Development of a bacteriophage cocktail with high specificity against high-risk avian pathogenic Escherichia coli

Abstract

Avian pathogenic Escherichia coli (APEC) is a widespread pathogen that poses a significant threat to the poultry industry globally. A recent typing scheme has proposed several APEC pathotypes, including non-APEC, High Risk non-APEC (HR non-APEC), APEC, and High Risk APEC (HR APEC), based on the presence of the ompT and hlyF genes, sequence types (ST) ST131, ST23, ST428, ST355, and the O78 antigen. In Brazilian broiler farms, a higher prevalence of lesions in carcasses has been linked to the presence of HR APEC pathotypes. Due to the growing concern over antimicrobial resistance, bacteriophage-based formulations are emerging as a promising alternative for controlling APEC outbreaks. In this study, we isolated and sequenced 142 strains from Brazilian commercial broiler farms, classifying them as APEC (70), HR APEC (12), HR non-APEC (3) and non-APEC (57). Notably, 38 % of the isolates were classified as multi-drug resistant (MDR), with serotypes H10 and ST155 being the most frequently identified. Additionally, we isolated, sequenced, and classified 66 bacteriophages that exhibited lytic activity against these bacterial strains in both qualitative and quantitative assays. Eight of the bacteriophages demonstrated complementary host ranges against the bacterial collection. Bacteriophage cocktails were assembled, tested in quantitative assays, and shown to be effective against APEC. A cocktail consisting of four bacteriophages (AC-01) displayed a broad lytic spectrum in vitro, inhibiting the growth of 56.3 % (n = 80/142) of the isolates with a mean inhibition of 32.9 %. Remarkably, the in vitro lytic activity of the cocktail was significantly more effective against HR APEC isolates (12/12, 65.9 % mean inhibition) and HR non-APEC isolates (3/3, 58.0 % mean inhibition). Our results emphasize the importance of genetically characterizing target bacteria when developing an effective and specific bacteriophage cocktail against APEC.

Introduction

Avian pathogenic Escherichia coli (APEC) is an extraintestinal strain of E. coli responsible for various local and systemic infections in poultry, including chickens, turkeys and many other avian species (Nolan et al., 2020). APEC has been identified as the primary etiologic agent of avian colibacillosis, a disease associated with high mortality rates (1–10 %). The infection is especially lethal for chicks, with mortality rates reaching up to 53.5 % (Mellata, 2013). Avian colibacillosis has remained one of the most significant bacterial diseases in poultry production for over 50 years (Johnson et al., 2022), leading to decreased live weight, reduced feed conversion efficiency, lower egg production, and increased carcass condemnations at slaughter. Brazil, the world's second largest producer and exporter of chicken meat, produces over 14 million tons annually (ABPA, 2024). Between 2013 and 2017, colibacillosis accounted for 9.1 % of total condemned carcasses in Brazil, totaling 8.7 million carcasses and resulting in losses exceeding USD$16 million (Souza et al., 2019).

Key virulence factors of APEC strains, including adhesins, iron acquisition systems, temperature-sensitive hemagglutinin, and capsules, have been thoroughly described (Dho-Moulin and Fairbrother, 1999). It is well established that not all E. coli isolated from colibacillosis cases are APEC, and their identification can only be confirmed through molecular characterization (Collingwood et al., 2014). Furthermore, the presence of APEC does not necessarily lead to the development of colibacillosis. Poultry face numerous stressors, which are often sufficient to enable any E. coli, whether virulent or not, to cause disease in the bird (Johnson et al., 2008). Therefore, identifying highly virulent APEC strains is critical for developing effective strategies against colibacillosis. While some studies have focused on clinically relevant isolates causing colibacillosis (Janßen et al., 2001; Ewers et al., 2004; Yaguchi et al., 2007), others have compared isolates from diseased and healthy birds (Vandekerchove et al., 2005; McPeake et al., 2005; Kawano et al., 2006). The presence of two plasmids containing pathogenicity associated island (PAI) genes has been proposed to distinguish APEC strains from commensal E. coli in birds (Rodriguez-Siek et al., 2005; Johnson et al., 2006a; 2006b; Tivendale et al., 2009). Based on in vivo challenges and the screening of isolates, Johnson et al. (2008) refined the PAI gene set to a subset of five defining APEC genes: iss, iroN, hlyF, ompT and iutA. However, more recent studies have challenged these approaches, as APEC plasmids are highly prevalent in isolates from asymptomatic birds (Mageiros et al., 2021). Other authors have proposed that multiple distinct clonal backgrounds among clinical poultry E. coli should be considered when defining APEC (Mehat et al., 2021). Recently, Johnson et al. (2022) refined the definition of APEC by including high-risk clonal groups, in a method known as APECtyper. They identified differences in the clonal backgrounds of turkey and broiler clinical versus cecal strains and determined the sequence types (ST) dominating both clinical settings (ST23, O78, ST117 + O78, ST131, ST355 and ST428). Based on the presence of the ompT and hlyF genes, they defined APEC, while the STs distinguished four pathotypes: APEC, High risk (HR) APEC, non-APEC and HR non-APEC.

Traditional intervention strategies for preventing and controlling avian colibacillosis include the management of environmental factors, vaccination and antibiotic use (Nolan et al., 2020). The main classes of antibiotics used in major poultry-producing countries (US, Brazil, China, Poland, UK, Germany, France and Spain) include aminoglycosides, ß-lactams (penicillins, first and third generation cephalosporins), phenicols, fluoroquinolones, ionophores, lincosamides, macrolides, polypeptides, polymyxins, sulfonamides and tetracyclines. The type and extent of antibiotics use vary by country, influenced by factors such as the country's economy, development level, and animal husbandry practices (Roth et al., 2019). In Europe and the US, the use of antibiotics growth promoters is banned, unlike in countries such as Brazil and China (Official Journal of the European Union, 2003; Access Science Editors, 2017). Moreover, the rising global antibiotic resistance crisis poses risks to both animal and human health (One Health Joint Plan of Action, 2022–2026, 2022). The use of antibiotics in poultry production increases the selective pressure for antibiotic-resistant bacteria (Diarra and Malouin, 2014). Commensal E. coli can act as reservoirs for antibiotic resistance genes in the chicken gut, potentially becoming pathogenic under stress (Ahmed et al., 2023). Studies worldwide have reported high resistance of E. coli from poultry to aminoglycosides, penicillins, quinolones, sulfonamides and tetracyclines (Roth et al., 2019). These resistant strains can be transferred to humans, which is concerning. To address this, the WHO collects epidemiological data and establishes guidelines to guide research. Recently, the WHO listed third-generation cephalosporin and carbapenem-resistant E. coli as critical priority pathogens (World Health Organization, 2024), and identified quinolones, third, fourth and fifth generation cephalosporins, macrolides, ketolides, glycopeptides, polypeptides and polymyxins as critically important antibiotics for human medicine (World Health Organization, 2019). In this scenario, the poultry industry is looking for innovative feeding and nonfeeding-based solutions to manage APEC in the chicken production chain (Christensen et al., 2021).

Bacteriophages have emerged as a promising alternative for controlling bacterial pathogens due to their safety, and high host specificity (Nolan et al., 2020). Some studies have reported the use of phages to control APEC both in vitro (Korf et al., 2020; Nicolas et al., 2023) and in vivo (Lau et al., 2010; Oliveira et al., 2010; Tawakol et al., 2019). Accurate bacterial genomic characterization and classification are critical for identifying true APEC and HR APEC pathotypes, and their potential susceptibility to phages. To the best of our knowledge, APECtyper classification has not been yet used to develop bacteriophage cocktails targeting APEC or HR APEC strains.

In this study, a cocktail composed of four bacteriophages (named FÓRMIDA) was developed to inhibit the growth of a collection of isolates (n = 142) obtained from Brazilian commercial broiler farms and classified by APECtyper. The cocktail demonstrated high in vitro efficacy against HR APEC. Our data supports the potential of the FÓRMIDA cocktail as a biocontrol agent in broiler farms, specifically targeting APEC, and HR APEC.

Materials and methods

Escherichia coli collection

Strains were obtained from shoe covers used on day 28 of rearing in aviaries at Brazilian commercial broiler farms, where chickens exhibited clinical symptoms of colibacillosis. PhageLab SpA requested the isolation of E. coli from a Brazilian diagnostic laboratory. Isolates were selected using MacConkey and blood agar (Koneman et al., 2000). The strain collection was screened for potential APEC isolates using Johnson's Pentaplex method (2008), which detects the presence of five virulence genes (iroN, iss, iutA, ompT, and hlyF). A total of 140 E. coli strains were identified, each containing at least one of the virulence genes. The strains were sourced from the states of Paraná (n = 135), Goiás (n = 4) and São Paulo (n = 1), with all isolates collected between July 2020 and February 2022 (Supplementary Table 1). In addition, two reference APEC strains (DSM 103263 and DSM 103266) were used in this study, obtained from the German Collection of Microorganisms and Cell Cultures (DSM, Braunschweig, Germany). All strains were stored frozen in Tryptic Soy Broth (TSB) with glycerol 5:1 v/v at −80 °C for further experiments.

Sequencing and bioinformatic analysis of E. coli collection

DNA was extracted from enriched cultures using the PureLink Genomic DNA Mini Kit (Thermofisher, Waltham, MA, USA) following manufacturer's instructions. Libraries were prepared using the Nextera XT DNA library preparation kit (Illumina, San Diego, CA, USA), and sequencing was carried out on a NextSeq 550 (Illumina). Genomic analyses were conducted using a custom workflow pipeline developed at PhageLab SpA. APEC was identified through classification with APECtyper (Jonhson et al., 2022), and the presence of antibiotic resistance genes was assessed using ResFinder (Bortolaia et al., 2020). Antimicrobials of interest were selected based on NARMS (National Antimicrobial Resistance Monitoring System for Enteric Bacteria) guidelines (CDC, 2024).

Isolation of bacteriophages

APEC strains, grouped into bacterial pools, were incubated with samples from various sources, including sewage water, cattle and pig rectal swabs, and feces, among others. The incubation was performed on TSB at 37 °C with agitation overnight to enrich viral particles (EVP). After incubation, EVPs were centrifuged at 5,000 x g for 10 min, and the supernatant was filtered through a 0.45 µm syringe filter. To purify the EVPs, 10 % of their volume of chloroform was added, followed by centrifugation at 5,000 x g for an additional 10 min. The aqueous phase was then carefully recovered as the final purified EVPs. These purified EVPs were assessed individually for bactericidal activity against bacterial strains using double layer agar assays in Tryptic Soy Agar (TSA), as described (Carlson, 2005; Kropinski et al., 2009). Any EVP displaying antimicrobial activity was used to isolate phages through 10-fold serial dilutions in SM buffer (dilution –1 to –8). To isolate the phages, plaque forming units (PFU) were collected from the EPV and isolated three consecutive times, selecting PFUs with the same morphology in the second and third rounds to ensure the isolation of the same bacteriophage. All isolations were performed with their respective host strains on double layer agar plates grown at 37 °C. Once the isolation process was complete, solid propagation was carried out on TSA plates to obtain a pure bacteriophage stock (Fortier and Moineau, 2009). Briefly, PFUs were propagated by mixing ∼10⁶ PFU with 100 µL of the host´s overnight culture and 6 mL of soft TSA (0.45 % agar). The mixture was poured onto a TSA plate and incubated overnight at 37 °C. To harvest the propagated bacteriophage, 5 mL of SM buffer was added to the TSA plate, and it was incubated at room temperature with gentle agitation (60-80 rpm) for 4 h. Finally, the bacteriophage was purified using the same procedure described to purify the EPV.

Sequencing of bacteriophages

Sequencing and bioinformatic analysis of the phage genomes were performed as previously described (Aguilera et al., 2023). Briefly, phage DNA was extracted from concentrated stocks (>1E+08 PFU/mL) using the Phage DNA isolation kit (Norgen, Ontario, Canada), following manufacturer's instructions. Library preparation and sequencing were carried out as outlined in the previous section. Sequence quality was assessed by analyzing raw FASTQ reads, with adapters trimmed, and low-quality reads discarded to ensure data integrity and reliability. High-quality reads were then used for de novo genome assembly, enabling the reconstruction of complete or near-complete genomes. Assembled genomes were evaluated for completeness and contamination levels. Finally, the resulting genomes were subjected to gene annotation, and identification of potential virulence factors and antibiotic resistance genes.

Double layer agar assays

Bacterial cultures were grown overnight at 37 °C in TSB with shaking. A 10-fold dilution was prepared by mixing 600 µL of the overnight culture with 6 mL of soft TSA, which was then poured onto TSA plates (1.5 % agar), and allowed to solidify at room temperature. PFU titration was performed by spotting 10-fold serial dilutions of bacteriophage onto the solidified plates (dilutions –1 to –8). The plates were incubated overnight at 37 °C. For spot assays, bacteriophage stocks were adjusted to 1E+08 PFU/mL, and 5 µL were spotted onto the solidified top-agar. The plates were then incubated overnight at 37°C. A qualitative lytic score was assigned based on the observed lysis zones, ranging from 0 (no plaque) to 4 (clear lysis) (Kutter, 2009).

Quantitative assays

A quantitative host range assay was performed in a 96-well plate as described by (Xie et al., 2018), with some modifications. Briefly, overnight cultures of bacteria were refreshed and grown to ∼1E+08 CFU/mL (OD₆₀₀∼0.1, as determined previously) in TSB at 37 °C with agitation, using an EPOCH2 microplate reader (Agilent Technologies, Santa Clara, CA, USA). The culture was then diluted to 1E+06 CFU/mL in TSB and mixed with bacteriophage at 2E+07 PFU/mL, achieving a final multiplicity of infection (MOI) input of 20. For cocktails, each bacteriophage was adjusted to the same final MOI input of 20. The microtiter plates were incubated in a BioTek LogPhase 600 multiple microplate reader (Agilent Technologies) at 37 °C with maximum agitation. OD₆₀₀ readings were taken every 10 min for a total of 18 h. Growth inhibition was calculated as the percentage reduction in the area under the curve of treated bacteria compared to that of untreated bacteria.

Physicochemical characterization of bacteriophages

Bacteriophage aliquots were adjusted to 1E+08 PFU/mL. To evaluate bacteriophage tolerance to pH, each bacteriophage was mixed with NaCl 0.9 % adjusted to pH values ranging from 2 to 10. The mixtures were incubated for 4 h at 37 °C and then neutralized by the addition of NaHCO₃ 0.07 M (pH 7). For temperature tolerance testing, each bacteriophage was mixed with NaCl 0.9 % at pH 7 and exposed to a range of temperatures from −80 to 70 °C for 4 h. The exposure was ended by the addition of NaCl 0.9 % pH 7 at 4 °C. After neutralization, antimicrobial activity was evaluated by double layer agar titration.

Transmission electron microscopy

A 20 μL sample of each purified bacteriophage (with titers ≥ 5E+09 PFU/mL) was applied to a carbon coated copper grid for 1 min, followed by negative staining with 2 % (m/v) uranyl acetate and drying at 60 °C for 20 min. Images were acquired at a magnification of 92000x in a Thermo Scientific Talos F200C G2 (Thermofisher) transmission electron microscope (TEM), equipped with a Ceta 16M CMOS camera and C-twin lens (Thermofisher). The images were analyzed and measurements taken with Fiji (Schindelin et al., 2012) and ImageJ softwares (Schneider et al., 2012). For each bacteriophage, the mean and standard deviation were calculated from measurements of n = 10 individual particles. The parameters assessed included the full length of the bacteriophage (from the top of the head to base plate), head length (from the top of the head to bottom of the head), and tail length (from the bottom of the head to base plate).

Statistical analysis

Data was analyzed using R, ensuring reproducibility by setting a fixed seed (set.seed(19931007)). Physicochemical experiments were performed in triplicate and analyzed using ANOVA with Dunnett post-test (reference group pH 7.0 or 5 °C). Bacteriophages were tested in uniplicate, while cocktail efficiency assessments were performed in triplicate. To compare the efficiency between cocktails, a mixed-model (lme4 package) was used, incorporating variable intercepts and slopes for each bacterium. Multiple comparisons were performed using post-hoc Tukey HSD.

Results

Classification and characterization of isolates

To characterize the strain collection, 142 E. coli isolates were sequenced and classified using APECtyper (Johnson et al., 2022). Based on this analysis, 12 isolates were identified as HR APEC, 3 as HR non-APEC, 70 as APEC and 57 as non-APEC. Additionally, 72 serotypes and 55 sequence types (ST) were identified. The most frequent serotype was H10 (n = 23/142), while clinically relevant serogroups O1 (n = 1/142), O2 (n = 1/142) and O78 (n = 9/142) were also identified. Sequence type ST155 was the most common (n = 15/142), followed by ST10 (n = 12/142) and ST752 (n = 12/142) (Fig. 1A, Supplementary Table 1). Further, to characterize the antibiotic resistance profile of the isolates, the presence of antibiotic resistance associated genes was evaluated. Overall, 90.8 % (n = 129/142) of isolates carried antibiotic resistance genes for at least one antibiotic class. The most frequently detected resistance genes were against aminoglycosides (n = 82/142), beta-lactams (n = 75/142), fluoroquinolones (n = 73/142), sulfonamides and tetracyclines (n = 65/142) (Fig. 1B, Supplementary Table 2). APEC strains (including HR APEC) presented a significantly higher prevalence of antibiotic resistance genes compared to non-APEC (including HR non-APEC) (χ² = 6.88, df = 1, p-value = 0.0087) (Fig. 1B). Furthermore, 62.6 % of isolates displayed resistance genes against at least 3 antibiotic classes and were classified as multidrug resistant (MDR) (Magiorakos et al., 2012). Among APEC and HR APEC isolates, 38.0 % were classified as MDR.

Fig. 1.

Genomic characterization of avian pathogenic Escherichia coli (APEC) collection. (A) Genomic distance-based tree of isolates. Distance was calculated as the average genomic divergence between each isolate. Colored circles represent key isolate features: from inner to outer, APECtyper classification (HR: High-Risk), presence of O78 (O78+) antigen and Sequence Type used for APECtyper classification. (B) Bioinformatic prediction of antibiotic resistance across different classes. The percentage of isolates with resistance genes (horizontal axis) is plotted against each antibiotic class (vertical axis), grouped by APECtyper classification. Each percentage is calculated based on the total number of isolates within each APECtyper classification, represented by different colors.

Isolation of lytic bacteriophages

To isolate bacteriophages targeting APEC, PFUs were collected from diverse sewage and environmental water samples. These PFUs were propagated, titrated, sequenced and analyzed using bioinformatics to determine their lifestyles. A total of 66 bacteriophages were confirmed to have a lytic lifestyle, with no detected antibiotic resistance genes, virulence genes or recombinases. To evaluate their lytic activity against E. coli, each bacteriophage was tested individually against 95 selected isolates using a double layer spot agar assay, generating a matrix of 6,270 interactions (Fig. 2). The tested isolates included 70 APEC, 2 HR APEC and 23 Non-APEC strains. Overall, individual bacteriophages showed narrow host ranges, with the most effective phages lysing up to 32/95 isolates, while many targeted less than 10 isolates. This discrete host range is likely due to the high genetic diversity of E. coli, which included 72 serotypes (Supplementary Table 1). Notably, some bacteriophages had complementary lytic activities against a large percentage of APEC and HR APEC isolates. Based on their host range, eight bacteriophages, AA2, APEC157P04, DEV-01, BK1, DEV-02, ETE131P06, FR5 and DEV-03 were selected for further experiments.

Fig. 2.

Heatmap of activity of 66 bacteriophages against 95 E. coli strains based on double-layer agar spot test. Lytic score values are shown as a gradient in green (4: complete clearing of the bacterial lawn, 3: complete clearing with faintly hazy background, 2: uncountable (>60) individual plaque-forming units, 1: countable plaque forming units, 0: No clearing of bacterial lawn nor plaque-forming units). Bacteria (vertical axis) and bacteriophages (horizontal axis) were clustered based on the average distance of the score matrix. The top and the left dendrograms show the relationship based on lytic score between phages and bacteria, respectively. The APECtyper classification of each strain is represented in the right bar (HR: High-Risk). Phages of interest are indicated by their name in the horizontal axis.

Quantitative activity of bacteriophages

To further characterize bacteriophage lytic activity, 142 isolates were tested using quantitative liquid co-culture assays, a reliable in vitro method for analyzing phage virulence (Henry et al., 2012; Xie et al., 2018; Martinez-Soto et al., 2021). The eight previously selected bacteriophages were individually co-cultured with each isolate in 96 well plates at an MOI input of 20, and bacterial growth was monitored for 18 h. A significant bacterial growth inhibition (>15 %) was observed in cases where double-layer agar also showed activity (χ² = 321.76, df = 1, p-value < 2.2E-16). Most isolates were inhibited by at least one of the eight bacteriophages (Fig. 3). Remarkably, bacteriophages APEC157P04 and AA2 displayed high activity against HR APEC. These findings highlight the lytic efficacy of individual bacteriophages AA2, APEC157P04, DEV-01, BK1, DEV-02, ETE131P06, FR5 and DEV-03 and support their potential combined use as a cocktail against APEC isolates.

Fig. 3.

Heatmap of activity of 8 bacteriophages against 121 E. coli strains based on co-culture assay. Growth inhibition values are shown as a gradient in green. Bacteria (vertical axis) and bacteriophages (horizontal axis) were clustered based on the average distance of the growth inhibition (unweighted pair group method with arithmetic mean, UPGMA). The left tree shows the relationship based on growth inhibition between bacteria. The APECtyper classification of each strain is represented in the right bar (HR: High-Risk).

Cocktail assembly

To assess the efficacy of bacteriophage cocktails against APEC, phages AA2, APEC157P04, DEV-01, BK1, DEV-02, ETE131P06, FR5 and DEV-03 were combined into four cocktails, based on host range results (Table 1). Liquid co-culture assays were used to quantify cocktail efficacies. Each bacterial strain (n = 142) was grown in the presence of a cocktail at an MOI input of 20 per bacteriophage, and bacterial growth was recorded until 18 h post infection. All four cocktails inhibited (≥15 %) the growth of at least 80 isolates, with mean inhibitions of 32.9, 32.5, 31.0 and 34.3 % for cocktails AC-01, AC-02, AC-03, and AC-04, respectively, with no significant differences among them (χ² = 3.49, df = 3, p-value = 0.322) (Fig. 4). To evaluate the specificity of cocktails against APEC, growth inhibition results were grouped by APECtyper classification (Table 2). Against APEC isolates, cocktail AC-04 was the most effective (n = 49/70), whereas AC-03 was the least effective (n = 41/70) and exhibited the lowest inhibition rate (mean 27.7 %). Against HR APEC, all 4 cocktails were 100 % effective (n = 12/12), with mean inhibition rates of 65.9, 62.8, 64.7 and 54.8 % for cocktails AC-01, AC-02, AC-03, and AC-04, respectively. Against HR Non-APEC, only AC-01 was 100 % effective (n = 3/3), with a mean inhibition of 58.0 %. Against Non-APEC isolates, cocktail AC-04 was the most effective (n = 33/60) and AC-01 the least (n = 25/60) with mean inhibitions of 30.7 and 25.9 %, respectively. The effectiveness and inhibition means of each cocktail against the entire collection or by its APECtyper classification is shown in Table 2. Based on these results, AC-01 was selected for further characterization because of its high specificity against APEC, HR APEC and HR Non-APEC.

Table 1.

Composition of cocktails (AC-01 to AC-04). Light blue squares represent the presence of each phage (bold) within the corresponding cocktail.

Fig. 4.

Cocktail performance by APECtyper classification based on liquid co-culture assay. Bacterial growth was monitored for 18 h with OD₆₀₀ readings every 10 min. Stacked bars represent the number of isolates on which the cocktails inhibited the growth of each strain (Lysis (L, blue), ≥ 15 % of growth inhibition) or had a null effect (Null (N, gray), <15 % of growth inhibition). APECtyper classification is represented on each panel (HR: High-Risk). Cocktails evaluated are shown in the x-axis (AC-01 to AC-04). Number of isolates is shown in the left y-axis. Mean growth inhibitions and standard error for each cocktail against each APECtyper classification are shown in red (right y-axis, n = 3 for each isolate).

Table 2.

Effectivity and mean growth inhibition exhibited by each cocktail against the entire isolate collection or by APECtyper classification.

	Effectivity - Growth Inhibition1
Cocktail	Collection	APEC	HR APEC	HR Non-APEC		Non-APEC
AC-01	80/142 - 32.9%	43/70 - 33.3%	12/12 - 65.9%	3/3 - 58.0%	22/57 - 25.9d
AC-02	86/142 - 32.5%	46/70 - 34.0%a	12/12 - 62.8%	2/3 - 41.9%c	26/57 - 24.7%
AC-03	84/142 - 31.0%	41/70 - 27.7%a,b	12/12 - 63.4%	2/3 - 41.9%	29/57 - 28.4%
AC-04	94/142 - 34.4%	49/70 - 34.3%b	12/12 - 53.9%	2/3 - 41.9%c	31/57 - 30.7%d

¹Superscript letters indicate statistically significant differences between growth inhibition with the same letters, determined by Post-hoc Tukey HSD.

^a: p-value = 0.047;

^b: p-value = 0.049;

^c: p-value < 0.001;

^d: p-value = 0.022. APEC: Avian pathogenic Escherichia coli. HR: High-Risk.

Characterization of AC-01 bacteriophages

To characterize the morphology of the phages in cocktail AC-01, TEM was performed on AA2, APEC157P04, BK1 and ETE131P06 (Fig. 5A, from left to right, respectively). The full length size of AA2 is 219.12 ± 6.04 nm, phage head length of 105.81 ± 2.96 nm, and tail length of 113.31 ± 5.96 nm; APEC157P04 has a full length size of 188.96 ± 4.75 nm, phage head length of 83.74 ± 1.96 nm, and tail length of 105.21 ± 4.61 nm; BK1 has a full length size of 220.08 ± 3.86 nm, phage head length of 108.95 ± 4.90 nm, and tail length of 111.13 ± 3.01 nm; the full length size of ETE131P06 was 216.67 ± 3.57 nm, phage head length of 106.45 ± 3.53 nm, and tail length of 110.22 ± 4.84 nm. For a more detailed genomic characterization, a pipeline developed in PhageLab was used (Table 3). Genome sizes of AC-01 phages ranged between 152.5 and 169.9 kbp. Phages AA2, BK1, and ETE131P06 belong to the Straboviridae subfamily and phage APEC157P04 belongs to the Chaseviridae subfamily.

Fig. 5.

Physicochemical characterization of FÓRMIDA bacteriophages. Left to right bacteriophages AA2, APEC157P04, BK1 and ETE131P06. (A) Transmission electron microscopy (TEM) of bacteriophages. (B) pH stability (horizontal axis indicate pH range from 2 to 10, vertical axis shows the concentration of bacteriophages in plaque-forming units per milliliter (PFU/mL). Reference group was pH 7. (C) Temperature stability (horizontal axis indicates temperature range from −80 to 70 °C, vertical axis shows the concentration of bacteriophages in plaque-forming units per milliliter (PFU/mL). Reference group was 5 °C. ANOVA with Dunnett test was performed to compare against the reference group. Bars represent mean with standard deviation as error bars. Experiments were performed in triplicates, with each replicate represented as a black dot. Significant statistical differences (p-value < 0.05) against the reference group are shown by a horizontal line indicating the p-value.

Table 3.

Origin and genomic characteristics of AC-01 bacteriophages.

Bacteriophage	AA2	APEC157P04	BK1	ETE131P06
Origin	Cattle rectal swab	Chicken faeces	Pig stomach content	Pig faeces
Genome size (bp)	169,147	152,572	166,545	169,917
GC content	35.31	39.01	40.41	40.53
Number of ORFs	289	311	296	286
Number of tRNAs	10	12	0	0
Taxonomy	Straboviridae	Chaseviridae	Straboviridae	Straboviridae
Closest NCBI ID Accession no.	Escherichia phage vB_EcoM_C2-3 OK076929.1	Escherichia phage ukendt MN850565.1	Escherichia phage kaaroe MN850574.1	Shigella phage Sf20 MF327006.1
Identity (%)	97.06	98.65	92.89	97.03
Antibiotic resistance	Absent	Absent	Absent	Absent
Integrases	Absent	Absent	Absent	Absent
Lifestyle	Lytic	Lytic	Lytic	Lytic

To characterize the stability of bacteriophages under different pH and temperature conditions, they were individually exposed for 4 h (see methods). All bacteriophages remained stable within a pH range of 4–10, with no significant changes in their titers. However, exposure to pH 3 led to a significant decrease of APEC157P04 and AA2 phage titers, while BK1 and ETE131P06 showed similar reductions at pH 2 (Fig. 5B). Regarding temperature stability, all bacteriophages retained infectivity in the range between 5 °C and 50 °C. Temperatures below −20 °C or above 50 °C reduced the titers of all 4 bacteriophages (Fig. 5C). Therefore, phages of cocktail AC-01 were found to be stable within a pH range 4–10 and temperatures between 5 and 50 °C.

Discussion

Our research highlights the importance of comprehensive bacterial genomic characterization in developing a lytic phage cocktail to improve control strategies against Avian Pathogenic E. coli. To our knowledge, this is the first study using APECtyper to design a phage cocktail, highlighting the genetic diversity of isolates from farm environments. Our results clearly demonstrate that not all E. coli isolates obtained from farms where colibacillosis was detected can be classified as APEC. Further, we found 12 isolates that were classified as HR APEC and 3 as HR Non-APEC, which we considered critical targets when assembling the phage cocktails.

Since APECtyper relies on genomic data for classification, we examined serotype and ST distributions. Our bacterial collection was highly diverse, comprising 72 serotypes, with H10 as the most prevalent, a finding that contrasts with earlier reports on pathogenic E. coli (Wolf, 1997). The clinically significant serogroups O1, O2, and O78 were less frequent than previously reported (Nolan et al., 2020), and many isolates had unidentifiable O serogroups. Additionally, 55 different STs were identified, with ST10 and ST752 being the most common. Previously, ST10, ST88, ST93, ST117, ST131, and ST155 were detected in avian isolates from Brazil (Galarce et al., 2023), all of which were also found in our collection. These findings underscore the need to account for bacterial genomic variability and pathogenic profiles when developing phage-based products.

Antibiotic resistance remains a major concern, with a substantial proportion of our isolates exhibiting multidrug resistance. APEC strains are known to harbor more resistance genes than commensal strains and frequently exhibit resistance to tetracyclines, sulfonamides, ampicillin or streptomycin, and MDR (Nolan et al., 2020). Other studies have reported high resistance levels to quinolones, beta-lactams and colistin (Galarce et al., 2023). Recent data from Brazil (2018-2023) revealed MDR prevalence ranging from 60 % to 88.8 % (Pilati et al., 2023, 2024). Similarly, our genomic analysis predicted that 62.6 % of the isolates were MDR, reinforcing the urgency of alternative solutions. The more frequently identified resistance genes targeted aminoglycosides (n = 82/142), beta-lactams (n = 75/142), fluoroquinolones (n = 73/142), sulfonamides and tetracyclines (n = 65/142), consistent with previous reports. APEC strains (APEC and HR APEC) identified by APECtyper, presented significantly higher antibiotic resistance frequencies than commensal strains (Non-APEC), supporting their clinical relevance. Notably, fluoroquinolone resistance genes were present in 51.4 % of isolates. Given that fluoroquinolones are classified as “highest priority critically important” antibiotics for human medicine by the WHO (World Health Organization, 2019), the high prevalence of resistance genes underscores the need for alternative interventions, such as bacteriophage-based therapies.

Bacteriophages were isolated using a mix of APEC and Non-APEC strains. Among the 66 bacteriophages isolated, only a few showed broad activity against APEC isolates. The overall discrete host range of the phage collection can be partially explained by the high genetic diversity of E. coli (Supplementary Table 1), a pattern previously observed in APEC-specific phages (Kazibwe et al., 2020; Nicolas et al., 2023). Despite this, eight phages demonstrated complementary host ranges, a crucial criterion for cocktail design. These phages were selected for further evaluation in cocktail assembly.

The four cocktails assayed displayed similar efficiencies against all E. coli isolates, despite variations in the number of phages used (between 4 and 6 per cocktail). These findings are relevant since minimization of interference between phages of a cocktail could potentially help to maximize its efficacy in field applications. Notably, cocktail AC-01 showed the higher efficiencies and growth inhibitions against HR APEC and HR Non-APEC, while maintaining the lowest efficiency against Non-APEC isolates. This specificity is a critical feature, as bacteriophage cocktails should minimize disruption to commensal bacteria, which contribute to the overall animal health status (Brisbin et al., 2008; Cieplak et al., 2018; Clavijo and Flórez, 2018). Given the high specificity displayed by AC-01 against APEC and HR APEC isolates, it was selected for further characterization. The phages in AC-01 remained stable across a pH range between 4 and 10, and temperatures between 5 °C and 50 °C, with morphological and genomic properties confirming their novelty. Even though minimal phage cocktails with maximal activity and specificity are crucial during development stages, the stability of phages is equally important for large-scale production and field applications.

Our findings have profound implications for the development of bacteriophage therapies against APEC. By specifically targeting APEC and HR APEC pathotypes, our cocktail -commercially named FÓRMIDA- shows high specificity against these pathogenic strains of E. coli. This approach can be applied to other bacterial systems, where distinguishing pathogens from commensals is critical. To our knowledge, FÓRMIDA is the first bacteriophage cocktail developed using APECtyper for isolate classification. While our study successfully designed and tested a bacteriophage cocktail against APEC and HR APEC in vitro, further studies are required to assess its safety and efficacy in vivo, as well as its potential interactions with commonly used antibiotics in farms. These findings will offer valuable insights to improve APEC control strategies in broiler production.

CRediT authorship contribution statement

Rodrigo Norambuena: Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft. Victoria Rojas-Martínez: Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft. Eduardo Tobar-Calfucoy: Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft, Software, Data curation, Visualization. Matías Aguilera: Supervision, Project administration, Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft. Andrea Sabag: Methodology, Investigation. María Sofía Zamudio: Methodology, Investigation. Pabla Lara: Methodology, Investigation. Daniel San Martín: Methodology, Investigation. Marcela Zabner: Methodology, Investigation. Daniel Tichy: Software, Data curation. Pamela Camejo: Software, Data curation. Felipe Rojas: Software, Data curation. Luis León: Software, Data curation. Michael Pino: Investigation. Paola Mora: Investigation. Soledad Ulloa: Investigation. Pablo Cifuentes: Supervision, Project administration. Hans Pieringer: Funding acquisition. Nicolás Cifuentes Muñoz: Conceptualization, Supervision, Project administration, Writing – review & editing.

Declaration of competing interest

Rodrigo Norambuena, Victoria Rojas, Eduardo Tobar, Matias Aguilera, Andrea Sabag, Maria Sofia Zamudio, Pabla Lara, Daniel San Martin, Marcela Zabner, Pamela Camejo, Felipe Rojas, Michael Pino, Paola Mora, Soledad Ulloa, Pablo Cifuentes, Hans Pieringer reports a relationship with PhageLab Chile SpA that includes: employment. Nicolas Cifuentes Munoz reports a relationship with PhageLab Chile SpA that includes: equity or stocks. Pablo Cifuentes, Hans Pieringer reports a relationship with PhageLab Chile SpA that includes: equity or stocks. Matias Aguilera, Rodrigo Norambuena, Eduardo Tobar, Andrea Sabag, Victoria Rojas, Pablo Cifuentes, Hans Pieringer, Nicolas Cifuentes has patent pending to Assignee. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.