Comparative Genomic Analysis Reveals That the 20K and 38K Prophages in Listeria monocytogenes Serovar 4a Strains Lm850658 and M7 Contribute to Genetic Diversity but Not to Virulence

Abstract

Listeria monocytogenes is a foodborne pathogen of considerable genetic diversity with varying pathogenicity. Initially, we found that the strain M7 was far less pathogenic than the strain Lm850658 though both are serovar 4a strains belonging to the lineage III. Comparative genomic approaches were then attempted to decipher the genetic basis that might govern the strain-dependent pathotypes. There are 2,761 coding sequences of 100% nucleotide identity between the two strains, accounting for 95.7% of the total genes in Lm850658 and 92.7% in M7. Lm850658 contains 33 specific genes, including a novel 20K prophage whereas strain M7 has 130 specific genes, including two large prophages (38K and 44K). To examine the roles of these specific prophages in pathogenicity, the 20K and 38K prophages were deleted from their respective strains. There were virtually no differences of pathogenicity between the deletion mutants and their parent strains, although some putative virulent factors like VirB4 are present in the 20K region or holin-lysin in the 38K region. In silico PCR analysis of 29 listeria genomes show that only strain SLCC2540 has the same 18 bp integration hotspot as Lm850658, whereas the sequence identity of their 20K prophages is very low (21.3%). The 38K and 44K prophages are located in two other different hotspots and are conserved in low virulent strains M7, HCC23, and L99. In conclusion, the 20K and 38K prophages of L. monocytogenes serovar 4a strains Lm850658 and M7 are not related to virulence but contribute to genetic diversity.

Introduction

Listeria monocytogenes is an opportunistic, intracellular pathogen that causes foodborne infections in humans [15,16]. Immunocompromised individuals, such as the elderly, pregnant, and neonates, are at particular risk for listerial infections with high mortality (up to 30%) due to ingestion of contaminated foods [21]. It has been considered as a severe threat to public health.

L. monocytogenes contains 13 serotypes and can be divided into four genetic lineages (I, II, III, and IV) based on multigene phylogenetic analysis [35,44]. Lineage I is primarily composed of serotypes 4b, 1/2b, and 3b; lineage II has 1/2a, 1/2c, and 3a; and lineages III and IV have serotypes 4a and 4c [34]. Lineage I and II strains account for the majority of human listeriosis cases (about 98%) [41], whereas lineage III and IV strains are rare and predominantly identified in animals [19,44,48].

Lineage III strains are believed to have a low pathogenic phenotype [13,35]. Early studies on L. monocytogenes strains from different lineages indicate that the major force driving the attenuated phenotype of lineage III strains, such as serotype 4a L99 and HCC23, is reductive evolution [19,38]. Acquisition and adaptation of prophage genes were speculated as novel virulence-associated factors of L. monocytogenes [9,26]. These accessory genes mainly represent gene-scale differences, resulting in short-term evolution [32]. However, phage-mediated horizontal transfer contributes not only to increased pathogenicity but also to inactivation of genes, which might reduce the pathogenicity of the microbes of interests if the genes inactivated are related, directly or indirectly, to virulence [1,3,10,11,42].

From the genomic sequencing data of two lineage III strains (Lm850658 and M7) that are substantially conserved in their genomes but different in virulence, we found three specific prophages in the two strains; one designated as the 20K prophage is specific in Lm850658, and the other two as 38K and 44K prophages are present in M7. By knockout in combination with virulence-related tests, we conclude that the 20K and 38K prophages contributed only to genetic diversity, but not to pathogenicity.

 

Materials and Methods

Bacterial Strains, Plasmids, Cells, and Culture Conditions

L. monocytogenes strain Lm850658 was a clinical isolate from diseased sheep in Australia (Dr. John Bowman) and M7 was isolated from pasteurized milk [8]. Both strains belong to lineage III serotype 4a. The L. monocytogenes strains were cultured in brain heart infusion medium (BHI; Oxoid, UK). Escherichia coli DH5α was employed as the host strain for cloning vectors and grown at 37℃ in Luria-Bertani (LB; Oxoid) medium. Stock solutions of ampicillin (50 mg/ml), gentamicin (50 mg/ml), and chloramphenicol (10 mg/ml) (Sangon Biotech, China) were added to media, where appropriate, at the required levels. Murine macrophage cell line RAW264.7, intestinal epithelial cell line Caco-2, and mouse fibroblast L929 were cultured at 37℃ and 5% CO2 in complete cell culture medium (RPMI 1640 medium; Gibco, USA) containing 10% fetal bovine serum (Gibco, USA).

Multilocus Sequence Typing (MLST)

Nine unlinked genes (including seven housekeeping genes gyrB, dapE, hisJ, ribC, purM, gap, and tuf, and two stress-response genes sigB and betL) were used for MLST analysis [5]. For each locus, an allele number was given to each distinct sequence variant, and a distinct sequence type (ST) number was given to each distinct combination of alleles of the nine genes. MEGA 4.0 was used to construct a neighbor-joining tree based on the concatenated sequences of nine loci with 1,000 bootstrap tests [43].

Genomic Sequencing, Annotation, and Comparison

High-quality DNA sample from the strain Lm850658 was prepared using the proteinase K and chloroform method (Genomic DNA Extraction Reagent; Sangon Biotech, China) as recommended by the manufacturer. Ten micrograms of genomic DNA was sheared either to a mean size of 300 bp for constructing a paired-end genomic DNA library or to a mean size of 2,000 bp for a mate-paired genomic DNA library (http://www.illumina.com). The two libraries were sequenced on an Illumina HiSeq 2000 Sequencer at the Berry Genomics Co. Ltd. (Beijing, China), and reads were discarded if the proportion of low-quality bases was beyond 50% or the ambiguous bases (N) beyond 10%.

A total of 10,443,517 mate-paired and 6,276,435 paired-end 100 bp reads from the strain Lm850658 genome were collected for de novo assembly using Velvet ver. 1.2.07 (http://www.ebi.ac.uk/~zerbino/velvet/) [47]. The best assembly was obtained using a k-mer of 85 (as determined by the VelvetOptimiser script) and comprised 67 contigs totaling 2,902,910 bp with an N50 of 327,182. These contigs were blasted against the genome of L. monocytogenes M7 and scaffolded into a circle with 19 gaps. PCR and sequencing were performed to confirm all these gaps.

GeneMark (http://exon.gatech.edu/) [50] was invoked to run an ab initio protein coding gene prediction based on the gene module of L. monocytogenes. The protein sequences were then blasted against the nonredundant protein database. All these sequences could hit to the genes in the M7 genome with high identities. Therefore, the M7 genes were used to annotate those of the strain Lm850658. The tRNA genes were predicted by the tRNAscan-SE ver. 1.3.1 [30,37] and rRNA genes by the RNAmmer ver. 1.2 [27] online service.

Deduced amino acid sequences of ORFs in the annotated genomes of strains Lm850658 and M7 were aligned by their amino acid sequences, and the identity codes were presented as I, M, L, and G, which mean coding sequence identity 100%, nonsynonymous mutation, premature truncation, and gaps in the ORF, respectively. Differences were classified according to the identity codes (Tables S5 and S6).

Construction of Mutant Strains

A homologous recombination strategy was used for mutant construction with the prophage region deletion in L. monocytogenes wild-type strain Lm850658 or M7, according to the protocol described previously [7,31]. Primers used to generate homologous fragments and to identify the deletions are listed in Table S1. The homologous fragments were gel-purified and ligated to pMD18-T vector. After sequencing confirmation, the inserted fragments were digested and subcloned into the temperature-sensitive shuttle vector pKSV7 and transformed into E. coli DH5α. Plasmids containing inserts were subsequently extracted and electroporated into L. monocytogenes competent cells. Transformants were grown at a non-permissive temperature (41°C) in BHI containing chloramphenicol (10 μg/ml) to promote chromosomal integration. The putative homologous recombinants were passaged successively in BHI without antibiotic at a permissive temperature (30°C) to enable plasmid excision and curing [4]. The recombinants were verified by PCR and sequencing.

Transcriptional Analysis Using Quantitative Real-Time PCR

Overnight cultures of L. monocytogenes were inoculated into fresh BHI broth and grown to the stationary phase (OD600nm = 0.6) at 37℃. RNA was prepared using the Trizol method (Sangon Biotech, China) and cDNA was synthesized with reverse transcriptase (TOYOBO, Japan). Quantitative RT-PCR was then performed in 20 μl reaction mixtures containing SYBR green qPCR mix (TOYOBO) to detect the transcriptional levels in the iCycler iQ5 real-time PCR detection system (Bio-Rad, USA) with specific primer pairs (Table S1). The housekeeping gene gyrB was used as an internal control for normalization as previously described [6,7]. RNA was extracted from three separate cultures and all PCRs were performed in three independent experiments.

Adhesion and Invasion Assays in Caco-2 Cells

Adhesion and invasion assays were conducted according to our previous protocol [22]. Overnight-cultured bacteria were harvested by centrifugation (5,000 ×g for 10 min), resuspended in 10 mM phosphate buffer saline (PBS, pH 7.4), and adjusted to 0.25 at OD600 nm. The Caco-2 cells were grown at 37℃ and 5% CO2 for 18-24 h to confluence (about 2 × 105) in 12-well plates (Corning, USA) and then infected with L. monocytogenes (MOI = 10) for 1 h. For adhesion, cells were lysed after three times of washing with PBS. For estimation of invasion, cells were washed with PBS after 1 h infection and incubated for an additional hour in complete cell culture medium containing 200 μg/ml gentamicin. The cells were lysed and 10-fold diluted for plating on BHI agar. The agar plates were incubated overnight at 37℃ for colony counting. Adhesion was expressed as the ratio of recovered colonies to those inoculated; whereas invasion was calculated as the ratio of recovered colonies after gentamicin treatment to those inoculated.

Plaque Formation Assay

Plaque-forming assay was performed on L929 cell monolayers in 6-well plates (Corning) according to the method described previously [18,40]. Briefly, overnight-cultured bacteria were added to cell monolayers (approximately 80% confluence) at MOI of 0.1. Extracellular bacteria were then killed with 200 μg/ml gentamicin for 1 h at 37℃ and 5% CO2. The cells were washed with PBS for three times and then overlaid with 3 ml of complete cell culture medium (without phenol red) containing 0.7% agarose and 20 μg/ml gentamicin. After 3 days of incubation, plaques were visualized by adding 2 ml of cell culture medium containing 0.7% agarose and 0.01% filter-sterilized neutral red (Sigma, Germany).

Confocal Imaging for Visualization of Actin Tails

The Caco-2 cells were cultured in complete cell culture medium for 12 h. Overnight cultures were treated as described above and adjusted to 0.25 at OD600nm. Cells were infected at MOI of 10 and incubated at 37℃ and 5% CO2 for 1 h. Extracellular bacteria were then killed with 200 μg/ml gentamicin for 1 h and incubated for an additional 5 h. Cells were washed gently with PBS, fixed with 4% paraformaldehyde, and then permeabilized with 0.5% Triton X-100. The bacterial cells were stained with polyclonal antibodies to L. monocytogenes (prepared in our laboratory) for 1 h at 37℃, washed twice with PBS, and probed with Alexa Fluor 488 conjugated donkey anti-rabbit antibody (Abcam, UK) for 1 h at 37℃. The F-actin was then stained with 6.6 μM phalloidin-Alexa Fluor 568 (Abcam). DAPI (4’,6-diamidino-2-phenylindole; Invitrogen, USA) was used to stain the cell nuclei. Actin tail formation was visualized by confocal microscopy (Olympus FLV 1000, Japan).

Virulence to Mice

The virulence of L. monocytogenes strains Lm850658 and M7 to mice was conducted as previously reported [28] with modifications. Fifty-six femal ICR mice weighing 20-22 g (Zhejiang College of Traditional Chinese Medicine, China) were randomly divided into seven groups of eight mice per group. Mice were acclimatized for 3 days in a standard class II laboratory animal facility. Overnight bacterial cultures were treated as above and adjusted to 0.6 at OD600nm. The actual bacterial numbers inoculated were determined by plating the serially diluted inoculum of each strain on BHI agar. Each strain was inoculated into three groups, with the last group receiving no inocula as sham control. Each mouse was inoculated intragastrically with 2 × 108 CFU bacteria. One group of mice from each strain test were euthanized at 24 h post infection (hpi) and another group at 48 hpi. The liver and spleen samples were homogenated for serial dilution and plate counting on BHI agar plates. The remaining two inoculated groups and the control mice were observed for mortalities twice a day for 7 days.

The LD50 assay on immunocompromised ICR mice was conducted as previously reported [48]. Briefly, all mice received intraperitoneal injection with 200 mg/kg carrageenan 24 h before challenge [39]. The mice (5 mice per group) were inoculated intraperitoneally with serial 10-fold dilutions of bacteria in PBS. Control mice received PBS only. The LD50 values were calculated by using the trimmed Spearman-Karber method on the mortality data during the 10 day observation period. All animal experiments in this study were approved by the Laboratory Animal Management Committee of Zhejiang University (Approval No. 20131025).

Statistical Analysis

All results were presented as the mean ± SD of triplicate experiments and data were analyzed using the two-tailed Student’s t-test with p values <0.05 as statistically significant and p values <0.01 as marked statistical significance.

 

Results

Different Virulence of Lineage III Strains Lm850658 and M7 in Mouse Model

The phylogenetic tree constructed by the neighbor-joining method based on concatenated gene fragment gyrB-dapE-hisJ-sigB-ribC-purM-betL-gap-tuf [5] shows that M7 and Lm850658 were located in the same branch (Fig. 1). All mice infected with strain M7 remained alive during the 7-day observation period, while 75% of those infected with strain Lm850658 died within 1 week (Fig. 2A), suggesting that the strain Lm850658 is more virulent than M7 although both belong to lineage III strains that are usually considered as low pathogenic [19,35,48]. Intragastric infection showed that the bacterial load in liver and spleen samples of the M7-infected mice was less than the Lm850658-infected counterparts at 24 hpi and became significant at 48 hpi (p < 0.01) (Figs. 2B and 2C).

Fig. 1.Neighbor-joining cladogram of L. monocytogenes strains based on the concatenated sequence gyrB-dapE-hisJ-sigB-ribC-purMbetL-gap-tuf.

Fig. 2.L. monocytogenes M7 is far less pathogenic than strain Lm850658.

Comparative Genomic Analysis of Strains Lm850658 and M7

The complete genome of L. monocytogenes Lm850658 was deposited in GenBank under the accession number CP009242 and was compared with our previous published genome CP002816 for strain M7 [8]. The general features of the sequenced circular genomes of L. monocytogenes Lm850658 and M7 are compared in Table 1. Although they are different in virulence in the mouse model, their genomes are remarkably syntenic: genome size, G+C content, percentage coding, and average length of protein-coding genes are similar between the two lineage III strains. Both genomes harbor six complete copies of rRNA operons (16S-23S-5S). Neither of them contains any mobile genetic element. The number of tRNA-coding genes in the strain Lm850658 is 72, but only 67 in M7.

Table 1.General genomic features of L. monocytogenes strains Lm850658 and M7.

Total coding sequences (CDS) of the strain Lm850658 and M7 are 2,885 and 2,978, respectively (Table 2). There are 2,761 CDSs that are identical in both strains, accounting for 95.7% of the total genes in strain Lm850658 and 92.7% in M7. There are 19 premature truncation CDSs in Lm850658 but 15 in M7. Thirty-three genes are specific in Lm850658, and 130 are specific in M7, thus contributing to the difference of their chromosome size. The specific genes in Lm850658 consist of 10 phage-associated genes and a 20,099 bp cryptic region containing 23 genes, which is actually a novel prophage (designated as 20K prophage) since there is no homologous sequence in the NCBI database (Tables S2 and S3). The specific genes in M7 are contained in two prophages; one in the 38,447 bp region covering 52 genes (designated as 38K prophage) and the other in the 43,920 bp region containing 69 genes (designated as 44K prophage) (Table S4). The 38K prophage is also present in other lineage III strains HCC23 and L99. However, the 44K prophage is seen in strains of other lineages as well, but with different sequence similarity. The 20K, 38K, and 44K prophages are integrated into genomes by the specific sequences known as hotspots in L. monocytogenes genomes [26]. There are six CDSs with gaps (Table S5) and 66 CDSs with non-synonymous mutations (Table S6) in both genomes. Besides this, there are 3,135 SNPs in Lm850658 relative to M7, which result in 66 non-synonymous mutation genes and 21 truncated genes in Lm850658 (Tables S5 and S6).

Table 2.Comparison of coding genes of L. monocytogenes strains Lm850658 and M7.

The 20K and 38K Prophages in L. monocytogenes Strains Lm850658 and M7 Did Not Contribute to Pathogenicity

A previous study speculated based on recurring patterns that frequent loss or truncation of genes could be vital for pathogenicity [19]. Since the 20K prophage in the virulent strain Lm850658 is absent in low-virulent M7 strain, we were curious whether genes in this novel prophage were regulated by the virulence regulator PrfA and contribute to pathogenicity. We mutated Lm850658 PrfA as the M7 version, which is constitutively activated owing to mutations at G145S [45]. Fig. 3 shows that PrfA constitutive activation did not change the transcriptional level of the 20K genes. This was within our expectations since we did not find putative PrfA binding sites upstream of these prophage genes. To confirm the possible roles of the 20K prophage in L. monocytogenes virulence, we constructed a 20K deletion mutant by homologous recombination. Adhesion and invasion assays show that deletion of the 20K prophage did not affect the ability of L. monocytogenes to infect Caco-2 cells (Figs. 4A and 4B). Such deletion did not affect actin-tail formation in Caco-2 cells (Fig. 4C), plaque formation in L929 cells (Fig. 4D), and virulence to mice (Fig. 4E), as compared with its parent strain Lm850658. Therefore, we conclude that the 20K prophage does not contribute to pathogenicity of the strain Lm850658.

Fig. 3.Major genes in 20K prophage were not regulated by virulent regulator PrfA in L. monocytogenes Lm850658.

Fig. 4.Deletion of the 20K and 38K prophages did not affect pathogenic traits of L. monocytogenes.

Next, we attempted to examine if the 38K and 44K prophages in the strain M7 are involved in low virulence phenotypes. We tried to knock out 38K and 44K prophages in M7. Finally, we only obtained the 38K deletion mutant, but failed in deleting the 44K prophage. Infection assays in cultured cells and in a mouse model indicate that the 38K prophage in the strain M7 did not affect its low virulence property as well (Fig. 4), suggesting that integration of this prophage into its host genome does not affect the genes that may be related to virulence.

Diversity of Prophages in L. monocytogenes Genomes

To uncover the genetic characteristics of these prophages, we analyzed the prophage sequences of Listeria genomes downloaded from the NCBI database. Among 29 Listeria genomes, only Lm850658 and SLCC2540 contain 20K prophages at the same 18 bp hotspot (TATATATTATGT AAACTA) and the nucleotide identity between the prophages is only 21.3% (Table 3). Four genomes have the 38K prophages at a hotspot located in tRNALys, and five other genomes contain a short sequence of about 300 bp. Nine genomes have the 44K prophages at another hotspot located in tRNAArg with their nucleotide identity ranging from 41.5% to 99.9% in comparison with the 44K prophage of strain M7 (Table 3). Genes in the 44K prophage among these strains vary widely from 1.6% to 100% by nucleotide identity (Table S7). These data indicate that the integration sites and the prophages in Listeria genomes are diverse and could largely expand the pan genome of L. monocytogenes.

Table 3.aGenomes of listed strains were downloaded from http://www.ncbi.nlm.nih.gov, and partial genomes are not listed. bIntegrated sequence identity (ISI) of the 20K (TATATATTATGTAAACTA), 38K (ACTCTTAATCAGCGGGTCGG GGGTTCGAAACCCTCACAACCCATA), and 44K (AATCCCTCTCAGGACGTTA) prophages, respectively, that have equivalent genomic location in diferent L. monocytogenes strains. cProphage identity (PI) to the 20K of Lm850658, or 38K and 44K of M7.

 

Discussion

Lineage III strains of L. monocytogenes are generally considered as low pathogenic [4]. Early studies have indicated that the attenuated phenotype of the lineage III strains might result from reductive evolution [19,26]. We compared the genomes of representative lineage III strains Lm850658 and M7, which were clustered in the same branch by multilocus sequence typing but with different pathogenicity (Table S9). Comparative genomic analysis further indicates that their genomes are similar in size, percentage coding, and average length of coding genes, though there are numerous SNPs across their genomes as well as three prophages; the 20K one in Lm850658, and those of 38K and 44K in M7.

Prophages are the sources of horizontal gene transfer among different species to facilitate their adaptability to environments [24,46]. They are located in defined chromosomal regions called hypervariable hotspots, constituting the primary loci of gene-scale species evolution [10,24,33]. We found that the three prophages in the two strains were integrated in different hotspots in the genomes (Table 3). These hotspots occupy the equivalent genomic locations in other L. monocytogenes strains. Although L. monocytogenes SLCC2540 has a prophage with similar size to the 20K of Lm850658 and their sequences of integration sites are identical, the low nucleotide identity (21.3%) of their coding sequences indicates a different source of acquisition. The 38K and 44K prophages are highly conserved in lineage III strains M7, HCC23, and L99. However, the 38K prophage is different from the one in L. innocua Clip 11262 [17], and the 44K prophage, from those of other listerial genomes (Tables 3, S7, and S8). Therefore, we provide clear evidence that these prophages, particularly the novel 20K one, contribute to the genetic diversity of L. monocytogenes.

Acquisition of prophages could empower bacteria to cope with hostile environments [24,46], and might contribute to pathogenic traits either by upregulation or by downregulation of virulence-related genes, which can be verified by targeted deletion [9,11,20,36,49]. Hain et al. [19] showed that deletion of two prophage genes (lmaB and lmaD) in 1/2a strain EGDe resulted in reduced growth in a murine infection model. It has been suggested that less pathogenic lineage III strains differ from other lineage strains mainly by loss of genes involved in metabolism, stress resistance, and surface-associated functions involved in adaptation to the complex host environments [12-14,19,26]. We found that within lineage III, strains also differed significantly in their virulence, a finding similar to an early report [29]. The novel 20K prophage attracted our attention because it contains genes encoding putative virulence factors such as VirB4, a component of the type 4 secretion system [25,49]. However, deletion of the whole 20K region did not affect the virulence of Lm850658, indicating that this prophage is not involved in pathogenicity.

It has been shown that prophage integration may also lead to inactivation of genes, hence impacting the pathogenicity when the genes inactivated are involved in virulence [1,3,9-11,42]. Of the 38K and 44K prophages in low-virulent strain M7 that are also contained in strains HCC23 and L99 [19,26], we were able to knock out the whole 38K region. Deletion of this region did not enhance the virulence of M7 to mice, suggesting that this prophage does not contain genes that negatively regulate virulence genes or their products and does not contribute to its low pathogenic phenotype. In Pseudomonas aeruginosa, the phage protein Tip was found to suppress the bacterial ATPase required for type IV pilus assembly, thus lessening its motility and possibly virulence [9].

In summary, this study clearly indicates that the 20K and 38K prophages in the lineage III strains Lm850658 and M7 contribute to genetic diversity, but not to virulence. The key players governing the low pathogenic phenotypes of strain M7 await further investigation, probably by transcriptomic profiling or by examining key SNPs on genes other than prfA, the mutation at 433A of which results in constitutive activated PrfA145S [8]. Constitutive activated PrfAM7 does result in PrfA-regulated factors being overexpressed, but it makes no difference on the virulence of M7 (data not show). Particular attention should be given to the genes involved in cell wall/cell surface metabolism or molecules related to protein trafficking and anchoring, since our preliminary analysis has indicated abnormal “fall-off’ into the culture supernatant of InlB, an important adhesion and invasion molecule that binds to lipoteichoic acid by its GW motif [2,23].