Each of the 253 UTI E. coli isolates was screened for static biofilm formation in 96-well untreated polystyrene microtiter plates following growth in artificial urine medium as described in the Experimental section. Isolates were considered positive for biofilm formation if the crystal violet-stained biofilm had an OD₆₀₀ equal to or greater than 3-fold the value obtained in the well containing bacteria-free medium. The UTI E. coli strain CTF073 was used as a positive control and 12 wells of this strain were included with each plate of samples of UPEC assayed for biofilm formation. The plotted results were the average of three independent experiments, each with six replicates per isolate. Only 42 out of the 253 E. coli isolates tested in this study produced biofilms in microtiter plate wells with the levels of biofilm formation varying among isolates (Figure 1). This is in contrast to a previous study [25] where a majority of the UTI strains tested produced biofilms in vitro. A possible reason for the different outcomes in the two studies is the fact that UPEC do not form biofilm well on polystyrene plates in urine or artificial urine medium as compared to M9 minimal medium used by [25]. Among the 42 biofilm-forming isolates, Can67 showed the highest level of biofilm formation (OD₆₀₀ above 1.0) whereas MSHS94 and Can72 resulted in weak biofilms formation with an OD₆₀₀ below 0.2.

The biofilm-forming UTI E. coli isolates totalized 22 different somatic O antigens and 14 different H antigens (Table 1). We then investigated their susceptibility profiles against 15 antimicrobials. The investigated isolates were frequently resistant with 32 (78%) isolates displaying resistance to antimicrobial agents, while 16 (39%) isolates displayed multidrug resistance phenotypes. These results are in agreement with previous studies that have shown a high frequency of antimicrobial resistance among UPEC [4,9,10,11,12,13]. However, inspection of the different levels of biofilm formation on microtiter plates (Figure 1) and comparing this to the results on antimicrobial susceptibility (Table 1) showed that there was no apparent direct correlation between antimicrobial resistance and the amount of biofilm formed by each isolate.

It is important to note that the ability of biofilms to withstand antibiotic killing is a function of their mode of growth. Antimicrobial agents have been shown to penetrate biofilms at different rates depending on the particular agent and the biofilm [26]. The antimicrobial susceptibility tests carried out in this study using the broth microdilution method did not test the susceptibility of the bacteria to the antimicrobial agent after biofilm formation. The antimicrobials were added to cell suspensions at time of media inoculation and not after the biofilms had formed. The results of [27] demonstrated that older biofilms of E. coli resisted ampicillin treatment to a greater extent than their younger counterparts. However, in our study, a majority of the biofilm forming E. coli UTI isolates (78%) were resistant to at least one antimicrobial agent. This compares to 53.1% of the non-biofilm forming E. coli UTI isolates that were resistant to at least one out of the 15 antimicrobials tested (results not shown). This is consistent with what is known about antimicrobial resistance among biofilm forming bacterial strains [28].

UPEC strains Can 91, Can 40 and MSHS1210 were used as propagating strains for the isolation of phages from serially diluted CsCl purified phage concentrate. All three isolates were from midstream urine specimens with Can 40 and Can 91 being part of the CANWARD study whereas MSHS1210 was from the McGill University student health service. The single unique phages isolated from UPEC strains Can 91, Can 40 and MSHS1210 were vB_EcoP_ACG-C91, vB_EcoM_ACG-C40 and vB_EcoS_ACG-M12, respectively.

Transmission electron microscopy revealed that phage ACG-C40 has an elongated head, a neck, and a contractile tail with tail fibers (Figure 2a). Its head is 110 × 82 nm, while the extended tail is 114 × 8 nm. Phage ACG-M12 (Figure 2b) has an isometric head of about 157 nm in diameter between opposite apices and a relatively flexible tail of 172 × 7 nm, which terminate in 1–2 fibers of 12 nm in length. Phage ACG-C91 (Figure 2c) has an isometric head of 65–68 nm and a short tail of 12 × 8 nm, which carries fibers of 13 nm in length with a terminal swelling. Based on their morphology, ACG-C40 is classified as a member of the family Myoviridae, ACG-M12 is part of the Siphoviridae and ACG-C91 belongs to the family Podoviridae.

The lytic activity of ACG-C91, ACG-C40 and ACG-M12 tested against the biofilm forming E. coli UTI isolates. In the spot tests, 33 (80.5%) of the biofilm forming E. coli UTI isolates were infected by at least one of the three phages (Table 1).

Out of the three phages isolated in this study, phage ACG-C40 had the broadest host range, with the ability to lyse 28 (66.67%) of the biofilm forming UTI E. coli isolates tested (Table 1). It is not unusual for phages in the family Myoviridae to have such a large host range. A previous study by [29], showed that bacteriophage T4 lysed 41.1% of the 69 clinical isolates of E. coli tested. Other T4-like phages such as AR1 have been observed to have similar broad host ranges [30]. The ability of phage ACG-C40 to cause lysis in a wide range of E. coli isolates with different serotypes makes it an important candidate for phage therapy applications. The broad host range of UTI E. coli isolates is an indicator of how well adapted phage ACG-C40 is to infecting UPEC, which in itself is a desirable property for a phage to be used in phage therapy against such strains. The criteria of choosing phages which are well adapted for infection of targeted hosts for therapeutic purposes have been applied in similar previous studies [31,32,33].

The correlation between host serotype and phage sensitivity is seen in the fact that among the biofilm forming strains, all the O25:H4 strains were resistant to the three tested phages. The O6:H1 strains were sensitive to phages ACG-C40 and ACG-M12. Further in depth studies to identify the receptors of the tested phages and the receptors spatial orientation on the surface of the host strains under tested conditions would give insight as to why the phages are resistant or susceptible to all strains of a given serotype.

Genome sequencing and bioinformatics analysis of the three isolated phages revealed that all three are non-temperate and that none carries any known bacterial virulence genes for humans or animals. None of the three phages possessed a demonstrable extracellular polysaccharide or exopolysaccharide (EPS) depolymerase gene. The importance of phage-associated depolymerases in biofilm eradication has been recognized in previous studies [22,23,24,34]. Depolymerase-producing phages were found to be more effective in eradicating mature biofilms than non-depolymerase producing phages [34]. Phages producing depolymerases have thus been used in concert with antimicrobials to facilitate deeper penetration of antimicrobials by degrading the EPS [23,34]. However, phages that do not produce EPS depolymerases have also found use in biofilm degradation. Such phages include naturally occurring phages such as T4 [16,17,35] and phages such as T7 engineered to express recombinant dispersin B (DsbB) [36].

Pairwise comparisons using CoreGenes 3.0 [39] at default stringency setting (“75”) revealed that at the protein level, ACG-C40, ACG-C91 and ACG_M12 were 85.9% ,78.8% and 77.3% similar to bacteriophages T4, SP6 and RTP, respectively. Bacteriophage ACG-C40 genome encodes 239 proteins homologous to T4 proteins, ACG-C91 genome has 41 encoded proteins homologous to those of SP6 and ACG-M12 genome encodes 58 proteins with homology to RTP proteins (supplementary files: Table A1; Table A2; Table A3).

A Mauve alignment [40,41] of phage ACG-C91 DNA against the genome of phage SP6 shows regions of sequence similarity (coloured red) with two regions where there are no DNA sequence similarities (coloured white). The regions of non-sequence similarity correspond to the regions encoding phage ACG-C91 internal virion protein proteins (orf 38), putative tail protein (orf 41) and the endosialidase (orf 53). The genome region encoding the endosialidase is lacking altogether in phage SP6 (supplimentary file: Table A1; supplementary file: Figure A2a). This protein is found in K1-specific phages, such as K1F and K1-5, that encode virion-associated endosialidases hydrolyzing the K1 polysialic acid structure of the K1 capsule-producing E.coli strains. Mauve alignment of ACG-M12 against the genome of phage RTP was also performed and it revealed regions of sequence similarity (red) interspersed with four regions where no sequence similarity exists (white). The regions of genome homology disparity include phage ACG-M12 genome regions encoding: hypothetical proteins orf2, orf3 and orf4 (1025 bp–1588 bp), the major tail protein orf26 (12435 bp–13091 bp), putative tail fibre orf47 (29255–32074) and conserved hypothetical proteins orf62 and orf63 (supplementary file: Figure A2 b).

UTI E. coli isolate Can 91 was selected to determine the effectiveness of the isolated phages to degrade established biofilms because of the isolate’s ability to form strong biofilms in microtiter wells (OD₆₀₀ 0.538) after 48 hrs and its sensitivity to all the three phages isolated. Phages ACG-C40, ACG-C91 and ACG-M12 treatment of preformed UTI E. coli Can 91 biofilms yielded reductions (based on OD₆₀₀ measurement of CV-stained cells) in biofilm mass compared to untreated controls (Figure 3). All of the three phages displayed, at low concentrations (10⁵ PFU/mL), a steady reduction of biofilm biomass over time with the largest percentage reduction of biofilm biomass being realized after 8 h incubation. The same pattern was also observed at higher phage concentrations (10⁷ PFU/mL and 10⁹ PFU/mL). At all phage concentrations, it was evident that the biofilm started re-establishing itself after 24 h (Figure 3).

These results indicate that under the conditions tested, the biofilm reduction by the phages is not dose-dependent, a fact that could be taken advantage of if the phages are to be applied in vivo to remove established biofilms. It would mean that low titers of the phages were as effective as using a higher titer in eradicating established biofilms. Re-establishment of biofilms after 24 h exposure to the phages in vitro may be attributed to development of bacterial resistance against the phages. This could be overcome by using a cocktail of phages. It may also be a good strategy to augment phage treatment with chemical antimicrobials which will be effective in prevention of the re-establishment of biofilms by the phage resistant cells.

Further studies need to be carried out to assess the efficacy of phage ACG-C40, ACG-C91 and ACG-M12 in the prevention of biofilm formation for comparison with this study where we have assayed their efficacy in the eradication of established biofilms.

A group of 123 strains of UTI E. coli were isolated from women with community-acquired UTI either from a student health service (McGill University, Montréal, QC, Canada) or at a community health center also in Montréal. To ensure no bias in the selection of isolates due to recurrent UTI or treatment failures, only one isolate per patient was used.

An additional 130 isolates were part of the ongoing Canadian national surveillance study (CANWARD) for testing urinary culture pathogens for antimicrobial susceptibilities. The study involved out-patients attending hospital clinics and emergency rooms, and in patients on medical and surgical wards, and in intensive care units [42]. Each study site was asked to submit only clinically significant isolates, as defined by local site criteria, from out- and in-patients with urinary infections. Fifty (2009) to 100 (2007, 2008) consecutive urinary tract isolates per year per medical center site (one isolate per patient) were collected. Primary isolate identification was performed by the submitting medical center site and confirmed by the coordinating laboratory, as required, based on morphological characteristics and spot tests [43]. If an isolate identification made by the coordinating laboratory did not match that provided by the submitting site, the isolate was removed from the study. All isolates were stored at −70 °C in skim milk.

At the Department of Epidemiology, Biostatistics and Occupational Health of McGill University, the isolates from the McGill University student health service and the community health center, also in Montréal, as well as those from the CANWARD study were recovered on MacConkey and CLED agar (Uricult dipslides: Orion Diagnostica, Oy, Finland). A single representative colony was picked from the MacConkey side of the Uricult slide and grown in LB broth and archived in 15% glycerol at −80 °C. When needed, further testing, based upon indole production and the presence of lysine or ornithine decarboxylase, was also performed to confirm the bacterial identification.

For all experiments, bacteria were grown in LB medium or synthetic urine medium. The latter medium was based on that devised by [44], and its composition and method of sterilization have been described previously [45]. Briefly the medium was composed of CaCl₂, 0.65 g/L; MgCl₂, 0.65 g/L; NaCl, 4.6 g/L; Na₂SO₄, 2.3 g/L; Na₃-citrate, 0.65 g/L; Na₂-oxalate, 0.02 g/L; KH₂PO₄, 2.8 g/L; KCl, 1.6 g/L; NH₄Cl, 1.0 g/L; urea, 25 g/L; creatinine, 1.1 g/L; Luria-Bertani (LB) broth (Difco, Sparks, MD, USA), 10 g/L. The pH of the medium is adjusted to pH 5.8 prior to sterilization by filtration through a 0.2-µm filter.

Identification of somatic (O) and flagellar (H) antigens was performed by standard agglutination methods [46]. The isolates were serotyped at the Public Health Agency of Canada, Laboratory for Foodborne Zoonoses, Guelph, ON, Canada.

The broth microdilution method was used (Sensititre System; Trek Diagnostics, Cleveland, OH), following the protocols of the Canadian Integrated Program for Antimicrobial Resistance Surveillance [47,48]. Each E. coli isolate was tested for the following antimicrobial agents (breakpoints are indicated in parentheses): amikacin (≥64 μg/mL), amoxicillin-clavulanic acid (≥32 and ≥16 µg/mL, respectively), ampicillin (≥32 µg/mL), cefoxitin (≥32 µg/mL), ceftiofur (≥8 μg/mL), ceftriaxone (≥2 µg/mL) (11), chloramphenicol (≥32 µg/mL), ciprofloxacin (≥4 µg/mL), gentamicin (≥16 μg/mL), kanamycin (≥64 μg/mL), naladixic acid (≥32 μg/mL), streptomycin (≥64 μg/mL), sulfisoxazole (≥512 μg/mL), tetracycline (≥16 μg/mL), and trimethoprim-sulfamethoxazole (≥4 and ≥76 μg/mL, respectively).

Three liters of preliminary treated sewage sample from the Guelph water treatment plant were centrifuged at 3,000 RPM for 20 min to eliminate solid debris. The clarified supernatant was concentrated to 400 mL using the tangential flow system as described in [49]. NaCl (20.5 g) was then added to the concentrate and stirred until it dissolved before 40 g of polyethylene glycol was added to precipitate the phage present in the concentrate. This was left overnight while being stirred gently with a magnetic stirrer at 4 °C. Precipitated phage was recovered by centrifugation at 3,000 RPM for 10 min and resuspended in 3 mL SM buffer [50]. This was subjected to two rounds of CsCl gradient purification steps and the resulting viral band isolated.

E. coli isolates Can 91, Can 40 and MSHS1210 were used as propagating strains for the isolation of single plaques from serially diluted CsCl purified phage concentrate. All three isolates were from midstream urine specimens with Can 40 and Can 91 being part of the CANWARD study whereas MSHS1210 was from the McGill University student health service.

LB agar (Difco) plugs from standard agar overlay method [51] were resuspended in 500 µL SM buffer, diluted and used for further single plaque isolation on the respective E. coli propagation strains. This was repeated three times for each strain to ensure unique phages for each UPEC isolate. The single unique phages isolated from E. coli isolates Can 91, Can 40 and MSHS1210 were named vB_EcoP_ACG-C91, vB_EcoM_ACG-C40 and vB_EcoS_ACG-M12, respectively, following the naming convention of [52].

Phages were pelleted at 25,000 × g for 1 hour, using a Beckman high-speed centrifuge and a JA-18.1 fixed-angle rotor (Beckman, Palo Alto, CA, USA). Phage pellets were washed twice under the same conditions in neutral 0.1 M ammonium acetate and resuspended. The phages were then deposited on copper grids with carbon-coated Formvar films and stained with 2% uranyl acetate (pH 4) or 2% phosphotungstate (pH 7.2). They were examined in a Philips EM electron microscope. Magnification was calibrated using T4 phage tails as size standards.

To separate phage from bacterial debris, crude phage lysates were centrifuged at 10, 000 × g for 15 min at 4 °C. Bacterial nucleic acids in the supernatants were digested with pancreatic DNase 1 and RNase A, each to a final concentration of 10 μg/mL (Sigma-Aldrich Canada Ltd., Oakville, ON, USA). Phage particles were then precipitated in the presence of 10% w/v (final concentration) polyethylene glycol (PEG-8000; Sigma-Aldrich) at 4 °C overnight. The precipitated phage particles were recovered by centrifugation, resuspended in TM buffer (10 mM Tris-HCl, pH 7.8, 1 mM MgSO₄). The DNA was extracted from a portion of the precipitated viral particles using the SDS/proteinase K method modified from [50], followed by extraction with phenol:chloroform:isoamyl alcohol (25:24:1, vol/vol), ethanol precipitation and resolution in 10 mM Tris-HCl (pH 7.5). The DNA was characterized spectrophotometrically.

Phage DNA was digested with restriction enzymes BamHI, BglI, EcoRV, HindIII, NaeI, XbaI, NdeI, PstI, and SmaI (New England Biolabs, Hertfordshire, UK), according to the manufacturer's instructions. Phage Lambda DNA (Fermentas Inc., Hanover, MD, USA) digested by the same enzymes was used as a positive control for the restriction digest analysis.

The DNA was subjected to pyrosequencing (454 technology) at the McGill University and Génome Québec Innovation Centre (Montréal, QC, Canada) to between 195 and 930 fold coverage.

The sequences were rearranged to resemble their GenBank homologs and annotated using MyRAST [53]. The generated gbk file was incorporated into Kodon (Applied Maths, Austin, TX, USA) and proof-read. Proteins were generated using gbk2faa program [54] and screened for homologs using BLAST algorithm [55]. In addition they were searched for conserved motifs in the Pfam database [56] and transmembrane helix predictions using TMHMM [57] and Phobius [58]. The sequence was also screened for tRNAs using tRNAscan-SE program [59]. The final sequences were converted to Sequin format [60] using gbk2sqn [61] before deposition into GenBank. The accession numbers are ACG-C91 JN986844, ACG-M12 JN986845 and ACG-C40 JN986846.

Each of the 253 E. coli isolates was screened for biofilm formation using the 96-well plate assay as described in [62], with minor modifications. Briefly, E. coli isolates were used to inoculate 5 mL of synthetic urine medium and grown for 16 h at 37 °C. A 1:100 dilution of each of the cultures was made in synthetic urine media and 200 µL of each diluted culture was added to six wells in an untreated 96-well polystyrene flat-bottomed, 96-well microtiter plates (Costar; Corning Inc., Corning, NY, USA). The plate was covered and incubated 37 °C for 48 h without agitation. Negative control wells that contained 200 µL of sterile synthetic urine media only were included. Following incubation, planktonic bacteria were removed by pipetting out the culture and washing the wells twice with PBS to remove loosely bound cells, using a multichannel pipettor. A 1% (wt/vol; 210 µL) crystal violet (CV) solution was added to the wells for 10 min. The CV stain was then decanted and the unbound stain removed by washing the plate in a water tray. The plates were left to dry following which 200 µL of 95% ethanol added in each well to solubilize the bound CV from the stained E. coli biofilms. The absorbance of CV at 600 nm was measured in a Wallac-Victor² 1420 Multilabel Counter (Perkin-Elmer, Boston, MA, USA).

The host range of phages ACG-C91, ACG-C40 and ACG-M12 against biofilm forming E. coli UTI isolates was determined by standard spot tests [63]. Briefly, this involved mixing 200 µL overnight culture of each isolate with 3 mL molten top LB agar (0.6% agar) and pouring on 96 mm diameter plates containing bottom agar. This was left to solidify then 10 μl of phage to be tested was spotted in the middle of the plate and left to dry before incubation inverted overnight at 37 °C.

E. coli isolate CLSC 94 was used to test the effect of the two environmentally isolated phages ACG-C40 and ACG-C91 on established biofilms. The method of [64] with minor modifications was used. Briefly, stationery-phase culture of E. coli CLSC94 was diluted 1:100 and 200 µL of the inoculum added to 12 wells of an untreated 96-well polystyrene microtiter plate. This was incubated for 24 h at 37 °C without agitation. Negative control wells that contained 200 µL of sterile synthetic urine media only were included. Following incubation period, planktonic bacteria were removed by pipetting out the culture, and washing the wells twice with PBS to remove loosely bound cells. The wells were then treated with 200 µL of ACG-C40 and ACG-C91 phage lysate at titers of 10⁵, 10⁷ and 10⁹ PFU/mL added to each of the 12 wells. Phages were also added to the negative control wells which contained only UPEC cells grown in sterile synthetic urine media. After incubation periods of 2 h, 4 h, 8 h and 24 h, planktonic bacteria were removed and wells were washed twice with PBS to remove loosely bound cells. The remaining biofilm was stained as described above.