Comparison between Listeria sensu stricto and Listeria sensu lato strains identifies novel determinants involved in infection

The human pathogen L. monocytogenes and the animal pathogen L. ivanovii, together with four other species isolated from symptom-free animals, form the “Listeria sensu stricto” clade. The members of the second clade, “Listeria sensu lato”, are believed to be solely environmental bacteria without the ability to colonize mammalian hosts. To identify novel determinants that contribute to infection by L. monocytogenes, the causative agent of the foodborne disease listeriosis, we performed a genome comparison of the two clades and found 151 candidate genes that are conserved in the Listeria sensu stricto species. Two factors were investigated further in vitro and in vivo. A mutant lacking an ATP-binding cassette transporter exhibited defective adhesion and invasion of human Caco-2 cells. Using a mouse model of foodborne L. monocytogenes infection, a reduced number of the mutant strain compared to the parental strain was observed in the small intestine and the liver. Another mutant with a defective 1,2-propanediol degradation pathway showed reduced persistence in the stool of infected mice, suggesting a role of 1,2-propanediol as a carbon and energy source of listeriae during infection. These findings reveal the relevance of novel factors for the colonization process of L. monocytogenes.


Listeria phylogenetic groups exhibit differences in colonization ability after oral infection in female BALB/c mice. To analyze the proliferation of different species from the Listeria sensu stricto and
Listeria sensu lato groups in the gastrointestinal tract and in other organs, female BALB/c mice were orally infected with 4-9 × 10 8 cell forming units (cfu) of the Listeria sensu stricto species L. monocytogenes and L. welshimeri, as well as the Listeria sensu lato species L. aquatica and L. booriae. The number of luminal Listeria in the ileum and colon was determined 2 days post infection (p.i.) ( Fig. 1a and b). There were significantly lower cfu numbers of the three non-pathogenic species in both compartments in comparison with the cfu numbers of L. monocytogenes strain EGDe. The cfu values of L. aquatica and L. welshimeri were similar in the ileum. L. booriae was barely detectable in both compartments and showed significantly lower mean values in comparison to L. welshimeri in the colon, but not in the ileum.
The number of tissue-associated bacteria, that is, in the mucus, the epithelial layer, or the lamina propria (LP) of the gastrointestinal tract, was also analyzed ( Fig. 1c and d). While L. monocytogenes was found in high numbers of 2.5-6.9 × 10 3 cfu, the three other species were not detectable in these tissues, confirming their non-pathogenic behavior as reported previously for L. welshimeri in mice after intravenous injection 33 . As expected, L. monocytogenes could also reach the mesenteric lymph nodes and spread to the liver and spleen of infected mice ( Supplementary Fig. 1).
The cfu/mg feces from stool samples taken 3 h p.i., 1 day p.i., and 2 days p.i. of the same mice were additionally analyzed. No significant difference in the cfu/mg feces was observed either immediately after infection or within 24 h of infection. In all cases, the number of cfu decreased 1000-fold, suggesting a species-independent initial clearance of listeriae shed in the feces. After an additional 24 h, L. monocytogenes persisted in stool pellets, but the number of cfu decreased significantly for the three non-pathogenic species (Fig. 1e). This effect was more pronounced in the Listeria sensu lato species. For example, for half of the mice fed L. aquatica, no cfu were detected, and only 1 of 6 mice fed L. booriae were still shedding Listeria.
Genome comparison between Listeria sensu stricto and Listeria sensu lato species reveals candidate genes for infection. The findings described above prompted us to identify as yet unknown genes associated with gut colonization by a genome sequence comparison of 16 out of the 17 known Listeria species type strains (all except L. grayi). For this purpose, we calculated the percentage of predicted amino acid sequence identity for each L. monocytogenes protein against the proteins of the other 15 species within both clades. Gene products with at least 70% sequence homology amongst the Listeria sensu stricto species were chosen for further analysis, and filtered against gene products with less than 30% sequence homology amongst the Listeria sensu lato species. Although homology of >40% is suggestive of functional homologs 34 , we chose the more stringent threshold to focus on the most robust candidate genes. The resulting Table 1 comprises 151 gene products that are present in all clade I species and absent (or significantly divergent) in clade II species. A majority of the gene products are involved in flagellar biosynthesis and the cobalamin (vitamin B 12 )-dependent utilization of ethanolamine and 1,2-PD. The gain of these clusters has already been described as an evolutionary event that contributed to the taxonomic definition of clade I 35 . Motility is well known to contribute to successful infection by L. monocytogenes [36][37][38][39][40] . Many candidate genes are involved in cell wall and membrane biogenesis, transport processes, transcription, and signaling (Table 1). There is experimental evidence that many of the genes specific for listeriae of clade I contribute to virulence properties, thus validating our approach: Two of the factors involved in peptidoglycan biosynthesis, Lmo0703 and Lmo0717, have been demonstrated to be controlled by the listerial virulence regulators DegU 41  Lmo2186 with putative heme transport capacity. SpvA is a listerial virulence factor that promotes escape from phagosomes of macrophages 42 and/or enables crossing of the digestive barrier 43 . The ATPase synthase encoded by lmo0090-lmo0093 is known to play a role in intracellular survival 44 , and deletion of the gene pdeD contributes to a decrease in listerial invasiveness in enterocytes 45 . Because of their predicted, but not experimentally evaluated function in vitro or in vivo, we chose the putative ATP-binding cassette (ABC) transporter Lmo1131-1132 and the 1,2-PD utilization genes from Table 1 for further functional and infection experiments.
The operon lmo1131-1132 encoding a putative ABC transporter facilitates adhesion and invasion of L. monocytogenes in Caco-2 cells. According to the Basic local alignment search tool on the NCBI homepage, the operon lmo1131-1132 encodes a putative ABC transporter. Both proteins Lmo1131 and Lmo1132 contain an ABC transporter-like transmembrane domain as well as a nucleotide binding domain. For further analysis of the functional role of the operon lmo1131-1132, we constructed the in-frame deletion mutant L. monocytogenes EGDe Δlmo1131-1132 and tested adhesive and invasive capabilities in cell culture assays using Caco-2 (human colon carcinoma) cells and HEp-2 (human larynx squamous cell carcinoma) cells. Adhesion to Caco-2 cells was reduced in the deletion mutant compared with that of the parental strain EGDe by a factor of 2 (Fig. 2a). In contrast, cell culture assays with HEp-2 cells revealed no significant adhesion differences between EGDe and EGDe Δlmo1131-1132 (Fig. 2b), suggesting a specific role of the ABC transporter in the interaction with Caco-2 cells. We also performed a gentamycin protection assay to test the invasion and the intracellular replication properties of the deletion mutant in Caco-2 and HEp-2 cells. Given that the number of intracellular cells detected 1 h after cell culture infection is a measure of the capacity of a strain to enter cells, we observed a more than 20-fold reduction of invasion of Caco-2 cells, but not of HEp-2 cells, by EGDe Δlmo1131-1132 compared with that of the parental strain ( Fig. 2c and d). The intracellular replication rates of EGDe [doubling time T d (EGDe) = 83 min] and EGDe Δlmo1131-1132 [T d (EGDe Δlmo1131-1132) = 83 min] however, were identical in the Caco-2 cells, indicating a specific role of the putative transporter in adhesion and/or invasion of this cell type. Members of the two listerial phylogenetic groups exhibit different cell numbers in the ileum and colon lumen. Female BALB/c mice were orally infected with 4-9 × 10 8 cfu of the Listeria sensu stricto species L. monocytogenes (•) and L. welshimeri (■), as well as of the Listeria sensu lato species L. aquatica (▲) and L. booriae (▼). After 2 days, cfu numbers for each species were determined in the ileum (a) and colon lumen (b), as well as in the ileum (c) and colon tissue (d). Stool samples (e) were collected at 3 h p.i., 1 day p.i. and 2 days p.i., and the number of cfu per mg feces was calculated. Symbols represent values for individual mice, while horizontal lines indicate the mean value that was pooled from two separate experiments (n = 3 mice per group). Dashed lines represent the detection limit for each sample. Statistical significance was assessed using two-tailed student's t-test with Welch's correction. Transcription of lmo1131-1132 shows oxygen-and temperature-dependent regulation. We then examined the transcription of lmo1131 and lmo1132 in L. monocytogenes EGDe at 24 °C and 37 °C under both aerobic and anaerobic growth conditions via quantitative real time (qRT)-polymerase chain reaction (PCR) (Fig. 3). The two transporter genes exhibited similar transcription levels under all four conditions, suggesting transcription of the operon from a common promoter located upstream of lmo1131. Setting the transcriptional level of both genes at 37 °C with oxygen as 100%, we observed a decrease of the lmo1131 and lmo1132 mRNA gene name/lmo number

Lack of lmo1131-1132 leads to attenuated colonization of BALB/c mice.
To verify the relevance of the candidate genes for the in vivo infection process, a food-borne transmission model in mice was used to enable study of the gastrointestinal phase of listeriosis 46 . We conducted co-infection experiments with female BALB/c mice aged 6-8 weeks using a 1:1 ratio of EGDe and its mutant EGDe Δlmo1131-1132 lacking the putative transporter genes. The strains were chromosomally tagged with pIMC3kan (EGDe) and pIMC3ery (mutant). Mice were fed a total of 1 × 10 9 cfu/mouse and the number of Listeria in the lumen of either the ileum or the colon was analyzed 2 days p.i. (Fig. 4a and b). The colon contained 25-fold more L. monocytogenes than the ileum, as was previously reported by others 46,47 . However, there was no significant difference in the ratio of the two strains recovered from the intestinal contents of either tissue. We then assessed the numbers of cfu associated with the flushed tissue, which includes L. monocytogenes trapped in the mucus layer, bacteria adhered to or inside intestinal epithelial cells (ECs) and bacteria that have reached the underlying LP. In both the ileum and the colon, a bimodal distribution was observed, with little or no L. monocytogenes recovered in approximately half of the mice, and up to 10 4 cfu in the other half (Fig. 4c). In the ileum, the average number of cfu for the latter group was six-fold higher for wildtype EGDe compared with the mutant. Overall, the competitive index (CI) revealed that the wild type strain outcompeted the mutant strain six-fold in the ileum (Fig. 4d). Interestingly, there was no significant difference between EGDe and EGDe Δlmo1131-1132 in the colon. To further investigate this difference in intestinal colonization, the gut tissues of infected mice were fractionated, and the number of cfu in either the mucus layer, ECs or the LP were determined (Supplementary Fig. 2a). The number of wild type and EGDe Δlmo1131-1132 in the mucus and the epithelial layer of the ileum did not differ significantly. However, wild type L. monocytogenes outnumbered the deletion mutant by five-fold in underlying LP of the ileum, similar to what was observed for whole tissue. As expected, there were no differences in the competitive indexes for any of the colon fractions. To assess the systemic spread of both strains, we also determined the bacterial load in the mesenteric lymph nodes (MLN), spleen and liver 2 days p.i. As shown in Fig. 4e and f, five-fold fewer EGDe Δlmo1131-1132 were recovered from the liver compared to the wild type strain. There was no significant difference between the two strains in the MLN or the spleen. To test whether dissemination of the transporter mutant to peripheral organs was impaired or if the mutant strain had a specific growth defect in the liver, mice were intravenously injected with a 1:1 mixture of EGDe and EGDe Δlmo1131-1132, and the total cfu in liver and spleen were determined 2 days p.i. Approximately equal numbers of both strains were recovered from the liver when the transmission route bypassed the gut ( Supplementary Fig. 2b). Together, these data suggest that the transporter Lmo1131-1132 is involved in efficient translocation across the intestinal mucosa in the small intestine, but not the large intestine.
The pduD gene of L. monocytogenes is required for growth with 1,2-PD. Three large gene clusters responsible for the cobalamin-dependent utilization of ethanolamine and 1,2-PD were identified to be specific to the clade sensu stricto species of the genus Listeria 22 (Table 1). These pathways have been demonstrated to play a role in the proliferation of S. Typhimurium in vitro and in vivo [48][49][50] , whereas no such experimental data have been available for L. monocytogenes. To investigate the role of 1,2-PD as a potential carbon and energy source for L. monocytogenes, strain EGDe and its mutant EGDe ΔpduD, which lacks a gene essential for 1,2-PD utilization, were grown in minimal medium 51 (MM) without glucose, but with 0.5% (w/v) yeast extract under anaerobic conditions at 37 °C. No significant differences in the growth behavior of the two strains were observed. However, when we added 1,2-PD and cobalamin, which is an essential cofactor of PduCDE and whose biosynthesis is encoded by the cob/cbi gene cluster, EGDe grew to an optical density (OD) 600 max = 0.35, whereas the cell density of EGDe ΔpduD did not exceed that in the medium without 1,2-PD (Fig. 5a). Therefore, it might be speculated that all nutrients preferred over 1,2-PD are used up during the first 4 h, and that L. monocytogenes only then degrades 1,2-PD to reach a higher cell density.
As a control, we compared the growth of both strains in MM containing 50 mM glucose in the absence and presence of 1,2-PD and cobalamin. Under these growth conditions with glucose as an energy source, the addition of 1,2-PD had no significant effect on the cell density of either EGDe or EGDe ΔpduD (Fig. 5b), suggesting that 1,2-PD is utilized by L. monocytogenes only in the absence of glucose.
Supplementation with 1,2-PD and/or cobalamin leads to induction of pdu and cob/cbi genes during the stationary phase. To investigate the transcriptional response of the pdu and the cob/cbi gene clusters to the presence of 1,2-PD and/or cobalamin, a global transcriptome analysis of L. monocytogenes EGDe was conducted via next-generation sequencing (NGS). During growth in brain heart infusion (BHI) medium, in BHI medium with 10 mM 1,2-PD, or in BHI medium with 10 mM 1,2-PD and 25 nM cobalamin, cells were harvested in the mid-logarithmic (exponential) and stationary phases. During the exponential phase, the pdu    Table S1). However, when 1,2-PD was added to the cultures, a strong increase of expression of most pdu and cob/cbi genes in the range from two-fold to 1,605-fold induction was observed (Fig. 6). The transcription of pdu genes was even more elevated (up to more than 10,942-fold) when cobalamin was added, and the presence of this cofactor in the medium resulted in repression of the genes responsible for its biosynthesis (Supplementary Table S1), a finding that confirms a regulatory model in which the presence of cobalamin represses its own synthesis post-transcriptionally via stabilization of an mRNA hairpin 52 . The transcription levels of pocR, the regulator of both clusters, were only slightly elevated under medium supplemented with 1,2-PD and/ or cobalamin. The eut genes of the ethanolamine degradation cluster, which are positioned between the pdu and cbi/cob clusters, were not significantly induced under the growth conditions applied (Supplementary Table S1).
The transcriptome data were succesfully validated via qRT-PCR for the selected genes lmo1146, pocR (lmo1150), pduC (lmo1153), lmo1190 and cbiA (lmo1199) in stationary phase ( Table 2). In exponential phase, there is a discrepancy between qRT-PCR and NGS results due to the higher sensitivity of qRT-PCR and the overall low number of RPKM values from the NGS (Suppl. Table S1). The results of both the transcriptome analysis and the qRT-PCR are compatible with the observation that the addition of 1,2-PD and cobalamin to L. monocytogenes EGDe in MM led to a growth advantage after exponential phase when other nutrients had been used up (Fig. 5a).

Deletion of pduD leads to faster clearance in BALB/c mice.
We hypothesized that the ability to metabolize 1,2-PD would enhance the ability of L. monocytogenes to compete with gut microbiota and to persist in the intestinal lumen following oral transmission. To test this, female BALB/c mice were co-infected with a 1:1 ratio of wild type EGDe and an isogenic ΔpduD mutant (total of 1 × 10 9 cfu/mouse) that were differentially tagged with antibiotic resistances. Stools were collected daily for up to 14 days p.i., and the total number of strain per mg of feces was determined. As expected, a significant portion of the inoculum was shed in feces within 3 hours after feeding the mice, and the number of cfu detected 21 hours later had decreased considerably (Fig. 7a). At each time point after that, the average number of ΔpduD mutant bacteria recovered was less than the wildtype strain. By 10 days post-infection, the ΔpduD mutant had been cleared, but wildtype bacteria continued to be shed in the feces in at least some of the mice for up to 4 more days. Accordingly, the competitive index decreased steadily over the course of the experiment with a maximum difference on day 8 when the wild type bacteria outcompeted the mutant bacteria by nearly 40-fold (Fig. 7b). Thus, the presence of the genes encoding the 1,2-PD utilization pathway prolonged the survival of L. monocytogenes EGDe in the mouse gastrointestinal tract.

Discussion
It is believed that a common ancestor of the Listeria sensu stricto group took up a set of virulence genes through horizontal gene transfer, and that these determinants were subsequently lost in the non-pathogenic species of this listerial clade I 53,54 . Other putative evolutionary processes differentiating Listeria sensu stricto from Listeria sensu lato are the expansion of internalin genes and the acquisition of the metabolically relevant eut/pdu/cbi/cob gene cluster as well as of flagellar genes, which were probably taken up via horizontal gene transfer from an ancestor of the Bacillus cereus complex 22 . The association of Listeria sensu stricto with fecal samples or the gastrointestinal tract of mostly symptom-free animals or food from animal origins [23][24][25][26][27] , a reduced colonization rate of the environmental species L. fleischmannii and L. floridensis compared to Listeria sensu stricto species in fecal samples from wild rodents 55 , as well as the alleviated colonization abilities of the Listeria sensu lato species in our BALB/c infection experiments, prompted us to speculate that a genome comparison of clade I versus clade II strains might identify as yet unknown listerial factors that play a role particularly during the gastrointestinal phase of infection.
Mice and other animal model organisms are considerably more resistant to oral infections than humans, and larger inocula need to be administered (10 9 -10 11 cfu/animal) to cause an intestinal infection 56 . A high species specificity of the surface proteins InlA, which interacts well with ECs of humans or guinea pigs, but not with ECs of mice 57,58 , contributes to inefficient oral transmission. Recent studies also show that competition with gut microbiota may limit opportunities for L. monocytogenes to invade the gut mucosa 47    relevance of as yet unknown listerial determinants for the gastrointestinal phase of listeriosis in more detail, we used a recently established and highly reproducible feeding model to mimic the natural course of infection 46 that overcomes the lack of reproducibility encountered by the conventional administration of listeriae by oral gavage. Gastrointestinal infection is a multifactorial process and, besides the already mentioned stress conditions, metabolism is one of the key factors of pathogenic bacteria that can only survive in the intestine by searching for a specific niche providing sufficient amounts of nutrients 60 . A hallmark of pathogens is therefore their metabolic flexibility during infection. Examples of this for L. monocytogenes include the utilization of sugar phosphates and glycerol as alternative carbon sources during intracellular replication [61][62][63] , the requirement of both thiamine uptake and biosynthesis of thiamine precursors 64 , or the possible exploitation of uncommon nitrogen sources such as ethanolamine or glucosamine 65 . L. monocytogenes possesses a 53-kb island harboring genes for the metabolism of 1,2-PD, ethanolamine, and cobalamin, which can be found in all Listeria sensu stricto species 22 . These compounds can be found in the gastrointestinal tracts of animals and have been shown to play an important role in Salmonella pathogenesis 49,66,67 . It has been discussed whether these gene clusters are also relevant for the virulence of listeriae, particularly for colonization 54 . So far, it has been shown that ethanolamine utilization contributes to intracellular replication in ECs 62 , and that 1,2-PD is metabolized by L. innocua 68 . Transcriptional studies in gnotobiotic mice infected with L. monocytogenes showed upregulation of most pdu genes during infection compared to growth in BHI 69 . Our study not only demonstrates that L. monocytogenes can use 1,2-PD as a nutrient but also provides strong experimental evidence that the ability of L. monocytogenes to catabolize 1,2-PD contributes to its persistence and proliferation during gastrointestinal infection. This is in line with the finding that S. Typhimurium expansion is driven by 1,2-PD and that its utilization requires not only intestinal inflammation, but also the presence of commensal bacteria 50 . These are assumed to provide 1,2-PD, for example, by the fermentation of fucose that has been cleaved from the mucosal glycans by commensal bacteria 70,71 .
Toledo-Arana and colleagues showed that lmo1131 but not lmo1132 expression is decreased in intestinal L. monocytogenes of gnotobiotic mice compared with that under BHI growth conditions 69 . Their reduced transcription at 24 °C and under anaerobic conditions suggests a role in an early stage of infection. A typical prokaryotic ABC transporter is composed of two hydrophobic transmembrane domains (TMDs) and two water soluble nucleotide binding domains (NBDs) at the cytosolic side of the cell membrane. An ABC transporter can be composed of four separate polypeptides, or two identical NBDs and/or TMDs can be present. NBDs and TMDs can also be fused together thus making up either the complete transporter from a single polypeptide or from two homo-or heterodimeric polypeptides 72 . Lmo1131 and Lmo1132 seem to belong to the group of "half transporters", with two heterodimeric halves, each containing one NBD and one TMD. In addition to import functions, ABC transporters are involved in sorting molecules to the outer membrane. These molecules include lipoproteins, polysaccharides or fimbriae that might contribute to the interaction with mammalian cells 73,74 . Examples from human pathogens are the export of a coat protein from EAEC contributing to bacterial dispersion, the polysaccharide-dependent adhesion of Kingella kingae to human ECs, and the fimbriae-mediated adhesion of Streptococcus parasanguis [75][76][77] . Another example is protein F as part of an ABC transporter from Haemophilus influenzae whose N-terminus promotes binding to ECs 78 .
To conclude, non-pathogenic Listeria species have been investigated here for the first time in a mouse model for oral infection, and the results suggest differences in the colonization ability of Listeria sensu stricto and Listeria sensu lato members. Novel Listeria sensu stricto-specific factors involved in the colonization of the gastrointestinal tract were identified, whereas as the Listeria sensu lato species seem to be less adapted to the conditions of the gastrointestinal tract. Because the non-pathogenic Listeria species of the sensu stricto group including L. welshimeri are equipped with these determinants, their potential to occupy niches of the pathogenic species has been underestimated.

Methods
Bacterial strains, plasmids, cell lines and growth conditions. The bacterial strains, cell lines, and plasmids used in this study are listed in Supplementary Table S2. E. coli was grown in Luria-Bertani broth, while L. monocytogenes EGDe was cultivated in BHI or MM 51 at 37 °C or 24 °C. If appropriate, the media were supplemented with the following antibiotics: erythromycin (300 μg/ml for E. coli or 10 μg/ml for L. monocytogenes), kanamycin (50 µg/ml), and chloramphenicol (10 µg/ml). For solid media, 1.5% agar (w/v) was added. Human colon ECs (Caco-2 cells, ATCC HTB-37) and human larynx squamous cell carcinoma cells (HEp-2 cells, ATCC CCL-23) were received from the American Type Culture Collection and were cultured at 37 °C and 5% CO 2 in Dulbecco's Modified Eagle Medium (Biochrom KG, Berlin, Germany) supplemented with 10% fetal calf serum (Pan Biotech, Aidenbach, Germany). For RNA isolation, 50.5 ml of BHI was inoculated with 0.5 ml of a L. monocytogenes overnight culture and incubated at 37 °C or 24 °C with shaking (150 rpm). Aerobic growth in broth was conducted in Erlenmeyer flasks with constant shaking; anaerobic growth was performed in sealed Falcon tubes. Growth was monitored by measuring OD 600 with a Lambda Bio + spectrophotometer (Perkin Elmer, Waltham, MA, USA). If appropriate, 3.5 ml of BHI supplemented with 10 mM 1,2-PD (Sigma-Aldrich, Taufkirchen, Germany) and/or 25 nm cobalamin (Applichem, Darmstadt, Germany) in a 15-ml glass tube was inoculated with 0.25 ml of an L. monocytogenes overnight culture and incubated at 37 °C with shaking.
Standard procedures. DNA manipulations and isolation of chromosomal DNA were performed in accordance with standard protocols 79 and following the manufacturer's instructions. GeneRuler TM 1 kb DNA Ladder from Thermo Scientific (Waltham, MA, USA) was used as a marker for DNA analysis. A Bio-Rad Gene pulser II was used for electroporation. PCR was carried out with Taq polymerase and with 100-400 ng of chromosomal DNA or an aliquot of a single colony resuspended in 50 µl of H 2 O as template. Quantitative real-time (qRT)-PCR and whole bacterial RNA isolation were perfomed as described recently 80   For genome comparison, type strain genomic sequences of the 16 Listeria species were downloaded from NCBI GenBank. Sequences were uploaded to the RAST server 81 and functionally annotated. L. grayi was excluded from this comparison since it is most closely related to Listeria sensu stricto, although it belongs to the Listeria sensu lato group.
In-frame gene deletions were performed as described recently 62 . Briefly, two flanking fragments of ~1000 bp were amplified from chromosomal DNA derived from the strain EGDe using the oligonucleotides indicated in Supplementary Table S3 and ligated via the introduced BglII sites. Following nested PCR and using the ligation mixture as a template, the resulting fragment was cloned into pLSV101 via SalI and XmaI. Following transformation of the resulting plasmids, erythromycin-resistant EGDe harboring the chromosomally integrated plasmid were selected upon incubation at 42 °C. Cointegrates were resolved, erythromycin-sensitive clones were screened by PCR, and the deletion sites were sequenced to identify the respective mutant strains.
Transcriptome analysis. Whole-transcriptome RNA library preparation was performed as described recently 82 . Briefly, RNA was extracted, ribosomal RNAs were depleted, and RNA was fragmented via sonication. After dephosphorylation and rephosphorylation, TruSeq Small RNA Sample Kit (Illumina, Munich, Germany) was used, and the resulting cDNAs were size-selected. Libraries were then diluted and sequenced on a MiSeq sequencer (Illumina, Munich, Germany) using a MiSeq Reagent Kit v2 (50 cycles), resulting in 50 bp single-end reads. Illumina FASTQ files were mapped to the reference genome of L. monocytogenes EGDe (GenBank: NC_003210) using Bowtie for Illumina implemented in Galaxy 83,84 . SAM files were converted to BAM files and indexed. Artemis 85,86 was used to visualize and calculate the number of reads mapping on each gene. Gene counts of each library were normalized to the smallest library in the comparison and RPKM (reads per kilobase per million mapped reads) values were calculated. Fold changes between the different conditions were calculated and visualized using a three-color scheme.

EC adhesion and invasion assays.
A total of 2.5 × 10 5 Caco-2 or HEp-2 cells per well were seeded in a 24-well culture plate and cultivated for 48 h until infection. Cells were washed twice with PBS/Mg 2+ Ca 2+ and covered for 35 min (adhesion assay) or 1 h (invasion assay) with 500 µl of DMEM containing approximately 2.5 × 10 6 bacteria (multiplicity of infection, MOI, =10) from a glycerol stock washed with PBS. For glycerol stocks, strains were grown in 20 ml of BHI medium to mid-log phase (OD 600 ~0.85-0.95) and supplemented with glycerol (15% final concentration). Aliquots of 1 ml were frozen at −80 °C. Prior to infection, samples were thawed, and the number of viable bacteria was determined as cfu per ml.
The average MOI was calculated immediately after infection and ranged from 8 to 11. For assessment of adhesion, the Caco-2 or HEp-2 cells were washed thrice with PBS/Mg 2+ Ca 2+ after a 35-min incubation period. Cell layers were lysed in 1 ml of cold Triton X-100 (0.1%) and vortexed for 1 min to disrupt the cells. For invasion assays, Caco-2 or HEp-2 cells were washed twice with PBS/Mg 2+ Ca 2+ after 1 h of incubation. Extracellular bacteria were removed by adding 0.5 ml of DMEM containing 100 μg/ml gentamycin for 1 h, and the medium was then replaced by DMEM with 10 µg/ml gentamycin. At appropriate time points of incubation in the presence of 10 μg/ ml gentamycin, the infected eukaryotic cells were washed again with PBS/Mg 2+ Ca 2+ and then lysed in 1 ml of cold Triton X-100 (0.1%) and vortexed. Adhesion and invasion characteristics as well as intracellular replication behavior of the mutants and the wild type were quantified by plating dilutions of the lysed cells on BHI agar plates that had ben incubated at 37 °C for one day. In all experiments, intact eukaryotic cell monolayers were observed prior to cell lysis.
In vitro growth analyses. For growth analysis of L. monocytogenes EGDe and EGDe ΔpduD, we used the Bioscreen C Automated Microbiology Growth Curve Analysis System (Oy Growth Curves Ab Ltd., Helsinki, Finland), allowing automated OD measurement in a microvolume of 200 µl. Overnight cultures of L. monocytogenes grown in BHI at 37 °C were washed with PBS and then diluted in PBS to obtain an OD 600 of 1. This cell suspension was further diluted 1:20 in MM 51 containing 0.5% (w/v) yeast extract (Oxoid, Wesel Germany). MM was supplemented with 0 or 50 mM glucose (Fluka, Neu-Ulm, Germany). Cultures were incubated at 37 °C with continuous medium shaking (shaking steps: 60), and were overlaid with 200 µl of paraffin oil (Roth, Karlsruhe, Germany) to establish anaerobic conditions. The OD 600 was automatically measured every 30 min over a period of 10 h.

Mouse infections.
Four-week-old female BALBc/By/J mice from The Jackson Laboratory (Bar Harbor, ME, USA) or BALB/cAnNCrl mice from Charles River Laboratories (Sulzfeld, Germany) were purchased and used for experiments at the age of 6-8 weeks. Mice were maintained in a specific-pathogen-free facility with a 14-h light and 10-h dark cycle. Mice were infected using a model for foodborne infection as described previously 46 . Mice were placed in a cage with raised wire flooring to prevent coprophagy and denied food for 18-22 h before the infection. Aliquots of frozen L. monocytogenes or other Listeria species were recovered in BHI medium for 1.5 h at 30 °C without shaking. The desired inoculum was resuspended in 2 µl of PBS mixed with 3 µl of salted butter (Kroger, Cincinatti, OH, USA or REWE, Cologne, Germany). Cell suspension was used to saturate a 2-to 3-mm piece of bread (Kroger or REWE). After the onset of the dark cycle, mice were transferred to an empty cage and fed the Listeria-contaminated pieces of bread with sterile forceps. Afterwards, the mice were returned to their raised wire flooring cages and mouse chow was replenished. Sample collection and handling was performed as described recently 46 . We confirm that all methods were carried out in accordance with relevant guidelines and regulations, and that all experimental protocols were approved by the Regierung von Oberbayern, München, Germany.