Taking into consideration the increasing antibiotic resistance of some pathogens, natural strategies such as phage control seem promising. Although reported effectiveness was variable, many studies have reported the isolation and use of E. coli bacteriophages as biocontrol tools in several food matrices, including meat and vegetables (O'Flynn et al. 2004; Abuladze et al. 2008; Viazis et al. 2011). Furthermore, the use of phage cocktails was highlighted as an enhanced tool when E. coli isolates exhibited resistance to some, but not all studied phages (Kudva et al. 1999; Viazis and Diez-Gonzalez 2011). Beyond the advantages of natural origin and high specificity, the ability of phages to act constantly, even during the manufacture process and food storage, make them a very interesting tool to be considered.
In this study, two bacteriophages isolated from faeces and selected on the basis of their host range were studied for their capability as biocontrol agents of pathogenic Escherichia coli during milk fermentation.
Results and discussion
Isolation, specificity and characterization of bacteriophages
Two bacteriophages (DT1 and DT6) obtained from diarrhoeic stool samples, which formed clearly defined plaques and exhibited different host ranges (Table 1), were selected for the experiments. Concentration of working stocks was 1·2 × 1010 PFU ml−1 and 4·8 × 1010 PFU ml−1 for phages DT1 and DT6, respectively.
|Phage||Escherichia coli sensitive strains||Total sensitive strains|
|DT1||Two EPEC; one O157:H7 STEC; one non-O157 STEC; two UDEC||6|
|DT6||Four EPEC; six O157:H7 STEC; three non-O157 STEC; three UDEC||16|
Awareness of the presence of virulence factors and phage-encoded genes is a prerequisite for phages to be included in food products for biocontrol purposes, since bacterial phenotype may be modified by phage gene expression (Miller and Day 2008). PCR amplification of major virulence factors (Stx1; Stx2; ST1; LT1 and Intimin) was negative for bacteriophages DT1 and DT6 (data not shown) making them suitable candidates for biocontrol applications. However, further sequencing and bioinformatic analysis are required to ensure that phages are free from other harmful factors and, in consequence, completely safe for use as biocontrol of pathogens in food. According to electron micrographs, phages DT1 and DT6 could be classified as T-even type of the Myoviridae family. Phages DT1 and DT6 had icosahedral heads and contractile tails. DT1 dimensions were of 89·3 ± 2·2 nm (head diameter), 127·8 ± 2·3 nm (tail length), 20·8 ± 1·0 nm (tail thickness) and 217·1 ± 4·3 nm (total length). DT6 measures were of 82·1 ± 1·5 nm (head diameter), 125·7 ± 2·0 nm (tail length), 17·7 ± 2·1 nm (tail thickness) and 207·8 ± 3·1 nm (total length) (Fig. 1).
The 35 non-E. coli and nonpathogenic E. coli strains tested for assessing the host range were resistant to the lytic action of both phages, while all O157:H7 strains, most non-O157 STEC and some EPEC isolated from diarrhoeic faeces or food, were sensitive to one or both bacteriophages (Table 1). Although these findings encourage the use of the phages for food conservation without altering the gastrointestinal tract normal flora, a broader host range needs to be assessed to ensure safety for commensal bacteria, since no correlation between nonpathogenic E. coli and phage sensitivity has been reported by other authors. Although prophage induction is usually related to the expression of virulence factors and many phage-encoded virulence genes exist as well, no correlation between E. coli strains lacking virulence genes and phage sensitivity has been reported. Despite the reported occurrence of O157-specific bacteriophages able to lyse all of the E. coli O157 (and none of the non-O157 E. coli or non-E. coli) strains tested (Kudva et al. 1999), no correlation was found between serogroups (O157 and non-O157) and phage sensitivity/resistance phenotype in our study. Likewise, other authors found phages to be species-specific, but usually capable of infecting more than one single serogroup (e.g. the virulent phage P100 can infect most Listeria monocytogenes strains; Carlton et al. 2005). On the other hand, one of the main drawbacks on the use of bacteriophages as biocontrol agents is their ability to lysogenize bacteria. In this study, phages DT1 and DT6 produced clear plaques, suggesting a lytic phenotype. In addition, BIMs (bacteriophage insensitive mutants) were isolated for both phages at a frequency similar to the normal occurrence of point mutations (ranging from 6·5 × 10−7 to 1·3 × 10−6) (Tomat, D., Mercanti, D., Balague, C., Quiberoni, A., unpublished data), whereas frequencies reported for temperate phages are much higher (about 10−4, Garcia et al., 2007). Besides, all of our BIMs reverted to phage sensitivity and no lysis plaques could be recovered after exposition to UV light, additional evidence of the incapability of the phages to lysogenize bacteria (Tomat, D., Mercanti, D., Balague, C., Quiberoni, A., unpublished results).
Biocontrol tests during milk fermentation
Phages DT1 and DT6, either alone or mixed in a cocktail, were evaluated for their efficiency to inhibit the growth of E. coli during milk fermentation. Phage DT1, with a narrower host range, was tested on the strains DH5α, EPEC920 and O157:H7 STEC464, whereas phage DT6 was tested on the two STEC strains (O157:H7 and non-O157). One additional assay was conducted by using a cocktail of DT1 + DT6 phages against the strain O157:H7 STEC464. The growth of Streptococcus thermophilus 10-C was similar in all the experiments, as well as the decrease of pH, regardless the presence of E. coli strains and phages tested (Fig. 2). Strep. thermophilus reached 109 PFU ml−1 after 4 h and maintained this level all throughout the experiment (24 h) (Fig. 2). The pH values evolved accordingly, falling to 4·5 at 8 h and 4·0 at the end of the fermentation process (24 h) (Fig. 2).
Excluding the phage cocktail (DT1 + DT6), which showed a slight but constant reduction (1·6 log PFU ml−1 after 24 h) (Fig. 3d), all phage titres remained constant (Fig. 3a,c,e) or increased slightly (0·5 log PFU ml−1) (Fig. 3b,f) throughout the first 8 h, with a subsequent decrease between 8 and 24 h (1–2 log PFU ml−1). The low pH and the accumulated lactic acid might be related to the partial phage inactivation observed at the final of the fermentation process. Regarding E. coli strains, phage DT1 and phage cocktail rapidly (<1 h) and completely inactivated DH5α [multiplicity of infection (MOI) = 2·3 × 105] and O157:H7 STEC (MOI = 2·2 × 104), respectively (Fig. 3a,d). Escherichia coli O157:H7 STEC464 was completely inactivated by either phage DT1 or phage DT6 at 8 h (Fig. 3e,f). Escherichia coli EPEC920 was not detected at 1 h (<10 CFU ml−1) after infection with phage DT1 at a MOI of 2·4 × 104 but showed a subsequent regrowth, although always lower than the control without phage, and yielding a reduction of 1·1 log CFU ml−1 after 24 h (Fig. 3b). For non-O157:H7 STEC ARG4827 (MOI = 3·5 × 103), reductions lower than 1 log CFU ml−1 were obtained for all sampling times (Fig. 3c). In control assays without phages, all the E. coli strains reached 4–6 log CFU ml−1 after 5–6 h of incubation (Fig. 3).
Regardless the presence of phages, a decrease on the viability of E. coli strains was observed as a result of the low pH attained during the acidification, resulting DH5α (1·5 log CFU ml−1 at 24 h) and O157:H7 STEC464 (2·3 log CFU ml−1 at 24 h) the most affected ones. Even if acidity is known to affect growth and survival of STEC and EPEC strains, the pH values reached during milk fermentation are not enough to assure their inactivation (Farrokh et al. 2012). Moreover, acid adaptation to nonlethal pH has been suggested to enhance E. coli O157:H7 survival (Jordan et al. 1999). The antagonistic effect of organic acids, hydrogen peroxide and bacteriocins produced by dairy starters may also contribute to E. coli inhibition (Akpinar et al. 2011). Nevertheless, the inhibitory effect of low pH was only observed after a considerable E. coli multiplication occurred, when the risk of Shiga toxin accumulation is high. Consequently, and considering our results, the use of phages DT1 and DT6, either individually or as a cocktail, could be a valuable tool to control E. coli early growth, especially O157:H7 STEC464, and therefore limiting toxin production and accumulation. It is worth noting that phages DT1 and DT6 were more effective as a cocktail than individually, as total inactivation of the pathogen O157:H7 STEC464 was achieved after 1 h of incubation in the first case, but required at least 8 h for individual phages. O'Flynn et al. (2004) also reported shorter inactivation times when using phage cocktails during in vitro challenge tests.
On the other hand, the pathogens EPEC920 and non-O157 STEC ARG4827 were not eliminated by phages DT1 and DT6, respectively. Successful phage infection and host killing is strongly dependent on MOI (Cairns et al. 2009). The MOI used for non-O157:H7 STEC (3·5 × 103) and EPEC (2·4 × 104) may be not high enough to achieve a full biocontrol, though reduction of viable cells counts of those strains would reduce the possibility of toxin production. Low MOI were tested during milk fermentation assays, but effectiveness was low and a subsequent bacterial regrowth was observed (preliminary results, not shown). Other authors largely agree in using high MOI to obtain large bacterial reductions on food matrices (e.g. MOI = 106, O'Flynn et al. 2004; 6 × 107 PFU/cm2, Carlton et al. 2005), as treatments with lower MOI values eventually led to bacterial regrowth. O'Flynn et al. (2004) assayed specific phages for E. coli O157:H7 on nine meat pieces, seven of which were completely host free after enrichment, but MOI used in that study were as high as 106. Kim et al. (2007) found that phage concentration between 107 and 109 PFU ml−1 (MOI = 105 to 107) was able to significantly inhibit the growth of Enterobacter sakazakii in reconstituted infant formula, although in a concentration-dependent way, as lower phage concentrations resulted in Enterobacter sakazakii regrowth. Taking into account the foregoing and evaluating the encouraging results obtained with DH5α and O157:H7 STEC strains, phages might be further concentrated up to about 1012 PFU ml−1 using PEG8000 and CsCl gradients, to test whether or not higher MOI values allow complete non-O157:H7 STEC and EPEC inactivation. Nowadays, it is possible to achieve such high concentrations at industrial scale. Indeed EcoShield™, a phage-based product commercially produced in large scale by Intralytix (www.intralytix.com/Intral_Food.htm), is composed by three lytic phages at a final concentration of 2·7 × 1011 PFU/ml.
Although many authors have reported the use of bacteriophages to control various foodborne bacterial pathogens, either in animals or in food matrices (Viazis and Diez-Gonzalez 2011), the present study is to our knowledge the first one focused on the biocontrol of EPEC and STEC during a milk fermentation process. Phages DT1 and DT6 were effective to reduce E. coli counts without compromising the performance of Streptococcus thermophilus, used as starter culture. In this sense, as Lactococcus lactis is even more extensively used than Streptococcus thermophilus worldwide (Mills et al. 2011), it is interesting to think about the usefulness of phages DT1 and DT6 as biocontrol tools in milk fermented by these mesophilic bacteria, either alone or as part of mixed cultures (Harnett et al. 2011). However, some parameters like temperature evolution and pH decrease during the fermentation process must be considered. Regarding temperature, and taking into account results previously obtained in meat during in vitro biocontrol assays, efficiency of both phages DT1 and DT6 at 30°C was not significantly different than at 37°C (unpublished data). With respect to acidification, final pH of milk is within the same range (4·0–4·5) regardless of the starter used (Streptococcus thermophilus or Lactococcus lactis). Therefore, a similar phage performance is expected in both cases.
Results suggest that novel phages such as those isolated in this study may indeed be applicable to the control of problematic pathogenic E. coli in dairy products. In summary, this research highlights that coliphages have significant potential as antimicrobial agents in milk and may function as an extra defence beyond traditional pasteurization. Their widespread occurrence and low cost of isolation and production may lead to their future exploitation in the dairy industry.
Materials and methods
Bacterial strains and culture conditions
Escherichia coli DH5α was used as sensitive strain to propagate all the bacteriophages used in this study. Three additional strains were used in the biocontrol experiments. Two of them, E. coli enteropathogenic (EPEC920) (eae) and E. coli Shiga toxigenic O157:H7 (STEC464) (stx2 and eae), were isolated from stool samples, identified using API-20E system (Biomerieux, Buenos Aires, Argentina), and further characterized by PCR. The third strain was E. coli Shiga toxigenic non-O157:H7 (STEC) (ARG4827; serogroup O18; stx1 and stx2) (Balagué et al. 2006). The E. coli strains were routinely reactivated overnight (37°C) in Hershey broth (8 g l−1 Bacto nutrient broth, 5 g l−1 Bacto peptone, 5 g l−1 NaCl and 1 g l−1 glucose) (Difco, Detroit, MI, USA) supplemented with MgSO4 (5 mmol l−1) (Cicarelli, San Lorenzo, Santa Fe, Argentina) (Hershey-Mg). The strain Streptococcus thermophilus 10-C was used as starter culture for milk fermentation (Suárez et al. 2002); it was routinely reactivated and grown overnight in Elliker broth (Biokar, Beauvais, France). All the strains were maintained as frozen (−80°C) stock cultures in Hershey (E. coli) or Elliker (Strep. thermophilus) broth supplemented with 15% (v/v) glycerol.
Escherichia coli DH5α was used to isolate the bacteriophages from 50 stool samples of patients with diarrhoea treated at the Centenary Hospital, Rosario. A portion of faeces (5 g) was added to 10 ml of a DH5α culture (OD600 = 1·0) grown in Hershey broth and the culture was incubated at 37°C for 12 h. Next, 0·5 ml of chloroform (Cicarelli) was added and the preparation was mixed and centrifuged at 4000 g for 10 min. The supernatant was sterilized by filtration through a 0·45-μm pore filter (Gamafil S.A., Buenos. Aires, Argentina) (Kennedy and Bitton 1987). Bacteriophage isolations were performed by the double-layer plaque technique; briefly, aliquots of 100 μl of phage stocks were mixed with 100 μl of recipient strain culture (OD600 = 1·0), then added with three ml of Hershey-Mg soft agar (Hershey-Mg with 0·7% agar, w/v) at 45°C. The mixture was poured into plates with Hershey-Mg agar (1·4%, w/v) and incubated overnight at 37°C (Jamalludeen et al. 2007). To isolate and purify phages, well-defined single plaques on the soft agar were picked and placed in 5 ml of Hershey-Mg broth (phage stock). High-titre phage suspensions were prepared as follows: Hershey-Mg broth was inoculated (1%, v/v) with an overnight culture of the recipient strain and aliquots of 100 μl of phage stocks were added. Incubation at 37°C with discontinuous shaking was performed until complete lysis. At that point, 0·1 ml of chloroform was added and cultures centrifuged at 4000 g for 10 min. Phage stocks were stored at 4°C and enumerated (plaque formation units per millilitre; PFU ml−1) by the double-layer plaque technique (Jamalludeen et al. 2007).
PCR amplification of virulence factors
Phages were tested for the presence of toxin-encoding genes (stx1, Shiga toxin 1; stx2, Shiga toxin 2; eaeA, attaching-and effacing; LT1, thermolabile toxin and ST1, thermostable toxin) of diarrheaogenic E. coli by the polymerase chain reaction (PCR) using primers detailed in Table 2. PCR conditions were as follows: initial denaturing step at 95°C for 2 min, followed by 25 cycles of 95°C for 30 s, annealing at 63°C for 30 s and elongation at 72°C for 30 s and a final elongation step at 72°C for 5 min. Escherichia coli ATCC43889 (stx2 and eaeA), ATCC43890 (stx1) and ATCC43895 (stx1, stx2 and eaeA, and also harbouring the stx2 phage, 933W) were used as positive controls, while enterotoxigenic E. coli ATCC35401 was used for LT1 and ST1 genes. Escherichia coli HB101 and ATCC98222 were utilized as negative controls. Amplified products were resolved by electrophoresis using 3% agarose gels in TBE buffer (89 mmol l−1 Tris borate, 2 mmol l−1 EDTA, pH 8·0) (Promega, Madison, WI, USA) at 100 V for 3 h. Gels were stained with ethidium bromide (0·5 μg/ml) (Sigma, St. Louis, MO, USA) and PCR products were visualized under UV light.
|Gene||Primera (Pass et al. 2000)||Product size (bp) expected|
fp: 5′- ACGTTACAGCGTGTTGCRGGGATC-3′
fp: 5′- TGTGGCTGGGTTCGTTTATACGGC-3′
bp: 5′- TCCGTTGTCATGGAAACCGTTGTC-3′
fp: 5′- TGAGCGGCTGGCATGATGCATAC-3′
bp: 5′- TCGATCCCCATCGTCACCAGAGG-3′
Phage electron micrographs were obtained by the procedure of Bolondi et al. (1995) using a JEOL 1200 EX II electron microscope (INTA Castelar, Buenos Aires, Argentina) operating at 85 kV. Phage morphologies and dimensions (head diameter, tail length and diameter) were recorded.
Host range of bacteriophages
The host range of each phage was determined by the double-layer agar technique using 70 strains isolated by streaking both stool samples (uncharacterized diarrheaogenic E. coli; UDEC) and urine cultures (uropathogenic E. coli; UPEC) in cystine lactose electrolyte–deficient agar plates. After incubation for 24 h at 35°C, lactose positive and citrate negative colonies were further identified using API-20E system. Sixteen E. coli strains from food (Balagué et al. 2006), one uropathogenic E. coli strain (E. coli T149) which expresses fimbriae P and α-haemolysin (Balagué et al. 2004) and five ATCC E. coli strains were also tested (ATCC 43890; 43889; 43895; 35401 and 98222). Previously characterized (API-20E system) isolates from stool samples were also tested: Shigella flexneri, Shigella sonnei, Proteus mirabilis, Citrobacter freundii, Klebsiella pneumoniae, Salmonella enteritidis, Salmonella Typhi and Salmonella Typhimurium. Strains tested against stock phages are listed in Table 3. Bacteriophage sensitivity was assayed by placing 10 μl of phage suspension on the solidified soft-agar layer inoculated with 100 μl of each bacterial culture, incubated for 24 h at 37°C, and the presence of lysis zones or plaques was examined (Goodridge et al. 2003).
|Source||Strains (No.)||Strain features/description|
|Food||Escherichia coli (16)||Eight non-O157 STEC (belonging to O8, O18, O44, O57 and O79 serogroups; Balagué et al. 2006); two O157:H7 STEC and six UDEC|
|Escherichia coli (35)||Four O157:H7 STEC; five EPEC and 26 UDEC|
|Escherichia coli (18)||Nonpathogenic (non-O157)|
|Stool samples|| |
Shigella spp. (5)
Salmonella spp. (5)
Proteus mirabilis (4)
Citrobacter freundii (1)
Klebsiella pneumoniae (2)
|Urine culture||Escherichia coli (18)||UPEC (7 fimH+, csgA+, papC+; 5 fimH+, csgA+; 6 fimH+; all bfp-) All papC+ (fimbria type P) and one papC- (fimbria type 1) were mannose resistant|
|ATCC||Escherichia coli (5)||35401; 43889; 43890; 43895 and 98222|
Biocontrol of Escherichia coli
During milk fermentation Biocontrol experiments were carried out at 37°C in parallel batches in sterile, commercial, reconstituted (10%, w/v) dry skim milk, added of CaCl2 (0·28 g l−1). All the batches were inoculated with overnight cultures of Streptococcus thermophilus 10-C at 1% (v/v), either alone (control assay) or together with overnight cultures of the tested E. coli strains (final concentration approx. 5 × 102–5 × 103 CFU ml−1; experimental assays). One aliquot of each E. coli-infected batch was inoculated with the corresponding phage or phage cocktail (107–108 PFU ml−1), so as to evaluate their potential as biocontrol agents, getting a MOI ranging from 3·5 × 103 to 2·5 × 105. The acidification proceeded during 24 h at 37°C in a thermostatic bath. During fermentation, changes in pH were assessed with a model SA 720 pH metre (Orion, Beverly, MA, USA), bacterial cell counts were performed in Elliker agar (37°C, 48 h) for streptococci or Hershey agar (37°C, 18 h) for E. coli, and phage enumeration (if applicable) was also carried out by the double-layer plaque titration method described above.
This research was supported by the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET; Project PIP 1206) and the Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT, Project PICT 2010, 0138) of Argentina.