In this study, we show that IFN-III are the major interferons produced and secreted by epithelial cells in response to L.
infection, by mechanisms involving both the internalization and cytosolic phases of infection. We report that other major bacterial pathogens, M. tuberculosis
and the Gram-positive cocci S. aureus
, also stimulate type III IFN expression in epithelial cells. In contrast, Gram-negative pathogens S. enterica
and C. trachomatis
have no or only weak effects. We also provide evidence for in vivo
induction of these genes during infection of placental tissue by L. monocytogenes
. IFN-I and IFN-III genes can potentially be expressed by all nucleated cells, following the activation of pattern recognition receptors (PRR) by microbial products 
. Most studies have nevertheless focused on the expression of these genes in immune cells or fibroblasts. The results presented here highlight that there are marked and intriguing differences between bacterial species in their ability to activate IFN-III genes in epithelial cells, and this could play an important role in epithelial immunobiology.
Among different bacteria tested, L. monocytogenes
highly induces IFN-λ1/λ2
transcription in intestinal cells and other cells of epithelial origin, such as cytotrophoblast and hepatocytes, extending our previous findings 
. Induction of IFN-III genes by L. monocytogenes
in epithelial LoVo cells occurs in two waves, the first probably involving InlA- and InlB-mediated internalization, and the second involving cellular events promoted by LLO-mediated vacuolar escape. Based on these observations, we hypothesize that both vacuolar and cytosolic immune surveillance pathways contribute to IFN-III production. Indeed, it has been shown that L. monocytogenes
infection induces distinct immune responses in macrophages, depending on whether it acts on the plasma membrane, the vacuole or the cytosol 
. However, LLO-deficient bacteria fail to induce type I IFN in these immune cells, suggesting that only the cytosolic surveillance pathway is responsible for L. monocytogenes
-induced IFN-I 
. Here, we found that while internalization contributes to IFN-λ induction in epithelial cells, LLO-deficient bacteria are only partially defective in eliciting this response. Moreover, IFN-λ1/λ2
genes start to be expressed earlier than IFN-β
upon infection, and IFN-λs are produced at higher level than IFN-β. These differences suggest that IFN-III induction during epithelial cell infection may not use the same mechanisms than those leading to IFN-I expression in macrophages 
. In this respect, Listeria
could be a useful tool to investigate such specificities.
We highlight that other Gram-positive bacteria also significantly induce the expression of IFN-III genes upon internalization in epithelial cells. In particular, two major pathogens, S. aureus
and M. tuberculosis
, induce IFN-λ in LoVo intestinal and A549 lung epithelial cells, respectively. Both species lead to chronic infections in humans, and an emerging body of evidence suggests that they can reside as intracellular pathogens in epithelial cells 
, which would constitute a reservoir involved in bacterial persistence in vivo
. Interestingly, IFN-III was recently described as a modulator of the T-helper 2 (Th2) response, with inhibitory effects on Th2 cell-mediated inflammation 
, as well as a suppressor of allergy in the lung 
. Therefore, it is possible that continuous induction of IFN-III by epithelial cells in mucosa may strategically help persistence of these pathogens.
In contrast, Gram-negative S. flexneri
, S. enterica
and C. trachomatis
species do not or only weakly induce IFN-λ genes in LoVo cells, suggesting that Gram-negative and Gram-positive organisms might differentially target PRR signaling cascades leading to IFN-III production in epithelial cells. Such differences in the induction of immune response genes between different groups of bacteria have been reported in other cells types; for instance, Gram-positive and Gram-negative bacteria induce different patterns of pro-inflammatory cytokines in human monocytes 
. The differences observed do not seem to result from different amounts of intracellular bacteria () or from localization in different cellular compartments. L. monocytogenes
and S. flexneri
replicate in the cytosol with similar efficiency, yet only L.
stimulates IFN-III production. The differences between Gram-positive and Gram-negative bacteria might result from production of distinct MAMPs and/or factors influencing IFN signaling pathways, or from epithelial cell specificities in the repertoire of PRRs. In fact, LPS that is produced by both Shigella
activates IFN-λ genes in other cell types 
and S. enterica
itself induces IFN-λ in DCs 
. However, as shown here, these enteropathogens do not trigger expression of IFN-λ genes in LoVo intestinal cells. They might use specific mechanisms to actively dampen IFN-III expression in this cell type.
So far, the function of IFN-III in bacterial infection is unknown. From viral infection studies 
and owing to the receptor restricted expression pattern, it is tempting to speculate that IFN-λ contributes to epithelial innate immunity in response to bacteria, but not necessarily for the host benefit. Indeed, while type II IFN (IFN-γ) has antibacterial activity, type I IFN favors L. monocytogenes
and M. tuberculosis
. A first step before understanding the role of IFN-III in bacterial infectious diseases, in particular listeriosis, was to find out whether these genes are induced in epithelial tissues in vivo
. To address this question, we used a mouse model in which L. monocytogenes
can efficiently invade epithelial cells due to the expression of its humanized receptor E-cadherin 
colonizes several tissues of epithelial origins such as the the liver, intestine and placenta. We chose to study the expression of IFN-III in the murine placenta for three reasons: (i
) the placenta has not yet been described as an IFN-λ-producing or -responsive tissue; (ii
) IFN-λ elicits a response in the mouse intestine 
, but this tissue is in contact with the numerous bacteria of the microbiota, which may affect IFN production; (iii
) IFN-λ receptor is expressed at very low levels in mouse liver, in contrast to human liver, and thus IFN-λ has no effect in this organ in the mouse model 
. We report that levels of IFN-λ2
mRNAs and that of the IFN-responsive genes IFIT1
are increased in the placenta infected with L. monocytogenes
, indicating that IFN-III may participate in the immune response at the fetoplacental barrier. Supporting this hypothesis, cells of the fetal membranes and decidual and labyrinth zones of the mouse placenta respond to IFN-λ2 treatment ().
Whether IFN-λ could mediate protection of the fetus from invading Listeria, or alternatively, whether this pathway is beneficial for the pathogen, for instance by stimulating abortion that leads to bacterial release in the environment, deserves future investigations. In this regard, in-depth analysis of infection kinetics and the establishment of new animal models are required, in particular generation of a mouse line that would be both permissive for Listeria infection of epithelia and impaired in IFN-III responses. However, one should keep in mind that the mouse model may not be optimal to address the role of IFN-III in human listeriosis, since IFN-λ1 is a pseudogene in mice, while human cells produce this cytokine upon infection with L. monocytogenes.
Most pathogenic bacteria target tissues of epithelial origin, such as skin, throat, gut, liver, lung, genital mucosa or placenta. We propose that some bacterial species allow epithelial cells to become a source of IFN-λs, acting as paracrine immunomodulators of mucosal surfaces. Dissecting the mechanisms of IFN-III production and function during bacterial diseases may have important implications for diagnostic and therapeutic developments.