Lactobacillus casei expressing Internalins A and B reduces Listeria monocytogenes interaction with Caco‐2 cells in vitro

Summary Listeria monocytogenes has been implicated in a number of outbreaks including the recent largest outbreak in South Africa. Current methods for prevention of foodborne L. monocytogenes infection are inadequate, thus raising a need for an alternative strategy. Probiotic bioengineering is considered a prevailing approach to enhance the efficacy of probiotics for targeted control of pathogens. Here, the ability of Lactobacillus casei expressing the L. monocytogenes invasion proteins Internalins A and B (inlAB) to prevent infection was investigated. The inlAB operon was cloned and surface‐expressed on L. casei resulting in a recombinant strain, LbcInl AB, and subsequently, its ability to inhibit adhesion, invasion and translocation of L. monocytogenes through enterocyte‐like Caco‐2 cells was examined. Cell surface expression of InlAB on the LbcInl AB was confirmed by Western blotting and immunofluorescence staining. The LbcInl AB strain showed significantly higher (P < 0.0001) adherence, invasion and translocation of Caco‐2 cells than the wild‐type L. casei strain (LbcWT), as well as reduced L. monocytogenes adhesion, invasion and transcellular passage through the cell monolayer than LbcWT. Furthermore, pre‐exposure of Caco‐2 cells to LbcInl AB significantly reduced L. monocytogenes‐induced cell cytotoxicity and epithelial barrier dysfunction. These results suggest that InlAB‐expressing L. casei could be a potential practical approach for prevention of listeriosis.


Introduction
Listeria monocytogenes is a Gram-positive, facultative intracellular foodborne pathogen that persists in the diverse environment within and outside mammalian host cells (Vazquez-Boland et al., 2001;Czuprynski, 2005). The severity of the disease listeriosis depends on the host immune status. The infection in immunocompetent individuals is commonly self-limiting febrile gastroenteritis, while it results in meningitis and encephalitis in immunocompromised individuals. In expectant women, it can spread to the uterus, thereby affecting the fetus, consequently causing complications such as spontaneous abortions, stillbirths or premature births (Schuchat et al., 1991;Wolfe et al., 2017). Incidences of listeriosis are much lower than diseases caused by most foodborne pathogens; however, its high case fatality rate (20-30%) has made it a considerable public health concern (Scallan et al., 2011;de Noordhout et al., 2014). The largest listeriosis outbreak ever recorded is the recent one reported in South Africa (2017)(2018), linked to consumption of ready-to-eat (RTE) sausage called Polony. It resulted in a total of 1060 cases, of which 216 were fatal (Allam et al., 2018) (http://www.nicd.ac.za/inde x.php/listeriosis-outbreak-situation-report-_4july2018/).
For host cell invasion, the pathogen uses InlA and InlB (Robbins et al., 2010;Stavru et al., 2011), which binds to the host cell receptor E-cadherin (Mengaud et al., 1996) and the hepatocyte growth factor receptor c-Met (Shen et al., 2000) respectively. InlA also aids crossing of the gut epithelial barrier by transcytosis (Nikitas et al., 2011), while InlB facilitates the invasion of human hepatic and M cells (Chiba et al., 2011;Disson and Lecuit, 2013). InlA and InlB are secreted proteins (Trost et al., 2005) and remain covalently attached to the peptidoglycan via LPXTG motif and teichoic acid via GW motif of L. monocytogenes cell wall respectively (Braun et al., 1997;Schubert et al., 2002). The bacterium then employs listeriolysin O (LLO) and phospholipases (PlcA and PlcB) to escape from the vacuoles and actin polymerization protein (ActA) to move from cell to cell (Portnoy et al., 1992;Camejo et al., 2011). There is currently no vaccine for this pathogen. Only precautionary guidance stated by the Centers for Disease Control and Prevention summarizes the importance of hygiene during food preparation and handling, as well as avoidance of certain RTE foods by high-risk groups.
Probiotics have been used to restore the balance of the gut microbial ecosystem and for control of pathogenic infections. They prevent or control foodborne illnesses through competitive exclusion of pathogens, stimulation of the host immune system and tightening of the gut barrier Behnsen et al., 2013). Several studies have reported their use to combat L. monocytogenes (Tour e et al., 2003;Corr et al., 2007;Aguilar et al., 2011). Despite the proven success of probiotics for control of enteric pathogens, they are not without shortcomings. Their disadvantages are that their action is non-specific in nature, they sometimes fail to block attachment of some pathogens to their specific receptors, and in certain instances, they induce low levels of an immune response (Bauer et al., 2002;McCarthy et al., 2003;Koo et al., 2012). Novel probiotic strains with enhanced desirable attributes can be designed by considering these limitations of traditional probiotics, as well as the behaviour and disease processes of the pathogens (O'Toole et al., 2017;do Carmo et al., 2018). These novel strains that can prevent pathogenic infections, deliver drugs or vaccines, mimic surface receptors and enhance host immune responses are developed using genetic modification (Steidler, 2003;Buccato et al., 2006;Kajikawa et al., 2007;Wells and Mercenier, 2008;Unnikrishnan et al., 2012;Amalaradjou and Bhunia, 2013;Ryan and Bhunia, 2017).
A recombinant Lactobacillus paracasei strain expressing the LAP of L. monocytogenes (Lbp LAP ) was previously developed in our laboratory, and it showed enhanced inhibition of L. monocytogenes interaction with Caco-2 cells when compared to its wild-type counterpart (Koo et al., 2012). Recently, this same gene was cloned and expressed into L. casei ATCC344 strain and the resultant recombinant strain Lbc LAP exhibited a similar anti-listeria effect (unpublished). Researchers elsewhere also cloned and expressed InlA into Lactococcus lactis for delivering DNA intracellularly (Guimaraes et al., 2005;Innocentin et al., 2009;De Azevedo et al., 2015;Yano et al., 2018). Paradoxically, none of these studies examined whether these InlA-expressing recombinant strains could prevent L. monocytogenes infection in a model system. Therefore, in the current study, our goal was to simultaneously clone and express both InlA and InlB (since both are required for cell invasion) into L. casei, a well-studied probiotic strain with proven health beneficial effects (Lenoir et al., 2016;Jacouton et al., 2017), and then investigate the ability of the resultant recombinant strain to inhibit adhesion, invasion and translocation of L. monocytogenes in vitro in a cell culture model.

InlAB was successfully cloned and expressed in Lactobacillus casei
To engineer the probiotic Lactobacillus casei expressing inlAB of L. monocytogenes, the PCR-amplified inlAB gene product and the plasmid pLP401-T were both digested with the restriction enzymes NotI and XhoI and then subsequently ligated to produce the recombinant vector designated pLP401-InlAB (Fig. S1A). This construct was electrotransferred into L. casei ATCC344 (Lbc WT ), and three selected transformants were confirmed by PCR to contain inlAB operon (Fig. 1A).
Western blot assay confirmed the expression of both InlA and InlB proteins in the different cellular fractions (supernatant, cell wall and intracellular) of Lbc InlAB while absent in Lbc WT or Lbc V (Lbc carrying only empty pLP401-T vector) cell fractions (Fig. 1B, Fig. S1B). Immunofluorescence staining also confirmed the surface expression of InlA and InlB in Lbc InlAB strain (Fig. 1C). Listeria monocytogenes F4244 (serotype 4b) was used as a positive control (Fig. 1B). These data indicate that both InlA and InlB were successfully expressed in Lbc InlAB strain and were associated with the cell wall. Transformant 1 (Lbc InlABÀ1 ) was used for the rest of the experiments.
The InlAB expression did not affect the growth rate of Lbc InlAB strain In order to determine whether the expression of InlAB affects the growth of L. casei, we compared growth curves of the Lbc WT , Lbc V and Lbc InlAB . Both optical density ( Fig. 2A) and the viable cell count (log CFU ml À1 ) ( Fig. 2B) data showed similar growth profiles for all three strains over time. Furthermore, in phase-contrast micrographs (Fig. 2C), all three strains Lbc WT , Lbc V and Lbc InlAB maintained a typical elongated curve-shaped morphology; however, Lbc V and Lbc InlAB formed slightly longer chains.

Adhesion, invasion and translocation characteristics of recombinant Lbc InlAB
We compared the abilities of the L. casei strains (Lbc WT , Lbc V and Lbc InlAB ) to adhere to, invade and translocate through or across the Caco-2 cells versus those of L. monocytogenes and Lbc LAP . Lbc WT (P = 0.8466) and Lbc V (P = 0.9964) showed similar adhesion profiles to Caco-2 cells when compared to L. monocytogenes (Fig. 3A); however, adhesion of Lbc InlAB was significantly higher than that of Lbc WT (P = 0.0153). As expected, Lbc LAP also showed higher adhesion (17.95%) than Lbc WT (11.13%). These data indicate that InlAB expression augmented the ability of Lbc InlAB strain to adhere to Caco-2 cells.
Likewise, Lbc InlAB also showed a significantly higher (P < 0.0001) transcellular translocation through epithelial (Caco-2) barrier in a trans-well set-up than the Lbc WT or  Lbc V strains (Fig. 3C). L. monocytogenes was able to invade and translocate across the Caco-2 cells at significantly higher levels (P < 0.0001) than those obtained for all the L. casei strains. Interestingly, Lbc LAP showed a significantly lower (P < 0.0001) paracellular translocation than the Lbc InlAB strain.  Probiotics inhibit pathogen colonization through the mechanism of competition, either for attachment site or for food. There are different ways by which probiotics can competitively inhibit pathogen adhesion and infection: competitive adhesion, inhibition of adhesion and displacement of adhesion (Fig. 4). Adhesion of L. monocytogenes to Caco-2 cells in the absence of L. casei strains was recorded as 100% in all the assays and was used to calculate the relative adhesion in the presence of these strains. In the competitive adhesion assay, adhesion of L. monocytogenes was significantly reduced (P < 0.0001) by 24% when coinoculated with Lbc InlAB for 1 h (Fig. 4A), while it was not reduced when it was added simultaneously with either Lbc WT (P = 0.9136) or Lbc V (P = 0.9986). Similar results were obtained for inhibition of adhesion assay where L. casei strains were allowed to adhere for 1 h before inoculation with L. monocytogenes for 1 h (Fig. 4B). Conversely, Lbc InlAB failed to displace already adhered L. monocytogenes to Caco-2 cells and showed no statistical differences (P < 0.05) when compared with Lbc WT or Lbc V (Fig. 4C). Interestingly, Lbc LAP showed significantly higher inhibition (P < 0.0007) of L. monocytogenes than Lbc InlAB (26% vs. 19%) and was unable to displace attached L. monocytogenes cells (Fig. 4).

Inhibition of L. monocytogenes adhesion, invasion and transcellular migration over time
Next, we compared the inhibitory effect of Lbc InlAB preexposed to Caco-2 cells for 1, 4, 16 and 24 h of duration against L. monocytogenes infection (adhesion, invasion and translocation) for 1 h. Overall, the results indicated that reduction of the interaction between L. monocytogenes and Caco-2 cells increased with increasing preexposure time of the Caco-2 monolayer to Lbc InlAB , while it was not or was negligibly affected by prolonged exposure to either Lbc WT or Lbc V (Fig. 5).
In the adhesion assay, Lbc InlAB reduced L. monocytogenes adhesion by 50-53.6% at 16 and 24 h, while Lbc WT and Lbc V reduced by only 8% (Fig. 5A). As a positive control, Lbc LAP showed about 64.43% reduction in L. monocytogenes adhesion to Caco-2 cells at 24 h, which is significantly higher (P < 0.0001) than that of the Lbc InlAB strain (Fig. 5A).
In the invasion assay, Lbc InlAB reduced L. monocytogenes invasion by 51.7% at 24 h, while Lbc WT and Lbc V reduced invasion by only 15%. As anticipated, Lbc LAP showed about a 32% reduction in L. monocytogenes invasion to Caco-2 cells at 24 h, which is significantly lower than Lbc InlAB (Fig. 5B).
In the transcellular translocation assay, Lbc InlAB reduced L. monocytogenes translocation by 57.14% at 24 h, while Lbc WT and Lbc V did not show any reduction at the same pre-exposure period. As a positive control, Lbc LAP showed about 52.46% reduction in L. monocytogenes translocation to Caco-2 cells at 24 h, similar to Lbc InlAB (P = 0.1595) (Fig. 5C). These results collectively indicate that InlAB-expressing L. casei reduced L. monocytogenes adhesion, invasion and transcellular translocation in the Caco-2 cell model showing a pronounced inhibitory effect after 16-24 h pre-exposure.
Inhibition of cytotoxic effects of L. monocytogenes on Caco-2 cells by L. casei We investigated the cytotoxic effect by measuring lactate dehydrogenase (LDH) release induced by Fig. 4. Competitive exclusion of Listeria monocytogenes (Lm) adhesion to Caco-2 cells by L. casei strains (Lbc WT , Lbc V , Lbc InlAB and Lbc LAP ), analysed by three different exclusion mechanisms: (A) competitive adhesion -Caco-2 cells were exposed to L. casei strains with Lm simultaneously; (B) inhibition of adhesion -Caco-2 cells were pre-exposed to L. casei strains for 1 h before infection with Lm; and (C) displacement of adhesion -Caco-2 cells were infected with Lm for 1 h before L. casei treatment (1 h). Adhesion of Lm alone to Caco-2 cells was presented as 100%, and per cent adhesion was calculated relative to that. For each time point, bars marked with different letters (a, b, c) indicate significant difference at P < 0.05. L. monocytogenes (1 h) from Caco-2 cells in the presence or absence of L. casei strains (Fig. 6). L. monocytogenes treatment for 1 h induced 64.38% cytotoxicity to Caco-2 cells in the absence of L. casei, while it induced only 5.93% and 28.7% cytotoxicity after 1 h and 24 h pre-exposure to Lbc InlAB , respectively, and by 57% and 62.3% after 1 and 24 h pre-exposure to Lbc WT respectively (Fig. 6). Interestingly, L. monocytogenes induced only 0.09% and 13.3% cytotoxicity after 1 and 24 h pre-exposure to Lbc LAP respectively. Pre-treatment of Caco-2 cells with recombinant L. casei strains resulted in their significant protection (P < 0.0001) against the cytotoxic effect of L. monocytogenes compared to pre-treatment with Lbc WT . Recombinant Lbc InlAB protects epithelial tight junction barrier integrity We further monitored the effect of recombinant L. casei strains on L. monocytogenes-mediated tight junction barrier function of Caco-2 cells by measuring the transepithelial electrical resistance (TEER) and permeability of 4 kDa of dextran FITC (FD4). The TEER value for Caco-2 cells exposed to L. monocytogenes for 2 h without L. casei pre-treatment was 16.9%. When Caco-2 cells were pre-exposed to L. casei strains, TEER values were between 9.5% and 16.7%, 2.6% and 8.53%, and 1.67% and 6.52% for Lbc WT , Lbc InlAB and Lbc LAP respectively (Fig. 7A). There was a significant (P < 0.0001) protection of epithelial barrier disruption by Lbc InlAB and Lbc LAP strains compared with Lbc WT . However, prolonged preexposure (24 h) to all L. casei strains resulted in a decrease in TEER values for all the treatments. We also measured the FD4 (paracellular marker) permeability through the epithelial barrier in a trans-well setup to assess the epithelial barrier integrity. When the Caco-2 cells were only infected with L. monocytogenes for 2 h, 2.76% of the FD4 was recovered at the basal side (Fig. 7B). The FD4 level decreased to 1.3% when Caco-2 cells were pre-exposed to Lbc InlAB , and 2.1% when pre-exposed to Lbc WT for 1 h. A similar trend was observed at 4, 16 and 24 h. As a positive control, Lbc LAP showed the highest protection against L. monocytogenesmediated epithelial barrier disruption showing FD4 permeability of only 0.1-0.3%. Nevertheless, these results show that Lbc InlAB can prevent epithelial barrier disruption from L. monocytogenes infection much greater than the Lbc WT .

Discussion
Most pathogens initiate infection of their host through the interaction of specific receptors using adhesive molecules on their surfaces (Kline et al., 2009). Beneficial bacteria (probiotics) prevent pathogen colonization by virtue of occupying the host cell surface receptors (Ryan and Bhunia, 2017;Jayashree et al., 2018). Therefore, the development of strategies to prevent pathogen interaction with the host provides a logical and effective intervention step. This can be achieved through expression of the virulence genes coding for molecules that bind to host cell receptors, in probiotic bacteria (Steidler, 2003;Paton et al., 2010;Aguilar et al., 2011;Kajikawa et al., 2011;Amalaradjou and Bhunia, 2013;Wolfe et al., 2017).
Listeria monocytogenes is responsible for a fatal infection in immunocompromised population, and pathogenesis depends on its ability to adhere and invade host cells in the gastrointestinal tract (Nikitas et al., 2011;Drolia et al., 2018;Drolia and Bhunia, 2019). Hence, blocking of adhesion and invasion events would be a logical robust option for preventing L. monocytogenes infection through its targeted inactivation (Amalaradjou and Bhunia, 2013;O'Toole et al., 2017). InlA and InlB are considered major invasion proteins required for L. monocytogenes adhesion and invasion into host cells (Radoshevich and Cossart, 2018). In this study, we successfully expressed InlA and InlB into L. casei (Lbc InlAB ) ( Fig. 1) to prevent L. monocytogenes interaction with an intestinal cell line. Molecular weight of InlB in Lbc InlAB in the cell wall fraction was found to be slightly higher (~80 kDa) than the actual MW in L. monocytogenes WT (Lm) (Fig. 1, Fig. S1) possibly because of coexpression of InlB (67 kDa) with the PrtP (PII-type Proteinase) anchor with cell wall (117 aa = 12.87 kDa) (Maassen et al., 1999), while the MW of InlA remained the same in Lbc InlAB because it possibly employed its own LPXTG motif to anchor the cell wall peptidoglycan (Bierne and Cossart, 2007). Often, the expression of new genes in a heterologous strain can result in changes in the growth and physiology of the recombinant strain (Ramos et al., 2004;Li et al., 2016). The growth rates of Lbc InlAB and Lbc WT were similar (Fig. 2), suggesting that expression of the extra genes by the recombinant L. casei did not affect its growth. This is a desirable outcome as it indicates that growth and potential consequent colonization of the recombinant would be comparable to those of the parental strain.
Expression of InlAB in Lbc InlAB strain enhanced its ability to adhere, invade and translocate across the epithelial cell barrier. Increased adhesion of Lbc InlAB to epithelial cells (Fig. 3) is highly desirable for its optimal functionality (Candela et al., 2008;Duary et al., 2011) and for creating a barrier for pathogen interaction with the host cells (Lee and Puong, 2002;Koo et al., 2012). Lbc InlAB strain exhibited higher invasion and paracellular translocation through the epithelial barrier than Lbc WT and Lbc LAP (Fig. 3). The lower invasion by Lbc LAP was expected since LAP is not involved in intracellular invasion . These findings, in part, corroborate with the previous studies where InlAexpressing L. lactis was able to invade enterocytes efficiently (Guimaraes et al., 2005;Innocentin et al., 2009;De Azevedo et al., 2015) by transcytosis (Nikitas et al., 2011;Drolia and Bhunia, 2019). The major concern remains if such a strain could cross the epithelial barrier to spread systemically and consequently cause undesirable effects such as bacteraemia or septicaemia (Didari et al., 2014). L. casei is a widely used non-pathogenic probiotic strain (Galdeano and Perdigon, 2006;Amalaradjou and Bhunia, 2012) lacking other virulence factors required for the systemic spread. It is thus anticipated that its recombinant strain expressing only InlAB will be cleared by immune cells rapidly from lamina propria. Previous studies indicated that some lactobacilli spontaneously translocate across the gut barrier; however, they were cleared within a short period by the host immune system, even when administered in higher dosages (Pavan et al., 2003;Liong, 2008). However, since these studies assessed natural infections using unmodified commensals, the safety of our recombinant strain will have to be tested and confirmed using in vivo experiments.
Others (Gueimonde et al., 2006;Collado et al., 2007) revealed that the degrees of probiotic strain adhesion and of its competitive adhesion, inhibition and/or displacement of the pathogen are not proportional. Therefore, adhesion of the probiotic should always be investigated simultaneously with its ability to reduce the adhesion of the pathogen to the same cells. Lee et al. (2003) reported that when incubated together, lactobacilli were able to compete with eight pathogens for adhesion to Caco-2 cells. However, Collado et al. (2007) found that co-incubation of probiotics and pathogens resulted in an increase in the adhesion of some pathogens. In our study, adhesion of L. monocytogenes to Caco-2 cells was similar in the presence or absence of Lbc WT , indicating the limitation of this wild-type strain to compete with this pathogen for the adhesion site on the cells (Fig. 4). Conversely, we found that both co-incubation with and pre-exposure to the recombinant Lbc InlAB and Lbc LAP significantly decreased L. monocytogenes adhesion, findings similar to previous reports (Lee and Puong, 2002;Jankowska et al., 2008;Koo et al., 2012). Furthermore, all L. casei strains were unable to displace L. monocytogenes already attached to Caco-2 cell monolayer, similar to previously published studies (Lee et al., 2003;Candela et al., 2008;Koo et al., 2012). Our results suggest that the recombinant L. casei will be effective as a prophylactic rather than a therapeutic intervention.
Next, we examined whether prolonged exposure to L. casei strains would offer higher protection against L. monocytogenes infection. A 16-24 h pre-exposure to Lbc InlAB showed the highest anti-listeria effect for all three stages of infection modalities: adhesion, invasion and translocation (Fig. 5). The anti-adhesive and antiinvasive activities of Lbc InlAB can be explained by its preoccupation of E-cadherin or c-Met receptors, which prevents L. monocytogenes adhesion and invasion by physical hindrance. Likewise, reduction in L. monocytogenes transcellular translocation is the consequential result of inhibition of its adhesion to the host cell receptor by Lbc InlAB . Lbc LAP also showed reduced L. monocytogenes translocation, which could be attributed to probiotic-induced physical hindrance and maintenance of tight junction integrity thus preventing pathogen passage (Pagnini et al., 2010;Bron et al., 2017). Indeed, both Lbc InlAB and Lbc LAP were able to prevent L. monocytogenes-mediated epithelial barrier dysfunction and helped maintain epithelial barrier integrity since dextran (paracellular marker) movement was significantly reduced in the Caco-2 monolayer from apical to the basal compartment in the trans-well set-up (Fig. 7).
In conclusion, expression of key virulence genes by probiotic strains offers an alternative strategy with potential for targeted control of L. monocytogenes infection. LAP-expressing probiotic provided protection against infection in our previous in vitro study (Koo et al., 2012). In this study, we also show that expression of InlAB by L. casei can also provide protection against infection in vitro. Therefore, recombinant Lactobacillus strains expressing different virulence genes of L. monocytogenes can be targeted at different stages of its infection cycle such as adhesion, invasion and translocation. These recombinant strains will be effective as a prophylactic rather than therapeutic intervention for pathogens and for conferring general health beneficial effects.
Although the findings reported in this paper for use of recombinant L. casei strain expressing inlA and inlB for control of L. monocytogenes infection are promising, additional in vivo studies are required to determine its suitability for direct application in humans. Such in vivo trials should determine the persistence of the recombinant strain, the stability of the plasmid and expression of foreign genes in the absence of antibiotic pressure and presence of glucose, demonstrate L. monocytogenes disease reduction and address safety issues relating to its applications.

Bacterial strains, plasmids and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 1. L. monocytogenes F4244 (serovar 4b, clinical epidemic strain) was cultured in tryptone soy broth supplemented with 0.6% yeast extract (TSB-YE) or brain heart infusion (BHI) broth at 37°C for 18 h. The vector pLP401-T (Pouwels et al., 2001) containing the pAmy promoter was used for the expression of InlAB in L. casei ATCC344. E. coli DH5a with vector was grown in Luria-Bertani (LB) broth supplemented with 50 lg ml À1 ampicillin. Wild-type L. casei (Lbc WT ) was grown in de Man Rogosa and Sharpe (MRS) broth, while the L. casei carrying the pLP401T empty vector (Lbc V ) and recombinant Lbc InlAB and Lbc LAP (unpublished) strains were grown anaerobically at 37°C for 16 h in MRS broth containing 2 lg ml À1 erythromycin. To induce expression of InlAB and LAP by recombinant L. casei strains, the recombinants were grown in modified MRS broth (1% w/v protease peptone, 0.5% w/v yeast extract, 0.2% w/v meat extract, 0.1% v/v Tween-80, 37 mM C 2 H 3 NaO 2 , 0.8 mM MgSO 4 , 0.24 mM MnSO 4 , 8.8 mM C 6 H 14 N 2 O 7 in 0.1 M potassium phosphate buffer, pH 7.0) supplemented with mannitol (1% w/v) (Koo et al., 2012) at 37°C for 16 h.

Construction of Lactobacillus casei harbouring Internalin A and B (inlAB) operons
The construction of the recombinant L. casei was done according to the methods described before (Maassen et al., 1999;Koo et al., 2012) with minor modifications. Briefly, chromosomal DNA of L. monocytogenes F4244 was extracted and inlAB operon was amplified with PCR using the primers: InlABExp-F (NotI): TAGCGGCCGC AACTATTGAAAAAGGAGTGTATATAGTG and InlA-BExp-R (XhoI): GTCTCGAGTTTCTGTGCCCTTAAATTA GC (Integrated DNA Technologies, Coralville, IA, USA) with an expected amplicon size of 4371 bp. The plasmid (pLP401T) containing the pAmy promoter was used for expression of inlAB in the probiotic L. casei ATCC344. InlA is anchored into L. casei through its own LPXTG motif while InlB has a GW motif, but it is possibly also fused to the C-terminal region of PrtP containing an LPXTG motif (Fig. S1). The plasmid and the purified DNA were digested using the restriction enzymes NotI and XhoI (NEB) and subsequently ligated (T4 DNA ligase). The product of ligation was then designated pLP401T-InlAB, which was used for electroporation into competent cells of E. coli and L. casei (Koo et al., 2012). Electroporated E. coli and L. casei cells were then incubated at 37°C for 1 h and 3 h respectively. Transformants harbouring pLP401T-inlAB were subsequently selected on LB agar containing 50 lg ml À1 ampicillin and MRS agar containing 2 lg ml À1 erythromycin for E. coli and L. casei respectively. The plates were incubated at 37°C overnight for E. coli and 72 h for L. casei. Confirmation of the identity of inlA and inlB genes in Lbc InlAB strain was done using PCR and sequencing.  (Burkholder et al., 2009;Koo et al., 2012). Equal amounts of proteins (10 lg) from each fraction were separated using SDS-PAGE(7.5%). Protein bands were transferred to an Immobilon-P membrane (Millipore, Billerica, MA, USA) and then immunoprobed with anti-InlA antibody mAb-2D12 (1.0 mg ml À1 ) (Mendonca et al., 2012) or anti-InlB pAb-404 (1:1000) (Lathrop et al., 2008) and reacted with horseradish peroxidase-coupled antimouse or anti-rabbit secondary antibodies (Jackson Immuno Research, West Grove, PA) at 37°C for 1 h. The membranes were developed with an enhanced chemiluminescence kit (Thermo Fisher, Canoga Park, CA, USA). Additionally, expression of InlA and InlB in the recombinant L. casei strains was determined by immunofluorescence staining. Overnight cultures were washed twice in PBS and incubated with the anti-InlA mAb 2D12 and anti-InlB pAb 404 (diluted 1:500 in PBS) at 37°C for 1 h. Subsequently, cells were treated with Alexa-conjugated anti-mouse IgG Fab2 Alexa Flour R555 and anti-rabbit IgG Fab2 Alexa Flour R488 (Cell Signaling, Danvers, MA, USA) secondary antibodies diluted 1:500 in PBS and incubated in the dark at 37°C for 1 h. Between the treatments, cells were washed at least four times with 0.5% PBS-Tween-20. The cells were viewed under a fluorescence microscope (Leica, Wetzlar, Germany) equipped with SPOT Software version 4.6.4.2 (Diagnostic Instruments, Sterling Heights, MI, USA).

Effect of InlAB expression on the growth of L. casei
The growth curve of Lbc WT , Lbc V and Lbc InlAB strains in MRS broth was conducted for 24 h by measuring the cell density (OD 600 nm ) in a spectrophotometer (Beckman DU80) and by plate counting. At each time point, the OD reading was taken, and culture (1 ml) was used for plating. Lbc WT cells were counted on MRS agar, while those of Lbc V and recombinant Lbc InlAB were counted on MRS agar containing 2 lg ml À1 erythromycin grown anaerobically at 37°C for 48 h. This experiment was performed twice in triplicates. Additionally, the morphologies of overnight L. casei cultures were examined using phase-contrast micrographs (Leica).

Recombinant L. casei strain adhesion and invasion into Caco-2 cells
Caco-2 cell culturing: Human colon carcinoma cell line Caco-2 (HTB37; American Type Culture Collection, Manassas, VA, USA) was cultured in Dulbecco's modified Eagle's medium (DMEM) with high glucose (HyClone TM ; GE, Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS) (Atlanta Biologicals, Flowery Branch, GA, USA) (D10F). The cells were grown in flasks (Greiner Bio-One) for up to 10-12 days or until differentiated and then trypsinized (Gaillard and Finlay, 1996). They were then seeded in 12-well plates at a density of 1 9 10 5 cells/well and incubated at 37°C in the presence of 7% CO 2 for 10-12 days until they were differentiated and reached confluence (10 6 cells/well). Adhesion and invasion assays: Overnight (18 h) grown bacterial cultures were washed twice with PBS, adjusted to OD 600 = 1 and were suspended in D10F to a final concentration of 1 9 10 7 CFU ml À1 to achieve a multiplicity of infection (MOI) or multiplicity of exposure (MOE), 10. The Caco-2 cell monolayer was washed three times using DMEM, and then exposed separately to the L. casei strains (Lbc WT , Lbc V , Lbc InlAB or Lbc LAP ) and L. monocytogenes and incubated for 1 h at 37°C in a gas atmosphere with 5% CO 2 (Koo et al., 2012). To enumerate bacterial adhesion, the Caco-2 cell monolayer was first washed thrice using DMEM and then treated with 0.1% Triton X-100 (37°C, 10 min). For the invasion assay, the monolayers were exposed to L. monocytogenes and L. casei and then washed as performed in the adhesion assay, and treated with gentamicin (50 lg ml À1 , 1 h) and with 0.1% Triton X-100 (37°C, 10 min). The lysed cell suspensions from both adhesion and invasion experiments were serially diluted in PBS before plating on MRS, supplemented with erythromycin (2 lg ml À1 ) and modified Oxford (MOX) agar for Lbc WT , recombinant L. casei and L. monocytogenes respectively. All the plates were incubated at 37°C for 24-48 h before bacterial enumeration.
Determination of L. monocytogenes exclusion mode by the recombinant L. casei strains The competitive exclusion assay was performed as before (Koo et al., 2012) with minor modifications. Bacterial cultures were prepared as above and were suspended in D10F to a final concentration of 1 9 10 7 CFU ml À1 . For competitive adhesion, L. monocytogenes was co-inoculated with each of the L. casei strains (Lbc WT , Lbc V , Lbc InlAB or Lbc LAP ) to Caco-2 cell monolayer (MOI, 10) and incubated for 1 h. Adherent bacteria were enumerated as above.
In the inhibition of adhesion assay, the Caco-2 monolayers were first inoculated with each L. casei strain (MOE, 10) and incubated for 1 h, and washed to remove unbound bacteria using DMEM. L. monocytogenes was then added to the wells, and plates were incubated for 1 h, followed by an enumeration of adherent bacteria by plating. For displacement of adhesion, Caco-2 cells were first inoculated with L. monocytogenes (MOI, 10) and incubated for 1 h, and washed to remove unbound bacteria. L. casei strains were then added to the wells, and plates were incubated for 1 h. Adhered bacteria were released by treatment with 0.1% Triton X-100 (37°C, 10 min) and plated on MRS, supplemented with 2 lg ml À1 of erythromycin and MOX agar plates for enumeration of Lbc WT , recombinant L. casei and L. monocytogenes respectively.

Inhibition of L. monocytogenes adhesion and invasion by L. casei strains
The Caco-2 cell monolayers were washed and then exposed to the L. casei strains (MOE, 10) for 1, 4, 16 and 24 h at 37°C in the humidified incubator with 5% CO 2 . Excess medium in the wells containing unbound L. casei was removed and replaced with 500 ll of L. monocytogenes suspended in D10F (MOI, 10), and the plates were incubated for 1 h at 37°C with 5% CO 2 . The adherent bacteria were enumerated by plating as above.
For inhibition of L. monocytogenes invasion, the Caco-2 cell monolayers were exposed to each L. casei strain (MOE, 10) for 1, 4, 16 and 24 h at 37°C with 5% CO 2 . Excess L. casei cells were removed and replaced with 500 ll of L. monocytogenes suspended in D10F (MOI, 10) and then incubated at 37°C with 5% CO 2 for 1 h . The cell monolayers were washed, treated with gentamicin (50 lg ml À1 ) for 1 h and determined for invading bacteria by plating.

Caco-2 cell cytotoxicity
Caco-2 cell cytotoxicity induced by L. monocytogenes after pre-exposure to L. casei strains over time was determined by using the lactate dehydrogenase (LDH) release assay as previously described (Koo et al., 2012).
Transcellular translocation of L. casei strains and subsequent inhibition of L. monocytogenes transepithelial translocation by recombinant L. casei The Caco-2 cells were grown in 12-well trans-well filter inserts (3 lm pore size) for 20-25 days to reach confluence Drolia et al., 2018). Transepithelial electrical resistance of Caco-2 cells was quantified using the Millicell ERS (Millipore), and a TEER value of more than 200 Ω cm À2 was used for all the experiments. For determining baseline translocation by L. casei strains or L. monocytogenes, the Caco-2 cells were washed, and then, the bacteria were added (MOI, 10) separately to the apical side of the trans-well at 37°C with 5% CO 2 for 2 h. The liquid from the basal well was collected, serially diluted in PBS and then plated for the enumeration of bacterial cells (CFU ml À1 ).
For the inhibition of L. monocytogenes translocation, L. casei cells were first added to the apical wells (MOE, 10) and incubated for 1, 4, 16 and 24 h at 37°C with 5% CO 2 and L. casei counts in the basal wells were determined by plating on MRS agar. Subsequently, excess L. casei cells were removed from the apical well, and replaced with 500 ll of L. monocytogenes (MOI, 10) and then incubated at 37°C with 5% CO 2 for 2 h. L. monocytogenes counts in the basal wells were determined by plating on MOX agar plates.

Epithelial tight junction integrity analysis
Quantification of TEER of Caco-2 cells before and after the exposure to the bacteria was performed using a Millicell ERS system (Millipore) as described before . Furthermore, the integrity of the tight junctions between Caco-2 cells was determined by measuring FD4 permeability in a spectrofluorometer Koo et al., 2012).

Statistical analysis
All data were analysed using Prism 7 software (Graph-Pad Software Inc., San Diego, CA, USA), and significance was assigned at P < 0.05. Where appropriate, Tukey's multiple comparisons test, with P < 0.05 as a significant difference, was used to identify statistically significant differences.