Selection and Characterization of Phage-Resistant Mutant Strains of Listeria monocytogenes Reveal Host Genes Linked to Phage Adsorption

Abstract

Listeria-infecting phages are readily isolated from Listeria-containing environments, yet little is known about the selective forces they exert on their host. Here, we identified that two virulent phages, LP-048 and LP-125, adsorb to the surface of Listeria monocytogenes strain 10403S through different mechanisms. We isolated and sequenced, using whole-genome sequencing, 69 spontaneous mutant strains of 10403S that were resistant to either one or both phages. Mutations from 56 phage-resistant mutant strains with only a single mutation mapped to 10 genes representing five loci on the 10403S chromosome. An additional 12 mutant strains showed two mutations, and one mutant strain showed three mutations. Two of the loci, containing seven of the genes, accumulated the majority (n = 64) of the mutations. A representative mutant strain for each of the 10 genes was shown to resist phage infection through mechanisms of adsorption inhibition. Complementation of mutant strains with the associated wild-type allele was able to rescue phage susceptibility for 6 out of the 10 representative mutant strains. Wheat germ agglutinin, which specifically binds to N-acetylglucosamine, bound to 10403S and mutant strains resistant to LP-048 but did not bind to mutant strains resistant to only LP-125. We conclude that mutant strains resistant to only LP-125 lack terminal N-acetylglucosamine in their wall teichoic acid (WTA), whereas mutant strains resistant to both phages have disruptive mutations in their rhamnose biosynthesis operon but still possess N-acetylglucosamine in their WTA.

INTRODUCTION

Virulent phages have been shown to present a tremendous selective pressure on their bacterial host populations. Not only is phage predation a major driver of bacterial diversification (1, 2), but it may also select for hypermutators, which could increase the frequency of mutations in bacterial populations (3, 4). Whereas bacteria are limited to one cell division per generation, a single phage-infected cell can produce a burst ranging from less than 5 to over 1,000 progeny phages in a similar period of time (5–7). Phages consequently have the capability to rapidly outgrow their bacterial hosts and can significantly reduce or eliminate susceptible bacteria in the local environment (8, 9). Therefore, the potential for bacterial strains to persist in an environment containing lytic phages may be contingent upon that strain accumulating spontaneous mutations that grant resistance to phage infection (10). These phage-resistant mutant strains most typically resist phage infection through mechanisms of adsorption inhibition, i.e., alterations of the cell surface that affect phage attachment (11). However, one study reported that nearly all phage-resistant mutant strains of Streptococcus thermophilus had acquired CRISPR spacers that matched invading phage genomes (12); these S. thermophilus mutant strains would be expected to resist phage infection after the adsorption step. Phage-resistant mutant strains that resist infection through mechanisms of adsorption inhibition have been well characterized at the genomic level for Gram-negative bacteria (13–16); however, fewer studies address the genetics of adsorption inhibiting phage-resistant mutant strains of Gram-positive bacteria (17).

Listeria monocytogenes is a Gram-positive bacterial food-borne pathogen that causes the potentially severe disease listeriosis (18). In the United States, an annual 1,445 hospitalizations and 255 deaths are attributed to L. monocytogenes (19), with an estimated negative economic impact at over $2.5 billion (20). One strategy that is being explored to control L. monocytogenes in food and food processing environments is to exploit lytic phages as agents to kill off contaminant Listeria (21–23). However, it has been shown that Listeria populations treated with phages can give rise to phage-resistant mutant strains that can grow in the presence of the applied phages (24, 25). To our knowledge, no study to date has characterized these Listeria mutant strains beyond determining their sensitivity to phage infection.

Listeria phages have been readily isolated from environmental sources, including dairy silage (26, 27), sewage effluent (28), sheep feces (29), and food processing plants (24, 30, 31). Currently characterized Listeria phages are all members of the order Caudovirales, i.e., tailed phages, and can be organized into evolutionarily conserved groups based on morphology and genome composition (32). The host ranges of Listeria phages have been shown to often correspond to host serotypes (26, 31, 33). For example, A118 has been reported as a predominantly serotype 1/2-infecting phage, and A500 has been reported as a predominantly serotype 4b-infecting phage (33). Differences between the serotypes of Listeria can be attributed to the composition of wall teichoic acids (WTA; cell surface polysaccharides): serotype 1/2 strains are decorated with terminal rhamnose and N-acetylglucosamine (GlcNAc) residues, whereas 4b strains are decorated with terminal glucose and galactose residues (34). Nearly all Listeria phages that have been evaluated for use as biocontrol agents belong to the genus Twortlikevirus of the family Myoviridae (25, 29, 35). Two Listeria-infecting twortlikeviruses, LP-048 and LP-125, share a very high nucleotide identity (∼97% average nucleotide identity across 93% of their genomes) (32) and display broad, yet different, host ranges against a panel of L. monocytogenes isolates representing a diversity of lineages and serotypes (26). As reported here, follow-up characterization of these two Listeria phages revealed very different rates of adsorption. We hypothesized that these closely related phages attach to their hosts through different mechanisms. Thus, we selected and characterized, using whole-genome sequencing (WGS), L. monocytogenes mutant strains resistant to phages LP-048 and/or LP-125 to identify different absorption mechanisms and phage-host interactions that are associated with these two twortlikeviruses.

MATERIALS AND METHODS

Growth conditions.

Bacterial strains were grown overnight (16 ± 2 h) on brain heart infusion (BHI) (Becton, Dickinson and Company, Franklin Lakes, NJ) agar at 37°C or in BHI broth with aeration (210 rpm) at the temperature indicated in the text. Strains were stored at −80°C in BHI broth containing 15% glycerol. Strains used in this study can be found in Table 1 and Table S1 in the supplemental material.

TABLE 1.

Strains, phages, and phage susceptibility

Strain or phage	Descriptiona	Phage used to select mutant	Phage sensitivityb		WGA bindingc
			LP-048	LP-125
L. monocytogenes strains
10403S	Lineage II, serotype 1/2a		++	++	+
FSL D4-0014	10403S, nonsense mutation in LMRG_00541	LP-125	++	−	−
FSL D4-0161	FSL D4-0014::pTD01 (pPL2::LMRG_00541)		++	++	+
FSL D4-0119	10403S, nonsense mutation in LMRG_00542	LP-048	−	−	+
FSL D4-0156	FSL D4-0119::pTD02 (pPL2::LMRG_00542)		++	++	NT
FSL D4-0118	10403S, nonsense mutation in LMRG_00543	LP-048	−	−	+
FSL D4-0160	FSL D4-0118::pTD11 (pPL2::LMRG_00543 with LMRG_00542 promoter)		−	−	NT
FSL D4-0126	10403S, nonsense mutation in LMRG_00545	LP-048	−	−	+
FSL D4-0155	FSL D4-0126::pTD08 (pPL2::LMRG_00545 with LMRG_00542 promoter)		+	++	NT
FSL D4-0028	10403S, missense mutation in LMRG_00546 (amino acid change of Thr to Ile)	LP-048	−	++	+
FSL D4-0158	FSL D4-0028::pTD09 (pPL2::LMRG_00546 with LMRG_00542 promoter)		++	++	NT
FSL D4-0082	10403S, missense mutation in LMRG_01009 (amino acid change of Pro to Gln)	LP-125	++	−	−
FSL D4-0159	FSL D4-0082::pTD10 (pPL2::LMRG_01009 with LMRG_01010 promoter)		++	−	−
FSL D4-0057	10403S, missense mutation in LMRG_01319 (amino acid change of Asn to Thr)	LP-125	++	−	−
FSL D4-0153	FSL D4-0057::pTD03 (pPL2::LMRG_01319)		++	−	−
FSL D4-0068	10403S, nonsense mutation in LMRG_01697	LP-125	++	−	−
FSL D4-0154	FSL D4-0068::pTD05 (pPL2::LMRG_01697)		++	++	+
FSL D4-0065	10403S, nonsense mutation in LMRG_01698	LP-125	++	−	−
FSL D4-0162	FSL D4-0065::pTD06 (pPL2::LMRG_01698 operon)		++	++	+
FSL D4-0087	10403S, missense mutation in LMRG_01709 (amino acid change of Ile to Met)	LP-125	++	−	−
FSL D4-0163	FSL D4-0087::pTD07 (pPL2::LMRG_01709)		++	−	−
FSL R9-0915	Serotype 7		−	−	−
Phages
LP-048	Twortlike Listeria phage, shown to infect serotype 1/2, 4a, 4b, and 4c strains (24, 26, 32)
LP-125	Twortlike Listeria phage, shown to infect serotype 1/2, 3a, 3b, 4a, and 4b strains (24, 26, 32)

^aSpecific location of the mutations in the phage-resistant mutant strains listed can be found in Table S1 in the supplemental material.

^bStrong lysis (++), weak lysis (+), or no lysis (−) was observed between the indicated strain and phage over three replicate experiments. Only minor variation was observed between replicates.

^cWGA (wheat germ agglutinin) binding (+) or lack of binding (−) was determined as shown in Fig. 4. NT, not tested.

Phage lysates were prepared as previously described (26) and stored in the dark at 4°C. Phage enumeration was conducted after serial dilution with SM buffer (100 mM NaCl, 8 mM MgSO₄·7H₂O, 0.002% [wt/vol] gelatin, and 50 mM Tris-Cl adjusted to a pH of 7.5), followed by a double-agar overlay plaque assay (36) using modified LB-MOPS (LB medium buffered with 50 mM morpholinepropanesulfonic acid [MOPS] at a pH of 7.6) as previously described (26). Briefly, agar overlays were made with 0.7% (wt/vol) LB-MOPS agar supplemented to give final concentrations of 0.1% (wt/vol) glucose and 10 mM each MgCl₂ and CaCl₂; agar underlays were made with 1.5% (wt/vol) LB-MOPS also supplemented with glucose and salts. Plated phage samples were incubated at 30°C for 16 ± 2 h. Phages used in this study can be found in Table 1.

One-step growth experiments.

In order to determine the growth kinetics of phages LP-048 and LP-125, standard one-step growth experiments were performed (37). A 5-ml liquid culture of L. monocytogenes was grown in LB-MOPS to an optical density at 600 nm (OD₆₀₀) of 0.1 and then supplemented with 50 μl of each 1 M CaCl₂ and 1 M MgCl₂. Following that, 1 × 10⁸ PFU of the appropriate phage was added to the culture (multiplicity of infection [MOI] of ∼0.1). The infected culture was incubated in a water bath at 30°C with aeration. At each time point, two samples were taken; one 100-μl sample was transferred into a tube containing several drops of chloroform, and the other sample was immediately diluted and enumerated (yielding the concentration of infected cells and free viable phages), using 10403S as the titering host. At the end of the growth experiment, chloroformed phage samples were enumerated (yielding the total concentration of viable phage particles in the sample, including intracellular phages). The average burst size was calculated by dividing the average concentration of infected cells and free viable phages from the three time points following the first step of lysis (time points 90 min, 100 min, and 110 min for LP-048 and time points 80 min, 90 min, and 100 min for LP-0125) by the average concentration of infected cells and free viable phage from the first three time points postinfection (as described by Hyman and Abedon [38]).

Isolation of phage-resistant mutant strains.

Individual colonies of L. monocytogenes 10403S were used to inoculate BHI broth. The liquid cultures were incubated overnight at 30°C, and then each culture was diluted 1:100 into 5 ml of fresh BHI broth and further incubated until an OD₆₀₀ of 0.85 was reached. Following that, 50 μl of filter-sterilized 1 M CaCl₂ and 1 M MgCl₂ and 1 × 10⁸ PFU of phage were added to each culture. After an additional incubation of 24 h, one sample from each infected culture was plated on BHI agar. A single colony was isolated from each plate; colonies confirmed to resist phage infection by spot assay (described below) were stored as phage-resistant mutant strains. All phage-resistant mutant strains were subsequently grown directly from freezer stocks as liquid cultures in order to reduce the number of cell divisions and thus the likelihood of a mutation reverting a phage-resistant phenotype prior to an experiment.

DNA extraction, sequencing, and bioinformatics.

DNA was extracted from Listeria using a QIAamp DNA minikit (Qiagen, Hilden, Germany). The manufacturer's recommended protocol for DNA extraction from Gram-positive bacteria was followed with the addition of an RNase treatment. After incubation with proteinase K and prior to addition of buffer AL (Qiagen), 4 μl of RNase A (100 mg/ml; Qiagen) was added to each sample, followed by incubation at 37°C for 10 min. Genomic DNA was submitted to the Cornell University Life Science Core Laboratory Facilities where library preparation and DNA sequencing were performed. A Nextera XT DNA sample preparation kit (Illumina, Inc., San Diego, CA) was used to prepare the library, and 100-base-pair reads were obtained by sequencing the library on an Illumina HiSeq 2500 platform. Single nucleotide polymorphisms (SNPs) were called using both a reference-based and de novo variant detection method. For the reference-based method, reads were mapped against the genome sequence of L. monocytogenes 10403S (GenBank accession number NC_017544.1) with the Burrows-Wheeler Aligner (BWA), version 0.7.3 (39), using the BWA-MEM (where MEM is maximal exact match) algorithm. SNPs were called using VarScan, version 2.3.4 (40). Only SNPs with a minimal coverage of 50% of the genome-wide average coverage (GAC), a minimal variant coverage of 50% of the GAC, a minimum alternative variant frequency of 95%, and a P value of ≤0.01 were considered for further analyses. The Cortex variation assembler, cortex_var, version 1.0.5.14 (41, 42), was used for the de novo variant detection (both SNP and insertion/deletion events) as outlined by den Bakker et al. (43).

L. monocytogenes 10403S is not known to harbor any plasmids; de novo assembly of sequencing reads obtained from this study further confirmed this.

Strain construction.

Integration plasmids for complementing phage-resistant mutant strains were constructed by cloning the desired wild-type (WT) open reading frame (ORF) and the desired promoter and 5′ untranslated region (UTR) into the multiple cloning site of pPL2 (44). Constructs with promoters and 5′ UTRs fused to a downstream ORF were created by spliced overhang extension (SOE) PCR (45). PCRs for cloning were carried out using Q5 DNA polymerase (New England BioLabs [NEB], Ipswich, MA). PCRs for Sanger sequencing were carried out with GoTaq Flexi DNA polymerase obtained from Promega (Madison, WI). Restriction enzymes (BamHI, SalI, and NotI) and ligase (T4 DNA ligase) used for cloning were obtained from New England BioLabs. Plasmid constructs were first replicated in NEB 5-alpha Escherichia coli (New England BioLabs) and then extracted with a Plasmid Midi Kit (Qiagen) and confirmed by Sanger sequencing at the Cornell University Life Science Core Laboratory Facilities (Ithaca, NY). The constructs were then transferred into L. monocytogenes either by conjugation with E. coli SM10 and selection of streptomycin- and chloramphenicol-resistant colonies or by electroporation (46). Constructed strains are shown in Table 1.

Spot tests and adsorption assays.

Spot tests of both LP-048 and LP-125 were conducted, as three independent replicates, on bacterial strains to determine the strains' susceptibilities to phage infection. Five microliters of phage lysate at 1 × 10⁸ PFU/ml was spotted in duplicate on duplicate lawns and then incubated at 30°C for 16 ± 2 h. Spots were then evaluated for strong lysis (++), some lysis (+), or no lysis (−).

Adsorption of LP-048 and LP-125 to bacteria was determined by enumeration of viable phages that failed to adsorb to the test bacteria after coincubation. Fifty-microliter volumes of bacterial culture grown at 30°C for 16 h (OD₆₀₀ values ranged from 1.4 to 1.7) were transferred into centrifuge tubes containing 912 μl of BHI broth, 20 μl of phage lysate at 1 × 10⁹ PFU/ml, 9 μl of 1 M CaCl₂, and 9 μl of 1 M MgCl₂ (salts were added immediately prior to the addition of bacteria). The bacteria and phage mixtures were incubated for 15 min at 30°C with aeration. Following that, bacteria and any adsorbed phages were sedimented by centrifugation at 17,000 × g for 1 min in an Eppendorf microcentrifuge 5417C (Hamburg, Germany). The supernatants were then filtered through 0.2-μm-pore size surfactant-free cellulose acetate (SFCA) syringe filters (Thermo Fisher Scientific, Waltham, MA). Viable phages left in the filtrates were enumerated. The percent adsorption was defined as the loss of phages (percent) from each sample after coincubation with bacteria, centrifugation, and filtration compared to the value for the sample with the greatest concentration of that respective phage remaining in the filtrate; the samples with the highest concentrations were set as 0% adsorption for the respective phage in the respective replicate experiment (these samples were not always the BHI broth controls). One-way analysis of variance (ANOVA) was used to analyze the effect of strain (WT 10403S, phage-resistant mutant strains, FSL R9-0915, and the BHI broth control were included in the analysis) on phage adsorption, and a Dunnett's post hoc test (α = 0.05) was used to identify significant differences in adsorption percentages between WT 10403S and the mutant strains and controls. In order to determine whether the complemented mutants showed partially restored phage adsorption, a t test (assuming unequal variances; α = 0.05) was performed between the value for each mutant strain (that showed significantly different phage adsorption from the WT) and that of the respective complemented mutant. All statistical analyses of phage adsorption were performed separately for LP-048 and LP-125 with JMP statistical software (JMP11; SAS Institute, Inc., Cary, NC).

WGA binding assay.

A wheat germ agglutinin (WGA)-Alexa Fluor 488 conjugate (Life Technologies, Carlsbad, CA) was used to detect the binding of WGA to Listeria. To fix cells, 17 μl of 16% (wt/vol), methanol-free, formaldehyde solution (Thermo Scientific) was added to 50 μl of an overnight Listeria culture (grown at 37°C), followed by incubation at room temperature for 15 min. Cells were then sedimented by centrifugation at 2,655 × g for 5 min in an Eppendorf microcentrifuge 5417C and resuspended in 100 μl of phosphate-buffered saline (PBS). The suspension was then mixed with 1 μl of WGA-Alexa Fluor 488 conjugate (1 mg/ml) and incubated for 15 min at room temperature. The samples were then sedimented again (same conditions) and resuspended in 100 μl of PBS. Bacterial cells were then mounted on glass slides and imaged on a confocal laser scanning microscope (Carl Zeiss, Peabody, MA).

Genome sequencing data accession number.

The raw sequencing reads generated in this study have been deposited in the Sequence Read Archive under BioProject number PRJNA261154.

RESULTS

One-step growth curves reveal different adsorption rates for LP-048 and LP-125.

To determine differences in infection kinetics of LP-048 and LP-125, one-step growth curves were performed on the serotype 1/2a L. monocytogenes strain 10403S (Fig. 1). The most striking difference between the two phages was observed in their adsorption rates. Whereas after 20 min of coincubation with host bacteria 78.2% (2.2% standard error [SE]) of LP-048 adsorbed to the host bacteria (Fig. 1A), 99.9% (0.0% SE) of LP-125 adsorbed to the host in the same period of time (Fig. 1B); this indicates a less efficient adsorption of LP-048 to 10403S under these conditions. Possible explanations for this difference could be different concentrations of the available receptors for the two phages or differences in the affinities of the receptors for the two phages; these explanations are consistent with observations of phage lambda adsorption under various receptor concentrations and receptor affinities (47). The eclipse period, defined as the period of time taken for the first viable phage particles to mature postinfection, was between 40 and 45 min for LP-048 and between 35 and 40 min for LP-125. The latent period, defined as the time taken for the infected cell to lyse postinfection, was between 55 and 60 min for LP-048 and between 50 and 55 min for LP-125. The average burst size, defined as the average number of phage particles produced per infected cell, was 13.6 (SE, 3.1) for LP-048 and 21.3 (SE, 4.5) for LP-125.

FIG 1.

One-step growth experiments of Listeria phages LP-048 (A) and LP-125 (B). Triangles are values for samples that were directly plated, representing the cumulative concentration of infected host cells and unadsorbed viable phages. Circles are values for samples that were treated with chloroform prior to plating, representing the total concentration of viable phages (including intracellular phage). Time point 0 values represent the theoretical input of phage, which was calculated by averaging values from the first three time points of directly plated samples. An initial drop in titer of chloroformed samples indicates adsorption of phage to bacteria. All values are the arithmetic means of three independent experiments, and error bars indicate the standard errors.

Whole-genome sequencing of phage-resistant mutant strains reveals host genes essential for phage infection.

Phage-resistant mutant strains derived from L. monocytogenes 10403S were selected for by confrontation with either LP-048 or LP-125. Out of a total of 110 confrontations, 95 resulted in the isolation of phage-resistant mutant strains. These mutant strains were later screened by spot assay, which confirmed them as true phage-resistant mutant strains. Sixty-nine phage-resistant mutant strains, as well as the parent strain 10403S, were sequenced on an Illumina HiSeq platform. Mutations that were detected in sequenced strains were mapped against the 10403S reference genome (Fig. 2A). A total of 83 mutations were identified; three mutations were each found in two separate mutant strains (see Table S1 in the supplemental material). Therefore, 80 unique mutations were identified. Fifty-six mutant strains showed a single mutation, 12 mutant strains showed two mutations, and 1 mutant strain showed three mutations. Mutations from phage-resistant mutant strains with only a single mutation were surmised to be the mutations most likely to cause a phage-resistant phenotype; these were termed mutations of interest (shown in red in Fig. 2). Out of the 80 unique mutations identified, 67 were found in 10 genes, located in five chromosomal loci (Table 2 shows the distribution of these mutations among the 10 genes); mutations from all 56 mutant strains with a single mutation were found in these five loci. In addition to these 67 mutations, 13 other mutations were found outside these five loci. All 13 of these mutations were found in mutant strains that contained more than one mutation and were thus not further characterized. These 13 mutations included five synonymous substitutions (shown in green in Fig. 2A) and seven other mutations (shown in blue in Fig. 2A) as well as one SNP found in an intergenic region flanked by LMRG_01577 and LMRG_01588 (shown in green in Fig. 2A) (see Table S1 in the supplemental material for details on these mutations).

FIG 2.

Mutations from phage-resistant mutant strains mapped against the 10403S reference genome (A) and key loci on the 10403S chromosome (B). Individual spontaneous mutations are identified as colored ovals. Red ovals represent mutations in strains with only a single mutation. Blue ovals represent nonsense mutations, missense mutations, frameshift mutations, or mutations in regulatory DNA sequences from mutant strains with more than one mutation. Green ovals represent synonymous mutations (no change in amino acid sequence) or an SNP in a featureless intergenic region from a mutant strain with more than one mutation. DEL, deletion.

TABLE 2.

Genes with mutations in phage-resistant L. monocytogenes strains

Genea	EGD-e homolog	Locusb	Function	No. of unique mutationsc
				Total	Mutation typed				Selected by:
					FS	MS	NS	O	LP-048	LP-125
LMRG_00541	lmo1079	I	Putative membrane protein	19	8	0	8	3	0	19
LMRG_00542	lmo1080	I	Putative glycosyltransferase (GT-A type)	16	2	9	4	1	13	3
LMRG_00543	lmo1081	I	Glucose-1-phosphate thymidylyltransferase (RmlA)	9	3	4	2	0	7	2
LMRG_00545	lmo1083	I	dTDP-glucose 4,6-dehydratase (RmlB)	8	0	6	1	1	6	2
LMRG_00546	lmo1084	I	dTDP-4-dehydrorhamnose reductase (RmlD)	1	0	1	0	0	1	0
LMRG_01319	lmo1647	II	1-Acyl-sn-glycerol-3-phosphate acyltransferase	1	0	1	0	0	0	1
LMRG_01009	lmo1862	III	Putative lipase/acylhydrolase	1	0	1	0	0	0	1
LMRG_01709	lmo2538	IV	Uracil phosphoribosyltransferase	1	0	1	0	1	0	1
LMRG_01697	lmo2550	V	Putative glycosyltransferase (family 2)	8	3	3	1	1	0	8
LMRG_01698	lmo2549	V	Cell wall teichoic acid glycosylation protein (GtcA-like)	3	0	1	2	0	0	3

^aOnly 10 genes of interest are listed here. A full list of all genes with mutations is included in Table S1 in the supplemental material.

^bLocus designation in this study (Fig. 2).

^cNumber of unique mutations found in this study that are present in the listed genes. See Table S1 for mutation-specific details.

^dFS, frameshift; MS, missense; NS, nonsense; O, other (large insertion or deletion, mutation in promoter, or mutation in the 5′ UTR).

A representative mutant strain for each of the 10 genes containing a mutation of interest was selected for further characterization; if possible the mutant strain containing the most upstream nonsense mutation in the gene of interest was selected as representative (Table 1). When these 10 representative mutant strains were characterized for phage susceptibility by spot assays, all mutant strains were found to be resistant to either one or both phages (Table 1). Sanger sequencing confirmed the mutations of interest in all 10 mutant strains. Additionally, each representative mutant strain was complemented in trans with the wild-type (WT) allele of the respective gene in which the mutation is located; phage susceptibility was at least partially restored in 6 out of 10 of the complemented mutants (as detailed below).

Two loci accumulated a majority of the unique mutations found in the strains sequenced in this study (64/80). One of these two loci (locus I) contains six genes, five of which accumulated a total of 53 unique mutations (Fig. 2B; Table 2). LMRG_00541, the first gene in the locus, encodes a putative membrane protein and makes up a one-gene operon that accumulated 19 unique mutations identified here (Table 2); all 19 mutations in LMRG_00541 were selected for in the presence of LP-125. Eight of the mutations in LMRG_00541 were single-base-pair deletion frameshift mutations, four of which were likely phase variants as the deletions occurred in homopolymeric tracts (≥6 bp in length) of adenine or thymine. One of these putative phase variants, at base position 1095105, was found in two sequenced strains (see Table S1 in the supplemental material). Another putative phase variant that was also found in two sequenced strains was located between the −10 and −35 promoter signals of LMRG_00541 (see Table S1 in the supplemental material). This mutation was also a single nucleotide deletion in a homopolymeric tract. This deletion reduced the gap between promoter signals from 17 nucleotides to 16 nucleotides, which could affect transcription of the operon. The representative mutant strain for LMRG_00541, FSL D4-0014, showed resistance to LP-125 but remained susceptible to LP-048; complementation of the mutation in trans with the WT allele of LMRG_00541 restored susceptibility of the mutant strain to LP-125 (Table 1).

The second operon in locus I (Fig. 2B) contains five genes, four of which accumulated a total of 34 unique mutations identified here (Table 2). LMRG_00542 encodes a putative GT-A type glycosyltransferase and accumulated a total of 16 unique mutations, 13 of which were selected for by LP-048. LMRG_00543, LMRG_00545, and LMRG_00546 encode rhamnose biosynthesis enzymes and together accumulated a total of 18 unique mutations, 14 of which were selected for by LP-048 (Table 2). One of the mutations, in LMRG_00542 (strain FSL D4-0114), is a likely phase variant as it is a single-base-pair deletion in a seven-base-pair homopolymeric tract of adenine residues (see Table S1 in the supplemental material). Whereas the LMRG_00542, LMRG_00543, and LMRG-00545 mutant strains were all shown to be resistant to both LP-048 and LP-125 by spot assay, the LMRG_00546 mutant strain was resistant to only LP-048. Complementation with the respective WT alleles restored phage susceptibility for the LMRG_00542, LMRG_00545, and LMRG_00546 mutant strains (Table 1) but not for the LMRG_00543 mutant strain. Failure to successfully complement the LMRG_00543 mutation could be due to a polar effect.

The other locus that accumulated a considerable number of mutations (n = 11), locus V, represents a two-gene operon that includes LMRG_01697, which encodes a putative glycosyltransferase, and LMRG_01698, which encodes a GtcA-like WTA glycosylation protein. One frameshift mutation, in LMRG_01697, was identified as a putative phase variant (strain D4-0093); this mutation consisted of an insertion of two nucleotides (TA) that extended a dinucleotide tandem repeat from four to five repeats. Another mutation in this locus was identified as a large deletion (431 bp) starting at the end of the flanking gene LMRG_01696 (encoding transcription termination factor Rho) and ending in the beginning of LMRG_01697. Representative mutant strains of both LMRG_01697 and LMRG_01698 were found to be resistant to only LP-125 by spot assay; complementation of mutations in both representative mutant strains with the respective WT alleles restored phage susceptibility (Table 1).

Three mutations of interest each mapped to a different locus on the chromosome (loci II, III, and IV) (Fig. 2); all of these mutations were nonsynonymous substitutions (Table 1 gives the amino acid substitutions). The three genes with these mutations were LMRG_01009, which encodes a putative lipase/acylhydrolase, LMRG_01319, which encodes a 1-acyl-sn-glycerol-3-phosphate acyltransferase, and LMRG_01709, which encodes a uracil phosphoribosyltransferase (Table 2). All three mutant strains were found to be resistant to only LP-125 by spot assay. Complementation of the mutations found in these mutant strains with the respective WT alleles failed to restore phage susceptibility (Table 1).

Phage-resistant mutant strains of L. monocytogenes resist phage by adsorption inhibition.

To determine if LP-048 and LP-125 could adsorb to the phage-resistant mutant strains isolated in this study, adsorption assays were performed for the parent strain and each representative mutant strain and its respective complemented strain (Fig. 3). After a 15-min coincubation, over 95% of both LP-048 and LP-125 adsorbed to WT 10403S; for the phage-resistant mutant strains, adsorption was severely reduced (Fig. 3). For example, mutant strains resistant to both LP-048 and LP-125, which have mutations in LMRG_00542, LMRG_00543, and LMRG_00545, showed ≤10% adsorption for both phages (Fig. 3). FSL D4-0028, which has a mutation in LMRG_00546 and is resistant to only LP-048, showed 2.7% (2.7% SE) adsorption of LP-48 and 37.3% (11.5% SE) adsorption of LP-125. All mutant strains resistant to only LP-125 showed ≤25% adsorption of LP-125 and ≥99% adsorption of LP-048 (Fig. 3). The six complemented mutants that showed restored phage susceptibility by spot assay, with mutations in LMRG_00541, LMRG_00542, LMRG_00545, LMRG_00546, LMRG_01697, and LMRG_01698, also showed restoration of phage adsorption although FSL D4-0155, which was complemented with a WT LMRG_00545 allele, showed that LP-048 adsorption was restored to only 68.9% (6.5% SE) (Fig. 3).

FIG 3.

Adsorption of phages LP-048 (blue) and LP-125 (red) to the L. monocytogenes parent strain 10403S, phage-resistant mutant strains, and complemented mutant strains. Phage-resistant mutant strains (−) and their respective complemented mutants (+) are adjacent to one another, and brackets above the bars are labeled with the affected genes. Striping on bars representing mutant strains and negative-control strains indicates a significant difference in adsorption percent values from the WT 10403S value (P < 0.0001). An asterisk (*) above a complemented mutant strain bar indicates that the complemented mutant showed a significant increase of adsorption percentage compared to that of the phage-resistant mutant strain (tested only for phage-resistant mutant strain and phage combinations for which the data were significantly different from the WT 10403S data). Values shown are the arithmetic means of three independent experiments, and error bars indicate the standard errors.

Phage-resistant mutant strains resistant to LP-125 and susceptible to LP-048 lack terminal N-acetylglucosamine in their wall teichoic acid.

As LMRG_01697 and LMRG_01698 (lmo2549 and lmo2550 in strain EGD-e) have been previously shown to be necessary for decoration of WTA with N-acetylglucosamine (GlcNAc) residues (53), we tested the representative phage-resistant mutant strains isolated in this study for the presence of terminal GlcNAc in their WTA. To do this, we determined whether the lectin wheat germ agglutinin (WGA), which specifically binds to GlcNAc, could bind to selected L. monocytogenes mutant strains that showed phage resistance. The serotype 7 L. monocytogenes strain FSL R9-0915, which was shown to resist infection to both LP-048 and LP-125 by spot assay (Table 1) and was adsorption deficient for both phages (Fig. 3), was included as a negative control; a serotype 7 strain was selected as the negative control as serotype 7 strains lack any substituents in their WTA polyol phosphate chains (34). As expected, we observed WGA binding to the L. monocytogenes parent strain 10403S (Fig. 4 and Table 1) but not to FSL R9-0915 or the representative LMRG_01697 and LMRG_01698 mutant strains. All six representative mutant strains resistant to LP-125 and susceptible to LP-048 (including the representative LMRG_01697 and LMRG_01698 mutant strains) did not bind WGA (Table 1; Fig. 4). All mutants complemented with the LMRG_00541, LMRG_01697, and LMRG_01698 WT alleles showed restoration of WGA binding (Table 1 and Fig. 4), whereas the mutants complemented with the LMRG_01009, LMRG_01319, and LMRG_01709 WT alleles did not show restoration of WGA binding, consistent with a lack of restoration of phage sensitivity in these mutants (Table 1 and Fig. 3). All four tested representative mutant strains that were resistant to LP-048 did bind WGA, suggesting that they still possessed terminal GlcNAc residues in their WTA (Table 1 and Fig. 4).

FIG 4.

Binding of wheat germ agglutinin (WGA)-Alexa Fluor 488 conjugate and bacterial cells observed by laser scanning microscopy. Fluorescent images are displayed in the left column, and differential interference contrast (DIC) microscopy images of the same field are shown in the middle column. The merged images show that either all or none of the cells fluoresce. Images are shown for selected strains; all 10 mutant strains and their respective complemented strains were tested. Strain FSL D4-0014 has a nonsense mutation in LMRG_00541; strain FSL D4-0119 has a nonsense mutation in LMRG_00542. Scale bar, 10 μm.

DISCUSSION

In this study, we fully sequenced an unprecedented 69 phage-resistant mutant strains of L. monocytogenes, which was followed by further genetic and phenotypic characterization of selected mutant strains. Our data show (i) that mutations of interest accumulated primarily in two chromosomal loci and in a total of 10 genes, (ii) that six genes were conclusively linked to phage adsorption, including three genes conclusively shown to contribute to WTA decoration, and (iii) that evidence of phase variation existed in three of the genes linked to phage adsorption. Overall, our results provide insight into phage-resistant mutant strains of L. monocytogenes and improve our understanding of the evolution of this important pathogen. Our results also will guide future studies needed to further assess the benefits and consequences of using phages as biocontrol agents.

Mutations associated with phage resistance were found primarily in two loci, which contain key genes linked to phage adsorption.

Phage-resistant mutant strains in many different phage-host systems have been shown to resist phage infection through mechanisms of adsorption inhibition (11, 14, 17), as opposed to mechanisms that inhibit phage DNA entry, replication, or escape (49). Similarly, we found that phage-resistant mutant strains of L. monocytogenes showed significant reduction in the adsorption of one or both phages used in this study. N-Acetylglucosamine (GlcNAc) has been previously characterized as a phage receptor for L. monocytogenes (33, 50–52); consistent with these previous observations, we found eight and three mutations, respectively, in the genes LMRG_01697 and LMRG_01698, both of which have been shown to be necessary for glycosylation of terminal GlcNAc in the wall teichoic acid of L. monocytogenes (53). However, a majority of mutations (n = 53) were found in a region not previously linked to phage susceptibility (locus I). Interestingly, genes within the two operons of locus I were associated with different phage resistance phenotypes. A nonsense mutation in LMRG_00541 was shown to affect LP-125 adsorption and glycosylation of WTA with terminal GlcNAc residues (as supported by WGA binding experiments) and was shown not to affect LP-048 adsorption; the same phenotype was observed in the LMRG_01697 and LMRG_01698 mutant strains. Mutations in genes from the neighboring operon (i.e., LMRG_00542, LMRG_00543, LMRG_00545 and LMRG_00546; here, the LMRG_00542 operon) were all shown to affect both LP-125 and LP-048 adsorption and did not affect glycosylation of WTA with GlcNAc. Interestingly, the missense mutation found in LMRG_00546 did not affect resistance to LP-125 in spot assay experiments; this mutation also had a lesser effect on the adsorption of LP-125 than other mutations in the operon. The LMRG_00542 operon encodes orthologs of well-characterized proteins with a role in rhamnose biosynthesis, such as RmlA (glucose-1-phosphate thymidylyltransferase; orthologous to LMRG_00543) (54), RmlB (dTDP-glucose 4,6-dehydratase; orthologous to LMRG_00545) (55), and RmlD (dTDP-4-dehydrorhamnose reductase; orthologous to LMRG_00546) (56); all of these are enzymes essential for the conversion of glucose-1-phosphate to dTDP-l-rhamnose (57). While Zhang et al. (58) proposed that this operon was responsible for the Listeria serovar-specific biosynthesis of the sugar nucleotide precursor needed for rhamnose decoration of WTA and while den Bakker et al. (59) identified it as belonging to a putative O-antigen-determining cluster, phenotypic studies are still needed to confirm that these enzymes contribute to rhamnose decoration of WTA. Our data that mutations in the rhamnose biosynthesis operon cause phage resistance is consistent, though, with previous observations that rhamnose inhibited Listeria phage A118 adsorption (33) and with a study by Habann et al. (52), which showed that the A511 receptor-binding protein binds to N-acetylglucosamine and suggested that it also binds to rhamnose. Although we did not restore phage susceptibility by complementing the nonsense mutation in LMRG_00543 (which encodes the RmlA ortholog), this failure to complement was most likely due to the mutation causing a polar effect on downstream genes LMRG_00545 and LMRG_00546 (which we conclusively linked to phage susceptibility and adsorption through complementation experiments). The nonsense mutation early in LMRG_00543 would likely leave a considerable length of mRNA free of ribosomes (>800 bp); such an effect could increase the probability of Rho-dependent transcriptional termination (60). Alternatively, but considerably less likely, the construct used to complement the mutation may not have expressed WT LMRG_00543 as intended, or rhamnose biosynthesis may not be necessary for phage susceptibility.

All the three phage-resistant mutant strains that contained the missense mutations of interest that were not successfully complemented (LMRG_01009, LMRG_01319, and LMRG_01709) were resistant to LP-125 and lacked terminal GlcNAc residues in their WTA. It is possible that phage susceptibility could not be restored by complementation for these mutant strains because their respective mutations had a dominant effect over the respective WT allele; it is also possible that these mutations are not responsible for the observed phage resistance phenotype and that resistance is due to polar effects or other, nonidentified, mutations. However, lmo2537 (the LMRG_01710 homolog in 10403S) encodes UDP-N-acetylglucosamine 2-epimerase, which is a precursor of the teichoic acid linkage unit (61, 62). This suggests that LMRG_01709 may have a direct effect on the composition of WTA and phage resistance as LMRG_01709 is part of the same operon as LMRG_01710; however, we cannot definitively exclude a polar effect on LMRG_01710. Future experiments to further characterize these three mutant strains will be necessary to confirm and understand specific functional links between the mutations and phage adsorption.

The five synonymous mutations identified in this study were not further characterized as they were not the sole mutations found in their respective phage-resistant mutant strains. Although unlikely, these mutations may still play a role in phage resistance as synonymous mutations can affect cellular processes (e.g., translation efficiency due to codon biases or mRNA structures) (63); future studies on these mutant strains will be needed to address this.

None of the mutations identified here mapped to either of the CRISPR systems present in 10403S (CRISPR-II or RliB-CRISPR) (64), leading us to conclude that phage-resistant mutant strains from this study did not develop CRISPR-mediated phage resistance. As our de novo assembly-based genome sequence analyses did identify a 60-bp insertion and two large deletions (130 bp and 420 bp), we are confident that our methodology would have detected the acquisition of a new CRISPR spacer in any of the phage-resistant mutant strains sequenced in this study.

Three genes conclusively linked to phage adsorption show evidence for phase variation.

Phase variation is a mechanism that has been shown to provide transient resistance to phage infection for Gram-negative bacteria (14, 65). We identified seven unique putative phase variant mutations in this study, two of which were each found in two separate mutant strains. All of the putative phase variants we identified would be generated by the general mechanism of slipped-strand mispairing (66), as opposed to other mechanisms of phase variation such as DNA rearrangement or gene conversion (67). Six of the putative phase variants were single nucleotide deletions in adenine or thymine homopolymeric tracts (one in the LMRG_00541 promoter, four in LMRG_00541, and one in LMRG_00542). These mutations are very similar to the phase variants found in phage-resistant Vibrio cholerae strains (14), where single nucleotide deletions in poly(A) tracts were identified in O1 antigen biosynthesis genes. One of the putative phase variants identified here extended a TA dinucleotide tandem repeat in LMRG_01697 from four to five repeats. A similar phase variation, which involved the loss or gain of a single repeat within a tetranucleotide tandem repeat, was found in Staphylococcus aureus icaG, which is linked to production of the polysaccharide adhesin β-1-6-linked N-acetylglucosamine (68); phage resistance phenotypes of this phase variation was not evaluated though. Previously, Orsi et al. (69) identified phase variation within a homopolymeric tract of seven adenine residues found within the internalin A gene (inlA) of L. monocytogenes. Based on the data reported by Orsi et al. (69) on the frequency of phase variation, we hypothesize that the putative phase variants identified here are much more likely than other frameshift mutations to revert to a WT genotype (restoring the full open reading frame). If so, such mutations could provide a transient genotypic escape for L. monocytogenes from phage predation; after the selective pressure of phage infection passes, any mutant strains that reverted back to the WT genotype would have the opportunity to outcompete the phage-resistant mutant strains.

Identification of distinct mutations in phage-resistant mutant strains of L. monocytogenes provides initial insight into the types of phage-resistant mutant strains that may emerge after exposure to or treatment with Listeria phages.

Whole-genome sequencing (WGS) has recently been used to identify mutations in phage-resistant mutants of Escherichia coli (13), Vibrio cholerae (14, 70), Bacillus anthracis (17), and Synechococcus (15), demonstrating the power of next-generation sequencing in improving our understanding of phage-host interactions. In this study, WGS of 69 mutant strains enabled us to observe the parallel evolution of phage resistance in L. monocytogenes 10403S under the selective pressure of lytic phages, as evidenced by mutations repeatedly and independently occurring in the same genes and causing the same phenotypic effects. As WTA have been shown to be associated with virulence functions (71), including the evidence that a teichoic acid biosynthesis gene is essential for virulence of L. monocytogenes in mice (62), further studies to determine how the mutations found in this study affect virulence, as well as fitness, under a variety of different environmental and stress conditions will be valuable. Additional studies should also examine the frequency of occurrence and stability of the mutations identified here across different environmental conditions. Together, these types of studies will facilitate better assessment of the safety of using Listeria phage products for control applications and will provide insight on how phages may drive the evolution and pathogenicity of L. monocytogenes. Information on conditional expression of genes linked here to phage adsorption (72) could also provide insight on whether L. monocytogenes is capable of escaping phage predation by gaining physiological refuge (where a bacterium becomes transiently resistant to phage infection). Transient phage resistance has been reported for E. coli; in one study phage resistance was induced under maltose-deficient conditions (73), while in another study phage resistance was induced by the quorum-sensing signal N-acyl-l-homoserine lactone (74). Additionally, phage-resistant mutant strains identified in this study may prove to be useful as hosts for isolating new phages that adsorb to different surface features and as screening tools to identify Listeria phages that use the same receptors as LP-048 and LP-125.

Conclusions.

Under the selective pressure of virulent phages, strains of L. monocytogenes harboring spontaneous mutations that grant phage resistance will survive and outcompete the susceptible parental strains. All phage-resistant mutant strains of L. monocytogenes from this study were shown to resist phage infection through mechanisms of adsorption inhibition. Postadsorption mechanisms of phage resistance, such as CRISPR-mediated phage immunity, were not found in any of the phage-resistant mutant strains; however, several of the mutations found were identified as putative phase variants, suggesting that phase variation may be an important genetic mechanism for the survival of L. monocytogenes under phage-mediated selective pressure.