Stable co-existence of Citrobacter rodentium with a lytic bacteriophage during in vivo murine infection

ABSTRACT

Bacteriophages are ubiquitously present in bacterial communities; however, phage-bacteria interactions in complex environments like the gut remain poorly understood. Although antibiotic resistance is driving a renewed interest in phage therapy, most studies have been conducted in in vitro systems, offering limited insight into the complexity of such dynamics in physiological contexts. Here, we use the mouse-restricted enteric pathogen Citrobacter rodentium (CR), a well-established model for human enteropathogenic and enterohemorrhagic Escherichia coli (EPEC and EHEC) infections, to investigate phage-pathogen interactions in a murine model with a complex microbiota. We isolate and characterize Eifel2, a novel lytic phage infecting CR, and generate anti-phage-specific antibodies that enable the visualization of phage infections in vitro. In a murine model of CR infection, oral administration of Eifel2 led to robust phage replication in the gut without reducing the bacterial burden or infection-associated inflammation, confirming the establishment of a stable co-existence in the gut. Despite the emergence of a sub-population of phage-resistant CR mutants in vivo, they did not undergo clonal expansion, indicating that additional selective pressures impaired their widespread dissemination in the gut. Together, our findings demonstrate that imaging approaches can capture key infection stages in vitro, although in vivo models are essential for capturing the complexity of phage-bacteria interactions. This work highlights the importance of studying phage therapy in host-pathogen contexts that include a normal microbiota and a suitable host environment, where dynamic co-existence rather than eradication may define therapeutic outcomes.

IMPORTANCE

Bacteriophages, or phages, are viruses that can either kill or persist inside bacteria. Current interests in phage biology are in part ignited by the fact that they could be used to treat infections caused by antibiotic-resistant bacteria. However, most of our understanding of phage-bacterial interactions comes from in vitro models and/or in vivo gut models relying on altering the endogenous microbiota. Here, we report the finding of a novel phage, Eifel2, which specifically targets Citrobacter rodentium (CR), the mouse equivalent of human diarrheagenic E. coli pathogens. Despite effectively killing CR in vitro, CR and Eifel2 develop a co-existence relationship in mice with an intact microbiota. Although CR phage-resistant mutants emerge, host and microbial factors constrain their expansion. This work highlights the importance of studying phage therapy in host-pathogen contexts that include the complete microbiota, where therapeutic outcomes may rely on dynamic co-existence and containment rather than eradication.

INTRODUCTION

Lytic bacteriophages or phages, viruses that specifically infect and lyse bacteria, are receiving renewed interest as an alternative to antibiotics, particularly to treat multidrug-resistant bacterial pathogens (1–3). However, translating promising in vitro results into successful therapeutic outcomes in vivo remains a challenge (2, 4). Recent studies in murine models have shown that introducing phages can lead to stable co-existence between phages and their bacterial hosts in the gut instead of bacterial eradication (5, 6). Such systems support a dynamic equilibrium, with stable levels of both phage and bacteria over time. This suggests active phage replication, potentially driven by protected bacterial replication niches in the gut (7), phage host-range expansion to microbiota members (8), or bacterial and phage coevolution, leading to resistant subpopulations (9, 10). Resistance can arise from the mutation of phage receptors and often involves the modification of surface structures such as lipopolysaccharide (LPS) or outer membrane proteins, which can carry fitness trade-offs in vivo and thus do not lead to a predominantly phage-resistant population (6).

One of the major limitations to studying phage-bacteria interactions in the gut is the availability of physiologically relevant infection models. Mice with an intact microbiota are usually refractory to exogenous bacterial infection, and therefore, antibiotic pre-treatment or other methods to interfere with colonization resistance are needed (11). This limits our ability to investigate phage-bacteria interactions and key processes such as the emergence and fitness cost of phage resistance in translational models, where they can be profoundly influenced by factors such as physicochemical conditions, host immune responses, spatial structure, and microbial community dynamics.

To overcome these limitations, Citrobacter rodentium (CR), a natural murine enteric pathogen, has been widely used as a surrogate model for EPEC and EHEC infections in vivo. CR shares key infection mechanisms with EPEC and EHEC, including the formation of attaching and effacing (A/E) lesions. CR, which robustly colonizes mice with an intact microbiota, triggers colonic crypt hyperplasia (CCH) and gut inflammation, manifested by increased fecal levels of lipocalin-2 (LCN-2) (12–14). As such, the CR model provides a valuable opportunity to study not only phage-bacteria interactions in a murine model with a complex microbiota but also the broader gut ecological and immunological dynamics that shape phage therapy outcomes in a physiologically representative context. Despite this potential, relatively few phages infecting CR have been isolated to date (15–18), and their application in murine models of enteric infection remained unexplored.

In this study, we aimed to isolate and characterize novel lytic phages that infect CR. We investigated phage-bacteria interactions in vitro and in vivo by combining immunofluorescence imaging with our well-established mouse model of CR infection.

RESULTS

Characterization of CR lytic phages Eifel1 and Eifel2

Phage Eifel1 infecting the CR strain ICC169 was isolated from sewage water samples collected in the Eifel region in Germany. Eifel1 forms small, translucent plaques (<1 mm in diameter) on soft agar CR lawns (Fig. 1A). During the purification process, we also recovered larger plaques (1.5–2 mm), appearing among the Eifel1 plaques (Fig. 1B). We isolated these separately and designated the new phage variant Eifel2.

Fig 1.

Morphology and replication characteristics of phages Eifel1 and Eifel2. (A and B) Plaque morphology of phages Eifel1 (A) and Eifel2 (B) on a double-layer agar (DLA) plate with CR. Scale bar: 1 cm. (C and D) Representative transmission electron micrographs of uranyl acetate negatively stained Eifel1 (C) and Eifel2 (D) virions. Scale bar: 50 nm. (E) Lysis kinetics of CR cultures infected with Eifel1 (solid lines) and Eifel2 (dashed lines) at the indicated MOIs or uninfected (UI) (solid black line). Data are shown as the mean of n = 3 independent biological repeats. (F and G) One-step growth curve of phages Eifel1 (F) and Eifel2 (G) in CR at a MOI of 1, showing latent period and burst size. Phages were allowed to adsorb to CR for 5 min, after which unbound phages were washed off, and phage titers were determined at the indicated time points. Data are shown as mean ± standard deviation (SD) of n = 5 independent biological repeats.

Transmission electron microscopy (TEM) revealed that Eifel1 and Eifel2 share a similar morphology, with virions formed of an icosahedral head and a long contractile tail (Fig. 1C and D). The head of Eifel1 measured 76.2 ± 2.8 nm (mean ± SD) in length and 73.8 ± 2.9 nm in width, whereas the tail measured 110.9 ± 4.5 nm and the neck 11.0 ± 2.5 nm (n = 23 virions). Eifel2 exhibited comparable dimensions: head length of 77.8 ± 3.0 nm, width of 74.2 ± 3.4 nm, tail length of 116.2 ± 9.2 nm, and neck size of 11.5 ± 2.5 nm (n = 56 virions). Head width and neck length showed no significant differences between Eifel1 and Eifel2 (P > 0.05 by Mann-Whitney test), whereas head length and tail length were significantly longer in Eifel2 (P = 0.0176 and P = 0.0005 by Mann-Whitney test, respectively).

To assess their lytic activity, we infected CR with Eifel1 and Eifel2 at multiplicities of infection (MOIs) ranging from 1 to 10⁻⁷ and monitored optical density (OD600) for 16 h (Fig. 1E; Fig. S1A and B). For most MOIs tested, Eifel1 and Eifel2 reduced OD₆₀₀ following an initial growth period of 2–10 h, with shorter delays at higher MOIs. After this initial decrease, indicating bacterial lysis, OD₆₀₀ levels increased again, suggesting the emergence of resistant bacteria. Notably, cultures infected with Eifel2 at MOI 1 showed no growth until 8 h post-infection (Fig. 1E). At the lowest MOIs (10⁻⁶ and 10⁻⁷), Eifel1 had no observable impact on bacterial growth compared with uninfected controls. Across all MOIs tested, Eifel2 caused a more rapid and pronounced OD₆₀₀ decrease than Eifel1, suggesting higher lytic activity.

To investigate whether the difference in infection kinetics was caused by variations in burst size or infection cycle length, we conducted one-step growth curve analyses. Both Eifel1 and Eifel2 exhibited a latent period of 25 min. However, Eifel1 released an average of 14 new virions per infected cell (Fig. 1F), whereas Eifel2 released 46 virions per infected cell (Fig. 1G), indicating a significantly higher burst size (P < 0.05 determined by lognormal Welch’s t test). During the latent period, Eifel1 maintained a concentration of ∼8.25 × 10⁵ PFU/mL, whereas Eifel2 reached ∼2.39 × 10⁶ PFU/mL, suggesting a higher adsorption rate for Eifel2.

We also tested the host range of Eifel1 and Eifel2 against 21 gram-negative strains (3 Citrobacter species, 11 Escherichia coli strains, 2 Klebsiella species, Enterobacter cloacae, Pseudomonas aeruginosa, Salmonella enterica, and Shigella sonnei); we also attempted to infect a gram-positive Staphylococcus aureus strain. None of the tested isolates were susceptible to either Eifel1 or Eifel2 infection, highlighting the specificity of both phages for CR.

Together, these results demonstrate that although Eifel1 and Eifel2 share morphology and host range, they exhibit a significant difference in their lytic efficiency.

Comparative genomic analysis of Eifel1 and Eifel2

To better understand the relationship between Eifel1 and Eifel2, their genomes were sequenced following phage purification. Assembly of Eifel1 revealed that its genome consists of a circular double-stranded DNA of 88,123 bp with a GC content of 39.1%. We identified 147 open reading frames (ORFs), of which 57 were assigned putative functions, along with 21 tRNA genes (Fig. 2A). We found no genes encoding integrases or other lysogeny-associated elements, and PhageAI (19) predicted a lytic lifestyle with 99.97% confidence.

Fig 2.

Genomic organization and comparison of phages Eifel1 and Eifel2. (A) Circular map of the Eifel1 genome, annotated based on ORFs predicted using Phold (https://github.com/gbouras13/phold) and displayed as directional arrows. ORFs are colored according to putative function, with predicted gene products indicated. (B) Whole genome alignment of Eifel1 and Eifel2 was performed using the progressive Mauve algorithm (20) and visualized with Easyfig (21). ORFs are represented as arrows and colored as in panel A. Genomic differences between the two phages are highlighted by black boxes.

Taxonomic classification using taxMyPhage (22) placed Eifel1 as a novel species in the Felixounavirus genus (subfamily Ounavirinae, family Andersonviridae, class Caudoviricetes) in line with the most recent taxonomy update from the International Committee on Taxonomy of Viruses (ICTV). According to the latest ICTV Master Species List (MSL40.v1), the Felixounavirus genus currently comprises 97 recognized species. To further support this classification, we constructed a phylogenetic tree using VICTOR (23), incorporating the representative genome of each Felixounavirus species alongside the Eifel1 genome. The resulting tree confirmed that Eifel1 forms a distinct clade, supporting its classification as a novel species within the genus (Fig. S2A).

Eifel2 has a circular double-stranded DNA genome of 88,736 bp, and PhageAI (19) predicted a lytic lifestyle at 99.96% confidence. Whole-genome alignment of Eifel1 and Eifel2 revealed almost identical sequences, supporting the classification of Eifel2 within the Felixounavirus genus as a member of the same novel species as Eifel1 (Fig. S2A). Only two differences were identified between the two genomes, and these were confirmed by Sanger sequencing (Fig. 2B). The first was a 12 bp deletion in an Eifel2 gene encoding a minor tail protein, which results in the loss of four amino acids (aa) in the predicted protein (Fig. S2B). Structural predictions generated with AlphaFold3 (24) suggested that this deletion would cause a loop in an α-helix at the C-terminal domain of the protein (Fig. S2C). Interestingly, BLAST analysis of the Eifel2 minor tail protein revealed homologous sequences with none containing the four aa deletion. The second difference was the presence of an additional 513 bp ORF in Eifel2, encoding a putative protein of unknown function (Fig. S2D). This gene showed no similarities with the CR encoded genes or endogenous plasmids. However, homologous sequences were detected in other Felixounavirus phage genomes within GenBank, such as Salmonella phage FelixO1 (accession number: AF320576) (Fig. S2E). The presence of an opaque halo observed around Eifel2 plaques (Fig. 1B) but not around Eifel1 plaques (Fig. 1A) suggests the production of a depolymerase (25). To investigate this, we analyzed the unique Eifel2 protein and the minor tail proteins of Eifel1 and Eifel2 using two specialized depolymerase prediction tools (DePP [26] and DepoScope [27]). Although neither protein was predicted to function as a depolymerase, a conserved tail fiber protein present in both phages was predicted to have depolymerase activity; however, this is unlikely to account for the halo observed specifically in Eifel2. Nonetheless, we cannot rule out that the unique Eifel2 protein may modulate the expression or activity of depolymerases, potentially contributing to this phenotype.

Tracking Eifel2 infection dynamics by immunofluorescence microscopy

Considering that Eifel1 and Eifel2 differ by only one four aa deletion and the acquisition of one 513 bp ORF and the higher lytic efficiency of Eifel2, we selected the latter for further characterization.

To investigate the interaction between Eifel2 and CR, we generated anti-sera to Eifel2. We immunized mice with purified Eifel2 phage particles according to the schedule in Fig. 3A, with control mice receiving a no-phage buffer control. Dot blot assays were performed using Eifel2, Eifel1, and two controls: (i) filtered supernatant from a CR overnight culture, and (ii) an additional tailed phage isolated during this study infecting CR, named ColRes, and belonging to a novel species within the Tequatrovirus genus (plaque morphology and TEM image are shown in Fig. S3A and B, respectively). Serum from immunized mice detected both Eifel1 and Eifel2, whereas no signal was observed using CR supernatant, and only a faint signal was detected for ColRes. No reactivity was seen with serum from control animals, indicating the generated antibodies specifically targeted Eifel2 and cross-reacted with Eifel1 (Fig. 3B).

Fig 3.

Generation and application of anti-Eifel2 antibodies in immunofluorescence microscopy. (A) Schematic representation of the mouse immunization protocol used to generate serum reactive to Eifel2. (B) Dot blot showing reactivity of serum from mock-immunized mice (top panel) or immunized mice (bottom panel) to CR overnight culture filtered supernatant and phages Eifel2, Eifel1, and ColRes. (C) Western blot analysis of purified Eifel2 lysed virions probed with anti-Eifel2 serum from immunized mice. (D) Immunofluorescence detection of phage Eifel2 particles. GFP-expressing CRs were either uninfected (left column), infected with Eifel2 (middle column), or infected with ColRes (right column) for 5 min. Samples were stained using either mock-immunized serum (top row) or anti-Eifel2 immunized serum (bottom row). The left panel shows the phage-specific channel (white); the right panel shows the merged images of the CR signal (GFP, green) and the phage signal (magenta). A selected region is highlighted with a white box and shown at higher magnification to visualize phage binding at the single-cell level. Scale bar: 10 µm. (E) Time-course immunofluorescence analysis of Eifel2 infecting CR. GFP-expressing CR cultures were infected with Eifel2, fixed, and stained at 15 min (top row) and 25 min (bottom row) post-infection. Columns show individual fluorescence channels: CR (GFP, green), phage (magenta), DNA (DAPI, white), and a merged image of all channels. Phage detection was performed using anti-Eifel2 serum. Scale bar: 5 µm. (B–D) Images are representative of n = 3 independent biological repeats. (E) Images are representative of n = 2 independent biological repeats. (A) Created in BioRender. Frankel, G. (2025) https://BioRender.com/nojbg9o.

ELISA revealed a ∼1,000-fold increase in Eifel2-specific IgG titers in the immunized group compared with controls (Fig. S3C). Western blot analysis of lysed Eifel2 virions revealed five distinct immunoreactive bands (Fig. 3C), indicating that the generated antibodies recognize multiple phage proteins.

We then used the anti-Eifel2 serum to follow phage binding and release by immunofluorescence microscopy (IF). To this end, we first infected CR expressing green fluorescent protein (GFP) with Eifel2. Cultures were fixed 5 min post-infection and stained with the serum of immunized or control mice. Uninfected bacteria and cultures incubated with ColRes were used as negative controls. No signal was detected for any of the samples stained with serum from mock-immunized mice. For samples stained with immunized serum, IF microscopy revealed strong fluorescent signals associated with Eifel2-infected bacteria (Fig. 3D), whereas no signal was detected in either control.

We next performed time-course infections using a one-step growth curve setup. Infected cultures were sampled at defined time points, fixed, and stained with serum from immunized mice for IF analysis (Fig. 3E). At 15 min post-infection, discrete dot signals are visible around CR, indicative of phage attachment. After 25 min, in line with the one-step growth curve data, we observed intense phage staining surrounding bacteria that displayed a reduced GFP signal and intense DAPI signal, consistent with bacterial lysis and phage release.

Eifel2 co-exists with CR in a murine model

We next aimed to study CR–Eifel2 interactions in a murine model with a complex microbiota. To this end, we first assessed its stability across different temperatures and pH values. Phage titers remained stable at 4°C and 37°C for at least 24 h (Fig. S4A). In contrast, higher temperatures progressively reduced phage viability, and no plaques were detectable after 5 min incubation at 90°C. Freeze-thawing phage preparations also led to a gradual decline in titer over time. Eifel2 titers were undetectable after 24 h of incubation at 37°C in buffers at pH 1, 2, 3, and 13, but titers remained stable for at least 24 h in buffers ranging from pH 4 to 12, indicating the phage can withstand the acidic pH in the gastrointestinal tract of fasted mice (28) (Fig. S4B).

We next investigated phage-bacteria interactions in vivo by oral administration of Eifel2 to C57BL/6 mice, raised under conventional specific-pathogen-free conditions with a complex microbiota. Mice infected with CR received daily phage treatment (1 × 10⁹ PFUs) by oral gavage, or SM buffer as a control, from 3 to 7 days post-infection (dpi), corresponding to the expansion phase, which is characterized by the spread to the colonic mucosa from the cecal patch and rapid bacterial proliferation (29). A non-infected group received phage treatment alone (Fig. 4A). We followed temporal bacterial and phage fecal shedding until no phage was detectable. No differences were observed in CR shedding between phage-treated and untreated mice, with both groups shedding between 10⁸ and 10⁹ CFU/g of stool until 13 dpi, after which bacterial loads gradually declined to 10³–10⁴ CFU/g by 23 dpi (Fig. 4B).

Fig 4.

Eifel2 replicates in a murine model of CR infection without reducing bacterial burden. (A) Schematic representation of the mouse phage treatment experiment. Mice were orally infected with CR or mock-infected and received daily oral gavage of either Eifel2 phage suspension or SM buffer (mock treatment) from 3 to 7 dpi. Bacterial and phage shedding in feces was monitored until phage clearance, which occurred at 21 and 23 dpi in two independent experiments. Mice were culled at 8 dpi and at the point of phage clearance. (B) Temporal quantification of CR shedding in feces, expressed as CFU per g of feces, in infected mice receiving phage treatment or mock treatment. From 0 to 8 dpi, sample sizes were n = 19 (untreated) and n = 20 (phage-treated) across three independent experiments. After 8 dpi, data are from n = 7 (untreated) and n = 8 (phage-treated) across two independent experiments. (C) Temporal quantification of Eifel2 shedding in feces, expressed as PFU per g of feces, in infected or mock-infected mice receiving phage treatment. For UI mice, sample sizes were n = 16 (0–8 dpi) and n = 8 (after 8 dpi) across two independent experiments. For infected mice, sample sizes were n = 20 (0–8 dpi) and n = 8 (after 8 dpi) across three independent experiments. (D) Correlation analysis of log₁₀-transformed PFU and CFU counts in feces of phage-treated infected mice. Each point represents values from an individual mouse at a specific time point after the end of phage treatment (9–22 dpi). Simple linear regression is shown with the corresponding equation and R². Data points below the LoD for either PFU or CFU were excluded. (B, C) Data are shown as mean ± standard error of the mean (SEM), and values below the limit of detection (LoD) were set to 1 for plotting. (A) Created in BioRender. Frankel, G. (2025) https://BioRender.com/q1nlu57.

Phage shedding, however, differed between CR-infected and uninfected animals. In both groups, fecal phage titers were 10⁷–10⁸ PFU/g 1 day after the beginning of phage treatment. In uninfected mice, phage titers remained at 10⁶ PFU/g for the duration of the treatment and then steadily declined once the treatment ended, becoming undetectable by 11 dpi (4 days after the end of the treatment). In contrast, CR-infected mice showed increasing phage titers that peaked at 10⁹ PFU/g of stool and remained elevated until 13 dpi (6 days after the end of the treatment). After this point, phage titers declined in parallel with CR loads and were undetectable by 23 dpi (Fig. 4C). CR and phage titers in the cecum at 8 dpi were consistent with fecal levels, with CR counts averaging ∼10⁹ CFU/g of cecal contents for both infected groups, although the phage-treated group exhibited slightly lower bacterial counts. Phage titers were ∼10⁸ PFU/g in infected mice and ∼10⁶ PFU/g in the phage-only group (Fig. S4C and D).

The observed differences suggest that phage replication occurs only in the presence of CR and that oral administration of Eifel2 alone does not result in sustained colonization due to the absence of a suitable commensal host. Correlation analysis of PFU and CFU counts for each mouse across different time points revealed a strong positive relationship between phage and bacterial levels in the gut (R² = 0.9296) (Fig. 4D), indicating a highly dynamic interaction in vivo. These findings confirm that Eifel2 can infect CR in vivo, thus maintaining high titers over time, and support the establishment of a stable co-existence between Eifel2 and CR within the gut environment.

Phage–CR co-existence does not affect host–pathogen interactions

To visualize phage–CR interactions in the gut, distal colon sections were stained using the Eifel2 antiserum alongside anti-CR serum at 8 dpi. Although no phage signal was detected, comparable levels of attached CR were observed in both phage-treated and untreated groups (Fig. 5A). Moreover, evaluation of colon histology samples revealed that both infected groups, regardless of phage treatment, exhibited CCH (crypt length ~200 µm) at 8 dpi and at the time of phage clearance (21–23 dpi), whereas uninfected controls showed normal crypt lengths (∼150 µm) (Fig. 5B). As CCH is a hallmark of CR infection that reflects activation of the damage repair response and subsequent cell proliferation and crypt elongation, phage treatment does not appear to impact CR-induced barrier damage.

Fig 5.

Eifel2 does not impact CR-induced colonic damage. (A) Representative immunofluorescence images of colonic sections at 8 dpi, with CR (green), anti-Eifel2 serum (red), WGA (purple), and DAPI (blue). Scale bars: 200 µm. (B) Crypt length measurements in distal colon sections at the indicated time points. Data are shown as mean ± SEM, and each dot represents the mean crypt length for an individual mouse. Statistical analysis was performed using two-way ANOVA with Tukey’s multiple comparison test. (C) LCN-2 levels in homogenized feces at the indicated time points, measured by ELISA. Data are shown as mean ± SEM, and dots represent values for individual mice. Statistical analysis was performed using a mixed-effects model with Geisser-Greenhouse correction and Tukey’s multiple comparison test. (B, C) Statistical significance is indicated as: P < 0.05 (*); P < 0.01 (**); P < 0.001 (***); P < 0.0001 (****); non-significant comparisons are not shown.

To confirm these results, we quantified fecal levels of LCN-2, a marker of intestinal inflammation (29). At 8 dpi, LCN-2 levels were elevated in both infected groups, regardless of phage treatment, whereas phage treatment alone did not significantly alter this response (Fig. 5C). At the time of phage clearance, LCN-2 levels were comparable with pre-infection levels across all groups. These results suggest that oral administration of phage alone does not elicit an inflammatory immune response and that during stable Eifel2-CR co-existence in the gut, the phage does not affect host-pathogen interactions.

Emergence and characterization of phage resistance in vivo

We next investigated the emergence of Eifel2-resistant CR in vivo. We screened bacterial isolates from mice feces at 8 and 13–15 dpi using cross-streak assays. From each mouse, we tested up to six colonies for their susceptibility to Eifel2. No resistant colonies were detected in CR-infected, phage-untreated mice. In the phage-treated-infected group, a low frequency of resistant CR was detected, with co-housed mice carrying between 0% and 8.33% resistant bacteria at 8 dpi (Fig. 6; each group in the table represents a cage of co-housed mice). At later time points, variability between mice increased, with 63.6% resistant colonies found in a phage-treated infected group from one biological replicate, whereas others remained at 0%. Overall, resistant colonies were detected in six of the 20 mice that received phage treatment. Four of these mice were co-housed and carried resistant CR at both 8 and 15 dpi. Although the other two mice, each from separate cages, shed resistant CR at 8 dpi (Fig. 6), no resistant bacteria were recovered at 13–15 dpi. Conversely, the three mice co-housed with one of these two mice shed resistant CR at 15 dpi (Fig. 6), suggesting that resistance could have spread through coprophagic behavior. Together, this suggests that not only was the selective pressure exerted by the presence of the phage not enough to enable clonal expansion of the resistant phenotype in vivo, but also that resistance seems disadvantageous in the context of co-existence. In agreement with these observations, the presence of resistant variants was not linked to slower CR clearance.

Fig 6.

Quantification of resistant CR colonies in mouse feces.

To investigate the genetic basis of resistance, one resistant colony from each of the six mice where resistance was observed (labeled R1–R6) was sequenced and compared with the reference CR ICC168 genome (NCBI accession number: NC_013716.1) (Table S1). We identified three mutations that could be implicated in resistance: rfaJ and rfaK, both involved in lipopolysaccharide (LPS) biosynthesis (Fig. 7A), and dsbC, a gene encoding a periplasmic disulfide bond isomerase. A premature stop codon at position 307 of 1,017 bp was identified in rfaJ in one isolate (R2), whereas transposon insertions were found at different positions in rfaK in two other isolates (R1 and R6). A large deletion affecting dsbC was detected in three isolates (R3, R4, and R5), with the deletion also disrupting the downstream gene recJ. None of the sequenced isolates exhibited a significant growth defect when grown in LB for 16 h (Fig. S5A).

Fig 7.

Genetic and structural basis of CR resistance to Eifel2. (A) Schematic representation of the CR LPS structure, highlighting the roles of RfaK, RfaJ, and RfaC in structure assembly, inferred from structural homology with the E. coli type R2 rfa (waa) locus (30–36). GlcNAc: N-acetyl-D-glucosamine; P: phosphate; Kdo: 3-deoxy-D-manno-oct-2-ulosonic acid; Hep: heptose; Glc: D-glucose; Gal: D-galactose; and Rha: L-rhamnose. (B–D) As the rfaJ mutant could not be generated, the sequenced mutated fecal isolate (R2) was used, along with its complementation. (B) Representative images of DLA spot assays of Eifel2 on CR mutants and their complemented strains grown under pBAD-inducing conditions (0.2% L-arabinose). Images are representative of n = 3 independent biological replicates. (C) PFU counts of DLA spot assays of Eifel2 on CR mutants and their complemented strains grown under pBAD-inducing conditions. Values below the LoD were set to one for plotting. Data are shown as mean ± SEM of n = 3 independent biological replicates, and dots represent PFU counts for individual repeats. Statistical analysis of strains with PFU counts above the LoD was performed using a nonparametric Kruskal-Wallis test with Dunn’s multiple comparisons test. ns, non-significant. (D) Silver-stained SDS-PAGE gel of LPS extracted from each strain following 3 h of growth in pBAD-inducing conditions. (A) Created in BioRender. Frankel, G. (2025) https://BioRender.com/ycp5cc3.

To further study the spread of the resistance mutations, we performed PCRs on the 18 isolated resistant colonies (including the six that were sequenced) to detect alterations in rfaK, rfaJ, and the dsbC-recJ region (Fig. S5B). In the case of rfaK, three colonies, including R1 and R6, produced fragments consistent with the insertion of a transposon. Interestingly, R1 and R6 were isolated from separate biological replicates, indicating that these insertions were independent events. For dsbC and recJ, 14 colonies isolated from the same cage produced PCR products consistent with the deletion observed in the sequenced isolates, suggesting that they spread between animals via the fecal-oral route.

To confirm the role of these mutations in resistance, we generated clean mutants in the wild-type CR background. A stop codon mutation was introduced in rfaK, and a deletion mutant was generated for dsbC. Because the deletion affecting dsbC also disrupted the downstream gene recJ in the sequenced isolate, we also generated a stop codon mutant in recJ. Plating assays showed that mutations in rfaK and dsbC conferred full resistance to Eifel2 infection (Fig. 7B and C), with the dsbC mutant showing low levels of lysis at high phage concentrations but no plaque formation (Fig. 7B and C), similar to the phenotype observed in the sequenced isolates (Fig. S5C). Complementation of the mutants with wild-type rfaK or dsbC restored full susceptibility, confirming the role of these mutations in resistance. In contrast, the recJ mutation did not affect phage susceptibility (Fig. S5D). As attempts to generate mutants for rfaJ were unsuccessful, we complemented the original resistant R2 isolate with a plasmid carrying the wild-type rfaJ gene instead. Susceptibility of the complemented R2 isolate to Eifel2 infection was fully restored, confirming the role of the mutation in rfaJ in resistance (Fig. 7B and C). To test whether these resistance mutations affected infection by other phages, we performed plating assays with ColRes. Although rfaK and rfaJ mutants showed a slightly lower plating efficiency, none of the mutants showed efficient resistance to ColRes infection, indicating the resistance mechanisms we identified are specific to Eifel2 (Fig. S5E).

The involvement of rfaJ and rfaK in core LPS biosynthesis points toward a component of LPS being the receptor for the phage. We examined LPS profiles of mutant and complemented strains by silver staining. We used the wild-type strain and a known deep-rough mutant (ΔrfaC), which was also resistant to Eifel2 (Fig. S5F), as controls. Both rfaJ and rfaK mutants showed truncated LPS with a lack of O-antigen, agreeing with the presence of the terminal GlcNAc, added by RfaK, being essential for O-antigen ligation (30, 31), whereas the complemented strains restored normal LPS length (Fig. 7D). The dsbC deletion mutant did not show any noticeable differences in LPS structure, suggesting that resistance is mediated by a different mechanism. To rule out the O-antigen component of the LPS as a receptor for Eifel2, a CR strain lacking O-antigen was tested. The O-antigen-deficient mutant remained sensitive to Eifel2 (Fig. S5F), suggesting that although the rfaJ and rfaK mutants lack O-antigen, the resistance is due to the mutation affecting the core LPS components.

DISCUSSION

CR has extensively been used as a translational model to study EPEC and EHEC infections in vivo, given its ability to infect immunocompetent mice in the presence of the natural microbiota (12–14). Although CR-specific phages have been isolated, their use in vivo remains unexplored (15–18). Thus, the aim of this study was to isolate lytic phages infecting CR and develop an in vivo model to accurately depict phage-pathogen interactions in a physiological context.

We isolated two closely related lytic phages, Eifel1 and Eifel2, infecting CR. Genomic analysis revealed they represent a new species within the Felixounavirus genus. Phages within the Felixounavirus genus are strictly lytic and primarily target gram-negative enterobacteria, including Salmonella spp. and E. coli, and their potential as antimicrobial agents has been previously explored in vitro (37, 38). Despite almost identical genomes, Eifel2 exhibited higher lytic activity compared with Eifel1, potentially due to a structural variation in a minor tail protein, which could enhance host attachment and/or DNA injection efficiency (39, 40). An additional gene in Eifel2, located among hypothetical proteins, remains functionally uncharacterized and needs further investigations to explore its potential implications in the infection cycle of Eifel2. Interestingly, both phages displayed strict host specificity for CR, in contrast to other members of the Felixounavirus genus (41).

To visualize phage infection dynamics, we developed a simplified immunization protocol using whole phage particles. Following two rounds of immunization, the resulting serum recognized multiple virion proteins by Western blot. Compared with previous approaches such as fusion of fluorescent tags to capsid proteins (42, 43), fluorescent labeling of virions (44, 45) or phage DNA (46–49), FISH-based phage DNA detection (50, 51), plasmid-based approaches (52), and generation of capsid protein specific antibodies (53), our method provides key advantages: no genetic manipulation of phages, avoiding potential disruptions to phage infectivity; visualization of multiple infection cycles and phage progeny, not limited to initial input phages; applicability to unsequenced phages, enabling broader use and broad antigen detection and bypassing the need to purify individual phage proteins.

In the CR mouse infection model, Eifel2 showed robust replication in the gut. However, phage treatment did not significantly reduce bacterial burden or alleviate infection-associated colonic pathology. A strong positive correlation between phage and bacterial titers was observed, suggesting the establishment of dynamic co-existence rather than phage-mediated bacterial clearance. This is consistent with previous in vivo studies in murine models with different E. coli strains, which showed stable co-existence of phages and host bacteria in the gut (5, 7, 54). Those studies relied on disturbing the microbiome to allow for pathogen colonization, either by antibiotic pre-treatment to use enteroaggregative O104:H4 Escherichia coli (EAEC)^9,50, or by using mice with a synthetic microbiota like the OMM¹² mouse model, to infect with E. coli Mt1B1 and EAEC¹¹. Notably, Lourenço et al. (7) showed that co-existence was enabled by the spatial partitioning of bacteria and phages in the gut: phage amplification occurred mainly in the lumen; thus, bacteria adhering to mucosal surfaces were protected from phage predation. Upon infection, CR intimately attaches to intestinal epithelial cells (IECs) and is shed into the lumen (55) and could thus support this model of niche partitioning.

Consistently, immunofluorescence imaging using the anti-Eifel2 serum failed to detect phage signal in distal colon sections, despite high fecal titers. This could reflect low phage abundance at the mucosal surface in the colon, consistent with the hypothesis of phage replication occurring in the lumen, where free bacteria are more exposed, whereas mucosa-associated communities remain relatively protected from phage predation. The absence of signal likely reflects biological distribution rather than limitation of the antibody, which efficiently recognized multiple virion proteins in vitro.

Building on earlier studies that demonstrated phage-bacteria co-existence under perturbed conditions (e.g., antibiotic pre-treatment, gnotobiotic or synthetic microbiota models), our findings provide the first evidence that a virulent phage and its pathogenic host can establish a stable co-existence within an unperturbed, immunocompetent host. This demonstrates that such dynamics are not artifacts of microbiota disruption but intrinsic to the ecological constraints of the natural gut environment. The CR model thus provides a unique physiological context to dissect the determinants that shape the balance between bacterial growth and phage replication and, by extension, the influence and limitations imposed by the ecological context on phage therapy outcomes.

A recent study using E. coli Mt1B1-colonized OMM¹² mice showed that the emergence of phage resistance is not a major contributor to phage-bacteria co-existence in the gut (7). In our model, phage-resistant mutants emerged at low frequencies. However, these mutants did not clonally expand and spread, suggesting a significant fitness cost for phage resistance mutants in the highly selective gut environment (1, 2). Spatial segregation of phages and bacteria in the gut could further explain the low emergence frequencies of resistant isolates by limiting selection (7). Whole genome sequencing of resistant isolates revealed resistance-conferring mutations in rfaJ and rfaK, genes involved in core LPS biosynthesis (56), which resulted in truncated LPS and loss of O-antigen. These mutations have previously been described in phage resistance mechanisms in gram-negative bacteria (57–59). However, O-antigen is an important virulence factor for in vivo colonization and immune evasion in gram-negative pathogens (60), potentially explaining the low frequency and limited expansion of these mutants. A large deletion affecting dsbC, a gene encoding a periplasmic disulfide bond isomerase (61–64), was observed more frequently. dsbC mutants exhibited intact LPS and O-antigen, possibly contributing to a lower fitness cost in vivo, and thus, a more important spread of this mutation between mice within a cage. However, these mutants also displayed partial sensitivity to high concentrations of phage lysate without supporting plaque formation. Absence of DsbC indirectly affects bacterial resistance to environmental elements, such as oxidative stress (65), suggesting that CR dsbC mutants could exhibit lower resistance to certain elements present in phage lysate.

The precise mechanisms by which core LPS and dsbC mutants confer resistance to Eifel2 remain unclear but seem to be highly specific. Given DsbC’s role in protein folding, loss of function could alter membrane proteins or membrane composition, potentially reducing phage receptor availability. In the case of LPS, O-antigen-deficient strains remained sensitive, suggesting that other major LPS core modifications, such as the ones observed in rfaK and rfaJ mutants, may also lead to resistance by either disrupting receptor recognition or modifying accessibility to the actual receptor.

Taken together, this study establishes a murine model to investigate phage-bacteria interactions in vivo in the context of the complete endogenous gut microbiota and natural host immunity. Our results showed that despite strong in vitro lytic activity, oral administration of Eifel2 resulted in the emergence of a dynamic and spatially structured system that supports co-existence rather than eradication of one or the other in the gut environment. These findings demonstrate that phage therapy outcomes are highly context-dependent, shaped by spatial structure, host immunity, and microbiota interactions rather than by phage lytic capacity alone. We have shown that although resistance emerged, the resistant CR mutants did not clonally expand, indicating an associated fitness cost and underscoring that co-existence can be an ecologically stable outcome in complex gut environments. From a translational perspective, our findings highlight the importance of studying phage-bacteria interactions in physiologically relevant models that reveal the true complexity of such interactions, where ecological and immune constraints may limit therapeutic efficacy. Recognizing these limitations is essential for designing effective phage-based treatments and for selecting appropriate infection contexts in which eradication versus co-existence may be realistic goals.

MATERIALS AND METHODS

Bacterial strains and growth conditions

All bacterial strains used in this study are listed in Table S2. Strains were cultured in Lysogeny broth (LB) at 37°C with shaking at 200  rpm, or on LB agar (1.5%) plates incubated statically at 37°C. Where appropriate, antibiotics were added to the media at the following final concentrations: nalidixic acid (NAL), 50  µg/mL; gentamicin (GEN), 10  µg/mL; kanamycin (KAN), 50  µg/mL; and streptomycin (STR), 50  µg/mL.

For experiments involving complemented strains harboring the pBAD expression plasmid, cultures were grown in LB supplemented with 0.2% wt/vol glucose to repress expression. To induce gene expression, 0.2% wt/vol L-arabinose was added to the media in place of glucose.

Generation of CR mutants and complementation

Plasmids and primers used in this study are listed in Table S2. All genomic deletions and point mutations were generated using a two-step recombination protocol, resulting in scarless and markerless mutants, as previously described by (66). Genomic DNA (gDNA) from CR ICC169 was extracted using the Monarch Genomic DNA Purification Kit (New England Biolabs, NEB), following the manufacturer’s instructions. For gene deletions, ∼500 bp homology regions (HRs) flanking the gene of interest were PCR amplified using 2× Phanta Flash Master Mix (Vazyme) from CR gDNA. To generate point mutations, ∼500 bp HRs flanking the mutation site were amplified using primers incorporating a stop codon. PCR products were purified using the Monarch PCR & DNA Cleanup Kit (NEB). The pSEVA612S_revISceI vector was linearized via PCR using 2× Phanta Flash Master Mix (Vazyme) and purified using the same cleanup kit. HRs were inserted into the linearized vector using Gibson Assembly (NEB), following the manufacturer’s instructions. Assembled vectors were transformed into chemically competent E. coli CC118λpir cells and maintained with appropriate antibiotics.

To introduce the desired mutations into CR, recipient strains pre-transformed with pACBSR (plasmid expressing the I-SceI endonuclease and lambda red fragment) were used. Tri-parental conjugation was performed by incubating 20 µL of the donor strain (E. coli CC118λpir carrying the mutagenesis vector) with 20 µL of the helper strain (E. coli 1047 carrying pRK2013) on LB agar at 37°C for 2 h. Then, 40 µL of the recipient strain (CR pACBSR) was added and incubated for at least 8 h at 37°C. Transconjugants were selected on LB agar plates containing appropriate antibiotics. Growing colonies were grown in LB supplemented with L-arabinose (0.4%) and appropriate antibiotics for 2 h to induce I-SceI expression. Cultures were plated on LB agar with appropriate antibiotics, and candidate deletion mutants were screened via colony PCR using 2X Rapid Taq Master Mix (Vazyme).

For genetic complementation, genes of interest were PCR-amplified from CR gDNA, and the pBAD vector was linearized by PCR using 2× Phanta Flash Master Mix (Vazyme). PCR amplicons were purified using Monarch PCR & DNA Cleanup Kit (NEB), and genes were inserted using Gibson Assembly (NEB). Resulting plasmids were transformed into electrocompetent mutant CR pACBSR cells prepared at room temperature as previously described (67) and selected on LB agar supplemented with glucose (0.2%) and appropriate antibiotics. For rfaJ complementation, the assembled plasmid was introduced into the sequenced fecal bacterial isolate, and selection was done on LB agar with glucose (0.2%) and appropriate antibiotics. All plasmids and mutants were validated by colony PCR using 2× Rapid Taq Master Mix (Vazyme) and Sanger sequencing (Eurofins Genomics).

Phage isolation and purification

Double-layer agar (DLA) assays were performed as previously described (68), with minor modifications. Briefly, 4 mL of soft LB agar (0.3%) supplemented with 10  mM MgSO₄ and 10  mM CaCl₂ was inoculated with 100  µL of an overnight culture of the bacterial host. For DLA plate assays, the inoculated soft agar was mixed with 100  µL of phage suspension before being poured onto LB agar (1.5%) plates supplemented with appropriate antibiotics in standard 90 mm Petri dishes. For DLA spot assays, only the bacterial culture was mixed into the soft agar, poured onto LB agar (1.5%) plates, and allowed to solidify before spotting 10  µL of serially diluted phage suspensions onto the surface. Plates were incubated overnight at 37°C under static conditions. For larger square Petri dish plates, reagent volumes were doubled.

All phages isolated in this study are listed in Table S2. Sewage water samples were initially centrifuged and filtered through 0.45  µm pore-size membranes. The filtrate was added to LB supplemented with Nal, inoculated with CR, and incubated overnight at 37°C with shaking. The resulting co-cultures were centrifuged and filtered through 0.45  µm membranes to obtain crude phage lysates, which were serially diluted, and DLA plate assays were performed. For single plaque isolation and phage purification, a prophage-free CR strain (CR Δ10, a derivative of ICC169) was used in all subsequent steps. Single plaques were picked and placed into saline-magnesium (SM) buffer (100  mM NaCl, 8  mM MgSO₄, 50  mM Tris-HCl, pH 7.5) for 2  h at room temperature. This purification process was repeated three times.

High-titer phage lysates were generated either by liquid co-culture (Eifel2) or large-scale DLA plate assays (Eifel1 and ColRes). For the latter, ∼20 DLA plates were inoculated with appropriate phage dilutions to achieve near-confluent lysis (“webbed” plates). The following day, 5  mL of SM buffer was added to each plate and incubated with gentle shaking for 2  h at room temperature. The buffer was then collected, pooled, centrifuged, and filtered through 0.45  µm membranes. High-titer lysates were used to inoculate large liquid co-cultures to obtain large high-titer phage preparations for downstream purification steps. These lysates were concentrated by ultracentrifugation and purified by CsCl density gradient ultracentrifugation. Briefly, pooled lysates were centrifuged at 40,000  ×  g for 2  h, and the supernatant was discarded. This step was repeated as needed to concentrate phage particles. The resulting pellets were resuspended in 1  mL SM buffer and incubated overnight at 4°C before being pooled. CsCl gradients were prepared with layers of 1.7, 1.5, and 1.3  g/cm³ CsCl solutions in SM buffer. Gradients were ultracentrifuged at 40,000  rpm for 3  h at 5°C. Visible phage bands were extracted using a sterile needle and syringe and subsequently washed and concentrated using centrifugal concentrators (100,000  kDa MWCO, Vivaspin) with sterile SM buffer according to the manufacturer’s instructions. Final phage preparations were filtered with 0.45 µm membranes and were stored in SM buffer at 4°C.

TEM images

TEM of phage preparations was performed by applying 2–3 μL of purified phage to glow-discharged, carbon-coated copper grids (200 mesh, Agar Scientific), and then staining with 2% uranyl acetate. Micrographs of phages were captured on an FEI T12 electron microscope at an acceleration voltage of 120 kV and a nominal magnification of 26,000 ×.

Images were processed and analyzed using Zen 3.5 Blue software (Carl Zeiss MicroImaging GmbH, Germany). Phage dimensions were determined by measuring at least 20 individual virions. Measurements included capsid length and width, tail length, and neck length, and the mean values were calculated for each parameter.

Phage lysis curves

Overnight cultures of CR_WT were diluted in LB supplemented with Nal to reach an initial OD₆₀₀ of 0.1. Dilutions of purified phage suspensions were prepared in SM buffer and added to the diluted CR cultures to achieve the indicated MOIs. Control wells received SM buffer without phage. Each culture was loaded into a 96-well flat-bottom microplate in technical triplicates, and OD₆₀₀ readings were recorded every 15 min over 16 h using a FLUOstar Omega microplate reader (BMG Biotech) at 37 °C with continuous shaking.

Bacterial growth curves

Overnight cultures of bacterial strains were diluted in LB supplemented with appropriate antibiotics to reach an initial OD₆₀₀ of 0.1. Cultures were aliquoted into a 96-well flat-bottom microplate in technical duplicates, and OD₆₀₀ measurements were recorded every 15 min for 16 h using a FLUOstar Omega microplate reader (BMG Biotech) at 37°C with continuous shaking.

One-step growth curve assay

One-step growth curves were performed using a modified protocol based on (69). Briefly, CR cultures were grown to mid-log phase (OD₆₀₀  = 0.2) and infected with purified phage suspensions at an MOI of 1. Cultures were incubated at 37°C with shaking for 5  min to allow phage adsorption. Following adsorption, cultures were centrifuged to remove unbound phages, and the pellets were resuspended in fresh LB supplemented with Nal. The infected cultures were then incubated at 37°C with shaking. Samples were collected at the indicated time points, immediately serially diluted in SM buffer, and phage titers were determined using DLA spot assays.

Phage DNA extraction and sequencing

Phage DNA was extracted using the MasterPure Complete DNA and RNA Purification Kit (Lucigen) following an optimized protocol adapted from the “DNA Purification Protocols-Fluid Samples” section of the manufacturer’s manual. Briefly, 150  µL of high-titer filter-sterilized purified phage lysate was mixed with 150 µL of 2× Tissue and Cell Lysis Solution supplemented with 1 µL RNase A (5  µg/mL) and 1 µL RNase-free DNase I (1 U/µL). Samples were incubated at 37°C for 30  min to remove contaminating DNA and RNA. To inactivate nucleases, 20 µL of 0.5 M EDTA was added, followed by 1 µL of Proteinase K, and samples were incubated at 65°C for 15 min, vortexing every 5 min. Samples were then cooled on ice for 5 min before the addition of 150  µL of MPC Protein Precipitation Reagent. Tubes were vortexed and centrifuged at 4°C. The supernatant was transferred to a clean microcentrifuge tube and mixed with 500 µL of isopropanol. DNA was pelleted by centrifugation at 4°C, and the isopropanol was carefully decanted. Pellets were washed twice with 70% ethanol, ensuring not to dislodge the pellet, and residual ethanol was removed completely. Purified DNA was resuspended in 15 µL of Elution Buffer (EB) (Qiagen) and submitted for whole genome sequencing through the Enhanced Genome Service provided by MicrobesNG, following their standard submission protocols. Sequencing was performed using a combination of Illumina short-read and Oxford Nanopore long-read technologies.

Assembly, annotation, visualization, and alignment of phage genomes

Prior to genome assembly, host-derived (Citrobacter rodentium ICC168, NC_013716.1) sequence contamination was removed from raw reads using bbduk (BBTools v39.13). Decontaminated reads were assembled using SPAdes genome assembler (70, 71) (v4.0.0, hybrid assembly using --nanopore option and --isolate options, default options were used for the other parameters) and annotation of assembled contigs was performed using Pharokka (72) (v1.7.4, default options with the addition of --dnaapler to reorder the genome on the terL gene) and Phold (v0.2.0, using the run module with the --ultra-sensitivity option) (https://github.com/gbouras13/phold). Assemblies and annotations were visualized and curated using Geneious Prime 2025.1.2 (GraphPad Software LLC d.b.a Geneious).

Taxonomic classification of phage genomes was carried out using the web-based taxMyPhage tool (22). Phylogenetic relationships between species within the same genus were inferred using VICTOR and visualized as taxonomic trees. The entire analysis was carried out by the VICTOR web service (https://victor.dsmz.de/), a method for the genome-based phylogeny and classification of prokaryotic viruses (23). All pairwise comparisons of the nucleotide sequences were conducted using the Genome-BLAST Distance Phylogeny (GBDP) method (73) under settings recommended for prokaryotic viruses (23). The resulting intergenomic distances were used to infer a balanced minimum evolution tree with branch support via FASTME, including SPR postprocessing (74) for each of the formulas D0, D4, and D6, respectively. Branch support was inferred from 100 pseudo-bootstrap replicates each. Trees were rooted at the midpoint (75) and visualized with ggtree (76). Taxon boundaries at the species, genus, and family levels were estimated with the OPTSIL program (77), the recommended clustering thresholds (23), and an F value (fraction of links required for cluster fusion) of 0.5 (73). The online PhageAI tool (19) was used to predict the lifestyle of phages.

Whole genome alignments were performed with the progressive Mauve algorithm (20) and visualized in Geneious Prime 2025.1.2 (GraphPad Software LLC d.b.a Geneious), and whole genome comparisons were visualized using Easyfig (21). Protein structural predictions were performed using the web version of AlphaFold 3 (24). Resulting models were visualized and analyzed using UCSF ChimeraX (v1.9) (78). Prediction of depolymerase activity for the unique Eifel2-encoded putative protein and the Eifel1 and Eifel2 minor tail proteins was performed using the web version of DePP (v1.0.0) (26) with default parameters. The web version of DepoScope (27) was used to screen the Eifel1 and Eifel2 genomes for putative depolymerase-encoding genes using default parameters.

Phage pH and temperature stability

To assess temperature stability, purified phage suspensions in SM buffer were incubated for 24 h at the indicated temperatures. At the designated time points, the samples were collected and PFUs were determined using a DLA spot assay. To assess phage stability upon freezing, multiple aliquots of the same phage suspension were stored at –20°C to avoid repeated thawing and refreezing. PFUs were determined after thawing a single-use aliquot. To assess pH stability, purified phage suspensions in SM buffer were diluted 1:100 in SM buffer adjusted to the indicated pH values and incubated at 37°C for 24 h in a static incubator. Following incubation, PFUs were quantified using DLA spot assays.

Mouse experiments

Mouse experiments were performed in accordance with the Animals Scientific Procedures Act of 1986 (79) and UK Home Office guidelines and were approved by the Imperial College Animal Welfare and Ethical Review Body. Experiments were designed in agreement with the ARRIVE guidelines (80) for the reporting and execution of animal experiments. License number PPL PP7392693.

Specific-pathogen-free C57BL/6 female mice (18–20 g; 6–8 weeks) and pathogen-free CD-1 female mice (29–31 g; 5–7 weeks) were purchased from Charles River Laboratories and housed in groups of 2–5 individuals in high-efficiency particulate air (HEPA)-filtered cages with bedding, nesting, and free access to food and water. The temperature, humidity, and light cycles were kept within the UK Home Office code of practice, with the temperature between 20°C and 24°C, the room humidity at 45%–65%, and a 12 h/12 h light cycle with a 30 min dawn and dusk period to provide a gradual change. We used female mice only to avoid the logistical issues and stress to animals associated with male in-cage fighting. For each experiment, mice were randomly assigned to experimental groups. Investigators were not blind to the allocation. Treatments and measurements were always performed following the same cage order.

Mouse infections and phage treatment

ICC169 (CR_WT) was cultured overnight in LB supplemented with Nal. The following day, saturated bacterial cultures were centrifuged, and pellets were resuspended in 1.5 mL of sterile phosphate-buffered saline (PBS). C57BL/6 mice were infected by oral gavage with 200 µL of the bacterial suspension (∼1  ×  10⁹ CFUs) or mock-infected with 200 µL of sterile PBS. The inoculum dose was retrospectively confirmed by serial dilution and CFU enumeration on LB agar supplemented with Nal.

On days 3, 4, 5, 6, and 7 post-infection, mice received 200 µL of a sterile purified phage Eifel2 suspension (1 × 10⁹ PFUs) in SM buffer by oral gavage. Control animals received 200 µL of sterile SM buffer. To prevent phage adsorption to mouse feed and increase stomach pH, thus enhancing phage treatment efficacy, food was withdrawn 3 h prior to phage administration and reinstated immediately afterward. Body weight and bacterial burden in feces were monitored throughout the infection as previously described (14). Briefly, fecal pellets were collected on indicated days, weighed, and homogenized in PBS (0.1 g of stool/mL of PBS) using a vortex mixer for 15 min at room temperature. Homogenates were serially diluted and plated on LB agar supplemented with Nal to determine CFU counts. For phage quantification, homogenized fecal samples were centrifuged and filtered through 0.45 µm pore-size membranes. Filtrates were serially diluted and spotted using a DLA-spot assay as described above.

At 8 dpi, mice were culled by cervical dislocation. At 21 or 23 dpi, mice were weighed and anesthetized with ketamine (100 mg/kg) and medetomidine (1 mg/kg) administered via intraperitoneal injection using a 27G, 13 mm needle (BD microlance). Once mice had lost pedal reflexes, whole blood was collected by cardiac puncture (25G, 16 mm needles, BD Microlance) and transferred to a serum collection tube containing a clot activator and separating gel (BD Microtainer SST, BD). Mice were then euthanized by cervical dislocation, and cecal contents and distal colon sections were harvested for downstream analyses. CFU and PFU counts in cecal contents were determined using the same procedures described above for fecal samples.

Mouse immunization

CD-1 mice were immunized via subcutaneous injection with 2.2 × 10⁸ PFUs of sterile, purified Eifel2 phage particles in 220 µL of SM buffer, supplemented with 30 µL of alum adjuvant (Alum Alhydrogel adjuvant 2%). Control mice received 220 µL of sterile SM buffer with 30 µL of alum. A booster injection was administered 21 days after the initial immunization. Body weight loss of mice was monitored throughout the experiment. At 21 days post-booster injection, ∼30 µL of blood was collected via tail vein sampling and allowed to clot at room temperature to assess serum reactivity to Eifel2. At 22 days post-booster injection, mice were first anesthetized as described above. Whole blood was then collected by cardiac puncture as above for subsequent serum processing and use.

Isolation of resistant colonies from faecal samples and phage host range determination

Resistance of fecal bacterial isolates and the host range of isolated phages were assessed using cross-streak assays. Colonies that grew from plated fecal samples were picked and cultured overnight in LB supplemented with Nal at 37°C with shaking. Bacterial strains for host range determination were cultured overnight in LB at 37°C with shaking. To perform cross-streak assays, 50  µL of purified high-titer phage suspension was streaked in a single horizontal line across a LB agar square plate and allowed to dry. Then, 10 µL of an overnight bacterial culture was streaked perpendicularly to the phage streak. Plates were incubated overnight at 37°C under static conditions. The following day, bacterial lysis at the intersection of the bacterial and phage streaks was recorded as an indicator of phage sensitivity.

Whole genome sequencing of resistant isolates and mutation identification

Resistant bacterial colonies isolated from fecal samples were submitted for whole genome sequencing through the Enhanced Genome Service provided by MicrobesNG, following their standard submission protocols. Sequencing was performed using a combination of Illumina short-read and Oxford Nanopore long-read technologies.

SNPs were identified using Snippy (https://github.com/tseemann/snippy), whereas larger genomic rearrangements and insertions/deletions were detected using Pilon (81). Both analyses were carried out using the Galaxy Europe platform (https://usegalaxy.eu) (82). To validate mutations or assess the integrity of target genes across all isolated resistant colonies, PCR amplification of the rfaK, rfaJ, and dsbC/recJ regions was performed using 2X Rapid Taq Master Mix (Vazyme) with appropriate primers annealing immediately upstream or downstream of each gene or region. Amplified products were resolved on 1% agarose Tris-Acetate-EDTA (TAE) gels containing SYBR Safe DNA Gel Stain (1:10,000) and electrophoresed under standard conditions. Gels were imaged using a Chemidoc imaging system (Bio-Rad) and analyzed using ImageLab (v6.1) (Bio-Rad).

LCN-2 ELISA

Mouse fecal pellets were weighed and homogenized in PBS (0.1 g of stool/1 mL of PBS) using a vortex mixer for 15 min at room temperature. Samples were centrifuged to remove debris, and the supernatants were collected and stored at −80 °C. Fecal LCN-2 concentrations were measured using a DuoSet Mouse Lipocalin-2 ELISA kit (Bio-Techne), following the manufacturer’s instructions. Absorbance readings were taken with a FLUOstar Omega microplate reader (BMG Labtech).

Serum collection and processing, and anti-Eifel2 IgG ELISA

Mice were anesthetized, and blood was collected via cardiac puncture as described above. Blood was left to clot for 1 h at room temperature, the serum was subsequently separated by centrifugation at 20,000 × g for 3 min and stored in aliquots at −80°C until use or further analysis.

For the quantification of anti-Eifel2 IgG levels in serum, ELISA clear flat-bottom Maxisorp immune 96-well plates (Thermo Fisher) were coated overnight at 4°C with either 100  µL/well of purified Eifel2 phage suspension (1  ×  10¹⁰ PFU/mL) for sample wells or 100  µL/well of anti-mouse IgG antibody (2  µg/mL) for standard wells. Plates were washed three times with PBS containing 0.1% Tween 20 (PBS-T) and then blocked for 1  h at room temperature with PBS-T supplemented with 5% bovine serum albumin (BSA). Serum samples and serial dilutions of purified mouse IgG as the standard were added to the appropriate wells and incubated overnight at 4°C. Plates were washed as described above and incubated with horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG antibody (1:2000; Jackson ImmunoResearch) for 2  h at room temperature. Plates were washed as described above, and detection was carried out by adding 100  µL/well of 3,3′,5,5′-tetramethylbenzidine (TMB) substrate. The reaction was stopped after 5–10  min with 2N orthophosphoric acid.

Absorbance was measured at 450  nm and at 540  nm using a FLUOstar Omega microplate reader (BMG Labtech). Anti-Eifel2 IgG concentrations were determined by comparison to the IgG standard curve.

Western blot and dot blot

For western blotting, purified high-titer Eifel2 lysate (10¹¹ PFU/mL) was mixed with 5× SDS loading buffer containing 0.1 M dithiothreitol (DTT), boiled at 95°C for 5 min, vortexed, then boiled again at 95°C for 5 min and vortexed again before centrifugation. Samples were resolved on a 12% SDS-PAGE gel under standard electrophoresis conditions and transferred onto pre-activated PVDF membranes using a semi-dry transfer system (Bio-Rad). Membranes were blocked for at least 4 h at room temperature in 10% skimmed milk in PBS-T, then incubated overnight at 4°C with polyclonal anti-Eifel2 mouse serum (1:500) in 5% milk PBS-T. Following incubation, membranes were washed three times (10 min each) in PBS-T and incubated for 1 h at room temperature with HRP-conjugated goat anti-mouse IgG antibody (1:5,000) in 5% milk PBS-T. After three additional PBS-T washes (10 min each), the membranes were developed using ECL substrate (Bio-Rad).

For dot blotting, 5 µL of purified phage suspensions (1 × 10⁹ PFU/mL) or 5 µL of filtered (0.45  µm membrane) CR_WT overnight culture supernatant were spotted onto dry nitrocellulose membranes. Once dry, membranes were blocked and processed identically to Western blot membranes, using the same antibody concentrations and incubation conditions. Blots were imaged using a Chemidoc imaging system (Bio-Rad) and were analyzed using ImageLab (v6.1) (Bio-Rad).

Phage in vitro staining

In total, 1 mL of the CR Δ10 strain constitutively expressing GFP overnight culture was harvested by centrifugation, and the bacterial pellet was resuspended in 1  mL of purified high-titer Eifel2 lysate (10¹⁰ PFU/mL). Samples were incubated for 5 min at 37°C with shaking to allow phage binding. Cells were pelleted again and resuspended in PBS with 1% paraformaldehyde (PFA) for 30 min on ice to fix. Following fixation, cells were washed twice with PBS and blocked by resuspension in PBS with 3% BSA for 1 h at room temperature with continuous agitation. After blocking, samples were pelleted and resuspended in the generated polyclonal anti-Eifel2 mouse serum (1:50 dilution in PBS with 1% BSA) and incubated overnight at 4 °C with continuous agitation. Cells were then washed twice in PBS and incubated with the secondary antibody (1:500) and DAPI (1:1,000) (time course infection only) in PBS with 1% BSA for 1 h at room temperature in the dark with continuous agitation. After two final PBS washes, the final cell pellet was resuspended in 50 µL of PBS and mounted on microscopy slides using agarose pads. Details about the reagents and antibodies used can be found in Table S2.

For time-course imaging of phage infection, the CR Δ10 strain constitutively expressing GFP was infected with phage Eifel2 following the protocol described above for one-step growth curve experiments (one-step growth curve assay section). At defined time points post-infection, the samples were collected and processed for imaging using the in vitro staining protocol described above. Images were acquired using a Zeiss AxioVision Z1 microscope with a Hamamatsu digital camera. Images were processed and analyzed using Zen (v3.5) Blue software (Carl Zeiss MicroImaging GmbH, Germany).

Histological analysis and immunofluorescence staining

The distal 0.5 cm of the colon was harvested and fixed in 4% paraformaldehyde (PFA) for 2 h at room temperature, then transferred to 70% ethanol for storage. Tissues were processed, paraffin-embedded, and sectioned at 5 µm. Sections were stained with either hematoxylin and eosin (H&E) or processed for immunofluorescence. Crypt hyperplasia was quantified by measuring the lengths of at least 20 well-oriented colonic crypts per mouse from H&E-stained sections. The mean crypt length of each mouse was used for statistical analysis.

For immunofluorescence staining, paraffin-embedded sections were dewaxed and rehydrated by sequential submersion in Histo-Clear solution (2 × 10  min), 100% ethanol (2 × 10  min), 95% ethanol (2 × 10  min), 80% ethanol (2 × 3  min), and PBS containing 0.1% Tween 20 and 0.1% saponin (PBS-TS; 2 × 3  min). Antigen retrieval was performed by heating the slides for 30 min in demasking solution (0.3% trisodium citrate and 0.05% Tween 20 in distilled water, pH 6.0). Slides were cooled and then blocked in PBS-TS supplemented with 10% normal donkey serum (NDS) for at least 6 h in a humid chamber at room temperature. Sections were incubated overnight at 4°C with primary antibodies diluted in PBS-TS containing 10% NDS. The following primary antibodies were used: rabbit polyclonal anti-CR serum (1:50) and mouse polyclonal anti-Eifel2 serum generated in this study (1:10). Slides were washed twice in PBS-TS (10 min each), followed by incubation with secondary antibodies (1:100), fluorescent wheat germ agglutinin (WGA) (1:200), and DAPI (1:1,000). After two final PBS-TS washes (10 min each), slides were mounted using ProLong Gold Antifade Mountant (Thermo Fisher Scientific) and allowed to dry in the dark. Details about the reagents and antibodies used can be found in Table S2.

Images were acquired using a Zeiss AxioVision Z1 microscope. Fluorescence images were acquired using a Hamamatsu digital camera. Brightfield images of H&E-stained sections were acquired with an Axiocam 105 color camera. Images were processed and analyzed using Zen (v3.5) Blue software (Carl Zeiss MicroImaging GmbH, Germany).

LPS extraction and silver staining

Overnight bacterial cultures were grown in LB supplemented with appropriate antibiotics and 0.2% glucose. The following day, cultures were diluted 1:100 into fresh LB containing the same antibiotics and 0.2% L-arabinose to induce gene expression from the pBAD promoter. Cultures were incubated at 37°C with shaking for 3 h. Cultures were then normalized to an OD₆₀₀ of 1.2 in 1 mL volumes and harvested by centrifugation. Supernatants were discarded, and cell pellets were resuspended in 200 µL of 1× Laemmli sample buffer (Bio-Rad), vortexed, and boiled for 5 min. Following heat treatment, 5 µL of proteinase K (20 mg/mL) was added to each sample. Samples were then incubated at 60°C for 2 h. After incubation, 10 µL of 2-mercaptoethanol was added, and samples were vortexed and boiled again for 5 min. Lysates were centrifuged, and the resulting supernatants were loaded onto a 12% SDS-PAGE gel for electrophoresis under standard conditions.

Silver staining was performed using the SilverQuest Silver Staining Kit (Invitrogen) with an adapted protocol. Gels were rinsed in Milli-Q (MQ) water for 5 min immediately after electrophoresis. Gels were then fixed overnight in 40% isopropanol and 5% acetic acid in MQ water. Fixed gels were washed twice with MQ water (5 min each) and oxidized for 5 min in a freshly prepared oxidizing solution containing 0.7% periodic acid, 40% isopropanol, and 5% acetic acid in MQ water. Following oxidation, gels were washed twice in MQ water (15 min each) and then washed in 30% ethanol for 10 min. Gels were sensitized in a solution composed of 30 mL ethanol, 10 mL kit-provided Sensitizer, and 60 mL MQ water for 10 min. After sensitization, the gels were washed once in 30% ethanol (10 min) and once in MQ water (10 min). Staining was carried out by incubating the gels in 100 mL of staining solution (1  mL kit-provided Stainer in 99  mL MQ water) for 15 min. Gels were developed in 100 mL of developing solution (10 mL kit-provided developer, 1 drop of enhancer, and 90 mL MQ water) until LPS bands became visible. Development was stopped by directly adding 10 mL of the kit-provided stopper into the staining container and gently agitating the gel for 10 min. Gels were then washed in MQ water for 10 min.

Images were acquired using a Chemidoc imaging system (Bio-Rad) and were analyzed using ImageLab (v6.1) (Bio-Rad).

Data analysis and visualization

No statistical methods were used to predetermine sample size. For ELISAs, two technical replicates were used to calculate experimental means. Statistical significance between two normally distributed groups was performed by Welch’s t-test; when two groups were not normally distributed but were lognormally distributed, a lognormal Welch’s t-test was used. When data were not normally or lognormally distributed (based on D’Agostino-Pearson or Shapiro-Wilk normality tests), a nonparametric Mann-Whitney test was applied. For comparisons involving more than two normally distributed groups, one-way analysis of variance (ANOVA) was used. When assessing more than two normally distributed groups across different time points, a two-way ANOVA (for independent measures) or a mixed-effects model (for repeated measures) was performed. When data were not normally distributed (based on D’Agostino-Pearson or Shapiro-Wilk normality tests), a logarithmic transformation was applied, and the data were then analyzed by parametric statistical methods (ANOVA). If normality was not achieved by transformation, a nonparametric test was applied (Kruskal-Wallis test followed by Dunn’s multiple comparisons test). Multiple comparisons were corrected using either Dunnett’s or Tukey’s post hoc test, as specified in the figure legends. Statistical analyses were only performed on data sets with a minimum of three or more biological replicates per group. Data visualization and the analyses were performed using Prism (v10.4.1) (GraphPad Software, LLC). A P value < 0.05 was considered statistically significant. Statistical tests used for each experiment are detailed in the corresponding figure legends. Schematics in Fig. 3A, 4A, and 7A were created in BioRender. Figure 3A: Frankel, G. (2025) https://BioRender.com/nojbg9o. Figure 4A: Frankel, G. (2025) https://BioRender.com/q1nlu57. Figure 7A: Frankel, G. (2025) https://BioRender.com/ycp5cc3.

DATA AVAILABILITY

The ColRes, Eifel1, and Eifel2 bacteriophage genome sequences are available in GenBank under accession numbers PX596622, PX596620, and PX596621, respectively. Assembled genomes are available at https://doi.org/10.5281/zenodo.15724312. All the other data are contained within the main text and supplemental material.

ETHICS APPROVAL

Mouse experiments were approved by the Imperial College Animal Welfare and Ethical Review Body.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mbio.01944-25.

Supplemental Figures. mbio.01944-25-s0001.pdf.

Figures S1 to S5.

Table S1. mbio.01944-25-s0002.xlsx.

SNP and variant detection in resistant colonies isolated from mouse feces.

Table S2. mbio.01944-25-s0003.xlsx.

Materials.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.