Salmonella enterica serovar Typhimurium LB5000 (SGSC181; University of Calgary) was used to propagate and quantify the virulent bacteriophages used in this study and for mating studies. A S. Typhimurium ATCC 14028 spontaneous mutant resistant to rifampicin (Rif^R), obtained by us, was used for the studies of the emergence of reduced phage-susceptibility of bacteria. Moreover, S. Typhimurium LT2 wild-type strain (University of Calgary), TA1537 strain (Dr. Bruce Ames, University of California, Berkeley), and a collection of lipopolysaccharide (LPS) mutants of S. Typhimurium LT2 (University of Calgary) were used to identify the possible receptors of phages UAB_Phi20, UAB_Phi78, and UAB_Phi87. Finally, the strain Escherichia coli DH5α was used for mutagenesis strategies (Clontech, Takara Bio, France) (Supplementary Table S1). The S. Typhimurium ATCC 14028 Rif^R strain and derived variants were cultured on Luria-Bertani (LB) or XLD Agar (Xylose-Lysine-Desoxycholate Agar; Becton Dickinson) media supplemented with Rifampicin (75 µg/mL). All the other bacterial strains were grown in LB broth or agar plates. In all cases, plates were incubated for 18 h at 37°C.

The bacteriophages studied belong to Lederbergvirus (UAB_Phi20, accession number: NC_031019), Zendervirus (UAB_Phi78, accession number: NC_020414) and Felixounavirus (UAB_Phi87, accession number: NC_027360) genera. These phages have been fully characterized and their genomes have been sequenced in previous studies (Bardina et al., 2012; Bardina et al., 2016).

Phage lysates were prepared following a previously described method (Colom et al., 2015) and filtered through 0.45 µm pore-size polyethersulfone (PES) membranes (Millex®-HP). The phage cocktail was prepared by mixing the UAB_Phi20, UAB_Phi78, and UAB_Phi87 bacteriophages with a ratio of 1:1:1, each phage at a titer of 3.3 log₁₀ PFU/mL. The phage cocktail was further diluted with MgSO₄ 10 mM to reach the desired concentration of experiments. Individual phages or phage cocktail titration was performed by plating ten-fold serial dilutions onto ATCC 14028 Rif^R LB plates using the double agar method (Kutter, 2009).

Detection of ATCC 14028 Rif^R isolates with reduced susceptibility to bacteriophages was studied in three different experimental conditions: i) Salmonella cultures infected with the phage cocktail (LAB), ii) cooked ham slices previously contaminated with Salmonella and treated with the phage cocktail (FOOD), and iii) oral phage therapy applied to broilers experimentally infected with Salmonella in experimental farms (PT) (Colom et al., 2017).

Infection of Salmonella in vitro cultures was performed as follows. A culture of ATCC 14028 Rif^R strain grown in LB medium at 37°C until an optical density at 550 nm (OD₅₅₀) of 0.2 was infected with the phage cocktail (MOI of 1 PFU/CFU) and incubated for 24 h at 37°C. An uninfected culture was included as a control. Samples of both cultures were taken at 0, 4.5, 6.5, and 24 h, and appropriate ten-fold serial dilutions were plated on XLD-Rif (75 µg/mL). After 18 h of incubation at 37°C, around 50 colonies were randomly isolated on LB plates per time and type of culture for further studies.

For FOOD experiments, slices of packed cooked ham were purchased in a local supermarket, and the absence of Salmonella and bacteriophages capable of infecting the parental strain used in this study was verified through enrichment methods (Guttman et al., 2005). All the processes related to Salmonella contamination, phage treatment and drying of the ham slices were carried out in a level II biological safety cabinet (Telstar ™ Bio II Advance 4, Telstar Technologies S.L.). Firstly, slices of cooked ham were cut into 49 cm² pieces using sterile material. To remove the remains of preservatives to ensure an efficient contamination with ATCC 14028 Rif^R strain, ham pieces were immersed in 0.9% NaCl (300 mL NaCl 0.9% for every 25 slices) and shaken at 50 r.p.m. for 5 min. Pieces of cooked ham were placed in trays covered with aluminum foil (previously disinfected with 70% ethanol and ultraviolet light for 15 min on each side) and dried in the cabin for 5 min (each side). One side of ham slices was contaminated with ATCC 14028 Rif^R strain by spraying a suspension in 0.9% NaCl [6.1 mL; 1 x 10⁷ colony-forming units (CFU)/mL] using an airbrush coupled to a compressor (Ventus Air-23) at a pressure of 1.5 bar. After 15 min of drying, pieces were treated by spraying with the phage cocktail in MgSO₄ 10 mM (6.1 mL; 1 x 10⁹ PFU/mL). Control samples, untreated, were sprayed with the same volume of MgSO₄ 10 mM. After an additional 15 min of drying, each slice of cooked ham was individually stored in Whirl-Pak filter bags (Nasco, Fort Atkinson, WI) at 4°C for 7 days. On days 0, 3, and 7, five pieces were taken, and 50 mL of buffered peptone water (BPW, Merck) was added to each bag. The contents were homogenized with a Bagmixer (Interscience) at maximum speed for 60 s. Determination of Salmonella concentration was done as previously described in XLD-Rif (75 µg/mL) (Colom et al., 2015). Finally, 240, 232, and 214 colonies from the plates of the control group on days 0, 3, and 7, respectively, were isolated on LB plates for further studies. Similarly, 235, 198, and 232 colonies were isolated at the same days from the treated group.

Concerning the PT experiment, Salmonella isolates came from a previous in vivo oral phage therapy study of Salmonella ATCC 14028 Rif^R-infected commercial broilers and treated with the phage cocktail (Colom et al., 2017). Before infection, broilers were Salmonella and ATCC 14028 Rif^R-specific phage free (Colom et al., 2017). Briefly, twenty-four cecum samples were collected on days 1, 8, and 15 after Salmonella infection from phage-treated and untreated group, respectively. After processing of samples (Colom et al., 2015), a total of 720 colonies were isolated from the control group (240 colonies per period) and 630 colonies from the bacteriophage-treated group (166 colonies on day 1, 225 on day 8, and 240 on day 15).

Randomly selected colonies were streaked on LB-Rif (75 µg/mL) and incubated for 18 h at 37°C. Afterwards, they were streaked on green plates (Chan et al., 1972) by three consecutive subcultures to ensure the absence of concomitant bacteriophages with the bacterial cells.

To determine the susceptibility to UAB_Phi78 phage, the isolates were grown in LB liquid medium in 96-well microtiter microplates (Sterilin™, Thermo Fisher Scientific, Massachusetts, USA) by incubation at 37°C with orbital shaking at 500 r.p.m (Kline; K3E model, Ovan, Barcelona, Spain) up to an absorbance at 550 nm (A₅₅₀) of 0.12-0.15 (equivalent to 1 x 10⁷ CFU/mL). Then, cultures were infected with the phage at a MOI of 0.1. The plates were incubated at 37°C with shaking at 500 r.p.m, and growth was monitored by measuring the A₅₅₀ every hour until 3 h, using the plate reader Sunrise (Tecan Trading AG, Switzerland). Moreover, uninfected clones were included as controls. Those isolates whose A₅₅₀ increased over time in a similar way to the uninfected culture were considered to show reduced phage-susceptibility. In contrast, those whose A₅₅₀ decreased over time were categorized as sensitive to phage. Strains ATCC 14028 Rif^R (sensitive to the three bacteriophages) and TA1537 (resistant to the three bacteriophages) were used as controls.

The sensitivity to UAB_Phi20 and UAB_Phi87 bacteriophages of all isolates showing reduced UAB_Phi78-susceptibility was determined by spot assay. Briefly, 10 µL of the bacteriophage lysate (1 x 10⁶ PFU/mL) were dropped onto the double layers of each isolate in LB agar plates and incubated at 37°C overnight (Kutter, 2009). Strains ATCC 14028 Rif^R (sensitive to the three bacteriophages) and TA1537 (resistant to the three bacteriophages) were used as controls.

When necessary, the effect of phage infection on bacterial growth kinetics of the ATCC 14028 Rif^R variants was checked on liquid LB cultures as above described monitoring the A₅₅₀ at 30-min or 1-h intervals after infection till 3.5 h, using the spectrophotometer Genesys 10S UV-Vis (Thermo Fisher Scientific, Massachusetts, USA). In addition, the concentration of viable cells (CFU/mL) was determined each time. Non-infected cultures of the studied strains were included as a control.

Assessment of the efficiency of plating (EOP) was tested under the stationary phase of both target (ATCC 14028 Rif^R variants) and host (S. Typhimurium ATCC 14028 Rif^R) strains. For this, the bacteria were cultured at 37°C for 18 h in LB medium, and 0.1 mL of the cultures were mixed with 0.1 mL of ten-fold serial dilutions of the appropriate phage lysate in LB molten agar (0.7%) and plated in LB agar by the double layer method. The plates were incubated at 37°C for 18 h and the number of PFU was counted. The EOP was calculated as described previously (Khan Mirzaei and Nilsson, 2015): average PFU on target bacteria/average PFU on host bacteria, and the standard deviation was also calculated for a minimum of three assays. For each phage, and according to the EOP value obtained, the variants were considered as high (EOP ≥0.5), medium (EOP < 0.5 but ≥0.1), low (EOP between 0.001 and 0.1) efficiency, and inefficient (<0.0001) (Khan Mirzaei and Nilsson, 2015).

Phage adsorption to target bacteria and burst size of phages was determined as described previously (Kropinski, 2009; Kropinski, 2018). Likewise, center of infection (COI) and ECOI were determined as described previously (Moineau et al., 1993).

The bacterial LPS was extracted following a protocol previously designed and kindly provided by Dr. J. Casadesús (Universidad de Sevilla) (Buendía-Clavería et al., 2003). The gels were digitally photographed with the Gel Doc XR system (Bio-Rad Laboratories, S.A).

Total DNA was prepared using the Easy-DNA^TM kit (Invitrogen), following the manufacturer’s instructions. Sequencing was performed by Sistemas Genómicos S.L. (Sistemas Genómicos S.L, Valencia, Spain). The pool of the libraries was sequenced by paired-end sequencing (100 x 2) in an Illumina Hiseq 2500 sequencer. The sequencing resulted in around 9 million 101-nt long paired-end reads for each sample, with a mean Phred quality score > 30.

To find the putative mutations responsible for phage reduced susceptibility, the trimmed paired-end reads were aligned against the chromosome and pSLT plasmid sequences of S. Typhimurium ATCC 14028s (Genbank accession numbers: CP001363.1 and CP001362.1, respectively) using Geneious Prime 2020.0.4 (https://www.geneious.com) with up to 5 fine-tuning iterations. Single nucleotide polymorphisms (SNPs) calling was performed using the Geneious Prime software while larger insertions/deletions (≥ 5bp) were determined using Mauve Contig Mover (MCM) (Darling et al., 2010). Only insertions/deletions that were not found at the ends of the contigs (more than 20 bp away from the ends) and were absent in S. Typhimurium ATCC 14028 Rif^R parental strain were retained.

The existence of unused reads resulting from mapping assemblies revealed the presence of exogenous DNA to the S. Typhimurium ATCC 14028s chromosome or pSLT plasmid. To further investigate this acquired DNA, trimmed paired-end reads were re-assembled de novo using SPAdes v3.11.1 (with ‘careful -k 21,33,55,77’ settings) (Bankevich et al., 2012), and the scaffolds matching the genome or plasmid of S. Typhimurium 14028s through BLASTn (coverage>75%, e-value<1E-10) (Altschul et al., 1997) were removed from further analysis. To obtain the complete sequences of the acquired DNA, the retained scaffolds were combined with PLACNETw assemblies (Vielva et al., 2017) using Velvet (Zerbino, 2010) through Geneious Prime. Finally, complete plasmid assemblies were annotated using Prokka version 1.12 (Seemann, 2014) against the COG (Tatusov et al., 2000), HAMAP (Pedruzzi et al., 2015) and Pfam (Finn et al., 2016) databases and using their three best BLASTn complete plasmids hits as reference. The replicon sequences of each plasmid were determined using the PlasmidFinder 2.1 web-tool (Carattoli et al., 2014).

Construction of plasmids for complementation of mutations of target genes of Salmonella genomes was carried out by amplification of each wild-type gene of interest by PCR from the chromosome of ATCC 14028Rif^R using the appropriate primers (Supplementary Table S2). The primers carried homology for appropriate cloning at the 5’-end of the sequence and the PCR was done using the Phusion High-Fidelity DNA Polymerase (Thermo Scientific™). The resulting PCR products were purified with NzyGelpure kit (NZYTech, Lisboa, Portugal) and cloned into a previously linearized and dephosphorylated pUA1108 vector (Supplementary Table S1) via double digestion with NdeI and BamHI enzymes, as previously described (Mayola et al., 2014). The resulting ligation was used to transform competent E. coli DH5α cells by electroporation. Transformants were selected by growth on 50 µg/mL ampicillin plates. Plasmid DNA was extracted with a Plasmid DNA purification kit (NZYTech, Lisboa, Portugal) and the gene sequences were confirmed by sequencing (Macrogen).

The confirmed recombinant plasmids were introduced by electroporation in target Salmonella electrocompetent cells. Following selection of the transformants, the presence of cloned products was confirmed again by PCR and sequencing (Macrogen). As a control, the empty pUA1108 vector was transformed into the target bacteria under the same conditions.

Transference of phage interference mechanisms codified in acquired plasmids of PT variants was assessed by mating assays on a solid surface by adapting the method previously described (García-Quintanilla et al., 2008). S. Typhimurium LB5000 Str^R was used as recipient strain. The number of CFUs was recorded and the conjugation frequencies were calculated.

The susceptibility to the phages of the cocktail of four transconjugants of each matting experiment was assayed by spot assay, as described before (Kutter, 2009). Furthermore, the presence of acquired plasmids in such transconjugants was confirmed by conventional PCR using appropriate primers for sequences specific to each plasmid (Supplementary Table S2).

Insertion of a kanamycin resistance cassette in the pUA1139 plasmid was performed using a λ-Red recombinase strategy (Datsenko and Wanner, 2000). Briefly, the kanamycin resistance cassette from pKD4 (Supplementary Table S1) was amplified by PCR using the primers with homology with the desired regions (Supplementary Table S2). The resulting linear PCR products were purified with NzyGelpure kit (NZYTech, Lisboa, Portugal), and were introduced by electroporation in the target bacteria carrying the plasmid pKOBEGA (Ap^R; Supplementary Table S1). Transformant clones were selected in LB agar plates supplemented with kanamycin (50µg/mL). The kanamycin cassette insertion was confirmed by PCR and sequencing using the appropriate primers (Supplementary Table S2).

The CRISPR/Cas9-based genome editing method (Chen et al., 2018) was adapted to inactivate pUA1135 and pUA1136 plasmids identified in S. Typhimurium variants. The pCasPA plasmid encoding the Cas9 protein and the λ-Red recombination system and the pACRISPR plasmid with the BsaI site for cloning the spacers were used. First, the bla gene that codifies the beta-lactamase in the pACRISRPR plasmid was replaced by a spectinomycin (spt) gene obtained from pSET4s plasmid (Takamatsu et al., 2001). The resulting plasmid was named pUA1148 (Supplementary Table S2). The spt gene and pACRISPR (without the bla gene) were amplified by PCR using the Phusion polymerase (ThermoFisher Scientific) with appropriate primers (Supplementary Table S2). Then, the PCR products were assembled using the NEBuilder® HiFi DNA Assembly Master Mix (New England Biolabs, Inc) and transformed by electroporation in DH5α competent cells. The transformants were selected on LB agar plates supplemented with spectinomycin (100 μg/mL) and checked by conventional PCR.

The 20 nt sequence (spacer) targeting the repA gene of target plasmids was selected using the Find CRISPR Sites tool of the Geneious Prime software. For each CRISPR site identified, off-target binding sites in the S. Typhimurium genome or pSLT plasmid were also searched. The spacer with the highest activity and specificity scores was selected. The phosphorylated and annealed primers containing the spacer sequences were designed as described (Chen et al., 2018), and cloned into the pUA1148 using the NEB® Golden Gate Assembly Kit (BsaI-HF®v2; New England Biolabs, Inc) following the manufacturer’s instructions. The assembled product pUA1148+gRNA48 (Supplementary Table S2) was transformed by electroporation in DH5α competent cells and transformants were selected on LB agar plates supplemented with spectinomycin (100 μg/mL) and checked by PCR and sequencing (Supplementary Table S2). Afterwards, 1 µg of pCasPA plasmid was transformed by electroporation in Salmonella variants and the colonies were selected on LB agar plates supplemented with tetracycline (Tet; 17 μg/mL). Then, 100 ng of pUA1148+gRNA48 were transformed by electroporation into those Salmonella electrocompetent cells and the colonies containing pCasPA and pUA1148+gRNA48 were selected on LB agar plates with tetracycline (17 μg/mL) and spectinomycin (200 μg/mL). The elimination of the target plasmid was checked by PCR with suitable primers (Supplementary Table S2). Finally, the pCasPA and pUA1148+gRNA48 were cured by sucrose counter-selection (Chen et al., 2018).

To eliminate the pUA1139 plasmid, the curing method based on λ-Red mutagenesis and I-SceI nuclease (de Moraes and Teplitski, 2015) was applied using the appropriate plasmids and primers (Supplementary Tables S1, S2). Briefly, one-step inactivation (Chaveroche et al., 2000; Datsenko and Wanner, 2000) was used to introduce the kanamycin resistance cassette (obtained from pKD4) and the restriction site for I-SceI deleting gene pUA1139_00007 from pUA1139 plasmid. At the same time, pUA1165 plasmid was constructed by removing the genes related to the λ-Red system from pKD46 and cloning the I-SceI nuclease controlled by the tetracycline inducible promoter (P_tetA) (synthesized at ATG:biosynthetics GmbH, Germany) into the NcoI restriction site using NEBuilder® HiFi DNA Assembly Master Mix (New England Biolabs, Inc) and transformed by electroporation in DH5α competent cells. The transformants were selected on LB agar plates supplemented with ampicillin (100 μg/mL) and checked by conventional PCR. Afterwards, 100 ng of pUA1165 were transformed by electroporation into the Salmonella IT2 variant containing the previously modified pUA1139 and made electrocompetent. Induction of I-SceI nuclease was done using anhydrotetracycline (AHT) as described previously (de Moraes and Teplitski, 2015), and clones that were kanamycin sensitive were checked by PCR with suitable primers. Finally, pUA1165 was cured by temperature.

To study the emergence of bacterial resistance to phages in different settings, we isolated S. Typhimurium ATCC14028 Rif^R clones from the LAB, FOOD and PT experiments, and first identified variants that exhibited reduced susceptibility to UAB_Phi78 by monitoring absorbance at A₅₅₀ of cultures infected with this phage. UAB_Phi78 was used for the initial screening due its fast multiplicate cycle (Bardina et al., 2012) and its ability to infect both smooth and rough Salmonella strains (Figure 1) (Bardina et al., 2016). From the LAB scenario, 200 and 197 isolates were randomly selected from uninfected and infected cultures, respectively, at different times after infection with the bacteriophage cocktail. All isolates from the uninfected cultures were sensitive to the UAB_Phi78 phage. However, 92% of isolates from the cocktail-infected culture displayed UAB_Phi78-reduced sensitivity 24 h after phage infection (Table 1).

In the FOOD scenario, 686 and 665 isolates were obtained from 15 untreated and 15 treated cooked ham slices, respectively (Table 2). All isolates from the untreated cooked ham slices were sensitive to UAB_Phi78 phage, whereas 21 (3.2%) isolates from the treated slices showed reduced susceptibility to this phage.

In the PT scenario, 720 and 631 isolates were recovered from the ceca of untreated and treated broilers, respectively (Colom et al., 2017) (Table 3). Most of them were sensitive to the UAB_Phi78 phage, but 70 (9.7%) of the isolates from the untreated group and 21 (3.3%) from the group treated with the cocktail showed reduced susceptibility to this phage (Table 3).

All variants with reduced UAB_Phi78-susceptibility isolated from LAB and FOOD scenarios were insensitive to UAB_Phi78 infection and their efficiency of plating (EOP) was zero. Data from the spot test analysis to the two other phages of the cocktail showed that the 96 variants isolated from treated LAB samples (Table 1) were also insensitive to infection by the other phages of the cocktail. In the case of the 21 variants isolated from treated FOOD samples (Table 2), 12 were also insensitive to UAB_Phi20 and 9 were insensitive to the three phages of the cocktail. Our initial hypothesis was that these variants were resistant to phage infection due to changes in their phage receptors, which are located in the LPS of the outer membrane. To study this, eight of these variants were selected (Table 4), and their LPS electrophoretic profile was assessed and compared with those of the parental strain S. Typhimurium ATCC 14028Rif^R and the S. Typhimurium LT2 LPS mutants (Supplementary Table S1). Results showed that the LPS electrophoretic profile of five variants isolated from both LAB (CI1, CI2 and CI3) and FOOD (HT3 and HT5) was similar to those of the rfaL-, rfaK-, rfaJ-, and rfaI- rough mutants (Figure 2). On the other hand, the HT1, HT2 and HT4 variants from FOOD showed a LPS electrophoretic profile resembling that of the rfc- rough mutant (Figure 2).

To further explore the hypothesis that the observed resistance was due to changes in phage receptors, we sequenced the genomes of the variants indicated in Table 4 and that of the parental strain ATCC 14028 Rif^R, and we mapped their sequencing reads to the reference ATCC 14028s genome (GenBank CP001363.1). Results revealed several point mutations in the genomes of all variants and the parental strain when compared to the NCBI reference genome (Supplementary Table S3). To test our hypothesis, we focused on identifying mutations in genes involved in LPS biosynthesis present in the variants but absent in the parental strain genome. In this regard, we found that the genomes of the CI2 and CI3 variants from LAB and the HT5 variant from FOOD, which display an identical phage sensitivity phenotype, contained point mutations in the rfaJ gene. Specifically, the CI2 and CI3 variants displayed a G→T transversion at position 910 of this gene, while the genome of the HT5 variant had an adenine deletion at position 742 of the rfaJ gene (Table 5). Furthermore, after PCR amplification and sequencing of the rfaJ gene, we found that the CI1 variant had the same G→T transversion identified in CI2 and CI3, while an adenine deletion at position 715 was detected in the HT3 variant (Table 5). All these point mutations give rise to a premature ochre stop codon (UAA) in the rfaJ gene.

On the other hand, the genome sequence of the HT1 variant from FOOD showed a deletion of 2,557 bp that encompassed the rfc gene. In the parental strain genome, this region is flanked by two direct repeats (DR) of 114 bp located upstream of the STM14_1615 gene (CP001363.1: 1,419,477–1,419,592) and downstream of the STM14_1618 gene (CP001363.1: 1,422,035–1,422,148) (Figure 3). The HT1 variant deletion involved the STM14_1615, rfc, STM14_1617, and STM14_1618 genes and the loss of one DR (Figure 3). Furthermore, by using PCR amplification and DNA sequencing, we searched for mutations in this DNA region in the genomes of HT2 and HT4 variants. We identified the same deletion in the HT4 variant and a C→A transversion at position 219 of the rfc gene of the HT2 variant giving rise to a premature ochre stop codon (UAA) (Table 5).

Finally, we demonstrated that complementation in trans of the rfc and rfaJ wild-type genes in the variants that presented mutations in these genes restored both the sensitivity to phages and LPS profiles seen in the parental strain.

In the PT scenario, the A₅₅₀ of cultures for S. Typhimurium ATCC 14028 Rif^R variants with reduced UAB_Phi78-susceptibility (Table 3) remained almost constant or increased after infection with the UAB_Phi78 phage, although their absorbances were lower than that of the non-infected parental strain (Supplementary Figure S1). The number of viable cells declined, especially in one of the two variants, but to a much lesser extent than observed in the UAB_Phi78-infected parental strain. These data suggested that phage infection could cause cell death of these variants, but with limited lysis activity. Spot test analyses of the two other cocktail phages revealed that, of the 70 variants with reduced susceptibility to UAB_Phi78 isolated from the untreated group (Table 3), 62 were insensitive to the other two phages of the cocktail (UAB_Phi20 and UAB_Phi87), and 8 only to the UAB_Phi87 phage. In contrast, all 21 variants isolated from the treated group (Table 3) were insensitive to UAB_Phi87 phage, but sensitive to UAB_Phi20 phage. Sixteen variants from untreated and treated animal groups were selected according to their susceptibility profile to the cocktail phages and to the date of isolation (Table 6), and their LPS electrophoretic profiles were examined. We observed no differences between the electrophoretic profiles of these variants and that of the parental strain (Supplementary Figure S2), suggesting that they have functional phage receptors. Consequently, we hypothesized that these variants might have some mechanism of interference with the multiplicative cycle of the phages. Therefore, seven of these variants (highlighted in bold in Table 6) were chosen for an in-depth study.

Firstly, we proceeded to determine the adsorption constant (K) and various infection parameters [efficiency of plating (EOP), efficiency of center of infection (ECOI) and burst size] of the three phages. The data obtained revealed that all seven variants had functional receptors for both UAB_Phi78 and UAB_Phi20 phages since K values of both phages in the variants were like those for the parental strain (Table 7). Regarding the EOP, ECOI and burst size parameters of both phages, variants were divided into two groups. One group, encompassing the IC3, IC5 and IC10 variants, was insensitive to these phages with EOP and ECOI values of 0 (Table 7). The other group, which included IC6, IC8, IT2 and IT3 variants, displayed a medium production efficiency [EOP values between 0.1 and 0.5 (Khan Mirzaei and Nilsson, 2015)] for both phages, with the exception of the IT3 variant, considered as low production efficiency [EOP values between 0.001 and 0.1 (Khan Mirzaei and Nilsson, 2015)] for UAB_Phi78 phage, and the IC6 variant, which achieved high production [EOP values ≥ 0.5 (Khan Mirzaei and Nilsson, 2015)] for the UAB_Phi20 phage. For both phages, ECOI values of the variants in this group were < 0.5 and burst sizes were lower than those obtained in the parental strain (Table 7). On the other hand, although we could determine that the UAB_Phi87 phage adsorbs to the parental strain with a low K value (2.3E-9 ± 7.5E-10 mL/min), the slopes of adsorption curves in these variants were lower than that of the parental strain, and the K values could not be calculated as described previously (Kropinski, 2009). In addition, no plaques were observed when this phage was titered on these variants.

Altogether, the results above confirmed our hypothesis that the variants isolated from PT with reduced sensitivity to phages must have some mechanism of interference that affects processes involved in phage multiplication in bacterial cells. To elucidate this, the genomes of these seven variants were sequenced. Analysis of these genome sequences revealed no point mutations or indels involving known genes related to phage defense mechanisms when compared to the genome sequence of the parental strain (Supplementary Table S3). However, analysis of unmapped Illumina reads demonstrated that all these variants carried several plasmids in addition to the pSLT plasmid, known to be present in the serovar Typhimurium of S. enterica (Jones et al., 1982). Sequencing determined that all variants carried a large conjugative IncI1α plasmid (~96-110 Kb) (Table 8). IC3, IC5, and IC10 variants harbored the pUA1135 plasmid, encoding a CTX-M-14 beta-lactamase. IC6 and IC8 variants carried the pUA1136 plasmid, encoding a TEM-1 beta-lactamase, and IT2 and IT3 variants contained the pUA1139 plasmid, which did not encode any known antibiotic-resistance mechanism (Table 8). Furthermore, some of these variants carried additional lower-size plasmids from different incompatibility groups (Table 8).

Since the IC5, IC6, and IT2 variants only carried, respectively, the pUA1135, pUA1136, and pUA1139 large plasmids, we hypothesized that these plasmids encode interference mechanisms. To demonstrate this, these variants were used as donors in conjugation experiments, using the phage-sensitive and streptomycin-resistant S. Typhimurium LB5000 strain as a recipient. Transconjugant strains receiving the pUA1135 and pUA1136 plasmids were selected on streptomycin and ampicillin plates, and streptomycin and kanamycin plates were used for transconjugants for the pUA1139 plasmid, to which a kanamycin resistance cassette had been previously introduced by one-step inactivation (Datsenko and Warner, 2000). The mating frequencies were 5.0E-3, 4.7E-3, and 9.0E-4 for the pUA1135, pUA1136, and pUA1139 plasmids, respectively. The presence of large plasmids in four transconjugants of each mating was confirmed by PCR, using specific primers for each plasmid (Supplementary Table S1). EOP and ECOI values for both UAB_Phi20 and UAB_Phi78 phages were determined for one transconjugant of each mating (Table 9). Results indicated that the acquisition by conjugation of the pUA1135 plasmid gave rise to transconjugants insensitive to both UAB_Phi20 and UAB_Phi78 phages. In addition, transconjugants that acquired the pUA1136 and pUA1139 plasmids also exhibited a reduced sensitivity to UAB_Phi20 and UAB_Phi78 phages, with EOP and ECOI values quite similar to those of the bacterial donors (Table 9). In this regard, it must be considered that the genetic backgrounds of the donor and recipient strains were different since the donor strain derived from S. Typhimurium ATCC 14028 Rif^R, and the recipient was S. Typhimurium LB5000. Concerning phage UAB_Phi87, the transconjugants exhibited a reduced sensitivity to this phage, with EOP values around 0.1-0.2 and lower size plaques compared to the ones observed on the parental strain (LB5000) (Table 9). To corroborate this, plasmids pUA1135, pUA1136, and pUA1139 were subsequently cured from IC5, IC6 and IT2 variants, respectively (Figure 4). Their loss entailed the recovery of sensitivity to both UAB_Phi20 and UAB_Phi78 phages, as evidenced by EOP values like those of S. Typhimurium ATCC 14028 Rif^R However, sensitivity to the UAB_Phi87 phage was not fully recovered, since lower EOP values (≤ 0.1) than those of the parental strains were observed.