In Vitro Demonstration of Targeted Phage Therapy and Competitive Exclusion as a Novel Strategy for Decolonization of Extended-Spectrum-Cephalosporin-Resistant Escherichia coli

ABSTRACT

Extended-spectrum cephalosporin-resistant (ESC-R) Escherichia coli have disseminated in food-producing animals globally, attributed to horizontal transmission of bla_CTX-M variants, as seen in the InCI1-bla_CTX-M-1 plasmid. This ease of transmission, coupled with its demonstrated long-term persistence, presents a significant One Health antimicrobial resistance (AMR) risk. Bacteriophage (phage) therapy is a potential strategy in eliminating ESC-R E. coli in food-producing animals; however, it is hindered by the development of phage-resistant bacteria and phage biosafety concerns. Another alternative to antimicrobials is probiotics, with this study demonstrating that AMR-free commensal E. coli, termed competitive exclusion clones (CECs), can be used to competitively exclude ESC-R E. coli. This study isolated and characterized phages that lysed E. coli clones harboring the InCI1-bla_CTX-M-1 plasmid, before investigation of the effect and synergy of phage therapy and competitive exclusion as a novel strategy for decolonizing ESC-resistant E. coli. In vitro testing demonstrated superiority in the combined therapy, reducing and possibly eliminating ESC-R E. coli through phage-mediated lysis coupled with simultaneous prevention of regrowth of phage-resistant mutants due to competitive exclusion with the CEC. Further investigation into this combined therapy in vivo is warranted, with on-farm application possibly reducing ESC-R prevalence, while constricting newly emergent ESC-R E. coli outbreaks prior to their dissemination throughout food-producing animals or humans.

IMPORTANCE The emergence and global dissemination of resistance toward critically important antimicrobials, including extended-spectrum cephalosporins in the livestock sector, deepens the One Health threat of antimicrobial resistance. This resistance has the potential to disseminate to humans, directly or indirectly, nullifying these last lines of defense in life-threatening human infections. This study explores a novel strategy, the coadministration of bacteriophages (phages) and a competitive exclusion clone (antimicrobial-susceptible commensal E. coli), to revert an antimicrobial-resistant population to a susceptible population. While phage therapy is vulnerable to the emergence of phage-resistant bacteria, no phage-resistant bacteria emerged when a competitive exclusion clone was used in combination with the phage. Novel strategies that reduce the prevalence and slow the dissemination of extended-spectrum cephalosporin-resistant E. coli in food-producing animals have the potential to extend the time frame in which antimicrobials remain available for effective use in animal and human health.

INTRODUCTION

Antimicrobial resistance (AMR) has become one of the most pressing issues affecting human and animal health globally (1). Without a concerted international effort to control AMR, it is predicted to cause 10 million human deaths per year by 2050, at a cumulative economic loss exceeding $100 trillion (2). One of the most pervasive AMR threats are critically important antimicrobial-resistant (CIA-R) Gram-negative bacteria, especially sepsis-causing pathogens such as Escherichia coli resistant to last-line-of-defense human therapies (extended-spectrum cephalosporins [ESCs], fluoroquinolones [FQs], carbapenems, and colistin). Livestock are a potential reservoir for the development and dissemination of CIA resistance, which can be transmitted to humans through the food chain or via more diffuse ecological pathways (3).

The emergence of ESC-R E. coli in food-producing animals has heightened concern regarding the use of CIAs in food animals and has resulted in tightened regulations. Despite these efforts, diverse forms of ESC resistance in E. coli have continued to spread globally (3, 4). Concerningly, resistance toward ESCs is predominantly plasmid mediated and commonly attributed to variants of the antimicrobial resistance gene bla_CTX-M (4). Globally, genetically similar and highly transferable IncI1-bla_CTX-M-1 plasmids have been detected in genetically diverse E. coli from various animal species, including detection in potentially pathogenic E. coli lineages (5–7). Furthermore, multiple studies have demonstrated the long-term persistence of ESC resistance in environments despite these environments having no direct antimicrobial selection pressures (5, 8, 9). These two characteristics increase the One Health risk associated with the spread of ESC resistance through livestock and to humans.

A significant decrease in CIA-resistant bacteria in food-producing animals is required to thwart its dissemination and minimize the associated One Health threat. While antimicrobial stewardship reduces selection pressures on bacteria, development of strategies to decolonize CIA resistance upon emergence are necessary. One such potential strategy is the use of lytic bacteriophages (phages). Phage therapy is increasing as a method for treatment of last-line resistant infections in immunocompromised human patients (10, 11) and has been investigated in swine for the clearance of pathogenic bacteria (12, 13). Recent studies have also started to pave the way for phage treatment targeting CIA-R bacteria in healthy hosts through the isolation, characterization, and investigation of phages capable of lysing CIA-R bacteria (14, 15). The advancement of next-generation sequencing continues to advance our understanding of phage-bacteria interactions and through phage genome annotation can enhance selection of phages for therapeutic use through determining the phage life cycle and the presence of any detrimental bacterial genes (16). However, the main factors that continue to jeopardize the success of phage therapy, against both pathogens and CIA-R bacteria, are the in vivo efficacy of phage therapy, the high specificity of phages, and the development of phage resistance among target bacteria. The efficacy of phages has been demonstrated to be dependent on the phage-processing method, dose schedule, host-neutralization, and administration method (17–19). Recent studies have begun to optimize these variables evident in the increased retention time of liposome-encapsulated phages throughout the gastrointestinal tract (18). Meanwhile, the issue of narrow target range has largely been overcome through the design of phage cocktails; a preparation of multiple phages capable of targeting different bacterial strains (17). Phage cocktails have also been demonstrated to lessen the emergence of phage-resistant bacteria; however, the emergence of phage-resistant bacteria continues to occur and hinders the application of this novel CIA-R control method (15, 17). While phages also have the evolutionary potential to mutate and can potentially regain lytic activity against these mutated bacterial strains, this relies on random activity, which is far from ideal for a therapeutic intervention; an ideal control method would prevent or minimize the emergence of phage resistant bacteria.

Another approach worthy of investigation is the potential displacement of resistant bacteria by commensal bacterial strains capable of outcompeting resistant strains in the gastrointestinal tract. This concept, termed probiotics, has also previously been used against pathogenic bacterial strains, reducing the availability of nutrients to pathogens and host cell adhesion sites through competitive exclusion (20). The modification of this approach to target CIA-R E. coli, using fully susceptible commensal E. coli clones capable of outcompeting CIA-R clones, has potential to prevent colonization by CIA-R bacteria or to displace colonized bacteria. Furthermore, these AMR-free clones, referred to in this context as competitive exclusion clones (CECs) could be selected to possess additional fitness advantages, such as the ability to produce bacteriocins, offering additional mechanisms for displacement of resistant E. coli. Bacteriocins are antimicrobial peptides, incorporating both colicins and microcins, and are produced in response to stressful conditions, targeting and killing E. coli strains that lack immunity proteins (21). While E. coli CECs may offer a CIA-R control method, a combination approach may have synergistic potential, through initial reduction of the target bacteria due to phage lysis followed by prevention of recolonization due to the competitive exclusion provided by the CEC. Overall, upon emergence of CIA-R bacteria in a herd, specifically in countries with no or low prevalence of CIA-R bacteria, the possibility to revert the bacterial population to a CIA-susceptible population would reduce the risk of transfer to humans through the food chain, ultimately reducing the associated public health risk.

We here show the isolation and genomic characterization of phages with specific lytic activity against ESC-R E. coli, all of which harbor the InCI1-bla_CTX-M-1 plasmid and belong to multiple E. coli lineages, followed by in vitro testing of phage, CEC, and the combination, on ESC-R E. coli growth (Fig. 1). We hypothesized that the combination of phage along with CEC isolates targeting ESC-R E. coli would demonstrate greater efficacy in controlling ESC-R E. coli growth than either intervention alone, offering a future ESC-R control method in livestock.

FIG 1.

Process for selection of phage and competitive exclusion clone for use as combination therapy for controlling critically important antimicrobial-resistant bacteria in livestock. Stages 1 and 2 can occur simultaneously followed by stage 3 to analyze their combined efficacy. CIA, critically important antimicrobial.

RESULTS

Phage isolation, specificity, and characterization.

Swine fecal samples were collected from two distinct locations and screened for phages capable of lysing the ESC-R E. coli strain E. coli_SA13. Overall, 20 phages targeting E. coli_SA13 were isolated from across both locations. The specificity of these phages against 14 ESC-R E. coli strains were analyzed with the majority of phages (n = 16) demonstrating a narrow host range, only capable of lysing the bacterial isolate used for phage isolation (Fig. 2). In comparison, phages 26 and 27 lysed 64.3% (n = 9) of the ESC-R E. coli tested with phage 30 and 31 lysing 28.6% (n = 4) and 14.3% (n = 2) ESC-R E. coli, respectively. On average, phages isolated from the same environment as the ESC-R E. coli were capable of lysing 1.0 ESC-R E. coli, while phages isolated from a separate location lysed 3.5 ± 1.3 standard error (SE) different ESC-R E. coli isolates. These ESC-R E. coli belonged to different MLSTs; however, all harbored the globally disseminated InCI1-bla_CTX-M-1 plasmid. Phages showed high specificity to ESC-R E. coli, with 15 phages lysing only one ESC-susceptible E. coli strain from the 44 ESC-susceptible E. coli tested. These E. coli represent commensal and pathogenic strains from multiple hosts, including swine, cattle, and companion animals, as well as laboratory strains. The lysed E. coli strain, E. coli J53, is a sodium azide-resistant laboratory strain and laboratory mutant of E. coli K-12. No lysis of other bacterial genera or species was detected.

FIG 2.

Overview of lytic activity of isolated phages (n = 20) against extended-spectrum cephalosporin-resistant E. coli isolates that harbor bla_CTX-M-1 (n = 14). Multilocus sequence types of E. coli are indicated after the underscore in the isolate name.

All phages were subjected to whole-genome sequencing with 19 phage genomes extracted from 17 samples, with two different phages extracted from phage lysates 3 and 4. A single phage belonged to the viral family Myoviridae, while nine belonged to the families Siphoviridae and Drexlerviridae each. The phages classed in the Drexlerviridae viral family were all isolated from location A, while phages classified as Siphoviridae were isolated from both sources. None of the phage genomes contained the lysogenic-associated integrase gene. Phage 26, belonging to the Siphoviridae family, was selected for all further testing based on its broad host range (Fig. 2) and lack of lysogenic genes, bacterial virulence genes, and antimicrobial resistance genes (Fig. 3). The genome of phage 26 was 42,262 bp in length with a G + C content of 50.4%. Seventy-three open reading frames (ORFs) were identified with annotation of ORFs identifying cell lysis-related proteins, structural proteins, and DNA metabolism-associated proteins.

FIG 3.

Annotated genome map of phage 26. The blue line indicates the G + C content. The functional categories and annotation of open reading frames are indicated by specific colors. The scale units are base pairs with genome length of 42,262 bp.

Phage assay.

The growth of E. coli_SA13 was inhibited in the presence of phage 26 as illustrated in Fig. 4. Bacterial growth in cultures inoculated with phage showed an initial increase before declining by the end of the assay. In comparison, noninfected E. coli_SA13 concentration increased exponentially before plateauing. The inhibition level of E. coli_SA13 was related to phage concentration with inoculum of 30 PFU/mL of phage 26, demonstrating the highest bacterial growth at T₂₁₀ and the longest time to clear E. coli_SA13. In comparison, the 10-fold increasing phage concentration series demonstrated lower bacterial growth rates and shorter durations to eliminate the ESC-R bacteria with minimal growth of E. coli_SA13 detected when inoculated with 3 × 10⁶ PFU/mL of phage 26.

FIG 4.

Bacterial growth curve of extended-spectrum cephalosporin-resistant E. coli strain SA13 when inoculated with 10-fold increasing concentrations of phage 26. Error bars indicate the standard error of the mean. The experiment was performed in triplicate. OD₆₀₀, optical density at 600 nm.

Competitive exclusion clone selection and assay.

Selection of E. coli strains for use as CECs was based on their antimicrobial susceptibility, absence of known virulence genes, growth rate measures, and presence of colicin genes. In a previous study, 800 commensal E. coli were isolated from Australian finisher pigs (22), with 80 of these isolates selected for the current study based on antimicrobial susceptibility data, availability of whole-genome sequencing (WGS) data, and the absence of the E. coli virulence genes eltA, est, stx2e, and eae. Growth curves of these 80 isolates were analyzed based on the rate of growth between absorbance readings of 0.2 and 0.5 and the concentration reached at plateau (Table 1). Isolate H10 was selected due to high performance in these two metrics, while B1 was selected as a negative control due to low performance.

TABLE 1.

Growth characteristics and presence of colicin genes of E. coli isolates selected for use as competitive exclusion clones

Isolate	Growth rate (absorbance OD600/min)	Concn at plateau (absorbance OD600)	Colicin genes
B1	0.0036	0.59
F1	0.0051	0.65	cba, cma
H10	0.0051	0.66

Isolates were also screened for colicin and microcin genes with the colicin genes cba and cma and the microcin gene mchB identified in 20 of the potential CEC isolates with all three detected in a single isolate, and cba and cma detected in eight isolates. No colicin genes were detected in isolates H10 and B1; therefore, a third isolate was detected to represent colicin-producing strains. The growth curves of isolates with two or more colicin genes were compared with F1 selected for use in combination assays due to its high growth performance. These three selected CECs were tested in vitro for colicin activity against E. coli_SA13. The culture supernatants of each isolate were tested for inhibitory activity against the growth of E. coli_SA13. The culture supernatant from F1 demonstrated the strongest inhibition with ESC-R E. coli growth at the final time point reduced from an absorbance of 1.28 to 1.07. In comparison, isolates B1 and H10 demonstrated mild inhibitory activity with both isolates having a final absorbance of 1.19 (Fig. 5).

FIG 5.

Inhibitory effect of culture supernatants, demonstrating colicin activity, from competitive exclusion clones (CECs) on the growth of extended spectrum cephalosporin-resistant E. coli strain SA13. The growth rate of pure extended-spectrum cephalosporin-resistant (ESC-R) E. coli is illustrated in red. The error bars indicate the standard error of the mean. The experiment was performed in triplicate. NC, culture supernatant from non-colicin-producing strain.

The effect of selected CECs on the growth of ESC-R E. coli when grown in combination was investigated. Overall, all three CECs were demonstrated to reduce the growth of ESC-R E. coli (Fig. 6). The growth of ESC-R E. coli was similar between cultures for the first 6 h of the assay with divergence between concentrations first detected at T₂₄ and again at T₄₈. The largest reduction in ESC-R E. coli concentration at T₂₄ was detected when grown in combination with CEC H10 with a concentration of log₁₀ 6.00 CFU/mL compared to the log₁₀ 8.80 CFU/mL when grown as a pure culture. Furthermore, ESC-R E. coli concentration was significantly reduced (P value < 0.001) when grown in combination with each CEC at the final time point (T₄₈), measured to be a reduction of 2.14, 1.74, and 1.39 log₁₀ CFU/mL for CEC F1, B1, and H10, respectively. Competitive exclusion clones F1 and B1 were selected for use in the combination assay with F1 significantly (P value = 0.012) reducing ESC-R concentration compared to H10 at the final time point. The final concentration of the CECs was significantly increased by 0.84 log₁₀ (P value = 0.002) and 0.73 log₁₀ (P value = 0.032) CFU/mL for CEC H10 and CEC F1 when grown in the presence of the ESC-R E. coli strain compared to the pure culture, respectively. In comparison, on average the final concentration of CEC B1 was significantly reduced (P value = 0.033) by 1.07 log₁₀ CFU/mL when grown in the mixed culture with ESC-R E. coli.

FIG 6.

Effect of different E. coli CECs on the growth of ESC-R E. coli. The growth rate of pure ESC-R E. coli is illustrated in red. The data points indicate average CFU/mL. Error bars indicate the standard error of the mean (CFU/mL).

Combination assay.

The combined effects of phage therapy and CECs were tested against each treatment as a single modality and against a control of pure ESC-R E. coli (Fig. 7). This was completed using the two different CECs, F1 and B1, due to these two CECs showing the lowest final ESC-R E. coli concentration in the above CEC assay. Treatment with CECs alone demonstrated results similar to those of the above CEC assays with exponential growth of the ESC-R E. coli strain for the initial growth; however, treatment with CECs alone showed a reduced concentration at T₄₈ compared to the pure ESC-R E. coli growth.

FIG 7.

Effect of different methods on the growth of ESC-R E. coli in a series of in vitro experiments. The methods analyzed were bacteriophages (green), E. coli CECs (orange), and the combination of bacteriophages and CECs (blue). (a) Isolate F1 used as CEC. (b) Isolate B1 used as CEC. Data points represent average log₁₀ CFU/mL with error bars representing standard error of means with the exception of each biological replicate (Rep) of phage-treated ESC-R growth illustrated for observation of time point in emergence of phage-resistant ESC-R E. coli. Inoculation of phage 26 occurred at T₃ and T₆. Assay detection limit (log₂^.3) represented by red dashed line. The experiments were performed in triplicate.

E. coli_SA13 was no longer detectable across all replicates 2 h after the first inoculation of phage (T₃). The concentration of E. coli_SA13 remained below the detection limit for 3 and 4 h in one and three biological replicates, respectively, while ESC-R E. coli was detected again at the final time point (T₄₈) for two biological replicates. The average ESC-R E. coli concentrations returned to within 0.67 and 0.09 log₁₀ CFU/mL of the negative control by the final time point. Compared to treatment with CEC alone, phage therapy initially demonstrated superior results through the clearance of E. coli_SA13. However, by the end of the assays, a greater reduction of ESC-R E. coli was detected when treated with CEC. Furthermore, treatment with both phages and CEC resulted in a failure to detect ESC-R E. coli (less than log 2.3 CFU/mL) by the final time point. An initial increase in E. coli_SA13 was observed before a sharp decline 2 h after administration of phage 26, similar to when treated with phages alone. However, E. coli_SA13 did not reemerge with no detectable growth for the remainder of the assay. Minimal effects were detected on the concentration of CEC strains in treatments compared to when grown in pure culture, with final concentrations within 0.32 and 0.38 log₁₀ CFU/mL of the concentration of the pure culture for all treatments for F1 and B1, respectively.

DISCUSSION

In this study, we demonstrate in vitro the efficacy of a promising novel strategy combining lytic phages and CECs for the control of CIA resistance in livestock. These therapies, having two different mechanisms of action, demonstrated a synergistic effect when administered in combination. This resulted in the significant reduction (below detectable levels) and possible elimination of ESC-R E. coli and was markedly more effective than when either were administered as single modalities.

Bacteria harboring resistance toward CIAs have been detected globally in food-producing animals. This emergence has occurred despite the implementation of numerous actions by the industry (4). Furthermore, studies have recently indicated human and bird isolates as sources of CIA resistance on farms, with this avenue of introduction difficult to control (4, 9, 23, 24). Upon emergence of CIA-resistant bacteria, the carriage of additional antimicrobial resistance genes allows for coselection by routinely administered antimicrobials, resulting in the long-term persistence of CIA-resistant bacteria in herds (5). Together, these results demonstrate that strict regulation of CIA usage cannot be solely relied upon for the control of CIA-R bacteria in food-producing animals. Of particular One Health concern are globally detected ESC-R E. coli, containing the dominant bla_CTX-M-1 gene detected within a highly transferable InCI1 plasmid. Development of novel strategies is urgently required to reduce current prevalence and further dissemination of resistance while remaining cost effective for application at the farm level. In the current study, we isolated phages capable of lysing ESC-R E. coli. Initial experiments demonstrated greater effects on reducing ESC-R E. coli with increasing phage concentration with the highest phage concentration tested, 3 × 10⁶ PFU/mL, demonstrating minimal bacterial growth (Fig. 4). While these low phage concentrations demonstrated in vitro efficacy, the ideal concentration of phage and the probability of successful phage therapy in vivo are dependent on many factors within the complex environment of the gastrointestinal tract. These include phage-microbiome dynamics, gut motility, and various pH levels that may reduce the titer of active phage reaching the target bacterium (18, 25). Therefore, while these low phage concentrations have the potential for successful phage therapy in optimum conditions, in vivo models commonly dose with increased phage titers ranging from 10⁷ to 10¹⁰ PFU to increase the chances of successful phage therapy.

Bernasconi et al. (15) also investigated the in vitro efficacy of phage therapy for the decolonization of an ESBL-producing ST131 E. coli strain (E. coli 4901.28). While repeated doses of the phage cocktail reduced E. coli 4901.28 to nondetectable concentrations (10¹ CFU/mL or less) in the first of two fecal pools tested, a phage-resistant strain emerged in the second experiment, returning the ESBL-resistant E. coli concentration to 10³ to 10⁴ CFU/mL (15). In the current study, we observed similar results with an initial reduction of the CIA-R bacteria 2 h after inoculation with phage, before phage-resistant strains emerged, returning ESC-R E. coli concentrations to levels similar to those of the untreated culture. In contrast, simultaneous administration of CECs prevented the reemergence of CIA-R bacteria through competitive exclusion. These results establish this combined therapy as a promising strategy to modify and modulate the resistant E. coli population in herds toward an AMR-susceptible population.

Currently, the requirement to access outbreaks of ESC-R E. coli for the isolation of phages has biosecurity implications and One Health risks due to the increased potential for dissemination between livestock enterprises. Recent studies have investigated overcoming this biosafety issue by isolating phages from locations separate from the occurrence of the target bacterium (14, 26, 27). Skaradzińska et al. (14) isolated phages from a Polish swine farm capable of lysing a range of E. coli resistant to extended spectrum β-lactams (ESBL) and AmpC β-lactam from German swine and turkey farms. The current study supports this, with phages isolated from a distinct location having an increased host range against the ESC-R E. coli, an ideal characteristic for phage therapy. These broad range phages were capable of lysing a diverse range of E. coli, with the common target attribute being the presence of the InCI1-bla_CTX-M-1 plasmid. The development of phage storage banks, a strategy repeatedly relied upon in extensively drug-resistant bacterial infections in human medicine (26, 27), is another method to overcome the biosecurity concerns. Prior development of phage storage banks and prescreening of potential CECs coupled with the evolution of robotic platforms for high-throughput screening (28) also ensure timely development of the targeted therapy in response to any future ESC-R emergence. Eliminating this resistance prior to global dissemination, as seen in the InCI1-bla_CTX-M-1 plasmid, is an ideal and obtainable scenario.

Genomic sequencing of all phages and CECs prior to addition to storage banks will also reduce response times in developing the target therapy upon emergence of CIA-R bacteria in food-producing animals. Second, WGS ensures therapeutic safety of both phages and CECs prior to administration. Analysis of phage genomes is considered an essential step in phage therapy, guaranteeing the absence of antimicrobial resistance genes and virulence genes while ensuring that phages follow a strictly lytic life cycle (16). Increasing publicly available phage sequencing data additionally allows continued research into phage-bacteria interactions and phage diversity (16, 29). Software programs such as HostPhinder (30) and WIsH (31) have begun to capitalize on this increased data collection, with the development of programs predicting phage host ranges. While current predictions are only 60 to 80% accurate (16, 30, 31), continued data collection and analysis will likely increase accuracy and aid in selection of phages from phage storage banks upon emergence of resistant bacterial strains on farms. Moreover, performing WGS on potential CECs is also required to ensure that bacterial strains do not harbor specific antimicrobial resistance or virulence genes. Additionally, the identification of acquired or lost genes in E. coli strains that result in enhanced colonization in guts of food-producing animals may optimize the selection of strains for use as a CEC (32). This includes the presence of bacteriocins in E. coli and would require in vivo trials to determine their benefits.

The combination of these existing strategies against AMR, proposed in relation to enterotoxigenic E. coli harboring CIA resistance in pigs (17), offer promising and cost-effective approaches for application in food-animal production. With this industry subject to limited profit margins, the economic feasibility of novel AMR strategies, both in cost and in ease of administration, is imperative for their continued success. Both phage therapy and probiotics have demonstrated relatively low-economic production costs in intensive production systems, with Torres-Acosta et al. (33) calculating production costs of phage therapy at $0.02 per chicken (33, 34). Meanwhile the production cost of the combined therapy in this study is approximated at $1.00 with application intended to be through freeze-drying and administration through drinking water or feed. This method for administration removes additional labor costs for repeated administration to individual animals while proving successful in previous studies of single modalities (12, 35). Furthermore, the economic burden of pathogens harboring AMR may be alleviated through decolonization of CIA-R bacteria prior to dissemination of CIA-R into pathogenic bacteria. Bacterial infections resistant to CIAs incur additional costs to industry due to extra laboratory tests, use of antimicrobials, and the increased rate of mortalities (17). While novel CIA-R control methods will initially be an extra cost, the long-term benefit for the industry has potentially huge economic paybacks through downstream decrease of these AMR-associated costs.

Overall, this study highlights the necessity to continue exploring new avenues for the control of CIA-R bacteria with this combined therapy utilizing the two different mechanisms of action of each single modality: the lytic capability of phages and the competitive exclusion of CECs. Further in vitro and in vivo studies are warranted to determine the efficacy and optimize this novel method in controlling other CIA-R bacteria. This includes evaluation of a phage cocktail in the combined therapy, the in vivo growth rate of the selected CEC, and the administration schedule of the combined therapy. Involvement of a phage cocktail would likely result in a broadened host range and therefore widen the applicability of this novel method. Meanwhile, although the CECs used in the current study were originally isolated from swine, their competitive nature may not translate in in vivo models. The E. coli population within the swine gut has been demonstrated to be highly diverse (22); therefore, these CECs will be competing against multiple E. coli lineages instead of a single ESC-R E. coli. The dosage and administration schedule would also need to be optimized to ensure that the reduction observed in the in vitro model is achieved and maintained in vivo. Additional studies exploring phage genetics are also required to further our understanding of phage-bacteria interactions. This study has demonstrated the initial success of the in vitro efficacy of phages and CECs as a combined therapy for the control of CIA-R bacteria and prevented the occurrence of phage-resistant bacterial strains through competitive exclusion. Investigation of novel CIA-R control methods in food-producing animals is warranted in the battle against AMR, potentially prolonging the use of existing antimicrobials for life-threatening human infections. Combined with the lack of direct selection pressures arising from minimizing or ceasing the use of CIAs, novel CIA-R control strategies may minimize the risk of the amplification and propagation of these forms of AMR within and beyond the livestock sector, yielding benefits for One Health.

MATERIALS AND METHODS

Study overview.

The study into the novel CIA-R control strategy was divided into three main components (Fig. 1): (i) the isolation and characterization of phages, (ii) selection of a CEC and preliminary CEC testing, and (iii) in vitro analysis of phages and CECs as single modalities and as a combined therapy. All three components required an ESC-R E. coli isolate with E. coli_SA13 selected. This isolate was selected from a previously characterized collection of ESC-R E. coli harboring an InCI1-bla_CTX-M-1 plasmid (pCTXM1-MU2) (5). Strain E. coli_SA13 belongs to ST6976; is phenotypically resistant to ampicillin, ceftiofur, ceftriaxone, tetracycline, and trimethoprim-sulfamethoxazole; and harbors the antimicrobial genes aadA1, aadA5, bla_CTX-M-1, dfrA17, tetA, sul1, and sul2.

Phage isolation.

Freshly defecated fecal samples were collected from swine from two distinct locations (over 3,000 km apart) and transported to Murdoch University on ice. Fecal samples were suspended in SM buffer (0.1 M NaCl, 1 mM MgSO₄, 0.2 M Tris [pH 7.5], 0.01% gelatin) at a ratio of 1:10 and stirred using a magnetic stirrer for 24 h at 4°C. The suspensions were centrifuged at 17,000 × g for 10 min followed by filtration through a 0.45-μm syringe-driven membrane filter unit (Pall Corporation, USA). For overnight enrichment, an equal volume of phage preparation and 2× Luria-Bertani (LB) (Thermo Fisher Scientific, Australia) broth were combined and inoculated with a single colony of the ESC-R E. coli isolate SA13. The samples were incubated overnight at 37°C on an orbital shaker (80 rpm) before filtration through a 0.25-μm filter unit.

Phage lysates were then spot tested onto lawn plates of E. coli_SA13. Briefly, 4 mL of LB broth was inoculated with a single colony of E. coli_SA13 and incubated at 37°C on an orbital shaker at 220 rpm for 5 h. The culture was spread onto a LB agar plate, excess broth was removed, and the plates were dried before 20-μL aliquots of phage lysates were dropped onto agar, allowed to dry, and incubated overnight at 37°C. Phage growth was indicated by the formation of phage plaques. A section of plaque was harvested using a sterile Pasteur pipette and suspended into a solution containing 1 mL of SM buffer and 25 μL of chloroform (Sigma-Aldrich, Australia). The phage suspension was held at room temperature for 4 h before storage at 4°C.

Phage stock preparation.

Phage stock was prepared by incubating 100 μL of phage lysates with 100 μL of E. coli_SA13 culture for 20 min at 37°C. This mixture was added to 3 mL of soft LB agar, mixed, and then poured onto a LB agar plate. The plates were dried and incubated overnight at 37°C. A solution containing 10 mL SM buffer and 200 μL chloroform (Sigma-Aldrich, Australia) was poured on top of the soft agar. The plates were stored at 4°C with constant agitation for 4 h. The supernatant was then collected, centrifuged, and filtered through a 0.25-μm membrane filter. Phage stocks were immediately used for characterization, with the remainder stored at −80°C.

Phage specificity.

Host range of phages was tested against 14 ESC-R E. coli originating from location A and harboring the InCI1-bla_CTX-M-1 plasmid (5). Further specificity testing was completed against 44 E. coli isolates from in-house archival stores, including both commensal and pathogenic strains and originating from swine, cattle, seagulls, companion animals, and laboratory strains. Additional testing was conducted testing phage lysis against multiple bacterial genera commonly comprising the gut microflora, including Salmonella spp. (n = 1) (36), Streptococcus spp. (n = 2) (37), Enterococcus spp. (n = 7) (38), Staphylococcus aureus (n = 4) (24), and in-house archival stores of Bacillus spp. (n = 7). This was conducted using spot testing as described above, with phage lysis measured as complete lysis or no lysis.

Phage characterization.

Phages were subjected to WGS for genetic characterization. Phage stock was concentrated 10-fold using 500 μL Vivaspin 10-kDa cutoff protein concentrator spin columns (GE Healthcare Life Sciences, Australia). DNA was extracted using a MagMAX viral isolation kit (Ambion, Australia). Library preparation was performed using a Nextera XT DNA library preparation kit (Illumina, USA) with sequencing performed on an Illumina Miseq platform using a V3 2 × 300 flow cell. Sequencing data were de novo assembled using Geneious Prime 2021.1.1 (https://www.geneious.com), resulting in 12 contigs with a N₅₀ of 42,589 and GC% content of 50.43. The single contig larger than 10,000 bp in length was characterized by contig sequence homology to known viral sequences in the BLASTn database. MultiPhATE2 (39) was used for gene calling and annotation using the blastp, phmmer, and jackhmmr databases. All phage sequences were screened for genes associated with lysogeny, antimicrobial resistance, and bacterial virulence.

Phage assay.

Phage 26 was selected for use in the phage assay and combination therapy assays due to its strong lytic activity against E. coli_SA13 and broad host range against ESC-R E. coli. Phage stocks were prepared using the double agar method, and stock was titrated by spotting 10-fold dilutions onto an E. coli_SA13 lawn LB agar plate in triplicate. Phage concentration was determined by analyzing the growth of E. coli_SA13 when inoculated with various concentrations of phage. This was conducted in a 96-well plate with wells containing 160 μL LB broth, 20 μL of an overnight culture of E. coli_SA13 diluted to a 0.5 McFarland standard, and 20 μL of the phage inoculum. Ten-fold dilutions of phage were used starting at a concentration of 3 × 10⁷ PFU/mL. The assays were completed in triplicate with positive (bacterial growth with no phage) and negative included. The plate was sealed and placed on a SPARK reader for 6 h with continued incubation at 37°C, constant shaking at 270 rpm, and absorbance (optical density at 600 nm [OD₆₀₀]) readings completed every 10 min.

Selection of competitive exclusion clones.

Commensal E. coli from Australian finisher pigs (22) were selected as potential CECs based upon the following selection criteria: (i) they were either fully susceptible or harbored resistance to only ampicillin or tetracycline when tested against ampicillin, ceftriaxone, ciprofloxacin, gentamicin, tetracycline, and trimethoprim-sulfamethoxazole; (ii) they had undergone WGS; and (iii) E. coli virulence genes eltA, est, stx2e, and eae, were not detected within the genome sequence.

Selection of isolates for inclusion in further assays was based on the bacterial growth rate with an additional isolate selected based on this isolate being the fastest growing strain that harbored two or more colicin or microcin genes. Colicin and microcin genes were determined by analyzing WGS data using Abricate (40). The E. coli virulence finder database was used with cutoffs of 95% or more for coverage and 99% or more for identity. Growth curves were determined by growing bacterial isolates overnight at 37°C in individual wells of a 96-well deep-well plate containing LB broth. Suspensions were diluted using a Tecan Evo liquid handling platform with an attached absorbance reader to an absorbance (OD₆₂₀) equivalent to a 0.5 McFarland standard. A further 1 in 10 dilution was completed using LB broth with transfer to a 200-μL 96-well U-bottom plate. Bacterial growth was measured using a Tecan SPARK reader programed as described above. Colicin production and activity was tested in vitro as described by Abraham et al. (41). The genomic sequences of the CECs are available in the NCBI database under BioProject number PRJNA720242. Isolates B1, F1, and H10 are isolates F1P10-008, F1P7-002, and F10P1-007, respectively (22). Quality control was performed on all sequencing data using FastQC (v0.11.09) and Nullarbor with average base quality scores of 33.5, 33.2, and 33.1 and average depth of 33×, 40×, and 51× for isolates B1, F1, and H10, respectively (42).

Competitive exclusion clone assay.

Overnight cultures of bacterial strains were standardized to an OD₆₀₀ of 0.1 using a SPARK (Tecan) reader followed by a 100× dilution (approximately 1.6 × 10⁶ CFU/mL). The assays were performed in 96-well deep-well plates with a starting assay volume of 2 mL. This consisted of 20 μL of the CEC and/or the ESC-R aliquoted into LB broth. Sampling was completed at the specified time intervals with 20 μL removed and used for 10-fold serial dilutions. These dilutions and a sample directly from the assay plate were used to quantify the bacterial strains, recorded as CFU/mL. This was repeated in triplicate by dropwise addition of 5 μL onto agar plates, followed by drying of the droplet and then overnight incubation of the agar plate. Two different agar plates were used: LB agar alone and LB agar + ceftriaxone (8 μg/mL). Cultures containing the CEC were spotted onto LB agar, cultures containing the ESC-R E. coli were spotted onto LB agar containing ceftriaxone, and cultures containing both isolates were spotted onto both agar plates. All plates were incubated overnight at 37°C, and the colonies were counted manually the next day. The CEC quantification in the mixed cultures was calculated using the ESC-R E. coli CFU/mL subtracted from the total bacterial CFU/mL.

Combination assay.

Combination assays were completed using ESC-R E. coli, phage 26, and either CEC F1 or B1. This was completed as described for CEC assays with the addition of phage inoculation at 3 and 6 h. The phage was inoculated into corresponding cultures, with 20 μL of 3 × 10⁷ PFU/mL added. Biological replicates were completed for combination assays to detect potential mutation of the ESC-R E. coli strain.

Statistical analysis.

All graphing and analysis was conducted using Stata 15.1 (43). Bacterial concentrations (CFU/mL) were log₁₀ transformed for graphing and statistical analysis. Comparisons of bacterial growth at final time points of the CEC assays were performed using linear regression model (analysis of variance [ANOVA]) using a continuous dependent variable (log₁₀ CFU/mL) and the interacting effects of the four-factor categorical variable (treatment) and continuous independent variable (time). ANOVA was followed by pairwise comparison of means with Tukey’s adjustments. The α significance level was set to 0.05.

Data availability.

The annotated genome of phage 26 was deposited in the NCBI database under GenBank accession number MZ832314.