Phage P100 resistance in clinical and foodborne Listeria monocytogenes isolates is associated with adsorption-inhibiting mutations and fitness trade-offs

TL;DR

The study shows how Listeria monocytogenes strains resist the P100 phage through mutations in cell wall genes, leading to reduced phage binding and fitness trade-offs.

Contribution

Identifies serovar-specific genetic mutations linked to phage resistance and reveals fitness trade-offs in Listeria monocytogenes.

Findings

Mutations in WTA glycosylation genes impair P100 phage binding in Listeria monocytogenes.

Serovar 4b-derived mutants show greater fitness loss compared to 2a or 2b.

Phage-resistant mutants exhibit increased antibiotic sensitivity and impaired growth under certain conditions.

Abstract

•Most mutations occurred in genes associated with WTA glycosylation.•All mutants exhibited impaired P100 phage binding.•Gene mutation patterns appeared serovar-dependent.•Serovar 4b-derived mutants showed greater fitness loss than 2a or 2b.•Phage-resistance induction led to increased antibiotics sensitivity.

Most mutations occurred in genes associated with WTA glycosylation.

All mutants exhibited impaired P100 phage binding.

Gene mutation patterns appeared serovar-dependent.

Serovar 4b-derived mutants showed greater fitness loss than 2a or 2b.

Phage-resistance induction led to increased antibiotics sensitivity.

Listeria monocytogenes is a common bacterial pathogen causing listeriosis. Phage-based products are increasingly used in food safety, but their use raises concerns about the emergence of phage-resistant bacteria. The broad-host-range phage P100 is applied in foods, however, little is known about the genetic basis of resistance and associated fitness trade-offs in L. monocytogenes.

In this study, recently (2014–2020) circulating, clinical and food-associated L. monocytogenes strains (n = 14) were exposed to a commercial P100 phage preparation in soft agar at 20 °C and 10 °C. Eight mutants were isolated and confirmed as phage-resistant by spot assays and flow cytometry. Whole-genome sequencing revealed that most mutants derived from serovar 4b strains carried mutations in genes for polysaccharide synthesis indirectly involved in cell wall teichoic acid glycosylation, whereas mutants from serovar 2a or 2b strains had mutations in a gene directly involved in WTA glycosylation. Confocal microscopy and flow cytometry confirmed reduced phage adsorption in all mutants. The resistant mutants displayed increased antibiotic sensitivity, with two showing pronounced sensitivity to β-lactam antibiotics. These two mutants also showed impaired growth at 37 °C. Together with four additional mutants, growth was also impaired at 10 °C, both without and with sodium chloride and sodium nitrite. Overall, serovar 4b-derived mutants exhibited stronger fitness trade-offs than those of serovar 2a or 2b

The findings of this study contribute to the understanding of the P100 phage-resistance mechanisms in different serovar strains of L. monocytogenes and reveal associated fitness trade-offs that may be relevant for future antimicrobial and biocontrol strategies.

Image, graphical abstract

Figures

Validation of P100 phage resistance in *L. monocytogenes* mutants.(A) Wild-types (blue bars) and mutants generated at 20 °C (red bars) or 10 °C (pink bars) were incubated with P100 phage (hatched bars) or without phage (controls). Viable cell counts were determined at 0, 3, 24, and 48 h of incubation at 20 °C using flow cytometric live staining. Data are presented as mean ± standard deviation from six (wild types and 20 °C mutants) or three (10 °C mutants) biological replicates. A two-tailed Student’s *t*-test was performed of the replicates for each isolate to test for statistical significanc

Illumina sequencing results of all eight P100 phage-resistant mutant strains generated from *L. monocytogenes* serovar 2a, 2b and 4b strains in soft agar at 20 °C and 10 °C.(A) Overview of the total number of mutations for each mutant. (B) Overview of mutations in four genes of interest. Gene mutations are shown in boxes; additionally, predicted protein level effects are indicated. DNA and protein changes follow Human Genome Variation Society (HGVS) nomenclature. c., coding DNA; p., protein; del, deletion; ins, insertion; *, stop-gained; dup, duplication; delins, deletion-insertion; UGPase, UT

Sequence conservation and 3D structure of phosphoglucomutase (PGM) of *L. monocytogenes* strain 475.(A) Multi-sequence alignment of PGM homologs of *L. monocytogenes* strain 475 and other bacteria: Rth – *Ruminiclostridium thermocellum* ([YP_001037690](YP_001037690)), Cdi – *Clostridioides difficile* ([YP_001089293](YP_001089293)), Fpr – *Faecalibacterium prausnitzii* ([ZP_02092750](ZP_02092750)), Tpa – *Treponema pallidum* (3322,941), Mxa – *Myxococcus xanthus* ([ABF88119](ABF88119)), Tle – *Thermotoga lettingae* ([YP_001470567](YP_001470567)), Pae – *Pseudomonas aeruginosa* (WP_410,924,565.1

Adsorption of fluorescence-labeled P100 phages to wild-type and mutant strains of *L. monocytogenes*. Bacteria were incubated with Atto488-labeled phages, followed by fixation and DAPI counterstaining.(A) Confocal laser scanning microscopy (CLSM) images of wild-type strain 446 and its respective mutant (generated at 20 °C), showing Atto488-labeled phages (green dots) associated with bacterial cells (magenta). (B) Quantification of wild-type and mutant strains associated with Atto488-labeled phages based on CLSM. Bacterial cell counts were obtained from 2–9 fields of view. Data are shown as mea

Growth kinetics of *L. monocytogenes* wild-type and mutant strains at 37 °C and under food-relevant preservation conditions (10 °C with and without sodium chloride and/or sodium nitrite).(A) Bacterial cultures were grown in BHI broth at 37 °C with shaking, and optical density at 600 nm (OD_600_) was measured every 30 min over 16 h. Growth performance was quantified as the area under the curve (AUC). Data are presented as mean ± standard deviation from six biological replicates. A two-tailed Student’s *t*-test was performed of the replicates for each isolate to test for statistical significance

Full text

Introduction

1

Listeria monocytogenes is a common bacterial pathogen that frequently causes outbreaks of foodborne illness and product recalls. In 2024, 27 European Union Member States reported 3041 confirmed invasive cases of human listeriosis, corresponding to a notification rate of 0.69 cases per 100,000 population (European Food Safety Authority 2025).

Consumption of contaminated foods can lead to listeriosis, ranging from mild gastroenteritis to severe, life-threatening disease (Koopmans et al. 2023). In 2024, listeriosis was the most severe zoonotic disease in the European Union, showing the highest case fatality and hospitalization rates among reported infections (European Food Safety Authority 2025). Listeria monocytogenes is currently divided into 14 serovars (Kathariou 2002; Doumith et al. 2004; Yin et al. 2019) with at least 95% of strains isolated from food and patients belonging to serovars 1/2a, 1/2b, and 4b (Kathariou 2002). Serovar classification of L. monocytogenes relies on the structural differences in glycosylated cell wall teichoic acids (WTA) and flagellum proteins (Seeliger 1975). Wall teichoic acids are built from repeated ribitol phosphate subunits which are covalently linked to the peptidoglycan (Uchikawa et al. 1986; Fiedler 1988). The glycosylation patterns are serovar-dependent and in serovar 1/2, N-acetylglucosamine (GlcNAc) and rhamnose decorate the ribitol residues of the ribitol phosphate monomers (Kamisango et al. 1983). In contrast, serovar 4b differs markedly as GlcNAc residues are incorporated into the ribitol phosphate backbone and are further decorated with glucose and galactose (Uchikawa et al. 1986; Fiedler 1988).

Glycosylation of WTAs confers multiple advantages to L. monocytogenes, e.g., virulence (Orndorff 2016; Spears et al. 2016; Sumrall et al. 2019; Monteiro et al. 2025), resistance to antimicrobial peptides (Carvalho et al. 2015) and cold tolerance (Chassaing and Auvray 2007).

However, glycosylation also confers susceptibility to bacteriophages (phages), since glycosylated WTAs serve as ligands for phage adsorption, as shown for serovar 2 (Tokman et al. 2016; Brown et al. 2021; Denes et al. 2015; Eugster et al. 2015; Trudelle et al. 2019, 2022) and 4b strains (Sumrall et al. 2019).

The application of phages offer multiple advantages as biocontrol agents in food settings: they are a feasible option for preventing and/or eradicating harmful bacteria with high specificity in various foods and food processing environments (Ranveer et al. 2024). However, bacteria can evade phage predation by developing a range of anti-phage defense systems that target different stages of the phage life cycle (Seed 2015; Hampton et al. 2020; Georjon and Bernheim 2023). A common resistance mechanism involves preventing phage adsorption through the loss, alteration or masking of bacterial surface receptors (Rostøl and Marraffini 2019).

The P100 phage, used for the biocontrol of L. monocytogenes, is well-characterized and has been shown to be effective in several studies (Carlton et al. 2005; Guenther et al. 2009; Soni et al. 2014; Fister et al. 2016; Gutiérrez et al. 2017; European Food Safety Authority 2016). The emergence of phage-insensitive L. monocytogenes has not been observed under experimental conditions when the P100 phage product was applied to food matrices artificially contaminated with L. monocytogenes (Carlton et al. 2005; Guenther et al. 2009; Chibeu et al. 2013; Lewis et al. 2019). Nevertheless, when Fister et al. (2016) screened 486 L. monocytogenes isolates obtained from 59 dairies over a 15-year period for P100-insensitive variants, 2.7% demonstrated resistance to P100. Their appearance did not occur randomly, but rather in association with phage treatment. The authors concluded that the observation of insensitive L. monocytogenes was likely linked to the application of the phage, rather than to incidental introduction into the production facilities (Fister et al. 2016).

This study employed an in vitro model in which recently circulating, clinical and food-associated L. monocytogenes strains of serovars 2a, 2b, 2c and 4b were exposed to a P100 phage-based product in semi-solid agar at 20 °C and 10 °C. Resistant mutants and their parental wild-type strains were analyzed by whole-genome sequencing to identify genes associated with resistance to the P100 phage. Phage binding in mutants and wild-type strains was examined using confocal laser scanning microscopy and flow cytometry. To assess potential fitness trade-offs, mutant clones were further characterized with respect to their sensitivity to various antibiotics, their growth ability at 37 °C, and their growth at 10 °C with and without the food preservatives sodium chloride (NaCl) and sodium nitrite (NaNO_2_).

Materials and methods

2

Microorganisms and growth conditions

2.1

The clinical and food-associated L. monocytogenes strains used in this study are listed in Table 1. The genomes of all strains were previously sequenced and were genetically distinct (Lüth et al. 2020; Halbedel et al. 2024). The clinical isolates represented highly prevalent sequence types (ST) that have been associated with listeriosis cases in Germany, as determined by multi-locus sequence typing (Halbedel et al. 2024). For all experiments, the commercial phage preparation Phageguard L (formerly Listex; Micreos Food Safety B.V., Wageningen, The Netherlands), targeting L. monocytogenes, was used. According to the manufacturer, it contains phage P100 at a concentration of 2 × 10¹¹ plaque-forming units (PFU)/mL.Table 1. Clinical and food-associated L. monocytogenes isolates used in this study.Table 1 dummy alt textIsolate no.RKI/BfR no.OriginSero-varReferenceabcd47014–03633Stool2b(Lüth et al. 2020)yes1 of 3no0 of 1447316–00461Stool2a(Halbedel et al. 2018)yes0 of 3ntnt47416–02860Stool2a(Halbedel et al. 2018)noN/Antnt47117–05075Stool2a(Lachmann et al. 2022)noN/Antnt47217–01398Stool2c(Lachmann et al. 2022)noN/Antnt47518–00792Stool4b(Halbedel et al. 2024)yes1 of 3yes3 of 343919-LI01475–0Teewurst2cN/AnoN/Antnt44020-LI00049–0Smoked Bratwurst4bN/AnoN/Antnt44120-LI00052–0Smoked Bratwurst2aN/AnoN/Antnt44220-LI00069–0Mettwurst, finely minced4bN/Ayes3 of 3yes3 of 344320-LI00075–0Bauernbratwurst2aN/Ayes1 of 3no0 of 1444420-LI00373–0Rohwurst schnittfest2aN/AnoN/Antnt44520-LI00412–0Landjäger2aN/AnoN/Antnt44620-LI00449–0Onion sausage4bN/Ayes2 of 3yes3 of 9a, single-colony development in softagar with phage at 20 °C; b and d, x of y picked colonies showed growth in BHI broth with phage; c, single-colony development in softagar with phage at 10 °C; N/A, not applicable; nt, not tested

Bacterial strains were routinely cultured in Brain Heart Infusion (BHI) broth and on BHI agar (Carl Roth, Karlsruhe, Germany). For long-term storage, cultures grown for 18 h at 20 °C (food-associated isolates) or 37 °C (clinical isolates) were supplemented with 20% glycerol and stored at −80 °C. For propagation, strains were streaked onto BHI agar and incubated for 2 days at 20 °C (food isolates) or 20 h at 37 °C (clinical isolates) and then kept at 4 °C. For experiments, single colonies were inoculated into BHI broth and incubated overnight (19 ± 1 h) at 20 °C, unless otherwise stated.

Generation and isolation of phage-insensitive L. monocytogenes mutants

2.2

The strains used to generate phage-insensitive mutants are listed in Table 1. For mutant generation, 100 µL of overnight culture (∼2 × 10^9^ cells/mL) was mixed with 10 µL phage preparation (∼2 × 10^11^ phages/mL) and 3 mL BHI soft agar (0.7% w/v agar) and poured onto BHI agar plates. This resulted in a multiplicity of infection (MOI) of ∼10, which was chosen to achieve a high phage concentration for confluent lysis. Plates were incubated at 20 °C for 5 days or at 10 °C for 13 days until colonies appeared within the large, clear lytic zone on the plate. These incubation temperatures were selected to compare the emergence of mutants under moderate versus low-temperature conditions, the latter representing food-relevant, elevated refrigeration temperatures that may occur during storage. Single colonies growing within this clear zone were randomly selected and cultured overnight at 20 °C in BHI supplemented with 1% (v/v) phage preparation to select for genuinely resistant isolates. Glycerol stocks (20% v/v) were prepared from overnight cultures and stored at −80 °C for further use.

In total, 5 of 7 resistant mutants isolated at 20 °C were selected for further analysis: 1 each from the clinical strains 470 and 475, and from the food-associated strains 442, 443 and 446. These mutants were designated 470–20–3 and 475–20–3, 442–20–1, 443–20–2 and 446–20–3, respectively, according to the incubation temperature and picked colony. For selection at 10 °C, the 5 wild-type strains were incubated with phage in soft agar for 13 days. All 5 produced single colonies, which were subsequently cultured in BHI with phage; picked colonies of 3 strains (out of 5) showed growth in the presence of phage: 1 colony from clinical strain 475 and 1 from each food strain 442 and 446, designated 475–10–1, 442–10–1 and 446–10–6. Thus, in total 8 mutants were selected, 5 from plates incubated at 20 °C and 3 from plates incubated at 10 °C. Phage insensitivity of the mutants was confirmed via flow cytometric live/dead staining and spot assays, as described in Section 2.3.

Validation of phage-insensitivity by flow cytometry and spot assays

2.3

To validate phage insensitivity, the mutants (8 strains) and their respective wild type strains (5 strains) were incubated with phage at 20 °C for a period of 48 h, and viable cells were quantified after 0, 3, 24 and 48 h using flow cytometric live staining, as described previously (Low et al. 2020; Brieske et al. 2025). Wild-type and mutant controls without phages were included to assess normal growth of the Listeria strains over time.

Overnight cultures were diluted 1:2000 in BHI (final concentration approx. 1 × 10⁶ cells/mL), then incubated with 1% (v/v) phage preparation in 400 µL total volume (deep well plates). At each sampling point, cells were diluted 1:50 in freshly prepared staining solution, incubated for 30 min at room temperature, and analyzed using a CytoFLEX flow cytometer (Beckman Coulter). The staining solution was prepared by diluting Syto 13 (1:4000) and 20 mM propidium iodide (PI, both Thermo Fisher Scientific, Waltham, USA) 1:4000 or 1:8000 in 0.25 × Ringer's solution, respectively. Gating was based on forward and side scatter, and Syto 13 fluorescence was recorded to determine the viable cell count. All experiments were performed in six biological replicates for wild types and 20 °C mutants and in three biological replicates for 10 °C mutants. A two-tailed Student’s t-test was performed to assess statistically significant differences between wild-type and phage-resistant mutant strains.

For spot assays, 100 µL of overnight culture was mixed with 3 mL soft BHI agar (0.4% w/v) and poured onto BHI plates (Kutter 2009). After solidification, 5 µL of each tenfold serial dilution of phage in SM buffer (0.05 M Tris–HCl, pH 7.5; 0.58% NaCl; 0.2% MgSO₄·7 H₂O) ranging from 2 × 10¹⁰ to 2 × 10⁷ PFU/mL was spotted. These inocula corresponded to ∼1 × 10^8^ to 1 × 10^5^ phages/spot and ∼2.8 × 10^6^ cells/spot, yielding an MOI of ∼36 to ∼4 × 10^–2^. Phage-free SM buffer served as a negative control. Plates were incubated at 20 °C for ∼24 h and evaluated for lysis as evidenced by clear spots. All spot assays were performed in 3 biological replicates.

DNA extraction, whole genome sequencing and genomic analysis

2.4

Mutants (8 strains) and wild-types (5 strains) were grown in BHI at 20 °C for 20 h. After centrifugation, DNA was extracted from bacterial cell pellets using the Micro AX Bacteria Gravity Kit (A&A Biotechnology, Gdańsk, Poland) following the manufacturer’s protocol, including RNase A digestion (0.1 µg/µL, 5 min at room temperature). DNA concentration and purity were measured with a Qubit 3 Fluorometer and a NanoDrop spectrophotometer (both Thermo Fisher Scientific), respectively.

Whole-genome sequencing was performed on an Illumina MiSeq platform. For library preparation the Illumina DNA Prep Kit was used according to the manufactures protocol. The Illumina MiSeq Reagent Kit V2 (500 cycles) was used for most samples but for strain 446 and its mutant the Illumina MiSeq Reagent Kit V2 micro (300 cycles) was used.

Raw data were analyzed using the Miseq reporter software for base calling, demultiplexing and adapter trimming directly on the instrument. Raw reads were assembled using the BV-BRC genome assembly tool (assembly strategy: unicycler v0.4.8) with default parameters. Genome annotation of the wild-type sequences was performed using the RAST tool kit for bacteria at BV-BRC.

Variant calling was performed using the SNP caller FreeBayes with the aligner BWA-MEM in BV-BRC. Raw read data and assemblies are available at NCBI under BioProject accession PRJNA1356489.

The mutated gene sequence of mutant 475–10–1 was further analyzed in order to predict whether the mutation could be located in the active center of the encoded enzyme. A multiple sequence alignment was performed with the amino acid sequence of the respective gene and homologs from the bacteria Ruminiclostridium thermocellum (YP_001037690), Clostridioides difficile (YP_001089293), Faecalibacterium prausnitzii (ZP_02092750), Treponema pallidum (3322941), Myxococcus xanthus (ABF88119), Thermotoga lettingae (YP_001470567) and Pseudomonas aeruginosa (WP_410924565.1) using HHpred. The protein’s 3D structure was predicted with AlphaFold2 (Abramson et al. 2024), and metal ion binding was predicted using AlphaFill (Hekkelman et al. 2023). The predicted protein’s 3D structure was visualized using ChimeraX (Meng et al. 2023).

Three mutants (443–20–2, 470–20–3, 475–20–3) were further analyzed using HHpred (Zimmermann et al. 2018). For mutant 475–20–3, an additional analysis using AlphaFold3 to predict the 3D structure was performed, followed by a structure-based homology comparison with Foldseek (van Kempen et al. 2024) using the AlphaFold3 generated structure.

To explore possible associations between the categorical variables serovar (2 categories: 2a/2b and 4b) and genes of interest directly (YfhO) or indirectly (UGPase, PGM) involved in WTA glycosylation (3 categories: YfhO, UGPase/PGM, hypothetical protein), a 2 × 3 contingency table was created for all 8 mutants. To identify potential associations, distributions were visually compared. A Fisher–Freeman–Halton test was performed for statistical evaluation of these associations, followed by pairwise Fisher’s exact tests with Holm correction. All analyses were carried out in R (version 4.5.0), and a p-value < 0.05 was considered statistically significant.

Phage adsorption analysis

2.5

Confocal laser scanning microscopy (CLSM)

2.5.1

For fluorescent labeling of the P100 phages, 200 µL of phage preparation was mixed with 5 µL of 10 mM Atto 488 NHS ester (Atto-Tec, Siegen, Germany) in DMSO and incubated for 1 h at room temperature to allow covalent binding of the dye to the phage proteins. Unbound dye was removed using Pierce dye removal columns with purification resin (Thermo Fisher Scientific, Waltham, USA). The labeled phages were stored at 4 °C up to 2 months for phage-adsorption experiments.

For phage adsorption tests, 100 µL of overnight culture was mixed with 1 µL labeled phage and incubated for 2 min at room temperature to allow binding of Atto 488-labeled phages to the bacteria. The mixture was vortexed for 10 s, after which the cells were fixed with 50 µL of formaldehyde (32% v/v) and centrifuged to remove unbound phages. The pellet containing the bacteria with bound phage was then resuspended in staining solution (SM buffer + 25% glycerol + DAPI 1:500) for bacterial counterstaining. After 10 min incubation, 10 µL of suspension was placed on a cover glass (thickness no. 1.5) and covered with an agarose pad to spread and immobilize the bacteria-phage mixture on the cover glass for imaging. Agarose pads were produced by melting 4% w/v agarose in SM-buffer, spread in a thin layer in a sterile petri dish and cut into squares of ∼1cm^2^.

The stained cells were examined using a TCS SP8 confocal laser scanning microscope (Leica Microsystems, Wetzlar, Germany). Bacterial cells, phage particles, and bacterial cells with associated phages were counted. For each sample, two to nine independent fields of view from confocal images were analyzed. A custom Python 3.13 script generated with the help of ChatGPT4.5 and Gemini Pro 2.5 (utilizing numpy, skimage, scipy, matplotlib, and tifffile) was used for image analysis. Bacteria were detected from multichannel TIFF images using the DAPI fluorescence channel (designated as the magenta channel). Image adjustment involved normalizing pixel intensities between a user-defined minimum and maximum. Objects were segmented using global Otsu thresholding, multiplied by a user-adjustable factor. Resulting binary objects smaller than a specified minimum pixel area were removed. The outlines of these detected bacteria were subsequently identified. Phages were detected in the green fluorescence channel. Similar to bacteria, pixel intensities were normalized between user-defined minimum and maximum values. Segmentation was performed using global Otsu thresholding, multiplied by a user-adjustable factor. The centroid of each remaining segmented object was identified as a point-like phage. Phage association to bacteria was confirmed if the Euclidean distance between a phage centroid and any point on the outline of a detected bacterium was lower than a user-specified association distance, which was determined through visual assessment. Each analysis was visually checked for correctness via an interactive graphical user interface, which allowed for real-time adjustment of detection thresholds, image intensity ranges, object size filters, aspect ratio, and the phage-bacteria association distance. Counts for total bacteria, bacteria associated with phages, non-associated bacteria, total phages, phages associated with bacteria, and non-associated phages were determined computationally. The percentage of phage-associated bacterial cells was then calculated based on the total bacteria counts and phage-associated bacteria counts. A two-tailed Student’s t-test was performed to assess statistically significant differences between wild-type and phage-resistant mutant strains.

Flow cytometry

2.5.2

For flow cytometric analysis of phage adsorption, 10 µL of overnight culture was mixed with 1 µL of Atto 488-labeled phage (prediluted 1:10 in SM buffer) and incubated for 2 min at room temperature. SM buffer alone served as unstained control. For heat-inactivated phage controls, Atto 488-labeled phages were inactivated at 90°C for 30 min and used in the same way. After vortexing, samples were fixed with 5 µL of formaldehyde (32 % v/v, Thermo Fischer Scientific, Waltham, USA), diluted 1:10 in SM buffer, and stained with DAPI (1:500 in SM buffer; final dilution 1:100) in 96-well plates. Following 5 min of incubation, Atto 488 fluorescence was measured using a CytoFLEX flow cytometer. To account for autofluorescence, mean fluorescence intensity (MFI) values of unstained controls were subtracted from corresponding sample values. The experiment was performed in three biological replicates. A two-tailed Student’s t-test was performed to assess statistically significant differences between wild-type and phage-resistant mutant strains.

Antibiotic susceptibility testing (minimum inhibitory concentration assay)

2.6

To assess antibiotic susceptibility, minimal inhibitory concentrations (MICs) of 8 antibiotics were determined via broth microdilution. The tested antibiotics included ampicillin, tetracycline (Carl Roth, Karlsruhe, Germany), oxacillin, streptomycin (Sigma-Aldrich, St. Louis, USA), meropenem, ciprofloxacin (Thermo Fisher, Waltham, USA), vancomycin (Apollo Scientific, Manchester, UK) and erythromycin (Serva, Heidelberg, Germany), all tested in the range of 0.125–128 µg/mL using a doubling dilution assay.

Overnight cultures grown in BHI at 37 °C for 19 h were diluted 1:1000 in BHI. Aliquots of 50 µL of the diluted antibiotics were mixed with 50 µL of bacterial suspension in 96-well plates. For growth controls, BHI medium without antibiotics was used. Turbidity was assessed after incubation at 37 °C for 20 h. The MIC was defined as the lowest concentration preventing visible growth. Experiments were performed in triplicate on three separate days.

Growth kinetics at different temperatures and in the presence of food preservatives

2.7

To assess bacterial growth under different temperature and food-relevant stress conditions, two experiments were conducted. For growth analysis at 37 °C, overnight cultures (16 h at 37 °C) were diluted 1:100 in pre-warmed BHI, and 100 µL was transferred into wells of a 96-well plate. Optical density at 600 nm (OD_600_) was measured every 30 min over 16 h with continuous shaking using a TECAN Spark 10 M microplate reader. To compare the growth performance of wild-type and mutant strains, the area under the curve (AUC) values were calculated from the growth curves using SigmaPlot 14.0. A two-tailed Student’s t-test was performed to assess statistically significant differences between wild-type and phage-resistant mutant strains.

For low-temperature growth experiments, overnight cultures (20 h at 20 °C) were first acclimatized for 3 h at 10 °C. They were then pre-diluted 1:100 in pre-cooled BHI. Subsequently, cultures were diluted 1:20,000 in (i) pre-cooled BHI, (ii) pre-cooled BHI supplemented with 3.1% NaCl, (iii) pre-cooled BHI supplemented with 150 ppm sodium nitrite, or (iv) pre-cooled BHI containing both 3.1% NaCl and 150 ppm NaNO_2_. These conditions were selected to simulate environments typically found in ready-to-eat foods such as smoked salmon or cured sausages. Cultures were incubated at 10 °C for a period of 9 days. Viable cells were quantified on day 0 (initial cell count), and after 2, 5 and 9 days using flow cytometry as described in Section 2.3. All experiments were performed in 6 biological replicates.

For statistical analysis, AUC values were log10-transformed to meet ANOVA assumptions. A two-way ANOVA with Tukey’s post hoc test was applied to replicates for each strain to test for significance. Comparisons were performed for (i) different conditions within each strain (wild type, mutant) and (ii) wild type vs. mutant(s) and, where applicable, 20 °C mutant vs. 10 °C mutant within each condition.

Software

2.8

Flow cytometry data were analyzed using CytExpert version 2.5 (Beckman Coulter, Krefeld, Germany). Confocal microscopy images were analyzed using ImageJ (NIH, Bethesda, USA). Genome assembly, annotation and variant calling were performed using the BV-BRC platform (https://www.bv-brc.org/) and sequence homology analyses were performed using HHpred, available at the MPI Bioinformatics Toolkit (https://toolkit.tuebingen.mpg.de/tools/hhpred). Statistical analyses were performed using two-tailed Student’s t-tests or two-way ANOVA followed by Tukey’s post hoc test. Graphs were generated with SigmaPlot version 14.0 (Systat Software Inc., San Jose, CA, USA). P-values < 0.05 were considered statistically significant.

Results

3

Isolation and validation of P100 phage-resistant mutants

3.1

To select for phage-resistant mutants, a total of 14 clinical and food L. monocytogenes strains were incubated with P100 phage in BHI soft agar. After 5 days at 20 °C, single-colony growth was observed for 11 of the 14 strains (for 3 strains no single colonies developed, instead after two days of incubation a thin lawn of growth appeared which became fully evident after 3–4 days of incubation). For each of the 11 strains, 3 well isolated colonies were picked from the BHI plate and were then each cultured in BHI together with the phage P100. For 5 of the 11 strains, at least 1 of the inoculated colonies showed growth (Table 1). These 5 mutant isolates were subsequently confirmed to be resistant by spot assays and flow cytometry, as they showed reduced lysis and maintained viability (Fig. 1A and B).Fig. 1. Validation of P100 phage resistance in L. monocytogenes mutants.(A) Wild-types (blue bars) and mutants generated at 20 °C (red bars) or 10 °C (pink bars) were incubated with P100 phage (hatched bars) or without phage (controls). Viable cell counts were determined at 0, 3, 24, and 48 h of incubation at 20 °C using flow cytometric live staining. Data are presented as mean ± standard deviation from six (wild types and 20 °C mutants) or three (10 °C mutants) biological replicates. A two-tailed Student’s t-test was performed of the replicates for each isolate to test for statistical significance between treatment with phage versus no-phage control. Significant reductions in bacterial counts are indicated as follows: * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001; ns, not significant.(B) Spot assays were performed in parallel. Four 10-fold serial dilutions (10⁻¹ to 10⁻⁴) of P100 phage suspension in SM buffer (corresponds to ∼2 × 10^10^ to ∼2 × 10^7^ PFU/mL) were spotted onto soft agar plates and incubated at 20 °C for 24 h. Phage-free SM buffer served as a negative control (spotted in the center of the plates). Clear zones indicate bacterial lysis. Spot assays were performed in three biological replicates.Fig 1 dummy alt text

For additional mutant selection at 10 °C, the 5 wild-type strains that were capable of developing resistance at 20 °C were incubated with phage in soft agar at 10 °C for 13 days. All 5 strains produced single colonies, which were then again grown together with phage in BHI broth culture at 20 °C for 18 h. Three of these 5 strains were able to grow again in the presence of phage (Table 1) and were thus confirmed as phage-resistant using flow cytometric live cell staining and spot assays. Flow cytometric live cell staining revealed only a minor reduction in the number of viable cells in all phage-resistant mutants during the 48-hour phage incubation (Fig. 1A). Conversely, the corresponding wild-type strains exhibited a marked decline in viable cell numbers, decreasing from approximately 10⁶ to 10⁵ cells/mL within a relatively short 3 h of phage exposure. After 24 h, cell numbers in wild-type cultures were further reduced from ∼10⁸ to ∼10⁴ cells/mL. Viable cell counts of mutants and wild types, both with and without phage, at various time points are shown in Fig. 1A. The corresponding raw data can be found in Supplementary Table S1.

In spot assays, phage susceptibility was assessed visually by monitoring the formation of clear lysis zones. No lysis or strongly reduced lysis (smaller and less clear zones) was observed for the mutants, while the phage-susceptible wild-type strains exhibited clear lysis (Fig. 1B).

Whole-genome sequencing and bioinformatic analyses of mutants and wild types

3.2

To elucidate the molecular mechanisms underlying phage resistance, whole-genome sequences of the 8 phage-resistant mutants were compared to their respective parental wild-type genomes. An overview of the total number of mutations for each strain is provided in Fig. 2A. Mutations predicted to have a high impact (such as frameshifts or stop codons, or missense mutations) on phage resistance development were found to be located in four different genes of interest, as shown in Fig. 2B. A table listing all mutations of each mutant can be found in Supplementary Table S2. The Human Genome Variation Society (HGVS) nomenclature was used to describe the mutations, as it is the global standard for describing and communicating variants in DNA and protein sequences (Hart et al. 2024).Fig. 2. Illumina sequencing results of all eight P100 phage-resistant mutant strains generated from L. monocytogenes serovar 2a, 2b and 4b strains in soft agar at 20 °C and 10 °C.(A) Overview of the total number of mutations for each mutant. (B) Overview of mutations in four genes of interest. Gene mutations are shown in boxes; additionally, predicted protein level effects are indicated. DNA and protein changes follow Human Genome Variation Society (HGVS) nomenclature. c., coding DNA; p., protein; del, deletion; ins, insertion; *, stop-gained; dup, duplication; delins, deletion-insertion; UGPase, UTP-glucose-1-phosphate uridylyltransferase; YfhO, YfhO family protein; PGM, phosphoglucomutase; HP, hypothetical protein.Fig 2 dummy alt text

Three mutants (442–20–1; 442–10–1 and 446–10–6) generated from serovar 4b strains showed frameshift mutations caused by a single-nucleotide deletion (c.221del and predicted p.Asn75fs) or a deletion-insertion (c.44_46delinsTT and predicted p.Thr15fs) in the gene encoding a UTP-glucose-1-phosphate uridylyltransferase (UGPase) (Fig. 2B).

Two mutants (446–20–3; 475–10–1) generated from serovar 4b strains harbored mutations in a gene encoding a phosphoglucomutase (PGM) (Fig. 2B). Of these PGM mutants, one (446–20–3) exhibited a premature stop codon (p.Gly511*) resulting from a nucleotide substitution (c.1531G>T), while the other (475–10–1) carried a missense mutation (c.925G>T), causing a predicted p.Asp309Tyr amino acid change. To determine whether the missense mutation in the PGM-gene of strain 475–10–1 was located in a conserved region, a multiple sequence alignment was performed using the amino acid sequence of PGM from this strain and PGM homologs from bacteria Ruminiclostridium thermocellum (YP_001037690), Clostridioides difficile (YP_001089293), Faecalibacterium prausnitzii (ZP_02092750), Treponema pallidum (3322,941), Myxococcus xanthus (ABF88119), Thermotoga lettingae (YP_001470567) and Pseudomonas aeruginosa (WP_410924565.1). The changed amino acid was found to be at the position of a highly conserved metal-binding site. The conserved aspartic acid residue at position 309 (p.Asp309) is shown in Fig. 3A. The 3D-structure of the metal-binding-site of the wild type PGM from strain 475 (structure prediction by AlphaFold2; Zn^2+^-binding predicted using AlphaFill) is shown in Fig. 3B The structure of the deficient metal-binding-site of the PGM variant hypothetically resulting from the phage-insensitive mutant is shown in Fig. 3C. In this variant, the aspartic acid at position 309 is replaced by tyrosine, and AlphaFill analysis could not predict the coordination of a metal ion at this position.Fig. 3. Sequence conservation and 3D structure of phosphoglucomutase (PGM) of L. monocytogenes strain 475.(A) Multi-sequence alignment of PGM homologs of L. monocytogenes strain 475 and other bacteria: Rth – Ruminiclostridium thermocellum (YP_001037690), Cdi – Clostridioides difficile (YP_001089293), Fpr – Faecalibacterium prausnitzii (ZP_02092750), Tpa – Treponema pallidum (3322,941), Mxa – Myxococcus xanthus (ABF88119), Tle – Thermotoga lettingae (YP_001470567), Pae – Pseudomonas aeruginosa (WP_410,924,565.1). Bar height indicates the level of amino acid conservation at any position. Grey boxes highlight conserved residues. The conserved amino acid Asp309 of strain 475 is indicated by a red box. (B) Structure of the metal-binding-site of the wild type PGM of strain 475 (structure prediction was conducted by AlphaFold2; Zn^2+^-binding was predicted using AlphaFill). (C) Structure of the deficient metal-binding-site of the PGM variant of the phage resistant mutant. The mutant harbours a tyrosine residue instead of aspartic acid at position 309 (structure prediction was conducted by AlphaFold2; metal-binding was analysed using AlphaFill but yielded no coordinated metal ion at the depicted position).Fig 3 dummy alt text

Two additional mutants (443–20–2; 470–20–3) generated from serovar 2a or 2b strains showed a frameshift insertion (c.1456_1457insA and predicted p.Val486fs) or a frameshift duplication (c.1938dup and predicted p.Glu647fs) in a hypothetical gene (Fig. 2B). Further analysis of its amino acid sequence using HHpred, with a probability of 100% and E-value of 2 × 10^–53^, revealed that the gene encodes a YfhO family protein (Fig. 2B).

One mutant (475–20–3), generated from a serovar 4b strain carried a missense mutation (c.187C>A and predicted p.Leu63Met) in a gene whose function was described by BV-BRC as a conserved domain protein; this gene is hereafter referred to as a gene encoding a hypothetical protein (HP) (Fig. 2B). HHpred analysis suggested a possible match to a WxL-interacting protein (WxLIP), but the relatively high E-value (0.089) indicates low confidence in this prediction. Additional analysis using AlphaFold3 to predict the 3D structure was performed; however, the predicted Local Distance Difference Test (plDDT) scores were low (70 > plDDT > 50) to very low (plDDT < 50), and the predicted template modeling (pTM) score for the full structure was < 0.5 (pTM = 0.25), indicating that the predicted structure is likely inaccurate. A structure-based homology comparison with Foldseek, using the AlphaFold3-generated structure, revealed a match to a BspA-family leucine-rich repeat surface protein of L. monocytogenes with a high probability of 0.99 and an E-value of 1.95 × 10^–4^, but the TM-score was < 0.5 (0.31) (RMSD of 13.9 Å) indicating a low degree of structural similarity.

Overall, no notable differences in the types or locations of resistance-associated mutations were observed between mutants generated at 10 °C and those generated at 20 °C (Fig. 2B).

Visually, associations between serovars of the parental wild-type strains and the mutated genes of the phage-resistant mutants were observed. Serovar 2a/2b strains developed mutations in YfhO, whereas 5 of 6 serovar 4b strains developed mutations in UGPase/PGM (Fig. 2B). The Fisher-Freeman-Halton test of all 8 mutants revealed a statistically significant (p < 0.05) association between the categorical variables serovar (2a/2b, 4b) and genes of interest (YfhO, UGPase/PGM, hypothetical protein) (Table S3). Subsequent pairwise Fisher tests, adjusted using Holm-correction, revealed no significant differences (p > 0.05) in the specific pairwise comparisons (Table S4) may be due to the small sample size.

CLSM and flow cytometry revealed reduced phage binding to mutants

3.3

Confocal laser scanning microscopy was used to assess phage-bacterium association of the wild-type and mutant strains. Here, “association” refers to the detectable binding of fluorescently labeled phages to bacterial cells. To assess phage binding, Atto 488-labeled P100 phages were incubated with the wild-type and phage-insensitive mutant strains. CLSM revealed distinct fluorescence signals from the phages appearing as fluorescent dots on the surface of wild-type cells, indicating successful phage binding. In contrast, mutant strains showed markedly reduced fluorescent dots on the bacterial surface, suggesting impaired phage adsorption, as shown for example in strain 446 (Fig. 4A). Confocal microscopic images of all mutants and wild types are shown in the Supplementary Figure S1. For quantification of phage binding, bacterial cells, phage particles, and bacterial cells associated with phages were counted in 2 to 9 fields of view per sample. Among the wild types, 64–84% of the counted bacteria were associated with phages, whereas the mutants showed significantly (p < 0.05) lower phage association, ranging from 4–42% (Fig. 4B). The corresponding raw data can be found in Supplementary Table S5.Fig. 4. Adsorption of fluorescence-labeled P100 phages to wild-type and mutant strains of L. monocytogenes. Bacteria were incubated with Atto488-labeled phages, followed by fixation and DAPI counterstaining.(A) Confocal laser scanning microscopy (CLSM) images of wild-type strain 446 and its respective mutant (generated at 20 °C), showing Atto488-labeled phages (green dots) associated with bacterial cells (magenta). (B) Quantification of wild-type and mutant strains associated with Atto488-labeled phages based on CLSM. Bacterial cell counts were obtained from 2–9 fields of view. Data are shown as mean ± standard deviation. A two-tailed Student’s t-test was performed on the replicates for each isolate to test for statistical significance between wild type and mutant. (C) Quantification of wild-type and mutant strains associated with Atto488-labeled phages based on mean fluorescence intensity (MFI) measured by flow cytometry. Data are shown as mean ± standard deviation from three biological replicates. A two-tailed Student’s t-test was performed on the replicates for each isolate to test for statistical significance between wild type and mutant.Significant reductions are indicated as follows: * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001.Fig 4 dummy alt text

These findings were supported by flow cytometry. While wild-type strains exhibited high Atto 488 fluorescence intensity, corresponding to efficient phage binding, the MFI of all mutants was significantly lower (p < 0.01, Fig. 4C and Figure S2). This reduction in fluorescence confirms that the observed phage resistance in mutants is associated with diminished surface binding of phage P100. The corresponding raw data can be found in Supplementary Table S6.

Increased antibiotic sensitivity of phage-resistant mutants

3.4

To assess potential effects of phage resistance on antibiotic susceptibility, the 5 mutants generated at 20 °C and their parental wild-types were tested against 8 antibiotics. All mutants exhibited increased sensitivity to at least one antibiotic, with mutant 442–20–1 showing the strongest effect, displaying increased susceptibility to 5 (ampicillin, oxacillin, meropenem, ciprofloxacin and streptomycin) of the 8 tested antibiotics (Table 2). Together with mutant 446–20–3, it demonstrated the most pronounced changes in susceptibility, which featured a fourfold reduction in MIC for ampicillin from 4 µg/ml to 1 µg/ml, and a twofold reduction for oxacillin from 32 µg/ml to 16 µg/ml (Table 2), both being β-lactam antibiotics. Mutant 446–20–3 also showed a fourfold reduction in MIC for oxacillin, from 64 µg/ml to 16 µg/ml (Table 2). Moreover, 2 mutants (443–20–2 and 446–20–3) showed a twofold increase in sensitivity to tetracycline, and 1 mutant (443–20–2) exhibited a twofold increase in sensitivity to erythromycin. In contrast, none of the mutants displayed any change in vancomycin susceptibility (Table 2). Detailed MIC values from individual replicates are listed in Table S7 (Supplementary Material).Table 2. Minimal inhibitory concentrations of 8 antibiotics in wild-type and mutant strains.Table 2 dummy alt textStrainMIC (µg/mL)AMPOXAMERCIPVANERYTET****STR442 Wild type432181612128442–20–1 Mutant1160.54161264443 Wild type23218161464443–20–2 Mutant21618160.5264446 Wild type464181612128446–20–3 Mutant116181611128470 Wild type232181612128470–20–3 Mutant23218161264475 Wild type116181612128475–20–3 Mutant11618161264AMP, ampicillin; OXA, oxacillin; MER, meropenem; CIP, ciprofloxacin; VAN, vancomycin; ERY, erythromycin; TET, tetracyclin; STR, streptomycin

Impaired growth kinetics of mutants under various environmental conditions

3.5

To investigate whether the development of phage resistance was accompanied by altered growth behavior, growth kinetics of all mutants and their respective wild types were assessed at 37 °C and under food-relevant stress conditions at 10 °C.

At 37 °C, OD_600_ was measured at 30-minute intervals over a period of 16 hours. The resulting growth curves (Figure S3) were converted into AUC values to compare wild-types and mutants (Fig. 5A). OD_600_ measurements revealed a significant (p < 0.001) reduction in growth in phage-resistant mutants 442–20–1 and 446–20–3 (both generated from serovar 4b strains) compared to their wild-type strains (Fig. 5A). A notably but not significant growth reduction was observed for 5 of the other 6 mutants (Fig. 5A). The corresponding raw data can be found in the Supplementary Table S8.Fig. 5. Growth kinetics of L. monocytogenes wild-type and mutant strains at 37 °C and under food-relevant preservation conditions (10 °C with and without sodium chloride and/or sodium nitrite).(A) Bacterial cultures were grown in BHI broth at 37 °C with shaking, and optical density at 600 nm (OD_600_) was measured every 30 min over 16 h. Growth performance was quantified as the area under the curve (AUC). Data are presented as mean ± standard deviation from six biological replicates. A two-tailed Student’s t-test was performed of the replicates for each isolate to test for statistical significance between wild-type versus mutant. Significant reductions in AUC are indicated as follows: *** p < 0.001; **** p < 0.0001; ns, not significant. (B) Bacterial cultures were incubated statically in BHI broth under four conditions: (i) 10 °C, (ii) 10 °C with 3.1% NaCl, (iii) 10 °C with 150 ppm NaNO₂, and (iv) 10 °C with 3.1% NaCl plus 150 ppm NaNO₂. Viable cell counts were determined by flow cytometric live cell staining at day 0 (initial cell number) and after 2, 5, and 9 days of incubation. Growth performance was quantified as AUC. Data are presented as means ± standard deviations from six biological replicates. For statistical analysis, AUC values were log10-transformed to meet ANOVA assumptions. A two-way ANOVA with Tukey’s post hoc test was applied to replicates for each strain to test for significance. Comparisons were performed for (i) different conditions within each strain (wild type, mutant) and (ii) wild type vs. mutant(s) and, where applicable, 20 °C mutant vs. 10 °C mutant within each condition. All comparisons showed significant differences (p < 0.05), except those marked ns, (ns, not significant p > 0.05). For the condition sodium nitrite of mutant 443, the ANOVA assumptions were not met and no statistical significance was considered.For phage-resistant mutants, the products of the respective mutated genes are indicated below the strain numbers in the same colors as the corresponding bars: UGPase, UTP-glucose-1-phosphate uridylyltransferase; YfhO, YfhO family protein; PGM, phosphoglucomutase; HP, hypothetical protein.Fig 5 dummy alt text

Low-temperature experiments at 10 °C simulated elevated refrigeration temperatures that may occur during storage, and NaCl or NaNO_2_ concentrations typical for ready-to-eat foods such as smoked salmon or cured sausages. Viable bacteria counts were determined by flow cytometry on days 0, 2, 5, and 9. Tests were performed with and without 3.1% NaCl, 150 ppm sodium nitrite or their combination. Across almost all conditions, mutants 442–20–1, 442–10–1, 446–20–3, 446–10–6, 475–20–3 and 475–10–1 showed significantly (p < 0.05) impaired growth compared to their parental serovar 4b wild types (Fig. 5B and Figure S3). No further growth restriction of the mutants relative to the wild types was observed in the presence of NaCl or NaNO_2_ (Fig. 5B and Figure S4). The two mutants 443–20–2 and 470–20–3 generated from serovar 2a or 2b strains did not show significant growth reduction across all conditions. The corresponding raw data can be found in Supplementary Tables S9 and S10.

Discussion

4

One primary concern regarding the use of phages for the biocontrol of pathogenic bacteria in food is the potential selection of phage-resistant mutants. To date, no study has investigated the resistance mechanisms and phenotypic effects of P100 resistance in L. monocytogenes in detail. This study employed an in vitro model in which recently circulated (2014–2020) clinical and food-associated L. monocytogenes strains of serovars 2a, 2b, 2c and 4b were exposed to a P100 phage preparation. A comprehensive analysis of the evolved resistant mutants was conducted to investigate the genetic basis of P100 phage-resistance and to assess phenotypic changes in terms of growth ability and antibiotic susceptibility.

Eight mutants genuinely resistant to P100 phage were obtained and confirmed by flow cytometry and spot assays. Flow cytometric live staining demonstrated clear growth of the mutants in the presence of the phage. However, the mutants still exhibited partial susceptibility, as bacterial cell numbers decreased slightly during 48 h of phage exposure compared to the no-phage controls. Confocal laser scanning microscopy showed that a small proportion of cells in the mutants could still bind phages, and spot assays revealed that some mutants exhibited slight lysis, albeit greatly attenuated compared to the wild type. This observation may be explained by a phenomenon known as “leaky resistance," as described by Chaudhry et al (2018). According to this concept, a dominant population of phage-resistant cells coexists with a subpopulation of phage-susceptible cells. These susceptible cells may arise through phenotypic or genetic reversion to susceptibility, thereby allowing continued phage propagation (Chaudhry et al. 2018).

To uncover the underlying genetic changes leading to P100 phage-resistance, the eight mutants were subjected to whole-genome sequencing. The mutations predicted to have a high impact on phage resistance development were located in four different genes. In the two mutants derived from serovar 2a or 2b strains, a single-nucleotide insertion or duplication caused a frameshift in a gene that exhibited 100% homology to a YfhO family protein. Wall teichoic acids of serovar 2a and 2b strains are decorated with GlcNAc and rhamnose (Kamisango et al. 1983), and the glycosyltransferase YfhO encoded by the lmo1079 gene in L. monocytogenes has been shown to be responsible for transferring GlcNAc onto the ribitol backbone of WTAs (Eugster et al. 2015; Rismondo et al. 2018; Rismondo et al. 2020). This enzyme catalyzes the final step of WTA O-GlcNAcylation in serovar 2 strains and is therefore considered to be directly involved in WTA glycosylation. Adsorption studies with the two P100-like phages LP-048 and LP-125 on L. monocytogenes strains have shown that GlcNAc and/or rhamnose is required for efficient phage adsorption (Tokman et al. 2016; Brown et al. 2021).

Wall teichoic acids of serovar 4b strains are more highly glycosylated, carrying GlcNAc residues further modified with glucose and galactose (Uchikawa et al. 1986; Fiedler 1988). Several genes are involved in the galactosylation process: galE catalyzes the synthesis of uridine diphosphate (UDP)-glucose to UDP-galactose (Sumrall et al. 2019), UDP-galactose is then added to an undecaprenol‐phosphate (UndP) lipid carrier by gttA (Sumrall et al. 2019), UndP-Gal is then flipped by the putative flippase GtcA (encoded by gtcA) to the outer leaflet of the membrane (Promadej et al. 1999; Rismondo et al. 2018), and gttB catalyzes the addition of galactose from UndP-galactose onto the integrated GlcNAc residue of WTA (Sumrall et al. 2020). For glucosylation, the genes gtcA (Promadej et al. 1999), gltA, and gltB are involved (Lei et al. 2001). Surprisingly, none of the six mutants derived from serovar 4b in our study had mutations in these genes. Instead, five mutants developed mutations in genes encoding UGPase or PGM, both of which are enzymes that play a key role in polysaccharide synthesis and act much further upstream than the above-mentioned glycosylation enzymes. These two enzymes are therefore classified to be indirectly involved in WTA glycosylation. In three mutants, a single nucleotide deletion induced a frameshift in the gene encoding UGPase. UGPase, encoded by the galU gene, catalyzes the conversion of glucose 1-phosphate and uridine triphosphate (UTP) into the nucleotide-activated sugar uridine diphosphate (UDP)-glucose (Weissborn et al. 1994) and pyrophosphate. UDP-glucose is an essential precursor for the biosynthesis of glycans across all domains of life (Zhang et al. 2025), and in bacteria it serves both as a signaling molecule and as a building block for the synthesis of complex polysaccharides (Berbís et al. 2014). In Gram-positive bacteria it also serves as a substrate for enzymes involved in the biosynthesis of WTAs and cell surface carbohydrates (Allison et al. 2011; Brown et al. 2012; Wu et al. 2021). UDP-glucose is essential for WTA glycosylation in L. monocytogenes (Spears et al. 2016; Sumrall et al. 2019) and Sumrall et al. (2019) showed that exposure of L. monocytogenes serovar 4b to the broad-host-range phage A511 selected for an insensitive mutant also harboring both a single nucleotide polymorphism and a frameshift in galU. Mass spectrometry confirmed the absence of galactose and glucose modifications on WTA GlcNAc residues, while genetic complementation of galU largely restored these decorations and phage adsorption.

UDP-glucose is synthesized from glucose 1-phosphate, which itself is generated from glucose 6-phosphate by PGM (Adhya and Schwartz 1971). Two additional phage-insensitive mutants derived from serovar 4b strains contained non-synonymous mutations in the PGM-encoding gene. One mutant strain harbored a nonsense mutation resulting in a premature stop codon (p.Gly511*), while the other carried a missense mutation predicted to cause an amino acid substitution (p.Asp309Tyr) in the enzyme’s metal binding site, likely affecting its function.

For Listeria phages other than P100, resistance mechanisms in L. monocytogenes isolates also involved inhibition of phage adsorption and most mutations identified in resistant strains in these studies were found in genes associated with WTA biosynthesis or WTA-decoration (Denes et al. 2015; Eugster et al. 2015; Sumrall et al. 2019; Trudelle et al. 2019, 2022). The genes affected in P100 phage-resistant L. monocytogenes in our study, encoding UGPase or YfhO, are the same as those reported in L. monocytogenes strains resistant to phages other than P100 (Eugster et al. 2015; Sumrall et al. 2019).

For one mutant derived from a serovar 4b strain, it was not possible to clearly identify which gene or genes were involved in resistance development. A missense mutation was found in a gene whose function was described by BV-BRC as a conserved domain protein. HHpred analysis suggested a possible match to WxLIP, but the high E-value (0.089) makes this prediction highly speculative. WxL proteins are located on the cell surface and contain peptidoglycan-binding WxL domains (Galloway-Peña et al. 2015; Hassan and Williamson 2023). WxLIP is thought to bind and stabilize WxL on the peptidoglycan surface (Hassan and Williamson 2023; Hassan et al. 2023), suggesting a potential, but unconfirmed, role in phage adsorption. Further analysis with AlphaFold3 and Foldseek matched with a BspA-family leucine-rich repeat surface protein, but with a low degree of structural similarity. This surface protein was previously detected in Tannerella forsythia and is speculated to mediate the binding of bacteria to host extracellular matrix components and clotting factors (Sharma et al. 1998). Homologs of the BspA protein have been identified in several bacterial and eukaryotic pathogens (Takkouche et al. 2023). Overall, the mutated conserved domain protein may be a surface protein or one required for anchoring a surface protein. Resolving this question would require more detailed investigation, which was beyond the scope of this study.

In summary, seven of eight P100 phage-resistant L. monocytogenes mutants in our study developed mutations in genes directly or indirectly involved in WTA glycosylation. In serovar 2a and 2b-derived mutants, the YfhO-encoding gene, directly involved in WTA glycosylation, was affected, whereas in five of the six serovar 4b-derived mutants, UGPase- or PGM encoding genes were affected. The Fisher-Freeman-Halton test revealed a significant association between serovars and the genes directly or indirectly involved in WTA glycosylation, indicating a clear visual trend. The non-significant pairwise comparisons may be due to the small sample size. Further studies with larger samples are required to validate the observed associations. All resistant mutants exhibited reduced phage binding, suggesting that surface modifications resulting from mutations in genes involved in WTA glycosylation appear to be a key factor in the development of resistance at both 20°C and at 10°C. Moreover, our present study shows for the first time, that in two mutants the PGM-encoding gene is affected in the development of phage resistance in L. monocytogenes. Since the mutant colonies were selected at random and not all emerging colonies were screened, one must question whether the selected mutant is representative of the entire mutant population. This raises the question of whether repeating the screening procedure under the same conditions would consistently yield the same or similar variants. The findings in this study provide a solid basis for the design of future studies, which should be planned with a substantially larger cohort of resistant mutants to comprehensively capture the range of resistance mechanisms. Our results provide strong evidence for the genetic cause of P100 phage resistance, but should be supplemented by knockout and complementation experiments to fill in the gaps.

All tested 20 °C-generated phage-resistant mutants isolated in this study exhibited increased sensitivity to at least one antibiotic when compared to their wild-type strain, with two mutants (derived from serovar 4b strains) demonstrating the most significant alterations, exhibiting a fourfold MIC reduction for ampicillin and two- to fourfold for oxacillin, both β-lactam antibiotics. In these mutants, mutations affected the UGPase or PGM genes, the products of which are both indirectly involved in WTA glycosylation. The observed link between genes related to WTA glycosylation and altered β-lactam sensitivity is consistent with findings of Meireles et al. (2020), who demonstrated that L. monocytogenes deletion mutants lacking WTA glycosylation exhibited increased sensitivity to ampicillin and penicillin compared to glycosylated wild-types (Meireles et al. 2020). The authors hypothesized that this increased β-lactam susceptibility may be due to enhanced peptidoglycan permeability.

Further observations reported by the European Food Safety Authority (2016) support the notion that P100 resistance can influence antimicrobial susceptibility. In one of two P100-resistant mutants derived from a single L. monocytogenes strain with intermediate rifamycin resistance, acquisition of phage resistance coincided with full resistance to rifamycin. Conversely, two strains originally resistant to ciprofloxacin and one strain resistant to erythromycin reverted to sensitivity following the development of P100 resistance. The mechanisms underlying these changes remain unclear (European Food Safety Authority 2016). In general, mutations conferring phage resistance in both Gram-positive and Gram-negative bacteria have been shown to alter antibiotic susceptibility, sometimes restoring sensitivity and in other cases leading to increased resistance (Chan et al. 2016; Burmeister et al. 2020; Canfield et al. 2023; McGee et al. 2023). The results of our study suggests that mutants which would surviving phage treatment in food and subsequently cause listeriosis in consumers may be more amenable to treatment with antibiotics due to their heightened susceptibility. This would particularly relate to the two mutants in our study that exhibited markedly elevated sensitivity to the β-lactam antibiotic ampicillin, given that the first-line treatment for listeriosis patients involves the administration of ampicillin, often in combination with gentamicin (Temple and Nahata 2000; Dos Reis et al. 2022). The increased sensitivity of the two mutants to ampicillin was also accompanied by a significant decrease in growth at 37 °C. We speculate, that this phenotypic change could also be advantageous for antibiotic treatment of listeriosis patients.

A current focus in phage therapy is the concept of phage steering, where selective phage pressure is applied not only to eliminate bacterial pathogens but also to guide resistance evolution in beneficial directions, e.g., toward reduced virulence or increased antibiotic susceptibility (Chan et al. 2016; Orndorff 2016; Burmeister et al. 2020; Chan et al. 2025). WTA glycosylation in L. monocytogenes is known to confer multiple advantages for the bacterium itself, including virulence (Orndorff 2016; Spears et al. 2016; Sumrall et al. 2019; Monteiro et al. 2025), resistance to antimicrobial peptides (Carvalho et al. 2015), and cold tolerance (Chassaing and Auvray 2007). Based on the observations of the aforementioned studies and those obtained in our study, we think that phage steering could potentially be applied in the food sector to drive the evolution of pathogenic bacteria that survived phage treatment towards less harmful phenotypes. In addition to elevated antibiotic susceptibility and virulence attenuation, reduced cold tolerance and impaired resistance to antimicrobial peptides would be particularly desirable outcomes. Six mutants, including the two with increased ampicillin sensitivity and reduced growth at 37 °C, showed significantly impaired growth at 10 °C. For most of these six mutants, the impairment was also observed at 10 °C in the presence of the food preservatives sodium chloride and sodium nitrite. Overall, serovar 4b-derived mutants exhibited stronger fitness trade-offs than those from serovar 2a or 2b It can be hypothesized that mutations in the key enzymes UGPase or PGM impose a higher metabolic burden on the bacterium than mutations in genes directly involved in WTA glycosylation such as lmo1099 encoding YfhO. This may explain the significantly more impaired growth observed in the 4b-derived mutants. However, this hypothesis requires experimental confirmation.

A limitation of this study is that the results of the in vitro experiments cannot be directly transferred to the context of phage application in food. However, it has recently been shown that phage-resistance patterns of Klebsiella pneumoniae did not differ substantially between in vitro and in vivo conditions (Fu et al. 2025). In Escherichia coli, phage-resistant clones that emerged from both in vitro (liquid medium) and in vivo (murine pneumonia), also exhibited a convergent mutational pattern in phage-resistance mechanisms under both conditions (Gaborieau et al. 2024). Whether similar resistance patterns to those observed in vitro can also be detected in food matrices should be investigated in future studies.

In conclusion, this study provides insights into the genetic basis of phage P100 resistance in recently circulating clinical and food-associated L. monocytogenes strains. Depending on the serovar, phage-resistance development affected genes involved either directly or indirectly in WTA glycosylation. These mutations were associated with impaired phage adsorption and fitness trade-offs, including reduced growth at both 37 °C and 10 °C, also under conditions relevant to food preservation, such as the presence of sodium chloride and sodium nitrite. All tested mutants (generated at 20 °C) developed increased sensitivity to one or more antibiotics, with two mutants showing particularly pronounced sensitivity to β-lactam antibiotics, exhibiting up to fourfold lower MICs compared to their respective wild types. These findings highlight exploitable vulnerabilities of resistant clones and may inform future antimicrobial strategies and biocontrol applications. Future work should focus on the detection of P100 resistance under real food system conditions and elucidate the mechanisms linking phage resistance to altered antibiotic sensitivity.

Data availability statement

All raw sequencing data and assemblies are available at NCBI under BioProject accession PRJNA1356489.

Funding

This project is made possible by the Junior Research Group program of the Max Rubner-Institut.

Declaration of generative AI and AI-assisted technologies in the manuscript preparation process

During the preparation of this work the authors used ChatGPT-5 in order to improve the language in select sections of the manuscript. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

CRediT authorship contribution statement

Christoph Brieske: Methodology, Investigation, Data curation, Formal analysis, Validation, Visualization, Writing – original draft, Writing – review & editing, Conceptualization. Frank Hille: Methodology, Writing – review & editing. Erik Brinks: Methodology, Software, Writing – review & editing. Hui-Zhi Low: Conceptualization, Methodology, Writing – review & editing, Funding acquisition, Project administration. Charles M.A.P. Franz: Conceptualization, Resources, Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.