Route of Injection Affects the Impact of InlB Internalin Domain Variants on Severity of Listeria monocytogenes Infection in Mice

The facultative intracellular pathogen Listeria monocytogenes causes a severe food-borne infection in humans and animals. L. monocytogenes invasion factor InlB interacts with the tyrosine kinase c-Met via the N-terminal internalin domain. Previously, distinct variants of the InlB internalin domain (idInlB) have been described in L. monocytogenes field isolates. Three variants were used to restore full-length InlB expression in the L. monocytogenes strain EGDeΔinlB. Obtained isogenic L. monocytogenes strains were tested in the invasion assay and intravenous, intraperitoneal, and intragastric models of infection in mice. All idInlBs were functional, restored InlB activity as an invasion factor, and improved invasion of the parental strain EGDeΔinlB into human kidney HEK23 cells. Meanwhile, distinct idInlBs provided different mortality rates and bacterial loads in internal organs. When recombinant strains were compared, the variant designated idInlB14 decreased severity of disease caused by intravenous and intraperitoneal bacterial administration, whereas this variant improved intestine colonization and stimulated intragastric infection. Obtained results demonstrated that naturally occurring idInlBs differed in their impact on severity of L. monocytogenes infection in mice in dependence on the infection route.


Introduction
The Gram-positive bacterium Listeria monocytogenes is a causative agent of food-borne infection, listeriosis, in humans and many domestic and wild animals [1,2]. Listeriosis is a rare but severe infection with lethality reaching 30% in humans [3,4]. The most serious clinical manifestations of listeriosis are CNS disorders and fetus impairment in pregnant females.
The intestine is a port of bacterial entry in the course of food-borne listeriosis [1,5]. L. monocytogenes crosses the intestinal barrier and colonizes the lamina propria from where it spreads to internal organs [5]. Systemic listeriosis includes colonization of the liver and spleen. If the infection is not controlled at this stage, a secondary bacteremia develops, and bacteria cross cerebral and/or maternal-fetal barriers to infect the brain and fetus [1,2,5,6].
L. monocytogenes is a facultative intracellular pathogen that infects macrophages and a wide range of nonprofessional phagocytes [7]. Internalin (InlA) and InlB are major factors that provide active invasion in nonprofessional phagocytes [7,8]. InlA and InlB belong to the internalin family. Internalin family proteins include an internalin domain, which consists of a central leucine rich repeat-(LRR-) domain flanked by amino-terminal N-cap and carboxy-terminal immunoglobulin-like (Ig-like) domains [9]. The internalin domain is directly involved in specific protein-protein interactions with mammalian cell surface receptors. InlB internalin domain specifically interacts with the eukaryotic receptor c-Met/HGFR (Hepatocyte Growth Factor Receptor) [10].
Mouse models are of key importance for studies of L. monocytogenes virulence. LD50 and bacterial counts in the liver and spleen are major ways used for assessment of L. monocytogenes strain virulence [11,12]. Intravenous (i.v.) and intraperitoneal (i.p.) infection routes usually require a smaller infective dose and are used more often in L. monocytogenes research than intragastric (i.g.) or oral routes [13][14][15]. When several strains were compared, their relative virulence was dependent on the route of infection. Certain strains highly virulent when injected i.p. or i.p. are less virulent when applied orally or i.g., and vice versa [16,17]. L. monocytogenes oral route infection in mice is dependent on InlB, which accelerates listerial invasion into M cells of ileal Peyer's patches (PPs) [18,19]. Previously, we and others described isolation of L. monocytogenes from internal organs of wild small rodents belonging to the Murinae subfamily, which is closely related to Mus musculus. These data suggest that mouse intestine barrier crossing is effective enough for systemic infection to be established in the natural environment [20,21]. Among 15 InlB variants found in L. monocytogenes strains, only two variants were revealed in L. monocytogenes isolated from wild small rodents [21,22]. Existence of specific InlB variants was demonstrated for L. monocytogenes strains characterized by reduced virulence [23] or associated with cardiovascular disease [24].
To evaluate an impact of naturally occurring InlB variants on L. monocytogenes virulence, we used a set of previously described isogenic strains that carried distinct InlB internalin domains (idInlBs) [25]. Virulence of the recombinant strains was compared in mouse models. Distinct idInlBs provided different mortality rates and bacterial loads in internal organs, and the difference was dependent on the route of infection.

Bacterial Strains and Growth
Conditions. Bacterial strains used in the work are listed in Table 1. L. monocytogenes were routinely cultivated in the Brain Heart Infusion (BHI, BD, USA) medium. To get a virulence regulon induction, L. monocytogenes was grown in BHI supplemented with 0,2% activated charcoal (Merck) [26]. E. coli used in cloning and expression procedures were cultivated in the LB medium (Sigma-Aldrich). All bacteria were grown at 37 ∘ C with agitation at 180 rpm. Antibiotics were added up to the following concentrations: erythromycin up to 10 and 300 g ml −1 for L. monocytogenes and E. coli, respectively; ampicillin up to 100 g ml −1 ; kanamycin up to 200 g ml −1 . All antibiotics were purchased from Sigma-Aldrich. To prepare a culture for infection, bacteria were grown up to mid-exponential phase, washed with PBS, aliquoted, and frozen in the presence of 10% glycerin. The concentration was determined by plating serial dilutions from the frozen culture the day before the experiment. Bacteria were thawed immediately before the experiment and resuspended in PBS up to the required concentration.

Recombinant Strains.
Recombinant strain construction was described previously [25]. Shortly, the idInlB encoding DNA fragments obtained on lysates of L. monocytogenes strains VIMHA004, VIMHA015, and VIMHA034 (Table 1) with primers InlB3: 5 -CAAGCGGGATCCATCACCGTG-CCAACGCC and InlB4: 5 -TATCCGGGATCCTGCTTC-TACTTTTG were cloned into the derivative of the pTRKH2 vector [29] carrying the inlB gene, lacking the internalin domain encoding part. The obtained plasmids encoded the full-length InlB with different idInlBs. The plasmids were incorporated into the L. monocytogenes EGDeΔinlB strain by electroporation.

Cell Invasion Assay.
Bacteria were prepared from the mid-exponential culture as it was described above. The "gentamycin assay" was performed as described previously [25]. Briefly, bacteria resuspended in DMEM were added with MOI 100 : 1 (bacteria to cell) to cells, incubated 1 h at 37 ∘ C in 5% CO 2 atmosphere, and washed with PBS, and then the fresh medium supplemented with 100 g ml −1 gentamycin (Fluka) was added. In 1 h, cells were washed and lysed with 1% Triton X-100. Serial dilutions of cell lysates were plated to count bacterial colonies. The invasion efficiency was determined as the ratio of intracellular bacteria to the number of applied bacteria. . Animals were sacrificed using ether inhalation. Female BALB/c mice of 16-18 g, purchased from the nursery Stolbovaya (Moscow region, Russia), were kept at our breeding facilities in specific-pathogen-free conditions and used in all experiments.

Experimental Infection in Mice.
For intravenous infection, mice were prewarmed with air heater and injected into the tail vein with bacteria suspended in 100 l PBS. The doses are indicated in the text. For intraperitoneal infection, mice were injected intraperitoneally with bacteria suspended in 200 l PBS. For intragastric infection, mice starved for 12 h were anesthetized by intraperitoneal injection of sodium pentobarbital (40 g/g) and infected with bacteria suspended in 200 l PBS by using the needle with the diameter 0.6 mm and the length 25 mm. The tip of the needle was cut and polished to smoothness to avoid stomach injury.

Mortality Rates and Bacterial Loads in Organs.
To determine mortality rates, groups of 20 animals were infected i.v. with doses of bacteria higher or near LD50. The experiment was stopped as soon as 50% lethality was observed in at least one of the groups and all remained animals were euthanized.
Groups of 5 animals were used to determine bacterial loads in infected organs. Animals were euthanized with ether at time points specified in the text. The liver and the spleen were aseptically harvested and homogenized in 1 and 0,2 ml sterile PBS, respectively. The intestine was washed out with sterile PBS; visible Peyer's patches were separated and homogenized in 5 ml of sterile PBS. Small intestine was separately homogenized in 5 ml of sterile PBS. Tenfold serial dilutions of organ homogenates were plated on BHI agar supplemented with antibiotics if required. Plates were incubated at 37 ∘ C for 24 h The detection limit of this procedure was 10, 10, and 10 2 colony forming units (CFU) per liver, spleen, and intestine, respectively.

Statistics.
All experiments were performed using duplicate samples. Results of repeated experiments were averaged. The one-tailed -test was used for assessment of statistical significance.

idInlB Variants Differentially Supported i.v. L. monocytogenes Infection.
Firstly, we evaluated an impact of idInlB variants on the final outcome of intravenous (i.v.) infection. To compare mortality rates, 2 × 10 6 CFU of parental EGDeΔinlB, wild-type EGDe, or each recombinant strain was injected into the tail vein of female BALB/c mice. This bacterial load was near LD50 for the parental strain EGDeΔinlB (LD50 ≈ 1 × 10 6 in our experimental settings) and much higher than LD50 for the wild-type strain EGDe (LD50 ≈ 1,5 × 10 4 ). The 50% mortality was reached in 72 h after infection with both EGDeΔinlB and the wild-type strain EGDe (Figure 1) that was in line with the high infection dose. Mortality caused by all but one recombinant strains had similar rates. However, bacteria carrying idInlB14 caused noticeably lower mortality rates with only 5% mortality during the first three days (Figure 1(a)).
To understand, whether the strain EGDeΔinlB::InlB14 was able to multiply within the host we followed bacterial loads in the liver and spleen of mice infected with 10 4 bacteria of this strain. A noticeable drop in bacterial loads was observed 24 hpi in comparison with the dose applied for infection: only (3,5 ± 2,3) × 10 3 CFU per mouse were found in the liver and (5,3 ± 1,8) × 10 2 CFU per mouse in the spleen (Figure 1(b)). However, after the initial drop, bacterial loads increased logarithmically suggesting that the strain is competent to multiply within the host.
Taken together, obtained results demonstrated that (i) InlB hyperproduction rather worsened than improved bacterial accumulation in the first 24 h; this effect was especially noticeable in the spleen 24 hpi; and (ii) InlB internalin domain variants differed in their ability to support i.v. infection with the idInlB14 demonstrating the worst results.

idInlB9 and idInlB14
Differentially Support Intestine Colonization. Previously, we have demonstrated that the idInlBs differed in their ability to support intragastric (i.g.) infection providing different bacterial loads in the liver but not in the spleen [25]. The idInlB14 gave rise to the highest loads in the liver 3 days after i.g. inoculation. In contrast, idInlB9, which as we demonstrated here provided high bacterial loads after i.p. injection, was low efficient in i.g. route infection [25]. The internal organ penetration after i.g. infection is not an instantaneous event as it happens after i.v. injection. Colonization of the intestine was shown to play an important role upon i.g. infection [30]. To check whether distinct idInlBs differentially affect gut colonization, EGDeΔinlB::InlB9 and EGDeΔinlB::InlB14 strains were applied i.g. in the dose of 10 8 CFU per mouse and the number of bacteria in the villous epithelium and Peyer's patches was counted 24 hpi. To count both intracellular and extracellular bacteria associated with the intestine, the intestine was washed thoroughly but it was not treated with gentamycin so both intracellular and surface attached bacteria could be counted. Counting of EGDeΔinlB::InlB14 bacteria was about 1,5 log 10 higher than EGDeΔinlB::InlB9 bacteria in Peyer's patches (Figure 3; < 0,05). The difference between strains for villous epithelium was not so high, about 3-fold (not significant).
Taken together, obtained results demonstrated that InlB internalin domain variants differed in their ability to support infection in dependence on the route of bacterial injection under otherwise equal conditions. The idInlB14 was less effective when bacteria were injected i.v. while idInlB9 was less effective upon i.g. infection. The third variant idInlB1 demonstrated medium results in all tests.  control bacteria (MOI 100 : 1, bacteria : cells). Relative invasion of EGDeΔinlB::InlB9 was 2,7-fold higher in HEK293 than EGDeΔinlB::InlB14 ( Figure 4). However, the difference between recombinant strains was minor in comparison with invasion of the parental strain EGDeΔinlB that was about 2 log 10 lower for both cell lines. Therefore, all idInlBs were functional to support invasion in epithelial cells.

Discussion
Mouse models are of key importance for studies of listeriosis. L. monocytogenes strains differ considerably in their virulence [11,12]. Moreover, an infection route affects severity of L. monocytogenes infection. The intragastric (i.g.) route is usually noticeably less effective than intravenous (i.v.) and intraperitoneal (i.p.) routes with LD50 differed by 4-6 log 10 [30,31]. Still, other works reported a smaller difference within 1-2 log 10 or a better effectiveness of the i.g. infection when alternative L. monocytogenes strains were used [16,17]. Isolation of L. monocytogenes from the internal organs of wild small rodents captured in the pristine habitats further supports the view that systemic L. monocytogenes infection in mice might take place via the oral route even at low doses of infection that might be expected in natural ecosystems [20,22,32].
Previously, we suggested that the sequence of the invasion factor InlB might be critical for L. monocytogenes spreading among small rodents in the pristine environment [21,25]. Indeed, InlB is a key invasion factor responsible for the intestine barrier crossing in mice [19]. Among 15 InlB internalin domain variants described in L. monocytogenes isolated from different hosts, only two variants were discovered in strains isolated from wild small rodents [21,25]. These are the InlB internalin domain variant 1 designated here the idInlB1, which was found in strains belonging to the phylogenetic lineage I, and the variant idInlB14 found in the lineage II strains. Comparison with available data established that idInlB14 was found in the widely used type strains 10403 and EGD but not in EGDe which was used in this work as a control and which carries the distinct idInlB [21,25,33]. Besides listed strains, idInlB14 was described by Témoin et al. in a group of strains that demonstrated low virulence in i.v. infected mice [23]. Complementation of low virulent strains with the plasmid encoding InlB from the strain EGDe improved virulence on the i.v. infection model [23]. Our data are in line with these results demonstrating that idInlB14 caused a decrease in bacterial counts after i.v. infection (see Figure 2). Meanwhile, the idInlB14 variant effectively supported the intragastric infection [25] and intestine colonization (see Figure 3).

Conclusions
Obtained results demonstrated that naturally occurring InlB internalin domain variants differed in their impact on the outcome of L. monocytogenes infection. The impact of distinct idInlBs was determined by the route of infection. The results supported the view that L. monocytogenes virulence on the models of intravenous and intraperitoneal mouse infection is not fully consistent with virulence on the intragastric route infection model. Particular idInlB variants might be responsible for this inconsistency.